diff --git a/CHANGELOG.md b/CHANGELOG.md --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -1,5 +1,10 @@ # libarchive +## 2.2.5.1 + + * Add `static` flag to cabal which builds from bundled sources rather than + `pkg-config` + ## 2.2.5.0 * Speed improvements in places diff --git a/README.md b/README.md --- a/README.md +++ b/README.md @@ -21,7 +21,7 @@ make ``` -so that you have appropriate test data downloaded. +to download test data. ## Performance diff --git a/c/archive.h b/c/archive.h new file mode 100644 --- /dev/null +++ b/c/archive.h @@ -0,0 +1,1197 @@ +/*- + * Copyright (c) 2003-2010 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD: src/lib/libarchive/archive.h.in,v 1.50 2008/05/26 17:00:22 kientzle Exp $ + */ + +#ifndef ARCHIVE_H_INCLUDED +#define ARCHIVE_H_INCLUDED + +/* + * The version number is expressed as a single integer that makes it + * easy to compare versions at build time: for version a.b.c, the + * version number is printf("%d%03d%03d",a,b,c). For example, if you + * know your application requires version 2.12.108 or later, you can + * assert that ARCHIVE_VERSION_NUMBER >= 2012108. + */ +/* Note: Compiler will complain if this does not match archive_entry.h! */ +#define ARCHIVE_VERSION_NUMBER 3004003 + +#include +#include /* for wchar_t */ +#include /* For FILE * */ +#include /* For time_t */ + +/* + * Note: archive.h is for use outside of libarchive; the configuration + * headers (config.h, archive_platform.h, etc.) are purely internal. + * Do NOT use HAVE_XXX configuration macros to control the behavior of + * this header! If you must conditionalize, use predefined compiler and/or + * platform macros. + */ +#if defined(__BORLANDC__) && __BORLANDC__ >= 0x560 +# include +#elif !defined(__WATCOMC__) && !defined(_MSC_VER) && !defined(__INTERIX) && !defined(__BORLANDC__) && !defined(_SCO_DS) && !defined(__osf__) && !defined(__CLANG_INTTYPES_H) +# include +#endif + +/* Get appropriate definitions of 64-bit integer */ +#if !defined(__LA_INT64_T_DEFINED) +/* Older code relied on the __LA_INT64_T macro; after 4.0 we'll switch to the typedef exclusively. */ +# if ARCHIVE_VERSION_NUMBER < 4000000 +#define __LA_INT64_T la_int64_t +# endif +#define __LA_INT64_T_DEFINED +# if defined(_WIN32) && !defined(__CYGWIN__) && !defined(__WATCOMC__) +typedef __int64 la_int64_t; +# else +# include /* ssize_t */ +# if defined(_SCO_DS) || defined(__osf__) +typedef long long la_int64_t; +# else +typedef int64_t la_int64_t; +# endif +# endif +#endif + +/* The la_ssize_t should match the type used in 'struct stat' */ +#if !defined(__LA_SSIZE_T_DEFINED) +/* Older code relied on the __LA_SSIZE_T macro; after 4.0 we'll switch to the typedef exclusively. */ +# if ARCHIVE_VERSION_NUMBER < 4000000 +#define __LA_SSIZE_T la_ssize_t +# endif +#define __LA_SSIZE_T_DEFINED +# if defined(_WIN32) && !defined(__CYGWIN__) && !defined(__WATCOMC__) +# if defined(_SSIZE_T_DEFINED) || defined(_SSIZE_T_) +typedef ssize_t la_ssize_t; +# elif defined(_WIN64) +typedef __int64 la_ssize_t; +# else +typedef long la_ssize_t; +# endif +# else +# include /* ssize_t */ +typedef ssize_t la_ssize_t; +# endif +#endif + +/* Large file support for Android */ +#ifdef __ANDROID__ +#include "android_lf.h" +#endif + +/* + * On Windows, define LIBARCHIVE_STATIC if you're building or using a + * .lib. The default here assumes you're building a DLL. Only + * libarchive source should ever define __LIBARCHIVE_BUILD. + */ +#if ((defined __WIN32__) || (defined _WIN32) || defined(__CYGWIN__)) && (!defined LIBARCHIVE_STATIC) +# ifdef __LIBARCHIVE_BUILD +# ifdef __GNUC__ +# define __LA_DECL __attribute__((dllexport)) extern +# else +# define __LA_DECL __declspec(dllexport) +# endif +# else +# ifdef __GNUC__ +# define __LA_DECL +# else +# define __LA_DECL __declspec(dllimport) +# endif +# endif +#else +/* Static libraries or non-Windows needs no special declaration. */ +# define __LA_DECL +#endif + +#if defined(__GNUC__) && __GNUC__ >= 3 && !defined(__MINGW32__) +#define __LA_PRINTF(fmtarg, firstvararg) \ + __attribute__((__format__ (__printf__, fmtarg, firstvararg))) +#else +#define __LA_PRINTF(fmtarg, firstvararg) /* nothing */ +#endif + +#if defined(__GNUC__) && __GNUC__ >= 3 && __GNUC_MINOR__ >= 1 +# define __LA_DEPRECATED __attribute__((deprecated)) +#else +# define __LA_DEPRECATED +#endif + +#ifdef __cplusplus +extern "C" { +#endif + +/* + * The version number is provided as both a macro and a function. + * The macro identifies the installed header; the function identifies + * the library version (which may not be the same if you're using a + * dynamically-linked version of the library). Of course, if the + * header and library are very different, you should expect some + * strangeness. Don't do that. + */ +__LA_DECL int archive_version_number(void); + +/* + * Textual name/version of the library, useful for version displays. + */ +#define ARCHIVE_VERSION_ONLY_STRING "3.4.3" +#define ARCHIVE_VERSION_STRING "libarchive " ARCHIVE_VERSION_ONLY_STRING +__LA_DECL const char * archive_version_string(void); + +/* + * Detailed textual name/version of the library and its dependencies. + * This has the form: + * "libarchive x.y.z zlib/a.b.c liblzma/d.e.f ... etc ..." + * the list of libraries described here will vary depending on how + * libarchive was compiled. + */ +__LA_DECL const char * archive_version_details(void); + +/* + * Returns NULL if libarchive was compiled without the associated library. + * Otherwise, returns the version number that libarchive was compiled + * against. + */ +__LA_DECL const char * archive_zlib_version(void); +__LA_DECL const char * archive_liblzma_version(void); +__LA_DECL const char * archive_bzlib_version(void); +__LA_DECL const char * archive_liblz4_version(void); +__LA_DECL const char * archive_libzstd_version(void); + +/* Declare our basic types. */ +struct archive; +struct archive_entry; + +/* + * Error codes: Use archive_errno() and archive_error_string() + * to retrieve details. Unless specified otherwise, all functions + * that return 'int' use these codes. + */ +#define ARCHIVE_EOF 1 /* Found end of archive. */ +#define ARCHIVE_OK 0 /* Operation was successful. */ +#define ARCHIVE_RETRY (-10) /* Retry might succeed. */ +#define ARCHIVE_WARN (-20) /* Partial success. */ +/* For example, if write_header "fails", then you can't push data. */ +#define ARCHIVE_FAILED (-25) /* Current operation cannot complete. */ +/* But if write_header is "fatal," then this archive is dead and useless. */ +#define ARCHIVE_FATAL (-30) /* No more operations are possible. */ + +/* + * As far as possible, archive_errno returns standard platform errno codes. + * Of course, the details vary by platform, so the actual definitions + * here are stored in "archive_platform.h". The symbols are listed here + * for reference; as a rule, clients should not need to know the exact + * platform-dependent error code. + */ +/* Unrecognized or invalid file format. */ +/* #define ARCHIVE_ERRNO_FILE_FORMAT */ +/* Illegal usage of the library. */ +/* #define ARCHIVE_ERRNO_PROGRAMMER_ERROR */ +/* Unknown or unclassified error. */ +/* #define ARCHIVE_ERRNO_MISC */ + +/* + * Callbacks are invoked to automatically read/skip/write/open/close the + * archive. You can provide your own for complex tasks (like breaking + * archives across multiple tapes) or use standard ones built into the + * library. + */ + +/* Returns pointer and size of next block of data from archive. */ +typedef la_ssize_t archive_read_callback(struct archive *, + void *_client_data, const void **_buffer); + +/* Skips at most request bytes from archive and returns the skipped amount. + * This may skip fewer bytes than requested; it may even skip zero bytes. + * If you do skip fewer bytes than requested, libarchive will invoke your + * read callback and discard data as necessary to make up the full skip. + */ +typedef la_int64_t archive_skip_callback(struct archive *, + void *_client_data, la_int64_t request); + +/* Seeks to specified location in the file and returns the position. + * Whence values are SEEK_SET, SEEK_CUR, SEEK_END from stdio.h. + * Return ARCHIVE_FATAL if the seek fails for any reason. + */ +typedef la_int64_t archive_seek_callback(struct archive *, + void *_client_data, la_int64_t offset, int whence); + +/* Returns size actually written, zero on EOF, -1 on error. */ +typedef la_ssize_t archive_write_callback(struct archive *, + void *_client_data, + const void *_buffer, size_t _length); + +typedef int archive_open_callback(struct archive *, void *_client_data); + +typedef int archive_close_callback(struct archive *, void *_client_data); + +/* Switches from one client data object to the next/prev client data object. + * This is useful for reading from different data blocks such as a set of files + * that make up one large file. + */ +typedef int archive_switch_callback(struct archive *, void *_client_data1, + void *_client_data2); + +/* + * Returns a passphrase used for encryption or decryption, NULL on nothing + * to do and give it up. + */ +typedef const char *archive_passphrase_callback(struct archive *, + void *_client_data); + +/* + * Codes to identify various stream filters. + */ +#define ARCHIVE_FILTER_NONE 0 +#define ARCHIVE_FILTER_GZIP 1 +#define ARCHIVE_FILTER_BZIP2 2 +#define ARCHIVE_FILTER_COMPRESS 3 +#define ARCHIVE_FILTER_PROGRAM 4 +#define ARCHIVE_FILTER_LZMA 5 +#define ARCHIVE_FILTER_XZ 6 +#define ARCHIVE_FILTER_UU 7 +#define ARCHIVE_FILTER_RPM 8 +#define ARCHIVE_FILTER_LZIP 9 +#define ARCHIVE_FILTER_LRZIP 10 +#define ARCHIVE_FILTER_LZOP 11 +#define ARCHIVE_FILTER_GRZIP 12 +#define ARCHIVE_FILTER_LZ4 13 +#define ARCHIVE_FILTER_ZSTD 14 + +#if ARCHIVE_VERSION_NUMBER < 4000000 +#define ARCHIVE_COMPRESSION_NONE ARCHIVE_FILTER_NONE +#define ARCHIVE_COMPRESSION_GZIP ARCHIVE_FILTER_GZIP +#define ARCHIVE_COMPRESSION_BZIP2 ARCHIVE_FILTER_BZIP2 +#define ARCHIVE_COMPRESSION_COMPRESS ARCHIVE_FILTER_COMPRESS +#define ARCHIVE_COMPRESSION_PROGRAM ARCHIVE_FILTER_PROGRAM +#define ARCHIVE_COMPRESSION_LZMA ARCHIVE_FILTER_LZMA +#define ARCHIVE_COMPRESSION_XZ ARCHIVE_FILTER_XZ +#define ARCHIVE_COMPRESSION_UU ARCHIVE_FILTER_UU +#define ARCHIVE_COMPRESSION_RPM ARCHIVE_FILTER_RPM +#define ARCHIVE_COMPRESSION_LZIP ARCHIVE_FILTER_LZIP +#define ARCHIVE_COMPRESSION_LRZIP ARCHIVE_FILTER_LRZIP +#endif + +/* + * Codes returned by archive_format. + * + * Top 16 bits identifies the format family (e.g., "tar"); lower + * 16 bits indicate the variant. This is updated by read_next_header. + * Note that the lower 16 bits will often vary from entry to entry. + * In some cases, this variation occurs as libarchive learns more about + * the archive (for example, later entries might utilize extensions that + * weren't necessary earlier in the archive; in this case, libarchive + * will change the format code to indicate the extended format that + * was used). In other cases, it's because different tools have + * modified the archive and so different parts of the archive + * actually have slightly different formats. (Both tar and cpio store + * format codes in each entry, so it is quite possible for each + * entry to be in a different format.) + */ +#define ARCHIVE_FORMAT_BASE_MASK 0xff0000 +#define ARCHIVE_FORMAT_CPIO 0x10000 +#define ARCHIVE_FORMAT_CPIO_POSIX (ARCHIVE_FORMAT_CPIO | 1) +#define ARCHIVE_FORMAT_CPIO_BIN_LE (ARCHIVE_FORMAT_CPIO | 2) +#define ARCHIVE_FORMAT_CPIO_BIN_BE (ARCHIVE_FORMAT_CPIO | 3) +#define ARCHIVE_FORMAT_CPIO_SVR4_NOCRC (ARCHIVE_FORMAT_CPIO | 4) +#define ARCHIVE_FORMAT_CPIO_SVR4_CRC (ARCHIVE_FORMAT_CPIO | 5) +#define ARCHIVE_FORMAT_CPIO_AFIO_LARGE (ARCHIVE_FORMAT_CPIO | 6) +#define ARCHIVE_FORMAT_SHAR 0x20000 +#define ARCHIVE_FORMAT_SHAR_BASE (ARCHIVE_FORMAT_SHAR | 1) +#define ARCHIVE_FORMAT_SHAR_DUMP (ARCHIVE_FORMAT_SHAR | 2) +#define ARCHIVE_FORMAT_TAR 0x30000 +#define ARCHIVE_FORMAT_TAR_USTAR (ARCHIVE_FORMAT_TAR | 1) +#define ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE (ARCHIVE_FORMAT_TAR | 2) +#define ARCHIVE_FORMAT_TAR_PAX_RESTRICTED (ARCHIVE_FORMAT_TAR | 3) +#define ARCHIVE_FORMAT_TAR_GNUTAR (ARCHIVE_FORMAT_TAR | 4) +#define ARCHIVE_FORMAT_ISO9660 0x40000 +#define ARCHIVE_FORMAT_ISO9660_ROCKRIDGE (ARCHIVE_FORMAT_ISO9660 | 1) +#define ARCHIVE_FORMAT_ZIP 0x50000 +#define ARCHIVE_FORMAT_EMPTY 0x60000 +#define ARCHIVE_FORMAT_AR 0x70000 +#define ARCHIVE_FORMAT_AR_GNU (ARCHIVE_FORMAT_AR | 1) +#define ARCHIVE_FORMAT_AR_BSD (ARCHIVE_FORMAT_AR | 2) +#define ARCHIVE_FORMAT_MTREE 0x80000 +#define ARCHIVE_FORMAT_RAW 0x90000 +#define ARCHIVE_FORMAT_XAR 0xA0000 +#define ARCHIVE_FORMAT_LHA 0xB0000 +#define ARCHIVE_FORMAT_CAB 0xC0000 +#define ARCHIVE_FORMAT_RAR 0xD0000 +#define ARCHIVE_FORMAT_7ZIP 0xE0000 +#define ARCHIVE_FORMAT_WARC 0xF0000 +#define ARCHIVE_FORMAT_RAR_V5 0x100000 + +/* + * Codes returned by archive_read_format_capabilities(). + * + * This list can be extended with values between 0 and 0xffff. + * The original purpose of this list was to let different archive + * format readers expose their general capabilities in terms of + * encryption. + */ +#define ARCHIVE_READ_FORMAT_CAPS_NONE (0) /* no special capabilities */ +#define ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA (1<<0) /* reader can detect encrypted data */ +#define ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA (1<<1) /* reader can detect encryptable metadata (pathname, mtime, etc.) */ + +/* + * Codes returned by archive_read_has_encrypted_entries(). + * + * In case the archive does not support encryption detection at all + * ARCHIVE_READ_FORMAT_ENCRYPTION_UNSUPPORTED is returned. If the reader + * for some other reason (e.g. not enough bytes read) cannot say if + * there are encrypted entries, ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW + * is returned. + */ +#define ARCHIVE_READ_FORMAT_ENCRYPTION_UNSUPPORTED -2 +#define ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW -1 + +/*- + * Basic outline for reading an archive: + * 1) Ask archive_read_new for an archive reader object. + * 2) Update any global properties as appropriate. + * In particular, you'll certainly want to call appropriate + * archive_read_support_XXX functions. + * 3) Call archive_read_open_XXX to open the archive + * 4) Repeatedly call archive_read_next_header to get information about + * successive archive entries. Call archive_read_data to extract + * data for entries of interest. + * 5) Call archive_read_free to end processing. + */ +__LA_DECL struct archive *archive_read_new(void); + +/* + * The archive_read_support_XXX calls enable auto-detect for this + * archive handle. They also link in the necessary support code. + * For example, if you don't want bzlib linked in, don't invoke + * support_compression_bzip2(). The "all" functions provide the + * obvious shorthand. + */ + +#if ARCHIVE_VERSION_NUMBER < 4000000 +__LA_DECL int archive_read_support_compression_all(struct archive *) + __LA_DEPRECATED; +__LA_DECL int archive_read_support_compression_bzip2(struct archive *) + __LA_DEPRECATED; +__LA_DECL int archive_read_support_compression_compress(struct archive *) + __LA_DEPRECATED; +__LA_DECL int archive_read_support_compression_gzip(struct archive *) + __LA_DEPRECATED; +__LA_DECL int archive_read_support_compression_lzip(struct archive *) + __LA_DEPRECATED; +__LA_DECL int archive_read_support_compression_lzma(struct archive *) + __LA_DEPRECATED; +__LA_DECL int archive_read_support_compression_none(struct archive *) + __LA_DEPRECATED; +__LA_DECL int archive_read_support_compression_program(struct archive *, + const char *command) __LA_DEPRECATED; +__LA_DECL int archive_read_support_compression_program_signature + (struct archive *, const char *, + const void * /* match */, size_t) __LA_DEPRECATED; + +__LA_DECL int archive_read_support_compression_rpm(struct archive *) + __LA_DEPRECATED; +__LA_DECL int archive_read_support_compression_uu(struct archive *) + __LA_DEPRECATED; +__LA_DECL int archive_read_support_compression_xz(struct archive *) + __LA_DEPRECATED; +#endif + +__LA_DECL int archive_read_support_filter_all(struct archive *); +__LA_DECL int archive_read_support_filter_bzip2(struct archive *); +__LA_DECL int archive_read_support_filter_compress(struct archive *); +__LA_DECL int archive_read_support_filter_gzip(struct archive *); +__LA_DECL int archive_read_support_filter_grzip(struct archive *); +__LA_DECL int archive_read_support_filter_lrzip(struct archive *); +__LA_DECL int archive_read_support_filter_lz4(struct archive *); +__LA_DECL int archive_read_support_filter_lzip(struct archive *); +__LA_DECL int archive_read_support_filter_lzma(struct archive *); +__LA_DECL int archive_read_support_filter_lzop(struct archive *); +__LA_DECL int archive_read_support_filter_none(struct archive *); +__LA_DECL int archive_read_support_filter_program(struct archive *, + const char *command); +__LA_DECL int archive_read_support_filter_program_signature + (struct archive *, const char * /* cmd */, + const void * /* match */, size_t); +__LA_DECL int archive_read_support_filter_rpm(struct archive *); +__LA_DECL int archive_read_support_filter_uu(struct archive *); +__LA_DECL int archive_read_support_filter_xz(struct archive *); +__LA_DECL int archive_read_support_filter_zstd(struct archive *); + +__LA_DECL int archive_read_support_format_7zip(struct archive *); +__LA_DECL int archive_read_support_format_all(struct archive *); +__LA_DECL int archive_read_support_format_ar(struct archive *); +__LA_DECL int archive_read_support_format_by_code(struct archive *, int); +__LA_DECL int archive_read_support_format_cab(struct archive *); +__LA_DECL int archive_read_support_format_cpio(struct archive *); +__LA_DECL int archive_read_support_format_empty(struct archive *); +__LA_DECL int archive_read_support_format_gnutar(struct archive *); +__LA_DECL int archive_read_support_format_iso9660(struct archive *); +__LA_DECL int archive_read_support_format_lha(struct archive *); +__LA_DECL int archive_read_support_format_mtree(struct archive *); +__LA_DECL int archive_read_support_format_rar(struct archive *); +__LA_DECL int archive_read_support_format_rar5(struct archive *); +__LA_DECL int archive_read_support_format_raw(struct archive *); +__LA_DECL int archive_read_support_format_tar(struct archive *); +__LA_DECL int archive_read_support_format_warc(struct archive *); +__LA_DECL int archive_read_support_format_xar(struct archive *); +/* archive_read_support_format_zip() enables both streamable and seekable + * zip readers. */ +__LA_DECL int archive_read_support_format_zip(struct archive *); +/* Reads Zip archives as stream from beginning to end. Doesn't + * correctly handle SFX ZIP files or ZIP archives that have been modified + * in-place. */ +__LA_DECL int archive_read_support_format_zip_streamable(struct archive *); +/* Reads starting from central directory; requires seekable input. */ +__LA_DECL int archive_read_support_format_zip_seekable(struct archive *); + +/* Functions to manually set the format and filters to be used. This is + * useful to bypass the bidding process when the format and filters to use + * is known in advance. + */ +__LA_DECL int archive_read_set_format(struct archive *, int); +__LA_DECL int archive_read_append_filter(struct archive *, int); +__LA_DECL int archive_read_append_filter_program(struct archive *, + const char *); +__LA_DECL int archive_read_append_filter_program_signature + (struct archive *, const char *, const void * /* match */, size_t); + +/* Set various callbacks. */ +__LA_DECL int archive_read_set_open_callback(struct archive *, + archive_open_callback *); +__LA_DECL int archive_read_set_read_callback(struct archive *, + archive_read_callback *); +__LA_DECL int archive_read_set_seek_callback(struct archive *, + archive_seek_callback *); +__LA_DECL int archive_read_set_skip_callback(struct archive *, + archive_skip_callback *); +__LA_DECL int archive_read_set_close_callback(struct archive *, + archive_close_callback *); +/* Callback used to switch between one data object to the next */ +__LA_DECL int archive_read_set_switch_callback(struct archive *, + archive_switch_callback *); + +/* This sets the first data object. */ +__LA_DECL int archive_read_set_callback_data(struct archive *, void *); +/* This sets data object at specified index */ +__LA_DECL int archive_read_set_callback_data2(struct archive *, void *, + unsigned int); +/* This adds a data object at the specified index. */ +__LA_DECL int archive_read_add_callback_data(struct archive *, void *, + unsigned int); +/* This appends a data object to the end of list */ +__LA_DECL int archive_read_append_callback_data(struct archive *, void *); +/* This prepends a data object to the beginning of list */ +__LA_DECL int archive_read_prepend_callback_data(struct archive *, void *); + +/* Opening freezes the callbacks. */ +__LA_DECL int archive_read_open1(struct archive *); + +/* Convenience wrappers around the above. */ +__LA_DECL int archive_read_open(struct archive *, void *_client_data, + archive_open_callback *, archive_read_callback *, + archive_close_callback *); +__LA_DECL int archive_read_open2(struct archive *, void *_client_data, + archive_open_callback *, archive_read_callback *, + archive_skip_callback *, archive_close_callback *); + +/* + * A variety of shortcuts that invoke archive_read_open() with + * canned callbacks suitable for common situations. The ones that + * accept a block size handle tape blocking correctly. + */ +/* Use this if you know the filename. Note: NULL indicates stdin. */ +__LA_DECL int archive_read_open_filename(struct archive *, + const char *_filename, size_t _block_size); +/* Use this for reading multivolume files by filenames. + * NOTE: Must be NULL terminated. Sorting is NOT done. */ +__LA_DECL int archive_read_open_filenames(struct archive *, + const char **_filenames, size_t _block_size); +__LA_DECL int archive_read_open_filename_w(struct archive *, + const wchar_t *_filename, size_t _block_size); +/* archive_read_open_file() is a deprecated synonym for ..._open_filename(). */ +__LA_DECL int archive_read_open_file(struct archive *, + const char *_filename, size_t _block_size) __LA_DEPRECATED; +/* Read an archive that's stored in memory. */ +__LA_DECL int archive_read_open_memory(struct archive *, + const void * buff, size_t size); +/* A more involved version that is only used for internal testing. */ +__LA_DECL int archive_read_open_memory2(struct archive *a, const void *buff, + size_t size, size_t read_size); +/* Read an archive that's already open, using the file descriptor. */ +__LA_DECL int archive_read_open_fd(struct archive *, int _fd, + size_t _block_size); +/* Read an archive that's already open, using a FILE *. */ +/* Note: DO NOT use this with tape drives. */ +__LA_DECL int archive_read_open_FILE(struct archive *, FILE *_file); + +/* Parses and returns next entry header. */ +__LA_DECL int archive_read_next_header(struct archive *, + struct archive_entry **); + +/* Parses and returns next entry header using the archive_entry passed in */ +__LA_DECL int archive_read_next_header2(struct archive *, + struct archive_entry *); + +/* + * Retrieve the byte offset in UNCOMPRESSED data where last-read + * header started. + */ +__LA_DECL la_int64_t archive_read_header_position(struct archive *); + +/* + * Returns 1 if the archive contains at least one encrypted entry. + * If the archive format not support encryption at all + * ARCHIVE_READ_FORMAT_ENCRYPTION_UNSUPPORTED is returned. + * If for any other reason (e.g. not enough data read so far) + * we cannot say whether there are encrypted entries, then + * ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW is returned. + * In general, this function will return values below zero when the + * reader is uncertain or totally incapable of encryption support. + * When this function returns 0 you can be sure that the reader + * supports encryption detection but no encrypted entries have + * been found yet. + * + * NOTE: If the metadata/header of an archive is also encrypted, you + * cannot rely on the number of encrypted entries. That is why this + * function does not return the number of encrypted entries but# + * just shows that there are some. + */ +__LA_DECL int archive_read_has_encrypted_entries(struct archive *); + +/* + * Returns a bitmask of capabilities that are supported by the archive format reader. + * If the reader has no special capabilities, ARCHIVE_READ_FORMAT_CAPS_NONE is returned. + */ +__LA_DECL int archive_read_format_capabilities(struct archive *); + +/* Read data from the body of an entry. Similar to read(2). */ +__LA_DECL la_ssize_t archive_read_data(struct archive *, + void *, size_t); + +/* Seek within the body of an entry. Similar to lseek(2). */ +__LA_DECL la_int64_t archive_seek_data(struct archive *, la_int64_t, int); + +/* + * A zero-copy version of archive_read_data that also exposes the file offset + * of each returned block. Note that the client has no way to specify + * the desired size of the block. The API does guarantee that offsets will + * be strictly increasing and that returned blocks will not overlap. + */ +__LA_DECL int archive_read_data_block(struct archive *a, + const void **buff, size_t *size, la_int64_t *offset); + +/*- + * Some convenience functions that are built on archive_read_data: + * 'skip': skips entire entry + * 'into_buffer': writes data into memory buffer that you provide + * 'into_fd': writes data to specified filedes + */ +__LA_DECL int archive_read_data_skip(struct archive *); +__LA_DECL int archive_read_data_into_fd(struct archive *, int fd); + +/* + * Set read options. + */ +/* Apply option to the format only. */ +__LA_DECL int archive_read_set_format_option(struct archive *_a, + const char *m, const char *o, + const char *v); +/* Apply option to the filter only. */ +__LA_DECL int archive_read_set_filter_option(struct archive *_a, + const char *m, const char *o, + const char *v); +/* Apply option to both the format and the filter. */ +__LA_DECL int archive_read_set_option(struct archive *_a, + const char *m, const char *o, + const char *v); +/* Apply option string to both the format and the filter. */ +__LA_DECL int archive_read_set_options(struct archive *_a, + const char *opts); + +/* + * Add a decryption passphrase. + */ +__LA_DECL int archive_read_add_passphrase(struct archive *, const char *); +__LA_DECL int archive_read_set_passphrase_callback(struct archive *, + void *client_data, archive_passphrase_callback *); + + +/*- + * Convenience function to recreate the current entry (whose header + * has just been read) on disk. + * + * This does quite a bit more than just copy data to disk. It also: + * - Creates intermediate directories as required. + * - Manages directory permissions: non-writable directories will + * be initially created with write permission enabled; when the + * archive is closed, dir permissions are edited to the values specified + * in the archive. + * - Checks hardlinks: hardlinks will not be extracted unless the + * linked-to file was also extracted within the same session. (TODO) + */ + +/* The "flags" argument selects optional behavior, 'OR' the flags you want. */ + +/* Default: Do not try to set owner/group. */ +#define ARCHIVE_EXTRACT_OWNER (0x0001) +/* Default: Do obey umask, do not restore SUID/SGID/SVTX bits. */ +#define ARCHIVE_EXTRACT_PERM (0x0002) +/* Default: Do not restore mtime/atime. */ +#define ARCHIVE_EXTRACT_TIME (0x0004) +/* Default: Replace existing files. */ +#define ARCHIVE_EXTRACT_NO_OVERWRITE (0x0008) +/* Default: Try create first, unlink only if create fails with EEXIST. */ +#define ARCHIVE_EXTRACT_UNLINK (0x0010) +/* Default: Do not restore ACLs. */ +#define ARCHIVE_EXTRACT_ACL (0x0020) +/* Default: Do not restore fflags. */ +#define ARCHIVE_EXTRACT_FFLAGS (0x0040) +/* Default: Do not restore xattrs. */ +#define ARCHIVE_EXTRACT_XATTR (0x0080) +/* Default: Do not try to guard against extracts redirected by symlinks. */ +/* Note: With ARCHIVE_EXTRACT_UNLINK, will remove any intermediate symlink. */ +#define ARCHIVE_EXTRACT_SECURE_SYMLINKS (0x0100) +/* Default: Do not reject entries with '..' as path elements. */ +#define ARCHIVE_EXTRACT_SECURE_NODOTDOT (0x0200) +/* Default: Create parent directories as needed. */ +#define ARCHIVE_EXTRACT_NO_AUTODIR (0x0400) +/* Default: Overwrite files, even if one on disk is newer. */ +#define ARCHIVE_EXTRACT_NO_OVERWRITE_NEWER (0x0800) +/* Detect blocks of 0 and write holes instead. */ +#define ARCHIVE_EXTRACT_SPARSE (0x1000) +/* Default: Do not restore Mac extended metadata. */ +/* This has no effect except on Mac OS. */ +#define ARCHIVE_EXTRACT_MAC_METADATA (0x2000) +/* Default: Use HFS+ compression if it was compressed. */ +/* This has no effect except on Mac OS v10.6 or later. */ +#define ARCHIVE_EXTRACT_NO_HFS_COMPRESSION (0x4000) +/* Default: Do not use HFS+ compression if it was not compressed. */ +/* This has no effect except on Mac OS v10.6 or later. */ +#define ARCHIVE_EXTRACT_HFS_COMPRESSION_FORCED (0x8000) +/* Default: Do not reject entries with absolute paths */ +#define ARCHIVE_EXTRACT_SECURE_NOABSOLUTEPATHS (0x10000) +/* Default: Do not clear no-change flags when unlinking object */ +#define ARCHIVE_EXTRACT_CLEAR_NOCHANGE_FFLAGS (0x20000) +/* Default: Do not extract atomically (using rename) */ +#define ARCHIVE_EXTRACT_SAFE_WRITES (0x40000) + +__LA_DECL int archive_read_extract(struct archive *, struct archive_entry *, + int flags); +__LA_DECL int archive_read_extract2(struct archive *, struct archive_entry *, + struct archive * /* dest */); +__LA_DECL void archive_read_extract_set_progress_callback(struct archive *, + void (*_progress_func)(void *), void *_user_data); + +/* Record the dev/ino of a file that will not be written. This is + * generally set to the dev/ino of the archive being read. */ +__LA_DECL void archive_read_extract_set_skip_file(struct archive *, + la_int64_t, la_int64_t); + +/* Close the file and release most resources. */ +__LA_DECL int archive_read_close(struct archive *); +/* Release all resources and destroy the object. */ +/* Note that archive_read_free will call archive_read_close for you. */ +__LA_DECL int archive_read_free(struct archive *); +#if ARCHIVE_VERSION_NUMBER < 4000000 +/* Synonym for archive_read_free() for backwards compatibility. */ +__LA_DECL int archive_read_finish(struct archive *) __LA_DEPRECATED; +#endif + +/*- + * To create an archive: + * 1) Ask archive_write_new for an archive writer object. + * 2) Set any global properties. In particular, you should set + * the compression and format to use. + * 3) Call archive_write_open to open the file (most people + * will use archive_write_open_file or archive_write_open_fd, + * which provide convenient canned I/O callbacks for you). + * 4) For each entry: + * - construct an appropriate struct archive_entry structure + * - archive_write_header to write the header + * - archive_write_data to write the entry data + * 5) archive_write_close to close the output + * 6) archive_write_free to cleanup the writer and release resources + */ +__LA_DECL struct archive *archive_write_new(void); +__LA_DECL int archive_write_set_bytes_per_block(struct archive *, + int bytes_per_block); +__LA_DECL int archive_write_get_bytes_per_block(struct archive *); +/* XXX This is badly misnamed; suggestions appreciated. XXX */ +__LA_DECL int archive_write_set_bytes_in_last_block(struct archive *, + int bytes_in_last_block); +__LA_DECL int archive_write_get_bytes_in_last_block(struct archive *); + +/* The dev/ino of a file that won't be archived. This is used + * to avoid recursively adding an archive to itself. */ +__LA_DECL int archive_write_set_skip_file(struct archive *, + la_int64_t, la_int64_t); + +#if ARCHIVE_VERSION_NUMBER < 4000000 +__LA_DECL int archive_write_set_compression_bzip2(struct archive *) + __LA_DEPRECATED; +__LA_DECL int archive_write_set_compression_compress(struct archive *) + __LA_DEPRECATED; +__LA_DECL int archive_write_set_compression_gzip(struct archive *) + __LA_DEPRECATED; +__LA_DECL int archive_write_set_compression_lzip(struct archive *) + __LA_DEPRECATED; +__LA_DECL int archive_write_set_compression_lzma(struct archive *) + __LA_DEPRECATED; +__LA_DECL int archive_write_set_compression_none(struct archive *) + __LA_DEPRECATED; +__LA_DECL int archive_write_set_compression_program(struct archive *, + const char *cmd) __LA_DEPRECATED; +__LA_DECL int archive_write_set_compression_xz(struct archive *) + __LA_DEPRECATED; +#endif + +/* A convenience function to set the filter based on the code. */ +__LA_DECL int archive_write_add_filter(struct archive *, int filter_code); +__LA_DECL int archive_write_add_filter_by_name(struct archive *, + const char *name); +__LA_DECL int archive_write_add_filter_b64encode(struct archive *); +__LA_DECL int archive_write_add_filter_bzip2(struct archive *); +__LA_DECL int archive_write_add_filter_compress(struct archive *); +__LA_DECL int archive_write_add_filter_grzip(struct archive *); +__LA_DECL int archive_write_add_filter_gzip(struct archive *); +__LA_DECL int archive_write_add_filter_lrzip(struct archive *); +__LA_DECL int archive_write_add_filter_lz4(struct archive *); +__LA_DECL int archive_write_add_filter_lzip(struct archive *); +__LA_DECL int archive_write_add_filter_lzma(struct archive *); +__LA_DECL int archive_write_add_filter_lzop(struct archive *); +__LA_DECL int archive_write_add_filter_none(struct archive *); +__LA_DECL int archive_write_add_filter_program(struct archive *, + const char *cmd); +__LA_DECL int archive_write_add_filter_uuencode(struct archive *); +__LA_DECL int archive_write_add_filter_xz(struct archive *); +__LA_DECL int archive_write_add_filter_zstd(struct archive *); + + +/* A convenience function to set the format based on the code or name. */ +__LA_DECL int archive_write_set_format(struct archive *, int format_code); +__LA_DECL int archive_write_set_format_by_name(struct archive *, + const char *name); +/* To minimize link pollution, use one or more of the following. */ +__LA_DECL int archive_write_set_format_7zip(struct archive *); +__LA_DECL int archive_write_set_format_ar_bsd(struct archive *); +__LA_DECL int archive_write_set_format_ar_svr4(struct archive *); +__LA_DECL int archive_write_set_format_cpio(struct archive *); +__LA_DECL int archive_write_set_format_cpio_newc(struct archive *); +__LA_DECL int archive_write_set_format_gnutar(struct archive *); +__LA_DECL int archive_write_set_format_iso9660(struct archive *); +__LA_DECL int archive_write_set_format_mtree(struct archive *); +__LA_DECL int archive_write_set_format_mtree_classic(struct archive *); +/* TODO: int archive_write_set_format_old_tar(struct archive *); */ +__LA_DECL int archive_write_set_format_pax(struct archive *); +__LA_DECL int archive_write_set_format_pax_restricted(struct archive *); +__LA_DECL int archive_write_set_format_raw(struct archive *); +__LA_DECL int archive_write_set_format_shar(struct archive *); +__LA_DECL int archive_write_set_format_shar_dump(struct archive *); +__LA_DECL int archive_write_set_format_ustar(struct archive *); +__LA_DECL int archive_write_set_format_v7tar(struct archive *); +__LA_DECL int archive_write_set_format_warc(struct archive *); +__LA_DECL int archive_write_set_format_xar(struct archive *); +__LA_DECL int archive_write_set_format_zip(struct archive *); +__LA_DECL int archive_write_set_format_filter_by_ext(struct archive *a, const char *filename); +__LA_DECL int archive_write_set_format_filter_by_ext_def(struct archive *a, const char *filename, const char * def_ext); +__LA_DECL int archive_write_zip_set_compression_deflate(struct archive *); +__LA_DECL int archive_write_zip_set_compression_store(struct archive *); +__LA_DECL int archive_write_open(struct archive *, void *, + archive_open_callback *, archive_write_callback *, + archive_close_callback *); +__LA_DECL int archive_write_open_fd(struct archive *, int _fd); +__LA_DECL int archive_write_open_filename(struct archive *, const char *_file); +__LA_DECL int archive_write_open_filename_w(struct archive *, + const wchar_t *_file); +/* A deprecated synonym for archive_write_open_filename() */ +__LA_DECL int archive_write_open_file(struct archive *, const char *_file) + __LA_DEPRECATED; +__LA_DECL int archive_write_open_FILE(struct archive *, FILE *); +/* _buffSize is the size of the buffer, _used refers to a variable that + * will be updated after each write into the buffer. */ +__LA_DECL int archive_write_open_memory(struct archive *, + void *_buffer, size_t _buffSize, size_t *_used); + +/* + * Note that the library will truncate writes beyond the size provided + * to archive_write_header or pad if the provided data is short. + */ +__LA_DECL int archive_write_header(struct archive *, + struct archive_entry *); +__LA_DECL la_ssize_t archive_write_data(struct archive *, + const void *, size_t); + +/* This interface is currently only available for archive_write_disk handles. */ +__LA_DECL la_ssize_t archive_write_data_block(struct archive *, + const void *, size_t, la_int64_t); + +__LA_DECL int archive_write_finish_entry(struct archive *); +__LA_DECL int archive_write_close(struct archive *); +/* Marks the archive as FATAL so that a subsequent free() operation + * won't try to close() cleanly. Provides a fast abort capability + * when the client discovers that things have gone wrong. */ +__LA_DECL int archive_write_fail(struct archive *); +/* This can fail if the archive wasn't already closed, in which case + * archive_write_free() will implicitly call archive_write_close(). */ +__LA_DECL int archive_write_free(struct archive *); +#if ARCHIVE_VERSION_NUMBER < 4000000 +/* Synonym for archive_write_free() for backwards compatibility. */ +__LA_DECL int archive_write_finish(struct archive *) __LA_DEPRECATED; +#endif + +/* + * Set write options. + */ +/* Apply option to the format only. */ +__LA_DECL int archive_write_set_format_option(struct archive *_a, + const char *m, const char *o, + const char *v); +/* Apply option to the filter only. */ +__LA_DECL int archive_write_set_filter_option(struct archive *_a, + const char *m, const char *o, + const char *v); +/* Apply option to both the format and the filter. */ +__LA_DECL int archive_write_set_option(struct archive *_a, + const char *m, const char *o, + const char *v); +/* Apply option string to both the format and the filter. */ +__LA_DECL int archive_write_set_options(struct archive *_a, + const char *opts); + +/* + * Set a encryption passphrase. + */ +__LA_DECL int archive_write_set_passphrase(struct archive *_a, const char *p); +__LA_DECL int archive_write_set_passphrase_callback(struct archive *, + void *client_data, archive_passphrase_callback *); + +/*- + * ARCHIVE_WRITE_DISK API + * + * To create objects on disk: + * 1) Ask archive_write_disk_new for a new archive_write_disk object. + * 2) Set any global properties. In particular, you probably + * want to set the options. + * 3) For each entry: + * - construct an appropriate struct archive_entry structure + * - archive_write_header to create the file/dir/etc on disk + * - archive_write_data to write the entry data + * 4) archive_write_free to cleanup the writer and release resources + * + * In particular, you can use this in conjunction with archive_read() + * to pull entries out of an archive and create them on disk. + */ +__LA_DECL struct archive *archive_write_disk_new(void); +/* This file will not be overwritten. */ +__LA_DECL int archive_write_disk_set_skip_file(struct archive *, + la_int64_t, la_int64_t); +/* Set flags to control how the next item gets created. + * This accepts a bitmask of ARCHIVE_EXTRACT_XXX flags defined above. */ +__LA_DECL int archive_write_disk_set_options(struct archive *, + int flags); +/* + * The lookup functions are given uname/uid (or gname/gid) pairs and + * return a uid (gid) suitable for this system. These are used for + * restoring ownership and for setting ACLs. The default functions + * are naive, they just return the uid/gid. These are small, so reasonable + * for applications that don't need to preserve ownership; they + * are probably also appropriate for applications that are doing + * same-system backup and restore. + */ +/* + * The "standard" lookup functions use common system calls to lookup + * the uname/gname, falling back to the uid/gid if the names can't be + * found. They cache lookups and are reasonably fast, but can be very + * large, so they are not used unless you ask for them. In + * particular, these match the specifications of POSIX "pax" and old + * POSIX "tar". + */ +__LA_DECL int archive_write_disk_set_standard_lookup(struct archive *); +/* + * If neither the default (naive) nor the standard (big) functions suit + * your needs, you can write your own and register them. Be sure to + * include a cleanup function if you have allocated private data. + */ +__LA_DECL int archive_write_disk_set_group_lookup(struct archive *, + void * /* private_data */, + la_int64_t (*)(void *, const char *, la_int64_t), + void (* /* cleanup */)(void *)); +__LA_DECL int archive_write_disk_set_user_lookup(struct archive *, + void * /* private_data */, + la_int64_t (*)(void *, const char *, la_int64_t), + void (* /* cleanup */)(void *)); +__LA_DECL la_int64_t archive_write_disk_gid(struct archive *, const char *, la_int64_t); +__LA_DECL la_int64_t archive_write_disk_uid(struct archive *, const char *, la_int64_t); + +/* + * ARCHIVE_READ_DISK API + * + * This is still evolving and somewhat experimental. + */ +__LA_DECL struct archive *archive_read_disk_new(void); +/* The names for symlink modes here correspond to an old BSD + * command-line argument convention: -L, -P, -H */ +/* Follow all symlinks. */ +__LA_DECL int archive_read_disk_set_symlink_logical(struct archive *); +/* Follow no symlinks. */ +__LA_DECL int archive_read_disk_set_symlink_physical(struct archive *); +/* Follow symlink initially, then not. */ +__LA_DECL int archive_read_disk_set_symlink_hybrid(struct archive *); +/* TODO: Handle Linux stat32/stat64 ugliness. */ +__LA_DECL int archive_read_disk_entry_from_file(struct archive *, + struct archive_entry *, int /* fd */, const struct stat *); +/* Look up gname for gid or uname for uid. */ +/* Default implementations are very, very stupid. */ +__LA_DECL const char *archive_read_disk_gname(struct archive *, la_int64_t); +__LA_DECL const char *archive_read_disk_uname(struct archive *, la_int64_t); +/* "Standard" implementation uses getpwuid_r, getgrgid_r and caches the + * results for performance. */ +__LA_DECL int archive_read_disk_set_standard_lookup(struct archive *); +/* You can install your own lookups if you like. */ +__LA_DECL int archive_read_disk_set_gname_lookup(struct archive *, + void * /* private_data */, + const char *(* /* lookup_fn */)(void *, la_int64_t), + void (* /* cleanup_fn */)(void *)); +__LA_DECL int archive_read_disk_set_uname_lookup(struct archive *, + void * /* private_data */, + const char *(* /* lookup_fn */)(void *, la_int64_t), + void (* /* cleanup_fn */)(void *)); +/* Start traversal. */ +__LA_DECL int archive_read_disk_open(struct archive *, const char *); +__LA_DECL int archive_read_disk_open_w(struct archive *, const wchar_t *); +/* + * Request that current entry be visited. If you invoke it on every + * directory, you'll get a physical traversal. This is ignored if the + * current entry isn't a directory or a link to a directory. So, if + * you invoke this on every returned path, you'll get a full logical + * traversal. + */ +__LA_DECL int archive_read_disk_descend(struct archive *); +__LA_DECL int archive_read_disk_can_descend(struct archive *); +__LA_DECL int archive_read_disk_current_filesystem(struct archive *); +__LA_DECL int archive_read_disk_current_filesystem_is_synthetic(struct archive *); +__LA_DECL int archive_read_disk_current_filesystem_is_remote(struct archive *); +/* Request that the access time of the entry visited by traversal be restored. */ +__LA_DECL int archive_read_disk_set_atime_restored(struct archive *); +/* + * Set behavior. The "flags" argument selects optional behavior. + */ +/* Request that the access time of the entry visited by traversal be restored. + * This is the same as archive_read_disk_set_atime_restored. */ +#define ARCHIVE_READDISK_RESTORE_ATIME (0x0001) +/* Default: Do not skip an entry which has nodump flags. */ +#define ARCHIVE_READDISK_HONOR_NODUMP (0x0002) +/* Default: Skip a mac resource fork file whose prefix is "._" because of + * using copyfile. */ +#define ARCHIVE_READDISK_MAC_COPYFILE (0x0004) +/* Default: Traverse mount points. */ +#define ARCHIVE_READDISK_NO_TRAVERSE_MOUNTS (0x0008) +/* Default: Xattrs are read from disk. */ +#define ARCHIVE_READDISK_NO_XATTR (0x0010) +/* Default: ACLs are read from disk. */ +#define ARCHIVE_READDISK_NO_ACL (0x0020) +/* Default: File flags are read from disk. */ +#define ARCHIVE_READDISK_NO_FFLAGS (0x0040) + +__LA_DECL int archive_read_disk_set_behavior(struct archive *, + int flags); + +/* + * Set archive_match object that will be used in archive_read_disk to + * know whether an entry should be skipped. The callback function + * _excluded_func will be invoked when an entry is skipped by the result + * of archive_match. + */ +__LA_DECL int archive_read_disk_set_matching(struct archive *, + struct archive *_matching, void (*_excluded_func) + (struct archive *, void *, struct archive_entry *), + void *_client_data); +__LA_DECL int archive_read_disk_set_metadata_filter_callback(struct archive *, + int (*_metadata_filter_func)(struct archive *, void *, + struct archive_entry *), void *_client_data); + +/* Simplified cleanup interface; + * This calls archive_read_free() or archive_write_free() as needed. */ +__LA_DECL int archive_free(struct archive *); + +/* + * Accessor functions to read/set various information in + * the struct archive object: + */ + +/* Number of filters in the current filter pipeline. */ +/* Filter #0 is the one closest to the format, -1 is a synonym for the + * last filter, which is always the pseudo-filter that wraps the + * client callbacks. */ +__LA_DECL int archive_filter_count(struct archive *); +__LA_DECL la_int64_t archive_filter_bytes(struct archive *, int); +__LA_DECL int archive_filter_code(struct archive *, int); +__LA_DECL const char * archive_filter_name(struct archive *, int); + +#if ARCHIVE_VERSION_NUMBER < 4000000 +/* These don't properly handle multiple filters, so are deprecated and + * will eventually be removed. */ +/* As of libarchive 3.0, this is an alias for archive_filter_bytes(a, -1); */ +__LA_DECL la_int64_t archive_position_compressed(struct archive *) + __LA_DEPRECATED; +/* As of libarchive 3.0, this is an alias for archive_filter_bytes(a, 0); */ +__LA_DECL la_int64_t archive_position_uncompressed(struct archive *) + __LA_DEPRECATED; +/* As of libarchive 3.0, this is an alias for archive_filter_name(a, 0); */ +__LA_DECL const char *archive_compression_name(struct archive *) + __LA_DEPRECATED; +/* As of libarchive 3.0, this is an alias for archive_filter_code(a, 0); */ +__LA_DECL int archive_compression(struct archive *) + __LA_DEPRECATED; +#endif + +__LA_DECL int archive_errno(struct archive *); +__LA_DECL const char *archive_error_string(struct archive *); +__LA_DECL const char *archive_format_name(struct archive *); +__LA_DECL int archive_format(struct archive *); +__LA_DECL void archive_clear_error(struct archive *); +__LA_DECL void archive_set_error(struct archive *, int _err, + const char *fmt, ...) __LA_PRINTF(3, 4); +__LA_DECL void archive_copy_error(struct archive *dest, + struct archive *src); +__LA_DECL int archive_file_count(struct archive *); + +/* + * ARCHIVE_MATCH API + */ +__LA_DECL struct archive *archive_match_new(void); +__LA_DECL int archive_match_free(struct archive *); + +/* + * Test if archive_entry is excluded. + * This is a convenience function. This is the same as calling all + * archive_match_path_excluded, archive_match_time_excluded + * and archive_match_owner_excluded. + */ +__LA_DECL int archive_match_excluded(struct archive *, + struct archive_entry *); + +/* + * Test if pathname is excluded. The conditions are set by following functions. + */ +__LA_DECL int archive_match_path_excluded(struct archive *, + struct archive_entry *); +/* Control recursive inclusion of directory content when directory is included. Default on. */ +__LA_DECL int archive_match_set_inclusion_recursion(struct archive *, int); +/* Add exclusion pathname pattern. */ +__LA_DECL int archive_match_exclude_pattern(struct archive *, const char *); +__LA_DECL int archive_match_exclude_pattern_w(struct archive *, + const wchar_t *); +/* Add exclusion pathname pattern from file. */ +__LA_DECL int archive_match_exclude_pattern_from_file(struct archive *, + const char *, int _nullSeparator); +__LA_DECL int archive_match_exclude_pattern_from_file_w(struct archive *, + const wchar_t *, int _nullSeparator); +/* Add inclusion pathname pattern. */ +__LA_DECL int archive_match_include_pattern(struct archive *, const char *); +__LA_DECL int archive_match_include_pattern_w(struct archive *, + const wchar_t *); +/* Add inclusion pathname pattern from file. */ +__LA_DECL int archive_match_include_pattern_from_file(struct archive *, + const char *, int _nullSeparator); +__LA_DECL int archive_match_include_pattern_from_file_w(struct archive *, + const wchar_t *, int _nullSeparator); +/* + * How to get statistic information for inclusion patterns. + */ +/* Return the amount number of unmatched inclusion patterns. */ +__LA_DECL int archive_match_path_unmatched_inclusions(struct archive *); +/* Return the pattern of unmatched inclusion with ARCHIVE_OK. + * Return ARCHIVE_EOF if there is no inclusion pattern. */ +__LA_DECL int archive_match_path_unmatched_inclusions_next( + struct archive *, const char **); +__LA_DECL int archive_match_path_unmatched_inclusions_next_w( + struct archive *, const wchar_t **); + +/* + * Test if a file is excluded by its time stamp. + * The conditions are set by following functions. + */ +__LA_DECL int archive_match_time_excluded(struct archive *, + struct archive_entry *); + +/* + * Flags to tell a matching type of time stamps. These are used for + * following functions. + */ +/* Time flag: mtime to be tested. */ +#define ARCHIVE_MATCH_MTIME (0x0100) +/* Time flag: ctime to be tested. */ +#define ARCHIVE_MATCH_CTIME (0x0200) +/* Comparison flag: Match the time if it is newer than. */ +#define ARCHIVE_MATCH_NEWER (0x0001) +/* Comparison flag: Match the time if it is older than. */ +#define ARCHIVE_MATCH_OLDER (0x0002) +/* Comparison flag: Match the time if it is equal to. */ +#define ARCHIVE_MATCH_EQUAL (0x0010) +/* Set inclusion time. */ +__LA_DECL int archive_match_include_time(struct archive *, int _flag, + time_t _sec, long _nsec); +/* Set inclusion time by a date string. */ +__LA_DECL int archive_match_include_date(struct archive *, int _flag, + const char *_datestr); +__LA_DECL int archive_match_include_date_w(struct archive *, int _flag, + const wchar_t *_datestr); +/* Set inclusion time by a particular file. */ +__LA_DECL int archive_match_include_file_time(struct archive *, + int _flag, const char *_pathname); +__LA_DECL int archive_match_include_file_time_w(struct archive *, + int _flag, const wchar_t *_pathname); +/* Add exclusion entry. */ +__LA_DECL int archive_match_exclude_entry(struct archive *, + int _flag, struct archive_entry *); + +/* + * Test if a file is excluded by its uid ,gid, uname or gname. + * The conditions are set by following functions. + */ +__LA_DECL int archive_match_owner_excluded(struct archive *, + struct archive_entry *); +/* Add inclusion uid, gid, uname and gname. */ +__LA_DECL int archive_match_include_uid(struct archive *, la_int64_t); +__LA_DECL int archive_match_include_gid(struct archive *, la_int64_t); +__LA_DECL int archive_match_include_uname(struct archive *, const char *); +__LA_DECL int archive_match_include_uname_w(struct archive *, + const wchar_t *); +__LA_DECL int archive_match_include_gname(struct archive *, const char *); +__LA_DECL int archive_match_include_gname_w(struct archive *, + const wchar_t *); + +/* Utility functions */ +/* Convenience function to sort a NULL terminated list of strings */ +__LA_DECL int archive_utility_string_sort(char **); + +#ifdef __cplusplus +} +#endif + +/* These are meaningless outside of this header. */ +#undef __LA_DECL + +#endif /* !ARCHIVE_H_INCLUDED */ diff --git a/c/archive_acl.c b/c/archive_acl.c new file mode 100644 --- /dev/null +++ b/c/archive_acl.c @@ -0,0 +1,2097 @@ +/*- + * Copyright (c) 2003-2010 Tim Kientzle + * Copyright (c) 2016 Martin Matuska + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_LIMITS_H +#include +#endif +#ifdef HAVE_WCHAR_H +#include +#endif + +#include "archive_acl_private.h" +#include "archive_entry.h" +#include "archive_private.h" + +#undef max +#define max(a, b) ((a)>(b)?(a):(b)) + +#ifndef HAVE_WMEMCMP +/* Good enough for simple equality testing, but not for sorting. */ +#define wmemcmp(a,b,i) memcmp((a), (b), (i) * sizeof(wchar_t)) +#endif + +static int acl_special(struct archive_acl *acl, + int type, int permset, int tag); +static struct archive_acl_entry *acl_new_entry(struct archive_acl *acl, + int type, int permset, int tag, int id); +static int archive_acl_add_entry_len_l(struct archive_acl *acl, + int type, int permset, int tag, int id, const char *name, + size_t len, struct archive_string_conv *sc); +static int archive_acl_text_want_type(struct archive_acl *acl, int flags); +static ssize_t archive_acl_text_len(struct archive_acl *acl, int want_type, + int flags, int wide, struct archive *a, + struct archive_string_conv *sc); +static int isint_w(const wchar_t *start, const wchar_t *end, int *result); +static int ismode_w(const wchar_t *start, const wchar_t *end, int *result); +static int is_nfs4_flags_w(const wchar_t *start, const wchar_t *end, + int *result); +static int is_nfs4_perms_w(const wchar_t *start, const wchar_t *end, + int *result); +static void next_field_w(const wchar_t **wp, const wchar_t **start, + const wchar_t **end, wchar_t *sep); +static void append_entry_w(wchar_t **wp, const wchar_t *prefix, int type, + int tag, int flags, const wchar_t *wname, int perm, int id); +static void append_id_w(wchar_t **wp, int id); +static int isint(const char *start, const char *end, int *result); +static int ismode(const char *start, const char *end, int *result); +static int is_nfs4_flags(const char *start, const char *end, + int *result); +static int is_nfs4_perms(const char *start, const char *end, + int *result); +static void next_field(const char **p, const char **start, + const char **end, char *sep); +static void append_entry(char **p, const char *prefix, int type, + int tag, int flags, const char *name, int perm, int id); +static void append_id(char **p, int id); + +static const struct { + const int perm; + const char c; + const wchar_t wc; +} nfsv4_acl_perm_map[] = { + { ARCHIVE_ENTRY_ACL_READ_DATA | ARCHIVE_ENTRY_ACL_LIST_DIRECTORY, 'r', + L'r' }, + { ARCHIVE_ENTRY_ACL_WRITE_DATA | ARCHIVE_ENTRY_ACL_ADD_FILE, 'w', + L'w' }, + { ARCHIVE_ENTRY_ACL_EXECUTE, 'x', L'x' }, + { ARCHIVE_ENTRY_ACL_APPEND_DATA | ARCHIVE_ENTRY_ACL_ADD_SUBDIRECTORY, + 'p', L'p' }, + { ARCHIVE_ENTRY_ACL_DELETE, 'd', L'd' }, + { ARCHIVE_ENTRY_ACL_DELETE_CHILD, 'D', L'D' }, + { ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES, 'a', L'a' }, + { ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES, 'A', L'A' }, + { ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS, 'R', L'R' }, + { ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS, 'W', L'W' }, + { ARCHIVE_ENTRY_ACL_READ_ACL, 'c', L'c' }, + { ARCHIVE_ENTRY_ACL_WRITE_ACL, 'C', L'C' }, + { ARCHIVE_ENTRY_ACL_WRITE_OWNER, 'o', L'o' }, + { ARCHIVE_ENTRY_ACL_SYNCHRONIZE, 's', L's' } +}; + +static const int nfsv4_acl_perm_map_size = (int)(sizeof(nfsv4_acl_perm_map) / + sizeof(nfsv4_acl_perm_map[0])); + +static const struct { + const int perm; + const char c; + const wchar_t wc; +} nfsv4_acl_flag_map[] = { + { ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT, 'f', L'f' }, + { ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT, 'd', L'd' }, + { ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY, 'i', L'i' }, + { ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT, 'n', L'n' }, + { ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS, 'S', L'S' }, + { ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS, 'F', L'F' }, + { ARCHIVE_ENTRY_ACL_ENTRY_INHERITED, 'I', L'I' } +}; + +static const int nfsv4_acl_flag_map_size = (int)(sizeof(nfsv4_acl_flag_map) / + sizeof(nfsv4_acl_flag_map[0])); + +void +archive_acl_clear(struct archive_acl *acl) +{ + struct archive_acl_entry *ap; + + while (acl->acl_head != NULL) { + ap = acl->acl_head->next; + archive_mstring_clean(&acl->acl_head->name); + free(acl->acl_head); + acl->acl_head = ap; + } + free(acl->acl_text_w); + acl->acl_text_w = NULL; + free(acl->acl_text); + acl->acl_text = NULL; + acl->acl_p = NULL; + acl->acl_types = 0; + acl->acl_state = 0; /* Not counting. */ +} + +void +archive_acl_copy(struct archive_acl *dest, struct archive_acl *src) +{ + struct archive_acl_entry *ap, *ap2; + + archive_acl_clear(dest); + + dest->mode = src->mode; + ap = src->acl_head; + while (ap != NULL) { + ap2 = acl_new_entry(dest, + ap->type, ap->permset, ap->tag, ap->id); + if (ap2 != NULL) + archive_mstring_copy(&ap2->name, &ap->name); + ap = ap->next; + } +} + +int +archive_acl_add_entry(struct archive_acl *acl, + int type, int permset, int tag, int id, const char *name) +{ + struct archive_acl_entry *ap; + + if (acl_special(acl, type, permset, tag) == 0) + return ARCHIVE_OK; + ap = acl_new_entry(acl, type, permset, tag, id); + if (ap == NULL) { + /* XXX Error XXX */ + return ARCHIVE_FAILED; + } + if (name != NULL && *name != '\0') + archive_mstring_copy_mbs(&ap->name, name); + else + archive_mstring_clean(&ap->name); + return ARCHIVE_OK; +} + +int +archive_acl_add_entry_w_len(struct archive_acl *acl, + int type, int permset, int tag, int id, const wchar_t *name, size_t len) +{ + struct archive_acl_entry *ap; + + if (acl_special(acl, type, permset, tag) == 0) + return ARCHIVE_OK; + ap = acl_new_entry(acl, type, permset, tag, id); + if (ap == NULL) { + /* XXX Error XXX */ + return ARCHIVE_FAILED; + } + if (name != NULL && *name != L'\0' && len > 0) + archive_mstring_copy_wcs_len(&ap->name, name, len); + else + archive_mstring_clean(&ap->name); + return ARCHIVE_OK; +} + +static int +archive_acl_add_entry_len_l(struct archive_acl *acl, + int type, int permset, int tag, int id, const char *name, size_t len, + struct archive_string_conv *sc) +{ + struct archive_acl_entry *ap; + int r; + + if (acl_special(acl, type, permset, tag) == 0) + return ARCHIVE_OK; + ap = acl_new_entry(acl, type, permset, tag, id); + if (ap == NULL) { + /* XXX Error XXX */ + return ARCHIVE_FAILED; + } + if (name != NULL && *name != '\0' && len > 0) { + r = archive_mstring_copy_mbs_len_l(&ap->name, name, len, sc); + } else { + r = 0; + archive_mstring_clean(&ap->name); + } + if (r == 0) + return (ARCHIVE_OK); + else if (errno == ENOMEM) + return (ARCHIVE_FATAL); + else + return (ARCHIVE_WARN); +} + +/* + * If this ACL entry is part of the standard POSIX permissions set, + * store the permissions in the stat structure and return zero. + */ +static int +acl_special(struct archive_acl *acl, int type, int permset, int tag) +{ + if (type == ARCHIVE_ENTRY_ACL_TYPE_ACCESS + && ((permset & ~007) == 0)) { + switch (tag) { + case ARCHIVE_ENTRY_ACL_USER_OBJ: + acl->mode &= ~0700; + acl->mode |= (permset & 7) << 6; + return (0); + case ARCHIVE_ENTRY_ACL_GROUP_OBJ: + acl->mode &= ~0070; + acl->mode |= (permset & 7) << 3; + return (0); + case ARCHIVE_ENTRY_ACL_OTHER: + acl->mode &= ~0007; + acl->mode |= permset & 7; + return (0); + } + } + return (1); +} + +/* + * Allocate and populate a new ACL entry with everything but the + * name. + */ +static struct archive_acl_entry * +acl_new_entry(struct archive_acl *acl, + int type, int permset, int tag, int id) +{ + struct archive_acl_entry *ap, *aq; + + /* Type argument must be a valid NFS4 or POSIX.1e type. + * The type must agree with anything already set and + * the permset must be compatible. */ + if (type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) { + if (acl->acl_types & ~ARCHIVE_ENTRY_ACL_TYPE_NFS4) { + return (NULL); + } + if (permset & + ~(ARCHIVE_ENTRY_ACL_PERMS_NFS4 + | ARCHIVE_ENTRY_ACL_INHERITANCE_NFS4)) { + return (NULL); + } + } else if (type & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) { + if (acl->acl_types & ~ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) { + return (NULL); + } + if (permset & ~ARCHIVE_ENTRY_ACL_PERMS_POSIX1E) { + return (NULL); + } + } else { + return (NULL); + } + + /* Verify the tag is valid and compatible with NFS4 or POSIX.1e. */ + switch (tag) { + case ARCHIVE_ENTRY_ACL_USER: + case ARCHIVE_ENTRY_ACL_USER_OBJ: + case ARCHIVE_ENTRY_ACL_GROUP: + case ARCHIVE_ENTRY_ACL_GROUP_OBJ: + /* Tags valid in both NFS4 and POSIX.1e */ + break; + case ARCHIVE_ENTRY_ACL_MASK: + case ARCHIVE_ENTRY_ACL_OTHER: + /* Tags valid only in POSIX.1e. */ + if (type & ~ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) { + return (NULL); + } + break; + case ARCHIVE_ENTRY_ACL_EVERYONE: + /* Tags valid only in NFS4. */ + if (type & ~ARCHIVE_ENTRY_ACL_TYPE_NFS4) { + return (NULL); + } + break; + default: + /* No other values are valid. */ + return (NULL); + } + + free(acl->acl_text_w); + acl->acl_text_w = NULL; + free(acl->acl_text); + acl->acl_text = NULL; + + /* + * If there's a matching entry already in the list, overwrite it. + * NFSv4 entries may be repeated and are not overwritten. + * + * TODO: compare names of no id is provided (needs more rework) + */ + ap = acl->acl_head; + aq = NULL; + while (ap != NULL) { + if (((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) == 0) && + ap->type == type && ap->tag == tag && ap->id == id) { + if (id != -1 || (tag != ARCHIVE_ENTRY_ACL_USER && + tag != ARCHIVE_ENTRY_ACL_GROUP)) { + ap->permset = permset; + return (ap); + } + } + aq = ap; + ap = ap->next; + } + + /* Add a new entry to the end of the list. */ + ap = (struct archive_acl_entry *)calloc(1, sizeof(*ap)); + if (ap == NULL) + return (NULL); + if (aq == NULL) + acl->acl_head = ap; + else + aq->next = ap; + ap->type = type; + ap->tag = tag; + ap->id = id; + ap->permset = permset; + acl->acl_types |= type; + return (ap); +} + +/* + * Return a count of entries matching "want_type". + */ +int +archive_acl_count(struct archive_acl *acl, int want_type) +{ + int count; + struct archive_acl_entry *ap; + + count = 0; + ap = acl->acl_head; + while (ap != NULL) { + if ((ap->type & want_type) != 0) + count++; + ap = ap->next; + } + + if (count > 0 && ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0)) + count += 3; + return (count); +} + +/* + * Return a bitmask of stored ACL types in an ACL list + */ +int +archive_acl_types(struct archive_acl *acl) +{ + return (acl->acl_types); +} + +/* + * Prepare for reading entries from the ACL data. Returns a count + * of entries matching "want_type", or zero if there are no + * non-extended ACL entries of that type. + */ +int +archive_acl_reset(struct archive_acl *acl, int want_type) +{ + int count, cutoff; + + count = archive_acl_count(acl, want_type); + + /* + * If the only entries are the three standard ones, + * then don't return any ACL data. (In this case, + * client can just use chmod(2) to set permissions.) + */ + if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) + cutoff = 3; + else + cutoff = 0; + + if (count > cutoff) + acl->acl_state = ARCHIVE_ENTRY_ACL_USER_OBJ; + else + acl->acl_state = 0; + acl->acl_p = acl->acl_head; + return (count); +} + + +/* + * Return the next ACL entry in the list. Fake entries for the + * standard permissions and include them in the returned list. + */ +int +archive_acl_next(struct archive *a, struct archive_acl *acl, int want_type, + int *type, int *permset, int *tag, int *id, const char **name) +{ + *name = NULL; + *id = -1; + + /* + * The acl_state is either zero (no entries available), -1 + * (reading from list), or an entry type (retrieve that type + * from ae_stat.aest_mode). + */ + if (acl->acl_state == 0) + return (ARCHIVE_WARN); + + /* The first three access entries are special. */ + if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) { + switch (acl->acl_state) { + case ARCHIVE_ENTRY_ACL_USER_OBJ: + *permset = (acl->mode >> 6) & 7; + *type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS; + *tag = ARCHIVE_ENTRY_ACL_USER_OBJ; + acl->acl_state = ARCHIVE_ENTRY_ACL_GROUP_OBJ; + return (ARCHIVE_OK); + case ARCHIVE_ENTRY_ACL_GROUP_OBJ: + *permset = (acl->mode >> 3) & 7; + *type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS; + *tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; + acl->acl_state = ARCHIVE_ENTRY_ACL_OTHER; + return (ARCHIVE_OK); + case ARCHIVE_ENTRY_ACL_OTHER: + *permset = acl->mode & 7; + *type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS; + *tag = ARCHIVE_ENTRY_ACL_OTHER; + acl->acl_state = -1; + acl->acl_p = acl->acl_head; + return (ARCHIVE_OK); + default: + break; + } + } + + while (acl->acl_p != NULL && (acl->acl_p->type & want_type) == 0) + acl->acl_p = acl->acl_p->next; + if (acl->acl_p == NULL) { + acl->acl_state = 0; + *type = 0; + *permset = 0; + *tag = 0; + *id = -1; + *name = NULL; + return (ARCHIVE_EOF); /* End of ACL entries. */ + } + *type = acl->acl_p->type; + *permset = acl->acl_p->permset; + *tag = acl->acl_p->tag; + *id = acl->acl_p->id; + if (archive_mstring_get_mbs(a, &acl->acl_p->name, name) != 0) { + if (errno == ENOMEM) + return (ARCHIVE_FATAL); + *name = NULL; + } + acl->acl_p = acl->acl_p->next; + return (ARCHIVE_OK); +} + +/* + * Determine what type of ACL do we want + */ +static int +archive_acl_text_want_type(struct archive_acl *acl, int flags) +{ + int want_type; + + /* Check if ACL is NFSv4 */ + if ((acl->acl_types & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { + /* NFSv4 should never mix with POSIX.1e */ + if ((acl->acl_types & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) + return (0); + else + return (ARCHIVE_ENTRY_ACL_TYPE_NFS4); + } + + /* Now deal with POSIX.1e ACLs */ + + want_type = 0; + if ((flags & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) + want_type |= ARCHIVE_ENTRY_ACL_TYPE_ACCESS; + if ((flags & ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) != 0) + want_type |= ARCHIVE_ENTRY_ACL_TYPE_DEFAULT; + + /* By default we want both access and default ACLs */ + if (want_type == 0) + return (ARCHIVE_ENTRY_ACL_TYPE_POSIX1E); + + return (want_type); +} + +/* + * Calculate ACL text string length + */ +static ssize_t +archive_acl_text_len(struct archive_acl *acl, int want_type, int flags, + int wide, struct archive *a, struct archive_string_conv *sc) { + struct archive_acl_entry *ap; + const char *name; + const wchar_t *wname; + int count, idlen, tmp, r; + ssize_t length; + size_t len; + + count = 0; + length = 0; + for (ap = acl->acl_head; ap != NULL; ap = ap->next) { + if ((ap->type & want_type) == 0) + continue; + /* + * Filemode-mapping ACL entries are stored exclusively in + * ap->mode so they should not be in the list + */ + if ((ap->type == ARCHIVE_ENTRY_ACL_TYPE_ACCESS) + && (ap->tag == ARCHIVE_ENTRY_ACL_USER_OBJ + || ap->tag == ARCHIVE_ENTRY_ACL_GROUP_OBJ + || ap->tag == ARCHIVE_ENTRY_ACL_OTHER)) + continue; + count++; + if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) != 0 + && (ap->type & ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) != 0) + length += 8; /* "default:" */ + switch (ap->tag) { + case ARCHIVE_ENTRY_ACL_USER_OBJ: + if (want_type == ARCHIVE_ENTRY_ACL_TYPE_NFS4) { + length += 6; /* "owner@" */ + break; + } + /* FALLTHROUGH */ + case ARCHIVE_ENTRY_ACL_USER: + case ARCHIVE_ENTRY_ACL_MASK: + length += 4; /* "user", "mask" */ + break; + case ARCHIVE_ENTRY_ACL_GROUP_OBJ: + if (want_type == ARCHIVE_ENTRY_ACL_TYPE_NFS4) { + length += 6; /* "group@" */ + break; + } + /* FALLTHROUGH */ + case ARCHIVE_ENTRY_ACL_GROUP: + case ARCHIVE_ENTRY_ACL_OTHER: + length += 5; /* "group", "other" */ + break; + case ARCHIVE_ENTRY_ACL_EVERYONE: + length += 9; /* "everyone@" */ + break; + } + length += 1; /* colon after tag */ + if (ap->tag == ARCHIVE_ENTRY_ACL_USER || + ap->tag == ARCHIVE_ENTRY_ACL_GROUP) { + if (wide) { + r = archive_mstring_get_wcs(a, &ap->name, + &wname); + if (r == 0 && wname != NULL) + length += wcslen(wname); + else if (r < 0 && errno == ENOMEM) + return (0); + else + length += sizeof(uid_t) * 3 + 1; + } else { + r = archive_mstring_get_mbs_l(&ap->name, &name, + &len, sc); + if (r != 0) + return (0); + if (len > 0 && name != NULL) + length += len; + else + length += sizeof(uid_t) * 3 + 1; + } + length += 1; /* colon after user or group name */ + } else if (want_type != ARCHIVE_ENTRY_ACL_TYPE_NFS4) + length += 1; /* 2nd colon empty user,group or other */ + + if (((flags & ARCHIVE_ENTRY_ACL_STYLE_SOLARIS) != 0) + && ((want_type & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) + && (ap->tag == ARCHIVE_ENTRY_ACL_OTHER + || ap->tag == ARCHIVE_ENTRY_ACL_MASK)) { + /* Solaris has no colon after other: and mask: */ + length = length - 1; + } + + if (want_type == ARCHIVE_ENTRY_ACL_TYPE_NFS4) { + /* rwxpdDaARWcCos:fdinSFI:deny */ + length += 27; + if ((ap->type & ARCHIVE_ENTRY_ACL_TYPE_DENY) == 0) + length += 1; /* allow, alarm, audit */ + } else + length += 3; /* rwx */ + + if ((ap->tag == ARCHIVE_ENTRY_ACL_USER || + ap->tag == ARCHIVE_ENTRY_ACL_GROUP) && + (flags & ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID) != 0) { + length += 1; /* colon */ + /* ID digit count */ + idlen = 1; + tmp = ap->id; + while (tmp > 9) { + tmp = tmp / 10; + idlen++; + } + length += idlen; + } + length ++; /* entry separator */ + } + + /* Add filemode-mapping access entries to the length */ + if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) { + if ((flags & ARCHIVE_ENTRY_ACL_STYLE_SOLARIS) != 0) { + /* "user::rwx\ngroup::rwx\nother:rwx\n" */ + length += 31; + } else { + /* "user::rwx\ngroup::rwx\nother::rwx\n" */ + length += 32; + } + } else if (count == 0) + return (0); + + /* The terminating character is included in count */ + return (length); +} + +/* + * Generate a wide text version of the ACL. The flags parameter controls + * the type and style of the generated ACL. + */ +wchar_t * +archive_acl_to_text_w(struct archive_acl *acl, ssize_t *text_len, int flags, + struct archive *a) +{ + int count; + ssize_t length; + size_t len; + const wchar_t *wname; + const wchar_t *prefix; + wchar_t separator; + struct archive_acl_entry *ap; + int id, r, want_type; + wchar_t *wp, *ws; + + want_type = archive_acl_text_want_type(acl, flags); + + /* Both NFSv4 and POSIX.1 types found */ + if (want_type == 0) + return (NULL); + + if (want_type == ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) + flags |= ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT; + + length = archive_acl_text_len(acl, want_type, flags, 1, a, NULL); + + if (length == 0) + return (NULL); + + if (flags & ARCHIVE_ENTRY_ACL_STYLE_SEPARATOR_COMMA) + separator = L','; + else + separator = L'\n'; + + /* Now, allocate the string and actually populate it. */ + wp = ws = (wchar_t *)malloc(length * sizeof(wchar_t)); + if (wp == NULL) { + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (NULL); + } + count = 0; + + if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) { + append_entry_w(&wp, NULL, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, + ARCHIVE_ENTRY_ACL_USER_OBJ, flags, NULL, + acl->mode & 0700, -1); + *wp++ = separator; + append_entry_w(&wp, NULL, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, + ARCHIVE_ENTRY_ACL_GROUP_OBJ, flags, NULL, + acl->mode & 0070, -1); + *wp++ = separator; + append_entry_w(&wp, NULL, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, + ARCHIVE_ENTRY_ACL_OTHER, flags, NULL, + acl->mode & 0007, -1); + count += 3; + } + + for (ap = acl->acl_head; ap != NULL; ap = ap->next) { + if ((ap->type & want_type) == 0) + continue; + /* + * Filemode-mapping ACL entries are stored exclusively in + * ap->mode so they should not be in the list + */ + if ((ap->type == ARCHIVE_ENTRY_ACL_TYPE_ACCESS) + && (ap->tag == ARCHIVE_ENTRY_ACL_USER_OBJ + || ap->tag == ARCHIVE_ENTRY_ACL_GROUP_OBJ + || ap->tag == ARCHIVE_ENTRY_ACL_OTHER)) + continue; + if (ap->type == ARCHIVE_ENTRY_ACL_TYPE_DEFAULT && + (flags & ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT) != 0) + prefix = L"default:"; + else + prefix = NULL; + r = archive_mstring_get_wcs(a, &ap->name, &wname); + if (r == 0) { + if (count > 0) + *wp++ = separator; + if (flags & ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID) + id = ap->id; + else + id = -1; + append_entry_w(&wp, prefix, ap->type, ap->tag, flags, + wname, ap->permset, id); + count++; + } else if (r < 0 && errno == ENOMEM) { + free(ws); + return (NULL); + } + } + + /* Add terminating character */ + *wp++ = L'\0'; + + len = wcslen(ws); + + if ((ssize_t)len > (length - 1)) + __archive_errx(1, "Buffer overrun"); + + if (text_len != NULL) + *text_len = len; + + return (ws); +} + +static void +append_id_w(wchar_t **wp, int id) +{ + if (id < 0) + id = 0; + if (id > 9) + append_id_w(wp, id / 10); + *(*wp)++ = L"0123456789"[id % 10]; +} + +static void +append_entry_w(wchar_t **wp, const wchar_t *prefix, int type, + int tag, int flags, const wchar_t *wname, int perm, int id) +{ + int i; + + if (prefix != NULL) { + wcscpy(*wp, prefix); + *wp += wcslen(*wp); + } + switch (tag) { + case ARCHIVE_ENTRY_ACL_USER_OBJ: + wname = NULL; + id = -1; + if ((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { + wcscpy(*wp, L"owner@"); + break; + } + /* FALLTHROUGH */ + case ARCHIVE_ENTRY_ACL_USER: + wcscpy(*wp, L"user"); + break; + case ARCHIVE_ENTRY_ACL_GROUP_OBJ: + wname = NULL; + id = -1; + if ((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { + wcscpy(*wp, L"group@"); + break; + } + /* FALLTHROUGH */ + case ARCHIVE_ENTRY_ACL_GROUP: + wcscpy(*wp, L"group"); + break; + case ARCHIVE_ENTRY_ACL_MASK: + wcscpy(*wp, L"mask"); + wname = NULL; + id = -1; + break; + case ARCHIVE_ENTRY_ACL_OTHER: + wcscpy(*wp, L"other"); + wname = NULL; + id = -1; + break; + case ARCHIVE_ENTRY_ACL_EVERYONE: + wcscpy(*wp, L"everyone@"); + wname = NULL; + id = -1; + break; + } + *wp += wcslen(*wp); + *(*wp)++ = L':'; + if (((type & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) || + tag == ARCHIVE_ENTRY_ACL_USER || + tag == ARCHIVE_ENTRY_ACL_GROUP) { + if (wname != NULL) { + wcscpy(*wp, wname); + *wp += wcslen(*wp); + } else if (tag == ARCHIVE_ENTRY_ACL_USER + || tag == ARCHIVE_ENTRY_ACL_GROUP) { + append_id_w(wp, id); + if ((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) == 0) + id = -1; + } + /* Solaris style has no second colon after other and mask */ + if (((flags & ARCHIVE_ENTRY_ACL_STYLE_SOLARIS) == 0) + || (tag != ARCHIVE_ENTRY_ACL_OTHER + && tag != ARCHIVE_ENTRY_ACL_MASK)) + *(*wp)++ = L':'; + } + if ((type & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) { + /* POSIX.1e ACL perms */ + *(*wp)++ = (perm & 0444) ? L'r' : L'-'; + *(*wp)++ = (perm & 0222) ? L'w' : L'-'; + *(*wp)++ = (perm & 0111) ? L'x' : L'-'; + } else { + /* NFSv4 ACL perms */ + for (i = 0; i < nfsv4_acl_perm_map_size; i++) { + if (perm & nfsv4_acl_perm_map[i].perm) + *(*wp)++ = nfsv4_acl_perm_map[i].wc; + else if ((flags & ARCHIVE_ENTRY_ACL_STYLE_COMPACT) == 0) + *(*wp)++ = L'-'; + } + *(*wp)++ = L':'; + for (i = 0; i < nfsv4_acl_flag_map_size; i++) { + if (perm & nfsv4_acl_flag_map[i].perm) + *(*wp)++ = nfsv4_acl_flag_map[i].wc; + else if ((flags & ARCHIVE_ENTRY_ACL_STYLE_COMPACT) == 0) + *(*wp)++ = L'-'; + } + *(*wp)++ = L':'; + switch (type) { + case ARCHIVE_ENTRY_ACL_TYPE_ALLOW: + wcscpy(*wp, L"allow"); + break; + case ARCHIVE_ENTRY_ACL_TYPE_DENY: + wcscpy(*wp, L"deny"); + break; + case ARCHIVE_ENTRY_ACL_TYPE_AUDIT: + wcscpy(*wp, L"audit"); + break; + case ARCHIVE_ENTRY_ACL_TYPE_ALARM: + wcscpy(*wp, L"alarm"); + break; + default: + break; + } + *wp += wcslen(*wp); + } + if (id != -1) { + *(*wp)++ = L':'; + append_id_w(wp, id); + } +} + +/* + * Generate a text version of the ACL. The flags parameter controls + * the type and style of the generated ACL. + */ +char * +archive_acl_to_text_l(struct archive_acl *acl, ssize_t *text_len, int flags, + struct archive_string_conv *sc) +{ + int count; + ssize_t length; + size_t len; + const char *name; + const char *prefix; + char separator; + struct archive_acl_entry *ap; + int id, r, want_type; + char *p, *s; + + want_type = archive_acl_text_want_type(acl, flags); + + /* Both NFSv4 and POSIX.1 types found */ + if (want_type == 0) + return (NULL); + + if (want_type == ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) + flags |= ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT; + + length = archive_acl_text_len(acl, want_type, flags, 0, NULL, sc); + + if (length == 0) + return (NULL); + + if (flags & ARCHIVE_ENTRY_ACL_STYLE_SEPARATOR_COMMA) + separator = ','; + else + separator = '\n'; + + /* Now, allocate the string and actually populate it. */ + p = s = (char *)malloc(length * sizeof(char)); + if (p == NULL) { + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (NULL); + } + count = 0; + + if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) { + append_entry(&p, NULL, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, + ARCHIVE_ENTRY_ACL_USER_OBJ, flags, NULL, + acl->mode & 0700, -1); + *p++ = separator; + append_entry(&p, NULL, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, + ARCHIVE_ENTRY_ACL_GROUP_OBJ, flags, NULL, + acl->mode & 0070, -1); + *p++ = separator; + append_entry(&p, NULL, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, + ARCHIVE_ENTRY_ACL_OTHER, flags, NULL, + acl->mode & 0007, -1); + count += 3; + } + + for (ap = acl->acl_head; ap != NULL; ap = ap->next) { + if ((ap->type & want_type) == 0) + continue; + /* + * Filemode-mapping ACL entries are stored exclusively in + * ap->mode so they should not be in the list + */ + if ((ap->type == ARCHIVE_ENTRY_ACL_TYPE_ACCESS) + && (ap->tag == ARCHIVE_ENTRY_ACL_USER_OBJ + || ap->tag == ARCHIVE_ENTRY_ACL_GROUP_OBJ + || ap->tag == ARCHIVE_ENTRY_ACL_OTHER)) + continue; + if (ap->type == ARCHIVE_ENTRY_ACL_TYPE_DEFAULT && + (flags & ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT) != 0) + prefix = "default:"; + else + prefix = NULL; + r = archive_mstring_get_mbs_l( + &ap->name, &name, &len, sc); + if (r != 0) { + free(s); + return (NULL); + } + if (count > 0) + *p++ = separator; + if (name == NULL || + (flags & ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID)) { + id = ap->id; + } else { + id = -1; + } + append_entry(&p, prefix, ap->type, ap->tag, flags, name, + ap->permset, id); + count++; + } + + /* Add terminating character */ + *p++ = '\0'; + + len = strlen(s); + + if ((ssize_t)len > (length - 1)) + __archive_errx(1, "Buffer overrun"); + + if (text_len != NULL) + *text_len = len; + + return (s); +} + +static void +append_id(char **p, int id) +{ + if (id < 0) + id = 0; + if (id > 9) + append_id(p, id / 10); + *(*p)++ = "0123456789"[id % 10]; +} + +static void +append_entry(char **p, const char *prefix, int type, + int tag, int flags, const char *name, int perm, int id) +{ + int i; + + if (prefix != NULL) { + strcpy(*p, prefix); + *p += strlen(*p); + } + switch (tag) { + case ARCHIVE_ENTRY_ACL_USER_OBJ: + name = NULL; + id = -1; + if ((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { + strcpy(*p, "owner@"); + break; + } + /* FALLTHROUGH */ + case ARCHIVE_ENTRY_ACL_USER: + strcpy(*p, "user"); + break; + case ARCHIVE_ENTRY_ACL_GROUP_OBJ: + name = NULL; + id = -1; + if ((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { + strcpy(*p, "group@"); + break; + } + /* FALLTHROUGH */ + case ARCHIVE_ENTRY_ACL_GROUP: + strcpy(*p, "group"); + break; + case ARCHIVE_ENTRY_ACL_MASK: + strcpy(*p, "mask"); + name = NULL; + id = -1; + break; + case ARCHIVE_ENTRY_ACL_OTHER: + strcpy(*p, "other"); + name = NULL; + id = -1; + break; + case ARCHIVE_ENTRY_ACL_EVERYONE: + strcpy(*p, "everyone@"); + name = NULL; + id = -1; + break; + } + *p += strlen(*p); + *(*p)++ = ':'; + if (((type & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) || + tag == ARCHIVE_ENTRY_ACL_USER || + tag == ARCHIVE_ENTRY_ACL_GROUP) { + if (name != NULL) { + strcpy(*p, name); + *p += strlen(*p); + } else if (tag == ARCHIVE_ENTRY_ACL_USER + || tag == ARCHIVE_ENTRY_ACL_GROUP) { + append_id(p, id); + if ((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) == 0) + id = -1; + } + /* Solaris style has no second colon after other and mask */ + if (((flags & ARCHIVE_ENTRY_ACL_STYLE_SOLARIS) == 0) + || (tag != ARCHIVE_ENTRY_ACL_OTHER + && tag != ARCHIVE_ENTRY_ACL_MASK)) + *(*p)++ = ':'; + } + if ((type & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) { + /* POSIX.1e ACL perms */ + *(*p)++ = (perm & 0444) ? 'r' : '-'; + *(*p)++ = (perm & 0222) ? 'w' : '-'; + *(*p)++ = (perm & 0111) ? 'x' : '-'; + } else { + /* NFSv4 ACL perms */ + for (i = 0; i < nfsv4_acl_perm_map_size; i++) { + if (perm & nfsv4_acl_perm_map[i].perm) + *(*p)++ = nfsv4_acl_perm_map[i].c; + else if ((flags & ARCHIVE_ENTRY_ACL_STYLE_COMPACT) == 0) + *(*p)++ = '-'; + } + *(*p)++ = ':'; + for (i = 0; i < nfsv4_acl_flag_map_size; i++) { + if (perm & nfsv4_acl_flag_map[i].perm) + *(*p)++ = nfsv4_acl_flag_map[i].c; + else if ((flags & ARCHIVE_ENTRY_ACL_STYLE_COMPACT) == 0) + *(*p)++ = '-'; + } + *(*p)++ = ':'; + switch (type) { + case ARCHIVE_ENTRY_ACL_TYPE_ALLOW: + strcpy(*p, "allow"); + break; + case ARCHIVE_ENTRY_ACL_TYPE_DENY: + strcpy(*p, "deny"); + break; + case ARCHIVE_ENTRY_ACL_TYPE_AUDIT: + strcpy(*p, "audit"); + break; + case ARCHIVE_ENTRY_ACL_TYPE_ALARM: + strcpy(*p, "alarm"); + break; + } + *p += strlen(*p); + } + if (id != -1) { + *(*p)++ = ':'; + append_id(p, id); + } +} + +/* + * Parse a wide ACL text string. + * + * The want_type argument may be one of the following: + * ARCHIVE_ENTRY_ACL_TYPE_ACCESS - text is a POSIX.1e ACL of type ACCESS + * ARCHIVE_ENTRY_ACL_TYPE_DEFAULT - text is a POSIX.1e ACL of type DEFAULT + * ARCHIVE_ENTRY_ACL_TYPE_NFS4 - text is as a NFSv4 ACL + * + * POSIX.1e ACL entries prefixed with "default:" are treated as + * ARCHIVE_ENTRY_ACL_TYPE_DEFAULT unless type is ARCHIVE_ENTRY_ACL_TYPE_NFS4 + */ +int +archive_acl_from_text_w(struct archive_acl *acl, const wchar_t *text, + int want_type) +{ + struct { + const wchar_t *start; + const wchar_t *end; + } field[6], name; + + const wchar_t *s, *st; + + int numfields, fields, n, r, sol, ret; + int type, types, tag, permset, id; + size_t len; + wchar_t sep; + + ret = ARCHIVE_OK; + types = 0; + + switch (want_type) { + case ARCHIVE_ENTRY_ACL_TYPE_POSIX1E: + want_type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS; + __LA_FALLTHROUGH; + case ARCHIVE_ENTRY_ACL_TYPE_ACCESS: + case ARCHIVE_ENTRY_ACL_TYPE_DEFAULT: + numfields = 5; + break; + case ARCHIVE_ENTRY_ACL_TYPE_NFS4: + numfields = 6; + break; + default: + return (ARCHIVE_FATAL); + } + + while (text != NULL && *text != L'\0') { + /* + * Parse the fields out of the next entry, + * advance 'text' to start of next entry. + */ + fields = 0; + do { + const wchar_t *start, *end; + next_field_w(&text, &start, &end, &sep); + if (fields < numfields) { + field[fields].start = start; + field[fields].end = end; + } + ++fields; + } while (sep == L':'); + + /* Set remaining fields to blank. */ + for (n = fields; n < numfields; ++n) + field[n].start = field[n].end = NULL; + + if (field[0].start != NULL && *(field[0].start) == L'#') { + /* Comment, skip entry */ + continue; + } + + n = 0; + sol = 0; + id = -1; + permset = 0; + name.start = name.end = NULL; + + if (want_type != ARCHIVE_ENTRY_ACL_TYPE_NFS4) { + /* POSIX.1e ACLs */ + /* + * Default keyword "default:user::rwx" + * if found, we have one more field + * + * We also support old Solaris extension: + * "defaultuser::rwx" is the default ACL corresponding + * to "user::rwx", etc. valid only for first field + */ + s = field[0].start; + len = field[0].end - field[0].start; + if (*s == L'd' && (len == 1 || (len >= 7 + && wmemcmp((s + 1), L"efault", 6) == 0))) { + type = ARCHIVE_ENTRY_ACL_TYPE_DEFAULT; + if (len > 7) + field[0].start += 7; + else + n = 1; + } else + type = want_type; + + /* Check for a numeric ID in field n+1 or n+3. */ + isint_w(field[n + 1].start, field[n + 1].end, &id); + /* Field n+3 is optional. */ + if (id == -1 && fields > n+3) + isint_w(field[n + 3].start, field[n + 3].end, + &id); + + tag = 0; + s = field[n].start; + st = field[n].start + 1; + len = field[n].end - field[n].start; + + switch (*s) { + case L'u': + if (len == 1 || (len == 4 + && wmemcmp(st, L"ser", 3) == 0)) + tag = ARCHIVE_ENTRY_ACL_USER_OBJ; + break; + case L'g': + if (len == 1 || (len == 5 + && wmemcmp(st, L"roup", 4) == 0)) + tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; + break; + case L'o': + if (len == 1 || (len == 5 + && wmemcmp(st, L"ther", 4) == 0)) + tag = ARCHIVE_ENTRY_ACL_OTHER; + break; + case L'm': + if (len == 1 || (len == 4 + && wmemcmp(st, L"ask", 3) == 0)) + tag = ARCHIVE_ENTRY_ACL_MASK; + break; + default: + break; + } + + switch (tag) { + case ARCHIVE_ENTRY_ACL_OTHER: + case ARCHIVE_ENTRY_ACL_MASK: + if (fields == (n + 2) + && field[n + 1].start < field[n + 1].end + && ismode_w(field[n + 1].start, + field[n + 1].end, &permset)) { + /* This is Solaris-style "other:rwx" */ + sol = 1; + } else if (fields == (n + 3) && + field[n + 1].start < field[n + 1].end) { + /* Invalid mask or other field */ + ret = ARCHIVE_WARN; + continue; + } + break; + case ARCHIVE_ENTRY_ACL_USER_OBJ: + case ARCHIVE_ENTRY_ACL_GROUP_OBJ: + if (id != -1 || + field[n + 1].start < field[n + 1].end) { + name = field[n + 1]; + if (tag == ARCHIVE_ENTRY_ACL_USER_OBJ) + tag = ARCHIVE_ENTRY_ACL_USER; + else + tag = ARCHIVE_ENTRY_ACL_GROUP; + } + break; + default: + /* Invalid tag, skip entry */ + ret = ARCHIVE_WARN; + continue; + } + + /* + * Without "default:" we expect mode in field 2 + * Exception: Solaris other and mask fields + */ + if (permset == 0 && !ismode_w(field[n + 2 - sol].start, + field[n + 2 - sol].end, &permset)) { + /* Invalid mode, skip entry */ + ret = ARCHIVE_WARN; + continue; + } + } else { + /* NFS4 ACLs */ + s = field[0].start; + len = field[0].end - field[0].start; + tag = 0; + + switch (len) { + case 4: + if (wmemcmp(s, L"user", 4) == 0) + tag = ARCHIVE_ENTRY_ACL_USER; + break; + case 5: + if (wmemcmp(s, L"group", 5) == 0) + tag = ARCHIVE_ENTRY_ACL_GROUP; + break; + case 6: + if (wmemcmp(s, L"owner@", 6) == 0) + tag = ARCHIVE_ENTRY_ACL_USER_OBJ; + else if (wmemcmp(s, L"group@", len) == 0) + tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; + break; + case 9: + if (wmemcmp(s, L"everyone@", 9) == 0) + tag = ARCHIVE_ENTRY_ACL_EVERYONE; + default: + break; + } + + if (tag == 0) { + /* Invalid tag, skip entry */ + ret = ARCHIVE_WARN; + continue; + } else if (tag == ARCHIVE_ENTRY_ACL_USER || + tag == ARCHIVE_ENTRY_ACL_GROUP) { + n = 1; + name = field[1]; + isint_w(name.start, name.end, &id); + } else + n = 0; + + if (!is_nfs4_perms_w(field[1 + n].start, + field[1 + n].end, &permset)) { + /* Invalid NFSv4 perms, skip entry */ + ret = ARCHIVE_WARN; + continue; + } + if (!is_nfs4_flags_w(field[2 + n].start, + field[2 + n].end, &permset)) { + /* Invalid NFSv4 flags, skip entry */ + ret = ARCHIVE_WARN; + continue; + } + s = field[3 + n].start; + len = field[3 + n].end - field[3 + n].start; + type = 0; + if (len == 4) { + if (wmemcmp(s, L"deny", 4) == 0) + type = ARCHIVE_ENTRY_ACL_TYPE_DENY; + } else if (len == 5) { + if (wmemcmp(s, L"allow", 5) == 0) + type = ARCHIVE_ENTRY_ACL_TYPE_ALLOW; + else if (wmemcmp(s, L"audit", 5) == 0) + type = ARCHIVE_ENTRY_ACL_TYPE_AUDIT; + else if (wmemcmp(s, L"alarm", 5) == 0) + type = ARCHIVE_ENTRY_ACL_TYPE_ALARM; + } + if (type == 0) { + /* Invalid entry type, skip entry */ + ret = ARCHIVE_WARN; + continue; + } + isint_w(field[4 + n].start, field[4 + n].end, &id); + } + + /* Add entry to the internal list. */ + r = archive_acl_add_entry_w_len(acl, type, permset, + tag, id, name.start, name.end - name.start); + if (r < ARCHIVE_WARN) + return (r); + if (r != ARCHIVE_OK) + ret = ARCHIVE_WARN; + types |= type; + } + + /* Reset ACL */ + archive_acl_reset(acl, types); + + return (ret); +} + +/* + * Parse a string to a positive decimal integer. Returns true if + * the string is non-empty and consists only of decimal digits, + * false otherwise. + */ +static int +isint_w(const wchar_t *start, const wchar_t *end, int *result) +{ + int n = 0; + if (start >= end) + return (0); + while (start < end) { + if (*start < '0' || *start > '9') + return (0); + if (n > (INT_MAX / 10) || + (n == INT_MAX / 10 && (*start - '0') > INT_MAX % 10)) { + n = INT_MAX; + } else { + n *= 10; + n += *start - '0'; + } + start++; + } + *result = n; + return (1); +} + +/* + * Parse a string as a mode field. Returns true if + * the string is non-empty and consists only of mode characters, + * false otherwise. + */ +static int +ismode_w(const wchar_t *start, const wchar_t *end, int *permset) +{ + const wchar_t *p; + + if (start >= end) + return (0); + p = start; + *permset = 0; + while (p < end) { + switch (*p++) { + case L'r': case L'R': + *permset |= ARCHIVE_ENTRY_ACL_READ; + break; + case L'w': case L'W': + *permset |= ARCHIVE_ENTRY_ACL_WRITE; + break; + case L'x': case L'X': + *permset |= ARCHIVE_ENTRY_ACL_EXECUTE; + break; + case L'-': + break; + default: + return (0); + } + } + return (1); +} + +/* + * Parse a string as a NFS4 ACL permission field. + * Returns true if the string is non-empty and consists only of NFS4 ACL + * permission characters, false otherwise + */ +static int +is_nfs4_perms_w(const wchar_t *start, const wchar_t *end, int *permset) +{ + const wchar_t *p = start; + + while (p < end) { + switch (*p++) { + case L'r': + *permset |= ARCHIVE_ENTRY_ACL_READ_DATA; + break; + case L'w': + *permset |= ARCHIVE_ENTRY_ACL_WRITE_DATA; + break; + case L'x': + *permset |= ARCHIVE_ENTRY_ACL_EXECUTE; + break; + case L'p': + *permset |= ARCHIVE_ENTRY_ACL_APPEND_DATA; + break; + case L'D': + *permset |= ARCHIVE_ENTRY_ACL_DELETE_CHILD; + break; + case L'd': + *permset |= ARCHIVE_ENTRY_ACL_DELETE; + break; + case L'a': + *permset |= ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES; + break; + case L'A': + *permset |= ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES; + break; + case L'R': + *permset |= ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS; + break; + case L'W': + *permset |= ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS; + break; + case L'c': + *permset |= ARCHIVE_ENTRY_ACL_READ_ACL; + break; + case L'C': + *permset |= ARCHIVE_ENTRY_ACL_WRITE_ACL; + break; + case L'o': + *permset |= ARCHIVE_ENTRY_ACL_WRITE_OWNER; + break; + case L's': + *permset |= ARCHIVE_ENTRY_ACL_SYNCHRONIZE; + break; + case L'-': + break; + default: + return(0); + } + } + return (1); +} + +/* + * Parse a string as a NFS4 ACL flags field. + * Returns true if the string is non-empty and consists only of NFS4 ACL + * flag characters, false otherwise + */ +static int +is_nfs4_flags_w(const wchar_t *start, const wchar_t *end, int *permset) +{ + const wchar_t *p = start; + + while (p < end) { + switch(*p++) { + case L'f': + *permset |= ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT; + break; + case L'd': + *permset |= ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT; + break; + case L'i': + *permset |= ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY; + break; + case L'n': + *permset |= + ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT; + break; + case L'S': + *permset |= ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS; + break; + case L'F': + *permset |= ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS; + break; + case L'I': + *permset |= ARCHIVE_ENTRY_ACL_ENTRY_INHERITED; + break; + case L'-': + break; + default: + return (0); + } + } + return (1); +} + +/* + * Match "[:whitespace:]*(.*)[:whitespace:]*[:,\n]". *wp is updated + * to point to just after the separator. *start points to the first + * character of the matched text and *end just after the last + * character of the matched identifier. In particular *end - *start + * is the length of the field body, not including leading or trailing + * whitespace. + */ +static void +next_field_w(const wchar_t **wp, const wchar_t **start, + const wchar_t **end, wchar_t *sep) +{ + /* Skip leading whitespace to find start of field. */ + while (**wp == L' ' || **wp == L'\t' || **wp == L'\n') { + (*wp)++; + } + *start = *wp; + + /* Scan for the separator. */ + while (**wp != L'\0' && **wp != L',' && **wp != L':' && + **wp != L'\n' && **wp != L'#') { + (*wp)++; + } + *sep = **wp; + + /* Locate end of field, trim trailing whitespace if necessary */ + if (*wp == *start) { + *end = *wp; + } else { + *end = *wp - 1; + while (**end == L' ' || **end == L'\t' || **end == L'\n') { + (*end)--; + } + (*end)++; + } + + /* Handle in-field comments */ + if (*sep == L'#') { + while (**wp != L'\0' && **wp != L',' && **wp != L'\n') { + (*wp)++; + } + *sep = **wp; + } + + /* Adjust scanner location. */ + if (**wp != L'\0') + (*wp)++; +} + +/* + * Parse an ACL text string. + * + * The want_type argument may be one of the following: + * ARCHIVE_ENTRY_ACL_TYPE_ACCESS - text is a POSIX.1e ACL of type ACCESS + * ARCHIVE_ENTRY_ACL_TYPE_DEFAULT - text is a POSIX.1e ACL of type DEFAULT + * ARCHIVE_ENTRY_ACL_TYPE_NFS4 - text is as a NFSv4 ACL + * + * POSIX.1e ACL entries prefixed with "default:" are treated as + * ARCHIVE_ENTRY_ACL_TYPE_DEFAULT unless type is ARCHIVE_ENTRY_ACL_TYPE_NFS4 + */ +int +archive_acl_from_text_l(struct archive_acl *acl, const char *text, + int want_type, struct archive_string_conv *sc) +{ + struct { + const char *start; + const char *end; + } field[6], name; + + const char *s, *st; + int numfields, fields, n, r, sol, ret; + int type, types, tag, permset, id; + size_t len; + char sep; + + switch (want_type) { + case ARCHIVE_ENTRY_ACL_TYPE_POSIX1E: + want_type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS; + __LA_FALLTHROUGH; + case ARCHIVE_ENTRY_ACL_TYPE_ACCESS: + case ARCHIVE_ENTRY_ACL_TYPE_DEFAULT: + numfields = 5; + break; + case ARCHIVE_ENTRY_ACL_TYPE_NFS4: + numfields = 6; + break; + default: + return (ARCHIVE_FATAL); + } + + ret = ARCHIVE_OK; + types = 0; + + while (text != NULL && *text != '\0') { + /* + * Parse the fields out of the next entry, + * advance 'text' to start of next entry. + */ + fields = 0; + do { + const char *start, *end; + next_field(&text, &start, &end, &sep); + if (fields < numfields) { + field[fields].start = start; + field[fields].end = end; + } + ++fields; + } while (sep == ':'); + + /* Set remaining fields to blank. */ + for (n = fields; n < numfields; ++n) + field[n].start = field[n].end = NULL; + + if (field[0].start != NULL && *(field[0].start) == '#') { + /* Comment, skip entry */ + continue; + } + + n = 0; + sol = 0; + id = -1; + permset = 0; + name.start = name.end = NULL; + + if (want_type != ARCHIVE_ENTRY_ACL_TYPE_NFS4) { + /* POSIX.1e ACLs */ + /* + * Default keyword "default:user::rwx" + * if found, we have one more field + * + * We also support old Solaris extension: + * "defaultuser::rwx" is the default ACL corresponding + * to "user::rwx", etc. valid only for first field + */ + s = field[0].start; + len = field[0].end - field[0].start; + if (*s == 'd' && (len == 1 || (len >= 7 + && memcmp((s + 1), "efault", 6) == 0))) { + type = ARCHIVE_ENTRY_ACL_TYPE_DEFAULT; + if (len > 7) + field[0].start += 7; + else + n = 1; + } else + type = want_type; + + /* Check for a numeric ID in field n+1 or n+3. */ + isint(field[n + 1].start, field[n + 1].end, &id); + /* Field n+3 is optional. */ + if (id == -1 && fields > (n + 3)) + isint(field[n + 3].start, field[n + 3].end, + &id); + + tag = 0; + s = field[n].start; + st = field[n].start + 1; + len = field[n].end - field[n].start; + + if (len == 0) { + ret = ARCHIVE_WARN; + continue; + } + + switch (*s) { + case 'u': + if (len == 1 || (len == 4 + && memcmp(st, "ser", 3) == 0)) + tag = ARCHIVE_ENTRY_ACL_USER_OBJ; + break; + case 'g': + if (len == 1 || (len == 5 + && memcmp(st, "roup", 4) == 0)) + tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; + break; + case 'o': + if (len == 1 || (len == 5 + && memcmp(st, "ther", 4) == 0)) + tag = ARCHIVE_ENTRY_ACL_OTHER; + break; + case 'm': + if (len == 1 || (len == 4 + && memcmp(st, "ask", 3) == 0)) + tag = ARCHIVE_ENTRY_ACL_MASK; + break; + default: + break; + } + + switch (tag) { + case ARCHIVE_ENTRY_ACL_OTHER: + case ARCHIVE_ENTRY_ACL_MASK: + if (fields == (n + 2) + && field[n + 1].start < field[n + 1].end + && ismode(field[n + 1].start, + field[n + 1].end, &permset)) { + /* This is Solaris-style "other:rwx" */ + sol = 1; + } else if (fields == (n + 3) && + field[n + 1].start < field[n + 1].end) { + /* Invalid mask or other field */ + ret = ARCHIVE_WARN; + continue; + } + break; + case ARCHIVE_ENTRY_ACL_USER_OBJ: + case ARCHIVE_ENTRY_ACL_GROUP_OBJ: + if (id != -1 || + field[n + 1].start < field[n + 1].end) { + name = field[n + 1]; + if (tag == ARCHIVE_ENTRY_ACL_USER_OBJ) + tag = ARCHIVE_ENTRY_ACL_USER; + else + tag = ARCHIVE_ENTRY_ACL_GROUP; + } + break; + default: + /* Invalid tag, skip entry */ + ret = ARCHIVE_WARN; + continue; + } + + /* + * Without "default:" we expect mode in field 3 + * Exception: Solaris other and mask fields + */ + if (permset == 0 && !ismode(field[n + 2 - sol].start, + field[n + 2 - sol].end, &permset)) { + /* Invalid mode, skip entry */ + ret = ARCHIVE_WARN; + continue; + } + } else { + /* NFS4 ACLs */ + s = field[0].start; + len = field[0].end - field[0].start; + tag = 0; + + switch (len) { + case 4: + if (memcmp(s, "user", 4) == 0) + tag = ARCHIVE_ENTRY_ACL_USER; + break; + case 5: + if (memcmp(s, "group", 5) == 0) + tag = ARCHIVE_ENTRY_ACL_GROUP; + break; + case 6: + if (memcmp(s, "owner@", 6) == 0) + tag = ARCHIVE_ENTRY_ACL_USER_OBJ; + else if (memcmp(s, "group@", 6) == 0) + tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; + break; + case 9: + if (memcmp(s, "everyone@", 9) == 0) + tag = ARCHIVE_ENTRY_ACL_EVERYONE; + break; + default: + break; + } + + if (tag == 0) { + /* Invalid tag, skip entry */ + ret = ARCHIVE_WARN; + continue; + } else if (tag == ARCHIVE_ENTRY_ACL_USER || + tag == ARCHIVE_ENTRY_ACL_GROUP) { + n = 1; + name = field[1]; + isint(name.start, name.end, &id); + } else + n = 0; + + if (!is_nfs4_perms(field[1 + n].start, + field[1 + n].end, &permset)) { + /* Invalid NFSv4 perms, skip entry */ + ret = ARCHIVE_WARN; + continue; + } + if (!is_nfs4_flags(field[2 + n].start, + field[2 + n].end, &permset)) { + /* Invalid NFSv4 flags, skip entry */ + ret = ARCHIVE_WARN; + continue; + } + s = field[3 + n].start; + len = field[3 + n].end - field[3 + n].start; + type = 0; + if (len == 4) { + if (memcmp(s, "deny", 4) == 0) + type = ARCHIVE_ENTRY_ACL_TYPE_DENY; + } else if (len == 5) { + if (memcmp(s, "allow", 5) == 0) + type = ARCHIVE_ENTRY_ACL_TYPE_ALLOW; + else if (memcmp(s, "audit", 5) == 0) + type = ARCHIVE_ENTRY_ACL_TYPE_AUDIT; + else if (memcmp(s, "alarm", 5) == 0) + type = ARCHIVE_ENTRY_ACL_TYPE_ALARM; + } + if (type == 0) { + /* Invalid entry type, skip entry */ + ret = ARCHIVE_WARN; + continue; + } + isint(field[4 + n].start, field[4 + n].end, + &id); + } + + /* Add entry to the internal list. */ + r = archive_acl_add_entry_len_l(acl, type, permset, + tag, id, name.start, name.end - name.start, sc); + if (r < ARCHIVE_WARN) + return (r); + if (r != ARCHIVE_OK) + ret = ARCHIVE_WARN; + types |= type; + } + + /* Reset ACL */ + archive_acl_reset(acl, types); + + return (ret); +} + +/* + * Parse a string to a positive decimal integer. Returns true if + * the string is non-empty and consists only of decimal digits, + * false otherwise. + */ +static int +isint(const char *start, const char *end, int *result) +{ + int n = 0; + if (start >= end) + return (0); + while (start < end) { + if (*start < '0' || *start > '9') + return (0); + if (n > (INT_MAX / 10) || + (n == INT_MAX / 10 && (*start - '0') > INT_MAX % 10)) { + n = INT_MAX; + } else { + n *= 10; + n += *start - '0'; + } + start++; + } + *result = n; + return (1); +} + +/* + * Parse a string as a mode field. Returns true if + * the string is non-empty and consists only of mode characters, + * false otherwise. + */ +static int +ismode(const char *start, const char *end, int *permset) +{ + const char *p; + + if (start >= end) + return (0); + p = start; + *permset = 0; + while (p < end) { + switch (*p++) { + case 'r': case 'R': + *permset |= ARCHIVE_ENTRY_ACL_READ; + break; + case 'w': case 'W': + *permset |= ARCHIVE_ENTRY_ACL_WRITE; + break; + case 'x': case 'X': + *permset |= ARCHIVE_ENTRY_ACL_EXECUTE; + break; + case '-': + break; + default: + return (0); + } + } + return (1); +} + +/* + * Parse a string as a NFS4 ACL permission field. + * Returns true if the string is non-empty and consists only of NFS4 ACL + * permission characters, false otherwise + */ +static int +is_nfs4_perms(const char *start, const char *end, int *permset) +{ + const char *p = start; + + while (p < end) { + switch (*p++) { + case 'r': + *permset |= ARCHIVE_ENTRY_ACL_READ_DATA; + break; + case 'w': + *permset |= ARCHIVE_ENTRY_ACL_WRITE_DATA; + break; + case 'x': + *permset |= ARCHIVE_ENTRY_ACL_EXECUTE; + break; + case 'p': + *permset |= ARCHIVE_ENTRY_ACL_APPEND_DATA; + break; + case 'D': + *permset |= ARCHIVE_ENTRY_ACL_DELETE_CHILD; + break; + case 'd': + *permset |= ARCHIVE_ENTRY_ACL_DELETE; + break; + case 'a': + *permset |= ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES; + break; + case 'A': + *permset |= ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES; + break; + case 'R': + *permset |= ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS; + break; + case 'W': + *permset |= ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS; + break; + case 'c': + *permset |= ARCHIVE_ENTRY_ACL_READ_ACL; + break; + case 'C': + *permset |= ARCHIVE_ENTRY_ACL_WRITE_ACL; + break; + case 'o': + *permset |= ARCHIVE_ENTRY_ACL_WRITE_OWNER; + break; + case 's': + *permset |= ARCHIVE_ENTRY_ACL_SYNCHRONIZE; + break; + case '-': + break; + default: + return(0); + } + } + return (1); +} + +/* + * Parse a string as a NFS4 ACL flags field. + * Returns true if the string is non-empty and consists only of NFS4 ACL + * flag characters, false otherwise + */ +static int +is_nfs4_flags(const char *start, const char *end, int *permset) +{ + const char *p = start; + + while (p < end) { + switch(*p++) { + case 'f': + *permset |= ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT; + break; + case 'd': + *permset |= ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT; + break; + case 'i': + *permset |= ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY; + break; + case 'n': + *permset |= + ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT; + break; + case 'S': + *permset |= ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS; + break; + case 'F': + *permset |= ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS; + break; + case 'I': + *permset |= ARCHIVE_ENTRY_ACL_ENTRY_INHERITED; + break; + case '-': + break; + default: + return (0); + } + } + return (1); +} + +/* + * Match "[:whitespace:]*(.*)[:whitespace:]*[:,\n]". *wp is updated + * to point to just after the separator. *start points to the first + * character of the matched text and *end just after the last + * character of the matched identifier. In particular *end - *start + * is the length of the field body, not including leading or trailing + * whitespace. + */ +static void +next_field(const char **p, const char **start, + const char **end, char *sep) +{ + /* Skip leading whitespace to find start of field. */ + while (**p == ' ' || **p == '\t' || **p == '\n') { + (*p)++; + } + *start = *p; + + /* Scan for the separator. */ + while (**p != '\0' && **p != ',' && **p != ':' && **p != '\n' && + **p != '#') { + (*p)++; + } + *sep = **p; + + /* Locate end of field, trim trailing whitespace if necessary */ + if (*p == *start) { + *end = *p; + } else { + *end = *p - 1; + while (**end == ' ' || **end == '\t' || **end == '\n') { + (*end)--; + } + (*end)++; + } + + /* Handle in-field comments */ + if (*sep == '#') { + while (**p != '\0' && **p != ',' && **p != '\n') { + (*p)++; + } + *sep = **p; + } + + /* Adjust scanner location. */ + if (**p != '\0') + (*p)++; +} diff --git a/c/archive_acl_private.h b/c/archive_acl_private.h new file mode 100644 --- /dev/null +++ b/c/archive_acl_private.h @@ -0,0 +1,83 @@ +/*- + * Copyright (c) 2003-2010 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD$ + */ + +#ifndef ARCHIVE_ACL_PRIVATE_H_INCLUDED +#define ARCHIVE_ACL_PRIVATE_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif + +#include "archive_string.h" + +struct archive_acl_entry { + struct archive_acl_entry *next; + int type; /* E.g., access or default */ + int tag; /* E.g., user/group/other/mask */ + int permset; /* r/w/x bits */ + int id; /* uid/gid for user/group */ + struct archive_mstring name; /* uname/gname */ +}; + +struct archive_acl { + mode_t mode; + struct archive_acl_entry *acl_head; + struct archive_acl_entry *acl_p; + int acl_state; /* See acl_next for details. */ + wchar_t *acl_text_w; + char *acl_text; + int acl_types; +}; + +void archive_acl_clear(struct archive_acl *); +void archive_acl_copy(struct archive_acl *, struct archive_acl *); +int archive_acl_count(struct archive_acl *, int); +int archive_acl_types(struct archive_acl *); +int archive_acl_reset(struct archive_acl *, int); +int archive_acl_next(struct archive *, struct archive_acl *, int, + int *, int *, int *, int *, const char **); + +int archive_acl_add_entry(struct archive_acl *, int, int, int, int, const char *); +int archive_acl_add_entry_w_len(struct archive_acl *, + int, int, int, int, const wchar_t *, size_t); +int archive_acl_add_entry_len(struct archive_acl *, + int, int, int, int, const char *, size_t); + +wchar_t *archive_acl_to_text_w(struct archive_acl *, ssize_t *, int, + struct archive *); +char *archive_acl_to_text_l(struct archive_acl *, ssize_t *, int, + struct archive_string_conv *); + +/* + * ACL text parser. + */ +int archive_acl_from_text_w(struct archive_acl *, const wchar_t * /* wtext */, + int /* type */); +int archive_acl_from_text_l(struct archive_acl *, const char * /* text */, + int /* type */, struct archive_string_conv *); + +#endif /* ARCHIVE_ENTRY_PRIVATE_H_INCLUDED */ diff --git a/c/archive_blake2.h b/c/archive_blake2.h new file mode 100644 --- /dev/null +++ b/c/archive_blake2.h @@ -0,0 +1,195 @@ +/* + BLAKE2 reference source code package - reference C implementations + + Copyright 2012, Samuel Neves . You may use this under the + terms of the CC0, the OpenSSL Licence, or the Apache Public License 2.0, at + your option. The terms of these licenses can be found at: + + - CC0 1.0 Universal : http://creativecommons.org/publicdomain/zero/1.0 + - OpenSSL license : https://www.openssl.org/source/license.html + - Apache 2.0 : http://www.apache.org/licenses/LICENSE-2.0 + + More information about the BLAKE2 hash function can be found at + https://blake2.net. +*/ + +#ifndef ARCHIVE_BLAKE2_H +#define ARCHIVE_BLAKE2_H + +#include +#include + +#if defined(_MSC_VER) +#define BLAKE2_PACKED(x) __pragma(pack(push, 1)) x __pragma(pack(pop)) +#else +#define BLAKE2_PACKED(x) x __attribute__((packed)) +#endif + +#if defined(__cplusplus) +extern "C" { +#endif + + enum blake2s_constant + { + BLAKE2S_BLOCKBYTES = 64, + BLAKE2S_OUTBYTES = 32, + BLAKE2S_KEYBYTES = 32, + BLAKE2S_SALTBYTES = 8, + BLAKE2S_PERSONALBYTES = 8 + }; + + enum blake2b_constant + { + BLAKE2B_BLOCKBYTES = 128, + BLAKE2B_OUTBYTES = 64, + BLAKE2B_KEYBYTES = 64, + BLAKE2B_SALTBYTES = 16, + BLAKE2B_PERSONALBYTES = 16 + }; + + typedef struct blake2s_state__ + { + uint32_t h[8]; + uint32_t t[2]; + uint32_t f[2]; + uint8_t buf[BLAKE2S_BLOCKBYTES]; + size_t buflen; + size_t outlen; + uint8_t last_node; + } blake2s_state; + + typedef struct blake2b_state__ + { + uint64_t h[8]; + uint64_t t[2]; + uint64_t f[2]; + uint8_t buf[BLAKE2B_BLOCKBYTES]; + size_t buflen; + size_t outlen; + uint8_t last_node; + } blake2b_state; + + typedef struct blake2sp_state__ + { + blake2s_state S[8][1]; + blake2s_state R[1]; + uint8_t buf[8 * BLAKE2S_BLOCKBYTES]; + size_t buflen; + size_t outlen; + } blake2sp_state; + + typedef struct blake2bp_state__ + { + blake2b_state S[4][1]; + blake2b_state R[1]; + uint8_t buf[4 * BLAKE2B_BLOCKBYTES]; + size_t buflen; + size_t outlen; + } blake2bp_state; + + BLAKE2_PACKED(struct blake2s_param__ + { + uint8_t digest_length; /* 1 */ + uint8_t key_length; /* 2 */ + uint8_t fanout; /* 3 */ + uint8_t depth; /* 4 */ + uint32_t leaf_length; /* 8 */ + uint32_t node_offset; /* 12 */ + uint16_t xof_length; /* 14 */ + uint8_t node_depth; /* 15 */ + uint8_t inner_length; /* 16 */ + /* uint8_t reserved[0]; */ + uint8_t salt[BLAKE2S_SALTBYTES]; /* 24 */ + uint8_t personal[BLAKE2S_PERSONALBYTES]; /* 32 */ + }); + + typedef struct blake2s_param__ blake2s_param; + + BLAKE2_PACKED(struct blake2b_param__ + { + uint8_t digest_length; /* 1 */ + uint8_t key_length; /* 2 */ + uint8_t fanout; /* 3 */ + uint8_t depth; /* 4 */ + uint32_t leaf_length; /* 8 */ + uint32_t node_offset; /* 12 */ + uint32_t xof_length; /* 16 */ + uint8_t node_depth; /* 17 */ + uint8_t inner_length; /* 18 */ + uint8_t reserved[14]; /* 32 */ + uint8_t salt[BLAKE2B_SALTBYTES]; /* 48 */ + uint8_t personal[BLAKE2B_PERSONALBYTES]; /* 64 */ + }); + + typedef struct blake2b_param__ blake2b_param; + + typedef struct blake2xs_state__ + { + blake2s_state S[1]; + blake2s_param P[1]; + } blake2xs_state; + + typedef struct blake2xb_state__ + { + blake2b_state S[1]; + blake2b_param P[1]; + } blake2xb_state; + + /* Padded structs result in a compile-time error */ + enum { + BLAKE2_DUMMY_1 = 1/(sizeof(blake2s_param) == BLAKE2S_OUTBYTES), + BLAKE2_DUMMY_2 = 1/(sizeof(blake2b_param) == BLAKE2B_OUTBYTES) + }; + + /* Streaming API */ + int blake2s_init( blake2s_state *S, size_t outlen ); + int blake2s_init_key( blake2s_state *S, size_t outlen, const void *key, size_t keylen ); + int blake2s_init_param( blake2s_state *S, const blake2s_param *P ); + int blake2s_update( blake2s_state *S, const void *in, size_t inlen ); + int blake2s_final( blake2s_state *S, void *out, size_t outlen ); + + int blake2b_init( blake2b_state *S, size_t outlen ); + int blake2b_init_key( blake2b_state *S, size_t outlen, const void *key, size_t keylen ); + int blake2b_init_param( blake2b_state *S, const blake2b_param *P ); + int blake2b_update( blake2b_state *S, const void *in, size_t inlen ); + int blake2b_final( blake2b_state *S, void *out, size_t outlen ); + + int blake2sp_init( blake2sp_state *S, size_t outlen ); + int blake2sp_init_key( blake2sp_state *S, size_t outlen, const void *key, size_t keylen ); + int blake2sp_update( blake2sp_state *S, const void *in, size_t inlen ); + int blake2sp_final( blake2sp_state *S, void *out, size_t outlen ); + + int blake2bp_init( blake2bp_state *S, size_t outlen ); + int blake2bp_init_key( blake2bp_state *S, size_t outlen, const void *key, size_t keylen ); + int blake2bp_update( blake2bp_state *S, const void *in, size_t inlen ); + int blake2bp_final( blake2bp_state *S, void *out, size_t outlen ); + + /* Variable output length API */ + int blake2xs_init( blake2xs_state *S, const size_t outlen ); + int blake2xs_init_key( blake2xs_state *S, const size_t outlen, const void *key, size_t keylen ); + int blake2xs_update( blake2xs_state *S, const void *in, size_t inlen ); + int blake2xs_final(blake2xs_state *S, void *out, size_t outlen); + + int blake2xb_init( blake2xb_state *S, const size_t outlen ); + int blake2xb_init_key( blake2xb_state *S, const size_t outlen, const void *key, size_t keylen ); + int blake2xb_update( blake2xb_state *S, const void *in, size_t inlen ); + int blake2xb_final(blake2xb_state *S, void *out, size_t outlen); + + /* Simple API */ + int blake2s( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen ); + int blake2b( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen ); + + int blake2sp( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen ); + int blake2bp( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen ); + + int blake2xs( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen ); + int blake2xb( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen ); + + /* This is simply an alias for blake2b */ + int blake2( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen ); + +#if defined(__cplusplus) +} +#endif + +#endif diff --git a/c/archive_blake2_impl.h b/c/archive_blake2_impl.h new file mode 100644 --- /dev/null +++ b/c/archive_blake2_impl.h @@ -0,0 +1,161 @@ +/* + BLAKE2 reference source code package - reference C implementations + + Copyright 2012, Samuel Neves . You may use this under the + terms of the CC0, the OpenSSL Licence, or the Apache Public License 2.0, at + your option. The terms of these licenses can be found at: + + - CC0 1.0 Universal : http://creativecommons.org/publicdomain/zero/1.0 + - OpenSSL license : https://www.openssl.org/source/license.html + - Apache 2.0 : http://www.apache.org/licenses/LICENSE-2.0 + + More information about the BLAKE2 hash function can be found at + https://blake2.net. +*/ + +#ifndef ARCHIVE_BLAKE2_IMPL_H +#define ARCHIVE_BLAKE2_IMPL_H + +#include +#include + +#if !defined(__cplusplus) && (!defined(__STDC_VERSION__) || __STDC_VERSION__ < 199901L) + #if defined(_MSC_VER) + #define BLAKE2_INLINE __inline + #elif defined(__GNUC__) + #define BLAKE2_INLINE __inline__ + #else + #define BLAKE2_INLINE + #endif +#else + #define BLAKE2_INLINE inline +#endif + +static BLAKE2_INLINE uint32_t load32( const void *src ) +{ +#if defined(NATIVE_LITTLE_ENDIAN) + uint32_t w; + memcpy(&w, src, sizeof w); + return w; +#else + const uint8_t *p = ( const uint8_t * )src; + return (( uint32_t )( p[0] ) << 0) | + (( uint32_t )( p[1] ) << 8) | + (( uint32_t )( p[2] ) << 16) | + (( uint32_t )( p[3] ) << 24) ; +#endif +} + +static BLAKE2_INLINE uint64_t load64( const void *src ) +{ +#if defined(NATIVE_LITTLE_ENDIAN) + uint64_t w; + memcpy(&w, src, sizeof w); + return w; +#else + const uint8_t *p = ( const uint8_t * )src; + return (( uint64_t )( p[0] ) << 0) | + (( uint64_t )( p[1] ) << 8) | + (( uint64_t )( p[2] ) << 16) | + (( uint64_t )( p[3] ) << 24) | + (( uint64_t )( p[4] ) << 32) | + (( uint64_t )( p[5] ) << 40) | + (( uint64_t )( p[6] ) << 48) | + (( uint64_t )( p[7] ) << 56) ; +#endif +} + +static BLAKE2_INLINE uint16_t load16( const void *src ) +{ +#if defined(NATIVE_LITTLE_ENDIAN) + uint16_t w; + memcpy(&w, src, sizeof w); + return w; +#else + const uint8_t *p = ( const uint8_t * )src; + return ( uint16_t )((( uint32_t )( p[0] ) << 0) | + (( uint32_t )( p[1] ) << 8)); +#endif +} + +static BLAKE2_INLINE void store16( void *dst, uint16_t w ) +{ +#if defined(NATIVE_LITTLE_ENDIAN) + memcpy(dst, &w, sizeof w); +#else + uint8_t *p = ( uint8_t * )dst; + *p++ = ( uint8_t )w; w >>= 8; + *p++ = ( uint8_t )w; +#endif +} + +static BLAKE2_INLINE void store32( void *dst, uint32_t w ) +{ +#if defined(NATIVE_LITTLE_ENDIAN) + memcpy(dst, &w, sizeof w); +#else + uint8_t *p = ( uint8_t * )dst; + p[0] = (uint8_t)(w >> 0); + p[1] = (uint8_t)(w >> 8); + p[2] = (uint8_t)(w >> 16); + p[3] = (uint8_t)(w >> 24); +#endif +} + +static BLAKE2_INLINE void store64( void *dst, uint64_t w ) +{ +#if defined(NATIVE_LITTLE_ENDIAN) + memcpy(dst, &w, sizeof w); +#else + uint8_t *p = ( uint8_t * )dst; + p[0] = (uint8_t)(w >> 0); + p[1] = (uint8_t)(w >> 8); + p[2] = (uint8_t)(w >> 16); + p[3] = (uint8_t)(w >> 24); + p[4] = (uint8_t)(w >> 32); + p[5] = (uint8_t)(w >> 40); + p[6] = (uint8_t)(w >> 48); + p[7] = (uint8_t)(w >> 56); +#endif +} + +static BLAKE2_INLINE uint64_t load48( const void *src ) +{ + const uint8_t *p = ( const uint8_t * )src; + return (( uint64_t )( p[0] ) << 0) | + (( uint64_t )( p[1] ) << 8) | + (( uint64_t )( p[2] ) << 16) | + (( uint64_t )( p[3] ) << 24) | + (( uint64_t )( p[4] ) << 32) | + (( uint64_t )( p[5] ) << 40) ; +} + +static BLAKE2_INLINE void store48( void *dst, uint64_t w ) +{ + uint8_t *p = ( uint8_t * )dst; + p[0] = (uint8_t)(w >> 0); + p[1] = (uint8_t)(w >> 8); + p[2] = (uint8_t)(w >> 16); + p[3] = (uint8_t)(w >> 24); + p[4] = (uint8_t)(w >> 32); + p[5] = (uint8_t)(w >> 40); +} + +static BLAKE2_INLINE uint32_t rotr32( const uint32_t w, const unsigned c ) +{ + return ( w >> c ) | ( w << ( 32 - c ) ); +} + +static BLAKE2_INLINE uint64_t rotr64( const uint64_t w, const unsigned c ) +{ + return ( w >> c ) | ( w << ( 64 - c ) ); +} + +/* prevents compiler optimizing out memset() */ +static BLAKE2_INLINE void secure_zero_memory(void *v, size_t n) +{ + static void *(*const volatile memset_v)(void *, int, size_t) = &memset; + memset_v(v, 0, n); +} + +#endif diff --git a/c/archive_blake2s_ref.c b/c/archive_blake2s_ref.c new file mode 100644 --- /dev/null +++ b/c/archive_blake2s_ref.c @@ -0,0 +1,367 @@ +/* + BLAKE2 reference source code package - reference C implementations + + Copyright 2012, Samuel Neves . You may use this under the + terms of the CC0, the OpenSSL Licence, or the Apache Public License 2.0, at + your option. The terms of these licenses can be found at: + + - CC0 1.0 Universal : http://creativecommons.org/publicdomain/zero/1.0 + - OpenSSL license : https://www.openssl.org/source/license.html + - Apache 2.0 : http://www.apache.org/licenses/LICENSE-2.0 + + More information about the BLAKE2 hash function can be found at + https://blake2.net. +*/ + +#include +#include +#include + +#include "archive_blake2.h" +#include "archive_blake2_impl.h" + +static const uint32_t blake2s_IV[8] = +{ + 0x6A09E667UL, 0xBB67AE85UL, 0x3C6EF372UL, 0xA54FF53AUL, + 0x510E527FUL, 0x9B05688CUL, 0x1F83D9ABUL, 0x5BE0CD19UL +}; + +static const uint8_t blake2s_sigma[10][16] = +{ + { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 } , + { 14, 10, 4, 8, 9, 15, 13, 6, 1, 12, 0, 2, 11, 7, 5, 3 } , + { 11, 8, 12, 0, 5, 2, 15, 13, 10, 14, 3, 6, 7, 1, 9, 4 } , + { 7, 9, 3, 1, 13, 12, 11, 14, 2, 6, 5, 10, 4, 0, 15, 8 } , + { 9, 0, 5, 7, 2, 4, 10, 15, 14, 1, 11, 12, 6, 8, 3, 13 } , + { 2, 12, 6, 10, 0, 11, 8, 3, 4, 13, 7, 5, 15, 14, 1, 9 } , + { 12, 5, 1, 15, 14, 13, 4, 10, 0, 7, 6, 3, 9, 2, 8, 11 } , + { 13, 11, 7, 14, 12, 1, 3, 9, 5, 0, 15, 4, 8, 6, 2, 10 } , + { 6, 15, 14, 9, 11, 3, 0, 8, 12, 2, 13, 7, 1, 4, 10, 5 } , + { 10, 2, 8, 4, 7, 6, 1, 5, 15, 11, 9, 14, 3, 12, 13 , 0 } , +}; + +static void blake2s_set_lastnode( blake2s_state *S ) +{ + S->f[1] = (uint32_t)-1; +} + +/* Some helper functions, not necessarily useful */ +static int blake2s_is_lastblock( const blake2s_state *S ) +{ + return S->f[0] != 0; +} + +static void blake2s_set_lastblock( blake2s_state *S ) +{ + if( S->last_node ) blake2s_set_lastnode( S ); + + S->f[0] = (uint32_t)-1; +} + +static void blake2s_increment_counter( blake2s_state *S, const uint32_t inc ) +{ + S->t[0] += inc; + S->t[1] += ( S->t[0] < inc ); +} + +static void blake2s_init0( blake2s_state *S ) +{ + size_t i; + memset( S, 0, sizeof( blake2s_state ) ); + + for( i = 0; i < 8; ++i ) S->h[i] = blake2s_IV[i]; +} + +/* init2 xors IV with input parameter block */ +int blake2s_init_param( blake2s_state *S, const blake2s_param *P ) +{ + const unsigned char *p = ( const unsigned char * )( P ); + size_t i; + + blake2s_init0( S ); + + /* IV XOR ParamBlock */ + for( i = 0; i < 8; ++i ) + S->h[i] ^= load32( &p[i * 4] ); + + S->outlen = P->digest_length; + return 0; +} + + +/* Sequential blake2s initialization */ +int blake2s_init( blake2s_state *S, size_t outlen ) +{ + blake2s_param P[1]; + + /* Move interval verification here? */ + if ( ( !outlen ) || ( outlen > BLAKE2S_OUTBYTES ) ) return -1; + + P->digest_length = (uint8_t)outlen; + P->key_length = 0; + P->fanout = 1; + P->depth = 1; + store32( &P->leaf_length, 0 ); + store32( &P->node_offset, 0 ); + store16( &P->xof_length, 0 ); + P->node_depth = 0; + P->inner_length = 0; + /* memset(P->reserved, 0, sizeof(P->reserved) ); */ + memset( P->salt, 0, sizeof( P->salt ) ); + memset( P->personal, 0, sizeof( P->personal ) ); + return blake2s_init_param( S, P ); +} + +int blake2s_init_key( blake2s_state *S, size_t outlen, const void *key, size_t keylen ) +{ + blake2s_param P[1]; + + if ( ( !outlen ) || ( outlen > BLAKE2S_OUTBYTES ) ) return -1; + + if ( !key || !keylen || keylen > BLAKE2S_KEYBYTES ) return -1; + + P->digest_length = (uint8_t)outlen; + P->key_length = (uint8_t)keylen; + P->fanout = 1; + P->depth = 1; + store32( &P->leaf_length, 0 ); + store32( &P->node_offset, 0 ); + store16( &P->xof_length, 0 ); + P->node_depth = 0; + P->inner_length = 0; + /* memset(P->reserved, 0, sizeof(P->reserved) ); */ + memset( P->salt, 0, sizeof( P->salt ) ); + memset( P->personal, 0, sizeof( P->personal ) ); + + if( blake2s_init_param( S, P ) < 0 ) return -1; + + { + uint8_t block[BLAKE2S_BLOCKBYTES]; + memset( block, 0, BLAKE2S_BLOCKBYTES ); + memcpy( block, key, keylen ); + blake2s_update( S, block, BLAKE2S_BLOCKBYTES ); + secure_zero_memory( block, BLAKE2S_BLOCKBYTES ); /* Burn the key from stack */ + } + return 0; +} + +#define G(r,i,a,b,c,d) \ + do { \ + a = a + b + m[blake2s_sigma[r][2*i+0]]; \ + d = rotr32(d ^ a, 16); \ + c = c + d; \ + b = rotr32(b ^ c, 12); \ + a = a + b + m[blake2s_sigma[r][2*i+1]]; \ + d = rotr32(d ^ a, 8); \ + c = c + d; \ + b = rotr32(b ^ c, 7); \ + } while(0) + +#define ROUND(r) \ + do { \ + G(r,0,v[ 0],v[ 4],v[ 8],v[12]); \ + G(r,1,v[ 1],v[ 5],v[ 9],v[13]); \ + G(r,2,v[ 2],v[ 6],v[10],v[14]); \ + G(r,3,v[ 3],v[ 7],v[11],v[15]); \ + G(r,4,v[ 0],v[ 5],v[10],v[15]); \ + G(r,5,v[ 1],v[ 6],v[11],v[12]); \ + G(r,6,v[ 2],v[ 7],v[ 8],v[13]); \ + G(r,7,v[ 3],v[ 4],v[ 9],v[14]); \ + } while(0) + +static void blake2s_compress( blake2s_state *S, const uint8_t in[BLAKE2S_BLOCKBYTES] ) +{ + uint32_t m[16]; + uint32_t v[16]; + size_t i; + + for( i = 0; i < 16; ++i ) { + m[i] = load32( in + i * sizeof( m[i] ) ); + } + + for( i = 0; i < 8; ++i ) { + v[i] = S->h[i]; + } + + v[ 8] = blake2s_IV[0]; + v[ 9] = blake2s_IV[1]; + v[10] = blake2s_IV[2]; + v[11] = blake2s_IV[3]; + v[12] = S->t[0] ^ blake2s_IV[4]; + v[13] = S->t[1] ^ blake2s_IV[5]; + v[14] = S->f[0] ^ blake2s_IV[6]; + v[15] = S->f[1] ^ blake2s_IV[7]; + + ROUND( 0 ); + ROUND( 1 ); + ROUND( 2 ); + ROUND( 3 ); + ROUND( 4 ); + ROUND( 5 ); + ROUND( 6 ); + ROUND( 7 ); + ROUND( 8 ); + ROUND( 9 ); + + for( i = 0; i < 8; ++i ) { + S->h[i] = S->h[i] ^ v[i] ^ v[i + 8]; + } +} + +#undef G +#undef ROUND + +int blake2s_update( blake2s_state *S, const void *pin, size_t inlen ) +{ + const unsigned char * in = (const unsigned char *)pin; + if( inlen > 0 ) + { + size_t left = S->buflen; + size_t fill = BLAKE2S_BLOCKBYTES - left; + if( inlen > fill ) + { + S->buflen = 0; + memcpy( S->buf + left, in, fill ); /* Fill buffer */ + blake2s_increment_counter( S, BLAKE2S_BLOCKBYTES ); + blake2s_compress( S, S->buf ); /* Compress */ + in += fill; inlen -= fill; + while(inlen > BLAKE2S_BLOCKBYTES) { + blake2s_increment_counter(S, BLAKE2S_BLOCKBYTES); + blake2s_compress( S, in ); + in += BLAKE2S_BLOCKBYTES; + inlen -= BLAKE2S_BLOCKBYTES; + } + } + memcpy( S->buf + S->buflen, in, inlen ); + S->buflen += inlen; + } + return 0; +} + +int blake2s_final( blake2s_state *S, void *out, size_t outlen ) +{ + uint8_t buffer[BLAKE2S_OUTBYTES] = {0}; + size_t i; + + if( out == NULL || outlen < S->outlen ) + return -1; + + if( blake2s_is_lastblock( S ) ) + return -1; + + blake2s_increment_counter( S, ( uint32_t )S->buflen ); + blake2s_set_lastblock( S ); + memset( S->buf + S->buflen, 0, BLAKE2S_BLOCKBYTES - S->buflen ); /* Padding */ + blake2s_compress( S, S->buf ); + + for( i = 0; i < 8; ++i ) /* Output full hash to temp buffer */ + store32( buffer + sizeof( S->h[i] ) * i, S->h[i] ); + + memcpy( out, buffer, outlen ); + secure_zero_memory(buffer, sizeof(buffer)); + return 0; +} + +int blake2s( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen ) +{ + blake2s_state S[1]; + + /* Verify parameters */ + if ( NULL == in && inlen > 0 ) return -1; + + if ( NULL == out ) return -1; + + if ( NULL == key && keylen > 0) return -1; + + if( !outlen || outlen > BLAKE2S_OUTBYTES ) return -1; + + if( keylen > BLAKE2S_KEYBYTES ) return -1; + + if( keylen > 0 ) + { + if( blake2s_init_key( S, outlen, key, keylen ) < 0 ) return -1; + } + else + { + if( blake2s_init( S, outlen ) < 0 ) return -1; + } + + blake2s_update( S, ( const uint8_t * )in, inlen ); + blake2s_final( S, out, outlen ); + return 0; +} + +#if defined(SUPERCOP) +int crypto_hash( unsigned char *out, unsigned char *in, unsigned long long inlen ) +{ + return blake2s( out, BLAKE2S_OUTBYTES, in, inlen, NULL, 0 ); +} +#endif + +#if defined(BLAKE2S_SELFTEST) +#include +#include "blake2-kat.h" +int main( void ) +{ + uint8_t key[BLAKE2S_KEYBYTES]; + uint8_t buf[BLAKE2_KAT_LENGTH]; + size_t i, step; + + for( i = 0; i < BLAKE2S_KEYBYTES; ++i ) + key[i] = ( uint8_t )i; + + for( i = 0; i < BLAKE2_KAT_LENGTH; ++i ) + buf[i] = ( uint8_t )i; + + /* Test simple API */ + for( i = 0; i < BLAKE2_KAT_LENGTH; ++i ) + { + uint8_t hash[BLAKE2S_OUTBYTES]; + blake2s( hash, BLAKE2S_OUTBYTES, buf, i, key, BLAKE2S_KEYBYTES ); + + if( 0 != memcmp( hash, blake2s_keyed_kat[i], BLAKE2S_OUTBYTES ) ) + { + goto fail; + } + } + + /* Test streaming API */ + for(step = 1; step < BLAKE2S_BLOCKBYTES; ++step) { + for (i = 0; i < BLAKE2_KAT_LENGTH; ++i) { + uint8_t hash[BLAKE2S_OUTBYTES]; + blake2s_state S; + uint8_t * p = buf; + size_t mlen = i; + int err = 0; + + if( (err = blake2s_init_key(&S, BLAKE2S_OUTBYTES, key, BLAKE2S_KEYBYTES)) < 0 ) { + goto fail; + } + + while (mlen >= step) { + if ( (err = blake2s_update(&S, p, step)) < 0 ) { + goto fail; + } + mlen -= step; + p += step; + } + if ( (err = blake2s_update(&S, p, mlen)) < 0) { + goto fail; + } + if ( (err = blake2s_final(&S, hash, BLAKE2S_OUTBYTES)) < 0) { + goto fail; + } + + if (0 != memcmp(hash, blake2s_keyed_kat[i], BLAKE2S_OUTBYTES)) { + goto fail; + } + } + } + + puts( "ok" ); + return 0; +fail: + puts("error"); + return -1; +} +#endif diff --git a/c/archive_blake2sp_ref.c b/c/archive_blake2sp_ref.c new file mode 100644 --- /dev/null +++ b/c/archive_blake2sp_ref.c @@ -0,0 +1,359 @@ +/* + BLAKE2 reference source code package - reference C implementations + + Copyright 2012, Samuel Neves . You may use this under the + terms of the CC0, the OpenSSL Licence, or the Apache Public License 2.0, at + your option. The terms of these licenses can be found at: + + - CC0 1.0 Universal : http://creativecommons.org/publicdomain/zero/1.0 + - OpenSSL license : https://www.openssl.org/source/license.html + - Apache 2.0 : http://www.apache.org/licenses/LICENSE-2.0 + + More information about the BLAKE2 hash function can be found at + https://blake2.net. +*/ + +#include +#include +#include + +#if defined(_OPENMP) +#include +#endif + +#include "archive_blake2.h" +#include "archive_blake2_impl.h" + +#define PARALLELISM_DEGREE 8 + +/* + blake2sp_init_param defaults to setting the expecting output length + from the digest_length parameter block field. + + In some cases, however, we do not want this, as the output length + of these instances is given by inner_length instead. +*/ +static int blake2sp_init_leaf_param( blake2s_state *S, const blake2s_param *P ) +{ + int err = blake2s_init_param(S, P); + S->outlen = P->inner_length; + return err; +} + +static int blake2sp_init_leaf( blake2s_state *S, size_t outlen, size_t keylen, uint32_t offset ) +{ + blake2s_param P[1]; + P->digest_length = (uint8_t)outlen; + P->key_length = (uint8_t)keylen; + P->fanout = PARALLELISM_DEGREE; + P->depth = 2; + store32( &P->leaf_length, 0 ); + store32( &P->node_offset, offset ); + store16( &P->xof_length, 0 ); + P->node_depth = 0; + P->inner_length = BLAKE2S_OUTBYTES; + memset( P->salt, 0, sizeof( P->salt ) ); + memset( P->personal, 0, sizeof( P->personal ) ); + return blake2sp_init_leaf_param( S, P ); +} + +static int blake2sp_init_root( blake2s_state *S, size_t outlen, size_t keylen ) +{ + blake2s_param P[1]; + P->digest_length = (uint8_t)outlen; + P->key_length = (uint8_t)keylen; + P->fanout = PARALLELISM_DEGREE; + P->depth = 2; + store32( &P->leaf_length, 0 ); + store32( &P->node_offset, 0 ); + store16( &P->xof_length, 0 ); + P->node_depth = 1; + P->inner_length = BLAKE2S_OUTBYTES; + memset( P->salt, 0, sizeof( P->salt ) ); + memset( P->personal, 0, sizeof( P->personal ) ); + return blake2s_init_param( S, P ); +} + + +int blake2sp_init( blake2sp_state *S, size_t outlen ) +{ + size_t i; + + if( !outlen || outlen > BLAKE2S_OUTBYTES ) return -1; + + memset( S->buf, 0, sizeof( S->buf ) ); + S->buflen = 0; + S->outlen = outlen; + + if( blake2sp_init_root( S->R, outlen, 0 ) < 0 ) + return -1; + + for( i = 0; i < PARALLELISM_DEGREE; ++i ) + if( blake2sp_init_leaf( S->S[i], outlen, 0, (uint32_t)i ) < 0 ) return -1; + + S->R->last_node = 1; + S->S[PARALLELISM_DEGREE - 1]->last_node = 1; + return 0; +} + +int blake2sp_init_key( blake2sp_state *S, size_t outlen, const void *key, size_t keylen ) +{ + size_t i; + + if( !outlen || outlen > BLAKE2S_OUTBYTES ) return -1; + + if( !key || !keylen || keylen > BLAKE2S_KEYBYTES ) return -1; + + memset( S->buf, 0, sizeof( S->buf ) ); + S->buflen = 0; + S->outlen = outlen; + + if( blake2sp_init_root( S->R, outlen, keylen ) < 0 ) + return -1; + + for( i = 0; i < PARALLELISM_DEGREE; ++i ) + if( blake2sp_init_leaf( S->S[i], outlen, keylen, (uint32_t)i ) < 0 ) return -1; + + S->R->last_node = 1; + S->S[PARALLELISM_DEGREE - 1]->last_node = 1; + { + uint8_t block[BLAKE2S_BLOCKBYTES]; + memset( block, 0, BLAKE2S_BLOCKBYTES ); + memcpy( block, key, keylen ); + + for( i = 0; i < PARALLELISM_DEGREE; ++i ) + blake2s_update( S->S[i], block, BLAKE2S_BLOCKBYTES ); + + secure_zero_memory( block, BLAKE2S_BLOCKBYTES ); /* Burn the key from stack */ + } + return 0; +} + + +int blake2sp_update( blake2sp_state *S, const void *pin, size_t inlen ) +{ + const unsigned char * in = (const unsigned char *)pin; + size_t left = S->buflen; + size_t fill = sizeof( S->buf ) - left; + size_t i; + + if( left && inlen >= fill ) + { + memcpy( S->buf + left, in, fill ); + + for( i = 0; i < PARALLELISM_DEGREE; ++i ) + blake2s_update( S->S[i], S->buf + i * BLAKE2S_BLOCKBYTES, BLAKE2S_BLOCKBYTES ); + + in += fill; + inlen -= fill; + left = 0; + } + +#if defined(_OPENMP) + #pragma omp parallel shared(S), num_threads(PARALLELISM_DEGREE) +#else + for( i = 0; i < PARALLELISM_DEGREE; ++i ) +#endif + { +#if defined(_OPENMP) + size_t i = omp_get_thread_num(); +#endif + size_t inlen__ = inlen; + const unsigned char *in__ = ( const unsigned char * )in; + in__ += i * BLAKE2S_BLOCKBYTES; + + while( inlen__ >= PARALLELISM_DEGREE * BLAKE2S_BLOCKBYTES ) + { + blake2s_update( S->S[i], in__, BLAKE2S_BLOCKBYTES ); + in__ += PARALLELISM_DEGREE * BLAKE2S_BLOCKBYTES; + inlen__ -= PARALLELISM_DEGREE * BLAKE2S_BLOCKBYTES; + } + } + + in += inlen - inlen % ( PARALLELISM_DEGREE * BLAKE2S_BLOCKBYTES ); + inlen %= PARALLELISM_DEGREE * BLAKE2S_BLOCKBYTES; + + if( inlen > 0 ) + memcpy( S->buf + left, in, inlen ); + + S->buflen = left + inlen; + return 0; +} + + +int blake2sp_final( blake2sp_state *S, void *out, size_t outlen ) +{ + uint8_t hash[PARALLELISM_DEGREE][BLAKE2S_OUTBYTES]; + size_t i; + + if(out == NULL || outlen < S->outlen) { + return -1; + } + + for( i = 0; i < PARALLELISM_DEGREE; ++i ) + { + if( S->buflen > i * BLAKE2S_BLOCKBYTES ) + { + size_t left = S->buflen - i * BLAKE2S_BLOCKBYTES; + + if( left > BLAKE2S_BLOCKBYTES ) left = BLAKE2S_BLOCKBYTES; + + blake2s_update( S->S[i], S->buf + i * BLAKE2S_BLOCKBYTES, left ); + } + + blake2s_final( S->S[i], hash[i], BLAKE2S_OUTBYTES ); + } + + for( i = 0; i < PARALLELISM_DEGREE; ++i ) + blake2s_update( S->R, hash[i], BLAKE2S_OUTBYTES ); + + return blake2s_final( S->R, out, S->outlen ); +} + + +int blake2sp( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen ) +{ + uint8_t hash[PARALLELISM_DEGREE][BLAKE2S_OUTBYTES]; + blake2s_state S[PARALLELISM_DEGREE][1]; + blake2s_state FS[1]; + size_t i; + + /* Verify parameters */ + if ( NULL == in && inlen > 0 ) return -1; + + if ( NULL == out ) return -1; + + if ( NULL == key && keylen > 0) return -1; + + if( !outlen || outlen > BLAKE2S_OUTBYTES ) return -1; + + if( keylen > BLAKE2S_KEYBYTES ) return -1; + + for( i = 0; i < PARALLELISM_DEGREE; ++i ) + if( blake2sp_init_leaf( S[i], outlen, keylen, (uint32_t)i ) < 0 ) return -1; + + S[PARALLELISM_DEGREE - 1]->last_node = 1; /* mark last node */ + + if( keylen > 0 ) + { + uint8_t block[BLAKE2S_BLOCKBYTES]; + memset( block, 0, BLAKE2S_BLOCKBYTES ); + memcpy( block, key, keylen ); + + for( i = 0; i < PARALLELISM_DEGREE; ++i ) + blake2s_update( S[i], block, BLAKE2S_BLOCKBYTES ); + + secure_zero_memory( block, BLAKE2S_BLOCKBYTES ); /* Burn the key from stack */ + } + +#if defined(_OPENMP) + #pragma omp parallel shared(S,hash), num_threads(PARALLELISM_DEGREE) +#else + + for( i = 0; i < PARALLELISM_DEGREE; ++i ) +#endif + { +#if defined(_OPENMP) + size_t i = omp_get_thread_num(); +#endif + size_t inlen__ = inlen; + const unsigned char *in__ = ( const unsigned char * )in; + in__ += i * BLAKE2S_BLOCKBYTES; + + while( inlen__ >= PARALLELISM_DEGREE * BLAKE2S_BLOCKBYTES ) + { + blake2s_update( S[i], in__, BLAKE2S_BLOCKBYTES ); + in__ += PARALLELISM_DEGREE * BLAKE2S_BLOCKBYTES; + inlen__ -= PARALLELISM_DEGREE * BLAKE2S_BLOCKBYTES; + } + + if( inlen__ > i * BLAKE2S_BLOCKBYTES ) + { + const size_t left = inlen__ - i * BLAKE2S_BLOCKBYTES; + const size_t len = left <= BLAKE2S_BLOCKBYTES ? left : BLAKE2S_BLOCKBYTES; + blake2s_update( S[i], in__, len ); + } + + blake2s_final( S[i], hash[i], BLAKE2S_OUTBYTES ); + } + + if( blake2sp_init_root( FS, outlen, keylen ) < 0 ) + return -1; + + FS->last_node = 1; + + for( i = 0; i < PARALLELISM_DEGREE; ++i ) + blake2s_update( FS, hash[i], BLAKE2S_OUTBYTES ); + + return blake2s_final( FS, out, outlen ); +} + + + +#if defined(BLAKE2SP_SELFTEST) +#include +#include "blake2-kat.h" +int main( void ) +{ + uint8_t key[BLAKE2S_KEYBYTES]; + uint8_t buf[BLAKE2_KAT_LENGTH]; + size_t i, step; + + for( i = 0; i < BLAKE2S_KEYBYTES; ++i ) + key[i] = ( uint8_t )i; + + for( i = 0; i < BLAKE2_KAT_LENGTH; ++i ) + buf[i] = ( uint8_t )i; + + /* Test simple API */ + for( i = 0; i < BLAKE2_KAT_LENGTH; ++i ) + { + uint8_t hash[BLAKE2S_OUTBYTES]; + blake2sp( hash, BLAKE2S_OUTBYTES, buf, i, key, BLAKE2S_KEYBYTES ); + + if( 0 != memcmp( hash, blake2sp_keyed_kat[i], BLAKE2S_OUTBYTES ) ) + { + goto fail; + } + } + + /* Test streaming API */ + for(step = 1; step < BLAKE2S_BLOCKBYTES; ++step) { + for (i = 0; i < BLAKE2_KAT_LENGTH; ++i) { + uint8_t hash[BLAKE2S_OUTBYTES]; + blake2sp_state S; + uint8_t * p = buf; + size_t mlen = i; + int err = 0; + + if( (err = blake2sp_init_key(&S, BLAKE2S_OUTBYTES, key, BLAKE2S_KEYBYTES)) < 0 ) { + goto fail; + } + + while (mlen >= step) { + if ( (err = blake2sp_update(&S, p, step)) < 0 ) { + goto fail; + } + mlen -= step; + p += step; + } + if ( (err = blake2sp_update(&S, p, mlen)) < 0) { + goto fail; + } + if ( (err = blake2sp_final(&S, hash, BLAKE2S_OUTBYTES)) < 0) { + goto fail; + } + + if (0 != memcmp(hash, blake2sp_keyed_kat[i], BLAKE2S_OUTBYTES)) { + goto fail; + } + } + } + + puts( "ok" ); + return 0; +fail: + puts("error"); + return -1; +} +#endif diff --git a/c/archive_check_magic.c b/c/archive_check_magic.c new file mode 100644 --- /dev/null +++ b/c/archive_check_magic.c @@ -0,0 +1,175 @@ +/*- + * Copyright (c) 2003-2010 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_check_magic.c 201089 2009-12-28 02:20:23Z kientzle $"); + +#ifdef HAVE_SYS_TYPES_H +#include +#endif + +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif +#if defined(_WIN32) && !defined(__CYGWIN__) +#include +#include +#endif + +#include "archive_private.h" + +static void +errmsg(const char *m) +{ + size_t s = strlen(m); + ssize_t written; + + while (s > 0) { + written = write(2, m, strlen(m)); + if (written <= 0) + return; + m += written; + s -= written; + } +} + +static __LA_DEAD void +diediedie(void) +{ +#if defined(_WIN32) && !defined(__CYGWIN__) && defined(_DEBUG) + /* Cause a breakpoint exception */ + DebugBreak(); +#endif + abort(); /* Terminate the program abnormally. */ +} + +static const char * +state_name(unsigned s) +{ + switch (s) { + case ARCHIVE_STATE_NEW: return ("new"); + case ARCHIVE_STATE_HEADER: return ("header"); + case ARCHIVE_STATE_DATA: return ("data"); + case ARCHIVE_STATE_EOF: return ("eof"); + case ARCHIVE_STATE_CLOSED: return ("closed"); + case ARCHIVE_STATE_FATAL: return ("fatal"); + default: return ("??"); + } +} + +static const char * +archive_handle_type_name(unsigned m) +{ + switch (m) { + case ARCHIVE_WRITE_MAGIC: return ("archive_write"); + case ARCHIVE_READ_MAGIC: return ("archive_read"); + case ARCHIVE_WRITE_DISK_MAGIC: return ("archive_write_disk"); + case ARCHIVE_READ_DISK_MAGIC: return ("archive_read_disk"); + case ARCHIVE_MATCH_MAGIC: return ("archive_match"); + default: return NULL; + } +} + + +static char * +write_all_states(char *buff, unsigned int states) +{ + unsigned int lowbit; + + buff[0] = '\0'; + + /* A trick for computing the lowest set bit. */ + while ((lowbit = states & (1 + ~states)) != 0) { + states &= ~lowbit; /* Clear the low bit. */ + strcat(buff, state_name(lowbit)); + if (states != 0) + strcat(buff, "/"); + } + return buff; +} + +/* + * Check magic value and current state. + * Magic value mismatches are fatal and result in calls to abort(). + * State mismatches return ARCHIVE_FATAL. + * Otherwise, returns ARCHIVE_OK. + * + * This is designed to catch serious programming errors that violate + * the libarchive API. + */ +int +__archive_check_magic(struct archive *a, unsigned int magic, + unsigned int state, const char *function) +{ + char states1[64]; + char states2[64]; + const char *handle_type; + + /* + * If this isn't some form of archive handle, + * then the library user has screwed up so bad that + * we don't even have a reliable way to report an error. + */ + handle_type = archive_handle_type_name(a->magic); + + if (!handle_type) { + errmsg("PROGRAMMER ERROR: Function "); + errmsg(function); + errmsg(" invoked with invalid archive handle.\n"); + diediedie(); + } + + if (a->magic != magic) { + archive_set_error(a, -1, + "PROGRAMMER ERROR: Function '%s' invoked" + " on '%s' archive object, which is not supported.", + function, + handle_type); + a->state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + } + + if ((a->state & state) == 0) { + /* If we're already FATAL, don't overwrite the error. */ + if (a->state != ARCHIVE_STATE_FATAL) + archive_set_error(a, -1, + "INTERNAL ERROR: Function '%s' invoked with" + " archive structure in state '%s'," + " should be in state '%s'", + function, + write_all_states(states1, a->state), + write_all_states(states2, state)); + a->state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + } + return ARCHIVE_OK; +} diff --git a/c/archive_cmdline.c b/c/archive_cmdline.c new file mode 100644 --- /dev/null +++ b/c/archive_cmdline.c @@ -0,0 +1,227 @@ +/*- + * Copyright (c) 2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_STRING_H +# include +#endif +#ifdef HAVE_STDLIB_H +# include +#endif + +#include "archive.h" +#include "archive_cmdline_private.h" +#include "archive_string.h" + +static int cmdline_set_path(struct archive_cmdline *, const char *); +static int cmdline_add_arg(struct archive_cmdline *, const char *); + +static ssize_t +extract_quotation(struct archive_string *as, const char *p) +{ + const char *s; + + for (s = p + 1; *s;) { + if (*s == '\\') { + if (s[1] != '\0') { + archive_strappend_char(as, s[1]); + s += 2; + } else + s++; + } else if (*s == '"') + break; + else { + archive_strappend_char(as, s[0]); + s++; + } + } + if (*s != '"') + return (ARCHIVE_FAILED);/* Invalid sequence. */ + return ((ssize_t)(s + 1 - p)); +} + +static ssize_t +get_argument(struct archive_string *as, const char *p) +{ + const char *s = p; + + archive_string_empty(as); + + /* Skip beginning space characters. */ + while (*s != '\0' && *s == ' ') + s++; + /* Copy non-space characters. */ + while (*s != '\0' && *s != ' ') { + if (*s == '\\') { + if (s[1] != '\0') { + archive_strappend_char(as, s[1]); + s += 2; + } else { + s++;/* Ignore this character.*/ + break; + } + } else if (*s == '"') { + ssize_t q = extract_quotation(as, s); + if (q < 0) + return (ARCHIVE_FAILED);/* Invalid sequence. */ + s += q; + } else { + archive_strappend_char(as, s[0]); + s++; + } + } + return ((ssize_t)(s - p)); +} + +/* + * Set up command line arguments. + * Returns ARCHIVE_OK if everything okey. + * Returns ARCHIVE_FAILED if there is a lack of the `"' terminator or an + * empty command line. + * Returns ARCHIVE_FATAL if no memory. + */ +int +__archive_cmdline_parse(struct archive_cmdline *data, const char *cmd) +{ + struct archive_string as; + const char *p; + ssize_t al; + int r; + + archive_string_init(&as); + + /* Get first argument as a command path. */ + al = get_argument(&as, cmd); + if (al < 0) { + r = ARCHIVE_FAILED;/* Invalid sequence. */ + goto exit_function; + } + if (archive_strlen(&as) == 0) { + r = ARCHIVE_FAILED;/* An empty command path. */ + goto exit_function; + } + r = cmdline_set_path(data, as.s); + if (r != ARCHIVE_OK) + goto exit_function; + p = strrchr(as.s, '/'); + if (p == NULL) + p = as.s; + else + p++; + r = cmdline_add_arg(data, p); + if (r != ARCHIVE_OK) + goto exit_function; + cmd += al; + + for (;;) { + al = get_argument(&as, cmd); + if (al < 0) { + r = ARCHIVE_FAILED;/* Invalid sequence. */ + goto exit_function; + } + if (al == 0) + break; + cmd += al; + if (archive_strlen(&as) == 0 && *cmd == '\0') + break; + r = cmdline_add_arg(data, as.s); + if (r != ARCHIVE_OK) + goto exit_function; + } + r = ARCHIVE_OK; +exit_function: + archive_string_free(&as); + return (r); +} + +/* + * Set the program path. + */ +static int +cmdline_set_path(struct archive_cmdline *data, const char *path) +{ + char *newptr; + + newptr = realloc(data->path, strlen(path) + 1); + if (newptr == NULL) + return (ARCHIVE_FATAL); + data->path = newptr; + strcpy(data->path, path); + return (ARCHIVE_OK); +} + +/* + * Add a argument for the program. + */ +static int +cmdline_add_arg(struct archive_cmdline *data, const char *arg) +{ + char **newargv; + + if (data->path == NULL) + return (ARCHIVE_FAILED); + + newargv = realloc(data->argv, (data->argc + 2) * sizeof(char *)); + if (newargv == NULL) + return (ARCHIVE_FATAL); + data->argv = newargv; + data->argv[data->argc] = strdup(arg); + if (data->argv[data->argc] == NULL) + return (ARCHIVE_FATAL); + /* Set the terminator of argv. */ + data->argv[++data->argc] = NULL; + return (ARCHIVE_OK); +} + +struct archive_cmdline * +__archive_cmdline_allocate(void) +{ + return (struct archive_cmdline *) + calloc(1, sizeof(struct archive_cmdline)); +} + +/* + * Release the resources. + */ +int +__archive_cmdline_free(struct archive_cmdline *data) +{ + + if (data) { + free(data->path); + if (data->argv != NULL) { + int i; + for (i = 0; data->argv[i] != NULL; i++) + free(data->argv[i]); + free(data->argv); + } + free(data); + } + return (ARCHIVE_OK); +} + diff --git a/c/archive_cmdline_private.h b/c/archive_cmdline_private.h new file mode 100644 --- /dev/null +++ b/c/archive_cmdline_private.h @@ -0,0 +1,47 @@ +/*- + * Copyright (c) 2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD$ + */ + +#ifndef ARCHIVE_CMDLINE_PRIVATE_H +#define ARCHIVE_CMDLINE_PRIVATE_H + +#ifndef __LIBARCHIVE_BUILD +#ifndef __LIBARCHIVE_TEST +#error This header is only to be used internally to libarchive. +#endif +#endif + +struct archive_cmdline { + char *path; + char **argv; + int argc; +}; + +struct archive_cmdline *__archive_cmdline_allocate(void); +int __archive_cmdline_parse(struct archive_cmdline *, const char *); +int __archive_cmdline_free(struct archive_cmdline *); + +#endif diff --git a/c/archive_crc32.h b/c/archive_crc32.h new file mode 100644 --- /dev/null +++ b/c/archive_crc32.h @@ -0,0 +1,83 @@ +/*- + * Copyright (c) 2009 Joerg Sonnenberger + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD: head/lib/libarchive/archive_crc32.h 201102 2009-12-28 03:11:36Z kientzle $ + */ + +#ifndef ARCHIVE_CRC32_H +#define ARCHIVE_CRC32_H + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif + +/* + * When zlib is unavailable, we should still be able to validate + * uncompressed zip archives. That requires us to be able to compute + * the CRC32 check value. This is a drop-in compatible replacement + * for crc32() from zlib. It's slower than the zlib implementation, + * but still pretty fast: This runs about 300MB/s on my 3GHz P4 + * compared to about 800MB/s for the zlib implementation. + */ +static unsigned long +crc32(unsigned long crc, const void *_p, size_t len) +{ + unsigned long crc2, b, i; + const unsigned char *p = _p; + static volatile int crc_tbl_inited = 0; + static unsigned long crc_tbl[256]; + + if (!crc_tbl_inited) { + for (b = 0; b < 256; ++b) { + crc2 = b; + for (i = 8; i > 0; --i) { + if (crc2 & 1) + crc2 = (crc2 >> 1) ^ 0xedb88320UL; + else + crc2 = (crc2 >> 1); + } + crc_tbl[b] = crc2; + } + crc_tbl_inited = 1; + } + + crc = crc ^ 0xffffffffUL; + /* A use of this loop is about 20% - 30% faster than + * no use version in any optimization option of gcc. */ + for (;len >= 8; len -= 8) { + crc = crc_tbl[(crc ^ *p++) & 0xff] ^ (crc >> 8); + crc = crc_tbl[(crc ^ *p++) & 0xff] ^ (crc >> 8); + crc = crc_tbl[(crc ^ *p++) & 0xff] ^ (crc >> 8); + crc = crc_tbl[(crc ^ *p++) & 0xff] ^ (crc >> 8); + crc = crc_tbl[(crc ^ *p++) & 0xff] ^ (crc >> 8); + crc = crc_tbl[(crc ^ *p++) & 0xff] ^ (crc >> 8); + crc = crc_tbl[(crc ^ *p++) & 0xff] ^ (crc >> 8); + crc = crc_tbl[(crc ^ *p++) & 0xff] ^ (crc >> 8); + } + while (len--) + crc = crc_tbl[(crc ^ *p++) & 0xff] ^ (crc >> 8); + return (crc ^ 0xffffffffUL); +} + +#endif diff --git a/c/archive_cryptor.c b/c/archive_cryptor.c new file mode 100644 --- /dev/null +++ b/c/archive_cryptor.c @@ -0,0 +1,519 @@ +/*- +* Copyright (c) 2014 Michihiro NAKAJIMA +* All rights reserved. +* +* Redistribution and use in source and binary forms, with or without +* modification, are permitted provided that the following conditions +* are met: +* 1. Redistributions of source code must retain the above copyright +* notice, this list of conditions and the following disclaimer. +* 2. Redistributions in binary form must reproduce the above copyright +* notice, this list of conditions and the following disclaimer in the +* documentation and/or other materials provided with the distribution. +* +* THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR +* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES +* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. +* IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, +* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT +* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF +* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include "archive_platform.h" + +#ifdef HAVE_STRING_H +#include +#endif +#include "archive.h" +#include "archive_cryptor_private.h" + +/* + * On systems that do not support any recognized crypto libraries, + * this file will normally define no usable symbols. + * + * But some compilers and linkers choke on empty object files, so + * define a public symbol that will always exist. This could + * be removed someday if this file gains another always-present + * symbol definition. + */ +int __libarchive_cryptor_build_hack(void) { + return 0; +} + +#ifdef ARCHIVE_CRYPTOR_USE_Apple_CommonCrypto + +static int +pbkdf2_sha1(const char *pw, size_t pw_len, const uint8_t *salt, + size_t salt_len, unsigned rounds, uint8_t *derived_key, + size_t derived_key_len) +{ + CCKeyDerivationPBKDF(kCCPBKDF2, (const char *)pw, + pw_len, salt, salt_len, kCCPRFHmacAlgSHA1, rounds, + derived_key, derived_key_len); + return 0; +} + +#elif defined(_WIN32) && !defined(__CYGWIN__) && defined(HAVE_BCRYPT_H) +#ifdef _MSC_VER +#pragma comment(lib, "Bcrypt.lib") +#endif + +static int +pbkdf2_sha1(const char *pw, size_t pw_len, const uint8_t *salt, + size_t salt_len, unsigned rounds, uint8_t *derived_key, + size_t derived_key_len) +{ + NTSTATUS status; + BCRYPT_ALG_HANDLE hAlg; + + status = BCryptOpenAlgorithmProvider(&hAlg, BCRYPT_SHA1_ALGORITHM, + MS_PRIMITIVE_PROVIDER, BCRYPT_ALG_HANDLE_HMAC_FLAG); + if (!BCRYPT_SUCCESS(status)) + return -1; + + status = BCryptDeriveKeyPBKDF2(hAlg, + (PUCHAR)(uintptr_t)pw, (ULONG)pw_len, + (PUCHAR)(uintptr_t)salt, (ULONG)salt_len, rounds, + (PUCHAR)derived_key, (ULONG)derived_key_len, 0); + + BCryptCloseAlgorithmProvider(hAlg, 0); + + return (BCRYPT_SUCCESS(status)) ? 0: -1; +} + +#elif defined(HAVE_LIBMBEDCRYPTO) && defined(HAVE_MBEDTLS_PKCS5_H) + +static int +pbkdf2_sha1(const char *pw, size_t pw_len, const uint8_t *salt, + size_t salt_len, unsigned rounds, uint8_t *derived_key, + size_t derived_key_len) +{ + mbedtls_md_context_t ctx; + const mbedtls_md_info_t *info; + int ret; + + mbedtls_md_init(&ctx); + info = mbedtls_md_info_from_type(MBEDTLS_MD_SHA1); + if (info == NULL) { + mbedtls_md_free(&ctx); + return (-1); + } + ret = mbedtls_md_setup(&ctx, info, 1); + if (ret != 0) { + mbedtls_md_free(&ctx); + return (-1); + } + ret = mbedtls_pkcs5_pbkdf2_hmac(&ctx, (const unsigned char *)pw, + pw_len, salt, salt_len, rounds, derived_key_len, derived_key); + + mbedtls_md_free(&ctx); + return (ret); +} + +#elif defined(HAVE_LIBNETTLE) && defined(HAVE_NETTLE_PBKDF2_H) + +static int +pbkdf2_sha1(const char *pw, size_t pw_len, const uint8_t *salt, + size_t salt_len, unsigned rounds, uint8_t *derived_key, + size_t derived_key_len) { + pbkdf2_hmac_sha1((unsigned)pw_len, (const uint8_t *)pw, rounds, + salt_len, salt, derived_key_len, derived_key); + return 0; +} + +#elif defined(HAVE_LIBCRYPTO) && defined(HAVE_PKCS5_PBKDF2_HMAC_SHA1) + +static int +pbkdf2_sha1(const char *pw, size_t pw_len, const uint8_t *salt, + size_t salt_len, unsigned rounds, uint8_t *derived_key, + size_t derived_key_len) { + + PKCS5_PBKDF2_HMAC_SHA1(pw, pw_len, salt, salt_len, rounds, + derived_key_len, derived_key); + return 0; +} + +#else + +/* Stub */ +static int +pbkdf2_sha1(const char *pw, size_t pw_len, const uint8_t *salt, + size_t salt_len, unsigned rounds, uint8_t *derived_key, + size_t derived_key_len) { + (void)pw; /* UNUSED */ + (void)pw_len; /* UNUSED */ + (void)salt; /* UNUSED */ + (void)salt_len; /* UNUSED */ + (void)rounds; /* UNUSED */ + (void)derived_key; /* UNUSED */ + (void)derived_key_len; /* UNUSED */ + return -1; /* UNSUPPORTED */ +} + +#endif + +#ifdef ARCHIVE_CRYPTOR_USE_Apple_CommonCrypto +# if MAC_OS_X_VERSION_MAX_ALLOWED < 1090 +# define kCCAlgorithmAES kCCAlgorithmAES128 +# endif + +static int +aes_ctr_init(archive_crypto_ctx *ctx, const uint8_t *key, size_t key_len) +{ + CCCryptorStatus r; + + ctx->key_len = key_len; + memcpy(ctx->key, key, key_len); + memset(ctx->nonce, 0, sizeof(ctx->nonce)); + ctx->encr_pos = AES_BLOCK_SIZE; + r = CCCryptorCreateWithMode(kCCEncrypt, kCCModeECB, kCCAlgorithmAES, + ccNoPadding, NULL, key, key_len, NULL, 0, 0, 0, &ctx->ctx); + return (r == kCCSuccess)? 0: -1; +} + +static int +aes_ctr_encrypt_counter(archive_crypto_ctx *ctx) +{ + CCCryptorRef ref = ctx->ctx; + CCCryptorStatus r; + + r = CCCryptorReset(ref, NULL); + if (r != kCCSuccess && r != kCCUnimplemented) + return -1; + r = CCCryptorUpdate(ref, ctx->nonce, AES_BLOCK_SIZE, ctx->encr_buf, + AES_BLOCK_SIZE, NULL); + return (r == kCCSuccess)? 0: -1; +} + +static int +aes_ctr_release(archive_crypto_ctx *ctx) +{ + memset(ctx->key, 0, ctx->key_len); + memset(ctx->nonce, 0, sizeof(ctx->nonce)); + return 0; +} + +#elif defined(_WIN32) && !defined(__CYGWIN__) && defined(HAVE_BCRYPT_H) + +static int +aes_ctr_init(archive_crypto_ctx *ctx, const uint8_t *key, size_t key_len) +{ + BCRYPT_ALG_HANDLE hAlg; + BCRYPT_KEY_HANDLE hKey; + DWORD keyObj_len, aes_key_len; + PBYTE keyObj; + ULONG result; + NTSTATUS status; + BCRYPT_KEY_LENGTHS_STRUCT key_lengths; + + ctx->hAlg = NULL; + ctx->hKey = NULL; + ctx->keyObj = NULL; + switch (key_len) { + case 16: aes_key_len = 128; break; + case 24: aes_key_len = 192; break; + case 32: aes_key_len = 256; break; + default: return -1; + } + status = BCryptOpenAlgorithmProvider(&hAlg, BCRYPT_AES_ALGORITHM, + MS_PRIMITIVE_PROVIDER, 0); + if (!BCRYPT_SUCCESS(status)) + return -1; + status = BCryptGetProperty(hAlg, BCRYPT_KEY_LENGTHS, (PUCHAR)&key_lengths, + sizeof(key_lengths), &result, 0); + if (!BCRYPT_SUCCESS(status)) { + BCryptCloseAlgorithmProvider(hAlg, 0); + return -1; + } + if (key_lengths.dwMinLength > aes_key_len + || key_lengths.dwMaxLength < aes_key_len) { + BCryptCloseAlgorithmProvider(hAlg, 0); + return -1; + } + status = BCryptGetProperty(hAlg, BCRYPT_OBJECT_LENGTH, (PUCHAR)&keyObj_len, + sizeof(keyObj_len), &result, 0); + if (!BCRYPT_SUCCESS(status)) { + BCryptCloseAlgorithmProvider(hAlg, 0); + return -1; + } + keyObj = (PBYTE)HeapAlloc(GetProcessHeap(), 0, keyObj_len); + if (keyObj == NULL) { + BCryptCloseAlgorithmProvider(hAlg, 0); + return -1; + } + status = BCryptSetProperty(hAlg, BCRYPT_CHAINING_MODE, + (PUCHAR)BCRYPT_CHAIN_MODE_ECB, sizeof(BCRYPT_CHAIN_MODE_ECB), 0); + if (!BCRYPT_SUCCESS(status)) { + BCryptCloseAlgorithmProvider(hAlg, 0); + HeapFree(GetProcessHeap(), 0, keyObj); + return -1; + } + status = BCryptGenerateSymmetricKey(hAlg, &hKey, + keyObj, keyObj_len, + (PUCHAR)(uintptr_t)key, (ULONG)key_len, 0); + if (!BCRYPT_SUCCESS(status)) { + BCryptCloseAlgorithmProvider(hAlg, 0); + HeapFree(GetProcessHeap(), 0, keyObj); + return -1; + } + + ctx->hAlg = hAlg; + ctx->hKey = hKey; + ctx->keyObj = keyObj; + ctx->keyObj_len = keyObj_len; + ctx->encr_pos = AES_BLOCK_SIZE; + + return 0; +} + +static int +aes_ctr_encrypt_counter(archive_crypto_ctx *ctx) +{ + NTSTATUS status; + ULONG result; + + status = BCryptEncrypt(ctx->hKey, (PUCHAR)ctx->nonce, AES_BLOCK_SIZE, + NULL, NULL, 0, (PUCHAR)ctx->encr_buf, AES_BLOCK_SIZE, + &result, 0); + return BCRYPT_SUCCESS(status) ? 0 : -1; +} + +static int +aes_ctr_release(archive_crypto_ctx *ctx) +{ + + if (ctx->hAlg != NULL) { + BCryptCloseAlgorithmProvider(ctx->hAlg, 0); + ctx->hAlg = NULL; + BCryptDestroyKey(ctx->hKey); + ctx->hKey = NULL; + HeapFree(GetProcessHeap(), 0, ctx->keyObj); + ctx->keyObj = NULL; + } + memset(ctx, 0, sizeof(*ctx)); + return 0; +} + +#elif defined(HAVE_LIBMBEDCRYPTO) && defined(HAVE_MBEDTLS_AES_H) + +static int +aes_ctr_init(archive_crypto_ctx *ctx, const uint8_t *key, size_t key_len) +{ + mbedtls_aes_init(&ctx->ctx); + ctx->key_len = key_len; + memcpy(ctx->key, key, key_len); + memset(ctx->nonce, 0, sizeof(ctx->nonce)); + ctx->encr_pos = AES_BLOCK_SIZE; + return 0; +} + +static int +aes_ctr_encrypt_counter(archive_crypto_ctx *ctx) +{ + if (mbedtls_aes_setkey_enc(&ctx->ctx, ctx->key, + ctx->key_len * 8) != 0) + return (-1); + if (mbedtls_aes_crypt_ecb(&ctx->ctx, MBEDTLS_AES_ENCRYPT, ctx->nonce, + ctx->encr_buf) != 0) + return (-1); + return 0; +} + +static int +aes_ctr_release(archive_crypto_ctx *ctx) +{ + mbedtls_aes_free(&ctx->ctx); + memset(ctx, 0, sizeof(*ctx)); + return 0; +} + +#elif defined(HAVE_LIBNETTLE) && defined(HAVE_NETTLE_AES_H) + +static int +aes_ctr_init(archive_crypto_ctx *ctx, const uint8_t *key, size_t key_len) +{ + ctx->key_len = key_len; + memcpy(ctx->key, key, key_len); + memset(ctx->nonce, 0, sizeof(ctx->nonce)); + ctx->encr_pos = AES_BLOCK_SIZE; + memset(&ctx->ctx, 0, sizeof(ctx->ctx)); + return 0; +} + +static int +aes_ctr_encrypt_counter(archive_crypto_ctx *ctx) +{ + aes_set_encrypt_key(&ctx->ctx, ctx->key_len, ctx->key); + aes_encrypt(&ctx->ctx, AES_BLOCK_SIZE, ctx->encr_buf, ctx->nonce); + return 0; +} + +static int +aes_ctr_release(archive_crypto_ctx *ctx) +{ + memset(ctx, 0, sizeof(*ctx)); + return 0; +} + +#elif defined(HAVE_LIBCRYPTO) + +static int +aes_ctr_init(archive_crypto_ctx *ctx, const uint8_t *key, size_t key_len) +{ + if ((ctx->ctx = EVP_CIPHER_CTX_new()) == NULL) + return -1; + + switch (key_len) { + case 16: ctx->type = EVP_aes_128_ecb(); break; + case 24: ctx->type = EVP_aes_192_ecb(); break; + case 32: ctx->type = EVP_aes_256_ecb(); break; + default: ctx->type = NULL; return -1; + } + + ctx->key_len = key_len; + memcpy(ctx->key, key, key_len); + memset(ctx->nonce, 0, sizeof(ctx->nonce)); + ctx->encr_pos = AES_BLOCK_SIZE; +#if OPENSSL_VERSION_NUMBER >= 0x10100000L && !defined(LIBRESSL_VERSION_NUMBER) + if (!EVP_CIPHER_CTX_reset(ctx->ctx)) { + EVP_CIPHER_CTX_free(ctx->ctx); + ctx->ctx = NULL; + } +#else + EVP_CIPHER_CTX_init(ctx->ctx); +#endif + return 0; +} + +static int +aes_ctr_encrypt_counter(archive_crypto_ctx *ctx) +{ + int outl = 0; + int r; + + r = EVP_EncryptInit_ex(ctx->ctx, ctx->type, NULL, ctx->key, NULL); + if (r == 0) + return -1; + r = EVP_EncryptUpdate(ctx->ctx, ctx->encr_buf, &outl, ctx->nonce, + AES_BLOCK_SIZE); + if (r == 0 || outl != AES_BLOCK_SIZE) + return -1; + return 0; +} + +static int +aes_ctr_release(archive_crypto_ctx *ctx) +{ + EVP_CIPHER_CTX_free(ctx->ctx); + memset(ctx->key, 0, ctx->key_len); + memset(ctx->nonce, 0, sizeof(ctx->nonce)); + return 0; +} + +#else + +#define ARCHIVE_CRYPTOR_STUB +/* Stub */ +static int +aes_ctr_init(archive_crypto_ctx *ctx, const uint8_t *key, size_t key_len) +{ + (void)ctx; /* UNUSED */ + (void)key; /* UNUSED */ + (void)key_len; /* UNUSED */ + return -1; +} + +static int +aes_ctr_encrypt_counter(archive_crypto_ctx *ctx) +{ + (void)ctx; /* UNUSED */ + return -1; +} + +static int +aes_ctr_release(archive_crypto_ctx *ctx) +{ + (void)ctx; /* UNUSED */ + return 0; +} + +#endif + +#ifdef ARCHIVE_CRYPTOR_STUB +static int +aes_ctr_update(archive_crypto_ctx *ctx, const uint8_t * const in, + size_t in_len, uint8_t * const out, size_t *out_len) +{ + (void)ctx; /* UNUSED */ + (void)in; /* UNUSED */ + (void)in_len; /* UNUSED */ + (void)out; /* UNUSED */ + (void)out_len; /* UNUSED */ + aes_ctr_encrypt_counter(ctx); /* UNUSED */ /* Fix unused function warning */ + return -1; +} + +#else +static void +aes_ctr_increase_counter(archive_crypto_ctx *ctx) +{ + uint8_t *const nonce = ctx->nonce; + int j; + + for (j = 0; j < 8; j++) { + if (++nonce[j]) + break; + } +} + +static int +aes_ctr_update(archive_crypto_ctx *ctx, const uint8_t * const in, + size_t in_len, uint8_t * const out, size_t *out_len) +{ + uint8_t *const ebuf = ctx->encr_buf; + unsigned pos = ctx->encr_pos; + unsigned max = (unsigned)((in_len < *out_len)? in_len: *out_len); + unsigned i; + + for (i = 0; i < max; ) { + if (pos == AES_BLOCK_SIZE) { + aes_ctr_increase_counter(ctx); + if (aes_ctr_encrypt_counter(ctx) != 0) + return -1; + while (max -i >= AES_BLOCK_SIZE) { + for (pos = 0; pos < AES_BLOCK_SIZE; pos++) + out[i+pos] = in[i+pos] ^ ebuf[pos]; + i += AES_BLOCK_SIZE; + aes_ctr_increase_counter(ctx); + if (aes_ctr_encrypt_counter(ctx) != 0) + return -1; + } + pos = 0; + if (i >= max) + break; + } + out[i] = in[i] ^ ebuf[pos++]; + i++; + } + ctx->encr_pos = pos; + *out_len = i; + + return 0; +} +#endif /* ARCHIVE_CRYPTOR_STUB */ + + +const struct archive_cryptor __archive_cryptor = +{ + &pbkdf2_sha1, + &aes_ctr_init, + &aes_ctr_update, + &aes_ctr_release, + &aes_ctr_init, + &aes_ctr_update, + &aes_ctr_release, +}; diff --git a/c/archive_cryptor_private.h b/c/archive_cryptor_private.h new file mode 100644 --- /dev/null +++ b/c/archive_cryptor_private.h @@ -0,0 +1,179 @@ +/*- +* Copyright (c) 2014 Michihiro NAKAJIMA +* All rights reserved. +* +* Redistribution and use in source and binary forms, with or without +* modification, are permitted provided that the following conditions +* are met: +* 1. Redistributions of source code must retain the above copyright +* notice, this list of conditions and the following disclaimer. +* 2. Redistributions in binary form must reproduce the above copyright +* notice, this list of conditions and the following disclaimer in the +* documentation and/or other materials provided with the distribution. +* +* THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR +* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES +* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. +* IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, +* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT +* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF +* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#ifndef ARCHIVE_CRYPTOR_PRIVATE_H_INCLUDED +#define ARCHIVE_CRYPTOR_PRIVATE_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif +/* + * On systems that do not support any recognized crypto libraries, + * the archive_cryptor.c file will normally define no usable symbols. + * + * But some compilers and linkers choke on empty object files, so + * define a public symbol that will always exist. This could + * be removed someday if this file gains another always-present + * symbol definition. + */ +int __libarchive_cryptor_build_hack(void); + +#ifdef __APPLE__ +# include +# if MAC_OS_X_VERSION_MAX_ALLOWED >= 1080 +# define ARCHIVE_CRYPTOR_USE_Apple_CommonCrypto +# endif +#endif + +#ifdef ARCHIVE_CRYPTOR_USE_Apple_CommonCrypto +#include +#include +#define AES_BLOCK_SIZE 16 +#define AES_MAX_KEY_SIZE kCCKeySizeAES256 + +typedef struct { + CCCryptorRef ctx; + uint8_t key[AES_MAX_KEY_SIZE]; + unsigned key_len; + uint8_t nonce[AES_BLOCK_SIZE]; + uint8_t encr_buf[AES_BLOCK_SIZE]; + unsigned encr_pos; +} archive_crypto_ctx; + +#elif defined(_WIN32) && !defined(__CYGWIN__) && defined(HAVE_BCRYPT_H) +#include + +/* Common in other bcrypt implementations, but missing from VS2008. */ +#ifndef BCRYPT_SUCCESS +#define BCRYPT_SUCCESS(r) ((NTSTATUS)(r) == STATUS_SUCCESS) +#endif + +#define AES_MAX_KEY_SIZE 32 +#define AES_BLOCK_SIZE 16 +typedef struct { + BCRYPT_ALG_HANDLE hAlg; + BCRYPT_KEY_HANDLE hKey; + PBYTE keyObj; + DWORD keyObj_len; + uint8_t nonce[AES_BLOCK_SIZE]; + uint8_t encr_buf[AES_BLOCK_SIZE]; + unsigned encr_pos; +} archive_crypto_ctx; + +#elif defined(HAVE_LIBMBEDCRYPTO) && defined(HAVE_MBEDTLS_AES_H) +#include +#include +#include + +#define AES_MAX_KEY_SIZE 32 +#define AES_BLOCK_SIZE 16 + +typedef struct { + mbedtls_aes_context ctx; + uint8_t key[AES_MAX_KEY_SIZE]; + unsigned key_len; + uint8_t nonce[AES_BLOCK_SIZE]; + uint8_t encr_buf[AES_BLOCK_SIZE]; + unsigned encr_pos; +} archive_crypto_ctx; + +#elif defined(HAVE_LIBNETTLE) && defined(HAVE_NETTLE_AES_H) +#if defined(HAVE_NETTLE_PBKDF2_H) +#include +#endif +#include + +typedef struct { + struct aes_ctx ctx; + uint8_t key[AES_MAX_KEY_SIZE]; + unsigned key_len; + uint8_t nonce[AES_BLOCK_SIZE]; + uint8_t encr_buf[AES_BLOCK_SIZE]; + unsigned encr_pos; +} archive_crypto_ctx; + +#elif defined(HAVE_LIBCRYPTO) +#include "archive_openssl_evp_private.h" +#define AES_BLOCK_SIZE 16 +#define AES_MAX_KEY_SIZE 32 + +typedef struct { + EVP_CIPHER_CTX *ctx; + const EVP_CIPHER *type; + uint8_t key[AES_MAX_KEY_SIZE]; + unsigned key_len; + uint8_t nonce[AES_BLOCK_SIZE]; + uint8_t encr_buf[AES_BLOCK_SIZE]; + unsigned encr_pos; +} archive_crypto_ctx; + +#else + +#define AES_BLOCK_SIZE 16 +#define AES_MAX_KEY_SIZE 32 +typedef int archive_crypto_ctx; + +#endif + +/* defines */ +#define archive_pbkdf2_sha1(pw, pw_len, salt, salt_len, rounds, dk, dk_len)\ + __archive_cryptor.pbkdf2sha1(pw, pw_len, salt, salt_len, rounds, dk, dk_len) + +#define archive_decrypto_aes_ctr_init(ctx, key, key_len) \ + __archive_cryptor.decrypto_aes_ctr_init(ctx, key, key_len) +#define archive_decrypto_aes_ctr_update(ctx, in, in_len, out, out_len) \ + __archive_cryptor.decrypto_aes_ctr_update(ctx, in, in_len, out, out_len) +#define archive_decrypto_aes_ctr_release(ctx) \ + __archive_cryptor.decrypto_aes_ctr_release(ctx) + +#define archive_encrypto_aes_ctr_init(ctx, key, key_len) \ + __archive_cryptor.encrypto_aes_ctr_init(ctx, key, key_len) +#define archive_encrypto_aes_ctr_update(ctx, in, in_len, out, out_len) \ + __archive_cryptor.encrypto_aes_ctr_update(ctx, in, in_len, out, out_len) +#define archive_encrypto_aes_ctr_release(ctx) \ + __archive_cryptor.encrypto_aes_ctr_release(ctx) + +/* Minimal interface to cryptographic functionality for internal use in + * libarchive */ +struct archive_cryptor +{ + /* PKCS5 PBKDF2 HMAC-SHA1 */ + int (*pbkdf2sha1)(const char *pw, size_t pw_len, const uint8_t *salt, + size_t salt_len, unsigned rounds, uint8_t *derived_key, + size_t derived_key_len); + /* AES CTR mode(little endian version) */ + int (*decrypto_aes_ctr_init)(archive_crypto_ctx *, const uint8_t *, size_t); + int (*decrypto_aes_ctr_update)(archive_crypto_ctx *, const uint8_t *, + size_t, uint8_t *, size_t *); + int (*decrypto_aes_ctr_release)(archive_crypto_ctx *); + int (*encrypto_aes_ctr_init)(archive_crypto_ctx *, const uint8_t *, size_t); + int (*encrypto_aes_ctr_update)(archive_crypto_ctx *, const uint8_t *, + size_t, uint8_t *, size_t *); + int (*encrypto_aes_ctr_release)(archive_crypto_ctx *); +}; + +extern const struct archive_cryptor __archive_cryptor; + +#endif diff --git a/c/archive_digest.c b/c/archive_digest.c new file mode 100644 --- /dev/null +++ b/c/archive_digest.c @@ -0,0 +1,1499 @@ +/*- +* Copyright (c) 2003-2007 Tim Kientzle +* Copyright (c) 2011 Andres Mejia +* Copyright (c) 2011 Michihiro NAKAJIMA +* All rights reserved. +* +* Redistribution and use in source and binary forms, with or without +* modification, are permitted provided that the following conditions +* are met: +* 1. Redistributions of source code must retain the above copyright +* notice, this list of conditions and the following disclaimer. +* 2. Redistributions in binary form must reproduce the above copyright +* notice, this list of conditions and the following disclaimer in the +* documentation and/or other materials provided with the distribution. +* +* THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR +* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES +* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. +* IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, +* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT +* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF +* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include "archive_platform.h" + +#include "archive.h" +#include "archive_digest_private.h" + +/* In particular, force the configure probe to break if it tries + * to test a combination of OpenSSL and libmd. */ +#if defined(ARCHIVE_CRYPTO_OPENSSL) && defined(ARCHIVE_CRYPTO_LIBMD) +#error Cannot use both OpenSSL and libmd. +#endif + +/* + * Message digest functions for Windows platform. + */ +#if defined(ARCHIVE_CRYPTO_MD5_WIN) ||\ + defined(ARCHIVE_CRYPTO_SHA1_WIN) ||\ + defined(ARCHIVE_CRYPTO_SHA256_WIN) ||\ + defined(ARCHIVE_CRYPTO_SHA384_WIN) ||\ + defined(ARCHIVE_CRYPTO_SHA512_WIN) + +/* + * Initialize a Message digest. + */ +static int +win_crypto_init(Digest_CTX *ctx, ALG_ID algId) +{ + + ctx->valid = 0; + if (!CryptAcquireContext(&ctx->cryptProv, NULL, NULL, + PROV_RSA_FULL, CRYPT_VERIFYCONTEXT)) { + if (GetLastError() != (DWORD)NTE_BAD_KEYSET) + return (ARCHIVE_FAILED); + if (!CryptAcquireContext(&ctx->cryptProv, NULL, NULL, + PROV_RSA_FULL, CRYPT_NEWKEYSET)) + return (ARCHIVE_FAILED); + } + + if (!CryptCreateHash(ctx->cryptProv, algId, 0, 0, &ctx->hash)) { + CryptReleaseContext(ctx->cryptProv, 0); + return (ARCHIVE_FAILED); + } + + ctx->valid = 1; + return (ARCHIVE_OK); +} + +/* + * Update a Message digest. + */ +static int +win_crypto_Update(Digest_CTX *ctx, const unsigned char *buf, size_t len) +{ + + if (!ctx->valid) + return (ARCHIVE_FAILED); + + CryptHashData(ctx->hash, + (unsigned char *)(uintptr_t)buf, + (DWORD)len, 0); + return (ARCHIVE_OK); +} + +static int +win_crypto_Final(unsigned char *buf, size_t bufsize, Digest_CTX *ctx) +{ + DWORD siglen = (DWORD)bufsize; + + if (!ctx->valid) + return (ARCHIVE_FAILED); + + CryptGetHashParam(ctx->hash, HP_HASHVAL, buf, &siglen, 0); + CryptDestroyHash(ctx->hash); + CryptReleaseContext(ctx->cryptProv, 0); + ctx->valid = 0; + return (ARCHIVE_OK); +} + +#endif /* defined(ARCHIVE_CRYPTO_*_WIN) */ + + +/* MD5 implementations */ +#if defined(ARCHIVE_CRYPTO_MD5_LIBC) + +static int +__archive_md5init(archive_md5_ctx *ctx) +{ + MD5Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_md5update(archive_md5_ctx *ctx, const void *indata, + size_t insize) +{ + MD5Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_md5final(archive_md5_ctx *ctx, void *md) +{ + MD5Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_MD5_LIBMD) + +static int +__archive_md5init(archive_md5_ctx *ctx) +{ + MD5Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_md5update(archive_md5_ctx *ctx, const void *indata, + size_t insize) +{ + MD5Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_md5final(archive_md5_ctx *ctx, void *md) +{ + MD5Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_MD5_LIBSYSTEM) + +static int +__archive_md5init(archive_md5_ctx *ctx) +{ + CC_MD5_Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_md5update(archive_md5_ctx *ctx, const void *indata, + size_t insize) +{ + CC_MD5_Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_md5final(archive_md5_ctx *ctx, void *md) +{ + CC_MD5_Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_MD5_MBEDTLS) + +static int +__archive_md5init(archive_md5_ctx *ctx) +{ + mbedtls_md5_init(ctx); + if (mbedtls_md5_starts_ret(ctx) == 0) + return (ARCHIVE_OK); + else + return (ARCHIVE_FATAL); +} + +static int +__archive_md5update(archive_md5_ctx *ctx, const void *indata, + size_t insize) +{ + if (mbedtls_md5_update_ret(ctx, indata, insize) == 0) + return (ARCHIVE_OK); + else + return (ARCHIVE_FATAL); +} + +static int +__archive_md5final(archive_md5_ctx *ctx, void *md) +{ + if (mbedtls_md5_finish_ret(ctx, md) == 0) { + mbedtls_md5_free(ctx); + return (ARCHIVE_OK); + } else { + mbedtls_md5_free(ctx); + return (ARCHIVE_FATAL); + } +} + +#elif defined(ARCHIVE_CRYPTO_MD5_NETTLE) + +static int +__archive_md5init(archive_md5_ctx *ctx) +{ + md5_init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_md5update(archive_md5_ctx *ctx, const void *indata, + size_t insize) +{ + md5_update(ctx, insize, indata); + return (ARCHIVE_OK); +} + +static int +__archive_md5final(archive_md5_ctx *ctx, void *md) +{ + md5_digest(ctx, MD5_DIGEST_SIZE, md); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_MD5_OPENSSL) + +static int +__archive_md5init(archive_md5_ctx *ctx) +{ + if ((*ctx = EVP_MD_CTX_new()) == NULL) + return (ARCHIVE_FAILED); + EVP_DigestInit(*ctx, EVP_md5()); + return (ARCHIVE_OK); +} + +static int +__archive_md5update(archive_md5_ctx *ctx, const void *indata, + size_t insize) +{ + EVP_DigestUpdate(*ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_md5final(archive_md5_ctx *ctx, void *md) +{ + /* HACK: archive_write_set_format_xar.c is finalizing empty contexts, so + * this is meant to cope with that. Real fix is probably to fix + * archive_write_set_format_xar.c + */ + if (*ctx) { + EVP_DigestFinal(*ctx, md, NULL); + EVP_MD_CTX_free(*ctx); + *ctx = NULL; + } + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_MD5_WIN) + +static int +__archive_md5init(archive_md5_ctx *ctx) +{ + return (win_crypto_init(ctx, CALG_MD5)); +} + +static int +__archive_md5update(archive_md5_ctx *ctx, const void *indata, + size_t insize) +{ + return (win_crypto_Update(ctx, indata, insize)); +} + +static int +__archive_md5final(archive_md5_ctx *ctx, void *md) +{ + return (win_crypto_Final(md, 16, ctx)); +} + +#else + +static int +__archive_md5init(archive_md5_ctx *ctx) +{ + (void)ctx; /* UNUSED */ + return (ARCHIVE_FAILED); +} + +static int +__archive_md5update(archive_md5_ctx *ctx, const void *indata, + size_t insize) +{ + (void)ctx; /* UNUSED */ + (void)indata; /* UNUSED */ + (void)insize; /* UNUSED */ + return (ARCHIVE_FAILED); +} + +static int +__archive_md5final(archive_md5_ctx *ctx, void *md) +{ + (void)ctx; /* UNUSED */ + (void)md; /* UNUSED */ + return (ARCHIVE_FAILED); +} + +#endif + +/* RIPEMD160 implementations */ +#if defined(ARCHIVE_CRYPTO_RMD160_LIBC) + +static int +__archive_ripemd160init(archive_rmd160_ctx *ctx) +{ + RMD160Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_ripemd160update(archive_rmd160_ctx *ctx, const void *indata, + size_t insize) +{ + RMD160Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_ripemd160final(archive_rmd160_ctx *ctx, void *md) +{ + RMD160Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_RMD160_LIBMD) + +static int +__archive_ripemd160init(archive_rmd160_ctx *ctx) +{ + RIPEMD160_Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_ripemd160update(archive_rmd160_ctx *ctx, const void *indata, + size_t insize) +{ + RIPEMD160_Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_ripemd160final(archive_rmd160_ctx *ctx, void *md) +{ + RIPEMD160_Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_RMD160_MBEDTLS) + +static int +__archive_ripemd160init(archive_rmd160_ctx *ctx) +{ + mbedtls_ripemd160_init(ctx); + if (mbedtls_ripemd160_starts_ret(ctx) == 0) + return (ARCHIVE_OK); + else + return (ARCHIVE_FATAL); +} + +static int +__archive_ripemd160update(archive_rmd160_ctx *ctx, const void *indata, + size_t insize) +{ + if (mbedtls_ripemd160_update_ret(ctx, indata, insize) == 0) + return (ARCHIVE_OK); + else + return (ARCHIVE_FATAL); +} + +static int +__archive_ripemd160final(archive_rmd160_ctx *ctx, void *md) +{ + if (mbedtls_ripemd160_finish_ret(ctx, md) == 0) { + mbedtls_ripemd160_free(ctx); + return (ARCHIVE_OK); + } else { + mbedtls_ripemd160_free(ctx); + return (ARCHIVE_FATAL); + } +} + +#elif defined(ARCHIVE_CRYPTO_RMD160_NETTLE) + +static int +__archive_ripemd160init(archive_rmd160_ctx *ctx) +{ + ripemd160_init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_ripemd160update(archive_rmd160_ctx *ctx, const void *indata, + size_t insize) +{ + ripemd160_update(ctx, insize, indata); + return (ARCHIVE_OK); +} + +static int +__archive_ripemd160final(archive_rmd160_ctx *ctx, void *md) +{ + ripemd160_digest(ctx, RIPEMD160_DIGEST_SIZE, md); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_RMD160_OPENSSL) + +static int +__archive_ripemd160init(archive_rmd160_ctx *ctx) +{ + if ((*ctx = EVP_MD_CTX_new()) == NULL) + return (ARCHIVE_FAILED); + EVP_DigestInit(*ctx, EVP_ripemd160()); + return (ARCHIVE_OK); +} + +static int +__archive_ripemd160update(archive_rmd160_ctx *ctx, const void *indata, + size_t insize) +{ + EVP_DigestUpdate(*ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_ripemd160final(archive_rmd160_ctx *ctx, void *md) +{ + if (*ctx) { + EVP_DigestFinal(*ctx, md, NULL); + EVP_MD_CTX_free(*ctx); + *ctx = NULL; + } + return (ARCHIVE_OK); +} + +#else + +static int +__archive_ripemd160init(archive_rmd160_ctx *ctx) +{ + (void)ctx; /* UNUSED */ + return (ARCHIVE_FAILED); +} + +static int +__archive_ripemd160update(archive_rmd160_ctx *ctx, const void *indata, + size_t insize) +{ + (void)ctx; /* UNUSED */ + (void)indata; /* UNUSED */ + (void)insize; /* UNUSED */ + return (ARCHIVE_FAILED); +} + +static int +__archive_ripemd160final(archive_rmd160_ctx *ctx, void *md) +{ + (void)ctx; /* UNUSED */ + (void)md; /* UNUSED */ + return (ARCHIVE_FAILED); +} + +#endif + +/* SHA1 implementations */ +#if defined(ARCHIVE_CRYPTO_SHA1_LIBC) + +static int +__archive_sha1init(archive_sha1_ctx *ctx) +{ + SHA1Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha1update(archive_sha1_ctx *ctx, const void *indata, + size_t insize) +{ + SHA1Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha1final(archive_sha1_ctx *ctx, void *md) +{ + SHA1Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA1_LIBMD) + +static int +__archive_sha1init(archive_sha1_ctx *ctx) +{ + SHA1_Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha1update(archive_sha1_ctx *ctx, const void *indata, + size_t insize) +{ + SHA1_Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha1final(archive_sha1_ctx *ctx, void *md) +{ + SHA1_Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA1_LIBSYSTEM) + +static int +__archive_sha1init(archive_sha1_ctx *ctx) +{ + CC_SHA1_Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha1update(archive_sha1_ctx *ctx, const void *indata, + size_t insize) +{ + CC_SHA1_Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha1final(archive_sha1_ctx *ctx, void *md) +{ + CC_SHA1_Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA1_MBEDTLS) + +static int +__archive_sha1init(archive_sha1_ctx *ctx) +{ + mbedtls_sha1_init(ctx); + if (mbedtls_sha1_starts_ret(ctx) == 0) + return (ARCHIVE_OK); + else + return (ARCHIVE_FATAL); +} + +static int +__archive_sha1update(archive_sha1_ctx *ctx, const void *indata, + size_t insize) +{ + if (mbedtls_sha1_update_ret(ctx, indata, insize) == 0) + return (ARCHIVE_OK); + else + return (ARCHIVE_FATAL); +} + +static int +__archive_sha1final(archive_sha1_ctx *ctx, void *md) +{ + if (mbedtls_sha1_finish_ret(ctx, md) == 0) { + mbedtls_sha1_free(ctx); + return (ARCHIVE_OK); + } else { + mbedtls_sha1_free(ctx); + return (ARCHIVE_FATAL); + } +} + +#elif defined(ARCHIVE_CRYPTO_SHA1_NETTLE) + +static int +__archive_sha1init(archive_sha1_ctx *ctx) +{ + sha1_init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha1update(archive_sha1_ctx *ctx, const void *indata, + size_t insize) +{ + sha1_update(ctx, insize, indata); + return (ARCHIVE_OK); +} + +static int +__archive_sha1final(archive_sha1_ctx *ctx, void *md) +{ + sha1_digest(ctx, SHA1_DIGEST_SIZE, md); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA1_OPENSSL) + +static int +__archive_sha1init(archive_sha1_ctx *ctx) +{ + if ((*ctx = EVP_MD_CTX_new()) == NULL) + return (ARCHIVE_FAILED); + EVP_DigestInit(*ctx, EVP_sha1()); + return (ARCHIVE_OK); +} + +static int +__archive_sha1update(archive_sha1_ctx *ctx, const void *indata, + size_t insize) +{ + EVP_DigestUpdate(*ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha1final(archive_sha1_ctx *ctx, void *md) +{ + /* HACK: archive_write_set_format_xar.c is finalizing empty contexts, so + * this is meant to cope with that. Real fix is probably to fix + * archive_write_set_format_xar.c + */ + if (*ctx) { + EVP_DigestFinal(*ctx, md, NULL); + EVP_MD_CTX_free(*ctx); + *ctx = NULL; + } + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA1_WIN) + +static int +__archive_sha1init(archive_sha1_ctx *ctx) +{ + return (win_crypto_init(ctx, CALG_SHA1)); +} + +static int +__archive_sha1update(archive_sha1_ctx *ctx, const void *indata, + size_t insize) +{ + return (win_crypto_Update(ctx, indata, insize)); +} + +static int +__archive_sha1final(archive_sha1_ctx *ctx, void *md) +{ + return (win_crypto_Final(md, 20, ctx)); +} + +#else + +static int +__archive_sha1init(archive_sha1_ctx *ctx) +{ + (void)ctx; /* UNUSED */ + return (ARCHIVE_FAILED); +} + +static int +__archive_sha1update(archive_sha1_ctx *ctx, const void *indata, + size_t insize) +{ + (void)ctx; /* UNUSED */ + (void)indata; /* UNUSED */ + (void)insize; /* UNUSED */ + return (ARCHIVE_FAILED); +} + +static int +__archive_sha1final(archive_sha1_ctx *ctx, void *md) +{ + (void)ctx; /* UNUSED */ + (void)md; /* UNUSED */ + return (ARCHIVE_FAILED); +} + +#endif + +/* SHA256 implementations */ +#if defined(ARCHIVE_CRYPTO_SHA256_LIBC) + +static int +__archive_sha256init(archive_sha256_ctx *ctx) +{ + SHA256_Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha256update(archive_sha256_ctx *ctx, const void *indata, + size_t insize) +{ + SHA256_Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha256final(archive_sha256_ctx *ctx, void *md) +{ + SHA256_Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA256_LIBC2) + +static int +__archive_sha256init(archive_sha256_ctx *ctx) +{ + SHA256Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha256update(archive_sha256_ctx *ctx, const void *indata, + size_t insize) +{ + SHA256Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha256final(archive_sha256_ctx *ctx, void *md) +{ + SHA256Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA256_LIBC3) + +static int +__archive_sha256init(archive_sha256_ctx *ctx) +{ + SHA256Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha256update(archive_sha256_ctx *ctx, const void *indata, + size_t insize) +{ + SHA256Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha256final(archive_sha256_ctx *ctx, void *md) +{ + SHA256Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA256_LIBMD) + +static int +__archive_sha256init(archive_sha256_ctx *ctx) +{ + SHA256_Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha256update(archive_sha256_ctx *ctx, const void *indata, + size_t insize) +{ + SHA256_Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha256final(archive_sha256_ctx *ctx, void *md) +{ + SHA256_Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA256_LIBSYSTEM) + +static int +__archive_sha256init(archive_sha256_ctx *ctx) +{ + CC_SHA256_Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha256update(archive_sha256_ctx *ctx, const void *indata, + size_t insize) +{ + CC_SHA256_Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha256final(archive_sha256_ctx *ctx, void *md) +{ + CC_SHA256_Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA256_MBEDTLS) + +static int +__archive_sha256init(archive_sha256_ctx *ctx) +{ + mbedtls_sha256_init(ctx); + if (mbedtls_sha256_starts_ret(ctx, 0) == 0) + return (ARCHIVE_OK); + else + return (ARCHIVE_FATAL); +} + +static int +__archive_sha256update(archive_sha256_ctx *ctx, const void *indata, + size_t insize) +{ + if (mbedtls_sha256_update_ret(ctx, indata, insize) == 0) + return (ARCHIVE_OK); + else + return (ARCHIVE_FATAL); +} + +static int +__archive_sha256final(archive_sha256_ctx *ctx, void *md) +{ + if (mbedtls_sha256_finish_ret(ctx, md) == 0) { + mbedtls_sha256_free(ctx); + return (ARCHIVE_OK); + } else { + mbedtls_sha256_free(ctx); + return (ARCHIVE_FATAL); + } +} + +#elif defined(ARCHIVE_CRYPTO_SHA256_NETTLE) + +static int +__archive_sha256init(archive_sha256_ctx *ctx) +{ + sha256_init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha256update(archive_sha256_ctx *ctx, const void *indata, + size_t insize) +{ + sha256_update(ctx, insize, indata); + return (ARCHIVE_OK); +} + +static int +__archive_sha256final(archive_sha256_ctx *ctx, void *md) +{ + sha256_digest(ctx, SHA256_DIGEST_SIZE, md); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA256_OPENSSL) + +static int +__archive_sha256init(archive_sha256_ctx *ctx) +{ + if ((*ctx = EVP_MD_CTX_new()) == NULL) + return (ARCHIVE_FAILED); + EVP_DigestInit(*ctx, EVP_sha256()); + return (ARCHIVE_OK); +} + +static int +__archive_sha256update(archive_sha256_ctx *ctx, const void *indata, + size_t insize) +{ + EVP_DigestUpdate(*ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha256final(archive_sha256_ctx *ctx, void *md) +{ + if (*ctx) { + EVP_DigestFinal(*ctx, md, NULL); + EVP_MD_CTX_free(*ctx); + *ctx = NULL; + } + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA256_WIN) + +static int +__archive_sha256init(archive_sha256_ctx *ctx) +{ + return (win_crypto_init(ctx, CALG_SHA_256)); +} + +static int +__archive_sha256update(archive_sha256_ctx *ctx, const void *indata, + size_t insize) +{ + return (win_crypto_Update(ctx, indata, insize)); +} + +static int +__archive_sha256final(archive_sha256_ctx *ctx, void *md) +{ + return (win_crypto_Final(md, 32, ctx)); +} + +#else + +static int +__archive_sha256init(archive_sha256_ctx *ctx) +{ + (void)ctx; /* UNUSED */ + return (ARCHIVE_FAILED); +} + +static int +__archive_sha256update(archive_sha256_ctx *ctx, const void *indata, + size_t insize) +{ + (void)ctx; /* UNUSED */ + (void)indata; /* UNUSED */ + (void)insize; /* UNUSED */ + return (ARCHIVE_FAILED); +} + +static int +__archive_sha256final(archive_sha256_ctx *ctx, void *md) +{ + (void)ctx; /* UNUSED */ + (void)md; /* UNUSED */ + return (ARCHIVE_FAILED); +} + +#endif + +/* SHA384 implementations */ +#if defined(ARCHIVE_CRYPTO_SHA384_LIBC) + +static int +__archive_sha384init(archive_sha384_ctx *ctx) +{ + SHA384_Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha384update(archive_sha384_ctx *ctx, const void *indata, + size_t insize) +{ + SHA384_Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha384final(archive_sha384_ctx *ctx, void *md) +{ + SHA384_Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA384_LIBC2) + +static int +__archive_sha384init(archive_sha384_ctx *ctx) +{ + SHA384Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha384update(archive_sha384_ctx *ctx, const void *indata, + size_t insize) +{ + SHA384Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha384final(archive_sha384_ctx *ctx, void *md) +{ + SHA384Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA384_LIBC3) + +static int +__archive_sha384init(archive_sha384_ctx *ctx) +{ + SHA384Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha384update(archive_sha384_ctx *ctx, const void *indata, + size_t insize) +{ + SHA384Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha384final(archive_sha384_ctx *ctx, void *md) +{ + SHA384Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA384_LIBSYSTEM) + +static int +__archive_sha384init(archive_sha384_ctx *ctx) +{ + CC_SHA384_Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha384update(archive_sha384_ctx *ctx, const void *indata, + size_t insize) +{ + CC_SHA384_Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha384final(archive_sha384_ctx *ctx, void *md) +{ + CC_SHA384_Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA384_MBEDTLS) + +static int +__archive_sha384init(archive_sha384_ctx *ctx) +{ + mbedtls_sha512_init(ctx); + if (mbedtls_sha512_starts_ret(ctx, 1) == 0) + return (ARCHIVE_OK); + else + return (ARCHIVE_FATAL); +} + +static int +__archive_sha384update(archive_sha384_ctx *ctx, const void *indata, + size_t insize) +{ + if (mbedtls_sha512_update_ret(ctx, indata, insize) == 0) + return (ARCHIVE_OK); + else + return (ARCHIVE_FATAL); +} + +static int +__archive_sha384final(archive_sha384_ctx *ctx, void *md) +{ + if (mbedtls_sha512_finish_ret(ctx, md) == 0) { + mbedtls_sha512_free(ctx); + return (ARCHIVE_OK); + } else { + mbedtls_sha512_free(ctx); + return (ARCHIVE_FATAL); + } +} + +#elif defined(ARCHIVE_CRYPTO_SHA384_NETTLE) + +static int +__archive_sha384init(archive_sha384_ctx *ctx) +{ + sha384_init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha384update(archive_sha384_ctx *ctx, const void *indata, + size_t insize) +{ + sha384_update(ctx, insize, indata); + return (ARCHIVE_OK); +} + +static int +__archive_sha384final(archive_sha384_ctx *ctx, void *md) +{ + sha384_digest(ctx, SHA384_DIGEST_SIZE, md); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA384_OPENSSL) + +static int +__archive_sha384init(archive_sha384_ctx *ctx) +{ + if ((*ctx = EVP_MD_CTX_new()) == NULL) + return (ARCHIVE_FAILED); + EVP_DigestInit(*ctx, EVP_sha384()); + return (ARCHIVE_OK); +} + +static int +__archive_sha384update(archive_sha384_ctx *ctx, const void *indata, + size_t insize) +{ + EVP_DigestUpdate(*ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha384final(archive_sha384_ctx *ctx, void *md) +{ + if (*ctx) { + EVP_DigestFinal(*ctx, md, NULL); + EVP_MD_CTX_free(*ctx); + *ctx = NULL; + } + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA384_WIN) + +static int +__archive_sha384init(archive_sha384_ctx *ctx) +{ + return (win_crypto_init(ctx, CALG_SHA_384)); +} + +static int +__archive_sha384update(archive_sha384_ctx *ctx, const void *indata, + size_t insize) +{ + return (win_crypto_Update(ctx, indata, insize)); +} + +static int +__archive_sha384final(archive_sha384_ctx *ctx, void *md) +{ + return (win_crypto_Final(md, 48, ctx)); +} + +#else + +static int +__archive_sha384init(archive_sha384_ctx *ctx) +{ + (void)ctx; /* UNUSED */ + return (ARCHIVE_FAILED); +} + +static int +__archive_sha384update(archive_sha384_ctx *ctx, const void *indata, + size_t insize) +{ + (void)ctx; /* UNUSED */ + (void)indata; /* UNUSED */ + (void)insize; /* UNUSED */ + return (ARCHIVE_FAILED); +} + +static int +__archive_sha384final(archive_sha384_ctx *ctx, void *md) +{ + (void)ctx; /* UNUSED */ + (void)md; /* UNUSED */ + return (ARCHIVE_FAILED); +} + +#endif + +/* SHA512 implementations */ +#if defined(ARCHIVE_CRYPTO_SHA512_LIBC) + +static int +__archive_sha512init(archive_sha512_ctx *ctx) +{ + SHA512_Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha512update(archive_sha512_ctx *ctx, const void *indata, + size_t insize) +{ + SHA512_Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha512final(archive_sha512_ctx *ctx, void *md) +{ + SHA512_Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA512_LIBC2) + +static int +__archive_sha512init(archive_sha512_ctx *ctx) +{ + SHA512Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha512update(archive_sha512_ctx *ctx, const void *indata, + size_t insize) +{ + SHA512Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha512final(archive_sha512_ctx *ctx, void *md) +{ + SHA512Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA512_LIBC3) + +static int +__archive_sha512init(archive_sha512_ctx *ctx) +{ + SHA512Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha512update(archive_sha512_ctx *ctx, const void *indata, + size_t insize) +{ + SHA512Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha512final(archive_sha512_ctx *ctx, void *md) +{ + SHA512Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA512_LIBMD) + +static int +__archive_sha512init(archive_sha512_ctx *ctx) +{ + SHA512_Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha512update(archive_sha512_ctx *ctx, const void *indata, + size_t insize) +{ + SHA512_Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha512final(archive_sha512_ctx *ctx, void *md) +{ + SHA512_Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA512_LIBSYSTEM) + +static int +__archive_sha512init(archive_sha512_ctx *ctx) +{ + CC_SHA512_Init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha512update(archive_sha512_ctx *ctx, const void *indata, + size_t insize) +{ + CC_SHA512_Update(ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha512final(archive_sha512_ctx *ctx, void *md) +{ + CC_SHA512_Final(md, ctx); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA512_MBEDTLS) + +static int +__archive_sha512init(archive_sha512_ctx *ctx) +{ + mbedtls_sha512_init(ctx); + if (mbedtls_sha512_starts_ret(ctx, 0) == 0) + return (ARCHIVE_OK); + else + return (ARCHIVE_FATAL); +} + +static int +__archive_sha512update(archive_sha512_ctx *ctx, const void *indata, + size_t insize) +{ + if (mbedtls_sha512_update_ret(ctx, indata, insize) == 0) + return (ARCHIVE_OK); + else + return (ARCHIVE_FATAL); +} + +static int +__archive_sha512final(archive_sha512_ctx *ctx, void *md) +{ + if (mbedtls_sha512_finish_ret(ctx, md) == 0) { + mbedtls_sha512_free(ctx); + return (ARCHIVE_OK); + } else { + mbedtls_sha512_free(ctx); + return (ARCHIVE_FATAL); + } +} + +#elif defined(ARCHIVE_CRYPTO_SHA512_NETTLE) + +static int +__archive_sha512init(archive_sha512_ctx *ctx) +{ + sha512_init(ctx); + return (ARCHIVE_OK); +} + +static int +__archive_sha512update(archive_sha512_ctx *ctx, const void *indata, + size_t insize) +{ + sha512_update(ctx, insize, indata); + return (ARCHIVE_OK); +} + +static int +__archive_sha512final(archive_sha512_ctx *ctx, void *md) +{ + sha512_digest(ctx, SHA512_DIGEST_SIZE, md); + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA512_OPENSSL) + +static int +__archive_sha512init(archive_sha512_ctx *ctx) +{ + if ((*ctx = EVP_MD_CTX_new()) == NULL) + return (ARCHIVE_FAILED); + EVP_DigestInit(*ctx, EVP_sha512()); + return (ARCHIVE_OK); +} + +static int +__archive_sha512update(archive_sha512_ctx *ctx, const void *indata, + size_t insize) +{ + EVP_DigestUpdate(*ctx, indata, insize); + return (ARCHIVE_OK); +} + +static int +__archive_sha512final(archive_sha512_ctx *ctx, void *md) +{ + if (*ctx) { + EVP_DigestFinal(*ctx, md, NULL); + EVP_MD_CTX_free(*ctx); + *ctx = NULL; + } + return (ARCHIVE_OK); +} + +#elif defined(ARCHIVE_CRYPTO_SHA512_WIN) + +static int +__archive_sha512init(archive_sha512_ctx *ctx) +{ + return (win_crypto_init(ctx, CALG_SHA_512)); +} + +static int +__archive_sha512update(archive_sha512_ctx *ctx, const void *indata, + size_t insize) +{ + return (win_crypto_Update(ctx, indata, insize)); +} + +static int +__archive_sha512final(archive_sha512_ctx *ctx, void *md) +{ + return (win_crypto_Final(md, 64, ctx)); +} + +#else + +static int +__archive_sha512init(archive_sha512_ctx *ctx) +{ + (void)ctx; /* UNUSED */ + return (ARCHIVE_FAILED); +} + +static int +__archive_sha512update(archive_sha512_ctx *ctx, const void *indata, + size_t insize) +{ + (void)ctx; /* UNUSED */ + (void)indata; /* UNUSED */ + (void)insize; /* UNUSED */ + return (ARCHIVE_FAILED); +} + +static int +__archive_sha512final(archive_sha512_ctx *ctx, void *md) +{ + (void)ctx; /* UNUSED */ + (void)md; /* UNUSED */ + return (ARCHIVE_FAILED); +} + +#endif + +/* NOTE: Message Digest functions are set based on availability and by the + * following order of preference. + * 1. libc + * 2. libc2 + * 3. libc3 + * 4. libSystem + * 5. Nettle + * 6. OpenSSL + * 7. libmd + * 8. Windows API + */ +const struct archive_digest __archive_digest = +{ +/* MD5 */ + &__archive_md5init, + &__archive_md5update, + &__archive_md5final, + +/* RIPEMD160 */ + &__archive_ripemd160init, + &__archive_ripemd160update, + &__archive_ripemd160final, + +/* SHA1 */ + &__archive_sha1init, + &__archive_sha1update, + &__archive_sha1final, + +/* SHA256 */ + &__archive_sha256init, + &__archive_sha256update, + &__archive_sha256final, + +/* SHA384 */ + &__archive_sha384init, + &__archive_sha384update, + &__archive_sha384final, + +/* SHA512 */ + &__archive_sha512init, + &__archive_sha512update, + &__archive_sha512final +}; diff --git a/c/archive_digest_private.h b/c/archive_digest_private.h new file mode 100644 --- /dev/null +++ b/c/archive_digest_private.h @@ -0,0 +1,412 @@ +/*- +* Copyright (c) 2003-2007 Tim Kientzle +* Copyright (c) 2011 Andres Mejia +* All rights reserved. +* +* Redistribution and use in source and binary forms, with or without +* modification, are permitted provided that the following conditions +* are met: +* 1. Redistributions of source code must retain the above copyright +* notice, this list of conditions and the following disclaimer. +* 2. Redistributions in binary form must reproduce the above copyright +* notice, this list of conditions and the following disclaimer in the +* documentation and/or other materials provided with the distribution. +* +* THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR +* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES +* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. +* IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, +* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT +* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF +* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#ifndef ARCHIVE_DIGEST_PRIVATE_H_INCLUDED +#define ARCHIVE_DIGEST_PRIVATE_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif +/* + * Crypto support in various Operating Systems: + * + * NetBSD: + * - MD5 and SHA1 in libc: without _ after algorithm name + * - SHA2 in libc: with _ after algorithm name + * + * OpenBSD: + * - MD5, SHA1 and SHA2 in libc: without _ after algorithm name + * - OpenBSD 4.4 and earlier have SHA2 in libc with _ after algorithm name + * + * DragonFly and FreeBSD: + * - MD5 libmd: without _ after algorithm name + * - SHA1, SHA256 and SHA512 in libmd: with _ after algorithm name + * + * Mac OS X (10.4 and later): + * - MD5, SHA1 and SHA2 in libSystem: with CC_ prefix and _ after algorithm name + * + * OpenSSL: + * - MD5, SHA1 and SHA2 in libcrypto: with _ after algorithm name + * + * Windows: + * - MD5, SHA1 and SHA2 in archive_crypto.c using Windows crypto API + */ + +/* libc crypto headers */ +#if defined(ARCHIVE_CRYPTO_MD5_LIBC) +#include +#endif +#if defined(ARCHIVE_CRYPTO_RMD160_LIBC) +#include +#endif +#if defined(ARCHIVE_CRYPTO_SHA1_LIBC) +#include +#endif +#if defined(ARCHIVE_CRYPTO_SHA256_LIBC) ||\ + defined(ARCHIVE_CRYPTO_SHA256_LIBC2) ||\ + defined(ARCHIVE_CRYPTO_SHA256_LIBC3) ||\ + defined(ARCHIVE_CRYPTO_SHA384_LIBC) ||\ + defined(ARCHIVE_CRYPTO_SHA384_LIBC2) ||\ + defined(ARCHIVE_CRYPTO_SHA384_LIBC3) ||\ + defined(ARCHIVE_CRYPTO_SHA512_LIBC) ||\ + defined(ARCHIVE_CRYPTO_SHA512_LIBC2) ||\ + defined(ARCHIVE_CRYPTO_SHA512_LIBC3) +#include +#endif + +/* libmd crypto headers */ +#if defined(ARCHIVE_CRYPTO_MD5_LIBMD) ||\ + defined(ARCHIVE_CRYPTO_RMD160_LIBMD) ||\ + defined(ARCHIVE_CRYPTO_SHA1_LIBMD) ||\ + defined(ARCHIVE_CRYPTO_SHA256_LIBMD) ||\ + defined(ARCHIVE_CRYPTO_SHA512_LIBMD) +#define ARCHIVE_CRYPTO_LIBMD 1 +#endif + +#if defined(ARCHIVE_CRYPTO_MD5_LIBMD) +#include +#endif +#if defined(ARCHIVE_CRYPTO_RMD160_LIBMD) +#include +#endif +#if defined(ARCHIVE_CRYPTO_SHA1_LIBMD) +#include +#endif +#if defined(ARCHIVE_CRYPTO_SHA256_LIBMD) +#include +#endif +#if defined(ARCHIVE_CRYPTO_SHA512_LIBMD) +#include +#endif + +/* libSystem crypto headers */ +#if defined(ARCHIVE_CRYPTO_MD5_LIBSYSTEM) ||\ + defined(ARCHIVE_CRYPTO_SHA1_LIBSYSTEM) ||\ + defined(ARCHIVE_CRYPTO_SHA256_LIBSYSTEM) ||\ + defined(ARCHIVE_CRYPTO_SHA384_LIBSYSTEM) ||\ + defined(ARCHIVE_CRYPTO_SHA512_LIBSYSTEM) +#include +#endif + +/* mbed TLS crypto headers */ +#if defined(ARCHIVE_CRYPTO_MD5_MBEDTLS) +#include +#endif +#if defined(ARCHIVE_CRYPTO_RMD160_MBEDTLS) +#include +#endif +#if defined(ARCHIVE_CRYPTO_SHA1_MBEDTLS) +#include +#endif +#if defined(ARCHIVE_CRYPTO_SHA256_MBEDTLS) +#include +#endif +#if defined(ARCHIVE_CRYPTO_SHA384_MBEDTLS) ||\ + defined(ARCHIVE_CRYPTO_SHA512_MBEDTLS) +#include +#endif + +/* Nettle crypto headers */ +#if defined(ARCHIVE_CRYPTO_MD5_NETTLE) +#include +#endif +#if defined(ARCHIVE_CRYPTO_RMD160_NETTLE) +#include +#endif +#if defined(ARCHIVE_CRYPTO_SHA1_NETTLE) ||\ + defined(ARCHIVE_CRYPTO_SHA256_NETTLE) ||\ + defined(ARCHIVE_CRYPTO_SHA384_NETTLE) ||\ + defined(ARCHIVE_CRYPTO_SHA512_NETTLE) +#include +#endif + +/* OpenSSL crypto headers */ +#if defined(ARCHIVE_CRYPTO_MD5_OPENSSL) ||\ + defined(ARCHIVE_CRYPTO_RMD160_OPENSSL) ||\ + defined(ARCHIVE_CRYPTO_SHA1_OPENSSL) ||\ + defined(ARCHIVE_CRYPTO_SHA256_OPENSSL) ||\ + defined(ARCHIVE_CRYPTO_SHA384_OPENSSL) ||\ + defined(ARCHIVE_CRYPTO_SHA512_OPENSSL) +#define ARCHIVE_CRYPTO_OPENSSL 1 +#include "archive_openssl_evp_private.h" +#endif + +/* Windows crypto headers */ +#if defined(ARCHIVE_CRYPTO_MD5_WIN) ||\ + defined(ARCHIVE_CRYPTO_SHA1_WIN) ||\ + defined(ARCHIVE_CRYPTO_SHA256_WIN) ||\ + defined(ARCHIVE_CRYPTO_SHA384_WIN) ||\ + defined(ARCHIVE_CRYPTO_SHA512_WIN) +#include +#include +typedef struct { + int valid; + HCRYPTPROV cryptProv; + HCRYPTHASH hash; +} Digest_CTX; +#endif + +/* typedefs */ +#if defined(ARCHIVE_CRYPTO_MD5_LIBC) +typedef MD5_CTX archive_md5_ctx; +#elif defined(ARCHIVE_CRYPTO_MD5_LIBMD) +typedef MD5_CTX archive_md5_ctx; +#elif defined(ARCHIVE_CRYPTO_MD5_LIBSYSTEM) +typedef CC_MD5_CTX archive_md5_ctx; +#elif defined(ARCHIVE_CRYPTO_MD5_MBEDTLS) +typedef mbedtls_md5_context archive_md5_ctx; +#elif defined(ARCHIVE_CRYPTO_MD5_NETTLE) +typedef struct md5_ctx archive_md5_ctx; +#elif defined(ARCHIVE_CRYPTO_MD5_OPENSSL) +typedef EVP_MD_CTX *archive_md5_ctx; +#elif defined(ARCHIVE_CRYPTO_MD5_WIN) +typedef Digest_CTX archive_md5_ctx; +#else +typedef unsigned char archive_md5_ctx; +#endif + +#if defined(ARCHIVE_CRYPTO_RMD160_LIBC) +typedef RMD160_CTX archive_rmd160_ctx; +#elif defined(ARCHIVE_CRYPTO_RMD160_LIBMD) +typedef RIPEMD160_CTX archive_rmd160_ctx; +#elif defined(ARCHIVE_CRYPTO_RMD160_MBEDTLS) +typedef mbedtls_ripemd160_context archive_rmd160_ctx; +#elif defined(ARCHIVE_CRYPTO_RMD160_NETTLE) +typedef struct ripemd160_ctx archive_rmd160_ctx; +#elif defined(ARCHIVE_CRYPTO_RMD160_OPENSSL) +typedef EVP_MD_CTX *archive_rmd160_ctx; +#else +typedef unsigned char archive_rmd160_ctx; +#endif + +#if defined(ARCHIVE_CRYPTO_SHA1_LIBC) +typedef SHA1_CTX archive_sha1_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA1_LIBMD) +typedef SHA1_CTX archive_sha1_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA1_LIBSYSTEM) +typedef CC_SHA1_CTX archive_sha1_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA1_MBEDTLS) +typedef mbedtls_sha1_context archive_sha1_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA1_NETTLE) +typedef struct sha1_ctx archive_sha1_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA1_OPENSSL) +typedef EVP_MD_CTX *archive_sha1_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA1_WIN) +typedef Digest_CTX archive_sha1_ctx; +#else +typedef unsigned char archive_sha1_ctx; +#endif + +#if defined(ARCHIVE_CRYPTO_SHA256_LIBC) +typedef SHA256_CTX archive_sha256_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA256_LIBC2) +typedef SHA256_CTX archive_sha256_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA256_LIBC3) +typedef SHA2_CTX archive_sha256_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA256_LIBMD) +typedef SHA256_CTX archive_sha256_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA256_LIBSYSTEM) +typedef CC_SHA256_CTX archive_sha256_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA256_MBEDTLS) +typedef mbedtls_sha256_context archive_sha256_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA256_NETTLE) +typedef struct sha256_ctx archive_sha256_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA256_OPENSSL) +typedef EVP_MD_CTX *archive_sha256_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA256_WIN) +typedef Digest_CTX archive_sha256_ctx; +#else +typedef unsigned char archive_sha256_ctx; +#endif + +#if defined(ARCHIVE_CRYPTO_SHA384_LIBC) +typedef SHA384_CTX archive_sha384_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA384_LIBC2) +typedef SHA384_CTX archive_sha384_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA384_LIBC3) +typedef SHA2_CTX archive_sha384_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA384_LIBSYSTEM) +typedef CC_SHA512_CTX archive_sha384_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA384_MBEDTLS) +typedef mbedtls_sha512_context archive_sha384_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA384_NETTLE) +typedef struct sha384_ctx archive_sha384_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA384_OPENSSL) +typedef EVP_MD_CTX *archive_sha384_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA384_WIN) +typedef Digest_CTX archive_sha384_ctx; +#else +typedef unsigned char archive_sha384_ctx; +#endif + +#if defined(ARCHIVE_CRYPTO_SHA512_LIBC) +typedef SHA512_CTX archive_sha512_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA512_LIBC2) +typedef SHA512_CTX archive_sha512_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA512_LIBC3) +typedef SHA2_CTX archive_sha512_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA512_LIBMD) +typedef SHA512_CTX archive_sha512_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA512_LIBSYSTEM) +typedef CC_SHA512_CTX archive_sha512_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA512_MBEDTLS) +typedef mbedtls_sha512_context archive_sha512_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA512_NETTLE) +typedef struct sha512_ctx archive_sha512_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA512_OPENSSL) +typedef EVP_MD_CTX *archive_sha512_ctx; +#elif defined(ARCHIVE_CRYPTO_SHA512_WIN) +typedef Digest_CTX archive_sha512_ctx; +#else +typedef unsigned char archive_sha512_ctx; +#endif + +/* defines */ +#if defined(ARCHIVE_CRYPTO_MD5_LIBC) ||\ + defined(ARCHIVE_CRYPTO_MD5_LIBMD) || \ + defined(ARCHIVE_CRYPTO_MD5_LIBSYSTEM) ||\ + defined(ARCHIVE_CRYPTO_MD5_MBEDTLS) ||\ + defined(ARCHIVE_CRYPTO_MD5_NETTLE) ||\ + defined(ARCHIVE_CRYPTO_MD5_OPENSSL) ||\ + defined(ARCHIVE_CRYPTO_MD5_WIN) +#define ARCHIVE_HAS_MD5 +#endif +#define archive_md5_init(ctx)\ + __archive_digest.md5init(ctx) +#define archive_md5_final(ctx, md)\ + __archive_digest.md5final(ctx, md) +#define archive_md5_update(ctx, buf, n)\ + __archive_digest.md5update(ctx, buf, n) + +#if defined(ARCHIVE_CRYPTO_RMD160_LIBC) ||\ + defined(ARCHIVE_CRYPTO_RMD160_MBEDTLS) ||\ + defined(ARCHIVE_CRYPTO_RMD160_NETTLE) ||\ + defined(ARCHIVE_CRYPTO_RMD160_OPENSSL) +#define ARCHIVE_HAS_RMD160 +#endif +#define archive_rmd160_init(ctx)\ + __archive_digest.rmd160init(ctx) +#define archive_rmd160_final(ctx, md)\ + __archive_digest.rmd160final(ctx, md) +#define archive_rmd160_update(ctx, buf, n)\ + __archive_digest.rmd160update(ctx, buf, n) + +#if defined(ARCHIVE_CRYPTO_SHA1_LIBC) ||\ + defined(ARCHIVE_CRYPTO_SHA1_LIBMD) || \ + defined(ARCHIVE_CRYPTO_SHA1_LIBSYSTEM) ||\ + defined(ARCHIVE_CRYPTO_SHA1_MBEDTLS) ||\ + defined(ARCHIVE_CRYPTO_SHA1_NETTLE) ||\ + defined(ARCHIVE_CRYPTO_SHA1_OPENSSL) ||\ + defined(ARCHIVE_CRYPTO_SHA1_WIN) +#define ARCHIVE_HAS_SHA1 +#endif +#define archive_sha1_init(ctx)\ + __archive_digest.sha1init(ctx) +#define archive_sha1_final(ctx, md)\ + __archive_digest.sha1final(ctx, md) +#define archive_sha1_update(ctx, buf, n)\ + __archive_digest.sha1update(ctx, buf, n) + +#if defined(ARCHIVE_CRYPTO_SHA256_LIBC) ||\ + defined(ARCHIVE_CRYPTO_SHA256_LIBC2) ||\ + defined(ARCHIVE_CRYPTO_SHA256_LIBC3) ||\ + defined(ARCHIVE_CRYPTO_SHA256_LIBMD) ||\ + defined(ARCHIVE_CRYPTO_SHA256_LIBSYSTEM) ||\ + defined(ARCHIVE_CRYPTO_SHA256_MBEDTLS) ||\ + defined(ARCHIVE_CRYPTO_SHA256_NETTLE) ||\ + defined(ARCHIVE_CRYPTO_SHA256_OPENSSL) ||\ + defined(ARCHIVE_CRYPTO_SHA256_WIN) +#define ARCHIVE_HAS_SHA256 +#endif +#define archive_sha256_init(ctx)\ + __archive_digest.sha256init(ctx) +#define archive_sha256_final(ctx, md)\ + __archive_digest.sha256final(ctx, md) +#define archive_sha256_update(ctx, buf, n)\ + __archive_digest.sha256update(ctx, buf, n) + +#if defined(ARCHIVE_CRYPTO_SHA384_LIBC) ||\ + defined(ARCHIVE_CRYPTO_SHA384_LIBC2) ||\ + defined(ARCHIVE_CRYPTO_SHA384_LIBC3) ||\ + defined(ARCHIVE_CRYPTO_SHA384_LIBSYSTEM) ||\ + defined(ARCHIVE_CRYPTO_SHA384_MBEDTLS) ||\ + defined(ARCHIVE_CRYPTO_SHA384_NETTLE) ||\ + defined(ARCHIVE_CRYPTO_SHA384_OPENSSL) ||\ + defined(ARCHIVE_CRYPTO_SHA384_WIN) +#define ARCHIVE_HAS_SHA384 +#endif +#define archive_sha384_init(ctx)\ + __archive_digest.sha384init(ctx) +#define archive_sha384_final(ctx, md)\ + __archive_digest.sha384final(ctx, md) +#define archive_sha384_update(ctx, buf, n)\ + __archive_digest.sha384update(ctx, buf, n) + +#if defined(ARCHIVE_CRYPTO_SHA512_LIBC) ||\ + defined(ARCHIVE_CRYPTO_SHA512_LIBC2) ||\ + defined(ARCHIVE_CRYPTO_SHA512_LIBC3) ||\ + defined(ARCHIVE_CRYPTO_SHA512_LIBMD) ||\ + defined(ARCHIVE_CRYPTO_SHA512_LIBSYSTEM) ||\ + defined(ARCHIVE_CRYPTO_SHA512_MBEDTLS) ||\ + defined(ARCHIVE_CRYPTO_SHA512_NETTLE) ||\ + defined(ARCHIVE_CRYPTO_SHA512_OPENSSL) ||\ + defined(ARCHIVE_CRYPTO_SHA512_WIN) +#define ARCHIVE_HAS_SHA512 +#endif +#define archive_sha512_init(ctx)\ + __archive_digest.sha512init(ctx) +#define archive_sha512_final(ctx, md)\ + __archive_digest.sha512final(ctx, md) +#define archive_sha512_update(ctx, buf, n)\ + __archive_digest.sha512update(ctx, buf, n) + +/* Minimal interface to digest functionality for internal use in libarchive */ +struct archive_digest +{ + /* Message Digest */ + int (*md5init)(archive_md5_ctx *ctx); + int (*md5update)(archive_md5_ctx *, const void *, size_t); + int (*md5final)(archive_md5_ctx *, void *); + int (*rmd160init)(archive_rmd160_ctx *); + int (*rmd160update)(archive_rmd160_ctx *, const void *, size_t); + int (*rmd160final)(archive_rmd160_ctx *, void *); + int (*sha1init)(archive_sha1_ctx *); + int (*sha1update)(archive_sha1_ctx *, const void *, size_t); + int (*sha1final)(archive_sha1_ctx *, void *); + int (*sha256init)(archive_sha256_ctx *); + int (*sha256update)(archive_sha256_ctx *, const void *, size_t); + int (*sha256final)(archive_sha256_ctx *, void *); + int (*sha384init)(archive_sha384_ctx *); + int (*sha384update)(archive_sha384_ctx *, const void *, size_t); + int (*sha384final)(archive_sha384_ctx *, void *); + int (*sha512init)(archive_sha512_ctx *); + int (*sha512update)(archive_sha512_ctx *, const void *, size_t); + int (*sha512final)(archive_sha512_ctx *, void *); +}; + +extern const struct archive_digest __archive_digest; + +#endif diff --git a/c/archive_disk_acl_darwin.c b/c/archive_disk_acl_darwin.c new file mode 100644 --- /dev/null +++ b/c/archive_disk_acl_darwin.c @@ -0,0 +1,559 @@ +/*- + * Copyright (c) 2017 Martin Matuska + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +#if ARCHIVE_ACL_DARWIN + +#ifdef HAVE_FCNTL_H +#include +#endif +#if HAVE_ERRNO_H +#include +#endif +#if HAVE_MEMBERSHIP_H +#include +#endif +#ifdef HAVE_SYS_TYPES_H +#include +#endif +#ifdef HAVE_SYS_ACL_H +#define _ACL_PRIVATE /* For debugging */ +#include +#endif + +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_read_disk_private.h" +#include "archive_write_disk_private.h" + +typedef struct { + const int a_perm; /* Libarchive permission or flag */ + const int p_perm; /* Platform permission or flag */ +} acl_perm_map_t; + +static const acl_perm_map_t acl_nfs4_perm_map[] = { + {ARCHIVE_ENTRY_ACL_READ_DATA, ACL_READ_DATA}, + {ARCHIVE_ENTRY_ACL_LIST_DIRECTORY, ACL_LIST_DIRECTORY}, + {ARCHIVE_ENTRY_ACL_WRITE_DATA, ACL_WRITE_DATA}, + {ARCHIVE_ENTRY_ACL_ADD_FILE, ACL_ADD_FILE}, + {ARCHIVE_ENTRY_ACL_EXECUTE, ACL_EXECUTE}, + {ARCHIVE_ENTRY_ACL_DELETE, ACL_DELETE}, + {ARCHIVE_ENTRY_ACL_APPEND_DATA, ACL_APPEND_DATA}, + {ARCHIVE_ENTRY_ACL_ADD_SUBDIRECTORY, ACL_ADD_SUBDIRECTORY}, + {ARCHIVE_ENTRY_ACL_DELETE_CHILD, ACL_DELETE_CHILD}, + {ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES, ACL_READ_ATTRIBUTES}, + {ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES, ACL_WRITE_ATTRIBUTES}, + {ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS, ACL_READ_EXTATTRIBUTES}, + {ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS, ACL_WRITE_EXTATTRIBUTES}, + {ARCHIVE_ENTRY_ACL_READ_ACL, ACL_READ_SECURITY}, + {ARCHIVE_ENTRY_ACL_WRITE_ACL, ACL_WRITE_SECURITY}, + {ARCHIVE_ENTRY_ACL_WRITE_OWNER, ACL_CHANGE_OWNER}, +#if HAVE_DECL_ACL_SYNCHRONIZE + {ARCHIVE_ENTRY_ACL_SYNCHRONIZE, ACL_SYNCHRONIZE} +#endif +}; + +static const int acl_nfs4_perm_map_size = + (int)(sizeof(acl_nfs4_perm_map)/sizeof(acl_nfs4_perm_map[0])); + +static const acl_perm_map_t acl_nfs4_flag_map[] = { + {ARCHIVE_ENTRY_ACL_ENTRY_INHERITED, ACL_ENTRY_INHERITED}, + {ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT, ACL_ENTRY_FILE_INHERIT}, + {ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT, ACL_ENTRY_DIRECTORY_INHERIT}, + {ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT, ACL_ENTRY_LIMIT_INHERIT}, + {ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY, ACL_ENTRY_ONLY_INHERIT} +}; + +static const int acl_nfs4_flag_map_size = + (int)(sizeof(acl_nfs4_flag_map)/sizeof(acl_nfs4_flag_map[0])); + +static int translate_guid(struct archive *a, acl_entry_t acl_entry, + int *ae_id, int *ae_tag, const char **ae_name) +{ + void *q; + uid_t ugid; + int r, idtype; + + q = acl_get_qualifier(acl_entry); + if (q == NULL) + return (1); + r = mbr_uuid_to_id((const unsigned char *)q, &ugid, &idtype); + if (r != 0) { + acl_free(q); + return (1); + } + if (idtype == ID_TYPE_UID) { + *ae_tag = ARCHIVE_ENTRY_ACL_USER; + *ae_id = ugid; + *ae_name = archive_read_disk_uname(a, *ae_id); + } else if (idtype == ID_TYPE_GID) { + *ae_tag = ARCHIVE_ENTRY_ACL_GROUP; + *ae_id = ugid; + *ae_name = archive_read_disk_gname(a, *ae_id); + } else + r = 1; + + acl_free(q); + return (r); +} + +static void +add_trivial_nfs4_acl(struct archive_entry *entry) +{ + mode_t mode; + int i; + const int rperm = ARCHIVE_ENTRY_ACL_READ_DATA; + const int wperm = ARCHIVE_ENTRY_ACL_WRITE_DATA | + ARCHIVE_ENTRY_ACL_APPEND_DATA; + const int eperm = ARCHIVE_ENTRY_ACL_EXECUTE; + const int pubset = ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES | + ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS | + ARCHIVE_ENTRY_ACL_READ_ACL | + ARCHIVE_ENTRY_ACL_SYNCHRONIZE; + const int ownset = pubset | ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES | + ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS | + ARCHIVE_ENTRY_ACL_WRITE_ACL | + ARCHIVE_ENTRY_ACL_WRITE_OWNER; + + struct { + const int type; + const int tag; + int permset; + } tacl_entry[] = { + {ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_USER_OBJ, 0}, + {ARCHIVE_ENTRY_ACL_TYPE_DENY, ARCHIVE_ENTRY_ACL_USER_OBJ, 0}, + {ARCHIVE_ENTRY_ACL_TYPE_DENY, ARCHIVE_ENTRY_ACL_GROUP_OBJ, 0}, + {ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_USER_OBJ, ownset}, + {ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_GROUP_OBJ, pubset}, + {ARCHIVE_ENTRY_ACL_TYPE_ALLOW, ARCHIVE_ENTRY_ACL_EVERYONE, pubset} + }; + + mode = archive_entry_mode(entry); + + /* Permissions for everyone@ */ + if (mode & 0004) + tacl_entry[5].permset |= rperm; + if (mode & 0002) + tacl_entry[5].permset |= wperm; + if (mode & 0001) + tacl_entry[5].permset |= eperm; + + /* Permissions for group@ */ + if (mode & 0040) + tacl_entry[4].permset |= rperm; + else if (mode & 0004) + tacl_entry[2].permset |= rperm; + if (mode & 0020) + tacl_entry[4].permset |= wperm; + else if (mode & 0002) + tacl_entry[2].permset |= wperm; + if (mode & 0010) + tacl_entry[4].permset |= eperm; + else if (mode & 0001) + tacl_entry[2].permset |= eperm; + + /* Permissions for owner@ */ + if (mode & 0400) { + tacl_entry[3].permset |= rperm; + if (!(mode & 0040) && (mode & 0004)) + tacl_entry[0].permset |= rperm; + } else if ((mode & 0040) || (mode & 0004)) + tacl_entry[1].permset |= rperm; + if (mode & 0200) { + tacl_entry[3].permset |= wperm; + if (!(mode & 0020) && (mode & 0002)) + tacl_entry[0].permset |= wperm; + } else if ((mode & 0020) || (mode & 0002)) + tacl_entry[1].permset |= wperm; + if (mode & 0100) { + tacl_entry[3].permset |= eperm; + if (!(mode & 0010) && (mode & 0001)) + tacl_entry[0].permset |= eperm; + } else if ((mode & 0010) || (mode & 0001)) + tacl_entry[1].permset |= eperm; + + for (i = 0; i < 6; i++) { + if (tacl_entry[i].permset != 0) { + archive_entry_acl_add_entry(entry, + tacl_entry[i].type, tacl_entry[i].permset, + tacl_entry[i].tag, -1, NULL); + } + } + + return; +} + +static int +translate_acl(struct archive_read_disk *a, + struct archive_entry *entry, acl_t acl) +{ + acl_tag_t acl_tag; + acl_flagset_t acl_flagset; + acl_entry_t acl_entry; + acl_permset_t acl_permset; + int i, entry_acl_type; + int r, s, ae_id, ae_tag, ae_perm; + const char *ae_name; + + s = acl_get_entry(acl, ACL_FIRST_ENTRY, &acl_entry); + if (s == -1) { + archive_set_error(&a->archive, errno, + "Failed to get first ACL entry"); + return (ARCHIVE_WARN); + } + + while (s == 0) { + ae_id = -1; + ae_name = NULL; + ae_perm = 0; + + if (acl_get_tag_type(acl_entry, &acl_tag) != 0) { + archive_set_error(&a->archive, errno, + "Failed to get ACL tag type"); + return (ARCHIVE_WARN); + } + switch (acl_tag) { + case ACL_EXTENDED_ALLOW: + entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_ALLOW; + r = translate_guid(&a->archive, acl_entry, + &ae_id, &ae_tag, &ae_name); + break; + case ACL_EXTENDED_DENY: + entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_DENY; + r = translate_guid(&a->archive, acl_entry, + &ae_id, &ae_tag, &ae_name); + break; + default: + /* Skip types that libarchive can't support. */ + s = acl_get_entry(acl, ACL_NEXT_ENTRY, &acl_entry); + continue; + } + + /* Skip if translate_guid() above failed */ + if (r != 0) { + s = acl_get_entry(acl, ACL_NEXT_ENTRY, &acl_entry); + continue; + } + + /* + * Libarchive stores "flag" (NFSv4 inheritance bits) + * in the ae_perm bitmap. + * + * acl_get_flagset_np() fails with non-NFSv4 ACLs + */ + if (acl_get_flagset_np(acl_entry, &acl_flagset) != 0) { + archive_set_error(&a->archive, errno, + "Failed to get flagset from a NFSv4 ACL entry"); + return (ARCHIVE_WARN); + } + for (i = 0; i < acl_nfs4_flag_map_size; ++i) { + r = acl_get_flag_np(acl_flagset, + acl_nfs4_flag_map[i].p_perm); + if (r == -1) { + archive_set_error(&a->archive, errno, + "Failed to check flag in a NFSv4 " + "ACL flagset"); + return (ARCHIVE_WARN); + } else if (r) + ae_perm |= acl_nfs4_flag_map[i].a_perm; + } + + if (acl_get_permset(acl_entry, &acl_permset) != 0) { + archive_set_error(&a->archive, errno, + "Failed to get ACL permission set"); + return (ARCHIVE_WARN); + } + + for (i = 0; i < acl_nfs4_perm_map_size; ++i) { + /* + * acl_get_perm() is spelled differently on different + * platforms; see above. + */ + r = acl_get_perm_np(acl_permset, + acl_nfs4_perm_map[i].p_perm); + if (r == -1) { + archive_set_error(&a->archive, errno, + "Failed to check permission in an ACL " + "permission set"); + return (ARCHIVE_WARN); + } else if (r) + ae_perm |= acl_nfs4_perm_map[i].a_perm; + } + +#if !HAVE_DECL_ACL_SYNCHRONIZE + /* On Mac OS X without ACL_SYNCHRONIZE assume it is set */ + ae_perm |= ARCHIVE_ENTRY_ACL_SYNCHRONIZE; +#endif + + archive_entry_acl_add_entry(entry, entry_acl_type, + ae_perm, ae_tag, + ae_id, ae_name); + + s = acl_get_entry(acl, ACL_NEXT_ENTRY, &acl_entry); + } + return (ARCHIVE_OK); +} + +static int +set_acl(struct archive *a, int fd, const char *name, + struct archive_acl *abstract_acl, + int ae_requested_type, const char *tname) +{ + acl_t acl; + acl_entry_t acl_entry; + acl_permset_t acl_permset; + acl_flagset_t acl_flagset; + int ret; + int ae_type, ae_permset, ae_tag, ae_id; + uuid_t ae_uuid; + uid_t ae_uid; + gid_t ae_gid; + const char *ae_name; + int entries; + int i; + + ret = ARCHIVE_OK; + entries = archive_acl_reset(abstract_acl, ae_requested_type); + if (entries == 0) + return (ARCHIVE_OK); + + if (ae_requested_type != ARCHIVE_ENTRY_ACL_TYPE_NFS4) { + errno = ENOENT; + archive_set_error(a, errno, "Unsupported ACL type"); + return (ARCHIVE_FAILED); + } + + acl = acl_init(entries); + if (acl == (acl_t)NULL) { + archive_set_error(a, errno, + "Failed to initialize ACL working storage"); + return (ARCHIVE_FAILED); + } + + while (archive_acl_next(a, abstract_acl, ae_requested_type, &ae_type, + &ae_permset, &ae_tag, &ae_id, &ae_name) == ARCHIVE_OK) { + /* + * Mac OS doesn't support NFSv4 ACLs for + * owner@, group@ and everyone@. + * We skip any of these ACLs found. + */ + if (ae_tag == ARCHIVE_ENTRY_ACL_USER_OBJ || + ae_tag == ARCHIVE_ENTRY_ACL_GROUP_OBJ || + ae_tag == ARCHIVE_ENTRY_ACL_EVERYONE) + continue; + + if (acl_create_entry(&acl, &acl_entry) != 0) { + archive_set_error(a, errno, + "Failed to create a new ACL entry"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + + switch (ae_type) { + case ARCHIVE_ENTRY_ACL_TYPE_ALLOW: + acl_set_tag_type(acl_entry, ACL_EXTENDED_ALLOW); + break; + case ARCHIVE_ENTRY_ACL_TYPE_DENY: + acl_set_tag_type(acl_entry, ACL_EXTENDED_DENY); + break; + default: + /* We don't support any other types on MacOS */ + continue; + } + + switch (ae_tag) { + case ARCHIVE_ENTRY_ACL_USER: + ae_uid = archive_write_disk_uid(a, ae_name, ae_id); + if (mbr_uid_to_uuid(ae_uid, ae_uuid) != 0) + continue; + if (acl_set_qualifier(acl_entry, &ae_uuid) != 0) + continue; + break; + case ARCHIVE_ENTRY_ACL_GROUP: + ae_gid = archive_write_disk_gid(a, ae_name, ae_id); + if (mbr_gid_to_uuid(ae_gid, ae_uuid) != 0) + continue; + if (acl_set_qualifier(acl_entry, &ae_uuid) != 0) + continue; + break; + default: + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Unsupported ACL tag"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + + if (acl_get_permset(acl_entry, &acl_permset) != 0) { + archive_set_error(a, errno, + "Failed to get ACL permission set"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + if (acl_clear_perms(acl_permset) != 0) { + archive_set_error(a, errno, + "Failed to clear ACL permissions"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + + for (i = 0; i < acl_nfs4_perm_map_size; ++i) { + if (ae_permset & acl_nfs4_perm_map[i].a_perm) { + if (acl_add_perm(acl_permset, + acl_nfs4_perm_map[i].p_perm) != 0) { + archive_set_error(a, errno, + "Failed to add ACL permission"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + } + } + + /* + * acl_get_flagset_np() fails with non-NFSv4 ACLs + */ + if (acl_get_flagset_np(acl_entry, &acl_flagset) != 0) { + archive_set_error(a, errno, + "Failed to get flagset from an NFSv4 ACL entry"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + if (acl_clear_flags_np(acl_flagset) != 0) { + archive_set_error(a, errno, + "Failed to clear flags from an NFSv4 ACL flagset"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + + for (i = 0; i < acl_nfs4_flag_map_size; ++i) { + if (ae_permset & acl_nfs4_flag_map[i].a_perm) { + if (acl_add_flag_np(acl_flagset, + acl_nfs4_flag_map[i].p_perm) != 0) { + archive_set_error(a, errno, + "Failed to add flag to " + "NFSv4 ACL flagset"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + } + } + } + + if (fd >= 0) { + if (acl_set_fd_np(fd, acl, ACL_TYPE_EXTENDED) == 0) + ret = ARCHIVE_OK; + else { + if (errno == EOPNOTSUPP) { + /* Filesystem doesn't support ACLs */ + ret = ARCHIVE_OK; + } else { + archive_set_error(a, errno, + "Failed to set acl on fd: %s", tname); + ret = ARCHIVE_WARN; + } + } + } else if (acl_set_link_np(name, ACL_TYPE_EXTENDED, acl) != 0) { + if (errno == EOPNOTSUPP) { + /* Filesystem doesn't support ACLs */ + ret = ARCHIVE_OK; + } else { + archive_set_error(a, errno, "Failed to set acl: %s", + tname); + ret = ARCHIVE_WARN; + } + } +exit_free: + acl_free(acl); + return (ret); +} + +int +archive_read_disk_entry_setup_acls(struct archive_read_disk *a, + struct archive_entry *entry, int *fd) +{ + const char *accpath; + acl_t acl; + int r; + + accpath = NULL; + + if (*fd < 0) { + accpath = archive_read_disk_entry_setup_path(a, entry, fd); + if (accpath == NULL) + return (ARCHIVE_WARN); + } + + archive_entry_acl_clear(entry); + + acl = NULL; + + if (*fd >= 0) + acl = acl_get_fd_np(*fd, ACL_TYPE_EXTENDED); + else if (!a->follow_symlinks) + acl = acl_get_link_np(accpath, ACL_TYPE_EXTENDED); + else + acl = acl_get_file(accpath, ACL_TYPE_EXTENDED); + + if (acl != NULL) { + r = translate_acl(a, entry, acl); + acl_free(acl); + acl = NULL; + + if (r != ARCHIVE_OK) { + archive_set_error(&a->archive, errno, + "Couldn't translate NFSv4 ACLs"); + } + + /* + * Because Mac OS doesn't support owner@, group@ and everyone@ + * ACLs we need to add NFSv4 ACLs mirroring the file mode to + * the archive entry. Otherwise extraction on non-Mac platforms + * would lead to an invalid file mode. + */ + if ((archive_entry_acl_types(entry) & + ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) + add_trivial_nfs4_acl(entry); + + return (r); + } + return (ARCHIVE_OK); +} + +int +archive_write_disk_set_acls(struct archive *a, int fd, const char *name, + struct archive_acl *abstract_acl, __LA_MODE_T mode) +{ + int ret = ARCHIVE_OK; + + (void)mode; /* UNUSED */ + + if ((archive_acl_types(abstract_acl) & + ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { + ret = set_acl(a, fd, name, abstract_acl, + ARCHIVE_ENTRY_ACL_TYPE_NFS4, "nfs4"); + } + return (ret); +} +#endif /* ARCHIVE_ACL_DARWIN */ diff --git a/c/archive_disk_acl_freebsd.c b/c/archive_disk_acl_freebsd.c new file mode 100644 --- /dev/null +++ b/c/archive_disk_acl_freebsd.c @@ -0,0 +1,702 @@ +/*- + * Copyright (c) 2003-2009 Tim Kientzle + * Copyright (c) 2010-2012 Michihiro NAKAJIMA + * Copyright (c) 2017 Martin Matuska + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +#if ARCHIVE_ACL_FREEBSD + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_FCNTL_H +#include +#endif +#ifdef HAVE_SYS_TYPES_H +#include +#endif +#ifdef HAVE_SYS_ACL_H +#define _ACL_PRIVATE /* For debugging */ +#include +#endif + +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_read_disk_private.h" +#include "archive_write_disk_private.h" + +typedef struct { + const int a_perm; /* Libarchive permission or flag */ + const int p_perm; /* Platform permission or flag */ +} acl_perm_map_t; + +static const acl_perm_map_t acl_posix_perm_map[] = { + {ARCHIVE_ENTRY_ACL_EXECUTE, ACL_EXECUTE}, + {ARCHIVE_ENTRY_ACL_WRITE, ACL_WRITE}, + {ARCHIVE_ENTRY_ACL_READ, ACL_READ}, +}; + +static const int acl_posix_perm_map_size = + (int)(sizeof(acl_posix_perm_map)/sizeof(acl_posix_perm_map[0])); + +#if ARCHIVE_ACL_FREEBSD_NFS4 +static const acl_perm_map_t acl_nfs4_perm_map[] = { + {ARCHIVE_ENTRY_ACL_EXECUTE, ACL_EXECUTE}, + {ARCHIVE_ENTRY_ACL_READ_DATA, ACL_READ_DATA}, + {ARCHIVE_ENTRY_ACL_LIST_DIRECTORY, ACL_LIST_DIRECTORY}, + {ARCHIVE_ENTRY_ACL_WRITE_DATA, ACL_WRITE_DATA}, + {ARCHIVE_ENTRY_ACL_ADD_FILE, ACL_ADD_FILE}, + {ARCHIVE_ENTRY_ACL_APPEND_DATA, ACL_APPEND_DATA}, + {ARCHIVE_ENTRY_ACL_ADD_SUBDIRECTORY, ACL_ADD_SUBDIRECTORY}, + {ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS, ACL_READ_NAMED_ATTRS}, + {ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS, ACL_WRITE_NAMED_ATTRS}, + {ARCHIVE_ENTRY_ACL_DELETE_CHILD, ACL_DELETE_CHILD}, + {ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES, ACL_READ_ATTRIBUTES}, + {ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES, ACL_WRITE_ATTRIBUTES}, + {ARCHIVE_ENTRY_ACL_DELETE, ACL_DELETE}, + {ARCHIVE_ENTRY_ACL_READ_ACL, ACL_READ_ACL}, + {ARCHIVE_ENTRY_ACL_WRITE_ACL, ACL_WRITE_ACL}, + {ARCHIVE_ENTRY_ACL_WRITE_OWNER, ACL_WRITE_OWNER}, + {ARCHIVE_ENTRY_ACL_SYNCHRONIZE, ACL_SYNCHRONIZE} +}; + +static const int acl_nfs4_perm_map_size = + (int)(sizeof(acl_nfs4_perm_map)/sizeof(acl_nfs4_perm_map[0])); + +static const acl_perm_map_t acl_nfs4_flag_map[] = { + {ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT, ACL_ENTRY_FILE_INHERIT}, + {ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT, ACL_ENTRY_DIRECTORY_INHERIT}, + {ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT, ACL_ENTRY_NO_PROPAGATE_INHERIT}, + {ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY, ACL_ENTRY_INHERIT_ONLY}, + {ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS, ACL_ENTRY_SUCCESSFUL_ACCESS}, + {ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS, ACL_ENTRY_FAILED_ACCESS}, +#ifdef ACL_ENTRY_INHERITED + {ARCHIVE_ENTRY_ACL_ENTRY_INHERITED, ACL_ENTRY_INHERITED} +#endif +}; + +static const int acl_nfs4_flag_map_size = + (int)(sizeof(acl_nfs4_flag_map)/sizeof(acl_nfs4_flag_map[0])); +#endif /* ARCHIVE_ACL_FREEBSD_NFS4 */ + +static int +translate_acl(struct archive_read_disk *a, + struct archive_entry *entry, acl_t acl, int default_entry_acl_type) +{ +#if ARCHIVE_ACL_FREEBSD_NFS4 + int brand; + acl_flagset_t acl_flagset; + acl_entry_type_t acl_type; +#endif + acl_tag_t acl_tag; + acl_entry_t acl_entry; + acl_permset_t acl_permset; + int i, entry_acl_type, perm_map_size; + const acl_perm_map_t *perm_map; + int r, s, ae_id, ae_tag, ae_perm; + void *q; + const char *ae_name; + +#if ARCHIVE_ACL_FREEBSD_NFS4 + // FreeBSD "brands" ACLs as POSIX.1e or NFSv4 + // Make sure the "brand" on this ACL is consistent + // with the default_entry_acl_type bits provided. + if (acl_get_brand_np(acl, &brand) != 0) { + archive_set_error(&a->archive, errno, + "Failed to read ACL brand"); + return (ARCHIVE_WARN); + } + switch (brand) { + case ACL_BRAND_POSIX: + switch (default_entry_acl_type) { + case ARCHIVE_ENTRY_ACL_TYPE_ACCESS: + case ARCHIVE_ENTRY_ACL_TYPE_DEFAULT: + break; + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Invalid ACL entry type for POSIX.1e ACL"); + return (ARCHIVE_WARN); + } + break; + case ACL_BRAND_NFS4: + if (default_entry_acl_type & ~ARCHIVE_ENTRY_ACL_TYPE_NFS4) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Invalid ACL entry type for NFSv4 ACL"); + return (ARCHIVE_WARN); + } + break; + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Unknown ACL brand"); + return (ARCHIVE_WARN); + } +#endif + + s = acl_get_entry(acl, ACL_FIRST_ENTRY, &acl_entry); + if (s == -1) { + archive_set_error(&a->archive, errno, + "Failed to get first ACL entry"); + return (ARCHIVE_WARN); + } + + while (s == 1) { + ae_id = -1; + ae_name = NULL; + ae_perm = 0; + + if (acl_get_tag_type(acl_entry, &acl_tag) != 0) { + archive_set_error(&a->archive, errno, + "Failed to get ACL tag type"); + return (ARCHIVE_WARN); + } + switch (acl_tag) { + case ACL_USER: + q = acl_get_qualifier(acl_entry); + if (q != NULL) { + ae_id = (int)*(uid_t *)q; + acl_free(q); + ae_name = archive_read_disk_uname(&a->archive, + ae_id); + } + ae_tag = ARCHIVE_ENTRY_ACL_USER; + break; + case ACL_GROUP: + q = acl_get_qualifier(acl_entry); + if (q != NULL) { + ae_id = (int)*(gid_t *)q; + acl_free(q); + ae_name = archive_read_disk_gname(&a->archive, + ae_id); + } + ae_tag = ARCHIVE_ENTRY_ACL_GROUP; + break; + case ACL_MASK: + ae_tag = ARCHIVE_ENTRY_ACL_MASK; + break; + case ACL_USER_OBJ: + ae_tag = ARCHIVE_ENTRY_ACL_USER_OBJ; + break; + case ACL_GROUP_OBJ: + ae_tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; + break; + case ACL_OTHER: + ae_tag = ARCHIVE_ENTRY_ACL_OTHER; + break; +#if ARCHIVE_ACL_FREEBSD_NFS4 + case ACL_EVERYONE: + ae_tag = ARCHIVE_ENTRY_ACL_EVERYONE; + break; +#endif + default: + /* Skip types that libarchive can't support. */ + s = acl_get_entry(acl, ACL_NEXT_ENTRY, &acl_entry); + continue; + } + + // XXX acl_type maps to allow/deny/audit/YYYY bits + entry_acl_type = default_entry_acl_type; + +#if ARCHIVE_ACL_FREEBSD_NFS4 + if (default_entry_acl_type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) { + /* + * acl_get_entry_type_np() fails with non-NFSv4 ACLs + */ + if (acl_get_entry_type_np(acl_entry, &acl_type) != 0) { + archive_set_error(&a->archive, errno, "Failed " + "to get ACL type from a NFSv4 ACL entry"); + return (ARCHIVE_WARN); + } + switch (acl_type) { + case ACL_ENTRY_TYPE_ALLOW: + entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_ALLOW; + break; + case ACL_ENTRY_TYPE_DENY: + entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_DENY; + break; + case ACL_ENTRY_TYPE_AUDIT: + entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_AUDIT; + break; + case ACL_ENTRY_TYPE_ALARM: + entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_ALARM; + break; + default: + archive_set_error(&a->archive, errno, + "Invalid NFSv4 ACL entry type"); + return (ARCHIVE_WARN); + } + + /* + * Libarchive stores "flag" (NFSv4 inheritance bits) + * in the ae_perm bitmap. + * + * acl_get_flagset_np() fails with non-NFSv4 ACLs + */ + if (acl_get_flagset_np(acl_entry, &acl_flagset) != 0) { + archive_set_error(&a->archive, errno, + "Failed to get flagset from a NFSv4 " + "ACL entry"); + return (ARCHIVE_WARN); + } + for (i = 0; i < acl_nfs4_flag_map_size; ++i) { + r = acl_get_flag_np(acl_flagset, + acl_nfs4_flag_map[i].p_perm); + if (r == -1) { + archive_set_error(&a->archive, errno, + "Failed to check flag in a NFSv4 " + "ACL flagset"); + return (ARCHIVE_WARN); + } else if (r) + ae_perm |= acl_nfs4_flag_map[i].a_perm; + } + } +#endif + + if (acl_get_permset(acl_entry, &acl_permset) != 0) { + archive_set_error(&a->archive, errno, + "Failed to get ACL permission set"); + return (ARCHIVE_WARN); + } + +#if ARCHIVE_ACL_FREEBSD_NFS4 + if (default_entry_acl_type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) { + perm_map_size = acl_nfs4_perm_map_size; + perm_map = acl_nfs4_perm_map; + } else { +#endif + perm_map_size = acl_posix_perm_map_size; + perm_map = acl_posix_perm_map; +#if ARCHIVE_ACL_FREEBSD_NFS4 + } +#endif + + for (i = 0; i < perm_map_size; ++i) { + r = acl_get_perm_np(acl_permset, perm_map[i].p_perm); + if (r == -1) { + archive_set_error(&a->archive, errno, + "Failed to check permission in an ACL " + "permission set"); + return (ARCHIVE_WARN); + } else if (r) + ae_perm |= perm_map[i].a_perm; + } + + archive_entry_acl_add_entry(entry, entry_acl_type, + ae_perm, ae_tag, + ae_id, ae_name); + + s = acl_get_entry(acl, ACL_NEXT_ENTRY, &acl_entry); + if (s == -1) { + archive_set_error(&a->archive, errno, + "Failed to get next ACL entry"); + return (ARCHIVE_WARN); + } + } + return (ARCHIVE_OK); +} + +static int +set_acl(struct archive *a, int fd, const char *name, + struct archive_acl *abstract_acl, + int ae_requested_type, const char *tname) +{ + int acl_type = 0; + acl_t acl; + acl_entry_t acl_entry; + acl_permset_t acl_permset; +#if ARCHIVE_ACL_FREEBSD_NFS4 + acl_flagset_t acl_flagset; + int r; +#endif + int ret; + int ae_type, ae_permset, ae_tag, ae_id; + int perm_map_size; + const acl_perm_map_t *perm_map; + uid_t ae_uid; + gid_t ae_gid; + const char *ae_name; + int entries; + int i; + + ret = ARCHIVE_OK; + entries = archive_acl_reset(abstract_acl, ae_requested_type); + if (entries == 0) + return (ARCHIVE_OK); + + + switch (ae_requested_type) { + case ARCHIVE_ENTRY_ACL_TYPE_ACCESS: + acl_type = ACL_TYPE_ACCESS; + break; + case ARCHIVE_ENTRY_ACL_TYPE_DEFAULT: + acl_type = ACL_TYPE_DEFAULT; + break; +#if ARCHIVE_ACL_FREEBSD_NFS4 + case ARCHIVE_ENTRY_ACL_TYPE_NFS4: + acl_type = ACL_TYPE_NFS4; + break; +#endif + default: + errno = ENOENT; + archive_set_error(a, errno, "Unsupported ACL type"); + return (ARCHIVE_FAILED); + } + + acl = acl_init(entries); + if (acl == (acl_t)NULL) { + archive_set_error(a, errno, + "Failed to initialize ACL working storage"); + return (ARCHIVE_FAILED); + } + + while (archive_acl_next(a, abstract_acl, ae_requested_type, &ae_type, + &ae_permset, &ae_tag, &ae_id, &ae_name) == ARCHIVE_OK) { + if (acl_create_entry(&acl, &acl_entry) != 0) { + archive_set_error(a, errno, + "Failed to create a new ACL entry"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + switch (ae_tag) { + case ARCHIVE_ENTRY_ACL_USER: + ae_uid = archive_write_disk_uid(a, ae_name, ae_id); + acl_set_tag_type(acl_entry, ACL_USER); + acl_set_qualifier(acl_entry, &ae_uid); + break; + case ARCHIVE_ENTRY_ACL_GROUP: + ae_gid = archive_write_disk_gid(a, ae_name, ae_id); + acl_set_tag_type(acl_entry, ACL_GROUP); + acl_set_qualifier(acl_entry, &ae_gid); + break; + case ARCHIVE_ENTRY_ACL_USER_OBJ: + acl_set_tag_type(acl_entry, ACL_USER_OBJ); + break; + case ARCHIVE_ENTRY_ACL_GROUP_OBJ: + acl_set_tag_type(acl_entry, ACL_GROUP_OBJ); + break; + case ARCHIVE_ENTRY_ACL_MASK: + acl_set_tag_type(acl_entry, ACL_MASK); + break; + case ARCHIVE_ENTRY_ACL_OTHER: + acl_set_tag_type(acl_entry, ACL_OTHER); + break; +#if ARCHIVE_ACL_FREEBSD_NFS4 + case ARCHIVE_ENTRY_ACL_EVERYONE: + acl_set_tag_type(acl_entry, ACL_EVERYONE); + break; +#endif + default: + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Unsupported ACL tag"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + +#if ARCHIVE_ACL_FREEBSD_NFS4 + r = 0; + switch (ae_type) { + case ARCHIVE_ENTRY_ACL_TYPE_ALLOW: + r = acl_set_entry_type_np(acl_entry, + ACL_ENTRY_TYPE_ALLOW); + break; + case ARCHIVE_ENTRY_ACL_TYPE_DENY: + r = acl_set_entry_type_np(acl_entry, + ACL_ENTRY_TYPE_DENY); + break; + case ARCHIVE_ENTRY_ACL_TYPE_AUDIT: + r = acl_set_entry_type_np(acl_entry, + ACL_ENTRY_TYPE_AUDIT); + break; + case ARCHIVE_ENTRY_ACL_TYPE_ALARM: + r = acl_set_entry_type_np(acl_entry, + ACL_ENTRY_TYPE_ALARM); + break; + case ARCHIVE_ENTRY_ACL_TYPE_ACCESS: + case ARCHIVE_ENTRY_ACL_TYPE_DEFAULT: + // These don't translate directly into the system ACL. + break; + default: + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Unsupported ACL entry type"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + + if (r != 0) { + archive_set_error(a, errno, + "Failed to set ACL entry type"); + ret = ARCHIVE_FAILED; + goto exit_free; + } +#endif + + if (acl_get_permset(acl_entry, &acl_permset) != 0) { + archive_set_error(a, errno, + "Failed to get ACL permission set"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + if (acl_clear_perms(acl_permset) != 0) { + archive_set_error(a, errno, + "Failed to clear ACL permissions"); + ret = ARCHIVE_FAILED; + goto exit_free; + } +#if ARCHIVE_ACL_FREEBSD_NFS4 + if (ae_requested_type == ARCHIVE_ENTRY_ACL_TYPE_NFS4) { + perm_map_size = acl_nfs4_perm_map_size; + perm_map = acl_nfs4_perm_map; + } else { +#endif + perm_map_size = acl_posix_perm_map_size; + perm_map = acl_posix_perm_map; +#if ARCHIVE_ACL_FREEBSD_NFS4 + } +#endif + + for (i = 0; i < perm_map_size; ++i) { + if (ae_permset & perm_map[i].a_perm) { + if (acl_add_perm(acl_permset, + perm_map[i].p_perm) != 0) { + archive_set_error(a, errno, + "Failed to add ACL permission"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + } + } + +#if ARCHIVE_ACL_FREEBSD_NFS4 + if (ae_requested_type == ARCHIVE_ENTRY_ACL_TYPE_NFS4) { + /* + * acl_get_flagset_np() fails with non-NFSv4 ACLs + */ + if (acl_get_flagset_np(acl_entry, &acl_flagset) != 0) { + archive_set_error(a, errno, + "Failed to get flagset from an NFSv4 " + "ACL entry"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + if (acl_clear_flags_np(acl_flagset) != 0) { + archive_set_error(a, errno, + "Failed to clear flags from an NFSv4 " + "ACL flagset"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + for (i = 0; i < acl_nfs4_flag_map_size; ++i) { + if (ae_permset & acl_nfs4_flag_map[i].a_perm) { + if (acl_add_flag_np(acl_flagset, + acl_nfs4_flag_map[i].p_perm) != 0) { + archive_set_error(a, errno, + "Failed to add flag to " + "NFSv4 ACL flagset"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + } + } + } +#endif + } + + /* Try restoring the ACL through 'fd' if we can. */ + if (fd >= 0) { + if (acl_set_fd_np(fd, acl, acl_type) == 0) + ret = ARCHIVE_OK; + else { + if (errno == EOPNOTSUPP) { + /* Filesystem doesn't support ACLs */ + ret = ARCHIVE_OK; + } else { + archive_set_error(a, errno, + "Failed to set acl on fd: %s", tname); + ret = ARCHIVE_WARN; + } + } + } +#if HAVE_ACL_SET_LINK_NP + else if (acl_set_link_np(name, acl_type, acl) != 0) +#else + /* FreeBSD older than 8.0 */ + else if (acl_set_file(name, acl_type, acl) != 0) +#endif + { + if (errno == EOPNOTSUPP) { + /* Filesystem doesn't support ACLs */ + ret = ARCHIVE_OK; + } else { + archive_set_error(a, errno, "Failed to set acl: %s", + tname); + ret = ARCHIVE_WARN; + } + } +exit_free: + acl_free(acl); + return (ret); +} + +int +archive_read_disk_entry_setup_acls(struct archive_read_disk *a, + struct archive_entry *entry, int *fd) +{ + const char *accpath; + acl_t acl; + int r; + + accpath = NULL; + + if (*fd < 0) { + accpath = archive_read_disk_entry_setup_path(a, entry, fd); + if (accpath == NULL) + return (ARCHIVE_WARN); + } + + archive_entry_acl_clear(entry); + + acl = NULL; + +#if ARCHIVE_ACL_FREEBSD_NFS4 + /* Try NFSv4 ACL first. */ + if (*fd >= 0) + acl = acl_get_fd_np(*fd, ACL_TYPE_NFS4); + else if (!a->follow_symlinks) + acl = acl_get_link_np(accpath, ACL_TYPE_NFS4); + else + acl = acl_get_file(accpath, ACL_TYPE_NFS4); + + /* Ignore "trivial" ACLs that just mirror the file mode. */ + if (acl != NULL && acl_is_trivial_np(acl, &r) == 0 && r == 1) { + acl_free(acl); + acl = NULL; + return (ARCHIVE_OK); + } + + if (acl != NULL) { + r = translate_acl(a, entry, acl, ARCHIVE_ENTRY_ACL_TYPE_NFS4); + acl_free(acl); + acl = NULL; + + if (r != ARCHIVE_OK) { + archive_set_error(&a->archive, errno, + "Couldn't translate NFSv4 ACLs"); + } + + return (r); + } +#endif + + /* Retrieve access ACL from file. */ + if (*fd >= 0) + acl = acl_get_fd_np(*fd, ACL_TYPE_ACCESS); +#if HAVE_ACL_GET_LINK_NP + else if (!a->follow_symlinks) + acl = acl_get_link_np(accpath, ACL_TYPE_ACCESS); +#else + else if ((!a->follow_symlinks) + && (archive_entry_filetype(entry) == AE_IFLNK)) + /* We can't get the ACL of a symlink, so we assume it can't + have one. */ + acl = NULL; +#endif + else + acl = acl_get_file(accpath, ACL_TYPE_ACCESS); + +#if HAVE_ACL_IS_TRIVIAL_NP + /* Ignore "trivial" ACLs that just mirror the file mode. */ + if (acl != NULL && acl_is_trivial_np(acl, &r) == 0 && r == 1) { + acl_free(acl); + acl = NULL; + } +#endif + + if (acl != NULL) { + r = translate_acl(a, entry, acl, ARCHIVE_ENTRY_ACL_TYPE_ACCESS); + acl_free(acl); + acl = NULL; + + if (r != ARCHIVE_OK) { + archive_set_error(&a->archive, errno, + "Couldn't translate access ACLs"); + return (r); + } + } + + /* Only directories can have default ACLs. */ + if (S_ISDIR(archive_entry_mode(entry))) { + if (*fd >= 0) + acl = acl_get_fd_np(*fd, ACL_TYPE_DEFAULT); + else + acl = acl_get_file(accpath, ACL_TYPE_DEFAULT); + if (acl != NULL) { + r = translate_acl(a, entry, acl, + ARCHIVE_ENTRY_ACL_TYPE_DEFAULT); + acl_free(acl); + if (r != ARCHIVE_OK) { + archive_set_error(&a->archive, errno, + "Couldn't translate default ACLs"); + return (r); + } + } + } + return (ARCHIVE_OK); +} + +int +archive_write_disk_set_acls(struct archive *a, int fd, const char *name, + struct archive_acl *abstract_acl, __LA_MODE_T mode) +{ + int ret = ARCHIVE_OK; + + (void)mode; /* UNUSED */ + + if ((archive_acl_types(abstract_acl) + & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) { + if ((archive_acl_types(abstract_acl) + & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) { + ret = set_acl(a, fd, name, abstract_acl, + ARCHIVE_ENTRY_ACL_TYPE_ACCESS, "access"); + if (ret != ARCHIVE_OK) + return (ret); + } + if ((archive_acl_types(abstract_acl) + & ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) != 0) + ret = set_acl(a, fd, name, abstract_acl, + ARCHIVE_ENTRY_ACL_TYPE_DEFAULT, "default"); + + /* Simultaneous POSIX.1e and NFSv4 is not supported */ + return (ret); + } +#if ARCHIVE_ACL_FREEBSD_NFS4 + else if ((archive_acl_types(abstract_acl) & + ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { + ret = set_acl(a, fd, name, abstract_acl, + ARCHIVE_ENTRY_ACL_TYPE_NFS4, "nfs4"); + } +#endif + return (ret); +} +#endif /* ARCHIVE_ACL_FREEBSD */ diff --git a/c/archive_disk_acl_linux.c b/c/archive_disk_acl_linux.c new file mode 100644 --- /dev/null +++ b/c/archive_disk_acl_linux.c @@ -0,0 +1,743 @@ +/*- + * Copyright (c) 2003-2009 Tim Kientzle + * Copyright (c) 2010-2012 Michihiro NAKAJIMA + * Copyright (c) 2017 Martin Matuska + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +#if ARCHIVE_ACL_LIBACL || ARCHIVE_ACL_LIBRICHACL + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_FCNTL_H +#include +#endif +#if HAVE_ACL_LIBACL_H +#include +#endif +#ifdef HAVE_SYS_ACL_H +#include +#endif +#ifdef HAVE_SYS_RICHACL_H +#include +#endif + +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_read_disk_private.h" +#include "archive_write_disk_private.h" + +typedef struct { + const int a_perm; /* Libarchive permission or flag */ + const int p_perm; /* Platform permission or flag */ +} acl_perm_map_t; + +#if ARCHIVE_ACL_LIBACL +static const acl_perm_map_t acl_posix_perm_map[] = { + {ARCHIVE_ENTRY_ACL_EXECUTE, ACL_EXECUTE}, + {ARCHIVE_ENTRY_ACL_WRITE, ACL_WRITE}, + {ARCHIVE_ENTRY_ACL_READ, ACL_READ}, +}; + +static const int acl_posix_perm_map_size = + (int)(sizeof(acl_posix_perm_map)/sizeof(acl_posix_perm_map[0])); +#endif /* ARCHIVE_ACL_LIBACL */ + +#if ARCHIVE_ACL_LIBRICHACL +static const acl_perm_map_t acl_nfs4_perm_map[] = { + {ARCHIVE_ENTRY_ACL_EXECUTE, RICHACE_EXECUTE}, + {ARCHIVE_ENTRY_ACL_READ_DATA, RICHACE_READ_DATA}, + {ARCHIVE_ENTRY_ACL_LIST_DIRECTORY, RICHACE_LIST_DIRECTORY}, + {ARCHIVE_ENTRY_ACL_WRITE_DATA, RICHACE_WRITE_DATA}, + {ARCHIVE_ENTRY_ACL_ADD_FILE, RICHACE_ADD_FILE}, + {ARCHIVE_ENTRY_ACL_APPEND_DATA, RICHACE_APPEND_DATA}, + {ARCHIVE_ENTRY_ACL_ADD_SUBDIRECTORY, RICHACE_ADD_SUBDIRECTORY}, + {ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS, RICHACE_READ_NAMED_ATTRS}, + {ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS, RICHACE_WRITE_NAMED_ATTRS}, + {ARCHIVE_ENTRY_ACL_DELETE_CHILD, RICHACE_DELETE_CHILD}, + {ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES, RICHACE_READ_ATTRIBUTES}, + {ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES, RICHACE_WRITE_ATTRIBUTES}, + {ARCHIVE_ENTRY_ACL_DELETE, RICHACE_DELETE}, + {ARCHIVE_ENTRY_ACL_READ_ACL, RICHACE_READ_ACL}, + {ARCHIVE_ENTRY_ACL_WRITE_ACL, RICHACE_WRITE_ACL}, + {ARCHIVE_ENTRY_ACL_WRITE_OWNER, RICHACE_WRITE_OWNER}, + {ARCHIVE_ENTRY_ACL_SYNCHRONIZE, RICHACE_SYNCHRONIZE} +}; + +static const int acl_nfs4_perm_map_size = + (int)(sizeof(acl_nfs4_perm_map)/sizeof(acl_nfs4_perm_map[0])); + +static const acl_perm_map_t acl_nfs4_flag_map[] = { + {ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT, RICHACE_FILE_INHERIT_ACE}, + {ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT, RICHACE_DIRECTORY_INHERIT_ACE}, + {ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT, RICHACE_NO_PROPAGATE_INHERIT_ACE}, + {ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY, RICHACE_INHERIT_ONLY_ACE}, + {ARCHIVE_ENTRY_ACL_ENTRY_INHERITED, RICHACE_INHERITED_ACE} +}; + +static const int acl_nfs4_flag_map_size = + (int)(sizeof(acl_nfs4_flag_map)/sizeof(acl_nfs4_flag_map[0])); +#endif /* ARCHIVE_ACL_LIBRICHACL */ + +#if ARCHIVE_ACL_LIBACL +/* + * Translate POSIX.1e ACLs into libarchive internal structure + */ +static int +translate_acl(struct archive_read_disk *a, + struct archive_entry *entry, acl_t acl, int default_entry_acl_type) +{ + acl_tag_t acl_tag; + acl_entry_t acl_entry; + acl_permset_t acl_permset; + int i, entry_acl_type; + int r, s, ae_id, ae_tag, ae_perm; + void *q; + const char *ae_name; + + s = acl_get_entry(acl, ACL_FIRST_ENTRY, &acl_entry); + if (s == -1) { + archive_set_error(&a->archive, errno, + "Failed to get first ACL entry"); + return (ARCHIVE_WARN); + } + + while (s == 1) { + ae_id = -1; + ae_name = NULL; + ae_perm = 0; + + if (acl_get_tag_type(acl_entry, &acl_tag) != 0) { + archive_set_error(&a->archive, errno, + "Failed to get ACL tag type"); + return (ARCHIVE_WARN); + } + switch (acl_tag) { + case ACL_USER: + q = acl_get_qualifier(acl_entry); + if (q != NULL) { + ae_id = (int)*(uid_t *)q; + acl_free(q); + ae_name = archive_read_disk_uname(&a->archive, + ae_id); + } + ae_tag = ARCHIVE_ENTRY_ACL_USER; + break; + case ACL_GROUP: + q = acl_get_qualifier(acl_entry); + if (q != NULL) { + ae_id = (int)*(gid_t *)q; + acl_free(q); + ae_name = archive_read_disk_gname(&a->archive, + ae_id); + } + ae_tag = ARCHIVE_ENTRY_ACL_GROUP; + break; + case ACL_MASK: + ae_tag = ARCHIVE_ENTRY_ACL_MASK; + break; + case ACL_USER_OBJ: + ae_tag = ARCHIVE_ENTRY_ACL_USER_OBJ; + break; + case ACL_GROUP_OBJ: + ae_tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; + break; + case ACL_OTHER: + ae_tag = ARCHIVE_ENTRY_ACL_OTHER; + break; + default: + /* Skip types that libarchive can't support. */ + s = acl_get_entry(acl, ACL_NEXT_ENTRY, &acl_entry); + continue; + } + + // XXX acl_type maps to allow/deny/audit/YYYY bits + entry_acl_type = default_entry_acl_type; + + if (acl_get_permset(acl_entry, &acl_permset) != 0) { + archive_set_error(&a->archive, errno, + "Failed to get ACL permission set"); + return (ARCHIVE_WARN); + } + + for (i = 0; i < acl_posix_perm_map_size; ++i) { + r = acl_get_perm(acl_permset, + acl_posix_perm_map[i].p_perm); + if (r == -1) { + archive_set_error(&a->archive, errno, + "Failed to check permission in an ACL " + "permission set"); + return (ARCHIVE_WARN); + } else if (r) + ae_perm |= acl_posix_perm_map[i].a_perm; + } + + archive_entry_acl_add_entry(entry, entry_acl_type, + ae_perm, ae_tag, + ae_id, ae_name); + + s = acl_get_entry(acl, ACL_NEXT_ENTRY, &acl_entry); + if (s == -1) { + archive_set_error(&a->archive, errno, + "Failed to get next ACL entry"); + return (ARCHIVE_WARN); + } + } + return (ARCHIVE_OK); +} +#endif /* ARCHIVE_ACL_LIBACL */ + +#if ARCHIVE_ACL_LIBRICHACL +/* + * Translate RichACL into libarchive internal ACL + */ +static int +translate_richacl(struct archive_read_disk *a, struct archive_entry *entry, + struct richacl *richacl) +{ + int ae_id, ae_tag, ae_perm; + int entry_acl_type, i; + const char *ae_name; + + struct richace *richace; + + richacl_for_each_entry(richace, richacl) { + ae_name = NULL; + ae_tag = 0; + ae_perm = 0; + ae_id = -1; + + switch (richace->e_type) { + case RICHACE_ACCESS_ALLOWED_ACE_TYPE: + entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_ALLOW; + break; + case RICHACE_ACCESS_DENIED_ACE_TYPE: + entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_DENY; + break; + default: /* Unknown entry type, skip */ + continue; + } + + /* Unsupported */ + if (richace->e_flags & RICHACE_UNMAPPED_WHO) + continue; + + if (richace->e_flags & RICHACE_SPECIAL_WHO) { + switch (richace->e_id) { + case RICHACE_OWNER_SPECIAL_ID: + ae_tag = ARCHIVE_ENTRY_ACL_USER_OBJ; + break; + case RICHACE_GROUP_SPECIAL_ID: + ae_tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; + break; + case RICHACE_EVERYONE_SPECIAL_ID: + ae_tag = ARCHIVE_ENTRY_ACL_EVERYONE; + break; + default: /* Unknown special ID type */ + continue; + } + } else { + ae_id = richace->e_id; + if (richace->e_flags & RICHACE_IDENTIFIER_GROUP) { + ae_tag = ARCHIVE_ENTRY_ACL_GROUP; + ae_name = archive_read_disk_gname(&a->archive, + (gid_t)(richace->e_id)); + } else { + ae_tag = ARCHIVE_ENTRY_ACL_USER; + ae_name = archive_read_disk_uname(&a->archive, + (uid_t)(richace->e_id)); + } + } + for (i = 0; i < acl_nfs4_flag_map_size; ++i) { + if ((richace->e_flags & + acl_nfs4_flag_map[i].p_perm) != 0) + ae_perm |= acl_nfs4_flag_map[i].a_perm; + } + for (i = 0; i < acl_nfs4_perm_map_size; ++i) { + if ((richace->e_mask & + acl_nfs4_perm_map[i].p_perm) != 0) + ae_perm |= + acl_nfs4_perm_map[i].a_perm; + } + + archive_entry_acl_add_entry(entry, entry_acl_type, + ae_perm, ae_tag, ae_id, ae_name); + } + return (ARCHIVE_OK); +} +#endif /* ARCHIVE_ACL_LIBRICHACL */ + +#if ARCHIVE_ACL_LIBRICHACL +static int +_richacl_mode_to_mask(short mode) +{ + int mask = 0; + + if (mode & S_IROTH) + mask |= RICHACE_POSIX_MODE_READ; + if (mode & S_IWOTH) + mask |= RICHACE_POSIX_MODE_WRITE; + if (mode & S_IXOTH) + mask |= RICHACE_POSIX_MODE_EXEC; + + return (mask); +} + +static void +_richacl_mode_to_masks(struct richacl *richacl, __LA_MODE_T mode) +{ + richacl->a_owner_mask = _richacl_mode_to_mask((mode & 0700) >> 6); + richacl->a_group_mask = _richacl_mode_to_mask((mode & 0070) >> 3); + richacl->a_other_mask = _richacl_mode_to_mask(mode & 0007); +} +#endif /* ARCHIVE_ACL_LIBRICHACL */ + +#if ARCHIVE_ACL_LIBRICHACL +static int +set_richacl(struct archive *a, int fd, const char *name, + struct archive_acl *abstract_acl, __LA_MODE_T mode, + int ae_requested_type, const char *tname) +{ + int ae_type, ae_permset, ae_tag, ae_id; + uid_t ae_uid; + gid_t ae_gid; + const char *ae_name; + int entries; + int i; + int ret; + int e = 0; + struct richacl *richacl = NULL; + struct richace *richace; + + ret = ARCHIVE_OK; + entries = archive_acl_reset(abstract_acl, ae_requested_type); + if (entries == 0) + return (ARCHIVE_OK); + + if (ae_requested_type != ARCHIVE_ENTRY_ACL_TYPE_NFS4) { + errno = ENOENT; + archive_set_error(a, errno, "Unsupported ACL type"); + return (ARCHIVE_FAILED); + } + + richacl = richacl_alloc(entries); + if (richacl == NULL) { + archive_set_error(a, errno, + "Failed to initialize RichACL working storage"); + return (ARCHIVE_FAILED); + } + + e = 0; + + while (archive_acl_next(a, abstract_acl, ae_requested_type, &ae_type, + &ae_permset, &ae_tag, &ae_id, &ae_name) == ARCHIVE_OK) { + richace = &(richacl->a_entries[e]); + + richace->e_flags = 0; + richace->e_mask = 0; + + switch (ae_tag) { + case ARCHIVE_ENTRY_ACL_USER: + ae_uid = archive_write_disk_uid(a, ae_name, ae_id); + richace->e_id = ae_uid; + break; + case ARCHIVE_ENTRY_ACL_GROUP: + ae_gid = archive_write_disk_gid(a, ae_name, ae_id); + richace->e_id = ae_gid; + richace->e_flags |= RICHACE_IDENTIFIER_GROUP; + break; + case ARCHIVE_ENTRY_ACL_USER_OBJ: + richace->e_flags |= RICHACE_SPECIAL_WHO; + richace->e_id = RICHACE_OWNER_SPECIAL_ID; + break; + case ARCHIVE_ENTRY_ACL_GROUP_OBJ: + richace->e_flags |= RICHACE_SPECIAL_WHO; + richace->e_id = RICHACE_GROUP_SPECIAL_ID; + break; + case ARCHIVE_ENTRY_ACL_EVERYONE: + richace->e_flags |= RICHACE_SPECIAL_WHO; + richace->e_id = RICHACE_EVERYONE_SPECIAL_ID; + break; + default: + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Unsupported ACL tag"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + + switch (ae_type) { + case ARCHIVE_ENTRY_ACL_TYPE_ALLOW: + richace->e_type = + RICHACE_ACCESS_ALLOWED_ACE_TYPE; + break; + case ARCHIVE_ENTRY_ACL_TYPE_DENY: + richace->e_type = + RICHACE_ACCESS_DENIED_ACE_TYPE; + break; + case ARCHIVE_ENTRY_ACL_TYPE_AUDIT: + case ARCHIVE_ENTRY_ACL_TYPE_ALARM: + break; + default: + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Unsupported ACL entry type"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + + for (i = 0; i < acl_nfs4_perm_map_size; ++i) { + if (ae_permset & acl_nfs4_perm_map[i].a_perm) + richace->e_mask |= acl_nfs4_perm_map[i].p_perm; + } + + for (i = 0; i < acl_nfs4_flag_map_size; ++i) { + if (ae_permset & + acl_nfs4_flag_map[i].a_perm) + richace->e_flags |= acl_nfs4_flag_map[i].p_perm; + } + e++; + } + + /* Fill RichACL masks */ + _richacl_mode_to_masks(richacl, mode); + + if (fd >= 0) { + if (richacl_set_fd(fd, richacl) == 0) + ret = ARCHIVE_OK; + else { + if (errno == EOPNOTSUPP) { + /* Filesystem doesn't support ACLs */ + ret = ARCHIVE_OK; + } else { + archive_set_error(a, errno, + "Failed to set richacl on fd: %s", tname); + ret = ARCHIVE_WARN; + } + } + } else if (richacl_set_file(name, richacl) != 0) { + if (errno == EOPNOTSUPP) { + /* Filesystem doesn't support ACLs */ + ret = ARCHIVE_OK; + } else { + archive_set_error(a, errno, "Failed to set richacl: %s", + tname); + ret = ARCHIVE_WARN; + } + } +exit_free: + richacl_free(richacl); + return (ret); +} +#endif /* ARCHIVE_ACL_RICHACL */ + +#if ARCHIVE_ACL_LIBACL +static int +set_acl(struct archive *a, int fd, const char *name, + struct archive_acl *abstract_acl, + int ae_requested_type, const char *tname) +{ + int acl_type = 0; + int ae_type, ae_permset, ae_tag, ae_id; + uid_t ae_uid; + gid_t ae_gid; + const char *ae_name; + int entries; + int i; + int ret; + acl_t acl = NULL; + acl_entry_t acl_entry; + acl_permset_t acl_permset; + + ret = ARCHIVE_OK; + entries = archive_acl_reset(abstract_acl, ae_requested_type); + if (entries == 0) + return (ARCHIVE_OK); + + switch (ae_requested_type) { + case ARCHIVE_ENTRY_ACL_TYPE_ACCESS: + acl_type = ACL_TYPE_ACCESS; + break; + case ARCHIVE_ENTRY_ACL_TYPE_DEFAULT: + acl_type = ACL_TYPE_DEFAULT; + break; + default: + errno = ENOENT; + archive_set_error(a, errno, "Unsupported ACL type"); + return (ARCHIVE_FAILED); + } + + acl = acl_init(entries); + if (acl == (acl_t)NULL) { + archive_set_error(a, errno, + "Failed to initialize ACL working storage"); + return (ARCHIVE_FAILED); + } + + while (archive_acl_next(a, abstract_acl, ae_requested_type, &ae_type, + &ae_permset, &ae_tag, &ae_id, &ae_name) == ARCHIVE_OK) { + + if (acl_create_entry(&acl, &acl_entry) != 0) { + archive_set_error(a, errno, + "Failed to create a new ACL entry"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + + switch (ae_tag) { + case ARCHIVE_ENTRY_ACL_USER: + ae_uid = archive_write_disk_uid(a, ae_name, ae_id); + acl_set_tag_type(acl_entry, ACL_USER); + acl_set_qualifier(acl_entry, &ae_uid); + break; + case ARCHIVE_ENTRY_ACL_GROUP: + ae_gid = archive_write_disk_gid(a, ae_name, ae_id); + acl_set_tag_type(acl_entry, ACL_GROUP); + acl_set_qualifier(acl_entry, &ae_gid); + break; + case ARCHIVE_ENTRY_ACL_USER_OBJ: + acl_set_tag_type(acl_entry, ACL_USER_OBJ); + break; + case ARCHIVE_ENTRY_ACL_GROUP_OBJ: + acl_set_tag_type(acl_entry, ACL_GROUP_OBJ); + break; + case ARCHIVE_ENTRY_ACL_MASK: + acl_set_tag_type(acl_entry, ACL_MASK); + break; + case ARCHIVE_ENTRY_ACL_OTHER: + acl_set_tag_type(acl_entry, ACL_OTHER); + break; + default: + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Unsupported ACL tag"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + + if (acl_get_permset(acl_entry, &acl_permset) != 0) { + archive_set_error(a, errno, + "Failed to get ACL permission set"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + if (acl_clear_perms(acl_permset) != 0) { + archive_set_error(a, errno, + "Failed to clear ACL permissions"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + + for (i = 0; i < acl_posix_perm_map_size; ++i) { + if (ae_permset & acl_posix_perm_map[i].a_perm) { + if (acl_add_perm(acl_permset, + acl_posix_perm_map[i].p_perm) != 0) { + archive_set_error(a, errno, + "Failed to add ACL permission"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + } + } + + } + + if (fd >= 0 && ae_requested_type == ARCHIVE_ENTRY_ACL_TYPE_ACCESS) { + if (acl_set_fd(fd, acl) == 0) + ret = ARCHIVE_OK; + else { + if (errno == EOPNOTSUPP) { + /* Filesystem doesn't support ACLs */ + ret = ARCHIVE_OK; + } else { + archive_set_error(a, errno, + "Failed to set acl on fd: %s", tname); + ret = ARCHIVE_WARN; + } + } + } else if (acl_set_file(name, acl_type, acl) != 0) { + if (errno == EOPNOTSUPP) { + /* Filesystem doesn't support ACLs */ + ret = ARCHIVE_OK; + } else { + archive_set_error(a, errno, "Failed to set acl: %s", + tname); + ret = ARCHIVE_WARN; + } + } +exit_free: + acl_free(acl); + return (ret); +} +#endif /* ARCHIVE_ACL_LIBACL */ + +int +archive_read_disk_entry_setup_acls(struct archive_read_disk *a, + struct archive_entry *entry, int *fd) +{ + const char *accpath; + int r; +#if ARCHIVE_ACL_LIBACL + acl_t acl; +#endif +#if ARCHIVE_ACL_LIBRICHACL + struct richacl *richacl; + mode_t mode; +#endif + + accpath = NULL; + r = ARCHIVE_OK; + + /* For default ACLs we need reachable accpath */ + if (*fd < 0 || S_ISDIR(archive_entry_mode(entry))) { + accpath = archive_read_disk_entry_setup_path(a, entry, fd); + if (accpath == NULL) + return (ARCHIVE_WARN); + } + + archive_entry_acl_clear(entry); + +#if ARCHIVE_ACL_LIBACL + acl = NULL; +#endif +#if ARCHIVE_ACL_LIBRICHACL + richacl = NULL; +#endif + +#if ARCHIVE_ACL_LIBRICHACL + /* Try NFSv4 ACL first. */ + if (*fd >= 0) + richacl = richacl_get_fd(*fd); + else if ((!a->follow_symlinks) + && (archive_entry_filetype(entry) == AE_IFLNK)) + /* We can't get the ACL of a symlink, so we assume it can't + have one */ + richacl = NULL; + else + richacl = richacl_get_file(accpath); + + /* Ignore "trivial" ACLs that just mirror the file mode. */ + if (richacl != NULL) { + mode = archive_entry_mode(entry); + if (richacl_equiv_mode(richacl, &mode) == 0) { + richacl_free(richacl); + richacl = NULL; + return (ARCHIVE_OK); + } + } + + if (richacl != NULL) { + r = translate_richacl(a, entry, richacl); + richacl_free(richacl); + richacl = NULL; + + if (r != ARCHIVE_OK) { + archive_set_error(&a->archive, errno, + "Couldn't translate NFSv4 ACLs"); + } + + return (r); + } +#endif /* ARCHIVE_ACL_LIBRICHACL */ + +#if ARCHIVE_ACL_LIBACL + /* Retrieve access ACL from file. */ + if (*fd >= 0) + acl = acl_get_fd(*fd); + else if ((!a->follow_symlinks) + && (archive_entry_filetype(entry) == AE_IFLNK)) + /* We can't get the ACL of a symlink, so we assume it can't + have one. */ + acl = NULL; + else + acl = acl_get_file(accpath, ACL_TYPE_ACCESS); + + if (acl != NULL) { + r = translate_acl(a, entry, acl, ARCHIVE_ENTRY_ACL_TYPE_ACCESS); + acl_free(acl); + acl = NULL; + + if (r != ARCHIVE_OK) { + archive_set_error(&a->archive, errno, + "Couldn't translate access ACLs"); + return (r); + } + } + + /* Only directories can have default ACLs. */ + if (S_ISDIR(archive_entry_mode(entry))) { + acl = acl_get_file(accpath, ACL_TYPE_DEFAULT); + if (acl != NULL) { + r = translate_acl(a, entry, acl, + ARCHIVE_ENTRY_ACL_TYPE_DEFAULT); + acl_free(acl); + if (r != ARCHIVE_OK) { + archive_set_error(&a->archive, errno, + "Couldn't translate default ACLs"); + return (r); + } + } + } +#endif /* ARCHIVE_ACL_LIBACL */ + return (r); +} + +int +archive_write_disk_set_acls(struct archive *a, int fd, const char *name, + struct archive_acl *abstract_acl, __LA_MODE_T mode) +{ + int ret = ARCHIVE_OK; + +#if !ARCHIVE_ACL_LIBRICHACL + (void)mode; /* UNUSED */ +#endif + +#if ARCHIVE_ACL_LIBRICHACL + if ((archive_acl_types(abstract_acl) + & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { + ret = set_richacl(a, fd, name, abstract_acl, mode, + ARCHIVE_ENTRY_ACL_TYPE_NFS4, "nfs4"); + } +#if ARCHIVE_ACL_LIBACL + else +#endif +#endif /* ARCHIVE_ACL_LIBRICHACL */ +#if ARCHIVE_ACL_LIBACL + if ((archive_acl_types(abstract_acl) + & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) { + if ((archive_acl_types(abstract_acl) + & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) { + ret = set_acl(a, fd, name, abstract_acl, + ARCHIVE_ENTRY_ACL_TYPE_ACCESS, "access"); + if (ret != ARCHIVE_OK) + return (ret); + } + if ((archive_acl_types(abstract_acl) + & ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) != 0) + ret = set_acl(a, fd, name, abstract_acl, + ARCHIVE_ENTRY_ACL_TYPE_DEFAULT, "default"); + } +#endif /* ARCHIVE_ACL_LIBACL */ + return (ret); +} +#endif /* ARCHIVE_ACL_LIBACL || ARCHIVE_ACL_LIBRICHACL */ diff --git a/c/archive_disk_acl_sunos.c b/c/archive_disk_acl_sunos.c new file mode 100644 --- /dev/null +++ b/c/archive_disk_acl_sunos.c @@ -0,0 +1,819 @@ +/*- + * Copyright (c) 2017 Martin Matuska + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +#if ARCHIVE_ACL_SUNOS + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_FCNTL_H +#include +#endif +#ifdef HAVE_SYS_TYPES_H +#include +#endif +#ifdef HAVE_SYS_ACL_H +#define _ACL_PRIVATE /* For debugging */ +#include +#endif + +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_read_disk_private.h" +#include "archive_write_disk_private.h" + +typedef struct { + const int a_perm; /* Libarchive permission or flag */ + const int p_perm; /* Platform permission or flag */ +} acl_perm_map_t; + +static const acl_perm_map_t acl_posix_perm_map[] = { + {ARCHIVE_ENTRY_ACL_EXECUTE, S_IXOTH }, + {ARCHIVE_ENTRY_ACL_WRITE, S_IWOTH }, + {ARCHIVE_ENTRY_ACL_READ, S_IROTH } +}; + +static const int acl_posix_perm_map_size = + (int)(sizeof(acl_posix_perm_map)/sizeof(acl_posix_perm_map[0])); + +#if ARCHIVE_ACL_SUNOS_NFS4 +static const acl_perm_map_t acl_nfs4_perm_map[] = { + {ARCHIVE_ENTRY_ACL_EXECUTE, ACE_EXECUTE}, + {ARCHIVE_ENTRY_ACL_READ_DATA, ACE_READ_DATA}, + {ARCHIVE_ENTRY_ACL_LIST_DIRECTORY, ACE_LIST_DIRECTORY}, + {ARCHIVE_ENTRY_ACL_WRITE_DATA, ACE_WRITE_DATA}, + {ARCHIVE_ENTRY_ACL_ADD_FILE, ACE_ADD_FILE}, + {ARCHIVE_ENTRY_ACL_APPEND_DATA, ACE_APPEND_DATA}, + {ARCHIVE_ENTRY_ACL_ADD_SUBDIRECTORY, ACE_ADD_SUBDIRECTORY}, + {ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS, ACE_READ_NAMED_ATTRS}, + {ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS, ACE_WRITE_NAMED_ATTRS}, + {ARCHIVE_ENTRY_ACL_DELETE_CHILD, ACE_DELETE_CHILD}, + {ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES, ACE_READ_ATTRIBUTES}, + {ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES, ACE_WRITE_ATTRIBUTES}, + {ARCHIVE_ENTRY_ACL_DELETE, ACE_DELETE}, + {ARCHIVE_ENTRY_ACL_READ_ACL, ACE_READ_ACL}, + {ARCHIVE_ENTRY_ACL_WRITE_ACL, ACE_WRITE_ACL}, + {ARCHIVE_ENTRY_ACL_WRITE_OWNER, ACE_WRITE_OWNER}, + {ARCHIVE_ENTRY_ACL_SYNCHRONIZE, ACE_SYNCHRONIZE} +}; + +static const int acl_nfs4_perm_map_size = + (int)(sizeof(acl_nfs4_perm_map)/sizeof(acl_nfs4_perm_map[0])); + +static const acl_perm_map_t acl_nfs4_flag_map[] = { + {ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT, ACE_FILE_INHERIT_ACE}, + {ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT, ACE_DIRECTORY_INHERIT_ACE}, + {ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT, ACE_NO_PROPAGATE_INHERIT_ACE}, + {ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY, ACE_INHERIT_ONLY_ACE}, + {ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS, ACE_SUCCESSFUL_ACCESS_ACE_FLAG}, + {ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS, ACE_FAILED_ACCESS_ACE_FLAG}, +#ifdef ACE_INHERITED_ACE + {ARCHIVE_ENTRY_ACL_ENTRY_INHERITED, ACE_INHERITED_ACE} +#endif +}; + +const int acl_nfs4_flag_map_size = + (int)(sizeof(acl_nfs4_flag_map)/sizeof(acl_nfs4_flag_map[0])); + +#endif /* ARCHIVE_ACL_SUNOS_NFS4 */ + +static void * +sunacl_get(int cmd, int *aclcnt, int fd, const char *path) +{ + int cnt, cntcmd; + size_t size; + void *aclp; + + if (cmd == GETACL) { + cntcmd = GETACLCNT; + size = sizeof(aclent_t); + } +#if ARCHIVE_ACL_SUNOS_NFS4 + else if (cmd == ACE_GETACL) { + cntcmd = ACE_GETACLCNT; + size = sizeof(ace_t); + } +#endif + else { + errno = EINVAL; + *aclcnt = -1; + return (NULL); + } + + aclp = NULL; + cnt = -2; + + while (cnt == -2 || (cnt == -1 && errno == ENOSPC)) { + if (path != NULL) + cnt = acl(path, cntcmd, 0, NULL); + else + cnt = facl(fd, cntcmd, 0, NULL); + + if (cnt > 0) { + if (aclp == NULL) + aclp = malloc(cnt * size); + else + aclp = realloc(NULL, cnt * size); + if (aclp != NULL) { + if (path != NULL) + cnt = acl(path, cmd, cnt, aclp); + else + cnt = facl(fd, cmd, cnt, aclp); + } + } else { + free(aclp); + aclp = NULL; + break; + } + } + + *aclcnt = cnt; + return (aclp); +} + +/* + * Check if acl is trivial + * This is a FreeBSD acl_is_trivial_np() implementation for Solaris + */ +static int +sun_acl_is_trivial(void *aclp, int aclcnt, mode_t mode, int is_nfs4, + int is_dir, int *trivialp) +{ +#if ARCHIVE_ACL_SUNOS_NFS4 + int i, p; + const uint32_t rperm = ACE_READ_DATA; + const uint32_t wperm = ACE_WRITE_DATA | ACE_APPEND_DATA; + const uint32_t eperm = ACE_EXECUTE; + const uint32_t pubset = ACE_READ_ATTRIBUTES | ACE_READ_NAMED_ATTRS | + ACE_READ_ACL | ACE_SYNCHRONIZE; + const uint32_t ownset = pubset | ACE_WRITE_ATTRIBUTES | + ACE_WRITE_NAMED_ATTRS | ACE_WRITE_ACL | ACE_WRITE_OWNER; + + ace_t *ace; + ace_t tace[6]; +#endif + + if (aclp == NULL || trivialp == NULL) + return (-1); + + *trivialp = 0; + + /* + * POSIX.1e ACLs marked with ACL_IS_TRIVIAL are compatible with + * FreeBSD acl_is_trivial_np(). On Solaris they have 4 entries, + * including mask. + */ + if (!is_nfs4) { + if (aclcnt == 4) + *trivialp = 1; + return (0); + } + +#if ARCHIVE_ACL_SUNOS_NFS4 + /* + * Continue with checking NFSv4 ACLs + * + * Create list of trivial ace's to be compared + */ + + /* owner@ allow pre */ + tace[0].a_flags = ACE_OWNER; + tace[0].a_type = ACE_ACCESS_ALLOWED_ACE_TYPE; + tace[0].a_access_mask = 0; + + /* owner@ deny */ + tace[1].a_flags = ACE_OWNER; + tace[1].a_type = ACE_ACCESS_DENIED_ACE_TYPE; + tace[1].a_access_mask = 0; + + /* group@ deny */ + tace[2].a_flags = ACE_GROUP | ACE_IDENTIFIER_GROUP; + tace[2].a_type = ACE_ACCESS_DENIED_ACE_TYPE; + tace[2].a_access_mask = 0; + + /* owner@ allow */ + tace[3].a_flags = ACE_OWNER; + tace[3].a_type = ACE_ACCESS_ALLOWED_ACE_TYPE; + tace[3].a_access_mask = ownset; + + /* group@ allow */ + tace[4].a_flags = ACE_GROUP | ACE_IDENTIFIER_GROUP; + tace[4].a_type = ACE_ACCESS_ALLOWED_ACE_TYPE; + tace[4].a_access_mask = pubset; + + /* everyone@ allow */ + tace[5].a_flags = ACE_EVERYONE; + tace[5].a_type = ACE_ACCESS_ALLOWED_ACE_TYPE; + tace[5].a_access_mask = pubset; + + /* Permissions for everyone@ */ + if (mode & 0004) + tace[5].a_access_mask |= rperm; + if (mode & 0002) + tace[5].a_access_mask |= wperm; + if (mode & 0001) + tace[5].a_access_mask |= eperm; + + /* Permissions for group@ */ + if (mode & 0040) + tace[4].a_access_mask |= rperm; + else if (mode & 0004) + tace[2].a_access_mask |= rperm; + if (mode & 0020) + tace[4].a_access_mask |= wperm; + else if (mode & 0002) + tace[2].a_access_mask |= wperm; + if (mode & 0010) + tace[4].a_access_mask |= eperm; + else if (mode & 0001) + tace[2].a_access_mask |= eperm; + + /* Permissions for owner@ */ + if (mode & 0400) { + tace[3].a_access_mask |= rperm; + if (!(mode & 0040) && (mode & 0004)) + tace[0].a_access_mask |= rperm; + } else if ((mode & 0040) || (mode & 0004)) + tace[1].a_access_mask |= rperm; + if (mode & 0200) { + tace[3].a_access_mask |= wperm; + if (!(mode & 0020) && (mode & 0002)) + tace[0].a_access_mask |= wperm; + } else if ((mode & 0020) || (mode & 0002)) + tace[1].a_access_mask |= wperm; + if (mode & 0100) { + tace[3].a_access_mask |= eperm; + if (!(mode & 0010) && (mode & 0001)) + tace[0].a_access_mask |= eperm; + } else if ((mode & 0010) || (mode & 0001)) + tace[1].a_access_mask |= eperm; + + /* Check if the acl count matches */ + p = 3; + for (i = 0; i < 3; i++) { + if (tace[i].a_access_mask != 0) + p++; + } + if (aclcnt != p) + return (0); + + p = 0; + for (i = 0; i < 6; i++) { + if (tace[i].a_access_mask != 0) { + ace = &((ace_t *)aclp)[p]; + /* + * Illumos added ACE_DELETE_CHILD to write perms for + * directories. We have to check against that, too. + */ + if (ace->a_flags != tace[i].a_flags || + ace->a_type != tace[i].a_type || + (ace->a_access_mask != tace[i].a_access_mask && + (!is_dir || (tace[i].a_access_mask & wperm) == 0 || + ace->a_access_mask != + (tace[i].a_access_mask | ACE_DELETE_CHILD)))) + return (0); + p++; + } + } + + *trivialp = 1; +#else /* !ARCHIVE_ACL_SUNOS_NFS4 */ + (void)is_dir; /* UNUSED */ + (void)aclp; /* UNUSED */ +#endif /* !ARCHIVE_ACL_SUNOS_NFS4 */ + return (0); +} + +/* + * Translate Solaris POSIX.1e and NFSv4 ACLs into libarchive internal ACL + */ +static int +translate_acl(struct archive_read_disk *a, + struct archive_entry *entry, void *aclp, int aclcnt, + int default_entry_acl_type) +{ + int e, i; + int ae_id, ae_tag, ae_perm; + int entry_acl_type; + const char *ae_name; + aclent_t *aclent; +#if ARCHIVE_ACL_SUNOS_NFS4 + ace_t *ace; +#endif + + if (aclcnt <= 0) + return (ARCHIVE_OK); + + for (e = 0; e < aclcnt; e++) { + ae_name = NULL; + ae_tag = 0; + ae_perm = 0; + +#if ARCHIVE_ACL_SUNOS_NFS4 + if (default_entry_acl_type == ARCHIVE_ENTRY_ACL_TYPE_NFS4) { + ace = &((ace_t *)aclp)[e]; + ae_id = ace->a_who; + + switch(ace->a_type) { + case ACE_ACCESS_ALLOWED_ACE_TYPE: + entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_ALLOW; + break; + case ACE_ACCESS_DENIED_ACE_TYPE: + entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_DENY; + break; + case ACE_SYSTEM_AUDIT_ACE_TYPE: + entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS; + break; + case ACE_SYSTEM_ALARM_ACE_TYPE: + entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_ALARM; + break; + default: + /* Unknown entry type, skip */ + continue; + } + + if ((ace->a_flags & ACE_OWNER) != 0) + ae_tag = ARCHIVE_ENTRY_ACL_USER_OBJ; + else if ((ace->a_flags & ACE_GROUP) != 0) + ae_tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; + else if ((ace->a_flags & ACE_EVERYONE) != 0) + ae_tag = ARCHIVE_ENTRY_ACL_EVERYONE; + else if ((ace->a_flags & ACE_IDENTIFIER_GROUP) != 0) { + ae_tag = ARCHIVE_ENTRY_ACL_GROUP; + ae_name = archive_read_disk_gname(&a->archive, + ae_id); + } else { + ae_tag = ARCHIVE_ENTRY_ACL_USER; + ae_name = archive_read_disk_uname(&a->archive, + ae_id); + } + + for (i = 0; i < acl_nfs4_flag_map_size; ++i) { + if ((ace->a_flags & + acl_nfs4_flag_map[i].p_perm) != 0) + ae_perm |= acl_nfs4_flag_map[i].a_perm; + } + + for (i = 0; i < acl_nfs4_perm_map_size; ++i) { + if ((ace->a_access_mask & + acl_nfs4_perm_map[i].p_perm) != 0) + ae_perm |= acl_nfs4_perm_map[i].a_perm; + } + } else +#endif /* ARCHIVE_ACL_SUNOS_NFS4 */ + if (default_entry_acl_type == ARCHIVE_ENTRY_ACL_TYPE_ACCESS) { + aclent = &((aclent_t *)aclp)[e]; + if ((aclent->a_type & ACL_DEFAULT) != 0) + entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_DEFAULT; + else + entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS; + ae_id = aclent->a_id; + + switch(aclent->a_type) { + case DEF_USER: + case USER: + ae_name = archive_read_disk_uname(&a->archive, + ae_id); + ae_tag = ARCHIVE_ENTRY_ACL_USER; + break; + case DEF_GROUP: + case GROUP: + ae_name = archive_read_disk_gname(&a->archive, + ae_id); + ae_tag = ARCHIVE_ENTRY_ACL_GROUP; + break; + case DEF_CLASS_OBJ: + case CLASS_OBJ: + ae_tag = ARCHIVE_ENTRY_ACL_MASK; + break; + case DEF_USER_OBJ: + case USER_OBJ: + ae_tag = ARCHIVE_ENTRY_ACL_USER_OBJ; + break; + case DEF_GROUP_OBJ: + case GROUP_OBJ: + ae_tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; + break; + case DEF_OTHER_OBJ: + case OTHER_OBJ: + ae_tag = ARCHIVE_ENTRY_ACL_OTHER; + break; + default: + /* Unknown tag type, skip */ + continue; + } + + for (i = 0; i < acl_posix_perm_map_size; ++i) { + if ((aclent->a_perm & + acl_posix_perm_map[i].p_perm) != 0) + ae_perm |= acl_posix_perm_map[i].a_perm; + } + } else + return (ARCHIVE_WARN); + + archive_entry_acl_add_entry(entry, entry_acl_type, + ae_perm, ae_tag, ae_id, ae_name); + } + return (ARCHIVE_OK); +} + +static int +set_acl(struct archive *a, int fd, const char *name, + struct archive_acl *abstract_acl, + int ae_requested_type, const char *tname) +{ + aclent_t *aclent; +#if ARCHIVE_ACL_SUNOS_NFS4 + ace_t *ace; +#endif + int cmd, e, r; + void *aclp; + int ret; + int ae_type, ae_permset, ae_tag, ae_id; + int perm_map_size; + const acl_perm_map_t *perm_map; + uid_t ae_uid; + gid_t ae_gid; + const char *ae_name; + int entries; + int i; + + ret = ARCHIVE_OK; + entries = archive_acl_reset(abstract_acl, ae_requested_type); + if (entries == 0) + return (ARCHIVE_OK); + + + switch (ae_requested_type) { + case ARCHIVE_ENTRY_ACL_TYPE_POSIX1E: + cmd = SETACL; + aclp = malloc(entries * sizeof(aclent_t)); + break; +#if ARCHIVE_ACL_SUNOS_NFS4 + case ARCHIVE_ENTRY_ACL_TYPE_NFS4: + cmd = ACE_SETACL; + aclp = malloc(entries * sizeof(ace_t)); + + break; +#endif + default: + errno = ENOENT; + archive_set_error(a, errno, "Unsupported ACL type"); + return (ARCHIVE_FAILED); + } + + if (aclp == NULL) { + archive_set_error(a, errno, + "Can't allocate memory for acl buffer"); + return (ARCHIVE_FAILED); + } + + e = 0; + + while (archive_acl_next(a, abstract_acl, ae_requested_type, &ae_type, + &ae_permset, &ae_tag, &ae_id, &ae_name) == ARCHIVE_OK) { + aclent = NULL; +#if ARCHIVE_ACL_SUNOS_NFS4 + ace = NULL; +#endif + if (cmd == SETACL) { + aclent = &((aclent_t *)aclp)[e]; + aclent->a_id = -1; + aclent->a_type = 0; + aclent->a_perm = 0; + } +#if ARCHIVE_ACL_SUNOS_NFS4 + else { /* cmd == ACE_SETACL */ + ace = &((ace_t *)aclp)[e]; + ace->a_who = -1; + ace->a_access_mask = 0; + ace->a_flags = 0; + } +#endif /* ARCHIVE_ACL_SUNOS_NFS4 */ + + switch (ae_tag) { + case ARCHIVE_ENTRY_ACL_USER: + ae_uid = archive_write_disk_uid(a, ae_name, ae_id); + if (aclent != NULL) { + aclent->a_id = ae_uid; + aclent->a_type |= USER; + } +#if ARCHIVE_ACL_SUNOS_NFS4 + else { + ace->a_who = ae_uid; + } +#endif + break; + case ARCHIVE_ENTRY_ACL_GROUP: + ae_gid = archive_write_disk_gid(a, ae_name, ae_id); + if (aclent != NULL) { + aclent->a_id = ae_gid; + aclent->a_type |= GROUP; + } +#if ARCHIVE_ACL_SUNOS_NFS4 + else { + ace->a_who = ae_gid; + ace->a_flags |= ACE_IDENTIFIER_GROUP; + } +#endif + break; + case ARCHIVE_ENTRY_ACL_USER_OBJ: + if (aclent != NULL) + aclent->a_type |= USER_OBJ; +#if ARCHIVE_ACL_SUNOS_NFS4 + else { + ace->a_flags |= ACE_OWNER; + } +#endif + break; + case ARCHIVE_ENTRY_ACL_GROUP_OBJ: + if (aclent != NULL) + aclent->a_type |= GROUP_OBJ; +#if ARCHIVE_ACL_SUNOS_NFS4 + else { + ace->a_flags |= ACE_GROUP; + ace->a_flags |= ACE_IDENTIFIER_GROUP; + } +#endif + break; + case ARCHIVE_ENTRY_ACL_MASK: + if (aclent != NULL) + aclent->a_type |= CLASS_OBJ; + break; + case ARCHIVE_ENTRY_ACL_OTHER: + if (aclent != NULL) + aclent->a_type |= OTHER_OBJ; + break; +#if ARCHIVE_ACL_SUNOS_NFS4 + case ARCHIVE_ENTRY_ACL_EVERYONE: + if (ace != NULL) + ace->a_flags |= ACE_EVERYONE; + break; +#endif + default: + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Unsupported ACL tag"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + + r = 0; + switch (ae_type) { +#if ARCHIVE_ACL_SUNOS_NFS4 + case ARCHIVE_ENTRY_ACL_TYPE_ALLOW: + if (ace != NULL) + ace->a_type = ACE_ACCESS_ALLOWED_ACE_TYPE; + else + r = -1; + break; + case ARCHIVE_ENTRY_ACL_TYPE_DENY: + if (ace != NULL) + ace->a_type = ACE_ACCESS_DENIED_ACE_TYPE; + else + r = -1; + break; + case ARCHIVE_ENTRY_ACL_TYPE_AUDIT: + if (ace != NULL) + ace->a_type = ACE_SYSTEM_AUDIT_ACE_TYPE; + else + r = -1; + break; + case ARCHIVE_ENTRY_ACL_TYPE_ALARM: + if (ace != NULL) + ace->a_type = ACE_SYSTEM_ALARM_ACE_TYPE; + else + r = -1; + break; +#endif + case ARCHIVE_ENTRY_ACL_TYPE_ACCESS: + if (aclent == NULL) + r = -1; + break; + case ARCHIVE_ENTRY_ACL_TYPE_DEFAULT: + if (aclent != NULL) + aclent->a_type |= ACL_DEFAULT; + else + r = -1; + break; + default: + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Unsupported ACL entry type"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + + if (r != 0) { + errno = EINVAL; + archive_set_error(a, errno, + "Failed to set ACL entry type"); + ret = ARCHIVE_FAILED; + goto exit_free; + } + +#if ARCHIVE_ACL_SUNOS_NFS4 + if (ae_requested_type == ARCHIVE_ENTRY_ACL_TYPE_NFS4) { + perm_map_size = acl_nfs4_perm_map_size; + perm_map = acl_nfs4_perm_map; + } else { +#endif + perm_map_size = acl_posix_perm_map_size; + perm_map = acl_posix_perm_map; +#if ARCHIVE_ACL_SUNOS_NFS4 + } +#endif + for (i = 0; i < perm_map_size; ++i) { + if (ae_permset & perm_map[i].a_perm) { +#if ARCHIVE_ACL_SUNOS_NFS4 + if (ae_requested_type == + ARCHIVE_ENTRY_ACL_TYPE_NFS4) + ace->a_access_mask |= + perm_map[i].p_perm; + else +#endif + aclent->a_perm |= perm_map[i].p_perm; + } + } + +#if ARCHIVE_ACL_SUNOS_NFS4 + if (ae_requested_type == ARCHIVE_ENTRY_ACL_TYPE_NFS4) { + for (i = 0; i < acl_nfs4_flag_map_size; ++i) { + if (ae_permset & acl_nfs4_flag_map[i].a_perm) { + ace->a_flags |= + acl_nfs4_flag_map[i].p_perm; + } + } + } +#endif + e++; + } + + /* Try restoring the ACL through 'fd' if we can. */ + if (fd >= 0) { + if (facl(fd, cmd, entries, aclp) == 0) + ret = ARCHIVE_OK; + else { + if (errno == EOPNOTSUPP) { + /* Filesystem doesn't support ACLs */ + ret = ARCHIVE_OK; + } else { + archive_set_error(a, errno, + "Failed to set acl on fd: %s", tname); + ret = ARCHIVE_WARN; + } + } + } else if (acl(name, cmd, entries, aclp) != 0) { + if (errno == EOPNOTSUPP) { + /* Filesystem doesn't support ACLs */ + ret = ARCHIVE_OK; + } else { + archive_set_error(a, errno, "Failed to set acl: %s", + tname); + ret = ARCHIVE_WARN; + } + } +exit_free: + free(aclp); + return (ret); +} + +int +archive_read_disk_entry_setup_acls(struct archive_read_disk *a, + struct archive_entry *entry, int *fd) +{ + const char *accpath; + void *aclp; + int aclcnt; + int r; + + accpath = NULL; + + if (*fd < 0) { + accpath = archive_read_disk_entry_setup_path(a, entry, fd); + if (accpath == NULL) + return (ARCHIVE_WARN); + } + + archive_entry_acl_clear(entry); + + aclp = NULL; + +#if ARCHIVE_ACL_SUNOS_NFS4 + if (*fd >= 0) + aclp = sunacl_get(ACE_GETACL, &aclcnt, *fd, NULL); + else if ((!a->follow_symlinks) + && (archive_entry_filetype(entry) == AE_IFLNK)) + /* We can't get the ACL of a symlink, so we assume it can't + have one. */ + aclp = NULL; + else + aclp = sunacl_get(ACE_GETACL, &aclcnt, 0, accpath); + + if (aclp != NULL && sun_acl_is_trivial(aclp, aclcnt, + archive_entry_mode(entry), 1, S_ISDIR(archive_entry_mode(entry)), + &r) == 0 && r == 1) { + free(aclp); + aclp = NULL; + return (ARCHIVE_OK); + } + + if (aclp != NULL) { + r = translate_acl(a, entry, aclp, aclcnt, + ARCHIVE_ENTRY_ACL_TYPE_NFS4); + free(aclp); + aclp = NULL; + + if (r != ARCHIVE_OK) { + archive_set_error(&a->archive, errno, + "Couldn't translate NFSv4 ACLs"); + } + return (r); + } +#endif /* ARCHIVE_ACL_SUNOS_NFS4 */ + + /* Retrieve POSIX.1e ACLs from file. */ + if (*fd >= 0) + aclp = sunacl_get(GETACL, &aclcnt, *fd, NULL); + else if ((!a->follow_symlinks) + && (archive_entry_filetype(entry) == AE_IFLNK)) + /* We can't get the ACL of a symlink, so we assume it can't + have one. */ + aclp = NULL; + else + aclp = sunacl_get(GETACL, &aclcnt, 0, accpath); + + /* Ignore "trivial" ACLs that just mirror the file mode. */ + if (aclp != NULL && sun_acl_is_trivial(aclp, aclcnt, + archive_entry_mode(entry), 0, S_ISDIR(archive_entry_mode(entry)), + &r) == 0 && r == 1) { + free(aclp); + aclp = NULL; + } + + if (aclp != NULL) + { + r = translate_acl(a, entry, aclp, aclcnt, + ARCHIVE_ENTRY_ACL_TYPE_ACCESS); + free(aclp); + aclp = NULL; + + if (r != ARCHIVE_OK) { + archive_set_error(&a->archive, errno, + "Couldn't translate access ACLs"); + return (r); + } + } + + return (ARCHIVE_OK); +} + +int +archive_write_disk_set_acls(struct archive *a, int fd, const char *name, + struct archive_acl *abstract_acl, __LA_MODE_T mode) +{ + int ret = ARCHIVE_OK; + + (void)mode; /* UNUSED */ + + if ((archive_acl_types(abstract_acl) + & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) { + /* Solaris writes POSIX.1e access and default ACLs together */ + ret = set_acl(a, fd, name, abstract_acl, + ARCHIVE_ENTRY_ACL_TYPE_POSIX1E, "posix1e"); + + /* Simultaneous POSIX.1e and NFSv4 is not supported */ + return (ret); + } +#if ARCHIVE_ACL_SUNOS_NFS4 + else if ((archive_acl_types(abstract_acl) & + ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { + ret = set_acl(a, fd, name, abstract_acl, + ARCHIVE_ENTRY_ACL_TYPE_NFS4, "nfs4"); + } +#endif + return (ret); +} +#endif /* ARCHIVE_ACL_SUNOS */ diff --git a/c/archive_endian.h b/c/archive_endian.h new file mode 100644 --- /dev/null +++ b/c/archive_endian.h @@ -0,0 +1,195 @@ +/*- + * Copyright (c) 2002 Thomas Moestl + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * $FreeBSD: head/lib/libarchive/archive_endian.h 201085 2009-12-28 02:17:15Z kientzle $ + * + * Borrowed from FreeBSD's + */ + +#ifndef ARCHIVE_ENDIAN_H_INCLUDED +#define ARCHIVE_ENDIAN_H_INCLUDED + +/* Note: This is a purely internal header! */ +/* Do not use this outside of libarchive internal code! */ + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif + +/* + * Disabling inline keyword for compilers known to choke on it: + * - Watcom C++ in C code. (For any version?) + * - SGI MIPSpro + * - Microsoft Visual C++ 6.0 (supposedly newer versions too) + * - IBM VisualAge 6 (XL v6) + * - Sun WorkShop C (SunPro) before 5.9 + */ +#if defined(__WATCOMC__) || defined(__sgi) || defined(__hpux) || defined(__BORLANDC__) +#define inline +#elif defined(__IBMC__) && __IBMC__ < 700 +#define inline +#elif defined(__SUNPRO_C) && __SUNPRO_C < 0x590 +#define inline +#elif defined(_MSC_VER) || defined(__osf__) +#define inline __inline +#endif + +/* Alignment-agnostic encode/decode bytestream to/from little/big endian. */ + +static inline uint16_t +archive_be16dec(const void *pp) +{ + unsigned char const *p = (unsigned char const *)pp; + + /* Store into unsigned temporaries before left shifting, to avoid + promotion to signed int and then left shifting into the sign bit, + which is undefined behaviour. */ + unsigned int p1 = p[1]; + unsigned int p0 = p[0]; + + return ((p0 << 8) | p1); +} + +static inline uint32_t +archive_be32dec(const void *pp) +{ + unsigned char const *p = (unsigned char const *)pp; + + /* Store into unsigned temporaries before left shifting, to avoid + promotion to signed int and then left shifting into the sign bit, + which is undefined behaviour. */ + unsigned int p3 = p[3]; + unsigned int p2 = p[2]; + unsigned int p1 = p[1]; + unsigned int p0 = p[0]; + + return ((p0 << 24) | (p1 << 16) | (p2 << 8) | p3); +} + +static inline uint64_t +archive_be64dec(const void *pp) +{ + unsigned char const *p = (unsigned char const *)pp; + + return (((uint64_t)archive_be32dec(p) << 32) | archive_be32dec(p + 4)); +} + +static inline uint16_t +archive_le16dec(const void *pp) +{ + unsigned char const *p = (unsigned char const *)pp; + + /* Store into unsigned temporaries before left shifting, to avoid + promotion to signed int and then left shifting into the sign bit, + which is undefined behaviour. */ + unsigned int p1 = p[1]; + unsigned int p0 = p[0]; + + return ((p1 << 8) | p0); +} + +static inline uint32_t +archive_le32dec(const void *pp) +{ + unsigned char const *p = (unsigned char const *)pp; + + /* Store into unsigned temporaries before left shifting, to avoid + promotion to signed int and then left shifting into the sign bit, + which is undefined behaviour. */ + unsigned int p3 = p[3]; + unsigned int p2 = p[2]; + unsigned int p1 = p[1]; + unsigned int p0 = p[0]; + + return ((p3 << 24) | (p2 << 16) | (p1 << 8) | p0); +} + +static inline uint64_t +archive_le64dec(const void *pp) +{ + unsigned char const *p = (unsigned char const *)pp; + + return (((uint64_t)archive_le32dec(p + 4) << 32) | archive_le32dec(p)); +} + +static inline void +archive_be16enc(void *pp, uint16_t u) +{ + unsigned char *p = (unsigned char *)pp; + + p[0] = (u >> 8) & 0xff; + p[1] = u & 0xff; +} + +static inline void +archive_be32enc(void *pp, uint32_t u) +{ + unsigned char *p = (unsigned char *)pp; + + p[0] = (u >> 24) & 0xff; + p[1] = (u >> 16) & 0xff; + p[2] = (u >> 8) & 0xff; + p[3] = u & 0xff; +} + +static inline void +archive_be64enc(void *pp, uint64_t u) +{ + unsigned char *p = (unsigned char *)pp; + + archive_be32enc(p, (uint32_t)(u >> 32)); + archive_be32enc(p + 4, (uint32_t)(u & 0xffffffff)); +} + +static inline void +archive_le16enc(void *pp, uint16_t u) +{ + unsigned char *p = (unsigned char *)pp; + + p[0] = u & 0xff; + p[1] = (u >> 8) & 0xff; +} + +static inline void +archive_le32enc(void *pp, uint32_t u) +{ + unsigned char *p = (unsigned char *)pp; + + p[0] = u & 0xff; + p[1] = (u >> 8) & 0xff; + p[2] = (u >> 16) & 0xff; + p[3] = (u >> 24) & 0xff; +} + +static inline void +archive_le64enc(void *pp, uint64_t u) +{ + unsigned char *p = (unsigned char *)pp; + + archive_le32enc(p, (uint32_t)(u & 0xffffffff)); + archive_le32enc(p + 4, (uint32_t)(u >> 32)); +} + +#endif diff --git a/c/archive_entry.c b/c/archive_entry.c new file mode 100644 --- /dev/null +++ b/c/archive_entry.c @@ -0,0 +1,2066 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * Copyright (c) 2016 Martin Matuska + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_entry.c 201096 2009-12-28 02:41:27Z kientzle $"); + +#ifdef HAVE_SYS_STAT_H +#include +#endif +#ifdef HAVE_SYS_TYPES_H +#include +#endif +#if MAJOR_IN_MKDEV +#include +#define HAVE_MAJOR +#elif MAJOR_IN_SYSMACROS +#include +#define HAVE_MAJOR +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_LIMITS_H +#include +#endif +#ifdef HAVE_LINUX_FS_H +#include /* for Linux file flags */ +#endif +/* + * Some Linux distributions have both linux/ext2_fs.h and ext2fs/ext2_fs.h. + * As the include guards don't agree, the order of include is important. + */ +#ifdef HAVE_LINUX_EXT2_FS_H +#include /* for Linux file flags */ +#endif +#if defined(HAVE_EXT2FS_EXT2_FS_H) && !defined(__CYGWIN__) +#include /* for Linux file flags */ +#endif +#include +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_WCHAR_H +#include +#endif + +#include "archive.h" +#include "archive_acl_private.h" +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_private.h" +#include "archive_entry_private.h" + +#if !defined(HAVE_MAJOR) && !defined(major) +/* Replacement for major/minor/makedev. */ +#define major(x) ((int)(0x00ff & ((x) >> 8))) +#define minor(x) ((int)(0xffff00ff & (x))) +#define makedev(maj,min) ((0xff00 & ((maj)<<8)) | (0xffff00ff & (min))) +#endif + +/* Play games to come up with a suitable makedev() definition. */ +#ifdef __QNXNTO__ +/* QNX. */ +#include +#define ae_makedev(maj, min) makedev(ND_LOCAL_NODE, (maj), (min)) +#elif defined makedev +/* There's a "makedev" macro. */ +#define ae_makedev(maj, min) makedev((maj), (min)) +#elif defined mkdev || ((defined _WIN32 || defined __WIN32__) && !defined(__CYGWIN__)) +/* Windows. */ +#define ae_makedev(maj, min) mkdev((maj), (min)) +#else +/* There's a "makedev" function. */ +#define ae_makedev(maj, min) makedev((maj), (min)) +#endif + +/* + * This adjustment is needed to support the following idiom for adding + * 1000ns to the stored time: + * archive_entry_set_atime(archive_entry_atime(), + * archive_entry_atime_nsec() + 1000) + * The additional if() here compensates for ambiguity in the C standard, + * which permits two possible interpretations of a % b when a is negative. + */ +#define FIX_NS(t,ns) \ + do { \ + t += ns / 1000000000; \ + ns %= 1000000000; \ + if (ns < 0) { --t; ns += 1000000000; } \ + } while (0) + +static char * ae_fflagstostr(unsigned long bitset, unsigned long bitclear); +static const wchar_t *ae_wcstofflags(const wchar_t *stringp, + unsigned long *setp, unsigned long *clrp); +static const char *ae_strtofflags(const char *stringp, + unsigned long *setp, unsigned long *clrp); + +#ifndef HAVE_WCSCPY +static wchar_t * wcscpy(wchar_t *s1, const wchar_t *s2) +{ + wchar_t *dest = s1; + while ((*s1 = *s2) != L'\0') + ++s1, ++s2; + return dest; +} +#endif +#ifndef HAVE_WCSLEN +static size_t wcslen(const wchar_t *s) +{ + const wchar_t *p = s; + while (*p != L'\0') + ++p; + return p - s; +} +#endif +#ifndef HAVE_WMEMCMP +/* Good enough for simple equality testing, but not for sorting. */ +#define wmemcmp(a,b,i) memcmp((a), (b), (i) * sizeof(wchar_t)) +#endif + +/**************************************************************************** + * + * Public Interface + * + ****************************************************************************/ + +struct archive_entry * +archive_entry_clear(struct archive_entry *entry) +{ + if (entry == NULL) + return (NULL); + archive_mstring_clean(&entry->ae_fflags_text); + archive_mstring_clean(&entry->ae_gname); + archive_mstring_clean(&entry->ae_hardlink); + archive_mstring_clean(&entry->ae_pathname); + archive_mstring_clean(&entry->ae_sourcepath); + archive_mstring_clean(&entry->ae_symlink); + archive_mstring_clean(&entry->ae_uname); + archive_entry_copy_mac_metadata(entry, NULL, 0); + archive_acl_clear(&entry->acl); + archive_entry_xattr_clear(entry); + archive_entry_sparse_clear(entry); + free(entry->stat); + entry->ae_symlink_type = AE_SYMLINK_TYPE_UNDEFINED; + memset(entry, 0, sizeof(*entry)); + return entry; +} + +struct archive_entry * +archive_entry_clone(struct archive_entry *entry) +{ + struct archive_entry *entry2; + struct ae_xattr *xp; + struct ae_sparse *sp; + size_t s; + const void *p; + + /* Allocate new structure and copy over all of the fields. */ + /* TODO: Should we copy the archive over? Or require a new archive + * as an argument? */ + entry2 = archive_entry_new2(entry->archive); + if (entry2 == NULL) + return (NULL); + entry2->ae_stat = entry->ae_stat; + entry2->ae_fflags_set = entry->ae_fflags_set; + entry2->ae_fflags_clear = entry->ae_fflags_clear; + + /* TODO: XXX If clone can have a different archive, what do we do here if + * character sets are different? XXX */ + archive_mstring_copy(&entry2->ae_fflags_text, &entry->ae_fflags_text); + archive_mstring_copy(&entry2->ae_gname, &entry->ae_gname); + archive_mstring_copy(&entry2->ae_hardlink, &entry->ae_hardlink); + archive_mstring_copy(&entry2->ae_pathname, &entry->ae_pathname); + archive_mstring_copy(&entry2->ae_sourcepath, &entry->ae_sourcepath); + archive_mstring_copy(&entry2->ae_symlink, &entry->ae_symlink); + entry2->ae_set = entry->ae_set; + archive_mstring_copy(&entry2->ae_uname, &entry->ae_uname); + + /* Copy symlink type */ + entry2->ae_symlink_type = entry->ae_symlink_type; + + /* Copy encryption status */ + entry2->encryption = entry->encryption; + + /* Copy ACL data over. */ + archive_acl_copy(&entry2->acl, &entry->acl); + + /* Copy Mac OS metadata. */ + p = archive_entry_mac_metadata(entry, &s); + archive_entry_copy_mac_metadata(entry2, p, s); + + /* Copy xattr data over. */ + xp = entry->xattr_head; + while (xp != NULL) { + archive_entry_xattr_add_entry(entry2, + xp->name, xp->value, xp->size); + xp = xp->next; + } + + /* Copy sparse data over. */ + sp = entry->sparse_head; + while (sp != NULL) { + archive_entry_sparse_add_entry(entry2, + sp->offset, sp->length); + sp = sp->next; + } + + return (entry2); +} + +void +archive_entry_free(struct archive_entry *entry) +{ + archive_entry_clear(entry); + free(entry); +} + +struct archive_entry * +archive_entry_new(void) +{ + return archive_entry_new2(NULL); +} + +struct archive_entry * +archive_entry_new2(struct archive *a) +{ + struct archive_entry *entry; + + entry = (struct archive_entry *)calloc(1, sizeof(*entry)); + if (entry == NULL) + return (NULL); + entry->archive = a; + entry->ae_symlink_type = AE_SYMLINK_TYPE_UNDEFINED; + return (entry); +} + +/* + * Functions for reading fields from an archive_entry. + */ + +time_t +archive_entry_atime(struct archive_entry *entry) +{ + return (entry->ae_stat.aest_atime); +} + +long +archive_entry_atime_nsec(struct archive_entry *entry) +{ + return (entry->ae_stat.aest_atime_nsec); +} + +int +archive_entry_atime_is_set(struct archive_entry *entry) +{ + return (entry->ae_set & AE_SET_ATIME); +} + +time_t +archive_entry_birthtime(struct archive_entry *entry) +{ + return (entry->ae_stat.aest_birthtime); +} + +long +archive_entry_birthtime_nsec(struct archive_entry *entry) +{ + return (entry->ae_stat.aest_birthtime_nsec); +} + +int +archive_entry_birthtime_is_set(struct archive_entry *entry) +{ + return (entry->ae_set & AE_SET_BIRTHTIME); +} + +time_t +archive_entry_ctime(struct archive_entry *entry) +{ + return (entry->ae_stat.aest_ctime); +} + +int +archive_entry_ctime_is_set(struct archive_entry *entry) +{ + return (entry->ae_set & AE_SET_CTIME); +} + +long +archive_entry_ctime_nsec(struct archive_entry *entry) +{ + return (entry->ae_stat.aest_ctime_nsec); +} + +dev_t +archive_entry_dev(struct archive_entry *entry) +{ + if (entry->ae_stat.aest_dev_is_broken_down) + return ae_makedev(entry->ae_stat.aest_devmajor, + entry->ae_stat.aest_devminor); + else + return (entry->ae_stat.aest_dev); +} + +int +archive_entry_dev_is_set(struct archive_entry *entry) +{ + return (entry->ae_set & AE_SET_DEV); +} + +dev_t +archive_entry_devmajor(struct archive_entry *entry) +{ + if (entry->ae_stat.aest_dev_is_broken_down) + return (entry->ae_stat.aest_devmajor); + else + return major(entry->ae_stat.aest_dev); +} + +dev_t +archive_entry_devminor(struct archive_entry *entry) +{ + if (entry->ae_stat.aest_dev_is_broken_down) + return (entry->ae_stat.aest_devminor); + else + return minor(entry->ae_stat.aest_dev); +} + +__LA_MODE_T +archive_entry_filetype(struct archive_entry *entry) +{ + return (AE_IFMT & entry->acl.mode); +} + +void +archive_entry_fflags(struct archive_entry *entry, + unsigned long *set, unsigned long *clear) +{ + *set = entry->ae_fflags_set; + *clear = entry->ae_fflags_clear; +} + +/* + * Note: if text was provided, this just returns that text. If you + * really need the text to be rebuilt in a canonical form, set the + * text, ask for the bitmaps, then set the bitmaps. (Setting the + * bitmaps clears any stored text.) This design is deliberate: if + * we're editing archives, we don't want to discard flags just because + * they aren't supported on the current system. The bitmap<->text + * conversions are platform-specific (see below). + */ +const char * +archive_entry_fflags_text(struct archive_entry *entry) +{ + const char *f; + char *p; + + if (archive_mstring_get_mbs(entry->archive, + &entry->ae_fflags_text, &f) == 0) { + if (f != NULL) + return (f); + } else if (errno == ENOMEM) + __archive_errx(1, "No memory"); + + if (entry->ae_fflags_set == 0 && entry->ae_fflags_clear == 0) + return (NULL); + + p = ae_fflagstostr(entry->ae_fflags_set, entry->ae_fflags_clear); + if (p == NULL) + return (NULL); + + archive_mstring_copy_mbs(&entry->ae_fflags_text, p); + free(p); + if (archive_mstring_get_mbs(entry->archive, + &entry->ae_fflags_text, &f) == 0) + return (f); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (NULL); +} + +la_int64_t +archive_entry_gid(struct archive_entry *entry) +{ + return (entry->ae_stat.aest_gid); +} + +const char * +archive_entry_gname(struct archive_entry *entry) +{ + const char *p; + if (archive_mstring_get_mbs(entry->archive, &entry->ae_gname, &p) == 0) + return (p); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (NULL); +} + +const char * +archive_entry_gname_utf8(struct archive_entry *entry) +{ + const char *p; + if (archive_mstring_get_utf8(entry->archive, &entry->ae_gname, &p) == 0) + return (p); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (NULL); +} + + +const wchar_t * +archive_entry_gname_w(struct archive_entry *entry) +{ + const wchar_t *p; + if (archive_mstring_get_wcs(entry->archive, &entry->ae_gname, &p) == 0) + return (p); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (NULL); +} + +int +_archive_entry_gname_l(struct archive_entry *entry, + const char **p, size_t *len, struct archive_string_conv *sc) +{ + return (archive_mstring_get_mbs_l(&entry->ae_gname, p, len, sc)); +} + +const char * +archive_entry_hardlink(struct archive_entry *entry) +{ + const char *p; + if ((entry->ae_set & AE_SET_HARDLINK) == 0) + return (NULL); + if (archive_mstring_get_mbs( + entry->archive, &entry->ae_hardlink, &p) == 0) + return (p); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (NULL); +} + +const char * +archive_entry_hardlink_utf8(struct archive_entry *entry) +{ + const char *p; + if ((entry->ae_set & AE_SET_HARDLINK) == 0) + return (NULL); + if (archive_mstring_get_utf8( + entry->archive, &entry->ae_hardlink, &p) == 0) + return (p); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (NULL); +} + +const wchar_t * +archive_entry_hardlink_w(struct archive_entry *entry) +{ + const wchar_t *p; + if ((entry->ae_set & AE_SET_HARDLINK) == 0) + return (NULL); + if (archive_mstring_get_wcs( + entry->archive, &entry->ae_hardlink, &p) == 0) + return (p); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (NULL); +} + +int +_archive_entry_hardlink_l(struct archive_entry *entry, + const char **p, size_t *len, struct archive_string_conv *sc) +{ + if ((entry->ae_set & AE_SET_HARDLINK) == 0) { + *p = NULL; + *len = 0; + return (0); + } + return (archive_mstring_get_mbs_l(&entry->ae_hardlink, p, len, sc)); +} + +la_int64_t +archive_entry_ino(struct archive_entry *entry) +{ + return (entry->ae_stat.aest_ino); +} + +int +archive_entry_ino_is_set(struct archive_entry *entry) +{ + return (entry->ae_set & AE_SET_INO); +} + +la_int64_t +archive_entry_ino64(struct archive_entry *entry) +{ + return (entry->ae_stat.aest_ino); +} + +__LA_MODE_T +archive_entry_mode(struct archive_entry *entry) +{ + return (entry->acl.mode); +} + +time_t +archive_entry_mtime(struct archive_entry *entry) +{ + return (entry->ae_stat.aest_mtime); +} + +long +archive_entry_mtime_nsec(struct archive_entry *entry) +{ + return (entry->ae_stat.aest_mtime_nsec); +} + +int +archive_entry_mtime_is_set(struct archive_entry *entry) +{ + return (entry->ae_set & AE_SET_MTIME); +} + +unsigned int +archive_entry_nlink(struct archive_entry *entry) +{ + return (entry->ae_stat.aest_nlink); +} + +const char * +archive_entry_pathname(struct archive_entry *entry) +{ + const char *p; + if (archive_mstring_get_mbs( + entry->archive, &entry->ae_pathname, &p) == 0) + return (p); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (NULL); +} + +const char * +archive_entry_pathname_utf8(struct archive_entry *entry) +{ + const char *p; + if (archive_mstring_get_utf8( + entry->archive, &entry->ae_pathname, &p) == 0) + return (p); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (NULL); +} + +const wchar_t * +archive_entry_pathname_w(struct archive_entry *entry) +{ + const wchar_t *p; + if (archive_mstring_get_wcs( + entry->archive, &entry->ae_pathname, &p) == 0) + return (p); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (NULL); +} + +int +_archive_entry_pathname_l(struct archive_entry *entry, + const char **p, size_t *len, struct archive_string_conv *sc) +{ + return (archive_mstring_get_mbs_l(&entry->ae_pathname, p, len, sc)); +} + +__LA_MODE_T +archive_entry_perm(struct archive_entry *entry) +{ + return (~AE_IFMT & entry->acl.mode); +} + +dev_t +archive_entry_rdev(struct archive_entry *entry) +{ + if (entry->ae_stat.aest_rdev_is_broken_down) + return ae_makedev(entry->ae_stat.aest_rdevmajor, + entry->ae_stat.aest_rdevminor); + else + return (entry->ae_stat.aest_rdev); +} + +dev_t +archive_entry_rdevmajor(struct archive_entry *entry) +{ + if (entry->ae_stat.aest_rdev_is_broken_down) + return (entry->ae_stat.aest_rdevmajor); + else + return major(entry->ae_stat.aest_rdev); +} + +dev_t +archive_entry_rdevminor(struct archive_entry *entry) +{ + if (entry->ae_stat.aest_rdev_is_broken_down) + return (entry->ae_stat.aest_rdevminor); + else + return minor(entry->ae_stat.aest_rdev); +} + +la_int64_t +archive_entry_size(struct archive_entry *entry) +{ + return (entry->ae_stat.aest_size); +} + +int +archive_entry_size_is_set(struct archive_entry *entry) +{ + return (entry->ae_set & AE_SET_SIZE); +} + +const char * +archive_entry_sourcepath(struct archive_entry *entry) +{ + const char *p; + if (archive_mstring_get_mbs( + entry->archive, &entry->ae_sourcepath, &p) == 0) + return (p); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (NULL); +} + +const wchar_t * +archive_entry_sourcepath_w(struct archive_entry *entry) +{ + const wchar_t *p; + if (archive_mstring_get_wcs( + entry->archive, &entry->ae_sourcepath, &p) == 0) + return (p); + return (NULL); +} + +const char * +archive_entry_symlink(struct archive_entry *entry) +{ + const char *p; + if ((entry->ae_set & AE_SET_SYMLINK) == 0) + return (NULL); + if (archive_mstring_get_mbs( + entry->archive, &entry->ae_symlink, &p) == 0) + return (p); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (NULL); +} + +int +archive_entry_symlink_type(struct archive_entry *entry) +{ + return (entry->ae_symlink_type); +} + +const char * +archive_entry_symlink_utf8(struct archive_entry *entry) +{ + const char *p; + if ((entry->ae_set & AE_SET_SYMLINK) == 0) + return (NULL); + if (archive_mstring_get_utf8( + entry->archive, &entry->ae_symlink, &p) == 0) + return (p); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (NULL); +} + +const wchar_t * +archive_entry_symlink_w(struct archive_entry *entry) +{ + const wchar_t *p; + if ((entry->ae_set & AE_SET_SYMLINK) == 0) + return (NULL); + if (archive_mstring_get_wcs( + entry->archive, &entry->ae_symlink, &p) == 0) + return (p); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (NULL); +} + +int +_archive_entry_symlink_l(struct archive_entry *entry, + const char **p, size_t *len, struct archive_string_conv *sc) +{ + if ((entry->ae_set & AE_SET_SYMLINK) == 0) { + *p = NULL; + *len = 0; + return (0); + } + return (archive_mstring_get_mbs_l( &entry->ae_symlink, p, len, sc)); +} + +la_int64_t +archive_entry_uid(struct archive_entry *entry) +{ + return (entry->ae_stat.aest_uid); +} + +const char * +archive_entry_uname(struct archive_entry *entry) +{ + const char *p; + if (archive_mstring_get_mbs(entry->archive, &entry->ae_uname, &p) == 0) + return (p); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (NULL); +} + +const char * +archive_entry_uname_utf8(struct archive_entry *entry) +{ + const char *p; + if (archive_mstring_get_utf8(entry->archive, &entry->ae_uname, &p) == 0) + return (p); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (NULL); +} + +const wchar_t * +archive_entry_uname_w(struct archive_entry *entry) +{ + const wchar_t *p; + if (archive_mstring_get_wcs(entry->archive, &entry->ae_uname, &p) == 0) + return (p); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (NULL); +} + +int +_archive_entry_uname_l(struct archive_entry *entry, + const char **p, size_t *len, struct archive_string_conv *sc) +{ + return (archive_mstring_get_mbs_l(&entry->ae_uname, p, len, sc)); +} + +int +archive_entry_is_data_encrypted(struct archive_entry *entry) +{ + return ((entry->encryption & AE_ENCRYPTION_DATA) == AE_ENCRYPTION_DATA); +} + +int +archive_entry_is_metadata_encrypted(struct archive_entry *entry) +{ + return ((entry->encryption & AE_ENCRYPTION_METADATA) == AE_ENCRYPTION_METADATA); +} + +int +archive_entry_is_encrypted(struct archive_entry *entry) +{ + return (entry->encryption & (AE_ENCRYPTION_DATA|AE_ENCRYPTION_METADATA)); +} + +/* + * Functions to set archive_entry properties. + */ + +void +archive_entry_set_filetype(struct archive_entry *entry, unsigned int type) +{ + entry->stat_valid = 0; + entry->acl.mode &= ~AE_IFMT; + entry->acl.mode |= AE_IFMT & type; +} + +void +archive_entry_set_fflags(struct archive_entry *entry, + unsigned long set, unsigned long clear) +{ + archive_mstring_clean(&entry->ae_fflags_text); + entry->ae_fflags_set = set; + entry->ae_fflags_clear = clear; +} + +const char * +archive_entry_copy_fflags_text(struct archive_entry *entry, + const char *flags) +{ + archive_mstring_copy_mbs(&entry->ae_fflags_text, flags); + return (ae_strtofflags(flags, + &entry->ae_fflags_set, &entry->ae_fflags_clear)); +} + +const wchar_t * +archive_entry_copy_fflags_text_w(struct archive_entry *entry, + const wchar_t *flags) +{ + archive_mstring_copy_wcs(&entry->ae_fflags_text, flags); + return (ae_wcstofflags(flags, + &entry->ae_fflags_set, &entry->ae_fflags_clear)); +} + +void +archive_entry_set_gid(struct archive_entry *entry, la_int64_t g) +{ + entry->stat_valid = 0; + entry->ae_stat.aest_gid = g; +} + +void +archive_entry_set_gname(struct archive_entry *entry, const char *name) +{ + archive_mstring_copy_mbs(&entry->ae_gname, name); +} + +void +archive_entry_set_gname_utf8(struct archive_entry *entry, const char *name) +{ + archive_mstring_copy_utf8(&entry->ae_gname, name); +} + +void +archive_entry_copy_gname(struct archive_entry *entry, const char *name) +{ + archive_mstring_copy_mbs(&entry->ae_gname, name); +} + +void +archive_entry_copy_gname_w(struct archive_entry *entry, const wchar_t *name) +{ + archive_mstring_copy_wcs(&entry->ae_gname, name); +} + +int +archive_entry_update_gname_utf8(struct archive_entry *entry, const char *name) +{ + if (archive_mstring_update_utf8(entry->archive, + &entry->ae_gname, name) == 0) + return (1); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (0); +} + +int +_archive_entry_copy_gname_l(struct archive_entry *entry, + const char *name, size_t len, struct archive_string_conv *sc) +{ + return (archive_mstring_copy_mbs_len_l(&entry->ae_gname, name, len, sc)); +} + +void +archive_entry_set_ino(struct archive_entry *entry, la_int64_t ino) +{ + entry->stat_valid = 0; + entry->ae_set |= AE_SET_INO; + entry->ae_stat.aest_ino = ino; +} + +void +archive_entry_set_ino64(struct archive_entry *entry, la_int64_t ino) +{ + entry->stat_valid = 0; + entry->ae_set |= AE_SET_INO; + entry->ae_stat.aest_ino = ino; +} + +void +archive_entry_set_hardlink(struct archive_entry *entry, const char *target) +{ + archive_mstring_copy_mbs(&entry->ae_hardlink, target); + if (target != NULL) + entry->ae_set |= AE_SET_HARDLINK; + else + entry->ae_set &= ~AE_SET_HARDLINK; +} + +void +archive_entry_set_hardlink_utf8(struct archive_entry *entry, const char *target) +{ + archive_mstring_copy_utf8(&entry->ae_hardlink, target); + if (target != NULL) + entry->ae_set |= AE_SET_HARDLINK; + else + entry->ae_set &= ~AE_SET_HARDLINK; +} + +void +archive_entry_copy_hardlink(struct archive_entry *entry, const char *target) +{ + archive_mstring_copy_mbs(&entry->ae_hardlink, target); + if (target != NULL) + entry->ae_set |= AE_SET_HARDLINK; + else + entry->ae_set &= ~AE_SET_HARDLINK; +} + +void +archive_entry_copy_hardlink_w(struct archive_entry *entry, const wchar_t *target) +{ + archive_mstring_copy_wcs(&entry->ae_hardlink, target); + if (target != NULL) + entry->ae_set |= AE_SET_HARDLINK; + else + entry->ae_set &= ~AE_SET_HARDLINK; +} + +int +archive_entry_update_hardlink_utf8(struct archive_entry *entry, const char *target) +{ + if (target != NULL) + entry->ae_set |= AE_SET_HARDLINK; + else + entry->ae_set &= ~AE_SET_HARDLINK; + if (archive_mstring_update_utf8(entry->archive, + &entry->ae_hardlink, target) == 0) + return (1); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (0); +} + +int +_archive_entry_copy_hardlink_l(struct archive_entry *entry, + const char *target, size_t len, struct archive_string_conv *sc) +{ + int r; + + r = archive_mstring_copy_mbs_len_l(&entry->ae_hardlink, + target, len, sc); + if (target != NULL && r == 0) + entry->ae_set |= AE_SET_HARDLINK; + else + entry->ae_set &= ~AE_SET_HARDLINK; + return (r); +} + +void +archive_entry_set_atime(struct archive_entry *entry, time_t t, long ns) +{ + FIX_NS(t, ns); + entry->stat_valid = 0; + entry->ae_set |= AE_SET_ATIME; + entry->ae_stat.aest_atime = t; + entry->ae_stat.aest_atime_nsec = ns; +} + +void +archive_entry_unset_atime(struct archive_entry *entry) +{ + archive_entry_set_atime(entry, 0, 0); + entry->ae_set &= ~AE_SET_ATIME; +} + +void +archive_entry_set_birthtime(struct archive_entry *entry, time_t t, long ns) +{ + FIX_NS(t, ns); + entry->stat_valid = 0; + entry->ae_set |= AE_SET_BIRTHTIME; + entry->ae_stat.aest_birthtime = t; + entry->ae_stat.aest_birthtime_nsec = ns; +} + +void +archive_entry_unset_birthtime(struct archive_entry *entry) +{ + archive_entry_set_birthtime(entry, 0, 0); + entry->ae_set &= ~AE_SET_BIRTHTIME; +} + +void +archive_entry_set_ctime(struct archive_entry *entry, time_t t, long ns) +{ + FIX_NS(t, ns); + entry->stat_valid = 0; + entry->ae_set |= AE_SET_CTIME; + entry->ae_stat.aest_ctime = t; + entry->ae_stat.aest_ctime_nsec = ns; +} + +void +archive_entry_unset_ctime(struct archive_entry *entry) +{ + archive_entry_set_ctime(entry, 0, 0); + entry->ae_set &= ~AE_SET_CTIME; +} + +void +archive_entry_set_dev(struct archive_entry *entry, dev_t d) +{ + entry->stat_valid = 0; + entry->ae_set |= AE_SET_DEV; + entry->ae_stat.aest_dev_is_broken_down = 0; + entry->ae_stat.aest_dev = d; +} + +void +archive_entry_set_devmajor(struct archive_entry *entry, dev_t m) +{ + entry->stat_valid = 0; + entry->ae_set |= AE_SET_DEV; + entry->ae_stat.aest_dev_is_broken_down = 1; + entry->ae_stat.aest_devmajor = m; +} + +void +archive_entry_set_devminor(struct archive_entry *entry, dev_t m) +{ + entry->stat_valid = 0; + entry->ae_set |= AE_SET_DEV; + entry->ae_stat.aest_dev_is_broken_down = 1; + entry->ae_stat.aest_devminor = m; +} + +/* Set symlink if symlink is already set, else set hardlink. */ +void +archive_entry_set_link(struct archive_entry *entry, const char *target) +{ + if (entry->ae_set & AE_SET_SYMLINK) + archive_mstring_copy_mbs(&entry->ae_symlink, target); + else + archive_mstring_copy_mbs(&entry->ae_hardlink, target); +} + +void +archive_entry_set_link_utf8(struct archive_entry *entry, const char *target) +{ + if (entry->ae_set & AE_SET_SYMLINK) + archive_mstring_copy_utf8(&entry->ae_symlink, target); + else + archive_mstring_copy_utf8(&entry->ae_hardlink, target); +} + +/* Set symlink if symlink is already set, else set hardlink. */ +void +archive_entry_copy_link(struct archive_entry *entry, const char *target) +{ + if (entry->ae_set & AE_SET_SYMLINK) + archive_mstring_copy_mbs(&entry->ae_symlink, target); + else + archive_mstring_copy_mbs(&entry->ae_hardlink, target); +} + +/* Set symlink if symlink is already set, else set hardlink. */ +void +archive_entry_copy_link_w(struct archive_entry *entry, const wchar_t *target) +{ + if (entry->ae_set & AE_SET_SYMLINK) + archive_mstring_copy_wcs(&entry->ae_symlink, target); + else + archive_mstring_copy_wcs(&entry->ae_hardlink, target); +} + +int +archive_entry_update_link_utf8(struct archive_entry *entry, const char *target) +{ + int r; + if (entry->ae_set & AE_SET_SYMLINK) + r = archive_mstring_update_utf8(entry->archive, + &entry->ae_symlink, target); + else + r = archive_mstring_update_utf8(entry->archive, + &entry->ae_hardlink, target); + if (r == 0) + return (1); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (0); +} + +int +_archive_entry_copy_link_l(struct archive_entry *entry, + const char *target, size_t len, struct archive_string_conv *sc) +{ + int r; + + if (entry->ae_set & AE_SET_SYMLINK) + r = archive_mstring_copy_mbs_len_l(&entry->ae_symlink, + target, len, sc); + else + r = archive_mstring_copy_mbs_len_l(&entry->ae_hardlink, + target, len, sc); + return (r); +} + +void +archive_entry_set_mode(struct archive_entry *entry, mode_t m) +{ + entry->stat_valid = 0; + entry->acl.mode = m; +} + +void +archive_entry_set_mtime(struct archive_entry *entry, time_t t, long ns) +{ + FIX_NS(t, ns); + entry->stat_valid = 0; + entry->ae_set |= AE_SET_MTIME; + entry->ae_stat.aest_mtime = t; + entry->ae_stat.aest_mtime_nsec = ns; +} + +void +archive_entry_unset_mtime(struct archive_entry *entry) +{ + archive_entry_set_mtime(entry, 0, 0); + entry->ae_set &= ~AE_SET_MTIME; +} + +void +archive_entry_set_nlink(struct archive_entry *entry, unsigned int nlink) +{ + entry->stat_valid = 0; + entry->ae_stat.aest_nlink = nlink; +} + +void +archive_entry_set_pathname(struct archive_entry *entry, const char *name) +{ + archive_mstring_copy_mbs(&entry->ae_pathname, name); +} + +void +archive_entry_set_pathname_utf8(struct archive_entry *entry, const char *name) +{ + archive_mstring_copy_utf8(&entry->ae_pathname, name); +} + +void +archive_entry_copy_pathname(struct archive_entry *entry, const char *name) +{ + archive_mstring_copy_mbs(&entry->ae_pathname, name); +} + +void +archive_entry_copy_pathname_w(struct archive_entry *entry, const wchar_t *name) +{ + archive_mstring_copy_wcs(&entry->ae_pathname, name); +} + +int +archive_entry_update_pathname_utf8(struct archive_entry *entry, const char *name) +{ + if (archive_mstring_update_utf8(entry->archive, + &entry->ae_pathname, name) == 0) + return (1); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (0); +} + +int +_archive_entry_copy_pathname_l(struct archive_entry *entry, + const char *name, size_t len, struct archive_string_conv *sc) +{ + return (archive_mstring_copy_mbs_len_l(&entry->ae_pathname, + name, len, sc)); +} + +void +archive_entry_set_perm(struct archive_entry *entry, mode_t p) +{ + entry->stat_valid = 0; + entry->acl.mode &= AE_IFMT; + entry->acl.mode |= ~AE_IFMT & p; +} + +void +archive_entry_set_rdev(struct archive_entry *entry, dev_t m) +{ + entry->stat_valid = 0; + entry->ae_stat.aest_rdev = m; + entry->ae_stat.aest_rdev_is_broken_down = 0; +} + +void +archive_entry_set_rdevmajor(struct archive_entry *entry, dev_t m) +{ + entry->stat_valid = 0; + entry->ae_stat.aest_rdev_is_broken_down = 1; + entry->ae_stat.aest_rdevmajor = m; +} + +void +archive_entry_set_rdevminor(struct archive_entry *entry, dev_t m) +{ + entry->stat_valid = 0; + entry->ae_stat.aest_rdev_is_broken_down = 1; + entry->ae_stat.aest_rdevminor = m; +} + +void +archive_entry_set_size(struct archive_entry *entry, la_int64_t s) +{ + entry->stat_valid = 0; + entry->ae_stat.aest_size = s; + entry->ae_set |= AE_SET_SIZE; +} + +void +archive_entry_unset_size(struct archive_entry *entry) +{ + archive_entry_set_size(entry, 0); + entry->ae_set &= ~AE_SET_SIZE; +} + +void +archive_entry_copy_sourcepath(struct archive_entry *entry, const char *path) +{ + archive_mstring_copy_mbs(&entry->ae_sourcepath, path); +} + +void +archive_entry_copy_sourcepath_w(struct archive_entry *entry, const wchar_t *path) +{ + archive_mstring_copy_wcs(&entry->ae_sourcepath, path); +} + +void +archive_entry_set_symlink(struct archive_entry *entry, const char *linkname) +{ + archive_mstring_copy_mbs(&entry->ae_symlink, linkname); + if (linkname != NULL) + entry->ae_set |= AE_SET_SYMLINK; + else + entry->ae_set &= ~AE_SET_SYMLINK; +} + +void +archive_entry_set_symlink_type(struct archive_entry *entry, int type) +{ + entry->ae_symlink_type = type; +} + +void +archive_entry_set_symlink_utf8(struct archive_entry *entry, const char *linkname) +{ + archive_mstring_copy_utf8(&entry->ae_symlink, linkname); + if (linkname != NULL) + entry->ae_set |= AE_SET_SYMLINK; + else + entry->ae_set &= ~AE_SET_SYMLINK; +} + +void +archive_entry_copy_symlink(struct archive_entry *entry, const char *linkname) +{ + archive_mstring_copy_mbs(&entry->ae_symlink, linkname); + if (linkname != NULL) + entry->ae_set |= AE_SET_SYMLINK; + else + entry->ae_set &= ~AE_SET_SYMLINK; +} + +void +archive_entry_copy_symlink_w(struct archive_entry *entry, const wchar_t *linkname) +{ + archive_mstring_copy_wcs(&entry->ae_symlink, linkname); + if (linkname != NULL) + entry->ae_set |= AE_SET_SYMLINK; + else + entry->ae_set &= ~AE_SET_SYMLINK; +} + +int +archive_entry_update_symlink_utf8(struct archive_entry *entry, const char *linkname) +{ + if (linkname != NULL) + entry->ae_set |= AE_SET_SYMLINK; + else + entry->ae_set &= ~AE_SET_SYMLINK; + if (archive_mstring_update_utf8(entry->archive, + &entry->ae_symlink, linkname) == 0) + return (1); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (0); +} + +int +_archive_entry_copy_symlink_l(struct archive_entry *entry, + const char *linkname, size_t len, struct archive_string_conv *sc) +{ + int r; + + r = archive_mstring_copy_mbs_len_l(&entry->ae_symlink, + linkname, len, sc); + if (linkname != NULL && r == 0) + entry->ae_set |= AE_SET_SYMLINK; + else + entry->ae_set &= ~AE_SET_SYMLINK; + return (r); +} + +void +archive_entry_set_uid(struct archive_entry *entry, la_int64_t u) +{ + entry->stat_valid = 0; + entry->ae_stat.aest_uid = u; +} + +void +archive_entry_set_uname(struct archive_entry *entry, const char *name) +{ + archive_mstring_copy_mbs(&entry->ae_uname, name); +} + +void +archive_entry_set_uname_utf8(struct archive_entry *entry, const char *name) +{ + archive_mstring_copy_utf8(&entry->ae_uname, name); +} + +void +archive_entry_copy_uname(struct archive_entry *entry, const char *name) +{ + archive_mstring_copy_mbs(&entry->ae_uname, name); +} + +void +archive_entry_copy_uname_w(struct archive_entry *entry, const wchar_t *name) +{ + archive_mstring_copy_wcs(&entry->ae_uname, name); +} + +int +archive_entry_update_uname_utf8(struct archive_entry *entry, const char *name) +{ + if (archive_mstring_update_utf8(entry->archive, + &entry->ae_uname, name) == 0) + return (1); + if (errno == ENOMEM) + __archive_errx(1, "No memory"); + return (0); +} + +void +archive_entry_set_is_data_encrypted(struct archive_entry *entry, char is_encrypted) +{ + if (is_encrypted) { + entry->encryption |= AE_ENCRYPTION_DATA; + } else { + entry->encryption &= ~AE_ENCRYPTION_DATA; + } +} + +void +archive_entry_set_is_metadata_encrypted(struct archive_entry *entry, char is_encrypted) +{ + if (is_encrypted) { + entry->encryption |= AE_ENCRYPTION_METADATA; + } else { + entry->encryption &= ~AE_ENCRYPTION_METADATA; + } +} + +int +_archive_entry_copy_uname_l(struct archive_entry *entry, + const char *name, size_t len, struct archive_string_conv *sc) +{ + return (archive_mstring_copy_mbs_len_l(&entry->ae_uname, + name, len, sc)); +} + +const void * +archive_entry_mac_metadata(struct archive_entry *entry, size_t *s) +{ + *s = entry->mac_metadata_size; + return entry->mac_metadata; +} + +void +archive_entry_copy_mac_metadata(struct archive_entry *entry, + const void *p, size_t s) +{ + free(entry->mac_metadata); + if (p == NULL || s == 0) { + entry->mac_metadata = NULL; + entry->mac_metadata_size = 0; + } else { + entry->mac_metadata_size = s; + entry->mac_metadata = malloc(s); + if (entry->mac_metadata == NULL) + abort(); + memcpy(entry->mac_metadata, p, s); + } +} + +/* + * ACL management. The following would, of course, be a lot simpler + * if: 1) the last draft of POSIX.1e were a really thorough and + * complete standard that addressed the needs of ACL archiving and 2) + * everyone followed it faithfully. Alas, neither is true, so the + * following is a lot more complex than might seem necessary to the + * uninitiated. + */ + +struct archive_acl * +archive_entry_acl(struct archive_entry *entry) +{ + return &entry->acl; +} + +void +archive_entry_acl_clear(struct archive_entry *entry) +{ + archive_acl_clear(&entry->acl); +} + +/* + * Add a single ACL entry to the internal list of ACL data. + */ +int +archive_entry_acl_add_entry(struct archive_entry *entry, + int type, int permset, int tag, int id, const char *name) +{ + return archive_acl_add_entry(&entry->acl, type, permset, tag, id, name); +} + +/* + * As above, but with a wide-character name. + */ +int +archive_entry_acl_add_entry_w(struct archive_entry *entry, + int type, int permset, int tag, int id, const wchar_t *name) +{ + return archive_acl_add_entry_w_len(&entry->acl, + type, permset, tag, id, name, wcslen(name)); +} + +/* + * Return a bitmask of ACL types in an archive entry ACL list + */ +int +archive_entry_acl_types(struct archive_entry *entry) +{ + return (archive_acl_types(&entry->acl)); +} + +/* + * Return a count of entries matching "want_type". + */ +int +archive_entry_acl_count(struct archive_entry *entry, int want_type) +{ + return archive_acl_count(&entry->acl, want_type); +} + +/* + * Prepare for reading entries from the ACL data. Returns a count + * of entries matching "want_type", or zero if there are no + * non-extended ACL entries of that type. + */ +int +archive_entry_acl_reset(struct archive_entry *entry, int want_type) +{ + return archive_acl_reset(&entry->acl, want_type); +} + +/* + * Return the next ACL entry in the list. Fake entries for the + * standard permissions and include them in the returned list. + */ +int +archive_entry_acl_next(struct archive_entry *entry, int want_type, int *type, + int *permset, int *tag, int *id, const char **name) +{ + int r; + r = archive_acl_next(entry->archive, &entry->acl, want_type, type, + permset, tag, id, name); + if (r == ARCHIVE_FATAL && errno == ENOMEM) + __archive_errx(1, "No memory"); + return (r); +} + +/* + * Generate a text version of the ACL. The flags parameter controls + * the style of the generated ACL. + */ +wchar_t * +archive_entry_acl_to_text_w(struct archive_entry *entry, la_ssize_t *len, + int flags) +{ + return (archive_acl_to_text_w(&entry->acl, len, flags, + entry->archive)); +} + +char * +archive_entry_acl_to_text(struct archive_entry *entry, la_ssize_t *len, + int flags) +{ + return (archive_acl_to_text_l(&entry->acl, len, flags, NULL)); +} + +char * +_archive_entry_acl_to_text_l(struct archive_entry *entry, ssize_t *len, + int flags, struct archive_string_conv *sc) +{ + return (archive_acl_to_text_l(&entry->acl, len, flags, sc)); +} + +/* + * ACL text parser. + */ +int +archive_entry_acl_from_text_w(struct archive_entry *entry, + const wchar_t *wtext, int type) +{ + return (archive_acl_from_text_w(&entry->acl, wtext, type)); +} + +int +archive_entry_acl_from_text(struct archive_entry *entry, + const char *text, int type) +{ + return (archive_acl_from_text_l(&entry->acl, text, type, NULL)); +} + +int +_archive_entry_acl_from_text_l(struct archive_entry *entry, const char *text, + int type, struct archive_string_conv *sc) +{ + return (archive_acl_from_text_l(&entry->acl, text, type, sc)); +} + +/* Deprecated */ +static int +archive_entry_acl_text_compat(int *flags) +{ + if ((*flags & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) == 0) + return (1); + + /* ABI compat with old ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID */ + if ((*flags & OLD_ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID) != 0) + *flags |= ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID; + + /* ABI compat with old ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT */ + if ((*flags & OLD_ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT) != 0) + *flags |= ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT; + + *flags |= ARCHIVE_ENTRY_ACL_STYLE_SEPARATOR_COMMA; + + return (0); +} + +/* Deprecated */ +const wchar_t * +archive_entry_acl_text_w(struct archive_entry *entry, int flags) +{ + free(entry->acl.acl_text_w); + entry->acl.acl_text_w = NULL; + if (archive_entry_acl_text_compat(&flags) == 0) + entry->acl.acl_text_w = archive_acl_to_text_w(&entry->acl, + NULL, flags, entry->archive); + return (entry->acl.acl_text_w); +} + +/* Deprecated */ +const char * +archive_entry_acl_text(struct archive_entry *entry, int flags) +{ + free(entry->acl.acl_text); + entry->acl.acl_text = NULL; + if (archive_entry_acl_text_compat(&flags) == 0) + entry->acl.acl_text = archive_acl_to_text_l(&entry->acl, NULL, + flags, NULL); + + return (entry->acl.acl_text); +} + +/* Deprecated */ +int +_archive_entry_acl_text_l(struct archive_entry *entry, int flags, + const char **acl_text, size_t *len, struct archive_string_conv *sc) +{ + free(entry->acl.acl_text); + entry->acl.acl_text = NULL; + + if (archive_entry_acl_text_compat(&flags) == 0) + entry->acl.acl_text = archive_acl_to_text_l(&entry->acl, + (ssize_t *)len, flags, sc); + + *acl_text = entry->acl.acl_text; + + return (0); +} + +/* + * Following code is modified from UC Berkeley sources, and + * is subject to the following copyright notice. + */ + +/*- + * Copyright (c) 1993 + * The Regents of the University of California. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + +/* + * Supported file flags on FreeBSD and Mac OS: + * sappnd,sappend SF_APPEND + * arch,archived SF_ARCHIVED + * schg,schange,simmutable SF_IMMUTABLE + * sunlnk,sunlink SF_NOUNLINK (FreeBSD only) + * uappnd,uappend UF_APPEND + * compressed UF_COMPRESSED (Mac OS only) + * hidden,uhidden UF_HIDDEN + * uchg,uchange,uimmutable UF_IMMUTABLE + * nodump UF_NODUMP + * uunlnk,uunlink UF_NOUNLINK (FreeBSD only) + * offline,uoffline UF_OFFLINE (FreeBSD only) + * opaque UF_OPAQUE + * rdonly,urdonly,readonly UF_READONLY (FreeBSD only) + * reparse,ureparse UF_REPARSE (FreeBSD only) + * sparse,usparse UF_SPARSE (FreeBSD only) + * system,usystem UF_SYSTEM (FreeBSD only) + * + * See chflags(2) for more information + * + * Supported file attributes on Linux: + * a append only FS_APPEND_FL sappnd + * A no atime updates FS_NOATIME_FL atime + * c compress FS_COMPR_FL compress + * C no copy on write FS_NOCOW_FL cow + * d no dump FS_NODUMP_FL dump + * D synchronous directory updates FS_DIRSYNC_FL dirsync + * i immutable FS_IMMUTABLE_FL schg + * j data journalling FS_JOURNAL_DATA_FL journal + * P project hierarchy FS_PROJINHERIT_FL projinherit + * s secure deletion FS_SECRM_FL securedeletion + * S synchronous updates FS_SYNC_FL sync + * t no tail-merging FS_NOTAIL_FL tail + * T top of directory hierarchy FS_TOPDIR_FL topdir + * u undeletable FS_UNRM_FL undel + * + * See ioctl_iflags(2) for more information + * + * Equivalent file flags supported on FreeBSD / Mac OS and Linux: + * SF_APPEND FS_APPEND_FL sappnd + * SF_IMMUTABLE FS_IMMUTABLE_FL schg + * UF_NODUMP FS_NODUMP_FL nodump + */ + +static const struct flag { + const char *name; + const wchar_t *wname; + unsigned long set; + unsigned long clear; +} fileflags[] = { + /* Preferred (shorter) names per flag first, all prefixed by "no" */ +#ifdef SF_APPEND + { "nosappnd", L"nosappnd", SF_APPEND, 0}, + { "nosappend", L"nosappend", SF_APPEND, 0}, +#endif +#if defined(FS_APPEND_FL) /* 'a' */ + { "nosappnd", L"nosappnd", FS_APPEND_FL, 0}, + { "nosappend", L"nosappend", FS_APPEND_FL, 0}, +#elif defined(EXT2_APPEND_FL) /* 'a' */ + { "nosappnd", L"nosappnd", EXT2_APPEND_FL, 0}, + { "nosappend", L"nosappend", EXT2_APPEND_FL, 0}, +#endif +#ifdef SF_ARCHIVED + { "noarch", L"noarch", SF_ARCHIVED, 0}, + { "noarchived", L"noarchived", SF_ARCHIVED, 0}, +#endif +#ifdef SF_IMMUTABLE + { "noschg", L"noschg", SF_IMMUTABLE, 0}, + { "noschange", L"noschange", SF_IMMUTABLE, 0}, + { "nosimmutable", L"nosimmutable", SF_IMMUTABLE, 0}, +#endif +#if defined(FS_IMMUTABLE_FL) /* 'i' */ + { "noschg", L"noschg", FS_IMMUTABLE_FL, 0}, + { "noschange", L"noschange", FS_IMMUTABLE_FL, 0}, + { "nosimmutable", L"nosimmutable", FS_IMMUTABLE_FL, 0}, +#elif defined(EXT2_IMMUTABLE_FL) /* 'i' */ + { "noschg", L"noschg", EXT2_IMMUTABLE_FL, 0}, + { "noschange", L"noschange", EXT2_IMMUTABLE_FL, 0}, + { "nosimmutable", L"nosimmutable", EXT2_IMMUTABLE_FL, 0}, +#endif +#ifdef SF_NOUNLINK + { "nosunlnk", L"nosunlnk", SF_NOUNLINK, 0}, + { "nosunlink", L"nosunlink", SF_NOUNLINK, 0}, +#endif +#ifdef UF_APPEND + { "nouappnd", L"nouappnd", UF_APPEND, 0}, + { "nouappend", L"nouappend", UF_APPEND, 0}, +#endif +#ifdef UF_IMMUTABLE + { "nouchg", L"nouchg", UF_IMMUTABLE, 0}, + { "nouchange", L"nouchange", UF_IMMUTABLE, 0}, + { "nouimmutable", L"nouimmutable", UF_IMMUTABLE, 0}, +#endif +#ifdef UF_NODUMP + { "nodump", L"nodump", 0, UF_NODUMP}, +#endif +#if defined(FS_NODUMP_FL) /* 'd' */ + { "nodump", L"nodump", 0, FS_NODUMP_FL}, +#elif defined(EXT2_NODUMP_FL) + { "nodump", L"nodump", 0, EXT2_NODUMP_FL}, +#endif +#ifdef UF_OPAQUE + { "noopaque", L"noopaque", UF_OPAQUE, 0}, +#endif +#ifdef UF_NOUNLINK + { "nouunlnk", L"nouunlnk", UF_NOUNLINK, 0}, + { "nouunlink", L"nouunlink", UF_NOUNLINK, 0}, +#endif +#ifdef UF_COMPRESSED + /* Mac OS */ + { "nocompressed", L"nocompressed", UF_COMPRESSED, 0}, +#endif +#ifdef UF_HIDDEN + { "nohidden", L"nohidden", UF_HIDDEN, 0}, + { "nouhidden", L"nouhidden", UF_HIDDEN, 0}, +#endif +#ifdef FILE_ATTRIBUTE_HIDDEN + { "nohidden", L"nohidden", FILE_ATTRIBUTE_HIDDEN, 0}, + { "nouhidden", L"nouhidden", FILE_ATTRIBUTE_HIDDEN, 0}, +#endif +#ifdef UF_OFFLINE + { "nooffline", L"nooffline", UF_OFFLINE, 0}, + { "nouoffline", L"nouoffline", UF_OFFLINE, 0}, +#endif +#ifdef UF_READONLY + { "nordonly", L"nordonly", UF_READONLY, 0}, + { "nourdonly", L"nourdonly", UF_READONLY, 0}, + { "noreadonly", L"noreadonly", UF_READONLY, 0}, +#endif +#ifdef FILE_ATTRIBUTE_READONLY + { "nordonly", L"nordonly", FILE_ATTRIBUTE_READONLY, 0}, + { "nourdonly", L"nourdonly", FILE_ATTRIBUTE_READONLY, 0}, + { "noreadonly", L"noreadonly", FILE_ATTRIBUTE_READONLY, 0}, +#endif +#ifdef UF_SPARSE + { "nosparse", L"nosparse", UF_SPARSE, 0}, + { "nousparse", L"nousparse", UF_SPARSE, 0}, +#endif +#ifdef UF_REPARSE + { "noreparse", L"noreparse", UF_REPARSE, 0}, + { "noureparse", L"noureparse", UF_REPARSE, 0}, +#endif +#ifdef UF_SYSTEM + { "nosystem", L"nosystem", UF_SYSTEM, 0}, + { "nousystem", L"nousystem", UF_SYSTEM, 0}, +#endif +#ifdef FILE_ATTRIBUTE_SYSTEM + { "nosystem", L"nosystem", FILE_ATTRIBUTE_SYSTEM, 0}, + { "nousystem", L"nousystem", FILE_ATTRIBUTE_SYSTEM, 0}, +#endif +#if defined(FS_UNRM_FL) /* 'u' */ + { "noundel", L"noundel", FS_UNRM_FL, 0}, +#elif defined(EXT2_UNRM_FL) + { "noundel", L"noundel", EXT2_UNRM_FL, 0}, +#endif + +#if defined(FS_COMPR_FL) /* 'c' */ + { "nocompress", L"nocompress", FS_COMPR_FL, 0}, +#elif defined(EXT2_COMPR_FL) + { "nocompress", L"nocompress", EXT2_COMPR_FL, 0}, +#endif + +#if defined(FS_NOATIME_FL) /* 'A' */ + { "noatime", L"noatime", 0, FS_NOATIME_FL}, +#elif defined(EXT2_NOATIME_FL) + { "noatime", L"noatime", 0, EXT2_NOATIME_FL}, +#endif +#if defined(FS_DIRSYNC_FL) /* 'D' */ + { "nodirsync", L"nodirsync", FS_DIRSYNC_FL, 0}, +#elif defined(EXT2_DIRSYNC_FL) + { "nodirsync", L"nodirsync", EXT2_DIRSYNC_FL, 0}, +#endif +#if defined(FS_JOURNAL_DATA_FL) /* 'j' */ + { "nojournal-data",L"nojournal-data", FS_JOURNAL_DATA_FL, 0}, + { "nojournal", L"nojournal", FS_JOURNAL_DATA_FL, 0}, +#elif defined(EXT3_JOURNAL_DATA_FL) + { "nojournal-data",L"nojournal-data", EXT3_JOURNAL_DATA_FL, 0}, + { "nojournal", L"nojournal", EXT3_JOURNAL_DATA_FL, 0}, +#endif +#if defined(FS_SECRM_FL) /* 's' */ + { "nosecdel", L"nosecdel", FS_SECRM_FL, 0}, + { "nosecuredeletion",L"nosecuredeletion",FS_SECRM_FL, 0}, +#elif defined(EXT2_SECRM_FL) + { "nosecdel", L"nosecdel", EXT2_SECRM_FL, 0}, + { "nosecuredeletion",L"nosecuredeletion",EXT2_SECRM_FL, 0}, +#endif +#if defined(FS_SYNC_FL) /* 'S' */ + { "nosync", L"nosync", FS_SYNC_FL, 0}, +#elif defined(EXT2_SYNC_FL) + { "nosync", L"nosync", EXT2_SYNC_FL, 0}, +#endif +#if defined(FS_NOTAIL_FL) /* 't' */ + { "notail", L"notail", 0, FS_NOTAIL_FL}, +#elif defined(EXT2_NOTAIL_FL) + { "notail", L"notail", 0, EXT2_NOTAIL_FL}, +#endif +#if defined(FS_TOPDIR_FL) /* 'T' */ + { "notopdir", L"notopdir", FS_TOPDIR_FL, 0}, +#elif defined(EXT2_TOPDIR_FL) + { "notopdir", L"notopdir", EXT2_TOPDIR_FL, 0}, +#endif +#ifdef FS_NOCOW_FL /* 'C' */ + { "nocow", L"nocow", 0, FS_NOCOW_FL}, +#endif +#ifdef FS_PROJINHERIT_FL /* 'P' */ + { "noprojinherit",L"noprojinherit", FS_PROJINHERIT_FL, 0}, +#endif + { NULL, NULL, 0, 0} +}; + +/* + * fflagstostr -- + * Convert file flags to a comma-separated string. If no flags + * are set, return the empty string. + */ +static char * +ae_fflagstostr(unsigned long bitset, unsigned long bitclear) +{ + char *string, *dp; + const char *sp; + unsigned long bits; + const struct flag *flag; + size_t length; + + bits = bitset | bitclear; + length = 0; + for (flag = fileflags; flag->name != NULL; flag++) + if (bits & (flag->set | flag->clear)) { + length += strlen(flag->name) + 1; + bits &= ~(flag->set | flag->clear); + } + + if (length == 0) + return (NULL); + string = (char *)malloc(length); + if (string == NULL) + return (NULL); + + dp = string; + for (flag = fileflags; flag->name != NULL; flag++) { + if (bitset & flag->set || bitclear & flag->clear) { + sp = flag->name + 2; + } else if (bitset & flag->clear || bitclear & flag->set) { + sp = flag->name; + } else + continue; + bitset &= ~(flag->set | flag->clear); + bitclear &= ~(flag->set | flag->clear); + if (dp > string) + *dp++ = ','; + while ((*dp++ = *sp++) != '\0') + ; + dp--; + } + + *dp = '\0'; + return (string); +} + +/* + * strtofflags -- + * Take string of arguments and return file flags. This + * version works a little differently than strtofflags(3). + * In particular, it always tests every token, skipping any + * unrecognized tokens. It returns a pointer to the first + * unrecognized token, or NULL if every token was recognized. + * This version is also const-correct and does not modify the + * provided string. + */ +static const char * +ae_strtofflags(const char *s, unsigned long *setp, unsigned long *clrp) +{ + const char *start, *end; + const struct flag *flag; + unsigned long set, clear; + const char *failed; + + set = clear = 0; + start = s; + failed = NULL; + /* Find start of first token. */ + while (*start == '\t' || *start == ' ' || *start == ',') + start++; + while (*start != '\0') { + size_t length; + /* Locate end of token. */ + end = start; + while (*end != '\0' && *end != '\t' && + *end != ' ' && *end != ',') + end++; + length = end - start; + for (flag = fileflags; flag->name != NULL; flag++) { + size_t flag_length = strlen(flag->name); + if (length == flag_length + && memcmp(start, flag->name, length) == 0) { + /* Matched "noXXXX", so reverse the sense. */ + clear |= flag->set; + set |= flag->clear; + break; + } else if (length == flag_length - 2 + && memcmp(start, flag->name + 2, length) == 0) { + /* Matched "XXXX", so don't reverse. */ + set |= flag->set; + clear |= flag->clear; + break; + } + } + /* Ignore unknown flag names. */ + if (flag->name == NULL && failed == NULL) + failed = start; + + /* Find start of next token. */ + start = end; + while (*start == '\t' || *start == ' ' || *start == ',') + start++; + + } + + if (setp) + *setp = set; + if (clrp) + *clrp = clear; + + /* Return location of first failure. */ + return (failed); +} + +/* + * wcstofflags -- + * Take string of arguments and return file flags. This + * version works a little differently than strtofflags(3). + * In particular, it always tests every token, skipping any + * unrecognized tokens. It returns a pointer to the first + * unrecognized token, or NULL if every token was recognized. + * This version is also const-correct and does not modify the + * provided string. + */ +static const wchar_t * +ae_wcstofflags(const wchar_t *s, unsigned long *setp, unsigned long *clrp) +{ + const wchar_t *start, *end; + const struct flag *flag; + unsigned long set, clear; + const wchar_t *failed; + + set = clear = 0; + start = s; + failed = NULL; + /* Find start of first token. */ + while (*start == L'\t' || *start == L' ' || *start == L',') + start++; + while (*start != L'\0') { + size_t length; + /* Locate end of token. */ + end = start; + while (*end != L'\0' && *end != L'\t' && + *end != L' ' && *end != L',') + end++; + length = end - start; + for (flag = fileflags; flag->wname != NULL; flag++) { + size_t flag_length = wcslen(flag->wname); + if (length == flag_length + && wmemcmp(start, flag->wname, length) == 0) { + /* Matched "noXXXX", so reverse the sense. */ + clear |= flag->set; + set |= flag->clear; + break; + } else if (length == flag_length - 2 + && wmemcmp(start, flag->wname + 2, length) == 0) { + /* Matched "XXXX", so don't reverse. */ + set |= flag->set; + clear |= flag->clear; + break; + } + } + /* Ignore unknown flag names. */ + if (flag->wname == NULL && failed == NULL) + failed = start; + + /* Find start of next token. */ + start = end; + while (*start == L'\t' || *start == L' ' || *start == L',') + start++; + + } + + if (setp) + *setp = set; + if (clrp) + *clrp = clear; + + /* Return location of first failure. */ + return (failed); +} + + +#ifdef TEST +#include +int +main(int argc, char **argv) +{ + struct archive_entry *entry = archive_entry_new(); + unsigned long set, clear; + const wchar_t *remainder; + + remainder = archive_entry_copy_fflags_text_w(entry, L"nosappnd dump archive,,,,,,,"); + archive_entry_fflags(entry, &set, &clear); + + wprintf(L"set=0x%lX clear=0x%lX remainder='%ls'\n", set, clear, remainder); + + wprintf(L"new flags='%s'\n", archive_entry_fflags_text(entry)); + return (0); +} +#endif diff --git a/c/archive_entry.h b/c/archive_entry.h new file mode 100644 --- /dev/null +++ b/c/archive_entry.h @@ -0,0 +1,708 @@ +/*- + * Copyright (c) 2003-2008 Tim Kientzle + * Copyright (c) 2016 Martin Matuska + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD: head/lib/libarchive/archive_entry.h 201096 2009-12-28 02:41:27Z kientzle $ + */ + +#ifndef ARCHIVE_ENTRY_H_INCLUDED +#define ARCHIVE_ENTRY_H_INCLUDED + +/* Note: Compiler will complain if this does not match archive.h! */ +#define ARCHIVE_VERSION_NUMBER 3004003 + +/* + * Note: archive_entry.h is for use outside of libarchive; the + * configuration headers (config.h, archive_platform.h, etc.) are + * purely internal. Do NOT use HAVE_XXX configuration macros to + * control the behavior of this header! If you must conditionalize, + * use predefined compiler and/or platform macros. + */ + +#include +#include /* for wchar_t */ +#include +#include + +#if defined(_WIN32) && !defined(__CYGWIN__) +#include +#endif + +/* Get a suitable 64-bit integer type. */ +#if !defined(__LA_INT64_T_DEFINED) +# if ARCHIVE_VERSION_NUMBER < 4000000 +#define __LA_INT64_T la_int64_t +# endif +#define __LA_INT64_T_DEFINED +# if defined(_WIN32) && !defined(__CYGWIN__) && !defined(__WATCOMC__) +typedef __int64 la_int64_t; +# else +#include +# if defined(_SCO_DS) || defined(__osf__) +typedef long long la_int64_t; +# else +typedef int64_t la_int64_t; +# endif +# endif +#endif + +/* The la_ssize_t should match the type used in 'struct stat' */ +#if !defined(__LA_SSIZE_T_DEFINED) +/* Older code relied on the __LA_SSIZE_T macro; after 4.0 we'll switch to the typedef exclusively. */ +# if ARCHIVE_VERSION_NUMBER < 4000000 +#define __LA_SSIZE_T la_ssize_t +# endif +#define __LA_SSIZE_T_DEFINED +# if defined(_WIN32) && !defined(__CYGWIN__) && !defined(__WATCOMC__) +# if defined(_SSIZE_T_DEFINED) || defined(_SSIZE_T_) +typedef ssize_t la_ssize_t; +# elif defined(_WIN64) +typedef __int64 la_ssize_t; +# else +typedef long la_ssize_t; +# endif +# else +# include /* ssize_t */ +typedef ssize_t la_ssize_t; +# endif +#endif + +/* Get a suitable definition for mode_t */ +#if ARCHIVE_VERSION_NUMBER >= 3999000 +/* Switch to plain 'int' for libarchive 4.0. It's less broken than 'mode_t' */ +# define __LA_MODE_T int +#elif defined(_WIN32) && !defined(__CYGWIN__) && !defined(__BORLANDC__) && !defined(__WATCOMC__) +# define __LA_MODE_T unsigned short +#else +# define __LA_MODE_T mode_t +#endif + +/* Large file support for Android */ +#ifdef __ANDROID__ +#include "android_lf.h" +#endif + +/* + * On Windows, define LIBARCHIVE_STATIC if you're building or using a + * .lib. The default here assumes you're building a DLL. Only + * libarchive source should ever define __LIBARCHIVE_BUILD. + */ +#if ((defined __WIN32__) || (defined _WIN32) || defined(__CYGWIN__)) && (!defined LIBARCHIVE_STATIC) +# ifdef __LIBARCHIVE_BUILD +# ifdef __GNUC__ +# define __LA_DECL __attribute__((dllexport)) extern +# else +# define __LA_DECL __declspec(dllexport) +# endif +# else +# ifdef __GNUC__ +# define __LA_DECL +# else +# define __LA_DECL __declspec(dllimport) +# endif +# endif +#else +/* Static libraries on all platforms and shared libraries on non-Windows. */ +# define __LA_DECL +#endif + +#if defined(__GNUC__) && __GNUC__ >= 3 && __GNUC_MINOR__ >= 1 +# define __LA_DEPRECATED __attribute__((deprecated)) +#else +# define __LA_DEPRECATED +#endif + +#ifdef __cplusplus +extern "C" { +#endif + +/* + * Description of an archive entry. + * + * You can think of this as "struct stat" with some text fields added in. + * + * TODO: Add "comment", "charset", and possibly other entries that are + * supported by "pax interchange" format. However, GNU, ustar, cpio, + * and other variants don't support these features, so they're not an + * excruciatingly high priority right now. + * + * TODO: "pax interchange" format allows essentially arbitrary + * key/value attributes to be attached to any entry. Supporting + * such extensions may make this library useful for special + * applications (e.g., a package manager could attach special + * package-management attributes to each entry). + */ +struct archive; +struct archive_entry; + +/* + * File-type constants. These are returned from archive_entry_filetype() + * and passed to archive_entry_set_filetype(). + * + * These values match S_XXX defines on every platform I've checked, + * including Windows, AIX, Linux, Solaris, and BSD. They're + * (re)defined here because platforms generally don't define the ones + * they don't support. For example, Windows doesn't define S_IFLNK or + * S_IFBLK. Instead of having a mass of conditional logic and system + * checks to define any S_XXX values that aren't supported locally, + * I've just defined a new set of such constants so that + * libarchive-based applications can manipulate and identify archive + * entries properly even if the hosting platform can't store them on + * disk. + * + * These values are also used directly within some portable formats, + * such as cpio. If you find a platform that varies from these, the + * correct solution is to leave these alone and translate from these + * portable values to platform-native values when entries are read from + * or written to disk. + */ +/* + * In libarchive 4.0, we can drop the casts here. + * They're needed to work around Borland C's broken mode_t. + */ +#define AE_IFMT ((__LA_MODE_T)0170000) +#define AE_IFREG ((__LA_MODE_T)0100000) +#define AE_IFLNK ((__LA_MODE_T)0120000) +#define AE_IFSOCK ((__LA_MODE_T)0140000) +#define AE_IFCHR ((__LA_MODE_T)0020000) +#define AE_IFBLK ((__LA_MODE_T)0060000) +#define AE_IFDIR ((__LA_MODE_T)0040000) +#define AE_IFIFO ((__LA_MODE_T)0010000) + +/* + * Symlink types + */ +#define AE_SYMLINK_TYPE_UNDEFINED 0 +#define AE_SYMLINK_TYPE_FILE 1 +#define AE_SYMLINK_TYPE_DIRECTORY 2 + +/* + * Basic object manipulation + */ + +__LA_DECL struct archive_entry *archive_entry_clear(struct archive_entry *); +/* The 'clone' function does a deep copy; all of the strings are copied too. */ +__LA_DECL struct archive_entry *archive_entry_clone(struct archive_entry *); +__LA_DECL void archive_entry_free(struct archive_entry *); +__LA_DECL struct archive_entry *archive_entry_new(void); + +/* + * This form of archive_entry_new2() will pull character-set + * conversion information from the specified archive handle. The + * older archive_entry_new(void) form is equivalent to calling + * archive_entry_new2(NULL) and will result in the use of an internal + * default character-set conversion. + */ +__LA_DECL struct archive_entry *archive_entry_new2(struct archive *); + +/* + * Retrieve fields from an archive_entry. + * + * There are a number of implicit conversions among these fields. For + * example, if a regular string field is set and you read the _w wide + * character field, the entry will implicitly convert narrow-to-wide + * using the current locale. Similarly, dev values are automatically + * updated when you write devmajor or devminor and vice versa. + * + * In addition, fields can be "set" or "unset." Unset string fields + * return NULL, non-string fields have _is_set() functions to test + * whether they've been set. You can "unset" a string field by + * assigning NULL; non-string fields have _unset() functions to + * unset them. + * + * Note: There is one ambiguity in the above; string fields will + * also return NULL when implicit character set conversions fail. + * This is usually what you want. + */ +__LA_DECL time_t archive_entry_atime(struct archive_entry *); +__LA_DECL long archive_entry_atime_nsec(struct archive_entry *); +__LA_DECL int archive_entry_atime_is_set(struct archive_entry *); +__LA_DECL time_t archive_entry_birthtime(struct archive_entry *); +__LA_DECL long archive_entry_birthtime_nsec(struct archive_entry *); +__LA_DECL int archive_entry_birthtime_is_set(struct archive_entry *); +__LA_DECL time_t archive_entry_ctime(struct archive_entry *); +__LA_DECL long archive_entry_ctime_nsec(struct archive_entry *); +__LA_DECL int archive_entry_ctime_is_set(struct archive_entry *); +__LA_DECL dev_t archive_entry_dev(struct archive_entry *); +__LA_DECL int archive_entry_dev_is_set(struct archive_entry *); +__LA_DECL dev_t archive_entry_devmajor(struct archive_entry *); +__LA_DECL dev_t archive_entry_devminor(struct archive_entry *); +__LA_DECL __LA_MODE_T archive_entry_filetype(struct archive_entry *); +__LA_DECL void archive_entry_fflags(struct archive_entry *, + unsigned long * /* set */, + unsigned long * /* clear */); +__LA_DECL const char *archive_entry_fflags_text(struct archive_entry *); +__LA_DECL la_int64_t archive_entry_gid(struct archive_entry *); +__LA_DECL const char *archive_entry_gname(struct archive_entry *); +__LA_DECL const char *archive_entry_gname_utf8(struct archive_entry *); +__LA_DECL const wchar_t *archive_entry_gname_w(struct archive_entry *); +__LA_DECL const char *archive_entry_hardlink(struct archive_entry *); +__LA_DECL const char *archive_entry_hardlink_utf8(struct archive_entry *); +__LA_DECL const wchar_t *archive_entry_hardlink_w(struct archive_entry *); +__LA_DECL la_int64_t archive_entry_ino(struct archive_entry *); +__LA_DECL la_int64_t archive_entry_ino64(struct archive_entry *); +__LA_DECL int archive_entry_ino_is_set(struct archive_entry *); +__LA_DECL __LA_MODE_T archive_entry_mode(struct archive_entry *); +__LA_DECL time_t archive_entry_mtime(struct archive_entry *); +__LA_DECL long archive_entry_mtime_nsec(struct archive_entry *); +__LA_DECL int archive_entry_mtime_is_set(struct archive_entry *); +__LA_DECL unsigned int archive_entry_nlink(struct archive_entry *); +__LA_DECL const char *archive_entry_pathname(struct archive_entry *); +__LA_DECL const char *archive_entry_pathname_utf8(struct archive_entry *); +__LA_DECL const wchar_t *archive_entry_pathname_w(struct archive_entry *); +__LA_DECL __LA_MODE_T archive_entry_perm(struct archive_entry *); +__LA_DECL dev_t archive_entry_rdev(struct archive_entry *); +__LA_DECL dev_t archive_entry_rdevmajor(struct archive_entry *); +__LA_DECL dev_t archive_entry_rdevminor(struct archive_entry *); +__LA_DECL const char *archive_entry_sourcepath(struct archive_entry *); +__LA_DECL const wchar_t *archive_entry_sourcepath_w(struct archive_entry *); +__LA_DECL la_int64_t archive_entry_size(struct archive_entry *); +__LA_DECL int archive_entry_size_is_set(struct archive_entry *); +__LA_DECL const char *archive_entry_strmode(struct archive_entry *); +__LA_DECL const char *archive_entry_symlink(struct archive_entry *); +__LA_DECL const char *archive_entry_symlink_utf8(struct archive_entry *); +__LA_DECL int archive_entry_symlink_type(struct archive_entry *); +__LA_DECL const wchar_t *archive_entry_symlink_w(struct archive_entry *); +__LA_DECL la_int64_t archive_entry_uid(struct archive_entry *); +__LA_DECL const char *archive_entry_uname(struct archive_entry *); +__LA_DECL const char *archive_entry_uname_utf8(struct archive_entry *); +__LA_DECL const wchar_t *archive_entry_uname_w(struct archive_entry *); +__LA_DECL int archive_entry_is_data_encrypted(struct archive_entry *); +__LA_DECL int archive_entry_is_metadata_encrypted(struct archive_entry *); +__LA_DECL int archive_entry_is_encrypted(struct archive_entry *); + +/* + * Set fields in an archive_entry. + * + * Note: Before libarchive 2.4, there were 'set' and 'copy' versions + * of the string setters. 'copy' copied the actual string, 'set' just + * stored the pointer. In libarchive 2.4 and later, strings are + * always copied. + */ + +__LA_DECL void archive_entry_set_atime(struct archive_entry *, time_t, long); +__LA_DECL void archive_entry_unset_atime(struct archive_entry *); +#if defined(_WIN32) && !defined(__CYGWIN__) +__LA_DECL void archive_entry_copy_bhfi(struct archive_entry *, BY_HANDLE_FILE_INFORMATION *); +#endif +__LA_DECL void archive_entry_set_birthtime(struct archive_entry *, time_t, long); +__LA_DECL void archive_entry_unset_birthtime(struct archive_entry *); +__LA_DECL void archive_entry_set_ctime(struct archive_entry *, time_t, long); +__LA_DECL void archive_entry_unset_ctime(struct archive_entry *); +__LA_DECL void archive_entry_set_dev(struct archive_entry *, dev_t); +__LA_DECL void archive_entry_set_devmajor(struct archive_entry *, dev_t); +__LA_DECL void archive_entry_set_devminor(struct archive_entry *, dev_t); +__LA_DECL void archive_entry_set_filetype(struct archive_entry *, unsigned int); +__LA_DECL void archive_entry_set_fflags(struct archive_entry *, + unsigned long /* set */, unsigned long /* clear */); +/* Returns pointer to start of first invalid token, or NULL if none. */ +/* Note that all recognized tokens are processed, regardless. */ +__LA_DECL const char *archive_entry_copy_fflags_text(struct archive_entry *, + const char *); +__LA_DECL const wchar_t *archive_entry_copy_fflags_text_w(struct archive_entry *, + const wchar_t *); +__LA_DECL void archive_entry_set_gid(struct archive_entry *, la_int64_t); +__LA_DECL void archive_entry_set_gname(struct archive_entry *, const char *); +__LA_DECL void archive_entry_set_gname_utf8(struct archive_entry *, const char *); +__LA_DECL void archive_entry_copy_gname(struct archive_entry *, const char *); +__LA_DECL void archive_entry_copy_gname_w(struct archive_entry *, const wchar_t *); +__LA_DECL int archive_entry_update_gname_utf8(struct archive_entry *, const char *); +__LA_DECL void archive_entry_set_hardlink(struct archive_entry *, const char *); +__LA_DECL void archive_entry_set_hardlink_utf8(struct archive_entry *, const char *); +__LA_DECL void archive_entry_copy_hardlink(struct archive_entry *, const char *); +__LA_DECL void archive_entry_copy_hardlink_w(struct archive_entry *, const wchar_t *); +__LA_DECL int archive_entry_update_hardlink_utf8(struct archive_entry *, const char *); +__LA_DECL void archive_entry_set_ino(struct archive_entry *, la_int64_t); +__LA_DECL void archive_entry_set_ino64(struct archive_entry *, la_int64_t); +__LA_DECL void archive_entry_set_link(struct archive_entry *, const char *); +__LA_DECL void archive_entry_set_link_utf8(struct archive_entry *, const char *); +__LA_DECL void archive_entry_copy_link(struct archive_entry *, const char *); +__LA_DECL void archive_entry_copy_link_w(struct archive_entry *, const wchar_t *); +__LA_DECL int archive_entry_update_link_utf8(struct archive_entry *, const char *); +__LA_DECL void archive_entry_set_mode(struct archive_entry *, __LA_MODE_T); +__LA_DECL void archive_entry_set_mtime(struct archive_entry *, time_t, long); +__LA_DECL void archive_entry_unset_mtime(struct archive_entry *); +__LA_DECL void archive_entry_set_nlink(struct archive_entry *, unsigned int); +__LA_DECL void archive_entry_set_pathname(struct archive_entry *, const char *); +__LA_DECL void archive_entry_set_pathname_utf8(struct archive_entry *, const char *); +__LA_DECL void archive_entry_copy_pathname(struct archive_entry *, const char *); +__LA_DECL void archive_entry_copy_pathname_w(struct archive_entry *, const wchar_t *); +__LA_DECL int archive_entry_update_pathname_utf8(struct archive_entry *, const char *); +__LA_DECL void archive_entry_set_perm(struct archive_entry *, __LA_MODE_T); +__LA_DECL void archive_entry_set_rdev(struct archive_entry *, dev_t); +__LA_DECL void archive_entry_set_rdevmajor(struct archive_entry *, dev_t); +__LA_DECL void archive_entry_set_rdevminor(struct archive_entry *, dev_t); +__LA_DECL void archive_entry_set_size(struct archive_entry *, la_int64_t); +__LA_DECL void archive_entry_unset_size(struct archive_entry *); +__LA_DECL void archive_entry_copy_sourcepath(struct archive_entry *, const char *); +__LA_DECL void archive_entry_copy_sourcepath_w(struct archive_entry *, const wchar_t *); +__LA_DECL void archive_entry_set_symlink(struct archive_entry *, const char *); +__LA_DECL void archive_entry_set_symlink_type(struct archive_entry *, int); +__LA_DECL void archive_entry_set_symlink_utf8(struct archive_entry *, const char *); +__LA_DECL void archive_entry_copy_symlink(struct archive_entry *, const char *); +__LA_DECL void archive_entry_copy_symlink_w(struct archive_entry *, const wchar_t *); +__LA_DECL int archive_entry_update_symlink_utf8(struct archive_entry *, const char *); +__LA_DECL void archive_entry_set_uid(struct archive_entry *, la_int64_t); +__LA_DECL void archive_entry_set_uname(struct archive_entry *, const char *); +__LA_DECL void archive_entry_set_uname_utf8(struct archive_entry *, const char *); +__LA_DECL void archive_entry_copy_uname(struct archive_entry *, const char *); +__LA_DECL void archive_entry_copy_uname_w(struct archive_entry *, const wchar_t *); +__LA_DECL int archive_entry_update_uname_utf8(struct archive_entry *, const char *); +__LA_DECL void archive_entry_set_is_data_encrypted(struct archive_entry *, char is_encrypted); +__LA_DECL void archive_entry_set_is_metadata_encrypted(struct archive_entry *, char is_encrypted); +/* + * Routines to bulk copy fields to/from a platform-native "struct + * stat." Libarchive used to just store a struct stat inside of each + * archive_entry object, but this created issues when trying to + * manipulate archives on systems different than the ones they were + * created on. + * + * TODO: On Linux and other LFS systems, provide both stat32 and + * stat64 versions of these functions and all of the macro glue so + * that archive_entry_stat is magically defined to + * archive_entry_stat32 or archive_entry_stat64 as appropriate. + */ +__LA_DECL const struct stat *archive_entry_stat(struct archive_entry *); +__LA_DECL void archive_entry_copy_stat(struct archive_entry *, const struct stat *); + +/* + * Storage for Mac OS-specific AppleDouble metadata information. + * Apple-format tar files store a separate binary blob containing + * encoded metadata with ACL, extended attributes, etc. + * This provides a place to store that blob. + */ + +__LA_DECL const void * archive_entry_mac_metadata(struct archive_entry *, size_t *); +__LA_DECL void archive_entry_copy_mac_metadata(struct archive_entry *, const void *, size_t); + +/* + * ACL routines. This used to simply store and return text-format ACL + * strings, but that proved insufficient for a number of reasons: + * = clients need control over uname/uid and gname/gid mappings + * = there are many different ACL text formats + * = would like to be able to read/convert archives containing ACLs + * on platforms that lack ACL libraries + * + * This last point, in particular, forces me to implement a reasonably + * complete set of ACL support routines. + */ + +/* + * Permission bits. + */ +#define ARCHIVE_ENTRY_ACL_EXECUTE 0x00000001 +#define ARCHIVE_ENTRY_ACL_WRITE 0x00000002 +#define ARCHIVE_ENTRY_ACL_READ 0x00000004 +#define ARCHIVE_ENTRY_ACL_READ_DATA 0x00000008 +#define ARCHIVE_ENTRY_ACL_LIST_DIRECTORY 0x00000008 +#define ARCHIVE_ENTRY_ACL_WRITE_DATA 0x00000010 +#define ARCHIVE_ENTRY_ACL_ADD_FILE 0x00000010 +#define ARCHIVE_ENTRY_ACL_APPEND_DATA 0x00000020 +#define ARCHIVE_ENTRY_ACL_ADD_SUBDIRECTORY 0x00000020 +#define ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS 0x00000040 +#define ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS 0x00000080 +#define ARCHIVE_ENTRY_ACL_DELETE_CHILD 0x00000100 +#define ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES 0x00000200 +#define ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES 0x00000400 +#define ARCHIVE_ENTRY_ACL_DELETE 0x00000800 +#define ARCHIVE_ENTRY_ACL_READ_ACL 0x00001000 +#define ARCHIVE_ENTRY_ACL_WRITE_ACL 0x00002000 +#define ARCHIVE_ENTRY_ACL_WRITE_OWNER 0x00004000 +#define ARCHIVE_ENTRY_ACL_SYNCHRONIZE 0x00008000 + +#define ARCHIVE_ENTRY_ACL_PERMS_POSIX1E \ + (ARCHIVE_ENTRY_ACL_EXECUTE \ + | ARCHIVE_ENTRY_ACL_WRITE \ + | ARCHIVE_ENTRY_ACL_READ) + +#define ARCHIVE_ENTRY_ACL_PERMS_NFS4 \ + (ARCHIVE_ENTRY_ACL_EXECUTE \ + | ARCHIVE_ENTRY_ACL_READ_DATA \ + | ARCHIVE_ENTRY_ACL_LIST_DIRECTORY \ + | ARCHIVE_ENTRY_ACL_WRITE_DATA \ + | ARCHIVE_ENTRY_ACL_ADD_FILE \ + | ARCHIVE_ENTRY_ACL_APPEND_DATA \ + | ARCHIVE_ENTRY_ACL_ADD_SUBDIRECTORY \ + | ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS \ + | ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS \ + | ARCHIVE_ENTRY_ACL_DELETE_CHILD \ + | ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES \ + | ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES \ + | ARCHIVE_ENTRY_ACL_DELETE \ + | ARCHIVE_ENTRY_ACL_READ_ACL \ + | ARCHIVE_ENTRY_ACL_WRITE_ACL \ + | ARCHIVE_ENTRY_ACL_WRITE_OWNER \ + | ARCHIVE_ENTRY_ACL_SYNCHRONIZE) + +/* + * Inheritance values (NFS4 ACLs only); included in permset. + */ +#define ARCHIVE_ENTRY_ACL_ENTRY_INHERITED 0x01000000 +#define ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT 0x02000000 +#define ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT 0x04000000 +#define ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT 0x08000000 +#define ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY 0x10000000 +#define ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS 0x20000000 +#define ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS 0x40000000 + +#define ARCHIVE_ENTRY_ACL_INHERITANCE_NFS4 \ + (ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT \ + | ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT \ + | ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT \ + | ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY \ + | ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS \ + | ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS \ + | ARCHIVE_ENTRY_ACL_ENTRY_INHERITED) + +/* We need to be able to specify combinations of these. */ +#define ARCHIVE_ENTRY_ACL_TYPE_ACCESS 0x00000100 /* POSIX.1e only */ +#define ARCHIVE_ENTRY_ACL_TYPE_DEFAULT 0x00000200 /* POSIX.1e only */ +#define ARCHIVE_ENTRY_ACL_TYPE_ALLOW 0x00000400 /* NFS4 only */ +#define ARCHIVE_ENTRY_ACL_TYPE_DENY 0x00000800 /* NFS4 only */ +#define ARCHIVE_ENTRY_ACL_TYPE_AUDIT 0x00001000 /* NFS4 only */ +#define ARCHIVE_ENTRY_ACL_TYPE_ALARM 0x00002000 /* NFS4 only */ +#define ARCHIVE_ENTRY_ACL_TYPE_POSIX1E (ARCHIVE_ENTRY_ACL_TYPE_ACCESS \ + | ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) +#define ARCHIVE_ENTRY_ACL_TYPE_NFS4 (ARCHIVE_ENTRY_ACL_TYPE_ALLOW \ + | ARCHIVE_ENTRY_ACL_TYPE_DENY \ + | ARCHIVE_ENTRY_ACL_TYPE_AUDIT \ + | ARCHIVE_ENTRY_ACL_TYPE_ALARM) + +/* Tag values mimic POSIX.1e */ +#define ARCHIVE_ENTRY_ACL_USER 10001 /* Specified user. */ +#define ARCHIVE_ENTRY_ACL_USER_OBJ 10002 /* User who owns the file. */ +#define ARCHIVE_ENTRY_ACL_GROUP 10003 /* Specified group. */ +#define ARCHIVE_ENTRY_ACL_GROUP_OBJ 10004 /* Group who owns the file. */ +#define ARCHIVE_ENTRY_ACL_MASK 10005 /* Modify group access (POSIX.1e only) */ +#define ARCHIVE_ENTRY_ACL_OTHER 10006 /* Public (POSIX.1e only) */ +#define ARCHIVE_ENTRY_ACL_EVERYONE 10107 /* Everyone (NFS4 only) */ + +/* + * Set the ACL by clearing it and adding entries one at a time. + * Unlike the POSIX.1e ACL routines, you must specify the type + * (access/default) for each entry. Internally, the ACL data is just + * a soup of entries. API calls here allow you to retrieve just the + * entries of interest. This design (which goes against the spirit of + * POSIX.1e) is useful for handling archive formats that combine + * default and access information in a single ACL list. + */ +__LA_DECL void archive_entry_acl_clear(struct archive_entry *); +__LA_DECL int archive_entry_acl_add_entry(struct archive_entry *, + int /* type */, int /* permset */, int /* tag */, + int /* qual */, const char * /* name */); +__LA_DECL int archive_entry_acl_add_entry_w(struct archive_entry *, + int /* type */, int /* permset */, int /* tag */, + int /* qual */, const wchar_t * /* name */); + +/* + * To retrieve the ACL, first "reset", then repeatedly ask for the + * "next" entry. The want_type parameter allows you to request only + * certain types of entries. + */ +__LA_DECL int archive_entry_acl_reset(struct archive_entry *, int /* want_type */); +__LA_DECL int archive_entry_acl_next(struct archive_entry *, int /* want_type */, + int * /* type */, int * /* permset */, int * /* tag */, + int * /* qual */, const char ** /* name */); + +/* + * Construct a text-format ACL. The flags argument is a bitmask that + * can include any of the following: + * + * Flags only for archive entries with POSIX.1e ACL: + * ARCHIVE_ENTRY_ACL_TYPE_ACCESS - Include POSIX.1e "access" entries. + * ARCHIVE_ENTRY_ACL_TYPE_DEFAULT - Include POSIX.1e "default" entries. + * ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT - Include "default:" before each + * default ACL entry. + * ARCHIVE_ENTRY_ACL_STYLE_SOLARIS - Output only one colon after "other" and + * "mask" entries. + * + * Flags only for archive entries with NFSv4 ACL: + * ARCHIVE_ENTRY_ACL_STYLE_COMPACT - Do not output the minus character for + * unset permissions and flags in NFSv4 ACL permission and flag fields + * + * Flags for for archive entries with POSIX.1e ACL or NFSv4 ACL: + * ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID - Include extra numeric ID field in + * each ACL entry. + * ARCHIVE_ENTRY_ACL_STYLE_SEPARATOR_COMMA - Separate entries with comma + * instead of newline. + */ +#define ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID 0x00000001 +#define ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT 0x00000002 +#define ARCHIVE_ENTRY_ACL_STYLE_SOLARIS 0x00000004 +#define ARCHIVE_ENTRY_ACL_STYLE_SEPARATOR_COMMA 0x00000008 +#define ARCHIVE_ENTRY_ACL_STYLE_COMPACT 0x00000010 + +__LA_DECL wchar_t *archive_entry_acl_to_text_w(struct archive_entry *, + la_ssize_t * /* len */, int /* flags */); +__LA_DECL char *archive_entry_acl_to_text(struct archive_entry *, + la_ssize_t * /* len */, int /* flags */); +__LA_DECL int archive_entry_acl_from_text_w(struct archive_entry *, + const wchar_t * /* wtext */, int /* type */); +__LA_DECL int archive_entry_acl_from_text(struct archive_entry *, + const char * /* text */, int /* type */); + +/* Deprecated constants */ +#define OLD_ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID 1024 +#define OLD_ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT 2048 + +/* Deprecated functions */ +__LA_DECL const wchar_t *archive_entry_acl_text_w(struct archive_entry *, + int /* flags */) __LA_DEPRECATED; +__LA_DECL const char *archive_entry_acl_text(struct archive_entry *, + int /* flags */) __LA_DEPRECATED; + +/* Return bitmask of ACL types in an archive entry */ +__LA_DECL int archive_entry_acl_types(struct archive_entry *); + +/* Return a count of entries matching 'want_type' */ +__LA_DECL int archive_entry_acl_count(struct archive_entry *, int /* want_type */); + +/* Return an opaque ACL object. */ +/* There's not yet anything clients can actually do with this... */ +struct archive_acl; +__LA_DECL struct archive_acl *archive_entry_acl(struct archive_entry *); + +/* + * extended attributes + */ + +__LA_DECL void archive_entry_xattr_clear(struct archive_entry *); +__LA_DECL void archive_entry_xattr_add_entry(struct archive_entry *, + const char * /* name */, const void * /* value */, + size_t /* size */); + +/* + * To retrieve the xattr list, first "reset", then repeatedly ask for the + * "next" entry. + */ + +__LA_DECL int archive_entry_xattr_count(struct archive_entry *); +__LA_DECL int archive_entry_xattr_reset(struct archive_entry *); +__LA_DECL int archive_entry_xattr_next(struct archive_entry *, + const char ** /* name */, const void ** /* value */, size_t *); + +/* + * sparse + */ + +__LA_DECL void archive_entry_sparse_clear(struct archive_entry *); +__LA_DECL void archive_entry_sparse_add_entry(struct archive_entry *, + la_int64_t /* offset */, la_int64_t /* length */); + +/* + * To retrieve the xattr list, first "reset", then repeatedly ask for the + * "next" entry. + */ + +__LA_DECL int archive_entry_sparse_count(struct archive_entry *); +__LA_DECL int archive_entry_sparse_reset(struct archive_entry *); +__LA_DECL int archive_entry_sparse_next(struct archive_entry *, + la_int64_t * /* offset */, la_int64_t * /* length */); + +/* + * Utility to match up hardlinks. + * + * The 'struct archive_entry_linkresolver' is a cache of archive entries + * for files with multiple links. Here's how to use it: + * 1. Create a lookup object with archive_entry_linkresolver_new() + * 2. Tell it the archive format you're using. + * 3. Hand each archive_entry to archive_entry_linkify(). + * That function will return 0, 1, or 2 entries that should + * be written. + * 4. Call archive_entry_linkify(resolver, NULL) until + * no more entries are returned. + * 5. Call archive_entry_linkresolver_free(resolver) to free resources. + * + * The entries returned have their hardlink and size fields updated + * appropriately. If an entry is passed in that does not refer to + * a file with multiple links, it is returned unchanged. The intention + * is that you should be able to simply filter all entries through + * this machine. + * + * To make things more efficient, be sure that each entry has a valid + * nlinks value. The hardlink cache uses this to track when all links + * have been found. If the nlinks value is zero, it will keep every + * name in the cache indefinitely, which can use a lot of memory. + * + * Note that archive_entry_size() is reset to zero if the file + * body should not be written to the archive. Pay attention! + */ +struct archive_entry_linkresolver; + +/* + * There are three different strategies for marking hardlinks. + * The descriptions below name them after the best-known + * formats that rely on each strategy: + * + * "Old cpio" is the simplest, it always returns any entry unmodified. + * As far as I know, only cpio formats use this. Old cpio archives + * store every link with the full body; the onus is on the dearchiver + * to detect and properly link the files as they are restored. + * "tar" is also pretty simple; it caches a copy the first time it sees + * any link. Subsequent appearances are modified to be hardlink + * references to the first one without any body. Used by all tar + * formats, although the newest tar formats permit the "old cpio" strategy + * as well. This strategy is very simple for the dearchiver, + * and reasonably straightforward for the archiver. + * "new cpio" is trickier. It stores the body only with the last + * occurrence. The complication is that we might not + * see every link to a particular file in a single session, so + * there's no easy way to know when we've seen the last occurrence. + * The solution here is to queue one link until we see the next. + * At the end of the session, you can enumerate any remaining + * entries by calling archive_entry_linkify(NULL) and store those + * bodies. If you have a file with three links l1, l2, and l3, + * you'll get the following behavior if you see all three links: + * linkify(l1) => NULL (the resolver stores l1 internally) + * linkify(l2) => l1 (resolver stores l2, you write l1) + * linkify(l3) => l2, l3 (all links seen, you can write both). + * If you only see l1 and l2, you'll get this behavior: + * linkify(l1) => NULL + * linkify(l2) => l1 + * linkify(NULL) => l2 (at end, you retrieve remaining links) + * As the name suggests, this strategy is used by newer cpio variants. + * It's noticeably more complex for the archiver, slightly more complex + * for the dearchiver than the tar strategy, but makes it straightforward + * to restore a file using any link by simply continuing to scan until + * you see a link that is stored with a body. In contrast, the tar + * strategy requires you to rescan the archive from the beginning to + * correctly extract an arbitrary link. + */ + +__LA_DECL struct archive_entry_linkresolver *archive_entry_linkresolver_new(void); +__LA_DECL void archive_entry_linkresolver_set_strategy( + struct archive_entry_linkresolver *, int /* format_code */); +__LA_DECL void archive_entry_linkresolver_free(struct archive_entry_linkresolver *); +__LA_DECL void archive_entry_linkify(struct archive_entry_linkresolver *, + struct archive_entry **, struct archive_entry **); +__LA_DECL struct archive_entry *archive_entry_partial_links( + struct archive_entry_linkresolver *res, unsigned int *links); +#ifdef __cplusplus +} +#endif + +/* This is meaningless outside of this header. */ +#undef __LA_DECL + +#endif /* !ARCHIVE_ENTRY_H_INCLUDED */ diff --git a/c/archive_entry_copy_bhfi.c b/c/archive_entry_copy_bhfi.c new file mode 100644 --- /dev/null +++ b/c/archive_entry_copy_bhfi.c @@ -0,0 +1,75 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#include "archive_private.h" +#include "archive_entry.h" + +#if defined(_WIN32) && !defined(__CYGWIN__) + +#define EPOC_TIME ARCHIVE_LITERAL_ULL(116444736000000000) + +__inline static void +fileTimeToUtc(const FILETIME *filetime, time_t *t, long *ns) +{ + ULARGE_INTEGER utc; + + utc.HighPart = filetime->dwHighDateTime; + utc.LowPart = filetime->dwLowDateTime; + if (utc.QuadPart >= EPOC_TIME) { + utc.QuadPart -= EPOC_TIME; + *t = (time_t)(utc.QuadPart / 10000000); /* milli seconds base */ + *ns = (long)(utc.QuadPart % 10000000) * 100;/* nano seconds base */ + } else { + *t = 0; + *ns = 0; + } +} + +void +archive_entry_copy_bhfi(struct archive_entry *entry, + BY_HANDLE_FILE_INFORMATION *bhfi) +{ + time_t secs; + long nsecs; + + fileTimeToUtc(&bhfi->ftLastAccessTime, &secs, &nsecs); + archive_entry_set_atime(entry, secs, nsecs); + fileTimeToUtc(&bhfi->ftLastWriteTime, &secs, &nsecs); + archive_entry_set_mtime(entry, secs, nsecs); + fileTimeToUtc(&bhfi->ftCreationTime, &secs, &nsecs); + archive_entry_set_birthtime(entry, secs, nsecs); + archive_entry_set_ctime(entry, secs, nsecs); + archive_entry_set_dev(entry, bhfi->dwVolumeSerialNumber); + archive_entry_set_ino64(entry, (((int64_t)bhfi->nFileIndexHigh) << 32) + + bhfi->nFileIndexLow); + archive_entry_set_nlink(entry, bhfi->nNumberOfLinks); + archive_entry_set_size(entry, (((int64_t)bhfi->nFileSizeHigh) << 32) + + bhfi->nFileSizeLow); + /* archive_entry_set_mode(entry, st->st_mode); */ +} +#endif diff --git a/c/archive_entry_copy_stat.c b/c/archive_entry_copy_stat.c new file mode 100644 --- /dev/null +++ b/c/archive_entry_copy_stat.c @@ -0,0 +1,83 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_entry_copy_stat.c 189466 2009-03-07 00:52:02Z kientzle $"); + +#ifdef HAVE_SYS_STAT_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" + +void +archive_entry_copy_stat(struct archive_entry *entry, const struct stat *st) +{ +#if HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC + archive_entry_set_atime(entry, st->st_atime, st->st_atimespec.tv_nsec); + archive_entry_set_ctime(entry, st->st_ctime, st->st_ctimespec.tv_nsec); + archive_entry_set_mtime(entry, st->st_mtime, st->st_mtimespec.tv_nsec); +#elif HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC + archive_entry_set_atime(entry, st->st_atime, st->st_atim.tv_nsec); + archive_entry_set_ctime(entry, st->st_ctime, st->st_ctim.tv_nsec); + archive_entry_set_mtime(entry, st->st_mtime, st->st_mtim.tv_nsec); +#elif HAVE_STRUCT_STAT_ST_MTIME_NSEC + archive_entry_set_atime(entry, st->st_atime, st->st_atime_nsec); + archive_entry_set_ctime(entry, st->st_ctime, st->st_ctime_nsec); + archive_entry_set_mtime(entry, st->st_mtime, st->st_mtime_nsec); +#elif HAVE_STRUCT_STAT_ST_MTIME_N + archive_entry_set_atime(entry, st->st_atime, st->st_atime_n); + archive_entry_set_ctime(entry, st->st_ctime, st->st_ctime_n); + archive_entry_set_mtime(entry, st->st_mtime, st->st_mtime_n); +#elif HAVE_STRUCT_STAT_ST_UMTIME + archive_entry_set_atime(entry, st->st_atime, st->st_uatime * 1000); + archive_entry_set_ctime(entry, st->st_ctime, st->st_uctime * 1000); + archive_entry_set_mtime(entry, st->st_mtime, st->st_umtime * 1000); +#elif HAVE_STRUCT_STAT_ST_MTIME_USEC + archive_entry_set_atime(entry, st->st_atime, st->st_atime_usec * 1000); + archive_entry_set_ctime(entry, st->st_ctime, st->st_ctime_usec * 1000); + archive_entry_set_mtime(entry, st->st_mtime, st->st_mtime_usec * 1000); +#else + archive_entry_set_atime(entry, st->st_atime, 0); + archive_entry_set_ctime(entry, st->st_ctime, 0); + archive_entry_set_mtime(entry, st->st_mtime, 0); +#endif +#if HAVE_STRUCT_STAT_ST_BIRTHTIMESPEC_TV_NSEC + archive_entry_set_birthtime(entry, st->st_birthtime, st->st_birthtimespec.tv_nsec); +#elif HAVE_STRUCT_STAT_ST_BIRTHTIME + archive_entry_set_birthtime(entry, st->st_birthtime, 0); +#else + archive_entry_unset_birthtime(entry); +#endif + archive_entry_set_dev(entry, st->st_dev); + archive_entry_set_gid(entry, st->st_gid); + archive_entry_set_uid(entry, st->st_uid); + archive_entry_set_ino(entry, st->st_ino); + archive_entry_set_nlink(entry, st->st_nlink); + archive_entry_set_rdev(entry, st->st_rdev); + archive_entry_set_size(entry, st->st_size); + archive_entry_set_mode(entry, st->st_mode); +} diff --git a/c/archive_entry_link_resolver.c b/c/archive_entry_link_resolver.c new file mode 100644 --- /dev/null +++ b/c/archive_entry_link_resolver.c @@ -0,0 +1,447 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_entry_link_resolver.c 201100 2009-12-28 03:05:31Z kientzle $"); + +#ifdef HAVE_SYS_STAT_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" + +/* + * This is mostly a pretty straightforward hash table implementation. + * The only interesting bit is the different strategies used to + * match up links. These strategies match those used by various + * archiving formats: + * tar - content stored with first link, remainder refer back to it. + * This requires us to match each subsequent link up with the + * first appearance. + * cpio - Old cpio just stored body with each link, match-ups were + * implicit. This is trivial. + * new cpio - New cpio only stores body with last link, match-ups + * are implicit. This is actually quite tricky; see the notes + * below. + */ + +/* Users pass us a format code, we translate that into a strategy here. */ +#define ARCHIVE_ENTRY_LINKIFY_LIKE_TAR 0 +#define ARCHIVE_ENTRY_LINKIFY_LIKE_MTREE 1 +#define ARCHIVE_ENTRY_LINKIFY_LIKE_OLD_CPIO 2 +#define ARCHIVE_ENTRY_LINKIFY_LIKE_NEW_CPIO 3 + +/* Initial size of link cache. */ +#define links_cache_initial_size 1024 + +struct links_entry { + struct links_entry *next; + struct links_entry *previous; + struct archive_entry *canonical; + struct archive_entry *entry; + size_t hash; + unsigned int links; /* # links not yet seen */ +}; + +struct archive_entry_linkresolver { + struct links_entry **buckets; + struct links_entry *spare; + unsigned long number_entries; + size_t number_buckets; + int strategy; +}; + +#define NEXT_ENTRY_DEFERRED 1 +#define NEXT_ENTRY_PARTIAL 2 +#define NEXT_ENTRY_ALL (NEXT_ENTRY_DEFERRED | NEXT_ENTRY_PARTIAL) + +static struct links_entry *find_entry(struct archive_entry_linkresolver *, + struct archive_entry *); +static void grow_hash(struct archive_entry_linkresolver *); +static struct links_entry *insert_entry(struct archive_entry_linkresolver *, + struct archive_entry *); +static struct links_entry *next_entry(struct archive_entry_linkresolver *, + int); + +struct archive_entry_linkresolver * +archive_entry_linkresolver_new(void) +{ + struct archive_entry_linkresolver *res; + + /* Check for positive power-of-two */ + if (links_cache_initial_size == 0 || + (links_cache_initial_size & (links_cache_initial_size - 1)) != 0) + return (NULL); + + res = calloc(1, sizeof(struct archive_entry_linkresolver)); + if (res == NULL) + return (NULL); + res->number_buckets = links_cache_initial_size; + res->buckets = calloc(res->number_buckets, sizeof(res->buckets[0])); + if (res->buckets == NULL) { + free(res); + return (NULL); + } + return (res); +} + +void +archive_entry_linkresolver_set_strategy(struct archive_entry_linkresolver *res, + int fmt) +{ + int fmtbase = fmt & ARCHIVE_FORMAT_BASE_MASK; + + switch (fmtbase) { + case ARCHIVE_FORMAT_7ZIP: + case ARCHIVE_FORMAT_AR: + case ARCHIVE_FORMAT_ZIP: + res->strategy = ARCHIVE_ENTRY_LINKIFY_LIKE_OLD_CPIO; + break; + case ARCHIVE_FORMAT_CPIO: + switch (fmt) { + case ARCHIVE_FORMAT_CPIO_SVR4_NOCRC: + case ARCHIVE_FORMAT_CPIO_SVR4_CRC: + res->strategy = ARCHIVE_ENTRY_LINKIFY_LIKE_NEW_CPIO; + break; + default: + res->strategy = ARCHIVE_ENTRY_LINKIFY_LIKE_OLD_CPIO; + break; + } + break; + case ARCHIVE_FORMAT_MTREE: + res->strategy = ARCHIVE_ENTRY_LINKIFY_LIKE_MTREE; + break; + case ARCHIVE_FORMAT_ISO9660: + case ARCHIVE_FORMAT_SHAR: + case ARCHIVE_FORMAT_TAR: + case ARCHIVE_FORMAT_XAR: + res->strategy = ARCHIVE_ENTRY_LINKIFY_LIKE_TAR; + break; + default: + res->strategy = ARCHIVE_ENTRY_LINKIFY_LIKE_OLD_CPIO; + break; + } +} + +void +archive_entry_linkresolver_free(struct archive_entry_linkresolver *res) +{ + struct links_entry *le; + + if (res == NULL) + return; + + while ((le = next_entry(res, NEXT_ENTRY_ALL)) != NULL) + archive_entry_free(le->entry); + free(res->buckets); + free(res); +} + +void +archive_entry_linkify(struct archive_entry_linkresolver *res, + struct archive_entry **e, struct archive_entry **f) +{ + struct links_entry *le; + struct archive_entry *t; + + *f = NULL; /* Default: Don't return a second entry. */ + + if (*e == NULL) { + le = next_entry(res, NEXT_ENTRY_DEFERRED); + if (le != NULL) { + *e = le->entry; + le->entry = NULL; + } + return; + } + + /* If it has only one link, then we're done. */ + if (archive_entry_nlink(*e) == 1) + return; + /* Directories, devices never have hardlinks. */ + if (archive_entry_filetype(*e) == AE_IFDIR + || archive_entry_filetype(*e) == AE_IFBLK + || archive_entry_filetype(*e) == AE_IFCHR) + return; + + switch (res->strategy) { + case ARCHIVE_ENTRY_LINKIFY_LIKE_TAR: + le = find_entry(res, *e); + if (le != NULL) { + archive_entry_unset_size(*e); + archive_entry_copy_hardlink(*e, + archive_entry_pathname(le->canonical)); + } else + insert_entry(res, *e); + return; + case ARCHIVE_ENTRY_LINKIFY_LIKE_MTREE: + le = find_entry(res, *e); + if (le != NULL) { + archive_entry_copy_hardlink(*e, + archive_entry_pathname(le->canonical)); + } else + insert_entry(res, *e); + return; + case ARCHIVE_ENTRY_LINKIFY_LIKE_OLD_CPIO: + /* This one is trivial. */ + return; + case ARCHIVE_ENTRY_LINKIFY_LIKE_NEW_CPIO: + le = find_entry(res, *e); + if (le != NULL) { + /* + * Put the new entry in le, return the + * old entry from le. + */ + t = *e; + *e = le->entry; + le->entry = t; + /* Make the old entry into a hardlink. */ + archive_entry_unset_size(*e); + archive_entry_copy_hardlink(*e, + archive_entry_pathname(le->canonical)); + /* If we ran out of links, return the + * final entry as well. */ + if (le->links == 0) { + *f = le->entry; + le->entry = NULL; + } + } else { + /* + * If we haven't seen it, tuck it away + * for future use. + */ + le = insert_entry(res, *e); + if (le == NULL) + /* XXX We should return an error code XXX */ + return; + le->entry = *e; + *e = NULL; + } + return; + default: + break; + } + return; +} + +static struct links_entry * +find_entry(struct archive_entry_linkresolver *res, + struct archive_entry *entry) +{ + struct links_entry *le; + size_t hash, bucket; + dev_t dev; + int64_t ino; + + /* Free a held entry. */ + if (res->spare != NULL) { + archive_entry_free(res->spare->canonical); + archive_entry_free(res->spare->entry); + free(res->spare); + res->spare = NULL; + } + + dev = archive_entry_dev(entry); + ino = archive_entry_ino64(entry); + hash = (size_t)(dev ^ ino); + + /* Try to locate this entry in the links cache. */ + bucket = hash & (res->number_buckets - 1); + for (le = res->buckets[bucket]; le != NULL; le = le->next) { + if (le->hash == hash + && dev == archive_entry_dev(le->canonical) + && ino == archive_entry_ino64(le->canonical)) { + /* + * Decrement link count each time and release + * the entry if it hits zero. This saves + * memory and is necessary for detecting + * missed links. + */ + --le->links; + if (le->links > 0) + return (le); + /* Remove it from this hash bucket. */ + if (le->previous != NULL) + le->previous->next = le->next; + if (le->next != NULL) + le->next->previous = le->previous; + if (res->buckets[bucket] == le) + res->buckets[bucket] = le->next; + res->number_entries--; + /* Defer freeing this entry. */ + res->spare = le; + return (le); + } + } + return (NULL); +} + +static struct links_entry * +next_entry(struct archive_entry_linkresolver *res, int mode) +{ + struct links_entry *le; + size_t bucket; + + /* Free a held entry. */ + if (res->spare != NULL) { + archive_entry_free(res->spare->canonical); + archive_entry_free(res->spare->entry); + free(res->spare); + res->spare = NULL; + } + + /* Look for next non-empty bucket in the links cache. */ + for (bucket = 0; bucket < res->number_buckets; bucket++) { + for (le = res->buckets[bucket]; le != NULL; le = le->next) { + if (le->entry != NULL && + (mode & NEXT_ENTRY_DEFERRED) == 0) + continue; + if (le->entry == NULL && + (mode & NEXT_ENTRY_PARTIAL) == 0) + continue; + /* Remove it from this hash bucket. */ + if (le->next != NULL) + le->next->previous = le->previous; + if (le->previous != NULL) + le->previous->next = le->next; + else + res->buckets[bucket] = le->next; + res->number_entries--; + /* Defer freeing this entry. */ + res->spare = le; + return (le); + } + } + return (NULL); +} + +static struct links_entry * +insert_entry(struct archive_entry_linkresolver *res, + struct archive_entry *entry) +{ + struct links_entry *le; + size_t hash, bucket; + + /* Add this entry to the links cache. */ + le = calloc(1, sizeof(struct links_entry)); + if (le == NULL) + return (NULL); + le->canonical = archive_entry_clone(entry); + + /* If the links cache is getting too full, enlarge the hash table. */ + if (res->number_entries > res->number_buckets * 2) + grow_hash(res); + + hash = (size_t)(archive_entry_dev(entry) ^ archive_entry_ino64(entry)); + bucket = hash & (res->number_buckets - 1); + + /* If we could allocate the entry, record it. */ + if (res->buckets[bucket] != NULL) + res->buckets[bucket]->previous = le; + res->number_entries++; + le->next = res->buckets[bucket]; + le->previous = NULL; + res->buckets[bucket] = le; + le->hash = hash; + le->links = archive_entry_nlink(entry) - 1; + return (le); +} + +static void +grow_hash(struct archive_entry_linkresolver *res) +{ + struct links_entry *le, **new_buckets; + size_t new_size; + size_t i, bucket; + + /* Try to enlarge the bucket list. */ + new_size = res->number_buckets * 2; + if (new_size < res->number_buckets) + return; + new_buckets = calloc(new_size, sizeof(struct links_entry *)); + + if (new_buckets == NULL) + return; + + for (i = 0; i < res->number_buckets; i++) { + while (res->buckets[i] != NULL) { + /* Remove entry from old bucket. */ + le = res->buckets[i]; + res->buckets[i] = le->next; + + /* Add entry to new bucket. */ + bucket = le->hash & (new_size - 1); + + if (new_buckets[bucket] != NULL) + new_buckets[bucket]->previous = le; + le->next = new_buckets[bucket]; + le->previous = NULL; + new_buckets[bucket] = le; + } + } + free(res->buckets); + res->buckets = new_buckets; + res->number_buckets = new_size; +} + +struct archive_entry * +archive_entry_partial_links(struct archive_entry_linkresolver *res, + unsigned int *links) +{ + struct archive_entry *e; + struct links_entry *le; + + /* Free a held entry. */ + if (res->spare != NULL) { + archive_entry_free(res->spare->canonical); + archive_entry_free(res->spare->entry); + free(res->spare); + res->spare = NULL; + } + + le = next_entry(res, NEXT_ENTRY_PARTIAL); + if (le != NULL) { + e = le->canonical; + if (links != NULL) + *links = le->links; + le->canonical = NULL; + } else { + e = NULL; + if (links != NULL) + *links = 0; + } + return (e); +} diff --git a/c/archive_entry_locale.h b/c/archive_entry_locale.h new file mode 100644 --- /dev/null +++ b/c/archive_entry_locale.h @@ -0,0 +1,92 @@ +/*- + * Copyright (c) 2011 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD$ + */ + +#ifndef ARCHIVE_ENTRY_LOCALE_H_INCLUDED +#define ARCHIVE_ENTRY_LOCALE_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif + +struct archive_entry; +struct archive_string_conv; + +/* + * Utility functions to set and get entry attributes by translating + * character-set. These are designed for use in format readers and writers. + * + * The return code and interface of these are quite different from other + * functions for archive_entry defined in archive_entry.h. + * Common return code are: + * Return 0 if the string conversion succeeded. + * Return -1 if the string conversion failed. + */ + +#define archive_entry_gname_l _archive_entry_gname_l +int _archive_entry_gname_l(struct archive_entry *, + const char **, size_t *, struct archive_string_conv *); +#define archive_entry_hardlink_l _archive_entry_hardlink_l +int _archive_entry_hardlink_l(struct archive_entry *, + const char **, size_t *, struct archive_string_conv *); +#define archive_entry_pathname_l _archive_entry_pathname_l +int _archive_entry_pathname_l(struct archive_entry *, + const char **, size_t *, struct archive_string_conv *); +#define archive_entry_symlink_l _archive_entry_symlink_l +int _archive_entry_symlink_l(struct archive_entry *, + const char **, size_t *, struct archive_string_conv *); +#define archive_entry_uname_l _archive_entry_uname_l +int _archive_entry_uname_l(struct archive_entry *, + const char **, size_t *, struct archive_string_conv *); +#define archive_entry_acl_text_l _archive_entry_acl_text_l +int _archive_entry_acl_text_l(struct archive_entry *, int, +const char **, size_t *, struct archive_string_conv *) __LA_DEPRECATED; +#define archive_entry_acl_to_text_l _archive_entry_acl_to_text_l +char *_archive_entry_acl_to_text_l(struct archive_entry *, ssize_t *, int, + struct archive_string_conv *); +#define archive_entry_acl_from_text_l _archive_entry_acl_from_text_l +int _archive_entry_acl_from_text_l(struct archive_entry *, const char* text, + int type, struct archive_string_conv *); +#define archive_entry_copy_gname_l _archive_entry_copy_gname_l +int _archive_entry_copy_gname_l(struct archive_entry *, + const char *, size_t, struct archive_string_conv *); +#define archive_entry_copy_hardlink_l _archive_entry_copy_hardlink_l +int _archive_entry_copy_hardlink_l(struct archive_entry *, + const char *, size_t, struct archive_string_conv *); +#define archive_entry_copy_link_l _archive_entry_copy_link_l +int _archive_entry_copy_link_l(struct archive_entry *, + const char *, size_t, struct archive_string_conv *); +#define archive_entry_copy_pathname_l _archive_entry_copy_pathname_l +int _archive_entry_copy_pathname_l(struct archive_entry *, + const char *, size_t, struct archive_string_conv *); +#define archive_entry_copy_symlink_l _archive_entry_copy_symlink_l +int _archive_entry_copy_symlink_l(struct archive_entry *, + const char *, size_t, struct archive_string_conv *); +#define archive_entry_copy_uname_l _archive_entry_copy_uname_l +int _archive_entry_copy_uname_l(struct archive_entry *, + const char *, size_t, struct archive_string_conv *); + +#endif /* ARCHIVE_ENTRY_LOCALE_H_INCLUDED */ diff --git a/c/archive_entry_private.h b/c/archive_entry_private.h new file mode 100644 --- /dev/null +++ b/c/archive_entry_private.h @@ -0,0 +1,184 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD: head/lib/libarchive/archive_entry_private.h 201096 2009-12-28 02:41:27Z kientzle $ + */ + +#ifndef ARCHIVE_ENTRY_PRIVATE_H_INCLUDED +#define ARCHIVE_ENTRY_PRIVATE_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif + +#include "archive_acl_private.h" +#include "archive_string.h" + +struct ae_xattr { + struct ae_xattr *next; + + char *name; + void *value; + size_t size; +}; + +struct ae_sparse { + struct ae_sparse *next; + + int64_t offset; + int64_t length; +}; + +/* + * Description of an archive entry. + * + * Basically, this is a "struct stat" with a few text fields added in. + * + * TODO: Add "comment", "charset", and possibly other entries + * that are supported by "pax interchange" format. However, GNU, ustar, + * cpio, and other variants don't support these features, so they're not an + * excruciatingly high priority right now. + * + * TODO: "pax interchange" format allows essentially arbitrary + * key/value attributes to be attached to any entry. Supporting + * such extensions may make this library useful for special + * applications (e.g., a package manager could attach special + * package-management attributes to each entry). There are tricky + * API issues involved, so this is not going to happen until + * there's a real demand for it. + * + * TODO: Design a good API for handling sparse files. + */ +struct archive_entry { + struct archive *archive; + + /* + * Note that ae_stat.st_mode & AE_IFMT can be 0! + * + * This occurs when the actual file type of the object is not + * in the archive. For example, 'tar' archives store + * hardlinks without marking the type of the underlying + * object. + */ + + /* + * We have a "struct aest" for holding file metadata rather than just + * a "struct stat" because on some platforms the "struct stat" has + * fields which are too narrow to hold the range of possible values; + * we don't want to lose information if we read an archive and write + * out another (e.g., in "tar -cf new.tar @old.tar"). + * + * The "stat" pointer points to some form of platform-specific struct + * stat; it is declared as a void * rather than a struct stat * as + * some platforms have multiple varieties of stat structures. + */ + void *stat; + int stat_valid; /* Set to 0 whenever a field in aest changes. */ + + struct aest { + int64_t aest_atime; + uint32_t aest_atime_nsec; + int64_t aest_ctime; + uint32_t aest_ctime_nsec; + int64_t aest_mtime; + uint32_t aest_mtime_nsec; + int64_t aest_birthtime; + uint32_t aest_birthtime_nsec; + int64_t aest_gid; + int64_t aest_ino; + uint32_t aest_nlink; + uint64_t aest_size; + int64_t aest_uid; + /* + * Because converting between device codes and + * major/minor values is platform-specific and + * inherently a bit risky, we only do that conversion + * lazily. That way, we will do a better job of + * preserving information in those cases where no + * conversion is actually required. + */ + int aest_dev_is_broken_down; + dev_t aest_dev; + dev_t aest_devmajor; + dev_t aest_devminor; + int aest_rdev_is_broken_down; + dev_t aest_rdev; + dev_t aest_rdevmajor; + dev_t aest_rdevminor; + } ae_stat; + + int ae_set; /* bitmap of fields that are currently set */ +#define AE_SET_HARDLINK 1 +#define AE_SET_SYMLINK 2 +#define AE_SET_ATIME 4 +#define AE_SET_CTIME 8 +#define AE_SET_MTIME 16 +#define AE_SET_BIRTHTIME 32 +#define AE_SET_SIZE 64 +#define AE_SET_INO 128 +#define AE_SET_DEV 256 + + /* + * Use aes here so that we get transparent mbs<->wcs conversions. + */ + struct archive_mstring ae_fflags_text; /* Text fflags per fflagstostr(3) */ + unsigned long ae_fflags_set; /* Bitmap fflags */ + unsigned long ae_fflags_clear; + struct archive_mstring ae_gname; /* Name of owning group */ + struct archive_mstring ae_hardlink; /* Name of target for hardlink */ + struct archive_mstring ae_pathname; /* Name of entry */ + struct archive_mstring ae_symlink; /* symlink contents */ + struct archive_mstring ae_uname; /* Name of owner */ + + /* Not used within libarchive; useful for some clients. */ + struct archive_mstring ae_sourcepath; /* Path this entry is sourced from. */ + +#define AE_ENCRYPTION_NONE 0 +#define AE_ENCRYPTION_DATA 1 +#define AE_ENCRYPTION_METADATA 2 + char encryption; + + void *mac_metadata; + size_t mac_metadata_size; + + /* ACL support. */ + struct archive_acl acl; + + /* extattr support. */ + struct ae_xattr *xattr_head; + struct ae_xattr *xattr_p; + + /* sparse support. */ + struct ae_sparse *sparse_head; + struct ae_sparse *sparse_tail; + struct ae_sparse *sparse_p; + + /* Miscellaneous. */ + char strmode[12]; + + /* Symlink type support */ + int ae_symlink_type; +}; + +#endif /* ARCHIVE_ENTRY_PRIVATE_H_INCLUDED */ diff --git a/c/archive_entry_sparse.c b/c/archive_entry_sparse.c new file mode 100644 --- /dev/null +++ b/c/archive_entry_sparse.c @@ -0,0 +1,156 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * Copyright (c) 2010-2011 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#include "archive.h" +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_entry_private.h" + +/* + * sparse handling + */ + +void +archive_entry_sparse_clear(struct archive_entry *entry) +{ + struct ae_sparse *sp; + + while (entry->sparse_head != NULL) { + sp = entry->sparse_head->next; + free(entry->sparse_head); + entry->sparse_head = sp; + } + entry->sparse_tail = NULL; +} + +void +archive_entry_sparse_add_entry(struct archive_entry *entry, + la_int64_t offset, la_int64_t length) +{ + struct ae_sparse *sp; + + if (offset < 0 || length < 0) + /* Invalid value */ + return; + if (offset > INT64_MAX - length || + offset + length > archive_entry_size(entry)) + /* A value of "length" parameter is too large. */ + return; + if ((sp = entry->sparse_tail) != NULL) { + if (sp->offset + sp->length > offset) + /* Invalid value. */ + return; + if (sp->offset + sp->length == offset) { + if (sp->offset + sp->length + length < 0) + /* A value of "length" parameter is + * too large. */ + return; + /* Expand existing sparse block size. */ + sp->length += length; + return; + } + } + + if ((sp = (struct ae_sparse *)malloc(sizeof(*sp))) == NULL) + /* XXX Error XXX */ + return; + + sp->offset = offset; + sp->length = length; + sp->next = NULL; + + if (entry->sparse_head == NULL) + entry->sparse_head = entry->sparse_tail = sp; + else { + /* Add a new sparse block to the tail of list. */ + if (entry->sparse_tail != NULL) + entry->sparse_tail->next = sp; + entry->sparse_tail = sp; + } +} + + +/* + * returns number of the sparse entries + */ +int +archive_entry_sparse_count(struct archive_entry *entry) +{ + struct ae_sparse *sp; + int count = 0; + + for (sp = entry->sparse_head; sp != NULL; sp = sp->next) + count++; + + /* + * Sanity check if this entry is exactly sparse. + * If amount of sparse blocks is just one and it indicates the whole + * file data, we should remove it and return zero. + */ + if (count == 1) { + sp = entry->sparse_head; + if (sp->offset == 0 && + sp->length >= archive_entry_size(entry)) { + count = 0; + archive_entry_sparse_clear(entry); + } + } + + return (count); +} + +int +archive_entry_sparse_reset(struct archive_entry * entry) +{ + entry->sparse_p = entry->sparse_head; + + return archive_entry_sparse_count(entry); +} + +int +archive_entry_sparse_next(struct archive_entry * entry, + la_int64_t *offset, la_int64_t *length) +{ + if (entry->sparse_p) { + *offset = entry->sparse_p->offset; + *length = entry->sparse_p->length; + + entry->sparse_p = entry->sparse_p->next; + + return (ARCHIVE_OK); + } else { + *offset = 0; + *length = 0; + return (ARCHIVE_WARN); + } +} + +/* + * end of sparse handling + */ diff --git a/c/archive_entry_stat.c b/c/archive_entry_stat.c new file mode 100644 --- /dev/null +++ b/c/archive_entry_stat.c @@ -0,0 +1,118 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_entry_stat.c 201100 2009-12-28 03:05:31Z kientzle $"); + +#ifdef HAVE_SYS_STAT_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif + +#include "archive_entry.h" +#include "archive_entry_private.h" + +const struct stat * +archive_entry_stat(struct archive_entry *entry) +{ + struct stat *st; + if (entry->stat == NULL) { + entry->stat = calloc(1, sizeof(*st)); + if (entry->stat == NULL) + return (NULL); + entry->stat_valid = 0; + } + + /* + * If none of the underlying fields have been changed, we + * don't need to regenerate. In theory, we could use a bitmap + * here to flag only those items that have changed, but the + * extra complexity probably isn't worth it. It will be very + * rare for anyone to change just one field then request a new + * stat structure. + */ + if (entry->stat_valid) + return (entry->stat); + + st = entry->stat; + /* + * Use the public interfaces to extract items, so that + * the appropriate conversions get invoked. + */ + st->st_atime = archive_entry_atime(entry); +#if HAVE_STRUCT_STAT_ST_BIRTHTIME + st->st_birthtime = archive_entry_birthtime(entry); +#endif + st->st_ctime = archive_entry_ctime(entry); + st->st_mtime = archive_entry_mtime(entry); + st->st_dev = archive_entry_dev(entry); + st->st_gid = (gid_t)archive_entry_gid(entry); + st->st_uid = (uid_t)archive_entry_uid(entry); + st->st_ino = (ino_t)archive_entry_ino64(entry); + st->st_nlink = archive_entry_nlink(entry); + st->st_rdev = archive_entry_rdev(entry); + st->st_size = (off_t)archive_entry_size(entry); + st->st_mode = archive_entry_mode(entry); + + /* + * On systems that support high-res timestamps, copy that + * information into struct stat. + */ +#if HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC + st->st_atimespec.tv_nsec = archive_entry_atime_nsec(entry); + st->st_ctimespec.tv_nsec = archive_entry_ctime_nsec(entry); + st->st_mtimespec.tv_nsec = archive_entry_mtime_nsec(entry); +#elif HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC + st->st_atim.tv_nsec = archive_entry_atime_nsec(entry); + st->st_ctim.tv_nsec = archive_entry_ctime_nsec(entry); + st->st_mtim.tv_nsec = archive_entry_mtime_nsec(entry); +#elif HAVE_STRUCT_STAT_ST_MTIME_N + st->st_atime_n = archive_entry_atime_nsec(entry); + st->st_ctime_n = archive_entry_ctime_nsec(entry); + st->st_mtime_n = archive_entry_mtime_nsec(entry); +#elif HAVE_STRUCT_STAT_ST_UMTIME + st->st_uatime = archive_entry_atime_nsec(entry) / 1000; + st->st_uctime = archive_entry_ctime_nsec(entry) / 1000; + st->st_umtime = archive_entry_mtime_nsec(entry) / 1000; +#elif HAVE_STRUCT_STAT_ST_MTIME_USEC + st->st_atime_usec = archive_entry_atime_nsec(entry) / 1000; + st->st_ctime_usec = archive_entry_ctime_nsec(entry) / 1000; + st->st_mtime_usec = archive_entry_mtime_nsec(entry) / 1000; +#endif +#if HAVE_STRUCT_STAT_ST_BIRTHTIMESPEC_TV_NSEC + st->st_birthtimespec.tv_nsec = archive_entry_birthtime_nsec(entry); +#endif + + /* + * TODO: On Linux, store 32 or 64 here depending on whether + * the cached stat structure is a stat32 or a stat64. This + * will allow us to support both variants interchangeably. + */ + entry->stat_valid = 1; + + return (st); +} diff --git a/c/archive_entry_strmode.c b/c/archive_entry_strmode.c new file mode 100644 --- /dev/null +++ b/c/archive_entry_strmode.c @@ -0,0 +1,87 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: src/lib/libarchive/archive_entry_strmode.c,v 1.4 2008/06/15 05:14:01 kientzle Exp $"); + +#ifdef HAVE_SYS_STAT_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive_entry.h" +#include "archive_entry_private.h" + +const char * +archive_entry_strmode(struct archive_entry *entry) +{ + static const mode_t permbits[] = + { 0400, 0200, 0100, 0040, 0020, 0010, 0004, 0002, 0001 }; + char *bp = entry->strmode; + mode_t mode; + int i; + + /* Fill in a default string, then selectively override. */ + strcpy(bp, "?rwxrwxrwx "); + + mode = archive_entry_mode(entry); + switch (archive_entry_filetype(entry)) { + case AE_IFREG: bp[0] = '-'; break; + case AE_IFBLK: bp[0] = 'b'; break; + case AE_IFCHR: bp[0] = 'c'; break; + case AE_IFDIR: bp[0] = 'd'; break; + case AE_IFLNK: bp[0] = 'l'; break; + case AE_IFSOCK: bp[0] = 's'; break; + case AE_IFIFO: bp[0] = 'p'; break; + default: + if (archive_entry_hardlink(entry) != NULL) { + bp[0] = 'h'; + break; + } + } + + for (i = 0; i < 9; i++) + if (!(mode & permbits[i])) + bp[i+1] = '-'; + + if (mode & S_ISUID) { + if (mode & 0100) bp[3] = 's'; + else bp[3] = 'S'; + } + if (mode & S_ISGID) { + if (mode & 0010) bp[6] = 's'; + else bp[6] = 'S'; + } + if (mode & S_ISVTX) { + if (mode & 0001) bp[9] = 't'; + else bp[9] = 'T'; + } + if (archive_entry_acl_types(entry) != 0) + bp[10] = '+'; + + return (bp); +} diff --git a/c/archive_entry_xattr.c b/c/archive_entry_xattr.c new file mode 100644 --- /dev/null +++ b/c/archive_entry_xattr.c @@ -0,0 +1,156 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_entry_xattr.c 201096 2009-12-28 02:41:27Z kientzle $"); + +#ifdef HAVE_SYS_STAT_H +#include +#endif +#ifdef HAVE_SYS_TYPES_H +#include +#endif +#ifdef HAVE_LIMITS_H +#include +#endif +#ifdef HAVE_LINUX_FS_H +#include /* for Linux file flags */ +#endif +/* + * Some Linux distributions have both linux/ext2_fs.h and ext2fs/ext2_fs.h. + * As the include guards don't agree, the order of include is important. + */ +#ifdef HAVE_LINUX_EXT2_FS_H +#include /* for Linux file flags */ +#endif +#if defined(HAVE_EXT2FS_EXT2_FS_H) && !defined(__CYGWIN__) +#include /* for Linux file flags */ +#endif +#include +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_WCHAR_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_entry_private.h" + +/* + * extended attribute handling + */ + +void +archive_entry_xattr_clear(struct archive_entry *entry) +{ + struct ae_xattr *xp; + + while (entry->xattr_head != NULL) { + xp = entry->xattr_head->next; + free(entry->xattr_head->name); + free(entry->xattr_head->value); + free(entry->xattr_head); + entry->xattr_head = xp; + } + + entry->xattr_head = NULL; +} + +void +archive_entry_xattr_add_entry(struct archive_entry *entry, + const char *name, const void *value, size_t size) +{ + struct ae_xattr *xp; + + if ((xp = (struct ae_xattr *)malloc(sizeof(struct ae_xattr))) == NULL) + __archive_errx(1, "Out of memory"); + + if ((xp->name = strdup(name)) == NULL) + __archive_errx(1, "Out of memory"); + + if ((xp->value = malloc(size)) != NULL) { + memcpy(xp->value, value, size); + xp->size = size; + } else + xp->size = 0; + + xp->next = entry->xattr_head; + entry->xattr_head = xp; +} + + +/* + * returns number of the extended attribute entries + */ +int +archive_entry_xattr_count(struct archive_entry *entry) +{ + struct ae_xattr *xp; + int count = 0; + + for (xp = entry->xattr_head; xp != NULL; xp = xp->next) + count++; + + return count; +} + +int +archive_entry_xattr_reset(struct archive_entry * entry) +{ + entry->xattr_p = entry->xattr_head; + + return archive_entry_xattr_count(entry); +} + +int +archive_entry_xattr_next(struct archive_entry * entry, + const char **name, const void **value, size_t *size) +{ + if (entry->xattr_p) { + *name = entry->xattr_p->name; + *value = entry->xattr_p->value; + *size = entry->xattr_p->size; + + entry->xattr_p = entry->xattr_p->next; + + return (ARCHIVE_OK); + } else { + *name = NULL; + *value = NULL; + *size = (size_t)0; + return (ARCHIVE_WARN); + } +} + +/* + * end of xattr handling + */ diff --git a/c/archive_getdate.c b/c/archive_getdate.c new file mode 100644 --- /dev/null +++ b/c/archive_getdate.c @@ -0,0 +1,1165 @@ +/* + * This code is in the public domain and has no copyright. + * + * This is a plain C recursive-descent translation of an old + * public-domain YACC grammar that has been used for parsing dates in + * very many open-source projects. + * + * Since the original authors were generous enough to donate their + * work to the public domain, I feel compelled to match their + * generosity. + * + * Tim Kientzle, February 2009. + */ + +/* + * Header comment from original getdate.y: + */ + +/* +** Originally written by Steven M. Bellovin while +** at the University of North Carolina at Chapel Hill. Later tweaked by +** a couple of people on Usenet. Completely overhauled by Rich $alz +** and Jim Berets in August, 1990; +** +** This grammar has 10 shift/reduce conflicts. +** +** This code is in the public domain and has no copyright. +*/ + +#include "archive_platform.h" +#ifdef __FreeBSD__ +#include +__FBSDID("$FreeBSD$"); +#endif + +#include +#include +#include +#include +#include + +#define __LIBARCHIVE_BUILD 1 +#include "archive_getdate.h" + +/* Basic time units. */ +#define EPOCH 1970 +#define MINUTE (60L) +#define HOUR (60L * MINUTE) +#define DAY (24L * HOUR) + +/* Daylight-savings mode: on, off, or not yet known. */ +enum DSTMODE { DSTon, DSToff, DSTmaybe }; +/* Meridian: am or pm. */ +enum { tAM, tPM }; +/* Token types returned by nexttoken() */ +enum { tAGO = 260, tDAY, tDAYZONE, tAMPM, tMONTH, tMONTH_UNIT, tSEC_UNIT, + tUNUMBER, tZONE, tDST }; +struct token { int token; time_t value; }; + +/* + * Parser state. + */ +struct gdstate { + struct token *tokenp; /* Pointer to next token. */ + /* HaveXxxx counts how many of this kind of phrase we've seen; + * it's a fatal error to have more than one time, zone, day, + * or date phrase. */ + int HaveYear; + int HaveMonth; + int HaveDay; + int HaveWeekDay; /* Day of week */ + int HaveTime; /* Hour/minute/second */ + int HaveZone; /* timezone and/or DST info */ + int HaveRel; /* time offset; we can have more than one */ + /* Absolute time values. */ + time_t Timezone; /* Seconds offset from GMT */ + time_t Day; + time_t Hour; + time_t Minutes; + time_t Month; + time_t Seconds; + time_t Year; + /* DST selection */ + enum DSTMODE DSTmode; + /* Day of week accounting, e.g., "3rd Tuesday" */ + time_t DayOrdinal; /* "3" in "3rd Tuesday" */ + time_t DayNumber; /* "Tuesday" in "3rd Tuesday" */ + /* Relative time values: hour/day/week offsets are measured in + * seconds, month/year are counted in months. */ + time_t RelMonth; + time_t RelSeconds; +}; + +/* + * A series of functions that recognize certain common time phrases. + * Each function returns 1 if it managed to make sense of some of the + * tokens, zero otherwise. + */ + +/* + * hour:minute or hour:minute:second with optional AM, PM, or numeric + * timezone offset + */ +static int +timephrase(struct gdstate *gds) +{ + if (gds->tokenp[0].token == tUNUMBER + && gds->tokenp[1].token == ':' + && gds->tokenp[2].token == tUNUMBER + && gds->tokenp[3].token == ':' + && gds->tokenp[4].token == tUNUMBER) { + /* "12:14:18" or "22:08:07" */ + ++gds->HaveTime; + gds->Hour = gds->tokenp[0].value; + gds->Minutes = gds->tokenp[2].value; + gds->Seconds = gds->tokenp[4].value; + gds->tokenp += 5; + } + else if (gds->tokenp[0].token == tUNUMBER + && gds->tokenp[1].token == ':' + && gds->tokenp[2].token == tUNUMBER) { + /* "12:14" or "22:08" */ + ++gds->HaveTime; + gds->Hour = gds->tokenp[0].value; + gds->Minutes = gds->tokenp[2].value; + gds->Seconds = 0; + gds->tokenp += 3; + } + else if (gds->tokenp[0].token == tUNUMBER + && gds->tokenp[1].token == tAMPM) { + /* "7" is a time if it's followed by "am" or "pm" */ + ++gds->HaveTime; + gds->Hour = gds->tokenp[0].value; + gds->Minutes = gds->Seconds = 0; + /* We'll handle the AM/PM below. */ + gds->tokenp += 1; + } else { + /* We can't handle this. */ + return 0; + } + + if (gds->tokenp[0].token == tAMPM) { + /* "7:12pm", "12:20:13am" */ + if (gds->Hour == 12) + gds->Hour = 0; + if (gds->tokenp[0].value == tPM) + gds->Hour += 12; + gds->tokenp += 1; + } + if (gds->tokenp[0].token == '+' + && gds->tokenp[1].token == tUNUMBER) { + /* "7:14+0700" */ + gds->HaveZone++; + gds->DSTmode = DSToff; + gds->Timezone = - ((gds->tokenp[1].value / 100) * HOUR + + (gds->tokenp[1].value % 100) * MINUTE); + gds->tokenp += 2; + } + if (gds->tokenp[0].token == '-' + && gds->tokenp[1].token == tUNUMBER) { + /* "19:14:12-0530" */ + gds->HaveZone++; + gds->DSTmode = DSToff; + gds->Timezone = + ((gds->tokenp[1].value / 100) * HOUR + + (gds->tokenp[1].value % 100) * MINUTE); + gds->tokenp += 2; + } + return 1; +} + +/* + * Timezone name, possibly including DST. + */ +static int +zonephrase(struct gdstate *gds) +{ + if (gds->tokenp[0].token == tZONE + && gds->tokenp[1].token == tDST) { + gds->HaveZone++; + gds->Timezone = gds->tokenp[0].value; + gds->DSTmode = DSTon; + gds->tokenp += 1; + return 1; + } + + if (gds->tokenp[0].token == tZONE) { + gds->HaveZone++; + gds->Timezone = gds->tokenp[0].value; + gds->DSTmode = DSToff; + gds->tokenp += 1; + return 1; + } + + if (gds->tokenp[0].token == tDAYZONE) { + gds->HaveZone++; + gds->Timezone = gds->tokenp[0].value; + gds->DSTmode = DSTon; + gds->tokenp += 1; + return 1; + } + return 0; +} + +/* + * Year/month/day in various combinations. + */ +static int +datephrase(struct gdstate *gds) +{ + if (gds->tokenp[0].token == tUNUMBER + && gds->tokenp[1].token == '/' + && gds->tokenp[2].token == tUNUMBER + && gds->tokenp[3].token == '/' + && gds->tokenp[4].token == tUNUMBER) { + gds->HaveYear++; + gds->HaveMonth++; + gds->HaveDay++; + if (gds->tokenp[0].value >= 13) { + /* First number is big: 2004/01/29, 99/02/17 */ + gds->Year = gds->tokenp[0].value; + gds->Month = gds->tokenp[2].value; + gds->Day = gds->tokenp[4].value; + } else if ((gds->tokenp[4].value >= 13) + || (gds->tokenp[2].value >= 13)) { + /* Last number is big: 01/07/98 */ + /* Middle number is big: 01/29/04 */ + gds->Month = gds->tokenp[0].value; + gds->Day = gds->tokenp[2].value; + gds->Year = gds->tokenp[4].value; + } else { + /* No significant clues: 02/03/04 */ + gds->Month = gds->tokenp[0].value; + gds->Day = gds->tokenp[2].value; + gds->Year = gds->tokenp[4].value; + } + gds->tokenp += 5; + return 1; + } + + if (gds->tokenp[0].token == tUNUMBER + && gds->tokenp[1].token == '/' + && gds->tokenp[2].token == tUNUMBER) { + /* "1/15" */ + gds->HaveMonth++; + gds->HaveDay++; + gds->Month = gds->tokenp[0].value; + gds->Day = gds->tokenp[2].value; + gds->tokenp += 3; + return 1; + } + + if (gds->tokenp[0].token == tUNUMBER + && gds->tokenp[1].token == '-' + && gds->tokenp[2].token == tUNUMBER + && gds->tokenp[3].token == '-' + && gds->tokenp[4].token == tUNUMBER) { + /* ISO 8601 format. yyyy-mm-dd. */ + gds->HaveYear++; + gds->HaveMonth++; + gds->HaveDay++; + gds->Year = gds->tokenp[0].value; + gds->Month = gds->tokenp[2].value; + gds->Day = gds->tokenp[4].value; + gds->tokenp += 5; + return 1; + } + + if (gds->tokenp[0].token == tUNUMBER + && gds->tokenp[1].token == '-' + && gds->tokenp[2].token == tMONTH + && gds->tokenp[3].token == '-' + && gds->tokenp[4].token == tUNUMBER) { + gds->HaveYear++; + gds->HaveMonth++; + gds->HaveDay++; + if (gds->tokenp[0].value > 31) { + /* e.g. 1992-Jun-17 */ + gds->Year = gds->tokenp[0].value; + gds->Month = gds->tokenp[2].value; + gds->Day = gds->tokenp[4].value; + } else { + /* e.g. 17-JUN-1992. */ + gds->Day = gds->tokenp[0].value; + gds->Month = gds->tokenp[2].value; + gds->Year = gds->tokenp[4].value; + } + gds->tokenp += 5; + return 1; + } + + if (gds->tokenp[0].token == tMONTH + && gds->tokenp[1].token == tUNUMBER + && gds->tokenp[2].token == ',' + && gds->tokenp[3].token == tUNUMBER) { + /* "June 17, 2001" */ + gds->HaveYear++; + gds->HaveMonth++; + gds->HaveDay++; + gds->Month = gds->tokenp[0].value; + gds->Day = gds->tokenp[1].value; + gds->Year = gds->tokenp[3].value; + gds->tokenp += 4; + return 1; + } + + if (gds->tokenp[0].token == tMONTH + && gds->tokenp[1].token == tUNUMBER) { + /* "May 3" */ + gds->HaveMonth++; + gds->HaveDay++; + gds->Month = gds->tokenp[0].value; + gds->Day = gds->tokenp[1].value; + gds->tokenp += 2; + return 1; + } + + if (gds->tokenp[0].token == tUNUMBER + && gds->tokenp[1].token == tMONTH + && gds->tokenp[2].token == tUNUMBER) { + /* "12 Sept 1997" */ + gds->HaveYear++; + gds->HaveMonth++; + gds->HaveDay++; + gds->Day = gds->tokenp[0].value; + gds->Month = gds->tokenp[1].value; + gds->Year = gds->tokenp[2].value; + gds->tokenp += 3; + return 1; + } + + if (gds->tokenp[0].token == tUNUMBER + && gds->tokenp[1].token == tMONTH) { + /* "12 Sept" */ + gds->HaveMonth++; + gds->HaveDay++; + gds->Day = gds->tokenp[0].value; + gds->Month = gds->tokenp[1].value; + gds->tokenp += 2; + return 1; + } + + return 0; +} + +/* + * Relative time phrase: "tomorrow", "yesterday", "+1 hour", etc. + */ +static int +relunitphrase(struct gdstate *gds) +{ + if (gds->tokenp[0].token == '-' + && gds->tokenp[1].token == tUNUMBER + && gds->tokenp[2].token == tSEC_UNIT) { + /* "-3 hours" */ + gds->HaveRel++; + gds->RelSeconds -= gds->tokenp[1].value * gds->tokenp[2].value; + gds->tokenp += 3; + return 1; + } + if (gds->tokenp[0].token == '+' + && gds->tokenp[1].token == tUNUMBER + && gds->tokenp[2].token == tSEC_UNIT) { + /* "+1 minute" */ + gds->HaveRel++; + gds->RelSeconds += gds->tokenp[1].value * gds->tokenp[2].value; + gds->tokenp += 3; + return 1; + } + if (gds->tokenp[0].token == tUNUMBER + && gds->tokenp[1].token == tSEC_UNIT) { + /* "1 day" */ + gds->HaveRel++; + gds->RelSeconds += gds->tokenp[0].value * gds->tokenp[1].value; + gds->tokenp += 2; + return 1; + } + if (gds->tokenp[0].token == '-' + && gds->tokenp[1].token == tUNUMBER + && gds->tokenp[2].token == tMONTH_UNIT) { + /* "-3 months" */ + gds->HaveRel++; + gds->RelMonth -= gds->tokenp[1].value * gds->tokenp[2].value; + gds->tokenp += 3; + return 1; + } + if (gds->tokenp[0].token == '+' + && gds->tokenp[1].token == tUNUMBER + && gds->tokenp[2].token == tMONTH_UNIT) { + /* "+5 years" */ + gds->HaveRel++; + gds->RelMonth += gds->tokenp[1].value * gds->tokenp[2].value; + gds->tokenp += 3; + return 1; + } + if (gds->tokenp[0].token == tUNUMBER + && gds->tokenp[1].token == tMONTH_UNIT) { + /* "2 years" */ + gds->HaveRel++; + gds->RelMonth += gds->tokenp[0].value * gds->tokenp[1].value; + gds->tokenp += 2; + return 1; + } + if (gds->tokenp[0].token == tSEC_UNIT) { + /* "now", "tomorrow" */ + gds->HaveRel++; + gds->RelSeconds += gds->tokenp[0].value; + gds->tokenp += 1; + return 1; + } + if (gds->tokenp[0].token == tMONTH_UNIT) { + /* "month" */ + gds->HaveRel++; + gds->RelMonth += gds->tokenp[0].value; + gds->tokenp += 1; + return 1; + } + return 0; +} + +/* + * Day of the week specification. + */ +static int +dayphrase(struct gdstate *gds) +{ + if (gds->tokenp[0].token == tDAY) { + /* "tues", "wednesday," */ + gds->HaveWeekDay++; + gds->DayOrdinal = 1; + gds->DayNumber = gds->tokenp[0].value; + gds->tokenp += 1; + if (gds->tokenp[0].token == ',') + gds->tokenp += 1; + return 1; + } + if (gds->tokenp[0].token == tUNUMBER + && gds->tokenp[1].token == tDAY) { + /* "second tues" "3 wed" */ + gds->HaveWeekDay++; + gds->DayOrdinal = gds->tokenp[0].value; + gds->DayNumber = gds->tokenp[1].value; + gds->tokenp += 2; + return 1; + } + return 0; +} + +/* + * Try to match a phrase using one of the above functions. + * This layer also deals with a couple of generic issues. + */ +static int +phrase(struct gdstate *gds) +{ + if (timephrase(gds)) + return 1; + if (zonephrase(gds)) + return 1; + if (datephrase(gds)) + return 1; + if (dayphrase(gds)) + return 1; + if (relunitphrase(gds)) { + if (gds->tokenp[0].token == tAGO) { + gds->RelSeconds = -gds->RelSeconds; + gds->RelMonth = -gds->RelMonth; + gds->tokenp += 1; + } + return 1; + } + + /* Bare numbers sometimes have meaning. */ + if (gds->tokenp[0].token == tUNUMBER) { + if (gds->HaveTime && !gds->HaveYear && !gds->HaveRel) { + gds->HaveYear++; + gds->Year = gds->tokenp[0].value; + gds->tokenp += 1; + return 1; + } + + if(gds->tokenp[0].value > 10000) { + /* "20040301" */ + gds->HaveYear++; + gds->HaveMonth++; + gds->HaveDay++; + gds->Day= (gds->tokenp[0].value)%100; + gds->Month= (gds->tokenp[0].value/100)%100; + gds->Year = gds->tokenp[0].value/10000; + gds->tokenp += 1; + return 1; + } + + if (gds->tokenp[0].value < 24) { + gds->HaveTime++; + gds->Hour = gds->tokenp[0].value; + gds->Minutes = 0; + gds->Seconds = 0; + gds->tokenp += 1; + return 1; + } + + if ((gds->tokenp[0].value / 100 < 24) + && (gds->tokenp[0].value % 100 < 60)) { + /* "513" is same as "5:13" */ + gds->Hour = gds->tokenp[0].value / 100; + gds->Minutes = gds->tokenp[0].value % 100; + gds->Seconds = 0; + gds->tokenp += 1; + return 1; + } + } + + return 0; +} + +/* + * A dictionary of time words. + */ +static struct LEXICON { + size_t abbrev; + const char *name; + int type; + time_t value; +} const TimeWords[] = { + /* am/pm */ + { 0, "am", tAMPM, tAM }, + { 0, "pm", tAMPM, tPM }, + + /* Month names. */ + { 3, "january", tMONTH, 1 }, + { 3, "february", tMONTH, 2 }, + { 3, "march", tMONTH, 3 }, + { 3, "april", tMONTH, 4 }, + { 3, "may", tMONTH, 5 }, + { 3, "june", tMONTH, 6 }, + { 3, "july", tMONTH, 7 }, + { 3, "august", tMONTH, 8 }, + { 3, "september", tMONTH, 9 }, + { 3, "october", tMONTH, 10 }, + { 3, "november", tMONTH, 11 }, + { 3, "december", tMONTH, 12 }, + + /* Days of the week. */ + { 2, "sunday", tDAY, 0 }, + { 3, "monday", tDAY, 1 }, + { 2, "tuesday", tDAY, 2 }, + { 3, "wednesday", tDAY, 3 }, + { 2, "thursday", tDAY, 4 }, + { 2, "friday", tDAY, 5 }, + { 2, "saturday", tDAY, 6 }, + + /* Timezones: Offsets are in seconds. */ + { 0, "gmt", tZONE, 0*HOUR }, /* Greenwich Mean */ + { 0, "ut", tZONE, 0*HOUR }, /* Universal (Coordinated) */ + { 0, "utc", tZONE, 0*HOUR }, + { 0, "wet", tZONE, 0*HOUR }, /* Western European */ + { 0, "bst", tDAYZONE, 0*HOUR }, /* British Summer */ + { 0, "wat", tZONE, 1*HOUR }, /* West Africa */ + { 0, "at", tZONE, 2*HOUR }, /* Azores */ + /* { 0, "bst", tZONE, 3*HOUR }, */ /* Brazil Standard: Conflict */ + /* { 0, "gst", tZONE, 3*HOUR }, */ /* Greenland Standard: Conflict*/ + { 0, "nft", tZONE, 3*HOUR+30*MINUTE }, /* Newfoundland */ + { 0, "nst", tZONE, 3*HOUR+30*MINUTE }, /* Newfoundland Standard */ + { 0, "ndt", tDAYZONE, 3*HOUR+30*MINUTE }, /* Newfoundland Daylight */ + { 0, "ast", tZONE, 4*HOUR }, /* Atlantic Standard */ + { 0, "adt", tDAYZONE, 4*HOUR }, /* Atlantic Daylight */ + { 0, "est", tZONE, 5*HOUR }, /* Eastern Standard */ + { 0, "edt", tDAYZONE, 5*HOUR }, /* Eastern Daylight */ + { 0, "cst", tZONE, 6*HOUR }, /* Central Standard */ + { 0, "cdt", tDAYZONE, 6*HOUR }, /* Central Daylight */ + { 0, "mst", tZONE, 7*HOUR }, /* Mountain Standard */ + { 0, "mdt", tDAYZONE, 7*HOUR }, /* Mountain Daylight */ + { 0, "pst", tZONE, 8*HOUR }, /* Pacific Standard */ + { 0, "pdt", tDAYZONE, 8*HOUR }, /* Pacific Daylight */ + { 0, "yst", tZONE, 9*HOUR }, /* Yukon Standard */ + { 0, "ydt", tDAYZONE, 9*HOUR }, /* Yukon Daylight */ + { 0, "hst", tZONE, 10*HOUR }, /* Hawaii Standard */ + { 0, "hdt", tDAYZONE, 10*HOUR }, /* Hawaii Daylight */ + { 0, "cat", tZONE, 10*HOUR }, /* Central Alaska */ + { 0, "ahst", tZONE, 10*HOUR }, /* Alaska-Hawaii Standard */ + { 0, "nt", tZONE, 11*HOUR }, /* Nome */ + { 0, "idlw", tZONE, 12*HOUR }, /* Intl Date Line West */ + { 0, "cet", tZONE, -1*HOUR }, /* Central European */ + { 0, "met", tZONE, -1*HOUR }, /* Middle European */ + { 0, "mewt", tZONE, -1*HOUR }, /* Middle European Winter */ + { 0, "mest", tDAYZONE, -1*HOUR }, /* Middle European Summer */ + { 0, "swt", tZONE, -1*HOUR }, /* Swedish Winter */ + { 0, "sst", tDAYZONE, -1*HOUR }, /* Swedish Summer */ + { 0, "fwt", tZONE, -1*HOUR }, /* French Winter */ + { 0, "fst", tDAYZONE, -1*HOUR }, /* French Summer */ + { 0, "eet", tZONE, -2*HOUR }, /* Eastern Eur, USSR Zone 1 */ + { 0, "bt", tZONE, -3*HOUR }, /* Baghdad, USSR Zone 2 */ + { 0, "it", tZONE, -3*HOUR-30*MINUTE },/* Iran */ + { 0, "zp4", tZONE, -4*HOUR }, /* USSR Zone 3 */ + { 0, "zp5", tZONE, -5*HOUR }, /* USSR Zone 4 */ + { 0, "ist", tZONE, -5*HOUR-30*MINUTE },/* Indian Standard */ + { 0, "zp6", tZONE, -6*HOUR }, /* USSR Zone 5 */ + /* { 0, "nst", tZONE, -6.5*HOUR }, */ /* North Sumatra: Conflict */ + /* { 0, "sst", tZONE, -7*HOUR }, */ /* So Sumatra, USSR 6: Conflict */ + { 0, "wast", tZONE, -7*HOUR }, /* West Australian Standard */ + { 0, "wadt", tDAYZONE, -7*HOUR }, /* West Australian Daylight */ + { 0, "jt", tZONE, -7*HOUR-30*MINUTE },/* Java (3pm in Cronusland!)*/ + { 0, "cct", tZONE, -8*HOUR }, /* China Coast, USSR Zone 7 */ + { 0, "jst", tZONE, -9*HOUR }, /* Japan Std, USSR Zone 8 */ + { 0, "cast", tZONE, -9*HOUR-30*MINUTE },/* Ctrl Australian Std */ + { 0, "cadt", tDAYZONE, -9*HOUR-30*MINUTE },/* Ctrl Australian Daylt */ + { 0, "east", tZONE, -10*HOUR }, /* Eastern Australian Std */ + { 0, "eadt", tDAYZONE, -10*HOUR }, /* Eastern Australian Daylt */ + { 0, "gst", tZONE, -10*HOUR }, /* Guam Std, USSR Zone 9 */ + { 0, "nzt", tZONE, -12*HOUR }, /* New Zealand */ + { 0, "nzst", tZONE, -12*HOUR }, /* New Zealand Standard */ + { 0, "nzdt", tDAYZONE, -12*HOUR }, /* New Zealand Daylight */ + { 0, "idle", tZONE, -12*HOUR }, /* Intl Date Line East */ + + { 0, "dst", tDST, 0 }, + + /* Time units. */ + { 4, "years", tMONTH_UNIT, 12 }, + { 5, "months", tMONTH_UNIT, 1 }, + { 9, "fortnights", tSEC_UNIT, 14 * DAY }, + { 4, "weeks", tSEC_UNIT, 7 * DAY }, + { 3, "days", tSEC_UNIT, DAY }, + { 4, "hours", tSEC_UNIT, HOUR }, + { 3, "minutes", tSEC_UNIT, MINUTE }, + { 3, "seconds", tSEC_UNIT, 1 }, + + /* Relative-time words. */ + { 0, "tomorrow", tSEC_UNIT, DAY }, + { 0, "yesterday", tSEC_UNIT, -DAY }, + { 0, "today", tSEC_UNIT, 0 }, + { 0, "now", tSEC_UNIT, 0 }, + { 0, "last", tUNUMBER, -1 }, + { 0, "this", tSEC_UNIT, 0 }, + { 0, "next", tUNUMBER, 2 }, + { 0, "first", tUNUMBER, 1 }, + { 0, "1st", tUNUMBER, 1 }, +/* { 0, "second", tUNUMBER, 2 }, */ + { 0, "2nd", tUNUMBER, 2 }, + { 0, "third", tUNUMBER, 3 }, + { 0, "3rd", tUNUMBER, 3 }, + { 0, "fourth", tUNUMBER, 4 }, + { 0, "4th", tUNUMBER, 4 }, + { 0, "fifth", tUNUMBER, 5 }, + { 0, "5th", tUNUMBER, 5 }, + { 0, "sixth", tUNUMBER, 6 }, + { 0, "seventh", tUNUMBER, 7 }, + { 0, "eighth", tUNUMBER, 8 }, + { 0, "ninth", tUNUMBER, 9 }, + { 0, "tenth", tUNUMBER, 10 }, + { 0, "eleventh", tUNUMBER, 11 }, + { 0, "twelfth", tUNUMBER, 12 }, + { 0, "ago", tAGO, 1 }, + + /* Military timezones. */ + { 0, "a", tZONE, 1*HOUR }, + { 0, "b", tZONE, 2*HOUR }, + { 0, "c", tZONE, 3*HOUR }, + { 0, "d", tZONE, 4*HOUR }, + { 0, "e", tZONE, 5*HOUR }, + { 0, "f", tZONE, 6*HOUR }, + { 0, "g", tZONE, 7*HOUR }, + { 0, "h", tZONE, 8*HOUR }, + { 0, "i", tZONE, 9*HOUR }, + { 0, "k", tZONE, 10*HOUR }, + { 0, "l", tZONE, 11*HOUR }, + { 0, "m", tZONE, 12*HOUR }, + { 0, "n", tZONE, -1*HOUR }, + { 0, "o", tZONE, -2*HOUR }, + { 0, "p", tZONE, -3*HOUR }, + { 0, "q", tZONE, -4*HOUR }, + { 0, "r", tZONE, -5*HOUR }, + { 0, "s", tZONE, -6*HOUR }, + { 0, "t", tZONE, -7*HOUR }, + { 0, "u", tZONE, -8*HOUR }, + { 0, "v", tZONE, -9*HOUR }, + { 0, "w", tZONE, -10*HOUR }, + { 0, "x", tZONE, -11*HOUR }, + { 0, "y", tZONE, -12*HOUR }, + { 0, "z", tZONE, 0*HOUR }, + + /* End of table. */ + { 0, NULL, 0, 0 } +}; + +/* + * Year is either: + * = A number from 0 to 99, which means a year from 1970 to 2069, or + * = The actual year (>=100). + */ +static time_t +Convert(time_t Month, time_t Day, time_t Year, + time_t Hours, time_t Minutes, time_t Seconds, + time_t Timezone, enum DSTMODE DSTmode) +{ + signed char DaysInMonth[12] = { + 31, 0, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31 + }; + time_t Julian; + int i; + struct tm *ltime; +#if defined(HAVE_LOCALTIME_R) || defined(HAVE__LOCALTIME64_S) + struct tm tmbuf; +#endif +#if defined(HAVE__LOCALTIME64_S) + errno_t terr; + __time64_t tmptime; +#endif + + if (Year < 69) + Year += 2000; + else if (Year < 100) + Year += 1900; + DaysInMonth[1] = Year % 4 == 0 && (Year % 100 != 0 || Year % 400 == 0) + ? 29 : 28; + /* Checking for 2038 bogusly assumes that time_t is 32 bits. But + I'm too lazy to try to check for time_t overflow in another way. */ + if (Year < EPOCH || Year > 2038 + || Month < 1 || Month > 12 + /* Lint fluff: "conversion from long may lose accuracy" */ + || Day < 1 || Day > DaysInMonth[(int)--Month] + || Hours < 0 || Hours > 23 + || Minutes < 0 || Minutes > 59 + || Seconds < 0 || Seconds > 59) + return -1; + + Julian = Day - 1; + for (i = 0; i < Month; i++) + Julian += DaysInMonth[i]; + for (i = EPOCH; i < Year; i++) + Julian += 365 + (i % 4 == 0); + Julian *= DAY; + Julian += Timezone; + Julian += Hours * HOUR + Minutes * MINUTE + Seconds; +#if defined(HAVE_LOCALTIME_R) + ltime = localtime_r(&Julian, &tmbuf); +#elif defined(HAVE__LOCALTIME64_S) + tmptime = Julian; + terr = _localtime64_s(&tmbuf, &tmptime); + if (terr) + ltime = NULL; + else + ltime = &tmbuf; +#else + ltime = localtime(&Julian); +#endif + if (DSTmode == DSTon + || (DSTmode == DSTmaybe && ltime->tm_isdst)) + Julian -= HOUR; + return Julian; +} + +static time_t +DSTcorrect(time_t Start, time_t Future) +{ + time_t StartDay; + time_t FutureDay; + struct tm *ltime; +#if defined(HAVE_LOCALTIME_R) || defined(HAVE__LOCALTIME64_S) + struct tm tmbuf; +#endif +#if defined(HAVE__LOCALTIME64_S) + errno_t terr; + __time64_t tmptime; +#endif + +#if defined(HAVE_LOCALTIME_R) + ltime = localtime_r(&Start, &tmbuf); +#elif defined(HAVE__LOCALTIME64_S) + tmptime = Start; + terr = _localtime64_s(&tmbuf, &tmptime); + if (terr) + ltime = NULL; + else + ltime = &tmbuf; +#else + ltime = localtime(&Start); +#endif + StartDay = (ltime->tm_hour + 1) % 24; +#if defined(HAVE_LOCALTIME_R) + ltime = localtime_r(&Future, &tmbuf); +#elif defined(HAVE__LOCALTIME64_S) + tmptime = Future; + terr = _localtime64_s(&tmbuf, &tmptime); + if (terr) + ltime = NULL; + else + ltime = &tmbuf; +#else + ltime = localtime(&Future); +#endif + FutureDay = (ltime->tm_hour + 1) % 24; + return (Future - Start) + (StartDay - FutureDay) * HOUR; +} + + +static time_t +RelativeDate(time_t Start, time_t zone, int dstmode, + time_t DayOrdinal, time_t DayNumber) +{ + struct tm *tm; + time_t t, now; +#if defined(HAVE_GMTIME_R) || defined(HAVE__GMTIME64_S) + struct tm tmbuf; +#endif +#if defined(HAVE__GMTIME64_S) + errno_t terr; + __time64_t tmptime; +#endif + + t = Start - zone; +#if defined(HAVE_GMTIME_R) + tm = gmtime_r(&t, &tmbuf); +#elif defined(HAVE__GMTIME64_S) + tmptime = t; + terr = _gmtime64_s(&tmbuf, &tmptime); + if (terr) + tm = NULL; + else + tm = &tmbuf; +#else + tm = gmtime(&t); +#endif + now = Start; + now += DAY * ((DayNumber - tm->tm_wday + 7) % 7); + now += 7 * DAY * (DayOrdinal <= 0 ? DayOrdinal : DayOrdinal - 1); + if (dstmode == DSTmaybe) + return DSTcorrect(Start, now); + return now - Start; +} + + +static time_t +RelativeMonth(time_t Start, time_t Timezone, time_t RelMonth) +{ + struct tm *tm; + time_t Month; + time_t Year; +#if defined(HAVE_LOCALTIME_R) || defined(HAVE__LOCALTIME64_S) + struct tm tmbuf; +#endif +#if defined(HAVE__LOCALTIME64_S) + errno_t terr; + __time64_t tmptime; +#endif + + if (RelMonth == 0) + return 0; +#if defined(HAVE_LOCALTIME_R) + tm = localtime_r(&Start, &tmbuf); +#elif defined(HAVE__LOCALTIME64_S) + tmptime = Start; + terr = _localtime64_s(&tmbuf, &tmptime); + if (terr) + tm = NULL; + else + tm = &tmbuf; +#else + tm = localtime(&Start); +#endif + Month = 12 * (tm->tm_year + 1900) + tm->tm_mon + RelMonth; + Year = Month / 12; + Month = Month % 12 + 1; + return DSTcorrect(Start, + Convert(Month, (time_t)tm->tm_mday, Year, + (time_t)tm->tm_hour, (time_t)tm->tm_min, (time_t)tm->tm_sec, + Timezone, DSTmaybe)); +} + +/* + * Tokenizer. + */ +static int +nexttoken(const char **in, time_t *value) +{ + char c; + char buff[64]; + + for ( ; ; ) { + while (isspace((unsigned char)**in)) + ++*in; + + /* Skip parenthesized comments. */ + if (**in == '(') { + int Count = 0; + do { + c = *(*in)++; + if (c == '\0') + return c; + if (c == '(') + Count++; + else if (c == ')') + Count--; + } while (Count > 0); + continue; + } + + /* Try the next token in the word table first. */ + /* This allows us to match "2nd", for example. */ + { + const char *src = *in; + const struct LEXICON *tp; + unsigned i = 0; + + /* Force to lowercase and strip '.' characters. */ + while (*src != '\0' + && (isalnum((unsigned char)*src) || *src == '.') + && i < sizeof(buff)-1) { + if (*src != '.') { + if (isupper((unsigned char)*src)) + buff[i++] = tolower((unsigned char)*src); + else + buff[i++] = *src; + } + src++; + } + buff[i] = '\0'; + + /* + * Find the first match. If the word can be + * abbreviated, make sure we match at least + * the minimum abbreviation. + */ + for (tp = TimeWords; tp->name; tp++) { + size_t abbrev = tp->abbrev; + if (abbrev == 0) + abbrev = strlen(tp->name); + if (strlen(buff) >= abbrev + && strncmp(tp->name, buff, strlen(buff)) + == 0) { + /* Skip over token. */ + *in = src; + /* Return the match. */ + *value = tp->value; + return tp->type; + } + } + } + + /* + * Not in the word table, maybe it's a number. Note: + * Because '-' and '+' have other special meanings, I + * don't deal with signed numbers here. + */ + if (isdigit((unsigned char)(c = **in))) { + for (*value = 0; isdigit((unsigned char)(c = *(*in)++)); ) + *value = 10 * *value + c - '0'; + (*in)--; + return (tUNUMBER); + } + + return *(*in)++; + } +} + +#define TM_YEAR_ORIGIN 1900 + +/* Yield A - B, measured in seconds. */ +static long +difftm (struct tm *a, struct tm *b) +{ + int ay = a->tm_year + (TM_YEAR_ORIGIN - 1); + int by = b->tm_year + (TM_YEAR_ORIGIN - 1); + int days = ( + /* difference in day of year */ + a->tm_yday - b->tm_yday + /* + intervening leap days */ + + ((ay >> 2) - (by >> 2)) + - (ay/100 - by/100) + + ((ay/100 >> 2) - (by/100 >> 2)) + /* + difference in years * 365 */ + + (long)(ay-by) * 365 + ); + return (days * DAY + (a->tm_hour - b->tm_hour) * HOUR + + (a->tm_min - b->tm_min) * MINUTE + + (a->tm_sec - b->tm_sec)); +} + +/* + * + * The public function. + * + * TODO: tokens[] array should be dynamically sized. + */ +time_t +__archive_get_date(time_t now, const char *p) +{ + struct token tokens[256]; + struct gdstate _gds; + struct token *lasttoken; + struct gdstate *gds; + struct tm local, *tm; + struct tm gmt, *gmt_ptr; + time_t Start; + time_t tod; + long tzone; +#if defined(HAVE__LOCALTIME64_S) || defined(HAVE__GMTIME64_S) + errno_t terr; + __time64_t tmptime; +#endif + + /* Clear out the parsed token array. */ + memset(tokens, 0, sizeof(tokens)); + /* Initialize the parser state. */ + memset(&_gds, 0, sizeof(_gds)); + gds = &_gds; + + /* Look up the current time. */ +#if defined(HAVE_LOCALTIME_R) + tm = localtime_r(&now, &local); +#elif defined(HAVE__LOCALTIME64_S) + tmptime = now; + terr = _localtime64_s(&local, &tmptime); + if (terr) + tm = NULL; + else + tm = &local; +#else + memset(&local, 0, sizeof(local)); + tm = localtime(&now); +#endif + if (tm == NULL) + return -1; +#if !defined(HAVE_LOCALTIME_R) && !defined(HAVE__LOCALTIME64_S) + local = *tm; +#endif + + /* Look up UTC if we can and use that to determine the current + * timezone offset. */ +#if defined(HAVE_GMTIME_R) + gmt_ptr = gmtime_r(&now, &gmt); +#elif defined(HAVE__GMTIME64_S) + tmptime = now; + terr = _gmtime64_s(&gmt, &tmptime); + if (terr) + gmt_ptr = NULL; + else + gmt_ptr = &gmt; +#else + memset(&gmt, 0, sizeof(gmt)); + gmt_ptr = gmtime(&now); + if (gmt_ptr != NULL) { + /* Copy, in case localtime and gmtime use the same buffer. */ + gmt = *gmt_ptr; + } +#endif + if (gmt_ptr != NULL) + tzone = difftm (&gmt, &local); + else + /* This system doesn't understand timezones; fake it. */ + tzone = 0; + if(local.tm_isdst) + tzone += HOUR; + + /* Tokenize the input string. */ + lasttoken = tokens; + while ((lasttoken->token = nexttoken(&p, &lasttoken->value)) != 0) { + ++lasttoken; + if (lasttoken > tokens + 255) + return -1; + } + gds->tokenp = tokens; + + /* Match phrases until we run out of input tokens. */ + while (gds->tokenp < lasttoken) { + if (!phrase(gds)) + return -1; + } + + /* Use current local timezone if none was specified. */ + if (!gds->HaveZone) { + gds->Timezone = tzone; + gds->DSTmode = DSTmaybe; + } + + /* If a timezone was specified, use that for generating the default + * time components instead of the local timezone. */ + if (gds->HaveZone && gmt_ptr != NULL) { + now -= gds->Timezone; +#if defined(HAVE_GMTIME_R) + gmt_ptr = gmtime_r(&now, &gmt); +#elif defined(HAVE__GMTIME64_S) + tmptime = now; + terr = _gmtime64_s(&gmt, &tmptime); + if (terr) + gmt_ptr = NULL; + else + gmt_ptr = &gmt; +#else + gmt_ptr = gmtime(&now); +#endif + if (gmt_ptr != NULL) + local = *gmt_ptr; + now += gds->Timezone; + } + + if (!gds->HaveYear) + gds->Year = local.tm_year + 1900; + if (!gds->HaveMonth) + gds->Month = local.tm_mon + 1; + if (!gds->HaveDay) + gds->Day = local.tm_mday; + /* Note: No default for hour/min/sec; a specifier that just + * gives date always refers to 00:00 on that date. */ + + /* If we saw more than one time, timezone, weekday, year, month, + * or day, then give up. */ + if (gds->HaveTime > 1 || gds->HaveZone > 1 || gds->HaveWeekDay > 1 + || gds->HaveYear > 1 || gds->HaveMonth > 1 || gds->HaveDay > 1) + return -1; + + /* Compute an absolute time based on whatever absolute information + * we collected. */ + if (gds->HaveYear || gds->HaveMonth || gds->HaveDay + || gds->HaveTime || gds->HaveWeekDay) { + Start = Convert(gds->Month, gds->Day, gds->Year, + gds->Hour, gds->Minutes, gds->Seconds, + gds->Timezone, gds->DSTmode); + if (Start < 0) + return -1; + } else { + Start = now; + if (!gds->HaveRel) + Start -= local.tm_hour * HOUR + local.tm_min * MINUTE + + local.tm_sec; + } + + /* Add the relative offset. */ + Start += gds->RelSeconds; + Start += RelativeMonth(Start, gds->Timezone, gds->RelMonth); + + /* Adjust for day-of-week offsets. */ + if (gds->HaveWeekDay + && !(gds->HaveYear || gds->HaveMonth || gds->HaveDay)) { + tod = RelativeDate(Start, gds->Timezone, + gds->DSTmode, gds->DayOrdinal, gds->DayNumber); + Start += tod; + } + + /* -1 is an error indicator, so return 0 instead of -1 if + * that's the actual time. */ + return Start == -1 ? 0 : Start; +} + + +#if defined(TEST) + +/* ARGSUSED */ +int +main(int argc, char **argv) +{ + time_t d; + time_t now = time(NULL); + + while (*++argv != NULL) { + (void)printf("Input: %s\n", *argv); + d = get_date(now, *argv); + if (d == -1) + (void)printf("Bad format - couldn't convert.\n"); + else + (void)printf("Output: %s\n", ctime(&d)); + } + exit(0); + /* NOTREACHED */ +} +#endif /* defined(TEST) */ diff --git a/c/archive_getdate.h b/c/archive_getdate.h new file mode 100644 --- /dev/null +++ b/c/archive_getdate.h @@ -0,0 +1,39 @@ +/*- + * Copyright (c) 2003-2015 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD$ + */ + +#ifndef ARCHIVE_GETDATE_H_INCLUDED +#define ARCHIVE_GETDATE_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif + +#include + +time_t __archive_get_date(time_t now, const char *); + +#endif diff --git a/c/archive_hmac.c b/c/archive_hmac.c new file mode 100644 --- /dev/null +++ b/c/archive_hmac.c @@ -0,0 +1,305 @@ +/*- +* Copyright (c) 2014 Michihiro NAKAJIMA +* All rights reserved. +* +* Redistribution and use in source and binary forms, with or without +* modification, are permitted provided that the following conditions +* are met: +* 1. Redistributions of source code must retain the above copyright +* notice, this list of conditions and the following disclaimer. +* 2. Redistributions in binary form must reproduce the above copyright +* notice, this list of conditions and the following disclaimer in the +* documentation and/or other materials provided with the distribution. +* +* THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR +* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES +* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. +* IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, +* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT +* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF +* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include "archive_platform.h" + +#ifdef HAVE_STRING_H +#include +#endif +#include "archive.h" +#include "archive_hmac_private.h" + +/* + * On systems that do not support any recognized crypto libraries, + * the archive_hmac.c file is expected to define no usable symbols. + * + * But some compilers and linkers choke on empty object files, so + * define a public symbol that will always exist. This could + * be removed someday if this file gains another always-present + * symbol definition. + */ +int __libarchive_hmac_build_hack(void) { + return 0; +} + + +#ifdef ARCHIVE_HMAC_USE_Apple_CommonCrypto + +static int +__hmac_sha1_init(archive_hmac_sha1_ctx *ctx, const uint8_t *key, size_t key_len) +{ + CCHmacInit(ctx, kCCHmacAlgSHA1, key, key_len); + return 0; +} + +static void +__hmac_sha1_update(archive_hmac_sha1_ctx *ctx, const uint8_t *data, + size_t data_len) +{ + CCHmacUpdate(ctx, data, data_len); +} + +static void +__hmac_sha1_final(archive_hmac_sha1_ctx *ctx, uint8_t *out, size_t *out_len) +{ + CCHmacFinal(ctx, out); + *out_len = 20; +} + +static void +__hmac_sha1_cleanup(archive_hmac_sha1_ctx *ctx) +{ + memset(ctx, 0, sizeof(*ctx)); +} + +#elif defined(_WIN32) && !defined(__CYGWIN__) && defined(HAVE_BCRYPT_H) + +#ifndef BCRYPT_HASH_REUSABLE_FLAG +# define BCRYPT_HASH_REUSABLE_FLAG 0x00000020 +#endif + +static int +__hmac_sha1_init(archive_hmac_sha1_ctx *ctx, const uint8_t *key, size_t key_len) +{ +#ifdef __GNUC__ +#pragma GCC diagnostic ignored "-Wcast-qual" +#endif + BCRYPT_ALG_HANDLE hAlg; + BCRYPT_HASH_HANDLE hHash; + DWORD hash_len; + PBYTE hash; + ULONG result; + NTSTATUS status; + + ctx->hAlg = NULL; + status = BCryptOpenAlgorithmProvider(&hAlg, BCRYPT_SHA1_ALGORITHM, + MS_PRIMITIVE_PROVIDER, BCRYPT_ALG_HANDLE_HMAC_FLAG); + if (!BCRYPT_SUCCESS(status)) + return -1; + status = BCryptGetProperty(hAlg, BCRYPT_HASH_LENGTH, (PUCHAR)&hash_len, + sizeof(hash_len), &result, 0); + if (!BCRYPT_SUCCESS(status)) { + BCryptCloseAlgorithmProvider(hAlg, 0); + return -1; + } + hash = (PBYTE)HeapAlloc(GetProcessHeap(), 0, hash_len); + if (hash == NULL) { + BCryptCloseAlgorithmProvider(hAlg, 0); + return -1; + } + status = BCryptCreateHash(hAlg, &hHash, NULL, 0, + (PUCHAR)key, (ULONG)key_len, BCRYPT_HASH_REUSABLE_FLAG); + if (!BCRYPT_SUCCESS(status)) { + BCryptCloseAlgorithmProvider(hAlg, 0); + HeapFree(GetProcessHeap(), 0, hash); + return -1; + } + + ctx->hAlg = hAlg; + ctx->hHash = hHash; + ctx->hash_len = hash_len; + ctx->hash = hash; + + return 0; +} + +static void +__hmac_sha1_update(archive_hmac_sha1_ctx *ctx, const uint8_t *data, + size_t data_len) +{ + BCryptHashData(ctx->hHash, (PUCHAR)(uintptr_t)data, (ULONG)data_len, 0); +} + +static void +__hmac_sha1_final(archive_hmac_sha1_ctx *ctx, uint8_t *out, size_t *out_len) +{ + BCryptFinishHash(ctx->hHash, ctx->hash, ctx->hash_len, 0); + if (ctx->hash_len == *out_len) + memcpy(out, ctx->hash, *out_len); +} + +static void +__hmac_sha1_cleanup(archive_hmac_sha1_ctx *ctx) +{ + if (ctx->hAlg != NULL) { + BCryptCloseAlgorithmProvider(ctx->hAlg, 0); + HeapFree(GetProcessHeap(), 0, ctx->hash); + ctx->hAlg = NULL; + } +} + +#elif defined(HAVE_LIBMBEDCRYPTO) && defined(HAVE_MBEDTLS_MD_H) + +static int +__hmac_sha1_init(archive_hmac_sha1_ctx *ctx, const uint8_t *key, size_t key_len) +{ + const mbedtls_md_info_t *info; + int ret; + + mbedtls_md_init(ctx); + info = mbedtls_md_info_from_type(MBEDTLS_MD_SHA1); + if (info == NULL) { + mbedtls_md_free(ctx); + return (-1); + } + ret = mbedtls_md_setup(ctx, info, 1); + if (ret != 0) { + mbedtls_md_free(ctx); + return (-1); + } + ret = mbedtls_md_hmac_starts(ctx, key, key_len); + if (ret != 0) { + mbedtls_md_free(ctx); + return (-1); + } + return 0; +} + +static void +__hmac_sha1_update(archive_hmac_sha1_ctx *ctx, const uint8_t *data, + size_t data_len) +{ + mbedtls_md_hmac_update(ctx, data, data_len); +} + +static void __hmac_sha1_final(archive_hmac_sha1_ctx *ctx, uint8_t *out, size_t *out_len) +{ + (void)out_len; /* UNUSED */ + + mbedtls_md_hmac_finish(ctx, out); +} + +static void __hmac_sha1_cleanup(archive_hmac_sha1_ctx *ctx) +{ + mbedtls_md_free(ctx); + memset(ctx, 0, sizeof(*ctx)); +} + +#elif defined(HAVE_LIBNETTLE) && defined(HAVE_NETTLE_HMAC_H) + +static int +__hmac_sha1_init(archive_hmac_sha1_ctx *ctx, const uint8_t *key, size_t key_len) +{ + hmac_sha1_set_key(ctx, key_len, key); + return 0; +} + +static void +__hmac_sha1_update(archive_hmac_sha1_ctx *ctx, const uint8_t *data, + size_t data_len) +{ + hmac_sha1_update(ctx, data_len, data); +} + +static void +__hmac_sha1_final(archive_hmac_sha1_ctx *ctx, uint8_t *out, size_t *out_len) +{ + hmac_sha1_digest(ctx, (unsigned)*out_len, out); +} + +static void +__hmac_sha1_cleanup(archive_hmac_sha1_ctx *ctx) +{ + memset(ctx, 0, sizeof(*ctx)); +} + +#elif defined(HAVE_LIBCRYPTO) + +static int +__hmac_sha1_init(archive_hmac_sha1_ctx *ctx, const uint8_t *key, size_t key_len) +{ + *ctx = HMAC_CTX_new(); + if (*ctx == NULL) + return -1; + HMAC_Init_ex(*ctx, key, key_len, EVP_sha1(), NULL); + return 0; +} + +static void +__hmac_sha1_update(archive_hmac_sha1_ctx *ctx, const uint8_t *data, + size_t data_len) +{ + HMAC_Update(*ctx, data, data_len); +} + +static void +__hmac_sha1_final(archive_hmac_sha1_ctx *ctx, uint8_t *out, size_t *out_len) +{ + unsigned int len = (unsigned int)*out_len; + + HMAC_Final(*ctx, out, &len); + *out_len = len; +} + +static void +__hmac_sha1_cleanup(archive_hmac_sha1_ctx *ctx) +{ + HMAC_CTX_free(*ctx); + *ctx = NULL; +} + +#else + +/* Stub */ +static int +__hmac_sha1_init(archive_hmac_sha1_ctx *ctx, const uint8_t *key, size_t key_len) +{ + (void)ctx;/* UNUSED */ + (void)key;/* UNUSED */ + (void)key_len;/* UNUSED */ + return -1; +} + +static void +__hmac_sha1_update(archive_hmac_sha1_ctx *ctx, const uint8_t *data, + size_t data_len) +{ + (void)ctx;/* UNUSED */ + (void)data;/* UNUSED */ + (void)data_len;/* UNUSED */ +} + +static void +__hmac_sha1_final(archive_hmac_sha1_ctx *ctx, uint8_t *out, size_t *out_len) +{ + (void)ctx;/* UNUSED */ + (void)out;/* UNUSED */ + (void)out_len;/* UNUSED */ +} + +static void +__hmac_sha1_cleanup(archive_hmac_sha1_ctx *ctx) +{ + (void)ctx;/* UNUSED */ +} + +#endif + +const struct archive_hmac __archive_hmac = { + &__hmac_sha1_init, + &__hmac_sha1_update, + &__hmac_sha1_final, + &__hmac_sha1_cleanup, +}; diff --git a/c/archive_hmac_private.h b/c/archive_hmac_private.h new file mode 100644 --- /dev/null +++ b/c/archive_hmac_private.h @@ -0,0 +1,110 @@ +/*- +* Copyright (c) 2014 Michihiro NAKAJIMA +* All rights reserved. +* +* Redistribution and use in source and binary forms, with or without +* modification, are permitted provided that the following conditions +* are met: +* 1. Redistributions of source code must retain the above copyright +* notice, this list of conditions and the following disclaimer. +* 2. Redistributions in binary form must reproduce the above copyright +* notice, this list of conditions and the following disclaimer in the +* documentation and/or other materials provided with the distribution. +* +* THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR +* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES +* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. +* IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, +* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT +* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF +* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#ifndef ARCHIVE_HMAC_PRIVATE_H_INCLUDED +#define ARCHIVE_HMAC_PRIVATE_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif +/* + * On systems that do not support any recognized crypto libraries, + * the archive_hmac.c file is expected to define no usable symbols. + * + * But some compilers and linkers choke on empty object files, so + * define a public symbol that will always exist. This could + * be removed someday if this file gains another always-present + * symbol definition. + */ +int __libarchive_hmac_build_hack(void); + +#ifdef __APPLE__ +# include +# if MAC_OS_X_VERSION_MAX_ALLOWED >= 1060 +# define ARCHIVE_HMAC_USE_Apple_CommonCrypto +# endif +#endif + +#ifdef ARCHIVE_HMAC_USE_Apple_CommonCrypto +#include + +typedef CCHmacContext archive_hmac_sha1_ctx; + +#elif defined(_WIN32) && !defined(__CYGWIN__) && defined(HAVE_BCRYPT_H) +#include + +typedef struct { + BCRYPT_ALG_HANDLE hAlg; + BCRYPT_HASH_HANDLE hHash; + DWORD hash_len; + PBYTE hash; + +} archive_hmac_sha1_ctx; + +#elif defined(HAVE_LIBMBEDCRYPTO) && defined(HAVE_MBEDTLS_MD_H) +#include + +typedef mbedtls_md_context_t archive_hmac_sha1_ctx; + +#elif defined(HAVE_LIBNETTLE) && defined(HAVE_NETTLE_HMAC_H) +#include + +typedef struct hmac_sha1_ctx archive_hmac_sha1_ctx; + +#elif defined(HAVE_LIBCRYPTO) +#include "archive_openssl_hmac_private.h" + +typedef HMAC_CTX* archive_hmac_sha1_ctx; + +#else + +typedef int archive_hmac_sha1_ctx; + +#endif + + +/* HMAC */ +#define archive_hmac_sha1_init(ctx, key, key_len)\ + __archive_hmac.__hmac_sha1_init(ctx, key, key_len) +#define archive_hmac_sha1_update(ctx, data, data_len)\ + __archive_hmac.__hmac_sha1_update(ctx, data, data_len) +#define archive_hmac_sha1_final(ctx, out, out_len)\ + __archive_hmac.__hmac_sha1_final(ctx, out, out_len) +#define archive_hmac_sha1_cleanup(ctx)\ + __archive_hmac.__hmac_sha1_cleanup(ctx) + + +struct archive_hmac { + /* HMAC */ + int (*__hmac_sha1_init)(archive_hmac_sha1_ctx *, const uint8_t *, + size_t); + void (*__hmac_sha1_update)(archive_hmac_sha1_ctx *, const uint8_t *, + size_t); + void (*__hmac_sha1_final)(archive_hmac_sha1_ctx *, uint8_t *, size_t *); + void (*__hmac_sha1_cleanup)(archive_hmac_sha1_ctx *); +}; + +extern const struct archive_hmac __archive_hmac; +#endif /* ARCHIVE_HMAC_PRIVATE_H_INCLUDED */ diff --git a/c/archive_match.c b/c/archive_match.c new file mode 100644 --- /dev/null +++ b/c/archive_match.c @@ -0,0 +1,1875 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * Copyright (c) 2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_entry.h" +#include "archive_getdate.h" +#include "archive_pathmatch.h" +#include "archive_rb.h" +#include "archive_string.h" + +struct match { + struct match *next; + int matches; + struct archive_mstring pattern; +}; + +struct match_list { + struct match *first; + struct match **last; + int count; + int unmatched_count; + struct match *unmatched_next; + int unmatched_eof; +}; + +struct match_file { + struct archive_rb_node node; + struct match_file *next; + struct archive_mstring pathname; + int flag; + time_t mtime_sec; + long mtime_nsec; + time_t ctime_sec; + long ctime_nsec; +}; + +struct entry_list { + struct match_file *first; + struct match_file **last; + int count; +}; + +struct id_array { + size_t size;/* Allocated size */ + size_t count; + int64_t *ids; +}; + +#define PATTERN_IS_SET 1 +#define TIME_IS_SET 2 +#define ID_IS_SET 4 + +struct archive_match { + struct archive archive; + + /* exclusion/inclusion set flag. */ + int setflag; + + /* Recursively include directory content? */ + int recursive_include; + + /* + * Matching filename patterns. + */ + struct match_list exclusions; + struct match_list inclusions; + + /* + * Matching time stamps. + */ + time_t now; + int newer_mtime_filter; + time_t newer_mtime_sec; + long newer_mtime_nsec; + int newer_ctime_filter; + time_t newer_ctime_sec; + long newer_ctime_nsec; + int older_mtime_filter; + time_t older_mtime_sec; + long older_mtime_nsec; + int older_ctime_filter; + time_t older_ctime_sec; + long older_ctime_nsec; + /* + * Matching time stamps with its filename. + */ + struct archive_rb_tree exclusion_tree; + struct entry_list exclusion_entry_list; + + /* + * Matching file owners. + */ + struct id_array inclusion_uids; + struct id_array inclusion_gids; + struct match_list inclusion_unames; + struct match_list inclusion_gnames; +}; + +static int add_pattern_from_file(struct archive_match *, + struct match_list *, int, const void *, int); +static int add_entry(struct archive_match *, int, + struct archive_entry *); +static int add_owner_id(struct archive_match *, struct id_array *, + int64_t); +static int add_owner_name(struct archive_match *, struct match_list *, + int, const void *); +static int add_pattern_mbs(struct archive_match *, struct match_list *, + const char *); +static int add_pattern_wcs(struct archive_match *, struct match_list *, + const wchar_t *); +static int cmp_key_mbs(const struct archive_rb_node *, const void *); +static int cmp_key_wcs(const struct archive_rb_node *, const void *); +static int cmp_node_mbs(const struct archive_rb_node *, + const struct archive_rb_node *); +static int cmp_node_wcs(const struct archive_rb_node *, + const struct archive_rb_node *); +static void entry_list_add(struct entry_list *, struct match_file *); +static void entry_list_free(struct entry_list *); +static void entry_list_init(struct entry_list *); +static int error_nomem(struct archive_match *); +static void match_list_add(struct match_list *, struct match *); +static void match_list_free(struct match_list *); +static void match_list_init(struct match_list *); +static int match_list_unmatched_inclusions_next(struct archive_match *, + struct match_list *, int, const void **); +static int match_owner_id(struct id_array *, int64_t); +#if !defined(_WIN32) || defined(__CYGWIN__) +static int match_owner_name_mbs(struct archive_match *, + struct match_list *, const char *); +#else +static int match_owner_name_wcs(struct archive_match *, + struct match_list *, const wchar_t *); +#endif +static int match_path_exclusion(struct archive_match *, + struct match *, int, const void *); +static int match_path_inclusion(struct archive_match *, + struct match *, int, const void *); +static int owner_excluded(struct archive_match *, + struct archive_entry *); +static int path_excluded(struct archive_match *, int, const void *); +static int set_timefilter(struct archive_match *, int, time_t, long, + time_t, long); +static int set_timefilter_pathname_mbs(struct archive_match *, + int, const char *); +static int set_timefilter_pathname_wcs(struct archive_match *, + int, const wchar_t *); +static int set_timefilter_date(struct archive_match *, int, const char *); +static int set_timefilter_date_w(struct archive_match *, int, + const wchar_t *); +static int time_excluded(struct archive_match *, + struct archive_entry *); +static int validate_time_flag(struct archive *, int, const char *); + +#define get_date __archive_get_date + +static const struct archive_rb_tree_ops rb_ops_mbs = { + cmp_node_mbs, cmp_key_mbs +}; + +static const struct archive_rb_tree_ops rb_ops_wcs = { + cmp_node_wcs, cmp_key_wcs +}; + +/* + * The matching logic here needs to be re-thought. I started out to + * try to mimic gtar's matching logic, but it's not entirely + * consistent. In particular 'tar -t' and 'tar -x' interpret patterns + * on the command line as anchored, but --exclude doesn't. + */ + +static int +error_nomem(struct archive_match *a) +{ + archive_set_error(&(a->archive), ENOMEM, "No memory"); + a->archive.state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); +} + +/* + * Create an ARCHIVE_MATCH object. + */ +struct archive * +archive_match_new(void) +{ + struct archive_match *a; + + a = (struct archive_match *)calloc(1, sizeof(*a)); + if (a == NULL) + return (NULL); + a->archive.magic = ARCHIVE_MATCH_MAGIC; + a->archive.state = ARCHIVE_STATE_NEW; + a->recursive_include = 1; + match_list_init(&(a->inclusions)); + match_list_init(&(a->exclusions)); + __archive_rb_tree_init(&(a->exclusion_tree), &rb_ops_mbs); + entry_list_init(&(a->exclusion_entry_list)); + match_list_init(&(a->inclusion_unames)); + match_list_init(&(a->inclusion_gnames)); + time(&a->now); + return (&(a->archive)); +} + +/* + * Free an ARCHIVE_MATCH object. + */ +int +archive_match_free(struct archive *_a) +{ + struct archive_match *a; + + if (_a == NULL) + return (ARCHIVE_OK); + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_match_free"); + a = (struct archive_match *)_a; + match_list_free(&(a->inclusions)); + match_list_free(&(a->exclusions)); + entry_list_free(&(a->exclusion_entry_list)); + free(a->inclusion_uids.ids); + free(a->inclusion_gids.ids); + match_list_free(&(a->inclusion_unames)); + match_list_free(&(a->inclusion_gnames)); + free(a); + return (ARCHIVE_OK); +} + +/* + * Convenience function to perform all exclusion tests. + * + * Returns 1 if archive entry is excluded. + * Returns 0 if archive entry is not excluded. + * Returns <0 if something error happened. + */ +int +archive_match_excluded(struct archive *_a, struct archive_entry *entry) +{ + struct archive_match *a; + int r; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_excluded_ae"); + + a = (struct archive_match *)_a; + if (entry == NULL) { + archive_set_error(&(a->archive), EINVAL, "entry is NULL"); + return (ARCHIVE_FAILED); + } + + r = 0; + if (a->setflag & PATTERN_IS_SET) { +#if defined(_WIN32) && !defined(__CYGWIN__) + r = path_excluded(a, 0, archive_entry_pathname_w(entry)); +#else + r = path_excluded(a, 1, archive_entry_pathname(entry)); +#endif + if (r != 0) + return (r); + } + + if (a->setflag & TIME_IS_SET) { + r = time_excluded(a, entry); + if (r != 0) + return (r); + } + + if (a->setflag & ID_IS_SET) + r = owner_excluded(a, entry); + return (r); +} + +/* + * Utility functions to manage exclusion/inclusion patterns + */ + +int +archive_match_exclude_pattern(struct archive *_a, const char *pattern) +{ + struct archive_match *a; + int r; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_exclude_pattern"); + a = (struct archive_match *)_a; + + if (pattern == NULL || *pattern == '\0') { + archive_set_error(&(a->archive), EINVAL, "pattern is empty"); + return (ARCHIVE_FAILED); + } + if ((r = add_pattern_mbs(a, &(a->exclusions), pattern)) != ARCHIVE_OK) + return (r); + return (ARCHIVE_OK); +} + +int +archive_match_exclude_pattern_w(struct archive *_a, const wchar_t *pattern) +{ + struct archive_match *a; + int r; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_exclude_pattern_w"); + a = (struct archive_match *)_a; + + if (pattern == NULL || *pattern == L'\0') { + archive_set_error(&(a->archive), EINVAL, "pattern is empty"); + return (ARCHIVE_FAILED); + } + if ((r = add_pattern_wcs(a, &(a->exclusions), pattern)) != ARCHIVE_OK) + return (r); + return (ARCHIVE_OK); +} + +int +archive_match_exclude_pattern_from_file(struct archive *_a, + const char *pathname, int nullSeparator) +{ + struct archive_match *a; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_exclude_pattern_from_file"); + a = (struct archive_match *)_a; + + return add_pattern_from_file(a, &(a->exclusions), 1, pathname, + nullSeparator); +} + +int +archive_match_exclude_pattern_from_file_w(struct archive *_a, + const wchar_t *pathname, int nullSeparator) +{ + struct archive_match *a; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_exclude_pattern_from_file_w"); + a = (struct archive_match *)_a; + + return add_pattern_from_file(a, &(a->exclusions), 0, pathname, + nullSeparator); +} + +int +archive_match_include_pattern(struct archive *_a, const char *pattern) +{ + struct archive_match *a; + int r; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_include_pattern"); + a = (struct archive_match *)_a; + + if (pattern == NULL || *pattern == '\0') { + archive_set_error(&(a->archive), EINVAL, "pattern is empty"); + return (ARCHIVE_FAILED); + } + if ((r = add_pattern_mbs(a, &(a->inclusions), pattern)) != ARCHIVE_OK) + return (r); + return (ARCHIVE_OK); +} + +int +archive_match_include_pattern_w(struct archive *_a, const wchar_t *pattern) +{ + struct archive_match *a; + int r; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_include_pattern_w"); + a = (struct archive_match *)_a; + + if (pattern == NULL || *pattern == L'\0') { + archive_set_error(&(a->archive), EINVAL, "pattern is empty"); + return (ARCHIVE_FAILED); + } + if ((r = add_pattern_wcs(a, &(a->inclusions), pattern)) != ARCHIVE_OK) + return (r); + return (ARCHIVE_OK); +} + +int +archive_match_include_pattern_from_file(struct archive *_a, + const char *pathname, int nullSeparator) +{ + struct archive_match *a; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_include_pattern_from_file"); + a = (struct archive_match *)_a; + + return add_pattern_from_file(a, &(a->inclusions), 1, pathname, + nullSeparator); +} + +int +archive_match_include_pattern_from_file_w(struct archive *_a, + const wchar_t *pathname, int nullSeparator) +{ + struct archive_match *a; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_include_pattern_from_file_w"); + a = (struct archive_match *)_a; + + return add_pattern_from_file(a, &(a->inclusions), 0, pathname, + nullSeparator); +} + +/* + * Test functions for pathname patterns. + * + * Returns 1 if archive entry is excluded. + * Returns 0 if archive entry is not excluded. + * Returns <0 if something error happened. + */ +int +archive_match_path_excluded(struct archive *_a, + struct archive_entry *entry) +{ + struct archive_match *a; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_path_excluded"); + + a = (struct archive_match *)_a; + if (entry == NULL) { + archive_set_error(&(a->archive), EINVAL, "entry is NULL"); + return (ARCHIVE_FAILED); + } + + /* If we don't have exclusion/inclusion pattern set at all, + * the entry is always not excluded. */ + if ((a->setflag & PATTERN_IS_SET) == 0) + return (0); +#if defined(_WIN32) && !defined(__CYGWIN__) + return (path_excluded(a, 0, archive_entry_pathname_w(entry))); +#else + return (path_excluded(a, 1, archive_entry_pathname(entry))); +#endif +} + +/* + * When recursive inclusion of directory content is enabled, + * an inclusion pattern that matches a directory will also + * include everything beneath that directory. Enabled by default. + * + * For compatibility with GNU tar, exclusion patterns always + * match if a subset of the full patch matches (i.e., they are + * are not rooted at the beginning of the path) and thus there + * is no corresponding non-recursive exclusion mode. + */ +int +archive_match_set_inclusion_recursion(struct archive *_a, int enabled) +{ + struct archive_match *a; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_set_inclusion_recursion"); + a = (struct archive_match *)_a; + a->recursive_include = enabled; + return (ARCHIVE_OK); +} + +/* + * Utility functions to get statistic information for inclusion patterns. + */ +int +archive_match_path_unmatched_inclusions(struct archive *_a) +{ + struct archive_match *a; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_unmatched_inclusions"); + a = (struct archive_match *)_a; + + return (a->inclusions.unmatched_count); +} + +int +archive_match_path_unmatched_inclusions_next(struct archive *_a, + const char **_p) +{ + struct archive_match *a; + const void *v; + int r; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_unmatched_inclusions_next"); + a = (struct archive_match *)_a; + + r = match_list_unmatched_inclusions_next(a, &(a->inclusions), 1, &v); + *_p = (const char *)v; + return (r); +} + +int +archive_match_path_unmatched_inclusions_next_w(struct archive *_a, + const wchar_t **_p) +{ + struct archive_match *a; + const void *v; + int r; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_unmatched_inclusions_next_w"); + a = (struct archive_match *)_a; + + r = match_list_unmatched_inclusions_next(a, &(a->inclusions), 0, &v); + *_p = (const wchar_t *)v; + return (r); +} + +/* + * Add inclusion/exclusion patterns. + */ +static int +add_pattern_mbs(struct archive_match *a, struct match_list *list, + const char *pattern) +{ + struct match *match; + size_t len; + + match = calloc(1, sizeof(*match)); + if (match == NULL) + return (error_nomem(a)); + /* Both "foo/" and "foo" should match "foo/bar". */ + len = strlen(pattern); + if (len && pattern[len - 1] == '/') + --len; + archive_mstring_copy_mbs_len(&(match->pattern), pattern, len); + match_list_add(list, match); + a->setflag |= PATTERN_IS_SET; + return (ARCHIVE_OK); +} + +static int +add_pattern_wcs(struct archive_match *a, struct match_list *list, + const wchar_t *pattern) +{ + struct match *match; + size_t len; + + match = calloc(1, sizeof(*match)); + if (match == NULL) + return (error_nomem(a)); + /* Both "foo/" and "foo" should match "foo/bar". */ + len = wcslen(pattern); + if (len && pattern[len - 1] == L'/') + --len; + archive_mstring_copy_wcs_len(&(match->pattern), pattern, len); + match_list_add(list, match); + a->setflag |= PATTERN_IS_SET; + return (ARCHIVE_OK); +} + +static int +add_pattern_from_file(struct archive_match *a, struct match_list *mlist, + int mbs, const void *pathname, int nullSeparator) +{ + struct archive *ar; + struct archive_entry *ae; + struct archive_string as; + const void *buff; + size_t size; + int64_t offset; + int r; + + ar = archive_read_new(); + if (ar == NULL) { + archive_set_error(&(a->archive), ENOMEM, "No memory"); + return (ARCHIVE_FATAL); + } + r = archive_read_support_format_raw(ar); + r = archive_read_support_format_empty(ar); + if (r != ARCHIVE_OK) { + archive_copy_error(&(a->archive), ar); + archive_read_free(ar); + return (r); + } + if (mbs) + r = archive_read_open_filename(ar, pathname, 512*20); + else + r = archive_read_open_filename_w(ar, pathname, 512*20); + if (r != ARCHIVE_OK) { + archive_copy_error(&(a->archive), ar); + archive_read_free(ar); + return (r); + } + r = archive_read_next_header(ar, &ae); + if (r != ARCHIVE_OK) { + archive_read_free(ar); + if (r == ARCHIVE_EOF) { + return (ARCHIVE_OK); + } else { + archive_copy_error(&(a->archive), ar); + return (r); + } + } + + archive_string_init(&as); + + while ((r = archive_read_data_block(ar, &buff, &size, &offset)) + == ARCHIVE_OK) { + const char *b = (const char *)buff; + + while (size) { + const char *s = (const char *)b; + size_t length = 0; + int found_separator = 0; + + while (length < size) { + if (nullSeparator) { + if (*b == '\0') { + found_separator = 1; + break; + } + } else { + if (*b == 0x0d || *b == 0x0a) { + found_separator = 1; + break; + } + } + b++; + length++; + } + if (!found_separator) { + archive_strncat(&as, s, length); + /* Read next data block. */ + break; + } + b++; + size -= length + 1; + archive_strncat(&as, s, length); + + /* If the line is not empty, add the pattern. */ + if (archive_strlen(&as) > 0) { + /* Add pattern. */ + r = add_pattern_mbs(a, mlist, as.s); + if (r != ARCHIVE_OK) { + archive_read_free(ar); + archive_string_free(&as); + return (r); + } + archive_string_empty(&as); + } + } + } + + /* If an error occurred, report it immediately. */ + if (r < ARCHIVE_OK) { + archive_copy_error(&(a->archive), ar); + archive_read_free(ar); + archive_string_free(&as); + return (r); + } + + /* If the line is not empty, add the pattern. */ + if (r == ARCHIVE_EOF && archive_strlen(&as) > 0) { + /* Add pattern. */ + r = add_pattern_mbs(a, mlist, as.s); + if (r != ARCHIVE_OK) { + archive_read_free(ar); + archive_string_free(&as); + return (r); + } + } + archive_read_free(ar); + archive_string_free(&as); + return (ARCHIVE_OK); +} + +/* + * Test if pathname is excluded by inclusion/exclusion patterns. + */ +static int +path_excluded(struct archive_match *a, int mbs, const void *pathname) +{ + struct match *match; + struct match *matched; + int r; + + if (a == NULL) + return (0); + + /* Mark off any unmatched inclusions. */ + /* In particular, if a filename does appear in the archive and + * is explicitly included and excluded, then we don't report + * it as missing even though we don't extract it. + */ + matched = NULL; + for (match = a->inclusions.first; match != NULL; + match = match->next){ + if (match->matches == 0 && + (r = match_path_inclusion(a, match, mbs, pathname)) != 0) { + if (r < 0) + return (r); + a->inclusions.unmatched_count--; + match->matches++; + matched = match; + } + } + + /* Exclusions take priority */ + for (match = a->exclusions.first; match != NULL; + match = match->next){ + r = match_path_exclusion(a, match, mbs, pathname); + if (r) + return (r); + } + + /* It's not excluded and we found an inclusion above, so it's + * included. */ + if (matched != NULL) + return (0); + + + /* We didn't find an unmatched inclusion, check the remaining ones. */ + for (match = a->inclusions.first; match != NULL; + match = match->next){ + /* We looked at previously-unmatched inclusions already. */ + if (match->matches > 0 && + (r = match_path_inclusion(a, match, mbs, pathname)) != 0) { + if (r < 0) + return (r); + match->matches++; + return (0); + } + } + + /* If there were inclusions, default is to exclude. */ + if (a->inclusions.first != NULL) + return (1); + + /* No explicit inclusions, default is to match. */ + return (0); +} + +/* + * This is a little odd, but it matches the default behavior of + * gtar. In particular, 'a*b' will match 'foo/a1111/222b/bar' + * + */ +static int +match_path_exclusion(struct archive_match *a, struct match *m, + int mbs, const void *pn) +{ + int flag = PATHMATCH_NO_ANCHOR_START | PATHMATCH_NO_ANCHOR_END; + int r; + + if (mbs) { + const char *p; + r = archive_mstring_get_mbs(&(a->archive), &(m->pattern), &p); + if (r == 0) + return (archive_pathmatch(p, (const char *)pn, flag)); + } else { + const wchar_t *p; + r = archive_mstring_get_wcs(&(a->archive), &(m->pattern), &p); + if (r == 0) + return (archive_pathmatch_w(p, (const wchar_t *)pn, + flag)); + } + if (errno == ENOMEM) + return (error_nomem(a)); + return (0); +} + +/* + * Again, mimic gtar: inclusions are always anchored (have to match + * the beginning of the path) even though exclusions are not anchored. + */ +static int +match_path_inclusion(struct archive_match *a, struct match *m, + int mbs, const void *pn) +{ + /* Recursive operation requires only a prefix match. */ + int flag = a->recursive_include ? + PATHMATCH_NO_ANCHOR_END : + 0; + int r; + + if (mbs) { + const char *p; + r = archive_mstring_get_mbs(&(a->archive), &(m->pattern), &p); + if (r == 0) + return (archive_pathmatch(p, (const char *)pn, flag)); + } else { + const wchar_t *p; + r = archive_mstring_get_wcs(&(a->archive), &(m->pattern), &p); + if (r == 0) + return (archive_pathmatch_w(p, (const wchar_t *)pn, + flag)); + } + if (errno == ENOMEM) + return (error_nomem(a)); + return (0); +} + +static void +match_list_init(struct match_list *list) +{ + list->first = NULL; + list->last = &(list->first); + list->count = 0; +} + +static void +match_list_free(struct match_list *list) +{ + struct match *p, *q; + + for (p = list->first; p != NULL; ) { + q = p; + p = p->next; + archive_mstring_clean(&(q->pattern)); + free(q); + } +} + +static void +match_list_add(struct match_list *list, struct match *m) +{ + *list->last = m; + list->last = &(m->next); + list->count++; + list->unmatched_count++; +} + +static int +match_list_unmatched_inclusions_next(struct archive_match *a, + struct match_list *list, int mbs, const void **vp) +{ + struct match *m; + + *vp = NULL; + if (list->unmatched_eof) { + list->unmatched_eof = 0; + return (ARCHIVE_EOF); + } + if (list->unmatched_next == NULL) { + if (list->unmatched_count == 0) + return (ARCHIVE_EOF); + list->unmatched_next = list->first; + } + + for (m = list->unmatched_next; m != NULL; m = m->next) { + int r; + + if (m->matches) + continue; + if (mbs) { + const char *p; + r = archive_mstring_get_mbs(&(a->archive), + &(m->pattern), &p); + if (r < 0 && errno == ENOMEM) + return (error_nomem(a)); + if (p == NULL) + p = ""; + *vp = p; + } else { + const wchar_t *p; + r = archive_mstring_get_wcs(&(a->archive), + &(m->pattern), &p); + if (r < 0 && errno == ENOMEM) + return (error_nomem(a)); + if (p == NULL) + p = L""; + *vp = p; + } + list->unmatched_next = m->next; + if (list->unmatched_next == NULL) + /* To return EOF next time. */ + list->unmatched_eof = 1; + return (ARCHIVE_OK); + } + list->unmatched_next = NULL; + return (ARCHIVE_EOF); +} + +/* + * Utility functions to manage inclusion timestamps. + */ +int +archive_match_include_time(struct archive *_a, int flag, time_t sec, + long nsec) +{ + int r; + + r = validate_time_flag(_a, flag, "archive_match_include_time"); + if (r != ARCHIVE_OK) + return (r); + return set_timefilter((struct archive_match *)_a, flag, + sec, nsec, sec, nsec); +} + +int +archive_match_include_date(struct archive *_a, int flag, + const char *datestr) +{ + int r; + + r = validate_time_flag(_a, flag, "archive_match_include_date"); + if (r != ARCHIVE_OK) + return (r); + return set_timefilter_date((struct archive_match *)_a, flag, datestr); +} + +int +archive_match_include_date_w(struct archive *_a, int flag, + const wchar_t *datestr) +{ + int r; + + r = validate_time_flag(_a, flag, "archive_match_include_date_w"); + if (r != ARCHIVE_OK) + return (r); + + return set_timefilter_date_w((struct archive_match *)_a, flag, datestr); +} + +int +archive_match_include_file_time(struct archive *_a, int flag, + const char *pathname) +{ + int r; + + r = validate_time_flag(_a, flag, "archive_match_include_file_time"); + if (r != ARCHIVE_OK) + return (r); + return set_timefilter_pathname_mbs((struct archive_match *)_a, + flag, pathname); +} + +int +archive_match_include_file_time_w(struct archive *_a, int flag, + const wchar_t *pathname) +{ + int r; + + r = validate_time_flag(_a, flag, "archive_match_include_file_time_w"); + if (r != ARCHIVE_OK) + return (r); + return set_timefilter_pathname_wcs((struct archive_match *)_a, + flag, pathname); +} + +int +archive_match_exclude_entry(struct archive *_a, int flag, + struct archive_entry *entry) +{ + struct archive_match *a; + int r; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_time_include_entry"); + a = (struct archive_match *)_a; + + if (entry == NULL) { + archive_set_error(&(a->archive), EINVAL, "entry is NULL"); + return (ARCHIVE_FAILED); + } + r = validate_time_flag(_a, flag, "archive_match_exclude_entry"); + if (r != ARCHIVE_OK) + return (r); + return (add_entry(a, flag, entry)); +} + +/* + * Test function for time stamps. + * + * Returns 1 if archive entry is excluded. + * Returns 0 if archive entry is not excluded. + * Returns <0 if something error happened. + */ +int +archive_match_time_excluded(struct archive *_a, + struct archive_entry *entry) +{ + struct archive_match *a; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_time_excluded_ae"); + + a = (struct archive_match *)_a; + if (entry == NULL) { + archive_set_error(&(a->archive), EINVAL, "entry is NULL"); + return (ARCHIVE_FAILED); + } + + /* If we don't have inclusion time set at all, the entry is always + * not excluded. */ + if ((a->setflag & TIME_IS_SET) == 0) + return (0); + return (time_excluded(a, entry)); +} + +static int +validate_time_flag(struct archive *_a, int flag, const char *_fn) +{ + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, _fn); + + /* Check a type of time. */ + if (flag & + ((~(ARCHIVE_MATCH_MTIME | ARCHIVE_MATCH_CTIME)) & 0xff00)) { + archive_set_error(_a, EINVAL, "Invalid time flag"); + return (ARCHIVE_FAILED); + } + if ((flag & (ARCHIVE_MATCH_MTIME | ARCHIVE_MATCH_CTIME)) == 0) { + archive_set_error(_a, EINVAL, "No time flag"); + return (ARCHIVE_FAILED); + } + + /* Check a type of comparison. */ + if (flag & + ((~(ARCHIVE_MATCH_NEWER | ARCHIVE_MATCH_OLDER + | ARCHIVE_MATCH_EQUAL)) & 0x00ff)) { + archive_set_error(_a, EINVAL, "Invalid comparison flag"); + return (ARCHIVE_FAILED); + } + if ((flag & (ARCHIVE_MATCH_NEWER | ARCHIVE_MATCH_OLDER + | ARCHIVE_MATCH_EQUAL)) == 0) { + archive_set_error(_a, EINVAL, "No comparison flag"); + return (ARCHIVE_FAILED); + } + + return (ARCHIVE_OK); +} + +#define JUST_EQUAL(t) (((t) & (ARCHIVE_MATCH_EQUAL |\ + ARCHIVE_MATCH_NEWER | ARCHIVE_MATCH_OLDER)) == ARCHIVE_MATCH_EQUAL) +static int +set_timefilter(struct archive_match *a, int timetype, + time_t mtime_sec, long mtime_nsec, time_t ctime_sec, long ctime_nsec) +{ + if (timetype & ARCHIVE_MATCH_MTIME) { + if ((timetype & ARCHIVE_MATCH_NEWER) || JUST_EQUAL(timetype)) { + a->newer_mtime_filter = timetype; + a->newer_mtime_sec = mtime_sec; + a->newer_mtime_nsec = mtime_nsec; + a->setflag |= TIME_IS_SET; + } + if ((timetype & ARCHIVE_MATCH_OLDER) || JUST_EQUAL(timetype)) { + a->older_mtime_filter = timetype; + a->older_mtime_sec = mtime_sec; + a->older_mtime_nsec = mtime_nsec; + a->setflag |= TIME_IS_SET; + } + } + if (timetype & ARCHIVE_MATCH_CTIME) { + if ((timetype & ARCHIVE_MATCH_NEWER) || JUST_EQUAL(timetype)) { + a->newer_ctime_filter = timetype; + a->newer_ctime_sec = ctime_sec; + a->newer_ctime_nsec = ctime_nsec; + a->setflag |= TIME_IS_SET; + } + if ((timetype & ARCHIVE_MATCH_OLDER) || JUST_EQUAL(timetype)) { + a->older_ctime_filter = timetype; + a->older_ctime_sec = ctime_sec; + a->older_ctime_nsec = ctime_nsec; + a->setflag |= TIME_IS_SET; + } + } + return (ARCHIVE_OK); +} + +static int +set_timefilter_date(struct archive_match *a, int timetype, const char *datestr) +{ + time_t t; + + if (datestr == NULL || *datestr == '\0') { + archive_set_error(&(a->archive), EINVAL, "date is empty"); + return (ARCHIVE_FAILED); + } + t = get_date(a->now, datestr); + if (t == (time_t)-1) { + archive_set_error(&(a->archive), EINVAL, "invalid date string"); + return (ARCHIVE_FAILED); + } + return set_timefilter(a, timetype, t, 0, t, 0); +} + +static int +set_timefilter_date_w(struct archive_match *a, int timetype, + const wchar_t *datestr) +{ + struct archive_string as; + time_t t; + + if (datestr == NULL || *datestr == L'\0') { + archive_set_error(&(a->archive), EINVAL, "date is empty"); + return (ARCHIVE_FAILED); + } + + archive_string_init(&as); + if (archive_string_append_from_wcs(&as, datestr, wcslen(datestr)) < 0) { + archive_string_free(&as); + if (errno == ENOMEM) + return (error_nomem(a)); + archive_set_error(&(a->archive), -1, + "Failed to convert WCS to MBS"); + return (ARCHIVE_FAILED); + } + t = get_date(a->now, as.s); + archive_string_free(&as); + if (t == (time_t)-1) { + archive_set_error(&(a->archive), EINVAL, "invalid date string"); + return (ARCHIVE_FAILED); + } + return set_timefilter(a, timetype, t, 0, t, 0); +} + +#if defined(_WIN32) && !defined(__CYGWIN__) +#define EPOC_TIME ARCHIVE_LITERAL_ULL(116444736000000000) +static int +set_timefilter_find_data(struct archive_match *a, int timetype, + DWORD ftLastWriteTime_dwHighDateTime, DWORD ftLastWriteTime_dwLowDateTime, + DWORD ftCreationTime_dwHighDateTime, DWORD ftCreationTime_dwLowDateTime) +{ + ULARGE_INTEGER utc; + time_t ctime_sec, mtime_sec; + long ctime_ns, mtime_ns; + + utc.HighPart = ftCreationTime_dwHighDateTime; + utc.LowPart = ftCreationTime_dwLowDateTime; + if (utc.QuadPart >= EPOC_TIME) { + utc.QuadPart -= EPOC_TIME; + ctime_sec = (time_t)(utc.QuadPart / 10000000); + ctime_ns = (long)(utc.QuadPart % 10000000) * 100; + } else { + ctime_sec = 0; + ctime_ns = 0; + } + utc.HighPart = ftLastWriteTime_dwHighDateTime; + utc.LowPart = ftLastWriteTime_dwLowDateTime; + if (utc.QuadPart >= EPOC_TIME) { + utc.QuadPart -= EPOC_TIME; + mtime_sec = (time_t)(utc.QuadPart / 10000000); + mtime_ns = (long)(utc.QuadPart % 10000000) * 100; + } else { + mtime_sec = 0; + mtime_ns = 0; + } + return set_timefilter(a, timetype, + mtime_sec, mtime_ns, ctime_sec, ctime_ns); +} + +static int +set_timefilter_pathname_mbs(struct archive_match *a, int timetype, + const char *path) +{ + /* NOTE: stat() on Windows cannot handle nano seconds. */ + HANDLE h; + WIN32_FIND_DATAA d; + + if (path == NULL || *path == '\0') { + archive_set_error(&(a->archive), EINVAL, "pathname is empty"); + return (ARCHIVE_FAILED); + } + h = FindFirstFileA(path, &d); + if (h == INVALID_HANDLE_VALUE) { + la_dosmaperr(GetLastError()); + archive_set_error(&(a->archive), errno, + "Failed to FindFirstFileA"); + return (ARCHIVE_FAILED); + } + FindClose(h); + return set_timefilter_find_data(a, timetype, + d.ftLastWriteTime.dwHighDateTime, d.ftLastWriteTime.dwLowDateTime, + d.ftCreationTime.dwHighDateTime, d.ftCreationTime.dwLowDateTime); +} + +static int +set_timefilter_pathname_wcs(struct archive_match *a, int timetype, + const wchar_t *path) +{ + HANDLE h; + WIN32_FIND_DATAW d; + + if (path == NULL || *path == L'\0') { + archive_set_error(&(a->archive), EINVAL, "pathname is empty"); + return (ARCHIVE_FAILED); + } + h = FindFirstFileW(path, &d); + if (h == INVALID_HANDLE_VALUE) { + la_dosmaperr(GetLastError()); + archive_set_error(&(a->archive), errno, + "Failed to FindFirstFile"); + return (ARCHIVE_FAILED); + } + FindClose(h); + return set_timefilter_find_data(a, timetype, + d.ftLastWriteTime.dwHighDateTime, d.ftLastWriteTime.dwLowDateTime, + d.ftCreationTime.dwHighDateTime, d.ftCreationTime.dwLowDateTime); +} + +#else /* _WIN32 && !__CYGWIN__ */ + +static int +set_timefilter_stat(struct archive_match *a, int timetype, struct stat *st) +{ + struct archive_entry *ae; + time_t ctime_sec, mtime_sec; + long ctime_ns, mtime_ns; + + ae = archive_entry_new(); + if (ae == NULL) + return (error_nomem(a)); + archive_entry_copy_stat(ae, st); + ctime_sec = archive_entry_ctime(ae); + ctime_ns = archive_entry_ctime_nsec(ae); + mtime_sec = archive_entry_mtime(ae); + mtime_ns = archive_entry_mtime_nsec(ae); + archive_entry_free(ae); + return set_timefilter(a, timetype, mtime_sec, mtime_ns, + ctime_sec, ctime_ns); +} + +static int +set_timefilter_pathname_mbs(struct archive_match *a, int timetype, + const char *path) +{ + struct stat st; + + if (path == NULL || *path == '\0') { + archive_set_error(&(a->archive), EINVAL, "pathname is empty"); + return (ARCHIVE_FAILED); + } + if (la_stat(path, &st) != 0) { + archive_set_error(&(a->archive), errno, "Failed to stat()"); + return (ARCHIVE_FAILED); + } + return (set_timefilter_stat(a, timetype, &st)); +} + +static int +set_timefilter_pathname_wcs(struct archive_match *a, int timetype, + const wchar_t *path) +{ + struct archive_string as; + int r; + + if (path == NULL || *path == L'\0') { + archive_set_error(&(a->archive), EINVAL, "pathname is empty"); + return (ARCHIVE_FAILED); + } + + /* Convert WCS filename to MBS filename. */ + archive_string_init(&as); + if (archive_string_append_from_wcs(&as, path, wcslen(path)) < 0) { + archive_string_free(&as); + if (errno == ENOMEM) + return (error_nomem(a)); + archive_set_error(&(a->archive), -1, + "Failed to convert WCS to MBS"); + return (ARCHIVE_FAILED); + } + + r = set_timefilter_pathname_mbs(a, timetype, as.s); + archive_string_free(&as); + + return (r); +} +#endif /* _WIN32 && !__CYGWIN__ */ + +/* + * Call back functions for archive_rb. + */ +static int +cmp_node_mbs(const struct archive_rb_node *n1, + const struct archive_rb_node *n2) +{ + struct match_file *f1 = (struct match_file *)(uintptr_t)n1; + struct match_file *f2 = (struct match_file *)(uintptr_t)n2; + const char *p1, *p2; + + archive_mstring_get_mbs(NULL, &(f1->pathname), &p1); + archive_mstring_get_mbs(NULL, &(f2->pathname), &p2); + if (p1 == NULL) + return (1); + if (p2 == NULL) + return (-1); + return (strcmp(p1, p2)); +} + +static int +cmp_key_mbs(const struct archive_rb_node *n, const void *key) +{ + struct match_file *f = (struct match_file *)(uintptr_t)n; + const char *p; + + archive_mstring_get_mbs(NULL, &(f->pathname), &p); + if (p == NULL) + return (-1); + return (strcmp(p, (const char *)key)); +} + +static int +cmp_node_wcs(const struct archive_rb_node *n1, + const struct archive_rb_node *n2) +{ + struct match_file *f1 = (struct match_file *)(uintptr_t)n1; + struct match_file *f2 = (struct match_file *)(uintptr_t)n2; + const wchar_t *p1, *p2; + + archive_mstring_get_wcs(NULL, &(f1->pathname), &p1); + archive_mstring_get_wcs(NULL, &(f2->pathname), &p2); + if (p1 == NULL) + return (1); + if (p2 == NULL) + return (-1); + return (wcscmp(p1, p2)); +} + +static int +cmp_key_wcs(const struct archive_rb_node *n, const void *key) +{ + struct match_file *f = (struct match_file *)(uintptr_t)n; + const wchar_t *p; + + archive_mstring_get_wcs(NULL, &(f->pathname), &p); + if (p == NULL) + return (-1); + return (wcscmp(p, (const wchar_t *)key)); +} + +static void +entry_list_init(struct entry_list *list) +{ + list->first = NULL; + list->last = &(list->first); + list->count = 0; +} + +static void +entry_list_free(struct entry_list *list) +{ + struct match_file *p, *q; + + for (p = list->first; p != NULL; ) { + q = p; + p = p->next; + archive_mstring_clean(&(q->pathname)); + free(q); + } +} + +static void +entry_list_add(struct entry_list *list, struct match_file *file) +{ + *list->last = file; + list->last = &(file->next); + list->count++; +} + +static int +add_entry(struct archive_match *a, int flag, + struct archive_entry *entry) +{ + struct match_file *f; + const void *pathname; + int r; + + f = calloc(1, sizeof(*f)); + if (f == NULL) + return (error_nomem(a)); + +#if defined(_WIN32) && !defined(__CYGWIN__) + pathname = archive_entry_pathname_w(entry); + if (pathname == NULL) { + free(f); + archive_set_error(&(a->archive), EINVAL, "pathname is NULL"); + return (ARCHIVE_FAILED); + } + archive_mstring_copy_wcs(&(f->pathname), pathname); + a->exclusion_tree.rbt_ops = &rb_ops_wcs; +#else + (void)rb_ops_wcs; + pathname = archive_entry_pathname(entry); + if (pathname == NULL) { + free(f); + archive_set_error(&(a->archive), EINVAL, "pathname is NULL"); + return (ARCHIVE_FAILED); + } + archive_mstring_copy_mbs(&(f->pathname), pathname); + a->exclusion_tree.rbt_ops = &rb_ops_mbs; +#endif + f->flag = flag; + f->mtime_sec = archive_entry_mtime(entry); + f->mtime_nsec = archive_entry_mtime_nsec(entry); + f->ctime_sec = archive_entry_ctime(entry); + f->ctime_nsec = archive_entry_ctime_nsec(entry); + r = __archive_rb_tree_insert_node(&(a->exclusion_tree), &(f->node)); + if (!r) { + struct match_file *f2; + + /* Get the duplicated file. */ + f2 = (struct match_file *)__archive_rb_tree_find_node( + &(a->exclusion_tree), pathname); + + /* + * We always overwrite comparison condition. + * If you do not want to overwrite it, you should not + * call archive_match_exclude_entry(). We cannot know + * what behavior you really expect since overwriting + * condition might be different with the flag. + */ + if (f2 != NULL) { + f2->flag = f->flag; + f2->mtime_sec = f->mtime_sec; + f2->mtime_nsec = f->mtime_nsec; + f2->ctime_sec = f->ctime_sec; + f2->ctime_nsec = f->ctime_nsec; + } + /* Release the duplicated file. */ + archive_mstring_clean(&(f->pathname)); + free(f); + return (ARCHIVE_OK); + } + entry_list_add(&(a->exclusion_entry_list), f); + a->setflag |= TIME_IS_SET; + return (ARCHIVE_OK); +} + +/* + * Test if entry is excluded by its timestamp. + */ +static int +time_excluded(struct archive_match *a, struct archive_entry *entry) +{ + struct match_file *f; + const void *pathname; + time_t sec; + long nsec; + + /* + * If this file/dir is excluded by a time comparison, skip it. + */ + if (a->newer_ctime_filter) { + /* If ctime is not set, use mtime instead. */ + if (archive_entry_ctime_is_set(entry)) + sec = archive_entry_ctime(entry); + else + sec = archive_entry_mtime(entry); + if (sec < a->newer_ctime_sec) + return (1); /* Too old, skip it. */ + if (sec == a->newer_ctime_sec) { + if (archive_entry_ctime_is_set(entry)) + nsec = archive_entry_ctime_nsec(entry); + else + nsec = archive_entry_mtime_nsec(entry); + if (nsec < a->newer_ctime_nsec) + return (1); /* Too old, skip it. */ + if (nsec == a->newer_ctime_nsec && + (a->newer_ctime_filter & ARCHIVE_MATCH_EQUAL) + == 0) + return (1); /* Equal, skip it. */ + } + } + if (a->older_ctime_filter) { + /* If ctime is not set, use mtime instead. */ + if (archive_entry_ctime_is_set(entry)) + sec = archive_entry_ctime(entry); + else + sec = archive_entry_mtime(entry); + if (sec > a->older_ctime_sec) + return (1); /* Too new, skip it. */ + if (sec == a->older_ctime_sec) { + if (archive_entry_ctime_is_set(entry)) + nsec = archive_entry_ctime_nsec(entry); + else + nsec = archive_entry_mtime_nsec(entry); + if (nsec > a->older_ctime_nsec) + return (1); /* Too new, skip it. */ + if (nsec == a->older_ctime_nsec && + (a->older_ctime_filter & ARCHIVE_MATCH_EQUAL) + == 0) + return (1); /* Equal, skip it. */ + } + } + if (a->newer_mtime_filter) { + sec = archive_entry_mtime(entry); + if (sec < a->newer_mtime_sec) + return (1); /* Too old, skip it. */ + if (sec == a->newer_mtime_sec) { + nsec = archive_entry_mtime_nsec(entry); + if (nsec < a->newer_mtime_nsec) + return (1); /* Too old, skip it. */ + if (nsec == a->newer_mtime_nsec && + (a->newer_mtime_filter & ARCHIVE_MATCH_EQUAL) + == 0) + return (1); /* Equal, skip it. */ + } + } + if (a->older_mtime_filter) { + sec = archive_entry_mtime(entry); + if (sec > a->older_mtime_sec) + return (1); /* Too new, skip it. */ + nsec = archive_entry_mtime_nsec(entry); + if (sec == a->older_mtime_sec) { + if (nsec > a->older_mtime_nsec) + return (1); /* Too new, skip it. */ + if (nsec == a->older_mtime_nsec && + (a->older_mtime_filter & ARCHIVE_MATCH_EQUAL) + == 0) + return (1); /* Equal, skip it. */ + } + } + + /* If there is no exclusion list, include the file. */ + if (a->exclusion_entry_list.count == 0) + return (0); + +#if defined(_WIN32) && !defined(__CYGWIN__) + pathname = archive_entry_pathname_w(entry); + a->exclusion_tree.rbt_ops = &rb_ops_wcs; +#else + (void)rb_ops_wcs; + pathname = archive_entry_pathname(entry); + a->exclusion_tree.rbt_ops = &rb_ops_mbs; +#endif + if (pathname == NULL) + return (0); + + f = (struct match_file *)__archive_rb_tree_find_node( + &(a->exclusion_tree), pathname); + /* If the file wasn't rejected, include it. */ + if (f == NULL) + return (0); + + if (f->flag & ARCHIVE_MATCH_CTIME) { + sec = archive_entry_ctime(entry); + if (f->ctime_sec > sec) { + if (f->flag & ARCHIVE_MATCH_OLDER) + return (1); + } else if (f->ctime_sec < sec) { + if (f->flag & ARCHIVE_MATCH_NEWER) + return (1); + } else { + nsec = archive_entry_ctime_nsec(entry); + if (f->ctime_nsec > nsec) { + if (f->flag & ARCHIVE_MATCH_OLDER) + return (1); + } else if (f->ctime_nsec < nsec) { + if (f->flag & ARCHIVE_MATCH_NEWER) + return (1); + } else if (f->flag & ARCHIVE_MATCH_EQUAL) + return (1); + } + } + if (f->flag & ARCHIVE_MATCH_MTIME) { + sec = archive_entry_mtime(entry); + if (f->mtime_sec > sec) { + if (f->flag & ARCHIVE_MATCH_OLDER) + return (1); + } else if (f->mtime_sec < sec) { + if (f->flag & ARCHIVE_MATCH_NEWER) + return (1); + } else { + nsec = archive_entry_mtime_nsec(entry); + if (f->mtime_nsec > nsec) { + if (f->flag & ARCHIVE_MATCH_OLDER) + return (1); + } else if (f->mtime_nsec < nsec) { + if (f->flag & ARCHIVE_MATCH_NEWER) + return (1); + } else if (f->flag & ARCHIVE_MATCH_EQUAL) + return (1); + } + } + return (0); +} + +/* + * Utility functions to manage inclusion owners + */ + +int +archive_match_include_uid(struct archive *_a, la_int64_t uid) +{ + struct archive_match *a; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_include_uid"); + a = (struct archive_match *)_a; + return (add_owner_id(a, &(a->inclusion_uids), uid)); +} + +int +archive_match_include_gid(struct archive *_a, la_int64_t gid) +{ + struct archive_match *a; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_include_gid"); + a = (struct archive_match *)_a; + return (add_owner_id(a, &(a->inclusion_gids), gid)); +} + +int +archive_match_include_uname(struct archive *_a, const char *uname) +{ + struct archive_match *a; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_include_uname"); + a = (struct archive_match *)_a; + return (add_owner_name(a, &(a->inclusion_unames), 1, uname)); +} + +int +archive_match_include_uname_w(struct archive *_a, const wchar_t *uname) +{ + struct archive_match *a; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_include_uname_w"); + a = (struct archive_match *)_a; + return (add_owner_name(a, &(a->inclusion_unames), 0, uname)); +} + +int +archive_match_include_gname(struct archive *_a, const char *gname) +{ + struct archive_match *a; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_include_gname"); + a = (struct archive_match *)_a; + return (add_owner_name(a, &(a->inclusion_gnames), 1, gname)); +} + +int +archive_match_include_gname_w(struct archive *_a, const wchar_t *gname) +{ + struct archive_match *a; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_include_gname_w"); + a = (struct archive_match *)_a; + return (add_owner_name(a, &(a->inclusion_gnames), 0, gname)); +} + +/* + * Test function for owner(uid, gid, uname, gname). + * + * Returns 1 if archive entry is excluded. + * Returns 0 if archive entry is not excluded. + * Returns <0 if something error happened. + */ +int +archive_match_owner_excluded(struct archive *_a, + struct archive_entry *entry) +{ + struct archive_match *a; + + archive_check_magic(_a, ARCHIVE_MATCH_MAGIC, + ARCHIVE_STATE_NEW, "archive_match_id_excluded_ae"); + + a = (struct archive_match *)_a; + if (entry == NULL) { + archive_set_error(&(a->archive), EINVAL, "entry is NULL"); + return (ARCHIVE_FAILED); + } + + /* If we don't have inclusion id set at all, the entry is always + * not excluded. */ + if ((a->setflag & ID_IS_SET) == 0) + return (0); + return (owner_excluded(a, entry)); +} + +static int +add_owner_id(struct archive_match *a, struct id_array *ids, int64_t id) +{ + unsigned i; + + if (ids->count + 1 >= ids->size) { + void *p; + + if (ids->size == 0) + ids->size = 8; + else + ids->size *= 2; + p = realloc(ids->ids, sizeof(*ids->ids) * ids->size); + if (p == NULL) + return (error_nomem(a)); + ids->ids = (int64_t *)p; + } + + /* Find an insert point. */ + for (i = 0; i < ids->count; i++) { + if (ids->ids[i] >= id) + break; + } + + /* Add owner id. */ + if (i == ids->count) + ids->ids[ids->count++] = id; + else if (ids->ids[i] != id) { + memmove(&(ids->ids[i+1]), &(ids->ids[i]), + (ids->count - i) * sizeof(ids->ids[0])); + ids->ids[i] = id; + ids->count++; + } + a->setflag |= ID_IS_SET; + return (ARCHIVE_OK); +} + +static int +match_owner_id(struct id_array *ids, int64_t id) +{ + unsigned b, m, t; + + t = 0; + b = (unsigned)ids->count; + while (t < b) { + m = (t + b)>>1; + if (ids->ids[m] == id) + return (1); + if (ids->ids[m] < id) + t = m + 1; + else + b = m; + } + return (0); +} + +static int +add_owner_name(struct archive_match *a, struct match_list *list, + int mbs, const void *name) +{ + struct match *match; + + match = calloc(1, sizeof(*match)); + if (match == NULL) + return (error_nomem(a)); + if (mbs) + archive_mstring_copy_mbs(&(match->pattern), name); + else + archive_mstring_copy_wcs(&(match->pattern), name); + match_list_add(list, match); + a->setflag |= ID_IS_SET; + return (ARCHIVE_OK); +} + +#if !defined(_WIN32) || defined(__CYGWIN__) +static int +match_owner_name_mbs(struct archive_match *a, struct match_list *list, + const char *name) +{ + struct match *m; + const char *p; + + if (name == NULL || *name == '\0') + return (0); + for (m = list->first; m; m = m->next) { + if (archive_mstring_get_mbs(&(a->archive), &(m->pattern), &p) + < 0 && errno == ENOMEM) + return (error_nomem(a)); + if (p != NULL && strcmp(p, name) == 0) { + m->matches++; + return (1); + } + } + return (0); +} +#else +static int +match_owner_name_wcs(struct archive_match *a, struct match_list *list, + const wchar_t *name) +{ + struct match *m; + const wchar_t *p; + + if (name == NULL || *name == L'\0') + return (0); + for (m = list->first; m; m = m->next) { + if (archive_mstring_get_wcs(&(a->archive), &(m->pattern), &p) + < 0 && errno == ENOMEM) + return (error_nomem(a)); + if (p != NULL && wcscmp(p, name) == 0) { + m->matches++; + return (1); + } + } + return (0); +} +#endif + +/* + * Test if entry is excluded by uid, gid, uname or gname. + */ +static int +owner_excluded(struct archive_match *a, struct archive_entry *entry) +{ + int r; + + if (a->inclusion_uids.count) { + if (!match_owner_id(&(a->inclusion_uids), + archive_entry_uid(entry))) + return (1); + } + + if (a->inclusion_gids.count) { + if (!match_owner_id(&(a->inclusion_gids), + archive_entry_gid(entry))) + return (1); + } + + if (a->inclusion_unames.count) { +#if defined(_WIN32) && !defined(__CYGWIN__) + r = match_owner_name_wcs(a, &(a->inclusion_unames), + archive_entry_uname_w(entry)); +#else + r = match_owner_name_mbs(a, &(a->inclusion_unames), + archive_entry_uname(entry)); +#endif + if (!r) + return (1); + else if (r < 0) + return (r); + } + + if (a->inclusion_gnames.count) { +#if defined(_WIN32) && !defined(__CYGWIN__) + r = match_owner_name_wcs(a, &(a->inclusion_gnames), + archive_entry_gname_w(entry)); +#else + r = match_owner_name_mbs(a, &(a->inclusion_gnames), + archive_entry_gname(entry)); +#endif + if (!r) + return (1); + else if (r < 0) + return (r); + } + return (0); +} + diff --git a/c/archive_openssl_evp_private.h b/c/archive_openssl_evp_private.h new file mode 100644 --- /dev/null +++ b/c/archive_openssl_evp_private.h @@ -0,0 +1,53 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#ifndef ARCHIVE_OPENSSL_EVP_PRIVATE_H_INCLUDED +#define ARCHIVE_OPENSSL_EVP_PRIVATE_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif + +#include +#include + +#if OPENSSL_VERSION_NUMBER < 0x10100000L +#include /* malloc, free */ +#include /* memset */ +static inline EVP_MD_CTX *EVP_MD_CTX_new(void) +{ + EVP_MD_CTX *ctx = (EVP_MD_CTX *)calloc(1, sizeof(EVP_MD_CTX)); + return ctx; +} + +static inline void EVP_MD_CTX_free(EVP_MD_CTX *ctx) +{ + EVP_MD_CTX_cleanup(ctx); + memset(ctx, 0, sizeof(*ctx)); + free(ctx); +} +#endif + +#endif diff --git a/c/archive_openssl_hmac_private.h b/c/archive_openssl_hmac_private.h new file mode 100644 --- /dev/null +++ b/c/archive_openssl_hmac_private.h @@ -0,0 +1,54 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#ifndef ARCHIVE_OPENSSL_HMAC_PRIVATE_H_INCLUDED +#define ARCHIVE_OPENSSL_HMAC_PRIVATE_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif + +#include +#include + +#if OPENSSL_VERSION_NUMBER < 0x10100000L || \ + (defined(LIBRESSL_VERSION_NUMBER) && LIBRESSL_VERSION_NUMBER < 0x20700000L) +#include /* malloc, free */ +#include /* memset */ +static inline HMAC_CTX *HMAC_CTX_new(void) +{ + HMAC_CTX *ctx = (HMAC_CTX *)calloc(1, sizeof(HMAC_CTX)); + return ctx; +} + +static inline void HMAC_CTX_free(HMAC_CTX *ctx) +{ + HMAC_CTX_cleanup(ctx); + memset(ctx, 0, sizeof(*ctx)); + free(ctx); +} +#endif + +#endif diff --git a/c/archive_options.c b/c/archive_options.c new file mode 100644 --- /dev/null +++ b/c/archive_options.c @@ -0,0 +1,218 @@ +/*- + * Copyright (c) 2011 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif + +#include "archive_options_private.h" + +static const char * +parse_option(const char **str, + const char **mod, const char **opt, const char **val); + +int +_archive_set_option(struct archive *a, + const char *m, const char *o, const char *v, + int magic, const char *fn, option_handler use_option) +{ + const char *mp, *op, *vp; + int r; + + archive_check_magic(a, magic, ARCHIVE_STATE_NEW, fn); + + mp = (m != NULL && m[0] != '\0') ? m : NULL; + op = (o != NULL && o[0] != '\0') ? o : NULL; + vp = (v != NULL && v[0] != '\0') ? v : NULL; + + if (op == NULL && vp == NULL) + return (ARCHIVE_OK); + if (op == NULL) { + archive_set_error(a, ARCHIVE_ERRNO_MISC, "Empty option"); + return (ARCHIVE_FAILED); + } + + r = use_option(a, mp, op, vp); + if (r == ARCHIVE_WARN - 1) { + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Unknown module name: `%s'", mp); + return (ARCHIVE_FAILED); + } + if (r == ARCHIVE_WARN) { + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Undefined option: `%s%s%s%s%s%s'", + vp?"":"!", mp?mp:"", mp?":":"", op, vp?"=":"", vp?vp:""); + return (ARCHIVE_FAILED); + } + return (r); +} + +int +_archive_set_either_option(struct archive *a, const char *m, const char *o, const char *v, + option_handler use_format_option, option_handler use_filter_option) +{ + int r1, r2; + + if (o == NULL && v == NULL) + return (ARCHIVE_OK); + if (o == NULL) + return (ARCHIVE_FAILED); + + r1 = use_format_option(a, m, o, v); + if (r1 == ARCHIVE_FATAL) + return (ARCHIVE_FATAL); + + r2 = use_filter_option(a, m, o, v); + if (r2 == ARCHIVE_FATAL) + return (ARCHIVE_FATAL); + + if (r2 == ARCHIVE_WARN - 1) + return r1; + return r1 > r2 ? r1 : r2; +} + +int +_archive_set_options(struct archive *a, const char *options, + int magic, const char *fn, option_handler use_option) +{ + int allok = 1, anyok = 0, ignore_mod_err = 0, r; + char *data; + const char *s, *mod, *opt, *val; + + archive_check_magic(a, magic, ARCHIVE_STATE_NEW, fn); + + if (options == NULL || options[0] == '\0') + return ARCHIVE_OK; + + if ((data = strdup(options)) == NULL) { + archive_set_error(a, + ENOMEM, "Out of memory adding file to list"); + return (ARCHIVE_FATAL); + } + s = (const char *)data; + + do { + mod = opt = val = NULL; + + parse_option(&s, &mod, &opt, &val); + if (mod == NULL && opt != NULL && + strcmp("__ignore_wrong_module_name__", opt) == 0) { + /* Ignore module name error */ + if (val != NULL) { + ignore_mod_err = 1; + anyok = 1; + } + continue; + } + + r = use_option(a, mod, opt, val); + if (r == ARCHIVE_FATAL) { + free(data); + return (ARCHIVE_FATAL); + } + if (r == ARCHIVE_FAILED && mod != NULL) { + free(data); + return (ARCHIVE_FAILED); + } + if (r == ARCHIVE_WARN - 1) { + if (ignore_mod_err) + continue; + /* The module name is wrong. */ + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Unknown module name: `%s'", mod); + free(data); + return (ARCHIVE_FAILED); + } + if (r == ARCHIVE_WARN) { + /* The option name is wrong. No-one used this. */ + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Undefined option: `%s%s%s'", + mod?mod:"", mod?":":"", opt); + free(data); + return (ARCHIVE_FAILED); + } + if (r == ARCHIVE_OK) + anyok = 1; + else + allok = 0; + } while (s != NULL); + + free(data); + return allok ? ARCHIVE_OK : anyok ? ARCHIVE_WARN : ARCHIVE_FAILED; +} + +static const char * +parse_option(const char **s, const char **m, const char **o, const char **v) +{ + const char *end, *mod, *opt, *val; + char *p; + + end = NULL; + mod = NULL; + opt = *s; + val = "1"; + + p = strchr(opt, ','); + + if (p != NULL) { + *p = '\0'; + end = ((const char *)p) + 1; + } + + if (0 == strlen(opt)) { + *s = end; + *m = NULL; + *o = NULL; + *v = NULL; + return end; + } + + p = strchr(opt, ':'); + if (p != NULL) { + *p = '\0'; + mod = opt; + opt = ++p; + } + + p = strchr(opt, '='); + if (p != NULL) { + *p = '\0'; + val = ++p; + } else if (opt[0] == '!') { + ++opt; + val = NULL; + } + + *s = end; + *m = mod; + *o = opt; + *v = val; + + return end; +} + diff --git a/c/archive_options_private.h b/c/archive_options_private.h new file mode 100644 --- /dev/null +++ b/c/archive_options_private.h @@ -0,0 +1,51 @@ +/*- + * Copyright (c) 2011 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#ifndef ARCHIVE_OPTIONS_PRIVATE_H_INCLUDED +#define ARCHIVE_OPTIONS_PRIVATE_H_INCLUDED + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#include "archive_private.h" + +typedef int (*option_handler)(struct archive *a, + const char *mod, const char *opt, const char *val); + +int +_archive_set_option(struct archive *a, + const char *mod, const char *opt, const char *val, + int magic, const char *fn, option_handler use_option); + +int +_archive_set_options(struct archive *a, const char *options, + int magic, const char *fn, option_handler use_option); + +int +_archive_set_either_option(struct archive *a, + const char *m, const char *o, const char *v, + option_handler use_format_option, option_handler use_filter_option); + +#endif diff --git a/c/archive_pack_dev.c b/c/archive_pack_dev.c new file mode 100644 --- /dev/null +++ b/c/archive_pack_dev.c @@ -0,0 +1,336 @@ +/* $NetBSD: pack_dev.c,v 1.12 2013/06/14 16:28:20 tsutsui Exp $ */ + +/*- + * Copyright (c) 1998, 2001 The NetBSD Foundation, Inc. + * All rights reserved. + * + * This code is derived from software contributed to The NetBSD Foundation + * by Charles M. Hannum. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS + * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED + * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS + * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR + * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF + * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS + * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN + * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) + * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE + * POSSIBILITY OF SUCH DAMAGE. + */ + +/* Originally from NetBSD's mknod(8) source. */ + +#include "archive_platform.h" + +#if HAVE_SYS_CDEFS_H +#include +#endif +#if !defined(lint) +__RCSID("$NetBSD$"); +#endif /* not lint */ + +#ifdef HAVE_LIMITS_H +#include +#endif + +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_SYS_TYPES_H +#include +#endif +#ifdef HAVE_SYS_STAT_H +#include +#endif +#if MAJOR_IN_MKDEV +#include +#define HAVE_MAJOR +#elif MAJOR_IN_SYSMACROS +#include +#define HAVE_MAJOR +#endif +#ifdef HAVE_UNISTD_H +#include +#endif + +#include "archive_pack_dev.h" + +static pack_t pack_netbsd; +static pack_t pack_freebsd; +static pack_t pack_8_8; +static pack_t pack_12_20; +static pack_t pack_14_18; +static pack_t pack_8_24; +static pack_t pack_bsdos; +static int compare_format(const void *, const void *); + +static const char iMajorError[] = "invalid major number"; +static const char iMinorError[] = "invalid minor number"; +static const char tooManyFields[] = "too many fields for format"; + +/* This is blatantly stolen from libarchive/archive_entry.c, + * in an attempt to get this to play nice on MinGW... */ +#if !defined(HAVE_MAJOR) && !defined(major) +/* Replacement for major/minor/makedev. */ +#define major(x) ((int)(0x00ff & ((x) >> 8))) +#define minor(x) ((int)(0xffff00ff & (x))) +#define makedev(maj,min) ((0xff00 & ((maj)<<8)) | (0xffff00ff & (min))) +#endif + +/* Play games to come up with a suitable makedev() definition. */ +#ifdef __QNXNTO__ +/* QNX. */ +#include +#define apd_makedev(maj, min) makedev(ND_LOCAL_NODE, (maj), (min)) +#elif defined makedev +/* There's a "makedev" macro. */ +#define apd_makedev(maj, min) makedev((maj), (min)) +#elif defined mkdev || ((defined _WIN32 || defined __WIN32__) && !defined(__CYGWIN__)) +/* Windows. */ +#define apd_makedev(maj, min) mkdev((maj), (min)) +#else +/* There's a "makedev" function. */ +#define apd_makedev(maj, min) makedev((maj), (min)) +#endif + +/* exported */ +dev_t +pack_native(int n, unsigned long numbers[], const char **error) +{ + dev_t dev = 0; + + if (n == 2) { + dev = apd_makedev(numbers[0], numbers[1]); + if ((unsigned long)major(dev) != numbers[0]) + *error = iMajorError; + else if ((unsigned long)minor(dev) != numbers[1]) + *error = iMinorError; + } else + *error = tooManyFields; + return (dev); +} + + +static dev_t +pack_netbsd(int n, unsigned long numbers[], const char **error) +{ + dev_t dev = 0; + + if (n == 2) { + dev = makedev_netbsd(numbers[0], numbers[1]); + if ((unsigned long)major_netbsd(dev) != numbers[0]) + *error = iMajorError; + else if ((unsigned long)minor_netbsd(dev) != numbers[1]) + *error = iMinorError; + } else + *error = tooManyFields; + return (dev); +} + + +#define major_freebsd(x) ((int32_t)(((x) & 0x0000ff00) >> 8)) +#define minor_freebsd(x) ((int32_t)(((x) & 0xffff00ff) >> 0)) +#define makedev_freebsd(x,y) ((dev_t)((((x) << 8) & 0x0000ff00) | \ + (((y) << 0) & 0xffff00ff))) + +static dev_t +pack_freebsd(int n, unsigned long numbers[], const char **error) +{ + dev_t dev = 0; + + if (n == 2) { + dev = makedev_freebsd(numbers[0], numbers[1]); + if ((unsigned long)major_freebsd(dev) != numbers[0]) + *error = iMajorError; + if ((unsigned long)minor_freebsd(dev) != numbers[1]) + *error = iMinorError; + } else + *error = tooManyFields; + return (dev); +} + + +#define major_8_8(x) ((int32_t)(((x) & 0x0000ff00) >> 8)) +#define minor_8_8(x) ((int32_t)(((x) & 0x000000ff) >> 0)) +#define makedev_8_8(x,y) ((dev_t)((((x) << 8) & 0x0000ff00) | \ + (((y) << 0) & 0x000000ff))) + +static dev_t +pack_8_8(int n, unsigned long numbers[], const char **error) +{ + dev_t dev = 0; + + if (n == 2) { + dev = makedev_8_8(numbers[0], numbers[1]); + if ((unsigned long)major_8_8(dev) != numbers[0]) + *error = iMajorError; + if ((unsigned long)minor_8_8(dev) != numbers[1]) + *error = iMinorError; + } else + *error = tooManyFields; + return (dev); +} + + +#define major_12_20(x) ((int32_t)(((x) & 0xfff00000) >> 20)) +#define minor_12_20(x) ((int32_t)(((x) & 0x000fffff) >> 0)) +#define makedev_12_20(x,y) ((dev_t)((((x) << 20) & 0xfff00000) | \ + (((y) << 0) & 0x000fffff))) + +static dev_t +pack_12_20(int n, unsigned long numbers[], const char **error) +{ + dev_t dev = 0; + + if (n == 2) { + dev = makedev_12_20(numbers[0], numbers[1]); + if ((unsigned long)major_12_20(dev) != numbers[0]) + *error = iMajorError; + if ((unsigned long)minor_12_20(dev) != numbers[1]) + *error = iMinorError; + } else + *error = tooManyFields; + return (dev); +} + + +#define major_14_18(x) ((int32_t)(((x) & 0xfffc0000) >> 18)) +#define minor_14_18(x) ((int32_t)(((x) & 0x0003ffff) >> 0)) +#define makedev_14_18(x,y) ((dev_t)((((x) << 18) & 0xfffc0000) | \ + (((y) << 0) & 0x0003ffff))) + +static dev_t +pack_14_18(int n, unsigned long numbers[], const char **error) +{ + dev_t dev = 0; + + if (n == 2) { + dev = makedev_14_18(numbers[0], numbers[1]); + if ((unsigned long)major_14_18(dev) != numbers[0]) + *error = iMajorError; + if ((unsigned long)minor_14_18(dev) != numbers[1]) + *error = iMinorError; + } else + *error = tooManyFields; + return (dev); +} + + +#define major_8_24(x) ((int32_t)(((x) & 0xff000000) >> 24)) +#define minor_8_24(x) ((int32_t)(((x) & 0x00ffffff) >> 0)) +#define makedev_8_24(x,y) ((dev_t)((((x) << 24) & 0xff000000) | \ + (((y) << 0) & 0x00ffffff))) + +static dev_t +pack_8_24(int n, unsigned long numbers[], const char **error) +{ + dev_t dev = 0; + + if (n == 2) { + dev = makedev_8_24(numbers[0], numbers[1]); + if ((unsigned long)major_8_24(dev) != numbers[0]) + *error = iMajorError; + if ((unsigned long)minor_8_24(dev) != numbers[1]) + *error = iMinorError; + } else + *error = tooManyFields; + return (dev); +} + + +#define major_12_12_8(x) ((int32_t)(((x) & 0xfff00000) >> 20)) +#define unit_12_12_8(x) ((int32_t)(((x) & 0x000fff00) >> 8)) +#define subunit_12_12_8(x) ((int32_t)(((x) & 0x000000ff) >> 0)) +#define makedev_12_12_8(x,y,z) ((dev_t)((((x) << 20) & 0xfff00000) | \ + (((y) << 8) & 0x000fff00) | \ + (((z) << 0) & 0x000000ff))) + +static dev_t +pack_bsdos(int n, unsigned long numbers[], const char **error) +{ + dev_t dev = 0; + + if (n == 2) { + dev = makedev_12_20(numbers[0], numbers[1]); + if ((unsigned long)major_12_20(dev) != numbers[0]) + *error = iMajorError; + if ((unsigned long)minor_12_20(dev) != numbers[1]) + *error = iMinorError; + } else if (n == 3) { + dev = makedev_12_12_8(numbers[0], numbers[1], numbers[2]); + if ((unsigned long)major_12_12_8(dev) != numbers[0]) + *error = iMajorError; + if ((unsigned long)unit_12_12_8(dev) != numbers[1]) + *error = "invalid unit number"; + if ((unsigned long)subunit_12_12_8(dev) != numbers[2]) + *error = "invalid subunit number"; + } else + *error = tooManyFields; + return (dev); +} + + + /* list of formats and pack functions */ + /* this list must be sorted lexically */ +static const struct format { + const char *name; + pack_t *pack; +} formats[] = { + {"386bsd", pack_8_8}, + {"4bsd", pack_8_8}, + {"bsdos", pack_bsdos}, + {"freebsd", pack_freebsd}, + {"hpux", pack_8_24}, + {"isc", pack_8_8}, + {"linux", pack_8_8}, + {"native", pack_native}, + {"netbsd", pack_netbsd}, + {"osf1", pack_12_20}, + {"sco", pack_8_8}, + {"solaris", pack_14_18}, + {"sunos", pack_8_8}, + {"svr3", pack_8_8}, + {"svr4", pack_14_18}, + {"ultrix", pack_8_8}, +}; + +static int +compare_format(const void *key, const void *element) +{ + const char *name; + const struct format *format; + + name = key; + format = element; + + return (strcmp(name, format->name)); +} + + +pack_t * +pack_find(const char *name) +{ + struct format *format; + + format = bsearch(name, formats, + sizeof(formats)/sizeof(formats[0]), + sizeof(formats[0]), compare_format); + if (format == 0) + return (NULL); + return (format->pack); +} diff --git a/c/archive_pack_dev.h b/c/archive_pack_dev.h new file mode 100644 --- /dev/null +++ b/c/archive_pack_dev.h @@ -0,0 +1,49 @@ +/* $NetBSD: pack_dev.h,v 1.8 2013/06/14 16:28:20 tsutsui Exp $ */ + +/*- + * Copyright (c) 1998, 2001 The NetBSD Foundation, Inc. + * All rights reserved. + * + * This code is derived from software contributed to The NetBSD Foundation + * by Charles M. Hannum. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS + * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED + * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS + * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR + * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF + * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS + * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN + * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) + * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE + * POSSIBILITY OF SUCH DAMAGE. + */ + +/* Originally from NetBSD's mknod(8) source. */ + +#ifndef ARCHIVE_PACK_DEV_H +#define ARCHIVE_PACK_DEV_H + +typedef dev_t pack_t(int, unsigned long [], const char **); + +pack_t *pack_find(const char *); +pack_t pack_native; + +#define major_netbsd(x) ((int32_t)((((x) & 0x000fff00) >> 8))) +#define minor_netbsd(x) ((int32_t)((((x) & 0xfff00000) >> 12) | \ + (((x) & 0x000000ff) >> 0))) +#define makedev_netbsd(x,y) ((dev_t)((((x) << 8) & 0x000fff00) | \ + (((y) << 12) & 0xfff00000) | \ + (((y) << 0) & 0x000000ff))) + +#endif /* ARCHIVE_PACK_DEV_H */ diff --git a/c/archive_pathmatch.c b/c/archive_pathmatch.c new file mode 100644 --- /dev/null +++ b/c/archive_pathmatch.c @@ -0,0 +1,459 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer + * in this position and unchanged. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_WCHAR_H +#include +#endif + +#include "archive_pathmatch.h" + +/* + * Check whether a character 'c' is matched by a list specification [...]: + * * Leading '!' or '^' negates the class. + * * - is a range of characters + * * \ removes any special meaning for + * + * Some interesting boundary cases: + * a-d-e is one range (a-d) followed by two single characters - and e. + * \a-\d is same as a-d + * a\-d is three single characters: a, d, - + * Trailing - is not special (so [a-] is two characters a and -). + * Initial - is not special ([a-] is same as [-a] is same as [\\-a]) + * This function never sees a trailing \. + * [] always fails + * [!] always succeeds + */ +static int +pm_list(const char *start, const char *end, const char c, int flags) +{ + const char *p = start; + char rangeStart = '\0', nextRangeStart; + int match = 1, nomatch = 0; + + /* This will be used soon... */ + (void)flags; /* UNUSED */ + + /* If this is a negated class, return success for nomatch. */ + if ((*p == '!' || *p == '^') && p < end) { + match = 0; + nomatch = 1; + ++p; + } + + while (p < end) { + nextRangeStart = '\0'; + switch (*p) { + case '-': + /* Trailing or initial '-' is not special. */ + if ((rangeStart == '\0') || (p == end - 1)) { + if (*p == c) + return (match); + } else { + char rangeEnd = *++p; + if (rangeEnd == '\\') + rangeEnd = *++p; + if ((rangeStart <= c) && (c <= rangeEnd)) + return (match); + } + break; + case '\\': + ++p; + /* Fall through */ + default: + if (*p == c) + return (match); + nextRangeStart = *p; /* Possible start of range. */ + } + rangeStart = nextRangeStart; + ++p; + } + return (nomatch); +} + +static int +pm_list_w(const wchar_t *start, const wchar_t *end, const wchar_t c, int flags) +{ + const wchar_t *p = start; + wchar_t rangeStart = L'\0', nextRangeStart; + int match = 1, nomatch = 0; + + /* This will be used soon... */ + (void)flags; /* UNUSED */ + + /* If this is a negated class, return success for nomatch. */ + if ((*p == L'!' || *p == L'^') && p < end) { + match = 0; + nomatch = 1; + ++p; + } + + while (p < end) { + nextRangeStart = L'\0'; + switch (*p) { + case L'-': + /* Trailing or initial '-' is not special. */ + if ((rangeStart == L'\0') || (p == end - 1)) { + if (*p == c) + return (match); + } else { + wchar_t rangeEnd = *++p; + if (rangeEnd == L'\\') + rangeEnd = *++p; + if ((rangeStart <= c) && (c <= rangeEnd)) + return (match); + } + break; + case L'\\': + ++p; + /* Fall through */ + default: + if (*p == c) + return (match); + nextRangeStart = *p; /* Possible start of range. */ + } + rangeStart = nextRangeStart; + ++p; + } + return (nomatch); +} + +/* + * If s is pointing to "./", ".//", "./././" or the like, skip it. + */ +static const char * +pm_slashskip(const char *s) { + while ((*s == '/') + || (s[0] == '.' && s[1] == '/') + || (s[0] == '.' && s[1] == '\0')) + ++s; + return (s); +} + +static const wchar_t * +pm_slashskip_w(const wchar_t *s) { + while ((*s == L'/') + || (s[0] == L'.' && s[1] == L'/') + || (s[0] == L'.' && s[1] == L'\0')) + ++s; + return (s); +} + +static int +pm(const char *p, const char *s, int flags) +{ + const char *end; + + /* + * Ignore leading './', './/', '././', etc. + */ + if (s[0] == '.' && s[1] == '/') + s = pm_slashskip(s + 1); + if (p[0] == '.' && p[1] == '/') + p = pm_slashskip(p + 1); + + for (;;) { + switch (*p) { + case '\0': + if (s[0] == '/') { + if (flags & PATHMATCH_NO_ANCHOR_END) + return (1); + /* "dir" == "dir/" == "dir/." */ + s = pm_slashskip(s); + } + return (*s == '\0'); + case '?': + /* ? always succeeds, unless we hit end of 's' */ + if (*s == '\0') + return (0); + break; + case '*': + /* "*" == "**" == "***" ... */ + while (*p == '*') + ++p; + /* Trailing '*' always succeeds. */ + if (*p == '\0') + return (1); + while (*s) { + if (archive_pathmatch(p, s, flags)) + return (1); + ++s; + } + return (0); + case '[': + /* + * Find the end of the [...] character class, + * ignoring \] that might occur within the class. + */ + end = p + 1; + while (*end != '\0' && *end != ']') { + if (*end == '\\' && end[1] != '\0') + ++end; + ++end; + } + if (*end == ']') { + /* We found [...], try to match it. */ + if (!pm_list(p + 1, end, *s, flags)) + return (0); + p = end; /* Jump to trailing ']' char. */ + break; + } else + /* No final ']', so just match '['. */ + if (*p != *s) + return (0); + break; + case '\\': + /* Trailing '\\' matches itself. */ + if (p[1] == '\0') { + if (*s != '\\') + return (0); + } else { + ++p; + if (*p != *s) + return (0); + } + break; + case '/': + if (*s != '/' && *s != '\0') + return (0); + /* Note: pattern "/\./" won't match "/"; + * pm_slashskip() correctly stops at backslash. */ + p = pm_slashskip(p); + s = pm_slashskip(s); + if (*p == '\0' && (flags & PATHMATCH_NO_ANCHOR_END)) + return (1); + --p; /* Counteract the increment below. */ + --s; + break; + case '$': + /* '$' is special only at end of pattern and only + * if PATHMATCH_NO_ANCHOR_END is specified. */ + if (p[1] == '\0' && (flags & PATHMATCH_NO_ANCHOR_END)){ + /* "dir" == "dir/" == "dir/." */ + return (*pm_slashskip(s) == '\0'); + } + /* Otherwise, '$' is not special. */ + /* FALL THROUGH */ + default: + if (*p != *s) + return (0); + break; + } + ++p; + ++s; + } +} + +static int +pm_w(const wchar_t *p, const wchar_t *s, int flags) +{ + const wchar_t *end; + + /* + * Ignore leading './', './/', '././', etc. + */ + if (s[0] == L'.' && s[1] == L'/') + s = pm_slashskip_w(s + 1); + if (p[0] == L'.' && p[1] == L'/') + p = pm_slashskip_w(p + 1); + + for (;;) { + switch (*p) { + case L'\0': + if (s[0] == L'/') { + if (flags & PATHMATCH_NO_ANCHOR_END) + return (1); + /* "dir" == "dir/" == "dir/." */ + s = pm_slashskip_w(s); + } + return (*s == L'\0'); + case L'?': + /* ? always succeeds, unless we hit end of 's' */ + if (*s == L'\0') + return (0); + break; + case L'*': + /* "*" == "**" == "***" ... */ + while (*p == L'*') + ++p; + /* Trailing '*' always succeeds. */ + if (*p == L'\0') + return (1); + while (*s) { + if (archive_pathmatch_w(p, s, flags)) + return (1); + ++s; + } + return (0); + case L'[': + /* + * Find the end of the [...] character class, + * ignoring \] that might occur within the class. + */ + end = p + 1; + while (*end != L'\0' && *end != L']') { + if (*end == L'\\' && end[1] != L'\0') + ++end; + ++end; + } + if (*end == L']') { + /* We found [...], try to match it. */ + if (!pm_list_w(p + 1, end, *s, flags)) + return (0); + p = end; /* Jump to trailing ']' char. */ + break; + } else + /* No final ']', so just match '['. */ + if (*p != *s) + return (0); + break; + case L'\\': + /* Trailing '\\' matches itself. */ + if (p[1] == L'\0') { + if (*s != L'\\') + return (0); + } else { + ++p; + if (*p != *s) + return (0); + } + break; + case L'/': + if (*s != L'/' && *s != L'\0') + return (0); + /* Note: pattern "/\./" won't match "/"; + * pm_slashskip() correctly stops at backslash. */ + p = pm_slashskip_w(p); + s = pm_slashskip_w(s); + if (*p == L'\0' && (flags & PATHMATCH_NO_ANCHOR_END)) + return (1); + --p; /* Counteract the increment below. */ + --s; + break; + case L'$': + /* '$' is special only at end of pattern and only + * if PATHMATCH_NO_ANCHOR_END is specified. */ + if (p[1] == L'\0' && (flags & PATHMATCH_NO_ANCHOR_END)){ + /* "dir" == "dir/" == "dir/." */ + return (*pm_slashskip_w(s) == L'\0'); + } + /* Otherwise, '$' is not special. */ + /* FALL THROUGH */ + default: + if (*p != *s) + return (0); + break; + } + ++p; + ++s; + } +} + +/* Main entry point. */ +int +__archive_pathmatch(const char *p, const char *s, int flags) +{ + /* Empty pattern only matches the empty string. */ + if (p == NULL || *p == '\0') + return (s == NULL || *s == '\0'); + + /* Leading '^' anchors the start of the pattern. */ + if (*p == '^') { + ++p; + flags &= ~PATHMATCH_NO_ANCHOR_START; + } + + if (*p == '/' && *s != '/') + return (0); + + /* Certain patterns anchor implicitly. */ + if (*p == '*' || *p == '/') { + while (*p == '/') + ++p; + while (*s == '/') + ++s; + return (pm(p, s, flags)); + } + + /* If start is unanchored, try to match start of each path element. */ + if (flags & PATHMATCH_NO_ANCHOR_START) { + for ( ; s != NULL; s = strchr(s, '/')) { + if (*s == '/') + s++; + if (pm(p, s, flags)) + return (1); + } + return (0); + } + + /* Default: Match from beginning. */ + return (pm(p, s, flags)); +} + +int +__archive_pathmatch_w(const wchar_t *p, const wchar_t *s, int flags) +{ + /* Empty pattern only matches the empty string. */ + if (p == NULL || *p == L'\0') + return (s == NULL || *s == L'\0'); + + /* Leading '^' anchors the start of the pattern. */ + if (*p == L'^') { + ++p; + flags &= ~PATHMATCH_NO_ANCHOR_START; + } + + if (*p == L'/' && *s != L'/') + return (0); + + /* Certain patterns anchor implicitly. */ + if (*p == L'*' || *p == L'/') { + while (*p == L'/') + ++p; + while (*s == L'/') + ++s; + return (pm_w(p, s, flags)); + } + + /* If start is unanchored, try to match start of each path element. */ + if (flags & PATHMATCH_NO_ANCHOR_START) { + for ( ; s != NULL; s = wcschr(s, L'/')) { + if (*s == L'/') + s++; + if (pm_w(p, s, flags)) + return (1); + } + return (0); + } + + /* Default: Match from beginning. */ + return (pm_w(p, s, flags)); +} diff --git a/c/archive_pathmatch.h b/c/archive_pathmatch.h new file mode 100644 --- /dev/null +++ b/c/archive_pathmatch.h @@ -0,0 +1,52 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer + * in this position and unchanged. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD$ + */ + +#ifndef ARCHIVE_PATHMATCH_H +#define ARCHIVE_PATHMATCH_H + +#ifndef __LIBARCHIVE_BUILD +#ifndef __LIBARCHIVE_TEST +#error This header is only to be used internally to libarchive. +#endif +#endif + +/* Don't anchor at beginning unless the pattern starts with "^" */ +#define PATHMATCH_NO_ANCHOR_START 1 +/* Don't anchor at end unless the pattern ends with "$" */ +#define PATHMATCH_NO_ANCHOR_END 2 + +/* Note that "^" and "$" are not special unless you set the corresponding + * flag above. */ + +int __archive_pathmatch(const char *p, const char *s, int flags); +int __archive_pathmatch_w(const wchar_t *p, const wchar_t *s, int flags); + +#define archive_pathmatch(p, s, f) __archive_pathmatch(p, s, f) +#define archive_pathmatch_w(p, s, f) __archive_pathmatch_w(p, s, f) + +#endif diff --git a/c/archive_platform.h b/c/archive_platform.h new file mode 100644 --- /dev/null +++ b/c/archive_platform.h @@ -0,0 +1,202 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD: head/lib/libarchive/archive_platform.h 201090 2009-12-28 02:22:04Z kientzle $ + */ + +/* !!ONLY FOR USE INTERNALLY TO LIBARCHIVE!! */ + +/* + * This header is the first thing included in any of the libarchive + * source files. As far as possible, platform-specific issues should + * be dealt with here and not within individual source files. I'm + * actively trying to minimize #if blocks within the main source, + * since they obfuscate the code. + */ + +#ifndef ARCHIVE_PLATFORM_H_INCLUDED +#define ARCHIVE_PLATFORM_H_INCLUDED + +/* archive.h and archive_entry.h require this. */ +#define __LIBARCHIVE_BUILD 1 + +#if defined(PLATFORM_CONFIG_H) +/* Use hand-built config.h in environments that need it. */ +#include PLATFORM_CONFIG_H +#elif defined(HAVE_CONFIG_H) +/* Most POSIX platforms use the 'configure' script to build config.h */ +#include "config.h" +#else +/* Warn if the library hasn't been (automatically or manually) configured. */ +#error Oops: No config.h and no pre-built configuration in archive_platform.h. +#endif + +/* On macOS check for some symbols based on the deployment target version. */ +#if defined(__APPLE__) +# undef HAVE_FUTIMENS +# undef HAVE_UTIMENSAT +# include +# if MAC_OS_X_VERSION_MIN_REQUIRED >= 101300 +# define HAVE_FUTIMENS 1 +# define HAVE_UTIMENSAT 1 +# endif +#endif + +/* It should be possible to get rid of this by extending the feature-test + * macros to cover Windows API functions, probably along with non-trivial + * refactoring of code to find structures that sit more cleanly on top of + * either Windows or Posix APIs. */ +#if (defined(__WIN32__) || defined(_WIN32) || defined(__WIN32)) && !defined(__CYGWIN__) +#include "archive_windows.h" +#else +#define la_stat(path,stref) stat(path,stref) +#endif + +/* + * The config files define a lot of feature macros. The following + * uses those macros to select/define replacements and include key + * headers as required. + */ + +/* Get a real definition for __FBSDID or __RCSID if we can */ +#if HAVE_SYS_CDEFS_H +#include +#endif + +/* If not, define them so as to avoid dangling semicolons. */ +#ifndef __FBSDID +#define __FBSDID(a) struct _undefined_hack +#endif +#ifndef __RCSID +#define __RCSID(a) struct _undefined_hack +#endif + +/* Try to get standard C99-style integer type definitions. */ +#if HAVE_INTTYPES_H +#include +#endif +#if HAVE_STDINT_H +#include +#endif + +/* Borland warns about its own constants! */ +#if defined(__BORLANDC__) +# if HAVE_DECL_UINT64_MAX +# undef UINT64_MAX +# undef HAVE_DECL_UINT64_MAX +# endif +# if HAVE_DECL_UINT64_MIN +# undef UINT64_MIN +# undef HAVE_DECL_UINT64_MIN +# endif +# if HAVE_DECL_INT64_MAX +# undef INT64_MAX +# undef HAVE_DECL_INT64_MAX +# endif +# if HAVE_DECL_INT64_MIN +# undef INT64_MIN +# undef HAVE_DECL_INT64_MIN +# endif +#endif + +/* Some platforms lack the standard *_MAX definitions. */ +#if !HAVE_DECL_SIZE_MAX +#define SIZE_MAX (~(size_t)0) +#endif +#if !HAVE_DECL_SSIZE_MAX +#define SSIZE_MAX ((ssize_t)(SIZE_MAX >> 1)) +#endif +#if !HAVE_DECL_UINT32_MAX +#define UINT32_MAX (~(uint32_t)0) +#endif +#if !HAVE_DECL_INT32_MAX +#define INT32_MAX ((int32_t)(UINT32_MAX >> 1)) +#endif +#if !HAVE_DECL_INT32_MIN +#define INT32_MIN ((int32_t)(~INT32_MAX)) +#endif +#if !HAVE_DECL_UINT64_MAX +#define UINT64_MAX (~(uint64_t)0) +#endif +#if !HAVE_DECL_INT64_MAX +#define INT64_MAX ((int64_t)(UINT64_MAX >> 1)) +#endif +#if !HAVE_DECL_INT64_MIN +#define INT64_MIN ((int64_t)(~INT64_MAX)) +#endif +#if !HAVE_DECL_UINTMAX_MAX +#define UINTMAX_MAX (~(uintmax_t)0) +#endif +#if !HAVE_DECL_INTMAX_MAX +#define INTMAX_MAX ((intmax_t)(UINTMAX_MAX >> 1)) +#endif +#if !HAVE_DECL_INTMAX_MIN +#define INTMAX_MIN ((intmax_t)(~INTMAX_MAX)) +#endif + +/* + * If we can't restore metadata using a file descriptor, then + * for compatibility's sake, close files before trying to restore metadata. + */ +#if defined(HAVE_FCHMOD) || defined(HAVE_FUTIMES) || defined(HAVE_ACL_SET_FD) || defined(HAVE_ACL_SET_FD_NP) || defined(HAVE_FCHOWN) +#define CAN_RESTORE_METADATA_FD +#endif + +/* + * glibc 2.24 deprecates readdir_r + */ +#if defined(HAVE_READDIR_R) && (!defined(__GLIBC__) || !defined(__GLIBC_MINOR__) || __GLIBC__ < 2 || (__GLIBC__ == 2 && __GLIBC_MINOR__ < 24)) +#define USE_READDIR_R 1 +#else +#undef USE_READDIR_R +#endif + +/* Set up defaults for internal error codes. */ +#ifndef ARCHIVE_ERRNO_FILE_FORMAT +#if HAVE_EFTYPE +#define ARCHIVE_ERRNO_FILE_FORMAT EFTYPE +#else +#if HAVE_EILSEQ +#define ARCHIVE_ERRNO_FILE_FORMAT EILSEQ +#else +#define ARCHIVE_ERRNO_FILE_FORMAT EINVAL +#endif +#endif +#endif + +#ifndef ARCHIVE_ERRNO_PROGRAMMER +#define ARCHIVE_ERRNO_PROGRAMMER EINVAL +#endif + +#ifndef ARCHIVE_ERRNO_MISC +#define ARCHIVE_ERRNO_MISC (-1) +#endif + +#if defined(__GNUC__) && (__GNUC__ >= 7) +#define __LA_FALLTHROUGH __attribute__((fallthrough)) +#else +#define __LA_FALLTHROUGH +#endif + +#endif /* !ARCHIVE_PLATFORM_H_INCLUDED */ diff --git a/c/archive_platform_acl.h b/c/archive_platform_acl.h new file mode 100644 --- /dev/null +++ b/c/archive_platform_acl.h @@ -0,0 +1,55 @@ +/*- + * Copyright (c) 2017 Martin Matuska + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD$ + */ + +/* !!ONLY FOR USE INTERNALLY TO LIBARCHIVE!! */ + +#ifndef ARCHIVE_PLATFORM_ACL_H_INCLUDED +#define ARCHIVE_PLATFORM_ACL_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#ifndef __LIBARCHIVE_TEST_COMMON +#error This header is only to be used internally to libarchive. +#endif +#endif + +/* + * Determine what ACL types are supported + */ +#if ARCHIVE_ACL_FREEBSD || ARCHIVE_ACL_SUNOS || ARCHIVE_ACL_LIBACL +#define ARCHIVE_ACL_POSIX1E 1 +#endif + +#if ARCHIVE_ACL_FREEBSD_NFS4 || ARCHIVE_ACL_SUNOS_NFS4 || \ + ARCHIVE_ACL_DARWIN || ARCHIVE_ACL_LIBRICHACL +#define ARCHIVE_ACL_NFS4 1 +#endif + +#if ARCHIVE_ACL_POSIX1E || ARCHIVE_ACL_NFS4 +#define ARCHIVE_ACL_SUPPORT 1 +#endif + +#endif /* ARCHIVE_PLATFORM_ACL_H_INCLUDED */ diff --git a/c/archive_platform_xattr.h b/c/archive_platform_xattr.h new file mode 100644 --- /dev/null +++ b/c/archive_platform_xattr.h @@ -0,0 +1,47 @@ +/*- + * Copyright (c) 2017 Martin Matuska + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD$ + */ + +/* !!ONLY FOR USE INTERNALLY TO LIBARCHIVE!! */ + +#ifndef ARCHIVE_PLATFORM_XATTR_H_INCLUDED +#define ARCHIVE_PLATFORM_XATTR_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#ifndef __LIBARCHIVE_TEST_COMMON +#error This header is only to be used internally to libarchive. +#endif +#endif + +/* + * Determine if we support extended attributes + */ +#if ARCHIVE_XATTR_LINUX || ARCHIVE_XATTR_DARWIN || ARCHIVE_XATTR_FREEBSD || \ + ARCHIVE_XATTR_AIX +#define ARCHIVE_XATTR_SUPPORT 1 +#endif + +#endif /* ARCHIVE_PLATFORM_XATTR_H_INCLUDED */ diff --git a/c/archive_ppmd7.c b/c/archive_ppmd7.c new file mode 100644 --- /dev/null +++ b/c/archive_ppmd7.c @@ -0,0 +1,1168 @@ +/* Ppmd7.c -- PPMdH codec +2010-03-12 : Igor Pavlov : Public domain +This code is based on PPMd var.H (2001): Dmitry Shkarin : Public domain */ + +#include "archive_platform.h" + +#include + +#include "archive_ppmd7_private.h" + +#ifdef PPMD_32BIT + #define Ppmd7_GetPtr(p, ptr) (ptr) + #define Ppmd7_GetContext(p, ptr) (ptr) + #define Ppmd7_GetStats(p, ctx) ((ctx)->Stats) +#else + #define Ppmd7_GetPtr(p, offs) ((void *)((p)->Base + (offs))) + #define Ppmd7_GetContext(p, offs) ((CPpmd7_Context *)Ppmd7_GetPtr((p), (offs))) + #define Ppmd7_GetStats(p, ctx) ((CPpmd_State *)Ppmd7_GetPtr((p), ((ctx)->Stats))) +#endif + +#define Ppmd7_GetBinSumm(p) \ + &p->BinSumm[Ppmd7Context_OneState(p->MinContext)->Freq - 1][p->PrevSuccess + \ + p->NS2BSIndx[Ppmd7_GetContext(p, p->MinContext->Suffix)->NumStats - 1] + \ + (p->HiBitsFlag = p->HB2Flag[p->FoundState->Symbol]) + \ + 2 * p->HB2Flag[Ppmd7Context_OneState(p->MinContext)->Symbol] + \ + ((p->RunLength >> 26) & 0x20)] + +#define kTopValue (1 << 24) +#define MAX_FREQ 124 +#define UNIT_SIZE 12 + +#define U2B(nu) ((UInt32)(nu) * UNIT_SIZE) +#define U2I(nu) (p->Units2Indx[(nu) - 1]) +#define I2U(indx) (p->Indx2Units[indx]) + +#ifdef PPMD_32BIT + #define REF(ptr) (ptr) +#else + #define REF(ptr) ((UInt32)((Byte *)(ptr) - (p)->Base)) +#endif + +#define STATS_REF(ptr) ((CPpmd_State_Ref)REF(ptr)) + +#define CTX(ref) ((CPpmd7_Context *)Ppmd7_GetContext(p, ref)) +#define STATS(ctx) Ppmd7_GetStats(p, ctx) +#define ONE_STATE(ctx) Ppmd7Context_OneState(ctx) +#define SUFFIX(ctx) CTX((ctx)->Suffix) + +static const UInt16 kInitBinEsc[] = { 0x3CDD, 0x1F3F, 0x59BF, 0x48F3, 0x64A1, 0x5ABC, 0x6632, 0x6051}; +static const Byte PPMD7_kExpEscape[16] = { 25, 14, 9, 7, 5, 5, 4, 4, 4, 3, 3, 3, 2, 2, 2, 2 }; + +typedef CPpmd7_Context * CTX_PTR; + +struct CPpmd7_Node_; + +typedef + #ifdef PPMD_32BIT + struct CPpmd7_Node_ * + #else + UInt32 + #endif + CPpmd7_Node_Ref; + +typedef struct CPpmd7_Node_ +{ + UInt16 Stamp; /* must be at offset 0 as CPpmd7_Context::NumStats. Stamp=0 means free */ + UInt16 NU; + CPpmd7_Node_Ref Next; /* must be at offset >= 4 */ + CPpmd7_Node_Ref Prev; +} CPpmd7_Node; + +#ifdef PPMD_32BIT + #define NODE(ptr) (ptr) +#else + #define NODE(offs) ((CPpmd7_Node *)(p->Base + (offs))) +#endif + +static void Ppmd7_Update1(CPpmd7 *p); +static void Ppmd7_Update1_0(CPpmd7 *p); +static void Ppmd7_Update2(CPpmd7 *p); +static void Ppmd7_UpdateBin(CPpmd7 *p); +static CPpmd_See *Ppmd7_MakeEscFreq(CPpmd7 *p, unsigned numMasked, + UInt32 *scale); + +/* ----------- Base ----------- */ + +static void Ppmd7_Construct(CPpmd7 *p) +{ + unsigned i, k, m; + + p->Base = 0; + + for (i = 0, k = 0; i < PPMD_NUM_INDEXES; i++) + { + unsigned step = (i >= 12 ? 4 : (i >> 2) + 1); + do { p->Units2Indx[k++] = (Byte)i; } while(--step); + p->Indx2Units[i] = (Byte)k; + } + + p->NS2BSIndx[0] = (0 << 1); + p->NS2BSIndx[1] = (1 << 1); + memset(p->NS2BSIndx + 2, (2 << 1), 9); + memset(p->NS2BSIndx + 11, (3 << 1), 256 - 11); + + for (i = 0; i < 3; i++) + p->NS2Indx[i] = (Byte)i; + for (m = i, k = 1; i < 256; i++) + { + p->NS2Indx[i] = (Byte)m; + if (--k == 0) + k = (++m) - 2; + } + + memset(p->HB2Flag, 0, 0x40); + memset(p->HB2Flag + 0x40, 8, 0x100 - 0x40); +} + +static void Ppmd7_Free(CPpmd7 *p) +{ + free(p->Base); + p->Size = 0; + p->Base = 0; +} + +static Bool Ppmd7_Alloc(CPpmd7 *p, UInt32 size) +{ + if (p->Base == 0 || p->Size != size) + { + /* RestartModel() below assumes that p->Size >= UNIT_SIZE + (see the calculation of m->MinContext). */ + if (size < UNIT_SIZE) { + return False; + } + Ppmd7_Free(p); + p->AlignOffset = + #ifdef PPMD_32BIT + (4 - size) & 3; + #else + 4 - (size & 3); + #endif + if ((p->Base = (Byte *)malloc(p->AlignOffset + size + #ifndef PPMD_32BIT + + UNIT_SIZE + #endif + )) == 0) + return False; + p->Size = size; + } + return True; +} + +static void InsertNode(CPpmd7 *p, void *node, unsigned indx) +{ + *((CPpmd_Void_Ref *)node) = p->FreeList[indx]; + p->FreeList[indx] = REF(node); +} + +static void *RemoveNode(CPpmd7 *p, unsigned indx) +{ + CPpmd_Void_Ref *node = (CPpmd_Void_Ref *)Ppmd7_GetPtr(p, p->FreeList[indx]); + p->FreeList[indx] = *node; + return node; +} + +static void SplitBlock(CPpmd7 *p, void *ptr, unsigned oldIndx, unsigned newIndx) +{ + unsigned i, nu = I2U(oldIndx) - I2U(newIndx); + ptr = (Byte *)ptr + U2B(I2U(newIndx)); + if (I2U(i = U2I(nu)) != nu) + { + unsigned k = I2U(--i); + InsertNode(p, ((Byte *)ptr) + U2B(k), nu - k - 1); + } + InsertNode(p, ptr, i); +} + +static void GlueFreeBlocks(CPpmd7 *p) +{ + #ifdef PPMD_32BIT + CPpmd7_Node headItem; + CPpmd7_Node_Ref head = &headItem; + #else + CPpmd7_Node_Ref head = p->AlignOffset + p->Size; + #endif + + CPpmd7_Node_Ref n = head; + unsigned i; + + p->GlueCount = 255; + + /* create doubly-linked list of free blocks */ + for (i = 0; i < PPMD_NUM_INDEXES; i++) + { + UInt16 nu = I2U(i); + CPpmd7_Node_Ref next = (CPpmd7_Node_Ref)p->FreeList[i]; + p->FreeList[i] = 0; + while (next != 0) + { + CPpmd7_Node *node = NODE(next); + node->Next = n; + n = NODE(n)->Prev = next; + next = *(const CPpmd7_Node_Ref *)node; + node->Stamp = 0; + node->NU = (UInt16)nu; + } + } + NODE(head)->Stamp = 1; + NODE(head)->Next = n; + NODE(n)->Prev = head; + if (p->LoUnit != p->HiUnit) + ((CPpmd7_Node *)p->LoUnit)->Stamp = 1; + + /* Glue free blocks */ + while (n != head) + { + CPpmd7_Node *node = NODE(n); + UInt32 nu = (UInt32)node->NU; + for (;;) + { + CPpmd7_Node *node2 = NODE(n) + nu; + nu += node2->NU; + if (node2->Stamp != 0 || nu >= 0x10000) + break; + NODE(node2->Prev)->Next = node2->Next; + NODE(node2->Next)->Prev = node2->Prev; + node->NU = (UInt16)nu; + } + n = node->Next; + } + + /* Fill lists of free blocks */ + for (n = NODE(head)->Next; n != head;) + { + CPpmd7_Node *node = NODE(n); + unsigned nu; + CPpmd7_Node_Ref next = node->Next; + for (nu = node->NU; nu > 128; nu -= 128, node += 128) + InsertNode(p, node, PPMD_NUM_INDEXES - 1); + if (I2U(i = U2I(nu)) != nu) + { + unsigned k = I2U(--i); + InsertNode(p, node + k, nu - k - 1); + } + InsertNode(p, node, i); + n = next; + } +} + +static void *AllocUnitsRare(CPpmd7 *p, unsigned indx) +{ + unsigned i; + void *retVal; + if (p->GlueCount == 0) + { + GlueFreeBlocks(p); + if (p->FreeList[indx] != 0) + return RemoveNode(p, indx); + } + i = indx; + do + { + if (++i == PPMD_NUM_INDEXES) + { + UInt32 numBytes = U2B(I2U(indx)); + p->GlueCount--; + return ((UInt32)(p->UnitsStart - p->Text) > numBytes) ? (p->UnitsStart -= numBytes) : (NULL); + } + } + while (p->FreeList[i] == 0); + retVal = RemoveNode(p, i); + SplitBlock(p, retVal, i, indx); + return retVal; +} + +static void *AllocUnits(CPpmd7 *p, unsigned indx) +{ + UInt32 numBytes; + if (p->FreeList[indx] != 0) + return RemoveNode(p, indx); + numBytes = U2B(I2U(indx)); + if (numBytes <= (UInt32)(p->HiUnit - p->LoUnit)) + { + void *retVal = p->LoUnit; + p->LoUnit += numBytes; + return retVal; + } + return AllocUnitsRare(p, indx); +} + +#define MyMem12Cpy(dest, src, num) \ + { UInt32 *d = (UInt32 *)dest; const UInt32 *s = (const UInt32 *)src; UInt32 n = num; \ + do { d[0] = s[0]; d[1] = s[1]; d[2] = s[2]; s += 3; d += 3; } while(--n); } + +static void *ShrinkUnits(CPpmd7 *p, void *oldPtr, unsigned oldNU, unsigned newNU) +{ + unsigned i0 = U2I(oldNU); + unsigned i1 = U2I(newNU); + if (i0 == i1) + return oldPtr; + if (p->FreeList[i1] != 0) + { + void *ptr = RemoveNode(p, i1); + MyMem12Cpy(ptr, oldPtr, newNU); + InsertNode(p, oldPtr, i0); + return ptr; + } + SplitBlock(p, oldPtr, i0, i1); + return oldPtr; +} + +#define SUCCESSOR(p) ((CPpmd_Void_Ref)((p)->SuccessorLow | ((UInt32)(p)->SuccessorHigh << 16))) + +static void SetSuccessor(CPpmd_State *p, CPpmd_Void_Ref v) +{ + (p)->SuccessorLow = (UInt16)((UInt32)(v) & 0xFFFF); + (p)->SuccessorHigh = (UInt16)(((UInt32)(v) >> 16) & 0xFFFF); +} + +static void RestartModel(CPpmd7 *p) +{ + unsigned i, k, m; + + memset(p->FreeList, 0, sizeof(p->FreeList)); + p->Text = p->Base + p->AlignOffset; + p->HiUnit = p->Text + p->Size; + p->LoUnit = p->UnitsStart = p->HiUnit - p->Size / 8 / UNIT_SIZE * 7 * UNIT_SIZE; + p->GlueCount = 0; + + p->OrderFall = p->MaxOrder; + p->RunLength = p->InitRL = -(Int32)((p->MaxOrder < 12) ? p->MaxOrder : 12) - 1; + p->PrevSuccess = 0; + + p->MinContext = p->MaxContext = (CTX_PTR)(p->HiUnit -= UNIT_SIZE); /* AllocContext(p); */ + p->MinContext->Suffix = 0; + p->MinContext->NumStats = 256; + p->MinContext->SummFreq = 256 + 1; + p->FoundState = (CPpmd_State *)p->LoUnit; /* AllocUnits(p, PPMD_NUM_INDEXES - 1); */ + p->LoUnit += U2B(256 / 2); + p->MinContext->Stats = REF(p->FoundState); + for (i = 0; i < 256; i++) + { + CPpmd_State *s = &p->FoundState[i]; + s->Symbol = (Byte)i; + s->Freq = 1; + SetSuccessor(s, 0); + } + + for (i = 0; i < 128; i++) + for (k = 0; k < 8; k++) + { + UInt16 *dest = p->BinSumm[i] + k; + UInt16 val = (UInt16)(PPMD_BIN_SCALE - kInitBinEsc[k] / (i + 2)); + for (m = 0; m < 64; m += 8) + dest[m] = val; + } + + for (i = 0; i < 25; i++) + for (k = 0; k < 16; k++) + { + CPpmd_See *s = &p->See[i][k]; + s->Summ = (UInt16)((5 * i + 10) << (s->Shift = PPMD_PERIOD_BITS - 4)); + s->Count = 4; + } +} + +static void Ppmd7_Init(CPpmd7 *p, unsigned maxOrder) +{ + p->MaxOrder = maxOrder; + RestartModel(p); + p->DummySee.Shift = PPMD_PERIOD_BITS; + p->DummySee.Summ = 0; /* unused */ + p->DummySee.Count = 64; /* unused */ +} + +static CTX_PTR CreateSuccessors(CPpmd7 *p, Bool skip) +{ + CPpmd_State upState; + CTX_PTR c = p->MinContext; + CPpmd_Byte_Ref upBranch = (CPpmd_Byte_Ref)SUCCESSOR(p->FoundState); + CPpmd_State *ps[PPMD7_MAX_ORDER]; + unsigned numPs = 0; + + if (!skip) + ps[numPs++] = p->FoundState; + + while (c->Suffix) + { + CPpmd_Void_Ref successor; + CPpmd_State *s; + c = SUFFIX(c); + if (c->NumStats != 1) + { + for (s = STATS(c); s->Symbol != p->FoundState->Symbol; s++); + } + else + s = ONE_STATE(c); + successor = SUCCESSOR(s); + if (successor != upBranch) + { + c = CTX(successor); + if (numPs == 0) + return c; + break; + } + ps[numPs++] = s; + } + + upState.Symbol = *(const Byte *)Ppmd7_GetPtr(p, upBranch); + SetSuccessor(&upState, upBranch + 1); + + if (c->NumStats == 1) + upState.Freq = ONE_STATE(c)->Freq; + else + { + UInt32 cf, s0; + CPpmd_State *s; + for (s = STATS(c); s->Symbol != upState.Symbol; s++); + cf = s->Freq - 1; + s0 = c->SummFreq - c->NumStats - cf; + upState.Freq = (Byte)(1 + ((2 * cf <= s0) ? (5 * cf > s0) : ((2 * cf + 3 * s0 - 1) / (2 * s0)))); + } + + while (numPs != 0) + { + /* Create Child */ + CTX_PTR c1; /* = AllocContext(p); */ + if (p->HiUnit != p->LoUnit) + c1 = (CTX_PTR)(p->HiUnit -= UNIT_SIZE); + else if (p->FreeList[0] != 0) + c1 = (CTX_PTR)RemoveNode(p, 0); + else + { + c1 = (CTX_PTR)AllocUnitsRare(p, 0); + if (!c1) + return NULL; + } + c1->NumStats = 1; + *ONE_STATE(c1) = upState; + c1->Suffix = REF(c); + SetSuccessor(ps[--numPs], REF(c1)); + c = c1; + } + + return c; +} + +static void SwapStates(CPpmd_State *t1, CPpmd_State *t2) +{ + CPpmd_State tmp = *t1; + *t1 = *t2; + *t2 = tmp; +} + +static void UpdateModel(CPpmd7 *p) +{ + CPpmd_Void_Ref successor, fSuccessor = SUCCESSOR(p->FoundState); + CTX_PTR c; + unsigned s0, ns; + + if (p->FoundState->Freq < MAX_FREQ / 4 && p->MinContext->Suffix != 0) + { + c = SUFFIX(p->MinContext); + + if (c->NumStats == 1) + { + CPpmd_State *s = ONE_STATE(c); + if (s->Freq < 32) + s->Freq++; + } + else + { + CPpmd_State *s = STATS(c); + if (s->Symbol != p->FoundState->Symbol) + { + do { s++; } while (s->Symbol != p->FoundState->Symbol); + if (s[0].Freq >= s[-1].Freq) + { + SwapStates(&s[0], &s[-1]); + s--; + } + } + if (s->Freq < MAX_FREQ - 9) + { + s->Freq += 2; + c->SummFreq += 2; + } + } + } + + if (p->OrderFall == 0) + { + p->MinContext = p->MaxContext = CreateSuccessors(p, True); + if (p->MinContext == 0) + { + RestartModel(p); + return; + } + SetSuccessor(p->FoundState, REF(p->MinContext)); + return; + } + + *p->Text++ = p->FoundState->Symbol; + successor = REF(p->Text); + if (p->Text >= p->UnitsStart) + { + RestartModel(p); + return; + } + + if (fSuccessor) + { + if (fSuccessor <= successor) + { + CTX_PTR cs = CreateSuccessors(p, False); + if (cs == NULL) + { + RestartModel(p); + return; + } + fSuccessor = REF(cs); + } + if (--p->OrderFall == 0) + { + successor = fSuccessor; + p->Text -= (p->MaxContext != p->MinContext); + } + } + else + { + SetSuccessor(p->FoundState, successor); + fSuccessor = REF(p->MinContext); + } + + s0 = p->MinContext->SummFreq - (ns = p->MinContext->NumStats) - (p->FoundState->Freq - 1); + + for (c = p->MaxContext; c != p->MinContext; c = SUFFIX(c)) + { + unsigned ns1; + UInt32 cf, sf; + if ((ns1 = c->NumStats) != 1) + { + if ((ns1 & 1) == 0) + { + /* Expand for one UNIT */ + unsigned oldNU = ns1 >> 1; + unsigned i = U2I(oldNU); + if (i != U2I(oldNU + 1)) + { + void *ptr = AllocUnits(p, i + 1); + void *oldPtr; + if (!ptr) + { + RestartModel(p); + return; + } + oldPtr = STATS(c); + MyMem12Cpy(ptr, oldPtr, oldNU); + InsertNode(p, oldPtr, i); + c->Stats = STATS_REF(ptr); + } + } + c->SummFreq = (UInt16)(c->SummFreq + (2 * ns1 < ns) + 2 * ((4 * ns1 <= ns) & (c->SummFreq <= 8 * ns1))); + } + else + { + CPpmd_State *s = (CPpmd_State*)AllocUnits(p, 0); + if (!s) + { + RestartModel(p); + return; + } + *s = *ONE_STATE(c); + c->Stats = REF(s); + if (s->Freq < MAX_FREQ / 4 - 1) + s->Freq <<= 1; + else + s->Freq = MAX_FREQ - 4; + c->SummFreq = (UInt16)(s->Freq + p->InitEsc + (ns > 3)); + } + cf = 2 * (UInt32)p->FoundState->Freq * (c->SummFreq + 6); + sf = (UInt32)s0 + c->SummFreq; + if (cf < 6 * sf) + { + cf = 1 + (cf > sf) + (cf >= 4 * sf); + c->SummFreq += 3; + } + else + { + cf = 4 + (cf >= 9 * sf) + (cf >= 12 * sf) + (cf >= 15 * sf); + c->SummFreq = (UInt16)(c->SummFreq + cf); + } + { + CPpmd_State *s = STATS(c) + ns1; + SetSuccessor(s, successor); + s->Symbol = p->FoundState->Symbol; + s->Freq = (Byte)cf; + c->NumStats = (UInt16)(ns1 + 1); + } + } + p->MaxContext = p->MinContext = CTX(fSuccessor); +} + +static void Rescale(CPpmd7 *p) +{ + unsigned i, adder, sumFreq, escFreq; + CPpmd_State *stats = STATS(p->MinContext); + CPpmd_State *s = p->FoundState; + { + CPpmd_State tmp = *s; + for (; s != stats; s--) + s[0] = s[-1]; + *s = tmp; + } + escFreq = p->MinContext->SummFreq - s->Freq; + s->Freq += 4; + adder = (p->OrderFall != 0); + s->Freq = (Byte)((s->Freq + adder) >> 1); + sumFreq = s->Freq; + + i = p->MinContext->NumStats - 1; + do + { + escFreq -= (++s)->Freq; + s->Freq = (Byte)((s->Freq + adder) >> 1); + sumFreq += s->Freq; + if (s[0].Freq > s[-1].Freq) + { + CPpmd_State *s1 = s; + CPpmd_State tmp = *s1; + do + s1[0] = s1[-1]; + while (--s1 != stats && tmp.Freq > s1[-1].Freq); + *s1 = tmp; + } + } + while (--i); + + if (s->Freq == 0) + { + unsigned numStats = p->MinContext->NumStats; + unsigned n0, n1; + do { i++; } while ((--s)->Freq == 0); + escFreq += i; + p->MinContext->NumStats = (UInt16)(p->MinContext->NumStats - i); + if (p->MinContext->NumStats == 1) + { + CPpmd_State tmp = *stats; + do + { + tmp.Freq = (Byte)(tmp.Freq - (tmp.Freq >> 1)); + escFreq >>= 1; + } + while (escFreq > 1); + InsertNode(p, stats, U2I(((numStats + 1) >> 1))); + *(p->FoundState = ONE_STATE(p->MinContext)) = tmp; + return; + } + n0 = (numStats + 1) >> 1; + n1 = (p->MinContext->NumStats + 1) >> 1; + if (n0 != n1) + p->MinContext->Stats = STATS_REF(ShrinkUnits(p, stats, n0, n1)); + } + p->MinContext->SummFreq = (UInt16)(sumFreq + escFreq - (escFreq >> 1)); + p->FoundState = STATS(p->MinContext); +} + +static CPpmd_See *Ppmd7_MakeEscFreq(CPpmd7 *p, unsigned numMasked, UInt32 *escFreq) +{ + CPpmd_See *see; + unsigned nonMasked = p->MinContext->NumStats - numMasked; + if (p->MinContext->NumStats != 256) + { + see = p->See[p->NS2Indx[nonMasked - 1]] + + (nonMasked < (unsigned)SUFFIX(p->MinContext)->NumStats - p->MinContext->NumStats) + + 2 * (p->MinContext->SummFreq < 11 * p->MinContext->NumStats) + + 4 * (numMasked > nonMasked) + + p->HiBitsFlag; + { + unsigned r = (see->Summ >> see->Shift); + see->Summ = (UInt16)(see->Summ - r); + *escFreq = r + (r == 0); + } + } + else + { + see = &p->DummySee; + *escFreq = 1; + } + return see; +} + +static void NextContext(CPpmd7 *p) +{ + CTX_PTR c = CTX(SUCCESSOR(p->FoundState)); + if (p->OrderFall == 0 && (Byte *)c > p->Text) + p->MinContext = p->MaxContext = c; + else + UpdateModel(p); +} + +static void Ppmd7_Update1(CPpmd7 *p) +{ + CPpmd_State *s = p->FoundState; + s->Freq += 4; + p->MinContext->SummFreq += 4; + if (s[0].Freq > s[-1].Freq) + { + SwapStates(&s[0], &s[-1]); + p->FoundState = --s; + if (s->Freq > MAX_FREQ) + Rescale(p); + } + NextContext(p); +} + +static void Ppmd7_Update1_0(CPpmd7 *p) +{ + p->PrevSuccess = (2 * p->FoundState->Freq > p->MinContext->SummFreq); + p->RunLength += p->PrevSuccess; + p->MinContext->SummFreq += 4; + if ((p->FoundState->Freq += 4) > MAX_FREQ) + Rescale(p); + NextContext(p); +} + +static void Ppmd7_UpdateBin(CPpmd7 *p) +{ + p->FoundState->Freq = (Byte)(p->FoundState->Freq + (p->FoundState->Freq < 128 ? 1: 0)); + p->PrevSuccess = 1; + p->RunLength++; + NextContext(p); +} + +static void Ppmd7_Update2(CPpmd7 *p) +{ + p->MinContext->SummFreq += 4; + if ((p->FoundState->Freq += 4) > MAX_FREQ) + Rescale(p); + p->RunLength = p->InitRL; + UpdateModel(p); +} + +/* ---------- Decode ---------- */ + +static Bool Ppmd_RangeDec_Init(CPpmd7z_RangeDec *p) +{ + unsigned i; + p->Low = p->Bottom = 0; + p->Range = 0xFFFFFFFF; + for (i = 0; i < 4; i++) + p->Code = (p->Code << 8) | p->Stream->Read((void *)p->Stream); + return (p->Code < 0xFFFFFFFF); +} + +static Bool Ppmd7z_RangeDec_Init(CPpmd7z_RangeDec *p) +{ + if (p->Stream->Read((void *)p->Stream) != 0) + return False; + return Ppmd_RangeDec_Init(p); +} + +static Bool PpmdRAR_RangeDec_Init(CPpmd7z_RangeDec *p) +{ + if (!Ppmd_RangeDec_Init(p)) + return False; + p->Bottom = 0x8000; + return True; +} + +static UInt32 Range_GetThreshold(void *pp, UInt32 total) +{ + CPpmd7z_RangeDec *p = (CPpmd7z_RangeDec *)pp; + return (p->Code - p->Low) / (p->Range /= total); +} + +static void Range_Normalize(CPpmd7z_RangeDec *p) +{ + while (1) + { + if((p->Low ^ (p->Low + p->Range)) >= kTopValue) + { + if(p->Range >= p->Bottom) + break; + else + p->Range = ((uint32_t)(-(int32_t)p->Low)) & (p->Bottom - 1); + } + p->Code = (p->Code << 8) | p->Stream->Read((void *)p->Stream); + p->Range <<= 8; + p->Low <<= 8; + } +} + +static void Range_Decode_7z(void *pp, UInt32 start, UInt32 size) +{ + CPpmd7z_RangeDec *p = (CPpmd7z_RangeDec *)pp; + p->Code -= start * p->Range; + p->Range *= size; + Range_Normalize(p); +} + +static void Range_Decode_RAR(void *pp, UInt32 start, UInt32 size) +{ + CPpmd7z_RangeDec *p = (CPpmd7z_RangeDec *)pp; + p->Low += start * p->Range; + p->Range *= size; + Range_Normalize(p); +} + +static UInt32 Range_DecodeBit_7z(void *pp, UInt32 size0) +{ + CPpmd7z_RangeDec *p = (CPpmd7z_RangeDec *)pp; + UInt32 newBound = (p->Range >> 14) * size0; + UInt32 symbol; + if (p->Code < newBound) + { + symbol = 0; + p->Range = newBound; + } + else + { + symbol = 1; + p->Code -= newBound; + p->Range -= newBound; + } + Range_Normalize(p); + return symbol; +} + +static UInt32 Range_DecodeBit_RAR(void *pp, UInt32 size0) +{ + CPpmd7z_RangeDec *p = (CPpmd7z_RangeDec *)pp; + UInt32 bit, value = p->p.GetThreshold(p, PPMD_BIN_SCALE); + if(value < size0) + { + bit = 0; + p->p.Decode(p, 0, size0); + } + else + { + bit = 1; + p->p.Decode(p, size0, PPMD_BIN_SCALE - size0); + } + return bit; +} + +static void Ppmd7z_RangeDec_CreateVTable(CPpmd7z_RangeDec *p) +{ + p->p.GetThreshold = Range_GetThreshold; + p->p.Decode = Range_Decode_7z; + p->p.DecodeBit = Range_DecodeBit_7z; +} + +static void PpmdRAR_RangeDec_CreateVTable(CPpmd7z_RangeDec *p) +{ + p->p.GetThreshold = Range_GetThreshold; + p->p.Decode = Range_Decode_RAR; + p->p.DecodeBit = Range_DecodeBit_RAR; +} + +#define MASK(sym) ((signed char *)charMask)[sym] + +static int Ppmd7_DecodeSymbol(CPpmd7 *p, IPpmd7_RangeDec *rc) +{ + size_t charMask[256 / sizeof(size_t)]; + if (p->MinContext->NumStats != 1) + { + CPpmd_State *s = Ppmd7_GetStats(p, p->MinContext); + unsigned i; + UInt32 count, hiCnt; + if ((count = rc->GetThreshold(rc, p->MinContext->SummFreq)) < (hiCnt = s->Freq)) + { + Byte symbol; + rc->Decode(rc, 0, s->Freq); + p->FoundState = s; + symbol = s->Symbol; + Ppmd7_Update1_0(p); + return symbol; + } + p->PrevSuccess = 0; + i = p->MinContext->NumStats - 1; + do + { + if ((hiCnt += (++s)->Freq) > count) + { + Byte symbol; + rc->Decode(rc, hiCnt - s->Freq, s->Freq); + p->FoundState = s; + symbol = s->Symbol; + Ppmd7_Update1(p); + return symbol; + } + } + while (--i); + if (count >= p->MinContext->SummFreq) + return -2; + p->HiBitsFlag = p->HB2Flag[p->FoundState->Symbol]; + rc->Decode(rc, hiCnt, p->MinContext->SummFreq - hiCnt); + PPMD_SetAllBitsIn256Bytes(charMask); + MASK(s->Symbol) = 0; + i = p->MinContext->NumStats - 1; + do { MASK((--s)->Symbol) = 0; } while (--i); + } + else + { + UInt16 *prob = Ppmd7_GetBinSumm(p); + if (rc->DecodeBit(rc, *prob) == 0) + { + Byte symbol; + *prob = (UInt16)PPMD_UPDATE_PROB_0(*prob); + symbol = (p->FoundState = Ppmd7Context_OneState(p->MinContext))->Symbol; + Ppmd7_UpdateBin(p); + return symbol; + } + *prob = (UInt16)PPMD_UPDATE_PROB_1(*prob); + p->InitEsc = PPMD7_kExpEscape[*prob >> 10]; + PPMD_SetAllBitsIn256Bytes(charMask); + MASK(Ppmd7Context_OneState(p->MinContext)->Symbol) = 0; + p->PrevSuccess = 0; + } + for (;;) + { + CPpmd_State *ps[256], *s; + UInt32 freqSum, count, hiCnt; + CPpmd_See *see; + unsigned i, num, numMasked = p->MinContext->NumStats; + do + { + p->OrderFall++; + if (!p->MinContext->Suffix) + return -1; + p->MinContext = Ppmd7_GetContext(p, p->MinContext->Suffix); + } + while (p->MinContext->NumStats == numMasked); + hiCnt = 0; + s = Ppmd7_GetStats(p, p->MinContext); + i = 0; + num = p->MinContext->NumStats - numMasked; + do + { + int k = (int)(MASK(s->Symbol)); + hiCnt += (s->Freq & k); + ps[i] = s++; + i -= k; + } + while (i != num); + + see = Ppmd7_MakeEscFreq(p, numMasked, &freqSum); + freqSum += hiCnt; + count = rc->GetThreshold(rc, freqSum); + + if (count < hiCnt) + { + Byte symbol; + CPpmd_State **pps = ps; + for (hiCnt = 0; (hiCnt += (*pps)->Freq) <= count; pps++); + s = *pps; + rc->Decode(rc, hiCnt - s->Freq, s->Freq); + Ppmd_See_Update(see); + p->FoundState = s; + symbol = s->Symbol; + Ppmd7_Update2(p); + return symbol; + } + if (count >= freqSum) + return -2; + rc->Decode(rc, hiCnt, freqSum - hiCnt); + see->Summ = (UInt16)(see->Summ + freqSum); + do { MASK(ps[--i]->Symbol) = 0; } while (i != 0); + } +} + +/* ---------- Encode ---------- Ppmd7Enc.c */ + +#define kTopValue (1 << 24) + +static void Ppmd7z_RangeEnc_Init(CPpmd7z_RangeEnc *p) +{ + p->Low = 0; + p->Range = 0xFFFFFFFF; + p->Cache = 0; + p->CacheSize = 1; +} + +static void RangeEnc_ShiftLow(CPpmd7z_RangeEnc *p) +{ + if ((UInt32)p->Low < (UInt32)0xFF000000 || (unsigned)(p->Low >> 32) != 0) + { + Byte temp = p->Cache; + do + { + p->Stream->Write(p->Stream, (Byte)(temp + (Byte)(p->Low >> 32))); + temp = 0xFF; + } + while(--p->CacheSize != 0); + p->Cache = (Byte)((UInt32)p->Low >> 24); + } + p->CacheSize++; + p->Low = ((UInt32)p->Low << 8) & 0xFFFFFFFF; +} + +static void RangeEnc_Encode(CPpmd7z_RangeEnc *p, UInt32 start, UInt32 size, UInt32 total) +{ + p->Low += (UInt64)start * (UInt64)(p->Range /= total); + p->Range *= size; + while (p->Range < kTopValue) + { + p->Range <<= 8; + RangeEnc_ShiftLow(p); + } +} + +static void RangeEnc_EncodeBit_0(CPpmd7z_RangeEnc *p, UInt32 size0) +{ + p->Range = (p->Range >> 14) * size0; + while (p->Range < kTopValue) + { + p->Range <<= 8; + RangeEnc_ShiftLow(p); + } +} + +static void RangeEnc_EncodeBit_1(CPpmd7z_RangeEnc *p, UInt32 size0) +{ + UInt32 newBound = (p->Range >> 14) * size0; + p->Low += newBound; + p->Range -= newBound; + while (p->Range < kTopValue) + { + p->Range <<= 8; + RangeEnc_ShiftLow(p); + } +} + +static void Ppmd7z_RangeEnc_FlushData(CPpmd7z_RangeEnc *p) +{ + unsigned i; + for (i = 0; i < 5; i++) + RangeEnc_ShiftLow(p); +} + + +#define MASK(sym) ((signed char *)charMask)[sym] + +static void Ppmd7_EncodeSymbol(CPpmd7 *p, CPpmd7z_RangeEnc *rc, int symbol) +{ + size_t charMask[256 / sizeof(size_t)]; + if (p->MinContext->NumStats != 1) + { + CPpmd_State *s = Ppmd7_GetStats(p, p->MinContext); + UInt32 sum; + unsigned i; + if (s->Symbol == symbol) + { + RangeEnc_Encode(rc, 0, s->Freq, p->MinContext->SummFreq); + p->FoundState = s; + Ppmd7_Update1_0(p); + return; + } + p->PrevSuccess = 0; + sum = s->Freq; + i = p->MinContext->NumStats - 1; + do + { + if ((++s)->Symbol == symbol) + { + RangeEnc_Encode(rc, sum, s->Freq, p->MinContext->SummFreq); + p->FoundState = s; + Ppmd7_Update1(p); + return; + } + sum += s->Freq; + } + while (--i); + + p->HiBitsFlag = p->HB2Flag[p->FoundState->Symbol]; + PPMD_SetAllBitsIn256Bytes(charMask); + MASK(s->Symbol) = 0; + i = p->MinContext->NumStats - 1; + do { MASK((--s)->Symbol) = 0; } while (--i); + RangeEnc_Encode(rc, sum, p->MinContext->SummFreq - sum, p->MinContext->SummFreq); + } + else + { + UInt16 *prob = Ppmd7_GetBinSumm(p); + CPpmd_State *s = Ppmd7Context_OneState(p->MinContext); + if (s->Symbol == symbol) + { + RangeEnc_EncodeBit_0(rc, *prob); + *prob = (UInt16)PPMD_UPDATE_PROB_0(*prob); + p->FoundState = s; + Ppmd7_UpdateBin(p); + return; + } + else + { + RangeEnc_EncodeBit_1(rc, *prob); + *prob = (UInt16)PPMD_UPDATE_PROB_1(*prob); + p->InitEsc = PPMD7_kExpEscape[*prob >> 10]; + PPMD_SetAllBitsIn256Bytes(charMask); + MASK(s->Symbol) = 0; + p->PrevSuccess = 0; + } + } + for (;;) + { + UInt32 escFreq; + CPpmd_See *see; + CPpmd_State *s; + UInt32 sum; + unsigned i, numMasked = p->MinContext->NumStats; + do + { + p->OrderFall++; + if (!p->MinContext->Suffix) + return; /* EndMarker (symbol = -1) */ + p->MinContext = Ppmd7_GetContext(p, p->MinContext->Suffix); + } + while (p->MinContext->NumStats == numMasked); + + see = Ppmd7_MakeEscFreq(p, numMasked, &escFreq); + s = Ppmd7_GetStats(p, p->MinContext); + sum = 0; + i = p->MinContext->NumStats; + do + { + int cur = s->Symbol; + if (cur == symbol) + { + UInt32 low = sum; + CPpmd_State *s1 = s; + do + { + sum += (s->Freq & (int)(MASK(s->Symbol))); + s++; + } + while (--i); + RangeEnc_Encode(rc, low, s1->Freq, sum + escFreq); + Ppmd_See_Update(see); + p->FoundState = s1; + Ppmd7_Update2(p); + return; + } + sum += (s->Freq & (int)(MASK(cur))); + MASK(cur) = 0; + s++; + } + while (--i); + + RangeEnc_Encode(rc, sum, escFreq, sum + escFreq); + see->Summ = (UInt16)(see->Summ + sum + escFreq); + } +} + +const IPpmd7 __archive_ppmd7_functions = +{ + &Ppmd7_Construct, + &Ppmd7_Alloc, + &Ppmd7_Free, + &Ppmd7_Init, + &Ppmd7z_RangeDec_CreateVTable, + &PpmdRAR_RangeDec_CreateVTable, + &Ppmd7z_RangeDec_Init, + &PpmdRAR_RangeDec_Init, + &Ppmd7_DecodeSymbol, + &Ppmd7z_RangeEnc_Init, + &Ppmd7z_RangeEnc_FlushData, + &Ppmd7_EncodeSymbol +}; diff --git a/c/archive_ppmd7_private.h b/c/archive_ppmd7_private.h new file mode 100644 --- /dev/null +++ b/c/archive_ppmd7_private.h @@ -0,0 +1,119 @@ +/* Ppmd7.h -- PPMdH compression codec +2010-03-12 : Igor Pavlov : Public domain +This code is based on PPMd var.H (2001): Dmitry Shkarin : Public domain */ + +/* This code supports virtual RangeDecoder and includes the implementation +of RangeCoder from 7z, instead of RangeCoder from original PPMd var.H. +If you need the compatibility with original PPMd var.H, you can use external RangeDecoder */ + +#ifndef ARCHIVE_PPMD7_PRIVATE_H_INCLUDED +#define ARCHIVE_PPMD7_PRIVATE_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif + +#include "archive_ppmd_private.h" + +#define PPMD7_MIN_ORDER 2 +#define PPMD7_MAX_ORDER 64 + +#define PPMD7_MIN_MEM_SIZE (1 << 11) +#define PPMD7_MAX_MEM_SIZE (0xFFFFFFFFu - 12 * 3) + +struct CPpmd7_Context_; + +typedef + #ifdef PPMD_32BIT + struct CPpmd7_Context_ * + #else + UInt32 + #endif + CPpmd7_Context_Ref; + +typedef struct CPpmd7_Context_ +{ + UInt16 NumStats; + UInt16 SummFreq; + CPpmd_State_Ref Stats; + CPpmd7_Context_Ref Suffix; +} CPpmd7_Context; + +#define Ppmd7Context_OneState(p) ((CPpmd_State *)&(p)->SummFreq) + +typedef struct +{ + CPpmd7_Context *MinContext, *MaxContext; + CPpmd_State *FoundState; + unsigned OrderFall, InitEsc, PrevSuccess, MaxOrder, HiBitsFlag; + Int32 RunLength, InitRL; /* must be 32-bit at least */ + + UInt32 Size; + UInt32 GlueCount; + Byte *Base, *LoUnit, *HiUnit, *Text, *UnitsStart; + UInt32 AlignOffset; + + Byte Indx2Units[PPMD_NUM_INDEXES]; + Byte Units2Indx[128]; + CPpmd_Void_Ref FreeList[PPMD_NUM_INDEXES]; + Byte NS2Indx[256], NS2BSIndx[256], HB2Flag[256]; + CPpmd_See DummySee, See[25][16]; + UInt16 BinSumm[128][64]; +} CPpmd7; + +/* ---------- Decode ---------- */ + +typedef struct +{ + UInt32 (*GetThreshold)(void *p, UInt32 total); + void (*Decode)(void *p, UInt32 start, UInt32 size); + UInt32 (*DecodeBit)(void *p, UInt32 size0); +} IPpmd7_RangeDec; + +typedef struct +{ + IPpmd7_RangeDec p; + UInt32 Range; + UInt32 Code; + UInt32 Low; + UInt32 Bottom; + IByteIn *Stream; +} CPpmd7z_RangeDec; + +/* ---------- Encode ---------- */ + +typedef struct +{ + UInt64 Low; + UInt32 Range; + Byte Cache; + UInt64 CacheSize; + IByteOut *Stream; +} CPpmd7z_RangeEnc; + +typedef struct +{ + /* Base Functions */ + void (*Ppmd7_Construct)(CPpmd7 *p); + Bool (*Ppmd7_Alloc)(CPpmd7 *p, UInt32 size); + void (*Ppmd7_Free)(CPpmd7 *p); + void (*Ppmd7_Init)(CPpmd7 *p, unsigned maxOrder); + #define Ppmd7_WasAllocated(p) ((p)->Base != NULL) + + /* Decode Functions */ + void (*Ppmd7z_RangeDec_CreateVTable)(CPpmd7z_RangeDec *p); + void (*PpmdRAR_RangeDec_CreateVTable)(CPpmd7z_RangeDec *p); + Bool (*Ppmd7z_RangeDec_Init)(CPpmd7z_RangeDec *p); + Bool (*PpmdRAR_RangeDec_Init)(CPpmd7z_RangeDec *p); + #define Ppmd7z_RangeDec_IsFinishedOK(p) ((p)->Code == 0) + int (*Ppmd7_DecodeSymbol)(CPpmd7 *p, IPpmd7_RangeDec *rc); + + /* Encode Functions */ + void (*Ppmd7z_RangeEnc_Init)(CPpmd7z_RangeEnc *p); + void (*Ppmd7z_RangeEnc_FlushData)(CPpmd7z_RangeEnc *p); + + void (*Ppmd7_EncodeSymbol)(CPpmd7 *p, CPpmd7z_RangeEnc *rc, int symbol); +} IPpmd7; + +extern const IPpmd7 __archive_ppmd7_functions; +#endif diff --git a/c/archive_ppmd8.c b/c/archive_ppmd8.c new file mode 100644 --- /dev/null +++ b/c/archive_ppmd8.c @@ -0,0 +1,1287 @@ +/* Ppmd8.c -- PPMdI codec +2016-05-21 : Igor Pavlov : Public domain +This code is based on PPMd var.I (2002): Dmitry Shkarin : Public domain */ + +#include "archive_platform.h" + +#include + +#include "archive_ppmd8_private.h" + +const Byte PPMD8_kExpEscape[16] = { 25, 14, 9, 7, 5, 5, 4, 4, 4, 3, 3, 3, 2, 2, 2, 2 }; +static const UInt16 kInitBinEsc[] = { 0x3CDD, 0x1F3F, 0x59BF, 0x48F3, 0x64A1, 0x5ABC, 0x6632, 0x6051}; + +#define MAX_FREQ 124 +#define UNIT_SIZE 12 + +#define U2B(nu) ((UInt32)(nu) * UNIT_SIZE) +#define U2I(nu) (p->Units2Indx[(nu) - 1]) +#define I2U(indx) (p->Indx2Units[indx]) + +#ifdef PPMD_32BIT + #define REF(ptr) (ptr) +#else + #define REF(ptr) ((UInt32)((Byte *)(ptr) - (p)->Base)) +#endif + +#define STATS_REF(ptr) ((CPpmd_State_Ref)REF(ptr)) + +#define CTX(ref) ((CPpmd8_Context *)Ppmd8_GetContext(p, ref)) +#define STATS(ctx) Ppmd8_GetStats(p, ctx) +#define ONE_STATE(ctx) Ppmd8Context_OneState(ctx) +#define SUFFIX(ctx) CTX((ctx)->Suffix) + +#define kTop (1 << 24) +#define kBot (1 << 15) + +typedef CPpmd8_Context * CTX_PTR; + +struct CPpmd8_Node_; + +typedef + #ifdef PPMD_32BIT + struct CPpmd8_Node_ * + #else + UInt32 + #endif + CPpmd8_Node_Ref; + +typedef struct CPpmd8_Node_ +{ + UInt32 Stamp; + CPpmd8_Node_Ref Next; + UInt32 NU; +} CPpmd8_Node; + +#ifdef PPMD_32BIT + #define NODE(ptr) (ptr) +#else + #define NODE(offs) ((CPpmd8_Node *)(p->Base + (offs))) +#endif + +#define EMPTY_NODE 0xFFFFFFFF + +void Ppmd8_Construct(CPpmd8 *p) +{ + unsigned i, k, m; + + p->Base = 0; + + for (i = 0, k = 0; i < PPMD_NUM_INDEXES; i++) + { + unsigned step = (i >= 12 ? 4 : (i >> 2) + 1); + do { p->Units2Indx[k++] = (Byte)i; } while (--step); + p->Indx2Units[i] = (Byte)k; + } + + p->NS2BSIndx[0] = (0 << 1); + p->NS2BSIndx[1] = (1 << 1); + memset(p->NS2BSIndx + 2, (2 << 1), 9); + memset(p->NS2BSIndx + 11, (3 << 1), 256 - 11); + + for (i = 0; i < 5; i++) + p->NS2Indx[i] = (Byte)i; + for (m = i, k = 1; i < 260; i++) + { + p->NS2Indx[i] = (Byte)m; + if (--k == 0) + k = (++m) - 4; + } +} + +void Ppmd8_Free(CPpmd8 *p) +{ + free(p->Base); + p->Size = 0; + p->Base = 0; +} + +Bool Ppmd8_Alloc(CPpmd8 *p, UInt32 size) +{ + if (p->Base == 0 || p->Size != size) + { + Ppmd8_Free(p); + p->AlignOffset = + #ifdef PPMD_32BIT + (4 - size) & 3; + #else + 4 - (size & 3); + #endif + if ((p->Base = (Byte *)malloc(p->AlignOffset + size)) == 0) + return False; + p->Size = size; + } + return True; +} + +static void InsertNode(CPpmd8 *p, void *node, unsigned indx) +{ + ((CPpmd8_Node *)node)->Stamp = EMPTY_NODE; + ((CPpmd8_Node *)node)->Next = (CPpmd8_Node_Ref)p->FreeList[indx]; + ((CPpmd8_Node *)node)->NU = I2U(indx); + p->FreeList[indx] = REF(node); + p->Stamps[indx]++; +} + +static void *RemoveNode(CPpmd8 *p, unsigned indx) +{ + CPpmd8_Node *node = NODE((CPpmd8_Node_Ref)p->FreeList[indx]); + p->FreeList[indx] = node->Next; + p->Stamps[indx]--; + return node; +} + +static void SplitBlock(CPpmd8 *p, void *ptr, unsigned oldIndx, unsigned newIndx) +{ + unsigned i, nu = I2U(oldIndx) - I2U(newIndx); + ptr = (Byte *)ptr + U2B(I2U(newIndx)); + if (I2U(i = U2I(nu)) != nu) + { + unsigned k = I2U(--i); + InsertNode(p, ((Byte *)ptr) + U2B(k), nu - k - 1); + } + InsertNode(p, ptr, i); +} + +static void GlueFreeBlocks(CPpmd8 *p) +{ + CPpmd8_Node_Ref head = 0; + CPpmd8_Node_Ref *prev = &head; + unsigned i; + + p->GlueCount = 1 << 13; + memset(p->Stamps, 0, sizeof(p->Stamps)); + + /* Order-0 context is always at top UNIT, so we don't need guard NODE at the end. + All blocks up to p->LoUnit can be free, so we need guard NODE at LoUnit. */ + if (p->LoUnit != p->HiUnit) + ((CPpmd8_Node *)p->LoUnit)->Stamp = 0; + + /* Glue free blocks */ + for (i = 0; i < PPMD_NUM_INDEXES; i++) + { + CPpmd8_Node_Ref next = (CPpmd8_Node_Ref)p->FreeList[i]; + p->FreeList[i] = 0; + while (next != 0) + { + CPpmd8_Node *node = NODE(next); + if (node->NU != 0) + { + CPpmd8_Node *node2; + *prev = next; + prev = &(node->Next); + while ((node2 = node + node->NU)->Stamp == EMPTY_NODE) + { + node->NU += node2->NU; + node2->NU = 0; + } + } + next = node->Next; + } + } + *prev = 0; + + /* Fill lists of free blocks */ + while (head != 0) + { + CPpmd8_Node *node = NODE(head); + unsigned nu; + head = node->Next; + nu = node->NU; + if (nu == 0) + continue; + for (; nu > 128; nu -= 128, node += 128) + InsertNode(p, node, PPMD_NUM_INDEXES - 1); + if (I2U(i = U2I(nu)) != nu) + { + unsigned k = I2U(--i); + InsertNode(p, node + k, nu - k - 1); + } + InsertNode(p, node, i); + } +} + +static void *AllocUnitsRare(CPpmd8 *p, unsigned indx) +{ + unsigned i; + void *retVal; + if (p->GlueCount == 0) + { + GlueFreeBlocks(p); + if (p->FreeList[indx] != 0) + return RemoveNode(p, indx); + } + i = indx; + do + { + if (++i == PPMD_NUM_INDEXES) + { + UInt32 numBytes = U2B(I2U(indx)); + p->GlueCount--; + return ((UInt32)(p->UnitsStart - p->Text) > numBytes) ? (p->UnitsStart -= numBytes) : (NULL); + } + } + while (p->FreeList[i] == 0); + retVal = RemoveNode(p, i); + SplitBlock(p, retVal, i, indx); + return retVal; +} + +static void *AllocUnits(CPpmd8 *p, unsigned indx) +{ + UInt32 numBytes; + if (p->FreeList[indx] != 0) + return RemoveNode(p, indx); + numBytes = U2B(I2U(indx)); + if (numBytes <= (UInt32)(p->HiUnit - p->LoUnit)) + { + void *retVal = p->LoUnit; + p->LoUnit += numBytes; + return retVal; + } + return AllocUnitsRare(p, indx); +} + +#define MyMem12Cpy(dest, src, num) \ + { UInt32 *d = (UInt32 *)dest; const UInt32 *z = (const UInt32 *)src; UInt32 n = num; \ + do { d[0] = z[0]; d[1] = z[1]; d[2] = z[2]; z += 3; d += 3; } while (--n); } + +static void *ShrinkUnits(CPpmd8 *p, void *oldPtr, unsigned oldNU, unsigned newNU) +{ + unsigned i0 = U2I(oldNU); + unsigned i1 = U2I(newNU); + if (i0 == i1) + return oldPtr; + if (p->FreeList[i1] != 0) + { + void *ptr = RemoveNode(p, i1); + MyMem12Cpy(ptr, oldPtr, newNU); + InsertNode(p, oldPtr, i0); + return ptr; + } + SplitBlock(p, oldPtr, i0, i1); + return oldPtr; +} + +static void FreeUnits(CPpmd8 *p, void *ptr, unsigned nu) +{ + InsertNode(p, ptr, U2I(nu)); +} + +static void SpecialFreeUnit(CPpmd8 *p, void *ptr) +{ + if ((Byte *)ptr != p->UnitsStart) + InsertNode(p, ptr, 0); + else + { + #ifdef PPMD8_FREEZE_SUPPORT + *(UInt32 *)ptr = EMPTY_NODE; /* it's used for (Flags == 0xFF) check in RemoveBinContexts */ + #endif + p->UnitsStart += UNIT_SIZE; + } +} + +static void *MoveUnitsUp(CPpmd8 *p, void *oldPtr, unsigned nu) +{ + unsigned indx = U2I(nu); + void *ptr; + if ((Byte *)oldPtr > p->UnitsStart + 16 * 1024 || REF(oldPtr) > p->FreeList[indx]) + return oldPtr; + ptr = RemoveNode(p, indx); + MyMem12Cpy(ptr, oldPtr, nu); + if ((Byte*)oldPtr != p->UnitsStart) + InsertNode(p, oldPtr, indx); + else + p->UnitsStart += U2B(I2U(indx)); + return ptr; +} + +static void ExpandTextArea(CPpmd8 *p) +{ + UInt32 count[PPMD_NUM_INDEXES]; + unsigned i; + memset(count, 0, sizeof(count)); + if (p->LoUnit != p->HiUnit) + ((CPpmd8_Node *)p->LoUnit)->Stamp = 0; + + { + CPpmd8_Node *node = (CPpmd8_Node *)p->UnitsStart; + for (; node->Stamp == EMPTY_NODE; node += node->NU) + { + node->Stamp = 0; + count[U2I(node->NU)]++; + } + p->UnitsStart = (Byte *)node; + } + + for (i = 0; i < PPMD_NUM_INDEXES; i++) + { + CPpmd8_Node_Ref *next = (CPpmd8_Node_Ref *)&p->FreeList[i]; + while (count[i] != 0) + { + CPpmd8_Node *node = NODE(*next); + while (node->Stamp == 0) + { + *next = node->Next; + node = NODE(*next); + p->Stamps[i]--; + if (--count[i] == 0) + break; + } + next = &node->Next; + } + } +} + +#define SUCCESSOR(p) ((CPpmd_Void_Ref)((p)->SuccessorLow | ((UInt32)(p)->SuccessorHigh << 16))) + +static void SetSuccessor(CPpmd_State *p, CPpmd_Void_Ref v) +{ + (p)->SuccessorLow = (UInt16)((UInt32)(v) & 0xFFFF); + (p)->SuccessorHigh = (UInt16)(((UInt32)(v) >> 16) & 0xFFFF); +} + +#define RESET_TEXT(offs) { p->Text = p->Base + p->AlignOffset + (offs); } + +static void RestartModel(CPpmd8 *p) +{ + unsigned i, k, m, r; + + memset(p->FreeList, 0, sizeof(p->FreeList)); + memset(p->Stamps, 0, sizeof(p->Stamps)); + RESET_TEXT(0); + p->HiUnit = p->Text + p->Size; + p->LoUnit = p->UnitsStart = p->HiUnit - p->Size / 8 / UNIT_SIZE * 7 * UNIT_SIZE; + p->GlueCount = 0; + + p->OrderFall = p->MaxOrder; + p->RunLength = p->InitRL = -(Int32)((p->MaxOrder < 12) ? p->MaxOrder : 12) - 1; + p->PrevSuccess = 0; + + p->MinContext = p->MaxContext = (CTX_PTR)(p->HiUnit -= UNIT_SIZE); /* AllocContext(p); */ + p->MinContext->Suffix = 0; + p->MinContext->NumStats = 255; + p->MinContext->Flags = 0; + p->MinContext->SummFreq = 256 + 1; + p->FoundState = (CPpmd_State *)p->LoUnit; /* AllocUnits(p, PPMD_NUM_INDEXES - 1); */ + p->LoUnit += U2B(256 / 2); + p->MinContext->Stats = REF(p->FoundState); + for (i = 0; i < 256; i++) + { + CPpmd_State *s = &p->FoundState[i]; + s->Symbol = (Byte)i; + s->Freq = 1; + SetSuccessor(s, 0); + } + + for (i = m = 0; m < 25; m++) + { + while (p->NS2Indx[i] == m) + i++; + for (k = 0; k < 8; k++) + { + UInt16 val = (UInt16)(PPMD_BIN_SCALE - kInitBinEsc[k] / (i + 1)); + UInt16 *dest = p->BinSumm[m] + k; + for (r = 0; r < 64; r += 8) + dest[r] = val; + } + } + + for (i = m = 0; m < 24; m++) + { + while (p->NS2Indx[i + 3] == m + 3) + i++; + for (k = 0; k < 32; k++) + { + CPpmd_See *s = &p->See[m][k]; + s->Summ = (UInt16)((2 * i + 5) << (s->Shift = PPMD_PERIOD_BITS - 4)); + s->Count = 7; + } + } +} + +void Ppmd8_Init(CPpmd8 *p, unsigned maxOrder, unsigned restoreMethod) +{ + p->MaxOrder = maxOrder; + p->RestoreMethod = restoreMethod; + RestartModel(p); + p->DummySee.Shift = PPMD_PERIOD_BITS; + p->DummySee.Summ = 0; /* unused */ + p->DummySee.Count = 64; /* unused */ +} + +static void Refresh(CPpmd8 *p, CTX_PTR ctx, unsigned oldNU, unsigned scale) +{ + unsigned i = ctx->NumStats, escFreq, sumFreq, flags; + CPpmd_State *s = (CPpmd_State *)ShrinkUnits(p, STATS(ctx), oldNU, (i + 2) >> 1); + ctx->Stats = REF(s); + #ifdef PPMD8_FREEZE_SUPPORT + /* fixed over Shkarin's code. Fixed code is not compatible with original code for some files in FREEZE mode. */ + scale |= (ctx->SummFreq >= ((UInt32)1 << 15)); + #endif + flags = (ctx->Flags & (0x10 + 0x04 * scale)) + 0x08 * (s->Symbol >= 0x40); + escFreq = ctx->SummFreq - s->Freq; + sumFreq = (s->Freq = (Byte)((s->Freq + scale) >> scale)); + do + { + escFreq -= (++s)->Freq; + sumFreq += (s->Freq = (Byte)((s->Freq + scale) >> scale)); + flags |= 0x08 * (s->Symbol >= 0x40); + } + while (--i); + ctx->SummFreq = (UInt16)(sumFreq + ((escFreq + scale) >> scale)); + ctx->Flags = (Byte)flags; +} + +static void SwapStates(CPpmd_State *t1, CPpmd_State *t2) +{ + CPpmd_State tmp = *t1; + *t1 = *t2; + *t2 = tmp; +} + +static CPpmd_Void_Ref CutOff(CPpmd8 *p, CTX_PTR ctx, unsigned order) +{ + int i; + unsigned tmp; + CPpmd_State *s; + + if (!ctx->NumStats) + { + s = ONE_STATE(ctx); + if ((Byte *)Ppmd8_GetPtr(p, SUCCESSOR(s)) >= p->UnitsStart) + { + if (order < p->MaxOrder) + SetSuccessor(s, CutOff(p, CTX(SUCCESSOR(s)), order + 1)); + else + SetSuccessor(s, 0); + if (SUCCESSOR(s) || order <= 9) /* O_BOUND */ + return REF(ctx); + } + SpecialFreeUnit(p, ctx); + return 0; + } + + ctx->Stats = STATS_REF(MoveUnitsUp(p, STATS(ctx), tmp = ((unsigned)ctx->NumStats + 2) >> 1)); + + for (s = STATS(ctx) + (i = ctx->NumStats); s >= STATS(ctx); s--) + if ((Byte *)Ppmd8_GetPtr(p, SUCCESSOR(s)) < p->UnitsStart) + { + CPpmd_State *s2 = STATS(ctx) + (i--); + SetSuccessor(s, 0); + SwapStates(s, s2); + } + else if (order < p->MaxOrder) + SetSuccessor(s, CutOff(p, CTX(SUCCESSOR(s)), order + 1)); + else + SetSuccessor(s, 0); + + if (i != ctx->NumStats && order) + { + ctx->NumStats = (Byte)i; + s = STATS(ctx); + if (i < 0) + { + FreeUnits(p, s, tmp); + SpecialFreeUnit(p, ctx); + return 0; + } + if (i == 0) + { + ctx->Flags = (Byte)((ctx->Flags & 0x10) + 0x08 * (s->Symbol >= 0x40)); + *ONE_STATE(ctx) = *s; + FreeUnits(p, s, tmp); + /* 9.31: the code was fixed. It's was not BUG, if Freq <= MAX_FREQ = 124 */ + ONE_STATE(ctx)->Freq = (Byte)(((unsigned)ONE_STATE(ctx)->Freq + 11) >> 3); + } + else + Refresh(p, ctx, tmp, ctx->SummFreq > 16 * i); + } + return REF(ctx); +} + +#ifdef PPMD8_FREEZE_SUPPORT +static CPpmd_Void_Ref RemoveBinContexts(CPpmd8 *p, CTX_PTR ctx, unsigned order) +{ + CPpmd_State *s; + if (!ctx->NumStats) + { + s = ONE_STATE(ctx); + if ((Byte *)Ppmd8_GetPtr(p, SUCCESSOR(s)) >= p->UnitsStart && order < p->MaxOrder) + SetSuccessor(s, RemoveBinContexts(p, CTX(SUCCESSOR(s)), order + 1)); + else + SetSuccessor(s, 0); + /* Suffix context can be removed already, since different (high-order) + Successors may refer to same context. So we check Flags == 0xFF (Stamp == EMPTY_NODE) */ + if (!SUCCESSOR(s) && (!SUFFIX(ctx)->NumStats || SUFFIX(ctx)->Flags == 0xFF)) + { + FreeUnits(p, ctx, 1); + return 0; + } + else + return REF(ctx); + } + + for (s = STATS(ctx) + ctx->NumStats; s >= STATS(ctx); s--) + if ((Byte *)Ppmd8_GetPtr(p, SUCCESSOR(s)) >= p->UnitsStart && order < p->MaxOrder) + SetSuccessor(s, RemoveBinContexts(p, CTX(SUCCESSOR(s)), order + 1)); + else + SetSuccessor(s, 0); + + return REF(ctx); +} +#endif + +static UInt32 GetUsedMemory(const CPpmd8 *p) +{ + UInt32 v = 0; + unsigned i; + for (i = 0; i < PPMD_NUM_INDEXES; i++) + v += p->Stamps[i] * I2U(i); + return p->Size - (UInt32)(p->HiUnit - p->LoUnit) - (UInt32)(p->UnitsStart - p->Text) - U2B(v); +} + +#ifdef PPMD8_FREEZE_SUPPORT + #define RESTORE_MODEL(c1, fSuccessor) RestoreModel(p, c1, fSuccessor) +#else + #define RESTORE_MODEL(c1, fSuccessor) RestoreModel(p, c1) +#endif + +static void RestoreModel(CPpmd8 *p, CTX_PTR c1 + #ifdef PPMD8_FREEZE_SUPPORT + , CTX_PTR fSuccessor + #endif + ) +{ + CTX_PTR c; + CPpmd_State *s; + RESET_TEXT(0); + for (c = p->MaxContext; c != c1; c = SUFFIX(c)) + if (--(c->NumStats) == 0) + { + s = STATS(c); + c->Flags = (Byte)((c->Flags & 0x10) + 0x08 * (s->Symbol >= 0x40)); + *ONE_STATE(c) = *s; + SpecialFreeUnit(p, s); + ONE_STATE(c)->Freq = (Byte)(((unsigned)ONE_STATE(c)->Freq + 11) >> 3); + } + else + Refresh(p, c, (c->NumStats+3) >> 1, 0); + + for (; c != p->MinContext; c = SUFFIX(c)) + if (!c->NumStats) + ONE_STATE(c)->Freq = (Byte)(ONE_STATE(c)->Freq - (ONE_STATE(c)->Freq >> 1)); + else if ((c->SummFreq += 4) > 128 + 4 * c->NumStats) + Refresh(p, c, (c->NumStats + 2) >> 1, 1); + + #ifdef PPMD8_FREEZE_SUPPORT + if (p->RestoreMethod > PPMD8_RESTORE_METHOD_FREEZE) + { + p->MaxContext = fSuccessor; + p->GlueCount += !(p->Stamps[1] & 1); + } + else if (p->RestoreMethod == PPMD8_RESTORE_METHOD_FREEZE) + { + while (p->MaxContext->Suffix) + p->MaxContext = SUFFIX(p->MaxContext); + RemoveBinContexts(p, p->MaxContext, 0); + p->RestoreMethod++; + p->GlueCount = 0; + p->OrderFall = p->MaxOrder; + } + else + #endif + if (p->RestoreMethod == PPMD8_RESTORE_METHOD_RESTART || GetUsedMemory(p) < (p->Size >> 1)) + RestartModel(p); + else + { + while (p->MaxContext->Suffix) + p->MaxContext = SUFFIX(p->MaxContext); + do + { + CutOff(p, p->MaxContext, 0); + ExpandTextArea(p); + } + while (GetUsedMemory(p) > 3 * (p->Size >> 2)); + p->GlueCount = 0; + p->OrderFall = p->MaxOrder; + } +} + +static CTX_PTR CreateSuccessors(CPpmd8 *p, Bool skip, CPpmd_State *s1, CTX_PTR c) +{ + CPpmd_State upState; + Byte flags; + CPpmd_Byte_Ref upBranch = (CPpmd_Byte_Ref)SUCCESSOR(p->FoundState); + /* fixed over Shkarin's code. Maybe it could work without + 1 too. */ + CPpmd_State *ps[PPMD8_MAX_ORDER + 1]; + unsigned numPs = 0; + + if (!skip) + ps[numPs++] = p->FoundState; + + while (c->Suffix) + { + CPpmd_Void_Ref successor; + CPpmd_State *s; + c = SUFFIX(c); + if (s1) + { + s = s1; + s1 = NULL; + } + else if (c->NumStats != 0) + { + for (s = STATS(c); s->Symbol != p->FoundState->Symbol; s++); + if (s->Freq < MAX_FREQ - 9) + { + s->Freq++; + c->SummFreq++; + } + } + else + { + s = ONE_STATE(c); + s->Freq = (Byte)(s->Freq + (!SUFFIX(c)->NumStats & (s->Freq < 24))); + } + successor = SUCCESSOR(s); + if (successor != upBranch) + { + c = CTX(successor); + if (numPs == 0) + return c; + break; + } + ps[numPs++] = s; + } + + upState.Symbol = *(const Byte *)Ppmd8_GetPtr(p, upBranch); + SetSuccessor(&upState, upBranch + 1); + flags = (Byte)(0x10 * (p->FoundState->Symbol >= 0x40) + 0x08 * (upState.Symbol >= 0x40)); + + if (c->NumStats == 0) + upState.Freq = ONE_STATE(c)->Freq; + else + { + UInt32 cf, s0; + CPpmd_State *s; + for (s = STATS(c); s->Symbol != upState.Symbol; s++); + cf = s->Freq - 1; + s0 = c->SummFreq - c->NumStats - cf; + upState.Freq = (Byte)(1 + ((2 * cf <= s0) ? (5 * cf > s0) : ((cf + 2 * s0 - 3) / s0))); + } + + do + { + /* Create Child */ + CTX_PTR c1; /* = AllocContext(p); */ + if (p->HiUnit != p->LoUnit) + c1 = (CTX_PTR)(p->HiUnit -= UNIT_SIZE); + else if (p->FreeList[0] != 0) + c1 = (CTX_PTR)RemoveNode(p, 0); + else + { + c1 = (CTX_PTR)AllocUnitsRare(p, 0); + if (!c1) + return NULL; + } + c1->NumStats = 0; + c1->Flags = flags; + *ONE_STATE(c1) = upState; + c1->Suffix = REF(c); + SetSuccessor(ps[--numPs], REF(c1)); + c = c1; + } + while (numPs != 0); + + return c; +} + +static CTX_PTR ReduceOrder(CPpmd8 *p, CPpmd_State *s1, CTX_PTR c) +{ + CPpmd_State *s = NULL; + CTX_PTR c1 = c; + CPpmd_Void_Ref upBranch = REF(p->Text); + + #ifdef PPMD8_FREEZE_SUPPORT + /* The BUG in Shkarin's code was fixed: ps could overflow in CUT_OFF mode. */ + CPpmd_State *ps[PPMD8_MAX_ORDER + 1]; + unsigned numPs = 0; + ps[numPs++] = p->FoundState; + #endif + + SetSuccessor(p->FoundState, upBranch); + p->OrderFall++; + + for (;;) + { + if (s1) + { + c = SUFFIX(c); + s = s1; + s1 = NULL; + } + else + { + if (!c->Suffix) + { + #ifdef PPMD8_FREEZE_SUPPORT + if (p->RestoreMethod > PPMD8_RESTORE_METHOD_FREEZE) + { + do { SetSuccessor(ps[--numPs], REF(c)); } while (numPs); + RESET_TEXT(1); + p->OrderFall = 1; + } + #endif + return c; + } + c = SUFFIX(c); + if (c->NumStats) + { + if ((s = STATS(c))->Symbol != p->FoundState->Symbol) + do { s++; } while (s->Symbol != p->FoundState->Symbol); + if (s->Freq < MAX_FREQ - 9) + { + s->Freq += 2; + c->SummFreq += 2; + } + } + else + { + s = ONE_STATE(c); + s->Freq = (Byte)(s->Freq + (s->Freq < 32)); + } + } + if (SUCCESSOR(s)) + break; + #ifdef PPMD8_FREEZE_SUPPORT + ps[numPs++] = s; + #endif + SetSuccessor(s, upBranch); + p->OrderFall++; + } + + #ifdef PPMD8_FREEZE_SUPPORT + if (p->RestoreMethod > PPMD8_RESTORE_METHOD_FREEZE) + { + c = CTX(SUCCESSOR(s)); + do { SetSuccessor(ps[--numPs], REF(c)); } while (numPs); + RESET_TEXT(1); + p->OrderFall = 1; + return c; + } + else + #endif + if (SUCCESSOR(s) <= upBranch) + { + CTX_PTR successor; + CPpmd_State *s2 = p->FoundState; + p->FoundState = s; + + successor = CreateSuccessors(p, False, NULL, c); + if (successor == NULL) + SetSuccessor(s, 0); + else + SetSuccessor(s, REF(successor)); + p->FoundState = s2; + } + + if (p->OrderFall == 1 && c1 == p->MaxContext) + { + SetSuccessor(p->FoundState, SUCCESSOR(s)); + p->Text--; + } + if (SUCCESSOR(s) == 0) + return NULL; + return CTX(SUCCESSOR(s)); +} + +static void UpdateModel(CPpmd8 *p) +{ + CPpmd_Void_Ref successor, fSuccessor = SUCCESSOR(p->FoundState); + CTX_PTR c; + unsigned s0, ns, fFreq = p->FoundState->Freq; + Byte flag, fSymbol = p->FoundState->Symbol; + CPpmd_State *s = NULL; + + if (p->FoundState->Freq < MAX_FREQ / 4 && p->MinContext->Suffix != 0) + { + c = SUFFIX(p->MinContext); + + if (c->NumStats == 0) + { + s = ONE_STATE(c); + if (s->Freq < 32) + s->Freq++; + } + else + { + s = STATS(c); + if (s->Symbol != p->FoundState->Symbol) + { + do { s++; } while (s->Symbol != p->FoundState->Symbol); + if (s[0].Freq >= s[-1].Freq) + { + SwapStates(&s[0], &s[-1]); + s--; + } + } + if (s->Freq < MAX_FREQ - 9) + { + s->Freq += 2; + c->SummFreq += 2; + } + } + } + + c = p->MaxContext; + if (p->OrderFall == 0 && fSuccessor) + { + CTX_PTR cs = CreateSuccessors(p, True, s, p->MinContext); + if (cs == 0) + { + SetSuccessor(p->FoundState, 0); + RESTORE_MODEL(c, CTX(fSuccessor)); + } + else + { + SetSuccessor(p->FoundState, REF(cs)); + p->MaxContext = cs; + } + return; + } + + *p->Text++ = p->FoundState->Symbol; + successor = REF(p->Text); + if (p->Text >= p->UnitsStart) + { + RESTORE_MODEL(c, CTX(fSuccessor)); /* check it */ + return; + } + + if (!fSuccessor) + { + CTX_PTR cs = ReduceOrder(p, s, p->MinContext); + if (cs == NULL) + { + RESTORE_MODEL(c, 0); + return; + } + fSuccessor = REF(cs); + } + else if ((Byte *)Ppmd8_GetPtr(p, fSuccessor) < p->UnitsStart) + { + CTX_PTR cs = CreateSuccessors(p, False, s, p->MinContext); + if (cs == NULL) + { + RESTORE_MODEL(c, 0); + return; + } + fSuccessor = REF(cs); + } + + if (--p->OrderFall == 0) + { + successor = fSuccessor; + p->Text -= (p->MaxContext != p->MinContext); + } + #ifdef PPMD8_FREEZE_SUPPORT + else if (p->RestoreMethod > PPMD8_RESTORE_METHOD_FREEZE) + { + successor = fSuccessor; + RESET_TEXT(0); + p->OrderFall = 0; + } + #endif + + s0 = p->MinContext->SummFreq - (ns = p->MinContext->NumStats) - fFreq; + flag = (Byte)(0x08 * (fSymbol >= 0x40)); + + for (; c != p->MinContext; c = SUFFIX(c)) + { + unsigned ns1; + UInt32 cf, sf; + if ((ns1 = c->NumStats) != 0) + { + if ((ns1 & 1) != 0) + { + /* Expand for one UNIT */ + unsigned oldNU = (ns1 + 1) >> 1; + unsigned i = U2I(oldNU); + if (i != U2I(oldNU + 1)) + { + void *ptr = AllocUnits(p, i + 1); + void *oldPtr; + if (!ptr) + { + RESTORE_MODEL(c, CTX(fSuccessor)); + return; + } + oldPtr = STATS(c); + MyMem12Cpy(ptr, oldPtr, oldNU); + InsertNode(p, oldPtr, i); + c->Stats = STATS_REF(ptr); + } + } + c->SummFreq = (UInt16)(c->SummFreq + (3 * ns1 + 1 < ns)); + } + else + { + CPpmd_State *s2 = (CPpmd_State*)AllocUnits(p, 0); + if (!s2) + { + RESTORE_MODEL(c, CTX(fSuccessor)); + return; + } + *s2 = *ONE_STATE(c); + c->Stats = REF(s2); + if (s2->Freq < MAX_FREQ / 4 - 1) + s2->Freq <<= 1; + else + s2->Freq = MAX_FREQ - 4; + c->SummFreq = (UInt16)(s2->Freq + p->InitEsc + (ns > 2)); + } + cf = 2 * fFreq * (c->SummFreq + 6); + sf = (UInt32)s0 + c->SummFreq; + if (cf < 6 * sf) + { + cf = 1 + (cf > sf) + (cf >= 4 * sf); + c->SummFreq += 4; + } + else + { + cf = 4 + (cf > 9 * sf) + (cf > 12 * sf) + (cf > 15 * sf); + c->SummFreq = (UInt16)(c->SummFreq + cf); + } + { + CPpmd_State *s2 = STATS(c) + ns1 + 1; + SetSuccessor(s2, successor); + s2->Symbol = fSymbol; + s2->Freq = (Byte)cf; + c->Flags |= flag; + c->NumStats = (Byte)(ns1 + 1); + } + } + p->MaxContext = p->MinContext = CTX(fSuccessor); +} + +static void Rescale(CPpmd8 *p) +{ + unsigned i, adder, sumFreq, escFreq; + CPpmd_State *stats = STATS(p->MinContext); + CPpmd_State *s = p->FoundState; + { + CPpmd_State tmp = *s; + for (; s != stats; s--) + s[0] = s[-1]; + *s = tmp; + } + escFreq = p->MinContext->SummFreq - s->Freq; + s->Freq += 4; + adder = (p->OrderFall != 0 + #ifdef PPMD8_FREEZE_SUPPORT + || p->RestoreMethod > PPMD8_RESTORE_METHOD_FREEZE + #endif + ); + s->Freq = (Byte)((s->Freq + adder) >> 1); + sumFreq = s->Freq; + + i = p->MinContext->NumStats; + do + { + escFreq -= (++s)->Freq; + s->Freq = (Byte)((s->Freq + adder) >> 1); + sumFreq += s->Freq; + if (s[0].Freq > s[-1].Freq) + { + CPpmd_State *s1 = s; + CPpmd_State tmp = *s1; + do + s1[0] = s1[-1]; + while (--s1 != stats && tmp.Freq > s1[-1].Freq); + *s1 = tmp; + } + } + while (--i); + + if (s->Freq == 0) + { + unsigned numStats = p->MinContext->NumStats; + unsigned n0, n1; + do { i++; } while ((--s)->Freq == 0); + escFreq += i; + p->MinContext->NumStats = (Byte)(p->MinContext->NumStats - i); + if (p->MinContext->NumStats == 0) + { + CPpmd_State tmp = *stats; + tmp.Freq = (Byte)((2 * tmp.Freq + escFreq - 1) / escFreq); + if (tmp.Freq > MAX_FREQ / 3) + tmp.Freq = MAX_FREQ / 3; + InsertNode(p, stats, U2I((numStats + 2) >> 1)); + p->MinContext->Flags = (Byte)((p->MinContext->Flags & 0x10) + 0x08 * (tmp.Symbol >= 0x40)); + *(p->FoundState = ONE_STATE(p->MinContext)) = tmp; + return; + } + n0 = (numStats + 2) >> 1; + n1 = (p->MinContext->NumStats + 2) >> 1; + if (n0 != n1) + p->MinContext->Stats = STATS_REF(ShrinkUnits(p, stats, n0, n1)); + p->MinContext->Flags &= ~0x08; + p->MinContext->Flags |= 0x08 * ((s = STATS(p->MinContext))->Symbol >= 0x40); + i = p->MinContext->NumStats; + do { p->MinContext->Flags |= 0x08*((++s)->Symbol >= 0x40); } while (--i); + } + p->MinContext->SummFreq = (UInt16)(sumFreq + escFreq - (escFreq >> 1)); + p->MinContext->Flags |= 0x4; + p->FoundState = STATS(p->MinContext); +} + +CPpmd_See *Ppmd8_MakeEscFreq(CPpmd8 *p, unsigned numMasked1, UInt32 *escFreq) +{ + CPpmd_See *see; + if (p->MinContext->NumStats != 0xFF) + { + see = p->See[(unsigned)p->NS2Indx[(unsigned)p->MinContext->NumStats + 2] - 3] + + (p->MinContext->SummFreq > 11 * ((unsigned)p->MinContext->NumStats + 1)) + + 2 * (unsigned)(2 * (unsigned)p->MinContext->NumStats < + ((unsigned)SUFFIX(p->MinContext)->NumStats + numMasked1)) + + p->MinContext->Flags; + { + unsigned r = (see->Summ >> see->Shift); + see->Summ = (UInt16)(see->Summ - r); + *escFreq = r + (r == 0); + } + } + else + { + see = &p->DummySee; + *escFreq = 1; + } + return see; +} + +static void NextContext(CPpmd8 *p) +{ + CTX_PTR c = CTX(SUCCESSOR(p->FoundState)); + if (p->OrderFall == 0 && (Byte *)c >= p->UnitsStart) + p->MinContext = p->MaxContext = c; + else + { + UpdateModel(p); + p->MinContext = p->MaxContext; + } +} + +void Ppmd8_Update1(CPpmd8 *p) +{ + CPpmd_State *s = p->FoundState; + s->Freq += 4; + p->MinContext->SummFreq += 4; + if (s[0].Freq > s[-1].Freq) + { + SwapStates(&s[0], &s[-1]); + p->FoundState = --s; + if (s->Freq > MAX_FREQ) + Rescale(p); + } + NextContext(p); +} + +void Ppmd8_Update1_0(CPpmd8 *p) +{ + p->PrevSuccess = (2 * p->FoundState->Freq >= p->MinContext->SummFreq); + p->RunLength += p->PrevSuccess; + p->MinContext->SummFreq += 4; + if ((p->FoundState->Freq += 4) > MAX_FREQ) + Rescale(p); + NextContext(p); +} + +void Ppmd8_UpdateBin(CPpmd8 *p) +{ + p->FoundState->Freq = (Byte)(p->FoundState->Freq + (p->FoundState->Freq < 196)); + p->PrevSuccess = 1; + p->RunLength++; + NextContext(p); +} + +void Ppmd8_Update2(CPpmd8 *p) +{ + p->MinContext->SummFreq += 4; + if ((p->FoundState->Freq += 4) > MAX_FREQ) + Rescale(p); + p->RunLength = p->InitRL; + UpdateModel(p); + p->MinContext = p->MaxContext; +} + +/* Ppmd8Dec.c -- PPMdI Decoder +2010-04-16 : Igor Pavlov : Public domain +This code is based on: + PPMd var.I (2002): Dmitry Shkarin : Public domain + Carryless rangecoder (1999): Dmitry Subbotin : Public domain */ + +Bool Ppmd8_RangeDec_Init(CPpmd8 *p) +{ + unsigned i; + p->Low = 0; + p->Range = 0xFFFFFFFF; + p->Code = 0; + for (i = 0; i < 4; i++) + p->Code = (p->Code << 8) | p->Stream.In->Read(p->Stream.In); + return (p->Code < 0xFFFFFFFF); +} + +static UInt32 RangeDec_GetThreshold(CPpmd8 *p, UInt32 total) +{ + return p->Code / (p->Range /= total); +} + +static void RangeDec_Decode(CPpmd8 *p, UInt32 start, UInt32 size) +{ + start *= p->Range; + p->Low += start; + p->Code -= start; + p->Range *= size; + + while ((p->Low ^ (p->Low + p->Range)) < kTop || + (p->Range < kBot && ((p->Range = (0 - p->Low) & (kBot - 1)), 1))) + { + p->Code = (p->Code << 8) | p->Stream.In->Read(p->Stream.In); + p->Range <<= 8; + p->Low <<= 8; + } +} + +#define MASK(sym) ((signed char *)charMask)[sym] + +int Ppmd8_DecodeSymbol(CPpmd8 *p) +{ + size_t charMask[256 / sizeof(size_t)]; + if (p->MinContext->NumStats != 0) + { + CPpmd_State *s = Ppmd8_GetStats(p, p->MinContext); + unsigned i; + UInt32 count, hiCnt; + if ((count = RangeDec_GetThreshold(p, p->MinContext->SummFreq)) < (hiCnt = s->Freq)) + { + Byte symbol; + RangeDec_Decode(p, 0, s->Freq); + p->FoundState = s; + symbol = s->Symbol; + Ppmd8_Update1_0(p); + return symbol; + } + p->PrevSuccess = 0; + i = p->MinContext->NumStats; + do + { + if ((hiCnt += (++s)->Freq) > count) + { + Byte symbol; + RangeDec_Decode(p, hiCnt - s->Freq, s->Freq); + p->FoundState = s; + symbol = s->Symbol; + Ppmd8_Update1(p); + return symbol; + } + } + while (--i); + if (count >= p->MinContext->SummFreq) + return -2; + RangeDec_Decode(p, hiCnt, p->MinContext->SummFreq - hiCnt); + PPMD_SetAllBitsIn256Bytes(charMask); + MASK(s->Symbol) = 0; + i = p->MinContext->NumStats; + do { MASK((--s)->Symbol) = 0; } while (--i); + } + else + { + UInt16 *prob = Ppmd8_GetBinSumm(p); + if (((p->Code / (p->Range >>= 14)) < *prob)) + { + Byte symbol; + RangeDec_Decode(p, 0, *prob); + *prob = (UInt16)PPMD_UPDATE_PROB_0(*prob); + symbol = (p->FoundState = Ppmd8Context_OneState(p->MinContext))->Symbol; + Ppmd8_UpdateBin(p); + return symbol; + } + RangeDec_Decode(p, *prob, (1 << 14) - *prob); + *prob = (UInt16)PPMD_UPDATE_PROB_1(*prob); + p->InitEsc = PPMD8_kExpEscape[*prob >> 10]; + PPMD_SetAllBitsIn256Bytes(charMask); + MASK(Ppmd8Context_OneState(p->MinContext)->Symbol) = 0; + p->PrevSuccess = 0; + } + for (;;) + { + CPpmd_State *ps[256], *s; + UInt32 freqSum, count, hiCnt; + CPpmd_See *see; + unsigned i, num, numMasked = p->MinContext->NumStats; + do + { + p->OrderFall++; + if (!p->MinContext->Suffix) + return -1; + p->MinContext = Ppmd8_GetContext(p, p->MinContext->Suffix); + } + while (p->MinContext->NumStats == numMasked); + hiCnt = 0; + s = Ppmd8_GetStats(p, p->MinContext); + i = 0; + num = p->MinContext->NumStats - numMasked; + do + { + int k = (int)(MASK(s->Symbol)); + hiCnt += (s->Freq & k); + ps[i] = s++; + i -= k; + } + while (i != num); + + see = Ppmd8_MakeEscFreq(p, numMasked, &freqSum); + freqSum += hiCnt; + count = RangeDec_GetThreshold(p, freqSum); + + if (count < hiCnt) + { + Byte symbol; + CPpmd_State **pps = ps; + for (hiCnt = 0; (hiCnt += (*pps)->Freq) <= count; pps++); + s = *pps; + RangeDec_Decode(p, hiCnt - s->Freq, s->Freq); + Ppmd_See_Update(see); + p->FoundState = s; + symbol = s->Symbol; + Ppmd8_Update2(p); + return symbol; + } + if (count >= freqSum) + return -2; + RangeDec_Decode(p, hiCnt, freqSum - hiCnt); + see->Summ = (UInt16)(see->Summ + freqSum); + do { MASK(ps[--i]->Symbol) = 0; } while (i != 0); + } +} + +/* H->I changes: + NS2Indx + GlewCount, and Glue method + BinSum + See / EscFreq + CreateSuccessors updates more suffix contexts + UpdateModel consts. + PrevSuccess Update +*/ + +const IPpmd8 __archive_ppmd8_functions = +{ + &Ppmd8_Construct, + &Ppmd8_Alloc, + &Ppmd8_Free, + &Ppmd8_Init, + &Ppmd8_RangeDec_Init, + &Ppmd8_DecodeSymbol, +}; diff --git a/c/archive_ppmd8_private.h b/c/archive_ppmd8_private.h new file mode 100644 --- /dev/null +++ b/c/archive_ppmd8_private.h @@ -0,0 +1,148 @@ +/* Ppmd8.h -- PPMdI codec +2011-01-27 : Igor Pavlov : Public domain +This code is based on: + PPMd var.I (2002): Dmitry Shkarin : Public domain + Carryless rangecoder (1999): Dmitry Subbotin : Public domain */ + +#ifndef ARCHIVE_PPMD8_PRIVATE_H_INCLUDED +#define ARCHIVE_PPMD8_PRIVATE_H_INCLUDED + +#include "archive_ppmd_private.h" + +#define PPMD8_MIN_ORDER 2 +#define PPMD8_MAX_ORDER 16 + +struct CPpmd8_Context_; + +typedef + #ifdef PPMD_32BIT + struct CPpmd8_Context_ * + #else + UInt32 + #endif + CPpmd8_Context_Ref; + +#pragma pack(push, 1) + +typedef struct CPpmd8_Context_ +{ + Byte NumStats; + Byte Flags; + UInt16 SummFreq; + CPpmd_State_Ref Stats; + CPpmd8_Context_Ref Suffix; +} CPpmd8_Context; + +#pragma pack(pop) + +#define Ppmd8Context_OneState(p) ((CPpmd_State *)&(p)->SummFreq) + +/* The BUG in Shkarin's code for FREEZE mode was fixed, but that fixed + code is not compatible with original code for some files compressed + in FREEZE mode. So we disable FREEZE mode support. */ + +enum +{ + PPMD8_RESTORE_METHOD_RESTART, + PPMD8_RESTORE_METHOD_CUT_OFF + #ifdef PPMD8_FREEZE_SUPPORT + , PPMD8_RESTORE_METHOD_FREEZE + #endif +}; + +typedef struct +{ + CPpmd8_Context *MinContext, *MaxContext; + CPpmd_State *FoundState; + unsigned OrderFall, InitEsc, PrevSuccess, MaxOrder; + Int32 RunLength, InitRL; /* must be 32-bit at least */ + + UInt32 Size; + UInt32 GlueCount; + Byte *Base, *LoUnit, *HiUnit, *Text, *UnitsStart; + UInt32 AlignOffset; + unsigned RestoreMethod; + + /* Range Coder */ + UInt32 Range; + UInt32 Code; + UInt32 Low; + union + { + IByteIn *In; + IByteOut *Out; + } Stream; + + Byte Indx2Units[PPMD_NUM_INDEXES]; + Byte Units2Indx[128]; + CPpmd_Void_Ref FreeList[PPMD_NUM_INDEXES]; + UInt32 Stamps[PPMD_NUM_INDEXES]; + + Byte NS2BSIndx[256], NS2Indx[260]; + CPpmd_See DummySee, See[24][32]; + UInt16 BinSumm[25][64]; +} CPpmd8; + +void Ppmd8_Construct(CPpmd8 *p); +Bool Ppmd8_Alloc(CPpmd8 *p, UInt32 size); +void Ppmd8_Free(CPpmd8 *p); +void Ppmd8_Init(CPpmd8 *p, unsigned maxOrder, unsigned restoreMethod); +#define Ppmd8_WasAllocated(p) ((p)->Base != NULL) + + +/* ---------- Internal Functions ---------- */ + +extern const Byte PPMD8_kExpEscape[16]; + +#ifdef PPMD_32BIT + #define Ppmd8_GetPtr(p, ptr) (ptr) + #define Ppmd8_GetContext(p, ptr) (ptr) + #define Ppmd8_GetStats(p, ctx) ((ctx)->Stats) +#else + #define Ppmd8_GetPtr(p, offs) ((void *)((p)->Base + (offs))) + #define Ppmd8_GetContext(p, offs) ((CPpmd8_Context *)Ppmd8_GetPtr((p), (offs))) + #define Ppmd8_GetStats(p, ctx) ((CPpmd_State *)Ppmd8_GetPtr((p), ((ctx)->Stats))) +#endif + +void Ppmd8_Update1(CPpmd8 *p); +void Ppmd8_Update1_0(CPpmd8 *p); +void Ppmd8_Update2(CPpmd8 *p); +void Ppmd8_UpdateBin(CPpmd8 *p); + +#define Ppmd8_GetBinSumm(p) \ + &p->BinSumm[p->NS2Indx[Ppmd8Context_OneState(p->MinContext)->Freq - 1]][ \ + p->NS2BSIndx[Ppmd8_GetContext(p, p->MinContext->Suffix)->NumStats] + \ + p->PrevSuccess + p->MinContext->Flags + ((p->RunLength >> 26) & 0x20)] + +CPpmd_See *Ppmd8_MakeEscFreq(CPpmd8 *p, unsigned numMasked, UInt32 *scale); + + +/* ---------- Decode ---------- */ + +Bool Ppmd8_RangeDec_Init(CPpmd8 *p); +#define Ppmd8_RangeDec_IsFinishedOK(p) ((p)->Code == 0) +int Ppmd8_DecodeSymbol(CPpmd8 *p); /* returns: -1 as EndMarker, -2 as DataError */ + +/* ---------- Encode ---------- */ + +#define Ppmd8_RangeEnc_Init(p) { (p)->Low = 0; (p)->Range = 0xFFFFFFFF; } +void Ppmd8_RangeEnc_FlushData(CPpmd8 *p); +void Ppmd8_EncodeSymbol(CPpmd8 *p, int symbol); /* symbol = -1 means EndMarker */ + +typedef struct +{ + /* Base Functions */ + void (*Ppmd8_Construct)(CPpmd8 *p); + Bool (*Ppmd8_Alloc)(CPpmd8 *p, UInt32 size); + void (*Ppmd8_Free)(CPpmd8 *p); + void (*Ppmd8_Init)(CPpmd8 *p, unsigned max_order, unsigned restore_method); + #define Ppmd7_WasAllocated(p) ((p)->Base != NULL) + + /* Decode Functions */ + int (*Ppmd8_RangeDec_Init)(CPpmd8 *p); + int (*Ppmd8_DecodeSymbol)(CPpmd8 *p); +} IPpmd8; + +extern const IPpmd8 __archive_ppmd8_functions; + +#endif diff --git a/c/archive_ppmd_private.h b/c/archive_ppmd_private.h new file mode 100644 --- /dev/null +++ b/c/archive_ppmd_private.h @@ -0,0 +1,151 @@ +/* Ppmd.h -- PPMD codec common code +2010-03-12 : Igor Pavlov : Public domain +This code is based on PPMd var.H (2001): Dmitry Shkarin : Public domain */ + +#ifndef ARCHIVE_PPMD_PRIVATE_H_INCLUDED +#define ARCHIVE_PPMD_PRIVATE_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif + +#include + +#include "archive_read_private.h" + +/*** Begin defined in Types.h ***/ + +#if !defined(ZCONF_H) +typedef unsigned char Byte; +#endif +typedef short Int16; +typedef unsigned short UInt16; + +#ifdef _LZMA_UINT32_IS_ULONG +typedef long Int32; +typedef unsigned long UInt32; +#else +typedef int Int32; +typedef unsigned int UInt32; +#endif + +#ifdef _SZ_NO_INT_64 + +/* define _SZ_NO_INT_64, if your compiler doesn't support 64-bit integers. + NOTES: Some code will work incorrectly in that case! */ + +typedef long Int64; +typedef unsigned long UInt64; + +#else + +#if defined(_MSC_VER) || defined(__BORLANDC__) +typedef __int64 Int64; +typedef unsigned __int64 UInt64; +#define UINT64_CONST(n) n +#else +typedef long long int Int64; +typedef unsigned long long int UInt64; +#define UINT64_CONST(n) n ## ULL +#endif + +#endif + +typedef int Bool; +#define True 1 +#define False 0 + +/* The following interfaces use first parameter as pointer to structure */ + +typedef struct +{ + struct archive_read *a; + Byte (*Read)(void *p); /* reads one byte, returns 0 in case of EOF or error */ +} IByteIn; + +typedef struct +{ + struct archive_write *a; + void (*Write)(void *p, Byte b); +} IByteOut; + +/*** End defined in Types.h ***/ +/*** Begin defined in CpuArch.h ***/ + +#if defined(_M_IX86) || defined(__i386__) +#define MY_CPU_X86 +#endif + +#if defined(MY_CPU_X86) || defined(_M_ARM) +#define MY_CPU_32BIT +#endif + +#ifdef MY_CPU_32BIT +#define PPMD_32BIT +#endif + +/*** End defined in CpuArch.h ***/ + +#define PPMD_INT_BITS 7 +#define PPMD_PERIOD_BITS 7 +#define PPMD_BIN_SCALE (1 << (PPMD_INT_BITS + PPMD_PERIOD_BITS)) + +#define PPMD_GET_MEAN_SPEC(summ, shift, round) (((summ) + (1 << ((shift) - (round)))) >> (shift)) +#define PPMD_GET_MEAN(summ) PPMD_GET_MEAN_SPEC((summ), PPMD_PERIOD_BITS, 2) +#define PPMD_UPDATE_PROB_0(prob) ((prob) + (1 << PPMD_INT_BITS) - PPMD_GET_MEAN(prob)) +#define PPMD_UPDATE_PROB_1(prob) ((prob) - PPMD_GET_MEAN(prob)) + +#define PPMD_N1 4 +#define PPMD_N2 4 +#define PPMD_N3 4 +#define PPMD_N4 ((128 + 3 - 1 * PPMD_N1 - 2 * PPMD_N2 - 3 * PPMD_N3) / 4) +#define PPMD_NUM_INDEXES (PPMD_N1 + PPMD_N2 + PPMD_N3 + PPMD_N4) + +/* SEE-contexts for PPM-contexts with masked symbols */ +typedef struct +{ + UInt16 Summ; /* Freq */ + Byte Shift; /* Speed of Freq change; low Shift is for fast change */ + Byte Count; /* Count to next change of Shift */ +} CPpmd_See; + +#define Ppmd_See_Update(p) if ((p)->Shift < PPMD_PERIOD_BITS && --(p)->Count == 0) \ + { (p)->Summ <<= 1; (p)->Count = (Byte)(3 << (p)->Shift++); } + +typedef struct +{ + Byte Symbol; + Byte Freq; + UInt16 SuccessorLow; + UInt16 SuccessorHigh; +} CPpmd_State; + +typedef + #ifdef PPMD_32BIT + CPpmd_State * + #else + UInt32 + #endif + CPpmd_State_Ref; + +typedef + #ifdef PPMD_32BIT + void * + #else + UInt32 + #endif + CPpmd_Void_Ref; + +typedef + #ifdef PPMD_32BIT + Byte * + #else + UInt32 + #endif + CPpmd_Byte_Ref; + +#define PPMD_SetAllBitsIn256Bytes(p) \ + { unsigned j; for (j = 0; j < 256 / sizeof(p[0]); j += 8) { \ + p[j+7] = p[j+6] = p[j+5] = p[j+4] = p[j+3] = p[j+2] = p[j+1] = p[j+0] = ~(size_t)0; }} + +#endif diff --git a/c/archive_private.h b/c/archive_private.h new file mode 100644 --- /dev/null +++ b/c/archive_private.h @@ -0,0 +1,176 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD: head/lib/libarchive/archive_private.h 201098 2009-12-28 02:58:14Z kientzle $ + */ + +#ifndef ARCHIVE_PRIVATE_H_INCLUDED +#define ARCHIVE_PRIVATE_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif + +#if HAVE_ICONV_H +#include +#endif + +#include "archive.h" +#include "archive_string.h" + +#if defined(__GNUC__) && (__GNUC__ > 2 || \ + (__GNUC__ == 2 && __GNUC_MINOR__ >= 5)) +#define __LA_DEAD __attribute__((__noreturn__)) +#else +#define __LA_DEAD +#endif + +#define ARCHIVE_WRITE_MAGIC (0xb0c5c0deU) +#define ARCHIVE_READ_MAGIC (0xdeb0c5U) +#define ARCHIVE_WRITE_DISK_MAGIC (0xc001b0c5U) +#define ARCHIVE_READ_DISK_MAGIC (0xbadb0c5U) +#define ARCHIVE_MATCH_MAGIC (0xcad11c9U) + +#define ARCHIVE_STATE_NEW 1U +#define ARCHIVE_STATE_HEADER 2U +#define ARCHIVE_STATE_DATA 4U +#define ARCHIVE_STATE_EOF 0x10U +#define ARCHIVE_STATE_CLOSED 0x20U +#define ARCHIVE_STATE_FATAL 0x8000U +#define ARCHIVE_STATE_ANY (0xFFFFU & ~ARCHIVE_STATE_FATAL) + +struct archive_vtable { + int (*archive_close)(struct archive *); + int (*archive_free)(struct archive *); + int (*archive_write_header)(struct archive *, + struct archive_entry *); + int (*archive_write_finish_entry)(struct archive *); + ssize_t (*archive_write_data)(struct archive *, + const void *, size_t); + ssize_t (*archive_write_data_block)(struct archive *, + const void *, size_t, int64_t); + + int (*archive_read_next_header)(struct archive *, + struct archive_entry **); + int (*archive_read_next_header2)(struct archive *, + struct archive_entry *); + int (*archive_read_data_block)(struct archive *, + const void **, size_t *, int64_t *); + + int (*archive_filter_count)(struct archive *); + int64_t (*archive_filter_bytes)(struct archive *, int); + int (*archive_filter_code)(struct archive *, int); + const char * (*archive_filter_name)(struct archive *, int); +}; + +struct archive_string_conv; + +struct archive { + /* + * The magic/state values are used to sanity-check the + * client's usage. If an API function is called at a + * ridiculous time, or the client passes us an invalid + * pointer, these values allow me to catch that. + */ + unsigned int magic; + unsigned int state; + + /* + * Some public API functions depend on the "real" type of the + * archive object. + */ + struct archive_vtable *vtable; + + int archive_format; + const char *archive_format_name; + + int compression_code; /* Currently active compression. */ + const char *compression_name; + + /* Number of file entries processed. */ + int file_count; + + int archive_error_number; + const char *error; + struct archive_string error_string; + + char *current_code; + unsigned current_codepage; /* Current ACP(ANSI CodePage). */ + unsigned current_oemcp; /* Current OEMCP(OEM CodePage). */ + struct archive_string_conv *sconv; + + /* + * Used by archive_read_data() to track blocks and copy + * data to client buffers, filling gaps with zero bytes. + */ + const char *read_data_block; + int64_t read_data_offset; + int64_t read_data_output_offset; + size_t read_data_remaining; + + /* + * Used by formats/filters to determine the amount of data + * requested from a call to archive_read_data(). This is only + * useful when the format/filter has seek support. + */ + char read_data_is_posix_read; + size_t read_data_requested; +}; + +/* Check magic value and state; return(ARCHIVE_FATAL) if it isn't valid. */ +int __archive_check_magic(struct archive *, unsigned int magic, + unsigned int state, const char *func); +#define archive_check_magic(a, expected_magic, allowed_states, function_name) \ + do { \ + int magic_test = __archive_check_magic((a), (expected_magic), \ + (allowed_states), (function_name)); \ + if (magic_test == ARCHIVE_FATAL) \ + return ARCHIVE_FATAL; \ + } while (0) + +void __archive_errx(int retvalue, const char *msg) __LA_DEAD; + +void __archive_ensure_cloexec_flag(int fd); +int __archive_mktemp(const char *tmpdir); +#if defined(_WIN32) && !defined(__CYGWIN__) +int __archive_mkstemp(wchar_t *template); +#else +int __archive_mkstemp(char *template); +#endif + +int __archive_clean(struct archive *); + +void __archive_reset_read_data(struct archive *); + +#define err_combine(a,b) ((a) < (b) ? (a) : (b)) + +#if defined(__BORLANDC__) || (defined(_MSC_VER) && _MSC_VER <= 1300) +# define ARCHIVE_LITERAL_LL(x) x##i64 +# define ARCHIVE_LITERAL_ULL(x) x##ui64 +#else +# define ARCHIVE_LITERAL_LL(x) x##ll +# define ARCHIVE_LITERAL_ULL(x) x##ull +#endif + +#endif diff --git a/c/archive_random.c b/c/archive_random.c new file mode 100644 --- /dev/null +++ b/c/archive_random.c @@ -0,0 +1,272 @@ +/*- + * Copyright (c) 2014 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_STDLIB_H +#include +#endif + +#if !defined(HAVE_ARC4RANDOM_BUF) && (!defined(_WIN32) || defined(__CYGWIN__)) + +#ifdef HAVE_FCNTL +#include +#endif +#ifdef HAVE_LIMITS_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif +#ifdef HAVE_SYS_TYPES_H +#include +#endif +#ifdef HAVE_SYS_TIME_H +#include +#endif +#ifdef HAVE_PTHREAD_H +#include +#endif + +static void arc4random_buf(void *, size_t); + +#endif /* HAVE_ARC4RANDOM_BUF */ + +#include "archive.h" +#include "archive_random_private.h" + +#if defined(HAVE_WINCRYPT_H) && !defined(__CYGWIN__) +#include +#endif + +#ifndef O_CLOEXEC +#define O_CLOEXEC 0 +#endif + +/* + * Random number generator function. + * This simply calls arc4random_buf function if the platform provides it. + */ + +int +archive_random(void *buf, size_t nbytes) +{ +#if defined(_WIN32) && !defined(__CYGWIN__) + HCRYPTPROV hProv; + BOOL success; + + success = CryptAcquireContext(&hProv, NULL, NULL, PROV_RSA_FULL, + CRYPT_VERIFYCONTEXT); + if (!success && GetLastError() == (DWORD)NTE_BAD_KEYSET) { + success = CryptAcquireContext(&hProv, NULL, NULL, + PROV_RSA_FULL, CRYPT_NEWKEYSET); + } + if (success) { + success = CryptGenRandom(hProv, (DWORD)nbytes, (BYTE*)buf); + CryptReleaseContext(hProv, 0); + if (success) + return ARCHIVE_OK; + } + /* TODO: Does this case really happen? */ + return ARCHIVE_FAILED; +#else + arc4random_buf(buf, nbytes); + return ARCHIVE_OK; +#endif +} + +#if !defined(HAVE_ARC4RANDOM_BUF) && (!defined(_WIN32) || defined(__CYGWIN__)) + +/* $OpenBSD: arc4random.c,v 1.24 2013/06/11 16:59:50 deraadt Exp $ */ +/* + * Copyright (c) 1996, David Mazieres + * Copyright (c) 2008, Damien Miller + * + * Permission to use, copy, modify, and distribute this software for any + * purpose with or without fee is hereby granted, provided that the above + * copyright notice and this permission notice appear in all copies. + * + * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES + * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF + * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR + * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN + * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF + * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. + */ + +/* + * Arc4 random number generator for OpenBSD. + * + * This code is derived from section 17.1 of Applied Cryptography, + * second edition, which describes a stream cipher allegedly + * compatible with RSA Labs "RC4" cipher (the actual description of + * which is a trade secret). The same algorithm is used as a stream + * cipher called "arcfour" in Tatu Ylonen's ssh package. + * + * RC4 is a registered trademark of RSA Laboratories. + */ + +#ifdef __GNUC__ +#define inline __inline +#else /* !__GNUC__ */ +#define inline +#endif /* !__GNUC__ */ + +struct arc4_stream { + uint8_t i; + uint8_t j; + uint8_t s[256]; +}; + +#define RANDOMDEV "/dev/urandom" +#define KEYSIZE 128 +#ifdef HAVE_PTHREAD_H +static pthread_mutex_t arc4random_mtx = PTHREAD_MUTEX_INITIALIZER; +#define _ARC4_LOCK() pthread_mutex_lock(&arc4random_mtx); +#define _ARC4_UNLOCK() pthread_mutex_unlock(&arc4random_mtx); +#else +#define _ARC4_LOCK() +#define _ARC4_UNLOCK() +#endif + +static int rs_initialized; +static struct arc4_stream rs; +static pid_t arc4_stir_pid; +static int arc4_count; + +static inline uint8_t arc4_getbyte(void); +static void arc4_stir(void); + +static inline void +arc4_init(void) +{ + int n; + + for (n = 0; n < 256; n++) + rs.s[n] = n; + rs.i = 0; + rs.j = 0; +} + +static inline void +arc4_addrandom(u_char *dat, int datlen) +{ + int n; + uint8_t si; + + rs.i--; + for (n = 0; n < 256; n++) { + rs.i = (rs.i + 1); + si = rs.s[rs.i]; + rs.j = (rs.j + si + dat[n % datlen]); + rs.s[rs.i] = rs.s[rs.j]; + rs.s[rs.j] = si; + } + rs.j = rs.i; +} + +static void +arc4_stir(void) +{ + int done, fd, i; + struct { + struct timeval tv; + pid_t pid; + u_char rnd[KEYSIZE]; + } rdat; + + if (!rs_initialized) { + arc4_init(); + rs_initialized = 1; + } + done = 0; + fd = open(RANDOMDEV, O_RDONLY | O_CLOEXEC, 0); + if (fd >= 0) { + if (read(fd, &rdat, KEYSIZE) == KEYSIZE) + done = 1; + (void)close(fd); + } + if (!done) { + (void)gettimeofday(&rdat.tv, NULL); + rdat.pid = getpid(); + /* We'll just take whatever was on the stack too... */ + } + + arc4_addrandom((u_char *)&rdat, KEYSIZE); + + /* + * Discard early keystream, as per recommendations in: + * "(Not So) Random Shuffles of RC4" by Ilya Mironov. + * As per the Network Operations Division, cryptographic requirements + * published on wikileaks on March 2017. + */ + + for (i = 0; i < 3072; i++) + (void)arc4_getbyte(); + arc4_count = 1600000; +} + +static void +arc4_stir_if_needed(void) +{ + pid_t pid = getpid(); + + if (arc4_count <= 0 || !rs_initialized || arc4_stir_pid != pid) { + arc4_stir_pid = pid; + arc4_stir(); + } +} + +static inline uint8_t +arc4_getbyte(void) +{ + uint8_t si, sj; + + rs.i = (rs.i + 1); + si = rs.s[rs.i]; + rs.j = (rs.j + si); + sj = rs.s[rs.j]; + rs.s[rs.i] = sj; + rs.s[rs.j] = si; + return (rs.s[(si + sj) & 0xff]); +} + +static void +arc4random_buf(void *_buf, size_t n) +{ + u_char *buf = (u_char *)_buf; + _ARC4_LOCK(); + arc4_stir_if_needed(); + while (n--) { + if (--arc4_count <= 0) + arc4_stir(); + buf[n] = arc4_getbyte(); + } + _ARC4_UNLOCK(); +} + +#endif /* !HAVE_ARC4RANDOM_BUF */ diff --git a/c/archive_random_private.h b/c/archive_random_private.h new file mode 100644 --- /dev/null +++ b/c/archive_random_private.h @@ -0,0 +1,36 @@ +/*- + * Copyright (c) 2014 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#ifndef ARCHIVE_RANDOM_PRIVATE_H_INCLUDED +#define ARCHIVE_RANDOM_PRIVATE_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif + +/* Random number generator. */ +int archive_random(void *buf, size_t nbytes); + +#endif /* ARCHIVE_RANDOM_PRIVATE_H_INCLUDED */ diff --git a/c/archive_rb.c b/c/archive_rb.c new file mode 100644 --- /dev/null +++ b/c/archive_rb.c @@ -0,0 +1,709 @@ +/*- + * Copyright (c) 2001 The NetBSD Foundation, Inc. + * All rights reserved. + * + * This code is derived from software contributed to The NetBSD Foundation + * by Matt Thomas . + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS + * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED + * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS + * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR + * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF + * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS + * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN + * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) + * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE + * POSSIBILITY OF SUCH DAMAGE. + * + * Based on: NetBSD: rb.c,v 1.6 2010/04/30 13:58:09 joerg Exp + */ + +#include "archive_platform.h" + +#include + +#include "archive_rb.h" + +/* Keep in sync with archive_rb.h */ +#define RB_DIR_LEFT 0 +#define RB_DIR_RIGHT 1 +#define RB_DIR_OTHER 1 +#define rb_left rb_nodes[RB_DIR_LEFT] +#define rb_right rb_nodes[RB_DIR_RIGHT] + +#define RB_FLAG_POSITION 0x2 +#define RB_FLAG_RED 0x1 +#define RB_FLAG_MASK (RB_FLAG_POSITION|RB_FLAG_RED) +#define RB_FATHER(rb) \ + ((struct archive_rb_node *)((rb)->rb_info & ~RB_FLAG_MASK)) +#define RB_SET_FATHER(rb, father) \ + ((void)((rb)->rb_info = (uintptr_t)(father)|((rb)->rb_info & RB_FLAG_MASK))) + +#define RB_SENTINEL_P(rb) ((rb) == NULL) +#define RB_LEFT_SENTINEL_P(rb) RB_SENTINEL_P((rb)->rb_left) +#define RB_RIGHT_SENTINEL_P(rb) RB_SENTINEL_P((rb)->rb_right) +#define RB_FATHER_SENTINEL_P(rb) RB_SENTINEL_P(RB_FATHER((rb))) +#define RB_CHILDLESS_P(rb) \ + (RB_SENTINEL_P(rb) || (RB_LEFT_SENTINEL_P(rb) && RB_RIGHT_SENTINEL_P(rb))) +#define RB_TWOCHILDREN_P(rb) \ + (!RB_SENTINEL_P(rb) && !RB_LEFT_SENTINEL_P(rb) && !RB_RIGHT_SENTINEL_P(rb)) + +#define RB_POSITION(rb) \ + (((rb)->rb_info & RB_FLAG_POSITION) ? RB_DIR_RIGHT : RB_DIR_LEFT) +#define RB_RIGHT_P(rb) (RB_POSITION(rb) == RB_DIR_RIGHT) +#define RB_LEFT_P(rb) (RB_POSITION(rb) == RB_DIR_LEFT) +#define RB_RED_P(rb) (!RB_SENTINEL_P(rb) && ((rb)->rb_info & RB_FLAG_RED) != 0) +#define RB_BLACK_P(rb) (RB_SENTINEL_P(rb) || ((rb)->rb_info & RB_FLAG_RED) == 0) +#define RB_MARK_RED(rb) ((void)((rb)->rb_info |= RB_FLAG_RED)) +#define RB_MARK_BLACK(rb) ((void)((rb)->rb_info &= ~RB_FLAG_RED)) +#define RB_INVERT_COLOR(rb) ((void)((rb)->rb_info ^= RB_FLAG_RED)) +#define RB_ROOT_P(rbt, rb) ((rbt)->rbt_root == (rb)) +#define RB_SET_POSITION(rb, position) \ + ((void)((position) ? ((rb)->rb_info |= RB_FLAG_POSITION) : \ + ((rb)->rb_info &= ~RB_FLAG_POSITION))) +#define RB_ZERO_PROPERTIES(rb) ((void)((rb)->rb_info &= ~RB_FLAG_MASK)) +#define RB_COPY_PROPERTIES(dst, src) \ + ((void)((dst)->rb_info ^= ((dst)->rb_info ^ (src)->rb_info) & RB_FLAG_MASK)) +#define RB_SWAP_PROPERTIES(a, b) do { \ + uintptr_t xorinfo = ((a)->rb_info ^ (b)->rb_info) & RB_FLAG_MASK; \ + (a)->rb_info ^= xorinfo; \ + (b)->rb_info ^= xorinfo; \ + } while (/*CONSTCOND*/ 0) + +static void __archive_rb_tree_insert_rebalance(struct archive_rb_tree *, + struct archive_rb_node *); +static void __archive_rb_tree_removal_rebalance(struct archive_rb_tree *, + struct archive_rb_node *, unsigned int); + +#define RB_SENTINEL_NODE NULL + +#define T 1 +#define F 0 + +void +__archive_rb_tree_init(struct archive_rb_tree *rbt, + const struct archive_rb_tree_ops *ops) +{ + rbt->rbt_ops = ops; + *((struct archive_rb_node **)&rbt->rbt_root) = RB_SENTINEL_NODE; +} + +struct archive_rb_node * +__archive_rb_tree_find_node(struct archive_rb_tree *rbt, const void *key) +{ + archive_rbto_compare_key_fn compare_key = rbt->rbt_ops->rbto_compare_key; + struct archive_rb_node *parent = rbt->rbt_root; + + while (!RB_SENTINEL_P(parent)) { + const signed int diff = (*compare_key)(parent, key); + if (diff == 0) + return parent; + parent = parent->rb_nodes[diff > 0]; + } + + return NULL; +} + +struct archive_rb_node * +__archive_rb_tree_find_node_geq(struct archive_rb_tree *rbt, const void *key) +{ + archive_rbto_compare_key_fn compare_key = rbt->rbt_ops->rbto_compare_key; + struct archive_rb_node *parent = rbt->rbt_root; + struct archive_rb_node *last = NULL; + + while (!RB_SENTINEL_P(parent)) { + const signed int diff = (*compare_key)(parent, key); + if (diff == 0) + return parent; + if (diff < 0) + last = parent; + parent = parent->rb_nodes[diff > 0]; + } + + return last; +} + +struct archive_rb_node * +__archive_rb_tree_find_node_leq(struct archive_rb_tree *rbt, const void *key) +{ + archive_rbto_compare_key_fn compare_key = rbt->rbt_ops->rbto_compare_key; + struct archive_rb_node *parent = rbt->rbt_root; + struct archive_rb_node *last = NULL; + + while (!RB_SENTINEL_P(parent)) { + const signed int diff = (*compare_key)(parent, key); + if (diff == 0) + return parent; + if (diff > 0) + last = parent; + parent = parent->rb_nodes[diff > 0]; + } + + return last; +} + +int +__archive_rb_tree_insert_node(struct archive_rb_tree *rbt, + struct archive_rb_node *self) +{ + archive_rbto_compare_nodes_fn compare_nodes = rbt->rbt_ops->rbto_compare_nodes; + struct archive_rb_node *parent, *tmp; + unsigned int position; + int rebalance; + + tmp = rbt->rbt_root; + /* + * This is a hack. Because rbt->rbt_root is just a + * struct archive_rb_node *, just like rb_node->rb_nodes[RB_DIR_LEFT], + * we can use this fact to avoid a lot of tests for root and know + * that even at root, updating + * RB_FATHER(rb_node)->rb_nodes[RB_POSITION(rb_node)] will + * update rbt->rbt_root. + */ + parent = (struct archive_rb_node *)(void *)&rbt->rbt_root; + position = RB_DIR_LEFT; + + /* + * Find out where to place this new leaf. + */ + while (!RB_SENTINEL_P(tmp)) { + const signed int diff = (*compare_nodes)(tmp, self); + if (diff == 0) { + /* + * Node already exists; don't insert. + */ + return F; + } + parent = tmp; + position = (diff > 0); + tmp = parent->rb_nodes[position]; + } + + /* + * Initialize the node and insert as a leaf into the tree. + */ + RB_SET_FATHER(self, parent); + RB_SET_POSITION(self, position); + if (parent == (struct archive_rb_node *)(void *)&rbt->rbt_root) { + RB_MARK_BLACK(self); /* root is always black */ + rebalance = F; + } else { + /* + * All new nodes are colored red. We only need to rebalance + * if our parent is also red. + */ + RB_MARK_RED(self); + rebalance = RB_RED_P(parent); + } + self->rb_left = parent->rb_nodes[position]; + self->rb_right = parent->rb_nodes[position]; + parent->rb_nodes[position] = self; + + /* + * Rebalance tree after insertion + */ + if (rebalance) + __archive_rb_tree_insert_rebalance(rbt, self); + + return T; +} + +/* + * Swap the location and colors of 'self' and its child @ which. The child + * can not be a sentinel node. This is our rotation function. However, + * since it preserves coloring, it great simplifies both insertion and + * removal since rotation almost always involves the exchanging of colors + * as a separate step. + */ +/*ARGSUSED*/ +static void +__archive_rb_tree_reparent_nodes( + struct archive_rb_node *old_father, const unsigned int which) +{ + const unsigned int other = which ^ RB_DIR_OTHER; + struct archive_rb_node * const grandpa = RB_FATHER(old_father); + struct archive_rb_node * const old_child = old_father->rb_nodes[which]; + struct archive_rb_node * const new_father = old_child; + struct archive_rb_node * const new_child = old_father; + + if (new_father == NULL) + return; + /* + * Exchange descendant linkages. + */ + grandpa->rb_nodes[RB_POSITION(old_father)] = new_father; + new_child->rb_nodes[which] = old_child->rb_nodes[other]; + new_father->rb_nodes[other] = new_child; + + /* + * Update ancestor linkages + */ + RB_SET_FATHER(new_father, grandpa); + RB_SET_FATHER(new_child, new_father); + + /* + * Exchange properties between new_father and new_child. The only + * change is that new_child's position is now on the other side. + */ + RB_SWAP_PROPERTIES(new_father, new_child); + RB_SET_POSITION(new_child, other); + + /* + * Make sure to reparent the new child to ourself. + */ + if (!RB_SENTINEL_P(new_child->rb_nodes[which])) { + RB_SET_FATHER(new_child->rb_nodes[which], new_child); + RB_SET_POSITION(new_child->rb_nodes[which], which); + } + +} + +static void +__archive_rb_tree_insert_rebalance(struct archive_rb_tree *rbt, + struct archive_rb_node *self) +{ + struct archive_rb_node * father = RB_FATHER(self); + struct archive_rb_node * grandpa; + struct archive_rb_node * uncle; + unsigned int which; + unsigned int other; + + for (;;) { + /* + * We are red and our parent is red, therefore we must have a + * grandfather and he must be black. + */ + grandpa = RB_FATHER(father); + which = (father == grandpa->rb_right); + other = which ^ RB_DIR_OTHER; + uncle = grandpa->rb_nodes[other]; + + if (RB_BLACK_P(uncle)) + break; + + /* + * Case 1: our uncle is red + * Simply invert the colors of our parent and + * uncle and make our grandparent red. And + * then solve the problem up at his level. + */ + RB_MARK_BLACK(uncle); + RB_MARK_BLACK(father); + if (RB_ROOT_P(rbt, grandpa)) { + /* + * If our grandpa is root, don't bother + * setting him to red, just return. + */ + return; + } + RB_MARK_RED(grandpa); + self = grandpa; + father = RB_FATHER(self); + if (RB_BLACK_P(father)) { + /* + * If our great-grandpa is black, we're done. + */ + return; + } + } + + /* + * Case 2&3: our uncle is black. + */ + if (self == father->rb_nodes[other]) { + /* + * Case 2: we are on the same side as our uncle + * Swap ourselves with our parent so this case + * becomes case 3. Basically our parent becomes our + * child. + */ + __archive_rb_tree_reparent_nodes(father, other); + } + /* + * Case 3: we are opposite a child of a black uncle. + * Swap our parent and grandparent. Since our grandfather + * is black, our father will become black and our new sibling + * (former grandparent) will become red. + */ + __archive_rb_tree_reparent_nodes(grandpa, which); + + /* + * Final step: Set the root to black. + */ + RB_MARK_BLACK(rbt->rbt_root); +} + +static void +__archive_rb_tree_prune_node(struct archive_rb_tree *rbt, + struct archive_rb_node *self, int rebalance) +{ + const unsigned int which = RB_POSITION(self); + struct archive_rb_node *father = RB_FATHER(self); + + /* + * Since we are childless, we know that self->rb_left is pointing + * to the sentinel node. + */ + father->rb_nodes[which] = self->rb_left; + + /* + * Rebalance if requested. + */ + if (rebalance) + __archive_rb_tree_removal_rebalance(rbt, father, which); +} + +/* + * When deleting an interior node + */ +static void +__archive_rb_tree_swap_prune_and_rebalance(struct archive_rb_tree *rbt, + struct archive_rb_node *self, struct archive_rb_node *standin) +{ + const unsigned int standin_which = RB_POSITION(standin); + unsigned int standin_other = standin_which ^ RB_DIR_OTHER; + struct archive_rb_node *standin_son; + struct archive_rb_node *standin_father = RB_FATHER(standin); + int rebalance = RB_BLACK_P(standin); + + if (standin_father == self) { + /* + * As a child of self, any children would be opposite of + * our parent. + */ + standin_son = standin->rb_nodes[standin_which]; + } else { + /* + * Since we aren't a child of self, any children would be + * on the same side as our parent. + */ + standin_son = standin->rb_nodes[standin_other]; + } + + if (RB_RED_P(standin_son)) { + /* + * We know we have a red child so if we flip it to black + * we don't have to rebalance. + */ + RB_MARK_BLACK(standin_son); + rebalance = F; + + if (standin_father != self) { + /* + * Change the son's parentage to point to his grandpa. + */ + RB_SET_FATHER(standin_son, standin_father); + RB_SET_POSITION(standin_son, standin_which); + } + } + + if (standin_father == self) { + /* + * If we are about to delete the standin's father, then when + * we call rebalance, we need to use ourselves as our father. + * Otherwise remember our original father. Also, since we are + * our standin's father we only need to reparent the standin's + * brother. + * + * | R --> S | + * | Q S --> Q T | + * | t --> | + * + * Have our son/standin adopt his brother as his new son. + */ + standin_father = standin; + } else { + /* + * | R --> S . | + * | / \ | T --> / \ | / | + * | ..... | S --> ..... | T | + * + * Sever standin's connection to his father. + */ + standin_father->rb_nodes[standin_which] = standin_son; + /* + * Adopt the far son. + */ + standin->rb_nodes[standin_other] = self->rb_nodes[standin_other]; + RB_SET_FATHER(standin->rb_nodes[standin_other], standin); + /* + * Use standin_other because we need to preserve standin_which + * for the removal_rebalance. + */ + standin_other = standin_which; + } + + /* + * Move the only remaining son to our standin. If our standin is our + * son, this will be the only son needed to be moved. + */ + standin->rb_nodes[standin_other] = self->rb_nodes[standin_other]; + RB_SET_FATHER(standin->rb_nodes[standin_other], standin); + + /* + * Now copy the result of self to standin and then replace + * self with standin in the tree. + */ + RB_COPY_PROPERTIES(standin, self); + RB_SET_FATHER(standin, RB_FATHER(self)); + RB_FATHER(standin)->rb_nodes[RB_POSITION(standin)] = standin; + + if (rebalance) + __archive_rb_tree_removal_rebalance(rbt, standin_father, standin_which); +} + +/* + * We could do this by doing + * __archive_rb_tree_node_swap(rbt, self, which); + * __archive_rb_tree_prune_node(rbt, self, F); + * + * But it's more efficient to just evaluate and recolor the child. + */ +static void +__archive_rb_tree_prune_blackred_branch( + struct archive_rb_node *self, unsigned int which) +{ + struct archive_rb_node *father = RB_FATHER(self); + struct archive_rb_node *son = self->rb_nodes[which]; + + /* + * Remove ourselves from the tree and give our former child our + * properties (position, color, root). + */ + RB_COPY_PROPERTIES(son, self); + father->rb_nodes[RB_POSITION(son)] = son; + RB_SET_FATHER(son, father); +} +/* + * + */ +void +__archive_rb_tree_remove_node(struct archive_rb_tree *rbt, + struct archive_rb_node *self) +{ + struct archive_rb_node *standin; + unsigned int which; + + /* + * In the following diagrams, we (the node to be removed) are S. Red + * nodes are lowercase. T could be either red or black. + * + * Remember the major axiom of the red-black tree: the number of + * black nodes from the root to each leaf is constant across all + * leaves, only the number of red nodes varies. + * + * Thus removing a red leaf doesn't require any other changes to a + * red-black tree. So if we must remove a node, attempt to rearrange + * the tree so we can remove a red node. + * + * The simplest case is a childless red node or a childless root node: + * + * | T --> T | or | R --> * | + * | s --> * | + */ + if (RB_CHILDLESS_P(self)) { + const int rebalance = RB_BLACK_P(self) && !RB_ROOT_P(rbt, self); + __archive_rb_tree_prune_node(rbt, self, rebalance); + return; + } + if (!RB_TWOCHILDREN_P(self)) { + /* + * The next simplest case is the node we are deleting is + * black and has one red child. + * + * | T --> T --> T | + * | S --> R --> R | + * | r --> s --> * | + */ + which = RB_LEFT_SENTINEL_P(self) ? RB_DIR_RIGHT : RB_DIR_LEFT; + __archive_rb_tree_prune_blackred_branch(self, which); + return; + } + + /* + * We invert these because we prefer to remove from the inside of + * the tree. + */ + which = RB_POSITION(self) ^ RB_DIR_OTHER; + + /* + * Let's find the node closes to us opposite of our parent + * Now swap it with ourself, "prune" it, and rebalance, if needed. + */ + standin = __archive_rb_tree_iterate(rbt, self, which); + __archive_rb_tree_swap_prune_and_rebalance(rbt, self, standin); +} + +static void +__archive_rb_tree_removal_rebalance(struct archive_rb_tree *rbt, + struct archive_rb_node *parent, unsigned int which) +{ + + while (RB_BLACK_P(parent->rb_nodes[which])) { + unsigned int other = which ^ RB_DIR_OTHER; + struct archive_rb_node *brother = parent->rb_nodes[other]; + + if (brother == NULL) + return;/* The tree may be broken. */ + /* + * For cases 1, 2a, and 2b, our brother's children must + * be black and our father must be black + */ + if (RB_BLACK_P(parent) + && RB_BLACK_P(brother->rb_left) + && RB_BLACK_P(brother->rb_right)) { + if (RB_RED_P(brother)) { + /* + * Case 1: Our brother is red, swap its + * position (and colors) with our parent. + * This should now be case 2b (unless C or E + * has a red child which is case 3; thus no + * explicit branch to case 2b). + * + * B -> D + * A d -> b E + * C E -> A C + */ + __archive_rb_tree_reparent_nodes(parent, other); + brother = parent->rb_nodes[other]; + if (brother == NULL) + return;/* The tree may be broken. */ + } else { + /* + * Both our parent and brother are black. + * Change our brother to red, advance up rank + * and go through the loop again. + * + * B -> *B + * *A D -> A d + * C E -> C E + */ + RB_MARK_RED(brother); + if (RB_ROOT_P(rbt, parent)) + return; /* root == parent == black */ + which = RB_POSITION(parent); + parent = RB_FATHER(parent); + continue; + } + } + /* + * Avoid an else here so that case 2a above can hit either + * case 2b, 3, or 4. + */ + if (RB_RED_P(parent) + && RB_BLACK_P(brother) + && RB_BLACK_P(brother->rb_left) + && RB_BLACK_P(brother->rb_right)) { + /* + * We are black, our father is red, our brother and + * both nephews are black. Simply invert/exchange the + * colors of our father and brother (to black and red + * respectively). + * + * | f --> F | + * | * B --> * b | + * | N N --> N N | + */ + RB_MARK_BLACK(parent); + RB_MARK_RED(brother); + break; /* We're done! */ + } else { + /* + * Our brother must be black and have at least one + * red child (it may have two). + */ + if (RB_BLACK_P(brother->rb_nodes[other])) { + /* + * Case 3: our brother is black, our near + * nephew is red, and our far nephew is black. + * Swap our brother with our near nephew. + * This result in a tree that matches case 4. + * (Our father could be red or black). + * + * | F --> F | + * | x B --> x B | + * | n --> n | + */ + __archive_rb_tree_reparent_nodes(brother, which); + brother = parent->rb_nodes[other]; + } + /* + * Case 4: our brother is black and our far nephew + * is red. Swap our father and brother locations and + * change our far nephew to black. (these can be + * done in either order so we change the color first). + * The result is a valid red-black tree and is a + * terminal case. (again we don't care about the + * father's color) + * + * If the father is red, we will get a red-black-black + * tree: + * | f -> f --> b | + * | B -> B --> F N | + * | n -> N --> | + * + * If the father is black, we will get an all black + * tree: + * | F -> F --> B | + * | B -> B --> F N | + * | n -> N --> | + * + * If we had two red nephews, then after the swap, + * our former father would have a red grandson. + */ + if (brother->rb_nodes[other] == NULL) + return;/* The tree may be broken. */ + RB_MARK_BLACK(brother->rb_nodes[other]); + __archive_rb_tree_reparent_nodes(parent, other); + break; /* We're done! */ + } + } +} + +struct archive_rb_node * +__archive_rb_tree_iterate(struct archive_rb_tree *rbt, + struct archive_rb_node *self, const unsigned int direction) +{ + const unsigned int other = direction ^ RB_DIR_OTHER; + + if (self == NULL) { + self = rbt->rbt_root; + if (RB_SENTINEL_P(self)) + return NULL; + while (!RB_SENTINEL_P(self->rb_nodes[direction])) + self = self->rb_nodes[direction]; + return self; + } + /* + * We can't go any further in this direction. We proceed up in the + * opposite direction until our parent is in direction we want to go. + */ + if (RB_SENTINEL_P(self->rb_nodes[direction])) { + while (!RB_ROOT_P(rbt, self)) { + if (other == (unsigned int)RB_POSITION(self)) + return RB_FATHER(self); + self = RB_FATHER(self); + } + return NULL; + } + + /* + * Advance down one in current direction and go down as far as possible + * in the opposite direction. + */ + self = self->rb_nodes[direction]; + while (!RB_SENTINEL_P(self->rb_nodes[other])) + self = self->rb_nodes[other]; + return self; +} diff --git a/c/archive_rb.h b/c/archive_rb.h new file mode 100644 --- /dev/null +++ b/c/archive_rb.h @@ -0,0 +1,113 @@ +/*- + * Copyright (c) 2001 The NetBSD Foundation, Inc. + * All rights reserved. + * + * This code is derived from software contributed to The NetBSD Foundation + * by Matt Thomas . + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS + * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED + * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS + * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR + * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF + * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS + * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN + * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) + * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE + * POSSIBILITY OF SUCH DAMAGE. + * + * Based on NetBSD: rb.h,v 1.13 2009/08/16 10:57:01 yamt Exp + */ + +#ifndef ARCHIVE_RB_H_INCLUDED +#define ARCHIVE_RB_H_INCLUDED + +struct archive_rb_node { + struct archive_rb_node *rb_nodes[2]; + /* + * rb_info contains the two flags and the parent back pointer. + * We put the two flags in the low two bits since we know that + * rb_node will have an alignment of 4 or 8 bytes. + */ + uintptr_t rb_info; +}; + +#define ARCHIVE_RB_DIR_LEFT 0 +#define ARCHIVE_RB_DIR_RIGHT 1 + +#define ARCHIVE_RB_TREE_MIN(T) \ + __archive_rb_tree_iterate((T), NULL, ARCHIVE_RB_DIR_LEFT) +#define ARCHIVE_RB_TREE_MAX(T) \ + __archive_rb_tree_iterate((T), NULL, ARCHIVE_RB_DIR_RIGHT) +#define ARCHIVE_RB_TREE_NEXT(T, N) \ + __archive_rb_tree_iterate((T), (N), ARCHIVE_RB_DIR_RIGHT) +#define ARCHIVE_RB_TREE_PREV(T, N) \ + __archive_rb_tree_iterate((T), (N), ARCHIVE_RB_DIR_LEFT) +#define ARCHIVE_RB_TREE_FOREACH(N, T) \ + for ((N) = ARCHIVE_RB_TREE_MIN(T); (N); \ + (N) = ARCHIVE_RB_TREE_NEXT((T), (N))) +#define ARCHIVE_RB_TREE_FOREACH_REVERSE(N, T) \ + for ((N) = ARCHIVE_RB_TREE_MAX(T); (N); \ + (N) = ARCHIVE_RB_TREE_PREV((T), (N))) +#define ARCHIVE_RB_TREE_FOREACH_SAFE(N, T, S) \ + for ((N) = ARCHIVE_RB_TREE_MIN(T); \ + (N) && ((S) = ARCHIVE_RB_TREE_NEXT((T), (N)), 1); \ + (N) = (S)) +#define ARCHIVE_RB_TREE_FOREACH_REVERSE_SAFE(N, T, S) \ + for ((N) = ARCHIVE_RB_TREE_MAX(T); \ + (N) && ((S) = ARCHIVE_RB_TREE_PREV((T), (N)), 1); \ + (N) = (S)) + +/* + * archive_rbto_compare_nodes_fn: + * return a positive value if the first node < the second node. + * return a negative value if the first node > the second node. + * return 0 if they are considered same. + * + * archive_rbto_compare_key_fn: + * return a positive value if the node < the key. + * return a negative value if the node > the key. + * return 0 if they are considered same. + */ + +typedef signed int (*const archive_rbto_compare_nodes_fn)(const struct archive_rb_node *, + const struct archive_rb_node *); +typedef signed int (*const archive_rbto_compare_key_fn)(const struct archive_rb_node *, + const void *); + +struct archive_rb_tree_ops { + archive_rbto_compare_nodes_fn rbto_compare_nodes; + archive_rbto_compare_key_fn rbto_compare_key; +}; + +struct archive_rb_tree { + struct archive_rb_node *rbt_root; + const struct archive_rb_tree_ops *rbt_ops; +}; + +void __archive_rb_tree_init(struct archive_rb_tree *, + const struct archive_rb_tree_ops *); +int __archive_rb_tree_insert_node(struct archive_rb_tree *, + struct archive_rb_node *); +struct archive_rb_node * + __archive_rb_tree_find_node(struct archive_rb_tree *, const void *); +struct archive_rb_node * + __archive_rb_tree_find_node_geq(struct archive_rb_tree *, const void *); +struct archive_rb_node * + __archive_rb_tree_find_node_leq(struct archive_rb_tree *, const void *); +void __archive_rb_tree_remove_node(struct archive_rb_tree *, struct archive_rb_node *); +struct archive_rb_node * + __archive_rb_tree_iterate(struct archive_rb_tree *, + struct archive_rb_node *, const unsigned int); + +#endif /* ARCHIVE_RB_H_*/ diff --git a/c/archive_read.c b/c/archive_read.c new file mode 100644 --- /dev/null +++ b/c/archive_read.c @@ -0,0 +1,1752 @@ +/*- + * Copyright (c) 2003-2011 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +/* + * This file contains the "essential" portions of the read API, that + * is, stuff that will probably always be used by any client that + * actually needs to read an archive. Optional pieces have been, as + * far as possible, separated out into separate files to avoid + * needlessly bloating statically-linked clients. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_read.c 201157 2009-12-29 05:30:23Z kientzle $"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_read_private.h" + +#define minimum(a, b) (a < b ? a : b) + +static int choose_filters(struct archive_read *); +static int choose_format(struct archive_read *); +static int close_filters(struct archive_read *); +static struct archive_vtable *archive_read_vtable(void); +static int64_t _archive_filter_bytes(struct archive *, int); +static int _archive_filter_code(struct archive *, int); +static const char *_archive_filter_name(struct archive *, int); +static int _archive_filter_count(struct archive *); +static int _archive_read_close(struct archive *); +static int _archive_read_data_block(struct archive *, + const void **, size_t *, int64_t *); +static int _archive_read_free(struct archive *); +static int _archive_read_next_header(struct archive *, + struct archive_entry **); +static int _archive_read_next_header2(struct archive *, + struct archive_entry *); +static int64_t advance_file_pointer(struct archive_read_filter *, int64_t); + +static struct archive_vtable * +archive_read_vtable(void) +{ + static struct archive_vtable av; + static int inited = 0; + + if (!inited) { + av.archive_filter_bytes = _archive_filter_bytes; + av.archive_filter_code = _archive_filter_code; + av.archive_filter_name = _archive_filter_name; + av.archive_filter_count = _archive_filter_count; + av.archive_read_data_block = _archive_read_data_block; + av.archive_read_next_header = _archive_read_next_header; + av.archive_read_next_header2 = _archive_read_next_header2; + av.archive_free = _archive_read_free; + av.archive_close = _archive_read_close; + inited = 1; + } + return (&av); +} + +/* + * Allocate, initialize and return a struct archive object. + */ +struct archive * +archive_read_new(void) +{ + struct archive_read *a; + + a = (struct archive_read *)calloc(1, sizeof(*a)); + if (a == NULL) + return (NULL); + a->archive.magic = ARCHIVE_READ_MAGIC; + + a->archive.state = ARCHIVE_STATE_NEW; + a->entry = archive_entry_new2(&a->archive); + a->archive.vtable = archive_read_vtable(); + + a->passphrases.last = &a->passphrases.first; + + return (&a->archive); +} + +/* + * Record the do-not-extract-to file. This belongs in archive_read_extract.c. + */ +void +archive_read_extract_set_skip_file(struct archive *_a, la_int64_t d, + la_int64_t i) +{ + struct archive_read *a = (struct archive_read *)_a; + + if (ARCHIVE_OK != __archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_extract_set_skip_file")) + return; + a->skip_file_set = 1; + a->skip_file_dev = d; + a->skip_file_ino = i; +} + +/* + * Open the archive + */ +int +archive_read_open(struct archive *a, void *client_data, + archive_open_callback *client_opener, archive_read_callback *client_reader, + archive_close_callback *client_closer) +{ + /* Old archive_read_open() is just a thin shell around + * archive_read_open1. */ + archive_read_set_open_callback(a, client_opener); + archive_read_set_read_callback(a, client_reader); + archive_read_set_close_callback(a, client_closer); + archive_read_set_callback_data(a, client_data); + return archive_read_open1(a); +} + + +int +archive_read_open2(struct archive *a, void *client_data, + archive_open_callback *client_opener, + archive_read_callback *client_reader, + archive_skip_callback *client_skipper, + archive_close_callback *client_closer) +{ + /* Old archive_read_open2() is just a thin shell around + * archive_read_open1. */ + archive_read_set_callback_data(a, client_data); + archive_read_set_open_callback(a, client_opener); + archive_read_set_read_callback(a, client_reader); + archive_read_set_skip_callback(a, client_skipper); + archive_read_set_close_callback(a, client_closer); + return archive_read_open1(a); +} + +static ssize_t +client_read_proxy(struct archive_read_filter *self, const void **buff) +{ + ssize_t r; + r = (self->archive->client.reader)(&self->archive->archive, + self->data, buff); + return (r); +} + +static int64_t +client_skip_proxy(struct archive_read_filter *self, int64_t request) +{ + if (request < 0) + __archive_errx(1, "Negative skip requested."); + if (request == 0) + return 0; + + if (self->archive->client.skipper != NULL) { + /* Seek requests over 1GiB are broken down into + * multiple seeks. This avoids overflows when the + * requests get passed through 32-bit arguments. */ + int64_t skip_limit = (int64_t)1 << 30; + int64_t total = 0; + for (;;) { + int64_t get, ask = request; + if (ask > skip_limit) + ask = skip_limit; + get = (self->archive->client.skipper) + (&self->archive->archive, self->data, ask); + total += get; + if (get == 0 || get == request) + return (total); + if (get > request) + return ARCHIVE_FATAL; + request -= get; + } + } else if (self->archive->client.seeker != NULL + && request > 64 * 1024) { + /* If the client provided a seeker but not a skipper, + * we can use the seeker to skip forward. + * + * Note: This isn't always a good idea. The client + * skipper is allowed to skip by less than requested + * if it needs to maintain block alignment. The + * seeker is not allowed to play such games, so using + * the seeker here may be a performance loss compared + * to just reading and discarding. That's why we + * only do this for skips of over 64k. + */ + int64_t before = self->position; + int64_t after = (self->archive->client.seeker) + (&self->archive->archive, self->data, request, SEEK_CUR); + if (after != before + request) + return ARCHIVE_FATAL; + return after - before; + } + return 0; +} + +static int64_t +client_seek_proxy(struct archive_read_filter *self, int64_t offset, int whence) +{ + /* DO NOT use the skipper here! If we transparently handled + * forward seek here by using the skipper, that will break + * other libarchive code that assumes a successful forward + * seek means it can also seek backwards. + */ + if (self->archive->client.seeker == NULL) { + archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, + "Current client reader does not support seeking a device"); + return (ARCHIVE_FAILED); + } + return (self->archive->client.seeker)(&self->archive->archive, + self->data, offset, whence); +} + +static int +client_close_proxy(struct archive_read_filter *self) +{ + int r = ARCHIVE_OK, r2; + unsigned int i; + + if (self->archive->client.closer == NULL) + return (r); + for (i = 0; i < self->archive->client.nodes; i++) + { + r2 = (self->archive->client.closer) + ((struct archive *)self->archive, + self->archive->client.dataset[i].data); + if (r > r2) + r = r2; + } + return (r); +} + +static int +client_open_proxy(struct archive_read_filter *self) +{ + int r = ARCHIVE_OK; + if (self->archive->client.opener != NULL) + r = (self->archive->client.opener)( + (struct archive *)self->archive, self->data); + return (r); +} + +static int +client_switch_proxy(struct archive_read_filter *self, unsigned int iindex) +{ + int r1 = ARCHIVE_OK, r2 = ARCHIVE_OK; + void *data2 = NULL; + + /* Don't do anything if already in the specified data node */ + if (self->archive->client.cursor == iindex) + return (ARCHIVE_OK); + + self->archive->client.cursor = iindex; + data2 = self->archive->client.dataset[self->archive->client.cursor].data; + if (self->archive->client.switcher != NULL) + { + r1 = r2 = (self->archive->client.switcher) + ((struct archive *)self->archive, self->data, data2); + self->data = data2; + } + else + { + /* Attempt to call close and open instead */ + if (self->archive->client.closer != NULL) + r1 = (self->archive->client.closer) + ((struct archive *)self->archive, self->data); + self->data = data2; + if (self->archive->client.opener != NULL) + r2 = (self->archive->client.opener) + ((struct archive *)self->archive, self->data); + } + return (r1 < r2) ? r1 : r2; +} + +int +archive_read_set_open_callback(struct archive *_a, + archive_open_callback *client_opener) +{ + struct archive_read *a = (struct archive_read *)_a; + archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, + "archive_read_set_open_callback"); + a->client.opener = client_opener; + return ARCHIVE_OK; +} + +int +archive_read_set_read_callback(struct archive *_a, + archive_read_callback *client_reader) +{ + struct archive_read *a = (struct archive_read *)_a; + archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, + "archive_read_set_read_callback"); + a->client.reader = client_reader; + return ARCHIVE_OK; +} + +int +archive_read_set_skip_callback(struct archive *_a, + archive_skip_callback *client_skipper) +{ + struct archive_read *a = (struct archive_read *)_a; + archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, + "archive_read_set_skip_callback"); + a->client.skipper = client_skipper; + return ARCHIVE_OK; +} + +int +archive_read_set_seek_callback(struct archive *_a, + archive_seek_callback *client_seeker) +{ + struct archive_read *a = (struct archive_read *)_a; + archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, + "archive_read_set_seek_callback"); + a->client.seeker = client_seeker; + return ARCHIVE_OK; +} + +int +archive_read_set_close_callback(struct archive *_a, + archive_close_callback *client_closer) +{ + struct archive_read *a = (struct archive_read *)_a; + archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, + "archive_read_set_close_callback"); + a->client.closer = client_closer; + return ARCHIVE_OK; +} + +int +archive_read_set_switch_callback(struct archive *_a, + archive_switch_callback *client_switcher) +{ + struct archive_read *a = (struct archive_read *)_a; + archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, + "archive_read_set_switch_callback"); + a->client.switcher = client_switcher; + return ARCHIVE_OK; +} + +int +archive_read_set_callback_data(struct archive *_a, void *client_data) +{ + return archive_read_set_callback_data2(_a, client_data, 0); +} + +int +archive_read_set_callback_data2(struct archive *_a, void *client_data, + unsigned int iindex) +{ + struct archive_read *a = (struct archive_read *)_a; + archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, + "archive_read_set_callback_data2"); + + if (a->client.nodes == 0) + { + a->client.dataset = (struct archive_read_data_node *) + calloc(1, sizeof(*a->client.dataset)); + if (a->client.dataset == NULL) + { + archive_set_error(&a->archive, ENOMEM, + "No memory."); + return ARCHIVE_FATAL; + } + a->client.nodes = 1; + } + + if (iindex > a->client.nodes - 1) + { + archive_set_error(&a->archive, EINVAL, + "Invalid index specified."); + return ARCHIVE_FATAL; + } + a->client.dataset[iindex].data = client_data; + a->client.dataset[iindex].begin_position = -1; + a->client.dataset[iindex].total_size = -1; + return ARCHIVE_OK; +} + +int +archive_read_add_callback_data(struct archive *_a, void *client_data, + unsigned int iindex) +{ + struct archive_read *a = (struct archive_read *)_a; + void *p; + unsigned int i; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, + "archive_read_add_callback_data"); + if (iindex > a->client.nodes) { + archive_set_error(&a->archive, EINVAL, + "Invalid index specified."); + return ARCHIVE_FATAL; + } + p = realloc(a->client.dataset, sizeof(*a->client.dataset) + * (++(a->client.nodes))); + if (p == NULL) { + archive_set_error(&a->archive, ENOMEM, + "No memory."); + return ARCHIVE_FATAL; + } + a->client.dataset = (struct archive_read_data_node *)p; + for (i = a->client.nodes - 1; i > iindex; i--) { + a->client.dataset[i].data = a->client.dataset[i-1].data; + a->client.dataset[i].begin_position = -1; + a->client.dataset[i].total_size = -1; + } + a->client.dataset[iindex].data = client_data; + a->client.dataset[iindex].begin_position = -1; + a->client.dataset[iindex].total_size = -1; + return ARCHIVE_OK; +} + +int +archive_read_append_callback_data(struct archive *_a, void *client_data) +{ + struct archive_read *a = (struct archive_read *)_a; + return archive_read_add_callback_data(_a, client_data, a->client.nodes); +} + +int +archive_read_prepend_callback_data(struct archive *_a, void *client_data) +{ + return archive_read_add_callback_data(_a, client_data, 0); +} + +int +archive_read_open1(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_filter *filter, *tmp; + int slot, e = ARCHIVE_OK; + unsigned int i; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, + "archive_read_open"); + archive_clear_error(&a->archive); + + if (a->client.reader == NULL) { + archive_set_error(&a->archive, EINVAL, + "No reader function provided to archive_read_open"); + a->archive.state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + } + + /* Open data source. */ + if (a->client.opener != NULL) { + e = (a->client.opener)(&a->archive, a->client.dataset[0].data); + if (e != 0) { + /* If the open failed, call the closer to clean up. */ + if (a->client.closer) { + for (i = 0; i < a->client.nodes; i++) + (a->client.closer)(&a->archive, + a->client.dataset[i].data); + } + return (e); + } + } + + filter = calloc(1, sizeof(*filter)); + if (filter == NULL) + return (ARCHIVE_FATAL); + filter->bidder = NULL; + filter->upstream = NULL; + filter->archive = a; + filter->data = a->client.dataset[0].data; + filter->open = client_open_proxy; + filter->read = client_read_proxy; + filter->skip = client_skip_proxy; + filter->seek = client_seek_proxy; + filter->close = client_close_proxy; + filter->sswitch = client_switch_proxy; + filter->name = "none"; + filter->code = ARCHIVE_FILTER_NONE; + + a->client.dataset[0].begin_position = 0; + if (!a->filter || !a->bypass_filter_bidding) + { + a->filter = filter; + /* Build out the input pipeline. */ + e = choose_filters(a); + if (e < ARCHIVE_WARN) { + a->archive.state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + } + } + else + { + /* Need to add "NONE" type filter at the end of the filter chain */ + tmp = a->filter; + while (tmp->upstream) + tmp = tmp->upstream; + tmp->upstream = filter; + } + + if (!a->format) + { + slot = choose_format(a); + if (slot < 0) { + close_filters(a); + a->archive.state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + } + a->format = &(a->formats[slot]); + } + + a->archive.state = ARCHIVE_STATE_HEADER; + + /* Ensure libarchive starts from the first node in a multivolume set */ + client_switch_proxy(a->filter, 0); + return (e); +} + +/* + * Allow each registered stream transform to bid on whether + * it wants to handle this stream. Repeat until we've finished + * building the pipeline. + */ + +/* We won't build a filter pipeline with more stages than this. */ +#define MAX_NUMBER_FILTERS 25 + +static int +choose_filters(struct archive_read *a) +{ + int number_bidders, i, bid, best_bid, number_filters; + struct archive_read_filter_bidder *bidder, *best_bidder; + struct archive_read_filter *filter; + ssize_t avail; + int r; + + for (number_filters = 0; number_filters < MAX_NUMBER_FILTERS; ++number_filters) { + number_bidders = sizeof(a->bidders) / sizeof(a->bidders[0]); + + best_bid = 0; + best_bidder = NULL; + + bidder = a->bidders; + for (i = 0; i < number_bidders; i++, bidder++) { + if (bidder->bid != NULL) { + bid = (bidder->bid)(bidder, a->filter); + if (bid > best_bid) { + best_bid = bid; + best_bidder = bidder; + } + } + } + + /* If no bidder, we're done. */ + if (best_bidder == NULL) { + /* Verify the filter by asking it for some data. */ + __archive_read_filter_ahead(a->filter, 1, &avail); + if (avail < 0) { + __archive_read_free_filters(a); + return (ARCHIVE_FATAL); + } + a->archive.compression_name = a->filter->name; + a->archive.compression_code = a->filter->code; + return (ARCHIVE_OK); + } + + filter + = (struct archive_read_filter *)calloc(1, sizeof(*filter)); + if (filter == NULL) + return (ARCHIVE_FATAL); + filter->bidder = best_bidder; + filter->archive = a; + filter->upstream = a->filter; + a->filter = filter; + r = (best_bidder->init)(a->filter); + if (r != ARCHIVE_OK) { + __archive_read_free_filters(a); + return (ARCHIVE_FATAL); + } + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Input requires too many filters for decoding"); + return (ARCHIVE_FATAL); +} + +int +__archive_read_header(struct archive_read *a, struct archive_entry *entry) +{ + if (a->filter->read_header) + return a->filter->read_header(a->filter, entry); + else + return (ARCHIVE_OK); +} + +/* + * Read header of next entry. + */ +static int +_archive_read_next_header2(struct archive *_a, struct archive_entry *entry) +{ + struct archive_read *a = (struct archive_read *)_a; + int r1 = ARCHIVE_OK, r2; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, + "archive_read_next_header"); + + archive_entry_clear(entry); + archive_clear_error(&a->archive); + + /* + * If client didn't consume entire data, skip any remainder + * (This is especially important for GNU incremental directories.) + */ + if (a->archive.state == ARCHIVE_STATE_DATA) { + r1 = archive_read_data_skip(&a->archive); + if (r1 == ARCHIVE_EOF) + archive_set_error(&a->archive, EIO, + "Premature end-of-file."); + if (r1 == ARCHIVE_EOF || r1 == ARCHIVE_FATAL) { + a->archive.state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + } + } + + /* Record start-of-header offset in uncompressed stream. */ + a->header_position = a->filter->position; + + ++_a->file_count; + r2 = (a->format->read_header)(a, entry); + + /* + * EOF and FATAL are persistent at this layer. By + * modifying the state, we guarantee that future calls to + * read a header or read data will fail. + */ + switch (r2) { + case ARCHIVE_EOF: + a->archive.state = ARCHIVE_STATE_EOF; + --_a->file_count;/* Revert a file counter. */ + break; + case ARCHIVE_OK: + a->archive.state = ARCHIVE_STATE_DATA; + break; + case ARCHIVE_WARN: + a->archive.state = ARCHIVE_STATE_DATA; + break; + case ARCHIVE_RETRY: + break; + case ARCHIVE_FATAL: + a->archive.state = ARCHIVE_STATE_FATAL; + break; + } + + __archive_reset_read_data(&a->archive); + + a->data_start_node = a->client.cursor; + /* EOF always wins; otherwise return the worst error. */ + return (r2 < r1 || r2 == ARCHIVE_EOF) ? r2 : r1; +} + +static int +_archive_read_next_header(struct archive *_a, struct archive_entry **entryp) +{ + int ret; + struct archive_read *a = (struct archive_read *)_a; + *entryp = NULL; + ret = _archive_read_next_header2(_a, a->entry); + *entryp = a->entry; + return ret; +} + +/* + * Allow each registered format to bid on whether it wants to handle + * the next entry. Return index of winning bidder. + */ +static int +choose_format(struct archive_read *a) +{ + int slots; + int i; + int bid, best_bid; + int best_bid_slot; + + slots = sizeof(a->formats) / sizeof(a->formats[0]); + best_bid = -1; + best_bid_slot = -1; + + /* Set up a->format for convenience of bidders. */ + a->format = &(a->formats[0]); + for (i = 0; i < slots; i++, a->format++) { + if (a->format->bid) { + bid = (a->format->bid)(a, best_bid); + if (bid == ARCHIVE_FATAL) + return (ARCHIVE_FATAL); + if (a->filter->position != 0) + __archive_read_seek(a, 0, SEEK_SET); + if ((bid > best_bid) || (best_bid_slot < 0)) { + best_bid = bid; + best_bid_slot = i; + } + } + } + + /* + * There were no bidders; this is a serious programmer error + * and demands a quick and definitive abort. + */ + if (best_bid_slot < 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "No formats registered"); + return (ARCHIVE_FATAL); + } + + /* + * There were bidders, but no non-zero bids; this means we + * can't support this stream. + */ + if (best_bid < 1) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Unrecognized archive format"); + return (ARCHIVE_FATAL); + } + + return (best_bid_slot); +} + +/* + * Return the file offset (within the uncompressed data stream) where + * the last header started. + */ +la_int64_t +archive_read_header_position(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_header_position"); + return (a->header_position); +} + +/* + * Returns 1 if the archive contains at least one encrypted entry. + * If the archive format not support encryption at all + * ARCHIVE_READ_FORMAT_ENCRYPTION_UNSUPPORTED is returned. + * If for any other reason (e.g. not enough data read so far) + * we cannot say whether there are encrypted entries, then + * ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW is returned. + * In general, this function will return values below zero when the + * reader is uncertain or totally incapable of encryption support. + * When this function returns 0 you can be sure that the reader + * supports encryption detection but no encrypted entries have + * been found yet. + * + * NOTE: If the metadata/header of an archive is also encrypted, you + * cannot rely on the number of encrypted entries. That is why this + * function does not return the number of encrypted entries but# + * just shows that there are some. + */ +int +archive_read_has_encrypted_entries(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + int format_supports_encryption = archive_read_format_capabilities(_a) + & (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); + + if (!_a || !format_supports_encryption) { + /* Format in general doesn't support encryption */ + return ARCHIVE_READ_FORMAT_ENCRYPTION_UNSUPPORTED; + } + + /* A reader potentially has read enough data now. */ + if (a->format && a->format->has_encrypted_entries) { + return (a->format->has_encrypted_entries)(a); + } + + /* For any other reason we cannot say how many entries are there. */ + return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; +} + +/* + * Returns a bitmask of capabilities that are supported by the archive format reader. + * If the reader has no special capabilities, ARCHIVE_READ_FORMAT_CAPS_NONE is returned. + */ +int +archive_read_format_capabilities(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + if (a && a->format && a->format->format_capabilties) { + return (a->format->format_capabilties)(a); + } + return ARCHIVE_READ_FORMAT_CAPS_NONE; +} + +/* + * Read data from an archive entry, using a read(2)-style interface. + * This is a convenience routine that just calls + * archive_read_data_block and copies the results into the client + * buffer, filling any gaps with zero bytes. Clients using this + * API can be completely ignorant of sparse-file issues; sparse files + * will simply be padded with nulls. + * + * DO NOT intermingle calls to this function and archive_read_data_block + * to read a single entry body. + */ +la_ssize_t +archive_read_data(struct archive *_a, void *buff, size_t s) +{ + struct archive *a = (struct archive *)_a; + char *dest; + const void *read_buf; + size_t bytes_read; + size_t len; + int r; + + bytes_read = 0; + dest = (char *)buff; + + while (s > 0) { + if (a->read_data_offset == a->read_data_output_offset && + a->read_data_remaining == 0) { + read_buf = a->read_data_block; + a->read_data_is_posix_read = 1; + a->read_data_requested = s; + r = archive_read_data_block(a, &read_buf, + &a->read_data_remaining, &a->read_data_offset); + a->read_data_block = read_buf; + if (r == ARCHIVE_EOF) + return (bytes_read); + /* + * Error codes are all negative, so the status + * return here cannot be confused with a valid + * byte count. (ARCHIVE_OK is zero.) + */ + if (r < ARCHIVE_OK) + return (r); + } + + if (a->read_data_offset < a->read_data_output_offset) { + archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT, + "Encountered out-of-order sparse blocks"); + return (ARCHIVE_RETRY); + } + + /* Compute the amount of zero padding needed. */ + if (a->read_data_output_offset + (int64_t)s < + a->read_data_offset) { + len = s; + } else if (a->read_data_output_offset < + a->read_data_offset) { + len = (size_t)(a->read_data_offset - + a->read_data_output_offset); + } else + len = 0; + + /* Add zeroes. */ + memset(dest, 0, len); + s -= len; + a->read_data_output_offset += len; + dest += len; + bytes_read += len; + + /* Copy data if there is any space left. */ + if (s > 0) { + len = a->read_data_remaining; + if (len > s) + len = s; + if (len) { + memcpy(dest, a->read_data_block, len); + s -= len; + a->read_data_block += len; + a->read_data_remaining -= len; + a->read_data_output_offset += len; + a->read_data_offset += len; + dest += len; + bytes_read += len; + } + } + } + a->read_data_is_posix_read = 0; + a->read_data_requested = 0; + return (bytes_read); +} + +/* + * Reset the read_data_* variables, used for starting a new entry. + */ +void __archive_reset_read_data(struct archive * a) +{ + a->read_data_output_offset = 0; + a->read_data_remaining = 0; + a->read_data_is_posix_read = 0; + a->read_data_requested = 0; + + /* extra resets, from rar.c */ + a->read_data_block = NULL; + a->read_data_offset = 0; +} + +/* + * Skip over all remaining data in this entry. + */ +int +archive_read_data_skip(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + int r; + const void *buff; + size_t size; + int64_t offset; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_DATA, + "archive_read_data_skip"); + + if (a->format->read_data_skip != NULL) + r = (a->format->read_data_skip)(a); + else { + while ((r = archive_read_data_block(&a->archive, + &buff, &size, &offset)) + == ARCHIVE_OK) + ; + } + + if (r == ARCHIVE_EOF) + r = ARCHIVE_OK; + + a->archive.state = ARCHIVE_STATE_HEADER; + return (r); +} + +la_int64_t +archive_seek_data(struct archive *_a, int64_t offset, int whence) +{ + struct archive_read *a = (struct archive_read *)_a; + archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_DATA, + "archive_seek_data_block"); + + if (a->format->seek_data == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, + "Internal error: " + "No format_seek_data_block function registered"); + return (ARCHIVE_FATAL); + } + + return (a->format->seek_data)(a, offset, whence); +} + +/* + * Read the next block of entry data from the archive. + * This is a zero-copy interface; the client receives a pointer, + * size, and file offset of the next available block of data. + * + * Returns ARCHIVE_OK if the operation is successful, ARCHIVE_EOF if + * the end of entry is encountered. + */ +static int +_archive_read_data_block(struct archive *_a, + const void **buff, size_t *size, int64_t *offset) +{ + struct archive_read *a = (struct archive_read *)_a; + archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_DATA, + "archive_read_data_block"); + + if (a->format->read_data == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, + "Internal error: " + "No format->read_data function registered"); + return (ARCHIVE_FATAL); + } + + return (a->format->read_data)(a, buff, size, offset); +} + +static int +close_filters(struct archive_read *a) +{ + struct archive_read_filter *f = a->filter; + int r = ARCHIVE_OK; + /* Close each filter in the pipeline. */ + while (f != NULL) { + struct archive_read_filter *t = f->upstream; + if (!f->closed && f->close != NULL) { + int r1 = (f->close)(f); + f->closed = 1; + if (r1 < r) + r = r1; + } + free(f->buffer); + f->buffer = NULL; + f = t; + } + return r; +} + +void +__archive_read_free_filters(struct archive_read *a) +{ + /* Make sure filters are closed and their buffers are freed */ + close_filters(a); + + while (a->filter != NULL) { + struct archive_read_filter *t = a->filter->upstream; + free(a->filter); + a->filter = t; + } +} + +/* + * return the count of # of filters in use + */ +static int +_archive_filter_count(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_filter *p = a->filter; + int count = 0; + while(p) { + count++; + p = p->upstream; + } + return count; +} + +/* + * Close the file and all I/O. + */ +static int +_archive_read_close(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + int r = ARCHIVE_OK, r1 = ARCHIVE_OK; + + archive_check_magic(&a->archive, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_read_close"); + if (a->archive.state == ARCHIVE_STATE_CLOSED) + return (ARCHIVE_OK); + archive_clear_error(&a->archive); + a->archive.state = ARCHIVE_STATE_CLOSED; + + /* TODO: Clean up the formatters. */ + + /* Release the filter objects. */ + r1 = close_filters(a); + if (r1 < r) + r = r1; + + return (r); +} + +/* + * Release memory and other resources. + */ +static int +_archive_read_free(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_passphrase *p; + int i, n; + int slots; + int r = ARCHIVE_OK; + + if (_a == NULL) + return (ARCHIVE_OK); + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_read_free"); + if (a->archive.state != ARCHIVE_STATE_CLOSED + && a->archive.state != ARCHIVE_STATE_FATAL) + r = archive_read_close(&a->archive); + + /* Call cleanup functions registered by optional components. */ + if (a->cleanup_archive_extract != NULL) + r = (a->cleanup_archive_extract)(a); + + /* Cleanup format-specific data. */ + slots = sizeof(a->formats) / sizeof(a->formats[0]); + for (i = 0; i < slots; i++) { + a->format = &(a->formats[i]); + if (a->formats[i].cleanup) + (a->formats[i].cleanup)(a); + } + + /* Free the filters */ + __archive_read_free_filters(a); + + /* Release the bidder objects. */ + n = sizeof(a->bidders)/sizeof(a->bidders[0]); + for (i = 0; i < n; i++) { + if (a->bidders[i].free != NULL) { + int r1 = (a->bidders[i].free)(&a->bidders[i]); + if (r1 < r) + r = r1; + } + } + + /* Release passphrase list. */ + p = a->passphrases.first; + while (p != NULL) { + struct archive_read_passphrase *np = p->next; + + /* A passphrase should be cleaned. */ + memset(p->passphrase, 0, strlen(p->passphrase)); + free(p->passphrase); + free(p); + p = np; + } + + archive_string_free(&a->archive.error_string); + archive_entry_free(a->entry); + a->archive.magic = 0; + __archive_clean(&a->archive); + free(a->client.dataset); + free(a); + return (r); +} + +static struct archive_read_filter * +get_filter(struct archive *_a, int n) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_filter *f = a->filter; + /* We use n == -1 for 'the last filter', which is always the + * client proxy. */ + if (n == -1 && f != NULL) { + struct archive_read_filter *last = f; + f = f->upstream; + while (f != NULL) { + last = f; + f = f->upstream; + } + return (last); + } + if (n < 0) + return NULL; + while (n > 0 && f != NULL) { + f = f->upstream; + --n; + } + return (f); +} + +static int +_archive_filter_code(struct archive *_a, int n) +{ + struct archive_read_filter *f = get_filter(_a, n); + return f == NULL ? -1 : f->code; +} + +static const char * +_archive_filter_name(struct archive *_a, int n) +{ + struct archive_read_filter *f = get_filter(_a, n); + return f != NULL ? f->name : NULL; +} + +static int64_t +_archive_filter_bytes(struct archive *_a, int n) +{ + struct archive_read_filter *f = get_filter(_a, n); + return f == NULL ? -1 : f->position; +} + +/* + * Used internally by read format handlers to register their bid and + * initialization functions. + */ +int +__archive_read_register_format(struct archive_read *a, + void *format_data, + const char *name, + int (*bid)(struct archive_read *, int), + int (*options)(struct archive_read *, const char *, const char *), + int (*read_header)(struct archive_read *, struct archive_entry *), + int (*read_data)(struct archive_read *, const void **, size_t *, int64_t *), + int (*read_data_skip)(struct archive_read *), + int64_t (*seek_data)(struct archive_read *, int64_t, int), + int (*cleanup)(struct archive_read *), + int (*format_capabilities)(struct archive_read *), + int (*has_encrypted_entries)(struct archive_read *)) +{ + int i, number_slots; + + archive_check_magic(&a->archive, + ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, + "__archive_read_register_format"); + + number_slots = sizeof(a->formats) / sizeof(a->formats[0]); + + for (i = 0; i < number_slots; i++) { + if (a->formats[i].bid == bid) + return (ARCHIVE_WARN); /* We've already installed */ + if (a->formats[i].bid == NULL) { + a->formats[i].bid = bid; + a->formats[i].options = options; + a->formats[i].read_header = read_header; + a->formats[i].read_data = read_data; + a->formats[i].read_data_skip = read_data_skip; + a->formats[i].seek_data = seek_data; + a->formats[i].cleanup = cleanup; + a->formats[i].data = format_data; + a->formats[i].name = name; + a->formats[i].format_capabilties = format_capabilities; + a->formats[i].has_encrypted_entries = has_encrypted_entries; + return (ARCHIVE_OK); + } + } + + archive_set_error(&a->archive, ENOMEM, + "Not enough slots for format registration"); + return (ARCHIVE_FATAL); +} + +/* + * Used internally by decompression routines to register their bid and + * initialization functions. + */ +int +__archive_read_get_bidder(struct archive_read *a, + struct archive_read_filter_bidder **bidder) +{ + int i, number_slots; + + number_slots = sizeof(a->bidders) / sizeof(a->bidders[0]); + + for (i = 0; i < number_slots; i++) { + if (a->bidders[i].bid == NULL) { + memset(a->bidders + i, 0, sizeof(a->bidders[0])); + *bidder = (a->bidders + i); + return (ARCHIVE_OK); + } + } + + archive_set_error(&a->archive, ENOMEM, + "Not enough slots for filter registration"); + return (ARCHIVE_FATAL); +} + +/* + * The next section implements the peek/consume internal I/O + * system used by archive readers. This system allows simple + * read-ahead for consumers while preserving zero-copy operation + * most of the time. + * + * The two key operations: + * * The read-ahead function returns a pointer to a block of data + * that satisfies a minimum request. + * * The consume function advances the file pointer. + * + * In the ideal case, filters generate blocks of data + * and __archive_read_ahead() just returns pointers directly into + * those blocks. Then __archive_read_consume() just bumps those + * pointers. Only if your request would span blocks does the I/O + * layer use a copy buffer to provide you with a contiguous block of + * data. + * + * A couple of useful idioms: + * * "I just want some data." Ask for 1 byte and pay attention to + * the "number of bytes available" from __archive_read_ahead(). + * Consume whatever you actually use. + * * "I want to output a large block of data." As above, ask for 1 byte, + * emit all that's available (up to whatever limit you have), consume + * it all, then repeat until you're done. This effectively means that + * you're passing along the blocks that came from your provider. + * * "I want to peek ahead by a large amount." Ask for 4k or so, then + * double and repeat until you get an error or have enough. Note + * that the I/O layer will likely end up expanding its copy buffer + * to fit your request, so use this technique cautiously. This + * technique is used, for example, by some of the format tasting + * code that has uncertain look-ahead needs. + */ + +/* + * Looks ahead in the input stream: + * * If 'avail' pointer is provided, that returns number of bytes available + * in the current buffer, which may be much larger than requested. + * * If end-of-file, *avail gets set to zero. + * * If error, *avail gets error code. + * * If request can be met, returns pointer to data. + * * If minimum request cannot be met, returns NULL. + * + * Note: If you just want "some data", ask for 1 byte and pay attention + * to *avail, which will have the actual amount available. If you + * know exactly how many bytes you need, just ask for that and treat + * a NULL return as an error. + * + * Important: This does NOT move the file pointer. See + * __archive_read_consume() below. + */ +const void * +__archive_read_ahead(struct archive_read *a, size_t min, ssize_t *avail) +{ + return (__archive_read_filter_ahead(a->filter, min, avail)); +} + +const void * +__archive_read_filter_ahead(struct archive_read_filter *filter, + size_t min, ssize_t *avail) +{ + ssize_t bytes_read; + size_t tocopy; + + if (filter->fatal) { + if (avail) + *avail = ARCHIVE_FATAL; + return (NULL); + } + + /* + * Keep pulling more data until we can satisfy the request. + */ + for (;;) { + + /* + * If we can satisfy from the copy buffer (and the + * copy buffer isn't empty), we're done. In particular, + * note that min == 0 is a perfectly well-defined + * request. + */ + if (filter->avail >= min && filter->avail > 0) { + if (avail != NULL) + *avail = filter->avail; + return (filter->next); + } + + /* + * We can satisfy directly from client buffer if everything + * currently in the copy buffer is still in the client buffer. + */ + if (filter->client_total >= filter->client_avail + filter->avail + && filter->client_avail + filter->avail >= min) { + /* "Roll back" to client buffer. */ + filter->client_avail += filter->avail; + filter->client_next -= filter->avail; + /* Copy buffer is now empty. */ + filter->avail = 0; + filter->next = filter->buffer; + /* Return data from client buffer. */ + if (avail != NULL) + *avail = filter->client_avail; + return (filter->client_next); + } + + /* Move data forward in copy buffer if necessary. */ + if (filter->next > filter->buffer && + filter->next + min > filter->buffer + filter->buffer_size) { + if (filter->avail > 0) + memmove(filter->buffer, filter->next, + filter->avail); + filter->next = filter->buffer; + } + + /* If we've used up the client data, get more. */ + if (filter->client_avail <= 0) { + if (filter->end_of_file) { + if (avail != NULL) + *avail = 0; + return (NULL); + } + bytes_read = (filter->read)(filter, + &filter->client_buff); + if (bytes_read < 0) { /* Read error. */ + filter->client_total = filter->client_avail = 0; + filter->client_next = + filter->client_buff = NULL; + filter->fatal = 1; + if (avail != NULL) + *avail = ARCHIVE_FATAL; + return (NULL); + } + if (bytes_read == 0) { + /* Check for another client object first */ + if (filter->archive->client.cursor != + filter->archive->client.nodes - 1) { + if (client_switch_proxy(filter, + filter->archive->client.cursor + 1) + == ARCHIVE_OK) + continue; + } + /* Premature end-of-file. */ + filter->client_total = filter->client_avail = 0; + filter->client_next = + filter->client_buff = NULL; + filter->end_of_file = 1; + /* Return whatever we do have. */ + if (avail != NULL) + *avail = filter->avail; + return (NULL); + } + filter->client_total = bytes_read; + filter->client_avail = filter->client_total; + filter->client_next = filter->client_buff; + } else { + /* + * We can't satisfy the request from the copy + * buffer or the existing client data, so we + * need to copy more client data over to the + * copy buffer. + */ + + /* Ensure the buffer is big enough. */ + if (min > filter->buffer_size) { + size_t s, t; + char *p; + + /* Double the buffer; watch for overflow. */ + s = t = filter->buffer_size; + if (s == 0) + s = min; + while (s < min) { + t *= 2; + if (t <= s) { /* Integer overflow! */ + archive_set_error( + &filter->archive->archive, + ENOMEM, + "Unable to allocate copy" + " buffer"); + filter->fatal = 1; + if (avail != NULL) + *avail = ARCHIVE_FATAL; + return (NULL); + } + s = t; + } + /* Now s >= min, so allocate a new buffer. */ + p = (char *)malloc(s); + if (p == NULL) { + archive_set_error( + &filter->archive->archive, + ENOMEM, + "Unable to allocate copy buffer"); + filter->fatal = 1; + if (avail != NULL) + *avail = ARCHIVE_FATAL; + return (NULL); + } + /* Move data into newly-enlarged buffer. */ + if (filter->avail > 0) + memmove(p, filter->next, filter->avail); + free(filter->buffer); + filter->next = filter->buffer = p; + filter->buffer_size = s; + } + + /* We can add client data to copy buffer. */ + /* First estimate: copy to fill rest of buffer. */ + tocopy = (filter->buffer + filter->buffer_size) + - (filter->next + filter->avail); + /* Don't waste time buffering more than we need to. */ + if (tocopy + filter->avail > min) + tocopy = min - filter->avail; + /* Don't copy more than is available. */ + if (tocopy > filter->client_avail) + tocopy = filter->client_avail; + + memcpy(filter->next + filter->avail, + filter->client_next, tocopy); + /* Remove this data from client buffer. */ + filter->client_next += tocopy; + filter->client_avail -= tocopy; + /* add it to copy buffer. */ + filter->avail += tocopy; + } + } +} + +/* + * Move the file pointer forward. + */ +int64_t +__archive_read_consume(struct archive_read *a, int64_t request) +{ + return (__archive_read_filter_consume(a->filter, request)); +} + +int64_t +__archive_read_filter_consume(struct archive_read_filter * filter, + int64_t request) +{ + int64_t skipped; + + if (request < 0) + return ARCHIVE_FATAL; + if (request == 0) + return 0; + + skipped = advance_file_pointer(filter, request); + if (skipped == request) + return (skipped); + /* We hit EOF before we satisfied the skip request. */ + if (skipped < 0) /* Map error code to 0 for error message below. */ + skipped = 0; + archive_set_error(&filter->archive->archive, + ARCHIVE_ERRNO_MISC, + "Truncated input file (needed %jd bytes, only %jd available)", + (intmax_t)request, (intmax_t)skipped); + return (ARCHIVE_FATAL); +} + +/* + * Advance the file pointer by the amount requested. + * Returns the amount actually advanced, which may be less than the + * request if EOF is encountered first. + * Returns a negative value if there's an I/O error. + */ +static int64_t +advance_file_pointer(struct archive_read_filter *filter, int64_t request) +{ + int64_t bytes_skipped, total_bytes_skipped = 0; + ssize_t bytes_read; + size_t min; + + if (filter->fatal) + return (-1); + + /* Use up the copy buffer first. */ + if (filter->avail > 0) { + min = (size_t)minimum(request, (int64_t)filter->avail); + filter->next += min; + filter->avail -= min; + request -= min; + filter->position += min; + total_bytes_skipped += min; + } + + /* Then use up the client buffer. */ + if (filter->client_avail > 0) { + min = (size_t)minimum(request, (int64_t)filter->client_avail); + filter->client_next += min; + filter->client_avail -= min; + request -= min; + filter->position += min; + total_bytes_skipped += min; + } + if (request == 0) + return (total_bytes_skipped); + + /* If there's an optimized skip function, use it. */ + if (filter->skip != NULL) { + bytes_skipped = (filter->skip)(filter, request); + if (bytes_skipped < 0) { /* error */ + filter->fatal = 1; + return (bytes_skipped); + } + filter->position += bytes_skipped; + total_bytes_skipped += bytes_skipped; + request -= bytes_skipped; + if (request == 0) + return (total_bytes_skipped); + } + + /* Use ordinary reads as necessary to complete the request. */ + for (;;) { + bytes_read = (filter->read)(filter, &filter->client_buff); + if (bytes_read < 0) { + filter->client_buff = NULL; + filter->fatal = 1; + return (bytes_read); + } + + if (bytes_read == 0) { + if (filter->archive->client.cursor != + filter->archive->client.nodes - 1) { + if (client_switch_proxy(filter, + filter->archive->client.cursor + 1) + == ARCHIVE_OK) + continue; + } + filter->client_buff = NULL; + filter->end_of_file = 1; + return (total_bytes_skipped); + } + + if (bytes_read >= request) { + filter->client_next = + ((const char *)filter->client_buff) + request; + filter->client_avail = (size_t)(bytes_read - request); + filter->client_total = bytes_read; + total_bytes_skipped += request; + filter->position += request; + return (total_bytes_skipped); + } + + filter->position += bytes_read; + total_bytes_skipped += bytes_read; + request -= bytes_read; + } +} + +/** + * Returns ARCHIVE_FAILED if seeking isn't supported. + */ +int64_t +__archive_read_seek(struct archive_read *a, int64_t offset, int whence) +{ + return __archive_read_filter_seek(a->filter, offset, whence); +} + +int64_t +__archive_read_filter_seek(struct archive_read_filter *filter, int64_t offset, + int whence) +{ + struct archive_read_client *client; + int64_t r; + unsigned int cursor; + + if (filter->closed || filter->fatal) + return (ARCHIVE_FATAL); + if (filter->seek == NULL) + return (ARCHIVE_FAILED); + + client = &(filter->archive->client); + switch (whence) { + case SEEK_CUR: + /* Adjust the offset and use SEEK_SET instead */ + offset += filter->position; + __LA_FALLTHROUGH; + case SEEK_SET: + cursor = 0; + while (1) + { + if (client->dataset[cursor].begin_position < 0 || + client->dataset[cursor].total_size < 0 || + client->dataset[cursor].begin_position + + client->dataset[cursor].total_size - 1 > offset || + cursor + 1 >= client->nodes) + break; + r = client->dataset[cursor].begin_position + + client->dataset[cursor].total_size; + client->dataset[++cursor].begin_position = r; + } + while (1) { + r = client_switch_proxy(filter, cursor); + if (r != ARCHIVE_OK) + return r; + if ((r = client_seek_proxy(filter, 0, SEEK_END)) < 0) + return r; + client->dataset[cursor].total_size = r; + if (client->dataset[cursor].begin_position + + client->dataset[cursor].total_size - 1 > offset || + cursor + 1 >= client->nodes) + break; + r = client->dataset[cursor].begin_position + + client->dataset[cursor].total_size; + client->dataset[++cursor].begin_position = r; + } + offset -= client->dataset[cursor].begin_position; + if (offset < 0 + || offset > client->dataset[cursor].total_size) + return ARCHIVE_FATAL; + if ((r = client_seek_proxy(filter, offset, SEEK_SET)) < 0) + return r; + break; + + case SEEK_END: + cursor = 0; + while (1) { + if (client->dataset[cursor].begin_position < 0 || + client->dataset[cursor].total_size < 0 || + cursor + 1 >= client->nodes) + break; + r = client->dataset[cursor].begin_position + + client->dataset[cursor].total_size; + client->dataset[++cursor].begin_position = r; + } + while (1) { + r = client_switch_proxy(filter, cursor); + if (r != ARCHIVE_OK) + return r; + if ((r = client_seek_proxy(filter, 0, SEEK_END)) < 0) + return r; + client->dataset[cursor].total_size = r; + r = client->dataset[cursor].begin_position + + client->dataset[cursor].total_size; + if (cursor + 1 >= client->nodes) + break; + client->dataset[++cursor].begin_position = r; + } + while (1) { + if (r + offset >= + client->dataset[cursor].begin_position) + break; + offset += client->dataset[cursor].total_size; + if (cursor == 0) + break; + cursor--; + r = client->dataset[cursor].begin_position + + client->dataset[cursor].total_size; + } + offset = (r + offset) - client->dataset[cursor].begin_position; + if ((r = client_switch_proxy(filter, cursor)) != ARCHIVE_OK) + return r; + r = client_seek_proxy(filter, offset, SEEK_SET); + if (r < ARCHIVE_OK) + return r; + break; + + default: + return (ARCHIVE_FATAL); + } + r += client->dataset[cursor].begin_position; + + if (r >= 0) { + /* + * Ouch. Clearing the buffer like this hurts, especially + * at bid time. A lot of our efficiency at bid time comes + * from having bidders reuse the data we've already read. + * + * TODO: If the seek request is in data we already + * have, then don't call the seek callback. + * + * TODO: Zip seeks to end-of-file at bid time. If + * other formats also start doing this, we may need to + * find a way for clients to fudge the seek offset to + * a block boundary. + * + * Hmmm... If whence was SEEK_END, we know the file + * size is (r - offset). Can we use that to simplify + * the TODO items above? + */ + filter->avail = filter->client_avail = 0; + filter->next = filter->buffer; + filter->position = r; + filter->end_of_file = 0; + } + return r; +} diff --git a/c/archive_read_add_passphrase.c b/c/archive_read_add_passphrase.c new file mode 100644 --- /dev/null +++ b/c/archive_read_add_passphrase.c @@ -0,0 +1,190 @@ +/*- + * Copyright (c) 2014 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#include "archive_read_private.h" + +static void +add_passphrase_to_tail(struct archive_read *a, + struct archive_read_passphrase *p) +{ + *a->passphrases.last = p; + a->passphrases.last = &p->next; + p->next = NULL; +} + +static struct archive_read_passphrase * +remove_passphrases_from_head(struct archive_read *a) +{ + struct archive_read_passphrase *p; + + p = a->passphrases.first; + if (p != NULL) + a->passphrases.first = p->next; + return (p); +} + +static void +insert_passphrase_to_head(struct archive_read *a, + struct archive_read_passphrase *p) +{ + p->next = a->passphrases.first; + a->passphrases.first = p; + if (&a->passphrases.first == a->passphrases.last) { + a->passphrases.last = &p->next; + p->next = NULL; + } +} + +static struct archive_read_passphrase * +new_read_passphrase(struct archive_read *a, const char *passphrase) +{ + struct archive_read_passphrase *p; + + p = malloc(sizeof(*p)); + if (p == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (NULL); + } + p->passphrase = strdup(passphrase); + if (p->passphrase == NULL) { + free(p); + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (NULL); + } + return (p); +} + +int +archive_read_add_passphrase(struct archive *_a, const char *passphrase) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_passphrase *p; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, + "archive_read_add_passphrase"); + + if (passphrase == NULL || passphrase[0] == '\0') { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Empty passphrase is unacceptable"); + return (ARCHIVE_FAILED); + } + + p = new_read_passphrase(a, passphrase); + if (p == NULL) + return (ARCHIVE_FATAL); + add_passphrase_to_tail(a, p); + + return (ARCHIVE_OK); +} + +int +archive_read_set_passphrase_callback(struct archive *_a, void *client_data, + archive_passphrase_callback *cb) +{ + struct archive_read *a = (struct archive_read *)_a; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, + "archive_read_set_passphrase_callback"); + + a->passphrases.callback = cb; + a->passphrases.client_data = client_data; + return (ARCHIVE_OK); +} + +/* + * Call this in advance when you start to get a passphrase for decryption + * for a entry. + */ +void +__archive_read_reset_passphrase(struct archive_read *a) +{ + + a->passphrases.candidate = -1; +} + +/* + * Get a passphrase for decryption. + */ +const char * +__archive_read_next_passphrase(struct archive_read *a) +{ + struct archive_read_passphrase *p; + const char *passphrase; + + if (a->passphrases.candidate < 0) { + /* Count out how many passphrases we have. */ + int cnt = 0; + + for (p = a->passphrases.first; p != NULL; p = p->next) + cnt++; + a->passphrases.candidate = cnt; + p = a->passphrases.first; + } else if (a->passphrases.candidate > 1) { + /* Rotate a passphrase list. */ + a->passphrases.candidate--; + p = remove_passphrases_from_head(a); + add_passphrase_to_tail(a, p); + /* Pick a new passphrase candidate up. */ + p = a->passphrases.first; + } else if (a->passphrases.candidate == 1) { + /* This case is that all candidates failed to decrypt. */ + a->passphrases.candidate = 0; + if (a->passphrases.first->next != NULL) { + /* Rotate a passphrase list. */ + p = remove_passphrases_from_head(a); + add_passphrase_to_tail(a, p); + } + p = NULL; + } else /* There is no passphrase candidate. */ + p = NULL; + + if (p != NULL) + passphrase = p->passphrase; + else if (a->passphrases.callback != NULL) { + /* Get a passphrase through a call-back function + * since we tried all passphrases out or we don't + * have it. */ + passphrase = a->passphrases.callback(&a->archive, + a->passphrases.client_data); + if (passphrase != NULL) { + p = new_read_passphrase(a, passphrase); + if (p == NULL) + return (NULL); + insert_passphrase_to_head(a, p); + a->passphrases.candidate = 1; + } + } else + passphrase = NULL; + + return (passphrase); +} diff --git a/c/archive_read_append_filter.c b/c/archive_read_append_filter.c new file mode 100644 --- /dev/null +++ b/c/archive_read_append_filter.c @@ -0,0 +1,204 @@ +/*- + * Copyright (c) 2003-2012 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_read_private.h" + +int +archive_read_append_filter(struct archive *_a, int code) +{ + int r1, r2, number_bidders, i; + char str[20]; + struct archive_read_filter_bidder *bidder; + struct archive_read_filter *filter; + struct archive_read *a = (struct archive_read *)_a; + + r2 = (ARCHIVE_OK); + switch (code) + { + case ARCHIVE_FILTER_NONE: + /* No filter to add, so do nothing. + * NOTE: An initial "NONE" type filter is always set at the end of the + * filter chain. + */ + r1 = (ARCHIVE_OK); + break; + case ARCHIVE_FILTER_GZIP: + strcpy(str, "gzip"); + r1 = archive_read_support_filter_gzip(_a); + break; + case ARCHIVE_FILTER_BZIP2: + strcpy(str, "bzip2"); + r1 = archive_read_support_filter_bzip2(_a); + break; + case ARCHIVE_FILTER_COMPRESS: + strcpy(str, "compress (.Z)"); + r1 = archive_read_support_filter_compress(_a); + break; + case ARCHIVE_FILTER_PROGRAM: + archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, + "Cannot append program filter using archive_read_append_filter"); + return (ARCHIVE_FATAL); + case ARCHIVE_FILTER_LZMA: + strcpy(str, "lzma"); + r1 = archive_read_support_filter_lzma(_a); + break; + case ARCHIVE_FILTER_XZ: + strcpy(str, "xz"); + r1 = archive_read_support_filter_xz(_a); + break; + case ARCHIVE_FILTER_UU: + strcpy(str, "uu"); + r1 = archive_read_support_filter_uu(_a); + break; + case ARCHIVE_FILTER_RPM: + strcpy(str, "rpm"); + r1 = archive_read_support_filter_rpm(_a); + break; + case ARCHIVE_FILTER_LZ4: + strcpy(str, "lz4"); + r1 = archive_read_support_filter_lz4(_a); + break; + case ARCHIVE_FILTER_ZSTD: + strcpy(str, "zstd"); + r1 = archive_read_support_filter_zstd(_a); + break; + case ARCHIVE_FILTER_LZIP: + strcpy(str, "lzip"); + r1 = archive_read_support_filter_lzip(_a); + break; + case ARCHIVE_FILTER_LRZIP: + strcpy(str, "lrzip"); + r1 = archive_read_support_filter_lrzip(_a); + break; + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, + "Invalid filter code specified"); + return (ARCHIVE_FATAL); + } + + if (code != ARCHIVE_FILTER_NONE) + { + number_bidders = sizeof(a->bidders) / sizeof(a->bidders[0]); + + bidder = a->bidders; + for (i = 0; i < number_bidders; i++, bidder++) + { + if (!bidder->name || !strcmp(bidder->name, str)) + break; + } + if (!bidder->name || strcmp(bidder->name, str)) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, + "Internal error: Unable to append filter"); + return (ARCHIVE_FATAL); + } + + filter + = (struct archive_read_filter *)calloc(1, sizeof(*filter)); + if (filter == NULL) + { + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + filter->bidder = bidder; + filter->archive = a; + filter->upstream = a->filter; + a->filter = filter; + r2 = (bidder->init)(a->filter); + if (r2 != ARCHIVE_OK) { + __archive_read_free_filters(a); + return (ARCHIVE_FATAL); + } + } + + a->bypass_filter_bidding = 1; + return (r1 < r2) ? r1 : r2; +} + +int +archive_read_append_filter_program(struct archive *_a, const char *cmd) +{ + return (archive_read_append_filter_program_signature(_a, cmd, NULL, 0)); +} + +int +archive_read_append_filter_program_signature(struct archive *_a, + const char *cmd, const void *signature, size_t signature_len) +{ + int r, number_bidders, i; + struct archive_read_filter_bidder *bidder; + struct archive_read_filter *filter; + struct archive_read *a = (struct archive_read *)_a; + + if (archive_read_support_filter_program_signature(_a, cmd, signature, + signature_len) != (ARCHIVE_OK)) + return (ARCHIVE_FATAL); + + number_bidders = sizeof(a->bidders) / sizeof(a->bidders[0]); + + bidder = a->bidders; + for (i = 0; i < number_bidders; i++, bidder++) + { + /* Program bidder name set to filter name after initialization */ + if (bidder->data && !bidder->name) + break; + } + if (!bidder->data) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, + "Internal error: Unable to append program filter"); + return (ARCHIVE_FATAL); + } + + filter + = (struct archive_read_filter *)calloc(1, sizeof(*filter)); + if (filter == NULL) + { + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + filter->bidder = bidder; + filter->archive = a; + filter->upstream = a->filter; + a->filter = filter; + r = (bidder->init)(a->filter); + if (r != ARCHIVE_OK) { + __archive_read_free_filters(a); + return (ARCHIVE_FATAL); + } + bidder->name = a->filter->name; + + a->bypass_filter_bidding = 1; + return r; +} diff --git a/c/archive_read_data_into_fd.c b/c/archive_read_data_into_fd.c new file mode 100644 --- /dev/null +++ b/c/archive_read_data_into_fd.c @@ -0,0 +1,139 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: src/lib/libarchive/archive_read_data_into_fd.c,v 1.16 2008/05/23 05:01:29 cperciva Exp $"); + +#ifdef HAVE_SYS_TYPES_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" + +/* Maximum amount of data to write at one time. */ +#define MAX_WRITE (1024 * 1024) + +/* + * This implementation minimizes copying of data and is sparse-file aware. + */ +static int +pad_to(struct archive *a, int fd, int can_lseek, + size_t nulls_size, const char *nulls, + int64_t target_offset, int64_t actual_offset) +{ + size_t to_write; + ssize_t bytes_written; + + if (can_lseek) { + actual_offset = lseek(fd, + target_offset - actual_offset, SEEK_CUR); + if (actual_offset != target_offset) { + archive_set_error(a, errno, "Seek error"); + return (ARCHIVE_FATAL); + } + return (ARCHIVE_OK); + } + while (target_offset > actual_offset) { + to_write = nulls_size; + if (target_offset < actual_offset + (int64_t)nulls_size) + to_write = (size_t)(target_offset - actual_offset); + bytes_written = write(fd, nulls, to_write); + if (bytes_written < 0) { + archive_set_error(a, errno, "Write error"); + return (ARCHIVE_FATAL); + } + actual_offset += bytes_written; + } + return (ARCHIVE_OK); +} + + +int +archive_read_data_into_fd(struct archive *a, int fd) +{ + struct stat st; + int r, r2; + const void *buff; + size_t size, bytes_to_write; + ssize_t bytes_written; + int64_t target_offset; + int64_t actual_offset = 0; + int can_lseek; + char *nulls = NULL; + size_t nulls_size = 16384; + + archive_check_magic(a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_DATA, + "archive_read_data_into_fd"); + + can_lseek = (fstat(fd, &st) == 0) && S_ISREG(st.st_mode); + if (!can_lseek) + nulls = calloc(1, nulls_size); + + while ((r = archive_read_data_block(a, &buff, &size, &target_offset)) == + ARCHIVE_OK) { + const char *p = buff; + if (target_offset > actual_offset) { + r = pad_to(a, fd, can_lseek, nulls_size, nulls, + target_offset, actual_offset); + if (r != ARCHIVE_OK) + break; + actual_offset = target_offset; + } + while (size > 0) { + bytes_to_write = size; + if (bytes_to_write > MAX_WRITE) + bytes_to_write = MAX_WRITE; + bytes_written = write(fd, p, bytes_to_write); + if (bytes_written < 0) { + archive_set_error(a, errno, "Write error"); + r = ARCHIVE_FATAL; + goto cleanup; + } + actual_offset += bytes_written; + p += bytes_written; + size -= bytes_written; + } + } + + if (r == ARCHIVE_EOF && target_offset > actual_offset) { + r2 = pad_to(a, fd, can_lseek, nulls_size, nulls, + target_offset, actual_offset); + if (r2 != ARCHIVE_OK) + r = r2; + } + +cleanup: + free(nulls); + if (r != ARCHIVE_EOF) + return (r); + return (ARCHIVE_OK); +} diff --git a/c/archive_read_disk_entry_from_file.c b/c/archive_read_disk_entry_from_file.c new file mode 100644 --- /dev/null +++ b/c/archive_read_disk_entry_from_file.c @@ -0,0 +1,1044 @@ +/*- + * Copyright (c) 2003-2009 Tim Kientzle + * Copyright (c) 2010-2012 Michihiro NAKAJIMA + * Copyright (c) 2016 Martin Matuska + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD"); + +/* This is the tree-walking code for POSIX systems. */ +#if !defined(_WIN32) || defined(__CYGWIN__) + +#ifdef HAVE_SYS_TYPES_H +#include +#endif +#ifdef HAVE_SYS_EXTATTR_H +#include +#endif +#ifdef HAVE_SYS_IOCTL_H +#include +#endif +#ifdef HAVE_SYS_PARAM_H +#include +#endif +#ifdef HAVE_SYS_STAT_H +#include +#endif +#if defined(HAVE_SYS_XATTR_H) +#include +#elif defined(HAVE_ATTR_XATTR_H) +#include +#endif +#ifdef HAVE_SYS_EA_H +#include +#endif +#ifdef HAVE_COPYFILE_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_FCNTL_H +#include +#endif +#ifdef HAVE_LIMITS_H +#include +#endif +#ifdef HAVE_LINUX_TYPES_H +#include +#endif +#ifdef HAVE_LINUX_FIEMAP_H +#include +#endif +#ifdef HAVE_LINUX_FS_H +#include +#endif +/* + * Some Linux distributions have both linux/ext2_fs.h and ext2fs/ext2_fs.h. + * As the include guards don't agree, the order of include is important. + */ +#ifdef HAVE_LINUX_EXT2_FS_H +#include /* for Linux file flags */ +#endif +#if defined(HAVE_EXT2FS_EXT2_FS_H) && !defined(__CYGWIN__) +#include /* Linux file flags, broken on Cygwin */ +#endif +#ifdef HAVE_PATHS_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_read_disk_private.h" + +#ifndef O_CLOEXEC +#define O_CLOEXEC 0 +#endif + +static int setup_mac_metadata(struct archive_read_disk *, + struct archive_entry *, int *fd); +static int setup_xattrs(struct archive_read_disk *, + struct archive_entry *, int *fd); +static int setup_sparse(struct archive_read_disk *, + struct archive_entry *, int *fd); +#if defined(HAVE_LINUX_FIEMAP_H) +static int setup_sparse_fiemap(struct archive_read_disk *, + struct archive_entry *, int *fd); +#endif + +#if !ARCHIVE_ACL_SUPPORT +int +archive_read_disk_entry_setup_acls(struct archive_read_disk *a, + struct archive_entry *entry, int *fd) +{ + (void)a; /* UNUSED */ + (void)entry; /* UNUSED */ + (void)fd; /* UNUSED */ + return (ARCHIVE_OK); +} +#endif + +/* + * Enter working directory and return working pathname of archive_entry. + * If a pointer to an integer is provided and its value is below zero + * open a file descriptor on this pathname. + */ +const char * +archive_read_disk_entry_setup_path(struct archive_read_disk *a, + struct archive_entry *entry, int *fd) +{ + const char *path; + + path = archive_entry_sourcepath(entry); + + if (path == NULL || (a->tree != NULL && + a->tree_enter_working_dir(a->tree) != 0)) + path = archive_entry_pathname(entry); + if (path == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Couldn't determine path"); + } else if (fd != NULL && *fd < 0 && a->tree != NULL && + (a->follow_symlinks || archive_entry_filetype(entry) != AE_IFLNK)) { + *fd = a->open_on_current_dir(a->tree, path, + O_RDONLY | O_NONBLOCK); + } + return (path); +} + +int +archive_read_disk_entry_from_file(struct archive *_a, + struct archive_entry *entry, + int fd, + const struct stat *st) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + const char *path, *name; + struct stat s; + int initial_fd = fd; + int r, r1; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, + "archive_read_disk_entry_from_file"); + + archive_clear_error(_a); + path = archive_entry_sourcepath(entry); + if (path == NULL) + path = archive_entry_pathname(entry); + + if (a->tree == NULL) { + if (st == NULL) { +#if HAVE_FSTAT + if (fd >= 0) { + if (fstat(fd, &s) != 0) { + archive_set_error(&a->archive, errno, + "Can't fstat"); + return (ARCHIVE_FAILED); + } + } else +#endif +#if HAVE_LSTAT + if (!a->follow_symlinks) { + if (lstat(path, &s) != 0) { + archive_set_error(&a->archive, errno, + "Can't lstat %s", path); + return (ARCHIVE_FAILED); + } + } else +#endif + if (la_stat(path, &s) != 0) { + archive_set_error(&a->archive, errno, + "Can't stat %s", path); + return (ARCHIVE_FAILED); + } + st = &s; + } + archive_entry_copy_stat(entry, st); + } + + /* Lookup uname/gname */ + name = archive_read_disk_uname(_a, archive_entry_uid(entry)); + if (name != NULL) + archive_entry_copy_uname(entry, name); + name = archive_read_disk_gname(_a, archive_entry_gid(entry)); + if (name != NULL) + archive_entry_copy_gname(entry, name); + +#ifdef HAVE_STRUCT_STAT_ST_FLAGS + /* On FreeBSD, we get flags for free with the stat. */ + /* TODO: Does this belong in copy_stat()? */ + if ((a->flags & ARCHIVE_READDISK_NO_FFLAGS) == 0 && st->st_flags != 0) + archive_entry_set_fflags(entry, st->st_flags, 0); +#endif + +#if (defined(FS_IOC_GETFLAGS) && defined(HAVE_WORKING_FS_IOC_GETFLAGS)) || \ + (defined(EXT2_IOC_GETFLAGS) && defined(HAVE_WORKING_EXT2_IOC_GETFLAGS)) + /* Linux requires an extra ioctl to pull the flags. Although + * this is an extra step, it has a nice side-effect: We get an + * open file descriptor which we can use in the subsequent lookups. */ + if ((a->flags & ARCHIVE_READDISK_NO_FFLAGS) == 0 && + (S_ISREG(st->st_mode) || S_ISDIR(st->st_mode))) { + if (fd < 0) { + if (a->tree != NULL) + fd = a->open_on_current_dir(a->tree, path, + O_RDONLY | O_NONBLOCK | O_CLOEXEC); + else + fd = open(path, O_RDONLY | O_NONBLOCK | + O_CLOEXEC); + __archive_ensure_cloexec_flag(fd); + } + if (fd >= 0) { + int stflags; + r = ioctl(fd, +#if defined(FS_IOC_GETFLAGS) + FS_IOC_GETFLAGS, +#else + EXT2_IOC_GETFLAGS, +#endif + &stflags); + if (r == 0 && stflags != 0) + archive_entry_set_fflags(entry, stflags, 0); + } + } +#endif + +#if defined(HAVE_READLINK) || defined(HAVE_READLINKAT) + if (S_ISLNK(st->st_mode)) { + size_t linkbuffer_len = st->st_size; + char *linkbuffer; + int lnklen; + + linkbuffer = malloc(linkbuffer_len + 1); + if (linkbuffer == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Couldn't read link data"); + return (ARCHIVE_FAILED); + } + if (a->tree != NULL) { +#ifdef HAVE_READLINKAT + lnklen = readlinkat(a->tree_current_dir_fd(a->tree), + path, linkbuffer, linkbuffer_len); +#else + if (a->tree_enter_working_dir(a->tree) != 0) { + archive_set_error(&a->archive, errno, + "Couldn't read link data"); + free(linkbuffer); + return (ARCHIVE_FAILED); + } + lnklen = readlink(path, linkbuffer, linkbuffer_len); +#endif /* HAVE_READLINKAT */ + } else + lnklen = readlink(path, linkbuffer, linkbuffer_len); + if (lnklen < 0) { + archive_set_error(&a->archive, errno, + "Couldn't read link data"); + free(linkbuffer); + return (ARCHIVE_FAILED); + } + linkbuffer[lnklen] = '\0'; + archive_entry_set_symlink(entry, linkbuffer); + free(linkbuffer); + } +#endif /* HAVE_READLINK || HAVE_READLINKAT */ + + r = 0; + if ((a->flags & ARCHIVE_READDISK_NO_ACL) == 0) + r = archive_read_disk_entry_setup_acls(a, entry, &fd); + if ((a->flags & ARCHIVE_READDISK_NO_XATTR) == 0) { + r1 = setup_xattrs(a, entry, &fd); + if (r1 < r) + r = r1; + } + if (a->flags & ARCHIVE_READDISK_MAC_COPYFILE) { + r1 = setup_mac_metadata(a, entry, &fd); + if (r1 < r) + r = r1; + } + r1 = setup_sparse(a, entry, &fd); + if (r1 < r) + r = r1; + + /* If we opened the file earlier in this function, close it. */ + if (initial_fd != fd) + close(fd); + return (r); +} + +#if defined(__APPLE__) && defined(HAVE_COPYFILE_H) +/* + * The Mac OS "copyfile()" API copies the extended metadata for a + * file into a separate file in AppleDouble format (see RFC 1740). + * + * Mac OS tar and cpio implementations store this extended + * metadata as a separate entry just before the regular entry + * with a "._" prefix added to the filename. + * + * Note that this is currently done unconditionally; the tar program has + * an option to discard this information before the archive is written. + * + * TODO: If there's a failure, report it and return ARCHIVE_WARN. + */ +static int +setup_mac_metadata(struct archive_read_disk *a, + struct archive_entry *entry, int *fd) +{ + int tempfd = -1; + int copyfile_flags = COPYFILE_NOFOLLOW | COPYFILE_ACL | COPYFILE_XATTR; + struct stat copyfile_stat; + int ret = ARCHIVE_OK; + void *buff = NULL; + int have_attrs; + const char *name, *tempdir; + struct archive_string tempfile; + + (void)fd; /* UNUSED */ + + name = archive_read_disk_entry_setup_path(a, entry, NULL); + if (name == NULL) + return (ARCHIVE_WARN); + + /* Short-circuit if there's nothing to do. */ + have_attrs = copyfile(name, NULL, 0, copyfile_flags | COPYFILE_CHECK); + if (have_attrs == -1) { + archive_set_error(&a->archive, errno, + "Could not check extended attributes"); + return (ARCHIVE_WARN); + } + if (have_attrs == 0) + return (ARCHIVE_OK); + + tempdir = NULL; + if (issetugid() == 0) + tempdir = getenv("TMPDIR"); + if (tempdir == NULL) + tempdir = _PATH_TMP; + archive_string_init(&tempfile); + archive_strcpy(&tempfile, tempdir); + archive_strcat(&tempfile, "tar.md.XXXXXX"); + tempfd = mkstemp(tempfile.s); + if (tempfd < 0) { + archive_set_error(&a->archive, errno, + "Could not open extended attribute file"); + ret = ARCHIVE_WARN; + goto cleanup; + } + __archive_ensure_cloexec_flag(tempfd); + + /* XXX I wish copyfile() could pack directly to a memory + * buffer; that would avoid the temp file here. For that + * matter, it would be nice if fcopyfile() actually worked, + * that would reduce the many open/close races here. */ + if (copyfile(name, tempfile.s, 0, copyfile_flags | COPYFILE_PACK)) { + archive_set_error(&a->archive, errno, + "Could not pack extended attributes"); + ret = ARCHIVE_WARN; + goto cleanup; + } + if (fstat(tempfd, ©file_stat)) { + archive_set_error(&a->archive, errno, + "Could not check size of extended attributes"); + ret = ARCHIVE_WARN; + goto cleanup; + } + buff = malloc(copyfile_stat.st_size); + if (buff == NULL) { + archive_set_error(&a->archive, errno, + "Could not allocate memory for extended attributes"); + ret = ARCHIVE_WARN; + goto cleanup; + } + if (copyfile_stat.st_size != read(tempfd, buff, copyfile_stat.st_size)) { + archive_set_error(&a->archive, errno, + "Could not read extended attributes into memory"); + ret = ARCHIVE_WARN; + goto cleanup; + } + archive_entry_copy_mac_metadata(entry, buff, copyfile_stat.st_size); + +cleanup: + if (tempfd >= 0) { + close(tempfd); + unlink(tempfile.s); + } + archive_string_free(&tempfile); + free(buff); + return (ret); +} + +#else + +/* + * Stub implementation for non-Mac systems. + */ +static int +setup_mac_metadata(struct archive_read_disk *a, + struct archive_entry *entry, int *fd) +{ + (void)a; /* UNUSED */ + (void)entry; /* UNUSED */ + (void)fd; /* UNUSED */ + return (ARCHIVE_OK); +} +#endif + +#if ARCHIVE_XATTR_LINUX || ARCHIVE_XATTR_DARWIN || ARCHIVE_XATTR_AIX + +/* + * Linux, Darwin and AIX extended attribute support. + * + * TODO: By using a stack-allocated buffer for the first + * call to getxattr(), we might be able to avoid the second + * call entirely. We only need the second call if the + * stack-allocated buffer is too small. But a modest buffer + * of 1024 bytes or so will often be big enough. Same applies + * to listxattr(). + */ + + +static int +setup_xattr(struct archive_read_disk *a, + struct archive_entry *entry, const char *name, int fd, const char *accpath) +{ + ssize_t size; + void *value = NULL; + + + if (fd >= 0) { +#if ARCHIVE_XATTR_LINUX + size = fgetxattr(fd, name, NULL, 0); +#elif ARCHIVE_XATTR_DARWIN + size = fgetxattr(fd, name, NULL, 0, 0, 0); +#elif ARCHIVE_XATTR_AIX + size = fgetea(fd, name, NULL, 0); +#endif + } else if (!a->follow_symlinks) { +#if ARCHIVE_XATTR_LINUX + size = lgetxattr(accpath, name, NULL, 0); +#elif ARCHIVE_XATTR_DARWIN + size = getxattr(accpath, name, NULL, 0, 0, XATTR_NOFOLLOW); +#elif ARCHIVE_XATTR_AIX + size = lgetea(accpath, name, NULL, 0); +#endif + } else { +#if ARCHIVE_XATTR_LINUX + size = getxattr(accpath, name, NULL, 0); +#elif ARCHIVE_XATTR_DARWIN + size = getxattr(accpath, name, NULL, 0, 0, 0); +#elif ARCHIVE_XATTR_AIX + size = getea(accpath, name, NULL, 0); +#endif + } + + if (size == -1) { + archive_set_error(&a->archive, errno, + "Couldn't query extended attribute"); + return (ARCHIVE_WARN); + } + + if (size > 0 && (value = malloc(size)) == NULL) { + archive_set_error(&a->archive, errno, "Out of memory"); + return (ARCHIVE_FATAL); + } + + + if (fd >= 0) { +#if ARCHIVE_XATTR_LINUX + size = fgetxattr(fd, name, value, size); +#elif ARCHIVE_XATTR_DARWIN + size = fgetxattr(fd, name, value, size, 0, 0); +#elif ARCHIVE_XATTR_AIX + size = fgetea(fd, name, value, size); +#endif + } else if (!a->follow_symlinks) { +#if ARCHIVE_XATTR_LINUX + size = lgetxattr(accpath, name, value, size); +#elif ARCHIVE_XATTR_DARWIN + size = getxattr(accpath, name, value, size, 0, XATTR_NOFOLLOW); +#elif ARCHIVE_XATTR_AIX + size = lgetea(accpath, name, value, size); +#endif + } else { +#if ARCHIVE_XATTR_LINUX + size = getxattr(accpath, name, value, size); +#elif ARCHIVE_XATTR_DARWIN + size = getxattr(accpath, name, value, size, 0, 0); +#elif ARCHIVE_XATTR_AIX + size = getea(accpath, name, value, size); +#endif + } + + if (size == -1) { + archive_set_error(&a->archive, errno, + "Couldn't read extended attribute"); + return (ARCHIVE_WARN); + } + + archive_entry_xattr_add_entry(entry, name, value, size); + + free(value); + return (ARCHIVE_OK); +} + +static int +setup_xattrs(struct archive_read_disk *a, + struct archive_entry *entry, int *fd) +{ + char *list, *p; + const char *path; + ssize_t list_size; + + path = NULL; + + if (*fd < 0) { + path = archive_read_disk_entry_setup_path(a, entry, fd); + if (path == NULL) + return (ARCHIVE_WARN); + } + + if (*fd >= 0) { +#if ARCHIVE_XATTR_LINUX + list_size = flistxattr(*fd, NULL, 0); +#elif ARCHIVE_XATTR_DARWIN + list_size = flistxattr(*fd, NULL, 0, 0); +#elif ARCHIVE_XATTR_AIX + list_size = flistea(*fd, NULL, 0); +#endif + } else if (!a->follow_symlinks) { +#if ARCHIVE_XATTR_LINUX + list_size = llistxattr(path, NULL, 0); +#elif ARCHIVE_XATTR_DARWIN + list_size = listxattr(path, NULL, 0, XATTR_NOFOLLOW); +#elif ARCHIVE_XATTR_AIX + list_size = llistea(path, NULL, 0); +#endif + } else { +#if ARCHIVE_XATTR_LINUX + list_size = listxattr(path, NULL, 0); +#elif ARCHIVE_XATTR_DARWIN + list_size = listxattr(path, NULL, 0, 0); +#elif ARCHIVE_XATTR_AIX + list_size = listea(path, NULL, 0); +#endif + } + + if (list_size == -1) { + if (errno == ENOTSUP || errno == ENOSYS) + return (ARCHIVE_OK); + archive_set_error(&a->archive, errno, + "Couldn't list extended attributes"); + return (ARCHIVE_WARN); + } + + if (list_size == 0) + return (ARCHIVE_OK); + + if ((list = malloc(list_size)) == NULL) { + archive_set_error(&a->archive, errno, "Out of memory"); + return (ARCHIVE_FATAL); + } + + if (*fd >= 0) { +#if ARCHIVE_XATTR_LINUX + list_size = flistxattr(*fd, list, list_size); +#elif ARCHIVE_XATTR_DARWIN + list_size = flistxattr(*fd, list, list_size, 0); +#elif ARCHIVE_XATTR_AIX + list_size = flistea(*fd, list, list_size); +#endif + } else if (!a->follow_symlinks) { +#if ARCHIVE_XATTR_LINUX + list_size = llistxattr(path, list, list_size); +#elif ARCHIVE_XATTR_DARWIN + list_size = listxattr(path, list, list_size, XATTR_NOFOLLOW); +#elif ARCHIVE_XATTR_AIX + list_size = llistea(path, list, list_size); +#endif + } else { +#if ARCHIVE_XATTR_LINUX + list_size = listxattr(path, list, list_size); +#elif ARCHIVE_XATTR_DARWIN + list_size = listxattr(path, list, list_size, 0); +#elif ARCHIVE_XATTR_AIX + list_size = listea(path, list, list_size); +#endif + } + + if (list_size == -1) { + archive_set_error(&a->archive, errno, + "Couldn't retrieve extended attributes"); + free(list); + return (ARCHIVE_WARN); + } + + for (p = list; (p - list) < list_size; p += strlen(p) + 1) { +#if ARCHIVE_XATTR_LINUX + /* Linux: skip POSIX.1e ACL extended attributes */ + if (strncmp(p, "system.", 7) == 0 && + (strcmp(p + 7, "posix_acl_access") == 0 || + strcmp(p + 7, "posix_acl_default") == 0)) + continue; + if (strncmp(p, "trusted.SGI_", 12) == 0 && + (strcmp(p + 12, "ACL_DEFAULT") == 0 || + strcmp(p + 12, "ACL_FILE") == 0)) + continue; + + /* Linux: xfsroot namespace is obsolete and unsupported */ + if (strncmp(p, "xfsroot.", 8) == 0) + continue; +#endif + setup_xattr(a, entry, p, *fd, path); + } + + free(list); + return (ARCHIVE_OK); +} + +#elif ARCHIVE_XATTR_FREEBSD + +/* + * FreeBSD extattr interface. + */ + +/* TODO: Implement this. Follow the Linux model above, but + * with FreeBSD-specific system calls, of course. Be careful + * to not include the system extattrs that hold ACLs; we handle + * those separately. + */ +static int +setup_xattr(struct archive_read_disk *a, struct archive_entry *entry, + int namespace, const char *name, const char *fullname, int fd, + const char *path); + +static int +setup_xattr(struct archive_read_disk *a, struct archive_entry *entry, + int namespace, const char *name, const char *fullname, int fd, + const char *accpath) +{ + ssize_t size; + void *value = NULL; + + if (fd >= 0) + size = extattr_get_fd(fd, namespace, name, NULL, 0); + else if (!a->follow_symlinks) + size = extattr_get_link(accpath, namespace, name, NULL, 0); + else + size = extattr_get_file(accpath, namespace, name, NULL, 0); + + if (size == -1) { + archive_set_error(&a->archive, errno, + "Couldn't query extended attribute"); + return (ARCHIVE_WARN); + } + + if (size > 0 && (value = malloc(size)) == NULL) { + archive_set_error(&a->archive, errno, "Out of memory"); + return (ARCHIVE_FATAL); + } + + if (fd >= 0) + size = extattr_get_fd(fd, namespace, name, value, size); + else if (!a->follow_symlinks) + size = extattr_get_link(accpath, namespace, name, value, size); + else + size = extattr_get_file(accpath, namespace, name, value, size); + + if (size == -1) { + free(value); + archive_set_error(&a->archive, errno, + "Couldn't read extended attribute"); + return (ARCHIVE_WARN); + } + + archive_entry_xattr_add_entry(entry, fullname, value, size); + + free(value); + return (ARCHIVE_OK); +} + +static int +setup_xattrs(struct archive_read_disk *a, + struct archive_entry *entry, int *fd) +{ + char buff[512]; + char *list, *p; + ssize_t list_size; + const char *path; + int namespace = EXTATTR_NAMESPACE_USER; + + path = NULL; + + if (*fd < 0) { + path = archive_read_disk_entry_setup_path(a, entry, fd); + if (path == NULL) + return (ARCHIVE_WARN); + } + + if (*fd >= 0) + list_size = extattr_list_fd(*fd, namespace, NULL, 0); + else if (!a->follow_symlinks) + list_size = extattr_list_link(path, namespace, NULL, 0); + else + list_size = extattr_list_file(path, namespace, NULL, 0); + + if (list_size == -1 && errno == EOPNOTSUPP) + return (ARCHIVE_OK); + if (list_size == -1) { + archive_set_error(&a->archive, errno, + "Couldn't list extended attributes"); + return (ARCHIVE_WARN); + } + + if (list_size == 0) + return (ARCHIVE_OK); + + if ((list = malloc(list_size)) == NULL) { + archive_set_error(&a->archive, errno, "Out of memory"); + return (ARCHIVE_FATAL); + } + + if (*fd >= 0) + list_size = extattr_list_fd(*fd, namespace, list, list_size); + else if (!a->follow_symlinks) + list_size = extattr_list_link(path, namespace, list, list_size); + else + list_size = extattr_list_file(path, namespace, list, list_size); + + if (list_size == -1) { + archive_set_error(&a->archive, errno, + "Couldn't retrieve extended attributes"); + free(list); + return (ARCHIVE_WARN); + } + + p = list; + while ((p - list) < list_size) { + size_t len = 255 & (int)*p; + char *name; + + strcpy(buff, "user."); + name = buff + strlen(buff); + memcpy(name, p + 1, len); + name[len] = '\0'; + setup_xattr(a, entry, namespace, name, buff, *fd, path); + p += 1 + len; + } + + free(list); + return (ARCHIVE_OK); +} + +#else + +/* + * Generic (stub) extended attribute support. + */ +static int +setup_xattrs(struct archive_read_disk *a, + struct archive_entry *entry, int *fd) +{ + (void)a; /* UNUSED */ + (void)entry; /* UNUSED */ + (void)fd; /* UNUSED */ + return (ARCHIVE_OK); +} + +#endif + +#if defined(HAVE_LINUX_FIEMAP_H) + +/* + * Linux FIEMAP sparse interface. + * + * The FIEMAP ioctl returns an "extent" for each physical allocation + * on disk. We need to process those to generate a more compact list + * of logical file blocks. We also need to be very careful to use + * FIEMAP_FLAG_SYNC here, since there are reports that Linux sometimes + * does not report allocations for newly-written data that hasn't + * been synced to disk. + * + * It's important to return a minimal sparse file list because we want + * to not trigger sparse file extensions if we don't have to, since + * not all readers support them. + */ + +static int +setup_sparse_fiemap(struct archive_read_disk *a, + struct archive_entry *entry, int *fd) +{ + char buff[4096]; + struct fiemap *fm; + struct fiemap_extent *fe; + int64_t size; + int count, do_fiemap, iters; + int exit_sts = ARCHIVE_OK; + const char *path; + + if (archive_entry_filetype(entry) != AE_IFREG + || archive_entry_size(entry) <= 0 + || archive_entry_hardlink(entry) != NULL) + return (ARCHIVE_OK); + + if (*fd < 0) { + path = archive_read_disk_entry_setup_path(a, entry, NULL); + if (path == NULL) + return (ARCHIVE_FAILED); + + if (a->tree != NULL) + *fd = a->open_on_current_dir(a->tree, path, + O_RDONLY | O_NONBLOCK | O_CLOEXEC); + else + *fd = open(path, O_RDONLY | O_NONBLOCK | O_CLOEXEC); + if (*fd < 0) { + archive_set_error(&a->archive, errno, + "Can't open `%s'", path); + return (ARCHIVE_FAILED); + } + __archive_ensure_cloexec_flag(*fd); + } + + /* Initialize buffer to avoid the error valgrind complains about. */ + memset(buff, 0, sizeof(buff)); + count = (sizeof(buff) - sizeof(*fm))/sizeof(*fe); + fm = (struct fiemap *)buff; + fm->fm_start = 0; + fm->fm_length = ~0ULL;; + fm->fm_flags = FIEMAP_FLAG_SYNC; + fm->fm_extent_count = count; + do_fiemap = 1; + size = archive_entry_size(entry); + for (iters = 0; ; ++iters) { + int i, r; + + r = ioctl(*fd, FS_IOC_FIEMAP, fm); + if (r < 0) { + /* When something error happens, it is better we + * should return ARCHIVE_OK because an earlier + * version(<2.6.28) cannot perform FS_IOC_FIEMAP. */ + goto exit_setup_sparse_fiemap; + } + if (fm->fm_mapped_extents == 0) { + if (iters == 0) { + /* Fully sparse file; insert a zero-length "data" entry */ + archive_entry_sparse_add_entry(entry, 0, 0); + } + break; + } + fe = fm->fm_extents; + for (i = 0; i < (int)fm->fm_mapped_extents; i++, fe++) { + if (!(fe->fe_flags & FIEMAP_EXTENT_UNWRITTEN)) { + /* The fe_length of the last block does not + * adjust itself to its size files. */ + int64_t length = fe->fe_length; + if (fe->fe_logical + length > (uint64_t)size) + length -= fe->fe_logical + length - size; + if (fe->fe_logical == 0 && length == size) { + /* This is not sparse. */ + do_fiemap = 0; + break; + } + if (length > 0) + archive_entry_sparse_add_entry(entry, + fe->fe_logical, length); + } + if (fe->fe_flags & FIEMAP_EXTENT_LAST) + do_fiemap = 0; + } + if (do_fiemap) { + fe = fm->fm_extents + fm->fm_mapped_extents -1; + fm->fm_start = fe->fe_logical + fe->fe_length; + } else + break; + } +exit_setup_sparse_fiemap: + return (exit_sts); +} + +#if !defined(SEEK_HOLE) || !defined(SEEK_DATA) +static int +setup_sparse(struct archive_read_disk *a, + struct archive_entry *entry, int *fd) +{ + return setup_sparse_fiemap(a, entry, fd); +} +#endif +#endif /* defined(HAVE_LINUX_FIEMAP_H) */ + +#if defined(SEEK_HOLE) && defined(SEEK_DATA) + +/* + * SEEK_HOLE sparse interface (FreeBSD, Linux, Solaris) + */ + +static int +setup_sparse(struct archive_read_disk *a, + struct archive_entry *entry, int *fd) +{ + int64_t size; + off_t initial_off; + off_t off_s, off_e; + int exit_sts = ARCHIVE_OK; + int check_fully_sparse = 0; + const char *path; + + if (archive_entry_filetype(entry) != AE_IFREG + || archive_entry_size(entry) <= 0 + || archive_entry_hardlink(entry) != NULL) + return (ARCHIVE_OK); + + /* Does filesystem support the reporting of hole ? */ + if (*fd < 0) + path = archive_read_disk_entry_setup_path(a, entry, fd); + else + path = NULL; + + if (*fd >= 0) { +#ifdef _PC_MIN_HOLE_SIZE + if (fpathconf(*fd, _PC_MIN_HOLE_SIZE) <= 0) + return (ARCHIVE_OK); +#endif + initial_off = lseek(*fd, 0, SEEK_CUR); + if (initial_off != 0) + lseek(*fd, 0, SEEK_SET); + } else { + if (path == NULL) + return (ARCHIVE_FAILED); +#ifdef _PC_MIN_HOLE_SIZE + if (pathconf(path, _PC_MIN_HOLE_SIZE) <= 0) + return (ARCHIVE_OK); +#endif + *fd = open(path, O_RDONLY | O_NONBLOCK | O_CLOEXEC); + if (*fd < 0) { + archive_set_error(&a->archive, errno, + "Can't open `%s'", path); + return (ARCHIVE_FAILED); + } + __archive_ensure_cloexec_flag(*fd); + initial_off = 0; + } + +#ifndef _PC_MIN_HOLE_SIZE + /* Check if the underlying filesystem supports seek hole */ + off_s = lseek(*fd, 0, SEEK_HOLE); + if (off_s < 0) +#if defined(HAVE_LINUX_FIEMAP_H) + return setup_sparse_fiemap(a, entry, fd); +#else + goto exit_setup_sparse; +#endif + else if (off_s > 0) + lseek(*fd, 0, SEEK_SET); +#endif + + off_s = 0; + size = archive_entry_size(entry); + while (off_s < size) { + off_s = lseek(*fd, off_s, SEEK_DATA); + if (off_s == (off_t)-1) { + if (errno == ENXIO) { + /* no more hole */ + if (archive_entry_sparse_count(entry) == 0) { + /* Potentially a fully-sparse file. */ + check_fully_sparse = 1; + } + break; + } + archive_set_error(&a->archive, errno, + "lseek(SEEK_HOLE) failed"); + exit_sts = ARCHIVE_FAILED; + goto exit_setup_sparse; + } + off_e = lseek(*fd, off_s, SEEK_HOLE); + if (off_e == (off_t)-1) { + if (errno == ENXIO) { + off_e = lseek(*fd, 0, SEEK_END); + if (off_e != (off_t)-1) + break;/* no more data */ + } + archive_set_error(&a->archive, errno, + "lseek(SEEK_DATA) failed"); + exit_sts = ARCHIVE_FAILED; + goto exit_setup_sparse; + } + if (off_s == 0 && off_e == size) + break;/* This is not sparse. */ + archive_entry_sparse_add_entry(entry, off_s, + off_e - off_s); + off_s = off_e; + } + + if (check_fully_sparse) { + if (lseek(*fd, 0, SEEK_HOLE) == 0 && + lseek(*fd, 0, SEEK_END) == size) { + /* Fully sparse file; insert a zero-length "data" entry */ + archive_entry_sparse_add_entry(entry, 0, 0); + } + } +exit_setup_sparse: + lseek(*fd, initial_off, SEEK_SET); + return (exit_sts); +} + +#elif !defined(HAVE_LINUX_FIEMAP_H) + +/* + * Generic (stub) sparse support. + */ +static int +setup_sparse(struct archive_read_disk *a, + struct archive_entry *entry, int *fd) +{ + (void)a; /* UNUSED */ + (void)entry; /* UNUSED */ + (void)fd; /* UNUSED */ + return (ARCHIVE_OK); +} + +#endif + +#endif /* !defined(_WIN32) || defined(__CYGWIN__) */ + diff --git a/c/archive_read_disk_posix.c b/c/archive_read_disk_posix.c new file mode 100644 --- /dev/null +++ b/c/archive_read_disk_posix.c @@ -0,0 +1,2722 @@ +/*- + * Copyright (c) 2003-2009 Tim Kientzle + * Copyright (c) 2010-2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer + * in this position and unchanged. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +/* This is the tree-walking code for POSIX systems. */ +#if !defined(_WIN32) || defined(__CYGWIN__) + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_SYS_PARAM_H +#include +#endif +#ifdef HAVE_SYS_MOUNT_H +#include +#endif +#ifdef HAVE_SYS_STAT_H +#include +#endif +#ifdef HAVE_SYS_STATFS_H +#include +#endif +#ifdef HAVE_SYS_STATVFS_H +#include +#endif +#ifdef HAVE_SYS_TIME_H +#include +#endif +#ifdef HAVE_LINUX_MAGIC_H +#include +#endif +#ifdef HAVE_LINUX_FS_H +#include +#endif +/* + * Some Linux distributions have both linux/ext2_fs.h and ext2fs/ext2_fs.h. + * As the include guards don't agree, the order of include is important. + */ +#ifdef HAVE_LINUX_EXT2_FS_H +#include /* for Linux file flags */ +#endif +#if defined(HAVE_EXT2FS_EXT2_FS_H) && !defined(__CYGWIN__) +#include /* Linux file flags, broken on Cygwin */ +#endif +#ifdef HAVE_DIRECT_H +#include +#endif +#ifdef HAVE_DIRENT_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_FCNTL_H +#include +#endif +#ifdef HAVE_LIMITS_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif +#ifdef HAVE_SYS_IOCTL_H +#include +#endif + +#include "archive.h" +#include "archive_string.h" +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_read_disk_private.h" + +#ifndef HAVE_FCHDIR +#error fchdir function required. +#endif +#ifndef O_BINARY +#define O_BINARY 0 +#endif +#ifndef O_CLOEXEC +#define O_CLOEXEC 0 +#endif + +/*- + * This is a new directory-walking system that addresses a number + * of problems I've had with fts(3). In particular, it has no + * pathname-length limits (other than the size of 'int'), handles + * deep logical traversals, uses considerably less memory, and has + * an opaque interface (easier to modify in the future). + * + * Internally, it keeps a single list of "tree_entry" items that + * represent filesystem objects that require further attention. + * Non-directories are not kept in memory: they are pulled from + * readdir(), returned to the client, then freed as soon as possible. + * Any directory entry to be traversed gets pushed onto the stack. + * + * There is surprisingly little information that needs to be kept for + * each item on the stack. Just the name, depth (represented here as the + * string length of the parent directory's pathname), and some markers + * indicating how to get back to the parent (via chdir("..") for a + * regular dir or via fchdir(2) for a symlink). + */ +/* + * TODO: + * 1) Loop checking. + * 3) Arbitrary logical traversals by closing/reopening intermediate fds. + */ + +struct restore_time { + const char *name; + time_t mtime; + long mtime_nsec; + time_t atime; + long atime_nsec; + mode_t filetype; + int noatime; +}; + +struct tree_entry { + int depth; + struct tree_entry *next; + struct tree_entry *parent; + struct archive_string name; + size_t dirname_length; + int64_t dev; + int64_t ino; + int flags; + int filesystem_id; + /* How to return back to the parent of a symlink. */ + int symlink_parent_fd; + /* How to restore time of a directory. */ + struct restore_time restore_time; +}; + +struct filesystem { + int64_t dev; + int synthetic; + int remote; + int noatime; +#if defined(USE_READDIR_R) + size_t name_max; +#endif + long incr_xfer_size; + long max_xfer_size; + long min_xfer_size; + long xfer_align; + + /* + * Buffer used for reading file contents. + */ + /* Exactly allocated memory pointer. */ + unsigned char *allocation_ptr; + /* Pointer adjusted to the filesystem alignment . */ + unsigned char *buff; + size_t buff_size; +}; + +/* Definitions for tree_entry.flags bitmap. */ +#define isDir 1 /* This entry is a regular directory. */ +#define isDirLink 2 /* This entry is a symbolic link to a directory. */ +#define needsFirstVisit 4 /* This is an initial entry. */ +#define needsDescent 8 /* This entry needs to be previsited. */ +#define needsOpen 16 /* This is a directory that needs to be opened. */ +#define needsAscent 32 /* This entry needs to be postvisited. */ + +/* + * Local data for this package. + */ +struct tree { + struct tree_entry *stack; + struct tree_entry *current; + DIR *d; +#define INVALID_DIR_HANDLE NULL + struct dirent *de; +#if defined(USE_READDIR_R) + struct dirent *dirent; + size_t dirent_allocated; +#endif + int flags; + int visit_type; + /* Error code from last failed operation. */ + int tree_errno; + + /* Dynamically-sized buffer for holding path */ + struct archive_string path; + + /* Last path element */ + const char *basename; + /* Leading dir length */ + size_t dirname_length; + + int depth; + int openCount; + int maxOpenCount; + int initial_dir_fd; + int working_dir_fd; + + struct stat lst; + struct stat st; + int descend; + int nlink; + /* How to restore time of a file. */ + struct restore_time restore_time; + + struct entry_sparse { + int64_t length; + int64_t offset; + } *sparse_list, *current_sparse; + int sparse_count; + int sparse_list_size; + + char initial_symlink_mode; + char symlink_mode; + struct filesystem *current_filesystem; + struct filesystem *filesystem_table; + int initial_filesystem_id; + int current_filesystem_id; + int max_filesystem_id; + int allocated_filesystem; + + int entry_fd; + int entry_eof; + int64_t entry_remaining_bytes; + int64_t entry_total; + unsigned char *entry_buff; + size_t entry_buff_size; +}; + +/* Definitions for tree.flags bitmap. */ +#define hasStat 16 /* The st entry is valid. */ +#define hasLstat 32 /* The lst entry is valid. */ +#define onWorkingDir 64 /* We are on the working dir where we are + * reading directory entry at this time. */ +#define needsRestoreTimes 128 +#define onInitialDir 256 /* We are on the initial dir. */ + +static int +tree_dir_next_posix(struct tree *t); + +#ifdef HAVE_DIRENT_D_NAMLEN +/* BSD extension; avoids need for a strlen() call. */ +#define D_NAMELEN(dp) (dp)->d_namlen +#else +#define D_NAMELEN(dp) (strlen((dp)->d_name)) +#endif + +/* Initiate/terminate a tree traversal. */ +static struct tree *tree_open(const char *, int, int); +static struct tree *tree_reopen(struct tree *, const char *, int); +static void tree_close(struct tree *); +static void tree_free(struct tree *); +static void tree_push(struct tree *, const char *, int, int64_t, int64_t, + struct restore_time *); +static int tree_enter_initial_dir(struct tree *); +static int tree_enter_working_dir(struct tree *); +static int tree_current_dir_fd(struct tree *); + +/* + * tree_next() returns Zero if there is no next entry, non-zero if + * there is. Note that directories are visited three times. + * Directories are always visited first as part of enumerating their + * parent; that is a "regular" visit. If tree_descend() is invoked at + * that time, the directory is added to a work list and will + * subsequently be visited two more times: once just after descending + * into the directory ("postdescent") and again just after ascending + * back to the parent ("postascent"). + * + * TREE_ERROR_DIR is returned if the descent failed (because the + * directory couldn't be opened, for instance). This is returned + * instead of TREE_POSTDESCENT/TREE_POSTASCENT. TREE_ERROR_DIR is not a + * fatal error, but it does imply that the relevant subtree won't be + * visited. TREE_ERROR_FATAL is returned for an error that left the + * traversal completely hosed. Right now, this is only returned for + * chdir() failures during ascent. + */ +#define TREE_REGULAR 1 +#define TREE_POSTDESCENT 2 +#define TREE_POSTASCENT 3 +#define TREE_ERROR_DIR -1 +#define TREE_ERROR_FATAL -2 + +static int tree_next(struct tree *); + +/* + * Return information about the current entry. + */ + +/* + * The current full pathname, length of the full pathname, and a name + * that can be used to access the file. Because tree does use chdir + * extensively, the access path is almost never the same as the full + * current path. + * + * TODO: On platforms that support it, use openat()-style operations + * to eliminate the chdir() operations entirely while still supporting + * arbitrarily deep traversals. This makes access_path troublesome to + * support, of course, which means we'll need a rich enough interface + * that clients can function without it. (In particular, we'll need + * tree_current_open() that returns an open file descriptor.) + * + */ +static const char *tree_current_path(struct tree *); +static const char *tree_current_access_path(struct tree *); + +/* + * Request the lstat() or stat() data for the current path. Since the + * tree package needs to do some of this anyway, and caches the + * results, you should take advantage of it here if you need it rather + * than make a redundant stat() or lstat() call of your own. + */ +static const struct stat *tree_current_stat(struct tree *); +static const struct stat *tree_current_lstat(struct tree *); +static int tree_current_is_symblic_link_target(struct tree *); + +/* The following functions use tricks to avoid a certain number of + * stat()/lstat() calls. */ +/* "is_physical_dir" is equivalent to S_ISDIR(tree_current_lstat()->st_mode) */ +static int tree_current_is_physical_dir(struct tree *); +/* "is_dir" is equivalent to S_ISDIR(tree_current_stat()->st_mode) */ +static int tree_current_is_dir(struct tree *); +static int update_current_filesystem(struct archive_read_disk *a, + int64_t dev); +static int setup_current_filesystem(struct archive_read_disk *); +static int tree_target_is_same_as_parent(struct tree *, const struct stat *); + +static int _archive_read_disk_open(struct archive *, const char *); +static int _archive_read_free(struct archive *); +static int _archive_read_close(struct archive *); +static int _archive_read_data_block(struct archive *, + const void **, size_t *, int64_t *); +static int _archive_read_next_header(struct archive *, + struct archive_entry **); +static int _archive_read_next_header2(struct archive *, + struct archive_entry *); +static const char *trivial_lookup_gname(void *, int64_t gid); +static const char *trivial_lookup_uname(void *, int64_t uid); +static int setup_sparse(struct archive_read_disk *, struct archive_entry *); +static int close_and_restore_time(int fd, struct tree *, + struct restore_time *); +static int open_on_current_dir(struct tree *, const char *, int); +static int tree_dup(int); + + +static struct archive_vtable * +archive_read_disk_vtable(void) +{ + static struct archive_vtable av; + static int inited = 0; + + if (!inited) { + av.archive_free = _archive_read_free; + av.archive_close = _archive_read_close; + av.archive_read_data_block = _archive_read_data_block; + av.archive_read_next_header = _archive_read_next_header; + av.archive_read_next_header2 = _archive_read_next_header2; + inited = 1; + } + return (&av); +} + +const char * +archive_read_disk_gname(struct archive *_a, la_int64_t gid) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + if (ARCHIVE_OK != __archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_disk_gname")) + return (NULL); + if (a->lookup_gname == NULL) + return (NULL); + return ((*a->lookup_gname)(a->lookup_gname_data, gid)); +} + +const char * +archive_read_disk_uname(struct archive *_a, la_int64_t uid) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + if (ARCHIVE_OK != __archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_disk_uname")) + return (NULL); + if (a->lookup_uname == NULL) + return (NULL); + return ((*a->lookup_uname)(a->lookup_uname_data, uid)); +} + +int +archive_read_disk_set_gname_lookup(struct archive *_a, + void *private_data, + const char * (*lookup_gname)(void *private, la_int64_t gid), + void (*cleanup_gname)(void *private)) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + archive_check_magic(&a->archive, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_disk_set_gname_lookup"); + + if (a->cleanup_gname != NULL && a->lookup_gname_data != NULL) + (a->cleanup_gname)(a->lookup_gname_data); + + a->lookup_gname = lookup_gname; + a->cleanup_gname = cleanup_gname; + a->lookup_gname_data = private_data; + return (ARCHIVE_OK); +} + +int +archive_read_disk_set_uname_lookup(struct archive *_a, + void *private_data, + const char * (*lookup_uname)(void *private, la_int64_t uid), + void (*cleanup_uname)(void *private)) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + archive_check_magic(&a->archive, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_disk_set_uname_lookup"); + + if (a->cleanup_uname != NULL && a->lookup_uname_data != NULL) + (a->cleanup_uname)(a->lookup_uname_data); + + a->lookup_uname = lookup_uname; + a->cleanup_uname = cleanup_uname; + a->lookup_uname_data = private_data; + return (ARCHIVE_OK); +} + +/* + * Create a new archive_read_disk object and initialize it with global state. + */ +struct archive * +archive_read_disk_new(void) +{ + struct archive_read_disk *a; + + a = (struct archive_read_disk *)calloc(1, sizeof(*a)); + if (a == NULL) + return (NULL); + a->archive.magic = ARCHIVE_READ_DISK_MAGIC; + a->archive.state = ARCHIVE_STATE_NEW; + a->archive.vtable = archive_read_disk_vtable(); + a->entry = archive_entry_new2(&a->archive); + a->lookup_uname = trivial_lookup_uname; + a->lookup_gname = trivial_lookup_gname; + a->flags = ARCHIVE_READDISK_MAC_COPYFILE; + a->open_on_current_dir = open_on_current_dir; + a->tree_current_dir_fd = tree_current_dir_fd; + a->tree_enter_working_dir = tree_enter_working_dir; + return (&a->archive); +} + +static int +_archive_read_free(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + int r; + + if (_a == NULL) + return (ARCHIVE_OK); + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_read_free"); + + if (a->archive.state != ARCHIVE_STATE_CLOSED) + r = _archive_read_close(&a->archive); + else + r = ARCHIVE_OK; + + tree_free(a->tree); + if (a->cleanup_gname != NULL && a->lookup_gname_data != NULL) + (a->cleanup_gname)(a->lookup_gname_data); + if (a->cleanup_uname != NULL && a->lookup_uname_data != NULL) + (a->cleanup_uname)(a->lookup_uname_data); + archive_string_free(&a->archive.error_string); + archive_entry_free(a->entry); + a->archive.magic = 0; + __archive_clean(&a->archive); + free(a); + return (r); +} + +static int +_archive_read_close(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_read_close"); + + if (a->archive.state != ARCHIVE_STATE_FATAL) + a->archive.state = ARCHIVE_STATE_CLOSED; + + tree_close(a->tree); + + return (ARCHIVE_OK); +} + +static void +setup_symlink_mode(struct archive_read_disk *a, char symlink_mode, + int follow_symlinks) +{ + a->symlink_mode = symlink_mode; + a->follow_symlinks = follow_symlinks; + if (a->tree != NULL) { + a->tree->initial_symlink_mode = a->symlink_mode; + a->tree->symlink_mode = a->symlink_mode; + } +} + +int +archive_read_disk_set_symlink_logical(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_disk_set_symlink_logical"); + setup_symlink_mode(a, 'L', 1); + return (ARCHIVE_OK); +} + +int +archive_read_disk_set_symlink_physical(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_disk_set_symlink_physical"); + setup_symlink_mode(a, 'P', 0); + return (ARCHIVE_OK); +} + +int +archive_read_disk_set_symlink_hybrid(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_disk_set_symlink_hybrid"); + setup_symlink_mode(a, 'H', 1);/* Follow symlinks initially. */ + return (ARCHIVE_OK); +} + +int +archive_read_disk_set_atime_restored(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_disk_restore_atime"); +#ifdef HAVE_UTIMES + a->flags |= ARCHIVE_READDISK_RESTORE_ATIME; + if (a->tree != NULL) + a->tree->flags |= needsRestoreTimes; + return (ARCHIVE_OK); +#else + /* Display warning and unset flag */ + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Cannot restore access time on this system"); + a->flags &= ~ARCHIVE_READDISK_RESTORE_ATIME; + return (ARCHIVE_WARN); +#endif +} + +int +archive_read_disk_set_behavior(struct archive *_a, int flags) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + int r = ARCHIVE_OK; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_disk_honor_nodump"); + + a->flags = flags; + + if (flags & ARCHIVE_READDISK_RESTORE_ATIME) + r = archive_read_disk_set_atime_restored(_a); + else { + if (a->tree != NULL) + a->tree->flags &= ~needsRestoreTimes; + } + return (r); +} + +/* + * Trivial implementations of gname/uname lookup functions. + * These are normally overridden by the client, but these stub + * versions ensure that we always have something that works. + */ +static const char * +trivial_lookup_gname(void *private_data, int64_t gid) +{ + (void)private_data; /* UNUSED */ + (void)gid; /* UNUSED */ + return (NULL); +} + +static const char * +trivial_lookup_uname(void *private_data, int64_t uid) +{ + (void)private_data; /* UNUSED */ + (void)uid; /* UNUSED */ + return (NULL); +} + +/* + * Allocate memory for the reading buffer adjusted to the filesystem + * alignment. + */ +static int +setup_suitable_read_buffer(struct archive_read_disk *a) +{ + struct tree *t = a->tree; + struct filesystem *cf = t->current_filesystem; + size_t asize; + size_t s; + + if (cf->allocation_ptr == NULL) { + /* If we couldn't get a filesystem alignment, + * we use 4096 as default value but we won't use + * O_DIRECT to open() and openat() operations. */ + long xfer_align = (cf->xfer_align == -1)?4096:cf->xfer_align; + + if (cf->max_xfer_size != -1) + asize = cf->max_xfer_size + xfer_align; + else { + long incr = cf->incr_xfer_size; + /* Some platform does not set a proper value to + * incr_xfer_size.*/ + if (incr < 0) + incr = cf->min_xfer_size; + if (cf->min_xfer_size < 0) { + incr = xfer_align; + asize = xfer_align; + } else + asize = cf->min_xfer_size; + + /* Increase a buffer size up to 64K bytes in + * a proper increment size. */ + while (asize < 1024*64) + asize += incr; + /* Take a margin to adjust to the filesystem + * alignment. */ + asize += xfer_align; + } + cf->allocation_ptr = malloc(asize); + if (cf->allocation_ptr == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Couldn't allocate memory"); + a->archive.state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + } + + /* + * Calculate proper address for the filesystem. + */ + s = (uintptr_t)cf->allocation_ptr; + s %= xfer_align; + if (s > 0) + s = xfer_align - s; + + /* + * Set a read buffer pointer in the proper alignment of + * the current filesystem. + */ + cf->buff = cf->allocation_ptr + s; + cf->buff_size = asize - xfer_align; + } + return (ARCHIVE_OK); +} + +static int +_archive_read_data_block(struct archive *_a, const void **buff, + size_t *size, int64_t *offset) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + struct tree *t = a->tree; + int r; + ssize_t bytes; + int64_t sparse_bytes; + size_t buffbytes; + int empty_sparse_region = 0; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_DATA, + "archive_read_data_block"); + + if (t->entry_eof || t->entry_remaining_bytes <= 0) { + r = ARCHIVE_EOF; + goto abort_read_data; + } + + /* + * Open the current file. + */ + if (t->entry_fd < 0) { + int flags = O_RDONLY | O_BINARY | O_CLOEXEC; + + /* + * Eliminate or reduce cache effects if we can. + * + * Carefully consider this to be enabled. + */ +#if defined(O_DIRECT) && 0/* Disabled for now */ + if (t->current_filesystem->xfer_align != -1 && + t->nlink == 1) + flags |= O_DIRECT; +#endif +#if defined(O_NOATIME) + /* + * Linux has O_NOATIME flag; use it if we need. + */ + if ((t->flags & needsRestoreTimes) != 0 && + t->restore_time.noatime == 0) + flags |= O_NOATIME; +#endif + t->entry_fd = open_on_current_dir(t, + tree_current_access_path(t), flags); + __archive_ensure_cloexec_flag(t->entry_fd); +#if defined(O_NOATIME) + /* + * When we did open the file with O_NOATIME flag, + * if successful, set 1 to t->restore_time.noatime + * not to restore an atime of the file later. + * if failed by EPERM, retry it without O_NOATIME flag. + */ + if (flags & O_NOATIME) { + if (t->entry_fd >= 0) + t->restore_time.noatime = 1; + else if (errno == EPERM) + flags &= ~O_NOATIME; + } +#endif + if (t->entry_fd < 0) { + archive_set_error(&a->archive, errno, + "Couldn't open %s", tree_current_path(t)); + r = ARCHIVE_FAILED; + tree_enter_initial_dir(t); + goto abort_read_data; + } + tree_enter_initial_dir(t); + } + + /* + * Allocate read buffer if not allocated. + */ + if (t->current_filesystem->allocation_ptr == NULL) { + r = setup_suitable_read_buffer(a); + if (r != ARCHIVE_OK) { + a->archive.state = ARCHIVE_STATE_FATAL; + goto abort_read_data; + } + } + t->entry_buff = t->current_filesystem->buff; + t->entry_buff_size = t->current_filesystem->buff_size; + + buffbytes = t->entry_buff_size; + if ((int64_t)buffbytes > t->current_sparse->length) + buffbytes = t->current_sparse->length; + + if (t->current_sparse->length == 0) + empty_sparse_region = 1; + + /* + * Skip hole. + * TODO: Should we consider t->current_filesystem->xfer_align? + */ + if (t->current_sparse->offset > t->entry_total) { + if (lseek(t->entry_fd, + (off_t)t->current_sparse->offset, SEEK_SET) < 0) { + archive_set_error(&a->archive, errno, "Seek error"); + r = ARCHIVE_FATAL; + a->archive.state = ARCHIVE_STATE_FATAL; + goto abort_read_data; + } + sparse_bytes = t->current_sparse->offset - t->entry_total; + t->entry_remaining_bytes -= sparse_bytes; + t->entry_total += sparse_bytes; + } + + /* + * Read file contents. + */ + if (buffbytes > 0) { + bytes = read(t->entry_fd, t->entry_buff, buffbytes); + if (bytes < 0) { + archive_set_error(&a->archive, errno, "Read error"); + r = ARCHIVE_FATAL; + a->archive.state = ARCHIVE_STATE_FATAL; + goto abort_read_data; + } + } else + bytes = 0; + /* + * Return an EOF unless we've read a leading empty sparse region, which + * is used to represent fully-sparse files. + */ + if (bytes == 0 && !empty_sparse_region) { + /* Get EOF */ + t->entry_eof = 1; + r = ARCHIVE_EOF; + goto abort_read_data; + } + *buff = t->entry_buff; + *size = bytes; + *offset = t->entry_total; + t->entry_total += bytes; + t->entry_remaining_bytes -= bytes; + if (t->entry_remaining_bytes == 0) { + /* Close the current file descriptor */ + close_and_restore_time(t->entry_fd, t, &t->restore_time); + t->entry_fd = -1; + t->entry_eof = 1; + } + t->current_sparse->offset += bytes; + t->current_sparse->length -= bytes; + if (t->current_sparse->length == 0 && !t->entry_eof) + t->current_sparse++; + return (ARCHIVE_OK); + +abort_read_data: + *buff = NULL; + *size = 0; + *offset = t->entry_total; + if (t->entry_fd >= 0) { + /* Close the current file descriptor */ + close_and_restore_time(t->entry_fd, t, &t->restore_time); + t->entry_fd = -1; + } + return (r); +} + +static int +next_entry(struct archive_read_disk *a, struct tree *t, + struct archive_entry *entry) +{ + const struct stat *st; /* info to use for this entry */ + const struct stat *lst;/* lstat() information */ + const char *name; + int delayed, delayed_errno, descend, r; + struct archive_string delayed_str; + + delayed = ARCHIVE_OK; + delayed_errno = 0; + archive_string_init(&delayed_str); + + st = NULL; + lst = NULL; + t->descend = 0; + do { + switch (tree_next(t)) { + case TREE_ERROR_FATAL: + archive_set_error(&a->archive, t->tree_errno, + "%s: Unable to continue traversing directory tree", + tree_current_path(t)); + a->archive.state = ARCHIVE_STATE_FATAL; + tree_enter_initial_dir(t); + return (ARCHIVE_FATAL); + case TREE_ERROR_DIR: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "%s: Couldn't visit directory", + tree_current_path(t)); + tree_enter_initial_dir(t); + return (ARCHIVE_FAILED); + case 0: + tree_enter_initial_dir(t); + return (ARCHIVE_EOF); + case TREE_POSTDESCENT: + case TREE_POSTASCENT: + break; + case TREE_REGULAR: + lst = tree_current_lstat(t); + if (lst == NULL) { + if (errno == ENOENT && t->depth > 0) { + delayed = ARCHIVE_WARN; + delayed_errno = errno; + if (delayed_str.length == 0) { + archive_string_sprintf(&delayed_str, + "%s", tree_current_path(t)); + } else { + archive_string_sprintf(&delayed_str, + " %s", tree_current_path(t)); + } + } else { + archive_set_error(&a->archive, errno, + "%s: Cannot stat", + tree_current_path(t)); + tree_enter_initial_dir(t); + return (ARCHIVE_FAILED); + } + } + break; + } + } while (lst == NULL); + +#ifdef __APPLE__ + if (a->flags & ARCHIVE_READDISK_MAC_COPYFILE) { + /* If we're using copyfile(), ignore "._XXX" files. */ + const char *bname = strrchr(tree_current_path(t), '/'); + if (bname == NULL) + bname = tree_current_path(t); + else + ++bname; + if (bname[0] == '.' && bname[1] == '_') + return (ARCHIVE_RETRY); + } +#endif + + archive_entry_copy_pathname(entry, tree_current_path(t)); + /* + * Perform path matching. + */ + if (a->matching) { + r = archive_match_path_excluded(a->matching, entry); + if (r < 0) { + archive_set_error(&(a->archive), errno, + "Failed : %s", archive_error_string(a->matching)); + return (r); + } + if (r) { + if (a->excluded_cb_func) + a->excluded_cb_func(&(a->archive), + a->excluded_cb_data, entry); + return (ARCHIVE_RETRY); + } + } + + /* + * Distinguish 'L'/'P'/'H' symlink following. + */ + switch(t->symlink_mode) { + case 'H': + /* 'H': After the first item, rest like 'P'. */ + t->symlink_mode = 'P'; + /* 'H': First item (from command line) like 'L'. */ + /* FALLTHROUGH */ + case 'L': + /* 'L': Do descend through a symlink to dir. */ + descend = tree_current_is_dir(t); + /* 'L': Follow symlinks to files. */ + a->symlink_mode = 'L'; + a->follow_symlinks = 1; + /* 'L': Archive symlinks as targets, if we can. */ + st = tree_current_stat(t); + if (st != NULL && !tree_target_is_same_as_parent(t, st)) + break; + /* If stat fails, we have a broken symlink; + * in that case, don't follow the link. */ + /* FALLTHROUGH */ + default: + /* 'P': Don't descend through a symlink to dir. */ + descend = tree_current_is_physical_dir(t); + /* 'P': Don't follow symlinks to files. */ + a->symlink_mode = 'P'; + a->follow_symlinks = 0; + /* 'P': Archive symlinks as symlinks. */ + st = lst; + break; + } + + if (update_current_filesystem(a, st->st_dev) != ARCHIVE_OK) { + a->archive.state = ARCHIVE_STATE_FATAL; + tree_enter_initial_dir(t); + return (ARCHIVE_FATAL); + } + if (t->initial_filesystem_id == -1) + t->initial_filesystem_id = t->current_filesystem_id; + if (a->flags & ARCHIVE_READDISK_NO_TRAVERSE_MOUNTS) { + if (t->initial_filesystem_id != t->current_filesystem_id) + descend = 0; + } + t->descend = descend; + + /* + * Honor nodump flag. + * If the file is marked with nodump flag, do not return this entry. + */ + if (a->flags & ARCHIVE_READDISK_HONOR_NODUMP) { +#if defined(HAVE_STRUCT_STAT_ST_FLAGS) && defined(UF_NODUMP) + if (st->st_flags & UF_NODUMP) + return (ARCHIVE_RETRY); +#elif (defined(FS_IOC_GETFLAGS) && defined(FS_NODUMP_FL) && \ + defined(HAVE_WORKING_FS_IOC_GETFLAGS)) || \ + (defined(EXT2_IOC_GETFLAGS) && defined(EXT2_NODUMP_FL) && \ + defined(HAVE_WORKING_EXT2_IOC_GETFLAGS)) + if (S_ISREG(st->st_mode) || S_ISDIR(st->st_mode)) { + int stflags; + + t->entry_fd = open_on_current_dir(t, + tree_current_access_path(t), + O_RDONLY | O_NONBLOCK | O_CLOEXEC); + __archive_ensure_cloexec_flag(t->entry_fd); + if (t->entry_fd >= 0) { + r = ioctl(t->entry_fd, +#ifdef FS_IOC_GETFLAGS + FS_IOC_GETFLAGS, +#else + EXT2_IOC_GETFLAGS, +#endif + &stflags); +#ifdef FS_NODUMP_FL + if (r == 0 && (stflags & FS_NODUMP_FL) != 0) +#else + if (r == 0 && (stflags & EXT2_NODUMP_FL) != 0) +#endif + return (ARCHIVE_RETRY); + } + } +#endif + } + + archive_entry_copy_stat(entry, st); + + /* Save the times to be restored. This must be in before + * calling archive_read_disk_descend() or any chance of it, + * especially, invoking a callback. */ + t->restore_time.mtime = archive_entry_mtime(entry); + t->restore_time.mtime_nsec = archive_entry_mtime_nsec(entry); + t->restore_time.atime = archive_entry_atime(entry); + t->restore_time.atime_nsec = archive_entry_atime_nsec(entry); + t->restore_time.filetype = archive_entry_filetype(entry); + t->restore_time.noatime = t->current_filesystem->noatime; + + /* + * Perform time matching. + */ + if (a->matching) { + r = archive_match_time_excluded(a->matching, entry); + if (r < 0) { + archive_set_error(&(a->archive), errno, + "Failed : %s", archive_error_string(a->matching)); + return (r); + } + if (r) { + if (a->excluded_cb_func) + a->excluded_cb_func(&(a->archive), + a->excluded_cb_data, entry); + return (ARCHIVE_RETRY); + } + } + + /* Lookup uname/gname */ + name = archive_read_disk_uname(&(a->archive), archive_entry_uid(entry)); + if (name != NULL) + archive_entry_copy_uname(entry, name); + name = archive_read_disk_gname(&(a->archive), archive_entry_gid(entry)); + if (name != NULL) + archive_entry_copy_gname(entry, name); + + /* + * Perform owner matching. + */ + if (a->matching) { + r = archive_match_owner_excluded(a->matching, entry); + if (r < 0) { + archive_set_error(&(a->archive), errno, + "Failed : %s", archive_error_string(a->matching)); + return (r); + } + if (r) { + if (a->excluded_cb_func) + a->excluded_cb_func(&(a->archive), + a->excluded_cb_data, entry); + return (ARCHIVE_RETRY); + } + } + + /* + * Invoke a meta data filter callback. + */ + if (a->metadata_filter_func) { + if (!a->metadata_filter_func(&(a->archive), + a->metadata_filter_data, entry)) + return (ARCHIVE_RETRY); + } + + /* + * Populate the archive_entry with metadata from the disk. + */ + archive_entry_copy_sourcepath(entry, tree_current_access_path(t)); + r = archive_read_disk_entry_from_file(&(a->archive), entry, + t->entry_fd, st); + + if (r == ARCHIVE_OK) { + r = delayed; + if (r != ARCHIVE_OK) { + archive_string_sprintf(&delayed_str, ": %s", + "File removed before we read it"); + archive_set_error(&(a->archive), delayed_errno, + "%s", delayed_str.s); + } + } + archive_string_free(&delayed_str); + + return (r); +} + +static int +_archive_read_next_header(struct archive *_a, struct archive_entry **entryp) +{ + int ret; + struct archive_read_disk *a = (struct archive_read_disk *)_a; + *entryp = NULL; + ret = _archive_read_next_header2(_a, a->entry); + *entryp = a->entry; + return ret; +} + +static int +_archive_read_next_header2(struct archive *_a, struct archive_entry *entry) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + struct tree *t; + int r; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, + "archive_read_next_header2"); + + t = a->tree; + if (t->entry_fd >= 0) { + close_and_restore_time(t->entry_fd, t, &t->restore_time); + t->entry_fd = -1; + } + + archive_entry_clear(entry); + + for (;;) { + r = next_entry(a, t, entry); + if (t->entry_fd >= 0) { + close(t->entry_fd); + t->entry_fd = -1; + } + + if (r == ARCHIVE_RETRY) { + archive_entry_clear(entry); + continue; + } + break; + } + + /* Return to the initial directory. */ + tree_enter_initial_dir(t); + + /* + * EOF and FATAL are persistent at this layer. By + * modifying the state, we guarantee that future calls to + * read a header or read data will fail. + */ + switch (r) { + case ARCHIVE_EOF: + a->archive.state = ARCHIVE_STATE_EOF; + break; + case ARCHIVE_OK: + case ARCHIVE_WARN: + /* Overwrite the sourcepath based on the initial directory. */ + archive_entry_copy_sourcepath(entry, tree_current_path(t)); + t->entry_total = 0; + if (archive_entry_filetype(entry) == AE_IFREG) { + t->nlink = archive_entry_nlink(entry); + t->entry_remaining_bytes = archive_entry_size(entry); + t->entry_eof = (t->entry_remaining_bytes == 0)? 1: 0; + if (!t->entry_eof && + setup_sparse(a, entry) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } else { + t->entry_remaining_bytes = 0; + t->entry_eof = 1; + } + a->archive.state = ARCHIVE_STATE_DATA; + break; + case ARCHIVE_RETRY: + break; + case ARCHIVE_FATAL: + a->archive.state = ARCHIVE_STATE_FATAL; + break; + } + + __archive_reset_read_data(&a->archive); + return (r); +} + +static int +setup_sparse(struct archive_read_disk *a, struct archive_entry *entry) +{ + struct tree *t = a->tree; + int64_t length, offset; + int i; + + t->sparse_count = archive_entry_sparse_reset(entry); + if (t->sparse_count+1 > t->sparse_list_size) { + free(t->sparse_list); + t->sparse_list_size = t->sparse_count + 1; + t->sparse_list = malloc(sizeof(t->sparse_list[0]) * + t->sparse_list_size); + if (t->sparse_list == NULL) { + t->sparse_list_size = 0; + archive_set_error(&a->archive, ENOMEM, + "Can't allocate data"); + a->archive.state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + } + } + for (i = 0; i < t->sparse_count; i++) { + archive_entry_sparse_next(entry, &offset, &length); + t->sparse_list[i].offset = offset; + t->sparse_list[i].length = length; + } + if (i == 0) { + t->sparse_list[i].offset = 0; + t->sparse_list[i].length = archive_entry_size(entry); + } else { + t->sparse_list[i].offset = archive_entry_size(entry); + t->sparse_list[i].length = 0; + } + t->current_sparse = t->sparse_list; + + return (ARCHIVE_OK); +} + +int +archive_read_disk_set_matching(struct archive *_a, struct archive *_ma, + void (*_excluded_func)(struct archive *, void *, struct archive_entry *), + void *_client_data) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_disk_set_matching"); + a->matching = _ma; + a->excluded_cb_func = _excluded_func; + a->excluded_cb_data = _client_data; + return (ARCHIVE_OK); +} + +int +archive_read_disk_set_metadata_filter_callback(struct archive *_a, + int (*_metadata_filter_func)(struct archive *, void *, + struct archive_entry *), void *_client_data) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, + "archive_read_disk_set_metadata_filter_callback"); + + a->metadata_filter_func = _metadata_filter_func; + a->metadata_filter_data = _client_data; + return (ARCHIVE_OK); +} + +int +archive_read_disk_can_descend(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + struct tree *t = a->tree; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, + "archive_read_disk_can_descend"); + + return (t->visit_type == TREE_REGULAR && t->descend); +} + +/* + * Called by the client to mark the directory just returned from + * tree_next() as needing to be visited. + */ +int +archive_read_disk_descend(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + struct tree *t = a->tree; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, + "archive_read_disk_descend"); + + if (t->visit_type != TREE_REGULAR || !t->descend) + return (ARCHIVE_OK); + + /* + * We must not treat the initial specified path as a physical dir, + * because if we do then we will try and ascend out of it by opening + * ".." which is (a) wrong and (b) causes spurious permissions errors + * if ".." is not readable by us. Instead, treat it as if it were a + * symlink. (This uses an extra fd, but it can only happen once at the + * top level of a traverse.) But we can't necessarily assume t->st is + * valid here (though t->lst is), which complicates the logic a + * little. + */ + if (tree_current_is_physical_dir(t)) { + tree_push(t, t->basename, t->current_filesystem_id, + t->lst.st_dev, t->lst.st_ino, &t->restore_time); + if (t->stack->parent->parent != NULL) + t->stack->flags |= isDir; + else + t->stack->flags |= isDirLink; + } else if (tree_current_is_dir(t)) { + tree_push(t, t->basename, t->current_filesystem_id, + t->st.st_dev, t->st.st_ino, &t->restore_time); + t->stack->flags |= isDirLink; + } + t->descend = 0; + return (ARCHIVE_OK); +} + +int +archive_read_disk_open(struct archive *_a, const char *pathname) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_NEW | ARCHIVE_STATE_CLOSED, + "archive_read_disk_open"); + archive_clear_error(&a->archive); + + return (_archive_read_disk_open(_a, pathname)); +} + +int +archive_read_disk_open_w(struct archive *_a, const wchar_t *pathname) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + struct archive_string path; + int ret; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_NEW | ARCHIVE_STATE_CLOSED, + "archive_read_disk_open_w"); + archive_clear_error(&a->archive); + + /* Make a char string from a wchar_t string. */ + archive_string_init(&path); + if (archive_string_append_from_wcs(&path, pathname, + wcslen(pathname)) != 0) { + if (errno == ENOMEM) + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + else + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Can't convert a path to a char string"); + a->archive.state = ARCHIVE_STATE_FATAL; + ret = ARCHIVE_FATAL; + } else + ret = _archive_read_disk_open(_a, path.s); + + archive_string_free(&path); + return (ret); +} + +static int +_archive_read_disk_open(struct archive *_a, const char *pathname) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + + if (a->tree != NULL) + a->tree = tree_reopen(a->tree, pathname, + a->flags & ARCHIVE_READDISK_RESTORE_ATIME); + else + a->tree = tree_open(pathname, a->symlink_mode, + a->flags & ARCHIVE_READDISK_RESTORE_ATIME); + if (a->tree == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate tar data"); + a->archive.state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + } + a->archive.state = ARCHIVE_STATE_HEADER; + + return (ARCHIVE_OK); +} + +/* + * Return a current filesystem ID which is index of the filesystem entry + * you've visited through archive_read_disk. + */ +int +archive_read_disk_current_filesystem(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_DATA, + "archive_read_disk_current_filesystem"); + + return (a->tree->current_filesystem_id); +} + +static int +update_current_filesystem(struct archive_read_disk *a, int64_t dev) +{ + struct tree *t = a->tree; + int i, fid; + + if (t->current_filesystem != NULL && + t->current_filesystem->dev == dev) + return (ARCHIVE_OK); + + for (i = 0; i < t->max_filesystem_id; i++) { + if (t->filesystem_table[i].dev == dev) { + /* There is the filesystem ID we've already generated. */ + t->current_filesystem_id = i; + t->current_filesystem = &(t->filesystem_table[i]); + return (ARCHIVE_OK); + } + } + + /* + * This is the new filesystem which we have to generate a new ID for. + */ + fid = t->max_filesystem_id++; + if (t->max_filesystem_id > t->allocated_filesystem) { + size_t s; + void *p; + + s = t->max_filesystem_id * 2; + p = realloc(t->filesystem_table, + s * sizeof(*t->filesystem_table)); + if (p == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate tar data"); + return (ARCHIVE_FATAL); + } + t->filesystem_table = (struct filesystem *)p; + t->allocated_filesystem = s; + } + t->current_filesystem_id = fid; + t->current_filesystem = &(t->filesystem_table[fid]); + t->current_filesystem->dev = dev; + t->current_filesystem->allocation_ptr = NULL; + t->current_filesystem->buff = NULL; + + /* Setup the current filesystem properties which depend on + * platform specific. */ + return (setup_current_filesystem(a)); +} + +/* + * Returns 1 if current filesystem is generated filesystem, 0 if it is not + * or -1 if it is unknown. + */ +int +archive_read_disk_current_filesystem_is_synthetic(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_DATA, + "archive_read_disk_current_filesystem"); + + return (a->tree->current_filesystem->synthetic); +} + +/* + * Returns 1 if current filesystem is remote filesystem, 0 if it is not + * or -1 if it is unknown. + */ +int +archive_read_disk_current_filesystem_is_remote(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_DATA, + "archive_read_disk_current_filesystem"); + + return (a->tree->current_filesystem->remote); +} + +#if defined(_PC_REC_INCR_XFER_SIZE) && defined(_PC_REC_MAX_XFER_SIZE) &&\ + defined(_PC_REC_MIN_XFER_SIZE) && defined(_PC_REC_XFER_ALIGN) +static int +get_xfer_size(struct tree *t, int fd, const char *path) +{ + t->current_filesystem->xfer_align = -1; + errno = 0; + if (fd >= 0) { + t->current_filesystem->incr_xfer_size = + fpathconf(fd, _PC_REC_INCR_XFER_SIZE); + t->current_filesystem->max_xfer_size = + fpathconf(fd, _PC_REC_MAX_XFER_SIZE); + t->current_filesystem->min_xfer_size = + fpathconf(fd, _PC_REC_MIN_XFER_SIZE); + t->current_filesystem->xfer_align = + fpathconf(fd, _PC_REC_XFER_ALIGN); + } else if (path != NULL) { + t->current_filesystem->incr_xfer_size = + pathconf(path, _PC_REC_INCR_XFER_SIZE); + t->current_filesystem->max_xfer_size = + pathconf(path, _PC_REC_MAX_XFER_SIZE); + t->current_filesystem->min_xfer_size = + pathconf(path, _PC_REC_MIN_XFER_SIZE); + t->current_filesystem->xfer_align = + pathconf(path, _PC_REC_XFER_ALIGN); + } + /* At least we need an alignment size. */ + if (t->current_filesystem->xfer_align == -1) + return ((errno == EINVAL)?1:-1); + else + return (0); +} +#else +static int +get_xfer_size(struct tree *t, int fd, const char *path) +{ + (void)t; /* UNUSED */ + (void)fd; /* UNUSED */ + (void)path; /* UNUSED */ + return (1);/* Not supported */ +} +#endif + +#if defined(HAVE_STATFS) && defined(HAVE_FSTATFS) && defined(MNT_LOCAL) \ + && !defined(ST_LOCAL) + +/* + * Gather current filesystem properties on FreeBSD, OpenBSD and Mac OS X. + */ +static int +setup_current_filesystem(struct archive_read_disk *a) +{ + struct tree *t = a->tree; + struct statfs sfs; +#if defined(HAVE_GETVFSBYNAME) && defined(VFCF_SYNTHETIC) +/* TODO: configure should set GETVFSBYNAME_ARG_TYPE to make + * this accurate; some platforms have both and we need the one that's + * used by getvfsbyname() + * + * Then the following would become: + * #if defined(GETVFSBYNAME_ARG_TYPE) + * GETVFSBYNAME_ARG_TYPE vfc; + * #endif + */ +# if defined(HAVE_STRUCT_XVFSCONF) + struct xvfsconf vfc; +# else + struct vfsconf vfc; +# endif +#endif + int r, xr = 0; +#if !defined(HAVE_STRUCT_STATFS_F_NAMEMAX) + long nm; +#endif + + t->current_filesystem->synthetic = -1; + t->current_filesystem->remote = -1; + if (tree_current_is_symblic_link_target(t)) { +#if defined(HAVE_OPENAT) + /* + * Get file system statistics on any directory + * where current is. + */ + int fd = openat(tree_current_dir_fd(t), + tree_current_access_path(t), O_RDONLY | O_CLOEXEC); + __archive_ensure_cloexec_flag(fd); + if (fd < 0) { + archive_set_error(&a->archive, errno, + "openat failed"); + return (ARCHIVE_FAILED); + } + r = fstatfs(fd, &sfs); + if (r == 0) + xr = get_xfer_size(t, fd, NULL); + close(fd); +#else + if (tree_enter_working_dir(t) != 0) { + archive_set_error(&a->archive, errno, "fchdir failed"); + return (ARCHIVE_FAILED); + } + r = statfs(tree_current_access_path(t), &sfs); + if (r == 0) + xr = get_xfer_size(t, -1, tree_current_access_path(t)); +#endif + } else { + r = fstatfs(tree_current_dir_fd(t), &sfs); + if (r == 0) + xr = get_xfer_size(t, tree_current_dir_fd(t), NULL); + } + if (r == -1 || xr == -1) { + archive_set_error(&a->archive, errno, "statfs failed"); + return (ARCHIVE_FAILED); + } else if (xr == 1) { + /* pathconf(_PC_REX_*) operations are not supported. */ + t->current_filesystem->xfer_align = sfs.f_bsize; + t->current_filesystem->max_xfer_size = -1; + t->current_filesystem->min_xfer_size = sfs.f_iosize; + t->current_filesystem->incr_xfer_size = sfs.f_iosize; + } + if (sfs.f_flags & MNT_LOCAL) + t->current_filesystem->remote = 0; + else + t->current_filesystem->remote = 1; + +#if defined(HAVE_GETVFSBYNAME) && defined(VFCF_SYNTHETIC) + r = getvfsbyname(sfs.f_fstypename, &vfc); + if (r == -1) { + archive_set_error(&a->archive, errno, "getvfsbyname failed"); + return (ARCHIVE_FAILED); + } + if (vfc.vfc_flags & VFCF_SYNTHETIC) + t->current_filesystem->synthetic = 1; + else + t->current_filesystem->synthetic = 0; +#endif + +#if defined(MNT_NOATIME) + if (sfs.f_flags & MNT_NOATIME) + t->current_filesystem->noatime = 1; + else +#endif + t->current_filesystem->noatime = 0; + +#if defined(USE_READDIR_R) + /* Set maximum filename length. */ +#if defined(HAVE_STRUCT_STATFS_F_NAMEMAX) + t->current_filesystem->name_max = sfs.f_namemax; +#else +# if defined(_PC_NAME_MAX) + /* Mac OS X does not have f_namemax in struct statfs. */ + if (tree_current_is_symblic_link_target(t)) { + if (tree_enter_working_dir(t) != 0) { + archive_set_error(&a->archive, errno, "fchdir failed"); + return (ARCHIVE_FAILED); + } + nm = pathconf(tree_current_access_path(t), _PC_NAME_MAX); + } else + nm = fpathconf(tree_current_dir_fd(t), _PC_NAME_MAX); +# else + nm = -1; +# endif + if (nm == -1) + t->current_filesystem->name_max = NAME_MAX; + else + t->current_filesystem->name_max = nm; +#endif +#endif /* USE_READDIR_R */ + return (ARCHIVE_OK); +} + +#elif (defined(HAVE_STATVFS) || defined(HAVE_FSTATVFS)) && defined(ST_LOCAL) + +/* + * Gather current filesystem properties on NetBSD + */ +static int +setup_current_filesystem(struct archive_read_disk *a) +{ + struct tree *t = a->tree; + struct statvfs svfs; + int r, xr = 0; + + t->current_filesystem->synthetic = -1; + if (tree_enter_working_dir(t) != 0) { + archive_set_error(&a->archive, errno, "fchdir failed"); + return (ARCHIVE_FAILED); + } + if (tree_current_is_symblic_link_target(t)) { + r = statvfs(tree_current_access_path(t), &svfs); + if (r == 0) + xr = get_xfer_size(t, -1, tree_current_access_path(t)); + } else { +#ifdef HAVE_FSTATVFS + r = fstatvfs(tree_current_dir_fd(t), &svfs); + if (r == 0) + xr = get_xfer_size(t, tree_current_dir_fd(t), NULL); +#else + r = statvfs(".", &svfs); + if (r == 0) + xr = get_xfer_size(t, -1, "."); +#endif + } + if (r == -1 || xr == -1) { + t->current_filesystem->remote = -1; + archive_set_error(&a->archive, errno, "statvfs failed"); + return (ARCHIVE_FAILED); + } else if (xr == 1) { + /* Usually come here unless NetBSD supports _PC_REC_XFER_ALIGN + * for pathconf() function. */ + t->current_filesystem->xfer_align = svfs.f_frsize; + t->current_filesystem->max_xfer_size = -1; +#if defined(HAVE_STRUCT_STATVFS_F_IOSIZE) + t->current_filesystem->min_xfer_size = svfs.f_iosize; + t->current_filesystem->incr_xfer_size = svfs.f_iosize; +#else + t->current_filesystem->min_xfer_size = svfs.f_bsize; + t->current_filesystem->incr_xfer_size = svfs.f_bsize; +#endif + } + if (svfs.f_flag & ST_LOCAL) + t->current_filesystem->remote = 0; + else + t->current_filesystem->remote = 1; + +#if defined(ST_NOATIME) + if (svfs.f_flag & ST_NOATIME) + t->current_filesystem->noatime = 1; + else +#endif + t->current_filesystem->noatime = 0; + + /* Set maximum filename length. */ + t->current_filesystem->name_max = svfs.f_namemax; + return (ARCHIVE_OK); +} + +#elif defined(HAVE_SYS_STATFS_H) && defined(HAVE_LINUX_MAGIC_H) &&\ + defined(HAVE_STATFS) && defined(HAVE_FSTATFS) +/* + * Note: statfs is deprecated since LSB 3.2 + */ + +#ifndef CIFS_SUPER_MAGIC +#define CIFS_SUPER_MAGIC 0xFF534D42 +#endif +#ifndef DEVFS_SUPER_MAGIC +#define DEVFS_SUPER_MAGIC 0x1373 +#endif + +/* + * Gather current filesystem properties on Linux + */ +static int +setup_current_filesystem(struct archive_read_disk *a) +{ + struct tree *t = a->tree; + struct statfs sfs; +#if defined(HAVE_STATVFS) + struct statvfs svfs; +#endif + int r, vr = 0, xr = 0; + + if (tree_current_is_symblic_link_target(t)) { +#if defined(HAVE_OPENAT) + /* + * Get file system statistics on any directory + * where current is. + */ + int fd = openat(tree_current_dir_fd(t), + tree_current_access_path(t), O_RDONLY | O_CLOEXEC); + __archive_ensure_cloexec_flag(fd); + if (fd < 0) { + archive_set_error(&a->archive, errno, + "openat failed"); + return (ARCHIVE_FAILED); + } +#if defined(HAVE_FSTATVFS) + vr = fstatvfs(fd, &svfs);/* for f_flag, mount flags */ +#endif + r = fstatfs(fd, &sfs); + if (r == 0) + xr = get_xfer_size(t, fd, NULL); + close(fd); +#else + if (tree_enter_working_dir(t) != 0) { + archive_set_error(&a->archive, errno, "fchdir failed"); + return (ARCHIVE_FAILED); + } +#if defined(HAVE_STATVFS) + vr = statvfs(tree_current_access_path(t), &svfs); +#endif + r = statfs(tree_current_access_path(t), &sfs); + if (r == 0) + xr = get_xfer_size(t, -1, tree_current_access_path(t)); +#endif + } else { +#ifdef HAVE_FSTATFS +#if defined(HAVE_FSTATVFS) + vr = fstatvfs(tree_current_dir_fd(t), &svfs); +#endif + r = fstatfs(tree_current_dir_fd(t), &sfs); + if (r == 0) + xr = get_xfer_size(t, tree_current_dir_fd(t), NULL); +#else + if (tree_enter_working_dir(t) != 0) { + archive_set_error(&a->archive, errno, "fchdir failed"); + return (ARCHIVE_FAILED); + } +#if defined(HAVE_STATVFS) + vr = statvfs(".", &svfs); +#endif + r = statfs(".", &sfs); + if (r == 0) + xr = get_xfer_size(t, -1, "."); +#endif + } + if (r == -1 || xr == -1 || vr == -1) { + t->current_filesystem->synthetic = -1; + t->current_filesystem->remote = -1; + archive_set_error(&a->archive, errno, "statfs failed"); + return (ARCHIVE_FAILED); + } else if (xr == 1) { + /* pathconf(_PC_REX_*) operations are not supported. */ +#if defined(HAVE_STATVFS) + t->current_filesystem->xfer_align = svfs.f_frsize; + t->current_filesystem->max_xfer_size = -1; + t->current_filesystem->min_xfer_size = svfs.f_bsize; + t->current_filesystem->incr_xfer_size = svfs.f_bsize; +#else + t->current_filesystem->xfer_align = sfs.f_frsize; + t->current_filesystem->max_xfer_size = -1; + t->current_filesystem->min_xfer_size = sfs.f_bsize; + t->current_filesystem->incr_xfer_size = sfs.f_bsize; +#endif + } + switch (sfs.f_type) { + case AFS_SUPER_MAGIC: + case CIFS_SUPER_MAGIC: + case CODA_SUPER_MAGIC: + case NCP_SUPER_MAGIC:/* NetWare */ + case NFS_SUPER_MAGIC: + case SMB_SUPER_MAGIC: + t->current_filesystem->remote = 1; + t->current_filesystem->synthetic = 0; + break; + case DEVFS_SUPER_MAGIC: + case PROC_SUPER_MAGIC: + case USBDEVICE_SUPER_MAGIC: + t->current_filesystem->remote = 0; + t->current_filesystem->synthetic = 1; + break; + default: + t->current_filesystem->remote = 0; + t->current_filesystem->synthetic = 0; + break; + } + +#if defined(ST_NOATIME) +#if defined(HAVE_STATVFS) + if (svfs.f_flag & ST_NOATIME) +#else + if (sfs.f_flags & ST_NOATIME) +#endif + t->current_filesystem->noatime = 1; + else +#endif + t->current_filesystem->noatime = 0; + +#if defined(USE_READDIR_R) + /* Set maximum filename length. */ + t->current_filesystem->name_max = sfs.f_namelen; +#endif + return (ARCHIVE_OK); +} + +#elif defined(HAVE_SYS_STATVFS_H) &&\ + (defined(HAVE_STATVFS) || defined(HAVE_FSTATVFS)) + +/* + * Gather current filesystem properties on other posix platform. + */ +static int +setup_current_filesystem(struct archive_read_disk *a) +{ + struct tree *t = a->tree; + struct statvfs svfs; + int r, xr = 0; + + t->current_filesystem->synthetic = -1;/* Not supported */ + t->current_filesystem->remote = -1;/* Not supported */ + if (tree_current_is_symblic_link_target(t)) { +#if defined(HAVE_OPENAT) + /* + * Get file system statistics on any directory + * where current is. + */ + int fd = openat(tree_current_dir_fd(t), + tree_current_access_path(t), O_RDONLY | O_CLOEXEC); + __archive_ensure_cloexec_flag(fd); + if (fd < 0) { + archive_set_error(&a->archive, errno, + "openat failed"); + return (ARCHIVE_FAILED); + } + r = fstatvfs(fd, &svfs); + if (r == 0) + xr = get_xfer_size(t, fd, NULL); + close(fd); +#else + if (tree_enter_working_dir(t) != 0) { + archive_set_error(&a->archive, errno, "fchdir failed"); + return (ARCHIVE_FAILED); + } + r = statvfs(tree_current_access_path(t), &svfs); + if (r == 0) + xr = get_xfer_size(t, -1, tree_current_access_path(t)); +#endif + } else { +#ifdef HAVE_FSTATVFS + r = fstatvfs(tree_current_dir_fd(t), &svfs); + if (r == 0) + xr = get_xfer_size(t, tree_current_dir_fd(t), NULL); +#else + if (tree_enter_working_dir(t) != 0) { + archive_set_error(&a->archive, errno, "fchdir failed"); + return (ARCHIVE_FAILED); + } + r = statvfs(".", &svfs); + if (r == 0) + xr = get_xfer_size(t, -1, "."); +#endif + } + if (r == -1 || xr == -1) { + t->current_filesystem->synthetic = -1; + t->current_filesystem->remote = -1; + archive_set_error(&a->archive, errno, "statvfs failed"); + return (ARCHIVE_FAILED); + } else if (xr == 1) { + /* pathconf(_PC_REX_*) operations are not supported. */ + t->current_filesystem->xfer_align = svfs.f_frsize; + t->current_filesystem->max_xfer_size = -1; + t->current_filesystem->min_xfer_size = svfs.f_bsize; + t->current_filesystem->incr_xfer_size = svfs.f_bsize; + } + +#if defined(ST_NOATIME) + if (svfs.f_flag & ST_NOATIME) + t->current_filesystem->noatime = 1; + else +#endif + t->current_filesystem->noatime = 0; + +#if defined(USE_READDIR_R) + /* Set maximum filename length. */ + t->current_filesystem->name_max = svfs.f_namemax; +#endif + return (ARCHIVE_OK); +} + +#else + +/* + * Generic: Gather current filesystem properties. + * TODO: Is this generic function really needed? + */ +static int +setup_current_filesystem(struct archive_read_disk *a) +{ + struct tree *t = a->tree; +#if defined(_PC_NAME_MAX) && defined(USE_READDIR_R) + long nm; +#endif + t->current_filesystem->synthetic = -1;/* Not supported */ + t->current_filesystem->remote = -1;/* Not supported */ + t->current_filesystem->noatime = 0; + (void)get_xfer_size(t, -1, ".");/* Dummy call to avoid build error. */ + t->current_filesystem->xfer_align = -1;/* Unknown */ + t->current_filesystem->max_xfer_size = -1; + t->current_filesystem->min_xfer_size = -1; + t->current_filesystem->incr_xfer_size = -1; + +#if defined(USE_READDIR_R) + /* Set maximum filename length. */ +# if defined(_PC_NAME_MAX) + if (tree_current_is_symblic_link_target(t)) { + if (tree_enter_working_dir(t) != 0) { + archive_set_error(&a->archive, errno, "fchdir failed"); + return (ARCHIVE_FAILED); + } + nm = pathconf(tree_current_access_path(t), _PC_NAME_MAX); + } else + nm = fpathconf(tree_current_dir_fd(t), _PC_NAME_MAX); + if (nm == -1) +# endif /* _PC_NAME_MAX */ + /* + * Some systems (HP-UX or others?) incorrectly defined + * NAME_MAX macro to be a smaller value. + */ +# if defined(NAME_MAX) && NAME_MAX >= 255 + t->current_filesystem->name_max = NAME_MAX; +# else + /* No way to get a trusted value of maximum filename + * length. */ + t->current_filesystem->name_max = PATH_MAX; +# endif /* NAME_MAX */ +# if defined(_PC_NAME_MAX) + else + t->current_filesystem->name_max = nm; +# endif /* _PC_NAME_MAX */ +#endif /* USE_READDIR_R */ + return (ARCHIVE_OK); +} + +#endif + +static int +close_and_restore_time(int fd, struct tree *t, struct restore_time *rt) +{ +#ifndef HAVE_UTIMES + (void)t; /* UNUSED */ + (void)rt; /* UNUSED */ + return (close(fd)); +#else +#if defined(HAVE_FUTIMENS) && !defined(__CYGWIN__) + struct timespec timespecs[2]; +#endif + struct timeval times[2]; + + if ((t->flags & needsRestoreTimes) == 0 || rt->noatime) { + if (fd >= 0) + return (close(fd)); + else + return (0); + } + +#if defined(HAVE_FUTIMENS) && !defined(__CYGWIN__) + timespecs[1].tv_sec = rt->mtime; + timespecs[1].tv_nsec = rt->mtime_nsec; + + timespecs[0].tv_sec = rt->atime; + timespecs[0].tv_nsec = rt->atime_nsec; + /* futimens() is defined in POSIX.1-2008. */ + if (futimens(fd, timespecs) == 0) + return (close(fd)); +#endif + + times[1].tv_sec = rt->mtime; + times[1].tv_usec = rt->mtime_nsec / 1000; + + times[0].tv_sec = rt->atime; + times[0].tv_usec = rt->atime_nsec / 1000; + +#if !defined(HAVE_FUTIMENS) && defined(HAVE_FUTIMES) && !defined(__CYGWIN__) + if (futimes(fd, times) == 0) + return (close(fd)); +#endif + close(fd); +#if defined(HAVE_FUTIMESAT) + if (futimesat(tree_current_dir_fd(t), rt->name, times) == 0) + return (0); +#endif +#ifdef HAVE_LUTIMES + if (lutimes(rt->name, times) != 0) +#else + if (AE_IFLNK != rt->filetype && utimes(rt->name, times) != 0) +#endif + return (-1); +#endif + return (0); +} + +static int +open_on_current_dir(struct tree *t, const char *path, int flags) +{ +#ifdef HAVE_OPENAT + return (openat(tree_current_dir_fd(t), path, flags)); +#else + if (tree_enter_working_dir(t) != 0) + return (-1); + return (open(path, flags)); +#endif +} + +static int +tree_dup(int fd) +{ + int new_fd; +#ifdef F_DUPFD_CLOEXEC + static volatile int can_dupfd_cloexec = 1; + + if (can_dupfd_cloexec) { + new_fd = fcntl(fd, F_DUPFD_CLOEXEC, 0); + if (new_fd != -1) + return (new_fd); + /* Linux 2.6.18 - 2.6.23 declare F_DUPFD_CLOEXEC, + * but it cannot be used. So we have to try dup(). */ + /* We won't try F_DUPFD_CLOEXEC. */ + can_dupfd_cloexec = 0; + } +#endif /* F_DUPFD_CLOEXEC */ + new_fd = dup(fd); + __archive_ensure_cloexec_flag(new_fd); + return (new_fd); +} + +/* + * Add a directory path to the current stack. + */ +static void +tree_push(struct tree *t, const char *path, int filesystem_id, + int64_t dev, int64_t ino, struct restore_time *rt) +{ + struct tree_entry *te; + + te = calloc(1, sizeof(*te)); + te->next = t->stack; + te->parent = t->current; + if (te->parent) + te->depth = te->parent->depth + 1; + t->stack = te; + archive_string_init(&te->name); + te->symlink_parent_fd = -1; + archive_strcpy(&te->name, path); + te->flags = needsDescent | needsOpen | needsAscent; + te->filesystem_id = filesystem_id; + te->dev = dev; + te->ino = ino; + te->dirname_length = t->dirname_length; + te->restore_time.name = te->name.s; + if (rt != NULL) { + te->restore_time.mtime = rt->mtime; + te->restore_time.mtime_nsec = rt->mtime_nsec; + te->restore_time.atime = rt->atime; + te->restore_time.atime_nsec = rt->atime_nsec; + te->restore_time.filetype = rt->filetype; + te->restore_time.noatime = rt->noatime; + } +} + +/* + * Append a name to the current dir path. + */ +static void +tree_append(struct tree *t, const char *name, size_t name_length) +{ + size_t size_needed; + + t->path.s[t->dirname_length] = '\0'; + t->path.length = t->dirname_length; + /* Strip trailing '/' from name, unless entire name is "/". */ + while (name_length > 1 && name[name_length - 1] == '/') + name_length--; + + /* Resize pathname buffer as needed. */ + size_needed = name_length + t->dirname_length + 2; + archive_string_ensure(&t->path, size_needed); + /* Add a separating '/' if it's needed. */ + if (t->dirname_length > 0 && t->path.s[archive_strlen(&t->path)-1] != '/') + archive_strappend_char(&t->path, '/'); + t->basename = t->path.s + archive_strlen(&t->path); + archive_strncat(&t->path, name, name_length); + t->restore_time.name = t->basename; +} + +/* + * Open a directory tree for traversal. + */ +static struct tree * +tree_open(const char *path, int symlink_mode, int restore_time) +{ + struct tree *t; + + if ((t = calloc(1, sizeof(*t))) == NULL) + return (NULL); + archive_string_init(&t->path); + archive_string_ensure(&t->path, 31); + t->initial_symlink_mode = symlink_mode; + return (tree_reopen(t, path, restore_time)); +} + +static struct tree * +tree_reopen(struct tree *t, const char *path, int restore_time) +{ +#if defined(O_PATH) + /* Linux */ + const int o_flag = O_PATH; +#elif defined(O_SEARCH) + /* SunOS */ + const int o_flag = O_SEARCH; +#elif defined(__FreeBSD__) && defined(O_EXEC) + /* FreeBSD */ + const int o_flag = O_EXEC; +#endif + + t->flags = (restore_time != 0)?needsRestoreTimes:0; + t->flags |= onInitialDir; + t->visit_type = 0; + t->tree_errno = 0; + t->dirname_length = 0; + t->depth = 0; + t->descend = 0; + t->current = NULL; + t->d = INVALID_DIR_HANDLE; + t->symlink_mode = t->initial_symlink_mode; + archive_string_empty(&t->path); + t->entry_fd = -1; + t->entry_eof = 0; + t->entry_remaining_bytes = 0; + t->initial_filesystem_id = -1; + + /* First item is set up a lot like a symlink traversal. */ + tree_push(t, path, 0, 0, 0, NULL); + t->stack->flags = needsFirstVisit; + t->maxOpenCount = t->openCount = 1; + t->initial_dir_fd = open(".", O_RDONLY | O_CLOEXEC); +#if defined(O_PATH) || defined(O_SEARCH) || \ + (defined(__FreeBSD__) && defined(O_EXEC)) + /* + * Most likely reason to fail opening "." is that it's not readable, + * so try again for execute. The consequences of not opening this are + * unhelpful and unnecessary errors later. + */ + if (t->initial_dir_fd < 0) + t->initial_dir_fd = open(".", o_flag | O_CLOEXEC); +#endif + __archive_ensure_cloexec_flag(t->initial_dir_fd); + t->working_dir_fd = tree_dup(t->initial_dir_fd); + return (t); +} + +static int +tree_descent(struct tree *t) +{ + int flag, new_fd, r = 0; + + t->dirname_length = archive_strlen(&t->path); + flag = O_RDONLY | O_CLOEXEC; +#if defined(O_DIRECTORY) + flag |= O_DIRECTORY; +#endif + new_fd = open_on_current_dir(t, t->stack->name.s, flag); + __archive_ensure_cloexec_flag(new_fd); + if (new_fd < 0) { + t->tree_errno = errno; + r = TREE_ERROR_DIR; + } else { + t->depth++; + /* If it is a link, set up fd for the ascent. */ + if (t->stack->flags & isDirLink) { + t->stack->symlink_parent_fd = t->working_dir_fd; + t->openCount++; + if (t->openCount > t->maxOpenCount) + t->maxOpenCount = t->openCount; + } else + close(t->working_dir_fd); + /* Renew the current working directory. */ + t->working_dir_fd = new_fd; + t->flags &= ~onWorkingDir; + } + return (r); +} + +/* + * We've finished a directory; ascend back to the parent. + */ +static int +tree_ascend(struct tree *t) +{ + struct tree_entry *te; + int new_fd, r = 0, prev_dir_fd; + + te = t->stack; + prev_dir_fd = t->working_dir_fd; + if (te->flags & isDirLink) + new_fd = te->symlink_parent_fd; + else { + new_fd = open_on_current_dir(t, "..", O_RDONLY | O_CLOEXEC); + __archive_ensure_cloexec_flag(new_fd); + } + if (new_fd < 0) { + t->tree_errno = errno; + r = TREE_ERROR_FATAL; + } else { + /* Renew the current working directory. */ + t->working_dir_fd = new_fd; + t->flags &= ~onWorkingDir; + /* Current directory has been changed, we should + * close an fd of previous working directory. */ + close_and_restore_time(prev_dir_fd, t, &te->restore_time); + if (te->flags & isDirLink) { + t->openCount--; + te->symlink_parent_fd = -1; + } + t->depth--; + } + return (r); +} + +/* + * Return to the initial directory where tree_open() was performed. + */ +static int +tree_enter_initial_dir(struct tree *t) +{ + int r = 0; + + if ((t->flags & onInitialDir) == 0) { + r = fchdir(t->initial_dir_fd); + if (r == 0) { + t->flags &= ~onWorkingDir; + t->flags |= onInitialDir; + } + } + return (r); +} + +/* + * Restore working directory of directory traversals. + */ +static int +tree_enter_working_dir(struct tree *t) +{ + int r = 0; + + /* + * Change the current directory if really needed. + * Sometimes this is unneeded when we did not do + * descent. + */ + if (t->depth > 0 && (t->flags & onWorkingDir) == 0) { + r = fchdir(t->working_dir_fd); + if (r == 0) { + t->flags &= ~onInitialDir; + t->flags |= onWorkingDir; + } + } + return (r); +} + +static int +tree_current_dir_fd(struct tree *t) +{ + return (t->working_dir_fd); +} + +/* + * Pop the working stack. + */ +static void +tree_pop(struct tree *t) +{ + struct tree_entry *te; + + t->path.s[t->dirname_length] = '\0'; + t->path.length = t->dirname_length; + if (t->stack == t->current && t->current != NULL) + t->current = t->current->parent; + te = t->stack; + t->stack = te->next; + t->dirname_length = te->dirname_length; + t->basename = t->path.s + t->dirname_length; + while (t->basename[0] == '/') + t->basename++; + archive_string_free(&te->name); + free(te); +} + +/* + * Get the next item in the tree traversal. + */ +static int +tree_next(struct tree *t) +{ + int r; + + while (t->stack != NULL) { + /* If there's an open dir, get the next entry from there. */ + if (t->d != INVALID_DIR_HANDLE) { + r = tree_dir_next_posix(t); + if (r == 0) + continue; + return (r); + } + + if (t->stack->flags & needsFirstVisit) { + /* Top stack item needs a regular visit. */ + t->current = t->stack; + tree_append(t, t->stack->name.s, + archive_strlen(&(t->stack->name))); + /* t->dirname_length = t->path_length; */ + /* tree_pop(t); */ + t->stack->flags &= ~needsFirstVisit; + return (t->visit_type = TREE_REGULAR); + } else if (t->stack->flags & needsDescent) { + /* Top stack item is dir to descend into. */ + t->current = t->stack; + tree_append(t, t->stack->name.s, + archive_strlen(&(t->stack->name))); + t->stack->flags &= ~needsDescent; + r = tree_descent(t); + if (r != 0) { + tree_pop(t); + t->visit_type = r; + } else + t->visit_type = TREE_POSTDESCENT; + return (t->visit_type); + } else if (t->stack->flags & needsOpen) { + t->stack->flags &= ~needsOpen; + r = tree_dir_next_posix(t); + if (r == 0) + continue; + return (r); + } else if (t->stack->flags & needsAscent) { + /* Top stack item is dir and we're done with it. */ + r = tree_ascend(t); + tree_pop(t); + t->visit_type = r != 0 ? r : TREE_POSTASCENT; + return (t->visit_type); + } else { + /* Top item on stack is dead. */ + tree_pop(t); + t->flags &= ~hasLstat; + t->flags &= ~hasStat; + } + } + return (t->visit_type = 0); +} + +static int +tree_dir_next_posix(struct tree *t) +{ + int r; + const char *name; + size_t namelen; + + if (t->d == NULL) { +#if defined(USE_READDIR_R) + size_t dirent_size; +#endif + +#if defined(HAVE_FDOPENDIR) + t->d = fdopendir(tree_dup(t->working_dir_fd)); +#else /* HAVE_FDOPENDIR */ + if (tree_enter_working_dir(t) == 0) { + t->d = opendir("."); +#if HAVE_DIRFD || defined(dirfd) + __archive_ensure_cloexec_flag(dirfd(t->d)); +#endif + } +#endif /* HAVE_FDOPENDIR */ + if (t->d == NULL) { + r = tree_ascend(t); /* Undo "chdir" */ + tree_pop(t); + t->tree_errno = errno; + t->visit_type = r != 0 ? r : TREE_ERROR_DIR; + return (t->visit_type); + } +#if defined(USE_READDIR_R) + dirent_size = offsetof(struct dirent, d_name) + + t->filesystem_table[t->current->filesystem_id].name_max + 1; + if (t->dirent == NULL || t->dirent_allocated < dirent_size) { + free(t->dirent); + t->dirent = malloc(dirent_size); + if (t->dirent == NULL) { + closedir(t->d); + t->d = INVALID_DIR_HANDLE; + (void)tree_ascend(t); + tree_pop(t); + t->tree_errno = ENOMEM; + t->visit_type = TREE_ERROR_DIR; + return (t->visit_type); + } + t->dirent_allocated = dirent_size; + } +#endif /* USE_READDIR_R */ + } + for (;;) { + errno = 0; +#if defined(USE_READDIR_R) + r = readdir_r(t->d, t->dirent, &t->de); +#ifdef _AIX + /* Note: According to the man page, return value 9 indicates + * that the readdir_r was not successful and the error code + * is set to the global errno variable. And then if the end + * of directory entries was reached, the return value is 9 + * and the third parameter is set to NULL and errno is + * unchanged. */ + if (r == 9) + r = errno; +#endif /* _AIX */ + if (r != 0 || t->de == NULL) { +#else + t->de = readdir(t->d); + if (t->de == NULL) { + r = errno; +#endif + closedir(t->d); + t->d = INVALID_DIR_HANDLE; + if (r != 0) { + t->tree_errno = r; + t->visit_type = TREE_ERROR_DIR; + return (t->visit_type); + } else + return (0); + } + name = t->de->d_name; + namelen = D_NAMELEN(t->de); + t->flags &= ~hasLstat; + t->flags &= ~hasStat; + if (name[0] == '.' && name[1] == '\0') + continue; + if (name[0] == '.' && name[1] == '.' && name[2] == '\0') + continue; + tree_append(t, name, namelen); + return (t->visit_type = TREE_REGULAR); + } +} + + +/* + * Get the stat() data for the entry just returned from tree_next(). + */ +static const struct stat * +tree_current_stat(struct tree *t) +{ + if (!(t->flags & hasStat)) { +#ifdef HAVE_FSTATAT + if (fstatat(tree_current_dir_fd(t), + tree_current_access_path(t), &t->st, 0) != 0) +#else + if (tree_enter_working_dir(t) != 0) + return NULL; + if (la_stat(tree_current_access_path(t), &t->st) != 0) +#endif + return NULL; + t->flags |= hasStat; + } + return (&t->st); +} + +/* + * Get the lstat() data for the entry just returned from tree_next(). + */ +static const struct stat * +tree_current_lstat(struct tree *t) +{ + if (!(t->flags & hasLstat)) { +#ifdef HAVE_FSTATAT + if (fstatat(tree_current_dir_fd(t), + tree_current_access_path(t), &t->lst, + AT_SYMLINK_NOFOLLOW) != 0) +#else + if (tree_enter_working_dir(t) != 0) + return NULL; + if (lstat(tree_current_access_path(t), &t->lst) != 0) +#endif + return NULL; + t->flags |= hasLstat; + } + return (&t->lst); +} + +/* + * Test whether current entry is a dir or link to a dir. + */ +static int +tree_current_is_dir(struct tree *t) +{ + const struct stat *st; + /* + * If we already have lstat() info, then try some + * cheap tests to determine if this is a dir. + */ + if (t->flags & hasLstat) { + /* If lstat() says it's a dir, it must be a dir. */ + st = tree_current_lstat(t); + if (st == NULL) + return 0; + if (S_ISDIR(st->st_mode)) + return 1; + /* Not a dir; might be a link to a dir. */ + /* If it's not a link, then it's not a link to a dir. */ + if (!S_ISLNK(st->st_mode)) + return 0; + /* + * It's a link, but we don't know what it's a link to, + * so we'll have to use stat(). + */ + } + + st = tree_current_stat(t); + /* If we can't stat it, it's not a dir. */ + if (st == NULL) + return 0; + /* Use the definitive test. Hopefully this is cached. */ + return (S_ISDIR(st->st_mode)); +} + +/* + * Test whether current entry is a physical directory. Usually, we + * already have at least one of stat() or lstat() in memory, so we + * use tricks to try to avoid an extra trip to the disk. + */ +static int +tree_current_is_physical_dir(struct tree *t) +{ + const struct stat *st; + + /* + * If stat() says it isn't a dir, then it's not a dir. + * If stat() data is cached, this check is free, so do it first. + */ + if (t->flags & hasStat) { + st = tree_current_stat(t); + if (st == NULL) + return (0); + if (!S_ISDIR(st->st_mode)) + return (0); + } + + /* + * Either stat() said it was a dir (in which case, we have + * to determine whether it's really a link to a dir) or + * stat() info wasn't available. So we use lstat(), which + * hopefully is already cached. + */ + + st = tree_current_lstat(t); + /* If we can't stat it, it's not a dir. */ + if (st == NULL) + return 0; + /* Use the definitive test. Hopefully this is cached. */ + return (S_ISDIR(st->st_mode)); +} + +/* + * Test whether the same file has been in the tree as its parent. + */ +static int +tree_target_is_same_as_parent(struct tree *t, const struct stat *st) +{ + struct tree_entry *te; + + for (te = t->current->parent; te != NULL; te = te->parent) { + if (te->dev == (int64_t)st->st_dev && + te->ino == (int64_t)st->st_ino) + return (1); + } + return (0); +} + +/* + * Test whether the current file is symbolic link target and + * on the other filesystem. + */ +static int +tree_current_is_symblic_link_target(struct tree *t) +{ + static const struct stat *lst, *st; + + lst = tree_current_lstat(t); + st = tree_current_stat(t); + return (st != NULL && lst != NULL && + (int64_t)st->st_dev == t->current_filesystem->dev && + st->st_dev != lst->st_dev); +} + +/* + * Return the access path for the entry just returned from tree_next(). + */ +static const char * +tree_current_access_path(struct tree *t) +{ + return (t->basename); +} + +/* + * Return the full path for the entry just returned from tree_next(). + */ +static const char * +tree_current_path(struct tree *t) +{ + return (t->path.s); +} + +/* + * Terminate the traversal. + */ +static void +tree_close(struct tree *t) +{ + + if (t == NULL) + return; + if (t->entry_fd >= 0) { + close_and_restore_time(t->entry_fd, t, &t->restore_time); + t->entry_fd = -1; + } + /* Close the handle of readdir(). */ + if (t->d != INVALID_DIR_HANDLE) { + closedir(t->d); + t->d = INVALID_DIR_HANDLE; + } + /* Release anything remaining in the stack. */ + while (t->stack != NULL) { + if (t->stack->flags & isDirLink) + close(t->stack->symlink_parent_fd); + tree_pop(t); + } + if (t->working_dir_fd >= 0) { + close(t->working_dir_fd); + t->working_dir_fd = -1; + } + if (t->initial_dir_fd >= 0) { + close(t->initial_dir_fd); + t->initial_dir_fd = -1; + } +} + +/* + * Release any resources. + */ +static void +tree_free(struct tree *t) +{ + int i; + + if (t == NULL) + return; + archive_string_free(&t->path); +#if defined(USE_READDIR_R) + free(t->dirent); +#endif + free(t->sparse_list); + for (i = 0; i < t->max_filesystem_id; i++) + free(t->filesystem_table[i].allocation_ptr); + free(t->filesystem_table); + free(t); +} + +#endif diff --git a/c/archive_read_disk_private.h b/c/archive_read_disk_private.h new file mode 100644 --- /dev/null +++ b/c/archive_read_disk_private.h @@ -0,0 +1,98 @@ +/*- + * Copyright (c) 2003-2009 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer + * in this position and unchanged. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD: head/lib/libarchive/archive_read_disk_private.h 201105 2009-12-28 03:20:54Z kientzle $ + */ + +#ifndef ARCHIVE_READ_DISK_PRIVATE_H_INCLUDED +#define ARCHIVE_READ_DISK_PRIVATE_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif + +#include "archive_platform_acl.h" + +struct tree; +struct archive_entry; + +struct archive_read_disk { + struct archive archive; + + /* Reused by archive_read_next_header() */ + struct archive_entry *entry; + + /* + * Symlink mode is one of 'L'ogical, 'P'hysical, or 'H'ybrid, + * following an old BSD convention. 'L' follows all symlinks, + * 'P' follows none, 'H' follows symlinks only for the first + * item. + */ + char symlink_mode; + + /* + * Since symlink interaction changes, we need to track whether + * we're following symlinks for the current item. 'L' mode above + * sets this true, 'P' sets it false, 'H' changes it as we traverse. + */ + char follow_symlinks; /* Either 'L' or 'P'. */ + + /* Directory traversals. */ + struct tree *tree; + int (*open_on_current_dir)(struct tree*, const char *, int); + int (*tree_current_dir_fd)(struct tree*); + int (*tree_enter_working_dir)(struct tree*); + + /* Bitfield with ARCHIVE_READDISK_* tunables */ + int flags; + + const char * (*lookup_gname)(void *private, int64_t gid); + void (*cleanup_gname)(void *private); + void *lookup_gname_data; + const char * (*lookup_uname)(void *private, int64_t uid); + void (*cleanup_uname)(void *private); + void *lookup_uname_data; + + int (*metadata_filter_func)(struct archive *, void *, + struct archive_entry *); + void *metadata_filter_data; + + /* ARCHIVE_MATCH object. */ + struct archive *matching; + /* Callback function, this will be invoked when ARCHIVE_MATCH + * archive_match_*_excluded_ae return true. */ + void (*excluded_cb_func)(struct archive *, void *, + struct archive_entry *); + void *excluded_cb_data; +}; + +const char * +archive_read_disk_entry_setup_path(struct archive_read_disk *, + struct archive_entry *, int *); + +int +archive_read_disk_entry_setup_acls(struct archive_read_disk *, + struct archive_entry *, int *); +#endif diff --git a/c/archive_read_disk_set_standard_lookup.c b/c/archive_read_disk_set_standard_lookup.c new file mode 100644 --- /dev/null +++ b/c/archive_read_disk_set_standard_lookup.c @@ -0,0 +1,313 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_read_disk_set_standard_lookup.c 201109 2009-12-28 03:30:31Z kientzle $"); + +#ifdef HAVE_SYS_TYPES_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_GRP_H +#include +#endif +#ifdef HAVE_PWD_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" + +#if defined(_WIN32) && !defined(__CYGWIN__) +int +archive_read_disk_set_standard_lookup(struct archive *a) +{ + archive_set_error(a, -1, "Standard lookups not available on Windows"); + return (ARCHIVE_FATAL); +} +#else /* ! (_WIN32 && !__CYGWIN__) */ +#define name_cache_size 127 + +static const char * const NO_NAME = "(noname)"; + +struct name_cache { + struct archive *archive; + char *buff; + size_t buff_size; + int probes; + int hits; + size_t size; + struct { + id_t id; + const char *name; + } cache[name_cache_size]; +}; + +static const char * lookup_gname(void *, int64_t); +static const char * lookup_uname(void *, int64_t); +static void cleanup(void *); +static const char * lookup_gname_helper(struct name_cache *, id_t gid); +static const char * lookup_uname_helper(struct name_cache *, id_t uid); + +/* + * Installs functions that use getpwuid()/getgrgid()---along with + * a simple cache to accelerate such lookups---into the archive_read_disk + * object. This is in a separate file because getpwuid()/getgrgid() + * can pull in a LOT of library code (including NIS/LDAP functions, which + * pull in DNS resolvers, etc). This can easily top 500kB, which makes + * it inappropriate for some space-constrained applications. + * + * Applications that are size-sensitive may want to just use the + * real default functions (defined in archive_read_disk.c) that just + * use the uid/gid without the lookup. Or define your own custom functions + * if you prefer. + */ +int +archive_read_disk_set_standard_lookup(struct archive *a) +{ + struct name_cache *ucache = malloc(sizeof(struct name_cache)); + struct name_cache *gcache = malloc(sizeof(struct name_cache)); + + if (ucache == NULL || gcache == NULL) { + archive_set_error(a, ENOMEM, + "Can't allocate uname/gname lookup cache"); + free(ucache); + free(gcache); + return (ARCHIVE_FATAL); + } + + memset(ucache, 0, sizeof(*ucache)); + ucache->archive = a; + ucache->size = name_cache_size; + memset(gcache, 0, sizeof(*gcache)); + gcache->archive = a; + gcache->size = name_cache_size; + + archive_read_disk_set_gname_lookup(a, gcache, lookup_gname, cleanup); + archive_read_disk_set_uname_lookup(a, ucache, lookup_uname, cleanup); + + return (ARCHIVE_OK); +} + +static void +cleanup(void *data) +{ + struct name_cache *cache = (struct name_cache *)data; + size_t i; + + if (cache != NULL) { + for (i = 0; i < cache->size; i++) { + if (cache->cache[i].name != NULL && + cache->cache[i].name != NO_NAME) + free((void *)(uintptr_t)cache->cache[i].name); + } + free(cache->buff); + free(cache); + } +} + +/* + * Lookup uid/gid from uname/gname, return NULL if no match. + */ +static const char * +lookup_name(struct name_cache *cache, + const char * (*lookup_fn)(struct name_cache *, id_t), id_t id) +{ + const char *name; + int slot; + + + cache->probes++; + + slot = id % cache->size; + if (cache->cache[slot].name != NULL) { + if (cache->cache[slot].id == id) { + cache->hits++; + if (cache->cache[slot].name == NO_NAME) + return (NULL); + return (cache->cache[slot].name); + } + if (cache->cache[slot].name != NO_NAME) + free((void *)(uintptr_t)cache->cache[slot].name); + cache->cache[slot].name = NULL; + } + + name = (lookup_fn)(cache, id); + if (name == NULL) { + /* Cache and return the negative response. */ + cache->cache[slot].name = NO_NAME; + cache->cache[slot].id = id; + return (NULL); + } + + cache->cache[slot].name = name; + cache->cache[slot].id = id; + return (cache->cache[slot].name); +} + +static const char * +lookup_uname(void *data, int64_t uid) +{ + struct name_cache *uname_cache = (struct name_cache *)data; + return (lookup_name(uname_cache, + &lookup_uname_helper, (id_t)uid)); +} + +#if HAVE_GETPWUID_R +static const char * +lookup_uname_helper(struct name_cache *cache, id_t id) +{ + struct passwd pwent, *result; + char * nbuff; + size_t nbuff_size; + int r; + + if (cache->buff_size == 0) { + cache->buff_size = 256; + cache->buff = malloc(cache->buff_size); + } + if (cache->buff == NULL) + return (NULL); + for (;;) { + result = &pwent; /* Old getpwuid_r ignores last arg. */ + r = getpwuid_r((uid_t)id, &pwent, + cache->buff, cache->buff_size, &result); + if (r == 0) + break; + if (r != ERANGE) + break; + /* ERANGE means our buffer was too small, but POSIX + * doesn't tell us how big the buffer should be, so + * we just double it and try again. Because the buffer + * is kept around in the cache object, we shouldn't + * have to do this very often. */ + nbuff_size = cache->buff_size * 2; + nbuff = realloc(cache->buff, nbuff_size); + if (nbuff == NULL) + break; + cache->buff = nbuff; + cache->buff_size = nbuff_size; + } + if (r != 0) { + archive_set_error(cache->archive, errno, + "Can't lookup user for id %d", (int)id); + return (NULL); + } + if (result == NULL) + return (NULL); + + return strdup(result->pw_name); +} +#else +static const char * +lookup_uname_helper(struct name_cache *cache, id_t id) +{ + struct passwd *result; + (void)cache; /* UNUSED */ + + result = getpwuid((uid_t)id); + + if (result == NULL) + return (NULL); + + return strdup(result->pw_name); +} +#endif + +static const char * +lookup_gname(void *data, int64_t gid) +{ + struct name_cache *gname_cache = (struct name_cache *)data; + return (lookup_name(gname_cache, + &lookup_gname_helper, (id_t)gid)); +} + +#if HAVE_GETGRGID_R +static const char * +lookup_gname_helper(struct name_cache *cache, id_t id) +{ + struct group grent, *result; + char * nbuff; + size_t nbuff_size; + int r; + + if (cache->buff_size == 0) { + cache->buff_size = 256; + cache->buff = malloc(cache->buff_size); + } + if (cache->buff == NULL) + return (NULL); + for (;;) { + result = &grent; /* Old getgrgid_r ignores last arg. */ + r = getgrgid_r((gid_t)id, &grent, + cache->buff, cache->buff_size, &result); + if (r == 0) + break; + if (r != ERANGE) + break; + /* ERANGE means our buffer was too small, but POSIX + * doesn't tell us how big the buffer should be, so + * we just double it and try again. */ + nbuff_size = cache->buff_size * 2; + nbuff = realloc(cache->buff, nbuff_size); + if (nbuff == NULL) + break; + cache->buff = nbuff; + cache->buff_size = nbuff_size; + } + if (r != 0) { + archive_set_error(cache->archive, errno, + "Can't lookup group for id %d", (int)id); + return (NULL); + } + if (result == NULL) + return (NULL); + + return strdup(result->gr_name); +} +#else +static const char * +lookup_gname_helper(struct name_cache *cache, id_t id) +{ + struct group *result; + (void)cache; /* UNUSED */ + + result = getgrgid((gid_t)id); + + if (result == NULL) + return (NULL); + + return strdup(result->gr_name); +} +#endif + +#endif /* ! (_WIN32 && !__CYGWIN__) */ diff --git a/c/archive_read_disk_windows.c b/c/archive_read_disk_windows.c new file mode 100644 --- /dev/null +++ b/c/archive_read_disk_windows.c @@ -0,0 +1,2477 @@ +/*- + * Copyright (c) 2003-2009 Tim Kientzle + * Copyright (c) 2010-2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer + * in this position and unchanged. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#if defined(_WIN32) && !defined(__CYGWIN__) + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#include + +#include "archive.h" +#include "archive_string.h" +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_read_disk_private.h" + +#ifndef O_BINARY +#define O_BINARY 0 +#endif +#ifndef IO_REPARSE_TAG_SYMLINK +/* Old SDKs do not provide IO_REPARSE_TAG_SYMLINK */ +#define IO_REPARSE_TAG_SYMLINK 0xA000000CL +#endif + +/*- + * This is a new directory-walking system that addresses a number + * of problems I've had with fts(3). In particular, it has no + * pathname-length limits (other than the size of 'int'), handles + * deep logical traversals, uses considerably less memory, and has + * an opaque interface (easier to modify in the future). + * + * Internally, it keeps a single list of "tree_entry" items that + * represent filesystem objects that require further attention. + * Non-directories are not kept in memory: they are pulled from + * readdir(), returned to the client, then freed as soon as possible. + * Any directory entry to be traversed gets pushed onto the stack. + * + * There is surprisingly little information that needs to be kept for + * each item on the stack. Just the name, depth (represented here as the + * string length of the parent directory's pathname), and some markers + * indicating how to get back to the parent (via chdir("..") for a + * regular dir or via fchdir(2) for a symlink). + */ + +struct restore_time { + const wchar_t *full_path; + FILETIME lastWriteTime; + FILETIME lastAccessTime; + mode_t filetype; +}; + +struct tree_entry { + int depth; + struct tree_entry *next; + struct tree_entry *parent; + size_t full_path_dir_length; + struct archive_wstring name; + struct archive_wstring full_path; + size_t dirname_length; + int64_t dev; + int64_t ino; + int flags; + int filesystem_id; + /* How to restore time of a directory. */ + struct restore_time restore_time; +}; + +struct filesystem { + int64_t dev; + int synthetic; + int remote; + DWORD bytesPerSector; +}; + +/* Definitions for tree_entry.flags bitmap. */ +#define isDir 1 /* This entry is a regular directory. */ +#define isDirLink 2 /* This entry is a symbolic link to a directory. */ +#define needsFirstVisit 4 /* This is an initial entry. */ +#define needsDescent 8 /* This entry needs to be previsited. */ +#define needsOpen 16 /* This is a directory that needs to be opened. */ +#define needsAscent 32 /* This entry needs to be postvisited. */ + +/* + * On Windows, "first visit" is handled as a pattern to be handed to + * _findfirst(). This is consistent with Windows conventions that + * file patterns are handled within the application. On Posix, + * "first visit" is just returned to the client. + */ + +#define MAX_OVERLAPPED 8 +#define READ_BUFFER_SIZE (1024 * 64) /* Default to 64KB per https://technet.microsoft.com/en-us/library/cc938632.aspx */ +#define DIRECT_IO 0/* Disabled */ +#define ASYNC_IO 1/* Enabled */ + +/* + * Local data for this package. + */ +struct tree { + struct tree_entry *stack; + struct tree_entry *current; + HANDLE d; + WIN32_FIND_DATAW _findData; + WIN32_FIND_DATAW *findData; + int flags; + int visit_type; + /* Error code from last failed operation. */ + int tree_errno; + + /* A full path with "\\?\" prefix. */ + struct archive_wstring full_path; + size_t full_path_dir_length; + /* Dynamically-sized buffer for holding path */ + struct archive_wstring path; + + /* Last path element */ + const wchar_t *basename; + /* Leading dir length */ + size_t dirname_length; + + int depth; + + BY_HANDLE_FILE_INFORMATION lst; + BY_HANDLE_FILE_INFORMATION st; + int descend; + /* How to restore time of a file. */ + struct restore_time restore_time; + + struct entry_sparse { + int64_t length; + int64_t offset; + } *sparse_list, *current_sparse; + int sparse_count; + int sparse_list_size; + + char initial_symlink_mode; + char symlink_mode; + struct filesystem *current_filesystem; + struct filesystem *filesystem_table; + int initial_filesystem_id; + int current_filesystem_id; + int max_filesystem_id; + int allocated_filesystem; + + HANDLE entry_fh; + int entry_eof; + int64_t entry_remaining_bytes; + int64_t entry_total; + + int ol_idx_doing; + int ol_idx_done; + int ol_num_doing; + int ol_num_done; + int64_t ol_remaining_bytes; + int64_t ol_total; + struct la_overlapped { + OVERLAPPED ol; + struct archive * _a; + unsigned char *buff; + size_t buff_size; + int64_t offset; + size_t bytes_expected; + size_t bytes_transferred; + } ol[MAX_OVERLAPPED]; + int direct_io; + int async_io; +}; + +#define bhfi_dev(bhfi) ((bhfi)->dwVolumeSerialNumber) +/* Treat FileIndex as i-node. We should remove a sequence number + * which is high-16-bits of nFileIndexHigh. */ +#define bhfi_ino(bhfi) \ + ((((int64_t)((bhfi)->nFileIndexHigh & 0x0000FFFFUL)) << 32) \ + + (bhfi)->nFileIndexLow) + +/* Definitions for tree.flags bitmap. */ +#define hasStat 16 /* The st entry is valid. */ +#define hasLstat 32 /* The lst entry is valid. */ +#define needsRestoreTimes 128 + +static int +tree_dir_next_windows(struct tree *t, const wchar_t *pattern); + +/* Initiate/terminate a tree traversal. */ +static struct tree *tree_open(const wchar_t *, int, int); +static struct tree *tree_reopen(struct tree *, const wchar_t *, int); +static void tree_close(struct tree *); +static void tree_free(struct tree *); +static void tree_push(struct tree *, const wchar_t *, const wchar_t *, + int, int64_t, int64_t, struct restore_time *); + +/* + * tree_next() returns Zero if there is no next entry, non-zero if + * there is. Note that directories are visited three times. + * Directories are always visited first as part of enumerating their + * parent; that is a "regular" visit. If tree_descend() is invoked at + * that time, the directory is added to a work list and will + * subsequently be visited two more times: once just after descending + * into the directory ("postdescent") and again just after ascending + * back to the parent ("postascent"). + * + * TREE_ERROR_DIR is returned if the descent failed (because the + * directory couldn't be opened, for instance). This is returned + * instead of TREE_POSTDESCENT/TREE_POSTASCENT. TREE_ERROR_DIR is not a + * fatal error, but it does imply that the relevant subtree won't be + * visited. TREE_ERROR_FATAL is returned for an error that left the + * traversal completely hosed. Right now, this is only returned for + * chdir() failures during ascent. + */ +#define TREE_REGULAR 1 +#define TREE_POSTDESCENT 2 +#define TREE_POSTASCENT 3 +#define TREE_ERROR_DIR -1 +#define TREE_ERROR_FATAL -2 + +static int tree_next(struct tree *); + +/* + * Return information about the current entry. + */ + +/* + * The current full pathname, length of the full pathname, and a name + * that can be used to access the file. Because tree does use chdir + * extensively, the access path is almost never the same as the full + * current path. + * + */ +static const wchar_t *tree_current_path(struct tree *); +static const wchar_t *tree_current_access_path(struct tree *); + +/* + * Request the lstat() or stat() data for the current path. Since the + * tree package needs to do some of this anyway, and caches the + * results, you should take advantage of it here if you need it rather + * than make a redundant stat() or lstat() call of your own. + */ +static const BY_HANDLE_FILE_INFORMATION *tree_current_stat(struct tree *); +static const BY_HANDLE_FILE_INFORMATION *tree_current_lstat(struct tree *); + +/* The following functions use tricks to avoid a certain number of + * stat()/lstat() calls. */ +/* "is_physical_dir" is equivalent to S_ISDIR(tree_current_lstat()->st_mode) */ +static int tree_current_is_physical_dir(struct tree *); +/* "is_physical_link" is equivalent to S_ISLNK(tree_current_lstat()->st_mode) */ +static int tree_current_is_physical_link(struct tree *); +/* Instead of archive_entry_copy_stat for BY_HANDLE_FILE_INFORMATION */ +static void tree_archive_entry_copy_bhfi(struct archive_entry *, + struct tree *, const BY_HANDLE_FILE_INFORMATION *); +/* "is_dir" is equivalent to S_ISDIR(tree_current_stat()->st_mode) */ +static int tree_current_is_dir(struct tree *); +static int update_current_filesystem(struct archive_read_disk *a, + int64_t dev); +static int setup_current_filesystem(struct archive_read_disk *); +static int tree_target_is_same_as_parent(struct tree *, + const BY_HANDLE_FILE_INFORMATION *); + +static int _archive_read_disk_open_w(struct archive *, const wchar_t *); +static int _archive_read_free(struct archive *); +static int _archive_read_close(struct archive *); +static int _archive_read_data_block(struct archive *, + const void **, size_t *, int64_t *); +static int _archive_read_next_header(struct archive *, + struct archive_entry **); +static int _archive_read_next_header2(struct archive *, + struct archive_entry *); +static const char *trivial_lookup_gname(void *, int64_t gid); +static const char *trivial_lookup_uname(void *, int64_t uid); +static int setup_sparse(struct archive_read_disk *, struct archive_entry *); +static int close_and_restore_time(HANDLE, struct tree *, + struct restore_time *); +static int setup_sparse_from_disk(struct archive_read_disk *, + struct archive_entry *, HANDLE); +static int la_linkname_from_handle(HANDLE, wchar_t **, int *); +static int la_linkname_from_pathw(const wchar_t *, wchar_t **, int *); +static void entry_symlink_from_pathw(struct archive_entry *, + const wchar_t *path); + +typedef struct _REPARSE_DATA_BUFFER { + ULONG ReparseTag; + USHORT ReparseDataLength; + USHORT Reserved; + union { + struct { + USHORT SubstituteNameOffset; + USHORT SubstituteNameLength; + USHORT PrintNameOffset; + USHORT PrintNameLength; + ULONG Flags; + WCHAR PathBuffer[1]; + } SymbolicLinkReparseBuffer; + struct { + USHORT SubstituteNameOffset; + USHORT SubstituteNameLength; + USHORT PrintNameOffset; + USHORT PrintNameLength; + WCHAR PathBuffer[1]; + } MountPointReparseBuffer; + struct { + UCHAR DataBuffer[1]; + } GenericReparseBuffer; + } DUMMYUNIONNAME; +} REPARSE_DATA_BUFFER, *PREPARSE_DATA_BUFFER; + +/* + * Reads the target of a symbolic link + * + * Returns 0 on success and -1 on failure + * outbuf is allocated in the function + */ +static int +la_linkname_from_handle(HANDLE h, wchar_t **linkname, int *linktype) +{ + DWORD inbytes; + REPARSE_DATA_BUFFER *buf; + BY_HANDLE_FILE_INFORMATION st; + size_t len; + BOOL ret; + BYTE *indata; + wchar_t *tbuf; + + ret = GetFileInformationByHandle(h, &st); + if (ret == 0 || + (st.dwFileAttributes & FILE_ATTRIBUTE_REPARSE_POINT) == 0) { + return (-1); + } + + indata = malloc(MAXIMUM_REPARSE_DATA_BUFFER_SIZE); + ret = DeviceIoControl(h, FSCTL_GET_REPARSE_POINT, NULL, 0, indata, + 1024, &inbytes, NULL); + if (ret == 0) { + la_dosmaperr(GetLastError()); + free(indata); + return (-1); + } + + buf = (REPARSE_DATA_BUFFER *) indata; + if (buf->ReparseTag != IO_REPARSE_TAG_SYMLINK) { + free(indata); + /* File is not a symbolic link */ + errno = EINVAL; + return (-1); + } + + len = buf->SymbolicLinkReparseBuffer.SubstituteNameLength; + if (len <= 0) { + free(indata); + return (-1); + } + + tbuf = malloc(len + 1 * sizeof(wchar_t)); + if (tbuf == NULL) { + free(indata); + return (-1); + } + + memcpy(tbuf, &((BYTE *)buf->SymbolicLinkReparseBuffer.PathBuffer) + [buf->SymbolicLinkReparseBuffer.SubstituteNameOffset], len); + free(indata); + + tbuf[len / sizeof(wchar_t)] = L'\0'; + + *linkname = tbuf; + + /* + * Translate backslashes to slashes for libarchive internal use + */ + while(*tbuf != L'\0') { + if (*tbuf == L'\\') + *tbuf = L'/'; + tbuf++; + } + + if ((st.dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY) == 0) + *linktype = AE_SYMLINK_TYPE_FILE; + else + *linktype = AE_SYMLINK_TYPE_DIRECTORY; + + return (0); +} + +/* + * Returns AE_SYMLINK_TYPE_FILE, AE_SYMLINK_TYPE_DIRECTORY or -1 on error + */ +static int +la_linkname_from_pathw(const wchar_t *path, wchar_t **outbuf, int *linktype) +{ + HANDLE h; + const DWORD flag = FILE_FLAG_BACKUP_SEMANTICS | + FILE_FLAG_OPEN_REPARSE_POINT; + int ret; + + h = CreateFileW(path, 0, FILE_SHARE_READ, NULL, OPEN_EXISTING, flag, + NULL); + if (h == INVALID_HANDLE_VALUE) { + la_dosmaperr(GetLastError()); + return (-1); + } + + ret = la_linkname_from_handle(h, outbuf, linktype); + CloseHandle(h); + + return (ret); +} + +static void +entry_symlink_from_pathw(struct archive_entry *entry, const wchar_t *path) +{ + wchar_t *linkname = NULL; + int ret, linktype; + + ret = la_linkname_from_pathw(path, &linkname, &linktype); + if (ret != 0) + return; + if (linktype >= 0) { + archive_entry_copy_symlink_w(entry, linkname); + archive_entry_set_symlink_type(entry, linktype); + } + free(linkname); + + return; +} + +static struct archive_vtable * +archive_read_disk_vtable(void) +{ + static struct archive_vtable av; + static int inited = 0; + + if (!inited) { + av.archive_free = _archive_read_free; + av.archive_close = _archive_read_close; + av.archive_read_data_block = _archive_read_data_block; + av.archive_read_next_header = _archive_read_next_header; + av.archive_read_next_header2 = _archive_read_next_header2; + inited = 1; + } + return (&av); +} + +const char * +archive_read_disk_gname(struct archive *_a, la_int64_t gid) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + if (ARCHIVE_OK != __archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_disk_gname")) + return (NULL); + if (a->lookup_gname == NULL) + return (NULL); + return ((*a->lookup_gname)(a->lookup_gname_data, gid)); +} + +const char * +archive_read_disk_uname(struct archive *_a, la_int64_t uid) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + if (ARCHIVE_OK != __archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_disk_uname")) + return (NULL); + if (a->lookup_uname == NULL) + return (NULL); + return ((*a->lookup_uname)(a->lookup_uname_data, uid)); +} + +int +archive_read_disk_set_gname_lookup(struct archive *_a, + void *private_data, + const char * (*lookup_gname)(void *private, la_int64_t gid), + void (*cleanup_gname)(void *private)) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + archive_check_magic(&a->archive, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_disk_set_gname_lookup"); + + if (a->cleanup_gname != NULL && a->lookup_gname_data != NULL) + (a->cleanup_gname)(a->lookup_gname_data); + + a->lookup_gname = lookup_gname; + a->cleanup_gname = cleanup_gname; + a->lookup_gname_data = private_data; + return (ARCHIVE_OK); +} + +int +archive_read_disk_set_uname_lookup(struct archive *_a, + void *private_data, + const char * (*lookup_uname)(void *private, int64_t uid), + void (*cleanup_uname)(void *private)) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + archive_check_magic(&a->archive, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_disk_set_uname_lookup"); + + if (a->cleanup_uname != NULL && a->lookup_uname_data != NULL) + (a->cleanup_uname)(a->lookup_uname_data); + + a->lookup_uname = lookup_uname; + a->cleanup_uname = cleanup_uname; + a->lookup_uname_data = private_data; + return (ARCHIVE_OK); +} + +/* + * Create a new archive_read_disk object and initialize it with global state. + */ +struct archive * +archive_read_disk_new(void) +{ + struct archive_read_disk *a; + + a = (struct archive_read_disk *)calloc(1, sizeof(*a)); + if (a == NULL) + return (NULL); + a->archive.magic = ARCHIVE_READ_DISK_MAGIC; + a->archive.state = ARCHIVE_STATE_NEW; + a->archive.vtable = archive_read_disk_vtable(); + a->entry = archive_entry_new2(&a->archive); + a->lookup_uname = trivial_lookup_uname; + a->lookup_gname = trivial_lookup_gname; + a->flags = ARCHIVE_READDISK_MAC_COPYFILE; + return (&a->archive); +} + +static int +_archive_read_free(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + int r; + + if (_a == NULL) + return (ARCHIVE_OK); + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_read_free"); + + if (a->archive.state != ARCHIVE_STATE_CLOSED) + r = _archive_read_close(&a->archive); + else + r = ARCHIVE_OK; + + tree_free(a->tree); + if (a->cleanup_gname != NULL && a->lookup_gname_data != NULL) + (a->cleanup_gname)(a->lookup_gname_data); + if (a->cleanup_uname != NULL && a->lookup_uname_data != NULL) + (a->cleanup_uname)(a->lookup_uname_data); + archive_string_free(&a->archive.error_string); + archive_entry_free(a->entry); + a->archive.magic = 0; + free(a); + return (r); +} + +static int +_archive_read_close(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_read_close"); + + if (a->archive.state != ARCHIVE_STATE_FATAL) + a->archive.state = ARCHIVE_STATE_CLOSED; + + tree_close(a->tree); + + return (ARCHIVE_OK); +} + +static void +setup_symlink_mode(struct archive_read_disk *a, char symlink_mode, + int follow_symlinks) +{ + a->symlink_mode = symlink_mode; + a->follow_symlinks = follow_symlinks; + if (a->tree != NULL) { + a->tree->initial_symlink_mode = a->symlink_mode; + a->tree->symlink_mode = a->symlink_mode; + } +} + +int +archive_read_disk_set_symlink_logical(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_disk_set_symlink_logical"); + setup_symlink_mode(a, 'L', 1); + return (ARCHIVE_OK); +} + +int +archive_read_disk_set_symlink_physical(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_disk_set_symlink_physical"); + setup_symlink_mode(a, 'P', 0); + return (ARCHIVE_OK); +} + +int +archive_read_disk_set_symlink_hybrid(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_disk_set_symlink_hybrid"); + setup_symlink_mode(a, 'H', 1);/* Follow symlinks initially. */ + return (ARCHIVE_OK); +} + +int +archive_read_disk_set_atime_restored(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_disk_restore_atime"); + a->flags |= ARCHIVE_READDISK_RESTORE_ATIME; + if (a->tree != NULL) + a->tree->flags |= needsRestoreTimes; + return (ARCHIVE_OK); +} + +int +archive_read_disk_set_behavior(struct archive *_a, int flags) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + int r = ARCHIVE_OK; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_disk_honor_nodump"); + + a->flags = flags; + + if (flags & ARCHIVE_READDISK_RESTORE_ATIME) + r = archive_read_disk_set_atime_restored(_a); + else { + if (a->tree != NULL) + a->tree->flags &= ~needsRestoreTimes; + } + return (r); +} + +/* + * Trivial implementations of gname/uname lookup functions. + * These are normally overridden by the client, but these stub + * versions ensure that we always have something that works. + */ +static const char * +trivial_lookup_gname(void *private_data, int64_t gid) +{ + (void)private_data; /* UNUSED */ + (void)gid; /* UNUSED */ + return (NULL); +} + +static const char * +trivial_lookup_uname(void *private_data, int64_t uid) +{ + (void)private_data; /* UNUSED */ + (void)uid; /* UNUSED */ + return (NULL); +} + +static int64_t +align_num_per_sector(struct tree *t, int64_t size) +{ + int64_t surplus; + + size += t->current_filesystem->bytesPerSector -1; + surplus = size % t->current_filesystem->bytesPerSector; + size -= surplus; + return (size); +} + +static int +start_next_async_read(struct archive_read_disk *a, struct tree *t) +{ + struct la_overlapped *olp; + DWORD buffbytes, rbytes; + + if (t->ol_remaining_bytes == 0) + return (ARCHIVE_EOF); + + olp = &(t->ol[t->ol_idx_doing]); + t->ol_idx_doing = (t->ol_idx_doing + 1) % MAX_OVERLAPPED; + + /* Allocate read buffer. */ + if (olp->buff == NULL) { + void *p; + size_t s = (size_t)align_num_per_sector(t, READ_BUFFER_SIZE); + p = VirtualAlloc(NULL, s, MEM_COMMIT, PAGE_READWRITE); + if (p == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Couldn't allocate memory"); + a->archive.state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + } + olp->buff = p; + olp->buff_size = s; + olp->_a = &a->archive; + olp->ol.hEvent = CreateEventW(NULL, TRUE, FALSE, NULL); + if (olp->ol.hEvent == NULL) { + la_dosmaperr(GetLastError()); + archive_set_error(&a->archive, errno, + "CreateEvent failed"); + a->archive.state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + } + } else + ResetEvent(olp->ol.hEvent); + + buffbytes = (DWORD)olp->buff_size; + if (buffbytes > t->current_sparse->length) + buffbytes = (DWORD)t->current_sparse->length; + + /* Skip hole. */ + if (t->current_sparse->offset > t->ol_total) { + t->ol_remaining_bytes -= + t->current_sparse->offset - t->ol_total; + } + + olp->offset = t->current_sparse->offset; + olp->ol.Offset = (DWORD)(olp->offset & 0xffffffff); + olp->ol.OffsetHigh = (DWORD)(olp->offset >> 32); + + if (t->ol_remaining_bytes > buffbytes) { + olp->bytes_expected = buffbytes; + t->ol_remaining_bytes -= buffbytes; + } else { + olp->bytes_expected = (size_t)t->ol_remaining_bytes; + t->ol_remaining_bytes = 0; + } + olp->bytes_transferred = 0; + t->current_sparse->offset += buffbytes; + t->current_sparse->length -= buffbytes; + t->ol_total = t->current_sparse->offset; + if (t->current_sparse->length == 0 && t->ol_remaining_bytes > 0) + t->current_sparse++; + + if (!ReadFile(t->entry_fh, olp->buff, buffbytes, &rbytes, &(olp->ol))) { + DWORD lasterr; + + lasterr = GetLastError(); + if (lasterr == ERROR_HANDLE_EOF) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Reading file truncated"); + a->archive.state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + } else if (lasterr != ERROR_IO_PENDING) { + if (lasterr == ERROR_NO_DATA) + errno = EAGAIN; + else if (lasterr == ERROR_ACCESS_DENIED) + errno = EBADF; + else + la_dosmaperr(lasterr); + archive_set_error(&a->archive, errno, "Read error"); + a->archive.state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + } + } else + olp->bytes_transferred = rbytes; + t->ol_num_doing++; + + return (t->ol_remaining_bytes == 0)? ARCHIVE_EOF: ARCHIVE_OK; +} + +static void +cancel_async(struct tree *t) +{ + if (t->ol_num_doing != t->ol_num_done) { + CancelIo(t->entry_fh); + t->ol_num_doing = t->ol_num_done = 0; + } +} + +static int +_archive_read_data_block(struct archive *_a, const void **buff, + size_t *size, int64_t *offset) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + struct tree *t = a->tree; + struct la_overlapped *olp; + DWORD bytes_transferred; + int r = ARCHIVE_FATAL; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_DATA, + "archive_read_data_block"); + + if (t->entry_eof || t->entry_remaining_bytes <= 0) { + r = ARCHIVE_EOF; + goto abort_read_data; + } + + /* + * Make a request to read the file in asynchronous. + */ + if (t->ol_num_doing == 0) { + do { + r = start_next_async_read(a, t); + if (r == ARCHIVE_FATAL) + goto abort_read_data; + if (!t->async_io) + break; + } while (r == ARCHIVE_OK && t->ol_num_doing < MAX_OVERLAPPED); + } else { + if ((r = start_next_async_read(a, t)) == ARCHIVE_FATAL) + goto abort_read_data; + } + + olp = &(t->ol[t->ol_idx_done]); + t->ol_idx_done = (t->ol_idx_done + 1) % MAX_OVERLAPPED; + if (olp->bytes_transferred) + bytes_transferred = (DWORD)olp->bytes_transferred; + else if (!GetOverlappedResult(t->entry_fh, &(olp->ol), + &bytes_transferred, TRUE)) { + la_dosmaperr(GetLastError()); + archive_set_error(&a->archive, errno, + "GetOverlappedResult failed"); + a->archive.state = ARCHIVE_STATE_FATAL; + r = ARCHIVE_FATAL; + goto abort_read_data; + } + t->ol_num_done++; + + if (bytes_transferred == 0 || + olp->bytes_expected != bytes_transferred) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Reading file truncated"); + a->archive.state = ARCHIVE_STATE_FATAL; + r = ARCHIVE_FATAL; + goto abort_read_data; + } + + *buff = olp->buff; + *size = bytes_transferred; + *offset = olp->offset; + if (olp->offset > t->entry_total) + t->entry_remaining_bytes -= olp->offset - t->entry_total; + t->entry_total = olp->offset + *size; + t->entry_remaining_bytes -= *size; + if (t->entry_remaining_bytes == 0) { + /* Close the current file descriptor */ + close_and_restore_time(t->entry_fh, t, &t->restore_time); + t->entry_fh = INVALID_HANDLE_VALUE; + t->entry_eof = 1; + } + return (ARCHIVE_OK); + +abort_read_data: + *buff = NULL; + *size = 0; + *offset = t->entry_total; + if (t->entry_fh != INVALID_HANDLE_VALUE) { + cancel_async(t); + /* Close the current file descriptor */ + close_and_restore_time(t->entry_fh, t, &t->restore_time); + t->entry_fh = INVALID_HANDLE_VALUE; + } + return (r); +} + +static int +next_entry(struct archive_read_disk *a, struct tree *t, + struct archive_entry *entry) +{ + const BY_HANDLE_FILE_INFORMATION *st; + const BY_HANDLE_FILE_INFORMATION *lst; + const char*name; + int descend, r; + + st = NULL; + lst = NULL; + t->descend = 0; + do { + switch (tree_next(t)) { + case TREE_ERROR_FATAL: + archive_set_error(&a->archive, t->tree_errno, + "%ls: Unable to continue traversing directory tree", + tree_current_path(t)); + a->archive.state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + case TREE_ERROR_DIR: + archive_set_error(&a->archive, t->tree_errno, + "%ls: Couldn't visit directory", + tree_current_path(t)); + return (ARCHIVE_FAILED); + case 0: + return (ARCHIVE_EOF); + case TREE_POSTDESCENT: + case TREE_POSTASCENT: + break; + case TREE_REGULAR: + lst = tree_current_lstat(t); + if (lst == NULL) { + archive_set_error(&a->archive, t->tree_errno, + "%ls: Cannot stat", + tree_current_path(t)); + return (ARCHIVE_FAILED); + } + break; + } + } while (lst == NULL); + + archive_entry_copy_pathname_w(entry, tree_current_path(t)); + + /* + * Perform path matching. + */ + if (a->matching) { + r = archive_match_path_excluded(a->matching, entry); + if (r < 0) { + archive_set_error(&(a->archive), errno, + "Failed : %s", archive_error_string(a->matching)); + return (r); + } + if (r) { + if (a->excluded_cb_func) + a->excluded_cb_func(&(a->archive), + a->excluded_cb_data, entry); + return (ARCHIVE_RETRY); + } + } + + /* + * Distinguish 'L'/'P'/'H' symlink following. + */ + switch(t->symlink_mode) { + case 'H': + /* 'H': After the first item, rest like 'P'. */ + t->symlink_mode = 'P'; + /* 'H': First item (from command line) like 'L'. */ + /* FALLTHROUGH */ + case 'L': + /* 'L': Do descend through a symlink to dir. */ + descend = tree_current_is_dir(t); + /* 'L': Follow symlinks to files. */ + a->symlink_mode = 'L'; + a->follow_symlinks = 1; + /* 'L': Archive symlinks as targets, if we can. */ + st = tree_current_stat(t); + if (st != NULL && !tree_target_is_same_as_parent(t, st)) + break; + /* If stat fails, we have a broken symlink; + * in that case, don't follow the link. */ + /* FALLTHROUGH */ + default: + /* 'P': Don't descend through a symlink to dir. */ + descend = tree_current_is_physical_dir(t); + /* 'P': Don't follow symlinks to files. */ + a->symlink_mode = 'P'; + a->follow_symlinks = 0; + /* 'P': Archive symlinks as symlinks. */ + st = lst; + break; + } + + if (update_current_filesystem(a, bhfi_dev(st)) != ARCHIVE_OK) { + a->archive.state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + } + if (t->initial_filesystem_id == -1) + t->initial_filesystem_id = t->current_filesystem_id; + if (a->flags & ARCHIVE_READDISK_NO_TRAVERSE_MOUNTS) { + if (t->initial_filesystem_id != t->current_filesystem_id) + return (ARCHIVE_RETRY); + } + t->descend = descend; + + tree_archive_entry_copy_bhfi(entry, t, st); + + /* Save the times to be restored. This must be in before + * calling archive_read_disk_descend() or any chance of it, + * especially, invoking a callback. */ + t->restore_time.lastWriteTime = st->ftLastWriteTime; + t->restore_time.lastAccessTime = st->ftLastAccessTime; + t->restore_time.filetype = archive_entry_filetype(entry); + + /* + * Perform time matching. + */ + if (a->matching) { + r = archive_match_time_excluded(a->matching, entry); + if (r < 0) { + archive_set_error(&(a->archive), errno, + "Failed : %s", archive_error_string(a->matching)); + return (r); + } + if (r) { + if (a->excluded_cb_func) + a->excluded_cb_func(&(a->archive), + a->excluded_cb_data, entry); + return (ARCHIVE_RETRY); + } + } + + /* Lookup uname/gname */ + name = archive_read_disk_uname(&(a->archive), archive_entry_uid(entry)); + if (name != NULL) + archive_entry_copy_uname(entry, name); + name = archive_read_disk_gname(&(a->archive), archive_entry_gid(entry)); + if (name != NULL) + archive_entry_copy_gname(entry, name); + + /* + * Perform owner matching. + */ + if (a->matching) { + r = archive_match_owner_excluded(a->matching, entry); + if (r < 0) { + archive_set_error(&(a->archive), errno, + "Failed : %s", archive_error_string(a->matching)); + return (r); + } + if (r) { + if (a->excluded_cb_func) + a->excluded_cb_func(&(a->archive), + a->excluded_cb_data, entry); + return (ARCHIVE_RETRY); + } + } + + /* + * File attributes + */ + if ((a->flags & ARCHIVE_READDISK_NO_FFLAGS) == 0) { + const int supported_attrs = + FILE_ATTRIBUTE_READONLY | + FILE_ATTRIBUTE_HIDDEN | + FILE_ATTRIBUTE_SYSTEM; + DWORD file_attrs = st->dwFileAttributes & supported_attrs; + if (file_attrs != 0) + archive_entry_set_fflags(entry, file_attrs, 0); + } + + /* + * Invoke a meta data filter callback. + */ + if (a->metadata_filter_func) { + if (!a->metadata_filter_func(&(a->archive), + a->metadata_filter_data, entry)) + return (ARCHIVE_RETRY); + } + + archive_entry_copy_sourcepath_w(entry, tree_current_access_path(t)); + + r = ARCHIVE_OK; + if (archive_entry_filetype(entry) == AE_IFREG && + archive_entry_size(entry) > 0) { + DWORD flags = FILE_FLAG_BACKUP_SEMANTICS; + if (t->async_io) + flags |= FILE_FLAG_OVERLAPPED; + if (t->direct_io) + flags |= FILE_FLAG_NO_BUFFERING; + else + flags |= FILE_FLAG_SEQUENTIAL_SCAN; + t->entry_fh = CreateFileW(tree_current_access_path(t), + GENERIC_READ, FILE_SHARE_READ, NULL, OPEN_EXISTING, flags, NULL); + if (t->entry_fh == INVALID_HANDLE_VALUE) { + la_dosmaperr(GetLastError()); + archive_set_error(&a->archive, errno, + "Couldn't open %ls", tree_current_path(a->tree)); + return (ARCHIVE_FAILED); + } + + /* Find sparse data from the disk. */ + if (archive_entry_hardlink(entry) == NULL && + (st->dwFileAttributes & FILE_ATTRIBUTE_SPARSE_FILE) != 0) + r = setup_sparse_from_disk(a, entry, t->entry_fh); + } + return (r); +} + +static int +_archive_read_next_header(struct archive *_a, struct archive_entry **entryp) +{ + int ret; + struct archive_read_disk *a = (struct archive_read_disk *)_a; + *entryp = NULL; + ret = _archive_read_next_header2(_a, a->entry); + *entryp = a->entry; + return ret; +} + +static int +_archive_read_next_header2(struct archive *_a, struct archive_entry *entry) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + struct tree *t; + int r; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, + "archive_read_next_header2"); + + t = a->tree; + if (t->entry_fh != INVALID_HANDLE_VALUE) { + cancel_async(t); + close_and_restore_time(t->entry_fh, t, &t->restore_time); + t->entry_fh = INVALID_HANDLE_VALUE; + } + + archive_entry_clear(entry); + + while ((r = next_entry(a, t, entry)) == ARCHIVE_RETRY) + archive_entry_clear(entry); + + /* + * EOF and FATAL are persistent at this layer. By + * modifying the state, we guarantee that future calls to + * read a header or read data will fail. + */ + switch (r) { + case ARCHIVE_EOF: + a->archive.state = ARCHIVE_STATE_EOF; + break; + case ARCHIVE_OK: + case ARCHIVE_WARN: + t->entry_total = 0; + if (archive_entry_filetype(entry) == AE_IFREG) { + t->entry_remaining_bytes = archive_entry_size(entry); + t->entry_eof = (t->entry_remaining_bytes == 0)? 1: 0; + if (!t->entry_eof && + setup_sparse(a, entry) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } else { + t->entry_remaining_bytes = 0; + t->entry_eof = 1; + } + t->ol_idx_doing = t->ol_idx_done = 0; + t->ol_num_doing = t->ol_num_done = 0; + t->ol_remaining_bytes = t->entry_remaining_bytes; + t->ol_total = 0; + a->archive.state = ARCHIVE_STATE_DATA; + break; + case ARCHIVE_RETRY: + break; + case ARCHIVE_FATAL: + a->archive.state = ARCHIVE_STATE_FATAL; + break; + } + + __archive_reset_read_data(&a->archive); + return (r); +} + +static int +setup_sparse(struct archive_read_disk *a, struct archive_entry *entry) +{ + struct tree *t = a->tree; + int64_t aligned, length, offset; + int i; + + t->sparse_count = archive_entry_sparse_reset(entry); + if (t->sparse_count+1 > t->sparse_list_size) { + free(t->sparse_list); + t->sparse_list_size = t->sparse_count + 1; + t->sparse_list = malloc(sizeof(t->sparse_list[0]) * + t->sparse_list_size); + if (t->sparse_list == NULL) { + t->sparse_list_size = 0; + archive_set_error(&a->archive, ENOMEM, + "Can't allocate data"); + a->archive.state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + } + } + /* + * Get sparse list and make sure those offsets and lengths are + * aligned by a sector size. + */ + for (i = 0; i < t->sparse_count; i++) { + archive_entry_sparse_next(entry, &offset, &length); + aligned = align_num_per_sector(t, offset); + if (aligned != offset) { + aligned -= t->current_filesystem->bytesPerSector; + length += offset - aligned; + } + t->sparse_list[i].offset = aligned; + aligned = align_num_per_sector(t, length); + t->sparse_list[i].length = aligned; + } + + aligned = align_num_per_sector(t, archive_entry_size(entry)); + if (i == 0) { + t->sparse_list[i].offset = 0; + t->sparse_list[i].length = aligned; + } else { + int j, last = i; + + t->sparse_list[i].offset = aligned; + t->sparse_list[i].length = 0; + for (i = 0; i < last; i++) { + if ((t->sparse_list[i].offset + + t->sparse_list[i].length) <= + t->sparse_list[i+1].offset) + continue; + /* + * Now sparse_list[i+1] is overlapped by sparse_list[i]. + * Merge those two. + */ + length = t->sparse_list[i+1].offset - + t->sparse_list[i].offset; + t->sparse_list[i+1].offset = t->sparse_list[i].offset; + t->sparse_list[i+1].length += length; + /* Remove sparse_list[i]. */ + for (j = i; j < last; j++) { + t->sparse_list[j].offset = + t->sparse_list[j+1].offset; + t->sparse_list[j].length = + t->sparse_list[j+1].length; + } + last--; + } + } + t->current_sparse = t->sparse_list; + + return (ARCHIVE_OK); +} + +int +archive_read_disk_set_matching(struct archive *_a, struct archive *_ma, + void (*_excluded_func)(struct archive *, void *, struct archive_entry *), + void *_client_data) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_read_disk_set_matching"); + a->matching = _ma; + a->excluded_cb_func = _excluded_func; + a->excluded_cb_data = _client_data; + return (ARCHIVE_OK); +} + +int +archive_read_disk_set_metadata_filter_callback(struct archive *_a, + int (*_metadata_filter_func)(struct archive *, void *, + struct archive_entry *), void *_client_data) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, + "archive_read_disk_set_metadata_filter_callback"); + + a->metadata_filter_func = _metadata_filter_func; + a->metadata_filter_data = _client_data; + return (ARCHIVE_OK); +} + +int +archive_read_disk_can_descend(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + struct tree *t = a->tree; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, + "archive_read_disk_can_descend"); + + return (t->visit_type == TREE_REGULAR && t->descend); +} + +/* + * Called by the client to mark the directory just returned from + * tree_next() as needing to be visited. + */ +int +archive_read_disk_descend(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + struct tree *t = a->tree; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, + "archive_read_disk_descend"); + + if (t->visit_type != TREE_REGULAR || !t->descend) + return (ARCHIVE_OK); + + if (tree_current_is_physical_dir(t)) { + tree_push(t, t->basename, t->full_path.s, + t->current_filesystem_id, + bhfi_dev(&(t->lst)), bhfi_ino(&(t->lst)), + &t->restore_time); + t->stack->flags |= isDir; + } else if (tree_current_is_dir(t)) { + tree_push(t, t->basename, t->full_path.s, + t->current_filesystem_id, + bhfi_dev(&(t->st)), bhfi_ino(&(t->st)), + &t->restore_time); + t->stack->flags |= isDirLink; + } + t->descend = 0; + return (ARCHIVE_OK); +} + +int +archive_read_disk_open(struct archive *_a, const char *pathname) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + struct archive_wstring wpath; + int ret; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_NEW | ARCHIVE_STATE_CLOSED, + "archive_read_disk_open"); + archive_clear_error(&a->archive); + + /* Make a wchar_t string from a char string. */ + archive_string_init(&wpath); + if (archive_wstring_append_from_mbs(&wpath, pathname, + strlen(pathname)) != 0) { + if (errno == ENOMEM) + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + else + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Can't convert a path to a wchar_t string"); + a->archive.state = ARCHIVE_STATE_FATAL; + ret = ARCHIVE_FATAL; + } else + ret = _archive_read_disk_open_w(_a, wpath.s); + + archive_wstring_free(&wpath); + return (ret); +} + +int +archive_read_disk_open_w(struct archive *_a, const wchar_t *pathname) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, + ARCHIVE_STATE_NEW | ARCHIVE_STATE_CLOSED, + "archive_read_disk_open_w"); + archive_clear_error(&a->archive); + + return (_archive_read_disk_open_w(_a, pathname)); +} + +static int +_archive_read_disk_open_w(struct archive *_a, const wchar_t *pathname) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + + if (a->tree != NULL) + a->tree = tree_reopen(a->tree, pathname, + a->flags & ARCHIVE_READDISK_RESTORE_ATIME); + else + a->tree = tree_open(pathname, a->symlink_mode, + a->flags & ARCHIVE_READDISK_RESTORE_ATIME); + if (a->tree == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate directory traversal data"); + a->archive.state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + } + a->archive.state = ARCHIVE_STATE_HEADER; + + return (ARCHIVE_OK); +} + +/* + * Return a current filesystem ID which is index of the filesystem entry + * you've visited through archive_read_disk. + */ +int +archive_read_disk_current_filesystem(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_DATA, + "archive_read_disk_current_filesystem"); + + return (a->tree->current_filesystem_id); +} + +static int +update_current_filesystem(struct archive_read_disk *a, int64_t dev) +{ + struct tree *t = a->tree; + int i, fid; + + if (t->current_filesystem != NULL && + t->current_filesystem->dev == dev) + return (ARCHIVE_OK); + + for (i = 0; i < t->max_filesystem_id; i++) { + if (t->filesystem_table[i].dev == dev) { + /* There is the filesystem ID we've already generated. */ + t->current_filesystem_id = i; + t->current_filesystem = &(t->filesystem_table[i]); + return (ARCHIVE_OK); + } + } + + /* + * There is a new filesystem, we generate a new ID for. + */ + fid = t->max_filesystem_id++; + if (t->max_filesystem_id > t->allocated_filesystem) { + size_t s; + void *p; + + s = t->max_filesystem_id * 2; + p = realloc(t->filesystem_table, + s * sizeof(*t->filesystem_table)); + if (p == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate tar data"); + return (ARCHIVE_FATAL); + } + t->filesystem_table = (struct filesystem *)p; + t->allocated_filesystem = (int)s; + } + t->current_filesystem_id = fid; + t->current_filesystem = &(t->filesystem_table[fid]); + t->current_filesystem->dev = dev; + + return (setup_current_filesystem(a)); +} + +/* + * Returns 1 if current filesystem is generated filesystem, 0 if it is not + * or -1 if it is unknown. + */ +int +archive_read_disk_current_filesystem_is_synthetic(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_DATA, + "archive_read_disk_current_filesystem"); + + return (a->tree->current_filesystem->synthetic); +} + +/* + * Returns 1 if current filesystem is remote filesystem, 0 if it is not + * or -1 if it is unknown. + */ +int +archive_read_disk_current_filesystem_is_remote(struct archive *_a) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + + archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_DATA, + "archive_read_disk_current_filesystem"); + + return (a->tree->current_filesystem->remote); +} + +/* + * If symlink is broken, statfs or statvfs will fail. + * Use its directory path instead. + */ +static wchar_t * +safe_path_for_statfs(struct tree *t) +{ + const wchar_t *path; + wchar_t *cp, *p = NULL; + + path = tree_current_access_path(t); + if (tree_current_stat(t) == NULL) { + p = _wcsdup(path); + cp = wcsrchr(p, '/'); + if (cp != NULL && wcslen(cp) >= 2) { + cp[1] = '.'; + cp[2] = '\0'; + path = p; + } + } else + p = _wcsdup(path); + return (p); +} + +/* + * Get conditions of synthetic and remote on Windows + */ +static int +setup_current_filesystem(struct archive_read_disk *a) +{ + struct tree *t = a->tree; + wchar_t vol[256]; + wchar_t *path; + + t->current_filesystem->synthetic = -1;/* Not supported */ + path = safe_path_for_statfs(t); + if (!GetVolumePathNameW(path, vol, sizeof(vol)/sizeof(vol[0]))) { + free(path); + t->current_filesystem->remote = -1; + t->current_filesystem->bytesPerSector = 0; + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "GetVolumePathName failed: %d", (int)GetLastError()); + return (ARCHIVE_FAILED); + } + free(path); + switch (GetDriveTypeW(vol)) { + case DRIVE_UNKNOWN: + case DRIVE_NO_ROOT_DIR: + t->current_filesystem->remote = -1; + break; + case DRIVE_REMOTE: + t->current_filesystem->remote = 1; + break; + default: + t->current_filesystem->remote = 0; + break; + } + + if (!GetDiskFreeSpaceW(vol, NULL, + &(t->current_filesystem->bytesPerSector), NULL, NULL)) { + t->current_filesystem->bytesPerSector = 0; + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "GetDiskFreeSpace failed: %d", (int)GetLastError()); + return (ARCHIVE_FAILED); + } + + return (ARCHIVE_OK); +} + +static int +close_and_restore_time(HANDLE h, struct tree *t, struct restore_time *rt) +{ + HANDLE handle; + int r = 0; + + if (h == INVALID_HANDLE_VALUE && AE_IFLNK == rt->filetype) + return (0); + + /* Close a file descriptor. + * It will not be used for SetFileTime() because it has been opened + * by a read only mode. + */ + if (h != INVALID_HANDLE_VALUE) + CloseHandle(h); + if ((t->flags & needsRestoreTimes) == 0) + return (r); + + handle = CreateFileW(rt->full_path, FILE_WRITE_ATTRIBUTES, + 0, NULL, OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS, NULL); + if (handle == INVALID_HANDLE_VALUE) { + errno = EINVAL; + return (-1); + } + + if (SetFileTime(handle, NULL, &rt->lastAccessTime, + &rt->lastWriteTime) == 0) { + errno = EINVAL; + r = -1; + } else + r = 0; + CloseHandle(handle); + return (r); +} + +/* + * Add a directory path to the current stack. + */ +static void +tree_push(struct tree *t, const wchar_t *path, const wchar_t *full_path, + int filesystem_id, int64_t dev, int64_t ino, struct restore_time *rt) +{ + struct tree_entry *te; + + te = calloc(1, sizeof(*te)); + te->next = t->stack; + te->parent = t->current; + if (te->parent) + te->depth = te->parent->depth + 1; + t->stack = te; + archive_string_init(&te->name); + archive_wstrcpy(&te->name, path); + archive_string_init(&te->full_path); + archive_wstrcpy(&te->full_path, full_path); + te->flags = needsDescent | needsOpen | needsAscent; + te->filesystem_id = filesystem_id; + te->dev = dev; + te->ino = ino; + te->dirname_length = t->dirname_length; + te->full_path_dir_length = t->full_path_dir_length; + te->restore_time.full_path = te->full_path.s; + if (rt != NULL) { + te->restore_time.lastWriteTime = rt->lastWriteTime; + te->restore_time.lastAccessTime = rt->lastAccessTime; + te->restore_time.filetype = rt->filetype; + } +} + +/* + * Append a name to the current dir path. + */ +static void +tree_append(struct tree *t, const wchar_t *name, size_t name_length) +{ + size_t size_needed; + + t->path.s[t->dirname_length] = L'\0'; + t->path.length = t->dirname_length; + /* Strip trailing '/' from name, unless entire name is "/". */ + while (name_length > 1 && name[name_length - 1] == L'/') + name_length--; + + /* Resize pathname buffer as needed. */ + size_needed = name_length + t->dirname_length + 2; + archive_wstring_ensure(&t->path, size_needed); + /* Add a separating '/' if it's needed. */ + if (t->dirname_length > 0 && + t->path.s[archive_strlen(&t->path)-1] != L'/') + archive_wstrappend_wchar(&t->path, L'/'); + t->basename = t->path.s + archive_strlen(&t->path); + archive_wstrncat(&t->path, name, name_length); + t->restore_time.full_path = t->basename; + if (t->full_path_dir_length > 0) { + t->full_path.s[t->full_path_dir_length] = L'\0'; + t->full_path.length = t->full_path_dir_length; + size_needed = name_length + t->full_path_dir_length + 2; + archive_wstring_ensure(&t->full_path, size_needed); + /* Add a separating '\' if it's needed. */ + if (t->full_path.s[archive_strlen(&t->full_path)-1] != L'\\') + archive_wstrappend_wchar(&t->full_path, L'\\'); + archive_wstrncat(&t->full_path, name, name_length); + t->restore_time.full_path = t->full_path.s; + } +} + +/* + * Open a directory tree for traversal. + */ +static struct tree * +tree_open(const wchar_t *path, int symlink_mode, int restore_time) +{ + struct tree *t; + + t = calloc(1, sizeof(*t)); + archive_string_init(&(t->full_path)); + archive_string_init(&t->path); + archive_wstring_ensure(&t->path, 15); + t->initial_symlink_mode = symlink_mode; + return (tree_reopen(t, path, restore_time)); +} + +static struct tree * +tree_reopen(struct tree *t, const wchar_t *path, int restore_time) +{ + struct archive_wstring ws; + wchar_t *pathname, *p, *base; + + t->flags = (restore_time != 0)?needsRestoreTimes:0; + t->visit_type = 0; + t->tree_errno = 0; + t->full_path_dir_length = 0; + t->dirname_length = 0; + t->depth = 0; + t->descend = 0; + t->current = NULL; + t->d = INVALID_HANDLE_VALUE; + t->symlink_mode = t->initial_symlink_mode; + archive_string_empty(&(t->full_path)); + archive_string_empty(&t->path); + t->entry_fh = INVALID_HANDLE_VALUE; + t->entry_eof = 0; + t->entry_remaining_bytes = 0; + t->initial_filesystem_id = -1; + + /* Get wchar_t strings from char strings. */ + archive_string_init(&ws); + archive_wstrcpy(&ws, path); + pathname = ws.s; + /* Get a full-path-name. */ + p = __la_win_permissive_name_w(pathname); + if (p == NULL) + goto failed; + archive_wstrcpy(&(t->full_path), p); + free(p); + + /* Convert path separators from '\' to '/' */ + for (p = pathname; *p != L'\0'; ++p) { + if (*p == L'\\') + *p = L'/'; + } + base = pathname; + + /* First item is set up a lot like a symlink traversal. */ + /* printf("Looking for wildcard in %s\n", path); */ + if ((base[0] == L'/' && base[1] == L'/' && + base[2] == L'?' && base[3] == L'/' && + (wcschr(base+4, L'*') || wcschr(base+4, L'?'))) || + (!(base[0] == L'/' && base[1] == L'/' && + base[2] == L'?' && base[3] == L'/') && + (wcschr(base, L'*') || wcschr(base, L'?')))) { + // It has a wildcard in it... + // Separate the last element. + p = wcsrchr(base, L'/'); + if (p != NULL) { + *p = L'\0'; + tree_append(t, base, p - base); + t->dirname_length = archive_strlen(&t->path); + base = p + 1; + } + p = wcsrchr(t->full_path.s, L'\\'); + if (p != NULL) { + *p = L'\0'; + t->full_path.length = wcslen(t->full_path.s); + t->full_path_dir_length = archive_strlen(&t->full_path); + } + } + tree_push(t, base, t->full_path.s, 0, 0, 0, NULL); + archive_wstring_free(&ws); + t->stack->flags = needsFirstVisit; + /* + * Debug flag for Direct IO(No buffering) or Async IO. + * Those dependent on environment variable switches + * will be removed until next release. + */ + { + const char *e; + if ((e = getenv("LIBARCHIVE_DIRECT_IO")) != NULL) { + if (e[0] == '0') + t->direct_io = 0; + else + t->direct_io = 1; + fprintf(stderr, "LIBARCHIVE_DIRECT_IO=%s\n", + (t->direct_io)?"Enabled":"Disabled"); + } else + t->direct_io = DIRECT_IO; + if ((e = getenv("LIBARCHIVE_ASYNC_IO")) != NULL) { + if (e[0] == '0') + t->async_io = 0; + else + t->async_io = 1; + fprintf(stderr, "LIBARCHIVE_ASYNC_IO=%s\n", + (t->async_io)?"Enabled":"Disabled"); + } else + t->async_io = ASYNC_IO; + } + return (t); +failed: + archive_wstring_free(&ws); + tree_free(t); + return (NULL); +} + +static int +tree_descent(struct tree *t) +{ + t->dirname_length = archive_strlen(&t->path); + t->full_path_dir_length = archive_strlen(&t->full_path); + t->depth++; + return (0); +} + +/* + * We've finished a directory; ascend back to the parent. + */ +static int +tree_ascend(struct tree *t) +{ + struct tree_entry *te; + + te = t->stack; + t->depth--; + close_and_restore_time(INVALID_HANDLE_VALUE, t, &te->restore_time); + return (0); +} + +/* + * Pop the working stack. + */ +static void +tree_pop(struct tree *t) +{ + struct tree_entry *te; + + t->full_path.s[t->full_path_dir_length] = L'\0'; + t->full_path.length = t->full_path_dir_length; + t->path.s[t->dirname_length] = L'\0'; + t->path.length = t->dirname_length; + if (t->stack == t->current && t->current != NULL) + t->current = t->current->parent; + te = t->stack; + t->stack = te->next; + t->dirname_length = te->dirname_length; + t->basename = t->path.s + t->dirname_length; + t->full_path_dir_length = te->full_path_dir_length; + while (t->basename[0] == L'/') + t->basename++; + archive_wstring_free(&te->name); + archive_wstring_free(&te->full_path); + free(te); +} + +/* + * Get the next item in the tree traversal. + */ +static int +tree_next(struct tree *t) +{ + int r; + + while (t->stack != NULL) { + /* If there's an open dir, get the next entry from there. */ + if (t->d != INVALID_HANDLE_VALUE) { + r = tree_dir_next_windows(t, NULL); + if (r == 0) + continue; + return (r); + } + + if (t->stack->flags & needsFirstVisit) { + wchar_t *d = t->stack->name.s; + t->stack->flags &= ~needsFirstVisit; + if (!(d[0] == L'/' && d[1] == L'/' && + d[2] == L'?' && d[3] == L'/') && + (wcschr(d, L'*') || wcschr(d, L'?'))) { + r = tree_dir_next_windows(t, d); + if (r == 0) + continue; + return (r); + } else { + HANDLE h = FindFirstFileW(d, &t->_findData); + if (h == INVALID_HANDLE_VALUE) { + la_dosmaperr(GetLastError()); + t->tree_errno = errno; + t->visit_type = TREE_ERROR_DIR; + return (t->visit_type); + } + t->findData = &t->_findData; + FindClose(h); + } + /* Top stack item needs a regular visit. */ + t->current = t->stack; + tree_append(t, t->stack->name.s, + archive_strlen(&(t->stack->name))); + //t->dirname_length = t->path_length; + //tree_pop(t); + t->stack->flags &= ~needsFirstVisit; + return (t->visit_type = TREE_REGULAR); + } else if (t->stack->flags & needsDescent) { + /* Top stack item is dir to descend into. */ + t->current = t->stack; + tree_append(t, t->stack->name.s, + archive_strlen(&(t->stack->name))); + t->stack->flags &= ~needsDescent; + r = tree_descent(t); + if (r != 0) { + tree_pop(t); + t->visit_type = r; + } else + t->visit_type = TREE_POSTDESCENT; + return (t->visit_type); + } else if (t->stack->flags & needsOpen) { + t->stack->flags &= ~needsOpen; + r = tree_dir_next_windows(t, L"*"); + if (r == 0) + continue; + return (r); + } else if (t->stack->flags & needsAscent) { + /* Top stack item is dir and we're done with it. */ + r = tree_ascend(t); + tree_pop(t); + t->visit_type = r != 0 ? r : TREE_POSTASCENT; + return (t->visit_type); + } else { + /* Top item on stack is dead. */ + tree_pop(t); + t->flags &= ~hasLstat; + t->flags &= ~hasStat; + } + } + return (t->visit_type = 0); +} + +static int +tree_dir_next_windows(struct tree *t, const wchar_t *pattern) +{ + const wchar_t *name; + size_t namelen; + int r; + + for (;;) { + if (pattern != NULL) { + struct archive_wstring pt; + + archive_string_init(&pt); + archive_wstring_ensure(&pt, + archive_strlen(&(t->full_path)) + + 2 + wcslen(pattern)); + archive_wstring_copy(&pt, &(t->full_path)); + archive_wstrappend_wchar(&pt, L'\\'); + archive_wstrcat(&pt, pattern); + t->d = FindFirstFileW(pt.s, &t->_findData); + archive_wstring_free(&pt); + if (t->d == INVALID_HANDLE_VALUE) { + la_dosmaperr(GetLastError()); + t->tree_errno = errno; + r = tree_ascend(t); /* Undo "chdir" */ + tree_pop(t); + t->visit_type = r != 0 ? r : TREE_ERROR_DIR; + return (t->visit_type); + } + t->findData = &t->_findData; + pattern = NULL; + } else if (!FindNextFileW(t->d, &t->_findData)) { + FindClose(t->d); + t->d = INVALID_HANDLE_VALUE; + t->findData = NULL; + return (0); + } + name = t->findData->cFileName; + namelen = wcslen(name); + t->flags &= ~hasLstat; + t->flags &= ~hasStat; + if (name[0] == L'.' && name[1] == L'\0') + continue; + if (name[0] == L'.' && name[1] == L'.' && name[2] == L'\0') + continue; + tree_append(t, name, namelen); + return (t->visit_type = TREE_REGULAR); + } +} + +#define EPOC_TIME ARCHIVE_LITERAL_ULL(116444736000000000) +static void +fileTimeToUtc(const FILETIME *filetime, time_t *t, long *ns) +{ + ULARGE_INTEGER utc; + + utc.HighPart = filetime->dwHighDateTime; + utc.LowPart = filetime->dwLowDateTime; + if (utc.QuadPart >= EPOC_TIME) { + utc.QuadPart -= EPOC_TIME; + /* milli seconds base */ + *t = (time_t)(utc.QuadPart / 10000000); + /* nano seconds base */ + *ns = (long)(utc.QuadPart % 10000000) * 100; + } else { + *t = 0; + *ns = 0; + } +} + +static void +entry_copy_bhfi(struct archive_entry *entry, const wchar_t *path, + const WIN32_FIND_DATAW *findData, + const BY_HANDLE_FILE_INFORMATION *bhfi) +{ + time_t secs; + long nsecs; + mode_t mode; + + fileTimeToUtc(&bhfi->ftLastAccessTime, &secs, &nsecs); + archive_entry_set_atime(entry, secs, nsecs); + fileTimeToUtc(&bhfi->ftLastWriteTime, &secs, &nsecs); + archive_entry_set_mtime(entry, secs, nsecs); + fileTimeToUtc(&bhfi->ftCreationTime, &secs, &nsecs); + archive_entry_set_birthtime(entry, secs, nsecs); + archive_entry_set_ctime(entry, secs, nsecs); + archive_entry_set_dev(entry, bhfi_dev(bhfi)); + archive_entry_set_ino64(entry, bhfi_ino(bhfi)); + if (bhfi->dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY) + archive_entry_set_nlink(entry, bhfi->nNumberOfLinks + 1); + else + archive_entry_set_nlink(entry, bhfi->nNumberOfLinks); + archive_entry_set_size(entry, + (((int64_t)bhfi->nFileSizeHigh) << 32) + + bhfi->nFileSizeLow); + archive_entry_set_uid(entry, 0); + archive_entry_set_gid(entry, 0); + archive_entry_set_rdev(entry, 0); + + mode = S_IRUSR | S_IRGRP | S_IROTH; + if ((bhfi->dwFileAttributes & FILE_ATTRIBUTE_READONLY) == 0) + mode |= S_IWUSR | S_IWGRP | S_IWOTH; + if ((bhfi->dwFileAttributes & FILE_ATTRIBUTE_REPARSE_POINT) && + findData != NULL && + findData->dwReserved0 == IO_REPARSE_TAG_SYMLINK) { + mode |= S_IFLNK; + entry_symlink_from_pathw(entry, path); + } else if (bhfi->dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY) + mode |= S_IFDIR | S_IXUSR | S_IXGRP | S_IXOTH; + else { + const wchar_t *p; + + mode |= S_IFREG; + p = wcsrchr(path, L'.'); + if (p != NULL && wcslen(p) == 4) { + switch (p[1]) { + case L'B': case L'b': + if ((p[2] == L'A' || p[2] == L'a' ) && + (p[3] == L'T' || p[3] == L't' )) + mode |= S_IXUSR | S_IXGRP | S_IXOTH; + break; + case L'C': case L'c': + if (((p[2] == L'M' || p[2] == L'm' ) && + (p[3] == L'D' || p[3] == L'd' ))) + mode |= S_IXUSR | S_IXGRP | S_IXOTH; + break; + case L'E': case L'e': + if ((p[2] == L'X' || p[2] == L'x' ) && + (p[3] == L'E' || p[3] == L'e' )) + mode |= S_IXUSR | S_IXGRP | S_IXOTH; + break; + default: + break; + } + } + } + archive_entry_set_mode(entry, mode); +} + +static void +tree_archive_entry_copy_bhfi(struct archive_entry *entry, struct tree *t, + const BY_HANDLE_FILE_INFORMATION *bhfi) +{ + entry_copy_bhfi(entry, tree_current_path(t), t->findData, bhfi); +} + +static int +tree_current_file_information(struct tree *t, BY_HANDLE_FILE_INFORMATION *st, + int sim_lstat) +{ + HANDLE h; + int r; + DWORD flag = FILE_FLAG_BACKUP_SEMANTICS; + + if (sim_lstat && tree_current_is_physical_link(t)) + flag |= FILE_FLAG_OPEN_REPARSE_POINT; + h = CreateFileW(tree_current_access_path(t), 0, FILE_SHARE_READ, NULL, + OPEN_EXISTING, flag, NULL); + if (h == INVALID_HANDLE_VALUE) { + la_dosmaperr(GetLastError()); + t->tree_errno = errno; + return (0); + } + r = GetFileInformationByHandle(h, st); + CloseHandle(h); + return (r); +} + +/* + * Get the stat() data for the entry just returned from tree_next(). + */ +static const BY_HANDLE_FILE_INFORMATION * +tree_current_stat(struct tree *t) +{ + if (!(t->flags & hasStat)) { + if (!tree_current_file_information(t, &t->st, 0)) + return NULL; + t->flags |= hasStat; + } + return (&t->st); +} + +/* + * Get the lstat() data for the entry just returned from tree_next(). + */ +static const BY_HANDLE_FILE_INFORMATION * +tree_current_lstat(struct tree *t) +{ + if (!(t->flags & hasLstat)) { + if (!tree_current_file_information(t, &t->lst, 1)) + return NULL; + t->flags |= hasLstat; + } + return (&t->lst); +} + +/* + * Test whether current entry is a dir or link to a dir. + */ +static int +tree_current_is_dir(struct tree *t) +{ + if (t->findData) + return (t->findData->dwFileAttributes + & FILE_ATTRIBUTE_DIRECTORY); + return (0); +} + +/* + * Test whether current entry is a physical directory. Usually, we + * already have at least one of stat() or lstat() in memory, so we + * use tricks to try to avoid an extra trip to the disk. + */ +static int +tree_current_is_physical_dir(struct tree *t) +{ + if (tree_current_is_physical_link(t)) + return (0); + return (tree_current_is_dir(t)); +} + +/* + * Test whether current entry is a symbolic link. + */ +static int +tree_current_is_physical_link(struct tree *t) +{ + if (t->findData) + return ((t->findData->dwFileAttributes + & FILE_ATTRIBUTE_REPARSE_POINT) && + (t->findData->dwReserved0 + == IO_REPARSE_TAG_SYMLINK)); + return (0); +} + +/* + * Test whether the same file has been in the tree as its parent. + */ +static int +tree_target_is_same_as_parent(struct tree *t, + const BY_HANDLE_FILE_INFORMATION *st) +{ + struct tree_entry *te; + int64_t dev = bhfi_dev(st); + int64_t ino = bhfi_ino(st); + + for (te = t->current->parent; te != NULL; te = te->parent) { + if (te->dev == dev && te->ino == ino) + return (1); + } + return (0); +} + +/* + * Return the access path for the entry just returned from tree_next(). + */ +static const wchar_t * +tree_current_access_path(struct tree *t) +{ + return (t->full_path.s); +} + +/* + * Return the full path for the entry just returned from tree_next(). + */ +static const wchar_t * +tree_current_path(struct tree *t) +{ + return (t->path.s); +} + +/* + * Terminate the traversal. + */ +static void +tree_close(struct tree *t) +{ + + if (t == NULL) + return; + if (t->entry_fh != INVALID_HANDLE_VALUE) { + cancel_async(t); + close_and_restore_time(t->entry_fh, t, &t->restore_time); + t->entry_fh = INVALID_HANDLE_VALUE; + } + /* Close the handle of FindFirstFileW */ + if (t->d != INVALID_HANDLE_VALUE) { + FindClose(t->d); + t->d = INVALID_HANDLE_VALUE; + t->findData = NULL; + } + /* Release anything remaining in the stack. */ + while (t->stack != NULL) + tree_pop(t); +} + +/* + * Release any resources. + */ +static void +tree_free(struct tree *t) +{ + int i; + + if (t == NULL) + return; + archive_wstring_free(&t->path); + archive_wstring_free(&t->full_path); + free(t->sparse_list); + free(t->filesystem_table); + for (i = 0; i < MAX_OVERLAPPED; i++) { + if (t->ol[i].buff) + VirtualFree(t->ol[i].buff, 0, MEM_RELEASE); + CloseHandle(t->ol[i].ol.hEvent); + } + free(t); +} + + +/* + * Populate the archive_entry with metadata from the disk. + */ +int +archive_read_disk_entry_from_file(struct archive *_a, + struct archive_entry *entry, int fd, const struct stat *st) +{ + struct archive_read_disk *a = (struct archive_read_disk *)_a; + const wchar_t *path; + const wchar_t *wname; + const char *name; + HANDLE h; + BY_HANDLE_FILE_INFORMATION bhfi; + DWORD fileAttributes = 0; + int r; + + archive_clear_error(_a); + wname = archive_entry_sourcepath_w(entry); + if (wname == NULL) + wname = archive_entry_pathname_w(entry); + if (wname == NULL) { + archive_set_error(&a->archive, EINVAL, + "Can't get a wide character version of the path"); + return (ARCHIVE_FAILED); + } + path = __la_win_permissive_name_w(wname); + + if (st == NULL) { + /* + * Get metadata through GetFileInformationByHandle(). + */ + if (fd >= 0) { + h = (HANDLE)_get_osfhandle(fd); + r = GetFileInformationByHandle(h, &bhfi); + if (r == 0) { + la_dosmaperr(GetLastError()); + archive_set_error(&a->archive, errno, + "Can't GetFileInformationByHandle"); + return (ARCHIVE_FAILED); + } + entry_copy_bhfi(entry, path, NULL, &bhfi); + } else { + WIN32_FIND_DATAW findData; + DWORD flag, desiredAccess; + + h = FindFirstFileW(path, &findData); + if (h == INVALID_HANDLE_VALUE) { + la_dosmaperr(GetLastError()); + archive_set_error(&a->archive, errno, + "Can't FindFirstFileW"); + return (ARCHIVE_FAILED); + } + FindClose(h); + + flag = FILE_FLAG_BACKUP_SEMANTICS; + if (!a->follow_symlinks && + (findData.dwFileAttributes + & FILE_ATTRIBUTE_REPARSE_POINT) && + (findData.dwReserved0 == IO_REPARSE_TAG_SYMLINK)) { + flag |= FILE_FLAG_OPEN_REPARSE_POINT; + desiredAccess = 0; + } else if (findData.dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY) { + desiredAccess = 0; + } else + desiredAccess = GENERIC_READ; + + h = CreateFileW(path, desiredAccess, FILE_SHARE_READ, NULL, + OPEN_EXISTING, flag, NULL); + if (h == INVALID_HANDLE_VALUE) { + la_dosmaperr(GetLastError()); + archive_set_error(&a->archive, errno, + "Can't CreateFileW"); + return (ARCHIVE_FAILED); + } + r = GetFileInformationByHandle(h, &bhfi); + if (r == 0) { + la_dosmaperr(GetLastError()); + archive_set_error(&a->archive, errno, + "Can't GetFileInformationByHandle"); + CloseHandle(h); + return (ARCHIVE_FAILED); + } + entry_copy_bhfi(entry, path, &findData, &bhfi); + } + fileAttributes = bhfi.dwFileAttributes; + } else { + archive_entry_copy_stat(entry, st); + if (st->st_mode & S_IFLNK) + entry_symlink_from_pathw(entry, path); + h = INVALID_HANDLE_VALUE; + } + + /* Lookup uname/gname */ + name = archive_read_disk_uname(_a, archive_entry_uid(entry)); + if (name != NULL) + archive_entry_copy_uname(entry, name); + name = archive_read_disk_gname(_a, archive_entry_gid(entry)); + if (name != NULL) + archive_entry_copy_gname(entry, name); + + /* + * File attributes + */ + if ((a->flags & ARCHIVE_READDISK_NO_FFLAGS) == 0) { + const int supported_attrs = + FILE_ATTRIBUTE_READONLY | + FILE_ATTRIBUTE_HIDDEN | + FILE_ATTRIBUTE_SYSTEM; + DWORD file_attrs = fileAttributes & supported_attrs; + if (file_attrs != 0) + archive_entry_set_fflags(entry, file_attrs, 0); + } + + /* + * Can this file be sparse file ? + */ + if (archive_entry_filetype(entry) != AE_IFREG + || archive_entry_size(entry) <= 0 + || archive_entry_hardlink(entry) != NULL) { + if (h != INVALID_HANDLE_VALUE && fd < 0) + CloseHandle(h); + return (ARCHIVE_OK); + } + + if (h == INVALID_HANDLE_VALUE) { + if (fd >= 0) { + h = (HANDLE)_get_osfhandle(fd); + } else { + h = CreateFileW(path, GENERIC_READ, FILE_SHARE_READ, NULL, + OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS, NULL); + if (h == INVALID_HANDLE_VALUE) { + la_dosmaperr(GetLastError()); + archive_set_error(&a->archive, errno, + "Can't CreateFileW"); + return (ARCHIVE_FAILED); + } + } + r = GetFileInformationByHandle(h, &bhfi); + if (r == 0) { + la_dosmaperr(GetLastError()); + archive_set_error(&a->archive, errno, + "Can't GetFileInformationByHandle"); + if (h != INVALID_HANDLE_VALUE && fd < 0) + CloseHandle(h); + return (ARCHIVE_FAILED); + } + fileAttributes = bhfi.dwFileAttributes; + } + + /* Sparse file must be set a mark, FILE_ATTRIBUTE_SPARSE_FILE */ + if ((fileAttributes & FILE_ATTRIBUTE_SPARSE_FILE) == 0) { + if (fd < 0) + CloseHandle(h); + return (ARCHIVE_OK); + } + + r = setup_sparse_from_disk(a, entry, h); + if (fd < 0) + CloseHandle(h); + + return (r); +} + +/* + * Windows sparse interface. + */ +#if defined(__MINGW32__) && !defined(FSCTL_QUERY_ALLOCATED_RANGES) +#define FSCTL_QUERY_ALLOCATED_RANGES 0x940CF +typedef struct { + LARGE_INTEGER FileOffset; + LARGE_INTEGER Length; +} FILE_ALLOCATED_RANGE_BUFFER; +#endif + +static int +setup_sparse_from_disk(struct archive_read_disk *a, + struct archive_entry *entry, HANDLE handle) +{ + FILE_ALLOCATED_RANGE_BUFFER range, *outranges = NULL; + size_t outranges_size; + int64_t entry_size = archive_entry_size(entry); + int exit_sts = ARCHIVE_OK; + + range.FileOffset.QuadPart = 0; + range.Length.QuadPart = entry_size; + outranges_size = 2048; + outranges = (FILE_ALLOCATED_RANGE_BUFFER *)malloc(outranges_size); + if (outranges == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Couldn't allocate memory"); + exit_sts = ARCHIVE_FATAL; + goto exit_setup_sparse; + } + + for (;;) { + DWORD retbytes; + BOOL ret; + + for (;;) { + ret = DeviceIoControl(handle, + FSCTL_QUERY_ALLOCATED_RANGES, + &range, sizeof(range), outranges, + (DWORD)outranges_size, &retbytes, NULL); + if (ret == 0 && GetLastError() == ERROR_MORE_DATA) { + free(outranges); + outranges_size *= 2; + outranges = (FILE_ALLOCATED_RANGE_BUFFER *) + malloc(outranges_size); + if (outranges == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Couldn't allocate memory"); + exit_sts = ARCHIVE_FATAL; + goto exit_setup_sparse; + } + continue; + } else + break; + } + if (ret != 0) { + if (retbytes > 0) { + DWORD i, n; + + n = retbytes / sizeof(outranges[0]); + if (n == 1 && + outranges[0].FileOffset.QuadPart == 0 && + outranges[0].Length.QuadPart == entry_size) + break;/* This is not sparse. */ + for (i = 0; i < n; i++) + archive_entry_sparse_add_entry(entry, + outranges[i].FileOffset.QuadPart, + outranges[i].Length.QuadPart); + range.FileOffset.QuadPart = + outranges[n-1].FileOffset.QuadPart + + outranges[n-1].Length.QuadPart; + range.Length.QuadPart = + entry_size - range.FileOffset.QuadPart; + if (range.Length.QuadPart > 0) + continue; + } else { + /* The entire file is a hole. Add one data block of size 0 at the end. */ + archive_entry_sparse_add_entry(entry, + entry_size, + 0); + } + break; + } else { + la_dosmaperr(GetLastError()); + archive_set_error(&a->archive, errno, + "DeviceIoControl Failed: %lu", GetLastError()); + exit_sts = ARCHIVE_FAILED; + goto exit_setup_sparse; + } + } +exit_setup_sparse: + free(outranges); + + return (exit_sts); +} + +#endif diff --git a/c/archive_read_extract.c b/c/archive_read_extract.c new file mode 100644 --- /dev/null +++ b/c/archive_read_extract.c @@ -0,0 +1,60 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: src/lib/libarchive/archive_read_extract.c,v 1.61 2008/05/26 17:00:22 kientzle Exp $"); + +#ifdef HAVE_ERRNO_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_read_private.h" + +int +archive_read_extract(struct archive *_a, struct archive_entry *entry, int flags) +{ + struct archive_read_extract *extract; + struct archive_read * a = (struct archive_read *)_a; + + extract = __archive_read_get_extract(a); + if (extract == NULL) + return (ARCHIVE_FATAL); + + /* If we haven't initialized the archive_write_disk object, do it now. */ + if (extract->ad == NULL) { + extract->ad = archive_write_disk_new(); + if (extract->ad == NULL) { + archive_set_error(&a->archive, ENOMEM, "Can't extract"); + return (ARCHIVE_FATAL); + } + archive_write_disk_set_standard_lookup(extract->ad); + } + + archive_write_disk_set_options(extract->ad, flags); + return (archive_read_extract2(&a->archive, entry, extract->ad)); +} diff --git a/c/archive_read_extract2.c b/c/archive_read_extract2.c new file mode 100644 --- /dev/null +++ b/c/archive_read_extract2.c @@ -0,0 +1,155 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: src/lib/libarchive/archive_read_extract.c,v 1.61 2008/05/26 17:00:22 kientzle Exp $"); + +#ifdef HAVE_SYS_TYPES_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_read_private.h" + +static int copy_data(struct archive *ar, struct archive *aw); +static int archive_read_extract_cleanup(struct archive_read *); + + +/* Retrieve an extract object without initialising the associated + * archive_write_disk object. + */ +struct archive_read_extract * +__archive_read_get_extract(struct archive_read *a) +{ + if (a->extract == NULL) { + a->extract = (struct archive_read_extract *)calloc(1, sizeof(*a->extract)); + if (a->extract == NULL) { + archive_set_error(&a->archive, ENOMEM, "Can't extract"); + return (NULL); + } + a->cleanup_archive_extract = archive_read_extract_cleanup; + } + return (a->extract); +} + +/* + * Cleanup function for archive_extract. + */ +static int +archive_read_extract_cleanup(struct archive_read *a) +{ + int ret = ARCHIVE_OK; + + if (a->extract->ad != NULL) { + ret = archive_write_free(a->extract->ad); + } + free(a->extract); + a->extract = NULL; + return (ret); +} + +int +archive_read_extract2(struct archive *_a, struct archive_entry *entry, + struct archive *ad) +{ + struct archive_read *a = (struct archive_read *)_a; + int r, r2; + + /* Set up for this particular entry. */ + if (a->skip_file_set) + archive_write_disk_set_skip_file(ad, + a->skip_file_dev, a->skip_file_ino); + r = archive_write_header(ad, entry); + if (r < ARCHIVE_WARN) + r = ARCHIVE_WARN; + if (r != ARCHIVE_OK) + /* If _write_header failed, copy the error. */ + archive_copy_error(&a->archive, ad); + else if (!archive_entry_size_is_set(entry) || archive_entry_size(entry) > 0) + /* Otherwise, pour data into the entry. */ + r = copy_data(_a, ad); + r2 = archive_write_finish_entry(ad); + if (r2 < ARCHIVE_WARN) + r2 = ARCHIVE_WARN; + /* Use the first message. */ + if (r2 != ARCHIVE_OK && r == ARCHIVE_OK) + archive_copy_error(&a->archive, ad); + /* Use the worst error return. */ + if (r2 < r) + r = r2; + return (r); +} + +void +archive_read_extract_set_progress_callback(struct archive *_a, + void (*progress_func)(void *), void *user_data) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_extract *extract = __archive_read_get_extract(a); + if (extract != NULL) { + extract->extract_progress = progress_func; + extract->extract_progress_user_data = user_data; + } +} + +static int +copy_data(struct archive *ar, struct archive *aw) +{ + int64_t offset; + const void *buff; + struct archive_read_extract *extract; + size_t size; + int r; + + extract = __archive_read_get_extract((struct archive_read *)ar); + if (extract == NULL) + return (ARCHIVE_FATAL); + for (;;) { + r = archive_read_data_block(ar, &buff, &size, &offset); + if (r == ARCHIVE_EOF) + return (ARCHIVE_OK); + if (r != ARCHIVE_OK) + return (r); + r = (int)archive_write_data_block(aw, buff, size, offset); + if (r < ARCHIVE_WARN) + r = ARCHIVE_WARN; + if (r < ARCHIVE_OK) { + archive_set_error(ar, archive_errno(aw), + "%s", archive_error_string(aw)); + return (r); + } + if (extract->extract_progress) + (extract->extract_progress) + (extract->extract_progress_user_data); + } +} diff --git a/c/archive_read_open_fd.c b/c/archive_read_open_fd.c new file mode 100644 --- /dev/null +++ b/c/archive_read_open_fd.c @@ -0,0 +1,211 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_read_open_fd.c 201103 2009-12-28 03:13:49Z kientzle $"); + +#ifdef HAVE_SYS_STAT_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_FCNTL_H +#include +#endif +#ifdef HAVE_IO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif + +#include "archive.h" + +struct read_fd_data { + int fd; + size_t block_size; + char use_lseek; + void *buffer; +}; + +static int file_close(struct archive *, void *); +static ssize_t file_read(struct archive *, void *, const void **buff); +static int64_t file_seek(struct archive *, void *, int64_t request, int); +static int64_t file_skip(struct archive *, void *, int64_t request); + +int +archive_read_open_fd(struct archive *a, int fd, size_t block_size) +{ + struct stat st; + struct read_fd_data *mine; + void *b; + + archive_clear_error(a); + if (fstat(fd, &st) != 0) { + archive_set_error(a, errno, "Can't stat fd %d", fd); + return (ARCHIVE_FATAL); + } + + mine = (struct read_fd_data *)calloc(1, sizeof(*mine)); + b = malloc(block_size); + if (mine == NULL || b == NULL) { + archive_set_error(a, ENOMEM, "No memory"); + free(mine); + free(b); + return (ARCHIVE_FATAL); + } + mine->block_size = block_size; + mine->buffer = b; + mine->fd = fd; + /* + * Skip support is a performance optimization for anything + * that supports lseek(). On FreeBSD, only regular files and + * raw disk devices support lseek() and there's no portable + * way to determine if a device is a raw disk device, so we + * only enable this optimization for regular files. + */ + if (S_ISREG(st.st_mode)) { + archive_read_extract_set_skip_file(a, st.st_dev, st.st_ino); + mine->use_lseek = 1; + } +#if defined(__CYGWIN__) || defined(_WIN32) + setmode(mine->fd, O_BINARY); +#endif + + archive_read_set_read_callback(a, file_read); + archive_read_set_skip_callback(a, file_skip); + archive_read_set_seek_callback(a, file_seek); + archive_read_set_close_callback(a, file_close); + archive_read_set_callback_data(a, mine); + return (archive_read_open1(a)); +} + +static ssize_t +file_read(struct archive *a, void *client_data, const void **buff) +{ + struct read_fd_data *mine = (struct read_fd_data *)client_data; + ssize_t bytes_read; + + *buff = mine->buffer; + for (;;) { + bytes_read = read(mine->fd, mine->buffer, mine->block_size); + if (bytes_read < 0) { + if (errno == EINTR) + continue; + archive_set_error(a, errno, "Error reading fd %d", + mine->fd); + } + return (bytes_read); + } +} + +static int64_t +file_skip(struct archive *a, void *client_data, int64_t request) +{ + struct read_fd_data *mine = (struct read_fd_data *)client_data; + int64_t skip = request; + int64_t old_offset, new_offset; + int skip_bits = sizeof(skip) * 8 - 1; /* off_t is a signed type. */ + + if (!mine->use_lseek) + return (0); + + /* Reduce a request that would overflow the 'skip' variable. */ + if (sizeof(request) > sizeof(skip)) { + int64_t max_skip = + (((int64_t)1 << (skip_bits - 1)) - 1) * 2 + 1; + if (request > max_skip) + skip = max_skip; + } + + /* Reduce request to the next smallest multiple of block_size */ + request = (request / mine->block_size) * mine->block_size; + if (request == 0) + return (0); + + if (((old_offset = lseek(mine->fd, 0, SEEK_CUR)) >= 0) && + ((new_offset = lseek(mine->fd, skip, SEEK_CUR)) >= 0)) + return (new_offset - old_offset); + + /* If seek failed once, it will probably fail again. */ + mine->use_lseek = 0; + + /* Let libarchive recover with read+discard. */ + if (errno == ESPIPE) + return (0); + + /* + * There's been an error other than ESPIPE. This is most + * likely caused by a programmer error (too large request) + * or a corrupted archive file. + */ + archive_set_error(a, errno, "Error seeking"); + return (-1); +} + +/* + * TODO: Store the offset and use it in the read callback. + */ +static int64_t +file_seek(struct archive *a, void *client_data, int64_t request, int whence) +{ + struct read_fd_data *mine = (struct read_fd_data *)client_data; + int64_t r; + + /* We use off_t here because lseek() is declared that way. */ + /* See above for notes about when off_t is less than 64 bits. */ + r = lseek(mine->fd, request, whence); + if (r >= 0) + return r; + + if (errno == ESPIPE) { + archive_set_error(a, errno, + "A file descriptor(%d) is not seekable(PIPE)", mine->fd); + return (ARCHIVE_FAILED); + } else { + /* If the input is corrupted or truncated, fail. */ + archive_set_error(a, errno, + "Error seeking in a file descriptor(%d)", mine->fd); + return (ARCHIVE_FATAL); + } +} + +static int +file_close(struct archive *a, void *client_data) +{ + struct read_fd_data *mine = (struct read_fd_data *)client_data; + + (void)a; /* UNUSED */ + free(mine->buffer); + free(mine); + return (ARCHIVE_OK); +} diff --git a/c/archive_read_open_file.c b/c/archive_read_open_file.c new file mode 100644 --- /dev/null +++ b/c/archive_read_open_file.c @@ -0,0 +1,180 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_read_open_file.c 201093 2009-12-28 02:28:44Z kientzle $"); + +#ifdef HAVE_SYS_STAT_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_FCNTL_H +#include +#endif +#ifdef HAVE_IO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif + +#include "archive.h" + +struct read_FILE_data { + FILE *f; + size_t block_size; + void *buffer; + char can_skip; +}; + +static int file_close(struct archive *, void *); +static ssize_t file_read(struct archive *, void *, const void **buff); +static int64_t file_skip(struct archive *, void *, int64_t request); + +int +archive_read_open_FILE(struct archive *a, FILE *f) +{ + struct stat st; + struct read_FILE_data *mine; + size_t block_size = 128 * 1024; + void *b; + + archive_clear_error(a); + mine = (struct read_FILE_data *)malloc(sizeof(*mine)); + b = malloc(block_size); + if (mine == NULL || b == NULL) { + archive_set_error(a, ENOMEM, "No memory"); + free(mine); + free(b); + return (ARCHIVE_FATAL); + } + mine->block_size = block_size; + mine->buffer = b; + mine->f = f; + /* + * If we can't fstat() the file, it may just be that it's not + * a file. (On some platforms, FILE * objects can wrap I/O + * streams that don't support fileno()). As a result, fileno() + * should be used cautiously.) + */ + if (fstat(fileno(mine->f), &st) == 0 && S_ISREG(st.st_mode)) { + archive_read_extract_set_skip_file(a, st.st_dev, st.st_ino); + /* Enable the seek optimization only for regular files. */ + mine->can_skip = 1; + } else + mine->can_skip = 0; + +#if defined(__CYGWIN__) || defined(_WIN32) + setmode(fileno(mine->f), O_BINARY); +#endif + + archive_read_set_read_callback(a, file_read); + archive_read_set_skip_callback(a, file_skip); + archive_read_set_close_callback(a, file_close); + archive_read_set_callback_data(a, mine); + return (archive_read_open1(a)); +} + +static ssize_t +file_read(struct archive *a, void *client_data, const void **buff) +{ + struct read_FILE_data *mine = (struct read_FILE_data *)client_data; + size_t bytes_read; + + *buff = mine->buffer; + bytes_read = fread(mine->buffer, 1, mine->block_size, mine->f); + if (bytes_read < mine->block_size && ferror(mine->f)) { + archive_set_error(a, errno, "Error reading file"); + } + return (bytes_read); +} + +static int64_t +file_skip(struct archive *a, void *client_data, int64_t request) +{ + struct read_FILE_data *mine = (struct read_FILE_data *)client_data; +#if HAVE_FSEEKO + off_t skip = (off_t)request; +#elif HAVE__FSEEKI64 + int64_t skip = request; +#else + long skip = (long)request; +#endif + int skip_bits = sizeof(skip) * 8 - 1; + + (void)a; /* UNUSED */ + + /* + * If we can't skip, return 0 as the amount we did step and + * the caller will work around by reading and discarding. + */ + if (!mine->can_skip) + return (0); + if (request == 0) + return (0); + + /* If request is too big for a long or an off_t, reduce it. */ + if (sizeof(request) > sizeof(skip)) { + int64_t max_skip = + (((int64_t)1 << (skip_bits - 1)) - 1) * 2 + 1; + if (request > max_skip) + skip = max_skip; + } + +#ifdef __ANDROID__ + /* fileno() isn't safe on all platforms ... see above. */ + if (lseek(fileno(mine->f), skip, SEEK_CUR) < 0) +#elif HAVE_FSEEKO + if (fseeko(mine->f, skip, SEEK_CUR) != 0) +#elif HAVE__FSEEKI64 + if (_fseeki64(mine->f, skip, SEEK_CUR) != 0) +#else + if (fseek(mine->f, skip, SEEK_CUR) != 0) +#endif + { + mine->can_skip = 0; + return (0); + } + return (request); +} + +static int +file_close(struct archive *a, void *client_data) +{ + struct read_FILE_data *mine = (struct read_FILE_data *)client_data; + + (void)a; /* UNUSED */ + free(mine->buffer); + free(mine); + return (ARCHIVE_OK); +} diff --git a/c/archive_read_open_filename.c b/c/archive_read_open_filename.c new file mode 100644 --- /dev/null +++ b/c/archive_read_open_filename.c @@ -0,0 +1,586 @@ +/*- + * Copyright (c) 2003-2010 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_read_open_filename.c 201093 2009-12-28 02:28:44Z kientzle $"); + +#ifdef HAVE_SYS_IOCTL_H +#include +#endif +#ifdef HAVE_SYS_STAT_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_FCNTL_H +#include +#endif +#ifdef HAVE_IO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif +#if defined(__FreeBSD__) || defined(__FreeBSD_kernel__) +#include +#elif defined(__NetBSD__) || defined(__OpenBSD__) +#include +#include +#elif defined(__DragonFly__) +#include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_string.h" + +#ifndef O_BINARY +#define O_BINARY 0 +#endif +#ifndef O_CLOEXEC +#define O_CLOEXEC 0 +#endif + +struct read_file_data { + int fd; + size_t block_size; + void *buffer; + mode_t st_mode; /* Mode bits for opened file. */ + char use_lseek; + enum fnt_e { FNT_STDIN, FNT_MBS, FNT_WCS } filename_type; + union { + char m[1];/* MBS filename. */ + wchar_t w[1];/* WCS filename. */ + } filename; /* Must be last! */ +}; + +static int file_open(struct archive *, void *); +static int file_close(struct archive *, void *); +static int file_close2(struct archive *, void *); +static int file_switch(struct archive *, void *, void *); +static ssize_t file_read(struct archive *, void *, const void **buff); +static int64_t file_seek(struct archive *, void *, int64_t request, int); +static int64_t file_skip(struct archive *, void *, int64_t request); +static int64_t file_skip_lseek(struct archive *, void *, int64_t request); + +int +archive_read_open_file(struct archive *a, const char *filename, + size_t block_size) +{ + return (archive_read_open_filename(a, filename, block_size)); +} + +int +archive_read_open_filename(struct archive *a, const char *filename, + size_t block_size) +{ + const char *filenames[2]; + filenames[0] = filename; + filenames[1] = NULL; + return archive_read_open_filenames(a, filenames, block_size); +} + +int +archive_read_open_filenames(struct archive *a, const char **filenames, + size_t block_size) +{ + struct read_file_data *mine; + const char *filename = NULL; + if (filenames) + filename = *(filenames++); + + archive_clear_error(a); + do + { + if (filename == NULL) + filename = ""; + mine = (struct read_file_data *)calloc(1, + sizeof(*mine) + strlen(filename)); + if (mine == NULL) + goto no_memory; + strcpy(mine->filename.m, filename); + mine->block_size = block_size; + mine->fd = -1; + mine->buffer = NULL; + mine->st_mode = mine->use_lseek = 0; + if (filename == NULL || filename[0] == '\0') { + mine->filename_type = FNT_STDIN; + } else + mine->filename_type = FNT_MBS; + if (archive_read_append_callback_data(a, mine) != (ARCHIVE_OK)) + return (ARCHIVE_FATAL); + if (filenames == NULL) + break; + filename = *(filenames++); + } while (filename != NULL && filename[0] != '\0'); + archive_read_set_open_callback(a, file_open); + archive_read_set_read_callback(a, file_read); + archive_read_set_skip_callback(a, file_skip); + archive_read_set_close_callback(a, file_close); + archive_read_set_switch_callback(a, file_switch); + archive_read_set_seek_callback(a, file_seek); + + return (archive_read_open1(a)); +no_memory: + archive_set_error(a, ENOMEM, "No memory"); + return (ARCHIVE_FATAL); +} + +int +archive_read_open_filename_w(struct archive *a, const wchar_t *wfilename, + size_t block_size) +{ + struct read_file_data *mine = (struct read_file_data *)calloc(1, + sizeof(*mine) + wcslen(wfilename) * sizeof(wchar_t)); + if (!mine) + { + archive_set_error(a, ENOMEM, "No memory"); + return (ARCHIVE_FATAL); + } + mine->fd = -1; + mine->block_size = block_size; + + if (wfilename == NULL || wfilename[0] == L'\0') { + mine->filename_type = FNT_STDIN; + } else { +#if defined(_WIN32) && !defined(__CYGWIN__) + mine->filename_type = FNT_WCS; + wcscpy(mine->filename.w, wfilename); +#else + /* + * POSIX system does not support a wchar_t interface for + * open() system call, so we have to translate a wchar_t + * filename to multi-byte one and use it. + */ + struct archive_string fn; + + archive_string_init(&fn); + if (archive_string_append_from_wcs(&fn, wfilename, + wcslen(wfilename)) != 0) { + if (errno == ENOMEM) + archive_set_error(a, errno, + "Can't allocate memory"); + else + archive_set_error(a, EINVAL, + "Failed to convert a wide-character" + " filename to a multi-byte filename"); + archive_string_free(&fn); + free(mine); + return (ARCHIVE_FATAL); + } + mine->filename_type = FNT_MBS; + strcpy(mine->filename.m, fn.s); + archive_string_free(&fn); +#endif + } + if (archive_read_append_callback_data(a, mine) != (ARCHIVE_OK)) + return (ARCHIVE_FATAL); + archive_read_set_open_callback(a, file_open); + archive_read_set_read_callback(a, file_read); + archive_read_set_skip_callback(a, file_skip); + archive_read_set_close_callback(a, file_close); + archive_read_set_switch_callback(a, file_switch); + archive_read_set_seek_callback(a, file_seek); + + return (archive_read_open1(a)); +} + +static int +file_open(struct archive *a, void *client_data) +{ + struct stat st; + struct read_file_data *mine = (struct read_file_data *)client_data; + void *buffer; + const char *filename = NULL; +#if defined(_WIN32) && !defined(__CYGWIN__) + const wchar_t *wfilename = NULL; +#endif + int fd = -1; + int is_disk_like = 0; +#if defined(__FreeBSD__) || defined(__FreeBSD_kernel__) + off_t mediasize = 0; /* FreeBSD-specific, so off_t okay here. */ +#elif defined(__NetBSD__) || defined(__OpenBSD__) + struct disklabel dl; +#elif defined(__DragonFly__) + struct partinfo pi; +#endif + + archive_clear_error(a); + if (mine->filename_type == FNT_STDIN) { + /* We used to delegate stdin support by + * directly calling archive_read_open_fd(a,0,block_size) + * here, but that doesn't (and shouldn't) handle the + * end-of-file flush when reading stdout from a pipe. + * Basically, read_open_fd() is intended for folks who + * are willing to handle such details themselves. This + * API is intended to be a little smarter for folks who + * want easy handling of the common case. + */ + fd = 0; +#if defined(__CYGWIN__) || defined(_WIN32) + setmode(0, O_BINARY); +#endif + filename = ""; + } else if (mine->filename_type == FNT_MBS) { + filename = mine->filename.m; + fd = open(filename, O_RDONLY | O_BINARY | O_CLOEXEC); + __archive_ensure_cloexec_flag(fd); + if (fd < 0) { + archive_set_error(a, errno, + "Failed to open '%s'", filename); + return (ARCHIVE_FATAL); + } + } else { +#if defined(_WIN32) && !defined(__CYGWIN__) + wfilename = mine->filename.w; + fd = _wopen(wfilename, O_RDONLY | O_BINARY); + if (fd < 0 && errno == ENOENT) { + wchar_t *fullpath; + fullpath = __la_win_permissive_name_w(wfilename); + if (fullpath != NULL) { + fd = _wopen(fullpath, O_RDONLY | O_BINARY); + free(fullpath); + } + } + if (fd < 0) { + archive_set_error(a, errno, + "Failed to open '%S'", wfilename); + return (ARCHIVE_FATAL); + } +#else + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Unexpedted operation in archive_read_open_filename"); + goto fail; +#endif + } + if (fstat(fd, &st) != 0) { +#if defined(_WIN32) && !defined(__CYGWIN__) + if (mine->filename_type == FNT_WCS) + archive_set_error(a, errno, "Can't stat '%S'", + wfilename); + else +#endif + archive_set_error(a, errno, "Can't stat '%s'", + filename); + goto fail; + } + + /* + * Determine whether the input looks like a disk device or a + * tape device. The results are used below to select an I/O + * strategy: + * = "disk-like" devices support arbitrary lseek() and will + * support I/O requests of any size. So we get easy skipping + * and can cheat on block sizes to get better performance. + * = "tape-like" devices require strict blocking and use + * specialized ioctls for seeking. + * = "socket-like" devices cannot seek at all but can improve + * performance by using nonblocking I/O to read "whatever is + * available right now". + * + * Right now, we only specially recognize disk-like devices, + * but it should be straightforward to add probes and strategy + * here for tape-like and socket-like devices. + */ + if (S_ISREG(st.st_mode)) { + /* Safety: Tell the extractor not to overwrite the input. */ + archive_read_extract_set_skip_file(a, st.st_dev, st.st_ino); + /* Regular files act like disks. */ + is_disk_like = 1; + } +#if defined(__FreeBSD__) || defined(__FreeBSD_kernel__) + /* FreeBSD: if it supports DIOCGMEDIASIZE ioctl, it's disk-like. */ + else if (S_ISCHR(st.st_mode) && + ioctl(fd, DIOCGMEDIASIZE, &mediasize) == 0 && + mediasize > 0) { + is_disk_like = 1; + } +#elif defined(__NetBSD__) || defined(__OpenBSD__) + /* Net/OpenBSD: if it supports DIOCGDINFO ioctl, it's disk-like. */ + else if ((S_ISCHR(st.st_mode) || S_ISBLK(st.st_mode)) && + ioctl(fd, DIOCGDINFO, &dl) == 0 && + dl.d_partitions[DISKPART(st.st_rdev)].p_size > 0) { + is_disk_like = 1; + } +#elif defined(__DragonFly__) + /* DragonFly BSD: if it supports DIOCGPART ioctl, it's disk-like. */ + else if (S_ISCHR(st.st_mode) && + ioctl(fd, DIOCGPART, &pi) == 0 && + pi.media_size > 0) { + is_disk_like = 1; + } +#elif defined(__linux__) + /* Linux: All block devices are disk-like. */ + else if (S_ISBLK(st.st_mode) && + lseek(fd, 0, SEEK_CUR) == 0 && + lseek(fd, 0, SEEK_SET) == 0 && + lseek(fd, 0, SEEK_END) > 0 && + lseek(fd, 0, SEEK_SET) == 0) { + is_disk_like = 1; + } +#endif + /* TODO: Add an "is_tape_like" variable and appropriate tests. */ + + /* Disk-like devices prefer power-of-two block sizes. */ + /* Use provided block_size as a guide so users have some control. */ + if (is_disk_like) { + size_t new_block_size = 64 * 1024; + while (new_block_size < mine->block_size + && new_block_size < 64 * 1024 * 1024) + new_block_size *= 2; + mine->block_size = new_block_size; + } + buffer = malloc(mine->block_size); + if (buffer == NULL) { + archive_set_error(a, ENOMEM, "No memory"); + goto fail; + } + mine->buffer = buffer; + mine->fd = fd; + /* Remember mode so close can decide whether to flush. */ + mine->st_mode = st.st_mode; + + /* Disk-like inputs can use lseek(). */ + if (is_disk_like) + mine->use_lseek = 1; + + return (ARCHIVE_OK); +fail: + /* + * Don't close file descriptors not opened or ones pointing referring + * to `FNT_STDIN`. + */ + if (fd != -1 && fd != 0) + close(fd); + return (ARCHIVE_FATAL); +} + +static ssize_t +file_read(struct archive *a, void *client_data, const void **buff) +{ + struct read_file_data *mine = (struct read_file_data *)client_data; + ssize_t bytes_read; + + /* TODO: If a recent lseek() operation has left us + * mis-aligned, read and return a short block to try to get + * us back in alignment. */ + + /* TODO: Someday, try mmap() here; if that succeeds, give + * the entire file to libarchive as a single block. That + * could be a lot faster than block-by-block manual I/O. */ + + /* TODO: We might be able to improve performance on pipes and + * sockets by setting non-blocking I/O and just accepting + * whatever we get here instead of waiting for a full block + * worth of data. */ + + *buff = mine->buffer; + for (;;) { + bytes_read = read(mine->fd, mine->buffer, mine->block_size); + if (bytes_read < 0) { + if (errno == EINTR) + continue; + else if (mine->filename_type == FNT_STDIN) + archive_set_error(a, errno, + "Error reading stdin"); + else if (mine->filename_type == FNT_MBS) + archive_set_error(a, errno, + "Error reading '%s'", mine->filename.m); + else + archive_set_error(a, errno, + "Error reading '%S'", mine->filename.w); + } + return (bytes_read); + } +} + +/* + * Regular files and disk-like block devices can use simple lseek + * without needing to round the request to the block size. + * + * TODO: This can leave future reads mis-aligned. Since we know the + * offset here, we should store it and use it in file_read() above + * to determine whether we should perform a short read to get back + * into alignment. Long series of mis-aligned reads can negatively + * impact disk throughput. (Of course, the performance impact should + * be carefully tested; extra code complexity is only worthwhile if + * it does provide measurable improvement.) + * + * TODO: Be lazy about the actual seek. There are a few pathological + * cases where libarchive makes a bunch of seek requests in a row + * without any intervening reads. This isn't a huge performance + * problem, since the kernel handles seeks lazily already, but + * it would be very slightly faster if we simply remembered the + * seek request here and then actually performed the seek at the + * top of the read callback above. + */ +static int64_t +file_skip_lseek(struct archive *a, void *client_data, int64_t request) +{ + struct read_file_data *mine = (struct read_file_data *)client_data; +#if defined(_WIN32) && !defined(__CYGWIN__) + /* We use _lseeki64() on Windows. */ + int64_t old_offset, new_offset; +#else + off_t old_offset, new_offset; +#endif + + /* We use off_t here because lseek() is declared that way. */ + + /* TODO: Deal with case where off_t isn't 64 bits. + * This shouldn't be a problem on Linux or other POSIX + * systems, since the configuration logic for libarchive + * tries to obtain a 64-bit off_t. + */ + if ((old_offset = lseek(mine->fd, 0, SEEK_CUR)) >= 0 && + (new_offset = lseek(mine->fd, request, SEEK_CUR)) >= 0) + return (new_offset - old_offset); + + /* If lseek() fails, don't bother trying again. */ + mine->use_lseek = 0; + + /* Let libarchive recover with read+discard */ + if (errno == ESPIPE) + return (0); + + /* If the input is corrupted or truncated, fail. */ + if (mine->filename_type == FNT_STDIN) + archive_set_error(a, errno, "Error seeking in stdin"); + else if (mine->filename_type == FNT_MBS) + archive_set_error(a, errno, "Error seeking in '%s'", + mine->filename.m); + else + archive_set_error(a, errno, "Error seeking in '%S'", + mine->filename.w); + return (-1); +} + + +/* + * TODO: Implement another file_skip_XXXX that uses MTIO ioctls to + * accelerate operation on tape drives. + */ + +static int64_t +file_skip(struct archive *a, void *client_data, int64_t request) +{ + struct read_file_data *mine = (struct read_file_data *)client_data; + + /* Delegate skip requests. */ + if (mine->use_lseek) + return (file_skip_lseek(a, client_data, request)); + + /* If we can't skip, return 0; libarchive will read+discard instead. */ + return (0); +} + +/* + * TODO: Store the offset and use it in the read callback. + */ +static int64_t +file_seek(struct archive *a, void *client_data, int64_t request, int whence) +{ + struct read_file_data *mine = (struct read_file_data *)client_data; + int64_t r; + + /* We use off_t here because lseek() is declared that way. */ + /* See above for notes about when off_t is less than 64 bits. */ + r = lseek(mine->fd, request, whence); + if (r >= 0) + return r; + + /* If the input is corrupted or truncated, fail. */ + if (mine->filename_type == FNT_STDIN) + archive_set_error(a, errno, "Error seeking in stdin"); + else if (mine->filename_type == FNT_MBS) + archive_set_error(a, errno, "Error seeking in '%s'", + mine->filename.m); + else + archive_set_error(a, errno, "Error seeking in '%S'", + mine->filename.w); + return (ARCHIVE_FATAL); +} + +static int +file_close2(struct archive *a, void *client_data) +{ + struct read_file_data *mine = (struct read_file_data *)client_data; + + (void)a; /* UNUSED */ + + /* Only flush and close if open succeeded. */ + if (mine->fd >= 0) { + /* + * Sometimes, we should flush the input before closing. + * Regular files: faster to just close without flush. + * Disk-like devices: Ditto. + * Tapes: must not flush (user might need to + * read the "next" item on a non-rewind device). + * Pipes and sockets: must flush (otherwise, the + * program feeding the pipe or socket may complain). + * Here, I flush everything except for regular files and + * device nodes. + */ + if (!S_ISREG(mine->st_mode) + && !S_ISCHR(mine->st_mode) + && !S_ISBLK(mine->st_mode)) { + ssize_t bytesRead; + do { + bytesRead = read(mine->fd, mine->buffer, + mine->block_size); + } while (bytesRead > 0); + } + /* If a named file was opened, then it needs to be closed. */ + if (mine->filename_type != FNT_STDIN) + close(mine->fd); + } + free(mine->buffer); + mine->buffer = NULL; + mine->fd = -1; + return (ARCHIVE_OK); +} + +static int +file_close(struct archive *a, void *client_data) +{ + struct read_file_data *mine = (struct read_file_data *)client_data; + file_close2(a, client_data); + free(mine); + return (ARCHIVE_OK); +} + +static int +file_switch(struct archive *a, void *client_data1, void *client_data2) +{ + file_close2(a, client_data1); + return file_open(a, client_data2); +} diff --git a/c/archive_read_open_memory.c b/c/archive_read_open_memory.c new file mode 100644 --- /dev/null +++ b/c/archive_read_open_memory.c @@ -0,0 +1,186 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: src/lib/libarchive/archive_read_open_memory.c,v 1.6 2007/07/06 15:51:59 kientzle Exp $"); + +#include +#include +#include + +#include "archive.h" + +/* + * Glue to read an archive from a block of memory. + * + * This is mostly a huge help in building test harnesses; + * test programs can build archives in memory and read them + * back again without having to mess with files on disk. + */ + +struct read_memory_data { + const unsigned char *start; + const unsigned char *p; + const unsigned char *end; + ssize_t read_size; +}; + +static int memory_read_close(struct archive *, void *); +static int memory_read_open(struct archive *, void *); +static int64_t memory_read_seek(struct archive *, void *, int64_t offset, int whence); +static int64_t memory_read_skip(struct archive *, void *, int64_t request); +static ssize_t memory_read(struct archive *, void *, const void **buff); + +int +archive_read_open_memory(struct archive *a, const void *buff, size_t size) +{ + return archive_read_open_memory2(a, buff, size, size); +} + +/* + * Don't use _open_memory2() in production code; the archive_read_open_memory() + * version is the one you really want. This is just here so that + * test harnesses can exercise block operations inside the library. + */ +int +archive_read_open_memory2(struct archive *a, const void *buff, + size_t size, size_t read_size) +{ + struct read_memory_data *mine; + + mine = (struct read_memory_data *)calloc(1, sizeof(*mine)); + if (mine == NULL) { + archive_set_error(a, ENOMEM, "No memory"); + return (ARCHIVE_FATAL); + } + mine->start = mine->p = (const unsigned char *)buff; + mine->end = mine->start + size; + mine->read_size = read_size; + archive_read_set_open_callback(a, memory_read_open); + archive_read_set_read_callback(a, memory_read); + archive_read_set_seek_callback(a, memory_read_seek); + archive_read_set_skip_callback(a, memory_read_skip); + archive_read_set_close_callback(a, memory_read_close); + archive_read_set_callback_data(a, mine); + return (archive_read_open1(a)); +} + +/* + * There's nothing to open. + */ +static int +memory_read_open(struct archive *a, void *client_data) +{ + (void)a; /* UNUSED */ + (void)client_data; /* UNUSED */ + return (ARCHIVE_OK); +} + +/* + * This is scary simple: Just advance a pointer. Limiting + * to read_size is not technically necessary, but it exercises + * more of the internal logic when used with a small block size + * in a test harness. Production use should not specify a block + * size; then this is much faster. + */ +static ssize_t +memory_read(struct archive *a, void *client_data, const void **buff) +{ + struct read_memory_data *mine = (struct read_memory_data *)client_data; + ssize_t size; + + (void)a; /* UNUSED */ + *buff = mine->p; + size = mine->end - mine->p; + if (size > mine->read_size) + size = mine->read_size; + mine->p += size; + return (size); +} + +/* + * Advancing is just as simple. Again, this is doing more than + * necessary in order to better exercise internal code when used + * as a test harness. + */ +static int64_t +memory_read_skip(struct archive *a, void *client_data, int64_t skip) +{ + struct read_memory_data *mine = (struct read_memory_data *)client_data; + + (void)a; /* UNUSED */ + if ((int64_t)skip > (int64_t)(mine->end - mine->p)) + skip = mine->end - mine->p; + /* Round down to block size. */ + skip /= mine->read_size; + skip *= mine->read_size; + mine->p += skip; + return (skip); +} + +/* + * Seeking. + */ +static int64_t +memory_read_seek(struct archive *a, void *client_data, int64_t offset, int whence) +{ + struct read_memory_data *mine = (struct read_memory_data *)client_data; + + (void)a; /* UNUSED */ + switch (whence) { + case SEEK_SET: + mine->p = mine->start + offset; + break; + case SEEK_CUR: + mine->p += offset; + break; + case SEEK_END: + mine->p = mine->end + offset; + break; + default: + return ARCHIVE_FATAL; + } + if (mine->p < mine->start) { + mine->p = mine->start; + return ARCHIVE_FAILED; + } + if (mine->p > mine->end) { + mine->p = mine->end; + return ARCHIVE_FAILED; + } + return (mine->p - mine->start); +} + +/* + * Close is just cleaning up our one small bit of data. + */ +static int +memory_read_close(struct archive *a, void *client_data) +{ + struct read_memory_data *mine = (struct read_memory_data *)client_data; + (void)a; /* UNUSED */ + free(mine); + return (ARCHIVE_OK); +} diff --git a/c/archive_read_private.h b/c/archive_read_private.h new file mode 100644 --- /dev/null +++ b/c/archive_read_private.h @@ -0,0 +1,266 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD: head/lib/libarchive/archive_read_private.h 201088 2009-12-28 02:18:55Z kientzle $ + */ + +#ifndef ARCHIVE_READ_PRIVATE_H_INCLUDED +#define ARCHIVE_READ_PRIVATE_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#ifndef __LIBARCHIVE_TEST +#error This header is only to be used internally to libarchive. +#endif +#endif + +#include "archive.h" +#include "archive_string.h" +#include "archive_private.h" + +struct archive_read; +struct archive_read_filter_bidder; +struct archive_read_filter; + +/* + * How bidding works for filters: + * * The bid manager initializes the client-provided reader as the + * first filter. + * * It invokes the bidder for each registered filter with the + * current head filter. + * * The bidders can use archive_read_filter_ahead() to peek ahead + * at the incoming data to compose their bids. + * * The bid manager creates a new filter structure for the winning + * bidder and gives the winning bidder a chance to initialize it. + * * The new filter becomes the new top filter and we repeat the + * process. + * This ends only when no bidder provides a non-zero bid. Then + * we perform a similar dance with the registered format handlers. + */ +struct archive_read_filter_bidder { + /* Configuration data for the bidder. */ + void *data; + /* Name of the filter */ + const char *name; + /* Taste the upstream filter to see if we handle this. */ + int (*bid)(struct archive_read_filter_bidder *, + struct archive_read_filter *); + /* Initialize a newly-created filter. */ + int (*init)(struct archive_read_filter *); + /* Set an option for the filter bidder. */ + int (*options)(struct archive_read_filter_bidder *, + const char *key, const char *value); + /* Release the bidder's configuration data. */ + int (*free)(struct archive_read_filter_bidder *); +}; + +/* + * This structure is allocated within the archive_read core + * and initialized by archive_read and the init() method of the + * corresponding bidder above. + */ +struct archive_read_filter { + int64_t position; + /* Essentially all filters will need these values, so + * just declare them here. */ + struct archive_read_filter_bidder *bidder; /* My bidder. */ + struct archive_read_filter *upstream; /* Who I read from. */ + struct archive_read *archive; /* Associated archive. */ + /* Open a block for reading */ + int (*open)(struct archive_read_filter *self); + /* Return next block. */ + ssize_t (*read)(struct archive_read_filter *, const void **); + /* Skip forward this many bytes. */ + int64_t (*skip)(struct archive_read_filter *self, int64_t request); + /* Seek to an absolute location. */ + int64_t (*seek)(struct archive_read_filter *self, int64_t offset, int whence); + /* Close (just this filter) and free(self). */ + int (*close)(struct archive_read_filter *self); + /* Function that handles switching from reading one block to the next/prev */ + int (*sswitch)(struct archive_read_filter *self, unsigned int iindex); + /* Read any header metadata if available. */ + int (*read_header)(struct archive_read_filter *self, struct archive_entry *entry); + /* My private data. */ + void *data; + + const char *name; + int code; + + /* Used by reblocking logic. */ + char *buffer; + size_t buffer_size; + char *next; /* Current read location. */ + size_t avail; /* Bytes in my buffer. */ + const void *client_buff; /* Client buffer information. */ + size_t client_total; + const char *client_next; + size_t client_avail; + char end_of_file; + char closed; + char fatal; +}; + +/* + * The client looks a lot like a filter, so we just wrap it here. + * + * TODO: Make archive_read_filter and archive_read_client identical so + * that users of the library can easily register their own + * transformation filters. This will probably break the API/ABI and + * so should be deferred at least until libarchive 3.0. + */ +struct archive_read_data_node { + int64_t begin_position; + int64_t total_size; + void *data; +}; +struct archive_read_client { + archive_open_callback *opener; + archive_read_callback *reader; + archive_skip_callback *skipper; + archive_seek_callback *seeker; + archive_close_callback *closer; + archive_switch_callback *switcher; + unsigned int nodes; + unsigned int cursor; + int64_t position; + struct archive_read_data_node *dataset; +}; +struct archive_read_passphrase { + char *passphrase; + struct archive_read_passphrase *next; +}; + +struct archive_read_extract { + struct archive *ad; /* archive_write_disk object */ + + /* Progress function invoked during extract. */ + void (*extract_progress)(void *); + void *extract_progress_user_data; +}; + +struct archive_read { + struct archive archive; + + struct archive_entry *entry; + + /* Dev/ino of the archive being read/written. */ + int skip_file_set; + int64_t skip_file_dev; + int64_t skip_file_ino; + + /* Callbacks to open/read/write/close client archive streams. */ + struct archive_read_client client; + + /* Registered filter bidders. */ + struct archive_read_filter_bidder bidders[16]; + + /* Last filter in chain */ + struct archive_read_filter *filter; + + /* Whether to bypass filter bidding process */ + int bypass_filter_bidding; + + /* File offset of beginning of most recently-read header. */ + int64_t header_position; + + /* Nodes and offsets of compressed data block */ + unsigned int data_start_node; + unsigned int data_end_node; + + /* + * Format detection is mostly the same as compression + * detection, with one significant difference: The bidders + * use the read_ahead calls above to examine the stream rather + * than having the supervisor hand them a block of data to + * examine. + */ + + struct archive_format_descriptor { + void *data; + const char *name; + int (*bid)(struct archive_read *, int best_bid); + int (*options)(struct archive_read *, const char *key, + const char *value); + int (*read_header)(struct archive_read *, struct archive_entry *); + int (*read_data)(struct archive_read *, const void **, size_t *, int64_t *); + int (*read_data_skip)(struct archive_read *); + int64_t (*seek_data)(struct archive_read *, int64_t, int); + int (*cleanup)(struct archive_read *); + int (*format_capabilties)(struct archive_read *); + int (*has_encrypted_entries)(struct archive_read *); + } formats[16]; + struct archive_format_descriptor *format; /* Active format. */ + + /* + * Various information needed by archive_extract. + */ + struct archive_read_extract *extract; + int (*cleanup_archive_extract)(struct archive_read *); + + /* + * Decryption passphrase. + */ + struct { + struct archive_read_passphrase *first; + struct archive_read_passphrase **last; + int candidate; + archive_passphrase_callback *callback; + void *client_data; + } passphrases; +}; + +int __archive_read_register_format(struct archive_read *a, + void *format_data, + const char *name, + int (*bid)(struct archive_read *, int), + int (*options)(struct archive_read *, const char *, const char *), + int (*read_header)(struct archive_read *, struct archive_entry *), + int (*read_data)(struct archive_read *, const void **, size_t *, int64_t *), + int (*read_data_skip)(struct archive_read *), + int64_t (*seek_data)(struct archive_read *, int64_t, int), + int (*cleanup)(struct archive_read *), + int (*format_capabilities)(struct archive_read *), + int (*has_encrypted_entries)(struct archive_read *)); + +int __archive_read_get_bidder(struct archive_read *a, + struct archive_read_filter_bidder **bidder); + +const void *__archive_read_ahead(struct archive_read *, size_t, ssize_t *); +const void *__archive_read_filter_ahead(struct archive_read_filter *, + size_t, ssize_t *); +int64_t __archive_read_seek(struct archive_read*, int64_t, int); +int64_t __archive_read_filter_seek(struct archive_read_filter *, int64_t, int); +int64_t __archive_read_consume(struct archive_read *, int64_t); +int64_t __archive_read_filter_consume(struct archive_read_filter *, int64_t); +int __archive_read_header(struct archive_read *, struct archive_entry *); +int __archive_read_program(struct archive_read_filter *, const char *); +void __archive_read_free_filters(struct archive_read *); +struct archive_read_extract *__archive_read_get_extract(struct archive_read *); + + +/* + * Get a decryption passphrase. + */ +void __archive_read_reset_passphrase(struct archive_read *a); +const char * __archive_read_next_passphrase(struct archive_read *a); +#endif diff --git a/c/archive_read_set_format.c b/c/archive_read_set_format.c new file mode 100644 --- /dev/null +++ b/c/archive_read_set_format.c @@ -0,0 +1,108 @@ +/*- + * Copyright (c) 2003-2012 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_read_private.h" + +int +archive_read_set_format(struct archive *_a, int code) +{ + int r1, r2, slots, i; + char str[10]; + struct archive_read *a = (struct archive_read *)_a; + + if ((r1 = archive_read_support_format_by_code(_a, code)) < (ARCHIVE_OK)) + return r1; + + r1 = r2 = (ARCHIVE_OK); + if (a->format) + r2 = (ARCHIVE_WARN); + switch (code & ARCHIVE_FORMAT_BASE_MASK) + { + case ARCHIVE_FORMAT_7ZIP: + strcpy(str, "7zip"); + break; + case ARCHIVE_FORMAT_AR: + strcpy(str, "ar"); + break; + case ARCHIVE_FORMAT_CAB: + strcpy(str, "cab"); + break; + case ARCHIVE_FORMAT_CPIO: + strcpy(str, "cpio"); + break; + case ARCHIVE_FORMAT_ISO9660: + strcpy(str, "iso9660"); + break; + case ARCHIVE_FORMAT_LHA: + strcpy(str, "lha"); + break; + case ARCHIVE_FORMAT_MTREE: + strcpy(str, "mtree"); + break; + case ARCHIVE_FORMAT_RAR: + strcpy(str, "rar"); + break; + case ARCHIVE_FORMAT_RAR_V5: + strcpy(str, "rar5"); + break; + case ARCHIVE_FORMAT_TAR: + strcpy(str, "tar"); + break; + case ARCHIVE_FORMAT_XAR: + strcpy(str, "xar"); + break; + case ARCHIVE_FORMAT_ZIP: + strcpy(str, "zip"); + break; + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, + "Invalid format code specified"); + return (ARCHIVE_FATAL); + } + + slots = sizeof(a->formats) / sizeof(a->formats[0]); + a->format = &(a->formats[0]); + for (i = 0; i < slots; i++, a->format++) { + if (!a->format->name || !strcmp(a->format->name, str)) + break; + } + if (!a->format->name || strcmp(a->format->name, str)) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, + "Internal error: Unable to set format"); + r1 = (ARCHIVE_FATAL); + } + + return (r1 < r2) ? r1 : r2; +} diff --git a/c/archive_read_set_options.c b/c/archive_read_set_options.c new file mode 100644 --- /dev/null +++ b/c/archive_read_set_options.c @@ -0,0 +1,155 @@ +/*- + * Copyright (c) 2011 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#include "archive_read_private.h" +#include "archive_options_private.h" + +static int archive_set_format_option(struct archive *a, + const char *m, const char *o, const char *v); +static int archive_set_filter_option(struct archive *a, + const char *m, const char *o, const char *v); +static int archive_set_option(struct archive *a, + const char *m, const char *o, const char *v); + +int +archive_read_set_format_option(struct archive *a, const char *m, const char *o, + const char *v) +{ + return _archive_set_option(a, m, o, v, + ARCHIVE_READ_MAGIC, "archive_read_set_format_option", + archive_set_format_option); +} + +int +archive_read_set_filter_option(struct archive *a, const char *m, const char *o, + const char *v) +{ + return _archive_set_option(a, m, o, v, + ARCHIVE_READ_MAGIC, "archive_read_set_filter_option", + archive_set_filter_option); +} + +int +archive_read_set_option(struct archive *a, const char *m, const char *o, + const char *v) +{ + return _archive_set_option(a, m, o, v, + ARCHIVE_READ_MAGIC, "archive_read_set_option", + archive_set_option); +} + +int +archive_read_set_options(struct archive *a, const char *options) +{ + return _archive_set_options(a, options, + ARCHIVE_READ_MAGIC, "archive_read_set_options", + archive_set_option); +} + +static int +archive_set_format_option(struct archive *_a, const char *m, const char *o, + const char *v) +{ + struct archive_read *a = (struct archive_read *)_a; + size_t i; + int r, rv = ARCHIVE_WARN, matched_modules = 0; + + for (i = 0; i < sizeof(a->formats)/sizeof(a->formats[0]); i++) { + struct archive_format_descriptor *format = &a->formats[i]; + + if (format->options == NULL || format->name == NULL) + /* This format does not support option. */ + continue; + if (m != NULL) { + if (strcmp(format->name, m) != 0) + continue; + ++matched_modules; + } + + a->format = format; + r = format->options(a, o, v); + a->format = NULL; + + if (r == ARCHIVE_FATAL) + return (ARCHIVE_FATAL); + + if (r == ARCHIVE_OK) + rv = ARCHIVE_OK; + } + /* If the format name didn't match, return a special code for + * _archive_set_option[s]. */ + if (m != NULL && matched_modules == 0) + return ARCHIVE_WARN - 1; + return (rv); +} + +static int +archive_set_filter_option(struct archive *_a, const char *m, const char *o, + const char *v) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_filter *filter; + struct archive_read_filter_bidder *bidder; + int r, rv = ARCHIVE_WARN, matched_modules = 0; + + for (filter = a->filter; filter != NULL; filter = filter->upstream) { + bidder = filter->bidder; + if (bidder == NULL) + continue; + if (bidder->options == NULL) + /* This bidder does not support option */ + continue; + if (m != NULL) { + if (strcmp(filter->name, m) != 0) + continue; + ++matched_modules; + } + + r = bidder->options(bidder, o, v); + + if (r == ARCHIVE_FATAL) + return (ARCHIVE_FATAL); + + if (r == ARCHIVE_OK) + rv = ARCHIVE_OK; + } + /* If the filter name didn't match, return a special code for + * _archive_set_option[s]. */ + if (m != NULL && matched_modules == 0) + return ARCHIVE_WARN - 1; + return (rv); +} + +static int +archive_set_option(struct archive *a, const char *m, const char *o, + const char *v) +{ + return _archive_set_either_option(a, m, o, v, + archive_set_format_option, + archive_set_filter_option); +} diff --git a/c/archive_read_support_filter_all.c b/c/archive_read_support_filter_all.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_filter_all.c @@ -0,0 +1,85 @@ +/*- + * Copyright (c) 2003-2011 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#include "archive.h" +#include "archive_private.h" + +#if ARCHIVE_VERSION_NUMBER < 4000000 +/* Deprecated; remove in libarchive 4.0 */ +int +archive_read_support_compression_all(struct archive *a) +{ + return archive_read_support_filter_all(a); +} +#endif + +int +archive_read_support_filter_all(struct archive *a) +{ + archive_check_magic(a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_filter_all"); + + /* Bzip falls back to "bunzip2" command-line */ + archive_read_support_filter_bzip2(a); + /* The decompress code doesn't use an outside library. */ + archive_read_support_filter_compress(a); + /* Gzip decompress falls back to "gzip -d" command-line. */ + archive_read_support_filter_gzip(a); + /* Lzip falls back to "unlzip" command-line program. */ + archive_read_support_filter_lzip(a); + /* The LZMA file format has a very weak signature, so it + * may not be feasible to keep this here, but we'll try. + * This will come back out if there are problems. */ + /* Lzma falls back to "unlzma" command-line program. */ + archive_read_support_filter_lzma(a); + /* Xz falls back to "unxz" command-line program. */ + archive_read_support_filter_xz(a); + /* The decode code doesn't use an outside library. */ + archive_read_support_filter_uu(a); + /* The decode code doesn't use an outside library. */ + archive_read_support_filter_rpm(a); + /* The decode code always uses "lrzip -q -d" command-line. */ + archive_read_support_filter_lrzip(a); + /* Lzop decompress falls back to "lzop -d" command-line. */ + archive_read_support_filter_lzop(a); + /* The decode code always uses "grzip -d" command-line. */ + archive_read_support_filter_grzip(a); + /* Lz4 falls back to "lz4 -d" command-line program. */ + archive_read_support_filter_lz4(a); + /* Zstd falls back to "zstd -d" command-line program. */ + archive_read_support_filter_zstd(a); + + /* Note: We always return ARCHIVE_OK here, even if some of the + * above return ARCHIVE_WARN. The intent here is to enable + * "as much as possible." Clients who need specific + * compression should enable those individually so they can + * verify the level of support. */ + /* Clear any warning messages set by the above functions. */ + archive_clear_error(a); + return (ARCHIVE_OK); +} diff --git a/c/archive_read_support_filter_bzip2.c b/c/archive_read_support_filter_bzip2.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_filter_bzip2.c @@ -0,0 +1,371 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif +#ifdef HAVE_BZLIB_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_read_private.h" + +#if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) +struct private_data { + bz_stream stream; + char *out_block; + size_t out_block_size; + char valid; /* True = decompressor is initialized */ + char eof; /* True = found end of compressed data. */ +}; + +/* Bzip2 filter */ +static ssize_t bzip2_filter_read(struct archive_read_filter *, const void **); +static int bzip2_filter_close(struct archive_read_filter *); +#endif + +/* + * Note that we can detect bzip2 archives even if we can't decompress + * them. (In fact, we like detecting them because we can give better + * error messages.) So the bid framework here gets compiled even + * if bzlib is unavailable. + */ +static int bzip2_reader_bid(struct archive_read_filter_bidder *, struct archive_read_filter *); +static int bzip2_reader_init(struct archive_read_filter *); +static int bzip2_reader_free(struct archive_read_filter_bidder *); + +#if ARCHIVE_VERSION_NUMBER < 4000000 +/* Deprecated; remove in libarchive 4.0 */ +int +archive_read_support_compression_bzip2(struct archive *a) +{ + return archive_read_support_filter_bzip2(a); +} +#endif + +int +archive_read_support_filter_bzip2(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_filter_bidder *reader; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_filter_bzip2"); + + if (__archive_read_get_bidder(a, &reader) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + reader->data = NULL; + reader->name = "bzip2"; + reader->bid = bzip2_reader_bid; + reader->init = bzip2_reader_init; + reader->options = NULL; + reader->free = bzip2_reader_free; +#if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) + return (ARCHIVE_OK); +#else + archive_set_error(_a, ARCHIVE_ERRNO_MISC, + "Using external bzip2 program"); + return (ARCHIVE_WARN); +#endif +} + +static int +bzip2_reader_free(struct archive_read_filter_bidder *self){ + (void)self; /* UNUSED */ + return (ARCHIVE_OK); +} + +/* + * Test whether we can handle this data. + * + * This logic returns zero if any part of the signature fails. It + * also tries to Do The Right Thing if a very short buffer prevents us + * from verifying as much as we would like. + */ +static int +bzip2_reader_bid(struct archive_read_filter_bidder *self, struct archive_read_filter *filter) +{ + const unsigned char *buffer; + ssize_t avail; + int bits_checked; + + (void)self; /* UNUSED */ + + /* Minimal bzip2 archive is 14 bytes. */ + buffer = __archive_read_filter_ahead(filter, 14, &avail); + if (buffer == NULL) + return (0); + + /* First three bytes must be "BZh" */ + bits_checked = 0; + if (memcmp(buffer, "BZh", 3) != 0) + return (0); + bits_checked += 24; + + /* Next follows a compression flag which must be an ASCII digit. */ + if (buffer[3] < '1' || buffer[3] > '9') + return (0); + bits_checked += 5; + + /* After BZh[1-9], there must be either a data block + * which begins with 0x314159265359 or an end-of-data + * marker of 0x177245385090. */ + if (memcmp(buffer + 4, "\x31\x41\x59\x26\x53\x59", 6) == 0) + bits_checked += 48; + else if (memcmp(buffer + 4, "\x17\x72\x45\x38\x50\x90", 6) == 0) + bits_checked += 48; + else + return (0); + + return (bits_checked); +} + +#if !defined(HAVE_BZLIB_H) || !defined(BZ_CONFIG_ERROR) + +/* + * If we don't have the library on this system, we can't actually do the + * decompression. We can, however, still detect compressed archives + * and emit a useful message. + */ +static int +bzip2_reader_init(struct archive_read_filter *self) +{ + int r; + + r = __archive_read_program(self, "bzip2 -d"); + /* Note: We set the format here even if __archive_read_program() + * above fails. We do, after all, know what the format is + * even if we weren't able to read it. */ + self->code = ARCHIVE_FILTER_BZIP2; + self->name = "bzip2"; + return (r); +} + + +#else + +/* + * Setup the callbacks. + */ +static int +bzip2_reader_init(struct archive_read_filter *self) +{ + static const size_t out_block_size = 64 * 1024; + void *out_block; + struct private_data *state; + + self->code = ARCHIVE_FILTER_BZIP2; + self->name = "bzip2"; + + state = (struct private_data *)calloc(sizeof(*state), 1); + out_block = (unsigned char *)malloc(out_block_size); + if (state == NULL || out_block == NULL) { + archive_set_error(&self->archive->archive, ENOMEM, + "Can't allocate data for bzip2 decompression"); + free(out_block); + free(state); + return (ARCHIVE_FATAL); + } + + self->data = state; + state->out_block_size = out_block_size; + state->out_block = out_block; + self->read = bzip2_filter_read; + self->skip = NULL; /* not supported */ + self->close = bzip2_filter_close; + + return (ARCHIVE_OK); +} + +/* + * Return the next block of decompressed data. + */ +static ssize_t +bzip2_filter_read(struct archive_read_filter *self, const void **p) +{ + struct private_data *state; + size_t decompressed; + const char *read_buf; + ssize_t ret; + + state = (struct private_data *)self->data; + + if (state->eof) { + *p = NULL; + return (0); + } + + /* Empty our output buffer. */ + state->stream.next_out = state->out_block; + state->stream.avail_out = state->out_block_size; + + /* Try to fill the output buffer. */ + for (;;) { + if (!state->valid) { + if (bzip2_reader_bid(self->bidder, self->upstream) == 0) { + state->eof = 1; + *p = state->out_block; + decompressed = state->stream.next_out + - state->out_block; + return (decompressed); + } + /* Initialize compression library. */ + ret = BZ2_bzDecompressInit(&(state->stream), + 0 /* library verbosity */, + 0 /* don't use low-mem algorithm */); + + /* If init fails, try low-memory algorithm instead. */ + if (ret == BZ_MEM_ERROR) + ret = BZ2_bzDecompressInit(&(state->stream), + 0 /* library verbosity */, + 1 /* do use low-mem algo */); + + if (ret != BZ_OK) { + const char *detail = NULL; + int err = ARCHIVE_ERRNO_MISC; + switch (ret) { + case BZ_PARAM_ERROR: + detail = "invalid setup parameter"; + break; + case BZ_MEM_ERROR: + err = ENOMEM; + detail = "out of memory"; + break; + case BZ_CONFIG_ERROR: + detail = "mis-compiled library"; + break; + } + archive_set_error(&self->archive->archive, err, + "Internal error initializing decompressor%s%s", + detail == NULL ? "" : ": ", + detail); + return (ARCHIVE_FATAL); + } + state->valid = 1; + } + + /* stream.next_in is really const, but bzlib + * doesn't declare it so. */ + read_buf = + __archive_read_filter_ahead(self->upstream, 1, &ret); + if (read_buf == NULL) { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "truncated bzip2 input"); + return (ARCHIVE_FATAL); + } + state->stream.next_in = (char *)(uintptr_t)read_buf; + state->stream.avail_in = ret; + /* There is no more data, return whatever we have. */ + if (ret == 0) { + state->eof = 1; + *p = state->out_block; + decompressed = state->stream.next_out + - state->out_block; + return (decompressed); + } + + /* Decompress as much as we can in one pass. */ + ret = BZ2_bzDecompress(&(state->stream)); + __archive_read_filter_consume(self->upstream, + state->stream.next_in - read_buf); + + switch (ret) { + case BZ_STREAM_END: /* Found end of stream. */ + switch (BZ2_bzDecompressEnd(&(state->stream))) { + case BZ_OK: + break; + default: + archive_set_error(&(self->archive->archive), + ARCHIVE_ERRNO_MISC, + "Failed to clean up decompressor"); + return (ARCHIVE_FATAL); + } + state->valid = 0; + /* FALLTHROUGH */ + case BZ_OK: /* Decompressor made some progress. */ + /* If we filled our buffer, update stats and return. */ + if (state->stream.avail_out == 0) { + *p = state->out_block; + decompressed = state->stream.next_out + - state->out_block; + return (decompressed); + } + break; + default: /* Return an error. */ + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, "bzip decompression failed"); + return (ARCHIVE_FATAL); + } + } +} + +/* + * Clean up the decompressor. + */ +static int +bzip2_filter_close(struct archive_read_filter *self) +{ + struct private_data *state; + int ret = ARCHIVE_OK; + + state = (struct private_data *)self->data; + + if (state->valid) { + switch (BZ2_bzDecompressEnd(&state->stream)) { + case BZ_OK: + break; + default: + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Failed to clean up decompressor"); + ret = ARCHIVE_FATAL; + } + state->valid = 0; + } + + free(state->out_block); + free(state); + return (ret); +} + +#endif /* HAVE_BZLIB_H && BZ_CONFIG_ERROR */ diff --git a/c/archive_read_support_filter_compress.c b/c/archive_read_support_filter_compress.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_filter_compress.c @@ -0,0 +1,465 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +/* + * This code borrows heavily from "compress" source code, which is + * protected by the following copyright. (Clause 3 dropped by request + * of the Regents.) + */ + +/*- + * Copyright (c) 1985, 1986, 1992, 1993 + * The Regents of the University of California. All rights reserved. + * + * This code is derived from software contributed to Berkeley by + * Diomidis Spinellis and James A. Woods, derived from original + * work by Spencer Thomas and Joseph Orost. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_read_private.h" + +/* + * Because LZW decompression is pretty simple, I've just implemented + * the whole decompressor here (cribbing from "compress" source code, + * of course), rather than relying on an external library. I have + * made an effort to clarify and simplify the algorithm, so the + * names and structure here don't exactly match those used by compress. + */ + +struct private_data { + /* Input variables. */ + const unsigned char *next_in; + size_t avail_in; + size_t consume_unnotified; + int bit_buffer; + int bits_avail; + size_t bytes_in_section; + + /* Output variables. */ + size_t out_block_size; + void *out_block; + + /* Decompression status variables. */ + int use_reset_code; + int end_of_stream; /* EOF status. */ + int maxcode; /* Largest code. */ + int maxcode_bits; /* Length of largest code. */ + int section_end_code; /* When to increase bits. */ + int bits; /* Current code length. */ + int oldcode; /* Previous code. */ + int finbyte; /* Last byte of prev code. */ + + /* Dictionary. */ + int free_ent; /* Next dictionary entry. */ + unsigned char suffix[65536]; + uint16_t prefix[65536]; + + /* + * Scratch area for expanding dictionary entries. Note: + * "worst" case here comes from compressing /dev/zero: the + * last code in the dictionary will code a sequence of + * 65536-256 zero bytes. Thus, we need stack space to expand + * a 65280-byte dictionary entry. (Of course, 32640:1 + * compression could also be considered the "best" case. ;-) + */ + unsigned char *stackp; + unsigned char stack[65300]; +}; + +static int compress_bidder_bid(struct archive_read_filter_bidder *, struct archive_read_filter *); +static int compress_bidder_init(struct archive_read_filter *); +static int compress_bidder_free(struct archive_read_filter_bidder *); + +static ssize_t compress_filter_read(struct archive_read_filter *, const void **); +static int compress_filter_close(struct archive_read_filter *); + +static int getbits(struct archive_read_filter *, int n); +static int next_code(struct archive_read_filter *); + +#if ARCHIVE_VERSION_NUMBER < 4000000 +/* Deprecated; remove in libarchive 4.0 */ +int +archive_read_support_compression_compress(struct archive *a) +{ + return archive_read_support_filter_compress(a); +} +#endif + +int +archive_read_support_filter_compress(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_filter_bidder *bidder; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_filter_compress"); + + if (__archive_read_get_bidder(a, &bidder) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + bidder->data = NULL; + bidder->name = "compress (.Z)"; + bidder->bid = compress_bidder_bid; + bidder->init = compress_bidder_init; + bidder->options = NULL; + bidder->free = compress_bidder_free; + return (ARCHIVE_OK); +} + +/* + * Test whether we can handle this data. + * This logic returns zero if any part of the signature fails. + */ +static int +compress_bidder_bid(struct archive_read_filter_bidder *self, + struct archive_read_filter *filter) +{ + const unsigned char *buffer; + ssize_t avail; + int bits_checked; + + (void)self; /* UNUSED */ + + /* Shortest valid compress file is 3 bytes. */ + buffer = __archive_read_filter_ahead(filter, 3, &avail); + + if (buffer == NULL) + return (0); + + bits_checked = 0; + /* First two bytes are the magic value */ + if (buffer[0] != 0x1F || buffer[1] != 0x9D) + return (0); + /* Third byte holds compression parameters. */ + if (buffer[2] & 0x20) /* Reserved bit, must be zero. */ + return (0); + if (buffer[2] & 0x40) /* Reserved bit, must be zero. */ + return (0); + bits_checked += 18; + + return (bits_checked); +} + +/* + * Setup the callbacks. + */ +static int +compress_bidder_init(struct archive_read_filter *self) +{ + struct private_data *state; + static const size_t out_block_size = 64 * 1024; + void *out_block; + int code; + + self->code = ARCHIVE_FILTER_COMPRESS; + self->name = "compress (.Z)"; + + state = (struct private_data *)calloc(sizeof(*state), 1); + out_block = malloc(out_block_size); + if (state == NULL || out_block == NULL) { + free(out_block); + free(state); + archive_set_error(&self->archive->archive, ENOMEM, + "Can't allocate data for %s decompression", + self->name); + return (ARCHIVE_FATAL); + } + + self->data = state; + state->out_block_size = out_block_size; + state->out_block = out_block; + self->read = compress_filter_read; + self->skip = NULL; /* not supported */ + self->close = compress_filter_close; + + /* XXX MOVE THE FOLLOWING OUT OF INIT() XXX */ + + (void)getbits(self, 8); /* Skip first signature byte. */ + (void)getbits(self, 8); /* Skip second signature byte. */ + + /* Get compression parameters. */ + code = getbits(self, 8); + if ((code & 0x1f) > 16) { + archive_set_error(&self->archive->archive, -1, + "Invalid compressed data"); + return (ARCHIVE_FATAL); + } + state->maxcode_bits = code & 0x1f; + state->maxcode = (1 << state->maxcode_bits); + state->use_reset_code = code & 0x80; + + /* Initialize decompressor. */ + state->free_ent = 256; + state->stackp = state->stack; + if (state->use_reset_code) + state->free_ent++; + state->bits = 9; + state->section_end_code = (1<bits) - 1; + state->oldcode = -1; + for (code = 255; code >= 0; code--) { + state->prefix[code] = 0; + state->suffix[code] = code; + } + next_code(self); + + return (ARCHIVE_OK); +} + +/* + * Return a block of data from the decompression buffer. Decompress more + * as necessary. + */ +static ssize_t +compress_filter_read(struct archive_read_filter *self, const void **pblock) +{ + struct private_data *state; + unsigned char *p, *start, *end; + int ret; + + state = (struct private_data *)self->data; + if (state->end_of_stream) { + *pblock = NULL; + return (0); + } + p = start = (unsigned char *)state->out_block; + end = start + state->out_block_size; + + while (p < end && !state->end_of_stream) { + if (state->stackp > state->stack) { + *p++ = *--state->stackp; + } else { + ret = next_code(self); + if (ret == -1) + state->end_of_stream = ret; + else if (ret != ARCHIVE_OK) + return (ret); + } + } + + *pblock = start; + return (p - start); +} + +/* + * Clean up the reader. + */ +static int +compress_bidder_free(struct archive_read_filter_bidder *self) +{ + self->data = NULL; + return (ARCHIVE_OK); +} + +/* + * Close and release the filter. + */ +static int +compress_filter_close(struct archive_read_filter *self) +{ + struct private_data *state = (struct private_data *)self->data; + + free(state->out_block); + free(state); + return (ARCHIVE_OK); +} + +/* + * Process the next code and fill the stack with the expansion + * of the code. Returns ARCHIVE_FATAL if there is a fatal I/O or + * format error, ARCHIVE_EOF if we hit end of data, ARCHIVE_OK otherwise. + */ +static int +next_code(struct archive_read_filter *self) +{ + struct private_data *state = (struct private_data *)self->data; + int code, newcode; + + static int debug_buff[1024]; + static unsigned debug_index; + + code = newcode = getbits(self, state->bits); + if (code < 0) + return (code); + + debug_buff[debug_index++] = code; + if (debug_index >= sizeof(debug_buff)/sizeof(debug_buff[0])) + debug_index = 0; + + /* If it's a reset code, reset the dictionary. */ + if ((code == 256) && state->use_reset_code) { + /* + * The original 'compress' implementation blocked its + * I/O in a manner that resulted in junk bytes being + * inserted after every reset. The next section skips + * this junk. (Yes, the number of *bytes* to skip is + * a function of the current *bit* length.) + */ + int skip_bytes = state->bits - + (state->bytes_in_section % state->bits); + skip_bytes %= state->bits; + state->bits_avail = 0; /* Discard rest of this byte. */ + while (skip_bytes-- > 0) { + code = getbits(self, 8); + if (code < 0) + return (code); + } + /* Now, actually do the reset. */ + state->bytes_in_section = 0; + state->bits = 9; + state->section_end_code = (1 << state->bits) - 1; + state->free_ent = 257; + state->oldcode = -1; + return (next_code(self)); + } + + if (code > state->free_ent + || (code == state->free_ent && state->oldcode < 0)) { + /* An invalid code is a fatal error. */ + archive_set_error(&(self->archive->archive), -1, + "Invalid compressed data"); + return (ARCHIVE_FATAL); + } + + /* Special case for KwKwK string. */ + if (code >= state->free_ent) { + *state->stackp++ = state->finbyte; + code = state->oldcode; + } + + /* Generate output characters in reverse order. */ + while (code >= 256) { + *state->stackp++ = state->suffix[code]; + code = state->prefix[code]; + } + *state->stackp++ = state->finbyte = code; + + /* Generate the new entry. */ + code = state->free_ent; + if (code < state->maxcode && state->oldcode >= 0) { + state->prefix[code] = state->oldcode; + state->suffix[code] = state->finbyte; + ++state->free_ent; + } + if (state->free_ent > state->section_end_code) { + state->bits++; + state->bytes_in_section = 0; + if (state->bits == state->maxcode_bits) + state->section_end_code = state->maxcode; + else + state->section_end_code = (1 << state->bits) - 1; + } + + /* Remember previous code. */ + state->oldcode = newcode; + return (ARCHIVE_OK); +} + +/* + * Return next 'n' bits from stream. + * + * -1 indicates end of available data. + */ +static int +getbits(struct archive_read_filter *self, int n) +{ + struct private_data *state = (struct private_data *)self->data; + int code; + ssize_t ret; + static const int mask[] = { + 0x00, 0x01, 0x03, 0x07, 0x0f, 0x1f, 0x3f, 0x7f, 0xff, + 0x1ff, 0x3ff, 0x7ff, 0xfff, 0x1fff, 0x3fff, 0x7fff, 0xffff + }; + + while (state->bits_avail < n) { + if (state->avail_in <= 0) { + if (state->consume_unnotified) { + __archive_read_filter_consume(self->upstream, + state->consume_unnotified); + state->consume_unnotified = 0; + } + state->next_in + = __archive_read_filter_ahead(self->upstream, + 1, &ret); + if (ret == 0) + return (-1); + if (ret < 0 || state->next_in == NULL) + return (ARCHIVE_FATAL); + state->consume_unnotified = state->avail_in = ret; + } + state->bit_buffer |= *state->next_in++ << state->bits_avail; + state->avail_in--; + state->bits_avail += 8; + state->bytes_in_section++; + } + + code = state->bit_buffer; + state->bit_buffer >>= n; + state->bits_avail -= n; + + return (code & mask[n]); +} diff --git a/c/archive_read_support_filter_grzip.c b/c/archive_read_support_filter_grzip.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_filter_grzip.c @@ -0,0 +1,121 @@ +/*- + * Copyright (c) 2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +__FBSDID("$FreeBSD$"); + + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_read_private.h" + +static const unsigned char grzip_magic[] = { + 0x47, 0x52, 0x5a, 0x69, 0x70, 0x49, 0x49, 0x00, + 0x02, 0x04, 0x3a, 0x29 }; + +static int grzip_bidder_bid(struct archive_read_filter_bidder *, + struct archive_read_filter *); +static int grzip_bidder_init(struct archive_read_filter *); + + +static int +grzip_reader_free(struct archive_read_filter_bidder *self) +{ + (void)self; /* UNUSED */ + return (ARCHIVE_OK); +} + +int +archive_read_support_filter_grzip(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_filter_bidder *reader; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_filter_grzip"); + + if (__archive_read_get_bidder(a, &reader) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + reader->data = NULL; + reader->bid = grzip_bidder_bid; + reader->init = grzip_bidder_init; + reader->options = NULL; + reader->free = grzip_reader_free; + /* This filter always uses an external program. */ + archive_set_error(_a, ARCHIVE_ERRNO_MISC, + "Using external grzip program for grzip decompression"); + return (ARCHIVE_WARN); +} + +/* + * Bidder just verifies the header and returns the number of verified bits. + */ +static int +grzip_bidder_bid(struct archive_read_filter_bidder *self, + struct archive_read_filter *filter) +{ + const unsigned char *p; + ssize_t avail; + + (void)self; /* UNUSED */ + + p = __archive_read_filter_ahead(filter, sizeof(grzip_magic), &avail); + if (p == NULL || avail == 0) + return (0); + + if (memcmp(p, grzip_magic, sizeof(grzip_magic))) + return (0); + + return (sizeof(grzip_magic) * 8); +} + +static int +grzip_bidder_init(struct archive_read_filter *self) +{ + int r; + + r = __archive_read_program(self, "grzip -d"); + /* Note: We set the format here even if __archive_read_program() + * above fails. We do, after all, know what the format is + * even if we weren't able to read it. */ + self->code = ARCHIVE_FILTER_GRZIP; + self->name = "grzip"; + return (r); +} diff --git a/c/archive_read_support_filter_gzip.c b/c/archive_read_support_filter_gzip.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_filter_gzip.c @@ -0,0 +1,535 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +__FBSDID("$FreeBSD$"); + + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_LIMITS_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif +#ifdef HAVE_ZLIB_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_endian.h" +#include "archive_private.h" +#include "archive_read_private.h" + +#ifdef HAVE_ZLIB_H +struct private_data { + z_stream stream; + char in_stream; + unsigned char *out_block; + size_t out_block_size; + int64_t total_out; + unsigned long crc; + uint32_t mtime; + char *name; + char eof; /* True = found end of compressed data. */ +}; + +/* Gzip Filter. */ +static ssize_t gzip_filter_read(struct archive_read_filter *, const void **); +static int gzip_filter_close(struct archive_read_filter *); +#endif + +/* + * Note that we can detect gzip archives even if we can't decompress + * them. (In fact, we like detecting them because we can give better + * error messages.) So the bid framework here gets compiled even + * if zlib is unavailable. + * + * TODO: If zlib is unavailable, gzip_bidder_init() should + * use the compress_program framework to try to fire up an external + * gzip program. + */ +static int gzip_bidder_bid(struct archive_read_filter_bidder *, + struct archive_read_filter *); +static int gzip_bidder_init(struct archive_read_filter *); + +#if ARCHIVE_VERSION_NUMBER < 4000000 +/* Deprecated; remove in libarchive 4.0 */ +int +archive_read_support_compression_gzip(struct archive *a) +{ + return archive_read_support_filter_gzip(a); +} +#endif + +int +archive_read_support_filter_gzip(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_filter_bidder *bidder; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_filter_gzip"); + + if (__archive_read_get_bidder(a, &bidder) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + bidder->data = NULL; + bidder->name = "gzip"; + bidder->bid = gzip_bidder_bid; + bidder->init = gzip_bidder_init; + bidder->options = NULL; + bidder->free = NULL; /* No data, so no cleanup necessary. */ + /* Signal the extent of gzip support with the return value here. */ +#if HAVE_ZLIB_H + return (ARCHIVE_OK); +#else + archive_set_error(_a, ARCHIVE_ERRNO_MISC, + "Using external gzip program"); + return (ARCHIVE_WARN); +#endif +} + +/* + * Read and verify the header. + * + * Returns zero if the header couldn't be validated, else returns + * number of bytes in header. If pbits is non-NULL, it receives a + * count of bits verified, suitable for use by bidder. + */ +static ssize_t +peek_at_header(struct archive_read_filter *filter, int *pbits, +#ifdef HAVE_ZLIB_H + struct private_data *state +#else + void *state +#endif + ) +{ + const unsigned char *p; + ssize_t avail, len; + int bits = 0; + int header_flags; +#ifndef HAVE_ZLIB_H + (void)state; /* UNUSED */ +#endif + + /* Start by looking at the first ten bytes of the header, which + * is all fixed layout. */ + len = 10; + p = __archive_read_filter_ahead(filter, len, &avail); + if (p == NULL || avail == 0) + return (0); + /* We only support deflation- third byte must be 0x08. */ + if (memcmp(p, "\x1F\x8B\x08", 3) != 0) + return (0); + bits += 24; + if ((p[3] & 0xE0)!= 0) /* No reserved flags set. */ + return (0); + bits += 3; + header_flags = p[3]; + /* Bytes 4-7 are mod time in little endian. */ +#ifdef HAVE_ZLIB_H + if (state) + state->mtime = archive_le32dec(p + 4); +#endif + /* Byte 8 is deflate flags. */ + /* XXXX TODO: return deflate flags back to consume_header for use + in initializing the decompressor. */ + /* Byte 9 is OS. */ + + /* Optional extra data: 2 byte length plus variable body. */ + if (header_flags & 4) { + p = __archive_read_filter_ahead(filter, len + 2, &avail); + if (p == NULL) + return (0); + len += ((int)p[len + 1] << 8) | (int)p[len]; + len += 2; + } + + /* Null-terminated optional filename. */ + if (header_flags & 8) { +#ifdef HAVE_ZLIB_H + ssize_t file_start = len; +#endif + do { + ++len; + if (avail < len) + p = __archive_read_filter_ahead(filter, + len, &avail); + if (p == NULL) + return (0); + } while (p[len - 1] != 0); + +#ifdef HAVE_ZLIB_H + if (state) { + /* Reset the name in case of repeat header reads. */ + free(state->name); + state->name = strdup((const char *)&p[file_start]); + } +#endif + } + + /* Null-terminated optional comment. */ + if (header_flags & 16) { + do { + ++len; + if (avail < len) + p = __archive_read_filter_ahead(filter, + len, &avail); + if (p == NULL) + return (0); + } while (p[len - 1] != 0); + } + + /* Optional header CRC */ + if ((header_flags & 2)) { + p = __archive_read_filter_ahead(filter, len + 2, &avail); + if (p == NULL) + return (0); +#if 0 + int hcrc = ((int)p[len + 1] << 8) | (int)p[len]; + int crc = /* XXX TODO: Compute header CRC. */; + if (crc != hcrc) + return (0); + bits += 16; +#endif + len += 2; + } + + if (pbits != NULL) + *pbits = bits; + return (len); +} + +/* + * Bidder just verifies the header and returns the number of verified bits. + */ +static int +gzip_bidder_bid(struct archive_read_filter_bidder *self, + struct archive_read_filter *filter) +{ + int bits_checked; + + (void)self; /* UNUSED */ + + if (peek_at_header(filter, &bits_checked, NULL)) + return (bits_checked); + return (0); +} + +#ifndef HAVE_ZLIB_H + +/* + * If we don't have the library on this system, we can't do the + * decompression directly. We can, however, try to run "gzip -d" + * in case that's available. + */ +static int +gzip_bidder_init(struct archive_read_filter *self) +{ + int r; + + r = __archive_read_program(self, "gzip -d"); + /* Note: We set the format here even if __archive_read_program() + * above fails. We do, after all, know what the format is + * even if we weren't able to read it. */ + self->code = ARCHIVE_FILTER_GZIP; + self->name = "gzip"; + return (r); +} + +#else + +static int +gzip_read_header(struct archive_read_filter *self, struct archive_entry *entry) +{ + struct private_data *state; + + state = (struct private_data *)self->data; + + /* A mtime of 0 is considered invalid/missing. */ + if (state->mtime != 0) + archive_entry_set_mtime(entry, state->mtime, 0); + + /* If the name is available, extract it. */ + if (state->name) + archive_entry_set_pathname(entry, state->name); + + return (ARCHIVE_OK); +} + +/* + * Initialize the filter object. + */ +static int +gzip_bidder_init(struct archive_read_filter *self) +{ + struct private_data *state; + static const size_t out_block_size = 64 * 1024; + void *out_block; + + self->code = ARCHIVE_FILTER_GZIP; + self->name = "gzip"; + + state = (struct private_data *)calloc(sizeof(*state), 1); + out_block = (unsigned char *)malloc(out_block_size); + if (state == NULL || out_block == NULL) { + free(out_block); + free(state); + archive_set_error(&self->archive->archive, ENOMEM, + "Can't allocate data for gzip decompression"); + return (ARCHIVE_FATAL); + } + + self->data = state; + state->out_block_size = out_block_size; + state->out_block = out_block; + self->read = gzip_filter_read; + self->skip = NULL; /* not supported */ + self->close = gzip_filter_close; +#ifdef HAVE_ZLIB_H + self->read_header = gzip_read_header; +#endif + + state->in_stream = 0; /* We're not actually within a stream yet. */ + + return (ARCHIVE_OK); +} + +static int +consume_header(struct archive_read_filter *self) +{ + struct private_data *state; + ssize_t avail; + size_t len; + int ret; + + state = (struct private_data *)self->data; + + /* If this is a real header, consume it. */ + len = peek_at_header(self->upstream, NULL, state); + if (len == 0) + return (ARCHIVE_EOF); + __archive_read_filter_consume(self->upstream, len); + + /* Initialize CRC accumulator. */ + state->crc = crc32(0L, NULL, 0); + + /* Initialize compression library. */ + state->stream.next_in = (unsigned char *)(uintptr_t) + __archive_read_filter_ahead(self->upstream, 1, &avail); + state->stream.avail_in = (uInt)avail; + ret = inflateInit2(&(state->stream), + -15 /* Don't check for zlib header */); + + /* Decipher the error code. */ + switch (ret) { + case Z_OK: + state->in_stream = 1; + return (ARCHIVE_OK); + case Z_STREAM_ERROR: + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Internal error initializing compression library: " + "invalid setup parameter"); + break; + case Z_MEM_ERROR: + archive_set_error(&self->archive->archive, ENOMEM, + "Internal error initializing compression library: " + "out of memory"); + break; + case Z_VERSION_ERROR: + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Internal error initializing compression library: " + "invalid library version"); + break; + default: + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Internal error initializing compression library: " + " Zlib error %d", ret); + break; + } + return (ARCHIVE_FATAL); +} + +static int +consume_trailer(struct archive_read_filter *self) +{ + struct private_data *state; + const unsigned char *p; + ssize_t avail; + + state = (struct private_data *)self->data; + + state->in_stream = 0; + switch (inflateEnd(&(state->stream))) { + case Z_OK: + break; + default: + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Failed to clean up gzip decompressor"); + return (ARCHIVE_FATAL); + } + + /* GZip trailer is a fixed 8 byte structure. */ + p = __archive_read_filter_ahead(self->upstream, 8, &avail); + if (p == NULL || avail == 0) + return (ARCHIVE_FATAL); + + /* XXX TODO: Verify the length and CRC. */ + + /* We've verified the trailer, so consume it now. */ + __archive_read_filter_consume(self->upstream, 8); + + return (ARCHIVE_OK); +} + +static ssize_t +gzip_filter_read(struct archive_read_filter *self, const void **p) +{ + struct private_data *state; + size_t decompressed; + ssize_t avail_in, max_in; + int ret; + + state = (struct private_data *)self->data; + + /* Empty our output buffer. */ + state->stream.next_out = state->out_block; + state->stream.avail_out = (uInt)state->out_block_size; + + /* Try to fill the output buffer. */ + while (state->stream.avail_out > 0 && !state->eof) { + /* If we're not in a stream, read a header + * and initialize the decompression library. */ + if (!state->in_stream) { + ret = consume_header(self); + if (ret == ARCHIVE_EOF) { + state->eof = 1; + break; + } + if (ret < ARCHIVE_OK) + return (ret); + } + + /* Peek at the next available data. */ + /* ZLib treats stream.next_in as const but doesn't declare + * it so, hence this ugly cast. */ + state->stream.next_in = (unsigned char *)(uintptr_t) + __archive_read_filter_ahead(self->upstream, 1, &avail_in); + if (state->stream.next_in == NULL) { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "truncated gzip input"); + return (ARCHIVE_FATAL); + } + if (UINT_MAX >= SSIZE_MAX) + max_in = SSIZE_MAX; + else + max_in = UINT_MAX; + if (avail_in > max_in) + avail_in = max_in; + state->stream.avail_in = (uInt)avail_in; + + /* Decompress and consume some of that data. */ + ret = inflate(&(state->stream), 0); + switch (ret) { + case Z_OK: /* Decompressor made some progress. */ + __archive_read_filter_consume(self->upstream, + avail_in - state->stream.avail_in); + break; + case Z_STREAM_END: /* Found end of stream. */ + __archive_read_filter_consume(self->upstream, + avail_in - state->stream.avail_in); + /* Consume the stream trailer; release the + * decompression library. */ + ret = consume_trailer(self); + if (ret < ARCHIVE_OK) + return (ret); + break; + default: + /* Return an error. */ + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "gzip decompression failed"); + return (ARCHIVE_FATAL); + } + } + + /* We've read as much as we can. */ + decompressed = state->stream.next_out - state->out_block; + state->total_out += decompressed; + if (decompressed == 0) + *p = NULL; + else + *p = state->out_block; + return (decompressed); +} + +/* + * Clean up the decompressor. + */ +static int +gzip_filter_close(struct archive_read_filter *self) +{ + struct private_data *state; + int ret; + + state = (struct private_data *)self->data; + ret = ARCHIVE_OK; + + if (state->in_stream) { + switch (inflateEnd(&(state->stream))) { + case Z_OK: + break; + default: + archive_set_error(&(self->archive->archive), + ARCHIVE_ERRNO_MISC, + "Failed to clean up gzip compressor"); + ret = ARCHIVE_FATAL; + } + } + + free(state->name); + free(state->out_block); + free(state); + return (ret); +} + +#endif /* HAVE_ZLIB_H */ diff --git a/c/archive_read_support_filter_lrzip.c b/c/archive_read_support_filter_lrzip.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_filter_lrzip.c @@ -0,0 +1,132 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +__FBSDID("$FreeBSD$"); + + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_read_private.h" + +#define LRZIP_HEADER_MAGIC "LRZI" +#define LRZIP_HEADER_MAGIC_LEN 4 + +static int lrzip_bidder_bid(struct archive_read_filter_bidder *, + struct archive_read_filter *); +static int lrzip_bidder_init(struct archive_read_filter *); + + +static int +lrzip_reader_free(struct archive_read_filter_bidder *self) +{ + (void)self; /* UNUSED */ + return (ARCHIVE_OK); +} + +int +archive_read_support_filter_lrzip(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_filter_bidder *reader; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_filter_lrzip"); + + if (__archive_read_get_bidder(a, &reader) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + reader->data = NULL; + reader->name = "lrzip"; + reader->bid = lrzip_bidder_bid; + reader->init = lrzip_bidder_init; + reader->options = NULL; + reader->free = lrzip_reader_free; + /* This filter always uses an external program. */ + archive_set_error(_a, ARCHIVE_ERRNO_MISC, + "Using external lrzip program for lrzip decompression"); + return (ARCHIVE_WARN); +} + +/* + * Bidder just verifies the header and returns the number of verified bits. + */ +static int +lrzip_bidder_bid(struct archive_read_filter_bidder *self, + struct archive_read_filter *filter) +{ + const unsigned char *p; + ssize_t avail, len; + int i; + + (void)self; /* UNUSED */ + /* Start by looking at the first six bytes of the header, which + * is all fixed layout. */ + len = 6; + p = __archive_read_filter_ahead(filter, len, &avail); + if (p == NULL || avail == 0) + return (0); + + if (memcmp(p, LRZIP_HEADER_MAGIC, LRZIP_HEADER_MAGIC_LEN)) + return (0); + + /* current major version is always 0, verify this */ + if (p[LRZIP_HEADER_MAGIC_LEN]) + return 0; + /* support only v0.6+ lrzip for sanity */ + i = p[LRZIP_HEADER_MAGIC_LEN + 1]; + if ((i < 6) || (i > 10)) + return 0; + + return (int)len; +} + +static int +lrzip_bidder_init(struct archive_read_filter *self) +{ + int r; + + r = __archive_read_program(self, "lrzip -d -q"); + /* Note: We set the format here even if __archive_read_program() + * above fails. We do, after all, know what the format is + * even if we weren't able to read it. */ + self->code = ARCHIVE_FILTER_LRZIP; + self->name = "lrzip"; + return (r); +} diff --git a/c/archive_read_support_filter_lz4.c b/c/archive_read_support_filter_lz4.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_filter_lz4.c @@ -0,0 +1,742 @@ +/*- + * Copyright (c) 2014 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif +#ifdef HAVE_LZ4_H +#include +#endif + +#include "archive.h" +#include "archive_endian.h" +#include "archive_private.h" +#include "archive_read_private.h" +#include "archive_xxhash.h" + +#define LZ4_MAGICNUMBER 0x184d2204 +#define LZ4_SKIPPABLED 0x184d2a50 +#define LZ4_LEGACY 0x184c2102 + +#if defined(HAVE_LIBLZ4) +struct private_data { + enum { SELECT_STREAM, + READ_DEFAULT_STREAM, + READ_DEFAULT_BLOCK, + READ_LEGACY_STREAM, + READ_LEGACY_BLOCK, + } stage; + struct { + unsigned block_independence:1; + unsigned block_checksum:3; + unsigned stream_size:1; + unsigned stream_checksum:1; + unsigned preset_dictionary:1; + int block_maximum_size; + } flags; + int64_t stream_size; + uint32_t dict_id; + char *out_block; + size_t out_block_size; + + /* Bytes read but not yet consumed via __archive_read_consume() */ + size_t unconsumed; + size_t decoded_size; + void *xxh32_state; + + char valid; /* True = decompressor is initialized */ + char eof; /* True = found end of compressed data. */ +}; + +#define LEGACY_BLOCK_SIZE (8 * 1024 * 1024) + +/* Lz4 filter */ +static ssize_t lz4_filter_read(struct archive_read_filter *, const void **); +static int lz4_filter_close(struct archive_read_filter *); +#endif + +/* + * Note that we can detect lz4 archives even if we can't decompress + * them. (In fact, we like detecting them because we can give better + * error messages.) So the bid framework here gets compiled even + * if liblz4 is unavailable. + */ +static int lz4_reader_bid(struct archive_read_filter_bidder *, struct archive_read_filter *); +static int lz4_reader_init(struct archive_read_filter *); +static int lz4_reader_free(struct archive_read_filter_bidder *); +#if defined(HAVE_LIBLZ4) +static ssize_t lz4_filter_read_default_stream(struct archive_read_filter *, + const void **); +static ssize_t lz4_filter_read_legacy_stream(struct archive_read_filter *, + const void **); +#endif + +int +archive_read_support_filter_lz4(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_filter_bidder *reader; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_filter_lz4"); + + if (__archive_read_get_bidder(a, &reader) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + reader->data = NULL; + reader->name = "lz4"; + reader->bid = lz4_reader_bid; + reader->init = lz4_reader_init; + reader->options = NULL; + reader->free = lz4_reader_free; +#if defined(HAVE_LIBLZ4) + return (ARCHIVE_OK); +#else + archive_set_error(_a, ARCHIVE_ERRNO_MISC, + "Using external lz4 program"); + return (ARCHIVE_WARN); +#endif +} + +static int +lz4_reader_free(struct archive_read_filter_bidder *self){ + (void)self; /* UNUSED */ + return (ARCHIVE_OK); +} + +/* + * Test whether we can handle this data. + * + * This logic returns zero if any part of the signature fails. It + * also tries to Do The Right Thing if a very short buffer prevents us + * from verifying as much as we would like. + */ +static int +lz4_reader_bid(struct archive_read_filter_bidder *self, + struct archive_read_filter *filter) +{ + const unsigned char *buffer; + ssize_t avail; + int bits_checked; + uint32_t number; + + (void)self; /* UNUSED */ + + /* Minimal lz4 archive is 11 bytes. */ + buffer = __archive_read_filter_ahead(filter, 11, &avail); + if (buffer == NULL) + return (0); + + /* First four bytes must be LZ4 magic numbers. */ + bits_checked = 0; + if ((number = archive_le32dec(buffer)) == LZ4_MAGICNUMBER) { + unsigned char flag, BD; + + bits_checked += 32; + /* Next follows a stream descriptor. */ + /* Descriptor Flags. */ + flag = buffer[4]; + /* A version number must be "01". */ + if (((flag & 0xc0) >> 6) != 1) + return (0); + /* A reserved bit must be "0". */ + if (flag & 2) + return (0); + bits_checked += 8; + BD = buffer[5]; + /* A block maximum size should be more than 3. */ + if (((BD & 0x70) >> 4) < 4) + return (0); + /* Reserved bits must be "0". */ + if (BD & ~0x70) + return (0); + bits_checked += 8; + } else if (number == LZ4_LEGACY) { + bits_checked += 32; + } + + return (bits_checked); +} + +#if !defined(HAVE_LIBLZ4) + +/* + * If we don't have the library on this system, we can't actually do the + * decompression. We can, however, still detect compressed archives + * and emit a useful message. + */ +static int +lz4_reader_init(struct archive_read_filter *self) +{ + int r; + + r = __archive_read_program(self, "lz4 -d -q"); + /* Note: We set the format here even if __archive_read_program() + * above fails. We do, after all, know what the format is + * even if we weren't able to read it. */ + self->code = ARCHIVE_FILTER_LZ4; + self->name = "lz4"; + return (r); +} + + +#else + +/* + * Setup the callbacks. + */ +static int +lz4_reader_init(struct archive_read_filter *self) +{ + struct private_data *state; + + self->code = ARCHIVE_FILTER_LZ4; + self->name = "lz4"; + + state = (struct private_data *)calloc(sizeof(*state), 1); + if (state == NULL) { + archive_set_error(&self->archive->archive, ENOMEM, + "Can't allocate data for lz4 decompression"); + return (ARCHIVE_FATAL); + } + + self->data = state; + state->stage = SELECT_STREAM; + self->read = lz4_filter_read; + self->skip = NULL; /* not supported */ + self->close = lz4_filter_close; + + return (ARCHIVE_OK); +} + +static int +lz4_allocate_out_block(struct archive_read_filter *self) +{ + struct private_data *state = (struct private_data *)self->data; + size_t out_block_size = state->flags.block_maximum_size; + void *out_block; + + if (!state->flags.block_independence) + out_block_size += 64 * 1024; + if (state->out_block_size < out_block_size) { + free(state->out_block); + out_block = (unsigned char *)malloc(out_block_size); + state->out_block_size = out_block_size; + if (out_block == NULL) { + archive_set_error(&self->archive->archive, ENOMEM, + "Can't allocate data for lz4 decompression"); + return (ARCHIVE_FATAL); + } + state->out_block = out_block; + } + if (!state->flags.block_independence) + memset(state->out_block, 0, 64 * 1024); + return (ARCHIVE_OK); +} + +static int +lz4_allocate_out_block_for_legacy(struct archive_read_filter *self) +{ + struct private_data *state = (struct private_data *)self->data; + size_t out_block_size = LEGACY_BLOCK_SIZE; + void *out_block; + + if (state->out_block_size < out_block_size) { + free(state->out_block); + out_block = (unsigned char *)malloc(out_block_size); + state->out_block_size = out_block_size; + if (out_block == NULL) { + archive_set_error(&self->archive->archive, ENOMEM, + "Can't allocate data for lz4 decompression"); + return (ARCHIVE_FATAL); + } + state->out_block = out_block; + } + return (ARCHIVE_OK); +} + +/* + * Return the next block of decompressed data. + */ +static ssize_t +lz4_filter_read(struct archive_read_filter *self, const void **p) +{ + struct private_data *state = (struct private_data *)self->data; + ssize_t ret; + + if (state->eof) { + *p = NULL; + return (0); + } + + __archive_read_filter_consume(self->upstream, state->unconsumed); + state->unconsumed = 0; + + switch (state->stage) { + case SELECT_STREAM: + break; + case READ_DEFAULT_STREAM: + case READ_LEGACY_STREAM: + /* Reading a lz4 stream already failed. */ + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, "Invalid sequence."); + return (ARCHIVE_FATAL); + case READ_DEFAULT_BLOCK: + ret = lz4_filter_read_default_stream(self, p); + if (ret != 0 || state->stage != SELECT_STREAM) + return ret; + break; + case READ_LEGACY_BLOCK: + ret = lz4_filter_read_legacy_stream(self, p); + if (ret != 0 || state->stage != SELECT_STREAM) + return ret; + break; + default: + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, "Program error."); + return (ARCHIVE_FATAL); + break; + } + + while (state->stage == SELECT_STREAM) { + const char *read_buf; + + /* Read a magic number. */ + read_buf = __archive_read_filter_ahead(self->upstream, 4, + NULL); + if (read_buf == NULL) { + state->eof = 1; + *p = NULL; + return (0); + } + uint32_t number = archive_le32dec(read_buf); + __archive_read_filter_consume(self->upstream, 4); + if (number == LZ4_MAGICNUMBER) + return lz4_filter_read_default_stream(self, p); + else if (number == LZ4_LEGACY) + return lz4_filter_read_legacy_stream(self, p); + else if ((number & ~0xF) == LZ4_SKIPPABLED) { + read_buf = __archive_read_filter_ahead( + self->upstream, 4, NULL); + if (read_buf == NULL) { + archive_set_error( + &self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Malformed lz4 data"); + return (ARCHIVE_FATAL); + } + uint32_t skip_bytes = archive_le32dec(read_buf); + __archive_read_filter_consume(self->upstream, + 4 + skip_bytes); + } else { + /* Ignore following unrecognized data. */ + state->eof = 1; + *p = NULL; + return (0); + } + } + state->eof = 1; + *p = NULL; + return (0); +} + +static int +lz4_filter_read_descriptor(struct archive_read_filter *self) +{ + struct private_data *state = (struct private_data *)self->data; + const char *read_buf; + ssize_t bytes_remaining; + ssize_t descriptor_bytes; + unsigned char flag, bd; + unsigned int chsum, chsum_verifier; + + /* Make sure we have 2 bytes for flags. */ + read_buf = __archive_read_filter_ahead(self->upstream, 2, + &bytes_remaining); + if (read_buf == NULL) { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "truncated lz4 input"); + return (ARCHIVE_FATAL); + } + + /* + Parse flags. + */ + flag = (unsigned char)read_buf[0]; + /* Verify version number. */ + if ((flag & 0xc0) != 1<<6) + goto malformed_error; + /* A reserved bit must be zero. */ + if (flag & 0x02) + goto malformed_error; + state->flags.block_independence = (flag & 0x20) != 0; + state->flags.block_checksum = (flag & 0x10)?4:0; + state->flags.stream_size = (flag & 0x08) != 0; + state->flags.stream_checksum = (flag & 0x04) != 0; + state->flags.preset_dictionary = (flag & 0x01) != 0; + + /* BD */ + bd = (unsigned char)read_buf[1]; + /* Reserved bits must be zero. */ + if (bd & 0x8f) + goto malformed_error; + /* Get a maximum block size. */ + switch (read_buf[1] >> 4) { + case 4: /* 64 KB */ + state->flags.block_maximum_size = 64 * 1024; + break; + case 5: /* 256 KB */ + state->flags.block_maximum_size = 256 * 1024; + break; + case 6: /* 1 MB */ + state->flags.block_maximum_size = 1024 * 1024; + break; + case 7: /* 4 MB */ + state->flags.block_maximum_size = 4 * 1024 * 1024; + break; + default: + goto malformed_error; + } + + /* Read the whole descriptor in a stream block. */ + descriptor_bytes = 3; + if (state->flags.stream_size) + descriptor_bytes += 8; + if (state->flags.preset_dictionary) + descriptor_bytes += 4; + if (bytes_remaining < descriptor_bytes) { + read_buf = __archive_read_filter_ahead(self->upstream, + descriptor_bytes, &bytes_remaining); + if (read_buf == NULL) { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "truncated lz4 input"); + return (ARCHIVE_FATAL); + } + } + /* Check if a descriptor is corrupted */ + chsum = __archive_xxhash.XXH32(read_buf, (int)descriptor_bytes -1, 0); + chsum = (chsum >> 8) & 0xff; + chsum_verifier = read_buf[descriptor_bytes-1] & 0xff; + if (chsum != chsum_verifier) + goto malformed_error; + + __archive_read_filter_consume(self->upstream, descriptor_bytes); + + /* Make sure we have a large enough buffer for uncompressed data. */ + if (lz4_allocate_out_block(self) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + if (state->flags.stream_checksum) + state->xxh32_state = __archive_xxhash.XXH32_init(0); + + state->decoded_size = 0; + /* Success */ + return (ARCHIVE_OK); +malformed_error: + archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, + "malformed lz4 data"); + return (ARCHIVE_FATAL); +} + +static ssize_t +lz4_filter_read_data_block(struct archive_read_filter *self, const void **p) +{ + struct private_data *state = (struct private_data *)self->data; + ssize_t compressed_size; + const char *read_buf; + ssize_t bytes_remaining; + int checksum_size; + ssize_t uncompressed_size; + size_t prefix64k; + + *p = NULL; + + /* Make sure we have 4 bytes for a block size. */ + read_buf = __archive_read_filter_ahead(self->upstream, 4, + &bytes_remaining); + if (read_buf == NULL) + goto truncated_error; + compressed_size = archive_le32dec(read_buf); + if ((compressed_size & 0x7fffffff) > state->flags.block_maximum_size) + goto malformed_error; + /* A compressed size == 0 means the end of stream blocks. */ + if (compressed_size == 0) { + __archive_read_filter_consume(self->upstream, 4); + return 0; + } + + checksum_size = state->flags.block_checksum; + /* Check if the block is uncompressed. */ + if (compressed_size & 0x80000000U) { + compressed_size &= 0x7fffffff; + uncompressed_size = compressed_size; + } else + uncompressed_size = 0;/* Unknown yet. */ + + /* + Unfortunately, lz4 decompression API requires a whole block + for its decompression speed, so we read a whole block and allocate + a huge buffer used for decoded data. + */ + read_buf = __archive_read_filter_ahead(self->upstream, + 4 + compressed_size + checksum_size, &bytes_remaining); + if (read_buf == NULL) + goto truncated_error; + + /* Optional processing, checking a block sum. */ + if (checksum_size) { + unsigned int chsum = __archive_xxhash.XXH32( + read_buf + 4, (int)compressed_size, 0); + unsigned int chsum_block = + archive_le32dec(read_buf + 4 + compressed_size); + if (chsum != chsum_block) + goto malformed_error; + } + + + /* If the block is uncompressed, there is nothing to do. */ + if (uncompressed_size) { + /* Prepare a prefix 64k block for next block. */ + if (!state->flags.block_independence) { + prefix64k = 64 * 1024; + if (uncompressed_size < (ssize_t)prefix64k) { + memcpy(state->out_block + + prefix64k - uncompressed_size, + read_buf + 4, + uncompressed_size); + memset(state->out_block, 0, + prefix64k - uncompressed_size); + } else { + memcpy(state->out_block, + read_buf + 4 + + uncompressed_size - prefix64k, + prefix64k); + } + state->decoded_size = 0; + } + state->unconsumed = 4 + uncompressed_size + checksum_size; + *p = read_buf + 4; + return uncompressed_size; + } + + /* + Decompress a block data. + */ + if (state->flags.block_independence) { + prefix64k = 0; + uncompressed_size = LZ4_decompress_safe(read_buf + 4, + state->out_block, (int)compressed_size, + state->flags.block_maximum_size); + } else { + prefix64k = 64 * 1024; + if (state->decoded_size) { + if (state->decoded_size < prefix64k) { + memmove(state->out_block + + prefix64k - state->decoded_size, + state->out_block + prefix64k, + state->decoded_size); + memset(state->out_block, 0, + prefix64k - state->decoded_size); + } else { + memmove(state->out_block, + state->out_block + state->decoded_size, + prefix64k); + } + } +#if LZ4_VERSION_MAJOR >= 1 && LZ4_VERSION_MINOR >= 7 + uncompressed_size = LZ4_decompress_safe_usingDict( + read_buf + 4, + state->out_block + prefix64k, (int)compressed_size, + state->flags.block_maximum_size, + state->out_block, + prefix64k); +#else + uncompressed_size = LZ4_decompress_safe_withPrefix64k( + read_buf + 4, + state->out_block + prefix64k, (int)compressed_size, + state->flags.block_maximum_size); +#endif + } + + /* Check if an error occurred in the decompression process. */ + if (uncompressed_size < 0) { + archive_set_error(&(self->archive->archive), + ARCHIVE_ERRNO_MISC, "lz4 decompression failed"); + return (ARCHIVE_FATAL); + } + + state->unconsumed = 4 + compressed_size + checksum_size; + *p = state->out_block + prefix64k; + state->decoded_size = uncompressed_size; + return uncompressed_size; + +malformed_error: + archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, + "malformed lz4 data"); + return (ARCHIVE_FATAL); +truncated_error: + archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, + "truncated lz4 input"); + return (ARCHIVE_FATAL); +} + +static ssize_t +lz4_filter_read_default_stream(struct archive_read_filter *self, const void **p) +{ + struct private_data *state = (struct private_data *)self->data; + const char *read_buf; + ssize_t bytes_remaining; + ssize_t ret; + + if (state->stage == SELECT_STREAM) { + state->stage = READ_DEFAULT_STREAM; + /* First, read a descriptor. */ + if((ret = lz4_filter_read_descriptor(self)) != ARCHIVE_OK) + return (ret); + state->stage = READ_DEFAULT_BLOCK; + } + /* Decompress a block. */ + ret = lz4_filter_read_data_block(self, p); + + /* If the end of block is detected, change the filter status + to read next stream. */ + if (ret == 0 && *p == NULL) + state->stage = SELECT_STREAM; + + /* Optional processing, checking a stream sum. */ + if (state->flags.stream_checksum) { + if (state->stage == SELECT_STREAM) { + unsigned int checksum; + unsigned int checksum_stream; + read_buf = __archive_read_filter_ahead(self->upstream, + 4, &bytes_remaining); + if (read_buf == NULL) { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, "truncated lz4 input"); + return (ARCHIVE_FATAL); + } + checksum = archive_le32dec(read_buf); + __archive_read_filter_consume(self->upstream, 4); + checksum_stream = __archive_xxhash.XXH32_digest( + state->xxh32_state); + state->xxh32_state = NULL; + if (checksum != checksum_stream) { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "lz4 stream checksum error"); + return (ARCHIVE_FATAL); + } + } else if (ret > 0) + __archive_xxhash.XXH32_update(state->xxh32_state, + *p, (int)ret); + } + return (ret); +} + +static ssize_t +lz4_filter_read_legacy_stream(struct archive_read_filter *self, const void **p) +{ + struct private_data *state = (struct private_data *)self->data; + uint32_t compressed; + const char *read_buf; + ssize_t ret; + + *p = NULL; + ret = lz4_allocate_out_block_for_legacy(self); + if (ret != ARCHIVE_OK) + return ret; + + /* Make sure we have 4 bytes for a block size. */ + read_buf = __archive_read_filter_ahead(self->upstream, 4, NULL); + if (read_buf == NULL) { + if (state->stage == SELECT_STREAM) { + state->stage = READ_LEGACY_STREAM; + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "truncated lz4 input"); + return (ARCHIVE_FATAL); + } + state->stage = SELECT_STREAM; + return 0; + } + state->stage = READ_LEGACY_BLOCK; + compressed = archive_le32dec(read_buf); + if (compressed > LZ4_COMPRESSBOUND(LEGACY_BLOCK_SIZE)) { + state->stage = SELECT_STREAM; + return 0; + } + + /* Make sure we have a whole block. */ + read_buf = __archive_read_filter_ahead(self->upstream, + 4 + compressed, NULL); + if (read_buf == NULL) { + archive_set_error(&(self->archive->archive), + ARCHIVE_ERRNO_MISC, "truncated lz4 input"); + return (ARCHIVE_FATAL); + } + ret = LZ4_decompress_safe(read_buf + 4, state->out_block, + compressed, (int)state->out_block_size); + if (ret < 0) { + archive_set_error(&(self->archive->archive), + ARCHIVE_ERRNO_MISC, "lz4 decompression failed"); + return (ARCHIVE_FATAL); + } + *p = state->out_block; + state->unconsumed = 4 + compressed; + return ret; +} + +/* + * Clean up the decompressor. + */ +static int +lz4_filter_close(struct archive_read_filter *self) +{ + struct private_data *state; + int ret = ARCHIVE_OK; + + state = (struct private_data *)self->data; + free(state->xxh32_state); + free(state->out_block); + free(state); + return (ret); +} + +#endif /* HAVE_LIBLZ4 */ diff --git a/c/archive_read_support_filter_lzop.c b/c/archive_read_support_filter_lzop.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_filter_lzop.c @@ -0,0 +1,494 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * Copyright (c) 2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_UNISTD_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif +#ifdef HAVE_LZO_LZOCONF_H +#include +#endif +#ifdef HAVE_LZO_LZO1X_H +#include +#endif +#ifdef HAVE_ZLIB_H +#include /* for crc32 and adler32 */ +#endif + +#include "archive.h" +#if !defined(HAVE_ZLIB_H) &&\ + defined(HAVE_LZO_LZOCONF_H) && defined(HAVE_LZO_LZO1X_H) +#include "archive_crc32.h" +#endif +#include "archive_endian.h" +#include "archive_private.h" +#include "archive_read_private.h" + +#ifndef HAVE_ZLIB_H +#define adler32 lzo_adler32 +#endif + +#define LZOP_HEADER_MAGIC "\x89\x4c\x5a\x4f\x00\x0d\x0a\x1a\x0a" +#define LZOP_HEADER_MAGIC_LEN 9 + +#if defined(HAVE_LZO_LZOCONF_H) && defined(HAVE_LZO_LZO1X_H) +struct read_lzop { + unsigned char *out_block; + size_t out_block_size; + int64_t total_out; + int flags; + uint32_t compressed_cksum; + uint32_t uncompressed_cksum; + size_t compressed_size; + size_t uncompressed_size; + size_t unconsumed_bytes; + char in_stream; + char eof; /* True = found end of compressed data. */ +}; + +#define FILTER 0x0800 +#define CRC32_HEADER 0x1000 +#define EXTRA_FIELD 0x0040 +#define ADLER32_UNCOMPRESSED 0x0001 +#define ADLER32_COMPRESSED 0x0002 +#define CRC32_UNCOMPRESSED 0x0100 +#define CRC32_COMPRESSED 0x0200 +#define MAX_BLOCK_SIZE (64 * 1024 * 1024) + +static ssize_t lzop_filter_read(struct archive_read_filter *, const void **); +static int lzop_filter_close(struct archive_read_filter *); +#endif + +static int lzop_bidder_bid(struct archive_read_filter_bidder *, + struct archive_read_filter *); +static int lzop_bidder_init(struct archive_read_filter *); + +int +archive_read_support_filter_lzop(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_filter_bidder *reader; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_filter_lzop"); + + if (__archive_read_get_bidder(a, &reader) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + reader->data = NULL; + reader->bid = lzop_bidder_bid; + reader->init = lzop_bidder_init; + reader->options = NULL; + reader->free = NULL; + /* Signal the extent of lzop support with the return value here. */ +#if defined(HAVE_LZO_LZOCONF_H) && defined(HAVE_LZO_LZO1X_H) + return (ARCHIVE_OK); +#else + /* Return ARCHIVE_WARN since this always uses an external program. */ + archive_set_error(_a, ARCHIVE_ERRNO_MISC, + "Using external lzop program for lzop decompression"); + return (ARCHIVE_WARN); +#endif +} + +/* + * Bidder just verifies the header and returns the number of verified bits. + */ +static int +lzop_bidder_bid(struct archive_read_filter_bidder *self, + struct archive_read_filter *filter) +{ + const unsigned char *p; + ssize_t avail; + + (void)self; /* UNUSED */ + + p = __archive_read_filter_ahead(filter, LZOP_HEADER_MAGIC_LEN, &avail); + if (p == NULL || avail == 0) + return (0); + + if (memcmp(p, LZOP_HEADER_MAGIC, LZOP_HEADER_MAGIC_LEN)) + return (0); + + return (LZOP_HEADER_MAGIC_LEN * 8); +} + +#if !defined(HAVE_LZO_LZOCONF_H) || !defined(HAVE_LZO_LZO1X_H) +/* + * If we don't have the library on this system, we can't do the + * decompression directly. We can, however, try to run "lzop -d" + * in case that's available. + */ +static int +lzop_bidder_init(struct archive_read_filter *self) +{ + int r; + + r = __archive_read_program(self, "lzop -d"); + /* Note: We set the format here even if __archive_read_program() + * above fails. We do, after all, know what the format is + * even if we weren't able to read it. */ + self->code = ARCHIVE_FILTER_LZOP; + self->name = "lzop"; + return (r); +} +#else +/* + * Initialize the filter object. + */ +static int +lzop_bidder_init(struct archive_read_filter *self) +{ + struct read_lzop *state; + + self->code = ARCHIVE_FILTER_LZOP; + self->name = "lzop"; + + state = (struct read_lzop *)calloc(sizeof(*state), 1); + if (state == NULL) { + archive_set_error(&self->archive->archive, ENOMEM, + "Can't allocate data for lzop decompression"); + return (ARCHIVE_FATAL); + } + + self->data = state; + self->read = lzop_filter_read; + self->skip = NULL; /* not supported */ + self->close = lzop_filter_close; + + return (ARCHIVE_OK); +} + +static int +consume_header(struct archive_read_filter *self) +{ + struct read_lzop *state = (struct read_lzop *)self->data; + const unsigned char *p, *_p; + unsigned checksum, flags, len, method, version; + + /* + * Check LZOP magic code. + */ + p = __archive_read_filter_ahead(self->upstream, + LZOP_HEADER_MAGIC_LEN, NULL); + if (p == NULL) + return (ARCHIVE_EOF); + + if (memcmp(p, LZOP_HEADER_MAGIC, LZOP_HEADER_MAGIC_LEN)) + return (ARCHIVE_EOF); + __archive_read_filter_consume(self->upstream, + LZOP_HEADER_MAGIC_LEN); + + p = __archive_read_filter_ahead(self->upstream, 29, NULL); + if (p == NULL) + goto truncated; + _p = p; + version = archive_be16dec(p); + p += 4;/* version(2 bytes) + library version(2 bytes) */ + + if (version >= 0x940) { + unsigned reqversion = archive_be16dec(p); p += 2; + if (reqversion < 0x900) { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, "Invalid required version"); + return (ARCHIVE_FAILED); + } + } + + method = *p++; + if (method < 1 || method > 3) { + archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, + "Unsupported method"); + return (ARCHIVE_FAILED); + } + + if (version >= 0x940) { + unsigned level = *p++; +#if 0 + unsigned default_level[] = {0, 3, 1, 9}; +#endif + if (level == 0) + /* Method is 1..3 here due to check above. */ +#if 0 /* Avoid an error Clang Static Analyzer claims + "Value stored to 'level' is never read". */ + level = default_level[method]; +#else + ;/* NOP */ +#endif + else if (level > 9) { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, "Invalid level"); + return (ARCHIVE_FAILED); + } + } + + flags = archive_be32dec(p); p += 4; + + if (flags & FILTER) + p += 4; /* Skip filter */ + p += 4; /* Skip mode */ + if (version >= 0x940) + p += 8; /* Skip mtime */ + else + p += 4; /* Skip mtime */ + len = *p++; /* Read filename length */ + len += p - _p; + /* Make sure we have all bytes we need to calculate checksum. */ + p = __archive_read_filter_ahead(self->upstream, len + 4, NULL); + if (p == NULL) + goto truncated; + if (flags & CRC32_HEADER) + checksum = crc32(crc32(0, NULL, 0), p, len); + else + checksum = adler32(adler32(0, NULL, 0), p, len); + if (archive_be32dec(p + len) != checksum) + goto corrupted; + __archive_read_filter_consume(self->upstream, len + 4); + if (flags & EXTRA_FIELD) { + /* Skip extra field */ + p = __archive_read_filter_ahead(self->upstream, 4, NULL); + if (p == NULL) + goto truncated; + len = archive_be32dec(p); + __archive_read_filter_consume(self->upstream, len + 4 + 4); + } + state->flags = flags; + state->in_stream = 1; + return (ARCHIVE_OK); +truncated: + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_FILE_FORMAT, "Truncated lzop data"); + return (ARCHIVE_FAILED); +corrupted: + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_FILE_FORMAT, "Corrupted lzop header"); + return (ARCHIVE_FAILED); +} + +static int +consume_block_info(struct archive_read_filter *self) +{ + struct read_lzop *state = (struct read_lzop *)self->data; + const unsigned char *p; + unsigned flags = state->flags; + + p = __archive_read_filter_ahead(self->upstream, 4, NULL); + if (p == NULL) + goto truncated; + state->uncompressed_size = archive_be32dec(p); + __archive_read_filter_consume(self->upstream, 4); + if (state->uncompressed_size == 0) + return (ARCHIVE_EOF); + if (state->uncompressed_size > MAX_BLOCK_SIZE) + goto corrupted; + + p = __archive_read_filter_ahead(self->upstream, 4, NULL); + if (p == NULL) + goto truncated; + state->compressed_size = archive_be32dec(p); + __archive_read_filter_consume(self->upstream, 4); + if (state->compressed_size > state->uncompressed_size) + goto corrupted; + + if (flags & (CRC32_UNCOMPRESSED | ADLER32_UNCOMPRESSED)) { + p = __archive_read_filter_ahead(self->upstream, 4, NULL); + if (p == NULL) + goto truncated; + state->compressed_cksum = state->uncompressed_cksum = + archive_be32dec(p); + __archive_read_filter_consume(self->upstream, 4); + } + if ((flags & (CRC32_COMPRESSED | ADLER32_COMPRESSED)) && + state->compressed_size < state->uncompressed_size) { + p = __archive_read_filter_ahead(self->upstream, 4, NULL); + if (p == NULL) + goto truncated; + state->compressed_cksum = archive_be32dec(p); + __archive_read_filter_consume(self->upstream, 4); + } + return (ARCHIVE_OK); +truncated: + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_FILE_FORMAT, "Truncated lzop data"); + return (ARCHIVE_FAILED); +corrupted: + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_FILE_FORMAT, "Corrupted lzop header"); + return (ARCHIVE_FAILED); +} + +static ssize_t +lzop_filter_read(struct archive_read_filter *self, const void **p) +{ + struct read_lzop *state = (struct read_lzop *)self->data; + const void *b; + lzo_uint out_size; + uint32_t cksum; + int ret, r; + + if (state->unconsumed_bytes) { + __archive_read_filter_consume(self->upstream, + state->unconsumed_bytes); + state->unconsumed_bytes = 0; + } + if (state->eof) + return (0); + + for (;;) { + if (!state->in_stream) { + ret = consume_header(self); + if (ret < ARCHIVE_OK) + return (ret); + if (ret == ARCHIVE_EOF) { + state->eof = 1; + return (0); + } + } + ret = consume_block_info(self); + if (ret < ARCHIVE_OK) + return (ret); + if (ret == ARCHIVE_EOF) + state->in_stream = 0; + else + break; + } + + if (state->out_block == NULL || + state->out_block_size < state->uncompressed_size) { + void *new_block; + + new_block = realloc(state->out_block, state->uncompressed_size); + if (new_block == NULL) { + archive_set_error(&self->archive->archive, ENOMEM, + "Can't allocate data for lzop decompression"); + return (ARCHIVE_FATAL); + } + state->out_block = new_block; + state->out_block_size = state->uncompressed_size; + } + + b = __archive_read_filter_ahead(self->upstream, + state->compressed_size, NULL); + if (b == NULL) { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_FILE_FORMAT, "Truncated lzop data"); + return (ARCHIVE_FATAL); + } + if (state->flags & CRC32_COMPRESSED) + cksum = crc32(crc32(0, NULL, 0), b, state->compressed_size); + else if (state->flags & ADLER32_COMPRESSED) + cksum = adler32(adler32(0, NULL, 0), b, state->compressed_size); + else + cksum = state->compressed_cksum; + if (cksum != state->compressed_cksum) { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, "Corrupted data"); + return (ARCHIVE_FATAL); + } + + /* + * If the both uncompressed size and compressed size are the same, + * we do not decompress this block. + */ + if (state->uncompressed_size == state->compressed_size) { + *p = b; + state->total_out += state->compressed_size; + state->unconsumed_bytes = state->compressed_size; + return ((ssize_t)state->uncompressed_size); + } + + /* + * Drive lzo uncompression. + */ + out_size = (lzo_uint)state->uncompressed_size; + r = lzo1x_decompress_safe(b, (lzo_uint)state->compressed_size, + state->out_block, &out_size, NULL); + switch (r) { + case LZO_E_OK: + if (out_size == state->uncompressed_size) + break; + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, "Corrupted data"); + return (ARCHIVE_FATAL); + case LZO_E_OUT_OF_MEMORY: + archive_set_error(&self->archive->archive, ENOMEM, + "lzop decompression failed: out of memory"); + return (ARCHIVE_FATAL); + default: + archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, + "lzop decompression failed: %d", r); + return (ARCHIVE_FATAL); + } + + if (state->flags & CRC32_UNCOMPRESSED) + cksum = crc32(crc32(0, NULL, 0), state->out_block, + state->uncompressed_size); + else if (state->flags & ADLER32_UNCOMPRESSED) + cksum = adler32(adler32(0, NULL, 0), state->out_block, + state->uncompressed_size); + else + cksum = state->uncompressed_cksum; + if (cksum != state->uncompressed_cksum) { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, "Corrupted data"); + return (ARCHIVE_FATAL); + } + + __archive_read_filter_consume(self->upstream, state->compressed_size); + *p = state->out_block; + state->total_out += out_size; + return ((ssize_t)out_size); +} + +/* + * Clean up the decompressor. + */ +static int +lzop_filter_close(struct archive_read_filter *self) +{ + struct read_lzop *state = (struct read_lzop *)self->data; + + free(state->out_block); + free(state); + return (ARCHIVE_OK); +} + +#endif diff --git a/c/archive_read_support_filter_none.c b/c/archive_read_support_filter_none.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_filter_none.c @@ -0,0 +1,52 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#include "archive.h" +#include "archive_private.h" + +#if ARCHIVE_VERSION_NUMBER < 4000000 +/* Deprecated; remove in libarchive 4.0 */ +int +archive_read_support_compression_none(struct archive *a) +{ + return archive_read_support_filter_none(a); +} +#endif + +/* + * Uncompressed streams are handled implicitly by the read core, + * so this is now a no-op. + */ +int +archive_read_support_filter_none(struct archive *a) +{ + archive_check_magic(a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_filter_none"); + + return (ARCHIVE_OK); +} diff --git a/c/archive_read_support_filter_program.c b/c/archive_read_support_filter_program.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_filter_program.c @@ -0,0 +1,503 @@ +/*- + * Copyright (c) 2007 Joerg Sonnenberger + * Copyright (c) 2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_SYS_WAIT_H +# include +#endif +#ifdef HAVE_ERRNO_H +# include +#endif +#ifdef HAVE_FCNTL_H +# include +#endif +#ifdef HAVE_LIMITS_H +# include +#endif +#ifdef HAVE_SIGNAL_H +# include +#endif +#ifdef HAVE_STDLIB_H +# include +#endif +#ifdef HAVE_STRING_H +# include +#endif +#ifdef HAVE_UNISTD_H +# include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_string.h" +#include "archive_read_private.h" +#include "filter_fork.h" + + +#if ARCHIVE_VERSION_NUMBER < 4000000 +/* Deprecated; remove in libarchive 4.0 */ +int +archive_read_support_compression_program(struct archive *a, const char *cmd) +{ + return archive_read_support_filter_program(a, cmd); +} + +int +archive_read_support_compression_program_signature(struct archive *a, + const char *cmd, const void *signature, size_t signature_len) +{ + return archive_read_support_filter_program_signature(a, + cmd, signature, signature_len); +} +#endif + +int +archive_read_support_filter_program(struct archive *a, const char *cmd) +{ + return (archive_read_support_filter_program_signature(a, cmd, NULL, 0)); +} + +/* + * The bidder object stores the command and the signature to watch for. + * The 'inhibit' entry here is used to ensure that unchecked filters never + * bid twice in the same pipeline. + */ +struct program_bidder { + char *description; + char *cmd; + void *signature; + size_t signature_len; + int inhibit; +}; + +static int program_bidder_bid(struct archive_read_filter_bidder *, + struct archive_read_filter *upstream); +static int program_bidder_init(struct archive_read_filter *); +static int program_bidder_free(struct archive_read_filter_bidder *); + +/* + * The actual filter needs to track input and output data. + */ +struct program_filter { + struct archive_string description; +#if defined(_WIN32) && !defined(__CYGWIN__) + HANDLE child; +#else + pid_t child; +#endif + int exit_status; + int waitpid_return; + int child_stdin, child_stdout; + + char *out_buf; + size_t out_buf_len; +}; + +static ssize_t program_filter_read(struct archive_read_filter *, + const void **); +static int program_filter_close(struct archive_read_filter *); +static void free_state(struct program_bidder *); + +static int +set_bidder_signature(struct archive_read_filter_bidder *bidder, + struct program_bidder *state, const void *signature, size_t signature_len) +{ + + if (signature != NULL && signature_len > 0) { + state->signature_len = signature_len; + state->signature = malloc(signature_len); + memcpy(state->signature, signature, signature_len); + } + + /* + * Fill in the bidder object. + */ + bidder->data = state; + bidder->bid = program_bidder_bid; + bidder->init = program_bidder_init; + bidder->options = NULL; + bidder->free = program_bidder_free; + return (ARCHIVE_OK); +} + +int +archive_read_support_filter_program_signature(struct archive *_a, + const char *cmd, const void *signature, size_t signature_len) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_filter_bidder *bidder; + struct program_bidder *state; + + /* + * Get a bidder object from the read core. + */ + if (__archive_read_get_bidder(a, &bidder) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + /* + * Allocate our private state. + */ + state = (struct program_bidder *)calloc(1, sizeof (*state)); + if (state == NULL) + goto memerr; + state->cmd = strdup(cmd); + if (state->cmd == NULL) + goto memerr; + + return set_bidder_signature(bidder, state, signature, signature_len); +memerr: + free_state(state); + archive_set_error(_a, ENOMEM, "Can't allocate memory"); + return (ARCHIVE_FATAL); +} + +static int +program_bidder_free(struct archive_read_filter_bidder *self) +{ + struct program_bidder *state = (struct program_bidder *)self->data; + + free_state(state); + return (ARCHIVE_OK); +} + +static void +free_state(struct program_bidder *state) +{ + + if (state) { + free(state->cmd); + free(state->signature); + free(state); + } +} + +/* + * If we do have a signature, bid only if that matches. + * + * If there's no signature, we bid INT_MAX the first time + * we're called, then never bid again. + */ +static int +program_bidder_bid(struct archive_read_filter_bidder *self, + struct archive_read_filter *upstream) +{ + struct program_bidder *state = self->data; + const char *p; + + /* If we have a signature, use that to match. */ + if (state->signature_len > 0) { + p = __archive_read_filter_ahead(upstream, + state->signature_len, NULL); + if (p == NULL) + return (0); + /* No match, so don't bid. */ + if (memcmp(p, state->signature, state->signature_len) != 0) + return (0); + return ((int)state->signature_len * 8); + } + + /* Otherwise, bid once and then never bid again. */ + if (state->inhibit) + return (0); + state->inhibit = 1; + return (INT_MAX); +} + +/* + * Shut down the child, return ARCHIVE_OK if it exited normally. + * + * Note that the return value is sticky; if we're called again, + * we won't reap the child again, but we will return the same status + * (including error message if the child came to a bad end). + */ +static int +child_stop(struct archive_read_filter *self, struct program_filter *state) +{ + /* Close our side of the I/O with the child. */ + if (state->child_stdin != -1) { + close(state->child_stdin); + state->child_stdin = -1; + } + if (state->child_stdout != -1) { + close(state->child_stdout); + state->child_stdout = -1; + } + + if (state->child != 0) { + /* Reap the child. */ + do { + state->waitpid_return + = waitpid(state->child, &state->exit_status, 0); + } while (state->waitpid_return == -1 && errno == EINTR); +#if defined(_WIN32) && !defined(__CYGWIN__) + CloseHandle(state->child); +#endif + state->child = 0; + } + + if (state->waitpid_return < 0) { + /* waitpid() failed? This is ugly. */ + archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, + "Child process exited badly"); + return (ARCHIVE_WARN); + } + +#if !defined(_WIN32) || defined(__CYGWIN__) + if (WIFSIGNALED(state->exit_status)) { +#ifdef SIGPIPE + /* If the child died because we stopped reading before + * it was done, that's okay. Some archive formats + * have padding at the end that we routinely ignore. */ + /* The alternative to this would be to add a step + * before close(child_stdout) above to read from the + * child until the child has no more to write. */ + if (WTERMSIG(state->exit_status) == SIGPIPE) + return (ARCHIVE_OK); +#endif + archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, + "Child process exited with signal %d", + WTERMSIG(state->exit_status)); + return (ARCHIVE_WARN); + } +#endif /* !_WIN32 || __CYGWIN__ */ + + if (WIFEXITED(state->exit_status)) { + if (WEXITSTATUS(state->exit_status) == 0) + return (ARCHIVE_OK); + + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Child process exited with status %d", + WEXITSTATUS(state->exit_status)); + return (ARCHIVE_WARN); + } + + return (ARCHIVE_WARN); +} + +/* + * Use select() to decide whether the child is ready for read or write. + */ +static ssize_t +child_read(struct archive_read_filter *self, char *buf, size_t buf_len) +{ + struct program_filter *state = self->data; + ssize_t ret, requested, avail; + const char *p; +#if defined(_WIN32) && !defined(__CYGWIN__) + HANDLE handle = (HANDLE)_get_osfhandle(state->child_stdout); +#endif + + requested = buf_len > SSIZE_MAX ? SSIZE_MAX : buf_len; + + for (;;) { + do { +#if defined(_WIN32) && !defined(__CYGWIN__) + /* Avoid infinity wait. + * Note: If there is no data in the pipe, ReadFile() + * called in read() never returns and so we won't + * write remaining encoded data to the pipe. + * Note: This way may cause performance problem. + * we are looking forward to great code to resolve + * this. */ + DWORD pipe_avail = -1; + int cnt = 2; + + while (PeekNamedPipe(handle, NULL, 0, NULL, + &pipe_avail, NULL) != 0 && pipe_avail == 0 && + cnt--) + Sleep(5); + if (pipe_avail == 0) { + ret = -1; + errno = EAGAIN; + break; + } +#endif + ret = read(state->child_stdout, buf, requested); + } while (ret == -1 && errno == EINTR); + + if (ret > 0) + return (ret); + if (ret == 0 || (ret == -1 && errno == EPIPE)) + /* Child has closed its output; reap the child + * and return the status. */ + return (child_stop(self, state)); + if (ret == -1 && errno != EAGAIN) + return (-1); + + if (state->child_stdin == -1) { + /* Block until child has some I/O ready. */ + __archive_check_child(state->child_stdin, + state->child_stdout); + continue; + } + + /* Get some more data from upstream. */ + p = __archive_read_filter_ahead(self->upstream, 1, &avail); + if (p == NULL) { + close(state->child_stdin); + state->child_stdin = -1; + fcntl(state->child_stdout, F_SETFL, 0); + if (avail < 0) + return (avail); + continue; + } + + do { + ret = write(state->child_stdin, p, avail); + } while (ret == -1 && errno == EINTR); + + if (ret > 0) { + /* Consume whatever we managed to write. */ + __archive_read_filter_consume(self->upstream, ret); + } else if (ret == -1 && errno == EAGAIN) { + /* Block until child has some I/O ready. */ + __archive_check_child(state->child_stdin, + state->child_stdout); + } else { + /* Write failed. */ + close(state->child_stdin); + state->child_stdin = -1; + fcntl(state->child_stdout, F_SETFL, 0); + /* If it was a bad error, we're done; otherwise + * it was EPIPE or EOF, and we can still read + * from the child. */ + if (ret == -1 && errno != EPIPE) + return (-1); + } + } +} + +int +__archive_read_program(struct archive_read_filter *self, const char *cmd) +{ + struct program_filter *state; + static const size_t out_buf_len = 65536; + char *out_buf; + const char *prefix = "Program: "; + int ret; + size_t l; + + l = strlen(prefix) + strlen(cmd) + 1; + state = (struct program_filter *)calloc(1, sizeof(*state)); + out_buf = (char *)malloc(out_buf_len); + if (state == NULL || out_buf == NULL || + archive_string_ensure(&state->description, l) == NULL) { + archive_set_error(&self->archive->archive, ENOMEM, + "Can't allocate input data"); + if (state != NULL) { + archive_string_free(&state->description); + free(state); + } + free(out_buf); + return (ARCHIVE_FATAL); + } + archive_strcpy(&state->description, prefix); + archive_strcat(&state->description, cmd); + + self->code = ARCHIVE_FILTER_PROGRAM; + self->name = state->description.s; + + state->out_buf = out_buf; + state->out_buf_len = out_buf_len; + + ret = __archive_create_child(cmd, &state->child_stdin, + &state->child_stdout, &state->child); + if (ret != ARCHIVE_OK) { + free(state->out_buf); + archive_string_free(&state->description); + free(state); + archive_set_error(&self->archive->archive, EINVAL, + "Can't initialize filter; unable to run program \"%s\"", + cmd); + return (ARCHIVE_FATAL); + } + + self->data = state; + self->read = program_filter_read; + self->skip = NULL; + self->close = program_filter_close; + + /* XXX Check that we can read at least one byte? */ + return (ARCHIVE_OK); +} + +static int +program_bidder_init(struct archive_read_filter *self) +{ + struct program_bidder *bidder_state; + + bidder_state = (struct program_bidder *)self->bidder->data; + return (__archive_read_program(self, bidder_state->cmd)); +} + +static ssize_t +program_filter_read(struct archive_read_filter *self, const void **buff) +{ + struct program_filter *state; + ssize_t bytes; + size_t total; + char *p; + + state = (struct program_filter *)self->data; + + total = 0; + p = state->out_buf; + while (state->child_stdout != -1 && total < state->out_buf_len) { + bytes = child_read(self, p, state->out_buf_len - total); + if (bytes < 0) + /* No recovery is possible if we can no longer + * read from the child. */ + return (ARCHIVE_FATAL); + if (bytes == 0) + /* We got EOF from the child. */ + break; + total += bytes; + p += bytes; + } + + *buff = state->out_buf; + return (total); +} + +static int +program_filter_close(struct archive_read_filter *self) +{ + struct program_filter *state; + int e; + + state = (struct program_filter *)self->data; + e = child_stop(self, state); + + /* Release our private data. */ + free(state->out_buf); + archive_string_free(&state->description); + free(state); + + return (e); +} diff --git a/c/archive_read_support_filter_rpm.c b/c/archive_read_support_filter_rpm.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_filter_rpm.c @@ -0,0 +1,289 @@ +/*- + * Copyright (c) 2009 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif + +#include "archive.h" +#include "archive_endian.h" +#include "archive_private.h" +#include "archive_read_private.h" + +struct rpm { + int64_t total_in; + size_t hpos; + size_t hlen; + unsigned char header[16]; + enum { + ST_LEAD, /* Skipping 'Lead' section. */ + ST_HEADER, /* Reading 'Header' section; + * first 16 bytes. */ + ST_HEADER_DATA, /* Skipping 'Header' section. */ + ST_PADDING, /* Skipping padding data after the + * 'Header' section. */ + ST_ARCHIVE /* Reading 'Archive' section. */ + } state; + int first_header; +}; +#define RPM_LEAD_SIZE 96 /* Size of 'Lead' section. */ + +static int rpm_bidder_bid(struct archive_read_filter_bidder *, + struct archive_read_filter *); +static int rpm_bidder_init(struct archive_read_filter *); + +static ssize_t rpm_filter_read(struct archive_read_filter *, + const void **); +static int rpm_filter_close(struct archive_read_filter *); + +#if ARCHIVE_VERSION_NUMBER < 4000000 +/* Deprecated; remove in libarchive 4.0 */ +int +archive_read_support_compression_rpm(struct archive *a) +{ + return archive_read_support_filter_rpm(a); +} +#endif + +int +archive_read_support_filter_rpm(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_filter_bidder *bidder; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_filter_rpm"); + + if (__archive_read_get_bidder(a, &bidder) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + bidder->data = NULL; + bidder->name = "rpm"; + bidder->bid = rpm_bidder_bid; + bidder->init = rpm_bidder_init; + bidder->options = NULL; + bidder->free = NULL; + return (ARCHIVE_OK); +} + +static int +rpm_bidder_bid(struct archive_read_filter_bidder *self, + struct archive_read_filter *filter) +{ + const unsigned char *b; + ssize_t avail; + int bits_checked; + + (void)self; /* UNUSED */ + + b = __archive_read_filter_ahead(filter, 8, &avail); + if (b == NULL) + return (0); + + bits_checked = 0; + /* + * Verify Header Magic Bytes : 0XED 0XAB 0XEE 0XDB + */ + if (memcmp(b, "\xED\xAB\xEE\xDB", 4) != 0) + return (0); + bits_checked += 32; + /* + * Check major version. + */ + if (b[4] != 3 && b[4] != 4) + return (0); + bits_checked += 8; + /* + * Check package type; binary or source. + */ + if (b[6] != 0) + return (0); + bits_checked += 8; + if (b[7] != 0 && b[7] != 1) + return (0); + bits_checked += 8; + + return (bits_checked); +} + +static int +rpm_bidder_init(struct archive_read_filter *self) +{ + struct rpm *rpm; + + self->code = ARCHIVE_FILTER_RPM; + self->name = "rpm"; + self->read = rpm_filter_read; + self->skip = NULL; /* not supported */ + self->close = rpm_filter_close; + + rpm = (struct rpm *)calloc(sizeof(*rpm), 1); + if (rpm == NULL) { + archive_set_error(&self->archive->archive, ENOMEM, + "Can't allocate data for rpm"); + return (ARCHIVE_FATAL); + } + + self->data = rpm; + rpm->state = ST_LEAD; + + return (ARCHIVE_OK); +} + +static ssize_t +rpm_filter_read(struct archive_read_filter *self, const void **buff) +{ + struct rpm *rpm; + const unsigned char *b; + ssize_t avail_in, total; + size_t used, n; + uint32_t section; + uint32_t bytes; + + rpm = (struct rpm *)self->data; + *buff = NULL; + total = avail_in = 0; + b = NULL; + used = 0; + do { + if (b == NULL) { + b = __archive_read_filter_ahead(self->upstream, 1, + &avail_in); + if (b == NULL) { + if (avail_in < 0) + return (ARCHIVE_FATAL); + else + break; + } + } + + switch (rpm->state) { + case ST_LEAD: + if (rpm->total_in + avail_in < RPM_LEAD_SIZE) + used += avail_in; + else { + n = (size_t)(RPM_LEAD_SIZE - rpm->total_in); + used += n; + b += n; + rpm->state = ST_HEADER; + rpm->hpos = 0; + rpm->hlen = 0; + rpm->first_header = 1; + } + break; + case ST_HEADER: + n = 16 - rpm->hpos; + if (n > avail_in - used) + n = avail_in - used; + memcpy(rpm->header+rpm->hpos, b, n); + b += n; + used += n; + rpm->hpos += n; + + if (rpm->hpos == 16) { + if (rpm->header[0] != 0x8e || + rpm->header[1] != 0xad || + rpm->header[2] != 0xe8 || + rpm->header[3] != 0x01) { + if (rpm->first_header) { + archive_set_error( + &self->archive->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Unrecoginized rpm header"); + return (ARCHIVE_FATAL); + } + rpm->state = ST_ARCHIVE; + *buff = rpm->header; + total = rpm->hpos; + break; + } + /* Calculate 'Header' length. */ + section = archive_be32dec(rpm->header+8); + bytes = archive_be32dec(rpm->header+12); + rpm->hlen = 16 + section * 16 + bytes; + rpm->state = ST_HEADER_DATA; + rpm->first_header = 0; + } + break; + case ST_HEADER_DATA: + n = rpm->hlen - rpm->hpos; + if (n > avail_in - used) + n = avail_in - used; + b += n; + used += n; + rpm->hpos += n; + if (rpm->hpos == rpm->hlen) + rpm->state = ST_PADDING; + break; + case ST_PADDING: + while (used < (size_t)avail_in) { + if (*b != 0) { + /* Read next header. */ + rpm->state = ST_HEADER; + rpm->hpos = 0; + rpm->hlen = 0; + break; + } + b++; + used++; + } + break; + case ST_ARCHIVE: + *buff = b; + total = avail_in; + used = avail_in; + break; + } + if (used == (size_t)avail_in) { + rpm->total_in += used; + __archive_read_filter_consume(self->upstream, used); + b = NULL; + used = 0; + } + } while (total == 0 && avail_in > 0); + + if (used > 0 && b != NULL) { + rpm->total_in += used; + __archive_read_filter_consume(self->upstream, used); + } + return (total); +} + +static int +rpm_filter_close(struct archive_read_filter *self) +{ + struct rpm *rpm; + + rpm = (struct rpm *)self->data; + free(rpm); + + return (ARCHIVE_OK); +} + diff --git a/c/archive_read_support_filter_uu.c b/c/archive_read_support_filter_uu.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_filter_uu.c @@ -0,0 +1,685 @@ +/*- + * Copyright (c) 2009-2011 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_read_private.h" + +/* Maximum lookahead during bid phase */ +#define UUENCODE_BID_MAX_READ 128*1024 /* in bytes */ + +struct uudecode { + int64_t total; + unsigned char *in_buff; +#define IN_BUFF_SIZE (1024) + int in_cnt; + size_t in_allocated; + unsigned char *out_buff; +#define OUT_BUFF_SIZE (64 * 1024) + int state; +#define ST_FIND_HEAD 0 +#define ST_READ_UU 1 +#define ST_UUEND 2 +#define ST_READ_BASE64 3 +#define ST_IGNORE 4 +}; + +static int uudecode_bidder_bid(struct archive_read_filter_bidder *, + struct archive_read_filter *filter); +static int uudecode_bidder_init(struct archive_read_filter *); + +static ssize_t uudecode_filter_read(struct archive_read_filter *, + const void **); +static int uudecode_filter_close(struct archive_read_filter *); + +#if ARCHIVE_VERSION_NUMBER < 4000000 +/* Deprecated; remove in libarchive 4.0 */ +int +archive_read_support_compression_uu(struct archive *a) +{ + return archive_read_support_filter_uu(a); +} +#endif + +int +archive_read_support_filter_uu(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_filter_bidder *bidder; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_filter_uu"); + + if (__archive_read_get_bidder(a, &bidder) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + bidder->data = NULL; + bidder->name = "uu"; + bidder->bid = uudecode_bidder_bid; + bidder->init = uudecode_bidder_init; + bidder->options = NULL; + bidder->free = NULL; + return (ARCHIVE_OK); +} + +static const unsigned char ascii[256] = { + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, '\n', 0, 0, '\r', 0, 0, /* 00 - 0F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 10 - 1F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 20 - 2F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 30 - 3F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 40 - 4F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 50 - 5F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 60 - 6F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, /* 70 - 7F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 80 - 8F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 90 - 9F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* A0 - AF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* B0 - BF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* C0 - CF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* D0 - DF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* E0 - EF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* F0 - FF */ +}; + +static const unsigned char uuchar[256] = { + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 00 - 0F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 10 - 1F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 20 - 2F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 30 - 3F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 40 - 4F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 50 - 5F */ + 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 60 - 6F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 70 - 7F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 80 - 8F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 90 - 9F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* A0 - AF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* B0 - BF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* C0 - CF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* D0 - DF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* E0 - EF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* F0 - FF */ +}; + +static const unsigned char base64[256] = { + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 00 - 0F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 10 - 1F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, /* 20 - 2F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0, /* 30 - 3F */ + 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 40 - 4F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, /* 50 - 5F */ + 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 60 - 6F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, /* 70 - 7F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 80 - 8F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 90 - 9F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* A0 - AF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* B0 - BF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* C0 - CF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* D0 - DF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* E0 - EF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* F0 - FF */ +}; + +static const int base64num[128] = { + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, /* 00 - 0F */ + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, /* 10 - 1F */ + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 62, 0, 0, 0, 63, /* 20 - 2F */ + 52, 53, 54, 55, 56, 57, 58, 59, + 60, 61, 0, 0, 0, 0, 0, 0, /* 30 - 3F */ + 0, 0, 1, 2, 3, 4, 5, 6, + 7, 8, 9, 10, 11, 12, 13, 14, /* 40 - 4F */ + 15, 16, 17, 18, 19, 20, 21, 22, + 23, 24, 25, 0, 0, 0, 0, 0, /* 50 - 5F */ + 0, 26, 27, 28, 29, 30, 31, 32, + 33, 34, 35, 36, 37, 38, 39, 40, /* 60 - 6F */ + 41, 42, 43, 44, 45, 46, 47, 48, + 49, 50, 51, 0, 0, 0, 0, 0, /* 70 - 7F */ +}; + +static ssize_t +get_line(const unsigned char *b, ssize_t avail, ssize_t *nlsize) +{ + ssize_t len; + + len = 0; + while (len < avail) { + switch (ascii[*b]) { + case 0: /* Non-ascii character or control character. */ + if (nlsize != NULL) + *nlsize = 0; + return (-1); + case '\r': + if (avail-len > 1 && b[1] == '\n') { + if (nlsize != NULL) + *nlsize = 2; + return (len+2); + } + /* FALL THROUGH */ + case '\n': + if (nlsize != NULL) + *nlsize = 1; + return (len+1); + case 1: + b++; + len++; + break; + } + } + if (nlsize != NULL) + *nlsize = 0; + return (avail); +} + +static ssize_t +bid_get_line(struct archive_read_filter *filter, + const unsigned char **b, ssize_t *avail, ssize_t *ravail, + ssize_t *nl, size_t* nbytes_read) +{ + ssize_t len; + int quit; + + quit = 0; + if (*avail == 0) { + *nl = 0; + len = 0; + } else + len = get_line(*b, *avail, nl); + + /* + * Read bytes more while it does not reach the end of line. + */ + while (*nl == 0 && len == *avail && !quit && + *nbytes_read < UUENCODE_BID_MAX_READ) { + ssize_t diff = *ravail - *avail; + size_t nbytes_req = (*ravail+1023) & ~1023U; + ssize_t tested; + + /* Increase reading bytes if it is not enough to at least + * new two lines. */ + if (nbytes_req < (size_t)*ravail + 160) + nbytes_req <<= 1; + + *b = __archive_read_filter_ahead(filter, nbytes_req, avail); + if (*b == NULL) { + if (*ravail >= *avail) + return (0); + /* Reading bytes reaches the end of a stream. */ + *b = __archive_read_filter_ahead(filter, *avail, avail); + quit = 1; + } + *nbytes_read = *avail; + *ravail = *avail; + *b += diff; + *avail -= diff; + tested = len;/* Skip some bytes we already determinated. */ + len = get_line(*b + tested, *avail - tested, nl); + if (len >= 0) + len += tested; + } + return (len); +} + +#define UUDECODE(c) (((c) - 0x20) & 0x3f) + +static int +uudecode_bidder_bid(struct archive_read_filter_bidder *self, + struct archive_read_filter *filter) +{ + const unsigned char *b; + ssize_t avail, ravail; + ssize_t len, nl; + int l; + int firstline; + size_t nbytes_read; + + (void)self; /* UNUSED */ + + b = __archive_read_filter_ahead(filter, 1, &avail); + if (b == NULL) + return (0); + + firstline = 20; + ravail = avail; + nbytes_read = avail; + for (;;) { + len = bid_get_line(filter, &b, &avail, &ravail, &nl, &nbytes_read); + if (len < 0 || nl == 0) + return (0); /* No match found. */ + if (len - nl >= 11 && memcmp(b, "begin ", 6) == 0) + l = 6; + else if (len -nl >= 18 && memcmp(b, "begin-base64 ", 13) == 0) + l = 13; + else + l = 0; + + if (l > 0 && (b[l] < '0' || b[l] > '7' || + b[l+1] < '0' || b[l+1] > '7' || + b[l+2] < '0' || b[l+2] > '7' || b[l+3] != ' ')) + l = 0; + + b += len; + avail -= len; + if (l) + break; + firstline = 0; + + /* Do not read more than UUENCODE_BID_MAX_READ bytes */ + if (nbytes_read >= UUENCODE_BID_MAX_READ) + return (0); + } + if (!avail) + return (0); + len = bid_get_line(filter, &b, &avail, &ravail, &nl, &nbytes_read); + if (len < 0 || nl == 0) + return (0);/* There are non-ascii characters. */ + avail -= len; + + if (l == 6) { + /* "begin " */ + if (!uuchar[*b]) + return (0); + /* Get a length of decoded bytes. */ + l = UUDECODE(*b++); len--; + if (l > 45) + /* Normally, maximum length is 45(character 'M'). */ + return (0); + if (l > len - nl) + return (0); /* Line too short. */ + while (l) { + if (!uuchar[*b++]) + return (0); + --len; + --l; + } + if (len-nl == 1 && + (uuchar[*b] || /* Check sum. */ + (*b >= 'a' && *b <= 'z'))) {/* Padding data(MINIX). */ + ++b; + --len; + } + b += nl; + if (avail && uuchar[*b]) + return (firstline+30); + } else if (l == 13) { + /* "begin-base64 " */ + while (len-nl > 0) { + if (!base64[*b++]) + return (0); + --len; + } + b += nl; + + if (avail >= 5 && memcmp(b, "====\n", 5) == 0) + return (firstline+40); + if (avail >= 6 && memcmp(b, "====\r\n", 6) == 0) + return (firstline+40); + if (avail > 0 && base64[*b]) + return (firstline+30); + } + + return (0); +} + +static int +uudecode_bidder_init(struct archive_read_filter *self) +{ + struct uudecode *uudecode; + void *out_buff; + void *in_buff; + + self->code = ARCHIVE_FILTER_UU; + self->name = "uu"; + self->read = uudecode_filter_read; + self->skip = NULL; /* not supported */ + self->close = uudecode_filter_close; + + uudecode = (struct uudecode *)calloc(sizeof(*uudecode), 1); + out_buff = malloc(OUT_BUFF_SIZE); + in_buff = malloc(IN_BUFF_SIZE); + if (uudecode == NULL || out_buff == NULL || in_buff == NULL) { + archive_set_error(&self->archive->archive, ENOMEM, + "Can't allocate data for uudecode"); + free(uudecode); + free(out_buff); + free(in_buff); + return (ARCHIVE_FATAL); + } + + self->data = uudecode; + uudecode->in_buff = in_buff; + uudecode->in_cnt = 0; + uudecode->in_allocated = IN_BUFF_SIZE; + uudecode->out_buff = out_buff; + uudecode->state = ST_FIND_HEAD; + + return (ARCHIVE_OK); +} + +static int +ensure_in_buff_size(struct archive_read_filter *self, + struct uudecode *uudecode, size_t size) +{ + + if (size > uudecode->in_allocated) { + unsigned char *ptr; + size_t newsize; + + /* + * Calculate a new buffer size for in_buff. + * Increase its value until it has enough size we need. + */ + newsize = uudecode->in_allocated; + do { + if (newsize < IN_BUFF_SIZE*32) + newsize <<= 1; + else + newsize += IN_BUFF_SIZE; + } while (size > newsize); + /* Allocate the new buffer. */ + ptr = malloc(newsize); + if (ptr == NULL) { + free(ptr); + archive_set_error(&self->archive->archive, + ENOMEM, + "Can't allocate data for uudecode"); + return (ARCHIVE_FATAL); + } + /* Move the remaining data in in_buff into the new buffer. */ + if (uudecode->in_cnt) + memmove(ptr, uudecode->in_buff, uudecode->in_cnt); + /* Replace in_buff with the new buffer. */ + free(uudecode->in_buff); + uudecode->in_buff = ptr; + uudecode->in_allocated = newsize; + } + return (ARCHIVE_OK); +} + +static ssize_t +uudecode_filter_read(struct archive_read_filter *self, const void **buff) +{ + struct uudecode *uudecode; + const unsigned char *b, *d; + unsigned char *out; + ssize_t avail_in, ravail; + ssize_t used; + ssize_t total; + ssize_t len, llen, nl; + + uudecode = (struct uudecode *)self->data; + +read_more: + d = __archive_read_filter_ahead(self->upstream, 1, &avail_in); + if (d == NULL && avail_in < 0) + return (ARCHIVE_FATAL); + /* Quiet a code analyzer; make sure avail_in must be zero + * when d is NULL. */ + if (d == NULL) + avail_in = 0; + used = 0; + total = 0; + out = uudecode->out_buff; + ravail = avail_in; + if (uudecode->state == ST_IGNORE) { + used = avail_in; + goto finish; + } + if (uudecode->in_cnt) { + /* + * If there is remaining data which is saved by + * previous calling, use it first. + */ + if (ensure_in_buff_size(self, uudecode, + avail_in + uudecode->in_cnt) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + memcpy(uudecode->in_buff + uudecode->in_cnt, + d, avail_in); + d = uudecode->in_buff; + avail_in += uudecode->in_cnt; + uudecode->in_cnt = 0; + } + for (;used < avail_in; d += llen, used += llen) { + int64_t l, body; + + b = d; + len = get_line(b, avail_in - used, &nl); + if (len < 0) { + /* Non-ascii character is found. */ + if (uudecode->state == ST_FIND_HEAD && + (uudecode->total > 0 || total > 0)) { + uudecode->state = ST_IGNORE; + used = avail_in; + goto finish; + } + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Insufficient compressed data"); + return (ARCHIVE_FATAL); + } + llen = len; + if ((nl == 0) && (uudecode->state != ST_UUEND)) { + if (total == 0 && ravail <= 0) { + /* There is nothing more to read, fail */ + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Missing format data"); + return (ARCHIVE_FATAL); + } + /* + * Save remaining data which does not contain + * NL('\n','\r'). + */ + if (ensure_in_buff_size(self, uudecode, len) + != ARCHIVE_OK) + return (ARCHIVE_FATAL); + if (uudecode->in_buff != b) + memmove(uudecode->in_buff, b, len); + uudecode->in_cnt = (int)len; + if (total == 0) { + /* Do not return 0; it means end-of-file. + * We should try to read bytes more. */ + __archive_read_filter_consume( + self->upstream, ravail); + goto read_more; + } + used += len; + break; + } + switch (uudecode->state) { + default: + case ST_FIND_HEAD: + /* Do not read more than UUENCODE_BID_MAX_READ bytes */ + if (total + len >= UUENCODE_BID_MAX_READ) { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid format data"); + return (ARCHIVE_FATAL); + } + if (len - nl >= 11 && memcmp(b, "begin ", 6) == 0) + l = 6; + else if (len - nl >= 18 && + memcmp(b, "begin-base64 ", 13) == 0) + l = 13; + else + l = 0; + if (l != 0 && b[l] >= '0' && b[l] <= '7' && + b[l+1] >= '0' && b[l+1] <= '7' && + b[l+2] >= '0' && b[l+2] <= '7' && b[l+3] == ' ') { + if (l == 6) + uudecode->state = ST_READ_UU; + else + uudecode->state = ST_READ_BASE64; + } + break; + case ST_READ_UU: + if (total + len * 2 > OUT_BUFF_SIZE) + goto finish; + body = len - nl; + if (!uuchar[*b] || body <= 0) { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Insufficient compressed data"); + return (ARCHIVE_FATAL); + } + /* Get length of undecoded bytes of current line. */ + l = UUDECODE(*b++); + body--; + if (l > body) { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Insufficient compressed data"); + return (ARCHIVE_FATAL); + } + if (l == 0) { + uudecode->state = ST_UUEND; + break; + } + while (l > 0) { + int n = 0; + + if (!uuchar[b[0]] || !uuchar[b[1]]) + break; + n = UUDECODE(*b++) << 18; + n |= UUDECODE(*b++) << 12; + *out++ = n >> 16; total++; + --l; + + if (l > 0) { + if (!uuchar[b[0]]) + break; + n |= UUDECODE(*b++) << 6; + *out++ = (n >> 8) & 0xFF; total++; + --l; + } + if (l > 0) { + if (!uuchar[b[0]]) + break; + n |= UUDECODE(*b++); + *out++ = n & 0xFF; total++; + --l; + } + } + if (l) { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Insufficient compressed data"); + return (ARCHIVE_FATAL); + } + break; + case ST_UUEND: + if (len - nl == 3 && memcmp(b, "end ", 3) == 0) + uudecode->state = ST_FIND_HEAD; + else { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Insufficient compressed data"); + return (ARCHIVE_FATAL); + } + break; + case ST_READ_BASE64: + if (total + len * 2 > OUT_BUFF_SIZE) + goto finish; + l = len - nl; + if (l >= 3 && b[0] == '=' && b[1] == '=' && + b[2] == '=') { + uudecode->state = ST_FIND_HEAD; + break; + } + while (l > 0) { + int n = 0; + + if (!base64[b[0]] || !base64[b[1]]) + break; + n = base64num[*b++] << 18; + n |= base64num[*b++] << 12; + *out++ = n >> 16; total++; + l -= 2; + + if (l > 0) { + if (*b == '=') + break; + if (!base64[*b]) + break; + n |= base64num[*b++] << 6; + *out++ = (n >> 8) & 0xFF; total++; + --l; + } + if (l > 0) { + if (*b == '=') + break; + if (!base64[*b]) + break; + n |= base64num[*b++]; + *out++ = n & 0xFF; total++; + --l; + } + } + if (l && *b != '=') { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Insufficient compressed data"); + return (ARCHIVE_FATAL); + } + break; + } + } +finish: + if (ravail < avail_in) + used -= avail_in - ravail; + __archive_read_filter_consume(self->upstream, used); + + *buff = uudecode->out_buff; + uudecode->total += total; + return (total); +} + +static int +uudecode_filter_close(struct archive_read_filter *self) +{ + struct uudecode *uudecode; + + uudecode = (struct uudecode *)self->data; + free(uudecode->in_buff); + free(uudecode->out_buff); + free(uudecode); + + return (ARCHIVE_OK); +} + diff --git a/c/archive_read_support_filter_xz.c b/c/archive_read_support_filter_xz.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_filter_xz.c @@ -0,0 +1,796 @@ +/*- + * Copyright (c) 2009-2011 Michihiro NAKAJIMA + * Copyright (c) 2003-2008 Tim Kientzle and Miklos Vajna + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif +#if HAVE_LZMA_H +#include +#endif + +#include "archive.h" +#include "archive_endian.h" +#include "archive_private.h" +#include "archive_read_private.h" + +#if HAVE_LZMA_H && HAVE_LIBLZMA + +struct private_data { + lzma_stream stream; + unsigned char *out_block; + size_t out_block_size; + int64_t total_out; + char eof; /* True = found end of compressed data. */ + char in_stream; + + /* Following variables are used for lzip only. */ + char lzip_ver; + uint32_t crc32; + int64_t member_in; + int64_t member_out; +}; + +#if LZMA_VERSION_MAJOR >= 5 +/* Effectively disable the limiter. */ +#define LZMA_MEMLIMIT UINT64_MAX +#else +/* NOTE: This needs to check memory size which running system has. */ +#define LZMA_MEMLIMIT (1U << 30) +#endif + +/* Combined lzip/lzma/xz filter */ +static ssize_t xz_filter_read(struct archive_read_filter *, const void **); +static int xz_filter_close(struct archive_read_filter *); +static int xz_lzma_bidder_init(struct archive_read_filter *); + +#endif + +/* + * Note that we can detect xz and lzma compressed files even if we + * can't decompress them. (In fact, we like detecting them because we + * can give better error messages.) So the bid framework here gets + * compiled even if no lzma library is available. + */ +static int xz_bidder_bid(struct archive_read_filter_bidder *, + struct archive_read_filter *); +static int xz_bidder_init(struct archive_read_filter *); +static int lzma_bidder_bid(struct archive_read_filter_bidder *, + struct archive_read_filter *); +static int lzma_bidder_init(struct archive_read_filter *); +static int lzip_has_member(struct archive_read_filter *); +static int lzip_bidder_bid(struct archive_read_filter_bidder *, + struct archive_read_filter *); +static int lzip_bidder_init(struct archive_read_filter *); + +#if ARCHIVE_VERSION_NUMBER < 4000000 +/* Deprecated; remove in libarchive 4.0 */ +int +archive_read_support_compression_xz(struct archive *a) +{ + return archive_read_support_filter_xz(a); +} +#endif + +int +archive_read_support_filter_xz(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_filter_bidder *bidder; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_filter_xz"); + + if (__archive_read_get_bidder(a, &bidder) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + bidder->data = NULL; + bidder->name = "xz"; + bidder->bid = xz_bidder_bid; + bidder->init = xz_bidder_init; + bidder->options = NULL; + bidder->free = NULL; +#if HAVE_LZMA_H && HAVE_LIBLZMA + return (ARCHIVE_OK); +#else + archive_set_error(_a, ARCHIVE_ERRNO_MISC, + "Using external xz program for xz decompression"); + return (ARCHIVE_WARN); +#endif +} + +#if ARCHIVE_VERSION_NUMBER < 4000000 +int +archive_read_support_compression_lzma(struct archive *a) +{ + return archive_read_support_filter_lzma(a); +} +#endif + +int +archive_read_support_filter_lzma(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_filter_bidder *bidder; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_filter_lzma"); + + if (__archive_read_get_bidder(a, &bidder) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + bidder->data = NULL; + bidder->name = "lzma"; + bidder->bid = lzma_bidder_bid; + bidder->init = lzma_bidder_init; + bidder->options = NULL; + bidder->free = NULL; +#if HAVE_LZMA_H && HAVE_LIBLZMA + return (ARCHIVE_OK); +#else + archive_set_error(_a, ARCHIVE_ERRNO_MISC, + "Using external lzma program for lzma decompression"); + return (ARCHIVE_WARN); +#endif +} + + +#if ARCHIVE_VERSION_NUMBER < 4000000 +int +archive_read_support_compression_lzip(struct archive *a) +{ + return archive_read_support_filter_lzip(a); +} +#endif + +int +archive_read_support_filter_lzip(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_filter_bidder *bidder; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_filter_lzip"); + + if (__archive_read_get_bidder(a, &bidder) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + bidder->data = NULL; + bidder->name = "lzip"; + bidder->bid = lzip_bidder_bid; + bidder->init = lzip_bidder_init; + bidder->options = NULL; + bidder->free = NULL; +#if HAVE_LZMA_H && HAVE_LIBLZMA + return (ARCHIVE_OK); +#else + archive_set_error(_a, ARCHIVE_ERRNO_MISC, + "Using external lzip program for lzip decompression"); + return (ARCHIVE_WARN); +#endif +} + +/* + * Test whether we can handle this data. + */ +static int +xz_bidder_bid(struct archive_read_filter_bidder *self, + struct archive_read_filter *filter) +{ + const unsigned char *buffer; + ssize_t avail; + + (void)self; /* UNUSED */ + + buffer = __archive_read_filter_ahead(filter, 6, &avail); + if (buffer == NULL) + return (0); + + /* + * Verify Header Magic Bytes : FD 37 7A 58 5A 00 + */ + if (memcmp(buffer, "\xFD\x37\x7A\x58\x5A\x00", 6) != 0) + return (0); + + return (48); +} + +/* + * Test whether we can handle this data. + * + * LZMA has a rather poor file signature. Zeros do not + * make good signature bytes as a rule, and the only non-zero byte + * here is an ASCII character. For example, an uncompressed tar + * archive whose first file is ']' would satisfy this check. It may + * be necessary to exclude LZMA from compression_all() because of + * this. Clients of libarchive would then have to explicitly enable + * LZMA checking instead of (or in addition to) compression_all() when + * they have other evidence (file name, command-line option) to go on. + */ +static int +lzma_bidder_bid(struct archive_read_filter_bidder *self, + struct archive_read_filter *filter) +{ + const unsigned char *buffer; + ssize_t avail; + uint32_t dicsize; + uint64_t uncompressed_size; + int bits_checked; + + (void)self; /* UNUSED */ + + buffer = __archive_read_filter_ahead(filter, 14, &avail); + if (buffer == NULL) + return (0); + + /* First byte of raw LZMA stream is commonly 0x5d. + * The first byte is a special number, which consists of + * three parameters of LZMA compression, a number of literal + * context bits(which is from 0 to 8, default is 3), a number + * of literal pos bits(which is from 0 to 4, default is 0), + * a number of pos bits(which is from 0 to 4, default is 2). + * The first byte is made by + * (pos bits * 5 + literal pos bit) * 9 + * literal contest bit, + * and so the default value in this field is + * (2 * 5 + 0) * 9 + 3 = 0x5d. + * lzma of LZMA SDK has options to change those parameters. + * It means a range of this field is from 0 to 224. And lzma of + * XZ Utils with option -e records 0x5e in this field. */ + /* NOTE: If this checking of the first byte increases false + * recognition, we should allow only 0x5d and 0x5e for the first + * byte of LZMA stream. */ + bits_checked = 0; + if (buffer[0] > (4 * 5 + 4) * 9 + 8) + return (0); + /* Most likely value in the first byte of LZMA stream. */ + if (buffer[0] == 0x5d || buffer[0] == 0x5e) + bits_checked += 8; + + /* Sixth through fourteenth bytes are uncompressed size, + * stored in little-endian order. `-1' means uncompressed + * size is unknown and lzma of XZ Utils always records `-1' + * in this field. */ + uncompressed_size = archive_le64dec(buffer+5); + if (uncompressed_size == (uint64_t)ARCHIVE_LITERAL_LL(-1)) + bits_checked += 64; + + /* Second through fifth bytes are dictionary size, stored in + * little-endian order. The minimum dictionary size is + * 1 << 12(4KiB) which the lzma of LZMA SDK uses with option + * -d12 and the maximum dictionary size is 1 << 27(128MiB) + * which the one uses with option -d27. + * NOTE: A comment of LZMA SDK source code says this dictionary + * range is from 1 << 12 to 1 << 30. */ + dicsize = archive_le32dec(buffer+1); + switch (dicsize) { + case 0x00001000:/* lzma of LZMA SDK option -d12. */ + case 0x00002000:/* lzma of LZMA SDK option -d13. */ + case 0x00004000:/* lzma of LZMA SDK option -d14. */ + case 0x00008000:/* lzma of LZMA SDK option -d15. */ + case 0x00010000:/* lzma of XZ Utils option -0 and -1. + * lzma of LZMA SDK option -d16. */ + case 0x00020000:/* lzma of LZMA SDK option -d17. */ + case 0x00040000:/* lzma of LZMA SDK option -d18. */ + case 0x00080000:/* lzma of XZ Utils option -2. + * lzma of LZMA SDK option -d19. */ + case 0x00100000:/* lzma of XZ Utils option -3. + * lzma of LZMA SDK option -d20. */ + case 0x00200000:/* lzma of XZ Utils option -4. + * lzma of LZMA SDK option -d21. */ + case 0x00400000:/* lzma of XZ Utils option -5. + * lzma of LZMA SDK option -d22. */ + case 0x00800000:/* lzma of XZ Utils option -6. + * lzma of LZMA SDK option -d23. */ + case 0x01000000:/* lzma of XZ Utils option -7. + * lzma of LZMA SDK option -d24. */ + case 0x02000000:/* lzma of XZ Utils option -8. + * lzma of LZMA SDK option -d25. */ + case 0x04000000:/* lzma of XZ Utils option -9. + * lzma of LZMA SDK option -d26. */ + case 0x08000000:/* lzma of LZMA SDK option -d27. */ + bits_checked += 32; + break; + default: + /* If a memory usage for encoding was not enough on + * the platform where LZMA stream was made, lzma of + * XZ Utils automatically decreased the dictionary + * size to enough memory for encoding by 1Mi bytes + * (1 << 20).*/ + if (dicsize <= 0x03F00000 && dicsize >= 0x00300000 && + (dicsize & ((1 << 20)-1)) == 0 && + bits_checked == 8 + 64) { + bits_checked += 32; + break; + } + /* Otherwise dictionary size is unlikely. But it is + * possible that someone makes lzma stream with + * liblzma/LZMA SDK in one's dictionary size. */ + return (0); + } + + /* TODO: The above test is still very weak. It would be + * good to do better. */ + + return (bits_checked); +} + +static int +lzip_has_member(struct archive_read_filter *filter) +{ + const unsigned char *buffer; + ssize_t avail; + int bits_checked; + int log2dic; + + buffer = __archive_read_filter_ahead(filter, 6, &avail); + if (buffer == NULL) + return (0); + + /* + * Verify Header Magic Bytes : 4C 5A 49 50 (`LZIP') + */ + bits_checked = 0; + if (memcmp(buffer, "LZIP", 4) != 0) + return (0); + bits_checked += 32; + + /* A version number must be 0 or 1 */ + if (buffer[4] != 0 && buffer[4] != 1) + return (0); + bits_checked += 8; + + /* Dictionary size. */ + log2dic = buffer[5] & 0x1f; + if (log2dic < 12 || log2dic > 27) + return (0); + bits_checked += 8; + + return (bits_checked); +} + +static int +lzip_bidder_bid(struct archive_read_filter_bidder *self, + struct archive_read_filter *filter) +{ + + (void)self; /* UNUSED */ + return (lzip_has_member(filter)); +} + +#if HAVE_LZMA_H && HAVE_LIBLZMA + +/* + * liblzma 4.999.7 and later support both lzma and xz streams. + */ +static int +xz_bidder_init(struct archive_read_filter *self) +{ + self->code = ARCHIVE_FILTER_XZ; + self->name = "xz"; + return (xz_lzma_bidder_init(self)); +} + +static int +lzma_bidder_init(struct archive_read_filter *self) +{ + self->code = ARCHIVE_FILTER_LZMA; + self->name = "lzma"; + return (xz_lzma_bidder_init(self)); +} + +static int +lzip_bidder_init(struct archive_read_filter *self) +{ + self->code = ARCHIVE_FILTER_LZIP; + self->name = "lzip"; + return (xz_lzma_bidder_init(self)); +} + +/* + * Set an error code and choose an error message + */ +static void +set_error(struct archive_read_filter *self, int ret) +{ + + switch (ret) { + case LZMA_STREAM_END: /* Found end of stream. */ + case LZMA_OK: /* Decompressor made some progress. */ + break; + case LZMA_MEM_ERROR: + archive_set_error(&self->archive->archive, ENOMEM, + "Lzma library error: Cannot allocate memory"); + break; + case LZMA_MEMLIMIT_ERROR: + archive_set_error(&self->archive->archive, ENOMEM, + "Lzma library error: Out of memory"); + break; + case LZMA_FORMAT_ERROR: + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Lzma library error: format not recognized"); + break; + case LZMA_OPTIONS_ERROR: + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Lzma library error: Invalid options"); + break; + case LZMA_DATA_ERROR: + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Lzma library error: Corrupted input data"); + break; + case LZMA_BUF_ERROR: + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Lzma library error: No progress is possible"); + break; + default: + /* Return an error. */ + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Lzma decompression failed: Unknown error"); + break; + } +} + +/* + * Setup the callbacks. + */ +static int +xz_lzma_bidder_init(struct archive_read_filter *self) +{ + static const size_t out_block_size = 64 * 1024; + void *out_block; + struct private_data *state; + int ret; + + state = (struct private_data *)calloc(sizeof(*state), 1); + out_block = (unsigned char *)malloc(out_block_size); + if (state == NULL || out_block == NULL) { + archive_set_error(&self->archive->archive, ENOMEM, + "Can't allocate data for xz decompression"); + free(out_block); + free(state); + return (ARCHIVE_FATAL); + } + + self->data = state; + state->out_block_size = out_block_size; + state->out_block = out_block; + self->read = xz_filter_read; + self->skip = NULL; /* not supported */ + self->close = xz_filter_close; + + state->stream.avail_in = 0; + + state->stream.next_out = state->out_block; + state->stream.avail_out = state->out_block_size; + + state->crc32 = 0; + if (self->code == ARCHIVE_FILTER_LZIP) { + /* + * We have to read a lzip header and use it to initialize + * compression library, thus we cannot initialize the + * library for lzip here. + */ + state->in_stream = 0; + return (ARCHIVE_OK); + } else + state->in_stream = 1; + + /* Initialize compression library. */ + if (self->code == ARCHIVE_FILTER_XZ) + ret = lzma_stream_decoder(&(state->stream), + LZMA_MEMLIMIT,/* memlimit */ + LZMA_CONCATENATED); + else + ret = lzma_alone_decoder(&(state->stream), + LZMA_MEMLIMIT);/* memlimit */ + + if (ret == LZMA_OK) + return (ARCHIVE_OK); + + /* Library setup failed: Choose an error message and clean up. */ + set_error(self, ret); + + free(state->out_block); + free(state); + self->data = NULL; + return (ARCHIVE_FATAL); +} + +static int +lzip_init(struct archive_read_filter *self) +{ + struct private_data *state; + const unsigned char *h; + lzma_filter filters[2]; + unsigned char props[5]; + ssize_t avail_in; + uint32_t dicsize; + int log2dic, ret; + + state = (struct private_data *)self->data; + h = __archive_read_filter_ahead(self->upstream, 6, &avail_in); + if (h == NULL) + return (ARCHIVE_FATAL); + + /* Get a version number. */ + state->lzip_ver = h[4]; + + /* + * Setup lzma property. + */ + props[0] = 0x5d; + + /* Get dictionary size. */ + log2dic = h[5] & 0x1f; + if (log2dic < 12 || log2dic > 27) + return (ARCHIVE_FATAL); + dicsize = 1U << log2dic; + if (log2dic > 12) + dicsize -= (dicsize / 16) * (h[5] >> 5); + archive_le32enc(props+1, dicsize); + + /* Consume lzip header. */ + __archive_read_filter_consume(self->upstream, 6); + state->member_in = 6; + + filters[0].id = LZMA_FILTER_LZMA1; + filters[0].options = NULL; + filters[1].id = LZMA_VLI_UNKNOWN; + filters[1].options = NULL; + + ret = lzma_properties_decode(&filters[0], NULL, props, sizeof(props)); + if (ret != LZMA_OK) { + set_error(self, ret); + return (ARCHIVE_FATAL); + } + ret = lzma_raw_decoder(&(state->stream), filters); + free(filters[0].options); + if (ret != LZMA_OK) { + set_error(self, ret); + return (ARCHIVE_FATAL); + } + return (ARCHIVE_OK); +} + +static int +lzip_tail(struct archive_read_filter *self) +{ + struct private_data *state; + const unsigned char *f; + ssize_t avail_in; + int tail; + + state = (struct private_data *)self->data; + if (state->lzip_ver == 0) + tail = 12; + else + tail = 20; + f = __archive_read_filter_ahead(self->upstream, tail, &avail_in); + if (f == NULL && avail_in < 0) + return (ARCHIVE_FATAL); + if (f == NULL || avail_in < tail) { + archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, + "Lzip: Remaining data is less bytes"); + return (ARCHIVE_FAILED); + } + + /* Check the crc32 value of the uncompressed data of the current + * member */ + if (state->crc32 != archive_le32dec(f)) { + archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, + "Lzip: CRC32 error"); + return (ARCHIVE_FAILED); + } + + /* Check the uncompressed size of the current member */ + if ((uint64_t)state->member_out != archive_le64dec(f + 4)) { + archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, + "Lzip: Uncompressed size error"); + return (ARCHIVE_FAILED); + } + + /* Check the total size of the current member */ + if (state->lzip_ver == 1 && + (uint64_t)state->member_in + tail != archive_le64dec(f + 12)) { + archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, + "Lzip: Member size error"); + return (ARCHIVE_FAILED); + } + __archive_read_filter_consume(self->upstream, tail); + + /* If current lzip data consists of multi member, try decompressing + * a next member. */ + if (lzip_has_member(self->upstream) != 0) { + state->in_stream = 0; + state->crc32 = 0; + state->member_out = 0; + state->member_in = 0; + state->eof = 0; + } + return (ARCHIVE_OK); +} + +/* + * Return the next block of decompressed data. + */ +static ssize_t +xz_filter_read(struct archive_read_filter *self, const void **p) +{ + struct private_data *state; + size_t decompressed; + ssize_t avail_in; + int ret; + + state = (struct private_data *)self->data; + + /* Empty our output buffer. */ + state->stream.next_out = state->out_block; + state->stream.avail_out = state->out_block_size; + + /* Try to fill the output buffer. */ + while (state->stream.avail_out > 0 && !state->eof) { + if (!state->in_stream) { + /* + * Initialize liblzma for lzip + */ + ret = lzip_init(self); + if (ret != ARCHIVE_OK) + return (ret); + state->in_stream = 1; + } + state->stream.next_in = + __archive_read_filter_ahead(self->upstream, 1, &avail_in); + if (state->stream.next_in == NULL && avail_in < 0) { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "truncated input"); + return (ARCHIVE_FATAL); + } + state->stream.avail_in = avail_in; + + /* Decompress as much as we can in one pass. */ + ret = lzma_code(&(state->stream), + (state->stream.avail_in == 0)? LZMA_FINISH: LZMA_RUN); + switch (ret) { + case LZMA_STREAM_END: /* Found end of stream. */ + state->eof = 1; + /* FALL THROUGH */ + case LZMA_OK: /* Decompressor made some progress. */ + __archive_read_filter_consume(self->upstream, + avail_in - state->stream.avail_in); + state->member_in += + avail_in - state->stream.avail_in; + break; + default: + set_error(self, ret); + return (ARCHIVE_FATAL); + } + } + + decompressed = state->stream.next_out - state->out_block; + state->total_out += decompressed; + state->member_out += decompressed; + if (decompressed == 0) + *p = NULL; + else { + *p = state->out_block; + if (self->code == ARCHIVE_FILTER_LZIP) { + state->crc32 = lzma_crc32(state->out_block, + decompressed, state->crc32); + if (state->eof) { + ret = lzip_tail(self); + if (ret != ARCHIVE_OK) + return (ret); + } + } + } + return (decompressed); +} + +/* + * Clean up the decompressor. + */ +static int +xz_filter_close(struct archive_read_filter *self) +{ + struct private_data *state; + + state = (struct private_data *)self->data; + lzma_end(&(state->stream)); + free(state->out_block); + free(state); + return (ARCHIVE_OK); +} + +#else + +/* + * + * If we have no suitable library on this system, we can't actually do + * the decompression. We can, however, still detect compressed + * archives and emit a useful message. + * + */ +static int +lzma_bidder_init(struct archive_read_filter *self) +{ + int r; + + r = __archive_read_program(self, "lzma -d -qq"); + /* Note: We set the format here even if __archive_read_program() + * above fails. We do, after all, know what the format is + * even if we weren't able to read it. */ + self->code = ARCHIVE_FILTER_LZMA; + self->name = "lzma"; + return (r); +} + +static int +xz_bidder_init(struct archive_read_filter *self) +{ + int r; + + r = __archive_read_program(self, "xz -d -qq"); + /* Note: We set the format here even if __archive_read_program() + * above fails. We do, after all, know what the format is + * even if we weren't able to read it. */ + self->code = ARCHIVE_FILTER_XZ; + self->name = "xz"; + return (r); +} + +static int +lzip_bidder_init(struct archive_read_filter *self) +{ + int r; + + r = __archive_read_program(self, "lzip -d -q"); + /* Note: We set the format here even if __archive_read_program() + * above fails. We do, after all, know what the format is + * even if we weren't able to read it. */ + self->code = ARCHIVE_FILTER_LZIP; + self->name = "lzip"; + return (r); +} + +#endif /* HAVE_LZMA_H */ diff --git a/c/archive_read_support_filter_zstd.c b/c/archive_read_support_filter_zstd.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_filter_zstd.c @@ -0,0 +1,296 @@ +/*- + * Copyright (c) 2009-2011 Sean Purcell + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif + +#ifdef HAVE_ERRNO_H +#include +#endif +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif +#if HAVE_ZSTD_H +#include +#endif + +#include "archive.h" +#include "archive_endian.h" +#include "archive_private.h" +#include "archive_read_private.h" + +#if HAVE_ZSTD_H && HAVE_LIBZSTD + +struct private_data { + ZSTD_DStream *dstream; + unsigned char *out_block; + size_t out_block_size; + int64_t total_out; + char in_frame; /* True = in the middle of a zstd frame. */ + char eof; /* True = found end of compressed data. */ +}; + +/* Zstd Filter. */ +static ssize_t zstd_filter_read(struct archive_read_filter *, const void**); +static int zstd_filter_close(struct archive_read_filter *); +#endif + +/* + * Note that we can detect zstd compressed files even if we can't decompress + * them. (In fact, we like detecting them because we can give better error + * messages.) So the bid framework here gets compiled even if no zstd library + * is available. + */ +static int zstd_bidder_bid(struct archive_read_filter_bidder *, + struct archive_read_filter *); +static int zstd_bidder_init(struct archive_read_filter *); + +int +archive_read_support_filter_zstd(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct archive_read_filter_bidder *bidder; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_filter_zstd"); + + if (__archive_read_get_bidder(a, &bidder) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + bidder->data = NULL; + bidder->name = "zstd"; + bidder->bid = zstd_bidder_bid; + bidder->init = zstd_bidder_init; + bidder->options = NULL; + bidder->free = NULL; +#if HAVE_ZSTD_H && HAVE_LIBZSTD + return (ARCHIVE_OK); +#else + archive_set_error(_a, ARCHIVE_ERRNO_MISC, + "Using external zstd program for zstd decompression"); + return (ARCHIVE_WARN); +#endif +} + +/* + * Test whether we can handle this data. + */ +static int +zstd_bidder_bid(struct archive_read_filter_bidder *self, + struct archive_read_filter *filter) +{ + const unsigned char *buffer; + ssize_t avail; + unsigned prefix; + + /* Zstd frame magic values */ + const unsigned zstd_magic = 0xFD2FB528U; + const unsigned zstd_magic_skippable_start = 0x184D2A50U; + const unsigned zstd_magic_skippable_mask = 0xFFFFFFF0; + + (void) self; /* UNUSED */ + + buffer = __archive_read_filter_ahead(filter, 4, &avail); + if (buffer == NULL) + return (0); + + prefix = archive_le32dec(buffer); + if (prefix == zstd_magic) + return (32); + if ((prefix & zstd_magic_skippable_mask) == zstd_magic_skippable_start) + return (32); + + return (0); +} + +#if !(HAVE_ZSTD_H && HAVE_LIBZSTD) + +/* + * If we don't have the library on this system, we can't do the + * decompression directly. We can, however, try to run "zstd -d" + * in case that's available. + */ +static int +zstd_bidder_init(struct archive_read_filter *self) +{ + int r; + + r = __archive_read_program(self, "zstd -d -qq"); + /* Note: We set the format here even if __archive_read_program() + * above fails. We do, after all, know what the format is + * even if we weren't able to read it. */ + self->code = ARCHIVE_FILTER_ZSTD; + self->name = "zstd"; + return (r); +} + +#else + +/* + * Initialize the filter object + */ +static int +zstd_bidder_init(struct archive_read_filter *self) +{ + struct private_data *state; + const size_t out_block_size = ZSTD_DStreamOutSize(); + void *out_block; + ZSTD_DStream *dstream; + + self->code = ARCHIVE_FILTER_ZSTD; + self->name = "zstd"; + + state = (struct private_data *)calloc(sizeof(*state), 1); + out_block = (unsigned char *)malloc(out_block_size); + dstream = ZSTD_createDStream(); + + if (state == NULL || out_block == NULL || dstream == NULL) { + free(out_block); + free(state); + ZSTD_freeDStream(dstream); /* supports free on NULL */ + archive_set_error(&self->archive->archive, ENOMEM, + "Can't allocate data for zstd decompression"); + return (ARCHIVE_FATAL); + } + + self->data = state; + + state->out_block_size = out_block_size; + state->out_block = out_block; + state->dstream = dstream; + self->read = zstd_filter_read; + self->skip = NULL; /* not supported */ + self->close = zstd_filter_close; + + state->eof = 0; + state->in_frame = 0; + + return (ARCHIVE_OK); +} + +static ssize_t +zstd_filter_read(struct archive_read_filter *self, const void **p) +{ + struct private_data *state; + size_t decompressed; + ssize_t avail_in; + ZSTD_outBuffer out; + ZSTD_inBuffer in; + + state = (struct private_data *)self->data; + + out = (ZSTD_outBuffer) { state->out_block, state->out_block_size, 0 }; + + /* Try to fill the output buffer. */ + while (out.pos < out.size && !state->eof) { + if (!state->in_frame) { + const size_t ret = ZSTD_initDStream(state->dstream); + if (ZSTD_isError(ret)) { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Error initializing zstd decompressor: %s", + ZSTD_getErrorName(ret)); + return (ARCHIVE_FATAL); + } + } + in.src = __archive_read_filter_ahead(self->upstream, 1, + &avail_in); + if (avail_in < 0) { + return avail_in; + } + if (in.src == NULL && avail_in == 0) { + if (!state->in_frame) { + /* end of stream */ + state->eof = 1; + break; + } else { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Truncated zstd input"); + return (ARCHIVE_FATAL); + } + } + in.size = avail_in; + in.pos = 0; + + { + const size_t ret = + ZSTD_decompressStream(state->dstream, &out, &in); + + if (ZSTD_isError(ret)) { + archive_set_error(&self->archive->archive, + ARCHIVE_ERRNO_MISC, + "Zstd decompression failed: %s", + ZSTD_getErrorName(ret)); + return (ARCHIVE_FATAL); + } + + /* Decompressor made some progress */ + __archive_read_filter_consume(self->upstream, in.pos); + + /* ret guaranteed to be > 0 if frame isn't done yet */ + state->in_frame = (ret != 0); + } + } + + decompressed = out.pos; + state->total_out += decompressed; + if (decompressed == 0) + *p = NULL; + else + *p = state->out_block; + return (decompressed); +} + +/* + * Clean up the decompressor. + */ +static int +zstd_filter_close(struct archive_read_filter *self) +{ + struct private_data *state; + + state = (struct private_data *)self->data; + + ZSTD_freeDStream(state->dstream); + free(state->out_block); + free(state); + + return (ARCHIVE_OK); +} + +#endif /* HAVE_ZLIB_H && HAVE_LIBZSTD */ diff --git a/c/archive_read_support_format_7zip.c b/c/archive_read_support_format_7zip.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_format_7zip.c @@ -0,0 +1,3873 @@ +/*- + * Copyright (c) 2011 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_BZLIB_H +#include +#endif +#ifdef HAVE_LZMA_H +#include +#endif +#ifdef HAVE_ZLIB_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_ppmd7_private.h" +#include "archive_private.h" +#include "archive_read_private.h" +#include "archive_endian.h" + +#ifndef HAVE_ZLIB_H +#include "archive_crc32.h" +#endif + +#define _7ZIP_SIGNATURE "7z\xBC\xAF\x27\x1C" +#define SFX_MIN_ADDR 0x27000 +#define SFX_MAX_ADDR 0x60000 + + +/* + * Codec ID + */ +#define _7Z_COPY 0 +#define _7Z_LZMA 0x030101 +#define _7Z_LZMA2 0x21 +#define _7Z_DEFLATE 0x040108 +#define _7Z_BZ2 0x040202 +#define _7Z_PPMD 0x030401 +#define _7Z_DELTA 0x03 +#define _7Z_CRYPTO_MAIN_ZIP 0x06F10101 /* Main Zip crypto algo */ +#define _7Z_CRYPTO_RAR_29 0x06F10303 /* Rar29 AES-128 + (modified SHA-1) */ +#define _7Z_CRYPTO_AES_256_SHA_256 0x06F10701 /* AES-256 + SHA-256 */ + + +#define _7Z_X86 0x03030103 +#define _7Z_X86_BCJ2 0x0303011B +#define _7Z_POWERPC 0x03030205 +#define _7Z_IA64 0x03030401 +#define _7Z_ARM 0x03030501 +#define _7Z_ARMTHUMB 0x03030701 +#define _7Z_SPARC 0x03030805 + +/* + * 7-Zip header property IDs. + */ +#define kEnd 0x00 +#define kHeader 0x01 +#define kArchiveProperties 0x02 +#define kAdditionalStreamsInfo 0x03 +#define kMainStreamsInfo 0x04 +#define kFilesInfo 0x05 +#define kPackInfo 0x06 +#define kUnPackInfo 0x07 +#define kSubStreamsInfo 0x08 +#define kSize 0x09 +#define kCRC 0x0A +#define kFolder 0x0B +#define kCodersUnPackSize 0x0C +#define kNumUnPackStream 0x0D +#define kEmptyStream 0x0E +#define kEmptyFile 0x0F +#define kAnti 0x10 +#define kName 0x11 +#define kCTime 0x12 +#define kATime 0x13 +#define kMTime 0x14 +#define kAttributes 0x15 +#define kEncodedHeader 0x17 +#define kDummy 0x19 + +struct _7z_digests { + unsigned char *defineds; + uint32_t *digests; +}; + + +struct _7z_folder { + uint64_t numCoders; + struct _7z_coder { + unsigned long codec; + uint64_t numInStreams; + uint64_t numOutStreams; + uint64_t propertiesSize; + unsigned char *properties; + } *coders; + uint64_t numBindPairs; + struct { + uint64_t inIndex; + uint64_t outIndex; + } *bindPairs; + uint64_t numPackedStreams; + uint64_t *packedStreams; + uint64_t numInStreams; + uint64_t numOutStreams; + uint64_t *unPackSize; + unsigned char digest_defined; + uint32_t digest; + uint64_t numUnpackStreams; + uint32_t packIndex; + /* Unoperated bytes. */ + uint64_t skipped_bytes; +}; + +struct _7z_coders_info { + uint64_t numFolders; + struct _7z_folder *folders; + uint64_t dataStreamIndex; +}; + +struct _7z_pack_info { + uint64_t pos; + uint64_t numPackStreams; + uint64_t *sizes; + struct _7z_digests digest; + /* Calculated from pos and numPackStreams. */ + uint64_t *positions; +}; + +struct _7z_substream_info { + size_t unpack_streams; + uint64_t *unpackSizes; + unsigned char *digestsDefined; + uint32_t *digests; +}; + +struct _7z_stream_info { + struct _7z_pack_info pi; + struct _7z_coders_info ci; + struct _7z_substream_info ss; +}; + +struct _7z_header_info { + uint64_t dataIndex; + + unsigned char *emptyStreamBools; + unsigned char *emptyFileBools; + unsigned char *antiBools; + unsigned char *attrBools; +}; + +struct _7zip_entry { + size_t name_len; + unsigned char *utf16name; +#if defined(_WIN32) && !defined(__CYGWIN__) && defined(_DEBUG) + const wchar_t *wname; +#endif + uint32_t folderIndex; + uint32_t ssIndex; + unsigned flg; +#define MTIME_IS_SET (1<<0) +#define ATIME_IS_SET (1<<1) +#define CTIME_IS_SET (1<<2) +#define CRC32_IS_SET (1<<3) +#define HAS_STREAM (1<<4) + + time_t mtime; + time_t atime; + time_t ctime; + long mtime_ns; + long atime_ns; + long ctime_ns; + uint32_t mode; + uint32_t attr; +}; + +struct _7zip { + /* Structural information about the archive. */ + struct _7z_stream_info si; + + int header_is_being_read; + int header_is_encoded; + uint64_t header_bytes_remaining; + unsigned long header_crc32; + /* Header offset to check that reading points of the file contents + * will not exceed the header. */ + uint64_t header_offset; + /* Base offset of the archive file for a seek in case reading SFX. */ + uint64_t seek_base; + + /* List of entries */ + size_t entries_remaining; + uint64_t numFiles; + struct _7zip_entry *entries; + struct _7zip_entry *entry; + unsigned char *entry_names; + + /* entry_bytes_remaining is the number of bytes we expect. */ + int64_t entry_offset; + uint64_t entry_bytes_remaining; + + /* Running CRC32 of the decompressed data */ + unsigned long entry_crc32; + + /* Flags to mark progress of decompression. */ + char end_of_entry; + + /* Uncompressed buffer control. */ +#define UBUFF_SIZE (64 * 1024) + unsigned char *uncompressed_buffer; + unsigned char *uncompressed_buffer_pointer; + size_t uncompressed_buffer_size; + size_t uncompressed_buffer_bytes_remaining; + + /* Offset of the compressed data. */ + int64_t stream_offset; + + /* + * Decompressing control data. + */ + unsigned folder_index; + uint64_t folder_outbytes_remaining; + unsigned pack_stream_index; + unsigned pack_stream_remaining; + uint64_t pack_stream_inbytes_remaining; + size_t pack_stream_bytes_unconsumed; + + /* The codec information of a folder. */ + unsigned long codec; + unsigned long codec2; + + /* + * Decompressor controllers. + */ + /* Decoding LZMA1 and LZMA2 data. */ +#ifdef HAVE_LZMA_H + lzma_stream lzstream; + int lzstream_valid; +#endif + /* Decoding bzip2 data. */ +#if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) + bz_stream bzstream; + int bzstream_valid; +#endif + /* Decoding deflate data. */ +#ifdef HAVE_ZLIB_H + z_stream stream; + int stream_valid; +#endif + /* Decoding PPMd data. */ + int ppmd7_stat; + CPpmd7 ppmd7_context; + CPpmd7z_RangeDec range_dec; + IByteIn bytein; + struct { + const unsigned char *next_in; + int64_t avail_in; + int64_t total_in; + unsigned char *next_out; + int64_t avail_out; + int64_t total_out; + int overconsumed; + } ppstream; + int ppmd7_valid; + + /* Decoding BCJ and BCJ2 data. */ + uint32_t bcj_state; + size_t odd_bcj_size; + unsigned char odd_bcj[4]; + /* Decoding BCJ data. */ + size_t bcj_prevPosT; + uint32_t bcj_prevMask; + uint32_t bcj_ip; + + /* Decoding BCJ2 data. */ + size_t main_stream_bytes_remaining; + unsigned char *sub_stream_buff[3]; + size_t sub_stream_size[3]; + size_t sub_stream_bytes_remaining[3]; + unsigned char *tmp_stream_buff; + size_t tmp_stream_buff_size; + size_t tmp_stream_bytes_avail; + size_t tmp_stream_bytes_remaining; +#ifdef _LZMA_PROB32 +#define CProb uint32_t +#else +#define CProb uint16_t +#endif + CProb bcj2_p[256 + 2]; + uint8_t bcj2_prevByte; + uint32_t bcj2_range; + uint32_t bcj2_code; + uint64_t bcj2_outPos; + + /* Filename character-set conversion data. */ + struct archive_string_conv *sconv; + + char format_name[64]; + + /* Custom value that is non-zero if this archive contains encrypted entries. */ + int has_encrypted_entries; +}; + +/* Maximum entry size. This limitation prevents reading intentional + * corrupted 7-zip files on assuming there are not so many entries in + * the files. */ +#define UMAX_ENTRY ARCHIVE_LITERAL_ULL(100000000) + +static int archive_read_format_7zip_has_encrypted_entries(struct archive_read *); +static int archive_read_support_format_7zip_capabilities(struct archive_read *a); +static int archive_read_format_7zip_bid(struct archive_read *, int); +static int archive_read_format_7zip_cleanup(struct archive_read *); +static int archive_read_format_7zip_read_data(struct archive_read *, + const void **, size_t *, int64_t *); +static int archive_read_format_7zip_read_data_skip(struct archive_read *); +static int archive_read_format_7zip_read_header(struct archive_read *, + struct archive_entry *); +static int check_7zip_header_in_sfx(const char *); +static unsigned long decode_codec_id(const unsigned char *, size_t); +static int decode_encoded_header_info(struct archive_read *, + struct _7z_stream_info *); +static int decompress(struct archive_read *, struct _7zip *, + void *, size_t *, const void *, size_t *); +static ssize_t extract_pack_stream(struct archive_read *, size_t); +static void fileTimeToUtc(uint64_t, time_t *, long *); +static uint64_t folder_uncompressed_size(struct _7z_folder *); +static void free_CodersInfo(struct _7z_coders_info *); +static void free_Digest(struct _7z_digests *); +static void free_Folder(struct _7z_folder *); +static void free_Header(struct _7z_header_info *); +static void free_PackInfo(struct _7z_pack_info *); +static void free_StreamsInfo(struct _7z_stream_info *); +static void free_SubStreamsInfo(struct _7z_substream_info *); +static int free_decompression(struct archive_read *, struct _7zip *); +static ssize_t get_uncompressed_data(struct archive_read *, const void **, + size_t, size_t); +static const unsigned char * header_bytes(struct archive_read *, size_t); +static int init_decompression(struct archive_read *, struct _7zip *, + const struct _7z_coder *, const struct _7z_coder *); +static int parse_7zip_uint64(struct archive_read *, uint64_t *); +static int read_Bools(struct archive_read *, unsigned char *, size_t); +static int read_CodersInfo(struct archive_read *, + struct _7z_coders_info *); +static int read_Digests(struct archive_read *, struct _7z_digests *, + size_t); +static int read_Folder(struct archive_read *, struct _7z_folder *); +static int read_Header(struct archive_read *, struct _7z_header_info *, + int); +static int read_PackInfo(struct archive_read *, struct _7z_pack_info *); +static int read_StreamsInfo(struct archive_read *, + struct _7z_stream_info *); +static int read_SubStreamsInfo(struct archive_read *, + struct _7z_substream_info *, struct _7z_folder *, size_t); +static int read_Times(struct archive_read *, struct _7z_header_info *, + int); +static void read_consume(struct archive_read *); +static ssize_t read_stream(struct archive_read *, const void **, size_t, + size_t); +static int seek_pack(struct archive_read *); +static int64_t skip_stream(struct archive_read *, size_t); +static int skip_sfx(struct archive_read *, ssize_t); +static int slurp_central_directory(struct archive_read *, struct _7zip *, + struct _7z_header_info *); +static int setup_decode_folder(struct archive_read *, struct _7z_folder *, + int); +static void x86_Init(struct _7zip *); +static size_t x86_Convert(struct _7zip *, uint8_t *, size_t); +static ssize_t Bcj2_Decode(struct _7zip *, uint8_t *, size_t); + + +int +archive_read_support_format_7zip(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct _7zip *zip; + int r; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_format_7zip"); + + zip = calloc(1, sizeof(*zip)); + if (zip == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate 7zip data"); + return (ARCHIVE_FATAL); + } + + /* + * Until enough data has been read, we cannot tell about + * any encrypted entries yet. + */ + zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; + + + r = __archive_read_register_format(a, + zip, + "7zip", + archive_read_format_7zip_bid, + NULL, + archive_read_format_7zip_read_header, + archive_read_format_7zip_read_data, + archive_read_format_7zip_read_data_skip, + NULL, + archive_read_format_7zip_cleanup, + archive_read_support_format_7zip_capabilities, + archive_read_format_7zip_has_encrypted_entries); + + if (r != ARCHIVE_OK) + free(zip); + return (ARCHIVE_OK); +} + +static int +archive_read_support_format_7zip_capabilities(struct archive_read * a) +{ + (void)a; /* UNUSED */ + return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | + ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); +} + + +static int +archive_read_format_7zip_has_encrypted_entries(struct archive_read *_a) +{ + if (_a && _a->format) { + struct _7zip * zip = (struct _7zip *)_a->format->data; + if (zip) { + return zip->has_encrypted_entries; + } + } + return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; +} + +static int +archive_read_format_7zip_bid(struct archive_read *a, int best_bid) +{ + const char *p; + + /* If someone has already bid more than 32, then avoid + trashing the look-ahead buffers with a seek. */ + if (best_bid > 32) + return (-1); + + if ((p = __archive_read_ahead(a, 6, NULL)) == NULL) + return (0); + + /* If first six bytes are the 7-Zip signature, + * return the bid right now. */ + if (memcmp(p, _7ZIP_SIGNATURE, 6) == 0) + return (48); + + /* + * It may a 7-Zip SFX archive file. If first two bytes are + * 'M' and 'Z' available on Windows or first four bytes are + * "\x7F\x45LF" available on posix like system, seek the 7-Zip + * signature. Although we will perform a seek when reading + * a header, what we do not use __archive_read_seek() here is + * due to a bidding performance. + */ + if ((p[0] == 'M' && p[1] == 'Z') || memcmp(p, "\x7F\x45LF", 4) == 0) { + ssize_t offset = SFX_MIN_ADDR; + ssize_t window = 4096; + ssize_t bytes_avail; + while (offset + window <= (SFX_MAX_ADDR)) { + const char *buff = __archive_read_ahead(a, + offset + window, &bytes_avail); + if (buff == NULL) { + /* Remaining bytes are less than window. */ + window >>= 1; + if (window < 0x40) + return (0); + continue; + } + p = buff + offset; + while (p + 32 < buff + bytes_avail) { + int step = check_7zip_header_in_sfx(p); + if (step == 0) + return (48); + p += step; + } + offset = p - buff; + } + } + return (0); +} + +static int +check_7zip_header_in_sfx(const char *p) +{ + switch ((unsigned char)p[5]) { + case 0x1C: + if (memcmp(p, _7ZIP_SIGNATURE, 6) != 0) + return (6); + /* + * Test the CRC because its extraction code has 7-Zip + * Magic Code, so we should do this in order not to + * make a mis-detection. + */ + if (crc32(0, (const unsigned char *)p + 12, 20) + != archive_le32dec(p + 8)) + return (6); + /* Hit the header! */ + return (0); + case 0x37: return (5); + case 0x7A: return (4); + case 0xBC: return (3); + case 0xAF: return (2); + case 0x27: return (1); + default: return (6); + } +} + +static int +skip_sfx(struct archive_read *a, ssize_t bytes_avail) +{ + const void *h; + const char *p, *q; + size_t skip, offset; + ssize_t bytes, window; + + /* + * If bytes_avail > SFX_MIN_ADDR we do not have to call + * __archive_read_seek() at this time since we have + * already had enough data. + */ + if (bytes_avail > SFX_MIN_ADDR) + __archive_read_consume(a, SFX_MIN_ADDR); + else if (__archive_read_seek(a, SFX_MIN_ADDR, SEEK_SET) < 0) + return (ARCHIVE_FATAL); + + offset = 0; + window = 1; + while (offset + window <= SFX_MAX_ADDR - SFX_MIN_ADDR) { + h = __archive_read_ahead(a, window, &bytes); + if (h == NULL) { + /* Remaining bytes are less than window. */ + window >>= 1; + if (window < 0x40) + goto fatal; + continue; + } + if (bytes < 6) { + /* This case might happen when window == 1. */ + window = 4096; + continue; + } + p = (const char *)h; + q = p + bytes; + + /* + * Scan ahead until we find something that looks + * like the 7-Zip header. + */ + while (p + 32 < q) { + int step = check_7zip_header_in_sfx(p); + if (step == 0) { + struct _7zip *zip = + (struct _7zip *)a->format->data; + skip = p - (const char *)h; + __archive_read_consume(a, skip); + zip->seek_base = SFX_MIN_ADDR + offset + skip; + return (ARCHIVE_OK); + } + p += step; + } + skip = p - (const char *)h; + __archive_read_consume(a, skip); + offset += skip; + if (window == 1) + window = 4096; + } +fatal: + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Couldn't find out 7-Zip header"); + return (ARCHIVE_FATAL); +} + +static int +archive_read_format_7zip_read_header(struct archive_read *a, + struct archive_entry *entry) +{ + struct _7zip *zip = (struct _7zip *)a->format->data; + struct _7zip_entry *zip_entry; + int r, ret = ARCHIVE_OK; + struct _7z_folder *folder = 0; + uint64_t fidx = 0; + + /* + * It should be sufficient to call archive_read_next_header() for + * a reader to determine if an entry is encrypted or not. If the + * encryption of an entry is only detectable when calling + * archive_read_data(), so be it. We'll do the same check there + * as well. + */ + if (zip->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { + zip->has_encrypted_entries = 0; + } + + a->archive.archive_format = ARCHIVE_FORMAT_7ZIP; + if (a->archive.archive_format_name == NULL) + a->archive.archive_format_name = "7-Zip"; + + if (zip->entries == NULL) { + struct _7z_header_info header; + + memset(&header, 0, sizeof(header)); + r = slurp_central_directory(a, zip, &header); + free_Header(&header); + if (r != ARCHIVE_OK) + return (r); + zip->entries_remaining = (size_t)zip->numFiles; + zip->entry = zip->entries; + } else { + ++zip->entry; + } + zip_entry = zip->entry; + + if (zip->entries_remaining <= 0 || zip_entry == NULL) + return ARCHIVE_EOF; + --zip->entries_remaining; + + zip->entry_offset = 0; + zip->end_of_entry = 0; + zip->entry_crc32 = crc32(0, NULL, 0); + + /* Setup a string conversion for a filename. */ + if (zip->sconv == NULL) { + zip->sconv = archive_string_conversion_from_charset( + &a->archive, "UTF-16LE", 1); + if (zip->sconv == NULL) + return (ARCHIVE_FATAL); + } + + /* Figure out if the entry is encrypted by looking at the folder + that is associated to the current 7zip entry. If the folder + has a coder with a _7Z_CRYPTO codec then the folder is encrypted. + Hence the entry must also be encrypted. */ + if (zip_entry && zip_entry->folderIndex < zip->si.ci.numFolders) { + folder = &(zip->si.ci.folders[zip_entry->folderIndex]); + for (fidx=0; folder && fidxnumCoders; fidx++) { + switch(folder->coders[fidx].codec) { + case _7Z_CRYPTO_MAIN_ZIP: + case _7Z_CRYPTO_RAR_29: + case _7Z_CRYPTO_AES_256_SHA_256: { + archive_entry_set_is_data_encrypted(entry, 1); + zip->has_encrypted_entries = 1; + break; + } + } + } + } + + /* Now that we've checked for encryption, if there were still no + * encrypted entries found we can say for sure that there are none. + */ + if (zip->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { + zip->has_encrypted_entries = 0; + } + + if (archive_entry_copy_pathname_l(entry, + (const char *)zip_entry->utf16name, + zip_entry->name_len, zip->sconv) != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Pathname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Pathname cannot be converted " + "from %s to current locale.", + archive_string_conversion_charset_name(zip->sconv)); + ret = ARCHIVE_WARN; + } + + /* Populate some additional entry fields: */ + archive_entry_set_mode(entry, zip_entry->mode); + if (zip_entry->flg & MTIME_IS_SET) + archive_entry_set_mtime(entry, zip_entry->mtime, + zip_entry->mtime_ns); + if (zip_entry->flg & CTIME_IS_SET) + archive_entry_set_ctime(entry, zip_entry->ctime, + zip_entry->ctime_ns); + if (zip_entry->flg & ATIME_IS_SET) + archive_entry_set_atime(entry, zip_entry->atime, + zip_entry->atime_ns); + if (zip_entry->ssIndex != (uint32_t)-1) { + zip->entry_bytes_remaining = + zip->si.ss.unpackSizes[zip_entry->ssIndex]; + archive_entry_set_size(entry, zip->entry_bytes_remaining); + } else { + zip->entry_bytes_remaining = 0; + archive_entry_set_size(entry, 0); + } + + /* If there's no body, force read_data() to return EOF immediately. */ + if (zip->entry_bytes_remaining < 1) + zip->end_of_entry = 1; + + if ((zip_entry->mode & AE_IFMT) == AE_IFLNK) { + unsigned char *symname = NULL; + size_t symsize = 0; + + /* + * Symbolic-name is recorded as its contents. We have to + * read the contents at this time. + */ + while (zip->entry_bytes_remaining > 0) { + const void *buff; + unsigned char *mem; + size_t size; + int64_t offset; + + r = archive_read_format_7zip_read_data(a, &buff, + &size, &offset); + if (r < ARCHIVE_WARN) { + free(symname); + return (r); + } + mem = realloc(symname, symsize + size + 1); + if (mem == NULL) { + free(symname); + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Symname"); + return (ARCHIVE_FATAL); + } + symname = mem; + memcpy(symname+symsize, buff, size); + symsize += size; + } + if (symsize == 0) { + /* If there is no symname, handle it as a regular + * file. */ + zip_entry->mode &= ~AE_IFMT; + zip_entry->mode |= AE_IFREG; + archive_entry_set_mode(entry, zip_entry->mode); + } else { + symname[symsize] = '\0'; + archive_entry_copy_symlink(entry, + (const char *)symname); + } + free(symname); + archive_entry_set_size(entry, 0); + } + + /* Set up a more descriptive format name. */ + sprintf(zip->format_name, "7-Zip"); + a->archive.archive_format_name = zip->format_name; + + return (ret); +} + +static int +archive_read_format_7zip_read_data(struct archive_read *a, + const void **buff, size_t *size, int64_t *offset) +{ + struct _7zip *zip; + ssize_t bytes; + int ret = ARCHIVE_OK; + + zip = (struct _7zip *)(a->format->data); + + if (zip->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { + zip->has_encrypted_entries = 0; + } + + if (zip->pack_stream_bytes_unconsumed) + read_consume(a); + + *offset = zip->entry_offset; + *size = 0; + *buff = NULL; + /* + * If we hit end-of-entry last time, clean up and return + * ARCHIVE_EOF this time. + */ + if (zip->end_of_entry) + return (ARCHIVE_EOF); + + bytes = read_stream(a, buff, + (size_t)zip->entry_bytes_remaining, 0); + if (bytes < 0) + return ((int)bytes); + if (bytes == 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated 7-Zip file body"); + return (ARCHIVE_FATAL); + } + zip->entry_bytes_remaining -= bytes; + if (zip->entry_bytes_remaining == 0) + zip->end_of_entry = 1; + + /* Update checksum */ + if ((zip->entry->flg & CRC32_IS_SET) && bytes) + zip->entry_crc32 = crc32(zip->entry_crc32, *buff, + (unsigned)bytes); + + /* If we hit the end, swallow any end-of-data marker. */ + if (zip->end_of_entry) { + /* Check computed CRC against file contents. */ + if ((zip->entry->flg & CRC32_IS_SET) && + zip->si.ss.digests[zip->entry->ssIndex] != + zip->entry_crc32) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "7-Zip bad CRC: 0x%lx should be 0x%lx", + (unsigned long)zip->entry_crc32, + (unsigned long)zip->si.ss.digests[ + zip->entry->ssIndex]); + ret = ARCHIVE_WARN; + } + } + + *size = bytes; + *offset = zip->entry_offset; + zip->entry_offset += bytes; + + return (ret); +} + +static int +archive_read_format_7zip_read_data_skip(struct archive_read *a) +{ + struct _7zip *zip; + int64_t bytes_skipped; + + zip = (struct _7zip *)(a->format->data); + + if (zip->pack_stream_bytes_unconsumed) + read_consume(a); + + /* If we've already read to end of data, we're done. */ + if (zip->end_of_entry) + return (ARCHIVE_OK); + + /* + * If the length is at the beginning, we can skip the + * compressed data much more quickly. + */ + bytes_skipped = skip_stream(a, (size_t)zip->entry_bytes_remaining); + if (bytes_skipped < 0) + return (ARCHIVE_FATAL); + zip->entry_bytes_remaining = 0; + + /* This entry is finished and done. */ + zip->end_of_entry = 1; + return (ARCHIVE_OK); +} + +static int +archive_read_format_7zip_cleanup(struct archive_read *a) +{ + struct _7zip *zip; + + zip = (struct _7zip *)(a->format->data); + free_StreamsInfo(&(zip->si)); + free(zip->entries); + free(zip->entry_names); + free_decompression(a, zip); + free(zip->uncompressed_buffer); + free(zip->sub_stream_buff[0]); + free(zip->sub_stream_buff[1]); + free(zip->sub_stream_buff[2]); + free(zip->tmp_stream_buff); + free(zip); + (a->format->data) = NULL; + return (ARCHIVE_OK); +} + +static void +read_consume(struct archive_read *a) +{ + struct _7zip *zip = (struct _7zip *)a->format->data; + + if (zip->pack_stream_bytes_unconsumed) { + __archive_read_consume(a, zip->pack_stream_bytes_unconsumed); + zip->stream_offset += zip->pack_stream_bytes_unconsumed; + zip->pack_stream_bytes_unconsumed = 0; + } +} + +#ifdef HAVE_LZMA_H + +/* + * Set an error code and choose an error message for liblzma. + */ +static void +set_error(struct archive_read *a, int ret) +{ + + switch (ret) { + case LZMA_STREAM_END: /* Found end of stream. */ + case LZMA_OK: /* Decompressor made some progress. */ + break; + case LZMA_MEM_ERROR: + archive_set_error(&a->archive, ENOMEM, + "Lzma library error: Cannot allocate memory"); + break; + case LZMA_MEMLIMIT_ERROR: + archive_set_error(&a->archive, ENOMEM, + "Lzma library error: Out of memory"); + break; + case LZMA_FORMAT_ERROR: + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Lzma library error: format not recognized"); + break; + case LZMA_OPTIONS_ERROR: + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Lzma library error: Invalid options"); + break; + case LZMA_DATA_ERROR: + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Lzma library error: Corrupted input data"); + break; + case LZMA_BUF_ERROR: + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Lzma library error: No progress is possible"); + break; + default: + /* Return an error. */ + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Lzma decompression failed: Unknown error"); + break; + } +} + +#endif + +static unsigned long +decode_codec_id(const unsigned char *codecId, size_t id_size) +{ + unsigned i; + unsigned long id = 0; + + for (i = 0; i < id_size; i++) { + id <<= 8; + id += codecId[i]; + } + return (id); +} + +static Byte +ppmd_read(void *p) +{ + struct archive_read *a = ((IByteIn*)p)->a; + struct _7zip *zip = (struct _7zip *)(a->format->data); + Byte b; + + if (zip->ppstream.avail_in == 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated RAR file data"); + zip->ppstream.overconsumed = 1; + return (0); + } + b = *zip->ppstream.next_in++; + zip->ppstream.avail_in--; + zip->ppstream.total_in++; + return (b); +} + +static int +init_decompression(struct archive_read *a, struct _7zip *zip, + const struct _7z_coder *coder1, const struct _7z_coder *coder2) +{ + int r; + + zip->codec = coder1->codec; + zip->codec2 = -1; + + switch (zip->codec) { + case _7Z_COPY: + case _7Z_BZ2: + case _7Z_DEFLATE: + case _7Z_PPMD: + if (coder2 != NULL) { + if (coder2->codec != _7Z_X86 && + coder2->codec != _7Z_X86_BCJ2) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Unsupported filter %lx for %lx", + coder2->codec, coder1->codec); + return (ARCHIVE_FAILED); + } + zip->codec2 = coder2->codec; + zip->bcj_state = 0; + if (coder2->codec == _7Z_X86) + x86_Init(zip); + } + break; + default: + break; + } + + switch (zip->codec) { + case _7Z_COPY: + break; + + case _7Z_LZMA: case _7Z_LZMA2: +#ifdef HAVE_LZMA_H +#if LZMA_VERSION_MAJOR >= 5 +/* Effectively disable the limiter. */ +#define LZMA_MEMLIMIT UINT64_MAX +#else +/* NOTE: This needs to check memory size which running system has. */ +#define LZMA_MEMLIMIT (1U << 30) +#endif + { + lzma_options_delta delta_opt; + lzma_filter filters[LZMA_FILTERS_MAX], *ff; + int fi = 0; + + if (zip->lzstream_valid) { + lzma_end(&(zip->lzstream)); + zip->lzstream_valid = 0; + } + + /* + * NOTE: liblzma incompletely handle the BCJ+LZMA compressed + * data made by 7-Zip because 7-Zip does not add End-Of- + * Payload Marker(EOPM) at the end of LZMA compressed data, + * and so liblzma cannot know the end of the compressed data + * without EOPM. So consequently liblzma will not return last + * three or four bytes of uncompressed data because + * LZMA_FILTER_X86 filter does not handle input data if its + * data size is less than five bytes. If liblzma detect EOPM + * or know the uncompressed data size, liblzma will flush out + * the remaining that three or four bytes of uncompressed + * data. That is why we have to use our converting program + * for BCJ+LZMA. If we were able to tell the uncompressed + * size to liblzma when using lzma_raw_decoder() liblzma + * could correctly deal with BCJ+LZMA. But unfortunately + * there is no way to do that. + * Discussion about this can be found at XZ Utils forum. + */ + if (coder2 != NULL) { + zip->codec2 = coder2->codec; + + filters[fi].options = NULL; + switch (zip->codec2) { + case _7Z_X86: + if (zip->codec == _7Z_LZMA2) { + filters[fi].id = LZMA_FILTER_X86; + fi++; + } else + /* Use our filter. */ + x86_Init(zip); + break; + case _7Z_X86_BCJ2: + /* Use our filter. */ + zip->bcj_state = 0; + break; + case _7Z_DELTA: + if (coder2->propertiesSize != 1) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Invalid Delta parameter"); + return (ARCHIVE_FAILED); + } + filters[fi].id = LZMA_FILTER_DELTA; + memset(&delta_opt, 0, sizeof(delta_opt)); + delta_opt.type = LZMA_DELTA_TYPE_BYTE; + delta_opt.dist = + (uint32_t)coder2->properties[0] + 1; + filters[fi].options = &delta_opt; + fi++; + break; + /* Following filters have not been tested yet. */ + case _7Z_POWERPC: + filters[fi].id = LZMA_FILTER_POWERPC; + fi++; + break; + case _7Z_IA64: + filters[fi].id = LZMA_FILTER_IA64; + fi++; + break; + case _7Z_ARM: + filters[fi].id = LZMA_FILTER_ARM; + fi++; + break; + case _7Z_ARMTHUMB: + filters[fi].id = LZMA_FILTER_ARMTHUMB; + fi++; + break; + case _7Z_SPARC: + filters[fi].id = LZMA_FILTER_SPARC; + fi++; + break; + default: + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Unexpected codec ID: %lX", zip->codec2); + return (ARCHIVE_FAILED); + } + } + + if (zip->codec == _7Z_LZMA2) + filters[fi].id = LZMA_FILTER_LZMA2; + else + filters[fi].id = LZMA_FILTER_LZMA1; + filters[fi].options = NULL; + ff = &filters[fi]; + r = lzma_properties_decode(&filters[fi], NULL, + coder1->properties, (size_t)coder1->propertiesSize); + if (r != LZMA_OK) { + set_error(a, r); + return (ARCHIVE_FAILED); + } + fi++; + + filters[fi].id = LZMA_VLI_UNKNOWN; + filters[fi].options = NULL; + r = lzma_raw_decoder(&(zip->lzstream), filters); + free(ff->options); + if (r != LZMA_OK) { + set_error(a, r); + return (ARCHIVE_FAILED); + } + zip->lzstream_valid = 1; + zip->lzstream.total_in = 0; + zip->lzstream.total_out = 0; + break; + } +#else + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "LZMA codec is unsupported"); + return (ARCHIVE_FAILED); +#endif + case _7Z_BZ2: +#if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) + if (zip->bzstream_valid) { + BZ2_bzDecompressEnd(&(zip->bzstream)); + zip->bzstream_valid = 0; + } + r = BZ2_bzDecompressInit(&(zip->bzstream), 0, 0); + if (r == BZ_MEM_ERROR) + r = BZ2_bzDecompressInit(&(zip->bzstream), 0, 1); + if (r != BZ_OK) { + int err = ARCHIVE_ERRNO_MISC; + const char *detail = NULL; + switch (r) { + case BZ_PARAM_ERROR: + detail = "invalid setup parameter"; + break; + case BZ_MEM_ERROR: + err = ENOMEM; + detail = "out of memory"; + break; + case BZ_CONFIG_ERROR: + detail = "mis-compiled library"; + break; + } + archive_set_error(&a->archive, err, + "Internal error initializing decompressor: %s", + detail != NULL ? detail : "??"); + zip->bzstream_valid = 0; + return (ARCHIVE_FAILED); + } + zip->bzstream_valid = 1; + zip->bzstream.total_in_lo32 = 0; + zip->bzstream.total_in_hi32 = 0; + zip->bzstream.total_out_lo32 = 0; + zip->bzstream.total_out_hi32 = 0; + break; +#else + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "BZ2 codec is unsupported"); + return (ARCHIVE_FAILED); +#endif + case _7Z_DEFLATE: +#ifdef HAVE_ZLIB_H + if (zip->stream_valid) + r = inflateReset(&(zip->stream)); + else + r = inflateInit2(&(zip->stream), + -15 /* Don't check for zlib header */); + if (r != Z_OK) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Couldn't initialize zlib stream."); + return (ARCHIVE_FAILED); + } + zip->stream_valid = 1; + zip->stream.total_in = 0; + zip->stream.total_out = 0; + break; +#else + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "DEFLATE codec is unsupported"); + return (ARCHIVE_FAILED); +#endif + case _7Z_PPMD: + { + unsigned order; + uint32_t msize; + + if (zip->ppmd7_valid) { + __archive_ppmd7_functions.Ppmd7_Free( + &zip->ppmd7_context); + zip->ppmd7_valid = 0; + } + + if (coder1->propertiesSize < 5) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Malformed PPMd parameter"); + return (ARCHIVE_FAILED); + } + order = coder1->properties[0]; + msize = archive_le32dec(&(coder1->properties[1])); + if (order < PPMD7_MIN_ORDER || order > PPMD7_MAX_ORDER || + msize < PPMD7_MIN_MEM_SIZE || msize > PPMD7_MAX_MEM_SIZE) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Malformed PPMd parameter"); + return (ARCHIVE_FAILED); + } + __archive_ppmd7_functions.Ppmd7_Construct(&zip->ppmd7_context); + r = __archive_ppmd7_functions.Ppmd7_Alloc( + &zip->ppmd7_context, msize); + if (r == 0) { + archive_set_error(&a->archive, ENOMEM, + "Coludn't allocate memory for PPMd"); + return (ARCHIVE_FATAL); + } + __archive_ppmd7_functions.Ppmd7_Init( + &zip->ppmd7_context, order); + __archive_ppmd7_functions.Ppmd7z_RangeDec_CreateVTable( + &zip->range_dec); + zip->ppmd7_valid = 1; + zip->ppmd7_stat = 0; + zip->ppstream.overconsumed = 0; + zip->ppstream.total_in = 0; + zip->ppstream.total_out = 0; + break; + } + case _7Z_X86: + case _7Z_X86_BCJ2: + case _7Z_POWERPC: + case _7Z_IA64: + case _7Z_ARM: + case _7Z_ARMTHUMB: + case _7Z_SPARC: + case _7Z_DELTA: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Unexpected codec ID: %lX", zip->codec); + return (ARCHIVE_FAILED); + case _7Z_CRYPTO_MAIN_ZIP: + case _7Z_CRYPTO_RAR_29: + case _7Z_CRYPTO_AES_256_SHA_256: + if (a->entry) { + archive_entry_set_is_metadata_encrypted(a->entry, 1); + archive_entry_set_is_data_encrypted(a->entry, 1); + zip->has_encrypted_entries = 1; + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Crypto codec not supported yet (ID: 0x%lX)", zip->codec); + return (ARCHIVE_FAILED); + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Unknown codec ID: %lX", zip->codec); + return (ARCHIVE_FAILED); + } + + return (ARCHIVE_OK); +} + +static int +decompress(struct archive_read *a, struct _7zip *zip, + void *buff, size_t *outbytes, const void *b, size_t *used) +{ + const uint8_t *t_next_in; + uint8_t *t_next_out; + size_t o_avail_in, o_avail_out; + size_t t_avail_in, t_avail_out; + uint8_t *bcj2_next_out; + size_t bcj2_avail_out; + int r, ret = ARCHIVE_OK; + + t_avail_in = o_avail_in = *used; + t_avail_out = o_avail_out = *outbytes; + t_next_in = b; + t_next_out = buff; + + if (zip->codec != _7Z_LZMA2 && zip->codec2 == _7Z_X86) { + int i; + + /* Do not copy out the BCJ remaining bytes when the output + * buffer size is less than five bytes. */ + if (o_avail_in != 0 && t_avail_out < 5 && zip->odd_bcj_size) { + *used = 0; + *outbytes = 0; + return (ret); + } + for (i = 0; zip->odd_bcj_size > 0 && t_avail_out; i++) { + *t_next_out++ = zip->odd_bcj[i]; + t_avail_out--; + zip->odd_bcj_size--; + } + if (o_avail_in == 0 || t_avail_out == 0) { + *used = o_avail_in - t_avail_in; + *outbytes = o_avail_out - t_avail_out; + if (o_avail_in == 0) + ret = ARCHIVE_EOF; + return (ret); + } + } + + bcj2_next_out = t_next_out; + bcj2_avail_out = t_avail_out; + if (zip->codec2 == _7Z_X86_BCJ2) { + /* + * Decord a remaining decompressed main stream for BCJ2. + */ + if (zip->tmp_stream_bytes_remaining) { + ssize_t bytes; + size_t remaining = zip->tmp_stream_bytes_remaining; + bytes = Bcj2_Decode(zip, t_next_out, t_avail_out); + if (bytes < 0) { + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "BCJ2 conversion Failed"); + return (ARCHIVE_FAILED); + } + zip->main_stream_bytes_remaining -= + remaining - zip->tmp_stream_bytes_remaining; + t_avail_out -= bytes; + if (o_avail_in == 0 || t_avail_out == 0) { + *used = 0; + *outbytes = o_avail_out - t_avail_out; + if (o_avail_in == 0 && + zip->tmp_stream_bytes_remaining) + ret = ARCHIVE_EOF; + return (ret); + } + t_next_out += bytes; + bcj2_next_out = t_next_out; + bcj2_avail_out = t_avail_out; + } + t_next_out = zip->tmp_stream_buff; + t_avail_out = zip->tmp_stream_buff_size; + } + + switch (zip->codec) { + case _7Z_COPY: + { + size_t bytes = + (t_avail_in > t_avail_out)?t_avail_out:t_avail_in; + + memcpy(t_next_out, t_next_in, bytes); + t_avail_in -= bytes; + t_avail_out -= bytes; + if (o_avail_in == 0) + ret = ARCHIVE_EOF; + break; + } +#ifdef HAVE_LZMA_H + case _7Z_LZMA: case _7Z_LZMA2: + zip->lzstream.next_in = t_next_in; + zip->lzstream.avail_in = t_avail_in; + zip->lzstream.next_out = t_next_out; + zip->lzstream.avail_out = t_avail_out; + + r = lzma_code(&(zip->lzstream), LZMA_RUN); + switch (r) { + case LZMA_STREAM_END: /* Found end of stream. */ + lzma_end(&(zip->lzstream)); + zip->lzstream_valid = 0; + ret = ARCHIVE_EOF; + break; + case LZMA_OK: /* Decompressor made some progress. */ + break; + default: + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "Decompression failed(%d)", + r); + return (ARCHIVE_FAILED); + } + t_avail_in = zip->lzstream.avail_in; + t_avail_out = zip->lzstream.avail_out; + break; +#endif +#if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) + case _7Z_BZ2: + zip->bzstream.next_in = (char *)(uintptr_t)t_next_in; + zip->bzstream.avail_in = t_avail_in; + zip->bzstream.next_out = (char *)(uintptr_t)t_next_out; + zip->bzstream.avail_out = t_avail_out; + r = BZ2_bzDecompress(&(zip->bzstream)); + switch (r) { + case BZ_STREAM_END: /* Found end of stream. */ + switch (BZ2_bzDecompressEnd(&(zip->bzstream))) { + case BZ_OK: + break; + default: + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "Failed to clean up decompressor"); + return (ARCHIVE_FAILED); + } + zip->bzstream_valid = 0; + ret = ARCHIVE_EOF; + break; + case BZ_OK: /* Decompressor made some progress. */ + break; + default: + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "bzip decompression failed"); + return (ARCHIVE_FAILED); + } + t_avail_in = zip->bzstream.avail_in; + t_avail_out = zip->bzstream.avail_out; + break; +#endif +#ifdef HAVE_ZLIB_H + case _7Z_DEFLATE: + zip->stream.next_in = (Bytef *)(uintptr_t)t_next_in; + zip->stream.avail_in = (uInt)t_avail_in; + zip->stream.next_out = t_next_out; + zip->stream.avail_out = (uInt)t_avail_out; + r = inflate(&(zip->stream), 0); + switch (r) { + case Z_STREAM_END: /* Found end of stream. */ + ret = ARCHIVE_EOF; + break; + case Z_OK: /* Decompressor made some progress.*/ + break; + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "File decompression failed (%d)", r); + return (ARCHIVE_FAILED); + } + t_avail_in = zip->stream.avail_in; + t_avail_out = zip->stream.avail_out; + break; +#endif + case _7Z_PPMD: + { + uint64_t flush_bytes; + + if (!zip->ppmd7_valid || zip->ppmd7_stat < 0 || + t_avail_out <= 0) { + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "Decompression internal error"); + return (ARCHIVE_FAILED); + } + zip->ppstream.next_in = t_next_in; + zip->ppstream.avail_in = t_avail_in; + zip->ppstream.next_out = t_next_out; + zip->ppstream.avail_out = t_avail_out; + if (zip->ppmd7_stat == 0) { + zip->bytein.a = a; + zip->bytein.Read = &ppmd_read; + zip->range_dec.Stream = &zip->bytein; + r = __archive_ppmd7_functions.Ppmd7z_RangeDec_Init( + &(zip->range_dec)); + if (r == 0) { + zip->ppmd7_stat = -1; + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Failed to initialize PPMd range decorder"); + return (ARCHIVE_FAILED); + } + if (zip->ppstream.overconsumed) { + zip->ppmd7_stat = -1; + return (ARCHIVE_FAILED); + } + zip->ppmd7_stat = 1; + } + + if (t_avail_in == 0) + /* XXX Flush out remaining decoded data XXX */ + flush_bytes = zip->folder_outbytes_remaining; + else + flush_bytes = 0; + + do { + int sym; + + sym = __archive_ppmd7_functions.Ppmd7_DecodeSymbol( + &(zip->ppmd7_context), &(zip->range_dec.p)); + if (sym < 0) { + zip->ppmd7_stat = -1; + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Failed to decode PPMd"); + return (ARCHIVE_FAILED); + } + if (zip->ppstream.overconsumed) { + zip->ppmd7_stat = -1; + return (ARCHIVE_FAILED); + } + *zip->ppstream.next_out++ = (unsigned char)sym; + zip->ppstream.avail_out--; + zip->ppstream.total_out++; + if (flush_bytes) + flush_bytes--; + } while (zip->ppstream.avail_out && + (zip->ppstream.avail_in || flush_bytes)); + + t_avail_in = (size_t)zip->ppstream.avail_in; + t_avail_out = (size_t)zip->ppstream.avail_out; + break; + } + default: + archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, + "Decompression internal error"); + return (ARCHIVE_FAILED); + } + if (ret != ARCHIVE_OK && ret != ARCHIVE_EOF) + return (ret); + + *used = o_avail_in - t_avail_in; + *outbytes = o_avail_out - t_avail_out; + + /* + * Decord BCJ. + */ + if (zip->codec != _7Z_LZMA2 && zip->codec2 == _7Z_X86) { + size_t l = x86_Convert(zip, buff, *outbytes); + zip->odd_bcj_size = *outbytes - l; + if (zip->odd_bcj_size > 0 && zip->odd_bcj_size <= 4 && + o_avail_in && ret != ARCHIVE_EOF) { + memcpy(zip->odd_bcj, ((unsigned char *)buff) + l, + zip->odd_bcj_size); + *outbytes = l; + } else + zip->odd_bcj_size = 0; + } + + /* + * Decord BCJ2 with a decompressed main stream. + */ + if (zip->codec2 == _7Z_X86_BCJ2) { + ssize_t bytes; + + zip->tmp_stream_bytes_avail = + zip->tmp_stream_buff_size - t_avail_out; + if (zip->tmp_stream_bytes_avail > + zip->main_stream_bytes_remaining) + zip->tmp_stream_bytes_avail = + zip->main_stream_bytes_remaining; + zip->tmp_stream_bytes_remaining = zip->tmp_stream_bytes_avail; + bytes = Bcj2_Decode(zip, bcj2_next_out, bcj2_avail_out); + if (bytes < 0) { + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, "BCJ2 conversion Failed"); + return (ARCHIVE_FAILED); + } + zip->main_stream_bytes_remaining -= + zip->tmp_stream_bytes_avail + - zip->tmp_stream_bytes_remaining; + bcj2_avail_out -= bytes; + *outbytes = o_avail_out - bcj2_avail_out; + } + + return (ret); +} + +static int +free_decompression(struct archive_read *a, struct _7zip *zip) +{ + int r = ARCHIVE_OK; + +#if !defined(HAVE_ZLIB_H) &&\ + !(defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR)) + (void)a;/* UNUSED */ +#endif +#ifdef HAVE_LZMA_H + if (zip->lzstream_valid) + lzma_end(&(zip->lzstream)); +#endif +#if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) + if (zip->bzstream_valid) { + if (BZ2_bzDecompressEnd(&(zip->bzstream)) != BZ_OK) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Failed to clean up bzip2 decompressor"); + r = ARCHIVE_FATAL; + } + zip->bzstream_valid = 0; + } +#endif +#ifdef HAVE_ZLIB_H + if (zip->stream_valid) { + if (inflateEnd(&(zip->stream)) != Z_OK) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Failed to clean up zlib decompressor"); + r = ARCHIVE_FATAL; + } + zip->stream_valid = 0; + } +#endif + if (zip->ppmd7_valid) { + __archive_ppmd7_functions.Ppmd7_Free( + &zip->ppmd7_context); + zip->ppmd7_valid = 0; + } + return (r); +} + +static int +parse_7zip_uint64(struct archive_read *a, uint64_t *val) +{ + const unsigned char *p; + unsigned char avail, mask; + int i; + + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + avail = *p; + mask = 0x80; + *val = 0; + for (i = 0; i < 8; i++) { + if (avail & mask) { + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + *val |= ((uint64_t)*p) << (8 * i); + mask >>= 1; + continue; + } + *val += ((uint64_t)(avail & (mask -1))) << (8 * i); + break; + } + return (0); +} + +static int +read_Bools(struct archive_read *a, unsigned char *data, size_t num) +{ + const unsigned char *p; + unsigned i, mask = 0, avail = 0; + + for (i = 0; i < num; i++) { + if (mask == 0) { + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + avail = *p; + mask = 0x80; + } + data[i] = (avail & mask)?1:0; + mask >>= 1; + } + return (0); +} + +static void +free_Digest(struct _7z_digests *d) +{ + free(d->defineds); + free(d->digests); +} + +static int +read_Digests(struct archive_read *a, struct _7z_digests *d, size_t num) +{ + const unsigned char *p; + unsigned i; + + if (num == 0) + return (-1); + memset(d, 0, sizeof(*d)); + + d->defineds = malloc(num); + if (d->defineds == NULL) + return (-1); + /* + * Read Bools. + */ + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + if (*p == 0) { + if (read_Bools(a, d->defineds, num) < 0) + return (-1); + } else + /* All are defined */ + memset(d->defineds, 1, num); + + d->digests = calloc(num, sizeof(*d->digests)); + if (d->digests == NULL) + return (-1); + for (i = 0; i < num; i++) { + if (d->defineds[i]) { + if ((p = header_bytes(a, 4)) == NULL) + return (-1); + d->digests[i] = archive_le32dec(p); + } + } + + return (0); +} + +static void +free_PackInfo(struct _7z_pack_info *pi) +{ + free(pi->sizes); + free(pi->positions); + free_Digest(&(pi->digest)); +} + +static int +read_PackInfo(struct archive_read *a, struct _7z_pack_info *pi) +{ + const unsigned char *p; + unsigned i; + + memset(pi, 0, sizeof(*pi)); + + /* + * Read PackPos. + */ + if (parse_7zip_uint64(a, &(pi->pos)) < 0) + return (-1); + + /* + * Read NumPackStreams. + */ + if (parse_7zip_uint64(a, &(pi->numPackStreams)) < 0) + return (-1); + if (pi->numPackStreams == 0) + return (-1); + if (UMAX_ENTRY < pi->numPackStreams) + return (-1); + + /* + * Read PackSizes[num] + */ + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + if (*p == kEnd) + /* PackSizes[num] are not present. */ + return (0); + if (*p != kSize) + return (-1); + pi->sizes = calloc((size_t)pi->numPackStreams, sizeof(uint64_t)); + pi->positions = calloc((size_t)pi->numPackStreams, sizeof(uint64_t)); + if (pi->sizes == NULL || pi->positions == NULL) + return (-1); + + for (i = 0; i < pi->numPackStreams; i++) { + if (parse_7zip_uint64(a, &(pi->sizes[i])) < 0) + return (-1); + } + + /* + * Read PackStreamDigests[num] + */ + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + if (*p == kEnd) { + /* PackStreamDigests[num] are not present. */ + pi->digest.defineds = + calloc((size_t)pi->numPackStreams, sizeof(*pi->digest.defineds)); + pi->digest.digests = + calloc((size_t)pi->numPackStreams, sizeof(*pi->digest.digests)); + if (pi->digest.defineds == NULL || pi->digest.digests == NULL) + return (-1); + return (0); + } + + if (*p != kCRC) + return (-1); + + if (read_Digests(a, &(pi->digest), (size_t)pi->numPackStreams) < 0) + return (-1); + + /* + * Must be marked by kEnd. + */ + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + if (*p != kEnd) + return (-1); + return (0); +} + +static void +free_Folder(struct _7z_folder *f) +{ + unsigned i; + + if (f->coders) { + for (i = 0; i< f->numCoders; i++) { + free(f->coders[i].properties); + } + free(f->coders); + } + free(f->bindPairs); + free(f->packedStreams); + free(f->unPackSize); +} + +static int +read_Folder(struct archive_read *a, struct _7z_folder *f) +{ + struct _7zip *zip = (struct _7zip *)a->format->data; + const unsigned char *p; + uint64_t numInStreamsTotal = 0; + uint64_t numOutStreamsTotal = 0; + unsigned i; + + memset(f, 0, sizeof(*f)); + + /* + * Read NumCoders. + */ + if (parse_7zip_uint64(a, &(f->numCoders)) < 0) + return (-1); + if (f->numCoders > 4) + /* Too many coders. */ + return (-1); + + f->coders = calloc((size_t)f->numCoders, sizeof(*f->coders)); + if (f->coders == NULL) + return (-1); + for (i = 0; i< f->numCoders; i++) { + size_t codec_size; + int simple, attr; + + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + /* + * 0:3 CodecIdSize + * 4: 0 - IsSimple + * 1 - Is not Simple + * 5: 0 - No Attributes + * 1 - There are Attributes; + * 7: Must be zero. + */ + codec_size = *p & 0xf; + simple = (*p & 0x10)?0:1; + attr = *p & 0x20; + if (*p & 0x80) + return (-1);/* Not supported. */ + + /* + * Read Decompression Method IDs. + */ + if ((p = header_bytes(a, codec_size)) == NULL) + return (-1); + + f->coders[i].codec = decode_codec_id(p, codec_size); + + if (simple) { + f->coders[i].numInStreams = 1; + f->coders[i].numOutStreams = 1; + } else { + if (parse_7zip_uint64( + a, &(f->coders[i].numInStreams)) < 0) + return (-1); + if (UMAX_ENTRY < f->coders[i].numInStreams) + return (-1); + if (parse_7zip_uint64( + a, &(f->coders[i].numOutStreams)) < 0) + return (-1); + if (UMAX_ENTRY < f->coders[i].numOutStreams) + return (-1); + } + + if (attr) { + if (parse_7zip_uint64( + a, &(f->coders[i].propertiesSize)) < 0) + return (-1); + if ((p = header_bytes( + a, (size_t)f->coders[i].propertiesSize)) == NULL) + return (-1); + f->coders[i].properties = + malloc((size_t)f->coders[i].propertiesSize); + if (f->coders[i].properties == NULL) + return (-1); + memcpy(f->coders[i].properties, p, + (size_t)f->coders[i].propertiesSize); + } + + numInStreamsTotal += f->coders[i].numInStreams; + numOutStreamsTotal += f->coders[i].numOutStreams; + } + + if (numOutStreamsTotal == 0 || + numInStreamsTotal < numOutStreamsTotal-1) + return (-1); + + f->numBindPairs = numOutStreamsTotal - 1; + if (zip->header_bytes_remaining < f->numBindPairs) + return (-1); + if (f->numBindPairs > 0) { + f->bindPairs = + calloc((size_t)f->numBindPairs, sizeof(*f->bindPairs)); + if (f->bindPairs == NULL) + return (-1); + } else + f->bindPairs = NULL; + for (i = 0; i < f->numBindPairs; i++) { + if (parse_7zip_uint64(a, &(f->bindPairs[i].inIndex)) < 0) + return (-1); + if (UMAX_ENTRY < f->bindPairs[i].inIndex) + return (-1); + if (parse_7zip_uint64(a, &(f->bindPairs[i].outIndex)) < 0) + return (-1); + if (UMAX_ENTRY < f->bindPairs[i].outIndex) + return (-1); + } + + f->numPackedStreams = numInStreamsTotal - f->numBindPairs; + f->packedStreams = + calloc((size_t)f->numPackedStreams, sizeof(*f->packedStreams)); + if (f->packedStreams == NULL) + return (-1); + if (f->numPackedStreams == 1) { + for (i = 0; i < numInStreamsTotal; i++) { + unsigned j; + for (j = 0; j < f->numBindPairs; j++) { + if (f->bindPairs[j].inIndex == i) + break; + } + if (j == f->numBindPairs) + break; + } + if (i == numInStreamsTotal) + return (-1); + f->packedStreams[0] = i; + } else { + for (i = 0; i < f->numPackedStreams; i++) { + if (parse_7zip_uint64(a, &(f->packedStreams[i])) < 0) + return (-1); + if (UMAX_ENTRY < f->packedStreams[i]) + return (-1); + } + } + f->numInStreams = numInStreamsTotal; + f->numOutStreams = numOutStreamsTotal; + + return (0); +} + +static void +free_CodersInfo(struct _7z_coders_info *ci) +{ + unsigned i; + + if (ci->folders) { + for (i = 0; i < ci->numFolders; i++) + free_Folder(&(ci->folders[i])); + free(ci->folders); + } +} + +static int +read_CodersInfo(struct archive_read *a, struct _7z_coders_info *ci) +{ + const unsigned char *p; + struct _7z_digests digest; + unsigned i; + + memset(ci, 0, sizeof(*ci)); + memset(&digest, 0, sizeof(digest)); + + if ((p = header_bytes(a, 1)) == NULL) + goto failed; + if (*p != kFolder) + goto failed; + + /* + * Read NumFolders. + */ + if (parse_7zip_uint64(a, &(ci->numFolders)) < 0) + goto failed; + if (UMAX_ENTRY < ci->numFolders) + return (-1); + + /* + * Read External. + */ + if ((p = header_bytes(a, 1)) == NULL) + goto failed; + switch (*p) { + case 0: + ci->folders = + calloc((size_t)ci->numFolders, sizeof(*ci->folders)); + if (ci->folders == NULL) + return (-1); + for (i = 0; i < ci->numFolders; i++) { + if (read_Folder(a, &(ci->folders[i])) < 0) + goto failed; + } + break; + case 1: + if (parse_7zip_uint64(a, &(ci->dataStreamIndex)) < 0) + return (-1); + if (UMAX_ENTRY < ci->dataStreamIndex) + return (-1); + if (ci->numFolders > 0) { + archive_set_error(&a->archive, -1, + "Malformed 7-Zip archive"); + goto failed; + } + break; + default: + archive_set_error(&a->archive, -1, + "Malformed 7-Zip archive"); + goto failed; + } + + if ((p = header_bytes(a, 1)) == NULL) + goto failed; + if (*p != kCodersUnPackSize) + goto failed; + + for (i = 0; i < ci->numFolders; i++) { + struct _7z_folder *folder = &(ci->folders[i]); + unsigned j; + + folder->unPackSize = + calloc((size_t)folder->numOutStreams, sizeof(*folder->unPackSize)); + if (folder->unPackSize == NULL) + goto failed; + for (j = 0; j < folder->numOutStreams; j++) { + if (parse_7zip_uint64(a, &(folder->unPackSize[j])) < 0) + goto failed; + } + } + + /* + * Read CRCs. + */ + if ((p = header_bytes(a, 1)) == NULL) + goto failed; + if (*p == kEnd) + return (0); + if (*p != kCRC) + goto failed; + if (read_Digests(a, &digest, (size_t)ci->numFolders) < 0) + goto failed; + for (i = 0; i < ci->numFolders; i++) { + ci->folders[i].digest_defined = digest.defineds[i]; + ci->folders[i].digest = digest.digests[i]; + } + + /* + * Must be kEnd. + */ + if ((p = header_bytes(a, 1)) == NULL) + goto failed; + if (*p != kEnd) + goto failed; + free_Digest(&digest); + return (0); +failed: + free_Digest(&digest); + return (-1); +} + +static uint64_t +folder_uncompressed_size(struct _7z_folder *f) +{ + int n = (int)f->numOutStreams; + unsigned pairs = (unsigned)f->numBindPairs; + + while (--n >= 0) { + unsigned i; + for (i = 0; i < pairs; i++) { + if (f->bindPairs[i].outIndex == (uint64_t)n) + break; + } + if (i >= pairs) + return (f->unPackSize[n]); + } + return (0); +} + +static void +free_SubStreamsInfo(struct _7z_substream_info *ss) +{ + free(ss->unpackSizes); + free(ss->digestsDefined); + free(ss->digests); +} + +static int +read_SubStreamsInfo(struct archive_read *a, struct _7z_substream_info *ss, + struct _7z_folder *f, size_t numFolders) +{ + const unsigned char *p; + uint64_t *usizes; + size_t unpack_streams; + int type; + unsigned i; + uint32_t numDigests; + + memset(ss, 0, sizeof(*ss)); + + for (i = 0; i < numFolders; i++) + f[i].numUnpackStreams = 1; + + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + type = *p; + + if (type == kNumUnPackStream) { + unpack_streams = 0; + for (i = 0; i < numFolders; i++) { + if (parse_7zip_uint64(a, &(f[i].numUnpackStreams)) < 0) + return (-1); + if (UMAX_ENTRY < f[i].numUnpackStreams) + return (-1); + if (unpack_streams > SIZE_MAX - UMAX_ENTRY) { + return (-1); + } + unpack_streams += (size_t)f[i].numUnpackStreams; + } + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + type = *p; + } else + unpack_streams = numFolders; + + ss->unpack_streams = unpack_streams; + if (unpack_streams) { + ss->unpackSizes = calloc(unpack_streams, + sizeof(*ss->unpackSizes)); + ss->digestsDefined = calloc(unpack_streams, + sizeof(*ss->digestsDefined)); + ss->digests = calloc(unpack_streams, + sizeof(*ss->digests)); + if (ss->unpackSizes == NULL || ss->digestsDefined == NULL || + ss->digests == NULL) + return (-1); + } + + usizes = ss->unpackSizes; + for (i = 0; i < numFolders; i++) { + unsigned pack; + uint64_t sum; + + if (f[i].numUnpackStreams == 0) + continue; + + sum = 0; + if (type == kSize) { + for (pack = 1; pack < f[i].numUnpackStreams; pack++) { + if (parse_7zip_uint64(a, usizes) < 0) + return (-1); + sum += *usizes++; + } + } + *usizes++ = folder_uncompressed_size(&f[i]) - sum; + } + + if (type == kSize) { + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + type = *p; + } + + for (i = 0; i < unpack_streams; i++) { + ss->digestsDefined[i] = 0; + ss->digests[i] = 0; + } + + numDigests = 0; + for (i = 0; i < numFolders; i++) { + if (f[i].numUnpackStreams != 1 || !f[i].digest_defined) + numDigests += (uint32_t)f[i].numUnpackStreams; + } + + if (type == kCRC) { + struct _7z_digests tmpDigests; + unsigned char *digestsDefined = ss->digestsDefined; + uint32_t * digests = ss->digests; + int di = 0; + + memset(&tmpDigests, 0, sizeof(tmpDigests)); + if (read_Digests(a, &(tmpDigests), numDigests) < 0) { + free_Digest(&tmpDigests); + return (-1); + } + for (i = 0; i < numFolders; i++) { + if (f[i].numUnpackStreams == 1 && f[i].digest_defined) { + *digestsDefined++ = 1; + *digests++ = f[i].digest; + } else { + unsigned j; + + for (j = 0; j < f[i].numUnpackStreams; + j++, di++) { + *digestsDefined++ = + tmpDigests.defineds[di]; + *digests++ = + tmpDigests.digests[di]; + } + } + } + free_Digest(&tmpDigests); + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + type = *p; + } + + /* + * Must be kEnd. + */ + if (type != kEnd) + return (-1); + return (0); +} + +static void +free_StreamsInfo(struct _7z_stream_info *si) +{ + free_PackInfo(&(si->pi)); + free_CodersInfo(&(si->ci)); + free_SubStreamsInfo(&(si->ss)); +} + +static int +read_StreamsInfo(struct archive_read *a, struct _7z_stream_info *si) +{ + struct _7zip *zip = (struct _7zip *)a->format->data; + const unsigned char *p; + unsigned i; + + memset(si, 0, sizeof(*si)); + + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + if (*p == kPackInfo) { + uint64_t packPos; + + if (read_PackInfo(a, &(si->pi)) < 0) + return (-1); + + if (si->pi.positions == NULL || si->pi.sizes == NULL) + return (-1); + /* + * Calculate packed stream positions. + */ + packPos = si->pi.pos; + for (i = 0; i < si->pi.numPackStreams; i++) { + si->pi.positions[i] = packPos; + packPos += si->pi.sizes[i]; + if (packPos > zip->header_offset) + return (-1); + } + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + } + if (*p == kUnPackInfo) { + uint32_t packIndex; + struct _7z_folder *f; + + if (read_CodersInfo(a, &(si->ci)) < 0) + return (-1); + + /* + * Calculate packed stream indexes. + */ + packIndex = 0; + f = si->ci.folders; + for (i = 0; i < si->ci.numFolders; i++) { + f[i].packIndex = packIndex; + packIndex += (uint32_t)f[i].numPackedStreams; + if (packIndex > si->pi.numPackStreams) + return (-1); + } + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + } + + if (*p == kSubStreamsInfo) { + if (read_SubStreamsInfo(a, &(si->ss), + si->ci.folders, (size_t)si->ci.numFolders) < 0) + return (-1); + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + } + + /* + * Must be kEnd. + */ + if (*p != kEnd) + return (-1); + return (0); +} + +static void +free_Header(struct _7z_header_info *h) +{ + free(h->emptyStreamBools); + free(h->emptyFileBools); + free(h->antiBools); + free(h->attrBools); +} + +static int +read_Header(struct archive_read *a, struct _7z_header_info *h, + int check_header_id) +{ + struct _7zip *zip = (struct _7zip *)a->format->data; + const unsigned char *p; + struct _7z_folder *folders; + struct _7z_stream_info *si = &(zip->si); + struct _7zip_entry *entries; + uint32_t folderIndex, indexInFolder; + unsigned i; + int eindex, empty_streams, sindex; + + if (check_header_id) { + /* + * Read Header. + */ + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + if (*p != kHeader) + return (-1); + } + + /* + * Read ArchiveProperties. + */ + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + if (*p == kArchiveProperties) { + for (;;) { + uint64_t size; + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + if (*p == 0) + break; + if (parse_7zip_uint64(a, &size) < 0) + return (-1); + } + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + } + + /* + * Read MainStreamsInfo. + */ + if (*p == kMainStreamsInfo) { + if (read_StreamsInfo(a, &(zip->si)) < 0) + return (-1); + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + } + if (*p == kEnd) + return (0); + + /* + * Read FilesInfo. + */ + if (*p != kFilesInfo) + return (-1); + + if (parse_7zip_uint64(a, &(zip->numFiles)) < 0) + return (-1); + if (UMAX_ENTRY < zip->numFiles) + return (-1); + + zip->entries = calloc((size_t)zip->numFiles, sizeof(*zip->entries)); + if (zip->entries == NULL) + return (-1); + entries = zip->entries; + + empty_streams = 0; + for (;;) { + int type; + uint64_t size; + size_t ll; + + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + type = *p; + if (type == kEnd) + break; + + if (parse_7zip_uint64(a, &size) < 0) + return (-1); + if (zip->header_bytes_remaining < size) + return (-1); + ll = (size_t)size; + + switch (type) { + case kEmptyStream: + if (h->emptyStreamBools != NULL) + return (-1); + h->emptyStreamBools = calloc((size_t)zip->numFiles, + sizeof(*h->emptyStreamBools)); + if (h->emptyStreamBools == NULL) + return (-1); + if (read_Bools( + a, h->emptyStreamBools, (size_t)zip->numFiles) < 0) + return (-1); + empty_streams = 0; + for (i = 0; i < zip->numFiles; i++) { + if (h->emptyStreamBools[i]) + empty_streams++; + } + break; + case kEmptyFile: + if (empty_streams <= 0) { + /* Unexcepted sequence. Skip this. */ + if (header_bytes(a, ll) == NULL) + return (-1); + break; + } + if (h->emptyFileBools != NULL) + return (-1); + h->emptyFileBools = calloc(empty_streams, + sizeof(*h->emptyFileBools)); + if (h->emptyFileBools == NULL) + return (-1); + if (read_Bools(a, h->emptyFileBools, empty_streams) < 0) + return (-1); + break; + case kAnti: + if (empty_streams <= 0) { + /* Unexcepted sequence. Skip this. */ + if (header_bytes(a, ll) == NULL) + return (-1); + break; + } + if (h->antiBools != NULL) + return (-1); + h->antiBools = calloc(empty_streams, + sizeof(*h->antiBools)); + if (h->antiBools == NULL) + return (-1); + if (read_Bools(a, h->antiBools, empty_streams) < 0) + return (-1); + break; + case kCTime: + case kATime: + case kMTime: + if (read_Times(a, h, type) < 0) + return (-1); + break; + case kName: + { + unsigned char *np; + size_t nl, nb; + + /* Skip one byte. */ + if ((p = header_bytes(a, 1)) == NULL) + return (-1); + ll--; + + if ((ll & 1) || ll < zip->numFiles * 4) + return (-1); + + if (zip->entry_names != NULL) + return (-1); + zip->entry_names = malloc(ll); + if (zip->entry_names == NULL) + return (-1); + np = zip->entry_names; + nb = ll; + /* + * Copy whole file names. + * NOTE: This loop prevents from expanding + * the uncompressed buffer in order not to + * use extra memory resource. + */ + while (nb) { + size_t b; + if (nb > UBUFF_SIZE) + b = UBUFF_SIZE; + else + b = nb; + if ((p = header_bytes(a, b)) == NULL) + return (-1); + memcpy(np, p, b); + np += b; + nb -= b; + } + np = zip->entry_names; + nl = ll; + + for (i = 0; i < zip->numFiles; i++) { + entries[i].utf16name = np; +#if defined(_WIN32) && !defined(__CYGWIN__) && defined(_DEBUG) + entries[i].wname = (wchar_t *)np; +#endif + + /* Find a terminator. */ + while (nl >= 2 && (np[0] || np[1])) { + np += 2; + nl -= 2; + } + if (nl < 2) + return (-1);/* Terminator not found */ + entries[i].name_len = np - entries[i].utf16name; + np += 2; + nl -= 2; + } + break; + } + case kAttributes: + { + int allAreDefined; + + if ((p = header_bytes(a, 2)) == NULL) + return (-1); + allAreDefined = *p; + if (h->attrBools != NULL) + return (-1); + h->attrBools = calloc((size_t)zip->numFiles, + sizeof(*h->attrBools)); + if (h->attrBools == NULL) + return (-1); + if (allAreDefined) + memset(h->attrBools, 1, (size_t)zip->numFiles); + else { + if (read_Bools(a, h->attrBools, + (size_t)zip->numFiles) < 0) + return (-1); + } + for (i = 0; i < zip->numFiles; i++) { + if (h->attrBools[i]) { + if ((p = header_bytes(a, 4)) == NULL) + return (-1); + entries[i].attr = archive_le32dec(p); + } + } + break; + } + case kDummy: + if (ll == 0) + break; + __LA_FALLTHROUGH; + default: + if (header_bytes(a, ll) == NULL) + return (-1); + break; + } + } + + /* + * Set up entry's attributes. + */ + folders = si->ci.folders; + eindex = sindex = 0; + folderIndex = indexInFolder = 0; + for (i = 0; i < zip->numFiles; i++) { + if (h->emptyStreamBools == NULL || h->emptyStreamBools[i] == 0) + entries[i].flg |= HAS_STREAM; + /* The high 16 bits of attributes is a posix file mode. */ + entries[i].mode = entries[i].attr >> 16; + if (entries[i].flg & HAS_STREAM) { + if ((size_t)sindex >= si->ss.unpack_streams) + return (-1); + if (entries[i].mode == 0) + entries[i].mode = AE_IFREG | 0666; + if (si->ss.digestsDefined[sindex]) + entries[i].flg |= CRC32_IS_SET; + entries[i].ssIndex = sindex; + sindex++; + } else { + int dir; + if (h->emptyFileBools == NULL) + dir = 1; + else { + if (h->emptyFileBools[eindex]) + dir = 0; + else + dir = 1; + eindex++; + } + if (entries[i].mode == 0) { + if (dir) + entries[i].mode = AE_IFDIR | 0777; + else + entries[i].mode = AE_IFREG | 0666; + } else if (dir && + (entries[i].mode & AE_IFMT) != AE_IFDIR) { + entries[i].mode &= ~AE_IFMT; + entries[i].mode |= AE_IFDIR; + } + if ((entries[i].mode & AE_IFMT) == AE_IFDIR && + entries[i].name_len >= 2 && + (entries[i].utf16name[entries[i].name_len-2] != '/' || + entries[i].utf16name[entries[i].name_len-1] != 0)) { + entries[i].utf16name[entries[i].name_len] = '/'; + entries[i].utf16name[entries[i].name_len+1] = 0; + entries[i].name_len += 2; + } + entries[i].ssIndex = -1; + } + if (entries[i].attr & 0x01) + entries[i].mode &= ~0222;/* Read only. */ + + if ((entries[i].flg & HAS_STREAM) == 0 && indexInFolder == 0) { + /* + * The entry is an empty file or a directory file, + * those both have no contents. + */ + entries[i].folderIndex = -1; + continue; + } + if (indexInFolder == 0) { + for (;;) { + if (folderIndex >= si->ci.numFolders) + return (-1); + if (folders[folderIndex].numUnpackStreams) + break; + folderIndex++; + } + } + entries[i].folderIndex = folderIndex; + if ((entries[i].flg & HAS_STREAM) == 0) + continue; + indexInFolder++; + if (indexInFolder >= folders[folderIndex].numUnpackStreams) { + folderIndex++; + indexInFolder = 0; + } + } + + return (0); +} + +#define EPOC_TIME ARCHIVE_LITERAL_ULL(116444736000000000) +static void +fileTimeToUtc(uint64_t fileTime, time_t *timep, long *ns) +{ + + if (fileTime >= EPOC_TIME) { + fileTime -= EPOC_TIME; + /* milli seconds base */ + *timep = (time_t)(fileTime / 10000000); + /* nano seconds base */ + *ns = (long)(fileTime % 10000000) * 100; + } else { + *timep = 0; + *ns = 0; + } +} + +static int +read_Times(struct archive_read *a, struct _7z_header_info *h, int type) +{ + struct _7zip *zip = (struct _7zip *)a->format->data; + const unsigned char *p; + struct _7zip_entry *entries = zip->entries; + unsigned char *timeBools; + int allAreDefined; + unsigned i; + + timeBools = calloc((size_t)zip->numFiles, sizeof(*timeBools)); + if (timeBools == NULL) + return (-1); + + /* Read allAreDefined. */ + if ((p = header_bytes(a, 1)) == NULL) + goto failed; + allAreDefined = *p; + if (allAreDefined) + memset(timeBools, 1, (size_t)zip->numFiles); + else { + if (read_Bools(a, timeBools, (size_t)zip->numFiles) < 0) + goto failed; + } + + /* Read external. */ + if ((p = header_bytes(a, 1)) == NULL) + goto failed; + if (*p) { + if (parse_7zip_uint64(a, &(h->dataIndex)) < 0) + goto failed; + if (UMAX_ENTRY < h->dataIndex) + goto failed; + } + + for (i = 0; i < zip->numFiles; i++) { + if (!timeBools[i]) + continue; + if ((p = header_bytes(a, 8)) == NULL) + goto failed; + switch (type) { + case kCTime: + fileTimeToUtc(archive_le64dec(p), + &(entries[i].ctime), + &(entries[i].ctime_ns)); + entries[i].flg |= CTIME_IS_SET; + break; + case kATime: + fileTimeToUtc(archive_le64dec(p), + &(entries[i].atime), + &(entries[i].atime_ns)); + entries[i].flg |= ATIME_IS_SET; + break; + case kMTime: + fileTimeToUtc(archive_le64dec(p), + &(entries[i].mtime), + &(entries[i].mtime_ns)); + entries[i].flg |= MTIME_IS_SET; + break; + } + } + + free(timeBools); + return (0); +failed: + free(timeBools); + return (-1); +} + +static int +decode_encoded_header_info(struct archive_read *a, struct _7z_stream_info *si) +{ + struct _7zip *zip = (struct _7zip *)a->format->data; + + errno = 0; + if (read_StreamsInfo(a, si) < 0) { + if (errno == ENOMEM) + archive_set_error(&a->archive, -1, + "Couldn't allocate memory"); + else + archive_set_error(&a->archive, -1, + "Malformed 7-Zip archive"); + return (ARCHIVE_FATAL); + } + + if (si->pi.numPackStreams == 0 || si->ci.numFolders == 0) { + archive_set_error(&a->archive, -1, "Malformed 7-Zip archive"); + return (ARCHIVE_FATAL); + } + + if (zip->header_offset < si->pi.pos + si->pi.sizes[0] || + (int64_t)(si->pi.pos + si->pi.sizes[0]) < 0 || + si->pi.sizes[0] == 0 || (int64_t)si->pi.pos < 0) { + archive_set_error(&a->archive, -1, "Malformed Header offset"); + return (ARCHIVE_FATAL); + } + + return (ARCHIVE_OK); +} + +static const unsigned char * +header_bytes(struct archive_read *a, size_t rbytes) +{ + struct _7zip *zip = (struct _7zip *)a->format->data; + const unsigned char *p; + + if (zip->header_bytes_remaining < rbytes) + return (NULL); + if (zip->pack_stream_bytes_unconsumed) + read_consume(a); + + if (zip->header_is_encoded == 0) { + p = __archive_read_ahead(a, rbytes, NULL); + if (p == NULL) + return (NULL); + zip->header_bytes_remaining -= rbytes; + zip->pack_stream_bytes_unconsumed = rbytes; + } else { + const void *buff; + ssize_t bytes; + + bytes = read_stream(a, &buff, rbytes, rbytes); + if (bytes <= 0) + return (NULL); + zip->header_bytes_remaining -= bytes; + p = buff; + } + + /* Update checksum */ + zip->header_crc32 = crc32(zip->header_crc32, p, (unsigned)rbytes); + return (p); +} + +static int +slurp_central_directory(struct archive_read *a, struct _7zip *zip, + struct _7z_header_info *header) +{ + const unsigned char *p; + uint64_t next_header_offset; + uint64_t next_header_size; + uint32_t next_header_crc; + ssize_t bytes_avail; + int check_header_crc, r; + + if ((p = __archive_read_ahead(a, 32, &bytes_avail)) == NULL) + return (ARCHIVE_FATAL); + + if ((p[0] == 'M' && p[1] == 'Z') || memcmp(p, "\x7F\x45LF", 4) == 0) { + /* This is an executable ? Must be self-extracting... */ + r = skip_sfx(a, bytes_avail); + if (r < ARCHIVE_WARN) + return (r); + if ((p = __archive_read_ahead(a, 32, &bytes_avail)) == NULL) + return (ARCHIVE_FATAL); + } + zip->seek_base += 32; + + if (memcmp(p, _7ZIP_SIGNATURE, 6) != 0) { + archive_set_error(&a->archive, -1, "Not 7-Zip archive file"); + return (ARCHIVE_FATAL); + } + + /* CRC check. */ + if (crc32(0, (const unsigned char *)p + 12, 20) + != archive_le32dec(p + 8)) { + archive_set_error(&a->archive, -1, "Header CRC error"); + return (ARCHIVE_FATAL); + } + + next_header_offset = archive_le64dec(p + 12); + next_header_size = archive_le64dec(p + 20); + next_header_crc = archive_le32dec(p + 28); + + if (next_header_size == 0) + /* There is no entry in an archive file. */ + return (ARCHIVE_EOF); + + if (((int64_t)next_header_offset) < 0) { + archive_set_error(&a->archive, -1, "Malformed 7-Zip archive"); + return (ARCHIVE_FATAL); + } + __archive_read_consume(a, 32); + if (next_header_offset != 0) { + if (bytes_avail >= (ssize_t)next_header_offset) + __archive_read_consume(a, next_header_offset); + else if (__archive_read_seek(a, + next_header_offset + zip->seek_base, SEEK_SET) < 0) + return (ARCHIVE_FATAL); + } + zip->stream_offset = next_header_offset; + zip->header_offset = next_header_offset; + zip->header_bytes_remaining = next_header_size; + zip->header_crc32 = 0; + zip->header_is_encoded = 0; + zip->header_is_being_read = 1; + zip->has_encrypted_entries = 0; + check_header_crc = 1; + + if ((p = header_bytes(a, 1)) == NULL) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated 7-Zip file body"); + return (ARCHIVE_FATAL); + } + /* Parse ArchiveProperties. */ + switch (p[0]) { + case kEncodedHeader: + /* + * The archive has an encoded header and we have to decode it + * in order to parse the header correctly. + */ + r = decode_encoded_header_info(a, &(zip->si)); + + /* Check the EncodedHeader CRC.*/ + if (r == 0 && zip->header_crc32 != next_header_crc) { + archive_set_error(&a->archive, -1, + "Damaged 7-Zip archive"); + r = -1; + } + if (r == 0) { + if (zip->si.ci.folders[0].digest_defined) + next_header_crc = zip->si.ci.folders[0].digest; + else + check_header_crc = 0; + if (zip->pack_stream_bytes_unconsumed) + read_consume(a); + r = setup_decode_folder(a, zip->si.ci.folders, 1); + if (r == 0) { + zip->header_bytes_remaining = + zip->folder_outbytes_remaining; + r = seek_pack(a); + } + } + /* Clean up StreamsInfo. */ + free_StreamsInfo(&(zip->si)); + memset(&(zip->si), 0, sizeof(zip->si)); + if (r < 0) + return (ARCHIVE_FATAL); + zip->header_is_encoded = 1; + zip->header_crc32 = 0; + /* FALL THROUGH */ + case kHeader: + /* + * Parse the header. + */ + errno = 0; + r = read_Header(a, header, zip->header_is_encoded); + if (r < 0) { + if (errno == ENOMEM) + archive_set_error(&a->archive, -1, + "Couldn't allocate memory"); + else + archive_set_error(&a->archive, -1, + "Damaged 7-Zip archive"); + return (ARCHIVE_FATAL); + } + + /* + * Must be kEnd. + */ + if ((p = header_bytes(a, 1)) == NULL ||*p != kEnd) { + archive_set_error(&a->archive, -1, + "Malformed 7-Zip archive"); + return (ARCHIVE_FATAL); + } + + /* Check the Header CRC.*/ + if (check_header_crc && zip->header_crc32 != next_header_crc) { + archive_set_error(&a->archive, -1, + "Malformed 7-Zip archive"); + return (ARCHIVE_FATAL); + } + break; + default: + archive_set_error(&a->archive, -1, + "Unexpected Property ID = %X", p[0]); + return (ARCHIVE_FATAL); + } + + /* Clean up variables be used for decoding the archive header */ + zip->pack_stream_remaining = 0; + zip->pack_stream_index = 0; + zip->folder_outbytes_remaining = 0; + zip->uncompressed_buffer_bytes_remaining = 0; + zip->pack_stream_bytes_unconsumed = 0; + zip->header_is_being_read = 0; + + return (ARCHIVE_OK); +} + +static ssize_t +get_uncompressed_data(struct archive_read *a, const void **buff, size_t size, + size_t minimum) +{ + struct _7zip *zip = (struct _7zip *)a->format->data; + ssize_t bytes_avail; + + if (zip->codec == _7Z_COPY && zip->codec2 == (unsigned long)-1) { + /* Copy mode. */ + + *buff = __archive_read_ahead(a, minimum, &bytes_avail); + if (bytes_avail <= 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated 7-Zip file data"); + return (ARCHIVE_FATAL); + } + if ((size_t)bytes_avail > + zip->uncompressed_buffer_bytes_remaining) + bytes_avail = (ssize_t) + zip->uncompressed_buffer_bytes_remaining; + if ((size_t)bytes_avail > size) + bytes_avail = (ssize_t)size; + + zip->pack_stream_bytes_unconsumed = bytes_avail; + } else if (zip->uncompressed_buffer_pointer == NULL) { + /* Decompression has failed. */ + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, "Damaged 7-Zip archive"); + return (ARCHIVE_FATAL); + } else { + /* Packed mode. */ + if (minimum > zip->uncompressed_buffer_bytes_remaining) { + /* + * If remaining uncompressed data size is less than + * the minimum size, fill the buffer up to the + * minimum size. + */ + if (extract_pack_stream(a, minimum) < 0) + return (ARCHIVE_FATAL); + } + if (size > zip->uncompressed_buffer_bytes_remaining) + bytes_avail = (ssize_t) + zip->uncompressed_buffer_bytes_remaining; + else + bytes_avail = (ssize_t)size; + *buff = zip->uncompressed_buffer_pointer; + zip->uncompressed_buffer_pointer += bytes_avail; + } + zip->uncompressed_buffer_bytes_remaining -= bytes_avail; + return (bytes_avail); +} + +static ssize_t +extract_pack_stream(struct archive_read *a, size_t minimum) +{ + struct _7zip *zip = (struct _7zip *)a->format->data; + ssize_t bytes_avail; + int r; + + if (zip->codec == _7Z_COPY && zip->codec2 == (unsigned long)-1) { + if (minimum == 0) + minimum = 1; + if (__archive_read_ahead(a, minimum, &bytes_avail) == NULL + || bytes_avail <= 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated 7-Zip file body"); + return (ARCHIVE_FATAL); + } + if (bytes_avail > (ssize_t)zip->pack_stream_inbytes_remaining) + bytes_avail = (ssize_t)zip->pack_stream_inbytes_remaining; + zip->pack_stream_inbytes_remaining -= bytes_avail; + if (bytes_avail > (ssize_t)zip->folder_outbytes_remaining) + bytes_avail = (ssize_t)zip->folder_outbytes_remaining; + zip->folder_outbytes_remaining -= bytes_avail; + zip->uncompressed_buffer_bytes_remaining = bytes_avail; + return (ARCHIVE_OK); + } + + /* If the buffer hasn't been allocated, allocate it now. */ + if (zip->uncompressed_buffer == NULL) { + zip->uncompressed_buffer_size = UBUFF_SIZE; + if (zip->uncompressed_buffer_size < minimum) { + zip->uncompressed_buffer_size = minimum + 1023; + zip->uncompressed_buffer_size &= ~0x3ff; + } + zip->uncompressed_buffer = + malloc(zip->uncompressed_buffer_size); + if (zip->uncompressed_buffer == NULL) { + archive_set_error(&a->archive, ENOMEM, + "No memory for 7-Zip decompression"); + return (ARCHIVE_FATAL); + } + zip->uncompressed_buffer_bytes_remaining = 0; + } else if (zip->uncompressed_buffer_size < minimum || + zip->uncompressed_buffer_bytes_remaining < minimum) { + /* + * Make sure the uncompressed buffer can have bytes + * at least `minimum' bytes. + * NOTE: This case happen when reading the header. + */ + size_t used; + if (zip->uncompressed_buffer_pointer != 0) + used = zip->uncompressed_buffer_pointer - + zip->uncompressed_buffer; + else + used = 0; + if (zip->uncompressed_buffer_size < minimum) { + /* + * Expand the uncompressed buffer up to + * the minimum size. + */ + void *p; + size_t new_size; + + new_size = minimum + 1023; + new_size &= ~0x3ff; + p = realloc(zip->uncompressed_buffer, new_size); + if (p == NULL) { + archive_set_error(&a->archive, ENOMEM, + "No memory for 7-Zip decompression"); + return (ARCHIVE_FATAL); + } + zip->uncompressed_buffer = (unsigned char *)p; + zip->uncompressed_buffer_size = new_size; + } + /* + * Move unconsumed bytes to the head. + */ + if (used) { + memmove(zip->uncompressed_buffer, + zip->uncompressed_buffer + used, + zip->uncompressed_buffer_bytes_remaining); + } + } else + zip->uncompressed_buffer_bytes_remaining = 0; + zip->uncompressed_buffer_pointer = NULL; + for (;;) { + size_t bytes_in, bytes_out; + const void *buff_in; + unsigned char *buff_out; + int end_of_data; + + /* + * Note: '1' here is a performance optimization. + * Recall that the decompression layer returns a count of + * available bytes; asking for more than that forces the + * decompressor to combine reads by copying data. + */ + buff_in = __archive_read_ahead(a, 1, &bytes_avail); + if (bytes_avail <= 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated 7-Zip file body"); + return (ARCHIVE_FATAL); + } + + buff_out = zip->uncompressed_buffer + + zip->uncompressed_buffer_bytes_remaining; + bytes_out = zip->uncompressed_buffer_size + - zip->uncompressed_buffer_bytes_remaining; + bytes_in = bytes_avail; + if (bytes_in > zip->pack_stream_inbytes_remaining) + bytes_in = (size_t)zip->pack_stream_inbytes_remaining; + /* Drive decompression. */ + r = decompress(a, zip, buff_out, &bytes_out, + buff_in, &bytes_in); + switch (r) { + case ARCHIVE_OK: + end_of_data = 0; + break; + case ARCHIVE_EOF: + end_of_data = 1; + break; + default: + return (ARCHIVE_FATAL); + } + zip->pack_stream_inbytes_remaining -= bytes_in; + if (bytes_out > zip->folder_outbytes_remaining) + bytes_out = (size_t)zip->folder_outbytes_remaining; + zip->folder_outbytes_remaining -= bytes_out; + zip->uncompressed_buffer_bytes_remaining += bytes_out; + zip->pack_stream_bytes_unconsumed = bytes_in; + + /* + * Continue decompression until uncompressed_buffer is full. + */ + if (zip->uncompressed_buffer_bytes_remaining == + zip->uncompressed_buffer_size) + break; + if (zip->codec2 == _7Z_X86 && zip->odd_bcj_size && + zip->uncompressed_buffer_bytes_remaining + 5 > + zip->uncompressed_buffer_size) + break; + if (zip->pack_stream_inbytes_remaining == 0 && + zip->folder_outbytes_remaining == 0) + break; + if (end_of_data || (bytes_in == 0 && bytes_out == 0)) { + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, "Damaged 7-Zip archive"); + return (ARCHIVE_FATAL); + } + read_consume(a); + } + if (zip->uncompressed_buffer_bytes_remaining < minimum) { + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, "Damaged 7-Zip archive"); + return (ARCHIVE_FATAL); + } + zip->uncompressed_buffer_pointer = zip->uncompressed_buffer; + return (ARCHIVE_OK); +} + +static int +seek_pack(struct archive_read *a) +{ + struct _7zip *zip = (struct _7zip *)a->format->data; + int64_t pack_offset; + + if (zip->pack_stream_remaining <= 0) { + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, "Damaged 7-Zip archive"); + return (ARCHIVE_FATAL); + } + zip->pack_stream_inbytes_remaining = + zip->si.pi.sizes[zip->pack_stream_index]; + pack_offset = zip->si.pi.positions[zip->pack_stream_index]; + if (zip->stream_offset != pack_offset) { + if (0 > __archive_read_seek(a, pack_offset + zip->seek_base, + SEEK_SET)) + return (ARCHIVE_FATAL); + zip->stream_offset = pack_offset; + } + zip->pack_stream_index++; + zip->pack_stream_remaining--; + return (ARCHIVE_OK); +} + +static ssize_t +read_stream(struct archive_read *a, const void **buff, size_t size, + size_t minimum) +{ + struct _7zip *zip = (struct _7zip *)a->format->data; + uint64_t skip_bytes = 0; + ssize_t r; + + if (zip->uncompressed_buffer_bytes_remaining == 0) { + if (zip->pack_stream_inbytes_remaining > 0) { + r = extract_pack_stream(a, 0); + if (r < 0) + return (r); + return (get_uncompressed_data(a, buff, size, minimum)); + } else if (zip->folder_outbytes_remaining > 0) { + /* Extract a remaining pack stream. */ + r = extract_pack_stream(a, 0); + if (r < 0) + return (r); + return (get_uncompressed_data(a, buff, size, minimum)); + } + } else + return (get_uncompressed_data(a, buff, size, minimum)); + + /* + * Current pack stream has been consumed. + */ + if (zip->pack_stream_remaining == 0) { + if (zip->header_is_being_read) { + /* Invalid sequence. This might happen when + * reading a malformed archive. */ + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, "Malformed 7-Zip archive"); + return (ARCHIVE_FATAL); + } + + /* + * All current folder's pack streams have been + * consumed. Switch to next folder. + */ + if (zip->folder_index == 0 && + (zip->si.ci.folders[zip->entry->folderIndex].skipped_bytes + || zip->folder_index != zip->entry->folderIndex)) { + zip->folder_index = zip->entry->folderIndex; + skip_bytes = + zip->si.ci.folders[zip->folder_index].skipped_bytes; + } + + if (zip->folder_index >= zip->si.ci.numFolders) { + /* + * We have consumed all folders and its pack streams. + */ + *buff = NULL; + return (0); + } + r = setup_decode_folder(a, + &(zip->si.ci.folders[zip->folder_index]), 0); + if (r != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + zip->folder_index++; + } + + /* + * Switch to next pack stream. + */ + r = seek_pack(a); + if (r < 0) + return (r); + + /* Extract a new pack stream. */ + r = extract_pack_stream(a, 0); + if (r < 0) + return (r); + + /* + * Skip the bytes we already has skipped in skip_stream(). + */ + while (skip_bytes) { + ssize_t skipped; + + if (zip->uncompressed_buffer_bytes_remaining == 0) { + if (zip->pack_stream_inbytes_remaining > 0) { + r = extract_pack_stream(a, 0); + if (r < 0) + return (r); + } else if (zip->folder_outbytes_remaining > 0) { + /* Extract a remaining pack stream. */ + r = extract_pack_stream(a, 0); + if (r < 0) + return (r); + } else { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated 7-Zip file body"); + return (ARCHIVE_FATAL); + } + } + skipped = get_uncompressed_data( + a, buff, (size_t)skip_bytes, 0); + if (skipped < 0) + return (skipped); + skip_bytes -= skipped; + if (zip->pack_stream_bytes_unconsumed) + read_consume(a); + } + + return (get_uncompressed_data(a, buff, size, minimum)); +} + +static int +setup_decode_folder(struct archive_read *a, struct _7z_folder *folder, + int header) +{ + struct _7zip *zip = (struct _7zip *)a->format->data; + const struct _7z_coder *coder1, *coder2; + const char *cname = (header)?"archive header":"file content"; + unsigned i; + int r, found_bcj2 = 0; + + /* + * Release the memory which the previous folder used for BCJ2. + */ + for (i = 0; i < 3; i++) { + free(zip->sub_stream_buff[i]); + zip->sub_stream_buff[i] = NULL; + } + + /* + * Initialize a stream reader. + */ + zip->pack_stream_remaining = (unsigned)folder->numPackedStreams; + zip->pack_stream_index = (unsigned)folder->packIndex; + zip->folder_outbytes_remaining = folder_uncompressed_size(folder); + zip->uncompressed_buffer_bytes_remaining = 0; + + /* + * Check coder types. + */ + for (i = 0; i < folder->numCoders; i++) { + switch(folder->coders[i].codec) { + case _7Z_CRYPTO_MAIN_ZIP: + case _7Z_CRYPTO_RAR_29: + case _7Z_CRYPTO_AES_256_SHA_256: { + /* For entry that is associated with this folder, mark + it as encrypted (data+metadata). */ + zip->has_encrypted_entries = 1; + if (a->entry) { + archive_entry_set_is_data_encrypted(a->entry, 1); + archive_entry_set_is_metadata_encrypted(a->entry, 1); + } + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "The %s is encrypted, " + "but currently not supported", cname); + return (ARCHIVE_FATAL); + } + case _7Z_X86_BCJ2: { + found_bcj2++; + break; + } + } + } + /* Now that we've checked for encryption, if there were still no + * encrypted entries found we can say for sure that there are none. + */ + if (zip->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { + zip->has_encrypted_entries = 0; + } + + if ((folder->numCoders > 2 && !found_bcj2) || found_bcj2 > 1) { + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "The %s is encoded with many filters, " + "but currently not supported", cname); + return (ARCHIVE_FATAL); + } + coder1 = &(folder->coders[0]); + if (folder->numCoders == 2) + coder2 = &(folder->coders[1]); + else + coder2 = NULL; + + if (found_bcj2) { + /* + * Preparation to decode BCJ2. + * Decoding BCJ2 requires four sources. Those are at least, + * as far as I know, two types of the storage form. + */ + const struct _7z_coder *fc = folder->coders; + static const struct _7z_coder coder_copy = {0, 1, 1, 0, NULL}; + const struct _7z_coder *scoder[3] = + {&coder_copy, &coder_copy, &coder_copy}; + const void *buff; + ssize_t bytes; + unsigned char *b[3] = {NULL, NULL, NULL}; + uint64_t sunpack[3] ={-1, -1, -1}; + size_t s[3] = {0, 0, 0}; + int idx[3] = {0, 1, 2}; + + if (folder->numCoders == 4 && fc[3].codec == _7Z_X86_BCJ2 && + folder->numInStreams == 7 && folder->numOutStreams == 4 && + zip->pack_stream_remaining == 4) { + /* Source type 1 made by 7zr or 7z with -m options. */ + if (folder->bindPairs[0].inIndex == 5) { + /* The form made by 7zr */ + idx[0] = 1; idx[1] = 2; idx[2] = 0; + scoder[1] = &(fc[1]); + scoder[2] = &(fc[0]); + sunpack[1] = folder->unPackSize[1]; + sunpack[2] = folder->unPackSize[0]; + coder1 = &(fc[2]); + } else { + /* + * NOTE: Some patterns do not work. + * work: + * 7z a -m0=BCJ2 -m1=COPY -m2=COPY + * -m3=(any) + * 7z a -m0=BCJ2 -m1=COPY -m2=(any) + * -m3=COPY + * 7z a -m0=BCJ2 -m1=(any) -m2=COPY + * -m3=COPY + * not work: + * other patterns. + * + * We have to handle this like `pipe' or + * our libarchive7s filter frame work, + * decoding the BCJ2 main stream sequentially, + * m3 -> m2 -> m1 -> BCJ2. + * + */ + if (fc[0].codec == _7Z_COPY && + fc[1].codec == _7Z_COPY) + coder1 = &(folder->coders[2]); + else if (fc[0].codec == _7Z_COPY && + fc[2].codec == _7Z_COPY) + coder1 = &(folder->coders[1]); + else if (fc[1].codec == _7Z_COPY && + fc[2].codec == _7Z_COPY) + coder1 = &(folder->coders[0]); + else { + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "Unsupported form of " + "BCJ2 streams"); + return (ARCHIVE_FATAL); + } + } + coder2 = &(fc[3]); + zip->main_stream_bytes_remaining = + (size_t)folder->unPackSize[2]; + } else if (coder2 != NULL && coder2->codec == _7Z_X86_BCJ2 && + zip->pack_stream_remaining == 4 && + folder->numInStreams == 5 && folder->numOutStreams == 2) { + /* Source type 0 made by 7z */ + zip->main_stream_bytes_remaining = + (size_t)folder->unPackSize[0]; + } else { + /* We got an unexpected form. */ + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "Unsupported form of BCJ2 streams"); + return (ARCHIVE_FATAL); + } + + /* Skip the main stream at this time. */ + if ((r = seek_pack(a)) < 0) + return (r); + zip->pack_stream_bytes_unconsumed = + (size_t)zip->pack_stream_inbytes_remaining; + read_consume(a); + + /* Read following three sub streams. */ + for (i = 0; i < 3; i++) { + const struct _7z_coder *coder = scoder[i]; + + if ((r = seek_pack(a)) < 0) { + free(b[0]); free(b[1]); free(b[2]); + return (r); + } + + if (sunpack[i] == (uint64_t)-1) + zip->folder_outbytes_remaining = + zip->pack_stream_inbytes_remaining; + else + zip->folder_outbytes_remaining = sunpack[i]; + + r = init_decompression(a, zip, coder, NULL); + if (r != ARCHIVE_OK) { + free(b[0]); free(b[1]); free(b[2]); + return (ARCHIVE_FATAL); + } + + /* Allocate memory for the decoded data of a sub + * stream. */ + b[i] = malloc((size_t)zip->folder_outbytes_remaining); + if (b[i] == NULL) { + free(b[0]); free(b[1]); free(b[2]); + archive_set_error(&a->archive, ENOMEM, + "No memory for 7-Zip decompression"); + return (ARCHIVE_FATAL); + } + + /* Extract a sub stream. */ + while (zip->pack_stream_inbytes_remaining > 0) { + r = (int)extract_pack_stream(a, 0); + if (r < 0) { + free(b[0]); free(b[1]); free(b[2]); + return (r); + } + bytes = get_uncompressed_data(a, &buff, + zip->uncompressed_buffer_bytes_remaining, + 0); + if (bytes < 0) { + free(b[0]); free(b[1]); free(b[2]); + return ((int)bytes); + } + memcpy(b[i]+s[i], buff, bytes); + s[i] += bytes; + if (zip->pack_stream_bytes_unconsumed) + read_consume(a); + } + } + + /* Set the sub streams to the right place. */ + for (i = 0; i < 3; i++) { + zip->sub_stream_buff[i] = b[idx[i]]; + zip->sub_stream_size[i] = s[idx[i]]; + zip->sub_stream_bytes_remaining[i] = s[idx[i]]; + } + + /* Allocate memory used for decoded main stream bytes. */ + if (zip->tmp_stream_buff == NULL) { + zip->tmp_stream_buff_size = 32 * 1024; + zip->tmp_stream_buff = + malloc(zip->tmp_stream_buff_size); + if (zip->tmp_stream_buff == NULL) { + archive_set_error(&a->archive, ENOMEM, + "No memory for 7-Zip decompression"); + return (ARCHIVE_FATAL); + } + } + zip->tmp_stream_bytes_avail = 0; + zip->tmp_stream_bytes_remaining = 0; + zip->odd_bcj_size = 0; + zip->bcj2_outPos = 0; + + /* + * Reset a stream reader in order to read the main stream + * of BCJ2. + */ + zip->pack_stream_remaining = 1; + zip->pack_stream_index = (unsigned)folder->packIndex; + zip->folder_outbytes_remaining = + folder_uncompressed_size(folder); + zip->uncompressed_buffer_bytes_remaining = 0; + } + + /* + * Initialize the decompressor for the new folder's pack streams. + */ + r = init_decompression(a, zip, coder1, coder2); + if (r != ARCHIVE_OK) + return (ARCHIVE_FATAL); + return (ARCHIVE_OK); +} + +static int64_t +skip_stream(struct archive_read *a, size_t skip_bytes) +{ + struct _7zip *zip = (struct _7zip *)a->format->data; + const void *p; + int64_t skipped_bytes; + size_t bytes = skip_bytes; + + if (zip->folder_index == 0) { + /* + * Optimization for a list mode. + * Avoid unnecessary decoding operations. + */ + zip->si.ci.folders[zip->entry->folderIndex].skipped_bytes + += skip_bytes; + return (skip_bytes); + } + + while (bytes) { + skipped_bytes = read_stream(a, &p, bytes, 0); + if (skipped_bytes < 0) + return (skipped_bytes); + if (skipped_bytes == 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated 7-Zip file body"); + return (ARCHIVE_FATAL); + } + bytes -= (size_t)skipped_bytes; + if (zip->pack_stream_bytes_unconsumed) + read_consume(a); + } + return (skip_bytes); +} + +/* + * Brought from LZMA SDK. + * + * Bra86.c -- Converter for x86 code (BCJ) + * 2008-10-04 : Igor Pavlov : Public domain + * + */ + +#define Test86MSByte(b) ((b) == 0 || (b) == 0xFF) + +static void +x86_Init(struct _7zip *zip) +{ + zip->bcj_state = 0; + zip->bcj_prevPosT = (size_t)0 - 1; + zip->bcj_prevMask = 0; + zip->bcj_ip = 5; +} + +static size_t +x86_Convert(struct _7zip *zip, uint8_t *data, size_t size) +{ + static const uint8_t kMaskToAllowedStatus[8] = {1, 1, 1, 0, 1, 0, 0, 0}; + static const uint8_t kMaskToBitNumber[8] = {0, 1, 2, 2, 3, 3, 3, 3}; + size_t bufferPos, prevPosT; + uint32_t ip, prevMask; + + if (size < 5) + return 0; + + bufferPos = 0; + prevPosT = zip->bcj_prevPosT; + prevMask = zip->bcj_prevMask; + ip = zip->bcj_ip; + + for (;;) { + uint8_t *p = data + bufferPos; + uint8_t *limit = data + size - 4; + + for (; p < limit; p++) + if ((*p & 0xFE) == 0xE8) + break; + bufferPos = (size_t)(p - data); + if (p >= limit) + break; + prevPosT = bufferPos - prevPosT; + if (prevPosT > 3) + prevMask = 0; + else { + prevMask = (prevMask << ((int)prevPosT - 1)) & 0x7; + if (prevMask != 0) { + unsigned char b = + p[4 - kMaskToBitNumber[prevMask]]; + if (!kMaskToAllowedStatus[prevMask] || + Test86MSByte(b)) { + prevPosT = bufferPos; + prevMask = ((prevMask << 1) & 0x7) | 1; + bufferPos++; + continue; + } + } + } + prevPosT = bufferPos; + + if (Test86MSByte(p[4])) { + uint32_t src = ((uint32_t)p[4] << 24) | + ((uint32_t)p[3] << 16) | ((uint32_t)p[2] << 8) | + ((uint32_t)p[1]); + uint32_t dest; + for (;;) { + uint8_t b; + int b_index; + + dest = src - (ip + (uint32_t)bufferPos); + if (prevMask == 0) + break; + b_index = kMaskToBitNumber[prevMask] * 8; + b = (uint8_t)(dest >> (24 - b_index)); + if (!Test86MSByte(b)) + break; + src = dest ^ ((1 << (32 - b_index)) - 1); + } + p[4] = (uint8_t)(~(((dest >> 24) & 1) - 1)); + p[3] = (uint8_t)(dest >> 16); + p[2] = (uint8_t)(dest >> 8); + p[1] = (uint8_t)dest; + bufferPos += 5; + } else { + prevMask = ((prevMask << 1) & 0x7) | 1; + bufferPos++; + } + } + zip->bcj_prevPosT = prevPosT; + zip->bcj_prevMask = prevMask; + zip->bcj_ip += (uint32_t)bufferPos; + return (bufferPos); +} + +/* + * Brought from LZMA SDK. + * + * Bcj2.c -- Converter for x86 code (BCJ2) + * 2008-10-04 : Igor Pavlov : Public domain + * + */ + +#define SZ_ERROR_DATA ARCHIVE_FAILED + +#define IsJcc(b0, b1) ((b0) == 0x0F && ((b1) & 0xF0) == 0x80) +#define IsJ(b0, b1) ((b1 & 0xFE) == 0xE8 || IsJcc(b0, b1)) + +#define kNumTopBits 24 +#define kTopValue ((uint32_t)1 << kNumTopBits) + +#define kNumBitModelTotalBits 11 +#define kBitModelTotal (1 << kNumBitModelTotalBits) +#define kNumMoveBits 5 + +#define RC_READ_BYTE (*buffer++) +#define RC_TEST { if (buffer == bufferLim) return SZ_ERROR_DATA; } +#define RC_INIT2 zip->bcj2_code = 0; zip->bcj2_range = 0xFFFFFFFF; \ + { int ii; for (ii = 0; ii < 5; ii++) { RC_TEST; zip->bcj2_code = (zip->bcj2_code << 8) | RC_READ_BYTE; }} + +#define NORMALIZE if (zip->bcj2_range < kTopValue) { RC_TEST; zip->bcj2_range <<= 8; zip->bcj2_code = (zip->bcj2_code << 8) | RC_READ_BYTE; } + +#define IF_BIT_0(p) ttt = *(p); bound = (zip->bcj2_range >> kNumBitModelTotalBits) * ttt; if (zip->bcj2_code < bound) +#define UPDATE_0(p) zip->bcj2_range = bound; *(p) = (CProb)(ttt + ((kBitModelTotal - ttt) >> kNumMoveBits)); NORMALIZE; +#define UPDATE_1(p) zip->bcj2_range -= bound; zip->bcj2_code -= bound; *(p) = (CProb)(ttt - (ttt >> kNumMoveBits)); NORMALIZE; + +static ssize_t +Bcj2_Decode(struct _7zip *zip, uint8_t *outBuf, size_t outSize) +{ + size_t inPos = 0, outPos = 0; + const uint8_t *buf0, *buf1, *buf2, *buf3; + size_t size0, size1, size2, size3; + const uint8_t *buffer, *bufferLim; + unsigned int i, j; + + size0 = zip->tmp_stream_bytes_remaining; + buf0 = zip->tmp_stream_buff + zip->tmp_stream_bytes_avail - size0; + size1 = zip->sub_stream_bytes_remaining[0]; + buf1 = zip->sub_stream_buff[0] + zip->sub_stream_size[0] - size1; + size2 = zip->sub_stream_bytes_remaining[1]; + buf2 = zip->sub_stream_buff[1] + zip->sub_stream_size[1] - size2; + size3 = zip->sub_stream_bytes_remaining[2]; + buf3 = zip->sub_stream_buff[2] + zip->sub_stream_size[2] - size3; + + buffer = buf3; + bufferLim = buffer + size3; + + if (zip->bcj_state == 0) { + /* + * Initialize. + */ + zip->bcj2_prevByte = 0; + for (i = 0; + i < sizeof(zip->bcj2_p) / sizeof(zip->bcj2_p[0]); i++) + zip->bcj2_p[i] = kBitModelTotal >> 1; + RC_INIT2; + zip->bcj_state = 1; + } + + /* + * Gather the odd bytes of a previous call. + */ + for (i = 0; zip->odd_bcj_size > 0 && outPos < outSize; i++) { + outBuf[outPos++] = zip->odd_bcj[i]; + zip->odd_bcj_size--; + } + + if (outSize == 0) { + zip->bcj2_outPos += outPos; + return (outPos); + } + + for (;;) { + uint8_t b; + CProb *prob; + uint32_t bound; + uint32_t ttt; + + size_t limit = size0 - inPos; + if (outSize - outPos < limit) + limit = outSize - outPos; + + if (zip->bcj_state == 1) { + while (limit != 0) { + uint8_t bb = buf0[inPos]; + outBuf[outPos++] = bb; + if (IsJ(zip->bcj2_prevByte, bb)) { + zip->bcj_state = 2; + break; + } + inPos++; + zip->bcj2_prevByte = bb; + limit--; + } + } + + if (limit == 0 || outPos == outSize) + break; + zip->bcj_state = 1; + + b = buf0[inPos++]; + + if (b == 0xE8) + prob = zip->bcj2_p + zip->bcj2_prevByte; + else if (b == 0xE9) + prob = zip->bcj2_p + 256; + else + prob = zip->bcj2_p + 257; + + IF_BIT_0(prob) { + UPDATE_0(prob) + zip->bcj2_prevByte = b; + } else { + uint32_t dest; + const uint8_t *v; + uint8_t out[4]; + + UPDATE_1(prob) + if (b == 0xE8) { + v = buf1; + if (size1 < 4) + return SZ_ERROR_DATA; + buf1 += 4; + size1 -= 4; + } else { + v = buf2; + if (size2 < 4) + return SZ_ERROR_DATA; + buf2 += 4; + size2 -= 4; + } + dest = (((uint32_t)v[0] << 24) | + ((uint32_t)v[1] << 16) | + ((uint32_t)v[2] << 8) | + ((uint32_t)v[3])) - + ((uint32_t)zip->bcj2_outPos + (uint32_t)outPos + 4); + out[0] = (uint8_t)dest; + out[1] = (uint8_t)(dest >> 8); + out[2] = (uint8_t)(dest >> 16); + out[3] = zip->bcj2_prevByte = (uint8_t)(dest >> 24); + + for (i = 0; i < 4 && outPos < outSize; i++) + outBuf[outPos++] = out[i]; + if (i < 4) { + /* + * Save odd bytes which we could not add into + * the output buffer because of out of space. + */ + zip->odd_bcj_size = 4 -i; + for (; i < 4; i++) { + j = i - 4 + (unsigned)zip->odd_bcj_size; + zip->odd_bcj[j] = out[i]; + } + break; + } + } + } + zip->tmp_stream_bytes_remaining -= inPos; + zip->sub_stream_bytes_remaining[0] = size1; + zip->sub_stream_bytes_remaining[1] = size2; + zip->sub_stream_bytes_remaining[2] = bufferLim - buffer; + zip->bcj2_outPos += outPos; + + return ((ssize_t)outPos); +} + diff --git a/c/archive_read_support_format_all.c b/c/archive_read_support_format_all.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_format_all.c @@ -0,0 +1,89 @@ +/*- + * Copyright (c) 2003-2011 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_read_support_format_all.c 174991 2007-12-30 04:58:22Z kientzle $"); + +#include "archive.h" +#include "archive_private.h" + +int +archive_read_support_format_all(struct archive *a) +{ + archive_check_magic(a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_format_all"); + + /* TODO: It would be nice to compute the ordering + * here automatically so that people who enable just + * a few formats can still get the benefits. That + * may just require the format registration to include + * a "maximum read-ahead" value (anything that uses seek + * would be essentially infinite read-ahead). The core + * bid management can then sort the bidders before calling + * them. + * + * If you implement the above, please return the list below + * to alphabetic order. + */ + + /* + * These bidders are all pretty cheap; they just examine a + * small initial part of the archive. If one of these bids + * high, we can maybe avoid running any of the more expensive + * bidders below. + */ + archive_read_support_format_ar(a); + archive_read_support_format_cpio(a); + archive_read_support_format_empty(a); + archive_read_support_format_lha(a); + archive_read_support_format_mtree(a); + archive_read_support_format_tar(a); + archive_read_support_format_xar(a); + archive_read_support_format_warc(a); + + /* + * Install expensive bidders last. By doing them last, we + * increase the chance that a high bid from someone else will + * make it unnecessary for these to do anything at all. + */ + /* These three have potentially large look-ahead. */ + archive_read_support_format_7zip(a); + archive_read_support_format_cab(a); + archive_read_support_format_rar(a); + archive_read_support_format_rar5(a); + archive_read_support_format_iso9660(a); + /* Seek is really bad, since it forces the read-ahead + * logic to discard buffered data. */ + archive_read_support_format_zip(a); + + /* Note: We always return ARCHIVE_OK here, even if some of the + * above return ARCHIVE_WARN. The intent here is to enable + * "as much as possible." Clients who need specific + * compression should enable those individually so they can + * verify the level of support. */ + /* Clear any warning messages set by the above functions. */ + archive_clear_error(a); + return (ARCHIVE_OK); +} diff --git a/c/archive_read_support_format_ar.c b/c/archive_read_support_format_ar.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_format_ar.c @@ -0,0 +1,638 @@ +/*- + * Copyright (c) 2007 Kai Wang + * Copyright (c) 2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer + * in this position and unchanged. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_read_support_format_ar.c 201101 2009-12-28 03:06:27Z kientzle $"); + +#ifdef HAVE_SYS_STAT_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_LIMITS_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_read_private.h" + +struct ar { + int64_t entry_bytes_remaining; + /* unconsumed is purely to track data we've gotten from readahead, + * but haven't yet marked as consumed. Must be paired with + * entry_bytes_remaining usage/modification. + */ + size_t entry_bytes_unconsumed; + int64_t entry_offset; + int64_t entry_padding; + char *strtab; + size_t strtab_size; + char read_global_header; +}; + +/* + * Define structure of the "ar" header. + */ +#define AR_name_offset 0 +#define AR_name_size 16 +#define AR_date_offset 16 +#define AR_date_size 12 +#define AR_uid_offset 28 +#define AR_uid_size 6 +#define AR_gid_offset 34 +#define AR_gid_size 6 +#define AR_mode_offset 40 +#define AR_mode_size 8 +#define AR_size_offset 48 +#define AR_size_size 10 +#define AR_fmag_offset 58 +#define AR_fmag_size 2 + +static int archive_read_format_ar_bid(struct archive_read *a, int); +static int archive_read_format_ar_cleanup(struct archive_read *a); +static int archive_read_format_ar_read_data(struct archive_read *a, + const void **buff, size_t *size, int64_t *offset); +static int archive_read_format_ar_skip(struct archive_read *a); +static int archive_read_format_ar_read_header(struct archive_read *a, + struct archive_entry *e); +static uint64_t ar_atol8(const char *p, unsigned char_cnt); +static uint64_t ar_atol10(const char *p, unsigned char_cnt); +static int ar_parse_gnu_filename_table(struct archive_read *a); +static int ar_parse_common_header(struct ar *ar, struct archive_entry *, + const char *h); + +int +archive_read_support_format_ar(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct ar *ar; + int r; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_format_ar"); + + ar = (struct ar *)calloc(1, sizeof(*ar)); + if (ar == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate ar data"); + return (ARCHIVE_FATAL); + } + ar->strtab = NULL; + + r = __archive_read_register_format(a, + ar, + "ar", + archive_read_format_ar_bid, + NULL, + archive_read_format_ar_read_header, + archive_read_format_ar_read_data, + archive_read_format_ar_skip, + NULL, + archive_read_format_ar_cleanup, + NULL, + NULL); + + if (r != ARCHIVE_OK) { + free(ar); + return (r); + } + return (ARCHIVE_OK); +} + +static int +archive_read_format_ar_cleanup(struct archive_read *a) +{ + struct ar *ar; + + ar = (struct ar *)(a->format->data); + free(ar->strtab); + free(ar); + (a->format->data) = NULL; + return (ARCHIVE_OK); +} + +static int +archive_read_format_ar_bid(struct archive_read *a, int best_bid) +{ + const void *h; + + (void)best_bid; /* UNUSED */ + + /* + * Verify the 8-byte file signature. + * TODO: Do we need to check more than this? + */ + if ((h = __archive_read_ahead(a, 8, NULL)) == NULL) + return (-1); + if (memcmp(h, "!\n", 8) == 0) { + return (64); + } + return (-1); +} + +static int +_ar_read_header(struct archive_read *a, struct archive_entry *entry, + struct ar *ar, const char *h, size_t *unconsumed) +{ + char filename[AR_name_size + 1]; + uint64_t number; /* Used to hold parsed numbers before validation. */ + size_t bsd_name_length, entry_size; + char *p, *st; + const void *b; + int r; + + /* Verify the magic signature on the file header. */ + if (strncmp(h + AR_fmag_offset, "`\n", 2) != 0) { + archive_set_error(&a->archive, EINVAL, + "Incorrect file header signature"); + return (ARCHIVE_FATAL); + } + + /* Copy filename into work buffer. */ + strncpy(filename, h + AR_name_offset, AR_name_size); + filename[AR_name_size] = '\0'; + + /* + * Guess the format variant based on the filename. + */ + if (a->archive.archive_format == ARCHIVE_FORMAT_AR) { + /* We don't already know the variant, so let's guess. */ + /* + * Biggest clue is presence of '/': GNU starts special + * filenames with '/', appends '/' as terminator to + * non-special names, so anything with '/' should be + * GNU except for BSD long filenames. + */ + if (strncmp(filename, "#1/", 3) == 0) + a->archive.archive_format = ARCHIVE_FORMAT_AR_BSD; + else if (strchr(filename, '/') != NULL) + a->archive.archive_format = ARCHIVE_FORMAT_AR_GNU; + else if (strncmp(filename, "__.SYMDEF", 9) == 0) + a->archive.archive_format = ARCHIVE_FORMAT_AR_BSD; + /* + * XXX Do GNU/SVR4 'ar' programs ever omit trailing '/' + * if name exactly fills 16-byte field? If so, we + * can't assume entries without '/' are BSD. XXX + */ + } + + /* Update format name from the code. */ + if (a->archive.archive_format == ARCHIVE_FORMAT_AR_GNU) + a->archive.archive_format_name = "ar (GNU/SVR4)"; + else if (a->archive.archive_format == ARCHIVE_FORMAT_AR_BSD) + a->archive.archive_format_name = "ar (BSD)"; + else + a->archive.archive_format_name = "ar"; + + /* + * Remove trailing spaces from the filename. GNU and BSD + * variants both pad filename area out with spaces. + * This will only be wrong if GNU/SVR4 'ar' implementations + * omit trailing '/' for 16-char filenames and we have + * a 16-char filename that ends in ' '. + */ + p = filename + AR_name_size - 1; + while (p >= filename && *p == ' ') { + *p = '\0'; + p--; + } + + /* + * Remove trailing slash unless first character is '/'. + * (BSD entries never end in '/', so this will only trim + * GNU-format entries. GNU special entries start with '/' + * and are not terminated in '/', so we don't trim anything + * that starts with '/'.) + */ + if (filename[0] != '/' && p > filename && *p == '/') { + *p = '\0'; + } + + if (p < filename) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Found entry with empty filename"); + return (ARCHIVE_FATAL); + } + + /* + * '//' is the GNU filename table. + * Later entries can refer to names in this table. + */ + if (strcmp(filename, "//") == 0) { + /* This must come before any call to _read_ahead. */ + ar_parse_common_header(ar, entry, h); + archive_entry_copy_pathname(entry, filename); + archive_entry_set_filetype(entry, AE_IFREG); + /* Get the size of the filename table. */ + number = ar_atol10(h + AR_size_offset, AR_size_size); + if (number > SIZE_MAX || number > 1024 * 1024 * 1024) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Filename table too large"); + return (ARCHIVE_FATAL); + } + entry_size = (size_t)number; + if (entry_size == 0) { + archive_set_error(&a->archive, EINVAL, + "Invalid string table"); + return (ARCHIVE_FATAL); + } + if (ar->strtab != NULL) { + archive_set_error(&a->archive, EINVAL, + "More than one string tables exist"); + return (ARCHIVE_FATAL); + } + + /* Read the filename table into memory. */ + st = malloc(entry_size); + if (st == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate filename table buffer"); + return (ARCHIVE_FATAL); + } + ar->strtab = st; + ar->strtab_size = entry_size; + + if (*unconsumed) { + __archive_read_consume(a, *unconsumed); + *unconsumed = 0; + } + + if ((b = __archive_read_ahead(a, entry_size, NULL)) == NULL) + return (ARCHIVE_FATAL); + memcpy(st, b, entry_size); + __archive_read_consume(a, entry_size); + /* All contents are consumed. */ + ar->entry_bytes_remaining = 0; + archive_entry_set_size(entry, ar->entry_bytes_remaining); + + /* Parse the filename table. */ + return (ar_parse_gnu_filename_table(a)); + } + + /* + * GNU variant handles long filenames by storing / + * to indicate a name stored in the filename table. + * XXX TODO: Verify that it's all digits... Don't be fooled + * by "/9xyz" XXX + */ + if (filename[0] == '/' && filename[1] >= '0' && filename[1] <= '9') { + number = ar_atol10(h + AR_name_offset + 1, AR_name_size - 1); + /* + * If we can't look up the real name, warn and return + * the entry with the wrong name. + */ + if (ar->strtab == NULL || number >= ar->strtab_size) { + archive_set_error(&a->archive, EINVAL, + "Can't find long filename for GNU/SVR4 archive entry"); + archive_entry_copy_pathname(entry, filename); + /* Parse the time, owner, mode, size fields. */ + ar_parse_common_header(ar, entry, h); + return (ARCHIVE_FATAL); + } + + archive_entry_copy_pathname(entry, &ar->strtab[(size_t)number]); + /* Parse the time, owner, mode, size fields. */ + return (ar_parse_common_header(ar, entry, h)); + } + + /* + * BSD handles long filenames by storing "#1/" followed by the + * length of filename as a decimal number, then prepends the + * the filename to the file contents. + */ + if (strncmp(filename, "#1/", 3) == 0) { + /* Parse the time, owner, mode, size fields. */ + /* This must occur before _read_ahead is called again. */ + ar_parse_common_header(ar, entry, h); + + /* Parse the size of the name, adjust the file size. */ + number = ar_atol10(h + AR_name_offset + 3, AR_name_size - 3); + /* Sanity check the filename length: + * = Must be <= SIZE_MAX - 1 + * = Must be <= 1MB + * = Cannot be bigger than the entire entry + */ + if (number > SIZE_MAX - 1 + || number > 1024 * 1024 + || (int64_t)number > ar->entry_bytes_remaining) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Bad input file size"); + return (ARCHIVE_FATAL); + } + bsd_name_length = (size_t)number; + ar->entry_bytes_remaining -= bsd_name_length; + /* Adjust file size reported to client. */ + archive_entry_set_size(entry, ar->entry_bytes_remaining); + + if (*unconsumed) { + __archive_read_consume(a, *unconsumed); + *unconsumed = 0; + } + + /* Read the long name into memory. */ + if ((b = __archive_read_ahead(a, bsd_name_length, NULL)) == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Truncated input file"); + return (ARCHIVE_FATAL); + } + /* Store it in the entry. */ + p = (char *)malloc(bsd_name_length + 1); + if (p == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate fname buffer"); + return (ARCHIVE_FATAL); + } + strncpy(p, b, bsd_name_length); + p[bsd_name_length] = '\0'; + + __archive_read_consume(a, bsd_name_length); + + archive_entry_copy_pathname(entry, p); + free(p); + return (ARCHIVE_OK); + } + + /* + * "/" is the SVR4/GNU archive symbol table. + * "/SYM64/" is the SVR4/GNU 64-bit variant archive symbol table. + */ + if (strcmp(filename, "/") == 0 || strcmp(filename, "/SYM64/") == 0) { + archive_entry_copy_pathname(entry, filename); + /* Parse the time, owner, mode, size fields. */ + r = ar_parse_common_header(ar, entry, h); + /* Force the file type to a regular file. */ + archive_entry_set_filetype(entry, AE_IFREG); + return (r); + } + + /* + * "__.SYMDEF" is a BSD archive symbol table. + */ + if (strcmp(filename, "__.SYMDEF") == 0) { + archive_entry_copy_pathname(entry, filename); + /* Parse the time, owner, mode, size fields. */ + return (ar_parse_common_header(ar, entry, h)); + } + + /* + * Otherwise, this is a standard entry. The filename + * has already been trimmed as much as possible, based + * on our current knowledge of the format. + */ + archive_entry_copy_pathname(entry, filename); + return (ar_parse_common_header(ar, entry, h)); +} + +static int +archive_read_format_ar_read_header(struct archive_read *a, + struct archive_entry *entry) +{ + struct ar *ar = (struct ar*)(a->format->data); + size_t unconsumed; + const void *header_data; + int ret; + + if (!ar->read_global_header) { + /* + * We are now at the beginning of the archive, + * so we need first consume the ar global header. + */ + __archive_read_consume(a, 8); + ar->read_global_header = 1; + /* Set a default format code for now. */ + a->archive.archive_format = ARCHIVE_FORMAT_AR; + } + + /* Read the header for the next file entry. */ + if ((header_data = __archive_read_ahead(a, 60, NULL)) == NULL) + /* Broken header. */ + return (ARCHIVE_EOF); + + unconsumed = 60; + + ret = _ar_read_header(a, entry, ar, (const char *)header_data, &unconsumed); + + if (unconsumed) + __archive_read_consume(a, unconsumed); + + return ret; +} + + +static int +ar_parse_common_header(struct ar *ar, struct archive_entry *entry, + const char *h) +{ + uint64_t n; + + /* Copy remaining header */ + archive_entry_set_filetype(entry, AE_IFREG); + archive_entry_set_mtime(entry, + (time_t)ar_atol10(h + AR_date_offset, AR_date_size), 0L); + archive_entry_set_uid(entry, + (uid_t)ar_atol10(h + AR_uid_offset, AR_uid_size)); + archive_entry_set_gid(entry, + (gid_t)ar_atol10(h + AR_gid_offset, AR_gid_size)); + archive_entry_set_mode(entry, + (mode_t)ar_atol8(h + AR_mode_offset, AR_mode_size)); + n = ar_atol10(h + AR_size_offset, AR_size_size); + + ar->entry_offset = 0; + ar->entry_padding = n % 2; + archive_entry_set_size(entry, n); + ar->entry_bytes_remaining = n; + return (ARCHIVE_OK); +} + +static int +archive_read_format_ar_read_data(struct archive_read *a, + const void **buff, size_t *size, int64_t *offset) +{ + ssize_t bytes_read; + struct ar *ar; + + ar = (struct ar *)(a->format->data); + + if (ar->entry_bytes_unconsumed) { + __archive_read_consume(a, ar->entry_bytes_unconsumed); + ar->entry_bytes_unconsumed = 0; + } + + if (ar->entry_bytes_remaining > 0) { + *buff = __archive_read_ahead(a, 1, &bytes_read); + if (bytes_read == 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Truncated ar archive"); + return (ARCHIVE_FATAL); + } + if (bytes_read < 0) + return (ARCHIVE_FATAL); + if (bytes_read > ar->entry_bytes_remaining) + bytes_read = (ssize_t)ar->entry_bytes_remaining; + *size = bytes_read; + ar->entry_bytes_unconsumed = bytes_read; + *offset = ar->entry_offset; + ar->entry_offset += bytes_read; + ar->entry_bytes_remaining -= bytes_read; + return (ARCHIVE_OK); + } else { + int64_t skipped = __archive_read_consume(a, ar->entry_padding); + if (skipped >= 0) { + ar->entry_padding -= skipped; + } + if (ar->entry_padding) { + if (skipped >= 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Truncated ar archive- failed consuming padding"); + } + return (ARCHIVE_FATAL); + } + *buff = NULL; + *size = 0; + *offset = ar->entry_offset; + return (ARCHIVE_EOF); + } +} + +static int +archive_read_format_ar_skip(struct archive_read *a) +{ + int64_t bytes_skipped; + struct ar* ar; + + ar = (struct ar *)(a->format->data); + + bytes_skipped = __archive_read_consume(a, + ar->entry_bytes_remaining + ar->entry_padding + + ar->entry_bytes_unconsumed); + if (bytes_skipped < 0) + return (ARCHIVE_FATAL); + + ar->entry_bytes_remaining = 0; + ar->entry_bytes_unconsumed = 0; + ar->entry_padding = 0; + + return (ARCHIVE_OK); +} + +static int +ar_parse_gnu_filename_table(struct archive_read *a) +{ + struct ar *ar; + char *p; + size_t size; + + ar = (struct ar*)(a->format->data); + size = ar->strtab_size; + + for (p = ar->strtab; p < ar->strtab + size - 1; ++p) { + if (*p == '/') { + *p++ = '\0'; + if (*p != '\n') + goto bad_string_table; + *p = '\0'; + } + } + /* + * GNU ar always pads the table to an even size. + * The pad character is either '\n' or '`'. + */ + if (p != ar->strtab + size && *p != '\n' && *p != '`') + goto bad_string_table; + + /* Enforce zero termination. */ + ar->strtab[size - 1] = '\0'; + + return (ARCHIVE_OK); + +bad_string_table: + archive_set_error(&a->archive, EINVAL, + "Invalid string table"); + free(ar->strtab); + ar->strtab = NULL; + return (ARCHIVE_FATAL); +} + +static uint64_t +ar_atol8(const char *p, unsigned char_cnt) +{ + uint64_t l, limit, last_digit_limit; + unsigned int digit, base; + + base = 8; + limit = UINT64_MAX / base; + last_digit_limit = UINT64_MAX % base; + + while ((*p == ' ' || *p == '\t') && char_cnt-- > 0) + p++; + + l = 0; + digit = *p - '0'; + while (*p >= '0' && digit < base && char_cnt-- > 0) { + if (l>limit || (l == limit && digit > last_digit_limit)) { + l = UINT64_MAX; /* Truncate on overflow. */ + break; + } + l = (l * base) + digit; + digit = *++p - '0'; + } + return (l); +} + +static uint64_t +ar_atol10(const char *p, unsigned char_cnt) +{ + uint64_t l, limit, last_digit_limit; + unsigned int base, digit; + + base = 10; + limit = UINT64_MAX / base; + last_digit_limit = UINT64_MAX % base; + + while ((*p == ' ' || *p == '\t') && char_cnt-- > 0) + p++; + l = 0; + digit = *p - '0'; + while (*p >= '0' && digit < base && char_cnt-- > 0) { + if (l > limit || (l == limit && digit > last_digit_limit)) { + l = UINT64_MAX; /* Truncate on overflow. */ + break; + } + l = (l * base) + digit; + digit = *++p - '0'; + } + return (l); +} diff --git a/c/archive_read_support_format_by_code.c b/c/archive_read_support_format_by_code.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_format_by_code.c @@ -0,0 +1,77 @@ +/*- + * Copyright (c) 2003-2011 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#include "archive.h" +#include "archive_private.h" + +int +archive_read_support_format_by_code(struct archive *a, int format_code) +{ + archive_check_magic(a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_format_by_code"); + + switch (format_code & ARCHIVE_FORMAT_BASE_MASK) { + case ARCHIVE_FORMAT_7ZIP: + return archive_read_support_format_7zip(a); + break; + case ARCHIVE_FORMAT_AR: + return archive_read_support_format_ar(a); + break; + case ARCHIVE_FORMAT_CAB: + return archive_read_support_format_cab(a); + break; + case ARCHIVE_FORMAT_CPIO: + return archive_read_support_format_cpio(a); + break; + case ARCHIVE_FORMAT_ISO9660: + return archive_read_support_format_iso9660(a); + break; + case ARCHIVE_FORMAT_LHA: + return archive_read_support_format_lha(a); + break; + case ARCHIVE_FORMAT_MTREE: + return archive_read_support_format_mtree(a); + break; + case ARCHIVE_FORMAT_RAR: + return archive_read_support_format_rar(a); + break; + case ARCHIVE_FORMAT_RAR_V5: + return archive_read_support_format_rar5(a); + break; + case ARCHIVE_FORMAT_TAR: + return archive_read_support_format_tar(a); + break; + case ARCHIVE_FORMAT_XAR: + return archive_read_support_format_xar(a); + break; + case ARCHIVE_FORMAT_ZIP: + return archive_read_support_format_zip(a); + break; + } + return (ARCHIVE_FATAL); +} diff --git a/c/archive_read_support_format_cab.c b/c/archive_read_support_format_cab.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_format_cab.c @@ -0,0 +1,3227 @@ +/*- + * Copyright (c) 2010-2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_LIMITS_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_ZLIB_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_private.h" +#include "archive_read_private.h" +#include "archive_endian.h" + + +struct lzx_dec { + /* Decoding status. */ + int state; + + /* + * Window to see last decoded data, from 32KBi to 2MBi. + */ + int w_size; + int w_mask; + /* Window buffer, which is a loop buffer. */ + unsigned char *w_buff; + /* The insert position to the window. */ + int w_pos; + /* The position where we can copy decoded code from the window. */ + int copy_pos; + /* The length how many bytes we can copy decoded code from + * the window. */ + int copy_len; + /* Translation reversal for x86 processor CALL byte sequence(E8). + * This is used for LZX only. */ + uint32_t translation_size; + char translation; + char block_type; +#define VERBATIM_BLOCK 1 +#define ALIGNED_OFFSET_BLOCK 2 +#define UNCOMPRESSED_BLOCK 3 + size_t block_size; + size_t block_bytes_avail; + /* Repeated offset. */ + int r0, r1, r2; + unsigned char rbytes[4]; + int rbytes_avail; + int length_header; + int position_slot; + int offset_bits; + + struct lzx_pos_tbl { + int base; + int footer_bits; + } *pos_tbl; + /* + * Bit stream reader. + */ + struct lzx_br { +#define CACHE_TYPE uint64_t +#define CACHE_BITS (8 * sizeof(CACHE_TYPE)) + /* Cache buffer. */ + CACHE_TYPE cache_buffer; + /* Indicates how many bits avail in cache_buffer. */ + int cache_avail; + unsigned char odd; + char have_odd; + } br; + + /* + * Huffman coding. + */ + struct huffman { + int len_size; + int freq[17]; + unsigned char *bitlen; + + /* + * Use a index table. It's faster than searching a huffman + * coding tree, which is a binary tree. But a use of a large + * index table causes L1 cache read miss many times. + */ + int max_bits; + int tbl_bits; + int tree_used; + /* Direct access table. */ + uint16_t *tbl; + } at, lt, mt, pt; + + int loop; + int error; +}; + +static const int slots[] = { + 30, 32, 34, 36, 38, 42, 50, 66, 98, 162, 290 +}; +#define SLOT_BASE 15 +#define SLOT_MAX 21/*->25*/ + +struct lzx_stream { + const unsigned char *next_in; + int64_t avail_in; + int64_t total_in; + unsigned char *next_out; + int64_t avail_out; + int64_t total_out; + struct lzx_dec *ds; +}; + +/* + * Cabinet file definitions. + */ +/* CFHEADER offset */ +#define CFHEADER_signature 0 +#define CFHEADER_cbCabinet 8 +#define CFHEADER_coffFiles 16 +#define CFHEADER_versionMinor 24 +#define CFHEADER_versionMajor 25 +#define CFHEADER_cFolders 26 +#define CFHEADER_cFiles 28 +#define CFHEADER_flags 30 +#define CFHEADER_setID 32 +#define CFHEADER_iCabinet 34 +#define CFHEADER_cbCFHeader 36 +#define CFHEADER_cbCFFolder 38 +#define CFHEADER_cbCFData 39 + +/* CFFOLDER offset */ +#define CFFOLDER_coffCabStart 0 +#define CFFOLDER_cCFData 4 +#define CFFOLDER_typeCompress 6 +#define CFFOLDER_abReserve 8 + +/* CFFILE offset */ +#define CFFILE_cbFile 0 +#define CFFILE_uoffFolderStart 4 +#define CFFILE_iFolder 8 +#define CFFILE_date_time 10 +#define CFFILE_attribs 14 + +/* CFDATA offset */ +#define CFDATA_csum 0 +#define CFDATA_cbData 4 +#define CFDATA_cbUncomp 6 + +static const char * const compression_name[] = { + "NONE", + "MSZIP", + "Quantum", + "LZX", +}; + +struct cfdata { + /* Sum value of this CFDATA. */ + uint32_t sum; + uint16_t compressed_size; + uint16_t compressed_bytes_remaining; + uint16_t uncompressed_size; + uint16_t uncompressed_bytes_remaining; + /* To know how many bytes we have decompressed. */ + uint16_t uncompressed_avail; + /* Offset from the beginning of compressed data of this CFDATA */ + uint16_t read_offset; + int64_t unconsumed; + /* To keep memory image of this CFDATA to compute the sum. */ + size_t memimage_size; + unsigned char *memimage; + /* Result of calculation of sum. */ + uint32_t sum_calculated; + unsigned char sum_extra[4]; + int sum_extra_avail; + const void *sum_ptr; +}; + +struct cffolder { + uint32_t cfdata_offset_in_cab; + uint16_t cfdata_count; + uint16_t comptype; +#define COMPTYPE_NONE 0x0000 +#define COMPTYPE_MSZIP 0x0001 +#define COMPTYPE_QUANTUM 0x0002 +#define COMPTYPE_LZX 0x0003 + uint16_t compdata; + const char *compname; + /* At the time reading CFDATA */ + struct cfdata cfdata; + int cfdata_index; + /* Flags to mark progress of decompression. */ + char decompress_init; +}; + +struct cffile { + uint32_t uncompressed_size; + uint32_t offset; + time_t mtime; + uint16_t folder; +#define iFoldCONTINUED_FROM_PREV 0xFFFD +#define iFoldCONTINUED_TO_NEXT 0xFFFE +#define iFoldCONTINUED_PREV_AND_NEXT 0xFFFF + unsigned char attr; +#define ATTR_RDONLY 0x01 +#define ATTR_NAME_IS_UTF 0x80 + struct archive_string pathname; +}; + +struct cfheader { + /* Total bytes of all file size in a Cabinet. */ + uint32_t total_bytes; + uint32_t files_offset; + uint16_t folder_count; + uint16_t file_count; + uint16_t flags; +#define PREV_CABINET 0x0001 +#define NEXT_CABINET 0x0002 +#define RESERVE_PRESENT 0x0004 + uint16_t setid; + uint16_t cabinet; + /* Version number. */ + unsigned char major; + unsigned char minor; + unsigned char cffolder; + unsigned char cfdata; + /* All folders in a cabinet. */ + struct cffolder *folder_array; + /* All files in a cabinet. */ + struct cffile *file_array; + int file_index; +}; + +struct cab { + /* entry_bytes_remaining is the number of bytes we expect. */ + int64_t entry_offset; + int64_t entry_bytes_remaining; + int64_t entry_unconsumed; + int64_t entry_compressed_bytes_read; + int64_t entry_uncompressed_bytes_read; + struct cffolder *entry_cffolder; + struct cffile *entry_cffile; + struct cfdata *entry_cfdata; + + /* Offset from beginning of a cabinet file. */ + int64_t cab_offset; + struct cfheader cfheader; + struct archive_wstring ws; + + /* Flag to mark progress that an archive was read their first header.*/ + char found_header; + char end_of_archive; + char end_of_entry; + char end_of_entry_cleanup; + char read_data_invoked; + int64_t bytes_skipped; + + unsigned char *uncompressed_buffer; + size_t uncompressed_buffer_size; + + int init_default_conversion; + struct archive_string_conv *sconv; + struct archive_string_conv *sconv_default; + struct archive_string_conv *sconv_utf8; + char format_name[64]; + +#ifdef HAVE_ZLIB_H + z_stream stream; + char stream_valid; +#endif + struct lzx_stream xstrm; +}; + +static int archive_read_format_cab_bid(struct archive_read *, int); +static int archive_read_format_cab_options(struct archive_read *, + const char *, const char *); +static int archive_read_format_cab_read_header(struct archive_read *, + struct archive_entry *); +static int archive_read_format_cab_read_data(struct archive_read *, + const void **, size_t *, int64_t *); +static int archive_read_format_cab_read_data_skip(struct archive_read *); +static int archive_read_format_cab_cleanup(struct archive_read *); + +static int cab_skip_sfx(struct archive_read *); +static time_t cab_dos_time(const unsigned char *); +static int cab_read_data(struct archive_read *, const void **, + size_t *, int64_t *); +static int cab_read_header(struct archive_read *); +static uint32_t cab_checksum_cfdata_4(const void *, size_t bytes, uint32_t); +static uint32_t cab_checksum_cfdata(const void *, size_t bytes, uint32_t); +static void cab_checksum_update(struct archive_read *, size_t); +static int cab_checksum_finish(struct archive_read *); +static int cab_next_cfdata(struct archive_read *); +static const void *cab_read_ahead_cfdata(struct archive_read *, ssize_t *); +static const void *cab_read_ahead_cfdata_none(struct archive_read *, ssize_t *); +static const void *cab_read_ahead_cfdata_deflate(struct archive_read *, + ssize_t *); +static const void *cab_read_ahead_cfdata_lzx(struct archive_read *, + ssize_t *); +static int64_t cab_consume_cfdata(struct archive_read *, int64_t); +static int64_t cab_minimum_consume_cfdata(struct archive_read *, int64_t); +static int lzx_decode_init(struct lzx_stream *, int); +static int lzx_read_blocks(struct lzx_stream *, int); +static int lzx_decode_blocks(struct lzx_stream *, int); +static void lzx_decode_free(struct lzx_stream *); +static void lzx_translation(struct lzx_stream *, void *, size_t, uint32_t); +static void lzx_cleanup_bitstream(struct lzx_stream *); +static int lzx_decode(struct lzx_stream *, int); +static int lzx_read_pre_tree(struct lzx_stream *); +static int lzx_read_bitlen(struct lzx_stream *, struct huffman *, int); +static int lzx_huffman_init(struct huffman *, size_t, int); +static void lzx_huffman_free(struct huffman *); +static int lzx_make_huffman_table(struct huffman *); +static inline int lzx_decode_huffman(struct huffman *, unsigned); + + +int +archive_read_support_format_cab(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct cab *cab; + int r; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_format_cab"); + + cab = (struct cab *)calloc(1, sizeof(*cab)); + if (cab == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate CAB data"); + return (ARCHIVE_FATAL); + } + archive_string_init(&cab->ws); + archive_wstring_ensure(&cab->ws, 256); + + r = __archive_read_register_format(a, + cab, + "cab", + archive_read_format_cab_bid, + archive_read_format_cab_options, + archive_read_format_cab_read_header, + archive_read_format_cab_read_data, + archive_read_format_cab_read_data_skip, + NULL, + archive_read_format_cab_cleanup, + NULL, + NULL); + + if (r != ARCHIVE_OK) + free(cab); + return (ARCHIVE_OK); +} + +static int +find_cab_magic(const char *p) +{ + switch (p[4]) { + case 0: + /* + * Note: Self-Extraction program has 'MSCF' string in their + * program. If we were finding 'MSCF' string only, we got + * wrong place for Cabinet header, thus, we have to check + * following four bytes which are reserved and must be set + * to zero. + */ + if (memcmp(p, "MSCF\0\0\0\0", 8) == 0) + return 0; + return 5; + case 'F': return 1; + case 'C': return 2; + case 'S': return 3; + case 'M': return 4; + default: return 5; + } +} + +static int +archive_read_format_cab_bid(struct archive_read *a, int best_bid) +{ + const char *p; + ssize_t bytes_avail, offset, window; + + /* If there's already a better bid than we can ever + make, don't bother testing. */ + if (best_bid > 64) + return (-1); + + if ((p = __archive_read_ahead(a, 8, NULL)) == NULL) + return (-1); + + if (memcmp(p, "MSCF\0\0\0\0", 8) == 0) + return (64); + + /* + * Attempt to handle self-extracting archives + * by noting a PE header and searching forward + * up to 128k for a 'MSCF' marker. + */ + if (p[0] == 'M' && p[1] == 'Z') { + offset = 0; + window = 4096; + while (offset < (1024 * 128)) { + const char *h = __archive_read_ahead(a, offset + window, + &bytes_avail); + if (h == NULL) { + /* Remaining bytes are less than window. */ + window >>= 1; + if (window < 128) + return (0); + continue; + } + p = h + offset; + while (p + 8 < h + bytes_avail) { + int next; + if ((next = find_cab_magic(p)) == 0) + return (64); + p += next; + } + offset = p - h; + } + } + return (0); +} + +static int +archive_read_format_cab_options(struct archive_read *a, + const char *key, const char *val) +{ + struct cab *cab; + int ret = ARCHIVE_FAILED; + + cab = (struct cab *)(a->format->data); + if (strcmp(key, "hdrcharset") == 0) { + if (val == NULL || val[0] == 0) + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "cab: hdrcharset option needs a character-set name"); + else { + cab->sconv = archive_string_conversion_from_charset( + &a->archive, val, 0); + if (cab->sconv != NULL) + ret = ARCHIVE_OK; + else + ret = ARCHIVE_FATAL; + } + return (ret); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +static int +cab_skip_sfx(struct archive_read *a) +{ + const char *p, *q; + size_t skip; + ssize_t bytes, window; + + window = 4096; + for (;;) { + const char *h = __archive_read_ahead(a, window, &bytes); + if (h == NULL) { + /* Remaining size are less than window. */ + window >>= 1; + if (window < 128) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Couldn't find out CAB header"); + return (ARCHIVE_FATAL); + } + continue; + } + p = h; + q = p + bytes; + + /* + * Scan ahead until we find something that looks + * like the cab header. + */ + while (p + 8 < q) { + int next; + if ((next = find_cab_magic(p)) == 0) { + skip = p - h; + __archive_read_consume(a, skip); + return (ARCHIVE_OK); + } + p += next; + } + skip = p - h; + __archive_read_consume(a, skip); + } +} + +static int +truncated_error(struct archive_read *a) +{ + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated CAB header"); + return (ARCHIVE_FATAL); +} + +static ssize_t +cab_strnlen(const unsigned char *p, size_t maxlen) +{ + size_t i; + + for (i = 0; i <= maxlen; i++) { + if (p[i] == 0) + break; + } + if (i > maxlen) + return (-1);/* invalid */ + return ((ssize_t)i); +} + +/* Read bytes as much as remaining. */ +static const void * +cab_read_ahead_remaining(struct archive_read *a, size_t min, ssize_t *avail) +{ + const void *p; + + while (min > 0) { + p = __archive_read_ahead(a, min, avail); + if (p != NULL) + return (p); + min--; + } + return (NULL); +} + +/* Convert a path separator '\' -> '/' */ +static int +cab_convert_path_separator_1(struct archive_string *fn, unsigned char attr) +{ + size_t i; + int mb; + + /* Easy check if we have '\' in multi-byte string. */ + mb = 0; + for (i = 0; i < archive_strlen(fn); i++) { + if (fn->s[i] == '\\') { + if (mb) { + /* This may be second byte of multi-byte + * character. */ + break; + } + fn->s[i] = '/'; + mb = 0; + } else if ((fn->s[i] & 0x80) && !(attr & ATTR_NAME_IS_UTF)) + mb = 1; + else + mb = 0; + } + if (i == archive_strlen(fn)) + return (0); + return (-1); +} + +/* + * Replace a character '\' with '/' in wide character. + */ +static void +cab_convert_path_separator_2(struct cab *cab, struct archive_entry *entry) +{ + const wchar_t *wp; + size_t i; + + /* If a conversion to wide character failed, force the replacement. */ + if ((wp = archive_entry_pathname_w(entry)) != NULL) { + archive_wstrcpy(&(cab->ws), wp); + for (i = 0; i < archive_strlen(&(cab->ws)); i++) { + if (cab->ws.s[i] == L'\\') + cab->ws.s[i] = L'/'; + } + archive_entry_copy_pathname_w(entry, cab->ws.s); + } +} + +/* + * Read CFHEADER, CFFOLDER and CFFILE. + */ +static int +cab_read_header(struct archive_read *a) +{ + const unsigned char *p; + struct cab *cab; + struct cfheader *hd; + size_t bytes, used; + ssize_t len; + int64_t skip; + int err, i; + int cur_folder, prev_folder; + uint32_t offset32; + + a->archive.archive_format = ARCHIVE_FORMAT_CAB; + if (a->archive.archive_format_name == NULL) + a->archive.archive_format_name = "CAB"; + + if ((p = __archive_read_ahead(a, 42, NULL)) == NULL) + return (truncated_error(a)); + + cab = (struct cab *)(a->format->data); + if (cab->found_header == 0 && + p[0] == 'M' && p[1] == 'Z') { + /* This is an executable? Must be self-extracting... */ + err = cab_skip_sfx(a); + if (err < ARCHIVE_WARN) + return (err); + + /* Re-read header after processing the SFX. */ + if ((p = __archive_read_ahead(a, 42, NULL)) == NULL) + return (truncated_error(a)); + } + + cab->cab_offset = 0; + /* + * Read CFHEADER. + */ + hd = &cab->cfheader; + if (p[CFHEADER_signature+0] != 'M' || p[CFHEADER_signature+1] != 'S' || + p[CFHEADER_signature+2] != 'C' || p[CFHEADER_signature+3] != 'F') { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Couldn't find out CAB header"); + return (ARCHIVE_FATAL); + } + hd->total_bytes = archive_le32dec(p + CFHEADER_cbCabinet); + hd->files_offset = archive_le32dec(p + CFHEADER_coffFiles); + hd->minor = p[CFHEADER_versionMinor]; + hd->major = p[CFHEADER_versionMajor]; + hd->folder_count = archive_le16dec(p + CFHEADER_cFolders); + if (hd->folder_count == 0) + goto invalid; + hd->file_count = archive_le16dec(p + CFHEADER_cFiles); + if (hd->file_count == 0) + goto invalid; + hd->flags = archive_le16dec(p + CFHEADER_flags); + hd->setid = archive_le16dec(p + CFHEADER_setID); + hd->cabinet = archive_le16dec(p + CFHEADER_iCabinet); + used = CFHEADER_iCabinet + 2; + if (hd->flags & RESERVE_PRESENT) { + uint16_t cfheader; + cfheader = archive_le16dec(p + CFHEADER_cbCFHeader); + if (cfheader > 60000U) + goto invalid; + hd->cffolder = p[CFHEADER_cbCFFolder]; + hd->cfdata = p[CFHEADER_cbCFData]; + used += 4;/* cbCFHeader, cbCFFolder and cbCFData */ + used += cfheader;/* abReserve */ + } else + hd->cffolder = 0;/* Avoid compiling warning. */ + if (hd->flags & PREV_CABINET) { + /* How many bytes are used for szCabinetPrev. */ + if ((p = __archive_read_ahead(a, used+256, NULL)) == NULL) + return (truncated_error(a)); + if ((len = cab_strnlen(p + used, 255)) <= 0) + goto invalid; + used += len + 1; + /* How many bytes are used for szDiskPrev. */ + if ((p = __archive_read_ahead(a, used+256, NULL)) == NULL) + return (truncated_error(a)); + if ((len = cab_strnlen(p + used, 255)) <= 0) + goto invalid; + used += len + 1; + } + if (hd->flags & NEXT_CABINET) { + /* How many bytes are used for szCabinetNext. */ + if ((p = __archive_read_ahead(a, used+256, NULL)) == NULL) + return (truncated_error(a)); + if ((len = cab_strnlen(p + used, 255)) <= 0) + goto invalid; + used += len + 1; + /* How many bytes are used for szDiskNext. */ + if ((p = __archive_read_ahead(a, used+256, NULL)) == NULL) + return (truncated_error(a)); + if ((len = cab_strnlen(p + used, 255)) <= 0) + goto invalid; + used += len + 1; + } + __archive_read_consume(a, used); + cab->cab_offset += used; + used = 0; + + /* + * Read CFFOLDER. + */ + hd->folder_array = (struct cffolder *)calloc( + hd->folder_count, sizeof(struct cffolder)); + if (hd->folder_array == NULL) + goto nomem; + + bytes = 8; + if (hd->flags & RESERVE_PRESENT) + bytes += hd->cffolder; + bytes *= hd->folder_count; + if ((p = __archive_read_ahead(a, bytes, NULL)) == NULL) + return (truncated_error(a)); + offset32 = 0; + for (i = 0; i < hd->folder_count; i++) { + struct cffolder *folder = &(hd->folder_array[i]); + folder->cfdata_offset_in_cab = + archive_le32dec(p + CFFOLDER_coffCabStart); + folder->cfdata_count = archive_le16dec(p+CFFOLDER_cCFData); + folder->comptype = + archive_le16dec(p+CFFOLDER_typeCompress) & 0x0F; + folder->compdata = + archive_le16dec(p+CFFOLDER_typeCompress) >> 8; + /* Get a compression name. */ + if (folder->comptype < + sizeof(compression_name) / sizeof(compression_name[0])) + folder->compname = compression_name[folder->comptype]; + else + folder->compname = "UNKNOWN"; + p += 8; + used += 8; + if (hd->flags & RESERVE_PRESENT) { + p += hd->cffolder;/* abReserve */ + used += hd->cffolder; + } + /* + * Sanity check if each data is acceptable. + */ + if (offset32 >= folder->cfdata_offset_in_cab) + goto invalid; + offset32 = folder->cfdata_offset_in_cab; + + /* Set a request to initialize zlib for the CFDATA of + * this folder. */ + folder->decompress_init = 0; + } + __archive_read_consume(a, used); + cab->cab_offset += used; + + /* + * Read CFFILE. + */ + /* Seek read pointer to the offset of CFFILE if needed. */ + skip = (int64_t)hd->files_offset - cab->cab_offset; + if (skip < 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Invalid offset of CFFILE %jd < %jd", + (intmax_t)hd->files_offset, (intmax_t)cab->cab_offset); + return (ARCHIVE_FATAL); + } + if (skip) { + __archive_read_consume(a, skip); + cab->cab_offset += skip; + } + /* Allocate memory for CFDATA */ + hd->file_array = (struct cffile *)calloc( + hd->file_count, sizeof(struct cffile)); + if (hd->file_array == NULL) + goto nomem; + + prev_folder = -1; + for (i = 0; i < hd->file_count; i++) { + struct cffile *file = &(hd->file_array[i]); + ssize_t avail; + + if ((p = __archive_read_ahead(a, 16, NULL)) == NULL) + return (truncated_error(a)); + file->uncompressed_size = archive_le32dec(p + CFFILE_cbFile); + file->offset = archive_le32dec(p + CFFILE_uoffFolderStart); + file->folder = archive_le16dec(p + CFFILE_iFolder); + file->mtime = cab_dos_time(p + CFFILE_date_time); + file->attr = (uint8_t)archive_le16dec(p + CFFILE_attribs); + __archive_read_consume(a, 16); + + cab->cab_offset += 16; + if ((p = cab_read_ahead_remaining(a, 256, &avail)) == NULL) + return (truncated_error(a)); + if ((len = cab_strnlen(p, avail-1)) <= 0) + goto invalid; + + /* Copy a pathname. */ + archive_string_init(&(file->pathname)); + archive_strncpy(&(file->pathname), p, len); + __archive_read_consume(a, len + 1); + cab->cab_offset += len + 1; + + /* + * Sanity check if each data is acceptable. + */ + if (file->uncompressed_size > 0x7FFF8000) + goto invalid;/* Too large */ + if ((int64_t)file->offset + (int64_t)file->uncompressed_size + > ARCHIVE_LITERAL_LL(0x7FFF8000)) + goto invalid;/* Too large */ + switch (file->folder) { + case iFoldCONTINUED_TO_NEXT: + /* This must be last file in a folder. */ + if (i != hd->file_count -1) + goto invalid; + cur_folder = hd->folder_count -1; + break; + case iFoldCONTINUED_PREV_AND_NEXT: + /* This must be only one file in a folder. */ + if (hd->file_count != 1) + goto invalid; + /* FALL THROUGH */ + case iFoldCONTINUED_FROM_PREV: + /* This must be first file in a folder. */ + if (i != 0) + goto invalid; + prev_folder = cur_folder = 0; + offset32 = file->offset; + break; + default: + if (file->folder >= hd->folder_count) + goto invalid; + cur_folder = file->folder; + break; + } + /* Dot not back track. */ + if (cur_folder < prev_folder) + goto invalid; + if (cur_folder != prev_folder) + offset32 = 0; + prev_folder = cur_folder; + + /* Make sure there are not any blanks from last file + * contents. */ + if (offset32 != file->offset) + goto invalid; + offset32 += file->uncompressed_size; + + /* CFDATA is available for file contents. */ + if (file->uncompressed_size > 0 && + hd->folder_array[cur_folder].cfdata_count == 0) + goto invalid; + } + + if (hd->cabinet != 0 || hd->flags & (PREV_CABINET | NEXT_CABINET)) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Multivolume cabinet file is unsupported"); + return (ARCHIVE_WARN); + } + return (ARCHIVE_OK); +invalid: + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid CAB header"); + return (ARCHIVE_FATAL); +nomem: + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for CAB data"); + return (ARCHIVE_FATAL); +} + +static int +archive_read_format_cab_read_header(struct archive_read *a, + struct archive_entry *entry) +{ + struct cab *cab; + struct cfheader *hd; + struct cffolder *prev_folder; + struct cffile *file; + struct archive_string_conv *sconv; + int err = ARCHIVE_OK, r; + + cab = (struct cab *)(a->format->data); + if (cab->found_header == 0) { + err = cab_read_header(a); + if (err < ARCHIVE_WARN) + return (err); + /* We've found the header. */ + cab->found_header = 1; + } + hd = &cab->cfheader; + + if (hd->file_index >= hd->file_count) { + cab->end_of_archive = 1; + return (ARCHIVE_EOF); + } + file = &hd->file_array[hd->file_index++]; + + cab->end_of_entry = 0; + cab->end_of_entry_cleanup = 0; + cab->entry_compressed_bytes_read = 0; + cab->entry_uncompressed_bytes_read = 0; + cab->entry_unconsumed = 0; + cab->entry_cffile = file; + + /* + * Choose a proper folder. + */ + prev_folder = cab->entry_cffolder; + switch (file->folder) { + case iFoldCONTINUED_FROM_PREV: + case iFoldCONTINUED_PREV_AND_NEXT: + cab->entry_cffolder = &hd->folder_array[0]; + break; + case iFoldCONTINUED_TO_NEXT: + cab->entry_cffolder = &hd->folder_array[hd->folder_count-1]; + break; + default: + cab->entry_cffolder = &hd->folder_array[file->folder]; + break; + } + /* If a cffolder of this file is changed, reset a cfdata to read + * file contents from next cfdata. */ + if (prev_folder != cab->entry_cffolder) + cab->entry_cfdata = NULL; + + /* If a pathname is UTF-8, prepare a string conversion object + * for UTF-8 and use it. */ + if (file->attr & ATTR_NAME_IS_UTF) { + if (cab->sconv_utf8 == NULL) { + cab->sconv_utf8 = + archive_string_conversion_from_charset( + &(a->archive), "UTF-8", 1); + if (cab->sconv_utf8 == NULL) + return (ARCHIVE_FATAL); + } + sconv = cab->sconv_utf8; + } else if (cab->sconv != NULL) { + /* Choose the conversion specified by the option. */ + sconv = cab->sconv; + } else { + /* Choose the default conversion. */ + if (!cab->init_default_conversion) { + cab->sconv_default = + archive_string_default_conversion_for_read( + &(a->archive)); + cab->init_default_conversion = 1; + } + sconv = cab->sconv_default; + } + + /* + * Set a default value and common data + */ + r = cab_convert_path_separator_1(&(file->pathname), file->attr); + if (archive_entry_copy_pathname_l(entry, file->pathname.s, + archive_strlen(&(file->pathname)), sconv) != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Pathname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Pathname cannot be converted " + "from %s to current locale.", + archive_string_conversion_charset_name(sconv)); + err = ARCHIVE_WARN; + } + if (r < 0) { + /* Convert a path separator '\' -> '/' */ + cab_convert_path_separator_2(cab, entry); + } + + archive_entry_set_size(entry, file->uncompressed_size); + if (file->attr & ATTR_RDONLY) + archive_entry_set_mode(entry, AE_IFREG | 0555); + else + archive_entry_set_mode(entry, AE_IFREG | 0666); + archive_entry_set_mtime(entry, file->mtime, 0); + + cab->entry_bytes_remaining = file->uncompressed_size; + cab->entry_offset = 0; + /* We don't need compress data. */ + if (file->uncompressed_size == 0) + cab->end_of_entry_cleanup = cab->end_of_entry = 1; + + /* Set up a more descriptive format name. */ + sprintf(cab->format_name, "CAB %d.%d (%s)", + hd->major, hd->minor, cab->entry_cffolder->compname); + a->archive.archive_format_name = cab->format_name; + + return (err); +} + +static int +archive_read_format_cab_read_data(struct archive_read *a, + const void **buff, size_t *size, int64_t *offset) +{ + struct cab *cab = (struct cab *)(a->format->data); + int r; + + switch (cab->entry_cffile->folder) { + case iFoldCONTINUED_FROM_PREV: + case iFoldCONTINUED_TO_NEXT: + case iFoldCONTINUED_PREV_AND_NEXT: + *buff = NULL; + *size = 0; + *offset = 0; + archive_clear_error(&a->archive); + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Cannot restore this file split in multivolume."); + return (ARCHIVE_FAILED); + default: + break; + } + if (cab->read_data_invoked == 0) { + if (cab->bytes_skipped) { + if (cab->entry_cfdata == NULL) { + r = cab_next_cfdata(a); + if (r < 0) + return (r); + } + if (cab_consume_cfdata(a, cab->bytes_skipped) < 0) + return (ARCHIVE_FATAL); + cab->bytes_skipped = 0; + } + cab->read_data_invoked = 1; + } + if (cab->entry_unconsumed) { + /* Consume as much as the compressor actually used. */ + r = (int)cab_consume_cfdata(a, cab->entry_unconsumed); + cab->entry_unconsumed = 0; + if (r < 0) + return (r); + } + if (cab->end_of_archive || cab->end_of_entry) { + if (!cab->end_of_entry_cleanup) { + /* End-of-entry cleanup done. */ + cab->end_of_entry_cleanup = 1; + } + *offset = cab->entry_offset; + *size = 0; + *buff = NULL; + return (ARCHIVE_EOF); + } + + return (cab_read_data(a, buff, size, offset)); +} + +static uint32_t +cab_checksum_cfdata_4(const void *p, size_t bytes, uint32_t seed) +{ + const unsigned char *b; + unsigned u32num; + uint32_t sum; + + u32num = (unsigned)bytes / 4; + sum = seed; + b = p; + for (;u32num > 0; --u32num) { + sum ^= archive_le32dec(b); + b += 4; + } + return (sum); +} + +static uint32_t +cab_checksum_cfdata(const void *p, size_t bytes, uint32_t seed) +{ + const unsigned char *b; + uint32_t sum; + uint32_t t; + + sum = cab_checksum_cfdata_4(p, bytes, seed); + b = p; + b += bytes & ~3; + t = 0; + switch (bytes & 3) { + case 3: + t |= ((uint32_t)(*b++)) << 16; + /* FALL THROUGH */ + case 2: + t |= ((uint32_t)(*b++)) << 8; + /* FALL THROUGH */ + case 1: + t |= *b; + /* FALL THROUGH */ + default: + break; + } + sum ^= t; + + return (sum); +} + +static void +cab_checksum_update(struct archive_read *a, size_t bytes) +{ + struct cab *cab = (struct cab *)(a->format->data); + struct cfdata *cfdata = cab->entry_cfdata; + const unsigned char *p; + size_t sumbytes; + + if (cfdata->sum == 0 || cfdata->sum_ptr == NULL) + return; + /* + * Calculate the sum of this CFDATA. + * Make sure CFDATA must be calculated in four bytes. + */ + p = cfdata->sum_ptr; + sumbytes = bytes; + if (cfdata->sum_extra_avail) { + while (cfdata->sum_extra_avail < 4 && sumbytes > 0) { + cfdata->sum_extra[ + cfdata->sum_extra_avail++] = *p++; + sumbytes--; + } + if (cfdata->sum_extra_avail == 4) { + cfdata->sum_calculated = cab_checksum_cfdata_4( + cfdata->sum_extra, 4, cfdata->sum_calculated); + cfdata->sum_extra_avail = 0; + } + } + if (sumbytes) { + int odd = sumbytes & 3; + if (sumbytes - odd > 0) + cfdata->sum_calculated = cab_checksum_cfdata_4( + p, sumbytes - odd, cfdata->sum_calculated); + if (odd) + memcpy(cfdata->sum_extra, p + sumbytes - odd, odd); + cfdata->sum_extra_avail = odd; + } + cfdata->sum_ptr = NULL; +} + +static int +cab_checksum_finish(struct archive_read *a) +{ + struct cab *cab = (struct cab *)(a->format->data); + struct cfdata *cfdata = cab->entry_cfdata; + int l; + + /* Do not need to compute a sum. */ + if (cfdata->sum == 0) + return (ARCHIVE_OK); + + /* + * Calculate the sum of remaining CFDATA. + */ + if (cfdata->sum_extra_avail) { + cfdata->sum_calculated = + cab_checksum_cfdata(cfdata->sum_extra, + cfdata->sum_extra_avail, cfdata->sum_calculated); + cfdata->sum_extra_avail = 0; + } + + l = 4; + if (cab->cfheader.flags & RESERVE_PRESENT) + l += cab->cfheader.cfdata; + cfdata->sum_calculated = cab_checksum_cfdata( + cfdata->memimage + CFDATA_cbData, l, cfdata->sum_calculated); + if (cfdata->sum_calculated != cfdata->sum) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Checksum error CFDATA[%d] %x:%x in %d bytes", + cab->entry_cffolder->cfdata_index -1, + cfdata->sum, cfdata->sum_calculated, + cfdata->compressed_size); + return (ARCHIVE_FAILED); + } + return (ARCHIVE_OK); +} + +/* + * Read CFDATA if needed. + */ +static int +cab_next_cfdata(struct archive_read *a) +{ + struct cab *cab = (struct cab *)(a->format->data); + struct cfdata *cfdata = cab->entry_cfdata; + + /* There are remaining bytes in current CFDATA, use it first. */ + if (cfdata != NULL && cfdata->uncompressed_bytes_remaining > 0) + return (ARCHIVE_OK); + + if (cfdata == NULL) { + int64_t skip; + + cab->entry_cffolder->cfdata_index = 0; + + /* Seek read pointer to the offset of CFDATA if needed. */ + skip = cab->entry_cffolder->cfdata_offset_in_cab + - cab->cab_offset; + if (skip < 0) { + int folder_index; + switch (cab->entry_cffile->folder) { + case iFoldCONTINUED_FROM_PREV: + case iFoldCONTINUED_PREV_AND_NEXT: + folder_index = 0; + break; + case iFoldCONTINUED_TO_NEXT: + folder_index = cab->cfheader.folder_count-1; + break; + default: + folder_index = cab->entry_cffile->folder; + break; + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Invalid offset of CFDATA in folder(%d) %jd < %jd", + folder_index, + (intmax_t)cab->entry_cffolder->cfdata_offset_in_cab, + (intmax_t)cab->cab_offset); + return (ARCHIVE_FATAL); + } + if (skip > 0) { + if (__archive_read_consume(a, skip) < 0) + return (ARCHIVE_FATAL); + cab->cab_offset = + cab->entry_cffolder->cfdata_offset_in_cab; + } + } + + /* + * Read a CFDATA. + */ + if (cab->entry_cffolder->cfdata_index < + cab->entry_cffolder->cfdata_count) { + const unsigned char *p; + int l; + + cfdata = &(cab->entry_cffolder->cfdata); + cab->entry_cffolder->cfdata_index++; + cab->entry_cfdata = cfdata; + cfdata->sum_calculated = 0; + cfdata->sum_extra_avail = 0; + cfdata->sum_ptr = NULL; + l = 8; + if (cab->cfheader.flags & RESERVE_PRESENT) + l += cab->cfheader.cfdata; + if ((p = __archive_read_ahead(a, l, NULL)) == NULL) + return (truncated_error(a)); + cfdata->sum = archive_le32dec(p + CFDATA_csum); + cfdata->compressed_size = archive_le16dec(p + CFDATA_cbData); + cfdata->compressed_bytes_remaining = cfdata->compressed_size; + cfdata->uncompressed_size = + archive_le16dec(p + CFDATA_cbUncomp); + cfdata->uncompressed_bytes_remaining = + cfdata->uncompressed_size; + cfdata->uncompressed_avail = 0; + cfdata->read_offset = 0; + cfdata->unconsumed = 0; + + /* + * Sanity check if data size is acceptable. + */ + if (cfdata->compressed_size == 0 || + cfdata->compressed_size > (0x8000+6144)) + goto invalid; + if (cfdata->uncompressed_size > 0x8000) + goto invalid; + if (cfdata->uncompressed_size == 0) { + switch (cab->entry_cffile->folder) { + case iFoldCONTINUED_PREV_AND_NEXT: + case iFoldCONTINUED_TO_NEXT: + break; + case iFoldCONTINUED_FROM_PREV: + default: + goto invalid; + } + } + /* If CFDATA is not last in a folder, an uncompressed + * size must be 0x8000(32KBi) */ + if ((cab->entry_cffolder->cfdata_index < + cab->entry_cffolder->cfdata_count) && + cfdata->uncompressed_size != 0x8000) + goto invalid; + + /* A compressed data size and an uncompressed data size must + * be the same in no compression mode. */ + if (cab->entry_cffolder->comptype == COMPTYPE_NONE && + cfdata->compressed_size != cfdata->uncompressed_size) + goto invalid; + + /* + * Save CFDATA image for sum check. + */ + if (cfdata->memimage_size < (size_t)l) { + free(cfdata->memimage); + cfdata->memimage = malloc(l); + if (cfdata->memimage == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for CAB data"); + return (ARCHIVE_FATAL); + } + cfdata->memimage_size = l; + } + memcpy(cfdata->memimage, p, l); + + /* Consume bytes as much as we used. */ + __archive_read_consume(a, l); + cab->cab_offset += l; + } else if (cab->entry_cffolder->cfdata_count > 0) { + /* Run out of all CFDATA in a folder. */ + cfdata->compressed_size = 0; + cfdata->uncompressed_size = 0; + cfdata->compressed_bytes_remaining = 0; + cfdata->uncompressed_bytes_remaining = 0; + } else { + /* Current folder does not have any CFDATA. */ + cfdata = &(cab->entry_cffolder->cfdata); + cab->entry_cfdata = cfdata; + memset(cfdata, 0, sizeof(*cfdata)); + } + return (ARCHIVE_OK); +invalid: + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid CFDATA"); + return (ARCHIVE_FATAL); +} + +/* + * Read ahead CFDATA. + */ +static const void * +cab_read_ahead_cfdata(struct archive_read *a, ssize_t *avail) +{ + struct cab *cab = (struct cab *)(a->format->data); + int err; + + err = cab_next_cfdata(a); + if (err < ARCHIVE_OK) { + *avail = err; + return (NULL); + } + + switch (cab->entry_cffolder->comptype) { + case COMPTYPE_NONE: + return (cab_read_ahead_cfdata_none(a, avail)); + case COMPTYPE_MSZIP: + return (cab_read_ahead_cfdata_deflate(a, avail)); + case COMPTYPE_LZX: + return (cab_read_ahead_cfdata_lzx(a, avail)); + default: /* Unsupported compression. */ + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Unsupported CAB compression : %s", + cab->entry_cffolder->compname); + *avail = ARCHIVE_FAILED; + return (NULL); + } +} + +/* + * Read ahead CFDATA as uncompressed data. + */ +static const void * +cab_read_ahead_cfdata_none(struct archive_read *a, ssize_t *avail) +{ + struct cab *cab = (struct cab *)(a->format->data); + struct cfdata *cfdata; + const void *d; + + cfdata = cab->entry_cfdata; + + /* + * Note: '1' here is a performance optimization. + * Recall that the decompression layer returns a count of + * available bytes; asking for more than that forces the + * decompressor to combine reads by copying data. + */ + d = __archive_read_ahead(a, 1, avail); + if (*avail <= 0) { + *avail = truncated_error(a); + return (NULL); + } + if (*avail > cfdata->uncompressed_bytes_remaining) + *avail = cfdata->uncompressed_bytes_remaining; + cfdata->uncompressed_avail = cfdata->uncompressed_size; + cfdata->unconsumed = *avail; + cfdata->sum_ptr = d; + return (d); +} + +/* + * Read ahead CFDATA as deflate data. + */ +#ifdef HAVE_ZLIB_H +static const void * +cab_read_ahead_cfdata_deflate(struct archive_read *a, ssize_t *avail) +{ + struct cab *cab = (struct cab *)(a->format->data); + struct cfdata *cfdata; + const void *d; + int r, mszip; + uint16_t uavail; + char eod = 0; + + cfdata = cab->entry_cfdata; + /* If the buffer hasn't been allocated, allocate it now. */ + if (cab->uncompressed_buffer == NULL) { + cab->uncompressed_buffer_size = 0x8000; + cab->uncompressed_buffer + = (unsigned char *)malloc(cab->uncompressed_buffer_size); + if (cab->uncompressed_buffer == NULL) { + archive_set_error(&a->archive, ENOMEM, + "No memory for CAB reader"); + *avail = ARCHIVE_FATAL; + return (NULL); + } + } + + uavail = cfdata->uncompressed_avail; + if (uavail == cfdata->uncompressed_size) { + d = cab->uncompressed_buffer + cfdata->read_offset; + *avail = uavail - cfdata->read_offset; + return (d); + } + + if (!cab->entry_cffolder->decompress_init) { + cab->stream.next_in = NULL; + cab->stream.avail_in = 0; + cab->stream.total_in = 0; + cab->stream.next_out = NULL; + cab->stream.avail_out = 0; + cab->stream.total_out = 0; + if (cab->stream_valid) + r = inflateReset(&cab->stream); + else + r = inflateInit2(&cab->stream, + -15 /* Don't check for zlib header */); + if (r != Z_OK) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Can't initialize deflate decompression."); + *avail = ARCHIVE_FATAL; + return (NULL); + } + /* Stream structure has been set up. */ + cab->stream_valid = 1; + /* We've initialized decompression for this stream. */ + cab->entry_cffolder->decompress_init = 1; + } + + if (cfdata->compressed_bytes_remaining == cfdata->compressed_size) + mszip = 2; + else + mszip = 0; + eod = 0; + cab->stream.total_out = uavail; + /* + * We always uncompress all data in current CFDATA. + */ + while (!eod && cab->stream.total_out < cfdata->uncompressed_size) { + ssize_t bytes_avail; + + cab->stream.next_out = + cab->uncompressed_buffer + cab->stream.total_out; + cab->stream.avail_out = + cfdata->uncompressed_size - cab->stream.total_out; + + d = __archive_read_ahead(a, 1, &bytes_avail); + if (bytes_avail <= 0) { + *avail = truncated_error(a); + return (NULL); + } + if (bytes_avail > cfdata->compressed_bytes_remaining) + bytes_avail = cfdata->compressed_bytes_remaining; + /* + * A bug in zlib.h: stream.next_in should be marked 'const' + * but isn't (the library never alters data through the + * next_in pointer, only reads it). The result: this ugly + * cast to remove 'const'. + */ + cab->stream.next_in = (Bytef *)(uintptr_t)d; + cab->stream.avail_in = (uInt)bytes_avail; + cab->stream.total_in = 0; + + /* Cut out a tow-byte MSZIP signature(0x43, 0x4b). */ + if (mszip > 0) { + if (bytes_avail <= 0) + goto nomszip; + if (bytes_avail <= mszip) { + if (mszip == 2) { + if (cab->stream.next_in[0] != 0x43) + goto nomszip; + if (bytes_avail > 1 && + cab->stream.next_in[1] != 0x4b) + goto nomszip; + } else if (cab->stream.next_in[0] != 0x4b) + goto nomszip; + cfdata->unconsumed = bytes_avail; + cfdata->sum_ptr = d; + if (cab_minimum_consume_cfdata( + a, cfdata->unconsumed) < 0) { + *avail = ARCHIVE_FATAL; + return (NULL); + } + mszip -= (int)bytes_avail; + continue; + } + if (mszip == 1 && cab->stream.next_in[0] != 0x4b) + goto nomszip; + else if (mszip == 2 && (cab->stream.next_in[0] != 0x43 || + cab->stream.next_in[1] != 0x4b)) + goto nomszip; + cab->stream.next_in += mszip; + cab->stream.avail_in -= mszip; + cab->stream.total_in += mszip; + mszip = 0; + } + + r = inflate(&cab->stream, 0); + switch (r) { + case Z_OK: + break; + case Z_STREAM_END: + eod = 1; + break; + default: + goto zlibfailed; + } + cfdata->unconsumed = cab->stream.total_in; + cfdata->sum_ptr = d; + if (cab_minimum_consume_cfdata(a, cfdata->unconsumed) < 0) { + *avail = ARCHIVE_FATAL; + return (NULL); + } + } + uavail = (uint16_t)cab->stream.total_out; + + if (uavail < cfdata->uncompressed_size) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Invalid uncompressed size (%d < %d)", + uavail, cfdata->uncompressed_size); + *avail = ARCHIVE_FATAL; + return (NULL); + } + + /* + * Note: I suspect there is a bug in makecab.exe because, in rare + * case, compressed bytes are still remaining regardless we have + * gotten all uncompressed bytes, which size is recorded in CFDATA, + * as much as we need, and we have to use the garbage so as to + * correctly compute the sum of CFDATA accordingly. + */ + if (cfdata->compressed_bytes_remaining > 0) { + ssize_t bytes_avail; + + d = __archive_read_ahead(a, cfdata->compressed_bytes_remaining, + &bytes_avail); + if (bytes_avail <= 0) { + *avail = truncated_error(a); + return (NULL); + } + cfdata->unconsumed = cfdata->compressed_bytes_remaining; + cfdata->sum_ptr = d; + if (cab_minimum_consume_cfdata(a, cfdata->unconsumed) < 0) { + *avail = ARCHIVE_FATAL; + return (NULL); + } + } + + /* + * Set dictionary data for decompressing of next CFDATA, which + * in the same folder. This is why we always do decompress CFDATA + * even if beginning CFDATA or some of CFDATA are not used in + * skipping file data. + */ + if (cab->entry_cffolder->cfdata_index < + cab->entry_cffolder->cfdata_count) { + r = inflateReset(&cab->stream); + if (r != Z_OK) + goto zlibfailed; + r = inflateSetDictionary(&cab->stream, + cab->uncompressed_buffer, cfdata->uncompressed_size); + if (r != Z_OK) + goto zlibfailed; + } + + d = cab->uncompressed_buffer + cfdata->read_offset; + *avail = uavail - cfdata->read_offset; + cfdata->uncompressed_avail = uavail; + + return (d); + +zlibfailed: + switch (r) { + case Z_MEM_ERROR: + archive_set_error(&a->archive, ENOMEM, + "Out of memory for deflate decompression"); + break; + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Deflate decompression failed (%d)", r); + break; + } + *avail = ARCHIVE_FATAL; + return (NULL); +nomszip: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "CFDATA incorrect(no MSZIP signature)"); + *avail = ARCHIVE_FATAL; + return (NULL); +} + +#else /* HAVE_ZLIB_H */ + +static const void * +cab_read_ahead_cfdata_deflate(struct archive_read *a, ssize_t *avail) +{ + *avail = ARCHIVE_FATAL; + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "libarchive compiled without deflate support (no libz)"); + return (NULL); +} + +#endif /* HAVE_ZLIB_H */ + +static const void * +cab_read_ahead_cfdata_lzx(struct archive_read *a, ssize_t *avail) +{ + struct cab *cab = (struct cab *)(a->format->data); + struct cfdata *cfdata; + const void *d; + int r; + uint16_t uavail; + + cfdata = cab->entry_cfdata; + /* If the buffer hasn't been allocated, allocate it now. */ + if (cab->uncompressed_buffer == NULL) { + cab->uncompressed_buffer_size = 0x8000; + cab->uncompressed_buffer + = (unsigned char *)malloc(cab->uncompressed_buffer_size); + if (cab->uncompressed_buffer == NULL) { + archive_set_error(&a->archive, ENOMEM, + "No memory for CAB reader"); + *avail = ARCHIVE_FATAL; + return (NULL); + } + } + + uavail = cfdata->uncompressed_avail; + if (uavail == cfdata->uncompressed_size) { + d = cab->uncompressed_buffer + cfdata->read_offset; + *avail = uavail - cfdata->read_offset; + return (d); + } + + if (!cab->entry_cffolder->decompress_init) { + r = lzx_decode_init(&cab->xstrm, + cab->entry_cffolder->compdata); + if (r != ARCHIVE_OK) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Can't initialize LZX decompression."); + *avail = ARCHIVE_FATAL; + return (NULL); + } + /* We've initialized decompression for this stream. */ + cab->entry_cffolder->decompress_init = 1; + } + + /* Clean up remaining bits of previous CFDATA. */ + lzx_cleanup_bitstream(&cab->xstrm); + cab->xstrm.total_out = uavail; + while (cab->xstrm.total_out < cfdata->uncompressed_size) { + ssize_t bytes_avail; + + cab->xstrm.next_out = + cab->uncompressed_buffer + cab->xstrm.total_out; + cab->xstrm.avail_out = + cfdata->uncompressed_size - cab->xstrm.total_out; + + d = __archive_read_ahead(a, 1, &bytes_avail); + if (bytes_avail <= 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated CAB file data"); + *avail = ARCHIVE_FATAL; + return (NULL); + } + if (bytes_avail > cfdata->compressed_bytes_remaining) + bytes_avail = cfdata->compressed_bytes_remaining; + + cab->xstrm.next_in = d; + cab->xstrm.avail_in = bytes_avail; + cab->xstrm.total_in = 0; + r = lzx_decode(&cab->xstrm, + cfdata->compressed_bytes_remaining == bytes_avail); + switch (r) { + case ARCHIVE_OK: + case ARCHIVE_EOF: + break; + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "LZX decompression failed (%d)", r); + *avail = ARCHIVE_FATAL; + return (NULL); + } + cfdata->unconsumed = cab->xstrm.total_in; + cfdata->sum_ptr = d; + if (cab_minimum_consume_cfdata(a, cfdata->unconsumed) < 0) { + *avail = ARCHIVE_FATAL; + return (NULL); + } + } + + uavail = (uint16_t)cab->xstrm.total_out; + /* + * Make sure a read pointer advances to next CFDATA. + */ + if (cfdata->compressed_bytes_remaining > 0) { + ssize_t bytes_avail; + + d = __archive_read_ahead(a, cfdata->compressed_bytes_remaining, + &bytes_avail); + if (bytes_avail <= 0) { + *avail = truncated_error(a); + return (NULL); + } + cfdata->unconsumed = cfdata->compressed_bytes_remaining; + cfdata->sum_ptr = d; + if (cab_minimum_consume_cfdata(a, cfdata->unconsumed) < 0) { + *avail = ARCHIVE_FATAL; + return (NULL); + } + } + + /* + * Translation reversal of x86 processor CALL byte sequence(E8). + */ + lzx_translation(&cab->xstrm, cab->uncompressed_buffer, + cfdata->uncompressed_size, + (cab->entry_cffolder->cfdata_index-1) * 0x8000); + + d = cab->uncompressed_buffer + cfdata->read_offset; + *avail = uavail - cfdata->read_offset; + cfdata->uncompressed_avail = uavail; + + return (d); +} + +/* + * Consume CFDATA. + * We always decompress CFDATA to consume CFDATA as much as we need + * in uncompressed bytes because all CFDATA in a folder are related + * so we do not skip any CFDATA without decompressing. + * Note: If the folder of a CFFILE is iFoldCONTINUED_PREV_AND_NEXT or + * iFoldCONTINUED_FROM_PREV, we won't decompress because a CFDATA for + * the CFFILE is remaining bytes of previous Multivolume CAB file. + */ +static int64_t +cab_consume_cfdata(struct archive_read *a, int64_t consumed_bytes) +{ + struct cab *cab = (struct cab *)(a->format->data); + struct cfdata *cfdata; + int64_t cbytes, rbytes; + int err; + + rbytes = cab_minimum_consume_cfdata(a, consumed_bytes); + if (rbytes < 0) + return (ARCHIVE_FATAL); + + cfdata = cab->entry_cfdata; + while (rbytes > 0) { + ssize_t avail; + + if (cfdata->compressed_size == 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid CFDATA"); + return (ARCHIVE_FATAL); + } + cbytes = cfdata->uncompressed_bytes_remaining; + if (cbytes > rbytes) + cbytes = rbytes; + rbytes -= cbytes; + + if (cfdata->uncompressed_avail == 0 && + (cab->entry_cffile->folder == iFoldCONTINUED_PREV_AND_NEXT || + cab->entry_cffile->folder == iFoldCONTINUED_FROM_PREV)) { + /* We have not read any data yet. */ + if (cbytes == cfdata->uncompressed_bytes_remaining) { + /* Skip whole current CFDATA. */ + __archive_read_consume(a, + cfdata->compressed_size); + cab->cab_offset += cfdata->compressed_size; + cfdata->compressed_bytes_remaining = 0; + cfdata->uncompressed_bytes_remaining = 0; + err = cab_next_cfdata(a); + if (err < 0) + return (err); + cfdata = cab->entry_cfdata; + if (cfdata->uncompressed_size == 0) { + switch (cab->entry_cffile->folder) { + case iFoldCONTINUED_PREV_AND_NEXT: + case iFoldCONTINUED_TO_NEXT: + case iFoldCONTINUED_FROM_PREV: + rbytes = 0; + break; + default: + break; + } + } + continue; + } + cfdata->read_offset += (uint16_t)cbytes; + cfdata->uncompressed_bytes_remaining -= (uint16_t)cbytes; + break; + } else if (cbytes == 0) { + err = cab_next_cfdata(a); + if (err < 0) + return (err); + cfdata = cab->entry_cfdata; + if (cfdata->uncompressed_size == 0) { + switch (cab->entry_cffile->folder) { + case iFoldCONTINUED_PREV_AND_NEXT: + case iFoldCONTINUED_TO_NEXT: + case iFoldCONTINUED_FROM_PREV: + return (ARCHIVE_FATAL); + default: + break; + } + } + continue; + } + while (cbytes > 0) { + (void)cab_read_ahead_cfdata(a, &avail); + if (avail <= 0) + return (ARCHIVE_FATAL); + if (avail > cbytes) + avail = (ssize_t)cbytes; + if (cab_minimum_consume_cfdata(a, avail) < 0) + return (ARCHIVE_FATAL); + cbytes -= avail; + } + } + return (consumed_bytes); +} + +/* + * Consume CFDATA as much as we have already gotten and + * compute the sum of CFDATA. + */ +static int64_t +cab_minimum_consume_cfdata(struct archive_read *a, int64_t consumed_bytes) +{ + struct cab *cab = (struct cab *)(a->format->data); + struct cfdata *cfdata; + int64_t cbytes, rbytes; + int err; + + cfdata = cab->entry_cfdata; + rbytes = consumed_bytes; + if (cab->entry_cffolder->comptype == COMPTYPE_NONE) { + if (consumed_bytes < cfdata->unconsumed) + cbytes = consumed_bytes; + else + cbytes = cfdata->unconsumed; + rbytes -= cbytes; + cfdata->read_offset += (uint16_t)cbytes; + cfdata->uncompressed_bytes_remaining -= (uint16_t)cbytes; + cfdata->unconsumed -= cbytes; + } else { + cbytes = cfdata->uncompressed_avail - cfdata->read_offset; + if (cbytes > 0) { + if (consumed_bytes < cbytes) + cbytes = consumed_bytes; + rbytes -= cbytes; + cfdata->read_offset += (uint16_t)cbytes; + cfdata->uncompressed_bytes_remaining -= (uint16_t)cbytes; + } + + if (cfdata->unconsumed) { + cbytes = cfdata->unconsumed; + cfdata->unconsumed = 0; + } else + cbytes = 0; + } + if (cbytes) { + /* Compute the sum. */ + cab_checksum_update(a, (size_t)cbytes); + + /* Consume as much as the compressor actually used. */ + __archive_read_consume(a, cbytes); + cab->cab_offset += cbytes; + cfdata->compressed_bytes_remaining -= (uint16_t)cbytes; + if (cfdata->compressed_bytes_remaining == 0) { + err = cab_checksum_finish(a); + if (err < 0) + return (err); + } + } + return (rbytes); +} + +/* + * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets + * cab->end_of_entry if it consumes all of the data. + */ +static int +cab_read_data(struct archive_read *a, const void **buff, + size_t *size, int64_t *offset) +{ + struct cab *cab = (struct cab *)(a->format->data); + ssize_t bytes_avail; + + if (cab->entry_bytes_remaining == 0) { + *buff = NULL; + *size = 0; + *offset = cab->entry_offset; + cab->end_of_entry = 1; + return (ARCHIVE_OK); + } + + *buff = cab_read_ahead_cfdata(a, &bytes_avail); + if (bytes_avail <= 0) { + *buff = NULL; + *size = 0; + *offset = 0; + if (bytes_avail == 0 && + cab->entry_cfdata->uncompressed_size == 0) { + /* All of CFDATA in a folder has been handled. */ + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, "Invalid CFDATA"); + return (ARCHIVE_FATAL); + } else + return ((int)bytes_avail); + } + if (bytes_avail > cab->entry_bytes_remaining) + bytes_avail = (ssize_t)cab->entry_bytes_remaining; + + *size = bytes_avail; + *offset = cab->entry_offset; + cab->entry_offset += bytes_avail; + cab->entry_bytes_remaining -= bytes_avail; + if (cab->entry_bytes_remaining == 0) + cab->end_of_entry = 1; + cab->entry_unconsumed = bytes_avail; + if (cab->entry_cffolder->comptype == COMPTYPE_NONE) { + /* Don't consume more than current entry used. */ + if (cab->entry_cfdata->unconsumed > cab->entry_unconsumed) + cab->entry_cfdata->unconsumed = cab->entry_unconsumed; + } + return (ARCHIVE_OK); +} + +static int +archive_read_format_cab_read_data_skip(struct archive_read *a) +{ + struct cab *cab; + int64_t bytes_skipped; + int r; + + cab = (struct cab *)(a->format->data); + + if (cab->end_of_archive) + return (ARCHIVE_EOF); + + if (!cab->read_data_invoked) { + cab->bytes_skipped += cab->entry_bytes_remaining; + cab->entry_bytes_remaining = 0; + /* This entry is finished and done. */ + cab->end_of_entry_cleanup = cab->end_of_entry = 1; + return (ARCHIVE_OK); + } + + if (cab->entry_unconsumed) { + /* Consume as much as the compressor actually used. */ + r = (int)cab_consume_cfdata(a, cab->entry_unconsumed); + cab->entry_unconsumed = 0; + if (r < 0) + return (r); + } else if (cab->entry_cfdata == NULL) { + r = cab_next_cfdata(a); + if (r < 0) + return (r); + } + + /* if we've already read to end of data, we're done. */ + if (cab->end_of_entry_cleanup) + return (ARCHIVE_OK); + + /* + * If the length is at the beginning, we can skip the + * compressed data much more quickly. + */ + bytes_skipped = cab_consume_cfdata(a, cab->entry_bytes_remaining); + if (bytes_skipped < 0) + return (ARCHIVE_FATAL); + + /* If the compression type is none(uncompressed), we've already + * consumed data as much as the current entry size. */ + if (cab->entry_cffolder->comptype == COMPTYPE_NONE && + cab->entry_cfdata != NULL) + cab->entry_cfdata->unconsumed = 0; + + /* This entry is finished and done. */ + cab->end_of_entry_cleanup = cab->end_of_entry = 1; + return (ARCHIVE_OK); +} + +static int +archive_read_format_cab_cleanup(struct archive_read *a) +{ + struct cab *cab = (struct cab *)(a->format->data); + struct cfheader *hd = &cab->cfheader; + int i; + + if (hd->folder_array != NULL) { + for (i = 0; i < hd->folder_count; i++) + free(hd->folder_array[i].cfdata.memimage); + free(hd->folder_array); + } + if (hd->file_array != NULL) { + for (i = 0; i < cab->cfheader.file_count; i++) + archive_string_free(&(hd->file_array[i].pathname)); + free(hd->file_array); + } +#ifdef HAVE_ZLIB_H + if (cab->stream_valid) + inflateEnd(&cab->stream); +#endif + lzx_decode_free(&cab->xstrm); + archive_wstring_free(&cab->ws); + free(cab->uncompressed_buffer); + free(cab); + (a->format->data) = NULL; + return (ARCHIVE_OK); +} + +/* Convert an MSDOS-style date/time into Unix-style time. */ +static time_t +cab_dos_time(const unsigned char *p) +{ + int msTime, msDate; + struct tm ts; + + msDate = archive_le16dec(p); + msTime = archive_le16dec(p+2); + + memset(&ts, 0, sizeof(ts)); + ts.tm_year = ((msDate >> 9) & 0x7f) + 80; /* Years since 1900. */ + ts.tm_mon = ((msDate >> 5) & 0x0f) - 1; /* Month number. */ + ts.tm_mday = msDate & 0x1f; /* Day of month. */ + ts.tm_hour = (msTime >> 11) & 0x1f; + ts.tm_min = (msTime >> 5) & 0x3f; + ts.tm_sec = (msTime << 1) & 0x3e; + ts.tm_isdst = -1; + return (mktime(&ts)); +} + +/***************************************************************** + * + * LZX decompression code. + * + *****************************************************************/ + +/* + * Initialize LZX decoder. + * + * Returns ARCHIVE_OK if initialization was successful. + * Returns ARCHIVE_FAILED if w_bits has unsupported value. + * Returns ARCHIVE_FATAL if initialization failed; memory allocation + * error occurred. + */ +static int +lzx_decode_init(struct lzx_stream *strm, int w_bits) +{ + struct lzx_dec *ds; + int slot, w_size, w_slot; + int base, footer; + int base_inc[18]; + + if (strm->ds == NULL) { + strm->ds = calloc(1, sizeof(*strm->ds)); + if (strm->ds == NULL) + return (ARCHIVE_FATAL); + } + ds = strm->ds; + ds->error = ARCHIVE_FAILED; + + /* Allow bits from 15(32KBi) up to 21(2MBi) */ + if (w_bits < SLOT_BASE || w_bits > SLOT_MAX) + return (ARCHIVE_FAILED); + + ds->error = ARCHIVE_FATAL; + + /* + * Alloc window + */ + w_size = ds->w_size; + w_slot = slots[w_bits - SLOT_BASE]; + ds->w_size = 1U << w_bits; + ds->w_mask = ds->w_size -1; + if (ds->w_buff == NULL || w_size != ds->w_size) { + free(ds->w_buff); + ds->w_buff = malloc(ds->w_size); + if (ds->w_buff == NULL) + return (ARCHIVE_FATAL); + free(ds->pos_tbl); + ds->pos_tbl = malloc(sizeof(ds->pos_tbl[0]) * w_slot); + if (ds->pos_tbl == NULL) + return (ARCHIVE_FATAL); + lzx_huffman_free(&(ds->mt)); + } + + for (footer = 0; footer < 18; footer++) + base_inc[footer] = 1 << footer; + base = footer = 0; + for (slot = 0; slot < w_slot; slot++) { + int n; + if (footer == 0) + base = slot; + else + base += base_inc[footer]; + if (footer < 17) { + footer = -2; + for (n = base; n; n >>= 1) + footer++; + if (footer <= 0) + footer = 0; + } + ds->pos_tbl[slot].base = base; + ds->pos_tbl[slot].footer_bits = footer; + } + + ds->w_pos = 0; + ds->state = 0; + ds->br.cache_buffer = 0; + ds->br.cache_avail = 0; + ds->r0 = ds->r1 = ds->r2 = 1; + + /* Initialize aligned offset tree. */ + if (lzx_huffman_init(&(ds->at), 8, 8) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + /* Initialize pre-tree. */ + if (lzx_huffman_init(&(ds->pt), 20, 10) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + /* Initialize Main tree. */ + if (lzx_huffman_init(&(ds->mt), 256+(w_slot<<3), 16) + != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + /* Initialize Length tree. */ + if (lzx_huffman_init(&(ds->lt), 249, 16) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + ds->error = 0; + + return (ARCHIVE_OK); +} + +/* + * Release LZX decoder. + */ +static void +lzx_decode_free(struct lzx_stream *strm) +{ + + if (strm->ds == NULL) + return; + free(strm->ds->w_buff); + free(strm->ds->pos_tbl); + lzx_huffman_free(&(strm->ds->at)); + lzx_huffman_free(&(strm->ds->pt)); + lzx_huffman_free(&(strm->ds->mt)); + lzx_huffman_free(&(strm->ds->lt)); + free(strm->ds); + strm->ds = NULL; +} + +/* + * E8 Call Translation reversal. + */ +static void +lzx_translation(struct lzx_stream *strm, void *p, size_t size, uint32_t offset) +{ + struct lzx_dec *ds = strm->ds; + unsigned char *b, *end; + + if (!ds->translation || size <= 10) + return; + b = p; + end = b + size - 10; + while (b < end && (b = memchr(b, 0xE8, end - b)) != NULL) { + size_t i = b - (unsigned char *)p; + int32_t cp, displacement, value; + + cp = (int32_t)(offset + (uint32_t)i); + value = archive_le32dec(&b[1]); + if (value >= -cp && value < (int32_t)ds->translation_size) { + if (value >= 0) + displacement = value - cp; + else + displacement = value + ds->translation_size; + archive_le32enc(&b[1], (uint32_t)displacement); + } + b += 5; + } +} + +/* + * Bit stream reader. + */ +/* Check that the cache buffer has enough bits. */ +#define lzx_br_has(br, n) ((br)->cache_avail >= n) +/* Get compressed data by bit. */ +#define lzx_br_bits(br, n) \ + (((uint32_t)((br)->cache_buffer >> \ + ((br)->cache_avail - (n)))) & cache_masks[n]) +#define lzx_br_bits_forced(br, n) \ + (((uint32_t)((br)->cache_buffer << \ + ((n) - (br)->cache_avail))) & cache_masks[n]) +/* Read ahead to make sure the cache buffer has enough compressed data we + * will use. + * True : completed, there is enough data in the cache buffer. + * False : we met that strm->next_in is empty, we have to get following + * bytes. */ +#define lzx_br_read_ahead_0(strm, br, n) \ + (lzx_br_has((br), (n)) || lzx_br_fillup(strm, br)) +/* True : the cache buffer has some bits as much as we need. + * False : there are no enough bits in the cache buffer to be used, + * we have to get following bytes if we could. */ +#define lzx_br_read_ahead(strm, br, n) \ + (lzx_br_read_ahead_0((strm), (br), (n)) || lzx_br_has((br), (n))) + +/* Notify how many bits we consumed. */ +#define lzx_br_consume(br, n) ((br)->cache_avail -= (n)) +#define lzx_br_consume_unaligned_bits(br) ((br)->cache_avail &= ~0x0f) + +#define lzx_br_is_unaligned(br) ((br)->cache_avail & 0x0f) + +static const uint32_t cache_masks[] = { + 0x00000000, 0x00000001, 0x00000003, 0x00000007, + 0x0000000F, 0x0000001F, 0x0000003F, 0x0000007F, + 0x000000FF, 0x000001FF, 0x000003FF, 0x000007FF, + 0x00000FFF, 0x00001FFF, 0x00003FFF, 0x00007FFF, + 0x0000FFFF, 0x0001FFFF, 0x0003FFFF, 0x0007FFFF, + 0x000FFFFF, 0x001FFFFF, 0x003FFFFF, 0x007FFFFF, + 0x00FFFFFF, 0x01FFFFFF, 0x03FFFFFF, 0x07FFFFFF, + 0x0FFFFFFF, 0x1FFFFFFF, 0x3FFFFFFF, 0x7FFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF +}; + +/* + * Shift away used bits in the cache data and fill it up with following bits. + * Call this when cache buffer does not have enough bits you need. + * + * Returns 1 if the cache buffer is full. + * Returns 0 if the cache buffer is not full; input buffer is empty. + */ +static int +lzx_br_fillup(struct lzx_stream *strm, struct lzx_br *br) +{ +/* + * x86 processor family can read misaligned data without an access error. + */ + int n = CACHE_BITS - br->cache_avail; + + for (;;) { + switch (n >> 4) { + case 4: + if (strm->avail_in >= 8) { + br->cache_buffer = + ((uint64_t)strm->next_in[1]) << 56 | + ((uint64_t)strm->next_in[0]) << 48 | + ((uint64_t)strm->next_in[3]) << 40 | + ((uint64_t)strm->next_in[2]) << 32 | + ((uint32_t)strm->next_in[5]) << 24 | + ((uint32_t)strm->next_in[4]) << 16 | + ((uint32_t)strm->next_in[7]) << 8 | + (uint32_t)strm->next_in[6]; + strm->next_in += 8; + strm->avail_in -= 8; + br->cache_avail += 8 * 8; + return (1); + } + break; + case 3: + if (strm->avail_in >= 6) { + br->cache_buffer = + (br->cache_buffer << 48) | + ((uint64_t)strm->next_in[1]) << 40 | + ((uint64_t)strm->next_in[0]) << 32 | + ((uint32_t)strm->next_in[3]) << 24 | + ((uint32_t)strm->next_in[2]) << 16 | + ((uint32_t)strm->next_in[5]) << 8 | + (uint32_t)strm->next_in[4]; + strm->next_in += 6; + strm->avail_in -= 6; + br->cache_avail += 6 * 8; + return (1); + } + break; + case 0: + /* We have enough compressed data in + * the cache buffer.*/ + return (1); + default: + break; + } + if (strm->avail_in < 2) { + /* There is not enough compressed data to + * fill up the cache buffer. */ + if (strm->avail_in == 1) { + br->odd = *strm->next_in++; + strm->avail_in--; + br->have_odd = 1; + } + return (0); + } + br->cache_buffer = + (br->cache_buffer << 16) | + archive_le16dec(strm->next_in); + strm->next_in += 2; + strm->avail_in -= 2; + br->cache_avail += 16; + n -= 16; + } +} + +static void +lzx_br_fixup(struct lzx_stream *strm, struct lzx_br *br) +{ + int n = CACHE_BITS - br->cache_avail; + + if (br->have_odd && n >= 16 && strm->avail_in > 0) { + br->cache_buffer = + (br->cache_buffer << 16) | + ((uint16_t)(*strm->next_in)) << 8 | br->odd; + strm->next_in++; + strm->avail_in--; + br->cache_avail += 16; + br->have_odd = 0; + } +} + +static void +lzx_cleanup_bitstream(struct lzx_stream *strm) +{ + strm->ds->br.cache_avail = 0; + strm->ds->br.have_odd = 0; +} + +/* + * Decode LZX. + * + * 1. Returns ARCHIVE_OK if output buffer or input buffer are empty. + * Please set available buffer and call this function again. + * 2. Returns ARCHIVE_EOF if decompression has been completed. + * 3. Returns ARCHIVE_FAILED if an error occurred; compressed data + * is broken or you do not set 'last' flag properly. + */ +#define ST_RD_TRANSLATION 0 +#define ST_RD_TRANSLATION_SIZE 1 +#define ST_RD_BLOCK_TYPE 2 +#define ST_RD_BLOCK_SIZE 3 +#define ST_RD_ALIGNMENT 4 +#define ST_RD_R0 5 +#define ST_RD_R1 6 +#define ST_RD_R2 7 +#define ST_COPY_UNCOMP1 8 +#define ST_COPY_UNCOMP2 9 +#define ST_RD_ALIGNED_OFFSET 10 +#define ST_RD_VERBATIM 11 +#define ST_RD_PRE_MAIN_TREE_256 12 +#define ST_MAIN_TREE_256 13 +#define ST_RD_PRE_MAIN_TREE_REM 14 +#define ST_MAIN_TREE_REM 15 +#define ST_RD_PRE_LENGTH_TREE 16 +#define ST_LENGTH_TREE 17 +#define ST_MAIN 18 +#define ST_LENGTH 19 +#define ST_OFFSET 20 +#define ST_REAL_POS 21 +#define ST_COPY 22 + +static int +lzx_decode(struct lzx_stream *strm, int last) +{ + struct lzx_dec *ds = strm->ds; + int64_t avail_in; + int r; + + if (ds->error) + return (ds->error); + + avail_in = strm->avail_in; + lzx_br_fixup(strm, &(ds->br)); + do { + if (ds->state < ST_MAIN) + r = lzx_read_blocks(strm, last); + else { + int64_t bytes_written = strm->avail_out; + r = lzx_decode_blocks(strm, last); + bytes_written -= strm->avail_out; + strm->next_out += bytes_written; + strm->total_out += bytes_written; + } + } while (r == 100); + strm->total_in += avail_in - strm->avail_in; + return (r); +} + +static int +lzx_read_blocks(struct lzx_stream *strm, int last) +{ + struct lzx_dec *ds = strm->ds; + struct lzx_br *br = &(ds->br); + int i, r; + + for (;;) { + switch (ds->state) { + case ST_RD_TRANSLATION: + if (!lzx_br_read_ahead(strm, br, 1)) { + ds->state = ST_RD_TRANSLATION; + if (last) + goto failed; + return (ARCHIVE_OK); + } + ds->translation = lzx_br_bits(br, 1); + lzx_br_consume(br, 1); + /* FALL THROUGH */ + case ST_RD_TRANSLATION_SIZE: + if (ds->translation) { + if (!lzx_br_read_ahead(strm, br, 32)) { + ds->state = ST_RD_TRANSLATION_SIZE; + if (last) + goto failed; + return (ARCHIVE_OK); + } + ds->translation_size = lzx_br_bits(br, 16); + lzx_br_consume(br, 16); + ds->translation_size <<= 16; + ds->translation_size |= lzx_br_bits(br, 16); + lzx_br_consume(br, 16); + } + /* FALL THROUGH */ + case ST_RD_BLOCK_TYPE: + if (!lzx_br_read_ahead(strm, br, 3)) { + ds->state = ST_RD_BLOCK_TYPE; + if (last) + goto failed; + return (ARCHIVE_OK); + } + ds->block_type = lzx_br_bits(br, 3); + lzx_br_consume(br, 3); + /* Check a block type. */ + switch (ds->block_type) { + case VERBATIM_BLOCK: + case ALIGNED_OFFSET_BLOCK: + case UNCOMPRESSED_BLOCK: + break; + default: + goto failed;/* Invalid */ + } + /* FALL THROUGH */ + case ST_RD_BLOCK_SIZE: + if (!lzx_br_read_ahead(strm, br, 24)) { + ds->state = ST_RD_BLOCK_SIZE; + if (last) + goto failed; + return (ARCHIVE_OK); + } + ds->block_size = lzx_br_bits(br, 8); + lzx_br_consume(br, 8); + ds->block_size <<= 16; + ds->block_size |= lzx_br_bits(br, 16); + lzx_br_consume(br, 16); + if (ds->block_size == 0) + goto failed; + ds->block_bytes_avail = ds->block_size; + if (ds->block_type != UNCOMPRESSED_BLOCK) { + if (ds->block_type == VERBATIM_BLOCK) + ds->state = ST_RD_VERBATIM; + else + ds->state = ST_RD_ALIGNED_OFFSET; + break; + } + /* FALL THROUGH */ + case ST_RD_ALIGNMENT: + /* + * Handle an Uncompressed Block. + */ + /* Skip padding to align following field on + * 16-bit boundary. */ + if (lzx_br_is_unaligned(br)) + lzx_br_consume_unaligned_bits(br); + else { + if (lzx_br_read_ahead(strm, br, 16)) + lzx_br_consume(br, 16); + else { + ds->state = ST_RD_ALIGNMENT; + if (last) + goto failed; + return (ARCHIVE_OK); + } + } + /* Preparation to read repeated offsets R0,R1 and R2. */ + ds->rbytes_avail = 0; + ds->state = ST_RD_R0; + /* FALL THROUGH */ + case ST_RD_R0: + case ST_RD_R1: + case ST_RD_R2: + do { + uint16_t u16; + /* Drain bits in the cache buffer of + * bit-stream. */ + if (lzx_br_has(br, 32)) { + u16 = lzx_br_bits(br, 16); + lzx_br_consume(br, 16); + archive_le16enc(ds->rbytes, u16); + u16 = lzx_br_bits(br, 16); + lzx_br_consume(br, 16); + archive_le16enc(ds->rbytes+2, u16); + ds->rbytes_avail = 4; + } else if (lzx_br_has(br, 16)) { + u16 = lzx_br_bits(br, 16); + lzx_br_consume(br, 16); + archive_le16enc(ds->rbytes, u16); + ds->rbytes_avail = 2; + } + if (ds->rbytes_avail < 4 && ds->br.have_odd) { + ds->rbytes[ds->rbytes_avail++] = + ds->br.odd; + ds->br.have_odd = 0; + } + while (ds->rbytes_avail < 4) { + if (strm->avail_in <= 0) { + if (last) + goto failed; + return (ARCHIVE_OK); + } + ds->rbytes[ds->rbytes_avail++] = + *strm->next_in++; + strm->avail_in--; + } + ds->rbytes_avail = 0; + if (ds->state == ST_RD_R0) { + ds->r0 = archive_le32dec(ds->rbytes); + if (ds->r0 < 0) + goto failed; + ds->state = ST_RD_R1; + } else if (ds->state == ST_RD_R1) { + ds->r1 = archive_le32dec(ds->rbytes); + if (ds->r1 < 0) + goto failed; + ds->state = ST_RD_R2; + } else if (ds->state == ST_RD_R2) { + ds->r2 = archive_le32dec(ds->rbytes); + if (ds->r2 < 0) + goto failed; + /* We've gotten all repeated offsets. */ + ds->state = ST_COPY_UNCOMP1; + } + } while (ds->state != ST_COPY_UNCOMP1); + /* FALL THROUGH */ + case ST_COPY_UNCOMP1: + /* + * Copy bytes form next_in to next_out directly. + */ + while (ds->block_bytes_avail) { + int l; + + if (strm->avail_out <= 0) + /* Output buffer is empty. */ + return (ARCHIVE_OK); + if (strm->avail_in <= 0) { + /* Input buffer is empty. */ + if (last) + goto failed; + return (ARCHIVE_OK); + } + l = (int)ds->block_bytes_avail; + if (l > ds->w_size - ds->w_pos) + l = ds->w_size - ds->w_pos; + if (l > strm->avail_out) + l = (int)strm->avail_out; + if (l > strm->avail_in) + l = (int)strm->avail_in; + memcpy(strm->next_out, strm->next_in, l); + memcpy(&(ds->w_buff[ds->w_pos]), + strm->next_in, l); + strm->next_in += l; + strm->avail_in -= l; + strm->next_out += l; + strm->avail_out -= l; + strm->total_out += l; + ds->w_pos = (ds->w_pos + l) & ds->w_mask; + ds->block_bytes_avail -= l; + } + /* FALL THROUGH */ + case ST_COPY_UNCOMP2: + /* Re-align; skip padding byte. */ + if (ds->block_size & 1) { + if (strm->avail_in <= 0) { + /* Input buffer is empty. */ + ds->state = ST_COPY_UNCOMP2; + if (last) + goto failed; + return (ARCHIVE_OK); + } + strm->next_in++; + strm->avail_in --; + } + /* This block ended. */ + ds->state = ST_RD_BLOCK_TYPE; + return (ARCHIVE_EOF); + /********************/ + case ST_RD_ALIGNED_OFFSET: + /* + * Read Aligned offset tree. + */ + if (!lzx_br_read_ahead(strm, br, 3 * ds->at.len_size)) { + ds->state = ST_RD_ALIGNED_OFFSET; + if (last) + goto failed; + return (ARCHIVE_OK); + } + memset(ds->at.freq, 0, sizeof(ds->at.freq)); + for (i = 0; i < ds->at.len_size; i++) { + ds->at.bitlen[i] = lzx_br_bits(br, 3); + ds->at.freq[ds->at.bitlen[i]]++; + lzx_br_consume(br, 3); + } + if (!lzx_make_huffman_table(&ds->at)) + goto failed; + /* FALL THROUGH */ + case ST_RD_VERBATIM: + ds->loop = 0; + /* FALL THROUGH */ + case ST_RD_PRE_MAIN_TREE_256: + /* + * Read Pre-tree for first 256 elements of main tree. + */ + if (!lzx_read_pre_tree(strm)) { + ds->state = ST_RD_PRE_MAIN_TREE_256; + if (last) + goto failed; + return (ARCHIVE_OK); + } + if (!lzx_make_huffman_table(&ds->pt)) + goto failed; + ds->loop = 0; + /* FALL THROUGH */ + case ST_MAIN_TREE_256: + /* + * Get path lengths of first 256 elements of main tree. + */ + r = lzx_read_bitlen(strm, &ds->mt, 256); + if (r < 0) + goto failed; + else if (!r) { + ds->state = ST_MAIN_TREE_256; + if (last) + goto failed; + return (ARCHIVE_OK); + } + ds->loop = 0; + /* FALL THROUGH */ + case ST_RD_PRE_MAIN_TREE_REM: + /* + * Read Pre-tree for remaining elements of main tree. + */ + if (!lzx_read_pre_tree(strm)) { + ds->state = ST_RD_PRE_MAIN_TREE_REM; + if (last) + goto failed; + return (ARCHIVE_OK); + } + if (!lzx_make_huffman_table(&ds->pt)) + goto failed; + ds->loop = 256; + /* FALL THROUGH */ + case ST_MAIN_TREE_REM: + /* + * Get path lengths of remaining elements of main tree. + */ + r = lzx_read_bitlen(strm, &ds->mt, -1); + if (r < 0) + goto failed; + else if (!r) { + ds->state = ST_MAIN_TREE_REM; + if (last) + goto failed; + return (ARCHIVE_OK); + } + if (!lzx_make_huffman_table(&ds->mt)) + goto failed; + ds->loop = 0; + /* FALL THROUGH */ + case ST_RD_PRE_LENGTH_TREE: + /* + * Read Pre-tree for remaining elements of main tree. + */ + if (!lzx_read_pre_tree(strm)) { + ds->state = ST_RD_PRE_LENGTH_TREE; + if (last) + goto failed; + return (ARCHIVE_OK); + } + if (!lzx_make_huffman_table(&ds->pt)) + goto failed; + ds->loop = 0; + /* FALL THROUGH */ + case ST_LENGTH_TREE: + /* + * Get path lengths of remaining elements of main tree. + */ + r = lzx_read_bitlen(strm, &ds->lt, -1); + if (r < 0) + goto failed; + else if (!r) { + ds->state = ST_LENGTH_TREE; + if (last) + goto failed; + return (ARCHIVE_OK); + } + if (!lzx_make_huffman_table(&ds->lt)) + goto failed; + ds->state = ST_MAIN; + return (100); + } + } +failed: + return (ds->error = ARCHIVE_FAILED); +} + +static int +lzx_decode_blocks(struct lzx_stream *strm, int last) +{ + struct lzx_dec *ds = strm->ds; + struct lzx_br bre = ds->br; + struct huffman *at = &(ds->at), *lt = &(ds->lt), *mt = &(ds->mt); + const struct lzx_pos_tbl *pos_tbl = ds->pos_tbl; + unsigned char *noutp = strm->next_out; + unsigned char *endp = noutp + strm->avail_out; + unsigned char *w_buff = ds->w_buff; + unsigned char *at_bitlen = at->bitlen; + unsigned char *lt_bitlen = lt->bitlen; + unsigned char *mt_bitlen = mt->bitlen; + size_t block_bytes_avail = ds->block_bytes_avail; + int at_max_bits = at->max_bits; + int lt_max_bits = lt->max_bits; + int mt_max_bits = mt->max_bits; + int c, copy_len = ds->copy_len, copy_pos = ds->copy_pos; + int w_pos = ds->w_pos, w_mask = ds->w_mask, w_size = ds->w_size; + int length_header = ds->length_header; + int offset_bits = ds->offset_bits; + int position_slot = ds->position_slot; + int r0 = ds->r0, r1 = ds->r1, r2 = ds->r2; + int state = ds->state; + char block_type = ds->block_type; + + for (;;) { + switch (state) { + case ST_MAIN: + for (;;) { + if (block_bytes_avail == 0) { + /* This block ended. */ + ds->state = ST_RD_BLOCK_TYPE; + ds->br = bre; + ds->block_bytes_avail = + block_bytes_avail; + ds->copy_len = copy_len; + ds->copy_pos = copy_pos; + ds->length_header = length_header; + ds->position_slot = position_slot; + ds->r0 = r0; ds->r1 = r1; ds->r2 = r2; + ds->w_pos = w_pos; + strm->avail_out = endp - noutp; + return (ARCHIVE_EOF); + } + if (noutp >= endp) + /* Output buffer is empty. */ + goto next_data; + + if (!lzx_br_read_ahead(strm, &bre, + mt_max_bits)) { + if (!last) + goto next_data; + /* Remaining bits are less than + * maximum bits(mt.max_bits) but maybe + * it still remains as much as we need, + * so we should try to use it with + * dummy bits. */ + c = lzx_decode_huffman(mt, + lzx_br_bits_forced( + &bre, mt_max_bits)); + lzx_br_consume(&bre, mt_bitlen[c]); + if (!lzx_br_has(&bre, 0)) + goto failed;/* Over read. */ + } else { + c = lzx_decode_huffman(mt, + lzx_br_bits(&bre, mt_max_bits)); + lzx_br_consume(&bre, mt_bitlen[c]); + } + if (c > UCHAR_MAX) + break; + /* + * 'c' is exactly literal code. + */ + /* Save a decoded code to reference it + * afterward. */ + w_buff[w_pos] = c; + w_pos = (w_pos + 1) & w_mask; + /* Store the decoded code to output buffer. */ + *noutp++ = c; + block_bytes_avail--; + } + /* + * Get a match code, its length and offset. + */ + c -= UCHAR_MAX + 1; + length_header = c & 7; + position_slot = c >> 3; + /* FALL THROUGH */ + case ST_LENGTH: + /* + * Get a length. + */ + if (length_header == 7) { + if (!lzx_br_read_ahead(strm, &bre, + lt_max_bits)) { + if (!last) { + state = ST_LENGTH; + goto next_data; + } + c = lzx_decode_huffman(lt, + lzx_br_bits_forced( + &bre, lt_max_bits)); + lzx_br_consume(&bre, lt_bitlen[c]); + if (!lzx_br_has(&bre, 0)) + goto failed;/* Over read. */ + } else { + c = lzx_decode_huffman(lt, + lzx_br_bits(&bre, lt_max_bits)); + lzx_br_consume(&bre, lt_bitlen[c]); + } + copy_len = c + 7 + 2; + } else + copy_len = length_header + 2; + if ((size_t)copy_len > block_bytes_avail) + goto failed; + /* + * Get an offset. + */ + switch (position_slot) { + case 0: /* Use repeated offset 0. */ + copy_pos = r0; + state = ST_REAL_POS; + continue; + case 1: /* Use repeated offset 1. */ + copy_pos = r1; + /* Swap repeated offset. */ + r1 = r0; + r0 = copy_pos; + state = ST_REAL_POS; + continue; + case 2: /* Use repeated offset 2. */ + copy_pos = r2; + /* Swap repeated offset. */ + r2 = r0; + r0 = copy_pos; + state = ST_REAL_POS; + continue; + default: + offset_bits = + pos_tbl[position_slot].footer_bits; + break; + } + /* FALL THROUGH */ + case ST_OFFSET: + /* + * Get the offset, which is a distance from + * current window position. + */ + if (block_type == ALIGNED_OFFSET_BLOCK && + offset_bits >= 3) { + int offbits = offset_bits - 3; + + if (!lzx_br_read_ahead(strm, &bre, offbits)) { + state = ST_OFFSET; + if (last) + goto failed; + goto next_data; + } + copy_pos = lzx_br_bits(&bre, offbits) << 3; + + /* Get an aligned number. */ + if (!lzx_br_read_ahead(strm, &bre, + offbits + at_max_bits)) { + if (!last) { + state = ST_OFFSET; + goto next_data; + } + lzx_br_consume(&bre, offbits); + c = lzx_decode_huffman(at, + lzx_br_bits_forced(&bre, + at_max_bits)); + lzx_br_consume(&bre, at_bitlen[c]); + if (!lzx_br_has(&bre, 0)) + goto failed;/* Over read. */ + } else { + lzx_br_consume(&bre, offbits); + c = lzx_decode_huffman(at, + lzx_br_bits(&bre, at_max_bits)); + lzx_br_consume(&bre, at_bitlen[c]); + } + /* Add an aligned number. */ + copy_pos += c; + } else { + if (!lzx_br_read_ahead(strm, &bre, + offset_bits)) { + state = ST_OFFSET; + if (last) + goto failed; + goto next_data; + } + copy_pos = lzx_br_bits(&bre, offset_bits); + lzx_br_consume(&bre, offset_bits); + } + copy_pos += pos_tbl[position_slot].base -2; + + /* Update repeated offset LRU queue. */ + r2 = r1; + r1 = r0; + r0 = copy_pos; + /* FALL THROUGH */ + case ST_REAL_POS: + /* + * Compute a real position in window. + */ + copy_pos = (w_pos - copy_pos) & w_mask; + /* FALL THROUGH */ + case ST_COPY: + /* + * Copy several bytes as extracted data from the window + * into the output buffer. + */ + for (;;) { + const unsigned char *s; + int l; + + l = copy_len; + if (copy_pos > w_pos) { + if (l > w_size - copy_pos) + l = w_size - copy_pos; + } else { + if (l > w_size - w_pos) + l = w_size - w_pos; + } + if (noutp + l >= endp) + l = (int)(endp - noutp); + s = w_buff + copy_pos; + if (l >= 8 && ((copy_pos + l < w_pos) + || (w_pos + l < copy_pos))) { + memcpy(w_buff + w_pos, s, l); + memcpy(noutp, s, l); + } else { + unsigned char *d; + int li; + + d = w_buff + w_pos; + for (li = 0; li < l; li++) + noutp[li] = d[li] = s[li]; + } + noutp += l; + copy_pos = (copy_pos + l) & w_mask; + w_pos = (w_pos + l) & w_mask; + block_bytes_avail -= l; + if (copy_len <= l) + /* A copy of current pattern ended. */ + break; + copy_len -= l; + if (noutp >= endp) { + /* Output buffer is empty. */ + state = ST_COPY; + goto next_data; + } + } + state = ST_MAIN; + break; + } + } +failed: + return (ds->error = ARCHIVE_FAILED); +next_data: + ds->br = bre; + ds->block_bytes_avail = block_bytes_avail; + ds->copy_len = copy_len; + ds->copy_pos = copy_pos; + ds->length_header = length_header; + ds->offset_bits = offset_bits; + ds->position_slot = position_slot; + ds->r0 = r0; ds->r1 = r1; ds->r2 = r2; + ds->state = state; + ds->w_pos = w_pos; + strm->avail_out = endp - noutp; + return (ARCHIVE_OK); +} + +static int +lzx_read_pre_tree(struct lzx_stream *strm) +{ + struct lzx_dec *ds = strm->ds; + struct lzx_br *br = &(ds->br); + int i; + + if (ds->loop == 0) + memset(ds->pt.freq, 0, sizeof(ds->pt.freq)); + for (i = ds->loop; i < ds->pt.len_size; i++) { + if (!lzx_br_read_ahead(strm, br, 4)) { + ds->loop = i; + return (0); + } + ds->pt.bitlen[i] = lzx_br_bits(br, 4); + ds->pt.freq[ds->pt.bitlen[i]]++; + lzx_br_consume(br, 4); + } + ds->loop = i; + return (1); +} + +/* + * Read a bunch of bit-lengths from pre-tree. + */ +static int +lzx_read_bitlen(struct lzx_stream *strm, struct huffman *d, int end) +{ + struct lzx_dec *ds = strm->ds; + struct lzx_br *br = &(ds->br); + int c, i, j, ret, same; + unsigned rbits; + + i = ds->loop; + if (i == 0) + memset(d->freq, 0, sizeof(d->freq)); + ret = 0; + if (end < 0) + end = d->len_size; + while (i < end) { + ds->loop = i; + if (!lzx_br_read_ahead(strm, br, ds->pt.max_bits)) + goto getdata; + rbits = lzx_br_bits(br, ds->pt.max_bits); + c = lzx_decode_huffman(&(ds->pt), rbits); + switch (c) { + case 17:/* several zero lengths, from 4 to 19. */ + if (!lzx_br_read_ahead(strm, br, ds->pt.bitlen[c]+4)) + goto getdata; + lzx_br_consume(br, ds->pt.bitlen[c]); + same = lzx_br_bits(br, 4) + 4; + if (i + same > end) + return (-1);/* Invalid */ + lzx_br_consume(br, 4); + for (j = 0; j < same; j++) + d->bitlen[i++] = 0; + break; + case 18:/* many zero lengths, from 20 to 51. */ + if (!lzx_br_read_ahead(strm, br, ds->pt.bitlen[c]+5)) + goto getdata; + lzx_br_consume(br, ds->pt.bitlen[c]); + same = lzx_br_bits(br, 5) + 20; + if (i + same > end) + return (-1);/* Invalid */ + lzx_br_consume(br, 5); + memset(d->bitlen + i, 0, same); + i += same; + break; + case 19:/* a few same lengths. */ + if (!lzx_br_read_ahead(strm, br, + ds->pt.bitlen[c]+1+ds->pt.max_bits)) + goto getdata; + lzx_br_consume(br, ds->pt.bitlen[c]); + same = lzx_br_bits(br, 1) + 4; + if (i + same > end) + return (-1); + lzx_br_consume(br, 1); + rbits = lzx_br_bits(br, ds->pt.max_bits); + c = lzx_decode_huffman(&(ds->pt), rbits); + lzx_br_consume(br, ds->pt.bitlen[c]); + c = (d->bitlen[i] - c + 17) % 17; + if (c < 0) + return (-1);/* Invalid */ + for (j = 0; j < same; j++) + d->bitlen[i++] = c; + d->freq[c] += same; + break; + default: + lzx_br_consume(br, ds->pt.bitlen[c]); + c = (d->bitlen[i] - c + 17) % 17; + if (c < 0) + return (-1);/* Invalid */ + d->freq[c]++; + d->bitlen[i++] = c; + break; + } + } + ret = 1; +getdata: + ds->loop = i; + return (ret); +} + +static int +lzx_huffman_init(struct huffman *hf, size_t len_size, int tbl_bits) +{ + + if (hf->bitlen == NULL || hf->len_size != (int)len_size) { + free(hf->bitlen); + hf->bitlen = calloc(len_size, sizeof(hf->bitlen[0])); + if (hf->bitlen == NULL) + return (ARCHIVE_FATAL); + hf->len_size = (int)len_size; + } else + memset(hf->bitlen, 0, len_size * sizeof(hf->bitlen[0])); + if (hf->tbl == NULL) { + hf->tbl = malloc(((size_t)1 << tbl_bits) * sizeof(hf->tbl[0])); + if (hf->tbl == NULL) + return (ARCHIVE_FATAL); + hf->tbl_bits = tbl_bits; + } + return (ARCHIVE_OK); +} + +static void +lzx_huffman_free(struct huffman *hf) +{ + free(hf->bitlen); + free(hf->tbl); +} + +/* + * Make a huffman coding table. + */ +static int +lzx_make_huffman_table(struct huffman *hf) +{ + uint16_t *tbl; + const unsigned char *bitlen; + int bitptn[17], weight[17]; + int i, maxbits = 0, ptn, tbl_size, w; + int len_avail; + + /* + * Initialize bit patterns. + */ + ptn = 0; + for (i = 1, w = 1 << 15; i <= 16; i++, w >>= 1) { + bitptn[i] = ptn; + weight[i] = w; + if (hf->freq[i]) { + ptn += hf->freq[i] * w; + maxbits = i; + } + } + if ((ptn & 0xffff) != 0 || maxbits > hf->tbl_bits) + return (0);/* Invalid */ + + hf->max_bits = maxbits; + + /* + * Cut out extra bits which we won't house in the table. + * This preparation reduces the same calculation in the for-loop + * making the table. + */ + if (maxbits < 16) { + int ebits = 16 - maxbits; + for (i = 1; i <= maxbits; i++) { + bitptn[i] >>= ebits; + weight[i] >>= ebits; + } + } + + /* + * Make the table. + */ + tbl_size = 1 << hf->tbl_bits; + tbl = hf->tbl; + bitlen = hf->bitlen; + len_avail = hf->len_size; + hf->tree_used = 0; + for (i = 0; i < len_avail; i++) { + uint16_t *p; + int len, cnt; + + if (bitlen[i] == 0) + continue; + /* Get a bit pattern */ + len = bitlen[i]; + if (len > tbl_size) + return (0); + ptn = bitptn[len]; + cnt = weight[len]; + /* Calculate next bit pattern */ + if ((bitptn[len] = ptn + cnt) > tbl_size) + return (0);/* Invalid */ + /* Update the table */ + p = &(tbl[ptn]); + while (--cnt >= 0) + p[cnt] = (uint16_t)i; + } + return (1); +} + +static inline int +lzx_decode_huffman(struct huffman *hf, unsigned rbits) +{ + int c; + c = hf->tbl[rbits]; + if (c < hf->len_size) + return (c); + return (0); +} diff --git a/c/archive_read_support_format_cpio.c b/c/archive_read_support_format_cpio.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_format_cpio.c @@ -0,0 +1,1086 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * Copyright (c) 2010-2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_read_support_format_cpio.c 201163 2009-12-29 05:50:34Z kientzle $"); + +#ifdef HAVE_ERRNO_H +#include +#endif +/* #include */ /* See archive_platform.h */ +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_private.h" +#include "archive_read_private.h" + +#define bin_magic_offset 0 +#define bin_magic_size 2 +#define bin_dev_offset 2 +#define bin_dev_size 2 +#define bin_ino_offset 4 +#define bin_ino_size 2 +#define bin_mode_offset 6 +#define bin_mode_size 2 +#define bin_uid_offset 8 +#define bin_uid_size 2 +#define bin_gid_offset 10 +#define bin_gid_size 2 +#define bin_nlink_offset 12 +#define bin_nlink_size 2 +#define bin_rdev_offset 14 +#define bin_rdev_size 2 +#define bin_mtime_offset 16 +#define bin_mtime_size 4 +#define bin_namesize_offset 20 +#define bin_namesize_size 2 +#define bin_filesize_offset 22 +#define bin_filesize_size 4 +#define bin_header_size 26 + +#define odc_magic_offset 0 +#define odc_magic_size 6 +#define odc_dev_offset 6 +#define odc_dev_size 6 +#define odc_ino_offset 12 +#define odc_ino_size 6 +#define odc_mode_offset 18 +#define odc_mode_size 6 +#define odc_uid_offset 24 +#define odc_uid_size 6 +#define odc_gid_offset 30 +#define odc_gid_size 6 +#define odc_nlink_offset 36 +#define odc_nlink_size 6 +#define odc_rdev_offset 42 +#define odc_rdev_size 6 +#define odc_mtime_offset 48 +#define odc_mtime_size 11 +#define odc_namesize_offset 59 +#define odc_namesize_size 6 +#define odc_filesize_offset 65 +#define odc_filesize_size 11 +#define odc_header_size 76 + +#define newc_magic_offset 0 +#define newc_magic_size 6 +#define newc_ino_offset 6 +#define newc_ino_size 8 +#define newc_mode_offset 14 +#define newc_mode_size 8 +#define newc_uid_offset 22 +#define newc_uid_size 8 +#define newc_gid_offset 30 +#define newc_gid_size 8 +#define newc_nlink_offset 38 +#define newc_nlink_size 8 +#define newc_mtime_offset 46 +#define newc_mtime_size 8 +#define newc_filesize_offset 54 +#define newc_filesize_size 8 +#define newc_devmajor_offset 62 +#define newc_devmajor_size 8 +#define newc_devminor_offset 70 +#define newc_devminor_size 8 +#define newc_rdevmajor_offset 78 +#define newc_rdevmajor_size 8 +#define newc_rdevminor_offset 86 +#define newc_rdevminor_size 8 +#define newc_namesize_offset 94 +#define newc_namesize_size 8 +#define newc_checksum_offset 102 +#define newc_checksum_size 8 +#define newc_header_size 110 + +/* + * An afio large ASCII header, which they named itself. + * afio utility uses this header, if a file size is larger than 2G bytes + * or inode/uid/gid is bigger than 65535(0xFFFF) or mtime is bigger than + * 0x7fffffff, which we cannot record to odc header because of its limit. + * If not, uses odc header. + */ +#define afiol_magic_offset 0 +#define afiol_magic_size 6 +#define afiol_dev_offset 6 +#define afiol_dev_size 8 /* hex */ +#define afiol_ino_offset 14 +#define afiol_ino_size 16 /* hex */ +#define afiol_ino_m_offset 30 /* 'm' */ +#define afiol_mode_offset 31 +#define afiol_mode_size 6 /* oct */ +#define afiol_uid_offset 37 +#define afiol_uid_size 8 /* hex */ +#define afiol_gid_offset 45 +#define afiol_gid_size 8 /* hex */ +#define afiol_nlink_offset 53 +#define afiol_nlink_size 8 /* hex */ +#define afiol_rdev_offset 61 +#define afiol_rdev_size 8 /* hex */ +#define afiol_mtime_offset 69 +#define afiol_mtime_size 16 /* hex */ +#define afiol_mtime_n_offset 85 /* 'n' */ +#define afiol_namesize_offset 86 +#define afiol_namesize_size 4 /* hex */ +#define afiol_flag_offset 90 +#define afiol_flag_size 4 /* hex */ +#define afiol_xsize_offset 94 +#define afiol_xsize_size 4 /* hex */ +#define afiol_xsize_s_offset 98 /* 's' */ +#define afiol_filesize_offset 99 +#define afiol_filesize_size 16 /* hex */ +#define afiol_filesize_c_offset 115 /* ':' */ +#define afiol_header_size 116 + + +struct links_entry { + struct links_entry *next; + struct links_entry *previous; + unsigned int links; + dev_t dev; + int64_t ino; + char *name; +}; + +#define CPIO_MAGIC 0x13141516 +struct cpio { + int magic; + int (*read_header)(struct archive_read *, struct cpio *, + struct archive_entry *, size_t *, size_t *); + struct links_entry *links_head; + int64_t entry_bytes_remaining; + int64_t entry_bytes_unconsumed; + int64_t entry_offset; + int64_t entry_padding; + + struct archive_string_conv *opt_sconv; + struct archive_string_conv *sconv_default; + int init_default_conversion; +}; + +static int64_t atol16(const char *, unsigned); +static int64_t atol8(const char *, unsigned); +static int archive_read_format_cpio_bid(struct archive_read *, int); +static int archive_read_format_cpio_options(struct archive_read *, + const char *, const char *); +static int archive_read_format_cpio_cleanup(struct archive_read *); +static int archive_read_format_cpio_read_data(struct archive_read *, + const void **, size_t *, int64_t *); +static int archive_read_format_cpio_read_header(struct archive_read *, + struct archive_entry *); +static int archive_read_format_cpio_skip(struct archive_read *); +static int64_t be4(const unsigned char *); +static int find_odc_header(struct archive_read *); +static int find_newc_header(struct archive_read *); +static int header_bin_be(struct archive_read *, struct cpio *, + struct archive_entry *, size_t *, size_t *); +static int header_bin_le(struct archive_read *, struct cpio *, + struct archive_entry *, size_t *, size_t *); +static int header_newc(struct archive_read *, struct cpio *, + struct archive_entry *, size_t *, size_t *); +static int header_odc(struct archive_read *, struct cpio *, + struct archive_entry *, size_t *, size_t *); +static int header_afiol(struct archive_read *, struct cpio *, + struct archive_entry *, size_t *, size_t *); +static int is_octal(const char *, size_t); +static int is_hex(const char *, size_t); +static int64_t le4(const unsigned char *); +static int record_hardlink(struct archive_read *a, + struct cpio *cpio, struct archive_entry *entry); + +int +archive_read_support_format_cpio(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct cpio *cpio; + int r; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_format_cpio"); + + cpio = (struct cpio *)calloc(1, sizeof(*cpio)); + if (cpio == NULL) { + archive_set_error(&a->archive, ENOMEM, "Can't allocate cpio data"); + return (ARCHIVE_FATAL); + } + cpio->magic = CPIO_MAGIC; + + r = __archive_read_register_format(a, + cpio, + "cpio", + archive_read_format_cpio_bid, + archive_read_format_cpio_options, + archive_read_format_cpio_read_header, + archive_read_format_cpio_read_data, + archive_read_format_cpio_skip, + NULL, + archive_read_format_cpio_cleanup, + NULL, + NULL); + + if (r != ARCHIVE_OK) + free(cpio); + return (ARCHIVE_OK); +} + + +static int +archive_read_format_cpio_bid(struct archive_read *a, int best_bid) +{ + const unsigned char *p; + struct cpio *cpio; + int bid; + + (void)best_bid; /* UNUSED */ + + cpio = (struct cpio *)(a->format->data); + + if ((p = __archive_read_ahead(a, 6, NULL)) == NULL) + return (-1); + + bid = 0; + if (memcmp(p, "070707", 6) == 0) { + /* ASCII cpio archive (odc, POSIX.1) */ + cpio->read_header = header_odc; + bid += 48; + /* + * XXX TODO: More verification; Could check that only octal + * digits appear in appropriate header locations. XXX + */ + } else if (memcmp(p, "070727", 6) == 0) { + /* afio large ASCII cpio archive */ + cpio->read_header = header_odc; + bid += 48; + /* + * XXX TODO: More verification; Could check that almost hex + * digits appear in appropriate header locations. XXX + */ + } else if (memcmp(p, "070701", 6) == 0) { + /* ASCII cpio archive (SVR4 without CRC) */ + cpio->read_header = header_newc; + bid += 48; + /* + * XXX TODO: More verification; Could check that only hex + * digits appear in appropriate header locations. XXX + */ + } else if (memcmp(p, "070702", 6) == 0) { + /* ASCII cpio archive (SVR4 with CRC) */ + /* XXX TODO: Flag that we should check the CRC. XXX */ + cpio->read_header = header_newc; + bid += 48; + /* + * XXX TODO: More verification; Could check that only hex + * digits appear in appropriate header locations. XXX + */ + } else if (p[0] * 256 + p[1] == 070707) { + /* big-endian binary cpio archives */ + cpio->read_header = header_bin_be; + bid += 16; + /* Is more verification possible here? */ + } else if (p[0] + p[1] * 256 == 070707) { + /* little-endian binary cpio archives */ + cpio->read_header = header_bin_le; + bid += 16; + /* Is more verification possible here? */ + } else + return (ARCHIVE_WARN); + + return (bid); +} + +static int +archive_read_format_cpio_options(struct archive_read *a, + const char *key, const char *val) +{ + struct cpio *cpio; + int ret = ARCHIVE_FAILED; + + cpio = (struct cpio *)(a->format->data); + if (strcmp(key, "compat-2x") == 0) { + /* Handle filenames as libarchive 2.x */ + cpio->init_default_conversion = (val != NULL)?1:0; + return (ARCHIVE_OK); + } else if (strcmp(key, "hdrcharset") == 0) { + if (val == NULL || val[0] == 0) + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "cpio: hdrcharset option needs a character-set name"); + else { + cpio->opt_sconv = + archive_string_conversion_from_charset( + &a->archive, val, 0); + if (cpio->opt_sconv != NULL) + ret = ARCHIVE_OK; + else + ret = ARCHIVE_FATAL; + } + return (ret); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +static int +archive_read_format_cpio_read_header(struct archive_read *a, + struct archive_entry *entry) +{ + struct cpio *cpio; + const void *h, *hl; + struct archive_string_conv *sconv; + size_t namelength; + size_t name_pad; + int r; + + cpio = (struct cpio *)(a->format->data); + sconv = cpio->opt_sconv; + if (sconv == NULL) { + if (!cpio->init_default_conversion) { + cpio->sconv_default = + archive_string_default_conversion_for_read( + &(a->archive)); + cpio->init_default_conversion = 1; + } + sconv = cpio->sconv_default; + } + + r = (cpio->read_header(a, cpio, entry, &namelength, &name_pad)); + + if (r < ARCHIVE_WARN) + return (r); + + /* Read name from buffer. */ + h = __archive_read_ahead(a, namelength + name_pad, NULL); + if (h == NULL) + return (ARCHIVE_FATAL); + if (archive_entry_copy_pathname_l(entry, + (const char *)h, namelength, sconv) != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Pathname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Pathname can't be converted from %s to current locale.", + archive_string_conversion_charset_name(sconv)); + r = ARCHIVE_WARN; + } + cpio->entry_offset = 0; + + __archive_read_consume(a, namelength + name_pad); + + /* If this is a symlink, read the link contents. */ + if (archive_entry_filetype(entry) == AE_IFLNK) { + if (cpio->entry_bytes_remaining > 1024 * 1024) { + archive_set_error(&a->archive, ENOMEM, + "Rejecting malformed cpio archive: symlink contents exceed 1 megabyte"); + return (ARCHIVE_FATAL); + } + hl = __archive_read_ahead(a, + (size_t)cpio->entry_bytes_remaining, NULL); + if (hl == NULL) + return (ARCHIVE_FATAL); + if (archive_entry_copy_symlink_l(entry, (const char *)hl, + (size_t)cpio->entry_bytes_remaining, sconv) != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Linkname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Linkname can't be converted from %s to " + "current locale.", + archive_string_conversion_charset_name(sconv)); + r = ARCHIVE_WARN; + } + __archive_read_consume(a, cpio->entry_bytes_remaining); + cpio->entry_bytes_remaining = 0; + } + + /* XXX TODO: If the full mode is 0160200, then this is a Solaris + * ACL description for the following entry. Read this body + * and parse it as a Solaris-style ACL, then read the next + * header. XXX */ + + /* Compare name to "TRAILER!!!" to test for end-of-archive. */ + if (namelength == 11 && strncmp((const char *)h, "TRAILER!!!", + 11) == 0) { + /* TODO: Store file location of start of block. */ + archive_clear_error(&a->archive); + return (ARCHIVE_EOF); + } + + /* Detect and record hardlinks to previously-extracted entries. */ + if (record_hardlink(a, cpio, entry) != ARCHIVE_OK) { + return (ARCHIVE_FATAL); + } + + return (r); +} + +static int +archive_read_format_cpio_read_data(struct archive_read *a, + const void **buff, size_t *size, int64_t *offset) +{ + ssize_t bytes_read; + struct cpio *cpio; + + cpio = (struct cpio *)(a->format->data); + + if (cpio->entry_bytes_unconsumed) { + __archive_read_consume(a, cpio->entry_bytes_unconsumed); + cpio->entry_bytes_unconsumed = 0; + } + + if (cpio->entry_bytes_remaining > 0) { + *buff = __archive_read_ahead(a, 1, &bytes_read); + if (bytes_read <= 0) + return (ARCHIVE_FATAL); + if (bytes_read > cpio->entry_bytes_remaining) + bytes_read = (ssize_t)cpio->entry_bytes_remaining; + *size = bytes_read; + cpio->entry_bytes_unconsumed = bytes_read; + *offset = cpio->entry_offset; + cpio->entry_offset += bytes_read; + cpio->entry_bytes_remaining -= bytes_read; + return (ARCHIVE_OK); + } else { + if (cpio->entry_padding != + __archive_read_consume(a, cpio->entry_padding)) { + return (ARCHIVE_FATAL); + } + cpio->entry_padding = 0; + *buff = NULL; + *size = 0; + *offset = cpio->entry_offset; + return (ARCHIVE_EOF); + } +} + +static int +archive_read_format_cpio_skip(struct archive_read *a) +{ + struct cpio *cpio = (struct cpio *)(a->format->data); + int64_t to_skip = cpio->entry_bytes_remaining + cpio->entry_padding + + cpio->entry_bytes_unconsumed; + + if (to_skip != __archive_read_consume(a, to_skip)) { + return (ARCHIVE_FATAL); + } + cpio->entry_bytes_remaining = 0; + cpio->entry_padding = 0; + cpio->entry_bytes_unconsumed = 0; + return (ARCHIVE_OK); +} + +/* + * Skip forward to the next cpio newc header by searching for the + * 07070[12] string. This should be generalized and merged with + * find_odc_header below. + */ +static int +is_hex(const char *p, size_t len) +{ + while (len-- > 0) { + if ((*p >= '0' && *p <= '9') + || (*p >= 'a' && *p <= 'f') + || (*p >= 'A' && *p <= 'F')) + ++p; + else + return (0); + } + return (1); +} + +static int +find_newc_header(struct archive_read *a) +{ + const void *h; + const char *p, *q; + size_t skip, skipped = 0; + ssize_t bytes; + + for (;;) { + h = __archive_read_ahead(a, newc_header_size, &bytes); + if (h == NULL) + return (ARCHIVE_FATAL); + p = h; + q = p + bytes; + + /* Try the typical case first, then go into the slow search.*/ + if (memcmp("07070", p, 5) == 0 + && (p[5] == '1' || p[5] == '2') + && is_hex(p, newc_header_size)) + return (ARCHIVE_OK); + + /* + * Scan ahead until we find something that looks + * like a newc header. + */ + while (p + newc_header_size <= q) { + switch (p[5]) { + case '1': + case '2': + if (memcmp("07070", p, 5) == 0 + && is_hex(p, newc_header_size)) { + skip = p - (const char *)h; + __archive_read_consume(a, skip); + skipped += skip; + if (skipped > 0) { + archive_set_error(&a->archive, + 0, + "Skipped %d bytes before " + "finding valid header", + (int)skipped); + return (ARCHIVE_WARN); + } + return (ARCHIVE_OK); + } + p += 2; + break; + case '0': + p++; + break; + default: + p += 6; + break; + } + } + skip = p - (const char *)h; + __archive_read_consume(a, skip); + skipped += skip; + } +} + +static int +header_newc(struct archive_read *a, struct cpio *cpio, + struct archive_entry *entry, size_t *namelength, size_t *name_pad) +{ + const void *h; + const char *header; + int r; + + r = find_newc_header(a); + if (r < ARCHIVE_WARN) + return (r); + + /* Read fixed-size portion of header. */ + h = __archive_read_ahead(a, newc_header_size, NULL); + if (h == NULL) + return (ARCHIVE_FATAL); + + /* Parse out hex fields. */ + header = (const char *)h; + + if (memcmp(header + newc_magic_offset, "070701", 6) == 0) { + a->archive.archive_format = ARCHIVE_FORMAT_CPIO_SVR4_NOCRC; + a->archive.archive_format_name = "ASCII cpio (SVR4 with no CRC)"; + } else if (memcmp(header + newc_magic_offset, "070702", 6) == 0) { + a->archive.archive_format = ARCHIVE_FORMAT_CPIO_SVR4_CRC; + a->archive.archive_format_name = "ASCII cpio (SVR4 with CRC)"; + } else { + /* TODO: Abort here? */ + } + + archive_entry_set_devmajor(entry, + (dev_t)atol16(header + newc_devmajor_offset, newc_devmajor_size)); + archive_entry_set_devminor(entry, + (dev_t)atol16(header + newc_devminor_offset, newc_devminor_size)); + archive_entry_set_ino(entry, atol16(header + newc_ino_offset, newc_ino_size)); + archive_entry_set_mode(entry, + (mode_t)atol16(header + newc_mode_offset, newc_mode_size)); + archive_entry_set_uid(entry, atol16(header + newc_uid_offset, newc_uid_size)); + archive_entry_set_gid(entry, atol16(header + newc_gid_offset, newc_gid_size)); + archive_entry_set_nlink(entry, + (unsigned int)atol16(header + newc_nlink_offset, newc_nlink_size)); + archive_entry_set_rdevmajor(entry, + (dev_t)atol16(header + newc_rdevmajor_offset, newc_rdevmajor_size)); + archive_entry_set_rdevminor(entry, + (dev_t)atol16(header + newc_rdevminor_offset, newc_rdevminor_size)); + archive_entry_set_mtime(entry, atol16(header + newc_mtime_offset, newc_mtime_size), 0); + *namelength = (size_t)atol16(header + newc_namesize_offset, newc_namesize_size); + /* Pad name to 2 more than a multiple of 4. */ + *name_pad = (2 - *namelength) & 3; + + /* Make sure that the padded name length fits into size_t. */ + if (*name_pad > SIZE_MAX - *namelength) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "cpio archive has invalid namelength"); + return (ARCHIVE_FATAL); + } + + /* + * Note: entry_bytes_remaining is at least 64 bits and + * therefore guaranteed to be big enough for a 33-bit file + * size. + */ + cpio->entry_bytes_remaining = + atol16(header + newc_filesize_offset, newc_filesize_size); + archive_entry_set_size(entry, cpio->entry_bytes_remaining); + /* Pad file contents to a multiple of 4. */ + cpio->entry_padding = 3 & -cpio->entry_bytes_remaining; + __archive_read_consume(a, newc_header_size); + return (r); +} + +/* + * Skip forward to the next cpio odc header by searching for the + * 070707 string. This is a hand-optimized search that could + * probably be easily generalized to handle all character-based + * cpio variants. + */ +static int +is_octal(const char *p, size_t len) +{ + while (len-- > 0) { + if (*p < '0' || *p > '7') + return (0); + ++p; + } + return (1); +} + +static int +is_afio_large(const char *h, size_t len) +{ + if (len < afiol_header_size) + return (0); + if (h[afiol_ino_m_offset] != 'm' + || h[afiol_mtime_n_offset] != 'n' + || h[afiol_xsize_s_offset] != 's' + || h[afiol_filesize_c_offset] != ':') + return (0); + if (!is_hex(h + afiol_dev_offset, afiol_ino_m_offset - afiol_dev_offset)) + return (0); + if (!is_hex(h + afiol_mode_offset, afiol_mtime_n_offset - afiol_mode_offset)) + return (0); + if (!is_hex(h + afiol_namesize_offset, afiol_xsize_s_offset - afiol_namesize_offset)) + return (0); + if (!is_hex(h + afiol_filesize_offset, afiol_filesize_size)) + return (0); + return (1); +} + +static int +find_odc_header(struct archive_read *a) +{ + const void *h; + const char *p, *q; + size_t skip, skipped = 0; + ssize_t bytes; + + for (;;) { + h = __archive_read_ahead(a, odc_header_size, &bytes); + if (h == NULL) + return (ARCHIVE_FATAL); + p = h; + q = p + bytes; + + /* Try the typical case first, then go into the slow search.*/ + if (memcmp("070707", p, 6) == 0 && is_octal(p, odc_header_size)) + return (ARCHIVE_OK); + if (memcmp("070727", p, 6) == 0 && is_afio_large(p, bytes)) { + a->archive.archive_format = ARCHIVE_FORMAT_CPIO_AFIO_LARGE; + return (ARCHIVE_OK); + } + + /* + * Scan ahead until we find something that looks + * like an odc header. + */ + while (p + odc_header_size <= q) { + switch (p[5]) { + case '7': + if ((memcmp("070707", p, 6) == 0 + && is_octal(p, odc_header_size)) + || (memcmp("070727", p, 6) == 0 + && is_afio_large(p, q - p))) { + skip = p - (const char *)h; + __archive_read_consume(a, skip); + skipped += skip; + if (p[4] == '2') + a->archive.archive_format = + ARCHIVE_FORMAT_CPIO_AFIO_LARGE; + if (skipped > 0) { + archive_set_error(&a->archive, + 0, + "Skipped %d bytes before " + "finding valid header", + (int)skipped); + return (ARCHIVE_WARN); + } + return (ARCHIVE_OK); + } + p += 2; + break; + case '0': + p++; + break; + default: + p += 6; + break; + } + } + skip = p - (const char *)h; + __archive_read_consume(a, skip); + skipped += skip; + } +} + +static int +header_odc(struct archive_read *a, struct cpio *cpio, + struct archive_entry *entry, size_t *namelength, size_t *name_pad) +{ + const void *h; + int r; + const char *header; + + a->archive.archive_format = ARCHIVE_FORMAT_CPIO_POSIX; + a->archive.archive_format_name = "POSIX octet-oriented cpio"; + + /* Find the start of the next header. */ + r = find_odc_header(a); + if (r < ARCHIVE_WARN) + return (r); + + if (a->archive.archive_format == ARCHIVE_FORMAT_CPIO_AFIO_LARGE) { + int r2 = (header_afiol(a, cpio, entry, namelength, name_pad)); + if (r2 == ARCHIVE_OK) + return (r); + else + return (r2); + } + + /* Read fixed-size portion of header. */ + h = __archive_read_ahead(a, odc_header_size, NULL); + if (h == NULL) + return (ARCHIVE_FATAL); + + /* Parse out octal fields. */ + header = (const char *)h; + + archive_entry_set_dev(entry, + (dev_t)atol8(header + odc_dev_offset, odc_dev_size)); + archive_entry_set_ino(entry, atol8(header + odc_ino_offset, odc_ino_size)); + archive_entry_set_mode(entry, + (mode_t)atol8(header + odc_mode_offset, odc_mode_size)); + archive_entry_set_uid(entry, atol8(header + odc_uid_offset, odc_uid_size)); + archive_entry_set_gid(entry, atol8(header + odc_gid_offset, odc_gid_size)); + archive_entry_set_nlink(entry, + (unsigned int)atol8(header + odc_nlink_offset, odc_nlink_size)); + archive_entry_set_rdev(entry, + (dev_t)atol8(header + odc_rdev_offset, odc_rdev_size)); + archive_entry_set_mtime(entry, atol8(header + odc_mtime_offset, odc_mtime_size), 0); + *namelength = (size_t)atol8(header + odc_namesize_offset, odc_namesize_size); + *name_pad = 0; /* No padding of filename. */ + + /* + * Note: entry_bytes_remaining is at least 64 bits and + * therefore guaranteed to be big enough for a 33-bit file + * size. + */ + cpio->entry_bytes_remaining = + atol8(header + odc_filesize_offset, odc_filesize_size); + archive_entry_set_size(entry, cpio->entry_bytes_remaining); + cpio->entry_padding = 0; + __archive_read_consume(a, odc_header_size); + return (r); +} + +/* + * NOTE: if a filename suffix is ".z", it is the file gziped by afio. + * it would be nice that we can show uncompressed file size and we can + * uncompressed file contents automatically, unfortunately we have nothing + * to get a uncompressed file size while reading each header. It means + * we also cannot uncompress file contents under our framework. + */ +static int +header_afiol(struct archive_read *a, struct cpio *cpio, + struct archive_entry *entry, size_t *namelength, size_t *name_pad) +{ + const void *h; + const char *header; + + a->archive.archive_format = ARCHIVE_FORMAT_CPIO_AFIO_LARGE; + a->archive.archive_format_name = "afio large ASCII"; + + /* Read fixed-size portion of header. */ + h = __archive_read_ahead(a, afiol_header_size, NULL); + if (h == NULL) + return (ARCHIVE_FATAL); + + /* Parse out octal fields. */ + header = (const char *)h; + + archive_entry_set_dev(entry, + (dev_t)atol16(header + afiol_dev_offset, afiol_dev_size)); + archive_entry_set_ino(entry, atol16(header + afiol_ino_offset, afiol_ino_size)); + archive_entry_set_mode(entry, + (mode_t)atol8(header + afiol_mode_offset, afiol_mode_size)); + archive_entry_set_uid(entry, atol16(header + afiol_uid_offset, afiol_uid_size)); + archive_entry_set_gid(entry, atol16(header + afiol_gid_offset, afiol_gid_size)); + archive_entry_set_nlink(entry, + (unsigned int)atol16(header + afiol_nlink_offset, afiol_nlink_size)); + archive_entry_set_rdev(entry, + (dev_t)atol16(header + afiol_rdev_offset, afiol_rdev_size)); + archive_entry_set_mtime(entry, atol16(header + afiol_mtime_offset, afiol_mtime_size), 0); + *namelength = (size_t)atol16(header + afiol_namesize_offset, afiol_namesize_size); + *name_pad = 0; /* No padding of filename. */ + + cpio->entry_bytes_remaining = + atol16(header + afiol_filesize_offset, afiol_filesize_size); + archive_entry_set_size(entry, cpio->entry_bytes_remaining); + cpio->entry_padding = 0; + __archive_read_consume(a, afiol_header_size); + return (ARCHIVE_OK); +} + + +static int +header_bin_le(struct archive_read *a, struct cpio *cpio, + struct archive_entry *entry, size_t *namelength, size_t *name_pad) +{ + const void *h; + const unsigned char *header; + + a->archive.archive_format = ARCHIVE_FORMAT_CPIO_BIN_LE; + a->archive.archive_format_name = "cpio (little-endian binary)"; + + /* Read fixed-size portion of header. */ + h = __archive_read_ahead(a, bin_header_size, NULL); + if (h == NULL) { + archive_set_error(&a->archive, 0, + "End of file trying to read next cpio header"); + return (ARCHIVE_FATAL); + } + + /* Parse out binary fields. */ + header = (const unsigned char *)h; + + archive_entry_set_dev(entry, header[bin_dev_offset] + header[bin_dev_offset + 1] * 256); + archive_entry_set_ino(entry, header[bin_ino_offset] + header[bin_ino_offset + 1] * 256); + archive_entry_set_mode(entry, header[bin_mode_offset] + header[bin_mode_offset + 1] * 256); + archive_entry_set_uid(entry, header[bin_uid_offset] + header[bin_uid_offset + 1] * 256); + archive_entry_set_gid(entry, header[bin_gid_offset] + header[bin_gid_offset + 1] * 256); + archive_entry_set_nlink(entry, header[bin_nlink_offset] + header[bin_nlink_offset + 1] * 256); + archive_entry_set_rdev(entry, header[bin_rdev_offset] + header[bin_rdev_offset + 1] * 256); + archive_entry_set_mtime(entry, le4(header + bin_mtime_offset), 0); + *namelength = header[bin_namesize_offset] + header[bin_namesize_offset + 1] * 256; + *name_pad = *namelength & 1; /* Pad to even. */ + + cpio->entry_bytes_remaining = le4(header + bin_filesize_offset); + archive_entry_set_size(entry, cpio->entry_bytes_remaining); + cpio->entry_padding = cpio->entry_bytes_remaining & 1; /* Pad to even. */ + __archive_read_consume(a, bin_header_size); + return (ARCHIVE_OK); +} + +static int +header_bin_be(struct archive_read *a, struct cpio *cpio, + struct archive_entry *entry, size_t *namelength, size_t *name_pad) +{ + const void *h; + const unsigned char *header; + + a->archive.archive_format = ARCHIVE_FORMAT_CPIO_BIN_BE; + a->archive.archive_format_name = "cpio (big-endian binary)"; + + /* Read fixed-size portion of header. */ + h = __archive_read_ahead(a, bin_header_size, NULL); + if (h == NULL) { + archive_set_error(&a->archive, 0, + "End of file trying to read next cpio header"); + return (ARCHIVE_FATAL); + } + + /* Parse out binary fields. */ + header = (const unsigned char *)h; + + archive_entry_set_dev(entry, header[bin_dev_offset] * 256 + header[bin_dev_offset + 1]); + archive_entry_set_ino(entry, header[bin_ino_offset] * 256 + header[bin_ino_offset + 1]); + archive_entry_set_mode(entry, header[bin_mode_offset] * 256 + header[bin_mode_offset + 1]); + archive_entry_set_uid(entry, header[bin_uid_offset] * 256 + header[bin_uid_offset + 1]); + archive_entry_set_gid(entry, header[bin_gid_offset] * 256 + header[bin_gid_offset + 1]); + archive_entry_set_nlink(entry, header[bin_nlink_offset] * 256 + header[bin_nlink_offset + 1]); + archive_entry_set_rdev(entry, header[bin_rdev_offset] * 256 + header[bin_rdev_offset + 1]); + archive_entry_set_mtime(entry, be4(header + bin_mtime_offset), 0); + *namelength = header[bin_namesize_offset] * 256 + header[bin_namesize_offset + 1]; + *name_pad = *namelength & 1; /* Pad to even. */ + + cpio->entry_bytes_remaining = be4(header + bin_filesize_offset); + archive_entry_set_size(entry, cpio->entry_bytes_remaining); + cpio->entry_padding = cpio->entry_bytes_remaining & 1; /* Pad to even. */ + __archive_read_consume(a, bin_header_size); + return (ARCHIVE_OK); +} + +static int +archive_read_format_cpio_cleanup(struct archive_read *a) +{ + struct cpio *cpio; + + cpio = (struct cpio *)(a->format->data); + /* Free inode->name map */ + while (cpio->links_head != NULL) { + struct links_entry *lp = cpio->links_head->next; + + free(cpio->links_head->name); + free(cpio->links_head); + cpio->links_head = lp; + } + free(cpio); + (a->format->data) = NULL; + return (ARCHIVE_OK); +} + +static int64_t +le4(const unsigned char *p) +{ + return ((p[0] << 16) + (((int64_t)p[1]) << 24) + (p[2] << 0) + (p[3] << 8)); +} + + +static int64_t +be4(const unsigned char *p) +{ + return ((((int64_t)p[0]) << 24) + (p[1] << 16) + (p[2] << 8) + (p[3])); +} + +/* + * Note that this implementation does not (and should not!) obey + * locale settings; you cannot simply substitute strtol here, since + * it does obey locale. + */ +static int64_t +atol8(const char *p, unsigned char_cnt) +{ + int64_t l; + int digit; + + l = 0; + while (char_cnt-- > 0) { + if (*p >= '0' && *p <= '7') + digit = *p - '0'; + else + return (l); + p++; + l <<= 3; + l |= digit; + } + return (l); +} + +static int64_t +atol16(const char *p, unsigned char_cnt) +{ + int64_t l; + int digit; + + l = 0; + while (char_cnt-- > 0) { + if (*p >= 'a' && *p <= 'f') + digit = *p - 'a' + 10; + else if (*p >= 'A' && *p <= 'F') + digit = *p - 'A' + 10; + else if (*p >= '0' && *p <= '9') + digit = *p - '0'; + else + return (l); + p++; + l <<= 4; + l |= digit; + } + return (l); +} + +static int +record_hardlink(struct archive_read *a, + struct cpio *cpio, struct archive_entry *entry) +{ + struct links_entry *le; + dev_t dev; + int64_t ino; + + if (archive_entry_nlink(entry) <= 1) + return (ARCHIVE_OK); + + dev = archive_entry_dev(entry); + ino = archive_entry_ino64(entry); + + /* + * First look in the list of multiply-linked files. If we've + * already dumped it, convert this entry to a hard link entry. + */ + for (le = cpio->links_head; le; le = le->next) { + if (le->dev == dev && le->ino == ino) { + archive_entry_copy_hardlink(entry, le->name); + + if (--le->links <= 0) { + if (le->previous != NULL) + le->previous->next = le->next; + if (le->next != NULL) + le->next->previous = le->previous; + if (cpio->links_head == le) + cpio->links_head = le->next; + free(le->name); + free(le); + } + + return (ARCHIVE_OK); + } + } + + le = (struct links_entry *)malloc(sizeof(struct links_entry)); + if (le == NULL) { + archive_set_error(&a->archive, + ENOMEM, "Out of memory adding file to list"); + return (ARCHIVE_FATAL); + } + if (cpio->links_head != NULL) + cpio->links_head->previous = le; + le->next = cpio->links_head; + le->previous = NULL; + cpio->links_head = le; + le->dev = dev; + le->ino = ino; + le->links = archive_entry_nlink(entry) - 1; + le->name = strdup(archive_entry_pathname(entry)); + if (le->name == NULL) { + archive_set_error(&a->archive, + ENOMEM, "Out of memory adding file to list"); + return (ARCHIVE_FATAL); + } + + return (ARCHIVE_OK); +} diff --git a/c/archive_read_support_format_empty.c b/c/archive_read_support_format_empty.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_format_empty.c @@ -0,0 +1,96 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_read_support_format_empty.c 191524 2009-04-26 18:24:14Z kientzle $"); + +#include "archive.h" +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_read_private.h" + +static int archive_read_format_empty_bid(struct archive_read *, int); +static int archive_read_format_empty_read_data(struct archive_read *, + const void **, size_t *, int64_t *); +static int archive_read_format_empty_read_header(struct archive_read *, + struct archive_entry *); +int +archive_read_support_format_empty(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + int r; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_format_empty"); + + r = __archive_read_register_format(a, + NULL, + NULL, + archive_read_format_empty_bid, + NULL, + archive_read_format_empty_read_header, + archive_read_format_empty_read_data, + NULL, + NULL, + NULL, + NULL, + NULL); + + return (r); +} + + +static int +archive_read_format_empty_bid(struct archive_read *a, int best_bid) +{ + if (best_bid < 1 && __archive_read_ahead(a, 1, NULL) == NULL) + return (1); + return (-1); +} + +static int +archive_read_format_empty_read_header(struct archive_read *a, + struct archive_entry *entry) +{ + (void)a; /* UNUSED */ + (void)entry; /* UNUSED */ + + a->archive.archive_format = ARCHIVE_FORMAT_EMPTY; + a->archive.archive_format_name = "Empty file"; + + return (ARCHIVE_EOF); +} + +static int +archive_read_format_empty_read_data(struct archive_read *a, + const void **buff, size_t *size, int64_t *offset) +{ + (void)a; /* UNUSED */ + (void)buff; /* UNUSED */ + (void)size; /* UNUSED */ + (void)offset; /* UNUSED */ + + return (ARCHIVE_EOF); +} diff --git a/c/archive_read_support_format_iso9660.c b/c/archive_read_support_format_iso9660.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_format_iso9660.c @@ -0,0 +1,3278 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * Copyright (c) 2009 Andreas Henriksson + * Copyright (c) 2009-2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_read_support_format_iso9660.c 201246 2009-12-30 05:30:35Z kientzle $"); + +#ifdef HAVE_ERRNO_H +#include +#endif +/* #include */ /* See archive_platform.h */ +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#include +#ifdef HAVE_ZLIB_H +#include +#endif + +#include "archive.h" +#include "archive_endian.h" +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_private.h" +#include "archive_read_private.h" +#include "archive_string.h" + +/* + * An overview of ISO 9660 format: + * + * Each disk is laid out as follows: + * * 32k reserved for private use + * * Volume descriptor table. Each volume descriptor + * is 2k and specifies basic format information. + * The "Primary Volume Descriptor" (PVD) is defined by the + * standard and should always be present; other volume + * descriptors include various vendor-specific extensions. + * * Files and directories. Each file/dir is specified by + * an "extent" (starting sector and length in bytes). + * Dirs are just files with directory records packed one + * after another. The PVD contains a single dir entry + * specifying the location of the root directory. Everything + * else follows from there. + * + * This module works by first reading the volume descriptors, then + * building a list of directory entries, sorted by starting + * sector. At each step, I look for the earliest dir entry that + * hasn't yet been read, seek forward to that location and read + * that entry. If it's a dir, I slurp in the new dir entries and + * add them to the heap; if it's a regular file, I return the + * corresponding archive_entry and wait for the client to request + * the file body. This strategy allows us to read most compliant + * CDs with a single pass through the data, as required by libarchive. + */ +#define LOGICAL_BLOCK_SIZE 2048 +#define SYSTEM_AREA_BLOCK 16 + +/* Structure of on-disk primary volume descriptor. */ +#define PVD_type_offset 0 +#define PVD_type_size 1 +#define PVD_id_offset (PVD_type_offset + PVD_type_size) +#define PVD_id_size 5 +#define PVD_version_offset (PVD_id_offset + PVD_id_size) +#define PVD_version_size 1 +#define PVD_reserved1_offset (PVD_version_offset + PVD_version_size) +#define PVD_reserved1_size 1 +#define PVD_system_id_offset (PVD_reserved1_offset + PVD_reserved1_size) +#define PVD_system_id_size 32 +#define PVD_volume_id_offset (PVD_system_id_offset + PVD_system_id_size) +#define PVD_volume_id_size 32 +#define PVD_reserved2_offset (PVD_volume_id_offset + PVD_volume_id_size) +#define PVD_reserved2_size 8 +#define PVD_volume_space_size_offset (PVD_reserved2_offset + PVD_reserved2_size) +#define PVD_volume_space_size_size 8 +#define PVD_reserved3_offset (PVD_volume_space_size_offset + PVD_volume_space_size_size) +#define PVD_reserved3_size 32 +#define PVD_volume_set_size_offset (PVD_reserved3_offset + PVD_reserved3_size) +#define PVD_volume_set_size_size 4 +#define PVD_volume_sequence_number_offset (PVD_volume_set_size_offset + PVD_volume_set_size_size) +#define PVD_volume_sequence_number_size 4 +#define PVD_logical_block_size_offset (PVD_volume_sequence_number_offset + PVD_volume_sequence_number_size) +#define PVD_logical_block_size_size 4 +#define PVD_path_table_size_offset (PVD_logical_block_size_offset + PVD_logical_block_size_size) +#define PVD_path_table_size_size 8 +#define PVD_type_1_path_table_offset (PVD_path_table_size_offset + PVD_path_table_size_size) +#define PVD_type_1_path_table_size 4 +#define PVD_opt_type_1_path_table_offset (PVD_type_1_path_table_offset + PVD_type_1_path_table_size) +#define PVD_opt_type_1_path_table_size 4 +#define PVD_type_m_path_table_offset (PVD_opt_type_1_path_table_offset + PVD_opt_type_1_path_table_size) +#define PVD_type_m_path_table_size 4 +#define PVD_opt_type_m_path_table_offset (PVD_type_m_path_table_offset + PVD_type_m_path_table_size) +#define PVD_opt_type_m_path_table_size 4 +#define PVD_root_directory_record_offset (PVD_opt_type_m_path_table_offset + PVD_opt_type_m_path_table_size) +#define PVD_root_directory_record_size 34 +#define PVD_volume_set_id_offset (PVD_root_directory_record_offset + PVD_root_directory_record_size) +#define PVD_volume_set_id_size 128 +#define PVD_publisher_id_offset (PVD_volume_set_id_offset + PVD_volume_set_id_size) +#define PVD_publisher_id_size 128 +#define PVD_preparer_id_offset (PVD_publisher_id_offset + PVD_publisher_id_size) +#define PVD_preparer_id_size 128 +#define PVD_application_id_offset (PVD_preparer_id_offset + PVD_preparer_id_size) +#define PVD_application_id_size 128 +#define PVD_copyright_file_id_offset (PVD_application_id_offset + PVD_application_id_size) +#define PVD_copyright_file_id_size 37 +#define PVD_abstract_file_id_offset (PVD_copyright_file_id_offset + PVD_copyright_file_id_size) +#define PVD_abstract_file_id_size 37 +#define PVD_bibliographic_file_id_offset (PVD_abstract_file_id_offset + PVD_abstract_file_id_size) +#define PVD_bibliographic_file_id_size 37 +#define PVD_creation_date_offset (PVD_bibliographic_file_id_offset + PVD_bibliographic_file_id_size) +#define PVD_creation_date_size 17 +#define PVD_modification_date_offset (PVD_creation_date_offset + PVD_creation_date_size) +#define PVD_modification_date_size 17 +#define PVD_expiration_date_offset (PVD_modification_date_offset + PVD_modification_date_size) +#define PVD_expiration_date_size 17 +#define PVD_effective_date_offset (PVD_expiration_date_offset + PVD_expiration_date_size) +#define PVD_effective_date_size 17 +#define PVD_file_structure_version_offset (PVD_effective_date_offset + PVD_effective_date_size) +#define PVD_file_structure_version_size 1 +#define PVD_reserved4_offset (PVD_file_structure_version_offset + PVD_file_structure_version_size) +#define PVD_reserved4_size 1 +#define PVD_application_data_offset (PVD_reserved4_offset + PVD_reserved4_size) +#define PVD_application_data_size 512 +#define PVD_reserved5_offset (PVD_application_data_offset + PVD_application_data_size) +#define PVD_reserved5_size (2048 - PVD_reserved5_offset) + +/* TODO: It would make future maintenance easier to just hardcode the + * above values. In particular, ECMA119 states the offsets as part of + * the standard. That would eliminate the need for the following check.*/ +#if PVD_reserved5_offset != 1395 +#error PVD offset and size definitions are wrong. +#endif + + +/* Structure of optional on-disk supplementary volume descriptor. */ +#define SVD_type_offset 0 +#define SVD_type_size 1 +#define SVD_id_offset (SVD_type_offset + SVD_type_size) +#define SVD_id_size 5 +#define SVD_version_offset (SVD_id_offset + SVD_id_size) +#define SVD_version_size 1 +/* ... */ +#define SVD_reserved1_offset 72 +#define SVD_reserved1_size 8 +#define SVD_volume_space_size_offset 80 +#define SVD_volume_space_size_size 8 +#define SVD_escape_sequences_offset (SVD_volume_space_size_offset + SVD_volume_space_size_size) +#define SVD_escape_sequences_size 32 +/* ... */ +#define SVD_logical_block_size_offset 128 +#define SVD_logical_block_size_size 4 +#define SVD_type_L_path_table_offset 140 +#define SVD_type_M_path_table_offset 148 +/* ... */ +#define SVD_root_directory_record_offset 156 +#define SVD_root_directory_record_size 34 +#define SVD_file_structure_version_offset 881 +#define SVD_reserved2_offset 882 +#define SVD_reserved2_size 1 +#define SVD_reserved3_offset 1395 +#define SVD_reserved3_size 653 +/* ... */ +/* FIXME: validate correctness of last SVD entry offset. */ + +/* Structure of an on-disk directory record. */ +/* Note: ISO9660 stores each multi-byte integer twice, once in + * each byte order. The sizes here are the size of just one + * of the two integers. (This is why the offset of a field isn't + * the same as the offset+size of the previous field.) */ +#define DR_length_offset 0 +#define DR_length_size 1 +#define DR_ext_attr_length_offset 1 +#define DR_ext_attr_length_size 1 +#define DR_extent_offset 2 +#define DR_extent_size 4 +#define DR_size_offset 10 +#define DR_size_size 4 +#define DR_date_offset 18 +#define DR_date_size 7 +#define DR_flags_offset 25 +#define DR_flags_size 1 +#define DR_file_unit_size_offset 26 +#define DR_file_unit_size_size 1 +#define DR_interleave_offset 27 +#define DR_interleave_size 1 +#define DR_volume_sequence_number_offset 28 +#define DR_volume_sequence_number_size 2 +#define DR_name_len_offset 32 +#define DR_name_len_size 1 +#define DR_name_offset 33 + +#ifdef HAVE_ZLIB_H +static const unsigned char zisofs_magic[8] = { + 0x37, 0xE4, 0x53, 0x96, 0xC9, 0xDB, 0xD6, 0x07 +}; + +struct zisofs { + /* Set 1 if this file compressed by paged zlib */ + int pz; + int pz_log2_bs; /* Log2 of block size */ + uint64_t pz_uncompressed_size; + + int initialized; + unsigned char *uncompressed_buffer; + size_t uncompressed_buffer_size; + + uint32_t pz_offset; + unsigned char header[16]; + size_t header_avail; + int header_passed; + unsigned char *block_pointers; + size_t block_pointers_alloc; + size_t block_pointers_size; + size_t block_pointers_avail; + size_t block_off; + uint32_t block_avail; + + z_stream stream; + int stream_valid; +}; +#else +struct zisofs { + /* Set 1 if this file compressed by paged zlib */ + int pz; +}; +#endif + +struct content { + uint64_t offset;/* Offset on disk. */ + uint64_t size; /* File size in bytes. */ + struct content *next; +}; + +/* In-memory storage for a directory record. */ +struct file_info { + struct file_info *use_next; + struct file_info *parent; + struct file_info *next; + struct file_info *re_next; + int subdirs; + uint64_t key; /* Heap Key. */ + uint64_t offset; /* Offset on disk. */ + uint64_t size; /* File size in bytes. */ + uint32_t ce_offset; /* Offset of CE. */ + uint32_t ce_size; /* Size of CE. */ + char rr_moved; /* Flag to rr_moved. */ + char rr_moved_has_re_only; + char re; /* Having RRIP "RE" extension. */ + char re_descendant; + uint64_t cl_offset; /* Having RRIP "CL" extension. */ + int birthtime_is_set; + time_t birthtime; /* File created time. */ + time_t mtime; /* File last modified time. */ + time_t atime; /* File last accessed time. */ + time_t ctime; /* File attribute change time. */ + uint64_t rdev; /* Device number. */ + mode_t mode; + uid_t uid; + gid_t gid; + int64_t number; + int nlinks; + struct archive_string name; /* Pathname */ + unsigned char *utf16be_name; + size_t utf16be_bytes; + char name_continues; /* Non-zero if name continues */ + struct archive_string symlink; + char symlink_continues; /* Non-zero if link continues */ + /* Set 1 if this file compressed by paged zlib(zisofs) */ + int pz; + int pz_log2_bs; /* Log2 of block size */ + uint64_t pz_uncompressed_size; + /* Set 1 if this file is multi extent. */ + int multi_extent; + struct { + struct content *first; + struct content **last; + } contents; + struct { + struct file_info *first; + struct file_info **last; + } rede_files; +}; + +struct heap_queue { + struct file_info **files; + int allocated; + int used; +}; + +struct iso9660 { + int magic; +#define ISO9660_MAGIC 0x96609660 + + int opt_support_joliet; + int opt_support_rockridge; + + struct archive_string pathname; + char seenRockridge; /* Set true if RR extensions are used. */ + char seenSUSP; /* Set true if SUSP is being used. */ + char seenJoliet; + + unsigned char suspOffset; + struct file_info *rr_moved; + struct read_ce_queue { + struct read_ce_req { + uint64_t offset;/* Offset of CE on disk. */ + struct file_info *file; + } *reqs; + int cnt; + int allocated; + } read_ce_req; + + int64_t previous_number; + struct archive_string previous_pathname; + + struct file_info *use_files; + struct heap_queue pending_files; + struct { + struct file_info *first; + struct file_info **last; + } cache_files; + struct { + struct file_info *first; + struct file_info **last; + } re_files; + + uint64_t current_position; + ssize_t logical_block_size; + uint64_t volume_size; /* Total size of volume in bytes. */ + int32_t volume_block;/* Total size of volume in logical blocks. */ + + struct vd { + int location; /* Location of Extent. */ + uint32_t size; + } primary, joliet; + + int64_t entry_sparse_offset; + int64_t entry_bytes_remaining; + size_t entry_bytes_unconsumed; + struct zisofs entry_zisofs; + struct content *entry_content; + struct archive_string_conv *sconv_utf16be; + /* + * Buffers for a full pathname in UTF-16BE in Joliet extensions. + */ +#define UTF16_NAME_MAX 1024 + unsigned char *utf16be_path; + size_t utf16be_path_len; + unsigned char *utf16be_previous_path; + size_t utf16be_previous_path_len; + /* Null buffer used in bidder to improve its performance. */ + unsigned char null[2048]; +}; + +static int archive_read_format_iso9660_bid(struct archive_read *, int); +static int archive_read_format_iso9660_options(struct archive_read *, + const char *, const char *); +static int archive_read_format_iso9660_cleanup(struct archive_read *); +static int archive_read_format_iso9660_read_data(struct archive_read *, + const void **, size_t *, int64_t *); +static int archive_read_format_iso9660_read_data_skip(struct archive_read *); +static int archive_read_format_iso9660_read_header(struct archive_read *, + struct archive_entry *); +static const char *build_pathname(struct archive_string *, struct file_info *, int); +static int build_pathname_utf16be(unsigned char *, size_t, size_t *, + struct file_info *); +#if DEBUG +static void dump_isodirrec(FILE *, const unsigned char *isodirrec); +#endif +static time_t time_from_tm(struct tm *); +static time_t isodate17(const unsigned char *); +static time_t isodate7(const unsigned char *); +static int isBootRecord(struct iso9660 *, const unsigned char *); +static int isVolumePartition(struct iso9660 *, const unsigned char *); +static int isVDSetTerminator(struct iso9660 *, const unsigned char *); +static int isJolietSVD(struct iso9660 *, const unsigned char *); +static int isSVD(struct iso9660 *, const unsigned char *); +static int isEVD(struct iso9660 *, const unsigned char *); +static int isPVD(struct iso9660 *, const unsigned char *); +static int next_cache_entry(struct archive_read *, struct iso9660 *, + struct file_info **); +static int next_entry_seek(struct archive_read *, struct iso9660 *, + struct file_info **); +static struct file_info * + parse_file_info(struct archive_read *a, + struct file_info *parent, const unsigned char *isodirrec, + size_t reclen); +static int parse_rockridge(struct archive_read *a, + struct file_info *file, const unsigned char *start, + const unsigned char *end); +static int register_CE(struct archive_read *a, int32_t location, + struct file_info *file); +static int read_CE(struct archive_read *a, struct iso9660 *iso9660); +static void parse_rockridge_NM1(struct file_info *, + const unsigned char *, int); +static void parse_rockridge_SL1(struct file_info *, + const unsigned char *, int); +static void parse_rockridge_TF1(struct file_info *, + const unsigned char *, int); +static void parse_rockridge_ZF1(struct file_info *, + const unsigned char *, int); +static void register_file(struct iso9660 *, struct file_info *); +static void release_files(struct iso9660 *); +static unsigned toi(const void *p, int n); +static inline void re_add_entry(struct iso9660 *, struct file_info *); +static inline struct file_info * re_get_entry(struct iso9660 *); +static inline int rede_add_entry(struct file_info *); +static inline struct file_info * rede_get_entry(struct file_info *); +static inline void cache_add_entry(struct iso9660 *iso9660, + struct file_info *file); +static inline struct file_info *cache_get_entry(struct iso9660 *iso9660); +static int heap_add_entry(struct archive_read *a, struct heap_queue *heap, + struct file_info *file, uint64_t key); +static struct file_info *heap_get_entry(struct heap_queue *heap); + +#define add_entry(arch, iso9660, file) \ + heap_add_entry(arch, &((iso9660)->pending_files), file, file->offset) +#define next_entry(iso9660) \ + heap_get_entry(&((iso9660)->pending_files)) + +int +archive_read_support_format_iso9660(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct iso9660 *iso9660; + int r; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_format_iso9660"); + + iso9660 = (struct iso9660 *)calloc(1, sizeof(*iso9660)); + if (iso9660 == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate iso9660 data"); + return (ARCHIVE_FATAL); + } + iso9660->magic = ISO9660_MAGIC; + iso9660->cache_files.first = NULL; + iso9660->cache_files.last = &(iso9660->cache_files.first); + iso9660->re_files.first = NULL; + iso9660->re_files.last = &(iso9660->re_files.first); + /* Enable to support Joliet extensions by default. */ + iso9660->opt_support_joliet = 1; + /* Enable to support Rock Ridge extensions by default. */ + iso9660->opt_support_rockridge = 1; + + r = __archive_read_register_format(a, + iso9660, + "iso9660", + archive_read_format_iso9660_bid, + archive_read_format_iso9660_options, + archive_read_format_iso9660_read_header, + archive_read_format_iso9660_read_data, + archive_read_format_iso9660_read_data_skip, + NULL, + archive_read_format_iso9660_cleanup, + NULL, + NULL); + + if (r != ARCHIVE_OK) { + free(iso9660); + return (r); + } + return (ARCHIVE_OK); +} + + +static int +archive_read_format_iso9660_bid(struct archive_read *a, int best_bid) +{ + struct iso9660 *iso9660; + ssize_t bytes_read; + const unsigned char *p; + int seenTerminator; + + /* If there's already a better bid than we can ever + make, don't bother testing. */ + if (best_bid > 48) + return (-1); + + iso9660 = (struct iso9660 *)(a->format->data); + + /* + * Skip the first 32k (reserved area) and get the first + * 8 sectors of the volume descriptor table. Of course, + * if the I/O layer gives us more, we'll take it. + */ +#define RESERVED_AREA (SYSTEM_AREA_BLOCK * LOGICAL_BLOCK_SIZE) + p = __archive_read_ahead(a, + RESERVED_AREA + 8 * LOGICAL_BLOCK_SIZE, + &bytes_read); + if (p == NULL) + return (-1); + + /* Skip the reserved area. */ + bytes_read -= RESERVED_AREA; + p += RESERVED_AREA; + + /* Check each volume descriptor. */ + seenTerminator = 0; + for (; bytes_read > LOGICAL_BLOCK_SIZE; + bytes_read -= LOGICAL_BLOCK_SIZE, p += LOGICAL_BLOCK_SIZE) { + /* Do not handle undefined Volume Descriptor Type. */ + if (p[0] >= 4 && p[0] <= 254) + return (0); + /* Standard Identifier must be "CD001" */ + if (memcmp(p + 1, "CD001", 5) != 0) + return (0); + if (isPVD(iso9660, p)) + continue; + if (!iso9660->joliet.location) { + if (isJolietSVD(iso9660, p)) + continue; + } + if (isBootRecord(iso9660, p)) + continue; + if (isEVD(iso9660, p)) + continue; + if (isSVD(iso9660, p)) + continue; + if (isVolumePartition(iso9660, p)) + continue; + if (isVDSetTerminator(iso9660, p)) { + seenTerminator = 1; + break; + } + return (0); + } + /* + * ISO 9660 format must have Primary Volume Descriptor and + * Volume Descriptor Set Terminator. + */ + if (seenTerminator && iso9660->primary.location > 16) + return (48); + + /* We didn't find a valid PVD; return a bid of zero. */ + return (0); +} + +static int +archive_read_format_iso9660_options(struct archive_read *a, + const char *key, const char *val) +{ + struct iso9660 *iso9660; + + iso9660 = (struct iso9660 *)(a->format->data); + + if (strcmp(key, "joliet") == 0) { + if (val == NULL || strcmp(val, "off") == 0 || + strcmp(val, "ignore") == 0 || + strcmp(val, "disable") == 0 || + strcmp(val, "0") == 0) + iso9660->opt_support_joliet = 0; + else + iso9660->opt_support_joliet = 1; + return (ARCHIVE_OK); + } + if (strcmp(key, "rockridge") == 0 || + strcmp(key, "Rockridge") == 0) { + iso9660->opt_support_rockridge = val != NULL; + return (ARCHIVE_OK); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +static int +isNull(struct iso9660 *iso9660, const unsigned char *h, unsigned offset, +unsigned bytes) +{ + + while (bytes >= sizeof(iso9660->null)) { + if (!memcmp(iso9660->null, h + offset, sizeof(iso9660->null))) + return (0); + offset += sizeof(iso9660->null); + bytes -= sizeof(iso9660->null); + } + if (bytes) + return memcmp(iso9660->null, h + offset, bytes) == 0; + else + return (1); +} + +static int +isBootRecord(struct iso9660 *iso9660, const unsigned char *h) +{ + (void)iso9660; /* UNUSED */ + + /* Type of the Volume Descriptor Boot Record must be 0. */ + if (h[0] != 0) + return (0); + + /* Volume Descriptor Version must be 1. */ + if (h[6] != 1) + return (0); + + return (1); +} + +static int +isVolumePartition(struct iso9660 *iso9660, const unsigned char *h) +{ + int32_t location; + + /* Type of the Volume Partition Descriptor must be 3. */ + if (h[0] != 3) + return (0); + + /* Volume Descriptor Version must be 1. */ + if (h[6] != 1) + return (0); + /* Unused Field */ + if (h[7] != 0) + return (0); + + location = archive_le32dec(h + 72); + if (location <= SYSTEM_AREA_BLOCK || + location >= iso9660->volume_block) + return (0); + if ((uint32_t)location != archive_be32dec(h + 76)) + return (0); + + return (1); +} + +static int +isVDSetTerminator(struct iso9660 *iso9660, const unsigned char *h) +{ + (void)iso9660; /* UNUSED */ + + /* Type of the Volume Descriptor Set Terminator must be 255. */ + if (h[0] != 255) + return (0); + + /* Volume Descriptor Version must be 1. */ + if (h[6] != 1) + return (0); + + /* Reserved field must be 0. */ + if (!isNull(iso9660, h, 7, 2048-7)) + return (0); + + return (1); +} + +static int +isJolietSVD(struct iso9660 *iso9660, const unsigned char *h) +{ + const unsigned char *p; + ssize_t logical_block_size; + int32_t volume_block; + + /* Check if current sector is a kind of Supplementary Volume + * Descriptor. */ + if (!isSVD(iso9660, h)) + return (0); + + /* FIXME: do more validations according to joliet spec. */ + + /* check if this SVD contains joliet extension! */ + p = h + SVD_escape_sequences_offset; + /* N.B. Joliet spec says p[1] == '\\', but.... */ + if (p[0] == '%' && p[1] == '/') { + int level = 0; + + if (p[2] == '@') + level = 1; + else if (p[2] == 'C') + level = 2; + else if (p[2] == 'E') + level = 3; + else /* not joliet */ + return (0); + + iso9660->seenJoliet = level; + + } else /* not joliet */ + return (0); + + logical_block_size = + archive_le16dec(h + SVD_logical_block_size_offset); + volume_block = archive_le32dec(h + SVD_volume_space_size_offset); + + iso9660->logical_block_size = logical_block_size; + iso9660->volume_block = volume_block; + iso9660->volume_size = logical_block_size * (uint64_t)volume_block; + /* Read Root Directory Record in Volume Descriptor. */ + p = h + SVD_root_directory_record_offset; + iso9660->joliet.location = archive_le32dec(p + DR_extent_offset); + iso9660->joliet.size = archive_le32dec(p + DR_size_offset); + + return (48); +} + +static int +isSVD(struct iso9660 *iso9660, const unsigned char *h) +{ + const unsigned char *p; + ssize_t logical_block_size; + int32_t volume_block; + int32_t location; + + (void)iso9660; /* UNUSED */ + + /* Type 2 means it's a SVD. */ + if (h[SVD_type_offset] != 2) + return (0); + + /* Reserved field must be 0. */ + if (!isNull(iso9660, h, SVD_reserved1_offset, SVD_reserved1_size)) + return (0); + if (!isNull(iso9660, h, SVD_reserved2_offset, SVD_reserved2_size)) + return (0); + if (!isNull(iso9660, h, SVD_reserved3_offset, SVD_reserved3_size)) + return (0); + + /* File structure version must be 1 for ISO9660/ECMA119. */ + if (h[SVD_file_structure_version_offset] != 1) + return (0); + + logical_block_size = + archive_le16dec(h + SVD_logical_block_size_offset); + if (logical_block_size <= 0) + return (0); + + volume_block = archive_le32dec(h + SVD_volume_space_size_offset); + if (volume_block <= SYSTEM_AREA_BLOCK+4) + return (0); + + /* Location of Occurrence of Type L Path Table must be + * available location, + * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ + location = archive_le32dec(h+SVD_type_L_path_table_offset); + if (location < SYSTEM_AREA_BLOCK+2 || location >= volume_block) + return (0); + + /* The Type M Path Table must be at a valid location (WinISO + * and probably other programs omit this, so we allow zero) + * + * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ + location = archive_be32dec(h+SVD_type_M_path_table_offset); + if ((location > 0 && location < SYSTEM_AREA_BLOCK+2) + || location >= volume_block) + return (0); + + /* Read Root Directory Record in Volume Descriptor. */ + p = h + SVD_root_directory_record_offset; + if (p[DR_length_offset] != 34) + return (0); + + return (48); +} + +static int +isEVD(struct iso9660 *iso9660, const unsigned char *h) +{ + const unsigned char *p; + ssize_t logical_block_size; + int32_t volume_block; + int32_t location; + + (void)iso9660; /* UNUSED */ + + /* Type of the Enhanced Volume Descriptor must be 2. */ + if (h[PVD_type_offset] != 2) + return (0); + + /* EVD version must be 2. */ + if (h[PVD_version_offset] != 2) + return (0); + + /* Reserved field must be 0. */ + if (h[PVD_reserved1_offset] != 0) + return (0); + + /* Reserved field must be 0. */ + if (!isNull(iso9660, h, PVD_reserved2_offset, PVD_reserved2_size)) + return (0); + + /* Reserved field must be 0. */ + if (!isNull(iso9660, h, PVD_reserved3_offset, PVD_reserved3_size)) + return (0); + + /* Logical block size must be > 0. */ + /* I've looked at Ecma 119 and can't find any stronger + * restriction on this field. */ + logical_block_size = + archive_le16dec(h + PVD_logical_block_size_offset); + if (logical_block_size <= 0) + return (0); + + volume_block = + archive_le32dec(h + PVD_volume_space_size_offset); + if (volume_block <= SYSTEM_AREA_BLOCK+4) + return (0); + + /* File structure version must be 2 for ISO9660:1999. */ + if (h[PVD_file_structure_version_offset] != 2) + return (0); + + /* Location of Occurrence of Type L Path Table must be + * available location, + * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ + location = archive_le32dec(h+PVD_type_1_path_table_offset); + if (location < SYSTEM_AREA_BLOCK+2 || location >= volume_block) + return (0); + + /* Location of Occurrence of Type M Path Table must be + * available location, + * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ + location = archive_be32dec(h+PVD_type_m_path_table_offset); + if ((location > 0 && location < SYSTEM_AREA_BLOCK+2) + || location >= volume_block) + return (0); + + /* Reserved field must be 0. */ + if (!isNull(iso9660, h, PVD_reserved4_offset, PVD_reserved4_size)) + return (0); + + /* Reserved field must be 0. */ + if (!isNull(iso9660, h, PVD_reserved5_offset, PVD_reserved5_size)) + return (0); + + /* Read Root Directory Record in Volume Descriptor. */ + p = h + PVD_root_directory_record_offset; + if (p[DR_length_offset] != 34) + return (0); + + return (48); +} + +static int +isPVD(struct iso9660 *iso9660, const unsigned char *h) +{ + const unsigned char *p; + ssize_t logical_block_size; + int32_t volume_block; + int32_t location; + int i; + + /* Type of the Primary Volume Descriptor must be 1. */ + if (h[PVD_type_offset] != 1) + return (0); + + /* PVD version must be 1. */ + if (h[PVD_version_offset] != 1) + return (0); + + /* Reserved field must be 0. */ + if (h[PVD_reserved1_offset] != 0) + return (0); + + /* Reserved field must be 0. */ + if (!isNull(iso9660, h, PVD_reserved2_offset, PVD_reserved2_size)) + return (0); + + /* Reserved field must be 0. */ + if (!isNull(iso9660, h, PVD_reserved3_offset, PVD_reserved3_size)) + return (0); + + /* Logical block size must be > 0. */ + /* I've looked at Ecma 119 and can't find any stronger + * restriction on this field. */ + logical_block_size = + archive_le16dec(h + PVD_logical_block_size_offset); + if (logical_block_size <= 0) + return (0); + + volume_block = archive_le32dec(h + PVD_volume_space_size_offset); + if (volume_block <= SYSTEM_AREA_BLOCK+4) + return (0); + + /* File structure version must be 1 for ISO9660/ECMA119. */ + if (h[PVD_file_structure_version_offset] != 1) + return (0); + + /* Location of Occurrence of Type L Path Table must be + * available location, + * > SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ + location = archive_le32dec(h+PVD_type_1_path_table_offset); + if (location < SYSTEM_AREA_BLOCK+2 || location >= volume_block) + return (0); + + /* The Type M Path Table must also be at a valid location + * (although ECMA 119 requires a Type M Path Table, WinISO and + * probably other programs omit it, so we permit a zero here) + * + * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ + location = archive_be32dec(h+PVD_type_m_path_table_offset); + if ((location > 0 && location < SYSTEM_AREA_BLOCK+2) + || location >= volume_block) + return (0); + + /* Reserved field must be 0. */ + /* But accept NetBSD/FreeBSD "makefs" images with 0x20 here. */ + for (i = 0; i < PVD_reserved4_size; ++i) + if (h[PVD_reserved4_offset + i] != 0 + && h[PVD_reserved4_offset + i] != 0x20) + return (0); + + /* Reserved field must be 0. */ + if (!isNull(iso9660, h, PVD_reserved5_offset, PVD_reserved5_size)) + return (0); + + /* XXX TODO: Check other values for sanity; reject more + * malformed PVDs. XXX */ + + /* Read Root Directory Record in Volume Descriptor. */ + p = h + PVD_root_directory_record_offset; + if (p[DR_length_offset] != 34) + return (0); + + if (!iso9660->primary.location) { + iso9660->logical_block_size = logical_block_size; + iso9660->volume_block = volume_block; + iso9660->volume_size = + logical_block_size * (uint64_t)volume_block; + iso9660->primary.location = + archive_le32dec(p + DR_extent_offset); + iso9660->primary.size = archive_le32dec(p + DR_size_offset); + } + + return (48); +} + +static int +read_children(struct archive_read *a, struct file_info *parent) +{ + struct iso9660 *iso9660; + const unsigned char *b, *p; + struct file_info *multi; + size_t step, skip_size; + + iso9660 = (struct iso9660 *)(a->format->data); + /* flush any remaining bytes from the last round to ensure + * we're positioned */ + if (iso9660->entry_bytes_unconsumed) { + __archive_read_consume(a, iso9660->entry_bytes_unconsumed); + iso9660->entry_bytes_unconsumed = 0; + } + if (iso9660->current_position > parent->offset) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Ignoring out-of-order directory (%s) %jd > %jd", + parent->name.s, + (intmax_t)iso9660->current_position, + (intmax_t)parent->offset); + return (ARCHIVE_WARN); + } + if (parent->offset + parent->size > iso9660->volume_size) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Directory is beyond end-of-media: %s", + parent->name.s); + return (ARCHIVE_WARN); + } + if (iso9660->current_position < parent->offset) { + int64_t skipsize; + + skipsize = parent->offset - iso9660->current_position; + skipsize = __archive_read_consume(a, skipsize); + if (skipsize < 0) + return ((int)skipsize); + iso9660->current_position = parent->offset; + } + + step = (size_t)(((parent->size + iso9660->logical_block_size -1) / + iso9660->logical_block_size) * iso9660->logical_block_size); + b = __archive_read_ahead(a, step, NULL); + if (b == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Failed to read full block when scanning " + "ISO9660 directory list"); + return (ARCHIVE_FATAL); + } + iso9660->current_position += step; + multi = NULL; + skip_size = step; + while (step) { + p = b; + b += iso9660->logical_block_size; + step -= iso9660->logical_block_size; + for (; *p != 0 && p < b && p + *p <= b; p += *p) { + struct file_info *child; + + /* N.B.: these special directory identifiers + * are 8 bit "values" even on a + * Joliet CD with UCS-2 (16bit) encoding. + */ + + /* Skip '.' entry. */ + if (*(p + DR_name_len_offset) == 1 + && *(p + DR_name_offset) == '\0') + continue; + /* Skip '..' entry. */ + if (*(p + DR_name_len_offset) == 1 + && *(p + DR_name_offset) == '\001') + continue; + child = parse_file_info(a, parent, p, b - p); + if (child == NULL) { + __archive_read_consume(a, skip_size); + return (ARCHIVE_FATAL); + } + if (child->cl_offset == 0 && + (child->multi_extent || multi != NULL)) { + struct content *con; + + if (multi == NULL) { + multi = child; + multi->contents.first = NULL; + multi->contents.last = + &(multi->contents.first); + } + con = malloc(sizeof(struct content)); + if (con == NULL) { + archive_set_error( + &a->archive, ENOMEM, + "No memory for multi extent"); + __archive_read_consume(a, skip_size); + return (ARCHIVE_FATAL); + } + con->offset = child->offset; + con->size = child->size; + con->next = NULL; + *multi->contents.last = con; + multi->contents.last = &(con->next); + if (multi == child) { + if (add_entry(a, iso9660, child) + != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } else { + multi->size += child->size; + if (!child->multi_extent) + multi = NULL; + } + } else + if (add_entry(a, iso9660, child) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + } + + __archive_read_consume(a, skip_size); + + /* Read data which recorded by RRIP "CE" extension. */ + if (read_CE(a, iso9660) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + return (ARCHIVE_OK); +} + +static int +choose_volume(struct archive_read *a, struct iso9660 *iso9660) +{ + struct file_info *file; + int64_t skipsize; + struct vd *vd; + const void *block; + char seenJoliet; + + vd = &(iso9660->primary); + if (!iso9660->opt_support_joliet) + iso9660->seenJoliet = 0; + if (iso9660->seenJoliet && + vd->location > iso9660->joliet.location) + /* This condition is unlikely; by way of caution. */ + vd = &(iso9660->joliet); + + skipsize = LOGICAL_BLOCK_SIZE * (int64_t)vd->location; + skipsize = __archive_read_consume(a, skipsize); + if (skipsize < 0) + return ((int)skipsize); + iso9660->current_position = skipsize; + + block = __archive_read_ahead(a, vd->size, NULL); + if (block == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Failed to read full block when scanning " + "ISO9660 directory list"); + return (ARCHIVE_FATAL); + } + + /* + * While reading Root Directory, flag seenJoliet must be zero to + * avoid converting special name 0x00(Current Directory) and + * next byte to UCS2. + */ + seenJoliet = iso9660->seenJoliet;/* Save flag. */ + iso9660->seenJoliet = 0; + file = parse_file_info(a, NULL, block, vd->size); + if (file == NULL) + return (ARCHIVE_FATAL); + iso9660->seenJoliet = seenJoliet; + + /* + * If the iso image has both RockRidge and Joliet, we preferentially + * use RockRidge Extensions rather than Joliet ones. + */ + if (vd == &(iso9660->primary) && iso9660->seenRockridge + && iso9660->seenJoliet) + iso9660->seenJoliet = 0; + + if (vd == &(iso9660->primary) && !iso9660->seenRockridge + && iso9660->seenJoliet) { + /* Switch reading data from primary to joliet. */ + vd = &(iso9660->joliet); + skipsize = LOGICAL_BLOCK_SIZE * (int64_t)vd->location; + skipsize -= iso9660->current_position; + skipsize = __archive_read_consume(a, skipsize); + if (skipsize < 0) + return ((int)skipsize); + iso9660->current_position += skipsize; + + block = __archive_read_ahead(a, vd->size, NULL); + if (block == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Failed to read full block when scanning " + "ISO9660 directory list"); + return (ARCHIVE_FATAL); + } + iso9660->seenJoliet = 0; + file = parse_file_info(a, NULL, block, vd->size); + if (file == NULL) + return (ARCHIVE_FATAL); + iso9660->seenJoliet = seenJoliet; + } + + /* Store the root directory in the pending list. */ + if (add_entry(a, iso9660, file) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + if (iso9660->seenRockridge) { + a->archive.archive_format = ARCHIVE_FORMAT_ISO9660_ROCKRIDGE; + a->archive.archive_format_name = + "ISO9660 with Rockridge extensions"; + } + + return (ARCHIVE_OK); +} + +static int +archive_read_format_iso9660_read_header(struct archive_read *a, + struct archive_entry *entry) +{ + struct iso9660 *iso9660; + struct file_info *file; + int r, rd_r = ARCHIVE_OK; + + iso9660 = (struct iso9660 *)(a->format->data); + + if (!a->archive.archive_format) { + a->archive.archive_format = ARCHIVE_FORMAT_ISO9660; + a->archive.archive_format_name = "ISO9660"; + } + + if (iso9660->current_position == 0) { + r = choose_volume(a, iso9660); + if (r != ARCHIVE_OK) + return (r); + } + + file = NULL;/* Eliminate a warning. */ + /* Get the next entry that appears after the current offset. */ + r = next_entry_seek(a, iso9660, &file); + if (r != ARCHIVE_OK) + return (r); + + if (iso9660->seenJoliet) { + /* + * Convert UTF-16BE of a filename to local locale MBS + * and store the result into a filename field. + */ + if (iso9660->sconv_utf16be == NULL) { + iso9660->sconv_utf16be = + archive_string_conversion_from_charset( + &(a->archive), "UTF-16BE", 1); + if (iso9660->sconv_utf16be == NULL) + /* Couldn't allocate memory */ + return (ARCHIVE_FATAL); + } + if (iso9660->utf16be_path == NULL) { + iso9660->utf16be_path = malloc(UTF16_NAME_MAX); + if (iso9660->utf16be_path == NULL) { + archive_set_error(&a->archive, ENOMEM, + "No memory"); + return (ARCHIVE_FATAL); + } + } + if (iso9660->utf16be_previous_path == NULL) { + iso9660->utf16be_previous_path = malloc(UTF16_NAME_MAX); + if (iso9660->utf16be_previous_path == NULL) { + archive_set_error(&a->archive, ENOMEM, + "No memory"); + return (ARCHIVE_FATAL); + } + } + + iso9660->utf16be_path_len = 0; + if (build_pathname_utf16be(iso9660->utf16be_path, + UTF16_NAME_MAX, &(iso9660->utf16be_path_len), file) != 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Pathname is too long"); + return (ARCHIVE_FATAL); + } + + r = archive_entry_copy_pathname_l(entry, + (const char *)iso9660->utf16be_path, + iso9660->utf16be_path_len, + iso9660->sconv_utf16be); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "No memory for Pathname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Pathname cannot be converted " + "from %s to current locale.", + archive_string_conversion_charset_name( + iso9660->sconv_utf16be)); + + rd_r = ARCHIVE_WARN; + } + } else { + const char *path = build_pathname(&iso9660->pathname, file, 0); + if (path == NULL) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Pathname is too long"); + return (ARCHIVE_FATAL); + } else { + archive_string_empty(&iso9660->pathname); + archive_entry_set_pathname(entry, path); + } + } + + iso9660->entry_bytes_remaining = file->size; + /* Offset for sparse-file-aware clients. */ + iso9660->entry_sparse_offset = 0; + + if (file->offset + file->size > iso9660->volume_size) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "File is beyond end-of-media: %s", + archive_entry_pathname(entry)); + iso9660->entry_bytes_remaining = 0; + return (ARCHIVE_WARN); + } + + /* Set up the entry structure with information about this entry. */ + archive_entry_set_mode(entry, file->mode); + archive_entry_set_uid(entry, file->uid); + archive_entry_set_gid(entry, file->gid); + archive_entry_set_nlink(entry, file->nlinks); + if (file->birthtime_is_set) + archive_entry_set_birthtime(entry, file->birthtime, 0); + else + archive_entry_unset_birthtime(entry); + archive_entry_set_mtime(entry, file->mtime, 0); + archive_entry_set_ctime(entry, file->ctime, 0); + archive_entry_set_atime(entry, file->atime, 0); + /* N.B.: Rock Ridge supports 64-bit device numbers. */ + archive_entry_set_rdev(entry, (dev_t)file->rdev); + archive_entry_set_size(entry, iso9660->entry_bytes_remaining); + if (file->symlink.s != NULL) + archive_entry_copy_symlink(entry, file->symlink.s); + + /* Note: If the input isn't seekable, we can't rewind to + * return the same body again, so if the next entry refers to + * the same data, we have to return it as a hardlink to the + * original entry. */ + if (file->number != -1 && + file->number == iso9660->previous_number) { + if (iso9660->seenJoliet) { + r = archive_entry_copy_hardlink_l(entry, + (const char *)iso9660->utf16be_previous_path, + iso9660->utf16be_previous_path_len, + iso9660->sconv_utf16be); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "No memory for Linkname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Linkname cannot be converted " + "from %s to current locale.", + archive_string_conversion_charset_name( + iso9660->sconv_utf16be)); + rd_r = ARCHIVE_WARN; + } + } else + archive_entry_set_hardlink(entry, + iso9660->previous_pathname.s); + archive_entry_unset_size(entry); + iso9660->entry_bytes_remaining = 0; + return (rd_r); + } + + if ((file->mode & AE_IFMT) != AE_IFDIR && + file->offset < iso9660->current_position) { + int64_t r64; + + r64 = __archive_read_seek(a, file->offset, SEEK_SET); + if (r64 != (int64_t)file->offset) { + /* We can't seek backwards to extract it, so issue + * a warning. Note that this can only happen if + * this entry was added to the heap after we passed + * this offset, that is, only if the directory + * mentioning this entry is later than the body of + * the entry. Such layouts are very unusual; most + * ISO9660 writers lay out and record all directory + * information first, then store all file bodies. */ + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Ignoring out-of-order file @%jx (%s) %jd < %jd", + (intmax_t)file->number, + iso9660->pathname.s, + (intmax_t)file->offset, + (intmax_t)iso9660->current_position); + iso9660->entry_bytes_remaining = 0; + return (ARCHIVE_WARN); + } + iso9660->current_position = (uint64_t)r64; + } + + /* Initialize zisofs variables. */ + iso9660->entry_zisofs.pz = file->pz; + if (file->pz) { +#ifdef HAVE_ZLIB_H + struct zisofs *zisofs; + + zisofs = &iso9660->entry_zisofs; + zisofs->initialized = 0; + zisofs->pz_log2_bs = file->pz_log2_bs; + zisofs->pz_uncompressed_size = file->pz_uncompressed_size; + zisofs->pz_offset = 0; + zisofs->header_avail = 0; + zisofs->header_passed = 0; + zisofs->block_pointers_avail = 0; +#endif + archive_entry_set_size(entry, file->pz_uncompressed_size); + } + + iso9660->previous_number = file->number; + if (iso9660->seenJoliet) { + memcpy(iso9660->utf16be_previous_path, iso9660->utf16be_path, + iso9660->utf16be_path_len); + iso9660->utf16be_previous_path_len = iso9660->utf16be_path_len; + } else + archive_strcpy( + &iso9660->previous_pathname, iso9660->pathname.s); + + /* Reset entry_bytes_remaining if the file is multi extent. */ + iso9660->entry_content = file->contents.first; + if (iso9660->entry_content != NULL) + iso9660->entry_bytes_remaining = iso9660->entry_content->size; + + if (archive_entry_filetype(entry) == AE_IFDIR) { + /* Overwrite nlinks by proper link number which is + * calculated from number of sub directories. */ + archive_entry_set_nlink(entry, 2 + file->subdirs); + /* Directory data has been read completely. */ + iso9660->entry_bytes_remaining = 0; + } + + if (rd_r != ARCHIVE_OK) + return (rd_r); + return (ARCHIVE_OK); +} + +static int +archive_read_format_iso9660_read_data_skip(struct archive_read *a) +{ + /* Because read_next_header always does an explicit skip + * to the next entry, we don't need to do anything here. */ + (void)a; /* UNUSED */ + return (ARCHIVE_OK); +} + +#ifdef HAVE_ZLIB_H + +static int +zisofs_read_data(struct archive_read *a, + const void **buff, size_t *size, int64_t *offset) +{ + struct iso9660 *iso9660; + struct zisofs *zisofs; + const unsigned char *p; + size_t avail; + ssize_t bytes_read; + size_t uncompressed_size; + int r; + + iso9660 = (struct iso9660 *)(a->format->data); + zisofs = &iso9660->entry_zisofs; + + p = __archive_read_ahead(a, 1, &bytes_read); + if (bytes_read <= 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated zisofs file body"); + return (ARCHIVE_FATAL); + } + if (bytes_read > iso9660->entry_bytes_remaining) + bytes_read = (ssize_t)iso9660->entry_bytes_remaining; + avail = bytes_read; + uncompressed_size = 0; + + if (!zisofs->initialized) { + size_t ceil, xsize; + + /* Allocate block pointers buffer. */ + ceil = (size_t)((zisofs->pz_uncompressed_size + + (((int64_t)1) << zisofs->pz_log2_bs) - 1) + >> zisofs->pz_log2_bs); + xsize = (ceil + 1) * 4; + if (zisofs->block_pointers_alloc < xsize) { + size_t alloc; + + if (zisofs->block_pointers != NULL) + free(zisofs->block_pointers); + alloc = ((xsize >> 10) + 1) << 10; + zisofs->block_pointers = malloc(alloc); + if (zisofs->block_pointers == NULL) { + archive_set_error(&a->archive, ENOMEM, + "No memory for zisofs decompression"); + return (ARCHIVE_FATAL); + } + zisofs->block_pointers_alloc = alloc; + } + zisofs->block_pointers_size = xsize; + + /* Allocate uncompressed data buffer. */ + xsize = (size_t)1UL << zisofs->pz_log2_bs; + if (zisofs->uncompressed_buffer_size < xsize) { + if (zisofs->uncompressed_buffer != NULL) + free(zisofs->uncompressed_buffer); + zisofs->uncompressed_buffer = malloc(xsize); + if (zisofs->uncompressed_buffer == NULL) { + archive_set_error(&a->archive, ENOMEM, + "No memory for zisofs decompression"); + return (ARCHIVE_FATAL); + } + } + zisofs->uncompressed_buffer_size = xsize; + + /* + * Read the file header, and check the magic code of zisofs. + */ + if (zisofs->header_avail < sizeof(zisofs->header)) { + xsize = sizeof(zisofs->header) - zisofs->header_avail; + if (avail < xsize) + xsize = avail; + memcpy(zisofs->header + zisofs->header_avail, p, xsize); + zisofs->header_avail += xsize; + avail -= xsize; + p += xsize; + } + if (!zisofs->header_passed && + zisofs->header_avail == sizeof(zisofs->header)) { + int err = 0; + + if (memcmp(zisofs->header, zisofs_magic, + sizeof(zisofs_magic)) != 0) + err = 1; + if (archive_le32dec(zisofs->header + 8) + != zisofs->pz_uncompressed_size) + err = 1; + if (zisofs->header[12] != 4) + err = 1; + if (zisofs->header[13] != zisofs->pz_log2_bs) + err = 1; + if (err) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Illegal zisofs file body"); + return (ARCHIVE_FATAL); + } + zisofs->header_passed = 1; + } + /* + * Read block pointers. + */ + if (zisofs->header_passed && + zisofs->block_pointers_avail < zisofs->block_pointers_size) { + xsize = zisofs->block_pointers_size + - zisofs->block_pointers_avail; + if (avail < xsize) + xsize = avail; + memcpy(zisofs->block_pointers + + zisofs->block_pointers_avail, p, xsize); + zisofs->block_pointers_avail += xsize; + avail -= xsize; + p += xsize; + if (zisofs->block_pointers_avail + == zisofs->block_pointers_size) { + /* We've got all block pointers and initialize + * related variables. */ + zisofs->block_off = 0; + zisofs->block_avail = 0; + /* Complete a initialization */ + zisofs->initialized = 1; + } + } + + if (!zisofs->initialized) + goto next_data; /* We need more data. */ + } + + /* + * Get block offsets from block pointers. + */ + if (zisofs->block_avail == 0) { + uint32_t bst, bed; + + if (zisofs->block_off + 4 >= zisofs->block_pointers_size) { + /* There isn't a pair of offsets. */ + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Illegal zisofs block pointers"); + return (ARCHIVE_FATAL); + } + bst = archive_le32dec( + zisofs->block_pointers + zisofs->block_off); + if (bst != zisofs->pz_offset + (bytes_read - avail)) { + /* TODO: Should we seek offset of current file + * by bst ? */ + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Illegal zisofs block pointers(cannot seek)"); + return (ARCHIVE_FATAL); + } + bed = archive_le32dec( + zisofs->block_pointers + zisofs->block_off + 4); + if (bed < bst) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Illegal zisofs block pointers"); + return (ARCHIVE_FATAL); + } + zisofs->block_avail = bed - bst; + zisofs->block_off += 4; + + /* Initialize compression library for new block. */ + if (zisofs->stream_valid) + r = inflateReset(&zisofs->stream); + else + r = inflateInit(&zisofs->stream); + if (r != Z_OK) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Can't initialize zisofs decompression."); + return (ARCHIVE_FATAL); + } + zisofs->stream_valid = 1; + zisofs->stream.total_in = 0; + zisofs->stream.total_out = 0; + } + + /* + * Make uncompressed data. + */ + if (zisofs->block_avail == 0) { + memset(zisofs->uncompressed_buffer, 0, + zisofs->uncompressed_buffer_size); + uncompressed_size = zisofs->uncompressed_buffer_size; + } else { + zisofs->stream.next_in = (Bytef *)(uintptr_t)(const void *)p; + if (avail > zisofs->block_avail) + zisofs->stream.avail_in = zisofs->block_avail; + else + zisofs->stream.avail_in = (uInt)avail; + zisofs->stream.next_out = zisofs->uncompressed_buffer; + zisofs->stream.avail_out = + (uInt)zisofs->uncompressed_buffer_size; + + r = inflate(&zisofs->stream, 0); + switch (r) { + case Z_OK: /* Decompressor made some progress.*/ + case Z_STREAM_END: /* Found end of stream. */ + break; + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "zisofs decompression failed (%d)", r); + return (ARCHIVE_FATAL); + } + uncompressed_size = + zisofs->uncompressed_buffer_size - zisofs->stream.avail_out; + avail -= zisofs->stream.next_in - p; + zisofs->block_avail -= (uint32_t)(zisofs->stream.next_in - p); + } +next_data: + bytes_read -= avail; + *buff = zisofs->uncompressed_buffer; + *size = uncompressed_size; + *offset = iso9660->entry_sparse_offset; + iso9660->entry_sparse_offset += uncompressed_size; + iso9660->entry_bytes_remaining -= bytes_read; + iso9660->current_position += bytes_read; + zisofs->pz_offset += (uint32_t)bytes_read; + iso9660->entry_bytes_unconsumed += bytes_read; + + return (ARCHIVE_OK); +} + +#else /* HAVE_ZLIB_H */ + +static int +zisofs_read_data(struct archive_read *a, + const void **buff, size_t *size, int64_t *offset) +{ + + (void)buff;/* UNUSED */ + (void)size;/* UNUSED */ + (void)offset;/* UNUSED */ + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "zisofs is not supported on this platform."); + return (ARCHIVE_FAILED); +} + +#endif /* HAVE_ZLIB_H */ + +static int +archive_read_format_iso9660_read_data(struct archive_read *a, + const void **buff, size_t *size, int64_t *offset) +{ + ssize_t bytes_read; + struct iso9660 *iso9660; + + iso9660 = (struct iso9660 *)(a->format->data); + + if (iso9660->entry_bytes_unconsumed) { + __archive_read_consume(a, iso9660->entry_bytes_unconsumed); + iso9660->entry_bytes_unconsumed = 0; + } + + if (iso9660->entry_bytes_remaining <= 0) { + if (iso9660->entry_content != NULL) + iso9660->entry_content = iso9660->entry_content->next; + if (iso9660->entry_content == NULL) { + *buff = NULL; + *size = 0; + *offset = iso9660->entry_sparse_offset; + return (ARCHIVE_EOF); + } + /* Seek forward to the start of the entry. */ + if (iso9660->current_position < iso9660->entry_content->offset) { + int64_t step; + + step = iso9660->entry_content->offset - + iso9660->current_position; + step = __archive_read_consume(a, step); + if (step < 0) + return ((int)step); + iso9660->current_position = + iso9660->entry_content->offset; + } + if (iso9660->entry_content->offset < iso9660->current_position) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Ignoring out-of-order file (%s) %jd < %jd", + iso9660->pathname.s, + (intmax_t)iso9660->entry_content->offset, + (intmax_t)iso9660->current_position); + *buff = NULL; + *size = 0; + *offset = iso9660->entry_sparse_offset; + return (ARCHIVE_WARN); + } + iso9660->entry_bytes_remaining = iso9660->entry_content->size; + } + if (iso9660->entry_zisofs.pz) + return (zisofs_read_data(a, buff, size, offset)); + + *buff = __archive_read_ahead(a, 1, &bytes_read); + if (bytes_read == 0) + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Truncated input file"); + if (*buff == NULL) + return (ARCHIVE_FATAL); + if (bytes_read > iso9660->entry_bytes_remaining) + bytes_read = (ssize_t)iso9660->entry_bytes_remaining; + *size = bytes_read; + *offset = iso9660->entry_sparse_offset; + iso9660->entry_sparse_offset += bytes_read; + iso9660->entry_bytes_remaining -= bytes_read; + iso9660->entry_bytes_unconsumed = bytes_read; + iso9660->current_position += bytes_read; + return (ARCHIVE_OK); +} + +static int +archive_read_format_iso9660_cleanup(struct archive_read *a) +{ + struct iso9660 *iso9660; + int r = ARCHIVE_OK; + + iso9660 = (struct iso9660 *)(a->format->data); + release_files(iso9660); + free(iso9660->read_ce_req.reqs); + archive_string_free(&iso9660->pathname); + archive_string_free(&iso9660->previous_pathname); + free(iso9660->pending_files.files); +#ifdef HAVE_ZLIB_H + free(iso9660->entry_zisofs.uncompressed_buffer); + free(iso9660->entry_zisofs.block_pointers); + if (iso9660->entry_zisofs.stream_valid) { + if (inflateEnd(&iso9660->entry_zisofs.stream) != Z_OK) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Failed to clean up zlib decompressor"); + r = ARCHIVE_FATAL; + } + } +#endif + free(iso9660->utf16be_path); + free(iso9660->utf16be_previous_path); + free(iso9660); + (a->format->data) = NULL; + return (r); +} + +/* + * This routine parses a single ISO directory record, makes sense + * of any extensions, and stores the result in memory. + */ +static struct file_info * +parse_file_info(struct archive_read *a, struct file_info *parent, + const unsigned char *isodirrec, size_t reclen) +{ + struct iso9660 *iso9660; + struct file_info *file, *filep; + size_t name_len; + const unsigned char *rr_start, *rr_end; + const unsigned char *p; + size_t dr_len; + uint64_t fsize, offset; + int32_t location; + int flags; + + iso9660 = (struct iso9660 *)(a->format->data); + + if (reclen != 0) + dr_len = (size_t)isodirrec[DR_length_offset]; + /* + * Sanity check that reclen is not zero and dr_len is greater than + * reclen but at least 34 + */ + if (reclen == 0 || reclen < dr_len || dr_len < 34) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Invalid length of directory record"); + return (NULL); + } + name_len = (size_t)isodirrec[DR_name_len_offset]; + location = archive_le32dec(isodirrec + DR_extent_offset); + fsize = toi(isodirrec + DR_size_offset, DR_size_size); + /* Sanity check that name_len doesn't exceed dr_len. */ + if (dr_len - 33 < name_len || name_len == 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Invalid length of file identifier"); + return (NULL); + } + /* Sanity check that location doesn't exceed volume block. + * Don't check lower limit of location; it's possibility + * the location has negative value when file type is symbolic + * link or file size is zero. As far as I know latest mkisofs + * do that. + */ + if (location > 0 && + (location + ((fsize + iso9660->logical_block_size -1) + / iso9660->logical_block_size)) + > (uint32_t)iso9660->volume_block) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Invalid location of extent of file"); + return (NULL); + } + /* Sanity check that location doesn't have a negative value + * when the file is not empty. it's too large. */ + if (fsize != 0 && location < 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Invalid location of extent of file"); + return (NULL); + } + + /* Sanity check that this entry does not create a cycle. */ + offset = iso9660->logical_block_size * (uint64_t)location; + for (filep = parent; filep != NULL; filep = filep->parent) { + if (filep->offset == offset) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Directory structure contains loop"); + return (NULL); + } + } + + /* Create a new file entry and copy data from the ISO dir record. */ + file = (struct file_info *)calloc(1, sizeof(*file)); + if (file == NULL) { + archive_set_error(&a->archive, ENOMEM, + "No memory for file entry"); + return (NULL); + } + file->parent = parent; + file->offset = offset; + file->size = fsize; + file->mtime = isodate7(isodirrec + DR_date_offset); + file->ctime = file->atime = file->mtime; + file->rede_files.first = NULL; + file->rede_files.last = &(file->rede_files.first); + + p = isodirrec + DR_name_offset; + /* Rockridge extensions (if any) follow name. Compute this + * before fidgeting the name_len below. */ + rr_start = p + name_len + (name_len & 1 ? 0 : 1); + rr_end = isodirrec + dr_len; + + if (iso9660->seenJoliet) { + /* Joliet names are max 64 chars (128 bytes) according to spec, + * but genisoimage/mkisofs allows recording longer Joliet + * names which are 103 UCS2 characters(206 bytes) by their + * option '-joliet-long'. + */ + if (name_len > 206) + name_len = 206; + name_len &= ~1; + + /* trim trailing first version and dot from filename. + * + * Remember we were in UTF-16BE land! + * SEPARATOR 1 (.) and SEPARATOR 2 (;) are both + * 16 bits big endian characters on Joliet. + * + * TODO: sanitize filename? + * Joliet allows any UCS-2 char except: + * *, /, :, ;, ? and \. + */ + /* Chop off trailing ';1' from files. */ + if (name_len > 4 && p[name_len-4] == 0 && p[name_len-3] == ';' + && p[name_len-2] == 0 && p[name_len-1] == '1') + name_len -= 4; +#if 0 /* XXX: this somehow manages to strip of single-character file extensions, like '.c'. */ + /* Chop off trailing '.' from filenames. */ + if (name_len > 2 && p[name_len-2] == 0 && p[name_len-1] == '.') + name_len -= 2; +#endif + if ((file->utf16be_name = malloc(name_len)) == NULL) { + archive_set_error(&a->archive, ENOMEM, + "No memory for file name"); + goto fail; + } + memcpy(file->utf16be_name, p, name_len); + file->utf16be_bytes = name_len; + } else { + /* Chop off trailing ';1' from files. */ + if (name_len > 2 && p[name_len - 2] == ';' && + p[name_len - 1] == '1') + name_len -= 2; + /* Chop off trailing '.' from filenames. */ + if (name_len > 1 && p[name_len - 1] == '.') + --name_len; + + archive_strncpy(&file->name, (const char *)p, name_len); + } + + flags = isodirrec[DR_flags_offset]; + if (flags & 0x02) + file->mode = AE_IFDIR | 0700; + else + file->mode = AE_IFREG | 0400; + if (flags & 0x80) + file->multi_extent = 1; + else + file->multi_extent = 0; + /* + * Use a location for the file number, which is treated as an inode + * number to find out hardlink target. If Rockridge extensions is + * being used, the file number will be overwritten by FILE SERIAL + * NUMBER of RRIP "PX" extension. + * Note: Old mkisofs did not record that FILE SERIAL NUMBER + * in ISO images. + * Note2: xorriso set 0 to the location of a symlink file. + */ + if (file->size == 0 && location >= 0) { + /* If file->size is zero, its location points wrong place, + * and so we should not use it for the file number. + * When the location has negative value, it can be used + * for the file number. + */ + file->number = -1; + /* Do not appear before any directory entries. */ + file->offset = -1; + } else + file->number = (int64_t)(uint32_t)location; + + /* Rockridge extensions overwrite information from above. */ + if (iso9660->opt_support_rockridge) { + if (parent == NULL && rr_end - rr_start >= 7) { + p = rr_start; + if (memcmp(p, "SP\x07\x01\xbe\xef", 6) == 0) { + /* + * SP extension stores the suspOffset + * (Number of bytes to skip between + * filename and SUSP records.) + * It is mandatory by the SUSP standard + * (IEEE 1281). + * + * It allows SUSP to coexist with + * non-SUSP uses of the System + * Use Area by placing non-SUSP data + * before SUSP data. + * + * SP extension must be in the root + * directory entry, disable all SUSP + * processing if not found. + */ + iso9660->suspOffset = p[6]; + iso9660->seenSUSP = 1; + rr_start += 7; + } + } + if (iso9660->seenSUSP) { + int r; + + file->name_continues = 0; + file->symlink_continues = 0; + rr_start += iso9660->suspOffset; + r = parse_rockridge(a, file, rr_start, rr_end); + if (r != ARCHIVE_OK) + goto fail; + /* + * A file size of symbolic link files in ISO images + * made by makefs is not zero and its location is + * the same as those of next regular file. That is + * the same as hard like file and it causes unexpected + * error. + */ + if (file->size > 0 && + (file->mode & AE_IFMT) == AE_IFLNK) { + file->size = 0; + file->number = -1; + file->offset = -1; + } + } else + /* If there isn't SUSP, disable parsing + * rock ridge extensions. */ + iso9660->opt_support_rockridge = 0; + } + + file->nlinks = 1;/* Reset nlink. we'll calculate it later. */ + /* Tell file's parent how many children that parent has. */ + if (parent != NULL && (flags & 0x02)) + parent->subdirs++; + + if (iso9660->seenRockridge) { + if (parent != NULL && parent->parent == NULL && + (flags & 0x02) && iso9660->rr_moved == NULL && + file->name.s && + (strcmp(file->name.s, "rr_moved") == 0 || + strcmp(file->name.s, ".rr_moved") == 0)) { + iso9660->rr_moved = file; + file->rr_moved = 1; + file->rr_moved_has_re_only = 1; + file->re = 0; + parent->subdirs--; + } else if (file->re) { + /* + * Sanity check: file's parent is rr_moved. + */ + if (parent == NULL || parent->rr_moved == 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Invalid Rockridge RE"); + goto fail; + } + /* + * Sanity check: file does not have "CL" extension. + */ + if (file->cl_offset) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Invalid Rockridge RE and CL"); + goto fail; + } + /* + * Sanity check: The file type must be a directory. + */ + if ((flags & 0x02) == 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Invalid Rockridge RE"); + goto fail; + } + } else if (parent != NULL && parent->rr_moved) + file->rr_moved_has_re_only = 0; + else if (parent != NULL && (flags & 0x02) && + (parent->re || parent->re_descendant)) + file->re_descendant = 1; + if (file->cl_offset) { + struct file_info *r; + + if (parent == NULL || parent->parent == NULL) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Invalid Rockridge CL"); + goto fail; + } + /* + * Sanity check: The file type must be a regular file. + */ + if ((flags & 0x02) != 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Invalid Rockridge CL"); + goto fail; + } + parent->subdirs++; + /* Overwrite an offset and a number of this "CL" entry + * to appear before other dirs. "+1" to those is to + * make sure to appear after "RE" entry which this + * "CL" entry should be connected with. */ + file->offset = file->number = file->cl_offset + 1; + + /* + * Sanity check: cl_offset does not point at its + * the parents or itself. + */ + for (r = parent; r; r = r->parent) { + if (r->offset == file->cl_offset) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Invalid Rockridge CL"); + goto fail; + } + } + if (file->cl_offset == file->offset || + parent->rr_moved) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Invalid Rockridge CL"); + goto fail; + } + } + } + +#if DEBUG + /* DEBUGGING: Warn about attributes I don't yet fully support. */ + if ((flags & ~0x02) != 0) { + fprintf(stderr, "\n ** Unrecognized flag: "); + dump_isodirrec(stderr, isodirrec); + fprintf(stderr, "\n"); + } else if (toi(isodirrec + DR_volume_sequence_number_offset, 2) != 1) { + fprintf(stderr, "\n ** Unrecognized sequence number: "); + dump_isodirrec(stderr, isodirrec); + fprintf(stderr, "\n"); + } else if (*(isodirrec + DR_file_unit_size_offset) != 0) { + fprintf(stderr, "\n ** Unexpected file unit size: "); + dump_isodirrec(stderr, isodirrec); + fprintf(stderr, "\n"); + } else if (*(isodirrec + DR_interleave_offset) != 0) { + fprintf(stderr, "\n ** Unexpected interleave: "); + dump_isodirrec(stderr, isodirrec); + fprintf(stderr, "\n"); + } else if (*(isodirrec + DR_ext_attr_length_offset) != 0) { + fprintf(stderr, "\n ** Unexpected extended attribute length: "); + dump_isodirrec(stderr, isodirrec); + fprintf(stderr, "\n"); + } +#endif + register_file(iso9660, file); + return (file); +fail: + archive_string_free(&file->name); + free(file); + return (NULL); +} + +static int +parse_rockridge(struct archive_read *a, struct file_info *file, + const unsigned char *p, const unsigned char *end) +{ + struct iso9660 *iso9660; + int entry_seen = 0; + + iso9660 = (struct iso9660 *)(a->format->data); + + while (p + 4 <= end /* Enough space for another entry. */ + && p[0] >= 'A' && p[0] <= 'Z' /* Sanity-check 1st char of name. */ + && p[1] >= 'A' && p[1] <= 'Z' /* Sanity-check 2nd char of name. */ + && p[2] >= 4 /* Sanity-check length. */ + && p + p[2] <= end) { /* Sanity-check length. */ + const unsigned char *data = p + 4; + int data_length = p[2] - 4; + int version = p[3]; + + switch(p[0]) { + case 'C': + if (p[1] == 'E') { + if (version == 1 && data_length == 24) { + /* + * CE extension comprises: + * 8 byte sector containing extension + * 8 byte offset w/in above sector + * 8 byte length of continuation + */ + int32_t location = + archive_le32dec(data); + file->ce_offset = + archive_le32dec(data+8); + file->ce_size = + archive_le32dec(data+16); + if (register_CE(a, location, file) + != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + } + else if (p[1] == 'L') { + if (version == 1 && data_length == 8) { + file->cl_offset = (uint64_t) + iso9660->logical_block_size * + (uint64_t)archive_le32dec(data); + iso9660->seenRockridge = 1; + } + } + break; + case 'N': + if (p[1] == 'M') { + if (version == 1) { + parse_rockridge_NM1(file, + data, data_length); + iso9660->seenRockridge = 1; + } + } + break; + case 'P': + /* + * PD extension is padding; + * contents are always ignored. + * + * PL extension won't appear; + * contents are always ignored. + */ + if (p[1] == 'N') { + if (version == 1 && data_length == 16) { + file->rdev = toi(data,4); + file->rdev <<= 32; + file->rdev |= toi(data + 8, 4); + iso9660->seenRockridge = 1; + } + } + else if (p[1] == 'X') { + /* + * PX extension comprises: + * 8 bytes for mode, + * 8 bytes for nlinks, + * 8 bytes for uid, + * 8 bytes for gid, + * 8 bytes for inode. + */ + if (version == 1) { + if (data_length >= 8) + file->mode + = toi(data, 4); + if (data_length >= 16) + file->nlinks + = toi(data + 8, 4); + if (data_length >= 24) + file->uid + = toi(data + 16, 4); + if (data_length >= 32) + file->gid + = toi(data + 24, 4); + if (data_length >= 40) + file->number + = toi(data + 32, 4); + iso9660->seenRockridge = 1; + } + } + break; + case 'R': + if (p[1] == 'E' && version == 1) { + file->re = 1; + iso9660->seenRockridge = 1; + } + else if (p[1] == 'R' && version == 1) { + /* + * RR extension comprises: + * one byte flag value + * This extension is obsolete, + * so contents are always ignored. + */ + } + break; + case 'S': + if (p[1] == 'L') { + if (version == 1) { + parse_rockridge_SL1(file, + data, data_length); + iso9660->seenRockridge = 1; + } + } + else if (p[1] == 'T' + && data_length == 0 && version == 1) { + /* + * ST extension marks end of this + * block of SUSP entries. + * + * It allows SUSP to coexist with + * non-SUSP uses of the System + * Use Area by placing non-SUSP data + * after SUSP data. + */ + iso9660->seenSUSP = 0; + iso9660->seenRockridge = 0; + return (ARCHIVE_OK); + } + break; + case 'T': + if (p[1] == 'F') { + if (version == 1) { + parse_rockridge_TF1(file, + data, data_length); + iso9660->seenRockridge = 1; + } + } + break; + case 'Z': + if (p[1] == 'F') { + if (version == 1) + parse_rockridge_ZF1(file, + data, data_length); + } + break; + default: + break; + } + + p += p[2]; + entry_seen = 1; + } + + if (entry_seen) + return (ARCHIVE_OK); + else { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Tried to parse Rockridge extensions, but none found"); + return (ARCHIVE_WARN); + } +} + +static int +register_CE(struct archive_read *a, int32_t location, + struct file_info *file) +{ + struct iso9660 *iso9660; + struct read_ce_queue *heap; + struct read_ce_req *p; + uint64_t offset, parent_offset; + int hole, parent; + + iso9660 = (struct iso9660 *)(a->format->data); + offset = ((uint64_t)location) * (uint64_t)iso9660->logical_block_size; + if (((file->mode & AE_IFMT) == AE_IFREG && + offset >= file->offset) || + offset < iso9660->current_position || + (((uint64_t)file->ce_offset) + file->ce_size) + > (uint64_t)iso9660->logical_block_size || + offset + file->ce_offset + file->ce_size + > iso9660->volume_size) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Invalid parameter in SUSP \"CE\" extension"); + return (ARCHIVE_FATAL); + } + + /* Expand our CE list as necessary. */ + heap = &(iso9660->read_ce_req); + if (heap->cnt >= heap->allocated) { + int new_size; + + if (heap->allocated < 16) + new_size = 16; + else + new_size = heap->allocated * 2; + /* Overflow might keep us from growing the list. */ + if (new_size <= heap->allocated) { + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + p = calloc(new_size, sizeof(p[0])); + if (p == NULL) { + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + if (heap->reqs != NULL) { + memcpy(p, heap->reqs, heap->cnt * sizeof(*p)); + free(heap->reqs); + } + heap->reqs = p; + heap->allocated = new_size; + } + + /* + * Start with hole at end, walk it up tree to find insertion point. + */ + hole = heap->cnt++; + while (hole > 0) { + parent = (hole - 1)/2; + parent_offset = heap->reqs[parent].offset; + if (offset >= parent_offset) { + heap->reqs[hole].offset = offset; + heap->reqs[hole].file = file; + return (ARCHIVE_OK); + } + /* Move parent into hole <==> move hole up tree. */ + heap->reqs[hole] = heap->reqs[parent]; + hole = parent; + } + heap->reqs[0].offset = offset; + heap->reqs[0].file = file; + return (ARCHIVE_OK); +} + +static void +next_CE(struct read_ce_queue *heap) +{ + uint64_t a_offset, b_offset, c_offset; + int a, b, c; + struct read_ce_req tmp; + + if (heap->cnt < 1) + return; + + /* + * Move the last item in the heap to the root of the tree + */ + heap->reqs[0] = heap->reqs[--(heap->cnt)]; + + /* + * Rebalance the heap. + */ + a = 0; /* Starting element and its offset */ + a_offset = heap->reqs[a].offset; + for (;;) { + b = a + a + 1; /* First child */ + if (b >= heap->cnt) + return; + b_offset = heap->reqs[b].offset; + c = b + 1; /* Use second child if it is smaller. */ + if (c < heap->cnt) { + c_offset = heap->reqs[c].offset; + if (c_offset < b_offset) { + b = c; + b_offset = c_offset; + } + } + if (a_offset <= b_offset) + return; + tmp = heap->reqs[a]; + heap->reqs[a] = heap->reqs[b]; + heap->reqs[b] = tmp; + a = b; + } +} + + +static int +read_CE(struct archive_read *a, struct iso9660 *iso9660) +{ + struct read_ce_queue *heap; + const unsigned char *b, *p, *end; + struct file_info *file; + size_t step; + int r; + + /* Read data which RRIP "CE" extension points. */ + heap = &(iso9660->read_ce_req); + step = iso9660->logical_block_size; + while (heap->cnt && + heap->reqs[0].offset == iso9660->current_position) { + b = __archive_read_ahead(a, step, NULL); + if (b == NULL) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Failed to read full block when scanning " + "ISO9660 directory list"); + return (ARCHIVE_FATAL); + } + do { + file = heap->reqs[0].file; + if (file->ce_offset + file->ce_size > step) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Malformed CE information"); + return (ARCHIVE_FATAL); + } + p = b + file->ce_offset; + end = p + file->ce_size; + next_CE(heap); + r = parse_rockridge(a, file, p, end); + if (r != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } while (heap->cnt && + heap->reqs[0].offset == iso9660->current_position); + /* NOTE: Do not move this consume's code to front of + * do-while loop. Registration of nested CE extension + * might cause error because of current position. */ + __archive_read_consume(a, step); + iso9660->current_position += step; + } + return (ARCHIVE_OK); +} + +static void +parse_rockridge_NM1(struct file_info *file, + const unsigned char *data, int data_length) +{ + if (!file->name_continues) + archive_string_empty(&file->name); + file->name_continues = 0; + if (data_length < 1) + return; + /* + * NM version 1 extension comprises: + * 1 byte flag, value is one of: + * = 0: remainder is name + * = 1: remainder is name, next NM entry continues name + * = 2: "." + * = 4: ".." + * = 32: Implementation specific + * All other values are reserved. + */ + switch(data[0]) { + case 0: + if (data_length < 2) + return; + archive_strncat(&file->name, + (const char *)data + 1, data_length - 1); + break; + case 1: + if (data_length < 2) + return; + archive_strncat(&file->name, + (const char *)data + 1, data_length - 1); + file->name_continues = 1; + break; + case 2: + archive_strcat(&file->name, "."); + break; + case 4: + archive_strcat(&file->name, ".."); + break; + default: + return; + } + +} + +static void +parse_rockridge_TF1(struct file_info *file, const unsigned char *data, + int data_length) +{ + char flag; + /* + * TF extension comprises: + * one byte flag + * create time (optional) + * modify time (optional) + * access time (optional) + * attribute time (optional) + * Time format and presence of fields + * is controlled by flag bits. + */ + if (data_length < 1) + return; + flag = data[0]; + ++data; + --data_length; + if (flag & 0x80) { + /* Use 17-byte time format. */ + if ((flag & 1) && data_length >= 17) { + /* Create time. */ + file->birthtime_is_set = 1; + file->birthtime = isodate17(data); + data += 17; + data_length -= 17; + } + if ((flag & 2) && data_length >= 17) { + /* Modify time. */ + file->mtime = isodate17(data); + data += 17; + data_length -= 17; + } + if ((flag & 4) && data_length >= 17) { + /* Access time. */ + file->atime = isodate17(data); + data += 17; + data_length -= 17; + } + if ((flag & 8) && data_length >= 17) { + /* Attribute change time. */ + file->ctime = isodate17(data); + } + } else { + /* Use 7-byte time format. */ + if ((flag & 1) && data_length >= 7) { + /* Create time. */ + file->birthtime_is_set = 1; + file->birthtime = isodate7(data); + data += 7; + data_length -= 7; + } + if ((flag & 2) && data_length >= 7) { + /* Modify time. */ + file->mtime = isodate7(data); + data += 7; + data_length -= 7; + } + if ((flag & 4) && data_length >= 7) { + /* Access time. */ + file->atime = isodate7(data); + data += 7; + data_length -= 7; + } + if ((flag & 8) && data_length >= 7) { + /* Attribute change time. */ + file->ctime = isodate7(data); + } + } +} + +static void +parse_rockridge_SL1(struct file_info *file, const unsigned char *data, + int data_length) +{ + const char *separator = ""; + + if (!file->symlink_continues || file->symlink.length < 1) + archive_string_empty(&file->symlink); + file->symlink_continues = 0; + + /* + * Defined flag values: + * 0: This is the last SL record for this symbolic link + * 1: this symbolic link field continues in next SL entry + * All other values are reserved. + */ + if (data_length < 1) + return; + switch(*data) { + case 0: + break; + case 1: + file->symlink_continues = 1; + break; + default: + return; + } + ++data; /* Skip flag byte. */ + --data_length; + + /* + * SL extension body stores "components". + * Basically, this is a complicated way of storing + * a POSIX path. It also interferes with using + * symlinks for storing non-path data. + * + * Each component is 2 bytes (flag and length) + * possibly followed by name data. + */ + while (data_length >= 2) { + unsigned char flag = *data++; + unsigned char nlen = *data++; + data_length -= 2; + + archive_strcat(&file->symlink, separator); + separator = "/"; + + switch(flag) { + case 0: /* Usual case, this is text. */ + if (data_length < nlen) + return; + archive_strncat(&file->symlink, + (const char *)data, nlen); + break; + case 0x01: /* Text continues in next component. */ + if (data_length < nlen) + return; + archive_strncat(&file->symlink, + (const char *)data, nlen); + separator = ""; + break; + case 0x02: /* Current dir. */ + archive_strcat(&file->symlink, "."); + break; + case 0x04: /* Parent dir. */ + archive_strcat(&file->symlink, ".."); + break; + case 0x08: /* Root of filesystem. */ + archive_strcat(&file->symlink, "/"); + separator = ""; + break; + case 0x10: /* Undefined (historically "volume root" */ + archive_string_empty(&file->symlink); + archive_strcat(&file->symlink, "ROOT"); + break; + case 0x20: /* Undefined (historically "hostname") */ + archive_strcat(&file->symlink, "hostname"); + break; + default: + /* TODO: issue a warning ? */ + return; + } + data += nlen; + data_length -= nlen; + } +} + +static void +parse_rockridge_ZF1(struct file_info *file, const unsigned char *data, + int data_length) +{ + + if (data[0] == 0x70 && data[1] == 0x7a && data_length == 12) { + /* paged zlib */ + file->pz = 1; + file->pz_log2_bs = data[3]; + file->pz_uncompressed_size = archive_le32dec(&data[4]); + } +} + +static void +register_file(struct iso9660 *iso9660, struct file_info *file) +{ + + file->use_next = iso9660->use_files; + iso9660->use_files = file; +} + +static void +release_files(struct iso9660 *iso9660) +{ + struct content *con, *connext; + struct file_info *file; + + file = iso9660->use_files; + while (file != NULL) { + struct file_info *next = file->use_next; + + archive_string_free(&file->name); + archive_string_free(&file->symlink); + free(file->utf16be_name); + con = file->contents.first; + while (con != NULL) { + connext = con->next; + free(con); + con = connext; + } + free(file); + file = next; + } +} + +static int +next_entry_seek(struct archive_read *a, struct iso9660 *iso9660, + struct file_info **pfile) +{ + struct file_info *file; + int r; + + r = next_cache_entry(a, iso9660, pfile); + if (r != ARCHIVE_OK) + return (r); + file = *pfile; + + /* Don't waste time seeking for zero-length bodies. */ + if (file->size == 0) + file->offset = iso9660->current_position; + + /* flush any remaining bytes from the last round to ensure + * we're positioned */ + if (iso9660->entry_bytes_unconsumed) { + __archive_read_consume(a, iso9660->entry_bytes_unconsumed); + iso9660->entry_bytes_unconsumed = 0; + } + + /* Seek forward to the start of the entry. */ + if (iso9660->current_position < file->offset) { + int64_t step; + + step = file->offset - iso9660->current_position; + step = __archive_read_consume(a, step); + if (step < 0) + return ((int)step); + iso9660->current_position = file->offset; + } + + /* We found body of file; handle it now. */ + return (ARCHIVE_OK); +} + +static int +next_cache_entry(struct archive_read *a, struct iso9660 *iso9660, + struct file_info **pfile) +{ + struct file_info *file; + struct { + struct file_info *first; + struct file_info **last; + } empty_files; + int64_t number; + int count; + + file = cache_get_entry(iso9660); + if (file != NULL) { + *pfile = file; + return (ARCHIVE_OK); + } + + for (;;) { + struct file_info *re, *d; + + *pfile = file = next_entry(iso9660); + if (file == NULL) { + /* + * If directory entries all which are descendant of + * rr_moved are still remaining, expose their. + */ + if (iso9660->re_files.first != NULL && + iso9660->rr_moved != NULL && + iso9660->rr_moved->rr_moved_has_re_only) + /* Expose "rr_moved" entry. */ + cache_add_entry(iso9660, iso9660->rr_moved); + while ((re = re_get_entry(iso9660)) != NULL) { + /* Expose its descendant dirs. */ + while ((d = rede_get_entry(re)) != NULL) + cache_add_entry(iso9660, d); + } + if (iso9660->cache_files.first != NULL) + return (next_cache_entry(a, iso9660, pfile)); + return (ARCHIVE_EOF); + } + + if (file->cl_offset) { + struct file_info *first_re = NULL; + int nexted_re = 0; + + /* + * Find "RE" dir for the current file, which + * has "CL" flag. + */ + while ((re = re_get_entry(iso9660)) + != first_re) { + if (first_re == NULL) + first_re = re; + if (re->offset == file->cl_offset) { + re->parent->subdirs--; + re->parent = file->parent; + re->re = 0; + if (re->parent->re_descendant) { + nexted_re = 1; + re->re_descendant = 1; + if (rede_add_entry(re) < 0) + goto fatal_rr; + /* Move a list of descendants + * to a new ancestor. */ + while ((d = rede_get_entry( + re)) != NULL) + if (rede_add_entry(d) + < 0) + goto fatal_rr; + break; + } + /* Replace the current file + * with "RE" dir */ + *pfile = file = re; + /* Expose its descendant */ + while ((d = rede_get_entry( + file)) != NULL) + cache_add_entry( + iso9660, d); + break; + } else + re_add_entry(iso9660, re); + } + if (nexted_re) { + /* + * Do not expose this at this time + * because we have not gotten its full-path + * name yet. + */ + continue; + } + } else if ((file->mode & AE_IFMT) == AE_IFDIR) { + int r; + + /* Read file entries in this dir. */ + r = read_children(a, file); + if (r != ARCHIVE_OK) + return (r); + + /* + * Handle a special dir of Rockridge extensions, + * "rr_moved". + */ + if (file->rr_moved) { + /* + * If this has only the subdirectories which + * have "RE" flags, do not expose at this time. + */ + if (file->rr_moved_has_re_only) + continue; + /* Otherwise expose "rr_moved" entry. */ + } else if (file->re) { + /* + * Do not expose this at this time + * because we have not gotten its full-path + * name yet. + */ + re_add_entry(iso9660, file); + continue; + } else if (file->re_descendant) { + /* + * If the top level "RE" entry of this entry + * is not exposed, we, accordingly, should not + * expose this entry at this time because + * we cannot make its proper full-path name. + */ + if (rede_add_entry(file) == 0) + continue; + /* Otherwise we can expose this entry because + * it seems its top level "RE" has already been + * exposed. */ + } + } + break; + } + + if ((file->mode & AE_IFMT) != AE_IFREG || file->number == -1) + return (ARCHIVE_OK); + + count = 0; + number = file->number; + iso9660->cache_files.first = NULL; + iso9660->cache_files.last = &(iso9660->cache_files.first); + empty_files.first = NULL; + empty_files.last = &empty_files.first; + /* Collect files which has the same file serial number. + * Peek pending_files so that file which number is different + * is not put back. */ + while (iso9660->pending_files.used > 0 && + (iso9660->pending_files.files[0]->number == -1 || + iso9660->pending_files.files[0]->number == number)) { + if (file->number == -1) { + /* This file has the same offset + * but it's wrong offset which empty files + * and symlink files have. + * NOTE: This wrong offset was recorded by + * old mkisofs utility. If ISO images is + * created by latest mkisofs, this does not + * happen. + */ + file->next = NULL; + *empty_files.last = file; + empty_files.last = &(file->next); + } else { + count++; + cache_add_entry(iso9660, file); + } + file = next_entry(iso9660); + } + + if (count == 0) { + *pfile = file; + return ((file == NULL)?ARCHIVE_EOF:ARCHIVE_OK); + } + if (file->number == -1) { + file->next = NULL; + *empty_files.last = file; + empty_files.last = &(file->next); + } else { + count++; + cache_add_entry(iso9660, file); + } + + if (count > 1) { + /* The count is the same as number of hardlink, + * so much so that each nlinks of files in cache_file + * is overwritten by value of the count. + */ + for (file = iso9660->cache_files.first; + file != NULL; file = file->next) + file->nlinks = count; + } + /* If there are empty files, that files are added + * to the tail of the cache_files. */ + if (empty_files.first != NULL) { + *iso9660->cache_files.last = empty_files.first; + iso9660->cache_files.last = empty_files.last; + } + *pfile = cache_get_entry(iso9660); + return ((*pfile == NULL)?ARCHIVE_EOF:ARCHIVE_OK); + +fatal_rr: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Failed to connect 'CL' pointer to 'RE' rr_moved pointer of " + "Rockridge extensions: current position = %jd, CL offset = %jd", + (intmax_t)iso9660->current_position, (intmax_t)file->cl_offset); + return (ARCHIVE_FATAL); +} + +static inline void +re_add_entry(struct iso9660 *iso9660, struct file_info *file) +{ + file->re_next = NULL; + *iso9660->re_files.last = file; + iso9660->re_files.last = &(file->re_next); +} + +static inline struct file_info * +re_get_entry(struct iso9660 *iso9660) +{ + struct file_info *file; + + if ((file = iso9660->re_files.first) != NULL) { + iso9660->re_files.first = file->re_next; + if (iso9660->re_files.first == NULL) + iso9660->re_files.last = + &(iso9660->re_files.first); + } + return (file); +} + +static inline int +rede_add_entry(struct file_info *file) +{ + struct file_info *re; + + /* + * Find "RE" entry. + */ + re = file->parent; + while (re != NULL && !re->re) + re = re->parent; + if (re == NULL) + return (-1); + + file->re_next = NULL; + *re->rede_files.last = file; + re->rede_files.last = &(file->re_next); + return (0); +} + +static inline struct file_info * +rede_get_entry(struct file_info *re) +{ + struct file_info *file; + + if ((file = re->rede_files.first) != NULL) { + re->rede_files.first = file->re_next; + if (re->rede_files.first == NULL) + re->rede_files.last = + &(re->rede_files.first); + } + return (file); +} + +static inline void +cache_add_entry(struct iso9660 *iso9660, struct file_info *file) +{ + file->next = NULL; + *iso9660->cache_files.last = file; + iso9660->cache_files.last = &(file->next); +} + +static inline struct file_info * +cache_get_entry(struct iso9660 *iso9660) +{ + struct file_info *file; + + if ((file = iso9660->cache_files.first) != NULL) { + iso9660->cache_files.first = file->next; + if (iso9660->cache_files.first == NULL) + iso9660->cache_files.last = + &(iso9660->cache_files.first); + } + return (file); +} + +static int +heap_add_entry(struct archive_read *a, struct heap_queue *heap, + struct file_info *file, uint64_t key) +{ + uint64_t file_key, parent_key; + int hole, parent; + + /* Expand our pending files list as necessary. */ + if (heap->used >= heap->allocated) { + struct file_info **new_pending_files; + int new_size = heap->allocated * 2; + + if (heap->allocated < 1024) + new_size = 1024; + /* Overflow might keep us from growing the list. */ + if (new_size <= heap->allocated) { + archive_set_error(&a->archive, + ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + new_pending_files = (struct file_info **) + malloc(new_size * sizeof(new_pending_files[0])); + if (new_pending_files == NULL) { + archive_set_error(&a->archive, + ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + if (heap->allocated) + memcpy(new_pending_files, heap->files, + heap->allocated * sizeof(new_pending_files[0])); + free(heap->files); + heap->files = new_pending_files; + heap->allocated = new_size; + } + + file_key = file->key = key; + + /* + * Start with hole at end, walk it up tree to find insertion point. + */ + hole = heap->used++; + while (hole > 0) { + parent = (hole - 1)/2; + parent_key = heap->files[parent]->key; + if (file_key >= parent_key) { + heap->files[hole] = file; + return (ARCHIVE_OK); + } + /* Move parent into hole <==> move hole up tree. */ + heap->files[hole] = heap->files[parent]; + hole = parent; + } + heap->files[0] = file; + + return (ARCHIVE_OK); +} + +static struct file_info * +heap_get_entry(struct heap_queue *heap) +{ + uint64_t a_key, b_key, c_key; + int a, b, c; + struct file_info *r, *tmp; + + if (heap->used < 1) + return (NULL); + + /* + * The first file in the list is the earliest; we'll return this. + */ + r = heap->files[0]; + + /* + * Move the last item in the heap to the root of the tree + */ + heap->files[0] = heap->files[--(heap->used)]; + + /* + * Rebalance the heap. + */ + a = 0; /* Starting element and its heap key */ + a_key = heap->files[a]->key; + for (;;) { + b = a + a + 1; /* First child */ + if (b >= heap->used) + return (r); + b_key = heap->files[b]->key; + c = b + 1; /* Use second child if it is smaller. */ + if (c < heap->used) { + c_key = heap->files[c]->key; + if (c_key < b_key) { + b = c; + b_key = c_key; + } + } + if (a_key <= b_key) + return (r); + tmp = heap->files[a]; + heap->files[a] = heap->files[b]; + heap->files[b] = tmp; + a = b; + } +} + +static unsigned int +toi(const void *p, int n) +{ + const unsigned char *v = (const unsigned char *)p; + if (n > 1) + return v[0] + 256 * toi(v + 1, n - 1); + if (n == 1) + return v[0]; + return (0); +} + +static time_t +isodate7(const unsigned char *v) +{ + struct tm tm; + int offset; + time_t t; + + memset(&tm, 0, sizeof(tm)); + tm.tm_year = v[0]; + tm.tm_mon = v[1] - 1; + tm.tm_mday = v[2]; + tm.tm_hour = v[3]; + tm.tm_min = v[4]; + tm.tm_sec = v[5]; + /* v[6] is the signed timezone offset, in 1/4-hour increments. */ + offset = ((const signed char *)v)[6]; + if (offset > -48 && offset < 52) { + tm.tm_hour -= offset / 4; + tm.tm_min -= (offset % 4) * 15; + } + t = time_from_tm(&tm); + if (t == (time_t)-1) + return ((time_t)0); + return (t); +} + +static time_t +isodate17(const unsigned char *v) +{ + struct tm tm; + int offset; + time_t t; + + memset(&tm, 0, sizeof(tm)); + tm.tm_year = (v[0] - '0') * 1000 + (v[1] - '0') * 100 + + (v[2] - '0') * 10 + (v[3] - '0') + - 1900; + tm.tm_mon = (v[4] - '0') * 10 + (v[5] - '0'); + tm.tm_mday = (v[6] - '0') * 10 + (v[7] - '0'); + tm.tm_hour = (v[8] - '0') * 10 + (v[9] - '0'); + tm.tm_min = (v[10] - '0') * 10 + (v[11] - '0'); + tm.tm_sec = (v[12] - '0') * 10 + (v[13] - '0'); + /* v[16] is the signed timezone offset, in 1/4-hour increments. */ + offset = ((const signed char *)v)[16]; + if (offset > -48 && offset < 52) { + tm.tm_hour -= offset / 4; + tm.tm_min -= (offset % 4) * 15; + } + t = time_from_tm(&tm); + if (t == (time_t)-1) + return ((time_t)0); + return (t); +} + +static time_t +time_from_tm(struct tm *t) +{ +#if HAVE_TIMEGM + /* Use platform timegm() if available. */ + return (timegm(t)); +#elif HAVE__MKGMTIME64 + return (_mkgmtime64(t)); +#else + /* Else use direct calculation using POSIX assumptions. */ + /* First, fix up tm_yday based on the year/month/day. */ + if (mktime(t) == (time_t)-1) + return ((time_t)-1); + /* Then we can compute timegm() from first principles. */ + return (t->tm_sec + + t->tm_min * 60 + + t->tm_hour * 3600 + + t->tm_yday * 86400 + + (t->tm_year - 70) * 31536000 + + ((t->tm_year - 69) / 4) * 86400 + - ((t->tm_year - 1) / 100) * 86400 + + ((t->tm_year + 299) / 400) * 86400); +#endif +} + +static const char * +build_pathname(struct archive_string *as, struct file_info *file, int depth) +{ + // Plain ISO9660 only allows 8 dir levels; if we get + // to 1000, then something is very, very wrong. + if (depth > 1000) { + return NULL; + } + if (file->parent != NULL && archive_strlen(&file->parent->name) > 0) { + if (build_pathname(as, file->parent, depth + 1) == NULL) { + return NULL; + } + archive_strcat(as, "/"); + } + if (archive_strlen(&file->name) == 0) + archive_strcat(as, "."); + else + archive_string_concat(as, &file->name); + return (as->s); +} + +static int +build_pathname_utf16be(unsigned char *p, size_t max, size_t *len, + struct file_info *file) +{ + if (file->parent != NULL && file->parent->utf16be_bytes > 0) { + if (build_pathname_utf16be(p, max, len, file->parent) != 0) + return (-1); + p[*len] = 0; + p[*len + 1] = '/'; + *len += 2; + } + if (file->utf16be_bytes == 0) { + if (*len + 2 > max) + return (-1);/* Path is too long! */ + p[*len] = 0; + p[*len + 1] = '.'; + *len += 2; + } else { + if (*len + file->utf16be_bytes > max) + return (-1);/* Path is too long! */ + memcpy(p + *len, file->utf16be_name, file->utf16be_bytes); + *len += file->utf16be_bytes; + } + return (0); +} + +#if DEBUG +static void +dump_isodirrec(FILE *out, const unsigned char *isodirrec) +{ + fprintf(out, " l %d,", + toi(isodirrec + DR_length_offset, DR_length_size)); + fprintf(out, " a %d,", + toi(isodirrec + DR_ext_attr_length_offset, DR_ext_attr_length_size)); + fprintf(out, " ext 0x%x,", + toi(isodirrec + DR_extent_offset, DR_extent_size)); + fprintf(out, " s %d,", + toi(isodirrec + DR_size_offset, DR_extent_size)); + fprintf(out, " f 0x%x,", + toi(isodirrec + DR_flags_offset, DR_flags_size)); + fprintf(out, " u %d,", + toi(isodirrec + DR_file_unit_size_offset, DR_file_unit_size_size)); + fprintf(out, " ilv %d,", + toi(isodirrec + DR_interleave_offset, DR_interleave_size)); + fprintf(out, " seq %d,", + toi(isodirrec + DR_volume_sequence_number_offset, + DR_volume_sequence_number_size)); + fprintf(out, " nl %d:", + toi(isodirrec + DR_name_len_offset, DR_name_len_size)); + fprintf(out, " `%.*s'", + toi(isodirrec + DR_name_len_offset, DR_name_len_size), + isodirrec + DR_name_offset); +} +#endif diff --git a/c/archive_read_support_format_lha.c b/c/archive_read_support_format_lha.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_format_lha.c @@ -0,0 +1,2912 @@ +/*- + * Copyright (c) 2008-2014 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_LIMITS_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_private.h" +#include "archive_read_private.h" +#include "archive_endian.h" + + +#define MAXMATCH 256 /* Maximum match length. */ +#define MINMATCH 3 /* Minimum match length. */ +/* + * Literal table format: + * +0 +256 +510 + * +---------------+-------------------------+ + * | literal code | match length | + * | 0 ... 255 | MINMATCH ... MAXMATCH | + * +---------------+-------------------------+ + * <--- LT_BITLEN_SIZE ---> + */ +/* Literal table size. */ +#define LT_BITLEN_SIZE (UCHAR_MAX + 1 + MAXMATCH - MINMATCH + 1) +/* Position table size. + * Note: this used for both position table and pre literal table.*/ +#define PT_BITLEN_SIZE (3 + 16) + +struct lzh_dec { + /* Decoding status. */ + int state; + + /* + * Window to see last 8Ki(lh5),32Ki(lh6),64Ki(lh7) bytes of decoded + * data. + */ + int w_size; + int w_mask; + /* Window buffer, which is a loop buffer. */ + unsigned char *w_buff; + /* The insert position to the window. */ + int w_pos; + /* The position where we can copy decoded code from the window. */ + int copy_pos; + /* The length how many bytes we can copy decoded code from + * the window. */ + int copy_len; + + /* + * Bit stream reader. + */ + struct lzh_br { +#define CACHE_TYPE uint64_t +#define CACHE_BITS (8 * sizeof(CACHE_TYPE)) + /* Cache buffer. */ + CACHE_TYPE cache_buffer; + /* Indicates how many bits avail in cache_buffer. */ + int cache_avail; + } br; + + /* + * Huffman coding. + */ + struct huffman { + int len_size; + int len_avail; + int len_bits; + int freq[17]; + unsigned char *bitlen; + + /* + * Use a index table. It's faster than searching a huffman + * coding tree, which is a binary tree. But a use of a large + * index table causes L1 cache read miss many times. + */ +#define HTBL_BITS 10 + int max_bits; + int shift_bits; + int tbl_bits; + int tree_used; + int tree_avail; + /* Direct access table. */ + uint16_t *tbl; + /* Binary tree table for extra bits over the direct access. */ + struct htree_t { + uint16_t left; + uint16_t right; + } *tree; + } lt, pt; + + int blocks_avail; + int pos_pt_len_size; + int pos_pt_len_bits; + int literal_pt_len_size; + int literal_pt_len_bits; + int reading_position; + int loop; + int error; +}; + +struct lzh_stream { + const unsigned char *next_in; + int avail_in; + int64_t total_in; + const unsigned char *ref_ptr; + int avail_out; + int64_t total_out; + struct lzh_dec *ds; +}; + +struct lha { + /* entry_bytes_remaining is the number of bytes we expect. */ + int64_t entry_offset; + int64_t entry_bytes_remaining; + int64_t entry_unconsumed; + uint16_t entry_crc_calculated; + + size_t header_size; /* header size */ + unsigned char level; /* header level */ + char method[3]; /* compress type */ + int64_t compsize; /* compressed data size */ + int64_t origsize; /* original file size */ + int setflag; +#define BIRTHTIME_IS_SET 1 +#define ATIME_IS_SET 2 +#define UNIX_MODE_IS_SET 4 +#define CRC_IS_SET 8 + time_t birthtime; + long birthtime_tv_nsec; + time_t mtime; + long mtime_tv_nsec; + time_t atime; + long atime_tv_nsec; + mode_t mode; + int64_t uid; + int64_t gid; + struct archive_string uname; + struct archive_string gname; + uint16_t header_crc; + uint16_t crc; + /* dirname and filename could be in different codepages */ + struct archive_string_conv *sconv_dir; + struct archive_string_conv *sconv_fname; + struct archive_string_conv *opt_sconv; + + struct archive_string dirname; + struct archive_string filename; + struct archive_wstring ws; + + unsigned char dos_attr; + + /* Flag to mark progress that an archive was read their first header.*/ + char found_first_header; + /* Flag to mark that indicates an empty directory. */ + char directory; + + /* Flags to mark progress of decompression. */ + char decompress_init; + char end_of_entry; + char end_of_entry_cleanup; + char entry_is_compressed; + + char format_name[64]; + + struct lzh_stream strm; +}; + +/* + * LHA header common member offset. + */ +#define H_METHOD_OFFSET 2 /* Compress type. */ +#define H_ATTR_OFFSET 19 /* DOS attribute. */ +#define H_LEVEL_OFFSET 20 /* Header Level. */ +#define H_SIZE 22 /* Minimum header size. */ + +static int archive_read_format_lha_bid(struct archive_read *, int); +static int archive_read_format_lha_options(struct archive_read *, + const char *, const char *); +static int archive_read_format_lha_read_header(struct archive_read *, + struct archive_entry *); +static int archive_read_format_lha_read_data(struct archive_read *, + const void **, size_t *, int64_t *); +static int archive_read_format_lha_read_data_skip(struct archive_read *); +static int archive_read_format_lha_cleanup(struct archive_read *); + +static void lha_replace_path_separator(struct lha *, + struct archive_entry *); +static int lha_read_file_header_0(struct archive_read *, struct lha *); +static int lha_read_file_header_1(struct archive_read *, struct lha *); +static int lha_read_file_header_2(struct archive_read *, struct lha *); +static int lha_read_file_header_3(struct archive_read *, struct lha *); +static int lha_read_file_extended_header(struct archive_read *, + struct lha *, uint16_t *, int, size_t, size_t *); +static size_t lha_check_header_format(const void *); +static int lha_skip_sfx(struct archive_read *); +static time_t lha_dos_time(const unsigned char *); +static time_t lha_win_time(uint64_t, long *); +static unsigned char lha_calcsum(unsigned char, const void *, + int, size_t); +static int lha_parse_linkname(struct archive_wstring *, + struct archive_wstring *); +static int lha_read_data_none(struct archive_read *, const void **, + size_t *, int64_t *); +static int lha_read_data_lzh(struct archive_read *, const void **, + size_t *, int64_t *); +static void lha_crc16_init(void); +static uint16_t lha_crc16(uint16_t, const void *, size_t); +static int lzh_decode_init(struct lzh_stream *, const char *); +static void lzh_decode_free(struct lzh_stream *); +static int lzh_decode(struct lzh_stream *, int); +static int lzh_br_fillup(struct lzh_stream *, struct lzh_br *); +static int lzh_huffman_init(struct huffman *, size_t, int); +static void lzh_huffman_free(struct huffman *); +static int lzh_read_pt_bitlen(struct lzh_stream *, int start, int end); +static int lzh_make_fake_table(struct huffman *, uint16_t); +static int lzh_make_huffman_table(struct huffman *); +static inline int lzh_decode_huffman(struct huffman *, unsigned); +static int lzh_decode_huffman_tree(struct huffman *, unsigned, int); + + +int +archive_read_support_format_lha(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct lha *lha; + int r; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_format_lha"); + + lha = (struct lha *)calloc(1, sizeof(*lha)); + if (lha == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate lha data"); + return (ARCHIVE_FATAL); + } + archive_string_init(&lha->ws); + + r = __archive_read_register_format(a, + lha, + "lha", + archive_read_format_lha_bid, + archive_read_format_lha_options, + archive_read_format_lha_read_header, + archive_read_format_lha_read_data, + archive_read_format_lha_read_data_skip, + NULL, + archive_read_format_lha_cleanup, + NULL, + NULL); + + if (r != ARCHIVE_OK) + free(lha); + return (ARCHIVE_OK); +} + +static size_t +lha_check_header_format(const void *h) +{ + const unsigned char *p = h; + size_t next_skip_bytes; + + switch (p[H_METHOD_OFFSET+3]) { + /* + * "-lh0-" ... "-lh7-" "-lhd-" + * "-lzs-" "-lz5-" + */ + case '0': case '1': case '2': case '3': + case '4': case '5': case '6': case '7': + case 'd': + case 's': + next_skip_bytes = 4; + + /* b0 == 0 means the end of an LHa archive file. */ + if (p[0] == 0) + break; + if (p[H_METHOD_OFFSET] != '-' || p[H_METHOD_OFFSET+1] != 'l' + || p[H_METHOD_OFFSET+4] != '-') + break; + + if (p[H_METHOD_OFFSET+2] == 'h') { + /* "-lh?-" */ + if (p[H_METHOD_OFFSET+3] == 's') + break; + if (p[H_LEVEL_OFFSET] == 0) + return (0); + if (p[H_LEVEL_OFFSET] <= 3 && p[H_ATTR_OFFSET] == 0x20) + return (0); + } + if (p[H_METHOD_OFFSET+2] == 'z') { + /* LArc extensions: -lzs-,-lz4- and -lz5- */ + if (p[H_LEVEL_OFFSET] != 0) + break; + if (p[H_METHOD_OFFSET+3] == 's' + || p[H_METHOD_OFFSET+3] == '4' + || p[H_METHOD_OFFSET+3] == '5') + return (0); + } + break; + case 'h': next_skip_bytes = 1; break; + case 'z': next_skip_bytes = 1; break; + case 'l': next_skip_bytes = 2; break; + case '-': next_skip_bytes = 3; break; + default : next_skip_bytes = 4; break; + } + + return (next_skip_bytes); +} + +static int +archive_read_format_lha_bid(struct archive_read *a, int best_bid) +{ + const char *p; + const void *buff; + ssize_t bytes_avail, offset, window; + size_t next; + + /* If there's already a better bid than we can ever + make, don't bother testing. */ + if (best_bid > 30) + return (-1); + + if ((p = __archive_read_ahead(a, H_SIZE, NULL)) == NULL) + return (-1); + + if (lha_check_header_format(p) == 0) + return (30); + + if (p[0] == 'M' && p[1] == 'Z') { + /* PE file */ + offset = 0; + window = 4096; + while (offset < (1024 * 20)) { + buff = __archive_read_ahead(a, offset + window, + &bytes_avail); + if (buff == NULL) { + /* Remaining bytes are less than window. */ + window >>= 1; + if (window < (H_SIZE + 3)) + return (0); + continue; + } + p = (const char *)buff + offset; + while (p + H_SIZE < (const char *)buff + bytes_avail) { + if ((next = lha_check_header_format(p)) == 0) + return (30); + p += next; + } + offset = p - (const char *)buff; + } + } + return (0); +} + +static int +archive_read_format_lha_options(struct archive_read *a, + const char *key, const char *val) +{ + struct lha *lha; + int ret = ARCHIVE_FAILED; + + lha = (struct lha *)(a->format->data); + if (strcmp(key, "hdrcharset") == 0) { + if (val == NULL || val[0] == 0) + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "lha: hdrcharset option needs a character-set name"); + else { + lha->opt_sconv = + archive_string_conversion_from_charset( + &a->archive, val, 0); + if (lha->opt_sconv != NULL) + ret = ARCHIVE_OK; + else + ret = ARCHIVE_FATAL; + } + return (ret); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +static int +lha_skip_sfx(struct archive_read *a) +{ + const void *h; + const char *p, *q; + size_t next, skip; + ssize_t bytes, window; + + window = 4096; + for (;;) { + h = __archive_read_ahead(a, window, &bytes); + if (h == NULL) { + /* Remaining bytes are less than window. */ + window >>= 1; + if (window < (H_SIZE + 3)) + goto fatal; + continue; + } + if (bytes < H_SIZE) + goto fatal; + p = h; + q = p + bytes; + + /* + * Scan ahead until we find something that looks + * like the lha header. + */ + while (p + H_SIZE < q) { + if ((next = lha_check_header_format(p)) == 0) { + skip = p - (const char *)h; + __archive_read_consume(a, skip); + return (ARCHIVE_OK); + } + p += next; + } + skip = p - (const char *)h; + __archive_read_consume(a, skip); + } +fatal: + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Couldn't find out LHa header"); + return (ARCHIVE_FATAL); +} + +static int +truncated_error(struct archive_read *a) +{ + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated LHa header"); + return (ARCHIVE_FATAL); +} + +static int +archive_read_format_lha_read_header(struct archive_read *a, + struct archive_entry *entry) +{ + struct archive_wstring linkname; + struct archive_wstring pathname; + struct lha *lha; + const unsigned char *p; + const char *signature; + int err; + struct archive_mstring conv_buffer; + const wchar_t *conv_buffer_p; + + lha_crc16_init(); + + a->archive.archive_format = ARCHIVE_FORMAT_LHA; + if (a->archive.archive_format_name == NULL) + a->archive.archive_format_name = "lha"; + + lha = (struct lha *)(a->format->data); + lha->decompress_init = 0; + lha->end_of_entry = 0; + lha->end_of_entry_cleanup = 0; + lha->entry_unconsumed = 0; + + if ((p = __archive_read_ahead(a, H_SIZE, NULL)) == NULL) { + /* + * LHa archiver added 0 to the tail of its archive file as + * the mark of the end of the archive. + */ + signature = __archive_read_ahead(a, sizeof(signature[0]), NULL); + if (signature == NULL || signature[0] == 0) + return (ARCHIVE_EOF); + return (truncated_error(a)); + } + + signature = (const char *)p; + if (lha->found_first_header == 0 && + signature[0] == 'M' && signature[1] == 'Z') { + /* This is an executable? Must be self-extracting... */ + err = lha_skip_sfx(a); + if (err < ARCHIVE_WARN) + return (err); + + if ((p = __archive_read_ahead(a, sizeof(*p), NULL)) == NULL) + return (truncated_error(a)); + signature = (const char *)p; + } + /* signature[0] == 0 means the end of an LHa archive file. */ + if (signature[0] == 0) + return (ARCHIVE_EOF); + + /* + * Check the header format and method type. + */ + if (lha_check_header_format(p) != 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Bad LHa file"); + return (ARCHIVE_FATAL); + } + + /* We've found the first header. */ + lha->found_first_header = 1; + /* Set a default value and common data */ + lha->header_size = 0; + lha->level = p[H_LEVEL_OFFSET]; + lha->method[0] = p[H_METHOD_OFFSET+1]; + lha->method[1] = p[H_METHOD_OFFSET+2]; + lha->method[2] = p[H_METHOD_OFFSET+3]; + if (memcmp(lha->method, "lhd", 3) == 0) + lha->directory = 1; + else + lha->directory = 0; + if (memcmp(lha->method, "lh0", 3) == 0 || + memcmp(lha->method, "lz4", 3) == 0) + lha->entry_is_compressed = 0; + else + lha->entry_is_compressed = 1; + + lha->compsize = 0; + lha->origsize = 0; + lha->setflag = 0; + lha->birthtime = 0; + lha->birthtime_tv_nsec = 0; + lha->mtime = 0; + lha->mtime_tv_nsec = 0; + lha->atime = 0; + lha->atime_tv_nsec = 0; + lha->mode = (lha->directory)? 0777 : 0666; + lha->uid = 0; + lha->gid = 0; + archive_string_empty(&lha->dirname); + archive_string_empty(&lha->filename); + lha->dos_attr = 0; + if (lha->opt_sconv != NULL) { + lha->sconv_dir = lha->opt_sconv; + lha->sconv_fname = lha->opt_sconv; + } else { + lha->sconv_dir = NULL; + lha->sconv_fname = NULL; + } + + switch (p[H_LEVEL_OFFSET]) { + case 0: + err = lha_read_file_header_0(a, lha); + break; + case 1: + err = lha_read_file_header_1(a, lha); + break; + case 2: + err = lha_read_file_header_2(a, lha); + break; + case 3: + err = lha_read_file_header_3(a, lha); + break; + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Unsupported LHa header level %d", p[H_LEVEL_OFFSET]); + err = ARCHIVE_FATAL; + break; + } + if (err < ARCHIVE_WARN) + return (err); + + + if (!lha->directory && archive_strlen(&lha->filename) == 0) + /* The filename has not been set */ + return (truncated_error(a)); + + /* + * Make a pathname from a dirname and a filename, after converting to Unicode. + * This is because codepages might differ between dirname and filename. + */ + archive_string_init(&pathname); + archive_string_init(&linkname); + archive_string_init(&conv_buffer.aes_mbs); + archive_string_init(&conv_buffer.aes_mbs_in_locale); + archive_string_init(&conv_buffer.aes_utf8); + archive_string_init(&conv_buffer.aes_wcs); + if (0 != archive_mstring_copy_mbs_len_l(&conv_buffer, lha->dirname.s, lha->dirname.length, lha->sconv_dir)) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Pathname cannot be converted " + "from %s to Unicode.", + archive_string_conversion_charset_name(lha->sconv_dir)); + err = ARCHIVE_FATAL; + } else if (0 != archive_mstring_get_wcs(&a->archive, &conv_buffer, &conv_buffer_p)) + err = ARCHIVE_FATAL; + if (err == ARCHIVE_FATAL) { + archive_mstring_clean(&conv_buffer); + archive_wstring_free(&pathname); + archive_wstring_free(&linkname); + return (err); + } + archive_wstring_copy(&pathname, &conv_buffer.aes_wcs); + + archive_string_empty(&conv_buffer.aes_mbs); + archive_string_empty(&conv_buffer.aes_mbs_in_locale); + archive_string_empty(&conv_buffer.aes_utf8); + archive_wstring_empty(&conv_buffer.aes_wcs); + if (0 != archive_mstring_copy_mbs_len_l(&conv_buffer, lha->filename.s, lha->filename.length, lha->sconv_fname)) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Pathname cannot be converted " + "from %s to Unicode.", + archive_string_conversion_charset_name(lha->sconv_fname)); + err = ARCHIVE_FATAL; + } + else if (0 != archive_mstring_get_wcs(&a->archive, &conv_buffer, &conv_buffer_p)) + err = ARCHIVE_FATAL; + if (err == ARCHIVE_FATAL) { + archive_mstring_clean(&conv_buffer); + archive_wstring_free(&pathname); + archive_wstring_free(&linkname); + return (err); + } + archive_wstring_concat(&pathname, &conv_buffer.aes_wcs); + archive_mstring_clean(&conv_buffer); + + if ((lha->mode & AE_IFMT) == AE_IFLNK) { + /* + * Extract the symlink-name if it's included in the pathname. + */ + if (!lha_parse_linkname(&linkname, &pathname)) { + /* We couldn't get the symlink-name. */ + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Unknown symlink-name"); + archive_wstring_free(&pathname); + archive_wstring_free(&linkname); + return (ARCHIVE_FAILED); + } + } else { + /* + * Make sure a file-type is set. + * The mode has been overridden if it is in the extended data. + */ + lha->mode = (lha->mode & ~AE_IFMT) | + ((lha->directory)? AE_IFDIR: AE_IFREG); + } + if ((lha->setflag & UNIX_MODE_IS_SET) == 0 && + (lha->dos_attr & 1) != 0) + lha->mode &= ~(0222);/* read only. */ + + /* + * Set basic file parameters. + */ + archive_entry_copy_pathname_w(entry, pathname.s); + archive_wstring_free(&pathname); + if (archive_strlen(&linkname) > 0) { + archive_entry_copy_symlink_w(entry, linkname.s); + } else + archive_entry_set_symlink(entry, NULL); + archive_wstring_free(&linkname); + /* + * When a header level is 0, there is a possibility that + * a pathname and a symlink has '\' character, a directory + * separator in DOS/Windows. So we should convert it to '/'. + */ + if (p[H_LEVEL_OFFSET] == 0) + lha_replace_path_separator(lha, entry); + + archive_entry_set_mode(entry, lha->mode); + archive_entry_set_uid(entry, lha->uid); + archive_entry_set_gid(entry, lha->gid); + if (archive_strlen(&lha->uname) > 0) + archive_entry_set_uname(entry, lha->uname.s); + if (archive_strlen(&lha->gname) > 0) + archive_entry_set_gname(entry, lha->gname.s); + if (lha->setflag & BIRTHTIME_IS_SET) { + archive_entry_set_birthtime(entry, lha->birthtime, + lha->birthtime_tv_nsec); + archive_entry_set_ctime(entry, lha->birthtime, + lha->birthtime_tv_nsec); + } else { + archive_entry_unset_birthtime(entry); + archive_entry_unset_ctime(entry); + } + archive_entry_set_mtime(entry, lha->mtime, lha->mtime_tv_nsec); + if (lha->setflag & ATIME_IS_SET) + archive_entry_set_atime(entry, lha->atime, + lha->atime_tv_nsec); + else + archive_entry_unset_atime(entry); + if (lha->directory || archive_entry_symlink(entry) != NULL) + archive_entry_unset_size(entry); + else + archive_entry_set_size(entry, lha->origsize); + + /* + * Prepare variables used to read a file content. + */ + lha->entry_bytes_remaining = lha->compsize; + if (lha->entry_bytes_remaining < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid LHa entry size"); + return (ARCHIVE_FATAL); + } + lha->entry_offset = 0; + lha->entry_crc_calculated = 0; + + /* + * This file does not have a content. + */ + if (lha->directory || lha->compsize == 0) + lha->end_of_entry = 1; + + sprintf(lha->format_name, "lha -%c%c%c-", + lha->method[0], lha->method[1], lha->method[2]); + a->archive.archive_format_name = lha->format_name; + + return (err); +} + +/* + * Replace a DOS path separator '\' by a character '/'. + * Some multi-byte character set have a character '\' in its second byte. + */ +static void +lha_replace_path_separator(struct lha *lha, struct archive_entry *entry) +{ + const wchar_t *wp; + size_t i; + + if ((wp = archive_entry_pathname_w(entry)) != NULL) { + archive_wstrcpy(&(lha->ws), wp); + for (i = 0; i < archive_strlen(&(lha->ws)); i++) { + if (lha->ws.s[i] == L'\\') + lha->ws.s[i] = L'/'; + } + archive_entry_copy_pathname_w(entry, lha->ws.s); + } + + if ((wp = archive_entry_symlink_w(entry)) != NULL) { + archive_wstrcpy(&(lha->ws), wp); + for (i = 0; i < archive_strlen(&(lha->ws)); i++) { + if (lha->ws.s[i] == L'\\') + lha->ws.s[i] = L'/'; + } + archive_entry_copy_symlink_w(entry, lha->ws.s); + } +} + +/* + * Header 0 format + * + * +0 +1 +2 +7 +11 + * +---------------+----------+----------------+-------------------+ + * |header size(*1)|header sum|compression type|compressed size(*2)| + * +---------------+----------+----------------+-------------------+ + * <---------------------(*1)----------* + * + * +11 +15 +17 +19 +20 +21 + * +-----------------+---------+---------+--------------+----------------+ + * |uncompressed size|time(DOS)|date(DOS)|attribute(DOS)|header level(=0)| + * +-----------------+---------+---------+--------------+----------------+ + * *--------------------------------(*1)---------------------------------* + * + * +21 +22 +22+(*3) +22+(*3)+2 +22+(*3)+2+(*4) + * +---------------+---------+----------+----------------+------------------+ + * |name length(*3)|file name|file CRC16|extra header(*4)| compressed data | + * +---------------+---------+----------+----------------+------------------+ + * <--(*3)-> <------(*2)------> + * *----------------------(*1)--------------------------> + * + */ +#define H0_HEADER_SIZE_OFFSET 0 +#define H0_HEADER_SUM_OFFSET 1 +#define H0_COMP_SIZE_OFFSET 7 +#define H0_ORIG_SIZE_OFFSET 11 +#define H0_DOS_TIME_OFFSET 15 +#define H0_NAME_LEN_OFFSET 21 +#define H0_FILE_NAME_OFFSET 22 +#define H0_FIXED_SIZE 24 +static int +lha_read_file_header_0(struct archive_read *a, struct lha *lha) +{ + const unsigned char *p; + int extdsize, namelen; + unsigned char headersum, sum_calculated; + + if ((p = __archive_read_ahead(a, H0_FIXED_SIZE, NULL)) == NULL) + return (truncated_error(a)); + lha->header_size = p[H0_HEADER_SIZE_OFFSET] + 2; + headersum = p[H0_HEADER_SUM_OFFSET]; + lha->compsize = archive_le32dec(p + H0_COMP_SIZE_OFFSET); + lha->origsize = archive_le32dec(p + H0_ORIG_SIZE_OFFSET); + lha->mtime = lha_dos_time(p + H0_DOS_TIME_OFFSET); + namelen = p[H0_NAME_LEN_OFFSET]; + extdsize = (int)lha->header_size - H0_FIXED_SIZE - namelen; + if ((namelen > 221 || extdsize < 0) && extdsize != -2) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid LHa header"); + return (ARCHIVE_FATAL); + } + if ((p = __archive_read_ahead(a, lha->header_size, NULL)) == NULL) + return (truncated_error(a)); + + archive_strncpy(&lha->filename, p + H0_FILE_NAME_OFFSET, namelen); + /* When extdsize == -2, A CRC16 value is not present in the header. */ + if (extdsize >= 0) { + lha->crc = archive_le16dec(p + H0_FILE_NAME_OFFSET + namelen); + lha->setflag |= CRC_IS_SET; + } + sum_calculated = lha_calcsum(0, p, 2, lha->header_size - 2); + + /* Read an extended header */ + if (extdsize > 0) { + /* This extended data is set by 'LHa for UNIX' only. + * Maybe fixed size. + */ + p += H0_FILE_NAME_OFFSET + namelen + 2; + if (p[0] == 'U' && extdsize == 12) { + /* p[1] is a minor version. */ + lha->mtime = archive_le32dec(&p[2]); + lha->mode = archive_le16dec(&p[6]); + lha->uid = archive_le16dec(&p[8]); + lha->gid = archive_le16dec(&p[10]); + lha->setflag |= UNIX_MODE_IS_SET; + } + } + __archive_read_consume(a, lha->header_size); + + if (sum_calculated != headersum) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "LHa header sum error"); + return (ARCHIVE_FATAL); + } + + return (ARCHIVE_OK); +} + +/* + * Header 1 format + * + * +0 +1 +2 +7 +11 + * +---------------+----------+----------------+-------------+ + * |header size(*1)|header sum|compression type|skip size(*2)| + * +---------------+----------+----------------+-------------+ + * <---------------(*1)----------* + * + * +11 +15 +17 +19 +20 +21 + * +-----------------+---------+---------+--------------+----------------+ + * |uncompressed size|time(DOS)|date(DOS)|attribute(DOS)|header level(=1)| + * +-----------------+---------+---------+--------------+----------------+ + * *-------------------------------(*1)----------------------------------* + * + * +21 +22 +22+(*3) +22+(*3)+2 +22+(*3)+3 +22+(*3)+3+(*4) + * +---------------+---------+----------+-----------+-----------+ + * |name length(*3)|file name|file CRC16| creator |padding(*4)| + * +---------------+---------+----------+-----------+-----------+ + * <--(*3)-> + * *----------------------------(*1)----------------------------* + * + * +22+(*3)+3+(*4) +22+(*3)+3+(*4)+2 +22+(*3)+3+(*4)+2+(*5) + * +----------------+---------------------+------------------------+ + * |next header size| extended header(*5) | compressed data | + * +----------------+---------------------+------------------------+ + * *------(*1)-----> <--------------------(*2)--------------------> + */ +#define H1_HEADER_SIZE_OFFSET 0 +#define H1_HEADER_SUM_OFFSET 1 +#define H1_COMP_SIZE_OFFSET 7 +#define H1_ORIG_SIZE_OFFSET 11 +#define H1_DOS_TIME_OFFSET 15 +#define H1_NAME_LEN_OFFSET 21 +#define H1_FILE_NAME_OFFSET 22 +#define H1_FIXED_SIZE 27 +static int +lha_read_file_header_1(struct archive_read *a, struct lha *lha) +{ + const unsigned char *p; + size_t extdsize; + int i, err, err2; + int namelen, padding; + unsigned char headersum, sum_calculated; + + err = ARCHIVE_OK; + + if ((p = __archive_read_ahead(a, H1_FIXED_SIZE, NULL)) == NULL) + return (truncated_error(a)); + + lha->header_size = p[H1_HEADER_SIZE_OFFSET] + 2; + headersum = p[H1_HEADER_SUM_OFFSET]; + /* Note: An extended header size is included in a compsize. */ + lha->compsize = archive_le32dec(p + H1_COMP_SIZE_OFFSET); + lha->origsize = archive_le32dec(p + H1_ORIG_SIZE_OFFSET); + lha->mtime = lha_dos_time(p + H1_DOS_TIME_OFFSET); + namelen = p[H1_NAME_LEN_OFFSET]; + /* Calculate a padding size. The result will be normally 0 only(?) */ + padding = ((int)lha->header_size) - H1_FIXED_SIZE - namelen; + + if (namelen > 230 || padding < 0) + goto invalid; + + if ((p = __archive_read_ahead(a, lha->header_size, NULL)) == NULL) + return (truncated_error(a)); + + for (i = 0; i < namelen; i++) { + if (p[i + H1_FILE_NAME_OFFSET] == 0xff) + goto invalid;/* Invalid filename. */ + } + archive_strncpy(&lha->filename, p + H1_FILE_NAME_OFFSET, namelen); + lha->crc = archive_le16dec(p + H1_FILE_NAME_OFFSET + namelen); + lha->setflag |= CRC_IS_SET; + + sum_calculated = lha_calcsum(0, p, 2, lha->header_size - 2); + /* Consume used bytes but not include `next header size' data + * since it will be consumed in lha_read_file_extended_header(). */ + __archive_read_consume(a, lha->header_size - 2); + + /* Read extended headers */ + err2 = lha_read_file_extended_header(a, lha, NULL, 2, + (size_t)(lha->compsize + 2), &extdsize); + if (err2 < ARCHIVE_WARN) + return (err2); + if (err2 < err) + err = err2; + /* Get a real compressed file size. */ + lha->compsize -= extdsize - 2; + + if (lha->compsize < 0) + goto invalid; /* Invalid compressed file size */ + + if (sum_calculated != headersum) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "LHa header sum error"); + return (ARCHIVE_FATAL); + } + return (err); +invalid: + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid LHa header"); + return (ARCHIVE_FATAL); +} + +/* + * Header 2 format + * + * +0 +2 +7 +11 +15 + * +---------------+----------------+-------------------+-----------------+ + * |header size(*1)|compression type|compressed size(*2)|uncompressed size| + * +---------------+----------------+-------------------+-----------------+ + * <--------------------------------(*1)---------------------------------* + * + * +15 +19 +20 +21 +23 +24 + * +-----------------+------------+----------------+----------+-----------+ + * |data/time(time_t)| 0x20 fixed |header level(=2)|file CRC16| creator | + * +-----------------+------------+----------------+----------+-----------+ + * *---------------------------------(*1)---------------------------------* + * + * +24 +26 +26+(*3) +26+(*3)+(*4) + * +----------------+-------------------+-------------+-------------------+ + * |next header size|extended header(*3)| padding(*4) | compressed data | + * +----------------+-------------------+-------------+-------------------+ + * *--------------------------(*1)-------------------> <------(*2)-------> + * + */ +#define H2_HEADER_SIZE_OFFSET 0 +#define H2_COMP_SIZE_OFFSET 7 +#define H2_ORIG_SIZE_OFFSET 11 +#define H2_TIME_OFFSET 15 +#define H2_CRC_OFFSET 21 +#define H2_FIXED_SIZE 24 +static int +lha_read_file_header_2(struct archive_read *a, struct lha *lha) +{ + const unsigned char *p; + size_t extdsize; + int err, padding; + uint16_t header_crc; + + if ((p = __archive_read_ahead(a, H2_FIXED_SIZE, NULL)) == NULL) + return (truncated_error(a)); + + lha->header_size =archive_le16dec(p + H2_HEADER_SIZE_OFFSET); + lha->compsize = archive_le32dec(p + H2_COMP_SIZE_OFFSET); + lha->origsize = archive_le32dec(p + H2_ORIG_SIZE_OFFSET); + lha->mtime = archive_le32dec(p + H2_TIME_OFFSET); + lha->crc = archive_le16dec(p + H2_CRC_OFFSET); + lha->setflag |= CRC_IS_SET; + + if (lha->header_size < H2_FIXED_SIZE) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid LHa header size"); + return (ARCHIVE_FATAL); + } + + header_crc = lha_crc16(0, p, H2_FIXED_SIZE); + __archive_read_consume(a, H2_FIXED_SIZE); + + /* Read extended headers */ + err = lha_read_file_extended_header(a, lha, &header_crc, 2, + lha->header_size - H2_FIXED_SIZE, &extdsize); + if (err < ARCHIVE_WARN) + return (err); + + /* Calculate a padding size. The result will be normally 0 or 1. */ + padding = (int)lha->header_size - (int)(H2_FIXED_SIZE + extdsize); + if (padding > 0) { + if ((p = __archive_read_ahead(a, padding, NULL)) == NULL) + return (truncated_error(a)); + header_crc = lha_crc16(header_crc, p, padding); + __archive_read_consume(a, padding); + } + + if (header_crc != lha->header_crc) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "LHa header CRC error"); + return (ARCHIVE_FATAL); + } + return (err); +} + +/* + * Header 3 format + * + * +0 +2 +7 +11 +15 + * +------------+----------------+-------------------+-----------------+ + * | 0x04 fixed |compression type|compressed size(*2)|uncompressed size| + * +------------+----------------+-------------------+-----------------+ + * <-------------------------------(*1)-------------------------------* + * + * +15 +19 +20 +21 +23 +24 + * +-----------------+------------+----------------+----------+-----------+ + * |date/time(time_t)| 0x20 fixed |header level(=3)|file CRC16| creator | + * +-----------------+------------+----------------+----------+-----------+ + * *--------------------------------(*1)----------------------------------* + * + * +24 +28 +32 +32+(*3) + * +---------------+----------------+-------------------+-----------------+ + * |header size(*1)|next header size|extended header(*3)| compressed data | + * +---------------+----------------+-------------------+-----------------+ + * *------------------------(*1)-----------------------> <------(*2)-----> + * + */ +#define H3_FIELD_LEN_OFFSET 0 +#define H3_COMP_SIZE_OFFSET 7 +#define H3_ORIG_SIZE_OFFSET 11 +#define H3_TIME_OFFSET 15 +#define H3_CRC_OFFSET 21 +#define H3_HEADER_SIZE_OFFSET 24 +#define H3_FIXED_SIZE 28 +static int +lha_read_file_header_3(struct archive_read *a, struct lha *lha) +{ + const unsigned char *p; + size_t extdsize; + int err; + uint16_t header_crc; + + if ((p = __archive_read_ahead(a, H3_FIXED_SIZE, NULL)) == NULL) + return (truncated_error(a)); + + if (archive_le16dec(p + H3_FIELD_LEN_OFFSET) != 4) + goto invalid; + lha->header_size =archive_le32dec(p + H3_HEADER_SIZE_OFFSET); + lha->compsize = archive_le32dec(p + H3_COMP_SIZE_OFFSET); + lha->origsize = archive_le32dec(p + H3_ORIG_SIZE_OFFSET); + lha->mtime = archive_le32dec(p + H3_TIME_OFFSET); + lha->crc = archive_le16dec(p + H3_CRC_OFFSET); + lha->setflag |= CRC_IS_SET; + + if (lha->header_size < H3_FIXED_SIZE + 4) + goto invalid; + header_crc = lha_crc16(0, p, H3_FIXED_SIZE); + __archive_read_consume(a, H3_FIXED_SIZE); + + /* Read extended headers */ + err = lha_read_file_extended_header(a, lha, &header_crc, 4, + lha->header_size - H3_FIXED_SIZE, &extdsize); + if (err < ARCHIVE_WARN) + return (err); + + if (header_crc != lha->header_crc) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "LHa header CRC error"); + return (ARCHIVE_FATAL); + } + return (err); +invalid: + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid LHa header"); + return (ARCHIVE_FATAL); +} + +/* + * Extended header format + * + * +0 +2 +3 -- used in header 1 and 2 + * +0 +4 +5 -- used in header 3 + * +--------------+---------+-------------------+--------------+-- + * |ex-header size|header id| data |ex-header size| ....... + * +--------------+---------+-------------------+--------------+-- + * <-------------( ex-header size)------------> <-- next extended header --* + * + * If the ex-header size is zero, it is the make of the end of extended + * headers. + * + */ +static int +lha_read_file_extended_header(struct archive_read *a, struct lha *lha, + uint16_t *crc, int sizefield_length, size_t limitsize, size_t *total_size) +{ + const void *h; + const unsigned char *extdheader; + size_t extdsize; + size_t datasize; + unsigned int i; + unsigned char extdtype; + +#define EXT_HEADER_CRC 0x00 /* Header CRC and information*/ +#define EXT_FILENAME 0x01 /* Filename */ +#define EXT_DIRECTORY 0x02 /* Directory name */ +#define EXT_DOS_ATTR 0x40 /* MS-DOS attribute */ +#define EXT_TIMESTAMP 0x41 /* Windows time stamp */ +#define EXT_FILESIZE 0x42 /* Large file size */ +#define EXT_TIMEZONE 0x43 /* Time zone */ +#define EXT_UTF16_FILENAME 0x44 /* UTF-16 filename */ +#define EXT_UTF16_DIRECTORY 0x45 /* UTF-16 directory name */ +#define EXT_CODEPAGE 0x46 /* Codepage */ +#define EXT_UNIX_MODE 0x50 /* File permission */ +#define EXT_UNIX_GID_UID 0x51 /* gid,uid */ +#define EXT_UNIX_GNAME 0x52 /* Group name */ +#define EXT_UNIX_UNAME 0x53 /* User name */ +#define EXT_UNIX_MTIME 0x54 /* Modified time */ +#define EXT_OS2_NEW_ATTR 0x7f /* new attribute(OS/2 only) */ +#define EXT_NEW_ATTR 0xff /* new attribute */ + + *total_size = sizefield_length; + + for (;;) { + /* Read an extended header size. */ + if ((h = + __archive_read_ahead(a, sizefield_length, NULL)) == NULL) + return (truncated_error(a)); + /* Check if the size is the zero indicates the end of the + * extended header. */ + if (sizefield_length == sizeof(uint16_t)) + extdsize = archive_le16dec(h); + else + extdsize = archive_le32dec(h); + if (extdsize == 0) { + /* End of extended header */ + if (crc != NULL) + *crc = lha_crc16(*crc, h, sizefield_length); + __archive_read_consume(a, sizefield_length); + return (ARCHIVE_OK); + } + + /* Sanity check to the extended header size. */ + if (((uint64_t)*total_size + extdsize) > + (uint64_t)limitsize || + extdsize <= (size_t)sizefield_length) + goto invalid; + + /* Read the extended header. */ + if ((h = __archive_read_ahead(a, extdsize, NULL)) == NULL) + return (truncated_error(a)); + *total_size += extdsize; + + extdheader = (const unsigned char *)h; + /* Get the extended header type. */ + extdtype = extdheader[sizefield_length]; + /* Calculate an extended data size. */ + datasize = extdsize - (1 + sizefield_length); + /* Skip an extended header size field and type field. */ + extdheader += sizefield_length + 1; + + if (crc != NULL && extdtype != EXT_HEADER_CRC) + *crc = lha_crc16(*crc, h, extdsize); + switch (extdtype) { + case EXT_HEADER_CRC: + /* We only use a header CRC. Following data will not + * be used. */ + if (datasize >= 2) { + lha->header_crc = archive_le16dec(extdheader); + if (crc != NULL) { + static const char zeros[2] = {0, 0}; + *crc = lha_crc16(*crc, h, + extdsize - datasize); + /* CRC value itself as zero */ + *crc = lha_crc16(*crc, zeros, 2); + *crc = lha_crc16(*crc, + extdheader+2, datasize - 2); + } + } + break; + case EXT_FILENAME: + if (datasize == 0) { + /* maybe directory header */ + archive_string_empty(&lha->filename); + break; + } + if (extdheader[0] == '\0') + goto invalid; + archive_strncpy(&lha->filename, + (const char *)extdheader, datasize); + break; + case EXT_UTF16_FILENAME: + if (datasize == 0) { + /* maybe directory header */ + archive_string_empty(&lha->filename); + break; + } else if (datasize & 1) { + /* UTF-16 characters take always 2 or 4 bytes */ + goto invalid; + } + if (extdheader[0] == '\0') + goto invalid; + archive_string_empty(&lha->filename); + archive_array_append(&lha->filename, + (const char *)extdheader, datasize); + /* Setup a string conversion for a filename. */ + lha->sconv_fname = + archive_string_conversion_from_charset(&a->archive, + "UTF-16LE", 1); + if (lha->sconv_fname == NULL) + return (ARCHIVE_FATAL); + break; + case EXT_DIRECTORY: + if (datasize == 0 || extdheader[0] == '\0') + /* no directory name data. exit this case. */ + goto invalid; + + archive_strncpy(&lha->dirname, + (const char *)extdheader, datasize); + /* + * Convert directory delimiter from 0xFF + * to '/' for local system. + */ + for (i = 0; i < lha->dirname.length; i++) { + if ((unsigned char)lha->dirname.s[i] == 0xFF) + lha->dirname.s[i] = '/'; + } + /* Is last character directory separator? */ + if (lha->dirname.s[lha->dirname.length-1] != '/') + /* invalid directory data */ + goto invalid; + break; + case EXT_UTF16_DIRECTORY: + /* UTF-16 characters take always 2 or 4 bytes */ + if (datasize == 0 || (datasize & 1) || + extdheader[0] == '\0') { + /* no directory name data. exit this case. */ + goto invalid; + } + + archive_string_empty(&lha->dirname); + archive_array_append(&lha->dirname, + (const char *)extdheader, datasize); + lha->sconv_dir = + archive_string_conversion_from_charset(&a->archive, + "UTF-16LE", 1); + if (lha->sconv_dir == NULL) + return (ARCHIVE_FATAL); + else { + /* + * Convert directory delimiter from 0xFFFF + * to '/' for local system. + */ + uint16_t dirSep; + uint16_t d = 1; + if (archive_be16dec(&d) == 1) + dirSep = 0x2F00; + else + dirSep = 0x002F; + + /* UTF-16LE character */ + uint16_t *utf16name = + (uint16_t *)lha->dirname.s; + for (i = 0; i < lha->dirname.length / 2; i++) { + if (utf16name[i] == 0xFFFF) { + utf16name[i] = dirSep; + } + } + /* Is last character directory separator? */ + if (utf16name[lha->dirname.length / 2 - 1] != + dirSep) { + /* invalid directory data */ + goto invalid; + } + } + break; + case EXT_DOS_ATTR: + if (datasize == 2) + lha->dos_attr = (unsigned char) + (archive_le16dec(extdheader) & 0xff); + break; + case EXT_TIMESTAMP: + if (datasize == (sizeof(uint64_t) * 3)) { + lha->birthtime = lha_win_time( + archive_le64dec(extdheader), + &lha->birthtime_tv_nsec); + extdheader += sizeof(uint64_t); + lha->mtime = lha_win_time( + archive_le64dec(extdheader), + &lha->mtime_tv_nsec); + extdheader += sizeof(uint64_t); + lha->atime = lha_win_time( + archive_le64dec(extdheader), + &lha->atime_tv_nsec); + lha->setflag |= BIRTHTIME_IS_SET | + ATIME_IS_SET; + } + break; + case EXT_FILESIZE: + if (datasize == sizeof(uint64_t) * 2) { + lha->compsize = archive_le64dec(extdheader); + extdheader += sizeof(uint64_t); + lha->origsize = archive_le64dec(extdheader); + } + break; + case EXT_CODEPAGE: + /* Get an archived filename charset from codepage. + * This overwrites the charset specified by + * hdrcharset option. */ + if (datasize == sizeof(uint32_t)) { + struct archive_string cp; + const char *charset; + + archive_string_init(&cp); + switch (archive_le32dec(extdheader)) { + case 65001: /* UTF-8 */ + charset = "UTF-8"; + break; + default: + archive_string_sprintf(&cp, "CP%d", + (int)archive_le32dec(extdheader)); + charset = cp.s; + break; + } + lha->sconv_dir = + archive_string_conversion_from_charset( + &(a->archive), charset, 1); + lha->sconv_fname = + archive_string_conversion_from_charset( + &(a->archive), charset, 1); + archive_string_free(&cp); + if (lha->sconv_dir == NULL) + return (ARCHIVE_FATAL); + if (lha->sconv_fname == NULL) + return (ARCHIVE_FATAL); + } + break; + case EXT_UNIX_MODE: + if (datasize == sizeof(uint16_t)) { + lha->mode = archive_le16dec(extdheader); + lha->setflag |= UNIX_MODE_IS_SET; + } + break; + case EXT_UNIX_GID_UID: + if (datasize == (sizeof(uint16_t) * 2)) { + lha->gid = archive_le16dec(extdheader); + lha->uid = archive_le16dec(extdheader+2); + } + break; + case EXT_UNIX_GNAME: + if (datasize > 0) + archive_strncpy(&lha->gname, + (const char *)extdheader, datasize); + break; + case EXT_UNIX_UNAME: + if (datasize > 0) + archive_strncpy(&lha->uname, + (const char *)extdheader, datasize); + break; + case EXT_UNIX_MTIME: + if (datasize == sizeof(uint32_t)) + lha->mtime = archive_le32dec(extdheader); + break; + case EXT_OS2_NEW_ATTR: + /* This extended header is OS/2 depend. */ + if (datasize == 16) { + lha->dos_attr = (unsigned char) + (archive_le16dec(extdheader) & 0xff); + lha->mode = archive_le16dec(extdheader+2); + lha->gid = archive_le16dec(extdheader+4); + lha->uid = archive_le16dec(extdheader+6); + lha->birthtime = archive_le32dec(extdheader+8); + lha->atime = archive_le32dec(extdheader+12); + lha->setflag |= UNIX_MODE_IS_SET + | BIRTHTIME_IS_SET | ATIME_IS_SET; + } + break; + case EXT_NEW_ATTR: + if (datasize == 20) { + lha->mode = (mode_t)archive_le32dec(extdheader); + lha->gid = archive_le32dec(extdheader+4); + lha->uid = archive_le32dec(extdheader+8); + lha->birthtime = archive_le32dec(extdheader+12); + lha->atime = archive_le32dec(extdheader+16); + lha->setflag |= UNIX_MODE_IS_SET + | BIRTHTIME_IS_SET | ATIME_IS_SET; + } + break; + case EXT_TIMEZONE: /* Not supported */ + break; + default: + break; + } + + __archive_read_consume(a, extdsize); + } +invalid: + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid extended LHa header"); + return (ARCHIVE_FATAL); +} + +static int +lha_end_of_entry(struct archive_read *a) +{ + struct lha *lha = (struct lha *)(a->format->data); + int r = ARCHIVE_EOF; + + if (!lha->end_of_entry_cleanup) { + if ((lha->setflag & CRC_IS_SET) && + lha->crc != lha->entry_crc_calculated) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "LHa data CRC error"); + r = ARCHIVE_WARN; + } + + /* End-of-entry cleanup done. */ + lha->end_of_entry_cleanup = 1; + } + return (r); +} + +static int +archive_read_format_lha_read_data(struct archive_read *a, + const void **buff, size_t *size, int64_t *offset) +{ + struct lha *lha = (struct lha *)(a->format->data); + int r; + + if (lha->entry_unconsumed) { + /* Consume as much as the decompressor actually used. */ + __archive_read_consume(a, lha->entry_unconsumed); + lha->entry_unconsumed = 0; + } + if (lha->end_of_entry) { + *offset = lha->entry_offset; + *size = 0; + *buff = NULL; + return (lha_end_of_entry(a)); + } + + if (lha->entry_is_compressed) + r = lha_read_data_lzh(a, buff, size, offset); + else + /* No compression. */ + r = lha_read_data_none(a, buff, size, offset); + return (r); +} + +/* + * Read a file content in no compression. + * + * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets + * lha->end_of_entry if it consumes all of the data. + */ +static int +lha_read_data_none(struct archive_read *a, const void **buff, + size_t *size, int64_t *offset) +{ + struct lha *lha = (struct lha *)(a->format->data); + ssize_t bytes_avail; + + if (lha->entry_bytes_remaining == 0) { + *buff = NULL; + *size = 0; + *offset = lha->entry_offset; + lha->end_of_entry = 1; + return (ARCHIVE_OK); + } + /* + * Note: '1' here is a performance optimization. + * Recall that the decompression layer returns a count of + * available bytes; asking for more than that forces the + * decompressor to combine reads by copying data. + */ + *buff = __archive_read_ahead(a, 1, &bytes_avail); + if (bytes_avail <= 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated LHa file data"); + return (ARCHIVE_FATAL); + } + if (bytes_avail > lha->entry_bytes_remaining) + bytes_avail = (ssize_t)lha->entry_bytes_remaining; + lha->entry_crc_calculated = + lha_crc16(lha->entry_crc_calculated, *buff, bytes_avail); + *size = bytes_avail; + *offset = lha->entry_offset; + lha->entry_offset += bytes_avail; + lha->entry_bytes_remaining -= bytes_avail; + if (lha->entry_bytes_remaining == 0) + lha->end_of_entry = 1; + lha->entry_unconsumed = bytes_avail; + return (ARCHIVE_OK); +} + +/* + * Read a file content in LZHUFF encoding. + * + * Returns ARCHIVE_OK if successful, returns ARCHIVE_WARN if compression is + * unsupported, ARCHIVE_FATAL otherwise, sets lha->end_of_entry if it consumes + * all of the data. + */ +static int +lha_read_data_lzh(struct archive_read *a, const void **buff, + size_t *size, int64_t *offset) +{ + struct lha *lha = (struct lha *)(a->format->data); + ssize_t bytes_avail; + int r; + + /* If we haven't yet read any data, initialize the decompressor. */ + if (!lha->decompress_init) { + r = lzh_decode_init(&(lha->strm), lha->method); + switch (r) { + case ARCHIVE_OK: + break; + case ARCHIVE_FAILED: + /* Unsupported compression. */ + *buff = NULL; + *size = 0; + *offset = 0; + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Unsupported lzh compression method -%c%c%c-", + lha->method[0], lha->method[1], lha->method[2]); + /* We know compressed size; just skip it. */ + archive_read_format_lha_read_data_skip(a); + return (ARCHIVE_WARN); + default: + archive_set_error(&a->archive, ENOMEM, + "Couldn't allocate memory " + "for lzh decompression"); + return (ARCHIVE_FATAL); + } + /* We've initialized decompression for this stream. */ + lha->decompress_init = 1; + lha->strm.avail_out = 0; + lha->strm.total_out = 0; + } + + /* + * Note: '1' here is a performance optimization. + * Recall that the decompression layer returns a count of + * available bytes; asking for more than that forces the + * decompressor to combine reads by copying data. + */ + lha->strm.next_in = __archive_read_ahead(a, 1, &bytes_avail); + if (bytes_avail <= 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated LHa file body"); + return (ARCHIVE_FATAL); + } + if (bytes_avail > lha->entry_bytes_remaining) + bytes_avail = (ssize_t)lha->entry_bytes_remaining; + + lha->strm.avail_in = (int)bytes_avail; + lha->strm.total_in = 0; + lha->strm.avail_out = 0; + + r = lzh_decode(&(lha->strm), bytes_avail == lha->entry_bytes_remaining); + switch (r) { + case ARCHIVE_OK: + break; + case ARCHIVE_EOF: + lha->end_of_entry = 1; + break; + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Bad lzh data"); + return (ARCHIVE_FAILED); + } + lha->entry_unconsumed = lha->strm.total_in; + lha->entry_bytes_remaining -= lha->strm.total_in; + + if (lha->strm.avail_out) { + *offset = lha->entry_offset; + *size = lha->strm.avail_out; + *buff = lha->strm.ref_ptr; + lha->entry_crc_calculated = + lha_crc16(lha->entry_crc_calculated, *buff, *size); + lha->entry_offset += *size; + } else { + *offset = lha->entry_offset; + *size = 0; + *buff = NULL; + if (lha->end_of_entry) + return (lha_end_of_entry(a)); + } + return (ARCHIVE_OK); +} + +/* + * Skip a file content. + */ +static int +archive_read_format_lha_read_data_skip(struct archive_read *a) +{ + struct lha *lha; + int64_t bytes_skipped; + + lha = (struct lha *)(a->format->data); + + if (lha->entry_unconsumed) { + /* Consume as much as the decompressor actually used. */ + __archive_read_consume(a, lha->entry_unconsumed); + lha->entry_unconsumed = 0; + } + + /* if we've already read to end of data, we're done. */ + if (lha->end_of_entry_cleanup) + return (ARCHIVE_OK); + + /* + * If the length is at the beginning, we can skip the + * compressed data much more quickly. + */ + bytes_skipped = __archive_read_consume(a, lha->entry_bytes_remaining); + if (bytes_skipped < 0) + return (ARCHIVE_FATAL); + + /* This entry is finished and done. */ + lha->end_of_entry_cleanup = lha->end_of_entry = 1; + return (ARCHIVE_OK); +} + +static int +archive_read_format_lha_cleanup(struct archive_read *a) +{ + struct lha *lha = (struct lha *)(a->format->data); + + lzh_decode_free(&(lha->strm)); + archive_string_free(&(lha->dirname)); + archive_string_free(&(lha->filename)); + archive_string_free(&(lha->uname)); + archive_string_free(&(lha->gname)); + archive_wstring_free(&(lha->ws)); + free(lha); + (a->format->data) = NULL; + return (ARCHIVE_OK); +} + +/* + * 'LHa for UNIX' utility has archived a symbolic-link name after + * a pathname with '|' character. + * This function extracts the symbolic-link name from the pathname. + * + * example. + * 1. a symbolic-name is 'aaa/bb/cc' + * 2. a filename is 'xxx/bbb' + * then a archived pathname is 'xxx/bbb|aaa/bb/cc' + */ +static int +lha_parse_linkname(struct archive_wstring *linkname, + struct archive_wstring *pathname) +{ + wchar_t * linkptr; + size_t symlen; + + linkptr = wcschr(pathname->s, L'|'); + if (linkptr != NULL) { + symlen = wcslen(linkptr + 1); + archive_wstrncpy(linkname, linkptr+1, symlen); + + *linkptr = 0; + pathname->length = wcslen(pathname->s); + + return (1); + } + return (0); +} + +/* Convert an MSDOS-style date/time into Unix-style time. */ +static time_t +lha_dos_time(const unsigned char *p) +{ + int msTime, msDate; + struct tm ts; + + msTime = archive_le16dec(p); + msDate = archive_le16dec(p+2); + + memset(&ts, 0, sizeof(ts)); + ts.tm_year = ((msDate >> 9) & 0x7f) + 80; /* Years since 1900. */ + ts.tm_mon = ((msDate >> 5) & 0x0f) - 1; /* Month number. */ + ts.tm_mday = msDate & 0x1f; /* Day of month. */ + ts.tm_hour = (msTime >> 11) & 0x1f; + ts.tm_min = (msTime >> 5) & 0x3f; + ts.tm_sec = (msTime << 1) & 0x3e; + ts.tm_isdst = -1; + return (mktime(&ts)); +} + +/* Convert an MS-Windows-style date/time into Unix-style time. */ +static time_t +lha_win_time(uint64_t wintime, long *ns) +{ +#define EPOC_TIME ARCHIVE_LITERAL_ULL(116444736000000000) + + if (wintime >= EPOC_TIME) { + wintime -= EPOC_TIME; /* 1970-01-01 00:00:00 (UTC) */ + if (ns != NULL) + *ns = (long)(wintime % 10000000) * 100; + return (wintime / 10000000); + } else { + if (ns != NULL) + *ns = 0; + return (0); + } +} + +static unsigned char +lha_calcsum(unsigned char sum, const void *pp, int offset, size_t size) +{ + unsigned char const *p = (unsigned char const *)pp; + + p += offset; + for (;size > 0; --size) + sum += *p++; + return (sum); +} + +static uint16_t crc16tbl[2][256]; +static void +lha_crc16_init(void) +{ + unsigned int i; + static int crc16init = 0; + + if (crc16init) + return; + crc16init = 1; + + for (i = 0; i < 256; i++) { + unsigned int j; + uint16_t crc = (uint16_t)i; + for (j = 8; j; j--) + crc = (crc >> 1) ^ ((crc & 1) * 0xA001); + crc16tbl[0][i] = crc; + } + + for (i = 0; i < 256; i++) { + crc16tbl[1][i] = (crc16tbl[0][i] >> 8) + ^ crc16tbl[0][crc16tbl[0][i] & 0xff]; + } +} + +static uint16_t +lha_crc16(uint16_t crc, const void *pp, size_t len) +{ + const unsigned char *p = (const unsigned char *)pp; + const uint16_t *buff; + const union { + uint32_t i; + char c[4]; + } u = { 0x01020304 }; + + if (len == 0) + return crc; + + /* Process unaligned address. */ + if (((uintptr_t)p) & (uintptr_t)0x1) { + crc = (crc >> 8) ^ crc16tbl[0][(crc ^ *p++) & 0xff]; + len--; + } + buff = (const uint16_t *)p; + /* + * Modern C compiler such as GCC does not unroll automatically yet + * without unrolling pragma, and Clang is so. So we should + * unroll this loop for its performance. + */ + for (;len >= 8; len -= 8) { + /* This if statement expects compiler optimization will + * remove the statement which will not be executed. */ +#undef bswap16 +#if defined(_MSC_VER) && _MSC_VER >= 1400 /* Visual Studio */ +# define bswap16(x) _byteswap_ushort(x) +#elif defined(__GNUC__) && ((__GNUC__ == 4 && __GNUC_MINOR__ >= 8) || __GNUC__ > 4) +/* GCC 4.8 and later has __builtin_bswap16() */ +# define bswap16(x) __builtin_bswap16(x) +#elif defined(__clang__) +/* All clang versions have __builtin_bswap16() */ +# define bswap16(x) __builtin_bswap16(x) +#else +# define bswap16(x) ((((x) >> 8) & 0xff) | ((x) << 8)) +#endif +#define CRC16W do { \ + if(u.c[0] == 1) { /* Big endian */ \ + crc ^= bswap16(*buff); buff++; \ + } else \ + crc ^= *buff++; \ + crc = crc16tbl[1][crc & 0xff] ^ crc16tbl[0][crc >> 8];\ +} while (0) + CRC16W; + CRC16W; + CRC16W; + CRC16W; +#undef CRC16W +#undef bswap16 + } + + p = (const unsigned char *)buff; + for (;len; len--) { + crc = (crc >> 8) ^ crc16tbl[0][(crc ^ *p++) & 0xff]; + } + return crc; +} + +/* + * Initialize LZHUF decoder. + * + * Returns ARCHIVE_OK if initialization was successful. + * Returns ARCHIVE_FAILED if method is unsupported. + * Returns ARCHIVE_FATAL if initialization failed; memory allocation + * error occurred. + */ +static int +lzh_decode_init(struct lzh_stream *strm, const char *method) +{ + struct lzh_dec *ds; + int w_bits, w_size; + + if (strm->ds == NULL) { + strm->ds = calloc(1, sizeof(*strm->ds)); + if (strm->ds == NULL) + return (ARCHIVE_FATAL); + } + ds = strm->ds; + ds->error = ARCHIVE_FAILED; + if (method == NULL || method[0] != 'l' || method[1] != 'h') + return (ARCHIVE_FAILED); + switch (method[2]) { + case '5': + w_bits = 13;/* 8KiB for window */ + break; + case '6': + w_bits = 15;/* 32KiB for window */ + break; + case '7': + w_bits = 16;/* 64KiB for window */ + break; + default: + return (ARCHIVE_FAILED);/* Not supported. */ + } + ds->error = ARCHIVE_FATAL; + /* Expand a window size up to 128 KiB for decompressing process + * performance whatever its original window size is. */ + ds->w_size = 1U << 17; + ds->w_mask = ds->w_size -1; + if (ds->w_buff == NULL) { + ds->w_buff = malloc(ds->w_size); + if (ds->w_buff == NULL) + return (ARCHIVE_FATAL); + } + w_size = 1U << w_bits; + memset(ds->w_buff + ds->w_size - w_size, 0x20, w_size); + ds->w_pos = 0; + ds->state = 0; + ds->pos_pt_len_size = w_bits + 1; + ds->pos_pt_len_bits = (w_bits == 15 || w_bits == 16)? 5: 4; + ds->literal_pt_len_size = PT_BITLEN_SIZE; + ds->literal_pt_len_bits = 5; + ds->br.cache_buffer = 0; + ds->br.cache_avail = 0; + + if (lzh_huffman_init(&(ds->lt), LT_BITLEN_SIZE, 16) + != ARCHIVE_OK) + return (ARCHIVE_FATAL); + ds->lt.len_bits = 9; + if (lzh_huffman_init(&(ds->pt), PT_BITLEN_SIZE, 16) + != ARCHIVE_OK) + return (ARCHIVE_FATAL); + ds->error = 0; + + return (ARCHIVE_OK); +} + +/* + * Release LZHUF decoder. + */ +static void +lzh_decode_free(struct lzh_stream *strm) +{ + + if (strm->ds == NULL) + return; + free(strm->ds->w_buff); + lzh_huffman_free(&(strm->ds->lt)); + lzh_huffman_free(&(strm->ds->pt)); + free(strm->ds); + strm->ds = NULL; +} + +/* + * Bit stream reader. + */ +/* Check that the cache buffer has enough bits. */ +#define lzh_br_has(br, n) ((br)->cache_avail >= n) +/* Get compressed data by bit. */ +#define lzh_br_bits(br, n) \ + (((uint16_t)((br)->cache_buffer >> \ + ((br)->cache_avail - (n)))) & cache_masks[n]) +#define lzh_br_bits_forced(br, n) \ + (((uint16_t)((br)->cache_buffer << \ + ((n) - (br)->cache_avail))) & cache_masks[n]) +/* Read ahead to make sure the cache buffer has enough compressed data we + * will use. + * True : completed, there is enough data in the cache buffer. + * False : we met that strm->next_in is empty, we have to get following + * bytes. */ +#define lzh_br_read_ahead_0(strm, br, n) \ + (lzh_br_has(br, (n)) || lzh_br_fillup(strm, br)) +/* True : the cache buffer has some bits as much as we need. + * False : there are no enough bits in the cache buffer to be used, + * we have to get following bytes if we could. */ +#define lzh_br_read_ahead(strm, br, n) \ + (lzh_br_read_ahead_0((strm), (br), (n)) || lzh_br_has((br), (n))) + +/* Notify how many bits we consumed. */ +#define lzh_br_consume(br, n) ((br)->cache_avail -= (n)) +#define lzh_br_unconsume(br, n) ((br)->cache_avail += (n)) + +static const uint16_t cache_masks[] = { + 0x0000, 0x0001, 0x0003, 0x0007, + 0x000F, 0x001F, 0x003F, 0x007F, + 0x00FF, 0x01FF, 0x03FF, 0x07FF, + 0x0FFF, 0x1FFF, 0x3FFF, 0x7FFF, + 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF +}; + +/* + * Shift away used bits in the cache data and fill it up with following bits. + * Call this when cache buffer does not have enough bits you need. + * + * Returns 1 if the cache buffer is full. + * Returns 0 if the cache buffer is not full; input buffer is empty. + */ +static int +lzh_br_fillup(struct lzh_stream *strm, struct lzh_br *br) +{ + int n = CACHE_BITS - br->cache_avail; + + for (;;) { + const int x = n >> 3; + if (strm->avail_in >= x) { + switch (x) { + case 8: + br->cache_buffer = + ((uint64_t)strm->next_in[0]) << 56 | + ((uint64_t)strm->next_in[1]) << 48 | + ((uint64_t)strm->next_in[2]) << 40 | + ((uint64_t)strm->next_in[3]) << 32 | + ((uint32_t)strm->next_in[4]) << 24 | + ((uint32_t)strm->next_in[5]) << 16 | + ((uint32_t)strm->next_in[6]) << 8 | + (uint32_t)strm->next_in[7]; + strm->next_in += 8; + strm->avail_in -= 8; + br->cache_avail += 8 * 8; + return (1); + case 7: + br->cache_buffer = + (br->cache_buffer << 56) | + ((uint64_t)strm->next_in[0]) << 48 | + ((uint64_t)strm->next_in[1]) << 40 | + ((uint64_t)strm->next_in[2]) << 32 | + ((uint32_t)strm->next_in[3]) << 24 | + ((uint32_t)strm->next_in[4]) << 16 | + ((uint32_t)strm->next_in[5]) << 8 | + (uint32_t)strm->next_in[6]; + strm->next_in += 7; + strm->avail_in -= 7; + br->cache_avail += 7 * 8; + return (1); + case 6: + br->cache_buffer = + (br->cache_buffer << 48) | + ((uint64_t)strm->next_in[0]) << 40 | + ((uint64_t)strm->next_in[1]) << 32 | + ((uint32_t)strm->next_in[2]) << 24 | + ((uint32_t)strm->next_in[3]) << 16 | + ((uint32_t)strm->next_in[4]) << 8 | + (uint32_t)strm->next_in[5]; + strm->next_in += 6; + strm->avail_in -= 6; + br->cache_avail += 6 * 8; + return (1); + case 0: + /* We have enough compressed data in + * the cache buffer.*/ + return (1); + default: + break; + } + } + if (strm->avail_in == 0) { + /* There is not enough compressed data to fill up the + * cache buffer. */ + return (0); + } + br->cache_buffer = + (br->cache_buffer << 8) | *strm->next_in++; + strm->avail_in--; + br->cache_avail += 8; + n -= 8; + } +} + +/* + * Decode LZHUF. + * + * 1. Returns ARCHIVE_OK if output buffer or input buffer are empty. + * Please set available buffer and call this function again. + * 2. Returns ARCHIVE_EOF if decompression has been completed. + * 3. Returns ARCHIVE_FAILED if an error occurred; compressed data + * is broken or you do not set 'last' flag properly. + * 4. 'last' flag is very important, you must set 1 to the flag if there + * is no input data. The lha compressed data format does not provide how + * to know the compressed data is really finished. + * Note: lha command utility check if the total size of output bytes is + * reached the uncompressed size recorded in its header. it does not mind + * that the decoding process is properly finished. + * GNU ZIP can decompress another compressed file made by SCO LZH compress. + * it handles EOF as null to fill read buffer with zero until the decoding + * process meet 2 bytes of zeros at reading a size of a next chunk, so the + * zeros are treated as the mark of the end of the data although the zeros + * is dummy, not the file data. + */ +static int lzh_read_blocks(struct lzh_stream *, int); +static int lzh_decode_blocks(struct lzh_stream *, int); +#define ST_RD_BLOCK 0 +#define ST_RD_PT_1 1 +#define ST_RD_PT_2 2 +#define ST_RD_PT_3 3 +#define ST_RD_PT_4 4 +#define ST_RD_LITERAL_1 5 +#define ST_RD_LITERAL_2 6 +#define ST_RD_LITERAL_3 7 +#define ST_RD_POS_DATA_1 8 +#define ST_GET_LITERAL 9 +#define ST_GET_POS_1 10 +#define ST_GET_POS_2 11 +#define ST_COPY_DATA 12 + +static int +lzh_decode(struct lzh_stream *strm, int last) +{ + struct lzh_dec *ds = strm->ds; + int avail_in; + int r; + + if (ds->error) + return (ds->error); + + avail_in = strm->avail_in; + do { + if (ds->state < ST_GET_LITERAL) + r = lzh_read_blocks(strm, last); + else + r = lzh_decode_blocks(strm, last); + } while (r == 100); + strm->total_in += avail_in - strm->avail_in; + return (r); +} + +static void +lzh_emit_window(struct lzh_stream *strm, size_t s) +{ + strm->ref_ptr = strm->ds->w_buff; + strm->avail_out = (int)s; + strm->total_out += s; +} + +static int +lzh_read_blocks(struct lzh_stream *strm, int last) +{ + struct lzh_dec *ds = strm->ds; + struct lzh_br *br = &(ds->br); + int c = 0, i; + unsigned rbits; + + for (;;) { + switch (ds->state) { + case ST_RD_BLOCK: + /* + * Read a block number indicates how many blocks + * we will handle. The block is composed of a + * literal and a match, sometimes a literal only + * in particular, there are no reference data at + * the beginning of the decompression. + */ + if (!lzh_br_read_ahead_0(strm, br, 16)) { + if (!last) + /* We need following data. */ + return (ARCHIVE_OK); + if (lzh_br_has(br, 8)) { + /* + * It seems there are extra bits. + * 1. Compressed data is broken. + * 2. `last' flag does not properly + * set. + */ + goto failed; + } + if (ds->w_pos > 0) { + lzh_emit_window(strm, ds->w_pos); + ds->w_pos = 0; + return (ARCHIVE_OK); + } + /* End of compressed data; we have completely + * handled all compressed data. */ + return (ARCHIVE_EOF); + } + ds->blocks_avail = lzh_br_bits(br, 16); + if (ds->blocks_avail == 0) + goto failed; + lzh_br_consume(br, 16); + /* + * Read a literal table compressed in huffman + * coding. + */ + ds->pt.len_size = ds->literal_pt_len_size; + ds->pt.len_bits = ds->literal_pt_len_bits; + ds->reading_position = 0; + /* FALL THROUGH */ + case ST_RD_PT_1: + /* Note: ST_RD_PT_1, ST_RD_PT_2 and ST_RD_PT_4 are + * used in reading both a literal table and a + * position table. */ + if (!lzh_br_read_ahead(strm, br, ds->pt.len_bits)) { + if (last) + goto failed;/* Truncated data. */ + ds->state = ST_RD_PT_1; + return (ARCHIVE_OK); + } + ds->pt.len_avail = lzh_br_bits(br, ds->pt.len_bits); + lzh_br_consume(br, ds->pt.len_bits); + /* FALL THROUGH */ + case ST_RD_PT_2: + if (ds->pt.len_avail == 0) { + /* There is no bitlen. */ + if (!lzh_br_read_ahead(strm, br, + ds->pt.len_bits)) { + if (last) + goto failed;/* Truncated data.*/ + ds->state = ST_RD_PT_2; + return (ARCHIVE_OK); + } + if (!lzh_make_fake_table(&(ds->pt), + lzh_br_bits(br, ds->pt.len_bits))) + goto failed;/* Invalid data. */ + lzh_br_consume(br, ds->pt.len_bits); + if (ds->reading_position) + ds->state = ST_GET_LITERAL; + else + ds->state = ST_RD_LITERAL_1; + break; + } else if (ds->pt.len_avail > ds->pt.len_size) + goto failed;/* Invalid data. */ + ds->loop = 0; + memset(ds->pt.freq, 0, sizeof(ds->pt.freq)); + if (ds->pt.len_avail < 3 || + ds->pt.len_size == ds->pos_pt_len_size) { + ds->state = ST_RD_PT_4; + break; + } + /* FALL THROUGH */ + case ST_RD_PT_3: + ds->loop = lzh_read_pt_bitlen(strm, ds->loop, 3); + if (ds->loop < 3) { + if (ds->loop < 0 || last) + goto failed;/* Invalid data. */ + /* Not completed, get following data. */ + ds->state = ST_RD_PT_3; + return (ARCHIVE_OK); + } + /* There are some null in bitlen of the literal. */ + if (!lzh_br_read_ahead(strm, br, 2)) { + if (last) + goto failed;/* Truncated data. */ + ds->state = ST_RD_PT_3; + return (ARCHIVE_OK); + } + c = lzh_br_bits(br, 2); + lzh_br_consume(br, 2); + if (c > ds->pt.len_avail - 3) + goto failed;/* Invalid data. */ + for (i = 3; c-- > 0 ;) + ds->pt.bitlen[i++] = 0; + ds->loop = i; + /* FALL THROUGH */ + case ST_RD_PT_4: + ds->loop = lzh_read_pt_bitlen(strm, ds->loop, + ds->pt.len_avail); + if (ds->loop < ds->pt.len_avail) { + if (ds->loop < 0 || last) + goto failed;/* Invalid data. */ + /* Not completed, get following data. */ + ds->state = ST_RD_PT_4; + return (ARCHIVE_OK); + } + if (!lzh_make_huffman_table(&(ds->pt))) + goto failed;/* Invalid data */ + if (ds->reading_position) { + ds->state = ST_GET_LITERAL; + break; + } + /* FALL THROUGH */ + case ST_RD_LITERAL_1: + if (!lzh_br_read_ahead(strm, br, ds->lt.len_bits)) { + if (last) + goto failed;/* Truncated data. */ + ds->state = ST_RD_LITERAL_1; + return (ARCHIVE_OK); + } + ds->lt.len_avail = lzh_br_bits(br, ds->lt.len_bits); + lzh_br_consume(br, ds->lt.len_bits); + /* FALL THROUGH */ + case ST_RD_LITERAL_2: + if (ds->lt.len_avail == 0) { + /* There is no bitlen. */ + if (!lzh_br_read_ahead(strm, br, + ds->lt.len_bits)) { + if (last) + goto failed;/* Truncated data.*/ + ds->state = ST_RD_LITERAL_2; + return (ARCHIVE_OK); + } + if (!lzh_make_fake_table(&(ds->lt), + lzh_br_bits(br, ds->lt.len_bits))) + goto failed;/* Invalid data */ + lzh_br_consume(br, ds->lt.len_bits); + ds->state = ST_RD_POS_DATA_1; + break; + } else if (ds->lt.len_avail > ds->lt.len_size) + goto failed;/* Invalid data */ + ds->loop = 0; + memset(ds->lt.freq, 0, sizeof(ds->lt.freq)); + /* FALL THROUGH */ + case ST_RD_LITERAL_3: + i = ds->loop; + while (i < ds->lt.len_avail) { + if (!lzh_br_read_ahead(strm, br, + ds->pt.max_bits)) { + if (last) + goto failed;/* Truncated data.*/ + ds->loop = i; + ds->state = ST_RD_LITERAL_3; + return (ARCHIVE_OK); + } + rbits = lzh_br_bits(br, ds->pt.max_bits); + c = lzh_decode_huffman(&(ds->pt), rbits); + if (c > 2) { + /* Note: 'c' will never be more than + * eighteen since it's limited by + * PT_BITLEN_SIZE, which is being set + * to ds->pt.len_size through + * ds->literal_pt_len_size. */ + lzh_br_consume(br, ds->pt.bitlen[c]); + c -= 2; + ds->lt.freq[c]++; + ds->lt.bitlen[i++] = c; + } else if (c == 0) { + lzh_br_consume(br, ds->pt.bitlen[c]); + ds->lt.bitlen[i++] = 0; + } else { + /* c == 1 or c == 2 */ + int n = (c == 1)?4:9; + if (!lzh_br_read_ahead(strm, br, + ds->pt.bitlen[c] + n)) { + if (last) /* Truncated data. */ + goto failed; + ds->loop = i; + ds->state = ST_RD_LITERAL_3; + return (ARCHIVE_OK); + } + lzh_br_consume(br, ds->pt.bitlen[c]); + c = lzh_br_bits(br, n); + lzh_br_consume(br, n); + c += (n == 4)?3:20; + if (i + c > ds->lt.len_avail) + goto failed;/* Invalid data */ + memset(&(ds->lt.bitlen[i]), 0, c); + i += c; + } + } + if (i > ds->lt.len_avail || + !lzh_make_huffman_table(&(ds->lt))) + goto failed;/* Invalid data */ + /* FALL THROUGH */ + case ST_RD_POS_DATA_1: + /* + * Read a position table compressed in huffman + * coding. + */ + ds->pt.len_size = ds->pos_pt_len_size; + ds->pt.len_bits = ds->pos_pt_len_bits; + ds->reading_position = 1; + ds->state = ST_RD_PT_1; + break; + case ST_GET_LITERAL: + return (100); + } + } +failed: + return (ds->error = ARCHIVE_FAILED); +} + +static int +lzh_decode_blocks(struct lzh_stream *strm, int last) +{ + struct lzh_dec *ds = strm->ds; + struct lzh_br bre = ds->br; + struct huffman *lt = &(ds->lt); + struct huffman *pt = &(ds->pt); + unsigned char *w_buff = ds->w_buff; + unsigned char *lt_bitlen = lt->bitlen; + unsigned char *pt_bitlen = pt->bitlen; + int blocks_avail = ds->blocks_avail, c = 0; + int copy_len = ds->copy_len, copy_pos = ds->copy_pos; + int w_pos = ds->w_pos, w_mask = ds->w_mask, w_size = ds->w_size; + int lt_max_bits = lt->max_bits, pt_max_bits = pt->max_bits; + int state = ds->state; + + for (;;) { + switch (state) { + case ST_GET_LITERAL: + for (;;) { + if (blocks_avail == 0) { + /* We have decoded all blocks. + * Let's handle next blocks. */ + ds->state = ST_RD_BLOCK; + ds->br = bre; + ds->blocks_avail = 0; + ds->w_pos = w_pos; + ds->copy_pos = 0; + return (100); + } + + /* lzh_br_read_ahead() always try to fill the + * cache buffer up. In specific situation we + * are close to the end of the data, the cache + * buffer will not be full and thus we have to + * determine if the cache buffer has some bits + * as much as we need after lzh_br_read_ahead() + * failed. */ + if (!lzh_br_read_ahead(strm, &bre, + lt_max_bits)) { + if (!last) + goto next_data; + /* Remaining bits are less than + * maximum bits(lt.max_bits) but maybe + * it still remains as much as we need, + * so we should try to use it with + * dummy bits. */ + c = lzh_decode_huffman(lt, + lzh_br_bits_forced(&bre, + lt_max_bits)); + lzh_br_consume(&bre, lt_bitlen[c]); + if (!lzh_br_has(&bre, 0)) + goto failed;/* Over read. */ + } else { + c = lzh_decode_huffman(lt, + lzh_br_bits(&bre, lt_max_bits)); + lzh_br_consume(&bre, lt_bitlen[c]); + } + blocks_avail--; + if (c > UCHAR_MAX) + /* Current block is a match data. */ + break; + /* + * 'c' is exactly a literal code. + */ + /* Save a decoded code to reference it + * afterward. */ + w_buff[w_pos] = c; + if (++w_pos >= w_size) { + w_pos = 0; + lzh_emit_window(strm, w_size); + goto next_data; + } + } + /* 'c' is the length of a match pattern we have + * already extracted, which has be stored in + * window(ds->w_buff). */ + copy_len = c - (UCHAR_MAX + 1) + MINMATCH; + /* FALL THROUGH */ + case ST_GET_POS_1: + /* + * Get a reference position. + */ + if (!lzh_br_read_ahead(strm, &bre, pt_max_bits)) { + if (!last) { + state = ST_GET_POS_1; + ds->copy_len = copy_len; + goto next_data; + } + copy_pos = lzh_decode_huffman(pt, + lzh_br_bits_forced(&bre, pt_max_bits)); + lzh_br_consume(&bre, pt_bitlen[copy_pos]); + if (!lzh_br_has(&bre, 0)) + goto failed;/* Over read. */ + } else { + copy_pos = lzh_decode_huffman(pt, + lzh_br_bits(&bre, pt_max_bits)); + lzh_br_consume(&bre, pt_bitlen[copy_pos]); + } + /* FALL THROUGH */ + case ST_GET_POS_2: + if (copy_pos > 1) { + /* We need an additional adjustment number to + * the position. */ + int p = copy_pos - 1; + if (!lzh_br_read_ahead(strm, &bre, p)) { + if (last) + goto failed;/* Truncated data.*/ + state = ST_GET_POS_2; + ds->copy_len = copy_len; + ds->copy_pos = copy_pos; + goto next_data; + } + copy_pos = (1 << p) + lzh_br_bits(&bre, p); + lzh_br_consume(&bre, p); + } + /* The position is actually a distance from the last + * code we had extracted and thus we have to convert + * it to a position of the window. */ + copy_pos = (w_pos - copy_pos - 1) & w_mask; + /* FALL THROUGH */ + case ST_COPY_DATA: + /* + * Copy `copy_len' bytes as extracted data from + * the window into the output buffer. + */ + for (;;) { + int l; + + l = copy_len; + if (copy_pos > w_pos) { + if (l > w_size - copy_pos) + l = w_size - copy_pos; + } else { + if (l > w_size - w_pos) + l = w_size - w_pos; + } + if ((copy_pos + l < w_pos) + || (w_pos + l < copy_pos)) { + /* No overlap. */ + memcpy(w_buff + w_pos, + w_buff + copy_pos, l); + } else { + const unsigned char *s; + unsigned char *d; + int li; + + d = w_buff + w_pos; + s = w_buff + copy_pos; + for (li = 0; li < l-1;) { + d[li] = s[li];li++; + d[li] = s[li];li++; + } + if (li < l) + d[li] = s[li]; + } + w_pos += l; + if (w_pos == w_size) { + w_pos = 0; + lzh_emit_window(strm, w_size); + if (copy_len <= l) + state = ST_GET_LITERAL; + else { + state = ST_COPY_DATA; + ds->copy_len = copy_len - l; + ds->copy_pos = + (copy_pos + l) & w_mask; + } + goto next_data; + } + if (copy_len <= l) + /* A copy of current pattern ended. */ + break; + copy_len -= l; + copy_pos = (copy_pos + l) & w_mask; + } + state = ST_GET_LITERAL; + break; + } + } +failed: + return (ds->error = ARCHIVE_FAILED); +next_data: + ds->br = bre; + ds->blocks_avail = blocks_avail; + ds->state = state; + ds->w_pos = w_pos; + return (ARCHIVE_OK); +} + +static int +lzh_huffman_init(struct huffman *hf, size_t len_size, int tbl_bits) +{ + int bits; + + if (hf->bitlen == NULL) { + hf->bitlen = malloc(len_size * sizeof(hf->bitlen[0])); + if (hf->bitlen == NULL) + return (ARCHIVE_FATAL); + } + if (hf->tbl == NULL) { + if (tbl_bits < HTBL_BITS) + bits = tbl_bits; + else + bits = HTBL_BITS; + hf->tbl = malloc(((size_t)1 << bits) * sizeof(hf->tbl[0])); + if (hf->tbl == NULL) + return (ARCHIVE_FATAL); + } + if (hf->tree == NULL && tbl_bits > HTBL_BITS) { + hf->tree_avail = 1 << (tbl_bits - HTBL_BITS + 4); + hf->tree = malloc(hf->tree_avail * sizeof(hf->tree[0])); + if (hf->tree == NULL) + return (ARCHIVE_FATAL); + } + hf->len_size = (int)len_size; + hf->tbl_bits = tbl_bits; + return (ARCHIVE_OK); +} + +static void +lzh_huffman_free(struct huffman *hf) +{ + free(hf->bitlen); + free(hf->tbl); + free(hf->tree); +} + +static const char bitlen_tbl[0x400] = { + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, + 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, + 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, + 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, + 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, + 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, + 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, + 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, + 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, + 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, + 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, + 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, + 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, + 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, + 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, + 13, 13, 13, 13, 13, 13, 13, 13, 14, 14, 14, 14, 15, 15, 16, 0 +}; +static int +lzh_read_pt_bitlen(struct lzh_stream *strm, int start, int end) +{ + struct lzh_dec *ds = strm->ds; + struct lzh_br *br = &(ds->br); + int c, i; + + for (i = start; i < end; ) { + /* + * bit pattern the number we need + * 000 -> 0 + * 001 -> 1 + * 010 -> 2 + * ... + * 110 -> 6 + * 1110 -> 7 + * 11110 -> 8 + * ... + * 1111111111110 -> 16 + */ + if (!lzh_br_read_ahead(strm, br, 3)) + return (i); + if ((c = lzh_br_bits(br, 3)) == 7) { + if (!lzh_br_read_ahead(strm, br, 13)) + return (i); + c = bitlen_tbl[lzh_br_bits(br, 13) & 0x3FF]; + if (c) + lzh_br_consume(br, c - 3); + else + return (-1);/* Invalid data. */ + } else + lzh_br_consume(br, 3); + ds->pt.bitlen[i++] = c; + ds->pt.freq[c]++; + } + return (i); +} + +static int +lzh_make_fake_table(struct huffman *hf, uint16_t c) +{ + if (c >= hf->len_size) + return (0); + hf->tbl[0] = c; + hf->max_bits = 0; + hf->shift_bits = 0; + hf->bitlen[hf->tbl[0]] = 0; + return (1); +} + +/* + * Make a huffman coding table. + */ +static int +lzh_make_huffman_table(struct huffman *hf) +{ + uint16_t *tbl; + const unsigned char *bitlen; + int bitptn[17], weight[17]; + int i, maxbits = 0, ptn, tbl_size, w; + int diffbits, len_avail; + + /* + * Initialize bit patterns. + */ + ptn = 0; + for (i = 1, w = 1 << 15; i <= 16; i++, w >>= 1) { + bitptn[i] = ptn; + weight[i] = w; + if (hf->freq[i]) { + ptn += hf->freq[i] * w; + maxbits = i; + } + } + if (ptn != 0x10000 || maxbits > hf->tbl_bits) + return (0);/* Invalid */ + + hf->max_bits = maxbits; + + /* + * Cut out extra bits which we won't house in the table. + * This preparation reduces the same calculation in the for-loop + * making the table. + */ + if (maxbits < 16) { + int ebits = 16 - maxbits; + for (i = 1; i <= maxbits; i++) { + bitptn[i] >>= ebits; + weight[i] >>= ebits; + } + } + if (maxbits > HTBL_BITS) { + unsigned htbl_max; + uint16_t *p; + + diffbits = maxbits - HTBL_BITS; + for (i = 1; i <= HTBL_BITS; i++) { + bitptn[i] >>= diffbits; + weight[i] >>= diffbits; + } + htbl_max = bitptn[HTBL_BITS] + + weight[HTBL_BITS] * hf->freq[HTBL_BITS]; + p = &(hf->tbl[htbl_max]); + while (p < &hf->tbl[1U<shift_bits = diffbits; + + /* + * Make the table. + */ + tbl_size = 1 << HTBL_BITS; + tbl = hf->tbl; + bitlen = hf->bitlen; + len_avail = hf->len_avail; + hf->tree_used = 0; + for (i = 0; i < len_avail; i++) { + uint16_t *p; + int len, cnt; + uint16_t bit; + int extlen; + struct htree_t *ht; + + if (bitlen[i] == 0) + continue; + /* Get a bit pattern */ + len = bitlen[i]; + ptn = bitptn[len]; + cnt = weight[len]; + if (len <= HTBL_BITS) { + /* Calculate next bit pattern */ + if ((bitptn[len] = ptn + cnt) > tbl_size) + return (0);/* Invalid */ + /* Update the table */ + p = &(tbl[ptn]); + if (cnt > 7) { + uint16_t *pc; + + cnt -= 8; + pc = &p[cnt]; + pc[0] = (uint16_t)i; + pc[1] = (uint16_t)i; + pc[2] = (uint16_t)i; + pc[3] = (uint16_t)i; + pc[4] = (uint16_t)i; + pc[5] = (uint16_t)i; + pc[6] = (uint16_t)i; + pc[7] = (uint16_t)i; + if (cnt > 7) { + cnt -= 8; + memcpy(&p[cnt], pc, + 8 * sizeof(uint16_t)); + pc = &p[cnt]; + while (cnt > 15) { + cnt -= 16; + memcpy(&p[cnt], pc, + 16 * sizeof(uint16_t)); + } + } + if (cnt) + memcpy(p, pc, cnt * sizeof(uint16_t)); + } else { + while (cnt > 1) { + p[--cnt] = (uint16_t)i; + p[--cnt] = (uint16_t)i; + } + if (cnt) + p[--cnt] = (uint16_t)i; + } + continue; + } + + /* + * A bit length is too big to be housed to a direct table, + * so we use a tree model for its extra bits. + */ + bitptn[len] = ptn + cnt; + bit = 1U << (diffbits -1); + extlen = len - HTBL_BITS; + + p = &(tbl[ptn >> diffbits]); + if (*p == 0) { + *p = len_avail + hf->tree_used; + ht = &(hf->tree[hf->tree_used++]); + if (hf->tree_used > hf->tree_avail) + return (0);/* Invalid */ + ht->left = 0; + ht->right = 0; + } else { + if (*p < len_avail || + *p >= (len_avail + hf->tree_used)) + return (0);/* Invalid */ + ht = &(hf->tree[*p - len_avail]); + } + while (--extlen > 0) { + if (ptn & bit) { + if (ht->left < len_avail) { + ht->left = len_avail + hf->tree_used; + ht = &(hf->tree[hf->tree_used++]); + if (hf->tree_used > hf->tree_avail) + return (0);/* Invalid */ + ht->left = 0; + ht->right = 0; + } else { + ht = &(hf->tree[ht->left - len_avail]); + } + } else { + if (ht->right < len_avail) { + ht->right = len_avail + hf->tree_used; + ht = &(hf->tree[hf->tree_used++]); + if (hf->tree_used > hf->tree_avail) + return (0);/* Invalid */ + ht->left = 0; + ht->right = 0; + } else { + ht = &(hf->tree[ht->right - len_avail]); + } + } + bit >>= 1; + } + if (ptn & bit) { + if (ht->left != 0) + return (0);/* Invalid */ + ht->left = (uint16_t)i; + } else { + if (ht->right != 0) + return (0);/* Invalid */ + ht->right = (uint16_t)i; + } + } + return (1); +} + +static int +lzh_decode_huffman_tree(struct huffman *hf, unsigned rbits, int c) +{ + struct htree_t *ht; + int extlen; + + ht = hf->tree; + extlen = hf->shift_bits; + while (c >= hf->len_avail) { + c -= hf->len_avail; + if (extlen-- <= 0 || c >= hf->tree_used) + return (0); + if (rbits & (1U << extlen)) + c = ht[c].left; + else + c = ht[c].right; + } + return (c); +} + +static inline int +lzh_decode_huffman(struct huffman *hf, unsigned rbits) +{ + int c; + /* + * At first search an index table for a bit pattern. + * If it fails, search a huffman tree for. + */ + c = hf->tbl[rbits >> hf->shift_bits]; + if (c < hf->len_avail || hf->len_avail == 0) + return (c); + /* This bit pattern needs to be found out at a huffman tree. */ + return (lzh_decode_huffman_tree(hf, rbits, c)); +} + diff --git a/c/archive_read_support_format_mtree.c b/c/archive_read_support_format_mtree.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_format_mtree.c @@ -0,0 +1,2025 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * Copyright (c) 2008 Joerg Sonnenberger + * Copyright (c) 2011-2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_read_support_format_mtree.c 201165 2009-12-29 05:52:13Z kientzle $"); + +#ifdef HAVE_SYS_STAT_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_FCNTL_H +#include +#endif +#include +/* #include */ /* See archive_platform.h */ +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_CTYPE_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_rb.h" +#include "archive_read_private.h" +#include "archive_string.h" +#include "archive_pack_dev.h" + +#ifndef O_BINARY +#define O_BINARY 0 +#endif +#ifndef O_CLOEXEC +#define O_CLOEXEC 0 +#endif + +#define MTREE_HAS_DEVICE 0x0001 +#define MTREE_HAS_FFLAGS 0x0002 +#define MTREE_HAS_GID 0x0004 +#define MTREE_HAS_GNAME 0x0008 +#define MTREE_HAS_MTIME 0x0010 +#define MTREE_HAS_NLINK 0x0020 +#define MTREE_HAS_PERM 0x0040 +#define MTREE_HAS_SIZE 0x0080 +#define MTREE_HAS_TYPE 0x0100 +#define MTREE_HAS_UID 0x0200 +#define MTREE_HAS_UNAME 0x0400 + +#define MTREE_HAS_OPTIONAL 0x0800 +#define MTREE_HAS_NOCHANGE 0x1000 /* FreeBSD specific */ + +#define MAX_LINE_LEN (1024 * 1024) + +struct mtree_option { + struct mtree_option *next; + char *value; +}; + +struct mtree_entry { + struct archive_rb_node rbnode; + struct mtree_entry *next_dup; + struct mtree_entry *next; + struct mtree_option *options; + char *name; + char full; + char used; +}; + +struct mtree { + struct archive_string line; + size_t buffsize; + char *buff; + int64_t offset; + int fd; + int archive_format; + const char *archive_format_name; + struct mtree_entry *entries; + struct mtree_entry *this_entry; + struct archive_rb_tree entry_rbtree; + struct archive_string current_dir; + struct archive_string contents_name; + + struct archive_entry_linkresolver *resolver; + struct archive_rb_tree rbtree; + + int64_t cur_size; + char checkfs; +}; + +static int bid_keycmp(const char *, const char *, ssize_t); +static int cleanup(struct archive_read *); +static int detect_form(struct archive_read *, int *); +static int mtree_bid(struct archive_read *, int); +static int parse_file(struct archive_read *, struct archive_entry *, + struct mtree *, struct mtree_entry *, int *); +static void parse_escapes(char *, struct mtree_entry *); +static int parse_line(struct archive_read *, struct archive_entry *, + struct mtree *, struct mtree_entry *, int *); +static int parse_keyword(struct archive_read *, struct mtree *, + struct archive_entry *, struct mtree_option *, int *); +static int read_data(struct archive_read *a, + const void **buff, size_t *size, int64_t *offset); +static ssize_t readline(struct archive_read *, struct mtree *, char **, ssize_t); +static int skip(struct archive_read *a); +static int read_header(struct archive_read *, + struct archive_entry *); +static int64_t mtree_atol(char **, int base); + +/* + * There's no standard for TIME_T_MAX/TIME_T_MIN. So we compute them + * here. TODO: Move this to configure time, but be careful + * about cross-compile environments. + */ +static int64_t +get_time_t_max(void) +{ +#if defined(TIME_T_MAX) + return TIME_T_MAX; +#else + /* ISO C allows time_t to be a floating-point type, + but POSIX requires an integer type. The following + should work on any system that follows the POSIX + conventions. */ + if (((time_t)0) < ((time_t)-1)) { + /* Time_t is unsigned */ + return (~(time_t)0); + } else { + /* Time_t is signed. */ + /* Assume it's the same as int64_t or int32_t */ + if (sizeof(time_t) == sizeof(int64_t)) { + return (time_t)INT64_MAX; + } else { + return (time_t)INT32_MAX; + } + } +#endif +} + +static int64_t +get_time_t_min(void) +{ +#if defined(TIME_T_MIN) + return TIME_T_MIN; +#else + if (((time_t)0) < ((time_t)-1)) { + /* Time_t is unsigned */ + return (time_t)0; + } else { + /* Time_t is signed. */ + if (sizeof(time_t) == sizeof(int64_t)) { + return (time_t)INT64_MIN; + } else { + return (time_t)INT32_MIN; + } + } +#endif +} + +static int +archive_read_format_mtree_options(struct archive_read *a, + const char *key, const char *val) +{ + struct mtree *mtree; + + mtree = (struct mtree *)(a->format->data); + if (strcmp(key, "checkfs") == 0) { + /* Allows to read information missing from the mtree from the file system */ + if (val == NULL || val[0] == 0) { + mtree->checkfs = 0; + } else { + mtree->checkfs = 1; + } + return (ARCHIVE_OK); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +static void +free_options(struct mtree_option *head) +{ + struct mtree_option *next; + + for (; head != NULL; head = next) { + next = head->next; + free(head->value); + free(head); + } +} + +static int +mtree_cmp_node(const struct archive_rb_node *n1, + const struct archive_rb_node *n2) +{ + const struct mtree_entry *e1 = (const struct mtree_entry *)n1; + const struct mtree_entry *e2 = (const struct mtree_entry *)n2; + + return (strcmp(e1->name, e2->name)); +} + +static int +mtree_cmp_key(const struct archive_rb_node *n, const void *key) +{ + const struct mtree_entry *e = (const struct mtree_entry *)n; + + return (strcmp(e->name, key)); +} + +int +archive_read_support_format_mtree(struct archive *_a) +{ + static const struct archive_rb_tree_ops rb_ops = { + mtree_cmp_node, mtree_cmp_key, + }; + struct archive_read *a = (struct archive_read *)_a; + struct mtree *mtree; + int r; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_format_mtree"); + + mtree = (struct mtree *)calloc(1, sizeof(*mtree)); + if (mtree == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate mtree data"); + return (ARCHIVE_FATAL); + } + mtree->checkfs = 0; + mtree->fd = -1; + + __archive_rb_tree_init(&mtree->rbtree, &rb_ops); + + r = __archive_read_register_format(a, mtree, "mtree", + mtree_bid, archive_read_format_mtree_options, read_header, read_data, skip, NULL, cleanup, NULL, NULL); + + if (r != ARCHIVE_OK) + free(mtree); + return (ARCHIVE_OK); +} + +static int +cleanup(struct archive_read *a) +{ + struct mtree *mtree; + struct mtree_entry *p, *q; + + mtree = (struct mtree *)(a->format->data); + + p = mtree->entries; + while (p != NULL) { + q = p->next; + free(p->name); + free_options(p->options); + free(p); + p = q; + } + archive_string_free(&mtree->line); + archive_string_free(&mtree->current_dir); + archive_string_free(&mtree->contents_name); + archive_entry_linkresolver_free(mtree->resolver); + + free(mtree->buff); + free(mtree); + (a->format->data) = NULL; + return (ARCHIVE_OK); +} + +static ssize_t +get_line_size(const char *b, ssize_t avail, ssize_t *nlsize) +{ + ssize_t len; + + len = 0; + while (len < avail) { + switch (*b) { + case '\0':/* Non-ascii character or control character. */ + if (nlsize != NULL) + *nlsize = 0; + return (-1); + case '\r': + if (avail-len > 1 && b[1] == '\n') { + if (nlsize != NULL) + *nlsize = 2; + return (len+2); + } + /* FALL THROUGH */ + case '\n': + if (nlsize != NULL) + *nlsize = 1; + return (len+1); + default: + b++; + len++; + break; + } + } + if (nlsize != NULL) + *nlsize = 0; + return (avail); +} + +/* + * <---------------- ravail ---------------------> + * <-- diff ------> <--- avail -----------------> + * <---- len -----------> + * | Previous lines | line being parsed nl extra | + * ^ + * b + * + */ +static ssize_t +next_line(struct archive_read *a, + const char **b, ssize_t *avail, ssize_t *ravail, ssize_t *nl) +{ + ssize_t len; + int quit; + + quit = 0; + if (*avail == 0) { + *nl = 0; + len = 0; + } else + len = get_line_size(*b, *avail, nl); + /* + * Read bytes more while it does not reach the end of line. + */ + while (*nl == 0 && len == *avail && !quit) { + ssize_t diff = *ravail - *avail; + size_t nbytes_req = (*ravail+1023) & ~1023U; + ssize_t tested; + + /* + * Place an arbitrary limit on the line length. + * mtree is almost free-form input and without line length limits, + * it can consume a lot of memory. + */ + if (len >= MAX_LINE_LEN) + return (-1); + + /* Increase reading bytes if it is not enough to at least + * new two lines. */ + if (nbytes_req < (size_t)*ravail + 160) + nbytes_req <<= 1; + + *b = __archive_read_ahead(a, nbytes_req, avail); + if (*b == NULL) { + if (*ravail >= *avail) + return (0); + /* Reading bytes reaches the end of file. */ + *b = __archive_read_ahead(a, *avail, avail); + quit = 1; + } + *ravail = *avail; + *b += diff; + *avail -= diff; + tested = len;/* Skip some bytes we already determinated. */ + len = get_line_size(*b + len, *avail - len, nl); + if (len >= 0) + len += tested; + } + return (len); +} + +/* + * Compare characters with a mtree keyword. + * Returns the length of a mtree keyword if matched. + * Returns 0 if not matched. + */ +static int +bid_keycmp(const char *p, const char *key, ssize_t len) +{ + int match_len = 0; + + while (len > 0 && *p && *key) { + if (*p == *key) { + --len; + ++p; + ++key; + ++match_len; + continue; + } + return (0);/* Not match */ + } + if (*key != '\0') + return (0);/* Not match */ + + /* A following character should be specified characters */ + if (p[0] == '=' || p[0] == ' ' || p[0] == '\t' || + p[0] == '\n' || p[0] == '\r' || + (p[0] == '\\' && (p[1] == '\n' || p[1] == '\r'))) + return (match_len); + return (0);/* Not match */ +} + +/* + * Test whether the characters 'p' has is mtree keyword. + * Returns the length of a detected keyword. + * Returns 0 if any keywords were not found. + */ +static int +bid_keyword(const char *p, ssize_t len) +{ + static const char * const keys_c[] = { + "content", "contents", "cksum", NULL + }; + static const char * const keys_df[] = { + "device", "flags", NULL + }; + static const char * const keys_g[] = { + "gid", "gname", NULL + }; + static const char * const keys_il[] = { + "ignore", "inode", "link", NULL + }; + static const char * const keys_m[] = { + "md5", "md5digest", "mode", NULL + }; + static const char * const keys_no[] = { + "nlink", "nochange", "optional", NULL + }; + static const char * const keys_r[] = { + "resdevice", "rmd160", "rmd160digest", NULL + }; + static const char * const keys_s[] = { + "sha1", "sha1digest", + "sha256", "sha256digest", + "sha384", "sha384digest", + "sha512", "sha512digest", + "size", NULL + }; + static const char * const keys_t[] = { + "tags", "time", "type", NULL + }; + static const char * const keys_u[] = { + "uid", "uname", NULL + }; + const char * const *keys; + int i; + + switch (*p) { + case 'c': keys = keys_c; break; + case 'd': case 'f': keys = keys_df; break; + case 'g': keys = keys_g; break; + case 'i': case 'l': keys = keys_il; break; + case 'm': keys = keys_m; break; + case 'n': case 'o': keys = keys_no; break; + case 'r': keys = keys_r; break; + case 's': keys = keys_s; break; + case 't': keys = keys_t; break; + case 'u': keys = keys_u; break; + default: return (0);/* Unknown key */ + } + + for (i = 0; keys[i] != NULL; i++) { + int l = bid_keycmp(p, keys[i], len); + if (l > 0) + return (l); + } + return (0);/* Unknown key */ +} + +/* + * Test whether there is a set of mtree keywords. + * Returns the number of keyword. + * Returns -1 if we got incorrect sequence. + * This function expects a set of "keyword=value". + * When "unset" is specified, expects a set of "keyword". + */ +static int +bid_keyword_list(const char *p, ssize_t len, int unset, int last_is_path) +{ + int l; + int keycnt = 0; + + while (len > 0 && *p) { + int blank = 0; + + /* Test whether there are blank characters in the line. */ + while (len >0 && (*p == ' ' || *p == '\t')) { + ++p; + --len; + blank = 1; + } + if (*p == '\n' || *p == '\r') + break; + if (p[0] == '\\' && (p[1] == '\n' || p[1] == '\r')) + break; + if (!blank && !last_is_path) /* No blank character. */ + return (-1); + if (last_is_path && len == 0) + return (keycnt); + + if (unset) { + l = bid_keycmp(p, "all", len); + if (l > 0) + return (1); + } + /* Test whether there is a correct key in the line. */ + l = bid_keyword(p, len); + if (l == 0) + return (-1);/* Unknown keyword was found. */ + p += l; + len -= l; + keycnt++; + + /* Skip value */ + if (*p == '=') { + int value = 0; + ++p; + --len; + while (len > 0 && *p != ' ' && *p != '\t') { + ++p; + --len; + value = 1; + } + /* A keyword should have a its value unless + * "/unset" operation. */ + if (!unset && value == 0) + return (-1); + } + } + return (keycnt); +} + +static int +bid_entry(const char *p, ssize_t len, ssize_t nl, int *last_is_path) +{ + int f = 0; + static const unsigned char safe_char[256] = { + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 00 - 0F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 10 - 1F */ + /* !"$%&'()*+,-./ EXCLUSION:( )(#) */ + 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 20 - 2F */ + /* 0123456789:;<>? EXCLUSION:(=) */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, /* 30 - 3F */ + /* @ABCDEFGHIJKLMNO */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 40 - 4F */ + /* PQRSTUVWXYZ[\]^_ */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 50 - 5F */ + /* `abcdefghijklmno */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 60 - 6F */ + /* pqrstuvwxyz{|}~ */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, /* 70 - 7F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 80 - 8F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 90 - 9F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* A0 - AF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* B0 - BF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* C0 - CF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* D0 - DF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* E0 - EF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* F0 - FF */ + }; + ssize_t ll; + const char *pp = p; + const char * const pp_end = pp + len; + + *last_is_path = 0; + /* + * Skip the path-name which is quoted. + */ + for (;pp < pp_end; ++pp) { + if (!safe_char[*(const unsigned char *)pp]) { + if (*pp != ' ' && *pp != '\t' && *pp != '\r' + && *pp != '\n') + f = 0; + break; + } + f = 1; + } + ll = pp_end - pp; + + /* If a path-name was not found at the first, try to check + * a mtree format(a.k.a form D) ``NetBSD's mtree -D'' creates, + * which places the path-name at the last. */ + if (f == 0) { + const char *pb = p + len - nl; + int name_len = 0; + int slash; + + /* The form D accepts only a single line for an entry. */ + if (pb-2 >= p && + pb[-1] == '\\' && (pb[-2] == ' ' || pb[-2] == '\t')) + return (-1); + if (pb-1 >= p && pb[-1] == '\\') + return (-1); + + slash = 0; + while (p <= --pb && *pb != ' ' && *pb != '\t') { + if (!safe_char[*(const unsigned char *)pb]) + return (-1); + name_len++; + /* The pathname should have a slash in this + * format. */ + if (*pb == '/') + slash = 1; + } + if (name_len == 0 || slash == 0) + return (-1); + /* If '/' is placed at the first in this field, this is not + * a valid filename. */ + if (pb[1] == '/') + return (-1); + ll = len - nl - name_len; + pp = p; + *last_is_path = 1; + } + + return (bid_keyword_list(pp, ll, 0, *last_is_path)); +} + +#define MAX_BID_ENTRY 3 + +static int +mtree_bid(struct archive_read *a, int best_bid) +{ + const char *signature = "#mtree"; + const char *p; + + (void)best_bid; /* UNUSED */ + + /* Now let's look at the actual header and see if it matches. */ + p = __archive_read_ahead(a, strlen(signature), NULL); + if (p == NULL) + return (-1); + + if (memcmp(p, signature, strlen(signature)) == 0) + return (8 * (int)strlen(signature)); + + /* + * There is not a mtree signature. Let's try to detect mtree format. + */ + return (detect_form(a, NULL)); +} + +static int +detect_form(struct archive_read *a, int *is_form_d) +{ + const char *p; + ssize_t avail, ravail; + ssize_t detected_bytes = 0, len, nl; + int entry_cnt = 0, multiline = 0; + int form_D = 0;/* The archive is generated by `NetBSD mtree -D' + * (In this source we call it `form D') . */ + + if (is_form_d != NULL) + *is_form_d = 0; + p = __archive_read_ahead(a, 1, &avail); + if (p == NULL) + return (-1); + ravail = avail; + for (;;) { + len = next_line(a, &p, &avail, &ravail, &nl); + /* The terminal character of the line should be + * a new line character, '\r\n' or '\n'. */ + if (len <= 0 || nl == 0) + break; + if (!multiline) { + /* Leading whitespace is never significant, + * ignore it. */ + while (len > 0 && (*p == ' ' || *p == '\t')) { + ++p; + --avail; + --len; + } + /* Skip comment or empty line. */ + if (p[0] == '#' || p[0] == '\n' || p[0] == '\r') { + p += len; + avail -= len; + continue; + } + } else { + /* A continuance line; the terminal + * character of previous line was '\' character. */ + if (bid_keyword_list(p, len, 0, 0) <= 0) + break; + if (multiline == 1) + detected_bytes += len; + if (p[len-nl-1] != '\\') { + if (multiline == 1 && + ++entry_cnt >= MAX_BID_ENTRY) + break; + multiline = 0; + } + p += len; + avail -= len; + continue; + } + if (p[0] != '/') { + int last_is_path, keywords; + + keywords = bid_entry(p, len, nl, &last_is_path); + if (keywords >= 0) { + detected_bytes += len; + if (form_D == 0) { + if (last_is_path) + form_D = 1; + else if (keywords > 0) + /* This line is not `form D'. */ + form_D = -1; + } else if (form_D == 1) { + if (!last_is_path && keywords > 0) + /* This this is not `form D' + * and We cannot accept mixed + * format. */ + break; + } + if (!last_is_path && p[len-nl-1] == '\\') + /* This line continues. */ + multiline = 1; + else { + /* We've got plenty of correct lines + * to assume that this file is a mtree + * format. */ + if (++entry_cnt >= MAX_BID_ENTRY) + break; + } + } else + break; + } else if (len > 4 && strncmp(p, "/set", 4) == 0) { + if (bid_keyword_list(p+4, len-4, 0, 0) <= 0) + break; + /* This line continues. */ + if (p[len-nl-1] == '\\') + multiline = 2; + } else if (len > 6 && strncmp(p, "/unset", 6) == 0) { + if (bid_keyword_list(p+6, len-6, 1, 0) <= 0) + break; + /* This line continues. */ + if (p[len-nl-1] == '\\') + multiline = 2; + } else + break; + + /* Test next line. */ + p += len; + avail -= len; + } + if (entry_cnt >= MAX_BID_ENTRY || (entry_cnt > 0 && len == 0)) { + if (is_form_d != NULL) { + if (form_D == 1) + *is_form_d = 1; + } + return (32); + } + + return (0); +} + +/* + * The extended mtree format permits multiple lines specifying + * attributes for each file. For those entries, only the last line + * is actually used. Practically speaking, that means we have + * to read the entire mtree file into memory up front. + * + * The parsing is done in two steps. First, it is decided if a line + * changes the global defaults and if it is, processed accordingly. + * Otherwise, the options of the line are merged with the current + * global options. + */ +static int +add_option(struct archive_read *a, struct mtree_option **global, + const char *value, size_t len) +{ + struct mtree_option *opt; + + if ((opt = malloc(sizeof(*opt))) == NULL) { + archive_set_error(&a->archive, errno, "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + if ((opt->value = malloc(len + 1)) == NULL) { + free(opt); + archive_set_error(&a->archive, errno, "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + memcpy(opt->value, value, len); + opt->value[len] = '\0'; + opt->next = *global; + *global = opt; + return (ARCHIVE_OK); +} + +static void +remove_option(struct mtree_option **global, const char *value, size_t len) +{ + struct mtree_option *iter, *last; + + last = NULL; + for (iter = *global; iter != NULL; last = iter, iter = iter->next) { + if (strncmp(iter->value, value, len) == 0 && + (iter->value[len] == '\0' || + iter->value[len] == '=')) + break; + } + if (iter == NULL) + return; + if (last == NULL) + *global = iter->next; + else + last->next = iter->next; + + free(iter->value); + free(iter); +} + +static int +process_global_set(struct archive_read *a, + struct mtree_option **global, const char *line) +{ + const char *next, *eq; + size_t len; + int r; + + line += 4; + for (;;) { + next = line + strspn(line, " \t\r\n"); + if (*next == '\0') + return (ARCHIVE_OK); + line = next; + next = line + strcspn(line, " \t\r\n"); + eq = strchr(line, '='); + if (eq > next) + len = next - line; + else + len = eq - line; + + remove_option(global, line, len); + r = add_option(a, global, line, next - line); + if (r != ARCHIVE_OK) + return (r); + line = next; + } +} + +static int +process_global_unset(struct archive_read *a, + struct mtree_option **global, const char *line) +{ + const char *next; + size_t len; + + line += 6; + if (strchr(line, '=') != NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "/unset shall not contain `='"); + return ARCHIVE_FATAL; + } + + for (;;) { + next = line + strspn(line, " \t\r\n"); + if (*next == '\0') + return (ARCHIVE_OK); + line = next; + len = strcspn(line, " \t\r\n"); + + if (len == 3 && strncmp(line, "all", 3) == 0) { + free_options(*global); + *global = NULL; + } else { + remove_option(global, line, len); + } + + line += len; + } +} + +static int +process_add_entry(struct archive_read *a, struct mtree *mtree, + struct mtree_option **global, const char *line, ssize_t line_len, + struct mtree_entry **last_entry, int is_form_d) +{ + struct mtree_entry *entry; + struct mtree_option *iter; + const char *next, *eq, *name, *end; + size_t name_len, len; + int r, i; + + if ((entry = malloc(sizeof(*entry))) == NULL) { + archive_set_error(&a->archive, errno, "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + entry->next = NULL; + entry->options = NULL; + entry->name = NULL; + entry->used = 0; + entry->full = 0; + + /* Add this entry to list. */ + if (*last_entry == NULL) + mtree->entries = entry; + else + (*last_entry)->next = entry; + *last_entry = entry; + + if (is_form_d) { + /* Filename is last item on line. */ + /* Adjust line_len to trim trailing whitespace */ + while (line_len > 0) { + char last_character = line[line_len - 1]; + if (last_character == '\r' + || last_character == '\n' + || last_character == '\t' + || last_character == ' ') { + line_len--; + } else { + break; + } + } + /* Name starts after the last whitespace separator */ + name = line; + for (i = 0; i < line_len; i++) { + if (line[i] == '\r' + || line[i] == '\n' + || line[i] == '\t' + || line[i] == ' ') { + name = line + i + 1; + } + } + name_len = line + line_len - name; + end = name; + } else { + /* Filename is first item on line */ + name_len = strcspn(line, " \t\r\n"); + name = line; + line += name_len; + end = line + line_len; + } + /* name/name_len is the name within the line. */ + /* line..end brackets the entire line except the name */ + + if ((entry->name = malloc(name_len + 1)) == NULL) { + archive_set_error(&a->archive, errno, "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + + memcpy(entry->name, name, name_len); + entry->name[name_len] = '\0'; + parse_escapes(entry->name, entry); + + entry->next_dup = NULL; + if (entry->full) { + if (!__archive_rb_tree_insert_node(&mtree->rbtree, &entry->rbnode)) { + struct mtree_entry *alt; + alt = (struct mtree_entry *)__archive_rb_tree_find_node( + &mtree->rbtree, entry->name); + while (alt->next_dup) + alt = alt->next_dup; + alt->next_dup = entry; + } + } + + for (iter = *global; iter != NULL; iter = iter->next) { + r = add_option(a, &entry->options, iter->value, + strlen(iter->value)); + if (r != ARCHIVE_OK) + return (r); + } + + for (;;) { + next = line + strspn(line, " \t\r\n"); + if (*next == '\0') + return (ARCHIVE_OK); + if (next >= end) + return (ARCHIVE_OK); + line = next; + next = line + strcspn(line, " \t\r\n"); + eq = strchr(line, '='); + if (eq == NULL || eq > next) + len = next - line; + else + len = eq - line; + + remove_option(&entry->options, line, len); + r = add_option(a, &entry->options, line, next - line); + if (r != ARCHIVE_OK) + return (r); + line = next; + } +} + +static int +read_mtree(struct archive_read *a, struct mtree *mtree) +{ + ssize_t len; + uintmax_t counter; + char *p, *s; + struct mtree_option *global; + struct mtree_entry *last_entry; + int r, is_form_d; + + mtree->archive_format = ARCHIVE_FORMAT_MTREE; + mtree->archive_format_name = "mtree"; + + global = NULL; + last_entry = NULL; + + (void)detect_form(a, &is_form_d); + + for (counter = 1; ; ++counter) { + r = ARCHIVE_OK; + len = readline(a, mtree, &p, 65536); + if (len == 0) { + mtree->this_entry = mtree->entries; + free_options(global); + return (ARCHIVE_OK); + } + if (len < 0) { + free_options(global); + return ((int)len); + } + /* Leading whitespace is never significant, ignore it. */ + while (*p == ' ' || *p == '\t') { + ++p; + --len; + } + /* Skip content lines and blank lines. */ + if (*p == '#') + continue; + if (*p == '\r' || *p == '\n' || *p == '\0') + continue; + /* Non-printable characters are not allowed */ + for (s = p;s < p + len - 1; s++) { + if (!isprint(*s)) { + r = ARCHIVE_FATAL; + break; + } + } + if (r != ARCHIVE_OK) + break; + if (*p != '/') { + r = process_add_entry(a, mtree, &global, p, len, + &last_entry, is_form_d); + } else if (len > 4 && strncmp(p, "/set", 4) == 0) { + if (p[4] != ' ' && p[4] != '\t') + break; + r = process_global_set(a, &global, p); + } else if (len > 6 && strncmp(p, "/unset", 6) == 0) { + if (p[6] != ' ' && p[6] != '\t') + break; + r = process_global_unset(a, &global, p); + } else + break; + + if (r != ARCHIVE_OK) { + free_options(global); + return r; + } + } + + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Can't parse line %ju", counter); + free_options(global); + return (ARCHIVE_FATAL); +} + +/* + * Read in the entire mtree file into memory on the first request. + * Then use the next unused file to satisfy each header request. + */ +static int +read_header(struct archive_read *a, struct archive_entry *entry) +{ + struct mtree *mtree; + char *p; + int r, use_next; + + mtree = (struct mtree *)(a->format->data); + + if (mtree->fd >= 0) { + close(mtree->fd); + mtree->fd = -1; + } + + if (mtree->entries == NULL) { + mtree->resolver = archive_entry_linkresolver_new(); + if (mtree->resolver == NULL) + return ARCHIVE_FATAL; + archive_entry_linkresolver_set_strategy(mtree->resolver, + ARCHIVE_FORMAT_MTREE); + r = read_mtree(a, mtree); + if (r != ARCHIVE_OK) + return (r); + } + + a->archive.archive_format = mtree->archive_format; + a->archive.archive_format_name = mtree->archive_format_name; + + for (;;) { + if (mtree->this_entry == NULL) + return (ARCHIVE_EOF); + if (strcmp(mtree->this_entry->name, "..") == 0) { + mtree->this_entry->used = 1; + if (archive_strlen(&mtree->current_dir) > 0) { + /* Roll back current path. */ + p = mtree->current_dir.s + + mtree->current_dir.length - 1; + while (p >= mtree->current_dir.s && *p != '/') + --p; + if (p >= mtree->current_dir.s) + --p; + mtree->current_dir.length + = p - mtree->current_dir.s + 1; + } + } + if (!mtree->this_entry->used) { + use_next = 0; + r = parse_file(a, entry, mtree, mtree->this_entry, + &use_next); + if (use_next == 0) + return (r); + } + mtree->this_entry = mtree->this_entry->next; + } +} + +/* + * A single file can have multiple lines contribute specifications. + * Parse as many lines as necessary, then pull additional information + * from a backing file on disk as necessary. + */ +static int +parse_file(struct archive_read *a, struct archive_entry *entry, + struct mtree *mtree, struct mtree_entry *mentry, int *use_next) +{ + const char *path; + struct stat st_storage, *st; + struct mtree_entry *mp; + struct archive_entry *sparse_entry; + int r = ARCHIVE_OK, r1, parsed_kws; + + mentry->used = 1; + + /* Initialize reasonable defaults. */ + archive_entry_set_filetype(entry, AE_IFREG); + archive_entry_set_size(entry, 0); + archive_string_empty(&mtree->contents_name); + + /* Parse options from this line. */ + parsed_kws = 0; + r = parse_line(a, entry, mtree, mentry, &parsed_kws); + + if (mentry->full) { + archive_entry_copy_pathname(entry, mentry->name); + /* + * "Full" entries are allowed to have multiple lines + * and those lines aren't required to be adjacent. We + * don't support multiple lines for "relative" entries + * nor do we make any attempt to merge data from + * separate "relative" and "full" entries. (Merging + * "relative" and "full" entries would require dealing + * with pathname canonicalization, which is a very + * tricky subject.) + */ + mp = (struct mtree_entry *)__archive_rb_tree_find_node( + &mtree->rbtree, mentry->name); + for (; mp; mp = mp->next_dup) { + if (mp->full && !mp->used) { + /* Later lines override earlier ones. */ + mp->used = 1; + r1 = parse_line(a, entry, mtree, mp, &parsed_kws); + if (r1 < r) + r = r1; + } + } + } else { + /* + * Relative entries require us to construct + * the full path and possibly update the + * current directory. + */ + size_t n = archive_strlen(&mtree->current_dir); + if (n > 0) + archive_strcat(&mtree->current_dir, "/"); + archive_strcat(&mtree->current_dir, mentry->name); + archive_entry_copy_pathname(entry, mtree->current_dir.s); + if (archive_entry_filetype(entry) != AE_IFDIR) + mtree->current_dir.length = n; + } + + if (mtree->checkfs) { + /* + * Try to open and stat the file to get the real size + * and other file info. It would be nice to avoid + * this here so that getting a listing of an mtree + * wouldn't require opening every referenced contents + * file. But then we wouldn't know the actual + * contents size, so I don't see a really viable way + * around this. (Also, we may want to someday pull + * other unspecified info from the contents file on + * disk.) + */ + mtree->fd = -1; + if (archive_strlen(&mtree->contents_name) > 0) + path = mtree->contents_name.s; + else + path = archive_entry_pathname(entry); + + if (archive_entry_filetype(entry) == AE_IFREG || + archive_entry_filetype(entry) == AE_IFDIR) { + mtree->fd = open(path, O_RDONLY | O_BINARY | O_CLOEXEC); + __archive_ensure_cloexec_flag(mtree->fd); + if (mtree->fd == -1 && + (errno != ENOENT || + archive_strlen(&mtree->contents_name) > 0)) { + archive_set_error(&a->archive, errno, + "Can't open %s", path); + r = ARCHIVE_WARN; + } + } + + st = &st_storage; + if (mtree->fd >= 0) { + if (fstat(mtree->fd, st) == -1) { + archive_set_error(&a->archive, errno, + "Could not fstat %s", path); + r = ARCHIVE_WARN; + /* If we can't stat it, don't keep it open. */ + close(mtree->fd); + mtree->fd = -1; + st = NULL; + } + } else if (lstat(path, st) == -1) { + st = NULL; + } + + /* + * Check for a mismatch between the type in the specification + * and the type of the contents object on disk. + */ + if (st != NULL) { + if (((st->st_mode & S_IFMT) == S_IFREG && + archive_entry_filetype(entry) == AE_IFREG) +#ifdef S_IFLNK + ||((st->st_mode & S_IFMT) == S_IFLNK && + archive_entry_filetype(entry) == AE_IFLNK) +#endif +#ifdef S_IFSOCK + ||((st->st_mode & S_IFSOCK) == S_IFSOCK && + archive_entry_filetype(entry) == AE_IFSOCK) +#endif +#ifdef S_IFCHR + ||((st->st_mode & S_IFMT) == S_IFCHR && + archive_entry_filetype(entry) == AE_IFCHR) +#endif +#ifdef S_IFBLK + ||((st->st_mode & S_IFMT) == S_IFBLK && + archive_entry_filetype(entry) == AE_IFBLK) +#endif + ||((st->st_mode & S_IFMT) == S_IFDIR && + archive_entry_filetype(entry) == AE_IFDIR) +#ifdef S_IFIFO + ||((st->st_mode & S_IFMT) == S_IFIFO && + archive_entry_filetype(entry) == AE_IFIFO) +#endif + ) { + /* Types match. */ + } else { + /* Types don't match; bail out gracefully. */ + if (mtree->fd >= 0) + close(mtree->fd); + mtree->fd = -1; + if (parsed_kws & MTREE_HAS_OPTIONAL) { + /* It's not an error for an optional + * entry to not match disk. */ + *use_next = 1; + } else if (r == ARCHIVE_OK) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "mtree specification has different" + " type for %s", + archive_entry_pathname(entry)); + r = ARCHIVE_WARN; + } + return (r); + } + } + + /* + * If there is a contents file on disk, pick some of the + * metadata from that file. For most of these, we only + * set it from the contents if it wasn't already parsed + * from the specification. + */ + if (st != NULL) { + if (((parsed_kws & MTREE_HAS_DEVICE) == 0 || + (parsed_kws & MTREE_HAS_NOCHANGE) != 0) && + (archive_entry_filetype(entry) == AE_IFCHR || + archive_entry_filetype(entry) == AE_IFBLK)) + archive_entry_set_rdev(entry, st->st_rdev); + if ((parsed_kws & (MTREE_HAS_GID | MTREE_HAS_GNAME)) + == 0 || + (parsed_kws & MTREE_HAS_NOCHANGE) != 0) + archive_entry_set_gid(entry, st->st_gid); + if ((parsed_kws & (MTREE_HAS_UID | MTREE_HAS_UNAME)) + == 0 || + (parsed_kws & MTREE_HAS_NOCHANGE) != 0) + archive_entry_set_uid(entry, st->st_uid); + if ((parsed_kws & MTREE_HAS_MTIME) == 0 || + (parsed_kws & MTREE_HAS_NOCHANGE) != 0) { +#if HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC + archive_entry_set_mtime(entry, st->st_mtime, + st->st_mtimespec.tv_nsec); +#elif HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC + archive_entry_set_mtime(entry, st->st_mtime, + st->st_mtim.tv_nsec); +#elif HAVE_STRUCT_STAT_ST_MTIME_N + archive_entry_set_mtime(entry, st->st_mtime, + st->st_mtime_n); +#elif HAVE_STRUCT_STAT_ST_UMTIME + archive_entry_set_mtime(entry, st->st_mtime, + st->st_umtime*1000); +#elif HAVE_STRUCT_STAT_ST_MTIME_USEC + archive_entry_set_mtime(entry, st->st_mtime, + st->st_mtime_usec*1000); +#else + archive_entry_set_mtime(entry, st->st_mtime, 0); +#endif + } + if ((parsed_kws & MTREE_HAS_NLINK) == 0 || + (parsed_kws & MTREE_HAS_NOCHANGE) != 0) + archive_entry_set_nlink(entry, st->st_nlink); + if ((parsed_kws & MTREE_HAS_PERM) == 0 || + (parsed_kws & MTREE_HAS_NOCHANGE) != 0) + archive_entry_set_perm(entry, st->st_mode); + if ((parsed_kws & MTREE_HAS_SIZE) == 0 || + (parsed_kws & MTREE_HAS_NOCHANGE) != 0) + archive_entry_set_size(entry, st->st_size); + archive_entry_set_ino(entry, st->st_ino); + archive_entry_set_dev(entry, st->st_dev); + + archive_entry_linkify(mtree->resolver, &entry, + &sparse_entry); + } else if (parsed_kws & MTREE_HAS_OPTIONAL) { + /* + * Couldn't open the entry, stat it or the on-disk type + * didn't match. If this entry is optional, just + * ignore it and read the next header entry. + */ + *use_next = 1; + return ARCHIVE_OK; + } + } + + mtree->cur_size = archive_entry_size(entry); + mtree->offset = 0; + + return r; +} + +/* + * Each line contains a sequence of keywords. + */ +static int +parse_line(struct archive_read *a, struct archive_entry *entry, + struct mtree *mtree, struct mtree_entry *mp, int *parsed_kws) +{ + struct mtree_option *iter; + int r = ARCHIVE_OK, r1; + + for (iter = mp->options; iter != NULL; iter = iter->next) { + r1 = parse_keyword(a, mtree, entry, iter, parsed_kws); + if (r1 < r) + r = r1; + } + if (r == ARCHIVE_OK && (*parsed_kws & MTREE_HAS_TYPE) == 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Missing type keyword in mtree specification"); + return (ARCHIVE_WARN); + } + return (r); +} + +/* + * Device entries have one of the following forms: + * - raw dev_t + * - format,major,minor[,subdevice] + * When parsing succeeded, `pdev' will contain the appropriate dev_t value. + */ + +/* strsep() is not in C90, but strcspn() is. */ +/* Taken from http://unixpapa.com/incnote/string.html */ +static char * +la_strsep(char **sp, const char *sep) +{ + char *p, *s; + if (sp == NULL || *sp == NULL || **sp == '\0') + return(NULL); + s = *sp; + p = s + strcspn(s, sep); + if (*p != '\0') + *p++ = '\0'; + *sp = p; + return(s); +} + +static int +parse_device(dev_t *pdev, struct archive *a, char *val) +{ +#define MAX_PACK_ARGS 3 + unsigned long numbers[MAX_PACK_ARGS]; + char *p, *dev; + int argc; + pack_t *pack; + dev_t result; + const char *error = NULL; + + memset(pdev, 0, sizeof(*pdev)); + if ((dev = strchr(val, ',')) != NULL) { + /* + * Device's major/minor are given in a specified format. + * Decode and pack it accordingly. + */ + *dev++ = '\0'; + if ((pack = pack_find(val)) == NULL) { + archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT, + "Unknown format `%s'", val); + return ARCHIVE_WARN; + } + argc = 0; + while ((p = la_strsep(&dev, ",")) != NULL) { + if (*p == '\0') { + archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT, + "Missing number"); + return ARCHIVE_WARN; + } + if (argc >= MAX_PACK_ARGS) { + archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT, + "Too many arguments"); + return ARCHIVE_WARN; + } + numbers[argc++] = (unsigned long)mtree_atol(&p, 0); + } + if (argc < 2) { + archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT, + "Not enough arguments"); + return ARCHIVE_WARN; + } + result = (*pack)(argc, numbers, &error); + if (error != NULL) { + archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT, + "%s", error); + return ARCHIVE_WARN; + } + } else { + /* file system raw value. */ + result = (dev_t)mtree_atol(&val, 0); + } + *pdev = result; + return ARCHIVE_OK; +#undef MAX_PACK_ARGS +} + +/* + * Parse a single keyword and its value. + */ +static int +parse_keyword(struct archive_read *a, struct mtree *mtree, + struct archive_entry *entry, struct mtree_option *opt, int *parsed_kws) +{ + char *val, *key; + + key = opt->value; + + if (*key == '\0') + return (ARCHIVE_OK); + + if (strcmp(key, "nochange") == 0) { + *parsed_kws |= MTREE_HAS_NOCHANGE; + return (ARCHIVE_OK); + } + if (strcmp(key, "optional") == 0) { + *parsed_kws |= MTREE_HAS_OPTIONAL; + return (ARCHIVE_OK); + } + if (strcmp(key, "ignore") == 0) { + /* + * The mtree processing is not recursive, so + * recursion will only happen for explicitly listed + * entries. + */ + return (ARCHIVE_OK); + } + + val = strchr(key, '='); + if (val == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Malformed attribute \"%s\" (%d)", key, key[0]); + return (ARCHIVE_WARN); + } + + *val = '\0'; + ++val; + + switch (key[0]) { + case 'c': + if (strcmp(key, "content") == 0 + || strcmp(key, "contents") == 0) { + parse_escapes(val, NULL); + archive_strcpy(&mtree->contents_name, val); + break; + } + if (strcmp(key, "cksum") == 0) + break; + __LA_FALLTHROUGH; + case 'd': + if (strcmp(key, "device") == 0) { + /* stat(2) st_rdev field, e.g. the major/minor IDs + * of a char/block special file */ + int r; + dev_t dev; + + *parsed_kws |= MTREE_HAS_DEVICE; + r = parse_device(&dev, &a->archive, val); + if (r == ARCHIVE_OK) + archive_entry_set_rdev(entry, dev); + return r; + } + __LA_FALLTHROUGH; + case 'f': + if (strcmp(key, "flags") == 0) { + *parsed_kws |= MTREE_HAS_FFLAGS; + archive_entry_copy_fflags_text(entry, val); + break; + } + __LA_FALLTHROUGH; + case 'g': + if (strcmp(key, "gid") == 0) { + *parsed_kws |= MTREE_HAS_GID; + archive_entry_set_gid(entry, mtree_atol(&val, 10)); + break; + } + if (strcmp(key, "gname") == 0) { + *parsed_kws |= MTREE_HAS_GNAME; + archive_entry_copy_gname(entry, val); + break; + } + __LA_FALLTHROUGH; + case 'i': + if (strcmp(key, "inode") == 0) { + archive_entry_set_ino(entry, mtree_atol(&val, 10)); + break; + } + __LA_FALLTHROUGH; + case 'l': + if (strcmp(key, "link") == 0) { + archive_entry_copy_symlink(entry, val); + break; + } + __LA_FALLTHROUGH; + case 'm': + if (strcmp(key, "md5") == 0 || strcmp(key, "md5digest") == 0) + break; + if (strcmp(key, "mode") == 0) { + if (val[0] >= '0' && val[0] <= '7') { + *parsed_kws |= MTREE_HAS_PERM; + archive_entry_set_perm(entry, + (mode_t)mtree_atol(&val, 8)); + } else { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Symbolic or non-octal mode \"%s\" unsupported", val); + return ARCHIVE_WARN; + } + break; + } + __LA_FALLTHROUGH; + case 'n': + if (strcmp(key, "nlink") == 0) { + *parsed_kws |= MTREE_HAS_NLINK; + archive_entry_set_nlink(entry, + (unsigned int)mtree_atol(&val, 10)); + break; + } + __LA_FALLTHROUGH; + case 'r': + if (strcmp(key, "resdevice") == 0) { + /* stat(2) st_dev field, e.g. the device ID where the + * inode resides */ + int r; + dev_t dev; + + r = parse_device(&dev, &a->archive, val); + if (r == ARCHIVE_OK) + archive_entry_set_dev(entry, dev); + return r; + } + if (strcmp(key, "rmd160") == 0 || + strcmp(key, "rmd160digest") == 0) + break; + __LA_FALLTHROUGH; + case 's': + if (strcmp(key, "sha1") == 0 || strcmp(key, "sha1digest") == 0) + break; + if (strcmp(key, "sha256") == 0 || + strcmp(key, "sha256digest") == 0) + break; + if (strcmp(key, "sha384") == 0 || + strcmp(key, "sha384digest") == 0) + break; + if (strcmp(key, "sha512") == 0 || + strcmp(key, "sha512digest") == 0) + break; + if (strcmp(key, "size") == 0) { + archive_entry_set_size(entry, mtree_atol(&val, 10)); + break; + } + __LA_FALLTHROUGH; + case 't': + if (strcmp(key, "tags") == 0) { + /* + * Comma delimited list of tags. + * Ignore the tags for now, but the interface + * should be extended to allow inclusion/exclusion. + */ + break; + } + if (strcmp(key, "time") == 0) { + int64_t m; + int64_t my_time_t_max = get_time_t_max(); + int64_t my_time_t_min = get_time_t_min(); + long ns = 0; + + *parsed_kws |= MTREE_HAS_MTIME; + m = mtree_atol(&val, 10); + /* Replicate an old mtree bug: + * 123456789.1 represents 123456789 + * seconds and 1 nanosecond. */ + if (*val == '.') { + ++val; + ns = (long)mtree_atol(&val, 10); + if (ns < 0) + ns = 0; + else if (ns > 999999999) + ns = 999999999; + } + if (m > my_time_t_max) + m = my_time_t_max; + else if (m < my_time_t_min) + m = my_time_t_min; + archive_entry_set_mtime(entry, (time_t)m, ns); + break; + } + if (strcmp(key, "type") == 0) { + switch (val[0]) { + case 'b': + if (strcmp(val, "block") == 0) { + archive_entry_set_filetype(entry, AE_IFBLK); + break; + } + __LA_FALLTHROUGH; + case 'c': + if (strcmp(val, "char") == 0) { + archive_entry_set_filetype(entry, + AE_IFCHR); + break; + } + __LA_FALLTHROUGH; + case 'd': + if (strcmp(val, "dir") == 0) { + archive_entry_set_filetype(entry, + AE_IFDIR); + break; + } + __LA_FALLTHROUGH; + case 'f': + if (strcmp(val, "fifo") == 0) { + archive_entry_set_filetype(entry, + AE_IFIFO); + break; + } + if (strcmp(val, "file") == 0) { + archive_entry_set_filetype(entry, + AE_IFREG); + break; + } + __LA_FALLTHROUGH; + case 'l': + if (strcmp(val, "link") == 0) { + archive_entry_set_filetype(entry, + AE_IFLNK); + break; + } + __LA_FALLTHROUGH; + default: + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Unrecognized file type \"%s\"; " + "assuming \"file\"", val); + archive_entry_set_filetype(entry, AE_IFREG); + return (ARCHIVE_WARN); + } + *parsed_kws |= MTREE_HAS_TYPE; + break; + } + __LA_FALLTHROUGH; + case 'u': + if (strcmp(key, "uid") == 0) { + *parsed_kws |= MTREE_HAS_UID; + archive_entry_set_uid(entry, mtree_atol(&val, 10)); + break; + } + if (strcmp(key, "uname") == 0) { + *parsed_kws |= MTREE_HAS_UNAME; + archive_entry_copy_uname(entry, val); + break; + } + __LA_FALLTHROUGH; + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Unrecognized key %s=%s", key, val); + return (ARCHIVE_WARN); + } + return (ARCHIVE_OK); +} + +static int +read_data(struct archive_read *a, const void **buff, size_t *size, + int64_t *offset) +{ + size_t bytes_to_read; + ssize_t bytes_read; + struct mtree *mtree; + + mtree = (struct mtree *)(a->format->data); + if (mtree->fd < 0) { + *buff = NULL; + *offset = 0; + *size = 0; + return (ARCHIVE_EOF); + } + if (mtree->buff == NULL) { + mtree->buffsize = 64 * 1024; + mtree->buff = malloc(mtree->buffsize); + if (mtree->buff == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + } + + *buff = mtree->buff; + *offset = mtree->offset; + if ((int64_t)mtree->buffsize > mtree->cur_size - mtree->offset) + bytes_to_read = (size_t)(mtree->cur_size - mtree->offset); + else + bytes_to_read = mtree->buffsize; + bytes_read = read(mtree->fd, mtree->buff, bytes_to_read); + if (bytes_read < 0) { + archive_set_error(&a->archive, errno, "Can't read"); + return (ARCHIVE_WARN); + } + if (bytes_read == 0) { + *size = 0; + return (ARCHIVE_EOF); + } + mtree->offset += bytes_read; + *size = bytes_read; + return (ARCHIVE_OK); +} + +/* Skip does nothing except possibly close the contents file. */ +static int +skip(struct archive_read *a) +{ + struct mtree *mtree; + + mtree = (struct mtree *)(a->format->data); + if (mtree->fd >= 0) { + close(mtree->fd); + mtree->fd = -1; + } + return (ARCHIVE_OK); +} + +/* + * Since parsing backslash sequences always makes strings shorter, + * we can always do this conversion in-place. + */ +static void +parse_escapes(char *src, struct mtree_entry *mentry) +{ + char *dest = src; + char c; + + if (mentry != NULL && strcmp(src, ".") == 0) + mentry->full = 1; + + while (*src != '\0') { + c = *src++; + if (c == '/' && mentry != NULL) + mentry->full = 1; + if (c == '\\') { + switch (src[0]) { + case '0': + if (src[1] < '0' || src[1] > '7') { + c = 0; + ++src; + break; + } + /* FALLTHROUGH */ + case '1': + case '2': + case '3': + if (src[1] >= '0' && src[1] <= '7' && + src[2] >= '0' && src[2] <= '7') { + c = (src[0] - '0') << 6; + c |= (src[1] - '0') << 3; + c |= (src[2] - '0'); + src += 3; + } + break; + case 'a': + c = '\a'; + ++src; + break; + case 'b': + c = '\b'; + ++src; + break; + case 'f': + c = '\f'; + ++src; + break; + case 'n': + c = '\n'; + ++src; + break; + case 'r': + c = '\r'; + ++src; + break; + case 's': + c = ' '; + ++src; + break; + case 't': + c = '\t'; + ++src; + break; + case 'v': + c = '\v'; + ++src; + break; + case '\\': + c = '\\'; + ++src; + break; + } + } + *dest++ = c; + } + *dest = '\0'; +} + +/* Parse a hex digit. */ +static int +parsedigit(char c) +{ + if (c >= '0' && c <= '9') + return c - '0'; + else if (c >= 'a' && c <= 'f') + return c - 'a'; + else if (c >= 'A' && c <= 'F') + return c - 'A'; + else + return -1; +} + +/* + * Note that this implementation does not (and should not!) obey + * locale settings; you cannot simply substitute strtol here, since + * it does obey locale. + */ +static int64_t +mtree_atol(char **p, int base) +{ + int64_t l, limit; + int digit, last_digit_limit; + + if (base == 0) { + if (**p != '0') + base = 10; + else if ((*p)[1] == 'x' || (*p)[1] == 'X') { + *p += 2; + base = 16; + } else { + base = 8; + } + } + + if (**p == '-') { + limit = INT64_MIN / base; + last_digit_limit = INT64_MIN % base; + ++(*p); + + l = 0; + digit = parsedigit(**p); + while (digit >= 0 && digit < base) { + if (l < limit || (l == limit && digit > last_digit_limit)) + return INT64_MIN; + l = (l * base) - digit; + digit = parsedigit(*++(*p)); + } + return l; + } else { + limit = INT64_MAX / base; + last_digit_limit = INT64_MAX % base; + + l = 0; + digit = parsedigit(**p); + while (digit >= 0 && digit < base) { + if (l > limit || (l == limit && digit > last_digit_limit)) + return INT64_MAX; + l = (l * base) + digit; + digit = parsedigit(*++(*p)); + } + return l; + } +} + +/* + * Returns length of line (including trailing newline) + * or negative on error. 'start' argument is updated to + * point to first character of line. + */ +static ssize_t +readline(struct archive_read *a, struct mtree *mtree, char **start, + ssize_t limit) +{ + ssize_t bytes_read; + ssize_t total_size = 0; + ssize_t find_off = 0; + const void *t; + void *nl; + char *u; + + /* Accumulate line in a line buffer. */ + for (;;) { + /* Read some more. */ + t = __archive_read_ahead(a, 1, &bytes_read); + if (t == NULL) + return (0); + if (bytes_read < 0) + return (ARCHIVE_FATAL); + nl = memchr(t, '\n', bytes_read); + /* If we found '\n', trim the read to end exactly there. */ + if (nl != NULL) { + bytes_read = ((const char *)nl) - ((const char *)t) + 1; + } + if (total_size + bytes_read + 1 > limit) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Line too long"); + return (ARCHIVE_FATAL); + } + if (archive_string_ensure(&mtree->line, + total_size + bytes_read + 1) == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate working buffer"); + return (ARCHIVE_FATAL); + } + /* Append new bytes to string. */ + memcpy(mtree->line.s + total_size, t, bytes_read); + __archive_read_consume(a, bytes_read); + total_size += bytes_read; + mtree->line.s[total_size] = '\0'; + + for (u = mtree->line.s + find_off; *u; ++u) { + if (u[0] == '\n') { + /* Ends with unescaped newline. */ + *start = mtree->line.s; + return total_size; + } else if (u[0] == '#') { + /* Ends with comment sequence #...\n */ + if (nl == NULL) { + /* But we've not found the \n yet */ + break; + } + } else if (u[0] == '\\') { + if (u[1] == '\n') { + /* Trim escaped newline. */ + total_size -= 2; + mtree->line.s[total_size] = '\0'; + break; + } else if (u[1] != '\0') { + /* Skip the two-char escape sequence */ + ++u; + } + } + } + find_off = u - mtree->line.s; + } +} diff --git a/c/archive_read_support_format_rar.c b/c/archive_read_support_format_rar.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_format_rar.c @@ -0,0 +1,2999 @@ +/*- +* Copyright (c) 2003-2007 Tim Kientzle +* Copyright (c) 2011 Andres Mejia +* All rights reserved. +* +* Redistribution and use in source and binary forms, with or without +* modification, are permitted provided that the following conditions +* are met: +* 1. Redistributions of source code must retain the above copyright +* notice, this list of conditions and the following disclaimer. +* 2. Redistributions in binary form must reproduce the above copyright +* notice, this list of conditions and the following disclaimer in the +* documentation and/or other materials provided with the distribution. +* +* THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR +* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES +* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. +* IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, +* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT +* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF +* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include "archive_platform.h" + +#ifdef HAVE_ERRNO_H +#include +#endif +#include +#include +#ifdef HAVE_ZLIB_H +#include /* crc32 */ +#endif + +#include "archive.h" +#ifndef HAVE_ZLIB_H +#include "archive_crc32.h" +#endif +#include "archive_endian.h" +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_ppmd7_private.h" +#include "archive_private.h" +#include "archive_read_private.h" + +/* RAR signature, also known as the mark header */ +#define RAR_SIGNATURE "\x52\x61\x72\x21\x1A\x07\x00" + +/* Header types */ +#define MARK_HEAD 0x72 +#define MAIN_HEAD 0x73 +#define FILE_HEAD 0x74 +#define COMM_HEAD 0x75 +#define AV_HEAD 0x76 +#define SUB_HEAD 0x77 +#define PROTECT_HEAD 0x78 +#define SIGN_HEAD 0x79 +#define NEWSUB_HEAD 0x7a +#define ENDARC_HEAD 0x7b + +/* Main Header Flags */ +#define MHD_VOLUME 0x0001 +#define MHD_COMMENT 0x0002 +#define MHD_LOCK 0x0004 +#define MHD_SOLID 0x0008 +#define MHD_NEWNUMBERING 0x0010 +#define MHD_AV 0x0020 +#define MHD_PROTECT 0x0040 +#define MHD_PASSWORD 0x0080 +#define MHD_FIRSTVOLUME 0x0100 +#define MHD_ENCRYPTVER 0x0200 + +/* Flags common to all headers */ +#define HD_MARKDELETION 0x4000 +#define HD_ADD_SIZE_PRESENT 0x8000 + +/* File Header Flags */ +#define FHD_SPLIT_BEFORE 0x0001 +#define FHD_SPLIT_AFTER 0x0002 +#define FHD_PASSWORD 0x0004 +#define FHD_COMMENT 0x0008 +#define FHD_SOLID 0x0010 +#define FHD_LARGE 0x0100 +#define FHD_UNICODE 0x0200 +#define FHD_SALT 0x0400 +#define FHD_VERSION 0x0800 +#define FHD_EXTTIME 0x1000 +#define FHD_EXTFLAGS 0x2000 + +/* File dictionary sizes */ +#define DICTIONARY_SIZE_64 0x00 +#define DICTIONARY_SIZE_128 0x20 +#define DICTIONARY_SIZE_256 0x40 +#define DICTIONARY_SIZE_512 0x60 +#define DICTIONARY_SIZE_1024 0x80 +#define DICTIONARY_SIZE_2048 0xA0 +#define DICTIONARY_SIZE_4096 0xC0 +#define FILE_IS_DIRECTORY 0xE0 +#define DICTIONARY_MASK FILE_IS_DIRECTORY + +/* OS Flags */ +#define OS_MSDOS 0 +#define OS_OS2 1 +#define OS_WIN32 2 +#define OS_UNIX 3 +#define OS_MAC_OS 4 +#define OS_BEOS 5 + +/* Compression Methods */ +#define COMPRESS_METHOD_STORE 0x30 +/* LZSS */ +#define COMPRESS_METHOD_FASTEST 0x31 +#define COMPRESS_METHOD_FAST 0x32 +#define COMPRESS_METHOD_NORMAL 0x33 +/* PPMd Variant H */ +#define COMPRESS_METHOD_GOOD 0x34 +#define COMPRESS_METHOD_BEST 0x35 + +#define CRC_POLYNOMIAL 0xEDB88320 + +#define NS_UNIT 10000000 + +#define DICTIONARY_MAX_SIZE 0x400000 + +#define MAINCODE_SIZE 299 +#define OFFSETCODE_SIZE 60 +#define LOWOFFSETCODE_SIZE 17 +#define LENGTHCODE_SIZE 28 +#define HUFFMAN_TABLE_SIZE \ + MAINCODE_SIZE + OFFSETCODE_SIZE + LOWOFFSETCODE_SIZE + LENGTHCODE_SIZE + +#define MAX_SYMBOL_LENGTH 0xF +#define MAX_SYMBOLS 20 + +/* + * Considering L1,L2 cache miss and a calling of write system-call, + * the best size of the output buffer(uncompressed buffer) is 128K. + * If the structure of extracting process is changed, this value + * might be researched again. + */ +#define UNP_BUFFER_SIZE (128 * 1024) + +/* Define this here for non-Windows platforms */ +#if !((defined(__WIN32__) || defined(_WIN32) || defined(__WIN32)) && !defined(__CYGWIN__)) +#define FILE_ATTRIBUTE_DIRECTORY 0x10 +#endif + +#undef minimum +#define minimum(a, b) ((a)<(b)?(a):(b)) + +/* Fields common to all headers */ +struct rar_header +{ + char crc[2]; + char type; + char flags[2]; + char size[2]; +}; + +/* Fields common to all file headers */ +struct rar_file_header +{ + char pack_size[4]; + char unp_size[4]; + char host_os; + char file_crc[4]; + char file_time[4]; + char unp_ver; + char method; + char name_size[2]; + char file_attr[4]; +}; + +struct huffman_tree_node +{ + int branches[2]; +}; + +struct huffman_table_entry +{ + unsigned int length; + int value; +}; + +struct huffman_code +{ + struct huffman_tree_node *tree; + int numentries; + int numallocatedentries; + int minlength; + int maxlength; + int tablesize; + struct huffman_table_entry *table; +}; + +struct lzss +{ + unsigned char *window; + int mask; + int64_t position; +}; + +struct data_block_offsets +{ + int64_t header_size; + int64_t start_offset; + int64_t end_offset; +}; + +struct rar +{ + /* Entries from main RAR header */ + unsigned main_flags; + unsigned long file_crc; + char reserved1[2]; + char reserved2[4]; + char encryptver; + + /* File header entries */ + char compression_method; + unsigned file_flags; + int64_t packed_size; + int64_t unp_size; + time_t mtime; + long mnsec; + mode_t mode; + char *filename; + char *filename_save; + size_t filename_save_size; + size_t filename_allocated; + + /* File header optional entries */ + char salt[8]; + time_t atime; + long ansec; + time_t ctime; + long cnsec; + time_t arctime; + long arcnsec; + + /* Fields to help with tracking decompression of files. */ + int64_t bytes_unconsumed; + int64_t bytes_remaining; + int64_t bytes_uncopied; + int64_t offset; + int64_t offset_outgoing; + int64_t offset_seek; + char valid; + unsigned int unp_offset; + unsigned int unp_buffer_size; + unsigned char *unp_buffer; + unsigned int dictionary_size; + char start_new_block; + char entry_eof; + unsigned long crc_calculated; + int found_first_header; + char has_endarc_header; + struct data_block_offsets *dbo; + unsigned int cursor; + unsigned int nodes; + char filename_must_match; + + /* LZSS members */ + struct huffman_code maincode; + struct huffman_code offsetcode; + struct huffman_code lowoffsetcode; + struct huffman_code lengthcode; + unsigned char lengthtable[HUFFMAN_TABLE_SIZE]; + struct lzss lzss; + char output_last_match; + unsigned int lastlength; + unsigned int lastoffset; + unsigned int oldoffset[4]; + unsigned int lastlowoffset; + unsigned int numlowoffsetrepeats; + int64_t filterstart; + char start_new_table; + + /* PPMd Variant H members */ + char ppmd_valid; + char ppmd_eod; + char is_ppmd_block; + int ppmd_escape; + CPpmd7 ppmd7_context; + CPpmd7z_RangeDec range_dec; + IByteIn bytein; + + /* + * String conversion object. + */ + int init_default_conversion; + struct archive_string_conv *sconv_default; + struct archive_string_conv *opt_sconv; + struct archive_string_conv *sconv_utf8; + struct archive_string_conv *sconv_utf16be; + + /* + * Bit stream reader. + */ + struct rar_br { +#define CACHE_TYPE uint64_t +#define CACHE_BITS (8 * sizeof(CACHE_TYPE)) + /* Cache buffer. */ + CACHE_TYPE cache_buffer; + /* Indicates how many bits avail in cache_buffer. */ + int cache_avail; + ssize_t avail_in; + const unsigned char *next_in; + } br; + + /* + * Custom field to denote that this archive contains encrypted entries + */ + int has_encrypted_entries; +}; + +static int archive_read_support_format_rar_capabilities(struct archive_read *); +static int archive_read_format_rar_has_encrypted_entries(struct archive_read *); +static int archive_read_format_rar_bid(struct archive_read *, int); +static int archive_read_format_rar_options(struct archive_read *, + const char *, const char *); +static int archive_read_format_rar_read_header(struct archive_read *, + struct archive_entry *); +static int archive_read_format_rar_read_data(struct archive_read *, + const void **, size_t *, int64_t *); +static int archive_read_format_rar_read_data_skip(struct archive_read *a); +static int64_t archive_read_format_rar_seek_data(struct archive_read *, int64_t, + int); +static int archive_read_format_rar_cleanup(struct archive_read *); + +/* Support functions */ +static int read_header(struct archive_read *, struct archive_entry *, char); +static time_t get_time(int); +static int read_exttime(const char *, struct rar *, const char *); +static int read_symlink_stored(struct archive_read *, struct archive_entry *, + struct archive_string_conv *); +static int read_data_stored(struct archive_read *, const void **, size_t *, + int64_t *); +static int read_data_compressed(struct archive_read *, const void **, size_t *, + int64_t *); +static int rar_br_preparation(struct archive_read *, struct rar_br *); +static int parse_codes(struct archive_read *); +static void free_codes(struct archive_read *); +static int read_next_symbol(struct archive_read *, struct huffman_code *); +static int create_code(struct archive_read *, struct huffman_code *, + unsigned char *, int, char); +static int add_value(struct archive_read *, struct huffman_code *, int, int, + int); +static int new_node(struct huffman_code *); +static int make_table(struct archive_read *, struct huffman_code *); +static int make_table_recurse(struct archive_read *, struct huffman_code *, int, + struct huffman_table_entry *, int, int); +static int64_t expand(struct archive_read *, int64_t); +static int copy_from_lzss_window(struct archive_read *, const void **, + int64_t, int); +static const void *rar_read_ahead(struct archive_read *, size_t, ssize_t *); + +/* + * Bit stream reader. + */ +/* Check that the cache buffer has enough bits. */ +#define rar_br_has(br, n) ((br)->cache_avail >= n) +/* Get compressed data by bit. */ +#define rar_br_bits(br, n) \ + (((uint32_t)((br)->cache_buffer >> \ + ((br)->cache_avail - (n)))) & cache_masks[n]) +#define rar_br_bits_forced(br, n) \ + (((uint32_t)((br)->cache_buffer << \ + ((n) - (br)->cache_avail))) & cache_masks[n]) +/* Read ahead to make sure the cache buffer has enough compressed data we + * will use. + * True : completed, there is enough data in the cache buffer. + * False : there is no data in the stream. */ +#define rar_br_read_ahead(a, br, n) \ + ((rar_br_has(br, (n)) || rar_br_fillup(a, br)) || rar_br_has(br, (n))) +/* Notify how many bits we consumed. */ +#define rar_br_consume(br, n) ((br)->cache_avail -= (n)) +#define rar_br_consume_unalined_bits(br) ((br)->cache_avail &= ~7) + +static const uint32_t cache_masks[] = { + 0x00000000, 0x00000001, 0x00000003, 0x00000007, + 0x0000000F, 0x0000001F, 0x0000003F, 0x0000007F, + 0x000000FF, 0x000001FF, 0x000003FF, 0x000007FF, + 0x00000FFF, 0x00001FFF, 0x00003FFF, 0x00007FFF, + 0x0000FFFF, 0x0001FFFF, 0x0003FFFF, 0x0007FFFF, + 0x000FFFFF, 0x001FFFFF, 0x003FFFFF, 0x007FFFFF, + 0x00FFFFFF, 0x01FFFFFF, 0x03FFFFFF, 0x07FFFFFF, + 0x0FFFFFFF, 0x1FFFFFFF, 0x3FFFFFFF, 0x7FFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF +}; + +/* + * Shift away used bits in the cache data and fill it up with following bits. + * Call this when cache buffer does not have enough bits you need. + * + * Returns 1 if the cache buffer is full. + * Returns 0 if the cache buffer is not full; input buffer is empty. + */ +static int +rar_br_fillup(struct archive_read *a, struct rar_br *br) +{ + struct rar *rar = (struct rar *)(a->format->data); + int n = CACHE_BITS - br->cache_avail; + + for (;;) { + switch (n >> 3) { + case 8: + if (br->avail_in >= 8) { + br->cache_buffer = + ((uint64_t)br->next_in[0]) << 56 | + ((uint64_t)br->next_in[1]) << 48 | + ((uint64_t)br->next_in[2]) << 40 | + ((uint64_t)br->next_in[3]) << 32 | + ((uint32_t)br->next_in[4]) << 24 | + ((uint32_t)br->next_in[5]) << 16 | + ((uint32_t)br->next_in[6]) << 8 | + (uint32_t)br->next_in[7]; + br->next_in += 8; + br->avail_in -= 8; + br->cache_avail += 8 * 8; + rar->bytes_unconsumed += 8; + rar->bytes_remaining -= 8; + return (1); + } + break; + case 7: + if (br->avail_in >= 7) { + br->cache_buffer = + (br->cache_buffer << 56) | + ((uint64_t)br->next_in[0]) << 48 | + ((uint64_t)br->next_in[1]) << 40 | + ((uint64_t)br->next_in[2]) << 32 | + ((uint32_t)br->next_in[3]) << 24 | + ((uint32_t)br->next_in[4]) << 16 | + ((uint32_t)br->next_in[5]) << 8 | + (uint32_t)br->next_in[6]; + br->next_in += 7; + br->avail_in -= 7; + br->cache_avail += 7 * 8; + rar->bytes_unconsumed += 7; + rar->bytes_remaining -= 7; + return (1); + } + break; + case 6: + if (br->avail_in >= 6) { + br->cache_buffer = + (br->cache_buffer << 48) | + ((uint64_t)br->next_in[0]) << 40 | + ((uint64_t)br->next_in[1]) << 32 | + ((uint32_t)br->next_in[2]) << 24 | + ((uint32_t)br->next_in[3]) << 16 | + ((uint32_t)br->next_in[4]) << 8 | + (uint32_t)br->next_in[5]; + br->next_in += 6; + br->avail_in -= 6; + br->cache_avail += 6 * 8; + rar->bytes_unconsumed += 6; + rar->bytes_remaining -= 6; + return (1); + } + break; + case 0: + /* We have enough compressed data in + * the cache buffer.*/ + return (1); + default: + break; + } + if (br->avail_in <= 0) { + + if (rar->bytes_unconsumed > 0) { + /* Consume as much as the decompressor + * actually used. */ + __archive_read_consume(a, rar->bytes_unconsumed); + rar->bytes_unconsumed = 0; + } + br->next_in = rar_read_ahead(a, 1, &(br->avail_in)); + if (br->next_in == NULL) + return (0); + if (br->avail_in == 0) + return (0); + } + br->cache_buffer = + (br->cache_buffer << 8) | *br->next_in++; + br->avail_in--; + br->cache_avail += 8; + n -= 8; + rar->bytes_unconsumed++; + rar->bytes_remaining--; + } +} + +static int +rar_br_preparation(struct archive_read *a, struct rar_br *br) +{ + struct rar *rar = (struct rar *)(a->format->data); + + if (rar->bytes_remaining > 0) { + br->next_in = rar_read_ahead(a, 1, &(br->avail_in)); + if (br->next_in == NULL) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated RAR file data"); + return (ARCHIVE_FATAL); + } + if (br->cache_avail == 0) + (void)rar_br_fillup(a, br); + } + return (ARCHIVE_OK); +} + +/* Find last bit set */ +static inline int +rar_fls(unsigned int word) +{ + word |= (word >> 1); + word |= (word >> 2); + word |= (word >> 4); + word |= (word >> 8); + word |= (word >> 16); + return word - (word >> 1); +} + +/* LZSS functions */ +static inline int64_t +lzss_position(struct lzss *lzss) +{ + return lzss->position; +} + +static inline int +lzss_mask(struct lzss *lzss) +{ + return lzss->mask; +} + +static inline int +lzss_size(struct lzss *lzss) +{ + return lzss->mask + 1; +} + +static inline int +lzss_offset_for_position(struct lzss *lzss, int64_t pos) +{ + return (int)(pos & lzss->mask); +} + +static inline unsigned char * +lzss_pointer_for_position(struct lzss *lzss, int64_t pos) +{ + return &lzss->window[lzss_offset_for_position(lzss, pos)]; +} + +static inline int +lzss_current_offset(struct lzss *lzss) +{ + return lzss_offset_for_position(lzss, lzss->position); +} + +static inline uint8_t * +lzss_current_pointer(struct lzss *lzss) +{ + return lzss_pointer_for_position(lzss, lzss->position); +} + +static inline void +lzss_emit_literal(struct rar *rar, uint8_t literal) +{ + *lzss_current_pointer(&rar->lzss) = literal; + rar->lzss.position++; +} + +static inline void +lzss_emit_match(struct rar *rar, int offset, int length) +{ + int dstoffs = lzss_current_offset(&rar->lzss); + int srcoffs = (dstoffs - offset) & lzss_mask(&rar->lzss); + int l, li, remaining; + unsigned char *d, *s; + + remaining = length; + while (remaining > 0) { + l = remaining; + if (dstoffs > srcoffs) { + if (l > lzss_size(&rar->lzss) - dstoffs) + l = lzss_size(&rar->lzss) - dstoffs; + } else { + if (l > lzss_size(&rar->lzss) - srcoffs) + l = lzss_size(&rar->lzss) - srcoffs; + } + d = &(rar->lzss.window[dstoffs]); + s = &(rar->lzss.window[srcoffs]); + if ((dstoffs + l < srcoffs) || (srcoffs + l < dstoffs)) + memcpy(d, s, l); + else { + for (li = 0; li < l; li++) + d[li] = s[li]; + } + remaining -= l; + dstoffs = (dstoffs + l) & lzss_mask(&(rar->lzss)); + srcoffs = (srcoffs + l) & lzss_mask(&(rar->lzss)); + } + rar->lzss.position += length; +} + +static Byte +ppmd_read(void *p) +{ + struct archive_read *a = ((IByteIn*)p)->a; + struct rar *rar = (struct rar *)(a->format->data); + struct rar_br *br = &(rar->br); + Byte b; + if (!rar_br_read_ahead(a, br, 8)) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated RAR file data"); + rar->valid = 0; + return 0; + } + b = rar_br_bits(br, 8); + rar_br_consume(br, 8); + return b; +} + +int +archive_read_support_format_rar(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct rar *rar; + int r; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, + "archive_read_support_format_rar"); + + rar = (struct rar *)calloc(sizeof(*rar), 1); + if (rar == NULL) + { + archive_set_error(&a->archive, ENOMEM, "Can't allocate rar data"); + return (ARCHIVE_FATAL); + } + + /* + * Until enough data has been read, we cannot tell about + * any encrypted entries yet. + */ + rar->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; + + r = __archive_read_register_format(a, + rar, + "rar", + archive_read_format_rar_bid, + archive_read_format_rar_options, + archive_read_format_rar_read_header, + archive_read_format_rar_read_data, + archive_read_format_rar_read_data_skip, + archive_read_format_rar_seek_data, + archive_read_format_rar_cleanup, + archive_read_support_format_rar_capabilities, + archive_read_format_rar_has_encrypted_entries); + + if (r != ARCHIVE_OK) + free(rar); + return (r); +} + +static int +archive_read_support_format_rar_capabilities(struct archive_read * a) +{ + (void)a; /* UNUSED */ + return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA + | ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); +} + +static int +archive_read_format_rar_has_encrypted_entries(struct archive_read *_a) +{ + if (_a && _a->format) { + struct rar * rar = (struct rar *)_a->format->data; + if (rar) { + return rar->has_encrypted_entries; + } + } + return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; +} + + +static int +archive_read_format_rar_bid(struct archive_read *a, int best_bid) +{ + const char *p; + + /* If there's already a bid > 30, we'll never win. */ + if (best_bid > 30) + return (-1); + + if ((p = __archive_read_ahead(a, 7, NULL)) == NULL) + return (-1); + + if (memcmp(p, RAR_SIGNATURE, 7) == 0) + return (30); + + if ((p[0] == 'M' && p[1] == 'Z') || memcmp(p, "\x7F\x45LF", 4) == 0) { + /* This is a PE file */ + ssize_t offset = 0x10000; + ssize_t window = 4096; + ssize_t bytes_avail; + while (offset + window <= (1024 * 128)) { + const char *buff = __archive_read_ahead(a, offset + window, &bytes_avail); + if (buff == NULL) { + /* Remaining bytes are less than window. */ + window >>= 1; + if (window < 0x40) + return (0); + continue; + } + p = buff + offset; + while (p + 7 < buff + bytes_avail) { + if (memcmp(p, RAR_SIGNATURE, 7) == 0) + return (30); + p += 0x10; + } + offset = p - buff; + } + } + return (0); +} + +static int +skip_sfx(struct archive_read *a) +{ + const void *h; + const char *p, *q; + size_t skip, total; + ssize_t bytes, window; + + total = 0; + window = 4096; + while (total + window <= (1024 * 128)) { + h = __archive_read_ahead(a, window, &bytes); + if (h == NULL) { + /* Remaining bytes are less than window. */ + window >>= 1; + if (window < 0x40) + goto fatal; + continue; + } + if (bytes < 0x40) + goto fatal; + p = h; + q = p + bytes; + + /* + * Scan ahead until we find something that looks + * like the RAR header. + */ + while (p + 7 < q) { + if (memcmp(p, RAR_SIGNATURE, 7) == 0) { + skip = p - (const char *)h; + __archive_read_consume(a, skip); + return (ARCHIVE_OK); + } + p += 0x10; + } + skip = p - (const char *)h; + __archive_read_consume(a, skip); + total += skip; + } +fatal: + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Couldn't find out RAR header"); + return (ARCHIVE_FATAL); +} + +static int +archive_read_format_rar_options(struct archive_read *a, + const char *key, const char *val) +{ + struct rar *rar; + int ret = ARCHIVE_FAILED; + + rar = (struct rar *)(a->format->data); + if (strcmp(key, "hdrcharset") == 0) { + if (val == NULL || val[0] == 0) + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "rar: hdrcharset option needs a character-set name"); + else { + rar->opt_sconv = + archive_string_conversion_from_charset( + &a->archive, val, 0); + if (rar->opt_sconv != NULL) + ret = ARCHIVE_OK; + else + ret = ARCHIVE_FATAL; + } + return (ret); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +static int +archive_read_format_rar_read_header(struct archive_read *a, + struct archive_entry *entry) +{ + const void *h; + const char *p; + struct rar *rar; + size_t skip; + char head_type; + int ret; + unsigned flags; + unsigned long crc32_expected; + + a->archive.archive_format = ARCHIVE_FORMAT_RAR; + if (a->archive.archive_format_name == NULL) + a->archive.archive_format_name = "RAR"; + + rar = (struct rar *)(a->format->data); + + /* + * It should be sufficient to call archive_read_next_header() for + * a reader to determine if an entry is encrypted or not. If the + * encryption of an entry is only detectable when calling + * archive_read_data(), so be it. We'll do the same check there + * as well. + */ + if (rar->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { + rar->has_encrypted_entries = 0; + } + + /* RAR files can be generated without EOF headers, so return ARCHIVE_EOF if + * this fails. + */ + if ((h = __archive_read_ahead(a, 7, NULL)) == NULL) + return (ARCHIVE_EOF); + + p = h; + if (rar->found_first_header == 0 && + ((p[0] == 'M' && p[1] == 'Z') || memcmp(p, "\x7F\x45LF", 4) == 0)) { + /* This is an executable ? Must be self-extracting... */ + ret = skip_sfx(a); + if (ret < ARCHIVE_WARN) + return (ret); + } + rar->found_first_header = 1; + + while (1) + { + unsigned long crc32_val; + + if ((h = __archive_read_ahead(a, 7, NULL)) == NULL) + return (ARCHIVE_FATAL); + p = h; + + head_type = p[2]; + switch(head_type) + { + case MARK_HEAD: + if (memcmp(p, RAR_SIGNATURE, 7) != 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid marker header"); + return (ARCHIVE_FATAL); + } + __archive_read_consume(a, 7); + break; + + case MAIN_HEAD: + rar->main_flags = archive_le16dec(p + 3); + skip = archive_le16dec(p + 5); + if (skip < 7 + sizeof(rar->reserved1) + sizeof(rar->reserved2)) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid header size"); + return (ARCHIVE_FATAL); + } + if ((h = __archive_read_ahead(a, skip, NULL)) == NULL) + return (ARCHIVE_FATAL); + p = h; + memcpy(rar->reserved1, p + 7, sizeof(rar->reserved1)); + memcpy(rar->reserved2, p + 7 + sizeof(rar->reserved1), + sizeof(rar->reserved2)); + if (rar->main_flags & MHD_ENCRYPTVER) { + if (skip < 7 + sizeof(rar->reserved1) + sizeof(rar->reserved2)+1) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid header size"); + return (ARCHIVE_FATAL); + } + rar->encryptver = *(p + 7 + sizeof(rar->reserved1) + + sizeof(rar->reserved2)); + } + + /* Main header is password encrypted, so we cannot read any + file names or any other info about files from the header. */ + if (rar->main_flags & MHD_PASSWORD) + { + archive_entry_set_is_metadata_encrypted(entry, 1); + archive_entry_set_is_data_encrypted(entry, 1); + rar->has_encrypted_entries = 1; + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "RAR encryption support unavailable."); + return (ARCHIVE_FATAL); + } + + crc32_val = crc32(0, (const unsigned char *)p + 2, (unsigned)skip - 2); + if ((crc32_val & 0xffff) != archive_le16dec(p)) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Header CRC error"); + return (ARCHIVE_FATAL); + } + __archive_read_consume(a, skip); + break; + + case FILE_HEAD: + return read_header(a, entry, head_type); + + case COMM_HEAD: + case AV_HEAD: + case SUB_HEAD: + case PROTECT_HEAD: + case SIGN_HEAD: + case ENDARC_HEAD: + flags = archive_le16dec(p + 3); + skip = archive_le16dec(p + 5); + if (skip < 7) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid header size too small"); + return (ARCHIVE_FATAL); + } + if (flags & HD_ADD_SIZE_PRESENT) + { + if (skip < 7 + 4) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid header size too small"); + return (ARCHIVE_FATAL); + } + if ((h = __archive_read_ahead(a, skip, NULL)) == NULL) + return (ARCHIVE_FATAL); + p = h; + skip += archive_le32dec(p + 7); + } + + /* Skip over the 2-byte CRC at the beginning of the header. */ + crc32_expected = archive_le16dec(p); + __archive_read_consume(a, 2); + skip -= 2; + + /* Skim the entire header and compute the CRC. */ + crc32_val = 0; + while (skip > 0) { + size_t to_read = skip; + ssize_t did_read; + if (to_read > 32 * 1024) { + to_read = 32 * 1024; + } + if ((h = __archive_read_ahead(a, to_read, &did_read)) == NULL) { + return (ARCHIVE_FATAL); + } + p = h; + crc32_val = crc32(crc32_val, (const unsigned char *)p, (unsigned)did_read); + __archive_read_consume(a, did_read); + skip -= did_read; + } + if ((crc32_val & 0xffff) != crc32_expected) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Header CRC error"); + return (ARCHIVE_FATAL); + } + if (head_type == ENDARC_HEAD) + return (ARCHIVE_EOF); + break; + + case NEWSUB_HEAD: + if ((ret = read_header(a, entry, head_type)) < ARCHIVE_WARN) + return ret; + break; + + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Bad RAR file"); + return (ARCHIVE_FATAL); + } + } +} + +static int +archive_read_format_rar_read_data(struct archive_read *a, const void **buff, + size_t *size, int64_t *offset) +{ + struct rar *rar = (struct rar *)(a->format->data); + int ret; + + if (rar->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { + rar->has_encrypted_entries = 0; + } + + if (rar->bytes_unconsumed > 0) { + /* Consume as much as the decompressor actually used. */ + __archive_read_consume(a, rar->bytes_unconsumed); + rar->bytes_unconsumed = 0; + } + + *buff = NULL; + if (rar->entry_eof || rar->offset_seek >= rar->unp_size) { + *size = 0; + *offset = rar->offset; + if (*offset < rar->unp_size) + *offset = rar->unp_size; + return (ARCHIVE_EOF); + } + + switch (rar->compression_method) + { + case COMPRESS_METHOD_STORE: + ret = read_data_stored(a, buff, size, offset); + break; + + case COMPRESS_METHOD_FASTEST: + case COMPRESS_METHOD_FAST: + case COMPRESS_METHOD_NORMAL: + case COMPRESS_METHOD_GOOD: + case COMPRESS_METHOD_BEST: + ret = read_data_compressed(a, buff, size, offset); + if (ret != ARCHIVE_OK && ret != ARCHIVE_WARN) { + __archive_ppmd7_functions.Ppmd7_Free(&rar->ppmd7_context); + rar->start_new_table = 1; + rar->ppmd_valid = 0; + } + break; + + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Unsupported compression method for RAR file."); + ret = ARCHIVE_FATAL; + break; + } + return (ret); +} + +static int +archive_read_format_rar_read_data_skip(struct archive_read *a) +{ + struct rar *rar; + int64_t bytes_skipped; + int ret; + + rar = (struct rar *)(a->format->data); + + if (rar->bytes_unconsumed > 0) { + /* Consume as much as the decompressor actually used. */ + __archive_read_consume(a, rar->bytes_unconsumed); + rar->bytes_unconsumed = 0; + } + + if (rar->bytes_remaining > 0) { + bytes_skipped = __archive_read_consume(a, rar->bytes_remaining); + if (bytes_skipped < 0) + return (ARCHIVE_FATAL); + } + + /* Compressed data to skip must be read from each header in a multivolume + * archive. + */ + if (rar->main_flags & MHD_VOLUME && rar->file_flags & FHD_SPLIT_AFTER) + { + ret = archive_read_format_rar_read_header(a, a->entry); + if (ret == (ARCHIVE_EOF)) + ret = archive_read_format_rar_read_header(a, a->entry); + if (ret != (ARCHIVE_OK)) + return ret; + return archive_read_format_rar_read_data_skip(a); + } + + return (ARCHIVE_OK); +} + +static int64_t +archive_read_format_rar_seek_data(struct archive_read *a, int64_t offset, + int whence) +{ + int64_t client_offset, ret; + unsigned int i; + struct rar *rar = (struct rar *)(a->format->data); + + if (rar->compression_method == COMPRESS_METHOD_STORE) + { + /* Modify the offset for use with SEEK_SET */ + switch (whence) + { + case SEEK_CUR: + client_offset = rar->offset_seek; + break; + case SEEK_END: + client_offset = rar->unp_size; + break; + case SEEK_SET: + default: + client_offset = 0; + } + client_offset += offset; + if (client_offset < 0) + { + /* Can't seek past beginning of data block */ + return -1; + } + else if (client_offset > rar->unp_size) + { + /* + * Set the returned offset but only seek to the end of + * the data block. + */ + rar->offset_seek = client_offset; + client_offset = rar->unp_size; + } + + client_offset += rar->dbo[0].start_offset; + i = 0; + while (i < rar->cursor) + { + i++; + client_offset += rar->dbo[i].start_offset - rar->dbo[i-1].end_offset; + } + if (rar->main_flags & MHD_VOLUME) + { + /* Find the appropriate offset among the multivolume archive */ + while (1) + { + if (client_offset < rar->dbo[rar->cursor].start_offset && + rar->file_flags & FHD_SPLIT_BEFORE) + { + /* Search backwards for the correct data block */ + if (rar->cursor == 0) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Attempt to seek past beginning of RAR data block"); + return (ARCHIVE_FAILED); + } + rar->cursor--; + client_offset -= rar->dbo[rar->cursor+1].start_offset - + rar->dbo[rar->cursor].end_offset; + if (client_offset < rar->dbo[rar->cursor].start_offset) + continue; + ret = __archive_read_seek(a, rar->dbo[rar->cursor].start_offset - + rar->dbo[rar->cursor].header_size, SEEK_SET); + if (ret < (ARCHIVE_OK)) + return ret; + ret = archive_read_format_rar_read_header(a, a->entry); + if (ret != (ARCHIVE_OK)) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Error during seek of RAR file"); + return (ARCHIVE_FAILED); + } + rar->cursor--; + break; + } + else if (client_offset > rar->dbo[rar->cursor].end_offset && + rar->file_flags & FHD_SPLIT_AFTER) + { + /* Search forward for the correct data block */ + rar->cursor++; + if (rar->cursor < rar->nodes && + client_offset > rar->dbo[rar->cursor].end_offset) + { + client_offset += rar->dbo[rar->cursor].start_offset - + rar->dbo[rar->cursor-1].end_offset; + continue; + } + rar->cursor--; + ret = __archive_read_seek(a, rar->dbo[rar->cursor].end_offset, + SEEK_SET); + if (ret < (ARCHIVE_OK)) + return ret; + ret = archive_read_format_rar_read_header(a, a->entry); + if (ret == (ARCHIVE_EOF)) + { + rar->has_endarc_header = 1; + ret = archive_read_format_rar_read_header(a, a->entry); + } + if (ret != (ARCHIVE_OK)) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Error during seek of RAR file"); + return (ARCHIVE_FAILED); + } + client_offset += rar->dbo[rar->cursor].start_offset - + rar->dbo[rar->cursor-1].end_offset; + continue; + } + break; + } + } + + ret = __archive_read_seek(a, client_offset, SEEK_SET); + if (ret < (ARCHIVE_OK)) + return ret; + rar->bytes_remaining = rar->dbo[rar->cursor].end_offset - ret; + i = rar->cursor; + while (i > 0) + { + i--; + ret -= rar->dbo[i+1].start_offset - rar->dbo[i].end_offset; + } + ret -= rar->dbo[0].start_offset; + + /* Always restart reading the file after a seek */ + __archive_reset_read_data(&a->archive); + + rar->bytes_unconsumed = 0; + rar->offset = 0; + + /* + * If a seek past the end of file was requested, return the requested + * offset. + */ + if (ret == rar->unp_size && rar->offset_seek > rar->unp_size) + return rar->offset_seek; + + /* Return the new offset */ + rar->offset_seek = ret; + return rar->offset_seek; + } + else + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Seeking of compressed RAR files is unsupported"); + } + return (ARCHIVE_FAILED); +} + +static int +archive_read_format_rar_cleanup(struct archive_read *a) +{ + struct rar *rar; + + rar = (struct rar *)(a->format->data); + free_codes(a); + free(rar->filename); + free(rar->filename_save); + free(rar->dbo); + free(rar->unp_buffer); + free(rar->lzss.window); + __archive_ppmd7_functions.Ppmd7_Free(&rar->ppmd7_context); + free(rar); + (a->format->data) = NULL; + return (ARCHIVE_OK); +} + +static int +read_header(struct archive_read *a, struct archive_entry *entry, + char head_type) +{ + const void *h; + const char *p, *endp; + struct rar *rar; + struct rar_header rar_header; + struct rar_file_header file_header; + int64_t header_size; + unsigned filename_size, end; + char *filename; + char *strp; + char packed_size[8]; + char unp_size[8]; + int ttime; + struct archive_string_conv *sconv, *fn_sconv; + unsigned long crc32_val; + int ret = (ARCHIVE_OK), ret2; + + rar = (struct rar *)(a->format->data); + + /* Setup a string conversion object for non-rar-unicode filenames. */ + sconv = rar->opt_sconv; + if (sconv == NULL) { + if (!rar->init_default_conversion) { + rar->sconv_default = + archive_string_default_conversion_for_read( + &(a->archive)); + rar->init_default_conversion = 1; + } + sconv = rar->sconv_default; + } + + + if ((h = __archive_read_ahead(a, 7, NULL)) == NULL) + return (ARCHIVE_FATAL); + p = h; + memcpy(&rar_header, p, sizeof(rar_header)); + rar->file_flags = archive_le16dec(rar_header.flags); + header_size = archive_le16dec(rar_header.size); + if (header_size < (int64_t)sizeof(file_header) + 7) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid header size"); + return (ARCHIVE_FATAL); + } + crc32_val = crc32(0, (const unsigned char *)p + 2, 7 - 2); + __archive_read_consume(a, 7); + + if (!(rar->file_flags & FHD_SOLID)) + { + rar->compression_method = 0; + rar->packed_size = 0; + rar->unp_size = 0; + rar->mtime = 0; + rar->ctime = 0; + rar->atime = 0; + rar->arctime = 0; + rar->mode = 0; + memset(&rar->salt, 0, sizeof(rar->salt)); + rar->atime = 0; + rar->ansec = 0; + rar->ctime = 0; + rar->cnsec = 0; + rar->mtime = 0; + rar->mnsec = 0; + rar->arctime = 0; + rar->arcnsec = 0; + } + else + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "RAR solid archive support unavailable."); + return (ARCHIVE_FATAL); + } + + if ((h = __archive_read_ahead(a, (size_t)header_size - 7, NULL)) == NULL) + return (ARCHIVE_FATAL); + + /* File Header CRC check. */ + crc32_val = crc32(crc32_val, h, (unsigned)(header_size - 7)); + if ((crc32_val & 0xffff) != archive_le16dec(rar_header.crc)) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Header CRC error"); + return (ARCHIVE_FATAL); + } + /* If no CRC error, Go on parsing File Header. */ + p = h; + endp = p + header_size - 7; + memcpy(&file_header, p, sizeof(file_header)); + p += sizeof(file_header); + + rar->compression_method = file_header.method; + + ttime = archive_le32dec(file_header.file_time); + rar->mtime = get_time(ttime); + + rar->file_crc = archive_le32dec(file_header.file_crc); + + if (rar->file_flags & FHD_PASSWORD) + { + archive_entry_set_is_data_encrypted(entry, 1); + rar->has_encrypted_entries = 1; + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "RAR encryption support unavailable."); + /* Since it is only the data part itself that is encrypted we can at least + extract information about the currently processed entry and don't need + to return ARCHIVE_FATAL here. */ + /*return (ARCHIVE_FATAL);*/ + } + + if (rar->file_flags & FHD_LARGE) + { + memcpy(packed_size, file_header.pack_size, 4); + memcpy(packed_size + 4, p, 4); /* High pack size */ + p += 4; + memcpy(unp_size, file_header.unp_size, 4); + memcpy(unp_size + 4, p, 4); /* High unpack size */ + p += 4; + rar->packed_size = archive_le64dec(&packed_size); + rar->unp_size = archive_le64dec(&unp_size); + } + else + { + rar->packed_size = archive_le32dec(file_header.pack_size); + rar->unp_size = archive_le32dec(file_header.unp_size); + } + + if (rar->packed_size < 0 || rar->unp_size < 0) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid sizes specified."); + return (ARCHIVE_FATAL); + } + + rar->bytes_remaining = rar->packed_size; + + /* TODO: RARv3 subblocks contain comments. For now the complete block is + * consumed at the end. + */ + if (head_type == NEWSUB_HEAD) { + size_t distance = p - (const char *)h; + header_size += rar->packed_size; + /* Make sure we have the extended data. */ + if ((h = __archive_read_ahead(a, (size_t)header_size - 7, NULL)) == NULL) + return (ARCHIVE_FATAL); + p = h; + endp = p + header_size - 7; + p += distance; + } + + filename_size = archive_le16dec(file_header.name_size); + if (p + filename_size > endp) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid filename size"); + return (ARCHIVE_FATAL); + } + if (rar->filename_allocated < filename_size * 2 + 2) { + char *newptr; + size_t newsize = filename_size * 2 + 2; + newptr = realloc(rar->filename, newsize); + if (newptr == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Couldn't allocate memory."); + return (ARCHIVE_FATAL); + } + rar->filename = newptr; + rar->filename_allocated = newsize; + } + filename = rar->filename; + memcpy(filename, p, filename_size); + filename[filename_size] = '\0'; + if (rar->file_flags & FHD_UNICODE) + { + if (filename_size != strlen(filename)) + { + unsigned char highbyte, flagbits, flagbyte; + unsigned fn_end, offset; + + end = filename_size; + fn_end = filename_size * 2; + filename_size = 0; + offset = (unsigned)strlen(filename) + 1; + highbyte = *(p + offset++); + flagbits = 0; + flagbyte = 0; + while (offset < end && filename_size < fn_end) + { + if (!flagbits) + { + flagbyte = *(p + offset++); + flagbits = 8; + } + + flagbits -= 2; + switch((flagbyte >> flagbits) & 3) + { + case 0: + filename[filename_size++] = '\0'; + filename[filename_size++] = *(p + offset++); + break; + case 1: + filename[filename_size++] = highbyte; + filename[filename_size++] = *(p + offset++); + break; + case 2: + filename[filename_size++] = *(p + offset + 1); + filename[filename_size++] = *(p + offset); + offset += 2; + break; + case 3: + { + char extra, high; + uint8_t length = *(p + offset++); + + if (length & 0x80) { + extra = *(p + offset++); + high = (char)highbyte; + } else + extra = high = 0; + length = (length & 0x7f) + 2; + while (length && filename_size < fn_end) { + unsigned cp = filename_size >> 1; + filename[filename_size++] = high; + filename[filename_size++] = p[cp] + extra; + length--; + } + } + break; + } + } + if (filename_size > fn_end) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid filename"); + return (ARCHIVE_FATAL); + } + filename[filename_size++] = '\0'; + /* + * Do not increment filename_size here as the computations below + * add the space for the terminating NUL explicitly. + */ + filename[filename_size] = '\0'; + + /* Decoded unicode form is UTF-16BE, so we have to update a string + * conversion object for it. */ + if (rar->sconv_utf16be == NULL) { + rar->sconv_utf16be = archive_string_conversion_from_charset( + &a->archive, "UTF-16BE", 1); + if (rar->sconv_utf16be == NULL) + return (ARCHIVE_FATAL); + } + fn_sconv = rar->sconv_utf16be; + + strp = filename; + while (memcmp(strp, "\x00\x00", 2)) + { + if (!memcmp(strp, "\x00\\", 2)) + *(strp + 1) = '/'; + strp += 2; + } + p += offset; + } else { + /* + * If FHD_UNICODE is set but no unicode data, this file name form + * is UTF-8, so we have to update a string conversion object for + * it accordingly. + */ + if (rar->sconv_utf8 == NULL) { + rar->sconv_utf8 = archive_string_conversion_from_charset( + &a->archive, "UTF-8", 1); + if (rar->sconv_utf8 == NULL) + return (ARCHIVE_FATAL); + } + fn_sconv = rar->sconv_utf8; + while ((strp = strchr(filename, '\\')) != NULL) + *strp = '/'; + p += filename_size; + } + } + else + { + fn_sconv = sconv; + while ((strp = strchr(filename, '\\')) != NULL) + *strp = '/'; + p += filename_size; + } + + /* Split file in multivolume RAR. No more need to process header. */ + if (rar->filename_save && + filename_size == rar->filename_save_size && + !memcmp(rar->filename, rar->filename_save, filename_size + 1)) + { + __archive_read_consume(a, header_size - 7); + rar->cursor++; + if (rar->cursor >= rar->nodes) + { + rar->nodes++; + if ((rar->dbo = + realloc(rar->dbo, sizeof(*rar->dbo) * rar->nodes)) == NULL) + { + archive_set_error(&a->archive, ENOMEM, "Couldn't allocate memory."); + return (ARCHIVE_FATAL); + } + rar->dbo[rar->cursor].header_size = header_size; + rar->dbo[rar->cursor].start_offset = -1; + rar->dbo[rar->cursor].end_offset = -1; + } + if (rar->dbo[rar->cursor].start_offset < 0) + { + rar->dbo[rar->cursor].start_offset = a->filter->position; + rar->dbo[rar->cursor].end_offset = rar->dbo[rar->cursor].start_offset + + rar->packed_size; + } + return ret; + } + else if (rar->filename_must_match) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Mismatch of file parts split across multi-volume archive"); + return (ARCHIVE_FATAL); + } + + rar->filename_save = (char*)realloc(rar->filename_save, + filename_size + 1); + memcpy(rar->filename_save, rar->filename, filename_size + 1); + rar->filename_save_size = filename_size; + + /* Set info for seeking */ + free(rar->dbo); + if ((rar->dbo = calloc(1, sizeof(*rar->dbo))) == NULL) + { + archive_set_error(&a->archive, ENOMEM, "Couldn't allocate memory."); + return (ARCHIVE_FATAL); + } + rar->dbo[0].header_size = header_size; + rar->dbo[0].start_offset = -1; + rar->dbo[0].end_offset = -1; + rar->cursor = 0; + rar->nodes = 1; + + if (rar->file_flags & FHD_SALT) + { + if (p + 8 > endp) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid header size"); + return (ARCHIVE_FATAL); + } + memcpy(rar->salt, p, 8); + p += 8; + } + + if (rar->file_flags & FHD_EXTTIME) { + if (read_exttime(p, rar, endp) < 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid header size"); + return (ARCHIVE_FATAL); + } + } + + __archive_read_consume(a, header_size - 7); + rar->dbo[0].start_offset = a->filter->position; + rar->dbo[0].end_offset = rar->dbo[0].start_offset + rar->packed_size; + + switch(file_header.host_os) + { + case OS_MSDOS: + case OS_OS2: + case OS_WIN32: + rar->mode = archive_le32dec(file_header.file_attr); + if (rar->mode & FILE_ATTRIBUTE_DIRECTORY) + rar->mode = AE_IFDIR | S_IXUSR | S_IXGRP | S_IXOTH; + else + rar->mode = AE_IFREG; + rar->mode |= S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH; + break; + + case OS_UNIX: + case OS_MAC_OS: + case OS_BEOS: + rar->mode = archive_le32dec(file_header.file_attr); + break; + + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Unknown file attributes from RAR file's host OS"); + return (ARCHIVE_FATAL); + } + + rar->bytes_uncopied = rar->bytes_unconsumed = 0; + rar->lzss.position = rar->offset = 0; + rar->offset_seek = 0; + rar->dictionary_size = 0; + rar->offset_outgoing = 0; + rar->br.cache_avail = 0; + rar->br.avail_in = 0; + rar->crc_calculated = 0; + rar->entry_eof = 0; + rar->valid = 1; + rar->is_ppmd_block = 0; + rar->start_new_table = 1; + free(rar->unp_buffer); + rar->unp_buffer = NULL; + rar->unp_offset = 0; + rar->unp_buffer_size = UNP_BUFFER_SIZE; + memset(rar->lengthtable, 0, sizeof(rar->lengthtable)); + __archive_ppmd7_functions.Ppmd7_Free(&rar->ppmd7_context); + rar->ppmd_valid = rar->ppmd_eod = 0; + + /* Don't set any archive entries for non-file header types */ + if (head_type == NEWSUB_HEAD) + return ret; + + archive_entry_set_mtime(entry, rar->mtime, rar->mnsec); + archive_entry_set_ctime(entry, rar->ctime, rar->cnsec); + archive_entry_set_atime(entry, rar->atime, rar->ansec); + archive_entry_set_size(entry, rar->unp_size); + archive_entry_set_mode(entry, rar->mode); + + if (archive_entry_copy_pathname_l(entry, filename, filename_size, fn_sconv)) + { + if (errno == ENOMEM) + { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Pathname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Pathname cannot be converted from %s to current locale.", + archive_string_conversion_charset_name(fn_sconv)); + ret = (ARCHIVE_WARN); + } + + if (((rar->mode) & AE_IFMT) == AE_IFLNK) + { + /* Make sure a symbolic-link file does not have its body. */ + rar->bytes_remaining = 0; + archive_entry_set_size(entry, 0); + + /* Read a symbolic-link name. */ + if ((ret2 = read_symlink_stored(a, entry, sconv)) < (ARCHIVE_WARN)) + return ret2; + if (ret > ret2) + ret = ret2; + } + + if (rar->bytes_remaining == 0) + rar->entry_eof = 1; + + return ret; +} + +static time_t +get_time(int ttime) +{ + struct tm tm; + tm.tm_sec = 2 * (ttime & 0x1f); + tm.tm_min = (ttime >> 5) & 0x3f; + tm.tm_hour = (ttime >> 11) & 0x1f; + tm.tm_mday = (ttime >> 16) & 0x1f; + tm.tm_mon = ((ttime >> 21) & 0x0f) - 1; + tm.tm_year = ((ttime >> 25) & 0x7f) + 80; + tm.tm_isdst = -1; + return mktime(&tm); +} + +static int +read_exttime(const char *p, struct rar *rar, const char *endp) +{ + unsigned rmode, flags, rem, j, count; + int ttime, i; + struct tm *tm; + time_t t; + long nsec; +#if defined(HAVE_LOCALTIME_R) || defined(HAVE__LOCALTIME64_S) + struct tm tmbuf; +#endif +#if defined(HAVE__LOCALTIME64_S) + errno_t terr; + __time64_t tmptime; +#endif + + if (p + 2 > endp) + return (-1); + flags = archive_le16dec(p); + p += 2; + + for (i = 3; i >= 0; i--) + { + t = 0; + if (i == 3) + t = rar->mtime; + rmode = flags >> i * 4; + if (rmode & 8) + { + if (!t) + { + if (p + 4 > endp) + return (-1); + ttime = archive_le32dec(p); + t = get_time(ttime); + p += 4; + } + rem = 0; + count = rmode & 3; + if (p + count > endp) + return (-1); + for (j = 0; j < count; j++) + { + rem = (((unsigned)(unsigned char)*p) << 16) | (rem >> 8); + p++; + } +#if defined(HAVE_LOCALTIME_R) + tm = localtime_r(&t, &tmbuf); +#elif defined(HAVE__LOCALTIME64_S) + tmptime = t; + terr = _localtime64_s(&tmbuf, &tmptime); + if (terr) + tm = NULL; + else + tm = &tmbuf; +#else + tm = localtime(&t); +#endif + nsec = tm->tm_sec + rem / NS_UNIT; + if (rmode & 4) + { + tm->tm_sec++; + t = mktime(tm); + } + if (i == 3) + { + rar->mtime = t; + rar->mnsec = nsec; + } + else if (i == 2) + { + rar->ctime = t; + rar->cnsec = nsec; + } + else if (i == 1) + { + rar->atime = t; + rar->ansec = nsec; + } + else + { + rar->arctime = t; + rar->arcnsec = nsec; + } + } + } + return (0); +} + +static int +read_symlink_stored(struct archive_read *a, struct archive_entry *entry, + struct archive_string_conv *sconv) +{ + const void *h; + const char *p; + struct rar *rar; + int ret = (ARCHIVE_OK); + + rar = (struct rar *)(a->format->data); + if ((h = rar_read_ahead(a, (size_t)rar->packed_size, NULL)) == NULL) + return (ARCHIVE_FATAL); + p = h; + + if (archive_entry_copy_symlink_l(entry, + p, (size_t)rar->packed_size, sconv)) + { + if (errno == ENOMEM) + { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for link"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "link cannot be converted from %s to current locale.", + archive_string_conversion_charset_name(sconv)); + ret = (ARCHIVE_WARN); + } + __archive_read_consume(a, rar->packed_size); + return ret; +} + +static int +read_data_stored(struct archive_read *a, const void **buff, size_t *size, + int64_t *offset) +{ + struct rar *rar; + ssize_t bytes_avail; + + rar = (struct rar *)(a->format->data); + if (rar->bytes_remaining == 0 && + !(rar->main_flags & MHD_VOLUME && rar->file_flags & FHD_SPLIT_AFTER)) + { + *buff = NULL; + *size = 0; + *offset = rar->offset; + if (rar->file_crc != rar->crc_calculated) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "File CRC error"); + return (ARCHIVE_FATAL); + } + rar->entry_eof = 1; + return (ARCHIVE_EOF); + } + + *buff = rar_read_ahead(a, 1, &bytes_avail); + if (bytes_avail <= 0) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated RAR file data"); + return (ARCHIVE_FATAL); + } + + *size = bytes_avail; + *offset = rar->offset; + rar->offset += bytes_avail; + rar->offset_seek += bytes_avail; + rar->bytes_remaining -= bytes_avail; + rar->bytes_unconsumed = bytes_avail; + /* Calculate File CRC. */ + rar->crc_calculated = crc32(rar->crc_calculated, *buff, + (unsigned)bytes_avail); + return (ARCHIVE_OK); +} + +static int +read_data_compressed(struct archive_read *a, const void **buff, size_t *size, + int64_t *offset) +{ + struct rar *rar; + int64_t start, end, actualend; + size_t bs; + int ret = (ARCHIVE_OK), sym, code, lzss_offset, length, i; + + rar = (struct rar *)(a->format->data); + + do { + if (!rar->valid) + return (ARCHIVE_FATAL); + if (rar->ppmd_eod || + (rar->dictionary_size && rar->offset >= rar->unp_size)) + { + if (rar->unp_offset > 0) { + /* + * We have unprocessed extracted data. write it out. + */ + *buff = rar->unp_buffer; + *size = rar->unp_offset; + *offset = rar->offset_outgoing; + rar->offset_outgoing += *size; + /* Calculate File CRC. */ + rar->crc_calculated = crc32(rar->crc_calculated, *buff, + (unsigned)*size); + rar->unp_offset = 0; + return (ARCHIVE_OK); + } + *buff = NULL; + *size = 0; + *offset = rar->offset; + if (rar->file_crc != rar->crc_calculated) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "File CRC error"); + return (ARCHIVE_FATAL); + } + rar->entry_eof = 1; + return (ARCHIVE_EOF); + } + + if (!rar->is_ppmd_block && rar->dictionary_size && rar->bytes_uncopied > 0) + { + if (rar->bytes_uncopied > (rar->unp_buffer_size - rar->unp_offset)) + bs = rar->unp_buffer_size - rar->unp_offset; + else + bs = (size_t)rar->bytes_uncopied; + ret = copy_from_lzss_window(a, buff, rar->offset, (int)bs); + if (ret != ARCHIVE_OK) + return (ret); + rar->offset += bs; + rar->bytes_uncopied -= bs; + if (*buff != NULL) { + rar->unp_offset = 0; + *size = rar->unp_buffer_size; + *offset = rar->offset_outgoing; + rar->offset_outgoing += *size; + /* Calculate File CRC. */ + rar->crc_calculated = crc32(rar->crc_calculated, *buff, + (unsigned)*size); + return (ret); + } + continue; + } + + if (!rar->br.next_in && + (ret = rar_br_preparation(a, &(rar->br))) < ARCHIVE_WARN) + return (ret); + if (rar->start_new_table && ((ret = parse_codes(a)) < (ARCHIVE_WARN))) + return (ret); + + if (rar->is_ppmd_block) + { + if ((sym = __archive_ppmd7_functions.Ppmd7_DecodeSymbol( + &rar->ppmd7_context, &rar->range_dec.p)) < 0) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid symbol"); + return (ARCHIVE_FATAL); + } + if(sym != rar->ppmd_escape) + { + lzss_emit_literal(rar, sym); + rar->bytes_uncopied++; + } + else + { + if ((code = __archive_ppmd7_functions.Ppmd7_DecodeSymbol( + &rar->ppmd7_context, &rar->range_dec.p)) < 0) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid symbol"); + return (ARCHIVE_FATAL); + } + + switch(code) + { + case 0: + rar->start_new_table = 1; + return read_data_compressed(a, buff, size, offset); + + case 2: + rar->ppmd_eod = 1;/* End Of ppmd Data. */ + continue; + + case 3: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Parsing filters is unsupported."); + return (ARCHIVE_FAILED); + + case 4: + lzss_offset = 0; + for (i = 2; i >= 0; i--) + { + if ((code = __archive_ppmd7_functions.Ppmd7_DecodeSymbol( + &rar->ppmd7_context, &rar->range_dec.p)) < 0) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid symbol"); + return (ARCHIVE_FATAL); + } + lzss_offset |= code << (i * 8); + } + if ((length = __archive_ppmd7_functions.Ppmd7_DecodeSymbol( + &rar->ppmd7_context, &rar->range_dec.p)) < 0) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid symbol"); + return (ARCHIVE_FATAL); + } + lzss_emit_match(rar, lzss_offset + 2, length + 32); + rar->bytes_uncopied += length + 32; + break; + + case 5: + if ((length = __archive_ppmd7_functions.Ppmd7_DecodeSymbol( + &rar->ppmd7_context, &rar->range_dec.p)) < 0) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid symbol"); + return (ARCHIVE_FATAL); + } + lzss_emit_match(rar, 1, length + 4); + rar->bytes_uncopied += length + 4; + break; + + default: + lzss_emit_literal(rar, sym); + rar->bytes_uncopied++; + } + } + } + else + { + start = rar->offset; + end = start + rar->dictionary_size; + rar->filterstart = INT64_MAX; + + if ((actualend = expand(a, end)) < 0) + return ((int)actualend); + + rar->bytes_uncopied = actualend - start; + if (rar->bytes_uncopied == 0) { + /* Broken RAR files cause this case. + * NOTE: If this case were possible on a normal RAR file + * we would find out where it was actually bad and + * what we would do to solve it. */ + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Internal error extracting RAR file"); + return (ARCHIVE_FATAL); + } + } + if (rar->bytes_uncopied > (rar->unp_buffer_size - rar->unp_offset)) + bs = rar->unp_buffer_size - rar->unp_offset; + else + bs = (size_t)rar->bytes_uncopied; + ret = copy_from_lzss_window(a, buff, rar->offset, (int)bs); + if (ret != ARCHIVE_OK) + return (ret); + rar->offset += bs; + rar->bytes_uncopied -= bs; + /* + * If *buff is NULL, it means unp_buffer is not full. + * So we have to continue extracting a RAR file. + */ + } while (*buff == NULL); + + rar->unp_offset = 0; + *size = rar->unp_buffer_size; + *offset = rar->offset_outgoing; + rar->offset_outgoing += *size; + /* Calculate File CRC. */ + rar->crc_calculated = crc32(rar->crc_calculated, *buff, (unsigned)*size); + return ret; +} + +static int +parse_codes(struct archive_read *a) +{ + int i, j, val, n, r; + unsigned char bitlengths[MAX_SYMBOLS], zerocount, ppmd_flags; + unsigned int maxorder; + struct huffman_code precode; + struct rar *rar = (struct rar *)(a->format->data); + struct rar_br *br = &(rar->br); + + free_codes(a); + + /* Skip to the next byte */ + rar_br_consume_unalined_bits(br); + + /* PPMd block flag */ + if (!rar_br_read_ahead(a, br, 1)) + goto truncated_data; + if ((rar->is_ppmd_block = rar_br_bits(br, 1)) != 0) + { + rar_br_consume(br, 1); + if (!rar_br_read_ahead(a, br, 7)) + goto truncated_data; + ppmd_flags = rar_br_bits(br, 7); + rar_br_consume(br, 7); + + /* Memory is allocated in MB */ + if (ppmd_flags & 0x20) + { + if (!rar_br_read_ahead(a, br, 8)) + goto truncated_data; + rar->dictionary_size = (rar_br_bits(br, 8) + 1) << 20; + rar_br_consume(br, 8); + } + + if (ppmd_flags & 0x40) + { + if (!rar_br_read_ahead(a, br, 8)) + goto truncated_data; + rar->ppmd_escape = rar->ppmd7_context.InitEsc = rar_br_bits(br, 8); + rar_br_consume(br, 8); + } + else + rar->ppmd_escape = 2; + + if (ppmd_flags & 0x20) + { + maxorder = (ppmd_flags & 0x1F) + 1; + if(maxorder > 16) + maxorder = 16 + (maxorder - 16) * 3; + + if (maxorder == 1) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated RAR file data"); + return (ARCHIVE_FATAL); + } + + /* Make sure ppmd7_contest is freed before Ppmd7_Construct + * because reading a broken file cause this abnormal sequence. */ + __archive_ppmd7_functions.Ppmd7_Free(&rar->ppmd7_context); + + rar->bytein.a = a; + rar->bytein.Read = &ppmd_read; + __archive_ppmd7_functions.PpmdRAR_RangeDec_CreateVTable(&rar->range_dec); + rar->range_dec.Stream = &rar->bytein; + __archive_ppmd7_functions.Ppmd7_Construct(&rar->ppmd7_context); + + if (rar->dictionary_size == 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid zero dictionary size"); + return (ARCHIVE_FATAL); + } + + if (!__archive_ppmd7_functions.Ppmd7_Alloc(&rar->ppmd7_context, + rar->dictionary_size)) + { + archive_set_error(&a->archive, ENOMEM, + "Out of memory"); + return (ARCHIVE_FATAL); + } + if (!__archive_ppmd7_functions.PpmdRAR_RangeDec_Init(&rar->range_dec)) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Unable to initialize PPMd range decoder"); + return (ARCHIVE_FATAL); + } + __archive_ppmd7_functions.Ppmd7_Init(&rar->ppmd7_context, maxorder); + rar->ppmd_valid = 1; + } + else + { + if (!rar->ppmd_valid) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid PPMd sequence"); + return (ARCHIVE_FATAL); + } + if (!__archive_ppmd7_functions.PpmdRAR_RangeDec_Init(&rar->range_dec)) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Unable to initialize PPMd range decoder"); + return (ARCHIVE_FATAL); + } + } + } + else + { + rar_br_consume(br, 1); + + /* Keep existing table flag */ + if (!rar_br_read_ahead(a, br, 1)) + goto truncated_data; + if (!rar_br_bits(br, 1)) + memset(rar->lengthtable, 0, sizeof(rar->lengthtable)); + rar_br_consume(br, 1); + + memset(&bitlengths, 0, sizeof(bitlengths)); + for (i = 0; i < MAX_SYMBOLS;) + { + if (!rar_br_read_ahead(a, br, 4)) + goto truncated_data; + bitlengths[i++] = rar_br_bits(br, 4); + rar_br_consume(br, 4); + if (bitlengths[i-1] == 0xF) + { + if (!rar_br_read_ahead(a, br, 4)) + goto truncated_data; + zerocount = rar_br_bits(br, 4); + rar_br_consume(br, 4); + if (zerocount) + { + i--; + for (j = 0; j < zerocount + 2 && i < MAX_SYMBOLS; j++) + bitlengths[i++] = 0; + } + } + } + + memset(&precode, 0, sizeof(precode)); + r = create_code(a, &precode, bitlengths, MAX_SYMBOLS, MAX_SYMBOL_LENGTH); + if (r != ARCHIVE_OK) { + free(precode.tree); + free(precode.table); + return (r); + } + + for (i = 0; i < HUFFMAN_TABLE_SIZE;) + { + if ((val = read_next_symbol(a, &precode)) < 0) { + free(precode.tree); + free(precode.table); + return (ARCHIVE_FATAL); + } + if (val < 16) + { + rar->lengthtable[i] = (rar->lengthtable[i] + val) & 0xF; + i++; + } + else if (val < 18) + { + if (i == 0) + { + free(precode.tree); + free(precode.table); + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Internal error extracting RAR file."); + return (ARCHIVE_FATAL); + } + + if(val == 16) { + if (!rar_br_read_ahead(a, br, 3)) { + free(precode.tree); + free(precode.table); + goto truncated_data; + } + n = rar_br_bits(br, 3) + 3; + rar_br_consume(br, 3); + } else { + if (!rar_br_read_ahead(a, br, 7)) { + free(precode.tree); + free(precode.table); + goto truncated_data; + } + n = rar_br_bits(br, 7) + 11; + rar_br_consume(br, 7); + } + + for (j = 0; j < n && i < HUFFMAN_TABLE_SIZE; j++) + { + rar->lengthtable[i] = rar->lengthtable[i-1]; + i++; + } + } + else + { + if(val == 18) { + if (!rar_br_read_ahead(a, br, 3)) { + free(precode.tree); + free(precode.table); + goto truncated_data; + } + n = rar_br_bits(br, 3) + 3; + rar_br_consume(br, 3); + } else { + if (!rar_br_read_ahead(a, br, 7)) { + free(precode.tree); + free(precode.table); + goto truncated_data; + } + n = rar_br_bits(br, 7) + 11; + rar_br_consume(br, 7); + } + + for(j = 0; j < n && i < HUFFMAN_TABLE_SIZE; j++) + rar->lengthtable[i++] = 0; + } + } + free(precode.tree); + free(precode.table); + + r = create_code(a, &rar->maincode, &rar->lengthtable[0], MAINCODE_SIZE, + MAX_SYMBOL_LENGTH); + if (r != ARCHIVE_OK) + return (r); + r = create_code(a, &rar->offsetcode, &rar->lengthtable[MAINCODE_SIZE], + OFFSETCODE_SIZE, MAX_SYMBOL_LENGTH); + if (r != ARCHIVE_OK) + return (r); + r = create_code(a, &rar->lowoffsetcode, + &rar->lengthtable[MAINCODE_SIZE + OFFSETCODE_SIZE], + LOWOFFSETCODE_SIZE, MAX_SYMBOL_LENGTH); + if (r != ARCHIVE_OK) + return (r); + r = create_code(a, &rar->lengthcode, + &rar->lengthtable[MAINCODE_SIZE + OFFSETCODE_SIZE + + LOWOFFSETCODE_SIZE], LENGTHCODE_SIZE, MAX_SYMBOL_LENGTH); + if (r != ARCHIVE_OK) + return (r); + } + + if (!rar->dictionary_size || !rar->lzss.window) + { + /* Seems as though dictionary sizes are not used. Even so, minimize + * memory usage as much as possible. + */ + void *new_window; + unsigned int new_size; + + if (rar->unp_size >= DICTIONARY_MAX_SIZE) + new_size = DICTIONARY_MAX_SIZE; + else + new_size = rar_fls((unsigned int)rar->unp_size) << 1; + if (new_size == 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Zero window size is invalid."); + return (ARCHIVE_FATAL); + } + new_window = realloc(rar->lzss.window, new_size); + if (new_window == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Unable to allocate memory for uncompressed data."); + return (ARCHIVE_FATAL); + } + rar->lzss.window = (unsigned char *)new_window; + rar->dictionary_size = new_size; + memset(rar->lzss.window, 0, rar->dictionary_size); + rar->lzss.mask = rar->dictionary_size - 1; + } + + rar->start_new_table = 0; + return (ARCHIVE_OK); +truncated_data: + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated RAR file data"); + rar->valid = 0; + return (ARCHIVE_FATAL); +} + +static void +free_codes(struct archive_read *a) +{ + struct rar *rar = (struct rar *)(a->format->data); + free(rar->maincode.tree); + free(rar->offsetcode.tree); + free(rar->lowoffsetcode.tree); + free(rar->lengthcode.tree); + free(rar->maincode.table); + free(rar->offsetcode.table); + free(rar->lowoffsetcode.table); + free(rar->lengthcode.table); + memset(&rar->maincode, 0, sizeof(rar->maincode)); + memset(&rar->offsetcode, 0, sizeof(rar->offsetcode)); + memset(&rar->lowoffsetcode, 0, sizeof(rar->lowoffsetcode)); + memset(&rar->lengthcode, 0, sizeof(rar->lengthcode)); +} + + +static int +read_next_symbol(struct archive_read *a, struct huffman_code *code) +{ + unsigned char bit; + unsigned int bits; + int length, value, node; + struct rar *rar; + struct rar_br *br; + + if (!code->table) + { + if (make_table(a, code) != (ARCHIVE_OK)) + return -1; + } + + rar = (struct rar *)(a->format->data); + br = &(rar->br); + + /* Look ahead (peek) at bits */ + if (!rar_br_read_ahead(a, br, code->tablesize)) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated RAR file data"); + rar->valid = 0; + return -1; + } + bits = rar_br_bits(br, code->tablesize); + + length = code->table[bits].length; + value = code->table[bits].value; + + if (length < 0) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid prefix code in bitstream"); + return -1; + } + + if (length <= code->tablesize) + { + /* Skip length bits */ + rar_br_consume(br, length); + return value; + } + + /* Skip tablesize bits */ + rar_br_consume(br, code->tablesize); + + node = value; + while (!(code->tree[node].branches[0] == + code->tree[node].branches[1])) + { + if (!rar_br_read_ahead(a, br, 1)) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated RAR file data"); + rar->valid = 0; + return -1; + } + bit = rar_br_bits(br, 1); + rar_br_consume(br, 1); + + if (code->tree[node].branches[bit] < 0) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid prefix code in bitstream"); + return -1; + } + node = code->tree[node].branches[bit]; + } + + return code->tree[node].branches[0]; +} + +static int +create_code(struct archive_read *a, struct huffman_code *code, + unsigned char *lengths, int numsymbols, char maxlength) +{ + int i, j, codebits = 0, symbolsleft = numsymbols; + + code->numentries = 0; + code->numallocatedentries = 0; + if (new_node(code) < 0) { + archive_set_error(&a->archive, ENOMEM, + "Unable to allocate memory for node data."); + return (ARCHIVE_FATAL); + } + code->numentries = 1; + code->minlength = INT_MAX; + code->maxlength = INT_MIN; + codebits = 0; + for(i = 1; i <= maxlength; i++) + { + for(j = 0; j < numsymbols; j++) + { + if (lengths[j] != i) continue; + if (add_value(a, code, j, codebits, i) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + codebits++; + if (--symbolsleft <= 0) + break; + } + if (symbolsleft <= 0) + break; + codebits <<= 1; + } + return (ARCHIVE_OK); +} + +static int +add_value(struct archive_read *a, struct huffman_code *code, int value, + int codebits, int length) +{ + int lastnode, bitpos, bit; + /* int repeatpos, repeatnode, nextnode; */ + + free(code->table); + code->table = NULL; + + if(length > code->maxlength) + code->maxlength = length; + if(length < code->minlength) + code->minlength = length; + + /* + * Dead code, repeatpos was is -1 + * + repeatpos = -1; + if (repeatpos == 0 || (repeatpos >= 0 + && (((codebits >> (repeatpos - 1)) & 3) == 0 + || ((codebits >> (repeatpos - 1)) & 3) == 3))) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid repeat position"); + return (ARCHIVE_FATAL); + } + */ + + lastnode = 0; + for (bitpos = length - 1; bitpos >= 0; bitpos--) + { + bit = (codebits >> bitpos) & 1; + + /* Leaf node check */ + if (code->tree[lastnode].branches[0] == + code->tree[lastnode].branches[1]) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Prefix found"); + return (ARCHIVE_FATAL); + } + + /* + * Dead code, repeatpos was -1, bitpos >=0 + * + if (bitpos == repeatpos) + { + * Open branch check * + if (!(code->tree[lastnode].branches[bit] < 0)) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid repeating code"); + return (ARCHIVE_FATAL); + } + + if ((repeatnode = new_node(code)) < 0) { + archive_set_error(&a->archive, ENOMEM, + "Unable to allocate memory for node data."); + return (ARCHIVE_FATAL); + } + if ((nextnode = new_node(code)) < 0) { + archive_set_error(&a->archive, ENOMEM, + "Unable to allocate memory for node data."); + return (ARCHIVE_FATAL); + } + + * Set branches * + code->tree[lastnode].branches[bit] = repeatnode; + code->tree[repeatnode].branches[bit] = repeatnode; + code->tree[repeatnode].branches[bit^1] = nextnode; + lastnode = nextnode; + + bitpos++; * terminating bit already handled, skip it * + } + else + { + */ + /* Open branch check */ + if (code->tree[lastnode].branches[bit] < 0) + { + if (new_node(code) < 0) { + archive_set_error(&a->archive, ENOMEM, + "Unable to allocate memory for node data."); + return (ARCHIVE_FATAL); + } + code->tree[lastnode].branches[bit] = code->numentries++; + } + + /* set to branch */ + lastnode = code->tree[lastnode].branches[bit]; + /* } */ + } + + if (!(code->tree[lastnode].branches[0] == -1 + && code->tree[lastnode].branches[1] == -2)) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Prefix found"); + return (ARCHIVE_FATAL); + } + + /* Set leaf value */ + code->tree[lastnode].branches[0] = value; + code->tree[lastnode].branches[1] = value; + + return (ARCHIVE_OK); +} + +static int +new_node(struct huffman_code *code) +{ + void *new_tree; + if (code->numallocatedentries == code->numentries) { + int new_num_entries = 256; + if (code->numentries > 0) { + new_num_entries = code->numentries * 2; + } + new_tree = realloc(code->tree, new_num_entries * sizeof(*code->tree)); + if (new_tree == NULL) + return (-1); + code->tree = (struct huffman_tree_node *)new_tree; + code->numallocatedentries = new_num_entries; + } + code->tree[code->numentries].branches[0] = -1; + code->tree[code->numentries].branches[1] = -2; + return 1; +} + +static int +make_table(struct archive_read *a, struct huffman_code *code) +{ + if (code->maxlength < code->minlength || code->maxlength > 10) + code->tablesize = 10; + else + code->tablesize = code->maxlength; + + code->table = + (struct huffman_table_entry *)calloc(1, sizeof(*code->table) + * ((size_t)1 << code->tablesize)); + + return make_table_recurse(a, code, 0, code->table, 0, code->tablesize); +} + +static int +make_table_recurse(struct archive_read *a, struct huffman_code *code, int node, + struct huffman_table_entry *table, int depth, + int maxdepth) +{ + int currtablesize, i, ret = (ARCHIVE_OK); + + if (!code->tree) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Huffman tree was not created."); + return (ARCHIVE_FATAL); + } + if (node < 0 || node >= code->numentries) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid location to Huffman tree specified."); + return (ARCHIVE_FATAL); + } + + currtablesize = 1 << (maxdepth - depth); + + if (code->tree[node].branches[0] == + code->tree[node].branches[1]) + { + for(i = 0; i < currtablesize; i++) + { + table[i].length = depth; + table[i].value = code->tree[node].branches[0]; + } + } + /* + * Dead code, node >= 0 + * + else if (node < 0) + { + for(i = 0; i < currtablesize; i++) + table[i].length = -1; + } + */ + else + { + if(depth == maxdepth) + { + table[0].length = maxdepth + 1; + table[0].value = node; + } + else + { + ret |= make_table_recurse(a, code, code->tree[node].branches[0], table, + depth + 1, maxdepth); + ret |= make_table_recurse(a, code, code->tree[node].branches[1], + table + currtablesize / 2, depth + 1, maxdepth); + } + } + return ret; +} + +static int64_t +expand(struct archive_read *a, int64_t end) +{ + static const unsigned char lengthbases[] = + { 0, 1, 2, 3, 4, 5, 6, + 7, 8, 10, 12, 14, 16, 20, + 24, 28, 32, 40, 48, 56, 64, + 80, 96, 112, 128, 160, 192, 224 }; + static const unsigned char lengthbits[] = + { 0, 0, 0, 0, 0, 0, 0, + 0, 1, 1, 1, 1, 2, 2, + 2, 2, 3, 3, 3, 3, 4, + 4, 4, 4, 5, 5, 5, 5 }; + static const int lengthb_min = minimum( + (int)(sizeof(lengthbases)/sizeof(lengthbases[0])), + (int)(sizeof(lengthbits)/sizeof(lengthbits[0])) + ); + static const unsigned int offsetbases[] = + { 0, 1, 2, 3, 4, 6, + 8, 12, 16, 24, 32, 48, + 64, 96, 128, 192, 256, 384, + 512, 768, 1024, 1536, 2048, 3072, + 4096, 6144, 8192, 12288, 16384, 24576, + 32768, 49152, 65536, 98304, 131072, 196608, + 262144, 327680, 393216, 458752, 524288, 589824, + 655360, 720896, 786432, 851968, 917504, 983040, + 1048576, 1310720, 1572864, 1835008, 2097152, 2359296, + 2621440, 2883584, 3145728, 3407872, 3670016, 3932160 }; + static const unsigned char offsetbits[] = + { 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, + 5, 5, 6, 6, 7, 7, 8, 8, 9, 9, 10, 10, + 11, 11, 12, 12, 13, 13, 14, 14, 15, 15, 16, 16, + 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, + 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18 }; + static const int offsetb_min = minimum( + (int)(sizeof(offsetbases)/sizeof(offsetbases[0])), + (int)(sizeof(offsetbits)/sizeof(offsetbits[0])) + ); + static const unsigned char shortbases[] = + { 0, 4, 8, 16, 32, 64, 128, 192 }; + static const unsigned char shortbits[] = + { 2, 2, 3, 4, 5, 6, 6, 6 }; + + int symbol, offs, len, offsindex, lensymbol, i, offssymbol, lowoffsetsymbol; + unsigned char newfile; + struct rar *rar = (struct rar *)(a->format->data); + struct rar_br *br = &(rar->br); + + if (rar->filterstart < end) + end = rar->filterstart; + + while (1) + { + if (rar->output_last_match && + lzss_position(&rar->lzss) + rar->lastlength <= end) + { + lzss_emit_match(rar, rar->lastoffset, rar->lastlength); + rar->output_last_match = 0; + } + + if(rar->is_ppmd_block || rar->output_last_match || + lzss_position(&rar->lzss) >= end) + return lzss_position(&rar->lzss); + + if ((symbol = read_next_symbol(a, &rar->maincode)) < 0) + return (ARCHIVE_FATAL); + rar->output_last_match = 0; + + if (symbol < 256) + { + lzss_emit_literal(rar, symbol); + continue; + } + else if (symbol == 256) + { + if (!rar_br_read_ahead(a, br, 1)) + goto truncated_data; + newfile = !rar_br_bits(br, 1); + rar_br_consume(br, 1); + + if(newfile) + { + rar->start_new_block = 1; + if (!rar_br_read_ahead(a, br, 1)) + goto truncated_data; + rar->start_new_table = rar_br_bits(br, 1); + rar_br_consume(br, 1); + return lzss_position(&rar->lzss); + } + else + { + if (parse_codes(a) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + continue; + } + } + else if(symbol==257) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Parsing filters is unsupported."); + return (ARCHIVE_FAILED); + } + else if(symbol==258) + { + if(rar->lastlength == 0) + continue; + + offs = rar->lastoffset; + len = rar->lastlength; + } + else if (symbol <= 262) + { + offsindex = symbol - 259; + offs = rar->oldoffset[offsindex]; + + if ((lensymbol = read_next_symbol(a, &rar->lengthcode)) < 0) + goto bad_data; + if (lensymbol > lengthb_min) + goto bad_data; + len = lengthbases[lensymbol] + 2; + if (lengthbits[lensymbol] > 0) { + if (!rar_br_read_ahead(a, br, lengthbits[lensymbol])) + goto truncated_data; + len += rar_br_bits(br, lengthbits[lensymbol]); + rar_br_consume(br, lengthbits[lensymbol]); + } + + for (i = offsindex; i > 0; i--) + rar->oldoffset[i] = rar->oldoffset[i-1]; + rar->oldoffset[0] = offs; + } + else if(symbol<=270) + { + offs = shortbases[symbol-263] + 1; + if(shortbits[symbol-263] > 0) { + if (!rar_br_read_ahead(a, br, shortbits[symbol-263])) + goto truncated_data; + offs += rar_br_bits(br, shortbits[symbol-263]); + rar_br_consume(br, shortbits[symbol-263]); + } + + len = 2; + + for(i = 3; i > 0; i--) + rar->oldoffset[i] = rar->oldoffset[i-1]; + rar->oldoffset[0] = offs; + } + else + { + if (symbol-271 > lengthb_min) + goto bad_data; + len = lengthbases[symbol-271]+3; + if(lengthbits[symbol-271] > 0) { + if (!rar_br_read_ahead(a, br, lengthbits[symbol-271])) + goto truncated_data; + len += rar_br_bits(br, lengthbits[symbol-271]); + rar_br_consume(br, lengthbits[symbol-271]); + } + + if ((offssymbol = read_next_symbol(a, &rar->offsetcode)) < 0) + goto bad_data; + if (offssymbol > offsetb_min) + goto bad_data; + offs = offsetbases[offssymbol]+1; + if(offsetbits[offssymbol] > 0) + { + if(offssymbol > 9) + { + if(offsetbits[offssymbol] > 4) { + if (!rar_br_read_ahead(a, br, offsetbits[offssymbol] - 4)) + goto truncated_data; + offs += rar_br_bits(br, offsetbits[offssymbol] - 4) << 4; + rar_br_consume(br, offsetbits[offssymbol] - 4); + } + + if(rar->numlowoffsetrepeats > 0) + { + rar->numlowoffsetrepeats--; + offs += rar->lastlowoffset; + } + else + { + if ((lowoffsetsymbol = + read_next_symbol(a, &rar->lowoffsetcode)) < 0) + return (ARCHIVE_FATAL); + if(lowoffsetsymbol == 16) + { + rar->numlowoffsetrepeats = 15; + offs += rar->lastlowoffset; + } + else + { + offs += lowoffsetsymbol; + rar->lastlowoffset = lowoffsetsymbol; + } + } + } + else { + if (!rar_br_read_ahead(a, br, offsetbits[offssymbol])) + goto truncated_data; + offs += rar_br_bits(br, offsetbits[offssymbol]); + rar_br_consume(br, offsetbits[offssymbol]); + } + } + + if (offs >= 0x40000) + len++; + if (offs >= 0x2000) + len++; + + for(i = 3; i > 0; i--) + rar->oldoffset[i] = rar->oldoffset[i-1]; + rar->oldoffset[0] = offs; + } + + rar->lastoffset = offs; + rar->lastlength = len; + rar->output_last_match = 1; + } +truncated_data: + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated RAR file data"); + rar->valid = 0; + return (ARCHIVE_FATAL); +bad_data: + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Bad RAR file data"); + return (ARCHIVE_FATAL); +} + +static int +copy_from_lzss_window(struct archive_read *a, const void **buffer, + int64_t startpos, int length) +{ + int windowoffs, firstpart; + struct rar *rar = (struct rar *)(a->format->data); + + if (!rar->unp_buffer) + { + if ((rar->unp_buffer = malloc(rar->unp_buffer_size)) == NULL) + { + archive_set_error(&a->archive, ENOMEM, + "Unable to allocate memory for uncompressed data."); + return (ARCHIVE_FATAL); + } + } + + windowoffs = lzss_offset_for_position(&rar->lzss, startpos); + if(windowoffs + length <= lzss_size(&rar->lzss)) { + memcpy(&rar->unp_buffer[rar->unp_offset], &rar->lzss.window[windowoffs], + length); + } else if (length <= lzss_size(&rar->lzss)) { + firstpart = lzss_size(&rar->lzss) - windowoffs; + if (firstpart < 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Bad RAR file data"); + return (ARCHIVE_FATAL); + } + if (firstpart < length) { + memcpy(&rar->unp_buffer[rar->unp_offset], + &rar->lzss.window[windowoffs], firstpart); + memcpy(&rar->unp_buffer[rar->unp_offset + firstpart], + &rar->lzss.window[0], length - firstpart); + } else { + memcpy(&rar->unp_buffer[rar->unp_offset], + &rar->lzss.window[windowoffs], length); + } + } else { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Bad RAR file data"); + return (ARCHIVE_FATAL); + } + rar->unp_offset += length; + if (rar->unp_offset >= rar->unp_buffer_size) + *buffer = rar->unp_buffer; + else + *buffer = NULL; + return (ARCHIVE_OK); +} + +static const void * +rar_read_ahead(struct archive_read *a, size_t min, ssize_t *avail) +{ + struct rar *rar = (struct rar *)(a->format->data); + const void *h = __archive_read_ahead(a, min, avail); + int ret; + if (avail) + { + if (a->archive.read_data_is_posix_read && *avail > (ssize_t)a->archive.read_data_requested) + *avail = a->archive.read_data_requested; + if (*avail > rar->bytes_remaining) + *avail = (ssize_t)rar->bytes_remaining; + if (*avail < 0) + return NULL; + else if (*avail == 0 && rar->main_flags & MHD_VOLUME && + rar->file_flags & FHD_SPLIT_AFTER) + { + rar->filename_must_match = 1; + ret = archive_read_format_rar_read_header(a, a->entry); + if (ret == (ARCHIVE_EOF)) + { + rar->has_endarc_header = 1; + ret = archive_read_format_rar_read_header(a, a->entry); + } + rar->filename_must_match = 0; + if (ret != (ARCHIVE_OK)) + return NULL; + return rar_read_ahead(a, min, avail); + } + } + return h; +} diff --git a/c/archive_read_support_format_rar5.c b/c/archive_read_support_format_rar5.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_format_rar5.c @@ -0,0 +1,4097 @@ +/*- +* Copyright (c) 2018 Grzegorz Antoniak (http://antoniak.org) +* All rights reserved. +* +* Redistribution and use in source and binary forms, with or without +* modification, are permitted provided that the following conditions +* are met: +* 1. Redistributions of source code must retain the above copyright +* notice, this list of conditions and the following disclaimer. +* 2. Redistributions in binary form must reproduce the above copyright +* notice, this list of conditions and the following disclaimer in the +* documentation and/or other materials provided with the distribution. +* +* THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR +* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES +* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. +* IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, +* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT +* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF +* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include "archive_platform.h" +#include "archive_endian.h" + +#ifdef HAVE_ERRNO_H +#include +#endif +#include +#ifdef HAVE_ZLIB_H +#include /* crc32 */ +#endif +#ifdef HAVE_LIMITS_H +#include +#endif + +#include "archive.h" +#ifndef HAVE_ZLIB_H +#include "archive_crc32.h" +#endif + +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_ppmd7_private.h" +#include "archive_entry_private.h" + +#ifdef HAVE_BLAKE2_H +#include +#else +#include "archive_blake2.h" +#endif + +/*#define CHECK_CRC_ON_SOLID_SKIP*/ +/*#define DONT_FAIL_ON_CRC_ERROR*/ +/*#define DEBUG*/ + +#define rar5_min(a, b) (((a) > (b)) ? (b) : (a)) +#define rar5_max(a, b) (((a) > (b)) ? (a) : (b)) +#define rar5_countof(X) ((const ssize_t) (sizeof(X) / sizeof(*X))) + +#if defined DEBUG +#define DEBUG_CODE if(1) +#define LOG(...) do { printf("rar5: " __VA_ARGS__); puts(""); } while(0) +#else +#define DEBUG_CODE if(0) +#endif + +/* Real RAR5 magic number is: + * + * 0x52, 0x61, 0x72, 0x21, 0x1a, 0x07, 0x01, 0x00 + * "Rar!→•☺·\x00" + * + * Retrieved with `rar5_signature()` by XOR'ing it with 0xA1, because I don't + * want to put this magic sequence in each binary that uses libarchive, so + * applications that scan through the file for this marker won't trigger on + * this "false" one. + * + * The array itself is decrypted in `rar5_init` function. */ + +static unsigned char rar5_signature_xor[] = { 243, 192, 211, 128, 187, 166, 160, 161 }; +static const size_t g_unpack_window_size = 0x20000; + +/* These could have been static const's, but they aren't, because of + * Visual Studio. */ +#define MAX_NAME_IN_CHARS 2048 +#define MAX_NAME_IN_BYTES (4 * MAX_NAME_IN_CHARS) + +struct file_header { + ssize_t bytes_remaining; + ssize_t unpacked_size; + int64_t last_offset; /* Used in sanity checks. */ + int64_t last_size; /* Used in sanity checks. */ + + uint8_t solid : 1; /* Is this a solid stream? */ + uint8_t service : 1; /* Is this file a service data? */ + uint8_t eof : 1; /* Did we finish unpacking the file? */ + uint8_t dir : 1; /* Is this file entry a directory? */ + + /* Optional time fields. */ + uint64_t e_mtime; + uint64_t e_ctime; + uint64_t e_atime; + uint32_t e_unix_ns; + + /* Optional hash fields. */ + uint32_t stored_crc32; + uint32_t calculated_crc32; + uint8_t blake2sp[32]; + blake2sp_state b2state; + char has_blake2; + + /* Optional redir fields */ + uint64_t redir_type; + uint64_t redir_flags; + + ssize_t solid_window_size; /* Used in file format check. */ +}; + +enum EXTRA { + EX_CRYPT = 0x01, + EX_HASH = 0x02, + EX_HTIME = 0x03, + EX_VERSION = 0x04, + EX_REDIR = 0x05, + EX_UOWNER = 0x06, + EX_SUBDATA = 0x07 +}; + +#define REDIR_SYMLINK_IS_DIR 1 + +enum REDIR_TYPE { + REDIR_TYPE_NONE = 0, + REDIR_TYPE_UNIXSYMLINK = 1, + REDIR_TYPE_WINSYMLINK = 2, + REDIR_TYPE_JUNCTION = 3, + REDIR_TYPE_HARDLINK = 4, + REDIR_TYPE_FILECOPY = 5, +}; + +#define OWNER_USER_NAME 0x01 +#define OWNER_GROUP_NAME 0x02 +#define OWNER_USER_UID 0x04 +#define OWNER_GROUP_GID 0x08 +#define OWNER_MAXNAMELEN 256 + +enum FILTER_TYPE { + FILTER_DELTA = 0, /* Generic pattern. */ + FILTER_E8 = 1, /* Intel x86 code. */ + FILTER_E8E9 = 2, /* Intel x86 code. */ + FILTER_ARM = 3, /* ARM code. */ + FILTER_AUDIO = 4, /* Audio filter, not used in RARv5. */ + FILTER_RGB = 5, /* Color palette, not used in RARv5. */ + FILTER_ITANIUM = 6, /* Intel's Itanium, not used in RARv5. */ + FILTER_PPM = 7, /* Predictive pattern matching, not used in + RARv5. */ + FILTER_NONE = 8, +}; + +struct filter_info { + int type; + int channels; + int pos_r; + + int64_t block_start; + ssize_t block_length; + uint16_t width; +}; + +struct data_ready { + char used; + const uint8_t* buf; + size_t size; + int64_t offset; +}; + +struct cdeque { + uint16_t beg_pos; + uint16_t end_pos; + uint16_t cap_mask; + uint16_t size; + size_t* arr; +}; + +struct decode_table { + uint32_t size; + int32_t decode_len[16]; + uint32_t decode_pos[16]; + uint32_t quick_bits; + uint8_t quick_len[1 << 10]; + uint16_t quick_num[1 << 10]; + uint16_t decode_num[306]; +}; + +struct comp_state { + /* Flag used to specify if unpacker needs to reinitialize the + uncompression context. */ + uint8_t initialized : 1; + + /* Flag used when applying filters. */ + uint8_t all_filters_applied : 1; + + /* Flag used to skip file context reinitialization, used when unpacker + is skipping through different multivolume archives. */ + uint8_t switch_multivolume : 1; + + /* Flag used to specify if unpacker has processed the whole data block + or just a part of it. */ + uint8_t block_parsing_finished : 1; + + signed int notused : 4; + + int flags; /* Uncompression flags. */ + int method; /* Uncompression algorithm method. */ + int version; /* Uncompression algorithm version. */ + ssize_t window_size; /* Size of window_buf. */ + uint8_t* window_buf; /* Circular buffer used during + decompression. */ + uint8_t* filtered_buf; /* Buffer used when applying filters. */ + const uint8_t* block_buf; /* Buffer used when merging blocks. */ + size_t window_mask; /* Convenience field; window_size - 1. */ + int64_t write_ptr; /* This amount of data has been unpacked + in the window buffer. */ + int64_t last_write_ptr; /* This amount of data has been stored in + the output file. */ + int64_t last_unstore_ptr; /* Counter of bytes extracted during + unstoring. This is separate from + last_write_ptr because of how SERVICE + base blocks are handled during skipping + in solid multiarchive archives. */ + int64_t solid_offset; /* Additional offset inside the window + buffer, used in unpacking solid + archives. */ + ssize_t cur_block_size; /* Size of current data block. */ + int last_len; /* Flag used in lzss decompression. */ + + /* Decode tables used during lzss uncompression. */ + +#define HUFF_BC 20 + struct decode_table bd; /* huffman bit lengths */ +#define HUFF_NC 306 + struct decode_table ld; /* literals */ +#define HUFF_DC 64 + struct decode_table dd; /* distances */ +#define HUFF_LDC 16 + struct decode_table ldd; /* lower bits of distances */ +#define HUFF_RC 44 + struct decode_table rd; /* repeating distances */ +#define HUFF_TABLE_SIZE (HUFF_NC + HUFF_DC + HUFF_RC + HUFF_LDC) + + /* Circular deque for storing filters. */ + struct cdeque filters; + int64_t last_block_start; /* Used for sanity checking. */ + ssize_t last_block_length; /* Used for sanity checking. */ + + /* Distance cache used during lzss uncompression. */ + int dist_cache[4]; + + /* Data buffer stack. */ + struct data_ready dready[2]; +}; + +/* Bit reader state. */ +struct bit_reader { + int8_t bit_addr; /* Current bit pointer inside current byte. */ + int in_addr; /* Current byte pointer. */ +}; + +/* RARv5 block header structure. Use bf_* functions to get values from + * block_flags_u8 field. I.e. bf_byte_count, etc. */ +struct compressed_block_header { + /* block_flags_u8 contain fields encoded in little-endian bitfield: + * + * - table present flag (shr 7, and 1), + * - last block flag (shr 6, and 1), + * - byte_count (shr 3, and 7), + * - bit_size (shr 0, and 7). + */ + uint8_t block_flags_u8; + uint8_t block_cksum; +}; + +/* RARv5 main header structure. */ +struct main_header { + /* Does the archive contain solid streams? */ + uint8_t solid : 1; + + /* If this a multi-file archive? */ + uint8_t volume : 1; + uint8_t endarc : 1; + uint8_t notused : 5; + + unsigned int vol_no; +}; + +struct generic_header { + uint8_t split_after : 1; + uint8_t split_before : 1; + uint8_t padding : 6; + int size; + int last_header_id; +}; + +struct multivolume { + unsigned int expected_vol_no; + uint8_t* push_buf; +}; + +/* Main context structure. */ +struct rar5 { + int header_initialized; + + /* Set to 1 if current file is positioned AFTER the magic value + * of the archive file. This is used in header reading functions. */ + int skipped_magic; + + /* Set to not zero if we're in skip mode (either by calling + * rar5_data_skip function or when skipping over solid streams). + * Set to 0 when in * extraction mode. This is used during checksum + * calculation functions. */ + int skip_mode; + + /* Set to not zero if we're in block merging mode (i.e. when switching + * to another file in multivolume archive, last block from 1st archive + * needs to be merged with 1st block from 2nd archive). This flag + * guards against recursive use of the merging function, which doesn't + * support recursive calls. */ + int merge_mode; + + /* An offset to QuickOpen list. This is not supported by this unpacker, + * because we're focusing on streaming interface. QuickOpen is designed + * to make things quicker for non-stream interfaces, so it's not our + * use case. */ + uint64_t qlist_offset; + + /* An offset to additional Recovery data. This is not supported by this + * unpacker. Recovery data are additional Reed-Solomon codes that could + * be used to calculate bytes that are missing in archive or are + * corrupted. */ + uint64_t rr_offset; + + /* Various context variables grouped to different structures. */ + struct generic_header generic; + struct main_header main; + struct comp_state cstate; + struct file_header file; + struct bit_reader bits; + struct multivolume vol; + + /* The header of currently processed RARv5 block. Used in main + * decompression logic loop. */ + struct compressed_block_header last_block_hdr; +}; + +/* Forward function declarations. */ + +static void rar5_signature(char *buf); +static int verify_global_checksums(struct archive_read* a); +static int rar5_read_data_skip(struct archive_read *a); +static int push_data_ready(struct archive_read* a, struct rar5* rar, + const uint8_t* buf, size_t size, int64_t offset); + +/* CDE_xxx = Circular Double Ended (Queue) return values. */ +enum CDE_RETURN_VALUES { + CDE_OK, CDE_ALLOC, CDE_PARAM, CDE_OUT_OF_BOUNDS, +}; + +/* Clears the contents of this circular deque. */ +static void cdeque_clear(struct cdeque* d) { + d->size = 0; + d->beg_pos = 0; + d->end_pos = 0; +} + +/* Creates a new circular deque object. Capacity must be power of 2: 8, 16, 32, + * 64, 256, etc. When the user will add another item above current capacity, + * the circular deque will overwrite the oldest entry. */ +static int cdeque_init(struct cdeque* d, int max_capacity_power_of_2) { + if(d == NULL || max_capacity_power_of_2 == 0) + return CDE_PARAM; + + d->cap_mask = max_capacity_power_of_2 - 1; + d->arr = NULL; + + if((max_capacity_power_of_2 & d->cap_mask) != 0) + return CDE_PARAM; + + cdeque_clear(d); + d->arr = malloc(sizeof(void*) * max_capacity_power_of_2); + + return d->arr ? CDE_OK : CDE_ALLOC; +} + +/* Return the current size (not capacity) of circular deque `d`. */ +static size_t cdeque_size(struct cdeque* d) { + return d->size; +} + +/* Returns the first element of current circular deque. Note that this function + * doesn't perform any bounds checking. If you need bounds checking, use + * `cdeque_front()` function instead. */ +static void cdeque_front_fast(struct cdeque* d, void** value) { + *value = (void*) d->arr[d->beg_pos]; +} + +/* Returns the first element of current circular deque. This function + * performs bounds checking. */ +static int cdeque_front(struct cdeque* d, void** value) { + if(d->size > 0) { + cdeque_front_fast(d, value); + return CDE_OK; + } else + return CDE_OUT_OF_BOUNDS; +} + +/* Pushes a new element into the end of this circular deque object. If current + * size will exceed capacity, the oldest element will be overwritten. */ +static int cdeque_push_back(struct cdeque* d, void* item) { + if(d == NULL) + return CDE_PARAM; + + if(d->size == d->cap_mask + 1) + return CDE_OUT_OF_BOUNDS; + + d->arr[d->end_pos] = (size_t) item; + d->end_pos = (d->end_pos + 1) & d->cap_mask; + d->size++; + + return CDE_OK; +} + +/* Pops a front element of this circular deque object and returns its value. + * This function doesn't perform any bounds checking. */ +static void cdeque_pop_front_fast(struct cdeque* d, void** value) { + *value = (void*) d->arr[d->beg_pos]; + d->beg_pos = (d->beg_pos + 1) & d->cap_mask; + d->size--; +} + +/* Pops a front element of this circular deque object and returns its value. + * This function performs bounds checking. */ +static int cdeque_pop_front(struct cdeque* d, void** value) { + if(!d || !value) + return CDE_PARAM; + + if(d->size == 0) + return CDE_OUT_OF_BOUNDS; + + cdeque_pop_front_fast(d, value); + return CDE_OK; +} + +/* Convenience function to cast filter_info** to void **. */ +static void** cdeque_filter_p(struct filter_info** f) { + return (void**) (size_t) f; +} + +/* Convenience function to cast filter_info* to void *. */ +static void* cdeque_filter(struct filter_info* f) { + return (void**) (size_t) f; +} + +/* Destroys this circular deque object. Deallocates the memory of the + * collection buffer, but doesn't deallocate the memory of any pointer passed + * to this deque as a value. */ +static void cdeque_free(struct cdeque* d) { + if(!d) + return; + + if(!d->arr) + return; + + free(d->arr); + + d->arr = NULL; + d->beg_pos = -1; + d->end_pos = -1; + d->cap_mask = 0; +} + +static inline +uint8_t bf_bit_size(const struct compressed_block_header* hdr) { + return hdr->block_flags_u8 & 7; +} + +static inline +uint8_t bf_byte_count(const struct compressed_block_header* hdr) { + return (hdr->block_flags_u8 >> 3) & 7; +} + +static inline +uint8_t bf_is_table_present(const struct compressed_block_header* hdr) { + return (hdr->block_flags_u8 >> 7) & 1; +} + +static inline struct rar5* get_context(struct archive_read* a) { + return (struct rar5*) a->format->data; +} + +/* Convenience functions used by filter implementations. */ +static void circular_memcpy(uint8_t* dst, uint8_t* window, const uint64_t mask, + int64_t start, int64_t end) +{ + if((start & mask) > (end & mask)) { + ssize_t len1 = mask + 1 - (start & mask); + ssize_t len2 = end & mask; + + memcpy(dst, &window[start & mask], len1); + memcpy(dst + len1, window, len2); + } else { + memcpy(dst, &window[start & mask], (size_t) (end - start)); + } +} + +static uint32_t read_filter_data(struct rar5* rar, uint32_t offset) { + uint8_t linear_buf[4]; + circular_memcpy(linear_buf, rar->cstate.window_buf, + rar->cstate.window_mask, offset, offset + 4); + return archive_le32dec(linear_buf); +} + +static void write_filter_data(struct rar5* rar, uint32_t offset, + uint32_t value) +{ + archive_le32enc(&rar->cstate.filtered_buf[offset], value); +} + +/* Allocates a new filter descriptor and adds it to the filter array. */ +static struct filter_info* add_new_filter(struct rar5* rar) { + struct filter_info* f = + (struct filter_info*) calloc(1, sizeof(struct filter_info)); + + if(!f) { + return NULL; + } + + cdeque_push_back(&rar->cstate.filters, cdeque_filter(f)); + return f; +} + +static int run_delta_filter(struct rar5* rar, struct filter_info* flt) { + int i; + ssize_t dest_pos, src_pos = 0; + + for(i = 0; i < flt->channels; i++) { + uint8_t prev_byte = 0; + for(dest_pos = i; + dest_pos < flt->block_length; + dest_pos += flt->channels) + { + uint8_t byte; + + byte = rar->cstate.window_buf[ + (rar->cstate.solid_offset + flt->block_start + + src_pos) & rar->cstate.window_mask]; + + prev_byte -= byte; + rar->cstate.filtered_buf[dest_pos] = prev_byte; + src_pos++; + } + } + + return ARCHIVE_OK; +} + +static int run_e8e9_filter(struct rar5* rar, struct filter_info* flt, + int extended) +{ + const uint32_t file_size = 0x1000000; + ssize_t i; + + circular_memcpy(rar->cstate.filtered_buf, + rar->cstate.window_buf, rar->cstate.window_mask, + rar->cstate.solid_offset + flt->block_start, + rar->cstate.solid_offset + flt->block_start + flt->block_length); + + for(i = 0; i < flt->block_length - 4;) { + uint8_t b = rar->cstate.window_buf[ + (rar->cstate.solid_offset + flt->block_start + + i++) & rar->cstate.window_mask]; + + /* + * 0xE8 = x86's call (function call) + * 0xE9 = x86's jmp (unconditional jump) + */ + if(b == 0xE8 || (extended && b == 0xE9)) { + + uint32_t addr; + uint32_t offset = (i + flt->block_start) % file_size; + + addr = read_filter_data(rar, + (uint32_t)(rar->cstate.solid_offset + + flt->block_start + i) & rar->cstate.window_mask); + + if(addr & 0x80000000) { + if(((addr + offset) & 0x80000000) == 0) { + write_filter_data(rar, (uint32_t)i, + addr + file_size); + } + } else { + if((addr - file_size) & 0x80000000) { + uint32_t naddr = addr - offset; + write_filter_data(rar, (uint32_t)i, + naddr); + } + } + + i += 4; + } + } + + return ARCHIVE_OK; +} + +static int run_arm_filter(struct rar5* rar, struct filter_info* flt) { + ssize_t i = 0; + uint32_t offset; + + circular_memcpy(rar->cstate.filtered_buf, + rar->cstate.window_buf, rar->cstate.window_mask, + rar->cstate.solid_offset + flt->block_start, + rar->cstate.solid_offset + flt->block_start + flt->block_length); + + for(i = 0; i < flt->block_length - 3; i += 4) { + uint8_t* b = &rar->cstate.window_buf[ + (rar->cstate.solid_offset + + flt->block_start + i + 3) & rar->cstate.window_mask]; + + if(*b == 0xEB) { + /* 0xEB = ARM's BL (branch + link) instruction. */ + offset = read_filter_data(rar, + (rar->cstate.solid_offset + flt->block_start + i) & + rar->cstate.window_mask) & 0x00ffffff; + + offset -= (uint32_t) ((i + flt->block_start) / 4); + offset = (offset & 0x00ffffff) | 0xeb000000; + write_filter_data(rar, (uint32_t)i, offset); + } + } + + return ARCHIVE_OK; +} + +static int run_filter(struct archive_read* a, struct filter_info* flt) { + int ret; + struct rar5* rar = get_context(a); + + free(rar->cstate.filtered_buf); + + rar->cstate.filtered_buf = malloc(flt->block_length); + if(!rar->cstate.filtered_buf) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for filter data."); + return ARCHIVE_FATAL; + } + + switch(flt->type) { + case FILTER_DELTA: + ret = run_delta_filter(rar, flt); + break; + + case FILTER_E8: + /* fallthrough */ + case FILTER_E8E9: + ret = run_e8e9_filter(rar, flt, + flt->type == FILTER_E8E9); + break; + + case FILTER_ARM: + ret = run_arm_filter(rar, flt); + break; + + default: + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Unsupported filter type: 0x%x", flt->type); + return ARCHIVE_FATAL; + } + + if(ret != ARCHIVE_OK) { + /* Filter has failed. */ + return ret; + } + + if(ARCHIVE_OK != push_data_ready(a, rar, rar->cstate.filtered_buf, + flt->block_length, rar->cstate.last_write_ptr)) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, + "Stack overflow when submitting unpacked data"); + + return ARCHIVE_FATAL; + } + + rar->cstate.last_write_ptr += flt->block_length; + return ARCHIVE_OK; +} + +/* The `push_data` function submits the selected data range to the user. + * Next call of `use_data` will use the pointer, size and offset arguments + * that are specified here. These arguments are pushed to the FIFO stack here, + * and popped from the stack by the `use_data` function. */ +static void push_data(struct archive_read* a, struct rar5* rar, + const uint8_t* buf, int64_t idx_begin, int64_t idx_end) +{ + const uint64_t wmask = rar->cstate.window_mask; + const ssize_t solid_write_ptr = (rar->cstate.solid_offset + + rar->cstate.last_write_ptr) & wmask; + + idx_begin += rar->cstate.solid_offset; + idx_end += rar->cstate.solid_offset; + + /* Check if our unpacked data is wrapped inside the window circular + * buffer. If it's not wrapped, it can be copied out by using + * a single memcpy, but when it's wrapped, we need to copy the first + * part with one memcpy, and the second part with another memcpy. */ + + if((idx_begin & wmask) > (idx_end & wmask)) { + /* The data is wrapped (begin offset sis bigger than end + * offset). */ + const ssize_t frag1_size = rar->cstate.window_size - + (idx_begin & wmask); + const ssize_t frag2_size = idx_end & wmask; + + /* Copy the first part of the buffer first. */ + push_data_ready(a, rar, buf + solid_write_ptr, frag1_size, + rar->cstate.last_write_ptr); + + /* Copy the second part of the buffer. */ + push_data_ready(a, rar, buf, frag2_size, + rar->cstate.last_write_ptr + frag1_size); + + rar->cstate.last_write_ptr += frag1_size + frag2_size; + } else { + /* Data is not wrapped, so we can just use one call to copy the + * data. */ + push_data_ready(a, rar, + buf + solid_write_ptr, (idx_end - idx_begin) & wmask, + rar->cstate.last_write_ptr); + + rar->cstate.last_write_ptr += idx_end - idx_begin; + } +} + +/* Convenience function that submits the data to the user. It uses the + * unpack window buffer as a source location. */ +static void push_window_data(struct archive_read* a, struct rar5* rar, + int64_t idx_begin, int64_t idx_end) +{ + push_data(a, rar, rar->cstate.window_buf, idx_begin, idx_end); +} + +static int apply_filters(struct archive_read* a) { + struct filter_info* flt; + struct rar5* rar = get_context(a); + int ret; + + rar->cstate.all_filters_applied = 0; + + /* Get the first filter that can be applied to our data. The data + * needs to be fully unpacked before the filter can be run. */ + if(CDE_OK == cdeque_front(&rar->cstate.filters, + cdeque_filter_p(&flt))) { + /* Check if our unpacked data fully covers this filter's + * range. */ + if(rar->cstate.write_ptr > flt->block_start && + rar->cstate.write_ptr >= flt->block_start + + flt->block_length) { + /* Check if we have some data pending to be written + * right before the filter's start offset. */ + if(rar->cstate.last_write_ptr == flt->block_start) { + /* Run the filter specified by descriptor + * `flt`. */ + ret = run_filter(a, flt); + if(ret != ARCHIVE_OK) { + /* Filter failure, return error. */ + return ret; + } + + /* Filter descriptor won't be needed anymore + * after it's used, * so remove it from the + * filter list and free its memory. */ + (void) cdeque_pop_front(&rar->cstate.filters, + cdeque_filter_p(&flt)); + + free(flt); + } else { + /* We can't run filters yet, dump the memory + * right before the filter. */ + push_window_data(a, rar, + rar->cstate.last_write_ptr, + flt->block_start); + } + + /* Return 'filter applied or not needed' state to the + * caller. */ + return ARCHIVE_RETRY; + } + } + + rar->cstate.all_filters_applied = 1; + return ARCHIVE_OK; +} + +static void dist_cache_push(struct rar5* rar, int value) { + int* q = rar->cstate.dist_cache; + + q[3] = q[2]; + q[2] = q[1]; + q[1] = q[0]; + q[0] = value; +} + +static int dist_cache_touch(struct rar5* rar, int idx) { + int* q = rar->cstate.dist_cache; + int i, dist = q[idx]; + + for(i = idx; i > 0; i--) + q[i] = q[i - 1]; + + q[0] = dist; + return dist; +} + +static void free_filters(struct rar5* rar) { + struct cdeque* d = &rar->cstate.filters; + + /* Free any remaining filters. All filters should be naturally + * consumed by the unpacking function, so remaining filters after + * unpacking normally mean that unpacking wasn't successful. + * But still of course we shouldn't leak memory in such case. */ + + /* cdeque_size() is a fast operation, so we can use it as a loop + * expression. */ + while(cdeque_size(d) > 0) { + struct filter_info* f = NULL; + + /* Pop_front will also decrease the collection's size. */ + if (CDE_OK == cdeque_pop_front(d, cdeque_filter_p(&f))) + free(f); + } + + cdeque_clear(d); + + /* Also clear out the variables needed for sanity checking. */ + rar->cstate.last_block_start = 0; + rar->cstate.last_block_length = 0; +} + +static void reset_file_context(struct rar5* rar) { + memset(&rar->file, 0, sizeof(rar->file)); + blake2sp_init(&rar->file.b2state, 32); + + if(rar->main.solid) { + rar->cstate.solid_offset += rar->cstate.write_ptr; + } else { + rar->cstate.solid_offset = 0; + } + + rar->cstate.write_ptr = 0; + rar->cstate.last_write_ptr = 0; + rar->cstate.last_unstore_ptr = 0; + + rar->file.redir_type = REDIR_TYPE_NONE; + rar->file.redir_flags = 0; + + free_filters(rar); +} + +static inline int get_archive_read(struct archive* a, + struct archive_read** ar) +{ + *ar = (struct archive_read*) a; + archive_check_magic(a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, + "archive_read_support_format_rar5"); + + return ARCHIVE_OK; +} + +static int read_ahead(struct archive_read* a, size_t how_many, + const uint8_t** ptr) +{ + ssize_t avail = -1; + if(!ptr) + return 0; + + *ptr = __archive_read_ahead(a, how_many, &avail); + if(*ptr == NULL) { + return 0; + } + + return 1; +} + +static int consume(struct archive_read* a, int64_t how_many) { + int ret; + + ret = how_many == __archive_read_consume(a, how_many) + ? ARCHIVE_OK + : ARCHIVE_FATAL; + + return ret; +} + +/** + * Read a RAR5 variable sized numeric value. This value will be stored in + * `pvalue`. The `pvalue_len` argument points to a variable that will receive + * the byte count that was consumed in order to decode the `pvalue` value, plus + * one. + * + * pvalue_len is optional and can be NULL. + * + * NOTE: if `pvalue_len` is NOT NULL, the caller needs to manually consume + * the number of bytes that `pvalue_len` value contains. If the `pvalue_len` + * is NULL, this consuming operation is done automatically. + * + * Returns 1 if *pvalue was successfully read. + * Returns 0 if there was an error. In this case, *pvalue contains an + * invalid value. + */ + +static int read_var(struct archive_read* a, uint64_t* pvalue, + uint64_t* pvalue_len) +{ + uint64_t result = 0; + size_t shift, i; + const uint8_t* p; + uint8_t b; + + /* We will read maximum of 8 bytes. We don't have to handle the + * situation to read the RAR5 variable-sized value stored at the end of + * the file, because such situation will never happen. */ + if(!read_ahead(a, 8, &p)) + return 0; + + for(shift = 0, i = 0; i < 8; i++, shift += 7) { + b = p[i]; + + /* Strip the MSB from the input byte and add the resulting + * number to the `result`. */ + result += (b & (uint64_t)0x7F) << shift; + + /* MSB set to 1 means we need to continue decoding process. + * MSB set to 0 means we're done. + * + * This conditional checks for the second case. */ + if((b & 0x80) == 0) { + if(pvalue) { + *pvalue = result; + } + + /* If the caller has passed the `pvalue_len` pointer, + * store the number of consumed bytes in it and do NOT + * consume those bytes, since the caller has all the + * information it needs to perform */ + if(pvalue_len) { + *pvalue_len = 1 + i; + } else { + /* If the caller did not provide the + * `pvalue_len` pointer, it will not have the + * possibility to advance the file pointer, + * because it will not know how many bytes it + * needs to consume. This is why we handle + * such situation here automatically. */ + if(ARCHIVE_OK != consume(a, 1 + i)) { + return 0; + } + } + + /* End of decoding process, return success. */ + return 1; + } + } + + /* The decoded value takes the maximum number of 8 bytes. + * It's a maximum number of bytes, so end decoding process here + * even if the first bit of last byte is 1. */ + if(pvalue) { + *pvalue = result; + } + + if(pvalue_len) { + *pvalue_len = 9; + } else { + if(ARCHIVE_OK != consume(a, 9)) { + return 0; + } + } + + return 1; +} + +static int read_var_sized(struct archive_read* a, size_t* pvalue, + size_t* pvalue_len) +{ + uint64_t v; + uint64_t v_size = 0; + + const int ret = pvalue_len ? read_var(a, &v, &v_size) + : read_var(a, &v, NULL); + + if(ret == 1 && pvalue) { + *pvalue = (size_t) v; + } + + if(pvalue_len) { + /* Possible data truncation should be safe. */ + *pvalue_len = (size_t) v_size; + } + + return ret; +} + +static int read_bits_32(struct rar5* rar, const uint8_t* p, uint32_t* value) { + uint32_t bits = ((uint32_t) p[rar->bits.in_addr]) << 24; + bits |= p[rar->bits.in_addr + 1] << 16; + bits |= p[rar->bits.in_addr + 2] << 8; + bits |= p[rar->bits.in_addr + 3]; + bits <<= rar->bits.bit_addr; + bits |= p[rar->bits.in_addr + 4] >> (8 - rar->bits.bit_addr); + *value = bits; + return ARCHIVE_OK; +} + +static int read_bits_16(struct rar5* rar, const uint8_t* p, uint16_t* value) { + int bits = (int) ((uint32_t) p[rar->bits.in_addr]) << 16; + bits |= (int) p[rar->bits.in_addr + 1] << 8; + bits |= (int) p[rar->bits.in_addr + 2]; + bits >>= (8 - rar->bits.bit_addr); + *value = bits & 0xffff; + return ARCHIVE_OK; +} + +static void skip_bits(struct rar5* rar, int bits) { + const int new_bits = rar->bits.bit_addr + bits; + rar->bits.in_addr += new_bits >> 3; + rar->bits.bit_addr = new_bits & 7; +} + +/* n = up to 16 */ +static int read_consume_bits(struct rar5* rar, const uint8_t* p, int n, + int* value) +{ + uint16_t v; + int ret, num; + + if(n == 0 || n > 16) { + /* This is a programmer error and should never happen + * in runtime. */ + return ARCHIVE_FATAL; + } + + ret = read_bits_16(rar, p, &v); + if(ret != ARCHIVE_OK) + return ret; + + num = (int) v; + num >>= 16 - n; + + skip_bits(rar, n); + + if(value) + *value = num; + + return ARCHIVE_OK; +} + +static int read_u32(struct archive_read* a, uint32_t* pvalue) { + const uint8_t* p; + if(!read_ahead(a, 4, &p)) + return 0; + + *pvalue = archive_le32dec(p); + return ARCHIVE_OK == consume(a, 4) ? 1 : 0; +} + +static int read_u64(struct archive_read* a, uint64_t* pvalue) { + const uint8_t* p; + if(!read_ahead(a, 8, &p)) + return 0; + + *pvalue = archive_le64dec(p); + return ARCHIVE_OK == consume(a, 8) ? 1 : 0; +} + +static int bid_standard(struct archive_read* a) { + const uint8_t* p; + char signature[sizeof(rar5_signature_xor)]; + + rar5_signature(signature); + + if(!read_ahead(a, sizeof(rar5_signature_xor), &p)) + return -1; + + if(!memcmp(signature, p, sizeof(rar5_signature_xor))) + return 30; + + return -1; +} + +static int rar5_bid(struct archive_read* a, int best_bid) { + int my_bid; + + if(best_bid > 30) + return -1; + + my_bid = bid_standard(a); + if(my_bid > -1) { + return my_bid; + } + + return -1; +} + +static int rar5_options(struct archive_read *a, const char *key, + const char *val) { + (void) a; + (void) key; + (void) val; + + /* No options supported in this version. Return the ARCHIVE_WARN code + * to signal the options supervisor that the unpacker didn't handle + * setting this option. */ + + return ARCHIVE_WARN; +} + +static void init_header(struct archive_read* a) { + a->archive.archive_format = ARCHIVE_FORMAT_RAR_V5; + a->archive.archive_format_name = "RAR5"; +} + +static void init_window_mask(struct rar5* rar) { + if (rar->cstate.window_size) + rar->cstate.window_mask = rar->cstate.window_size - 1; + else + rar->cstate.window_mask = 0; +} + +enum HEADER_FLAGS { + HFL_EXTRA_DATA = 0x0001, + HFL_DATA = 0x0002, + HFL_SKIP_IF_UNKNOWN = 0x0004, + HFL_SPLIT_BEFORE = 0x0008, + HFL_SPLIT_AFTER = 0x0010, + HFL_CHILD = 0x0020, + HFL_INHERITED = 0x0040 +}; + +static int process_main_locator_extra_block(struct archive_read* a, + struct rar5* rar) +{ + uint64_t locator_flags; + + enum LOCATOR_FLAGS { + QLIST = 0x01, RECOVERY = 0x02, + }; + + if(!read_var(a, &locator_flags, NULL)) { + return ARCHIVE_EOF; + } + + if(locator_flags & QLIST) { + if(!read_var(a, &rar->qlist_offset, NULL)) { + return ARCHIVE_EOF; + } + + /* qlist is not used */ + } + + if(locator_flags & RECOVERY) { + if(!read_var(a, &rar->rr_offset, NULL)) { + return ARCHIVE_EOF; + } + + /* rr is not used */ + } + + return ARCHIVE_OK; +} + +static int parse_file_extra_hash(struct archive_read* a, struct rar5* rar, + ssize_t* extra_data_size) +{ + size_t hash_type = 0; + size_t value_len; + + enum HASH_TYPE { + BLAKE2sp = 0x00 + }; + + if(!read_var_sized(a, &hash_type, &value_len)) + return ARCHIVE_EOF; + + *extra_data_size -= value_len; + if(ARCHIVE_OK != consume(a, value_len)) { + return ARCHIVE_EOF; + } + + /* The file uses BLAKE2sp checksum algorithm instead of plain old + * CRC32. */ + if(hash_type == BLAKE2sp) { + const uint8_t* p; + const int hash_size = sizeof(rar->file.blake2sp); + + if(!read_ahead(a, hash_size, &p)) + return ARCHIVE_EOF; + + rar->file.has_blake2 = 1; + memcpy(&rar->file.blake2sp, p, hash_size); + + if(ARCHIVE_OK != consume(a, hash_size)) { + return ARCHIVE_EOF; + } + + *extra_data_size -= hash_size; + } else { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Unsupported hash type (0x%x)", (int) hash_type); + return ARCHIVE_FATAL; + } + + return ARCHIVE_OK; +} + +static uint64_t time_win_to_unix(uint64_t win_time) { + const size_t ns_in_sec = 10000000; + const uint64_t sec_to_unix = 11644473600LL; + return win_time / ns_in_sec - sec_to_unix; +} + +static int parse_htime_item(struct archive_read* a, char unix_time, + uint64_t* where, ssize_t* extra_data_size) +{ + if(unix_time) { + uint32_t time_val; + if(!read_u32(a, &time_val)) + return ARCHIVE_EOF; + + *extra_data_size -= 4; + *where = (uint64_t) time_val; + } else { + uint64_t windows_time; + if(!read_u64(a, &windows_time)) + return ARCHIVE_EOF; + + *where = time_win_to_unix(windows_time); + *extra_data_size -= 8; + } + + return ARCHIVE_OK; +} + +static int parse_file_extra_version(struct archive_read* a, + struct archive_entry* e, ssize_t* extra_data_size) +{ + size_t flags = 0; + size_t version = 0; + size_t value_len = 0; + struct archive_string version_string; + struct archive_string name_utf8_string; + const char* cur_filename; + + /* Flags are ignored. */ + if(!read_var_sized(a, &flags, &value_len)) + return ARCHIVE_EOF; + + *extra_data_size -= value_len; + if(ARCHIVE_OK != consume(a, value_len)) + return ARCHIVE_EOF; + + if(!read_var_sized(a, &version, &value_len)) + return ARCHIVE_EOF; + + *extra_data_size -= value_len; + if(ARCHIVE_OK != consume(a, value_len)) + return ARCHIVE_EOF; + + /* extra_data_size should be zero here. */ + + cur_filename = archive_entry_pathname_utf8(e); + if(cur_filename == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, + "Version entry without file name"); + return ARCHIVE_FATAL; + } + + archive_string_init(&version_string); + archive_string_init(&name_utf8_string); + + /* Prepare a ;123 suffix for the filename, where '123' is the version + * value of this file. */ + archive_string_sprintf(&version_string, ";%zu", version); + + /* Build the new filename. */ + archive_strcat(&name_utf8_string, cur_filename); + archive_strcat(&name_utf8_string, version_string.s); + + /* Apply the new filename into this file's context. */ + archive_entry_update_pathname_utf8(e, name_utf8_string.s); + + /* Free buffers. */ + archive_string_free(&version_string); + archive_string_free(&name_utf8_string); + return ARCHIVE_OK; +} + +static int parse_file_extra_htime(struct archive_read* a, + struct archive_entry* e, struct rar5* rar, ssize_t* extra_data_size) +{ + char unix_time = 0; + size_t flags = 0; + size_t value_len; + + enum HTIME_FLAGS { + IS_UNIX = 0x01, + HAS_MTIME = 0x02, + HAS_CTIME = 0x04, + HAS_ATIME = 0x08, + HAS_UNIX_NS = 0x10, + }; + + if(!read_var_sized(a, &flags, &value_len)) + return ARCHIVE_EOF; + + *extra_data_size -= value_len; + if(ARCHIVE_OK != consume(a, value_len)) { + return ARCHIVE_EOF; + } + + unix_time = flags & IS_UNIX; + + if(flags & HAS_MTIME) { + parse_htime_item(a, unix_time, &rar->file.e_mtime, + extra_data_size); + archive_entry_set_mtime(e, rar->file.e_mtime, 0); + } + + if(flags & HAS_CTIME) { + parse_htime_item(a, unix_time, &rar->file.e_ctime, + extra_data_size); + archive_entry_set_ctime(e, rar->file.e_ctime, 0); + } + + if(flags & HAS_ATIME) { + parse_htime_item(a, unix_time, &rar->file.e_atime, + extra_data_size); + archive_entry_set_atime(e, rar->file.e_atime, 0); + } + + if(flags & HAS_UNIX_NS) { + if(!read_u32(a, &rar->file.e_unix_ns)) + return ARCHIVE_EOF; + + *extra_data_size -= 4; + } + + return ARCHIVE_OK; +} + +static int parse_file_extra_redir(struct archive_read* a, + struct archive_entry* e, struct rar5* rar, ssize_t* extra_data_size) +{ + uint64_t value_size = 0; + size_t target_size = 0; + char target_utf8_buf[MAX_NAME_IN_BYTES]; + const uint8_t* p; + + if(!read_var(a, &rar->file.redir_type, &value_size)) + return ARCHIVE_EOF; + if(ARCHIVE_OK != consume(a, (int64_t)value_size)) + return ARCHIVE_EOF; + *extra_data_size -= value_size; + + if(!read_var(a, &rar->file.redir_flags, &value_size)) + return ARCHIVE_EOF; + if(ARCHIVE_OK != consume(a, (int64_t)value_size)) + return ARCHIVE_EOF; + *extra_data_size -= value_size; + + if(!read_var_sized(a, &target_size, NULL)) + return ARCHIVE_EOF; + *extra_data_size -= target_size + 1; + + if(!read_ahead(a, target_size, &p)) + return ARCHIVE_EOF; + + if(target_size > (MAX_NAME_IN_CHARS - 1)) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Link target is too long"); + return ARCHIVE_FATAL; + } + + if(target_size == 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "No link target specified"); + return ARCHIVE_FATAL; + } + + memcpy(target_utf8_buf, p, target_size); + target_utf8_buf[target_size] = 0; + + if(ARCHIVE_OK != consume(a, (int64_t)target_size)) + return ARCHIVE_EOF; + + switch(rar->file.redir_type) { + case REDIR_TYPE_UNIXSYMLINK: + case REDIR_TYPE_WINSYMLINK: + archive_entry_set_filetype(e, AE_IFLNK); + archive_entry_update_symlink_utf8(e, target_utf8_buf); + if (rar->file.redir_flags & REDIR_SYMLINK_IS_DIR) { + archive_entry_set_symlink_type(e, + AE_SYMLINK_TYPE_DIRECTORY); + } else { + archive_entry_set_symlink_type(e, + AE_SYMLINK_TYPE_FILE); + } + break; + + case REDIR_TYPE_HARDLINK: + archive_entry_set_filetype(e, AE_IFREG); + archive_entry_update_hardlink_utf8(e, target_utf8_buf); + break; + + default: + /* Unknown redir type, skip it. */ + break; + } + return ARCHIVE_OK; +} + +static int parse_file_extra_owner(struct archive_read* a, + struct archive_entry* e, ssize_t* extra_data_size) +{ + uint64_t flags = 0; + uint64_t value_size = 0; + uint64_t id = 0; + size_t name_len = 0; + size_t name_size = 0; + char namebuf[OWNER_MAXNAMELEN]; + const uint8_t* p; + + if(!read_var(a, &flags, &value_size)) + return ARCHIVE_EOF; + if(ARCHIVE_OK != consume(a, (int64_t)value_size)) + return ARCHIVE_EOF; + *extra_data_size -= value_size; + + if ((flags & OWNER_USER_NAME) != 0) { + if(!read_var_sized(a, &name_size, NULL)) + return ARCHIVE_EOF; + *extra_data_size -= name_size + 1; + + if(!read_ahead(a, name_size, &p)) + return ARCHIVE_EOF; + + if (name_size >= OWNER_MAXNAMELEN) { + name_len = OWNER_MAXNAMELEN - 1; + } else { + name_len = name_size; + } + + memcpy(namebuf, p, name_len); + namebuf[name_len] = 0; + if(ARCHIVE_OK != consume(a, (int64_t)name_size)) + return ARCHIVE_EOF; + + archive_entry_set_uname(e, namebuf); + } + if ((flags & OWNER_GROUP_NAME) != 0) { + if(!read_var_sized(a, &name_size, NULL)) + return ARCHIVE_EOF; + *extra_data_size -= name_size + 1; + + if(!read_ahead(a, name_size, &p)) + return ARCHIVE_EOF; + + if (name_size >= OWNER_MAXNAMELEN) { + name_len = OWNER_MAXNAMELEN - 1; + } else { + name_len = name_size; + } + + memcpy(namebuf, p, name_len); + namebuf[name_len] = 0; + if(ARCHIVE_OK != consume(a, (int64_t)name_size)) + return ARCHIVE_EOF; + + archive_entry_set_gname(e, namebuf); + } + if ((flags & OWNER_USER_UID) != 0) { + if(!read_var(a, &id, &value_size)) + return ARCHIVE_EOF; + if(ARCHIVE_OK != consume(a, (int64_t)value_size)) + return ARCHIVE_EOF; + *extra_data_size -= value_size; + + archive_entry_set_uid(e, (la_int64_t)id); + } + if ((flags & OWNER_GROUP_GID) != 0) { + if(!read_var(a, &id, &value_size)) + return ARCHIVE_EOF; + if(ARCHIVE_OK != consume(a, (int64_t)value_size)) + return ARCHIVE_EOF; + *extra_data_size -= value_size; + + archive_entry_set_gid(e, (la_int64_t)id); + } + return ARCHIVE_OK; +} + +static int process_head_file_extra(struct archive_read* a, + struct archive_entry* e, struct rar5* rar, ssize_t extra_data_size) +{ + size_t extra_field_size; + size_t extra_field_id = 0; + int ret = ARCHIVE_FATAL; + size_t var_size; + + while(extra_data_size > 0) { + if(!read_var_sized(a, &extra_field_size, &var_size)) + return ARCHIVE_EOF; + + extra_data_size -= var_size; + if(ARCHIVE_OK != consume(a, var_size)) { + return ARCHIVE_EOF; + } + + if(!read_var_sized(a, &extra_field_id, &var_size)) + return ARCHIVE_EOF; + + extra_data_size -= var_size; + if(ARCHIVE_OK != consume(a, var_size)) { + return ARCHIVE_EOF; + } + + switch(extra_field_id) { + case EX_HASH: + ret = parse_file_extra_hash(a, rar, + &extra_data_size); + break; + case EX_HTIME: + ret = parse_file_extra_htime(a, e, rar, + &extra_data_size); + break; + case EX_REDIR: + ret = parse_file_extra_redir(a, e, rar, + &extra_data_size); + break; + case EX_UOWNER: + ret = parse_file_extra_owner(a, e, + &extra_data_size); + break; + case EX_VERSION: + ret = parse_file_extra_version(a, e, + &extra_data_size); + break; + case EX_CRYPT: + /* fallthrough */ + case EX_SUBDATA: + /* fallthrough */ + default: + /* Skip unsupported entry. */ + return consume(a, extra_data_size); + } + } + + if(ret != ARCHIVE_OK) { + /* Attribute not implemented. */ + return ret; + } + + return ARCHIVE_OK; +} + +static int process_head_file(struct archive_read* a, struct rar5* rar, + struct archive_entry* entry, size_t block_flags) +{ + ssize_t extra_data_size = 0; + size_t data_size = 0; + size_t file_flags = 0; + size_t file_attr = 0; + size_t compression_info = 0; + size_t host_os = 0; + size_t name_size = 0; + uint64_t unpacked_size, window_size; + uint32_t mtime = 0, crc = 0; + int c_method = 0, c_version = 0; + char name_utf8_buf[MAX_NAME_IN_BYTES]; + const uint8_t* p; + + enum FILE_FLAGS { + DIRECTORY = 0x0001, UTIME = 0x0002, CRC32 = 0x0004, + UNKNOWN_UNPACKED_SIZE = 0x0008, + }; + + enum FILE_ATTRS { + ATTR_READONLY = 0x1, ATTR_HIDDEN = 0x2, ATTR_SYSTEM = 0x4, + ATTR_DIRECTORY = 0x10, + }; + + enum COMP_INFO_FLAGS { + SOLID = 0x0040, + }; + + enum HOST_OS { + HOST_WINDOWS = 0, + HOST_UNIX = 1, + }; + + archive_entry_clear(entry); + + /* Do not reset file context if we're switching archives. */ + if(!rar->cstate.switch_multivolume) { + reset_file_context(rar); + } + + if(block_flags & HFL_EXTRA_DATA) { + size_t edata_size = 0; + if(!read_var_sized(a, &edata_size, NULL)) + return ARCHIVE_EOF; + + /* Intentional type cast from unsigned to signed. */ + extra_data_size = (ssize_t) edata_size; + } + + if(block_flags & HFL_DATA) { + if(!read_var_sized(a, &data_size, NULL)) + return ARCHIVE_EOF; + + rar->file.bytes_remaining = data_size; + } else { + rar->file.bytes_remaining = 0; + + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "no data found in file/service block"); + return ARCHIVE_FATAL; + } + + if(!read_var_sized(a, &file_flags, NULL)) + return ARCHIVE_EOF; + + if(!read_var(a, &unpacked_size, NULL)) + return ARCHIVE_EOF; + + if(file_flags & UNKNOWN_UNPACKED_SIZE) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, + "Files with unknown unpacked size are not supported"); + return ARCHIVE_FATAL; + } + + rar->file.dir = (uint8_t) ((file_flags & DIRECTORY) > 0); + + if(!read_var_sized(a, &file_attr, NULL)) + return ARCHIVE_EOF; + + if(file_flags & UTIME) { + if(!read_u32(a, &mtime)) + return ARCHIVE_EOF; + } + + if(file_flags & CRC32) { + if(!read_u32(a, &crc)) + return ARCHIVE_EOF; + } + + if(!read_var_sized(a, &compression_info, NULL)) + return ARCHIVE_EOF; + + c_method = (int) (compression_info >> 7) & 0x7; + c_version = (int) (compression_info & 0x3f); + + /* RAR5 seems to limit the dictionary size to 64MB. */ + window_size = (rar->file.dir > 0) ? + 0 : + g_unpack_window_size << ((compression_info >> 10) & 15); + rar->cstate.method = c_method; + rar->cstate.version = c_version + 50; + rar->file.solid = (compression_info & SOLID) > 0; + + /* Archives which declare solid files without initializing the window + * buffer first are invalid. */ + + if(rar->file.solid > 0 && rar->cstate.window_buf == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Declared solid file, but no window buffer " + "initialized yet."); + return ARCHIVE_FATAL; + } + + /* Check if window_size is a sane value. Also, if the file is not + * declared as a directory, disallow window_size == 0. */ + if(window_size > (64 * 1024 * 1024) || + (rar->file.dir == 0 && window_size == 0)) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Declared dictionary size is not supported."); + return ARCHIVE_FATAL; + } + + if(rar->file.solid > 0) { + /* Re-check if current window size is the same as previous + * window size (for solid files only). */ + if(rar->file.solid_window_size > 0 && + rar->file.solid_window_size != (ssize_t) window_size) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Window size for this solid file doesn't match " + "the window size used in previous solid file. "); + return ARCHIVE_FATAL; + } + } + + /* If we're currently switching volumes, ignore the new definition of + * window_size. */ + if(rar->cstate.switch_multivolume == 0) { + /* Values up to 64M should fit into ssize_t on every + * architecture. */ + rar->cstate.window_size = (ssize_t) window_size; + } + + if(rar->file.solid > 0 && rar->file.solid_window_size == 0) { + /* Solid files have to have the same window_size across + whole archive. Remember the window_size parameter + for first solid file found. */ + rar->file.solid_window_size = rar->cstate.window_size; + } + + init_window_mask(rar); + + rar->file.service = 0; + + if(!read_var_sized(a, &host_os, NULL)) + return ARCHIVE_EOF; + + if(host_os == HOST_WINDOWS) { + /* Host OS is Windows */ + + __LA_MODE_T mode; + + if(file_attr & ATTR_DIRECTORY) { + if (file_attr & ATTR_READONLY) { + mode = 0555 | AE_IFDIR; + } else { + mode = 0755 | AE_IFDIR; + } + } else { + if (file_attr & ATTR_READONLY) { + mode = 0444 | AE_IFREG; + } else { + mode = 0644 | AE_IFREG; + } + } + + archive_entry_set_mode(entry, mode); + + if (file_attr & (ATTR_READONLY | ATTR_HIDDEN | ATTR_SYSTEM)) { + char *fflags_text, *ptr; + /* allocate for "rdonly,hidden,system," */ + fflags_text = malloc(22 * sizeof(char)); + if (fflags_text != NULL) { + ptr = fflags_text; + if (file_attr & ATTR_READONLY) { + strcpy(ptr, "rdonly,"); + ptr = ptr + 7; + } + if (file_attr & ATTR_HIDDEN) { + strcpy(ptr, "hidden,"); + ptr = ptr + 7; + } + if (file_attr & ATTR_SYSTEM) { + strcpy(ptr, "system,"); + ptr = ptr + 7; + } + if (ptr > fflags_text) { + /* Delete trailing comma */ + *(ptr - 1) = '\0'; + archive_entry_copy_fflags_text(entry, + fflags_text); + } + free(fflags_text); + } + } + } else if(host_os == HOST_UNIX) { + /* Host OS is Unix */ + archive_entry_set_mode(entry, (__LA_MODE_T) file_attr); + } else { + /* Unknown host OS */ + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Unsupported Host OS: 0x%x", (int) host_os); + + return ARCHIVE_FATAL; + } + + if(!read_var_sized(a, &name_size, NULL)) + return ARCHIVE_EOF; + + if(!read_ahead(a, name_size, &p)) + return ARCHIVE_EOF; + + if(name_size > (MAX_NAME_IN_CHARS - 1)) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Filename is too long"); + + return ARCHIVE_FATAL; + } + + if(name_size == 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "No filename specified"); + + return ARCHIVE_FATAL; + } + + memcpy(name_utf8_buf, p, name_size); + name_utf8_buf[name_size] = 0; + if(ARCHIVE_OK != consume(a, name_size)) { + return ARCHIVE_EOF; + } + + archive_entry_update_pathname_utf8(entry, name_utf8_buf); + + if(extra_data_size > 0) { + int ret = process_head_file_extra(a, entry, rar, + extra_data_size); + + /* + * TODO: rewrite or remove useless sanity check + * as extra_data_size is not passed as a pointer + * + if(extra_data_size < 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, + "File extra data size is not zero"); + return ARCHIVE_FATAL; + } + */ + + if(ret != ARCHIVE_OK) + return ret; + } + + if((file_flags & UNKNOWN_UNPACKED_SIZE) == 0) { + rar->file.unpacked_size = (ssize_t) unpacked_size; + if(rar->file.redir_type == REDIR_TYPE_NONE) + archive_entry_set_size(entry, unpacked_size); + } + + if(file_flags & UTIME) { + archive_entry_set_mtime(entry, (time_t) mtime, 0); + } + + if(file_flags & CRC32) { + rar->file.stored_crc32 = crc; + } + + if(!rar->cstate.switch_multivolume) { + /* Do not reinitialize unpacking state if we're switching + * archives. */ + rar->cstate.block_parsing_finished = 1; + rar->cstate.all_filters_applied = 1; + rar->cstate.initialized = 0; + } + + if(rar->generic.split_before > 0) { + /* If now we're standing on a header that has a 'split before' + * mark, it means we're standing on a 'continuation' file + * header. Signal the caller that if it wants to move to + * another file, it must call rar5_read_header() function + * again. */ + + return ARCHIVE_RETRY; + } else { + return ARCHIVE_OK; + } +} + +static int process_head_service(struct archive_read* a, struct rar5* rar, + struct archive_entry* entry, size_t block_flags) +{ + /* Process this SERVICE block the same way as FILE blocks. */ + int ret = process_head_file(a, rar, entry, block_flags); + if(ret != ARCHIVE_OK) + return ret; + + rar->file.service = 1; + + /* But skip the data part automatically. It's no use for the user + * anyway. It contains only service data, not even needed to + * properly unpack the file. */ + ret = rar5_read_data_skip(a); + if(ret != ARCHIVE_OK) + return ret; + + /* After skipping, try parsing another block automatically. */ + return ARCHIVE_RETRY; +} + +static int process_head_main(struct archive_read* a, struct rar5* rar, + struct archive_entry* entry, size_t block_flags) +{ + int ret; + size_t extra_data_size = 0; + size_t extra_field_size = 0; + size_t extra_field_id = 0; + size_t archive_flags = 0; + + enum MAIN_FLAGS { + VOLUME = 0x0001, /* multi-volume archive */ + VOLUME_NUMBER = 0x0002, /* volume number, first vol doesn't + * have it */ + SOLID = 0x0004, /* solid archive */ + PROTECT = 0x0008, /* contains Recovery info */ + LOCK = 0x0010, /* readonly flag, not used */ + }; + + enum MAIN_EXTRA { + // Just one attribute here. + LOCATOR = 0x01, + }; + + (void) entry; + + if(block_flags & HFL_EXTRA_DATA) { + if(!read_var_sized(a, &extra_data_size, NULL)) + return ARCHIVE_EOF; + } else { + extra_data_size = 0; + } + + if(!read_var_sized(a, &archive_flags, NULL)) { + return ARCHIVE_EOF; + } + + rar->main.volume = (archive_flags & VOLUME) > 0; + rar->main.solid = (archive_flags & SOLID) > 0; + + if(archive_flags & VOLUME_NUMBER) { + size_t v = 0; + if(!read_var_sized(a, &v, NULL)) { + return ARCHIVE_EOF; + } + + if (v > UINT_MAX) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid volume number"); + return ARCHIVE_FATAL; + } + + rar->main.vol_no = (unsigned int) v; + } else { + rar->main.vol_no = 0; + } + + if(rar->vol.expected_vol_no > 0 && + rar->main.vol_no != rar->vol.expected_vol_no) + { + /* Returning EOF instead of FATAL because of strange + * libarchive behavior. When opening multiple files via + * archive_read_open_filenames(), after reading up the whole + * last file, the __archive_read_ahead function wraps up to + * the first archive instead of returning EOF. */ + return ARCHIVE_EOF; + } + + if(extra_data_size == 0) { + /* Early return. */ + return ARCHIVE_OK; + } + + if(!read_var_sized(a, &extra_field_size, NULL)) { + return ARCHIVE_EOF; + } + + if(!read_var_sized(a, &extra_field_id, NULL)) { + return ARCHIVE_EOF; + } + + if(extra_field_size == 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid extra field size"); + return ARCHIVE_FATAL; + } + + switch(extra_field_id) { + case LOCATOR: + ret = process_main_locator_extra_block(a, rar); + if(ret != ARCHIVE_OK) { + /* Error while parsing main locator extra + * block. */ + return ret; + } + + break; + default: + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Unsupported extra type (0x%x)", + (int) extra_field_id); + return ARCHIVE_FATAL; + } + + return ARCHIVE_OK; +} + +static int skip_unprocessed_bytes(struct archive_read* a) { + struct rar5* rar = get_context(a); + int ret; + + if(rar->file.bytes_remaining) { + /* Use different skipping method in block merging mode than in + * normal mode. If merge mode is active, rar5_read_data_skip + * can't be used, because it could allow recursive use of + * merge_block() * function, and this function doesn't support + * recursive use. */ + if(rar->merge_mode) { + /* Discard whole merged block. This is valid in solid + * mode as well, because the code will discard blocks + * only if those blocks are safe to discard (i.e. + * they're not FILE blocks). */ + ret = consume(a, rar->file.bytes_remaining); + if(ret != ARCHIVE_OK) { + return ret; + } + rar->file.bytes_remaining = 0; + } else { + /* If we're not in merge mode, use safe skipping code. + * This will ensure we'll handle solid archives + * properly. */ + ret = rar5_read_data_skip(a); + if(ret != ARCHIVE_OK) { + return ret; + } + } + } + + return ARCHIVE_OK; +} + +static int scan_for_signature(struct archive_read* a); + +/* Base block processing function. A 'base block' is a RARv5 header block + * that tells the reader what kind of data is stored inside the block. + * + * From the birds-eye view a RAR file looks file this: + * + * ... + * + * There are a few types of base blocks. Those types are specified inside + * the 'switch' statement in this function. For example purposes, I'll write + * how a standard RARv5 file could look like here: + * + *
+ * + * The structure above could describe an archive file with 3 files in it, + * one service "QuickOpen" block (that is ignored by this parser), and an + * end of file base block marker. + * + * If the file is stored in multiple archive files ("multiarchive"), it might + * look like this: + * + * .part01.rar:
+ * .part02.rar:
+ * .part03.rar:
+ * + * This example could describe 3 RAR files that contain ONE archived file. + * Or it could describe 3 RAR files that contain 3 different files. Or 3 + * RAR files than contain 2 files. It all depends what metadata is stored in + * the headers of blocks. + * + * Each block contains info about its size, the name of the file it's + * storing inside, and whether this FILE block is a continuation block of + * previous archive ('split before'), and is this FILE block should be + * continued in another archive ('split after'). By parsing the 'split before' + * and 'split after' flags, we're able to tell if multiple base blocks + * are describing one file, or multiple files (with the same filename, for + * example). + * + * One thing to note is that if we're parsing the first block, and + * we see 'split after' flag, then we need to jump over to another + * block to be able to decompress rest of the data. To do this, we need + * to skip the block, then switch to another file, then skip the + * block,
block, and then we're standing on the proper + * block. + */ + +static int process_base_block(struct archive_read* a, + struct archive_entry* entry) +{ + const size_t SMALLEST_RAR5_BLOCK_SIZE = 3; + + struct rar5* rar = get_context(a); + uint32_t hdr_crc, computed_crc; + size_t raw_hdr_size = 0, hdr_size_len, hdr_size; + size_t header_id = 0; + size_t header_flags = 0; + const uint8_t* p; + int ret; + + enum HEADER_TYPE { + HEAD_MARK = 0x00, HEAD_MAIN = 0x01, HEAD_FILE = 0x02, + HEAD_SERVICE = 0x03, HEAD_CRYPT = 0x04, HEAD_ENDARC = 0x05, + HEAD_UNKNOWN = 0xff, + }; + + /* Skip any unprocessed data for this file. */ + ret = skip_unprocessed_bytes(a); + if(ret != ARCHIVE_OK) + return ret; + + /* Read the expected CRC32 checksum. */ + if(!read_u32(a, &hdr_crc)) { + return ARCHIVE_EOF; + } + + /* Read header size. */ + if(!read_var_sized(a, &raw_hdr_size, &hdr_size_len)) { + return ARCHIVE_EOF; + } + + hdr_size = raw_hdr_size + hdr_size_len; + + /* Sanity check, maximum header size for RAR5 is 2MB. */ + if(hdr_size > (2 * 1024 * 1024)) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Base block header is too large"); + + return ARCHIVE_FATAL; + } + + /* Additional sanity checks to weed out invalid files. */ + if(raw_hdr_size == 0 || hdr_size_len == 0 || + hdr_size < SMALLEST_RAR5_BLOCK_SIZE) + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Too small block encountered (%zu bytes)", + raw_hdr_size); + + return ARCHIVE_FATAL; + } + + /* Read the whole header data into memory, maximum memory use here is + * 2MB. */ + if(!read_ahead(a, hdr_size, &p)) { + return ARCHIVE_EOF; + } + + /* Verify the CRC32 of the header data. */ + computed_crc = (uint32_t) crc32(0, p, (int) hdr_size); + if(computed_crc != hdr_crc) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Header CRC error"); + + return ARCHIVE_FATAL; + } + + /* If the checksum is OK, we proceed with parsing. */ + if(ARCHIVE_OK != consume(a, hdr_size_len)) { + return ARCHIVE_EOF; + } + + if(!read_var_sized(a, &header_id, NULL)) + return ARCHIVE_EOF; + + if(!read_var_sized(a, &header_flags, NULL)) + return ARCHIVE_EOF; + + rar->generic.split_after = (header_flags & HFL_SPLIT_AFTER) > 0; + rar->generic.split_before = (header_flags & HFL_SPLIT_BEFORE) > 0; + rar->generic.size = (int)hdr_size; + rar->generic.last_header_id = (int)header_id; + rar->main.endarc = 0; + + /* Those are possible header ids in RARv5. */ + switch(header_id) { + case HEAD_MAIN: + ret = process_head_main(a, rar, entry, header_flags); + + /* Main header doesn't have any files in it, so it's + * pointless to return to the caller. Retry to next + * header, which should be HEAD_FILE/HEAD_SERVICE. */ + if(ret == ARCHIVE_OK) + return ARCHIVE_RETRY; + + return ret; + case HEAD_SERVICE: + ret = process_head_service(a, rar, entry, header_flags); + return ret; + case HEAD_FILE: + ret = process_head_file(a, rar, entry, header_flags); + return ret; + case HEAD_CRYPT: + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Encryption is not supported"); + return ARCHIVE_FATAL; + case HEAD_ENDARC: + rar->main.endarc = 1; + + /* After encountering an end of file marker, we need + * to take into consideration if this archive is + * continued in another file (i.e. is it part01.rar: + * is there a part02.rar?) */ + if(rar->main.volume) { + /* In case there is part02.rar, position the + * read pointer in a proper place, so we can + * resume parsing. */ + ret = scan_for_signature(a); + if(ret == ARCHIVE_FATAL) { + return ARCHIVE_EOF; + } else { + if(rar->vol.expected_vol_no == + UINT_MAX) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Header error"); + return ARCHIVE_FATAL; + } + + rar->vol.expected_vol_no = + rar->main.vol_no + 1; + return ARCHIVE_OK; + } + } else { + return ARCHIVE_EOF; + } + case HEAD_MARK: + return ARCHIVE_EOF; + default: + if((header_flags & HFL_SKIP_IF_UNKNOWN) == 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Header type error"); + return ARCHIVE_FATAL; + } else { + /* If the block is marked as 'skip if unknown', + * do as the flag says: skip the block + * instead on failing on it. */ + return ARCHIVE_RETRY; + } + } + +#if !defined WIN32 + // Not reached. + archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, + "Internal unpacker error"); + return ARCHIVE_FATAL; +#endif +} + +static int skip_base_block(struct archive_read* a) { + int ret; + struct rar5* rar = get_context(a); + + /* Create a new local archive_entry structure that will be operated on + * by header reader; operations on this archive_entry will be discarded. + */ + struct archive_entry* entry = archive_entry_new(); + ret = process_base_block(a, entry); + + /* Discard operations on this archive_entry structure. */ + archive_entry_free(entry); + if(ret == ARCHIVE_FATAL) + return ret; + + if(rar->generic.last_header_id == 2 && rar->generic.split_before > 0) + return ARCHIVE_OK; + + if(ret == ARCHIVE_OK) + return ARCHIVE_RETRY; + else + return ret; +} + +static int rar5_read_header(struct archive_read *a, + struct archive_entry *entry) +{ + struct rar5* rar = get_context(a); + int ret; + + if(rar->header_initialized == 0) { + init_header(a); + rar->header_initialized = 1; + } + + if(rar->skipped_magic == 0) { + if(ARCHIVE_OK != consume(a, sizeof(rar5_signature_xor))) { + return ARCHIVE_EOF; + } + + rar->skipped_magic = 1; + } + + do { + ret = process_base_block(a, entry); + } while(ret == ARCHIVE_RETRY || + (rar->main.endarc > 0 && ret == ARCHIVE_OK)); + + return ret; +} + +static void init_unpack(struct rar5* rar) { + rar->file.calculated_crc32 = 0; + init_window_mask(rar); + + free(rar->cstate.window_buf); + free(rar->cstate.filtered_buf); + + if(rar->cstate.window_size > 0) { + rar->cstate.window_buf = calloc(1, rar->cstate.window_size); + rar->cstate.filtered_buf = calloc(1, rar->cstate.window_size); + } else { + rar->cstate.window_buf = NULL; + rar->cstate.filtered_buf = NULL; + } + + rar->cstate.write_ptr = 0; + rar->cstate.last_write_ptr = 0; + + memset(&rar->cstate.bd, 0, sizeof(rar->cstate.bd)); + memset(&rar->cstate.ld, 0, sizeof(rar->cstate.ld)); + memset(&rar->cstate.dd, 0, sizeof(rar->cstate.dd)); + memset(&rar->cstate.ldd, 0, sizeof(rar->cstate.ldd)); + memset(&rar->cstate.rd, 0, sizeof(rar->cstate.rd)); +} + +static void update_crc(struct rar5* rar, const uint8_t* p, size_t to_read) { + int verify_crc; + + if(rar->skip_mode) { +#if defined CHECK_CRC_ON_SOLID_SKIP + verify_crc = 1; +#else + verify_crc = 0; +#endif + } else + verify_crc = 1; + + if(verify_crc) { + /* Don't update CRC32 if the file doesn't have the + * `stored_crc32` info filled in. */ + if(rar->file.stored_crc32 > 0) { + rar->file.calculated_crc32 = + crc32(rar->file.calculated_crc32, p, to_read); + } + + /* Check if the file uses an optional BLAKE2sp checksum + * algorithm. */ + if(rar->file.has_blake2 > 0) { + /* Return value of the `update` function is always 0, + * so we can explicitly ignore it here. */ + (void) blake2sp_update(&rar->file.b2state, p, to_read); + } + } +} + +static int create_decode_tables(uint8_t* bit_length, + struct decode_table* table, int size) +{ + int code, upper_limit = 0, i, lc[16]; + uint32_t decode_pos_clone[rar5_countof(table->decode_pos)]; + ssize_t cur_len, quick_data_size; + + memset(&lc, 0, sizeof(lc)); + memset(table->decode_num, 0, sizeof(table->decode_num)); + table->size = size; + table->quick_bits = size == HUFF_NC ? 10 : 7; + + for(i = 0; i < size; i++) { + lc[bit_length[i] & 15]++; + } + + lc[0] = 0; + table->decode_pos[0] = 0; + table->decode_len[0] = 0; + + for(i = 1; i < 16; i++) { + upper_limit += lc[i]; + + table->decode_len[i] = upper_limit << (16 - i); + table->decode_pos[i] = table->decode_pos[i - 1] + lc[i - 1]; + + upper_limit <<= 1; + } + + memcpy(decode_pos_clone, table->decode_pos, sizeof(decode_pos_clone)); + + for(i = 0; i < size; i++) { + uint8_t clen = bit_length[i] & 15; + if(clen > 0) { + int last_pos = decode_pos_clone[clen]; + table->decode_num[last_pos] = i; + decode_pos_clone[clen]++; + } + } + + quick_data_size = (int64_t)1 << table->quick_bits; + cur_len = 1; + for(code = 0; code < quick_data_size; code++) { + int bit_field = code << (16 - table->quick_bits); + int dist, pos; + + while(cur_len < rar5_countof(table->decode_len) && + bit_field >= table->decode_len[cur_len]) { + cur_len++; + } + + table->quick_len[code] = (uint8_t) cur_len; + + dist = bit_field - table->decode_len[cur_len - 1]; + dist >>= (16 - cur_len); + + pos = table->decode_pos[cur_len & 15] + dist; + if(cur_len < rar5_countof(table->decode_pos) && pos < size) { + table->quick_num[code] = table->decode_num[pos]; + } else { + table->quick_num[code] = 0; + } + } + + return ARCHIVE_OK; +} + +static int decode_number(struct archive_read* a, struct decode_table* table, + const uint8_t* p, uint16_t* num) +{ + int i, bits, dist; + uint16_t bitfield; + uint32_t pos; + struct rar5* rar = get_context(a); + + if(ARCHIVE_OK != read_bits_16(rar, p, &bitfield)) { + return ARCHIVE_EOF; + } + + bitfield &= 0xfffe; + + if(bitfield < table->decode_len[table->quick_bits]) { + int code = bitfield >> (16 - table->quick_bits); + skip_bits(rar, table->quick_len[code]); + *num = table->quick_num[code]; + return ARCHIVE_OK; + } + + bits = 15; + + for(i = table->quick_bits + 1; i < 15; i++) { + if(bitfield < table->decode_len[i]) { + bits = i; + break; + } + } + + skip_bits(rar, bits); + + dist = bitfield - table->decode_len[bits - 1]; + dist >>= (16 - bits); + pos = table->decode_pos[bits] + dist; + + if(pos >= table->size) + pos = 0; + + *num = table->decode_num[pos]; + return ARCHIVE_OK; +} + +/* Reads and parses Huffman tables from the beginning of the block. */ +static int parse_tables(struct archive_read* a, struct rar5* rar, + const uint8_t* p) +{ + int ret, value, i, w, idx = 0; + uint8_t bit_length[HUFF_BC], + table[HUFF_TABLE_SIZE], + nibble_mask = 0xF0, + nibble_shift = 4; + + enum { ESCAPE = 15 }; + + /* The data for table generation is compressed using a simple RLE-like + * algorithm when storing zeroes, so we need to unpack it first. */ + for(w = 0, i = 0; w < HUFF_BC;) { + if(i >= rar->cstate.cur_block_size) { + /* Truncated data, can't continue. */ + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated data in huffman tables"); + return ARCHIVE_FATAL; + } + + value = (p[i] & nibble_mask) >> nibble_shift; + + if(nibble_mask == 0x0F) + ++i; + + nibble_mask ^= 0xFF; + nibble_shift ^= 4; + + /* Values smaller than 15 is data, so we write it directly. + * Value 15 is a flag telling us that we need to unpack more + * bytes. */ + if(value == ESCAPE) { + value = (p[i] & nibble_mask) >> nibble_shift; + if(nibble_mask == 0x0F) + ++i; + nibble_mask ^= 0xFF; + nibble_shift ^= 4; + + if(value == 0) { + /* We sometimes need to write the actual value + * of 15, so this case handles that. */ + bit_length[w++] = ESCAPE; + } else { + int k; + + /* Fill zeroes. */ + for(k = 0; (k < value + 2) && (w < HUFF_BC); + k++) { + bit_length[w++] = 0; + } + } + } else { + bit_length[w++] = value; + } + } + + rar->bits.in_addr = i; + rar->bits.bit_addr = nibble_shift ^ 4; + + ret = create_decode_tables(bit_length, &rar->cstate.bd, HUFF_BC); + if(ret != ARCHIVE_OK) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Decoding huffman tables failed"); + return ARCHIVE_FATAL; + } + + for(i = 0; i < HUFF_TABLE_SIZE;) { + uint16_t num; + + if((rar->bits.in_addr + 6) >= rar->cstate.cur_block_size) { + /* Truncated data, can't continue. */ + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated data in huffman tables (#2)"); + return ARCHIVE_FATAL; + } + + ret = decode_number(a, &rar->cstate.bd, p, &num); + if(ret != ARCHIVE_OK) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Decoding huffman tables failed"); + return ARCHIVE_FATAL; + } + + if(num < 16) { + /* 0..15: store directly */ + table[i] = (uint8_t) num; + i++; + } else if(num < 18) { + /* 16..17: repeat previous code */ + uint16_t n; + + if(ARCHIVE_OK != read_bits_16(rar, p, &n)) + return ARCHIVE_EOF; + + if(num == 16) { + n >>= 13; + n += 3; + skip_bits(rar, 3); + } else { + n >>= 9; + n += 11; + skip_bits(rar, 7); + } + + if(i > 0) { + while(n-- > 0 && i < HUFF_TABLE_SIZE) { + table[i] = table[i - 1]; + i++; + } + } else { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Unexpected error when decoding " + "huffman tables"); + return ARCHIVE_FATAL; + } + } else { + /* other codes: fill with zeroes `n` times */ + uint16_t n; + + if(ARCHIVE_OK != read_bits_16(rar, p, &n)) + return ARCHIVE_EOF; + + if(num == 18) { + n >>= 13; + n += 3; + skip_bits(rar, 3); + } else { + n >>= 9; + n += 11; + skip_bits(rar, 7); + } + + while(n-- > 0 && i < HUFF_TABLE_SIZE) + table[i++] = 0; + } + } + + ret = create_decode_tables(&table[idx], &rar->cstate.ld, HUFF_NC); + if(ret != ARCHIVE_OK) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Failed to create literal table"); + return ARCHIVE_FATAL; + } + + idx += HUFF_NC; + + ret = create_decode_tables(&table[idx], &rar->cstate.dd, HUFF_DC); + if(ret != ARCHIVE_OK) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Failed to create distance table"); + return ARCHIVE_FATAL; + } + + idx += HUFF_DC; + + ret = create_decode_tables(&table[idx], &rar->cstate.ldd, HUFF_LDC); + if(ret != ARCHIVE_OK) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Failed to create lower bits of distances table"); + return ARCHIVE_FATAL; + } + + idx += HUFF_LDC; + + ret = create_decode_tables(&table[idx], &rar->cstate.rd, HUFF_RC); + if(ret != ARCHIVE_OK) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Failed to create repeating distances table"); + return ARCHIVE_FATAL; + } + + return ARCHIVE_OK; +} + +/* Parses the block header, verifies its CRC byte, and saves the header + * fields inside the `hdr` pointer. */ +static int parse_block_header(struct archive_read* a, const uint8_t* p, + ssize_t* block_size, struct compressed_block_header* hdr) +{ + uint8_t calculated_cksum; + memcpy(hdr, p, sizeof(struct compressed_block_header)); + + if(bf_byte_count(hdr) > 2) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Unsupported block header size (was %d, max is 2)", + bf_byte_count(hdr)); + return ARCHIVE_FATAL; + } + + /* This should probably use bit reader interface in order to be more + * future-proof. */ + *block_size = 0; + switch(bf_byte_count(hdr)) { + /* 1-byte block size */ + case 0: + *block_size = *(const uint8_t*) &p[2]; + break; + + /* 2-byte block size */ + case 1: + *block_size = archive_le16dec(&p[2]); + break; + + /* 3-byte block size */ + case 2: + *block_size = archive_le32dec(&p[2]); + *block_size &= 0x00FFFFFF; + break; + + /* Other block sizes are not supported. This case is not + * reached, because we have an 'if' guard before the switch + * that makes sure of it. */ + default: + return ARCHIVE_FATAL; + } + + /* Verify the block header checksum. 0x5A is a magic value and is + * always * constant. */ + calculated_cksum = 0x5A + ^ (uint8_t) hdr->block_flags_u8 + ^ (uint8_t) *block_size + ^ (uint8_t) (*block_size >> 8) + ^ (uint8_t) (*block_size >> 16); + + if(calculated_cksum != hdr->block_cksum) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Block checksum error: got 0x%x, expected 0x%x", + hdr->block_cksum, calculated_cksum); + + return ARCHIVE_FATAL; + } + + return ARCHIVE_OK; +} + +/* Convenience function used during filter processing. */ +static int parse_filter_data(struct rar5* rar, const uint8_t* p, + uint32_t* filter_data) +{ + int i, bytes; + uint32_t data = 0; + + if(ARCHIVE_OK != read_consume_bits(rar, p, 2, &bytes)) + return ARCHIVE_EOF; + + bytes++; + + for(i = 0; i < bytes; i++) { + uint16_t byte; + + if(ARCHIVE_OK != read_bits_16(rar, p, &byte)) { + return ARCHIVE_EOF; + } + + /* Cast to uint32_t will ensure the shift operation will not + * produce undefined result. */ + data += ((uint32_t) byte >> 8) << (i * 8); + skip_bits(rar, 8); + } + + *filter_data = data; + return ARCHIVE_OK; +} + +/* Function is used during sanity checking. */ +static int is_valid_filter_block_start(struct rar5* rar, + uint32_t start) +{ + const int64_t block_start = (ssize_t) start + rar->cstate.write_ptr; + const int64_t last_bs = rar->cstate.last_block_start; + const ssize_t last_bl = rar->cstate.last_block_length; + + if(last_bs == 0 || last_bl == 0) { + /* We didn't have any filters yet, so accept this offset. */ + return 1; + } + + if(block_start >= last_bs + last_bl) { + /* Current offset is bigger than last block's end offset, so + * accept current offset. */ + return 1; + } + + /* Any other case is not a normal situation and we should fail. */ + return 0; +} + +/* The function will create a new filter, read its parameters from the input + * stream and add it to the filter collection. */ +static int parse_filter(struct archive_read* ar, const uint8_t* p) { + uint32_t block_start, block_length; + uint16_t filter_type; + struct filter_info* filt = NULL; + struct rar5* rar = get_context(ar); + + /* Read the parameters from the input stream. */ + if(ARCHIVE_OK != parse_filter_data(rar, p, &block_start)) + return ARCHIVE_EOF; + + if(ARCHIVE_OK != parse_filter_data(rar, p, &block_length)) + return ARCHIVE_EOF; + + if(ARCHIVE_OK != read_bits_16(rar, p, &filter_type)) + return ARCHIVE_EOF; + + filter_type >>= 13; + skip_bits(rar, 3); + + /* Perform some sanity checks on this filter parameters. Note that we + * allow only DELTA, E8/E9 and ARM filters here, because rest of + * filters are not used in RARv5. */ + + if(block_length < 4 || + block_length > 0x400000 || + filter_type > FILTER_ARM || + !is_valid_filter_block_start(rar, block_start)) + { + archive_set_error(&ar->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid filter encountered"); + return ARCHIVE_FATAL; + } + + /* Allocate a new filter. */ + filt = add_new_filter(rar); + if(filt == NULL) { + archive_set_error(&ar->archive, ENOMEM, + "Can't allocate memory for a filter descriptor."); + return ARCHIVE_FATAL; + } + + filt->type = filter_type; + filt->block_start = rar->cstate.write_ptr + block_start; + filt->block_length = block_length; + + rar->cstate.last_block_start = filt->block_start; + rar->cstate.last_block_length = filt->block_length; + + /* Read some more data in case this is a DELTA filter. Other filter + * types don't require any additional data over what was already + * read. */ + if(filter_type == FILTER_DELTA) { + int channels; + + if(ARCHIVE_OK != read_consume_bits(rar, p, 5, &channels)) + return ARCHIVE_EOF; + + filt->channels = channels + 1; + } + + return ARCHIVE_OK; +} + +static int decode_code_length(struct rar5* rar, const uint8_t* p, + uint16_t code) +{ + int lbits, length = 2; + if(code < 8) { + lbits = 0; + length += code; + } else { + lbits = code / 4 - 1; + length += (4 | (code & 3)) << lbits; + } + + if(lbits > 0) { + int add; + + if(ARCHIVE_OK != read_consume_bits(rar, p, lbits, &add)) + return -1; + + length += add; + } + + return length; +} + +static int copy_string(struct archive_read* a, int len, int dist) { + struct rar5* rar = get_context(a); + const uint64_t cmask = rar->cstate.window_mask; + const uint64_t write_ptr = rar->cstate.write_ptr + + rar->cstate.solid_offset; + int i; + + if (rar->cstate.window_buf == NULL) + return ARCHIVE_FATAL; + + /* The unpacker spends most of the time in this function. It would be + * a good idea to introduce some optimizations here. + * + * Just remember that this loop treats buffers that overlap differently + * than buffers that do not overlap. This is why a simple memcpy(3) + * call will not be enough. */ + + for(i = 0; i < len; i++) { + const ssize_t write_idx = (write_ptr + i) & cmask; + const ssize_t read_idx = (write_ptr + i - dist) & cmask; + rar->cstate.window_buf[write_idx] = + rar->cstate.window_buf[read_idx]; + } + + rar->cstate.write_ptr += len; + return ARCHIVE_OK; +} + +static int do_uncompress_block(struct archive_read* a, const uint8_t* p) { + struct rar5* rar = get_context(a); + uint16_t num; + int ret; + + const uint64_t cmask = rar->cstate.window_mask; + const struct compressed_block_header* hdr = &rar->last_block_hdr; + const uint8_t bit_size = 1 + bf_bit_size(hdr); + + while(1) { + if(rar->cstate.write_ptr - rar->cstate.last_write_ptr > + (rar->cstate.window_size >> 1)) { + /* Don't allow growing data by more than half of the + * window size at a time. In such case, break the loop; + * next call to this function will continue processing + * from this moment. */ + break; + } + + if(rar->bits.in_addr > rar->cstate.cur_block_size - 1 || + (rar->bits.in_addr == rar->cstate.cur_block_size - 1 && + rar->bits.bit_addr >= bit_size)) + { + /* If the program counter is here, it means the + * function has finished processing the block. */ + rar->cstate.block_parsing_finished = 1; + break; + } + + /* Decode the next literal. */ + if(ARCHIVE_OK != decode_number(a, &rar->cstate.ld, p, &num)) { + return ARCHIVE_EOF; + } + + /* Num holds a decompression literal, or 'command code'. + * + * - Values lower than 256 are just bytes. Those codes + * can be stored in the output buffer directly. + * + * - Code 256 defines a new filter, which is later used to + * ransform the data block accordingly to the filter type. + * The data block needs to be fully uncompressed first. + * + * - Code bigger than 257 and smaller than 262 define + * a repetition pattern that should be copied from + * an already uncompressed chunk of data. + */ + + if(num < 256) { + /* Directly store the byte. */ + int64_t write_idx = rar->cstate.solid_offset + + rar->cstate.write_ptr++; + + rar->cstate.window_buf[write_idx & cmask] = + (uint8_t) num; + continue; + } else if(num >= 262) { + uint16_t dist_slot; + int len = decode_code_length(rar, p, num - 262), + dbits, + dist = 1; + + if(len == -1) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_PROGRAMMER, + "Failed to decode the code length"); + + return ARCHIVE_FATAL; + } + + if(ARCHIVE_OK != decode_number(a, &rar->cstate.dd, p, + &dist_slot)) + { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_PROGRAMMER, + "Failed to decode the distance slot"); + + return ARCHIVE_FATAL; + } + + if(dist_slot < 4) { + dbits = 0; + dist += dist_slot; + } else { + dbits = dist_slot / 2 - 1; + + /* Cast to uint32_t will make sure the shift + * left operation won't produce undefined + * result. Then, the uint32_t type will + * be implicitly casted to int. */ + dist += (uint32_t) (2 | + (dist_slot & 1)) << dbits; + } + + if(dbits > 0) { + if(dbits >= 4) { + uint32_t add = 0; + uint16_t low_dist; + + if(dbits > 4) { + if(ARCHIVE_OK != read_bits_32( + rar, p, &add)) { + /* Return EOF if we + * can't read more + * data. */ + return ARCHIVE_EOF; + } + + skip_bits(rar, dbits - 4); + add = (add >> ( + 36 - dbits)) << 4; + dist += add; + } + + if(ARCHIVE_OK != decode_number(a, + &rar->cstate.ldd, p, &low_dist)) + { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_PROGRAMMER, + "Failed to decode the " + "distance slot"); + + return ARCHIVE_FATAL; + } + + if(dist >= INT_MAX - low_dist - 1) { + /* This only happens in + * invalid archives. */ + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Distance pointer " + "overflow"); + return ARCHIVE_FATAL; + } + + dist += low_dist; + } else { + /* dbits is one of [0,1,2,3] */ + int add; + + if(ARCHIVE_OK != read_consume_bits(rar, + p, dbits, &add)) { + /* Return EOF if we can't read + * more data. */ + return ARCHIVE_EOF; + } + + dist += add; + } + } + + if(dist > 0x100) { + len++; + + if(dist > 0x2000) { + len++; + + if(dist > 0x40000) { + len++; + } + } + } + + dist_cache_push(rar, dist); + rar->cstate.last_len = len; + + if(ARCHIVE_OK != copy_string(a, len, dist)) + return ARCHIVE_FATAL; + + continue; + } else if(num == 256) { + /* Create a filter. */ + ret = parse_filter(a, p); + if(ret != ARCHIVE_OK) + return ret; + + continue; + } else if(num == 257) { + if(rar->cstate.last_len != 0) { + if(ARCHIVE_OK != copy_string(a, + rar->cstate.last_len, + rar->cstate.dist_cache[0])) + { + return ARCHIVE_FATAL; + } + } + + continue; + } else { + /* num < 262 */ + const int idx = num - 258; + const int dist = dist_cache_touch(rar, idx); + + uint16_t len_slot; + int len; + + if(ARCHIVE_OK != decode_number(a, &rar->cstate.rd, p, + &len_slot)) { + return ARCHIVE_FATAL; + } + + len = decode_code_length(rar, p, len_slot); + rar->cstate.last_len = len; + + if(ARCHIVE_OK != copy_string(a, len, dist)) + return ARCHIVE_FATAL; + + continue; + } + } + + return ARCHIVE_OK; +} + +/* Binary search for the RARv5 signature. */ +static int scan_for_signature(struct archive_read* a) { + const uint8_t* p; + const int chunk_size = 512; + ssize_t i; + char signature[sizeof(rar5_signature_xor)]; + + /* If we're here, it means we're on an 'unknown territory' data. + * There's no indication what kind of data we're reading here. + * It could be some text comment, any kind of binary data, + * digital sign, dragons, etc. + * + * We want to find a valid RARv5 magic header inside this unknown + * data. */ + + /* Is it possible in libarchive to just skip everything until the + * end of the file? If so, it would be a better approach than the + * current implementation of this function. */ + + rar5_signature(signature); + + while(1) { + if(!read_ahead(a, chunk_size, &p)) + return ARCHIVE_EOF; + + for(i = 0; i < chunk_size - (int)sizeof(rar5_signature_xor); + i++) { + if(memcmp(&p[i], signature, + sizeof(rar5_signature_xor)) == 0) { + /* Consume the number of bytes we've used to + * search for the signature, as well as the + * number of bytes used by the signature + * itself. After this we should be standing + * on a valid base block header. */ + (void) consume(a, + i + sizeof(rar5_signature_xor)); + return ARCHIVE_OK; + } + } + + consume(a, chunk_size); + } + + return ARCHIVE_FATAL; +} + +/* This function will switch the multivolume archive file to another file, + * i.e. from part03 to part 04. */ +static int advance_multivolume(struct archive_read* a) { + int lret; + struct rar5* rar = get_context(a); + + /* A small state machine that will skip unnecessary data, needed to + * switch from one multivolume to another. Such skipping is needed if + * we want to be an stream-oriented (instead of file-oriented) + * unpacker. + * + * The state machine starts with `rar->main.endarc` == 0. It also + * assumes that current stream pointer points to some base block + * header. + * + * The `endarc` field is being set when the base block parsing + * function encounters the 'end of archive' marker. + */ + + while(1) { + if(rar->main.endarc == 1) { + int looping = 1; + + rar->main.endarc = 0; + + while(looping) { + lret = skip_base_block(a); + switch(lret) { + case ARCHIVE_RETRY: + /* Continue looping. */ + break; + case ARCHIVE_OK: + /* Break loop. */ + looping = 0; + break; + default: + /* Forward any errors to the + * caller. */ + return lret; + } + } + + break; + } else { + /* Skip current base block. In order to properly skip + * it, we really need to simply parse it and discard + * the results. */ + + lret = skip_base_block(a); + if(lret == ARCHIVE_FATAL || lret == ARCHIVE_FAILED) + return lret; + + /* The `skip_base_block` function tells us if we + * should continue with skipping, or we should stop + * skipping. We're trying to skip everything up to + * a base FILE block. */ + + if(lret != ARCHIVE_RETRY) { + /* If there was an error during skipping, or we + * have just skipped a FILE base block... */ + + if(rar->main.endarc == 0) { + return lret; + } else { + continue; + } + } + } + } + + return ARCHIVE_OK; +} + +/* Merges the partial block from the first multivolume archive file, and + * partial block from the second multivolume archive file. The result is + * a chunk of memory containing the whole block, and the stream pointer + * is advanced to the next block in the second multivolume archive file. */ +static int merge_block(struct archive_read* a, ssize_t block_size, + const uint8_t** p) +{ + struct rar5* rar = get_context(a); + ssize_t cur_block_size, partial_offset = 0; + const uint8_t* lp; + int ret; + + if(rar->merge_mode) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, + "Recursive merge is not allowed"); + + return ARCHIVE_FATAL; + } + + /* Set a flag that we're in the switching mode. */ + rar->cstate.switch_multivolume = 1; + + /* Reallocate the memory which will hold the whole block. */ + if(rar->vol.push_buf) + free((void*) rar->vol.push_buf); + + /* Increasing the allocation block by 8 is due to bit reading functions, + * which are using additional 2 or 4 bytes. Allocating the block size + * by exact value would make bit reader perform reads from invalid + * memory block when reading the last byte from the buffer. */ + rar->vol.push_buf = malloc(block_size + 8); + if(!rar->vol.push_buf) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for a merge block buffer."); + return ARCHIVE_FATAL; + } + + /* Valgrind complains if the extension block for bit reader is not + * initialized, so initialize it. */ + memset(&rar->vol.push_buf[block_size], 0, 8); + + /* A single block can span across multiple multivolume archive files, + * so we use a loop here. This loop will consume enough multivolume + * archive files until the whole block is read. */ + + while(1) { + /* Get the size of current block chunk in this multivolume + * archive file and read it. */ + cur_block_size = rar5_min(rar->file.bytes_remaining, + block_size - partial_offset); + + if(cur_block_size == 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Encountered block size == 0 during block merge"); + return ARCHIVE_FATAL; + } + + if(!read_ahead(a, cur_block_size, &lp)) + return ARCHIVE_EOF; + + /* Sanity check; there should never be a situation where this + * function reads more data than the block's size. */ + if(partial_offset + cur_block_size > block_size) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_PROGRAMMER, + "Consumed too much data when merging blocks."); + return ARCHIVE_FATAL; + } + + /* Merge previous block chunk with current block chunk, + * or create first block chunk if this is our first + * iteration. */ + memcpy(&rar->vol.push_buf[partial_offset], lp, cur_block_size); + + /* Advance the stream read pointer by this block chunk size. */ + if(ARCHIVE_OK != consume(a, cur_block_size)) + return ARCHIVE_EOF; + + /* Update the pointers. `partial_offset` contains information + * about the sum of merged block chunks. */ + partial_offset += cur_block_size; + rar->file.bytes_remaining -= cur_block_size; + + /* If `partial_offset` is the same as `block_size`, this means + * we've merged all block chunks and we have a valid full + * block. */ + if(partial_offset == block_size) { + break; + } + + /* If we don't have any bytes to read, this means we should + * switch to another multivolume archive file. */ + if(rar->file.bytes_remaining == 0) { + rar->merge_mode++; + ret = advance_multivolume(a); + rar->merge_mode--; + if(ret != ARCHIVE_OK) { + return ret; + } + } + } + + *p = rar->vol.push_buf; + + /* If we're here, we can resume unpacking by processing the block + * pointed to by the `*p` memory pointer. */ + + return ARCHIVE_OK; +} + +static int process_block(struct archive_read* a) { + const uint8_t* p; + struct rar5* rar = get_context(a); + int ret; + + /* If we don't have any data to be processed, this most probably means + * we need to switch to the next volume. */ + if(rar->main.volume && rar->file.bytes_remaining == 0) { + ret = advance_multivolume(a); + if(ret != ARCHIVE_OK) + return ret; + } + + if(rar->cstate.block_parsing_finished) { + ssize_t block_size; + ssize_t to_skip; + ssize_t cur_block_size; + + /* The header size won't be bigger than 6 bytes. */ + if(!read_ahead(a, 6, &p)) { + /* Failed to prefetch data block header. */ + return ARCHIVE_EOF; + } + + /* + * Read block_size by parsing block header. Validate the header + * by calculating CRC byte stored inside the header. Size of + * the header is not constant (block size can be stored either + * in 1 or 2 bytes), that's why block size is left out from the + * `compressed_block_header` structure and returned by + * `parse_block_header` as the second argument. */ + + ret = parse_block_header(a, p, &block_size, + &rar->last_block_hdr); + if(ret != ARCHIVE_OK) { + return ret; + } + + /* Skip block header. Next data is huffman tables, + * if present. */ + to_skip = sizeof(struct compressed_block_header) + + bf_byte_count(&rar->last_block_hdr) + 1; + + if(ARCHIVE_OK != consume(a, to_skip)) + return ARCHIVE_EOF; + + rar->file.bytes_remaining -= to_skip; + + /* The block size gives information about the whole block size, + * but the block could be stored in split form when using + * multi-volume archives. In this case, the block size will be + * bigger than the actual data stored in this file. Remaining + * part of the data will be in another file. */ + + cur_block_size = + rar5_min(rar->file.bytes_remaining, block_size); + + if(block_size > rar->file.bytes_remaining) { + /* If current blocks' size is bigger than our data + * size, this means we have a multivolume archive. + * In this case, skip all base headers until the end + * of the file, proceed to next "partXXX.rar" volume, + * find its signature, skip all headers up to the first + * FILE base header, and continue from there. + * + * Note that `merge_block` will update the `rar` + * context structure quite extensively. */ + + ret = merge_block(a, block_size, &p); + if(ret != ARCHIVE_OK) { + return ret; + } + + cur_block_size = block_size; + + /* Current stream pointer should be now directly + * *after* the block that spanned through multiple + * archive files. `p` pointer should have the data of + * the *whole* block (merged from partial blocks + * stored in multiple archives files). */ + } else { + rar->cstate.switch_multivolume = 0; + + /* Read the whole block size into memory. This can take + * up to 8 megabytes of memory in theoretical cases. + * Might be worth to optimize this and use a standard + * chunk of 4kb's. */ + if(!read_ahead(a, 4 + cur_block_size, &p)) { + /* Failed to prefetch block data. */ + return ARCHIVE_EOF; + } + } + + rar->cstate.block_buf = p; + rar->cstate.cur_block_size = cur_block_size; + rar->cstate.block_parsing_finished = 0; + + rar->bits.in_addr = 0; + rar->bits.bit_addr = 0; + + if(bf_is_table_present(&rar->last_block_hdr)) { + /* Load Huffman tables. */ + ret = parse_tables(a, rar, p); + if(ret != ARCHIVE_OK) { + /* Error during decompression of Huffman + * tables. */ + return ret; + } + } + } else { + /* Block parsing not finished, reuse previous memory buffer. */ + p = rar->cstate.block_buf; + } + + /* Uncompress the block, or a part of it, depending on how many bytes + * will be generated by uncompressing the block. + * + * In case too many bytes will be generated, calling this function + * again will resume the uncompression operation. */ + ret = do_uncompress_block(a, p); + if(ret != ARCHIVE_OK) { + return ret; + } + + if(rar->cstate.block_parsing_finished && + rar->cstate.switch_multivolume == 0 && + rar->cstate.cur_block_size > 0) + { + /* If we're processing a normal block, consume the whole + * block. We can do this because we've already read the whole + * block to memory. */ + if(ARCHIVE_OK != consume(a, rar->cstate.cur_block_size)) + return ARCHIVE_FATAL; + + rar->file.bytes_remaining -= rar->cstate.cur_block_size; + } else if(rar->cstate.switch_multivolume) { + /* Don't consume the block if we're doing multivolume + * processing. The volume switching function will consume + * the proper count of bytes instead. */ + rar->cstate.switch_multivolume = 0; + } + + return ARCHIVE_OK; +} + +/* Pops the `buf`, `size` and `offset` from the "data ready" stack. + * + * Returns ARCHIVE_OK when those arguments can be used, ARCHIVE_RETRY + * when there is no data on the stack. */ +static int use_data(struct rar5* rar, const void** buf, size_t* size, + int64_t* offset) +{ + int i; + + for(i = 0; i < rar5_countof(rar->cstate.dready); i++) { + struct data_ready *d = &rar->cstate.dready[i]; + + if(d->used) { + if(buf) *buf = d->buf; + if(size) *size = d->size; + if(offset) *offset = d->offset; + + d->used = 0; + return ARCHIVE_OK; + } + } + + return ARCHIVE_RETRY; +} + +/* Pushes the `buf`, `size` and `offset` arguments to the rar->cstate.dready + * FIFO stack. Those values will be popped from this stack by the `use_data` + * function. */ +static int push_data_ready(struct archive_read* a, struct rar5* rar, + const uint8_t* buf, size_t size, int64_t offset) +{ + int i; + + /* Don't push if we're in skip mode. This is needed because solid + * streams need full processing even if we're skipping data. After + * fully processing the stream, we need to discard the generated bytes, + * because we're interested only in the side effect: building up the + * internal window circular buffer. This window buffer will be used + * later during unpacking of requested data. */ + if(rar->skip_mode) + return ARCHIVE_OK; + + /* Sanity check. */ + if(offset != rar->file.last_offset + rar->file.last_size) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, + "Sanity check error: output stream is not continuous"); + return ARCHIVE_FATAL; + } + + for(i = 0; i < rar5_countof(rar->cstate.dready); i++) { + struct data_ready* d = &rar->cstate.dready[i]; + if(!d->used) { + d->used = 1; + d->buf = buf; + d->size = size; + d->offset = offset; + + /* These fields are used only in sanity checking. */ + rar->file.last_offset = offset; + rar->file.last_size = size; + + /* Calculate the checksum of this new block before + * submitting data to libarchive's engine. */ + update_crc(rar, d->buf, d->size); + + return ARCHIVE_OK; + } + } + + /* Program counter will reach this code if the `rar->cstate.data_ready` + * stack will be filled up so that no new entries will be allowed. The + * code shouldn't allow such situation to occur. So we treat this case + * as an internal error. */ + + archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, + "Error: premature end of data_ready stack"); + return ARCHIVE_FATAL; +} + +/* This function uncompresses the data that is stored in the base + * block. + * + * The FILE base block looks like this: + * + *
... + * + * The
is a block header, that is parsed in parse_block_header(). + * It's a "compressed_block_header" structure, containing metadata needed + * to know when we should stop looking for more blocks. + * + * contain data needed to set up the huffman tables, needed + * for the actual decompression. + * + * Each consists of series of literals: + * + * ... + * + * Those literals generate the uncompression data. They operate on a circular + * buffer, sometimes writing raw data into it, sometimes referencing + * some previous data inside this buffer, and sometimes declaring a filter + * that will need to be executed on the data stored in the circular buffer. + * It all depends on the literal that is used. + * + * Sometimes blocks produce output data, sometimes they don't. For example, for + * some huge files that use lots of filters, sometimes a block is filled with + * only filter declaration literals. Such blocks won't produce any data in the + * circular buffer. + * + * Sometimes blocks will produce 4 bytes of data, and sometimes 1 megabyte, + * because a literal can reference previously decompressed data. For example, + * there can be a literal that says: 'append a byte 0xFE here', and after + * it another literal can say 'append 1 megabyte of data from circular buffer + * offset 0x12345'. This is how RAR format handles compressing repeated + * patterns. + * + * The RAR compressor creates those literals and the actual efficiency of + * compression depends on what those literals are. The literals can also + * be seen as a kind of a non-turing-complete virtual machine that simply + * tells the decompressor what it should do. + * */ + +static int do_uncompress_file(struct archive_read* a) { + struct rar5* rar = get_context(a); + int ret; + int64_t max_end_pos; + + if(!rar->cstate.initialized) { + /* Don't perform full context reinitialization if we're + * processing a solid archive. */ + if(!rar->main.solid || !rar->cstate.window_buf) { + init_unpack(rar); + } + + rar->cstate.initialized = 1; + } + + if(rar->cstate.all_filters_applied == 1) { + /* We use while(1) here, but standard case allows for just 1 + * iteration. The loop will iterate if process_block() didn't + * generate any data at all. This can happen if the block + * contains only filter definitions (this is common in big + * files). */ + while(1) { + ret = process_block(a); + if(ret == ARCHIVE_EOF || ret == ARCHIVE_FATAL) + return ret; + + if(rar->cstate.last_write_ptr == + rar->cstate.write_ptr) { + /* The block didn't generate any new data, + * so just process a new block. */ + continue; + } + + /* The block has generated some new data, so break + * the loop. */ + break; + } + } + + /* Try to run filters. If filters won't be applied, it means that + * insufficient data was generated. */ + ret = apply_filters(a); + if(ret == ARCHIVE_RETRY) { + return ARCHIVE_OK; + } else if(ret == ARCHIVE_FATAL) { + return ARCHIVE_FATAL; + } + + /* If apply_filters() will return ARCHIVE_OK, we can continue here. */ + + if(cdeque_size(&rar->cstate.filters) > 0) { + /* Check if we can write something before hitting first + * filter. */ + struct filter_info* flt; + + /* Get the block_start offset from the first filter. */ + if(CDE_OK != cdeque_front(&rar->cstate.filters, + cdeque_filter_p(&flt))) + { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_PROGRAMMER, + "Can't read first filter"); + return ARCHIVE_FATAL; + } + + max_end_pos = rar5_min(flt->block_start, + rar->cstate.write_ptr); + } else { + /* There are no filters defined, or all filters were applied. + * This means we can just store the data without any + * postprocessing. */ + max_end_pos = rar->cstate.write_ptr; + } + + if(max_end_pos == rar->cstate.last_write_ptr) { + /* We can't write anything yet. The block uncompression + * function did not generate enough data, and no filter can be + * applied. At the same time we don't have any data that can be + * stored without filter postprocessing. This means we need to + * wait for more data to be generated, so we can apply the + * filters. + * + * Signal the caller that we need more data to be able to do + * anything. + */ + return ARCHIVE_RETRY; + } else { + /* We can write the data before hitting the first filter. + * So let's do it. The push_window_data() function will + * effectively return the selected data block to the user + * application. */ + push_window_data(a, rar, rar->cstate.last_write_ptr, + max_end_pos); + rar->cstate.last_write_ptr = max_end_pos; + } + + return ARCHIVE_OK; +} + +static int uncompress_file(struct archive_read* a) { + int ret; + + while(1) { + /* Sometimes the uncompression function will return a + * 'retry' signal. If this will happen, we have to retry + * the function. */ + ret = do_uncompress_file(a); + if(ret != ARCHIVE_RETRY) + return ret; + } +} + + +static int do_unstore_file(struct archive_read* a, + struct rar5* rar, const void** buf, size_t* size, int64_t* offset) +{ + size_t to_read; + const uint8_t* p; + + if(rar->file.bytes_remaining == 0 && rar->main.volume > 0 && + rar->generic.split_after > 0) + { + int ret; + + rar->cstate.switch_multivolume = 1; + ret = advance_multivolume(a); + rar->cstate.switch_multivolume = 0; + + if(ret != ARCHIVE_OK) { + /* Failed to advance to next multivolume archive + * file. */ + return ret; + } + } + + to_read = rar5_min(rar->file.bytes_remaining, 64 * 1024); + if(to_read == 0) { + return ARCHIVE_EOF; + } + + if(!read_ahead(a, to_read, &p)) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "I/O error when unstoring file"); + return ARCHIVE_FATAL; + } + + if(ARCHIVE_OK != consume(a, to_read)) { + return ARCHIVE_EOF; + } + + if(buf) *buf = p; + if(size) *size = to_read; + if(offset) *offset = rar->cstate.last_unstore_ptr; + + rar->file.bytes_remaining -= to_read; + rar->cstate.last_unstore_ptr += to_read; + + update_crc(rar, p, to_read); + return ARCHIVE_OK; +} + +static int do_unpack(struct archive_read* a, struct rar5* rar, + const void** buf, size_t* size, int64_t* offset) +{ + enum COMPRESSION_METHOD { + STORE = 0, FASTEST = 1, FAST = 2, NORMAL = 3, GOOD = 4, + BEST = 5 + }; + + if(rar->file.service > 0) { + return do_unstore_file(a, rar, buf, size, offset); + } else { + switch(rar->cstate.method) { + case STORE: + return do_unstore_file(a, rar, buf, size, + offset); + case FASTEST: + /* fallthrough */ + case FAST: + /* fallthrough */ + case NORMAL: + /* fallthrough */ + case GOOD: + /* fallthrough */ + case BEST: + return uncompress_file(a); + default: + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Compression method not supported: 0x%x", + rar->cstate.method); + + return ARCHIVE_FATAL; + } + } + +#if !defined WIN32 + /* Not reached. */ + return ARCHIVE_OK; +#endif +} + +static int verify_checksums(struct archive_read* a) { + int verify_crc; + struct rar5* rar = get_context(a); + + /* Check checksums only when actually unpacking the data. There's no + * need to calculate checksum when we're skipping data in solid archives + * (skipping in solid archives is the same thing as unpacking compressed + * data and discarding the result). */ + + if(!rar->skip_mode) { + /* Always check checksums if we're not in skip mode */ + verify_crc = 1; + } else { + /* We can override the logic above with a compile-time option + * NO_CRC_ON_SOLID_SKIP. This option is used during debugging, + * and it will check checksums of unpacked data even when + * we're skipping it. */ + +#if defined CHECK_CRC_ON_SOLID_SKIP + /* Debug case */ + verify_crc = 1; +#else + /* Normal case */ + verify_crc = 0; +#endif + } + + if(verify_crc) { + /* During unpacking, on each unpacked block we're calling the + * update_crc() function. Since we are here, the unpacking + * process is already over and we can check if calculated + * checksum (CRC32 or BLAKE2sp) is the same as what is stored + * in the archive. */ + if(rar->file.stored_crc32 > 0) { + /* Check CRC32 only when the file contains a CRC32 + * value for this file. */ + + if(rar->file.calculated_crc32 != + rar->file.stored_crc32) { + /* Checksums do not match; the unpacked file + * is corrupted. */ + + DEBUG_CODE { + printf("Checksum error: CRC32 " + "(was: %08x, expected: %08x)\n", + rar->file.calculated_crc32, + rar->file.stored_crc32); + } + +#ifndef DONT_FAIL_ON_CRC_ERROR + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Checksum error: CRC32"); + return ARCHIVE_FATAL; +#endif + } else { + DEBUG_CODE { + printf("Checksum OK: CRC32 " + "(%08x/%08x)\n", + rar->file.stored_crc32, + rar->file.calculated_crc32); + } + } + } + + if(rar->file.has_blake2 > 0) { + /* BLAKE2sp is an optional checksum algorithm that is + * added to RARv5 archives when using the `-htb` switch + * during creation of archive. + * + * We now finalize the hash calculation by calling the + * `final` function. This will generate the final hash + * value we can use to compare it with the BLAKE2sp + * checksum that is stored in the archive. + * + * The return value of this `final` function is not + * very helpful, as it guards only against improper use. + * This is why we're explicitly ignoring it. */ + + uint8_t b2_buf[32]; + (void) blake2sp_final(&rar->file.b2state, b2_buf, 32); + + if(memcmp(&rar->file.blake2sp, b2_buf, 32) != 0) { +#ifndef DONT_FAIL_ON_CRC_ERROR + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Checksum error: BLAKE2"); + + return ARCHIVE_FATAL; +#endif + } + } + } + + /* Finalization for this file has been successfully completed. */ + return ARCHIVE_OK; +} + +static int verify_global_checksums(struct archive_read* a) { + return verify_checksums(a); +} + +/* + * Decryption function for the magic signature pattern. Check the comment near + * the `rar5_signature_xor` symbol to read the rationale behind this. + */ +static void rar5_signature(char *buf) { + size_t i; + + for(i = 0; i < sizeof(rar5_signature_xor); i++) { + buf[i] = rar5_signature_xor[i] ^ 0xA1; + } +} + +static int rar5_read_data(struct archive_read *a, const void **buff, + size_t *size, int64_t *offset) { + int ret; + struct rar5* rar = get_context(a); + + if(rar->file.dir > 0) { + /* Don't process any data if this file entry was declared + * as a directory. This is needed, because entries marked as + * directory doesn't have any dictionary buffer allocated, so + * it's impossible to perform any decompression. */ + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Can't decompress an entry marked as a directory"); + return ARCHIVE_FAILED; + } + + if(!rar->skip_mode && (rar->cstate.last_write_ptr > rar->file.unpacked_size)) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, + "Unpacker has written too many bytes"); + return ARCHIVE_FATAL; + } + + ret = use_data(rar, buff, size, offset); + if(ret == ARCHIVE_OK) { + return ret; + } + + if(rar->file.eof == 1) { + return ARCHIVE_EOF; + } + + ret = do_unpack(a, rar, buff, size, offset); + if(ret != ARCHIVE_OK) { + return ret; + } + + if(rar->file.bytes_remaining == 0 && + rar->cstate.last_write_ptr == rar->file.unpacked_size) + { + /* If all bytes of current file were processed, run + * finalization. + * + * Finalization will check checksum against proper values. If + * some of the checksums will not match, we'll return an error + * value in the last `archive_read_data` call to signal an error + * to the user. */ + + rar->file.eof = 1; + return verify_global_checksums(a); + } + + return ARCHIVE_OK; +} + +static int rar5_read_data_skip(struct archive_read *a) { + struct rar5* rar = get_context(a); + + if(rar->main.solid) { + /* In solid archives, instead of skipping the data, we need to + * extract it, and dispose the result. The side effect of this + * operation will be setting up the initial window buffer state + * needed to be able to extract the selected file. */ + + int ret; + + /* Make sure to process all blocks in the compressed stream. */ + while(rar->file.bytes_remaining > 0) { + /* Setting the "skip mode" will allow us to skip + * checksum checks during data skipping. Checking the + * checksum of skipped data isn't really necessary and + * it's only slowing things down. + * + * This is incremented instead of setting to 1 because + * this data skipping function can be called + * recursively. */ + rar->skip_mode++; + + /* We're disposing 1 block of data, so we use triple + * NULLs in arguments. */ + ret = rar5_read_data(a, NULL, NULL, NULL); + + /* Turn off "skip mode". */ + rar->skip_mode--; + + if(ret < 0 || ret == ARCHIVE_EOF) { + /* Propagate any potential error conditions + * to the caller. */ + return ret; + } + } + } else { + /* In standard archives, we can just jump over the compressed + * stream. Each file in non-solid archives starts from an empty + * window buffer. */ + + if(ARCHIVE_OK != consume(a, rar->file.bytes_remaining)) { + return ARCHIVE_FATAL; + } + + rar->file.bytes_remaining = 0; + } + + return ARCHIVE_OK; +} + +static int64_t rar5_seek_data(struct archive_read *a, int64_t offset, + int whence) +{ + (void) a; + (void) offset; + (void) whence; + + /* We're a streaming unpacker, and we don't support seeking. */ + + return ARCHIVE_FATAL; +} + +static int rar5_cleanup(struct archive_read *a) { + struct rar5* rar = get_context(a); + + free(rar->cstate.window_buf); + free(rar->cstate.filtered_buf); + + free(rar->vol.push_buf); + + free_filters(rar); + cdeque_free(&rar->cstate.filters); + + free(rar); + a->format->data = NULL; + + return ARCHIVE_OK; +} + +static int rar5_capabilities(struct archive_read * a) { + (void) a; + return 0; +} + +static int rar5_has_encrypted_entries(struct archive_read *_a) { + (void) _a; + + /* Unsupported for now. */ + return ARCHIVE_READ_FORMAT_ENCRYPTION_UNSUPPORTED; +} + +static int rar5_init(struct rar5* rar) { + memset(rar, 0, sizeof(struct rar5)); + + if(CDE_OK != cdeque_init(&rar->cstate.filters, 8192)) + return ARCHIVE_FATAL; + + return ARCHIVE_OK; +} + +int archive_read_support_format_rar5(struct archive *_a) { + struct archive_read* ar; + int ret; + struct rar5* rar; + + if(ARCHIVE_OK != (ret = get_archive_read(_a, &ar))) + return ret; + + rar = malloc(sizeof(*rar)); + if(rar == NULL) { + archive_set_error(&ar->archive, ENOMEM, + "Can't allocate rar5 data"); + return ARCHIVE_FATAL; + } + + if(ARCHIVE_OK != rar5_init(rar)) { + archive_set_error(&ar->archive, ENOMEM, + "Can't allocate rar5 filter buffer"); + return ARCHIVE_FATAL; + } + + ret = __archive_read_register_format(ar, + rar, + "rar5", + rar5_bid, + rar5_options, + rar5_read_header, + rar5_read_data, + rar5_read_data_skip, + rar5_seek_data, + rar5_cleanup, + rar5_capabilities, + rar5_has_encrypted_entries); + + if(ret != ARCHIVE_OK) { + (void) rar5_cleanup(ar); + } + + return ret; +} diff --git a/c/archive_read_support_format_raw.c b/c/archive_read_support_format_raw.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_format_raw.c @@ -0,0 +1,192 @@ +/*- + * Copyright (c) 2003-2009 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_read_support_format_raw.c 201107 2009-12-28 03:25:33Z kientzle $"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#include +#ifdef HAVE_STDLIB_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_read_private.h" + +struct raw_info { + int64_t offset; /* Current position in the file. */ + int64_t unconsumed; + int end_of_file; +}; + +static int archive_read_format_raw_bid(struct archive_read *, int); +static int archive_read_format_raw_cleanup(struct archive_read *); +static int archive_read_format_raw_read_data(struct archive_read *, + const void **, size_t *, int64_t *); +static int archive_read_format_raw_read_data_skip(struct archive_read *); +static int archive_read_format_raw_read_header(struct archive_read *, + struct archive_entry *); + +int +archive_read_support_format_raw(struct archive *_a) +{ + struct raw_info *info; + struct archive_read *a = (struct archive_read *)_a; + int r; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_format_raw"); + + info = (struct raw_info *)calloc(1, sizeof(*info)); + if (info == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate raw_info data"); + return (ARCHIVE_FATAL); + } + + r = __archive_read_register_format(a, + info, + "raw", + archive_read_format_raw_bid, + NULL, + archive_read_format_raw_read_header, + archive_read_format_raw_read_data, + archive_read_format_raw_read_data_skip, + NULL, + archive_read_format_raw_cleanup, + NULL, + NULL); + if (r != ARCHIVE_OK) + free(info); + return (r); +} + +/* + * Bid 1 if this is a non-empty file. Anyone who can really support + * this should outbid us, so it should generally be safe to use "raw" + * in conjunction with other formats. But, this could really confuse + * folks if there are bid errors or minor file damage, so we don't + * include "raw" as part of support_format_all(). + */ +static int +archive_read_format_raw_bid(struct archive_read *a, int best_bid) +{ + if (best_bid < 1 && __archive_read_ahead(a, 1, NULL) != NULL) + return (1); + return (-1); +} + +/* + * Mock up a fake header. + */ +static int +archive_read_format_raw_read_header(struct archive_read *a, + struct archive_entry *entry) +{ + struct raw_info *info; + + info = (struct raw_info *)(a->format->data); + if (info->end_of_file) + return (ARCHIVE_EOF); + + a->archive.archive_format = ARCHIVE_FORMAT_RAW; + a->archive.archive_format_name = "raw"; + archive_entry_set_pathname(entry, "data"); + archive_entry_set_filetype(entry, AE_IFREG); + archive_entry_set_perm(entry, 0644); + /* I'm deliberately leaving most fields unset here. */ + + /* Let the filter fill out any fields it might have. */ + return __archive_read_header(a, entry); +} + +static int +archive_read_format_raw_read_data(struct archive_read *a, + const void **buff, size_t *size, int64_t *offset) +{ + struct raw_info *info; + ssize_t avail; + + info = (struct raw_info *)(a->format->data); + + /* Consume the bytes we read last time. */ + if (info->unconsumed) { + __archive_read_consume(a, info->unconsumed); + info->unconsumed = 0; + } + + if (info->end_of_file) + return (ARCHIVE_EOF); + + /* Get whatever bytes are immediately available. */ + *buff = __archive_read_ahead(a, 1, &avail); + if (avail > 0) { + /* Return the bytes we just read */ + *size = avail; + *offset = info->offset; + info->offset += *size; + info->unconsumed = avail; + return (ARCHIVE_OK); + } else if (0 == avail) { + /* Record and return end-of-file. */ + info->end_of_file = 1; + *size = 0; + *offset = info->offset; + return (ARCHIVE_EOF); + } else { + /* Record and return an error. */ + *size = 0; + *offset = info->offset; + return ((int)avail); + } +} + +static int +archive_read_format_raw_read_data_skip(struct archive_read *a) +{ + struct raw_info *info = (struct raw_info *)(a->format->data); + + /* Consume the bytes we read last time. */ + if (info->unconsumed) { + __archive_read_consume(a, info->unconsumed); + info->unconsumed = 0; + } + info->end_of_file = 1; + return (ARCHIVE_OK); +} + +static int +archive_read_format_raw_cleanup(struct archive_read *a) +{ + struct raw_info *info; + + info = (struct raw_info *)(a->format->data); + free(info); + a->format->data = NULL; + return (ARCHIVE_OK); +} diff --git a/c/archive_read_support_format_tar.c b/c/archive_read_support_format_tar.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_format_tar.c @@ -0,0 +1,2922 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * Copyright (c) 2011-2012 Michihiro NAKAJIMA + * Copyright (c) 2016 Martin Matuska + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_read_support_format_tar.c 201161 2009-12-29 05:44:39Z kientzle $"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_acl_private.h" /* For ACL parsing routines. */ +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_private.h" +#include "archive_read_private.h" + +#define tar_min(a,b) ((a) < (b) ? (a) : (b)) + +/* + * Layout of POSIX 'ustar' tar header. + */ +struct archive_entry_header_ustar { + char name[100]; + char mode[8]; + char uid[8]; + char gid[8]; + char size[12]; + char mtime[12]; + char checksum[8]; + char typeflag[1]; + char linkname[100]; /* "old format" header ends here */ + char magic[6]; /* For POSIX: "ustar\0" */ + char version[2]; /* For POSIX: "00" */ + char uname[32]; + char gname[32]; + char rdevmajor[8]; + char rdevminor[8]; + char prefix[155]; +}; + +/* + * Structure of GNU tar header + */ +struct gnu_sparse { + char offset[12]; + char numbytes[12]; +}; + +struct archive_entry_header_gnutar { + char name[100]; + char mode[8]; + char uid[8]; + char gid[8]; + char size[12]; + char mtime[12]; + char checksum[8]; + char typeflag[1]; + char linkname[100]; + char magic[8]; /* "ustar \0" (note blank/blank/null at end) */ + char uname[32]; + char gname[32]; + char rdevmajor[8]; + char rdevminor[8]; + char atime[12]; + char ctime[12]; + char offset[12]; + char longnames[4]; + char unused[1]; + struct gnu_sparse sparse[4]; + char isextended[1]; + char realsize[12]; + /* + * Old GNU format doesn't use POSIX 'prefix' field; they use + * the 'L' (longname) entry instead. + */ +}; + +/* + * Data specific to this format. + */ +struct sparse_block { + struct sparse_block *next; + int64_t offset; + int64_t remaining; + int hole; +}; + +struct tar { + struct archive_string acl_text; + struct archive_string entry_pathname; + /* For "GNU.sparse.name" and other similar path extensions. */ + struct archive_string entry_pathname_override; + struct archive_string entry_linkpath; + struct archive_string entry_uname; + struct archive_string entry_gname; + struct archive_string longlink; + struct archive_string longname; + struct archive_string pax_header; + struct archive_string pax_global; + struct archive_string line; + int pax_hdrcharset_binary; + int header_recursion_depth; + int64_t entry_bytes_remaining; + int64_t entry_offset; + int64_t entry_padding; + int64_t entry_bytes_unconsumed; + int64_t realsize; + int sparse_allowed; + struct sparse_block *sparse_list; + struct sparse_block *sparse_last; + int64_t sparse_offset; + int64_t sparse_numbytes; + int sparse_gnu_major; + int sparse_gnu_minor; + char sparse_gnu_pending; + + struct archive_string localname; + struct archive_string_conv *opt_sconv; + struct archive_string_conv *sconv; + struct archive_string_conv *sconv_acl; + struct archive_string_conv *sconv_default; + int init_default_conversion; + int compat_2x; + int process_mac_extensions; + int read_concatenated_archives; + int realsize_override; +}; + +static int archive_block_is_null(const char *p); +static char *base64_decode(const char *, size_t, size_t *); +static int gnu_add_sparse_entry(struct archive_read *, struct tar *, + int64_t offset, int64_t remaining); + +static void gnu_clear_sparse_list(struct tar *); +static int gnu_sparse_old_read(struct archive_read *, struct tar *, + const struct archive_entry_header_gnutar *header, size_t *); +static int gnu_sparse_old_parse(struct archive_read *, struct tar *, + const struct gnu_sparse *sparse, int length); +static int gnu_sparse_01_parse(struct archive_read *, struct tar *, + const char *); +static ssize_t gnu_sparse_10_read(struct archive_read *, struct tar *, + size_t *); +static int header_Solaris_ACL(struct archive_read *, struct tar *, + struct archive_entry *, const void *, size_t *); +static int header_common(struct archive_read *, struct tar *, + struct archive_entry *, const void *); +static int header_old_tar(struct archive_read *, struct tar *, + struct archive_entry *, const void *); +static int header_pax_extensions(struct archive_read *, struct tar *, + struct archive_entry *, const void *, size_t *); +static int header_pax_global(struct archive_read *, struct tar *, + struct archive_entry *, const void *h, size_t *); +static int header_longlink(struct archive_read *, struct tar *, + struct archive_entry *, const void *h, size_t *); +static int header_longname(struct archive_read *, struct tar *, + struct archive_entry *, const void *h, size_t *); +static int read_mac_metadata_blob(struct archive_read *, struct tar *, + struct archive_entry *, const void *h, size_t *); +static int header_volume(struct archive_read *, struct tar *, + struct archive_entry *, const void *h, size_t *); +static int header_ustar(struct archive_read *, struct tar *, + struct archive_entry *, const void *h); +static int header_gnutar(struct archive_read *, struct tar *, + struct archive_entry *, const void *h, size_t *); +static int archive_read_format_tar_bid(struct archive_read *, int); +static int archive_read_format_tar_options(struct archive_read *, + const char *, const char *); +static int archive_read_format_tar_cleanup(struct archive_read *); +static int archive_read_format_tar_read_data(struct archive_read *a, + const void **buff, size_t *size, int64_t *offset); +static int archive_read_format_tar_skip(struct archive_read *a); +static int archive_read_format_tar_read_header(struct archive_read *, + struct archive_entry *); +static int checksum(struct archive_read *, const void *); +static int pax_attribute(struct archive_read *, struct tar *, + struct archive_entry *, const char *key, const char *value, + size_t value_length); +static int pax_attribute_acl(struct archive_read *, struct tar *, + struct archive_entry *, const char *, int); +static int pax_attribute_xattr(struct archive_entry *, const char *, + const char *); +static int pax_header(struct archive_read *, struct tar *, + struct archive_entry *, struct archive_string *); +static void pax_time(const char *, int64_t *sec, long *nanos); +static ssize_t readline(struct archive_read *, struct tar *, const char **, + ssize_t limit, size_t *); +static int read_body_to_string(struct archive_read *, struct tar *, + struct archive_string *, const void *h, size_t *); +static int solaris_sparse_parse(struct archive_read *, struct tar *, + struct archive_entry *, const char *); +static int64_t tar_atol(const char *, size_t); +static int64_t tar_atol10(const char *, size_t); +static int64_t tar_atol256(const char *, size_t); +static int64_t tar_atol8(const char *, size_t); +static int tar_read_header(struct archive_read *, struct tar *, + struct archive_entry *, size_t *); +static int tohex(int c); +static char *url_decode(const char *); +static void tar_flush_unconsumed(struct archive_read *, size_t *); + + +int +archive_read_support_format_gnutar(struct archive *a) +{ + archive_check_magic(a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_format_gnutar"); + return (archive_read_support_format_tar(a)); +} + + +int +archive_read_support_format_tar(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct tar *tar; + int r; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_format_tar"); + + tar = (struct tar *)calloc(1, sizeof(*tar)); + if (tar == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate tar data"); + return (ARCHIVE_FATAL); + } +#ifdef HAVE_COPYFILE_H + /* Set this by default on Mac OS. */ + tar->process_mac_extensions = 1; +#endif + + r = __archive_read_register_format(a, tar, "tar", + archive_read_format_tar_bid, + archive_read_format_tar_options, + archive_read_format_tar_read_header, + archive_read_format_tar_read_data, + archive_read_format_tar_skip, + NULL, + archive_read_format_tar_cleanup, + NULL, + NULL); + + if (r != ARCHIVE_OK) + free(tar); + return (ARCHIVE_OK); +} + +static int +archive_read_format_tar_cleanup(struct archive_read *a) +{ + struct tar *tar; + + tar = (struct tar *)(a->format->data); + gnu_clear_sparse_list(tar); + archive_string_free(&tar->acl_text); + archive_string_free(&tar->entry_pathname); + archive_string_free(&tar->entry_pathname_override); + archive_string_free(&tar->entry_linkpath); + archive_string_free(&tar->entry_uname); + archive_string_free(&tar->entry_gname); + archive_string_free(&tar->line); + archive_string_free(&tar->pax_global); + archive_string_free(&tar->pax_header); + archive_string_free(&tar->longname); + archive_string_free(&tar->longlink); + archive_string_free(&tar->localname); + free(tar); + (a->format->data) = NULL; + return (ARCHIVE_OK); +} + +/* + * Validate number field + * + * This has to be pretty lenient in order to accommodate the enormous + * variety of tar writers in the world: + * = POSIX (IEEE Std 1003.1-1988) ustar requires octal values with leading + * zeros and allows fields to be terminated with space or null characters + * = Many writers use different termination (in particular, libarchive + * omits terminator bytes to squeeze one or two more digits) + * = Many writers pad with space and omit leading zeros + * = GNU tar and star write base-256 values if numbers are too + * big to be represented in octal + * + * Examples of specific tar headers that we should support: + * = Perl Archive::Tar terminates uid, gid, devminor and devmajor with two + * null bytes, pads size with spaces and other numeric fields with zeroes + * = plexus-archiver prior to 2.6.3 (before switching to commons-compress) + * may have uid and gid fields filled with spaces without any octal digits + * at all and pads all numeric fields with spaces + * + * This should tolerate all variants in use. It will reject a field + * where the writer just left garbage after a trailing NUL. + */ +static int +validate_number_field(const char* p_field, size_t i_size) +{ + unsigned char marker = (unsigned char)p_field[0]; + if (marker == 128 || marker == 255 || marker == 0) { + /* Base-256 marker, there's nothing we can check. */ + return 1; + } else { + /* Must be octal */ + size_t i = 0; + /* Skip any leading spaces */ + while (i < i_size && p_field[i] == ' ') { + ++i; + } + /* Skip octal digits. */ + while (i < i_size && p_field[i] >= '0' && p_field[i] <= '7') { + ++i; + } + /* Any remaining characters must be space or NUL padding. */ + while (i < i_size) { + if (p_field[i] != ' ' && p_field[i] != 0) { + return 0; + } + ++i; + } + return 1; + } +} + +static int +archive_read_format_tar_bid(struct archive_read *a, int best_bid) +{ + int bid; + const char *h; + const struct archive_entry_header_ustar *header; + + (void)best_bid; /* UNUSED */ + + bid = 0; + + /* Now let's look at the actual header and see if it matches. */ + h = __archive_read_ahead(a, 512, NULL); + if (h == NULL) + return (-1); + + /* If it's an end-of-archive mark, we can handle it. */ + if (h[0] == 0 && archive_block_is_null(h)) { + /* + * Usually, I bid the number of bits verified, but + * in this case, 4096 seems excessive so I picked 10 as + * an arbitrary but reasonable-seeming value. + */ + return (10); + } + + /* If it's not an end-of-archive mark, it must have a valid checksum.*/ + if (!checksum(a, h)) + return (0); + bid += 48; /* Checksum is usually 6 octal digits. */ + + header = (const struct archive_entry_header_ustar *)h; + + /* Recognize POSIX formats. */ + if ((memcmp(header->magic, "ustar\0", 6) == 0) + && (memcmp(header->version, "00", 2) == 0)) + bid += 56; + + /* Recognize GNU tar format. */ + if ((memcmp(header->magic, "ustar ", 6) == 0) + && (memcmp(header->version, " \0", 2) == 0)) + bid += 56; + + /* Type flag must be null, digit or A-Z, a-z. */ + if (header->typeflag[0] != 0 && + !( header->typeflag[0] >= '0' && header->typeflag[0] <= '9') && + !( header->typeflag[0] >= 'A' && header->typeflag[0] <= 'Z') && + !( header->typeflag[0] >= 'a' && header->typeflag[0] <= 'z') ) + return (0); + bid += 2; /* 6 bits of variation in an 8-bit field leaves 2 bits. */ + + /* + * Check format of mode/uid/gid/mtime/size/rdevmajor/rdevminor fields. + */ + if (bid > 0 && ( + validate_number_field(header->mode, sizeof(header->mode)) == 0 + || validate_number_field(header->uid, sizeof(header->uid)) == 0 + || validate_number_field(header->gid, sizeof(header->gid)) == 0 + || validate_number_field(header->mtime, sizeof(header->mtime)) == 0 + || validate_number_field(header->size, sizeof(header->size)) == 0 + || validate_number_field(header->rdevmajor, sizeof(header->rdevmajor)) == 0 + || validate_number_field(header->rdevminor, sizeof(header->rdevminor)) == 0)) { + bid = 0; + } + + return (bid); +} + +static int +archive_read_format_tar_options(struct archive_read *a, + const char *key, const char *val) +{ + struct tar *tar; + int ret = ARCHIVE_FAILED; + + tar = (struct tar *)(a->format->data); + if (strcmp(key, "compat-2x") == 0) { + /* Handle UTF-8 filenames as libarchive 2.x */ + tar->compat_2x = (val != NULL && val[0] != 0); + tar->init_default_conversion = tar->compat_2x; + return (ARCHIVE_OK); + } else if (strcmp(key, "hdrcharset") == 0) { + if (val == NULL || val[0] == 0) + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "tar: hdrcharset option needs a character-set name"); + else { + tar->opt_sconv = + archive_string_conversion_from_charset( + &a->archive, val, 0); + if (tar->opt_sconv != NULL) + ret = ARCHIVE_OK; + else + ret = ARCHIVE_FATAL; + } + return (ret); + } else if (strcmp(key, "mac-ext") == 0) { + tar->process_mac_extensions = (val != NULL && val[0] != 0); + return (ARCHIVE_OK); + } else if (strcmp(key, "read_concatenated_archives") == 0) { + tar->read_concatenated_archives = (val != NULL && val[0] != 0); + return (ARCHIVE_OK); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +/* utility function- this exists to centralize the logic of tracking + * how much unconsumed data we have floating around, and to consume + * anything outstanding since we're going to do read_aheads + */ +static void +tar_flush_unconsumed(struct archive_read *a, size_t *unconsumed) +{ + if (*unconsumed) { +/* + void *data = (void *)__archive_read_ahead(a, *unconsumed, NULL); + * this block of code is to poison claimed unconsumed space, ensuring + * things break if it is in use still. + * currently it WILL break things, so enable it only for debugging this issue + if (data) { + memset(data, 0xff, *unconsumed); + } +*/ + __archive_read_consume(a, *unconsumed); + *unconsumed = 0; + } +} + +/* + * The function invoked by archive_read_next_header(). This + * just sets up a few things and then calls the internal + * tar_read_header() function below. + */ +static int +archive_read_format_tar_read_header(struct archive_read *a, + struct archive_entry *entry) +{ + /* + * When converting tar archives to cpio archives, it is + * essential that each distinct file have a distinct inode + * number. To simplify this, we keep a static count here to + * assign fake dev/inode numbers to each tar entry. Note that + * pax format archives may overwrite this with something more + * useful. + * + * Ideally, we would track every file read from the archive so + * that we could assign the same dev/ino pair to hardlinks, + * but the memory required to store a complete lookup table is + * probably not worthwhile just to support the relatively + * obscure tar->cpio conversion case. + */ + static int default_inode; + static int default_dev; + struct tar *tar; + const char *p; + const wchar_t *wp; + int r; + size_t l, unconsumed = 0; + + /* Assign default device/inode values. */ + archive_entry_set_dev(entry, 1 + default_dev); /* Don't use zero. */ + archive_entry_set_ino(entry, ++default_inode); /* Don't use zero. */ + /* Limit generated st_ino number to 16 bits. */ + if (default_inode >= 0xffff) { + ++default_dev; + default_inode = 0; + } + + tar = (struct tar *)(a->format->data); + tar->entry_offset = 0; + gnu_clear_sparse_list(tar); + tar->realsize = -1; /* Mark this as "unset" */ + tar->realsize_override = 0; + + /* Setup default string conversion. */ + tar->sconv = tar->opt_sconv; + if (tar->sconv == NULL) { + if (!tar->init_default_conversion) { + tar->sconv_default = + archive_string_default_conversion_for_read(&(a->archive)); + tar->init_default_conversion = 1; + } + tar->sconv = tar->sconv_default; + } + + r = tar_read_header(a, tar, entry, &unconsumed); + + tar_flush_unconsumed(a, &unconsumed); + + /* + * "non-sparse" files are really just sparse files with + * a single block. + */ + if (tar->sparse_list == NULL) { + if (gnu_add_sparse_entry(a, tar, 0, tar->entry_bytes_remaining) + != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } else { + struct sparse_block *sb; + + for (sb = tar->sparse_list; sb != NULL; sb = sb->next) { + if (!sb->hole) + archive_entry_sparse_add_entry(entry, + sb->offset, sb->remaining); + } + } + + if (r == ARCHIVE_OK && archive_entry_filetype(entry) == AE_IFREG) { + /* + * "Regular" entry with trailing '/' is really + * directory: This is needed for certain old tar + * variants and even for some broken newer ones. + */ + if ((wp = archive_entry_pathname_w(entry)) != NULL) { + l = wcslen(wp); + if (l > 0 && wp[l - 1] == L'/') { + archive_entry_set_filetype(entry, AE_IFDIR); + } + } else if ((p = archive_entry_pathname(entry)) != NULL) { + l = strlen(p); + if (l > 0 && p[l - 1] == '/') { + archive_entry_set_filetype(entry, AE_IFDIR); + } + } + } + return (r); +} + +static int +archive_read_format_tar_read_data(struct archive_read *a, + const void **buff, size_t *size, int64_t *offset) +{ + ssize_t bytes_read; + struct tar *tar; + struct sparse_block *p; + + tar = (struct tar *)(a->format->data); + + for (;;) { + /* Remove exhausted entries from sparse list. */ + while (tar->sparse_list != NULL && + tar->sparse_list->remaining == 0) { + p = tar->sparse_list; + tar->sparse_list = p->next; + free(p); + } + + if (tar->entry_bytes_unconsumed) { + __archive_read_consume(a, tar->entry_bytes_unconsumed); + tar->entry_bytes_unconsumed = 0; + } + + /* If we're at end of file, return EOF. */ + if (tar->sparse_list == NULL || + tar->entry_bytes_remaining == 0) { + if (__archive_read_consume(a, tar->entry_padding) < 0) + return (ARCHIVE_FATAL); + tar->entry_padding = 0; + *buff = NULL; + *size = 0; + *offset = tar->realsize; + return (ARCHIVE_EOF); + } + + *buff = __archive_read_ahead(a, 1, &bytes_read); + if (bytes_read < 0) + return (ARCHIVE_FATAL); + if (*buff == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Truncated tar archive"); + return (ARCHIVE_FATAL); + } + if (bytes_read > tar->entry_bytes_remaining) + bytes_read = (ssize_t)tar->entry_bytes_remaining; + /* Don't read more than is available in the + * current sparse block. */ + if (tar->sparse_list->remaining < bytes_read) + bytes_read = (ssize_t)tar->sparse_list->remaining; + *size = bytes_read; + *offset = tar->sparse_list->offset; + tar->sparse_list->remaining -= bytes_read; + tar->sparse_list->offset += bytes_read; + tar->entry_bytes_remaining -= bytes_read; + tar->entry_bytes_unconsumed = bytes_read; + + if (!tar->sparse_list->hole) + return (ARCHIVE_OK); + /* Current is hole data and skip this. */ + } +} + +static int +archive_read_format_tar_skip(struct archive_read *a) +{ + int64_t bytes_skipped; + int64_t request; + struct sparse_block *p; + struct tar* tar; + + tar = (struct tar *)(a->format->data); + + /* Do not consume the hole of a sparse file. */ + request = 0; + for (p = tar->sparse_list; p != NULL; p = p->next) { + if (!p->hole) { + if (p->remaining >= INT64_MAX - request) { + return ARCHIVE_FATAL; + } + request += p->remaining; + } + } + if (request > tar->entry_bytes_remaining) + request = tar->entry_bytes_remaining; + request += tar->entry_padding + tar->entry_bytes_unconsumed; + + bytes_skipped = __archive_read_consume(a, request); + if (bytes_skipped < 0) + return (ARCHIVE_FATAL); + + tar->entry_bytes_remaining = 0; + tar->entry_bytes_unconsumed = 0; + tar->entry_padding = 0; + + /* Free the sparse list. */ + gnu_clear_sparse_list(tar); + + return (ARCHIVE_OK); +} + +/* + * This function recursively interprets all of the headers associated + * with a single entry. + */ +static int +tar_read_header(struct archive_read *a, struct tar *tar, + struct archive_entry *entry, size_t *unconsumed) +{ + ssize_t bytes; + int err, eof_vol_header; + const char *h; + const struct archive_entry_header_ustar *header; + const struct archive_entry_header_gnutar *gnuheader; + + eof_vol_header = 0; + + /* Loop until we find a workable header record. */ + for (;;) { + tar_flush_unconsumed(a, unconsumed); + + /* Read 512-byte header record */ + h = __archive_read_ahead(a, 512, &bytes); + if (bytes < 0) + return ((int)bytes); + if (bytes == 0) { /* EOF at a block boundary. */ + /* Some writers do omit the block of nulls. */ + return (ARCHIVE_EOF); + } + if (bytes < 512) { /* Short block at EOF; this is bad. */ + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated tar archive"); + return (ARCHIVE_FATAL); + } + *unconsumed = 512; + + /* Header is workable if it's not an end-of-archive mark. */ + if (h[0] != 0 || !archive_block_is_null(h)) + break; + + /* Ensure format is set for archives with only null blocks. */ + if (a->archive.archive_format_name == NULL) { + a->archive.archive_format = ARCHIVE_FORMAT_TAR; + a->archive.archive_format_name = "tar"; + } + + if (!tar->read_concatenated_archives) { + /* Try to consume a second all-null record, as well. */ + tar_flush_unconsumed(a, unconsumed); + h = __archive_read_ahead(a, 512, NULL); + if (h != NULL && h[0] == 0 && archive_block_is_null(h)) + __archive_read_consume(a, 512); + archive_clear_error(&a->archive); + return (ARCHIVE_EOF); + } + + /* + * We're reading concatenated archives, ignore this block and + * loop to get the next. + */ + } + + /* + * Note: If the checksum fails and we return ARCHIVE_RETRY, + * then the client is likely to just retry. This is a very + * crude way to search for the next valid header! + * + * TODO: Improve this by implementing a real header scan. + */ + if (!checksum(a, h)) { + tar_flush_unconsumed(a, unconsumed); + archive_set_error(&a->archive, EINVAL, "Damaged tar archive"); + return (ARCHIVE_RETRY); /* Retryable: Invalid header */ + } + + if (++tar->header_recursion_depth > 32) { + tar_flush_unconsumed(a, unconsumed); + archive_set_error(&a->archive, EINVAL, "Too many special headers"); + return (ARCHIVE_WARN); + } + + /* Determine the format variant. */ + header = (const struct archive_entry_header_ustar *)h; + + switch(header->typeflag[0]) { + case 'A': /* Solaris tar ACL */ + a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE; + a->archive.archive_format_name = "Solaris tar"; + err = header_Solaris_ACL(a, tar, entry, h, unconsumed); + break; + case 'g': /* POSIX-standard 'g' header. */ + a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE; + a->archive.archive_format_name = "POSIX pax interchange format"; + err = header_pax_global(a, tar, entry, h, unconsumed); + if (err == ARCHIVE_EOF) + return (err); + break; + case 'K': /* Long link name (GNU tar, others) */ + err = header_longlink(a, tar, entry, h, unconsumed); + break; + case 'L': /* Long filename (GNU tar, others) */ + err = header_longname(a, tar, entry, h, unconsumed); + break; + case 'V': /* GNU volume header */ + err = header_volume(a, tar, entry, h, unconsumed); + if (err == ARCHIVE_EOF) + eof_vol_header = 1; + break; + case 'X': /* Used by SUN tar; same as 'x'. */ + a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE; + a->archive.archive_format_name = + "POSIX pax interchange format (Sun variant)"; + err = header_pax_extensions(a, tar, entry, h, unconsumed); + break; + case 'x': /* POSIX-standard 'x' header. */ + a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE; + a->archive.archive_format_name = "POSIX pax interchange format"; + err = header_pax_extensions(a, tar, entry, h, unconsumed); + break; + default: + gnuheader = (const struct archive_entry_header_gnutar *)h; + if (memcmp(gnuheader->magic, "ustar \0", 8) == 0) { + a->archive.archive_format = ARCHIVE_FORMAT_TAR_GNUTAR; + a->archive.archive_format_name = "GNU tar format"; + err = header_gnutar(a, tar, entry, h, unconsumed); + } else if (memcmp(header->magic, "ustar", 5) == 0) { + if (a->archive.archive_format != ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE) { + a->archive.archive_format = ARCHIVE_FORMAT_TAR_USTAR; + a->archive.archive_format_name = "POSIX ustar format"; + } + err = header_ustar(a, tar, entry, h); + } else { + a->archive.archive_format = ARCHIVE_FORMAT_TAR; + a->archive.archive_format_name = "tar (non-POSIX)"; + err = header_old_tar(a, tar, entry, h); + } + } + if (err == ARCHIVE_FATAL) + return (err); + + tar_flush_unconsumed(a, unconsumed); + + h = NULL; + header = NULL; + + --tar->header_recursion_depth; + /* Yuck. Apple's design here ends up storing long pathname + * extensions for both the AppleDouble extension entry and the + * regular entry. + */ + if ((err == ARCHIVE_WARN || err == ARCHIVE_OK) && + tar->header_recursion_depth == 0 && + tar->process_mac_extensions) { + int err2 = read_mac_metadata_blob(a, tar, entry, h, unconsumed); + if (err2 < err) + err = err2; + } + + /* We return warnings or success as-is. Anything else is fatal. */ + if (err == ARCHIVE_WARN || err == ARCHIVE_OK) { + if (tar->sparse_gnu_pending) { + if (tar->sparse_gnu_major == 1 && + tar->sparse_gnu_minor == 0) { + ssize_t bytes_read; + + tar->sparse_gnu_pending = 0; + /* Read initial sparse map. */ + bytes_read = gnu_sparse_10_read(a, tar, unconsumed); + if (bytes_read < 0) + return ((int)bytes_read); + tar->entry_bytes_remaining -= bytes_read; + } else { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Unrecognized GNU sparse file format"); + return (ARCHIVE_WARN); + } + tar->sparse_gnu_pending = 0; + } + return (err); + } + if (err == ARCHIVE_EOF) { + if (!eof_vol_header) { + /* EOF when recursively reading a header is bad. */ + archive_set_error(&a->archive, EINVAL, + "Damaged tar archive"); + } else { + /* If we encounter just a GNU volume header treat + * this situation as an empty archive */ + return (ARCHIVE_EOF); + } + } + return (ARCHIVE_FATAL); +} + +/* + * Return true if block checksum is correct. + */ +static int +checksum(struct archive_read *a, const void *h) +{ + const unsigned char *bytes; + const struct archive_entry_header_ustar *header; + int check, sum; + size_t i; + + (void)a; /* UNUSED */ + bytes = (const unsigned char *)h; + header = (const struct archive_entry_header_ustar *)h; + + /* Checksum field must hold an octal number */ + for (i = 0; i < sizeof(header->checksum); ++i) { + char c = header->checksum[i]; + if (c != ' ' && c != '\0' && (c < '0' || c > '7')) + return 0; + } + + /* + * Test the checksum. Note that POSIX specifies _unsigned_ + * bytes for this calculation. + */ + sum = (int)tar_atol(header->checksum, sizeof(header->checksum)); + check = 0; + for (i = 0; i < 148; i++) + check += (unsigned char)bytes[i]; + for (; i < 156; i++) + check += 32; + for (; i < 512; i++) + check += (unsigned char)bytes[i]; + if (sum == check) + return (1); + + /* + * Repeat test with _signed_ bytes, just in case this archive + * was created by an old BSD, Solaris, or HP-UX tar with a + * broken checksum calculation. + */ + check = 0; + for (i = 0; i < 148; i++) + check += (signed char)bytes[i]; + for (; i < 156; i++) + check += 32; + for (; i < 512; i++) + check += (signed char)bytes[i]; + if (sum == check) + return (1); + + return (0); +} + +/* + * Return true if this block contains only nulls. + */ +static int +archive_block_is_null(const char *p) +{ + unsigned i; + + for (i = 0; i < 512; i++) + if (*p++) + return (0); + return (1); +} + +/* + * Interpret 'A' Solaris ACL header + */ +static int +header_Solaris_ACL(struct archive_read *a, struct tar *tar, + struct archive_entry *entry, const void *h, size_t *unconsumed) +{ + const struct archive_entry_header_ustar *header; + size_t size; + int err, acl_type; + int64_t type; + char *acl, *p; + + /* + * read_body_to_string adds a NUL terminator, but we need a little + * more to make sure that we don't overrun acl_text later. + */ + header = (const struct archive_entry_header_ustar *)h; + size = (size_t)tar_atol(header->size, sizeof(header->size)); + err = read_body_to_string(a, tar, &(tar->acl_text), h, unconsumed); + if (err != ARCHIVE_OK) + return (err); + + /* Recursively read next header */ + err = tar_read_header(a, tar, entry, unconsumed); + if ((err != ARCHIVE_OK) && (err != ARCHIVE_WARN)) + return (err); + + /* TODO: Examine the first characters to see if this + * is an AIX ACL descriptor. We'll likely never support + * them, but it would be polite to recognize and warn when + * we do see them. */ + + /* Leading octal number indicates ACL type and number of entries. */ + p = acl = tar->acl_text.s; + type = 0; + while (*p != '\0' && p < acl + size) { + if (*p < '0' || *p > '7') { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Malformed Solaris ACL attribute (invalid digit)"); + return(ARCHIVE_WARN); + } + type <<= 3; + type += *p - '0'; + if (type > 077777777) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Malformed Solaris ACL attribute (count too large)"); + return (ARCHIVE_WARN); + } + p++; + } + switch ((int)type & ~0777777) { + case 01000000: + /* POSIX.1e ACL */ + acl_type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS; + break; + case 03000000: + /* NFSv4 ACL */ + acl_type = ARCHIVE_ENTRY_ACL_TYPE_NFS4; + break; + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Malformed Solaris ACL attribute (unsupported type %o)", + (int)type); + return (ARCHIVE_WARN); + } + p++; + + if (p >= acl + size) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Malformed Solaris ACL attribute (body overflow)"); + return(ARCHIVE_WARN); + } + + /* ACL text is null-terminated; find the end. */ + size -= (p - acl); + acl = p; + + while (*p != '\0' && p < acl + size) + p++; + + if (tar->sconv_acl == NULL) { + tar->sconv_acl = archive_string_conversion_from_charset( + &(a->archive), "UTF-8", 1); + if (tar->sconv_acl == NULL) + return (ARCHIVE_FATAL); + } + archive_strncpy(&(tar->localname), acl, p - acl); + err = archive_acl_from_text_l(archive_entry_acl(entry), + tar->localname.s, acl_type, tar->sconv_acl); + if (err != ARCHIVE_OK) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for ACL"); + } else + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Malformed Solaris ACL attribute (unparsable)"); + } + return (err); +} + +/* + * Interpret 'K' long linkname header. + */ +static int +header_longlink(struct archive_read *a, struct tar *tar, + struct archive_entry *entry, const void *h, size_t *unconsumed) +{ + int err; + + err = read_body_to_string(a, tar, &(tar->longlink), h, unconsumed); + if (err != ARCHIVE_OK) + return (err); + err = tar_read_header(a, tar, entry, unconsumed); + if ((err != ARCHIVE_OK) && (err != ARCHIVE_WARN)) + return (err); + /* Set symlink if symlink already set, else hardlink. */ + archive_entry_copy_link(entry, tar->longlink.s); + return (ARCHIVE_OK); +} + +static int +set_conversion_failed_error(struct archive_read *a, + struct archive_string_conv *sconv, const char *name) +{ + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for %s", name); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "%s can't be converted from %s to current locale.", + name, archive_string_conversion_charset_name(sconv)); + return (ARCHIVE_WARN); +} + +/* + * Interpret 'L' long filename header. + */ +static int +header_longname(struct archive_read *a, struct tar *tar, + struct archive_entry *entry, const void *h, size_t *unconsumed) +{ + int err; + + err = read_body_to_string(a, tar, &(tar->longname), h, unconsumed); + if (err != ARCHIVE_OK) + return (err); + /* Read and parse "real" header, then override name. */ + err = tar_read_header(a, tar, entry, unconsumed); + if ((err != ARCHIVE_OK) && (err != ARCHIVE_WARN)) + return (err); + if (archive_entry_copy_pathname_l(entry, tar->longname.s, + archive_strlen(&(tar->longname)), tar->sconv) != 0) + err = set_conversion_failed_error(a, tar->sconv, "Pathname"); + return (err); +} + + +/* + * Interpret 'V' GNU tar volume header. + */ +static int +header_volume(struct archive_read *a, struct tar *tar, + struct archive_entry *entry, const void *h, size_t *unconsumed) +{ + (void)h; + + /* Just skip this and read the next header. */ + return (tar_read_header(a, tar, entry, unconsumed)); +} + +/* + * Read body of an archive entry into an archive_string object. + */ +static int +read_body_to_string(struct archive_read *a, struct tar *tar, + struct archive_string *as, const void *h, size_t *unconsumed) +{ + int64_t size; + const struct archive_entry_header_ustar *header; + const void *src; + + (void)tar; /* UNUSED */ + header = (const struct archive_entry_header_ustar *)h; + size = tar_atol(header->size, sizeof(header->size)); + if ((size > 1048576) || (size < 0)) { + archive_set_error(&a->archive, EINVAL, + "Special header too large"); + return (ARCHIVE_FATAL); + } + + /* Fail if we can't make our buffer big enough. */ + if (archive_string_ensure(as, (size_t)size+1) == NULL) { + archive_set_error(&a->archive, ENOMEM, + "No memory"); + return (ARCHIVE_FATAL); + } + + tar_flush_unconsumed(a, unconsumed); + + /* Read the body into the string. */ + *unconsumed = (size_t)((size + 511) & ~ 511); + src = __archive_read_ahead(a, *unconsumed, NULL); + if (src == NULL) { + *unconsumed = 0; + return (ARCHIVE_FATAL); + } + memcpy(as->s, src, (size_t)size); + as->s[size] = '\0'; + as->length = (size_t)size; + return (ARCHIVE_OK); +} + +/* + * Parse out common header elements. + * + * This would be the same as header_old_tar, except that the + * filename is handled slightly differently for old and POSIX + * entries (POSIX entries support a 'prefix'). This factoring + * allows header_old_tar and header_ustar + * to handle filenames differently, while still putting most of the + * common parsing into one place. + */ +static int +header_common(struct archive_read *a, struct tar *tar, + struct archive_entry *entry, const void *h) +{ + const struct archive_entry_header_ustar *header; + char tartype; + int err = ARCHIVE_OK; + + header = (const struct archive_entry_header_ustar *)h; + if (header->linkname[0]) + archive_strncpy(&(tar->entry_linkpath), + header->linkname, sizeof(header->linkname)); + else + archive_string_empty(&(tar->entry_linkpath)); + + /* Parse out the numeric fields (all are octal) */ + archive_entry_set_mode(entry, + (mode_t)tar_atol(header->mode, sizeof(header->mode))); + archive_entry_set_uid(entry, tar_atol(header->uid, sizeof(header->uid))); + archive_entry_set_gid(entry, tar_atol(header->gid, sizeof(header->gid))); + tar->entry_bytes_remaining = tar_atol(header->size, sizeof(header->size)); + if (tar->entry_bytes_remaining < 0) { + tar->entry_bytes_remaining = 0; + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Tar entry has negative size"); + return (ARCHIVE_FATAL); + } + if (tar->entry_bytes_remaining == INT64_MAX) { + /* Note: tar_atol returns INT64_MAX on overflow */ + tar->entry_bytes_remaining = 0; + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Tar entry size overflow"); + return (ARCHIVE_FATAL); + } + tar->realsize = tar->entry_bytes_remaining; + archive_entry_set_size(entry, tar->entry_bytes_remaining); + archive_entry_set_mtime(entry, tar_atol(header->mtime, sizeof(header->mtime)), 0); + + /* Handle the tar type flag appropriately. */ + tartype = header->typeflag[0]; + + switch (tartype) { + case '1': /* Hard link */ + if (archive_entry_copy_hardlink_l(entry, tar->entry_linkpath.s, + archive_strlen(&(tar->entry_linkpath)), tar->sconv) != 0) { + err = set_conversion_failed_error(a, tar->sconv, + "Linkname"); + if (err == ARCHIVE_FATAL) + return (err); + } + /* + * The following may seem odd, but: Technically, tar + * does not store the file type for a "hard link" + * entry, only the fact that it is a hard link. So, I + * leave the type zero normally. But, pax interchange + * format allows hard links to have data, which + * implies that the underlying entry is a regular + * file. + */ + if (archive_entry_size(entry) > 0) + archive_entry_set_filetype(entry, AE_IFREG); + + /* + * A tricky point: Traditionally, tar readers have + * ignored the size field when reading hardlink + * entries, and some writers put non-zero sizes even + * though the body is empty. POSIX blessed this + * convention in the 1988 standard, but broke with + * this tradition in 2001 by permitting hardlink + * entries to store valid bodies in pax interchange + * format, but not in ustar format. Since there is no + * hard and fast way to distinguish pax interchange + * from earlier archives (the 'x' and 'g' entries are + * optional, after all), we need a heuristic. + */ + if (archive_entry_size(entry) == 0) { + /* If the size is already zero, we're done. */ + } else if (a->archive.archive_format + == ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE) { + /* Definitely pax extended; must obey hardlink size. */ + } else if (a->archive.archive_format == ARCHIVE_FORMAT_TAR + || a->archive.archive_format == ARCHIVE_FORMAT_TAR_GNUTAR) + { + /* Old-style or GNU tar: we must ignore the size. */ + archive_entry_set_size(entry, 0); + tar->entry_bytes_remaining = 0; + } else if (archive_read_format_tar_bid(a, 50) > 50) { + /* + * We don't know if it's pax: If the bid + * function sees a valid ustar header + * immediately following, then let's ignore + * the hardlink size. + */ + archive_entry_set_size(entry, 0); + tar->entry_bytes_remaining = 0; + } + /* + * TODO: There are still two cases I'd like to handle: + * = a ustar non-pax archive with a hardlink entry at + * end-of-archive. (Look for block of nulls following?) + * = a pax archive that has not seen any pax headers + * and has an entry which is a hardlink entry storing + * a body containing an uncompressed tar archive. + * The first is worth addressing; I don't see any reliable + * way to deal with the second possibility. + */ + break; + case '2': /* Symlink */ + archive_entry_set_filetype(entry, AE_IFLNK); + archive_entry_set_size(entry, 0); + tar->entry_bytes_remaining = 0; + if (archive_entry_copy_symlink_l(entry, tar->entry_linkpath.s, + archive_strlen(&(tar->entry_linkpath)), tar->sconv) != 0) { + err = set_conversion_failed_error(a, tar->sconv, + "Linkname"); + if (err == ARCHIVE_FATAL) + return (err); + } + break; + case '3': /* Character device */ + archive_entry_set_filetype(entry, AE_IFCHR); + archive_entry_set_size(entry, 0); + tar->entry_bytes_remaining = 0; + break; + case '4': /* Block device */ + archive_entry_set_filetype(entry, AE_IFBLK); + archive_entry_set_size(entry, 0); + tar->entry_bytes_remaining = 0; + break; + case '5': /* Dir */ + archive_entry_set_filetype(entry, AE_IFDIR); + archive_entry_set_size(entry, 0); + tar->entry_bytes_remaining = 0; + break; + case '6': /* FIFO device */ + archive_entry_set_filetype(entry, AE_IFIFO); + archive_entry_set_size(entry, 0); + tar->entry_bytes_remaining = 0; + break; + case 'D': /* GNU incremental directory type */ + /* + * No special handling is actually required here. + * It might be nice someday to preprocess the file list and + * provide it to the client, though. + */ + archive_entry_set_filetype(entry, AE_IFDIR); + break; + case 'M': /* GNU "Multi-volume" (remainder of file from last archive)*/ + /* + * As far as I can tell, this is just like a regular file + * entry, except that the contents should be _appended_ to + * the indicated file at the indicated offset. This may + * require some API work to fully support. + */ + break; + case 'N': /* Old GNU "long filename" entry. */ + /* The body of this entry is a script for renaming + * previously-extracted entries. Ugh. It will never + * be supported by libarchive. */ + archive_entry_set_filetype(entry, AE_IFREG); + break; + case 'S': /* GNU sparse files */ + /* + * Sparse files are really just regular files with + * sparse information in the extended area. + */ + /* FALLTHROUGH */ + case '0': + /* + * Enable sparse file "read" support only for regular + * files and explicit GNU sparse files. However, we + * don't allow non-standard file types to be sparse. + */ + tar->sparse_allowed = 1; + /* FALLTHROUGH */ + default: /* Regular file and non-standard types */ + /* + * Per POSIX: non-recognized types should always be + * treated as regular files. + */ + archive_entry_set_filetype(entry, AE_IFREG); + break; + } + return (err); +} + +/* + * Parse out header elements for "old-style" tar archives. + */ +static int +header_old_tar(struct archive_read *a, struct tar *tar, + struct archive_entry *entry, const void *h) +{ + const struct archive_entry_header_ustar *header; + int err = ARCHIVE_OK, err2; + + /* Copy filename over (to ensure null termination). */ + header = (const struct archive_entry_header_ustar *)h; + if (archive_entry_copy_pathname_l(entry, + header->name, sizeof(header->name), tar->sconv) != 0) { + err = set_conversion_failed_error(a, tar->sconv, "Pathname"); + if (err == ARCHIVE_FATAL) + return (err); + } + + /* Grab rest of common fields */ + err2 = header_common(a, tar, entry, h); + if (err > err2) + err = err2; + + tar->entry_padding = 0x1ff & (-tar->entry_bytes_remaining); + return (err); +} + +/* + * Read a Mac AppleDouble-encoded blob of file metadata, + * if there is one. + */ +static int +read_mac_metadata_blob(struct archive_read *a, struct tar *tar, + struct archive_entry *entry, const void *h, size_t *unconsumed) +{ + int64_t size; + const void *data; + const char *p, *name; + const wchar_t *wp, *wname; + + (void)h; /* UNUSED */ + + wname = wp = archive_entry_pathname_w(entry); + if (wp != NULL) { + /* Find the last path element. */ + for (; *wp != L'\0'; ++wp) { + if (wp[0] == '/' && wp[1] != L'\0') + wname = wp + 1; + } + /* + * If last path element starts with "._", then + * this is a Mac extension. + */ + if (wname[0] != L'.' || wname[1] != L'_' || wname[2] == L'\0') + return ARCHIVE_OK; + } else { + /* Find the last path element. */ + name = p = archive_entry_pathname(entry); + if (p == NULL) + return (ARCHIVE_FAILED); + for (; *p != '\0'; ++p) { + if (p[0] == '/' && p[1] != '\0') + name = p + 1; + } + /* + * If last path element starts with "._", then + * this is a Mac extension. + */ + if (name[0] != '.' || name[1] != '_' || name[2] == '\0') + return ARCHIVE_OK; + } + + /* Read the body as a Mac OS metadata blob. */ + size = archive_entry_size(entry); + + /* + * TODO: Look beyond the body here to peek at the next header. + * If it's a regular header (not an extension header) + * that has the wrong name, just return the current + * entry as-is, without consuming the body here. + * That would reduce the risk of us mis-identifying + * an ordinary file that just happened to have + * a name starting with "._". + * + * Q: Is the above idea really possible? Even + * when there are GNU or pax extension entries? + */ + data = __archive_read_ahead(a, (size_t)size, NULL); + if (data == NULL) { + *unconsumed = 0; + return (ARCHIVE_FATAL); + } + archive_entry_copy_mac_metadata(entry, data, (size_t)size); + *unconsumed = (size_t)((size + 511) & ~ 511); + tar_flush_unconsumed(a, unconsumed); + return (tar_read_header(a, tar, entry, unconsumed)); +} + +/* + * Parse a file header for a pax extended archive entry. + */ +static int +header_pax_global(struct archive_read *a, struct tar *tar, + struct archive_entry *entry, const void *h, size_t *unconsumed) +{ + int err; + + err = read_body_to_string(a, tar, &(tar->pax_global), h, unconsumed); + if (err != ARCHIVE_OK) + return (err); + err = tar_read_header(a, tar, entry, unconsumed); + return (err); +} + +static int +header_pax_extensions(struct archive_read *a, struct tar *tar, + struct archive_entry *entry, const void *h, size_t *unconsumed) +{ + int err, err2; + + err = read_body_to_string(a, tar, &(tar->pax_header), h, unconsumed); + if (err != ARCHIVE_OK) + return (err); + + /* Parse the next header. */ + err = tar_read_header(a, tar, entry, unconsumed); + if ((err != ARCHIVE_OK) && (err != ARCHIVE_WARN)) + return (err); + + /* + * TODO: Parse global/default options into 'entry' struct here + * before handling file-specific options. + * + * This design (parse standard header, then overwrite with pax + * extended attribute data) usually works well, but isn't ideal; + * it would be better to parse the pax extended attributes first + * and then skip any fields in the standard header that were + * defined in the pax header. + */ + err2 = pax_header(a, tar, entry, &tar->pax_header); + err = err_combine(err, err2); + tar->entry_padding = 0x1ff & (-tar->entry_bytes_remaining); + return (err); +} + + +/* + * Parse a file header for a Posix "ustar" archive entry. This also + * handles "pax" or "extended ustar" entries. + */ +static int +header_ustar(struct archive_read *a, struct tar *tar, + struct archive_entry *entry, const void *h) +{ + const struct archive_entry_header_ustar *header; + struct archive_string *as; + int err = ARCHIVE_OK, r; + + header = (const struct archive_entry_header_ustar *)h; + + /* Copy name into an internal buffer to ensure null-termination. */ + as = &(tar->entry_pathname); + if (header->prefix[0]) { + archive_strncpy(as, header->prefix, sizeof(header->prefix)); + if (as->s[archive_strlen(as) - 1] != '/') + archive_strappend_char(as, '/'); + archive_strncat(as, header->name, sizeof(header->name)); + } else { + archive_strncpy(as, header->name, sizeof(header->name)); + } + if (archive_entry_copy_pathname_l(entry, as->s, archive_strlen(as), + tar->sconv) != 0) { + err = set_conversion_failed_error(a, tar->sconv, "Pathname"); + if (err == ARCHIVE_FATAL) + return (err); + } + + /* Handle rest of common fields. */ + r = header_common(a, tar, entry, h); + if (r == ARCHIVE_FATAL) + return (r); + if (r < err) + err = r; + + /* Handle POSIX ustar fields. */ + if (archive_entry_copy_uname_l(entry, + header->uname, sizeof(header->uname), tar->sconv) != 0) { + err = set_conversion_failed_error(a, tar->sconv, "Uname"); + if (err == ARCHIVE_FATAL) + return (err); + } + + if (archive_entry_copy_gname_l(entry, + header->gname, sizeof(header->gname), tar->sconv) != 0) { + err = set_conversion_failed_error(a, tar->sconv, "Gname"); + if (err == ARCHIVE_FATAL) + return (err); + } + + /* Parse out device numbers only for char and block specials. */ + if (header->typeflag[0] == '3' || header->typeflag[0] == '4') { + archive_entry_set_rdevmajor(entry, (dev_t) + tar_atol(header->rdevmajor, sizeof(header->rdevmajor))); + archive_entry_set_rdevminor(entry, (dev_t) + tar_atol(header->rdevminor, sizeof(header->rdevminor))); + } + + tar->entry_padding = 0x1ff & (-tar->entry_bytes_remaining); + + return (err); +} + + +/* + * Parse the pax extended attributes record. + * + * Returns non-zero if there's an error in the data. + */ +static int +pax_header(struct archive_read *a, struct tar *tar, + struct archive_entry *entry, struct archive_string *in_as) +{ + size_t attr_length, l, line_length, value_length; + char *p; + char *key, *value; + struct archive_string *as; + struct archive_string_conv *sconv; + int err, err2; + char *attr = in_as->s; + + attr_length = in_as->length; + tar->pax_hdrcharset_binary = 0; + archive_string_empty(&(tar->entry_gname)); + archive_string_empty(&(tar->entry_linkpath)); + archive_string_empty(&(tar->entry_pathname)); + archive_string_empty(&(tar->entry_pathname_override)); + archive_string_empty(&(tar->entry_uname)); + err = ARCHIVE_OK; + while (attr_length > 0) { + /* Parse decimal length field at start of line. */ + line_length = 0; + l = attr_length; + p = attr; /* Record start of line. */ + while (l>0) { + if (*p == ' ') { + p++; + l--; + break; + } + if (*p < '0' || *p > '9') { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Ignoring malformed pax extended attributes"); + return (ARCHIVE_WARN); + } + line_length *= 10; + line_length += *p - '0'; + if (line_length > 999999) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Rejecting pax extended attribute > 1MB"); + return (ARCHIVE_WARN); + } + p++; + l--; + } + + /* + * Parsed length must be no bigger than available data, + * at least 1, and the last character of the line must + * be '\n'. + */ + if (line_length > attr_length + || line_length < 1 + || attr[line_length - 1] != '\n') + { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Ignoring malformed pax extended attribute"); + return (ARCHIVE_WARN); + } + + /* Null-terminate the line. */ + attr[line_length - 1] = '\0'; + + /* Find end of key and null terminate it. */ + key = p; + if (key[0] == '=') + return (-1); + while (*p && *p != '=') + ++p; + if (*p == '\0') { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Invalid pax extended attributes"); + return (ARCHIVE_WARN); + } + *p = '\0'; + + value = p + 1; + + /* Some values may be binary data */ + value_length = attr + line_length - 1 - value; + + /* Identify this attribute and set it in the entry. */ + err2 = pax_attribute(a, tar, entry, key, value, value_length); + if (err2 == ARCHIVE_FATAL) + return (err2); + err = err_combine(err, err2); + + /* Skip to next line */ + attr += line_length; + attr_length -= line_length; + } + + /* + * PAX format uses UTF-8 as default charset for its metadata + * unless hdrcharset=BINARY is present in its header. + * We apply the charset specified by the hdrcharset option only + * when the hdrcharset attribute(in PAX header) is BINARY because + * we respect the charset described in PAX header and BINARY also + * means that metadata(filename,uname and gname) character-set + * is unknown. + */ + if (tar->pax_hdrcharset_binary) + sconv = tar->opt_sconv; + else { + sconv = archive_string_conversion_from_charset( + &(a->archive), "UTF-8", 1); + if (sconv == NULL) + return (ARCHIVE_FATAL); + if (tar->compat_2x) + archive_string_conversion_set_opt(sconv, + SCONV_SET_OPT_UTF8_LIBARCHIVE2X); + } + + if (archive_strlen(&(tar->entry_gname)) > 0) { + if (archive_entry_copy_gname_l(entry, tar->entry_gname.s, + archive_strlen(&(tar->entry_gname)), sconv) != 0) { + err = set_conversion_failed_error(a, sconv, "Gname"); + if (err == ARCHIVE_FATAL) + return (err); + /* Use a converted an original name. */ + archive_entry_copy_gname(entry, tar->entry_gname.s); + } + } + if (archive_strlen(&(tar->entry_linkpath)) > 0) { + if (archive_entry_copy_link_l(entry, tar->entry_linkpath.s, + archive_strlen(&(tar->entry_linkpath)), sconv) != 0) { + err = set_conversion_failed_error(a, sconv, "Linkname"); + if (err == ARCHIVE_FATAL) + return (err); + /* Use a converted an original name. */ + archive_entry_copy_link(entry, tar->entry_linkpath.s); + } + } + /* + * Some extensions (such as the GNU sparse file extensions) + * deliberately store a synthetic name under the regular 'path' + * attribute and the real file name under a different attribute. + * Since we're supposed to not care about the order, we + * have no choice but to store all of the various filenames + * we find and figure it all out afterwards. This is the + * figuring out part. + */ + as = NULL; + if (archive_strlen(&(tar->entry_pathname_override)) > 0) + as = &(tar->entry_pathname_override); + else if (archive_strlen(&(tar->entry_pathname)) > 0) + as = &(tar->entry_pathname); + if (as != NULL) { + if (archive_entry_copy_pathname_l(entry, as->s, + archive_strlen(as), sconv) != 0) { + err = set_conversion_failed_error(a, sconv, "Pathname"); + if (err == ARCHIVE_FATAL) + return (err); + /* Use a converted an original name. */ + archive_entry_copy_pathname(entry, as->s); + } + } + if (archive_strlen(&(tar->entry_uname)) > 0) { + if (archive_entry_copy_uname_l(entry, tar->entry_uname.s, + archive_strlen(&(tar->entry_uname)), sconv) != 0) { + err = set_conversion_failed_error(a, sconv, "Uname"); + if (err == ARCHIVE_FATAL) + return (err); + /* Use a converted an original name. */ + archive_entry_copy_uname(entry, tar->entry_uname.s); + } + } + return (err); +} + +static int +pax_attribute_xattr(struct archive_entry *entry, + const char *name, const char *value) +{ + char *name_decoded; + void *value_decoded; + size_t value_len; + + if (strlen(name) < 18 || (memcmp(name, "LIBARCHIVE.xattr.", 17)) != 0) + return 3; + + name += 17; + + /* URL-decode name */ + name_decoded = url_decode(name); + if (name_decoded == NULL) + return 2; + + /* Base-64 decode value */ + value_decoded = base64_decode(value, strlen(value), &value_len); + if (value_decoded == NULL) { + free(name_decoded); + return 1; + } + + archive_entry_xattr_add_entry(entry, name_decoded, + value_decoded, value_len); + + free(name_decoded); + free(value_decoded); + return 0; +} + +static int +pax_attribute_schily_xattr(struct archive_entry *entry, + const char *name, const char *value, size_t value_length) +{ + if (strlen(name) < 14 || (memcmp(name, "SCHILY.xattr.", 13)) != 0) + return 1; + + name += 13; + + archive_entry_xattr_add_entry(entry, name, value, value_length); + + return 0; +} + +static int +pax_attribute_rht_security_selinux(struct archive_entry *entry, + const char *value, size_t value_length) +{ + archive_entry_xattr_add_entry(entry, "security.selinux", + value, value_length); + + return 0; +} + +static int +pax_attribute_acl(struct archive_read *a, struct tar *tar, + struct archive_entry *entry, const char *value, int type) +{ + int r; + const char* errstr; + + switch (type) { + case ARCHIVE_ENTRY_ACL_TYPE_ACCESS: + errstr = "SCHILY.acl.access"; + break; + case ARCHIVE_ENTRY_ACL_TYPE_DEFAULT: + errstr = "SCHILY.acl.default"; + break; + case ARCHIVE_ENTRY_ACL_TYPE_NFS4: + errstr = "SCHILY.acl.ace"; + break; + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Unknown ACL type: %d", type); + return(ARCHIVE_FATAL); + } + + if (tar->sconv_acl == NULL) { + tar->sconv_acl = + archive_string_conversion_from_charset( + &(a->archive), "UTF-8", 1); + if (tar->sconv_acl == NULL) + return (ARCHIVE_FATAL); + } + + r = archive_acl_from_text_l(archive_entry_acl(entry), value, type, + tar->sconv_acl); + if (r != ARCHIVE_OK) { + if (r == ARCHIVE_FATAL) { + archive_set_error(&a->archive, ENOMEM, + "%s %s", "Can't allocate memory for ", + errstr); + return (r); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, "%s %s", "Parse error: ", errstr); + } + return (r); +} + +/* + * Parse a single key=value attribute. key/value pointers are + * assumed to point into reasonably long-lived storage. + * + * Note that POSIX reserves all-lowercase keywords. Vendor-specific + * extensions should always have keywords of the form "VENDOR.attribute" + * In particular, it's quite feasible to support many different + * vendor extensions here. I'm using "LIBARCHIVE" for extensions + * unique to this library. + * + * Investigate other vendor-specific extensions and see if + * any of them look useful. + */ +static int +pax_attribute(struct archive_read *a, struct tar *tar, + struct archive_entry *entry, const char *key, const char *value, size_t value_length) +{ + int64_t s; + long n; + int err = ARCHIVE_OK, r; + + if (value == NULL) + value = ""; /* Disable compiler warning; do not pass + * NULL pointer to strlen(). */ + switch (key[0]) { + case 'G': + /* Reject GNU.sparse.* headers on non-regular files. */ + if (strncmp(key, "GNU.sparse", 10) == 0 && + !tar->sparse_allowed) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Non-regular file cannot be sparse"); + return (ARCHIVE_FATAL); + } + + /* GNU "0.0" sparse pax format. */ + if (strcmp(key, "GNU.sparse.numblocks") == 0) { + tar->sparse_offset = -1; + tar->sparse_numbytes = -1; + tar->sparse_gnu_major = 0; + tar->sparse_gnu_minor = 0; + } + if (strcmp(key, "GNU.sparse.offset") == 0) { + tar->sparse_offset = tar_atol10(value, strlen(value)); + if (tar->sparse_numbytes != -1) { + if (gnu_add_sparse_entry(a, tar, + tar->sparse_offset, tar->sparse_numbytes) + != ARCHIVE_OK) + return (ARCHIVE_FATAL); + tar->sparse_offset = -1; + tar->sparse_numbytes = -1; + } + } + if (strcmp(key, "GNU.sparse.numbytes") == 0) { + tar->sparse_numbytes = tar_atol10(value, strlen(value)); + if (tar->sparse_numbytes != -1) { + if (gnu_add_sparse_entry(a, tar, + tar->sparse_offset, tar->sparse_numbytes) + != ARCHIVE_OK) + return (ARCHIVE_FATAL); + tar->sparse_offset = -1; + tar->sparse_numbytes = -1; + } + } + if (strcmp(key, "GNU.sparse.size") == 0) { + tar->realsize = tar_atol10(value, strlen(value)); + archive_entry_set_size(entry, tar->realsize); + tar->realsize_override = 1; + } + + /* GNU "0.1" sparse pax format. */ + if (strcmp(key, "GNU.sparse.map") == 0) { + tar->sparse_gnu_major = 0; + tar->sparse_gnu_minor = 1; + if (gnu_sparse_01_parse(a, tar, value) != ARCHIVE_OK) + return (ARCHIVE_WARN); + } + + /* GNU "1.0" sparse pax format */ + if (strcmp(key, "GNU.sparse.major") == 0) { + tar->sparse_gnu_major = (int)tar_atol10(value, strlen(value)); + tar->sparse_gnu_pending = 1; + } + if (strcmp(key, "GNU.sparse.minor") == 0) { + tar->sparse_gnu_minor = (int)tar_atol10(value, strlen(value)); + tar->sparse_gnu_pending = 1; + } + if (strcmp(key, "GNU.sparse.name") == 0) { + /* + * The real filename; when storing sparse + * files, GNU tar puts a synthesized name into + * the regular 'path' attribute in an attempt + * to limit confusion. ;-) + */ + archive_strcpy(&(tar->entry_pathname_override), value); + } + if (strcmp(key, "GNU.sparse.realsize") == 0) { + tar->realsize = tar_atol10(value, strlen(value)); + archive_entry_set_size(entry, tar->realsize); + tar->realsize_override = 1; + } + break; + case 'L': + /* Our extensions */ +/* TODO: Handle arbitrary extended attributes... */ +/* + if (strcmp(key, "LIBARCHIVE.xxxxxxx") == 0) + archive_entry_set_xxxxxx(entry, value); +*/ + if (strcmp(key, "LIBARCHIVE.creationtime") == 0) { + pax_time(value, &s, &n); + archive_entry_set_birthtime(entry, s, n); + } + if (strcmp(key, "LIBARCHIVE.symlinktype") == 0) { + if (strcmp(value, "file") == 0) { + archive_entry_set_symlink_type(entry, + AE_SYMLINK_TYPE_FILE); + } else if (strcmp(value, "dir") == 0) { + archive_entry_set_symlink_type(entry, + AE_SYMLINK_TYPE_DIRECTORY); + } + } + if (memcmp(key, "LIBARCHIVE.xattr.", 17) == 0) + pax_attribute_xattr(entry, key, value); + break; + case 'R': + /* GNU tar uses RHT.security header to store SELinux xattrs + * SCHILY.xattr.security.selinux == RHT.security.selinux */ + if (strcmp(key, "RHT.security.selinux") == 0) { + pax_attribute_rht_security_selinux(entry, value, + value_length); + } + break; + case 'S': + /* We support some keys used by the "star" archiver */ + if (strcmp(key, "SCHILY.acl.access") == 0) { + r = pax_attribute_acl(a, tar, entry, value, + ARCHIVE_ENTRY_ACL_TYPE_ACCESS); + if (r == ARCHIVE_FATAL) + return (r); + } else if (strcmp(key, "SCHILY.acl.default") == 0) { + r = pax_attribute_acl(a, tar, entry, value, + ARCHIVE_ENTRY_ACL_TYPE_DEFAULT); + if (r == ARCHIVE_FATAL) + return (r); + } else if (strcmp(key, "SCHILY.acl.ace") == 0) { + r = pax_attribute_acl(a, tar, entry, value, + ARCHIVE_ENTRY_ACL_TYPE_NFS4); + if (r == ARCHIVE_FATAL) + return (r); + } else if (strcmp(key, "SCHILY.devmajor") == 0) { + archive_entry_set_rdevmajor(entry, + (dev_t)tar_atol10(value, strlen(value))); + } else if (strcmp(key, "SCHILY.devminor") == 0) { + archive_entry_set_rdevminor(entry, + (dev_t)tar_atol10(value, strlen(value))); + } else if (strcmp(key, "SCHILY.fflags") == 0) { + archive_entry_copy_fflags_text(entry, value); + } else if (strcmp(key, "SCHILY.dev") == 0) { + archive_entry_set_dev(entry, + (dev_t)tar_atol10(value, strlen(value))); + } else if (strcmp(key, "SCHILY.ino") == 0) { + archive_entry_set_ino(entry, + tar_atol10(value, strlen(value))); + } else if (strcmp(key, "SCHILY.nlink") == 0) { + archive_entry_set_nlink(entry, (unsigned) + tar_atol10(value, strlen(value))); + } else if (strcmp(key, "SCHILY.realsize") == 0) { + tar->realsize = tar_atol10(value, strlen(value)); + tar->realsize_override = 1; + archive_entry_set_size(entry, tar->realsize); + } else if (strncmp(key, "SCHILY.xattr.", 13) == 0) { + pax_attribute_schily_xattr(entry, key, value, + value_length); + } else if (strcmp(key, "SUN.holesdata") == 0) { + /* A Solaris extension for sparse. */ + r = solaris_sparse_parse(a, tar, entry, value); + if (r < err) { + if (r == ARCHIVE_FATAL) + return (r); + err = r; + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Parse error: SUN.holesdata"); + } + } + break; + case 'a': + if (strcmp(key, "atime") == 0) { + pax_time(value, &s, &n); + archive_entry_set_atime(entry, s, n); + } + break; + case 'c': + if (strcmp(key, "ctime") == 0) { + pax_time(value, &s, &n); + archive_entry_set_ctime(entry, s, n); + } else if (strcmp(key, "charset") == 0) { + /* TODO: Publish charset information in entry. */ + } else if (strcmp(key, "comment") == 0) { + /* TODO: Publish comment in entry. */ + } + break; + case 'g': + if (strcmp(key, "gid") == 0) { + archive_entry_set_gid(entry, + tar_atol10(value, strlen(value))); + } else if (strcmp(key, "gname") == 0) { + archive_strcpy(&(tar->entry_gname), value); + } + break; + case 'h': + if (strcmp(key, "hdrcharset") == 0) { + if (strcmp(value, "BINARY") == 0) + /* Binary mode. */ + tar->pax_hdrcharset_binary = 1; + else if (strcmp(value, "ISO-IR 10646 2000 UTF-8") == 0) + tar->pax_hdrcharset_binary = 0; + } + break; + case 'l': + /* pax interchange doesn't distinguish hardlink vs. symlink. */ + if (strcmp(key, "linkpath") == 0) { + archive_strcpy(&(tar->entry_linkpath), value); + } + break; + case 'm': + if (strcmp(key, "mtime") == 0) { + pax_time(value, &s, &n); + archive_entry_set_mtime(entry, s, n); + } + break; + case 'p': + if (strcmp(key, "path") == 0) { + archive_strcpy(&(tar->entry_pathname), value); + } + break; + case 'r': + /* POSIX has reserved 'realtime.*' */ + break; + case 's': + /* POSIX has reserved 'security.*' */ + /* Someday: if (strcmp(key, "security.acl") == 0) { ... } */ + if (strcmp(key, "size") == 0) { + /* "size" is the size of the data in the entry. */ + tar->entry_bytes_remaining + = tar_atol10(value, strlen(value)); + /* + * The "size" pax header keyword always overrides the + * "size" field in the tar header. + * GNU.sparse.realsize, GNU.sparse.size and + * SCHILY.realsize override this value. + */ + if (!tar->realsize_override) { + archive_entry_set_size(entry, + tar->entry_bytes_remaining); + tar->realsize + = tar->entry_bytes_remaining; + } + } + break; + case 'u': + if (strcmp(key, "uid") == 0) { + archive_entry_set_uid(entry, + tar_atol10(value, strlen(value))); + } else if (strcmp(key, "uname") == 0) { + archive_strcpy(&(tar->entry_uname), value); + } + break; + } + return (err); +} + + + +/* + * parse a decimal time value, which may include a fractional portion + */ +static void +pax_time(const char *p, int64_t *ps, long *pn) +{ + char digit; + int64_t s; + unsigned long l; + int sign; + int64_t limit, last_digit_limit; + + limit = INT64_MAX / 10; + last_digit_limit = INT64_MAX % 10; + + s = 0; + sign = 1; + if (*p == '-') { + sign = -1; + p++; + } + while (*p >= '0' && *p <= '9') { + digit = *p - '0'; + if (s > limit || + (s == limit && digit > last_digit_limit)) { + s = INT64_MAX; + break; + } + s = (s * 10) + digit; + ++p; + } + + *ps = s * sign; + + /* Calculate nanoseconds. */ + *pn = 0; + + if (*p != '.') + return; + + l = 100000000UL; + do { + ++p; + if (*p >= '0' && *p <= '9') + *pn += (*p - '0') * l; + else + break; + } while (l /= 10); +} + +/* + * Parse GNU tar header + */ +static int +header_gnutar(struct archive_read *a, struct tar *tar, + struct archive_entry *entry, const void *h, size_t *unconsumed) +{ + const struct archive_entry_header_gnutar *header; + int64_t t; + int err = ARCHIVE_OK; + + /* + * GNU header is like POSIX ustar, except 'prefix' is + * replaced with some other fields. This also means the + * filename is stored as in old-style archives. + */ + + /* Grab fields common to all tar variants. */ + err = header_common(a, tar, entry, h); + if (err == ARCHIVE_FATAL) + return (err); + + /* Copy filename over (to ensure null termination). */ + header = (const struct archive_entry_header_gnutar *)h; + if (archive_entry_copy_pathname_l(entry, + header->name, sizeof(header->name), tar->sconv) != 0) { + err = set_conversion_failed_error(a, tar->sconv, "Pathname"); + if (err == ARCHIVE_FATAL) + return (err); + } + + /* Fields common to ustar and GNU */ + /* XXX Can the following be factored out since it's common + * to ustar and gnu tar? Is it okay to move it down into + * header_common, perhaps? */ + if (archive_entry_copy_uname_l(entry, + header->uname, sizeof(header->uname), tar->sconv) != 0) { + err = set_conversion_failed_error(a, tar->sconv, "Uname"); + if (err == ARCHIVE_FATAL) + return (err); + } + + if (archive_entry_copy_gname_l(entry, + header->gname, sizeof(header->gname), tar->sconv) != 0) { + err = set_conversion_failed_error(a, tar->sconv, "Gname"); + if (err == ARCHIVE_FATAL) + return (err); + } + + /* Parse out device numbers only for char and block specials */ + if (header->typeflag[0] == '3' || header->typeflag[0] == '4') { + archive_entry_set_rdevmajor(entry, (dev_t) + tar_atol(header->rdevmajor, sizeof(header->rdevmajor))); + archive_entry_set_rdevminor(entry, (dev_t) + tar_atol(header->rdevminor, sizeof(header->rdevminor))); + } else + archive_entry_set_rdev(entry, 0); + + tar->entry_padding = 0x1ff & (-tar->entry_bytes_remaining); + + /* Grab GNU-specific fields. */ + t = tar_atol(header->atime, sizeof(header->atime)); + if (t > 0) + archive_entry_set_atime(entry, t, 0); + t = tar_atol(header->ctime, sizeof(header->ctime)); + if (t > 0) + archive_entry_set_ctime(entry, t, 0); + + if (header->realsize[0] != 0) { + tar->realsize + = tar_atol(header->realsize, sizeof(header->realsize)); + archive_entry_set_size(entry, tar->realsize); + tar->realsize_override = 1; + } + + if (header->sparse[0].offset[0] != 0) { + if (gnu_sparse_old_read(a, tar, header, unconsumed) + != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } else { + if (header->isextended[0] != 0) { + /* XXX WTF? XXX */ + } + } + + return (err); +} + +static int +gnu_add_sparse_entry(struct archive_read *a, struct tar *tar, + int64_t offset, int64_t remaining) +{ + struct sparse_block *p; + + p = (struct sparse_block *)calloc(1, sizeof(*p)); + if (p == NULL) { + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + if (tar->sparse_last != NULL) + tar->sparse_last->next = p; + else + tar->sparse_list = p; + tar->sparse_last = p; + if (remaining < 0 || offset < 0 || offset > INT64_MAX - remaining) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed sparse map data"); + return (ARCHIVE_FATAL); + } + p->offset = offset; + p->remaining = remaining; + return (ARCHIVE_OK); +} + +static void +gnu_clear_sparse_list(struct tar *tar) +{ + struct sparse_block *p; + + while (tar->sparse_list != NULL) { + p = tar->sparse_list; + tar->sparse_list = p->next; + free(p); + } + tar->sparse_last = NULL; +} + +/* + * GNU tar old-format sparse data. + * + * GNU old-format sparse data is stored in a fixed-field + * format. Offset/size values are 11-byte octal fields (same + * format as 'size' field in ustart header). These are + * stored in the header, allocating subsequent header blocks + * as needed. Extending the header in this way is a pretty + * severe POSIX violation; this design has earned GNU tar a + * lot of criticism. + */ + +static int +gnu_sparse_old_read(struct archive_read *a, struct tar *tar, + const struct archive_entry_header_gnutar *header, size_t *unconsumed) +{ + ssize_t bytes_read; + const void *data; + struct extended { + struct gnu_sparse sparse[21]; + char isextended[1]; + char padding[7]; + }; + const struct extended *ext; + + if (gnu_sparse_old_parse(a, tar, header->sparse, 4) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + if (header->isextended[0] == 0) + return (ARCHIVE_OK); + + do { + tar_flush_unconsumed(a, unconsumed); + data = __archive_read_ahead(a, 512, &bytes_read); + if (bytes_read < 0) + return (ARCHIVE_FATAL); + if (bytes_read < 512) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated tar archive " + "detected while reading sparse file data"); + return (ARCHIVE_FATAL); + } + *unconsumed = 512; + ext = (const struct extended *)data; + if (gnu_sparse_old_parse(a, tar, ext->sparse, 21) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } while (ext->isextended[0] != 0); + if (tar->sparse_list != NULL) + tar->entry_offset = tar->sparse_list->offset; + return (ARCHIVE_OK); +} + +static int +gnu_sparse_old_parse(struct archive_read *a, struct tar *tar, + const struct gnu_sparse *sparse, int length) +{ + while (length > 0 && sparse->offset[0] != 0) { + if (gnu_add_sparse_entry(a, tar, + tar_atol(sparse->offset, sizeof(sparse->offset)), + tar_atol(sparse->numbytes, sizeof(sparse->numbytes))) + != ARCHIVE_OK) + return (ARCHIVE_FATAL); + sparse++; + length--; + } + return (ARCHIVE_OK); +} + +/* + * GNU tar sparse format 0.0 + * + * Beginning with GNU tar 1.15, sparse files are stored using + * information in the pax extended header. The GNU tar maintainers + * have gone through a number of variations in the process of working + * out this scheme; fortunately, they're all numbered. + * + * Sparse format 0.0 uses attribute GNU.sparse.numblocks to store the + * number of blocks, and GNU.sparse.offset/GNU.sparse.numbytes to + * store offset/size for each block. The repeated instances of these + * latter fields violate the pax specification (which frowns on + * duplicate keys), so this format was quickly replaced. + */ + +/* + * GNU tar sparse format 0.1 + * + * This version replaced the offset/numbytes attributes with + * a single "map" attribute that stored a list of integers. This + * format had two problems: First, the "map" attribute could be very + * long, which caused problems for some implementations. More + * importantly, the sparse data was lost when extracted by archivers + * that didn't recognize this extension. + */ + +static int +gnu_sparse_01_parse(struct archive_read *a, struct tar *tar, const char *p) +{ + const char *e; + int64_t offset = -1, size = -1; + + for (;;) { + e = p; + while (*e != '\0' && *e != ',') { + if (*e < '0' || *e > '9') + return (ARCHIVE_WARN); + e++; + } + if (offset < 0) { + offset = tar_atol10(p, e - p); + if (offset < 0) + return (ARCHIVE_WARN); + } else { + size = tar_atol10(p, e - p); + if (size < 0) + return (ARCHIVE_WARN); + if (gnu_add_sparse_entry(a, tar, offset, size) + != ARCHIVE_OK) + return (ARCHIVE_FATAL); + offset = -1; + } + if (*e == '\0') + return (ARCHIVE_OK); + p = e + 1; + } +} + +/* + * GNU tar sparse format 1.0 + * + * The idea: The offset/size data is stored as a series of base-10 + * ASCII numbers prepended to the file data, so that dearchivers that + * don't support this format will extract the block map along with the + * data and a separate post-process can restore the sparseness. + * + * Unfortunately, GNU tar 1.16 had a bug that added unnecessary + * padding to the body of the file when using this format. GNU tar + * 1.17 corrected this bug without bumping the version number, so + * it's not possible to support both variants. This code supports + * the later variant at the expense of not supporting the former. + * + * This variant also replaced GNU.sparse.size with GNU.sparse.realsize + * and introduced the GNU.sparse.major/GNU.sparse.minor attributes. + */ + +/* + * Read the next line from the input, and parse it as a decimal + * integer followed by '\n'. Returns positive integer value or + * negative on error. + */ +static int64_t +gnu_sparse_10_atol(struct archive_read *a, struct tar *tar, + int64_t *remaining, size_t *unconsumed) +{ + int64_t l, limit, last_digit_limit; + const char *p; + ssize_t bytes_read; + int base, digit; + + base = 10; + limit = INT64_MAX / base; + last_digit_limit = INT64_MAX % base; + + /* + * Skip any lines starting with '#'; GNU tar specs + * don't require this, but they should. + */ + do { + bytes_read = readline(a, tar, &p, + (ssize_t)tar_min(*remaining, 100), unconsumed); + if (bytes_read <= 0) + return (ARCHIVE_FATAL); + *remaining -= bytes_read; + } while (p[0] == '#'); + + l = 0; + while (bytes_read > 0) { + if (*p == '\n') + return (l); + if (*p < '0' || *p >= '0' + base) + return (ARCHIVE_WARN); + digit = *p - '0'; + if (l > limit || (l == limit && digit > last_digit_limit)) + l = INT64_MAX; /* Truncate on overflow. */ + else + l = (l * base) + digit; + p++; + bytes_read--; + } + /* TODO: Error message. */ + return (ARCHIVE_WARN); +} + +/* + * Returns length (in bytes) of the sparse data description + * that was read. + */ +static ssize_t +gnu_sparse_10_read(struct archive_read *a, struct tar *tar, size_t *unconsumed) +{ + ssize_t bytes_read; + int entries; + int64_t offset, size, to_skip, remaining; + + /* Clear out the existing sparse list. */ + gnu_clear_sparse_list(tar); + + remaining = tar->entry_bytes_remaining; + + /* Parse entries. */ + entries = (int)gnu_sparse_10_atol(a, tar, &remaining, unconsumed); + if (entries < 0) + return (ARCHIVE_FATAL); + /* Parse the individual entries. */ + while (entries-- > 0) { + /* Parse offset/size */ + offset = gnu_sparse_10_atol(a, tar, &remaining, unconsumed); + if (offset < 0) + return (ARCHIVE_FATAL); + size = gnu_sparse_10_atol(a, tar, &remaining, unconsumed); + if (size < 0) + return (ARCHIVE_FATAL); + /* Add a new sparse entry. */ + if (gnu_add_sparse_entry(a, tar, offset, size) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + /* Skip rest of block... */ + tar_flush_unconsumed(a, unconsumed); + bytes_read = (ssize_t)(tar->entry_bytes_remaining - remaining); + to_skip = 0x1ff & -bytes_read; + /* Fail if tar->entry_bytes_remaing would get negative */ + if (to_skip > remaining) + return (ARCHIVE_FATAL); + if (to_skip != __archive_read_consume(a, to_skip)) + return (ARCHIVE_FATAL); + return ((ssize_t)(bytes_read + to_skip)); +} + +/* + * Solaris pax extension for a sparse file. This is recorded with the + * data and hole pairs. The way recording sparse information by Solaris' + * pax simply indicates where data and sparse are, so the stored contents + * consist of both data and hole. + */ +static int +solaris_sparse_parse(struct archive_read *a, struct tar *tar, + struct archive_entry *entry, const char *p) +{ + const char *e; + int64_t start, end; + int hole = 1; + + (void)entry; /* UNUSED */ + + end = 0; + if (*p == ' ') + p++; + else + return (ARCHIVE_WARN); + for (;;) { + e = p; + while (*e != '\0' && *e != ' ') { + if (*e < '0' || *e > '9') + return (ARCHIVE_WARN); + e++; + } + start = end; + end = tar_atol10(p, e - p); + if (end < 0) + return (ARCHIVE_WARN); + if (start < end) { + if (gnu_add_sparse_entry(a, tar, start, + end - start) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + tar->sparse_last->hole = hole; + } + if (*e == '\0') + return (ARCHIVE_OK); + p = e + 1; + hole = hole == 0; + } +} + +/*- + * Convert text->integer. + * + * Traditional tar formats (including POSIX) specify base-8 for + * all of the standard numeric fields. This is a significant limitation + * in practice: + * = file size is limited to 8GB + * = rdevmajor and rdevminor are limited to 21 bits + * = uid/gid are limited to 21 bits + * + * There are two workarounds for this: + * = pax extended headers, which use variable-length string fields + * = GNU tar and STAR both allow either base-8 or base-256 in + * most fields. The high bit is set to indicate base-256. + * + * On read, this implementation supports both extensions. + */ +static int64_t +tar_atol(const char *p, size_t char_cnt) +{ + /* + * Technically, GNU tar considers a field to be in base-256 + * only if the first byte is 0xff or 0x80. + */ + if (*p & 0x80) + return (tar_atol256(p, char_cnt)); + return (tar_atol8(p, char_cnt)); +} + +/* + * Note that this implementation does not (and should not!) obey + * locale settings; you cannot simply substitute strtol here, since + * it does obey locale. + */ +static int64_t +tar_atol_base_n(const char *p, size_t char_cnt, int base) +{ + int64_t l, maxval, limit, last_digit_limit; + int digit, sign; + + maxval = INT64_MAX; + limit = INT64_MAX / base; + last_digit_limit = INT64_MAX % base; + + /* the pointer will not be dereferenced if char_cnt is zero + * due to the way the && operator is evaluated. + */ + while (char_cnt != 0 && (*p == ' ' || *p == '\t')) { + p++; + char_cnt--; + } + + sign = 1; + if (char_cnt != 0 && *p == '-') { + sign = -1; + p++; + char_cnt--; + + maxval = INT64_MIN; + limit = -(INT64_MIN / base); + last_digit_limit = INT64_MIN % base; + } + + l = 0; + if (char_cnt != 0) { + digit = *p - '0'; + while (digit >= 0 && digit < base && char_cnt != 0) { + if (l>limit || (l == limit && digit > last_digit_limit)) { + return maxval; /* Truncate on overflow. */ + } + l = (l * base) + digit; + digit = *++p - '0'; + char_cnt--; + } + } + return (sign < 0) ? -l : l; +} + +static int64_t +tar_atol8(const char *p, size_t char_cnt) +{ + return tar_atol_base_n(p, char_cnt, 8); +} + +static int64_t +tar_atol10(const char *p, size_t char_cnt) +{ + return tar_atol_base_n(p, char_cnt, 10); +} + +/* + * Parse a base-256 integer. This is just a variable-length + * twos-complement signed binary value in big-endian order, except + * that the high-order bit is ignored. The values here can be up to + * 12 bytes, so we need to be careful about overflowing 64-bit + * (8-byte) integers. + * + * This code unashamedly assumes that the local machine uses 8-bit + * bytes and twos-complement arithmetic. + */ +static int64_t +tar_atol256(const char *_p, size_t char_cnt) +{ + uint64_t l; + const unsigned char *p = (const unsigned char *)_p; + unsigned char c, neg; + + /* Extend 7-bit 2s-comp to 8-bit 2s-comp, decide sign. */ + c = *p; + if (c & 0x40) { + neg = 0xff; + c |= 0x80; + l = ~ARCHIVE_LITERAL_ULL(0); + } else { + neg = 0; + c &= 0x7f; + l = 0; + } + + /* If more than 8 bytes, check that we can ignore + * high-order bits without overflow. */ + while (char_cnt > sizeof(int64_t)) { + --char_cnt; + if (c != neg) + return neg ? INT64_MIN : INT64_MAX; + c = *++p; + } + + /* c is first byte that fits; if sign mismatch, return overflow */ + if ((c ^ neg) & 0x80) { + return neg ? INT64_MIN : INT64_MAX; + } + + /* Accumulate remaining bytes. */ + while (--char_cnt > 0) { + l = (l << 8) | c; + c = *++p; + } + l = (l << 8) | c; + /* Return signed twos-complement value. */ + return (int64_t)(l); +} + +/* + * Returns length of line (including trailing newline) + * or negative on error. 'start' argument is updated to + * point to first character of line. This avoids copying + * when possible. + */ +static ssize_t +readline(struct archive_read *a, struct tar *tar, const char **start, + ssize_t limit, size_t *unconsumed) +{ + ssize_t bytes_read; + ssize_t total_size = 0; + const void *t; + const char *s; + void *p; + + tar_flush_unconsumed(a, unconsumed); + + t = __archive_read_ahead(a, 1, &bytes_read); + if (bytes_read <= 0) + return (ARCHIVE_FATAL); + s = t; /* Start of line? */ + p = memchr(t, '\n', bytes_read); + /* If we found '\n' in the read buffer, return pointer to that. */ + if (p != NULL) { + bytes_read = 1 + ((const char *)p) - s; + if (bytes_read > limit) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Line too long"); + return (ARCHIVE_FATAL); + } + *unconsumed = bytes_read; + *start = s; + return (bytes_read); + } + *unconsumed = bytes_read; + /* Otherwise, we need to accumulate in a line buffer. */ + for (;;) { + if (total_size + bytes_read > limit) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Line too long"); + return (ARCHIVE_FATAL); + } + if (archive_string_ensure(&tar->line, total_size + bytes_read) == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate working buffer"); + return (ARCHIVE_FATAL); + } + memcpy(tar->line.s + total_size, t, bytes_read); + tar_flush_unconsumed(a, unconsumed); + total_size += bytes_read; + /* If we found '\n', clean up and return. */ + if (p != NULL) { + *start = tar->line.s; + return (total_size); + } + /* Read some more. */ + t = __archive_read_ahead(a, 1, &bytes_read); + if (bytes_read <= 0) + return (ARCHIVE_FATAL); + s = t; /* Start of line? */ + p = memchr(t, '\n', bytes_read); + /* If we found '\n', trim the read. */ + if (p != NULL) { + bytes_read = 1 + ((const char *)p) - s; + } + *unconsumed = bytes_read; + } +} + +/* + * base64_decode - Base64 decode + * + * This accepts most variations of base-64 encoding, including: + * * with or without line breaks + * * with or without the final group padded with '=' or '_' characters + * (The most economical Base-64 variant does not pad the last group and + * omits line breaks; RFC1341 used for MIME requires both.) + */ +static char * +base64_decode(const char *s, size_t len, size_t *out_len) +{ + static const unsigned char digits[64] = { + 'A','B','C','D','E','F','G','H','I','J','K','L','M','N', + 'O','P','Q','R','S','T','U','V','W','X','Y','Z','a','b', + 'c','d','e','f','g','h','i','j','k','l','m','n','o','p', + 'q','r','s','t','u','v','w','x','y','z','0','1','2','3', + '4','5','6','7','8','9','+','/' }; + static unsigned char decode_table[128]; + char *out, *d; + const unsigned char *src = (const unsigned char *)s; + + /* If the decode table is not yet initialized, prepare it. */ + if (decode_table[digits[1]] != 1) { + unsigned i; + memset(decode_table, 0xff, sizeof(decode_table)); + for (i = 0; i < sizeof(digits); i++) + decode_table[digits[i]] = i; + } + + /* Allocate enough space to hold the entire output. */ + /* Note that we may not use all of this... */ + out = (char *)malloc(len - len / 4 + 1); + if (out == NULL) { + *out_len = 0; + return (NULL); + } + d = out; + + while (len > 0) { + /* Collect the next group of (up to) four characters. */ + int v = 0; + int group_size = 0; + while (group_size < 4 && len > 0) { + /* '=' or '_' padding indicates final group. */ + if (*src == '=' || *src == '_') { + len = 0; + break; + } + /* Skip illegal characters (including line breaks) */ + if (*src > 127 || *src < 32 + || decode_table[*src] == 0xff) { + len--; + src++; + continue; + } + v <<= 6; + v |= decode_table[*src++]; + len --; + group_size++; + } + /* Align a short group properly. */ + v <<= 6 * (4 - group_size); + /* Unpack the group we just collected. */ + switch (group_size) { + case 4: d[2] = v & 0xff; + /* FALLTHROUGH */ + case 3: d[1] = (v >> 8) & 0xff; + /* FALLTHROUGH */ + case 2: d[0] = (v >> 16) & 0xff; + break; + case 1: /* this is invalid! */ + break; + } + d += group_size * 3 / 4; + } + + *out_len = d - out; + return (out); +} + +static char * +url_decode(const char *in) +{ + char *out, *d; + const char *s; + + out = (char *)malloc(strlen(in) + 1); + if (out == NULL) + return (NULL); + for (s = in, d = out; *s != '\0'; ) { + if (s[0] == '%' && s[1] != '\0' && s[2] != '\0') { + /* Try to convert % escape */ + int digit1 = tohex(s[1]); + int digit2 = tohex(s[2]); + if (digit1 >= 0 && digit2 >= 0) { + /* Looks good, consume three chars */ + s += 3; + /* Convert output */ + *d++ = ((digit1 << 4) | digit2); + continue; + } + /* Else fall through and treat '%' as normal char */ + } + *d++ = *s++; + } + *d = '\0'; + return (out); +} + +static int +tohex(int c) +{ + if (c >= '0' && c <= '9') + return (c - '0'); + else if (c >= 'A' && c <= 'F') + return (c - 'A' + 10); + else if (c >= 'a' && c <= 'f') + return (c - 'a' + 10); + else + return (-1); +} diff --git a/c/archive_read_support_format_warc.c b/c/archive_read_support_format_warc.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_format_warc.c @@ -0,0 +1,832 @@ +/*- + * Copyright (c) 2014 Sebastian Freundt + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +/** + * WARC is standardised by ISO TC46/SC4/WG12 and currently available as + * ISO 28500:2009. + * For the purposes of this file we used the final draft from: + * http://bibnum.bnf.fr/warc/WARC_ISO_28500_version1_latestdraft.pdf + * + * Todo: + * [ ] real-world warcs can contain resources at endpoints ending in / + * e.g. http://bibnum.bnf.fr/warc/ + * if you're lucky their response contains a Content-Location: header + * pointing to a unix-compliant filename, in the example above it's + * Content-Location: http://bibnum.bnf.fr/warc/index.html + * however, that's not mandated and github for example doesn't follow + * this convention. + * We need a set of archive options to control what to do with + * entries like these, at the moment care is taken to skip them. + * + **/ + +#ifdef HAVE_SYS_STAT_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_LIMITS_H +#include +#endif +#ifdef HAVE_CTYPE_H +#include +#endif +#ifdef HAVE_TIME_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_read_private.h" + +typedef enum { + WT_NONE, + /* warcinfo */ + WT_INFO, + /* metadata */ + WT_META, + /* resource */ + WT_RSRC, + /* request, unsupported */ + WT_REQ, + /* response, unsupported */ + WT_RSP, + /* revisit, unsupported */ + WT_RVIS, + /* conversion, unsupported */ + WT_CONV, + /* continuation, unsupported at the moment */ + WT_CONT, + /* invalid type */ + LAST_WT +} warc_type_t; + +typedef struct { + size_t len; + const char *str; +} warc_string_t; + +typedef struct { + size_t len; + char *str; +} warc_strbuf_t; + +struct warc_s { + /* content length ahead */ + size_t cntlen; + /* and how much we've processed so far */ + size_t cntoff; + /* and how much we need to consume between calls */ + size_t unconsumed; + + /* string pool */ + warc_strbuf_t pool; + /* previous version */ + unsigned int pver; + /* stringified format name */ + struct archive_string sver; +}; + +static int _warc_bid(struct archive_read *a, int); +static int _warc_cleanup(struct archive_read *a); +static int _warc_read(struct archive_read*, const void**, size_t*, int64_t*); +static int _warc_skip(struct archive_read *a); +static int _warc_rdhdr(struct archive_read *a, struct archive_entry *e); + +/* private routines */ +static unsigned int _warc_rdver(const char buf[10], size_t bsz); +static unsigned int _warc_rdtyp(const char *buf, size_t bsz); +static warc_string_t _warc_rduri(const char *buf, size_t bsz); +static ssize_t _warc_rdlen(const char *buf, size_t bsz); +static time_t _warc_rdrtm(const char *buf, size_t bsz); +static time_t _warc_rdmtm(const char *buf, size_t bsz); +static const char *_warc_find_eoh(const char *buf, size_t bsz); +static const char *_warc_find_eol(const char *buf, size_t bsz); + +int +archive_read_support_format_warc(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct warc_s *w; + int r; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_format_warc"); + + if ((w = calloc(1, sizeof(*w))) == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate warc data"); + return (ARCHIVE_FATAL); + } + + r = __archive_read_register_format( + a, w, "warc", + _warc_bid, NULL, _warc_rdhdr, _warc_read, + _warc_skip, NULL, _warc_cleanup, NULL, NULL); + + if (r != ARCHIVE_OK) { + free(w); + return (r); + } + return (ARCHIVE_OK); +} + +static int +_warc_cleanup(struct archive_read *a) +{ + struct warc_s *w = a->format->data; + + if (w->pool.len > 0U) { + free(w->pool.str); + } + archive_string_free(&w->sver); + free(w); + a->format->data = NULL; + return (ARCHIVE_OK); +} + +static int +_warc_bid(struct archive_read *a, int best_bid) +{ + const char *hdr; + ssize_t nrd; + unsigned int ver; + + (void)best_bid; /* UNUSED */ + + /* check first line of file, it should be a record already */ + if ((hdr = __archive_read_ahead(a, 12U, &nrd)) == NULL) { + /* no idea what to do */ + return -1; + } else if (nrd < 12) { + /* nah, not for us, our magic cookie is at least 12 bytes */ + return -1; + } + + /* otherwise snarf the record's version number */ + ver = _warc_rdver(hdr, nrd); + if (ver < 1200U || ver > 10000U) { + /* we only support WARC 0.12 to 1.0 */ + return -1; + } + + /* otherwise be confident */ + return (64); +} + +static int +_warc_rdhdr(struct archive_read *a, struct archive_entry *entry) +{ +#define HDR_PROBE_LEN (12U) + struct warc_s *w = a->format->data; + unsigned int ver; + const char *buf; + ssize_t nrd; + const char *eoh; + /* for the file name, saves some strndup()'ing */ + warc_string_t fnam; + /* warc record type, not that we really use it a lot */ + warc_type_t ftyp; + /* content-length+error monad */ + ssize_t cntlen; + /* record time is the WARC-Date time we reinterpret it as ctime */ + time_t rtime; + /* mtime is the Last-Modified time which will be the entry's mtime */ + time_t mtime; + +start_over: + /* just use read_ahead() they keep track of unconsumed + * bits and bobs for us; no need to put an extra shift in + * and reproduce that functionality here */ + buf = __archive_read_ahead(a, HDR_PROBE_LEN, &nrd); + + if (nrd < 0) { + /* no good */ + archive_set_error( + &a->archive, ARCHIVE_ERRNO_MISC, + "Bad record header"); + return (ARCHIVE_FATAL); + } else if (buf == NULL) { + /* there should be room for at least WARC/bla\r\n + * must be EOF therefore */ + return (ARCHIVE_EOF); + } + /* looks good so far, try and find the end of the header now */ + eoh = _warc_find_eoh(buf, nrd); + if (eoh == NULL) { + /* still no good, the header end might be beyond the + * probe we've requested, but then again who'd cram + * so much stuff into the header *and* be 28500-compliant */ + archive_set_error( + &a->archive, ARCHIVE_ERRNO_MISC, + "Bad record header"); + return (ARCHIVE_FATAL); + } + ver = _warc_rdver(buf, eoh - buf); + /* we currently support WARC 0.12 to 1.0 */ + if (ver == 0U) { + archive_set_error( + &a->archive, ARCHIVE_ERRNO_MISC, + "Invalid record version"); + return (ARCHIVE_FATAL); + } else if (ver < 1200U || ver > 10000U) { + archive_set_error( + &a->archive, ARCHIVE_ERRNO_MISC, + "Unsupported record version: %u.%u", + ver / 10000, (ver % 10000) / 100); + return (ARCHIVE_FATAL); + } + cntlen = _warc_rdlen(buf, eoh - buf); + if (cntlen < 0) { + /* nightmare! the specs say content-length is mandatory + * so I don't feel overly bad stopping the reader here */ + archive_set_error( + &a->archive, EINVAL, + "Bad content length"); + return (ARCHIVE_FATAL); + } + rtime = _warc_rdrtm(buf, eoh - buf); + if (rtime == (time_t)-1) { + /* record time is mandatory as per WARC/1.0, + * so just barf here, fast and loud */ + archive_set_error( + &a->archive, EINVAL, + "Bad record time"); + return (ARCHIVE_FATAL); + } + + /* let the world know we're a WARC archive */ + a->archive.archive_format = ARCHIVE_FORMAT_WARC; + if (ver != w->pver) { + /* stringify this entry's version */ + archive_string_sprintf(&w->sver, + "WARC/%u.%u", ver / 10000, (ver % 10000) / 100); + /* remember the version */ + w->pver = ver; + } + /* start off with the type */ + ftyp = _warc_rdtyp(buf, eoh - buf); + /* and let future calls know about the content */ + w->cntlen = cntlen; + w->cntoff = 0U; + mtime = 0;/* Avoid compiling error on some platform. */ + + switch (ftyp) { + case WT_RSRC: + case WT_RSP: + /* only try and read the filename in the cases that are + * guaranteed to have one */ + fnam = _warc_rduri(buf, eoh - buf); + /* check the last character in the URI to avoid creating + * directory endpoints as files, see Todo above */ + if (fnam.len == 0 || fnam.str[fnam.len - 1] == '/') { + /* break here for now */ + fnam.len = 0U; + fnam.str = NULL; + break; + } + /* bang to our string pool, so we save a + * malloc()+free() roundtrip */ + if (fnam.len + 1U > w->pool.len) { + w->pool.len = ((fnam.len + 64U) / 64U) * 64U; + w->pool.str = realloc(w->pool.str, w->pool.len); + } + memcpy(w->pool.str, fnam.str, fnam.len); + w->pool.str[fnam.len] = '\0'; + /* let no one else know about the pool, it's a secret, shhh */ + fnam.str = w->pool.str; + + /* snarf mtime or deduce from rtime + * this is a custom header added by our writer, it's quite + * hard to believe anyone else would go through with it + * (apart from being part of some http responses of course) */ + if ((mtime = _warc_rdmtm(buf, eoh - buf)) == (time_t)-1) { + mtime = rtime; + } + break; + default: + fnam.len = 0U; + fnam.str = NULL; + break; + } + + /* now eat some of those delicious buffer bits */ + __archive_read_consume(a, eoh - buf); + + switch (ftyp) { + case WT_RSRC: + case WT_RSP: + if (fnam.len > 0U) { + /* populate entry object */ + archive_entry_set_filetype(entry, AE_IFREG); + archive_entry_copy_pathname(entry, fnam.str); + archive_entry_set_size(entry, cntlen); + archive_entry_set_perm(entry, 0644); + /* rtime is the new ctime, mtime stays mtime */ + archive_entry_set_ctime(entry, rtime, 0L); + archive_entry_set_mtime(entry, mtime, 0L); + break; + } + /* FALLTHROUGH */ + default: + /* consume the content and start over */ + _warc_skip(a); + goto start_over; + } + return (ARCHIVE_OK); +} + +static int +_warc_read(struct archive_read *a, const void **buf, size_t *bsz, int64_t *off) +{ + struct warc_s *w = a->format->data; + const char *rab; + ssize_t nrd; + + if (w->cntoff >= w->cntlen) { + eof: + /* it's our lucky day, no work, we can leave early */ + *buf = NULL; + *bsz = 0U; + *off = w->cntoff + 4U/*for \r\n\r\n separator*/; + w->unconsumed = 0U; + return (ARCHIVE_EOF); + } + + if (w->unconsumed) { + __archive_read_consume(a, w->unconsumed); + w->unconsumed = 0U; + } + + rab = __archive_read_ahead(a, 1U, &nrd); + if (nrd < 0) { + *bsz = 0U; + /* big catastrophe */ + return (int)nrd; + } else if (nrd == 0) { + goto eof; + } else if ((size_t)nrd > w->cntlen - w->cntoff) { + /* clamp to content-length */ + nrd = w->cntlen - w->cntoff; + } + *off = w->cntoff; + *bsz = nrd; + *buf = rab; + + w->cntoff += nrd; + w->unconsumed = (size_t)nrd; + return (ARCHIVE_OK); +} + +static int +_warc_skip(struct archive_read *a) +{ + struct warc_s *w = a->format->data; + + __archive_read_consume(a, w->cntlen + 4U/*\r\n\r\n separator*/); + w->cntlen = 0U; + w->cntoff = 0U; + return (ARCHIVE_OK); +} + + +/* private routines */ +static void* +deconst(const void *c) +{ + return (char *)0x1 + (((const char *)c) - (const char *)0x1); +} + +static char* +xmemmem(const char *hay, const size_t haysize, + const char *needle, const size_t needlesize) +{ + const char *const eoh = hay + haysize; + const char *const eon = needle + needlesize; + const char *hp; + const char *np; + const char *cand; + unsigned int hsum; + unsigned int nsum; + unsigned int eqp; + + /* trivial checks first + * a 0-sized needle is defined to be found anywhere in haystack + * then run strchr() to find a candidate in HAYSTACK (i.e. a portion + * that happens to begin with *NEEDLE) */ + if (needlesize == 0UL) { + return deconst(hay); + } else if ((hay = memchr(hay, *needle, haysize)) == NULL) { + /* trivial */ + return NULL; + } + + /* First characters of haystack and needle are the same now. Both are + * guaranteed to be at least one character long. Now computes the sum + * of characters values of needle together with the sum of the first + * needle_len characters of haystack. */ + for (hp = hay + 1U, np = needle + 1U, hsum = *hay, nsum = *hay, eqp = 1U; + hp < eoh && np < eon; + hsum ^= *hp, nsum ^= *np, eqp &= *hp == *np, hp++, np++); + + /* HP now references the (NEEDLESIZE + 1)-th character. */ + if (np < eon) { + /* haystack is smaller than needle, :O */ + return NULL; + } else if (eqp) { + /* found a match */ + return deconst(hay); + } + + /* now loop through the rest of haystack, + * updating the sum iteratively */ + for (cand = hay; hp < eoh; hp++) { + hsum ^= *cand++; + hsum ^= *hp; + + /* Since the sum of the characters is already known to be + * equal at that point, it is enough to check just NEEDLESIZE - 1 + * characters for equality, + * also CAND is by design < HP, so no need for range checks */ + if (hsum == nsum && memcmp(cand, needle, needlesize - 1U) == 0) { + return deconst(cand); + } + } + return NULL; +} + +static int +strtoi_lim(const char *str, const char **ep, int llim, int ulim) +{ + int res = 0; + const char *sp; + /* we keep track of the number of digits via rulim */ + int rulim; + + for (sp = str, rulim = ulim > 10 ? ulim : 10; + res * 10 <= ulim && rulim && *sp >= '0' && *sp <= '9'; + sp++, rulim /= 10) { + res *= 10; + res += *sp - '0'; + } + if (sp == str) { + res = -1; + } else if (res < llim || res > ulim) { + res = -2; + } + *ep = (const char*)sp; + return res; +} + +static time_t +time_from_tm(struct tm *t) +{ +#if HAVE_TIMEGM + /* Use platform timegm() if available. */ + return (timegm(t)); +#elif HAVE__MKGMTIME64 + return (_mkgmtime64(t)); +#else + /* Else use direct calculation using POSIX assumptions. */ + /* First, fix up tm_yday based on the year/month/day. */ + if (mktime(t) == (time_t)-1) + return ((time_t)-1); + /* Then we can compute timegm() from first principles. */ + return (t->tm_sec + + t->tm_min * 60 + + t->tm_hour * 3600 + + t->tm_yday * 86400 + + (t->tm_year - 70) * 31536000 + + ((t->tm_year - 69) / 4) * 86400 + - ((t->tm_year - 1) / 100) * 86400 + + ((t->tm_year + 299) / 400) * 86400); +#endif +} + +static time_t +xstrpisotime(const char *s, char **endptr) +{ +/** like strptime() but strictly for ISO 8601 Zulu strings */ + struct tm tm; + time_t res = (time_t)-1; + + /* make sure tm is clean */ + memset(&tm, 0, sizeof(tm)); + + /* as a courtesy to our callers, and since this is a non-standard + * routine, we skip leading whitespace */ + while (*s == ' ' || *s == '\t') + ++s; + + /* read year */ + if ((tm.tm_year = strtoi_lim(s, &s, 1583, 4095)) < 0 || *s++ != '-') { + goto out; + } + /* read month */ + if ((tm.tm_mon = strtoi_lim(s, &s, 1, 12)) < 0 || *s++ != '-') { + goto out; + } + /* read day-of-month */ + if ((tm.tm_mday = strtoi_lim(s, &s, 1, 31)) < 0 || *s++ != 'T') { + goto out; + } + /* read hour */ + if ((tm.tm_hour = strtoi_lim(s, &s, 0, 23)) < 0 || *s++ != ':') { + goto out; + } + /* read minute */ + if ((tm.tm_min = strtoi_lim(s, &s, 0, 59)) < 0 || *s++ != ':') { + goto out; + } + /* read second */ + if ((tm.tm_sec = strtoi_lim(s, &s, 0, 60)) < 0 || *s++ != 'Z') { + goto out; + } + + /* massage TM to fulfill some of POSIX' constraints */ + tm.tm_year -= 1900; + tm.tm_mon--; + + /* now convert our custom tm struct to a unix stamp using UTC */ + res = time_from_tm(&tm); + +out: + if (endptr != NULL) { + *endptr = deconst(s); + } + return res; +} + +static unsigned int +_warc_rdver(const char *buf, size_t bsz) +{ + static const char magic[] = "WARC/"; + const char *c; + unsigned int ver = 0U; + unsigned int end = 0U; + + if (bsz < 12 || memcmp(buf, magic, sizeof(magic) - 1U) != 0) { + /* buffer too small or invalid magic */ + return ver; + } + /* looks good so far, read the version number for a laugh */ + buf += sizeof(magic) - 1U; + + if (isdigit((unsigned char)buf[0U]) && (buf[1U] == '.') && + isdigit((unsigned char)buf[2U])) { + /* we support a maximum of 2 digits in the minor version */ + if (isdigit((unsigned char)buf[3U])) + end = 1U; + /* set up major version */ + ver = (buf[0U] - '0') * 10000U; + /* set up minor version */ + if (end == 1U) { + ver += (buf[2U] - '0') * 1000U; + ver += (buf[3U] - '0') * 100U; + } else + ver += (buf[2U] - '0') * 100U; + /* + * WARC below version 0.12 has a space-separated header + * WARC 0.12 and above terminates the version with a CRLF + */ + c = buf + 3U + end; + if (ver >= 1200U) { + if (memcmp(c, "\r\n", 2U) != 0) + ver = 0U; + } else { + /* ver < 1200U */ + if (*c != ' ' && *c != '\t') + ver = 0U; + } + } + return ver; +} + +static unsigned int +_warc_rdtyp(const char *buf, size_t bsz) +{ + static const char _key[] = "\r\nWARC-Type:"; + const char *val, *eol; + + if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { + /* no bother */ + return WT_NONE; + } + val += sizeof(_key) - 1U; + if ((eol = _warc_find_eol(val, buf + bsz - val)) == NULL) { + /* no end of line */ + return WT_NONE; + } + + /* overread whitespace */ + while (val < eol && (*val == ' ' || *val == '\t')) + ++val; + + if (val + 8U == eol) { + if (memcmp(val, "resource", 8U) == 0) + return WT_RSRC; + else if (memcmp(val, "response", 8U) == 0) + return WT_RSP; + } + return WT_NONE; +} + +static warc_string_t +_warc_rduri(const char *buf, size_t bsz) +{ + static const char _key[] = "\r\nWARC-Target-URI:"; + const char *val, *uri, *eol, *p; + warc_string_t res = {0U, NULL}; + + if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { + /* no bother */ + return res; + } + /* overread whitespace */ + val += sizeof(_key) - 1U; + if ((eol = _warc_find_eol(val, buf + bsz - val)) == NULL) { + /* no end of line */ + return res; + } + + while (val < eol && (*val == ' ' || *val == '\t')) + ++val; + + /* overread URL designators */ + if ((uri = xmemmem(val, eol - val, "://", 3U)) == NULL) { + /* not touching that! */ + return res; + } + + /* spaces inside uri are not allowed, CRLF should follow */ + for (p = val; p < eol; p++) { + if (isspace((unsigned char)*p)) + return res; + } + + /* there must be at least space for ftp */ + if (uri < (val + 3U)) + return res; + + /* move uri to point to after :// */ + uri += 3U; + + /* now then, inspect the URI */ + if (memcmp(val, "file", 4U) == 0) { + /* perfect, nothing left to do here */ + + } else if (memcmp(val, "http", 4U) == 0 || + memcmp(val, "ftp", 3U) == 0) { + /* overread domain, and the first / */ + while (uri < eol && *uri++ != '/'); + } else { + /* not sure what to do? best to bugger off */ + return res; + } + res.str = uri; + res.len = eol - uri; + return res; +} + +static ssize_t +_warc_rdlen(const char *buf, size_t bsz) +{ + static const char _key[] = "\r\nContent-Length:"; + const char *val, *eol; + char *on = NULL; + long int len; + + if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { + /* no bother */ + return -1; + } + val += sizeof(_key) - 1U; + if ((eol = _warc_find_eol(val, buf + bsz - val)) == NULL) { + /* no end of line */ + return -1; + } + + /* skip leading whitespace */ + while (val < eol && (*val == ' ' || *val == '\t')) + val++; + /* there must be at least one digit */ + if (!isdigit((unsigned char)*val)) + return -1; + errno = 0; + len = strtol(val, &on, 10); + if (errno != 0 || on != eol) { + /* line must end here */ + return -1; + } + + return (size_t)len; +} + +static time_t +_warc_rdrtm(const char *buf, size_t bsz) +{ + static const char _key[] = "\r\nWARC-Date:"; + const char *val, *eol; + char *on = NULL; + time_t res; + + if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { + /* no bother */ + return (time_t)-1; + } + val += sizeof(_key) - 1U; + if ((eol = _warc_find_eol(val, buf + bsz - val)) == NULL ) { + /* no end of line */ + return -1; + } + + /* xstrpisotime() kindly overreads whitespace for us, so use that */ + res = xstrpisotime(val, &on); + if (on != eol) { + /* line must end here */ + return -1; + } + return res; +} + +static time_t +_warc_rdmtm(const char *buf, size_t bsz) +{ + static const char _key[] = "\r\nLast-Modified:"; + const char *val, *eol; + char *on = NULL; + time_t res; + + if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { + /* no bother */ + return (time_t)-1; + } + val += sizeof(_key) - 1U; + if ((eol = _warc_find_eol(val, buf + bsz - val)) == NULL ) { + /* no end of line */ + return -1; + } + + /* xstrpisotime() kindly overreads whitespace for us, so use that */ + res = xstrpisotime(val, &on); + if (on != eol) { + /* line must end here */ + return -1; + } + return res; +} + +static const char* +_warc_find_eoh(const char *buf, size_t bsz) +{ + static const char _marker[] = "\r\n\r\n"; + const char *hit = xmemmem(buf, bsz, _marker, sizeof(_marker) - 1U); + + if (hit != NULL) { + hit += sizeof(_marker) - 1U; + } + return hit; +} + +static const char* +_warc_find_eol(const char *buf, size_t bsz) +{ + static const char _marker[] = "\r\n"; + const char *hit = xmemmem(buf, bsz, _marker, sizeof(_marker) - 1U); + + return hit; +} +/* archive_read_support_format_warc.c ends here */ diff --git a/c/archive_read_support_format_xar.c b/c/archive_read_support_format_xar.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_format_xar.c @@ -0,0 +1,3328 @@ +/*- + * Copyright (c) 2009 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#if HAVE_LIBXML_XMLREADER_H +#include +#elif HAVE_BSDXML_H +#include +#elif HAVE_EXPAT_H +#include +#endif +#ifdef HAVE_BZLIB_H +#include +#endif +#if HAVE_LZMA_H +#include +#endif +#ifdef HAVE_ZLIB_H +#include +#endif + +#include "archive.h" +#include "archive_digest_private.h" +#include "archive_endian.h" +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_private.h" +#include "archive_read_private.h" + +#if (!defined(HAVE_LIBXML_XMLREADER_H) && \ + !defined(HAVE_BSDXML_H) && !defined(HAVE_EXPAT_H)) ||\ + !defined(HAVE_ZLIB_H) || \ + !defined(ARCHIVE_HAS_MD5) || !defined(ARCHIVE_HAS_SHA1) +/* + * xar needs several external libraries. + * o libxml2 or expat --- XML parser + * o openssl or MD5/SHA1 hash function + * o zlib + * o bzlib2 (option) + * o liblzma (option) + */ +int +archive_read_support_format_xar(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_format_xar"); + + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Xar not supported on this platform"); + return (ARCHIVE_WARN); +} + +#else /* Support xar format */ + +/* #define DEBUG 1 */ +/* #define DEBUG_PRINT_TOC 1 */ +#if DEBUG_PRINT_TOC +#define PRINT_TOC(d, outbytes) do { \ + unsigned char *x = (unsigned char *)(uintptr_t)d; \ + unsigned char c = x[outbytes-1]; \ + x[outbytes - 1] = 0; \ + fprintf(stderr, "%s", x); \ + fprintf(stderr, "%c", c); \ + x[outbytes - 1] = c; \ +} while (0) +#else +#define PRINT_TOC(d, outbytes) +#endif + +#define HEADER_MAGIC 0x78617221 +#define HEADER_SIZE 28 +#define HEADER_VERSION 1 +#define CKSUM_NONE 0 +#define CKSUM_SHA1 1 +#define CKSUM_MD5 2 + +#define MD5_SIZE 16 +#define SHA1_SIZE 20 +#define MAX_SUM_SIZE 20 + +enum enctype { + NONE, + GZIP, + BZIP2, + LZMA, + XZ, +}; + +struct chksumval { + int alg; + size_t len; + unsigned char val[MAX_SUM_SIZE]; +}; + +struct chksumwork { + int alg; +#ifdef ARCHIVE_HAS_MD5 + archive_md5_ctx md5ctx; +#endif +#ifdef ARCHIVE_HAS_SHA1 + archive_sha1_ctx sha1ctx; +#endif +}; + +struct xattr { + struct xattr *next; + struct archive_string name; + uint64_t id; + uint64_t length; + uint64_t offset; + uint64_t size; + enum enctype encoding; + struct chksumval a_sum; + struct chksumval e_sum; + struct archive_string fstype; +}; + +struct xar_file { + struct xar_file *next; + struct xar_file *hdnext; + struct xar_file *parent; + int subdirs; + + unsigned int has; +#define HAS_DATA 0x00001 +#define HAS_PATHNAME 0x00002 +#define HAS_SYMLINK 0x00004 +#define HAS_TIME 0x00008 +#define HAS_UID 0x00010 +#define HAS_GID 0x00020 +#define HAS_MODE 0x00040 +#define HAS_TYPE 0x00080 +#define HAS_DEV 0x00100 +#define HAS_DEVMAJOR 0x00200 +#define HAS_DEVMINOR 0x00400 +#define HAS_INO 0x00800 +#define HAS_FFLAGS 0x01000 +#define HAS_XATTR 0x02000 +#define HAS_ACL 0x04000 +#define HAS_CTIME 0x08000 +#define HAS_MTIME 0x10000 +#define HAS_ATIME 0x20000 + + uint64_t id; + uint64_t length; + uint64_t offset; + uint64_t size; + enum enctype encoding; + struct chksumval a_sum; + struct chksumval e_sum; + struct archive_string pathname; + struct archive_string symlink; + time_t ctime; + time_t mtime; + time_t atime; + struct archive_string uname; + int64_t uid; + struct archive_string gname; + int64_t gid; + mode_t mode; + dev_t dev; + dev_t devmajor; + dev_t devminor; + int64_t ino64; + struct archive_string fflags_text; + unsigned int link; + unsigned int nlink; + struct archive_string hardlink; + struct xattr *xattr_list; +}; + +struct hdlink { + struct hdlink *next; + + unsigned int id; + int cnt; + struct xar_file *files; +}; + +struct heap_queue { + struct xar_file **files; + int allocated; + int used; +}; + +enum xmlstatus { + INIT, + XAR, + TOC, + TOC_CREATION_TIME, + TOC_CHECKSUM, + TOC_CHECKSUM_OFFSET, + TOC_CHECKSUM_SIZE, + TOC_FILE, + FILE_DATA, + FILE_DATA_LENGTH, + FILE_DATA_OFFSET, + FILE_DATA_SIZE, + FILE_DATA_ENCODING, + FILE_DATA_A_CHECKSUM, + FILE_DATA_E_CHECKSUM, + FILE_DATA_CONTENT, + FILE_EA, + FILE_EA_LENGTH, + FILE_EA_OFFSET, + FILE_EA_SIZE, + FILE_EA_ENCODING, + FILE_EA_A_CHECKSUM, + FILE_EA_E_CHECKSUM, + FILE_EA_NAME, + FILE_EA_FSTYPE, + FILE_CTIME, + FILE_MTIME, + FILE_ATIME, + FILE_GROUP, + FILE_GID, + FILE_USER, + FILE_UID, + FILE_MODE, + FILE_DEVICE, + FILE_DEVICE_MAJOR, + FILE_DEVICE_MINOR, + FILE_DEVICENO, + FILE_INODE, + FILE_LINK, + FILE_TYPE, + FILE_NAME, + FILE_ACL, + FILE_ACL_DEFAULT, + FILE_ACL_ACCESS, + FILE_ACL_APPLEEXTENDED, + /* BSD file flags. */ + FILE_FLAGS, + FILE_FLAGS_USER_NODUMP, + FILE_FLAGS_USER_IMMUTABLE, + FILE_FLAGS_USER_APPEND, + FILE_FLAGS_USER_OPAQUE, + FILE_FLAGS_USER_NOUNLINK, + FILE_FLAGS_SYS_ARCHIVED, + FILE_FLAGS_SYS_IMMUTABLE, + FILE_FLAGS_SYS_APPEND, + FILE_FLAGS_SYS_NOUNLINK, + FILE_FLAGS_SYS_SNAPSHOT, + /* Linux file flags. */ + FILE_EXT2, + FILE_EXT2_SecureDeletion, + FILE_EXT2_Undelete, + FILE_EXT2_Compress, + FILE_EXT2_Synchronous, + FILE_EXT2_Immutable, + FILE_EXT2_AppendOnly, + FILE_EXT2_NoDump, + FILE_EXT2_NoAtime, + FILE_EXT2_CompDirty, + FILE_EXT2_CompBlock, + FILE_EXT2_NoCompBlock, + FILE_EXT2_CompError, + FILE_EXT2_BTree, + FILE_EXT2_HashIndexed, + FILE_EXT2_iMagic, + FILE_EXT2_Journaled, + FILE_EXT2_NoTail, + FILE_EXT2_DirSync, + FILE_EXT2_TopDir, + FILE_EXT2_Reserved, + UNKNOWN, +}; + +struct unknown_tag { + struct unknown_tag *next; + struct archive_string name; +}; + +struct xar { + uint64_t offset; /* Current position in the file. */ + int64_t total; + uint64_t h_base; + int end_of_file; +#define OUTBUFF_SIZE (1024 * 64) + unsigned char *outbuff; + + enum xmlstatus xmlsts; + enum xmlstatus xmlsts_unknown; + struct unknown_tag *unknowntags; + int base64text; + + /* + * TOC + */ + uint64_t toc_remaining; + uint64_t toc_total; + uint64_t toc_chksum_offset; + uint64_t toc_chksum_size; + + /* + * For Decoding data. + */ + enum enctype rd_encoding; + z_stream stream; + int stream_valid; +#if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) + bz_stream bzstream; + int bzstream_valid; +#endif +#if HAVE_LZMA_H && HAVE_LIBLZMA + lzma_stream lzstream; + int lzstream_valid; +#endif + /* + * For Checksum data. + */ + struct chksumwork a_sumwrk; + struct chksumwork e_sumwrk; + + struct xar_file *file; /* current reading file. */ + struct xattr *xattr; /* current reading extended attribute. */ + struct heap_queue file_queue; + struct xar_file *hdlink_orgs; + struct hdlink *hdlink_list; + + int entry_init; + uint64_t entry_total; + uint64_t entry_remaining; + size_t entry_unconsumed; + uint64_t entry_size; + enum enctype entry_encoding; + struct chksumval entry_a_sum; + struct chksumval entry_e_sum; + + struct archive_string_conv *sconv; +}; + +struct xmlattr { + struct xmlattr *next; + char *name; + char *value; +}; + +struct xmlattr_list { + struct xmlattr *first; + struct xmlattr **last; +}; + +static int xar_bid(struct archive_read *, int); +static int xar_read_header(struct archive_read *, + struct archive_entry *); +static int xar_read_data(struct archive_read *, + const void **, size_t *, int64_t *); +static int xar_read_data_skip(struct archive_read *); +static int xar_cleanup(struct archive_read *); +static int move_reading_point(struct archive_read *, uint64_t); +static int rd_contents_init(struct archive_read *, + enum enctype, int, int); +static int rd_contents(struct archive_read *, const void **, + size_t *, size_t *, uint64_t); +static uint64_t atol10(const char *, size_t); +static int64_t atol8(const char *, size_t); +static size_t atohex(unsigned char *, size_t, const char *, size_t); +static time_t parse_time(const char *p, size_t n); +static int heap_add_entry(struct archive_read *a, + struct heap_queue *, struct xar_file *); +static struct xar_file *heap_get_entry(struct heap_queue *); +static int add_link(struct archive_read *, + struct xar *, struct xar_file *); +static void checksum_init(struct archive_read *, int, int); +static void checksum_update(struct archive_read *, const void *, + size_t, const void *, size_t); +static int checksum_final(struct archive_read *, const void *, + size_t, const void *, size_t); +static void checksum_cleanup(struct archive_read *); +static int decompression_init(struct archive_read *, enum enctype); +static int decompress(struct archive_read *, const void **, + size_t *, const void *, size_t *); +static int decompression_cleanup(struct archive_read *); +static void xmlattr_cleanup(struct xmlattr_list *); +static int file_new(struct archive_read *, + struct xar *, struct xmlattr_list *); +static void file_free(struct xar_file *); +static int xattr_new(struct archive_read *, + struct xar *, struct xmlattr_list *); +static void xattr_free(struct xattr *); +static int getencoding(struct xmlattr_list *); +static int getsumalgorithm(struct xmlattr_list *); +static int unknowntag_start(struct archive_read *, + struct xar *, const char *); +static void unknowntag_end(struct xar *, const char *); +static int xml_start(struct archive_read *, + const char *, struct xmlattr_list *); +static void xml_end(void *, const char *); +static void xml_data(void *, const char *, int); +static int xml_parse_file_flags(struct xar *, const char *); +static int xml_parse_file_ext2(struct xar *, const char *); +#if defined(HAVE_LIBXML_XMLREADER_H) +static int xml2_xmlattr_setup(struct archive_read *, + struct xmlattr_list *, xmlTextReaderPtr); +static int xml2_read_cb(void *, char *, int); +static int xml2_close_cb(void *); +static void xml2_error_hdr(void *, const char *, xmlParserSeverities, + xmlTextReaderLocatorPtr); +static int xml2_read_toc(struct archive_read *); +#elif defined(HAVE_BSDXML_H) || defined(HAVE_EXPAT_H) +struct expat_userData { + int state; + struct archive_read *archive; +}; +static int expat_xmlattr_setup(struct archive_read *, + struct xmlattr_list *, const XML_Char **); +static void expat_start_cb(void *, const XML_Char *, const XML_Char **); +static void expat_end_cb(void *, const XML_Char *); +static void expat_data_cb(void *, const XML_Char *, int); +static int expat_read_toc(struct archive_read *); +#endif + +int +archive_read_support_format_xar(struct archive *_a) +{ + struct xar *xar; + struct archive_read *a = (struct archive_read *)_a; + int r; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_format_xar"); + + xar = (struct xar *)calloc(1, sizeof(*xar)); + if (xar == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate xar data"); + return (ARCHIVE_FATAL); + } + + /* initialize xar->file_queue */ + xar->file_queue.allocated = 0; + xar->file_queue.used = 0; + xar->file_queue.files = NULL; + + r = __archive_read_register_format(a, + xar, + "xar", + xar_bid, + NULL, + xar_read_header, + xar_read_data, + xar_read_data_skip, + NULL, + xar_cleanup, + NULL, + NULL); + if (r != ARCHIVE_OK) + free(xar); + return (r); +} + +static int +xar_bid(struct archive_read *a, int best_bid) +{ + const unsigned char *b; + int bid; + + (void)best_bid; /* UNUSED */ + + b = __archive_read_ahead(a, HEADER_SIZE, NULL); + if (b == NULL) + return (-1); + + bid = 0; + /* + * Verify magic code + */ + if (archive_be32dec(b) != HEADER_MAGIC) + return (0); + bid += 32; + /* + * Verify header size + */ + if (archive_be16dec(b+4) != HEADER_SIZE) + return (0); + bid += 16; + /* + * Verify header version + */ + if (archive_be16dec(b+6) != HEADER_VERSION) + return (0); + bid += 16; + /* + * Verify type of checksum + */ + switch (archive_be32dec(b+24)) { + case CKSUM_NONE: + case CKSUM_SHA1: + case CKSUM_MD5: + bid += 32; + break; + default: + return (0); + } + + return (bid); +} + +static int +read_toc(struct archive_read *a) +{ + struct xar *xar; + struct xar_file *file; + const unsigned char *b; + uint64_t toc_compressed_size; + uint64_t toc_uncompressed_size; + uint32_t toc_chksum_alg; + ssize_t bytes; + int r; + + xar = (struct xar *)(a->format->data); + + /* + * Read xar header. + */ + b = __archive_read_ahead(a, HEADER_SIZE, &bytes); + if (bytes < 0) + return ((int)bytes); + if (bytes < HEADER_SIZE) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated archive header"); + return (ARCHIVE_FATAL); + } + + if (archive_be32dec(b) != HEADER_MAGIC) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid header magic"); + return (ARCHIVE_FATAL); + } + if (archive_be16dec(b+6) != HEADER_VERSION) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Unsupported header version(%d)", + archive_be16dec(b+6)); + return (ARCHIVE_FATAL); + } + toc_compressed_size = archive_be64dec(b+8); + xar->toc_remaining = toc_compressed_size; + toc_uncompressed_size = archive_be64dec(b+16); + toc_chksum_alg = archive_be32dec(b+24); + __archive_read_consume(a, HEADER_SIZE); + xar->offset += HEADER_SIZE; + xar->toc_total = 0; + + /* + * Read TOC(Table of Contents). + */ + /* Initialize reading contents. */ + r = move_reading_point(a, HEADER_SIZE); + if (r != ARCHIVE_OK) + return (r); + r = rd_contents_init(a, GZIP, toc_chksum_alg, CKSUM_NONE); + if (r != ARCHIVE_OK) + return (r); + +#ifdef HAVE_LIBXML_XMLREADER_H + r = xml2_read_toc(a); +#elif defined(HAVE_BSDXML_H) || defined(HAVE_EXPAT_H) + r = expat_read_toc(a); +#endif + if (r != ARCHIVE_OK) + return (r); + + /* Set 'The HEAP' base. */ + xar->h_base = xar->offset; + if (xar->toc_total != toc_uncompressed_size) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "TOC uncompressed size error"); + return (ARCHIVE_FATAL); + } + + /* + * Checksum TOC + */ + if (toc_chksum_alg != CKSUM_NONE) { + r = move_reading_point(a, xar->toc_chksum_offset); + if (r != ARCHIVE_OK) + return (r); + b = __archive_read_ahead(a, + (size_t)xar->toc_chksum_size, &bytes); + if (bytes < 0) + return ((int)bytes); + if ((uint64_t)bytes < xar->toc_chksum_size) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated archive file"); + return (ARCHIVE_FATAL); + } + r = checksum_final(a, b, + (size_t)xar->toc_chksum_size, NULL, 0); + __archive_read_consume(a, xar->toc_chksum_size); + xar->offset += xar->toc_chksum_size; + if (r != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + + /* + * Connect hardlinked files. + */ + for (file = xar->hdlink_orgs; file != NULL; file = file->hdnext) { + struct hdlink **hdlink; + + for (hdlink = &(xar->hdlink_list); *hdlink != NULL; + hdlink = &((*hdlink)->next)) { + if ((*hdlink)->id == file->id) { + struct hdlink *hltmp; + struct xar_file *f2; + int nlink = (*hdlink)->cnt + 1; + + file->nlink = nlink; + for (f2 = (*hdlink)->files; f2 != NULL; + f2 = f2->hdnext) { + f2->nlink = nlink; + archive_string_copy( + &(f2->hardlink), &(file->pathname)); + } + /* Remove resolved files from hdlist_list. */ + hltmp = *hdlink; + *hdlink = hltmp->next; + free(hltmp); + break; + } + } + } + a->archive.archive_format = ARCHIVE_FORMAT_XAR; + a->archive.archive_format_name = "xar"; + + return (ARCHIVE_OK); +} + +static int +xar_read_header(struct archive_read *a, struct archive_entry *entry) +{ + struct xar *xar; + struct xar_file *file; + struct xattr *xattr; + int r; + + xar = (struct xar *)(a->format->data); + r = ARCHIVE_OK; + + if (xar->offset == 0) { + /* Create a character conversion object. */ + if (xar->sconv == NULL) { + xar->sconv = archive_string_conversion_from_charset( + &(a->archive), "UTF-8", 1); + if (xar->sconv == NULL) + return (ARCHIVE_FATAL); + } + + /* Read TOC. */ + r = read_toc(a); + if (r != ARCHIVE_OK) + return (r); + } + + for (;;) { + file = xar->file = heap_get_entry(&(xar->file_queue)); + if (file == NULL) { + xar->end_of_file = 1; + return (ARCHIVE_EOF); + } + if ((file->mode & AE_IFMT) != AE_IFDIR) + break; + if (file->has != (HAS_PATHNAME | HAS_TYPE)) + break; + /* + * If a file type is a directory and it does not have + * any metadata, do not export. + */ + file_free(file); + } + if (file->has & HAS_ATIME) { + archive_entry_set_atime(entry, file->atime, 0); + } + if (file->has & HAS_CTIME) { + archive_entry_set_ctime(entry, file->ctime, 0); + } + if (file->has & HAS_MTIME) { + archive_entry_set_mtime(entry, file->mtime, 0); + } + archive_entry_set_gid(entry, file->gid); + if (file->gname.length > 0 && + archive_entry_copy_gname_l(entry, file->gname.s, + archive_strlen(&(file->gname)), xar->sconv) != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Gname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Gname cannot be converted from %s to current locale.", + archive_string_conversion_charset_name(xar->sconv)); + r = ARCHIVE_WARN; + } + archive_entry_set_uid(entry, file->uid); + if (file->uname.length > 0 && + archive_entry_copy_uname_l(entry, file->uname.s, + archive_strlen(&(file->uname)), xar->sconv) != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Uname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Uname cannot be converted from %s to current locale.", + archive_string_conversion_charset_name(xar->sconv)); + r = ARCHIVE_WARN; + } + archive_entry_set_mode(entry, file->mode); + if (archive_entry_copy_pathname_l(entry, file->pathname.s, + archive_strlen(&(file->pathname)), xar->sconv) != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Pathname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Pathname cannot be converted from %s to current locale.", + archive_string_conversion_charset_name(xar->sconv)); + r = ARCHIVE_WARN; + } + + + if (file->symlink.length > 0 && + archive_entry_copy_symlink_l(entry, file->symlink.s, + archive_strlen(&(file->symlink)), xar->sconv) != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Linkname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Linkname cannot be converted from %s to current locale.", + archive_string_conversion_charset_name(xar->sconv)); + r = ARCHIVE_WARN; + } + /* Set proper nlink. */ + if ((file->mode & AE_IFMT) == AE_IFDIR) + archive_entry_set_nlink(entry, file->subdirs + 2); + else + archive_entry_set_nlink(entry, file->nlink); + archive_entry_set_size(entry, file->size); + if (archive_strlen(&(file->hardlink)) > 0) + archive_entry_set_hardlink(entry, file->hardlink.s); + archive_entry_set_ino64(entry, file->ino64); + if (file->has & HAS_DEV) + archive_entry_set_dev(entry, file->dev); + if (file->has & HAS_DEVMAJOR) + archive_entry_set_devmajor(entry, file->devmajor); + if (file->has & HAS_DEVMINOR) + archive_entry_set_devminor(entry, file->devminor); + if (archive_strlen(&(file->fflags_text)) > 0) + archive_entry_copy_fflags_text(entry, file->fflags_text.s); + + xar->entry_init = 1; + xar->entry_total = 0; + xar->entry_remaining = file->length; + xar->entry_size = file->size; + xar->entry_encoding = file->encoding; + xar->entry_a_sum = file->a_sum; + xar->entry_e_sum = file->e_sum; + /* + * Read extended attributes. + */ + xattr = file->xattr_list; + while (xattr != NULL) { + const void *d; + size_t outbytes = 0; + size_t used = 0; + + r = move_reading_point(a, xattr->offset); + if (r != ARCHIVE_OK) + break; + r = rd_contents_init(a, xattr->encoding, + xattr->a_sum.alg, xattr->e_sum.alg); + if (r != ARCHIVE_OK) + break; + d = NULL; + r = rd_contents(a, &d, &outbytes, &used, xattr->length); + if (r != ARCHIVE_OK) + break; + if (outbytes != xattr->size) { + archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, + "Decompressed size error"); + r = ARCHIVE_FATAL; + break; + } + r = checksum_final(a, + xattr->a_sum.val, xattr->a_sum.len, + xattr->e_sum.val, xattr->e_sum.len); + if (r != ARCHIVE_OK) { + archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, + "Xattr checksum error"); + r = ARCHIVE_WARN; + break; + } + if (xattr->name.s == NULL) { + archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, + "Xattr name error"); + r = ARCHIVE_WARN; + break; + } + archive_entry_xattr_add_entry(entry, + xattr->name.s, d, outbytes); + xattr = xattr->next; + } + if (r != ARCHIVE_OK) { + file_free(file); + return (r); + } + + if (xar->entry_remaining > 0) + /* Move reading point to the beginning of current + * file contents. */ + r = move_reading_point(a, file->offset); + else + r = ARCHIVE_OK; + + file_free(file); + return (r); +} + +static int +xar_read_data(struct archive_read *a, + const void **buff, size_t *size, int64_t *offset) +{ + struct xar *xar; + size_t used = 0; + int r; + + xar = (struct xar *)(a->format->data); + + if (xar->entry_unconsumed) { + __archive_read_consume(a, xar->entry_unconsumed); + xar->entry_unconsumed = 0; + } + + if (xar->end_of_file || xar->entry_remaining <= 0) { + r = ARCHIVE_EOF; + goto abort_read_data; + } + + if (xar->entry_init) { + r = rd_contents_init(a, xar->entry_encoding, + xar->entry_a_sum.alg, xar->entry_e_sum.alg); + if (r != ARCHIVE_OK) { + xar->entry_remaining = 0; + return (r); + } + xar->entry_init = 0; + } + + *buff = NULL; + r = rd_contents(a, buff, size, &used, xar->entry_remaining); + if (r != ARCHIVE_OK) + goto abort_read_data; + + *offset = xar->entry_total; + xar->entry_total += *size; + xar->total += *size; + xar->offset += used; + xar->entry_remaining -= used; + xar->entry_unconsumed = used; + + if (xar->entry_remaining == 0) { + if (xar->entry_total != xar->entry_size) { + archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, + "Decompressed size error"); + r = ARCHIVE_FATAL; + goto abort_read_data; + } + r = checksum_final(a, + xar->entry_a_sum.val, xar->entry_a_sum.len, + xar->entry_e_sum.val, xar->entry_e_sum.len); + if (r != ARCHIVE_OK) + goto abort_read_data; + } + + return (ARCHIVE_OK); +abort_read_data: + *buff = NULL; + *size = 0; + *offset = xar->total; + return (r); +} + +static int +xar_read_data_skip(struct archive_read *a) +{ + struct xar *xar; + int64_t bytes_skipped; + + xar = (struct xar *)(a->format->data); + if (xar->end_of_file) + return (ARCHIVE_EOF); + bytes_skipped = __archive_read_consume(a, xar->entry_remaining + + xar->entry_unconsumed); + if (bytes_skipped < 0) + return (ARCHIVE_FATAL); + xar->offset += bytes_skipped; + xar->entry_unconsumed = 0; + return (ARCHIVE_OK); +} + +static int +xar_cleanup(struct archive_read *a) +{ + struct xar *xar; + struct hdlink *hdlink; + int i; + int r; + + xar = (struct xar *)(a->format->data); + checksum_cleanup(a); + r = decompression_cleanup(a); + hdlink = xar->hdlink_list; + while (hdlink != NULL) { + struct hdlink *next = hdlink->next; + + free(hdlink); + hdlink = next; + } + for (i = 0; i < xar->file_queue.used; i++) + file_free(xar->file_queue.files[i]); + free(xar->file_queue.files); + while (xar->unknowntags != NULL) { + struct unknown_tag *tag; + + tag = xar->unknowntags; + xar->unknowntags = tag->next; + archive_string_free(&(tag->name)); + free(tag); + } + free(xar->outbuff); + free(xar); + a->format->data = NULL; + return (r); +} + +static int +move_reading_point(struct archive_read *a, uint64_t offset) +{ + struct xar *xar; + + xar = (struct xar *)(a->format->data); + if (xar->offset - xar->h_base != offset) { + /* Seek forward to the start of file contents. */ + int64_t step; + + step = offset - (xar->offset - xar->h_base); + if (step > 0) { + step = __archive_read_consume(a, step); + if (step < 0) + return ((int)step); + xar->offset += step; + } else { + int64_t pos = __archive_read_seek(a, xar->h_base + offset, SEEK_SET); + if (pos == ARCHIVE_FAILED) { + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "Cannot seek."); + return (ARCHIVE_FAILED); + } + xar->offset = pos; + } + } + return (ARCHIVE_OK); +} + +static int +rd_contents_init(struct archive_read *a, enum enctype encoding, + int a_sum_alg, int e_sum_alg) +{ + int r; + + /* Init decompress library. */ + if ((r = decompression_init(a, encoding)) != ARCHIVE_OK) + return (r); + /* Init checksum library. */ + checksum_init(a, a_sum_alg, e_sum_alg); + return (ARCHIVE_OK); +} + +static int +rd_contents(struct archive_read *a, const void **buff, size_t *size, + size_t *used, uint64_t remaining) +{ + const unsigned char *b; + ssize_t bytes; + + /* Get whatever bytes are immediately available. */ + b = __archive_read_ahead(a, 1, &bytes); + if (bytes < 0) + return ((int)bytes); + if (bytes == 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Truncated archive file"); + return (ARCHIVE_FATAL); + } + if ((uint64_t)bytes > remaining) + bytes = (ssize_t)remaining; + + /* + * Decompress contents of file. + */ + *used = bytes; + if (decompress(a, buff, size, b, used) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + /* + * Update checksum of a compressed data and a extracted data. + */ + checksum_update(a, b, *used, *buff, *size); + + return (ARCHIVE_OK); +} + +/* + * Note that this implementation does not (and should not!) obey + * locale settings; you cannot simply substitute strtol here, since + * it does obey locale. + */ + +static uint64_t +atol10(const char *p, size_t char_cnt) +{ + uint64_t l; + int digit; + + if (char_cnt == 0) + return (0); + + l = 0; + digit = *p - '0'; + while (digit >= 0 && digit < 10 && char_cnt-- > 0) { + l = (l * 10) + digit; + digit = *++p - '0'; + } + return (l); +} + +static int64_t +atol8(const char *p, size_t char_cnt) +{ + int64_t l; + int digit; + + if (char_cnt == 0) + return (0); + + l = 0; + while (char_cnt-- > 0) { + if (*p >= '0' && *p <= '7') + digit = *p - '0'; + else + break; + p++; + l <<= 3; + l |= digit; + } + return (l); +} + +static size_t +atohex(unsigned char *b, size_t bsize, const char *p, size_t psize) +{ + size_t fbsize = bsize; + + while (bsize && psize > 1) { + unsigned char x; + + if (p[0] >= 'a' && p[0] <= 'z') + x = (p[0] - 'a' + 0x0a) << 4; + else if (p[0] >= 'A' && p[0] <= 'Z') + x = (p[0] - 'A' + 0x0a) << 4; + else if (p[0] >= '0' && p[0] <= '9') + x = (p[0] - '0') << 4; + else + return (-1); + if (p[1] >= 'a' && p[1] <= 'z') + x |= p[1] - 'a' + 0x0a; + else if (p[1] >= 'A' && p[1] <= 'Z') + x |= p[1] - 'A' + 0x0a; + else if (p[1] >= '0' && p[1] <= '9') + x |= p[1] - '0'; + else + return (-1); + + *b++ = x; + bsize--; + p += 2; + psize -= 2; + } + return (fbsize - bsize); +} + +static time_t +time_from_tm(struct tm *t) +{ +#if HAVE_TIMEGM + /* Use platform timegm() if available. */ + return (timegm(t)); +#elif HAVE__MKGMTIME64 + return (_mkgmtime64(t)); +#else + /* Else use direct calculation using POSIX assumptions. */ + /* First, fix up tm_yday based on the year/month/day. */ + mktime(t); + /* Then we can compute timegm() from first principles. */ + return (t->tm_sec + + t->tm_min * 60 + + t->tm_hour * 3600 + + t->tm_yday * 86400 + + (t->tm_year - 70) * 31536000 + + ((t->tm_year - 69) / 4) * 86400 + - ((t->tm_year - 1) / 100) * 86400 + + ((t->tm_year + 299) / 400) * 86400); +#endif +} + +static time_t +parse_time(const char *p, size_t n) +{ + struct tm tm; + time_t t = 0; + int64_t data; + + memset(&tm, 0, sizeof(tm)); + if (n != 20) + return (t); + data = atol10(p, 4); + if (data < 1900) + return (t); + tm.tm_year = (int)data - 1900; + p += 4; + if (*p++ != '-') + return (t); + data = atol10(p, 2); + if (data < 1 || data > 12) + return (t); + tm.tm_mon = (int)data -1; + p += 2; + if (*p++ != '-') + return (t); + data = atol10(p, 2); + if (data < 1 || data > 31) + return (t); + tm.tm_mday = (int)data; + p += 2; + if (*p++ != 'T') + return (t); + data = atol10(p, 2); + if (data < 0 || data > 23) + return (t); + tm.tm_hour = (int)data; + p += 2; + if (*p++ != ':') + return (t); + data = atol10(p, 2); + if (data < 0 || data > 59) + return (t); + tm.tm_min = (int)data; + p += 2; + if (*p++ != ':') + return (t); + data = atol10(p, 2); + if (data < 0 || data > 60) + return (t); + tm.tm_sec = (int)data; +#if 0 + p += 2; + if (*p != 'Z') + return (t); +#endif + + t = time_from_tm(&tm); + + return (t); +} + +static int +heap_add_entry(struct archive_read *a, + struct heap_queue *heap, struct xar_file *file) +{ + uint64_t file_id, parent_id; + int hole, parent; + + /* Expand our pending files list as necessary. */ + if (heap->used >= heap->allocated) { + struct xar_file **new_pending_files; + int new_size; + + if (heap->allocated < 1024) + new_size = 1024; + else + new_size = heap->allocated * 2; + /* Overflow might keep us from growing the list. */ + if (new_size <= heap->allocated) { + archive_set_error(&a->archive, + ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + new_pending_files = (struct xar_file **) + malloc(new_size * sizeof(new_pending_files[0])); + if (new_pending_files == NULL) { + archive_set_error(&a->archive, + ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + if (heap->allocated) { + memcpy(new_pending_files, heap->files, + heap->allocated * sizeof(new_pending_files[0])); + free(heap->files); + } + heap->files = new_pending_files; + heap->allocated = new_size; + } + + file_id = file->id; + + /* + * Start with hole at end, walk it up tree to find insertion point. + */ + hole = heap->used++; + while (hole > 0) { + parent = (hole - 1)/2; + parent_id = heap->files[parent]->id; + if (file_id >= parent_id) { + heap->files[hole] = file; + return (ARCHIVE_OK); + } + /* Move parent into hole <==> move hole up tree. */ + heap->files[hole] = heap->files[parent]; + hole = parent; + } + heap->files[0] = file; + + return (ARCHIVE_OK); +} + +static struct xar_file * +heap_get_entry(struct heap_queue *heap) +{ + uint64_t a_id, b_id, c_id; + int a, b, c; + struct xar_file *r, *tmp; + + if (heap->used < 1) + return (NULL); + + /* + * The first file in the list is the earliest; we'll return this. + */ + r = heap->files[0]; + + /* + * Move the last item in the heap to the root of the tree + */ + heap->files[0] = heap->files[--(heap->used)]; + + /* + * Rebalance the heap. + */ + a = 0; /* Starting element and its heap key */ + a_id = heap->files[a]->id; + for (;;) { + b = a + a + 1; /* First child */ + if (b >= heap->used) + return (r); + b_id = heap->files[b]->id; + c = b + 1; /* Use second child if it is smaller. */ + if (c < heap->used) { + c_id = heap->files[c]->id; + if (c_id < b_id) { + b = c; + b_id = c_id; + } + } + if (a_id <= b_id) + return (r); + tmp = heap->files[a]; + heap->files[a] = heap->files[b]; + heap->files[b] = tmp; + a = b; + } +} + +static int +add_link(struct archive_read *a, struct xar *xar, struct xar_file *file) +{ + struct hdlink *hdlink; + + for (hdlink = xar->hdlink_list; hdlink != NULL; hdlink = hdlink->next) { + if (hdlink->id == file->link) { + file->hdnext = hdlink->files; + hdlink->cnt++; + hdlink->files = file; + return (ARCHIVE_OK); + } + } + hdlink = malloc(sizeof(*hdlink)); + if (hdlink == NULL) { + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + file->hdnext = NULL; + hdlink->id = file->link; + hdlink->cnt = 1; + hdlink->files = file; + hdlink->next = xar->hdlink_list; + xar->hdlink_list = hdlink; + return (ARCHIVE_OK); +} + +static void +_checksum_init(struct chksumwork *sumwrk, int sum_alg) +{ + sumwrk->alg = sum_alg; + switch (sum_alg) { + case CKSUM_NONE: + break; + case CKSUM_SHA1: + archive_sha1_init(&(sumwrk->sha1ctx)); + break; + case CKSUM_MD5: + archive_md5_init(&(sumwrk->md5ctx)); + break; + } +} + +static void +_checksum_update(struct chksumwork *sumwrk, const void *buff, size_t size) +{ + + switch (sumwrk->alg) { + case CKSUM_NONE: + break; + case CKSUM_SHA1: + archive_sha1_update(&(sumwrk->sha1ctx), buff, size); + break; + case CKSUM_MD5: + archive_md5_update(&(sumwrk->md5ctx), buff, size); + break; + } +} + +static int +_checksum_final(struct chksumwork *sumwrk, const void *val, size_t len) +{ + unsigned char sum[MAX_SUM_SIZE]; + int r = ARCHIVE_OK; + + switch (sumwrk->alg) { + case CKSUM_NONE: + break; + case CKSUM_SHA1: + archive_sha1_final(&(sumwrk->sha1ctx), sum); + if (len != SHA1_SIZE || + memcmp(val, sum, SHA1_SIZE) != 0) + r = ARCHIVE_FAILED; + break; + case CKSUM_MD5: + archive_md5_final(&(sumwrk->md5ctx), sum); + if (len != MD5_SIZE || + memcmp(val, sum, MD5_SIZE) != 0) + r = ARCHIVE_FAILED; + break; + } + return (r); +} + +static void +checksum_init(struct archive_read *a, int a_sum_alg, int e_sum_alg) +{ + struct xar *xar; + + xar = (struct xar *)(a->format->data); + _checksum_init(&(xar->a_sumwrk), a_sum_alg); + _checksum_init(&(xar->e_sumwrk), e_sum_alg); +} + +static void +checksum_update(struct archive_read *a, const void *abuff, size_t asize, + const void *ebuff, size_t esize) +{ + struct xar *xar; + + xar = (struct xar *)(a->format->data); + _checksum_update(&(xar->a_sumwrk), abuff, asize); + _checksum_update(&(xar->e_sumwrk), ebuff, esize); +} + +static int +checksum_final(struct archive_read *a, const void *a_sum_val, + size_t a_sum_len, const void *e_sum_val, size_t e_sum_len) +{ + struct xar *xar; + int r; + + xar = (struct xar *)(a->format->data); + r = _checksum_final(&(xar->a_sumwrk), a_sum_val, a_sum_len); + if (r == ARCHIVE_OK) + r = _checksum_final(&(xar->e_sumwrk), e_sum_val, e_sum_len); + if (r != ARCHIVE_OK) + archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, + "Sumcheck error"); + return (r); +} + +static int +decompression_init(struct archive_read *a, enum enctype encoding) +{ + struct xar *xar; + const char *detail; + int r; + + xar = (struct xar *)(a->format->data); + xar->rd_encoding = encoding; + switch (encoding) { + case NONE: + break; + case GZIP: + if (xar->stream_valid) + r = inflateReset(&(xar->stream)); + else + r = inflateInit(&(xar->stream)); + if (r != Z_OK) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Couldn't initialize zlib stream."); + return (ARCHIVE_FATAL); + } + xar->stream_valid = 1; + xar->stream.total_in = 0; + xar->stream.total_out = 0; + break; +#if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) + case BZIP2: + if (xar->bzstream_valid) { + BZ2_bzDecompressEnd(&(xar->bzstream)); + xar->bzstream_valid = 0; + } + r = BZ2_bzDecompressInit(&(xar->bzstream), 0, 0); + if (r == BZ_MEM_ERROR) + r = BZ2_bzDecompressInit(&(xar->bzstream), 0, 1); + if (r != BZ_OK) { + int err = ARCHIVE_ERRNO_MISC; + detail = NULL; + switch (r) { + case BZ_PARAM_ERROR: + detail = "invalid setup parameter"; + break; + case BZ_MEM_ERROR: + err = ENOMEM; + detail = "out of memory"; + break; + case BZ_CONFIG_ERROR: + detail = "mis-compiled library"; + break; + } + archive_set_error(&a->archive, err, + "Internal error initializing decompressor: %s", + detail == NULL ? "??" : detail); + xar->bzstream_valid = 0; + return (ARCHIVE_FATAL); + } + xar->bzstream_valid = 1; + xar->bzstream.total_in_lo32 = 0; + xar->bzstream.total_in_hi32 = 0; + xar->bzstream.total_out_lo32 = 0; + xar->bzstream.total_out_hi32 = 0; + break; +#endif +#if defined(HAVE_LZMA_H) && defined(HAVE_LIBLZMA) +#if LZMA_VERSION_MAJOR >= 5 +/* Effectively disable the limiter. */ +#define LZMA_MEMLIMIT UINT64_MAX +#else +/* NOTE: This needs to check memory size which running system has. */ +#define LZMA_MEMLIMIT (1U << 30) +#endif + case XZ: + case LZMA: + if (xar->lzstream_valid) { + lzma_end(&(xar->lzstream)); + xar->lzstream_valid = 0; + } + if (xar->entry_encoding == XZ) + r = lzma_stream_decoder(&(xar->lzstream), + LZMA_MEMLIMIT,/* memlimit */ + LZMA_CONCATENATED); + else + r = lzma_alone_decoder(&(xar->lzstream), + LZMA_MEMLIMIT);/* memlimit */ + if (r != LZMA_OK) { + switch (r) { + case LZMA_MEM_ERROR: + archive_set_error(&a->archive, + ENOMEM, + "Internal error initializing " + "compression library: " + "Cannot allocate memory"); + break; + case LZMA_OPTIONS_ERROR: + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Internal error initializing " + "compression library: " + "Invalid or unsupported options"); + break; + default: + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Internal error initializing " + "lzma library"); + break; + } + return (ARCHIVE_FATAL); + } + xar->lzstream_valid = 1; + xar->lzstream.total_in = 0; + xar->lzstream.total_out = 0; + break; +#endif + /* + * Unsupported compression. + */ + default: +#if !defined(HAVE_BZLIB_H) || !defined(BZ_CONFIG_ERROR) + case BZIP2: +#endif +#if !defined(HAVE_LZMA_H) || !defined(HAVE_LIBLZMA) + case LZMA: + case XZ: +#endif + switch (xar->entry_encoding) { + case BZIP2: detail = "bzip2"; break; + case LZMA: detail = "lzma"; break; + case XZ: detail = "xz"; break; + default: detail = "??"; break; + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "%s compression not supported on this platform", + detail); + return (ARCHIVE_FAILED); + } + return (ARCHIVE_OK); +} + +static int +decompress(struct archive_read *a, const void **buff, size_t *outbytes, + const void *b, size_t *used) +{ + struct xar *xar; + void *outbuff; + size_t avail_in, avail_out; + int r; + + xar = (struct xar *)(a->format->data); + avail_in = *used; + outbuff = (void *)(uintptr_t)*buff; + if (outbuff == NULL) { + if (xar->outbuff == NULL) { + xar->outbuff = malloc(OUTBUFF_SIZE); + if (xar->outbuff == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Couldn't allocate memory for out buffer"); + return (ARCHIVE_FATAL); + } + } + outbuff = xar->outbuff; + *buff = outbuff; + avail_out = OUTBUFF_SIZE; + } else + avail_out = *outbytes; + switch (xar->rd_encoding) { + case GZIP: + xar->stream.next_in = (Bytef *)(uintptr_t)b; + xar->stream.avail_in = avail_in; + xar->stream.next_out = (unsigned char *)outbuff; + xar->stream.avail_out = avail_out; + r = inflate(&(xar->stream), 0); + switch (r) { + case Z_OK: /* Decompressor made some progress.*/ + case Z_STREAM_END: /* Found end of stream. */ + break; + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "File decompression failed (%d)", r); + return (ARCHIVE_FATAL); + } + *used = avail_in - xar->stream.avail_in; + *outbytes = avail_out - xar->stream.avail_out; + break; +#if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) + case BZIP2: + xar->bzstream.next_in = (char *)(uintptr_t)b; + xar->bzstream.avail_in = avail_in; + xar->bzstream.next_out = (char *)outbuff; + xar->bzstream.avail_out = avail_out; + r = BZ2_bzDecompress(&(xar->bzstream)); + switch (r) { + case BZ_STREAM_END: /* Found end of stream. */ + switch (BZ2_bzDecompressEnd(&(xar->bzstream))) { + case BZ_OK: + break; + default: + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "Failed to clean up decompressor"); + return (ARCHIVE_FATAL); + } + xar->bzstream_valid = 0; + /* FALLTHROUGH */ + case BZ_OK: /* Decompressor made some progress. */ + break; + default: + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "bzip decompression failed"); + return (ARCHIVE_FATAL); + } + *used = avail_in - xar->bzstream.avail_in; + *outbytes = avail_out - xar->bzstream.avail_out; + break; +#endif +#if defined(HAVE_LZMA_H) && defined(HAVE_LIBLZMA) + case LZMA: + case XZ: + xar->lzstream.next_in = b; + xar->lzstream.avail_in = avail_in; + xar->lzstream.next_out = (unsigned char *)outbuff; + xar->lzstream.avail_out = avail_out; + r = lzma_code(&(xar->lzstream), LZMA_RUN); + switch (r) { + case LZMA_STREAM_END: /* Found end of stream. */ + lzma_end(&(xar->lzstream)); + xar->lzstream_valid = 0; + /* FALLTHROUGH */ + case LZMA_OK: /* Decompressor made some progress. */ + break; + default: + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "%s decompression failed(%d)", + (xar->entry_encoding == XZ)?"xz":"lzma", + r); + return (ARCHIVE_FATAL); + } + *used = avail_in - xar->lzstream.avail_in; + *outbytes = avail_out - xar->lzstream.avail_out; + break; +#endif +#if !defined(HAVE_BZLIB_H) || !defined(BZ_CONFIG_ERROR) + case BZIP2: +#endif +#if !defined(HAVE_LZMA_H) || !defined(HAVE_LIBLZMA) + case LZMA: + case XZ: +#endif + case NONE: + default: + if (outbuff == xar->outbuff) { + *buff = b; + *used = avail_in; + *outbytes = avail_in; + } else { + if (avail_out > avail_in) + avail_out = avail_in; + memcpy(outbuff, b, avail_out); + *used = avail_out; + *outbytes = avail_out; + } + break; + } + return (ARCHIVE_OK); +} + +static int +decompression_cleanup(struct archive_read *a) +{ + struct xar *xar; + int r; + + xar = (struct xar *)(a->format->data); + r = ARCHIVE_OK; + if (xar->stream_valid) { + if (inflateEnd(&(xar->stream)) != Z_OK) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Failed to clean up zlib decompressor"); + r = ARCHIVE_FATAL; + } + } +#if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) + if (xar->bzstream_valid) { + if (BZ2_bzDecompressEnd(&(xar->bzstream)) != BZ_OK) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Failed to clean up bzip2 decompressor"); + r = ARCHIVE_FATAL; + } + } +#endif +#if defined(HAVE_LZMA_H) && defined(HAVE_LIBLZMA) + if (xar->lzstream_valid) + lzma_end(&(xar->lzstream)); +#elif defined(HAVE_LZMA_H) && defined(HAVE_LIBLZMA) + if (xar->lzstream_valid) { + if (lzmadec_end(&(xar->lzstream)) != LZMADEC_OK) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Failed to clean up lzmadec decompressor"); + r = ARCHIVE_FATAL; + } + } +#endif + return (r); +} + +static void +checksum_cleanup(struct archive_read *a) { + struct xar *xar; + + xar = (struct xar *)(a->format->data); + + _checksum_final(&(xar->a_sumwrk), NULL, 0); + _checksum_final(&(xar->e_sumwrk), NULL, 0); +} + +static void +xmlattr_cleanup(struct xmlattr_list *list) +{ + struct xmlattr *attr, *next; + + attr = list->first; + while (attr != NULL) { + next = attr->next; + free(attr->name); + free(attr->value); + free(attr); + attr = next; + } + list->first = NULL; + list->last = &(list->first); +} + +static int +file_new(struct archive_read *a, struct xar *xar, struct xmlattr_list *list) +{ + struct xar_file *file; + struct xmlattr *attr; + + file = calloc(1, sizeof(*file)); + if (file == NULL) { + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + file->parent = xar->file; + file->mode = 0777 | AE_IFREG; + file->atime = 0; + file->mtime = 0; + xar->file = file; + xar->xattr = NULL; + for (attr = list->first; attr != NULL; attr = attr->next) { + if (strcmp(attr->name, "id") == 0) + file->id = atol10(attr->value, strlen(attr->value)); + } + file->nlink = 1; + if (heap_add_entry(a, &(xar->file_queue), file) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + return (ARCHIVE_OK); +} + +static void +file_free(struct xar_file *file) +{ + struct xattr *xattr; + + archive_string_free(&(file->pathname)); + archive_string_free(&(file->symlink)); + archive_string_free(&(file->uname)); + archive_string_free(&(file->gname)); + archive_string_free(&(file->hardlink)); + xattr = file->xattr_list; + while (xattr != NULL) { + struct xattr *next; + + next = xattr->next; + xattr_free(xattr); + xattr = next; + } + + free(file); +} + +static int +xattr_new(struct archive_read *a, struct xar *xar, struct xmlattr_list *list) +{ + struct xattr *xattr, **nx; + struct xmlattr *attr; + + xattr = calloc(1, sizeof(*xattr)); + if (xattr == NULL) { + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + xar->xattr = xattr; + for (attr = list->first; attr != NULL; attr = attr->next) { + if (strcmp(attr->name, "id") == 0) + xattr->id = atol10(attr->value, strlen(attr->value)); + } + /* Chain to xattr list. */ + for (nx = &(xar->file->xattr_list); + *nx != NULL; nx = &((*nx)->next)) { + if (xattr->id < (*nx)->id) + break; + } + xattr->next = *nx; + *nx = xattr; + + return (ARCHIVE_OK); +} + +static void +xattr_free(struct xattr *xattr) +{ + archive_string_free(&(xattr->name)); + free(xattr); +} + +static int +getencoding(struct xmlattr_list *list) +{ + struct xmlattr *attr; + enum enctype encoding = NONE; + + for (attr = list->first; attr != NULL; attr = attr->next) { + if (strcmp(attr->name, "style") == 0) { + if (strcmp(attr->value, "application/octet-stream") == 0) + encoding = NONE; + else if (strcmp(attr->value, "application/x-gzip") == 0) + encoding = GZIP; + else if (strcmp(attr->value, "application/x-bzip2") == 0) + encoding = BZIP2; + else if (strcmp(attr->value, "application/x-lzma") == 0) + encoding = LZMA; + else if (strcmp(attr->value, "application/x-xz") == 0) + encoding = XZ; + } + } + return (encoding); +} + +static int +getsumalgorithm(struct xmlattr_list *list) +{ + struct xmlattr *attr; + int alg = CKSUM_NONE; + + for (attr = list->first; attr != NULL; attr = attr->next) { + if (strcmp(attr->name, "style") == 0) { + const char *v = attr->value; + if ((v[0] == 'S' || v[0] == 's') && + (v[1] == 'H' || v[1] == 'h') && + (v[2] == 'A' || v[2] == 'a') && + v[3] == '1' && v[4] == '\0') + alg = CKSUM_SHA1; + if ((v[0] == 'M' || v[0] == 'm') && + (v[1] == 'D' || v[1] == 'd') && + v[2] == '5' && v[3] == '\0') + alg = CKSUM_MD5; + } + } + return (alg); +} + +static int +unknowntag_start(struct archive_read *a, struct xar *xar, const char *name) +{ + struct unknown_tag *tag; + + tag = malloc(sizeof(*tag)); + if (tag == NULL) { + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + tag->next = xar->unknowntags; + archive_string_init(&(tag->name)); + archive_strcpy(&(tag->name), name); + if (xar->unknowntags == NULL) { +#if DEBUG + fprintf(stderr, "UNKNOWNTAG_START:%s\n", name); +#endif + xar->xmlsts_unknown = xar->xmlsts; + xar->xmlsts = UNKNOWN; + } + xar->unknowntags = tag; + return (ARCHIVE_OK); +} + +static void +unknowntag_end(struct xar *xar, const char *name) +{ + struct unknown_tag *tag; + + tag = xar->unknowntags; + if (tag == NULL || name == NULL) + return; + if (strcmp(tag->name.s, name) == 0) { + xar->unknowntags = tag->next; + archive_string_free(&(tag->name)); + free(tag); + if (xar->unknowntags == NULL) { +#if DEBUG + fprintf(stderr, "UNKNOWNTAG_END:%s\n", name); +#endif + xar->xmlsts = xar->xmlsts_unknown; + } + } +} + +static int +xml_start(struct archive_read *a, const char *name, struct xmlattr_list *list) +{ + struct xar *xar; + struct xmlattr *attr; + + xar = (struct xar *)(a->format->data); + +#if DEBUG + fprintf(stderr, "xml_sta:[%s]\n", name); + for (attr = list->first; attr != NULL; attr = attr->next) + fprintf(stderr, " attr:\"%s\"=\"%s\"\n", + attr->name, attr->value); +#endif + xar->base64text = 0; + switch (xar->xmlsts) { + case INIT: + if (strcmp(name, "xar") == 0) + xar->xmlsts = XAR; + else + if (unknowntag_start(a, xar, name) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + break; + case XAR: + if (strcmp(name, "toc") == 0) + xar->xmlsts = TOC; + else + if (unknowntag_start(a, xar, name) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + break; + case TOC: + if (strcmp(name, "creation-time") == 0) + xar->xmlsts = TOC_CREATION_TIME; + else if (strcmp(name, "checksum") == 0) + xar->xmlsts = TOC_CHECKSUM; + else if (strcmp(name, "file") == 0) { + if (file_new(a, xar, list) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + xar->xmlsts = TOC_FILE; + } + else + if (unknowntag_start(a, xar, name) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + break; + case TOC_CHECKSUM: + if (strcmp(name, "offset") == 0) + xar->xmlsts = TOC_CHECKSUM_OFFSET; + else if (strcmp(name, "size") == 0) + xar->xmlsts = TOC_CHECKSUM_SIZE; + else + if (unknowntag_start(a, xar, name) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + break; + case TOC_FILE: + if (strcmp(name, "file") == 0) { + if (file_new(a, xar, list) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + else if (strcmp(name, "data") == 0) + xar->xmlsts = FILE_DATA; + else if (strcmp(name, "ea") == 0) { + if (xattr_new(a, xar, list) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + xar->xmlsts = FILE_EA; + } + else if (strcmp(name, "ctime") == 0) + xar->xmlsts = FILE_CTIME; + else if (strcmp(name, "mtime") == 0) + xar->xmlsts = FILE_MTIME; + else if (strcmp(name, "atime") == 0) + xar->xmlsts = FILE_ATIME; + else if (strcmp(name, "group") == 0) + xar->xmlsts = FILE_GROUP; + else if (strcmp(name, "gid") == 0) + xar->xmlsts = FILE_GID; + else if (strcmp(name, "user") == 0) + xar->xmlsts = FILE_USER; + else if (strcmp(name, "uid") == 0) + xar->xmlsts = FILE_UID; + else if (strcmp(name, "mode") == 0) + xar->xmlsts = FILE_MODE; + else if (strcmp(name, "device") == 0) + xar->xmlsts = FILE_DEVICE; + else if (strcmp(name, "deviceno") == 0) + xar->xmlsts = FILE_DEVICENO; + else if (strcmp(name, "inode") == 0) + xar->xmlsts = FILE_INODE; + else if (strcmp(name, "link") == 0) + xar->xmlsts = FILE_LINK; + else if (strcmp(name, "type") == 0) { + xar->xmlsts = FILE_TYPE; + for (attr = list->first; attr != NULL; + attr = attr->next) { + if (strcmp(attr->name, "link") != 0) + continue; + if (strcmp(attr->value, "original") == 0) { + xar->file->hdnext = xar->hdlink_orgs; + xar->hdlink_orgs = xar->file; + } else { + xar->file->link = (unsigned)atol10(attr->value, + strlen(attr->value)); + if (xar->file->link > 0) + if (add_link(a, xar, xar->file) != ARCHIVE_OK) { + return (ARCHIVE_FATAL); + }; + } + } + } + else if (strcmp(name, "name") == 0) { + xar->xmlsts = FILE_NAME; + for (attr = list->first; attr != NULL; + attr = attr->next) { + if (strcmp(attr->name, "enctype") == 0 && + strcmp(attr->value, "base64") == 0) + xar->base64text = 1; + } + } + else if (strcmp(name, "acl") == 0) + xar->xmlsts = FILE_ACL; + else if (strcmp(name, "flags") == 0) + xar->xmlsts = FILE_FLAGS; + else if (strcmp(name, "ext2") == 0) + xar->xmlsts = FILE_EXT2; + else + if (unknowntag_start(a, xar, name) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + break; + case FILE_DATA: + if (strcmp(name, "length") == 0) + xar->xmlsts = FILE_DATA_LENGTH; + else if (strcmp(name, "offset") == 0) + xar->xmlsts = FILE_DATA_OFFSET; + else if (strcmp(name, "size") == 0) + xar->xmlsts = FILE_DATA_SIZE; + else if (strcmp(name, "encoding") == 0) { + xar->xmlsts = FILE_DATA_ENCODING; + xar->file->encoding = getencoding(list); + } + else if (strcmp(name, "archived-checksum") == 0) { + xar->xmlsts = FILE_DATA_A_CHECKSUM; + xar->file->a_sum.alg = getsumalgorithm(list); + } + else if (strcmp(name, "extracted-checksum") == 0) { + xar->xmlsts = FILE_DATA_E_CHECKSUM; + xar->file->e_sum.alg = getsumalgorithm(list); + } + else if (strcmp(name, "content") == 0) + xar->xmlsts = FILE_DATA_CONTENT; + else + if (unknowntag_start(a, xar, name) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + break; + case FILE_DEVICE: + if (strcmp(name, "major") == 0) + xar->xmlsts = FILE_DEVICE_MAJOR; + else if (strcmp(name, "minor") == 0) + xar->xmlsts = FILE_DEVICE_MINOR; + else + if (unknowntag_start(a, xar, name) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + break; + case FILE_DATA_CONTENT: + if (unknowntag_start(a, xar, name) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + break; + case FILE_EA: + if (strcmp(name, "length") == 0) + xar->xmlsts = FILE_EA_LENGTH; + else if (strcmp(name, "offset") == 0) + xar->xmlsts = FILE_EA_OFFSET; + else if (strcmp(name, "size") == 0) + xar->xmlsts = FILE_EA_SIZE; + else if (strcmp(name, "encoding") == 0) { + xar->xmlsts = FILE_EA_ENCODING; + xar->xattr->encoding = getencoding(list); + } else if (strcmp(name, "archived-checksum") == 0) + xar->xmlsts = FILE_EA_A_CHECKSUM; + else if (strcmp(name, "extracted-checksum") == 0) + xar->xmlsts = FILE_EA_E_CHECKSUM; + else if (strcmp(name, "name") == 0) + xar->xmlsts = FILE_EA_NAME; + else if (strcmp(name, "fstype") == 0) + xar->xmlsts = FILE_EA_FSTYPE; + else + if (unknowntag_start(a, xar, name) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + break; + case FILE_ACL: + if (strcmp(name, "appleextended") == 0) + xar->xmlsts = FILE_ACL_APPLEEXTENDED; + else if (strcmp(name, "default") == 0) + xar->xmlsts = FILE_ACL_DEFAULT; + else if (strcmp(name, "access") == 0) + xar->xmlsts = FILE_ACL_ACCESS; + else + if (unknowntag_start(a, xar, name) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + break; + case FILE_FLAGS: + if (!xml_parse_file_flags(xar, name)) + if (unknowntag_start(a, xar, name) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + break; + case FILE_EXT2: + if (!xml_parse_file_ext2(xar, name)) + if (unknowntag_start(a, xar, name) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + break; + case TOC_CREATION_TIME: + case TOC_CHECKSUM_OFFSET: + case TOC_CHECKSUM_SIZE: + case FILE_DATA_LENGTH: + case FILE_DATA_OFFSET: + case FILE_DATA_SIZE: + case FILE_DATA_ENCODING: + case FILE_DATA_A_CHECKSUM: + case FILE_DATA_E_CHECKSUM: + case FILE_EA_LENGTH: + case FILE_EA_OFFSET: + case FILE_EA_SIZE: + case FILE_EA_ENCODING: + case FILE_EA_A_CHECKSUM: + case FILE_EA_E_CHECKSUM: + case FILE_EA_NAME: + case FILE_EA_FSTYPE: + case FILE_CTIME: + case FILE_MTIME: + case FILE_ATIME: + case FILE_GROUP: + case FILE_GID: + case FILE_USER: + case FILE_UID: + case FILE_INODE: + case FILE_DEVICE_MAJOR: + case FILE_DEVICE_MINOR: + case FILE_DEVICENO: + case FILE_MODE: + case FILE_TYPE: + case FILE_LINK: + case FILE_NAME: + case FILE_ACL_DEFAULT: + case FILE_ACL_ACCESS: + case FILE_ACL_APPLEEXTENDED: + case FILE_FLAGS_USER_NODUMP: + case FILE_FLAGS_USER_IMMUTABLE: + case FILE_FLAGS_USER_APPEND: + case FILE_FLAGS_USER_OPAQUE: + case FILE_FLAGS_USER_NOUNLINK: + case FILE_FLAGS_SYS_ARCHIVED: + case FILE_FLAGS_SYS_IMMUTABLE: + case FILE_FLAGS_SYS_APPEND: + case FILE_FLAGS_SYS_NOUNLINK: + case FILE_FLAGS_SYS_SNAPSHOT: + case FILE_EXT2_SecureDeletion: + case FILE_EXT2_Undelete: + case FILE_EXT2_Compress: + case FILE_EXT2_Synchronous: + case FILE_EXT2_Immutable: + case FILE_EXT2_AppendOnly: + case FILE_EXT2_NoDump: + case FILE_EXT2_NoAtime: + case FILE_EXT2_CompDirty: + case FILE_EXT2_CompBlock: + case FILE_EXT2_NoCompBlock: + case FILE_EXT2_CompError: + case FILE_EXT2_BTree: + case FILE_EXT2_HashIndexed: + case FILE_EXT2_iMagic: + case FILE_EXT2_Journaled: + case FILE_EXT2_NoTail: + case FILE_EXT2_DirSync: + case FILE_EXT2_TopDir: + case FILE_EXT2_Reserved: + case UNKNOWN: + if (unknowntag_start(a, xar, name) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + break; + } + return (ARCHIVE_OK); +} + +static void +xml_end(void *userData, const char *name) +{ + struct archive_read *a; + struct xar *xar; + + a = (struct archive_read *)userData; + xar = (struct xar *)(a->format->data); + +#if DEBUG + fprintf(stderr, "xml_end:[%s]\n", name); +#endif + switch (xar->xmlsts) { + case INIT: + break; + case XAR: + if (strcmp(name, "xar") == 0) + xar->xmlsts = INIT; + break; + case TOC: + if (strcmp(name, "toc") == 0) + xar->xmlsts = XAR; + break; + case TOC_CREATION_TIME: + if (strcmp(name, "creation-time") == 0) + xar->xmlsts = TOC; + break; + case TOC_CHECKSUM: + if (strcmp(name, "checksum") == 0) + xar->xmlsts = TOC; + break; + case TOC_CHECKSUM_OFFSET: + if (strcmp(name, "offset") == 0) + xar->xmlsts = TOC_CHECKSUM; + break; + case TOC_CHECKSUM_SIZE: + if (strcmp(name, "size") == 0) + xar->xmlsts = TOC_CHECKSUM; + break; + case TOC_FILE: + if (strcmp(name, "file") == 0) { + if (xar->file->parent != NULL && + ((xar->file->mode & AE_IFMT) == AE_IFDIR)) + xar->file->parent->subdirs++; + xar->file = xar->file->parent; + if (xar->file == NULL) + xar->xmlsts = TOC; + } + break; + case FILE_DATA: + if (strcmp(name, "data") == 0) + xar->xmlsts = TOC_FILE; + break; + case FILE_DATA_LENGTH: + if (strcmp(name, "length") == 0) + xar->xmlsts = FILE_DATA; + break; + case FILE_DATA_OFFSET: + if (strcmp(name, "offset") == 0) + xar->xmlsts = FILE_DATA; + break; + case FILE_DATA_SIZE: + if (strcmp(name, "size") == 0) + xar->xmlsts = FILE_DATA; + break; + case FILE_DATA_ENCODING: + if (strcmp(name, "encoding") == 0) + xar->xmlsts = FILE_DATA; + break; + case FILE_DATA_A_CHECKSUM: + if (strcmp(name, "archived-checksum") == 0) + xar->xmlsts = FILE_DATA; + break; + case FILE_DATA_E_CHECKSUM: + if (strcmp(name, "extracted-checksum") == 0) + xar->xmlsts = FILE_DATA; + break; + case FILE_DATA_CONTENT: + if (strcmp(name, "content") == 0) + xar->xmlsts = FILE_DATA; + break; + case FILE_EA: + if (strcmp(name, "ea") == 0) { + xar->xmlsts = TOC_FILE; + xar->xattr = NULL; + } + break; + case FILE_EA_LENGTH: + if (strcmp(name, "length") == 0) + xar->xmlsts = FILE_EA; + break; + case FILE_EA_OFFSET: + if (strcmp(name, "offset") == 0) + xar->xmlsts = FILE_EA; + break; + case FILE_EA_SIZE: + if (strcmp(name, "size") == 0) + xar->xmlsts = FILE_EA; + break; + case FILE_EA_ENCODING: + if (strcmp(name, "encoding") == 0) + xar->xmlsts = FILE_EA; + break; + case FILE_EA_A_CHECKSUM: + if (strcmp(name, "archived-checksum") == 0) + xar->xmlsts = FILE_EA; + break; + case FILE_EA_E_CHECKSUM: + if (strcmp(name, "extracted-checksum") == 0) + xar->xmlsts = FILE_EA; + break; + case FILE_EA_NAME: + if (strcmp(name, "name") == 0) + xar->xmlsts = FILE_EA; + break; + case FILE_EA_FSTYPE: + if (strcmp(name, "fstype") == 0) + xar->xmlsts = FILE_EA; + break; + case FILE_CTIME: + if (strcmp(name, "ctime") == 0) + xar->xmlsts = TOC_FILE; + break; + case FILE_MTIME: + if (strcmp(name, "mtime") == 0) + xar->xmlsts = TOC_FILE; + break; + case FILE_ATIME: + if (strcmp(name, "atime") == 0) + xar->xmlsts = TOC_FILE; + break; + case FILE_GROUP: + if (strcmp(name, "group") == 0) + xar->xmlsts = TOC_FILE; + break; + case FILE_GID: + if (strcmp(name, "gid") == 0) + xar->xmlsts = TOC_FILE; + break; + case FILE_USER: + if (strcmp(name, "user") == 0) + xar->xmlsts = TOC_FILE; + break; + case FILE_UID: + if (strcmp(name, "uid") == 0) + xar->xmlsts = TOC_FILE; + break; + case FILE_MODE: + if (strcmp(name, "mode") == 0) + xar->xmlsts = TOC_FILE; + break; + case FILE_DEVICE: + if (strcmp(name, "device") == 0) + xar->xmlsts = TOC_FILE; + break; + case FILE_DEVICE_MAJOR: + if (strcmp(name, "major") == 0) + xar->xmlsts = FILE_DEVICE; + break; + case FILE_DEVICE_MINOR: + if (strcmp(name, "minor") == 0) + xar->xmlsts = FILE_DEVICE; + break; + case FILE_DEVICENO: + if (strcmp(name, "deviceno") == 0) + xar->xmlsts = TOC_FILE; + break; + case FILE_INODE: + if (strcmp(name, "inode") == 0) + xar->xmlsts = TOC_FILE; + break; + case FILE_LINK: + if (strcmp(name, "link") == 0) + xar->xmlsts = TOC_FILE; + break; + case FILE_TYPE: + if (strcmp(name, "type") == 0) + xar->xmlsts = TOC_FILE; + break; + case FILE_NAME: + if (strcmp(name, "name") == 0) + xar->xmlsts = TOC_FILE; + break; + case FILE_ACL: + if (strcmp(name, "acl") == 0) + xar->xmlsts = TOC_FILE; + break; + case FILE_ACL_DEFAULT: + if (strcmp(name, "default") == 0) + xar->xmlsts = FILE_ACL; + break; + case FILE_ACL_ACCESS: + if (strcmp(name, "access") == 0) + xar->xmlsts = FILE_ACL; + break; + case FILE_ACL_APPLEEXTENDED: + if (strcmp(name, "appleextended") == 0) + xar->xmlsts = FILE_ACL; + break; + case FILE_FLAGS: + if (strcmp(name, "flags") == 0) + xar->xmlsts = TOC_FILE; + break; + case FILE_FLAGS_USER_NODUMP: + if (strcmp(name, "UserNoDump") == 0) + xar->xmlsts = FILE_FLAGS; + break; + case FILE_FLAGS_USER_IMMUTABLE: + if (strcmp(name, "UserImmutable") == 0) + xar->xmlsts = FILE_FLAGS; + break; + case FILE_FLAGS_USER_APPEND: + if (strcmp(name, "UserAppend") == 0) + xar->xmlsts = FILE_FLAGS; + break; + case FILE_FLAGS_USER_OPAQUE: + if (strcmp(name, "UserOpaque") == 0) + xar->xmlsts = FILE_FLAGS; + break; + case FILE_FLAGS_USER_NOUNLINK: + if (strcmp(name, "UserNoUnlink") == 0) + xar->xmlsts = FILE_FLAGS; + break; + case FILE_FLAGS_SYS_ARCHIVED: + if (strcmp(name, "SystemArchived") == 0) + xar->xmlsts = FILE_FLAGS; + break; + case FILE_FLAGS_SYS_IMMUTABLE: + if (strcmp(name, "SystemImmutable") == 0) + xar->xmlsts = FILE_FLAGS; + break; + case FILE_FLAGS_SYS_APPEND: + if (strcmp(name, "SystemAppend") == 0) + xar->xmlsts = FILE_FLAGS; + break; + case FILE_FLAGS_SYS_NOUNLINK: + if (strcmp(name, "SystemNoUnlink") == 0) + xar->xmlsts = FILE_FLAGS; + break; + case FILE_FLAGS_SYS_SNAPSHOT: + if (strcmp(name, "SystemSnapshot") == 0) + xar->xmlsts = FILE_FLAGS; + break; + case FILE_EXT2: + if (strcmp(name, "ext2") == 0) + xar->xmlsts = TOC_FILE; + break; + case FILE_EXT2_SecureDeletion: + if (strcmp(name, "SecureDeletion") == 0) + xar->xmlsts = FILE_EXT2; + break; + case FILE_EXT2_Undelete: + if (strcmp(name, "Undelete") == 0) + xar->xmlsts = FILE_EXT2; + break; + case FILE_EXT2_Compress: + if (strcmp(name, "Compress") == 0) + xar->xmlsts = FILE_EXT2; + break; + case FILE_EXT2_Synchronous: + if (strcmp(name, "Synchronous") == 0) + xar->xmlsts = FILE_EXT2; + break; + case FILE_EXT2_Immutable: + if (strcmp(name, "Immutable") == 0) + xar->xmlsts = FILE_EXT2; + break; + case FILE_EXT2_AppendOnly: + if (strcmp(name, "AppendOnly") == 0) + xar->xmlsts = FILE_EXT2; + break; + case FILE_EXT2_NoDump: + if (strcmp(name, "NoDump") == 0) + xar->xmlsts = FILE_EXT2; + break; + case FILE_EXT2_NoAtime: + if (strcmp(name, "NoAtime") == 0) + xar->xmlsts = FILE_EXT2; + break; + case FILE_EXT2_CompDirty: + if (strcmp(name, "CompDirty") == 0) + xar->xmlsts = FILE_EXT2; + break; + case FILE_EXT2_CompBlock: + if (strcmp(name, "CompBlock") == 0) + xar->xmlsts = FILE_EXT2; + break; + case FILE_EXT2_NoCompBlock: + if (strcmp(name, "NoCompBlock") == 0) + xar->xmlsts = FILE_EXT2; + break; + case FILE_EXT2_CompError: + if (strcmp(name, "CompError") == 0) + xar->xmlsts = FILE_EXT2; + break; + case FILE_EXT2_BTree: + if (strcmp(name, "BTree") == 0) + xar->xmlsts = FILE_EXT2; + break; + case FILE_EXT2_HashIndexed: + if (strcmp(name, "HashIndexed") == 0) + xar->xmlsts = FILE_EXT2; + break; + case FILE_EXT2_iMagic: + if (strcmp(name, "iMagic") == 0) + xar->xmlsts = FILE_EXT2; + break; + case FILE_EXT2_Journaled: + if (strcmp(name, "Journaled") == 0) + xar->xmlsts = FILE_EXT2; + break; + case FILE_EXT2_NoTail: + if (strcmp(name, "NoTail") == 0) + xar->xmlsts = FILE_EXT2; + break; + case FILE_EXT2_DirSync: + if (strcmp(name, "DirSync") == 0) + xar->xmlsts = FILE_EXT2; + break; + case FILE_EXT2_TopDir: + if (strcmp(name, "TopDir") == 0) + xar->xmlsts = FILE_EXT2; + break; + case FILE_EXT2_Reserved: + if (strcmp(name, "Reserved") == 0) + xar->xmlsts = FILE_EXT2; + break; + case UNKNOWN: + unknowntag_end(xar, name); + break; + } +} + +static const int base64[256] = { + -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, /* 00 - 0F */ + -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, /* 10 - 1F */ + -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, 62, -1, -1, -1, 63, /* 20 - 2F */ + 52, 53, 54, 55, 56, 57, 58, 59, + 60, 61, -1, -1, -1, -1, -1, -1, /* 30 - 3F */ + -1, 0, 1, 2, 3, 4, 5, 6, + 7, 8, 9, 10, 11, 12, 13, 14, /* 40 - 4F */ + 15, 16, 17, 18, 19, 20, 21, 22, + 23, 24, 25, -1, -1, -1, -1, -1, /* 50 - 5F */ + -1, 26, 27, 28, 29, 30, 31, 32, + 33, 34, 35, 36, 37, 38, 39, 40, /* 60 - 6F */ + 41, 42, 43, 44, 45, 46, 47, 48, + 49, 50, 51, -1, -1, -1, -1, -1, /* 70 - 7F */ + -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, /* 80 - 8F */ + -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, /* 90 - 9F */ + -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, /* A0 - AF */ + -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, /* B0 - BF */ + -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, /* C0 - CF */ + -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, /* D0 - DF */ + -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, /* E0 - EF */ + -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, /* F0 - FF */ +}; + +static void +strappend_base64(struct xar *xar, + struct archive_string *as, const char *s, size_t l) +{ + unsigned char buff[256]; + unsigned char *out; + const unsigned char *b; + size_t len; + + (void)xar; /* UNUSED */ + len = 0; + out = buff; + b = (const unsigned char *)s; + while (l > 0) { + int n = 0; + + if (base64[b[0]] < 0 || base64[b[1]] < 0) + break; + n = base64[*b++] << 18; + n |= base64[*b++] << 12; + *out++ = n >> 16; + len++; + l -= 2; + + if (l > 0) { + if (base64[*b] < 0) + break; + n |= base64[*b++] << 6; + *out++ = (n >> 8) & 0xFF; + len++; + --l; + } + if (l > 0) { + if (base64[*b] < 0) + break; + n |= base64[*b++]; + *out++ = n & 0xFF; + len++; + --l; + } + if (len+3 >= sizeof(buff)) { + archive_strncat(as, (const char *)buff, len); + len = 0; + out = buff; + } + } + if (len > 0) + archive_strncat(as, (const char *)buff, len); +} + +static int +is_string(const char *known, const char *data, size_t len) +{ + if (strlen(known) != len) + return -1; + return memcmp(data, known, len); +} + +static void +xml_data(void *userData, const char *s, int len) +{ + struct archive_read *a; + struct xar *xar; + + a = (struct archive_read *)userData; + xar = (struct xar *)(a->format->data); + +#if DEBUG + { + char buff[1024]; + if (len > (int)(sizeof(buff)-1)) + len = (int)(sizeof(buff)-1); + strncpy(buff, s, len); + buff[len] = 0; + fprintf(stderr, "\tlen=%d:\"%s\"\n", len, buff); + } +#endif + switch (xar->xmlsts) { + case TOC_CHECKSUM_OFFSET: + xar->toc_chksum_offset = atol10(s, len); + break; + case TOC_CHECKSUM_SIZE: + xar->toc_chksum_size = atol10(s, len); + break; + default: + break; + } + if (xar->file == NULL) + return; + + switch (xar->xmlsts) { + case FILE_NAME: + if (xar->file->parent != NULL) { + archive_string_concat(&(xar->file->pathname), + &(xar->file->parent->pathname)); + archive_strappend_char(&(xar->file->pathname), '/'); + } + xar->file->has |= HAS_PATHNAME; + if (xar->base64text) { + strappend_base64(xar, + &(xar->file->pathname), s, len); + } else + archive_strncat(&(xar->file->pathname), s, len); + break; + case FILE_LINK: + xar->file->has |= HAS_SYMLINK; + archive_strncpy(&(xar->file->symlink), s, len); + break; + case FILE_TYPE: + if (is_string("file", s, len) == 0 || + is_string("hardlink", s, len) == 0) + xar->file->mode = + (xar->file->mode & ~AE_IFMT) | AE_IFREG; + if (is_string("directory", s, len) == 0) + xar->file->mode = + (xar->file->mode & ~AE_IFMT) | AE_IFDIR; + if (is_string("symlink", s, len) == 0) + xar->file->mode = + (xar->file->mode & ~AE_IFMT) | AE_IFLNK; + if (is_string("character special", s, len) == 0) + xar->file->mode = + (xar->file->mode & ~AE_IFMT) | AE_IFCHR; + if (is_string("block special", s, len) == 0) + xar->file->mode = + (xar->file->mode & ~AE_IFMT) | AE_IFBLK; + if (is_string("socket", s, len) == 0) + xar->file->mode = + (xar->file->mode & ~AE_IFMT) | AE_IFSOCK; + if (is_string("fifo", s, len) == 0) + xar->file->mode = + (xar->file->mode & ~AE_IFMT) | AE_IFIFO; + xar->file->has |= HAS_TYPE; + break; + case FILE_INODE: + xar->file->has |= HAS_INO; + xar->file->ino64 = atol10(s, len); + break; + case FILE_DEVICE_MAJOR: + xar->file->has |= HAS_DEVMAJOR; + xar->file->devmajor = (dev_t)atol10(s, len); + break; + case FILE_DEVICE_MINOR: + xar->file->has |= HAS_DEVMINOR; + xar->file->devminor = (dev_t)atol10(s, len); + break; + case FILE_DEVICENO: + xar->file->has |= HAS_DEV; + xar->file->dev = (dev_t)atol10(s, len); + break; + case FILE_MODE: + xar->file->has |= HAS_MODE; + xar->file->mode = + (xar->file->mode & AE_IFMT) | + ((mode_t)(atol8(s, len)) & ~AE_IFMT); + break; + case FILE_GROUP: + xar->file->has |= HAS_GID; + archive_strncpy(&(xar->file->gname), s, len); + break; + case FILE_GID: + xar->file->has |= HAS_GID; + xar->file->gid = atol10(s, len); + break; + case FILE_USER: + xar->file->has |= HAS_UID; + archive_strncpy(&(xar->file->uname), s, len); + break; + case FILE_UID: + xar->file->has |= HAS_UID; + xar->file->uid = atol10(s, len); + break; + case FILE_CTIME: + xar->file->has |= HAS_TIME | HAS_CTIME; + xar->file->ctime = parse_time(s, len); + break; + case FILE_MTIME: + xar->file->has |= HAS_TIME | HAS_MTIME; + xar->file->mtime = parse_time(s, len); + break; + case FILE_ATIME: + xar->file->has |= HAS_TIME | HAS_ATIME; + xar->file->atime = parse_time(s, len); + break; + case FILE_DATA_LENGTH: + xar->file->has |= HAS_DATA; + xar->file->length = atol10(s, len); + break; + case FILE_DATA_OFFSET: + xar->file->has |= HAS_DATA; + xar->file->offset = atol10(s, len); + break; + case FILE_DATA_SIZE: + xar->file->has |= HAS_DATA; + xar->file->size = atol10(s, len); + break; + case FILE_DATA_A_CHECKSUM: + xar->file->a_sum.len = atohex(xar->file->a_sum.val, + sizeof(xar->file->a_sum.val), s, len); + break; + case FILE_DATA_E_CHECKSUM: + xar->file->e_sum.len = atohex(xar->file->e_sum.val, + sizeof(xar->file->e_sum.val), s, len); + break; + case FILE_EA_LENGTH: + xar->file->has |= HAS_XATTR; + xar->xattr->length = atol10(s, len); + break; + case FILE_EA_OFFSET: + xar->file->has |= HAS_XATTR; + xar->xattr->offset = atol10(s, len); + break; + case FILE_EA_SIZE: + xar->file->has |= HAS_XATTR; + xar->xattr->size = atol10(s, len); + break; + case FILE_EA_A_CHECKSUM: + xar->file->has |= HAS_XATTR; + xar->xattr->a_sum.len = atohex(xar->xattr->a_sum.val, + sizeof(xar->xattr->a_sum.val), s, len); + break; + case FILE_EA_E_CHECKSUM: + xar->file->has |= HAS_XATTR; + xar->xattr->e_sum.len = atohex(xar->xattr->e_sum.val, + sizeof(xar->xattr->e_sum.val), s, len); + break; + case FILE_EA_NAME: + xar->file->has |= HAS_XATTR; + archive_strncpy(&(xar->xattr->name), s, len); + break; + case FILE_EA_FSTYPE: + xar->file->has |= HAS_XATTR; + archive_strncpy(&(xar->xattr->fstype), s, len); + break; + break; + case FILE_ACL_DEFAULT: + case FILE_ACL_ACCESS: + case FILE_ACL_APPLEEXTENDED: + xar->file->has |= HAS_ACL; + /* TODO */ + break; + case INIT: + case XAR: + case TOC: + case TOC_CREATION_TIME: + case TOC_CHECKSUM: + case TOC_CHECKSUM_OFFSET: + case TOC_CHECKSUM_SIZE: + case TOC_FILE: + case FILE_DATA: + case FILE_DATA_ENCODING: + case FILE_DATA_CONTENT: + case FILE_DEVICE: + case FILE_EA: + case FILE_EA_ENCODING: + case FILE_ACL: + case FILE_FLAGS: + case FILE_FLAGS_USER_NODUMP: + case FILE_FLAGS_USER_IMMUTABLE: + case FILE_FLAGS_USER_APPEND: + case FILE_FLAGS_USER_OPAQUE: + case FILE_FLAGS_USER_NOUNLINK: + case FILE_FLAGS_SYS_ARCHIVED: + case FILE_FLAGS_SYS_IMMUTABLE: + case FILE_FLAGS_SYS_APPEND: + case FILE_FLAGS_SYS_NOUNLINK: + case FILE_FLAGS_SYS_SNAPSHOT: + case FILE_EXT2: + case FILE_EXT2_SecureDeletion: + case FILE_EXT2_Undelete: + case FILE_EXT2_Compress: + case FILE_EXT2_Synchronous: + case FILE_EXT2_Immutable: + case FILE_EXT2_AppendOnly: + case FILE_EXT2_NoDump: + case FILE_EXT2_NoAtime: + case FILE_EXT2_CompDirty: + case FILE_EXT2_CompBlock: + case FILE_EXT2_NoCompBlock: + case FILE_EXT2_CompError: + case FILE_EXT2_BTree: + case FILE_EXT2_HashIndexed: + case FILE_EXT2_iMagic: + case FILE_EXT2_Journaled: + case FILE_EXT2_NoTail: + case FILE_EXT2_DirSync: + case FILE_EXT2_TopDir: + case FILE_EXT2_Reserved: + case UNKNOWN: + break; + } +} + +/* + * BSD file flags. + */ +static int +xml_parse_file_flags(struct xar *xar, const char *name) +{ + const char *flag = NULL; + + if (strcmp(name, "UserNoDump") == 0) { + xar->xmlsts = FILE_FLAGS_USER_NODUMP; + flag = "nodump"; + } + else if (strcmp(name, "UserImmutable") == 0) { + xar->xmlsts = FILE_FLAGS_USER_IMMUTABLE; + flag = "uimmutable"; + } + else if (strcmp(name, "UserAppend") == 0) { + xar->xmlsts = FILE_FLAGS_USER_APPEND; + flag = "uappend"; + } + else if (strcmp(name, "UserOpaque") == 0) { + xar->xmlsts = FILE_FLAGS_USER_OPAQUE; + flag = "opaque"; + } + else if (strcmp(name, "UserNoUnlink") == 0) { + xar->xmlsts = FILE_FLAGS_USER_NOUNLINK; + flag = "nouunlink"; + } + else if (strcmp(name, "SystemArchived") == 0) { + xar->xmlsts = FILE_FLAGS_SYS_ARCHIVED; + flag = "archived"; + } + else if (strcmp(name, "SystemImmutable") == 0) { + xar->xmlsts = FILE_FLAGS_SYS_IMMUTABLE; + flag = "simmutable"; + } + else if (strcmp(name, "SystemAppend") == 0) { + xar->xmlsts = FILE_FLAGS_SYS_APPEND; + flag = "sappend"; + } + else if (strcmp(name, "SystemNoUnlink") == 0) { + xar->xmlsts = FILE_FLAGS_SYS_NOUNLINK; + flag = "nosunlink"; + } + else if (strcmp(name, "SystemSnapshot") == 0) { + xar->xmlsts = FILE_FLAGS_SYS_SNAPSHOT; + flag = "snapshot"; + } + + if (flag == NULL) + return (0); + xar->file->has |= HAS_FFLAGS; + if (archive_strlen(&(xar->file->fflags_text)) > 0) + archive_strappend_char(&(xar->file->fflags_text), ','); + archive_strcat(&(xar->file->fflags_text), flag); + return (1); +} + +/* + * Linux file flags. + */ +static int +xml_parse_file_ext2(struct xar *xar, const char *name) +{ + const char *flag = NULL; + + if (strcmp(name, "SecureDeletion") == 0) { + xar->xmlsts = FILE_EXT2_SecureDeletion; + flag = "securedeletion"; + } + else if (strcmp(name, "Undelete") == 0) { + xar->xmlsts = FILE_EXT2_Undelete; + flag = "nouunlink"; + } + else if (strcmp(name, "Compress") == 0) { + xar->xmlsts = FILE_EXT2_Compress; + flag = "compress"; + } + else if (strcmp(name, "Synchronous") == 0) { + xar->xmlsts = FILE_EXT2_Synchronous; + flag = "sync"; + } + else if (strcmp(name, "Immutable") == 0) { + xar->xmlsts = FILE_EXT2_Immutable; + flag = "simmutable"; + } + else if (strcmp(name, "AppendOnly") == 0) { + xar->xmlsts = FILE_EXT2_AppendOnly; + flag = "sappend"; + } + else if (strcmp(name, "NoDump") == 0) { + xar->xmlsts = FILE_EXT2_NoDump; + flag = "nodump"; + } + else if (strcmp(name, "NoAtime") == 0) { + xar->xmlsts = FILE_EXT2_NoAtime; + flag = "noatime"; + } + else if (strcmp(name, "CompDirty") == 0) { + xar->xmlsts = FILE_EXT2_CompDirty; + flag = "compdirty"; + } + else if (strcmp(name, "CompBlock") == 0) { + xar->xmlsts = FILE_EXT2_CompBlock; + flag = "comprblk"; + } + else if (strcmp(name, "NoCompBlock") == 0) { + xar->xmlsts = FILE_EXT2_NoCompBlock; + flag = "nocomprblk"; + } + else if (strcmp(name, "CompError") == 0) { + xar->xmlsts = FILE_EXT2_CompError; + flag = "comperr"; + } + else if (strcmp(name, "BTree") == 0) { + xar->xmlsts = FILE_EXT2_BTree; + flag = "btree"; + } + else if (strcmp(name, "HashIndexed") == 0) { + xar->xmlsts = FILE_EXT2_HashIndexed; + flag = "hashidx"; + } + else if (strcmp(name, "iMagic") == 0) { + xar->xmlsts = FILE_EXT2_iMagic; + flag = "imagic"; + } + else if (strcmp(name, "Journaled") == 0) { + xar->xmlsts = FILE_EXT2_Journaled; + flag = "journal"; + } + else if (strcmp(name, "NoTail") == 0) { + xar->xmlsts = FILE_EXT2_NoTail; + flag = "notail"; + } + else if (strcmp(name, "DirSync") == 0) { + xar->xmlsts = FILE_EXT2_DirSync; + flag = "dirsync"; + } + else if (strcmp(name, "TopDir") == 0) { + xar->xmlsts = FILE_EXT2_TopDir; + flag = "topdir"; + } + else if (strcmp(name, "Reserved") == 0) { + xar->xmlsts = FILE_EXT2_Reserved; + flag = "reserved"; + } + + if (flag == NULL) + return (0); + if (archive_strlen(&(xar->file->fflags_text)) > 0) + archive_strappend_char(&(xar->file->fflags_text), ','); + archive_strcat(&(xar->file->fflags_text), flag); + return (1); +} + +#ifdef HAVE_LIBXML_XMLREADER_H + +static int +xml2_xmlattr_setup(struct archive_read *a, + struct xmlattr_list *list, xmlTextReaderPtr reader) +{ + struct xmlattr *attr; + int r; + + list->first = NULL; + list->last = &(list->first); + r = xmlTextReaderMoveToFirstAttribute(reader); + while (r == 1) { + attr = malloc(sizeof*(attr)); + if (attr == NULL) { + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + attr->name = strdup( + (const char *)xmlTextReaderConstLocalName(reader)); + if (attr->name == NULL) { + free(attr); + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + attr->value = strdup( + (const char *)xmlTextReaderConstValue(reader)); + if (attr->value == NULL) { + free(attr->name); + free(attr); + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + attr->next = NULL; + *list->last = attr; + list->last = &(attr->next); + r = xmlTextReaderMoveToNextAttribute(reader); + } + return (r); +} + +static int +xml2_read_cb(void *context, char *buffer, int len) +{ + struct archive_read *a; + struct xar *xar; + const void *d; + size_t outbytes; + size_t used = 0; + int r; + + a = (struct archive_read *)context; + xar = (struct xar *)(a->format->data); + + if (xar->toc_remaining <= 0) + return (0); + d = buffer; + outbytes = len; + r = rd_contents(a, &d, &outbytes, &used, xar->toc_remaining); + if (r != ARCHIVE_OK) + return (r); + __archive_read_consume(a, used); + xar->toc_remaining -= used; + xar->offset += used; + xar->toc_total += outbytes; + PRINT_TOC(buffer, len); + + return ((int)outbytes); +} + +static int +xml2_close_cb(void *context) +{ + + (void)context; /* UNUSED */ + return (0); +} + +static void +xml2_error_hdr(void *arg, const char *msg, xmlParserSeverities severity, + xmlTextReaderLocatorPtr locator) +{ + struct archive_read *a; + + (void)locator; /* UNUSED */ + a = (struct archive_read *)arg; + switch (severity) { + case XML_PARSER_SEVERITY_VALIDITY_WARNING: + case XML_PARSER_SEVERITY_WARNING: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "XML Parsing error: %s", msg); + break; + case XML_PARSER_SEVERITY_VALIDITY_ERROR: + case XML_PARSER_SEVERITY_ERROR: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "XML Parsing error: %s", msg); + break; + } +} + +static int +xml2_read_toc(struct archive_read *a) +{ + xmlTextReaderPtr reader; + struct xmlattr_list list; + int r; + + reader = xmlReaderForIO(xml2_read_cb, xml2_close_cb, a, NULL, NULL, 0); + if (reader == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Couldn't allocate memory for xml parser"); + return (ARCHIVE_FATAL); + } + xmlTextReaderSetErrorHandler(reader, xml2_error_hdr, a); + + while ((r = xmlTextReaderRead(reader)) == 1) { + const char *name, *value; + int type, empty; + + type = xmlTextReaderNodeType(reader); + name = (const char *)xmlTextReaderConstLocalName(reader); + switch (type) { + case XML_READER_TYPE_ELEMENT: + empty = xmlTextReaderIsEmptyElement(reader); + r = xml2_xmlattr_setup(a, &list, reader); + if (r == ARCHIVE_OK) + r = xml_start(a, name, &list); + xmlattr_cleanup(&list); + if (r != ARCHIVE_OK) + return (r); + if (empty) + xml_end(a, name); + break; + case XML_READER_TYPE_END_ELEMENT: + xml_end(a, name); + break; + case XML_READER_TYPE_TEXT: + value = (const char *)xmlTextReaderConstValue(reader); + xml_data(a, value, strlen(value)); + break; + case XML_READER_TYPE_SIGNIFICANT_WHITESPACE: + default: + break; + } + if (r < 0) + break; + } + xmlFreeTextReader(reader); + xmlCleanupParser(); + + return ((r == 0)?ARCHIVE_OK:ARCHIVE_FATAL); +} + +#elif defined(HAVE_BSDXML_H) || defined(HAVE_EXPAT_H) + +static int +expat_xmlattr_setup(struct archive_read *a, + struct xmlattr_list *list, const XML_Char **atts) +{ + struct xmlattr *attr; + char *name, *value; + + list->first = NULL; + list->last = &(list->first); + if (atts == NULL) + return (ARCHIVE_OK); + while (atts[0] != NULL && atts[1] != NULL) { + attr = malloc(sizeof*(attr)); + name = strdup(atts[0]); + value = strdup(atts[1]); + if (attr == NULL || name == NULL || value == NULL) { + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + free(attr); + free(name); + free(value); + return (ARCHIVE_FATAL); + } + attr->name = name; + attr->value = value; + attr->next = NULL; + *list->last = attr; + list->last = &(attr->next); + atts += 2; + } + return (ARCHIVE_OK); +} + +static void +expat_start_cb(void *userData, const XML_Char *name, const XML_Char **atts) +{ + struct expat_userData *ud = (struct expat_userData *)userData; + struct archive_read *a = ud->archive; + struct xmlattr_list list; + int r; + + r = expat_xmlattr_setup(a, &list, atts); + if (r == ARCHIVE_OK) + r = xml_start(a, (const char *)name, &list); + xmlattr_cleanup(&list); + ud->state = r; +} + +static void +expat_end_cb(void *userData, const XML_Char *name) +{ + struct expat_userData *ud = (struct expat_userData *)userData; + + xml_end(ud->archive, (const char *)name); +} + +static void +expat_data_cb(void *userData, const XML_Char *s, int len) +{ + struct expat_userData *ud = (struct expat_userData *)userData; + + xml_data(ud->archive, s, len); +} + +static int +expat_read_toc(struct archive_read *a) +{ + struct xar *xar; + XML_Parser parser; + struct expat_userData ud; + + ud.state = ARCHIVE_OK; + ud.archive = a; + + xar = (struct xar *)(a->format->data); + + /* Initialize XML Parser library. */ + parser = XML_ParserCreate(NULL); + if (parser == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Couldn't allocate memory for xml parser"); + return (ARCHIVE_FATAL); + } + XML_SetUserData(parser, &ud); + XML_SetElementHandler(parser, expat_start_cb, expat_end_cb); + XML_SetCharacterDataHandler(parser, expat_data_cb); + xar->xmlsts = INIT; + + while (xar->toc_remaining && ud.state == ARCHIVE_OK) { + enum XML_Status xr; + const void *d; + size_t outbytes; + size_t used; + int r; + + d = NULL; + r = rd_contents(a, &d, &outbytes, &used, xar->toc_remaining); + if (r != ARCHIVE_OK) + return (r); + xar->toc_remaining -= used; + xar->offset += used; + xar->toc_total += outbytes; + PRINT_TOC(d, outbytes); + + xr = XML_Parse(parser, d, outbytes, xar->toc_remaining == 0); + __archive_read_consume(a, used); + if (xr == XML_STATUS_ERROR) { + XML_ParserFree(parser); + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "XML Parsing failed"); + return (ARCHIVE_FATAL); + } + } + XML_ParserFree(parser); + return (ud.state); +} +#endif /* defined(HAVE_BSDXML_H) || defined(HAVE_EXPAT_H) */ + +#endif /* Support xar format */ diff --git a/c/archive_read_support_format_zip.c b/c/archive_read_support_format_zip.c new file mode 100644 --- /dev/null +++ b/c/archive_read_support_format_zip.c @@ -0,0 +1,4044 @@ +/*- + * Copyright (c) 2004-2013 Tim Kientzle + * Copyright (c) 2011-2012,2014 Michihiro NAKAJIMA + * Copyright (c) 2013 Konrad Kleine + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_read_support_format_zip.c 201102 2009-12-28 03:11:36Z kientzle $"); + +/* + * The definitive documentation of the Zip file format is: + * http://www.pkware.com/documents/casestudies/APPNOTE.TXT + * + * The Info-Zip project has pioneered various extensions to better + * support Zip on Unix, including the 0x5455 "UT", 0x5855 "UX", 0x7855 + * "Ux", and 0x7875 "ux" extensions for time and ownership + * information. + * + * History of this code: The streaming Zip reader was first added to + * libarchive in January 2005. Support for seekable input sources was + * added in Nov 2011. Zip64 support (including a significant code + * refactoring) was added in 2014. + */ + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_ZLIB_H +#include +#endif +#ifdef HAVE_BZLIB_H +#include +#endif +#ifdef HAVE_LZMA_H +#include +#endif + +#include "archive.h" +#include "archive_digest_private.h" +#include "archive_cryptor_private.h" +#include "archive_endian.h" +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_hmac_private.h" +#include "archive_private.h" +#include "archive_rb.h" +#include "archive_read_private.h" +#include "archive_ppmd8_private.h" + +#ifndef HAVE_ZLIB_H +#include "archive_crc32.h" +#endif + +struct zip_entry { + struct archive_rb_node node; + struct zip_entry *next; + int64_t local_header_offset; + int64_t compressed_size; + int64_t uncompressed_size; + int64_t gid; + int64_t uid; + struct archive_string rsrcname; + time_t mtime; + time_t atime; + time_t ctime; + uint32_t crc32; + uint16_t mode; + uint16_t zip_flags; /* From GP Flags Field */ + unsigned char compression; + unsigned char system; /* From "version written by" */ + unsigned char flags; /* Our extra markers. */ + unsigned char decdat;/* Used for Decryption check */ + + /* WinZip AES encryption extra field should be available + * when compression is 99. */ + struct { + /* Vendor version: AE-1 - 0x0001, AE-2 - 0x0002 */ + unsigned vendor; +#define AES_VENDOR_AE_1 0x0001 +#define AES_VENDOR_AE_2 0x0002 + /* AES encryption strength: + * 1 - 128 bits, 2 - 192 bits, 2 - 256 bits. */ + unsigned strength; + /* Actual compression method. */ + unsigned char compression; + } aes_extra; +}; + +struct trad_enc_ctx { + uint32_t keys[3]; +}; + +/* Bits used in zip_flags. */ +#define ZIP_ENCRYPTED (1 << 0) +#define ZIP_LENGTH_AT_END (1 << 3) +#define ZIP_STRONG_ENCRYPTED (1 << 6) +#define ZIP_UTF8_NAME (1 << 11) +/* See "7.2 Single Password Symmetric Encryption Method" + in http://www.pkware.com/documents/casestudies/APPNOTE.TXT */ +#define ZIP_CENTRAL_DIRECTORY_ENCRYPTED (1 << 13) + +/* Bits used in flags. */ +#define LA_USED_ZIP64 (1 << 0) +#define LA_FROM_CENTRAL_DIRECTORY (1 << 1) + +/* + * See "WinZip - AES Encryption Information" + * http://www.winzip.com/aes_info.htm + */ +/* Value used in compression method. */ +#define WINZIP_AES_ENCRYPTION 99 +/* Authentication code size. */ +#define AUTH_CODE_SIZE 10 +/**/ +#define MAX_DERIVED_KEY_BUF_SIZE (AES_MAX_KEY_SIZE * 2 + 2) + +struct zip { + /* Structural information about the archive. */ + struct archive_string format_name; + int64_t central_directory_offset; + size_t central_directory_entries_total; + size_t central_directory_entries_on_this_disk; + int has_encrypted_entries; + + /* List of entries (seekable Zip only) */ + struct zip_entry *zip_entries; + struct archive_rb_tree tree; + struct archive_rb_tree tree_rsrc; + + /* Bytes read but not yet consumed via __archive_read_consume() */ + size_t unconsumed; + + /* Information about entry we're currently reading. */ + struct zip_entry *entry; + int64_t entry_bytes_remaining; + + /* These count the number of bytes actually read for the entry. */ + int64_t entry_compressed_bytes_read; + int64_t entry_uncompressed_bytes_read; + + /* Running CRC32 of the decompressed data */ + unsigned long entry_crc32; + unsigned long (*crc32func)(unsigned long, const void *, + size_t); + char ignore_crc32; + + /* Flags to mark progress of decompression. */ + char decompress_init; + char end_of_entry; + + unsigned char *uncompressed_buffer; + size_t uncompressed_buffer_size; + +#ifdef HAVE_ZLIB_H + z_stream stream; + char stream_valid; +#endif + +#if HAVE_LZMA_H && HAVE_LIBLZMA + lzma_stream zipx_lzma_stream; + char zipx_lzma_valid; +#endif + +#ifdef HAVE_BZLIB_H + bz_stream bzstream; + char bzstream_valid; +#endif + + IByteIn zipx_ppmd_stream; + ssize_t zipx_ppmd_read_compressed; + CPpmd8 ppmd8; + char ppmd8_valid; + char ppmd8_stream_failed; + + struct archive_string_conv *sconv; + struct archive_string_conv *sconv_default; + struct archive_string_conv *sconv_utf8; + int init_default_conversion; + int process_mac_extensions; + + char init_decryption; + + /* Decryption buffer. */ + /* + * The decrypted data starts at decrypted_ptr and + * extends for decrypted_bytes_remaining. Decryption + * adds new data to the end of this block, data is returned + * to clients from the beginning. When the block hits the + * end of decrypted_buffer, it has to be shuffled back to + * the beginning of the buffer. + */ + unsigned char *decrypted_buffer; + unsigned char *decrypted_ptr; + size_t decrypted_buffer_size; + size_t decrypted_bytes_remaining; + size_t decrypted_unconsumed_bytes; + + /* Traditional PKWARE decryption. */ + struct trad_enc_ctx tctx; + char tctx_valid; + + /* WinZip AES decryption. */ + /* Contexts used for AES decryption. */ + archive_crypto_ctx cctx; + char cctx_valid; + archive_hmac_sha1_ctx hctx; + char hctx_valid; + + /* Strong encryption's decryption header information. */ + unsigned iv_size; + unsigned alg_id; + unsigned bit_len; + unsigned flags; + unsigned erd_size; + unsigned v_size; + unsigned v_crc32; + uint8_t *iv; + uint8_t *erd; + uint8_t *v_data; +}; + +/* Many systems define min or MIN, but not all. */ +#define zipmin(a,b) ((a) < (b) ? (a) : (b)) + +/* This function is used by Ppmd8_DecodeSymbol during decompression of Ppmd8 + * streams inside ZIP files. It has 2 purposes: one is to fetch the next + * compressed byte from the stream, second one is to increase the counter how + * many compressed bytes were read. */ +static Byte +ppmd_read(void* p) { + /* Get the handle to current decompression context. */ + struct archive_read *a = ((IByteIn*)p)->a; + struct zip *zip = (struct zip*) a->format->data; + ssize_t bytes_avail = 0; + + /* Fetch next byte. */ + const uint8_t* data = __archive_read_ahead(a, 1, &bytes_avail); + if(bytes_avail < 1) { + zip->ppmd8_stream_failed = 1; + return 0; + } + + __archive_read_consume(a, 1); + + /* Increment the counter. */ + ++zip->zipx_ppmd_read_compressed; + + /* Return the next compressed byte. */ + return data[0]; +} + +/* ------------------------------------------------------------------------ */ + +/* + Traditional PKWARE Decryption functions. + */ + +static void +trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c) +{ + uint8_t t; +#define CRC32(c, b) (crc32(c ^ 0xffffffffUL, &b, 1) ^ 0xffffffffUL) + + ctx->keys[0] = CRC32(ctx->keys[0], c); + ctx->keys[1] = (ctx->keys[1] + (ctx->keys[0] & 0xff)) * 134775813L + 1; + t = (ctx->keys[1] >> 24) & 0xff; + ctx->keys[2] = CRC32(ctx->keys[2], t); +#undef CRC32 +} + +static uint8_t +trad_enc_decrypt_byte(struct trad_enc_ctx *ctx) +{ + unsigned temp = ctx->keys[2] | 2; + return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff; +} + +static void +trad_enc_decrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in, + size_t in_len, uint8_t *out, size_t out_len) +{ + unsigned i, max; + + max = (unsigned)((in_len < out_len)? in_len: out_len); + + for (i = 0; i < max; i++) { + uint8_t t = in[i] ^ trad_enc_decrypt_byte(ctx); + out[i] = t; + trad_enc_update_keys(ctx, t); + } +} + +static int +trad_enc_init(struct trad_enc_ctx *ctx, const char *pw, size_t pw_len, + const uint8_t *key, size_t key_len, uint8_t *crcchk) +{ + uint8_t header[12]; + + if (key_len < 12) { + *crcchk = 0xff; + return -1; + } + + ctx->keys[0] = 305419896L; + ctx->keys[1] = 591751049L; + ctx->keys[2] = 878082192L; + + for (;pw_len; --pw_len) + trad_enc_update_keys(ctx, *pw++); + + trad_enc_decrypt_update(ctx, key, 12, header, 12); + /* Return the last byte for CRC check. */ + *crcchk = header[11]; + return 0; +} + +#if 0 +static void +crypt_derive_key_sha1(const void *p, int size, unsigned char *key, + int key_size) +{ +#define MD_SIZE 20 + archive_sha1_ctx ctx; + unsigned char md1[MD_SIZE]; + unsigned char md2[MD_SIZE * 2]; + unsigned char mkb[64]; + int i; + + archive_sha1_init(&ctx); + archive_sha1_update(&ctx, p, size); + archive_sha1_final(&ctx, md1); + + memset(mkb, 0x36, sizeof(mkb)); + for (i = 0; i < MD_SIZE; i++) + mkb[i] ^= md1[i]; + archive_sha1_init(&ctx); + archive_sha1_update(&ctx, mkb, sizeof(mkb)); + archive_sha1_final(&ctx, md2); + + memset(mkb, 0x5C, sizeof(mkb)); + for (i = 0; i < MD_SIZE; i++) + mkb[i] ^= md1[i]; + archive_sha1_init(&ctx); + archive_sha1_update(&ctx, mkb, sizeof(mkb)); + archive_sha1_final(&ctx, md2 + MD_SIZE); + + if (key_size > 32) + key_size = 32; + memcpy(key, md2, key_size); +#undef MD_SIZE +} +#endif + +/* + * Common code for streaming or seeking modes. + * + * Includes code to read local file headers, decompress data + * from entry bodies, and common API. + */ + +static unsigned long +real_crc32(unsigned long crc, const void *buff, size_t len) +{ + return crc32(crc, buff, (unsigned int)len); +} + +/* Used by "ignorecrc32" option to speed up tests. */ +static unsigned long +fake_crc32(unsigned long crc, const void *buff, size_t len) +{ + (void)crc; /* UNUSED */ + (void)buff; /* UNUSED */ + (void)len; /* UNUSED */ + return 0; +} + +static const struct { + int id; + const char * name; +} compression_methods[] = { + {0, "uncompressed"}, /* The file is stored (no compression) */ + {1, "shrinking"}, /* The file is Shrunk */ + {2, "reduced-1"}, /* The file is Reduced with compression factor 1 */ + {3, "reduced-2"}, /* The file is Reduced with compression factor 2 */ + {4, "reduced-3"}, /* The file is Reduced with compression factor 3 */ + {5, "reduced-4"}, /* The file is Reduced with compression factor 4 */ + {6, "imploded"}, /* The file is Imploded */ + {7, "reserved"}, /* Reserved for Tokenizing compression algorithm */ + {8, "deflation"}, /* The file is Deflated */ + {9, "deflation-64-bit"}, /* Enhanced Deflating using Deflate64(tm) */ + {10, "ibm-terse"},/* PKWARE Data Compression Library Imploding + * (old IBM TERSE) */ + {11, "reserved"}, /* Reserved by PKWARE */ + {12, "bzip"}, /* File is compressed using BZIP2 algorithm */ + {13, "reserved"}, /* Reserved by PKWARE */ + {14, "lzma"}, /* LZMA (EFS) */ + {15, "reserved"}, /* Reserved by PKWARE */ + {16, "reserved"}, /* Reserved by PKWARE */ + {17, "reserved"}, /* Reserved by PKWARE */ + {18, "ibm-terse-new"}, /* File is compressed using IBM TERSE (new) */ + {19, "ibm-lz777"},/* IBM LZ77 z Architecture (PFS) */ + {95, "xz"}, /* XZ compressed data */ + {96, "jpeg"}, /* JPEG compressed data */ + {97, "wav-pack"}, /* WavPack compressed data */ + {98, "ppmd-1"}, /* PPMd version I, Rev 1 */ + {99, "aes"} /* WinZip AES encryption */ +}; + +static const char * +compression_name(const int compression) +{ + static const int num_compression_methods = + sizeof(compression_methods)/sizeof(compression_methods[0]); + int i=0; + + while(compression >= 0 && i < num_compression_methods) { + if (compression_methods[i].id == compression) + return compression_methods[i].name; + i++; + } + return "??"; +} + +/* Convert an MSDOS-style date/time into Unix-style time. */ +static time_t +zip_time(const char *p) +{ + int msTime, msDate; + struct tm ts; + + msTime = (0xff & (unsigned)p[0]) + 256 * (0xff & (unsigned)p[1]); + msDate = (0xff & (unsigned)p[2]) + 256 * (0xff & (unsigned)p[3]); + + memset(&ts, 0, sizeof(ts)); + ts.tm_year = ((msDate >> 9) & 0x7f) + 80; /* Years since 1900. */ + ts.tm_mon = ((msDate >> 5) & 0x0f) - 1; /* Month number. */ + ts.tm_mday = msDate & 0x1f; /* Day of month. */ + ts.tm_hour = (msTime >> 11) & 0x1f; + ts.tm_min = (msTime >> 5) & 0x3f; + ts.tm_sec = (msTime << 1) & 0x3e; + ts.tm_isdst = -1; + return mktime(&ts); +} + +/* + * The extra data is stored as a list of + * id1+size1+data1 + id2+size2+data2 ... + * triplets. id and size are 2 bytes each. + */ +static int +process_extra(struct archive_read *a, struct archive_entry *entry, + const char *p, size_t extra_length, struct zip_entry* zip_entry) +{ + unsigned offset = 0; + struct zip *zip = (struct zip *)(a->format->data); + + if (extra_length == 0) { + return ARCHIVE_OK; + } + + if (extra_length < 4) { + size_t i = 0; + /* Some ZIP files may have trailing 0 bytes. Let's check they + * are all 0 and ignore them instead of returning an error. + * + * This is not technically correct, but some ZIP files look + * like this and other tools support those files - so let's + * also support them. + */ + for (; i < extra_length; i++) { + if (p[i] != 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Too-small extra data: " + "Need at least 4 bytes, " + "but only found %d bytes", + (int)extra_length); + return ARCHIVE_FAILED; + } + } + + return ARCHIVE_OK; + } + + while (offset <= extra_length - 4) { + unsigned short headerid = archive_le16dec(p + offset); + unsigned short datasize = archive_le16dec(p + offset + 2); + + offset += 4; + if (offset + datasize > extra_length) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, "Extra data overflow: " + "Need %d bytes but only found %d bytes", + (int)datasize, (int)(extra_length - offset)); + return ARCHIVE_FAILED; + } +#ifdef DEBUG + fprintf(stderr, "Header id 0x%04x, length %d\n", + headerid, datasize); +#endif + switch (headerid) { + case 0x0001: + /* Zip64 extended information extra field. */ + zip_entry->flags |= LA_USED_ZIP64; + if (zip_entry->uncompressed_size == 0xffffffff) { + uint64_t t = 0; + if (datasize < 8 + || (t = archive_le64dec(p + offset)) > + INT64_MAX) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Malformed 64-bit " + "uncompressed size"); + return ARCHIVE_FAILED; + } + zip_entry->uncompressed_size = t; + offset += 8; + datasize -= 8; + } + if (zip_entry->compressed_size == 0xffffffff) { + uint64_t t = 0; + if (datasize < 8 + || (t = archive_le64dec(p + offset)) > + INT64_MAX) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Malformed 64-bit " + "compressed size"); + return ARCHIVE_FAILED; + } + zip_entry->compressed_size = t; + offset += 8; + datasize -= 8; + } + if (zip_entry->local_header_offset == 0xffffffff) { + uint64_t t = 0; + if (datasize < 8 + || (t = archive_le64dec(p + offset)) > + INT64_MAX) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Malformed 64-bit " + "local header offset"); + return ARCHIVE_FAILED; + } + zip_entry->local_header_offset = t; + offset += 8; + datasize -= 8; + } + /* archive_le32dec(p + offset) gives disk + * on which file starts, but we don't handle + * multi-volume Zip files. */ + break; +#ifdef DEBUG + case 0x0017: + { + /* Strong encryption field. */ + if (archive_le16dec(p + offset) == 2) { + unsigned algId = + archive_le16dec(p + offset + 2); + unsigned bitLen = + archive_le16dec(p + offset + 4); + int flags = + archive_le16dec(p + offset + 6); + fprintf(stderr, "algId=0x%04x, bitLen=%u, " + "flgas=%d\n", algId, bitLen,flags); + } + break; + } +#endif + case 0x5455: + { + /* Extended time field "UT". */ + int flags; + if (datasize == 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Incomplete extended time field"); + return ARCHIVE_FAILED; + } + flags = p[offset]; + offset++; + datasize--; + /* Flag bits indicate which dates are present. */ + if (flags & 0x01) + { +#ifdef DEBUG + fprintf(stderr, "mtime: %lld -> %d\n", + (long long)zip_entry->mtime, + archive_le32dec(p + offset)); +#endif + if (datasize < 4) + break; + zip_entry->mtime = archive_le32dec(p + offset); + offset += 4; + datasize -= 4; + } + if (flags & 0x02) + { + if (datasize < 4) + break; + zip_entry->atime = archive_le32dec(p + offset); + offset += 4; + datasize -= 4; + } + if (flags & 0x04) + { + if (datasize < 4) + break; + zip_entry->ctime = archive_le32dec(p + offset); + offset += 4; + datasize -= 4; + } + break; + } + case 0x5855: + { + /* Info-ZIP Unix Extra Field (old version) "UX". */ + if (datasize >= 8) { + zip_entry->atime = archive_le32dec(p + offset); + zip_entry->mtime = + archive_le32dec(p + offset + 4); + } + if (datasize >= 12) { + zip_entry->uid = + archive_le16dec(p + offset + 8); + zip_entry->gid = + archive_le16dec(p + offset + 10); + } + break; + } + case 0x6c78: + { + /* Experimental 'xl' field */ + /* + * Introduced Dec 2013 to provide a way to + * include external file attributes (and other + * fields that ordinarily appear only in + * central directory) in local file header. + * This provides file type and permission + * information necessary to support full + * streaming extraction. Currently being + * discussed with other Zip developers + * ... subject to change. + * + * Format: + * The field starts with a bitmap that specifies + * which additional fields are included. The + * bitmap is variable length and can be extended in + * the future. + * + * n bytes - feature bitmap: first byte has low-order + * 7 bits. If high-order bit is set, a subsequent + * byte holds the next 7 bits, etc. + * + * if bitmap & 1, 2 byte "version made by" + * if bitmap & 2, 2 byte "internal file attributes" + * if bitmap & 4, 4 byte "external file attributes" + * if bitmap & 8, 2 byte comment length + n byte + * comment + */ + int bitmap, bitmap_last; + + if (datasize < 1) + break; + bitmap_last = bitmap = 0xff & p[offset]; + offset += 1; + datasize -= 1; + + /* We only support first 7 bits of bitmap; skip rest. */ + while ((bitmap_last & 0x80) != 0 + && datasize >= 1) { + bitmap_last = p[offset]; + offset += 1; + datasize -= 1; + } + + if (bitmap & 1) { + /* 2 byte "version made by" */ + if (datasize < 2) + break; + zip_entry->system + = archive_le16dec(p + offset) >> 8; + offset += 2; + datasize -= 2; + } + if (bitmap & 2) { + /* 2 byte "internal file attributes" */ + uint32_t internal_attributes; + if (datasize < 2) + break; + internal_attributes + = archive_le16dec(p + offset); + /* Not used by libarchive at present. */ + (void)internal_attributes; /* UNUSED */ + offset += 2; + datasize -= 2; + } + if (bitmap & 4) { + /* 4 byte "external file attributes" */ + uint32_t external_attributes; + if (datasize < 4) + break; + external_attributes + = archive_le32dec(p + offset); + if (zip_entry->system == 3) { + zip_entry->mode + = external_attributes >> 16; + } else if (zip_entry->system == 0) { + // Interpret MSDOS directory bit + if (0x10 == (external_attributes & + 0x10)) { + zip_entry->mode = + AE_IFDIR | 0775; + } else { + zip_entry->mode = + AE_IFREG | 0664; + } + if (0x01 == (external_attributes & + 0x01)) { + /* Read-only bit; + * strip write permissions */ + zip_entry->mode &= 0555; + } + } else { + zip_entry->mode = 0; + } + offset += 4; + datasize -= 4; + } + if (bitmap & 8) { + /* 2 byte comment length + comment */ + uint32_t comment_length; + if (datasize < 2) + break; + comment_length + = archive_le16dec(p + offset); + offset += 2; + datasize -= 2; + + if (datasize < comment_length) + break; + /* Comment is not supported by libarchive */ + offset += comment_length; + datasize -= comment_length; + } + break; + } + case 0x7075: + { + /* Info-ZIP Unicode Path Extra Field. */ + if (datasize < 5 || entry == NULL) + break; + offset += 5; + datasize -= 5; + + /* The path name in this field is always encoded + * in UTF-8. */ + if (zip->sconv_utf8 == NULL) { + zip->sconv_utf8 = + archive_string_conversion_from_charset( + &a->archive, "UTF-8", 1); + /* If the converter from UTF-8 is not + * available, then the path name from the main + * field will more likely be correct. */ + if (zip->sconv_utf8 == NULL) + break; + } + + /* Make sure the CRC32 of the filename matches. */ + if (!zip->ignore_crc32) { + const char *cp = archive_entry_pathname(entry); + if (cp) { + unsigned long file_crc = + zip->crc32func(0, cp, strlen(cp)); + unsigned long utf_crc = + archive_le32dec(p + offset - 4); + if (file_crc != utf_crc) { +#ifdef DEBUG + fprintf(stderr, + "CRC filename mismatch; " + "CDE is %lx, but UTF8 " + "is outdated with %lx\n", + file_crc, utf_crc); +#endif + break; + } + } + } + + if (archive_entry_copy_pathname_l(entry, + p + offset, datasize, zip->sconv_utf8) != 0) { + /* Ignore the error, and fallback to the path + * name from the main field. */ +#ifdef DEBUG + fprintf(stderr, "Failed to read the ZIP " + "0x7075 extra field path.\n"); +#endif + } + break; + } + case 0x7855: + /* Info-ZIP Unix Extra Field (type 2) "Ux". */ +#ifdef DEBUG + fprintf(stderr, "uid %d gid %d\n", + archive_le16dec(p + offset), + archive_le16dec(p + offset + 2)); +#endif + if (datasize >= 2) + zip_entry->uid = archive_le16dec(p + offset); + if (datasize >= 4) + zip_entry->gid = + archive_le16dec(p + offset + 2); + break; + case 0x7875: + { + /* Info-Zip Unix Extra Field (type 3) "ux". */ + int uidsize = 0, gidsize = 0; + + /* TODO: support arbitrary uidsize/gidsize. */ + if (datasize >= 1 && p[offset] == 1) {/* version=1 */ + if (datasize >= 4) { + /* get a uid size. */ + uidsize = 0xff & (int)p[offset+1]; + if (uidsize == 2) + zip_entry->uid = + archive_le16dec( + p + offset + 2); + else if (uidsize == 4 && datasize >= 6) + zip_entry->uid = + archive_le32dec( + p + offset + 2); + } + if (datasize >= (2 + uidsize + 3)) { + /* get a gid size. */ + gidsize = 0xff & + (int)p[offset+2+uidsize]; + if (gidsize == 2) + zip_entry->gid = + archive_le16dec( + p+offset+2+uidsize+1); + else if (gidsize == 4 && + datasize >= (2 + uidsize + 5)) + zip_entry->gid = + archive_le32dec( + p+offset+2+uidsize+1); + } + } + break; + } + case 0x9901: + /* WinZip AES extra data field. */ + if (datasize < 6) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Incomplete AES field"); + return ARCHIVE_FAILED; + } + if (p[offset + 2] == 'A' && p[offset + 3] == 'E') { + /* Vendor version. */ + zip_entry->aes_extra.vendor = + archive_le16dec(p + offset); + /* AES encryption strength. */ + zip_entry->aes_extra.strength = p[offset + 4]; + /* Actual compression method. */ + zip_entry->aes_extra.compression = + p[offset + 5]; + } + break; + default: + break; + } + offset += datasize; + } + return ARCHIVE_OK; +} + +/* + * Assumes file pointer is at beginning of local file header. + */ +static int +zip_read_local_file_header(struct archive_read *a, struct archive_entry *entry, + struct zip *zip) +{ + const char *p; + const void *h; + const wchar_t *wp; + const char *cp; + size_t len, filename_length, extra_length; + struct archive_string_conv *sconv; + struct zip_entry *zip_entry = zip->entry; + struct zip_entry zip_entry_central_dir; + int ret = ARCHIVE_OK; + char version; + + /* Save a copy of the original for consistency checks. */ + zip_entry_central_dir = *zip_entry; + + zip->decompress_init = 0; + zip->end_of_entry = 0; + zip->entry_uncompressed_bytes_read = 0; + zip->entry_compressed_bytes_read = 0; + zip->entry_crc32 = zip->crc32func(0, NULL, 0); + + /* Setup default conversion. */ + if (zip->sconv == NULL && !zip->init_default_conversion) { + zip->sconv_default = + archive_string_default_conversion_for_read(&(a->archive)); + zip->init_default_conversion = 1; + } + + if ((p = __archive_read_ahead(a, 30, NULL)) == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated ZIP file header"); + return (ARCHIVE_FATAL); + } + + if (memcmp(p, "PK\003\004", 4) != 0) { + archive_set_error(&a->archive, -1, "Damaged Zip archive"); + return ARCHIVE_FATAL; + } + version = p[4]; + zip_entry->system = p[5]; + zip_entry->zip_flags = archive_le16dec(p + 6); + if (zip_entry->zip_flags & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)) { + zip->has_encrypted_entries = 1; + archive_entry_set_is_data_encrypted(entry, 1); + if (zip_entry->zip_flags & ZIP_CENTRAL_DIRECTORY_ENCRYPTED && + zip_entry->zip_flags & ZIP_ENCRYPTED && + zip_entry->zip_flags & ZIP_STRONG_ENCRYPTED) { + archive_entry_set_is_metadata_encrypted(entry, 1); + return ARCHIVE_FATAL; + } + } + zip->init_decryption = (zip_entry->zip_flags & ZIP_ENCRYPTED); + zip_entry->compression = (char)archive_le16dec(p + 8); + zip_entry->mtime = zip_time(p + 10); + zip_entry->crc32 = archive_le32dec(p + 14); + if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) + zip_entry->decdat = p[11]; + else + zip_entry->decdat = p[17]; + zip_entry->compressed_size = archive_le32dec(p + 18); + zip_entry->uncompressed_size = archive_le32dec(p + 22); + filename_length = archive_le16dec(p + 26); + extra_length = archive_le16dec(p + 28); + + __archive_read_consume(a, 30); + + /* Read the filename. */ + if ((h = __archive_read_ahead(a, filename_length, NULL)) == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated ZIP file header"); + return (ARCHIVE_FATAL); + } + if (zip_entry->zip_flags & ZIP_UTF8_NAME) { + /* The filename is stored to be UTF-8. */ + if (zip->sconv_utf8 == NULL) { + zip->sconv_utf8 = + archive_string_conversion_from_charset( + &a->archive, "UTF-8", 1); + if (zip->sconv_utf8 == NULL) + return (ARCHIVE_FATAL); + } + sconv = zip->sconv_utf8; + } else if (zip->sconv != NULL) + sconv = zip->sconv; + else + sconv = zip->sconv_default; + + if (archive_entry_copy_pathname_l(entry, + h, filename_length, sconv) != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Pathname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Pathname cannot be converted " + "from %s to current locale.", + archive_string_conversion_charset_name(sconv)); + ret = ARCHIVE_WARN; + } + __archive_read_consume(a, filename_length); + + /* Read the extra data. */ + if ((h = __archive_read_ahead(a, extra_length, NULL)) == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated ZIP file header"); + return (ARCHIVE_FATAL); + } + + if (ARCHIVE_OK != process_extra(a, entry, h, extra_length, + zip_entry)) { + return ARCHIVE_FATAL; + } + __archive_read_consume(a, extra_length); + + /* Work around a bug in Info-Zip: When reading from a pipe, it + * stats the pipe instead of synthesizing a file entry. */ + if ((zip_entry->mode & AE_IFMT) == AE_IFIFO) { + zip_entry->mode &= ~ AE_IFMT; + zip_entry->mode |= AE_IFREG; + } + + /* If the mode is totally empty, set some sane default. */ + if (zip_entry->mode == 0) { + zip_entry->mode |= 0664; + } + + /* Windows archivers sometimes use backslash as the directory + * separator. Normalize to slash. */ + if (zip_entry->system == 0 && + (wp = archive_entry_pathname_w(entry)) != NULL) { + if (wcschr(wp, L'/') == NULL && wcschr(wp, L'\\') != NULL) { + size_t i; + struct archive_wstring s; + archive_string_init(&s); + archive_wstrcpy(&s, wp); + for (i = 0; i < archive_strlen(&s); i++) { + if (s.s[i] == '\\') + s.s[i] = '/'; + } + archive_entry_copy_pathname_w(entry, s.s); + archive_wstring_free(&s); + } + } + + /* Make sure that entries with a trailing '/' are marked as directories + * even if the External File Attributes contains bogus values. If this + * is not a directory and there is no type, assume a regular file. */ + if ((zip_entry->mode & AE_IFMT) != AE_IFDIR) { + int has_slash; + + wp = archive_entry_pathname_w(entry); + if (wp != NULL) { + len = wcslen(wp); + has_slash = len > 0 && wp[len - 1] == L'/'; + } else { + cp = archive_entry_pathname(entry); + len = (cp != NULL)?strlen(cp):0; + has_slash = len > 0 && cp[len - 1] == '/'; + } + /* Correct file type as needed. */ + if (has_slash) { + zip_entry->mode &= ~AE_IFMT; + zip_entry->mode |= AE_IFDIR; + zip_entry->mode |= 0111; + } else if ((zip_entry->mode & AE_IFMT) == 0) { + zip_entry->mode |= AE_IFREG; + } + } + + /* Make sure directories end in '/' */ + if ((zip_entry->mode & AE_IFMT) == AE_IFDIR) { + wp = archive_entry_pathname_w(entry); + if (wp != NULL) { + len = wcslen(wp); + if (len > 0 && wp[len - 1] != L'/') { + struct archive_wstring s; + archive_string_init(&s); + archive_wstrcat(&s, wp); + archive_wstrappend_wchar(&s, L'/'); + archive_entry_copy_pathname_w(entry, s.s); + archive_wstring_free(&s); + } + } else { + cp = archive_entry_pathname(entry); + len = (cp != NULL)?strlen(cp):0; + if (len > 0 && cp[len - 1] != '/') { + struct archive_string s; + archive_string_init(&s); + archive_strcat(&s, cp); + archive_strappend_char(&s, '/'); + archive_entry_set_pathname(entry, s.s); + archive_string_free(&s); + } + } + } + + if (zip_entry->flags & LA_FROM_CENTRAL_DIRECTORY) { + /* If this came from the central dir, its size info + * is definitive, so ignore the length-at-end flag. */ + zip_entry->zip_flags &= ~ZIP_LENGTH_AT_END; + /* If local header is missing a value, use the one from + the central directory. If both have it, warn about + mismatches. */ + if (zip_entry->crc32 == 0) { + zip_entry->crc32 = zip_entry_central_dir.crc32; + } else if (!zip->ignore_crc32 + && zip_entry->crc32 != zip_entry_central_dir.crc32) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Inconsistent CRC32 values"); + ret = ARCHIVE_WARN; + } + if (zip_entry->compressed_size == 0) { + zip_entry->compressed_size + = zip_entry_central_dir.compressed_size; + } else if (zip_entry->compressed_size + != zip_entry_central_dir.compressed_size) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Inconsistent compressed size: " + "%jd in central directory, %jd in local header", + (intmax_t)zip_entry_central_dir.compressed_size, + (intmax_t)zip_entry->compressed_size); + ret = ARCHIVE_WARN; + } + if (zip_entry->uncompressed_size == 0) { + zip_entry->uncompressed_size + = zip_entry_central_dir.uncompressed_size; + } else if (zip_entry->uncompressed_size + != zip_entry_central_dir.uncompressed_size) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Inconsistent uncompressed size: " + "%jd in central directory, %jd in local header", + (intmax_t)zip_entry_central_dir.uncompressed_size, + (intmax_t)zip_entry->uncompressed_size); + ret = ARCHIVE_WARN; + } + } + + /* Populate some additional entry fields: */ + archive_entry_set_mode(entry, zip_entry->mode); + archive_entry_set_uid(entry, zip_entry->uid); + archive_entry_set_gid(entry, zip_entry->gid); + archive_entry_set_mtime(entry, zip_entry->mtime, 0); + archive_entry_set_ctime(entry, zip_entry->ctime, 0); + archive_entry_set_atime(entry, zip_entry->atime, 0); + + if ((zip->entry->mode & AE_IFMT) == AE_IFLNK) { + size_t linkname_length; + + if (zip_entry->compressed_size > 64 * 1024) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Zip file with oversized link entry"); + return ARCHIVE_FATAL; + } + + linkname_length = (size_t)zip_entry->compressed_size; + + archive_entry_set_size(entry, 0); + p = __archive_read_ahead(a, linkname_length, NULL); + if (p == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Truncated Zip file"); + return ARCHIVE_FATAL; + } + + sconv = zip->sconv; + if (sconv == NULL && (zip->entry->zip_flags & ZIP_UTF8_NAME)) + sconv = zip->sconv_utf8; + if (sconv == NULL) + sconv = zip->sconv_default; + if (archive_entry_copy_symlink_l(entry, p, linkname_length, + sconv) != 0) { + if (errno != ENOMEM && sconv == zip->sconv_utf8 && + (zip->entry->zip_flags & ZIP_UTF8_NAME)) + archive_entry_copy_symlink_l(entry, p, + linkname_length, NULL); + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Symlink"); + return (ARCHIVE_FATAL); + } + /* + * Since there is no character-set regulation for + * symlink name, do not report the conversion error + * in an automatic conversion. + */ + if (sconv != zip->sconv_utf8 || + (zip->entry->zip_flags & ZIP_UTF8_NAME) == 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Symlink cannot be converted " + "from %s to current locale.", + archive_string_conversion_charset_name( + sconv)); + ret = ARCHIVE_WARN; + } + } + zip_entry->uncompressed_size = zip_entry->compressed_size = 0; + + if (__archive_read_consume(a, linkname_length) < 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Read error skipping symlink target name"); + return ARCHIVE_FATAL; + } + } else if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) + || zip_entry->uncompressed_size > 0) { + /* Set the size only if it's meaningful. */ + archive_entry_set_size(entry, zip_entry->uncompressed_size); + } + zip->entry_bytes_remaining = zip_entry->compressed_size; + + /* If there's no body, force read_data() to return EOF immediately. */ + if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) + && zip->entry_bytes_remaining < 1) + zip->end_of_entry = 1; + + /* Set up a more descriptive format name. */ + archive_string_empty(&zip->format_name); + archive_string_sprintf(&zip->format_name, "ZIP %d.%d (%s)", + version / 10, version % 10, + compression_name(zip->entry->compression)); + a->archive.archive_format_name = zip->format_name.s; + + return (ret); +} + +static int +check_authentication_code(struct archive_read *a, const void *_p) +{ + struct zip *zip = (struct zip *)(a->format->data); + + /* Check authentication code. */ + if (zip->hctx_valid) { + const void *p; + uint8_t hmac[20]; + size_t hmac_len = 20; + int cmp; + + archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len); + if (_p == NULL) { + /* Read authentication code. */ + p = __archive_read_ahead(a, AUTH_CODE_SIZE, NULL); + if (p == NULL) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated ZIP file data"); + return (ARCHIVE_FATAL); + } + } else { + p = _p; + } + cmp = memcmp(hmac, p, AUTH_CODE_SIZE); + __archive_read_consume(a, AUTH_CODE_SIZE); + if (cmp != 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "ZIP bad Authentication code"); + return (ARCHIVE_WARN); + } + } + return (ARCHIVE_OK); +} + +/* + * Read "uncompressed" data. There are three cases: + * 1) We know the size of the data. This is always true for the + * seeking reader (we've examined the Central Directory already). + * 2) ZIP_LENGTH_AT_END was set, but only the CRC was deferred. + * Info-ZIP seems to do this; we know the size but have to grab + * the CRC from the data descriptor afterwards. + * 3) We're streaming and ZIP_LENGTH_AT_END was specified and + * we have no size information. In this case, we can do pretty + * well by watching for the data descriptor record. The data + * descriptor is 16 bytes and includes a computed CRC that should + * provide a strong check. + * + * TODO: Technically, the PK\007\010 signature is optional. + * In the original spec, the data descriptor contained CRC + * and size fields but had no leading signature. In practice, + * newer writers seem to provide the signature pretty consistently. + * + * For uncompressed data, the PK\007\010 marker seems essential + * to be sure we've actually seen the end of the entry. + * + * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets + * zip->end_of_entry if it consumes all of the data. + */ +static int +zip_read_data_none(struct archive_read *a, const void **_buff, + size_t *size, int64_t *offset) +{ + struct zip *zip; + const char *buff; + ssize_t bytes_avail; + int r; + + (void)offset; /* UNUSED */ + + zip = (struct zip *)(a->format->data); + + if (zip->entry->zip_flags & ZIP_LENGTH_AT_END) { + const char *p; + ssize_t grabbing_bytes = 24; + + if (zip->hctx_valid) + grabbing_bytes += AUTH_CODE_SIZE; + /* Grab at least 24 bytes. */ + buff = __archive_read_ahead(a, grabbing_bytes, &bytes_avail); + if (bytes_avail < grabbing_bytes) { + /* Zip archives have end-of-archive markers + that are longer than this, so a failure to get at + least 24 bytes really does indicate a truncated + file. */ + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated ZIP file data"); + return (ARCHIVE_FATAL); + } + /* Check for a complete PK\007\010 signature, followed + * by the correct 4-byte CRC. */ + p = buff; + if (zip->hctx_valid) + p += AUTH_CODE_SIZE; + if (p[0] == 'P' && p[1] == 'K' + && p[2] == '\007' && p[3] == '\010' + && (archive_le32dec(p + 4) == zip->entry_crc32 + || zip->ignore_crc32 + || (zip->hctx_valid + && zip->entry->aes_extra.vendor == AES_VENDOR_AE_2))) { + if (zip->entry->flags & LA_USED_ZIP64) { + uint64_t compressed, uncompressed; + zip->entry->crc32 = archive_le32dec(p + 4); + compressed = archive_le64dec(p + 8); + uncompressed = archive_le64dec(p + 16); + if (compressed > INT64_MAX || uncompressed > + INT64_MAX) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Overflow of 64-bit file sizes"); + return ARCHIVE_FAILED; + } + zip->entry->compressed_size = compressed; + zip->entry->uncompressed_size = uncompressed; + zip->unconsumed = 24; + } else { + zip->entry->crc32 = archive_le32dec(p + 4); + zip->entry->compressed_size = + archive_le32dec(p + 8); + zip->entry->uncompressed_size = + archive_le32dec(p + 12); + zip->unconsumed = 16; + } + if (zip->hctx_valid) { + r = check_authentication_code(a, buff); + if (r != ARCHIVE_OK) + return (r); + } + zip->end_of_entry = 1; + return (ARCHIVE_OK); + } + /* If not at EOF, ensure we consume at least one byte. */ + ++p; + + /* Scan forward until we see where a PK\007\010 signature + * might be. */ + /* Return bytes up until that point. On the next call, + * the code above will verify the data descriptor. */ + while (p < buff + bytes_avail - 4) { + if (p[3] == 'P') { p += 3; } + else if (p[3] == 'K') { p += 2; } + else if (p[3] == '\007') { p += 1; } + else if (p[3] == '\010' && p[2] == '\007' + && p[1] == 'K' && p[0] == 'P') { + if (zip->hctx_valid) + p -= AUTH_CODE_SIZE; + break; + } else { p += 4; } + } + bytes_avail = p - buff; + } else { + if (zip->entry_bytes_remaining == 0) { + zip->end_of_entry = 1; + if (zip->hctx_valid) { + r = check_authentication_code(a, NULL); + if (r != ARCHIVE_OK) + return (r); + } + return (ARCHIVE_OK); + } + /* Grab a bunch of bytes. */ + buff = __archive_read_ahead(a, 1, &bytes_avail); + if (bytes_avail <= 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated ZIP file data"); + return (ARCHIVE_FATAL); + } + if (bytes_avail > zip->entry_bytes_remaining) + bytes_avail = (ssize_t)zip->entry_bytes_remaining; + } + if (zip->tctx_valid || zip->cctx_valid) { + size_t dec_size = bytes_avail; + + if (dec_size > zip->decrypted_buffer_size) + dec_size = zip->decrypted_buffer_size; + if (zip->tctx_valid) { + trad_enc_decrypt_update(&zip->tctx, + (const uint8_t *)buff, dec_size, + zip->decrypted_buffer, dec_size); + } else { + size_t dsize = dec_size; + archive_hmac_sha1_update(&zip->hctx, + (const uint8_t *)buff, dec_size); + archive_decrypto_aes_ctr_update(&zip->cctx, + (const uint8_t *)buff, dec_size, + zip->decrypted_buffer, &dsize); + } + bytes_avail = dec_size; + buff = (const char *)zip->decrypted_buffer; + } + *size = bytes_avail; + zip->entry_bytes_remaining -= bytes_avail; + zip->entry_uncompressed_bytes_read += bytes_avail; + zip->entry_compressed_bytes_read += bytes_avail; + zip->unconsumed += bytes_avail; + *_buff = buff; + return (ARCHIVE_OK); +} + +static int +consume_optional_marker(struct archive_read *a, struct zip *zip) +{ + if (zip->end_of_entry && (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { + const char *p; + + if (NULL == (p = __archive_read_ahead(a, 24, NULL))) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated ZIP end-of-file record"); + return (ARCHIVE_FATAL); + } + /* Consume the optional PK\007\010 marker. */ + if (p[0] == 'P' && p[1] == 'K' && + p[2] == '\007' && p[3] == '\010') { + p += 4; + zip->unconsumed = 4; + } + if (zip->entry->flags & LA_USED_ZIP64) { + uint64_t compressed, uncompressed; + zip->entry->crc32 = archive_le32dec(p); + compressed = archive_le64dec(p + 4); + uncompressed = archive_le64dec(p + 12); + if (compressed > INT64_MAX || + uncompressed > INT64_MAX) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Overflow of 64-bit file sizes"); + return ARCHIVE_FAILED; + } + zip->entry->compressed_size = compressed; + zip->entry->uncompressed_size = uncompressed; + zip->unconsumed += 20; + } else { + zip->entry->crc32 = archive_le32dec(p); + zip->entry->compressed_size = archive_le32dec(p + 4); + zip->entry->uncompressed_size = archive_le32dec(p + 8); + zip->unconsumed += 12; + } + } + + return (ARCHIVE_OK); +} + +#if HAVE_LZMA_H && HAVE_LIBLZMA +static int +zipx_xz_init(struct archive_read *a, struct zip *zip) +{ + lzma_ret r; + + if(zip->zipx_lzma_valid) { + lzma_end(&zip->zipx_lzma_stream); + zip->zipx_lzma_valid = 0; + } + + memset(&zip->zipx_lzma_stream, 0, sizeof(zip->zipx_lzma_stream)); + r = lzma_stream_decoder(&zip->zipx_lzma_stream, UINT64_MAX, 0); + if (r != LZMA_OK) { + archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, + "xz initialization failed(%d)", + r); + + return (ARCHIVE_FAILED); + } + + zip->zipx_lzma_valid = 1; + + free(zip->uncompressed_buffer); + + zip->uncompressed_buffer_size = 256 * 1024; + zip->uncompressed_buffer = + (uint8_t*) malloc(zip->uncompressed_buffer_size); + if (zip->uncompressed_buffer == NULL) { + archive_set_error(&a->archive, ENOMEM, + "No memory for xz decompression"); + return (ARCHIVE_FATAL); + } + + zip->decompress_init = 1; + return (ARCHIVE_OK); +} + +static int +zipx_lzma_alone_init(struct archive_read *a, struct zip *zip) +{ + lzma_ret r; + const uint8_t* p; + +#pragma pack(push) +#pragma pack(1) + struct _alone_header { + uint8_t bytes[5]; + uint64_t uncompressed_size; + } alone_header; +#pragma pack(pop) + + if(zip->zipx_lzma_valid) { + lzma_end(&zip->zipx_lzma_stream); + zip->zipx_lzma_valid = 0; + } + + /* To unpack ZIPX's "LZMA" (id 14) stream we can use standard liblzma + * that is a part of XZ Utils. The stream format stored inside ZIPX + * file is a modified "lzma alone" file format, that was used by the + * `lzma` utility which was later deprecated in favour of `xz` utility. * Since those formats are nearly the same, we can use a standard + * "lzma alone" decoder from XZ Utils. */ + + memset(&zip->zipx_lzma_stream, 0, sizeof(zip->zipx_lzma_stream)); + r = lzma_alone_decoder(&zip->zipx_lzma_stream, UINT64_MAX); + if (r != LZMA_OK) { + archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, + "lzma initialization failed(%d)", r); + + return (ARCHIVE_FAILED); + } + + /* Flag the cleanup function that we want our lzma-related structures + * to be freed later. */ + zip->zipx_lzma_valid = 1; + + /* The "lzma alone" file format and the stream format inside ZIPx are + * almost the same. Here's an example of a structure of "lzma alone" + * format: + * + * $ cat /bin/ls | lzma | xxd | head -n 1 + * 00000000: 5d00 0080 00ff ffff ffff ffff ff00 2814 + * + * 5 bytes 8 bytes n bytes + * + * + * lzma_params is a 5-byte blob that has to be decoded to extract + * parameters of this LZMA stream. The uncompressed_size field is an + * uint64_t value that contains information about the size of the + * uncompressed file, or UINT64_MAX if this value is unknown. + * The part is the actual lzma-compressed data stream. + * + * Now here's the structure of the stream inside the ZIPX file: + * + * $ cat stream_inside_zipx | xxd | head -n 1 + * 00000000: 0914 0500 5d00 8000 0000 2814 .... .... + * + * 2byte 2byte 5 bytes n bytes + * + * + * This means that the ZIPX file contains an additional magic1 and + * magic2 headers, the lzma_params field contains the same parameter + * set as in the "lzma alone" format, and the field is the + * same as in the "lzma alone" format as well. Note that also the zipx + * format is missing the uncompressed_size field. + * + * So, in order to use the "lzma alone" decoder for the zipx lzma + * stream, we simply need to shuffle around some fields, prepare a new + * lzma alone header, feed it into lzma alone decoder so it will + * initialize itself properly, and then we can start feeding normal + * zipx lzma stream into the decoder. + */ + + /* Read magic1,magic2,lzma_params from the ZIPX stream. */ + if((p = __archive_read_ahead(a, 9, NULL)) == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated lzma data"); + return (ARCHIVE_FATAL); + } + + if(p[2] != 0x05 || p[3] != 0x00) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid lzma data"); + return (ARCHIVE_FATAL); + } + + /* Prepare an lzma alone header: copy the lzma_params blob into + * a proper place into the lzma alone header. */ + memcpy(&alone_header.bytes[0], p + 4, 5); + + /* Initialize the 'uncompressed size' field to unknown; we'll manually + * monitor how many bytes there are still to be uncompressed. */ + alone_header.uncompressed_size = UINT64_MAX; + + if(!zip->uncompressed_buffer) { + zip->uncompressed_buffer_size = 256 * 1024; + zip->uncompressed_buffer = + (uint8_t*) malloc(zip->uncompressed_buffer_size); + + if (zip->uncompressed_buffer == NULL) { + archive_set_error(&a->archive, ENOMEM, + "No memory for lzma decompression"); + return (ARCHIVE_FATAL); + } + } + + zip->zipx_lzma_stream.next_in = (void*) &alone_header; + zip->zipx_lzma_stream.avail_in = sizeof(alone_header); + zip->zipx_lzma_stream.total_in = 0; + zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer; + zip->zipx_lzma_stream.avail_out = zip->uncompressed_buffer_size; + zip->zipx_lzma_stream.total_out = 0; + + /* Feed only the header into the lzma alone decoder. This will + * effectively initialize the decoder, and will not produce any + * output bytes yet. */ + r = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN); + if (r != LZMA_OK) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, + "lzma stream initialization error"); + return ARCHIVE_FATAL; + } + + /* We've already consumed some bytes, so take this into account. */ + __archive_read_consume(a, 9); + zip->entry_bytes_remaining -= 9; + zip->entry_compressed_bytes_read += 9; + + zip->decompress_init = 1; + return (ARCHIVE_OK); +} + +static int +zip_read_data_zipx_xz(struct archive_read *a, const void **buff, + size_t *size, int64_t *offset) +{ + struct zip* zip = (struct zip *)(a->format->data); + int ret; + lzma_ret lz_ret; + const void* compressed_buf; + ssize_t bytes_avail, in_bytes, to_consume = 0; + + (void) offset; /* UNUSED */ + + /* Initialize decompressor if not yet initialized. */ + if (!zip->decompress_init) { + ret = zipx_xz_init(a, zip); + if (ret != ARCHIVE_OK) + return (ret); + } + + compressed_buf = __archive_read_ahead(a, 1, &bytes_avail); + if (bytes_avail < 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated xz file body"); + return (ARCHIVE_FATAL); + } + + in_bytes = zipmin(zip->entry_bytes_remaining, bytes_avail); + zip->zipx_lzma_stream.next_in = compressed_buf; + zip->zipx_lzma_stream.avail_in = in_bytes; + zip->zipx_lzma_stream.total_in = 0; + zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer; + zip->zipx_lzma_stream.avail_out = zip->uncompressed_buffer_size; + zip->zipx_lzma_stream.total_out = 0; + + /* Perform the decompression. */ + lz_ret = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN); + switch(lz_ret) { + case LZMA_DATA_ERROR: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "xz data error (error %d)", (int) lz_ret); + return (ARCHIVE_FATAL); + + case LZMA_NO_CHECK: + case LZMA_OK: + break; + + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "xz unknown error %d", (int) lz_ret); + return (ARCHIVE_FATAL); + + case LZMA_STREAM_END: + lzma_end(&zip->zipx_lzma_stream); + zip->zipx_lzma_valid = 0; + + if((int64_t) zip->zipx_lzma_stream.total_in != + zip->entry_bytes_remaining) + { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xz premature end of stream"); + return (ARCHIVE_FATAL); + } + + zip->end_of_entry = 1; + break; + } + + to_consume = zip->zipx_lzma_stream.total_in; + + __archive_read_consume(a, to_consume); + zip->entry_bytes_remaining -= to_consume; + zip->entry_compressed_bytes_read += to_consume; + zip->entry_uncompressed_bytes_read += zip->zipx_lzma_stream.total_out; + + *size = zip->zipx_lzma_stream.total_out; + *buff = zip->uncompressed_buffer; + + ret = consume_optional_marker(a, zip); + if (ret != ARCHIVE_OK) + return (ret); + + return (ARCHIVE_OK); +} + +static int +zip_read_data_zipx_lzma_alone(struct archive_read *a, const void **buff, + size_t *size, int64_t *offset) +{ + struct zip* zip = (struct zip *)(a->format->data); + int ret; + lzma_ret lz_ret; + const void* compressed_buf; + ssize_t bytes_avail, in_bytes, to_consume; + + (void) offset; /* UNUSED */ + + /* Initialize decompressor if not yet initialized. */ + if (!zip->decompress_init) { + ret = zipx_lzma_alone_init(a, zip); + if (ret != ARCHIVE_OK) + return (ret); + } + + /* Fetch more compressed data. The same note as in deflate handler + * applies here as well: + * + * Note: '1' here is a performance optimization. Recall that the + * decompression layer returns a count of available bytes; asking for + * more than that forces the decompressor to combine reads by copying + * data. + */ + compressed_buf = __archive_read_ahead(a, 1, &bytes_avail); + if (bytes_avail < 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated lzma file body"); + return (ARCHIVE_FATAL); + } + + /* Set decompressor parameters. */ + in_bytes = zipmin(zip->entry_bytes_remaining, bytes_avail); + + zip->zipx_lzma_stream.next_in = compressed_buf; + zip->zipx_lzma_stream.avail_in = in_bytes; + zip->zipx_lzma_stream.total_in = 0; + zip->zipx_lzma_stream.next_out = zip->uncompressed_buffer; + zip->zipx_lzma_stream.avail_out = + /* These lzma_alone streams lack end of stream marker, so let's + * make sure the unpacker won't try to unpack more than it's + * supposed to. */ + zipmin((int64_t) zip->uncompressed_buffer_size, + zip->entry->uncompressed_size - + zip->entry_uncompressed_bytes_read); + zip->zipx_lzma_stream.total_out = 0; + + /* Perform the decompression. */ + lz_ret = lzma_code(&zip->zipx_lzma_stream, LZMA_RUN); + switch(lz_ret) { + case LZMA_DATA_ERROR: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "lzma data error (error %d)", (int) lz_ret); + return (ARCHIVE_FATAL); + + /* This case is optional in lzma alone format. It can happen, + * but most of the files don't have it. (GitHub #1257) */ + case LZMA_STREAM_END: + lzma_end(&zip->zipx_lzma_stream); + zip->zipx_lzma_valid = 0; + if((int64_t) zip->zipx_lzma_stream.total_in != + zip->entry_bytes_remaining) + { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "lzma alone premature end of stream"); + return (ARCHIVE_FATAL); + } + + zip->end_of_entry = 1; + break; + + case LZMA_OK: + break; + + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "lzma unknown error %d", (int) lz_ret); + return (ARCHIVE_FATAL); + } + + to_consume = zip->zipx_lzma_stream.total_in; + + /* Update pointers. */ + __archive_read_consume(a, to_consume); + zip->entry_bytes_remaining -= to_consume; + zip->entry_compressed_bytes_read += to_consume; + zip->entry_uncompressed_bytes_read += zip->zipx_lzma_stream.total_out; + + if(zip->entry_bytes_remaining == 0) { + zip->end_of_entry = 1; + } + + /* Return values. */ + *size = zip->zipx_lzma_stream.total_out; + *buff = zip->uncompressed_buffer; + + /* Behave the same way as during deflate decompression. */ + ret = consume_optional_marker(a, zip); + if (ret != ARCHIVE_OK) + return (ret); + + /* Free lzma decoder handle because we'll no longer need it. */ + if(zip->end_of_entry) { + lzma_end(&zip->zipx_lzma_stream); + zip->zipx_lzma_valid = 0; + } + + /* If we're here, then we're good! */ + return (ARCHIVE_OK); +} +#endif /* HAVE_LZMA_H && HAVE_LIBLZMA */ + +static int +zipx_ppmd8_init(struct archive_read *a, struct zip *zip) +{ + const void* p; + uint32_t val; + uint32_t order; + uint32_t mem; + uint32_t restore_method; + + /* Remove previous decompression context if it exists. */ + if(zip->ppmd8_valid) { + __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8); + zip->ppmd8_valid = 0; + } + + /* Create a new decompression context. */ + __archive_ppmd8_functions.Ppmd8_Construct(&zip->ppmd8); + zip->ppmd8_stream_failed = 0; + + /* Setup function pointers required by Ppmd8 decompressor. The + * 'ppmd_read' function will feed new bytes to the decompressor, + * and will increment the 'zip->zipx_ppmd_read_compressed' counter. */ + zip->ppmd8.Stream.In = &zip->zipx_ppmd_stream; + zip->zipx_ppmd_stream.a = a; + zip->zipx_ppmd_stream.Read = &ppmd_read; + + /* Reset number of read bytes to 0. */ + zip->zipx_ppmd_read_compressed = 0; + + /* Read Ppmd8 header (2 bytes). */ + p = __archive_read_ahead(a, 2, NULL); + if(!p) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated file data in PPMd8 stream"); + return (ARCHIVE_FATAL); + } + __archive_read_consume(a, 2); + + /* Decode the stream's compression parameters. */ + val = archive_le16dec(p); + order = (val & 15) + 1; + mem = ((val >> 4) & 0xff) + 1; + restore_method = (val >> 12); + + if(order < 2 || restore_method > 2) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Invalid parameter set in PPMd8 stream (order=%d, " + "restore=%d)", order, restore_method); + return (ARCHIVE_FAILED); + } + + /* Allocate the memory needed to properly decompress the file. */ + if(!__archive_ppmd8_functions.Ppmd8_Alloc(&zip->ppmd8, mem << 20)) { + archive_set_error(&a->archive, ENOMEM, + "Unable to allocate memory for PPMd8 stream: %d bytes", + mem << 20); + return (ARCHIVE_FATAL); + } + + /* Signal the cleanup function to release Ppmd8 context in the + * cleanup phase. */ + zip->ppmd8_valid = 1; + + /* Perform further Ppmd8 initialization. */ + if(!__archive_ppmd8_functions.Ppmd8_RangeDec_Init(&zip->ppmd8)) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER, + "PPMd8 stream range decoder initialization error"); + return (ARCHIVE_FATAL); + } + + __archive_ppmd8_functions.Ppmd8_Init(&zip->ppmd8, order, + restore_method); + + /* Allocate the buffer that will hold uncompressed data. */ + free(zip->uncompressed_buffer); + + zip->uncompressed_buffer_size = 256 * 1024; + zip->uncompressed_buffer = + (uint8_t*) malloc(zip->uncompressed_buffer_size); + + if(zip->uncompressed_buffer == NULL) { + archive_set_error(&a->archive, ENOMEM, + "No memory for PPMd8 decompression"); + return ARCHIVE_FATAL; + } + + /* Ppmd8 initialization is done. */ + zip->decompress_init = 1; + + /* We've already read 2 bytes in the output stream. Additionally, + * Ppmd8 initialization code could read some data as well. So we + * are advancing the stream by 2 bytes plus whatever number of + * bytes Ppmd8 init function used. */ + zip->entry_compressed_bytes_read += 2 + zip->zipx_ppmd_read_compressed; + + return ARCHIVE_OK; +} + +static int +zip_read_data_zipx_ppmd(struct archive_read *a, const void **buff, + size_t *size, int64_t *offset) +{ + struct zip* zip = (struct zip *)(a->format->data); + int ret; + size_t consumed_bytes = 0; + ssize_t bytes_avail = 0; + + (void) offset; /* UNUSED */ + + /* If we're here for the first time, initialize Ppmd8 decompression + * context first. */ + if(!zip->decompress_init) { + ret = zipx_ppmd8_init(a, zip); + if(ret != ARCHIVE_OK) + return ret; + } + + /* Fetch for more data. We're reading 1 byte here, but libarchive + * should prefetch more bytes. */ + (void) __archive_read_ahead(a, 1, &bytes_avail); + if(bytes_avail < 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated PPMd8 file body"); + return (ARCHIVE_FATAL); + } + + /* This counter will be updated inside ppmd_read(), which at one + * point will be called by Ppmd8_DecodeSymbol. */ + zip->zipx_ppmd_read_compressed = 0; + + /* Decompression loop. */ + do { + int sym = __archive_ppmd8_functions.Ppmd8_DecodeSymbol( + &zip->ppmd8); + if(sym < 0) { + zip->end_of_entry = 1; + break; + } + + /* This field is set by ppmd_read() when there was no more data + * to be read. */ + if(zip->ppmd8_stream_failed) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated PPMd8 file body"); + return (ARCHIVE_FATAL); + } + + zip->uncompressed_buffer[consumed_bytes] = (uint8_t) sym; + ++consumed_bytes; + } while(consumed_bytes < zip->uncompressed_buffer_size); + + /* Update pointers for libarchive. */ + *buff = zip->uncompressed_buffer; + *size = consumed_bytes; + + /* Update pointers so we can continue decompression in another call. */ + zip->entry_bytes_remaining -= zip->zipx_ppmd_read_compressed; + zip->entry_compressed_bytes_read += zip->zipx_ppmd_read_compressed; + zip->entry_uncompressed_bytes_read += consumed_bytes; + + /* If we're at the end of stream, deinitialize Ppmd8 context. */ + if(zip->end_of_entry) { + __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8); + zip->ppmd8_valid = 0; + } + + /* Seek for optional marker, same way as in each zip entry. */ + ret = consume_optional_marker(a, zip); + if (ret != ARCHIVE_OK) + return ret; + + return ARCHIVE_OK; +} + +#ifdef HAVE_BZLIB_H +static int +zipx_bzip2_init(struct archive_read *a, struct zip *zip) +{ + int r; + + /* Deallocate already existing BZ2 decompression context if it + * exists. */ + if(zip->bzstream_valid) { + BZ2_bzDecompressEnd(&zip->bzstream); + zip->bzstream_valid = 0; + } + + /* Allocate a new BZ2 decompression context. */ + memset(&zip->bzstream, 0, sizeof(bz_stream)); + r = BZ2_bzDecompressInit(&zip->bzstream, 0, 1); + if(r != BZ_OK) { + archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, + "bzip2 initialization failed(%d)", + r); + + return ARCHIVE_FAILED; + } + + /* Mark the bzstream field to be released in cleanup phase. */ + zip->bzstream_valid = 1; + + /* (Re)allocate the buffer that will contain decompressed bytes. */ + free(zip->uncompressed_buffer); + + zip->uncompressed_buffer_size = 256 * 1024; + zip->uncompressed_buffer = + (uint8_t*) malloc(zip->uncompressed_buffer_size); + if (zip->uncompressed_buffer == NULL) { + archive_set_error(&a->archive, ENOMEM, + "No memory for bzip2 decompression"); + return ARCHIVE_FATAL; + } + + /* Initialization done. */ + zip->decompress_init = 1; + return ARCHIVE_OK; +} + +static int +zip_read_data_zipx_bzip2(struct archive_read *a, const void **buff, + size_t *size, int64_t *offset) +{ + struct zip *zip = (struct zip *)(a->format->data); + ssize_t bytes_avail = 0, in_bytes, to_consume; + const void *compressed_buff; + int r; + uint64_t total_out; + + (void) offset; /* UNUSED */ + + /* Initialize decompression context if we're here for the first time. */ + if(!zip->decompress_init) { + r = zipx_bzip2_init(a, zip); + if(r != ARCHIVE_OK) + return r; + } + + /* Fetch more compressed bytes. */ + compressed_buff = __archive_read_ahead(a, 1, &bytes_avail); + if(bytes_avail < 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated bzip2 file body"); + return (ARCHIVE_FATAL); + } + + in_bytes = zipmin(zip->entry_bytes_remaining, bytes_avail); + if(in_bytes < 1) { + /* libbz2 doesn't complain when caller feeds avail_in == 0. + * It will actually return success in this case, which is + * undesirable. This is why we need to make this check + * manually. */ + + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated bzip2 file body"); + return (ARCHIVE_FATAL); + } + + /* Setup buffer boundaries. */ + zip->bzstream.next_in = (char*)(uintptr_t) compressed_buff; + zip->bzstream.avail_in = in_bytes; + zip->bzstream.total_in_hi32 = 0; + zip->bzstream.total_in_lo32 = 0; + zip->bzstream.next_out = (char*) zip->uncompressed_buffer; + zip->bzstream.avail_out = zip->uncompressed_buffer_size; + zip->bzstream.total_out_hi32 = 0; + zip->bzstream.total_out_lo32 = 0; + + /* Perform the decompression. */ + r = BZ2_bzDecompress(&zip->bzstream); + switch(r) { + case BZ_STREAM_END: + /* If we're at the end of the stream, deinitialize the + * decompression context now. */ + switch(BZ2_bzDecompressEnd(&zip->bzstream)) { + case BZ_OK: + break; + default: + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Failed to clean up bzip2 " + "decompressor"); + return ARCHIVE_FATAL; + } + + zip->end_of_entry = 1; + break; + case BZ_OK: + /* The decompressor has successfully decoded this + * chunk of data, but more data is still in queue. */ + break; + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "bzip2 decompression failed"); + return ARCHIVE_FATAL; + } + + /* Update the pointers so decompressor can continue decoding. */ + to_consume = zip->bzstream.total_in_lo32; + __archive_read_consume(a, to_consume); + + total_out = ((uint64_t) zip->bzstream.total_out_hi32 << 32) + + zip->bzstream.total_out_lo32; + + zip->entry_bytes_remaining -= to_consume; + zip->entry_compressed_bytes_read += to_consume; + zip->entry_uncompressed_bytes_read += total_out; + + /* Give libarchive its due. */ + *size = total_out; + *buff = zip->uncompressed_buffer; + + /* Seek for optional marker, like in other entries. */ + r = consume_optional_marker(a, zip); + if(r != ARCHIVE_OK) + return r; + + return ARCHIVE_OK; +} + +#endif + +#ifdef HAVE_ZLIB_H +static int +zip_deflate_init(struct archive_read *a, struct zip *zip) +{ + int r; + + /* If we haven't yet read any data, initialize the decompressor. */ + if (!zip->decompress_init) { + if (zip->stream_valid) + r = inflateReset(&zip->stream); + else + r = inflateInit2(&zip->stream, + -15 /* Don't check for zlib header */); + if (r != Z_OK) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Can't initialize ZIP decompression."); + return (ARCHIVE_FATAL); + } + /* Stream structure has been set up. */ + zip->stream_valid = 1; + /* We've initialized decompression for this stream. */ + zip->decompress_init = 1; + } + return (ARCHIVE_OK); +} + +static int +zip_read_data_deflate(struct archive_read *a, const void **buff, + size_t *size, int64_t *offset) +{ + struct zip *zip; + ssize_t bytes_avail; + const void *compressed_buff, *sp; + int r; + + (void)offset; /* UNUSED */ + + zip = (struct zip *)(a->format->data); + + /* If the buffer hasn't been allocated, allocate it now. */ + if (zip->uncompressed_buffer == NULL) { + zip->uncompressed_buffer_size = 256 * 1024; + zip->uncompressed_buffer + = (unsigned char *)malloc(zip->uncompressed_buffer_size); + if (zip->uncompressed_buffer == NULL) { + archive_set_error(&a->archive, ENOMEM, + "No memory for ZIP decompression"); + return (ARCHIVE_FATAL); + } + } + + r = zip_deflate_init(a, zip); + if (r != ARCHIVE_OK) + return (r); + + /* + * Note: '1' here is a performance optimization. + * Recall that the decompression layer returns a count of + * available bytes; asking for more than that forces the + * decompressor to combine reads by copying data. + */ + compressed_buff = sp = __archive_read_ahead(a, 1, &bytes_avail); + if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) + && bytes_avail > zip->entry_bytes_remaining) { + bytes_avail = (ssize_t)zip->entry_bytes_remaining; + } + if (bytes_avail < 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated ZIP file body"); + return (ARCHIVE_FATAL); + } + + if (zip->tctx_valid || zip->cctx_valid) { + if (zip->decrypted_bytes_remaining < (size_t)bytes_avail) { + size_t buff_remaining = + (zip->decrypted_buffer + + zip->decrypted_buffer_size) + - (zip->decrypted_ptr + + zip->decrypted_bytes_remaining); + + if (buff_remaining > (size_t)bytes_avail) + buff_remaining = (size_t)bytes_avail; + + if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) && + zip->entry_bytes_remaining > 0) { + if ((int64_t)(zip->decrypted_bytes_remaining + + buff_remaining) + > zip->entry_bytes_remaining) { + if (zip->entry_bytes_remaining < + (int64_t)zip->decrypted_bytes_remaining) + buff_remaining = 0; + else + buff_remaining = + (size_t)zip->entry_bytes_remaining + - zip->decrypted_bytes_remaining; + } + } + if (buff_remaining > 0) { + if (zip->tctx_valid) { + trad_enc_decrypt_update(&zip->tctx, + compressed_buff, buff_remaining, + zip->decrypted_ptr + + zip->decrypted_bytes_remaining, + buff_remaining); + } else { + size_t dsize = buff_remaining; + archive_decrypto_aes_ctr_update( + &zip->cctx, + compressed_buff, buff_remaining, + zip->decrypted_ptr + + zip->decrypted_bytes_remaining, + &dsize); + } + zip->decrypted_bytes_remaining += + buff_remaining; + } + } + bytes_avail = zip->decrypted_bytes_remaining; + compressed_buff = (const char *)zip->decrypted_ptr; + } + + /* + * A bug in zlib.h: stream.next_in should be marked 'const' + * but isn't (the library never alters data through the + * next_in pointer, only reads it). The result: this ugly + * cast to remove 'const'. + */ + zip->stream.next_in = (Bytef *)(uintptr_t)(const void *)compressed_buff; + zip->stream.avail_in = (uInt)bytes_avail; + zip->stream.total_in = 0; + zip->stream.next_out = zip->uncompressed_buffer; + zip->stream.avail_out = (uInt)zip->uncompressed_buffer_size; + zip->stream.total_out = 0; + + r = inflate(&zip->stream, 0); + switch (r) { + case Z_OK: + break; + case Z_STREAM_END: + zip->end_of_entry = 1; + break; + case Z_MEM_ERROR: + archive_set_error(&a->archive, ENOMEM, + "Out of memory for ZIP decompression"); + return (ARCHIVE_FATAL); + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "ZIP decompression failed (%d)", r); + return (ARCHIVE_FATAL); + } + + /* Consume as much as the compressor actually used. */ + bytes_avail = zip->stream.total_in; + if (zip->tctx_valid || zip->cctx_valid) { + zip->decrypted_bytes_remaining -= bytes_avail; + if (zip->decrypted_bytes_remaining == 0) + zip->decrypted_ptr = zip->decrypted_buffer; + else + zip->decrypted_ptr += bytes_avail; + } + /* Calculate compressed data as much as we used.*/ + if (zip->hctx_valid) + archive_hmac_sha1_update(&zip->hctx, sp, bytes_avail); + __archive_read_consume(a, bytes_avail); + zip->entry_bytes_remaining -= bytes_avail; + zip->entry_compressed_bytes_read += bytes_avail; + + *size = zip->stream.total_out; + zip->entry_uncompressed_bytes_read += zip->stream.total_out; + *buff = zip->uncompressed_buffer; + + if (zip->end_of_entry && zip->hctx_valid) { + r = check_authentication_code(a, NULL); + if (r != ARCHIVE_OK) + return (r); + } + + r = consume_optional_marker(a, zip); + if (r != ARCHIVE_OK) + return (r); + + return (ARCHIVE_OK); +} +#endif + +static int +read_decryption_header(struct archive_read *a) +{ + struct zip *zip = (struct zip *)(a->format->data); + const char *p; + unsigned int remaining_size; + unsigned int ts; + + /* + * Read an initialization vector data field. + */ + p = __archive_read_ahead(a, 2, NULL); + if (p == NULL) + goto truncated; + ts = zip->iv_size; + zip->iv_size = archive_le16dec(p); + __archive_read_consume(a, 2); + if (ts < zip->iv_size) { + free(zip->iv); + zip->iv = NULL; + } + p = __archive_read_ahead(a, zip->iv_size, NULL); + if (p == NULL) + goto truncated; + if (zip->iv == NULL) { + zip->iv = malloc(zip->iv_size); + if (zip->iv == NULL) + goto nomem; + } + memcpy(zip->iv, p, zip->iv_size); + __archive_read_consume(a, zip->iv_size); + + /* + * Read a size of remaining decryption header field. + */ + p = __archive_read_ahead(a, 14, NULL); + if (p == NULL) + goto truncated; + remaining_size = archive_le32dec(p); + if (remaining_size < 16 || remaining_size > (1 << 18)) + goto corrupted; + + /* Check if format version is supported. */ + if (archive_le16dec(p+4) != 3) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Unsupported encryption format version: %u", + archive_le16dec(p+4)); + return (ARCHIVE_FAILED); + } + + /* + * Read an encryption algorithm field. + */ + zip->alg_id = archive_le16dec(p+6); + switch (zip->alg_id) { + case 0x6601:/* DES */ + case 0x6602:/* RC2 */ + case 0x6603:/* 3DES 168 */ + case 0x6609:/* 3DES 112 */ + case 0x660E:/* AES 128 */ + case 0x660F:/* AES 192 */ + case 0x6610:/* AES 256 */ + case 0x6702:/* RC2 (version >= 5.2) */ + case 0x6720:/* Blowfish */ + case 0x6721:/* Twofish */ + case 0x6801:/* RC4 */ + /* Supported encryption algorithm. */ + break; + default: + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Unknown encryption algorithm: %u", zip->alg_id); + return (ARCHIVE_FAILED); + } + + /* + * Read a bit length field. + */ + zip->bit_len = archive_le16dec(p+8); + + /* + * Read a flags field. + */ + zip->flags = archive_le16dec(p+10); + switch (zip->flags & 0xf000) { + case 0x0001: /* Password is required to decrypt. */ + case 0x0002: /* Certificates only. */ + case 0x0003: /* Password or certificate required to decrypt. */ + break; + default: + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Unknown encryption flag: %u", zip->flags); + return (ARCHIVE_FAILED); + } + if ((zip->flags & 0xf000) == 0 || + (zip->flags & 0xf000) == 0x4000) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Unknown encryption flag: %u", zip->flags); + return (ARCHIVE_FAILED); + } + + /* + * Read an encrypted random data field. + */ + ts = zip->erd_size; + zip->erd_size = archive_le16dec(p+12); + __archive_read_consume(a, 14); + if ((zip->erd_size & 0xf) != 0 || + (zip->erd_size + 16) > remaining_size || + (zip->erd_size + 16) < zip->erd_size) + goto corrupted; + + if (ts < zip->erd_size) { + free(zip->erd); + zip->erd = NULL; + } + p = __archive_read_ahead(a, zip->erd_size, NULL); + if (p == NULL) + goto truncated; + if (zip->erd == NULL) { + zip->erd = malloc(zip->erd_size); + if (zip->erd == NULL) + goto nomem; + } + memcpy(zip->erd, p, zip->erd_size); + __archive_read_consume(a, zip->erd_size); + + /* + * Read a reserved data field. + */ + p = __archive_read_ahead(a, 4, NULL); + if (p == NULL) + goto truncated; + /* Reserved data size should be zero. */ + if (archive_le32dec(p) != 0) + goto corrupted; + __archive_read_consume(a, 4); + + /* + * Read a password validation data field. + */ + p = __archive_read_ahead(a, 2, NULL); + if (p == NULL) + goto truncated; + ts = zip->v_size; + zip->v_size = archive_le16dec(p); + __archive_read_consume(a, 2); + if ((zip->v_size & 0x0f) != 0 || + (zip->erd_size + zip->v_size + 16) > remaining_size || + (zip->erd_size + zip->v_size + 16) < (zip->erd_size + zip->v_size)) + goto corrupted; + if (ts < zip->v_size) { + free(zip->v_data); + zip->v_data = NULL; + } + p = __archive_read_ahead(a, zip->v_size, NULL); + if (p == NULL) + goto truncated; + if (zip->v_data == NULL) { + zip->v_data = malloc(zip->v_size); + if (zip->v_data == NULL) + goto nomem; + } + memcpy(zip->v_data, p, zip->v_size); + __archive_read_consume(a, zip->v_size); + + p = __archive_read_ahead(a, 4, NULL); + if (p == NULL) + goto truncated; + zip->v_crc32 = archive_le32dec(p); + __archive_read_consume(a, 4); + + /*return (ARCHIVE_OK); + * This is not fully implemented yet.*/ + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Encrypted file is unsupported"); + return (ARCHIVE_FAILED); +truncated: + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated ZIP file data"); + return (ARCHIVE_FATAL); +corrupted: + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Corrupted ZIP file data"); + return (ARCHIVE_FATAL); +nomem: + archive_set_error(&a->archive, ENOMEM, + "No memory for ZIP decryption"); + return (ARCHIVE_FATAL); +} + +static int +zip_alloc_decryption_buffer(struct archive_read *a) +{ + struct zip *zip = (struct zip *)(a->format->data); + size_t bs = 256 * 1024; + + if (zip->decrypted_buffer == NULL) { + zip->decrypted_buffer_size = bs; + zip->decrypted_buffer = malloc(bs); + if (zip->decrypted_buffer == NULL) { + archive_set_error(&a->archive, ENOMEM, + "No memory for ZIP decryption"); + return (ARCHIVE_FATAL); + } + } + zip->decrypted_ptr = zip->decrypted_buffer; + return (ARCHIVE_OK); +} + +static int +init_traditional_PKWARE_decryption(struct archive_read *a) +{ + struct zip *zip = (struct zip *)(a->format->data); + const void *p; + int retry; + int r; + + if (zip->tctx_valid) + return (ARCHIVE_OK); + + /* + Read the 12 bytes encryption header stored at + the start of the data area. + */ +#define ENC_HEADER_SIZE 12 + if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) + && zip->entry_bytes_remaining < ENC_HEADER_SIZE) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated Zip encrypted body: only %jd bytes available", + (intmax_t)zip->entry_bytes_remaining); + return (ARCHIVE_FATAL); + } + + p = __archive_read_ahead(a, ENC_HEADER_SIZE, NULL); + if (p == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated ZIP file data"); + return (ARCHIVE_FATAL); + } + + for (retry = 0;; retry++) { + const char *passphrase; + uint8_t crcchk; + + passphrase = __archive_read_next_passphrase(a); + if (passphrase == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + (retry > 0)? + "Incorrect passphrase": + "Passphrase required for this entry"); + return (ARCHIVE_FAILED); + } + + /* + * Initialize ctx for Traditional PKWARE Decryption. + */ + r = trad_enc_init(&zip->tctx, passphrase, strlen(passphrase), + p, ENC_HEADER_SIZE, &crcchk); + if (r == 0 && crcchk == zip->entry->decdat) + break;/* The passphrase is OK. */ + if (retry > 10000) { + /* Avoid infinity loop. */ + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Too many incorrect passphrases"); + return (ARCHIVE_FAILED); + } + } + + __archive_read_consume(a, ENC_HEADER_SIZE); + zip->tctx_valid = 1; + if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { + zip->entry_bytes_remaining -= ENC_HEADER_SIZE; + } + /*zip->entry_uncompressed_bytes_read += ENC_HEADER_SIZE;*/ + zip->entry_compressed_bytes_read += ENC_HEADER_SIZE; + zip->decrypted_bytes_remaining = 0; + + return (zip_alloc_decryption_buffer(a)); +#undef ENC_HEADER_SIZE +} + +static int +init_WinZip_AES_decryption(struct archive_read *a) +{ + struct zip *zip = (struct zip *)(a->format->data); + const void *p; + const uint8_t *pv; + size_t key_len, salt_len; + uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; + int retry; + int r; + + if (zip->cctx_valid || zip->hctx_valid) + return (ARCHIVE_OK); + + switch (zip->entry->aes_extra.strength) { + case 1: salt_len = 8; key_len = 16; break; + case 2: salt_len = 12; key_len = 24; break; + case 3: salt_len = 16; key_len = 32; break; + default: goto corrupted; + } + p = __archive_read_ahead(a, salt_len + 2, NULL); + if (p == NULL) + goto truncated; + + for (retry = 0;; retry++) { + const char *passphrase; + + passphrase = __archive_read_next_passphrase(a); + if (passphrase == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + (retry > 0)? + "Incorrect passphrase": + "Passphrase required for this entry"); + return (ARCHIVE_FAILED); + } + memset(derived_key, 0, sizeof(derived_key)); + r = archive_pbkdf2_sha1(passphrase, strlen(passphrase), + p, salt_len, 1000, derived_key, key_len * 2 + 2); + if (r != 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Decryption is unsupported due to lack of " + "crypto library"); + return (ARCHIVE_FAILED); + } + + /* Check password verification value. */ + pv = ((const uint8_t *)p) + salt_len; + if (derived_key[key_len * 2] == pv[0] && + derived_key[key_len * 2 + 1] == pv[1]) + break;/* The passphrase is OK. */ + if (retry > 10000) { + /* Avoid infinity loop. */ + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Too many incorrect passphrases"); + return (ARCHIVE_FAILED); + } + } + + r = archive_decrypto_aes_ctr_init(&zip->cctx, derived_key, key_len); + if (r != 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Decryption is unsupported due to lack of crypto library"); + return (ARCHIVE_FAILED); + } + r = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len); + if (r != 0) { + archive_decrypto_aes_ctr_release(&zip->cctx); + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Failed to initialize HMAC-SHA1"); + return (ARCHIVE_FAILED); + } + zip->cctx_valid = zip->hctx_valid = 1; + __archive_read_consume(a, salt_len + 2); + zip->entry_bytes_remaining -= salt_len + 2 + AUTH_CODE_SIZE; + if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) + && zip->entry_bytes_remaining < 0) + goto corrupted; + zip->entry_compressed_bytes_read += salt_len + 2 + AUTH_CODE_SIZE; + zip->decrypted_bytes_remaining = 0; + + zip->entry->compression = zip->entry->aes_extra.compression; + return (zip_alloc_decryption_buffer(a)); + +truncated: + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated ZIP file data"); + return (ARCHIVE_FATAL); +corrupted: + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Corrupted ZIP file data"); + return (ARCHIVE_FATAL); +} + +static int +archive_read_format_zip_read_data(struct archive_read *a, + const void **buff, size_t *size, int64_t *offset) +{ + int r; + struct zip *zip = (struct zip *)(a->format->data); + + if (zip->has_encrypted_entries == + ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { + zip->has_encrypted_entries = 0; + } + + *offset = zip->entry_uncompressed_bytes_read; + *size = 0; + *buff = NULL; + + /* If we hit end-of-entry last time, return ARCHIVE_EOF. */ + if (zip->end_of_entry) + return (ARCHIVE_EOF); + + /* Return EOF immediately if this is a non-regular file. */ + if (AE_IFREG != (zip->entry->mode & AE_IFMT)) + return (ARCHIVE_EOF); + + __archive_read_consume(a, zip->unconsumed); + zip->unconsumed = 0; + + if (zip->init_decryption) { + zip->has_encrypted_entries = 1; + if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) + r = read_decryption_header(a); + else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) + r = init_WinZip_AES_decryption(a); + else + r = init_traditional_PKWARE_decryption(a); + if (r != ARCHIVE_OK) + return (r); + zip->init_decryption = 0; + } + + switch(zip->entry->compression) { + case 0: /* No compression. */ + r = zip_read_data_none(a, buff, size, offset); + break; +#ifdef HAVE_BZLIB_H + case 12: /* ZIPx bzip2 compression. */ + r = zip_read_data_zipx_bzip2(a, buff, size, offset); + break; +#endif +#if HAVE_LZMA_H && HAVE_LIBLZMA + case 14: /* ZIPx LZMA compression. */ + r = zip_read_data_zipx_lzma_alone(a, buff, size, offset); + break; + case 95: /* ZIPx XZ compression. */ + r = zip_read_data_zipx_xz(a, buff, size, offset); + break; +#endif + /* PPMd support is built-in, so we don't need any #if guards. */ + case 98: /* ZIPx PPMd compression. */ + r = zip_read_data_zipx_ppmd(a, buff, size, offset); + break; + +#ifdef HAVE_ZLIB_H + case 8: /* Deflate compression. */ + r = zip_read_data_deflate(a, buff, size, offset); + break; +#endif + default: /* Unsupported compression. */ + /* Return a warning. */ + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Unsupported ZIP compression method (%d: %s)", + zip->entry->compression, compression_name(zip->entry->compression)); + /* We can't decompress this entry, but we will + * be able to skip() it and try the next entry. */ + return (ARCHIVE_FAILED); + break; + } + if (r != ARCHIVE_OK) + return (r); + /* Update checksum */ + if (*size) + zip->entry_crc32 = zip->crc32func(zip->entry_crc32, *buff, + (unsigned)*size); + /* If we hit the end, swallow any end-of-data marker. */ + if (zip->end_of_entry) { + /* Check file size, CRC against these values. */ + if (zip->entry->compressed_size != + zip->entry_compressed_bytes_read) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "ZIP compressed data is wrong size " + "(read %jd, expected %jd)", + (intmax_t)zip->entry_compressed_bytes_read, + (intmax_t)zip->entry->compressed_size); + return (ARCHIVE_WARN); + } + /* Size field only stores the lower 32 bits of the actual + * size. */ + if ((zip->entry->uncompressed_size & UINT32_MAX) + != (zip->entry_uncompressed_bytes_read & UINT32_MAX)) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "ZIP uncompressed data is wrong size " + "(read %jd, expected %jd)\n", + (intmax_t)zip->entry_uncompressed_bytes_read, + (intmax_t)zip->entry->uncompressed_size); + return (ARCHIVE_WARN); + } + /* Check computed CRC against header */ + if ((!zip->hctx_valid || + zip->entry->aes_extra.vendor != AES_VENDOR_AE_2) && + zip->entry->crc32 != zip->entry_crc32 + && !zip->ignore_crc32) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "ZIP bad CRC: 0x%lx should be 0x%lx", + (unsigned long)zip->entry_crc32, + (unsigned long)zip->entry->crc32); + return (ARCHIVE_WARN); + } + } + + return (ARCHIVE_OK); +} + +static int +archive_read_format_zip_cleanup(struct archive_read *a) +{ + struct zip *zip; + struct zip_entry *zip_entry, *next_zip_entry; + + zip = (struct zip *)(a->format->data); + +#ifdef HAVE_ZLIB_H + if (zip->stream_valid) + inflateEnd(&zip->stream); +#endif + +#if HAVE_LZMA_H && HAVE_LIBLZMA + if (zip->zipx_lzma_valid) { + lzma_end(&zip->zipx_lzma_stream); + } +#endif + +#ifdef HAVE_BZLIB_H + if (zip->bzstream_valid) { + BZ2_bzDecompressEnd(&zip->bzstream); + } +#endif + + free(zip->uncompressed_buffer); + + if (zip->ppmd8_valid) + __archive_ppmd8_functions.Ppmd8_Free(&zip->ppmd8); + + if (zip->zip_entries) { + zip_entry = zip->zip_entries; + while (zip_entry != NULL) { + next_zip_entry = zip_entry->next; + archive_string_free(&zip_entry->rsrcname); + free(zip_entry); + zip_entry = next_zip_entry; + } + } + free(zip->decrypted_buffer); + if (zip->cctx_valid) + archive_decrypto_aes_ctr_release(&zip->cctx); + if (zip->hctx_valid) + archive_hmac_sha1_cleanup(&zip->hctx); + free(zip->iv); + free(zip->erd); + free(zip->v_data); + archive_string_free(&zip->format_name); + free(zip); + (a->format->data) = NULL; + return (ARCHIVE_OK); +} + +static int +archive_read_format_zip_has_encrypted_entries(struct archive_read *_a) +{ + if (_a && _a->format) { + struct zip * zip = (struct zip *)_a->format->data; + if (zip) { + return zip->has_encrypted_entries; + } + } + return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; +} + +static int +archive_read_format_zip_options(struct archive_read *a, + const char *key, const char *val) +{ + struct zip *zip; + int ret = ARCHIVE_FAILED; + + zip = (struct zip *)(a->format->data); + if (strcmp(key, "compat-2x") == 0) { + /* Handle filenames as libarchive 2.x */ + zip->init_default_conversion = (val != NULL) ? 1 : 0; + return (ARCHIVE_OK); + } else if (strcmp(key, "hdrcharset") == 0) { + if (val == NULL || val[0] == 0) + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "zip: hdrcharset option needs a character-set name" + ); + else { + zip->sconv = archive_string_conversion_from_charset( + &a->archive, val, 0); + if (zip->sconv != NULL) { + if (strcmp(val, "UTF-8") == 0) + zip->sconv_utf8 = zip->sconv; + ret = ARCHIVE_OK; + } else + ret = ARCHIVE_FATAL; + } + return (ret); + } else if (strcmp(key, "ignorecrc32") == 0) { + /* Mostly useful for testing. */ + if (val == NULL || val[0] == 0) { + zip->crc32func = real_crc32; + zip->ignore_crc32 = 0; + } else { + zip->crc32func = fake_crc32; + zip->ignore_crc32 = 1; + } + return (ARCHIVE_OK); + } else if (strcmp(key, "mac-ext") == 0) { + zip->process_mac_extensions = (val != NULL && val[0] != 0); + return (ARCHIVE_OK); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +int +archive_read_support_format_zip(struct archive *a) +{ + int r; + r = archive_read_support_format_zip_streamable(a); + if (r != ARCHIVE_OK) + return r; + return (archive_read_support_format_zip_seekable(a)); +} + +/* ------------------------------------------------------------------------ */ + +/* + * Streaming-mode support + */ + + +static int +archive_read_support_format_zip_capabilities_streamable(struct archive_read * a) +{ + (void)a; /* UNUSED */ + return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | + ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); +} + +static int +archive_read_format_zip_streamable_bid(struct archive_read *a, int best_bid) +{ + const char *p; + + (void)best_bid; /* UNUSED */ + + if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) + return (-1); + + /* + * Bid of 29 here comes from: + * + 16 bits for "PK", + * + next 16-bit field has 6 options so contributes + * about 16 - log_2(6) ~= 16 - 2.6 ~= 13 bits + * + * So we've effectively verified ~29 total bits of check data. + */ + if (p[0] == 'P' && p[1] == 'K') { + if ((p[2] == '\001' && p[3] == '\002') + || (p[2] == '\003' && p[3] == '\004') + || (p[2] == '\005' && p[3] == '\006') + || (p[2] == '\006' && p[3] == '\006') + || (p[2] == '\007' && p[3] == '\010') + || (p[2] == '0' && p[3] == '0')) + return (29); + } + + /* TODO: It's worth looking ahead a little bit for a valid + * PK signature. In particular, that would make it possible + * to read some UUEncoded SFX files or SFX files coming from + * a network socket. */ + + return (0); +} + +static int +archive_read_format_zip_streamable_read_header(struct archive_read *a, + struct archive_entry *entry) +{ + struct zip *zip; + + a->archive.archive_format = ARCHIVE_FORMAT_ZIP; + if (a->archive.archive_format_name == NULL) + a->archive.archive_format_name = "ZIP"; + + zip = (struct zip *)(a->format->data); + + /* + * It should be sufficient to call archive_read_next_header() for + * a reader to determine if an entry is encrypted or not. If the + * encryption of an entry is only detectable when calling + * archive_read_data(), so be it. We'll do the same check there + * as well. + */ + if (zip->has_encrypted_entries == + ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) + zip->has_encrypted_entries = 0; + + /* Make sure we have a zip_entry structure to use. */ + if (zip->zip_entries == NULL) { + zip->zip_entries = malloc(sizeof(struct zip_entry)); + if (zip->zip_entries == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Out of memory"); + return ARCHIVE_FATAL; + } + } + zip->entry = zip->zip_entries; + memset(zip->entry, 0, sizeof(struct zip_entry)); + + if (zip->cctx_valid) + archive_decrypto_aes_ctr_release(&zip->cctx); + if (zip->hctx_valid) + archive_hmac_sha1_cleanup(&zip->hctx); + zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; + __archive_read_reset_passphrase(a); + + /* Search ahead for the next local file header. */ + __archive_read_consume(a, zip->unconsumed); + zip->unconsumed = 0; + for (;;) { + int64_t skipped = 0; + const char *p, *end; + ssize_t bytes; + + p = __archive_read_ahead(a, 4, &bytes); + if (p == NULL) + return (ARCHIVE_FATAL); + end = p + bytes; + + while (p + 4 <= end) { + if (p[0] == 'P' && p[1] == 'K') { + if (p[2] == '\003' && p[3] == '\004') { + /* Regular file entry. */ + __archive_read_consume(a, skipped); + return zip_read_local_file_header(a, + entry, zip); + } + + /* + * TODO: We cannot restore permissions + * based only on the local file headers. + * Consider scanning the central + * directory and returning additional + * entries for at least directories. + * This would allow us to properly set + * directory permissions. + * + * This won't help us fix symlinks + * and may not help with regular file + * permissions, either. + */ + if (p[2] == '\001' && p[3] == '\002') { + return (ARCHIVE_EOF); + } + + /* End of central directory? Must be an + * empty archive. */ + if ((p[2] == '\005' && p[3] == '\006') + || (p[2] == '\006' && p[3] == '\006')) + return (ARCHIVE_EOF); + } + ++p; + ++skipped; + } + __archive_read_consume(a, skipped); + } +} + +static int +archive_read_format_zip_read_data_skip_streamable(struct archive_read *a) +{ + struct zip *zip; + int64_t bytes_skipped; + + zip = (struct zip *)(a->format->data); + bytes_skipped = __archive_read_consume(a, zip->unconsumed); + zip->unconsumed = 0; + if (bytes_skipped < 0) + return (ARCHIVE_FATAL); + + /* If we've already read to end of data, we're done. */ + if (zip->end_of_entry) + return (ARCHIVE_OK); + + /* So we know we're streaming... */ + if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) + || zip->entry->compressed_size > 0) { + /* We know the compressed length, so we can just skip. */ + bytes_skipped = __archive_read_consume(a, + zip->entry_bytes_remaining); + if (bytes_skipped < 0) + return (ARCHIVE_FATAL); + return (ARCHIVE_OK); + } + + if (zip->init_decryption) { + int r; + + zip->has_encrypted_entries = 1; + if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) + r = read_decryption_header(a); + else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) + r = init_WinZip_AES_decryption(a); + else + r = init_traditional_PKWARE_decryption(a); + if (r != ARCHIVE_OK) + return (r); + zip->init_decryption = 0; + } + + /* We're streaming and we don't know the length. */ + /* If the body is compressed and we know the format, we can + * find an exact end-of-entry by decompressing it. */ + switch (zip->entry->compression) { +#ifdef HAVE_ZLIB_H + case 8: /* Deflate compression. */ + while (!zip->end_of_entry) { + int64_t offset = 0; + const void *buff = NULL; + size_t size = 0; + int r; + r = zip_read_data_deflate(a, &buff, &size, &offset); + if (r != ARCHIVE_OK) + return (r); + } + return ARCHIVE_OK; +#endif + default: /* Uncompressed or unknown. */ + /* Scan for a PK\007\010 signature. */ + for (;;) { + const char *p, *buff; + ssize_t bytes_avail; + buff = __archive_read_ahead(a, 16, &bytes_avail); + if (bytes_avail < 16) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated ZIP file data"); + return (ARCHIVE_FATAL); + } + p = buff; + while (p <= buff + bytes_avail - 16) { + if (p[3] == 'P') { p += 3; } + else if (p[3] == 'K') { p += 2; } + else if (p[3] == '\007') { p += 1; } + else if (p[3] == '\010' && p[2] == '\007' + && p[1] == 'K' && p[0] == 'P') { + if (zip->entry->flags & LA_USED_ZIP64) + __archive_read_consume(a, + p - buff + 24); + else + __archive_read_consume(a, + p - buff + 16); + return ARCHIVE_OK; + } else { p += 4; } + } + __archive_read_consume(a, p - buff); + } + } +} + +int +archive_read_support_format_zip_streamable(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct zip *zip; + int r; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_format_zip"); + + zip = (struct zip *)calloc(1, sizeof(*zip)); + if (zip == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate zip data"); + return (ARCHIVE_FATAL); + } + + /* Streamable reader doesn't support mac extensions. */ + zip->process_mac_extensions = 0; + + /* + * Until enough data has been read, we cannot tell about + * any encrypted entries yet. + */ + zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; + zip->crc32func = real_crc32; + + r = __archive_read_register_format(a, + zip, + "zip", + archive_read_format_zip_streamable_bid, + archive_read_format_zip_options, + archive_read_format_zip_streamable_read_header, + archive_read_format_zip_read_data, + archive_read_format_zip_read_data_skip_streamable, + NULL, + archive_read_format_zip_cleanup, + archive_read_support_format_zip_capabilities_streamable, + archive_read_format_zip_has_encrypted_entries); + + if (r != ARCHIVE_OK) + free(zip); + return (ARCHIVE_OK); +} + +/* ------------------------------------------------------------------------ */ + +/* + * Seeking-mode support + */ + +static int +archive_read_support_format_zip_capabilities_seekable(struct archive_read * a) +{ + (void)a; /* UNUSED */ + return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | + ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); +} + +/* + * TODO: This is a performance sink because it forces the read core to + * drop buffered data from the start of file, which will then have to + * be re-read again if this bidder loses. + * + * We workaround this a little by passing in the best bid so far so + * that later bidders can do nothing if they know they'll never + * outbid. But we can certainly do better... + */ +static int +read_eocd(struct zip *zip, const char *p, int64_t current_offset) +{ + /* Sanity-check the EOCD we've found. */ + + /* This must be the first volume. */ + if (archive_le16dec(p + 4) != 0) + return 0; + /* Central directory must be on this volume. */ + if (archive_le16dec(p + 4) != archive_le16dec(p + 6)) + return 0; + /* All central directory entries must be on this volume. */ + if (archive_le16dec(p + 10) != archive_le16dec(p + 8)) + return 0; + /* Central directory can't extend beyond start of EOCD record. */ + if (archive_le32dec(p + 16) + archive_le32dec(p + 12) + > current_offset) + return 0; + + /* Save the central directory location for later use. */ + zip->central_directory_offset = archive_le32dec(p + 16); + + /* This is just a tiny bit higher than the maximum + returned by the streaming Zip bidder. This ensures + that the more accurate seeking Zip parser wins + whenever seek is available. */ + return 32; +} + +/* + * Examine Zip64 EOCD locator: If it's valid, store the information + * from it. + */ +static int +read_zip64_eocd(struct archive_read *a, struct zip *zip, const char *p) +{ + int64_t eocd64_offset; + int64_t eocd64_size; + + /* Sanity-check the locator record. */ + + /* Central dir must be on first volume. */ + if (archive_le32dec(p + 4) != 0) + return 0; + /* Must be only a single volume. */ + if (archive_le32dec(p + 16) != 1) + return 0; + + /* Find the Zip64 EOCD record. */ + eocd64_offset = archive_le64dec(p + 8); + if (__archive_read_seek(a, eocd64_offset, SEEK_SET) < 0) + return 0; + if ((p = __archive_read_ahead(a, 56, NULL)) == NULL) + return 0; + /* Make sure we can read all of it. */ + eocd64_size = archive_le64dec(p + 4) + 12; + if (eocd64_size < 56 || eocd64_size > 16384) + return 0; + if ((p = __archive_read_ahead(a, (size_t)eocd64_size, NULL)) == NULL) + return 0; + + /* Sanity-check the EOCD64 */ + if (archive_le32dec(p + 16) != 0) /* Must be disk #0 */ + return 0; + if (archive_le32dec(p + 20) != 0) /* CD must be on disk #0 */ + return 0; + /* CD can't be split. */ + if (archive_le64dec(p + 24) != archive_le64dec(p + 32)) + return 0; + + /* Save the central directory offset for later use. */ + zip->central_directory_offset = archive_le64dec(p + 48); + + return 32; +} + +static int +archive_read_format_zip_seekable_bid(struct archive_read *a, int best_bid) +{ + struct zip *zip = (struct zip *)a->format->data; + int64_t file_size, current_offset; + const char *p; + int i, tail; + + /* If someone has already bid more than 32, then avoid + trashing the look-ahead buffers with a seek. */ + if (best_bid > 32) + return (-1); + + file_size = __archive_read_seek(a, 0, SEEK_END); + if (file_size <= 0) + return 0; + + /* Search last 16k of file for end-of-central-directory + * record (which starts with PK\005\006) */ + tail = (int)zipmin(1024 * 16, file_size); + current_offset = __archive_read_seek(a, -tail, SEEK_END); + if (current_offset < 0) + return 0; + if ((p = __archive_read_ahead(a, (size_t)tail, NULL)) == NULL) + return 0; + /* Boyer-Moore search backwards from the end, since we want + * to match the last EOCD in the file (there can be more than + * one if there is an uncompressed Zip archive as a member + * within this Zip archive). */ + for (i = tail - 22; i > 0;) { + switch (p[i]) { + case 'P': + if (memcmp(p + i, "PK\005\006", 4) == 0) { + int ret = read_eocd(zip, p + i, + current_offset + i); + /* Zip64 EOCD locator precedes + * regular EOCD if present. */ + if (i >= 20 && memcmp(p + i - 20, "PK\006\007", 4) == 0) { + int ret_zip64 = read_zip64_eocd(a, zip, p + i - 20); + if (ret_zip64 > ret) + ret = ret_zip64; + } + return (ret); + } + i -= 4; + break; + case 'K': i -= 1; break; + case 005: i -= 2; break; + case 006: i -= 3; break; + default: i -= 4; break; + } + } + return 0; +} + +/* The red-black trees are only used in seeking mode to manage + * the in-memory copy of the central directory. */ + +static int +cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) +{ + const struct zip_entry *e1 = (const struct zip_entry *)n1; + const struct zip_entry *e2 = (const struct zip_entry *)n2; + + if (e1->local_header_offset > e2->local_header_offset) + return -1; + if (e1->local_header_offset < e2->local_header_offset) + return 1; + return 0; +} + +static int +cmp_key(const struct archive_rb_node *n, const void *key) +{ + /* This function won't be called */ + (void)n; /* UNUSED */ + (void)key; /* UNUSED */ + return 1; +} + +static const struct archive_rb_tree_ops rb_ops = { + &cmp_node, &cmp_key +}; + +static int +rsrc_cmp_node(const struct archive_rb_node *n1, + const struct archive_rb_node *n2) +{ + const struct zip_entry *e1 = (const struct zip_entry *)n1; + const struct zip_entry *e2 = (const struct zip_entry *)n2; + + return (strcmp(e2->rsrcname.s, e1->rsrcname.s)); +} + +static int +rsrc_cmp_key(const struct archive_rb_node *n, const void *key) +{ + const struct zip_entry *e = (const struct zip_entry *)n; + return (strcmp((const char *)key, e->rsrcname.s)); +} + +static const struct archive_rb_tree_ops rb_rsrc_ops = { + &rsrc_cmp_node, &rsrc_cmp_key +}; + +static const char * +rsrc_basename(const char *name, size_t name_length) +{ + const char *s, *r; + + r = s = name; + for (;;) { + s = memchr(s, '/', name_length - (s - name)); + if (s == NULL) + break; + r = ++s; + } + return (r); +} + +static void +expose_parent_dirs(struct zip *zip, const char *name, size_t name_length) +{ + struct archive_string str; + struct zip_entry *dir; + char *s; + + archive_string_init(&str); + archive_strncpy(&str, name, name_length); + for (;;) { + s = strrchr(str.s, '/'); + if (s == NULL) + break; + *s = '\0'; + /* Transfer the parent directory from zip->tree_rsrc RB + * tree to zip->tree RB tree to expose. */ + dir = (struct zip_entry *) + __archive_rb_tree_find_node(&zip->tree_rsrc, str.s); + if (dir == NULL) + break; + __archive_rb_tree_remove_node(&zip->tree_rsrc, &dir->node); + archive_string_free(&dir->rsrcname); + __archive_rb_tree_insert_node(&zip->tree, &dir->node); + } + archive_string_free(&str); +} + +static int +slurp_central_directory(struct archive_read *a, struct archive_entry* entry, + struct zip *zip) +{ + ssize_t i; + unsigned found; + int64_t correction; + ssize_t bytes_avail; + const char *p; + + /* + * Find the start of the central directory. The end-of-CD + * record has our starting point, but there are lots of + * Zip archives which have had other data prepended to the + * file, which makes the recorded offsets all too small. + * So we search forward from the specified offset until we + * find the real start of the central directory. Then we + * know the correction we need to apply to account for leading + * padding. + */ + if (__archive_read_seek(a, zip->central_directory_offset, SEEK_SET) < 0) + return ARCHIVE_FATAL; + + found = 0; + while (!found) { + if ((p = __archive_read_ahead(a, 20, &bytes_avail)) == NULL) + return ARCHIVE_FATAL; + for (found = 0, i = 0; !found && i < bytes_avail - 4;) { + switch (p[i + 3]) { + case 'P': i += 3; break; + case 'K': i += 2; break; + case 001: i += 1; break; + case 002: + if (memcmp(p + i, "PK\001\002", 4) == 0) { + p += i; + found = 1; + } else + i += 4; + break; + case 005: i += 1; break; + case 006: + if (memcmp(p + i, "PK\005\006", 4) == 0) { + p += i; + found = 1; + } else if (memcmp(p + i, "PK\006\006", 4) == 0) { + p += i; + found = 1; + } else + i += 1; + break; + default: i += 4; break; + } + } + __archive_read_consume(a, i); + } + correction = archive_filter_bytes(&a->archive, 0) + - zip->central_directory_offset; + + __archive_rb_tree_init(&zip->tree, &rb_ops); + __archive_rb_tree_init(&zip->tree_rsrc, &rb_rsrc_ops); + + zip->central_directory_entries_total = 0; + while (1) { + struct zip_entry *zip_entry; + size_t filename_length, extra_length, comment_length; + uint32_t external_attributes; + const char *name, *r; + + if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) + return ARCHIVE_FATAL; + if (memcmp(p, "PK\006\006", 4) == 0 + || memcmp(p, "PK\005\006", 4) == 0) { + break; + } else if (memcmp(p, "PK\001\002", 4) != 0) { + archive_set_error(&a->archive, + -1, "Invalid central directory signature"); + return ARCHIVE_FATAL; + } + if ((p = __archive_read_ahead(a, 46, NULL)) == NULL) + return ARCHIVE_FATAL; + + zip_entry = calloc(1, sizeof(struct zip_entry)); + if (zip_entry == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate zip entry"); + return ARCHIVE_FATAL; + } + zip_entry->next = zip->zip_entries; + zip_entry->flags |= LA_FROM_CENTRAL_DIRECTORY; + zip->zip_entries = zip_entry; + zip->central_directory_entries_total++; + + /* version = p[4]; */ + zip_entry->system = p[5]; + /* version_required = archive_le16dec(p + 6); */ + zip_entry->zip_flags = archive_le16dec(p + 8); + if (zip_entry->zip_flags + & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)){ + zip->has_encrypted_entries = 1; + } + zip_entry->compression = (char)archive_le16dec(p + 10); + zip_entry->mtime = zip_time(p + 12); + zip_entry->crc32 = archive_le32dec(p + 16); + if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) + zip_entry->decdat = p[13]; + else + zip_entry->decdat = p[19]; + zip_entry->compressed_size = archive_le32dec(p + 20); + zip_entry->uncompressed_size = archive_le32dec(p + 24); + filename_length = archive_le16dec(p + 28); + extra_length = archive_le16dec(p + 30); + comment_length = archive_le16dec(p + 32); + /* disk_start = archive_le16dec(p + 34); + * Better be zero. + * internal_attributes = archive_le16dec(p + 36); + * text bit */ + external_attributes = archive_le32dec(p + 38); + zip_entry->local_header_offset = + archive_le32dec(p + 42) + correction; + + /* If we can't guess the mode, leave it zero here; + when we read the local file header we might get + more information. */ + if (zip_entry->system == 3) { + zip_entry->mode = external_attributes >> 16; + } else if (zip_entry->system == 0) { + // Interpret MSDOS directory bit + if (0x10 == (external_attributes & 0x10)) { + zip_entry->mode = AE_IFDIR | 0775; + } else { + zip_entry->mode = AE_IFREG | 0664; + } + if (0x01 == (external_attributes & 0x01)) { + // Read-only bit; strip write permissions + zip_entry->mode &= 0555; + } + } else { + zip_entry->mode = 0; + } + + /* We're done with the regular data; get the filename and + * extra data. */ + __archive_read_consume(a, 46); + p = __archive_read_ahead(a, filename_length + extra_length, + NULL); + if (p == NULL) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated ZIP file header"); + return ARCHIVE_FATAL; + } + if (ARCHIVE_OK != process_extra(a, entry, p + filename_length, + extra_length, zip_entry)) { + return ARCHIVE_FATAL; + } + + /* + * Mac resource fork files are stored under the + * "__MACOSX/" directory, so we should check if + * it is. + */ + if (!zip->process_mac_extensions) { + /* Treat every entry as a regular entry. */ + __archive_rb_tree_insert_node(&zip->tree, + &zip_entry->node); + } else { + name = p; + r = rsrc_basename(name, filename_length); + if (filename_length >= 9 && + strncmp("__MACOSX/", name, 9) == 0) { + /* If this file is not a resource fork nor + * a directory. We should treat it as a non + * resource fork file to expose it. */ + if (name[filename_length-1] != '/' && + (r - name < 3 || r[0] != '.' || + r[1] != '_')) { + __archive_rb_tree_insert_node( + &zip->tree, &zip_entry->node); + /* Expose its parent directories. */ + expose_parent_dirs(zip, name, + filename_length); + } else { + /* This file is a resource fork file or + * a directory. */ + archive_strncpy(&(zip_entry->rsrcname), + name, filename_length); + __archive_rb_tree_insert_node( + &zip->tree_rsrc, &zip_entry->node); + } + } else { + /* Generate resource fork name to find its + * resource file at zip->tree_rsrc. */ + archive_strcpy(&(zip_entry->rsrcname), + "__MACOSX/"); + archive_strncat(&(zip_entry->rsrcname), + name, r - name); + archive_strcat(&(zip_entry->rsrcname), "._"); + archive_strncat(&(zip_entry->rsrcname), + name + (r - name), + filename_length - (r - name)); + /* Register an entry to RB tree to sort it by + * file offset. */ + __archive_rb_tree_insert_node(&zip->tree, + &zip_entry->node); + } + } + + /* Skip the comment too ... */ + __archive_read_consume(a, + filename_length + extra_length + comment_length); + } + + return ARCHIVE_OK; +} + +static ssize_t +zip_get_local_file_header_size(struct archive_read *a, size_t extra) +{ + const char *p; + ssize_t filename_length, extra_length; + + if ((p = __archive_read_ahead(a, extra + 30, NULL)) == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated ZIP file header"); + return (ARCHIVE_WARN); + } + p += extra; + + if (memcmp(p, "PK\003\004", 4) != 0) { + archive_set_error(&a->archive, -1, "Damaged Zip archive"); + return ARCHIVE_WARN; + } + filename_length = archive_le16dec(p + 26); + extra_length = archive_le16dec(p + 28); + + return (30 + filename_length + extra_length); +} + +static int +zip_read_mac_metadata(struct archive_read *a, struct archive_entry *entry, + struct zip_entry *rsrc) +{ + struct zip *zip = (struct zip *)a->format->data; + unsigned char *metadata, *mp; + int64_t offset = archive_filter_bytes(&a->archive, 0); + size_t remaining_bytes, metadata_bytes; + ssize_t hsize; + int ret = ARCHIVE_OK, eof; + + switch(rsrc->compression) { + case 0: /* No compression. */ + if (rsrc->uncompressed_size != rsrc->compressed_size) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Malformed OS X metadata entry: " + "inconsistent size"); + return (ARCHIVE_FATAL); + } +#ifdef HAVE_ZLIB_H + case 8: /* Deflate compression. */ +#endif + break; + default: /* Unsupported compression. */ + /* Return a warning. */ + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Unsupported ZIP compression method (%s)", + compression_name(rsrc->compression)); + /* We can't decompress this entry, but we will + * be able to skip() it and try the next entry. */ + return (ARCHIVE_WARN); + } + + if (rsrc->uncompressed_size > (4 * 1024 * 1024)) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Mac metadata is too large: %jd > 4M bytes", + (intmax_t)rsrc->uncompressed_size); + return (ARCHIVE_WARN); + } + if (rsrc->compressed_size > (4 * 1024 * 1024)) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Mac metadata is too large: %jd > 4M bytes", + (intmax_t)rsrc->compressed_size); + return (ARCHIVE_WARN); + } + + metadata = malloc((size_t)rsrc->uncompressed_size); + if (metadata == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Mac metadata"); + return (ARCHIVE_FATAL); + } + + if (offset < rsrc->local_header_offset) + __archive_read_consume(a, rsrc->local_header_offset - offset); + else if (offset != rsrc->local_header_offset) { + __archive_read_seek(a, rsrc->local_header_offset, SEEK_SET); + } + + hsize = zip_get_local_file_header_size(a, 0); + __archive_read_consume(a, hsize); + + remaining_bytes = (size_t)rsrc->compressed_size; + metadata_bytes = (size_t)rsrc->uncompressed_size; + mp = metadata; + eof = 0; + while (!eof && remaining_bytes) { + const unsigned char *p; + ssize_t bytes_avail; + size_t bytes_used; + + p = __archive_read_ahead(a, 1, &bytes_avail); + if (p == NULL) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Truncated ZIP file header"); + ret = ARCHIVE_WARN; + goto exit_mac_metadata; + } + if ((size_t)bytes_avail > remaining_bytes) + bytes_avail = remaining_bytes; + switch(rsrc->compression) { + case 0: /* No compression. */ + if ((size_t)bytes_avail > metadata_bytes) + bytes_avail = metadata_bytes; + memcpy(mp, p, bytes_avail); + bytes_used = (size_t)bytes_avail; + metadata_bytes -= bytes_used; + mp += bytes_used; + if (metadata_bytes == 0) + eof = 1; + break; +#ifdef HAVE_ZLIB_H + case 8: /* Deflate compression. */ + { + int r; + + ret = zip_deflate_init(a, zip); + if (ret != ARCHIVE_OK) + goto exit_mac_metadata; + zip->stream.next_in = + (Bytef *)(uintptr_t)(const void *)p; + zip->stream.avail_in = (uInt)bytes_avail; + zip->stream.total_in = 0; + zip->stream.next_out = mp; + zip->stream.avail_out = (uInt)metadata_bytes; + zip->stream.total_out = 0; + + r = inflate(&zip->stream, 0); + switch (r) { + case Z_OK: + break; + case Z_STREAM_END: + eof = 1; + break; + case Z_MEM_ERROR: + archive_set_error(&a->archive, ENOMEM, + "Out of memory for ZIP decompression"); + ret = ARCHIVE_FATAL; + goto exit_mac_metadata; + default: + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "ZIP decompression failed (%d)", r); + ret = ARCHIVE_FATAL; + goto exit_mac_metadata; + } + bytes_used = zip->stream.total_in; + metadata_bytes -= zip->stream.total_out; + mp += zip->stream.total_out; + break; + } +#endif + default: + bytes_used = 0; + break; + } + __archive_read_consume(a, bytes_used); + remaining_bytes -= bytes_used; + } + archive_entry_copy_mac_metadata(entry, metadata, + (size_t)rsrc->uncompressed_size - metadata_bytes); + +exit_mac_metadata: + __archive_read_seek(a, offset, SEEK_SET); + zip->decompress_init = 0; + free(metadata); + return (ret); +} + +static int +archive_read_format_zip_seekable_read_header(struct archive_read *a, + struct archive_entry *entry) +{ + struct zip *zip = (struct zip *)a->format->data; + struct zip_entry *rsrc; + int64_t offset; + int r, ret = ARCHIVE_OK; + + /* + * It should be sufficient to call archive_read_next_header() for + * a reader to determine if an entry is encrypted or not. If the + * encryption of an entry is only detectable when calling + * archive_read_data(), so be it. We'll do the same check there + * as well. + */ + if (zip->has_encrypted_entries == + ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) + zip->has_encrypted_entries = 0; + + a->archive.archive_format = ARCHIVE_FORMAT_ZIP; + if (a->archive.archive_format_name == NULL) + a->archive.archive_format_name = "ZIP"; + + if (zip->zip_entries == NULL) { + r = slurp_central_directory(a, entry, zip); + if (r != ARCHIVE_OK) + return r; + /* Get first entry whose local header offset is lower than + * other entries in the archive file. */ + zip->entry = + (struct zip_entry *)ARCHIVE_RB_TREE_MIN(&zip->tree); + } else if (zip->entry != NULL) { + /* Get next entry in local header offset order. */ + zip->entry = (struct zip_entry *)__archive_rb_tree_iterate( + &zip->tree, &zip->entry->node, ARCHIVE_RB_DIR_RIGHT); + } + + if (zip->entry == NULL) + return ARCHIVE_EOF; + + if (zip->entry->rsrcname.s) + rsrc = (struct zip_entry *)__archive_rb_tree_find_node( + &zip->tree_rsrc, zip->entry->rsrcname.s); + else + rsrc = NULL; + + if (zip->cctx_valid) + archive_decrypto_aes_ctr_release(&zip->cctx); + if (zip->hctx_valid) + archive_hmac_sha1_cleanup(&zip->hctx); + zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; + __archive_read_reset_passphrase(a); + + /* File entries are sorted by the header offset, we should mostly + * use __archive_read_consume to advance a read point to avoid + * redundant data reading. */ + offset = archive_filter_bytes(&a->archive, 0); + if (offset < zip->entry->local_header_offset) + __archive_read_consume(a, + zip->entry->local_header_offset - offset); + else if (offset != zip->entry->local_header_offset) { + __archive_read_seek(a, zip->entry->local_header_offset, + SEEK_SET); + } + zip->unconsumed = 0; + r = zip_read_local_file_header(a, entry, zip); + if (r != ARCHIVE_OK) + return r; + if (rsrc) { + int ret2 = zip_read_mac_metadata(a, entry, rsrc); + if (ret2 < ret) + ret = ret2; + } + return (ret); +} + +/* + * We're going to seek for the next header anyway, so we don't + * need to bother doing anything here. + */ +static int +archive_read_format_zip_read_data_skip_seekable(struct archive_read *a) +{ + struct zip *zip; + zip = (struct zip *)(a->format->data); + + zip->unconsumed = 0; + return (ARCHIVE_OK); +} + +int +archive_read_support_format_zip_seekable(struct archive *_a) +{ + struct archive_read *a = (struct archive_read *)_a; + struct zip *zip; + int r; + + archive_check_magic(_a, ARCHIVE_READ_MAGIC, + ARCHIVE_STATE_NEW, "archive_read_support_format_zip_seekable"); + + zip = (struct zip *)calloc(1, sizeof(*zip)); + if (zip == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate zip data"); + return (ARCHIVE_FATAL); + } + +#ifdef HAVE_COPYFILE_H + /* Set this by default on Mac OS. */ + zip->process_mac_extensions = 1; +#endif + + /* + * Until enough data has been read, we cannot tell about + * any encrypted entries yet. + */ + zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; + zip->crc32func = real_crc32; + + r = __archive_read_register_format(a, + zip, + "zip", + archive_read_format_zip_seekable_bid, + archive_read_format_zip_options, + archive_read_format_zip_seekable_read_header, + archive_read_format_zip_read_data, + archive_read_format_zip_read_data_skip_seekable, + NULL, + archive_read_format_zip_cleanup, + archive_read_support_format_zip_capabilities_seekable, + archive_read_format_zip_has_encrypted_entries); + + if (r != ARCHIVE_OK) + free(zip); + return (ARCHIVE_OK); +} + +/*# vim:set noet:*/ diff --git a/c/archive_string.c b/c/archive_string.c new file mode 100644 --- /dev/null +++ b/c/archive_string.c @@ -0,0 +1,4224 @@ +/*- + * Copyright (c) 2003-2011 Tim Kientzle + * Copyright (c) 2011-2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_string.c 201095 2009-12-28 02:33:22Z kientzle $"); + +/* + * Basic resizable string support, to simplify manipulating arbitrary-sized + * strings while minimizing heap activity. + * + * In particular, the buffer used by a string object is only grown, it + * never shrinks, so you can clear and reuse the same string object + * without incurring additional memory allocations. + */ + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_ICONV_H +#include +#endif +#ifdef HAVE_LANGINFO_H +#include +#endif +#ifdef HAVE_LOCALCHARSET_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_WCHAR_H +#include +#endif +#if defined(_WIN32) && !defined(__CYGWIN__) +#include +#include +#endif + +#include "archive_endian.h" +#include "archive_private.h" +#include "archive_string.h" +#include "archive_string_composition.h" + +#if !defined(HAVE_WMEMCPY) && !defined(wmemcpy) +#define wmemcpy(a,b,i) (wchar_t *)memcpy((a), (b), (i) * sizeof(wchar_t)) +#endif + +#if !defined(HAVE_WMEMMOVE) && !defined(wmemmove) +#define wmemmove(a,b,i) (wchar_t *)memmove((a), (b), (i) * sizeof(wchar_t)) +#endif + +#undef max +#define max(a, b) ((a)>(b)?(a):(b)) + +struct archive_string_conv { + struct archive_string_conv *next; + char *from_charset; + char *to_charset; + unsigned from_cp; + unsigned to_cp; + /* Set 1 if from_charset and to_charset are the same. */ + int same; + int flag; +#define SCONV_TO_CHARSET 1 /* MBS is being converted to specified + * charset. */ +#define SCONV_FROM_CHARSET (1<<1) /* MBS is being converted from + * specified charset. */ +#define SCONV_BEST_EFFORT (1<<2) /* Copy at least ASCII code. */ +#define SCONV_WIN_CP (1<<3) /* Use Windows API for converting + * MBS. */ +#define SCONV_UTF8_LIBARCHIVE_2 (1<<4) /* Incorrect UTF-8 made by libarchive + * 2.x in the wrong assumption. */ +#define SCONV_NORMALIZATION_C (1<<6) /* Need normalization to be Form C. + * Before UTF-8 characters are actually + * processed. */ +#define SCONV_NORMALIZATION_D (1<<7) /* Need normalization to be Form D. + * Before UTF-8 characters are actually + * processed. + * Currently this only for MAC OS X. */ +#define SCONV_TO_UTF8 (1<<8) /* "to charset" side is UTF-8. */ +#define SCONV_FROM_UTF8 (1<<9) /* "from charset" side is UTF-8. */ +#define SCONV_TO_UTF16BE (1<<10) /* "to charset" side is UTF-16BE. */ +#define SCONV_FROM_UTF16BE (1<<11) /* "from charset" side is UTF-16BE. */ +#define SCONV_TO_UTF16LE (1<<12) /* "to charset" side is UTF-16LE. */ +#define SCONV_FROM_UTF16LE (1<<13) /* "from charset" side is UTF-16LE. */ +#define SCONV_TO_UTF16 (SCONV_TO_UTF16BE | SCONV_TO_UTF16LE) +#define SCONV_FROM_UTF16 (SCONV_FROM_UTF16BE | SCONV_FROM_UTF16LE) + +#if HAVE_ICONV + iconv_t cd; + iconv_t cd_w;/* Use at archive_mstring on + * Windows. */ +#endif + /* A temporary buffer for normalization. */ + struct archive_string utftmp; + int (*converter[2])(struct archive_string *, const void *, size_t, + struct archive_string_conv *); + int nconverter; +}; + +#define CP_C_LOCALE 0 /* "C" locale only for this file. */ +#define CP_UTF16LE 1200 +#define CP_UTF16BE 1201 + +#define IS_HIGH_SURROGATE_LA(uc) ((uc) >= 0xD800 && (uc) <= 0xDBFF) +#define IS_LOW_SURROGATE_LA(uc) ((uc) >= 0xDC00 && (uc) <= 0xDFFF) +#define IS_SURROGATE_PAIR_LA(uc) ((uc) >= 0xD800 && (uc) <= 0xDFFF) +#define UNICODE_MAX 0x10FFFF +#define UNICODE_R_CHAR 0xFFFD /* Replacement character. */ +/* Set U+FFFD(Replacement character) in UTF-8. */ +static const char utf8_replacement_char[] = {0xef, 0xbf, 0xbd}; + +static struct archive_string_conv *find_sconv_object(struct archive *, + const char *, const char *); +static void add_sconv_object(struct archive *, struct archive_string_conv *); +static struct archive_string_conv *create_sconv_object(const char *, + const char *, unsigned, int); +static void free_sconv_object(struct archive_string_conv *); +static struct archive_string_conv *get_sconv_object(struct archive *, + const char *, const char *, int); +static unsigned make_codepage_from_charset(const char *); +static unsigned get_current_codepage(void); +static unsigned get_current_oemcp(void); +static size_t mbsnbytes(const void *, size_t); +static size_t utf16nbytes(const void *, size_t); +#if defined(_WIN32) && !defined(__CYGWIN__) +static int archive_wstring_append_from_mbs_in_codepage( + struct archive_wstring *, const char *, size_t, + struct archive_string_conv *); +static int archive_string_append_from_wcs_in_codepage(struct archive_string *, + const wchar_t *, size_t, struct archive_string_conv *); +static int is_big_endian(void); +static int strncat_in_codepage(struct archive_string *, const void *, + size_t, struct archive_string_conv *); +static int win_strncat_from_utf16be(struct archive_string *, const void *, + size_t, struct archive_string_conv *); +static int win_strncat_from_utf16le(struct archive_string *, const void *, + size_t, struct archive_string_conv *); +static int win_strncat_to_utf16be(struct archive_string *, const void *, + size_t, struct archive_string_conv *); +static int win_strncat_to_utf16le(struct archive_string *, const void *, + size_t, struct archive_string_conv *); +#endif +static int best_effort_strncat_from_utf16be(struct archive_string *, + const void *, size_t, struct archive_string_conv *); +static int best_effort_strncat_from_utf16le(struct archive_string *, + const void *, size_t, struct archive_string_conv *); +static int best_effort_strncat_to_utf16be(struct archive_string *, + const void *, size_t, struct archive_string_conv *); +static int best_effort_strncat_to_utf16le(struct archive_string *, + const void *, size_t, struct archive_string_conv *); +#if defined(HAVE_ICONV) +static int iconv_strncat_in_locale(struct archive_string *, const void *, + size_t, struct archive_string_conv *); +#endif +static int best_effort_strncat_in_locale(struct archive_string *, + const void *, size_t, struct archive_string_conv *); +static int _utf8_to_unicode(uint32_t *, const char *, size_t); +static int utf8_to_unicode(uint32_t *, const char *, size_t); +static inline uint32_t combine_surrogate_pair(uint32_t, uint32_t); +static int cesu8_to_unicode(uint32_t *, const char *, size_t); +static size_t unicode_to_utf8(char *, size_t, uint32_t); +static int utf16_to_unicode(uint32_t *, const char *, size_t, int); +static size_t unicode_to_utf16be(char *, size_t, uint32_t); +static size_t unicode_to_utf16le(char *, size_t, uint32_t); +static int strncat_from_utf8_libarchive2(struct archive_string *, + const void *, size_t, struct archive_string_conv *); +static int strncat_from_utf8_to_utf8(struct archive_string *, const void *, + size_t, struct archive_string_conv *); +static int archive_string_normalize_C(struct archive_string *, const void *, + size_t, struct archive_string_conv *); +static int archive_string_normalize_D(struct archive_string *, const void *, + size_t, struct archive_string_conv *); +static int archive_string_append_unicode(struct archive_string *, + const void *, size_t, struct archive_string_conv *); + +static struct archive_string * +archive_string_append(struct archive_string *as, const char *p, size_t s) +{ + if (archive_string_ensure(as, as->length + s + 1) == NULL) + return (NULL); + if (s) + memmove(as->s + as->length, p, s); + as->length += s; + as->s[as->length] = 0; + return (as); +} + +static struct archive_wstring * +archive_wstring_append(struct archive_wstring *as, const wchar_t *p, size_t s) +{ + if (archive_wstring_ensure(as, as->length + s + 1) == NULL) + return (NULL); + if (s) + wmemmove(as->s + as->length, p, s); + as->length += s; + as->s[as->length] = 0; + return (as); +} + +struct archive_string * +archive_array_append(struct archive_string *as, const char *p, size_t s) +{ + return archive_string_append(as, p, s); +} + +void +archive_string_concat(struct archive_string *dest, struct archive_string *src) +{ + if (archive_string_append(dest, src->s, src->length) == NULL) + __archive_errx(1, "Out of memory"); +} + +void +archive_wstring_concat(struct archive_wstring *dest, + struct archive_wstring *src) +{ + if (archive_wstring_append(dest, src->s, src->length) == NULL) + __archive_errx(1, "Out of memory"); +} + +void +archive_string_free(struct archive_string *as) +{ + as->length = 0; + as->buffer_length = 0; + free(as->s); + as->s = NULL; +} + +void +archive_wstring_free(struct archive_wstring *as) +{ + as->length = 0; + as->buffer_length = 0; + free(as->s); + as->s = NULL; +} + +struct archive_wstring * +archive_wstring_ensure(struct archive_wstring *as, size_t s) +{ + return (struct archive_wstring *) + archive_string_ensure((struct archive_string *)as, + s * sizeof(wchar_t)); +} + +/* Returns NULL on any allocation failure. */ +struct archive_string * +archive_string_ensure(struct archive_string *as, size_t s) +{ + char *p; + size_t new_length; + + /* If buffer is already big enough, don't reallocate. */ + if (as->s && (s <= as->buffer_length)) + return (as); + + /* + * Growing the buffer at least exponentially ensures that + * append operations are always linear in the number of + * characters appended. Using a smaller growth rate for + * larger buffers reduces memory waste somewhat at the cost of + * a larger constant factor. + */ + if (as->buffer_length < 32) + /* Start with a minimum 32-character buffer. */ + new_length = 32; + else if (as->buffer_length < 8192) + /* Buffers under 8k are doubled for speed. */ + new_length = as->buffer_length + as->buffer_length; + else { + /* Buffers 8k and over grow by at least 25% each time. */ + new_length = as->buffer_length + as->buffer_length / 4; + /* Be safe: If size wraps, fail. */ + if (new_length < as->buffer_length) { + /* On failure, wipe the string and return NULL. */ + archive_string_free(as); + errno = ENOMEM;/* Make sure errno has ENOMEM. */ + return (NULL); + } + } + /* + * The computation above is a lower limit to how much we'll + * grow the buffer. In any case, we have to grow it enough to + * hold the request. + */ + if (new_length < s) + new_length = s; + /* Now we can reallocate the buffer. */ + p = (char *)realloc(as->s, new_length); + if (p == NULL) { + /* On failure, wipe the string and return NULL. */ + archive_string_free(as); + errno = ENOMEM;/* Make sure errno has ENOMEM. */ + return (NULL); + } + + as->s = p; + as->buffer_length = new_length; + return (as); +} + +/* + * TODO: See if there's a way to avoid scanning + * the source string twice. Then test to see + * if it actually helps (remember that we're almost + * always called with pretty short arguments, so + * such an optimization might not help). + */ +struct archive_string * +archive_strncat(struct archive_string *as, const void *_p, size_t n) +{ + size_t s; + const char *p, *pp; + + p = (const char *)_p; + + /* Like strlen(p), except won't examine positions beyond p[n]. */ + s = 0; + pp = p; + while (s < n && *pp) { + pp++; + s++; + } + if ((as = archive_string_append(as, p, s)) == NULL) + __archive_errx(1, "Out of memory"); + return (as); +} + +struct archive_wstring * +archive_wstrncat(struct archive_wstring *as, const wchar_t *p, size_t n) +{ + size_t s; + const wchar_t *pp; + + /* Like strlen(p), except won't examine positions beyond p[n]. */ + s = 0; + pp = p; + while (s < n && *pp) { + pp++; + s++; + } + if ((as = archive_wstring_append(as, p, s)) == NULL) + __archive_errx(1, "Out of memory"); + return (as); +} + +struct archive_string * +archive_strcat(struct archive_string *as, const void *p) +{ + /* strcat is just strncat without an effective limit. + * Assert that we'll never get called with a source + * string over 16MB. + * TODO: Review all uses of strcat in the source + * and try to replace them with strncat(). + */ + return archive_strncat(as, p, 0x1000000); +} + +struct archive_wstring * +archive_wstrcat(struct archive_wstring *as, const wchar_t *p) +{ + /* Ditto. */ + return archive_wstrncat(as, p, 0x1000000); +} + +struct archive_string * +archive_strappend_char(struct archive_string *as, char c) +{ + if ((as = archive_string_append(as, &c, 1)) == NULL) + __archive_errx(1, "Out of memory"); + return (as); +} + +struct archive_wstring * +archive_wstrappend_wchar(struct archive_wstring *as, wchar_t c) +{ + if ((as = archive_wstring_append(as, &c, 1)) == NULL) + __archive_errx(1, "Out of memory"); + return (as); +} + +/* + * Get the "current character set" name to use with iconv. + * On FreeBSD, the empty character set name "" chooses + * the correct character encoding for the current locale, + * so this isn't necessary. + * But iconv on Mac OS 10.6 doesn't seem to handle this correctly; + * on that system, we have to explicitly call nl_langinfo() + * to get the right name. Not sure about other platforms. + * + * NOTE: GNU libiconv does not recognize the character-set name + * which some platform nl_langinfo(CODESET) returns, so we should + * use locale_charset() instead of nl_langinfo(CODESET) for GNU libiconv. + */ +static const char * +default_iconv_charset(const char *charset) { + if (charset != NULL && charset[0] != '\0') + return charset; +#if HAVE_LOCALE_CHARSET && !defined(__APPLE__) + /* locale_charset() is broken on Mac OS */ + return locale_charset(); +#elif HAVE_NL_LANGINFO + return nl_langinfo(CODESET); +#else + return ""; +#endif +} + +#if defined(_WIN32) && !defined(__CYGWIN__) + +/* + * Convert MBS to WCS. + * Note: returns -1 if conversion fails. + */ +int +archive_wstring_append_from_mbs(struct archive_wstring *dest, + const char *p, size_t len) +{ + return archive_wstring_append_from_mbs_in_codepage(dest, p, len, NULL); +} + +static int +archive_wstring_append_from_mbs_in_codepage(struct archive_wstring *dest, + const char *s, size_t length, struct archive_string_conv *sc) +{ + int count, ret = 0; + UINT from_cp; + + if (sc != NULL) + from_cp = sc->from_cp; + else + from_cp = get_current_codepage(); + + if (from_cp == CP_C_LOCALE) { + /* + * "C" locale special processing. + */ + wchar_t *ws; + const unsigned char *mp; + + if (NULL == archive_wstring_ensure(dest, + dest->length + length + 1)) + return (-1); + + ws = dest->s + dest->length; + mp = (const unsigned char *)s; + count = 0; + while (count < (int)length && *mp) { + *ws++ = (wchar_t)*mp++; + count++; + } + } else if (sc != NULL && + (sc->flag & (SCONV_NORMALIZATION_C | SCONV_NORMALIZATION_D))) { + /* + * Normalize UTF-8 and UTF-16BE and convert it directly + * to UTF-16 as wchar_t. + */ + struct archive_string u16; + int saved_flag = sc->flag;/* save current flag. */ + + if (is_big_endian()) + sc->flag |= SCONV_TO_UTF16BE; + else + sc->flag |= SCONV_TO_UTF16LE; + + if (sc->flag & SCONV_FROM_UTF16) { + /* + * UTF-16BE/LE NFD ===> UTF-16 NFC + * UTF-16BE/LE NFC ===> UTF-16 NFD + */ + count = (int)utf16nbytes(s, length); + } else { + /* + * UTF-8 NFD ===> UTF-16 NFC + * UTF-8 NFC ===> UTF-16 NFD + */ + count = (int)mbsnbytes(s, length); + } + u16.s = (char *)dest->s; + u16.length = dest->length << 1;; + u16.buffer_length = dest->buffer_length; + if (sc->flag & SCONV_NORMALIZATION_C) + ret = archive_string_normalize_C(&u16, s, count, sc); + else + ret = archive_string_normalize_D(&u16, s, count, sc); + dest->s = (wchar_t *)u16.s; + dest->length = u16.length >> 1; + dest->buffer_length = u16.buffer_length; + sc->flag = saved_flag;/* restore the saved flag. */ + return (ret); + } else if (sc != NULL && (sc->flag & SCONV_FROM_UTF16)) { + count = (int)utf16nbytes(s, length); + count >>= 1; /* to be WCS length */ + /* Allocate memory for WCS. */ + if (NULL == archive_wstring_ensure(dest, + dest->length + count + 1)) + return (-1); + wmemcpy(dest->s + dest->length, (const wchar_t *)s, count); + if ((sc->flag & SCONV_FROM_UTF16BE) && !is_big_endian()) { + uint16_t *u16 = (uint16_t *)(dest->s + dest->length); + int b; + for (b = 0; b < count; b++) { + uint16_t val = archive_le16dec(u16+b); + archive_be16enc(u16+b, val); + } + } else if ((sc->flag & SCONV_FROM_UTF16LE) && is_big_endian()) { + uint16_t *u16 = (uint16_t *)(dest->s + dest->length); + int b; + for (b = 0; b < count; b++) { + uint16_t val = archive_be16dec(u16+b); + archive_le16enc(u16+b, val); + } + } + } else { + DWORD mbflag; + size_t buffsize; + + if (sc == NULL) + mbflag = 0; + else if (sc->flag & SCONV_FROM_CHARSET) { + /* Do not trust the length which comes from + * an archive file. */ + length = mbsnbytes(s, length); + mbflag = 0; + } else + mbflag = MB_PRECOMPOSED; + + buffsize = dest->length + length + 1; + do { + /* Allocate memory for WCS. */ + if (NULL == archive_wstring_ensure(dest, buffsize)) + return (-1); + /* Convert MBS to WCS. */ + count = MultiByteToWideChar(from_cp, + mbflag, s, (int)length, dest->s + dest->length, + (int)(dest->buffer_length >> 1) -1); + if (count == 0 && + GetLastError() == ERROR_INSUFFICIENT_BUFFER) { + /* Expand the WCS buffer. */ + buffsize = dest->buffer_length << 1; + continue; + } + if (count == 0 && length != 0) + ret = -1; + break; + } while (1); + } + dest->length += count; + dest->s[dest->length] = L'\0'; + return (ret); +} + +#else + +/* + * Convert MBS to WCS. + * Note: returns -1 if conversion fails. + */ +int +archive_wstring_append_from_mbs(struct archive_wstring *dest, + const char *p, size_t len) +{ + size_t r; + int ret_val = 0; + /* + * No single byte will be more than one wide character, + * so this length estimate will always be big enough. + */ + // size_t wcs_length = len; + size_t mbs_length = len; + const char *mbs = p; + wchar_t *wcs; +#if HAVE_MBRTOWC + mbstate_t shift_state; + + memset(&shift_state, 0, sizeof(shift_state)); +#endif + /* + * As we decided to have wcs_length == mbs_length == len + * we can use len here instead of wcs_length + */ + if (NULL == archive_wstring_ensure(dest, dest->length + len + 1)) + return (-1); + wcs = dest->s + dest->length; + /* + * We cannot use mbsrtowcs/mbstowcs here because those may convert + * extra MBS when strlen(p) > len and one wide character consists of + * multi bytes. + */ + while (*mbs && mbs_length > 0) { + /* + * The buffer we allocated is always big enough. + * Keep this code path in a comment if we decide to choose + * smaller wcs_length in the future + */ +/* + if (wcs_length == 0) { + dest->length = wcs - dest->s; + dest->s[dest->length] = L'\0'; + wcs_length = mbs_length; + if (NULL == archive_wstring_ensure(dest, + dest->length + wcs_length + 1)) + return (-1); + wcs = dest->s + dest->length; + } +*/ +#if HAVE_MBRTOWC + r = mbrtowc(wcs, mbs, mbs_length, &shift_state); +#else + r = mbtowc(wcs, mbs, mbs_length); +#endif + if (r == (size_t)-1 || r == (size_t)-2) { + ret_val = -1; + break; + } + if (r == 0 || r > mbs_length) + break; + wcs++; + // wcs_length--; + mbs += r; + mbs_length -= r; + } + dest->length = wcs - dest->s; + dest->s[dest->length] = L'\0'; + return (ret_val); +} + +#endif + +#if defined(_WIN32) && !defined(__CYGWIN__) + +/* + * WCS ==> MBS. + * Note: returns -1 if conversion fails. + * + * Win32 builds use WideCharToMultiByte from the Windows API. + * (Maybe Cygwin should too? WideCharToMultiByte will know a + * lot more about local character encodings than the wcrtomb() + * wrapper is going to know.) + */ +int +archive_string_append_from_wcs(struct archive_string *as, + const wchar_t *w, size_t len) +{ + return archive_string_append_from_wcs_in_codepage(as, w, len, NULL); +} + +static int +archive_string_append_from_wcs_in_codepage(struct archive_string *as, + const wchar_t *ws, size_t len, struct archive_string_conv *sc) +{ + BOOL defchar_used, *dp; + int count, ret = 0; + UINT to_cp; + int wslen = (int)len; + + if (sc != NULL) + to_cp = sc->to_cp; + else + to_cp = get_current_codepage(); + + if (to_cp == CP_C_LOCALE) { + /* + * "C" locale special processing. + */ + const wchar_t *wp = ws; + char *p; + + if (NULL == archive_string_ensure(as, + as->length + wslen +1)) + return (-1); + p = as->s + as->length; + count = 0; + defchar_used = 0; + while (count < wslen && *wp) { + if (*wp > 255) { + *p++ = '?'; + wp++; + defchar_used = 1; + } else + *p++ = (char)*wp++; + count++; + } + } else if (sc != NULL && (sc->flag & SCONV_TO_UTF16)) { + uint16_t *u16; + + if (NULL == + archive_string_ensure(as, as->length + len * 2 + 2)) + return (-1); + u16 = (uint16_t *)(as->s + as->length); + count = 0; + defchar_used = 0; + if (sc->flag & SCONV_TO_UTF16BE) { + while (count < (int)len && *ws) { + archive_be16enc(u16+count, *ws); + ws++; + count++; + } + } else { + while (count < (int)len && *ws) { + archive_le16enc(u16+count, *ws); + ws++; + count++; + } + } + count <<= 1; /* to be byte size */ + } else { + /* Make sure the MBS buffer has plenty to set. */ + if (NULL == + archive_string_ensure(as, as->length + len * 2 + 1)) + return (-1); + do { + defchar_used = 0; + if (to_cp == CP_UTF8 || sc == NULL) + dp = NULL; + else + dp = &defchar_used; + count = WideCharToMultiByte(to_cp, 0, ws, wslen, + as->s + as->length, + (int)as->buffer_length - as->length - 1, NULL, dp); + if (count == 0 && + GetLastError() == ERROR_INSUFFICIENT_BUFFER) { + /* Expand the MBS buffer and retry. */ + if (NULL == archive_string_ensure(as, + as->buffer_length + len)) + return (-1); + continue; + } + if (count == 0) + ret = -1; + break; + } while (1); + } + as->length += count; + as->s[as->length] = '\0'; + return (defchar_used?-1:ret); +} + +#elif defined(HAVE_WCTOMB) || defined(HAVE_WCRTOMB) + +/* + * Translates a wide character string into current locale character set + * and appends to the archive_string. Note: returns -1 if conversion + * fails. + */ +int +archive_string_append_from_wcs(struct archive_string *as, + const wchar_t *w, size_t len) +{ + /* We cannot use the standard wcstombs() here because it + * cannot tell us how big the output buffer should be. So + * I've built a loop around wcrtomb() or wctomb() that + * converts a character at a time and resizes the string as + * needed. We prefer wcrtomb() when it's available because + * it's thread-safe. */ + int n, ret_val = 0; + char *p; + char *end; +#if HAVE_WCRTOMB + mbstate_t shift_state; + + memset(&shift_state, 0, sizeof(shift_state)); +#else + /* Clear the shift state before starting. */ + wctomb(NULL, L'\0'); +#endif + /* + * Allocate buffer for MBS. + * We need this allocation here since it is possible that + * as->s is still NULL. + */ + if (archive_string_ensure(as, as->length + len + 1) == NULL) + return (-1); + + p = as->s + as->length; + end = as->s + as->buffer_length - MB_CUR_MAX -1; + while (*w != L'\0' && len > 0) { + if (p >= end) { + as->length = p - as->s; + as->s[as->length] = '\0'; + /* Re-allocate buffer for MBS. */ + if (archive_string_ensure(as, + as->length + max(len * 2, + (size_t)MB_CUR_MAX) + 1) == NULL) + return (-1); + p = as->s + as->length; + end = as->s + as->buffer_length - MB_CUR_MAX -1; + } +#if HAVE_WCRTOMB + n = wcrtomb(p, *w++, &shift_state); +#else + n = wctomb(p, *w++); +#endif + if (n == -1) { + if (errno == EILSEQ) { + /* Skip an illegal wide char. */ + *p++ = '?'; + ret_val = -1; + } else { + ret_val = -1; + break; + } + } else + p += n; + len--; + } + as->length = p - as->s; + as->s[as->length] = '\0'; + return (ret_val); +} + +#else /* HAVE_WCTOMB || HAVE_WCRTOMB */ + +/* + * TODO: Test if __STDC_ISO_10646__ is defined. + * Non-Windows uses ISO C wcrtomb() or wctomb() to perform the conversion + * one character at a time. If a non-Windows platform doesn't have + * either of these, fall back to the built-in UTF8 conversion. + */ +int +archive_string_append_from_wcs(struct archive_string *as, + const wchar_t *w, size_t len) +{ + (void)as;/* UNUSED */ + (void)w;/* UNUSED */ + (void)len;/* UNUSED */ + errno = ENOSYS; + return (-1); +} + +#endif /* HAVE_WCTOMB || HAVE_WCRTOMB */ + +/* + * Find a string conversion object by a pair of 'from' charset name + * and 'to' charset name from an archive object. + * Return NULL if not found. + */ +static struct archive_string_conv * +find_sconv_object(struct archive *a, const char *fc, const char *tc) +{ + struct archive_string_conv *sc; + + if (a == NULL) + return (NULL); + + for (sc = a->sconv; sc != NULL; sc = sc->next) { + if (strcmp(sc->from_charset, fc) == 0 && + strcmp(sc->to_charset, tc) == 0) + break; + } + return (sc); +} + +/* + * Register a string object to an archive object. + */ +static void +add_sconv_object(struct archive *a, struct archive_string_conv *sc) +{ + struct archive_string_conv **psc; + + /* Add a new sconv to sconv list. */ + psc = &(a->sconv); + while (*psc != NULL) + psc = &((*psc)->next); + *psc = sc; +} + +static void +add_converter(struct archive_string_conv *sc, int (*converter) + (struct archive_string *, const void *, size_t, + struct archive_string_conv *)) +{ + if (sc == NULL || sc->nconverter >= 2) + __archive_errx(1, "Programming error"); + sc->converter[sc->nconverter++] = converter; +} + +static void +setup_converter(struct archive_string_conv *sc) +{ + + /* Reset. */ + sc->nconverter = 0; + + /* + * Perform special sequence for the incorrect UTF-8 filenames + * made by libarchive2.x. + */ + if (sc->flag & SCONV_UTF8_LIBARCHIVE_2) { + add_converter(sc, strncat_from_utf8_libarchive2); + return; + } + + /* + * Convert a string to UTF-16BE/LE. + */ + if (sc->flag & SCONV_TO_UTF16) { + /* + * If the current locale is UTF-8, we can translate + * a UTF-8 string into a UTF-16BE string. + */ + if (sc->flag & SCONV_FROM_UTF8) { + add_converter(sc, archive_string_append_unicode); + return; + } + +#if defined(_WIN32) && !defined(__CYGWIN__) + if (sc->flag & SCONV_WIN_CP) { + if (sc->flag & SCONV_TO_UTF16BE) + add_converter(sc, win_strncat_to_utf16be); + else + add_converter(sc, win_strncat_to_utf16le); + return; + } +#endif + +#if defined(HAVE_ICONV) + if (sc->cd != (iconv_t)-1) { + add_converter(sc, iconv_strncat_in_locale); + return; + } +#endif + + if (sc->flag & SCONV_BEST_EFFORT) { + if (sc->flag & SCONV_TO_UTF16BE) + add_converter(sc, + best_effort_strncat_to_utf16be); + else + add_converter(sc, + best_effort_strncat_to_utf16le); + } else + /* Make sure we have no converter. */ + sc->nconverter = 0; + return; + } + + /* + * Convert a string from UTF-16BE/LE. + */ + if (sc->flag & SCONV_FROM_UTF16) { + /* + * At least we should normalize a UTF-16BE string. + */ + if (sc->flag & SCONV_NORMALIZATION_D) + add_converter(sc,archive_string_normalize_D); + else if (sc->flag & SCONV_NORMALIZATION_C) + add_converter(sc, archive_string_normalize_C); + + if (sc->flag & SCONV_TO_UTF8) { + /* + * If the current locale is UTF-8, we can translate + * a UTF-16BE/LE string into a UTF-8 string directly. + */ + if (!(sc->flag & + (SCONV_NORMALIZATION_D |SCONV_NORMALIZATION_C))) + add_converter(sc, + archive_string_append_unicode); + return; + } + +#if defined(_WIN32) && !defined(__CYGWIN__) + if (sc->flag & SCONV_WIN_CP) { + if (sc->flag & SCONV_FROM_UTF16BE) + add_converter(sc, win_strncat_from_utf16be); + else + add_converter(sc, win_strncat_from_utf16le); + return; + } +#endif + +#if defined(HAVE_ICONV) + if (sc->cd != (iconv_t)-1) { + add_converter(sc, iconv_strncat_in_locale); + return; + } +#endif + + if ((sc->flag & (SCONV_BEST_EFFORT | SCONV_FROM_UTF16BE)) + == (SCONV_BEST_EFFORT | SCONV_FROM_UTF16BE)) + add_converter(sc, best_effort_strncat_from_utf16be); + else if ((sc->flag & (SCONV_BEST_EFFORT | SCONV_FROM_UTF16LE)) + == (SCONV_BEST_EFFORT | SCONV_FROM_UTF16LE)) + add_converter(sc, best_effort_strncat_from_utf16le); + else + /* Make sure we have no converter. */ + sc->nconverter = 0; + return; + } + + if (sc->flag & SCONV_FROM_UTF8) { + /* + * At least we should normalize a UTF-8 string. + */ + if (sc->flag & SCONV_NORMALIZATION_D) + add_converter(sc,archive_string_normalize_D); + else if (sc->flag & SCONV_NORMALIZATION_C) + add_converter(sc, archive_string_normalize_C); + + /* + * Copy UTF-8 string with a check of CESU-8. + * Apparently, iconv does not check surrogate pairs in UTF-8 + * when both from-charset and to-charset are UTF-8, and then + * we use our UTF-8 copy code. + */ + if (sc->flag & SCONV_TO_UTF8) { + /* + * If the current locale is UTF-8, we can translate + * a UTF-16BE string into a UTF-8 string directly. + */ + if (!(sc->flag & + (SCONV_NORMALIZATION_D |SCONV_NORMALIZATION_C))) + add_converter(sc, strncat_from_utf8_to_utf8); + return; + } + } + +#if defined(_WIN32) && !defined(__CYGWIN__) + /* + * On Windows we can use Windows API for a string conversion. + */ + if (sc->flag & SCONV_WIN_CP) { + add_converter(sc, strncat_in_codepage); + return; + } +#endif + +#if HAVE_ICONV + if (sc->cd != (iconv_t)-1) { + add_converter(sc, iconv_strncat_in_locale); + /* + * iconv generally does not support UTF-8-MAC and so + * we have to the output of iconv from NFC to NFD if + * need. + */ + if ((sc->flag & SCONV_FROM_CHARSET) && + (sc->flag & SCONV_TO_UTF8)) { + if (sc->flag & SCONV_NORMALIZATION_D) + add_converter(sc, archive_string_normalize_D); + } + return; + } +#endif + + /* + * Try conversion in the best effort or no conversion. + */ + if ((sc->flag & SCONV_BEST_EFFORT) || sc->same) + add_converter(sc, best_effort_strncat_in_locale); + else + /* Make sure we have no converter. */ + sc->nconverter = 0; +} + +/* + * Return canonicalized charset-name but this supports just UTF-8, UTF-16BE + * and CP932 which are referenced in create_sconv_object(). + */ +static const char * +canonical_charset_name(const char *charset) +{ + char cs[16]; + char *p; + const char *s; + + if (charset == NULL || charset[0] == '\0' + || strlen(charset) > 15) + return (charset); + + /* Copy name to uppercase. */ + p = cs; + s = charset; + while (*s) { + char c = *s++; + if (c >= 'a' && c <= 'z') + c -= 'a' - 'A'; + *p++ = c; + } + *p++ = '\0'; + + if (strcmp(cs, "UTF-8") == 0 || + strcmp(cs, "UTF8") == 0) + return ("UTF-8"); + if (strcmp(cs, "UTF-16BE") == 0 || + strcmp(cs, "UTF16BE") == 0) + return ("UTF-16BE"); + if (strcmp(cs, "UTF-16LE") == 0 || + strcmp(cs, "UTF16LE") == 0) + return ("UTF-16LE"); + if (strcmp(cs, "CP932") == 0) + return ("CP932"); + return (charset); +} + +/* + * Create a string conversion object. + */ +static struct archive_string_conv * +create_sconv_object(const char *fc, const char *tc, + unsigned current_codepage, int flag) +{ + struct archive_string_conv *sc; + + sc = calloc(1, sizeof(*sc)); + if (sc == NULL) + return (NULL); + sc->next = NULL; + sc->from_charset = strdup(fc); + if (sc->from_charset == NULL) { + free(sc); + return (NULL); + } + sc->to_charset = strdup(tc); + if (sc->to_charset == NULL) { + free(sc->from_charset); + free(sc); + return (NULL); + } + archive_string_init(&sc->utftmp); + + if (flag & SCONV_TO_CHARSET) { + /* + * Convert characters from the current locale charset to + * a specified charset. + */ + sc->from_cp = current_codepage; + sc->to_cp = make_codepage_from_charset(tc); +#if defined(_WIN32) && !defined(__CYGWIN__) + if (IsValidCodePage(sc->to_cp)) + flag |= SCONV_WIN_CP; +#endif + } else if (flag & SCONV_FROM_CHARSET) { + /* + * Convert characters from a specified charset to + * the current locale charset. + */ + sc->to_cp = current_codepage; + sc->from_cp = make_codepage_from_charset(fc); +#if defined(_WIN32) && !defined(__CYGWIN__) + if (IsValidCodePage(sc->from_cp)) + flag |= SCONV_WIN_CP; +#endif + } + + /* + * Check if "from charset" and "to charset" are the same. + */ + if (strcmp(fc, tc) == 0 || + (sc->from_cp != (unsigned)-1 && sc->from_cp == sc->to_cp)) + sc->same = 1; + else + sc->same = 0; + + /* + * Mark if "from charset" or "to charset" are UTF-8 or UTF-16BE/LE. + */ + if (strcmp(tc, "UTF-8") == 0) + flag |= SCONV_TO_UTF8; + else if (strcmp(tc, "UTF-16BE") == 0) + flag |= SCONV_TO_UTF16BE; + else if (strcmp(tc, "UTF-16LE") == 0) + flag |= SCONV_TO_UTF16LE; + if (strcmp(fc, "UTF-8") == 0) + flag |= SCONV_FROM_UTF8; + else if (strcmp(fc, "UTF-16BE") == 0) + flag |= SCONV_FROM_UTF16BE; + else if (strcmp(fc, "UTF-16LE") == 0) + flag |= SCONV_FROM_UTF16LE; +#if defined(_WIN32) && !defined(__CYGWIN__) + if (sc->to_cp == CP_UTF8) + flag |= SCONV_TO_UTF8; + else if (sc->to_cp == CP_UTF16BE) + flag |= SCONV_TO_UTF16BE | SCONV_WIN_CP; + else if (sc->to_cp == CP_UTF16LE) + flag |= SCONV_TO_UTF16LE | SCONV_WIN_CP; + if (sc->from_cp == CP_UTF8) + flag |= SCONV_FROM_UTF8; + else if (sc->from_cp == CP_UTF16BE) + flag |= SCONV_FROM_UTF16BE | SCONV_WIN_CP; + else if (sc->from_cp == CP_UTF16LE) + flag |= SCONV_FROM_UTF16LE | SCONV_WIN_CP; +#endif + + /* + * Set a flag for Unicode NFD. Usually iconv cannot correctly + * handle it. So we have to translate NFD characters to NFC ones + * ourselves before iconv handles. Another reason is to prevent + * that the same sight of two filenames, one is NFC and other + * is NFD, would be in its directory. + * On Mac OS X, although its filesystem layer automatically + * convert filenames to NFD, it would be useful for filename + * comparing to find out the same filenames that we normalize + * that to be NFD ourselves. + */ + if ((flag & SCONV_FROM_CHARSET) && + (flag & (SCONV_FROM_UTF16 | SCONV_FROM_UTF8))) { +#if defined(__APPLE__) + if (flag & SCONV_TO_UTF8) + flag |= SCONV_NORMALIZATION_D; + else +#endif + flag |= SCONV_NORMALIZATION_C; + } +#if defined(__APPLE__) + /* + * In case writing an archive file, make sure that a filename + * going to be passed to iconv is a Unicode NFC string since + * a filename in HFS Plus filesystem is a Unicode NFD one and + * iconv cannot handle it with "UTF-8" charset. It is simpler + * than a use of "UTF-8-MAC" charset. + */ + if ((flag & SCONV_TO_CHARSET) && + (flag & (SCONV_FROM_UTF16 | SCONV_FROM_UTF8)) && + !(flag & (SCONV_TO_UTF16 | SCONV_TO_UTF8))) + flag |= SCONV_NORMALIZATION_C; + /* + * In case reading an archive file. make sure that a filename + * will be passed to users is a Unicode NFD string in order to + * correctly compare the filename with other one which comes + * from HFS Plus filesystem. + */ + if ((flag & SCONV_FROM_CHARSET) && + !(flag & (SCONV_FROM_UTF16 | SCONV_FROM_UTF8)) && + (flag & SCONV_TO_UTF8)) + flag |= SCONV_NORMALIZATION_D; +#endif + +#if defined(HAVE_ICONV) + sc->cd_w = (iconv_t)-1; + /* + * Create an iconv object. + */ + if (((flag & (SCONV_TO_UTF8 | SCONV_TO_UTF16)) && + (flag & (SCONV_FROM_UTF8 | SCONV_FROM_UTF16))) || + (flag & SCONV_WIN_CP)) { + /* This case we won't use iconv. */ + sc->cd = (iconv_t)-1; + } else { + sc->cd = iconv_open(tc, fc); + if (sc->cd == (iconv_t)-1 && (sc->flag & SCONV_BEST_EFFORT)) { + /* + * Unfortunately, all of iconv implements do support + * "CP932" character-set, so we should use "SJIS" + * instead if iconv_open failed. + */ + if (strcmp(tc, "CP932") == 0) + sc->cd = iconv_open("SJIS", fc); + else if (strcmp(fc, "CP932") == 0) + sc->cd = iconv_open(tc, "SJIS"); + } +#if defined(_WIN32) && !defined(__CYGWIN__) + /* + * archive_mstring on Windows directly convert multi-bytes + * into archive_wstring in order not to depend on locale + * so that you can do a I18N programming. This will be + * used only in archive_mstring_copy_mbs_len_l so far. + */ + if (flag & SCONV_FROM_CHARSET) { + sc->cd_w = iconv_open("UTF-8", fc); + if (sc->cd_w == (iconv_t)-1 && + (sc->flag & SCONV_BEST_EFFORT)) { + if (strcmp(fc, "CP932") == 0) + sc->cd_w = iconv_open("UTF-8", "SJIS"); + } + } +#endif /* _WIN32 && !__CYGWIN__ */ + } +#endif /* HAVE_ICONV */ + + sc->flag = flag; + + /* + * Set up converters. + */ + setup_converter(sc); + + return (sc); +} + +/* + * Free a string conversion object. + */ +static void +free_sconv_object(struct archive_string_conv *sc) +{ + free(sc->from_charset); + free(sc->to_charset); + archive_string_free(&sc->utftmp); +#if HAVE_ICONV + if (sc->cd != (iconv_t)-1) + iconv_close(sc->cd); + if (sc->cd_w != (iconv_t)-1) + iconv_close(sc->cd_w); +#endif + free(sc); +} + +#if defined(_WIN32) && !defined(__CYGWIN__) +static unsigned +my_atoi(const char *p) +{ + unsigned cp; + + cp = 0; + while (*p) { + if (*p >= '0' && *p <= '9') + cp = cp * 10 + (*p - '0'); + else + return (-1); + p++; + } + return (cp); +} + +/* + * Translate Charset name (as used by iconv) into CodePage (as used by Windows) + * Return -1 if failed. + * + * Note: This translation code may be insufficient. + */ +static struct charset { + const char *name; + unsigned cp; +} charsets[] = { + /* MUST BE SORTED! */ + {"ASCII", 1252}, + {"ASMO-708", 708}, + {"BIG5", 950}, + {"CHINESE", 936}, + {"CP367", 1252}, + {"CP819", 1252}, + {"CP1025", 21025}, + {"DOS-720", 720}, + {"DOS-862", 862}, + {"EUC-CN", 51936}, + {"EUC-JP", 51932}, + {"EUC-KR", 949}, + {"EUCCN", 51936}, + {"EUCJP", 51932}, + {"EUCKR", 949}, + {"GB18030", 54936}, + {"GB2312", 936}, + {"HEBREW", 1255}, + {"HZ-GB-2312", 52936}, + {"IBM273", 20273}, + {"IBM277", 20277}, + {"IBM278", 20278}, + {"IBM280", 20280}, + {"IBM284", 20284}, + {"IBM285", 20285}, + {"IBM290", 20290}, + {"IBM297", 20297}, + {"IBM367", 1252}, + {"IBM420", 20420}, + {"IBM423", 20423}, + {"IBM424", 20424}, + {"IBM819", 1252}, + {"IBM871", 20871}, + {"IBM880", 20880}, + {"IBM905", 20905}, + {"IBM924", 20924}, + {"ISO-8859-1", 28591}, + {"ISO-8859-13", 28603}, + {"ISO-8859-15", 28605}, + {"ISO-8859-2", 28592}, + {"ISO-8859-3", 28593}, + {"ISO-8859-4", 28594}, + {"ISO-8859-5", 28595}, + {"ISO-8859-6", 28596}, + {"ISO-8859-7", 28597}, + {"ISO-8859-8", 28598}, + {"ISO-8859-9", 28599}, + {"ISO8859-1", 28591}, + {"ISO8859-13", 28603}, + {"ISO8859-15", 28605}, + {"ISO8859-2", 28592}, + {"ISO8859-3", 28593}, + {"ISO8859-4", 28594}, + {"ISO8859-5", 28595}, + {"ISO8859-6", 28596}, + {"ISO8859-7", 28597}, + {"ISO8859-8", 28598}, + {"ISO8859-9", 28599}, + {"JOHAB", 1361}, + {"KOI8-R", 20866}, + {"KOI8-U", 21866}, + {"KS_C_5601-1987", 949}, + {"LATIN1", 1252}, + {"LATIN2", 28592}, + {"MACINTOSH", 10000}, + {"SHIFT-JIS", 932}, + {"SHIFT_JIS", 932}, + {"SJIS", 932}, + {"US", 1252}, + {"US-ASCII", 1252}, + {"UTF-16", 1200}, + {"UTF-16BE", 1201}, + {"UTF-16LE", 1200}, + {"UTF-8", CP_UTF8}, + {"X-EUROPA", 29001}, + {"X-MAC-ARABIC", 10004}, + {"X-MAC-CE", 10029}, + {"X-MAC-CHINESEIMP", 10008}, + {"X-MAC-CHINESETRAD", 10002}, + {"X-MAC-CROATIAN", 10082}, + {"X-MAC-CYRILLIC", 10007}, + {"X-MAC-GREEK", 10006}, + {"X-MAC-HEBREW", 10005}, + {"X-MAC-ICELANDIC", 10079}, + {"X-MAC-JAPANESE", 10001}, + {"X-MAC-KOREAN", 10003}, + {"X-MAC-ROMANIAN", 10010}, + {"X-MAC-THAI", 10021}, + {"X-MAC-TURKISH", 10081}, + {"X-MAC-UKRAINIAN", 10017}, +}; +static unsigned +make_codepage_from_charset(const char *charset) +{ + char cs[16]; + char *p; + unsigned cp; + int a, b; + + if (charset == NULL || strlen(charset) > 15) + return -1; + + /* Copy name to uppercase. */ + p = cs; + while (*charset) { + char c = *charset++; + if (c >= 'a' && c <= 'z') + c -= 'a' - 'A'; + *p++ = c; + } + *p++ = '\0'; + cp = -1; + + /* Look it up in the table first, so that we can easily + * override CP367, which we map to 1252 instead of 367. */ + a = 0; + b = sizeof(charsets)/sizeof(charsets[0]); + while (b > a) { + int c = (b + a) / 2; + int r = strcmp(charsets[c].name, cs); + if (r < 0) + a = c + 1; + else if (r > 0) + b = c; + else + return charsets[c].cp; + } + + /* If it's not in the table, try to parse it. */ + switch (*cs) { + case 'C': + if (cs[1] == 'P' && cs[2] >= '0' && cs[2] <= '9') { + cp = my_atoi(cs + 2); + } else if (strcmp(cs, "CP_ACP") == 0) + cp = get_current_codepage(); + else if (strcmp(cs, "CP_OEMCP") == 0) + cp = get_current_oemcp(); + break; + case 'I': + if (cs[1] == 'B' && cs[2] == 'M' && + cs[3] >= '0' && cs[3] <= '9') { + cp = my_atoi(cs + 3); + } + break; + case 'W': + if (strncmp(cs, "WINDOWS-", 8) == 0) { + cp = my_atoi(cs + 8); + if (cp != 874 && (cp < 1250 || cp > 1258)) + cp = -1;/* This may invalid code. */ + } + break; + } + return (cp); +} + +/* + * Return ANSI Code Page of current locale set by setlocale(). + */ +static unsigned +get_current_codepage(void) +{ + char *locale, *p; + unsigned cp; + + locale = setlocale(LC_CTYPE, NULL); + if (locale == NULL) + return (GetACP()); + if (locale[0] == 'C' && locale[1] == '\0') + return (CP_C_LOCALE); + p = strrchr(locale, '.'); + if (p == NULL) + return (GetACP()); + if (strcmp(p+1, "utf8") == 0) + return CP_UTF8; + cp = my_atoi(p+1); + if ((int)cp <= 0) + return (GetACP()); + return (cp); +} + +/* + * Translation table between Locale Name and ACP/OEMCP. + */ +static struct { + unsigned acp; + unsigned ocp; + const char *locale; +} acp_ocp_map[] = { + { 950, 950, "Chinese_Taiwan" }, + { 936, 936, "Chinese_People's Republic of China" }, + { 950, 950, "Chinese_Taiwan" }, + { 1250, 852, "Czech_Czech Republic" }, + { 1252, 850, "Danish_Denmark" }, + { 1252, 850, "Dutch_Netherlands" }, + { 1252, 850, "Dutch_Belgium" }, + { 1252, 437, "English_United States" }, + { 1252, 850, "English_Australia" }, + { 1252, 850, "English_Canada" }, + { 1252, 850, "English_New Zealand" }, + { 1252, 850, "English_United Kingdom" }, + { 1252, 437, "English_United States" }, + { 1252, 850, "Finnish_Finland" }, + { 1252, 850, "French_France" }, + { 1252, 850, "French_Belgium" }, + { 1252, 850, "French_Canada" }, + { 1252, 850, "French_Switzerland" }, + { 1252, 850, "German_Germany" }, + { 1252, 850, "German_Austria" }, + { 1252, 850, "German_Switzerland" }, + { 1253, 737, "Greek_Greece" }, + { 1250, 852, "Hungarian_Hungary" }, + { 1252, 850, "Icelandic_Iceland" }, + { 1252, 850, "Italian_Italy" }, + { 1252, 850, "Italian_Switzerland" }, + { 932, 932, "Japanese_Japan" }, + { 949, 949, "Korean_Korea" }, + { 1252, 850, "Norwegian (BokmOl)_Norway" }, + { 1252, 850, "Norwegian (BokmOl)_Norway" }, + { 1252, 850, "Norwegian-Nynorsk_Norway" }, + { 1250, 852, "Polish_Poland" }, + { 1252, 850, "Portuguese_Portugal" }, + { 1252, 850, "Portuguese_Brazil" }, + { 1251, 866, "Russian_Russia" }, + { 1250, 852, "Slovak_Slovakia" }, + { 1252, 850, "Spanish_Spain" }, + { 1252, 850, "Spanish_Mexico" }, + { 1252, 850, "Spanish_Spain" }, + { 1252, 850, "Swedish_Sweden" }, + { 1254, 857, "Turkish_Turkey" }, + { 0, 0, NULL} +}; + +/* + * Return OEM Code Page of current locale set by setlocale(). + */ +static unsigned +get_current_oemcp(void) +{ + int i; + char *locale, *p; + size_t len; + + locale = setlocale(LC_CTYPE, NULL); + if (locale == NULL) + return (GetOEMCP()); + if (locale[0] == 'C' && locale[1] == '\0') + return (CP_C_LOCALE); + + p = strrchr(locale, '.'); + if (p == NULL) + return (GetOEMCP()); + len = p - locale; + for (i = 0; acp_ocp_map[i].acp; i++) { + if (strncmp(acp_ocp_map[i].locale, locale, len) == 0) + return (acp_ocp_map[i].ocp); + } + return (GetOEMCP()); +} +#else + +/* + * POSIX platform does not use CodePage. + */ + +static unsigned +get_current_codepage(void) +{ + return (-1);/* Unknown */ +} +static unsigned +make_codepage_from_charset(const char *charset) +{ + (void)charset; /* UNUSED */ + return (-1);/* Unknown */ +} +static unsigned +get_current_oemcp(void) +{ + return (-1);/* Unknown */ +} + +#endif /* defined(_WIN32) && !defined(__CYGWIN__) */ + +/* + * Return a string conversion object. + */ +static struct archive_string_conv * +get_sconv_object(struct archive *a, const char *fc, const char *tc, int flag) +{ + struct archive_string_conv *sc; + unsigned current_codepage; + + /* Check if we have made the sconv object. */ + sc = find_sconv_object(a, fc, tc); + if (sc != NULL) + return (sc); + + if (a == NULL) + current_codepage = get_current_codepage(); + else + current_codepage = a->current_codepage; + + sc = create_sconv_object(canonical_charset_name(fc), + canonical_charset_name(tc), current_codepage, flag); + if (sc == NULL) { + if (a != NULL) + archive_set_error(a, ENOMEM, + "Could not allocate memory for " + "a string conversion object"); + return (NULL); + } + + /* + * If there is no converter for current string conversion object, + * we cannot handle this conversion. + */ + if (sc->nconverter == 0) { + if (a != NULL) { +#if HAVE_ICONV + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "iconv_open failed : Cannot handle ``%s''", + (flag & SCONV_TO_CHARSET)?tc:fc); +#else + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "A character-set conversion not fully supported " + "on this platform"); +#endif + } + /* Failed; free a sconv object. */ + free_sconv_object(sc); + return (NULL); + } + + /* + * Success! + */ + if (a != NULL) + add_sconv_object(a, sc); + return (sc); +} + +static const char * +get_current_charset(struct archive *a) +{ + const char *cur_charset; + + if (a == NULL) + cur_charset = default_iconv_charset(""); + else { + cur_charset = default_iconv_charset(a->current_code); + if (a->current_code == NULL) { + a->current_code = strdup(cur_charset); + a->current_codepage = get_current_codepage(); + a->current_oemcp = get_current_oemcp(); + } + } + return (cur_charset); +} + +/* + * Make and Return a string conversion object. + * Return NULL if the platform does not support the specified conversion + * and best_effort is 0. + * If best_effort is set, A string conversion object must be returned + * unless memory allocation for the object fails, but the conversion + * might fail when non-ASCII code is found. + */ +struct archive_string_conv * +archive_string_conversion_to_charset(struct archive *a, const char *charset, + int best_effort) +{ + int flag = SCONV_TO_CHARSET; + + if (best_effort) + flag |= SCONV_BEST_EFFORT; + return (get_sconv_object(a, get_current_charset(a), charset, flag)); +} + +struct archive_string_conv * +archive_string_conversion_from_charset(struct archive *a, const char *charset, + int best_effort) +{ + int flag = SCONV_FROM_CHARSET; + + if (best_effort) + flag |= SCONV_BEST_EFFORT; + return (get_sconv_object(a, charset, get_current_charset(a), flag)); +} + +/* + * archive_string_default_conversion_*_archive() are provided for Windows + * platform because other archiver application use CP_OEMCP for + * MultiByteToWideChar() and WideCharToMultiByte() for the filenames + * in tar or zip files. But mbstowcs/wcstombs(CRT) usually use CP_ACP + * unless you use setlocale(LC_ALL, ".OCP")(specify CP_OEMCP). + * So we should make a string conversion between CP_ACP and CP_OEMCP + * for compatibility. + */ +#if defined(_WIN32) && !defined(__CYGWIN__) +struct archive_string_conv * +archive_string_default_conversion_for_read(struct archive *a) +{ + const char *cur_charset = get_current_charset(a); + char oemcp[16]; + + /* NOTE: a check of cur_charset is unneeded but we need + * that get_current_charset() has been surely called at + * this time whatever C compiler optimized. */ + if (cur_charset != NULL && + (a->current_codepage == CP_C_LOCALE || + a->current_codepage == a->current_oemcp)) + return (NULL);/* no conversion. */ + + _snprintf(oemcp, sizeof(oemcp)-1, "CP%d", a->current_oemcp); + /* Make sure a null termination must be set. */ + oemcp[sizeof(oemcp)-1] = '\0'; + return (get_sconv_object(a, oemcp, cur_charset, + SCONV_FROM_CHARSET)); +} + +struct archive_string_conv * +archive_string_default_conversion_for_write(struct archive *a) +{ + const char *cur_charset = get_current_charset(a); + char oemcp[16]; + + /* NOTE: a check of cur_charset is unneeded but we need + * that get_current_charset() has been surely called at + * this time whatever C compiler optimized. */ + if (cur_charset != NULL && + (a->current_codepage == CP_C_LOCALE || + a->current_codepage == a->current_oemcp)) + return (NULL);/* no conversion. */ + + _snprintf(oemcp, sizeof(oemcp)-1, "CP%d", a->current_oemcp); + /* Make sure a null termination must be set. */ + oemcp[sizeof(oemcp)-1] = '\0'; + return (get_sconv_object(a, cur_charset, oemcp, + SCONV_TO_CHARSET)); +} +#else +struct archive_string_conv * +archive_string_default_conversion_for_read(struct archive *a) +{ + (void)a; /* UNUSED */ + return (NULL); +} + +struct archive_string_conv * +archive_string_default_conversion_for_write(struct archive *a) +{ + (void)a; /* UNUSED */ + return (NULL); +} +#endif + +/* + * Dispose of all character conversion objects in the archive object. + */ +void +archive_string_conversion_free(struct archive *a) +{ + struct archive_string_conv *sc; + struct archive_string_conv *sc_next; + + for (sc = a->sconv; sc != NULL; sc = sc_next) { + sc_next = sc->next; + free_sconv_object(sc); + } + a->sconv = NULL; + free(a->current_code); + a->current_code = NULL; +} + +/* + * Return a conversion charset name. + */ +const char * +archive_string_conversion_charset_name(struct archive_string_conv *sc) +{ + if (sc->flag & SCONV_TO_CHARSET) + return (sc->to_charset); + else + return (sc->from_charset); +} + +/* + * Change the behavior of a string conversion. + */ +void +archive_string_conversion_set_opt(struct archive_string_conv *sc, int opt) +{ + switch (opt) { + /* + * A filename in UTF-8 was made with libarchive 2.x in a wrong + * assumption that wchar_t was Unicode. + * This option enables simulating the assumption in order to read + * that filename correctly. + */ + case SCONV_SET_OPT_UTF8_LIBARCHIVE2X: +#if (defined(_WIN32) && !defined(__CYGWIN__)) \ + || defined(__STDC_ISO_10646__) || defined(__APPLE__) + /* + * Nothing to do for it since wchar_t on these platforms + * is really Unicode. + */ + (void)sc; /* UNUSED */ +#else + if ((sc->flag & SCONV_UTF8_LIBARCHIVE_2) == 0) { + sc->flag |= SCONV_UTF8_LIBARCHIVE_2; + /* Set up string converters. */ + setup_converter(sc); + } +#endif + break; + case SCONV_SET_OPT_NORMALIZATION_C: + if ((sc->flag & SCONV_NORMALIZATION_C) == 0) { + sc->flag |= SCONV_NORMALIZATION_C; + sc->flag &= ~SCONV_NORMALIZATION_D; + /* Set up string converters. */ + setup_converter(sc); + } + break; + case SCONV_SET_OPT_NORMALIZATION_D: +#if defined(HAVE_ICONV) + /* + * If iconv will take the string, do not change the + * setting of the normalization. + */ + if (!(sc->flag & SCONV_WIN_CP) && + (sc->flag & (SCONV_FROM_UTF16 | SCONV_FROM_UTF8)) && + !(sc->flag & (SCONV_TO_UTF16 | SCONV_TO_UTF8))) + break; +#endif + if ((sc->flag & SCONV_NORMALIZATION_D) == 0) { + sc->flag |= SCONV_NORMALIZATION_D; + sc->flag &= ~SCONV_NORMALIZATION_C; + /* Set up string converters. */ + setup_converter(sc); + } + break; + default: + break; + } +} + +/* + * + * Copy one archive_string to another in locale conversion. + * + * archive_strncat_l(); + * archive_strncpy_l(); + * + */ + +static size_t +mbsnbytes(const void *_p, size_t n) +{ + size_t s; + const char *p, *pp; + + if (_p == NULL) + return (0); + p = (const char *)_p; + + /* Like strlen(p), except won't examine positions beyond p[n]. */ + s = 0; + pp = p; + while (s < n && *pp) { + pp++; + s++; + } + return (s); +} + +static size_t +utf16nbytes(const void *_p, size_t n) +{ + size_t s; + const char *p, *pp; + + if (_p == NULL) + return (0); + p = (const char *)_p; + + /* Like strlen(p), except won't examine positions beyond p[n]. */ + s = 0; + pp = p; + n >>= 1; + while (s < n && (pp[0] || pp[1])) { + pp += 2; + s++; + } + return (s<<1); +} + +int +archive_strncpy_l(struct archive_string *as, const void *_p, size_t n, + struct archive_string_conv *sc) +{ + as->length = 0; + return (archive_strncat_l(as, _p, n, sc)); +} + +int +archive_strncat_l(struct archive_string *as, const void *_p, size_t n, + struct archive_string_conv *sc) +{ + const void *s; + size_t length = 0; + int i, r = 0, r2; + + if (_p != NULL && n > 0) { + if (sc != NULL && (sc->flag & SCONV_FROM_UTF16)) + length = utf16nbytes(_p, n); + else + length = mbsnbytes(_p, n); + } + + /* We must allocate memory even if there is no data for conversion + * or copy. This simulates archive_string_append behavior. */ + if (length == 0) { + int tn = 1; + if (sc != NULL && (sc->flag & SCONV_TO_UTF16)) + tn = 2; + if (archive_string_ensure(as, as->length + tn) == NULL) + return (-1); + as->s[as->length] = 0; + if (tn == 2) + as->s[as->length+1] = 0; + return (0); + } + + /* + * If sc is NULL, we just make a copy. + */ + if (sc == NULL) { + if (archive_string_append(as, _p, length) == NULL) + return (-1);/* No memory */ + return (0); + } + + s = _p; + i = 0; + if (sc->nconverter > 1) { + sc->utftmp.length = 0; + r2 = sc->converter[0](&(sc->utftmp), s, length, sc); + if (r2 != 0 && errno == ENOMEM) + return (r2); + if (r > r2) + r = r2; + s = sc->utftmp.s; + length = sc->utftmp.length; + ++i; + } + r2 = sc->converter[i](as, s, length, sc); + if (r > r2) + r = r2; + return (r); +} + +#if HAVE_ICONV + +/* + * Return -1 if conversion fails. + */ +static int +iconv_strncat_in_locale(struct archive_string *as, const void *_p, + size_t length, struct archive_string_conv *sc) +{ + ICONV_CONST char *itp; + size_t remaining; + iconv_t cd; + char *outp; + size_t avail, bs; + int return_value = 0; /* success */ + int to_size, from_size; + + if (sc->flag & SCONV_TO_UTF16) + to_size = 2; + else + to_size = 1; + if (sc->flag & SCONV_FROM_UTF16) + from_size = 2; + else + from_size = 1; + + if (archive_string_ensure(as, as->length + length*2+to_size) == NULL) + return (-1); + + cd = sc->cd; + itp = (char *)(uintptr_t)_p; + remaining = length; + outp = as->s + as->length; + avail = as->buffer_length - as->length - to_size; + while (remaining >= (size_t)from_size) { + size_t result = iconv(cd, &itp, &remaining, &outp, &avail); + + if (result != (size_t)-1) + break; /* Conversion completed. */ + + if (errno == EILSEQ || errno == EINVAL) { + /* + * If an output charset is UTF-8 or UTF-16BE/LE, + * unknown character should be U+FFFD + * (replacement character). + */ + if (sc->flag & (SCONV_TO_UTF8 | SCONV_TO_UTF16)) { + size_t rbytes; + if (sc->flag & SCONV_TO_UTF8) + rbytes = sizeof(utf8_replacement_char); + else + rbytes = 2; + + if (avail < rbytes) { + as->length = outp - as->s; + bs = as->buffer_length + + (remaining * to_size) + rbytes; + if (NULL == + archive_string_ensure(as, bs)) + return (-1); + outp = as->s + as->length; + avail = as->buffer_length + - as->length - to_size; + } + if (sc->flag & SCONV_TO_UTF8) + memcpy(outp, utf8_replacement_char, sizeof(utf8_replacement_char)); + else if (sc->flag & SCONV_TO_UTF16BE) + archive_be16enc(outp, UNICODE_R_CHAR); + else + archive_le16enc(outp, UNICODE_R_CHAR); + outp += rbytes; + avail -= rbytes; + } else { + /* Skip the illegal input bytes. */ + *outp++ = '?'; + avail--; + } + itp += from_size; + remaining -= from_size; + return_value = -1; /* failure */ + } else { + /* E2BIG no output buffer, + * Increase an output buffer. */ + as->length = outp - as->s; + bs = as->buffer_length + remaining * 2; + if (NULL == archive_string_ensure(as, bs)) + return (-1); + outp = as->s + as->length; + avail = as->buffer_length - as->length - to_size; + } + } + as->length = outp - as->s; + as->s[as->length] = 0; + if (to_size == 2) + as->s[as->length+1] = 0; + return (return_value); +} + +#endif /* HAVE_ICONV */ + + +#if defined(_WIN32) && !defined(__CYGWIN__) + +/* + * Translate a string from a some CodePage to an another CodePage by + * Windows APIs, and copy the result. Return -1 if conversion fails. + */ +static int +strncat_in_codepage(struct archive_string *as, + const void *_p, size_t length, struct archive_string_conv *sc) +{ + const char *s = (const char *)_p; + struct archive_wstring aws; + size_t l; + int r, saved_flag; + + archive_string_init(&aws); + saved_flag = sc->flag; + sc->flag &= ~(SCONV_NORMALIZATION_D | SCONV_NORMALIZATION_C); + r = archive_wstring_append_from_mbs_in_codepage(&aws, s, length, sc); + sc->flag = saved_flag; + if (r != 0) { + archive_wstring_free(&aws); + if (errno != ENOMEM) + archive_string_append(as, s, length); + return (-1); + } + + l = as->length; + r = archive_string_append_from_wcs_in_codepage( + as, aws.s, aws.length, sc); + if (r != 0 && errno != ENOMEM && l == as->length) + archive_string_append(as, s, length); + archive_wstring_free(&aws); + return (r); +} + +/* + * Test whether MBS ==> WCS is okay. + */ +static int +invalid_mbs(const void *_p, size_t n, struct archive_string_conv *sc) +{ + const char *p = (const char *)_p; + unsigned codepage; + DWORD mbflag = MB_ERR_INVALID_CHARS; + + if (sc->flag & SCONV_FROM_CHARSET) + codepage = sc->to_cp; + else + codepage = sc->from_cp; + + if (codepage == CP_C_LOCALE) + return (0); + if (codepage != CP_UTF8) + mbflag |= MB_PRECOMPOSED; + + if (MultiByteToWideChar(codepage, mbflag, p, (int)n, NULL, 0) == 0) + return (-1); /* Invalid */ + return (0); /* Okay */ +} + +#else + +/* + * Test whether MBS ==> WCS is okay. + */ +static int +invalid_mbs(const void *_p, size_t n, struct archive_string_conv *sc) +{ + const char *p = (const char *)_p; + size_t r; + +#if HAVE_MBRTOWC + mbstate_t shift_state; + + memset(&shift_state, 0, sizeof(shift_state)); +#else + /* Clear the shift state before starting. */ + mbtowc(NULL, NULL, 0); +#endif + while (n) { + wchar_t wc; + +#if HAVE_MBRTOWC + r = mbrtowc(&wc, p, n, &shift_state); +#else + r = mbtowc(&wc, p, n); +#endif + if (r == (size_t)-1 || r == (size_t)-2) + return (-1);/* Invalid. */ + if (r == 0) + break; + p += r; + n -= r; + } + (void)sc; /* UNUSED */ + return (0); /* All Okey. */ +} + +#endif /* defined(_WIN32) && !defined(__CYGWIN__) */ + +/* + * Basically returns -1 because we cannot make a conversion of charset + * without iconv but in some cases this would return 0. + * Returns 0 if all copied characters are ASCII. + * Returns 0 if both from-locale and to-locale are the same and those + * can be WCS with no error. + */ +static int +best_effort_strncat_in_locale(struct archive_string *as, const void *_p, + size_t length, struct archive_string_conv *sc) +{ + size_t remaining; + const uint8_t *itp; + int return_value = 0; /* success */ + + /* + * If both from-locale and to-locale is the same, this makes a copy. + * And then this checks all copied MBS can be WCS if so returns 0. + */ + if (sc->same) { + if (archive_string_append(as, _p, length) == NULL) + return (-1);/* No memory */ + return (invalid_mbs(_p, length, sc)); + } + + /* + * If a character is ASCII, this just copies it. If not, this + * assigns '?' character instead but in UTF-8 locale this assigns + * byte sequence 0xEF 0xBD 0xBD, which are code point U+FFFD, + * a Replacement Character in Unicode. + */ + + remaining = length; + itp = (const uint8_t *)_p; + while (*itp && remaining > 0) { + if (*itp > 127) { + // Non-ASCII: Substitute with suitable replacement + if (sc->flag & SCONV_TO_UTF8) { + if (archive_string_append(as, utf8_replacement_char, sizeof(utf8_replacement_char)) == NULL) { + __archive_errx(1, "Out of memory"); + } + } else { + archive_strappend_char(as, '?'); + } + return_value = -1; + } else { + archive_strappend_char(as, *itp); + } + ++itp; + } + return (return_value); +} + + +/* + * Unicode conversion functions. + * - UTF-8 <===> UTF-8 in removing surrogate pairs. + * - UTF-8 NFD ===> UTF-8 NFC in removing surrogate pairs. + * - UTF-8 made by libarchive 2.x ===> UTF-8. + * - UTF-16BE <===> UTF-8. + * + */ + +/* + * Utility to convert a single UTF-8 sequence. + * + * Usually return used bytes, return used byte in negative value when + * a unicode character is replaced with U+FFFD. + * See also http://unicode.org/review/pr-121.html Public Review Issue #121 + * Recommended Practice for Replacement Characters. + */ +static int +_utf8_to_unicode(uint32_t *pwc, const char *s, size_t n) +{ + static const char utf8_count[256] = { + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 00 - 0F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 10 - 1F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 20 - 2F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 30 - 3F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 40 - 4F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 50 - 5F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 60 - 6F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 70 - 7F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 80 - 8F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 90 - 9F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* A0 - AF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* B0 - BF */ + 0, 0, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,/* C0 - CF */ + 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,/* D0 - DF */ + 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,/* E0 - EF */ + 4, 4, 4, 4, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 /* F0 - FF */ + }; + int ch, i; + int cnt; + uint32_t wc; + + /* Sanity check. */ + if (n == 0) + return (0); + /* + * Decode 1-4 bytes depending on the value of the first byte. + */ + ch = (unsigned char)*s; + if (ch == 0) + return (0); /* Standard: return 0 for end-of-string. */ + cnt = utf8_count[ch]; + + /* Invalid sequence or there are not plenty bytes. */ + if ((int)n < cnt) { + cnt = (int)n; + for (i = 1; i < cnt; i++) { + if ((s[i] & 0xc0) != 0x80) { + cnt = i; + break; + } + } + goto invalid_sequence; + } + + /* Make a Unicode code point from a single UTF-8 sequence. */ + switch (cnt) { + case 1: /* 1 byte sequence. */ + *pwc = ch & 0x7f; + return (cnt); + case 2: /* 2 bytes sequence. */ + if ((s[1] & 0xc0) != 0x80) { + cnt = 1; + goto invalid_sequence; + } + *pwc = ((ch & 0x1f) << 6) | (s[1] & 0x3f); + return (cnt); + case 3: /* 3 bytes sequence. */ + if ((s[1] & 0xc0) != 0x80) { + cnt = 1; + goto invalid_sequence; + } + if ((s[2] & 0xc0) != 0x80) { + cnt = 2; + goto invalid_sequence; + } + wc = ((ch & 0x0f) << 12) + | ((s[1] & 0x3f) << 6) + | (s[2] & 0x3f); + if (wc < 0x800) + goto invalid_sequence;/* Overlong sequence. */ + break; + case 4: /* 4 bytes sequence. */ + if ((s[1] & 0xc0) != 0x80) { + cnt = 1; + goto invalid_sequence; + } + if ((s[2] & 0xc0) != 0x80) { + cnt = 2; + goto invalid_sequence; + } + if ((s[3] & 0xc0) != 0x80) { + cnt = 3; + goto invalid_sequence; + } + wc = ((ch & 0x07) << 18) + | ((s[1] & 0x3f) << 12) + | ((s[2] & 0x3f) << 6) + | (s[3] & 0x3f); + if (wc < 0x10000) + goto invalid_sequence;/* Overlong sequence. */ + break; + default: /* Others are all invalid sequence. */ + if (ch == 0xc0 || ch == 0xc1) + cnt = 2; + else if (ch >= 0xf5 && ch <= 0xf7) + cnt = 4; + else if (ch >= 0xf8 && ch <= 0xfb) + cnt = 5; + else if (ch == 0xfc || ch == 0xfd) + cnt = 6; + else + cnt = 1; + if ((int)n < cnt) + cnt = (int)n; + for (i = 1; i < cnt; i++) { + if ((s[i] & 0xc0) != 0x80) { + cnt = i; + break; + } + } + goto invalid_sequence; + } + + /* The code point larger than 0x10FFFF is not legal + * Unicode values. */ + if (wc > UNICODE_MAX) + goto invalid_sequence; + /* Correctly gets a Unicode, returns used bytes. */ + *pwc = wc; + return (cnt); +invalid_sequence: + *pwc = UNICODE_R_CHAR;/* set the Replacement Character instead. */ + return (cnt * -1); +} + +static int +utf8_to_unicode(uint32_t *pwc, const char *s, size_t n) +{ + int cnt; + + cnt = _utf8_to_unicode(pwc, s, n); + /* Any of Surrogate pair is not legal Unicode values. */ + if (cnt == 3 && IS_SURROGATE_PAIR_LA(*pwc)) + return (-3); + return (cnt); +} + +static inline uint32_t +combine_surrogate_pair(uint32_t uc, uint32_t uc2) +{ + uc -= 0xD800; + uc *= 0x400; + uc += uc2 - 0xDC00; + uc += 0x10000; + return (uc); +} + +/* + * Convert a single UTF-8/CESU-8 sequence to a Unicode code point in + * removing surrogate pairs. + * + * CESU-8: The Compatibility Encoding Scheme for UTF-16. + * + * Usually return used bytes, return used byte in negative value when + * a unicode character is replaced with U+FFFD. + */ +static int +cesu8_to_unicode(uint32_t *pwc, const char *s, size_t n) +{ + uint32_t wc = 0; + int cnt; + + cnt = _utf8_to_unicode(&wc, s, n); + if (cnt == 3 && IS_HIGH_SURROGATE_LA(wc)) { + uint32_t wc2 = 0; + if (n - 3 < 3) { + /* Invalid byte sequence. */ + goto invalid_sequence; + } + cnt = _utf8_to_unicode(&wc2, s+3, n-3); + if (cnt != 3 || !IS_LOW_SURROGATE_LA(wc2)) { + /* Invalid byte sequence. */ + goto invalid_sequence; + } + wc = combine_surrogate_pair(wc, wc2); + cnt = 6; + } else if (cnt == 3 && IS_LOW_SURROGATE_LA(wc)) { + /* Invalid byte sequence. */ + goto invalid_sequence; + } + *pwc = wc; + return (cnt); +invalid_sequence: + *pwc = UNICODE_R_CHAR;/* set the Replacement Character instead. */ + if (cnt > 0) + cnt *= -1; + return (cnt); +} + +/* + * Convert a Unicode code point to a single UTF-8 sequence. + * + * NOTE:This function does not check if the Unicode is legal or not. + * Please you definitely check it before calling this. + */ +static size_t +unicode_to_utf8(char *p, size_t remaining, uint32_t uc) +{ + char *_p = p; + + /* Invalid Unicode char maps to Replacement character */ + if (uc > UNICODE_MAX) + uc = UNICODE_R_CHAR; + /* Translate code point to UTF8 */ + if (uc <= 0x7f) { + if (remaining == 0) + return (0); + *p++ = (char)uc; + } else if (uc <= 0x7ff) { + if (remaining < 2) + return (0); + *p++ = 0xc0 | ((uc >> 6) & 0x1f); + *p++ = 0x80 | (uc & 0x3f); + } else if (uc <= 0xffff) { + if (remaining < 3) + return (0); + *p++ = 0xe0 | ((uc >> 12) & 0x0f); + *p++ = 0x80 | ((uc >> 6) & 0x3f); + *p++ = 0x80 | (uc & 0x3f); + } else { + if (remaining < 4) + return (0); + *p++ = 0xf0 | ((uc >> 18) & 0x07); + *p++ = 0x80 | ((uc >> 12) & 0x3f); + *p++ = 0x80 | ((uc >> 6) & 0x3f); + *p++ = 0x80 | (uc & 0x3f); + } + return (p - _p); +} + +static int +utf16be_to_unicode(uint32_t *pwc, const char *s, size_t n) +{ + return (utf16_to_unicode(pwc, s, n, 1)); +} + +static int +utf16le_to_unicode(uint32_t *pwc, const char *s, size_t n) +{ + return (utf16_to_unicode(pwc, s, n, 0)); +} + +static int +utf16_to_unicode(uint32_t *pwc, const char *s, size_t n, int be) +{ + const char *utf16 = s; + unsigned uc; + + if (n == 0) + return (0); + if (n == 1) { + /* set the Replacement Character instead. */ + *pwc = UNICODE_R_CHAR; + return (-1); + } + + if (be) + uc = archive_be16dec(utf16); + else + uc = archive_le16dec(utf16); + utf16 += 2; + + /* If this is a surrogate pair, assemble the full code point.*/ + if (IS_HIGH_SURROGATE_LA(uc)) { + unsigned uc2; + + if (n >= 4) { + if (be) + uc2 = archive_be16dec(utf16); + else + uc2 = archive_le16dec(utf16); + } else + uc2 = 0; + if (IS_LOW_SURROGATE_LA(uc2)) { + uc = combine_surrogate_pair(uc, uc2); + utf16 += 2; + } else { + /* Undescribed code point should be U+FFFD + * (replacement character). */ + *pwc = UNICODE_R_CHAR; + return (-2); + } + } + + /* + * Surrogate pair values(0xd800 through 0xdfff) are only + * used by UTF-16, so, after above calculation, the code + * must not be surrogate values, and Unicode has no codes + * larger than 0x10ffff. Thus, those are not legal Unicode + * values. + */ + if (IS_SURROGATE_PAIR_LA(uc) || uc > UNICODE_MAX) { + /* Undescribed code point should be U+FFFD + * (replacement character). */ + *pwc = UNICODE_R_CHAR; + return (((int)(utf16 - s)) * -1); + } + *pwc = uc; + return ((int)(utf16 - s)); +} + +static size_t +unicode_to_utf16be(char *p, size_t remaining, uint32_t uc) +{ + char *utf16 = p; + + if (uc > 0xffff) { + /* We have a code point that won't fit into a + * wchar_t; convert it to a surrogate pair. */ + if (remaining < 4) + return (0); + uc -= 0x10000; + archive_be16enc(utf16, ((uc >> 10) & 0x3ff) + 0xD800); + archive_be16enc(utf16+2, (uc & 0x3ff) + 0xDC00); + return (4); + } else { + if (remaining < 2) + return (0); + archive_be16enc(utf16, uc); + return (2); + } +} + +static size_t +unicode_to_utf16le(char *p, size_t remaining, uint32_t uc) +{ + char *utf16 = p; + + if (uc > 0xffff) { + /* We have a code point that won't fit into a + * wchar_t; convert it to a surrogate pair. */ + if (remaining < 4) + return (0); + uc -= 0x10000; + archive_le16enc(utf16, ((uc >> 10) & 0x3ff) + 0xD800); + archive_le16enc(utf16+2, (uc & 0x3ff) + 0xDC00); + return (4); + } else { + if (remaining < 2) + return (0); + archive_le16enc(utf16, uc); + return (2); + } +} + +/* + * Copy UTF-8 string in checking surrogate pair. + * If any surrogate pair are found, it would be canonicalized. + */ +static int +strncat_from_utf8_to_utf8(struct archive_string *as, const void *_p, + size_t len, struct archive_string_conv *sc) +{ + const char *s; + char *p, *endp; + int n, ret = 0; + + (void)sc; /* UNUSED */ + + if (archive_string_ensure(as, as->length + len + 1) == NULL) + return (-1); + + s = (const char *)_p; + p = as->s + as->length; + endp = as->s + as->buffer_length -1; + do { + uint32_t uc; + const char *ss = s; + size_t w; + + /* + * Forward byte sequence until a conversion of that is needed. + */ + while ((n = utf8_to_unicode(&uc, s, len)) > 0) { + s += n; + len -= n; + } + if (ss < s) { + if (p + (s - ss) > endp) { + as->length = p - as->s; + if (archive_string_ensure(as, + as->buffer_length + len + 1) == NULL) + return (-1); + p = as->s + as->length; + endp = as->s + as->buffer_length -1; + } + + memcpy(p, ss, s - ss); + p += s - ss; + } + + /* + * If n is negative, current byte sequence needs a replacement. + */ + if (n < 0) { + if (n == -3 && IS_SURROGATE_PAIR_LA(uc)) { + /* Current byte sequence may be CESU-8. */ + n = cesu8_to_unicode(&uc, s, len); + } + if (n < 0) { + ret = -1; + n *= -1;/* Use a replaced unicode character. */ + } + + /* Rebuild UTF-8 byte sequence. */ + while ((w = unicode_to_utf8(p, endp - p, uc)) == 0) { + as->length = p - as->s; + if (archive_string_ensure(as, + as->buffer_length + len + 1) == NULL) + return (-1); + p = as->s + as->length; + endp = as->s + as->buffer_length -1; + } + p += w; + s += n; + len -= n; + } + } while (n > 0); + as->length = p - as->s; + as->s[as->length] = '\0'; + return (ret); +} + +static int +archive_string_append_unicode(struct archive_string *as, const void *_p, + size_t len, struct archive_string_conv *sc) +{ + const char *s; + char *p, *endp; + uint32_t uc; + size_t w; + int n, ret = 0, ts, tm; + int (*parse)(uint32_t *, const char *, size_t); + size_t (*unparse)(char *, size_t, uint32_t); + + if (sc->flag & SCONV_TO_UTF16BE) { + unparse = unicode_to_utf16be; + ts = 2; + } else if (sc->flag & SCONV_TO_UTF16LE) { + unparse = unicode_to_utf16le; + ts = 2; + } else if (sc->flag & SCONV_TO_UTF8) { + unparse = unicode_to_utf8; + ts = 1; + } else { + /* + * This case is going to be converted to another + * character-set through iconv. + */ + if (sc->flag & SCONV_FROM_UTF16BE) { + unparse = unicode_to_utf16be; + ts = 2; + } else if (sc->flag & SCONV_FROM_UTF16LE) { + unparse = unicode_to_utf16le; + ts = 2; + } else { + unparse = unicode_to_utf8; + ts = 1; + } + } + + if (sc->flag & SCONV_FROM_UTF16BE) { + parse = utf16be_to_unicode; + tm = 1; + } else if (sc->flag & SCONV_FROM_UTF16LE) { + parse = utf16le_to_unicode; + tm = 1; + } else { + parse = cesu8_to_unicode; + tm = ts; + } + + if (archive_string_ensure(as, as->length + len * tm + ts) == NULL) + return (-1); + + s = (const char *)_p; + p = as->s + as->length; + endp = as->s + as->buffer_length - ts; + while ((n = parse(&uc, s, len)) != 0) { + if (n < 0) { + /* Use a replaced unicode character. */ + n *= -1; + ret = -1; + } + s += n; + len -= n; + while ((w = unparse(p, endp - p, uc)) == 0) { + /* There is not enough output buffer so + * we have to expand it. */ + as->length = p - as->s; + if (archive_string_ensure(as, + as->buffer_length + len * tm + ts) == NULL) + return (-1); + p = as->s + as->length; + endp = as->s + as->buffer_length - ts; + } + p += w; + } + as->length = p - as->s; + as->s[as->length] = '\0'; + if (ts == 2) + as->s[as->length+1] = '\0'; + return (ret); +} + +/* + * Following Constants for Hangul compositions this information comes from + * Unicode Standard Annex #15 http://unicode.org/reports/tr15/ + */ +#define HC_SBASE 0xAC00 +#define HC_LBASE 0x1100 +#define HC_VBASE 0x1161 +#define HC_TBASE 0x11A7 +#define HC_LCOUNT 19 +#define HC_VCOUNT 21 +#define HC_TCOUNT 28 +#define HC_NCOUNT (HC_VCOUNT * HC_TCOUNT) +#define HC_SCOUNT (HC_LCOUNT * HC_NCOUNT) + +static uint32_t +get_nfc(uint32_t uc, uint32_t uc2) +{ + int t, b; + + t = 0; + b = sizeof(u_composition_table)/sizeof(u_composition_table[0]) -1; + while (b >= t) { + int m = (t + b) / 2; + if (u_composition_table[m].cp1 < uc) + t = m + 1; + else if (u_composition_table[m].cp1 > uc) + b = m - 1; + else if (u_composition_table[m].cp2 < uc2) + t = m + 1; + else if (u_composition_table[m].cp2 > uc2) + b = m - 1; + else + return (u_composition_table[m].nfc); + } + return (0); +} + +#define FDC_MAX 10 /* The maximum number of Following Decomposable + * Characters. */ + +/* + * Update first code point. + */ +#define UPDATE_UC(new_uc) do { \ + uc = new_uc; \ + ucptr = NULL; \ +} while (0) + +/* + * Replace first code point with second code point. + */ +#define REPLACE_UC_WITH_UC2() do { \ + uc = uc2; \ + ucptr = uc2ptr; \ + n = n2; \ +} while (0) + +#define EXPAND_BUFFER() do { \ + as->length = p - as->s; \ + if (archive_string_ensure(as, \ + as->buffer_length + len * tm + ts) == NULL)\ + return (-1); \ + p = as->s + as->length; \ + endp = as->s + as->buffer_length - ts; \ +} while (0) + +#define UNPARSE(p, endp, uc) do { \ + while ((w = unparse(p, (endp) - (p), uc)) == 0) {\ + EXPAND_BUFFER(); \ + } \ + p += w; \ +} while (0) + +/* + * Write first code point. + * If the code point has not be changed from its original code, + * this just copies it from its original buffer pointer. + * If not, this converts it to UTF-8 byte sequence and copies it. + */ +#define WRITE_UC() do { \ + if (ucptr) { \ + if (p + n > endp) \ + EXPAND_BUFFER(); \ + switch (n) { \ + case 4: \ + *p++ = *ucptr++; \ + /* FALL THROUGH */ \ + case 3: \ + *p++ = *ucptr++; \ + /* FALL THROUGH */ \ + case 2: \ + *p++ = *ucptr++; \ + /* FALL THROUGH */ \ + case 1: \ + *p++ = *ucptr; \ + break; \ + } \ + ucptr = NULL; \ + } else { \ + UNPARSE(p, endp, uc); \ + } \ +} while (0) + +/* + * Collect following decomposable code points. + */ +#define COLLECT_CPS(start) do { \ + int _i; \ + for (_i = start; _i < FDC_MAX ; _i++) { \ + nx = parse(&ucx[_i], s, len); \ + if (nx <= 0) \ + break; \ + cx = CCC(ucx[_i]); \ + if (cl >= cx && cl != 228 && cx != 228)\ + break; \ + s += nx; \ + len -= nx; \ + cl = cx; \ + ccx[_i] = cx; \ + } \ + if (_i >= FDC_MAX) { \ + ret = -1; \ + ucx_size = FDC_MAX; \ + } else \ + ucx_size = _i; \ +} while (0) + +/* + * Normalize UTF-8/UTF-16BE characters to Form C and copy the result. + * + * TODO: Convert composition exclusions, which are never converted + * from NFC,NFD,NFKC and NFKD, to Form C. + */ +static int +archive_string_normalize_C(struct archive_string *as, const void *_p, + size_t len, struct archive_string_conv *sc) +{ + const char *s = (const char *)_p; + char *p, *endp; + uint32_t uc, uc2; + size_t w; + int always_replace, n, n2, ret = 0, spair, ts, tm; + int (*parse)(uint32_t *, const char *, size_t); + size_t (*unparse)(char *, size_t, uint32_t); + + always_replace = 1; + ts = 1;/* text size. */ + if (sc->flag & SCONV_TO_UTF16BE) { + unparse = unicode_to_utf16be; + ts = 2; + if (sc->flag & SCONV_FROM_UTF16BE) + always_replace = 0; + } else if (sc->flag & SCONV_TO_UTF16LE) { + unparse = unicode_to_utf16le; + ts = 2; + if (sc->flag & SCONV_FROM_UTF16LE) + always_replace = 0; + } else if (sc->flag & SCONV_TO_UTF8) { + unparse = unicode_to_utf8; + if (sc->flag & SCONV_FROM_UTF8) + always_replace = 0; + } else { + /* + * This case is going to be converted to another + * character-set through iconv. + */ + always_replace = 0; + if (sc->flag & SCONV_FROM_UTF16BE) { + unparse = unicode_to_utf16be; + ts = 2; + } else if (sc->flag & SCONV_FROM_UTF16LE) { + unparse = unicode_to_utf16le; + ts = 2; + } else { + unparse = unicode_to_utf8; + } + } + + if (sc->flag & SCONV_FROM_UTF16BE) { + parse = utf16be_to_unicode; + tm = 1; + spair = 4;/* surrogate pair size in UTF-16. */ + } else if (sc->flag & SCONV_FROM_UTF16LE) { + parse = utf16le_to_unicode; + tm = 1; + spair = 4;/* surrogate pair size in UTF-16. */ + } else { + parse = cesu8_to_unicode; + tm = ts; + spair = 6;/* surrogate pair size in UTF-8. */ + } + + if (archive_string_ensure(as, as->length + len * tm + ts) == NULL) + return (-1); + + p = as->s + as->length; + endp = as->s + as->buffer_length - ts; + while ((n = parse(&uc, s, len)) != 0) { + const char *ucptr, *uc2ptr; + + if (n < 0) { + /* Use a replaced unicode character. */ + UNPARSE(p, endp, uc); + s += n*-1; + len -= n*-1; + ret = -1; + continue; + } else if (n == spair || always_replace) + /* uc is converted from a surrogate pair. + * this should be treated as a changed code. */ + ucptr = NULL; + else + ucptr = s; + s += n; + len -= n; + + /* Read second code point. */ + while ((n2 = parse(&uc2, s, len)) > 0) { + uint32_t ucx[FDC_MAX]; + int ccx[FDC_MAX]; + int cl, cx, i, nx, ucx_size; + int LIndex,SIndex; + uint32_t nfc; + + if (n2 == spair || always_replace) + /* uc2 is converted from a surrogate pair. + * this should be treated as a changed code. */ + uc2ptr = NULL; + else + uc2ptr = s; + s += n2; + len -= n2; + + /* + * If current second code point is out of decomposable + * code points, finding compositions is unneeded. + */ + if (!IS_DECOMPOSABLE_BLOCK(uc2)) { + WRITE_UC(); + REPLACE_UC_WITH_UC2(); + continue; + } + + /* + * Try to combine current code points. + */ + /* + * We have to combine Hangul characters according to + * http://uniicode.org/reports/tr15/#Hangul + */ + if (0 <= (LIndex = uc - HC_LBASE) && + LIndex < HC_LCOUNT) { + /* + * Hangul Composition. + * 1. Two current code points are L and V. + */ + int VIndex = uc2 - HC_VBASE; + if (0 <= VIndex && VIndex < HC_VCOUNT) { + /* Make syllable of form LV. */ + UPDATE_UC(HC_SBASE + + (LIndex * HC_VCOUNT + VIndex) * + HC_TCOUNT); + } else { + WRITE_UC(); + REPLACE_UC_WITH_UC2(); + } + continue; + } else if (0 <= (SIndex = uc - HC_SBASE) && + SIndex < HC_SCOUNT && (SIndex % HC_TCOUNT) == 0) { + /* + * Hangul Composition. + * 2. Two current code points are LV and T. + */ + int TIndex = uc2 - HC_TBASE; + if (0 < TIndex && TIndex < HC_TCOUNT) { + /* Make syllable of form LVT. */ + UPDATE_UC(uc + TIndex); + } else { + WRITE_UC(); + REPLACE_UC_WITH_UC2(); + } + continue; + } else if ((nfc = get_nfc(uc, uc2)) != 0) { + /* A composition to current code points + * is found. */ + UPDATE_UC(nfc); + continue; + } else if ((cl = CCC(uc2)) == 0) { + /* Clearly 'uc2' the second code point is not + * a decomposable code. */ + WRITE_UC(); + REPLACE_UC_WITH_UC2(); + continue; + } + + /* + * Collect following decomposable code points. + */ + cx = 0; + ucx[0] = uc2; + ccx[0] = cl; + COLLECT_CPS(1); + + /* + * Find a composed code in the collected code points. + */ + i = 1; + while (i < ucx_size) { + int j; + + if ((nfc = get_nfc(uc, ucx[i])) == 0) { + i++; + continue; + } + + /* + * nfc is composed of uc and ucx[i]. + */ + UPDATE_UC(nfc); + + /* + * Remove ucx[i] by shifting + * following code points. + */ + for (j = i; j+1 < ucx_size; j++) { + ucx[j] = ucx[j+1]; + ccx[j] = ccx[j+1]; + } + ucx_size --; + + /* + * Collect following code points blocked + * by ucx[i] the removed code point. + */ + if (ucx_size > 0 && i == ucx_size && + nx > 0 && cx == cl) { + cl = ccx[ucx_size-1]; + COLLECT_CPS(ucx_size); + } + /* + * Restart finding a composed code with + * the updated uc from the top of the + * collected code points. + */ + i = 0; + } + + /* + * Apparently the current code points are not + * decomposed characters or already composed. + */ + WRITE_UC(); + for (i = 0; i < ucx_size; i++) + UNPARSE(p, endp, ucx[i]); + + /* + * Flush out remaining canonical combining characters. + */ + if (nx > 0 && cx == cl && len > 0) { + while ((nx = parse(&ucx[0], s, len)) + > 0) { + cx = CCC(ucx[0]); + if (cl > cx) + break; + s += nx; + len -= nx; + cl = cx; + UNPARSE(p, endp, ucx[0]); + } + } + break; + } + if (n2 < 0) { + WRITE_UC(); + /* Use a replaced unicode character. */ + UNPARSE(p, endp, uc2); + s += n2*-1; + len -= n2*-1; + ret = -1; + continue; + } else if (n2 == 0) { + WRITE_UC(); + break; + } + } + as->length = p - as->s; + as->s[as->length] = '\0'; + if (ts == 2) + as->s[as->length+1] = '\0'; + return (ret); +} + +static int +get_nfd(uint32_t *cp1, uint32_t *cp2, uint32_t uc) +{ + int t, b; + + /* + * These are not converted to NFD on Mac OS. + */ + if ((uc >= 0x2000 && uc <= 0x2FFF) || + (uc >= 0xF900 && uc <= 0xFAFF) || + (uc >= 0x2F800 && uc <= 0x2FAFF)) + return (0); + /* + * Those code points are not converted to NFD on Mac OS. + * I do not know the reason because it is undocumented. + * NFC NFD + * 1109A ==> 11099 110BA + * 1109C ==> 1109B 110BA + * 110AB ==> 110A5 110BA + */ + if (uc == 0x1109A || uc == 0x1109C || uc == 0x110AB) + return (0); + + t = 0; + b = sizeof(u_decomposition_table)/sizeof(u_decomposition_table[0]) -1; + while (b >= t) { + int m = (t + b) / 2; + if (u_decomposition_table[m].nfc < uc) + t = m + 1; + else if (u_decomposition_table[m].nfc > uc) + b = m - 1; + else { + *cp1 = u_decomposition_table[m].cp1; + *cp2 = u_decomposition_table[m].cp2; + return (1); + } + } + return (0); +} + +#define REPLACE_UC_WITH(cp) do { \ + uc = cp; \ + ucptr = NULL; \ +} while (0) + +/* + * Normalize UTF-8 characters to Form D and copy the result. + */ +static int +archive_string_normalize_D(struct archive_string *as, const void *_p, + size_t len, struct archive_string_conv *sc) +{ + const char *s = (const char *)_p; + char *p, *endp; + uint32_t uc, uc2; + size_t w; + int always_replace, n, n2, ret = 0, spair, ts, tm; + int (*parse)(uint32_t *, const char *, size_t); + size_t (*unparse)(char *, size_t, uint32_t); + + always_replace = 1; + ts = 1;/* text size. */ + if (sc->flag & SCONV_TO_UTF16BE) { + unparse = unicode_to_utf16be; + ts = 2; + if (sc->flag & SCONV_FROM_UTF16BE) + always_replace = 0; + } else if (sc->flag & SCONV_TO_UTF16LE) { + unparse = unicode_to_utf16le; + ts = 2; + if (sc->flag & SCONV_FROM_UTF16LE) + always_replace = 0; + } else if (sc->flag & SCONV_TO_UTF8) { + unparse = unicode_to_utf8; + if (sc->flag & SCONV_FROM_UTF8) + always_replace = 0; + } else { + /* + * This case is going to be converted to another + * character-set through iconv. + */ + always_replace = 0; + if (sc->flag & SCONV_FROM_UTF16BE) { + unparse = unicode_to_utf16be; + ts = 2; + } else if (sc->flag & SCONV_FROM_UTF16LE) { + unparse = unicode_to_utf16le; + ts = 2; + } else { + unparse = unicode_to_utf8; + } + } + + if (sc->flag & SCONV_FROM_UTF16BE) { + parse = utf16be_to_unicode; + tm = 1; + spair = 4;/* surrogate pair size in UTF-16. */ + } else if (sc->flag & SCONV_FROM_UTF16LE) { + parse = utf16le_to_unicode; + tm = 1; + spair = 4;/* surrogate pair size in UTF-16. */ + } else { + parse = cesu8_to_unicode; + tm = ts; + spair = 6;/* surrogate pair size in UTF-8. */ + } + + if (archive_string_ensure(as, as->length + len * tm + ts) == NULL) + return (-1); + + p = as->s + as->length; + endp = as->s + as->buffer_length - ts; + while ((n = parse(&uc, s, len)) != 0) { + const char *ucptr; + uint32_t cp1, cp2; + int SIndex; + struct { + uint32_t uc; + int ccc; + } fdc[FDC_MAX]; + int fdi, fdj; + int ccc; + +check_first_code: + if (n < 0) { + /* Use a replaced unicode character. */ + UNPARSE(p, endp, uc); + s += n*-1; + len -= n*-1; + ret = -1; + continue; + } else if (n == spair || always_replace) + /* uc is converted from a surrogate pair. + * this should be treated as a changed code. */ + ucptr = NULL; + else + ucptr = s; + s += n; + len -= n; + + /* Hangul Decomposition. */ + if ((SIndex = uc - HC_SBASE) >= 0 && SIndex < HC_SCOUNT) { + int L = HC_LBASE + SIndex / HC_NCOUNT; + int V = HC_VBASE + (SIndex % HC_NCOUNT) / HC_TCOUNT; + int T = HC_TBASE + SIndex % HC_TCOUNT; + + REPLACE_UC_WITH(L); + WRITE_UC(); + REPLACE_UC_WITH(V); + WRITE_UC(); + if (T != HC_TBASE) { + REPLACE_UC_WITH(T); + WRITE_UC(); + } + continue; + } + if (IS_DECOMPOSABLE_BLOCK(uc) && CCC(uc) != 0) { + WRITE_UC(); + continue; + } + + fdi = 0; + while (get_nfd(&cp1, &cp2, uc) && fdi < FDC_MAX) { + int k; + + for (k = fdi; k > 0; k--) + fdc[k] = fdc[k-1]; + fdc[0].ccc = CCC(cp2); + fdc[0].uc = cp2; + fdi++; + REPLACE_UC_WITH(cp1); + } + + /* Read following code points. */ + while ((n2 = parse(&uc2, s, len)) > 0 && + (ccc = CCC(uc2)) != 0 && fdi < FDC_MAX) { + int j, k; + + s += n2; + len -= n2; + for (j = 0; j < fdi; j++) { + if (fdc[j].ccc > ccc) + break; + } + if (j < fdi) { + for (k = fdi; k > j; k--) + fdc[k] = fdc[k-1]; + fdc[j].ccc = ccc; + fdc[j].uc = uc2; + } else { + fdc[fdi].ccc = ccc; + fdc[fdi].uc = uc2; + } + fdi++; + } + + WRITE_UC(); + for (fdj = 0; fdj < fdi; fdj++) { + REPLACE_UC_WITH(fdc[fdj].uc); + WRITE_UC(); + } + + if (n2 == 0) + break; + REPLACE_UC_WITH(uc2); + n = n2; + goto check_first_code; + } + as->length = p - as->s; + as->s[as->length] = '\0'; + if (ts == 2) + as->s[as->length+1] = '\0'; + return (ret); +} + +/* + * libarchive 2.x made incorrect UTF-8 strings in the wrong assumption + * that WCS is Unicode. It is true for several platforms but some are false. + * And then people who did not use UTF-8 locale on the non Unicode WCS + * platform and made a tar file with libarchive(mostly bsdtar) 2.x. Those + * now cannot get right filename from libarchive 3.x and later since we + * fixed the wrong assumption and it is incompatible to older its versions. + * So we provide special option, "compat-2x.x", for resolving it. + * That option enable the string conversion of libarchive 2.x. + * + * Translates the wrong UTF-8 string made by libarchive 2.x into current + * locale character set and appends to the archive_string. + * Note: returns -1 if conversion fails. + */ +static int +strncat_from_utf8_libarchive2(struct archive_string *as, + const void *_p, size_t len, struct archive_string_conv *sc) +{ + const char *s; + int n; + char *p; + char *end; + uint32_t unicode; +#if HAVE_WCRTOMB + mbstate_t shift_state; + + memset(&shift_state, 0, sizeof(shift_state)); +#else + /* Clear the shift state before starting. */ + wctomb(NULL, L'\0'); +#endif + (void)sc; /* UNUSED */ + /* + * Allocate buffer for MBS. + * We need this allocation here since it is possible that + * as->s is still NULL. + */ + if (archive_string_ensure(as, as->length + len + 1) == NULL) + return (-1); + + s = (const char *)_p; + p = as->s + as->length; + end = as->s + as->buffer_length - MB_CUR_MAX -1; + while ((n = _utf8_to_unicode(&unicode, s, len)) != 0) { + wchar_t wc; + + if (p >= end) { + as->length = p - as->s; + /* Re-allocate buffer for MBS. */ + if (archive_string_ensure(as, + as->length + max(len * 2, + (size_t)MB_CUR_MAX) + 1) == NULL) + return (-1); + p = as->s + as->length; + end = as->s + as->buffer_length - MB_CUR_MAX -1; + } + + /* + * As libarchive 2.x, translates the UTF-8 characters into + * wide-characters in the assumption that WCS is Unicode. + */ + if (n < 0) { + n *= -1; + wc = L'?'; + } else + wc = (wchar_t)unicode; + + s += n; + len -= n; + /* + * Translates the wide-character into the current locale MBS. + */ +#if HAVE_WCRTOMB + n = (int)wcrtomb(p, wc, &shift_state); +#else + n = (int)wctomb(p, wc); +#endif + if (n == -1) + return (-1); + p += n; + } + as->length = p - as->s; + as->s[as->length] = '\0'; + return (0); +} + + +/* + * Conversion functions between current locale dependent MBS and UTF-16BE. + * strncat_from_utf16be() : UTF-16BE --> MBS + * strncat_to_utf16be() : MBS --> UTF16BE + */ + +#if defined(_WIN32) && !defined(__CYGWIN__) + +/* + * Convert a UTF-16BE/LE string to current locale and copy the result. + * Return -1 if conversion fails. + */ +static int +win_strncat_from_utf16(struct archive_string *as, const void *_p, size_t bytes, + struct archive_string_conv *sc, int be) +{ + struct archive_string tmp; + const char *u16; + int ll; + BOOL defchar; + char *mbs; + size_t mbs_size, b; + int ret = 0; + + bytes &= ~1; + if (archive_string_ensure(as, as->length + bytes +1) == NULL) + return (-1); + + mbs = as->s + as->length; + mbs_size = as->buffer_length - as->length -1; + + if (sc->to_cp == CP_C_LOCALE) { + /* + * "C" locale special process. + */ + u16 = _p; + ll = 0; + for (b = 0; b < bytes; b += 2) { + uint16_t val; + if (be) + val = archive_be16dec(u16+b); + else + val = archive_le16dec(u16+b); + if (val > 255) { + *mbs++ = '?'; + ret = -1; + } else + *mbs++ = (char)(val&0xff); + ll++; + } + as->length += ll; + as->s[as->length] = '\0'; + return (ret); + } + + archive_string_init(&tmp); + if (be) { + if (is_big_endian()) { + u16 = _p; + } else { + if (archive_string_ensure(&tmp, bytes+2) == NULL) + return (-1); + memcpy(tmp.s, _p, bytes); + for (b = 0; b < bytes; b += 2) { + uint16_t val = archive_be16dec(tmp.s+b); + archive_le16enc(tmp.s+b, val); + } + u16 = tmp.s; + } + } else { + if (!is_big_endian()) { + u16 = _p; + } else { + if (archive_string_ensure(&tmp, bytes+2) == NULL) + return (-1); + memcpy(tmp.s, _p, bytes); + for (b = 0; b < bytes; b += 2) { + uint16_t val = archive_le16dec(tmp.s+b); + archive_be16enc(tmp.s+b, val); + } + u16 = tmp.s; + } + } + + do { + defchar = 0; + ll = WideCharToMultiByte(sc->to_cp, 0, + (LPCWSTR)u16, (int)bytes>>1, mbs, (int)mbs_size, + NULL, &defchar); + /* Exit loop if we succeeded */ + if (ll != 0 || + GetLastError() != ERROR_INSUFFICIENT_BUFFER) { + break; + } + /* Else expand buffer and loop to try again. */ + ll = WideCharToMultiByte(sc->to_cp, 0, + (LPCWSTR)u16, (int)bytes, NULL, 0, NULL, NULL); + if (archive_string_ensure(as, ll +1) == NULL) + return (-1); + mbs = as->s + as->length; + mbs_size = as->buffer_length - as->length -1; + } while (1); + archive_string_free(&tmp); + as->length += ll; + as->s[as->length] = '\0'; + if (ll == 0 || defchar) + ret = -1; + return (ret); +} + +static int +win_strncat_from_utf16be(struct archive_string *as, const void *_p, + size_t bytes, struct archive_string_conv *sc) +{ + return (win_strncat_from_utf16(as, _p, bytes, sc, 1)); +} + +static int +win_strncat_from_utf16le(struct archive_string *as, const void *_p, + size_t bytes, struct archive_string_conv *sc) +{ + return (win_strncat_from_utf16(as, _p, bytes, sc, 0)); +} + +static int +is_big_endian(void) +{ + uint16_t d = 1; + + return (archive_be16dec(&d) == 1); +} + +/* + * Convert a current locale string to UTF-16BE/LE and copy the result. + * Return -1 if conversion fails. + */ +static int +win_strncat_to_utf16(struct archive_string *as16, const void *_p, + size_t length, struct archive_string_conv *sc, int bigendian) +{ + const char *s = (const char *)_p; + char *u16; + size_t count, avail; + + if (archive_string_ensure(as16, + as16->length + (length + 1) * 2) == NULL) + return (-1); + + u16 = as16->s + as16->length; + avail = as16->buffer_length - 2; + if (sc->from_cp == CP_C_LOCALE) { + /* + * "C" locale special process. + */ + count = 0; + while (count < length && *s) { + if (bigendian) + archive_be16enc(u16, *s); + else + archive_le16enc(u16, *s); + u16 += 2; + s++; + count++; + } + as16->length += count << 1; + as16->s[as16->length] = 0; + as16->s[as16->length+1] = 0; + return (0); + } + do { + count = MultiByteToWideChar(sc->from_cp, + MB_PRECOMPOSED, s, (int)length, (LPWSTR)u16, (int)avail>>1); + /* Exit loop if we succeeded */ + if (count != 0 || + GetLastError() != ERROR_INSUFFICIENT_BUFFER) { + break; + } + /* Expand buffer and try again */ + count = MultiByteToWideChar(sc->from_cp, + MB_PRECOMPOSED, s, (int)length, NULL, 0); + if (archive_string_ensure(as16, (count +1) * 2) + == NULL) + return (-1); + u16 = as16->s + as16->length; + avail = as16->buffer_length - 2; + } while (1); + as16->length += count * 2; + as16->s[as16->length] = 0; + as16->s[as16->length+1] = 0; + if (count == 0) + return (-1); + + if (is_big_endian()) { + if (!bigendian) { + while (count > 0) { + uint16_t v = archive_be16dec(u16); + archive_le16enc(u16, v); + u16 += 2; + count--; + } + } + } else { + if (bigendian) { + while (count > 0) { + uint16_t v = archive_le16dec(u16); + archive_be16enc(u16, v); + u16 += 2; + count--; + } + } + } + return (0); +} + +static int +win_strncat_to_utf16be(struct archive_string *as16, const void *_p, + size_t length, struct archive_string_conv *sc) +{ + return (win_strncat_to_utf16(as16, _p, length, sc, 1)); +} + +static int +win_strncat_to_utf16le(struct archive_string *as16, const void *_p, + size_t length, struct archive_string_conv *sc) +{ + return (win_strncat_to_utf16(as16, _p, length, sc, 0)); +} + +#endif /* _WIN32 && !__CYGWIN__ */ + +/* + * Do the best effort for conversions. + * We cannot handle UTF-16BE character-set without such iconv, + * but there is a chance if a string consists just ASCII code or + * a current locale is UTF-8. + */ + +/* + * Convert a UTF-16BE string to current locale and copy the result. + * Return -1 if conversion fails. + */ +static int +best_effort_strncat_from_utf16(struct archive_string *as, const void *_p, + size_t bytes, struct archive_string_conv *sc, int be) +{ + const char *utf16 = (const char *)_p; + char *mbs; + uint32_t uc; + int n, ret; + + (void)sc; /* UNUSED */ + /* + * Other case, we should do the best effort. + * If all character are ASCII(<0x7f), we can convert it. + * if not , we set a alternative character and return -1. + */ + ret = 0; + if (archive_string_ensure(as, as->length + bytes +1) == NULL) + return (-1); + mbs = as->s + as->length; + + while ((n = utf16_to_unicode(&uc, utf16, bytes, be)) != 0) { + if (n < 0) { + n *= -1; + ret = -1; + } + bytes -= n; + utf16 += n; + + if (uc > 127) { + /* We cannot handle it. */ + *mbs++ = '?'; + ret = -1; + } else + *mbs++ = (char)uc; + } + as->length = mbs - as->s; + as->s[as->length] = '\0'; + return (ret); +} + +static int +best_effort_strncat_from_utf16be(struct archive_string *as, const void *_p, + size_t bytes, struct archive_string_conv *sc) +{ + return (best_effort_strncat_from_utf16(as, _p, bytes, sc, 1)); +} + +static int +best_effort_strncat_from_utf16le(struct archive_string *as, const void *_p, + size_t bytes, struct archive_string_conv *sc) +{ + return (best_effort_strncat_from_utf16(as, _p, bytes, sc, 0)); +} + +/* + * Convert a current locale string to UTF-16BE/LE and copy the result. + * Return -1 if conversion fails. + */ +static int +best_effort_strncat_to_utf16(struct archive_string *as16, const void *_p, + size_t length, struct archive_string_conv *sc, int bigendian) +{ + const char *s = (const char *)_p; + char *utf16; + size_t remaining; + int ret; + + (void)sc; /* UNUSED */ + /* + * Other case, we should do the best effort. + * If all character are ASCII(<0x7f), we can convert it. + * if not , we set a alternative character and return -1. + */ + ret = 0; + remaining = length; + + if (archive_string_ensure(as16, + as16->length + (length + 1) * 2) == NULL) + return (-1); + + utf16 = as16->s + as16->length; + while (remaining--) { + unsigned c = *s++; + if (c > 127) { + /* We cannot handle it. */ + c = UNICODE_R_CHAR; + ret = -1; + } + if (bigendian) + archive_be16enc(utf16, c); + else + archive_le16enc(utf16, c); + utf16 += 2; + } + as16->length = utf16 - as16->s; + as16->s[as16->length] = 0; + as16->s[as16->length+1] = 0; + return (ret); +} + +static int +best_effort_strncat_to_utf16be(struct archive_string *as16, const void *_p, + size_t length, struct archive_string_conv *sc) +{ + return (best_effort_strncat_to_utf16(as16, _p, length, sc, 1)); +} + +static int +best_effort_strncat_to_utf16le(struct archive_string *as16, const void *_p, + size_t length, struct archive_string_conv *sc) +{ + return (best_effort_strncat_to_utf16(as16, _p, length, sc, 0)); +} + + +/* + * Multistring operations. + */ + +void +archive_mstring_clean(struct archive_mstring *aes) +{ + archive_wstring_free(&(aes->aes_wcs)); + archive_string_free(&(aes->aes_mbs)); + archive_string_free(&(aes->aes_utf8)); + archive_string_free(&(aes->aes_mbs_in_locale)); + aes->aes_set = 0; +} + +void +archive_mstring_copy(struct archive_mstring *dest, struct archive_mstring *src) +{ + dest->aes_set = src->aes_set; + archive_string_copy(&(dest->aes_mbs), &(src->aes_mbs)); + archive_string_copy(&(dest->aes_utf8), &(src->aes_utf8)); + archive_wstring_copy(&(dest->aes_wcs), &(src->aes_wcs)); +} + +int +archive_mstring_get_utf8(struct archive *a, struct archive_mstring *aes, + const char **p) +{ + struct archive_string_conv *sc; + int r; + + /* If we already have a UTF8 form, return that immediately. */ + if (aes->aes_set & AES_SET_UTF8) { + *p = aes->aes_utf8.s; + return (0); + } + + *p = NULL; + if (aes->aes_set & AES_SET_MBS) { + sc = archive_string_conversion_to_charset(a, "UTF-8", 1); + if (sc == NULL) + return (-1);/* Couldn't allocate memory for sc. */ + r = archive_strncpy_l(&(aes->aes_utf8), aes->aes_mbs.s, + aes->aes_mbs.length, sc); + if (a == NULL) + free_sconv_object(sc); + if (r == 0) { + aes->aes_set |= AES_SET_UTF8; + *p = aes->aes_utf8.s; + return (0);/* success. */ + } else + return (-1);/* failure. */ + } + return (0);/* success. */ +} + +int +archive_mstring_get_mbs(struct archive *a, struct archive_mstring *aes, + const char **p) +{ + int r, ret = 0; + + (void)a; /* UNUSED */ + /* If we already have an MBS form, return that immediately. */ + if (aes->aes_set & AES_SET_MBS) { + *p = aes->aes_mbs.s; + return (ret); + } + + *p = NULL; + /* If there's a WCS form, try converting with the native locale. */ + if (aes->aes_set & AES_SET_WCS) { + archive_string_empty(&(aes->aes_mbs)); + r = archive_string_append_from_wcs(&(aes->aes_mbs), + aes->aes_wcs.s, aes->aes_wcs.length); + *p = aes->aes_mbs.s; + if (r == 0) { + aes->aes_set |= AES_SET_MBS; + return (ret); + } else + ret = -1; + } + + /* + * Only a UTF-8 form cannot avail because its conversion already + * failed at archive_mstring_update_utf8(). + */ + return (ret); +} + +int +archive_mstring_get_wcs(struct archive *a, struct archive_mstring *aes, + const wchar_t **wp) +{ + int r, ret = 0; + + (void)a;/* UNUSED */ + /* Return WCS form if we already have it. */ + if (aes->aes_set & AES_SET_WCS) { + *wp = aes->aes_wcs.s; + return (ret); + } + + *wp = NULL; + /* Try converting MBS to WCS using native locale. */ + if (aes->aes_set & AES_SET_MBS) { + archive_wstring_empty(&(aes->aes_wcs)); + r = archive_wstring_append_from_mbs(&(aes->aes_wcs), + aes->aes_mbs.s, aes->aes_mbs.length); + if (r == 0) { + aes->aes_set |= AES_SET_WCS; + *wp = aes->aes_wcs.s; + } else + ret = -1;/* failure. */ + } + return (ret); +} + +int +archive_mstring_get_mbs_l(struct archive_mstring *aes, + const char **p, size_t *length, struct archive_string_conv *sc) +{ + int r, ret = 0; + +#if defined(_WIN32) && !defined(__CYGWIN__) + /* + * Internationalization programming on Windows must use Wide + * characters because Windows platform cannot make locale UTF-8. + */ + if (sc != NULL && (aes->aes_set & AES_SET_WCS) != 0) { + archive_string_empty(&(aes->aes_mbs_in_locale)); + r = archive_string_append_from_wcs_in_codepage( + &(aes->aes_mbs_in_locale), aes->aes_wcs.s, + aes->aes_wcs.length, sc); + if (r == 0) { + *p = aes->aes_mbs_in_locale.s; + if (length != NULL) + *length = aes->aes_mbs_in_locale.length; + return (0); + } else if (errno == ENOMEM) + return (-1); + else + ret = -1; + } +#endif + + /* If there is not an MBS form but is a WCS form, try converting + * with the native locale to be used for translating it to specified + * character-set. */ + if ((aes->aes_set & AES_SET_MBS) == 0 && + (aes->aes_set & AES_SET_WCS) != 0) { + archive_string_empty(&(aes->aes_mbs)); + r = archive_string_append_from_wcs(&(aes->aes_mbs), + aes->aes_wcs.s, aes->aes_wcs.length); + if (r == 0) + aes->aes_set |= AES_SET_MBS; + else if (errno == ENOMEM) + return (-1); + else + ret = -1; + } + /* If we already have an MBS form, use it to be translated to + * specified character-set. */ + if (aes->aes_set & AES_SET_MBS) { + if (sc == NULL) { + /* Conversion is unneeded. */ + *p = aes->aes_mbs.s; + if (length != NULL) + *length = aes->aes_mbs.length; + return (0); + } + ret = archive_strncpy_l(&(aes->aes_mbs_in_locale), + aes->aes_mbs.s, aes->aes_mbs.length, sc); + *p = aes->aes_mbs_in_locale.s; + if (length != NULL) + *length = aes->aes_mbs_in_locale.length; + } else { + *p = NULL; + if (length != NULL) + *length = 0; + } + return (ret); +} + +int +archive_mstring_copy_mbs(struct archive_mstring *aes, const char *mbs) +{ + if (mbs == NULL) { + aes->aes_set = 0; + return (0); + } + return (archive_mstring_copy_mbs_len(aes, mbs, strlen(mbs))); +} + +int +archive_mstring_copy_mbs_len(struct archive_mstring *aes, const char *mbs, + size_t len) +{ + if (mbs == NULL) { + aes->aes_set = 0; + return (0); + } + aes->aes_set = AES_SET_MBS; /* Only MBS form is set now. */ + archive_strncpy(&(aes->aes_mbs), mbs, len); + archive_string_empty(&(aes->aes_utf8)); + archive_wstring_empty(&(aes->aes_wcs)); + return (0); +} + +int +archive_mstring_copy_wcs(struct archive_mstring *aes, const wchar_t *wcs) +{ + return archive_mstring_copy_wcs_len(aes, wcs, + wcs == NULL ? 0 : wcslen(wcs)); +} + +int +archive_mstring_copy_utf8(struct archive_mstring *aes, const char *utf8) +{ + if (utf8 == NULL) { + aes->aes_set = 0; + return (0); + } + aes->aes_set = AES_SET_UTF8; + archive_string_empty(&(aes->aes_mbs)); + archive_string_empty(&(aes->aes_wcs)); + archive_strncpy(&(aes->aes_utf8), utf8, strlen(utf8)); + return (int)strlen(utf8); +} + +int +archive_mstring_copy_wcs_len(struct archive_mstring *aes, const wchar_t *wcs, + size_t len) +{ + if (wcs == NULL) { + aes->aes_set = 0; + return (0); + } + aes->aes_set = AES_SET_WCS; /* Only WCS form set. */ + archive_string_empty(&(aes->aes_mbs)); + archive_string_empty(&(aes->aes_utf8)); + archive_wstrncpy(&(aes->aes_wcs), wcs, len); + return (0); +} + +int +archive_mstring_copy_mbs_len_l(struct archive_mstring *aes, + const char *mbs, size_t len, struct archive_string_conv *sc) +{ + int r; + + if (mbs == NULL) { + aes->aes_set = 0; + return (0); + } + archive_string_empty(&(aes->aes_mbs)); + archive_wstring_empty(&(aes->aes_wcs)); + archive_string_empty(&(aes->aes_utf8)); +#if defined(_WIN32) && !defined(__CYGWIN__) + /* + * Internationalization programming on Windows must use Wide + * characters because Windows platform cannot make locale UTF-8. + */ + if (sc == NULL) { + if (archive_string_append(&(aes->aes_mbs), + mbs, mbsnbytes(mbs, len)) == NULL) { + aes->aes_set = 0; + r = -1; + } else { + aes->aes_set = AES_SET_MBS; + r = 0; + } +#if defined(HAVE_ICONV) + } else if (sc != NULL && sc->cd_w != (iconv_t)-1) { + /* + * This case happens only when MultiByteToWideChar() cannot + * handle sc->from_cp, and we have to iconv in order to + * translate character-set to wchar_t,UTF-16. + */ + iconv_t cd = sc->cd; + unsigned from_cp; + int flag; + + /* + * Translate multi-bytes from some character-set to UTF-8. + */ + sc->cd = sc->cd_w; + r = archive_strncpy_l(&(aes->aes_utf8), mbs, len, sc); + sc->cd = cd; + if (r != 0) { + aes->aes_set = 0; + return (r); + } + aes->aes_set = AES_SET_UTF8; + + /* + * Append the UTF-8 string into wstring. + */ + flag = sc->flag; + sc->flag &= ~(SCONV_NORMALIZATION_C + | SCONV_TO_UTF16| SCONV_FROM_UTF16); + from_cp = sc->from_cp; + sc->from_cp = CP_UTF8; + r = archive_wstring_append_from_mbs_in_codepage(&(aes->aes_wcs), + aes->aes_utf8.s, aes->aes_utf8.length, sc); + sc->flag = flag; + sc->from_cp = from_cp; + if (r == 0) + aes->aes_set |= AES_SET_WCS; +#endif + } else { + r = archive_wstring_append_from_mbs_in_codepage( + &(aes->aes_wcs), mbs, len, sc); + if (r == 0) + aes->aes_set = AES_SET_WCS; + else + aes->aes_set = 0; + } +#else + r = archive_strncpy_l(&(aes->aes_mbs), mbs, len, sc); + if (r == 0) + aes->aes_set = AES_SET_MBS; /* Only MBS form is set now. */ + else + aes->aes_set = 0; +#endif + return (r); +} + +/* + * The 'update' form tries to proactively update all forms of + * this string (WCS and MBS) and returns an error if any of + * them fail. This is used by the 'pax' handler, for instance, + * to detect and report character-conversion failures early while + * still allowing clients to get potentially useful values from + * the more tolerant lazy conversions. (get_mbs and get_wcs will + * strive to give the user something useful, so you can get hopefully + * usable values even if some of the character conversions are failing.) + */ +int +archive_mstring_update_utf8(struct archive *a, struct archive_mstring *aes, + const char *utf8) +{ + struct archive_string_conv *sc; + int r; + + if (utf8 == NULL) { + aes->aes_set = 0; + return (0); /* Succeeded in clearing everything. */ + } + + /* Save the UTF8 string. */ + archive_strcpy(&(aes->aes_utf8), utf8); + + /* Empty the mbs and wcs strings. */ + archive_string_empty(&(aes->aes_mbs)); + archive_wstring_empty(&(aes->aes_wcs)); + + aes->aes_set = AES_SET_UTF8; /* Only UTF8 is set now. */ + + /* Try converting UTF-8 to MBS, return false on failure. */ + sc = archive_string_conversion_from_charset(a, "UTF-8", 1); + if (sc == NULL) + return (-1);/* Couldn't allocate memory for sc. */ + r = archive_strcpy_l(&(aes->aes_mbs), utf8, sc); + if (a == NULL) + free_sconv_object(sc); + if (r != 0) + return (-1); + aes->aes_set = AES_SET_UTF8 | AES_SET_MBS; /* Both UTF8 and MBS set. */ + + /* Try converting MBS to WCS, return false on failure. */ + if (archive_wstring_append_from_mbs(&(aes->aes_wcs), aes->aes_mbs.s, + aes->aes_mbs.length)) + return (-1); + aes->aes_set = AES_SET_UTF8 | AES_SET_WCS | AES_SET_MBS; + + /* All conversions succeeded. */ + return (0); +} diff --git a/c/archive_string.h b/c/archive_string.h new file mode 100644 --- /dev/null +++ b/c/archive_string.h @@ -0,0 +1,243 @@ +/*- + * Copyright (c) 2003-2010 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD: head/lib/libarchive/archive_string.h 201092 2009-12-28 02:26:06Z kientzle $ + * + */ + +#ifndef ARCHIVE_STRING_H_INCLUDED +#define ARCHIVE_STRING_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#ifndef __LIBARCHIVE_TEST +#error This header is only to be used internally to libarchive. +#endif +#endif + +#include +#ifdef HAVE_STDLIB_H +#include /* required for wchar_t on some systems */ +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_WCHAR_H +#include +#endif + +#include "archive.h" + +/* + * Basic resizable/reusable string support similar to Java's "StringBuffer." + * + * Unlike sbuf(9), the buffers here are fully reusable and track the + * length throughout. + */ + +struct archive_string { + char *s; /* Pointer to the storage */ + size_t length; /* Length of 's' in characters */ + size_t buffer_length; /* Length of malloc-ed storage in bytes. */ +}; + +struct archive_wstring { + wchar_t *s; /* Pointer to the storage */ + size_t length; /* Length of 's' in characters */ + size_t buffer_length; /* Length of malloc-ed storage in bytes. */ +}; + +struct archive_string_conv; + +/* Initialize an archive_string object on the stack or elsewhere. */ +#define archive_string_init(a) \ + do { (a)->s = NULL; (a)->length = 0; (a)->buffer_length = 0; } while(0) + +/* Append a C char to an archive_string, resizing as necessary. */ +struct archive_string * +archive_strappend_char(struct archive_string *, char); + +/* Ditto for a wchar_t and an archive_wstring. */ +struct archive_wstring * +archive_wstrappend_wchar(struct archive_wstring *, wchar_t); + +/* Append a raw array to an archive_string, resizing as necessary */ +struct archive_string * +archive_array_append(struct archive_string *, const char *, size_t); + +/* Convert a Unicode string to current locale and append the result. */ +/* Returns -1 if conversion fails. */ +int +archive_string_append_from_wcs(struct archive_string *, const wchar_t *, size_t); + + +/* Create a string conversion object. + * Return NULL and set a error message if the conversion is not supported + * on the platform. */ +struct archive_string_conv * +archive_string_conversion_to_charset(struct archive *, const char *, int); +struct archive_string_conv * +archive_string_conversion_from_charset(struct archive *, const char *, int); +/* Create the default string conversion object for reading/writing an archive. + * Return NULL if the conversion is unneeded. + * Note: On non Windows platform this always returns NULL. + */ +struct archive_string_conv * +archive_string_default_conversion_for_read(struct archive *); +struct archive_string_conv * +archive_string_default_conversion_for_write(struct archive *); +/* Dispose of a string conversion object. */ +void +archive_string_conversion_free(struct archive *); +const char * +archive_string_conversion_charset_name(struct archive_string_conv *); +void +archive_string_conversion_set_opt(struct archive_string_conv *, int); +#define SCONV_SET_OPT_UTF8_LIBARCHIVE2X 1 +#define SCONV_SET_OPT_NORMALIZATION_C 2 +#define SCONV_SET_OPT_NORMALIZATION_D 4 + + +/* Copy one archive_string to another in locale conversion. + * Return -1 if conversion fails. */ +int +archive_strncpy_l(struct archive_string *, const void *, size_t, + struct archive_string_conv *); + +/* Copy one archive_string to another in locale conversion. + * Return -1 if conversion fails. */ +int +archive_strncat_l(struct archive_string *, const void *, size_t, + struct archive_string_conv *); + + +/* Copy one archive_string to another */ +#define archive_string_copy(dest, src) \ + ((dest)->length = 0, archive_string_concat((dest), (src))) +#define archive_wstring_copy(dest, src) \ + ((dest)->length = 0, archive_wstring_concat((dest), (src))) + +/* Concatenate one archive_string to another */ +void archive_string_concat(struct archive_string *dest, struct archive_string *src); +void archive_wstring_concat(struct archive_wstring *dest, struct archive_wstring *src); + +/* Ensure that the underlying buffer is at least as large as the request. */ +struct archive_string * +archive_string_ensure(struct archive_string *, size_t); +struct archive_wstring * +archive_wstring_ensure(struct archive_wstring *, size_t); + +/* Append C string, which may lack trailing \0. */ +/* The source is declared void * here because this gets used with + * "signed char *", "unsigned char *" and "char *" arguments. + * Declaring it "char *" as with some of the other functions just + * leads to a lot of extra casts. */ +struct archive_string * +archive_strncat(struct archive_string *, const void *, size_t); +struct archive_wstring * +archive_wstrncat(struct archive_wstring *, const wchar_t *, size_t); + +/* Append a C string to an archive_string, resizing as necessary. */ +struct archive_string * +archive_strcat(struct archive_string *, const void *); +struct archive_wstring * +archive_wstrcat(struct archive_wstring *, const wchar_t *); + +/* Copy a C string to an archive_string, resizing as necessary. */ +#define archive_strcpy(as,p) \ + archive_strncpy((as), (p), ((p) == NULL ? 0 : strlen(p))) +#define archive_wstrcpy(as,p) \ + archive_wstrncpy((as), (p), ((p) == NULL ? 0 : wcslen(p))) +#define archive_strcpy_l(as,p,lo) \ + archive_strncpy_l((as), (p), ((p) == NULL ? 0 : strlen(p)), (lo)) + +/* Copy a C string to an archive_string with limit, resizing as necessary. */ +#define archive_strncpy(as,p,l) \ + ((as)->length=0, archive_strncat((as), (p), (l))) +#define archive_wstrncpy(as,p,l) \ + ((as)->length = 0, archive_wstrncat((as), (p), (l))) + +/* Return length of string. */ +#define archive_strlen(a) ((a)->length) + +/* Set string length to zero. */ +#define archive_string_empty(a) ((a)->length = 0) +#define archive_wstring_empty(a) ((a)->length = 0) + +/* Release any allocated storage resources. */ +void archive_string_free(struct archive_string *); +void archive_wstring_free(struct archive_wstring *); + +/* Like 'vsprintf', but resizes the underlying string as necessary. */ +/* Note: This only implements a small subset of standard printf functionality. */ +void archive_string_vsprintf(struct archive_string *, const char *, + va_list) __LA_PRINTF(2, 0); +void archive_string_sprintf(struct archive_string *, const char *, ...) + __LA_PRINTF(2, 3); + +/* Translates from MBS to Unicode. */ +/* Returns non-zero if conversion failed in any way. */ +int archive_wstring_append_from_mbs(struct archive_wstring *dest, + const char *, size_t); + + +/* A "multistring" can hold Unicode, UTF8, or MBS versions of + * the string. If you set and read the same version, no translation + * is done. If you set and read different versions, the library + * will attempt to transparently convert. + */ +struct archive_mstring { + struct archive_string aes_mbs; + struct archive_string aes_utf8; + struct archive_wstring aes_wcs; + struct archive_string aes_mbs_in_locale; + /* Bitmap of which of the above are valid. Because we're lazy + * about malloc-ing and reusing the underlying storage, we + * can't rely on NULL pointers to indicate whether a string + * has been set. */ + int aes_set; +#define AES_SET_MBS 1 +#define AES_SET_UTF8 2 +#define AES_SET_WCS 4 +}; + +void archive_mstring_clean(struct archive_mstring *); +void archive_mstring_copy(struct archive_mstring *dest, struct archive_mstring *src); +int archive_mstring_get_mbs(struct archive *, struct archive_mstring *, const char **); +int archive_mstring_get_utf8(struct archive *, struct archive_mstring *, const char **); +int archive_mstring_get_wcs(struct archive *, struct archive_mstring *, const wchar_t **); +int archive_mstring_get_mbs_l(struct archive_mstring *, const char **, + size_t *, struct archive_string_conv *); +int archive_mstring_copy_mbs(struct archive_mstring *, const char *mbs); +int archive_mstring_copy_mbs_len(struct archive_mstring *, const char *mbs, + size_t); +int archive_mstring_copy_utf8(struct archive_mstring *, const char *utf8); +int archive_mstring_copy_wcs(struct archive_mstring *, const wchar_t *wcs); +int archive_mstring_copy_wcs_len(struct archive_mstring *, + const wchar_t *wcs, size_t); +int archive_mstring_copy_mbs_len_l(struct archive_mstring *, + const char *mbs, size_t, struct archive_string_conv *); +int archive_mstring_update_utf8(struct archive *, struct archive_mstring *aes, const char *utf8); + + +#endif diff --git a/c/archive_string_composition.h b/c/archive_string_composition.h new file mode 100644 --- /dev/null +++ b/c/archive_string_composition.h @@ -0,0 +1,2292 @@ +/*- + * Copyright (c) 2011-2012 libarchive Project + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD$ + * + */ + +/* + * ATTENTION! + * This file is generated by build/utils/gen_archive_string_composition_h.sh + * from http://unicode.org/Public/6.0.0/ucd/UnicodeData.txt + * + * See also http://unicode.org/report/tr15/ + */ + +#ifndef ARCHIVE_STRING_COMPOSITION_H_INCLUDED +#define ARCHIVE_STRING_COMPOSITION_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif + +struct unicode_composition_table { + uint32_t cp1; + uint32_t cp2; + uint32_t nfc; +}; + +static const struct unicode_composition_table u_composition_table[] = { + { 0x0003C , 0x00338 , 0x0226E }, + { 0x0003D , 0x00338 , 0x02260 }, + { 0x0003E , 0x00338 , 0x0226F }, + { 0x00041 , 0x00300 , 0x000C0 }, + { 0x00041 , 0x00301 , 0x000C1 }, + { 0x00041 , 0x00302 , 0x000C2 }, + { 0x00041 , 0x00303 , 0x000C3 }, + { 0x00041 , 0x00304 , 0x00100 }, + { 0x00041 , 0x00306 , 0x00102 }, + { 0x00041 , 0x00307 , 0x00226 }, + { 0x00041 , 0x00308 , 0x000C4 }, + { 0x00041 , 0x00309 , 0x01EA2 }, + { 0x00041 , 0x0030A , 0x000C5 }, + { 0x00041 , 0x0030C , 0x001CD }, + { 0x00041 , 0x0030F , 0x00200 }, + { 0x00041 , 0x00311 , 0x00202 }, + { 0x00041 , 0x00323 , 0x01EA0 }, + { 0x00041 , 0x00325 , 0x01E00 }, + { 0x00041 , 0x00328 , 0x00104 }, + { 0x00042 , 0x00307 , 0x01E02 }, + { 0x00042 , 0x00323 , 0x01E04 }, + { 0x00042 , 0x00331 , 0x01E06 }, + { 0x00043 , 0x00301 , 0x00106 }, + { 0x00043 , 0x00302 , 0x00108 }, + { 0x00043 , 0x00307 , 0x0010A }, + { 0x00043 , 0x0030C , 0x0010C }, + { 0x00043 , 0x00327 , 0x000C7 }, + { 0x00044 , 0x00307 , 0x01E0A }, + { 0x00044 , 0x0030C , 0x0010E }, + { 0x00044 , 0x00323 , 0x01E0C }, + { 0x00044 , 0x00327 , 0x01E10 }, + { 0x00044 , 0x0032D , 0x01E12 }, + { 0x00044 , 0x00331 , 0x01E0E }, + { 0x00045 , 0x00300 , 0x000C8 }, + { 0x00045 , 0x00301 , 0x000C9 }, + { 0x00045 , 0x00302 , 0x000CA }, + { 0x00045 , 0x00303 , 0x01EBC }, + { 0x00045 , 0x00304 , 0x00112 }, + { 0x00045 , 0x00306 , 0x00114 }, + { 0x00045 , 0x00307 , 0x00116 }, + { 0x00045 , 0x00308 , 0x000CB }, + { 0x00045 , 0x00309 , 0x01EBA }, + { 0x00045 , 0x0030C , 0x0011A }, + { 0x00045 , 0x0030F , 0x00204 }, + { 0x00045 , 0x00311 , 0x00206 }, + { 0x00045 , 0x00323 , 0x01EB8 }, + { 0x00045 , 0x00327 , 0x00228 }, + { 0x00045 , 0x00328 , 0x00118 }, + { 0x00045 , 0x0032D , 0x01E18 }, + { 0x00045 , 0x00330 , 0x01E1A }, + { 0x00046 , 0x00307 , 0x01E1E }, + { 0x00047 , 0x00301 , 0x001F4 }, + { 0x00047 , 0x00302 , 0x0011C }, + { 0x00047 , 0x00304 , 0x01E20 }, + { 0x00047 , 0x00306 , 0x0011E }, + { 0x00047 , 0x00307 , 0x00120 }, + { 0x00047 , 0x0030C , 0x001E6 }, + { 0x00047 , 0x00327 , 0x00122 }, + { 0x00048 , 0x00302 , 0x00124 }, + { 0x00048 , 0x00307 , 0x01E22 }, + { 0x00048 , 0x00308 , 0x01E26 }, + { 0x00048 , 0x0030C , 0x0021E }, + { 0x00048 , 0x00323 , 0x01E24 }, + { 0x00048 , 0x00327 , 0x01E28 }, + { 0x00048 , 0x0032E , 0x01E2A }, + { 0x00049 , 0x00300 , 0x000CC }, + { 0x00049 , 0x00301 , 0x000CD }, + { 0x00049 , 0x00302 , 0x000CE }, + { 0x00049 , 0x00303 , 0x00128 }, + { 0x00049 , 0x00304 , 0x0012A }, + { 0x00049 , 0x00306 , 0x0012C }, + { 0x00049 , 0x00307 , 0x00130 }, + { 0x00049 , 0x00308 , 0x000CF }, + { 0x00049 , 0x00309 , 0x01EC8 }, + { 0x00049 , 0x0030C , 0x001CF }, + { 0x00049 , 0x0030F , 0x00208 }, + { 0x00049 , 0x00311 , 0x0020A }, + { 0x00049 , 0x00323 , 0x01ECA }, + { 0x00049 , 0x00328 , 0x0012E }, + { 0x00049 , 0x00330 , 0x01E2C }, + { 0x0004A , 0x00302 , 0x00134 }, + { 0x0004B , 0x00301 , 0x01E30 }, + { 0x0004B , 0x0030C , 0x001E8 }, + { 0x0004B , 0x00323 , 0x01E32 }, + { 0x0004B , 0x00327 , 0x00136 }, + { 0x0004B , 0x00331 , 0x01E34 }, + { 0x0004C , 0x00301 , 0x00139 }, + { 0x0004C , 0x0030C , 0x0013D }, + { 0x0004C , 0x00323 , 0x01E36 }, + { 0x0004C , 0x00327 , 0x0013B }, + { 0x0004C , 0x0032D , 0x01E3C }, + { 0x0004C , 0x00331 , 0x01E3A }, + { 0x0004D , 0x00301 , 0x01E3E }, + { 0x0004D , 0x00307 , 0x01E40 }, + { 0x0004D , 0x00323 , 0x01E42 }, + { 0x0004E , 0x00300 , 0x001F8 }, + { 0x0004E , 0x00301 , 0x00143 }, + { 0x0004E , 0x00303 , 0x000D1 }, + { 0x0004E , 0x00307 , 0x01E44 }, + { 0x0004E , 0x0030C , 0x00147 }, + { 0x0004E , 0x00323 , 0x01E46 }, + { 0x0004E , 0x00327 , 0x00145 }, + { 0x0004E , 0x0032D , 0x01E4A }, + { 0x0004E , 0x00331 , 0x01E48 }, + { 0x0004F , 0x00300 , 0x000D2 }, + { 0x0004F , 0x00301 , 0x000D3 }, + { 0x0004F , 0x00302 , 0x000D4 }, + { 0x0004F , 0x00303 , 0x000D5 }, + { 0x0004F , 0x00304 , 0x0014C }, + { 0x0004F , 0x00306 , 0x0014E }, + { 0x0004F , 0x00307 , 0x0022E }, + { 0x0004F , 0x00308 , 0x000D6 }, + { 0x0004F , 0x00309 , 0x01ECE }, + { 0x0004F , 0x0030B , 0x00150 }, + { 0x0004F , 0x0030C , 0x001D1 }, + { 0x0004F , 0x0030F , 0x0020C }, + { 0x0004F , 0x00311 , 0x0020E }, + { 0x0004F , 0x0031B , 0x001A0 }, + { 0x0004F , 0x00323 , 0x01ECC }, + { 0x0004F , 0x00328 , 0x001EA }, + { 0x00050 , 0x00301 , 0x01E54 }, + { 0x00050 , 0x00307 , 0x01E56 }, + { 0x00052 , 0x00301 , 0x00154 }, + { 0x00052 , 0x00307 , 0x01E58 }, + { 0x00052 , 0x0030C , 0x00158 }, + { 0x00052 , 0x0030F , 0x00210 }, + { 0x00052 , 0x00311 , 0x00212 }, + { 0x00052 , 0x00323 , 0x01E5A }, + { 0x00052 , 0x00327 , 0x00156 }, + { 0x00052 , 0x00331 , 0x01E5E }, + { 0x00053 , 0x00301 , 0x0015A }, + { 0x00053 , 0x00302 , 0x0015C }, + { 0x00053 , 0x00307 , 0x01E60 }, + { 0x00053 , 0x0030C , 0x00160 }, + { 0x00053 , 0x00323 , 0x01E62 }, + { 0x00053 , 0x00326 , 0x00218 }, + { 0x00053 , 0x00327 , 0x0015E }, + { 0x00054 , 0x00307 , 0x01E6A }, + { 0x00054 , 0x0030C , 0x00164 }, + { 0x00054 , 0x00323 , 0x01E6C }, + { 0x00054 , 0x00326 , 0x0021A }, + { 0x00054 , 0x00327 , 0x00162 }, + { 0x00054 , 0x0032D , 0x01E70 }, + { 0x00054 , 0x00331 , 0x01E6E }, + { 0x00055 , 0x00300 , 0x000D9 }, + { 0x00055 , 0x00301 , 0x000DA }, + { 0x00055 , 0x00302 , 0x000DB }, + { 0x00055 , 0x00303 , 0x00168 }, + { 0x00055 , 0x00304 , 0x0016A }, + { 0x00055 , 0x00306 , 0x0016C }, + { 0x00055 , 0x00308 , 0x000DC }, + { 0x00055 , 0x00309 , 0x01EE6 }, + { 0x00055 , 0x0030A , 0x0016E }, + { 0x00055 , 0x0030B , 0x00170 }, + { 0x00055 , 0x0030C , 0x001D3 }, + { 0x00055 , 0x0030F , 0x00214 }, + { 0x00055 , 0x00311 , 0x00216 }, + { 0x00055 , 0x0031B , 0x001AF }, + { 0x00055 , 0x00323 , 0x01EE4 }, + { 0x00055 , 0x00324 , 0x01E72 }, + { 0x00055 , 0x00328 , 0x00172 }, + { 0x00055 , 0x0032D , 0x01E76 }, + { 0x00055 , 0x00330 , 0x01E74 }, + { 0x00056 , 0x00303 , 0x01E7C }, + { 0x00056 , 0x00323 , 0x01E7E }, + { 0x00057 , 0x00300 , 0x01E80 }, + { 0x00057 , 0x00301 , 0x01E82 }, + { 0x00057 , 0x00302 , 0x00174 }, + { 0x00057 , 0x00307 , 0x01E86 }, + { 0x00057 , 0x00308 , 0x01E84 }, + { 0x00057 , 0x00323 , 0x01E88 }, + { 0x00058 , 0x00307 , 0x01E8A }, + { 0x00058 , 0x00308 , 0x01E8C }, + { 0x00059 , 0x00300 , 0x01EF2 }, + { 0x00059 , 0x00301 , 0x000DD }, + { 0x00059 , 0x00302 , 0x00176 }, + { 0x00059 , 0x00303 , 0x01EF8 }, + { 0x00059 , 0x00304 , 0x00232 }, + { 0x00059 , 0x00307 , 0x01E8E }, + { 0x00059 , 0x00308 , 0x00178 }, + { 0x00059 , 0x00309 , 0x01EF6 }, + { 0x00059 , 0x00323 , 0x01EF4 }, + { 0x0005A , 0x00301 , 0x00179 }, + { 0x0005A , 0x00302 , 0x01E90 }, + { 0x0005A , 0x00307 , 0x0017B }, + { 0x0005A , 0x0030C , 0x0017D }, + { 0x0005A , 0x00323 , 0x01E92 }, + { 0x0005A , 0x00331 , 0x01E94 }, + { 0x00061 , 0x00300 , 0x000E0 }, + { 0x00061 , 0x00301 , 0x000E1 }, + { 0x00061 , 0x00302 , 0x000E2 }, + { 0x00061 , 0x00303 , 0x000E3 }, + { 0x00061 , 0x00304 , 0x00101 }, + { 0x00061 , 0x00306 , 0x00103 }, + { 0x00061 , 0x00307 , 0x00227 }, + { 0x00061 , 0x00308 , 0x000E4 }, + { 0x00061 , 0x00309 , 0x01EA3 }, + { 0x00061 , 0x0030A , 0x000E5 }, + { 0x00061 , 0x0030C , 0x001CE }, + { 0x00061 , 0x0030F , 0x00201 }, + { 0x00061 , 0x00311 , 0x00203 }, + { 0x00061 , 0x00323 , 0x01EA1 }, + { 0x00061 , 0x00325 , 0x01E01 }, + { 0x00061 , 0x00328 , 0x00105 }, + { 0x00062 , 0x00307 , 0x01E03 }, + { 0x00062 , 0x00323 , 0x01E05 }, + { 0x00062 , 0x00331 , 0x01E07 }, + { 0x00063 , 0x00301 , 0x00107 }, + { 0x00063 , 0x00302 , 0x00109 }, + { 0x00063 , 0x00307 , 0x0010B }, + { 0x00063 , 0x0030C , 0x0010D }, + { 0x00063 , 0x00327 , 0x000E7 }, + { 0x00064 , 0x00307 , 0x01E0B }, + { 0x00064 , 0x0030C , 0x0010F }, + { 0x00064 , 0x00323 , 0x01E0D }, + { 0x00064 , 0x00327 , 0x01E11 }, + { 0x00064 , 0x0032D , 0x01E13 }, + { 0x00064 , 0x00331 , 0x01E0F }, + { 0x00065 , 0x00300 , 0x000E8 }, + { 0x00065 , 0x00301 , 0x000E9 }, + { 0x00065 , 0x00302 , 0x000EA }, + { 0x00065 , 0x00303 , 0x01EBD }, + { 0x00065 , 0x00304 , 0x00113 }, + { 0x00065 , 0x00306 , 0x00115 }, + { 0x00065 , 0x00307 , 0x00117 }, + { 0x00065 , 0x00308 , 0x000EB }, + { 0x00065 , 0x00309 , 0x01EBB }, + { 0x00065 , 0x0030C , 0x0011B }, + { 0x00065 , 0x0030F , 0x00205 }, + { 0x00065 , 0x00311 , 0x00207 }, + { 0x00065 , 0x00323 , 0x01EB9 }, + { 0x00065 , 0x00327 , 0x00229 }, + { 0x00065 , 0x00328 , 0x00119 }, + { 0x00065 , 0x0032D , 0x01E19 }, + { 0x00065 , 0x00330 , 0x01E1B }, + { 0x00066 , 0x00307 , 0x01E1F }, + { 0x00067 , 0x00301 , 0x001F5 }, + { 0x00067 , 0x00302 , 0x0011D }, + { 0x00067 , 0x00304 , 0x01E21 }, + { 0x00067 , 0x00306 , 0x0011F }, + { 0x00067 , 0x00307 , 0x00121 }, + { 0x00067 , 0x0030C , 0x001E7 }, + { 0x00067 , 0x00327 , 0x00123 }, + { 0x00068 , 0x00302 , 0x00125 }, + { 0x00068 , 0x00307 , 0x01E23 }, + { 0x00068 , 0x00308 , 0x01E27 }, + { 0x00068 , 0x0030C , 0x0021F }, + { 0x00068 , 0x00323 , 0x01E25 }, + { 0x00068 , 0x00327 , 0x01E29 }, + { 0x00068 , 0x0032E , 0x01E2B }, + { 0x00068 , 0x00331 , 0x01E96 }, + { 0x00069 , 0x00300 , 0x000EC }, + { 0x00069 , 0x00301 , 0x000ED }, + { 0x00069 , 0x00302 , 0x000EE }, + { 0x00069 , 0x00303 , 0x00129 }, + { 0x00069 , 0x00304 , 0x0012B }, + { 0x00069 , 0x00306 , 0x0012D }, + { 0x00069 , 0x00308 , 0x000EF }, + { 0x00069 , 0x00309 , 0x01EC9 }, + { 0x00069 , 0x0030C , 0x001D0 }, + { 0x00069 , 0x0030F , 0x00209 }, + { 0x00069 , 0x00311 , 0x0020B }, + { 0x00069 , 0x00323 , 0x01ECB }, + { 0x00069 , 0x00328 , 0x0012F }, + { 0x00069 , 0x00330 , 0x01E2D }, + { 0x0006A , 0x00302 , 0x00135 }, + { 0x0006A , 0x0030C , 0x001F0 }, + { 0x0006B , 0x00301 , 0x01E31 }, + { 0x0006B , 0x0030C , 0x001E9 }, + { 0x0006B , 0x00323 , 0x01E33 }, + { 0x0006B , 0x00327 , 0x00137 }, + { 0x0006B , 0x00331 , 0x01E35 }, + { 0x0006C , 0x00301 , 0x0013A }, + { 0x0006C , 0x0030C , 0x0013E }, + { 0x0006C , 0x00323 , 0x01E37 }, + { 0x0006C , 0x00327 , 0x0013C }, + { 0x0006C , 0x0032D , 0x01E3D }, + { 0x0006C , 0x00331 , 0x01E3B }, + { 0x0006D , 0x00301 , 0x01E3F }, + { 0x0006D , 0x00307 , 0x01E41 }, + { 0x0006D , 0x00323 , 0x01E43 }, + { 0x0006E , 0x00300 , 0x001F9 }, + { 0x0006E , 0x00301 , 0x00144 }, + { 0x0006E , 0x00303 , 0x000F1 }, + { 0x0006E , 0x00307 , 0x01E45 }, + { 0x0006E , 0x0030C , 0x00148 }, + { 0x0006E , 0x00323 , 0x01E47 }, + { 0x0006E , 0x00327 , 0x00146 }, + { 0x0006E , 0x0032D , 0x01E4B }, + { 0x0006E , 0x00331 , 0x01E49 }, + { 0x0006F , 0x00300 , 0x000F2 }, + { 0x0006F , 0x00301 , 0x000F3 }, + { 0x0006F , 0x00302 , 0x000F4 }, + { 0x0006F , 0x00303 , 0x000F5 }, + { 0x0006F , 0x00304 , 0x0014D }, + { 0x0006F , 0x00306 , 0x0014F }, + { 0x0006F , 0x00307 , 0x0022F }, + { 0x0006F , 0x00308 , 0x000F6 }, + { 0x0006F , 0x00309 , 0x01ECF }, + { 0x0006F , 0x0030B , 0x00151 }, + { 0x0006F , 0x0030C , 0x001D2 }, + { 0x0006F , 0x0030F , 0x0020D }, + { 0x0006F , 0x00311 , 0x0020F }, + { 0x0006F , 0x0031B , 0x001A1 }, + { 0x0006F , 0x00323 , 0x01ECD }, + { 0x0006F , 0x00328 , 0x001EB }, + { 0x00070 , 0x00301 , 0x01E55 }, + { 0x00070 , 0x00307 , 0x01E57 }, + { 0x00072 , 0x00301 , 0x00155 }, + { 0x00072 , 0x00307 , 0x01E59 }, + { 0x00072 , 0x0030C , 0x00159 }, + { 0x00072 , 0x0030F , 0x00211 }, + { 0x00072 , 0x00311 , 0x00213 }, + { 0x00072 , 0x00323 , 0x01E5B }, + { 0x00072 , 0x00327 , 0x00157 }, + { 0x00072 , 0x00331 , 0x01E5F }, + { 0x00073 , 0x00301 , 0x0015B }, + { 0x00073 , 0x00302 , 0x0015D }, + { 0x00073 , 0x00307 , 0x01E61 }, + { 0x00073 , 0x0030C , 0x00161 }, + { 0x00073 , 0x00323 , 0x01E63 }, + { 0x00073 , 0x00326 , 0x00219 }, + { 0x00073 , 0x00327 , 0x0015F }, + { 0x00074 , 0x00307 , 0x01E6B }, + { 0x00074 , 0x00308 , 0x01E97 }, + { 0x00074 , 0x0030C , 0x00165 }, + { 0x00074 , 0x00323 , 0x01E6D }, + { 0x00074 , 0x00326 , 0x0021B }, + { 0x00074 , 0x00327 , 0x00163 }, + { 0x00074 , 0x0032D , 0x01E71 }, + { 0x00074 , 0x00331 , 0x01E6F }, + { 0x00075 , 0x00300 , 0x000F9 }, + { 0x00075 , 0x00301 , 0x000FA }, + { 0x00075 , 0x00302 , 0x000FB }, + { 0x00075 , 0x00303 , 0x00169 }, + { 0x00075 , 0x00304 , 0x0016B }, + { 0x00075 , 0x00306 , 0x0016D }, + { 0x00075 , 0x00308 , 0x000FC }, + { 0x00075 , 0x00309 , 0x01EE7 }, + { 0x00075 , 0x0030A , 0x0016F }, + { 0x00075 , 0x0030B , 0x00171 }, + { 0x00075 , 0x0030C , 0x001D4 }, + { 0x00075 , 0x0030F , 0x00215 }, + { 0x00075 , 0x00311 , 0x00217 }, + { 0x00075 , 0x0031B , 0x001B0 }, + { 0x00075 , 0x00323 , 0x01EE5 }, + { 0x00075 , 0x00324 , 0x01E73 }, + { 0x00075 , 0x00328 , 0x00173 }, + { 0x00075 , 0x0032D , 0x01E77 }, + { 0x00075 , 0x00330 , 0x01E75 }, + { 0x00076 , 0x00303 , 0x01E7D }, + { 0x00076 , 0x00323 , 0x01E7F }, + { 0x00077 , 0x00300 , 0x01E81 }, + { 0x00077 , 0x00301 , 0x01E83 }, + { 0x00077 , 0x00302 , 0x00175 }, + { 0x00077 , 0x00307 , 0x01E87 }, + { 0x00077 , 0x00308 , 0x01E85 }, + { 0x00077 , 0x0030A , 0x01E98 }, + { 0x00077 , 0x00323 , 0x01E89 }, + { 0x00078 , 0x00307 , 0x01E8B }, + { 0x00078 , 0x00308 , 0x01E8D }, + { 0x00079 , 0x00300 , 0x01EF3 }, + { 0x00079 , 0x00301 , 0x000FD }, + { 0x00079 , 0x00302 , 0x00177 }, + { 0x00079 , 0x00303 , 0x01EF9 }, + { 0x00079 , 0x00304 , 0x00233 }, + { 0x00079 , 0x00307 , 0x01E8F }, + { 0x00079 , 0x00308 , 0x000FF }, + { 0x00079 , 0x00309 , 0x01EF7 }, + { 0x00079 , 0x0030A , 0x01E99 }, + { 0x00079 , 0x00323 , 0x01EF5 }, + { 0x0007A , 0x00301 , 0x0017A }, + { 0x0007A , 0x00302 , 0x01E91 }, + { 0x0007A , 0x00307 , 0x0017C }, + { 0x0007A , 0x0030C , 0x0017E }, + { 0x0007A , 0x00323 , 0x01E93 }, + { 0x0007A , 0x00331 , 0x01E95 }, + { 0x000A8 , 0x00300 , 0x01FED }, + { 0x000A8 , 0x00301 , 0x00385 }, + { 0x000A8 , 0x00342 , 0x01FC1 }, + { 0x000C2 , 0x00300 , 0x01EA6 }, + { 0x000C2 , 0x00301 , 0x01EA4 }, + { 0x000C2 , 0x00303 , 0x01EAA }, + { 0x000C2 , 0x00309 , 0x01EA8 }, + { 0x000C4 , 0x00304 , 0x001DE }, + { 0x000C5 , 0x00301 , 0x001FA }, + { 0x000C6 , 0x00301 , 0x001FC }, + { 0x000C6 , 0x00304 , 0x001E2 }, + { 0x000C7 , 0x00301 , 0x01E08 }, + { 0x000CA , 0x00300 , 0x01EC0 }, + { 0x000CA , 0x00301 , 0x01EBE }, + { 0x000CA , 0x00303 , 0x01EC4 }, + { 0x000CA , 0x00309 , 0x01EC2 }, + { 0x000CF , 0x00301 , 0x01E2E }, + { 0x000D4 , 0x00300 , 0x01ED2 }, + { 0x000D4 , 0x00301 , 0x01ED0 }, + { 0x000D4 , 0x00303 , 0x01ED6 }, + { 0x000D4 , 0x00309 , 0x01ED4 }, + { 0x000D5 , 0x00301 , 0x01E4C }, + { 0x000D5 , 0x00304 , 0x0022C }, + { 0x000D5 , 0x00308 , 0x01E4E }, + { 0x000D6 , 0x00304 , 0x0022A }, + { 0x000D8 , 0x00301 , 0x001FE }, + { 0x000DC , 0x00300 , 0x001DB }, + { 0x000DC , 0x00301 , 0x001D7 }, + { 0x000DC , 0x00304 , 0x001D5 }, + { 0x000DC , 0x0030C , 0x001D9 }, + { 0x000E2 , 0x00300 , 0x01EA7 }, + { 0x000E2 , 0x00301 , 0x01EA5 }, + { 0x000E2 , 0x00303 , 0x01EAB }, + { 0x000E2 , 0x00309 , 0x01EA9 }, + { 0x000E4 , 0x00304 , 0x001DF }, + { 0x000E5 , 0x00301 , 0x001FB }, + { 0x000E6 , 0x00301 , 0x001FD }, + { 0x000E6 , 0x00304 , 0x001E3 }, + { 0x000E7 , 0x00301 , 0x01E09 }, + { 0x000EA , 0x00300 , 0x01EC1 }, + { 0x000EA , 0x00301 , 0x01EBF }, + { 0x000EA , 0x00303 , 0x01EC5 }, + { 0x000EA , 0x00309 , 0x01EC3 }, + { 0x000EF , 0x00301 , 0x01E2F }, + { 0x000F4 , 0x00300 , 0x01ED3 }, + { 0x000F4 , 0x00301 , 0x01ED1 }, + { 0x000F4 , 0x00303 , 0x01ED7 }, + { 0x000F4 , 0x00309 , 0x01ED5 }, + { 0x000F5 , 0x00301 , 0x01E4D }, + { 0x000F5 , 0x00304 , 0x0022D }, + { 0x000F5 , 0x00308 , 0x01E4F }, + { 0x000F6 , 0x00304 , 0x0022B }, + { 0x000F8 , 0x00301 , 0x001FF }, + { 0x000FC , 0x00300 , 0x001DC }, + { 0x000FC , 0x00301 , 0x001D8 }, + { 0x000FC , 0x00304 , 0x001D6 }, + { 0x000FC , 0x0030C , 0x001DA }, + { 0x00102 , 0x00300 , 0x01EB0 }, + { 0x00102 , 0x00301 , 0x01EAE }, + { 0x00102 , 0x00303 , 0x01EB4 }, + { 0x00102 , 0x00309 , 0x01EB2 }, + { 0x00103 , 0x00300 , 0x01EB1 }, + { 0x00103 , 0x00301 , 0x01EAF }, + { 0x00103 , 0x00303 , 0x01EB5 }, + { 0x00103 , 0x00309 , 0x01EB3 }, + { 0x00112 , 0x00300 , 0x01E14 }, + { 0x00112 , 0x00301 , 0x01E16 }, + { 0x00113 , 0x00300 , 0x01E15 }, + { 0x00113 , 0x00301 , 0x01E17 }, + { 0x0014C , 0x00300 , 0x01E50 }, + { 0x0014C , 0x00301 , 0x01E52 }, + { 0x0014D , 0x00300 , 0x01E51 }, + { 0x0014D , 0x00301 , 0x01E53 }, + { 0x0015A , 0x00307 , 0x01E64 }, + { 0x0015B , 0x00307 , 0x01E65 }, + { 0x00160 , 0x00307 , 0x01E66 }, + { 0x00161 , 0x00307 , 0x01E67 }, + { 0x00168 , 0x00301 , 0x01E78 }, + { 0x00169 , 0x00301 , 0x01E79 }, + { 0x0016A , 0x00308 , 0x01E7A }, + { 0x0016B , 0x00308 , 0x01E7B }, + { 0x0017F , 0x00307 , 0x01E9B }, + { 0x001A0 , 0x00300 , 0x01EDC }, + { 0x001A0 , 0x00301 , 0x01EDA }, + { 0x001A0 , 0x00303 , 0x01EE0 }, + { 0x001A0 , 0x00309 , 0x01EDE }, + { 0x001A0 , 0x00323 , 0x01EE2 }, + { 0x001A1 , 0x00300 , 0x01EDD }, + { 0x001A1 , 0x00301 , 0x01EDB }, + { 0x001A1 , 0x00303 , 0x01EE1 }, + { 0x001A1 , 0x00309 , 0x01EDF }, + { 0x001A1 , 0x00323 , 0x01EE3 }, + { 0x001AF , 0x00300 , 0x01EEA }, + { 0x001AF , 0x00301 , 0x01EE8 }, + { 0x001AF , 0x00303 , 0x01EEE }, + { 0x001AF , 0x00309 , 0x01EEC }, + { 0x001AF , 0x00323 , 0x01EF0 }, + { 0x001B0 , 0x00300 , 0x01EEB }, + { 0x001B0 , 0x00301 , 0x01EE9 }, + { 0x001B0 , 0x00303 , 0x01EEF }, + { 0x001B0 , 0x00309 , 0x01EED }, + { 0x001B0 , 0x00323 , 0x01EF1 }, + { 0x001B7 , 0x0030C , 0x001EE }, + { 0x001EA , 0x00304 , 0x001EC }, + { 0x001EB , 0x00304 , 0x001ED }, + { 0x00226 , 0x00304 , 0x001E0 }, + { 0x00227 , 0x00304 , 0x001E1 }, + { 0x00228 , 0x00306 , 0x01E1C }, + { 0x00229 , 0x00306 , 0x01E1D }, + { 0x0022E , 0x00304 , 0x00230 }, + { 0x0022F , 0x00304 , 0x00231 }, + { 0x00292 , 0x0030C , 0x001EF }, + { 0x00391 , 0x00300 , 0x01FBA }, + { 0x00391 , 0x00301 , 0x00386 }, + { 0x00391 , 0x00304 , 0x01FB9 }, + { 0x00391 , 0x00306 , 0x01FB8 }, + { 0x00391 , 0x00313 , 0x01F08 }, + { 0x00391 , 0x00314 , 0x01F09 }, + { 0x00391 , 0x00345 , 0x01FBC }, + { 0x00395 , 0x00300 , 0x01FC8 }, + { 0x00395 , 0x00301 , 0x00388 }, + { 0x00395 , 0x00313 , 0x01F18 }, + { 0x00395 , 0x00314 , 0x01F19 }, + { 0x00397 , 0x00300 , 0x01FCA }, + { 0x00397 , 0x00301 , 0x00389 }, + { 0x00397 , 0x00313 , 0x01F28 }, + { 0x00397 , 0x00314 , 0x01F29 }, + { 0x00397 , 0x00345 , 0x01FCC }, + { 0x00399 , 0x00300 , 0x01FDA }, + { 0x00399 , 0x00301 , 0x0038A }, + { 0x00399 , 0x00304 , 0x01FD9 }, + { 0x00399 , 0x00306 , 0x01FD8 }, + { 0x00399 , 0x00308 , 0x003AA }, + { 0x00399 , 0x00313 , 0x01F38 }, + { 0x00399 , 0x00314 , 0x01F39 }, + { 0x0039F , 0x00300 , 0x01FF8 }, + { 0x0039F , 0x00301 , 0x0038C }, + { 0x0039F , 0x00313 , 0x01F48 }, + { 0x0039F , 0x00314 , 0x01F49 }, + { 0x003A1 , 0x00314 , 0x01FEC }, + { 0x003A5 , 0x00300 , 0x01FEA }, + { 0x003A5 , 0x00301 , 0x0038E }, + { 0x003A5 , 0x00304 , 0x01FE9 }, + { 0x003A5 , 0x00306 , 0x01FE8 }, + { 0x003A5 , 0x00308 , 0x003AB }, + { 0x003A5 , 0x00314 , 0x01F59 }, + { 0x003A9 , 0x00300 , 0x01FFA }, + { 0x003A9 , 0x00301 , 0x0038F }, + { 0x003A9 , 0x00313 , 0x01F68 }, + { 0x003A9 , 0x00314 , 0x01F69 }, + { 0x003A9 , 0x00345 , 0x01FFC }, + { 0x003AC , 0x00345 , 0x01FB4 }, + { 0x003AE , 0x00345 , 0x01FC4 }, + { 0x003B1 , 0x00300 , 0x01F70 }, + { 0x003B1 , 0x00301 , 0x003AC }, + { 0x003B1 , 0x00304 , 0x01FB1 }, + { 0x003B1 , 0x00306 , 0x01FB0 }, + { 0x003B1 , 0x00313 , 0x01F00 }, + { 0x003B1 , 0x00314 , 0x01F01 }, + { 0x003B1 , 0x00342 , 0x01FB6 }, + { 0x003B1 , 0x00345 , 0x01FB3 }, + { 0x003B5 , 0x00300 , 0x01F72 }, + { 0x003B5 , 0x00301 , 0x003AD }, + { 0x003B5 , 0x00313 , 0x01F10 }, + { 0x003B5 , 0x00314 , 0x01F11 }, + { 0x003B7 , 0x00300 , 0x01F74 }, + { 0x003B7 , 0x00301 , 0x003AE }, + { 0x003B7 , 0x00313 , 0x01F20 }, + { 0x003B7 , 0x00314 , 0x01F21 }, + { 0x003B7 , 0x00342 , 0x01FC6 }, + { 0x003B7 , 0x00345 , 0x01FC3 }, + { 0x003B9 , 0x00300 , 0x01F76 }, + { 0x003B9 , 0x00301 , 0x003AF }, + { 0x003B9 , 0x00304 , 0x01FD1 }, + { 0x003B9 , 0x00306 , 0x01FD0 }, + { 0x003B9 , 0x00308 , 0x003CA }, + { 0x003B9 , 0x00313 , 0x01F30 }, + { 0x003B9 , 0x00314 , 0x01F31 }, + { 0x003B9 , 0x00342 , 0x01FD6 }, + { 0x003BF , 0x00300 , 0x01F78 }, + { 0x003BF , 0x00301 , 0x003CC }, + { 0x003BF , 0x00313 , 0x01F40 }, + { 0x003BF , 0x00314 , 0x01F41 }, + { 0x003C1 , 0x00313 , 0x01FE4 }, + { 0x003C1 , 0x00314 , 0x01FE5 }, + { 0x003C5 , 0x00300 , 0x01F7A }, + { 0x003C5 , 0x00301 , 0x003CD }, + { 0x003C5 , 0x00304 , 0x01FE1 }, + { 0x003C5 , 0x00306 , 0x01FE0 }, + { 0x003C5 , 0x00308 , 0x003CB }, + { 0x003C5 , 0x00313 , 0x01F50 }, + { 0x003C5 , 0x00314 , 0x01F51 }, + { 0x003C5 , 0x00342 , 0x01FE6 }, + { 0x003C9 , 0x00300 , 0x01F7C }, + { 0x003C9 , 0x00301 , 0x003CE }, + { 0x003C9 , 0x00313 , 0x01F60 }, + { 0x003C9 , 0x00314 , 0x01F61 }, + { 0x003C9 , 0x00342 , 0x01FF6 }, + { 0x003C9 , 0x00345 , 0x01FF3 }, + { 0x003CA , 0x00300 , 0x01FD2 }, + { 0x003CA , 0x00301 , 0x00390 }, + { 0x003CA , 0x00342 , 0x01FD7 }, + { 0x003CB , 0x00300 , 0x01FE2 }, + { 0x003CB , 0x00301 , 0x003B0 }, + { 0x003CB , 0x00342 , 0x01FE7 }, + { 0x003CE , 0x00345 , 0x01FF4 }, + { 0x003D2 , 0x00301 , 0x003D3 }, + { 0x003D2 , 0x00308 , 0x003D4 }, + { 0x00406 , 0x00308 , 0x00407 }, + { 0x00410 , 0x00306 , 0x004D0 }, + { 0x00410 , 0x00308 , 0x004D2 }, + { 0x00413 , 0x00301 , 0x00403 }, + { 0x00415 , 0x00300 , 0x00400 }, + { 0x00415 , 0x00306 , 0x004D6 }, + { 0x00415 , 0x00308 , 0x00401 }, + { 0x00416 , 0x00306 , 0x004C1 }, + { 0x00416 , 0x00308 , 0x004DC }, + { 0x00417 , 0x00308 , 0x004DE }, + { 0x00418 , 0x00300 , 0x0040D }, + { 0x00418 , 0x00304 , 0x004E2 }, + { 0x00418 , 0x00306 , 0x00419 }, + { 0x00418 , 0x00308 , 0x004E4 }, + { 0x0041A , 0x00301 , 0x0040C }, + { 0x0041E , 0x00308 , 0x004E6 }, + { 0x00423 , 0x00304 , 0x004EE }, + { 0x00423 , 0x00306 , 0x0040E }, + { 0x00423 , 0x00308 , 0x004F0 }, + { 0x00423 , 0x0030B , 0x004F2 }, + { 0x00427 , 0x00308 , 0x004F4 }, + { 0x0042B , 0x00308 , 0x004F8 }, + { 0x0042D , 0x00308 , 0x004EC }, + { 0x00430 , 0x00306 , 0x004D1 }, + { 0x00430 , 0x00308 , 0x004D3 }, + { 0x00433 , 0x00301 , 0x00453 }, + { 0x00435 , 0x00300 , 0x00450 }, + { 0x00435 , 0x00306 , 0x004D7 }, + { 0x00435 , 0x00308 , 0x00451 }, + { 0x00436 , 0x00306 , 0x004C2 }, + { 0x00436 , 0x00308 , 0x004DD }, + { 0x00437 , 0x00308 , 0x004DF }, + { 0x00438 , 0x00300 , 0x0045D }, + { 0x00438 , 0x00304 , 0x004E3 }, + { 0x00438 , 0x00306 , 0x00439 }, + { 0x00438 , 0x00308 , 0x004E5 }, + { 0x0043A , 0x00301 , 0x0045C }, + { 0x0043E , 0x00308 , 0x004E7 }, + { 0x00443 , 0x00304 , 0x004EF }, + { 0x00443 , 0x00306 , 0x0045E }, + { 0x00443 , 0x00308 , 0x004F1 }, + { 0x00443 , 0x0030B , 0x004F3 }, + { 0x00447 , 0x00308 , 0x004F5 }, + { 0x0044B , 0x00308 , 0x004F9 }, + { 0x0044D , 0x00308 , 0x004ED }, + { 0x00456 , 0x00308 , 0x00457 }, + { 0x00474 , 0x0030F , 0x00476 }, + { 0x00475 , 0x0030F , 0x00477 }, + { 0x004D8 , 0x00308 , 0x004DA }, + { 0x004D9 , 0x00308 , 0x004DB }, + { 0x004E8 , 0x00308 , 0x004EA }, + { 0x004E9 , 0x00308 , 0x004EB }, + { 0x00627 , 0x00653 , 0x00622 }, + { 0x00627 , 0x00654 , 0x00623 }, + { 0x00627 , 0x00655 , 0x00625 }, + { 0x00648 , 0x00654 , 0x00624 }, + { 0x0064A , 0x00654 , 0x00626 }, + { 0x006C1 , 0x00654 , 0x006C2 }, + { 0x006D2 , 0x00654 , 0x006D3 }, + { 0x006D5 , 0x00654 , 0x006C0 }, + { 0x00928 , 0x0093C , 0x00929 }, + { 0x00930 , 0x0093C , 0x00931 }, + { 0x00933 , 0x0093C , 0x00934 }, + { 0x009C7 , 0x009BE , 0x009CB }, + { 0x009C7 , 0x009D7 , 0x009CC }, + { 0x00B47 , 0x00B3E , 0x00B4B }, + { 0x00B47 , 0x00B56 , 0x00B48 }, + { 0x00B47 , 0x00B57 , 0x00B4C }, + { 0x00B92 , 0x00BD7 , 0x00B94 }, + { 0x00BC6 , 0x00BBE , 0x00BCA }, + { 0x00BC6 , 0x00BD7 , 0x00BCC }, + { 0x00BC7 , 0x00BBE , 0x00BCB }, + { 0x00C46 , 0x00C56 , 0x00C48 }, + { 0x00CBF , 0x00CD5 , 0x00CC0 }, + { 0x00CC6 , 0x00CC2 , 0x00CCA }, + { 0x00CC6 , 0x00CD5 , 0x00CC7 }, + { 0x00CC6 , 0x00CD6 , 0x00CC8 }, + { 0x00CCA , 0x00CD5 , 0x00CCB }, + { 0x00D46 , 0x00D3E , 0x00D4A }, + { 0x00D46 , 0x00D57 , 0x00D4C }, + { 0x00D47 , 0x00D3E , 0x00D4B }, + { 0x00DD9 , 0x00DCA , 0x00DDA }, + { 0x00DD9 , 0x00DCF , 0x00DDC }, + { 0x00DD9 , 0x00DDF , 0x00DDE }, + { 0x00DDC , 0x00DCA , 0x00DDD }, + { 0x01025 , 0x0102E , 0x01026 }, + { 0x01B05 , 0x01B35 , 0x01B06 }, + { 0x01B07 , 0x01B35 , 0x01B08 }, + { 0x01B09 , 0x01B35 , 0x01B0A }, + { 0x01B0B , 0x01B35 , 0x01B0C }, + { 0x01B0D , 0x01B35 , 0x01B0E }, + { 0x01B11 , 0x01B35 , 0x01B12 }, + { 0x01B3A , 0x01B35 , 0x01B3B }, + { 0x01B3C , 0x01B35 , 0x01B3D }, + { 0x01B3E , 0x01B35 , 0x01B40 }, + { 0x01B3F , 0x01B35 , 0x01B41 }, + { 0x01B42 , 0x01B35 , 0x01B43 }, + { 0x01E36 , 0x00304 , 0x01E38 }, + { 0x01E37 , 0x00304 , 0x01E39 }, + { 0x01E5A , 0x00304 , 0x01E5C }, + { 0x01E5B , 0x00304 , 0x01E5D }, + { 0x01E62 , 0x00307 , 0x01E68 }, + { 0x01E63 , 0x00307 , 0x01E69 }, + { 0x01EA0 , 0x00302 , 0x01EAC }, + { 0x01EA0 , 0x00306 , 0x01EB6 }, + { 0x01EA1 , 0x00302 , 0x01EAD }, + { 0x01EA1 , 0x00306 , 0x01EB7 }, + { 0x01EB8 , 0x00302 , 0x01EC6 }, + { 0x01EB9 , 0x00302 , 0x01EC7 }, + { 0x01ECC , 0x00302 , 0x01ED8 }, + { 0x01ECD , 0x00302 , 0x01ED9 }, + { 0x01F00 , 0x00300 , 0x01F02 }, + { 0x01F00 , 0x00301 , 0x01F04 }, + { 0x01F00 , 0x00342 , 0x01F06 }, + { 0x01F00 , 0x00345 , 0x01F80 }, + { 0x01F01 , 0x00300 , 0x01F03 }, + { 0x01F01 , 0x00301 , 0x01F05 }, + { 0x01F01 , 0x00342 , 0x01F07 }, + { 0x01F01 , 0x00345 , 0x01F81 }, + { 0x01F02 , 0x00345 , 0x01F82 }, + { 0x01F03 , 0x00345 , 0x01F83 }, + { 0x01F04 , 0x00345 , 0x01F84 }, + { 0x01F05 , 0x00345 , 0x01F85 }, + { 0x01F06 , 0x00345 , 0x01F86 }, + { 0x01F07 , 0x00345 , 0x01F87 }, + { 0x01F08 , 0x00300 , 0x01F0A }, + { 0x01F08 , 0x00301 , 0x01F0C }, + { 0x01F08 , 0x00342 , 0x01F0E }, + { 0x01F08 , 0x00345 , 0x01F88 }, + { 0x01F09 , 0x00300 , 0x01F0B }, + { 0x01F09 , 0x00301 , 0x01F0D }, + { 0x01F09 , 0x00342 , 0x01F0F }, + { 0x01F09 , 0x00345 , 0x01F89 }, + { 0x01F0A , 0x00345 , 0x01F8A }, + { 0x01F0B , 0x00345 , 0x01F8B }, + { 0x01F0C , 0x00345 , 0x01F8C }, + { 0x01F0D , 0x00345 , 0x01F8D }, + { 0x01F0E , 0x00345 , 0x01F8E }, + { 0x01F0F , 0x00345 , 0x01F8F }, + { 0x01F10 , 0x00300 , 0x01F12 }, + { 0x01F10 , 0x00301 , 0x01F14 }, + { 0x01F11 , 0x00300 , 0x01F13 }, + { 0x01F11 , 0x00301 , 0x01F15 }, + { 0x01F18 , 0x00300 , 0x01F1A }, + { 0x01F18 , 0x00301 , 0x01F1C }, + { 0x01F19 , 0x00300 , 0x01F1B }, + { 0x01F19 , 0x00301 , 0x01F1D }, + { 0x01F20 , 0x00300 , 0x01F22 }, + { 0x01F20 , 0x00301 , 0x01F24 }, + { 0x01F20 , 0x00342 , 0x01F26 }, + { 0x01F20 , 0x00345 , 0x01F90 }, + { 0x01F21 , 0x00300 , 0x01F23 }, + { 0x01F21 , 0x00301 , 0x01F25 }, + { 0x01F21 , 0x00342 , 0x01F27 }, + { 0x01F21 , 0x00345 , 0x01F91 }, + { 0x01F22 , 0x00345 , 0x01F92 }, + { 0x01F23 , 0x00345 , 0x01F93 }, + { 0x01F24 , 0x00345 , 0x01F94 }, + { 0x01F25 , 0x00345 , 0x01F95 }, + { 0x01F26 , 0x00345 , 0x01F96 }, + { 0x01F27 , 0x00345 , 0x01F97 }, + { 0x01F28 , 0x00300 , 0x01F2A }, + { 0x01F28 , 0x00301 , 0x01F2C }, + { 0x01F28 , 0x00342 , 0x01F2E }, + { 0x01F28 , 0x00345 , 0x01F98 }, + { 0x01F29 , 0x00300 , 0x01F2B }, + { 0x01F29 , 0x00301 , 0x01F2D }, + { 0x01F29 , 0x00342 , 0x01F2F }, + { 0x01F29 , 0x00345 , 0x01F99 }, + { 0x01F2A , 0x00345 , 0x01F9A }, + { 0x01F2B , 0x00345 , 0x01F9B }, + { 0x01F2C , 0x00345 , 0x01F9C }, + { 0x01F2D , 0x00345 , 0x01F9D }, + { 0x01F2E , 0x00345 , 0x01F9E }, + { 0x01F2F , 0x00345 , 0x01F9F }, + { 0x01F30 , 0x00300 , 0x01F32 }, + { 0x01F30 , 0x00301 , 0x01F34 }, + { 0x01F30 , 0x00342 , 0x01F36 }, + { 0x01F31 , 0x00300 , 0x01F33 }, + { 0x01F31 , 0x00301 , 0x01F35 }, + { 0x01F31 , 0x00342 , 0x01F37 }, + { 0x01F38 , 0x00300 , 0x01F3A }, + { 0x01F38 , 0x00301 , 0x01F3C }, + { 0x01F38 , 0x00342 , 0x01F3E }, + { 0x01F39 , 0x00300 , 0x01F3B }, + { 0x01F39 , 0x00301 , 0x01F3D }, + { 0x01F39 , 0x00342 , 0x01F3F }, + { 0x01F40 , 0x00300 , 0x01F42 }, + { 0x01F40 , 0x00301 , 0x01F44 }, + { 0x01F41 , 0x00300 , 0x01F43 }, + { 0x01F41 , 0x00301 , 0x01F45 }, + { 0x01F48 , 0x00300 , 0x01F4A }, + { 0x01F48 , 0x00301 , 0x01F4C }, + { 0x01F49 , 0x00300 , 0x01F4B }, + { 0x01F49 , 0x00301 , 0x01F4D }, + { 0x01F50 , 0x00300 , 0x01F52 }, + { 0x01F50 , 0x00301 , 0x01F54 }, + { 0x01F50 , 0x00342 , 0x01F56 }, + { 0x01F51 , 0x00300 , 0x01F53 }, + { 0x01F51 , 0x00301 , 0x01F55 }, + { 0x01F51 , 0x00342 , 0x01F57 }, + { 0x01F59 , 0x00300 , 0x01F5B }, + { 0x01F59 , 0x00301 , 0x01F5D }, + { 0x01F59 , 0x00342 , 0x01F5F }, + { 0x01F60 , 0x00300 , 0x01F62 }, + { 0x01F60 , 0x00301 , 0x01F64 }, + { 0x01F60 , 0x00342 , 0x01F66 }, + { 0x01F60 , 0x00345 , 0x01FA0 }, + { 0x01F61 , 0x00300 , 0x01F63 }, + { 0x01F61 , 0x00301 , 0x01F65 }, + { 0x01F61 , 0x00342 , 0x01F67 }, + { 0x01F61 , 0x00345 , 0x01FA1 }, + { 0x01F62 , 0x00345 , 0x01FA2 }, + { 0x01F63 , 0x00345 , 0x01FA3 }, + { 0x01F64 , 0x00345 , 0x01FA4 }, + { 0x01F65 , 0x00345 , 0x01FA5 }, + { 0x01F66 , 0x00345 , 0x01FA6 }, + { 0x01F67 , 0x00345 , 0x01FA7 }, + { 0x01F68 , 0x00300 , 0x01F6A }, + { 0x01F68 , 0x00301 , 0x01F6C }, + { 0x01F68 , 0x00342 , 0x01F6E }, + { 0x01F68 , 0x00345 , 0x01FA8 }, + { 0x01F69 , 0x00300 , 0x01F6B }, + { 0x01F69 , 0x00301 , 0x01F6D }, + { 0x01F69 , 0x00342 , 0x01F6F }, + { 0x01F69 , 0x00345 , 0x01FA9 }, + { 0x01F6A , 0x00345 , 0x01FAA }, + { 0x01F6B , 0x00345 , 0x01FAB }, + { 0x01F6C , 0x00345 , 0x01FAC }, + { 0x01F6D , 0x00345 , 0x01FAD }, + { 0x01F6E , 0x00345 , 0x01FAE }, + { 0x01F6F , 0x00345 , 0x01FAF }, + { 0x01F70 , 0x00345 , 0x01FB2 }, + { 0x01F74 , 0x00345 , 0x01FC2 }, + { 0x01F7C , 0x00345 , 0x01FF2 }, + { 0x01FB6 , 0x00345 , 0x01FB7 }, + { 0x01FBF , 0x00300 , 0x01FCD }, + { 0x01FBF , 0x00301 , 0x01FCE }, + { 0x01FBF , 0x00342 , 0x01FCF }, + { 0x01FC6 , 0x00345 , 0x01FC7 }, + { 0x01FF6 , 0x00345 , 0x01FF7 }, + { 0x01FFE , 0x00300 , 0x01FDD }, + { 0x01FFE , 0x00301 , 0x01FDE }, + { 0x01FFE , 0x00342 , 0x01FDF }, + { 0x02190 , 0x00338 , 0x0219A }, + { 0x02192 , 0x00338 , 0x0219B }, + { 0x02194 , 0x00338 , 0x021AE }, + { 0x021D0 , 0x00338 , 0x021CD }, + { 0x021D2 , 0x00338 , 0x021CF }, + { 0x021D4 , 0x00338 , 0x021CE }, + { 0x02203 , 0x00338 , 0x02204 }, + { 0x02208 , 0x00338 , 0x02209 }, + { 0x0220B , 0x00338 , 0x0220C }, + { 0x02223 , 0x00338 , 0x02224 }, + { 0x02225 , 0x00338 , 0x02226 }, + { 0x0223C , 0x00338 , 0x02241 }, + { 0x02243 , 0x00338 , 0x02244 }, + { 0x02245 , 0x00338 , 0x02247 }, + { 0x02248 , 0x00338 , 0x02249 }, + { 0x0224D , 0x00338 , 0x0226D }, + { 0x02261 , 0x00338 , 0x02262 }, + { 0x02264 , 0x00338 , 0x02270 }, + { 0x02265 , 0x00338 , 0x02271 }, + { 0x02272 , 0x00338 , 0x02274 }, + { 0x02273 , 0x00338 , 0x02275 }, + { 0x02276 , 0x00338 , 0x02278 }, + { 0x02277 , 0x00338 , 0x02279 }, + { 0x0227A , 0x00338 , 0x02280 }, + { 0x0227B , 0x00338 , 0x02281 }, + { 0x0227C , 0x00338 , 0x022E0 }, + { 0x0227D , 0x00338 , 0x022E1 }, + { 0x02282 , 0x00338 , 0x02284 }, + { 0x02283 , 0x00338 , 0x02285 }, + { 0x02286 , 0x00338 , 0x02288 }, + { 0x02287 , 0x00338 , 0x02289 }, + { 0x02291 , 0x00338 , 0x022E2 }, + { 0x02292 , 0x00338 , 0x022E3 }, + { 0x022A2 , 0x00338 , 0x022AC }, + { 0x022A8 , 0x00338 , 0x022AD }, + { 0x022A9 , 0x00338 , 0x022AE }, + { 0x022AB , 0x00338 , 0x022AF }, + { 0x022B2 , 0x00338 , 0x022EA }, + { 0x022B3 , 0x00338 , 0x022EB }, + { 0x022B4 , 0x00338 , 0x022EC }, + { 0x022B5 , 0x00338 , 0x022ED }, + { 0x03046 , 0x03099 , 0x03094 }, + { 0x0304B , 0x03099 , 0x0304C }, + { 0x0304D , 0x03099 , 0x0304E }, + { 0x0304F , 0x03099 , 0x03050 }, + { 0x03051 , 0x03099 , 0x03052 }, + { 0x03053 , 0x03099 , 0x03054 }, + { 0x03055 , 0x03099 , 0x03056 }, + { 0x03057 , 0x03099 , 0x03058 }, + { 0x03059 , 0x03099 , 0x0305A }, + { 0x0305B , 0x03099 , 0x0305C }, + { 0x0305D , 0x03099 , 0x0305E }, + { 0x0305F , 0x03099 , 0x03060 }, + { 0x03061 , 0x03099 , 0x03062 }, + { 0x03064 , 0x03099 , 0x03065 }, + { 0x03066 , 0x03099 , 0x03067 }, + { 0x03068 , 0x03099 , 0x03069 }, + { 0x0306F , 0x03099 , 0x03070 }, + { 0x0306F , 0x0309A , 0x03071 }, + { 0x03072 , 0x03099 , 0x03073 }, + { 0x03072 , 0x0309A , 0x03074 }, + { 0x03075 , 0x03099 , 0x03076 }, + { 0x03075 , 0x0309A , 0x03077 }, + { 0x03078 , 0x03099 , 0x03079 }, + { 0x03078 , 0x0309A , 0x0307A }, + { 0x0307B , 0x03099 , 0x0307C }, + { 0x0307B , 0x0309A , 0x0307D }, + { 0x0309D , 0x03099 , 0x0309E }, + { 0x030A6 , 0x03099 , 0x030F4 }, + { 0x030AB , 0x03099 , 0x030AC }, + { 0x030AD , 0x03099 , 0x030AE }, + { 0x030AF , 0x03099 , 0x030B0 }, + { 0x030B1 , 0x03099 , 0x030B2 }, + { 0x030B3 , 0x03099 , 0x030B4 }, + { 0x030B5 , 0x03099 , 0x030B6 }, + { 0x030B7 , 0x03099 , 0x030B8 }, + { 0x030B9 , 0x03099 , 0x030BA }, + { 0x030BB , 0x03099 , 0x030BC }, + { 0x030BD , 0x03099 , 0x030BE }, + { 0x030BF , 0x03099 , 0x030C0 }, + { 0x030C1 , 0x03099 , 0x030C2 }, + { 0x030C4 , 0x03099 , 0x030C5 }, + { 0x030C6 , 0x03099 , 0x030C7 }, + { 0x030C8 , 0x03099 , 0x030C9 }, + { 0x030CF , 0x03099 , 0x030D0 }, + { 0x030CF , 0x0309A , 0x030D1 }, + { 0x030D2 , 0x03099 , 0x030D3 }, + { 0x030D2 , 0x0309A , 0x030D4 }, + { 0x030D5 , 0x03099 , 0x030D6 }, + { 0x030D5 , 0x0309A , 0x030D7 }, + { 0x030D8 , 0x03099 , 0x030D9 }, + { 0x030D8 , 0x0309A , 0x030DA }, + { 0x030DB , 0x03099 , 0x030DC }, + { 0x030DB , 0x0309A , 0x030DD }, + { 0x030EF , 0x03099 , 0x030F7 }, + { 0x030F0 , 0x03099 , 0x030F8 }, + { 0x030F1 , 0x03099 , 0x030F9 }, + { 0x030F2 , 0x03099 , 0x030FA }, + { 0x030FD , 0x03099 , 0x030FE }, + { 0x11099 , 0x110BA , 0x1109A }, + { 0x1109B , 0x110BA , 0x1109C }, + { 0x110A5 , 0x110BA , 0x110AB }, +}; + +#define CANONICAL_CLASS_MIN 0x0300 +#define CANONICAL_CLASS_MAX 0x1D244 + +#define IS_DECOMPOSABLE_BLOCK(uc) \ + (((uc)>>8) <= 0x1D2 && u_decomposable_blocks[(uc)>>8]) +static const char u_decomposable_blocks[0x1D2+1] = { + 0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,1,0,0,0,1,1,1,1,1,1,1,0,0, + 1,0,0,0,0,0,0,0,0,0,0,0,1,1,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,1,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, + 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,1,0, + 0,1,0,0,0,0,0,0,0,0,1,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1, +}; + +/* Get Canonical Combining Class(CCC). */ +#define CCC(uc) \ + (((uc) > 0x1D244)?0:\ + ccc_val[ccc_val_index[ccc_index[(uc)>>8]][((uc)>>4)&0x0F]][(uc)&0x0F]) + +/* The table of the value of Canonical Combining Class */ +static const unsigned char ccc_val[][16] = { + /* idx=0: XXXX0 - XXXXF */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=1: 00300 - 0030F */ + {230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230 }, + /* idx=2: 00310 - 0031F */ + {230, 230, 230, 230, 230, 232, 220, 220, 220, 220, 232, 216, 220, 220, 220, 220 }, + /* idx=3: 00320 - 0032F */ + {220, 202, 202, 220, 220, 220, 220, 202, 202, 220, 220, 220, 220, 220, 220, 220 }, + /* idx=4: 00330 - 0033F */ + {220, 220, 220, 220, 1, 1, 1, 1, 1, 220, 220, 220, 220, 230, 230, 230 }, + /* idx=5: 00340 - 0034F */ + {230, 230, 230, 230, 230, 240, 230, 220, 220, 220, 230, 230, 230, 220, 220, 0 }, + /* idx=6: 00350 - 0035F */ + {230, 230, 230, 220, 220, 220, 220, 230, 232, 220, 220, 230, 233, 234, 234, 233 }, + /* idx=7: 00360 - 0036F */ + {234, 234, 233, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230 }, + /* idx=8: 00480 - 0048F */ + {0, 0, 0, 230, 230, 230, 230, 230, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=9: 00590 - 0059F */ + {0, 220, 230, 230, 230, 230, 220, 230, 230, 230, 222, 220, 230, 230, 230, 230 }, + /* idx=10: 005A0 - 005AF */ + {230, 230, 220, 220, 220, 220, 220, 220, 230, 230, 220, 230, 230, 222, 228, 230 }, + /* idx=11: 005B0 - 005BF */ + {10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 19, 20, 21, 22, 0, 23 }, + /* idx=12: 005C0 - 005CF */ + {0, 24, 25, 0, 230, 220, 0, 18, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=13: 00610 - 0061F */ + {230, 230, 230, 230, 230, 230, 230, 230, 30, 31, 32, 0, 0, 0, 0, 0 }, + /* idx=14: 00640 - 0064F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 27, 28, 29, 30, 31 }, + /* idx=15: 00650 - 0065F */ + {32, 33, 34, 230, 230, 220, 220, 230, 230, 230, 230, 230, 220, 230, 230, 220 }, + /* idx=16: 00670 - 0067F */ + {35, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=17: 006D0 - 006DF */ + {0, 0, 0, 0, 0, 0, 230, 230, 230, 230, 230, 230, 230, 0, 0, 230 }, + /* idx=18: 006E0 - 006EF */ + {230, 230, 230, 220, 230, 0, 0, 230, 230, 0, 220, 230, 230, 220, 0, 0 }, + /* idx=19: 00710 - 0071F */ + {0, 36, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=20: 00730 - 0073F */ + {230, 220, 230, 230, 220, 230, 230, 220, 220, 220, 230, 220, 220, 230, 220, 230 }, + /* idx=21: 00740 - 0074F */ + {230, 230, 220, 230, 220, 230, 220, 230, 220, 230, 230, 0, 0, 0, 0, 0 }, + /* idx=22: 007E0 - 007EF */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 230, 230, 230, 230, 230 }, + /* idx=23: 007F0 - 007FF */ + {230, 230, 220, 230, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=24: 00810 - 0081F */ + {0, 0, 0, 0, 0, 0, 230, 230, 230, 230, 0, 230, 230, 230, 230, 230 }, + /* idx=25: 00820 - 0082F */ + {230, 230, 230, 230, 0, 230, 230, 230, 0, 230, 230, 230, 230, 230, 0, 0 }, + /* idx=26: 00850 - 0085F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 220, 220, 220, 0, 0, 0, 0 }, + /* idx=27: 00930 - 0093F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7, 0, 0, 0 }, + /* idx=28: 00940 - 0094F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0 }, + /* idx=29: 00950 - 0095F */ + {0, 230, 220, 230, 230, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=30: 009B0 - 009BF */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7, 0, 0, 0 }, + /* idx=31: 009C0 - 009CF */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0 }, + /* idx=32: 00A30 - 00A3F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7, 0, 0, 0 }, + /* idx=33: 00A40 - 00A4F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0 }, + /* idx=34: 00AB0 - 00ABF */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7, 0, 0, 0 }, + /* idx=35: 00AC0 - 00ACF */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0 }, + /* idx=36: 00B30 - 00B3F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7, 0, 0, 0 }, + /* idx=37: 00B40 - 00B4F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0 }, + /* idx=38: 00BC0 - 00BCF */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0 }, + /* idx=39: 00C40 - 00C4F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0 }, + /* idx=40: 00C50 - 00C5F */ + {0, 0, 0, 0, 0, 84, 91, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=41: 00CB0 - 00CBF */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7, 0, 0, 0 }, + /* idx=42: 00CC0 - 00CCF */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0 }, + /* idx=43: 00D40 - 00D4F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0 }, + /* idx=44: 00DC0 - 00DCF */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0, 0, 0, 0 }, + /* idx=45: 00E30 - 00E3F */ + {0, 0, 0, 0, 0, 0, 0, 0, 103, 103, 9, 0, 0, 0, 0, 0 }, + /* idx=46: 00E40 - 00E4F */ + {0, 0, 0, 0, 0, 0, 0, 0, 107, 107, 107, 107, 0, 0, 0, 0 }, + /* idx=47: 00EB0 - 00EBF */ + {0, 0, 0, 0, 0, 0, 0, 0, 118, 118, 0, 0, 0, 0, 0, 0 }, + /* idx=48: 00EC0 - 00ECF */ + {0, 0, 0, 0, 0, 0, 0, 0, 122, 122, 122, 122, 0, 0, 0, 0 }, + /* idx=49: 00F10 - 00F1F */ + {0, 0, 0, 0, 0, 0, 0, 0, 220, 220, 0, 0, 0, 0, 0, 0 }, + /* idx=50: 00F30 - 00F3F */ + {0, 0, 0, 0, 0, 220, 0, 220, 0, 216, 0, 0, 0, 0, 0, 0 }, + /* idx=51: 00F70 - 00F7F */ + {0, 129, 130, 0, 132, 0, 0, 0, 0, 0, 130, 130, 130, 130, 0, 0 }, + /* idx=52: 00F80 - 00F8F */ + {130, 0, 230, 230, 9, 0, 230, 230, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=53: 00FC0 - 00FCF */ + {0, 0, 0, 0, 0, 0, 220, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=54: 01030 - 0103F */ + {0, 0, 0, 0, 0, 0, 0, 7, 0, 9, 9, 0, 0, 0, 0, 0 }, + /* idx=55: 01080 - 0108F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 220, 0, 0 }, + /* idx=56: 01350 - 0135F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 230, 230, 230 }, + /* idx=57: 01710 - 0171F */ + {0, 0, 0, 0, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=58: 01730 - 0173F */ + {0, 0, 0, 0, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=59: 017D0 - 017DF */ + {0, 0, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 230, 0, 0 }, + /* idx=60: 018A0 - 018AF */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 228, 0, 0, 0, 0, 0, 0 }, + /* idx=61: 01930 - 0193F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 222, 230, 220, 0, 0, 0, 0 }, + /* idx=62: 01A10 - 01A1F */ + {0, 0, 0, 0, 0, 0, 0, 230, 220, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=63: 01A60 - 01A6F */ + {9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=64: 01A70 - 01A7F */ + {0, 0, 0, 0, 0, 230, 230, 230, 230, 230, 230, 230, 230, 0, 0, 220 }, + /* idx=65: 01B30 - 01B3F */ + {0, 0, 0, 0, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=66: 01B40 - 01B4F */ + {0, 0, 0, 0, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=67: 01B60 - 01B6F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 230, 220, 230, 230, 230 }, + /* idx=68: 01B70 - 01B7F */ + {230, 230, 230, 230, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=69: 01BA0 - 01BAF */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0, 0, 0, 0 }, + /* idx=70: 01BE0 - 01BEF */ + {0, 0, 0, 0, 0, 0, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=71: 01BF0 - 01BFF */ + {0, 0, 9, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=72: 01C30 - 01C3F */ + {0, 0, 0, 0, 0, 0, 0, 7, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=73: 01CD0 - 01CDF */ + {230, 230, 230, 0, 1, 220, 220, 220, 220, 220, 230, 230, 220, 220, 220, 220 }, + /* idx=74: 01CE0 - 01CEF */ + {230, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 220, 0, 0 }, + /* idx=75: 01DC0 - 01DCF */ + {230, 230, 220, 230, 230, 230, 230, 230, 230, 230, 220, 230, 230, 234, 214, 220 }, + /* idx=76: 01DD0 - 01DDF */ + {202, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230 }, + /* idx=77: 01DE0 - 01DEF */ + {230, 230, 230, 230, 230, 230, 230, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=78: 01DF0 - 01DFF */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 233, 220, 230, 220 }, + /* idx=79: 020D0 - 020DF */ + {230, 230, 1, 1, 230, 230, 230, 230, 1, 1, 1, 230, 230, 0, 0, 0 }, + /* idx=80: 020E0 - 020EF */ + {0, 230, 0, 0, 0, 1, 1, 230, 220, 230, 1, 1, 220, 220, 220, 220 }, + /* idx=81: 020F0 - 020FF */ + {230, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=82: 02CE0 - 02CEF */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 230 }, + /* idx=83: 02CF0 - 02CFF */ + {230, 230, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=84: 02D70 - 02D7F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9 }, + /* idx=85: 02DE0 - 02DEF */ + {230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230 }, + /* idx=86: 02DF0 - 02DFF */ + {230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230 }, + /* idx=87: 03020 - 0302F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 218, 228, 232, 222, 224, 224 }, + /* idx=88: 03090 - 0309F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 8, 8, 0, 0, 0, 0, 0 }, + /* idx=89: 0A660 - 0A66F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 230 }, + /* idx=90: 0A670 - 0A67F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 230, 230, 0, 0 }, + /* idx=91: 0A6F0 - 0A6FF */ + {230, 230, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=92: 0A800 - 0A80F */ + {0, 0, 0, 0, 0, 0, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=93: 0A8C0 - 0A8CF */ + {0, 0, 0, 0, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=94: 0A8E0 - 0A8EF */ + {230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230 }, + /* idx=95: 0A8F0 - 0A8FF */ + {230, 230, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=96: 0A920 - 0A92F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 220, 220, 220, 0, 0 }, + /* idx=97: 0A950 - 0A95F */ + {0, 0, 0, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=98: 0A9B0 - 0A9BF */ + {0, 0, 0, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=99: 0A9C0 - 0A9CF */ + {9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=100: 0AAB0 - 0AABF */ + {230, 0, 230, 230, 220, 0, 0, 230, 230, 0, 0, 0, 0, 0, 230, 230 }, + /* idx=101: 0AAC0 - 0AACF */ + {0, 230, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=102: 0ABE0 - 0ABEF */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0 }, + /* idx=103: 0FB10 - 0FB1F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 26, 0 }, + /* idx=104: 0FE20 - 0FE2F */ + {230, 230, 230, 230, 230, 230, 230, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=105: 101F0 - 101FF */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 220, 0, 0 }, + /* idx=106: 10A00 - 10A0F */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 220, 0, 230 }, + /* idx=107: 10A30 - 10A3F */ + {0, 0, 0, 0, 0, 0, 0, 0, 230, 1, 220, 0, 0, 0, 0, 9 }, + /* idx=108: 11040 - 1104F */ + {0, 0, 0, 0, 0, 0, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=109: 110B0 - 110BF */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 7, 0, 0, 0, 0, 0 }, + /* idx=110: 1D160 - 1D16F */ + {0, 0, 0, 0, 0, 216, 216, 1, 1, 1, 0, 0, 0, 226, 216, 216 }, + /* idx=111: 1D170 - 1D17F */ + {216, 216, 216, 0, 0, 0, 0, 0, 0, 0, 0, 220, 220, 220, 220, 220 }, + /* idx=112: 1D180 - 1D18F */ + {220, 220, 220, 0, 0, 230, 230, 230, 230, 230, 220, 220, 0, 0, 0, 0 }, + /* idx=113: 1D1A0 - 1D1AF */ + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 230, 230, 230, 230, 0, 0 }, + /* idx=114: 1D240 - 1D24F */ + {0, 0, 230, 230, 230, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, +}; + +/* The index table to ccc_val[*][16] */ +static const unsigned char ccc_val_index[][16] = { + /* idx=0: XXX00 - XXXFF */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=1: 00300 - 003FF */ + { 1, 2, 3, 4, 5, 6, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=2: 00400 - 004FF */ + { 0, 0, 0, 0, 0, 0, 0, 0, 8, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=3: 00500 - 005FF */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 9,10,11,12, 0, 0, 0 }, + /* idx=4: 00600 - 006FF */ + { 0,13, 0, 0,14,15, 0,16, 0, 0, 0, 0, 0,17,18, 0 }, + /* idx=5: 00700 - 007FF */ + { 0,19, 0,20,21, 0, 0, 0, 0, 0, 0, 0, 0, 0,22,23 }, + /* idx=6: 00800 - 008FF */ + { 0,24,25, 0, 0,26, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=7: 00900 - 009FF */ + { 0, 0, 0,27,28,29, 0, 0, 0, 0, 0,30,31, 0, 0, 0 }, + /* idx=8: 00A00 - 00AFF */ + { 0, 0, 0,32,33, 0, 0, 0, 0, 0, 0,34,35, 0, 0, 0 }, + /* idx=9: 00B00 - 00BFF */ + { 0, 0, 0,36,37, 0, 0, 0, 0, 0, 0, 0,38, 0, 0, 0 }, + /* idx=10: 00C00 - 00CFF */ + { 0, 0, 0, 0,39,40, 0, 0, 0, 0, 0,41,42, 0, 0, 0 }, + /* idx=11: 00D00 - 00DFF */ + { 0, 0, 0, 0,43, 0, 0, 0, 0, 0, 0, 0,44, 0, 0, 0 }, + /* idx=12: 00E00 - 00EFF */ + { 0, 0, 0,45,46, 0, 0, 0, 0, 0, 0,47,48, 0, 0, 0 }, + /* idx=13: 00F00 - 00FFF */ + { 0,49, 0,50, 0, 0, 0,51,52, 0, 0, 0,53, 0, 0, 0 }, + /* idx=14: 01000 - 010FF */ + { 0, 0, 0,54, 0, 0, 0, 0,55, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=15: 01300 - 013FF */ + { 0, 0, 0, 0, 0,56, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=16: 01700 - 017FF */ + { 0,57, 0,58, 0, 0, 0, 0, 0, 0, 0, 0, 0,59, 0, 0 }, + /* idx=17: 01800 - 018FF */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,60, 0, 0, 0, 0, 0 }, + /* idx=18: 01900 - 019FF */ + { 0, 0, 0,61, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=19: 01A00 - 01AFF */ + { 0,62, 0, 0, 0, 0,63,64, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=20: 01B00 - 01BFF */ + { 0, 0, 0,65,66, 0,67,68, 0, 0,69, 0, 0, 0,70,71 }, + /* idx=21: 01C00 - 01CFF */ + { 0, 0, 0,72, 0, 0, 0, 0, 0, 0, 0, 0, 0,73,74, 0 }, + /* idx=22: 01D00 - 01DFF */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,75,76,77,78 }, + /* idx=23: 02000 - 020FF */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,79,80,81 }, + /* idx=24: 02C00 - 02CFF */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,82,83 }, + /* idx=25: 02D00 - 02DFF */ + { 0, 0, 0, 0, 0, 0, 0,84, 0, 0, 0, 0, 0, 0,85,86 }, + /* idx=26: 03000 - 030FF */ + { 0, 0,87, 0, 0, 0, 0, 0, 0,88, 0, 0, 0, 0, 0, 0 }, + /* idx=27: 0A600 - 0A6FF */ + { 0, 0, 0, 0, 0, 0,89,90, 0, 0, 0, 0, 0, 0, 0,91 }, + /* idx=28: 0A800 - 0A8FF */ + {92, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,93, 0,94,95 }, + /* idx=29: 0A900 - 0A9FF */ + { 0, 0,96, 0, 0,97, 0, 0, 0, 0, 0,98,99, 0, 0, 0 }, + /* idx=30: 0AA00 - 0AAFF */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,100,101, 0, 0, 0 }, + /* idx=31: 0AB00 - 0ABFF */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,102, 0 }, + /* idx=32: 0FB00 - 0FBFF */ + { 0,103, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=33: 0FE00 - 0FEFF */ + { 0, 0,104, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=34: 10100 - 101FF */ + { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,105 }, + /* idx=35: 10A00 - 10AFF */ + {106, 0, 0,107, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, + /* idx=36: 11000 - 110FF */ + { 0, 0, 0, 0,108, 0, 0, 0, 0, 0, 0,109, 0, 0, 0, 0 }, + /* idx=37: 1D100 - 1D1FF */ + { 0, 0, 0, 0, 0, 0,110,111,112, 0,113, 0, 0, 0, 0, 0 }, + /* idx=38: 1D200 - 1D2FF */ + { 0, 0, 0, 0,114, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, +}; + +/* The index table to ccc_val_index[*][16] */ +static const unsigned char ccc_index[] = { + 0, 0, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9,10,11,12,13,14, 0, 0,15, 0, 0, 0,16, + 17,18,19,20,21,22, 0, 0,23, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,24,25, 0, 0, + 26, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,27, 0, + 28,29,30,31, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,32, 0, 0,33, 0, 0,34, 0, 0, 0, 0, 0, 0, + 0, 0,35, 0, 0, 0, 0, 0,36, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0,37,38,}; + +struct unicode_decomposition_table { + uint32_t nfc; + uint32_t cp1; + uint32_t cp2; +}; + +static const struct unicode_decomposition_table u_decomposition_table[] = { + { 0x000C0 , 0x00041 , 0x00300 }, + { 0x000C1 , 0x00041 , 0x00301 }, + { 0x000C2 , 0x00041 , 0x00302 }, + { 0x000C3 , 0x00041 , 0x00303 }, + { 0x000C4 , 0x00041 , 0x00308 }, + { 0x000C5 , 0x00041 , 0x0030A }, + { 0x000C7 , 0x00043 , 0x00327 }, + { 0x000C8 , 0x00045 , 0x00300 }, + { 0x000C9 , 0x00045 , 0x00301 }, + { 0x000CA , 0x00045 , 0x00302 }, + { 0x000CB , 0x00045 , 0x00308 }, + { 0x000CC , 0x00049 , 0x00300 }, + { 0x000CD , 0x00049 , 0x00301 }, + { 0x000CE , 0x00049 , 0x00302 }, + { 0x000CF , 0x00049 , 0x00308 }, + { 0x000D1 , 0x0004E , 0x00303 }, + { 0x000D2 , 0x0004F , 0x00300 }, + { 0x000D3 , 0x0004F , 0x00301 }, + { 0x000D4 , 0x0004F , 0x00302 }, + { 0x000D5 , 0x0004F , 0x00303 }, + { 0x000D6 , 0x0004F , 0x00308 }, + { 0x000D9 , 0x00055 , 0x00300 }, + { 0x000DA , 0x00055 , 0x00301 }, + { 0x000DB , 0x00055 , 0x00302 }, + { 0x000DC , 0x00055 , 0x00308 }, + { 0x000DD , 0x00059 , 0x00301 }, + { 0x000E0 , 0x00061 , 0x00300 }, + { 0x000E1 , 0x00061 , 0x00301 }, + { 0x000E2 , 0x00061 , 0x00302 }, + { 0x000E3 , 0x00061 , 0x00303 }, + { 0x000E4 , 0x00061 , 0x00308 }, + { 0x000E5 , 0x00061 , 0x0030A }, + { 0x000E7 , 0x00063 , 0x00327 }, + { 0x000E8 , 0x00065 , 0x00300 }, + { 0x000E9 , 0x00065 , 0x00301 }, + { 0x000EA , 0x00065 , 0x00302 }, + { 0x000EB , 0x00065 , 0x00308 }, + { 0x000EC , 0x00069 , 0x00300 }, + { 0x000ED , 0x00069 , 0x00301 }, + { 0x000EE , 0x00069 , 0x00302 }, + { 0x000EF , 0x00069 , 0x00308 }, + { 0x000F1 , 0x0006E , 0x00303 }, + { 0x000F2 , 0x0006F , 0x00300 }, + { 0x000F3 , 0x0006F , 0x00301 }, + { 0x000F4 , 0x0006F , 0x00302 }, + { 0x000F5 , 0x0006F , 0x00303 }, + { 0x000F6 , 0x0006F , 0x00308 }, + { 0x000F9 , 0x00075 , 0x00300 }, + { 0x000FA , 0x00075 , 0x00301 }, + { 0x000FB , 0x00075 , 0x00302 }, + { 0x000FC , 0x00075 , 0x00308 }, + { 0x000FD , 0x00079 , 0x00301 }, + { 0x000FF , 0x00079 , 0x00308 }, + { 0x00100 , 0x00041 , 0x00304 }, + { 0x00101 , 0x00061 , 0x00304 }, + { 0x00102 , 0x00041 , 0x00306 }, + { 0x00103 , 0x00061 , 0x00306 }, + { 0x00104 , 0x00041 , 0x00328 }, + { 0x00105 , 0x00061 , 0x00328 }, + { 0x00106 , 0x00043 , 0x00301 }, + { 0x00107 , 0x00063 , 0x00301 }, + { 0x00108 , 0x00043 , 0x00302 }, + { 0x00109 , 0x00063 , 0x00302 }, + { 0x0010A , 0x00043 , 0x00307 }, + { 0x0010B , 0x00063 , 0x00307 }, + { 0x0010C , 0x00043 , 0x0030C }, + { 0x0010D , 0x00063 , 0x0030C }, + { 0x0010E , 0x00044 , 0x0030C }, + { 0x0010F , 0x00064 , 0x0030C }, + { 0x00112 , 0x00045 , 0x00304 }, + { 0x00113 , 0x00065 , 0x00304 }, + { 0x00114 , 0x00045 , 0x00306 }, + { 0x00115 , 0x00065 , 0x00306 }, + { 0x00116 , 0x00045 , 0x00307 }, + { 0x00117 , 0x00065 , 0x00307 }, + { 0x00118 , 0x00045 , 0x00328 }, + { 0x00119 , 0x00065 , 0x00328 }, + { 0x0011A , 0x00045 , 0x0030C }, + { 0x0011B , 0x00065 , 0x0030C }, + { 0x0011C , 0x00047 , 0x00302 }, + { 0x0011D , 0x00067 , 0x00302 }, + { 0x0011E , 0x00047 , 0x00306 }, + { 0x0011F , 0x00067 , 0x00306 }, + { 0x00120 , 0x00047 , 0x00307 }, + { 0x00121 , 0x00067 , 0x00307 }, + { 0x00122 , 0x00047 , 0x00327 }, + { 0x00123 , 0x00067 , 0x00327 }, + { 0x00124 , 0x00048 , 0x00302 }, + { 0x00125 , 0x00068 , 0x00302 }, + { 0x00128 , 0x00049 , 0x00303 }, + { 0x00129 , 0x00069 , 0x00303 }, + { 0x0012A , 0x00049 , 0x00304 }, + { 0x0012B , 0x00069 , 0x00304 }, + { 0x0012C , 0x00049 , 0x00306 }, + { 0x0012D , 0x00069 , 0x00306 }, + { 0x0012E , 0x00049 , 0x00328 }, + { 0x0012F , 0x00069 , 0x00328 }, + { 0x00130 , 0x00049 , 0x00307 }, + { 0x00134 , 0x0004A , 0x00302 }, + { 0x00135 , 0x0006A , 0x00302 }, + { 0x00136 , 0x0004B , 0x00327 }, + { 0x00137 , 0x0006B , 0x00327 }, + { 0x00139 , 0x0004C , 0x00301 }, + { 0x0013A , 0x0006C , 0x00301 }, + { 0x0013B , 0x0004C , 0x00327 }, + { 0x0013C , 0x0006C , 0x00327 }, + { 0x0013D , 0x0004C , 0x0030C }, + { 0x0013E , 0x0006C , 0x0030C }, + { 0x00143 , 0x0004E , 0x00301 }, + { 0x00144 , 0x0006E , 0x00301 }, + { 0x00145 , 0x0004E , 0x00327 }, + { 0x00146 , 0x0006E , 0x00327 }, + { 0x00147 , 0x0004E , 0x0030C }, + { 0x00148 , 0x0006E , 0x0030C }, + { 0x0014C , 0x0004F , 0x00304 }, + { 0x0014D , 0x0006F , 0x00304 }, + { 0x0014E , 0x0004F , 0x00306 }, + { 0x0014F , 0x0006F , 0x00306 }, + { 0x00150 , 0x0004F , 0x0030B }, + { 0x00151 , 0x0006F , 0x0030B }, + { 0x00154 , 0x00052 , 0x00301 }, + { 0x00155 , 0x00072 , 0x00301 }, + { 0x00156 , 0x00052 , 0x00327 }, + { 0x00157 , 0x00072 , 0x00327 }, + { 0x00158 , 0x00052 , 0x0030C }, + { 0x00159 , 0x00072 , 0x0030C }, + { 0x0015A , 0x00053 , 0x00301 }, + { 0x0015B , 0x00073 , 0x00301 }, + { 0x0015C , 0x00053 , 0x00302 }, + { 0x0015D , 0x00073 , 0x00302 }, + { 0x0015E , 0x00053 , 0x00327 }, + { 0x0015F , 0x00073 , 0x00327 }, + { 0x00160 , 0x00053 , 0x0030C }, + { 0x00161 , 0x00073 , 0x0030C }, + { 0x00162 , 0x00054 , 0x00327 }, + { 0x00163 , 0x00074 , 0x00327 }, + { 0x00164 , 0x00054 , 0x0030C }, + { 0x00165 , 0x00074 , 0x0030C }, + { 0x00168 , 0x00055 , 0x00303 }, + { 0x00169 , 0x00075 , 0x00303 }, + { 0x0016A , 0x00055 , 0x00304 }, + { 0x0016B , 0x00075 , 0x00304 }, + { 0x0016C , 0x00055 , 0x00306 }, + { 0x0016D , 0x00075 , 0x00306 }, + { 0x0016E , 0x00055 , 0x0030A }, + { 0x0016F , 0x00075 , 0x0030A }, + { 0x00170 , 0x00055 , 0x0030B }, + { 0x00171 , 0x00075 , 0x0030B }, + { 0x00172 , 0x00055 , 0x00328 }, + { 0x00173 , 0x00075 , 0x00328 }, + { 0x00174 , 0x00057 , 0x00302 }, + { 0x00175 , 0x00077 , 0x00302 }, + { 0x00176 , 0x00059 , 0x00302 }, + { 0x00177 , 0x00079 , 0x00302 }, + { 0x00178 , 0x00059 , 0x00308 }, + { 0x00179 , 0x0005A , 0x00301 }, + { 0x0017A , 0x0007A , 0x00301 }, + { 0x0017B , 0x0005A , 0x00307 }, + { 0x0017C , 0x0007A , 0x00307 }, + { 0x0017D , 0x0005A , 0x0030C }, + { 0x0017E , 0x0007A , 0x0030C }, + { 0x001A0 , 0x0004F , 0x0031B }, + { 0x001A1 , 0x0006F , 0x0031B }, + { 0x001AF , 0x00055 , 0x0031B }, + { 0x001B0 , 0x00075 , 0x0031B }, + { 0x001CD , 0x00041 , 0x0030C }, + { 0x001CE , 0x00061 , 0x0030C }, + { 0x001CF , 0x00049 , 0x0030C }, + { 0x001D0 , 0x00069 , 0x0030C }, + { 0x001D1 , 0x0004F , 0x0030C }, + { 0x001D2 , 0x0006F , 0x0030C }, + { 0x001D3 , 0x00055 , 0x0030C }, + { 0x001D4 , 0x00075 , 0x0030C }, + { 0x001D5 , 0x000DC , 0x00304 }, + { 0x001D6 , 0x000FC , 0x00304 }, + { 0x001D7 , 0x000DC , 0x00301 }, + { 0x001D8 , 0x000FC , 0x00301 }, + { 0x001D9 , 0x000DC , 0x0030C }, + { 0x001DA , 0x000FC , 0x0030C }, + { 0x001DB , 0x000DC , 0x00300 }, + { 0x001DC , 0x000FC , 0x00300 }, + { 0x001DE , 0x000C4 , 0x00304 }, + { 0x001DF , 0x000E4 , 0x00304 }, + { 0x001E0 , 0x00226 , 0x00304 }, + { 0x001E1 , 0x00227 , 0x00304 }, + { 0x001E2 , 0x000C6 , 0x00304 }, + { 0x001E3 , 0x000E6 , 0x00304 }, + { 0x001E6 , 0x00047 , 0x0030C }, + { 0x001E7 , 0x00067 , 0x0030C }, + { 0x001E8 , 0x0004B , 0x0030C }, + { 0x001E9 , 0x0006B , 0x0030C }, + { 0x001EA , 0x0004F , 0x00328 }, + { 0x001EB , 0x0006F , 0x00328 }, + { 0x001EC , 0x001EA , 0x00304 }, + { 0x001ED , 0x001EB , 0x00304 }, + { 0x001EE , 0x001B7 , 0x0030C }, + { 0x001EF , 0x00292 , 0x0030C }, + { 0x001F0 , 0x0006A , 0x0030C }, + { 0x001F4 , 0x00047 , 0x00301 }, + { 0x001F5 , 0x00067 , 0x00301 }, + { 0x001F8 , 0x0004E , 0x00300 }, + { 0x001F9 , 0x0006E , 0x00300 }, + { 0x001FA , 0x000C5 , 0x00301 }, + { 0x001FB , 0x000E5 , 0x00301 }, + { 0x001FC , 0x000C6 , 0x00301 }, + { 0x001FD , 0x000E6 , 0x00301 }, + { 0x001FE , 0x000D8 , 0x00301 }, + { 0x001FF , 0x000F8 , 0x00301 }, + { 0x00200 , 0x00041 , 0x0030F }, + { 0x00201 , 0x00061 , 0x0030F }, + { 0x00202 , 0x00041 , 0x00311 }, + { 0x00203 , 0x00061 , 0x00311 }, + { 0x00204 , 0x00045 , 0x0030F }, + { 0x00205 , 0x00065 , 0x0030F }, + { 0x00206 , 0x00045 , 0x00311 }, + { 0x00207 , 0x00065 , 0x00311 }, + { 0x00208 , 0x00049 , 0x0030F }, + { 0x00209 , 0x00069 , 0x0030F }, + { 0x0020A , 0x00049 , 0x00311 }, + { 0x0020B , 0x00069 , 0x00311 }, + { 0x0020C , 0x0004F , 0x0030F }, + { 0x0020D , 0x0006F , 0x0030F }, + { 0x0020E , 0x0004F , 0x00311 }, + { 0x0020F , 0x0006F , 0x00311 }, + { 0x00210 , 0x00052 , 0x0030F }, + { 0x00211 , 0x00072 , 0x0030F }, + { 0x00212 , 0x00052 , 0x00311 }, + { 0x00213 , 0x00072 , 0x00311 }, + { 0x00214 , 0x00055 , 0x0030F }, + { 0x00215 , 0x00075 , 0x0030F }, + { 0x00216 , 0x00055 , 0x00311 }, + { 0x00217 , 0x00075 , 0x00311 }, + { 0x00218 , 0x00053 , 0x00326 }, + { 0x00219 , 0x00073 , 0x00326 }, + { 0x0021A , 0x00054 , 0x00326 }, + { 0x0021B , 0x00074 , 0x00326 }, + { 0x0021E , 0x00048 , 0x0030C }, + { 0x0021F , 0x00068 , 0x0030C }, + { 0x00226 , 0x00041 , 0x00307 }, + { 0x00227 , 0x00061 , 0x00307 }, + { 0x00228 , 0x00045 , 0x00327 }, + { 0x00229 , 0x00065 , 0x00327 }, + { 0x0022A , 0x000D6 , 0x00304 }, + { 0x0022B , 0x000F6 , 0x00304 }, + { 0x0022C , 0x000D5 , 0x00304 }, + { 0x0022D , 0x000F5 , 0x00304 }, + { 0x0022E , 0x0004F , 0x00307 }, + { 0x0022F , 0x0006F , 0x00307 }, + { 0x00230 , 0x0022E , 0x00304 }, + { 0x00231 , 0x0022F , 0x00304 }, + { 0x00232 , 0x00059 , 0x00304 }, + { 0x00233 , 0x00079 , 0x00304 }, + { 0x00385 , 0x000A8 , 0x00301 }, + { 0x00386 , 0x00391 , 0x00301 }, + { 0x00388 , 0x00395 , 0x00301 }, + { 0x00389 , 0x00397 , 0x00301 }, + { 0x0038A , 0x00399 , 0x00301 }, + { 0x0038C , 0x0039F , 0x00301 }, + { 0x0038E , 0x003A5 , 0x00301 }, + { 0x0038F , 0x003A9 , 0x00301 }, + { 0x00390 , 0x003CA , 0x00301 }, + { 0x003AA , 0x00399 , 0x00308 }, + { 0x003AB , 0x003A5 , 0x00308 }, + { 0x003AC , 0x003B1 , 0x00301 }, + { 0x003AD , 0x003B5 , 0x00301 }, + { 0x003AE , 0x003B7 , 0x00301 }, + { 0x003AF , 0x003B9 , 0x00301 }, + { 0x003B0 , 0x003CB , 0x00301 }, + { 0x003CA , 0x003B9 , 0x00308 }, + { 0x003CB , 0x003C5 , 0x00308 }, + { 0x003CC , 0x003BF , 0x00301 }, + { 0x003CD , 0x003C5 , 0x00301 }, + { 0x003CE , 0x003C9 , 0x00301 }, + { 0x003D3 , 0x003D2 , 0x00301 }, + { 0x003D4 , 0x003D2 , 0x00308 }, + { 0x00400 , 0x00415 , 0x00300 }, + { 0x00401 , 0x00415 , 0x00308 }, + { 0x00403 , 0x00413 , 0x00301 }, + { 0x00407 , 0x00406 , 0x00308 }, + { 0x0040C , 0x0041A , 0x00301 }, + { 0x0040D , 0x00418 , 0x00300 }, + { 0x0040E , 0x00423 , 0x00306 }, + { 0x00419 , 0x00418 , 0x00306 }, + { 0x00439 , 0x00438 , 0x00306 }, + { 0x00450 , 0x00435 , 0x00300 }, + { 0x00451 , 0x00435 , 0x00308 }, + { 0x00453 , 0x00433 , 0x00301 }, + { 0x00457 , 0x00456 , 0x00308 }, + { 0x0045C , 0x0043A , 0x00301 }, + { 0x0045D , 0x00438 , 0x00300 }, + { 0x0045E , 0x00443 , 0x00306 }, + { 0x00476 , 0x00474 , 0x0030F }, + { 0x00477 , 0x00475 , 0x0030F }, + { 0x004C1 , 0x00416 , 0x00306 }, + { 0x004C2 , 0x00436 , 0x00306 }, + { 0x004D0 , 0x00410 , 0x00306 }, + { 0x004D1 , 0x00430 , 0x00306 }, + { 0x004D2 , 0x00410 , 0x00308 }, + { 0x004D3 , 0x00430 , 0x00308 }, + { 0x004D6 , 0x00415 , 0x00306 }, + { 0x004D7 , 0x00435 , 0x00306 }, + { 0x004DA , 0x004D8 , 0x00308 }, + { 0x004DB , 0x004D9 , 0x00308 }, + { 0x004DC , 0x00416 , 0x00308 }, + { 0x004DD , 0x00436 , 0x00308 }, + { 0x004DE , 0x00417 , 0x00308 }, + { 0x004DF , 0x00437 , 0x00308 }, + { 0x004E2 , 0x00418 , 0x00304 }, + { 0x004E3 , 0x00438 , 0x00304 }, + { 0x004E4 , 0x00418 , 0x00308 }, + { 0x004E5 , 0x00438 , 0x00308 }, + { 0x004E6 , 0x0041E , 0x00308 }, + { 0x004E7 , 0x0043E , 0x00308 }, + { 0x004EA , 0x004E8 , 0x00308 }, + { 0x004EB , 0x004E9 , 0x00308 }, + { 0x004EC , 0x0042D , 0x00308 }, + { 0x004ED , 0x0044D , 0x00308 }, + { 0x004EE , 0x00423 , 0x00304 }, + { 0x004EF , 0x00443 , 0x00304 }, + { 0x004F0 , 0x00423 , 0x00308 }, + { 0x004F1 , 0x00443 , 0x00308 }, + { 0x004F2 , 0x00423 , 0x0030B }, + { 0x004F3 , 0x00443 , 0x0030B }, + { 0x004F4 , 0x00427 , 0x00308 }, + { 0x004F5 , 0x00447 , 0x00308 }, + { 0x004F8 , 0x0042B , 0x00308 }, + { 0x004F9 , 0x0044B , 0x00308 }, + { 0x00622 , 0x00627 , 0x00653 }, + { 0x00623 , 0x00627 , 0x00654 }, + { 0x00624 , 0x00648 , 0x00654 }, + { 0x00625 , 0x00627 , 0x00655 }, + { 0x00626 , 0x0064A , 0x00654 }, + { 0x006C0 , 0x006D5 , 0x00654 }, + { 0x006C2 , 0x006C1 , 0x00654 }, + { 0x006D3 , 0x006D2 , 0x00654 }, + { 0x00929 , 0x00928 , 0x0093C }, + { 0x00931 , 0x00930 , 0x0093C }, + { 0x00934 , 0x00933 , 0x0093C }, + { 0x009CB , 0x009C7 , 0x009BE }, + { 0x009CC , 0x009C7 , 0x009D7 }, + { 0x00B48 , 0x00B47 , 0x00B56 }, + { 0x00B4B , 0x00B47 , 0x00B3E }, + { 0x00B4C , 0x00B47 , 0x00B57 }, + { 0x00B94 , 0x00B92 , 0x00BD7 }, + { 0x00BCA , 0x00BC6 , 0x00BBE }, + { 0x00BCB , 0x00BC7 , 0x00BBE }, + { 0x00BCC , 0x00BC6 , 0x00BD7 }, + { 0x00C48 , 0x00C46 , 0x00C56 }, + { 0x00CC0 , 0x00CBF , 0x00CD5 }, + { 0x00CC7 , 0x00CC6 , 0x00CD5 }, + { 0x00CC8 , 0x00CC6 , 0x00CD6 }, + { 0x00CCA , 0x00CC6 , 0x00CC2 }, + { 0x00CCB , 0x00CCA , 0x00CD5 }, + { 0x00D4A , 0x00D46 , 0x00D3E }, + { 0x00D4B , 0x00D47 , 0x00D3E }, + { 0x00D4C , 0x00D46 , 0x00D57 }, + { 0x00DDA , 0x00DD9 , 0x00DCA }, + { 0x00DDC , 0x00DD9 , 0x00DCF }, + { 0x00DDD , 0x00DDC , 0x00DCA }, + { 0x00DDE , 0x00DD9 , 0x00DDF }, + { 0x01026 , 0x01025 , 0x0102E }, + { 0x01B06 , 0x01B05 , 0x01B35 }, + { 0x01B08 , 0x01B07 , 0x01B35 }, + { 0x01B0A , 0x01B09 , 0x01B35 }, + { 0x01B0C , 0x01B0B , 0x01B35 }, + { 0x01B0E , 0x01B0D , 0x01B35 }, + { 0x01B12 , 0x01B11 , 0x01B35 }, + { 0x01B3B , 0x01B3A , 0x01B35 }, + { 0x01B3D , 0x01B3C , 0x01B35 }, + { 0x01B40 , 0x01B3E , 0x01B35 }, + { 0x01B41 , 0x01B3F , 0x01B35 }, + { 0x01B43 , 0x01B42 , 0x01B35 }, + { 0x01E00 , 0x00041 , 0x00325 }, + { 0x01E01 , 0x00061 , 0x00325 }, + { 0x01E02 , 0x00042 , 0x00307 }, + { 0x01E03 , 0x00062 , 0x00307 }, + { 0x01E04 , 0x00042 , 0x00323 }, + { 0x01E05 , 0x00062 , 0x00323 }, + { 0x01E06 , 0x00042 , 0x00331 }, + { 0x01E07 , 0x00062 , 0x00331 }, + { 0x01E08 , 0x000C7 , 0x00301 }, + { 0x01E09 , 0x000E7 , 0x00301 }, + { 0x01E0A , 0x00044 , 0x00307 }, + { 0x01E0B , 0x00064 , 0x00307 }, + { 0x01E0C , 0x00044 , 0x00323 }, + { 0x01E0D , 0x00064 , 0x00323 }, + { 0x01E0E , 0x00044 , 0x00331 }, + { 0x01E0F , 0x00064 , 0x00331 }, + { 0x01E10 , 0x00044 , 0x00327 }, + { 0x01E11 , 0x00064 , 0x00327 }, + { 0x01E12 , 0x00044 , 0x0032D }, + { 0x01E13 , 0x00064 , 0x0032D }, + { 0x01E14 , 0x00112 , 0x00300 }, + { 0x01E15 , 0x00113 , 0x00300 }, + { 0x01E16 , 0x00112 , 0x00301 }, + { 0x01E17 , 0x00113 , 0x00301 }, + { 0x01E18 , 0x00045 , 0x0032D }, + { 0x01E19 , 0x00065 , 0x0032D }, + { 0x01E1A , 0x00045 , 0x00330 }, + { 0x01E1B , 0x00065 , 0x00330 }, + { 0x01E1C , 0x00228 , 0x00306 }, + { 0x01E1D , 0x00229 , 0x00306 }, + { 0x01E1E , 0x00046 , 0x00307 }, + { 0x01E1F , 0x00066 , 0x00307 }, + { 0x01E20 , 0x00047 , 0x00304 }, + { 0x01E21 , 0x00067 , 0x00304 }, + { 0x01E22 , 0x00048 , 0x00307 }, + { 0x01E23 , 0x00068 , 0x00307 }, + { 0x01E24 , 0x00048 , 0x00323 }, + { 0x01E25 , 0x00068 , 0x00323 }, + { 0x01E26 , 0x00048 , 0x00308 }, + { 0x01E27 , 0x00068 , 0x00308 }, + { 0x01E28 , 0x00048 , 0x00327 }, + { 0x01E29 , 0x00068 , 0x00327 }, + { 0x01E2A , 0x00048 , 0x0032E }, + { 0x01E2B , 0x00068 , 0x0032E }, + { 0x01E2C , 0x00049 , 0x00330 }, + { 0x01E2D , 0x00069 , 0x00330 }, + { 0x01E2E , 0x000CF , 0x00301 }, + { 0x01E2F , 0x000EF , 0x00301 }, + { 0x01E30 , 0x0004B , 0x00301 }, + { 0x01E31 , 0x0006B , 0x00301 }, + { 0x01E32 , 0x0004B , 0x00323 }, + { 0x01E33 , 0x0006B , 0x00323 }, + { 0x01E34 , 0x0004B , 0x00331 }, + { 0x01E35 , 0x0006B , 0x00331 }, + { 0x01E36 , 0x0004C , 0x00323 }, + { 0x01E37 , 0x0006C , 0x00323 }, + { 0x01E38 , 0x01E36 , 0x00304 }, + { 0x01E39 , 0x01E37 , 0x00304 }, + { 0x01E3A , 0x0004C , 0x00331 }, + { 0x01E3B , 0x0006C , 0x00331 }, + { 0x01E3C , 0x0004C , 0x0032D }, + { 0x01E3D , 0x0006C , 0x0032D }, + { 0x01E3E , 0x0004D , 0x00301 }, + { 0x01E3F , 0x0006D , 0x00301 }, + { 0x01E40 , 0x0004D , 0x00307 }, + { 0x01E41 , 0x0006D , 0x00307 }, + { 0x01E42 , 0x0004D , 0x00323 }, + { 0x01E43 , 0x0006D , 0x00323 }, + { 0x01E44 , 0x0004E , 0x00307 }, + { 0x01E45 , 0x0006E , 0x00307 }, + { 0x01E46 , 0x0004E , 0x00323 }, + { 0x01E47 , 0x0006E , 0x00323 }, + { 0x01E48 , 0x0004E , 0x00331 }, + { 0x01E49 , 0x0006E , 0x00331 }, + { 0x01E4A , 0x0004E , 0x0032D }, + { 0x01E4B , 0x0006E , 0x0032D }, + { 0x01E4C , 0x000D5 , 0x00301 }, + { 0x01E4D , 0x000F5 , 0x00301 }, + { 0x01E4E , 0x000D5 , 0x00308 }, + { 0x01E4F , 0x000F5 , 0x00308 }, + { 0x01E50 , 0x0014C , 0x00300 }, + { 0x01E51 , 0x0014D , 0x00300 }, + { 0x01E52 , 0x0014C , 0x00301 }, + { 0x01E53 , 0x0014D , 0x00301 }, + { 0x01E54 , 0x00050 , 0x00301 }, + { 0x01E55 , 0x00070 , 0x00301 }, + { 0x01E56 , 0x00050 , 0x00307 }, + { 0x01E57 , 0x00070 , 0x00307 }, + { 0x01E58 , 0x00052 , 0x00307 }, + { 0x01E59 , 0x00072 , 0x00307 }, + { 0x01E5A , 0x00052 , 0x00323 }, + { 0x01E5B , 0x00072 , 0x00323 }, + { 0x01E5C , 0x01E5A , 0x00304 }, + { 0x01E5D , 0x01E5B , 0x00304 }, + { 0x01E5E , 0x00052 , 0x00331 }, + { 0x01E5F , 0x00072 , 0x00331 }, + { 0x01E60 , 0x00053 , 0x00307 }, + { 0x01E61 , 0x00073 , 0x00307 }, + { 0x01E62 , 0x00053 , 0x00323 }, + { 0x01E63 , 0x00073 , 0x00323 }, + { 0x01E64 , 0x0015A , 0x00307 }, + { 0x01E65 , 0x0015B , 0x00307 }, + { 0x01E66 , 0x00160 , 0x00307 }, + { 0x01E67 , 0x00161 , 0x00307 }, + { 0x01E68 , 0x01E62 , 0x00307 }, + { 0x01E69 , 0x01E63 , 0x00307 }, + { 0x01E6A , 0x00054 , 0x00307 }, + { 0x01E6B , 0x00074 , 0x00307 }, + { 0x01E6C , 0x00054 , 0x00323 }, + { 0x01E6D , 0x00074 , 0x00323 }, + { 0x01E6E , 0x00054 , 0x00331 }, + { 0x01E6F , 0x00074 , 0x00331 }, + { 0x01E70 , 0x00054 , 0x0032D }, + { 0x01E71 , 0x00074 , 0x0032D }, + { 0x01E72 , 0x00055 , 0x00324 }, + { 0x01E73 , 0x00075 , 0x00324 }, + { 0x01E74 , 0x00055 , 0x00330 }, + { 0x01E75 , 0x00075 , 0x00330 }, + { 0x01E76 , 0x00055 , 0x0032D }, + { 0x01E77 , 0x00075 , 0x0032D }, + { 0x01E78 , 0x00168 , 0x00301 }, + { 0x01E79 , 0x00169 , 0x00301 }, + { 0x01E7A , 0x0016A , 0x00308 }, + { 0x01E7B , 0x0016B , 0x00308 }, + { 0x01E7C , 0x00056 , 0x00303 }, + { 0x01E7D , 0x00076 , 0x00303 }, + { 0x01E7E , 0x00056 , 0x00323 }, + { 0x01E7F , 0x00076 , 0x00323 }, + { 0x01E80 , 0x00057 , 0x00300 }, + { 0x01E81 , 0x00077 , 0x00300 }, + { 0x01E82 , 0x00057 , 0x00301 }, + { 0x01E83 , 0x00077 , 0x00301 }, + { 0x01E84 , 0x00057 , 0x00308 }, + { 0x01E85 , 0x00077 , 0x00308 }, + { 0x01E86 , 0x00057 , 0x00307 }, + { 0x01E87 , 0x00077 , 0x00307 }, + { 0x01E88 , 0x00057 , 0x00323 }, + { 0x01E89 , 0x00077 , 0x00323 }, + { 0x01E8A , 0x00058 , 0x00307 }, + { 0x01E8B , 0x00078 , 0x00307 }, + { 0x01E8C , 0x00058 , 0x00308 }, + { 0x01E8D , 0x00078 , 0x00308 }, + { 0x01E8E , 0x00059 , 0x00307 }, + { 0x01E8F , 0x00079 , 0x00307 }, + { 0x01E90 , 0x0005A , 0x00302 }, + { 0x01E91 , 0x0007A , 0x00302 }, + { 0x01E92 , 0x0005A , 0x00323 }, + { 0x01E93 , 0x0007A , 0x00323 }, + { 0x01E94 , 0x0005A , 0x00331 }, + { 0x01E95 , 0x0007A , 0x00331 }, + { 0x01E96 , 0x00068 , 0x00331 }, + { 0x01E97 , 0x00074 , 0x00308 }, + { 0x01E98 , 0x00077 , 0x0030A }, + { 0x01E99 , 0x00079 , 0x0030A }, + { 0x01E9B , 0x0017F , 0x00307 }, + { 0x01EA0 , 0x00041 , 0x00323 }, + { 0x01EA1 , 0x00061 , 0x00323 }, + { 0x01EA2 , 0x00041 , 0x00309 }, + { 0x01EA3 , 0x00061 , 0x00309 }, + { 0x01EA4 , 0x000C2 , 0x00301 }, + { 0x01EA5 , 0x000E2 , 0x00301 }, + { 0x01EA6 , 0x000C2 , 0x00300 }, + { 0x01EA7 , 0x000E2 , 0x00300 }, + { 0x01EA8 , 0x000C2 , 0x00309 }, + { 0x01EA9 , 0x000E2 , 0x00309 }, + { 0x01EAA , 0x000C2 , 0x00303 }, + { 0x01EAB , 0x000E2 , 0x00303 }, + { 0x01EAC , 0x01EA0 , 0x00302 }, + { 0x01EAD , 0x01EA1 , 0x00302 }, + { 0x01EAE , 0x00102 , 0x00301 }, + { 0x01EAF , 0x00103 , 0x00301 }, + { 0x01EB0 , 0x00102 , 0x00300 }, + { 0x01EB1 , 0x00103 , 0x00300 }, + { 0x01EB2 , 0x00102 , 0x00309 }, + { 0x01EB3 , 0x00103 , 0x00309 }, + { 0x01EB4 , 0x00102 , 0x00303 }, + { 0x01EB5 , 0x00103 , 0x00303 }, + { 0x01EB6 , 0x01EA0 , 0x00306 }, + { 0x01EB7 , 0x01EA1 , 0x00306 }, + { 0x01EB8 , 0x00045 , 0x00323 }, + { 0x01EB9 , 0x00065 , 0x00323 }, + { 0x01EBA , 0x00045 , 0x00309 }, + { 0x01EBB , 0x00065 , 0x00309 }, + { 0x01EBC , 0x00045 , 0x00303 }, + { 0x01EBD , 0x00065 , 0x00303 }, + { 0x01EBE , 0x000CA , 0x00301 }, + { 0x01EBF , 0x000EA , 0x00301 }, + { 0x01EC0 , 0x000CA , 0x00300 }, + { 0x01EC1 , 0x000EA , 0x00300 }, + { 0x01EC2 , 0x000CA , 0x00309 }, + { 0x01EC3 , 0x000EA , 0x00309 }, + { 0x01EC4 , 0x000CA , 0x00303 }, + { 0x01EC5 , 0x000EA , 0x00303 }, + { 0x01EC6 , 0x01EB8 , 0x00302 }, + { 0x01EC7 , 0x01EB9 , 0x00302 }, + { 0x01EC8 , 0x00049 , 0x00309 }, + { 0x01EC9 , 0x00069 , 0x00309 }, + { 0x01ECA , 0x00049 , 0x00323 }, + { 0x01ECB , 0x00069 , 0x00323 }, + { 0x01ECC , 0x0004F , 0x00323 }, + { 0x01ECD , 0x0006F , 0x00323 }, + { 0x01ECE , 0x0004F , 0x00309 }, + { 0x01ECF , 0x0006F , 0x00309 }, + { 0x01ED0 , 0x000D4 , 0x00301 }, + { 0x01ED1 , 0x000F4 , 0x00301 }, + { 0x01ED2 , 0x000D4 , 0x00300 }, + { 0x01ED3 , 0x000F4 , 0x00300 }, + { 0x01ED4 , 0x000D4 , 0x00309 }, + { 0x01ED5 , 0x000F4 , 0x00309 }, + { 0x01ED6 , 0x000D4 , 0x00303 }, + { 0x01ED7 , 0x000F4 , 0x00303 }, + { 0x01ED8 , 0x01ECC , 0x00302 }, + { 0x01ED9 , 0x01ECD , 0x00302 }, + { 0x01EDA , 0x001A0 , 0x00301 }, + { 0x01EDB , 0x001A1 , 0x00301 }, + { 0x01EDC , 0x001A0 , 0x00300 }, + { 0x01EDD , 0x001A1 , 0x00300 }, + { 0x01EDE , 0x001A0 , 0x00309 }, + { 0x01EDF , 0x001A1 , 0x00309 }, + { 0x01EE0 , 0x001A0 , 0x00303 }, + { 0x01EE1 , 0x001A1 , 0x00303 }, + { 0x01EE2 , 0x001A0 , 0x00323 }, + { 0x01EE3 , 0x001A1 , 0x00323 }, + { 0x01EE4 , 0x00055 , 0x00323 }, + { 0x01EE5 , 0x00075 , 0x00323 }, + { 0x01EE6 , 0x00055 , 0x00309 }, + { 0x01EE7 , 0x00075 , 0x00309 }, + { 0x01EE8 , 0x001AF , 0x00301 }, + { 0x01EE9 , 0x001B0 , 0x00301 }, + { 0x01EEA , 0x001AF , 0x00300 }, + { 0x01EEB , 0x001B0 , 0x00300 }, + { 0x01EEC , 0x001AF , 0x00309 }, + { 0x01EED , 0x001B0 , 0x00309 }, + { 0x01EEE , 0x001AF , 0x00303 }, + { 0x01EEF , 0x001B0 , 0x00303 }, + { 0x01EF0 , 0x001AF , 0x00323 }, + { 0x01EF1 , 0x001B0 , 0x00323 }, + { 0x01EF2 , 0x00059 , 0x00300 }, + { 0x01EF3 , 0x00079 , 0x00300 }, + { 0x01EF4 , 0x00059 , 0x00323 }, + { 0x01EF5 , 0x00079 , 0x00323 }, + { 0x01EF6 , 0x00059 , 0x00309 }, + { 0x01EF7 , 0x00079 , 0x00309 }, + { 0x01EF8 , 0x00059 , 0x00303 }, + { 0x01EF9 , 0x00079 , 0x00303 }, + { 0x01F00 , 0x003B1 , 0x00313 }, + { 0x01F01 , 0x003B1 , 0x00314 }, + { 0x01F02 , 0x01F00 , 0x00300 }, + { 0x01F03 , 0x01F01 , 0x00300 }, + { 0x01F04 , 0x01F00 , 0x00301 }, + { 0x01F05 , 0x01F01 , 0x00301 }, + { 0x01F06 , 0x01F00 , 0x00342 }, + { 0x01F07 , 0x01F01 , 0x00342 }, + { 0x01F08 , 0x00391 , 0x00313 }, + { 0x01F09 , 0x00391 , 0x00314 }, + { 0x01F0A , 0x01F08 , 0x00300 }, + { 0x01F0B , 0x01F09 , 0x00300 }, + { 0x01F0C , 0x01F08 , 0x00301 }, + { 0x01F0D , 0x01F09 , 0x00301 }, + { 0x01F0E , 0x01F08 , 0x00342 }, + { 0x01F0F , 0x01F09 , 0x00342 }, + { 0x01F10 , 0x003B5 , 0x00313 }, + { 0x01F11 , 0x003B5 , 0x00314 }, + { 0x01F12 , 0x01F10 , 0x00300 }, + { 0x01F13 , 0x01F11 , 0x00300 }, + { 0x01F14 , 0x01F10 , 0x00301 }, + { 0x01F15 , 0x01F11 , 0x00301 }, + { 0x01F18 , 0x00395 , 0x00313 }, + { 0x01F19 , 0x00395 , 0x00314 }, + { 0x01F1A , 0x01F18 , 0x00300 }, + { 0x01F1B , 0x01F19 , 0x00300 }, + { 0x01F1C , 0x01F18 , 0x00301 }, + { 0x01F1D , 0x01F19 , 0x00301 }, + { 0x01F20 , 0x003B7 , 0x00313 }, + { 0x01F21 , 0x003B7 , 0x00314 }, + { 0x01F22 , 0x01F20 , 0x00300 }, + { 0x01F23 , 0x01F21 , 0x00300 }, + { 0x01F24 , 0x01F20 , 0x00301 }, + { 0x01F25 , 0x01F21 , 0x00301 }, + { 0x01F26 , 0x01F20 , 0x00342 }, + { 0x01F27 , 0x01F21 , 0x00342 }, + { 0x01F28 , 0x00397 , 0x00313 }, + { 0x01F29 , 0x00397 , 0x00314 }, + { 0x01F2A , 0x01F28 , 0x00300 }, + { 0x01F2B , 0x01F29 , 0x00300 }, + { 0x01F2C , 0x01F28 , 0x00301 }, + { 0x01F2D , 0x01F29 , 0x00301 }, + { 0x01F2E , 0x01F28 , 0x00342 }, + { 0x01F2F , 0x01F29 , 0x00342 }, + { 0x01F30 , 0x003B9 , 0x00313 }, + { 0x01F31 , 0x003B9 , 0x00314 }, + { 0x01F32 , 0x01F30 , 0x00300 }, + { 0x01F33 , 0x01F31 , 0x00300 }, + { 0x01F34 , 0x01F30 , 0x00301 }, + { 0x01F35 , 0x01F31 , 0x00301 }, + { 0x01F36 , 0x01F30 , 0x00342 }, + { 0x01F37 , 0x01F31 , 0x00342 }, + { 0x01F38 , 0x00399 , 0x00313 }, + { 0x01F39 , 0x00399 , 0x00314 }, + { 0x01F3A , 0x01F38 , 0x00300 }, + { 0x01F3B , 0x01F39 , 0x00300 }, + { 0x01F3C , 0x01F38 , 0x00301 }, + { 0x01F3D , 0x01F39 , 0x00301 }, + { 0x01F3E , 0x01F38 , 0x00342 }, + { 0x01F3F , 0x01F39 , 0x00342 }, + { 0x01F40 , 0x003BF , 0x00313 }, + { 0x01F41 , 0x003BF , 0x00314 }, + { 0x01F42 , 0x01F40 , 0x00300 }, + { 0x01F43 , 0x01F41 , 0x00300 }, + { 0x01F44 , 0x01F40 , 0x00301 }, + { 0x01F45 , 0x01F41 , 0x00301 }, + { 0x01F48 , 0x0039F , 0x00313 }, + { 0x01F49 , 0x0039F , 0x00314 }, + { 0x01F4A , 0x01F48 , 0x00300 }, + { 0x01F4B , 0x01F49 , 0x00300 }, + { 0x01F4C , 0x01F48 , 0x00301 }, + { 0x01F4D , 0x01F49 , 0x00301 }, + { 0x01F50 , 0x003C5 , 0x00313 }, + { 0x01F51 , 0x003C5 , 0x00314 }, + { 0x01F52 , 0x01F50 , 0x00300 }, + { 0x01F53 , 0x01F51 , 0x00300 }, + { 0x01F54 , 0x01F50 , 0x00301 }, + { 0x01F55 , 0x01F51 , 0x00301 }, + { 0x01F56 , 0x01F50 , 0x00342 }, + { 0x01F57 , 0x01F51 , 0x00342 }, + { 0x01F59 , 0x003A5 , 0x00314 }, + { 0x01F5B , 0x01F59 , 0x00300 }, + { 0x01F5D , 0x01F59 , 0x00301 }, + { 0x01F5F , 0x01F59 , 0x00342 }, + { 0x01F60 , 0x003C9 , 0x00313 }, + { 0x01F61 , 0x003C9 , 0x00314 }, + { 0x01F62 , 0x01F60 , 0x00300 }, + { 0x01F63 , 0x01F61 , 0x00300 }, + { 0x01F64 , 0x01F60 , 0x00301 }, + { 0x01F65 , 0x01F61 , 0x00301 }, + { 0x01F66 , 0x01F60 , 0x00342 }, + { 0x01F67 , 0x01F61 , 0x00342 }, + { 0x01F68 , 0x003A9 , 0x00313 }, + { 0x01F69 , 0x003A9 , 0x00314 }, + { 0x01F6A , 0x01F68 , 0x00300 }, + { 0x01F6B , 0x01F69 , 0x00300 }, + { 0x01F6C , 0x01F68 , 0x00301 }, + { 0x01F6D , 0x01F69 , 0x00301 }, + { 0x01F6E , 0x01F68 , 0x00342 }, + { 0x01F6F , 0x01F69 , 0x00342 }, + { 0x01F70 , 0x003B1 , 0x00300 }, + { 0x01F72 , 0x003B5 , 0x00300 }, + { 0x01F74 , 0x003B7 , 0x00300 }, + { 0x01F76 , 0x003B9 , 0x00300 }, + { 0x01F78 , 0x003BF , 0x00300 }, + { 0x01F7A , 0x003C5 , 0x00300 }, + { 0x01F7C , 0x003C9 , 0x00300 }, + { 0x01F80 , 0x01F00 , 0x00345 }, + { 0x01F81 , 0x01F01 , 0x00345 }, + { 0x01F82 , 0x01F02 , 0x00345 }, + { 0x01F83 , 0x01F03 , 0x00345 }, + { 0x01F84 , 0x01F04 , 0x00345 }, + { 0x01F85 , 0x01F05 , 0x00345 }, + { 0x01F86 , 0x01F06 , 0x00345 }, + { 0x01F87 , 0x01F07 , 0x00345 }, + { 0x01F88 , 0x01F08 , 0x00345 }, + { 0x01F89 , 0x01F09 , 0x00345 }, + { 0x01F8A , 0x01F0A , 0x00345 }, + { 0x01F8B , 0x01F0B , 0x00345 }, + { 0x01F8C , 0x01F0C , 0x00345 }, + { 0x01F8D , 0x01F0D , 0x00345 }, + { 0x01F8E , 0x01F0E , 0x00345 }, + { 0x01F8F , 0x01F0F , 0x00345 }, + { 0x01F90 , 0x01F20 , 0x00345 }, + { 0x01F91 , 0x01F21 , 0x00345 }, + { 0x01F92 , 0x01F22 , 0x00345 }, + { 0x01F93 , 0x01F23 , 0x00345 }, + { 0x01F94 , 0x01F24 , 0x00345 }, + { 0x01F95 , 0x01F25 , 0x00345 }, + { 0x01F96 , 0x01F26 , 0x00345 }, + { 0x01F97 , 0x01F27 , 0x00345 }, + { 0x01F98 , 0x01F28 , 0x00345 }, + { 0x01F99 , 0x01F29 , 0x00345 }, + { 0x01F9A , 0x01F2A , 0x00345 }, + { 0x01F9B , 0x01F2B , 0x00345 }, + { 0x01F9C , 0x01F2C , 0x00345 }, + { 0x01F9D , 0x01F2D , 0x00345 }, + { 0x01F9E , 0x01F2E , 0x00345 }, + { 0x01F9F , 0x01F2F , 0x00345 }, + { 0x01FA0 , 0x01F60 , 0x00345 }, + { 0x01FA1 , 0x01F61 , 0x00345 }, + { 0x01FA2 , 0x01F62 , 0x00345 }, + { 0x01FA3 , 0x01F63 , 0x00345 }, + { 0x01FA4 , 0x01F64 , 0x00345 }, + { 0x01FA5 , 0x01F65 , 0x00345 }, + { 0x01FA6 , 0x01F66 , 0x00345 }, + { 0x01FA7 , 0x01F67 , 0x00345 }, + { 0x01FA8 , 0x01F68 , 0x00345 }, + { 0x01FA9 , 0x01F69 , 0x00345 }, + { 0x01FAA , 0x01F6A , 0x00345 }, + { 0x01FAB , 0x01F6B , 0x00345 }, + { 0x01FAC , 0x01F6C , 0x00345 }, + { 0x01FAD , 0x01F6D , 0x00345 }, + { 0x01FAE , 0x01F6E , 0x00345 }, + { 0x01FAF , 0x01F6F , 0x00345 }, + { 0x01FB0 , 0x003B1 , 0x00306 }, + { 0x01FB1 , 0x003B1 , 0x00304 }, + { 0x01FB2 , 0x01F70 , 0x00345 }, + { 0x01FB3 , 0x003B1 , 0x00345 }, + { 0x01FB4 , 0x003AC , 0x00345 }, + { 0x01FB6 , 0x003B1 , 0x00342 }, + { 0x01FB7 , 0x01FB6 , 0x00345 }, + { 0x01FB8 , 0x00391 , 0x00306 }, + { 0x01FB9 , 0x00391 , 0x00304 }, + { 0x01FBA , 0x00391 , 0x00300 }, + { 0x01FBC , 0x00391 , 0x00345 }, + { 0x01FC1 , 0x000A8 , 0x00342 }, + { 0x01FC2 , 0x01F74 , 0x00345 }, + { 0x01FC3 , 0x003B7 , 0x00345 }, + { 0x01FC4 , 0x003AE , 0x00345 }, + { 0x01FC6 , 0x003B7 , 0x00342 }, + { 0x01FC7 , 0x01FC6 , 0x00345 }, + { 0x01FC8 , 0x00395 , 0x00300 }, + { 0x01FCA , 0x00397 , 0x00300 }, + { 0x01FCC , 0x00397 , 0x00345 }, + { 0x01FCD , 0x01FBF , 0x00300 }, + { 0x01FCE , 0x01FBF , 0x00301 }, + { 0x01FCF , 0x01FBF , 0x00342 }, + { 0x01FD0 , 0x003B9 , 0x00306 }, + { 0x01FD1 , 0x003B9 , 0x00304 }, + { 0x01FD2 , 0x003CA , 0x00300 }, + { 0x01FD6 , 0x003B9 , 0x00342 }, + { 0x01FD7 , 0x003CA , 0x00342 }, + { 0x01FD8 , 0x00399 , 0x00306 }, + { 0x01FD9 , 0x00399 , 0x00304 }, + { 0x01FDA , 0x00399 , 0x00300 }, + { 0x01FDD , 0x01FFE , 0x00300 }, + { 0x01FDE , 0x01FFE , 0x00301 }, + { 0x01FDF , 0x01FFE , 0x00342 }, + { 0x01FE0 , 0x003C5 , 0x00306 }, + { 0x01FE1 , 0x003C5 , 0x00304 }, + { 0x01FE2 , 0x003CB , 0x00300 }, + { 0x01FE4 , 0x003C1 , 0x00313 }, + { 0x01FE5 , 0x003C1 , 0x00314 }, + { 0x01FE6 , 0x003C5 , 0x00342 }, + { 0x01FE7 , 0x003CB , 0x00342 }, + { 0x01FE8 , 0x003A5 , 0x00306 }, + { 0x01FE9 , 0x003A5 , 0x00304 }, + { 0x01FEA , 0x003A5 , 0x00300 }, + { 0x01FEC , 0x003A1 , 0x00314 }, + { 0x01FED , 0x000A8 , 0x00300 }, + { 0x01FF2 , 0x01F7C , 0x00345 }, + { 0x01FF3 , 0x003C9 , 0x00345 }, + { 0x01FF4 , 0x003CE , 0x00345 }, + { 0x01FF6 , 0x003C9 , 0x00342 }, + { 0x01FF7 , 0x01FF6 , 0x00345 }, + { 0x01FF8 , 0x0039F , 0x00300 }, + { 0x01FFA , 0x003A9 , 0x00300 }, + { 0x01FFC , 0x003A9 , 0x00345 }, + { 0x0219A , 0x02190 , 0x00338 }, + { 0x0219B , 0x02192 , 0x00338 }, + { 0x021AE , 0x02194 , 0x00338 }, + { 0x021CD , 0x021D0 , 0x00338 }, + { 0x021CE , 0x021D4 , 0x00338 }, + { 0x021CF , 0x021D2 , 0x00338 }, + { 0x02204 , 0x02203 , 0x00338 }, + { 0x02209 , 0x02208 , 0x00338 }, + { 0x0220C , 0x0220B , 0x00338 }, + { 0x02224 , 0x02223 , 0x00338 }, + { 0x02226 , 0x02225 , 0x00338 }, + { 0x02241 , 0x0223C , 0x00338 }, + { 0x02244 , 0x02243 , 0x00338 }, + { 0x02247 , 0x02245 , 0x00338 }, + { 0x02249 , 0x02248 , 0x00338 }, + { 0x02260 , 0x0003D , 0x00338 }, + { 0x02262 , 0x02261 , 0x00338 }, + { 0x0226D , 0x0224D , 0x00338 }, + { 0x0226E , 0x0003C , 0x00338 }, + { 0x0226F , 0x0003E , 0x00338 }, + { 0x02270 , 0x02264 , 0x00338 }, + { 0x02271 , 0x02265 , 0x00338 }, + { 0x02274 , 0x02272 , 0x00338 }, + { 0x02275 , 0x02273 , 0x00338 }, + { 0x02278 , 0x02276 , 0x00338 }, + { 0x02279 , 0x02277 , 0x00338 }, + { 0x02280 , 0x0227A , 0x00338 }, + { 0x02281 , 0x0227B , 0x00338 }, + { 0x02284 , 0x02282 , 0x00338 }, + { 0x02285 , 0x02283 , 0x00338 }, + { 0x02288 , 0x02286 , 0x00338 }, + { 0x02289 , 0x02287 , 0x00338 }, + { 0x022AC , 0x022A2 , 0x00338 }, + { 0x022AD , 0x022A8 , 0x00338 }, + { 0x022AE , 0x022A9 , 0x00338 }, + { 0x022AF , 0x022AB , 0x00338 }, + { 0x022E0 , 0x0227C , 0x00338 }, + { 0x022E1 , 0x0227D , 0x00338 }, + { 0x022E2 , 0x02291 , 0x00338 }, + { 0x022E3 , 0x02292 , 0x00338 }, + { 0x022EA , 0x022B2 , 0x00338 }, + { 0x022EB , 0x022B3 , 0x00338 }, + { 0x022EC , 0x022B4 , 0x00338 }, + { 0x022ED , 0x022B5 , 0x00338 }, + { 0x0304C , 0x0304B , 0x03099 }, + { 0x0304E , 0x0304D , 0x03099 }, + { 0x03050 , 0x0304F , 0x03099 }, + { 0x03052 , 0x03051 , 0x03099 }, + { 0x03054 , 0x03053 , 0x03099 }, + { 0x03056 , 0x03055 , 0x03099 }, + { 0x03058 , 0x03057 , 0x03099 }, + { 0x0305A , 0x03059 , 0x03099 }, + { 0x0305C , 0x0305B , 0x03099 }, + { 0x0305E , 0x0305D , 0x03099 }, + { 0x03060 , 0x0305F , 0x03099 }, + { 0x03062 , 0x03061 , 0x03099 }, + { 0x03065 , 0x03064 , 0x03099 }, + { 0x03067 , 0x03066 , 0x03099 }, + { 0x03069 , 0x03068 , 0x03099 }, + { 0x03070 , 0x0306F , 0x03099 }, + { 0x03071 , 0x0306F , 0x0309A }, + { 0x03073 , 0x03072 , 0x03099 }, + { 0x03074 , 0x03072 , 0x0309A }, + { 0x03076 , 0x03075 , 0x03099 }, + { 0x03077 , 0x03075 , 0x0309A }, + { 0x03079 , 0x03078 , 0x03099 }, + { 0x0307A , 0x03078 , 0x0309A }, + { 0x0307C , 0x0307B , 0x03099 }, + { 0x0307D , 0x0307B , 0x0309A }, + { 0x03094 , 0x03046 , 0x03099 }, + { 0x0309E , 0x0309D , 0x03099 }, + { 0x030AC , 0x030AB , 0x03099 }, + { 0x030AE , 0x030AD , 0x03099 }, + { 0x030B0 , 0x030AF , 0x03099 }, + { 0x030B2 , 0x030B1 , 0x03099 }, + { 0x030B4 , 0x030B3 , 0x03099 }, + { 0x030B6 , 0x030B5 , 0x03099 }, + { 0x030B8 , 0x030B7 , 0x03099 }, + { 0x030BA , 0x030B9 , 0x03099 }, + { 0x030BC , 0x030BB , 0x03099 }, + { 0x030BE , 0x030BD , 0x03099 }, + { 0x030C0 , 0x030BF , 0x03099 }, + { 0x030C2 , 0x030C1 , 0x03099 }, + { 0x030C5 , 0x030C4 , 0x03099 }, + { 0x030C7 , 0x030C6 , 0x03099 }, + { 0x030C9 , 0x030C8 , 0x03099 }, + { 0x030D0 , 0x030CF , 0x03099 }, + { 0x030D1 , 0x030CF , 0x0309A }, + { 0x030D3 , 0x030D2 , 0x03099 }, + { 0x030D4 , 0x030D2 , 0x0309A }, + { 0x030D6 , 0x030D5 , 0x03099 }, + { 0x030D7 , 0x030D5 , 0x0309A }, + { 0x030D9 , 0x030D8 , 0x03099 }, + { 0x030DA , 0x030D8 , 0x0309A }, + { 0x030DC , 0x030DB , 0x03099 }, + { 0x030DD , 0x030DB , 0x0309A }, + { 0x030F4 , 0x030A6 , 0x03099 }, + { 0x030F7 , 0x030EF , 0x03099 }, + { 0x030F8 , 0x030F0 , 0x03099 }, + { 0x030F9 , 0x030F1 , 0x03099 }, + { 0x030FA , 0x030F2 , 0x03099 }, + { 0x030FE , 0x030FD , 0x03099 }, + { 0x1109A , 0x11099 , 0x110BA }, + { 0x1109C , 0x1109B , 0x110BA }, + { 0x110AB , 0x110A5 , 0x110BA }, +}; + +#endif /* ARCHIVE_STRING_COMPOSITION_H_INCLUDED */ + diff --git a/c/archive_string_sprintf.c b/c/archive_string_sprintf.c new file mode 100644 --- /dev/null +++ b/c/archive_string_sprintf.c @@ -0,0 +1,192 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_string_sprintf.c 189435 2009-03-06 05:14:55Z kientzle $"); + +/* + * The use of printf()-family functions can be troublesome + * for space-constrained applications. In addition, correctly + * implementing this function in terms of vsnprintf() requires + * two calls (one to determine the size, another to format the + * result), which in turn requires duplicating the argument list + * using va_copy, which isn't yet universally available. + * + * So, I've implemented a bare minimum of printf()-like capability + * here. This is only used to format error messages, so doesn't + * require any floating-point support or field-width handling. + */ +#ifdef HAVE_ERRNO_H +#include +#endif +#include + +#include "archive_string.h" +#include "archive_private.h" + +/* + * Utility functions to format signed/unsigned integers and append + * them to an archive_string. + */ +static void +append_uint(struct archive_string *as, uintmax_t d, unsigned base) +{ + static const char digits[] = "0123456789abcdef"; + if (d >= base) + append_uint(as, d/base, base); + archive_strappend_char(as, digits[d % base]); +} + +static void +append_int(struct archive_string *as, intmax_t d, unsigned base) +{ + uintmax_t ud; + + if (d < 0) { + archive_strappend_char(as, '-'); + ud = (d == INTMAX_MIN) ? (uintmax_t)(INTMAX_MAX) + 1 : (uintmax_t)(-d); + } else + ud = d; + append_uint(as, ud, base); +} + + +void +archive_string_sprintf(struct archive_string *as, const char *fmt, ...) +{ + va_list ap; + + va_start(ap, fmt); + archive_string_vsprintf(as, fmt, ap); + va_end(ap); +} + +/* + * Like 'vsprintf', but ensures the target is big enough, resizing if + * necessary. + */ +void +archive_string_vsprintf(struct archive_string *as, const char *fmt, + va_list ap) +{ + char long_flag; + intmax_t s; /* Signed integer temp. */ + uintmax_t u; /* Unsigned integer temp. */ + const char *p, *p2; + const wchar_t *pw; + + if (archive_string_ensure(as, 64) == NULL) + __archive_errx(1, "Out of memory"); + + if (fmt == NULL) { + as->s[0] = 0; + return; + } + + for (p = fmt; *p != '\0'; p++) { + const char *saved_p = p; + + if (*p != '%') { + archive_strappend_char(as, *p); + continue; + } + + p++; + + long_flag = '\0'; + switch(*p) { + case 'j': + case 'l': + case 'z': + long_flag = *p; + p++; + break; + } + + switch (*p) { + case '%': + archive_strappend_char(as, '%'); + break; + case 'c': + s = va_arg(ap, int); + archive_strappend_char(as, (char)s); + break; + case 'd': + switch(long_flag) { + case 'j': s = va_arg(ap, intmax_t); break; + case 'l': s = va_arg(ap, long); break; + case 'z': s = va_arg(ap, ssize_t); break; + default: s = va_arg(ap, int); break; + } + append_int(as, s, 10); + break; + case 's': + switch(long_flag) { + case 'l': + pw = va_arg(ap, wchar_t *); + if (pw == NULL) + pw = L"(null)"; + if (archive_string_append_from_wcs(as, pw, + wcslen(pw)) != 0 && errno == ENOMEM) + __archive_errx(1, "Out of memory"); + break; + default: + p2 = va_arg(ap, char *); + if (p2 == NULL) + p2 = "(null)"; + archive_strcat(as, p2); + break; + } + break; + case 'S': + pw = va_arg(ap, wchar_t *); + if (pw == NULL) + pw = L"(null)"; + if (archive_string_append_from_wcs(as, pw, + wcslen(pw)) != 0 && errno == ENOMEM) + __archive_errx(1, "Out of memory"); + break; + case 'o': case 'u': case 'x': case 'X': + /* Common handling for unsigned integer formats. */ + switch(long_flag) { + case 'j': u = va_arg(ap, uintmax_t); break; + case 'l': u = va_arg(ap, unsigned long); break; + case 'z': u = va_arg(ap, size_t); break; + default: u = va_arg(ap, unsigned int); break; + } + /* Format it in the correct base. */ + switch (*p) { + case 'o': append_uint(as, u, 8); break; + case 'u': append_uint(as, u, 10); break; + default: append_uint(as, u, 16); break; + } + break; + default: + /* Rewind and print the initial '%' literally. */ + p = saved_p; + archive_strappend_char(as, *p); + } + } +} diff --git a/c/archive_util.c b/c/archive_util.c new file mode 100644 --- /dev/null +++ b/c/archive_util.c @@ -0,0 +1,650 @@ +/*- + * Copyright (c) 2009-2012,2014 Michihiro NAKAJIMA + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_util.c 201098 2009-12-28 02:58:14Z kientzle $"); + +#ifdef HAVE_SYS_TYPES_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_FCNTL_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#if defined(HAVE_WINCRYPT_H) && !defined(__CYGWIN__) +#include +#endif +#ifdef HAVE_ZLIB_H +#include +#endif +#ifdef HAVE_LZMA_H +#include +#endif +#ifdef HAVE_BZLIB_H +#include +#endif +#ifdef HAVE_LZ4_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_random_private.h" +#include "archive_string.h" + +#ifndef O_CLOEXEC +#define O_CLOEXEC 0 +#endif + +static int archive_utility_string_sort_helper(char **, unsigned int); + +/* Generic initialization of 'struct archive' objects. */ +int +__archive_clean(struct archive *a) +{ + archive_string_conversion_free(a); + return (ARCHIVE_OK); +} + +int +archive_version_number(void) +{ + return (ARCHIVE_VERSION_NUMBER); +} + +const char * +archive_version_string(void) +{ + return (ARCHIVE_VERSION_STRING); +} + +int +archive_errno(struct archive *a) +{ + return (a->archive_error_number); +} + +const char * +archive_error_string(struct archive *a) +{ + + if (a->error != NULL && *a->error != '\0') + return (a->error); + else + return (NULL); +} + +int +archive_file_count(struct archive *a) +{ + return (a->file_count); +} + +int +archive_format(struct archive *a) +{ + return (a->archive_format); +} + +const char * +archive_format_name(struct archive *a) +{ + return (a->archive_format_name); +} + + +int +archive_compression(struct archive *a) +{ + return archive_filter_code(a, 0); +} + +const char * +archive_compression_name(struct archive *a) +{ + return archive_filter_name(a, 0); +} + + +/* + * Return a count of the number of compressed bytes processed. + */ +la_int64_t +archive_position_compressed(struct archive *a) +{ + return archive_filter_bytes(a, -1); +} + +/* + * Return a count of the number of uncompressed bytes processed. + */ +la_int64_t +archive_position_uncompressed(struct archive *a) +{ + return archive_filter_bytes(a, 0); +} + +void +archive_clear_error(struct archive *a) +{ + archive_string_empty(&a->error_string); + a->error = NULL; + a->archive_error_number = 0; +} + +void +archive_set_error(struct archive *a, int error_number, const char *fmt, ...) +{ + va_list ap; + + a->archive_error_number = error_number; + if (fmt == NULL) { + a->error = NULL; + return; + } + + archive_string_empty(&(a->error_string)); + va_start(ap, fmt); + archive_string_vsprintf(&(a->error_string), fmt, ap); + va_end(ap); + a->error = a->error_string.s; +} + +void +archive_copy_error(struct archive *dest, struct archive *src) +{ + dest->archive_error_number = src->archive_error_number; + + archive_string_copy(&dest->error_string, &src->error_string); + dest->error = dest->error_string.s; +} + +void +__archive_errx(int retvalue, const char *msg) +{ + static const char msg1[] = "Fatal Internal Error in libarchive: "; + size_t s; + + s = write(2, msg1, strlen(msg1)); + (void)s; /* UNUSED */ + s = write(2, msg, strlen(msg)); + (void)s; /* UNUSED */ + s = write(2, "\n", 1); + (void)s; /* UNUSED */ + exit(retvalue); +} + +/* + * Create a temporary file + */ +#if defined(_WIN32) && !defined(__CYGWIN__) + +/* + * Do not use Windows tmpfile() function. + * It will make a temporary file under the root directory + * and it'll cause permission error if a user who is + * non-Administrator creates temporary files. + * Also Windows version of mktemp family including _mktemp_s + * are not secure. + */ +static int +__archive_mktempx(const char *tmpdir, wchar_t *template) +{ + static const wchar_t prefix[] = L"libarchive_"; + static const wchar_t suffix[] = L"XXXXXXXXXX"; + static const wchar_t num[] = { + L'0', L'1', L'2', L'3', L'4', L'5', L'6', L'7', + L'8', L'9', L'A', L'B', L'C', L'D', L'E', L'F', + L'G', L'H', L'I', L'J', L'K', L'L', L'M', L'N', + L'O', L'P', L'Q', L'R', L'S', L'T', L'U', L'V', + L'W', L'X', L'Y', L'Z', L'a', L'b', L'c', L'd', + L'e', L'f', L'g', L'h', L'i', L'j', L'k', L'l', + L'm', L'n', L'o', L'p', L'q', L'r', L's', L't', + L'u', L'v', L'w', L'x', L'y', L'z' + }; + HCRYPTPROV hProv; + struct archive_wstring temp_name; + wchar_t *ws; + DWORD attr; + wchar_t *xp, *ep; + int fd; + + hProv = (HCRYPTPROV)NULL; + fd = -1; + ws = NULL; + + if (template == NULL) { + archive_string_init(&temp_name); + + /* Get a temporary directory. */ + if (tmpdir == NULL) { + size_t l; + wchar_t *tmp; + + l = GetTempPathW(0, NULL); + if (l == 0) { + la_dosmaperr(GetLastError()); + goto exit_tmpfile; + } + tmp = malloc(l*sizeof(wchar_t)); + if (tmp == NULL) { + errno = ENOMEM; + goto exit_tmpfile; + } + GetTempPathW((DWORD)l, tmp); + archive_wstrcpy(&temp_name, tmp); + free(tmp); + } else { + if (archive_wstring_append_from_mbs(&temp_name, tmpdir, + strlen(tmpdir)) < 0) + goto exit_tmpfile; + if (temp_name.s[temp_name.length-1] != L'/') + archive_wstrappend_wchar(&temp_name, L'/'); + } + + /* Check if temp_name is a directory. */ + attr = GetFileAttributesW(temp_name.s); + if (attr == (DWORD)-1) { + if (GetLastError() != ERROR_FILE_NOT_FOUND) { + la_dosmaperr(GetLastError()); + goto exit_tmpfile; + } + ws = __la_win_permissive_name_w(temp_name.s); + if (ws == NULL) { + errno = EINVAL; + goto exit_tmpfile; + } + attr = GetFileAttributesW(ws); + if (attr == (DWORD)-1) { + la_dosmaperr(GetLastError()); + goto exit_tmpfile; + } + } + if (!(attr & FILE_ATTRIBUTE_DIRECTORY)) { + errno = ENOTDIR; + goto exit_tmpfile; + } + + /* + * Create a temporary file. + */ + archive_wstrcat(&temp_name, prefix); + archive_wstrcat(&temp_name, suffix); + ep = temp_name.s + archive_strlen(&temp_name); + xp = ep - wcslen(suffix); + template = temp_name.s; + } else { + xp = wcschr(template, L'X'); + if (xp == NULL) /* No X, programming error */ + abort(); + for (ep = xp; *ep == L'X'; ep++) + continue; + if (*ep) /* X followed by non X, programming error */ + abort(); + } + + if (!CryptAcquireContext(&hProv, NULL, NULL, PROV_RSA_FULL, + CRYPT_VERIFYCONTEXT)) { + la_dosmaperr(GetLastError()); + goto exit_tmpfile; + } + + for (;;) { + wchar_t *p; + HANDLE h; + + /* Generate a random file name through CryptGenRandom(). */ + p = xp; + if (!CryptGenRandom(hProv, (DWORD)(ep - p)*sizeof(wchar_t), + (BYTE*)p)) { + la_dosmaperr(GetLastError()); + goto exit_tmpfile; + } + for (; p < ep; p++) + *p = num[((DWORD)*p) % (sizeof(num)/sizeof(num[0]))]; + + free(ws); + ws = __la_win_permissive_name_w(template); + if (ws == NULL) { + errno = EINVAL; + goto exit_tmpfile; + } + if (template == temp_name.s) { + attr = FILE_ATTRIBUTE_TEMPORARY | + FILE_FLAG_DELETE_ON_CLOSE; + } else { + /* mkstemp */ + attr = FILE_ATTRIBUTE_NORMAL; + } + h = CreateFileW(ws, + GENERIC_READ | GENERIC_WRITE | DELETE, + 0,/* Not share */ + NULL, + CREATE_NEW,/* Create a new file only */ + attr, + NULL); + if (h == INVALID_HANDLE_VALUE) { + /* The same file already exists. retry with + * a new filename. */ + if (GetLastError() == ERROR_FILE_EXISTS) + continue; + /* Otherwise, fail creation temporary file. */ + la_dosmaperr(GetLastError()); + goto exit_tmpfile; + } + fd = _open_osfhandle((intptr_t)h, _O_BINARY | _O_RDWR); + if (fd == -1) { + la_dosmaperr(GetLastError()); + CloseHandle(h); + goto exit_tmpfile; + } else + break;/* success! */ + } +exit_tmpfile: + if (hProv != (HCRYPTPROV)NULL) + CryptReleaseContext(hProv, 0); + free(ws); + if (template == temp_name.s) + archive_wstring_free(&temp_name); + return (fd); +} + +int +__archive_mktemp(const char *tmpdir) +{ + return __archive_mktempx(tmpdir, NULL); +} + +int +__archive_mkstemp(wchar_t *template) +{ + return __archive_mktempx(NULL, template); +} + +#else + +static int +get_tempdir(struct archive_string *temppath) +{ + const char *tmp; + + tmp = getenv("TMPDIR"); + if (tmp == NULL) +#ifdef _PATH_TMP + tmp = _PATH_TMP; +#else + tmp = "/tmp"; +#endif + archive_strcpy(temppath, tmp); + if (temppath->s[temppath->length-1] != '/') + archive_strappend_char(temppath, '/'); + return (ARCHIVE_OK); +} + +#if defined(HAVE_MKSTEMP) + +/* + * We can use mkstemp(). + */ + +int +__archive_mktemp(const char *tmpdir) +{ + struct archive_string temp_name; + int fd = -1; + + archive_string_init(&temp_name); + if (tmpdir == NULL) { + if (get_tempdir(&temp_name) != ARCHIVE_OK) + goto exit_tmpfile; + } else { + archive_strcpy(&temp_name, tmpdir); + if (temp_name.s[temp_name.length-1] != '/') + archive_strappend_char(&temp_name, '/'); + } + archive_strcat(&temp_name, "libarchive_XXXXXX"); + fd = mkstemp(temp_name.s); + if (fd < 0) + goto exit_tmpfile; + __archive_ensure_cloexec_flag(fd); + unlink(temp_name.s); +exit_tmpfile: + archive_string_free(&temp_name); + return (fd); +} + +int +__archive_mkstemp(char *template) +{ + int fd = -1; + fd = mkstemp(template); + if (fd >= 0) + __archive_ensure_cloexec_flag(fd); + return (fd); +} + +#else /* !HAVE_MKSTEMP */ + +/* + * We use a private routine. + */ + +static int +__archive_mktempx(const char *tmpdir, char *template) +{ + static const char num[] = { + '0', '1', '2', '3', '4', '5', '6', '7', + '8', '9', 'A', 'B', 'C', 'D', 'E', 'F', + 'G', 'H', 'I', 'J', 'K', 'L', 'M', 'N', + 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', + 'W', 'X', 'Y', 'Z', 'a', 'b', 'c', 'd', + 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', + 'm', 'n', 'o', 'p', 'q', 'r', 's', 't', + 'u', 'v', 'w', 'x', 'y', 'z' + }; + struct archive_string temp_name; + struct stat st; + int fd; + char *tp, *ep; + + fd = -1; + if (template == NULL) { + archive_string_init(&temp_name); + if (tmpdir == NULL) { + if (get_tempdir(&temp_name) != ARCHIVE_OK) + goto exit_tmpfile; + } else + archive_strcpy(&temp_name, tmpdir); + if (temp_name.s[temp_name.length-1] == '/') { + temp_name.s[temp_name.length-1] = '\0'; + temp_name.length --; + } + if (la_stat(temp_name.s, &st) < 0) + goto exit_tmpfile; + if (!S_ISDIR(st.st_mode)) { + errno = ENOTDIR; + goto exit_tmpfile; + } + archive_strcat(&temp_name, "/libarchive_"); + tp = temp_name.s + archive_strlen(&temp_name); + archive_strcat(&temp_name, "XXXXXXXXXX"); + ep = temp_name.s + archive_strlen(&temp_name); + template = temp_name.s; + } else { + tp = strchr(template, 'X'); + if (tp == NULL) /* No X, programming error */ + abort(); + for (ep = tp; *ep == 'X'; ep++) + continue; + if (*ep) /* X followed by non X, programming error */ + abort(); + } + + do { + char *p; + + p = tp; + archive_random(p, ep - p); + while (p < ep) { + int d = *((unsigned char *)p) % sizeof(num); + *p++ = num[d]; + } + fd = open(template, O_CREAT | O_EXCL | O_RDWR | O_CLOEXEC, + 0600); + } while (fd < 0 && errno == EEXIST); + if (fd < 0) + goto exit_tmpfile; + __archive_ensure_cloexec_flag(fd); + if (template == temp_name.s) + unlink(temp_name.s); +exit_tmpfile: + if (template == temp_name.s) + archive_string_free(&temp_name); + return (fd); +} + +int +__archive_mktemp(const char *tmpdir) +{ + return __archive_mktempx(tmpdir, NULL); +} + +int +__archive_mkstemp(char *template) +{ + return __archive_mktempx(NULL, template); +} + +#endif /* !HAVE_MKSTEMP */ +#endif /* !_WIN32 || __CYGWIN__ */ + +/* + * Set FD_CLOEXEC flag to a file descriptor if it is not set. + * We have to set the flag if the platform does not provide O_CLOEXEC + * or F_DUPFD_CLOEXEC flags. + * + * Note: This function is absolutely called after creating a new file + * descriptor even if the platform seemingly provides O_CLOEXEC or + * F_DUPFD_CLOEXEC macros because it is possible that the platform + * merely declares those macros, especially Linux 2.6.18 - 2.6.24 do it. + */ +void +__archive_ensure_cloexec_flag(int fd) +{ +#if defined(_WIN32) && !defined(__CYGWIN__) + (void)fd; /* UNUSED */ +#else + int flags; + + if (fd >= 0) { + flags = fcntl(fd, F_GETFD); + if (flags != -1 && (flags & FD_CLOEXEC) == 0) + fcntl(fd, F_SETFD, flags | FD_CLOEXEC); + } +#endif +} + +/* + * Utility function to sort a group of strings using quicksort. + */ +static int +archive_utility_string_sort_helper(char **strings, unsigned int n) +{ + unsigned int i, lesser_count, greater_count; + char **lesser, **greater, **tmp, *pivot; + int retval1, retval2; + + /* A list of 0 or 1 elements is already sorted */ + if (n <= 1) + return (ARCHIVE_OK); + + lesser_count = greater_count = 0; + lesser = greater = NULL; + pivot = strings[0]; + for (i = 1; i < n; i++) + { + if (strcmp(strings[i], pivot) < 0) + { + lesser_count++; + tmp = (char **)realloc(lesser, + lesser_count * sizeof(char *)); + if (!tmp) { + free(greater); + free(lesser); + return (ARCHIVE_FATAL); + } + lesser = tmp; + lesser[lesser_count - 1] = strings[i]; + } + else + { + greater_count++; + tmp = (char **)realloc(greater, + greater_count * sizeof(char *)); + if (!tmp) { + free(greater); + free(lesser); + return (ARCHIVE_FATAL); + } + greater = tmp; + greater[greater_count - 1] = strings[i]; + } + } + + /* quicksort(lesser) */ + retval1 = archive_utility_string_sort_helper(lesser, lesser_count); + for (i = 0; i < lesser_count; i++) + strings[i] = lesser[i]; + free(lesser); + + /* pivot */ + strings[lesser_count] = pivot; + + /* quicksort(greater) */ + retval2 = archive_utility_string_sort_helper(greater, greater_count); + for (i = 0; i < greater_count; i++) + strings[lesser_count + 1 + i] = greater[i]; + free(greater); + + return (retval1 < retval2) ? retval1 : retval2; +} + +int +archive_utility_string_sort(char **strings) +{ + unsigned int size = 0; + while (strings[size] != NULL) + size++; + return archive_utility_string_sort_helper(strings, size); +} diff --git a/c/archive_version_details.c b/c/archive_version_details.c new file mode 100644 --- /dev/null +++ b/c/archive_version_details.c @@ -0,0 +1,151 @@ +/*- + * Copyright (c) 2009-2012,2014 Michihiro NAKAJIMA + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_util.c 201098 2009-12-28 02:58:14Z kientzle $"); + +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_ZLIB_H +#include +#endif +#ifdef HAVE_LZMA_H +#include +#endif +#ifdef HAVE_BZLIB_H +#include +#endif +#ifdef HAVE_LZ4_H +#include +#endif +#ifdef HAVE_ZSTD_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_string.h" + +const char * +archive_version_details(void) +{ + static struct archive_string str; + static int init = 0; + const char *zlib = archive_zlib_version(); + const char *liblzma = archive_liblzma_version(); + const char *bzlib = archive_bzlib_version(); + const char *liblz4 = archive_liblz4_version(); + const char *libzstd = archive_libzstd_version(); + + if (!init) { + archive_string_init(&str); + + archive_strcat(&str, ARCHIVE_VERSION_STRING); + if (zlib != NULL) { + archive_strcat(&str, " zlib/"); + archive_strcat(&str, zlib); + } + if (liblzma) { + archive_strcat(&str, " liblzma/"); + archive_strcat(&str, liblzma); + } + if (bzlib) { + const char *p = bzlib; + const char *sep = strchr(p, ','); + if (sep == NULL) + sep = p + strlen(p); + archive_strcat(&str, " bz2lib/"); + archive_strncat(&str, p, sep - p); + } + if (liblz4) { + archive_strcat(&str, " liblz4/"); + archive_strcat(&str, liblz4); + } + if (libzstd) { + archive_strcat(&str, " libzstd/"); + archive_strcat(&str, libzstd); + } + } + return str.s; +} + +const char * +archive_zlib_version(void) +{ +#ifdef HAVE_ZLIB_H + return ZLIB_VERSION; +#else + return NULL; +#endif +} + +const char * +archive_liblzma_version(void) +{ +#ifdef HAVE_LZMA_H + return LZMA_VERSION_STRING; +#else + return NULL; +#endif +} + +const char * +archive_bzlib_version(void) +{ +#ifdef HAVE_BZLIB_H + return BZ2_bzlibVersion(); +#else + return NULL; +#endif +} + +const char * +archive_liblz4_version(void) +{ +#if defined(HAVE_LZ4_H) && defined(HAVE_LIBLZ4) +#define str(s) #s +#define NUMBER(x) str(x) + return NUMBER(LZ4_VERSION_MAJOR) "." NUMBER(LZ4_VERSION_MINOR) "." NUMBER(LZ4_VERSION_RELEASE); +#undef NUMBER +#undef str +#else + return NULL; +#endif +} + +const char * +archive_libzstd_version(void) +{ +#if HAVE_ZSTD_H && HAVE_LIBZSTD + return ZSTD_VERSION_STRING; +#else + return NULL; +#endif +} diff --git a/c/archive_virtual.c b/c/archive_virtual.c new file mode 100644 --- /dev/null +++ b/c/archive_virtual.c @@ -0,0 +1,163 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_virtual.c 201098 2009-12-28 02:58:14Z kientzle $"); + +#include "archive.h" +#include "archive_entry.h" +#include "archive_private.h" + +int +archive_filter_code(struct archive *a, int n) +{ + return ((a->vtable->archive_filter_code)(a, n)); +} + +int +archive_filter_count(struct archive *a) +{ + return ((a->vtable->archive_filter_count)(a)); +} + +const char * +archive_filter_name(struct archive *a, int n) +{ + return ((a->vtable->archive_filter_name)(a, n)); +} + +la_int64_t +archive_filter_bytes(struct archive *a, int n) +{ + return ((a->vtable->archive_filter_bytes)(a, n)); +} + +int +archive_free(struct archive *a) +{ + if (a == NULL) + return (ARCHIVE_OK); + return ((a->vtable->archive_free)(a)); +} + +int +archive_write_close(struct archive *a) +{ + return ((a->vtable->archive_close)(a)); +} + +int +archive_read_close(struct archive *a) +{ + return ((a->vtable->archive_close)(a)); +} + +int +archive_write_fail(struct archive *a) +{ + a->state = ARCHIVE_STATE_FATAL; + return a->state; +} + +int +archive_write_free(struct archive *a) +{ + return archive_free(a); +} + +#if ARCHIVE_VERSION_NUMBER < 4000000 +/* For backwards compatibility; will be removed with libarchive 4.0. */ +int +archive_write_finish(struct archive *a) +{ + return archive_write_free(a); +} +#endif + +int +archive_read_free(struct archive *a) +{ + return archive_free(a); +} + +#if ARCHIVE_VERSION_NUMBER < 4000000 +/* For backwards compatibility; will be removed with libarchive 4.0. */ +int +archive_read_finish(struct archive *a) +{ + return archive_read_free(a); +} +#endif + +int +archive_write_header(struct archive *a, struct archive_entry *entry) +{ + ++a->file_count; + return ((a->vtable->archive_write_header)(a, entry)); +} + +int +archive_write_finish_entry(struct archive *a) +{ + return ((a->vtable->archive_write_finish_entry)(a)); +} + +la_ssize_t +archive_write_data(struct archive *a, const void *buff, size_t s) +{ + return ((a->vtable->archive_write_data)(a, buff, s)); +} + +la_ssize_t +archive_write_data_block(struct archive *a, const void *buff, size_t s, + la_int64_t o) +{ + if (a->vtable->archive_write_data_block == NULL) { + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "archive_write_data_block not supported"); + a->state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + } + return ((a->vtable->archive_write_data_block)(a, buff, s, o)); +} + +int +archive_read_next_header(struct archive *a, struct archive_entry **entry) +{ + return ((a->vtable->archive_read_next_header)(a, entry)); +} + +int +archive_read_next_header2(struct archive *a, struct archive_entry *entry) +{ + return ((a->vtable->archive_read_next_header2)(a, entry)); +} + +int +archive_read_data_block(struct archive *a, + const void **buff, size_t *s, la_int64_t *o) +{ + return ((a->vtable->archive_read_data_block)(a, buff, s, o)); +} diff --git a/c/archive_windows.c b/c/archive_windows.c new file mode 100644 --- /dev/null +++ b/c/archive_windows.c @@ -0,0 +1,909 @@ +/*- + * Copyright (c) 2009-2011 Michihiro NAKAJIMA + * Copyright (c) 2003-2007 Kees Zeelenberg + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD$ + */ + +/* + * A set of compatibility glue for building libarchive on Windows platforms. + * + * Originally created as "libarchive-nonposix.c" by Kees Zeelenberg + * for the GnuWin32 project, trimmed significantly by Tim Kientzle. + * + * Much of the original file was unnecessary for libarchive, because + * many of the features it emulated were not strictly necessary for + * libarchive. I hope for this to shrink further as libarchive + * internals are gradually reworked to sit more naturally on both + * POSIX and Windows. Any ideas for this are greatly appreciated. + * + * The biggest remaining issue is the dev/ino emulation; libarchive + * has a couple of public APIs that rely on dev/ino uniquely + * identifying a file. This doesn't match well with Windows. I'm + * considering alternative APIs. + */ + +#if defined(_WIN32) && !defined(__CYGWIN__) + +#include "archive_platform.h" +#include "archive_private.h" +#include "archive_entry.h" +#include +#include +#include +#ifdef HAVE_SYS_UTIME_H +#include +#endif +#include +#include +#include +#include +#include +#include +#include + +#define EPOC_TIME ARCHIVE_LITERAL_ULL(116444736000000000) + +#if defined(__LA_LSEEK_NEEDED) +static BOOL SetFilePointerEx_perso(HANDLE hFile, + LARGE_INTEGER liDistanceToMove, + PLARGE_INTEGER lpNewFilePointer, + DWORD dwMoveMethod) +{ + LARGE_INTEGER li; + li.QuadPart = liDistanceToMove.QuadPart; + li.LowPart = SetFilePointer( + hFile, li.LowPart, &li.HighPart, dwMoveMethod); + if(lpNewFilePointer) { + lpNewFilePointer->QuadPart = li.QuadPart; + } + return li.LowPart != -1 || GetLastError() == NO_ERROR; +} +#endif + +struct ustat { + int64_t st_atime; + uint32_t st_atime_nsec; + int64_t st_ctime; + uint32_t st_ctime_nsec; + int64_t st_mtime; + uint32_t st_mtime_nsec; + gid_t st_gid; + /* 64bits ino */ + int64_t st_ino; + mode_t st_mode; + uint32_t st_nlink; + uint64_t st_size; + uid_t st_uid; + dev_t st_dev; + dev_t st_rdev; +}; + +/* Transform 64-bits ino into 32-bits by hashing. + * You do not forget that really unique number size is 64-bits. + */ +#define INOSIZE (8*sizeof(ino_t)) /* 32 */ +static __inline ino_t +getino(struct ustat *ub) +{ + ULARGE_INTEGER ino64; + ino64.QuadPart = ub->st_ino; + /* I don't know this hashing is correct way */ + return ((ino_t)(ino64.LowPart ^ (ino64.LowPart >> INOSIZE))); +} + +/* + * Prepend "\\?\" to the path name and convert it to unicode to permit + * an extended-length path for a maximum total path length of 32767 + * characters. + * see also http://msdn.microsoft.com/en-us/library/aa365247.aspx + */ +wchar_t * +__la_win_permissive_name(const char *name) +{ + wchar_t *wn; + wchar_t *ws; + size_t ll; + + ll = strlen(name); + wn = malloc((ll + 1) * sizeof(wchar_t)); + if (wn == NULL) + return (NULL); + ll = mbstowcs(wn, name, ll); + if (ll == (size_t)-1) { + free(wn); + return (NULL); + } + wn[ll] = L'\0'; + ws = __la_win_permissive_name_w(wn); + free(wn); + return (ws); +} + +wchar_t * +__la_win_permissive_name_w(const wchar_t *wname) +{ + wchar_t *wn, *wnp; + wchar_t *ws, *wsp; + DWORD l, len, slen; + int unc; + + /* Get a full-pathname. */ + l = GetFullPathNameW(wname, 0, NULL, NULL); + if (l == 0) + return (NULL); + /* NOTE: GetFullPathNameW has a bug that if the length of the file + * name is just 1 then it returns incomplete buffer size. Thus, we + * have to add three to the size to allocate a sufficient buffer + * size for the full-pathname of the file name. */ + l += 3; + wnp = malloc(l * sizeof(wchar_t)); + if (wnp == NULL) + return (NULL); + len = GetFullPathNameW(wname, l, wnp, NULL); + wn = wnp; + + if (wnp[0] == L'\\' && wnp[1] == L'\\' && + wnp[2] == L'?' && wnp[3] == L'\\') + /* We have already a permissive name. */ + return (wn); + + if (wnp[0] == L'\\' && wnp[1] == L'\\' && + wnp[2] == L'.' && wnp[3] == L'\\') { + /* This is a device name */ + if (((wnp[4] >= L'a' && wnp[4] <= L'z') || + (wnp[4] >= L'A' && wnp[4] <= L'Z')) && + wnp[5] == L':' && wnp[6] == L'\\') + wnp[2] = L'?';/* Not device name. */ + return (wn); + } + + unc = 0; + if (wnp[0] == L'\\' && wnp[1] == L'\\' && wnp[2] != L'\\') { + wchar_t *p = &wnp[2]; + + /* Skip server-name letters. */ + while (*p != L'\\' && *p != L'\0') + ++p; + if (*p == L'\\') { + wchar_t *rp = ++p; + /* Skip share-name letters. */ + while (*p != L'\\' && *p != L'\0') + ++p; + if (*p == L'\\' && p != rp) { + /* Now, match patterns such as + * "\\server-name\share-name\" */ + wnp += 2; + len -= 2; + unc = 1; + } + } + } + + slen = 4 + (unc * 4) + len + 1; + ws = wsp = malloc(slen * sizeof(wchar_t)); + if (ws == NULL) { + free(wn); + return (NULL); + } + /* prepend "\\?\" */ + wcsncpy(wsp, L"\\\\?\\", 4); + wsp += 4; + slen -= 4; + if (unc) { + /* append "UNC\" ---> "\\?\UNC\" */ + wcsncpy(wsp, L"UNC\\", 4); + wsp += 4; + slen -= 4; + } + wcsncpy(wsp, wnp, slen); + wsp[slen - 1] = L'\0'; /* Ensure null termination. */ + free(wn); + return (ws); +} + +/* + * Create a file handle. + * This can exceed MAX_PATH limitation. + */ +static HANDLE +la_CreateFile(const char *path, DWORD dwDesiredAccess, DWORD dwShareMode, + LPSECURITY_ATTRIBUTES lpSecurityAttributes, DWORD dwCreationDisposition, + DWORD dwFlagsAndAttributes, HANDLE hTemplateFile) +{ + wchar_t *wpath; + HANDLE handle; + + handle = CreateFileA(path, dwDesiredAccess, dwShareMode, + lpSecurityAttributes, dwCreationDisposition, dwFlagsAndAttributes, + hTemplateFile); + if (handle != INVALID_HANDLE_VALUE) + return (handle); + if (GetLastError() != ERROR_PATH_NOT_FOUND) + return (handle); + wpath = __la_win_permissive_name(path); + if (wpath == NULL) + return (handle); + handle = CreateFileW(wpath, dwDesiredAccess, dwShareMode, + lpSecurityAttributes, dwCreationDisposition, dwFlagsAndAttributes, + hTemplateFile); + free(wpath); + return (handle); +} + +#if defined(__LA_LSEEK_NEEDED) +__int64 +__la_lseek(int fd, __int64 offset, int whence) +{ + LARGE_INTEGER distance; + LARGE_INTEGER newpointer; + HANDLE handle; + + if (fd < 0) { + errno = EBADF; + return (-1); + } + handle = (HANDLE)_get_osfhandle(fd); + if (GetFileType(handle) != FILE_TYPE_DISK) { + errno = EBADF; + return (-1); + } + distance.QuadPart = offset; + if (!SetFilePointerEx_perso(handle, distance, &newpointer, whence)) { + DWORD lasterr; + + lasterr = GetLastError(); + if (lasterr == ERROR_BROKEN_PIPE) + return (0); + if (lasterr == ERROR_ACCESS_DENIED) + errno = EBADF; + else + la_dosmaperr(lasterr); + return (-1); + } + return (newpointer.QuadPart); +} +#endif + +/* This can exceed MAX_PATH limitation. */ +int +__la_open(const char *path, int flags, ...) +{ + va_list ap; + wchar_t *ws; + int r, pmode; + DWORD attr; + + va_start(ap, flags); + pmode = va_arg(ap, int); + va_end(ap); + ws = NULL; + if ((flags & ~O_BINARY) == O_RDONLY) { + /* + * When we open a directory, _open function returns + * "Permission denied" error. + */ + attr = GetFileAttributesA(path); + if (attr == (DWORD)-1 && GetLastError() == ERROR_PATH_NOT_FOUND) { + ws = __la_win_permissive_name(path); + if (ws == NULL) { + errno = EINVAL; + return (-1); + } + attr = GetFileAttributesW(ws); + } + if (attr == (DWORD)-1) { + la_dosmaperr(GetLastError()); + free(ws); + return (-1); + } + if (attr & FILE_ATTRIBUTE_DIRECTORY) { + HANDLE handle; + + if (ws != NULL) + handle = CreateFileW(ws, 0, 0, NULL, + OPEN_EXISTING, + FILE_FLAG_BACKUP_SEMANTICS | + FILE_ATTRIBUTE_READONLY, + NULL); + else + handle = CreateFileA(path, 0, 0, NULL, + OPEN_EXISTING, + FILE_FLAG_BACKUP_SEMANTICS | + FILE_ATTRIBUTE_READONLY, + NULL); + free(ws); + if (handle == INVALID_HANDLE_VALUE) { + la_dosmaperr(GetLastError()); + return (-1); + } + r = _open_osfhandle((intptr_t)handle, _O_RDONLY); + return (r); + } + } + if (ws == NULL) { +#if defined(__BORLANDC__) + /* Borland has no mode argument. + TODO: Fix mode of new file. */ + r = _open(path, flags); +#else + r = _open(path, flags, pmode); +#endif + if (r < 0 && errno == EACCES && (flags & O_CREAT) != 0) { + /* Simulate other POSIX system action to pass our test suite. */ + attr = GetFileAttributesA(path); + if (attr == (DWORD)-1) + la_dosmaperr(GetLastError()); + else if (attr & FILE_ATTRIBUTE_DIRECTORY) + errno = EISDIR; + else + errno = EACCES; + return (-1); + } + if (r >= 0 || errno != ENOENT) + return (r); + ws = __la_win_permissive_name(path); + if (ws == NULL) { + errno = EINVAL; + return (-1); + } + } + r = _wopen(ws, flags, pmode); + if (r < 0 && errno == EACCES && (flags & O_CREAT) != 0) { + /* Simulate other POSIX system action to pass our test suite. */ + attr = GetFileAttributesW(ws); + if (attr == (DWORD)-1) + la_dosmaperr(GetLastError()); + else if (attr & FILE_ATTRIBUTE_DIRECTORY) + errno = EISDIR; + else + errno = EACCES; + } + free(ws); + return (r); +} + +ssize_t +__la_read(int fd, void *buf, size_t nbytes) +{ + HANDLE handle; + DWORD bytes_read, lasterr; + int r; + +#ifdef _WIN64 + if (nbytes > UINT32_MAX) + nbytes = UINT32_MAX; +#endif + if (fd < 0) { + errno = EBADF; + return (-1); + } + /* Do not pass 0 to third parameter of ReadFile(), read bytes. + * This will not return to application side. */ + if (nbytes == 0) + return (0); + handle = (HANDLE)_get_osfhandle(fd); + r = ReadFile(handle, buf, (uint32_t)nbytes, + &bytes_read, NULL); + if (r == 0) { + lasterr = GetLastError(); + if (lasterr == ERROR_NO_DATA) { + errno = EAGAIN; + return (-1); + } + if (lasterr == ERROR_BROKEN_PIPE) + return (0); + if (lasterr == ERROR_ACCESS_DENIED) + errno = EBADF; + else + la_dosmaperr(lasterr); + return (-1); + } + return ((ssize_t)bytes_read); +} + +/* Convert Windows FILETIME to UTC */ +__inline static void +fileTimeToUTC(const FILETIME *filetime, time_t *t, long *ns) +{ + ULARGE_INTEGER utc; + + utc.HighPart = filetime->dwHighDateTime; + utc.LowPart = filetime->dwLowDateTime; + if (utc.QuadPart >= EPOC_TIME) { + utc.QuadPart -= EPOC_TIME; + *t = (time_t)(utc.QuadPart / 10000000); /* milli seconds base */ + *ns = (long)(utc.QuadPart % 10000000) * 100;/* nano seconds base */ + } else { + *t = 0; + *ns = 0; + } +} + +/* Stat by handle + * Windows' stat() does not accept the path added "\\?\" especially "?" + * character. + * It means we cannot access the long name path longer than MAX_PATH. + * So I've implemented a function similar to Windows' stat() to access the + * long name path. + * And I've added some feature. + * 1. set st_ino by nFileIndexHigh and nFileIndexLow of + * BY_HANDLE_FILE_INFORMATION. + * 2. set st_nlink by nNumberOfLinks of BY_HANDLE_FILE_INFORMATION. + * 3. set st_dev by dwVolumeSerialNumber by BY_HANDLE_FILE_INFORMATION. + */ +static int +__hstat(HANDLE handle, struct ustat *st) +{ + BY_HANDLE_FILE_INFORMATION info; + ULARGE_INTEGER ino64; + DWORD ftype; + mode_t mode; + time_t t; + long ns; + + switch (ftype = GetFileType(handle)) { + case FILE_TYPE_UNKNOWN: + errno = EBADF; + return (-1); + case FILE_TYPE_CHAR: + case FILE_TYPE_PIPE: + if (ftype == FILE_TYPE_CHAR) { + st->st_mode = S_IFCHR; + st->st_size = 0; + } else { + DWORD avail; + + st->st_mode = S_IFIFO; + if (PeekNamedPipe(handle, NULL, 0, NULL, &avail, NULL)) + st->st_size = avail; + else + st->st_size = 0; + } + st->st_atime = 0; + st->st_atime_nsec = 0; + st->st_mtime = 0; + st->st_mtime_nsec = 0; + st->st_ctime = 0; + st->st_ctime_nsec = 0; + st->st_ino = 0; + st->st_nlink = 1; + st->st_uid = 0; + st->st_gid = 0; + st->st_rdev = 0; + st->st_dev = 0; + return (0); + case FILE_TYPE_DISK: + break; + default: + /* This ftype is undocumented type. */ + la_dosmaperr(GetLastError()); + return (-1); + } + + ZeroMemory(&info, sizeof(info)); + if (!GetFileInformationByHandle (handle, &info)) { + la_dosmaperr(GetLastError()); + return (-1); + } + + mode = S_IRUSR | S_IRGRP | S_IROTH; + if ((info.dwFileAttributes & FILE_ATTRIBUTE_READONLY) == 0) + mode |= S_IWUSR | S_IWGRP | S_IWOTH; + if (info.dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY) + mode |= S_IFDIR | S_IXUSR | S_IXGRP | S_IXOTH; + else + mode |= S_IFREG; + st->st_mode = mode; + + fileTimeToUTC(&info.ftLastAccessTime, &t, &ns); + st->st_atime = t; + st->st_atime_nsec = ns; + fileTimeToUTC(&info.ftLastWriteTime, &t, &ns); + st->st_mtime = t; + st->st_mtime_nsec = ns; + fileTimeToUTC(&info.ftCreationTime, &t, &ns); + st->st_ctime = t; + st->st_ctime_nsec = ns; + st->st_size = + ((int64_t)(info.nFileSizeHigh) * ((int64_t)MAXDWORD + 1)) + + (int64_t)(info.nFileSizeLow); +#ifdef SIMULATE_WIN_STAT + st->st_ino = 0; + st->st_nlink = 1; + st->st_dev = 0; +#else + /* Getting FileIndex as i-node. We should remove a sequence which + * is high-16-bits of nFileIndexHigh. */ + ino64.HighPart = info.nFileIndexHigh & 0x0000FFFFUL; + ino64.LowPart = info.nFileIndexLow; + st->st_ino = ino64.QuadPart; + st->st_nlink = info.nNumberOfLinks; + if (info.dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY) + ++st->st_nlink;/* Add parent directory. */ + st->st_dev = info.dwVolumeSerialNumber; +#endif + st->st_uid = 0; + st->st_gid = 0; + st->st_rdev = 0; + return (0); +} + +static void +copy_stat(struct stat *st, struct ustat *us) +{ + st->st_atime = us->st_atime; + st->st_ctime = us->st_ctime; + st->st_mtime = us->st_mtime; + st->st_gid = us->st_gid; + st->st_ino = getino(us); + st->st_mode = us->st_mode; + st->st_nlink = us->st_nlink; + st->st_size = (off_t)us->st_size; + st->st_uid = us->st_uid; + st->st_dev = us->st_dev; + st->st_rdev = us->st_rdev; +} + +/* + * TODO: Remove a use of __la_fstat and __la_stat. + * We should use GetFileInformationByHandle in place + * where We still use the *stat functions. + */ +int +__la_fstat(int fd, struct stat *st) +{ + struct ustat u; + int ret; + + if (fd < 0) { + errno = EBADF; + return (-1); + } + ret = __hstat((HANDLE)_get_osfhandle(fd), &u); + if (ret >= 0) { + copy_stat(st, &u); + if (u.st_mode & (S_IFCHR | S_IFIFO)) { + st->st_dev = fd; + st->st_rdev = fd; + } + } + return (ret); +} + +/* This can exceed MAX_PATH limitation. */ +int +__la_stat(const char *path, struct stat *st) +{ + HANDLE handle; + struct ustat u; + int ret; + + handle = la_CreateFile(path, 0, FILE_SHARE_READ, NULL, OPEN_EXISTING, + FILE_FLAG_BACKUP_SEMANTICS, + NULL); + if (handle == INVALID_HANDLE_VALUE) { + la_dosmaperr(GetLastError()); + return (-1); + } + ret = __hstat(handle, &u); + CloseHandle(handle); + if (ret >= 0) { + char *p; + + copy_stat(st, &u); + p = strrchr(path, '.'); + if (p != NULL && strlen(p) == 4) { + char exttype[4]; + + ++ p; + exttype[0] = toupper(*p++); + exttype[1] = toupper(*p++); + exttype[2] = toupper(*p++); + exttype[3] = '\0'; + if (!strcmp(exttype, "EXE") || !strcmp(exttype, "CMD") || + !strcmp(exttype, "BAT") || !strcmp(exttype, "COM")) + st->st_mode |= S_IXUSR | S_IXGRP | S_IXOTH; + } + } + return (ret); +} + +/* + * This waitpid is limited implementation. + */ +pid_t +__la_waitpid(HANDLE child, int *status, int option) +{ + DWORD cs; + + (void)option;/* UNUSED */ + do { + if (GetExitCodeProcess(child, &cs) == 0) { + CloseHandle(child); + la_dosmaperr(GetLastError()); + *status = 0; + return (-1); + } + } while (cs == STILL_ACTIVE); + + *status = (int)(cs & 0xff); + return (0); +} + +ssize_t +__la_write(int fd, const void *buf, size_t nbytes) +{ + DWORD bytes_written; + +#ifdef _WIN64 + if (nbytes > UINT32_MAX) + nbytes = UINT32_MAX; +#endif + if (fd < 0) { + errno = EBADF; + return (-1); + } + if (!WriteFile((HANDLE)_get_osfhandle(fd), buf, (uint32_t)nbytes, + &bytes_written, NULL)) { + DWORD lasterr; + + lasterr = GetLastError(); + if (lasterr == ERROR_ACCESS_DENIED) + errno = EBADF; + else + la_dosmaperr(lasterr); + return (-1); + } + return (bytes_written); +} + +/* + * Replace the Windows path separator '\' with '/'. + */ +static int +replace_pathseparator(struct archive_wstring *ws, const wchar_t *wp) +{ + wchar_t *w; + size_t path_length; + + if (wp == NULL) + return(0); + if (wcschr(wp, L'\\') == NULL) + return(0); + path_length = wcslen(wp); + if (archive_wstring_ensure(ws, path_length) == NULL) + return(-1); + archive_wstrncpy(ws, wp, path_length); + for (w = ws->s; *w; w++) { + if (*w == L'\\') + *w = L'/'; + } + return(1); +} + +static int +fix_pathseparator(struct archive_entry *entry) +{ + struct archive_wstring ws; + const wchar_t *wp; + int ret = ARCHIVE_OK; + + archive_string_init(&ws); + wp = archive_entry_pathname_w(entry); + switch (replace_pathseparator(&ws, wp)) { + case 0: /* Not replaced. */ + break; + case 1: /* Replaced. */ + archive_entry_copy_pathname_w(entry, ws.s); + break; + default: + ret = ARCHIVE_FAILED; + } + wp = archive_entry_hardlink_w(entry); + switch (replace_pathseparator(&ws, wp)) { + case 0: /* Not replaced. */ + break; + case 1: /* Replaced. */ + archive_entry_copy_hardlink_w(entry, ws.s); + break; + default: + ret = ARCHIVE_FAILED; + } + wp = archive_entry_symlink_w(entry); + switch (replace_pathseparator(&ws, wp)) { + case 0: /* Not replaced. */ + break; + case 1: /* Replaced. */ + archive_entry_copy_symlink_w(entry, ws.s); + break; + default: + ret = ARCHIVE_FAILED; + } + archive_wstring_free(&ws); + return(ret); +} + +struct archive_entry * +__la_win_entry_in_posix_pathseparator(struct archive_entry *entry) +{ + struct archive_entry *entry_main; + const wchar_t *wp; + int has_backslash = 0; + int ret; + + wp = archive_entry_pathname_w(entry); + if (wp != NULL && wcschr(wp, L'\\') != NULL) + has_backslash = 1; + if (!has_backslash) { + wp = archive_entry_hardlink_w(entry); + if (wp != NULL && wcschr(wp, L'\\') != NULL) + has_backslash = 1; + } + if (!has_backslash) { + wp = archive_entry_symlink_w(entry); + if (wp != NULL && wcschr(wp, L'\\') != NULL) + has_backslash = 1; + } + /* + * If there is no backslash chars, return the original. + */ + if (!has_backslash) + return (entry); + + /* Copy entry so we can modify it as needed. */ + entry_main = archive_entry_clone(entry); + if (entry_main == NULL) + return (NULL); + /* Replace the Windows path-separator '\' with '/'. */ + ret = fix_pathseparator(entry_main); + if (ret < ARCHIVE_WARN) { + archive_entry_free(entry_main); + return (NULL); + } + return (entry_main); +} + + +/* + * The following function was modified from PostgreSQL sources and is + * subject to the copyright below. + */ +/*------------------------------------------------------------------------- + * + * win32error.c + * Map win32 error codes to errno values + * + * Portions Copyright (c) 1996-2008, PostgreSQL Global Development Group + * + * IDENTIFICATION + * $PostgreSQL: pgsql/src/port/win32error.c,v 1.4 2008/01/01 19:46:00 momjian Exp $ + * + *------------------------------------------------------------------------- + */ +/* +PostgreSQL Database Management System +(formerly known as Postgres, then as Postgres95) + +Portions Copyright (c) 1996-2008, PostgreSQL Global Development Group + +Portions Copyright (c) 1994, The Regents of the University of California + +Permission to use, copy, modify, and distribute this software and its +documentation for any purpose, without fee, and without a written agreement +is hereby granted, provided that the above copyright notice and this +paragraph and the following two paragraphs appear in all copies. + +IN NO EVENT SHALL THE UNIVERSITY OF CALIFORNIA BE LIABLE TO ANY PARTY FOR +DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES, INCLUDING +LOST PROFITS, ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS +DOCUMENTATION, EVEN IF THE UNIVERSITY OF CALIFORNIA HAS BEEN ADVISED OF THE +POSSIBILITY OF SUCH DAMAGE. + +THE UNIVERSITY OF CALIFORNIA SPECIFICALLY DISCLAIMS ANY WARRANTIES, +INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY +AND FITNESS FOR A PARTICULAR PURPOSE. THE SOFTWARE PROVIDED HEREUNDER IS +ON AN "AS IS" BASIS, AND THE UNIVERSITY OF CALIFORNIA HAS NO OBLIGATIONS TO +PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS. +*/ + +static const struct { + DWORD winerr; + int doserr; +} doserrors[] = +{ + { ERROR_INVALID_FUNCTION, EINVAL }, + { ERROR_FILE_NOT_FOUND, ENOENT }, + { ERROR_PATH_NOT_FOUND, ENOENT }, + { ERROR_TOO_MANY_OPEN_FILES, EMFILE }, + { ERROR_ACCESS_DENIED, EACCES }, + { ERROR_INVALID_HANDLE, EBADF }, + { ERROR_ARENA_TRASHED, ENOMEM }, + { ERROR_NOT_ENOUGH_MEMORY, ENOMEM }, + { ERROR_INVALID_BLOCK, ENOMEM }, + { ERROR_BAD_ENVIRONMENT, E2BIG }, + { ERROR_BAD_FORMAT, ENOEXEC }, + { ERROR_INVALID_ACCESS, EINVAL }, + { ERROR_INVALID_DATA, EINVAL }, + { ERROR_INVALID_DRIVE, ENOENT }, + { ERROR_CURRENT_DIRECTORY, EACCES }, + { ERROR_NOT_SAME_DEVICE, EXDEV }, + { ERROR_NO_MORE_FILES, ENOENT }, + { ERROR_LOCK_VIOLATION, EACCES }, + { ERROR_SHARING_VIOLATION, EACCES }, + { ERROR_BAD_NETPATH, ENOENT }, + { ERROR_NETWORK_ACCESS_DENIED, EACCES }, + { ERROR_BAD_NET_NAME, ENOENT }, + { ERROR_FILE_EXISTS, EEXIST }, + { ERROR_CANNOT_MAKE, EACCES }, + { ERROR_FAIL_I24, EACCES }, + { ERROR_INVALID_PARAMETER, EINVAL }, + { ERROR_NO_PROC_SLOTS, EAGAIN }, + { ERROR_DRIVE_LOCKED, EACCES }, + { ERROR_BROKEN_PIPE, EPIPE }, + { ERROR_DISK_FULL, ENOSPC }, + { ERROR_INVALID_TARGET_HANDLE, EBADF }, + { ERROR_INVALID_HANDLE, EINVAL }, + { ERROR_WAIT_NO_CHILDREN, ECHILD }, + { ERROR_CHILD_NOT_COMPLETE, ECHILD }, + { ERROR_DIRECT_ACCESS_HANDLE, EBADF }, + { ERROR_NEGATIVE_SEEK, EINVAL }, + { ERROR_SEEK_ON_DEVICE, EACCES }, + { ERROR_DIR_NOT_EMPTY, ENOTEMPTY }, + { ERROR_NOT_LOCKED, EACCES }, + { ERROR_BAD_PATHNAME, ENOENT }, + { ERROR_MAX_THRDS_REACHED, EAGAIN }, + { ERROR_LOCK_FAILED, EACCES }, + { ERROR_ALREADY_EXISTS, EEXIST }, + { ERROR_FILENAME_EXCED_RANGE, ENOENT }, + { ERROR_NESTING_NOT_ALLOWED, EAGAIN }, + { ERROR_NOT_ENOUGH_QUOTA, ENOMEM } +}; + +void +__la_dosmaperr(unsigned long e) +{ + int i; + + if (e == 0) + { + errno = 0; + return; + } + + for (i = 0; i < (int)(sizeof(doserrors)/sizeof(doserrors[0])); i++) + { + if (doserrors[i].winerr == e) + { + errno = doserrors[i].doserr; + return; + } + } + + /* fprintf(stderr, "unrecognized win32 error code: %lu", e); */ + errno = EINVAL; + return; +} + +#endif /* _WIN32 && !__CYGWIN__ */ diff --git a/c/archive_windows.h b/c/archive_windows.h new file mode 100644 --- /dev/null +++ b/c/archive_windows.h @@ -0,0 +1,315 @@ +/*- + * Copyright (c) 2009-2011 Michihiro NAKAJIMA + * Copyright (c) 2003-2006 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer + * in this position and unchanged. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD$ + */ + +/* + * TODO: A lot of stuff in here isn't actually used by libarchive and + * can be trimmed out. Note that this file is used by libarchive and + * libarchive_test but nowhere else. (But note that it gets compiled + * with many different Windows environments, including MinGW, Visual + * Studio, and Cygwin. Significant changes should be tested in all three.) + */ + +/* + * TODO: Don't use off_t in here. Use __int64 instead. Note that + * Visual Studio and the Windows SDK define off_t as 32 bits; Win32's + * more modern file handling APIs all use __int64 instead of off_t. + */ + +#ifndef LIBARCHIVE_ARCHIVE_WINDOWS_H_INCLUDED +#define LIBARCHIVE_ARCHIVE_WINDOWS_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif + +/* Start of configuration for native Win32 */ +#ifndef MINGW_HAS_SECURE_API +#define MINGW_HAS_SECURE_API 1 +#endif + +#include +#define set_errno(val) ((errno)=val) +#include +#include //brings in NULL +#if defined(HAVE_STDINT_H) +#include +#endif +#include +#include +#include +#include +#include +#if defined(__MINGW32__) && defined(HAVE_UNISTD_H) +/* Prevent build error from a type mismatch of ftruncate(). + * This unistd.h defines it as ftruncate(int, off_t). */ +#include +#endif +#define NOCRYPT +#include +//#define EFTYPE 7 + +#if defined(__BORLANDC__) +#pragma warn -8068 /* Constant out of range in comparison. */ +#pragma warn -8072 /* Suspicious pointer arithmetic. */ +#endif + +#ifndef NULL +#ifdef __cplusplus +#define NULL 0 +#else +#define NULL ((void *)0) +#endif +#endif + +/* Alias the Windows _function to the POSIX equivalent. */ +#define close _close +#define fcntl(fd, cmd, flg) /* No operation. */ +#ifndef fileno +#define fileno _fileno +#endif +#ifdef fstat +#undef fstat +#endif +#define fstat __la_fstat +#if !defined(__BORLANDC__) +#ifdef lseek +#undef lseek +#endif +#define lseek _lseeki64 +#else +#define lseek __la_lseek +#define __LA_LSEEK_NEEDED +#endif +#define lstat __la_stat +#define open __la_open +#define read __la_read +#if !defined(__BORLANDC__) && !defined(__WATCOMC__) +#define setmode _setmode +#endif +#define la_stat(path,stref) __la_stat(path,stref) +#if !defined(__WATCOMC__) +#if !defined(__BORLANDC__) +#define strdup _strdup +#endif +#define tzset _tzset +#if !defined(__BORLANDC__) +#define umask _umask +#endif +#endif +#define waitpid __la_waitpid +#define write __la_write + +#if !defined(__WATCOMC__) + +#ifndef O_RDONLY +#define O_RDONLY _O_RDONLY +#define O_WRONLY _O_WRONLY +#define O_TRUNC _O_TRUNC +#define O_CREAT _O_CREAT +#define O_EXCL _O_EXCL +#define O_BINARY _O_BINARY +#endif + +#ifndef _S_IFIFO + #define _S_IFIFO 0010000 /* pipe */ +#endif +#ifndef _S_IFCHR + #define _S_IFCHR 0020000 /* character special */ +#endif +#ifndef _S_IFDIR + #define _S_IFDIR 0040000 /* directory */ +#endif +#ifndef _S_IFBLK + #define _S_IFBLK 0060000 /* block special */ +#endif +#ifndef _S_IFLNK + #define _S_IFLNK 0120000 /* symbolic link */ +#endif +#ifndef _S_IFSOCK + #define _S_IFSOCK 0140000 /* socket */ +#endif +#ifndef _S_IFREG + #define _S_IFREG 0100000 /* regular */ +#endif +#ifndef _S_IFMT + #define _S_IFMT 0170000 /* file type mask */ +#endif + +#ifndef S_IFIFO +#define S_IFIFO _S_IFIFO +#endif +//#define S_IFCHR _S_IFCHR +//#define S_IFDIR _S_IFDIR +#ifndef S_IFBLK +#define S_IFBLK _S_IFBLK +#endif +#ifndef S_IFLNK +#define S_IFLNK _S_IFLNK +#endif +#ifndef S_IFSOCK +#define S_IFSOCK _S_IFSOCK +#endif +//#define S_IFREG _S_IFREG +//#define S_IFMT _S_IFMT + +#ifndef S_ISBLK +#define S_ISBLK(m) (((m) & S_IFMT) == S_IFBLK) /* block special */ +#define S_ISFIFO(m) (((m) & S_IFMT) == S_IFIFO) /* fifo or socket */ +#define S_ISCHR(m) (((m) & S_IFMT) == S_IFCHR) /* char special */ +#define S_ISDIR(m) (((m) & S_IFMT) == S_IFDIR) /* directory */ +#define S_ISREG(m) (((m) & S_IFMT) == S_IFREG) /* regular file */ +#endif +#define S_ISLNK(m) (((m) & S_IFMT) == S_IFLNK) /* Symbolic link */ +#define S_ISSOCK(m) (((m) & S_IFMT) == S_IFSOCK) /* Socket */ + +#define _S_ISUID 0004000 /* set user id on execution */ +#define _S_ISGID 0002000 /* set group id on execution */ +#define _S_ISVTX 0001000 /* save swapped text even after use */ + +#define S_ISUID _S_ISUID +#define S_ISGID _S_ISGID +#define S_ISVTX _S_ISVTX + +#define _S_IRWXU (_S_IREAD | _S_IWRITE | _S_IEXEC) +#define _S_IXUSR _S_IEXEC /* read permission, user */ +#define _S_IWUSR _S_IWRITE /* write permission, user */ +#define _S_IRUSR _S_IREAD /* execute/search permission, user */ +#define _S_IRWXG (_S_IRWXU >> 3) +#define _S_IXGRP (_S_IXUSR >> 3) /* read permission, group */ +#define _S_IWGRP (_S_IWUSR >> 3) /* write permission, group */ +#define _S_IRGRP (_S_IRUSR >> 3) /* execute/search permission, group */ +#define _S_IRWXO (_S_IRWXG >> 3) +#define _S_IXOTH (_S_IXGRP >> 3) /* read permission, other */ +#define _S_IWOTH (_S_IWGRP >> 3) /* write permission, other */ +#define _S_IROTH (_S_IRGRP >> 3) /* execute/search permission, other */ + +#ifndef S_IRWXU +#define S_IRWXU _S_IRWXU +#define S_IXUSR _S_IXUSR +#define S_IWUSR _S_IWUSR +#define S_IRUSR _S_IRUSR +#endif +#ifndef S_IRWXG +#define S_IRWXG _S_IRWXG +#define S_IXGRP _S_IXGRP +#define S_IWGRP _S_IWGRP +#endif +#ifndef S_IRGRP +#define S_IRGRP _S_IRGRP +#endif +#ifndef S_IRWXO +#define S_IRWXO _S_IRWXO +#define S_IXOTH _S_IXOTH +#define S_IWOTH _S_IWOTH +#define S_IROTH _S_IROTH +#endif + +#endif + +#define F_DUPFD 0 /* Duplicate file descriptor. */ +#define F_GETFD 1 /* Get file descriptor flags. */ +#define F_SETFD 2 /* Set file descriptor flags. */ +#define F_GETFL 3 /* Get file status flags. */ +#define F_SETFL 4 /* Set file status flags. */ +#define F_GETOWN 5 /* Get owner (receiver of SIGIO). */ +#define F_SETOWN 6 /* Set owner (receiver of SIGIO). */ +#define F_GETLK 7 /* Get record locking info. */ +#define F_SETLK 8 /* Set record locking info (non-blocking). */ +#define F_SETLKW 9 /* Set record locking info (blocking). */ + +/* XXX missing */ +#define F_GETLK64 7 /* Get record locking info. */ +#define F_SETLK64 8 /* Set record locking info (non-blocking). */ +#define F_SETLKW64 9 /* Set record locking info (blocking). */ + +/* File descriptor flags used with F_GETFD and F_SETFD. */ +#define FD_CLOEXEC 1 /* Close on exec. */ + +//NOT SURE IF O_NONBLOCK is OK here but at least the 0x0004 flag is not used by anything else... +#define O_NONBLOCK 0x0004 /* Non-blocking I/O. */ +//#define O_NDELAY O_NONBLOCK + +/* Symbolic constants for the access() function */ +#if !defined(F_OK) + #define R_OK 4 /* Test for read permission */ + #define W_OK 2 /* Test for write permission */ + #define X_OK 1 /* Test for execute permission */ + #define F_OK 0 /* Test for existence of file */ +#endif + + +/* Replacement POSIX function */ +extern int __la_fstat(int fd, struct stat *st); +extern int __la_lstat(const char *path, struct stat *st); +#if defined(__LA_LSEEK_NEEDED) +extern __int64 __la_lseek(int fd, __int64 offset, int whence); +#endif +extern int __la_open(const char *path, int flags, ...); +extern ssize_t __la_read(int fd, void *buf, size_t nbytes); +extern int __la_stat(const char *path, struct stat *st); +extern pid_t __la_waitpid(HANDLE child, int *status, int option); +extern ssize_t __la_write(int fd, const void *buf, size_t nbytes); + +#define _stat64i32(path, st) __la_stat(path, st) +#define _stat64(path, st) __la_stat(path, st) +/* for status returned by la_waitpid */ +#define WIFEXITED(sts) ((sts & 0x100) == 0) +#define WEXITSTATUS(sts) (sts & 0x0FF) + +extern wchar_t *__la_win_permissive_name(const char *name); +extern wchar_t *__la_win_permissive_name_w(const wchar_t *wname); +extern void __la_dosmaperr(unsigned long e); +#define la_dosmaperr(e) __la_dosmaperr(e) +extern struct archive_entry *__la_win_entry_in_posix_pathseparator( + struct archive_entry *); + +#if defined(HAVE_WCRTOMB) && defined(__BORLANDC__) +typedef int mbstate_t; +size_t wcrtomb(char *, wchar_t, mbstate_t *); +#endif + +#if defined(_MSC_VER) && _MSC_VER < 1300 +WINBASEAPI BOOL WINAPI GetVolumePathNameW( + LPCWSTR lpszFileName, + LPWSTR lpszVolumePathName, + DWORD cchBufferLength + ); +# if _WIN32_WINNT < 0x0500 /* windows.h not providing 0x500 API */ +typedef struct _FILE_ALLOCATED_RANGE_BUFFER { + LARGE_INTEGER FileOffset; + LARGE_INTEGER Length; +} FILE_ALLOCATED_RANGE_BUFFER, *PFILE_ALLOCATED_RANGE_BUFFER; +# define FSCTL_SET_SPARSE \ + CTL_CODE(FILE_DEVICE_FILE_SYSTEM, 49, METHOD_BUFFERED, FILE_WRITE_DATA) +# define FSCTL_QUERY_ALLOCATED_RANGES \ + CTL_CODE(FILE_DEVICE_FILE_SYSTEM, 51, METHOD_NEITHER, FILE_READ_DATA) +# endif +#endif + +#endif /* LIBARCHIVE_ARCHIVE_WINDOWS_H_INCLUDED */ diff --git a/c/archive_write.c b/c/archive_write.c new file mode 100644 --- /dev/null +++ b/c/archive_write.c @@ -0,0 +1,789 @@ +/*- + * Copyright (c) 2003-2010 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write.c 201099 2009-12-28 03:03:00Z kientzle $"); + +/* + * This file contains the "essential" portions of the write API, that + * is, stuff that will essentially always be used by any client that + * actually needs to write an archive. Optional pieces have been, as + * far as possible, separated out into separate files to reduce + * needlessly bloating statically-linked clients. + */ + +#ifdef HAVE_SYS_WAIT_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_LIMITS_H +#include +#endif +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#include +#ifdef HAVE_UNISTD_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_write_private.h" + +static struct archive_vtable *archive_write_vtable(void); + +static int _archive_filter_code(struct archive *, int); +static const char *_archive_filter_name(struct archive *, int); +static int64_t _archive_filter_bytes(struct archive *, int); +static int _archive_write_filter_count(struct archive *); +static int _archive_write_close(struct archive *); +static int _archive_write_free(struct archive *); +static int _archive_write_header(struct archive *, struct archive_entry *); +static int _archive_write_finish_entry(struct archive *); +static ssize_t _archive_write_data(struct archive *, const void *, size_t); + +struct archive_none { + size_t buffer_size; + size_t avail; + char *buffer; + char *next; +}; + +static struct archive_vtable * +archive_write_vtable(void) +{ + static struct archive_vtable av; + static int inited = 0; + + if (!inited) { + av.archive_close = _archive_write_close; + av.archive_filter_bytes = _archive_filter_bytes; + av.archive_filter_code = _archive_filter_code; + av.archive_filter_name = _archive_filter_name; + av.archive_filter_count = _archive_write_filter_count; + av.archive_free = _archive_write_free; + av.archive_write_header = _archive_write_header; + av.archive_write_finish_entry = _archive_write_finish_entry; + av.archive_write_data = _archive_write_data; + inited = 1; + } + return (&av); +} + +/* + * Allocate, initialize and return an archive object. + */ +struct archive * +archive_write_new(void) +{ + struct archive_write *a; + unsigned char *nulls; + + a = (struct archive_write *)calloc(1, sizeof(*a)); + if (a == NULL) + return (NULL); + a->archive.magic = ARCHIVE_WRITE_MAGIC; + a->archive.state = ARCHIVE_STATE_NEW; + a->archive.vtable = archive_write_vtable(); + /* + * The value 10240 here matches the traditional tar default, + * but is otherwise arbitrary. + * TODO: Set the default block size from the format selected. + */ + a->bytes_per_block = 10240; + a->bytes_in_last_block = -1; /* Default */ + + /* Initialize a block of nulls for padding purposes. */ + a->null_length = 1024; + nulls = (unsigned char *)calloc(1, a->null_length); + if (nulls == NULL) { + free(a); + return (NULL); + } + a->nulls = nulls; + return (&a->archive); +} + +/* + * Set the block size. Returns 0 if successful. + */ +int +archive_write_set_bytes_per_block(struct archive *_a, int bytes_per_block) +{ + struct archive_write *a = (struct archive_write *)_a; + archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_set_bytes_per_block"); + a->bytes_per_block = bytes_per_block; + return (ARCHIVE_OK); +} + +/* + * Get the current block size. -1 if it has never been set. + */ +int +archive_write_get_bytes_per_block(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_ANY, "archive_write_get_bytes_per_block"); + return (a->bytes_per_block); +} + +/* + * Set the size for the last block. + * Returns 0 if successful. + */ +int +archive_write_set_bytes_in_last_block(struct archive *_a, int bytes) +{ + struct archive_write *a = (struct archive_write *)_a; + archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_ANY, "archive_write_set_bytes_in_last_block"); + a->bytes_in_last_block = bytes; + return (ARCHIVE_OK); +} + +/* + * Return the value set above. -1 indicates it has not been set. + */ +int +archive_write_get_bytes_in_last_block(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_ANY, "archive_write_get_bytes_in_last_block"); + return (a->bytes_in_last_block); +} + +/* + * dev/ino of a file to be rejected. Used to prevent adding + * an archive to itself recursively. + */ +int +archive_write_set_skip_file(struct archive *_a, la_int64_t d, la_int64_t i) +{ + struct archive_write *a = (struct archive_write *)_a; + archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_ANY, "archive_write_set_skip_file"); + a->skip_file_set = 1; + a->skip_file_dev = d; + a->skip_file_ino = i; + return (ARCHIVE_OK); +} + +/* + * Allocate and return the next filter structure. + */ +struct archive_write_filter * +__archive_write_allocate_filter(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct archive_write_filter *f; + + f = calloc(1, sizeof(*f)); + f->archive = _a; + f->state = ARCHIVE_WRITE_FILTER_STATE_NEW; + if (a->filter_first == NULL) + a->filter_first = f; + else + a->filter_last->next_filter = f; + a->filter_last = f; + return f; +} + +/* + * Write data to a particular filter. + */ +int +__archive_write_filter(struct archive_write_filter *f, + const void *buff, size_t length) +{ + int r; + /* Never write to non-open filters */ + if (f->state != ARCHIVE_WRITE_FILTER_STATE_OPEN) + return(ARCHIVE_FATAL); + if (length == 0) + return(ARCHIVE_OK); + if (f->write == NULL) + /* If unset, a fatal error has already occurred, so this filter + * didn't open. We cannot write anything. */ + return(ARCHIVE_FATAL); + r = (f->write)(f, buff, length); + f->bytes_written += length; + return (r); +} + +/* + * Recursive function for opening the filter chain + * Last filter is opened first + */ +static int +__archive_write_open_filter(struct archive_write_filter *f) +{ + int ret; + + ret = ARCHIVE_OK; + if (f->next_filter != NULL) + ret = __archive_write_open_filter(f->next_filter); + if (ret != ARCHIVE_OK) + return (ret); + if (f->state != ARCHIVE_WRITE_FILTER_STATE_NEW) + return (ARCHIVE_FATAL); + if (f->open == NULL) { + f->state = ARCHIVE_WRITE_FILTER_STATE_OPEN; + return (ARCHIVE_OK); + } + ret = (f->open)(f); + if (ret == ARCHIVE_OK) + f->state = ARCHIVE_WRITE_FILTER_STATE_OPEN; + else + f->state = ARCHIVE_WRITE_FILTER_STATE_FATAL; + return (ret); +} + +/* + * Open all filters + */ +static int +__archive_write_filters_open(struct archive_write *a) +{ + return (__archive_write_open_filter(a->filter_first)); +} + +/* + * Close all filtes + */ +static int +__archive_write_filters_close(struct archive_write *a) +{ + struct archive_write_filter *f; + int ret, ret1; + ret = ARCHIVE_OK; + for (f = a->filter_first; f != NULL; f = f->next_filter) { + /* Do not close filters that are not open */ + if (f->state == ARCHIVE_WRITE_FILTER_STATE_OPEN) { + if (f->close != NULL) { + ret1 = (f->close)(f); + if (ret1 < ret) + ret = ret1; + if (ret1 == ARCHIVE_OK) { + f->state = + ARCHIVE_WRITE_FILTER_STATE_CLOSED; + } else { + f->state = + ARCHIVE_WRITE_FILTER_STATE_FATAL; + } + } else + f->state = ARCHIVE_WRITE_FILTER_STATE_CLOSED; + } + } + return (ret); +} + +int +__archive_write_output(struct archive_write *a, const void *buff, size_t length) +{ + return (__archive_write_filter(a->filter_first, buff, length)); +} + +int +__archive_write_nulls(struct archive_write *a, size_t length) +{ + if (length == 0) + return (ARCHIVE_OK); + + while (length > 0) { + size_t to_write = length < a->null_length ? length : a->null_length; + int r = __archive_write_output(a, a->nulls, to_write); + if (r < ARCHIVE_OK) + return (r); + length -= to_write; + } + return (ARCHIVE_OK); +} + +static int +archive_write_client_open(struct archive_write_filter *f) +{ + struct archive_write *a = (struct archive_write *)f->archive; + struct archive_none *state; + void *buffer; + size_t buffer_size; + int ret; + + f->bytes_per_block = archive_write_get_bytes_per_block(f->archive); + f->bytes_in_last_block = + archive_write_get_bytes_in_last_block(f->archive); + buffer_size = f->bytes_per_block; + + state = (struct archive_none *)calloc(1, sizeof(*state)); + buffer = (char *)malloc(buffer_size); + if (state == NULL || buffer == NULL) { + free(state); + free(buffer); + archive_set_error(f->archive, ENOMEM, + "Can't allocate data for output buffering"); + return (ARCHIVE_FATAL); + } + + state->buffer_size = buffer_size; + state->buffer = buffer; + state->next = state->buffer; + state->avail = state->buffer_size; + f->data = state; + + if (a->client_opener == NULL) + return (ARCHIVE_OK); + ret = a->client_opener(f->archive, a->client_data); + if (ret != ARCHIVE_OK) { + free(state->buffer); + free(state); + f->data = NULL; + } + return (ret); +} + +static int +archive_write_client_write(struct archive_write_filter *f, + const void *_buff, size_t length) +{ + struct archive_write *a = (struct archive_write *)f->archive; + struct archive_none *state = (struct archive_none *)f->data; + const char *buff = (const char *)_buff; + ssize_t remaining, to_copy; + ssize_t bytes_written; + + remaining = length; + + /* + * If there is no buffer for blocking, just pass the data + * straight through to the client write callback. In + * particular, this supports "no write delay" operation for + * special applications. Just set the block size to zero. + */ + if (state->buffer_size == 0) { + while (remaining > 0) { + bytes_written = (a->client_writer)(&a->archive, + a->client_data, buff, remaining); + if (bytes_written <= 0) + return (ARCHIVE_FATAL); + remaining -= bytes_written; + buff += bytes_written; + } + return (ARCHIVE_OK); + } + + /* If the copy buffer isn't empty, try to fill it. */ + if (state->avail < state->buffer_size) { + /* If buffer is not empty... */ + /* ... copy data into buffer ... */ + to_copy = ((size_t)remaining > state->avail) ? + state->avail : (size_t)remaining; + memcpy(state->next, buff, to_copy); + state->next += to_copy; + state->avail -= to_copy; + buff += to_copy; + remaining -= to_copy; + /* ... if it's full, write it out. */ + if (state->avail == 0) { + char *p = state->buffer; + size_t to_write = state->buffer_size; + while (to_write > 0) { + bytes_written = (a->client_writer)(&a->archive, + a->client_data, p, to_write); + if (bytes_written <= 0) + return (ARCHIVE_FATAL); + if ((size_t)bytes_written > to_write) { + archive_set_error(&(a->archive), + -1, "write overrun"); + return (ARCHIVE_FATAL); + } + p += bytes_written; + to_write -= bytes_written; + } + state->next = state->buffer; + state->avail = state->buffer_size; + } + } + + while ((size_t)remaining >= state->buffer_size) { + /* Write out full blocks directly to client. */ + bytes_written = (a->client_writer)(&a->archive, + a->client_data, buff, state->buffer_size); + if (bytes_written <= 0) + return (ARCHIVE_FATAL); + buff += bytes_written; + remaining -= bytes_written; + } + + if (remaining > 0) { + /* Copy last bit into copy buffer. */ + memcpy(state->next, buff, remaining); + state->next += remaining; + state->avail -= remaining; + } + return (ARCHIVE_OK); +} + +static int +archive_write_client_close(struct archive_write_filter *f) +{ + struct archive_write *a = (struct archive_write *)f->archive; + struct archive_none *state = (struct archive_none *)f->data; + ssize_t block_length; + ssize_t target_block_length; + ssize_t bytes_written; + int ret = ARCHIVE_OK; + + /* If there's pending data, pad and write the last block */ + if (state->next != state->buffer) { + block_length = state->buffer_size - state->avail; + + /* Tricky calculation to determine size of last block */ + if (a->bytes_in_last_block <= 0) + /* Default or Zero: pad to full block */ + target_block_length = a->bytes_per_block; + else + /* Round to next multiple of bytes_in_last_block. */ + target_block_length = a->bytes_in_last_block * + ( (block_length + a->bytes_in_last_block - 1) / + a->bytes_in_last_block); + if (target_block_length > a->bytes_per_block) + target_block_length = a->bytes_per_block; + if (block_length < target_block_length) { + memset(state->next, 0, + target_block_length - block_length); + block_length = target_block_length; + } + bytes_written = (a->client_writer)(&a->archive, + a->client_data, state->buffer, block_length); + ret = bytes_written <= 0 ? ARCHIVE_FATAL : ARCHIVE_OK; + } + if (a->client_closer) + (*a->client_closer)(&a->archive, a->client_data); + free(state->buffer); + free(state); + a->client_data = NULL; + /* Clear passphrase. */ + if (a->passphrase != NULL) { + memset(a->passphrase, 0, strlen(a->passphrase)); + free(a->passphrase); + a->passphrase = NULL; + } + /* Clear the close handler myself not to be called again. */ + f->state = ARCHIVE_WRITE_FILTER_STATE_CLOSED; + return (ret); +} + +/* + * Open the archive using the current settings. + */ +int +archive_write_open(struct archive *_a, void *client_data, + archive_open_callback *opener, archive_write_callback *writer, + archive_close_callback *closer) +{ + struct archive_write *a = (struct archive_write *)_a; + struct archive_write_filter *client_filter; + int ret, r1; + + archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_open"); + archive_clear_error(&a->archive); + + a->client_writer = writer; + a->client_opener = opener; + a->client_closer = closer; + a->client_data = client_data; + + client_filter = __archive_write_allocate_filter(_a); + client_filter->open = archive_write_client_open; + client_filter->write = archive_write_client_write; + client_filter->close = archive_write_client_close; + + ret = __archive_write_filters_open(a); + if (ret < ARCHIVE_WARN) { + r1 = __archive_write_filters_close(a); + __archive_write_filters_free(_a); + return (r1 < ret ? r1 : ret); + } + + a->archive.state = ARCHIVE_STATE_HEADER; + if (a->format_init) + ret = (a->format_init)(a); + return (ret); +} + +/* + * Close out the archive. + */ +static int +_archive_write_close(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + int r = ARCHIVE_OK, r1 = ARCHIVE_OK; + + archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, + "archive_write_close"); + if (a->archive.state == ARCHIVE_STATE_NEW + || a->archive.state == ARCHIVE_STATE_CLOSED) + return (ARCHIVE_OK); /* Okay to close() when not open. */ + + archive_clear_error(&a->archive); + + /* Finish the last entry if a finish callback is specified */ + if (a->archive.state == ARCHIVE_STATE_DATA + && a->format_finish_entry != NULL) + r = ((a->format_finish_entry)(a)); + + /* Finish off the archive. */ + /* TODO: have format closers invoke compression close. */ + if (a->format_close != NULL) { + r1 = (a->format_close)(a); + if (r1 < r) + r = r1; + } + + /* Finish the compression and close the stream. */ + r1 = __archive_write_filters_close(a); + if (r1 < r) + r = r1; + + if (a->archive.state != ARCHIVE_STATE_FATAL) + a->archive.state = ARCHIVE_STATE_CLOSED; + return (r); +} + +static int +_archive_write_filter_count(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct archive_write_filter *p = a->filter_first; + int count = 0; + while(p) { + count++; + p = p->next_filter; + } + return count; +} + +void +__archive_write_filters_free(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + int r = ARCHIVE_OK, r1; + + while (a->filter_first != NULL) { + struct archive_write_filter *next + = a->filter_first->next_filter; + if (a->filter_first->free != NULL) { + r1 = (*a->filter_first->free)(a->filter_first); + if (r > r1) + r = r1; + } + free(a->filter_first); + a->filter_first = next; + } + a->filter_last = NULL; +} + +/* + * Destroy the archive structure. + * + * Be careful: user might just call write_new and then write_free. + * Don't assume we actually wrote anything or performed any non-trivial + * initialization. + */ +static int +_archive_write_free(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + int r = ARCHIVE_OK, r1; + + if (_a == NULL) + return (ARCHIVE_OK); + /* It is okay to call free() in state FATAL. */ + archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_write_free"); + if (a->archive.state != ARCHIVE_STATE_FATAL) + r = archive_write_close(&a->archive); + + /* Release format resources. */ + if (a->format_free != NULL) { + r1 = (a->format_free)(a); + if (r1 < r) + r = r1; + } + + __archive_write_filters_free(_a); + + /* Release various dynamic buffers. */ + free((void *)(uintptr_t)(const void *)a->nulls); + archive_string_free(&a->archive.error_string); + if (a->passphrase != NULL) { + /* A passphrase should be cleaned. */ + memset(a->passphrase, 0, strlen(a->passphrase)); + free(a->passphrase); + } + a->archive.magic = 0; + __archive_clean(&a->archive); + free(a); + return (r); +} + +/* + * Write the appropriate header. + */ +static int +_archive_write_header(struct archive *_a, struct archive_entry *entry) +{ + struct archive_write *a = (struct archive_write *)_a; + int ret, r2; + + archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_DATA | ARCHIVE_STATE_HEADER, "archive_write_header"); + archive_clear_error(&a->archive); + + if (a->format_write_header == NULL) { + archive_set_error(&(a->archive), -1, + "Format must be set before you can write to an archive."); + a->archive.state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + } + + /* In particular, "retry" and "fatal" get returned immediately. */ + ret = archive_write_finish_entry(&a->archive); + if (ret == ARCHIVE_FATAL) { + a->archive.state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + } + if (ret < ARCHIVE_OK && ret != ARCHIVE_WARN) + return (ret); + + if (a->skip_file_set && + archive_entry_dev_is_set(entry) && + archive_entry_ino_is_set(entry) && + archive_entry_dev(entry) == (dev_t)a->skip_file_dev && + archive_entry_ino64(entry) == a->skip_file_ino) { + archive_set_error(&a->archive, 0, + "Can't add archive to itself"); + return (ARCHIVE_FAILED); + } + + /* Format and write header. */ + r2 = ((a->format_write_header)(a, entry)); + if (r2 == ARCHIVE_FAILED) { + return (ARCHIVE_FAILED); + } + if (r2 == ARCHIVE_FATAL) { + a->archive.state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); + } + if (r2 < ret) + ret = r2; + + a->archive.state = ARCHIVE_STATE_DATA; + return (ret); +} + +static int +_archive_write_finish_entry(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + int ret = ARCHIVE_OK; + + archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, + "archive_write_finish_entry"); + if (a->archive.state & ARCHIVE_STATE_DATA + && a->format_finish_entry != NULL) + ret = (a->format_finish_entry)(a); + a->archive.state = ARCHIVE_STATE_HEADER; + return (ret); +} + +/* + * Note that the compressor is responsible for blocking. + */ +static ssize_t +_archive_write_data(struct archive *_a, const void *buff, size_t s) +{ + struct archive_write *a = (struct archive_write *)_a; + const size_t max_write = INT_MAX; + + archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_DATA, "archive_write_data"); + /* In particular, this catches attempts to pass negative values. */ + if (s > max_write) + s = max_write; + archive_clear_error(&a->archive); + return ((a->format_write_data)(a, buff, s)); +} + +static struct archive_write_filter * +filter_lookup(struct archive *_a, int n) +{ + struct archive_write *a = (struct archive_write *)_a; + struct archive_write_filter *f = a->filter_first; + if (n == -1) + return a->filter_last; + if (n < 0) + return NULL; + while (n > 0 && f != NULL) { + f = f->next_filter; + --n; + } + return f; +} + +static int +_archive_filter_code(struct archive *_a, int n) +{ + struct archive_write_filter *f = filter_lookup(_a, n); + return f == NULL ? -1 : f->code; +} + +static const char * +_archive_filter_name(struct archive *_a, int n) +{ + struct archive_write_filter *f = filter_lookup(_a, n); + return f != NULL ? f->name : NULL; +} + +static int64_t +_archive_filter_bytes(struct archive *_a, int n) +{ + struct archive_write_filter *f = filter_lookup(_a, n); + return f == NULL ? -1 : f->bytes_written; +} diff --git a/c/archive_write_add_filter.c b/c/archive_write_add_filter.c new file mode 100644 --- /dev/null +++ b/c/archive_write_add_filter.c @@ -0,0 +1,72 @@ +/*- + * Copyright (c) 2012 Ondrej Holy + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_SYS_TYPES_H +#include +#endif + +#ifdef HAVE_ERRNO_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" + +/* A table that maps filter codes to functions. */ +static const +struct { int code; int (*setter)(struct archive *); } codes[] = +{ + { ARCHIVE_FILTER_NONE, archive_write_add_filter_none }, + { ARCHIVE_FILTER_GZIP, archive_write_add_filter_gzip }, + { ARCHIVE_FILTER_BZIP2, archive_write_add_filter_bzip2 }, + { ARCHIVE_FILTER_COMPRESS, archive_write_add_filter_compress }, + { ARCHIVE_FILTER_GRZIP, archive_write_add_filter_grzip }, + { ARCHIVE_FILTER_LRZIP, archive_write_add_filter_lrzip }, + { ARCHIVE_FILTER_LZ4, archive_write_add_filter_lz4 }, + { ARCHIVE_FILTER_LZIP, archive_write_add_filter_lzip }, + { ARCHIVE_FILTER_LZMA, archive_write_add_filter_lzma }, + { ARCHIVE_FILTER_LZOP, archive_write_add_filter_lzip }, + { ARCHIVE_FILTER_UU, archive_write_add_filter_uuencode }, + { ARCHIVE_FILTER_XZ, archive_write_add_filter_xz }, + { ARCHIVE_FILTER_ZSTD, archive_write_add_filter_zstd }, + { -1, NULL } +}; + +int +archive_write_add_filter(struct archive *a, int code) +{ + int i; + + for (i = 0; codes[i].code != -1; i++) { + if (code == codes[i].code) + return ((codes[i].setter)(a)); + } + + archive_set_error(a, EINVAL, "No such filter"); + return (ARCHIVE_FATAL); +} diff --git a/c/archive_write_add_filter_b64encode.c b/c/archive_write_add_filter_b64encode.c new file mode 100644 --- /dev/null +++ b/c/archive_write_add_filter_b64encode.c @@ -0,0 +1,304 @@ +/*- + * Copyright (c) 2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_string.h" +#include "archive_write_private.h" + +#define LBYTES 57 + +struct private_b64encode { + int mode; + struct archive_string name; + struct archive_string encoded_buff; + size_t bs; + size_t hold_len; + unsigned char hold[LBYTES]; +}; + +static int archive_filter_b64encode_options(struct archive_write_filter *, + const char *, const char *); +static int archive_filter_b64encode_open(struct archive_write_filter *); +static int archive_filter_b64encode_write(struct archive_write_filter *, + const void *, size_t); +static int archive_filter_b64encode_close(struct archive_write_filter *); +static int archive_filter_b64encode_free(struct archive_write_filter *); +static void la_b64_encode(struct archive_string *, const unsigned char *, size_t); +static int64_t atol8(const char *, size_t); + +static const char base64[] = { + 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', + 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P', + 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', + 'Y', 'Z', 'a', 'b', 'c', 'd', 'e', 'f', + 'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n', + 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', + 'w', 'x', 'y', 'z', '0', '1', '2', '3', + '4', '5', '6', '7', '8', '9', '+', '/' +}; + +/* + * Add a compress filter to this write handle. + */ +int +archive_write_add_filter_b64encode(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct archive_write_filter *f = __archive_write_allocate_filter(_a); + struct private_b64encode *state; + + archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_add_filter_uu"); + + state = (struct private_b64encode *)calloc(1, sizeof(*state)); + if (state == NULL) { + archive_set_error(f->archive, ENOMEM, + "Can't allocate data for b64encode filter"); + return (ARCHIVE_FATAL); + } + archive_strcpy(&state->name, "-"); + state->mode = 0644; + + f->data = state; + f->name = "b64encode"; + f->code = ARCHIVE_FILTER_UU; + f->open = archive_filter_b64encode_open; + f->options = archive_filter_b64encode_options; + f->write = archive_filter_b64encode_write; + f->close = archive_filter_b64encode_close; + f->free = archive_filter_b64encode_free; + + return (ARCHIVE_OK); +} + +/* + * Set write options. + */ +static int +archive_filter_b64encode_options(struct archive_write_filter *f, const char *key, + const char *value) +{ + struct private_b64encode *state = (struct private_b64encode *)f->data; + + if (strcmp(key, "mode") == 0) { + if (value == NULL) { + archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, + "mode option requires octal digits"); + return (ARCHIVE_FAILED); + } + state->mode = (int)atol8(value, strlen(value)) & 0777; + return (ARCHIVE_OK); + } else if (strcmp(key, "name") == 0) { + if (value == NULL) { + archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, + "name option requires a string"); + return (ARCHIVE_FAILED); + } + archive_strcpy(&state->name, value); + return (ARCHIVE_OK); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +/* + * Setup callback. + */ +static int +archive_filter_b64encode_open(struct archive_write_filter *f) +{ + struct private_b64encode *state = (struct private_b64encode *)f->data; + size_t bs = 65536, bpb; + + if (f->archive->magic == ARCHIVE_WRITE_MAGIC) { + /* Buffer size should be a multiple number of the of bytes + * per block for performance. */ + bpb = archive_write_get_bytes_per_block(f->archive); + if (bpb > bs) + bs = bpb; + else if (bpb != 0) + bs -= bs % bpb; + } + + state->bs = bs; + if (archive_string_ensure(&state->encoded_buff, bs + 512) == NULL) { + archive_set_error(f->archive, ENOMEM, + "Can't allocate data for b64encode buffer"); + return (ARCHIVE_FATAL); + } + + archive_string_sprintf(&state->encoded_buff, "begin-base64 %o %s\n", + state->mode, state->name.s); + + f->data = state; + return (0); +} + +static void +la_b64_encode(struct archive_string *as, const unsigned char *p, size_t len) +{ + int c; + + for (; len >= 3; p += 3, len -= 3) { + c = p[0] >> 2; + archive_strappend_char(as, base64[c]); + c = ((p[0] & 0x03) << 4) | ((p[1] & 0xf0) >> 4); + archive_strappend_char(as, base64[c]); + c = ((p[1] & 0x0f) << 2) | ((p[2] & 0xc0) >> 6); + archive_strappend_char(as, base64[c]); + c = p[2] & 0x3f; + archive_strappend_char(as, base64[c]); + } + if (len > 0) { + c = p[0] >> 2; + archive_strappend_char(as, base64[c]); + c = (p[0] & 0x03) << 4; + if (len == 1) { + archive_strappend_char(as, base64[c]); + archive_strappend_char(as, '='); + archive_strappend_char(as, '='); + } else { + c |= (p[1] & 0xf0) >> 4; + archive_strappend_char(as, base64[c]); + c = (p[1] & 0x0f) << 2; + archive_strappend_char(as, base64[c]); + archive_strappend_char(as, '='); + } + } + archive_strappend_char(as, '\n'); +} + +/* + * Write data to the encoded stream. + */ +static int +archive_filter_b64encode_write(struct archive_write_filter *f, const void *buff, + size_t length) +{ + struct private_b64encode *state = (struct private_b64encode *)f->data; + const unsigned char *p = buff; + int ret = ARCHIVE_OK; + + if (length == 0) + return (ret); + + if (state->hold_len) { + while (state->hold_len < LBYTES && length > 0) { + state->hold[state->hold_len++] = *p++; + length--; + } + if (state->hold_len < LBYTES) + return (ret); + la_b64_encode(&state->encoded_buff, state->hold, LBYTES); + state->hold_len = 0; + } + + for (; length >= LBYTES; length -= LBYTES, p += LBYTES) + la_b64_encode(&state->encoded_buff, p, LBYTES); + + /* Save remaining bytes. */ + if (length > 0) { + memcpy(state->hold, p, length); + state->hold_len = length; + } + while (archive_strlen(&state->encoded_buff) >= state->bs) { + ret = __archive_write_filter(f->next_filter, + state->encoded_buff.s, state->bs); + memmove(state->encoded_buff.s, + state->encoded_buff.s + state->bs, + state->encoded_buff.length - state->bs); + state->encoded_buff.length -= state->bs; + } + + return (ret); +} + + +/* + * Finish the compression... + */ +static int +archive_filter_b64encode_close(struct archive_write_filter *f) +{ + struct private_b64encode *state = (struct private_b64encode *)f->data; + + /* Flush remaining bytes. */ + if (state->hold_len != 0) + la_b64_encode(&state->encoded_buff, state->hold, state->hold_len); + archive_string_sprintf(&state->encoded_buff, "====\n"); + /* Write the last block */ + archive_write_set_bytes_in_last_block(f->archive, 1); + return __archive_write_filter(f->next_filter, + state->encoded_buff.s, archive_strlen(&state->encoded_buff)); +} + +static int +archive_filter_b64encode_free(struct archive_write_filter *f) +{ + struct private_b64encode *state = (struct private_b64encode *)f->data; + + archive_string_free(&state->name); + archive_string_free(&state->encoded_buff); + free(state); + return (ARCHIVE_OK); +} + +static int64_t +atol8(const char *p, size_t char_cnt) +{ + int64_t l; + int digit; + + l = 0; + while (char_cnt-- > 0) { + if (*p >= '0' && *p <= '7') + digit = *p - '0'; + else + break; + p++; + l <<= 3; + l |= digit; + } + return (l); +} + diff --git a/c/archive_write_add_filter_by_name.c b/c/archive_write_add_filter_by_name.c new file mode 100644 --- /dev/null +++ b/c/archive_write_add_filter_by_name.c @@ -0,0 +1,77 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * Copyright (c) 2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_SYS_TYPES_H +#include +#endif + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" + +/* A table that maps names to functions. */ +static const +struct { const char *name; int (*setter)(struct archive *); } names[] = +{ + { "b64encode", archive_write_add_filter_b64encode }, + { "bzip2", archive_write_add_filter_bzip2 }, + { "compress", archive_write_add_filter_compress }, + { "grzip", archive_write_add_filter_grzip }, + { "gzip", archive_write_add_filter_gzip }, + { "lrzip", archive_write_add_filter_lrzip }, + { "lz4", archive_write_add_filter_lz4 }, + { "lzip", archive_write_add_filter_lzip }, + { "lzma", archive_write_add_filter_lzma }, + { "lzop", archive_write_add_filter_lzop }, + { "uuencode", archive_write_add_filter_uuencode }, + { "xz", archive_write_add_filter_xz }, + { "zstd", archive_write_add_filter_zstd }, + { NULL, NULL } +}; + +int +archive_write_add_filter_by_name(struct archive *a, const char *name) +{ + int i; + + for (i = 0; names[i].name != NULL; i++) { + if (strcmp(name, names[i].name) == 0) + return ((names[i].setter)(a)); + } + + archive_set_error(a, EINVAL, "No such filter '%s'", name); + a->state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); +} diff --git a/c/archive_write_add_filter_bzip2.c b/c/archive_write_add_filter_bzip2.c new file mode 100644 --- /dev/null +++ b/c/archive_write_add_filter_bzip2.c @@ -0,0 +1,401 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * Copyright (c) 2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_compression_bzip2.c 201091 2009-12-28 02:22:41Z kientzle $"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_BZLIB_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_write_private.h" + +#if ARCHIVE_VERSION_NUMBER < 4000000 +int +archive_write_set_compression_bzip2(struct archive *a) +{ + __archive_write_filters_free(a); + return (archive_write_add_filter_bzip2(a)); +} +#endif + +struct private_data { + int compression_level; +#if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) + bz_stream stream; + int64_t total_in; + char *compressed; + size_t compressed_buffer_size; +#else + struct archive_write_program_data *pdata; +#endif +}; + +static int archive_compressor_bzip2_close(struct archive_write_filter *); +static int archive_compressor_bzip2_free(struct archive_write_filter *); +static int archive_compressor_bzip2_open(struct archive_write_filter *); +static int archive_compressor_bzip2_options(struct archive_write_filter *, + const char *, const char *); +static int archive_compressor_bzip2_write(struct archive_write_filter *, + const void *, size_t); + +/* + * Add a bzip2 compression filter to this write handle. + */ +int +archive_write_add_filter_bzip2(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct archive_write_filter *f = __archive_write_allocate_filter(_a); + struct private_data *data; + + archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_add_filter_bzip2"); + + data = calloc(1, sizeof(*data)); + if (data == NULL) { + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + data->compression_level = 9; /* default */ + + f->data = data; + f->options = &archive_compressor_bzip2_options; + f->close = &archive_compressor_bzip2_close; + f->free = &archive_compressor_bzip2_free; + f->open = &archive_compressor_bzip2_open; + f->code = ARCHIVE_FILTER_BZIP2; + f->name = "bzip2"; +#if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) + return (ARCHIVE_OK); +#else + data->pdata = __archive_write_program_allocate("bzip2"); + if (data->pdata == NULL) { + free(data); + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + data->compression_level = 0; + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Using external bzip2 program"); + return (ARCHIVE_WARN); +#endif +} + +/* + * Set write options. + */ +static int +archive_compressor_bzip2_options(struct archive_write_filter *f, + const char *key, const char *value) +{ + struct private_data *data = (struct private_data *)f->data; + + if (strcmp(key, "compression-level") == 0) { + if (value == NULL || !(value[0] >= '0' && value[0] <= '9') || + value[1] != '\0') + return (ARCHIVE_WARN); + data->compression_level = value[0] - '0'; + /* Make '0' be a synonym for '1'. */ + /* This way, bzip2 compressor supports the same 0..9 + * range of levels as gzip. */ + if (data->compression_level < 1) + data->compression_level = 1; + return (ARCHIVE_OK); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +#if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) +/* Don't compile this if we don't have bzlib. */ + +/* + * Yuck. bzlib.h is not const-correct, so I need this one bit + * of ugly hackery to convert a const * pointer to a non-const pointer. + */ +#define SET_NEXT_IN(st,src) \ + (st)->stream.next_in = (char *)(uintptr_t)(const void *)(src) +static int drive_compressor(struct archive_write_filter *, + struct private_data *, int finishing); + +/* + * Setup callback. + */ +static int +archive_compressor_bzip2_open(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + int ret; + + if (data->compressed == NULL) { + size_t bs = 65536, bpb; + if (f->archive->magic == ARCHIVE_WRITE_MAGIC) { + /* Buffer size should be a multiple number of the of bytes + * per block for performance. */ + bpb = archive_write_get_bytes_per_block(f->archive); + if (bpb > bs) + bs = bpb; + else if (bpb != 0) + bs -= bs % bpb; + } + data->compressed_buffer_size = bs; + data->compressed + = (char *)malloc(data->compressed_buffer_size); + if (data->compressed == NULL) { + archive_set_error(f->archive, ENOMEM, + "Can't allocate data for compression buffer"); + return (ARCHIVE_FATAL); + } + } + + memset(&data->stream, 0, sizeof(data->stream)); + data->stream.next_out = data->compressed; + data->stream.avail_out = data->compressed_buffer_size; + f->write = archive_compressor_bzip2_write; + + /* Initialize compression library */ + ret = BZ2_bzCompressInit(&(data->stream), + data->compression_level, 0, 30); + if (ret == BZ_OK) { + f->data = data; + return (ARCHIVE_OK); + } + + /* Library setup failed: clean up. */ + archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, + "Internal error initializing compression library"); + + /* Override the error message if we know what really went wrong. */ + switch (ret) { + case BZ_PARAM_ERROR: + archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, + "Internal error initializing compression library: " + "invalid setup parameter"); + break; + case BZ_MEM_ERROR: + archive_set_error(f->archive, ENOMEM, + "Internal error initializing compression library: " + "out of memory"); + break; + case BZ_CONFIG_ERROR: + archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, + "Internal error initializing compression library: " + "mis-compiled library"); + break; + } + + return (ARCHIVE_FATAL); + +} + +/* + * Write data to the compressed stream. + * + * Returns ARCHIVE_OK if all data written, error otherwise. + */ +static int +archive_compressor_bzip2_write(struct archive_write_filter *f, + const void *buff, size_t length) +{ + struct private_data *data = (struct private_data *)f->data; + + /* Update statistics */ + data->total_in += length; + + /* Compress input data to output buffer */ + SET_NEXT_IN(data, buff); + data->stream.avail_in = length; + if (drive_compressor(f, data, 0)) + return (ARCHIVE_FATAL); + return (ARCHIVE_OK); +} + + +/* + * Finish the compression. + */ +static int +archive_compressor_bzip2_close(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + int ret; + + /* Finish compression cycle. */ + ret = drive_compressor(f, data, 1); + if (ret == ARCHIVE_OK) { + /* Write the last block */ + ret = __archive_write_filter(f->next_filter, + data->compressed, + data->compressed_buffer_size - data->stream.avail_out); + } + + switch (BZ2_bzCompressEnd(&(data->stream))) { + case BZ_OK: + break; + default: + archive_set_error(f->archive, ARCHIVE_ERRNO_PROGRAMMER, + "Failed to clean up compressor"); + ret = ARCHIVE_FATAL; + } + return ret; +} + +static int +archive_compressor_bzip2_free(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + free(data->compressed); + free(data); + f->data = NULL; + return (ARCHIVE_OK); +} + +/* + * Utility function to push input data through compressor, writing + * full output blocks as necessary. + * + * Note that this handles both the regular write case (finishing == + * false) and the end-of-archive case (finishing == true). + */ +static int +drive_compressor(struct archive_write_filter *f, + struct private_data *data, int finishing) +{ + int ret; + + for (;;) { + if (data->stream.avail_out == 0) { + ret = __archive_write_filter(f->next_filter, + data->compressed, + data->compressed_buffer_size); + if (ret != ARCHIVE_OK) { + /* TODO: Handle this write failure */ + return (ARCHIVE_FATAL); + } + data->stream.next_out = data->compressed; + data->stream.avail_out = data->compressed_buffer_size; + } + + /* If there's nothing to do, we're done. */ + if (!finishing && data->stream.avail_in == 0) + return (ARCHIVE_OK); + + ret = BZ2_bzCompress(&(data->stream), + finishing ? BZ_FINISH : BZ_RUN); + + switch (ret) { + case BZ_RUN_OK: + /* In non-finishing case, did compressor + * consume everything? */ + if (!finishing && data->stream.avail_in == 0) + return (ARCHIVE_OK); + break; + case BZ_FINISH_OK: /* Finishing: There's more work to do */ + break; + case BZ_STREAM_END: /* Finishing: all done */ + /* Only occurs in finishing case */ + return (ARCHIVE_OK); + default: + /* Any other return value indicates an error */ + archive_set_error(f->archive, + ARCHIVE_ERRNO_PROGRAMMER, + "Bzip2 compression failed;" + " BZ2_bzCompress() returned %d", + ret); + return (ARCHIVE_FATAL); + } + } +} + +#else /* HAVE_BZLIB_H && BZ_CONFIG_ERROR */ + +static int +archive_compressor_bzip2_open(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + struct archive_string as; + int r; + + archive_string_init(&as); + archive_strcpy(&as, "bzip2"); + + /* Specify compression level. */ + if (data->compression_level > 0) { + archive_strcat(&as, " -"); + archive_strappend_char(&as, '0' + data->compression_level); + } + f->write = archive_compressor_bzip2_write; + + r = __archive_write_program_open(f, data->pdata, as.s); + archive_string_free(&as); + return (r); +} + +static int +archive_compressor_bzip2_write(struct archive_write_filter *f, const void *buff, + size_t length) +{ + struct private_data *data = (struct private_data *)f->data; + + return __archive_write_program_write(f, data->pdata, buff, length); +} + +static int +archive_compressor_bzip2_close(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + + return __archive_write_program_close(f, data->pdata); +} + +static int +archive_compressor_bzip2_free(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + + __archive_write_program_free(data->pdata); + free(data); + return (ARCHIVE_OK); +} + +#endif /* HAVE_BZLIB_H && BZ_CONFIG_ERROR */ diff --git a/c/archive_write_add_filter_compress.c b/c/archive_write_add_filter_compress.c new file mode 100644 --- /dev/null +++ b/c/archive_write_add_filter_compress.c @@ -0,0 +1,447 @@ +/*- + * Copyright (c) 2008 Joerg Sonnenberger + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +/*- + * Copyright (c) 1985, 1986, 1992, 1993 + * The Regents of the University of California. All rights reserved. + * + * This code is derived from software contributed to Berkeley by + * Diomidis Spinellis and James A. Woods, derived from original + * work by Spencer Thomas and Joseph Orost. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * 3. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + +#include "archive_platform.h" + +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_compression_compress.c 201111 2009-12-28 03:33:05Z kientzle $"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_write_private.h" + +#define HSIZE 69001 /* 95% occupancy */ +#define HSHIFT 8 /* 8 - trunc(log2(HSIZE / 65536)) */ +#define CHECK_GAP 10000 /* Ratio check interval. */ + +#define MAXCODE(bits) ((1 << (bits)) - 1) + +/* + * the next two codes should not be changed lightly, as they must not + * lie within the contiguous general code space. + */ +#define FIRST 257 /* First free entry. */ +#define CLEAR 256 /* Table clear output code. */ + +struct private_data { + int64_t in_count, out_count, checkpoint; + + int code_len; /* Number of bits/code. */ + int cur_maxcode; /* Maximum code, given n_bits. */ + int max_maxcode; /* Should NEVER generate this code. */ + int hashtab [HSIZE]; + unsigned short codetab [HSIZE]; + int first_free; /* First unused entry. */ + int compress_ratio; + + int cur_code, cur_fcode; + + int bit_offset; + unsigned char bit_buf; + + unsigned char *compressed; + size_t compressed_buffer_size; + size_t compressed_offset; +}; + +static int archive_compressor_compress_open(struct archive_write_filter *); +static int archive_compressor_compress_write(struct archive_write_filter *, + const void *, size_t); +static int archive_compressor_compress_close(struct archive_write_filter *); +static int archive_compressor_compress_free(struct archive_write_filter *); + +#if ARCHIVE_VERSION_NUMBER < 4000000 +int +archive_write_set_compression_compress(struct archive *a) +{ + __archive_write_filters_free(a); + return (archive_write_add_filter_compress(a)); +} +#endif + +/* + * Add a compress filter to this write handle. + */ +int +archive_write_add_filter_compress(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct archive_write_filter *f = __archive_write_allocate_filter(_a); + + archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_add_filter_compress"); + f->open = &archive_compressor_compress_open; + f->code = ARCHIVE_FILTER_COMPRESS; + f->name = "compress"; + return (ARCHIVE_OK); +} + +/* + * Setup callback. + */ +static int +archive_compressor_compress_open(struct archive_write_filter *f) +{ + struct private_data *state; + size_t bs = 65536, bpb; + + f->code = ARCHIVE_FILTER_COMPRESS; + f->name = "compress"; + + state = (struct private_data *)calloc(1, sizeof(*state)); + if (state == NULL) { + archive_set_error(f->archive, ENOMEM, + "Can't allocate data for compression"); + return (ARCHIVE_FATAL); + } + + if (f->archive->magic == ARCHIVE_WRITE_MAGIC) { + /* Buffer size should be a multiple number of the of bytes + * per block for performance. */ + bpb = archive_write_get_bytes_per_block(f->archive); + if (bpb > bs) + bs = bpb; + else if (bpb != 0) + bs -= bs % bpb; + } + state->compressed_buffer_size = bs; + state->compressed = malloc(state->compressed_buffer_size); + + if (state->compressed == NULL) { + archive_set_error(f->archive, ENOMEM, + "Can't allocate data for compression buffer"); + free(state); + return (ARCHIVE_FATAL); + } + + f->write = archive_compressor_compress_write; + f->close = archive_compressor_compress_close; + f->free = archive_compressor_compress_free; + + state->max_maxcode = 0x10000; /* Should NEVER generate this code. */ + state->in_count = 0; /* Length of input. */ + state->bit_buf = 0; + state->bit_offset = 0; + state->out_count = 3; /* Includes 3-byte header mojo. */ + state->compress_ratio = 0; + state->checkpoint = CHECK_GAP; + state->code_len = 9; + state->cur_maxcode = MAXCODE(state->code_len); + state->first_free = FIRST; + + memset(state->hashtab, 0xff, sizeof(state->hashtab)); + + /* Prime output buffer with a gzip header. */ + state->compressed[0] = 0x1f; /* Compress */ + state->compressed[1] = 0x9d; + state->compressed[2] = 0x90; /* Block mode, 16bit max */ + state->compressed_offset = 3; + + f->data = state; + return (0); +} + +/*- + * Output the given code. + * Inputs: + * code: A n_bits-bit integer. If == -1, then EOF. This assumes + * that n_bits <= (long)wordsize - 1. + * Outputs: + * Outputs code to the file. + * Assumptions: + * Chars are 8 bits long. + * Algorithm: + * Maintain a BITS character long buffer (so that 8 codes will + * fit in it exactly). Use the VAX insv instruction to insert each + * code in turn. When the buffer fills up empty it and start over. + */ + +static const unsigned char rmask[9] = + {0x00, 0x01, 0x03, 0x07, 0x0f, 0x1f, 0x3f, 0x7f, 0xff}; + +static int +output_byte(struct archive_write_filter *f, unsigned char c) +{ + struct private_data *state = f->data; + + state->compressed[state->compressed_offset++] = c; + ++state->out_count; + + if (state->compressed_buffer_size == state->compressed_offset) { + int ret = __archive_write_filter(f->next_filter, + state->compressed, state->compressed_buffer_size); + if (ret != ARCHIVE_OK) + return ARCHIVE_FATAL; + state->compressed_offset = 0; + } + + return ARCHIVE_OK; +} + +static int +output_code(struct archive_write_filter *f, int ocode) +{ + struct private_data *state = f->data; + int bits, ret, clear_flg, bit_offset; + + clear_flg = ocode == CLEAR; + + /* + * Since ocode is always >= 8 bits, only need to mask the first + * hunk on the left. + */ + bit_offset = state->bit_offset % 8; + state->bit_buf |= (ocode << bit_offset) & 0xff; + output_byte(f, state->bit_buf); + + bits = state->code_len - (8 - bit_offset); + ocode >>= 8 - bit_offset; + /* Get any 8 bit parts in the middle (<=1 for up to 16 bits). */ + if (bits >= 8) { + output_byte(f, ocode & 0xff); + ocode >>= 8; + bits -= 8; + } + /* Last bits. */ + state->bit_offset += state->code_len; + state->bit_buf = ocode & rmask[bits]; + if (state->bit_offset == state->code_len * 8) + state->bit_offset = 0; + + /* + * If the next entry is going to be too big for the ocode size, + * then increase it, if possible. + */ + if (clear_flg || state->first_free > state->cur_maxcode) { + /* + * Write the whole buffer, because the input side won't + * discover the size increase until after it has read it. + */ + if (state->bit_offset > 0) { + while (state->bit_offset < state->code_len * 8) { + ret = output_byte(f, state->bit_buf); + if (ret != ARCHIVE_OK) + return ret; + state->bit_offset += 8; + state->bit_buf = 0; + } + } + state->bit_buf = 0; + state->bit_offset = 0; + + if (clear_flg) { + state->code_len = 9; + state->cur_maxcode = MAXCODE(state->code_len); + } else { + state->code_len++; + if (state->code_len == 16) + state->cur_maxcode = state->max_maxcode; + else + state->cur_maxcode = MAXCODE(state->code_len); + } + } + + return (ARCHIVE_OK); +} + +static int +output_flush(struct archive_write_filter *f) +{ + struct private_data *state = f->data; + int ret; + + /* At EOF, write the rest of the buffer. */ + if (state->bit_offset % 8) { + state->code_len = (state->bit_offset % 8 + 7) / 8; + ret = output_byte(f, state->bit_buf); + if (ret != ARCHIVE_OK) + return ret; + } + + return (ARCHIVE_OK); +} + +/* + * Write data to the compressed stream. + */ +static int +archive_compressor_compress_write(struct archive_write_filter *f, + const void *buff, size_t length) +{ + struct private_data *state = (struct private_data *)f->data; + int i; + int ratio; + int c, disp, ret; + const unsigned char *bp; + + if (length == 0) + return ARCHIVE_OK; + + bp = buff; + + if (state->in_count == 0) { + state->cur_code = *bp++; + ++state->in_count; + --length; + } + + while (length--) { + c = *bp++; + state->in_count++; + state->cur_fcode = (c << 16) + state->cur_code; + i = ((c << HSHIFT) ^ state->cur_code); /* Xor hashing. */ + + if (state->hashtab[i] == state->cur_fcode) { + state->cur_code = state->codetab[i]; + continue; + } + if (state->hashtab[i] < 0) /* Empty slot. */ + goto nomatch; + /* Secondary hash (after G. Knott). */ + if (i == 0) + disp = 1; + else + disp = HSIZE - i; + probe: + if ((i -= disp) < 0) + i += HSIZE; + + if (state->hashtab[i] == state->cur_fcode) { + state->cur_code = state->codetab[i]; + continue; + } + if (state->hashtab[i] >= 0) + goto probe; + nomatch: + ret = output_code(f, state->cur_code); + if (ret != ARCHIVE_OK) + return ret; + state->cur_code = c; + if (state->first_free < state->max_maxcode) { + state->codetab[i] = state->first_free++; /* code -> hashtable */ + state->hashtab[i] = state->cur_fcode; + continue; + } + if (state->in_count < state->checkpoint) + continue; + + state->checkpoint = state->in_count + CHECK_GAP; + + if (state->in_count <= 0x007fffff && state->out_count != 0) + ratio = (int)(state->in_count * 256 / state->out_count); + else if ((ratio = (int)(state->out_count / 256)) == 0) + ratio = 0x7fffffff; + else + ratio = (int)(state->in_count / ratio); + + if (ratio > state->compress_ratio) + state->compress_ratio = ratio; + else { + state->compress_ratio = 0; + memset(state->hashtab, 0xff, sizeof(state->hashtab)); + state->first_free = FIRST; + ret = output_code(f, CLEAR); + if (ret != ARCHIVE_OK) + return ret; + } + } + + return (ARCHIVE_OK); +} + + +/* + * Finish the compression... + */ +static int +archive_compressor_compress_close(struct archive_write_filter *f) +{ + struct private_data *state = (struct private_data *)f->data; + int ret; + + ret = output_code(f, state->cur_code); + if (ret != ARCHIVE_OK) + return ret; + ret = output_flush(f); + if (ret != ARCHIVE_OK) + return ret; + + /* Write the last block */ + ret = __archive_write_filter(f->next_filter, + state->compressed, state->compressed_offset); + return (ret); +} + +static int +archive_compressor_compress_free(struct archive_write_filter *f) +{ + struct private_data *state = (struct private_data *)f->data; + + free(state->compressed); + free(state); + return (ARCHIVE_OK); +} diff --git a/c/archive_write_add_filter_grzip.c b/c/archive_write_add_filter_grzip.c new file mode 100644 --- /dev/null +++ b/c/archive_write_add_filter_grzip.c @@ -0,0 +1,135 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif + +#include "archive.h" +#include "archive_write_private.h" + +struct write_grzip { + struct archive_write_program_data *pdata; +}; + +static int archive_write_grzip_open(struct archive_write_filter *); +static int archive_write_grzip_options(struct archive_write_filter *, + const char *, const char *); +static int archive_write_grzip_write(struct archive_write_filter *, + const void *, size_t); +static int archive_write_grzip_close(struct archive_write_filter *); +static int archive_write_grzip_free(struct archive_write_filter *); + +int +archive_write_add_filter_grzip(struct archive *_a) +{ + struct archive_write_filter *f = __archive_write_allocate_filter(_a); + struct write_grzip *data; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_add_filter_grzip"); + + data = calloc(1, sizeof(*data)); + if (data == NULL) { + archive_set_error(_a, ENOMEM, "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + data->pdata = __archive_write_program_allocate("grzip"); + if (data->pdata == NULL) { + free(data); + archive_set_error(_a, ENOMEM, "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + + f->name = "grzip"; + f->code = ARCHIVE_FILTER_GRZIP; + f->data = data; + f->open = archive_write_grzip_open; + f->options = archive_write_grzip_options; + f->write = archive_write_grzip_write; + f->close = archive_write_grzip_close; + f->free = archive_write_grzip_free; + + /* Note: This filter always uses an external program, so we + * return "warn" to inform of the fact. */ + archive_set_error(_a, ARCHIVE_ERRNO_MISC, + "Using external grzip program for grzip compression"); + return (ARCHIVE_WARN); +} + +static int +archive_write_grzip_options(struct archive_write_filter *f, const char *key, + const char *value) +{ + (void)f; /* UNUSED */ + (void)key; /* UNUSED */ + (void)value; /* UNUSED */ + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +static int +archive_write_grzip_open(struct archive_write_filter *f) +{ + struct write_grzip *data = (struct write_grzip *)f->data; + + return __archive_write_program_open(f, data->pdata, "grzip"); +} + +static int +archive_write_grzip_write(struct archive_write_filter *f, + const void *buff, size_t length) +{ + struct write_grzip *data = (struct write_grzip *)f->data; + + return __archive_write_program_write(f, data->pdata, buff, length); +} + +static int +archive_write_grzip_close(struct archive_write_filter *f) +{ + struct write_grzip *data = (struct write_grzip *)f->data; + + return __archive_write_program_close(f, data->pdata); +} + +static int +archive_write_grzip_free(struct archive_write_filter *f) +{ + struct write_grzip *data = (struct write_grzip *)f->data; + + __archive_write_program_free(data->pdata); + free(data); + return (ARCHIVE_OK); +} diff --git a/c/archive_write_add_filter_gzip.c b/c/archive_write_add_filter_gzip.c new file mode 100644 --- /dev/null +++ b/c/archive_write_add_filter_gzip.c @@ -0,0 +1,442 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_compression_gzip.c 201081 2009-12-28 02:04:42Z kientzle $"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#include +#ifdef HAVE_ZLIB_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_string.h" +#include "archive_write_private.h" + +#if ARCHIVE_VERSION_NUMBER < 4000000 +int +archive_write_set_compression_gzip(struct archive *a) +{ + __archive_write_filters_free(a); + return (archive_write_add_filter_gzip(a)); +} +#endif + +/* Don't compile this if we don't have zlib. */ + +struct private_data { + int compression_level; + int timestamp; +#ifdef HAVE_ZLIB_H + z_stream stream; + int64_t total_in; + unsigned char *compressed; + size_t compressed_buffer_size; + unsigned long crc; +#else + struct archive_write_program_data *pdata; +#endif +}; + +/* + * Yuck. zlib.h is not const-correct, so I need this one bit + * of ugly hackery to convert a const * pointer to a non-const pointer. + */ +#define SET_NEXT_IN(st,src) \ + (st)->stream.next_in = (Bytef *)(uintptr_t)(const void *)(src) + +static int archive_compressor_gzip_options(struct archive_write_filter *, + const char *, const char *); +static int archive_compressor_gzip_open(struct archive_write_filter *); +static int archive_compressor_gzip_write(struct archive_write_filter *, + const void *, size_t); +static int archive_compressor_gzip_close(struct archive_write_filter *); +static int archive_compressor_gzip_free(struct archive_write_filter *); +#ifdef HAVE_ZLIB_H +static int drive_compressor(struct archive_write_filter *, + struct private_data *, int finishing); +#endif + + +/* + * Add a gzip compression filter to this write handle. + */ +int +archive_write_add_filter_gzip(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct archive_write_filter *f = __archive_write_allocate_filter(_a); + struct private_data *data; + archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_add_filter_gzip"); + + data = calloc(1, sizeof(*data)); + if (data == NULL) { + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + f->data = data; + f->open = &archive_compressor_gzip_open; + f->options = &archive_compressor_gzip_options; + f->close = &archive_compressor_gzip_close; + f->free = &archive_compressor_gzip_free; + f->code = ARCHIVE_FILTER_GZIP; + f->name = "gzip"; +#ifdef HAVE_ZLIB_H + data->compression_level = Z_DEFAULT_COMPRESSION; + return (ARCHIVE_OK); +#else + data->pdata = __archive_write_program_allocate("gzip"); + if (data->pdata == NULL) { + free(data); + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + data->compression_level = 0; + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Using external gzip program"); + return (ARCHIVE_WARN); +#endif +} + +static int +archive_compressor_gzip_free(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + +#ifdef HAVE_ZLIB_H + free(data->compressed); +#else + __archive_write_program_free(data->pdata); +#endif + free(data); + f->data = NULL; + return (ARCHIVE_OK); +} + +/* + * Set write options. + */ +static int +archive_compressor_gzip_options(struct archive_write_filter *f, const char *key, + const char *value) +{ + struct private_data *data = (struct private_data *)f->data; + + if (strcmp(key, "compression-level") == 0) { + if (value == NULL || !(value[0] >= '0' && value[0] <= '9') || + value[1] != '\0') + return (ARCHIVE_WARN); + data->compression_level = value[0] - '0'; + return (ARCHIVE_OK); + } + if (strcmp(key, "timestamp") == 0) { + data->timestamp = (value == NULL)?-1:1; + return (ARCHIVE_OK); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +#ifdef HAVE_ZLIB_H +/* + * Setup callback. + */ +static int +archive_compressor_gzip_open(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + int ret; + + if (data->compressed == NULL) { + size_t bs = 65536, bpb; + if (f->archive->magic == ARCHIVE_WRITE_MAGIC) { + /* Buffer size should be a multiple number of + * the of bytes per block for performance. */ + bpb = archive_write_get_bytes_per_block(f->archive); + if (bpb > bs) + bs = bpb; + else if (bpb != 0) + bs -= bs % bpb; + } + data->compressed_buffer_size = bs; + data->compressed + = (unsigned char *)malloc(data->compressed_buffer_size); + if (data->compressed == NULL) { + archive_set_error(f->archive, ENOMEM, + "Can't allocate data for compression buffer"); + return (ARCHIVE_FATAL); + } + } + + data->crc = crc32(0L, NULL, 0); + data->stream.next_out = data->compressed; + data->stream.avail_out = (uInt)data->compressed_buffer_size; + + /* Prime output buffer with a gzip header. */ + data->compressed[0] = 0x1f; /* GZip signature bytes */ + data->compressed[1] = 0x8b; + data->compressed[2] = 0x08; /* "Deflate" compression */ + data->compressed[3] = 0; /* No options */ + if (data->timestamp >= 0) { + time_t t = time(NULL); + data->compressed[4] = (uint8_t)(t)&0xff; /* Timestamp */ + data->compressed[5] = (uint8_t)(t>>8)&0xff; + data->compressed[6] = (uint8_t)(t>>16)&0xff; + data->compressed[7] = (uint8_t)(t>>24)&0xff; + } else + memset(&data->compressed[4], 0, 4); + if (data->compression_level == 9) + data->compressed[8] = 2; + else if(data->compression_level == 1) + data->compressed[8] = 4; + else + data->compressed[8] = 0; + data->compressed[9] = 3; /* OS=Unix */ + data->stream.next_out += 10; + data->stream.avail_out -= 10; + + f->write = archive_compressor_gzip_write; + + /* Initialize compression library. */ + ret = deflateInit2(&(data->stream), + data->compression_level, + Z_DEFLATED, + -15 /* < 0 to suppress zlib header */, + 8, + Z_DEFAULT_STRATEGY); + + if (ret == Z_OK) { + f->data = data; + return (ARCHIVE_OK); + } + + /* Library setup failed: clean up. */ + archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, "Internal error " + "initializing compression library"); + + /* Override the error message if we know what really went wrong. */ + switch (ret) { + case Z_STREAM_ERROR: + archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, + "Internal error initializing " + "compression library: invalid setup parameter"); + break; + case Z_MEM_ERROR: + archive_set_error(f->archive, ENOMEM, + "Internal error initializing compression library"); + break; + case Z_VERSION_ERROR: + archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, + "Internal error initializing " + "compression library: invalid library version"); + break; + } + + return (ARCHIVE_FATAL); +} + +/* + * Write data to the compressed stream. + */ +static int +archive_compressor_gzip_write(struct archive_write_filter *f, const void *buff, + size_t length) +{ + struct private_data *data = (struct private_data *)f->data; + int ret; + + /* Update statistics */ + data->crc = crc32(data->crc, (const Bytef *)buff, (uInt)length); + data->total_in += length; + + /* Compress input data to output buffer */ + SET_NEXT_IN(data, buff); + data->stream.avail_in = (uInt)length; + if ((ret = drive_compressor(f, data, 0)) != ARCHIVE_OK) + return (ret); + + return (ARCHIVE_OK); +} + +/* + * Finish the compression... + */ +static int +archive_compressor_gzip_close(struct archive_write_filter *f) +{ + unsigned char trailer[8]; + struct private_data *data = (struct private_data *)f->data; + int ret; + + /* Finish compression cycle */ + ret = drive_compressor(f, data, 1); + if (ret == ARCHIVE_OK) { + /* Write the last compressed data. */ + ret = __archive_write_filter(f->next_filter, + data->compressed, + data->compressed_buffer_size - data->stream.avail_out); + } + if (ret == ARCHIVE_OK) { + /* Build and write out 8-byte trailer. */ + trailer[0] = (uint8_t)(data->crc)&0xff; + trailer[1] = (uint8_t)(data->crc >> 8)&0xff; + trailer[2] = (uint8_t)(data->crc >> 16)&0xff; + trailer[3] = (uint8_t)(data->crc >> 24)&0xff; + trailer[4] = (uint8_t)(data->total_in)&0xff; + trailer[5] = (uint8_t)(data->total_in >> 8)&0xff; + trailer[6] = (uint8_t)(data->total_in >> 16)&0xff; + trailer[7] = (uint8_t)(data->total_in >> 24)&0xff; + ret = __archive_write_filter(f->next_filter, trailer, 8); + } + + switch (deflateEnd(&(data->stream))) { + case Z_OK: + break; + default: + archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, + "Failed to clean up compressor"); + ret = ARCHIVE_FATAL; + } + return ret; +} + +/* + * Utility function to push input data through compressor, + * writing full output blocks as necessary. + * + * Note that this handles both the regular write case (finishing == + * false) and the end-of-archive case (finishing == true). + */ +static int +drive_compressor(struct archive_write_filter *f, + struct private_data *data, int finishing) +{ + int ret; + + for (;;) { + if (data->stream.avail_out == 0) { + ret = __archive_write_filter(f->next_filter, + data->compressed, + data->compressed_buffer_size); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + data->stream.next_out = data->compressed; + data->stream.avail_out = + (uInt)data->compressed_buffer_size; + } + + /* If there's nothing to do, we're done. */ + if (!finishing && data->stream.avail_in == 0) + return (ARCHIVE_OK); + + ret = deflate(&(data->stream), + finishing ? Z_FINISH : Z_NO_FLUSH ); + + switch (ret) { + case Z_OK: + /* In non-finishing case, check if compressor + * consumed everything */ + if (!finishing && data->stream.avail_in == 0) + return (ARCHIVE_OK); + /* In finishing case, this return always means + * there's more work */ + break; + case Z_STREAM_END: + /* This return can only occur in finishing case. */ + return (ARCHIVE_OK); + default: + /* Any other return value indicates an error. */ + archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, + "GZip compression failed:" + " deflate() call returned status %d", + ret); + return (ARCHIVE_FATAL); + } + } +} + +#else /* HAVE_ZLIB_H */ + +static int +archive_compressor_gzip_open(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + struct archive_string as; + int r; + + archive_string_init(&as); + archive_strcpy(&as, "gzip"); + + /* Specify compression level. */ + if (data->compression_level > 0) { + archive_strcat(&as, " -"); + archive_strappend_char(&as, '0' + data->compression_level); + } + if (data->timestamp < 0) + /* Do not save timestamp. */ + archive_strcat(&as, " -n"); + else if (data->timestamp > 0) + /* Save timestamp. */ + archive_strcat(&as, " -N"); + + f->write = archive_compressor_gzip_write; + r = __archive_write_program_open(f, data->pdata, as.s); + archive_string_free(&as); + return (r); +} + +static int +archive_compressor_gzip_write(struct archive_write_filter *f, const void *buff, + size_t length) +{ + struct private_data *data = (struct private_data *)f->data; + + return __archive_write_program_write(f, data->pdata, buff, length); +} + +static int +archive_compressor_gzip_close(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + + return __archive_write_program_close(f, data->pdata); +} + +#endif /* HAVE_ZLIB_H */ diff --git a/c/archive_write_add_filter_lrzip.c b/c/archive_write_add_filter_lrzip.c new file mode 100644 --- /dev/null +++ b/c/archive_write_add_filter_lrzip.c @@ -0,0 +1,197 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_string.h" +#include "archive_write_private.h" + +struct write_lrzip { + struct archive_write_program_data *pdata; + int compression_level; + enum { lzma = 0, bzip2, gzip, lzo, none, zpaq } compression; +}; + +static int archive_write_lrzip_open(struct archive_write_filter *); +static int archive_write_lrzip_options(struct archive_write_filter *, + const char *, const char *); +static int archive_write_lrzip_write(struct archive_write_filter *, + const void *, size_t); +static int archive_write_lrzip_close(struct archive_write_filter *); +static int archive_write_lrzip_free(struct archive_write_filter *); + +int +archive_write_add_filter_lrzip(struct archive *_a) +{ + struct archive_write_filter *f = __archive_write_allocate_filter(_a); + struct write_lrzip *data; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_add_filter_lrzip"); + + data = calloc(1, sizeof(*data)); + if (data == NULL) { + archive_set_error(_a, ENOMEM, "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + data->pdata = __archive_write_program_allocate("lrzip"); + if (data->pdata == NULL) { + free(data); + archive_set_error(_a, ENOMEM, "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + + f->name = "lrzip"; + f->code = ARCHIVE_FILTER_LRZIP; + f->data = data; + f->open = archive_write_lrzip_open; + f->options = archive_write_lrzip_options; + f->write = archive_write_lrzip_write; + f->close = archive_write_lrzip_close; + f->free = archive_write_lrzip_free; + + /* Note: This filter always uses an external program, so we + * return "warn" to inform of the fact. */ + archive_set_error(_a, ARCHIVE_ERRNO_MISC, + "Using external lrzip program for lrzip compression"); + return (ARCHIVE_WARN); +} + +static int +archive_write_lrzip_options(struct archive_write_filter *f, const char *key, + const char *value) +{ + struct write_lrzip *data = (struct write_lrzip *)f->data; + + if (strcmp(key, "compression") == 0) { + if (value == NULL) + return (ARCHIVE_WARN); + else if (strcmp(value, "bzip2") == 0) + data->compression = bzip2; + else if (strcmp(value, "gzip") == 0) + data->compression = gzip; + else if (strcmp(value, "lzo") == 0) + data->compression = lzo; + else if (strcmp(value, "none") == 0) + data->compression = none; + else if (strcmp(value, "zpaq") == 0) + data->compression = zpaq; + else + return (ARCHIVE_WARN); + return (ARCHIVE_OK); + } else if (strcmp(key, "compression-level") == 0) { + if (value == NULL || !(value[0] >= '1' && value[0] <= '9') || + value[1] != '\0') + return (ARCHIVE_WARN); + data->compression_level = value[0] - '0'; + return (ARCHIVE_OK); + } + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +static int +archive_write_lrzip_open(struct archive_write_filter *f) +{ + struct write_lrzip *data = (struct write_lrzip *)f->data; + struct archive_string as; + int r; + + archive_string_init(&as); + archive_strcpy(&as, "lrzip -q"); + + /* Specify compression type. */ + switch (data->compression) { + case lzma:/* default compression */ + break; + case bzip2: + archive_strcat(&as, " -b"); + break; + case gzip: + archive_strcat(&as, " -g"); + break; + case lzo: + archive_strcat(&as, " -l"); + break; + case none: + archive_strcat(&as, " -n"); + break; + case zpaq: + archive_strcat(&as, " -z"); + break; + } + + /* Specify compression level. */ + if (data->compression_level > 0) { + archive_strcat(&as, " -L "); + archive_strappend_char(&as, '0' + data->compression_level); + } + + r = __archive_write_program_open(f, data->pdata, as.s); + archive_string_free(&as); + return (r); +} + +static int +archive_write_lrzip_write(struct archive_write_filter *f, + const void *buff, size_t length) +{ + struct write_lrzip *data = (struct write_lrzip *)f->data; + + return __archive_write_program_write(f, data->pdata, buff, length); +} + +static int +archive_write_lrzip_close(struct archive_write_filter *f) +{ + struct write_lrzip *data = (struct write_lrzip *)f->data; + + return __archive_write_program_close(f, data->pdata); +} + +static int +archive_write_lrzip_free(struct archive_write_filter *f) +{ + struct write_lrzip *data = (struct write_lrzip *)f->data; + + __archive_write_program_free(data->pdata); + free(data); + return (ARCHIVE_OK); +} diff --git a/c/archive_write_add_filter_lz4.c b/c/archive_write_add_filter_lz4.c new file mode 100644 --- /dev/null +++ b/c/archive_write_add_filter_lz4.c @@ -0,0 +1,700 @@ +/*- + * Copyright (c) 2014 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_LZ4_H +#include +#endif +#ifdef HAVE_LZ4HC_H +#include +#endif + +#include "archive.h" +#include "archive_endian.h" +#include "archive_private.h" +#include "archive_write_private.h" +#include "archive_xxhash.h" + +#define LZ4_MAGICNUMBER 0x184d2204 + +struct private_data { + int compression_level; + unsigned header_written:1; + unsigned version_number:1; + unsigned block_independence:1; + unsigned block_checksum:1; + unsigned stream_size:1; + unsigned stream_checksum:1; + unsigned preset_dictionary:1; + unsigned block_maximum_size:3; +#if defined(HAVE_LIBLZ4) && LZ4_VERSION_MAJOR >= 1 && LZ4_VERSION_MINOR >= 2 + int64_t total_in; + char *out; + char *out_buffer; + size_t out_buffer_size; + size_t out_block_size; + char *in; + char *in_buffer_allocated; + char *in_buffer; + size_t in_buffer_size; + size_t block_size; + + void *xxh32_state; + void *lz4_stream; +#else + struct archive_write_program_data *pdata; +#endif +}; + +static int archive_filter_lz4_close(struct archive_write_filter *); +static int archive_filter_lz4_free(struct archive_write_filter *); +static int archive_filter_lz4_open(struct archive_write_filter *); +static int archive_filter_lz4_options(struct archive_write_filter *, + const char *, const char *); +static int archive_filter_lz4_write(struct archive_write_filter *, + const void *, size_t); + +/* + * Add a lz4 compression filter to this write handle. + */ +int +archive_write_add_filter_lz4(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct archive_write_filter *f = __archive_write_allocate_filter(_a); + struct private_data *data; + + archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_add_filter_lz4"); + + data = calloc(1, sizeof(*data)); + if (data == NULL) { + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + + /* + * Setup default settings. + */ + data->compression_level = 1; + data->version_number = 0x01; + data->block_independence = 1; + data->block_checksum = 0; + data->stream_size = 0; + data->stream_checksum = 1; + data->preset_dictionary = 0; + data->block_maximum_size = 7; + + /* + * Setup a filter setting. + */ + f->data = data; + f->options = &archive_filter_lz4_options; + f->close = &archive_filter_lz4_close; + f->free = &archive_filter_lz4_free; + f->open = &archive_filter_lz4_open; + f->code = ARCHIVE_FILTER_LZ4; + f->name = "lz4"; +#if defined(HAVE_LIBLZ4) && LZ4_VERSION_MAJOR >= 1 && LZ4_VERSION_MINOR >= 2 + return (ARCHIVE_OK); +#else + /* + * We don't have lz4 library, and execute external lz4 program + * instead. + */ + data->pdata = __archive_write_program_allocate("lz4"); + if (data->pdata == NULL) { + free(data); + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + data->compression_level = 0; + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Using external lz4 program"); + return (ARCHIVE_WARN); +#endif +} + +/* + * Set write options. + */ +static int +archive_filter_lz4_options(struct archive_write_filter *f, + const char *key, const char *value) +{ + struct private_data *data = (struct private_data *)f->data; + + if (strcmp(key, "compression-level") == 0) { + int val; + if (value == NULL || !((val = value[0] - '0') >= 1 && val <= 9) || + value[1] != '\0') + return (ARCHIVE_WARN); + +#ifndef HAVE_LZ4HC_H + if(val >= 3) + { + archive_set_error(f->archive, ARCHIVE_ERRNO_PROGRAMMER, + "High compression not included in this build"); + return (ARCHIVE_FATAL); + } +#endif + data->compression_level = val; + return (ARCHIVE_OK); + } + if (strcmp(key, "stream-checksum") == 0) { + data->stream_checksum = value != NULL; + return (ARCHIVE_OK); + } + if (strcmp(key, "block-checksum") == 0) { + data->block_checksum = value != NULL; + return (ARCHIVE_OK); + } + if (strcmp(key, "block-size") == 0) { + if (value == NULL || !(value[0] >= '4' && value[0] <= '7') || + value[1] != '\0') + return (ARCHIVE_WARN); + data->block_maximum_size = value[0] - '0'; + return (ARCHIVE_OK); + } + if (strcmp(key, "block-dependence") == 0) { + data->block_independence = value == NULL; + return (ARCHIVE_OK); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +#if defined(HAVE_LIBLZ4) && LZ4_VERSION_MAJOR >= 1 && LZ4_VERSION_MINOR >= 2 +/* Don't compile this if we don't have liblz4. */ + +static int drive_compressor(struct archive_write_filter *, const char *, + size_t); +static int drive_compressor_independence(struct archive_write_filter *, + const char *, size_t); +static int drive_compressor_dependence(struct archive_write_filter *, + const char *, size_t); +static int lz4_write_stream_descriptor(struct archive_write_filter *); +static ssize_t lz4_write_one_block(struct archive_write_filter *, const char *, + size_t); + + +/* + * Setup callback. + */ +static int +archive_filter_lz4_open(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + size_t required_size; + static size_t const bkmap[] = { 64 * 1024, 256 * 1024, 1 * 1024 * 1024, + 4 * 1024 * 1024 }; + size_t pre_block_size; + + if (data->block_maximum_size < 4) + data->block_size = bkmap[0]; + else + data->block_size = bkmap[data->block_maximum_size - 4]; + + required_size = 4 + 15 + 4 + data->block_size + 4 + 4; + if (data->out_buffer_size < required_size) { + size_t bs = required_size, bpb; + free(data->out_buffer); + if (f->archive->magic == ARCHIVE_WRITE_MAGIC) { + /* Buffer size should be a multiple number of + * the of bytes per block for performance. */ + bpb = archive_write_get_bytes_per_block(f->archive); + if (bpb > bs) + bs = bpb; + else if (bpb != 0) { + bs += bpb; + bs -= bs % bpb; + } + } + data->out_block_size = bs; + bs += required_size; + data->out_buffer = malloc(bs); + data->out = data->out_buffer; + data->out_buffer_size = bs; + } + + pre_block_size = (data->block_independence)? 0: 64 * 1024; + if (data->in_buffer_size < data->block_size + pre_block_size) { + free(data->in_buffer_allocated); + data->in_buffer_size = data->block_size; + data->in_buffer_allocated = + malloc(data->in_buffer_size + pre_block_size); + data->in_buffer = data->in_buffer_allocated + pre_block_size; + if (!data->block_independence && data->compression_level >= 3) + data->in_buffer = data->in_buffer_allocated; + data->in = data->in_buffer; + data->in_buffer_size = data->block_size; + } + + if (data->out_buffer == NULL || data->in_buffer_allocated == NULL) { + archive_set_error(f->archive, ENOMEM, + "Can't allocate data for compression buffer"); + return (ARCHIVE_FATAL); + } + + f->write = archive_filter_lz4_write; + + return (ARCHIVE_OK); +} + +/* + * Write data to the out stream. + * + * Returns ARCHIVE_OK if all data written, error otherwise. + */ +static int +archive_filter_lz4_write(struct archive_write_filter *f, + const void *buff, size_t length) +{ + struct private_data *data = (struct private_data *)f->data; + int ret = ARCHIVE_OK; + const char *p; + size_t remaining; + ssize_t size; + + /* If we haven't written a stream descriptor, we have to do it first. */ + if (!data->header_written) { + ret = lz4_write_stream_descriptor(f); + if (ret != ARCHIVE_OK) + return (ret); + data->header_written = 1; + } + + /* Update statistics */ + data->total_in += length; + + p = (const char *)buff; + remaining = length; + while (remaining) { + size_t l; + /* Compress input data to output buffer */ + size = lz4_write_one_block(f, p, remaining); + if (size < ARCHIVE_OK) + return (ARCHIVE_FATAL); + l = data->out - data->out_buffer; + if (l >= data->out_block_size) { + ret = __archive_write_filter(f->next_filter, + data->out_buffer, data->out_block_size); + l -= data->out_block_size; + memcpy(data->out_buffer, + data->out_buffer + data->out_block_size, l); + data->out = data->out_buffer + l; + if (ret < ARCHIVE_WARN) + break; + } + p += size; + remaining -= size; + } + + return (ret); +} + +/* + * Finish the compression. + */ +static int +archive_filter_lz4_close(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + int ret; + + /* Finish compression cycle. */ + ret = (int)lz4_write_one_block(f, NULL, 0); + if (ret >= 0) { + /* + * Write the last block and the end of the stream data. + */ + + /* Write End Of Stream. */ + memset(data->out, 0, 4); data->out += 4; + /* Write Stream checksum if needed. */ + if (data->stream_checksum) { + unsigned int checksum; + checksum = __archive_xxhash.XXH32_digest( + data->xxh32_state); + data->xxh32_state = NULL; + archive_le32enc(data->out, checksum); + data->out += 4; + } + ret = __archive_write_filter(f->next_filter, + data->out_buffer, data->out - data->out_buffer); + } + return ret; +} + +static int +archive_filter_lz4_free(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + + if (data->lz4_stream != NULL) { +#ifdef HAVE_LZ4HC_H + if (data->compression_level >= 3) +#if LZ4_VERSION_MAJOR >= 1 && LZ4_VERSION_MINOR >= 7 + LZ4_freeStreamHC(data->lz4_stream); +#else + LZ4_freeHC(data->lz4_stream); +#endif + else +#endif +#if LZ4_VERSION_MINOR >= 3 + LZ4_freeStream(data->lz4_stream); +#else + LZ4_free(data->lz4_stream); +#endif + } + free(data->out_buffer); + free(data->in_buffer_allocated); + free(data->xxh32_state); + free(data); + f->data = NULL; + return (ARCHIVE_OK); +} + +static int +lz4_write_stream_descriptor(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + uint8_t *sd; + + sd = (uint8_t *)data->out; + /* Write Magic Number. */ + archive_le32enc(&sd[0], LZ4_MAGICNUMBER); + /* FLG */ + sd[4] = (data->version_number << 6) + | (data->block_independence << 5) + | (data->block_checksum << 4) + | (data->stream_size << 3) + | (data->stream_checksum << 2) + | (data->preset_dictionary << 0); + /* BD */ + sd[5] = (data->block_maximum_size << 4); + sd[6] = (__archive_xxhash.XXH32(&sd[4], 2, 0) >> 8) & 0xff; + data->out += 7; + if (data->stream_checksum) + data->xxh32_state = __archive_xxhash.XXH32_init(0); + else + data->xxh32_state = NULL; + return (ARCHIVE_OK); +} + +static ssize_t +lz4_write_one_block(struct archive_write_filter *f, const char *p, + size_t length) +{ + struct private_data *data = (struct private_data *)f->data; + ssize_t r; + + if (p == NULL) { + /* Compress remaining uncompressed data. */ + if (data->in_buffer == data->in) + return 0; + else { + size_t l = data->in - data->in_buffer; + r = drive_compressor(f, data->in_buffer, l); + if (r == ARCHIVE_OK) + r = (ssize_t)l; + } + } else if ((data->block_independence || data->compression_level < 3) && + data->in_buffer == data->in && length >= data->block_size) { + r = drive_compressor(f, p, data->block_size); + if (r == ARCHIVE_OK) + r = (ssize_t)data->block_size; + } else { + size_t remaining_size = data->in_buffer_size - + (data->in - data->in_buffer); + size_t l = (remaining_size > length)? length: remaining_size; + memcpy(data->in, p, l); + data->in += l; + if (l == remaining_size) { + r = drive_compressor(f, data->in_buffer, + data->block_size); + if (r == ARCHIVE_OK) + r = (ssize_t)l; + data->in = data->in_buffer; + } else + r = (ssize_t)l; + } + + return (r); +} + + +/* + * Utility function to push input data through compressor, writing + * full output blocks as necessary. + * + * Note that this handles both the regular write case (finishing == + * false) and the end-of-archive case (finishing == true). + */ +static int +drive_compressor(struct archive_write_filter *f, const char *p, size_t length) +{ + struct private_data *data = (struct private_data *)f->data; + + if (data->stream_checksum) + __archive_xxhash.XXH32_update(data->xxh32_state, + p, (int)length); + if (data->block_independence) + return drive_compressor_independence(f, p, length); + else + return drive_compressor_dependence(f, p, length); +} + +static int +drive_compressor_independence(struct archive_write_filter *f, const char *p, + size_t length) +{ + struct private_data *data = (struct private_data *)f->data; + unsigned int outsize; + +#ifdef HAVE_LZ4HC_H + if (data->compression_level >= 3) +#if LZ4_VERSION_MAJOR >= 1 && LZ4_VERSION_MINOR >= 7 + outsize = LZ4_compress_HC(p, data->out + 4, + (int)length, (int)data->block_size, + data->compression_level); +#else + outsize = LZ4_compressHC2_limitedOutput(p, data->out + 4, + (int)length, (int)data->block_size, + data->compression_level); +#endif + else +#endif +#if LZ4_VERSION_MAJOR >= 1 && LZ4_VERSION_MINOR >= 7 + outsize = LZ4_compress_default(p, data->out + 4, + (int)length, (int)data->block_size); +#else + outsize = LZ4_compress_limitedOutput(p, data->out + 4, + (int)length, (int)data->block_size); +#endif + + if (outsize) { + /* The buffer is compressed. */ + archive_le32enc(data->out, outsize); + data->out += 4; + } else { + /* The buffer is not compressed. The compressed size was + * bigger than its uncompressed size. */ + archive_le32enc(data->out, length | 0x80000000); + data->out += 4; + memcpy(data->out, p, length); + outsize = length; + } + data->out += outsize; + if (data->block_checksum) { + unsigned int checksum = + __archive_xxhash.XXH32(data->out - outsize, outsize, 0); + archive_le32enc(data->out, checksum); + data->out += 4; + } + return (ARCHIVE_OK); +} + +static int +drive_compressor_dependence(struct archive_write_filter *f, const char *p, + size_t length) +{ + struct private_data *data = (struct private_data *)f->data; + int outsize; + +#define DICT_SIZE (64 * 1024) +#ifdef HAVE_LZ4HC_H + if (data->compression_level >= 3) { + if (data->lz4_stream == NULL) { +#if LZ4_VERSION_MAJOR >= 1 && LZ4_VERSION_MINOR >= 7 + data->lz4_stream = LZ4_createStreamHC(); + LZ4_resetStreamHC(data->lz4_stream, data->compression_level); +#else + data->lz4_stream = + LZ4_createHC(data->in_buffer_allocated); +#endif + if (data->lz4_stream == NULL) { + archive_set_error(f->archive, ENOMEM, + "Can't allocate data for compression" + " buffer"); + return (ARCHIVE_FATAL); + } + } + else + LZ4_loadDictHC(data->lz4_stream, data->in_buffer_allocated, DICT_SIZE); + +#if LZ4_VERSION_MAJOR >= 1 && LZ4_VERSION_MINOR >= 7 + outsize = LZ4_compress_HC_continue( + data->lz4_stream, p, data->out + 4, (int)length, + (int)data->block_size); +#else + outsize = LZ4_compressHC2_limitedOutput_continue( + data->lz4_stream, p, data->out + 4, (int)length, + (int)data->block_size, data->compression_level); +#endif + } else +#endif + { + if (data->lz4_stream == NULL) { + data->lz4_stream = LZ4_createStream(); + if (data->lz4_stream == NULL) { + archive_set_error(f->archive, ENOMEM, + "Can't allocate data for compression" + " buffer"); + return (ARCHIVE_FATAL); + } + } + else + LZ4_loadDict(data->lz4_stream, data->in_buffer_allocated, DICT_SIZE); + +#if LZ4_VERSION_MAJOR >= 1 && LZ4_VERSION_MINOR >= 7 + outsize = LZ4_compress_fast_continue( + data->lz4_stream, p, data->out + 4, (int)length, + (int)data->block_size, 1); +#else + outsize = LZ4_compress_limitedOutput_continue( + data->lz4_stream, p, data->out + 4, (int)length, + (int)data->block_size); +#endif + } + + if (outsize) { + /* The buffer is compressed. */ + archive_le32enc(data->out, outsize); + data->out += 4; + } else { + /* The buffer is not compressed. The compressed size was + * bigger than its uncompressed size. */ + archive_le32enc(data->out, length | 0x80000000); + data->out += 4; + memcpy(data->out, p, length); + outsize = length; + } + data->out += outsize; + if (data->block_checksum) { + unsigned int checksum = + __archive_xxhash.XXH32(data->out - outsize, outsize, 0); + archive_le32enc(data->out, checksum); + data->out += 4; + } + + if (length == data->block_size) { +#ifdef HAVE_LZ4HC_H + if (data->compression_level >= 3) { +#if LZ4_VERSION_MAJOR >= 1 && LZ4_VERSION_MINOR >= 7 + LZ4_saveDictHC(data->lz4_stream, data->in_buffer_allocated, DICT_SIZE); +#else + LZ4_slideInputBufferHC(data->lz4_stream); +#endif + data->in_buffer = data->in_buffer_allocated + DICT_SIZE; + } + else +#endif + LZ4_saveDict(data->lz4_stream, + data->in_buffer_allocated, DICT_SIZE); +#undef DICT_SIZE + } + return (ARCHIVE_OK); +} + +#else /* HAVE_LIBLZ4 */ + +static int +archive_filter_lz4_open(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + struct archive_string as; + int r; + + archive_string_init(&as); + archive_strcpy(&as, "lz4 -z -q -q"); + + /* Specify a compression level. */ + if (data->compression_level > 0) { + archive_strcat(&as, " -"); + archive_strappend_char(&as, '0' + data->compression_level); + } + /* Specify a block size. */ + archive_strcat(&as, " -B"); + archive_strappend_char(&as, '0' + data->block_maximum_size); + + if (data->block_checksum) + archive_strcat(&as, " -BX"); + if (data->stream_checksum == 0) + archive_strcat(&as, " --no-frame-crc"); + if (data->block_independence == 0) + archive_strcat(&as, " -BD"); + + f->write = archive_filter_lz4_write; + + r = __archive_write_program_open(f, data->pdata, as.s); + archive_string_free(&as); + return (r); +} + +static int +archive_filter_lz4_write(struct archive_write_filter *f, const void *buff, + size_t length) +{ + struct private_data *data = (struct private_data *)f->data; + + return __archive_write_program_write(f, data->pdata, buff, length); +} + +static int +archive_filter_lz4_close(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + + return __archive_write_program_close(f, data->pdata); +} + +static int +archive_filter_lz4_free(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + + __archive_write_program_free(data->pdata); + free(data); + return (ARCHIVE_OK); +} + +#endif /* HAVE_LIBLZ4 */ diff --git a/c/archive_write_add_filter_lzop.c b/c/archive_write_add_filter_lzop.c new file mode 100644 --- /dev/null +++ b/c/archive_write_add_filter_lzop.c @@ -0,0 +1,478 @@ +/*- + * Copyright (c) 2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +__FBSDID("$FreeBSD$"); +//#undef HAVE_LZO_LZOCONF_H +//#undef HAVE_LZO_LZO1X_H + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#include +#ifdef HAVE_LZO_LZOCONF_H +#include +#endif +#ifdef HAVE_LZO_LZO1X_H +#include +#endif + +#include "archive.h" +#include "archive_string.h" +#include "archive_endian.h" +#include "archive_write_private.h" + +enum lzo_method { + METHOD_LZO1X_1 = 1, + METHOD_LZO1X_1_15 = 2, + METHOD_LZO1X_999 = 3 +}; +struct write_lzop { + int compression_level; +#if defined(HAVE_LZO_LZOCONF_H) && defined(HAVE_LZO_LZO1X_H) + unsigned char *uncompressed; + size_t uncompressed_buffer_size; + size_t uncompressed_avail_bytes; + unsigned char *compressed; + size_t compressed_buffer_size; + enum lzo_method method; + unsigned char level; + lzo_voidp work_buffer; + lzo_uint32 work_buffer_size; + char header_written; +#else + struct archive_write_program_data *pdata; +#endif +}; + +static int archive_write_lzop_open(struct archive_write_filter *); +static int archive_write_lzop_options(struct archive_write_filter *, + const char *, const char *); +static int archive_write_lzop_write(struct archive_write_filter *, + const void *, size_t); +static int archive_write_lzop_close(struct archive_write_filter *); +static int archive_write_lzop_free(struct archive_write_filter *); + +#if defined(HAVE_LZO_LZOCONF_H) && defined(HAVE_LZO_LZO1X_H) +/* Maximum block size. */ +#define BLOCK_SIZE (256 * 1024) +/* Block information is composed of uncompressed size(4 bytes), + * compressed size(4 bytes) and the checksum of uncompressed data(4 bytes) + * in this lzop writer. */ +#define BLOCK_INfO_SIZE 12 + +#define HEADER_VERSION 9 +#define HEADER_LIBVERSION 11 +#define HEADER_METHOD 15 +#define HEADER_LEVEL 16 +#define HEADER_MTIME_LOW 25 +#define HEADER_MTIME_HIGH 29 +#define HEADER_H_CHECKSUM 34 + +/* + * Header template. + */ +static const unsigned char header[] = { + /* LZOP Magic code 9 bytes */ + 0x89, 0x4c, 0x5a, 0x4f, 0x00, 0x0d, 0x0a, 0x1a, 0x0a, + /* LZOP utility version(fake data) 2 bytes */ + 0x10, 0x30, + /* LZO library version 2 bytes */ + 0x09, 0x40, + /* Minimum required LZO library version 2 bytes */ + 0x09, 0x40, + /* Method */ + 1, + /* Level */ + 5, + /* Flags 4 bytes + * -OS Unix + * -Stdout + * -Stdin + * -Adler32 used for uncompressed data 4 bytes */ + 0x03, 0x00, 0x00, 0x0d, + /* Mode (AE_IFREG | 0644) 4 bytes */ + 0x00, 0x00, 0x81, 0xa4, + /* Mtime low 4 bytes */ + 0x00, 0x00, 0x00, 0x00, + /* Mtime high 4 bytes */ + 0x00, 0x00, 0x00, 0x00, + /* Filename length */ + 0x00, + /* Header checksum 4 bytes */ + 0x00, 0x00, 0x00, 0x00, +}; +#endif + +int +archive_write_add_filter_lzop(struct archive *_a) +{ + struct archive_write_filter *f = __archive_write_allocate_filter(_a); + struct write_lzop *data; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_add_filter_lzop"); + + data = calloc(1, sizeof(*data)); + if (data == NULL) { + archive_set_error(_a, ENOMEM, "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + + f->name = "lzop"; + f->code = ARCHIVE_FILTER_LZOP; + f->data = data; + f->open = archive_write_lzop_open; + f->options = archive_write_lzop_options; + f->write = archive_write_lzop_write; + f->close = archive_write_lzop_close; + f->free = archive_write_lzop_free; +#if defined(HAVE_LZO_LZOCONF_H) && defined(HAVE_LZO_LZO1X_H) + if (lzo_init() != LZO_E_OK) { + free(data); + archive_set_error(_a, ARCHIVE_ERRNO_MISC, + "lzo_init(type check) failed"); + return (ARCHIVE_FATAL); + } + if (lzo_version() < 0x940) { + free(data); + archive_set_error(_a, ARCHIVE_ERRNO_MISC, + "liblzo library is too old(%s < 0.940)", + lzo_version_string()); + return (ARCHIVE_FATAL); + } + data->compression_level = 5; + return (ARCHIVE_OK); +#else + data->pdata = __archive_write_program_allocate("lzop"); + if (data->pdata == NULL) { + free(data); + archive_set_error(_a, ENOMEM, "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + data->compression_level = 0; + /* Note: We return "warn" to inform of using an external lzop + * program. */ + archive_set_error(_a, ARCHIVE_ERRNO_MISC, + "Using external lzop program for lzop compression"); + return (ARCHIVE_WARN); +#endif +} + +static int +archive_write_lzop_free(struct archive_write_filter *f) +{ + struct write_lzop *data = (struct write_lzop *)f->data; + +#if defined(HAVE_LZO_LZOCONF_H) && defined(HAVE_LZO_LZO1X_H) + free(data->uncompressed); + free(data->compressed); + free(data->work_buffer); +#else + __archive_write_program_free(data->pdata); +#endif + free(data); + return (ARCHIVE_OK); +} + +static int +archive_write_lzop_options(struct archive_write_filter *f, const char *key, + const char *value) +{ + struct write_lzop *data = (struct write_lzop *)f->data; + + if (strcmp(key, "compression-level") == 0) { + if (value == NULL || !(value[0] >= '1' && value[0] <= '9') || + value[1] != '\0') + return (ARCHIVE_WARN); + data->compression_level = value[0] - '0'; + return (ARCHIVE_OK); + } + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +#if defined(HAVE_LZO_LZOCONF_H) && defined(HAVE_LZO_LZO1X_H) +static int +archive_write_lzop_open(struct archive_write_filter *f) +{ + struct write_lzop *data = (struct write_lzop *)f->data; + + switch (data->compression_level) { + case 1: + data->method = METHOD_LZO1X_1_15; data->level = 1; break; + default: + case 2: case 3: case 4: case 5: case 6: + data->method = METHOD_LZO1X_1; data->level = 5; break; + case 7: + data->method = METHOD_LZO1X_999; data->level = 7; break; + case 8: + data->method = METHOD_LZO1X_999; data->level = 8; break; + case 9: + data->method = METHOD_LZO1X_999; data->level = 9; break; + } + switch (data->method) { + case METHOD_LZO1X_1: + data->work_buffer_size = LZO1X_1_MEM_COMPRESS; break; + case METHOD_LZO1X_1_15: + data->work_buffer_size = LZO1X_1_15_MEM_COMPRESS; break; + case METHOD_LZO1X_999: + data->work_buffer_size = LZO1X_999_MEM_COMPRESS; break; + } + if (data->work_buffer == NULL) { + data->work_buffer = (lzo_voidp)malloc(data->work_buffer_size); + if (data->work_buffer == NULL) { + archive_set_error(f->archive, ENOMEM, + "Can't allocate data for compression buffer"); + return (ARCHIVE_FATAL); + } + } + if (data->compressed == NULL) { + data->compressed_buffer_size = sizeof(header) + + BLOCK_SIZE + (BLOCK_SIZE >> 4) + 64 + 3; + data->compressed = (unsigned char *) + malloc(data->compressed_buffer_size); + if (data->compressed == NULL) { + archive_set_error(f->archive, ENOMEM, + "Can't allocate data for compression buffer"); + return (ARCHIVE_FATAL); + } + } + if (data->uncompressed == NULL) { + data->uncompressed_buffer_size = BLOCK_SIZE; + data->uncompressed = (unsigned char *) + malloc(data->uncompressed_buffer_size); + if (data->uncompressed == NULL) { + archive_set_error(f->archive, ENOMEM, + "Can't allocate data for compression buffer"); + return (ARCHIVE_FATAL); + } + data->uncompressed_avail_bytes = BLOCK_SIZE; + } + return (ARCHIVE_OK); +} + +static int +make_header(struct archive_write_filter *f) +{ + struct write_lzop *data = (struct write_lzop *)f->data; + int64_t t; + uint32_t checksum; + + memcpy(data->compressed, header, sizeof(header)); + /* Overwrite library version. */ + data->compressed[HEADER_LIBVERSION] = (unsigned char ) + (lzo_version() >> 8) & 0xff; + data->compressed[HEADER_LIBVERSION + 1] = (unsigned char ) + lzo_version() & 0xff; + /* Overwrite method and level. */ + data->compressed[HEADER_METHOD] = (unsigned char)data->method; + data->compressed[HEADER_LEVEL] = data->level; + /* Overwrite mtime with current time. */ + t = (int64_t)time(NULL); + archive_be32enc(&data->compressed[HEADER_MTIME_LOW], + (uint32_t)(t & 0xffffffff)); + archive_be32enc(&data->compressed[HEADER_MTIME_HIGH], + (uint32_t)((t >> 32) & 0xffffffff)); + /* Overwrite header checksum with calculated value. */ + checksum = lzo_adler32(1, data->compressed + HEADER_VERSION, + (lzo_uint)(HEADER_H_CHECKSUM - HEADER_VERSION)); + archive_be32enc(&data->compressed[HEADER_H_CHECKSUM], checksum); + return (sizeof(header)); +} + +static int +drive_compressor(struct archive_write_filter *f) +{ + struct write_lzop *data = (struct write_lzop *)f->data; + unsigned char *p; + const int block_info_bytes = 12; + int header_bytes, r; + lzo_uint usize, csize; + uint32_t checksum; + + if (!data->header_written) { + header_bytes = make_header(f); + data->header_written = 1; + } else + header_bytes = 0; + p = data->compressed; + + usize = (lzo_uint) + (data->uncompressed_buffer_size - data->uncompressed_avail_bytes); + csize = 0; + switch (data->method) { + default: + case METHOD_LZO1X_1: + r = lzo1x_1_compress(data->uncompressed, usize, + p + header_bytes + block_info_bytes, &csize, + data->work_buffer); + break; + case METHOD_LZO1X_1_15: + r = lzo1x_1_15_compress(data->uncompressed, usize, + p + header_bytes + block_info_bytes, &csize, + data->work_buffer); + break; + case METHOD_LZO1X_999: + r = lzo1x_999_compress_level(data->uncompressed, usize, + p + header_bytes + block_info_bytes, &csize, + data->work_buffer, NULL, 0, 0, data->level); + break; + } + if (r != LZO_E_OK) { + archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, + "Lzop compression failed: returned status %d", r); + return (ARCHIVE_FATAL); + } + + /* Store uncompressed size. */ + archive_be32enc(p + header_bytes, (uint32_t)usize); + /* Store the checksum of the uncompressed data. */ + checksum = lzo_adler32(1, data->uncompressed, usize); + archive_be32enc(p + header_bytes + 8, checksum); + + if (csize < usize) { + /* Store compressed size. */ + archive_be32enc(p + header_bytes + 4, (uint32_t)csize); + r = __archive_write_filter(f->next_filter, data->compressed, + header_bytes + block_info_bytes + csize); + } else { + /* + * This case, we output uncompressed data instead. + */ + /* Store uncompressed size as compressed size. */ + archive_be32enc(p + header_bytes + 4, (uint32_t)usize); + r = __archive_write_filter(f->next_filter, data->compressed, + header_bytes + block_info_bytes); + if (r != ARCHIVE_OK) + return (ARCHIVE_FATAL); + r = __archive_write_filter(f->next_filter, data->uncompressed, + usize); + } + + if (r != ARCHIVE_OK) + return (ARCHIVE_FATAL); + return (ARCHIVE_OK); +} + +static int +archive_write_lzop_write(struct archive_write_filter *f, + const void *buff, size_t length) +{ + struct write_lzop *data = (struct write_lzop *)f->data; + const char *p = buff; + int r; + + do { + if (data->uncompressed_avail_bytes > length) { + memcpy(data->uncompressed + + data->uncompressed_buffer_size + - data->uncompressed_avail_bytes, + p, length); + data->uncompressed_avail_bytes -= length; + return (ARCHIVE_OK); + } + + memcpy(data->uncompressed + data->uncompressed_buffer_size + - data->uncompressed_avail_bytes, + p, data->uncompressed_avail_bytes); + length -= data->uncompressed_avail_bytes; + p += data->uncompressed_avail_bytes; + data->uncompressed_avail_bytes = 0; + + r = drive_compressor(f); + if (r != ARCHIVE_OK) return (r); + data->uncompressed_avail_bytes = BLOCK_SIZE; + } while (length); + + return (ARCHIVE_OK); +} + +static int +archive_write_lzop_close(struct archive_write_filter *f) +{ + struct write_lzop *data = (struct write_lzop *)f->data; + const uint32_t endmark = 0; + int r; + + if (data->uncompressed_avail_bytes < BLOCK_SIZE) { + /* Compress and output remaining data. */ + r = drive_compressor(f); + if (r != ARCHIVE_OK) + return (r); + } + /* Write a zero uncompressed size as the end mark of the series of + * compressed block. */ + return __archive_write_filter(f->next_filter, &endmark, sizeof(endmark)); +} + +#else +static int +archive_write_lzop_open(struct archive_write_filter *f) +{ + struct write_lzop *data = (struct write_lzop *)f->data; + struct archive_string as; + int r; + + archive_string_init(&as); + archive_strcpy(&as, "lzop"); + /* Specify compression level. */ + if (data->compression_level > 0) { + archive_strappend_char(&as, ' '); + archive_strappend_char(&as, '-'); + archive_strappend_char(&as, '0' + data->compression_level); + } + + r = __archive_write_program_open(f, data->pdata, as.s); + archive_string_free(&as); + return (r); +} + +static int +archive_write_lzop_write(struct archive_write_filter *f, + const void *buff, size_t length) +{ + struct write_lzop *data = (struct write_lzop *)f->data; + + return __archive_write_program_write(f, data->pdata, buff, length); +} + +static int +archive_write_lzop_close(struct archive_write_filter *f) +{ + struct write_lzop *data = (struct write_lzop *)f->data; + + return __archive_write_program_close(f, data->pdata); +} +#endif diff --git a/c/archive_write_add_filter_none.c b/c/archive_write_add_filter_none.c new file mode 100644 --- /dev/null +++ b/c/archive_write_add_filter_none.c @@ -0,0 +1,43 @@ +/*- + * Copyright (c) 2003-2010 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_compression_none.c 201080 2009-12-28 02:03:54Z kientzle $"); + +#include "archive.h" + +int +archive_write_set_compression_none(struct archive *a) +{ + (void)a; /* UNUSED */ + return (ARCHIVE_OK); +} + +int +archive_write_add_filter_none(struct archive *a) +{ + (void)a; /* UNUSED */ + return (ARCHIVE_OK); +} diff --git a/c/archive_write_add_filter_program.c b/c/archive_write_add_filter_program.c new file mode 100644 --- /dev/null +++ b/c/archive_write_add_filter_program.c @@ -0,0 +1,391 @@ +/*- + * Copyright (c) 2007 Joerg Sonnenberger + * Copyright (c) 2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_compression_program.c 201104 2009-12-28 03:14:30Z kientzle $"); + +#ifdef HAVE_SYS_WAIT_H +# include +#endif +#ifdef HAVE_ERRNO_H +# include +#endif +#ifdef HAVE_FCNTL_H +# include +#endif +#ifdef HAVE_STDLIB_H +# include +#endif +#ifdef HAVE_STRING_H +# include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_string.h" +#include "archive_write_private.h" +#include "filter_fork.h" + +#if ARCHIVE_VERSION_NUMBER < 4000000 +int +archive_write_set_compression_program(struct archive *a, const char *cmd) +{ + __archive_write_filters_free(a); + return (archive_write_add_filter_program(a, cmd)); +} +#endif + +struct archive_write_program_data { +#if defined(_WIN32) && !defined(__CYGWIN__) + HANDLE child; +#else + pid_t child; +#endif + int child_stdin, child_stdout; + + char *child_buf; + size_t child_buf_len, child_buf_avail; + char *program_name; +}; + +struct private_data { + struct archive_write_program_data *pdata; + struct archive_string description; + char *cmd; +}; + +static int archive_compressor_program_open(struct archive_write_filter *); +static int archive_compressor_program_write(struct archive_write_filter *, + const void *, size_t); +static int archive_compressor_program_close(struct archive_write_filter *); +static int archive_compressor_program_free(struct archive_write_filter *); + +/* + * Add a filter to this write handle that passes all data through an + * external program. + */ +int +archive_write_add_filter_program(struct archive *_a, const char *cmd) +{ + struct archive_write_filter *f = __archive_write_allocate_filter(_a); + struct private_data *data; + static const char prefix[] = "Program: "; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_add_filter_program"); + + f->data = calloc(1, sizeof(*data)); + if (f->data == NULL) + goto memerr; + data = (struct private_data *)f->data; + + data->cmd = strdup(cmd); + if (data->cmd == NULL) + goto memerr; + + data->pdata = __archive_write_program_allocate(cmd); + if (data->pdata == NULL) + goto memerr; + + /* Make up a description string. */ + if (archive_string_ensure(&data->description, + strlen(prefix) + strlen(cmd) + 1) == NULL) + goto memerr; + archive_strcpy(&data->description, prefix); + archive_strcat(&data->description, cmd); + + f->name = data->description.s; + f->code = ARCHIVE_FILTER_PROGRAM; + f->open = archive_compressor_program_open; + f->write = archive_compressor_program_write; + f->close = archive_compressor_program_close; + f->free = archive_compressor_program_free; + return (ARCHIVE_OK); +memerr: + archive_compressor_program_free(f); + archive_set_error(_a, ENOMEM, + "Can't allocate memory for filter program"); + return (ARCHIVE_FATAL); +} + +static int +archive_compressor_program_open(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + + return __archive_write_program_open(f, data->pdata, data->cmd); +} + +static int +archive_compressor_program_write(struct archive_write_filter *f, + const void *buff, size_t length) +{ + struct private_data *data = (struct private_data *)f->data; + + return __archive_write_program_write(f, data->pdata, buff, length); +} + +static int +archive_compressor_program_close(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + + return __archive_write_program_close(f, data->pdata); +} + +static int +archive_compressor_program_free(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + + if (data) { + free(data->cmd); + archive_string_free(&data->description); + __archive_write_program_free(data->pdata); + free(data); + f->data = NULL; + } + return (ARCHIVE_OK); +} + +/* + * Allocate resources for executing an external program. + */ +struct archive_write_program_data * +__archive_write_program_allocate(const char *program) +{ + struct archive_write_program_data *data; + + data = calloc(1, sizeof(struct archive_write_program_data)); + if (data == NULL) + return (data); + data->child_stdin = -1; + data->child_stdout = -1; + data->program_name = strdup(program); + return (data); +} + +/* + * Release the resources. + */ +int +__archive_write_program_free(struct archive_write_program_data *data) +{ + + if (data) { + free(data->program_name); + free(data->child_buf); + free(data); + } + return (ARCHIVE_OK); +} + +int +__archive_write_program_open(struct archive_write_filter *f, + struct archive_write_program_data *data, const char *cmd) +{ + int ret; + + if (data->child_buf == NULL) { + data->child_buf_len = 65536; + data->child_buf_avail = 0; + data->child_buf = malloc(data->child_buf_len); + + if (data->child_buf == NULL) { + archive_set_error(f->archive, ENOMEM, + "Can't allocate compression buffer"); + return (ARCHIVE_FATAL); + } + } + + ret = __archive_create_child(cmd, &data->child_stdin, + &data->child_stdout, &data->child); + if (ret != ARCHIVE_OK) { + archive_set_error(f->archive, EINVAL, + "Can't launch external program: %s", cmd); + return (ARCHIVE_FATAL); + } + return (ARCHIVE_OK); +} + +static ssize_t +child_write(struct archive_write_filter *f, + struct archive_write_program_data *data, const char *buf, size_t buf_len) +{ + ssize_t ret; + + if (data->child_stdin == -1) + return (-1); + + if (buf_len == 0) + return (-1); + + for (;;) { + do { + ret = write(data->child_stdin, buf, buf_len); + } while (ret == -1 && errno == EINTR); + + if (ret > 0) + return (ret); + if (ret == 0) { + close(data->child_stdin); + data->child_stdin = -1; + fcntl(data->child_stdout, F_SETFL, 0); + return (0); + } + if (ret == -1 && errno != EAGAIN) + return (-1); + + if (data->child_stdout == -1) { + fcntl(data->child_stdin, F_SETFL, 0); + __archive_check_child(data->child_stdin, + data->child_stdout); + continue; + } + + do { + ret = read(data->child_stdout, + data->child_buf + data->child_buf_avail, + data->child_buf_len - data->child_buf_avail); + } while (ret == -1 && errno == EINTR); + + if (ret == 0 || (ret == -1 && errno == EPIPE)) { + close(data->child_stdout); + data->child_stdout = -1; + fcntl(data->child_stdin, F_SETFL, 0); + continue; + } + if (ret == -1 && errno == EAGAIN) { + __archive_check_child(data->child_stdin, + data->child_stdout); + continue; + } + if (ret == -1) + return (-1); + + data->child_buf_avail += ret; + + ret = __archive_write_filter(f->next_filter, + data->child_buf, data->child_buf_avail); + if (ret != ARCHIVE_OK) + return (-1); + data->child_buf_avail = 0; + } +} + +/* + * Write data to the filter stream. + */ +int +__archive_write_program_write(struct archive_write_filter *f, + struct archive_write_program_data *data, const void *buff, size_t length) +{ + ssize_t ret; + const char *buf; + + if (data->child == 0) + return (ARCHIVE_OK); + + buf = buff; + while (length > 0) { + ret = child_write(f, data, buf, length); + if (ret == -1 || ret == 0) { + archive_set_error(f->archive, EIO, + "Can't write to program: %s", data->program_name); + return (ARCHIVE_FATAL); + } + length -= ret; + buf += ret; + } + return (ARCHIVE_OK); +} + +/* + * Finish the filtering... + */ +int +__archive_write_program_close(struct archive_write_filter *f, + struct archive_write_program_data *data) +{ + int ret, status; + ssize_t bytes_read; + + if (data->child == 0) + return ARCHIVE_OK; + + ret = 0; + close(data->child_stdin); + data->child_stdin = -1; + fcntl(data->child_stdout, F_SETFL, 0); + + for (;;) { + do { + bytes_read = read(data->child_stdout, + data->child_buf + data->child_buf_avail, + data->child_buf_len - data->child_buf_avail); + } while (bytes_read == -1 && errno == EINTR); + + if (bytes_read == 0 || (bytes_read == -1 && errno == EPIPE)) + break; + + if (bytes_read == -1) { + archive_set_error(f->archive, errno, + "Error reading from program: %s", data->program_name); + ret = ARCHIVE_FATAL; + goto cleanup; + } + data->child_buf_avail += bytes_read; + + ret = __archive_write_filter(f->next_filter, + data->child_buf, data->child_buf_avail); + if (ret != ARCHIVE_OK) { + ret = ARCHIVE_FATAL; + goto cleanup; + } + data->child_buf_avail = 0; + } + +cleanup: + /* Shut down the child. */ + if (data->child_stdin != -1) + close(data->child_stdin); + if (data->child_stdout != -1) + close(data->child_stdout); + while (waitpid(data->child, &status, 0) == -1 && errno == EINTR) + continue; +#if defined(_WIN32) && !defined(__CYGWIN__) + CloseHandle(data->child); +#endif + data->child = 0; + + if (status != 0) { + archive_set_error(f->archive, EIO, + "Error closing program: %s", data->program_name); + ret = ARCHIVE_FATAL; + } + return ret; +} + diff --git a/c/archive_write_add_filter_uuencode.c b/c/archive_write_add_filter_uuencode.c new file mode 100644 --- /dev/null +++ b/c/archive_write_add_filter_uuencode.c @@ -0,0 +1,295 @@ +/*- + * Copyright (c) 2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_string.h" +#include "archive_write_private.h" + +#define LBYTES 45 + +struct private_uuencode { + int mode; + struct archive_string name; + struct archive_string encoded_buff; + size_t bs; + size_t hold_len; + unsigned char hold[LBYTES]; +}; + +static int archive_filter_uuencode_options(struct archive_write_filter *, + const char *, const char *); +static int archive_filter_uuencode_open(struct archive_write_filter *); +static int archive_filter_uuencode_write(struct archive_write_filter *, + const void *, size_t); +static int archive_filter_uuencode_close(struct archive_write_filter *); +static int archive_filter_uuencode_free(struct archive_write_filter *); +static void uu_encode(struct archive_string *, const unsigned char *, size_t); +static int64_t atol8(const char *, size_t); + +/* + * Add a compress filter to this write handle. + */ +int +archive_write_add_filter_uuencode(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct archive_write_filter *f = __archive_write_allocate_filter(_a); + struct private_uuencode *state; + + archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_add_filter_uu"); + + state = (struct private_uuencode *)calloc(1, sizeof(*state)); + if (state == NULL) { + archive_set_error(f->archive, ENOMEM, + "Can't allocate data for uuencode filter"); + return (ARCHIVE_FATAL); + } + archive_strcpy(&state->name, "-"); + state->mode = 0644; + + f->data = state; + f->name = "uuencode"; + f->code = ARCHIVE_FILTER_UU; + f->open = archive_filter_uuencode_open; + f->options = archive_filter_uuencode_options; + f->write = archive_filter_uuencode_write; + f->close = archive_filter_uuencode_close; + f->free = archive_filter_uuencode_free; + + return (ARCHIVE_OK); +} + +/* + * Set write options. + */ +static int +archive_filter_uuencode_options(struct archive_write_filter *f, const char *key, + const char *value) +{ + struct private_uuencode *state = (struct private_uuencode *)f->data; + + if (strcmp(key, "mode") == 0) { + if (value == NULL) { + archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, + "mode option requires octal digits"); + return (ARCHIVE_FAILED); + } + state->mode = (int)atol8(value, strlen(value)) & 0777; + return (ARCHIVE_OK); + } else if (strcmp(key, "name") == 0) { + if (value == NULL) { + archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, + "name option requires a string"); + return (ARCHIVE_FAILED); + } + archive_strcpy(&state->name, value); + return (ARCHIVE_OK); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +/* + * Setup callback. + */ +static int +archive_filter_uuencode_open(struct archive_write_filter *f) +{ + struct private_uuencode *state = (struct private_uuencode *)f->data; + size_t bs = 65536, bpb; + + if (f->archive->magic == ARCHIVE_WRITE_MAGIC) { + /* Buffer size should be a multiple number of the of bytes + * per block for performance. */ + bpb = archive_write_get_bytes_per_block(f->archive); + if (bpb > bs) + bs = bpb; + else if (bpb != 0) + bs -= bs % bpb; + } + + state->bs = bs; + if (archive_string_ensure(&state->encoded_buff, bs + 512) == NULL) { + archive_set_error(f->archive, ENOMEM, + "Can't allocate data for uuencode buffer"); + return (ARCHIVE_FATAL); + } + + archive_string_sprintf(&state->encoded_buff, "begin %o %s\n", + state->mode, state->name.s); + + f->data = state; + return (0); +} + +static void +uu_encode(struct archive_string *as, const unsigned char *p, size_t len) +{ + int c; + + c = (int)len; + archive_strappend_char(as, c?c + 0x20:'`'); + for (; len >= 3; p += 3, len -= 3) { + c = p[0] >> 2; + archive_strappend_char(as, c?c + 0x20:'`'); + c = ((p[0] & 0x03) << 4) | ((p[1] & 0xf0) >> 4); + archive_strappend_char(as, c?c + 0x20:'`'); + c = ((p[1] & 0x0f) << 2) | ((p[2] & 0xc0) >> 6); + archive_strappend_char(as, c?c + 0x20:'`'); + c = p[2] & 0x3f; + archive_strappend_char(as, c?c + 0x20:'`'); + } + if (len > 0) { + c = p[0] >> 2; + archive_strappend_char(as, c?c + 0x20:'`'); + c = (p[0] & 0x03) << 4; + if (len == 1) { + archive_strappend_char(as, c?c + 0x20:'`'); + archive_strappend_char(as, '`'); + archive_strappend_char(as, '`'); + } else { + c |= (p[1] & 0xf0) >> 4; + archive_strappend_char(as, c?c + 0x20:'`'); + c = (p[1] & 0x0f) << 2; + archive_strappend_char(as, c?c + 0x20:'`'); + archive_strappend_char(as, '`'); + } + } + archive_strappend_char(as, '\n'); +} + +/* + * Write data to the encoded stream. + */ +static int +archive_filter_uuencode_write(struct archive_write_filter *f, const void *buff, + size_t length) +{ + struct private_uuencode *state = (struct private_uuencode *)f->data; + const unsigned char *p = buff; + int ret = ARCHIVE_OK; + + if (length == 0) + return (ret); + + if (state->hold_len) { + while (state->hold_len < LBYTES && length > 0) { + state->hold[state->hold_len++] = *p++; + length--; + } + if (state->hold_len < LBYTES) + return (ret); + uu_encode(&state->encoded_buff, state->hold, LBYTES); + state->hold_len = 0; + } + + for (; length >= LBYTES; length -= LBYTES, p += LBYTES) + uu_encode(&state->encoded_buff, p, LBYTES); + + /* Save remaining bytes. */ + if (length > 0) { + memcpy(state->hold, p, length); + state->hold_len = length; + } + while (archive_strlen(&state->encoded_buff) >= state->bs) { + ret = __archive_write_filter(f->next_filter, + state->encoded_buff.s, state->bs); + memmove(state->encoded_buff.s, + state->encoded_buff.s + state->bs, + state->encoded_buff.length - state->bs); + state->encoded_buff.length -= state->bs; + } + + return (ret); +} + + +/* + * Finish the compression... + */ +static int +archive_filter_uuencode_close(struct archive_write_filter *f) +{ + struct private_uuencode *state = (struct private_uuencode *)f->data; + + /* Flush remaining bytes. */ + if (state->hold_len != 0) + uu_encode(&state->encoded_buff, state->hold, state->hold_len); + archive_string_sprintf(&state->encoded_buff, "`\nend\n"); + /* Write the last block */ + archive_write_set_bytes_in_last_block(f->archive, 1); + return __archive_write_filter(f->next_filter, + state->encoded_buff.s, archive_strlen(&state->encoded_buff)); +} + +static int +archive_filter_uuencode_free(struct archive_write_filter *f) +{ + struct private_uuencode *state = (struct private_uuencode *)f->data; + + archive_string_free(&state->name); + archive_string_free(&state->encoded_buff); + free(state); + return (ARCHIVE_OK); +} + +static int64_t +atol8(const char *p, size_t char_cnt) +{ + int64_t l; + int digit; + + l = 0; + while (char_cnt-- > 0) { + if (*p >= '0' && *p <= '7') + digit = *p - '0'; + else + break; + p++; + l <<= 3; + l |= digit; + } + return (l); +} + diff --git a/c/archive_write_add_filter_xz.c b/c/archive_write_add_filter_xz.c new file mode 100644 --- /dev/null +++ b/c/archive_write_add_filter_xz.c @@ -0,0 +1,545 @@ +/*- + * Copyright (c) 2003-2010 Tim Kientzle + * Copyright (c) 2009-2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_compression_xz.c 201108 2009-12-28 03:28:21Z kientzle $"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#include +#ifdef HAVE_LZMA_H +#include +#endif + +#include "archive.h" +#include "archive_endian.h" +#include "archive_private.h" +#include "archive_write_private.h" + +#if ARCHIVE_VERSION_NUMBER < 4000000 +int +archive_write_set_compression_lzip(struct archive *a) +{ + __archive_write_filters_free(a); + return (archive_write_add_filter_lzip(a)); +} + +int +archive_write_set_compression_lzma(struct archive *a) +{ + __archive_write_filters_free(a); + return (archive_write_add_filter_lzma(a)); +} + +int +archive_write_set_compression_xz(struct archive *a) +{ + __archive_write_filters_free(a); + return (archive_write_add_filter_xz(a)); +} + +#endif + +#ifndef HAVE_LZMA_H +int +archive_write_add_filter_xz(struct archive *a) +{ + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "xz compression not supported on this platform"); + return (ARCHIVE_FATAL); +} + +int +archive_write_add_filter_lzma(struct archive *a) +{ + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "lzma compression not supported on this platform"); + return (ARCHIVE_FATAL); +} + +int +archive_write_add_filter_lzip(struct archive *a) +{ + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "lzma compression not supported on this platform"); + return (ARCHIVE_FATAL); +} +#else +/* Don't compile this if we don't have liblzma. */ + +struct private_data { + int compression_level; + uint32_t threads; + lzma_stream stream; + lzma_filter lzmafilters[2]; + lzma_options_lzma lzma_opt; + int64_t total_in; + unsigned char *compressed; + size_t compressed_buffer_size; + int64_t total_out; + /* the CRC32 value of uncompressed data for lzip */ + uint32_t crc32; +}; + +static int archive_compressor_xz_options(struct archive_write_filter *, + const char *, const char *); +static int archive_compressor_xz_open(struct archive_write_filter *); +static int archive_compressor_xz_write(struct archive_write_filter *, + const void *, size_t); +static int archive_compressor_xz_close(struct archive_write_filter *); +static int archive_compressor_xz_free(struct archive_write_filter *); +static int drive_compressor(struct archive_write_filter *, + struct private_data *, int finishing); + +struct option_value { + uint32_t dict_size; + uint32_t nice_len; + lzma_match_finder mf; +}; +static const struct option_value option_values[] = { + { 1 << 16, 32, LZMA_MF_HC3}, + { 1 << 20, 32, LZMA_MF_HC3}, + { 3 << 19, 32, LZMA_MF_HC4}, + { 1 << 21, 32, LZMA_MF_BT4}, + { 3 << 20, 32, LZMA_MF_BT4}, + { 1 << 22, 32, LZMA_MF_BT4}, + { 1 << 23, 64, LZMA_MF_BT4}, + { 1 << 24, 64, LZMA_MF_BT4}, + { 3 << 23, 64, LZMA_MF_BT4}, + { 1 << 25, 64, LZMA_MF_BT4} +}; + +static int +common_setup(struct archive_write_filter *f) +{ + struct private_data *data; + struct archive_write *a = (struct archive_write *)f->archive; + data = calloc(1, sizeof(*data)); + if (data == NULL) { + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + f->data = data; + data->compression_level = LZMA_PRESET_DEFAULT; + data->threads = 1; + f->open = &archive_compressor_xz_open; + f->close = archive_compressor_xz_close; + f->free = archive_compressor_xz_free; + f->options = &archive_compressor_xz_options; + return (ARCHIVE_OK); +} + +/* + * Add an xz compression filter to this write handle. + */ +int +archive_write_add_filter_xz(struct archive *_a) +{ + struct archive_write_filter *f; + int r; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_add_filter_xz"); + f = __archive_write_allocate_filter(_a); + r = common_setup(f); + if (r == ARCHIVE_OK) { + f->code = ARCHIVE_FILTER_XZ; + f->name = "xz"; + } + return (r); +} + +/* LZMA is handled identically, we just need a different compression + * code set. (The liblzma setup looks at the code to determine + * the one place that XZ and LZMA require different handling.) */ +int +archive_write_add_filter_lzma(struct archive *_a) +{ + struct archive_write_filter *f; + int r; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_add_filter_lzma"); + f = __archive_write_allocate_filter(_a); + r = common_setup(f); + if (r == ARCHIVE_OK) { + f->code = ARCHIVE_FILTER_LZMA; + f->name = "lzma"; + } + return (r); +} + +int +archive_write_add_filter_lzip(struct archive *_a) +{ + struct archive_write_filter *f; + int r; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_add_filter_lzip"); + f = __archive_write_allocate_filter(_a); + r = common_setup(f); + if (r == ARCHIVE_OK) { + f->code = ARCHIVE_FILTER_LZIP; + f->name = "lzip"; + } + return (r); +} + +static int +archive_compressor_xz_init_stream(struct archive_write_filter *f, + struct private_data *data) +{ + static const lzma_stream lzma_stream_init_data = LZMA_STREAM_INIT; + int ret; +#ifdef HAVE_LZMA_STREAM_ENCODER_MT + lzma_mt mt_options; +#endif + + data->stream = lzma_stream_init_data; + data->stream.next_out = data->compressed; + data->stream.avail_out = data->compressed_buffer_size; + if (f->code == ARCHIVE_FILTER_XZ) { +#ifdef HAVE_LZMA_STREAM_ENCODER_MT + if (data->threads != 1) { + memset(&mt_options, 0, sizeof(mt_options)); + mt_options.threads = data->threads; + mt_options.timeout = 300; + mt_options.filters = data->lzmafilters; + mt_options.check = LZMA_CHECK_CRC64; + ret = lzma_stream_encoder_mt(&(data->stream), + &mt_options); + } else +#endif + ret = lzma_stream_encoder(&(data->stream), + data->lzmafilters, LZMA_CHECK_CRC64); + } else if (f->code == ARCHIVE_FILTER_LZMA) { + ret = lzma_alone_encoder(&(data->stream), &data->lzma_opt); + } else { /* ARCHIVE_FILTER_LZIP */ + int dict_size = data->lzma_opt.dict_size; + int ds, log2dic, wedges; + + /* Calculate a coded dictionary size */ + if (dict_size < (1 << 12) || dict_size > (1 << 27)) { + archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, + "Unacceptable dictionary size for lzip: %d", + dict_size); + return (ARCHIVE_FATAL); + } + for (log2dic = 27; log2dic >= 12; log2dic--) { + if (dict_size & (1 << log2dic)) + break; + } + if (dict_size > (1 << log2dic)) { + log2dic++; + wedges = + ((1 << log2dic) - dict_size) / (1 << (log2dic - 4)); + } else + wedges = 0; + ds = ((wedges << 5) & 0xe0) | (log2dic & 0x1f); + + data->crc32 = 0; + /* Make a header */ + data->compressed[0] = 0x4C; + data->compressed[1] = 0x5A; + data->compressed[2] = 0x49; + data->compressed[3] = 0x50; + data->compressed[4] = 1;/* Version */ + data->compressed[5] = (unsigned char)ds; + data->stream.next_out += 6; + data->stream.avail_out -= 6; + + ret = lzma_raw_encoder(&(data->stream), data->lzmafilters); + } + if (ret == LZMA_OK) + return (ARCHIVE_OK); + + switch (ret) { + case LZMA_MEM_ERROR: + archive_set_error(f->archive, ENOMEM, + "Internal error initializing compression library: " + "Cannot allocate memory"); + break; + default: + archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, + "Internal error initializing compression library: " + "It's a bug in liblzma"); + break; + } + return (ARCHIVE_FATAL); +} + +/* + * Setup callback. + */ +static int +archive_compressor_xz_open(struct archive_write_filter *f) +{ + struct private_data *data = f->data; + int ret; + + if (data->compressed == NULL) { + size_t bs = 65536, bpb; + if (f->archive->magic == ARCHIVE_WRITE_MAGIC) { + /* Buffer size should be a multiple number of the of bytes + * per block for performance. */ + bpb = archive_write_get_bytes_per_block(f->archive); + if (bpb > bs) + bs = bpb; + else if (bpb != 0) + bs -= bs % bpb; + } + data->compressed_buffer_size = bs; + data->compressed + = (unsigned char *)malloc(data->compressed_buffer_size); + if (data->compressed == NULL) { + archive_set_error(f->archive, ENOMEM, + "Can't allocate data for compression buffer"); + return (ARCHIVE_FATAL); + } + } + + f->write = archive_compressor_xz_write; + + /* Initialize compression library. */ + if (f->code == ARCHIVE_FILTER_LZIP) { + const struct option_value *val = + &option_values[data->compression_level]; + + data->lzma_opt.dict_size = val->dict_size; + data->lzma_opt.preset_dict = NULL; + data->lzma_opt.preset_dict_size = 0; + data->lzma_opt.lc = LZMA_LC_DEFAULT; + data->lzma_opt.lp = LZMA_LP_DEFAULT; + data->lzma_opt.pb = LZMA_PB_DEFAULT; + data->lzma_opt.mode = + data->compression_level<= 2? LZMA_MODE_FAST:LZMA_MODE_NORMAL; + data->lzma_opt.nice_len = val->nice_len; + data->lzma_opt.mf = val->mf; + data->lzma_opt.depth = 0; + data->lzmafilters[0].id = LZMA_FILTER_LZMA1; + data->lzmafilters[0].options = &data->lzma_opt; + data->lzmafilters[1].id = LZMA_VLI_UNKNOWN;/* Terminate */ + } else { + if (lzma_lzma_preset(&data->lzma_opt, data->compression_level)) { + archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, + "Internal error initializing compression library"); + } + data->lzmafilters[0].id = LZMA_FILTER_LZMA2; + data->lzmafilters[0].options = &data->lzma_opt; + data->lzmafilters[1].id = LZMA_VLI_UNKNOWN;/* Terminate */ + } + ret = archive_compressor_xz_init_stream(f, data); + if (ret == LZMA_OK) { + f->data = data; + return (0); + } + return (ARCHIVE_FATAL); +} + +/* + * Set write options. + */ +static int +archive_compressor_xz_options(struct archive_write_filter *f, + const char *key, const char *value) +{ + struct private_data *data = (struct private_data *)f->data; + + if (strcmp(key, "compression-level") == 0) { + if (value == NULL || !(value[0] >= '0' && value[0] <= '9') || + value[1] != '\0') + return (ARCHIVE_WARN); + data->compression_level = value[0] - '0'; + if (data->compression_level > 6) + data->compression_level = 6; + return (ARCHIVE_OK); + } else if (strcmp(key, "threads") == 0) { + char *endptr; + + if (value == NULL) + return (ARCHIVE_WARN); + errno = 0; + data->threads = (int)strtoul(value, &endptr, 10); + if (errno != 0 || *endptr != '\0') { + data->threads = 1; + return (ARCHIVE_WARN); + } + if (data->threads == 0) { +#ifdef HAVE_LZMA_STREAM_ENCODER_MT + data->threads = lzma_cputhreads(); +#else + data->threads = 1; +#endif + } + return (ARCHIVE_OK); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +/* + * Write data to the compressed stream. + */ +static int +archive_compressor_xz_write(struct archive_write_filter *f, + const void *buff, size_t length) +{ + struct private_data *data = (struct private_data *)f->data; + int ret; + + /* Update statistics */ + data->total_in += length; + if (f->code == ARCHIVE_FILTER_LZIP) + data->crc32 = lzma_crc32(buff, length, data->crc32); + + /* Compress input data to output buffer */ + data->stream.next_in = buff; + data->stream.avail_in = length; + if ((ret = drive_compressor(f, data, 0)) != ARCHIVE_OK) + return (ret); + + return (ARCHIVE_OK); +} + + +/* + * Finish the compression... + */ +static int +archive_compressor_xz_close(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + int ret; + + ret = drive_compressor(f, data, 1); + if (ret == ARCHIVE_OK) { + data->total_out += + data->compressed_buffer_size - data->stream.avail_out; + ret = __archive_write_filter(f->next_filter, + data->compressed, + data->compressed_buffer_size - data->stream.avail_out); + if (f->code == ARCHIVE_FILTER_LZIP && ret == ARCHIVE_OK) { + archive_le32enc(data->compressed, data->crc32); + archive_le64enc(data->compressed+4, data->total_in); + archive_le64enc(data->compressed+12, data->total_out + 20); + ret = __archive_write_filter(f->next_filter, + data->compressed, 20); + } + } + lzma_end(&(data->stream)); + return ret; +} + +static int +archive_compressor_xz_free(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + free(data->compressed); + free(data); + f->data = NULL; + return (ARCHIVE_OK); +} + +/* + * Utility function to push input data through compressor, + * writing full output blocks as necessary. + * + * Note that this handles both the regular write case (finishing == + * false) and the end-of-archive case (finishing == true). + */ +static int +drive_compressor(struct archive_write_filter *f, + struct private_data *data, int finishing) +{ + int ret; + + for (;;) { + if (data->stream.avail_out == 0) { + data->total_out += data->compressed_buffer_size; + ret = __archive_write_filter(f->next_filter, + data->compressed, + data->compressed_buffer_size); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + data->stream.next_out = data->compressed; + data->stream.avail_out = data->compressed_buffer_size; + } + + /* If there's nothing to do, we're done. */ + if (!finishing && data->stream.avail_in == 0) + return (ARCHIVE_OK); + + ret = lzma_code(&(data->stream), + finishing ? LZMA_FINISH : LZMA_RUN ); + + switch (ret) { + case LZMA_OK: + /* In non-finishing case, check if compressor + * consumed everything */ + if (!finishing && data->stream.avail_in == 0) + return (ARCHIVE_OK); + /* In finishing case, this return always means + * there's more work */ + break; + case LZMA_STREAM_END: + /* This return can only occur in finishing case. */ + if (finishing) + return (ARCHIVE_OK); + archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, + "lzma compression data error"); + return (ARCHIVE_FATAL); + case LZMA_MEMLIMIT_ERROR: + archive_set_error(f->archive, ENOMEM, + "lzma compression error: " + "%ju MiB would have been needed", + (uintmax_t)((lzma_memusage(&(data->stream)) + + 1024 * 1024 -1) + / (1024 * 1024))); + return (ARCHIVE_FATAL); + default: + /* Any other return value indicates an error. */ + archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, + "lzma compression failed:" + " lzma_code() call returned status %d", + ret); + return (ARCHIVE_FATAL); + } + } +} + +#endif /* HAVE_LZMA_H */ diff --git a/c/archive_write_add_filter_zstd.c b/c/archive_write_add_filter_zstd.c new file mode 100644 --- /dev/null +++ b/c/archive_write_add_filter_zstd.c @@ -0,0 +1,392 @@ +/*- + * Copyright (c) 2017 Sean Purcell + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +__FBSDID("$FreeBSD$"); + + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_ZSTD_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_string.h" +#include "archive_write_private.h" + +/* Don't compile this if we don't have zstd.h */ + +struct private_data { + int compression_level; +#if HAVE_ZSTD_H && HAVE_LIBZSTD + ZSTD_CStream *cstream; + int64_t total_in; + ZSTD_outBuffer out; +#else + struct archive_write_program_data *pdata; +#endif +}; + +/* If we don't have the library use default range values (zstdcli.c v1.4.0) */ +#define CLEVEL_MIN -99 +#define CLEVEL_STD_MIN 0 /* prior to 1.3.4 and more recent without using --fast */ +#define CLEVEL_DEFAULT 3 +#define CLEVEL_STD_MAX 19 /* without using --ultra */ +#define CLEVEL_MAX 22 + +#define MINVER_NEGCLEVEL 10304 +#define MINVER_MINCLEVEL 10306 + +static int archive_compressor_zstd_options(struct archive_write_filter *, + const char *, const char *); +static int archive_compressor_zstd_open(struct archive_write_filter *); +static int archive_compressor_zstd_write(struct archive_write_filter *, + const void *, size_t); +static int archive_compressor_zstd_close(struct archive_write_filter *); +static int archive_compressor_zstd_free(struct archive_write_filter *); +#if HAVE_ZSTD_H && HAVE_LIBZSTD +static int drive_compressor(struct archive_write_filter *, + struct private_data *, int, const void *, size_t); +#endif + + +/* + * Add a zstd compression filter to this write handle. + */ +int +archive_write_add_filter_zstd(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct archive_write_filter *f = __archive_write_allocate_filter(_a); + struct private_data *data; + archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_add_filter_zstd"); + + data = calloc(1, sizeof(*data)); + if (data == NULL) { + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + f->data = data; + f->open = &archive_compressor_zstd_open; + f->options = &archive_compressor_zstd_options; + f->close = &archive_compressor_zstd_close; + f->free = &archive_compressor_zstd_free; + f->code = ARCHIVE_FILTER_ZSTD; + f->name = "zstd"; + data->compression_level = CLEVEL_DEFAULT; +#if HAVE_ZSTD_H && HAVE_LIBZSTD + data->cstream = ZSTD_createCStream(); + if (data->cstream == NULL) { + free(data); + archive_set_error(&a->archive, ENOMEM, + "Failed to allocate zstd compressor object"); + return (ARCHIVE_FATAL); + } + + return (ARCHIVE_OK); +#else + data->pdata = __archive_write_program_allocate("zstd"); + if (data->pdata == NULL) { + free(data); + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Using external zstd program"); + return (ARCHIVE_WARN); +#endif +} + +static int +archive_compressor_zstd_free(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; +#if HAVE_ZSTD_H && HAVE_LIBZSTD + ZSTD_freeCStream(data->cstream); + free(data->out.dst); +#else + __archive_write_program_free(data->pdata); +#endif + free(data); + f->data = NULL; + return (ARCHIVE_OK); +} + +static int string_is_numeric (const char* value) +{ + size_t len = strlen(value); + size_t i; + + if (len == 0) { + return (ARCHIVE_WARN); + } + else if (len == 1 && !(value[0] >= '0' && value[0] <= '9')) { + return (ARCHIVE_WARN); + } + else if (!(value[0] >= '0' && value[0] <= '9') && + value[0] != '-' && value[0] != '+') { + return (ARCHIVE_WARN); + } + + for (i = 1; i < len; i++) { + if (!(value[i] >= '0' && value[i] <= '9')) { + return (ARCHIVE_WARN); + } + } + + return (ARCHIVE_OK); +} + +/* + * Set write options. + */ +static int +archive_compressor_zstd_options(struct archive_write_filter *f, const char *key, + const char *value) +{ + struct private_data *data = (struct private_data *)f->data; + + if (strcmp(key, "compression-level") == 0) { + int level = atoi(value); + /* If we don't have the library, hard-code the max level */ + int minimum = CLEVEL_MIN; + int maximum = CLEVEL_MAX; + if (string_is_numeric(value) != ARCHIVE_OK) { + return (ARCHIVE_WARN); + } +#if HAVE_ZSTD_H && HAVE_LIBZSTD + maximum = ZSTD_maxCLevel(); +#if ZSTD_VERSION_NUMBER >= MINVER_MINCLEVEL + if (ZSTD_versionNumber() >= MINVER_MINCLEVEL) { + minimum = ZSTD_minCLevel(); + } + else +#endif + if (ZSTD_versionNumber() < MINVER_NEGCLEVEL) { + minimum = CLEVEL_STD_MIN; + } +#endif + if (level < minimum || level > maximum) { + return (ARCHIVE_WARN); + } + data->compression_level = level; + return (ARCHIVE_OK); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +#if HAVE_ZSTD_H && HAVE_LIBZSTD +/* + * Setup callback. + */ +static int +archive_compressor_zstd_open(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + + if (data->out.dst == NULL) { + size_t bs = ZSTD_CStreamOutSize(), bpb; + if (f->archive->magic == ARCHIVE_WRITE_MAGIC) { + /* Buffer size should be a multiple number of + * the of bytes per block for performance. */ + bpb = archive_write_get_bytes_per_block(f->archive); + if (bpb > bs) + bs = bpb; + else if (bpb != 0) + bs -= bs % bpb; + } + data->out.size = bs; + data->out.pos = 0; + data->out.dst + = (unsigned char *)malloc(data->out.size); + if (data->out.dst == NULL) { + archive_set_error(f->archive, ENOMEM, + "Can't allocate data for compression buffer"); + return (ARCHIVE_FATAL); + } + } + + f->write = archive_compressor_zstd_write; + + if (ZSTD_isError(ZSTD_initCStream(data->cstream, + data->compression_level))) { + archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, + "Internal error initializing zstd compressor object"); + return (ARCHIVE_FATAL); + } + + return (ARCHIVE_OK); +} + +/* + * Write data to the compressed stream. + */ +static int +archive_compressor_zstd_write(struct archive_write_filter *f, const void *buff, + size_t length) +{ + struct private_data *data = (struct private_data *)f->data; + int ret; + + /* Update statistics */ + data->total_in += length; + + if ((ret = drive_compressor(f, data, 0, buff, length)) != ARCHIVE_OK) + return (ret); + + return (ARCHIVE_OK); +} + +/* + * Finish the compression... + */ +static int +archive_compressor_zstd_close(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + + /* Finish zstd frame */ + return drive_compressor(f, data, 1, NULL, 0); +} + +/* + * Utility function to push input data through compressor, + * writing full output blocks as necessary. + * + * Note that this handles both the regular write case (finishing == + * false) and the end-of-archive case (finishing == true). + */ +static int +drive_compressor(struct archive_write_filter *f, + struct private_data *data, int finishing, const void *src, size_t length) +{ + ZSTD_inBuffer in = (ZSTD_inBuffer) { src, length, 0 }; + + for (;;) { + if (data->out.pos == data->out.size) { + const int ret = __archive_write_filter(f->next_filter, + data->out.dst, data->out.size); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + data->out.pos = 0; + } + + /* If there's nothing to do, we're done. */ + if (!finishing && in.pos == in.size) + return (ARCHIVE_OK); + + { + const size_t zstdret = !finishing ? + ZSTD_compressStream(data->cstream, &data->out, &in) + : ZSTD_endStream(data->cstream, &data->out); + + if (ZSTD_isError(zstdret)) { + archive_set_error(f->archive, + ARCHIVE_ERRNO_MISC, + "Zstd compression failed: %s", + ZSTD_getErrorName(zstdret)); + return (ARCHIVE_FATAL); + } + + /* If we're finishing, 0 means nothing left to flush */ + if (finishing && zstdret == 0) { + const int ret = __archive_write_filter(f->next_filter, + data->out.dst, data->out.pos); + return (ret); + } + } + } +} + +#else /* HAVE_ZSTD_H && HAVE_LIBZSTD */ + +static int +archive_compressor_zstd_open(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + struct archive_string as; + int r; + + archive_string_init(&as); + /* --no-check matches library default */ + archive_strcpy(&as, "zstd --no-check"); + + if (data->compression_level < CLEVEL_STD_MIN) { + struct archive_string as2; + archive_string_init(&as2); + archive_string_sprintf(&as2, " --fast=%d", -data->compression_level); + archive_string_concat(&as, &as2); + archive_string_free(&as2); + } else { + struct archive_string as2; + archive_string_init(&as2); + archive_string_sprintf(&as2, " -%d", data->compression_level); + archive_string_concat(&as, &as2); + archive_string_free(&as2); + } + + if (data->compression_level > CLEVEL_STD_MAX) { + archive_strcat(&as, " --ultra"); + } + + f->write = archive_compressor_zstd_write; + r = __archive_write_program_open(f, data->pdata, as.s); + archive_string_free(&as); + return (r); +} + +static int +archive_compressor_zstd_write(struct archive_write_filter *f, const void *buff, + size_t length) +{ + struct private_data *data = (struct private_data *)f->data; + + return __archive_write_program_write(f, data->pdata, buff, length); +} + +static int +archive_compressor_zstd_close(struct archive_write_filter *f) +{ + struct private_data *data = (struct private_data *)f->data; + + return __archive_write_program_close(f, data->pdata); +} + +#endif /* HAVE_ZSTD_H && HAVE_LIBZSTD */ diff --git a/c/archive_write_disk_posix.c b/c/archive_write_disk_posix.c new file mode 100644 --- /dev/null +++ b/c/archive_write_disk_posix.c @@ -0,0 +1,4540 @@ +/*- + * Copyright (c) 2003-2010 Tim Kientzle + * Copyright (c) 2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer + * in this position and unchanged. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#if !defined(_WIN32) || defined(__CYGWIN__) + +#ifdef HAVE_SYS_TYPES_H +#include +#endif +#ifdef HAVE_SYS_ACL_H +#include +#endif +#ifdef HAVE_SYS_EXTATTR_H +#include +#endif +#if HAVE_SYS_XATTR_H +#include +#elif HAVE_ATTR_XATTR_H +#include +#endif +#ifdef HAVE_SYS_EA_H +#include +#endif +#ifdef HAVE_SYS_IOCTL_H +#include +#endif +#ifdef HAVE_SYS_STAT_H +#include +#endif +#ifdef HAVE_SYS_TIME_H +#include +#endif +#ifdef HAVE_SYS_UTIME_H +#include +#endif +#ifdef HAVE_COPYFILE_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_FCNTL_H +#include +#endif +#ifdef HAVE_GRP_H +#include +#endif +#ifdef HAVE_LANGINFO_H +#include +#endif +#ifdef HAVE_LINUX_FS_H +#include /* for Linux file flags */ +#endif +/* + * Some Linux distributions have both linux/ext2_fs.h and ext2fs/ext2_fs.h. + * As the include guards don't agree, the order of include is important. + */ +#ifdef HAVE_LINUX_EXT2_FS_H +#include /* for Linux file flags */ +#endif +#if defined(HAVE_EXT2FS_EXT2_FS_H) && !defined(__CYGWIN__) +#include /* Linux file flags, broken on Cygwin */ +#endif +#ifdef HAVE_LIMITS_H +#include +#endif +#ifdef HAVE_PWD_H +#include +#endif +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif +#ifdef HAVE_UTIME_H +#include +#endif +#ifdef F_GETTIMES /* Tru64 specific */ +#include +#endif + +/* + * Macro to cast st_mtime and time_t to an int64 so that 2 numbers can reliably be compared. + * + * It assumes that the input is an integer type of no more than 64 bits. + * If the number is less than zero, t must be a signed type, so it fits in + * int64_t. Otherwise, it's a nonnegative value so we can cast it to uint64_t + * without loss. But it could be a large unsigned value, so we have to clip it + * to INT64_MAX.* + */ +#define to_int64_time(t) \ + ((t) < 0 ? (int64_t)(t) : (uint64_t)(t) > (uint64_t)INT64_MAX ? INT64_MAX : (int64_t)(t)) + +#if __APPLE__ +#include +#if TARGET_OS_MAC && !TARGET_OS_EMBEDDED && HAVE_QUARANTINE_H +#include +#define HAVE_QUARANTINE 1 +#endif +#endif + +#ifdef HAVE_ZLIB_H +#include +#endif + +/* TODO: Support Mac OS 'quarantine' feature. This is really just a + * standard tag to mark files that have been downloaded as "tainted". + * On Mac OS, we should mark the extracted files as tainted if the + * archive being read was tainted. Windows has a similar feature; we + * should investigate ways to support this generically. */ + +#include "archive.h" +#include "archive_acl_private.h" +#include "archive_string.h" +#include "archive_endian.h" +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_write_disk_private.h" + +#ifndef O_BINARY +#define O_BINARY 0 +#endif +#ifndef O_CLOEXEC +#define O_CLOEXEC 0 +#endif + +/* Ignore non-int O_NOFOLLOW constant. */ +/* gnulib's fcntl.h does this on AIX, but it seems practical everywhere */ +#if defined O_NOFOLLOW && !(INT_MIN <= O_NOFOLLOW && O_NOFOLLOW <= INT_MAX) +#undef O_NOFOLLOW +#endif + +#ifndef O_NOFOLLOW +#define O_NOFOLLOW 0 +#endif + +#ifndef AT_FDCWD +#define AT_FDCWD -100 +#endif + +struct fixup_entry { + struct fixup_entry *next; + struct archive_acl acl; + mode_t mode; + int64_t atime; + int64_t birthtime; + int64_t mtime; + int64_t ctime; + unsigned long atime_nanos; + unsigned long birthtime_nanos; + unsigned long mtime_nanos; + unsigned long ctime_nanos; + unsigned long fflags_set; + size_t mac_metadata_size; + void *mac_metadata; + int fixup; /* bitmask of what needs fixing */ + char *name; +}; + +/* + * We use a bitmask to track which operations remain to be done for + * this file. In particular, this helps us avoid unnecessary + * operations when it's possible to take care of one step as a + * side-effect of another. For example, mkdir() can specify the mode + * for the newly-created object but symlink() cannot. This means we + * can skip chmod() if mkdir() succeeded, but we must explicitly + * chmod() if we're trying to create a directory that already exists + * (mkdir() failed) or if we're restoring a symlink. Similarly, we + * need to verify UID/GID before trying to restore SUID/SGID bits; + * that verification can occur explicitly through a stat() call or + * implicitly because of a successful chown() call. + */ +#define TODO_MODE_FORCE 0x40000000 +#define TODO_MODE_BASE 0x20000000 +#define TODO_SUID 0x10000000 +#define TODO_SUID_CHECK 0x08000000 +#define TODO_SGID 0x04000000 +#define TODO_SGID_CHECK 0x02000000 +#define TODO_APPLEDOUBLE 0x01000000 +#define TODO_MODE (TODO_MODE_BASE|TODO_SUID|TODO_SGID) +#define TODO_TIMES ARCHIVE_EXTRACT_TIME +#define TODO_OWNER ARCHIVE_EXTRACT_OWNER +#define TODO_FFLAGS ARCHIVE_EXTRACT_FFLAGS +#define TODO_ACLS ARCHIVE_EXTRACT_ACL +#define TODO_XATTR ARCHIVE_EXTRACT_XATTR +#define TODO_MAC_METADATA ARCHIVE_EXTRACT_MAC_METADATA +#define TODO_HFS_COMPRESSION ARCHIVE_EXTRACT_HFS_COMPRESSION_FORCED + +struct archive_write_disk { + struct archive archive; + + mode_t user_umask; + struct fixup_entry *fixup_list; + struct fixup_entry *current_fixup; + int64_t user_uid; + int skip_file_set; + int64_t skip_file_dev; + int64_t skip_file_ino; + time_t start_time; + + int64_t (*lookup_gid)(void *private, const char *gname, int64_t gid); + void (*cleanup_gid)(void *private); + void *lookup_gid_data; + int64_t (*lookup_uid)(void *private, const char *uname, int64_t uid); + void (*cleanup_uid)(void *private); + void *lookup_uid_data; + + /* + * Full path of last file to satisfy symlink checks. + */ + struct archive_string path_safe; + + /* + * Cached stat data from disk for the current entry. + * If this is valid, pst points to st. Otherwise, + * pst is null. + */ + struct stat st; + struct stat *pst; + + /* Information about the object being restored right now. */ + struct archive_entry *entry; /* Entry being extracted. */ + char *name; /* Name of entry, possibly edited. */ + struct archive_string _name_data; /* backing store for 'name' */ + char *tmpname; /* Temporary name * */ + struct archive_string _tmpname_data; /* backing store for 'tmpname' */ + /* Tasks remaining for this object. */ + int todo; + /* Tasks deferred until end-of-archive. */ + int deferred; + /* Options requested by the client. */ + int flags; + /* Handle for the file we're restoring. */ + int fd; + /* Current offset for writing data to the file. */ + int64_t offset; + /* Last offset actually written to disk. */ + int64_t fd_offset; + /* Total bytes actually written to files. */ + int64_t total_bytes_written; + /* Maximum size of file, -1 if unknown. */ + int64_t filesize; + /* Dir we were in before this restore; only for deep paths. */ + int restore_pwd; + /* Mode we should use for this entry; affected by _PERM and umask. */ + mode_t mode; + /* UID/GID to use in restoring this entry. */ + int64_t uid; + int64_t gid; + /* + * HFS+ Compression. + */ + /* Xattr "com.apple.decmpfs". */ + uint32_t decmpfs_attr_size; + unsigned char *decmpfs_header_p; + /* ResourceFork set options used for fsetxattr. */ + int rsrc_xattr_options; + /* Xattr "com.apple.ResourceFork". */ + unsigned char *resource_fork; + size_t resource_fork_allocated_size; + unsigned int decmpfs_block_count; + uint32_t *decmpfs_block_info; + /* Buffer for compressed data. */ + unsigned char *compressed_buffer; + size_t compressed_buffer_size; + size_t compressed_buffer_remaining; + /* The offset of the ResourceFork where compressed data will + * be placed. */ + uint32_t compressed_rsrc_position; + uint32_t compressed_rsrc_position_v; + /* Buffer for uncompressed data. */ + char *uncompressed_buffer; + size_t block_remaining_bytes; + size_t file_remaining_bytes; +#ifdef HAVE_ZLIB_H + z_stream stream; + int stream_valid; + int decmpfs_compression_level; +#endif +}; + +/* + * Default mode for dirs created automatically (will be modified by umask). + * Note that POSIX specifies 0777 for implicitly-created dirs, "modified + * by the process' file creation mask." + */ +#define DEFAULT_DIR_MODE 0777 +/* + * Dir modes are restored in two steps: During the extraction, the permissions + * in the archive are modified to match the following limits. During + * the post-extract fixup pass, the permissions from the archive are + * applied. + */ +#define MINIMUM_DIR_MODE 0700 +#define MAXIMUM_DIR_MODE 0775 + +/* + * Maximum uncompressed size of a decmpfs block. + */ +#define MAX_DECMPFS_BLOCK_SIZE (64 * 1024) +/* + * HFS+ compression type. + */ +#define CMP_XATTR 3/* Compressed data in xattr. */ +#define CMP_RESOURCE_FORK 4/* Compressed data in resource fork. */ +/* + * HFS+ compression resource fork. + */ +#define RSRC_H_SIZE 260 /* Base size of Resource fork header. */ +#define RSRC_F_SIZE 50 /* Size of Resource fork footer. */ +/* Size to write compressed data to resource fork. */ +#define COMPRESSED_W_SIZE (64 * 1024) +/* decmpfs definitions. */ +#define MAX_DECMPFS_XATTR_SIZE 3802 +#ifndef DECMPFS_XATTR_NAME +#define DECMPFS_XATTR_NAME "com.apple.decmpfs" +#endif +#define DECMPFS_MAGIC 0x636d7066 +#define DECMPFS_COMPRESSION_MAGIC 0 +#define DECMPFS_COMPRESSION_TYPE 4 +#define DECMPFS_UNCOMPRESSED_SIZE 8 +#define DECMPFS_HEADER_SIZE 16 + +#define HFS_BLOCKS(s) ((s) >> 12) + + +static int la_opendirat(int, const char *); +static int la_mktemp(struct archive_write_disk *); +static void fsobj_error(int *, struct archive_string *, int, const char *, + const char *); +static int check_symlinks_fsobj(char *, int *, struct archive_string *, + int); +static int check_symlinks(struct archive_write_disk *); +static int create_filesystem_object(struct archive_write_disk *); +static struct fixup_entry *current_fixup(struct archive_write_disk *, + const char *pathname); +#if defined(HAVE_FCHDIR) && defined(PATH_MAX) +static void edit_deep_directories(struct archive_write_disk *ad); +#endif +static int cleanup_pathname_fsobj(char *, int *, struct archive_string *, + int); +static int cleanup_pathname(struct archive_write_disk *); +static int create_dir(struct archive_write_disk *, char *); +static int create_parent_dir(struct archive_write_disk *, char *); +static ssize_t hfs_write_data_block(struct archive_write_disk *, + const char *, size_t); +static int fixup_appledouble(struct archive_write_disk *, const char *); +static int older(struct stat *, struct archive_entry *); +static int restore_entry(struct archive_write_disk *); +static int set_mac_metadata(struct archive_write_disk *, const char *, + const void *, size_t); +static int set_xattrs(struct archive_write_disk *); +static int clear_nochange_fflags(struct archive_write_disk *); +static int set_fflags(struct archive_write_disk *); +static int set_fflags_platform(struct archive_write_disk *, int fd, + const char *name, mode_t mode, + unsigned long fflags_set, unsigned long fflags_clear); +static int set_ownership(struct archive_write_disk *); +static int set_mode(struct archive_write_disk *, int mode); +static int set_time(int, int, const char *, time_t, long, time_t, long); +static int set_times(struct archive_write_disk *, int, int, const char *, + time_t, long, time_t, long, time_t, long, time_t, long); +static int set_times_from_entry(struct archive_write_disk *); +static struct fixup_entry *sort_dir_list(struct fixup_entry *p); +static ssize_t write_data_block(struct archive_write_disk *, + const char *, size_t); + +static struct archive_vtable *archive_write_disk_vtable(void); + +static int _archive_write_disk_close(struct archive *); +static int _archive_write_disk_free(struct archive *); +static int _archive_write_disk_header(struct archive *, + struct archive_entry *); +static int64_t _archive_write_disk_filter_bytes(struct archive *, int); +static int _archive_write_disk_finish_entry(struct archive *); +static ssize_t _archive_write_disk_data(struct archive *, const void *, + size_t); +static ssize_t _archive_write_disk_data_block(struct archive *, const void *, + size_t, int64_t); + +static int +la_mktemp(struct archive_write_disk *a) +{ + int oerrno, fd; + mode_t mode; + + archive_string_empty(&a->_tmpname_data); + archive_string_sprintf(&a->_tmpname_data, "%s.XXXXXX", a->name); + a->tmpname = a->_tmpname_data.s; + + fd = __archive_mkstemp(a->tmpname); + if (fd == -1) + return -1; + + mode = a->mode & 0777 & ~a->user_umask; + if (fchmod(fd, mode) == -1) { + oerrno = errno; + close(fd); + errno = oerrno; + return -1; + } + return fd; +} + +static int +la_opendirat(int fd, const char *path) { + const int flags = O_CLOEXEC +#if defined(O_BINARY) + | O_BINARY +#endif +#if defined(O_DIRECTORY) + | O_DIRECTORY +#endif +#if defined(O_PATH) + | O_PATH +#elif defined(O_SEARCH) + | O_SEARCH +#elif defined(__FreeBSD__) && defined(O_EXEC) + | O_EXEC +#else + | O_RDONLY +#endif + ; + +#if !defined(HAVE_OPENAT) + if (fd != AT_FDCWD) { + errno = ENOTSUP; + return (-1); + } else + return (open(path, flags)); +#else + return (openat(fd, path, flags)); +#endif +} + +static int +lazy_stat(struct archive_write_disk *a) +{ + if (a->pst != NULL) { + /* Already have stat() data available. */ + return (ARCHIVE_OK); + } +#ifdef HAVE_FSTAT + if (a->fd >= 0 && fstat(a->fd, &a->st) == 0) { + a->pst = &a->st; + return (ARCHIVE_OK); + } +#endif + /* + * XXX At this point, symlinks should not be hit, otherwise + * XXX a race occurred. Do we want to check explicitly for that? + */ + if (lstat(a->name, &a->st) == 0) { + a->pst = &a->st; + return (ARCHIVE_OK); + } + archive_set_error(&a->archive, errno, "Couldn't stat file"); + return (ARCHIVE_WARN); +} + +static struct archive_vtable * +archive_write_disk_vtable(void) +{ + static struct archive_vtable av; + static int inited = 0; + + if (!inited) { + av.archive_close = _archive_write_disk_close; + av.archive_filter_bytes = _archive_write_disk_filter_bytes; + av.archive_free = _archive_write_disk_free; + av.archive_write_header = _archive_write_disk_header; + av.archive_write_finish_entry + = _archive_write_disk_finish_entry; + av.archive_write_data = _archive_write_disk_data; + av.archive_write_data_block = _archive_write_disk_data_block; + inited = 1; + } + return (&av); +} + +static int64_t +_archive_write_disk_filter_bytes(struct archive *_a, int n) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + (void)n; /* UNUSED */ + if (n == -1 || n == 0) + return (a->total_bytes_written); + return (-1); +} + + +int +archive_write_disk_set_options(struct archive *_a, int flags) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + + a->flags = flags; + return (ARCHIVE_OK); +} + + +/* + * Extract this entry to disk. + * + * TODO: Validate hardlinks. According to the standards, we're + * supposed to check each extracted hardlink and squawk if it refers + * to a file that we didn't restore. I'm not entirely convinced this + * is a good idea, but more importantly: Is there any way to validate + * hardlinks without keeping a complete list of filenames from the + * entire archive?? Ugh. + * + */ +static int +_archive_write_disk_header(struct archive *_a, struct archive_entry *entry) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + struct fixup_entry *fe; + int ret, r; + + archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, + "archive_write_disk_header"); + archive_clear_error(&a->archive); + if (a->archive.state & ARCHIVE_STATE_DATA) { + r = _archive_write_disk_finish_entry(&a->archive); + if (r == ARCHIVE_FATAL) + return (r); + } + + /* Set up for this particular entry. */ + a->pst = NULL; + a->current_fixup = NULL; + a->deferred = 0; + if (a->entry) { + archive_entry_free(a->entry); + a->entry = NULL; + } + a->entry = archive_entry_clone(entry); + a->fd = -1; + a->fd_offset = 0; + a->offset = 0; + a->restore_pwd = -1; + a->uid = a->user_uid; + a->mode = archive_entry_mode(a->entry); + if (archive_entry_size_is_set(a->entry)) + a->filesize = archive_entry_size(a->entry); + else + a->filesize = -1; + archive_strcpy(&(a->_name_data), archive_entry_pathname(a->entry)); + a->name = a->_name_data.s; + archive_clear_error(&a->archive); + + /* + * Clean up the requested path. This is necessary for correct + * dir restores; the dir restore logic otherwise gets messed + * up by nonsense like "dir/.". + */ + ret = cleanup_pathname(a); + if (ret != ARCHIVE_OK) + return (ret); + + /* + * Query the umask so we get predictable mode settings. + * This gets done on every call to _write_header in case the + * user edits their umask during the extraction for some + * reason. + */ + umask(a->user_umask = umask(0)); + + /* Figure out what we need to do for this entry. */ + a->todo = TODO_MODE_BASE; + if (a->flags & ARCHIVE_EXTRACT_PERM) { + a->todo |= TODO_MODE_FORCE; /* Be pushy about permissions. */ + /* + * SGID requires an extra "check" step because we + * cannot easily predict the GID that the system will + * assign. (Different systems assign GIDs to files + * based on a variety of criteria, including process + * credentials and the gid of the enclosing + * directory.) We can only restore the SGID bit if + * the file has the right GID, and we only know the + * GID if we either set it (see set_ownership) or if + * we've actually called stat() on the file after it + * was restored. Since there are several places at + * which we might verify the GID, we need a TODO bit + * to keep track. + */ + if (a->mode & S_ISGID) + a->todo |= TODO_SGID | TODO_SGID_CHECK; + /* + * Verifying the SUID is simpler, but can still be + * done in multiple ways, hence the separate "check" bit. + */ + if (a->mode & S_ISUID) + a->todo |= TODO_SUID | TODO_SUID_CHECK; + } else { + /* + * User didn't request full permissions, so don't + * restore SUID, SGID bits and obey umask. + */ + a->mode &= ~S_ISUID; + a->mode &= ~S_ISGID; + a->mode &= ~S_ISVTX; + a->mode &= ~a->user_umask; + } + if (a->flags & ARCHIVE_EXTRACT_OWNER) + a->todo |= TODO_OWNER; + if (a->flags & ARCHIVE_EXTRACT_TIME) + a->todo |= TODO_TIMES; + if (a->flags & ARCHIVE_EXTRACT_ACL) { +#if ARCHIVE_ACL_DARWIN + /* + * On MacOS, platform ACLs get stored in mac_metadata, too. + * If we intend to extract mac_metadata and it is present + * we skip extracting libarchive NFSv4 ACLs. + */ + size_t metadata_size; + + if ((a->flags & ARCHIVE_EXTRACT_MAC_METADATA) == 0 || + archive_entry_mac_metadata(a->entry, + &metadata_size) == NULL || metadata_size == 0) +#endif +#if ARCHIVE_ACL_LIBRICHACL + /* + * RichACLs are stored in an extended attribute. + * If we intend to extract extended attributes and have this + * attribute we skip extracting libarchive NFSv4 ACLs. + */ + short extract_acls = 1; + if (a->flags & ARCHIVE_EXTRACT_XATTR && ( + archive_entry_acl_types(a->entry) & + ARCHIVE_ENTRY_ACL_TYPE_NFS4)) { + const char *attr_name; + const void *attr_value; + size_t attr_size; + int i = archive_entry_xattr_reset(a->entry); + while (i--) { + archive_entry_xattr_next(a->entry, &attr_name, + &attr_value, &attr_size); + if (attr_name != NULL && attr_value != NULL && + attr_size > 0 && strcmp(attr_name, + "trusted.richacl") == 0) { + extract_acls = 0; + break; + } + } + } + if (extract_acls) +#endif +#if ARCHIVE_ACL_DARWIN || ARCHIVE_ACL_LIBRICHACL + { +#endif + if (archive_entry_filetype(a->entry) == AE_IFDIR) + a->deferred |= TODO_ACLS; + else + a->todo |= TODO_ACLS; +#if ARCHIVE_ACL_DARWIN || ARCHIVE_ACL_LIBRICHACL + } +#endif + } + if (a->flags & ARCHIVE_EXTRACT_MAC_METADATA) { + if (archive_entry_filetype(a->entry) == AE_IFDIR) + a->deferred |= TODO_MAC_METADATA; + else + a->todo |= TODO_MAC_METADATA; + } +#if defined(__APPLE__) && defined(UF_COMPRESSED) && defined(HAVE_ZLIB_H) + if ((a->flags & ARCHIVE_EXTRACT_NO_HFS_COMPRESSION) == 0) { + unsigned long set, clear; + archive_entry_fflags(a->entry, &set, &clear); + if ((set & ~clear) & UF_COMPRESSED) { + a->todo |= TODO_HFS_COMPRESSION; + a->decmpfs_block_count = (unsigned)-1; + } + } + if ((a->flags & ARCHIVE_EXTRACT_HFS_COMPRESSION_FORCED) != 0 && + (a->mode & AE_IFMT) == AE_IFREG && a->filesize > 0) { + a->todo |= TODO_HFS_COMPRESSION; + a->decmpfs_block_count = (unsigned)-1; + } + { + const char *p; + + /* Check if the current file name is a type of the + * resource fork file. */ + p = strrchr(a->name, '/'); + if (p == NULL) + p = a->name; + else + p++; + if (p[0] == '.' && p[1] == '_') { + /* Do not compress "._XXX" files. */ + a->todo &= ~TODO_HFS_COMPRESSION; + if (a->filesize > 0) + a->todo |= TODO_APPLEDOUBLE; + } + } +#endif + + if (a->flags & ARCHIVE_EXTRACT_XATTR) { +#if ARCHIVE_XATTR_DARWIN + /* + * On MacOS, extended attributes get stored in mac_metadata, + * too. If we intend to extract mac_metadata and it is present + * we skip extracting extended attributes. + */ + size_t metadata_size; + + if ((a->flags & ARCHIVE_EXTRACT_MAC_METADATA) == 0 || + archive_entry_mac_metadata(a->entry, + &metadata_size) == NULL || metadata_size == 0) +#endif + a->todo |= TODO_XATTR; + } + if (a->flags & ARCHIVE_EXTRACT_FFLAGS) + a->todo |= TODO_FFLAGS; + if (a->flags & ARCHIVE_EXTRACT_SECURE_SYMLINKS) { + ret = check_symlinks(a); + if (ret != ARCHIVE_OK) + return (ret); + } +#if defined(HAVE_FCHDIR) && defined(PATH_MAX) + /* If path exceeds PATH_MAX, shorten the path. */ + edit_deep_directories(a); +#endif + + ret = restore_entry(a); + +#if defined(__APPLE__) && defined(UF_COMPRESSED) && defined(HAVE_ZLIB_H) + /* + * Check if the filesystem the file is restoring on supports + * HFS+ Compression. If not, cancel HFS+ Compression. + */ + if (a->todo | TODO_HFS_COMPRESSION) { + /* + * NOTE: UF_COMPRESSED is ignored even if the filesystem + * supports HFS+ Compression because the file should + * have at least an extended attribute "com.apple.decmpfs" + * before the flag is set to indicate that the file have + * been compressed. If the filesystem does not support + * HFS+ Compression the system call will fail. + */ + if (a->fd < 0 || fchflags(a->fd, UF_COMPRESSED) != 0) + a->todo &= ~TODO_HFS_COMPRESSION; + } +#endif + + /* + * TODO: There are rumours that some extended attributes must + * be restored before file data is written. If this is true, + * then we either need to write all extended attributes both + * before and after restoring the data, or find some rule for + * determining which must go first and which last. Due to the + * many ways people are using xattrs, this may prove to be an + * intractable problem. + */ + +#ifdef HAVE_FCHDIR + /* If we changed directory above, restore it here. */ + if (a->restore_pwd >= 0) { + r = fchdir(a->restore_pwd); + if (r != 0) { + archive_set_error(&a->archive, errno, + "chdir() failure"); + ret = ARCHIVE_FATAL; + } + close(a->restore_pwd); + a->restore_pwd = -1; + } +#endif + + /* + * Fixup uses the unedited pathname from archive_entry_pathname(), + * because it is relative to the base dir and the edited path + * might be relative to some intermediate dir as a result of the + * deep restore logic. + */ + if (a->deferred & TODO_MODE) { + fe = current_fixup(a, archive_entry_pathname(entry)); + if (fe == NULL) + return (ARCHIVE_FATAL); + fe->fixup |= TODO_MODE_BASE; + fe->mode = a->mode; + } + + if ((a->deferred & TODO_TIMES) + && (archive_entry_mtime_is_set(entry) + || archive_entry_atime_is_set(entry))) { + fe = current_fixup(a, archive_entry_pathname(entry)); + if (fe == NULL) + return (ARCHIVE_FATAL); + fe->mode = a->mode; + fe->fixup |= TODO_TIMES; + if (archive_entry_atime_is_set(entry)) { + fe->atime = archive_entry_atime(entry); + fe->atime_nanos = archive_entry_atime_nsec(entry); + } else { + /* If atime is unset, use start time. */ + fe->atime = a->start_time; + fe->atime_nanos = 0; + } + if (archive_entry_mtime_is_set(entry)) { + fe->mtime = archive_entry_mtime(entry); + fe->mtime_nanos = archive_entry_mtime_nsec(entry); + } else { + /* If mtime is unset, use start time. */ + fe->mtime = a->start_time; + fe->mtime_nanos = 0; + } + if (archive_entry_birthtime_is_set(entry)) { + fe->birthtime = archive_entry_birthtime(entry); + fe->birthtime_nanos = archive_entry_birthtime_nsec( + entry); + } else { + /* If birthtime is unset, use mtime. */ + fe->birthtime = fe->mtime; + fe->birthtime_nanos = fe->mtime_nanos; + } + } + + if (a->deferred & TODO_ACLS) { + fe = current_fixup(a, archive_entry_pathname(entry)); + if (fe == NULL) + return (ARCHIVE_FATAL); + fe->fixup |= TODO_ACLS; + archive_acl_copy(&fe->acl, archive_entry_acl(entry)); + } + + if (a->deferred & TODO_MAC_METADATA) { + const void *metadata; + size_t metadata_size; + metadata = archive_entry_mac_metadata(a->entry, &metadata_size); + if (metadata != NULL && metadata_size > 0) { + fe = current_fixup(a, archive_entry_pathname(entry)); + if (fe == NULL) + return (ARCHIVE_FATAL); + fe->mac_metadata = malloc(metadata_size); + if (fe->mac_metadata != NULL) { + memcpy(fe->mac_metadata, metadata, + metadata_size); + fe->mac_metadata_size = metadata_size; + fe->fixup |= TODO_MAC_METADATA; + } + } + } + + if (a->deferred & TODO_FFLAGS) { + fe = current_fixup(a, archive_entry_pathname(entry)); + if (fe == NULL) + return (ARCHIVE_FATAL); + fe->fixup |= TODO_FFLAGS; + /* TODO: Complete this.. defer fflags from below. */ + } + + /* We've created the object and are ready to pour data into it. */ + if (ret >= ARCHIVE_WARN) + a->archive.state = ARCHIVE_STATE_DATA; + /* + * If it's not open, tell our client not to try writing. + * In particular, dirs, links, etc, don't get written to. + */ + if (a->fd < 0) { + archive_entry_set_size(entry, 0); + a->filesize = 0; + } + + return (ret); +} + +int +archive_write_disk_set_skip_file(struct archive *_a, la_int64_t d, la_int64_t i) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_write_disk_set_skip_file"); + a->skip_file_set = 1; + a->skip_file_dev = d; + a->skip_file_ino = i; + return (ARCHIVE_OK); +} + +static ssize_t +write_data_block(struct archive_write_disk *a, const char *buff, size_t size) +{ + uint64_t start_size = size; + ssize_t bytes_written = 0; + ssize_t block_size = 0, bytes_to_write; + + if (size == 0) + return (ARCHIVE_OK); + + if (a->filesize == 0 || a->fd < 0) { + archive_set_error(&a->archive, 0, + "Attempt to write to an empty file"); + return (ARCHIVE_WARN); + } + + if (a->flags & ARCHIVE_EXTRACT_SPARSE) { +#if HAVE_STRUCT_STAT_ST_BLKSIZE + int r; + if ((r = lazy_stat(a)) != ARCHIVE_OK) + return (r); + block_size = a->pst->st_blksize; +#else + /* XXX TODO XXX Is there a more appropriate choice here ? */ + /* This needn't match the filesystem allocation size. */ + block_size = 16*1024; +#endif + } + + /* If this write would run beyond the file size, truncate it. */ + if (a->filesize >= 0 && (int64_t)(a->offset + size) > a->filesize) + start_size = size = (size_t)(a->filesize - a->offset); + + /* Write the data. */ + while (size > 0) { + if (block_size == 0) { + bytes_to_write = size; + } else { + /* We're sparsifying the file. */ + const char *p, *end; + int64_t block_end; + + /* Skip leading zero bytes. */ + for (p = buff, end = buff + size; p < end; ++p) { + if (*p != '\0') + break; + } + a->offset += p - buff; + size -= p - buff; + buff = p; + if (size == 0) + break; + + /* Calculate next block boundary after offset. */ + block_end + = (a->offset / block_size + 1) * block_size; + + /* If the adjusted write would cross block boundary, + * truncate it to the block boundary. */ + bytes_to_write = size; + if (a->offset + bytes_to_write > block_end) + bytes_to_write = block_end - a->offset; + } + /* Seek if necessary to the specified offset. */ + if (a->offset != a->fd_offset) { + if (lseek(a->fd, a->offset, SEEK_SET) < 0) { + archive_set_error(&a->archive, errno, + "Seek failed"); + return (ARCHIVE_FATAL); + } + a->fd_offset = a->offset; + } + bytes_written = write(a->fd, buff, bytes_to_write); + if (bytes_written < 0) { + archive_set_error(&a->archive, errno, "Write failed"); + return (ARCHIVE_WARN); + } + buff += bytes_written; + size -= bytes_written; + a->total_bytes_written += bytes_written; + a->offset += bytes_written; + a->fd_offset = a->offset; + } + return (start_size - size); +} + +#if defined(__APPLE__) && defined(UF_COMPRESSED) && defined(HAVE_SYS_XATTR_H)\ + && defined(HAVE_ZLIB_H) + +/* + * Set UF_COMPRESSED file flag. + * This have to be called after hfs_write_decmpfs() because if the + * file does not have "com.apple.decmpfs" xattr the flag is ignored. + */ +static int +hfs_set_compressed_fflag(struct archive_write_disk *a) +{ + int r; + + if ((r = lazy_stat(a)) != ARCHIVE_OK) + return (r); + + a->st.st_flags |= UF_COMPRESSED; + if (fchflags(a->fd, a->st.st_flags) != 0) { + archive_set_error(&a->archive, errno, + "Failed to set UF_COMPRESSED file flag"); + return (ARCHIVE_WARN); + } + return (ARCHIVE_OK); +} + +/* + * HFS+ Compression decmpfs + * + * +------------------------------+ +0 + * | Magic(LE 4 bytes) | + * +------------------------------+ + * | Type(LE 4 bytes) | + * +------------------------------+ + * | Uncompressed size(LE 8 bytes)| + * +------------------------------+ +16 + * | | + * | Compressed data | + * | (Placed only if Type == 3) | + * | | + * +------------------------------+ +3802 = MAX_DECMPFS_XATTR_SIZE + * + * Type is 3: decmpfs has compressed data. + * Type is 4: Resource Fork has compressed data. + */ +/* + * Write "com.apple.decmpfs" + */ +static int +hfs_write_decmpfs(struct archive_write_disk *a) +{ + int r; + uint32_t compression_type; + + r = fsetxattr(a->fd, DECMPFS_XATTR_NAME, a->decmpfs_header_p, + a->decmpfs_attr_size, 0, 0); + if (r < 0) { + archive_set_error(&a->archive, errno, + "Cannot restore xattr:%s", DECMPFS_XATTR_NAME); + compression_type = archive_le32dec( + &a->decmpfs_header_p[DECMPFS_COMPRESSION_TYPE]); + if (compression_type == CMP_RESOURCE_FORK) + fremovexattr(a->fd, XATTR_RESOURCEFORK_NAME, + XATTR_SHOWCOMPRESSION); + return (ARCHIVE_WARN); + } + return (ARCHIVE_OK); +} + +/* + * HFS+ Compression Resource Fork + * + * +-----------------------------+ + * | Header(260 bytes) | + * +-----------------------------+ + * | Block count(LE 4 bytes) | + * +-----------------------------+ --+ + * +-- | Offset (LE 4 bytes) | | + * | | [distance from Block count] | | Block 0 + * | +-----------------------------+ | + * | | Compressed size(LE 4 bytes) | | + * | +-----------------------------+ --+ + * | | | + * | | .................. | + * | | | + * | +-----------------------------+ --+ + * | | Offset (LE 4 bytes) | | + * | +-----------------------------+ | Block (Block count -1) + * | | Compressed size(LE 4 bytes) | | + * +-> +-----------------------------+ --+ + * | Compressed data(n bytes) | Block 0 + * +-----------------------------+ + * | | + * | .................. | + * | | + * +-----------------------------+ + * | Compressed data(n bytes) | Block (Block count -1) + * +-----------------------------+ + * | Footer(50 bytes) | + * +-----------------------------+ + * + */ +/* + * Write the header of "com.apple.ResourceFork" + */ +static int +hfs_write_resource_fork(struct archive_write_disk *a, unsigned char *buff, + size_t bytes, uint32_t position) +{ + int ret; + + ret = fsetxattr(a->fd, XATTR_RESOURCEFORK_NAME, buff, bytes, + position, a->rsrc_xattr_options); + if (ret < 0) { + archive_set_error(&a->archive, errno, + "Cannot restore xattr: %s at %u pos %u bytes", + XATTR_RESOURCEFORK_NAME, + (unsigned)position, + (unsigned)bytes); + return (ARCHIVE_WARN); + } + a->rsrc_xattr_options &= ~XATTR_CREATE; + return (ARCHIVE_OK); +} + +static int +hfs_write_compressed_data(struct archive_write_disk *a, size_t bytes_compressed) +{ + int ret; + + ret = hfs_write_resource_fork(a, a->compressed_buffer, + bytes_compressed, a->compressed_rsrc_position); + if (ret == ARCHIVE_OK) + a->compressed_rsrc_position += bytes_compressed; + return (ret); +} + +static int +hfs_write_resource_fork_header(struct archive_write_disk *a) +{ + unsigned char *buff; + uint32_t rsrc_bytes; + uint32_t rsrc_header_bytes; + + /* + * Write resource fork header + block info. + */ + buff = a->resource_fork; + rsrc_bytes = a->compressed_rsrc_position - RSRC_F_SIZE; + rsrc_header_bytes = + RSRC_H_SIZE + /* Header base size. */ + 4 + /* Block count. */ + (a->decmpfs_block_count * 8);/* Block info */ + archive_be32enc(buff, 0x100); + archive_be32enc(buff + 4, rsrc_bytes); + archive_be32enc(buff + 8, rsrc_bytes - 256); + archive_be32enc(buff + 12, 0x32); + memset(buff + 16, 0, 240); + archive_be32enc(buff + 256, rsrc_bytes - 260); + return hfs_write_resource_fork(a, buff, rsrc_header_bytes, 0); +} + +static size_t +hfs_set_resource_fork_footer(unsigned char *buff, size_t buff_size) +{ + static const char rsrc_footer[RSRC_F_SIZE] = { + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x1c, 0x00, 0x32, 0x00, 0x00, 'c', 'm', + 'p', 'f', 0x00, 0x00, 0x00, 0x0a, 0x00, 0x01, + 0xff, 0xff, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, + 0x00, 0x00 + }; + if (buff_size < sizeof(rsrc_footer)) + return (0); + memcpy(buff, rsrc_footer, sizeof(rsrc_footer)); + return (sizeof(rsrc_footer)); +} + +static int +hfs_reset_compressor(struct archive_write_disk *a) +{ + int ret; + + if (a->stream_valid) + ret = deflateReset(&a->stream); + else + ret = deflateInit(&a->stream, a->decmpfs_compression_level); + + if (ret != Z_OK) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Failed to initialize compressor"); + return (ARCHIVE_FATAL); + } else + a->stream_valid = 1; + + return (ARCHIVE_OK); +} + +static int +hfs_decompress(struct archive_write_disk *a) +{ + uint32_t *block_info; + unsigned int block_count; + uint32_t data_pos, data_size; + ssize_t r; + ssize_t bytes_written, bytes_to_write; + unsigned char *b; + + block_info = (uint32_t *)(a->resource_fork + RSRC_H_SIZE); + block_count = archive_le32dec(block_info++); + while (block_count--) { + data_pos = RSRC_H_SIZE + archive_le32dec(block_info++); + data_size = archive_le32dec(block_info++); + r = fgetxattr(a->fd, XATTR_RESOURCEFORK_NAME, + a->compressed_buffer, data_size, data_pos, 0); + if (r != data_size) { + archive_set_error(&a->archive, + (r < 0)?errno:ARCHIVE_ERRNO_MISC, + "Failed to read resource fork"); + return (ARCHIVE_WARN); + } + if (a->compressed_buffer[0] == 0xff) { + bytes_to_write = data_size -1; + b = a->compressed_buffer + 1; + } else { + uLong dest_len = MAX_DECMPFS_BLOCK_SIZE; + int zr; + + zr = uncompress((Bytef *)a->uncompressed_buffer, + &dest_len, a->compressed_buffer, data_size); + if (zr != Z_OK) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Failed to decompress resource fork"); + return (ARCHIVE_WARN); + } + bytes_to_write = dest_len; + b = (unsigned char *)a->uncompressed_buffer; + } + do { + bytes_written = write(a->fd, b, bytes_to_write); + if (bytes_written < 0) { + archive_set_error(&a->archive, errno, + "Write failed"); + return (ARCHIVE_WARN); + } + bytes_to_write -= bytes_written; + b += bytes_written; + } while (bytes_to_write > 0); + } + r = fremovexattr(a->fd, XATTR_RESOURCEFORK_NAME, 0); + if (r == -1) { + archive_set_error(&a->archive, errno, + "Failed to remove resource fork"); + return (ARCHIVE_WARN); + } + return (ARCHIVE_OK); +} + +static int +hfs_drive_compressor(struct archive_write_disk *a, const char *buff, + size_t size) +{ + unsigned char *buffer_compressed; + size_t bytes_compressed; + size_t bytes_used; + int ret; + + ret = hfs_reset_compressor(a); + if (ret != ARCHIVE_OK) + return (ret); + + if (a->compressed_buffer == NULL) { + size_t block_size; + + block_size = COMPRESSED_W_SIZE + RSRC_F_SIZE + + + compressBound(MAX_DECMPFS_BLOCK_SIZE); + a->compressed_buffer = malloc(block_size); + if (a->compressed_buffer == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Resource Fork"); + return (ARCHIVE_FATAL); + } + a->compressed_buffer_size = block_size; + a->compressed_buffer_remaining = block_size; + } + + buffer_compressed = a->compressed_buffer + + a->compressed_buffer_size - a->compressed_buffer_remaining; + a->stream.next_in = (Bytef *)(uintptr_t)(const void *)buff; + a->stream.avail_in = size; + a->stream.next_out = buffer_compressed; + a->stream.avail_out = a->compressed_buffer_remaining; + do { + ret = deflate(&a->stream, Z_FINISH); + switch (ret) { + case Z_OK: + case Z_STREAM_END: + break; + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Failed to compress data"); + return (ARCHIVE_FAILED); + } + } while (ret == Z_OK); + bytes_compressed = a->compressed_buffer_remaining - a->stream.avail_out; + + /* + * If the compressed size is larger than the original size, + * throw away compressed data, use uncompressed data instead. + */ + if (bytes_compressed > size) { + buffer_compressed[0] = 0xFF;/* uncompressed marker. */ + memcpy(buffer_compressed + 1, buff, size); + bytes_compressed = size + 1; + } + a->compressed_buffer_remaining -= bytes_compressed; + + /* + * If the compressed size is smaller than MAX_DECMPFS_XATTR_SIZE + * and the block count in the file is only one, store compressed + * data to decmpfs xattr instead of the resource fork. + */ + if (a->decmpfs_block_count == 1 && + (a->decmpfs_attr_size + bytes_compressed) + <= MAX_DECMPFS_XATTR_SIZE) { + archive_le32enc(&a->decmpfs_header_p[DECMPFS_COMPRESSION_TYPE], + CMP_XATTR); + memcpy(a->decmpfs_header_p + DECMPFS_HEADER_SIZE, + buffer_compressed, bytes_compressed); + a->decmpfs_attr_size += bytes_compressed; + a->compressed_buffer_remaining = a->compressed_buffer_size; + /* + * Finish HFS+ Compression. + * - Write the decmpfs xattr. + * - Set the UF_COMPRESSED file flag. + */ + ret = hfs_write_decmpfs(a); + if (ret == ARCHIVE_OK) + ret = hfs_set_compressed_fflag(a); + return (ret); + } + + /* Update block info. */ + archive_le32enc(a->decmpfs_block_info++, + a->compressed_rsrc_position_v - RSRC_H_SIZE); + archive_le32enc(a->decmpfs_block_info++, bytes_compressed); + a->compressed_rsrc_position_v += bytes_compressed; + + /* + * Write the compressed data to the resource fork. + */ + bytes_used = a->compressed_buffer_size - a->compressed_buffer_remaining; + while (bytes_used >= COMPRESSED_W_SIZE) { + ret = hfs_write_compressed_data(a, COMPRESSED_W_SIZE); + if (ret != ARCHIVE_OK) + return (ret); + bytes_used -= COMPRESSED_W_SIZE; + if (bytes_used > COMPRESSED_W_SIZE) + memmove(a->compressed_buffer, + a->compressed_buffer + COMPRESSED_W_SIZE, + bytes_used); + else + memcpy(a->compressed_buffer, + a->compressed_buffer + COMPRESSED_W_SIZE, + bytes_used); + } + a->compressed_buffer_remaining = a->compressed_buffer_size - bytes_used; + + /* + * If the current block is the last block, write the remaining + * compressed data and the resource fork footer. + */ + if (a->file_remaining_bytes == 0) { + size_t rsrc_size; + int64_t bk; + + /* Append the resource footer. */ + rsrc_size = hfs_set_resource_fork_footer( + a->compressed_buffer + bytes_used, + a->compressed_buffer_remaining); + ret = hfs_write_compressed_data(a, bytes_used + rsrc_size); + a->compressed_buffer_remaining = a->compressed_buffer_size; + + /* If the compressed size is not enough smaller than + * the uncompressed size. cancel HFS+ compression. + * TODO: study a behavior of ditto utility and improve + * the condition to fall back into no HFS+ compression. */ + bk = HFS_BLOCKS(a->compressed_rsrc_position); + bk += bk >> 7; + if (bk > HFS_BLOCKS(a->filesize)) + return hfs_decompress(a); + /* + * Write the resourcefork header. + */ + if (ret == ARCHIVE_OK) + ret = hfs_write_resource_fork_header(a); + /* + * Finish HFS+ Compression. + * - Write the decmpfs xattr. + * - Set the UF_COMPRESSED file flag. + */ + if (ret == ARCHIVE_OK) + ret = hfs_write_decmpfs(a); + if (ret == ARCHIVE_OK) + ret = hfs_set_compressed_fflag(a); + } + return (ret); +} + +static ssize_t +hfs_write_decmpfs_block(struct archive_write_disk *a, const char *buff, + size_t size) +{ + const char *buffer_to_write; + size_t bytes_to_write; + int ret; + + if (a->decmpfs_block_count == (unsigned)-1) { + void *new_block; + size_t new_size; + unsigned int block_count; + + if (a->decmpfs_header_p == NULL) { + new_block = malloc(MAX_DECMPFS_XATTR_SIZE + + sizeof(uint32_t)); + if (new_block == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for decmpfs"); + return (ARCHIVE_FATAL); + } + a->decmpfs_header_p = new_block; + } + a->decmpfs_attr_size = DECMPFS_HEADER_SIZE; + archive_le32enc(&a->decmpfs_header_p[DECMPFS_COMPRESSION_MAGIC], + DECMPFS_MAGIC); + archive_le32enc(&a->decmpfs_header_p[DECMPFS_COMPRESSION_TYPE], + CMP_RESOURCE_FORK); + archive_le64enc(&a->decmpfs_header_p[DECMPFS_UNCOMPRESSED_SIZE], + a->filesize); + + /* Calculate a block count of the file. */ + block_count = + (a->filesize + MAX_DECMPFS_BLOCK_SIZE -1) / + MAX_DECMPFS_BLOCK_SIZE; + /* + * Allocate buffer for resource fork. + * Set up related pointers; + */ + new_size = + RSRC_H_SIZE + /* header */ + 4 + /* Block count */ + (block_count * sizeof(uint32_t) * 2) + + RSRC_F_SIZE; /* footer */ + if (new_size > a->resource_fork_allocated_size) { + new_block = realloc(a->resource_fork, new_size); + if (new_block == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for ResourceFork"); + return (ARCHIVE_FATAL); + } + a->resource_fork_allocated_size = new_size; + a->resource_fork = new_block; + } + + /* Allocate uncompressed buffer */ + if (a->uncompressed_buffer == NULL) { + new_block = malloc(MAX_DECMPFS_BLOCK_SIZE); + if (new_block == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for decmpfs"); + return (ARCHIVE_FATAL); + } + a->uncompressed_buffer = new_block; + } + a->block_remaining_bytes = MAX_DECMPFS_BLOCK_SIZE; + a->file_remaining_bytes = a->filesize; + a->compressed_buffer_remaining = a->compressed_buffer_size; + + /* + * Set up a resource fork. + */ + a->rsrc_xattr_options = XATTR_CREATE; + /* Get the position where we are going to set a bunch + * of block info. */ + a->decmpfs_block_info = + (uint32_t *)(a->resource_fork + RSRC_H_SIZE); + /* Set the block count to the resource fork. */ + archive_le32enc(a->decmpfs_block_info++, block_count); + /* Get the position where we are going to set compressed + * data. */ + a->compressed_rsrc_position = + RSRC_H_SIZE + 4 + (block_count * 8); + a->compressed_rsrc_position_v = a->compressed_rsrc_position; + a->decmpfs_block_count = block_count; + } + + /* Ignore redundant bytes. */ + if (a->file_remaining_bytes == 0) + return ((ssize_t)size); + + /* Do not overrun a block size. */ + if (size > a->block_remaining_bytes) + bytes_to_write = a->block_remaining_bytes; + else + bytes_to_write = size; + /* Do not overrun the file size. */ + if (bytes_to_write > a->file_remaining_bytes) + bytes_to_write = a->file_remaining_bytes; + + /* For efficiency, if a copy length is full of the uncompressed + * buffer size, do not copy writing data to it. */ + if (bytes_to_write == MAX_DECMPFS_BLOCK_SIZE) + buffer_to_write = buff; + else { + memcpy(a->uncompressed_buffer + + MAX_DECMPFS_BLOCK_SIZE - a->block_remaining_bytes, + buff, bytes_to_write); + buffer_to_write = a->uncompressed_buffer; + } + a->block_remaining_bytes -= bytes_to_write; + a->file_remaining_bytes -= bytes_to_write; + + if (a->block_remaining_bytes == 0 || a->file_remaining_bytes == 0) { + ret = hfs_drive_compressor(a, buffer_to_write, + MAX_DECMPFS_BLOCK_SIZE - a->block_remaining_bytes); + if (ret < 0) + return (ret); + a->block_remaining_bytes = MAX_DECMPFS_BLOCK_SIZE; + } + /* Ignore redundant bytes. */ + if (a->file_remaining_bytes == 0) + return ((ssize_t)size); + return (bytes_to_write); +} + +static ssize_t +hfs_write_data_block(struct archive_write_disk *a, const char *buff, + size_t size) +{ + uint64_t start_size = size; + ssize_t bytes_written = 0; + ssize_t bytes_to_write; + + if (size == 0) + return (ARCHIVE_OK); + + if (a->filesize == 0 || a->fd < 0) { + archive_set_error(&a->archive, 0, + "Attempt to write to an empty file"); + return (ARCHIVE_WARN); + } + + /* If this write would run beyond the file size, truncate it. */ + if (a->filesize >= 0 && (int64_t)(a->offset + size) > a->filesize) + start_size = size = (size_t)(a->filesize - a->offset); + + /* Write the data. */ + while (size > 0) { + bytes_to_write = size; + /* Seek if necessary to the specified offset. */ + if (a->offset < a->fd_offset) { + /* Can't support backward move. */ + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Seek failed"); + return (ARCHIVE_FATAL); + } else if (a->offset > a->fd_offset) { + int64_t skip = a->offset - a->fd_offset; + char nullblock[1024]; + + memset(nullblock, 0, sizeof(nullblock)); + while (skip > 0) { + if (skip > (int64_t)sizeof(nullblock)) + bytes_written = hfs_write_decmpfs_block( + a, nullblock, sizeof(nullblock)); + else + bytes_written = hfs_write_decmpfs_block( + a, nullblock, skip); + if (bytes_written < 0) { + archive_set_error(&a->archive, errno, + "Write failed"); + return (ARCHIVE_WARN); + } + skip -= bytes_written; + } + + a->fd_offset = a->offset; + } + bytes_written = + hfs_write_decmpfs_block(a, buff, bytes_to_write); + if (bytes_written < 0) + return (bytes_written); + buff += bytes_written; + size -= bytes_written; + a->total_bytes_written += bytes_written; + a->offset += bytes_written; + a->fd_offset = a->offset; + } + return (start_size - size); +} +#else +static ssize_t +hfs_write_data_block(struct archive_write_disk *a, const char *buff, + size_t size) +{ + return (write_data_block(a, buff, size)); +} +#endif + +static ssize_t +_archive_write_disk_data_block(struct archive *_a, + const void *buff, size_t size, int64_t offset) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + ssize_t r; + + archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_DATA, "archive_write_data_block"); + + a->offset = offset; + if (a->todo & TODO_HFS_COMPRESSION) + r = hfs_write_data_block(a, buff, size); + else + r = write_data_block(a, buff, size); + if (r < ARCHIVE_OK) + return (r); + if ((size_t)r < size) { + archive_set_error(&a->archive, 0, + "Too much data: Truncating file at %ju bytes", + (uintmax_t)a->filesize); + return (ARCHIVE_WARN); + } +#if ARCHIVE_VERSION_NUMBER < 3999000 + return (ARCHIVE_OK); +#else + return (size); +#endif +} + +static ssize_t +_archive_write_disk_data(struct archive *_a, const void *buff, size_t size) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + + archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_DATA, "archive_write_data"); + + if (a->todo & TODO_HFS_COMPRESSION) + return (hfs_write_data_block(a, buff, size)); + return (write_data_block(a, buff, size)); +} + +static int +_archive_write_disk_finish_entry(struct archive *_a) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + int ret = ARCHIVE_OK; + + archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, + "archive_write_finish_entry"); + if (a->archive.state & ARCHIVE_STATE_HEADER) + return (ARCHIVE_OK); + archive_clear_error(&a->archive); + + /* Pad or truncate file to the right size. */ + if (a->fd < 0) { + /* There's no file. */ + } else if (a->filesize < 0) { + /* File size is unknown, so we can't set the size. */ + } else if (a->fd_offset == a->filesize) { + /* Last write ended at exactly the filesize; we're done. */ + /* Hopefully, this is the common case. */ +#if defined(__APPLE__) && defined(UF_COMPRESSED) && defined(HAVE_ZLIB_H) + } else if (a->todo & TODO_HFS_COMPRESSION) { + char null_d[1024]; + ssize_t r; + + if (a->file_remaining_bytes) + memset(null_d, 0, sizeof(null_d)); + while (a->file_remaining_bytes) { + if (a->file_remaining_bytes > sizeof(null_d)) + r = hfs_write_data_block( + a, null_d, sizeof(null_d)); + else + r = hfs_write_data_block( + a, null_d, a->file_remaining_bytes); + if (r < 0) + return ((int)r); + } +#endif + } else { +#if HAVE_FTRUNCATE + if (ftruncate(a->fd, a->filesize) == -1 && + a->filesize == 0) { + archive_set_error(&a->archive, errno, + "File size could not be restored"); + return (ARCHIVE_FAILED); + } +#endif + /* + * Not all platforms implement the XSI option to + * extend files via ftruncate. Stat() the file again + * to see what happened. + */ + a->pst = NULL; + if ((ret = lazy_stat(a)) != ARCHIVE_OK) + return (ret); + /* We can use lseek()/write() to extend the file if + * ftruncate didn't work or isn't available. */ + if (a->st.st_size < a->filesize) { + const char nul = '\0'; + if (lseek(a->fd, a->filesize - 1, SEEK_SET) < 0) { + archive_set_error(&a->archive, errno, + "Seek failed"); + return (ARCHIVE_FATAL); + } + if (write(a->fd, &nul, 1) < 0) { + archive_set_error(&a->archive, errno, + "Write to restore size failed"); + return (ARCHIVE_FATAL); + } + a->pst = NULL; + } + } + + /* Restore metadata. */ + + /* + * This is specific to Mac OS X. + * If the current file is an AppleDouble file, it should be + * linked with the data fork file and remove it. + */ + if (a->todo & TODO_APPLEDOUBLE) { + int r2 = fixup_appledouble(a, a->name); + if (r2 == ARCHIVE_EOF) { + /* The current file has been successfully linked + * with the data fork file and removed. So there + * is nothing to do on the current file. */ + goto finish_metadata; + } + if (r2 < ret) ret = r2; + } + + /* + * Look up the "real" UID only if we're going to need it. + * TODO: the TODO_SGID condition can be dropped here, can't it? + */ + if (a->todo & (TODO_OWNER | TODO_SUID | TODO_SGID)) { + a->uid = archive_write_disk_uid(&a->archive, + archive_entry_uname(a->entry), + archive_entry_uid(a->entry)); + } + /* Look up the "real" GID only if we're going to need it. */ + /* TODO: the TODO_SUID condition can be dropped here, can't it? */ + if (a->todo & (TODO_OWNER | TODO_SGID | TODO_SUID)) { + a->gid = archive_write_disk_gid(&a->archive, + archive_entry_gname(a->entry), + archive_entry_gid(a->entry)); + } + + /* + * Restore ownership before set_mode tries to restore suid/sgid + * bits. If we set the owner, we know what it is and can skip + * a stat() call to examine the ownership of the file on disk. + */ + if (a->todo & TODO_OWNER) { + int r2 = set_ownership(a); + if (r2 < ret) ret = r2; + } + + /* + * HYPOTHESIS: + * If we're not root, we won't be setting any security + * attributes that may be wiped by the set_mode() routine + * below. We also can't set xattr on non-owner-writable files, + * which may be the state after set_mode(). Perform + * set_xattrs() first based on these constraints. + */ + if (a->user_uid != 0 && + (a->todo & TODO_XATTR)) { + int r2 = set_xattrs(a); + if (r2 < ret) ret = r2; + } + + /* + * set_mode must precede ACLs on systems such as Solaris and + * FreeBSD where setting the mode implicitly clears extended ACLs + */ + if (a->todo & TODO_MODE) { + int r2 = set_mode(a, a->mode); + if (r2 < ret) ret = r2; + } + + /* + * Security-related extended attributes (such as + * security.capability on Linux) have to be restored last, + * since they're implicitly removed by other file changes. + * We do this last only when root. + */ + if (a->user_uid == 0 && + (a->todo & TODO_XATTR)) { + int r2 = set_xattrs(a); + if (r2 < ret) ret = r2; + } + + /* + * Some flags prevent file modification; they must be restored after + * file contents are written. + */ + if (a->todo & TODO_FFLAGS) { + int r2 = set_fflags(a); + if (r2 < ret) ret = r2; + } + + /* + * Time must follow most other metadata; + * otherwise atime will get changed. + */ + if (a->todo & TODO_TIMES) { + int r2 = set_times_from_entry(a); + if (r2 < ret) ret = r2; + } + + /* + * Mac extended metadata includes ACLs. + */ + if (a->todo & TODO_MAC_METADATA) { + const void *metadata; + size_t metadata_size; + metadata = archive_entry_mac_metadata(a->entry, &metadata_size); + if (metadata != NULL && metadata_size > 0) { + int r2 = set_mac_metadata(a, archive_entry_pathname( + a->entry), metadata, metadata_size); + if (r2 < ret) ret = r2; + } + } + + /* + * ACLs must be restored after timestamps because there are + * ACLs that prevent attribute changes (including time). + */ + if (a->todo & TODO_ACLS) { + int r2; + r2 = archive_write_disk_set_acls(&a->archive, a->fd, + archive_entry_pathname(a->entry), + archive_entry_acl(a->entry), + archive_entry_mode(a->entry)); + if (r2 < ret) ret = r2; + } + +finish_metadata: + /* If there's an fd, we can close it now. */ + if (a->fd >= 0) { + close(a->fd); + a->fd = -1; + if (a->tmpname) { + if (rename(a->tmpname, a->name) == -1) { + archive_set_error(&a->archive, errno, + "Failed to rename temporary file"); + ret = ARCHIVE_FAILED; + unlink(a->tmpname); + } + a->tmpname = NULL; + } + } + /* If there's an entry, we can release it now. */ + archive_entry_free(a->entry); + a->entry = NULL; + a->archive.state = ARCHIVE_STATE_HEADER; + return (ret); +} + +int +archive_write_disk_set_group_lookup(struct archive *_a, + void *private_data, + la_int64_t (*lookup_gid)(void *private, const char *gname, la_int64_t gid), + void (*cleanup_gid)(void *private)) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_write_disk_set_group_lookup"); + + if (a->cleanup_gid != NULL && a->lookup_gid_data != NULL) + (a->cleanup_gid)(a->lookup_gid_data); + + a->lookup_gid = lookup_gid; + a->cleanup_gid = cleanup_gid; + a->lookup_gid_data = private_data; + return (ARCHIVE_OK); +} + +int +archive_write_disk_set_user_lookup(struct archive *_a, + void *private_data, + int64_t (*lookup_uid)(void *private, const char *uname, int64_t uid), + void (*cleanup_uid)(void *private)) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_write_disk_set_user_lookup"); + + if (a->cleanup_uid != NULL && a->lookup_uid_data != NULL) + (a->cleanup_uid)(a->lookup_uid_data); + + a->lookup_uid = lookup_uid; + a->cleanup_uid = cleanup_uid; + a->lookup_uid_data = private_data; + return (ARCHIVE_OK); +} + +int64_t +archive_write_disk_gid(struct archive *_a, const char *name, la_int64_t id) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_write_disk_gid"); + if (a->lookup_gid) + return (a->lookup_gid)(a->lookup_gid_data, name, id); + return (id); +} + +int64_t +archive_write_disk_uid(struct archive *_a, const char *name, la_int64_t id) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_write_disk_uid"); + if (a->lookup_uid) + return (a->lookup_uid)(a->lookup_uid_data, name, id); + return (id); +} + +/* + * Create a new archive_write_disk object and initialize it with global state. + */ +struct archive * +archive_write_disk_new(void) +{ + struct archive_write_disk *a; + + a = (struct archive_write_disk *)calloc(1, sizeof(*a)); + if (a == NULL) + return (NULL); + a->archive.magic = ARCHIVE_WRITE_DISK_MAGIC; + /* We're ready to write a header immediately. */ + a->archive.state = ARCHIVE_STATE_HEADER; + a->archive.vtable = archive_write_disk_vtable(); + a->start_time = time(NULL); + /* Query and restore the umask. */ + umask(a->user_umask = umask(0)); +#ifdef HAVE_GETEUID + a->user_uid = geteuid(); +#endif /* HAVE_GETEUID */ + if (archive_string_ensure(&a->path_safe, 512) == NULL) { + free(a); + return (NULL); + } +#ifdef HAVE_ZLIB_H + a->decmpfs_compression_level = 5; +#endif + return (&a->archive); +} + + +/* + * If pathname is longer than PATH_MAX, chdir to a suitable + * intermediate dir and edit the path down to a shorter suffix. Note + * that this routine never returns an error; if the chdir() attempt + * fails for any reason, we just go ahead with the long pathname. The + * object creation is likely to fail, but any error will get handled + * at that time. + */ +#if defined(HAVE_FCHDIR) && defined(PATH_MAX) +static void +edit_deep_directories(struct archive_write_disk *a) +{ + int ret; + char *tail = a->name; + + /* If path is short, avoid the open() below. */ + if (strlen(tail) < PATH_MAX) + return; + + /* Try to record our starting dir. */ + a->restore_pwd = la_opendirat(AT_FDCWD, "."); + __archive_ensure_cloexec_flag(a->restore_pwd); + if (a->restore_pwd < 0) + return; + + /* As long as the path is too long... */ + while (strlen(tail) >= PATH_MAX) { + /* Locate a dir prefix shorter than PATH_MAX. */ + tail += PATH_MAX - 8; + while (tail > a->name && *tail != '/') + tail--; + /* Exit if we find a too-long path component. */ + if (tail <= a->name) + return; + /* Create the intermediate dir and chdir to it. */ + *tail = '\0'; /* Terminate dir portion */ + ret = create_dir(a, a->name); + if (ret == ARCHIVE_OK && chdir(a->name) != 0) + ret = ARCHIVE_FAILED; + *tail = '/'; /* Restore the / we removed. */ + if (ret != ARCHIVE_OK) + return; + tail++; + /* The chdir() succeeded; we've now shortened the path. */ + a->name = tail; + } + return; +} +#endif + +/* + * The main restore function. + */ +static int +restore_entry(struct archive_write_disk *a) +{ + int ret = ARCHIVE_OK, en; + + if (a->flags & ARCHIVE_EXTRACT_UNLINK && !S_ISDIR(a->mode)) { + /* + * TODO: Fix this. Apparently, there are platforms + * that still allow root to hose the entire filesystem + * by unlinking a dir. The S_ISDIR() test above + * prevents us from using unlink() here if the new + * object is a dir, but that doesn't mean the old + * object isn't a dir. + */ + if (a->flags & ARCHIVE_EXTRACT_CLEAR_NOCHANGE_FFLAGS) + (void)clear_nochange_fflags(a); + if (unlink(a->name) == 0) { + /* We removed it, reset cached stat. */ + a->pst = NULL; + } else if (errno == ENOENT) { + /* File didn't exist, that's just as good. */ + } else if (rmdir(a->name) == 0) { + /* It was a dir, but now it's gone. */ + a->pst = NULL; + } else { + /* We tried, but couldn't get rid of it. */ + archive_set_error(&a->archive, errno, + "Could not unlink"); + return(ARCHIVE_FAILED); + } + } + + /* Try creating it first; if this fails, we'll try to recover. */ + en = create_filesystem_object(a); + + if ((en == ENOTDIR || en == ENOENT) + && !(a->flags & ARCHIVE_EXTRACT_NO_AUTODIR)) { + /* If the parent dir doesn't exist, try creating it. */ + create_parent_dir(a, a->name); + /* Now try to create the object again. */ + en = create_filesystem_object(a); + } + + if ((en == ENOENT) && (archive_entry_hardlink(a->entry) != NULL)) { + archive_set_error(&a->archive, en, + "Hard-link target '%s' does not exist.", + archive_entry_hardlink(a->entry)); + return (ARCHIVE_FAILED); + } + + if ((en == EISDIR || en == EEXIST) + && (a->flags & ARCHIVE_EXTRACT_NO_OVERWRITE)) { + /* If we're not overwriting, we're done. */ + if (S_ISDIR(a->mode)) { + /* Don't overwrite any settings on existing directories. */ + a->todo = 0; + } + archive_entry_unset_size(a->entry); + return (ARCHIVE_OK); + } + + /* + * Some platforms return EISDIR if you call + * open(O_WRONLY | O_EXCL | O_CREAT) on a directory, some + * return EEXIST. POSIX is ambiguous, requiring EISDIR + * for open(O_WRONLY) on a dir and EEXIST for open(O_EXCL | O_CREAT) + * on an existing item. + */ + if (en == EISDIR) { + /* A dir is in the way of a non-dir, rmdir it. */ + if (rmdir(a->name) != 0) { + archive_set_error(&a->archive, errno, + "Can't remove already-existing dir"); + return (ARCHIVE_FAILED); + } + a->pst = NULL; + /* Try again. */ + en = create_filesystem_object(a); + } else if (en == EEXIST) { + /* + * We know something is in the way, but we don't know what; + * we need to find out before we go any further. + */ + int r = 0; + /* + * The SECURE_SYMLINKS logic has already removed a + * symlink to a dir if the client wants that. So + * follow the symlink if we're creating a dir. + */ + if (S_ISDIR(a->mode)) + r = la_stat(a->name, &a->st); + /* + * If it's not a dir (or it's a broken symlink), + * then don't follow it. + */ + if (r != 0 || !S_ISDIR(a->mode)) + r = lstat(a->name, &a->st); + if (r != 0) { + archive_set_error(&a->archive, errno, + "Can't stat existing object"); + return (ARCHIVE_FAILED); + } + + /* + * NO_OVERWRITE_NEWER doesn't apply to directories. + */ + if ((a->flags & ARCHIVE_EXTRACT_NO_OVERWRITE_NEWER) + && !S_ISDIR(a->st.st_mode)) { + if (!older(&(a->st), a->entry)) { + archive_entry_unset_size(a->entry); + return (ARCHIVE_OK); + } + } + + /* If it's our archive, we're done. */ + if (a->skip_file_set && + a->st.st_dev == (dev_t)a->skip_file_dev && + a->st.st_ino == (ino_t)a->skip_file_ino) { + archive_set_error(&a->archive, 0, + "Refusing to overwrite archive"); + return (ARCHIVE_FAILED); + } + + if (!S_ISDIR(a->st.st_mode)) { + if (a->flags & ARCHIVE_EXTRACT_CLEAR_NOCHANGE_FFLAGS) + (void)clear_nochange_fflags(a); + + if ((a->flags & ARCHIVE_EXTRACT_SAFE_WRITES) && + S_ISREG(a->st.st_mode)) { + /* Use a temporary file to extract */ + if ((a->fd = la_mktemp(a)) == -1) { + archive_set_error(&a->archive, errno, + "Can't create temporary file"); + return ARCHIVE_FAILED; + } + a->pst = NULL; + en = 0; + } else { + /* A non-dir is in the way, unlink it. */ + if (unlink(a->name) != 0) { + archive_set_error(&a->archive, errno, + "Can't unlink already-existing " + "object"); + return (ARCHIVE_FAILED); + } + a->pst = NULL; + /* Try again. */ + en = create_filesystem_object(a); + } + } else if (!S_ISDIR(a->mode)) { + /* A dir is in the way of a non-dir, rmdir it. */ + if (a->flags & ARCHIVE_EXTRACT_CLEAR_NOCHANGE_FFLAGS) + (void)clear_nochange_fflags(a); + if (rmdir(a->name) != 0) { + archive_set_error(&a->archive, errno, + "Can't replace existing directory with non-directory"); + return (ARCHIVE_FAILED); + } + /* Try again. */ + en = create_filesystem_object(a); + } else { + /* + * There's a dir in the way of a dir. Don't + * waste time with rmdir()/mkdir(), just fix + * up the permissions on the existing dir. + * Note that we don't change perms on existing + * dirs unless _EXTRACT_PERM is specified. + */ + if ((a->mode != a->st.st_mode) + && (a->todo & TODO_MODE_FORCE)) + a->deferred |= (a->todo & TODO_MODE); + /* Ownership doesn't need deferred fixup. */ + en = 0; /* Forget the EEXIST. */ + } + } + + if (en) { + /* Everything failed; give up here. */ + if ((&a->archive)->error == NULL) + archive_set_error(&a->archive, en, "Can't create '%s'", + a->name); + return (ARCHIVE_FAILED); + } + + a->pst = NULL; /* Cached stat data no longer valid. */ + return (ret); +} + +/* + * Returns 0 if creation succeeds, or else returns errno value from + * the failed system call. Note: This function should only ever perform + * a single system call. + */ +static int +create_filesystem_object(struct archive_write_disk *a) +{ + /* Create the entry. */ + const char *linkname; + mode_t final_mode, mode; + int r; + /* these for check_symlinks_fsobj */ + char *linkname_copy; /* non-const copy of linkname */ + struct stat st; + struct archive_string error_string; + int error_number; + + /* We identify hard/symlinks according to the link names. */ + /* Since link(2) and symlink(2) don't handle modes, we're done here. */ + linkname = archive_entry_hardlink(a->entry); + if (linkname != NULL) { +#if !HAVE_LINK + return (EPERM); +#else + archive_string_init(&error_string); + linkname_copy = strdup(linkname); + if (linkname_copy == NULL) { + return (EPERM); + } + /* + * TODO: consider using the cleaned-up path as the link + * target? + */ + r = cleanup_pathname_fsobj(linkname_copy, &error_number, + &error_string, a->flags); + if (r != ARCHIVE_OK) { + archive_set_error(&a->archive, error_number, "%s", + error_string.s); + free(linkname_copy); + archive_string_free(&error_string); + /* + * EPERM is more appropriate than error_number for our + * callers + */ + return (EPERM); + } + r = check_symlinks_fsobj(linkname_copy, &error_number, + &error_string, a->flags); + if (r != ARCHIVE_OK) { + archive_set_error(&a->archive, error_number, "%s", + error_string.s); + free(linkname_copy); + archive_string_free(&error_string); + /* + * EPERM is more appropriate than error_number for our + * callers + */ + return (EPERM); + } + free(linkname_copy); + archive_string_free(&error_string); + /* + * Unlinking and linking here is really not atomic, + * but doing it right, would require us to construct + * an mktemplink() function, and then use rename(2). + */ + if (a->flags & ARCHIVE_EXTRACT_SAFE_WRITES) + unlink(a->name); + r = link(linkname, a->name) ? errno : 0; + /* + * New cpio and pax formats allow hardlink entries + * to carry data, so we may have to open the file + * for hardlink entries. + * + * If the hardlink was successfully created and + * the archive doesn't have carry data for it, + * consider it to be non-authoritative for meta data. + * This is consistent with GNU tar and BSD pax. + * If the hardlink does carry data, let the last + * archive entry decide ownership. + */ + if (r == 0 && a->filesize <= 0) { + a->todo = 0; + a->deferred = 0; + } else if (r == 0 && a->filesize > 0) { +#ifdef HAVE_LSTAT + r = lstat(a->name, &st); +#else + r = la_stat(a->name, &st); +#endif + if (r != 0) + r = errno; + else if ((st.st_mode & AE_IFMT) == AE_IFREG) { + a->fd = open(a->name, O_WRONLY | O_TRUNC | + O_BINARY | O_CLOEXEC | O_NOFOLLOW); + __archive_ensure_cloexec_flag(a->fd); + if (a->fd < 0) + r = errno; + } + } + return (r); +#endif + } + linkname = archive_entry_symlink(a->entry); + if (linkname != NULL) { +#if HAVE_SYMLINK + /* + * Unlinking and linking here is really not atomic, + * but doing it right, would require us to construct + * an mktempsymlink() function, and then use rename(2). + */ + if (a->flags & ARCHIVE_EXTRACT_SAFE_WRITES) + unlink(a->name); + return symlink(linkname, a->name) ? errno : 0; +#else + return (EPERM); +#endif + } + + /* + * The remaining system calls all set permissions, so let's + * try to take advantage of that to avoid an extra chmod() + * call. (Recall that umask is set to zero right now!) + */ + + /* Mode we want for the final restored object (w/o file type bits). */ + final_mode = a->mode & 07777; + /* + * The mode that will actually be restored in this step. Note + * that SUID, SGID, etc, require additional work to ensure + * security, so we never restore them at this point. + */ + mode = final_mode & 0777 & ~a->user_umask; + + /* + * Always create writable such that [f]setxattr() works if we're not + * root. + */ + if (a->user_uid != 0 && + a->todo & (TODO_HFS_COMPRESSION | TODO_XATTR)) { + mode |= 0200; + } + + switch (a->mode & AE_IFMT) { + default: + /* POSIX requires that we fall through here. */ + /* FALLTHROUGH */ + case AE_IFREG: + a->tmpname = NULL; + a->fd = open(a->name, + O_WRONLY | O_CREAT | O_EXCL | O_BINARY | O_CLOEXEC, mode); + __archive_ensure_cloexec_flag(a->fd); + r = (a->fd < 0); + break; + case AE_IFCHR: +#ifdef HAVE_MKNOD + /* Note: we use AE_IFCHR for the case label, and + * S_IFCHR for the mknod() call. This is correct. */ + r = mknod(a->name, mode | S_IFCHR, + archive_entry_rdev(a->entry)); + break; +#else + /* TODO: Find a better way to warn about our inability + * to restore a char device node. */ + return (EINVAL); +#endif /* HAVE_MKNOD */ + case AE_IFBLK: +#ifdef HAVE_MKNOD + r = mknod(a->name, mode | S_IFBLK, + archive_entry_rdev(a->entry)); + break; +#else + /* TODO: Find a better way to warn about our inability + * to restore a block device node. */ + return (EINVAL); +#endif /* HAVE_MKNOD */ + case AE_IFDIR: + mode = (mode | MINIMUM_DIR_MODE) & MAXIMUM_DIR_MODE; + r = mkdir(a->name, mode); + if (r == 0) { + /* Defer setting dir times. */ + a->deferred |= (a->todo & TODO_TIMES); + a->todo &= ~TODO_TIMES; + /* Never use an immediate chmod(). */ + /* We can't avoid the chmod() entirely if EXTRACT_PERM + * because of SysV SGID inheritance. */ + if ((mode != final_mode) + || (a->flags & ARCHIVE_EXTRACT_PERM)) + a->deferred |= (a->todo & TODO_MODE); + a->todo &= ~TODO_MODE; + } + break; + case AE_IFIFO: +#ifdef HAVE_MKFIFO + r = mkfifo(a->name, mode); + break; +#else + /* TODO: Find a better way to warn about our inability + * to restore a fifo. */ + return (EINVAL); +#endif /* HAVE_MKFIFO */ + } + + /* All the system calls above set errno on failure. */ + if (r) + return (errno); + + /* If we managed to set the final mode, we've avoided a chmod(). */ + if (mode == final_mode) + a->todo &= ~TODO_MODE; + return (0); +} + +/* + * Cleanup function for archive_extract. Mostly, this involves processing + * the fixup list, which is used to address a number of problems: + * * Dir permissions might prevent us from restoring a file in that + * dir, so we restore the dir with minimum 0700 permissions first, + * then correct the mode at the end. + * * Similarly, the act of restoring a file touches the directory + * and changes the timestamp on the dir, so we have to touch-up dir + * timestamps at the end as well. + * * Some file flags can interfere with the restore by, for example, + * preventing the creation of hardlinks to those files. + * * Mac OS extended metadata includes ACLs, so must be deferred on dirs. + * + * Note that tar/cpio do not require that archives be in a particular + * order; there is no way to know when the last file has been restored + * within a directory, so there's no way to optimize the memory usage + * here by fixing up the directory any earlier than the + * end-of-archive. + * + * XXX TODO: Directory ACLs should be restored here, for the same + * reason we set directory perms here. XXX + */ +static int +_archive_write_disk_close(struct archive *_a) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + struct fixup_entry *next, *p; + int fd, ret; + + archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, + "archive_write_disk_close"); + ret = _archive_write_disk_finish_entry(&a->archive); + + /* Sort dir list so directories are fixed up in depth-first order. */ + p = sort_dir_list(a->fixup_list); + + while (p != NULL) { + fd = -1; + a->pst = NULL; /* Mark stat cache as out-of-date. */ + if (p->fixup & + (TODO_TIMES | TODO_MODE_BASE | TODO_ACLS | TODO_FFLAGS)) { + fd = open(p->name, + O_WRONLY | O_BINARY | O_NOFOLLOW | O_CLOEXEC); + } + if (p->fixup & TODO_TIMES) { + set_times(a, fd, p->mode, p->name, + p->atime, p->atime_nanos, + p->birthtime, p->birthtime_nanos, + p->mtime, p->mtime_nanos, + p->ctime, p->ctime_nanos); + } + if (p->fixup & TODO_MODE_BASE) { +#ifdef HAVE_FCHMOD + if (fd >= 0) + fchmod(fd, p->mode); + else +#endif + chmod(p->name, p->mode); + } + if (p->fixup & TODO_ACLS) + archive_write_disk_set_acls(&a->archive, fd, + p->name, &p->acl, p->mode); + if (p->fixup & TODO_FFLAGS) + set_fflags_platform(a, fd, p->name, + p->mode, p->fflags_set, 0); + if (p->fixup & TODO_MAC_METADATA) + set_mac_metadata(a, p->name, p->mac_metadata, + p->mac_metadata_size); + next = p->next; + archive_acl_clear(&p->acl); + free(p->mac_metadata); + free(p->name); + if (fd >= 0) + close(fd); + free(p); + p = next; + } + a->fixup_list = NULL; + return (ret); +} + +static int +_archive_write_disk_free(struct archive *_a) +{ + struct archive_write_disk *a; + int ret; + if (_a == NULL) + return (ARCHIVE_OK); + archive_check_magic(_a, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_write_disk_free"); + a = (struct archive_write_disk *)_a; + ret = _archive_write_disk_close(&a->archive); + archive_write_disk_set_group_lookup(&a->archive, NULL, NULL, NULL); + archive_write_disk_set_user_lookup(&a->archive, NULL, NULL, NULL); + archive_entry_free(a->entry); + archive_string_free(&a->_name_data); + archive_string_free(&a->_tmpname_data); + archive_string_free(&a->archive.error_string); + archive_string_free(&a->path_safe); + a->archive.magic = 0; + __archive_clean(&a->archive); + free(a->decmpfs_header_p); + free(a->resource_fork); + free(a->compressed_buffer); + free(a->uncompressed_buffer); +#if defined(__APPLE__) && defined(UF_COMPRESSED) && defined(HAVE_SYS_XATTR_H)\ + && defined(HAVE_ZLIB_H) + if (a->stream_valid) { + switch (deflateEnd(&a->stream)) { + case Z_OK: + break; + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Failed to clean up compressor"); + ret = ARCHIVE_FATAL; + break; + } + } +#endif + free(a); + return (ret); +} + +/* + * Simple O(n log n) merge sort to order the fixup list. In + * particular, we want to restore dir timestamps depth-first. + */ +static struct fixup_entry * +sort_dir_list(struct fixup_entry *p) +{ + struct fixup_entry *a, *b, *t; + + if (p == NULL) + return (NULL); + /* A one-item list is already sorted. */ + if (p->next == NULL) + return (p); + + /* Step 1: split the list. */ + t = p; + a = p->next->next; + while (a != NULL) { + /* Step a twice, t once. */ + a = a->next; + if (a != NULL) + a = a->next; + t = t->next; + } + /* Now, t is at the mid-point, so break the list here. */ + b = t->next; + t->next = NULL; + a = p; + + /* Step 2: Recursively sort the two sub-lists. */ + a = sort_dir_list(a); + b = sort_dir_list(b); + + /* Step 3: Merge the returned lists. */ + /* Pick the first element for the merged list. */ + if (strcmp(a->name, b->name) > 0) { + t = p = a; + a = a->next; + } else { + t = p = b; + b = b->next; + } + + /* Always put the later element on the list first. */ + while (a != NULL && b != NULL) { + if (strcmp(a->name, b->name) > 0) { + t->next = a; + a = a->next; + } else { + t->next = b; + b = b->next; + } + t = t->next; + } + + /* Only one list is non-empty, so just splice it on. */ + if (a != NULL) + t->next = a; + if (b != NULL) + t->next = b; + + return (p); +} + +/* + * Returns a new, initialized fixup entry. + * + * TODO: Reduce the memory requirements for this list by using a tree + * structure rather than a simple list of names. + */ +static struct fixup_entry * +new_fixup(struct archive_write_disk *a, const char *pathname) +{ + struct fixup_entry *fe; + + fe = (struct fixup_entry *)calloc(1, sizeof(struct fixup_entry)); + if (fe == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for a fixup"); + return (NULL); + } + fe->next = a->fixup_list; + a->fixup_list = fe; + fe->fixup = 0; + fe->name = strdup(pathname); + return (fe); +} + +/* + * Returns a fixup structure for the current entry. + */ +static struct fixup_entry * +current_fixup(struct archive_write_disk *a, const char *pathname) +{ + if (a->current_fixup == NULL) + a->current_fixup = new_fixup(a, pathname); + return (a->current_fixup); +} + +/* Error helper for new *_fsobj functions */ +static void +fsobj_error(int *a_eno, struct archive_string *a_estr, + int err, const char *errstr, const char *path) +{ + if (a_eno) + *a_eno = err; + if (a_estr) + archive_string_sprintf(a_estr, "%s%s", errstr, path); +} + +/* + * TODO: Someday, integrate this with the deep dir support; they both + * scan the path and both can be optimized by comparing against other + * recent paths. + */ +/* + * Checks the given path to see if any elements along it are symlinks. Returns + * ARCHIVE_OK if there are none, otherwise puts an error in errmsg. + */ +static int +check_symlinks_fsobj(char *path, int *a_eno, struct archive_string *a_estr, + int flags) +{ +#if !defined(HAVE_LSTAT) && \ + !(defined(HAVE_OPENAT) && defined(HAVE_FSTATAT) && defined(HAVE_UNLINKAT)) + /* Platform doesn't have lstat, so we can't look for symlinks. */ + (void)path; /* UNUSED */ + (void)error_number; /* UNUSED */ + (void)error_string; /* UNUSED */ + (void)flags; /* UNUSED */ + return (ARCHIVE_OK); +#else + int res = ARCHIVE_OK; + char *tail; + char *head; + int last; + char c; + int r; + struct stat st; + int chdir_fd; +#if defined(HAVE_OPENAT) && defined(HAVE_FSTATAT) && defined(HAVE_UNLINKAT) + int fd; +#endif + + /* Nothing to do here if name is empty */ + if(path[0] == '\0') + return (ARCHIVE_OK); + + /* + * Guard against symlink tricks. Reject any archive entry whose + * destination would be altered by a symlink. + * + * Walk the filename in chunks separated by '/'. For each segment: + * - if it doesn't exist, continue + * - if it's symlink, abort or remove it + * - if it's a directory and it's not the last chunk, cd into it + * As we go: + * head points to the current (relative) path + * tail points to the temporary \0 terminating the segment we're + * currently examining + * c holds what used to be in *tail + * last is 1 if this is the last tail + */ + chdir_fd = la_opendirat(AT_FDCWD, "."); + __archive_ensure_cloexec_flag(chdir_fd); + if (chdir_fd < 0) { + fsobj_error(a_eno, a_estr, errno, + "Could not open ", path); + return (ARCHIVE_FATAL); + } + head = path; + tail = path; + last = 0; + /* TODO: reintroduce a safe cache here? */ + /* Skip the root directory if the path is absolute. */ + if(tail == path && tail[0] == '/') + ++tail; + /* Keep going until we've checked the entire name. + * head, tail, path all alias the same string, which is + * temporarily zeroed at tail, so be careful restoring the + * stashed (c=tail[0]) for error messages. + * Exiting the loop with break is okay; continue is not. + */ + while (!last) { + /* + * Skip the separator we just consumed, plus any adjacent ones + */ + while (*tail == '/') + ++tail; + /* Skip the next path element. */ + while (*tail != '\0' && *tail != '/') + ++tail; + /* is this the last path component? */ + last = (tail[0] == '\0') || (tail[0] == '/' && tail[1] == '\0'); + /* temporarily truncate the string here */ + c = tail[0]; + tail[0] = '\0'; + /* Check that we haven't hit a symlink. */ +#if defined(HAVE_OPENAT) && defined(HAVE_FSTATAT) && defined(HAVE_UNLINKAT) + r = fstatat(chdir_fd, head, &st, AT_SYMLINK_NOFOLLOW); +#else + r = lstat(head, &st); +#endif + if (r != 0) { + tail[0] = c; + /* We've hit a dir that doesn't exist; stop now. */ + if (errno == ENOENT) { + break; + } else { + /* + * Treat any other error as fatal - best to be + * paranoid here. + * Note: This effectively disables deep + * directory support when security checks are + * enabled. Otherwise, very long pathnames that + * trigger an error here could evade the + * sandbox. + * TODO: We could do better, but it would + * probably require merging the symlink checks + * with the deep-directory editing. + */ + fsobj_error(a_eno, a_estr, errno, + "Could not stat ", path); + res = ARCHIVE_FAILED; + break; + } + } else if (S_ISDIR(st.st_mode)) { + if (!last) { +#if defined(HAVE_OPENAT) && defined(HAVE_FSTATAT) && defined(HAVE_UNLINKAT) + fd = la_opendirat(chdir_fd, head); + if (fd < 0) + r = -1; + else { + r = 0; + close(chdir_fd); + chdir_fd = fd; + } +#else + r = chdir(head); +#endif + if (r != 0) { + tail[0] = c; + fsobj_error(a_eno, a_estr, errno, + "Could not chdir ", path); + res = (ARCHIVE_FATAL); + break; + } + /* Our view is now from inside this dir: */ + head = tail + 1; + } + } else if (S_ISLNK(st.st_mode)) { + if (last) { + /* + * Last element is symlink; remove it + * so we can overwrite it with the + * item being extracted. + */ +#if defined(HAVE_OPENAT) && defined(HAVE_FSTATAT) && defined(HAVE_UNLINKAT) + r = unlinkat(chdir_fd, head, 0); +#else + r = unlink(head); +#endif + if (r != 0) { + tail[0] = c; + fsobj_error(a_eno, a_estr, errno, + "Could not remove symlink ", + path); + res = ARCHIVE_FAILED; + break; + } + /* + * Even if we did remove it, a warning + * is in order. The warning is silly, + * though, if we're just replacing one + * symlink with another symlink. + */ + tail[0] = c; + /* + * FIXME: not sure how important this is to + * restore + */ + /* + if (!S_ISLNK(path)) { + fsobj_error(a_eno, a_estr, 0, + "Removing symlink ", path); + } + */ + /* Symlink gone. No more problem! */ + res = ARCHIVE_OK; + break; + } else if (flags & ARCHIVE_EXTRACT_UNLINK) { + /* User asked us to remove problems. */ +#if defined(HAVE_OPENAT) && defined(HAVE_FSTATAT) && defined(HAVE_UNLINKAT) + r = unlinkat(chdir_fd, head, 0); +#else + r = unlink(head); +#endif + if (r != 0) { + tail[0] = c; + fsobj_error(a_eno, a_estr, 0, + "Cannot remove intervening " + "symlink ", path); + res = ARCHIVE_FAILED; + break; + } + tail[0] = c; + } else if ((flags & + ARCHIVE_EXTRACT_SECURE_SYMLINKS) == 0) { + /* + * We are not the last element and we want to + * follow symlinks if they are a directory. + * + * This is needed to extract hardlinks over + * symlinks. + */ +#if defined(HAVE_OPENAT) && defined(HAVE_FSTATAT) && defined(HAVE_UNLINKAT) + r = fstatat(chdir_fd, head, &st, 0); +#else + r = la_stat(head, &st); +#endif + if (r != 0) { + tail[0] = c; + if (errno == ENOENT) { + break; + } else { + fsobj_error(a_eno, a_estr, + errno, + "Could not stat ", path); + res = (ARCHIVE_FAILED); + break; + } + } else if (S_ISDIR(st.st_mode)) { +#if defined(HAVE_OPENAT) && defined(HAVE_FSTATAT) && defined(HAVE_UNLINKAT) + fd = la_opendirat(chdir_fd, head); + if (fd < 0) + r = -1; + else { + r = 0; + close(chdir_fd); + chdir_fd = fd; + } +#else + r = chdir(head); +#endif + if (r != 0) { + tail[0] = c; + fsobj_error(a_eno, a_estr, + errno, + "Could not chdir ", path); + res = (ARCHIVE_FATAL); + break; + } + /* + * Our view is now from inside + * this dir: + */ + head = tail + 1; + } else { + tail[0] = c; + fsobj_error(a_eno, a_estr, 0, + "Cannot extract through " + "symlink ", path); + res = ARCHIVE_FAILED; + break; + } + } else { + tail[0] = c; + fsobj_error(a_eno, a_estr, 0, + "Cannot extract through symlink ", path); + res = ARCHIVE_FAILED; + break; + } + } + /* be sure to always maintain this */ + tail[0] = c; + if (tail[0] != '\0') + tail++; /* Advance to the next segment. */ + } + /* Catches loop exits via break */ + tail[0] = c; +#if defined(HAVE_OPENAT) && defined(HAVE_FSTATAT) && defined(HAVE_UNLINKAT) + /* If we operate with openat(), fstatat() and unlinkat() there was + * no chdir(), so just close the fd */ + if (chdir_fd >= 0) + close(chdir_fd); +#elif HAVE_FCHDIR + /* If we changed directory above, restore it here. */ + if (chdir_fd >= 0) { + r = fchdir(chdir_fd); + if (r != 0) { + fsobj_error(a_eno, a_estr, errno, + "chdir() failure", ""); + } + close(chdir_fd); + chdir_fd = -1; + if (r != 0) { + res = (ARCHIVE_FATAL); + } + } +#endif + /* TODO: reintroduce a safe cache here? */ + return res; +#endif +} + +/* + * Check a->name for symlinks, returning ARCHIVE_OK if its clean, otherwise + * calls archive_set_error and returns ARCHIVE_{FATAL,FAILED} + */ +static int +check_symlinks(struct archive_write_disk *a) +{ + struct archive_string error_string; + int error_number; + int rc; + archive_string_init(&error_string); + rc = check_symlinks_fsobj(a->name, &error_number, &error_string, + a->flags); + if (rc != ARCHIVE_OK) { + archive_set_error(&a->archive, error_number, "%s", + error_string.s); + } + archive_string_free(&error_string); + a->pst = NULL; /* to be safe */ + return rc; +} + + +#if defined(__CYGWIN__) +/* + * 1. Convert a path separator from '\' to '/' . + * We shouldn't check multibyte character directly because some + * character-set have been using the '\' character for a part of + * its multibyte character code. + * 2. Replace unusable characters in Windows with underscore('_'). + * See also : http://msdn.microsoft.com/en-us/library/aa365247.aspx + */ +static void +cleanup_pathname_win(char *path) +{ + wchar_t wc; + char *p; + size_t alen, l; + int mb, complete, utf8; + + alen = 0; + mb = 0; + complete = 1; + utf8 = (strcmp(nl_langinfo(CODESET), "UTF-8") == 0)? 1: 0; + for (p = path; *p != '\0'; p++) { + ++alen; + if (*p == '\\') { + /* If previous byte is smaller than 128, + * this is not second byte of multibyte characters, + * so we can replace '\' with '/'. */ + if (utf8 || !mb) + *p = '/'; + else + complete = 0;/* uncompleted. */ + } else if (*(unsigned char *)p > 127) + mb = 1; + else + mb = 0; + /* Rewrite the path name if its next character is unusable. */ + if (*p == ':' || *p == '*' || *p == '?' || *p == '"' || + *p == '<' || *p == '>' || *p == '|') + *p = '_'; + } + if (complete) + return; + + /* + * Convert path separator in wide-character. + */ + p = path; + while (*p != '\0' && alen) { + l = mbtowc(&wc, p, alen); + if (l == (size_t)-1) { + while (*p != '\0') { + if (*p == '\\') + *p = '/'; + ++p; + } + break; + } + if (l == 1 && wc == L'\\') + *p = '/'; + p += l; + alen -= l; + } +} +#endif + +/* + * Canonicalize the pathname. In particular, this strips duplicate + * '/' characters, '.' elements, and trailing '/'. It also raises an + * error for an empty path, a trailing '..', (if _SECURE_NODOTDOT is + * set) any '..' in the path or (if ARCHIVE_EXTRACT_SECURE_NOABSOLUTEPATHS + * is set) if the path is absolute. + */ +static int +cleanup_pathname_fsobj(char *path, int *a_eno, struct archive_string *a_estr, + int flags) +{ + char *dest, *src; + char separator = '\0'; + + dest = src = path; + if (*src == '\0') { + fsobj_error(a_eno, a_estr, ARCHIVE_ERRNO_MISC, + "Invalid empty ", "pathname"); + return (ARCHIVE_FAILED); + } + +#if defined(__CYGWIN__) + cleanup_pathname_win(path); +#endif + /* Skip leading '/'. */ + if (*src == '/') { + if (flags & ARCHIVE_EXTRACT_SECURE_NOABSOLUTEPATHS) { + fsobj_error(a_eno, a_estr, ARCHIVE_ERRNO_MISC, + "Path is ", "absolute"); + return (ARCHIVE_FAILED); + } + + separator = *src++; + } + + /* Scan the pathname one element at a time. */ + for (;;) { + /* src points to first char after '/' */ + if (src[0] == '\0') { + break; + } else if (src[0] == '/') { + /* Found '//', ignore second one. */ + src++; + continue; + } else if (src[0] == '.') { + if (src[1] == '\0') { + /* Ignore trailing '.' */ + break; + } else if (src[1] == '/') { + /* Skip './'. */ + src += 2; + continue; + } else if (src[1] == '.') { + if (src[2] == '/' || src[2] == '\0') { + /* Conditionally warn about '..' */ + if (flags + & ARCHIVE_EXTRACT_SECURE_NODOTDOT) { + fsobj_error(a_eno, a_estr, + ARCHIVE_ERRNO_MISC, + "Path contains ", "'..'"); + return (ARCHIVE_FAILED); + } + } + /* + * Note: Under no circumstances do we + * remove '..' elements. In + * particular, restoring + * '/foo/../bar/' should create the + * 'foo' dir as a side-effect. + */ + } + } + + /* Copy current element, including leading '/'. */ + if (separator) + *dest++ = '/'; + while (*src != '\0' && *src != '/') { + *dest++ = *src++; + } + + if (*src == '\0') + break; + + /* Skip '/' separator. */ + separator = *src++; + } + /* + * We've just copied zero or more path elements, not including the + * final '/'. + */ + if (dest == path) { + /* + * Nothing got copied. The path must have been something + * like '.' or '/' or './' or '/././././/./'. + */ + if (separator) + *dest++ = '/'; + else + *dest++ = '.'; + } + /* Terminate the result. */ + *dest = '\0'; + return (ARCHIVE_OK); +} + +static int +cleanup_pathname(struct archive_write_disk *a) +{ + struct archive_string error_string; + int error_number; + int rc; + archive_string_init(&error_string); + rc = cleanup_pathname_fsobj(a->name, &error_number, &error_string, + a->flags); + if (rc != ARCHIVE_OK) { + archive_set_error(&a->archive, error_number, "%s", + error_string.s); + } + archive_string_free(&error_string); + return rc; +} + +/* + * Create the parent directory of the specified path, assuming path + * is already in mutable storage. + */ +static int +create_parent_dir(struct archive_write_disk *a, char *path) +{ + char *slash; + int r; + + /* Remove tail element to obtain parent name. */ + slash = strrchr(path, '/'); + if (slash == NULL) + return (ARCHIVE_OK); + *slash = '\0'; + r = create_dir(a, path); + *slash = '/'; + return (r); +} + +/* + * Create the specified dir, recursing to create parents as necessary. + * + * Returns ARCHIVE_OK if the path exists when we're done here. + * Otherwise, returns ARCHIVE_FAILED. + * Assumes path is in mutable storage; path is unchanged on exit. + */ +static int +create_dir(struct archive_write_disk *a, char *path) +{ + struct stat st; + struct fixup_entry *le; + char *slash, *base; + mode_t mode_final, mode; + int r; + + /* Check for special names and just skip them. */ + slash = strrchr(path, '/'); + if (slash == NULL) + base = path; + else + base = slash + 1; + + if (base[0] == '\0' || + (base[0] == '.' && base[1] == '\0') || + (base[0] == '.' && base[1] == '.' && base[2] == '\0')) { + /* Don't bother trying to create null path, '.', or '..'. */ + if (slash != NULL) { + *slash = '\0'; + r = create_dir(a, path); + *slash = '/'; + return (r); + } + return (ARCHIVE_OK); + } + + /* + * Yes, this should be stat() and not lstat(). Using lstat() + * here loses the ability to extract through symlinks. Also note + * that this should not use the a->st cache. + */ + if (la_stat(path, &st) == 0) { + if (S_ISDIR(st.st_mode)) + return (ARCHIVE_OK); + if ((a->flags & ARCHIVE_EXTRACT_NO_OVERWRITE)) { + archive_set_error(&a->archive, EEXIST, + "Can't create directory '%s'", path); + return (ARCHIVE_FAILED); + } + if (unlink(path) != 0) { + archive_set_error(&a->archive, errno, + "Can't create directory '%s': " + "Conflicting file cannot be removed", + path); + return (ARCHIVE_FAILED); + } + } else if (errno != ENOENT && errno != ENOTDIR) { + /* Stat failed? */ + archive_set_error(&a->archive, errno, + "Can't test directory '%s'", path); + return (ARCHIVE_FAILED); + } else if (slash != NULL) { + *slash = '\0'; + r = create_dir(a, path); + *slash = '/'; + if (r != ARCHIVE_OK) + return (r); + } + + /* + * Mode we want for the final restored directory. Per POSIX, + * implicitly-created dirs must be created obeying the umask. + * There's no mention whether this is different for privileged + * restores (which the rest of this code handles by pretending + * umask=0). I've chosen here to always obey the user's umask for + * implicit dirs, even if _EXTRACT_PERM was specified. + */ + mode_final = DEFAULT_DIR_MODE & ~a->user_umask; + /* Mode we want on disk during the restore process. */ + mode = mode_final; + mode |= MINIMUM_DIR_MODE; + mode &= MAXIMUM_DIR_MODE; + if (mkdir(path, mode) == 0) { + if (mode != mode_final) { + le = new_fixup(a, path); + if (le == NULL) + return (ARCHIVE_FATAL); + le->fixup |=TODO_MODE_BASE; + le->mode = mode_final; + } + return (ARCHIVE_OK); + } + + /* + * Without the following check, a/b/../b/c/d fails at the + * second visit to 'b', so 'd' can't be created. Note that we + * don't add it to the fixup list here, as it's already been + * added. + */ + if (la_stat(path, &st) == 0 && S_ISDIR(st.st_mode)) + return (ARCHIVE_OK); + + archive_set_error(&a->archive, errno, "Failed to create dir '%s'", + path); + return (ARCHIVE_FAILED); +} + +/* + * Note: Although we can skip setting the user id if the desired user + * id matches the current user, we cannot skip setting the group, as + * many systems set the gid based on the containing directory. So + * we have to perform a chown syscall if we want to set the SGID + * bit. (The alternative is to stat() and then possibly chown(); it's + * more efficient to skip the stat() and just always chown().) Note + * that a successful chown() here clears the TODO_SGID_CHECK bit, which + * allows set_mode to skip the stat() check for the GID. + */ +static int +set_ownership(struct archive_write_disk *a) +{ +#if !defined(__CYGWIN__) && !defined(__linux__) +/* + * On Linux, a process may have the CAP_CHOWN capability. + * On Windows there is no 'root' user with uid 0. + * Elsewhere we can skip calling chown if we are not root and the desired + * user id does not match the current user. + */ + if (a->user_uid != 0 && a->user_uid != a->uid) { + archive_set_error(&a->archive, errno, + "Can't set UID=%jd", (intmax_t)a->uid); + return (ARCHIVE_WARN); + } +#endif + +#ifdef HAVE_FCHOWN + /* If we have an fd, we can avoid a race. */ + if (a->fd >= 0 && fchown(a->fd, a->uid, a->gid) == 0) { + /* We've set owner and know uid/gid are correct. */ + a->todo &= ~(TODO_OWNER | TODO_SGID_CHECK | TODO_SUID_CHECK); + return (ARCHIVE_OK); + } +#endif + + /* We prefer lchown() but will use chown() if that's all we have. */ + /* Of course, if we have neither, this will always fail. */ +#ifdef HAVE_LCHOWN + if (lchown(a->name, a->uid, a->gid) == 0) { + /* We've set owner and know uid/gid are correct. */ + a->todo &= ~(TODO_OWNER | TODO_SGID_CHECK | TODO_SUID_CHECK); + return (ARCHIVE_OK); + } +#elif HAVE_CHOWN + if (!S_ISLNK(a->mode) && chown(a->name, a->uid, a->gid) == 0) { + /* We've set owner and know uid/gid are correct. */ + a->todo &= ~(TODO_OWNER | TODO_SGID_CHECK | TODO_SUID_CHECK); + return (ARCHIVE_OK); + } +#endif + + archive_set_error(&a->archive, errno, + "Can't set user=%jd/group=%jd for %s", + (intmax_t)a->uid, (intmax_t)a->gid, a->name); + return (ARCHIVE_WARN); +} + +/* + * Note: Returns 0 on success, non-zero on failure. + */ +static int +set_time(int fd, int mode, const char *name, + time_t atime, long atime_nsec, + time_t mtime, long mtime_nsec) +{ + /* Select the best implementation for this platform. */ +#if defined(HAVE_UTIMENSAT) && defined(HAVE_FUTIMENS) + /* + * utimensat() and futimens() are defined in + * POSIX.1-2008. They support ns resolution and setting times + * on fds and symlinks. + */ + struct timespec ts[2]; + (void)mode; /* UNUSED */ + ts[0].tv_sec = atime; + ts[0].tv_nsec = atime_nsec; + ts[1].tv_sec = mtime; + ts[1].tv_nsec = mtime_nsec; + if (fd >= 0) + return futimens(fd, ts); + return utimensat(AT_FDCWD, name, ts, AT_SYMLINK_NOFOLLOW); + +#elif HAVE_UTIMES + /* + * The utimes()-family functions support µs-resolution and + * setting times fds and symlinks. utimes() is documented as + * LEGACY by POSIX, futimes() and lutimes() are not described + * in POSIX. + */ + struct timeval times[2]; + + times[0].tv_sec = atime; + times[0].tv_usec = atime_nsec / 1000; + times[1].tv_sec = mtime; + times[1].tv_usec = mtime_nsec / 1000; + +#ifdef HAVE_FUTIMES + if (fd >= 0) + return (futimes(fd, times)); +#else + (void)fd; /* UNUSED */ +#endif +#ifdef HAVE_LUTIMES + (void)mode; /* UNUSED */ + return (lutimes(name, times)); +#else + if (S_ISLNK(mode)) + return (0); + return (utimes(name, times)); +#endif + +#elif defined(HAVE_UTIME) + /* + * utime() is POSIX-standard but only supports 1s resolution and + * does not support fds or symlinks. + */ + struct utimbuf times; + (void)fd; /* UNUSED */ + (void)name; /* UNUSED */ + (void)atime_nsec; /* UNUSED */ + (void)mtime_nsec; /* UNUSED */ + times.actime = atime; + times.modtime = mtime; + if (S_ISLNK(mode)) + return (ARCHIVE_OK); + return (utime(name, ×)); + +#else + /* + * We don't know how to set the time on this platform. + */ + (void)fd; /* UNUSED */ + (void)mode; /* UNUSED */ + (void)name; /* UNUSED */ + (void)atime_nsec; /* UNUSED */ + (void)mtime_nsec; /* UNUSED */ + return (ARCHIVE_WARN); +#endif +} + +#ifdef F_SETTIMES +static int +set_time_tru64(int fd, int mode, const char *name, + time_t atime, long atime_nsec, + time_t mtime, long mtime_nsec, + time_t ctime, long ctime_nsec) +{ + struct attr_timbuf tstamp; + tstamp.atime.tv_sec = atime; + tstamp.mtime.tv_sec = mtime; + tstamp.ctime.tv_sec = ctime; +#if defined (__hpux) && defined (__ia64) + tstamp.atime.tv_nsec = atime_nsec; + tstamp.mtime.tv_nsec = mtime_nsec; + tstamp.ctime.tv_nsec = ctime_nsec; +#else + tstamp.atime.tv_usec = atime_nsec / 1000; + tstamp.mtime.tv_usec = mtime_nsec / 1000; + tstamp.ctime.tv_usec = ctime_nsec / 1000; +#endif + return (fcntl(fd,F_SETTIMES,&tstamp)); +} +#endif /* F_SETTIMES */ + +static int +set_times(struct archive_write_disk *a, + int fd, int mode, const char *name, + time_t atime, long atime_nanos, + time_t birthtime, long birthtime_nanos, + time_t mtime, long mtime_nanos, + time_t cctime, long ctime_nanos) +{ + /* Note: set_time doesn't use libarchive return conventions! + * It uses syscall conventions. So 0 here instead of ARCHIVE_OK. */ + int r1 = 0, r2 = 0; + +#ifdef F_SETTIMES + /* + * on Tru64 try own fcntl first which can restore even the + * ctime, fall back to default code path below if it fails + * or if we are not running as root + */ + if (a->user_uid == 0 && + set_time_tru64(fd, mode, name, + atime, atime_nanos, mtime, + mtime_nanos, cctime, ctime_nanos) == 0) { + return (ARCHIVE_OK); + } +#else /* Tru64 */ + (void)cctime; /* UNUSED */ + (void)ctime_nanos; /* UNUSED */ +#endif /* Tru64 */ + +#ifdef HAVE_STRUCT_STAT_ST_BIRTHTIME + /* + * If you have struct stat.st_birthtime, we assume BSD + * birthtime semantics, in which {f,l,}utimes() updates + * birthtime to earliest mtime. So we set the time twice, + * first using the birthtime, then using the mtime. If + * birthtime == mtime, this isn't necessary, so we skip it. + * If birthtime > mtime, then this won't work, so we skip it. + */ + if (birthtime < mtime + || (birthtime == mtime && birthtime_nanos < mtime_nanos)) + r1 = set_time(fd, mode, name, + atime, atime_nanos, + birthtime, birthtime_nanos); +#else + (void)birthtime; /* UNUSED */ + (void)birthtime_nanos; /* UNUSED */ +#endif + r2 = set_time(fd, mode, name, + atime, atime_nanos, + mtime, mtime_nanos); + if (r1 != 0 || r2 != 0) { + archive_set_error(&a->archive, errno, + "Can't restore time"); + return (ARCHIVE_WARN); + } + return (ARCHIVE_OK); +} + +static int +set_times_from_entry(struct archive_write_disk *a) +{ + time_t atime, birthtime, mtime, cctime; + long atime_nsec, birthtime_nsec, mtime_nsec, ctime_nsec; + + /* Suitable defaults. */ + atime = birthtime = mtime = cctime = a->start_time; + atime_nsec = birthtime_nsec = mtime_nsec = ctime_nsec = 0; + + /* If no time was provided, we're done. */ + if (!archive_entry_atime_is_set(a->entry) +#if HAVE_STRUCT_STAT_ST_BIRTHTIME + && !archive_entry_birthtime_is_set(a->entry) +#endif + && !archive_entry_mtime_is_set(a->entry)) + return (ARCHIVE_OK); + + if (archive_entry_atime_is_set(a->entry)) { + atime = archive_entry_atime(a->entry); + atime_nsec = archive_entry_atime_nsec(a->entry); + } + if (archive_entry_birthtime_is_set(a->entry)) { + birthtime = archive_entry_birthtime(a->entry); + birthtime_nsec = archive_entry_birthtime_nsec(a->entry); + } + if (archive_entry_mtime_is_set(a->entry)) { + mtime = archive_entry_mtime(a->entry); + mtime_nsec = archive_entry_mtime_nsec(a->entry); + } + if (archive_entry_ctime_is_set(a->entry)) { + cctime = archive_entry_ctime(a->entry); + ctime_nsec = archive_entry_ctime_nsec(a->entry); + } + + return set_times(a, a->fd, a->mode, a->name, + atime, atime_nsec, + birthtime, birthtime_nsec, + mtime, mtime_nsec, + cctime, ctime_nsec); +} + +static int +set_mode(struct archive_write_disk *a, int mode) +{ + int r = ARCHIVE_OK; + int r2; + mode &= 07777; /* Strip off file type bits. */ + + if (a->todo & TODO_SGID_CHECK) { + /* + * If we don't know the GID is right, we must stat() + * to verify it. We can't just check the GID of this + * process, since systems sometimes set GID from + * the enclosing dir or based on ACLs. + */ + if ((r = lazy_stat(a)) != ARCHIVE_OK) + return (r); + if (a->pst->st_gid != a->gid) { + mode &= ~ S_ISGID; + if (a->flags & ARCHIVE_EXTRACT_OWNER) { + /* + * This is only an error if you + * requested owner restore. If you + * didn't, we'll try to restore + * sgid/suid, but won't consider it a + * problem if we can't. + */ + archive_set_error(&a->archive, -1, + "Can't restore SGID bit"); + r = ARCHIVE_WARN; + } + } + /* While we're here, double-check the UID. */ + if (a->pst->st_uid != a->uid + && (a->todo & TODO_SUID)) { + mode &= ~ S_ISUID; + if (a->flags & ARCHIVE_EXTRACT_OWNER) { + archive_set_error(&a->archive, -1, + "Can't restore SUID bit"); + r = ARCHIVE_WARN; + } + } + a->todo &= ~TODO_SGID_CHECK; + a->todo &= ~TODO_SUID_CHECK; + } else if (a->todo & TODO_SUID_CHECK) { + /* + * If we don't know the UID is right, we can just check + * the user, since all systems set the file UID from + * the process UID. + */ + if (a->user_uid != a->uid) { + mode &= ~ S_ISUID; + if (a->flags & ARCHIVE_EXTRACT_OWNER) { + archive_set_error(&a->archive, -1, + "Can't make file SUID"); + r = ARCHIVE_WARN; + } + } + a->todo &= ~TODO_SUID_CHECK; + } + + if (S_ISLNK(a->mode)) { +#ifdef HAVE_LCHMOD + /* + * If this is a symlink, use lchmod(). If the + * platform doesn't support lchmod(), just skip it. A + * platform that doesn't provide a way to set + * permissions on symlinks probably ignores + * permissions on symlinks, so a failure here has no + * impact. + */ + if (lchmod(a->name, mode) != 0) { + switch (errno) { + case ENOTSUP: + case ENOSYS: +#if ENOTSUP != EOPNOTSUPP + case EOPNOTSUPP: +#endif + /* + * if lchmod is defined but the platform + * doesn't support it, silently ignore + * error + */ + break; + default: + archive_set_error(&a->archive, errno, + "Can't set permissions to 0%o", (int)mode); + r = ARCHIVE_WARN; + } + } +#endif + } else if (!S_ISDIR(a->mode)) { + /* + * If it's not a symlink and not a dir, then use + * fchmod() or chmod(), depending on whether we have + * an fd. Dirs get their perms set during the + * post-extract fixup, which is handled elsewhere. + */ +#ifdef HAVE_FCHMOD + if (a->fd >= 0) + r2 = fchmod(a->fd, mode); + else +#endif + /* If this platform lacks fchmod(), then + * we'll just use chmod(). */ + r2 = chmod(a->name, mode); + + if (r2 != 0) { + archive_set_error(&a->archive, errno, + "Can't set permissions to 0%o", (int)mode); + r = ARCHIVE_WARN; + } + } + return (r); +} + +static int +set_fflags(struct archive_write_disk *a) +{ + struct fixup_entry *le; + unsigned long set, clear; + int r; + mode_t mode = archive_entry_mode(a->entry); + /* + * Make 'critical_flags' hold all file flags that can't be + * immediately restored. For example, on BSD systems, + * SF_IMMUTABLE prevents hardlinks from being created, so + * should not be set until after any hardlinks are created. To + * preserve some semblance of portability, this uses #ifdef + * extensively. Ugly, but it works. + * + * Yes, Virginia, this does create a security race. It's mitigated + * somewhat by the practice of creating dirs 0700 until the extract + * is done, but it would be nice if we could do more than that. + * People restoring critical file systems should be wary of + * other programs that might try to muck with files as they're + * being restored. + */ + const int critical_flags = 0 +#ifdef SF_IMMUTABLE + | SF_IMMUTABLE +#endif +#ifdef UF_IMMUTABLE + | UF_IMMUTABLE +#endif +#ifdef SF_APPEND + | SF_APPEND +#endif +#ifdef UF_APPEND + | UF_APPEND +#endif +#if defined(FS_APPEND_FL) + | FS_APPEND_FL +#elif defined(EXT2_APPEND_FL) + | EXT2_APPEND_FL +#endif +#if defined(FS_IMMUTABLE_FL) + | FS_IMMUTABLE_FL +#elif defined(EXT2_IMMUTABLE_FL) + | EXT2_IMMUTABLE_FL +#endif +#ifdef FS_JOURNAL_DATA_FL + | FS_JOURNAL_DATA_FL +#endif + ; + + if (a->todo & TODO_FFLAGS) { + archive_entry_fflags(a->entry, &set, &clear); + + /* + * The first test encourages the compiler to eliminate + * all of this if it's not necessary. + */ + if ((critical_flags != 0) && (set & critical_flags)) { + le = current_fixup(a, a->name); + if (le == NULL) + return (ARCHIVE_FATAL); + le->fixup |= TODO_FFLAGS; + le->fflags_set = set; + /* Store the mode if it's not already there. */ + if ((le->fixup & TODO_MODE) == 0) + le->mode = mode; + } else { + r = set_fflags_platform(a, a->fd, + a->name, mode, set, clear); + if (r != ARCHIVE_OK) + return (r); + } + } + return (ARCHIVE_OK); +} + +static int +clear_nochange_fflags(struct archive_write_disk *a) +{ + mode_t mode = archive_entry_mode(a->entry); + const int nochange_flags = 0 +#ifdef SF_IMMUTABLE + | SF_IMMUTABLE +#endif +#ifdef UF_IMMUTABLE + | UF_IMMUTABLE +#endif +#ifdef SF_APPEND + | SF_APPEND +#endif +#ifdef UF_APPEND + | UF_APPEND +#endif +#ifdef EXT2_APPEND_FL + | EXT2_APPEND_FL +#endif +#ifdef EXT2_IMMUTABLE_FL + | EXT2_IMMUTABLE_FL +#endif + ; + + return (set_fflags_platform(a, a->fd, a->name, mode, 0, + nochange_flags)); +} + + +#if ( defined(HAVE_LCHFLAGS) || defined(HAVE_CHFLAGS) || defined(HAVE_FCHFLAGS) ) && defined(HAVE_STRUCT_STAT_ST_FLAGS) +/* + * BSD reads flags using stat() and sets them with one of {f,l,}chflags() + */ +static int +set_fflags_platform(struct archive_write_disk *a, int fd, const char *name, + mode_t mode, unsigned long set, unsigned long clear) +{ + int r; + const int sf_mask = 0 +#ifdef SF_APPEND + | SF_APPEND +#endif +#ifdef SF_ARCHIVED + | SF_ARCHIVED +#endif +#ifdef SF_IMMUTABLE + | SF_IMMUTABLE +#endif +#ifdef SF_NOUNLINK + | SF_NOUNLINK +#endif + ; + (void)mode; /* UNUSED */ + + if (set == 0 && clear == 0) + return (ARCHIVE_OK); + + /* + * XXX Is the stat here really necessary? Or can I just use + * the 'set' flags directly? In particular, I'm not sure + * about the correct approach if we're overwriting an existing + * file that already has flags on it. XXX + */ + if ((r = lazy_stat(a)) != ARCHIVE_OK) + return (r); + + a->st.st_flags &= ~clear; + a->st.st_flags |= set; + + /* Only super-user may change SF_* flags */ + + if (a->user_uid != 0) + a->st.st_flags &= ~sf_mask; + +#ifdef HAVE_FCHFLAGS + /* If platform has fchflags() and we were given an fd, use it. */ + if (fd >= 0 && fchflags(fd, a->st.st_flags) == 0) + return (ARCHIVE_OK); +#endif + /* + * If we can't use the fd to set the flags, we'll use the + * pathname to set flags. We prefer lchflags() but will use + * chflags() if we must. + */ +#ifdef HAVE_LCHFLAGS + if (lchflags(name, a->st.st_flags) == 0) + return (ARCHIVE_OK); +#elif defined(HAVE_CHFLAGS) + if (S_ISLNK(a->st.st_mode)) { + archive_set_error(&a->archive, errno, + "Can't set file flags on symlink."); + return (ARCHIVE_WARN); + } + if (chflags(name, a->st.st_flags) == 0) + return (ARCHIVE_OK); +#endif + archive_set_error(&a->archive, errno, + "Failed to set file flags"); + return (ARCHIVE_WARN); +} + +#elif (defined(FS_IOC_GETFLAGS) && defined(FS_IOC_SETFLAGS) && \ + defined(HAVE_WORKING_FS_IOC_GETFLAGS)) || \ + (defined(EXT2_IOC_GETFLAGS) && defined(EXT2_IOC_SETFLAGS) && \ + defined(HAVE_WORKING_EXT2_IOC_GETFLAGS)) +/* + * Linux uses ioctl() to read and write file flags. + */ +static int +set_fflags_platform(struct archive_write_disk *a, int fd, const char *name, + mode_t mode, unsigned long set, unsigned long clear) +{ + int ret; + int myfd = fd; + int newflags, oldflags; + /* + * Linux has no define for the flags that are only settable by + * the root user. This code may seem a little complex, but + * there seem to be some Linux systems that lack these + * defines. (?) The code below degrades reasonably gracefully + * if sf_mask is incomplete. + */ + const int sf_mask = 0 +#if defined(FS_IMMUTABLE_FL) + | FS_IMMUTABLE_FL +#elif defined(EXT2_IMMUTABLE_FL) + | EXT2_IMMUTABLE_FL +#endif +#if defined(FS_APPEND_FL) + | FS_APPEND_FL +#elif defined(EXT2_APPEND_FL) + | EXT2_APPEND_FL +#endif +#if defined(FS_JOURNAL_DATA_FL) + | FS_JOURNAL_DATA_FL +#endif + ; + + if (set == 0 && clear == 0) + return (ARCHIVE_OK); + /* Only regular files and dirs can have flags. */ + if (!S_ISREG(mode) && !S_ISDIR(mode)) + return (ARCHIVE_OK); + + /* If we weren't given an fd, open it ourselves. */ + if (myfd < 0) { + myfd = open(name, O_RDONLY | O_NONBLOCK | O_BINARY | O_CLOEXEC); + __archive_ensure_cloexec_flag(myfd); + } + if (myfd < 0) + return (ARCHIVE_OK); + + /* + * XXX As above, this would be way simpler if we didn't have + * to read the current flags from disk. XXX + */ + ret = ARCHIVE_OK; + + /* Read the current file flags. */ + if (ioctl(myfd, +#ifdef FS_IOC_GETFLAGS + FS_IOC_GETFLAGS, +#else + EXT2_IOC_GETFLAGS, +#endif + &oldflags) < 0) + goto fail; + + /* Try setting the flags as given. */ + newflags = (oldflags & ~clear) | set; + if (ioctl(myfd, +#ifdef FS_IOC_SETFLAGS + FS_IOC_SETFLAGS, +#else + EXT2_IOC_SETFLAGS, +#endif + &newflags) >= 0) + goto cleanup; + if (errno != EPERM) + goto fail; + + /* If we couldn't set all the flags, try again with a subset. */ + newflags &= ~sf_mask; + oldflags &= sf_mask; + newflags |= oldflags; + if (ioctl(myfd, +#ifdef FS_IOC_SETFLAGS + FS_IOC_SETFLAGS, +#else + EXT2_IOC_SETFLAGS, +#endif + &newflags) >= 0) + goto cleanup; + + /* We couldn't set the flags, so report the failure. */ +fail: + archive_set_error(&a->archive, errno, + "Failed to set file flags"); + ret = ARCHIVE_WARN; +cleanup: + if (fd < 0) + close(myfd); + return (ret); +} + +#else + +/* + * Of course, some systems have neither BSD chflags() nor Linux' flags + * support through ioctl(). + */ +static int +set_fflags_platform(struct archive_write_disk *a, int fd, const char *name, + mode_t mode, unsigned long set, unsigned long clear) +{ + (void)a; /* UNUSED */ + (void)fd; /* UNUSED */ + (void)name; /* UNUSED */ + (void)mode; /* UNUSED */ + (void)set; /* UNUSED */ + (void)clear; /* UNUSED */ + return (ARCHIVE_OK); +} + +#endif /* __linux */ + +#ifndef HAVE_COPYFILE_H +/* Default is to simply drop Mac extended metadata. */ +static int +set_mac_metadata(struct archive_write_disk *a, const char *pathname, + const void *metadata, size_t metadata_size) +{ + (void)a; /* UNUSED */ + (void)pathname; /* UNUSED */ + (void)metadata; /* UNUSED */ + (void)metadata_size; /* UNUSED */ + return (ARCHIVE_OK); +} + +static int +fixup_appledouble(struct archive_write_disk *a, const char *pathname) +{ + (void)a; /* UNUSED */ + (void)pathname; /* UNUSED */ + return (ARCHIVE_OK); +} +#else + +/* + * On Mac OS, we use copyfile() to unpack the metadata and + * apply it to the target file. + */ + +#if defined(HAVE_SYS_XATTR_H) +static int +copy_xattrs(struct archive_write_disk *a, int tmpfd, int dffd) +{ + ssize_t xattr_size; + char *xattr_names = NULL, *xattr_val = NULL; + int ret = ARCHIVE_OK, xattr_i; + + xattr_size = flistxattr(tmpfd, NULL, 0, 0); + if (xattr_size == -1) { + archive_set_error(&a->archive, errno, + "Failed to read metadata(xattr)"); + ret = ARCHIVE_WARN; + goto exit_xattr; + } + xattr_names = malloc(xattr_size); + if (xattr_names == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for metadata(xattr)"); + ret = ARCHIVE_FATAL; + goto exit_xattr; + } + xattr_size = flistxattr(tmpfd, xattr_names, xattr_size, 0); + if (xattr_size == -1) { + archive_set_error(&a->archive, errno, + "Failed to read metadata(xattr)"); + ret = ARCHIVE_WARN; + goto exit_xattr; + } + for (xattr_i = 0; xattr_i < xattr_size; + xattr_i += strlen(xattr_names + xattr_i) + 1) { + char *xattr_val_saved; + ssize_t s; + int f; + + s = fgetxattr(tmpfd, xattr_names + xattr_i, NULL, 0, 0, 0); + if (s == -1) { + archive_set_error(&a->archive, errno, + "Failed to get metadata(xattr)"); + ret = ARCHIVE_WARN; + goto exit_xattr; + } + xattr_val_saved = xattr_val; + xattr_val = realloc(xattr_val, s); + if (xattr_val == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Failed to get metadata(xattr)"); + ret = ARCHIVE_WARN; + free(xattr_val_saved); + goto exit_xattr; + } + s = fgetxattr(tmpfd, xattr_names + xattr_i, xattr_val, s, 0, 0); + if (s == -1) { + archive_set_error(&a->archive, errno, + "Failed to get metadata(xattr)"); + ret = ARCHIVE_WARN; + goto exit_xattr; + } + f = fsetxattr(dffd, xattr_names + xattr_i, xattr_val, s, 0, 0); + if (f == -1) { + archive_set_error(&a->archive, errno, + "Failed to get metadata(xattr)"); + ret = ARCHIVE_WARN; + goto exit_xattr; + } + } +exit_xattr: + free(xattr_names); + free(xattr_val); + return (ret); +} +#endif + +static int +copy_acls(struct archive_write_disk *a, int tmpfd, int dffd) +{ +#ifndef HAVE_SYS_ACL_H + return 0; +#else + acl_t acl, dfacl = NULL; + int acl_r, ret = ARCHIVE_OK; + + acl = acl_get_fd(tmpfd); + if (acl == NULL) { + if (errno == ENOENT) + /* There are not any ACLs. */ + return (ret); + archive_set_error(&a->archive, errno, + "Failed to get metadata(acl)"); + ret = ARCHIVE_WARN; + goto exit_acl; + } + dfacl = acl_dup(acl); + acl_r = acl_set_fd(dffd, dfacl); + if (acl_r == -1) { + archive_set_error(&a->archive, errno, + "Failed to get metadata(acl)"); + ret = ARCHIVE_WARN; + goto exit_acl; + } +exit_acl: + if (acl) + acl_free(acl); + if (dfacl) + acl_free(dfacl); + return (ret); +#endif +} + +static int +create_tempdatafork(struct archive_write_disk *a, const char *pathname) +{ + struct archive_string tmpdatafork; + int tmpfd; + + archive_string_init(&tmpdatafork); + archive_strcpy(&tmpdatafork, "tar.md.XXXXXX"); + tmpfd = mkstemp(tmpdatafork.s); + if (tmpfd < 0) { + archive_set_error(&a->archive, errno, + "Failed to mkstemp"); + archive_string_free(&tmpdatafork); + return (-1); + } + if (copyfile(pathname, tmpdatafork.s, 0, + COPYFILE_UNPACK | COPYFILE_NOFOLLOW + | COPYFILE_ACL | COPYFILE_XATTR) < 0) { + archive_set_error(&a->archive, errno, + "Failed to restore metadata"); + close(tmpfd); + tmpfd = -1; + } + unlink(tmpdatafork.s); + archive_string_free(&tmpdatafork); + return (tmpfd); +} + +static int +copy_metadata(struct archive_write_disk *a, const char *metadata, + const char *datafork, int datafork_compressed) +{ + int ret = ARCHIVE_OK; + + if (datafork_compressed) { + int dffd, tmpfd; + + tmpfd = create_tempdatafork(a, metadata); + if (tmpfd == -1) + return (ARCHIVE_WARN); + + /* + * Do not open the data fork compressed by HFS+ compression + * with at least a writing mode(O_RDWR or O_WRONLY). it + * makes the data fork uncompressed. + */ + dffd = open(datafork, 0); + if (dffd == -1) { + archive_set_error(&a->archive, errno, + "Failed to open the data fork for metadata"); + close(tmpfd); + return (ARCHIVE_WARN); + } + +#if defined(HAVE_SYS_XATTR_H) + ret = copy_xattrs(a, tmpfd, dffd); + if (ret == ARCHIVE_OK) +#endif + ret = copy_acls(a, tmpfd, dffd); + close(tmpfd); + close(dffd); + } else { + if (copyfile(metadata, datafork, 0, + COPYFILE_UNPACK | COPYFILE_NOFOLLOW + | COPYFILE_ACL | COPYFILE_XATTR) < 0) { + archive_set_error(&a->archive, errno, + "Failed to restore metadata"); + ret = ARCHIVE_WARN; + } + } + return (ret); +} + +static int +set_mac_metadata(struct archive_write_disk *a, const char *pathname, + const void *metadata, size_t metadata_size) +{ + struct archive_string tmp; + ssize_t written; + int fd; + int ret = ARCHIVE_OK; + + /* This would be simpler if copyfile() could just accept the + * metadata as a block of memory; then we could sidestep this + * silly dance of writing the data to disk just so that + * copyfile() can read it back in again. */ + archive_string_init(&tmp); + archive_strcpy(&tmp, pathname); + archive_strcat(&tmp, ".XXXXXX"); + fd = mkstemp(tmp.s); + + if (fd < 0) { + archive_set_error(&a->archive, errno, + "Failed to restore metadata"); + archive_string_free(&tmp); + return (ARCHIVE_WARN); + } + written = write(fd, metadata, metadata_size); + close(fd); + if ((size_t)written != metadata_size) { + archive_set_error(&a->archive, errno, + "Failed to restore metadata"); + ret = ARCHIVE_WARN; + } else { + int compressed; + +#if defined(UF_COMPRESSED) + if ((a->todo & TODO_HFS_COMPRESSION) != 0 && + (ret = lazy_stat(a)) == ARCHIVE_OK) + compressed = a->st.st_flags & UF_COMPRESSED; + else +#endif + compressed = 0; + ret = copy_metadata(a, tmp.s, pathname, compressed); + } + unlink(tmp.s); + archive_string_free(&tmp); + return (ret); +} + +static int +fixup_appledouble(struct archive_write_disk *a, const char *pathname) +{ + char buff[8]; + struct stat st; + const char *p; + struct archive_string datafork; + int fd = -1, ret = ARCHIVE_OK; + + archive_string_init(&datafork); + /* Check if the current file name is a type of the resource + * fork file. */ + p = strrchr(pathname, '/'); + if (p == NULL) + p = pathname; + else + p++; + if (p[0] != '.' || p[1] != '_') + goto skip_appledouble; + + /* + * Check if the data fork file exists. + * + * TODO: Check if this write disk object has handled it. + */ + archive_strncpy(&datafork, pathname, p - pathname); + archive_strcat(&datafork, p + 2); + if (lstat(datafork.s, &st) == -1 || + (st.st_mode & AE_IFMT) != AE_IFREG) + goto skip_appledouble; + + /* + * Check if the file is in the AppleDouble form. + */ + fd = open(pathname, O_RDONLY | O_BINARY | O_CLOEXEC); + __archive_ensure_cloexec_flag(fd); + if (fd == -1) { + archive_set_error(&a->archive, errno, + "Failed to open a restoring file"); + ret = ARCHIVE_WARN; + goto skip_appledouble; + } + if (read(fd, buff, 8) == -1) { + archive_set_error(&a->archive, errno, + "Failed to read a restoring file"); + close(fd); + ret = ARCHIVE_WARN; + goto skip_appledouble; + } + close(fd); + /* Check AppleDouble Magic Code. */ + if (archive_be32dec(buff) != 0x00051607) + goto skip_appledouble; + /* Check AppleDouble Version. */ + if (archive_be32dec(buff+4) != 0x00020000) + goto skip_appledouble; + + ret = copy_metadata(a, pathname, datafork.s, +#if defined(UF_COMPRESSED) + st.st_flags & UF_COMPRESSED); +#else + 0); +#endif + if (ret == ARCHIVE_OK) { + unlink(pathname); + ret = ARCHIVE_EOF; + } +skip_appledouble: + archive_string_free(&datafork); + return (ret); +} +#endif + +#if ARCHIVE_XATTR_LINUX || ARCHIVE_XATTR_DARWIN || ARCHIVE_XATTR_AIX +/* + * Restore extended attributes - Linux, Darwin and AIX implementations: + * AIX' ea interface is syntaxwise identical to the Linux xattr interface. + */ +static int +set_xattrs(struct archive_write_disk *a) +{ + struct archive_entry *entry = a->entry; + struct archive_string errlist; + int ret = ARCHIVE_OK; + int i = archive_entry_xattr_reset(entry); + short fail = 0; + + archive_string_init(&errlist); + + while (i--) { + const char *name; + const void *value; + size_t size; + int e; + + archive_entry_xattr_next(entry, &name, &value, &size); + + if (name == NULL) + continue; +#if ARCHIVE_XATTR_LINUX + /* Linux: quietly skip POSIX.1e ACL extended attributes */ + if (strncmp(name, "system.", 7) == 0 && + (strcmp(name + 7, "posix_acl_access") == 0 || + strcmp(name + 7, "posix_acl_default") == 0)) + continue; + if (strncmp(name, "trusted.SGI_", 12) == 0 && + (strcmp(name + 12, "ACL_DEFAULT") == 0 || + strcmp(name + 12, "ACL_FILE") == 0)) + continue; + + /* Linux: xfsroot namespace is obsolete and unsupported */ + if (strncmp(name, "xfsroot.", 8) == 0) { + fail = 1; + archive_strcat(&errlist, name); + archive_strappend_char(&errlist, ' '); + continue; + } +#endif + + if (a->fd >= 0) { +#if ARCHIVE_XATTR_LINUX + e = fsetxattr(a->fd, name, value, size, 0); +#elif ARCHIVE_XATTR_DARWIN + e = fsetxattr(a->fd, name, value, size, 0, 0); +#elif ARCHIVE_XATTR_AIX + e = fsetea(a->fd, name, value, size, 0); +#endif + } else { +#if ARCHIVE_XATTR_LINUX + e = lsetxattr(archive_entry_pathname(entry), + name, value, size, 0); +#elif ARCHIVE_XATTR_DARWIN + e = setxattr(archive_entry_pathname(entry), + name, value, size, 0, XATTR_NOFOLLOW); +#elif ARCHIVE_XATTR_AIX + e = lsetea(archive_entry_pathname(entry), + name, value, size, 0); +#endif + } + if (e == -1) { + ret = ARCHIVE_WARN; + archive_strcat(&errlist, name); + archive_strappend_char(&errlist, ' '); + if (errno != ENOTSUP && errno != ENOSYS) + fail = 1; + } + } + + if (ret == ARCHIVE_WARN) { + if (fail && errlist.length > 0) { + errlist.length--; + errlist.s[errlist.length] = '\0'; + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Cannot restore extended attributes: %s", + errlist.s); + } else + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Cannot restore extended " + "attributes on this file system."); + } + + archive_string_free(&errlist); + return (ret); +} +#elif ARCHIVE_XATTR_FREEBSD +/* + * Restore extended attributes - FreeBSD implementation + */ +static int +set_xattrs(struct archive_write_disk *a) +{ + struct archive_entry *entry = a->entry; + struct archive_string errlist; + int ret = ARCHIVE_OK; + int i = archive_entry_xattr_reset(entry); + short fail = 0; + + archive_string_init(&errlist); + + while (i--) { + const char *name; + const void *value; + size_t size; + archive_entry_xattr_next(entry, &name, &value, &size); + if (name != NULL) { + int e; + int namespace; + + if (strncmp(name, "user.", 5) == 0) { + /* "user." attributes go to user namespace */ + name += 5; + namespace = EXTATTR_NAMESPACE_USER; + } else { + /* Other namespaces are unsupported */ + archive_strcat(&errlist, name); + archive_strappend_char(&errlist, ' '); + fail = 1; + ret = ARCHIVE_WARN; + continue; + } + + if (a->fd >= 0) { + e = extattr_set_fd(a->fd, namespace, name, + value, size); + } else { + e = extattr_set_link( + archive_entry_pathname(entry), namespace, + name, value, size); + } + if (e != (int)size) { + archive_strcat(&errlist, name); + archive_strappend_char(&errlist, ' '); + ret = ARCHIVE_WARN; + if (errno != ENOTSUP && errno != ENOSYS) + fail = 1; + } + } + } + + if (ret == ARCHIVE_WARN) { + if (fail && errlist.length > 0) { + errlist.length--; + errlist.s[errlist.length] = '\0'; + + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Cannot restore extended attributes: %s", + errlist.s); + } else + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Cannot restore extended " + "attributes on this file system."); + } + + archive_string_free(&errlist); + return (ret); +} +#else +/* + * Restore extended attributes - stub implementation for unsupported systems + */ +static int +set_xattrs(struct archive_write_disk *a) +{ + static int warning_done = 0; + + /* If there aren't any extended attributes, then it's okay not + * to extract them, otherwise, issue a single warning. */ + if (archive_entry_xattr_count(a->entry) != 0 && !warning_done) { + warning_done = 1; + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Cannot restore extended attributes on this system"); + return (ARCHIVE_WARN); + } + /* Warning was already emitted; suppress further warnings. */ + return (ARCHIVE_OK); +} +#endif + +/* + * Test if file on disk is older than entry. + */ +static int +older(struct stat *st, struct archive_entry *entry) +{ + /* First, test the seconds and return if we have a definite answer. */ + /* Definitely older. */ + if (to_int64_time(st->st_mtime) < to_int64_time(archive_entry_mtime(entry))) + return (1); + /* Definitely younger. */ + if (to_int64_time(st->st_mtime) > to_int64_time(archive_entry_mtime(entry))) + return (0); + /* If this platform supports fractional seconds, try those. */ +#if HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC + /* Definitely older. */ + if (st->st_mtimespec.tv_nsec < archive_entry_mtime_nsec(entry)) + return (1); +#elif HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC + /* Definitely older. */ + if (st->st_mtim.tv_nsec < archive_entry_mtime_nsec(entry)) + return (1); +#elif HAVE_STRUCT_STAT_ST_MTIME_N + /* older. */ + if (st->st_mtime_n < archive_entry_mtime_nsec(entry)) + return (1); +#elif HAVE_STRUCT_STAT_ST_UMTIME + /* older. */ + if (st->st_umtime * 1000 < archive_entry_mtime_nsec(entry)) + return (1); +#elif HAVE_STRUCT_STAT_ST_MTIME_USEC + /* older. */ + if (st->st_mtime_usec * 1000 < archive_entry_mtime_nsec(entry)) + return (1); +#else + /* This system doesn't have high-res timestamps. */ +#endif + /* Same age or newer, so not older. */ + return (0); +} + +#ifndef ARCHIVE_ACL_SUPPORT +int +archive_write_disk_set_acls(struct archive *a, int fd, const char *name, + struct archive_acl *abstract_acl, __LA_MODE_T mode) +{ + (void)a; /* UNUSED */ + (void)fd; /* UNUSED */ + (void)name; /* UNUSED */ + (void)abstract_acl; /* UNUSED */ + (void)mode; /* UNUSED */ + return (ARCHIVE_OK); +} +#endif + +#endif /* !_WIN32 || __CYGWIN__ */ + diff --git a/c/archive_write_disk_private.h b/c/archive_write_disk_private.h new file mode 100644 --- /dev/null +++ b/c/archive_write_disk_private.h @@ -0,0 +1,45 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer + * in this position and unchanged. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD: head/lib/libarchive/archive_write_disk_private.h 201086 2009-12-28 02:17:53Z kientzle $ + */ + +#ifndef ARCHIVE_WRITE_DISK_PRIVATE_H_INCLUDED +#define ARCHIVE_WRITE_DISK_PRIVATE_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif + +#include "archive_platform_acl.h" +#include "archive_acl_private.h" +#include "archive_entry.h" + +struct archive_write_disk; + +int archive_write_disk_set_acls(struct archive *, int, const char *, + struct archive_acl *, __LA_MODE_T); + +#endif diff --git a/c/archive_write_disk_set_standard_lookup.c b/c/archive_write_disk_set_standard_lookup.c new file mode 100644 --- /dev/null +++ b/c/archive_write_disk_set_standard_lookup.c @@ -0,0 +1,263 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_disk_set_standard_lookup.c 201083 2009-12-28 02:09:57Z kientzle $"); + +#ifdef HAVE_SYS_TYPES_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_GRP_H +#include +#endif +#ifdef HAVE_PWD_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_read_private.h" +#include "archive_write_disk_private.h" + +struct bucket { + char *name; + int hash; + id_t id; +}; + +static const size_t cache_size = 127; +static unsigned int hash(const char *); +static int64_t lookup_gid(void *, const char *uname, int64_t); +static int64_t lookup_uid(void *, const char *uname, int64_t); +static void cleanup(void *); + +/* + * Installs functions that use getpwnam()/getgrnam()---along with + * a simple cache to accelerate such lookups---into the archive_write_disk + * object. This is in a separate file because getpwnam()/getgrnam() + * can pull in a LOT of library code (including NIS/LDAP functions, which + * pull in DNS resolvers, etc). This can easily top 500kB, which makes + * it inappropriate for some space-constrained applications. + * + * Applications that are size-sensitive may want to just use the + * real default functions (defined in archive_write_disk.c) that just + * use the uid/gid without the lookup. Or define your own custom functions + * if you prefer. + * + * TODO: Replace these hash tables with simpler move-to-front LRU + * lists with a bounded size (128 items?). The hash is a bit faster, + * but has a bad pathology in which it thrashes a single bucket. Even + * walking a list of 128 items is a lot faster than calling + * getpwnam()! + */ +int +archive_write_disk_set_standard_lookup(struct archive *a) +{ + struct bucket *ucache = calloc(cache_size, sizeof(struct bucket)); + struct bucket *gcache = calloc(cache_size, sizeof(struct bucket)); + if (ucache == NULL || gcache == NULL) { + free(ucache); + free(gcache); + return (ARCHIVE_FATAL); + } + archive_write_disk_set_group_lookup(a, gcache, lookup_gid, cleanup); + archive_write_disk_set_user_lookup(a, ucache, lookup_uid, cleanup); + return (ARCHIVE_OK); +} + +static int64_t +lookup_gid(void *private_data, const char *gname, int64_t gid) +{ + int h; + struct bucket *b; + struct bucket *gcache = (struct bucket *)private_data; + + /* If no gname, just use the gid provided. */ + if (gname == NULL || *gname == '\0') + return (gid); + + /* Try to find gname in the cache. */ + h = hash(gname); + b = &gcache[h % cache_size ]; + if (b->name != NULL && b->hash == h && strcmp(gname, b->name) == 0) + return ((gid_t)b->id); + + /* Free the cache slot for a new entry. */ + free(b->name); + b->name = strdup(gname); + /* Note: If strdup fails, that's okay; we just won't cache. */ + b->hash = h; +#if HAVE_GRP_H +# if HAVE_GETGRNAM_R + { + char _buffer[128]; + size_t bufsize = 128; + char *buffer = _buffer; + char *allocated = NULL; + struct group grent, *result; + int r; + + for (;;) { + result = &grent; /* Old getgrnam_r ignores last arg. */ + r = getgrnam_r(gname, &grent, buffer, bufsize, &result); + if (r == 0) + break; + if (r != ERANGE) + break; + bufsize *= 2; + free(allocated); + allocated = malloc(bufsize); + if (allocated == NULL) + break; + buffer = allocated; + } + if (result != NULL) + gid = result->gr_gid; + free(allocated); + } +# else /* HAVE_GETGRNAM_R */ + { + struct group *result; + + result = getgrnam(gname); + if (result != NULL) + gid = result->gr_gid; + } +# endif /* HAVE_GETGRNAM_R */ +#elif defined(_WIN32) && !defined(__CYGWIN__) + /* TODO: do a gname->gid lookup for Windows. */ +#else + #error No way to perform gid lookups on this platform +#endif + b->id = (gid_t)gid; + + return (gid); +} + +static int64_t +lookup_uid(void *private_data, const char *uname, int64_t uid) +{ + int h; + struct bucket *b; + struct bucket *ucache = (struct bucket *)private_data; + + /* If no uname, just use the uid provided. */ + if (uname == NULL || *uname == '\0') + return (uid); + + /* Try to find uname in the cache. */ + h = hash(uname); + b = &ucache[h % cache_size ]; + if (b->name != NULL && b->hash == h && strcmp(uname, b->name) == 0) + return ((uid_t)b->id); + + /* Free the cache slot for a new entry. */ + free(b->name); + b->name = strdup(uname); + /* Note: If strdup fails, that's okay; we just won't cache. */ + b->hash = h; +#if HAVE_PWD_H +# if HAVE_GETPWNAM_R + { + char _buffer[128]; + size_t bufsize = 128; + char *buffer = _buffer; + char *allocated = NULL; + struct passwd pwent, *result; + int r; + + for (;;) { + result = &pwent; /* Old getpwnam_r ignores last arg. */ + r = getpwnam_r(uname, &pwent, buffer, bufsize, &result); + if (r == 0) + break; + if (r != ERANGE) + break; + bufsize *= 2; + free(allocated); + allocated = malloc(bufsize); + if (allocated == NULL) + break; + buffer = allocated; + } + if (result != NULL) + uid = result->pw_uid; + free(allocated); + } +# else /* HAVE_GETPWNAM_R */ + { + struct passwd *result; + + result = getpwnam(uname); + if (result != NULL) + uid = result->pw_uid; + } +#endif /* HAVE_GETPWNAM_R */ +#elif defined(_WIN32) && !defined(__CYGWIN__) + /* TODO: do a uname->uid lookup for Windows. */ +#else + #error No way to look up uids on this platform +#endif + b->id = (uid_t)uid; + + return (uid); +} + +static void +cleanup(void *private) +{ + size_t i; + struct bucket *cache = (struct bucket *)private; + + for (i = 0; i < cache_size; i++) + free(cache[i].name); + free(cache); +} + + +static unsigned int +hash(const char *p) +{ + /* A 32-bit version of Peter Weinberger's (PJW) hash algorithm, + as used by ELF for hashing function names. */ + unsigned g, h = 0; + while (*p != '\0') { + h = (h << 4) + *p++; + if ((g = h & 0xF0000000) != 0) { + h ^= g >> 24; + h &= 0x0FFFFFFF; + } + } + return h; +} diff --git a/c/archive_write_disk_windows.c b/c/archive_write_disk_windows.c new file mode 100644 --- /dev/null +++ b/c/archive_write_disk_windows.c @@ -0,0 +1,2834 @@ +/*- + * Copyright (c) 2003-2010 Tim Kientzle + * Copyright (c) 2011-2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer + * in this position and unchanged. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#if defined(_WIN32) && !defined(__CYGWIN__) + +#ifdef HAVE_SYS_TYPES_H +#include +#endif +#ifdef HAVE_SYS_UTIME_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_FCNTL_H +#include +#endif +#ifdef HAVE_LIMITS_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#include + +/* TODO: Support Mac OS 'quarantine' feature. This is really just a + * standard tag to mark files that have been downloaded as "tainted". + * On Mac OS, we should mark the extracted files as tainted if the + * archive being read was tainted. Windows has a similar feature; we + * should investigate ways to support this generically. */ + +#include "archive.h" +#include "archive_acl_private.h" +#include "archive_string.h" +#include "archive_entry.h" +#include "archive_private.h" + +#ifndef O_BINARY +#define O_BINARY 0 +#endif +#ifndef IO_REPARSE_TAG_SYMLINK +/* Old SDKs do not provide IO_REPARSE_TAG_SYMLINK */ +#define IO_REPARSE_TAG_SYMLINK 0xA000000CL +#endif + +static BOOL SetFilePointerEx_perso(HANDLE hFile, + LARGE_INTEGER liDistanceToMove, + PLARGE_INTEGER lpNewFilePointer, + DWORD dwMoveMethod) +{ + LARGE_INTEGER li; + li.QuadPart = liDistanceToMove.QuadPart; + li.LowPart = SetFilePointer( + hFile, li.LowPart, &li.HighPart, dwMoveMethod); + if(lpNewFilePointer) { + lpNewFilePointer->QuadPart = li.QuadPart; + } + return li.LowPart != (DWORD)-1 || GetLastError() == NO_ERROR; +} + +struct fixup_entry { + struct fixup_entry *next; + struct archive_acl acl; + mode_t mode; + int64_t atime; + int64_t birthtime; + int64_t mtime; + int64_t ctime; + unsigned long atime_nanos; + unsigned long birthtime_nanos; + unsigned long mtime_nanos; + unsigned long ctime_nanos; + unsigned long fflags_set; + int fixup; /* bitmask of what needs fixing */ + wchar_t *name; +}; + +/* + * We use a bitmask to track which operations remain to be done for + * this file. In particular, this helps us avoid unnecessary + * operations when it's possible to take care of one step as a + * side-effect of another. For example, mkdir() can specify the mode + * for the newly-created object but symlink() cannot. This means we + * can skip chmod() if mkdir() succeeded, but we must explicitly + * chmod() if we're trying to create a directory that already exists + * (mkdir() failed) or if we're restoring a symlink. Similarly, we + * need to verify UID/GID before trying to restore SUID/SGID bits; + * that verification can occur explicitly through a stat() call or + * implicitly because of a successful chown() call. + */ +#define TODO_MODE_FORCE 0x40000000 +#define TODO_MODE_BASE 0x20000000 +#define TODO_SUID 0x10000000 +#define TODO_SUID_CHECK 0x08000000 +#define TODO_SGID 0x04000000 +#define TODO_SGID_CHECK 0x02000000 +#define TODO_MODE (TODO_MODE_BASE|TODO_SUID|TODO_SGID) +#define TODO_TIMES ARCHIVE_EXTRACT_TIME +#define TODO_OWNER ARCHIVE_EXTRACT_OWNER +#define TODO_FFLAGS ARCHIVE_EXTRACT_FFLAGS +#define TODO_ACLS ARCHIVE_EXTRACT_ACL +#define TODO_XATTR ARCHIVE_EXTRACT_XATTR +#define TODO_MAC_METADATA ARCHIVE_EXTRACT_MAC_METADATA + +struct archive_write_disk { + struct archive archive; + + mode_t user_umask; + struct fixup_entry *fixup_list; + struct fixup_entry *current_fixup; + int64_t user_uid; + int skip_file_set; + int64_t skip_file_dev; + int64_t skip_file_ino; + time_t start_time; + + int64_t (*lookup_gid)(void *private, const char *gname, int64_t gid); + void (*cleanup_gid)(void *private); + void *lookup_gid_data; + int64_t (*lookup_uid)(void *private, const char *uname, int64_t uid); + void (*cleanup_uid)(void *private); + void *lookup_uid_data; + + /* + * Full path of last file to satisfy symlink checks. + */ + struct archive_wstring path_safe; + + /* + * Cached stat data from disk for the current entry. + * If this is valid, pst points to st. Otherwise, + * pst is null. + */ + BY_HANDLE_FILE_INFORMATION st; + BY_HANDLE_FILE_INFORMATION *pst; + + /* Information about the object being restored right now. */ + struct archive_entry *entry; /* Entry being extracted. */ + wchar_t *name; /* Name of entry, possibly edited. */ + struct archive_wstring _name_data; /* backing store for 'name' */ + wchar_t *tmpname; /* Temporary name */ + struct archive_wstring _tmpname_data; /* backing store for 'tmpname' */ + /* Tasks remaining for this object. */ + int todo; + /* Tasks deferred until end-of-archive. */ + int deferred; + /* Options requested by the client. */ + int flags; + /* Handle for the file we're restoring. */ + HANDLE fh; + /* Current offset for writing data to the file. */ + int64_t offset; + /* Last offset actually written to disk. */ + int64_t fd_offset; + /* Total bytes actually written to files. */ + int64_t total_bytes_written; + /* Maximum size of file, -1 if unknown. */ + int64_t filesize; + /* Dir we were in before this restore; only for deep paths. */ + int restore_pwd; + /* Mode we should use for this entry; affected by _PERM and umask. */ + mode_t mode; + /* UID/GID to use in restoring this entry. */ + int64_t uid; + int64_t gid; +}; + +/* + * Default mode for dirs created automatically (will be modified by umask). + * Note that POSIX specifies 0777 for implicitly-created dirs, "modified + * by the process' file creation mask." + */ +#define DEFAULT_DIR_MODE 0777 +/* + * Dir modes are restored in two steps: During the extraction, the permissions + * in the archive are modified to match the following limits. During + * the post-extract fixup pass, the permissions from the archive are + * applied. + */ +#define MINIMUM_DIR_MODE 0700 +#define MAXIMUM_DIR_MODE 0775 + +static int disk_unlink(const wchar_t *); +static int disk_rmdir(const wchar_t *); +static int check_symlinks(struct archive_write_disk *); +static int create_filesystem_object(struct archive_write_disk *); +static struct fixup_entry *current_fixup(struct archive_write_disk *, + const wchar_t *pathname); +static int cleanup_pathname(struct archive_write_disk *); +static int create_dir(struct archive_write_disk *, wchar_t *); +static int create_parent_dir(struct archive_write_disk *, wchar_t *); +static int la_chmod(const wchar_t *, mode_t); +static int la_mktemp(struct archive_write_disk *); +static int older(BY_HANDLE_FILE_INFORMATION *, struct archive_entry *); +static int permissive_name_w(struct archive_write_disk *); +static int restore_entry(struct archive_write_disk *); +static int set_acls(struct archive_write_disk *, HANDLE h, + const wchar_t *, struct archive_acl *); +static int set_xattrs(struct archive_write_disk *); +static int clear_nochange_fflags(struct archive_write_disk *); +static int set_fflags(struct archive_write_disk *); +static int set_fflags_platform(const wchar_t *, unsigned long, + unsigned long); +static int set_ownership(struct archive_write_disk *); +static int set_mode(struct archive_write_disk *, int mode); +static int set_times(struct archive_write_disk *, HANDLE, int, + const wchar_t *, time_t, long, time_t, long, time_t, + long, time_t, long); +static int set_times_from_entry(struct archive_write_disk *); +static struct fixup_entry *sort_dir_list(struct fixup_entry *p); +static ssize_t write_data_block(struct archive_write_disk *, + const char *, size_t); + +static struct archive_vtable *archive_write_disk_vtable(void); + +static int _archive_write_disk_close(struct archive *); +static int _archive_write_disk_free(struct archive *); +static int _archive_write_disk_header(struct archive *, + struct archive_entry *); +static int64_t _archive_write_disk_filter_bytes(struct archive *, int); +static int _archive_write_disk_finish_entry(struct archive *); +static ssize_t _archive_write_disk_data(struct archive *, const void *, + size_t); +static ssize_t _archive_write_disk_data_block(struct archive *, const void *, + size_t, int64_t); + +#define bhfi_dev(bhfi) ((bhfi)->dwVolumeSerialNumber) +/* Treat FileIndex as i-node. We should remove a sequence number + * which is high-16-bits of nFileIndexHigh. */ +#define bhfi_ino(bhfi) \ + ((((int64_t)((bhfi)->nFileIndexHigh & 0x0000FFFFUL)) << 32) \ + + (bhfi)->nFileIndexLow) +#define bhfi_size(bhfi) \ + ((((int64_t)(bhfi)->nFileSizeHigh) << 32) + (bhfi)->nFileSizeLow) + +static int +file_information(struct archive_write_disk *a, wchar_t *path, + BY_HANDLE_FILE_INFORMATION *st, mode_t *mode, int sim_lstat) +{ + HANDLE h; + int r; + DWORD flag = FILE_FLAG_BACKUP_SEMANTICS; + WIN32_FIND_DATAW findData; + + if (sim_lstat || mode != NULL) { + h = FindFirstFileW(path, &findData); + if (h == INVALID_HANDLE_VALUE && + GetLastError() == ERROR_INVALID_NAME) { + wchar_t *full; + full = __la_win_permissive_name_w(path); + h = FindFirstFileW(full, &findData); + free(full); + } + if (h == INVALID_HANDLE_VALUE) { + la_dosmaperr(GetLastError()); + return (-1); + } + FindClose(h); + } + + /* Is symlink file ? */ + if (sim_lstat && + ((findData.dwFileAttributes + & FILE_ATTRIBUTE_REPARSE_POINT) && + (findData.dwReserved0 == IO_REPARSE_TAG_SYMLINK))) + flag |= FILE_FLAG_OPEN_REPARSE_POINT; + + h = CreateFileW(a->name, 0, 0, NULL, + OPEN_EXISTING, flag, NULL); + if (h == INVALID_HANDLE_VALUE && + GetLastError() == ERROR_INVALID_NAME) { + wchar_t *full; + full = __la_win_permissive_name_w(path); + h = CreateFileW(full, 0, 0, NULL, + OPEN_EXISTING, flag, NULL); + free(full); + } + if (h == INVALID_HANDLE_VALUE) { + la_dosmaperr(GetLastError()); + return (-1); + } + r = GetFileInformationByHandle(h, st); + CloseHandle(h); + if (r == 0) { + la_dosmaperr(GetLastError()); + return (-1); + } + + if (mode == NULL) + return (0); + + *mode = S_IRUSR | S_IRGRP | S_IROTH; + if ((st->dwFileAttributes & FILE_ATTRIBUTE_READONLY) == 0) + *mode |= S_IWUSR | S_IWGRP | S_IWOTH; + if ((st->dwFileAttributes & FILE_ATTRIBUTE_REPARSE_POINT) && + findData.dwReserved0 == IO_REPARSE_TAG_SYMLINK) + *mode |= S_IFLNK; + else if (st->dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY) + *mode |= S_IFDIR | S_IXUSR | S_IXGRP | S_IXOTH; + else { + const wchar_t *p; + + *mode |= S_IFREG; + p = wcsrchr(path, L'.'); + if (p != NULL && wcslen(p) == 4) { + switch (p[1]) { + case L'B': case L'b': + if ((p[2] == L'A' || p[2] == L'a' ) && + (p[3] == L'T' || p[3] == L't' )) + *mode |= S_IXUSR | S_IXGRP | S_IXOTH; + break; + case L'C': case L'c': + if (((p[2] == L'M' || p[2] == L'm' ) && + (p[3] == L'D' || p[3] == L'd' ))) + *mode |= S_IXUSR | S_IXGRP | S_IXOTH; + break; + case L'E': case L'e': + if ((p[2] == L'X' || p[2] == L'x' ) && + (p[3] == L'E' || p[3] == L'e' )) + *mode |= S_IXUSR | S_IXGRP | S_IXOTH; + break; + default: + break; + } + } + } + return (0); +} + +/* + * Note: The path, for example, "aa/a/../b../c" will be converted to "aa/c" + * by GetFullPathNameW() W32 API, which __la_win_permissive_name_w uses. + * It means we cannot handle multiple dirs in one archive_entry. + * So we have to make the full-pathname in another way, which does not + * break "../" path string. + */ +static int +permissive_name_w(struct archive_write_disk *a) +{ + wchar_t *wn, *wnp; + wchar_t *ws, *wsp; + DWORD l; + + wnp = a->name; + if (wnp[0] == L'\\' && wnp[1] == L'\\' && + wnp[2] == L'?' && wnp[3] == L'\\') + /* We have already a permissive name. */ + return (0); + + if (wnp[0] == L'\\' && wnp[1] == L'\\' && + wnp[2] == L'.' && wnp[3] == L'\\') { + /* This is a device name */ + if (((wnp[4] >= L'a' && wnp[4] <= L'z') || + (wnp[4] >= L'A' && wnp[4] <= L'Z')) && + wnp[5] == L':' && wnp[6] == L'\\') { + wnp[2] = L'?';/* Not device name. */ + return (0); + } + } + + /* + * A full-pathname starting with a drive name like "C:\abc". + */ + if (((wnp[0] >= L'a' && wnp[0] <= L'z') || + (wnp[0] >= L'A' && wnp[0] <= L'Z')) && + wnp[1] == L':' && wnp[2] == L'\\') { + wn = _wcsdup(wnp); + if (wn == NULL) + return (-1); + archive_wstring_ensure(&(a->_name_data), 4 + wcslen(wn) + 1); + a->name = a->_name_data.s; + /* Prepend "\\?\" */ + archive_wstrncpy(&(a->_name_data), L"\\\\?\\", 4); + archive_wstrcat(&(a->_name_data), wn); + free(wn); + return (0); + } + + /* + * A full-pathname pointing to a network drive + * like "\\\\file". + */ + if (wnp[0] == L'\\' && wnp[1] == L'\\' && wnp[2] != L'\\') { + const wchar_t *p = &wnp[2]; + + /* Skip server-name letters. */ + while (*p != L'\\' && *p != L'\0') + ++p; + if (*p == L'\\') { + const wchar_t *rp = ++p; + /* Skip share-name letters. */ + while (*p != L'\\' && *p != L'\0') + ++p; + if (*p == L'\\' && p != rp) { + /* Now, match patterns such as + * "\\server-name\share-name\" */ + wn = _wcsdup(wnp); + if (wn == NULL) + return (-1); + archive_wstring_ensure(&(a->_name_data), + 8 + wcslen(wn) + 1); + a->name = a->_name_data.s; + /* Prepend "\\?\UNC\" */ + archive_wstrncpy(&(a->_name_data), + L"\\\\?\\UNC\\", 8); + archive_wstrcat(&(a->_name_data), wn+2); + free(wn); + return (0); + } + } + return (0); + } + + /* + * Get current working directory. + */ + l = GetCurrentDirectoryW(0, NULL); + if (l == 0) + return (-1); + ws = malloc(l * sizeof(wchar_t)); + l = GetCurrentDirectoryW(l, ws); + if (l == 0) { + free(ws); + return (-1); + } + wsp = ws; + + /* + * A full-pathname starting without a drive name like "\abc". + */ + if (wnp[0] == L'\\') { + wn = _wcsdup(wnp); + if (wn == NULL) + return (-1); + archive_wstring_ensure(&(a->_name_data), + 4 + 2 + wcslen(wn) + 1); + a->name = a->_name_data.s; + /* Prepend "\\?\" and drive name. */ + archive_wstrncpy(&(a->_name_data), L"\\\\?\\", 4); + archive_wstrncat(&(a->_name_data), wsp, 2); + archive_wstrcat(&(a->_name_data), wn); + free(wsp); + free(wn); + return (0); + } + + wn = _wcsdup(wnp); + if (wn == NULL) + return (-1); + archive_wstring_ensure(&(a->_name_data), 4 + l + 1 + wcslen(wn) + 1); + a->name = a->_name_data.s; + /* Prepend "\\?\" and drive name if not already added. */ + if (l > 3 && wsp[0] == L'\\' && wsp[1] == L'\\' && + wsp[2] == L'?' && wsp[3] == L'\\') + { + archive_wstrncpy(&(a->_name_data), wsp, l); + } + else if (l > 2 && wsp[0] == L'\\' && wsp[1] == L'\\' && wsp[2] != L'\\') + { + archive_wstrncpy(&(a->_name_data), L"\\\\?\\UNC\\", 8); + archive_wstrncat(&(a->_name_data), wsp+2, l-2); + } + else + { + archive_wstrncpy(&(a->_name_data), L"\\\\?\\", 4); + archive_wstrncat(&(a->_name_data), wsp, l); + } + archive_wstrncat(&(a->_name_data), L"\\", 1); + archive_wstrcat(&(a->_name_data), wn); + a->name = a->_name_data.s; + free(wsp); + free(wn); + return (0); +} + +static int +la_chmod(const wchar_t *path, mode_t mode) +{ + DWORD attr; + BOOL r; + wchar_t *fullname; + int ret = 0; + + fullname = NULL; + attr = GetFileAttributesW(path); + if (attr == (DWORD)-1 && + GetLastError() == ERROR_INVALID_NAME) { + fullname = __la_win_permissive_name_w(path); + attr = GetFileAttributesW(fullname); + } + if (attr == (DWORD)-1) { + la_dosmaperr(GetLastError()); + ret = -1; + goto exit_chmode; + } + if (mode & _S_IWRITE) + attr &= ~FILE_ATTRIBUTE_READONLY; + else + attr |= FILE_ATTRIBUTE_READONLY; + if (fullname != NULL) + r = SetFileAttributesW(fullname, attr); + else + r = SetFileAttributesW(path, attr); + if (r == 0) { + la_dosmaperr(GetLastError()); + ret = -1; + } +exit_chmode: + free(fullname); + return (ret); +} + +static int +la_mktemp(struct archive_write_disk *a) +{ + int fd; + mode_t mode; + + archive_wstring_empty(&(a->_tmpname_data)); + archive_wstrcpy(&(a->_tmpname_data), a->name); + archive_wstrcat(&(a->_tmpname_data), L".XXXXXX"); + a->tmpname = a->_tmpname_data.s; + + fd = __archive_mkstemp(a->tmpname); + if (fd == -1) + return -1; + + mode = a->mode & 0777 & ~a->user_umask; + if (la_chmod(a->tmpname, mode) == -1) { + la_dosmaperr(GetLastError()); + _close(fd); + return -1; + } + return (fd); +} + +static void * +la_GetFunctionKernel32(const char *name) +{ + static HINSTANCE lib; + static int set; + if (!set) { + set = 1; + lib = LoadLibrary(TEXT("kernel32.dll")); + } + if (lib == NULL) { + fprintf(stderr, "Can't load kernel32.dll?!\n"); + exit(1); + } + return (void *)GetProcAddress(lib, name); +} + +static int +la_CreateHardLinkW(wchar_t *linkname, wchar_t *target) +{ + static BOOLEAN (WINAPI *f)(LPWSTR, LPWSTR, LPSECURITY_ATTRIBUTES); + static int set; + BOOL ret; + + if (!set) { + set = 1; + f = la_GetFunctionKernel32("CreateHardLinkW"); + } + if (!f) { + errno = ENOTSUP; + return (0); + } + ret = (*f)(linkname, target, NULL); + if (!ret) { + /* Under windows 2000, it is necessary to remove + * the "\\?\" prefix. */ +#define IS_UNC(name) ((name[0] == L'U' || name[0] == L'u') && \ + (name[1] == L'N' || name[1] == L'n') && \ + (name[2] == L'C' || name[2] == L'c') && \ + name[3] == L'\\') + if (!wcsncmp(linkname,L"\\\\?\\", 4)) { + linkname += 4; + if (IS_UNC(linkname)) + linkname += 4; + } + if (!wcsncmp(target,L"\\\\?\\", 4)) { + target += 4; + if (IS_UNC(target)) + target += 4; + } +#undef IS_UNC + ret = (*f)(linkname, target, NULL); + } + return (ret); +} + +/* + * Create file or directory symolic link + * + * If linktype is AE_SYMLINK_TYPE_UNDEFINED (or unknown), guess linktype from + * the link target + */ +static int +la_CreateSymbolicLinkW(const wchar_t *linkname, const wchar_t *target, + int linktype) { + static BOOLEAN (WINAPI *f)(LPCWSTR, LPCWSTR, DWORD); + static int set; + wchar_t *ttarget, *p; + int len; + DWORD attrs = 0; + DWORD flags = 0; + DWORD newflags = 0; + BOOL ret = 0; + + if (!set) { + set = 1; + f = la_GetFunctionKernel32("CreateSymbolicLinkW"); + } + if (!f) + return (0); + + len = wcslen(target); + if (len == 0) { + errno = EINVAL; + return(0); + } + /* + * When writing path targets, we need to translate slashes + * to backslashes + */ + ttarget = malloc((len + 1) * sizeof(wchar_t)); + if (ttarget == NULL) + return(0); + + p = ttarget; + + while(*target != L'\0') { + if (*target == L'/') + *p = L'\\'; + else + *p = *target; + target++; + p++; + } + *p = L'\0'; + + /* + * In case of undefined symlink type we guess it from the target. + * If the target equals ".", "..", ends with a backslash or a + * backslash followed by "." or ".." we assume it is a directory + * symlink. In all other cases we assume a file symlink. + */ + if (linktype != AE_SYMLINK_TYPE_FILE && ( + linktype == AE_SYMLINK_TYPE_DIRECTORY || + *(p - 1) == L'\\' || (*(p - 1) == L'.' && ( + len == 1 || *(p - 2) == L'\\' || ( *(p - 2) == L'.' && ( + len == 2 || *(p - 3) == L'\\')))))) { +#if defined(SYMBOLIC_LINK_FLAG_DIRECTORY) + flags |= SYMBOLIC_LINK_FLAG_DIRECTORY; +#else + flags |= 0x1; +#endif + } + +#if defined(SYMBOLIC_LINK_FLAG_ALLOW_UNPRIVILEGED_CREATE) + newflags = flags | SYMBOLIC_LINK_FLAG_ALLOW_UNPRIVILEGED_CREATE; +#else + newflags = flags | 0x2; +#endif + + /* + * Windows won't overwrite existing links + */ + attrs = GetFileAttributesW(linkname); + if (attrs != INVALID_FILE_ATTRIBUTES) { + if (attrs & FILE_ATTRIBUTE_DIRECTORY) + disk_rmdir(linkname); + else + disk_unlink(linkname); + } + + ret = (*f)(linkname, ttarget, newflags); + /* + * Prior to Windows 10 calling CreateSymbolicLinkW() will fail + * if SYMBOLIC_LINK_FLAG_ALLOW_UNPRIVILEGED_CREATE is set + */ + if (!ret) { + ret = (*f)(linkname, ttarget, flags); + } + free(ttarget); + return (ret); +} + +static int +la_ftruncate(HANDLE handle, int64_t length) +{ + LARGE_INTEGER distance; + + if (GetFileType(handle) != FILE_TYPE_DISK) { + errno = EBADF; + return (-1); + } + distance.QuadPart = length; + if (!SetFilePointerEx_perso(handle, distance, NULL, FILE_BEGIN)) { + la_dosmaperr(GetLastError()); + return (-1); + } + if (!SetEndOfFile(handle)) { + la_dosmaperr(GetLastError()); + return (-1); + } + return (0); +} + +static int +lazy_stat(struct archive_write_disk *a) +{ + if (a->pst != NULL) { + /* Already have stat() data available. */ + return (ARCHIVE_OK); + } + if (a->fh != INVALID_HANDLE_VALUE && + GetFileInformationByHandle(a->fh, &a->st) == 0) { + a->pst = &a->st; + return (ARCHIVE_OK); + } + + /* + * XXX At this point, symlinks should not be hit, otherwise + * XXX a race occurred. Do we want to check explicitly for that? + */ + if (file_information(a, a->name, &a->st, NULL, 1) == 0) { + a->pst = &a->st; + return (ARCHIVE_OK); + } + archive_set_error(&a->archive, errno, "Couldn't stat file"); + return (ARCHIVE_WARN); +} + +static struct archive_vtable * +archive_write_disk_vtable(void) +{ + static struct archive_vtable av; + static int inited = 0; + + if (!inited) { + av.archive_close = _archive_write_disk_close; + av.archive_filter_bytes = _archive_write_disk_filter_bytes; + av.archive_free = _archive_write_disk_free; + av.archive_write_header = _archive_write_disk_header; + av.archive_write_finish_entry + = _archive_write_disk_finish_entry; + av.archive_write_data = _archive_write_disk_data; + av.archive_write_data_block = _archive_write_disk_data_block; + inited = 1; + } + return (&av); +} + +static int64_t +_archive_write_disk_filter_bytes(struct archive *_a, int n) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + (void)n; /* UNUSED */ + if (n == -1 || n == 0) + return (a->total_bytes_written); + return (-1); +} + + +int +archive_write_disk_set_options(struct archive *_a, int flags) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + + a->flags = flags; + return (ARCHIVE_OK); +} + + +/* + * Extract this entry to disk. + * + * TODO: Validate hardlinks. According to the standards, we're + * supposed to check each extracted hardlink and squawk if it refers + * to a file that we didn't restore. I'm not entirely convinced this + * is a good idea, but more importantly: Is there any way to validate + * hardlinks without keeping a complete list of filenames from the + * entire archive?? Ugh. + * + */ +static int +_archive_write_disk_header(struct archive *_a, struct archive_entry *entry) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + struct fixup_entry *fe; + int ret, r; + + archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, + "archive_write_disk_header"); + archive_clear_error(&a->archive); + if (a->archive.state & ARCHIVE_STATE_DATA) { + r = _archive_write_disk_finish_entry(&a->archive); + if (r == ARCHIVE_FATAL) + return (r); + } + + /* Set up for this particular entry. */ + a->pst = NULL; + a->current_fixup = NULL; + a->deferred = 0; + archive_entry_free(a->entry); + a->entry = NULL; + a->entry = archive_entry_clone(entry); + a->fh = INVALID_HANDLE_VALUE; + a->fd_offset = 0; + a->offset = 0; + a->restore_pwd = -1; + a->uid = a->user_uid; + a->mode = archive_entry_mode(a->entry); + if (archive_entry_size_is_set(a->entry)) + a->filesize = archive_entry_size(a->entry); + else + a->filesize = -1; + archive_wstrcpy(&(a->_name_data), archive_entry_pathname_w(a->entry)); + a->name = a->_name_data.s; + archive_clear_error(&a->archive); + + /* + * Clean up the requested path. This is necessary for correct + * dir restores; the dir restore logic otherwise gets messed + * up by nonsense like "dir/.". + */ + ret = cleanup_pathname(a); + if (ret != ARCHIVE_OK) + return (ret); + + /* + * Generate a full-pathname and use it from here. + */ + if (permissive_name_w(a) < 0) { + errno = EINVAL; + return (ARCHIVE_FAILED); + } + + /* + * Query the umask so we get predictable mode settings. + * This gets done on every call to _write_header in case the + * user edits their umask during the extraction for some + * reason. + */ + umask(a->user_umask = umask(0)); + + /* Figure out what we need to do for this entry. */ + a->todo = TODO_MODE_BASE; + if (a->flags & ARCHIVE_EXTRACT_PERM) { + a->todo |= TODO_MODE_FORCE; /* Be pushy about permissions. */ + /* + * SGID requires an extra "check" step because we + * cannot easily predict the GID that the system will + * assign. (Different systems assign GIDs to files + * based on a variety of criteria, including process + * credentials and the gid of the enclosing + * directory.) We can only restore the SGID bit if + * the file has the right GID, and we only know the + * GID if we either set it (see set_ownership) or if + * we've actually called stat() on the file after it + * was restored. Since there are several places at + * which we might verify the GID, we need a TODO bit + * to keep track. + */ + if (a->mode & S_ISGID) + a->todo |= TODO_SGID | TODO_SGID_CHECK; + /* + * Verifying the SUID is simpler, but can still be + * done in multiple ways, hence the separate "check" bit. + */ + if (a->mode & S_ISUID) + a->todo |= TODO_SUID | TODO_SUID_CHECK; + } else { + /* + * User didn't request full permissions, so don't + * restore SUID, SGID bits and obey umask. + */ + a->mode &= ~S_ISUID; + a->mode &= ~S_ISGID; + a->mode &= ~S_ISVTX; + a->mode &= ~a->user_umask; + } +#if 0 + if (a->flags & ARCHIVE_EXTRACT_OWNER) + a->todo |= TODO_OWNER; +#endif + if (a->flags & ARCHIVE_EXTRACT_TIME) + a->todo |= TODO_TIMES; + if (a->flags & ARCHIVE_EXTRACT_ACL) { + if (archive_entry_filetype(a->entry) == AE_IFDIR) + a->deferred |= TODO_ACLS; + else + a->todo |= TODO_ACLS; + } + if (a->flags & ARCHIVE_EXTRACT_XATTR) + a->todo |= TODO_XATTR; + if (a->flags & ARCHIVE_EXTRACT_FFLAGS) + a->todo |= TODO_FFLAGS; + if (a->flags & ARCHIVE_EXTRACT_SECURE_SYMLINKS) { + ret = check_symlinks(a); + if (ret != ARCHIVE_OK) + return (ret); + } + + ret = restore_entry(a); + + /* + * TODO: There are rumours that some extended attributes must + * be restored before file data is written. If this is true, + * then we either need to write all extended attributes both + * before and after restoring the data, or find some rule for + * determining which must go first and which last. Due to the + * many ways people are using xattrs, this may prove to be an + * intractable problem. + */ + + /* + * Fixup uses the unedited pathname from archive_entry_pathname(), + * because it is relative to the base dir and the edited path + * might be relative to some intermediate dir as a result of the + * deep restore logic. + */ + if (a->deferred & TODO_MODE) { + fe = current_fixup(a, archive_entry_pathname_w(entry)); + fe->fixup |= TODO_MODE_BASE; + fe->mode = a->mode; + } + + if ((a->deferred & TODO_TIMES) + && (archive_entry_mtime_is_set(entry) + || archive_entry_atime_is_set(entry))) { + fe = current_fixup(a, archive_entry_pathname_w(entry)); + fe->mode = a->mode; + fe->fixup |= TODO_TIMES; + if (archive_entry_atime_is_set(entry)) { + fe->atime = archive_entry_atime(entry); + fe->atime_nanos = archive_entry_atime_nsec(entry); + } else { + /* If atime is unset, use start time. */ + fe->atime = a->start_time; + fe->atime_nanos = 0; + } + if (archive_entry_mtime_is_set(entry)) { + fe->mtime = archive_entry_mtime(entry); + fe->mtime_nanos = archive_entry_mtime_nsec(entry); + } else { + /* If mtime is unset, use start time. */ + fe->mtime = a->start_time; + fe->mtime_nanos = 0; + } + if (archive_entry_birthtime_is_set(entry)) { + fe->birthtime = archive_entry_birthtime(entry); + fe->birthtime_nanos = archive_entry_birthtime_nsec(entry); + } else { + /* If birthtime is unset, use mtime. */ + fe->birthtime = fe->mtime; + fe->birthtime_nanos = fe->mtime_nanos; + } + } + + if (a->deferred & TODO_ACLS) { + fe = current_fixup(a, archive_entry_pathname_w(entry)); + archive_acl_copy(&fe->acl, archive_entry_acl(entry)); + } + + if (a->deferred & TODO_FFLAGS) { + unsigned long set, clear; + + fe = current_fixup(a, archive_entry_pathname_w(entry)); + archive_entry_fflags(entry, &set, &clear); + fe->fflags_set = set; + } + + /* + * On Windows, A creating sparse file requires a special mark. + */ + if (a->fh != INVALID_HANDLE_VALUE && + archive_entry_sparse_count(entry) > 0) { + int64_t base = 0, offset, length; + int i, cnt = archive_entry_sparse_reset(entry); + int sparse = 0; + + for (i = 0; i < cnt; i++) { + archive_entry_sparse_next(entry, &offset, &length); + if (offset - base >= 4096) { + sparse = 1;/* we have a hole. */ + break; + } + base = offset + length; + } + if (sparse) { + DWORD dmy; + /* Mark this file as sparse. */ + DeviceIoControl(a->fh, FSCTL_SET_SPARSE, + NULL, 0, NULL, 0, &dmy, NULL); + } + } + + /* We've created the object and are ready to pour data into it. */ + if (ret >= ARCHIVE_WARN) + a->archive.state = ARCHIVE_STATE_DATA; + /* + * If it's not open, tell our client not to try writing. + * In particular, dirs, links, etc, don't get written to. + */ + if (a->fh == INVALID_HANDLE_VALUE) { + archive_entry_set_size(entry, 0); + a->filesize = 0; + } + + return (ret); +} + +int +archive_write_disk_set_skip_file(struct archive *_a, la_int64_t d, la_int64_t i) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_write_disk_set_skip_file"); + a->skip_file_set = 1; + a->skip_file_dev = d; + a->skip_file_ino = i; + return (ARCHIVE_OK); +} + +static ssize_t +write_data_block(struct archive_write_disk *a, const char *buff, size_t size) +{ + OVERLAPPED ol; + uint64_t start_size = size; + DWORD bytes_written = 0; + ssize_t block_size = 0, bytes_to_write; + + if (size == 0) + return (ARCHIVE_OK); + + if (a->filesize == 0 || a->fh == INVALID_HANDLE_VALUE) { + archive_set_error(&a->archive, 0, + "Attempt to write to an empty file"); + return (ARCHIVE_WARN); + } + + if (a->flags & ARCHIVE_EXTRACT_SPARSE) { + /* XXX TODO XXX Is there a more appropriate choice here ? */ + /* This needn't match the filesystem allocation size. */ + block_size = 16*1024; + } + + /* If this write would run beyond the file size, truncate it. */ + if (a->filesize >= 0 && (int64_t)(a->offset + size) > a->filesize) + start_size = size = (size_t)(a->filesize - a->offset); + + /* Write the data. */ + while (size > 0) { + if (block_size == 0) { + bytes_to_write = size; + } else { + /* We're sparsifying the file. */ + const char *p, *end; + int64_t block_end; + + /* Skip leading zero bytes. */ + for (p = buff, end = buff + size; p < end; ++p) { + if (*p != '\0') + break; + } + a->offset += p - buff; + size -= p - buff; + buff = p; + if (size == 0) + break; + + /* Calculate next block boundary after offset. */ + block_end + = (a->offset / block_size + 1) * block_size; + + /* If the adjusted write would cross block boundary, + * truncate it to the block boundary. */ + bytes_to_write = size; + if (a->offset + bytes_to_write > block_end) + bytes_to_write = (DWORD)(block_end - a->offset); + } + memset(&ol, 0, sizeof(ol)); + ol.Offset = (DWORD)(a->offset & 0xFFFFFFFF); + ol.OffsetHigh = (DWORD)(a->offset >> 32); + if (!WriteFile(a->fh, buff, (uint32_t)bytes_to_write, + &bytes_written, &ol)) { + DWORD lasterr; + + lasterr = GetLastError(); + if (lasterr == ERROR_ACCESS_DENIED) + errno = EBADF; + else + la_dosmaperr(lasterr); + archive_set_error(&a->archive, errno, "Write failed"); + return (ARCHIVE_WARN); + } + buff += bytes_written; + size -= bytes_written; + a->total_bytes_written += bytes_written; + a->offset += bytes_written; + a->fd_offset = a->offset; + } + return ((ssize_t)(start_size - size)); +} + +static ssize_t +_archive_write_disk_data_block(struct archive *_a, + const void *buff, size_t size, int64_t offset) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + ssize_t r; + + archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_DATA, "archive_write_data_block"); + + a->offset = offset; + r = write_data_block(a, buff, size); + if (r < ARCHIVE_OK) + return (r); + if ((size_t)r < size) { + archive_set_error(&a->archive, 0, + "Write request too large"); + return (ARCHIVE_WARN); + } +#if ARCHIVE_VERSION_NUMBER < 3999000 + return (ARCHIVE_OK); +#else + return (size); +#endif +} + +static ssize_t +_archive_write_disk_data(struct archive *_a, const void *buff, size_t size) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + + archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_DATA, "archive_write_data"); + + return (write_data_block(a, buff, size)); +} + +static int +_archive_write_disk_finish_entry(struct archive *_a) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + int ret = ARCHIVE_OK; + + archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, + "archive_write_finish_entry"); + if (a->archive.state & ARCHIVE_STATE_HEADER) + return (ARCHIVE_OK); + archive_clear_error(&a->archive); + + /* Pad or truncate file to the right size. */ + if (a->fh == INVALID_HANDLE_VALUE) { + /* There's no file. */ + } else if (a->filesize < 0) { + /* File size is unknown, so we can't set the size. */ + } else if (a->fd_offset == a->filesize) { + /* Last write ended at exactly the filesize; we're done. */ + /* Hopefully, this is the common case. */ + } else { + if (la_ftruncate(a->fh, a->filesize) == -1) { + archive_set_error(&a->archive, errno, + "File size could not be restored"); + return (ARCHIVE_FAILED); + } + } + + /* Restore metadata. */ + + /* + * Look up the "real" UID only if we're going to need it. + * TODO: the TODO_SGID condition can be dropped here, can't it? + */ + if (a->todo & (TODO_OWNER | TODO_SUID | TODO_SGID)) { + a->uid = archive_write_disk_uid(&a->archive, + archive_entry_uname(a->entry), + archive_entry_uid(a->entry)); + } + /* Look up the "real" GID only if we're going to need it. */ + /* TODO: the TODO_SUID condition can be dropped here, can't it? */ + if (a->todo & (TODO_OWNER | TODO_SGID | TODO_SUID)) { + a->gid = archive_write_disk_gid(&a->archive, + archive_entry_gname(a->entry), + archive_entry_gid(a->entry)); + } + + /* + * Restore ownership before set_mode tries to restore suid/sgid + * bits. If we set the owner, we know what it is and can skip + * a stat() call to examine the ownership of the file on disk. + */ + if (a->todo & TODO_OWNER) + ret = set_ownership(a); + + /* + * set_mode must precede ACLs on systems such as Solaris and + * FreeBSD where setting the mode implicitly clears extended ACLs + */ + if (a->todo & TODO_MODE) { + int r2 = set_mode(a, a->mode); + if (r2 < ret) ret = r2; + } + + /* + * Security-related extended attributes (such as + * security.capability on Linux) have to be restored last, + * since they're implicitly removed by other file changes. + */ + if (a->todo & TODO_XATTR) { + int r2 = set_xattrs(a); + if (r2 < ret) ret = r2; + } + + /* + * Some flags prevent file modification; they must be restored after + * file contents are written. + */ + if (a->todo & TODO_FFLAGS) { + int r2 = set_fflags(a); + if (r2 < ret) ret = r2; + } + + /* + * Time must follow most other metadata; + * otherwise atime will get changed. + */ + if (a->todo & TODO_TIMES) { + int r2 = set_times_from_entry(a); + if (r2 < ret) ret = r2; + } + + /* + * ACLs must be restored after timestamps because there are + * ACLs that prevent attribute changes (including time). + */ + if (a->todo & TODO_ACLS) { + int r2 = set_acls(a, a->fh, + archive_entry_pathname_w(a->entry), + archive_entry_acl(a->entry)); + if (r2 < ret) ret = r2; + } + + /* If there's an fd, we can close it now. */ + if (a->fh != INVALID_HANDLE_VALUE) { + CloseHandle(a->fh); + a->fh = INVALID_HANDLE_VALUE; + if (a->tmpname) { + /* Windows does not support atomic rename */ + disk_unlink(a->name); + if (_wrename(a->tmpname, a->name) != 0) { + la_dosmaperr(GetLastError()); + archive_set_error(&a->archive, errno, + "Failed to rename temporary file"); + ret = ARCHIVE_FAILED; + disk_unlink(a->tmpname); + } + a->tmpname = NULL; + } + } + /* If there's an entry, we can release it now. */ + archive_entry_free(a->entry); + a->entry = NULL; + a->archive.state = ARCHIVE_STATE_HEADER; + return (ret); +} + +int +archive_write_disk_set_group_lookup(struct archive *_a, + void *private_data, + la_int64_t (*lookup_gid)(void *private, const char *gname, la_int64_t gid), + void (*cleanup_gid)(void *private)) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_write_disk_set_group_lookup"); + + if (a->cleanup_gid != NULL && a->lookup_gid_data != NULL) + (a->cleanup_gid)(a->lookup_gid_data); + + a->lookup_gid = lookup_gid; + a->cleanup_gid = cleanup_gid; + a->lookup_gid_data = private_data; + return (ARCHIVE_OK); +} + +int +archive_write_disk_set_user_lookup(struct archive *_a, + void *private_data, + int64_t (*lookup_uid)(void *private, const char *uname, int64_t uid), + void (*cleanup_uid)(void *private)) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_write_disk_set_user_lookup"); + + if (a->cleanup_uid != NULL && a->lookup_uid_data != NULL) + (a->cleanup_uid)(a->lookup_uid_data); + + a->lookup_uid = lookup_uid; + a->cleanup_uid = cleanup_uid; + a->lookup_uid_data = private_data; + return (ARCHIVE_OK); +} + +int64_t +archive_write_disk_gid(struct archive *_a, const char *name, la_int64_t id) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_write_disk_gid"); + if (a->lookup_gid) + return (a->lookup_gid)(a->lookup_gid_data, name, id); + return (id); +} + +int64_t +archive_write_disk_uid(struct archive *_a, const char *name, la_int64_t id) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_ANY, "archive_write_disk_uid"); + if (a->lookup_uid) + return (a->lookup_uid)(a->lookup_uid_data, name, id); + return (id); +} + +/* + * Create a new archive_write_disk object and initialize it with global state. + */ +struct archive * +archive_write_disk_new(void) +{ + struct archive_write_disk *a; + + a = (struct archive_write_disk *)calloc(1, sizeof(*a)); + if (a == NULL) + return (NULL); + a->archive.magic = ARCHIVE_WRITE_DISK_MAGIC; + /* We're ready to write a header immediately. */ + a->archive.state = ARCHIVE_STATE_HEADER; + a->archive.vtable = archive_write_disk_vtable(); + a->start_time = time(NULL); + /* Query and restore the umask. */ + umask(a->user_umask = umask(0)); + if (archive_wstring_ensure(&a->path_safe, 512) == NULL) { + free(a); + return (NULL); + } + return (&a->archive); +} + +static int +disk_unlink(const wchar_t *path) +{ + wchar_t *fullname; + int r; + + r = _wunlink(path); + if (r != 0 && GetLastError() == ERROR_INVALID_NAME) { + fullname = __la_win_permissive_name_w(path); + r = _wunlink(fullname); + free(fullname); + } + return (r); +} + +static int +disk_rmdir(const wchar_t *path) +{ + wchar_t *fullname; + int r; + + r = _wrmdir(path); + if (r != 0 && GetLastError() == ERROR_INVALID_NAME) { + fullname = __la_win_permissive_name_w(path); + r = _wrmdir(fullname); + free(fullname); + } + return (r); +} + +/* + * The main restore function. + */ +static int +restore_entry(struct archive_write_disk *a) +{ + int ret = ARCHIVE_OK, en; + + if (a->flags & ARCHIVE_EXTRACT_UNLINK && !S_ISDIR(a->mode)) { + /* + * TODO: Fix this. Apparently, there are platforms + * that still allow root to hose the entire filesystem + * by unlinking a dir. The S_ISDIR() test above + * prevents us from using unlink() here if the new + * object is a dir, but that doesn't mean the old + * object isn't a dir. + */ + if (a->flags & ARCHIVE_EXTRACT_CLEAR_NOCHANGE_FFLAGS) + (void)clear_nochange_fflags(a); + if (disk_unlink(a->name) == 0) { + /* We removed it, reset cached stat. */ + a->pst = NULL; + } else if (errno == ENOENT) { + /* File didn't exist, that's just as good. */ + } else if (disk_rmdir(a->name) == 0) { + /* It was a dir, but now it's gone. */ + a->pst = NULL; + } else { + /* We tried, but couldn't get rid of it. */ + archive_set_error(&a->archive, errno, + "Could not unlink"); + return(ARCHIVE_FAILED); + } + } + + /* Try creating it first; if this fails, we'll try to recover. */ + en = create_filesystem_object(a); + + if ((en == ENOTDIR || en == ENOENT) + && !(a->flags & ARCHIVE_EXTRACT_NO_AUTODIR)) { + wchar_t *full; + /* If the parent dir doesn't exist, try creating it. */ + create_parent_dir(a, a->name); + /* Now try to create the object again. */ + full = __la_win_permissive_name_w(a->name); + if (full == NULL) { + en = EINVAL; + } else { + /* Remove multiple directories such as "a/../b../c" */ + archive_wstrcpy(&(a->_name_data), full); + a->name = a->_name_data.s; + free(full); + en = create_filesystem_object(a); + } + } + + if ((en == ENOENT) && (archive_entry_hardlink(a->entry) != NULL)) { + archive_set_error(&a->archive, en, + "Hard-link target '%s' does not exist.", + archive_entry_hardlink(a->entry)); + return (ARCHIVE_FAILED); + } + + if ((en == EISDIR || en == EEXIST) + && (a->flags & ARCHIVE_EXTRACT_NO_OVERWRITE)) { + /* If we're not overwriting, we're done. */ + if (S_ISDIR(a->mode)) { + /* Don't overwrite any settings on existing directories. */ + a->todo = 0; + } + archive_entry_unset_size(a->entry); + return (ARCHIVE_OK); + } + + /* + * Some platforms return EISDIR if you call + * open(O_WRONLY | O_EXCL | O_CREAT) on a directory, some + * return EEXIST. POSIX is ambiguous, requiring EISDIR + * for open(O_WRONLY) on a dir and EEXIST for open(O_EXCL | O_CREAT) + * on an existing item. + */ + if (en == EISDIR) { + /* A dir is in the way of a non-dir, rmdir it. */ + if (disk_rmdir(a->name) != 0) { + archive_set_error(&a->archive, errno, + "Can't remove already-existing dir"); + return (ARCHIVE_FAILED); + } + a->pst = NULL; + /* Try again. */ + en = create_filesystem_object(a); + } else if (en == EEXIST) { + mode_t st_mode; + mode_t lst_mode; + BY_HANDLE_FILE_INFORMATION lst; + /* + * We know something is in the way, but we don't know what; + * we need to find out before we go any further. + */ + int r = 0; + int dirlnk = 0; + + /* + * The SECURE_SYMLINK logic has already removed a + * symlink to a dir if the client wants that. So + * follow the symlink if we're creating a dir. + * If it's not a dir (or it's a broken symlink), + * then don't follow it. + * + * Windows distinguishes file and directory symlinks. + * A file symlink may erroneously point to a directory + * and a directory symlink to a file. Windows does not follow + * such symlinks. We always need both source and target + * information. + */ + r = file_information(a, a->name, &lst, &lst_mode, 1); + if (r != 0) { + archive_set_error(&a->archive, errno, + "Can't stat existing object"); + return (ARCHIVE_FAILED); + } else if (S_ISLNK(lst_mode)) { + if (lst.dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY) + dirlnk = 1; + /* In case of a symlink we need target information */ + r = file_information(a, a->name, &a->st, &st_mode, 0); + if (r != 0) { + a->st = lst; + st_mode = lst_mode; + } + } else { + a->st = lst; + st_mode = lst_mode; + } + + /* + * NO_OVERWRITE_NEWER doesn't apply to directories. + */ + if ((a->flags & ARCHIVE_EXTRACT_NO_OVERWRITE_NEWER) + && !S_ISDIR(st_mode)) { + if (!older(&(a->st), a->entry)) { + archive_entry_unset_size(a->entry); + return (ARCHIVE_OK); + } + } + + /* If it's our archive, we're done. */ + if (a->skip_file_set && + bhfi_dev(&a->st) == a->skip_file_dev && + bhfi_ino(&a->st) == a->skip_file_ino) { + archive_set_error(&a->archive, 0, + "Refusing to overwrite archive"); + return (ARCHIVE_FAILED); + } + + if (!S_ISDIR(st_mode)) { + if (a->flags & + ARCHIVE_EXTRACT_CLEAR_NOCHANGE_FFLAGS) { + (void)clear_nochange_fflags(a); + } + if ((a->flags & ARCHIVE_EXTRACT_SAFE_WRITES) && + S_ISREG(st_mode)) { + int fd = la_mktemp(a); + + if (fd == -1) { + la_dosmaperr(GetLastError()); + archive_set_error(&a->archive, errno, + "Can't create temporary file"); + return (ARCHIVE_FAILED); + } + a->fh = (HANDLE)_get_osfhandle(fd); + if (a->fh == INVALID_HANDLE_VALUE) { + la_dosmaperr(GetLastError()); + return (ARCHIVE_FAILED); + } + a->pst = NULL; + en = 0; + } else { + if (dirlnk) { + /* Edge case: dir symlink pointing + * to a file */ + if (disk_rmdir(a->name) != 0) { + archive_set_error(&a->archive, + errno, "Can't unlink " + "directory symlink"); + return (ARCHIVE_FAILED); + } + } else { + if (disk_unlink(a->name) != 0) { + /* A non-dir is in the way, + * unlink it. */ + archive_set_error(&a->archive, + errno, "Can't unlink " + "already-existing object"); + return (ARCHIVE_FAILED); + } + } + a->pst = NULL; + /* Try again. */ + en = create_filesystem_object(a); + } + } else if (!S_ISDIR(a->mode)) { + /* A dir is in the way of a non-dir, rmdir it. */ + if (a->flags & ARCHIVE_EXTRACT_CLEAR_NOCHANGE_FFLAGS) + (void)clear_nochange_fflags(a); + if (disk_rmdir(a->name) != 0) { + archive_set_error(&a->archive, errno, + "Can't remove already-existing dir"); + return (ARCHIVE_FAILED); + } + /* Try again. */ + en = create_filesystem_object(a); + } else { + /* + * There's a dir in the way of a dir. Don't + * waste time with rmdir()/mkdir(), just fix + * up the permissions on the existing dir. + * Note that we don't change perms on existing + * dirs unless _EXTRACT_PERM is specified. + */ + if ((a->mode != st_mode) + && (a->todo & TODO_MODE_FORCE)) + a->deferred |= (a->todo & TODO_MODE); + /* Ownership doesn't need deferred fixup. */ + en = 0; /* Forget the EEXIST. */ + } + } + + if (en) { + /* Everything failed; give up here. */ + archive_set_error(&a->archive, en, "Can't create '%ls'", + a->name); + return (ARCHIVE_FAILED); + } + + a->pst = NULL; /* Cached stat data no longer valid. */ + return (ret); +} + +/* + * Returns 0 if creation succeeds, or else returns errno value from + * the failed system call. Note: This function should only ever perform + * a single system call. + */ +static int +create_filesystem_object(struct archive_write_disk *a) +{ + /* Create the entry. */ + const wchar_t *linkname; + wchar_t *fullname; + mode_t final_mode, mode; + int r; + DWORD attrs = 0; + + /* We identify hard/symlinks according to the link names. */ + /* Since link(2) and symlink(2) don't handle modes, we're done here. */ + linkname = archive_entry_hardlink_w(a->entry); + if (linkname != NULL) { + wchar_t *linkfull, *namefull; + + linkfull = __la_win_permissive_name_w(linkname); + namefull = __la_win_permissive_name_w(a->name); + if (linkfull == NULL || namefull == NULL) { + errno = EINVAL; + r = -1; + } else { + /* + * Unlinking and linking here is really not atomic, + * but doing it right, would require us to construct + * an mktemplink() function, and then use _wrename(). + */ + if (a->flags & ARCHIVE_EXTRACT_SAFE_WRITES) { + attrs = GetFileAttributesW(namefull); + if (attrs != INVALID_FILE_ATTRIBUTES) { + if (attrs & FILE_ATTRIBUTE_DIRECTORY) + disk_rmdir(namefull); + else + disk_unlink(namefull); + } + } + r = la_CreateHardLinkW(namefull, linkfull); + if (r == 0) { + la_dosmaperr(GetLastError()); + r = errno; + } else + r = 0; + } + /* + * New cpio and pax formats allow hardlink entries + * to carry data, so we may have to open the file + * for hardlink entries. + * + * If the hardlink was successfully created and + * the archive doesn't have carry data for it, + * consider it to be non-authoritative for meta data. + * This is consistent with GNU tar and BSD pax. + * If the hardlink does carry data, let the last + * archive entry decide ownership. + */ + if (r == 0 && a->filesize <= 0) { + a->todo = 0; + a->deferred = 0; + } else if (r == 0 && a->filesize > 0) { + a->fh = CreateFileW(namefull, GENERIC_WRITE, 0, NULL, + TRUNCATE_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL); + if (a->fh == INVALID_HANDLE_VALUE) { + la_dosmaperr(GetLastError()); + r = errno; + } + } + free(linkfull); + free(namefull); + return (r); + } + linkname = archive_entry_symlink_w(a->entry); + if (linkname != NULL) { + /* + * Unlinking and linking here is really not atomic, + * but doing it right, would require us to construct + * an mktemplink() function, and then use _wrename(). + */ + attrs = GetFileAttributesW(a->name); + if (attrs != INVALID_FILE_ATTRIBUTES) { + if (attrs & FILE_ATTRIBUTE_DIRECTORY) + disk_rmdir(a->name); + else + disk_unlink(a->name); + } +#if HAVE_SYMLINK + return symlink(linkname, a->name) ? errno : 0; +#else + errno = 0; + r = la_CreateSymbolicLinkW((const wchar_t *)a->name, linkname, + archive_entry_symlink_type(a->entry)); + if (r == 0) { + if (errno == 0) + la_dosmaperr(GetLastError()); + r = errno; + } else + r = 0; + return (r); +#endif + } + + /* + * The remaining system calls all set permissions, so let's + * try to take advantage of that to avoid an extra chmod() + * call. (Recall that umask is set to zero right now!) + */ + + /* Mode we want for the final restored object (w/o file type bits). */ + final_mode = a->mode & 07777; + /* + * The mode that will actually be restored in this step. Note + * that SUID, SGID, etc, require additional work to ensure + * security, so we never restore them at this point. + */ + mode = final_mode & 0777 & ~a->user_umask; + + switch (a->mode & AE_IFMT) { + default: + /* POSIX requires that we fall through here. */ + /* FALLTHROUGH */ + case AE_IFREG: + a->tmpname = NULL; + fullname = a->name; + /* O_WRONLY | O_CREAT | O_EXCL */ + a->fh = CreateFileW(fullname, GENERIC_WRITE, 0, NULL, + CREATE_NEW, FILE_ATTRIBUTE_NORMAL, NULL); + if (a->fh == INVALID_HANDLE_VALUE && + GetLastError() == ERROR_INVALID_NAME && + fullname == a->name) { + fullname = __la_win_permissive_name_w(a->name); + a->fh = CreateFileW(fullname, GENERIC_WRITE, 0, NULL, + CREATE_NEW, FILE_ATTRIBUTE_NORMAL, NULL); + } + if (a->fh == INVALID_HANDLE_VALUE) { + if (GetLastError() == ERROR_ACCESS_DENIED) { + DWORD attr; + /* Simulate an errno of POSIX system. */ + attr = GetFileAttributesW(fullname); + if (attr == (DWORD)-1) + la_dosmaperr(GetLastError()); + else if (attr & FILE_ATTRIBUTE_DIRECTORY) + errno = EISDIR; + else + errno = EACCES; + } else + la_dosmaperr(GetLastError()); + r = 1; + } else + r = 0; + if (fullname != a->name) + free(fullname); + break; + case AE_IFCHR: + case AE_IFBLK: + /* TODO: Find a better way to warn about our inability + * to restore a block device node. */ + return (EINVAL); + case AE_IFDIR: + mode = (mode | MINIMUM_DIR_MODE) & MAXIMUM_DIR_MODE; + fullname = a->name; + r = CreateDirectoryW(fullname, NULL); + if (r == 0 && GetLastError() == ERROR_INVALID_NAME && + fullname == a->name) { + fullname = __la_win_permissive_name_w(a->name); + r = CreateDirectoryW(fullname, NULL); + } + if (r != 0) { + r = 0; + /* Defer setting dir times. */ + a->deferred |= (a->todo & TODO_TIMES); + a->todo &= ~TODO_TIMES; + /* Never use an immediate chmod(). */ + /* We can't avoid the chmod() entirely if EXTRACT_PERM + * because of SysV SGID inheritance. */ + if ((mode != final_mode) + || (a->flags & ARCHIVE_EXTRACT_PERM)) + a->deferred |= (a->todo & TODO_MODE); + a->todo &= ~TODO_MODE; + } else { + la_dosmaperr(GetLastError()); + r = -1; + } + if (fullname != a->name) + free(fullname); + break; + case AE_IFIFO: + /* TODO: Find a better way to warn about our inability + * to restore a fifo. */ + return (EINVAL); + } + + /* All the system calls above set errno on failure. */ + if (r) + return (errno); + + /* If we managed to set the final mode, we've avoided a chmod(). */ + if (mode == final_mode) + a->todo &= ~TODO_MODE; + return (0); +} + +/* + * Cleanup function for archive_extract. Mostly, this involves processing + * the fixup list, which is used to address a number of problems: + * * Dir permissions might prevent us from restoring a file in that + * dir, so we restore the dir with minimum 0700 permissions first, + * then correct the mode at the end. + * * Similarly, the act of restoring a file touches the directory + * and changes the timestamp on the dir, so we have to touch-up dir + * timestamps at the end as well. + * * Some file flags can interfere with the restore by, for example, + * preventing the creation of hardlinks to those files. + * * Mac OS extended metadata includes ACLs, so must be deferred on dirs. + * + * Note that tar/cpio do not require that archives be in a particular + * order; there is no way to know when the last file has been restored + * within a directory, so there's no way to optimize the memory usage + * here by fixing up the directory any earlier than the + * end-of-archive. + * + * XXX TODO: Directory ACLs should be restored here, for the same + * reason we set directory perms here. XXX + */ +static int +_archive_write_disk_close(struct archive *_a) +{ + struct archive_write_disk *a = (struct archive_write_disk *)_a; + struct fixup_entry *next, *p; + int ret; + + archive_check_magic(&a->archive, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, + "archive_write_disk_close"); + ret = _archive_write_disk_finish_entry(&a->archive); + + /* Sort dir list so directories are fixed up in depth-first order. */ + p = sort_dir_list(a->fixup_list); + + while (p != NULL) { + a->pst = NULL; /* Mark stat cache as out-of-date. */ + if (p->fixup & TODO_TIMES) { + set_times(a, INVALID_HANDLE_VALUE, p->mode, p->name, + p->atime, p->atime_nanos, + p->birthtime, p->birthtime_nanos, + p->mtime, p->mtime_nanos, + p->ctime, p->ctime_nanos); + } + if (p->fixup & TODO_MODE_BASE) + la_chmod(p->name, p->mode); + if (p->fixup & TODO_ACLS) + set_acls(a, INVALID_HANDLE_VALUE, p->name, &p->acl); + if (p->fixup & TODO_FFLAGS) + set_fflags_platform(p->name, p->fflags_set, 0); + next = p->next; + archive_acl_clear(&p->acl); + free(p->name); + free(p); + p = next; + } + a->fixup_list = NULL; + return (ret); +} + +static int +_archive_write_disk_free(struct archive *_a) +{ + struct archive_write_disk *a; + int ret; + if (_a == NULL) + return (ARCHIVE_OK); + archive_check_magic(_a, ARCHIVE_WRITE_DISK_MAGIC, + ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_write_disk_free"); + a = (struct archive_write_disk *)_a; + ret = _archive_write_disk_close(&a->archive); + archive_write_disk_set_group_lookup(&a->archive, NULL, NULL, NULL); + archive_write_disk_set_user_lookup(&a->archive, NULL, NULL, NULL); + archive_entry_free(a->entry); + archive_wstring_free(&a->_name_data); + archive_wstring_free(&a->_tmpname_data); + archive_string_free(&a->archive.error_string); + archive_wstring_free(&a->path_safe); + a->archive.magic = 0; + __archive_clean(&a->archive); + free(a); + return (ret); +} + +/* + * Simple O(n log n) merge sort to order the fixup list. In + * particular, we want to restore dir timestamps depth-first. + */ +static struct fixup_entry * +sort_dir_list(struct fixup_entry *p) +{ + struct fixup_entry *a, *b, *t; + + if (p == NULL) + return (NULL); + /* A one-item list is already sorted. */ + if (p->next == NULL) + return (p); + + /* Step 1: split the list. */ + t = p; + a = p->next->next; + while (a != NULL) { + /* Step a twice, t once. */ + a = a->next; + if (a != NULL) + a = a->next; + t = t->next; + } + /* Now, t is at the mid-point, so break the list here. */ + b = t->next; + t->next = NULL; + a = p; + + /* Step 2: Recursively sort the two sub-lists. */ + a = sort_dir_list(a); + b = sort_dir_list(b); + + /* Step 3: Merge the returned lists. */ + /* Pick the first element for the merged list. */ + if (wcscmp(a->name, b->name) > 0) { + t = p = a; + a = a->next; + } else { + t = p = b; + b = b->next; + } + + /* Always put the later element on the list first. */ + while (a != NULL && b != NULL) { + if (wcscmp(a->name, b->name) > 0) { + t->next = a; + a = a->next; + } else { + t->next = b; + b = b->next; + } + t = t->next; + } + + /* Only one list is non-empty, so just splice it on. */ + if (a != NULL) + t->next = a; + if (b != NULL) + t->next = b; + + return (p); +} + +/* + * Returns a new, initialized fixup entry. + * + * TODO: Reduce the memory requirements for this list by using a tree + * structure rather than a simple list of names. + */ +static struct fixup_entry * +new_fixup(struct archive_write_disk *a, const wchar_t *pathname) +{ + struct fixup_entry *fe; + + fe = (struct fixup_entry *)calloc(1, sizeof(struct fixup_entry)); + if (fe == NULL) + return (NULL); + fe->next = a->fixup_list; + a->fixup_list = fe; + fe->fixup = 0; + fe->name = _wcsdup(pathname); + fe->fflags_set = 0; + return (fe); +} + +/* + * Returns a fixup structure for the current entry. + */ +static struct fixup_entry * +current_fixup(struct archive_write_disk *a, const wchar_t *pathname) +{ + if (a->current_fixup == NULL) + a->current_fixup = new_fixup(a, pathname); + return (a->current_fixup); +} + +/* + * TODO: The deep-directory support bypasses this; disable deep directory + * support if we're doing symlink checks. + */ +/* + * TODO: Someday, integrate this with the deep dir support; they both + * scan the path and both can be optimized by comparing against other + * recent paths. + */ +static int +check_symlinks(struct archive_write_disk *a) +{ + wchar_t *pn, *p; + wchar_t c; + int r; + BY_HANDLE_FILE_INFORMATION st; + mode_t st_mode; + + /* + * Guard against symlink tricks. Reject any archive entry whose + * destination would be altered by a symlink. + */ + /* Whatever we checked last time doesn't need to be re-checked. */ + pn = a->name; + p = a->path_safe.s; + while ((*pn != '\0') && (*p == *pn)) + ++p, ++pn; + /* Skip leading backslashes */ + while (*pn == '\\') + ++pn; + c = pn[0]; + /* Keep going until we've checked the entire name. */ + while (pn[0] != '\0' && (pn[0] != '\\' || pn[1] != '\0')) { + /* Skip the next path element. */ + while (*pn != '\0' && *pn != '\\') + ++pn; + c = pn[0]; + pn[0] = '\0'; + /* Check that we haven't hit a symlink. */ + r = file_information(a, a->name, &st, &st_mode, 1); + if (r != 0) { + /* We've hit a dir that doesn't exist; stop now. */ + if (errno == ENOENT) + break; + } else if (S_ISLNK(st_mode)) { + if (c == '\0') { + /* + * Last element is a file or directory symlink. + * Remove it so we can overwrite it with the + * item being extracted. + */ + if (a->flags & + ARCHIVE_EXTRACT_CLEAR_NOCHANGE_FFLAGS) { + (void)clear_nochange_fflags(a); + } + if (st.dwFileAttributes & + FILE_ATTRIBUTE_DIRECTORY) { + r = disk_rmdir(a->name); + } else { + r = disk_unlink(a->name); + } + if (r) { + archive_set_error(&a->archive, errno, + "Could not remove symlink %ls", + a->name); + pn[0] = c; + return (ARCHIVE_FAILED); + } + a->pst = NULL; + /* + * Even if we did remove it, a warning + * is in order. The warning is silly, + * though, if we're just replacing one + * symlink with another symlink. + */ + if (!S_ISLNK(a->mode)) { + archive_set_error(&a->archive, 0, + "Removing symlink %ls", + a->name); + } + /* Symlink gone. No more problem! */ + pn[0] = c; + return (0); + } else if (a->flags & ARCHIVE_EXTRACT_UNLINK) { + /* User asked us to remove problems. */ + if (a->flags & + ARCHIVE_EXTRACT_CLEAR_NOCHANGE_FFLAGS) { + (void)clear_nochange_fflags(a); + } + if (st.dwFileAttributes & + FILE_ATTRIBUTE_DIRECTORY) { + r = disk_rmdir(a->name); + } else { + r = disk_unlink(a->name); + } + if (r != 0) { + archive_set_error(&a->archive, 0, + "Cannot remove intervening " + "symlink %ls", a->name); + pn[0] = c; + return (ARCHIVE_FAILED); + } + a->pst = NULL; + } else { + archive_set_error(&a->archive, 0, + "Cannot extract through symlink %ls", + a->name); + pn[0] = c; + return (ARCHIVE_FAILED); + } + } + pn[0] = c; + pn++; + } + pn[0] = c; + /* We've checked and/or cleaned the whole path, so remember it. */ + archive_wstrcpy(&a->path_safe, a->name); + return (ARCHIVE_OK); +} + +static int +guidword(wchar_t *p, int n) +{ + int i; + + for (i = 0; i < n; i++) { + if ((*p >= L'0' && *p <= L'9') || + (*p >= L'a' && *p <= L'f') || + (*p >= L'A' && *p <= L'F')) + p++; + else + return (-1); + } + return (0); +} + +/* + * Canonicalize the pathname. In particular, this strips duplicate + * '\' characters, '.' elements, and trailing '\'. It also raises an + * error for an empty path, a trailing '..' or (if _SECURE_NODOTDOT is + * set) any '..' in the path. + */ +static int +cleanup_pathname(struct archive_write_disk *a) +{ + wchar_t *dest, *src, *p, *top; + wchar_t separator = L'\0'; + + p = a->name; + if (*p == L'\0') { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Invalid empty pathname"); + return (ARCHIVE_FAILED); + } + + /* Replace '/' by '\' */ + for (; *p != L'\0'; p++) { + if (*p == L'/') + *p = L'\\'; + } + p = a->name; + + /* Skip leading "\\.\" or "\\?\" or "\\?\UNC\" or + * "\\?\Volume{GUID}\" + * (absolute path prefixes used by Windows API) */ + if (p[0] == L'\\' && p[1] == L'\\' && + (p[2] == L'.' || p[2] == L'?') && p[3] == L'\\') + { + /* A path begin with "\\?\UNC\" */ + if (p[2] == L'?' && + (p[4] == L'U' || p[4] == L'u') && + (p[5] == L'N' || p[5] == L'n') && + (p[6] == L'C' || p[6] == L'c') && + p[7] == L'\\') + p += 8; + /* A path begin with "\\?\Volume{GUID}\" */ + else if (p[2] == L'?' && + (p[4] == L'V' || p[4] == L'v') && + (p[5] == L'O' || p[5] == L'o') && + (p[6] == L'L' || p[6] == L'l') && + (p[7] == L'U' || p[7] == L'u') && + (p[8] == L'M' || p[8] == L'm') && + (p[9] == L'E' || p[9] == L'e') && + p[10] == L'{') { + if (guidword(p+11, 8) == 0 && p[19] == L'-' && + guidword(p+20, 4) == 0 && p[24] == L'-' && + guidword(p+25, 4) == 0 && p[29] == L'-' && + guidword(p+30, 4) == 0 && p[34] == L'-' && + guidword(p+35, 12) == 0 && p[47] == L'}' && + p[48] == L'\\') + p += 49; + else + p += 4; + /* A path begin with "\\.\PhysicalDriveX" */ + } else if (p[2] == L'.' && + (p[4] == L'P' || p[4] == L'p') && + (p[5] == L'H' || p[5] == L'h') && + (p[6] == L'Y' || p[6] == L'y') && + (p[7] == L'S' || p[7] == L's') && + (p[8] == L'I' || p[8] == L'i') && + (p[9] == L'C' || p[9] == L'c') && + (p[9] == L'A' || p[9] == L'a') && + (p[9] == L'L' || p[9] == L'l') && + (p[9] == L'D' || p[9] == L'd') && + (p[9] == L'R' || p[9] == L'r') && + (p[9] == L'I' || p[9] == L'i') && + (p[9] == L'V' || p[9] == L'v') && + (p[9] == L'E' || p[9] == L'e') && + (p[10] >= L'0' && p[10] <= L'9') && + p[11] == L'\0') { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Path is a physical drive name"); + return (ARCHIVE_FAILED); + } else + p += 4; + } + + /* Skip leading drive letter from archives created + * on Windows. */ + if (((p[0] >= L'a' && p[0] <= L'z') || + (p[0] >= L'A' && p[0] <= L'Z')) && + p[1] == L':') { + if (p[2] == L'\0') { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Path is a drive name"); + return (ARCHIVE_FAILED); + } + if (p[2] == L'\\') + p += 2; + } + + top = dest = src = p; + /* Rewrite the path name if its character is a unusable. */ + for (; *p != L'\0'; p++) { + if (*p == L':' || *p == L'*' || *p == L'?' || *p == L'"' || + *p == L'<' || *p == L'>' || *p == L'|') + *p = L'_'; + } + /* Skip leading '\'. */ + if (*src == L'\\') + separator = *src++; + + /* Scan the pathname one element at a time. */ + for (;;) { + /* src points to first char after '\' */ + if (src[0] == L'\0') { + break; + } else if (src[0] == L'\\') { + /* Found '\\'('//'), ignore second one. */ + src++; + continue; + } else if (src[0] == L'.') { + if (src[1] == L'\0') { + /* Ignore trailing '.' */ + break; + } else if (src[1] == L'\\') { + /* Skip '.\'. */ + src += 2; + continue; + } else if (src[1] == L'.') { + if (src[2] == L'\\' || src[2] == L'\0') { + /* Conditionally warn about '..' */ + if (a->flags & + ARCHIVE_EXTRACT_SECURE_NODOTDOT) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Path contains '..'"); + return (ARCHIVE_FAILED); + } + } + /* + * Note: Under no circumstances do we + * remove '..' elements. In + * particular, restoring + * '\foo\..\bar\' should create the + * 'foo' dir as a side-effect. + */ + } + } + + /* Copy current element, including leading '\'. */ + if (separator) + *dest++ = L'\\'; + while (*src != L'\0' && *src != L'\\') { + *dest++ = *src++; + } + + if (*src == L'\0') + break; + + /* Skip '\' separator. */ + separator = *src++; + } + /* + * We've just copied zero or more path elements, not including the + * final '\'. + */ + if (dest == top) { + /* + * Nothing got copied. The path must have been something + * like '.' or '\' or './' or '/././././/./'. + */ + if (separator) + *dest++ = L'\\'; + else + *dest++ = L'.'; + } + /* Terminate the result. */ + *dest = L'\0'; + return (ARCHIVE_OK); +} + +/* + * Create the parent directory of the specified path, assuming path + * is already in mutable storage. + */ +static int +create_parent_dir(struct archive_write_disk *a, wchar_t *path) +{ + wchar_t *slash; + int r; + + /* Remove tail element to obtain parent name. */ + slash = wcsrchr(path, L'\\'); + if (slash == NULL) + return (ARCHIVE_OK); + *slash = L'\0'; + r = create_dir(a, path); + *slash = L'\\'; + return (r); +} + +/* + * Create the specified dir, recursing to create parents as necessary. + * + * Returns ARCHIVE_OK if the path exists when we're done here. + * Otherwise, returns ARCHIVE_FAILED. + * Assumes path is in mutable storage; path is unchanged on exit. + */ +static int +create_dir(struct archive_write_disk *a, wchar_t *path) +{ + BY_HANDLE_FILE_INFORMATION st; + struct fixup_entry *le; + wchar_t *slash, *base, *full; + mode_t mode_final, mode, st_mode; + int r; + + /* Check for special names and just skip them. */ + slash = wcsrchr(path, L'\\'); + if (slash == NULL) + base = path; + else + base = slash + 1; + + if (base[0] == L'\0' || + (base[0] == L'.' && base[1] == L'\0') || + (base[0] == L'.' && base[1] == L'.' && base[2] == L'\0')) { + /* Don't bother trying to create null path, '.', or '..'. */ + if (slash != NULL) { + *slash = L'\0'; + r = create_dir(a, path); + *slash = L'\\'; + return (r); + } + return (ARCHIVE_OK); + } + + /* + * Yes, this should be stat() and not lstat(). Using lstat() + * here loses the ability to extract through symlinks. Also note + * that this should not use the a->st cache. + */ + if (file_information(a, path, &st, &st_mode, 0) == 0) { + if (S_ISDIR(st_mode)) + return (ARCHIVE_OK); + if ((a->flags & ARCHIVE_EXTRACT_NO_OVERWRITE)) { + archive_set_error(&a->archive, EEXIST, + "Can't create directory '%ls'", path); + return (ARCHIVE_FAILED); + } + if (disk_unlink(path) != 0) { + archive_set_error(&a->archive, errno, + "Can't create directory '%ls': " + "Conflicting file cannot be removed", + path); + return (ARCHIVE_FAILED); + } + } else if (errno != ENOENT && errno != ENOTDIR) { + /* Stat failed? */ + archive_set_error(&a->archive, errno, + "Can't test directory '%ls'", path); + return (ARCHIVE_FAILED); + } else if (slash != NULL) { + *slash = '\0'; + r = create_dir(a, path); + *slash = '\\'; + if (r != ARCHIVE_OK) + return (r); + } + + /* + * Mode we want for the final restored directory. Per POSIX, + * implicitly-created dirs must be created obeying the umask. + * There's no mention whether this is different for privileged + * restores (which the rest of this code handles by pretending + * umask=0). I've chosen here to always obey the user's umask for + * implicit dirs, even if _EXTRACT_PERM was specified. + */ + mode_final = DEFAULT_DIR_MODE & ~a->user_umask; + /* Mode we want on disk during the restore process. */ + mode = mode_final; + mode |= MINIMUM_DIR_MODE; + mode &= MAXIMUM_DIR_MODE; + /* + * Apply __la_win_permissive_name_w to path in order to + * remove '../' path string. + */ + full = __la_win_permissive_name_w(path); + if (full == NULL) + errno = EINVAL; + else if (CreateDirectoryW(full, NULL) != 0) { + if (mode != mode_final) { + le = new_fixup(a, path); + le->fixup |=TODO_MODE_BASE; + le->mode = mode_final; + } + free(full); + return (ARCHIVE_OK); + } else { + la_dosmaperr(GetLastError()); + } + free(full); + + /* + * Without the following check, a/b/../b/c/d fails at the + * second visit to 'b', so 'd' can't be created. Note that we + * don't add it to the fixup list here, as it's already been + * added. + */ + if (file_information(a, path, &st, &st_mode, 0) == 0 && + S_ISDIR(st_mode)) + return (ARCHIVE_OK); + + archive_set_error(&a->archive, errno, "Failed to create dir '%ls'", + path); + return (ARCHIVE_FAILED); +} + +/* + * Note: Although we can skip setting the user id if the desired user + * id matches the current user, we cannot skip setting the group, as + * many systems set the gid based on the containing directory. So + * we have to perform a chown syscall if we want to set the SGID + * bit. (The alternative is to stat() and then possibly chown(); it's + * more efficient to skip the stat() and just always chown().) Note + * that a successful chown() here clears the TODO_SGID_CHECK bit, which + * allows set_mode to skip the stat() check for the GID. + */ +static int +set_ownership(struct archive_write_disk *a) +{ +/* unfortunately, on win32 there is no 'root' user with uid 0, + so we just have to try the chown and see if it works */ + + /* If we know we can't change it, don't bother trying. */ + if (a->user_uid != 0 && a->user_uid != a->uid) { + archive_set_error(&a->archive, errno, + "Can't set UID=%jd", (intmax_t)a->uid); + return (ARCHIVE_WARN); + } + + archive_set_error(&a->archive, errno, + "Can't set user=%jd/group=%jd for %ls", + (intmax_t)a->uid, (intmax_t)a->gid, a->name); + return (ARCHIVE_WARN); +} + +static int +set_times(struct archive_write_disk *a, + HANDLE h, int mode, const wchar_t *name, + time_t atime, long atime_nanos, + time_t birthtime, long birthtime_nanos, + time_t mtime, long mtime_nanos, + time_t ctime_sec, long ctime_nanos) +{ +#define EPOC_TIME ARCHIVE_LITERAL_ULL(116444736000000000) +#define WINTIME(sec, nsec) ((Int32x32To64(sec, 10000000) + EPOC_TIME)\ + + (((nsec)/1000)*10)) + + HANDLE hw = 0; + ULARGE_INTEGER wintm; + FILETIME *pfbtime; + FILETIME fatime, fbtime, fmtime; + + (void)ctime_sec; /* UNUSED */ + (void)ctime_nanos; /* UNUSED */ + + if (h != INVALID_HANDLE_VALUE) { + hw = NULL; + } else { + wchar_t *ws; + + if (S_ISLNK(mode)) + return (ARCHIVE_OK); + ws = __la_win_permissive_name_w(name); + if (ws == NULL) + goto settimes_failed; + hw = CreateFileW(ws, FILE_WRITE_ATTRIBUTES, + 0, NULL, OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS, NULL); + free(ws); + if (hw == INVALID_HANDLE_VALUE) + goto settimes_failed; + h = hw; + } + + wintm.QuadPart = WINTIME(atime, atime_nanos); + fatime.dwLowDateTime = wintm.LowPart; + fatime.dwHighDateTime = wintm.HighPart; + wintm.QuadPart = WINTIME(mtime, mtime_nanos); + fmtime.dwLowDateTime = wintm.LowPart; + fmtime.dwHighDateTime = wintm.HighPart; + /* + * SetFileTime() supports birthtime. + */ + if (birthtime > 0 || birthtime_nanos > 0) { + wintm.QuadPart = WINTIME(birthtime, birthtime_nanos); + fbtime.dwLowDateTime = wintm.LowPart; + fbtime.dwHighDateTime = wintm.HighPart; + pfbtime = &fbtime; + } else + pfbtime = NULL; + if (SetFileTime(h, pfbtime, &fatime, &fmtime) == 0) + goto settimes_failed; + CloseHandle(hw); + return (ARCHIVE_OK); + +settimes_failed: + CloseHandle(hw); + archive_set_error(&a->archive, EINVAL, "Can't restore time"); + return (ARCHIVE_WARN); +} + +static int +set_times_from_entry(struct archive_write_disk *a) +{ + time_t atime, birthtime, mtime, ctime_sec; + long atime_nsec, birthtime_nsec, mtime_nsec, ctime_nsec; + + /* Suitable defaults. */ + atime = birthtime = mtime = ctime_sec = a->start_time; + atime_nsec = birthtime_nsec = mtime_nsec = ctime_nsec = 0; + + /* If no time was provided, we're done. */ + if (!archive_entry_atime_is_set(a->entry) + && !archive_entry_birthtime_is_set(a->entry) + && !archive_entry_mtime_is_set(a->entry)) + return (ARCHIVE_OK); + + if (archive_entry_atime_is_set(a->entry)) { + atime = archive_entry_atime(a->entry); + atime_nsec = archive_entry_atime_nsec(a->entry); + } + if (archive_entry_birthtime_is_set(a->entry)) { + birthtime = archive_entry_birthtime(a->entry); + birthtime_nsec = archive_entry_birthtime_nsec(a->entry); + } + if (archive_entry_mtime_is_set(a->entry)) { + mtime = archive_entry_mtime(a->entry); + mtime_nsec = archive_entry_mtime_nsec(a->entry); + } + if (archive_entry_ctime_is_set(a->entry)) { + ctime_sec = archive_entry_ctime(a->entry); + ctime_nsec = archive_entry_ctime_nsec(a->entry); + } + + return set_times(a, a->fh, a->mode, a->name, + atime, atime_nsec, + birthtime, birthtime_nsec, + mtime, mtime_nsec, + ctime_sec, ctime_nsec); +} + +static int +set_mode(struct archive_write_disk *a, int mode) +{ + int r = ARCHIVE_OK; + mode &= 07777; /* Strip off file type bits. */ + + if (a->todo & TODO_SGID_CHECK) { + /* + * If we don't know the GID is right, we must stat() + * to verify it. We can't just check the GID of this + * process, since systems sometimes set GID from + * the enclosing dir or based on ACLs. + */ + if ((r = lazy_stat(a)) != ARCHIVE_OK) + return (r); + if (0 != a->gid) { + mode &= ~ S_ISGID; + } + /* While we're here, double-check the UID. */ + if (0 != a->uid + && (a->todo & TODO_SUID)) { + mode &= ~ S_ISUID; + } + a->todo &= ~TODO_SGID_CHECK; + a->todo &= ~TODO_SUID_CHECK; + } else if (a->todo & TODO_SUID_CHECK) { + /* + * If we don't know the UID is right, we can just check + * the user, since all systems set the file UID from + * the process UID. + */ + if (a->user_uid != a->uid) { + mode &= ~ S_ISUID; + } + a->todo &= ~TODO_SUID_CHECK; + } + + if (S_ISLNK(a->mode)) { +#ifdef HAVE_LCHMOD + /* + * If this is a symlink, use lchmod(). If the + * platform doesn't support lchmod(), just skip it. A + * platform that doesn't provide a way to set + * permissions on symlinks probably ignores + * permissions on symlinks, so a failure here has no + * impact. + */ + if (lchmod(a->name, mode) != 0) { + archive_set_error(&a->archive, errno, + "Can't set permissions to 0%o", (int)mode); + r = ARCHIVE_WARN; + } +#endif + } else if (!S_ISDIR(a->mode)) { + /* + * If it's not a symlink and not a dir, then use + * fchmod() or chmod(), depending on whether we have + * an fd. Dirs get their perms set during the + * post-extract fixup, which is handled elsewhere. + */ +#ifdef HAVE_FCHMOD + if (a->fd >= 0) { + if (fchmod(a->fd, mode) != 0) { + archive_set_error(&a->archive, errno, + "Can't set permissions to 0%o", (int)mode); + r = ARCHIVE_WARN; + } + } else +#endif + /* If this platform lacks fchmod(), then + * we'll just use chmod(). */ + if (la_chmod(a->name, mode) != 0) { + archive_set_error(&a->archive, errno, + "Can't set permissions to 0%o", (int)mode); + r = ARCHIVE_WARN; + } + } + return (r); +} + +static int set_fflags_platform(const wchar_t *name, unsigned long fflags_set, + unsigned long fflags_clear) +{ + DWORD oldflags, newflags; + wchar_t *fullname; + + const DWORD settable_flags = + FILE_ATTRIBUTE_ARCHIVE | + FILE_ATTRIBUTE_HIDDEN | + FILE_ATTRIBUTE_NORMAL | + FILE_ATTRIBUTE_NOT_CONTENT_INDEXED | + FILE_ATTRIBUTE_OFFLINE | + FILE_ATTRIBUTE_READONLY | + FILE_ATTRIBUTE_SYSTEM | + FILE_ATTRIBUTE_TEMPORARY; + + oldflags = GetFileAttributesW(name); + if (oldflags == (DWORD)-1 && + GetLastError() == ERROR_INVALID_NAME) { + fullname = __la_win_permissive_name_w(name); + oldflags = GetFileAttributesW(fullname); + } + if (oldflags == (DWORD)-1) { + la_dosmaperr(GetLastError()); + return (ARCHIVE_WARN); + } + newflags = ((oldflags & ~fflags_clear) | fflags_set) & settable_flags; + if(SetFileAttributesW(name, newflags) == 0) + return (ARCHIVE_WARN); + return (ARCHIVE_OK); +} + +static int +clear_nochange_fflags(struct archive_write_disk *a) +{ + return (set_fflags_platform(a->name, 0, FILE_ATTRIBUTE_READONLY)); +} + +static int +set_fflags(struct archive_write_disk *a) +{ + unsigned long set, clear; + + if (a->todo & TODO_FFLAGS) { + archive_entry_fflags(a->entry, &set, &clear); + if (set == 0 && clear == 0) + return (ARCHIVE_OK); + return (set_fflags_platform(a->name, set, clear)); + + } + return (ARCHIVE_OK); +} + +/* Default empty function body to satisfy mainline code. */ +static int +set_acls(struct archive_write_disk *a, HANDLE h, const wchar_t *name, + struct archive_acl *acl) +{ + (void)a; /* UNUSED */ + (void)h; /* UNUSED */ + (void)name; /* UNUSED */ + (void)acl; /* UNUSED */ + return (ARCHIVE_OK); +} + +/* + * Restore extended attributes - stub implementation for unsupported systems + */ +static int +set_xattrs(struct archive_write_disk *a) +{ + static int warning_done = 0; + + /* If there aren't any extended attributes, then it's okay not + * to extract them, otherwise, issue a single warning. */ + if (archive_entry_xattr_count(a->entry) != 0 && !warning_done) { + warning_done = 1; + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Cannot restore extended attributes on this system"); + return (ARCHIVE_WARN); + } + /* Warning was already emitted; suppress further warnings. */ + return (ARCHIVE_OK); +} + +static void +fileTimeToUtc(const FILETIME *filetime, time_t *t, long *ns) +{ + ULARGE_INTEGER utc; + + utc.HighPart = filetime->dwHighDateTime; + utc.LowPart = filetime->dwLowDateTime; + if (utc.QuadPart >= EPOC_TIME) { + utc.QuadPart -= EPOC_TIME; + /* milli seconds base */ + *t = (time_t)(utc.QuadPart / 10000000); + /* nano seconds base */ + *ns = (long)(utc.QuadPart % 10000000) * 100; + } else { + *t = 0; + *ns = 0; + } +} +/* + * Test if file on disk is older than entry. + */ +static int +older(BY_HANDLE_FILE_INFORMATION *st, struct archive_entry *entry) +{ + time_t sec; + long nsec; + + fileTimeToUtc(&st->ftLastWriteTime, &sec, &nsec); + /* First, test the seconds and return if we have a definite answer. */ + /* Definitely older. */ + if (sec < archive_entry_mtime(entry)) + return (1); + /* Definitely younger. */ + if (sec > archive_entry_mtime(entry)) + return (0); + if (nsec < archive_entry_mtime_nsec(entry)) + return (1); + /* Same age or newer, so not older. */ + return (0); +} + +#endif /* _WIN32 && !__CYGWIN__ */ + diff --git a/c/archive_write_open_fd.c b/c/archive_write_open_fd.c new file mode 100644 --- /dev/null +++ b/c/archive_write_open_fd.c @@ -0,0 +1,144 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_open_fd.c 201093 2009-12-28 02:28:44Z kientzle $"); + +#ifdef HAVE_SYS_STAT_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_FCNTL_H +#include +#endif +#ifdef HAVE_IO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif + +#include "archive.h" + +struct write_fd_data { + int fd; +}; + +static int file_close(struct archive *, void *); +static int file_open(struct archive *, void *); +static ssize_t file_write(struct archive *, void *, const void *buff, size_t); + +int +archive_write_open_fd(struct archive *a, int fd) +{ + struct write_fd_data *mine; + + mine = (struct write_fd_data *)malloc(sizeof(*mine)); + if (mine == NULL) { + archive_set_error(a, ENOMEM, "No memory"); + return (ARCHIVE_FATAL); + } + mine->fd = fd; +#if defined(__CYGWIN__) || defined(_WIN32) + setmode(mine->fd, O_BINARY); +#endif + return (archive_write_open(a, mine, + file_open, file_write, file_close)); +} + +static int +file_open(struct archive *a, void *client_data) +{ + struct write_fd_data *mine; + struct stat st; + + mine = (struct write_fd_data *)client_data; + + if (fstat(mine->fd, &st) != 0) { + archive_set_error(a, errno, "Couldn't stat fd %d", mine->fd); + return (ARCHIVE_FATAL); + } + + /* + * If this is a regular file, don't add it to itself. + */ + if (S_ISREG(st.st_mode)) + archive_write_set_skip_file(a, st.st_dev, st.st_ino); + + /* + * If client hasn't explicitly set the last block handling, + * then set it here. + */ + if (archive_write_get_bytes_in_last_block(a) < 0) { + /* If the output is a block or character device, fifo, + * or stdout, pad the last block, otherwise leave it + * unpadded. */ + if (S_ISCHR(st.st_mode) || S_ISBLK(st.st_mode) || + S_ISFIFO(st.st_mode) || (mine->fd == 1)) + /* Last block will be fully padded. */ + archive_write_set_bytes_in_last_block(a, 0); + else + archive_write_set_bytes_in_last_block(a, 1); + } + + return (ARCHIVE_OK); +} + +static ssize_t +file_write(struct archive *a, void *client_data, const void *buff, size_t length) +{ + struct write_fd_data *mine; + ssize_t bytesWritten; + + mine = (struct write_fd_data *)client_data; + for (;;) { + bytesWritten = write(mine->fd, buff, length); + if (bytesWritten <= 0) { + if (errno == EINTR) + continue; + archive_set_error(a, errno, "Write error"); + return (-1); + } + return (bytesWritten); + } +} + +static int +file_close(struct archive *a, void *client_data) +{ + struct write_fd_data *mine = (struct write_fd_data *)client_data; + + (void)a; /* UNUSED */ + free(mine); + return (ARCHIVE_OK); +} diff --git a/c/archive_write_open_file.c b/c/archive_write_open_file.c new file mode 100644 --- /dev/null +++ b/c/archive_write_open_file.c @@ -0,0 +1,109 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: src/lib/libarchive/archive_write_open_file.c,v 1.19 2007/01/09 08:05:56 kientzle Exp $"); + +#ifdef HAVE_SYS_STAT_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_FCNTL_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif + +#include "archive.h" + +struct write_FILE_data { + FILE *f; +}; + +static int file_close(struct archive *, void *); +static int file_open(struct archive *, void *); +static ssize_t file_write(struct archive *, void *, const void *buff, size_t); + +int +archive_write_open_FILE(struct archive *a, FILE *f) +{ + struct write_FILE_data *mine; + + mine = (struct write_FILE_data *)malloc(sizeof(*mine)); + if (mine == NULL) { + archive_set_error(a, ENOMEM, "No memory"); + return (ARCHIVE_FATAL); + } + mine->f = f; + return (archive_write_open(a, mine, + file_open, file_write, file_close)); +} + +static int +file_open(struct archive *a, void *client_data) +{ + (void)a; /* UNUSED */ + (void)client_data; /* UNUSED */ + + return (ARCHIVE_OK); +} + +static ssize_t +file_write(struct archive *a, void *client_data, const void *buff, size_t length) +{ + struct write_FILE_data *mine; + size_t bytesWritten; + + mine = client_data; + for (;;) { + bytesWritten = fwrite(buff, 1, length, mine->f); + if (bytesWritten <= 0) { + if (errno == EINTR) + continue; + archive_set_error(a, errno, "Write error"); + return (-1); + } + return (bytesWritten); + } +} + +static int +file_close(struct archive *a, void *client_data) +{ + struct write_FILE_data *mine = client_data; + + (void)a; /* UNUSED */ + free(mine); + return (ARCHIVE_OK); +} diff --git a/c/archive_write_open_filename.c b/c/archive_write_open_filename.c new file mode 100644 --- /dev/null +++ b/c/archive_write_open_filename.c @@ -0,0 +1,253 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_open_filename.c 191165 2009-04-17 00:39:35Z kientzle $"); + +#ifdef HAVE_SYS_STAT_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_FCNTL_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_string.h" + +#ifndef O_BINARY +#define O_BINARY 0 +#endif +#ifndef O_CLOEXEC +#define O_CLOEXEC 0 +#endif + +struct write_file_data { + int fd; + struct archive_mstring filename; +}; + +static int file_close(struct archive *, void *); +static int file_open(struct archive *, void *); +static ssize_t file_write(struct archive *, void *, const void *buff, size_t); +static int open_filename(struct archive *, int, const void *); + +int +archive_write_open_file(struct archive *a, const char *filename) +{ + return (archive_write_open_filename(a, filename)); +} + +int +archive_write_open_filename(struct archive *a, const char *filename) +{ + + if (filename == NULL || filename[0] == '\0') + return (archive_write_open_fd(a, 1)); + + return (open_filename(a, 1, filename)); +} + +int +archive_write_open_filename_w(struct archive *a, const wchar_t *filename) +{ + + if (filename == NULL || filename[0] == L'\0') + return (archive_write_open_fd(a, 1)); + + return (open_filename(a, 0, filename)); +} + +static int +open_filename(struct archive *a, int mbs_fn, const void *filename) +{ + struct write_file_data *mine; + int r; + + mine = (struct write_file_data *)calloc(1, sizeof(*mine)); + if (mine == NULL) { + archive_set_error(a, ENOMEM, "No memory"); + return (ARCHIVE_FATAL); + } + if (mbs_fn) + r = archive_mstring_copy_mbs(&mine->filename, filename); + else + r = archive_mstring_copy_wcs(&mine->filename, filename); + if (r < 0) { + if (errno == ENOMEM) { + archive_set_error(a, ENOMEM, "No memory"); + return (ARCHIVE_FATAL); + } + if (mbs_fn) + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Can't convert '%s' to WCS", + (const char *)filename); + else + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Can't convert '%S' to MBS", + (const wchar_t *)filename); + return (ARCHIVE_FAILED); + } + mine->fd = -1; + return (archive_write_open(a, mine, + file_open, file_write, file_close)); +} + +static int +file_open(struct archive *a, void *client_data) +{ + int flags; + struct write_file_data *mine; + struct stat st; +#if defined(_WIN32) && !defined(__CYGWIN__) + wchar_t *fullpath; +#endif + const wchar_t *wcs; + const char *mbs; + + mine = (struct write_file_data *)client_data; + flags = O_WRONLY | O_CREAT | O_TRUNC | O_BINARY | O_CLOEXEC; + + /* + * Open the file. + */ + mbs = NULL; wcs = NULL; +#if defined(_WIN32) && !defined(__CYGWIN__) + if (archive_mstring_get_wcs(a, &mine->filename, &wcs) != 0) { + if (errno == ENOMEM) + archive_set_error(a, errno, "No memory"); + else { + archive_mstring_get_mbs(a, &mine->filename, &mbs); + archive_set_error(a, errno, + "Can't convert '%s' to WCS", mbs); + } + return (ARCHIVE_FATAL); + } + fullpath = __la_win_permissive_name_w(wcs); + if (fullpath != NULL) { + mine->fd = _wopen(fullpath, flags, 0666); + free(fullpath); + } else + mine->fd = _wopen(wcs, flags, 0666); +#else + if (archive_mstring_get_mbs(a, &mine->filename, &mbs) != 0) { + if (errno == ENOMEM) + archive_set_error(a, errno, "No memory"); + else { + archive_mstring_get_wcs(a, &mine->filename, &wcs); + archive_set_error(a, errno, + "Can't convert '%S' to MBS", wcs); + } + return (ARCHIVE_FATAL); + } + mine->fd = open(mbs, flags, 0666); + __archive_ensure_cloexec_flag(mine->fd); +#endif + if (mine->fd < 0) { + if (mbs != NULL) + archive_set_error(a, errno, "Failed to open '%s'", mbs); + else + archive_set_error(a, errno, "Failed to open '%S'", wcs); + return (ARCHIVE_FATAL); + } + + if (fstat(mine->fd, &st) != 0) { + if (mbs != NULL) + archive_set_error(a, errno, "Couldn't stat '%s'", mbs); + else + archive_set_error(a, errno, "Couldn't stat '%S'", wcs); + return (ARCHIVE_FATAL); + } + + /* + * Set up default last block handling. + */ + if (archive_write_get_bytes_in_last_block(a) < 0) { + if (S_ISCHR(st.st_mode) || S_ISBLK(st.st_mode) || + S_ISFIFO(st.st_mode)) + /* Pad last block when writing to device or FIFO. */ + archive_write_set_bytes_in_last_block(a, 0); + else + /* Don't pad last block otherwise. */ + archive_write_set_bytes_in_last_block(a, 1); + } + + /* + * If the output file is a regular file, don't add it to + * itself. If it's a device file, it's okay to add the device + * entry to the output archive. + */ + if (S_ISREG(st.st_mode)) + archive_write_set_skip_file(a, st.st_dev, st.st_ino); + + return (ARCHIVE_OK); +} + +static ssize_t +file_write(struct archive *a, void *client_data, const void *buff, + size_t length) +{ + struct write_file_data *mine; + ssize_t bytesWritten; + + mine = (struct write_file_data *)client_data; + for (;;) { + bytesWritten = write(mine->fd, buff, length); + if (bytesWritten <= 0) { + if (errno == EINTR) + continue; + archive_set_error(a, errno, "Write error"); + return (-1); + } + return (bytesWritten); + } +} + +static int +file_close(struct archive *a, void *client_data) +{ + struct write_file_data *mine = (struct write_file_data *)client_data; + + (void)a; /* UNUSED */ + + if (mine->fd >= 0) + close(mine->fd); + + archive_mstring_clean(&mine->filename); + free(mine); + return (ARCHIVE_OK); +} diff --git a/c/archive_write_open_memory.c b/c/archive_write_open_memory.c new file mode 100644 --- /dev/null +++ b/c/archive_write_open_memory.c @@ -0,0 +1,113 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: src/lib/libarchive/archive_write_open_memory.c,v 1.3 2007/01/09 08:05:56 kientzle Exp $"); + +#include +#include +#include + +#include "archive.h" + +struct write_memory_data { + size_t used; + size_t size; + size_t * client_size; + unsigned char * buff; +}; + +static int memory_write_close(struct archive *, void *); +static int memory_write_open(struct archive *, void *); +static ssize_t memory_write(struct archive *, void *, const void *buff, size_t); + +/* + * Client provides a pointer to a block of memory to receive + * the data. The 'size' param both tells us the size of the + * client buffer and lets us tell the client the final size. + */ +int +archive_write_open_memory(struct archive *a, void *buff, size_t buffSize, size_t *used) +{ + struct write_memory_data *mine; + + mine = (struct write_memory_data *)calloc(1, sizeof(*mine)); + if (mine == NULL) { + archive_set_error(a, ENOMEM, "No memory"); + return (ARCHIVE_FATAL); + } + mine->buff = buff; + mine->size = buffSize; + mine->client_size = used; + return (archive_write_open(a, mine, + memory_write_open, memory_write, memory_write_close)); +} + +static int +memory_write_open(struct archive *a, void *client_data) +{ + struct write_memory_data *mine; + mine = client_data; + mine->used = 0; + if (mine->client_size != NULL) + *mine->client_size = mine->used; + /* Disable padding if it hasn't been set explicitly. */ + if (-1 == archive_write_get_bytes_in_last_block(a)) + archive_write_set_bytes_in_last_block(a, 1); + return (ARCHIVE_OK); +} + +/* + * Copy the data into the client buffer. + * Note that we update mine->client_size on every write. + * In particular, this means the client can follow exactly + * how much has been written into their buffer at any time. + */ +static ssize_t +memory_write(struct archive *a, void *client_data, const void *buff, size_t length) +{ + struct write_memory_data *mine; + mine = client_data; + + if (mine->used + length > mine->size) { + archive_set_error(a, ENOMEM, "Buffer exhausted"); + return (ARCHIVE_FATAL); + } + memcpy(mine->buff + mine->used, buff, length); + mine->used += length; + if (mine->client_size != NULL) + *mine->client_size = mine->used; + return (length); +} + +static int +memory_write_close(struct archive *a, void *client_data) +{ + struct write_memory_data *mine; + (void)a; /* UNUSED */ + mine = client_data; + free(mine); + return (ARCHIVE_OK); +} diff --git a/c/archive_write_private.h b/c/archive_write_private.h new file mode 100644 --- /dev/null +++ b/c/archive_write_private.h @@ -0,0 +1,164 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD: head/lib/libarchive/archive_write_private.h 201155 2009-12-29 05:20:12Z kientzle $ + */ + +#ifndef ARCHIVE_WRITE_PRIVATE_H_INCLUDED +#define ARCHIVE_WRITE_PRIVATE_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#ifndef __LIBARCHIVE_TEST +#error This header is only to be used internally to libarchive. +#endif +#endif + +#include "archive.h" +#include "archive_string.h" +#include "archive_private.h" + +#define ARCHIVE_WRITE_FILTER_STATE_NEW 1U +#define ARCHIVE_WRITE_FILTER_STATE_OPEN 2U +#define ARCHIVE_WRITE_FILTER_STATE_CLOSED 4U +#define ARCHIVE_WRITE_FILTER_STATE_FATAL 0x8000U + +struct archive_write; + +struct archive_write_filter { + int64_t bytes_written; + struct archive *archive; /* Associated archive. */ + struct archive_write_filter *next_filter; /* Who I write to. */ + int (*options)(struct archive_write_filter *, + const char *key, const char *value); + int (*open)(struct archive_write_filter *); + int (*write)(struct archive_write_filter *, const void *, size_t); + int (*close)(struct archive_write_filter *); + int (*free)(struct archive_write_filter *); + void *data; + const char *name; + int code; + int bytes_per_block; + int bytes_in_last_block; + int state; +}; + +#if ARCHIVE_VERSION < 4000000 +void __archive_write_filters_free(struct archive *); +#endif + +struct archive_write_filter *__archive_write_allocate_filter(struct archive *); + +int __archive_write_output(struct archive_write *, const void *, size_t); +int __archive_write_nulls(struct archive_write *, size_t); +int __archive_write_filter(struct archive_write_filter *, const void *, size_t); + +struct archive_write { + struct archive archive; + + /* Dev/ino of the archive being written. */ + int skip_file_set; + int64_t skip_file_dev; + int64_t skip_file_ino; + + /* Utility: Pointer to a block of nulls. */ + const unsigned char *nulls; + size_t null_length; + + /* Callbacks to open/read/write/close archive stream. */ + archive_open_callback *client_opener; + archive_write_callback *client_writer; + archive_close_callback *client_closer; + void *client_data; + + /* + * Blocking information. Note that bytes_in_last_block is + * misleadingly named; I should find a better name. These + * control the final output from all compressors, including + * compression_none. + */ + int bytes_per_block; + int bytes_in_last_block; + + /* + * First and last write filters in the pipeline. + */ + struct archive_write_filter *filter_first; + struct archive_write_filter *filter_last; + + /* + * Pointers to format-specific functions for writing. They're + * initialized by archive_write_set_format_XXX() calls. + */ + void *format_data; + const char *format_name; + int (*format_init)(struct archive_write *); + int (*format_options)(struct archive_write *, + const char *key, const char *value); + int (*format_finish_entry)(struct archive_write *); + int (*format_write_header)(struct archive_write *, + struct archive_entry *); + ssize_t (*format_write_data)(struct archive_write *, + const void *buff, size_t); + int (*format_close)(struct archive_write *); + int (*format_free)(struct archive_write *); + + + /* + * Encryption passphrase. + */ + char *passphrase; + archive_passphrase_callback *passphrase_callback; + void *passphrase_client_data; +}; + +/* + * Utility function to format a USTAR header into a buffer. If + * "strict" is set, this tries to create the absolutely most portable + * version of a ustar header. If "strict" is set to 0, then it will + * relax certain requirements. + * + * Generally, format-specific declarations don't belong in this + * header; this is a rare example of a function that is shared by + * two very similar formats (ustar and pax). + */ +int +__archive_write_format_header_ustar(struct archive_write *, char buff[512], + struct archive_entry *, int tartype, int strict, + struct archive_string_conv *); + +struct archive_write_program_data; +struct archive_write_program_data * __archive_write_program_allocate(const char *program_name); +int __archive_write_program_free(struct archive_write_program_data *); +int __archive_write_program_open(struct archive_write_filter *, + struct archive_write_program_data *, const char *); +int __archive_write_program_close(struct archive_write_filter *, + struct archive_write_program_data *); +int __archive_write_program_write(struct archive_write_filter *, + struct archive_write_program_data *, const void *, size_t); + +/* + * Get a encryption passphrase. + */ +const char * __archive_write_get_passphrase(struct archive_write *a); +#endif diff --git a/c/archive_write_set_format.c b/c/archive_write_set_format.c new file mode 100644 --- /dev/null +++ b/c/archive_write_set_format.c @@ -0,0 +1,123 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_format.c 201168 2009-12-29 06:15:32Z kientzle $"); + +#ifdef HAVE_SYS_TYPES_H +#include +#endif + +#ifdef HAVE_ERRNO_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" +#include "archive_write_set_format_private.h" + +/* A table that maps format codes to functions. */ +static const +struct { int code; int (*setter)(struct archive *); } codes[] = +{ + { ARCHIVE_FORMAT_7ZIP, archive_write_set_format_7zip }, + { ARCHIVE_FORMAT_CPIO, archive_write_set_format_cpio }, + { ARCHIVE_FORMAT_CPIO_POSIX, archive_write_set_format_cpio }, + { ARCHIVE_FORMAT_CPIO_SVR4_NOCRC, archive_write_set_format_cpio_newc }, + { ARCHIVE_FORMAT_ISO9660, archive_write_set_format_iso9660 }, + { ARCHIVE_FORMAT_MTREE, archive_write_set_format_mtree }, + { ARCHIVE_FORMAT_RAW, archive_write_set_format_raw }, + { ARCHIVE_FORMAT_SHAR, archive_write_set_format_shar }, + { ARCHIVE_FORMAT_SHAR_BASE, archive_write_set_format_shar }, + { ARCHIVE_FORMAT_SHAR_DUMP, archive_write_set_format_shar_dump }, + { ARCHIVE_FORMAT_TAR, archive_write_set_format_pax_restricted }, + { ARCHIVE_FORMAT_TAR_GNUTAR, archive_write_set_format_gnutar }, + { ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE, archive_write_set_format_pax }, + { ARCHIVE_FORMAT_TAR_PAX_RESTRICTED, + archive_write_set_format_pax_restricted }, + { ARCHIVE_FORMAT_TAR_USTAR, archive_write_set_format_ustar }, + { ARCHIVE_FORMAT_WARC, archive_write_set_format_warc }, + { ARCHIVE_FORMAT_XAR, archive_write_set_format_xar }, + { ARCHIVE_FORMAT_ZIP, archive_write_set_format_zip }, + { 0, NULL } +}; + +int +archive_write_set_format(struct archive *a, int code) +{ + int i; + + for (i = 0; codes[i].code != 0; i++) { + if (code == codes[i].code) + return ((codes[i].setter)(a)); + } + + archive_set_error(a, EINVAL, "No such format"); + return (ARCHIVE_FATAL); +} + +void +__archive_write_entry_filetype_unsupported(struct archive *a, + struct archive_entry *entry, const char *format) +{ + const char *name = NULL; + + switch (archive_entry_filetype(entry)) { + /* + * All formats should be able to archive regular files (AE_IFREG) + */ + case AE_IFDIR: + name = "directories"; + break; + case AE_IFLNK: + name = "symbolic links"; + break; + case AE_IFCHR: + name = "character devices"; + break; + case AE_IFBLK: + name = "block devices"; + break; + case AE_IFIFO: + name = "named pipes"; + break; + case AE_IFSOCK: + name = "sockets"; + break; + default: + break; + } + + if (name != NULL) { + archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT, + "%s: %s format cannot archive %s", + archive_entry_pathname(entry), format, name); + } else { + archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT, + "%s: %s format cannot archive files with mode 0%lo", + archive_entry_pathname(entry), format, + (unsigned long)archive_entry_mode(entry)); + } +} diff --git a/c/archive_write_set_format_7zip.c b/c/archive_write_set_format_7zip.c new file mode 100644 --- /dev/null +++ b/c/archive_write_set_format_7zip.c @@ -0,0 +1,2317 @@ +/*- + * Copyright (c) 2011-2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#include +#ifdef HAVE_BZLIB_H +#include +#endif +#if HAVE_LZMA_H +#include +#endif +#ifdef HAVE_ZLIB_H +#include +#endif + +#include "archive.h" +#ifndef HAVE_ZLIB_H +#include "archive_crc32.h" +#endif +#include "archive_endian.h" +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_ppmd7_private.h" +#include "archive_private.h" +#include "archive_rb.h" +#include "archive_string.h" +#include "archive_write_private.h" +#include "archive_write_set_format_private.h" + +/* + * Codec ID + */ +#define _7Z_COPY 0 +#define _7Z_LZMA1 0x030101 +#define _7Z_LZMA2 0x21 +#define _7Z_DEFLATE 0x040108 +#define _7Z_BZIP2 0x040202 +#define _7Z_PPMD 0x030401 + +/* + * 7-Zip header property IDs. + */ +#define kEnd 0x00 +#define kHeader 0x01 +#define kArchiveProperties 0x02 +#define kAdditionalStreamsInfo 0x03 +#define kMainStreamsInfo 0x04 +#define kFilesInfo 0x05 +#define kPackInfo 0x06 +#define kUnPackInfo 0x07 +#define kSubStreamsInfo 0x08 +#define kSize 0x09 +#define kCRC 0x0A +#define kFolder 0x0B +#define kCodersUnPackSize 0x0C +#define kNumUnPackStream 0x0D +#define kEmptyStream 0x0E +#define kEmptyFile 0x0F +#define kAnti 0x10 +#define kName 0x11 +#define kCTime 0x12 +#define kATime 0x13 +#define kMTime 0x14 +#define kAttributes 0x15 +#define kEncodedHeader 0x17 + +enum la_zaction { + ARCHIVE_Z_FINISH, + ARCHIVE_Z_RUN +}; + +/* + * A stream object of universal compressor. + */ +struct la_zstream { + const uint8_t *next_in; + size_t avail_in; + uint64_t total_in; + + uint8_t *next_out; + size_t avail_out; + uint64_t total_out; + + uint32_t prop_size; + uint8_t *props; + + int valid; + void *real_stream; + int (*code) (struct archive *a, + struct la_zstream *lastrm, + enum la_zaction action); + int (*end)(struct archive *a, + struct la_zstream *lastrm); +}; + +#define PPMD7_DEFAULT_ORDER 6 +#define PPMD7_DEFAULT_MEM_SIZE (1 << 24) + +struct ppmd_stream { + int stat; + CPpmd7 ppmd7_context; + CPpmd7z_RangeEnc range_enc; + IByteOut byteout; + uint8_t *buff; + uint8_t *buff_ptr; + uint8_t *buff_end; + size_t buff_bytes; +}; + +struct coder { + unsigned codec; + size_t prop_size; + uint8_t *props; +}; + +struct file { + struct archive_rb_node rbnode; + + struct file *next; + unsigned name_len; + uint8_t *utf16name;/* UTF16-LE name. */ + uint64_t size; + unsigned flg; +#define MTIME_IS_SET (1<<0) +#define ATIME_IS_SET (1<<1) +#define CTIME_IS_SET (1<<2) +#define CRC32_IS_SET (1<<3) +#define HAS_STREAM (1<<4) + + struct { + time_t time; + long time_ns; + } times[3]; +#define MTIME 0 +#define ATIME 1 +#define CTIME 2 + + mode_t mode; + uint32_t crc32; + + signed int dir:1; +}; + +struct _7zip { + int temp_fd; + uint64_t temp_offset; + + struct file *cur_file; + size_t total_number_entry; + size_t total_number_nonempty_entry; + size_t total_number_empty_entry; + size_t total_number_dir_entry; + size_t total_bytes_entry_name; + size_t total_number_time_defined[3]; + uint64_t total_bytes_compressed; + uint64_t total_bytes_uncompressed; + uint64_t entry_bytes_remaining; + uint32_t entry_crc32; + uint32_t precode_crc32; + uint32_t encoded_crc32; + int crc32flg; +#define PRECODE_CRC32 1 +#define ENCODED_CRC32 2 + + unsigned opt_compression; + int opt_compression_level; + + struct la_zstream stream; + struct coder coder; + + struct archive_string_conv *sconv; + + /* + * Compressed data buffer. + */ + unsigned char wbuff[512 * 20 * 6]; + size_t wbuff_remaining; + + /* + * The list of the file entries which has its contents is used to + * manage struct file objects. + * We use 'next' (a member of struct file) to chain. + */ + struct { + struct file *first; + struct file **last; + } file_list, empty_list; + struct archive_rb_tree rbtree;/* for empty files */ +}; + +static int _7z_options(struct archive_write *, + const char *, const char *); +static int _7z_write_header(struct archive_write *, + struct archive_entry *); +static ssize_t _7z_write_data(struct archive_write *, + const void *, size_t); +static int _7z_finish_entry(struct archive_write *); +static int _7z_close(struct archive_write *); +static int _7z_free(struct archive_write *); +static int file_cmp_node(const struct archive_rb_node *, + const struct archive_rb_node *); +static int file_cmp_key(const struct archive_rb_node *, const void *); +static int file_new(struct archive_write *a, struct archive_entry *, + struct file **); +static void file_free(struct file *); +static void file_register(struct _7zip *, struct file *); +static void file_register_empty(struct _7zip *, struct file *); +static void file_init_register(struct _7zip *); +static void file_init_register_empty(struct _7zip *); +static void file_free_register(struct _7zip *); +static ssize_t compress_out(struct archive_write *, const void *, size_t , + enum la_zaction); +static int compression_init_encoder_copy(struct archive *, + struct la_zstream *); +static int compression_code_copy(struct archive *, + struct la_zstream *, enum la_zaction); +static int compression_end_copy(struct archive *, struct la_zstream *); +static int compression_init_encoder_deflate(struct archive *, + struct la_zstream *, int, int); +#ifdef HAVE_ZLIB_H +static int compression_code_deflate(struct archive *, + struct la_zstream *, enum la_zaction); +static int compression_end_deflate(struct archive *, struct la_zstream *); +#endif +static int compression_init_encoder_bzip2(struct archive *, + struct la_zstream *, int); +#if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) +static int compression_code_bzip2(struct archive *, + struct la_zstream *, enum la_zaction); +static int compression_end_bzip2(struct archive *, struct la_zstream *); +#endif +static int compression_init_encoder_lzma1(struct archive *, + struct la_zstream *, int); +static int compression_init_encoder_lzma2(struct archive *, + struct la_zstream *, int); +#if defined(HAVE_LZMA_H) +static int compression_code_lzma(struct archive *, + struct la_zstream *, enum la_zaction); +static int compression_end_lzma(struct archive *, struct la_zstream *); +#endif +static int compression_init_encoder_ppmd(struct archive *, + struct la_zstream *, unsigned, uint32_t); +static int compression_code_ppmd(struct archive *, + struct la_zstream *, enum la_zaction); +static int compression_end_ppmd(struct archive *, struct la_zstream *); +static int _7z_compression_init_encoder(struct archive_write *, unsigned, + int); +static int compression_code(struct archive *, + struct la_zstream *, enum la_zaction); +static int compression_end(struct archive *, + struct la_zstream *); +static int enc_uint64(struct archive_write *, uint64_t); +static int make_header(struct archive_write *, uint64_t, uint64_t, + uint64_t, int, struct coder *); +static int make_streamsInfo(struct archive_write *, uint64_t, uint64_t, + uint64_t, int, struct coder *, int, uint32_t); + +int +archive_write_set_format_7zip(struct archive *_a) +{ + static const struct archive_rb_tree_ops rb_ops = { + file_cmp_node, file_cmp_key + }; + struct archive_write *a = (struct archive_write *)_a; + struct _7zip *zip; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_set_format_7zip"); + + /* If another format was already registered, unregister it. */ + if (a->format_free != NULL) + (a->format_free)(a); + + zip = calloc(1, sizeof(*zip)); + if (zip == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate 7-Zip data"); + return (ARCHIVE_FATAL); + } + zip->temp_fd = -1; + __archive_rb_tree_init(&(zip->rbtree), &rb_ops); + file_init_register(zip); + file_init_register_empty(zip); + + /* Set default compression type and its level. */ +#if HAVE_LZMA_H + zip->opt_compression = _7Z_LZMA1; +#elif defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) + zip->opt_compression = _7Z_BZIP2; +#elif defined(HAVE_ZLIB_H) + zip->opt_compression = _7Z_DEFLATE; +#else + zip->opt_compression = _7Z_COPY; +#endif + zip->opt_compression_level = 6; + + a->format_data = zip; + + a->format_name = "7zip"; + a->format_options = _7z_options; + a->format_write_header = _7z_write_header; + a->format_write_data = _7z_write_data; + a->format_finish_entry = _7z_finish_entry; + a->format_close = _7z_close; + a->format_free = _7z_free; + a->archive.archive_format = ARCHIVE_FORMAT_7ZIP; + a->archive.archive_format_name = "7zip"; + + return (ARCHIVE_OK); +} + +static int +_7z_options(struct archive_write *a, const char *key, const char *value) +{ + struct _7zip *zip; + + zip = (struct _7zip *)a->format_data; + + if (strcmp(key, "compression") == 0) { + const char *name = NULL; + + if (value == NULL || strcmp(value, "copy") == 0 || + strcmp(value, "COPY") == 0 || + strcmp(value, "store") == 0 || + strcmp(value, "STORE") == 0) + zip->opt_compression = _7Z_COPY; + else if (strcmp(value, "deflate") == 0 || + strcmp(value, "DEFLATE") == 0) +#if HAVE_ZLIB_H + zip->opt_compression = _7Z_DEFLATE; +#else + name = "deflate"; +#endif + else if (strcmp(value, "bzip2") == 0 || + strcmp(value, "BZIP2") == 0) +#if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) + zip->opt_compression = _7Z_BZIP2; +#else + name = "bzip2"; +#endif + else if (strcmp(value, "lzma1") == 0 || + strcmp(value, "LZMA1") == 0) +#if HAVE_LZMA_H + zip->opt_compression = _7Z_LZMA1; +#else + name = "lzma1"; +#endif + else if (strcmp(value, "lzma2") == 0 || + strcmp(value, "LZMA2") == 0) +#if HAVE_LZMA_H + zip->opt_compression = _7Z_LZMA2; +#else + name = "lzma2"; +#endif + else if (strcmp(value, "ppmd") == 0 || + strcmp(value, "PPMD") == 0 || + strcmp(value, "PPMd") == 0) + zip->opt_compression = _7Z_PPMD; + else { + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "Unknown compression name: `%s'", + value); + return (ARCHIVE_FAILED); + } + if (name != NULL) { + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "`%s' compression not supported " + "on this platform", + name); + return (ARCHIVE_FAILED); + } + return (ARCHIVE_OK); + } + if (strcmp(key, "compression-level") == 0) { + if (value == NULL || + !(value[0] >= '0' && value[0] <= '9') || + value[1] != '\0') { + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "Illegal value `%s'", + value); + return (ARCHIVE_FAILED); + } + zip->opt_compression_level = value[0] - '0'; + return (ARCHIVE_OK); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +static int +_7z_write_header(struct archive_write *a, struct archive_entry *entry) +{ + struct _7zip *zip; + struct file *file; + int r; + + zip = (struct _7zip *)a->format_data; + zip->cur_file = NULL; + zip->entry_bytes_remaining = 0; + + if (zip->sconv == NULL) { + zip->sconv = archive_string_conversion_to_charset( + &a->archive, "UTF-16LE", 1); + if (zip->sconv == NULL) + return (ARCHIVE_FATAL); + } + + r = file_new(a, entry, &file); + if (r < ARCHIVE_WARN) { + if (file != NULL) + file_free(file); + return (r); + } + if (file->size == 0 && file->dir) { + if (!__archive_rb_tree_insert_node(&(zip->rbtree), + (struct archive_rb_node *)file)) { + /* We have already had the same file. */ + file_free(file); + return (ARCHIVE_OK); + } + } + + if (file->flg & MTIME_IS_SET) + zip->total_number_time_defined[MTIME]++; + if (file->flg & CTIME_IS_SET) + zip->total_number_time_defined[CTIME]++; + if (file->flg & ATIME_IS_SET) + zip->total_number_time_defined[ATIME]++; + + zip->total_number_entry++; + zip->total_bytes_entry_name += file->name_len + 2; + if (file->size == 0) { + /* Count up the number of empty files. */ + zip->total_number_empty_entry++; + if (file->dir) + zip->total_number_dir_entry++; + else + file_register_empty(zip, file); + return (r); + } + + /* + * Init compression. + */ + if ((zip->total_number_entry - zip->total_number_empty_entry) == 1) { + r = _7z_compression_init_encoder(a, zip->opt_compression, + zip->opt_compression_level); + if (r < 0) { + file_free(file); + return (ARCHIVE_FATAL); + } + } + + /* Register a non-empty file. */ + file_register(zip, file); + + /* + * Set the current file to cur_file to read its contents. + */ + zip->cur_file = file; + + + /* Save a offset of current file in temporary file. */ + zip->entry_bytes_remaining = file->size; + zip->entry_crc32 = 0; + + /* + * Store a symbolic link name as file contents. + */ + if (archive_entry_filetype(entry) == AE_IFLNK) { + ssize_t bytes; + const void *p = (const void *)archive_entry_symlink(entry); + bytes = compress_out(a, p, (size_t)file->size, ARCHIVE_Z_RUN); + if (bytes < 0) + return ((int)bytes); + zip->entry_crc32 = crc32(zip->entry_crc32, p, (unsigned)bytes); + zip->entry_bytes_remaining -= bytes; + } + + return (r); +} + +/* + * Write data to a temporary file. + */ +static int +write_to_temp(struct archive_write *a, const void *buff, size_t s) +{ + struct _7zip *zip; + const unsigned char *p; + ssize_t ws; + + zip = (struct _7zip *)a->format_data; + + /* + * Open a temporary file. + */ + if (zip->temp_fd == -1) { + zip->temp_offset = 0; + zip->temp_fd = __archive_mktemp(NULL); + if (zip->temp_fd < 0) { + archive_set_error(&a->archive, errno, + "Couldn't create temporary file"); + return (ARCHIVE_FATAL); + } + } + + p = (const unsigned char *)buff; + while (s) { + ws = write(zip->temp_fd, p, s); + if (ws < 0) { + archive_set_error(&(a->archive), errno, + "fwrite function failed"); + return (ARCHIVE_FATAL); + } + s -= ws; + p += ws; + zip->temp_offset += ws; + } + return (ARCHIVE_OK); +} + +static ssize_t +compress_out(struct archive_write *a, const void *buff, size_t s, + enum la_zaction run) +{ + struct _7zip *zip = (struct _7zip *)a->format_data; + int r; + + if (run == ARCHIVE_Z_FINISH && zip->stream.total_in == 0 && s == 0) + return (0); + + if ((zip->crc32flg & PRECODE_CRC32) && s) + zip->precode_crc32 = crc32(zip->precode_crc32, buff, + (unsigned)s); + zip->stream.next_in = (const unsigned char *)buff; + zip->stream.avail_in = s; + for (;;) { + /* Compress file data. */ + r = compression_code(&(a->archive), &(zip->stream), run); + if (r != ARCHIVE_OK && r != ARCHIVE_EOF) + return (ARCHIVE_FATAL); + if (zip->stream.avail_out == 0) { + if (write_to_temp(a, zip->wbuff, sizeof(zip->wbuff)) + != ARCHIVE_OK) + return (ARCHIVE_FATAL); + zip->stream.next_out = zip->wbuff; + zip->stream.avail_out = sizeof(zip->wbuff); + if (zip->crc32flg & ENCODED_CRC32) + zip->encoded_crc32 = crc32(zip->encoded_crc32, + zip->wbuff, sizeof(zip->wbuff)); + if (run == ARCHIVE_Z_FINISH && r != ARCHIVE_EOF) + continue; + } + if (zip->stream.avail_in == 0) + break; + } + if (run == ARCHIVE_Z_FINISH) { + uint64_t bytes = sizeof(zip->wbuff) - zip->stream.avail_out; + if (write_to_temp(a, zip->wbuff, (size_t)bytes) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + if ((zip->crc32flg & ENCODED_CRC32) && bytes) + zip->encoded_crc32 = crc32(zip->encoded_crc32, + zip->wbuff, (unsigned)bytes); + } + + return (s); +} + +static ssize_t +_7z_write_data(struct archive_write *a, const void *buff, size_t s) +{ + struct _7zip *zip; + ssize_t bytes; + + zip = (struct _7zip *)a->format_data; + + if (s > zip->entry_bytes_remaining) + s = (size_t)zip->entry_bytes_remaining; + if (s == 0 || zip->cur_file == NULL) + return (0); + bytes = compress_out(a, buff, s, ARCHIVE_Z_RUN); + if (bytes < 0) + return (bytes); + zip->entry_crc32 = crc32(zip->entry_crc32, buff, (unsigned)bytes); + zip->entry_bytes_remaining -= bytes; + return (bytes); +} + +static int +_7z_finish_entry(struct archive_write *a) +{ + struct _7zip *zip; + size_t s; + ssize_t r; + + zip = (struct _7zip *)a->format_data; + if (zip->cur_file == NULL) + return (ARCHIVE_OK); + + while (zip->entry_bytes_remaining > 0) { + s = (size_t)zip->entry_bytes_remaining; + if (s > a->null_length) + s = a->null_length; + r = _7z_write_data(a, a->nulls, s); + if (r < 0) + return ((int)r); + } + zip->total_bytes_compressed += zip->stream.total_in; + zip->total_bytes_uncompressed += zip->stream.total_out; + zip->cur_file->crc32 = zip->entry_crc32; + zip->cur_file = NULL; + + return (ARCHIVE_OK); +} + +static int +flush_wbuff(struct archive_write *a) +{ + struct _7zip *zip; + int r; + size_t s; + + zip = (struct _7zip *)a->format_data; + s = sizeof(zip->wbuff) - zip->wbuff_remaining; + r = __archive_write_output(a, zip->wbuff, s); + if (r != ARCHIVE_OK) + return (r); + zip->wbuff_remaining = sizeof(zip->wbuff); + return (r); +} + +static int +copy_out(struct archive_write *a, uint64_t offset, uint64_t length) +{ + struct _7zip *zip; + int r; + + zip = (struct _7zip *)a->format_data; + if (zip->temp_offset > 0 && + lseek(zip->temp_fd, offset, SEEK_SET) < 0) { + archive_set_error(&(a->archive), errno, "lseek failed"); + return (ARCHIVE_FATAL); + } + while (length) { + size_t rsize; + ssize_t rs; + unsigned char *wb; + + if (length > zip->wbuff_remaining) + rsize = zip->wbuff_remaining; + else + rsize = (size_t)length; + wb = zip->wbuff + (sizeof(zip->wbuff) - zip->wbuff_remaining); + rs = read(zip->temp_fd, wb, rsize); + if (rs < 0) { + archive_set_error(&(a->archive), errno, + "Can't read temporary file(%jd)", + (intmax_t)rs); + return (ARCHIVE_FATAL); + } + if (rs == 0) { + archive_set_error(&(a->archive), 0, + "Truncated 7-Zip archive"); + return (ARCHIVE_FATAL); + } + zip->wbuff_remaining -= rs; + length -= rs; + if (zip->wbuff_remaining == 0) { + r = flush_wbuff(a); + if (r != ARCHIVE_OK) + return (r); + } + } + return (ARCHIVE_OK); +} + +static int +_7z_close(struct archive_write *a) +{ + struct _7zip *zip; + unsigned char *wb; + uint64_t header_offset, header_size, header_unpacksize; + uint64_t length; + uint32_t header_crc32; + int r; + + zip = (struct _7zip *)a->format_data; + + if (zip->total_number_entry > 0) { + struct archive_rb_node *n; + uint64_t data_offset, data_size, data_unpacksize; + unsigned header_compression; + + r = (int)compress_out(a, NULL, 0, ARCHIVE_Z_FINISH); + if (r < 0) + return (r); + data_offset = 0; + data_size = zip->stream.total_out; + data_unpacksize = zip->stream.total_in; + zip->coder.codec = zip->opt_compression; + zip->coder.prop_size = zip->stream.prop_size; + zip->coder.props = zip->stream.props; + zip->stream.prop_size = 0; + zip->stream.props = NULL; + zip->total_number_nonempty_entry = + zip->total_number_entry - zip->total_number_empty_entry; + + /* Connect an empty file list. */ + if (zip->empty_list.first != NULL) { + *zip->file_list.last = zip->empty_list.first; + zip->file_list.last = zip->empty_list.last; + } + /* Connect a directory file list. */ + ARCHIVE_RB_TREE_FOREACH(n, &(zip->rbtree)) { + file_register(zip, (struct file *)n); + } + + /* + * NOTE: 7z command supports just LZMA1, LZMA2 and COPY for + * the compression type for encoding the header. + */ +#if HAVE_LZMA_H + header_compression = _7Z_LZMA1; + /* If the stored file is only one, do not encode the header. + * This is the same way 7z command does. */ + if (zip->total_number_entry == 1) + header_compression = _7Z_COPY; +#else + header_compression = _7Z_COPY; +#endif + r = _7z_compression_init_encoder(a, header_compression, 6); + if (r < 0) + return (r); + zip->crc32flg = PRECODE_CRC32; + zip->precode_crc32 = 0; + r = make_header(a, data_offset, data_size, data_unpacksize, + 1, &(zip->coder)); + if (r < 0) + return (r); + r = (int)compress_out(a, NULL, 0, ARCHIVE_Z_FINISH); + if (r < 0) + return (r); + header_offset = data_offset + data_size; + header_size = zip->stream.total_out; + header_crc32 = zip->precode_crc32; + header_unpacksize = zip->stream.total_in; + + if (header_compression != _7Z_COPY) { + /* + * Encode the header in order to reduce the size + * of the archive. + */ + free(zip->coder.props); + zip->coder.codec = header_compression; + zip->coder.prop_size = zip->stream.prop_size; + zip->coder.props = zip->stream.props; + zip->stream.prop_size = 0; + zip->stream.props = NULL; + + r = _7z_compression_init_encoder(a, _7Z_COPY, 0); + if (r < 0) + return (r); + zip->crc32flg = ENCODED_CRC32; + zip->encoded_crc32 = 0; + + /* + * Make EncodedHeader. + */ + r = enc_uint64(a, kEncodedHeader); + if (r < 0) + return (r); + r = make_streamsInfo(a, header_offset, header_size, + header_unpacksize, 1, &(zip->coder), 0, + header_crc32); + if (r < 0) + return (r); + r = (int)compress_out(a, NULL, 0, ARCHIVE_Z_FINISH); + if (r < 0) + return (r); + header_offset = header_offset + header_size; + header_size = zip->stream.total_out; + header_crc32 = zip->encoded_crc32; + } + zip->crc32flg = 0; + } else { + header_offset = header_size = 0; + header_crc32 = 0; + } + + length = zip->temp_offset; + + /* + * Make the zip header on wbuff(write buffer). + */ + wb = zip->wbuff; + zip->wbuff_remaining = sizeof(zip->wbuff); + memcpy(&wb[0], "7z\xBC\xAF\x27\x1C", 6); + wb[6] = 0;/* Major version. */ + wb[7] = 3;/* Minor version. */ + archive_le64enc(&wb[12], header_offset);/* Next Header Offset */ + archive_le64enc(&wb[20], header_size);/* Next Header Size */ + archive_le32enc(&wb[28], header_crc32);/* Next Header CRC */ + archive_le32enc(&wb[8], crc32(0, &wb[12], 20));/* Start Header CRC */ + zip->wbuff_remaining -= 32; + + /* + * Read all file contents and an encoded header from the temporary + * file and write out it. + */ + r = copy_out(a, 0, length); + if (r != ARCHIVE_OK) + return (r); + r = flush_wbuff(a); + return (r); +} + +/* + * Encode 64 bits value into 7-Zip's encoded UINT64 value. + */ +static int +enc_uint64(struct archive_write *a, uint64_t val) +{ + unsigned mask = 0x80; + uint8_t numdata[9]; + int i; + + numdata[0] = 0; + for (i = 1; i < (int)sizeof(numdata); i++) { + if (val < mask) { + numdata[0] |= (uint8_t)val; + break; + } + numdata[i] = (uint8_t)val; + val >>= 8; + numdata[0] |= mask; + mask >>= 1; + } + return ((int)compress_out(a, numdata, i, ARCHIVE_Z_RUN)); +} + +static int +make_substreamsInfo(struct archive_write *a, struct coder *coders) +{ + struct _7zip *zip = (struct _7zip *)a->format_data; + struct file *file; + int r; + + /* + * Make SubStreamsInfo. + */ + r = enc_uint64(a, kSubStreamsInfo); + if (r < 0) + return (r); + + if (zip->total_number_nonempty_entry > 1 && coders->codec != _7Z_COPY) { + /* + * Make NumUnPackStream. + */ + r = enc_uint64(a, kNumUnPackStream); + if (r < 0) + return (r); + + /* Write numUnpackStreams */ + r = enc_uint64(a, zip->total_number_nonempty_entry); + if (r < 0) + return (r); + + /* + * Make kSize. + */ + r = enc_uint64(a, kSize); + if (r < 0) + return (r); + file = zip->file_list.first; + for (;file != NULL; file = file->next) { + if (file->next == NULL || + file->next->size == 0) + break; + r = enc_uint64(a, file->size); + if (r < 0) + return (r); + } + } + + /* + * Make CRC. + */ + r = enc_uint64(a, kCRC); + if (r < 0) + return (r); + + + /* All are defined */ + r = enc_uint64(a, 1); + if (r < 0) + return (r); + file = zip->file_list.first; + for (;file != NULL; file = file->next) { + uint8_t crc[4]; + if (file->size == 0) + break; + archive_le32enc(crc, file->crc32); + r = (int)compress_out(a, crc, 4, ARCHIVE_Z_RUN); + if (r < 0) + return (r); + } + + /* Write End. */ + r = enc_uint64(a, kEnd); + if (r < 0) + return (r); + return (ARCHIVE_OK); +} + +static int +make_streamsInfo(struct archive_write *a, uint64_t offset, uint64_t pack_size, + uint64_t unpack_size, int num_coder, struct coder *coders, int substrm, + uint32_t header_crc) +{ + struct _7zip *zip = (struct _7zip *)a->format_data; + uint8_t codec_buff[8]; + int numFolders, fi; + int codec_size; + int i, r; + + if (coders->codec == _7Z_COPY) + numFolders = (int)zip->total_number_nonempty_entry; + else + numFolders = 1; + + /* + * Make PackInfo. + */ + r = enc_uint64(a, kPackInfo); + if (r < 0) + return (r); + + /* Write PackPos. */ + r = enc_uint64(a, offset); + if (r < 0) + return (r); + + /* Write NumPackStreams. */ + r = enc_uint64(a, numFolders); + if (r < 0) + return (r); + + /* Make Size. */ + r = enc_uint64(a, kSize); + if (r < 0) + return (r); + + if (numFolders > 1) { + struct file *file = zip->file_list.first; + for (;file != NULL; file = file->next) { + if (file->size == 0) + break; + r = enc_uint64(a, file->size); + if (r < 0) + return (r); + } + } else { + /* Write size. */ + r = enc_uint64(a, pack_size); + if (r < 0) + return (r); + } + + r = enc_uint64(a, kEnd); + if (r < 0) + return (r); + + /* + * Make UnPackInfo. + */ + r = enc_uint64(a, kUnPackInfo); + if (r < 0) + return (r); + + /* + * Make Folder. + */ + r = enc_uint64(a, kFolder); + if (r < 0) + return (r); + + /* Write NumFolders. */ + r = enc_uint64(a, numFolders); + if (r < 0) + return (r); + + /* Write External. */ + r = enc_uint64(a, 0); + if (r < 0) + return (r); + + for (fi = 0; fi < numFolders; fi++) { + /* Write NumCoders. */ + r = enc_uint64(a, num_coder); + if (r < 0) + return (r); + + for (i = 0; i < num_coder; i++) { + unsigned codec_id = coders[i].codec; + + /* Write Codec flag. */ + archive_be64enc(codec_buff, codec_id); + for (codec_size = 8; codec_size > 0; codec_size--) { + if (codec_buff[8 - codec_size]) + break; + } + if (codec_size == 0) + codec_size = 1; + if (coders[i].prop_size) + r = enc_uint64(a, codec_size | 0x20); + else + r = enc_uint64(a, codec_size); + if (r < 0) + return (r); + + /* Write Codec ID. */ + codec_size &= 0x0f; + r = (int)compress_out(a, &codec_buff[8-codec_size], + codec_size, ARCHIVE_Z_RUN); + if (r < 0) + return (r); + + if (coders[i].prop_size) { + /* Write Codec property size. */ + r = enc_uint64(a, coders[i].prop_size); + if (r < 0) + return (r); + + /* Write Codec properties. */ + r = (int)compress_out(a, coders[i].props, + coders[i].prop_size, ARCHIVE_Z_RUN); + if (r < 0) + return (r); + } + } + } + + /* + * Make CodersUnPackSize. + */ + r = enc_uint64(a, kCodersUnPackSize); + if (r < 0) + return (r); + + if (numFolders > 1) { + struct file *file = zip->file_list.first; + for (;file != NULL; file = file->next) { + if (file->size == 0) + break; + r = enc_uint64(a, file->size); + if (r < 0) + return (r); + } + + } else { + /* Write UnPackSize. */ + r = enc_uint64(a, unpack_size); + if (r < 0) + return (r); + } + + if (!substrm) { + uint8_t crc[4]; + /* + * Make CRC. + */ + r = enc_uint64(a, kCRC); + if (r < 0) + return (r); + + /* All are defined */ + r = enc_uint64(a, 1); + if (r < 0) + return (r); + archive_le32enc(crc, header_crc); + r = (int)compress_out(a, crc, 4, ARCHIVE_Z_RUN); + if (r < 0) + return (r); + } + + /* Write End. */ + r = enc_uint64(a, kEnd); + if (r < 0) + return (r); + + if (substrm) { + /* + * Make SubStreamsInfo. + */ + r = make_substreamsInfo(a, coders); + if (r < 0) + return (r); + } + + + /* Write End. */ + r = enc_uint64(a, kEnd); + if (r < 0) + return (r); + + return (ARCHIVE_OK); +} + + +#define EPOC_TIME ARCHIVE_LITERAL_ULL(116444736000000000) +static uint64_t +utcToFiletime(time_t t, long ns) +{ + uint64_t fileTime; + + fileTime = t; + fileTime *= 10000000; + fileTime += ns / 100; + fileTime += EPOC_TIME; + return (fileTime); +} + +static int +make_time(struct archive_write *a, uint8_t type, unsigned flg, int ti) +{ + uint8_t filetime[8]; + struct _7zip *zip = (struct _7zip *)a->format_data; + struct file *file; + int r; + uint8_t b, mask; + + /* + * Make Time Bools. + */ + if (zip->total_number_time_defined[ti] == zip->total_number_entry) { + /* Write Time Type. */ + r = enc_uint64(a, type); + if (r < 0) + return (r); + /* Write EmptyStream Size. */ + r = enc_uint64(a, 2 + zip->total_number_entry * 8); + if (r < 0) + return (r); + /* All are defined. */ + r = enc_uint64(a, 1); + if (r < 0) + return (r); + } else { + if (zip->total_number_time_defined[ti] == 0) + return (ARCHIVE_OK); + + /* Write Time Type. */ + r = enc_uint64(a, type); + if (r < 0) + return (r); + /* Write EmptyStream Size. */ + r = enc_uint64(a, 2 + ((zip->total_number_entry + 7) >> 3) + + zip->total_number_time_defined[ti] * 8); + if (r < 0) + return (r); + + /* All are not defined. */ + r = enc_uint64(a, 0); + if (r < 0) + return (r); + + b = 0; + mask = 0x80; + file = zip->file_list.first; + for (;file != NULL; file = file->next) { + if (file->flg & flg) + b |= mask; + mask >>= 1; + if (mask == 0) { + r = (int)compress_out(a, &b, 1, ARCHIVE_Z_RUN); + if (r < 0) + return (r); + mask = 0x80; + b = 0; + } + } + if (mask != 0x80) { + r = (int)compress_out(a, &b, 1, ARCHIVE_Z_RUN); + if (r < 0) + return (r); + } + } + + /* External. */ + r = enc_uint64(a, 0); + if (r < 0) + return (r); + + + /* + * Make Times. + */ + file = zip->file_list.first; + for (;file != NULL; file = file->next) { + if ((file->flg & flg) == 0) + continue; + archive_le64enc(filetime, utcToFiletime(file->times[ti].time, + file->times[ti].time_ns)); + r = (int)compress_out(a, filetime, 8, ARCHIVE_Z_RUN); + if (r < 0) + return (r); + } + + return (ARCHIVE_OK); +} + +static int +make_header(struct archive_write *a, uint64_t offset, uint64_t pack_size, + uint64_t unpack_size, int codernum, struct coder *coders) +{ + struct _7zip *zip = (struct _7zip *)a->format_data; + struct file *file; + int r; + uint8_t b, mask; + + /* + * Make FilesInfo. + */ + r = enc_uint64(a, kHeader); + if (r < 0) + return (r); + + /* + * If there are empty files only, do not write MainStreamInfo. + */ + if (zip->total_number_nonempty_entry) { + /* + * Make MainStreamInfo. + */ + r = enc_uint64(a, kMainStreamsInfo); + if (r < 0) + return (r); + r = make_streamsInfo(a, offset, pack_size, unpack_size, + codernum, coders, 1, 0); + if (r < 0) + return (r); + } + + /* + * Make FilesInfo. + */ + r = enc_uint64(a, kFilesInfo); + if (r < 0) + return (r); + + /* Write numFiles. */ + r = enc_uint64(a, zip->total_number_entry); + if (r < 0) + return (r); + + if (zip->total_number_empty_entry > 0) { + /* Make EmptyStream. */ + r = enc_uint64(a, kEmptyStream); + if (r < 0) + return (r); + + /* Write EmptyStream Size. */ + r = enc_uint64(a, (zip->total_number_entry+7)>>3); + if (r < 0) + return (r); + + b = 0; + mask = 0x80; + file = zip->file_list.first; + for (;file != NULL; file = file->next) { + if (file->size == 0) + b |= mask; + mask >>= 1; + if (mask == 0) { + r = (int)compress_out(a, &b, 1, ARCHIVE_Z_RUN); + if (r < 0) + return (r); + mask = 0x80; + b = 0; + } + } + if (mask != 0x80) { + r = (int)compress_out(a, &b, 1, ARCHIVE_Z_RUN); + if (r < 0) + return (r); + } + } + + if (zip->total_number_empty_entry > zip->total_number_dir_entry) { + /* Make EmptyFile. */ + r = enc_uint64(a, kEmptyFile); + if (r < 0) + return (r); + + /* Write EmptyFile Size. */ + r = enc_uint64(a, (zip->total_number_empty_entry + 7) >> 3); + if (r < 0) + return (r); + + b = 0; + mask = 0x80; + file = zip->file_list.first; + for (;file != NULL; file = file->next) { + if (file->size) + continue; + if (!file->dir) + b |= mask; + mask >>= 1; + if (mask == 0) { + r = (int)compress_out(a, &b, 1, ARCHIVE_Z_RUN); + if (r < 0) + return (r); + mask = 0x80; + b = 0; + } + } + if (mask != 0x80) { + r = (int)compress_out(a, &b, 1, ARCHIVE_Z_RUN); + if (r < 0) + return (r); + } + } + + /* Make Name. */ + r = enc_uint64(a, kName); + if (r < 0) + return (r); + + /* Write Name size. */ + r = enc_uint64(a, zip->total_bytes_entry_name+1); + if (r < 0) + return (r); + + /* Write dmy byte. */ + r = enc_uint64(a, 0); + if (r < 0) + return (r); + + file = zip->file_list.first; + for (;file != NULL; file = file->next) { + r = (int)compress_out(a, file->utf16name, file->name_len+2, + ARCHIVE_Z_RUN); + if (r < 0) + return (r); + } + + /* Make MTime. */ + r = make_time(a, kMTime, MTIME_IS_SET, MTIME); + if (r < 0) + return (r); + + /* Make CTime. */ + r = make_time(a, kCTime, CTIME_IS_SET, CTIME); + if (r < 0) + return (r); + + /* Make ATime. */ + r = make_time(a, kATime, ATIME_IS_SET, ATIME); + if (r < 0) + return (r); + + /* Make Attributes. */ + r = enc_uint64(a, kAttributes); + if (r < 0) + return (r); + + /* Write Attributes size. */ + r = enc_uint64(a, 2 + zip->total_number_entry * 4); + if (r < 0) + return (r); + + /* Write "All Are Defined". */ + r = enc_uint64(a, 1); + if (r < 0) + return (r); + + /* Write dmy byte. */ + r = enc_uint64(a, 0); + if (r < 0) + return (r); + + file = zip->file_list.first; + for (;file != NULL; file = file->next) { + /* + * High 16bits is unix mode. + * Low 16bits is Windows attributes. + */ + uint32_t encattr, attr; + if (file->dir) + attr = 0x8010; + else + attr = 0x8020; + if ((file->mode & 0222) == 0) + attr |= 1;/* Read Only. */ + attr |= ((uint32_t)file->mode) << 16; + archive_le32enc(&encattr, attr); + r = (int)compress_out(a, &encattr, 4, ARCHIVE_Z_RUN); + if (r < 0) + return (r); + } + + /* Write End. */ + r = enc_uint64(a, kEnd); + if (r < 0) + return (r); + + /* Write End. */ + r = enc_uint64(a, kEnd); + if (r < 0) + return (r); + + return (ARCHIVE_OK); +} + + +static int +_7z_free(struct archive_write *a) +{ + struct _7zip *zip = (struct _7zip *)a->format_data; + + /* Close the temporary file. */ + if (zip->temp_fd >= 0) + close(zip->temp_fd); + + file_free_register(zip); + compression_end(&(a->archive), &(zip->stream)); + free(zip->coder.props); + free(zip); + + return (ARCHIVE_OK); +} + +static int +file_cmp_node(const struct archive_rb_node *n1, + const struct archive_rb_node *n2) +{ + const struct file *f1 = (const struct file *)n1; + const struct file *f2 = (const struct file *)n2; + + if (f1->name_len == f2->name_len) + return (memcmp(f1->utf16name, f2->utf16name, f1->name_len)); + return (f1->name_len > f2->name_len)?1:-1; +} + +static int +file_cmp_key(const struct archive_rb_node *n, const void *key) +{ + const struct file *f = (const struct file *)n; + + return (f->name_len - *(const char *)key); +} + +static int +file_new(struct archive_write *a, struct archive_entry *entry, + struct file **newfile) +{ + struct _7zip *zip; + struct file *file; + const char *u16; + size_t u16len; + int ret = ARCHIVE_OK; + + zip = (struct _7zip *)a->format_data; + *newfile = NULL; + + file = calloc(1, sizeof(*file)); + if (file == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + + if (0 > archive_entry_pathname_l(entry, &u16, &u16len, zip->sconv)) { + if (errno == ENOMEM) { + free(file); + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for UTF-16LE"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "A filename cannot be converted to UTF-16LE;" + "You should disable making Joliet extension"); + ret = ARCHIVE_WARN; + } + file->utf16name = malloc(u16len + 2); + if (file->utf16name == NULL) { + free(file); + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Name"); + return (ARCHIVE_FATAL); + } + memcpy(file->utf16name, u16, u16len); + file->utf16name[u16len+0] = 0; + file->utf16name[u16len+1] = 0; + file->name_len = (unsigned)u16len; + file->mode = archive_entry_mode(entry); + if (archive_entry_filetype(entry) == AE_IFREG) + file->size = archive_entry_size(entry); + else + archive_entry_set_size(entry, 0); + if (archive_entry_filetype(entry) == AE_IFDIR) + file->dir = 1; + else if (archive_entry_filetype(entry) == AE_IFLNK) + file->size = strlen(archive_entry_symlink(entry)); + if (archive_entry_mtime_is_set(entry)) { + file->flg |= MTIME_IS_SET; + file->times[MTIME].time = archive_entry_mtime(entry); + file->times[MTIME].time_ns = archive_entry_mtime_nsec(entry); + } + if (archive_entry_atime_is_set(entry)) { + file->flg |= ATIME_IS_SET; + file->times[ATIME].time = archive_entry_atime(entry); + file->times[ATIME].time_ns = archive_entry_atime_nsec(entry); + } + if (archive_entry_ctime_is_set(entry)) { + file->flg |= CTIME_IS_SET; + file->times[CTIME].time = archive_entry_ctime(entry); + file->times[CTIME].time_ns = archive_entry_ctime_nsec(entry); + } + + *newfile = file; + return (ret); +} + +static void +file_free(struct file *file) +{ + free(file->utf16name); + free(file); +} + +static void +file_register(struct _7zip *zip, struct file *file) +{ + file->next = NULL; + *zip->file_list.last = file; + zip->file_list.last = &(file->next); +} + +static void +file_init_register(struct _7zip *zip) +{ + zip->file_list.first = NULL; + zip->file_list.last = &(zip->file_list.first); +} + +static void +file_free_register(struct _7zip *zip) +{ + struct file *file, *file_next; + + file = zip->file_list.first; + while (file != NULL) { + file_next = file->next; + file_free(file); + file = file_next; + } +} + +static void +file_register_empty(struct _7zip *zip, struct file *file) +{ + file->next = NULL; + *zip->empty_list.last = file; + zip->empty_list.last = &(file->next); +} + +static void +file_init_register_empty(struct _7zip *zip) +{ + zip->empty_list.first = NULL; + zip->empty_list.last = &(zip->empty_list.first); +} + +#if !defined(HAVE_ZLIB_H) || !defined(HAVE_BZLIB_H) ||\ + !defined(BZ_CONFIG_ERROR) || !defined(HAVE_LZMA_H) +static int +compression_unsupported_encoder(struct archive *a, + struct la_zstream *lastrm, const char *name) +{ + + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "%s compression not supported on this platform", name); + lastrm->valid = 0; + lastrm->real_stream = NULL; + return (ARCHIVE_FAILED); +} +#endif + +/* + * _7_COPY compressor. + */ +static int +compression_init_encoder_copy(struct archive *a, struct la_zstream *lastrm) +{ + + if (lastrm->valid) + compression_end(a, lastrm); + lastrm->valid = 1; + lastrm->code = compression_code_copy; + lastrm->end = compression_end_copy; + return (ARCHIVE_OK); +} + +static int +compression_code_copy(struct archive *a, + struct la_zstream *lastrm, enum la_zaction action) +{ + size_t bytes; + + (void)a; /* UNUSED */ + if (lastrm->avail_out > lastrm->avail_in) + bytes = lastrm->avail_in; + else + bytes = lastrm->avail_out; + if (bytes) { + memcpy(lastrm->next_out, lastrm->next_in, bytes); + lastrm->next_in += bytes; + lastrm->avail_in -= bytes; + lastrm->total_in += bytes; + lastrm->next_out += bytes; + lastrm->avail_out -= bytes; + lastrm->total_out += bytes; + } + if (action == ARCHIVE_Z_FINISH && lastrm->avail_in == 0) + return (ARCHIVE_EOF); + return (ARCHIVE_OK); +} + +static int +compression_end_copy(struct archive *a, struct la_zstream *lastrm) +{ + (void)a; /* UNUSED */ + lastrm->valid = 0; + return (ARCHIVE_OK); +} + +/* + * _7_DEFLATE compressor. + */ +#ifdef HAVE_ZLIB_H +static int +compression_init_encoder_deflate(struct archive *a, + struct la_zstream *lastrm, int level, int withheader) +{ + z_stream *strm; + + if (lastrm->valid) + compression_end(a, lastrm); + strm = calloc(1, sizeof(*strm)); + if (strm == NULL) { + archive_set_error(a, ENOMEM, + "Can't allocate memory for gzip stream"); + return (ARCHIVE_FATAL); + } + /* zlib.h is not const-correct, so we need this one bit + * of ugly hackery to convert a const * pointer to + * a non-const pointer. */ + strm->next_in = (Bytef *)(uintptr_t)(const void *)lastrm->next_in; + strm->avail_in = (uInt)lastrm->avail_in; + strm->total_in = (uLong)lastrm->total_in; + strm->next_out = lastrm->next_out; + strm->avail_out = (uInt)lastrm->avail_out; + strm->total_out = (uLong)lastrm->total_out; + if (deflateInit2(strm, level, Z_DEFLATED, + (withheader)?15:-15, + 8, Z_DEFAULT_STRATEGY) != Z_OK) { + free(strm); + lastrm->real_stream = NULL; + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Internal error initializing compression library"); + return (ARCHIVE_FATAL); + } + lastrm->real_stream = strm; + lastrm->valid = 1; + lastrm->code = compression_code_deflate; + lastrm->end = compression_end_deflate; + return (ARCHIVE_OK); +} + +static int +compression_code_deflate(struct archive *a, + struct la_zstream *lastrm, enum la_zaction action) +{ + z_stream *strm; + int r; + + strm = (z_stream *)lastrm->real_stream; + /* zlib.h is not const-correct, so we need this one bit + * of ugly hackery to convert a const * pointer to + * a non-const pointer. */ + strm->next_in = (Bytef *)(uintptr_t)(const void *)lastrm->next_in; + strm->avail_in = (uInt)lastrm->avail_in; + strm->total_in = (uLong)lastrm->total_in; + strm->next_out = lastrm->next_out; + strm->avail_out = (uInt)lastrm->avail_out; + strm->total_out = (uLong)lastrm->total_out; + r = deflate(strm, + (action == ARCHIVE_Z_FINISH)? Z_FINISH: Z_NO_FLUSH); + lastrm->next_in = strm->next_in; + lastrm->avail_in = strm->avail_in; + lastrm->total_in = strm->total_in; + lastrm->next_out = strm->next_out; + lastrm->avail_out = strm->avail_out; + lastrm->total_out = strm->total_out; + switch (r) { + case Z_OK: + return (ARCHIVE_OK); + case Z_STREAM_END: + return (ARCHIVE_EOF); + default: + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "GZip compression failed:" + " deflate() call returned status %d", r); + return (ARCHIVE_FATAL); + } +} + +static int +compression_end_deflate(struct archive *a, struct la_zstream *lastrm) +{ + z_stream *strm; + int r; + + strm = (z_stream *)lastrm->real_stream; + r = deflateEnd(strm); + free(strm); + lastrm->real_stream = NULL; + lastrm->valid = 0; + if (r != Z_OK) { + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Failed to clean up compressor"); + return (ARCHIVE_FATAL); + } + return (ARCHIVE_OK); +} +#else +static int +compression_init_encoder_deflate(struct archive *a, + struct la_zstream *lastrm, int level, int withheader) +{ + + (void) level; /* UNUSED */ + (void) withheader; /* UNUSED */ + if (lastrm->valid) + compression_end(a, lastrm); + return (compression_unsupported_encoder(a, lastrm, "deflate")); +} +#endif + +/* + * _7_BZIP2 compressor. + */ +#if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) +static int +compression_init_encoder_bzip2(struct archive *a, + struct la_zstream *lastrm, int level) +{ + bz_stream *strm; + + if (lastrm->valid) + compression_end(a, lastrm); + strm = calloc(1, sizeof(*strm)); + if (strm == NULL) { + archive_set_error(a, ENOMEM, + "Can't allocate memory for bzip2 stream"); + return (ARCHIVE_FATAL); + } + /* bzlib.h is not const-correct, so we need this one bit + * of ugly hackery to convert a const * pointer to + * a non-const pointer. */ + strm->next_in = (char *)(uintptr_t)(const void *)lastrm->next_in; + strm->avail_in = lastrm->avail_in; + strm->total_in_lo32 = (uint32_t)(lastrm->total_in & 0xffffffff); + strm->total_in_hi32 = (uint32_t)(lastrm->total_in >> 32); + strm->next_out = (char *)lastrm->next_out; + strm->avail_out = lastrm->avail_out; + strm->total_out_lo32 = (uint32_t)(lastrm->total_out & 0xffffffff); + strm->total_out_hi32 = (uint32_t)(lastrm->total_out >> 32); + if (BZ2_bzCompressInit(strm, level, 0, 30) != BZ_OK) { + free(strm); + lastrm->real_stream = NULL; + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Internal error initializing compression library"); + return (ARCHIVE_FATAL); + } + lastrm->real_stream = strm; + lastrm->valid = 1; + lastrm->code = compression_code_bzip2; + lastrm->end = compression_end_bzip2; + return (ARCHIVE_OK); +} + +static int +compression_code_bzip2(struct archive *a, + struct la_zstream *lastrm, enum la_zaction action) +{ + bz_stream *strm; + int r; + + strm = (bz_stream *)lastrm->real_stream; + /* bzlib.h is not const-correct, so we need this one bit + * of ugly hackery to convert a const * pointer to + * a non-const pointer. */ + strm->next_in = (char *)(uintptr_t)(const void *)lastrm->next_in; + strm->avail_in = lastrm->avail_in; + strm->total_in_lo32 = (uint32_t)(lastrm->total_in & 0xffffffff); + strm->total_in_hi32 = (uint32_t)(lastrm->total_in >> 32); + strm->next_out = (char *)lastrm->next_out; + strm->avail_out = lastrm->avail_out; + strm->total_out_lo32 = (uint32_t)(lastrm->total_out & 0xffffffff); + strm->total_out_hi32 = (uint32_t)(lastrm->total_out >> 32); + r = BZ2_bzCompress(strm, + (action == ARCHIVE_Z_FINISH)? BZ_FINISH: BZ_RUN); + lastrm->next_in = (const unsigned char *)strm->next_in; + lastrm->avail_in = strm->avail_in; + lastrm->total_in = + (((uint64_t)(uint32_t)strm->total_in_hi32) << 32) + + (uint64_t)(uint32_t)strm->total_in_lo32; + lastrm->next_out = (unsigned char *)strm->next_out; + lastrm->avail_out = strm->avail_out; + lastrm->total_out = + (((uint64_t)(uint32_t)strm->total_out_hi32) << 32) + + (uint64_t)(uint32_t)strm->total_out_lo32; + switch (r) { + case BZ_RUN_OK: /* Non-finishing */ + case BZ_FINISH_OK: /* Finishing: There's more work to do */ + return (ARCHIVE_OK); + case BZ_STREAM_END: /* Finishing: all done */ + /* Only occurs in finishing case */ + return (ARCHIVE_EOF); + default: + /* Any other return value indicates an error */ + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Bzip2 compression failed:" + " BZ2_bzCompress() call returned status %d", r); + return (ARCHIVE_FATAL); + } +} + +static int +compression_end_bzip2(struct archive *a, struct la_zstream *lastrm) +{ + bz_stream *strm; + int r; + + strm = (bz_stream *)lastrm->real_stream; + r = BZ2_bzCompressEnd(strm); + free(strm); + lastrm->real_stream = NULL; + lastrm->valid = 0; + if (r != BZ_OK) { + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Failed to clean up compressor"); + return (ARCHIVE_FATAL); + } + return (ARCHIVE_OK); +} + +#else +static int +compression_init_encoder_bzip2(struct archive *a, + struct la_zstream *lastrm, int level) +{ + + (void) level; /* UNUSED */ + if (lastrm->valid) + compression_end(a, lastrm); + return (compression_unsupported_encoder(a, lastrm, "bzip2")); +} +#endif + +/* + * _7_LZMA1, _7_LZMA2 compressor. + */ +#if defined(HAVE_LZMA_H) +static int +compression_init_encoder_lzma(struct archive *a, + struct la_zstream *lastrm, int level, uint64_t filter_id) +{ + static const lzma_stream lzma_init_data = LZMA_STREAM_INIT; + lzma_stream *strm; + lzma_filter *lzmafilters; + lzma_options_lzma lzma_opt; + int r; + + if (lastrm->valid) + compression_end(a, lastrm); + strm = calloc(1, sizeof(*strm) + sizeof(*lzmafilters) * 2); + if (strm == NULL) { + archive_set_error(a, ENOMEM, + "Can't allocate memory for lzma stream"); + return (ARCHIVE_FATAL); + } + lzmafilters = (lzma_filter *)(strm+1); + if (level > 6) + level = 6; + if (lzma_lzma_preset(&lzma_opt, level)) { + free(strm); + lastrm->real_stream = NULL; + archive_set_error(a, ENOMEM, + "Internal error initializing compression library"); + return (ARCHIVE_FATAL); + } + lzmafilters[0].id = filter_id; + lzmafilters[0].options = &lzma_opt; + lzmafilters[1].id = LZMA_VLI_UNKNOWN;/* Terminate */ + + r = lzma_properties_size(&(lastrm->prop_size), lzmafilters); + if (r != LZMA_OK) { + free(strm); + lastrm->real_stream = NULL; + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "lzma_properties_size failed"); + return (ARCHIVE_FATAL); + } + if (lastrm->prop_size) { + lastrm->props = malloc(lastrm->prop_size); + if (lastrm->props == NULL) { + free(strm); + lastrm->real_stream = NULL; + archive_set_error(a, ENOMEM, + "Cannot allocate memory"); + return (ARCHIVE_FATAL); + } + r = lzma_properties_encode(lzmafilters, lastrm->props); + if (r != LZMA_OK) { + free(strm); + lastrm->real_stream = NULL; + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "lzma_properties_encode failed"); + return (ARCHIVE_FATAL); + } + } + + *strm = lzma_init_data; + r = lzma_raw_encoder(strm, lzmafilters); + switch (r) { + case LZMA_OK: + lastrm->real_stream = strm; + lastrm->valid = 1; + lastrm->code = compression_code_lzma; + lastrm->end = compression_end_lzma; + r = ARCHIVE_OK; + break; + case LZMA_MEM_ERROR: + free(strm); + lastrm->real_stream = NULL; + archive_set_error(a, ENOMEM, + "Internal error initializing compression library: " + "Cannot allocate memory"); + r = ARCHIVE_FATAL; + break; + default: + free(strm); + lastrm->real_stream = NULL; + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Internal error initializing compression library: " + "It's a bug in liblzma"); + r = ARCHIVE_FATAL; + break; + } + return (r); +} + +static int +compression_init_encoder_lzma1(struct archive *a, + struct la_zstream *lastrm, int level) +{ + return compression_init_encoder_lzma(a, lastrm, level, + LZMA_FILTER_LZMA1); +} + +static int +compression_init_encoder_lzma2(struct archive *a, + struct la_zstream *lastrm, int level) +{ + return compression_init_encoder_lzma(a, lastrm, level, + LZMA_FILTER_LZMA2); +} + +static int +compression_code_lzma(struct archive *a, + struct la_zstream *lastrm, enum la_zaction action) +{ + lzma_stream *strm; + int r; + + strm = (lzma_stream *)lastrm->real_stream; + strm->next_in = lastrm->next_in; + strm->avail_in = lastrm->avail_in; + strm->total_in = lastrm->total_in; + strm->next_out = lastrm->next_out; + strm->avail_out = lastrm->avail_out; + strm->total_out = lastrm->total_out; + r = lzma_code(strm, + (action == ARCHIVE_Z_FINISH)? LZMA_FINISH: LZMA_RUN); + lastrm->next_in = strm->next_in; + lastrm->avail_in = strm->avail_in; + lastrm->total_in = strm->total_in; + lastrm->next_out = strm->next_out; + lastrm->avail_out = strm->avail_out; + lastrm->total_out = strm->total_out; + switch (r) { + case LZMA_OK: + /* Non-finishing case */ + return (ARCHIVE_OK); + case LZMA_STREAM_END: + /* This return can only occur in finishing case. */ + return (ARCHIVE_EOF); + case LZMA_MEMLIMIT_ERROR: + archive_set_error(a, ENOMEM, + "lzma compression error:" + " %ju MiB would have been needed", + (uintmax_t)((lzma_memusage(strm) + 1024 * 1024 -1) + / (1024 * 1024))); + return (ARCHIVE_FATAL); + default: + /* Any other return value indicates an error */ + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "lzma compression failed:" + " lzma_code() call returned status %d", r); + return (ARCHIVE_FATAL); + } +} + +static int +compression_end_lzma(struct archive *a, struct la_zstream *lastrm) +{ + lzma_stream *strm; + + (void)a; /* UNUSED */ + strm = (lzma_stream *)lastrm->real_stream; + lzma_end(strm); + free(strm); + lastrm->valid = 0; + lastrm->real_stream = NULL; + return (ARCHIVE_OK); +} +#else +static int +compression_init_encoder_lzma1(struct archive *a, + struct la_zstream *lastrm, int level) +{ + + (void) level; /* UNUSED */ + if (lastrm->valid) + compression_end(a, lastrm); + return (compression_unsupported_encoder(a, lastrm, "lzma")); +} +static int +compression_init_encoder_lzma2(struct archive *a, + struct la_zstream *lastrm, int level) +{ + + (void) level; /* UNUSED */ + if (lastrm->valid) + compression_end(a, lastrm); + return (compression_unsupported_encoder(a, lastrm, "lzma")); +} +#endif + +/* + * _7_PPMD compressor. + */ +static void +ppmd_write(void *p, Byte b) +{ + struct archive_write *a = ((IByteOut *)p)->a; + struct _7zip *zip = (struct _7zip *)(a->format_data); + struct la_zstream *lastrm = &(zip->stream); + struct ppmd_stream *strm; + + if (lastrm->avail_out) { + *lastrm->next_out++ = b; + lastrm->avail_out--; + lastrm->total_out++; + return; + } + strm = (struct ppmd_stream *)lastrm->real_stream; + if (strm->buff_ptr < strm->buff_end) { + *strm->buff_ptr++ = b; + strm->buff_bytes++; + } +} + +static int +compression_init_encoder_ppmd(struct archive *a, + struct la_zstream *lastrm, unsigned maxOrder, uint32_t msize) +{ + struct ppmd_stream *strm; + uint8_t *props; + int r; + + if (lastrm->valid) + compression_end(a, lastrm); + strm = calloc(1, sizeof(*strm)); + if (strm == NULL) { + archive_set_error(a, ENOMEM, + "Can't allocate memory for PPMd"); + return (ARCHIVE_FATAL); + } + strm->buff = malloc(32); + if (strm->buff == NULL) { + free(strm); + archive_set_error(a, ENOMEM, + "Can't allocate memory for PPMd"); + return (ARCHIVE_FATAL); + } + strm->buff_ptr = strm->buff; + strm->buff_end = strm->buff + 32; + + props = malloc(1+4); + if (props == NULL) { + free(strm->buff); + free(strm); + archive_set_error(a, ENOMEM, + "Coludn't allocate memory for PPMd"); + return (ARCHIVE_FATAL); + } + props[0] = maxOrder; + archive_le32enc(props+1, msize); + __archive_ppmd7_functions.Ppmd7_Construct(&strm->ppmd7_context); + r = __archive_ppmd7_functions.Ppmd7_Alloc( + &strm->ppmd7_context, msize); + if (r == 0) { + free(strm->buff); + free(strm); + free(props); + archive_set_error(a, ENOMEM, + "Coludn't allocate memory for PPMd"); + return (ARCHIVE_FATAL); + } + __archive_ppmd7_functions.Ppmd7_Init(&(strm->ppmd7_context), maxOrder); + strm->byteout.a = (struct archive_write *)a; + strm->byteout.Write = ppmd_write; + strm->range_enc.Stream = &(strm->byteout); + __archive_ppmd7_functions.Ppmd7z_RangeEnc_Init(&(strm->range_enc)); + strm->stat = 0; + + lastrm->real_stream = strm; + lastrm->valid = 1; + lastrm->code = compression_code_ppmd; + lastrm->end = compression_end_ppmd; + lastrm->prop_size = 5; + lastrm->props = props; + return (ARCHIVE_OK); +} + +static int +compression_code_ppmd(struct archive *a, + struct la_zstream *lastrm, enum la_zaction action) +{ + struct ppmd_stream *strm; + + (void)a; /* UNUSED */ + + strm = (struct ppmd_stream *)lastrm->real_stream; + + /* Copy encoded data if there are remaining bytes from previous call. */ + if (strm->buff_bytes) { + uint8_t *p = strm->buff_ptr - strm->buff_bytes; + while (lastrm->avail_out && strm->buff_bytes) { + *lastrm->next_out++ = *p++; + lastrm->avail_out--; + lastrm->total_out++; + strm->buff_bytes--; + } + if (strm->buff_bytes) + return (ARCHIVE_OK); + if (strm->stat == 1) + return (ARCHIVE_EOF); + strm->buff_ptr = strm->buff; + } + while (lastrm->avail_in && lastrm->avail_out) { + __archive_ppmd7_functions.Ppmd7_EncodeSymbol( + &(strm->ppmd7_context), &(strm->range_enc), + *lastrm->next_in++); + lastrm->avail_in--; + lastrm->total_in++; + } + if (lastrm->avail_in == 0 && action == ARCHIVE_Z_FINISH) { + __archive_ppmd7_functions.Ppmd7z_RangeEnc_FlushData( + &(strm->range_enc)); + strm->stat = 1; + /* Return EOF if there are no remaining bytes. */ + if (strm->buff_bytes == 0) + return (ARCHIVE_EOF); + } + return (ARCHIVE_OK); +} + +static int +compression_end_ppmd(struct archive *a, struct la_zstream *lastrm) +{ + struct ppmd_stream *strm; + + (void)a; /* UNUSED */ + + strm = (struct ppmd_stream *)lastrm->real_stream; + __archive_ppmd7_functions.Ppmd7_Free(&strm->ppmd7_context); + free(strm->buff); + free(strm); + lastrm->real_stream = NULL; + lastrm->valid = 0; + return (ARCHIVE_OK); +} + +/* + * Universal compressor initializer. + */ +static int +_7z_compression_init_encoder(struct archive_write *a, unsigned compression, + int compression_level) +{ + struct _7zip *zip; + int r; + + zip = (struct _7zip *)a->format_data; + switch (compression) { + case _7Z_DEFLATE: + r = compression_init_encoder_deflate( + &(a->archive), &(zip->stream), + compression_level, 0); + break; + case _7Z_BZIP2: + r = compression_init_encoder_bzip2( + &(a->archive), &(zip->stream), + compression_level); + break; + case _7Z_LZMA1: + r = compression_init_encoder_lzma1( + &(a->archive), &(zip->stream), + compression_level); + break; + case _7Z_LZMA2: + r = compression_init_encoder_lzma2( + &(a->archive), &(zip->stream), + compression_level); + break; + case _7Z_PPMD: + r = compression_init_encoder_ppmd( + &(a->archive), &(zip->stream), + PPMD7_DEFAULT_ORDER, PPMD7_DEFAULT_MEM_SIZE); + break; + case _7Z_COPY: + default: + r = compression_init_encoder_copy( + &(a->archive), &(zip->stream)); + break; + } + if (r == ARCHIVE_OK) { + zip->stream.total_in = 0; + zip->stream.next_out = zip->wbuff; + zip->stream.avail_out = sizeof(zip->wbuff); + zip->stream.total_out = 0; + } + + return (r); +} + +static int +compression_code(struct archive *a, struct la_zstream *lastrm, + enum la_zaction action) +{ + if (lastrm->valid) + return (lastrm->code(a, lastrm, action)); + return (ARCHIVE_OK); +} + +static int +compression_end(struct archive *a, struct la_zstream *lastrm) +{ + if (lastrm->valid) { + lastrm->prop_size = 0; + free(lastrm->props); + lastrm->props = NULL; + return (lastrm->end(a, lastrm)); + } + return (ARCHIVE_OK); +} + + diff --git a/c/archive_write_set_format_ar.c b/c/archive_write_set_format_ar.c new file mode 100644 --- /dev/null +++ b/c/archive_write_set_format_ar.c @@ -0,0 +1,570 @@ +/*- + * Copyright (c) 2007 Kai Wang + * Copyright (c) 2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer + * in this position and unchanged. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_format_ar.c 201108 2009-12-28 03:28:21Z kientzle $"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_write_private.h" +#include "archive_write_set_format_private.h" + +struct ar_w { + uint64_t entry_bytes_remaining; + uint64_t entry_padding; + int is_strtab; + int has_strtab; + char wrote_global_header; + char *strtab; +}; + +/* + * Define structure of the "ar" header. + */ +#define AR_name_offset 0 +#define AR_name_size 16 +#define AR_date_offset 16 +#define AR_date_size 12 +#define AR_uid_offset 28 +#define AR_uid_size 6 +#define AR_gid_offset 34 +#define AR_gid_size 6 +#define AR_mode_offset 40 +#define AR_mode_size 8 +#define AR_size_offset 48 +#define AR_size_size 10 +#define AR_fmag_offset 58 +#define AR_fmag_size 2 + +static int archive_write_set_format_ar(struct archive_write *); +static int archive_write_ar_header(struct archive_write *, + struct archive_entry *); +static ssize_t archive_write_ar_data(struct archive_write *, + const void *buff, size_t s); +static int archive_write_ar_free(struct archive_write *); +static int archive_write_ar_close(struct archive_write *); +static int archive_write_ar_finish_entry(struct archive_write *); +static const char *ar_basename(const char *path); +static int format_octal(int64_t v, char *p, int s); +static int format_decimal(int64_t v, char *p, int s); + +int +archive_write_set_format_ar_bsd(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + int r; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_set_format_ar_bsd"); + r = archive_write_set_format_ar(a); + if (r == ARCHIVE_OK) { + a->archive.archive_format = ARCHIVE_FORMAT_AR_BSD; + a->archive.archive_format_name = "ar (BSD)"; + } + return (r); +} + +int +archive_write_set_format_ar_svr4(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + int r; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_set_format_ar_svr4"); + r = archive_write_set_format_ar(a); + if (r == ARCHIVE_OK) { + a->archive.archive_format = ARCHIVE_FORMAT_AR_GNU; + a->archive.archive_format_name = "ar (GNU/SVR4)"; + } + return (r); +} + +/* + * Generic initialization. + */ +static int +archive_write_set_format_ar(struct archive_write *a) +{ + struct ar_w *ar; + + /* If someone else was already registered, unregister them. */ + if (a->format_free != NULL) + (a->format_free)(a); + + ar = (struct ar_w *)calloc(1, sizeof(*ar)); + if (ar == NULL) { + archive_set_error(&a->archive, ENOMEM, "Can't allocate ar data"); + return (ARCHIVE_FATAL); + } + a->format_data = ar; + + a->format_name = "ar"; + a->format_write_header = archive_write_ar_header; + a->format_write_data = archive_write_ar_data; + a->format_close = archive_write_ar_close; + a->format_free = archive_write_ar_free; + a->format_finish_entry = archive_write_ar_finish_entry; + return (ARCHIVE_OK); +} + +static int +archive_write_ar_header(struct archive_write *a, struct archive_entry *entry) +{ + int ret, append_fn; + char buff[60]; + char *ss, *se; + struct ar_w *ar; + const char *pathname; + const char *filename; + int64_t size; + + append_fn = 0; + ar = (struct ar_w *)a->format_data; + ar->is_strtab = 0; + filename = NULL; + size = archive_entry_size(entry); + + + /* + * Reject files with empty name. + */ + pathname = archive_entry_pathname(entry); + if (pathname == NULL || *pathname == '\0') { + archive_set_error(&a->archive, EINVAL, + "Invalid filename"); + return (ARCHIVE_WARN); + } + + /* + * If we are now at the beginning of the archive, + * we need first write the ar global header. + */ + if (!ar->wrote_global_header) { + __archive_write_output(a, "!\n", 8); + ar->wrote_global_header = 1; + } + + memset(buff, ' ', 60); + memcpy(&buff[AR_fmag_offset], "`\n", 2); + + if (strcmp(pathname, "/") == 0 ) { + /* Entry is archive symbol table in GNU format */ + buff[AR_name_offset] = '/'; + goto stat; + } + if (strcmp(pathname, "/SYM64/") == 0) { + /* Entry is archive symbol table in GNU 64-bit format */ + memcpy(buff + AR_name_offset, "/SYM64/", 7); + goto stat; + } + if (strcmp(pathname, "__.SYMDEF") == 0) { + /* Entry is archive symbol table in BSD format */ + memcpy(buff + AR_name_offset, "__.SYMDEF", 9); + goto stat; + } + if (strcmp(pathname, "//") == 0) { + /* + * Entry is archive filename table, inform that we should + * collect strtab in next _data call. + */ + ar->is_strtab = 1; + buff[AR_name_offset] = buff[AR_name_offset + 1] = '/'; + /* + * For archive string table, only ar_size field should + * be set. + */ + goto size; + } + + /* + * Otherwise, entry is a normal archive member. + * Strip leading paths from filenames, if any. + */ + if ((filename = ar_basename(pathname)) == NULL) { + /* Reject filenames with trailing "/" */ + archive_set_error(&a->archive, EINVAL, + "Invalid filename"); + return (ARCHIVE_WARN); + } + + if (a->archive.archive_format == ARCHIVE_FORMAT_AR_GNU) { + /* + * SVR4/GNU variant use a "/" to mark then end of the filename, + * make it possible to have embedded spaces in the filename. + * So, the longest filename here (without extension) is + * actually 15 bytes. + */ + if (strlen(filename) <= 15) { + memcpy(&buff[AR_name_offset], + filename, strlen(filename)); + buff[AR_name_offset + strlen(filename)] = '/'; + } else { + /* + * For filename longer than 15 bytes, GNU variant + * makes use of a string table and instead stores the + * offset of the real filename to in the ar_name field. + * The string table should have been written before. + */ + if (ar->has_strtab <= 0) { + archive_set_error(&a->archive, EINVAL, + "Can't find string table"); + return (ARCHIVE_WARN); + } + + se = (char *)malloc(strlen(filename) + 3); + if (se == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate filename buffer"); + return (ARCHIVE_FATAL); + } + + memcpy(se, filename, strlen(filename)); + strcpy(se + strlen(filename), "/\n"); + + ss = strstr(ar->strtab, se); + free(se); + + if (ss == NULL) { + archive_set_error(&a->archive, EINVAL, + "Invalid string table"); + return (ARCHIVE_WARN); + } + + /* + * GNU variant puts "/" followed by digits into + * ar_name field. These digits indicates the real + * filename string's offset to the string table. + */ + buff[AR_name_offset] = '/'; + if (format_decimal(ss - ar->strtab, + buff + AR_name_offset + 1, + AR_name_size - 1)) { + archive_set_error(&a->archive, ERANGE, + "string table offset too large"); + return (ARCHIVE_WARN); + } + } + } else if (a->archive.archive_format == ARCHIVE_FORMAT_AR_BSD) { + /* + * BSD variant: for any file name which is more than + * 16 chars or contains one or more embedded space(s), the + * string "#1/" followed by the ASCII length of the name is + * put into the ar_name field. The file size (stored in the + * ar_size field) is incremented by the length of the name. + * The name is then written immediately following the + * archive header. + */ + if (strlen(filename) <= 16 && strchr(filename, ' ') == NULL) { + memcpy(&buff[AR_name_offset], filename, strlen(filename)); + buff[AR_name_offset + strlen(filename)] = ' '; + } + else { + memcpy(buff + AR_name_offset, "#1/", 3); + if (format_decimal(strlen(filename), + buff + AR_name_offset + 3, + AR_name_size - 3)) { + archive_set_error(&a->archive, ERANGE, + "File name too long"); + return (ARCHIVE_WARN); + } + append_fn = 1; + size += strlen(filename); + } + } + +stat: + if (format_decimal(archive_entry_mtime(entry), buff + AR_date_offset, AR_date_size)) { + archive_set_error(&a->archive, ERANGE, + "File modification time too large"); + return (ARCHIVE_WARN); + } + if (format_decimal(archive_entry_uid(entry), buff + AR_uid_offset, AR_uid_size)) { + archive_set_error(&a->archive, ERANGE, + "Numeric user ID too large"); + return (ARCHIVE_WARN); + } + if (format_decimal(archive_entry_gid(entry), buff + AR_gid_offset, AR_gid_size)) { + archive_set_error(&a->archive, ERANGE, + "Numeric group ID too large"); + return (ARCHIVE_WARN); + } + if (format_octal(archive_entry_mode(entry), buff + AR_mode_offset, AR_mode_size)) { + archive_set_error(&a->archive, ERANGE, + "Numeric mode too large"); + return (ARCHIVE_WARN); + } + /* + * Sanity Check: A non-pseudo archive member should always be + * a regular file. + */ + if (filename != NULL && archive_entry_filetype(entry) != AE_IFREG) { + archive_set_error(&a->archive, EINVAL, + "Regular file required for non-pseudo member"); + return (ARCHIVE_WARN); + } + +size: + if (format_decimal(size, buff + AR_size_offset, AR_size_size)) { + archive_set_error(&a->archive, ERANGE, + "File size out of range"); + return (ARCHIVE_WARN); + } + + ret = __archive_write_output(a, buff, 60); + if (ret != ARCHIVE_OK) + return (ret); + + ar->entry_bytes_remaining = size; + ar->entry_padding = ar->entry_bytes_remaining % 2; + + if (append_fn > 0) { + ret = __archive_write_output(a, filename, strlen(filename)); + if (ret != ARCHIVE_OK) + return (ret); + ar->entry_bytes_remaining -= strlen(filename); + } + + return (ARCHIVE_OK); +} + +static ssize_t +archive_write_ar_data(struct archive_write *a, const void *buff, size_t s) +{ + struct ar_w *ar; + int ret; + + ar = (struct ar_w *)a->format_data; + if (s > ar->entry_bytes_remaining) + s = (size_t)ar->entry_bytes_remaining; + + if (ar->is_strtab > 0) { + if (ar->has_strtab > 0) { + archive_set_error(&a->archive, EINVAL, + "More than one string tables exist"); + return (ARCHIVE_WARN); + } + + ar->strtab = (char *)malloc(s + 1); + if (ar->strtab == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate strtab buffer"); + return (ARCHIVE_FATAL); + } + memcpy(ar->strtab, buff, s); + ar->strtab[s] = '\0'; + ar->has_strtab = 1; + } + + ret = __archive_write_output(a, buff, s); + if (ret != ARCHIVE_OK) + return (ret); + + ar->entry_bytes_remaining -= s; + return (s); +} + +static int +archive_write_ar_free(struct archive_write *a) +{ + struct ar_w *ar; + + ar = (struct ar_w *)a->format_data; + + if (ar == NULL) + return (ARCHIVE_OK); + + if (ar->has_strtab > 0) { + free(ar->strtab); + ar->strtab = NULL; + } + + free(ar); + a->format_data = NULL; + return (ARCHIVE_OK); +} + +static int +archive_write_ar_close(struct archive_write *a) +{ + struct ar_w *ar; + int ret; + + /* + * If we haven't written anything yet, we need to write + * the ar global header now to make it a valid ar archive. + */ + ar = (struct ar_w *)a->format_data; + if (!ar->wrote_global_header) { + ar->wrote_global_header = 1; + ret = __archive_write_output(a, "!\n", 8); + return (ret); + } + + return (ARCHIVE_OK); +} + +static int +archive_write_ar_finish_entry(struct archive_write *a) +{ + struct ar_w *ar; + int ret; + + ar = (struct ar_w *)a->format_data; + + if (ar->entry_bytes_remaining != 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Entry remaining bytes larger than 0"); + return (ARCHIVE_WARN); + } + + if (ar->entry_padding == 0) { + return (ARCHIVE_OK); + } + + if (ar->entry_padding != 1) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Padding wrong size: %ju should be 1 or 0", + (uintmax_t)ar->entry_padding); + return (ARCHIVE_WARN); + } + + ret = __archive_write_output(a, "\n", 1); + return (ret); +} + +/* + * Format a number into the specified field using base-8. + * NB: This version is slightly different from the one in + * _ustar.c + */ +static int +format_octal(int64_t v, char *p, int s) +{ + int len; + char *h; + + len = s; + h = p; + + /* Octal values can't be negative, so use 0. */ + if (v < 0) { + while (len-- > 0) + *p++ = '0'; + return (-1); + } + + p += s; /* Start at the end and work backwards. */ + do { + *--p = (char)('0' + (v & 7)); + v >>= 3; + } while (--s > 0 && v > 0); + + if (v == 0) { + memmove(h, p, len - s); + p = h + len - s; + while (s-- > 0) + *p++ = ' '; + return (0); + } + /* If it overflowed, fill field with max value. */ + while (len-- > 0) + *p++ = '7'; + + return (-1); +} + +/* + * Format a number into the specified field using base-10. + */ +static int +format_decimal(int64_t v, char *p, int s) +{ + int len; + char *h; + + len = s; + h = p; + + /* Negative values in ar header are meaningless, so use 0. */ + if (v < 0) { + while (len-- > 0) + *p++ = '0'; + return (-1); + } + + p += s; + do { + *--p = (char)('0' + (v % 10)); + v /= 10; + } while (--s > 0 && v > 0); + + if (v == 0) { + memmove(h, p, len - s); + p = h + len - s; + while (s-- > 0) + *p++ = ' '; + return (0); + } + /* If it overflowed, fill field with max value. */ + while (len-- > 0) + *p++ = '9'; + + return (-1); +} + +static const char * +ar_basename(const char *path) +{ + const char *endp, *startp; + + endp = path + strlen(path) - 1; + /* + * For filename with trailing slash(es), we return + * NULL indicating an error. + */ + if (*endp == '/') + return (NULL); + + /* Find the start of the base */ + startp = endp; + while (startp > path && *(startp - 1) != '/') + startp--; + + return (startp); +} diff --git a/c/archive_write_set_format_by_name.c b/c/archive_write_set_format_by_name.c new file mode 100644 --- /dev/null +++ b/c/archive_write_set_format_by_name.c @@ -0,0 +1,92 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_format_by_name.c 201168 2009-12-29 06:15:32Z kientzle $"); + +#ifdef HAVE_SYS_TYPES_H +#include +#endif + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" + +/* A table that maps names to functions. */ +static const +struct { const char *name; int (*setter)(struct archive *); } names[] = +{ + { "7zip", archive_write_set_format_7zip }, + { "ar", archive_write_set_format_ar_bsd }, + { "arbsd", archive_write_set_format_ar_bsd }, + { "argnu", archive_write_set_format_ar_svr4 }, + { "arsvr4", archive_write_set_format_ar_svr4 }, + { "bsdtar", archive_write_set_format_pax_restricted }, + { "cd9660", archive_write_set_format_iso9660 }, + { "cpio", archive_write_set_format_cpio }, + { "gnutar", archive_write_set_format_gnutar }, + { "iso", archive_write_set_format_iso9660 }, + { "iso9660", archive_write_set_format_iso9660 }, + { "mtree", archive_write_set_format_mtree }, + { "mtree-classic", archive_write_set_format_mtree_classic }, + { "newc", archive_write_set_format_cpio_newc }, + { "odc", archive_write_set_format_cpio }, + { "oldtar", archive_write_set_format_v7tar }, + { "pax", archive_write_set_format_pax }, + { "paxr", archive_write_set_format_pax_restricted }, + { "posix", archive_write_set_format_pax }, + { "raw", archive_write_set_format_raw }, + { "rpax", archive_write_set_format_pax_restricted }, + { "shar", archive_write_set_format_shar }, + { "shardump", archive_write_set_format_shar_dump }, + { "ustar", archive_write_set_format_ustar }, + { "v7tar", archive_write_set_format_v7tar }, + { "v7", archive_write_set_format_v7tar }, + { "warc", archive_write_set_format_warc }, + { "xar", archive_write_set_format_xar }, + { "zip", archive_write_set_format_zip }, + { NULL, NULL } +}; + +int +archive_write_set_format_by_name(struct archive *a, const char *name) +{ + int i; + + for (i = 0; names[i].name != NULL; i++) { + if (strcmp(name, names[i].name) == 0) + return ((names[i].setter)(a)); + } + + archive_set_error(a, EINVAL, "No such format '%s'", name); + a->state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); +} diff --git a/c/archive_write_set_format_cpio.c b/c/archive_write_set_format_cpio.c new file mode 100644 --- /dev/null +++ b/c/archive_write_set_format_cpio.c @@ -0,0 +1,500 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * Copyright (c) 2011-2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_format_cpio.c 201170 2009-12-29 06:34:23Z kientzle $"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_private.h" +#include "archive_write_private.h" +#include "archive_write_set_format_private.h" + +static ssize_t archive_write_cpio_data(struct archive_write *, + const void *buff, size_t s); +static int archive_write_cpio_close(struct archive_write *); +static int archive_write_cpio_free(struct archive_write *); +static int archive_write_cpio_finish_entry(struct archive_write *); +static int archive_write_cpio_header(struct archive_write *, + struct archive_entry *); +static int archive_write_cpio_options(struct archive_write *, + const char *, const char *); +static int format_octal(int64_t, void *, int); +static int64_t format_octal_recursive(int64_t, char *, int); +static int write_header(struct archive_write *, struct archive_entry *); + +struct cpio { + uint64_t entry_bytes_remaining; + + int64_t ino_next; + + struct { int64_t old; int new;} *ino_list; + size_t ino_list_size; + size_t ino_list_next; + + struct archive_string_conv *opt_sconv; + struct archive_string_conv *sconv_default; + int init_default_conversion; +}; + +#define c_magic_offset 0 +#define c_magic_size 6 +#define c_dev_offset 6 +#define c_dev_size 6 +#define c_ino_offset 12 +#define c_ino_size 6 +#define c_mode_offset 18 +#define c_mode_size 6 +#define c_uid_offset 24 +#define c_uid_size 6 +#define c_gid_offset 30 +#define c_gid_size 6 +#define c_nlink_offset 36 +#define c_nlink_size 6 +#define c_rdev_offset 42 +#define c_rdev_size 6 +#define c_mtime_offset 48 +#define c_mtime_size 11 +#define c_namesize_offset 59 +#define c_namesize_size 6 +#define c_filesize_offset 65 +#define c_filesize_size 11 + +/* + * Set output format to 'cpio' format. + */ +int +archive_write_set_format_cpio(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct cpio *cpio; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_set_format_cpio"); + + /* If someone else was already registered, unregister them. */ + if (a->format_free != NULL) + (a->format_free)(a); + + cpio = (struct cpio *)calloc(1, sizeof(*cpio)); + if (cpio == NULL) { + archive_set_error(&a->archive, ENOMEM, "Can't allocate cpio data"); + return (ARCHIVE_FATAL); + } + a->format_data = cpio; + a->format_name = "cpio"; + a->format_options = archive_write_cpio_options; + a->format_write_header = archive_write_cpio_header; + a->format_write_data = archive_write_cpio_data; + a->format_finish_entry = archive_write_cpio_finish_entry; + a->format_close = archive_write_cpio_close; + a->format_free = archive_write_cpio_free; + a->archive.archive_format = ARCHIVE_FORMAT_CPIO_POSIX; + a->archive.archive_format_name = "POSIX cpio"; + return (ARCHIVE_OK); +} + +static int +archive_write_cpio_options(struct archive_write *a, const char *key, + const char *val) +{ + struct cpio *cpio = (struct cpio *)a->format_data; + int ret = ARCHIVE_FAILED; + + if (strcmp(key, "hdrcharset") == 0) { + if (val == NULL || val[0] == 0) + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "%s: hdrcharset option needs a character-set name", + a->format_name); + else { + cpio->opt_sconv = archive_string_conversion_to_charset( + &a->archive, val, 0); + if (cpio->opt_sconv != NULL) + ret = ARCHIVE_OK; + else + ret = ARCHIVE_FATAL; + } + return (ret); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +/* + * Ino values are as long as 64 bits on some systems; cpio format + * only allows 18 bits and relies on the ino values to identify hardlinked + * files. So, we can't merely "hash" the ino numbers since collisions + * would corrupt the archive. Instead, we generate synthetic ino values + * to store in the archive and maintain a map of original ino values to + * synthetic ones so we can preserve hardlink information. + * + * TODO: Make this more efficient. It's not as bad as it looks (most + * files don't have any hardlinks and we don't do any work here for those), + * but it wouldn't be hard to do better. + * + * TODO: Work with dev/ino pairs here instead of just ino values. + */ +static int +synthesize_ino_value(struct cpio *cpio, struct archive_entry *entry) +{ + int64_t ino = archive_entry_ino64(entry); + int ino_new; + size_t i; + + /* + * If no index number was given, don't assign one. In + * particular, this handles the end-of-archive marker + * correctly by giving it a zero index value. (This is also + * why we start our synthetic index numbers with one below.) + */ + if (ino == 0) + return (0); + + /* Don't store a mapping if we don't need to. */ + if (archive_entry_nlink(entry) < 2) { + return (int)(++cpio->ino_next); + } + + /* Look up old ino; if we have it, this is a hardlink + * and we reuse the same value. */ + for (i = 0; i < cpio->ino_list_next; ++i) { + if (cpio->ino_list[i].old == ino) + return (cpio->ino_list[i].new); + } + + /* Assign a new index number. */ + ino_new = (int)(++cpio->ino_next); + + /* Ensure space for the new mapping. */ + if (cpio->ino_list_size <= cpio->ino_list_next) { + size_t newsize = cpio->ino_list_size < 512 + ? 512 : cpio->ino_list_size * 2; + void *newlist = realloc(cpio->ino_list, + sizeof(cpio->ino_list[0]) * newsize); + if (newlist == NULL) + return (-1); + + cpio->ino_list_size = newsize; + cpio->ino_list = newlist; + } + + /* Record and return the new value. */ + cpio->ino_list[cpio->ino_list_next].old = ino; + cpio->ino_list[cpio->ino_list_next].new = ino_new; + ++cpio->ino_list_next; + return (ino_new); +} + + +static struct archive_string_conv * +get_sconv(struct archive_write *a) +{ + struct cpio *cpio; + struct archive_string_conv *sconv; + + cpio = (struct cpio *)a->format_data; + sconv = cpio->opt_sconv; + if (sconv == NULL) { + if (!cpio->init_default_conversion) { + cpio->sconv_default = + archive_string_default_conversion_for_write( + &(a->archive)); + cpio->init_default_conversion = 1; + } + sconv = cpio->sconv_default; + } + return (sconv); +} + +static int +archive_write_cpio_header(struct archive_write *a, struct archive_entry *entry) +{ + const char *path; + size_t len; + + if (archive_entry_filetype(entry) == 0) { + archive_set_error(&a->archive, -1, "Filetype required"); + return (ARCHIVE_FAILED); + } + + if (archive_entry_pathname_l(entry, &path, &len, get_sconv(a)) != 0 + && errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Pathname"); + return (ARCHIVE_FATAL); + } + if (len == 0 || path == NULL || path[0] == '\0') { + archive_set_error(&a->archive, -1, "Pathname required"); + return (ARCHIVE_FAILED); + } + + if (!archive_entry_size_is_set(entry) || archive_entry_size(entry) < 0) { + archive_set_error(&a->archive, -1, "Size required"); + return (ARCHIVE_FAILED); + } + return write_header(a, entry); +} + +static int +write_header(struct archive_write *a, struct archive_entry *entry) +{ + struct cpio *cpio; + const char *p, *path; + int pathlength, ret, ret_final; + int64_t ino; + char h[76]; + struct archive_string_conv *sconv; + struct archive_entry *entry_main; + size_t len; + + cpio = (struct cpio *)a->format_data; + ret_final = ARCHIVE_OK; + sconv = get_sconv(a); + +#if defined(_WIN32) && !defined(__CYGWIN__) + /* Make sure the path separators in pathname, hardlink and symlink + * are all slash '/', not the Windows path separator '\'. */ + entry_main = __la_win_entry_in_posix_pathseparator(entry); + if (entry_main == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate ustar data"); + return(ARCHIVE_FATAL); + } + if (entry != entry_main) + entry = entry_main; + else + entry_main = NULL; +#else + entry_main = NULL; +#endif + + ret = archive_entry_pathname_l(entry, &path, &len, sconv); + if (ret != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Pathname"); + ret_final = ARCHIVE_FATAL; + goto exit_write_header; + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate pathname '%s' to %s", + archive_entry_pathname(entry), + archive_string_conversion_charset_name(sconv)); + ret_final = ARCHIVE_WARN; + } + /* Include trailing null. */ + pathlength = (int)len + 1; + + memset(h, 0, sizeof(h)); + format_octal(070707, h + c_magic_offset, c_magic_size); + format_octal(archive_entry_dev(entry), h + c_dev_offset, c_dev_size); + + ino = synthesize_ino_value(cpio, entry); + if (ino < 0) { + archive_set_error(&a->archive, ENOMEM, + "No memory for ino translation table"); + ret_final = ARCHIVE_FATAL; + goto exit_write_header; + } else if (ino > 0777777) { + archive_set_error(&a->archive, ERANGE, + "Too many files for this cpio format"); + ret_final = ARCHIVE_FATAL; + goto exit_write_header; + } + format_octal(ino & 0777777, h + c_ino_offset, c_ino_size); + + /* TODO: Set ret_final to ARCHIVE_WARN if any of these overflow. */ + format_octal(archive_entry_mode(entry), h + c_mode_offset, c_mode_size); + format_octal(archive_entry_uid(entry), h + c_uid_offset, c_uid_size); + format_octal(archive_entry_gid(entry), h + c_gid_offset, c_gid_size); + format_octal(archive_entry_nlink(entry), h + c_nlink_offset, c_nlink_size); + if (archive_entry_filetype(entry) == AE_IFBLK + || archive_entry_filetype(entry) == AE_IFCHR) + format_octal(archive_entry_dev(entry), h + c_rdev_offset, c_rdev_size); + else + format_octal(0, h + c_rdev_offset, c_rdev_size); + format_octal(archive_entry_mtime(entry), h + c_mtime_offset, c_mtime_size); + format_octal(pathlength, h + c_namesize_offset, c_namesize_size); + + /* Non-regular files don't store bodies. */ + if (archive_entry_filetype(entry) != AE_IFREG) + archive_entry_set_size(entry, 0); + + /* Symlinks get the link written as the body of the entry. */ + ret = archive_entry_symlink_l(entry, &p, &len, sconv); + if (ret != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Linkname"); + ret_final = ARCHIVE_FATAL; + goto exit_write_header; + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate linkname '%s' to %s", + archive_entry_symlink(entry), + archive_string_conversion_charset_name(sconv)); + ret_final = ARCHIVE_WARN; + } + if (len > 0 && p != NULL && *p != '\0') + ret = format_octal(strlen(p), h + c_filesize_offset, + c_filesize_size); + else + ret = format_octal(archive_entry_size(entry), + h + c_filesize_offset, c_filesize_size); + if (ret) { + archive_set_error(&a->archive, ERANGE, + "File is too large for cpio format."); + ret_final = ARCHIVE_FAILED; + goto exit_write_header; + } + + ret = __archive_write_output(a, h, sizeof(h)); + if (ret != ARCHIVE_OK) { + ret_final = ARCHIVE_FATAL; + goto exit_write_header; + } + + ret = __archive_write_output(a, path, pathlength); + if (ret != ARCHIVE_OK) { + ret_final = ARCHIVE_FATAL; + goto exit_write_header; + } + + cpio->entry_bytes_remaining = archive_entry_size(entry); + + /* Write the symlink now. */ + if (p != NULL && *p != '\0') { + ret = __archive_write_output(a, p, strlen(p)); + if (ret != ARCHIVE_OK) { + ret_final = ARCHIVE_FATAL; + goto exit_write_header; + } + } +exit_write_header: + archive_entry_free(entry_main); + return (ret_final); +} + +static ssize_t +archive_write_cpio_data(struct archive_write *a, const void *buff, size_t s) +{ + struct cpio *cpio; + int ret; + + cpio = (struct cpio *)a->format_data; + if (s > cpio->entry_bytes_remaining) + s = (size_t)cpio->entry_bytes_remaining; + + ret = __archive_write_output(a, buff, s); + cpio->entry_bytes_remaining -= s; + if (ret >= 0) + return (s); + else + return (ret); +} + +/* + * Format a number into the specified field. + */ +static int +format_octal(int64_t v, void *p, int digits) +{ + int64_t max; + int ret; + + max = (((int64_t)1) << (digits * 3)) - 1; + if (v >= 0 && v <= max) { + format_octal_recursive(v, (char *)p, digits); + ret = 0; + } else { + format_octal_recursive(max, (char *)p, digits); + ret = -1; + } + return (ret); +} + +static int64_t +format_octal_recursive(int64_t v, char *p, int s) +{ + if (s == 0) + return (v); + v = format_octal_recursive(v, p+1, s-1); + *p = '0' + ((char)v & 7); + return (v >> 3); +} + +static int +archive_write_cpio_close(struct archive_write *a) +{ + int er; + struct archive_entry *trailer; + + trailer = archive_entry_new2(NULL); + /* nlink = 1 here for GNU cpio compat. */ + archive_entry_set_nlink(trailer, 1); + archive_entry_set_size(trailer, 0); + archive_entry_set_pathname(trailer, "TRAILER!!!"); + er = write_header(a, trailer); + archive_entry_free(trailer); + return (er); +} + +static int +archive_write_cpio_free(struct archive_write *a) +{ + struct cpio *cpio; + + cpio = (struct cpio *)a->format_data; + free(cpio->ino_list); + free(cpio); + a->format_data = NULL; + return (ARCHIVE_OK); +} + +static int +archive_write_cpio_finish_entry(struct archive_write *a) +{ + struct cpio *cpio; + + cpio = (struct cpio *)a->format_data; + return (__archive_write_nulls(a, + (size_t)cpio->entry_bytes_remaining)); +} diff --git a/c/archive_write_set_format_cpio_newc.c b/c/archive_write_set_format_cpio_newc.c new file mode 100644 --- /dev/null +++ b/c/archive_write_set_format_cpio_newc.c @@ -0,0 +1,457 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * Copyright (c) 2006 Rudolf Marek SYSGO s.r.o. + * Copyright (c) 2011-2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_format_cpio_newc.c 201160 2009-12-29 05:41:57Z kientzle $"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_private.h" +#include "archive_write_private.h" +#include "archive_write_set_format_private.h" + +static ssize_t archive_write_newc_data(struct archive_write *, + const void *buff, size_t s); +static int archive_write_newc_close(struct archive_write *); +static int archive_write_newc_free(struct archive_write *); +static int archive_write_newc_finish_entry(struct archive_write *); +static int archive_write_newc_header(struct archive_write *, + struct archive_entry *); +static int archive_write_newc_options(struct archive_write *, + const char *, const char *); +static int format_hex(int64_t, void *, int); +static int64_t format_hex_recursive(int64_t, char *, int); +static int write_header(struct archive_write *, struct archive_entry *); + +struct cpio { + uint64_t entry_bytes_remaining; + int padding; + + struct archive_string_conv *opt_sconv; + struct archive_string_conv *sconv_default; + int init_default_conversion; +}; + +#define c_magic_offset 0 +#define c_magic_size 6 +#define c_ino_offset 6 +#define c_ino_size 8 +#define c_mode_offset 14 +#define c_mode_size 8 +#define c_uid_offset 22 +#define c_uid_size 8 +#define c_gid_offset 30 +#define c_gid_size 8 +#define c_nlink_offset 38 +#define c_nlink_size 8 +#define c_mtime_offset 46 +#define c_mtime_size 8 +#define c_filesize_offset 54 +#define c_filesize_size 8 +#define c_devmajor_offset 62 +#define c_devmajor_size 8 +#define c_devminor_offset 70 +#define c_devminor_size 8 +#define c_rdevmajor_offset 78 +#define c_rdevmajor_size 8 +#define c_rdevminor_offset 86 +#define c_rdevminor_size 8 +#define c_namesize_offset 94 +#define c_namesize_size 8 +#define c_checksum_offset 102 +#define c_checksum_size 8 +#define c_header_size 110 + +/* Logic trick: difference between 'n' and next multiple of 4 */ +#define PAD4(n) (3 & (1 + ~(n))) + +/* + * Set output format to 'cpio' format. + */ +int +archive_write_set_format_cpio_newc(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct cpio *cpio; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_set_format_cpio_newc"); + + /* If someone else was already registered, unregister them. */ + if (a->format_free != NULL) + (a->format_free)(a); + + cpio = (struct cpio *)calloc(1, sizeof(*cpio)); + if (cpio == NULL) { + archive_set_error(&a->archive, ENOMEM, "Can't allocate cpio data"); + return (ARCHIVE_FATAL); + } + a->format_data = cpio; + a->format_name = "cpio"; + a->format_options = archive_write_newc_options; + a->format_write_header = archive_write_newc_header; + a->format_write_data = archive_write_newc_data; + a->format_finish_entry = archive_write_newc_finish_entry; + a->format_close = archive_write_newc_close; + a->format_free = archive_write_newc_free; + a->archive.archive_format = ARCHIVE_FORMAT_CPIO_SVR4_NOCRC; + a->archive.archive_format_name = "SVR4 cpio nocrc"; + return (ARCHIVE_OK); +} + +static int +archive_write_newc_options(struct archive_write *a, const char *key, + const char *val) +{ + struct cpio *cpio = (struct cpio *)a->format_data; + int ret = ARCHIVE_FAILED; + + if (strcmp(key, "hdrcharset") == 0) { + if (val == NULL || val[0] == 0) + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "%s: hdrcharset option needs a character-set name", + a->format_name); + else { + cpio->opt_sconv = archive_string_conversion_to_charset( + &a->archive, val, 0); + if (cpio->opt_sconv != NULL) + ret = ARCHIVE_OK; + else + ret = ARCHIVE_FATAL; + } + return (ret); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +static struct archive_string_conv * +get_sconv(struct archive_write *a) +{ + struct cpio *cpio; + struct archive_string_conv *sconv; + + cpio = (struct cpio *)a->format_data; + sconv = cpio->opt_sconv; + if (sconv == NULL) { + if (!cpio->init_default_conversion) { + cpio->sconv_default = + archive_string_default_conversion_for_write( + &(a->archive)); + cpio->init_default_conversion = 1; + } + sconv = cpio->sconv_default; + } + return (sconv); +} + +static int +archive_write_newc_header(struct archive_write *a, struct archive_entry *entry) +{ + const char *path; + size_t len; + + if (archive_entry_filetype(entry) == 0) { + archive_set_error(&a->archive, -1, "Filetype required"); + return (ARCHIVE_FAILED); + } + + if (archive_entry_pathname_l(entry, &path, &len, get_sconv(a)) != 0 + && errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Pathname"); + return (ARCHIVE_FATAL); + } + if (len == 0 || path == NULL || path[0] == '\0') { + archive_set_error(&a->archive, -1, "Pathname required"); + return (ARCHIVE_FAILED); + } + + if (archive_entry_hardlink(entry) == NULL + && (!archive_entry_size_is_set(entry) || archive_entry_size(entry) < 0)) { + archive_set_error(&a->archive, -1, "Size required"); + return (ARCHIVE_FAILED); + } + return write_header(a, entry); +} + +static int +write_header(struct archive_write *a, struct archive_entry *entry) +{ + int64_t ino; + struct cpio *cpio; + const char *p, *path; + int pathlength, ret, ret_final; + char h[c_header_size]; + struct archive_string_conv *sconv; + struct archive_entry *entry_main; + size_t len; + int pad; + + cpio = (struct cpio *)a->format_data; + ret_final = ARCHIVE_OK; + sconv = get_sconv(a); + +#if defined(_WIN32) && !defined(__CYGWIN__) + /* Make sure the path separators in pathname, hardlink and symlink + * are all slash '/', not the Windows path separator '\'. */ + entry_main = __la_win_entry_in_posix_pathseparator(entry); + if (entry_main == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate ustar data"); + return(ARCHIVE_FATAL); + } + if (entry != entry_main) + entry = entry_main; + else + entry_main = NULL; +#else + entry_main = NULL; +#endif + + ret = archive_entry_pathname_l(entry, &path, &len, sconv); + if (ret != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Pathname"); + ret_final = ARCHIVE_FATAL; + goto exit_write_header; + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate pathname '%s' to %s", + archive_entry_pathname(entry), + archive_string_conversion_charset_name(sconv)); + ret_final = ARCHIVE_WARN; + } + pathlength = (int)len + 1; /* Include trailing null. */ + + memset(h, 0, c_header_size); + format_hex(0x070701, h + c_magic_offset, c_magic_size); + format_hex(archive_entry_devmajor(entry), h + c_devmajor_offset, + c_devmajor_size); + format_hex(archive_entry_devminor(entry), h + c_devminor_offset, + c_devminor_size); + + ino = archive_entry_ino64(entry); + if (ino > 0xffffffff) { + archive_set_error(&a->archive, ERANGE, + "large inode number truncated"); + ret_final = ARCHIVE_WARN; + } + + /* TODO: Set ret_final to ARCHIVE_WARN if any of these overflow. */ + format_hex(ino & 0xffffffff, h + c_ino_offset, c_ino_size); + format_hex(archive_entry_mode(entry), h + c_mode_offset, c_mode_size); + format_hex(archive_entry_uid(entry), h + c_uid_offset, c_uid_size); + format_hex(archive_entry_gid(entry), h + c_gid_offset, c_gid_size); + format_hex(archive_entry_nlink(entry), h + c_nlink_offset, c_nlink_size); + if (archive_entry_filetype(entry) == AE_IFBLK + || archive_entry_filetype(entry) == AE_IFCHR) { + format_hex(archive_entry_rdevmajor(entry), h + c_rdevmajor_offset, c_rdevmajor_size); + format_hex(archive_entry_rdevminor(entry), h + c_rdevminor_offset, c_rdevminor_size); + } else { + format_hex(0, h + c_rdevmajor_offset, c_rdevmajor_size); + format_hex(0, h + c_rdevminor_offset, c_rdevminor_size); + } + format_hex(archive_entry_mtime(entry), h + c_mtime_offset, c_mtime_size); + format_hex(pathlength, h + c_namesize_offset, c_namesize_size); + format_hex(0, h + c_checksum_offset, c_checksum_size); + + /* Non-regular files don't store bodies. */ + if (archive_entry_filetype(entry) != AE_IFREG) + archive_entry_set_size(entry, 0); + + /* Symlinks get the link written as the body of the entry. */ + ret = archive_entry_symlink_l(entry, &p, &len, sconv); + if (ret != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Likname"); + ret_final = ARCHIVE_FATAL; + goto exit_write_header; + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate linkname '%s' to %s", + archive_entry_symlink(entry), + archive_string_conversion_charset_name(sconv)); + ret_final = ARCHIVE_WARN; + } + if (len > 0 && p != NULL && *p != '\0') + ret = format_hex(strlen(p), h + c_filesize_offset, + c_filesize_size); + else + ret = format_hex(archive_entry_size(entry), + h + c_filesize_offset, c_filesize_size); + if (ret) { + archive_set_error(&a->archive, ERANGE, + "File is too large for this format."); + ret_final = ARCHIVE_FAILED; + goto exit_write_header; + } + + ret = __archive_write_output(a, h, c_header_size); + if (ret != ARCHIVE_OK) { + ret_final = ARCHIVE_FATAL; + goto exit_write_header; + } + + /* Pad pathname to even length. */ + ret = __archive_write_output(a, path, pathlength); + if (ret != ARCHIVE_OK) { + ret_final = ARCHIVE_FATAL; + goto exit_write_header; + } + pad = PAD4(pathlength + c_header_size); + if (pad) { + ret = __archive_write_output(a, "\0\0\0", pad); + if (ret != ARCHIVE_OK) { + ret_final = ARCHIVE_FATAL; + goto exit_write_header; + } + } + + cpio->entry_bytes_remaining = archive_entry_size(entry); + cpio->padding = (int)PAD4(cpio->entry_bytes_remaining); + + /* Write the symlink now. */ + if (p != NULL && *p != '\0') { + ret = __archive_write_output(a, p, strlen(p)); + if (ret != ARCHIVE_OK) { + ret_final = ARCHIVE_FATAL; + goto exit_write_header; + } + pad = PAD4(strlen(p)); + ret = __archive_write_output(a, "\0\0\0", pad); + if (ret != ARCHIVE_OK) { + ret_final = ARCHIVE_FATAL; + goto exit_write_header; + } + } +exit_write_header: + archive_entry_free(entry_main); + return (ret_final); +} + +static ssize_t +archive_write_newc_data(struct archive_write *a, const void *buff, size_t s) +{ + struct cpio *cpio; + int ret; + + cpio = (struct cpio *)a->format_data; + if (s > cpio->entry_bytes_remaining) + s = (size_t)cpio->entry_bytes_remaining; + + ret = __archive_write_output(a, buff, s); + cpio->entry_bytes_remaining -= s; + if (ret >= 0) + return (s); + else + return (ret); +} + +/* + * Format a number into the specified field. + */ +static int +format_hex(int64_t v, void *p, int digits) +{ + int64_t max; + int ret; + + max = (((int64_t)1) << (digits * 4)) - 1; + if (v >= 0 && v <= max) { + format_hex_recursive(v, (char *)p, digits); + ret = 0; + } else { + format_hex_recursive(max, (char *)p, digits); + ret = -1; + } + return (ret); +} + +static int64_t +format_hex_recursive(int64_t v, char *p, int s) +{ + if (s == 0) + return (v); + v = format_hex_recursive(v, p+1, s-1); + *p = "0123456789abcdef"[v & 0xf]; + return (v >> 4); +} + +static int +archive_write_newc_close(struct archive_write *a) +{ + int er; + struct archive_entry *trailer; + + trailer = archive_entry_new(); + archive_entry_set_nlink(trailer, 1); + archive_entry_set_size(trailer, 0); + archive_entry_set_pathname(trailer, "TRAILER!!!"); + /* Bypass the required data checks. */ + er = write_header(a, trailer); + archive_entry_free(trailer); + return (er); +} + +static int +archive_write_newc_free(struct archive_write *a) +{ + struct cpio *cpio; + + cpio = (struct cpio *)a->format_data; + free(cpio); + a->format_data = NULL; + return (ARCHIVE_OK); +} + +static int +archive_write_newc_finish_entry(struct archive_write *a) +{ + struct cpio *cpio; + + cpio = (struct cpio *)a->format_data; + return (__archive_write_nulls(a, + (size_t)cpio->entry_bytes_remaining + cpio->padding)); +} diff --git a/c/archive_write_set_format_filter_by_ext.c b/c/archive_write_set_format_filter_by_ext.c new file mode 100644 --- /dev/null +++ b/c/archive_write_set_format_filter_by_ext.c @@ -0,0 +1,142 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * Copyright (c) 2015 Okhotnikov Kirill + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_format_by_name.c 201168 2009-12-29 06:15:32Z kientzle $"); + +#ifdef HAVE_SYS_TYPES_H +#include +#endif + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_private.h" + +/* A table that maps names to functions. */ +static const +struct { const char *name; int (*format)(struct archive *); int (*filter)(struct archive *); } names[] = +{ + { ".7z", archive_write_set_format_7zip, archive_write_add_filter_none}, + { ".zip", archive_write_set_format_zip, archive_write_add_filter_none}, + { ".jar", archive_write_set_format_zip, archive_write_add_filter_none}, + { ".cpio", archive_write_set_format_cpio, archive_write_add_filter_none}, + { ".iso", archive_write_set_format_iso9660, archive_write_add_filter_none}, +#if defined(__FreeBSD__) || defined(__FreeBSD_kernel__) || defined(__NetBSD__) || defined(__OpenBSD__) + { ".a", archive_write_set_format_ar_bsd, archive_write_add_filter_none}, + { ".ar", archive_write_set_format_ar_bsd, archive_write_add_filter_none}, +#else + { ".a", archive_write_set_format_ar_svr4, archive_write_add_filter_none}, + { ".ar", archive_write_set_format_ar_svr4, archive_write_add_filter_none}, +#endif + { ".tar", archive_write_set_format_pax_restricted, archive_write_add_filter_none}, + { ".tgz", archive_write_set_format_pax_restricted, archive_write_add_filter_gzip}, + { ".tar.gz", archive_write_set_format_pax_restricted, archive_write_add_filter_gzip}, + { ".tar.bz2", archive_write_set_format_pax_restricted, archive_write_add_filter_bzip2}, + { ".tar.xz", archive_write_set_format_pax_restricted, archive_write_add_filter_xz}, + { NULL, NULL, NULL } +}; + +static +int cmpsuff(const char *str, const char *suffix) +{ + size_t length_str, length_suffix; + + if ((str == NULL) || (suffix == NULL)) + return -1; + + length_str = strlen(str); + length_suffix = strlen(suffix); + + if (length_str >= length_suffix) { + return strcmp(str + (length_str - length_suffix), suffix); + } else { + return -1; + } +} + +static int get_array_index(const char *name) +{ + int i; + + for (i = 0; names[i].name != NULL; i++) + { + if (cmpsuff(name, names[i].name) == 0) + return i; + } + return -1; + +} + +int +archive_write_set_format_filter_by_ext(struct archive *a, const char *filename) +{ + int names_index = get_array_index(filename); + + if (names_index >= 0) + { + int format_state = (names[names_index].format)(a); + if (format_state == ARCHIVE_OK) + return ((names[names_index].filter)(a)); + else + return format_state; + } + + archive_set_error(a, EINVAL, "No such format '%s'", filename); + a->state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); +} + +int +archive_write_set_format_filter_by_ext_def(struct archive *a, const char *filename, const char * def_ext) +{ + int names_index = get_array_index(filename); + + if (names_index < 0) + names_index = get_array_index(def_ext); + + if (names_index >= 0) + { + int format_state = (names[names_index].format)(a); + if (format_state == ARCHIVE_OK) + return ((names[names_index].filter)(a)); + else + return format_state; + } + + archive_set_error(a, EINVAL, "No such format '%s'", filename); + a->state = ARCHIVE_STATE_FATAL; + return (ARCHIVE_FATAL); +} + + + + diff --git a/c/archive_write_set_format_gnutar.c b/c/archive_write_set_format_gnutar.c new file mode 100644 --- /dev/null +++ b/c/archive_write_set_format_gnutar.c @@ -0,0 +1,755 @@ +/*- + * Copyright (c) 2010 Nokia Corporation and/or its subsidiary(-ies). + * Author: Jonas Gastal + * Copyright (c) 2011-2012 Michihiro NAKAJIMA + * + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_format_gnu_tar.c 191579 2009-04-27 18:35:03Z gastal $"); + + +#ifdef HAVE_ERRNO_H +#include +#endif +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_private.h" +#include "archive_write_private.h" +#include "archive_write_set_format_private.h" + +struct gnutar { + uint64_t entry_bytes_remaining; + uint64_t entry_padding; + const char * linkname; + size_t linkname_length; + const char * pathname; + size_t pathname_length; + const char * uname; + size_t uname_length; + const char * gname; + size_t gname_length; + struct archive_string_conv *opt_sconv; + struct archive_string_conv *sconv_default; + int init_default_conversion; +}; + +/* + * Define structure of GNU tar header. + */ +#define GNUTAR_name_offset 0 +#define GNUTAR_name_size 100 +#define GNUTAR_mode_offset 100 +#define GNUTAR_mode_size 7 +#define GNUTAR_mode_max_size 8 +#define GNUTAR_uid_offset 108 +#define GNUTAR_uid_size 7 +#define GNUTAR_uid_max_size 8 +#define GNUTAR_gid_offset 116 +#define GNUTAR_gid_size 7 +#define GNUTAR_gid_max_size 8 +#define GNUTAR_size_offset 124 +#define GNUTAR_size_size 11 +#define GNUTAR_size_max_size 12 +#define GNUTAR_mtime_offset 136 +#define GNUTAR_mtime_size 11 +#define GNUTAR_mtime_max_size 11 +#define GNUTAR_checksum_offset 148 +#define GNUTAR_checksum_size 8 +#define GNUTAR_typeflag_offset 156 +#define GNUTAR_typeflag_size 1 +#define GNUTAR_linkname_offset 157 +#define GNUTAR_linkname_size 100 +#define GNUTAR_magic_offset 257 +#define GNUTAR_magic_size 6 +#define GNUTAR_version_offset 263 +#define GNUTAR_version_size 2 +#define GNUTAR_uname_offset 265 +#define GNUTAR_uname_size 32 +#define GNUTAR_gname_offset 297 +#define GNUTAR_gname_size 32 +#define GNUTAR_rdevmajor_offset 329 +#define GNUTAR_rdevmajor_size 6 +#define GNUTAR_rdevmajor_max_size 8 +#define GNUTAR_rdevminor_offset 337 +#define GNUTAR_rdevminor_size 6 +#define GNUTAR_rdevminor_max_size 8 + +/* + * A filled-in copy of the header for initialization. + */ +static const char template_header[] = { + /* name: 100 bytes */ + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0, + /* Mode, null termination: 8 bytes */ + '0','0','0','0','0','0', '0','\0', + /* uid, null termination: 8 bytes */ + '0','0','0','0','0','0', '0','\0', + /* gid, null termination: 8 bytes */ + '0','0','0','0','0','0', '0','\0', + /* size, space termination: 12 bytes */ + '0','0','0','0','0','0','0','0','0','0','0', '\0', + /* mtime, space termination: 12 bytes */ + '0','0','0','0','0','0','0','0','0','0','0', '\0', + /* Initial checksum value: 8 spaces */ + ' ',' ',' ',' ',' ',' ',' ',' ', + /* Typeflag: 1 byte */ + '0', /* '0' = regular file */ + /* Linkname: 100 bytes */ + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0, + /* Magic: 8 bytes */ + 'u','s','t','a','r',' ', ' ','\0', + /* Uname: 32 bytes */ + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + /* Gname: 32 bytes */ + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + /* rdevmajor + null padding: 8 bytes */ + '\0','\0','\0','\0','\0','\0', '\0','\0', + /* rdevminor + null padding: 8 bytes */ + '\0','\0','\0','\0','\0','\0', '\0','\0', + /* Padding: 167 bytes */ + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0 +}; + +static int archive_write_gnutar_options(struct archive_write *, + const char *, const char *); +static int archive_format_gnutar_header(struct archive_write *, char h[512], + struct archive_entry *, int tartype); +static int archive_write_gnutar_header(struct archive_write *, + struct archive_entry *entry); +static ssize_t archive_write_gnutar_data(struct archive_write *a, const void *buff, + size_t s); +static int archive_write_gnutar_free(struct archive_write *); +static int archive_write_gnutar_close(struct archive_write *); +static int archive_write_gnutar_finish_entry(struct archive_write *); +static int format_256(int64_t, char *, int); +static int format_number(int64_t, char *, int size, int maxsize); +static int format_octal(int64_t, char *, int); + +/* + * Set output format to 'GNU tar' format. + */ +int +archive_write_set_format_gnutar(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct gnutar *gnutar; + + gnutar = (struct gnutar *)calloc(1, sizeof(*gnutar)); + if (gnutar == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate gnutar data"); + return (ARCHIVE_FATAL); + } + a->format_data = gnutar; + a->format_name = "gnutar"; + a->format_options = archive_write_gnutar_options; + a->format_write_header = archive_write_gnutar_header; + a->format_write_data = archive_write_gnutar_data; + a->format_close = archive_write_gnutar_close; + a->format_free = archive_write_gnutar_free; + a->format_finish_entry = archive_write_gnutar_finish_entry; + a->archive.archive_format = ARCHIVE_FORMAT_TAR_GNUTAR; + a->archive.archive_format_name = "GNU tar"; + return (ARCHIVE_OK); +} + +static int +archive_write_gnutar_options(struct archive_write *a, const char *key, + const char *val) +{ + struct gnutar *gnutar = (struct gnutar *)a->format_data; + int ret = ARCHIVE_FAILED; + + if (strcmp(key, "hdrcharset") == 0) { + if (val == NULL || val[0] == 0) + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "%s: hdrcharset option needs a character-set name", + a->format_name); + else { + gnutar->opt_sconv = archive_string_conversion_to_charset( + &a->archive, val, 0); + if (gnutar->opt_sconv != NULL) + ret = ARCHIVE_OK; + else + ret = ARCHIVE_FATAL; + } + return (ret); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +static int +archive_write_gnutar_close(struct archive_write *a) +{ + return (__archive_write_nulls(a, 512*2)); +} + +static int +archive_write_gnutar_free(struct archive_write *a) +{ + struct gnutar *gnutar; + + gnutar = (struct gnutar *)a->format_data; + free(gnutar); + a->format_data = NULL; + return (ARCHIVE_OK); +} + +static int +archive_write_gnutar_finish_entry(struct archive_write *a) +{ + struct gnutar *gnutar; + int ret; + + gnutar = (struct gnutar *)a->format_data; + ret = __archive_write_nulls(a, (size_t) + (gnutar->entry_bytes_remaining + gnutar->entry_padding)); + gnutar->entry_bytes_remaining = gnutar->entry_padding = 0; + return (ret); +} + +static ssize_t +archive_write_gnutar_data(struct archive_write *a, const void *buff, size_t s) +{ + struct gnutar *gnutar; + int ret; + + gnutar = (struct gnutar *)a->format_data; + if (s > gnutar->entry_bytes_remaining) + s = (size_t)gnutar->entry_bytes_remaining; + ret = __archive_write_output(a, buff, s); + gnutar->entry_bytes_remaining -= s; + if (ret != ARCHIVE_OK) + return (ret); + return (s); +} + +static int +archive_write_gnutar_header(struct archive_write *a, + struct archive_entry *entry) +{ + char buff[512]; + int r, ret, ret2 = ARCHIVE_OK; + int tartype; + struct gnutar *gnutar; + struct archive_string_conv *sconv; + struct archive_entry *entry_main; + + gnutar = (struct gnutar *)a->format_data; + + /* Setup default string conversion. */ + if (gnutar->opt_sconv == NULL) { + if (!gnutar->init_default_conversion) { + gnutar->sconv_default = + archive_string_default_conversion_for_write( + &(a->archive)); + gnutar->init_default_conversion = 1; + } + sconv = gnutar->sconv_default; + } else + sconv = gnutar->opt_sconv; + + /* Only regular files (not hardlinks) have data. */ + if (archive_entry_hardlink(entry) != NULL || + archive_entry_symlink(entry) != NULL || + !(archive_entry_filetype(entry) == AE_IFREG)) + archive_entry_set_size(entry, 0); + + if (AE_IFDIR == archive_entry_filetype(entry)) { + const char *p; + size_t path_length; + /* + * Ensure a trailing '/'. Modify the entry so + * the client sees the change. + */ +#if defined(_WIN32) && !defined(__CYGWIN__) + const wchar_t *wp; + + wp = archive_entry_pathname_w(entry); + if (wp != NULL && wp[wcslen(wp) -1] != L'/') { + struct archive_wstring ws; + + archive_string_init(&ws); + path_length = wcslen(wp); + if (archive_wstring_ensure(&ws, + path_length + 2) == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate ustar data"); + archive_wstring_free(&ws); + return(ARCHIVE_FATAL); + } + /* Should we keep '\' ? */ + if (wp[path_length -1] == L'\\') + path_length--; + archive_wstrncpy(&ws, wp, path_length); + archive_wstrappend_wchar(&ws, L'/'); + archive_entry_copy_pathname_w(entry, ws.s); + archive_wstring_free(&ws); + p = NULL; + } else +#endif + p = archive_entry_pathname(entry); + /* + * On Windows, this is a backup operation just in + * case getting WCS failed. On POSIX, this is a + * normal operation. + */ + if (p != NULL && p[0] != '\0' && p[strlen(p) - 1] != '/') { + struct archive_string as; + + archive_string_init(&as); + path_length = strlen(p); + if (archive_string_ensure(&as, + path_length + 2) == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate ustar data"); + archive_string_free(&as); + return(ARCHIVE_FATAL); + } +#if defined(_WIN32) && !defined(__CYGWIN__) + /* NOTE: This might break the pathname + * if the current code page is CP932 and + * the pathname includes a character '\' + * as a part of its multibyte pathname. */ + if (p[strlen(p) -1] == '\\') + path_length--; + else +#endif + archive_strncpy(&as, p, path_length); + archive_strappend_char(&as, '/'); + archive_entry_copy_pathname(entry, as.s); + archive_string_free(&as); + } + } + +#if defined(_WIN32) && !defined(__CYGWIN__) + /* Make sure the path separators in pathname, hardlink and symlink + * are all slash '/', not the Windows path separator '\'. */ + entry_main = __la_win_entry_in_posix_pathseparator(entry); + if (entry_main == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate ustar data"); + return(ARCHIVE_FATAL); + } + if (entry != entry_main) + entry = entry_main; + else + entry_main = NULL; +#else + entry_main = NULL; +#endif + r = archive_entry_pathname_l(entry, &(gnutar->pathname), + &(gnutar->pathname_length), sconv); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Pathame"); + ret = ARCHIVE_FATAL; + goto exit_write_header; + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate pathname '%s' to %s", + archive_entry_pathname(entry), + archive_string_conversion_charset_name(sconv)); + ret2 = ARCHIVE_WARN; + } + r = archive_entry_uname_l(entry, &(gnutar->uname), + &(gnutar->uname_length), sconv); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Uname"); + ret = ARCHIVE_FATAL; + goto exit_write_header; + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate uname '%s' to %s", + archive_entry_uname(entry), + archive_string_conversion_charset_name(sconv)); + ret2 = ARCHIVE_WARN; + } + r = archive_entry_gname_l(entry, &(gnutar->gname), + &(gnutar->gname_length), sconv); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Gname"); + ret = ARCHIVE_FATAL; + goto exit_write_header; + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate gname '%s' to %s", + archive_entry_gname(entry), + archive_string_conversion_charset_name(sconv)); + ret2 = ARCHIVE_WARN; + } + + /* If linkname is longer than 100 chars we need to add a 'K' header. */ + r = archive_entry_hardlink_l(entry, &(gnutar->linkname), + &(gnutar->linkname_length), sconv); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Linkname"); + ret = ARCHIVE_FATAL; + goto exit_write_header; + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate linkname '%s' to %s", + archive_entry_hardlink(entry), + archive_string_conversion_charset_name(sconv)); + ret2 = ARCHIVE_WARN; + } + if (gnutar->linkname_length == 0) { + r = archive_entry_symlink_l(entry, &(gnutar->linkname), + &(gnutar->linkname_length), sconv); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Linkname"); + ret = ARCHIVE_FATAL; + goto exit_write_header; + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate linkname '%s' to %s", + archive_entry_hardlink(entry), + archive_string_conversion_charset_name(sconv)); + ret2 = ARCHIVE_WARN; + } + } + if (gnutar->linkname_length > GNUTAR_linkname_size) { + size_t length = gnutar->linkname_length + 1; + struct archive_entry *temp = archive_entry_new2(&a->archive); + + /* Uname/gname here don't really matter since no one reads them; + * these are the values that GNU tar happens to use on FreeBSD. */ + archive_entry_set_uname(temp, "root"); + archive_entry_set_gname(temp, "wheel"); + + archive_entry_set_pathname(temp, "././@LongLink"); + archive_entry_set_size(temp, length); + ret = archive_format_gnutar_header(a, buff, temp, 'K'); + archive_entry_free(temp); + if (ret < ARCHIVE_WARN) + goto exit_write_header; + ret = __archive_write_output(a, buff, 512); + if (ret < ARCHIVE_WARN) + goto exit_write_header; + /* Write name and trailing null byte. */ + ret = __archive_write_output(a, gnutar->linkname, length); + if (ret < ARCHIVE_WARN) + goto exit_write_header; + /* Pad to 512 bytes */ + ret = __archive_write_nulls(a, 0x1ff & (-(ssize_t)length)); + if (ret < ARCHIVE_WARN) + goto exit_write_header; + } + + /* If pathname is longer than 100 chars we need to add an 'L' header. */ + if (gnutar->pathname_length > GNUTAR_name_size) { + const char *pathname = gnutar->pathname; + size_t length = gnutar->pathname_length + 1; + struct archive_entry *temp = archive_entry_new2(&a->archive); + + /* Uname/gname here don't really matter since no one reads them; + * these are the values that GNU tar happens to use on FreeBSD. */ + archive_entry_set_uname(temp, "root"); + archive_entry_set_gname(temp, "wheel"); + + archive_entry_set_pathname(temp, "././@LongLink"); + archive_entry_set_size(temp, length); + ret = archive_format_gnutar_header(a, buff, temp, 'L'); + archive_entry_free(temp); + if (ret < ARCHIVE_WARN) + goto exit_write_header; + ret = __archive_write_output(a, buff, 512); + if(ret < ARCHIVE_WARN) + goto exit_write_header; + /* Write pathname + trailing null byte. */ + ret = __archive_write_output(a, pathname, length); + if(ret < ARCHIVE_WARN) + goto exit_write_header; + /* Pad to multiple of 512 bytes. */ + ret = __archive_write_nulls(a, 0x1ff & (-(ssize_t)length)); + if (ret < ARCHIVE_WARN) + goto exit_write_header; + } + + if (archive_entry_hardlink(entry) != NULL) { + tartype = '1'; + } else + switch (archive_entry_filetype(entry)) { + case AE_IFREG: tartype = '0' ; break; + case AE_IFLNK: tartype = '2' ; break; + case AE_IFCHR: tartype = '3' ; break; + case AE_IFBLK: tartype = '4' ; break; + case AE_IFDIR: tartype = '5' ; break; + case AE_IFIFO: tartype = '6' ; break; + default: /* AE_IFSOCK and unknown */ + __archive_write_entry_filetype_unsupported( + &a->archive, entry, "gnutar"); + ret = ARCHIVE_FAILED; + goto exit_write_header; + } + + ret = archive_format_gnutar_header(a, buff, entry, tartype); + if (ret < ARCHIVE_WARN) + goto exit_write_header; + if (ret2 < ret) + ret = ret2; + ret2 = __archive_write_output(a, buff, 512); + if (ret2 < ARCHIVE_WARN) { + ret = ret2; + goto exit_write_header; + } + if (ret2 < ret) + ret = ret2; + + gnutar->entry_bytes_remaining = archive_entry_size(entry); + gnutar->entry_padding = 0x1ff & (-(int64_t)gnutar->entry_bytes_remaining); +exit_write_header: + archive_entry_free(entry_main); + return (ret); +} + +static int +archive_format_gnutar_header(struct archive_write *a, char h[512], + struct archive_entry *entry, int tartype) +{ + unsigned int checksum; + int i, ret; + size_t copy_length; + const char *p; + struct gnutar *gnutar; + + gnutar = (struct gnutar *)a->format_data; + + ret = 0; + + /* + * The "template header" already includes the signature, + * various end-of-field markers, and other required elements. + */ + memcpy(h, &template_header, 512); + + /* + * Because the block is already null-filled, and strings + * are allowed to exactly fill their destination (without null), + * I use memcpy(dest, src, strlen()) here a lot to copy strings. + */ + + if (tartype == 'K' || tartype == 'L') { + p = archive_entry_pathname(entry); + copy_length = strlen(p); + } else { + p = gnutar->pathname; + copy_length = gnutar->pathname_length; + } + if (copy_length > GNUTAR_name_size) + copy_length = GNUTAR_name_size; + memcpy(h + GNUTAR_name_offset, p, copy_length); + + if ((copy_length = gnutar->linkname_length) > 0) { + if (copy_length > GNUTAR_linkname_size) + copy_length = GNUTAR_linkname_size; + memcpy(h + GNUTAR_linkname_offset, gnutar->linkname, + copy_length); + } + + /* TODO: How does GNU tar handle unames longer than GNUTAR_uname_size? */ + if (tartype == 'K' || tartype == 'L') { + p = archive_entry_uname(entry); + copy_length = strlen(p); + } else { + p = gnutar->uname; + copy_length = gnutar->uname_length; + } + if (copy_length > 0) { + if (copy_length > GNUTAR_uname_size) + copy_length = GNUTAR_uname_size; + memcpy(h + GNUTAR_uname_offset, p, copy_length); + } + + /* TODO: How does GNU tar handle gnames longer than GNUTAR_gname_size? */ + if (tartype == 'K' || tartype == 'L') { + p = archive_entry_gname(entry); + copy_length = strlen(p); + } else { + p = gnutar->gname; + copy_length = gnutar->gname_length; + } + if (copy_length > 0) { + if (strlen(p) > GNUTAR_gname_size) + copy_length = GNUTAR_gname_size; + memcpy(h + GNUTAR_gname_offset, p, copy_length); + } + + /* By truncating the mode here, we ensure it always fits. */ + format_octal(archive_entry_mode(entry) & 07777, + h + GNUTAR_mode_offset, GNUTAR_mode_size); + + /* GNU tar supports base-256 here, so should never overflow. */ + if (format_number(archive_entry_uid(entry), h + GNUTAR_uid_offset, + GNUTAR_uid_size, GNUTAR_uid_max_size)) { + archive_set_error(&a->archive, ERANGE, + "Numeric user ID %jd too large", + (intmax_t)archive_entry_uid(entry)); + ret = ARCHIVE_FAILED; + } + + /* GNU tar supports base-256 here, so should never overflow. */ + if (format_number(archive_entry_gid(entry), h + GNUTAR_gid_offset, + GNUTAR_gid_size, GNUTAR_gid_max_size)) { + archive_set_error(&a->archive, ERANGE, + "Numeric group ID %jd too large", + (intmax_t)archive_entry_gid(entry)); + ret = ARCHIVE_FAILED; + } + + /* GNU tar supports base-256 here, so should never overflow. */ + if (format_number(archive_entry_size(entry), h + GNUTAR_size_offset, + GNUTAR_size_size, GNUTAR_size_max_size)) { + archive_set_error(&a->archive, ERANGE, + "File size out of range"); + ret = ARCHIVE_FAILED; + } + + /* Shouldn't overflow before 2106, since mtime field is 33 bits. */ + format_octal(archive_entry_mtime(entry), + h + GNUTAR_mtime_offset, GNUTAR_mtime_size); + + if (archive_entry_filetype(entry) == AE_IFBLK + || archive_entry_filetype(entry) == AE_IFCHR) { + if (format_octal(archive_entry_rdevmajor(entry), + h + GNUTAR_rdevmajor_offset, + GNUTAR_rdevmajor_size)) { + archive_set_error(&a->archive, ERANGE, + "Major device number too large"); + ret = ARCHIVE_FAILED; + } + + if (format_octal(archive_entry_rdevminor(entry), + h + GNUTAR_rdevminor_offset, + GNUTAR_rdevminor_size)) { + archive_set_error(&a->archive, ERANGE, + "Minor device number too large"); + ret = ARCHIVE_FAILED; + } + } + + h[GNUTAR_typeflag_offset] = tartype; + + checksum = 0; + for (i = 0; i < 512; i++) + checksum += 255 & (unsigned int)h[i]; + h[GNUTAR_checksum_offset + 6] = '\0'; /* Can't be pre-set in the template. */ + /* h[GNUTAR_checksum_offset + 7] = ' '; */ /* This is pre-set in the template. */ + format_octal(checksum, h + GNUTAR_checksum_offset, 6); + return (ret); +} + +/* + * Format a number into a field, falling back to base-256 if necessary. + */ +static int +format_number(int64_t v, char *p, int s, int maxsize) +{ + int64_t limit = ((int64_t)1 << (s*3)); + + if (v < limit) + return (format_octal(v, p, s)); + return (format_256(v, p, maxsize)); +} + +/* + * Format a number into the specified field using base-256. + */ +static int +format_256(int64_t v, char *p, int s) +{ + p += s; + while (s-- > 0) { + *--p = (char)(v & 0xff); + v >>= 8; + } + *p |= 0x80; /* Set the base-256 marker bit. */ + return (0); +} + +/* + * Format a number into the specified field using octal. + */ +static int +format_octal(int64_t v, char *p, int s) +{ + int len = s; + + /* Octal values can't be negative, so use 0. */ + if (v < 0) + v = 0; + + p += s; /* Start at the end and work backwards. */ + while (s-- > 0) { + *--p = (char)('0' + (v & 7)); + v >>= 3; + } + + if (v == 0) + return (0); + + /* If it overflowed, fill field with max value. */ + while (len-- > 0) + *p++ = '7'; + + return (-1); +} diff --git a/c/archive_write_set_format_iso9660.c b/c/archive_write_set_format_iso9660.c new file mode 100644 --- /dev/null +++ b/c/archive_write_set_format_iso9660.c @@ -0,0 +1,8162 @@ +/*- + * Copyright (c) 2009-2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +#ifdef HAVE_SYS_TYPES_H +#include +#endif +#ifdef HAVE_SYS_UTSNAME_H +#include +#endif +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_LIMITS_H +#include +#endif +#include +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#include +#ifdef HAVE_UNISTD_H +#include +#endif +#ifdef HAVE_ZLIB_H +#include +#endif + +#include "archive.h" +#include "archive_endian.h" +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_private.h" +#include "archive_rb.h" +#include "archive_write_private.h" + +#if defined(_WIN32) && !defined(__CYGWIN__) +#define getuid() 0 +#define getgid() 0 +#endif + +/*#define DEBUG 1*/ +#ifdef DEBUG +/* To compare to the ISO image file made by mkisofs. */ +#define COMPAT_MKISOFS 1 +#endif + +#define LOGICAL_BLOCK_BITS 11 +#define LOGICAL_BLOCK_SIZE 2048 +#define PATH_TABLE_BLOCK_SIZE 4096 + +#define SYSTEM_AREA_BLOCK 16 +#define PRIMARY_VOLUME_DESCRIPTOR_BLOCK 1 +#define SUPPLEMENTARY_VOLUME_DESCRIPTOR_BLOCK 1 +#define BOOT_RECORD_DESCRIPTOR_BLOCK 1 +#define VOLUME_DESCRIPTOR_SET_TERMINATOR_BLOCK 1 +#define NON_ISO_FILE_SYSTEM_INFORMATION_BLOCK 1 +#define RRIP_ER_BLOCK 1 +#define PADDING_BLOCK 150 + +#define FD_1_2M_SIZE (1024 * 1200) +#define FD_1_44M_SIZE (1024 * 1440) +#define FD_2_88M_SIZE (1024 * 2880) +#define MULTI_EXTENT_SIZE (ARCHIVE_LITERAL_LL(1) << 32) /* 4Gi bytes. */ +#define MAX_DEPTH 8 +#define RR_CE_SIZE 28 /* SUSP "CE" extension size */ + +#define FILE_FLAG_EXISTENCE 0x01 +#define FILE_FLAG_DIRECTORY 0x02 +#define FILE_FLAG_ASSOCIATED 0x04 +#define FILE_FLAG_RECORD 0x08 +#define FILE_FLAG_PROTECTION 0x10 +#define FILE_FLAG_MULTI_EXTENT 0x80 + +static const char rrip_identifier[] = + "RRIP_1991A"; +static const char rrip_descriptor[] = + "THE ROCK RIDGE INTERCHANGE PROTOCOL PROVIDES SUPPORT FOR " + "POSIX FILE SYSTEM SEMANTICS"; +static const char rrip_source[] = + "PLEASE CONTACT DISC PUBLISHER FOR SPECIFICATION SOURCE. " + "SEE PUBLISHER IDENTIFIER IN PRIMARY VOLUME DESCRIPTOR FOR " + "CONTACT INFORMATION."; +#define RRIP_ER_ID_SIZE (sizeof(rrip_identifier)-1) +#define RRIP_ER_DSC_SIZE (sizeof(rrip_descriptor)-1) +#define RRIP_ER_SRC_SIZE (sizeof(rrip_source)-1) +#define RRIP_ER_SIZE (8 + RRIP_ER_ID_SIZE + \ + RRIP_ER_DSC_SIZE + RRIP_ER_SRC_SIZE) + +static const unsigned char zisofs_magic[8] = { + 0x37, 0xE4, 0x53, 0x96, 0xC9, 0xDB, 0xD6, 0x07 +}; + +#define ZF_HEADER_SIZE 16 /* zisofs header size. */ +#define ZF_LOG2_BS 15 /* log2 block size; 32K bytes. */ +#define ZF_BLOCK_SIZE (1UL << ZF_LOG2_BS) + +/* + * Manage extra records. + */ +struct extr_rec { + int location; + int offset; + unsigned char buf[LOGICAL_BLOCK_SIZE]; + struct extr_rec *next; +}; + +struct ctl_extr_rec { + int use_extr; + unsigned char *bp; + struct isoent *isoent; + unsigned char *ce_ptr; + int cur_len; + int dr_len; + int limit; + int extr_off; + int extr_loc; +}; +#define DR_SAFETY RR_CE_SIZE +#define DR_LIMIT (254 - DR_SAFETY) + +/* + * The relation of struct isofile and isoent and archive_entry. + * + * Primary volume tree --> struct isoent + * | + * v + * struct isofile --> archive_entry + * ^ + * | + * Joliet volume tree --> struct isoent + * + * struct isoent has specific information for volume. + */ + +struct isofile { + /* Used for managing struct isofile list. */ + struct isofile *allnext; + struct isofile *datanext; + /* Used for managing a hardlinked struct isofile list. */ + struct isofile *hlnext; + struct isofile *hardlink_target; + + struct archive_entry *entry; + + /* + * Used for making a directory tree. + */ + struct archive_string parentdir; + struct archive_string basename; + struct archive_string basename_utf16; + struct archive_string symlink; + int dircnt; /* The number of elements of + * its parent directory */ + + /* + * Used for a Directory Record. + */ + struct content { + int64_t offset_of_temp; + int64_t size; + int blocks; + uint32_t location; + /* + * One extent equals one content. + * If this entry has multi extent, `next' variable points + * next content data. + */ + struct content *next; /* next content */ + } content, *cur_content; + int write_content; + + enum { + NO = 0, + BOOT_CATALOG, + BOOT_IMAGE + } boot; + + /* + * Used for a zisofs. + */ + struct { + unsigned char header_size; + unsigned char log2_bs; + uint32_t uncompressed_size; + } zisofs; +}; + +struct isoent { + /* Keep `rbnode' at the first member of struct isoent. */ + struct archive_rb_node rbnode; + + struct isofile *file; + + struct isoent *parent; + /* A list of children.(use chnext) */ + struct { + struct isoent *first; + struct isoent **last; + int cnt; + } children; + struct archive_rb_tree rbtree; + + /* A list of sub directories.(use drnext) */ + struct { + struct isoent *first; + struct isoent **last; + int cnt; + } subdirs; + /* A sorted list of sub directories. */ + struct isoent **children_sorted; + /* Used for managing struct isoent list. */ + struct isoent *chnext; + struct isoent *drnext; + struct isoent *ptnext; + + /* + * Used for making a Directory Record. + */ + int dir_number; + struct { + int vd; + int self; + int parent; + int normal; + } dr_len; + uint32_t dir_location; + int dir_block; + + /* + * Identifier: + * on primary, ISO9660 file/directory name. + * on joliet, UCS2 file/directory name. + * ext_off : offset of identifier extension. + * ext_len : length of identifier extension. + * id_len : byte size of identifier. + * on primary, this is ext_off + ext_len + version length. + * on joliet, this is ext_off + ext_len. + * mb_len : length of multibyte-character of identifier. + * on primary, mb_len and id_len are always the same. + * on joliet, mb_len and id_len are different. + */ + char *identifier; + int ext_off; + int ext_len; + int id_len; + int mb_len; + + /* + * Used for making a Rockridge extension. + * This is a part of Directory Records. + */ + struct isoent *rr_parent; + struct isoent *rr_child; + + /* Extra Record.(which we call in this source file) + * A maximum size of the Directory Record is 254. + * so, if generated RRIP data of a file cannot into a Directory + * Record because of its size, that surplus data relocate this + * Extra Record. + */ + struct { + struct extr_rec *first; + struct extr_rec **last; + struct extr_rec *current; + } extr_rec_list; + + signed int virtual:1; + /* If set to one, this file type is a directory. + * A convenience flag to be used as + * "archive_entry_filetype(isoent->file->entry) == AE_IFDIR". + */ + signed int dir:1; +}; + +struct hardlink { + struct archive_rb_node rbnode; + int nlink; + struct { + struct isofile *first; + struct isofile **last; + } file_list; +}; + +/* + * ISO writer options + */ +struct iso_option { + /* + * Usage : abstract-file= + * Type : string, max 37 bytes + * Default: Not specified + * COMPAT : mkisofs -abstract + * + * Specifies Abstract Filename. + * This file shall be described in the Root Directory + * and containing a abstract statement. + */ + unsigned int abstract_file:1; +#define OPT_ABSTRACT_FILE_DEFAULT 0 /* Not specified */ +#define ABSTRACT_FILE_SIZE 37 + + /* + * Usage : application-id= + * Type : string, max 128 bytes + * Default: Not specified + * COMPAT : mkisofs -A/-appid . + * + * Specifies Application Identifier. + * If the first byte is set to '_'(5F), the remaining + * bytes of this option shall specify an identifier + * for a file containing the identification of the + * application. + * This file shall be described in the Root Directory. + */ + unsigned int application_id:1; +#define OPT_APPLICATION_ID_DEFAULT 0 /* Use default identifier */ +#define APPLICATION_IDENTIFIER_SIZE 128 + + /* + * Usage : !allow-vernum + * Type : boolean + * Default: Enabled + * : Violates the ISO9660 standard if disable. + * COMPAT: mkisofs -N + * + * Allow filenames to use version numbers. + */ + unsigned int allow_vernum:1; +#define OPT_ALLOW_VERNUM_DEFAULT 1 /* Enabled */ + + /* + * Usage : biblio-file= + * Type : string, max 37 bytes + * Default: Not specified + * COMPAT : mkisofs -biblio + * + * Specifies Bibliographic Filename. + * This file shall be described in the Root Directory + * and containing bibliographic records. + */ + unsigned int biblio_file:1; +#define OPT_BIBLIO_FILE_DEFAULT 0 /* Not specified */ +#define BIBLIO_FILE_SIZE 37 + + /* + * Usage : boot= + * Type : string + * Default: Not specified + * COMPAT : mkisofs -b/-eltorito-boot + * + * Specifies "El Torito" boot image file to make + * a bootable CD. + */ + unsigned int boot:1; +#define OPT_BOOT_DEFAULT 0 /* Not specified */ + + /* + * Usage : boot-catalog= + * Type : string + * Default: "boot.catalog" + * COMPAT : mkisofs -c/-eltorito-catalog + * + * Specifies a fullpath of El Torito boot catalog. + */ + unsigned int boot_catalog:1; +#define OPT_BOOT_CATALOG_DEFAULT 0 /* Not specified */ + + /* + * Usage : boot-info-table + * Type : boolean + * Default: Disabled + * COMPAT : mkisofs -boot-info-table + * + * Modify the boot image file specified by `boot' + * option; ISO writer stores boot file information + * into the boot file in ISO image at offset 8 + * through offset 64. + */ + unsigned int boot_info_table:1; +#define OPT_BOOT_INFO_TABLE_DEFAULT 0 /* Disabled */ + + /* + * Usage : boot-load-seg= + * Type : hexadecimal + * Default: Not specified + * COMPAT : mkisofs -boot-load-seg + * + * Specifies a load segment for boot image. + * This is used with no-emulation mode. + */ + unsigned int boot_load_seg:1; +#define OPT_BOOT_LOAD_SEG_DEFAULT 0 /* Not specified */ + + /* + * Usage : boot-load-size= + * Type : decimal + * Default: Not specified + * COMPAT : mkisofs -boot-load-size + * + * Specifies a sector count for boot image. + * This is used with no-emulation mode. + */ + unsigned int boot_load_size:1; +#define OPT_BOOT_LOAD_SIZE_DEFAULT 0 /* Not specified */ + + /* + * Usage : boot-type= + * : 'no-emulation' : 'no emulation' image + * : 'fd' : floppy disk image + * : 'hard-disk' : hard disk image + * Type : string + * Default: Auto detect + * : We check a size of boot image; + * : If the size is just 1.22M/1.44M/2.88M, + * : we assume boot_type is 'fd'; + * : otherwise boot_type is 'no-emulation'. + * COMPAT : + * boot=no-emulation + * mkisofs -no-emul-boot + * boot=fd + * This is a default on the mkisofs. + * boot=hard-disk + * mkisofs -hard-disk-boot + * + * Specifies a type of "El Torito" boot image. + */ + unsigned int boot_type:2; +#define OPT_BOOT_TYPE_AUTO 0 /* auto detect */ +#define OPT_BOOT_TYPE_NO_EMU 1 /* ``no emulation'' image */ +#define OPT_BOOT_TYPE_FD 2 /* floppy disk image */ +#define OPT_BOOT_TYPE_HARD_DISK 3 /* hard disk image */ +#define OPT_BOOT_TYPE_DEFAULT OPT_BOOT_TYPE_AUTO + + /* + * Usage : compression-level= + * Type : decimal + * Default: Not specified + * COMPAT : NONE + * + * Specifies compression level for option zisofs=direct. + */ + unsigned int compression_level:1; +#define OPT_COMPRESSION_LEVEL_DEFAULT 0 /* Not specified */ + + /* + * Usage : copyright-file= + * Type : string, max 37 bytes + * Default: Not specified + * COMPAT : mkisofs -copyright + * + * Specifies Copyright Filename. + * This file shall be described in the Root Directory + * and containing a copyright statement. + */ + unsigned int copyright_file:1; +#define OPT_COPYRIGHT_FILE_DEFAULT 0 /* Not specified */ +#define COPYRIGHT_FILE_SIZE 37 + + /* + * Usage : gid= + * Type : decimal + * Default: Not specified + * COMPAT : mkisofs -gid + * + * Specifies a group id to rewrite the group id of all files. + */ + unsigned int gid:1; +#define OPT_GID_DEFAULT 0 /* Not specified */ + + /* + * Usage : iso-level=[1234] + * Type : decimal + * Default: 1 + * COMPAT : mkisofs -iso-level + * + * Specifies ISO9600 Level. + * Level 1: [DEFAULT] + * - limits each file size less than 4Gi bytes; + * - a File Name shall not contain more than eight + * d-characters or eight d1-characters; + * - a File Name Extension shall not contain more than + * three d-characters or three d1-characters; + * - a Directory Identifier shall not contain more + * than eight d-characters or eight d1-characters. + * Level 2: + * - limits each file size less than 4Giga bytes; + * - a File Name shall not contain more than thirty + * d-characters or thirty d1-characters; + * - a File Name Extension shall not contain more than + * thirty d-characters or thirty d1-characters; + * - a Directory Identifier shall not contain more + * than thirty-one d-characters or thirty-one + * d1-characters. + * Level 3: + * - no limit of file size; use multi extent. + * Level 4: + * - this level 4 simulates mkisofs option + * '-iso-level 4'; + * - crate a enhanced volume as mkisofs doing; + * - allow a File Name to have leading dot; + * - allow a File Name to have all ASCII letters; + * - allow a File Name to have multiple dots; + * - allow more then 8 depths of directory trees; + * - disable a version number to a File Name; + * - disable a forced period to the tail of a File Name; + * - the maximum length of files and directories is raised to 193. + * if rockridge option is disabled, raised to 207. + */ + unsigned int iso_level:3; +#define OPT_ISO_LEVEL_DEFAULT 1 /* ISO Level 1 */ + + /* + * Usage : joliet[=long] + * : !joliet + * : Do not generate Joliet Volume and Records. + * : joliet [DEFAULT] + * : Generates Joliet Volume and Directory Records. + * : [COMPAT: mkisofs -J/-joliet] + * : joliet=long + * : The joliet filenames are up to 103 Unicode + * : characters. + * : This option breaks the Joliet specification. + * : [COMPAT: mkisofs -J -joliet-long] + * Type : boolean/string + * Default: Enabled + * COMPAT : mkisofs -J / -joliet-long + * + * Generates Joliet Volume and Directory Records. + */ + unsigned int joliet:2; +#define OPT_JOLIET_DISABLE 0 /* Not generate Joliet Records. */ +#define OPT_JOLIET_ENABLE 1 /* Generate Joliet Records. */ +#define OPT_JOLIET_LONGNAME 2 /* Use long joliet filenames.*/ +#define OPT_JOLIET_DEFAULT OPT_JOLIET_ENABLE + + /* + * Usage : !limit-depth + * Type : boolean + * Default: Enabled + * : Violates the ISO9660 standard if disable. + * COMPAT : mkisofs -D/-disable-deep-relocation + * + * The number of levels in hierarchy cannot exceed eight. + */ + unsigned int limit_depth:1; +#define OPT_LIMIT_DEPTH_DEFAULT 1 /* Enabled */ + + /* + * Usage : !limit-dirs + * Type : boolean + * Default: Enabled + * : Violates the ISO9660 standard if disable. + * COMPAT : mkisofs -no-limit-pathtables + * + * Limits the number of directories less than 65536 due + * to the size of the Parent Directory Number of Path + * Table. + */ + unsigned int limit_dirs:1; +#define OPT_LIMIT_DIRS_DEFAULT 1 /* Enabled */ + + /* + * Usage : !pad + * Type : boolean + * Default: Enabled + * COMPAT : -pad/-no-pad + * + * Pads the end of the ISO image by null of 300Ki bytes. + */ + unsigned int pad:1; +#define OPT_PAD_DEFAULT 1 /* Enabled */ + + /* + * Usage : publisher= + * Type : string, max 128 bytes + * Default: Not specified + * COMPAT : mkisofs -publisher + * + * Specifies Publisher Identifier. + * If the first byte is set to '_'(5F), the remaining + * bytes of this option shall specify an identifier + * for a file containing the identification of the user. + * This file shall be described in the Root Directory. + */ + unsigned int publisher:1; +#define OPT_PUBLISHER_DEFAULT 0 /* Not specified */ +#define PUBLISHER_IDENTIFIER_SIZE 128 + + /* + * Usage : rockridge + * : !rockridge + * : disable to generate SUSP and RR records. + * : rockridge + * : the same as 'rockridge=useful'. + * : rockridge=strict + * : generate SUSP and RR records. + * : [COMPAT: mkisofs -R] + * : rockridge=useful [DEFAULT] + * : generate SUSP and RR records. + * : [COMPAT: mkisofs -r] + * : NOTE Our rockridge=useful option does not set a zero + * : to uid and gid, you should use application + * : option such as --gid,--gname,--uid and --uname + * : bsdtar options instead. + * Type : boolean/string + * Default: Enabled as rockridge=useful + * COMPAT : mkisofs -r / -R + * + * Generates SUSP and RR records. + */ + unsigned int rr:2; +#define OPT_RR_DISABLED 0 +#define OPT_RR_STRICT 1 +#define OPT_RR_USEFUL 2 +#define OPT_RR_DEFAULT OPT_RR_USEFUL + + /* + * Usage : volume-id= + * Type : string, max 32 bytes + * Default: Not specified + * COMPAT : mkisofs -V + * + * Specifies Volume Identifier. + */ + unsigned int volume_id:1; +#define OPT_VOLUME_ID_DEFAULT 0 /* Use default identifier */ +#define VOLUME_IDENTIFIER_SIZE 32 + + /* + * Usage : !zisofs [DEFAULT] + * : Disable to generate RRIP 'ZF' extension. + * : zisofs + * : Make files zisofs file and generate RRIP 'ZF' + * : extension. So you do not need mkzftree utility + * : for making zisofs. + * : When the file size is less than one Logical Block + * : size, that file will not zisofs'ed since it does + * : reduce an ISO-image size. + * : + * : When you specify option 'boot=', that + * : 'boot-image' file won't be converted to zisofs file. + * Type : boolean + * Default: Disabled + * + * Generates RRIP 'ZF' System Use Entry. + */ + unsigned int zisofs:1; +#define OPT_ZISOFS_DISABLED 0 +#define OPT_ZISOFS_DIRECT 1 +#define OPT_ZISOFS_DEFAULT OPT_ZISOFS_DISABLED + +}; + +struct iso9660 { + /* The creation time of ISO image. */ + time_t birth_time; + /* A file stream of a temporary file, which file contents + * save to until ISO image can be created. */ + int temp_fd; + + struct isofile *cur_file; + struct isoent *cur_dirent; + struct archive_string cur_dirstr; + uint64_t bytes_remaining; + int need_multi_extent; + + /* Temporary string buffer for Joliet extension. */ + struct archive_string utf16be; + struct archive_string mbs; + + struct archive_string_conv *sconv_to_utf16be; + struct archive_string_conv *sconv_from_utf16be; + + /* A list of all of struct isofile entries. */ + struct { + struct isofile *first; + struct isofile **last; + } all_file_list; + + /* A list of struct isofile entries which have its + * contents and are not a directory, a hardlinked file + * and a symlink file. */ + struct { + struct isofile *first; + struct isofile **last; + } data_file_list; + + /* Used for managing to find hardlinking files. */ + struct archive_rb_tree hardlink_rbtree; + + /* Used for making the Path Table Record. */ + struct vdd { + /* the root of entry tree. */ + struct isoent *rootent; + enum vdd_type { + VDD_PRIMARY, + VDD_JOLIET, + VDD_ENHANCED + } vdd_type; + + struct path_table { + struct isoent *first; + struct isoent **last; + struct isoent **sorted; + int cnt; + } *pathtbl; + int max_depth; + + int path_table_block; + int path_table_size; + int location_type_L_path_table; + int location_type_M_path_table; + int total_dir_block; + } primary, joliet; + + /* Used for making a Volume Descriptor. */ + int volume_space_size; + int volume_sequence_number; + int total_file_block; + struct archive_string volume_identifier; + struct archive_string publisher_identifier; + struct archive_string data_preparer_identifier; + struct archive_string application_identifier; + struct archive_string copyright_file_identifier; + struct archive_string abstract_file_identifier; + struct archive_string bibliographic_file_identifier; + + /* Used for making rockridge extensions. */ + int location_rrip_er; + + /* Used for making zisofs. */ + struct { + signed int detect_magic:1; + signed int making:1; + signed int allzero:1; + unsigned char magic_buffer[64]; + int magic_cnt; + +#ifdef HAVE_ZLIB_H + /* + * Copy a compressed file to iso9660.zisofs.temp_fd + * and also copy a uncompressed file(original file) to + * iso9660.temp_fd . If the number of logical block + * of the compressed file is less than the number of + * logical block of the uncompressed file, use it and + * remove the copy of the uncompressed file. + * but if not, we use uncompressed file and remove + * the copy of the compressed file. + */ + uint32_t *block_pointers; + size_t block_pointers_allocated; + int block_pointers_cnt; + int block_pointers_idx; + int64_t total_size; + int64_t block_offset; + + z_stream stream; + int stream_valid; + int64_t remaining; + int compression_level; +#endif + } zisofs; + + struct isoent *directories_too_deep; + int dircnt_max; + + /* Write buffer. */ +#define wb_buffmax() (LOGICAL_BLOCK_SIZE * 32) +#define wb_remaining(a) (((struct iso9660 *)(a)->format_data)->wbuff_remaining) +#define wb_offset(a) (((struct iso9660 *)(a)->format_data)->wbuff_offset \ + + wb_buffmax() - wb_remaining(a)) + unsigned char wbuff[LOGICAL_BLOCK_SIZE * 32]; + size_t wbuff_remaining; + enum { + WB_TO_STREAM, + WB_TO_TEMP + } wbuff_type; + int64_t wbuff_offset; + int64_t wbuff_written; + int64_t wbuff_tail; + + /* 'El Torito' boot data. */ + struct { + /* boot catalog file */ + struct archive_string catalog_filename; + struct isoent *catalog; + /* boot image file */ + struct archive_string boot_filename; + struct isoent *boot; + + unsigned char platform_id; +#define BOOT_PLATFORM_X86 0 +#define BOOT_PLATFORM_PPC 1 +#define BOOT_PLATFORM_MAC 2 + struct archive_string id; + unsigned char media_type; +#define BOOT_MEDIA_NO_EMULATION 0 +#define BOOT_MEDIA_1_2M_DISKETTE 1 +#define BOOT_MEDIA_1_44M_DISKETTE 2 +#define BOOT_MEDIA_2_88M_DISKETTE 3 +#define BOOT_MEDIA_HARD_DISK 4 + unsigned char system_type; + uint16_t boot_load_seg; + uint16_t boot_load_size; +#define BOOT_LOAD_SIZE 4 + } el_torito; + + struct iso_option opt; +}; + +/* + * Types of Volume Descriptor + */ +enum VD_type { + VDT_BOOT_RECORD=0, /* Boot Record Volume Descriptor */ + VDT_PRIMARY=1, /* Primary Volume Descriptor */ + VDT_SUPPLEMENTARY=2, /* Supplementary Volume Descriptor */ + VDT_TERMINATOR=255 /* Volume Descriptor Set Terminator */ +}; + +/* + * Types of Directory Record + */ +enum dir_rec_type { + DIR_REC_VD, /* Stored in Volume Descriptor. */ + DIR_REC_SELF, /* Stored as Current Directory. */ + DIR_REC_PARENT, /* Stored as Parent Directory. */ + DIR_REC_NORMAL /* Stored as Child. */ +}; + +/* + * Kinds of Volume Descriptor Character + */ +enum vdc { + VDC_STD, + VDC_LOWERCASE, + VDC_UCS2, + VDC_UCS2_DIRECT +}; + +/* + * IDentifier Resolver. + * Used for resolving duplicated filenames. + */ +struct idr { + struct idrent { + struct archive_rb_node rbnode; + /* Used in wait_list. */ + struct idrent *wnext; + struct idrent *avail; + + struct isoent *isoent; + int weight; + int noff; + int rename_num; + } *idrent_pool; + + struct archive_rb_tree rbtree; + + struct { + struct idrent *first; + struct idrent **last; + } wait_list; + + int pool_size; + int pool_idx; + int num_size; + int null_size; + + char char_map[0x80]; +}; + +enum char_type { + A_CHAR, + D_CHAR +}; + + +static int iso9660_options(struct archive_write *, + const char *, const char *); +static int iso9660_write_header(struct archive_write *, + struct archive_entry *); +static ssize_t iso9660_write_data(struct archive_write *, + const void *, size_t); +static int iso9660_finish_entry(struct archive_write *); +static int iso9660_close(struct archive_write *); +static int iso9660_free(struct archive_write *); + +static void get_system_identitier(char *, size_t); +static void set_str(unsigned char *, const char *, size_t, char, + const char *); +static inline int joliet_allowed_char(unsigned char, unsigned char); +static int set_str_utf16be(struct archive_write *, unsigned char *, + const char *, size_t, uint16_t, enum vdc); +static int set_str_a_characters_bp(struct archive_write *, + unsigned char *, int, int, const char *, enum vdc); +static int set_str_d_characters_bp(struct archive_write *, + unsigned char *, int, int, const char *, enum vdc); +static void set_VD_bp(unsigned char *, enum VD_type, unsigned char); +static inline void set_unused_field_bp(unsigned char *, int, int); + +static unsigned char *extra_open_record(unsigned char *, int, + struct isoent *, struct ctl_extr_rec *); +static void extra_close_record(struct ctl_extr_rec *, int); +static unsigned char * extra_next_record(struct ctl_extr_rec *, int); +static unsigned char *extra_get_record(struct isoent *, int *, int *, int *); +static void extra_tell_used_size(struct ctl_extr_rec *, int); +static int extra_setup_location(struct isoent *, int); +static int set_directory_record_rr(unsigned char *, int, + struct isoent *, struct iso9660 *, enum dir_rec_type); +static int set_directory_record(unsigned char *, size_t, + struct isoent *, struct iso9660 *, enum dir_rec_type, + enum vdd_type); +static inline int get_dir_rec_size(struct iso9660 *, struct isoent *, + enum dir_rec_type, enum vdd_type); +static inline unsigned char *wb_buffptr(struct archive_write *); +static int wb_write_out(struct archive_write *); +static int wb_consume(struct archive_write *, size_t); +#ifdef HAVE_ZLIB_H +static int wb_set_offset(struct archive_write *, int64_t); +#endif +static int write_null(struct archive_write *, size_t); +static int write_VD_terminator(struct archive_write *); +static int set_file_identifier(unsigned char *, int, int, enum vdc, + struct archive_write *, struct vdd *, + struct archive_string *, const char *, int, + enum char_type); +static int write_VD(struct archive_write *, struct vdd *); +static int write_VD_boot_record(struct archive_write *); +static int write_information_block(struct archive_write *); +static int write_path_table(struct archive_write *, int, + struct vdd *); +static int write_directory_descriptors(struct archive_write *, + struct vdd *); +static int write_file_descriptors(struct archive_write *); +static int write_rr_ER(struct archive_write *); +static void calculate_path_table_size(struct vdd *); + +static void isofile_init_entry_list(struct iso9660 *); +static void isofile_add_entry(struct iso9660 *, struct isofile *); +static void isofile_free_all_entries(struct iso9660 *); +static void isofile_init_entry_data_file_list(struct iso9660 *); +static void isofile_add_data_file(struct iso9660 *, struct isofile *); +static struct isofile * isofile_new(struct archive_write *, + struct archive_entry *); +static void isofile_free(struct isofile *); +static int isofile_gen_utility_names(struct archive_write *, + struct isofile *); +static int isofile_register_hardlink(struct archive_write *, + struct isofile *); +static void isofile_connect_hardlink_files(struct iso9660 *); +static void isofile_init_hardlinks(struct iso9660 *); +static void isofile_free_hardlinks(struct iso9660 *); + +static struct isoent *isoent_new(struct isofile *); +static int isoent_clone_tree(struct archive_write *, + struct isoent **, struct isoent *); +static void _isoent_free(struct isoent *isoent); +static void isoent_free_all(struct isoent *); +static struct isoent * isoent_create_virtual_dir(struct archive_write *, + struct iso9660 *, const char *); +static int isoent_cmp_node(const struct archive_rb_node *, + const struct archive_rb_node *); +static int isoent_cmp_key(const struct archive_rb_node *, + const void *); +static int isoent_add_child_head(struct isoent *, struct isoent *); +static int isoent_add_child_tail(struct isoent *, struct isoent *); +static void isoent_remove_child(struct isoent *, struct isoent *); +static void isoent_setup_directory_location(struct iso9660 *, + int, struct vdd *); +static void isoent_setup_file_location(struct iso9660 *, int); +static int get_path_component(char *, size_t, const char *); +static int isoent_tree(struct archive_write *, struct isoent **); +static struct isoent *isoent_find_child(struct isoent *, const char *); +static struct isoent *isoent_find_entry(struct isoent *, const char *); +static void idr_relaxed_filenames(char *); +static void idr_init(struct iso9660 *, struct vdd *, struct idr *); +static void idr_cleanup(struct idr *); +static int idr_ensure_poolsize(struct archive_write *, struct idr *, + int); +static int idr_start(struct archive_write *, struct idr *, + int, int, int, int, const struct archive_rb_tree_ops *); +static void idr_register(struct idr *, struct isoent *, int, + int); +static void idr_extend_identifier(struct idrent *, int, int); +static void idr_resolve(struct idr *, void (*)(unsigned char *, int)); +static void idr_set_num(unsigned char *, int); +static void idr_set_num_beutf16(unsigned char *, int); +static int isoent_gen_iso9660_identifier(struct archive_write *, + struct isoent *, struct idr *); +static int isoent_gen_joliet_identifier(struct archive_write *, + struct isoent *, struct idr *); +static int isoent_cmp_iso9660_identifier(const struct isoent *, + const struct isoent *); +static int isoent_cmp_node_iso9660(const struct archive_rb_node *, + const struct archive_rb_node *); +static int isoent_cmp_key_iso9660(const struct archive_rb_node *, + const void *); +static int isoent_cmp_joliet_identifier(const struct isoent *, + const struct isoent *); +static int isoent_cmp_node_joliet(const struct archive_rb_node *, + const struct archive_rb_node *); +static int isoent_cmp_key_joliet(const struct archive_rb_node *, + const void *); +static inline void path_table_add_entry(struct path_table *, struct isoent *); +static inline struct isoent * path_table_last_entry(struct path_table *); +static int isoent_make_path_table(struct archive_write *); +static int isoent_find_out_boot_file(struct archive_write *, + struct isoent *); +static int isoent_create_boot_catalog(struct archive_write *, + struct isoent *); +static size_t fd_boot_image_size(int); +static int make_boot_catalog(struct archive_write *); +static int setup_boot_information(struct archive_write *); + +static int zisofs_init(struct archive_write *, struct isofile *); +static void zisofs_detect_magic(struct archive_write *, + const void *, size_t); +static int zisofs_write_to_temp(struct archive_write *, + const void *, size_t); +static int zisofs_finish_entry(struct archive_write *); +static int zisofs_rewind_boot_file(struct archive_write *); +static int zisofs_free(struct archive_write *); + +int +archive_write_set_format_iso9660(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct iso9660 *iso9660; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_set_format_iso9660"); + + /* If another format was already registered, unregister it. */ + if (a->format_free != NULL) + (a->format_free)(a); + + iso9660 = calloc(1, sizeof(*iso9660)); + if (iso9660 == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate iso9660 data"); + return (ARCHIVE_FATAL); + } + iso9660->birth_time = 0; + iso9660->temp_fd = -1; + iso9660->cur_file = NULL; + iso9660->primary.max_depth = 0; + iso9660->primary.vdd_type = VDD_PRIMARY; + iso9660->primary.pathtbl = NULL; + iso9660->joliet.rootent = NULL; + iso9660->joliet.max_depth = 0; + iso9660->joliet.vdd_type = VDD_JOLIET; + iso9660->joliet.pathtbl = NULL; + isofile_init_entry_list(iso9660); + isofile_init_entry_data_file_list(iso9660); + isofile_init_hardlinks(iso9660); + iso9660->directories_too_deep = NULL; + iso9660->dircnt_max = 1; + iso9660->wbuff_remaining = wb_buffmax(); + iso9660->wbuff_type = WB_TO_TEMP; + iso9660->wbuff_offset = 0; + iso9660->wbuff_written = 0; + iso9660->wbuff_tail = 0; + archive_string_init(&(iso9660->utf16be)); + archive_string_init(&(iso9660->mbs)); + + /* + * Init Identifiers used for PVD and SVD. + */ + archive_string_init(&(iso9660->volume_identifier)); + archive_strcpy(&(iso9660->volume_identifier), "CDROM"); + archive_string_init(&(iso9660->publisher_identifier)); + archive_string_init(&(iso9660->data_preparer_identifier)); + archive_string_init(&(iso9660->application_identifier)); + archive_strcpy(&(iso9660->application_identifier), + archive_version_string()); + archive_string_init(&(iso9660->copyright_file_identifier)); + archive_string_init(&(iso9660->abstract_file_identifier)); + archive_string_init(&(iso9660->bibliographic_file_identifier)); + + /* + * Init El Torito bootable CD variables. + */ + archive_string_init(&(iso9660->el_torito.catalog_filename)); + iso9660->el_torito.catalog = NULL; + /* Set default file name of boot catalog */ + archive_strcpy(&(iso9660->el_torito.catalog_filename), + "boot.catalog"); + archive_string_init(&(iso9660->el_torito.boot_filename)); + iso9660->el_torito.boot = NULL; + iso9660->el_torito.platform_id = BOOT_PLATFORM_X86; + archive_string_init(&(iso9660->el_torito.id)); + iso9660->el_torito.boot_load_seg = 0; + iso9660->el_torito.boot_load_size = BOOT_LOAD_SIZE; + + /* + * Init zisofs variables. + */ +#ifdef HAVE_ZLIB_H + iso9660->zisofs.block_pointers = NULL; + iso9660->zisofs.block_pointers_allocated = 0; + iso9660->zisofs.stream_valid = 0; + iso9660->zisofs.compression_level = 9; + memset(&(iso9660->zisofs.stream), 0, + sizeof(iso9660->zisofs.stream)); +#endif + + /* + * Set default value of iso9660 options. + */ + iso9660->opt.abstract_file = OPT_ABSTRACT_FILE_DEFAULT; + iso9660->opt.application_id = OPT_APPLICATION_ID_DEFAULT; + iso9660->opt.allow_vernum = OPT_ALLOW_VERNUM_DEFAULT; + iso9660->opt.biblio_file = OPT_BIBLIO_FILE_DEFAULT; + iso9660->opt.boot = OPT_BOOT_DEFAULT; + iso9660->opt.boot_catalog = OPT_BOOT_CATALOG_DEFAULT; + iso9660->opt.boot_info_table = OPT_BOOT_INFO_TABLE_DEFAULT; + iso9660->opt.boot_load_seg = OPT_BOOT_LOAD_SEG_DEFAULT; + iso9660->opt.boot_load_size = OPT_BOOT_LOAD_SIZE_DEFAULT; + iso9660->opt.boot_type = OPT_BOOT_TYPE_DEFAULT; + iso9660->opt.compression_level = OPT_COMPRESSION_LEVEL_DEFAULT; + iso9660->opt.copyright_file = OPT_COPYRIGHT_FILE_DEFAULT; + iso9660->opt.iso_level = OPT_ISO_LEVEL_DEFAULT; + iso9660->opt.joliet = OPT_JOLIET_DEFAULT; + iso9660->opt.limit_depth = OPT_LIMIT_DEPTH_DEFAULT; + iso9660->opt.limit_dirs = OPT_LIMIT_DIRS_DEFAULT; + iso9660->opt.pad = OPT_PAD_DEFAULT; + iso9660->opt.publisher = OPT_PUBLISHER_DEFAULT; + iso9660->opt.rr = OPT_RR_DEFAULT; + iso9660->opt.volume_id = OPT_VOLUME_ID_DEFAULT; + iso9660->opt.zisofs = OPT_ZISOFS_DEFAULT; + + /* Create the root directory. */ + iso9660->primary.rootent = + isoent_create_virtual_dir(a, iso9660, ""); + if (iso9660->primary.rootent == NULL) { + free(iso9660); + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + iso9660->primary.rootent->parent = iso9660->primary.rootent; + iso9660->cur_dirent = iso9660->primary.rootent; + archive_string_init(&(iso9660->cur_dirstr)); + archive_string_ensure(&(iso9660->cur_dirstr), 1); + iso9660->cur_dirstr.s[0] = 0; + iso9660->sconv_to_utf16be = NULL; + iso9660->sconv_from_utf16be = NULL; + + a->format_data = iso9660; + a->format_name = "iso9660"; + a->format_options = iso9660_options; + a->format_write_header = iso9660_write_header; + a->format_write_data = iso9660_write_data; + a->format_finish_entry = iso9660_finish_entry; + a->format_close = iso9660_close; + a->format_free = iso9660_free; + a->archive.archive_format = ARCHIVE_FORMAT_ISO9660; + a->archive.archive_format_name = "ISO9660"; + + return (ARCHIVE_OK); +} + +static int +get_str_opt(struct archive_write *a, struct archive_string *s, + size_t maxsize, const char *key, const char *value) +{ + + if (strlen(value) > maxsize) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Value is longer than %zu characters " + "for option ``%s''", maxsize, key); + return (ARCHIVE_FATAL); + } + archive_strcpy(s, value); + return (ARCHIVE_OK); +} + +static int +get_num_opt(struct archive_write *a, int *num, int high, int low, + const char *key, const char *value) +{ + const char *p = value; + int data = 0; + int neg = 0; + + if (p == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Invalid value(empty) for option ``%s''", key); + return (ARCHIVE_FATAL); + } + if (*p == '-') { + neg = 1; + p++; + } + while (*p) { + if (*p >= '0' && *p <= '9') + data = data * 10 + *p - '0'; + else { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Invalid value for option ``%s''", key); + return (ARCHIVE_FATAL); + } + if (data > high) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Invalid value(over %d) for " + "option ``%s''", high, key); + return (ARCHIVE_FATAL); + } + if (data < low) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Invalid value(under %d) for " + "option ``%s''", low, key); + return (ARCHIVE_FATAL); + } + p++; + } + if (neg) + data *= -1; + *num = data; + + return (ARCHIVE_OK); +} + +static int +iso9660_options(struct archive_write *a, const char *key, const char *value) +{ + struct iso9660 *iso9660 = a->format_data; + const char *p; + int r; + + switch (key[0]) { + case 'a': + if (strcmp(key, "abstract-file") == 0) { + r = get_str_opt(a, + &(iso9660->abstract_file_identifier), + ABSTRACT_FILE_SIZE, key, value); + iso9660->opt.abstract_file = r == ARCHIVE_OK; + return (r); + } + if (strcmp(key, "application-id") == 0) { + r = get_str_opt(a, + &(iso9660->application_identifier), + APPLICATION_IDENTIFIER_SIZE, key, value); + iso9660->opt.application_id = r == ARCHIVE_OK; + return (r); + } + if (strcmp(key, "allow-vernum") == 0) { + iso9660->opt.allow_vernum = value != NULL; + return (ARCHIVE_OK); + } + break; + case 'b': + if (strcmp(key, "biblio-file") == 0) { + r = get_str_opt(a, + &(iso9660->bibliographic_file_identifier), + BIBLIO_FILE_SIZE, key, value); + iso9660->opt.biblio_file = r == ARCHIVE_OK; + return (r); + } + if (strcmp(key, "boot") == 0) { + if (value == NULL) + iso9660->opt.boot = 0; + else { + iso9660->opt.boot = 1; + archive_strcpy( + &(iso9660->el_torito.boot_filename), + value); + } + return (ARCHIVE_OK); + } + if (strcmp(key, "boot-catalog") == 0) { + r = get_str_opt(a, + &(iso9660->el_torito.catalog_filename), + 1024, key, value); + iso9660->opt.boot_catalog = r == ARCHIVE_OK; + return (r); + } + if (strcmp(key, "boot-info-table") == 0) { + iso9660->opt.boot_info_table = value != NULL; + return (ARCHIVE_OK); + } + if (strcmp(key, "boot-load-seg") == 0) { + uint32_t seg; + + iso9660->opt.boot_load_seg = 0; + if (value == NULL) + goto invalid_value; + seg = 0; + p = value; + if (p[0] == '0' && (p[1] == 'x' || p[1] == 'X')) + p += 2; + while (*p) { + if (seg) + seg <<= 4; + if (*p >= 'A' && *p <= 'F') + seg += *p - 'A' + 0x0a; + else if (*p >= 'a' && *p <= 'f') + seg += *p - 'a' + 0x0a; + else if (*p >= '0' && *p <= '9') + seg += *p - '0'; + else + goto invalid_value; + if (seg > 0xffff) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Invalid value(over 0xffff) for " + "option ``%s''", key); + return (ARCHIVE_FATAL); + } + p++; + } + iso9660->el_torito.boot_load_seg = (uint16_t)seg; + iso9660->opt.boot_load_seg = 1; + return (ARCHIVE_OK); + } + if (strcmp(key, "boot-load-size") == 0) { + int num = 0; + r = get_num_opt(a, &num, 0xffff, 1, key, value); + iso9660->opt.boot_load_size = r == ARCHIVE_OK; + if (r != ARCHIVE_OK) + return (ARCHIVE_FATAL); + iso9660->el_torito.boot_load_size = (uint16_t)num; + return (ARCHIVE_OK); + } + if (strcmp(key, "boot-type") == 0) { + if (value == NULL) + goto invalid_value; + if (strcmp(value, "no-emulation") == 0) + iso9660->opt.boot_type = OPT_BOOT_TYPE_NO_EMU; + else if (strcmp(value, "fd") == 0) + iso9660->opt.boot_type = OPT_BOOT_TYPE_FD; + else if (strcmp(value, "hard-disk") == 0) + iso9660->opt.boot_type = OPT_BOOT_TYPE_HARD_DISK; + else + goto invalid_value; + return (ARCHIVE_OK); + } + break; + case 'c': + if (strcmp(key, "compression-level") == 0) { +#ifdef HAVE_ZLIB_H + if (value == NULL || + !(value[0] >= '0' && value[0] <= '9') || + value[1] != '\0') + goto invalid_value; + iso9660->zisofs.compression_level = value[0] - '0'; + iso9660->opt.compression_level = 1; + return (ARCHIVE_OK); +#else + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Option ``%s'' " + "is not supported on this platform.", key); + return (ARCHIVE_FATAL); +#endif + } + if (strcmp(key, "copyright-file") == 0) { + r = get_str_opt(a, + &(iso9660->copyright_file_identifier), + COPYRIGHT_FILE_SIZE, key, value); + iso9660->opt.copyright_file = r == ARCHIVE_OK; + return (r); + } +#ifdef DEBUG + /* Specifies Volume creation date and time; + * year(4),month(2),day(2),hour(2),minute(2),second(2). + * e.g. "20090929033757" + */ + if (strcmp(key, "creation") == 0) { + struct tm tm; + char buf[5]; + + p = value; + if (p == NULL || strlen(p) < 14) + goto invalid_value; + memset(&tm, 0, sizeof(tm)); + memcpy(buf, p, 4); buf[4] = '\0'; p += 4; + tm.tm_year = strtol(buf, NULL, 10) - 1900; + memcpy(buf, p, 2); buf[2] = '\0'; p += 2; + tm.tm_mon = strtol(buf, NULL, 10) - 1; + memcpy(buf, p, 2); buf[2] = '\0'; p += 2; + tm.tm_mday = strtol(buf, NULL, 10); + memcpy(buf, p, 2); buf[2] = '\0'; p += 2; + tm.tm_hour = strtol(buf, NULL, 10); + memcpy(buf, p, 2); buf[2] = '\0'; p += 2; + tm.tm_min = strtol(buf, NULL, 10); + memcpy(buf, p, 2); buf[2] = '\0'; + tm.tm_sec = strtol(buf, NULL, 10); + iso9660->birth_time = mktime(&tm); + return (ARCHIVE_OK); + } +#endif + break; + case 'i': + if (strcmp(key, "iso-level") == 0) { + if (value != NULL && value[1] == '\0' && + (value[0] >= '1' && value[0] <= '4')) { + iso9660->opt.iso_level = value[0]-'0'; + return (ARCHIVE_OK); + } + goto invalid_value; + } + break; + case 'j': + if (strcmp(key, "joliet") == 0) { + if (value == NULL) + iso9660->opt.joliet = OPT_JOLIET_DISABLE; + else if (strcmp(value, "1") == 0) + iso9660->opt.joliet = OPT_JOLIET_ENABLE; + else if (strcmp(value, "long") == 0) + iso9660->opt.joliet = OPT_JOLIET_LONGNAME; + else + goto invalid_value; + return (ARCHIVE_OK); + } + break; + case 'l': + if (strcmp(key, "limit-depth") == 0) { + iso9660->opt.limit_depth = value != NULL; + return (ARCHIVE_OK); + } + if (strcmp(key, "limit-dirs") == 0) { + iso9660->opt.limit_dirs = value != NULL; + return (ARCHIVE_OK); + } + break; + case 'p': + if (strcmp(key, "pad") == 0) { + iso9660->opt.pad = value != NULL; + return (ARCHIVE_OK); + } + if (strcmp(key, "publisher") == 0) { + r = get_str_opt(a, + &(iso9660->publisher_identifier), + PUBLISHER_IDENTIFIER_SIZE, key, value); + iso9660->opt.publisher = r == ARCHIVE_OK; + return (r); + } + break; + case 'r': + if (strcmp(key, "rockridge") == 0 || + strcmp(key, "Rockridge") == 0) { + if (value == NULL) + iso9660->opt.rr = OPT_RR_DISABLED; + else if (strcmp(value, "1") == 0) + iso9660->opt.rr = OPT_RR_USEFUL; + else if (strcmp(value, "strict") == 0) + iso9660->opt.rr = OPT_RR_STRICT; + else if (strcmp(value, "useful") == 0) + iso9660->opt.rr = OPT_RR_USEFUL; + else + goto invalid_value; + return (ARCHIVE_OK); + } + break; + case 'v': + if (strcmp(key, "volume-id") == 0) { + r = get_str_opt(a, &(iso9660->volume_identifier), + VOLUME_IDENTIFIER_SIZE, key, value); + iso9660->opt.volume_id = r == ARCHIVE_OK; + return (r); + } + break; + case 'z': + if (strcmp(key, "zisofs") == 0) { + if (value == NULL) + iso9660->opt.zisofs = OPT_ZISOFS_DISABLED; + else { +#ifdef HAVE_ZLIB_H + iso9660->opt.zisofs = OPT_ZISOFS_DIRECT; +#else + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "``zisofs'' " + "is not supported on this platform."); + return (ARCHIVE_FATAL); +#endif + } + return (ARCHIVE_OK); + } + break; + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); + +invalid_value: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Invalid value for option ``%s''", key); + return (ARCHIVE_FAILED); +} + +static int +iso9660_write_header(struct archive_write *a, struct archive_entry *entry) +{ + struct iso9660 *iso9660; + struct isofile *file; + struct isoent *isoent; + int r, ret = ARCHIVE_OK; + + iso9660 = a->format_data; + + iso9660->cur_file = NULL; + iso9660->bytes_remaining = 0; + iso9660->need_multi_extent = 0; + if (archive_entry_filetype(entry) == AE_IFLNK + && iso9660->opt.rr == OPT_RR_DISABLED) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Ignore symlink file."); + iso9660->cur_file = NULL; + return (ARCHIVE_WARN); + } + if (archive_entry_filetype(entry) == AE_IFREG && + archive_entry_size(entry) >= MULTI_EXTENT_SIZE) { + if (iso9660->opt.iso_level < 3) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Ignore over %lld bytes file. " + "This file too large.", + MULTI_EXTENT_SIZE); + iso9660->cur_file = NULL; + return (ARCHIVE_WARN); + } + iso9660->need_multi_extent = 1; + } + + file = isofile_new(a, entry); + if (file == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate data"); + return (ARCHIVE_FATAL); + } + r = isofile_gen_utility_names(a, file); + if (r < ARCHIVE_WARN) { + isofile_free(file); + return (r); + } + else if (r < ret) + ret = r; + + /* + * Ignore a path which looks like the top of directory name + * since we have already made the root directory of an ISO image. + */ + if (archive_strlen(&(file->parentdir)) == 0 && + archive_strlen(&(file->basename)) == 0) { + isofile_free(file); + return (r); + } + + isofile_add_entry(iso9660, file); + isoent = isoent_new(file); + if (isoent == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate data"); + return (ARCHIVE_FATAL); + } + if (isoent->file->dircnt > iso9660->dircnt_max) + iso9660->dircnt_max = isoent->file->dircnt; + + /* Add the current file into tree */ + r = isoent_tree(a, &isoent); + if (r != ARCHIVE_OK) + return (r); + + /* If there is the same file in tree and + * the current file is older than the file in tree. + * So we don't need the current file data anymore. */ + if (isoent->file != file) + return (ARCHIVE_OK); + + /* Non regular files contents are unneeded to be saved to + * temporary files. */ + if (archive_entry_filetype(file->entry) != AE_IFREG) + return (ret); + + /* + * Set the current file to cur_file to read its contents. + */ + iso9660->cur_file = file; + + if (archive_entry_nlink(file->entry) > 1) { + r = isofile_register_hardlink(a, file); + if (r != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + + /* + * Prepare to save the contents of the file. + */ + if (iso9660->temp_fd < 0) { + iso9660->temp_fd = __archive_mktemp(NULL); + if (iso9660->temp_fd < 0) { + archive_set_error(&a->archive, errno, + "Couldn't create temporary file"); + return (ARCHIVE_FATAL); + } + } + + /* Save an offset of current file in temporary file. */ + file->content.offset_of_temp = wb_offset(a); + file->cur_content = &(file->content); + r = zisofs_init(a, file); + if (r < ret) + ret = r; + iso9660->bytes_remaining = archive_entry_size(file->entry); + + return (ret); +} + +static int +write_to_temp(struct archive_write *a, const void *buff, size_t s) +{ + struct iso9660 *iso9660 = a->format_data; + ssize_t written; + const unsigned char *b; + + b = (const unsigned char *)buff; + while (s) { + written = write(iso9660->temp_fd, b, s); + if (written < 0) { + archive_set_error(&a->archive, errno, + "Can't write to temporary file"); + return (ARCHIVE_FATAL); + } + s -= written; + b += written; + } + return (ARCHIVE_OK); +} + +static int +wb_write_to_temp(struct archive_write *a, const void *buff, size_t s) +{ + const char *xp = buff; + size_t xs = s; + + /* + * If a written data size is big enough to use system-call + * and there is no waiting data, this calls write_to_temp() in + * order to reduce a extra memory copy. + */ + if (wb_remaining(a) == wb_buffmax() && s > (1024 * 16)) { + struct iso9660 *iso9660 = (struct iso9660 *)a->format_data; + xs = s % LOGICAL_BLOCK_SIZE; + iso9660->wbuff_offset += s - xs; + if (write_to_temp(a, buff, s - xs) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + if (xs == 0) + return (ARCHIVE_OK); + xp += s - xs; + } + + while (xs) { + size_t size = xs; + if (size > wb_remaining(a)) + size = wb_remaining(a); + memcpy(wb_buffptr(a), xp, size); + if (wb_consume(a, size) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + xs -= size; + xp += size; + } + return (ARCHIVE_OK); +} + +static int +wb_write_padding_to_temp(struct archive_write *a, int64_t csize) +{ + size_t ns; + int ret; + + ns = (size_t)(csize % LOGICAL_BLOCK_SIZE); + if (ns != 0) + ret = write_null(a, LOGICAL_BLOCK_SIZE - ns); + else + ret = ARCHIVE_OK; + return (ret); +} + +static ssize_t +write_iso9660_data(struct archive_write *a, const void *buff, size_t s) +{ + struct iso9660 *iso9660 = a->format_data; + size_t ws; + + if (iso9660->temp_fd < 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Couldn't create temporary file"); + return (ARCHIVE_FATAL); + } + + ws = s; + if (iso9660->need_multi_extent && + (iso9660->cur_file->cur_content->size + ws) >= + (MULTI_EXTENT_SIZE - LOGICAL_BLOCK_SIZE)) { + struct content *con; + size_t ts; + + ts = (size_t)(MULTI_EXTENT_SIZE - LOGICAL_BLOCK_SIZE - + iso9660->cur_file->cur_content->size); + + if (iso9660->zisofs.detect_magic) + zisofs_detect_magic(a, buff, ts); + + if (iso9660->zisofs.making) { + if (zisofs_write_to_temp(a, buff, ts) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } else { + if (wb_write_to_temp(a, buff, ts) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + iso9660->cur_file->cur_content->size += ts; + } + + /* Write padding. */ + if (wb_write_padding_to_temp(a, + iso9660->cur_file->cur_content->size) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + /* Compute the logical block number. */ + iso9660->cur_file->cur_content->blocks = (int) + ((iso9660->cur_file->cur_content->size + + LOGICAL_BLOCK_SIZE -1) >> LOGICAL_BLOCK_BITS); + + /* + * Make next extent. + */ + ws -= ts; + buff = (const void *)(((const unsigned char *)buff) + ts); + /* Make a content for next extent. */ + con = calloc(1, sizeof(*con)); + if (con == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate content data"); + return (ARCHIVE_FATAL); + } + con->offset_of_temp = wb_offset(a); + iso9660->cur_file->cur_content->next = con; + iso9660->cur_file->cur_content = con; +#ifdef HAVE_ZLIB_H + iso9660->zisofs.block_offset = 0; +#endif + } + + if (iso9660->zisofs.detect_magic) + zisofs_detect_magic(a, buff, ws); + + if (iso9660->zisofs.making) { + if (zisofs_write_to_temp(a, buff, ws) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } else { + if (wb_write_to_temp(a, buff, ws) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + iso9660->cur_file->cur_content->size += ws; + } + + return (s); +} + +static ssize_t +iso9660_write_data(struct archive_write *a, const void *buff, size_t s) +{ + struct iso9660 *iso9660 = a->format_data; + ssize_t r; + + if (iso9660->cur_file == NULL) + return (0); + if (archive_entry_filetype(iso9660->cur_file->entry) != AE_IFREG) + return (0); + if (s > iso9660->bytes_remaining) + s = (size_t)iso9660->bytes_remaining; + if (s == 0) + return (0); + + r = write_iso9660_data(a, buff, s); + if (r > 0) + iso9660->bytes_remaining -= r; + return (r); +} + +static int +iso9660_finish_entry(struct archive_write *a) +{ + struct iso9660 *iso9660 = a->format_data; + + if (iso9660->cur_file == NULL) + return (ARCHIVE_OK); + if (archive_entry_filetype(iso9660->cur_file->entry) != AE_IFREG) + return (ARCHIVE_OK); + if (iso9660->cur_file->content.size == 0) + return (ARCHIVE_OK); + + /* If there are unwritten data, write null data instead. */ + while (iso9660->bytes_remaining > 0) { + size_t s; + + s = (iso9660->bytes_remaining > a->null_length)? + a->null_length: (size_t)iso9660->bytes_remaining; + if (write_iso9660_data(a, a->nulls, s) < 0) + return (ARCHIVE_FATAL); + iso9660->bytes_remaining -= s; + } + + if (iso9660->zisofs.making && zisofs_finish_entry(a) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + /* Write padding. */ + if (wb_write_padding_to_temp(a, iso9660->cur_file->cur_content->size) + != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + /* Compute the logical block number. */ + iso9660->cur_file->cur_content->blocks = (int) + ((iso9660->cur_file->cur_content->size + + LOGICAL_BLOCK_SIZE -1) >> LOGICAL_BLOCK_BITS); + + /* Add the current file to data file list. */ + isofile_add_data_file(iso9660, iso9660->cur_file); + + return (ARCHIVE_OK); +} + +static int +iso9660_close(struct archive_write *a) +{ + struct iso9660 *iso9660; + int ret, blocks; + + iso9660 = a->format_data; + + /* + * Write remaining data out to the temporary file. + */ + if (wb_remaining(a) > 0) { + ret = wb_write_out(a); + if (ret < 0) + return (ret); + } + + /* + * Preparations... + */ +#ifdef DEBUG + if (iso9660->birth_time == 0) +#endif + time(&(iso9660->birth_time)); + + /* + * Prepare a bootable ISO image. + */ + if (iso9660->opt.boot) { + /* Find out the boot file entry. */ + ret = isoent_find_out_boot_file(a, iso9660->primary.rootent); + if (ret < 0) + return (ret); + /* Reconvert the boot file from zisofs'ed form to + * plain form. */ + ret = zisofs_rewind_boot_file(a); + if (ret < 0) + return (ret); + /* Write remaining data out to the temporary file. */ + if (wb_remaining(a) > 0) { + ret = wb_write_out(a); + if (ret < 0) + return (ret); + } + /* Create the boot catalog. */ + ret = isoent_create_boot_catalog(a, iso9660->primary.rootent); + if (ret < 0) + return (ret); + } + + /* + * Prepare joliet extensions. + */ + if (iso9660->opt.joliet) { + /* Make a new tree for joliet. */ + ret = isoent_clone_tree(a, &(iso9660->joliet.rootent), + iso9660->primary.rootent); + if (ret < 0) + return (ret); + /* Make sure we have UTF-16BE converters. + * if there is no file entry, converters are still + * uninitialized. */ + if (iso9660->sconv_to_utf16be == NULL) { + iso9660->sconv_to_utf16be = + archive_string_conversion_to_charset( + &(a->archive), "UTF-16BE", 1); + if (iso9660->sconv_to_utf16be == NULL) + /* Couldn't allocate memory */ + return (ARCHIVE_FATAL); + iso9660->sconv_from_utf16be = + archive_string_conversion_from_charset( + &(a->archive), "UTF-16BE", 1); + if (iso9660->sconv_from_utf16be == NULL) + /* Couldn't allocate memory */ + return (ARCHIVE_FATAL); + } + } + + /* + * Make Path Tables. + */ + ret = isoent_make_path_table(a); + if (ret < 0) + return (ret); + + /* + * Calculate a total volume size and setup all locations of + * contents of an iso9660 image. + */ + blocks = SYSTEM_AREA_BLOCK + + PRIMARY_VOLUME_DESCRIPTOR_BLOCK + + VOLUME_DESCRIPTOR_SET_TERMINATOR_BLOCK + + NON_ISO_FILE_SYSTEM_INFORMATION_BLOCK; + if (iso9660->opt.boot) + blocks += BOOT_RECORD_DESCRIPTOR_BLOCK; + if (iso9660->opt.joliet) + blocks += SUPPLEMENTARY_VOLUME_DESCRIPTOR_BLOCK; + if (iso9660->opt.iso_level == 4) + blocks += SUPPLEMENTARY_VOLUME_DESCRIPTOR_BLOCK; + + /* Setup the locations of Path Table. */ + iso9660->primary.location_type_L_path_table = blocks; + blocks += iso9660->primary.path_table_block; + iso9660->primary.location_type_M_path_table = blocks; + blocks += iso9660->primary.path_table_block; + if (iso9660->opt.joliet) { + iso9660->joliet.location_type_L_path_table = blocks; + blocks += iso9660->joliet.path_table_block; + iso9660->joliet.location_type_M_path_table = blocks; + blocks += iso9660->joliet.path_table_block; + } + + /* Setup the locations of directories. */ + isoent_setup_directory_location(iso9660, blocks, + &(iso9660->primary)); + blocks += iso9660->primary.total_dir_block; + if (iso9660->opt.joliet) { + isoent_setup_directory_location(iso9660, blocks, + &(iso9660->joliet)); + blocks += iso9660->joliet.total_dir_block; + } + + if (iso9660->opt.rr) { + iso9660->location_rrip_er = blocks; + blocks += RRIP_ER_BLOCK; + } + + /* Setup the locations of all file contents. */ + isoent_setup_file_location(iso9660, blocks); + blocks += iso9660->total_file_block; + if (iso9660->opt.boot && iso9660->opt.boot_info_table) { + ret = setup_boot_information(a); + if (ret < 0) + return (ret); + } + + /* Now we have a total volume size. */ + iso9660->volume_space_size = blocks; + if (iso9660->opt.pad) + iso9660->volume_space_size += PADDING_BLOCK; + iso9660->volume_sequence_number = 1; + + + /* + * Write an ISO 9660 image. + */ + + /* Switch to start using wbuff as file buffer. */ + iso9660->wbuff_remaining = wb_buffmax(); + iso9660->wbuff_type = WB_TO_STREAM; + iso9660->wbuff_offset = 0; + iso9660->wbuff_written = 0; + iso9660->wbuff_tail = 0; + + /* Write The System Area */ + ret = write_null(a, SYSTEM_AREA_BLOCK * LOGICAL_BLOCK_SIZE); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + /* Write Primary Volume Descriptor */ + ret = write_VD(a, &(iso9660->primary)); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + if (iso9660->opt.boot) { + /* Write Boot Record Volume Descriptor */ + ret = write_VD_boot_record(a); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + + if (iso9660->opt.iso_level == 4) { + /* Write Enhanced Volume Descriptor */ + iso9660->primary.vdd_type = VDD_ENHANCED; + ret = write_VD(a, &(iso9660->primary)); + iso9660->primary.vdd_type = VDD_PRIMARY; + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + + if (iso9660->opt.joliet) { + ret = write_VD(a, &(iso9660->joliet)); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + + /* Write Volume Descriptor Set Terminator */ + ret = write_VD_terminator(a); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + /* Write Non-ISO File System Information */ + ret = write_information_block(a); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + /* Write Type L Path Table */ + ret = write_path_table(a, 0, &(iso9660->primary)); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + /* Write Type M Path Table */ + ret = write_path_table(a, 1, &(iso9660->primary)); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + if (iso9660->opt.joliet) { + /* Write Type L Path Table */ + ret = write_path_table(a, 0, &(iso9660->joliet)); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + /* Write Type M Path Table */ + ret = write_path_table(a, 1, &(iso9660->joliet)); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + + /* Write Directory Descriptors */ + ret = write_directory_descriptors(a, &(iso9660->primary)); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + if (iso9660->opt.joliet) { + ret = write_directory_descriptors(a, &(iso9660->joliet)); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + + if (iso9660->opt.rr) { + /* Write Rockridge ER(Extensions Reference) */ + ret = write_rr_ER(a); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + + /* Write File Descriptors */ + ret = write_file_descriptors(a); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + /* Write Padding */ + if (iso9660->opt.pad) { + ret = write_null(a, PADDING_BLOCK * LOGICAL_BLOCK_SIZE); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + + if (iso9660->directories_too_deep != NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "%s: Directories too deep.", + archive_entry_pathname( + iso9660->directories_too_deep->file->entry)); + return (ARCHIVE_WARN); + } + + /* Write remaining data out. */ + ret = wb_write_out(a); + + return (ret); +} + +static int +iso9660_free(struct archive_write *a) +{ + struct iso9660 *iso9660; + int i, ret; + + iso9660 = a->format_data; + + /* Close the temporary file. */ + if (iso9660->temp_fd >= 0) + close(iso9660->temp_fd); + + /* Free some stuff for zisofs operations. */ + ret = zisofs_free(a); + + /* Remove directory entries in tree which includes file entries. */ + isoent_free_all(iso9660->primary.rootent); + for (i = 0; i < iso9660->primary.max_depth; i++) + free(iso9660->primary.pathtbl[i].sorted); + free(iso9660->primary.pathtbl); + + if (iso9660->opt.joliet) { + isoent_free_all(iso9660->joliet.rootent); + for (i = 0; i < iso9660->joliet.max_depth; i++) + free(iso9660->joliet.pathtbl[i].sorted); + free(iso9660->joliet.pathtbl); + } + + /* Remove isofile entries. */ + isofile_free_all_entries(iso9660); + isofile_free_hardlinks(iso9660); + + archive_string_free(&(iso9660->cur_dirstr)); + archive_string_free(&(iso9660->volume_identifier)); + archive_string_free(&(iso9660->publisher_identifier)); + archive_string_free(&(iso9660->data_preparer_identifier)); + archive_string_free(&(iso9660->application_identifier)); + archive_string_free(&(iso9660->copyright_file_identifier)); + archive_string_free(&(iso9660->abstract_file_identifier)); + archive_string_free(&(iso9660->bibliographic_file_identifier)); + archive_string_free(&(iso9660->el_torito.catalog_filename)); + archive_string_free(&(iso9660->el_torito.boot_filename)); + archive_string_free(&(iso9660->el_torito.id)); + archive_string_free(&(iso9660->utf16be)); + archive_string_free(&(iso9660->mbs)); + + free(iso9660); + a->format_data = NULL; + + return (ret); +} + +/* + * Get the System Identifier + */ +static void +get_system_identitier(char *system_id, size_t size) +{ +#if defined(HAVE_SYS_UTSNAME_H) + struct utsname u; + + uname(&u); + strncpy(system_id, u.sysname, size-1); + system_id[size-1] = '\0'; +#elif defined(_WIN32) && !defined(__CYGWIN__) + strncpy(system_id, "Windows", size-1); + system_id[size-1] = '\0'; +#else +#error no way to get the system identifier on your platform. +#endif +} + +static void +set_str(unsigned char *p, const char *s, size_t l, char f, const char *map) +{ + unsigned char c; + + if (s == NULL) + s = ""; + while ((c = *s++) != 0 && l > 0) { + if (c >= 0x80 || map[c] == 0) + { + /* illegal character */ + if (c >= 'a' && c <= 'z') { + /* convert c from a-z to A-Z */ + c -= 0x20; + } else + c = 0x5f; + } + *p++ = c; + l--; + } + /* If l isn't zero, fill p buffer by the character + * which indicated by f. */ + if (l > 0) + memset(p , f, l); +} + +static inline int +joliet_allowed_char(unsigned char high, unsigned char low) +{ + int utf16 = (high << 8) | low; + + if (utf16 <= 0x001F) + return (0); + + switch (utf16) { + case 0x002A: /* '*' */ + case 0x002F: /* '/' */ + case 0x003A: /* ':' */ + case 0x003B: /* ';' */ + case 0x003F: /* '?' */ + case 0x005C: /* '\' */ + return (0);/* Not allowed. */ + } + return (1); +} + +static int +set_str_utf16be(struct archive_write *a, unsigned char *p, const char *s, + size_t l, uint16_t uf, enum vdc vdc) +{ + size_t size, i; + int onepad; + + if (s == NULL) + s = ""; + if (l & 0x01) { + onepad = 1; + l &= ~1; + } else + onepad = 0; + if (vdc == VDC_UCS2) { + struct iso9660 *iso9660 = a->format_data; + if (archive_strncpy_l(&iso9660->utf16be, s, strlen(s), + iso9660->sconv_to_utf16be) != 0 && errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for UTF-16BE"); + return (ARCHIVE_FATAL); + } + size = iso9660->utf16be.length; + if (size > l) + size = l; + memcpy(p, iso9660->utf16be.s, size); + } else { + const uint16_t *u16 = (const uint16_t *)s; + + size = 0; + while (*u16++) + size += 2; + if (size > l) + size = l; + memcpy(p, s, size); + } + for (i = 0; i < size; i += 2, p += 2) { + if (!joliet_allowed_char(p[0], p[1])) + archive_be16enc(p, 0x005F);/* '_' */ + } + l -= size; + while (l > 0) { + archive_be16enc(p, uf); + p += 2; + l -= 2; + } + if (onepad) + *p = 0; + return (ARCHIVE_OK); +} + +static const char a_characters_map[0x80] = { +/* 0 1 2 3 4 5 6 7 8 9 A B C D E F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 00-0F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 10-1F */ + 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 20-2F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 30-3F */ + 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 40-4F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1,/* 50-5F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 60-6F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 70-7F */ +}; + +static const char a1_characters_map[0x80] = { +/* 0 1 2 3 4 5 6 7 8 9 A B C D E F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 00-0F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 10-1F */ + 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 20-2F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 30-3F */ + 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 40-4F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1,/* 50-5F */ + 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 60-6F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0,/* 70-7F */ +}; + +static const char d_characters_map[0x80] = { +/* 0 1 2 3 4 5 6 7 8 9 A B C D E F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 00-0F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 10-1F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 20-2F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0,/* 30-3F */ + 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 40-4F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1,/* 50-5F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 60-6F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 70-7F */ +}; + +static const char d1_characters_map[0x80] = { +/* 0 1 2 3 4 5 6 7 8 9 A B C D E F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 00-0F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 10-1F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 20-2F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0,/* 30-3F */ + 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 40-4F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1,/* 50-5F */ + 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 60-6F */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0,/* 70-7F */ +}; + +static int +set_str_a_characters_bp(struct archive_write *a, unsigned char *bp, + int from, int to, const char *s, enum vdc vdc) +{ + int r; + + switch (vdc) { + case VDC_STD: + set_str(bp+from, s, to - from + 1, 0x20, + a_characters_map); + r = ARCHIVE_OK; + break; + case VDC_LOWERCASE: + set_str(bp+from, s, to - from + 1, 0x20, + a1_characters_map); + r = ARCHIVE_OK; + break; + case VDC_UCS2: + case VDC_UCS2_DIRECT: + r = set_str_utf16be(a, bp+from, s, to - from + 1, + 0x0020, vdc); + break; + default: + r = ARCHIVE_FATAL; + } + return (r); +} + +static int +set_str_d_characters_bp(struct archive_write *a, unsigned char *bp, + int from, int to, const char *s, enum vdc vdc) +{ + int r; + + switch (vdc) { + case VDC_STD: + set_str(bp+from, s, to - from + 1, 0x20, + d_characters_map); + r = ARCHIVE_OK; + break; + case VDC_LOWERCASE: + set_str(bp+from, s, to - from + 1, 0x20, + d1_characters_map); + r = ARCHIVE_OK; + break; + case VDC_UCS2: + case VDC_UCS2_DIRECT: + r = set_str_utf16be(a, bp+from, s, to - from + 1, + 0x0020, vdc); + break; + default: + r = ARCHIVE_FATAL; + } + return (r); +} + +static void +set_VD_bp(unsigned char *bp, enum VD_type type, unsigned char ver) +{ + + /* Volume Descriptor Type */ + bp[1] = (unsigned char)type; + /* Standard Identifier */ + memcpy(bp + 2, "CD001", 5); + /* Volume Descriptor Version */ + bp[7] = ver; +} + +static inline void +set_unused_field_bp(unsigned char *bp, int from, int to) +{ + memset(bp + from, 0, to - from + 1); +} + +/* + * 8-bit unsigned numerical values. + * ISO9660 Standard 7.1.1 + */ +static inline void +set_num_711(unsigned char *p, unsigned char value) +{ + *p = value; +} + +/* + * 8-bit signed numerical values. + * ISO9660 Standard 7.1.2 + */ +static inline void +set_num_712(unsigned char *p, char value) +{ + *((char *)p) = value; +} + +/* + * Least significant byte first. + * ISO9660 Standard 7.2.1 + */ +static inline void +set_num_721(unsigned char *p, uint16_t value) +{ + archive_le16enc(p, value); +} + +/* + * Most significant byte first. + * ISO9660 Standard 7.2.2 + */ +static inline void +set_num_722(unsigned char *p, uint16_t value) +{ + archive_be16enc(p, value); +} + +/* + * Both-byte orders. + * ISO9660 Standard 7.2.3 + */ +static void +set_num_723(unsigned char *p, uint16_t value) +{ + archive_le16enc(p, value); + archive_be16enc(p+2, value); +} + +/* + * Least significant byte first. + * ISO9660 Standard 7.3.1 + */ +static inline void +set_num_731(unsigned char *p, uint32_t value) +{ + archive_le32enc(p, value); +} + +/* + * Most significant byte first. + * ISO9660 Standard 7.3.2 + */ +static inline void +set_num_732(unsigned char *p, uint32_t value) +{ + archive_be32enc(p, value); +} + +/* + * Both-byte orders. + * ISO9660 Standard 7.3.3 + */ +static inline void +set_num_733(unsigned char *p, uint32_t value) +{ + archive_le32enc(p, value); + archive_be32enc(p+4, value); +} + +static void +set_digit(unsigned char *p, size_t s, int value) +{ + + while (s--) { + p[s] = '0' + (value % 10); + value /= 10; + } +} + +#if defined(HAVE_STRUCT_TM_TM_GMTOFF) +#define get_gmoffset(tm) ((tm)->tm_gmtoff) +#elif defined(HAVE_STRUCT_TM___TM_GMTOFF) +#define get_gmoffset(tm) ((tm)->__tm_gmtoff) +#else +static long +get_gmoffset(struct tm *tm) +{ + long offset; + +#if defined(HAVE__GET_TIMEZONE) + _get_timezone(&offset); +#elif defined(__CYGWIN__) || defined(__MINGW32__) || defined(__BORLANDC__) + offset = _timezone; +#else + offset = timezone; +#endif + offset *= -1; + if (tm->tm_isdst) + offset += 3600; + return (offset); +} +#endif + +static void +get_tmfromtime(struct tm *tm, time_t *t) +{ +#if HAVE_LOCALTIME_R + tzset(); + localtime_r(t, tm); +#elif HAVE__LOCALTIME64_S + __time64_t tmp_t = (__time64_t) *t; //time_t may be shorter than 64 bits + _localtime64_s(tm, &tmp_t); +#else + memcpy(tm, localtime(t), sizeof(*tm)); +#endif +} + +/* + * Date and Time Format. + * ISO9660 Standard 8.4.26.1 + */ +static void +set_date_time(unsigned char *p, time_t t) +{ + struct tm tm; + + get_tmfromtime(&tm, &t); + set_digit(p, 4, tm.tm_year + 1900); + set_digit(p+4, 2, tm.tm_mon + 1); + set_digit(p+6, 2, tm.tm_mday); + set_digit(p+8, 2, tm.tm_hour); + set_digit(p+10, 2, tm.tm_min); + set_digit(p+12, 2, tm.tm_sec); + set_digit(p+14, 2, 0); + set_num_712(p+16, (char)(get_gmoffset(&tm)/(60*15))); +} + +static void +set_date_time_null(unsigned char *p) +{ + memset(p, (int)'0', 16); + p[16] = 0; +} + +static void +set_time_915(unsigned char *p, time_t t) +{ + struct tm tm; + + get_tmfromtime(&tm, &t); + set_num_711(p+0, tm.tm_year); + set_num_711(p+1, tm.tm_mon+1); + set_num_711(p+2, tm.tm_mday); + set_num_711(p+3, tm.tm_hour); + set_num_711(p+4, tm.tm_min); + set_num_711(p+5, tm.tm_sec); + set_num_712(p+6, (char)(get_gmoffset(&tm)/(60*15))); +} + + +/* + * Write SUSP "CE" System Use Entry. + */ +static int +set_SUSP_CE(unsigned char *p, int location, int offset, int size) +{ + unsigned char *bp = p -1; + /* Extend the System Use Area + * "CE" Format: + * len ver + * +----+----+----+----+-----------+-----------+ + * | 'C'| 'E'| 1C | 01 | LOCATION1 | LOCATION2 | + * +----+----+----+----+-----------+-----------+ + * 0 1 2 3 4 12 20 + * +-----------+ + * | LOCATION3 | + * +-----------+ + * 20 28 + * LOCATION1 : Location of Continuation of System Use Area. + * LOCATION2 : Offset to Start of Continuation. + * LOCATION3 : Length of the Continuation. + */ + + bp[1] = 'C'; + bp[2] = 'E'; + bp[3] = RR_CE_SIZE; /* length */ + bp[4] = 1; /* version */ + set_num_733(bp+5, location); + set_num_733(bp+13, offset); + set_num_733(bp+21, size); + return (RR_CE_SIZE); +} + +/* + * The functions, which names are beginning with extra_, are used to + * control extra records. + * The maximum size of a Directory Record is 254. When a filename is + * very long, all of RRIP data of a file won't stored to the Directory + * Record and so remaining RRIP data store to an extra record instead. + */ +static unsigned char * +extra_open_record(unsigned char *bp, int dr_len, struct isoent *isoent, + struct ctl_extr_rec *ctl) +{ + ctl->bp = bp; + if (bp != NULL) + bp += dr_len; + ctl->use_extr = 0; + ctl->isoent = isoent; + ctl->ce_ptr = NULL; + ctl->cur_len = ctl->dr_len = dr_len; + ctl->limit = DR_LIMIT; + + return (bp); +} + +static void +extra_close_record(struct ctl_extr_rec *ctl, int ce_size) +{ + int padding = 0; + + if (ce_size > 0) + extra_tell_used_size(ctl, ce_size); + /* Padding. */ + if (ctl->cur_len & 0x01) { + ctl->cur_len++; + if (ctl->bp != NULL) + ctl->bp[ctl->cur_len] = 0; + padding = 1; + } + if (ctl->use_extr) { + if (ctl->ce_ptr != NULL) + set_SUSP_CE(ctl->ce_ptr, ctl->extr_loc, + ctl->extr_off, ctl->cur_len - padding); + } else + ctl->dr_len = ctl->cur_len; +} + +#define extra_space(ctl) ((ctl)->limit - (ctl)->cur_len) + +static unsigned char * +extra_next_record(struct ctl_extr_rec *ctl, int length) +{ + int cur_len = ctl->cur_len;/* save cur_len */ + + /* Close the current extra record or Directory Record. */ + extra_close_record(ctl, RR_CE_SIZE); + + /* Get a next extra record. */ + ctl->use_extr = 1; + if (ctl->bp != NULL) { + /* Storing data into an extra record. */ + unsigned char *p; + + /* Save the pointer where a CE extension will be + * stored to. */ + ctl->ce_ptr = &ctl->bp[cur_len+1]; + p = extra_get_record(ctl->isoent, + &ctl->limit, &ctl->extr_off, &ctl->extr_loc); + ctl->bp = p - 1;/* the base of bp offset is 1. */ + } else + /* Calculating the size of an extra record. */ + (void)extra_get_record(ctl->isoent, + &ctl->limit, NULL, NULL); + ctl->cur_len = 0; + /* Check if an extra record is almost full. + * If so, get a next one. */ + if (extra_space(ctl) < length) + (void)extra_next_record(ctl, length); + + return (ctl->bp); +} + +static inline struct extr_rec * +extra_last_record(struct isoent *isoent) +{ + if (isoent->extr_rec_list.first == NULL) + return (NULL); + return ((struct extr_rec *)(void *) + ((char *)(isoent->extr_rec_list.last) + - offsetof(struct extr_rec, next))); +} + +static unsigned char * +extra_get_record(struct isoent *isoent, int *space, int *off, int *loc) +{ + struct extr_rec *rec; + + isoent = isoent->parent; + if (off != NULL) { + /* Storing data into an extra record. */ + rec = isoent->extr_rec_list.current; + if (DR_SAFETY > LOGICAL_BLOCK_SIZE - rec->offset) + rec = rec->next; + } else { + /* Calculating the size of an extra record. */ + rec = extra_last_record(isoent); + if (rec == NULL || + DR_SAFETY > LOGICAL_BLOCK_SIZE - rec->offset) { + rec = malloc(sizeof(*rec)); + if (rec == NULL) + return (NULL); + rec->location = 0; + rec->offset = 0; + /* Insert `rec` into the tail of isoent->extr_rec_list */ + rec->next = NULL; + /* + * Note: testing isoent->extr_rec_list.last == NULL + * here is really unneeded since it has been already + * initialized at isoent_new function but Clang Static + * Analyzer claims that it is dereference of null + * pointer. + */ + if (isoent->extr_rec_list.last == NULL) + isoent->extr_rec_list.last = + &(isoent->extr_rec_list.first); + *isoent->extr_rec_list.last = rec; + isoent->extr_rec_list.last = &(rec->next); + } + } + *space = LOGICAL_BLOCK_SIZE - rec->offset - DR_SAFETY; + if (*space & 0x01) + *space -= 1;/* Keep padding space. */ + if (off != NULL) + *off = rec->offset; + if (loc != NULL) + *loc = rec->location; + isoent->extr_rec_list.current = rec; + + return (&rec->buf[rec->offset]); +} + +static void +extra_tell_used_size(struct ctl_extr_rec *ctl, int size) +{ + struct isoent *isoent; + struct extr_rec *rec; + + if (ctl->use_extr) { + isoent = ctl->isoent->parent; + rec = isoent->extr_rec_list.current; + if (rec != NULL) + rec->offset += size; + } + ctl->cur_len += size; +} + +static int +extra_setup_location(struct isoent *isoent, int location) +{ + struct extr_rec *rec; + int cnt; + + cnt = 0; + rec = isoent->extr_rec_list.first; + isoent->extr_rec_list.current = rec; + while (rec) { + cnt++; + rec->location = location++; + rec->offset = 0; + rec = rec->next; + } + return (cnt); +} + +/* + * Create the RRIP entries. + */ +static int +set_directory_record_rr(unsigned char *bp, int dr_len, + struct isoent *isoent, struct iso9660 *iso9660, enum dir_rec_type t) +{ + /* Flags(BP 5) of the Rockridge "RR" System Use Field */ + unsigned char rr_flag; +#define RR_USE_PX 0x01 +#define RR_USE_PN 0x02 +#define RR_USE_SL 0x04 +#define RR_USE_NM 0x08 +#define RR_USE_CL 0x10 +#define RR_USE_PL 0x20 +#define RR_USE_RE 0x40 +#define RR_USE_TF 0x80 + int length; + struct ctl_extr_rec ctl; + struct isoent *rr_parent, *pxent; + struct isofile *file; + + bp = extra_open_record(bp, dr_len, isoent, &ctl); + + if (t == DIR_REC_PARENT) { + rr_parent = isoent->rr_parent; + pxent = isoent->parent; + if (rr_parent != NULL) + isoent = rr_parent; + else + isoent = isoent->parent; + } else { + rr_parent = NULL; + pxent = isoent; + } + file = isoent->file; + + if (t != DIR_REC_NORMAL) { + rr_flag = RR_USE_PX | RR_USE_TF; + if (rr_parent != NULL) + rr_flag |= RR_USE_PL; + } else { + rr_flag = RR_USE_PX | RR_USE_NM | RR_USE_TF; + if (archive_entry_filetype(file->entry) == AE_IFLNK) + rr_flag |= RR_USE_SL; + if (isoent->rr_parent != NULL) + rr_flag |= RR_USE_RE; + if (isoent->rr_child != NULL) + rr_flag |= RR_USE_CL; + if (archive_entry_filetype(file->entry) == AE_IFCHR || + archive_entry_filetype(file->entry) == AE_IFBLK) + rr_flag |= RR_USE_PN; +#ifdef COMPAT_MKISOFS + /* + * mkisofs 2.01.01a63 records "RE" extension to + * the entry of "rr_moved" directory. + * I don't understand this behavior. + */ + if (isoent->virtual && + isoent->parent == iso9660->primary.rootent && + strcmp(isoent->file->basename.s, "rr_moved") == 0) + rr_flag |= RR_USE_RE; +#endif + } + + /* Write "SP" System Use Entry. */ + if (t == DIR_REC_SELF && isoent == isoent->parent) { + length = 7; + if (bp != NULL) { + bp[1] = 'S'; + bp[2] = 'P'; + bp[3] = length; + bp[4] = 1; /* version */ + bp[5] = 0xBE; /* Check Byte */ + bp[6] = 0xEF; /* Check Byte */ + bp[7] = 0; + bp += length; + } + extra_tell_used_size(&ctl, length); + } + + /* Write "RR" System Use Entry. */ + length = 5; + if (extra_space(&ctl) < length) + bp = extra_next_record(&ctl, length); + if (bp != NULL) { + bp[1] = 'R'; + bp[2] = 'R'; + bp[3] = length; + bp[4] = 1; /* version */ + bp[5] = rr_flag; + bp += length; + } + extra_tell_used_size(&ctl, length); + + /* Write "NM" System Use Entry. */ + if (rr_flag & RR_USE_NM) { + /* + * "NM" Format: + * e.g. a basename is 'foo' + * len ver flg + * +----+----+----+----+----+----+----+----+ + * | 'N'| 'M'| 08 | 01 | 00 | 'f'| 'o'| 'o'| + * +----+----+----+----+----+----+----+----+ + * <----------------- len -----------------> + */ + size_t nmlen = file->basename.length; + const char *nm = file->basename.s; + size_t nmmax; + + if (extra_space(&ctl) < 6) + bp = extra_next_record(&ctl, 6); + if (bp != NULL) { + bp[1] = 'N'; + bp[2] = 'M'; + bp[4] = 1; /* version */ + } + nmmax = extra_space(&ctl); + if (nmmax > 0xff) + nmmax = 0xff; + while (nmlen + 5 > nmmax) { + length = (int)nmmax; + if (bp != NULL) { + bp[3] = length; + bp[5] = 0x01;/* Alternate Name continues + * in next "NM" field */ + memcpy(bp+6, nm, length - 5); + bp += length; + } + nmlen -= length - 5; + nm += length - 5; + extra_tell_used_size(&ctl, length); + if (extra_space(&ctl) < 6) { + bp = extra_next_record(&ctl, 6); + nmmax = extra_space(&ctl); + if (nmmax > 0xff) + nmmax = 0xff; + } + if (bp != NULL) { + bp[1] = 'N'; + bp[2] = 'M'; + bp[4] = 1; /* version */ + } + } + length = 5 + (int)nmlen; + if (bp != NULL) { + bp[3] = length; + bp[5] = 0; + memcpy(bp+6, nm, nmlen); + bp += length; + } + extra_tell_used_size(&ctl, length); + } + + /* Write "PX" System Use Entry. */ + if (rr_flag & RR_USE_PX) { + /* + * "PX" Format: + * len ver + * +----+----+----+----+-----------+-----------+ + * | 'P'| 'X'| 2C | 01 | FILE MODE | LINKS | + * +----+----+----+----+-----------+-----------+ + * 0 1 2 3 4 12 20 + * +-----------+-----------+------------------+ + * | USER ID | GROUP ID |FILE SERIAL NUMBER| + * +-----------+-----------+------------------+ + * 20 28 36 44 + */ + length = 44; + if (extra_space(&ctl) < length) + bp = extra_next_record(&ctl, length); + if (bp != NULL) { + mode_t mode; + int64_t uid; + int64_t gid; + + mode = archive_entry_mode(file->entry); + uid = archive_entry_uid(file->entry); + gid = archive_entry_gid(file->entry); + if (iso9660->opt.rr == OPT_RR_USEFUL) { + /* + * This action is similar to mkisofs -r option + * but our rockridge=useful option does not + * set a zero to uid and gid. + */ + /* set all read bit ON */ + mode |= 0444; +#if !defined(_WIN32) && !defined(__CYGWIN__) + if (mode & 0111) +#endif + /* set all exec bit ON */ + mode |= 0111; + /* clear all write bits. */ + mode &= ~0222; + /* clear setuid,setgid,sticky bits. */ + mode &= ~07000; + } + + bp[1] = 'P'; + bp[2] = 'X'; + bp[3] = length; + bp[4] = 1; /* version */ + /* file mode */ + set_num_733(bp+5, mode); + /* file links (stat.st_nlink) */ + set_num_733(bp+13, + archive_entry_nlink(file->entry)); + set_num_733(bp+21, (uint32_t)uid); + set_num_733(bp+29, (uint32_t)gid); + /* File Serial Number */ + if (pxent->dir) + set_num_733(bp+37, pxent->dir_location); + else if (file->hardlink_target != NULL) + set_num_733(bp+37, + file->hardlink_target->cur_content->location); + else + set_num_733(bp+37, + file->cur_content->location); + bp += length; + } + extra_tell_used_size(&ctl, length); + } + + /* Write "SL" System Use Entry. */ + if (rr_flag & RR_USE_SL) { + /* + * "SL" Format: + * e.g. a symbolic name is 'foo/bar' + * len ver flg + * +----+----+----+----+----+------------+ + * | 'S'| 'L'| 0F | 01 | 00 | components | + * +----+----+----+----+----+-----+------+ + * 0 1 2 3 4 5 ...|... 15 + * <----------------- len --------+------> + * components : | + * cflg clen | + * +----+----+----+----+----+ | + * | 00 | 03 | 'f'| 'o'| 'o'| <---+ + * +----+----+----+----+----+ | + * 5 6 7 8 9 10 | + * cflg clen | + * +----+----+----+----+----+ | + * | 00 | 03 | 'b'| 'a'| 'r'| <---+ + * +----+----+----+----+----+ + * 10 11 12 13 14 15 + * + * - cflg : flag of component + * - clen : length of component + */ + const char *sl; + char sl_last; + + if (extra_space(&ctl) < 7) + bp = extra_next_record(&ctl, 7); + sl = file->symlink.s; + sl_last = '\0'; + if (bp != NULL) { + bp[1] = 'S'; + bp[2] = 'L'; + bp[4] = 1; /* version */ + } + for (;;) { + unsigned char *nc, *cf, *cl, cldmy = 0; + int sllen, slmax; + + slmax = extra_space(&ctl); + if (slmax > 0xff) + slmax = 0xff; + if (bp != NULL) + nc = &bp[6]; + else + nc = NULL; + cf = cl = NULL; + sllen = 0; + while (*sl && sllen + 11 < slmax) { + if (sl_last == '\0' && sl[0] == '/') { + /* + * flg len + * +----+----+ + * | 08 | 00 | ROOT component. + * +----+----+ ("/") + * + * Root component has to appear + * at the first component only. + */ + if (nc != NULL) { + cf = nc++; + *cf = 0x08; /* ROOT */ + *nc++ = 0; + } + sllen += 2; + sl++; + sl_last = '/'; + cl = NULL; + continue; + } + if (((sl_last == '\0' || sl_last == '/') && + sl[0] == '.' && sl[1] == '.' && + (sl[2] == '/' || sl[2] == '\0')) || + (sl[0] == '/' && + sl[1] == '.' && sl[2] == '.' && + (sl[3] == '/' || sl[3] == '\0'))) { + /* + * flg len + * +----+----+ + * | 04 | 00 | PARENT component. + * +----+----+ ("..") + */ + if (nc != NULL) { + cf = nc++; + *cf = 0x04; /* PARENT */ + *nc++ = 0; + } + sllen += 2; + if (sl[0] == '/') + sl += 3;/* skip "/.." */ + else + sl += 2;/* skip ".." */ + sl_last = '.'; + cl = NULL; + continue; + } + if (((sl_last == '\0' || sl_last == '/') && + sl[0] == '.' && + (sl[1] == '/' || sl[1] == '\0')) || + (sl[0] == '/' && sl[1] == '.' && + (sl[2] == '/' || sl[2] == '\0'))) { + /* + * flg len + * +----+----+ + * | 02 | 00 | CURRENT component. + * +----+----+ (".") + */ + if (nc != NULL) { + cf = nc++; + *cf = 0x02; /* CURRENT */ + *nc++ = 0; + } + sllen += 2; + if (sl[0] == '/') + sl += 2;/* skip "/." */ + else + sl ++; /* skip "." */ + sl_last = '.'; + cl = NULL; + continue; + } + if (sl[0] == '/' || cl == NULL) { + if (nc != NULL) { + cf = nc++; + *cf = 0; + cl = nc++; + *cl = 0; + } else + cl = &cldmy; + sllen += 2; + if (sl[0] == '/') { + sl_last = *sl++; + continue; + } + } + sl_last = *sl++; + if (nc != NULL) { + *nc++ = sl_last; + (*cl) ++; + } + sllen++; + } + if (*sl) { + length = 5 + sllen; + if (bp != NULL) { + /* + * Mark flg as CONTINUE component. + */ + *cf |= 0x01; + /* + * len ver flg + * +----+----+----+----+----+- + * | 'S'| 'L'| XX | 01 | 01 | + * +----+----+----+----+----+- + * ^ + * continues in next "SL" + */ + bp[3] = length; + bp[5] = 0x01;/* This Symbolic Link + * continues in next + * "SL" field */ + bp += length; + } + extra_tell_used_size(&ctl, length); + if (extra_space(&ctl) < 11) + bp = extra_next_record(&ctl, 11); + if (bp != NULL) { + /* Next 'SL' */ + bp[1] = 'S'; + bp[2] = 'L'; + bp[4] = 1; /* version */ + } + } else { + length = 5 + sllen; + if (bp != NULL) { + bp[3] = length; + bp[5] = 0; + bp += length; + } + extra_tell_used_size(&ctl, length); + break; + } + } + } + + /* Write "TF" System Use Entry. */ + if (rr_flag & RR_USE_TF) { + /* + * "TF" Format: + * len ver + * +----+----+----+----+-----+-------------+ + * | 'T'| 'F'| XX | 01 |FLAGS| TIME STAMPS | + * +----+----+----+----+-----+-------------+ + * 0 1 2 3 4 5 XX + * TIME STAMPS : ISO 9660 Standard 9.1.5. + * If TF_LONG_FORM FLAGS is set, + * use ISO9660 Standard 8.4.26.1. + */ +#define TF_CREATION 0x01 /* Creation time recorded */ +#define TF_MODIFY 0x02 /* Modification time recorded */ +#define TF_ACCESS 0x04 /* Last Access time recorded */ +#define TF_ATTRIBUTES 0x08 /* Last Attribute Change time recorded */ +#define TF_BACKUP 0x10 /* Last Backup time recorded */ +#define TF_EXPIRATION 0x20 /* Expiration time recorded */ +#define TF_EFFECTIVE 0x40 /* Effective time recorded */ +#define TF_LONG_FORM 0x80 /* ISO 9660 17-byte time format used */ + unsigned char tf_flags; + + length = 5; + tf_flags = 0; +#ifndef COMPAT_MKISOFS + if (archive_entry_birthtime_is_set(file->entry) && + archive_entry_birthtime(file->entry) <= + archive_entry_mtime(file->entry)) { + length += 7; + tf_flags |= TF_CREATION; + } +#endif + if (archive_entry_mtime_is_set(file->entry)) { + length += 7; + tf_flags |= TF_MODIFY; + } + if (archive_entry_atime_is_set(file->entry)) { + length += 7; + tf_flags |= TF_ACCESS; + } + if (archive_entry_ctime_is_set(file->entry)) { + length += 7; + tf_flags |= TF_ATTRIBUTES; + } + if (extra_space(&ctl) < length) + bp = extra_next_record(&ctl, length); + if (bp != NULL) { + bp[1] = 'T'; + bp[2] = 'F'; + bp[3] = length; + bp[4] = 1; /* version */ + bp[5] = tf_flags; + bp += 5; + /* Creation time */ + if (tf_flags & TF_CREATION) { + set_time_915(bp+1, + archive_entry_birthtime(file->entry)); + bp += 7; + } + /* Modification time */ + if (tf_flags & TF_MODIFY) { + set_time_915(bp+1, + archive_entry_mtime(file->entry)); + bp += 7; + } + /* Last Access time */ + if (tf_flags & TF_ACCESS) { + set_time_915(bp+1, + archive_entry_atime(file->entry)); + bp += 7; + } + /* Last Attribute Change time */ + if (tf_flags & TF_ATTRIBUTES) { + set_time_915(bp+1, + archive_entry_ctime(file->entry)); + bp += 7; + } + } + extra_tell_used_size(&ctl, length); + } + + /* Write "RE" System Use Entry. */ + if (rr_flag & RR_USE_RE) { + /* + * "RE" Format: + * len ver + * +----+----+----+----+ + * | 'R'| 'E'| 04 | 01 | + * +----+----+----+----+ + * 0 1 2 3 4 + */ + length = 4; + if (extra_space(&ctl) < length) + bp = extra_next_record(&ctl, length); + if (bp != NULL) { + bp[1] = 'R'; + bp[2] = 'E'; + bp[3] = length; + bp[4] = 1; /* version */ + bp += length; + } + extra_tell_used_size(&ctl, length); + } + + /* Write "PL" System Use Entry. */ + if (rr_flag & RR_USE_PL) { + /* + * "PL" Format: + * len ver + * +----+----+----+----+------------+ + * | 'P'| 'L'| 0C | 01 | *LOCATION | + * +----+----+----+----+------------+ + * 0 1 2 3 4 12 + * *LOCATION: location of parent directory + */ + length = 12; + if (extra_space(&ctl) < length) + bp = extra_next_record(&ctl, length); + if (bp != NULL) { + bp[1] = 'P'; + bp[2] = 'L'; + bp[3] = length; + bp[4] = 1; /* version */ + set_num_733(bp + 5, + rr_parent->dir_location); + bp += length; + } + extra_tell_used_size(&ctl, length); + } + + /* Write "CL" System Use Entry. */ + if (rr_flag & RR_USE_CL) { + /* + * "CL" Format: + * len ver + * +----+----+----+----+------------+ + * | 'C'| 'L'| 0C | 01 | *LOCATION | + * +----+----+----+----+------------+ + * 0 1 2 3 4 12 + * *LOCATION: location of child directory + */ + length = 12; + if (extra_space(&ctl) < length) + bp = extra_next_record(&ctl, length); + if (bp != NULL) { + bp[1] = 'C'; + bp[2] = 'L'; + bp[3] = length; + bp[4] = 1; /* version */ + set_num_733(bp + 5, + isoent->rr_child->dir_location); + bp += length; + } + extra_tell_used_size(&ctl, length); + } + + /* Write "PN" System Use Entry. */ + if (rr_flag & RR_USE_PN) { + /* + * "PN" Format: + * len ver + * +----+----+----+----+------------+------------+ + * | 'P'| 'N'| 14 | 01 | dev_t high | dev_t low | + * +----+----+----+----+------------+------------+ + * 0 1 2 3 4 12 20 + */ + length = 20; + if (extra_space(&ctl) < length) + bp = extra_next_record(&ctl, length); + if (bp != NULL) { + uint64_t dev; + + bp[1] = 'P'; + bp[2] = 'N'; + bp[3] = length; + bp[4] = 1; /* version */ + dev = (uint64_t)archive_entry_rdev(file->entry); + set_num_733(bp + 5, (uint32_t)(dev >> 32)); + set_num_733(bp + 13, (uint32_t)(dev & 0xFFFFFFFF)); + bp += length; + } + extra_tell_used_size(&ctl, length); + } + + /* Write "ZF" System Use Entry. */ + if (file->zisofs.header_size) { + /* + * "ZF" Format: + * len ver + * +----+----+----+----+----+----+-------------+ + * | 'Z'| 'F'| 10 | 01 | 'p'| 'z'| Header Size | + * +----+----+----+----+----+----+-------------+ + * 0 1 2 3 4 5 6 7 + * +--------------------+-------------------+ + * | Log2 of block Size | Uncompressed Size | + * +--------------------+-------------------+ + * 7 8 16 + */ + length = 16; + if (extra_space(&ctl) < length) + bp = extra_next_record(&ctl, length); + if (bp != NULL) { + bp[1] = 'Z'; + bp[2] = 'F'; + bp[3] = length; + bp[4] = 1; /* version */ + bp[5] = 'p'; + bp[6] = 'z'; + bp[7] = file->zisofs.header_size; + bp[8] = file->zisofs.log2_bs; + set_num_733(bp + 9, file->zisofs.uncompressed_size); + bp += length; + } + extra_tell_used_size(&ctl, length); + } + + /* Write "CE" System Use Entry. */ + if (t == DIR_REC_SELF && isoent == isoent->parent) { + length = RR_CE_SIZE; + if (bp != NULL) + set_SUSP_CE(bp+1, iso9660->location_rrip_er, + 0, RRIP_ER_SIZE); + extra_tell_used_size(&ctl, length); + } + + extra_close_record(&ctl, 0); + + return (ctl.dr_len); +} + +/* + * Write data of a Directory Record or calculate writing bytes itself. + * If parameter `p' is NULL, calculates the size of writing data, which + * a Directory Record needs to write, then it saved and return + * the calculated size. + * Parameter `n' is a remaining size of buffer. when parameter `p' is + * not NULL, check whether that `n' is not less than the saved size. + * if that `n' is small, return zero. + * + * This format of the Directory Record is according to + * ISO9660 Standard 9.1 + */ +static int +set_directory_record(unsigned char *p, size_t n, struct isoent *isoent, + struct iso9660 *iso9660, enum dir_rec_type t, + enum vdd_type vdd_type) +{ + unsigned char *bp; + size_t dr_len; + size_t fi_len; + + if (p != NULL) { + /* + * Check whether a write buffer size is less than the + * saved size which is needed to write this Directory + * Record. + */ + switch (t) { + case DIR_REC_VD: + dr_len = isoent->dr_len.vd; break; + case DIR_REC_SELF: + dr_len = isoent->dr_len.self; break; + case DIR_REC_PARENT: + dr_len = isoent->dr_len.parent; break; + case DIR_REC_NORMAL: + default: + dr_len = isoent->dr_len.normal; break; + } + if (dr_len > n) + return (0);/* Needs more buffer size. */ + } + + if (t == DIR_REC_NORMAL && isoent->identifier != NULL) + fi_len = isoent->id_len; + else + fi_len = 1; + + if (p != NULL) { + struct isoent *xisoent; + struct isofile *file; + unsigned char flag; + + if (t == DIR_REC_PARENT) + xisoent = isoent->parent; + else + xisoent = isoent; + file = isoent->file; + if (file->hardlink_target != NULL) + file = file->hardlink_target; + /* Make a file flag. */ + if (xisoent->dir) + flag = FILE_FLAG_DIRECTORY; + else { + if (file->cur_content->next != NULL) + flag = FILE_FLAG_MULTI_EXTENT; + else + flag = 0; + } + + bp = p -1; + /* Extended Attribute Record Length */ + set_num_711(bp+2, 0); + /* Location of Extent */ + if (xisoent->dir) + set_num_733(bp+3, xisoent->dir_location); + else + set_num_733(bp+3, file->cur_content->location); + /* Data Length */ + if (xisoent->dir) + set_num_733(bp+11, + xisoent->dir_block * LOGICAL_BLOCK_SIZE); + else + set_num_733(bp+11, (uint32_t)file->cur_content->size); + /* Recording Date and Time */ + /* NOTE: + * If a file type is symbolic link, you are seeing this + * field value is different from a value mkisofs makes. + * libarchive uses lstat to get this one, but it + * seems mkisofs uses stat to get. + */ + set_time_915(bp+19, + archive_entry_mtime(xisoent->file->entry)); + /* File Flags */ + bp[26] = flag; + /* File Unit Size */ + set_num_711(bp+27, 0); + /* Interleave Gap Size */ + set_num_711(bp+28, 0); + /* Volume Sequence Number */ + set_num_723(bp+29, iso9660->volume_sequence_number); + /* Length of File Identifier */ + set_num_711(bp+33, (unsigned char)fi_len); + /* File Identifier */ + switch (t) { + case DIR_REC_VD: + case DIR_REC_SELF: + set_num_711(bp+34, 0); + break; + case DIR_REC_PARENT: + set_num_711(bp+34, 1); + break; + case DIR_REC_NORMAL: + if (isoent->identifier != NULL) + memcpy(bp+34, isoent->identifier, fi_len); + else + set_num_711(bp+34, 0); + break; + } + } else + bp = NULL; + dr_len = 33 + fi_len; + /* Padding Field */ + if (dr_len & 0x01) { + dr_len ++; + if (p != NULL) + bp[dr_len] = 0; + } + + /* Volume Descriptor does not record extension. */ + if (t == DIR_REC_VD) { + if (p != NULL) + /* Length of Directory Record */ + set_num_711(p, (unsigned char)dr_len); + else + isoent->dr_len.vd = (int)dr_len; + return ((int)dr_len); + } + + /* Rockridge */ + if (iso9660->opt.rr && vdd_type != VDD_JOLIET) + dr_len = set_directory_record_rr(bp, (int)dr_len, + isoent, iso9660, t); + + if (p != NULL) + /* Length of Directory Record */ + set_num_711(p, (unsigned char)dr_len); + else { + /* + * Save the size which is needed to write this + * Directory Record. + */ + switch (t) { + case DIR_REC_VD: + /* This case does not come, but compiler + * complains that DIR_REC_VD not handled + * in switch .... */ + break; + case DIR_REC_SELF: + isoent->dr_len.self = (int)dr_len; break; + case DIR_REC_PARENT: + isoent->dr_len.parent = (int)dr_len; break; + case DIR_REC_NORMAL: + isoent->dr_len.normal = (int)dr_len; break; + } + } + + return ((int)dr_len); +} + +/* + * Calculate the size of a directory record. + */ +static inline int +get_dir_rec_size(struct iso9660 *iso9660, struct isoent *isoent, + enum dir_rec_type t, enum vdd_type vdd_type) +{ + + return (set_directory_record(NULL, SIZE_MAX, + isoent, iso9660, t, vdd_type)); +} + +/* + * Manage to write ISO-image data with wbuff to reduce calling + * __archive_write_output() for performance. + */ + + +static inline unsigned char * +wb_buffptr(struct archive_write *a) +{ + struct iso9660 *iso9660 = (struct iso9660 *)a->format_data; + + return (&(iso9660->wbuff[sizeof(iso9660->wbuff) + - iso9660->wbuff_remaining])); +} + +static int +wb_write_out(struct archive_write *a) +{ + struct iso9660 *iso9660 = (struct iso9660 *)a->format_data; + size_t wsize, nw; + int r; + + wsize = sizeof(iso9660->wbuff) - iso9660->wbuff_remaining; + nw = wsize % LOGICAL_BLOCK_SIZE; + if (iso9660->wbuff_type == WB_TO_STREAM) + r = __archive_write_output(a, iso9660->wbuff, wsize - nw); + else + r = write_to_temp(a, iso9660->wbuff, wsize - nw); + /* Increase the offset. */ + iso9660->wbuff_offset += wsize - nw; + if (iso9660->wbuff_offset > iso9660->wbuff_written) + iso9660->wbuff_written = iso9660->wbuff_offset; + iso9660->wbuff_remaining = sizeof(iso9660->wbuff); + if (nw) { + iso9660->wbuff_remaining -= nw; + memmove(iso9660->wbuff, iso9660->wbuff + wsize - nw, nw); + } + return (r); +} + +static int +wb_consume(struct archive_write *a, size_t size) +{ + struct iso9660 *iso9660 = (struct iso9660 *)a->format_data; + + if (size > iso9660->wbuff_remaining || + iso9660->wbuff_remaining == 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Internal Programming error: iso9660:wb_consume()" + " size=%jd, wbuff_remaining=%jd", + (intmax_t)size, (intmax_t)iso9660->wbuff_remaining); + return (ARCHIVE_FATAL); + } + iso9660->wbuff_remaining -= size; + if (iso9660->wbuff_remaining < LOGICAL_BLOCK_SIZE) + return (wb_write_out(a)); + return (ARCHIVE_OK); +} + +#ifdef HAVE_ZLIB_H + +static int +wb_set_offset(struct archive_write *a, int64_t off) +{ + struct iso9660 *iso9660 = (struct iso9660 *)a->format_data; + int64_t used, ext_bytes; + + if (iso9660->wbuff_type != WB_TO_TEMP) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Internal Programming error: iso9660:wb_set_offset()"); + return (ARCHIVE_FATAL); + } + + used = sizeof(iso9660->wbuff) - iso9660->wbuff_remaining; + if (iso9660->wbuff_offset + used > iso9660->wbuff_tail) + iso9660->wbuff_tail = iso9660->wbuff_offset + used; + if (iso9660->wbuff_offset < iso9660->wbuff_written) { + if (used > 0 && + write_to_temp(a, iso9660->wbuff, (size_t)used) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + iso9660->wbuff_offset = iso9660->wbuff_written; + lseek(iso9660->temp_fd, iso9660->wbuff_offset, SEEK_SET); + iso9660->wbuff_remaining = sizeof(iso9660->wbuff); + used = 0; + } + if (off < iso9660->wbuff_offset) { + /* + * Write out waiting data. + */ + if (used > 0) { + if (wb_write_out(a) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + lseek(iso9660->temp_fd, off, SEEK_SET); + iso9660->wbuff_offset = off; + iso9660->wbuff_remaining = sizeof(iso9660->wbuff); + } else if (off <= iso9660->wbuff_tail) { + iso9660->wbuff_remaining = (size_t) + (sizeof(iso9660->wbuff) - (off - iso9660->wbuff_offset)); + } else { + ext_bytes = off - iso9660->wbuff_tail; + iso9660->wbuff_remaining = (size_t)(sizeof(iso9660->wbuff) + - (iso9660->wbuff_tail - iso9660->wbuff_offset)); + while (ext_bytes >= (int64_t)iso9660->wbuff_remaining) { + if (write_null(a, (size_t)iso9660->wbuff_remaining) + != ARCHIVE_OK) + return (ARCHIVE_FATAL); + ext_bytes -= iso9660->wbuff_remaining; + } + if (ext_bytes > 0) { + if (write_null(a, (size_t)ext_bytes) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + } + return (ARCHIVE_OK); +} + +#endif /* HAVE_ZLIB_H */ + +static int +write_null(struct archive_write *a, size_t size) +{ + size_t remaining; + unsigned char *p, *old; + int r; + + remaining = wb_remaining(a); + p = wb_buffptr(a); + if (size <= remaining) { + memset(p, 0, size); + return (wb_consume(a, size)); + } + memset(p, 0, remaining); + r = wb_consume(a, remaining); + if (r != ARCHIVE_OK) + return (r); + size -= remaining; + old = p; + p = wb_buffptr(a); + memset(p, 0, old - p); + remaining = wb_remaining(a); + while (size) { + size_t wsize = size; + + if (wsize > remaining) + wsize = remaining; + r = wb_consume(a, wsize); + if (r != ARCHIVE_OK) + return (r); + size -= wsize; + } + return (ARCHIVE_OK); +} + +/* + * Write Volume Descriptor Set Terminator + */ +static int +write_VD_terminator(struct archive_write *a) +{ + unsigned char *bp; + + bp = wb_buffptr(a) -1; + set_VD_bp(bp, VDT_TERMINATOR, 1); + set_unused_field_bp(bp, 8, LOGICAL_BLOCK_SIZE); + + return (wb_consume(a, LOGICAL_BLOCK_SIZE)); +} + +static int +set_file_identifier(unsigned char *bp, int from, int to, enum vdc vdc, + struct archive_write *a, struct vdd *vdd, struct archive_string *id, + const char *label, int leading_under, enum char_type char_type) +{ + char identifier[256]; + struct isoent *isoent; + const char *ids; + size_t len; + int r; + + if (id->length > 0 && leading_under && id->s[0] != '_') { + if (char_type == A_CHAR) + r = set_str_a_characters_bp(a, bp, from, to, id->s, vdc); + else + r = set_str_d_characters_bp(a, bp, from, to, id->s, vdc); + } else if (id->length > 0) { + ids = id->s; + if (leading_under) + ids++; + isoent = isoent_find_entry(vdd->rootent, ids); + if (isoent == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Not Found %s `%s'.", + label, ids); + return (ARCHIVE_FATAL); + } + len = isoent->ext_off + isoent->ext_len; + if (vdd->vdd_type == VDD_JOLIET) { + if (len > sizeof(identifier)-2) + len = sizeof(identifier)-2; + } else { + if (len > sizeof(identifier)-1) + len = sizeof(identifier)-1; + } + memcpy(identifier, isoent->identifier, len); + identifier[len] = '\0'; + if (vdd->vdd_type == VDD_JOLIET) { + identifier[len+1] = 0; + vdc = VDC_UCS2_DIRECT; + } + if (char_type == A_CHAR) + r = set_str_a_characters_bp(a, bp, from, to, + identifier, vdc); + else + r = set_str_d_characters_bp(a, bp, from, to, + identifier, vdc); + } else { + if (char_type == A_CHAR) + r = set_str_a_characters_bp(a, bp, from, to, NULL, vdc); + else + r = set_str_d_characters_bp(a, bp, from, to, NULL, vdc); + } + return (r); +} + +/* + * Write Primary/Supplementary Volume Descriptor + */ +static int +write_VD(struct archive_write *a, struct vdd *vdd) +{ + struct iso9660 *iso9660; + unsigned char *bp; + uint16_t volume_set_size = 1; + char identifier[256]; + enum VD_type vdt; + enum vdc vdc; + unsigned char vd_ver, fst_ver; + int r; + + iso9660 = a->format_data; + switch (vdd->vdd_type) { + case VDD_JOLIET: + vdt = VDT_SUPPLEMENTARY; + vd_ver = fst_ver = 1; + vdc = VDC_UCS2; + break; + case VDD_ENHANCED: + vdt = VDT_SUPPLEMENTARY; + vd_ver = fst_ver = 2; + vdc = VDC_LOWERCASE; + break; + case VDD_PRIMARY: + default: + vdt = VDT_PRIMARY; + vd_ver = fst_ver = 1; +#ifdef COMPAT_MKISOFS + vdc = VDC_LOWERCASE; +#else + vdc = VDC_STD; +#endif + break; + } + + bp = wb_buffptr(a) -1; + /* Volume Descriptor Type */ + set_VD_bp(bp, vdt, vd_ver); + /* Unused Field */ + set_unused_field_bp(bp, 8, 8); + /* System Identifier */ + get_system_identitier(identifier, sizeof(identifier)); + r = set_str_a_characters_bp(a, bp, 9, 40, identifier, vdc); + if (r != ARCHIVE_OK) + return (r); + /* Volume Identifier */ + r = set_str_d_characters_bp(a, bp, 41, 72, + iso9660->volume_identifier.s, vdc); + if (r != ARCHIVE_OK) + return (r); + /* Unused Field */ + set_unused_field_bp(bp, 73, 80); + /* Volume Space Size */ + set_num_733(bp+81, iso9660->volume_space_size); + if (vdd->vdd_type == VDD_JOLIET) { + /* Escape Sequences */ + bp[89] = 0x25;/* UCS-2 Level 3 */ + bp[90] = 0x2F; + bp[91] = 0x45; + memset(bp + 92, 0, 120 - 92 + 1); + } else { + /* Unused Field */ + set_unused_field_bp(bp, 89, 120); + } + /* Volume Set Size */ + set_num_723(bp+121, volume_set_size); + /* Volume Sequence Number */ + set_num_723(bp+125, iso9660->volume_sequence_number); + /* Logical Block Size */ + set_num_723(bp+129, LOGICAL_BLOCK_SIZE); + /* Path Table Size */ + set_num_733(bp+133, vdd->path_table_size); + /* Location of Occurrence of Type L Path Table */ + set_num_731(bp+141, vdd->location_type_L_path_table); + /* Location of Optional Occurrence of Type L Path Table */ + set_num_731(bp+145, 0); + /* Location of Occurrence of Type M Path Table */ + set_num_732(bp+149, vdd->location_type_M_path_table); + /* Location of Optional Occurrence of Type M Path Table */ + set_num_732(bp+153, 0); + /* Directory Record for Root Directory(BP 157 to 190) */ + set_directory_record(bp+157, 190-157+1, vdd->rootent, + iso9660, DIR_REC_VD, vdd->vdd_type); + /* Volume Set Identifier */ + r = set_str_d_characters_bp(a, bp, 191, 318, "", vdc); + if (r != ARCHIVE_OK) + return (r); + /* Publisher Identifier */ + r = set_file_identifier(bp, 319, 446, vdc, a, vdd, + &(iso9660->publisher_identifier), + "Publisher File", 1, A_CHAR); + if (r != ARCHIVE_OK) + return (r); + /* Data Preparer Identifier */ + r = set_file_identifier(bp, 447, 574, vdc, a, vdd, + &(iso9660->data_preparer_identifier), + "Data Preparer File", 1, A_CHAR); + if (r != ARCHIVE_OK) + return (r); + /* Application Identifier */ + r = set_file_identifier(bp, 575, 702, vdc, a, vdd, + &(iso9660->application_identifier), + "Application File", 1, A_CHAR); + if (r != ARCHIVE_OK) + return (r); + /* Copyright File Identifier */ + r = set_file_identifier(bp, 703, 739, vdc, a, vdd, + &(iso9660->copyright_file_identifier), + "Copyright File", 0, D_CHAR); + if (r != ARCHIVE_OK) + return (r); + /* Abstract File Identifier */ + r = set_file_identifier(bp, 740, 776, vdc, a, vdd, + &(iso9660->abstract_file_identifier), + "Abstract File", 0, D_CHAR); + if (r != ARCHIVE_OK) + return (r); + /* Bibliographic File Identifier */ + r = set_file_identifier(bp, 777, 813, vdc, a, vdd, + &(iso9660->bibliographic_file_identifier), + "Bibliongraphic File", 0, D_CHAR); + if (r != ARCHIVE_OK) + return (r); + /* Volume Creation Date and Time */ + set_date_time(bp+814, iso9660->birth_time); + /* Volume Modification Date and Time */ + set_date_time(bp+831, iso9660->birth_time); + /* Volume Expiration Date and Time(obsolete) */ + set_date_time_null(bp+848); + /* Volume Effective Date and Time */ + set_date_time(bp+865, iso9660->birth_time); + /* File Structure Version */ + bp[882] = fst_ver; + /* Reserved */ + bp[883] = 0; + /* Application Use */ + memset(bp + 884, 0x20, 1395 - 884 + 1); + /* Reserved */ + set_unused_field_bp(bp, 1396, LOGICAL_BLOCK_SIZE); + + return (wb_consume(a, LOGICAL_BLOCK_SIZE)); +} + +/* + * Write Boot Record Volume Descriptor + */ +static int +write_VD_boot_record(struct archive_write *a) +{ + struct iso9660 *iso9660; + unsigned char *bp; + + iso9660 = a->format_data; + bp = wb_buffptr(a) -1; + /* Volume Descriptor Type */ + set_VD_bp(bp, VDT_BOOT_RECORD, 1); + /* Boot System Identifier */ + memcpy(bp+8, "EL TORITO SPECIFICATION", 23); + set_unused_field_bp(bp, 8+23, 39); + /* Unused */ + set_unused_field_bp(bp, 40, 71); + /* Absolute pointer to first sector of Boot Catalog */ + set_num_731(bp+72, + iso9660->el_torito.catalog->file->content.location); + /* Unused */ + set_unused_field_bp(bp, 76, LOGICAL_BLOCK_SIZE); + + return (wb_consume(a, LOGICAL_BLOCK_SIZE)); +} + +enum keytype { + KEY_FLG, + KEY_STR, + KEY_INT, + KEY_HEX +}; +static void +set_option_info(struct archive_string *info, int *opt, const char *key, + enum keytype type, ...) +{ + va_list ap; + char prefix; + const char *s; + int d; + + prefix = (*opt==0)? ' ':','; + va_start(ap, type); + switch (type) { + case KEY_FLG: + d = va_arg(ap, int); + archive_string_sprintf(info, "%c%s%s", + prefix, (d == 0)?"!":"", key); + break; + case KEY_STR: + s = va_arg(ap, const char *); + archive_string_sprintf(info, "%c%s=%s", + prefix, key, s); + break; + case KEY_INT: + d = va_arg(ap, int); + archive_string_sprintf(info, "%c%s=%d", + prefix, key, d); + break; + case KEY_HEX: + d = va_arg(ap, int); + archive_string_sprintf(info, "%c%s=%x", + prefix, key, d); + break; + } + va_end(ap); + + *opt = 1; +} + +/* + * Make Non-ISO File System Information + */ +static int +write_information_block(struct archive_write *a) +{ + struct iso9660 *iso9660; + char buf[128]; + const char *v; + int opt, r; + struct archive_string info; + size_t info_size = LOGICAL_BLOCK_SIZE * + NON_ISO_FILE_SYSTEM_INFORMATION_BLOCK; + + iso9660 = (struct iso9660 *)a->format_data; + if (info_size > wb_remaining(a)) { + r = wb_write_out(a); + if (r != ARCHIVE_OK) + return (r); + } + archive_string_init(&info); + if (archive_string_ensure(&info, info_size) == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + memset(info.s, 0, info_size); + opt = 0; +#if defined(HAVE__CTIME64_S) + { + __time64_t iso9660_birth_time_tmp = (__time64_t) iso9660->birth_time; //time_t may be shorter than 64 bits + _ctime64_s(buf, sizeof(buf), &(iso9660_birth_time_tmp)); + } +#elif defined(HAVE_CTIME_R) + ctime_r(&(iso9660->birth_time), buf); +#else + strncpy(buf, ctime(&(iso9660->birth_time)), sizeof(buf)-1); + buf[sizeof(buf)-1] = '\0'; +#endif + archive_string_sprintf(&info, + "INFO %s%s", buf, archive_version_string()); + if (iso9660->opt.abstract_file != OPT_ABSTRACT_FILE_DEFAULT) + set_option_info(&info, &opt, "abstract-file", + KEY_STR, iso9660->abstract_file_identifier.s); + if (iso9660->opt.application_id != OPT_APPLICATION_ID_DEFAULT) + set_option_info(&info, &opt, "application-id", + KEY_STR, iso9660->application_identifier.s); + if (iso9660->opt.allow_vernum != OPT_ALLOW_VERNUM_DEFAULT) + set_option_info(&info, &opt, "allow-vernum", + KEY_FLG, iso9660->opt.allow_vernum); + if (iso9660->opt.biblio_file != OPT_BIBLIO_FILE_DEFAULT) + set_option_info(&info, &opt, "biblio-file", + KEY_STR, iso9660->bibliographic_file_identifier.s); + if (iso9660->opt.boot != OPT_BOOT_DEFAULT) + set_option_info(&info, &opt, "boot", + KEY_STR, iso9660->el_torito.boot_filename.s); + if (iso9660->opt.boot_catalog != OPT_BOOT_CATALOG_DEFAULT) + set_option_info(&info, &opt, "boot-catalog", + KEY_STR, iso9660->el_torito.catalog_filename.s); + if (iso9660->opt.boot_info_table != OPT_BOOT_INFO_TABLE_DEFAULT) + set_option_info(&info, &opt, "boot-info-table", + KEY_FLG, iso9660->opt.boot_info_table); + if (iso9660->opt.boot_load_seg != OPT_BOOT_LOAD_SEG_DEFAULT) + set_option_info(&info, &opt, "boot-load-seg", + KEY_HEX, iso9660->el_torito.boot_load_seg); + if (iso9660->opt.boot_load_size != OPT_BOOT_LOAD_SIZE_DEFAULT) + set_option_info(&info, &opt, "boot-load-size", + KEY_INT, iso9660->el_torito.boot_load_size); + if (iso9660->opt.boot_type != OPT_BOOT_TYPE_DEFAULT) { + v = "no-emulation"; + if (iso9660->opt.boot_type == OPT_BOOT_TYPE_FD) + v = "fd"; + if (iso9660->opt.boot_type == OPT_BOOT_TYPE_HARD_DISK) + v = "hard-disk"; + set_option_info(&info, &opt, "boot-type", + KEY_STR, v); + } +#ifdef HAVE_ZLIB_H + if (iso9660->opt.compression_level != OPT_COMPRESSION_LEVEL_DEFAULT) + set_option_info(&info, &opt, "compression-level", + KEY_INT, iso9660->zisofs.compression_level); +#endif + if (iso9660->opt.copyright_file != OPT_COPYRIGHT_FILE_DEFAULT) + set_option_info(&info, &opt, "copyright-file", + KEY_STR, iso9660->copyright_file_identifier.s); + if (iso9660->opt.iso_level != OPT_ISO_LEVEL_DEFAULT) + set_option_info(&info, &opt, "iso-level", + KEY_INT, iso9660->opt.iso_level); + if (iso9660->opt.joliet != OPT_JOLIET_DEFAULT) { + if (iso9660->opt.joliet == OPT_JOLIET_LONGNAME) + set_option_info(&info, &opt, "joliet", + KEY_STR, "long"); + else + set_option_info(&info, &opt, "joliet", + KEY_FLG, iso9660->opt.joliet); + } + if (iso9660->opt.limit_depth != OPT_LIMIT_DEPTH_DEFAULT) + set_option_info(&info, &opt, "limit-depth", + KEY_FLG, iso9660->opt.limit_depth); + if (iso9660->opt.limit_dirs != OPT_LIMIT_DIRS_DEFAULT) + set_option_info(&info, &opt, "limit-dirs", + KEY_FLG, iso9660->opt.limit_dirs); + if (iso9660->opt.pad != OPT_PAD_DEFAULT) + set_option_info(&info, &opt, "pad", + KEY_FLG, iso9660->opt.pad); + if (iso9660->opt.publisher != OPT_PUBLISHER_DEFAULT) + set_option_info(&info, &opt, "publisher", + KEY_STR, iso9660->publisher_identifier.s); + if (iso9660->opt.rr != OPT_RR_DEFAULT) { + if (iso9660->opt.rr == OPT_RR_DISABLED) + set_option_info(&info, &opt, "rockridge", + KEY_FLG, iso9660->opt.rr); + else if (iso9660->opt.rr == OPT_RR_STRICT) + set_option_info(&info, &opt, "rockridge", + KEY_STR, "strict"); + else if (iso9660->opt.rr == OPT_RR_USEFUL) + set_option_info(&info, &opt, "rockridge", + KEY_STR, "useful"); + } + if (iso9660->opt.volume_id != OPT_VOLUME_ID_DEFAULT) + set_option_info(&info, &opt, "volume-id", + KEY_STR, iso9660->volume_identifier.s); + if (iso9660->opt.zisofs != OPT_ZISOFS_DEFAULT) + set_option_info(&info, &opt, "zisofs", + KEY_FLG, iso9660->opt.zisofs); + + memcpy(wb_buffptr(a), info.s, info_size); + archive_string_free(&info); + return (wb_consume(a, info_size)); +} + +static int +write_rr_ER(struct archive_write *a) +{ + unsigned char *p; + + p = wb_buffptr(a); + + memset(p, 0, LOGICAL_BLOCK_SIZE); + p[0] = 'E'; + p[1] = 'R'; + p[3] = 0x01; + p[2] = RRIP_ER_SIZE; + p[4] = RRIP_ER_ID_SIZE; + p[5] = RRIP_ER_DSC_SIZE; + p[6] = RRIP_ER_SRC_SIZE; + p[7] = 0x01; + memcpy(&p[8], rrip_identifier, p[4]); + memcpy(&p[8+p[4]], rrip_descriptor, p[5]); + memcpy(&p[8+p[4]+p[5]], rrip_source, p[6]); + + return (wb_consume(a, LOGICAL_BLOCK_SIZE)); +} + +static void +calculate_path_table_size(struct vdd *vdd) +{ + int depth, size; + struct path_table *pt; + + pt = vdd->pathtbl; + size = 0; + for (depth = 0; depth < vdd->max_depth; depth++) { + struct isoent **ptbl; + int i, cnt; + + if ((cnt = pt[depth].cnt) == 0) + break; + + ptbl = pt[depth].sorted; + for (i = 0; i < cnt; i++) { + int len; + + if (ptbl[i]->identifier == NULL) + len = 1; /* root directory */ + else + len = ptbl[i]->id_len; + if (len & 0x01) + len++; /* Padding Field */ + size += 8 + len; + } + } + vdd->path_table_size = size; + vdd->path_table_block = + ((size + PATH_TABLE_BLOCK_SIZE -1) / + PATH_TABLE_BLOCK_SIZE) * + (PATH_TABLE_BLOCK_SIZE / LOGICAL_BLOCK_SIZE); +} + +static int +_write_path_table(struct archive_write *a, int type_m, int depth, + struct vdd *vdd) +{ + unsigned char *bp, *wb; + struct isoent **ptbl; + size_t wbremaining; + int i, r, wsize; + + if (vdd->pathtbl[depth].cnt == 0) + return (0); + + wsize = 0; + wb = wb_buffptr(a); + wbremaining = wb_remaining(a); + bp = wb - 1; + ptbl = vdd->pathtbl[depth].sorted; + for (i = 0; i < vdd->pathtbl[depth].cnt; i++) { + struct isoent *np; + size_t len; + + np = ptbl[i]; + if (np->identifier == NULL) + len = 1; /* root directory */ + else + len = np->id_len; + if (wbremaining - ((bp+1) - wb) < (len + 1 + 8)) { + r = wb_consume(a, (bp+1) - wb); + if (r < 0) + return (r); + wb = wb_buffptr(a); + wbremaining = wb_remaining(a); + bp = wb -1; + } + /* Length of Directory Identifier */ + set_num_711(bp+1, (unsigned char)len); + /* Extended Attribute Record Length */ + set_num_711(bp+2, 0); + /* Location of Extent */ + if (type_m) + set_num_732(bp+3, np->dir_location); + else + set_num_731(bp+3, np->dir_location); + /* Parent Directory Number */ + if (type_m) + set_num_722(bp+7, np->parent->dir_number); + else + set_num_721(bp+7, np->parent->dir_number); + /* Directory Identifier */ + if (np->identifier == NULL) + bp[9] = 0; + else + memcpy(&bp[9], np->identifier, len); + if (len & 0x01) { + /* Padding Field */ + bp[9+len] = 0; + len++; + } + wsize += 8 + (int)len; + bp += 8 + len; + } + if ((bp + 1) > wb) { + r = wb_consume(a, (bp+1)-wb); + if (r < 0) + return (r); + } + return (wsize); +} + +static int +write_path_table(struct archive_write *a, int type_m, struct vdd *vdd) +{ + int depth, r; + size_t path_table_size; + + r = ARCHIVE_OK; + path_table_size = 0; + for (depth = 0; depth < vdd->max_depth; depth++) { + r = _write_path_table(a, type_m, depth, vdd); + if (r < 0) + return (r); + path_table_size += r; + } + + /* Write padding data. */ + path_table_size = path_table_size % PATH_TABLE_BLOCK_SIZE; + if (path_table_size > 0) + r = write_null(a, PATH_TABLE_BLOCK_SIZE - path_table_size); + return (r); +} + +static int +calculate_directory_descriptors(struct iso9660 *iso9660, struct vdd *vdd, + struct isoent *isoent, int depth) +{ + struct isoent **enttbl; + int bs, block, i; + + block = 1; + bs = get_dir_rec_size(iso9660, isoent, DIR_REC_SELF, vdd->vdd_type); + bs += get_dir_rec_size(iso9660, isoent, DIR_REC_PARENT, vdd->vdd_type); + + if (isoent->children.cnt <= 0 || (vdd->vdd_type != VDD_JOLIET && + !iso9660->opt.rr && depth + 1 >= vdd->max_depth)) + return (block); + + enttbl = isoent->children_sorted; + for (i = 0; i < isoent->children.cnt; i++) { + struct isoent *np = enttbl[i]; + struct isofile *file; + + file = np->file; + if (file->hardlink_target != NULL) + file = file->hardlink_target; + file->cur_content = &(file->content); + do { + int dr_l; + + dr_l = get_dir_rec_size(iso9660, np, DIR_REC_NORMAL, + vdd->vdd_type); + if ((bs + dr_l) > LOGICAL_BLOCK_SIZE) { + block ++; + bs = dr_l; + } else + bs += dr_l; + file->cur_content = file->cur_content->next; + } while (file->cur_content != NULL); + } + return (block); +} + +static int +_write_directory_descriptors(struct archive_write *a, struct vdd *vdd, + struct isoent *isoent, int depth) +{ + struct iso9660 *iso9660 = a->format_data; + struct isoent **enttbl; + unsigned char *p, *wb; + int i, r; + int dr_l; + + p = wb = wb_buffptr(a); +#define WD_REMAINING (LOGICAL_BLOCK_SIZE - (p - wb)) + p += set_directory_record(p, WD_REMAINING, isoent, + iso9660, DIR_REC_SELF, vdd->vdd_type); + p += set_directory_record(p, WD_REMAINING, isoent, + iso9660, DIR_REC_PARENT, vdd->vdd_type); + + if (isoent->children.cnt <= 0 || (vdd->vdd_type != VDD_JOLIET && + !iso9660->opt.rr && depth + 1 >= vdd->max_depth)) { + memset(p, 0, WD_REMAINING); + return (wb_consume(a, LOGICAL_BLOCK_SIZE)); + } + + enttbl = isoent->children_sorted; + for (i = 0; i < isoent->children.cnt; i++) { + struct isoent *np = enttbl[i]; + struct isofile *file = np->file; + + if (file->hardlink_target != NULL) + file = file->hardlink_target; + file->cur_content = &(file->content); + do { + dr_l = set_directory_record(p, WD_REMAINING, + np, iso9660, DIR_REC_NORMAL, + vdd->vdd_type); + if (dr_l == 0) { + memset(p, 0, WD_REMAINING); + r = wb_consume(a, LOGICAL_BLOCK_SIZE); + if (r < 0) + return (r); + p = wb = wb_buffptr(a); + dr_l = set_directory_record(p, + WD_REMAINING, np, iso9660, + DIR_REC_NORMAL, vdd->vdd_type); + } + p += dr_l; + file->cur_content = file->cur_content->next; + } while (file->cur_content != NULL); + } + memset(p, 0, WD_REMAINING); + return (wb_consume(a, LOGICAL_BLOCK_SIZE)); +} + +static int +write_directory_descriptors(struct archive_write *a, struct vdd *vdd) +{ + struct isoent *np; + int depth, r; + + depth = 0; + np = vdd->rootent; + do { + struct extr_rec *extr; + + r = _write_directory_descriptors(a, vdd, np, depth); + if (r < 0) + return (r); + if (vdd->vdd_type != VDD_JOLIET) { + /* + * This extract record is used by SUSP,RRIP. + * Not for joliet. + */ + for (extr = np->extr_rec_list.first; + extr != NULL; + extr = extr->next) { + unsigned char *wb; + + wb = wb_buffptr(a); + memcpy(wb, extr->buf, extr->offset); + memset(wb + extr->offset, 0, + LOGICAL_BLOCK_SIZE - extr->offset); + r = wb_consume(a, LOGICAL_BLOCK_SIZE); + if (r < 0) + return (r); + } + } + + if (np->subdirs.first != NULL && depth + 1 < vdd->max_depth) { + /* Enter to sub directories. */ + np = np->subdirs.first; + depth++; + continue; + } + while (np != np->parent) { + if (np->drnext == NULL) { + /* Return to the parent directory. */ + np = np->parent; + depth--; + } else { + np = np->drnext; + break; + } + } + } while (np != np->parent); + + return (ARCHIVE_OK); +} + +/* + * Read file contents from the temporary file, and write it. + */ +static int +write_file_contents(struct archive_write *a, int64_t offset, int64_t size) +{ + struct iso9660 *iso9660 = a->format_data; + int r; + + lseek(iso9660->temp_fd, offset, SEEK_SET); + + while (size) { + size_t rsize; + ssize_t rs; + unsigned char *wb; + + wb = wb_buffptr(a); + rsize = wb_remaining(a); + if (rsize > (size_t)size) + rsize = (size_t)size; + rs = read(iso9660->temp_fd, wb, rsize); + if (rs <= 0) { + archive_set_error(&a->archive, errno, + "Can't read temporary file(%jd)", (intmax_t)rs); + return (ARCHIVE_FATAL); + } + size -= rs; + r = wb_consume(a, rs); + if (r < 0) + return (r); + } + return (ARCHIVE_OK); +} + +static int +write_file_descriptors(struct archive_write *a) +{ + struct iso9660 *iso9660 = a->format_data; + struct isofile *file; + int64_t blocks, offset; + int r; + + blocks = 0; + offset = 0; + + /* Make the boot catalog contents, and write it. */ + if (iso9660->el_torito.catalog != NULL) { + r = make_boot_catalog(a); + if (r < 0) + return (r); + } + + /* Write the boot file contents. */ + if (iso9660->el_torito.boot != NULL) { + file = iso9660->el_torito.boot->file; + blocks = file->content.blocks; + offset = file->content.offset_of_temp; + if (offset != 0) { + r = write_file_contents(a, offset, + blocks << LOGICAL_BLOCK_BITS); + if (r < 0) + return (r); + blocks = 0; + offset = 0; + } + } + + /* Write out all file contents. */ + for (file = iso9660->data_file_list.first; + file != NULL; file = file->datanext) { + + if (!file->write_content) + continue; + + if ((offset + (blocks << LOGICAL_BLOCK_BITS)) < + file->content.offset_of_temp) { + if (blocks > 0) { + r = write_file_contents(a, offset, + blocks << LOGICAL_BLOCK_BITS); + if (r < 0) + return (r); + } + blocks = 0; + offset = file->content.offset_of_temp; + } + + file->cur_content = &(file->content); + do { + blocks += file->cur_content->blocks; + /* Next fragment */ + file->cur_content = file->cur_content->next; + } while (file->cur_content != NULL); + } + + /* Flush out remaining blocks. */ + if (blocks > 0) { + r = write_file_contents(a, offset, + blocks << LOGICAL_BLOCK_BITS); + if (r < 0) + return (r); + } + + return (ARCHIVE_OK); +} + +static void +isofile_init_entry_list(struct iso9660 *iso9660) +{ + iso9660->all_file_list.first = NULL; + iso9660->all_file_list.last = &(iso9660->all_file_list.first); +} + +static void +isofile_add_entry(struct iso9660 *iso9660, struct isofile *file) +{ + file->allnext = NULL; + *iso9660->all_file_list.last = file; + iso9660->all_file_list.last = &(file->allnext); +} + +static void +isofile_free_all_entries(struct iso9660 *iso9660) +{ + struct isofile *file, *file_next; + + file = iso9660->all_file_list.first; + while (file != NULL) { + file_next = file->allnext; + isofile_free(file); + file = file_next; + } +} + +static void +isofile_init_entry_data_file_list(struct iso9660 *iso9660) +{ + iso9660->data_file_list.first = NULL; + iso9660->data_file_list.last = &(iso9660->data_file_list.first); +} + +static void +isofile_add_data_file(struct iso9660 *iso9660, struct isofile *file) +{ + file->datanext = NULL; + *iso9660->data_file_list.last = file; + iso9660->data_file_list.last = &(file->datanext); +} + + +static struct isofile * +isofile_new(struct archive_write *a, struct archive_entry *entry) +{ + struct isofile *file; + + file = calloc(1, sizeof(*file)); + if (file == NULL) + return (NULL); + + if (entry != NULL) + file->entry = archive_entry_clone(entry); + else + file->entry = archive_entry_new2(&a->archive); + if (file->entry == NULL) { + free(file); + return (NULL); + } + archive_string_init(&(file->parentdir)); + archive_string_init(&(file->basename)); + archive_string_init(&(file->basename_utf16)); + archive_string_init(&(file->symlink)); + file->cur_content = &(file->content); + + return (file); +} + +static void +isofile_free(struct isofile *file) +{ + struct content *con, *tmp; + + con = file->content.next; + while (con != NULL) { + tmp = con; + con = con->next; + free(tmp); + } + archive_entry_free(file->entry); + archive_string_free(&(file->parentdir)); + archive_string_free(&(file->basename)); + archive_string_free(&(file->basename_utf16)); + archive_string_free(&(file->symlink)); + free(file); +} + +#if defined(_WIN32) || defined(__CYGWIN__) +static int +cleanup_backslash_1(char *p) +{ + int mb, dos; + + mb = dos = 0; + while (*p) { + if (*(unsigned char *)p > 127) + mb = 1; + if (*p == '\\') { + /* If we have not met any multi-byte characters, + * we can replace '\' with '/'. */ + if (!mb) + *p = '/'; + dos = 1; + } + p++; + } + if (!mb || !dos) + return (0); + return (-1); +} + +static void +cleanup_backslash_2(wchar_t *p) +{ + + /* Convert a path-separator from '\' to '/' */ + while (*p != L'\0') { + if (*p == L'\\') + *p = L'/'; + p++; + } +} +#endif + +/* + * Generate a parent directory name and a base name from a pathname. + */ +static int +isofile_gen_utility_names(struct archive_write *a, struct isofile *file) +{ + struct iso9660 *iso9660; + const char *pathname; + char *p, *dirname, *slash; + size_t len; + int ret = ARCHIVE_OK; + + iso9660 = a->format_data; + + archive_string_empty(&(file->parentdir)); + archive_string_empty(&(file->basename)); + archive_string_empty(&(file->basename_utf16)); + archive_string_empty(&(file->symlink)); + + pathname = archive_entry_pathname(file->entry); + if (pathname == NULL || pathname[0] == '\0') {/* virtual root */ + file->dircnt = 0; + return (ret); + } + + /* + * Make a UTF-16BE basename if Joliet extension enabled. + */ + if (iso9660->opt.joliet) { + const char *u16, *ulast; + size_t u16len, ulen_last; + + if (iso9660->sconv_to_utf16be == NULL) { + iso9660->sconv_to_utf16be = + archive_string_conversion_to_charset( + &(a->archive), "UTF-16BE", 1); + if (iso9660->sconv_to_utf16be == NULL) + /* Couldn't allocate memory */ + return (ARCHIVE_FATAL); + iso9660->sconv_from_utf16be = + archive_string_conversion_from_charset( + &(a->archive), "UTF-16BE", 1); + if (iso9660->sconv_from_utf16be == NULL) + /* Couldn't allocate memory */ + return (ARCHIVE_FATAL); + } + + /* + * Convert a filename to UTF-16BE. + */ + if (0 > archive_entry_pathname_l(file->entry, &u16, &u16len, + iso9660->sconv_to_utf16be)) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for UTF-16BE"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "A filename cannot be converted to UTF-16BE;" + "You should disable making Joliet extension"); + ret = ARCHIVE_WARN; + } + + /* + * Make sure a path separator is not in the last; + * Remove trailing '/'. + */ + while (u16len >= 2) { +#if defined(_WIN32) || defined(__CYGWIN__) + if (u16[u16len-2] == 0 && + (u16[u16len-1] == '/' || u16[u16len-1] == '\\')) +#else + if (u16[u16len-2] == 0 && u16[u16len-1] == '/') +#endif + { + u16len -= 2; + } else + break; + } + + /* + * Find a basename in UTF-16BE. + */ + ulast = u16; + u16len >>= 1; + ulen_last = u16len; + while (u16len > 0) { +#if defined(_WIN32) || defined(__CYGWIN__) + if (u16[0] == 0 && (u16[1] == '/' || u16[1] == '\\')) +#else + if (u16[0] == 0 && u16[1] == '/') +#endif + { + ulast = u16 + 2; + ulen_last = u16len -1; + } + u16 += 2; + u16len --; + } + ulen_last <<= 1; + if (archive_string_ensure(&(file->basename_utf16), + ulen_last) == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for UTF-16BE"); + return (ARCHIVE_FATAL); + } + + /* + * Set UTF-16BE basename. + */ + memcpy(file->basename_utf16.s, ulast, ulen_last); + file->basename_utf16.length = ulen_last; + } + + archive_strcpy(&(file->parentdir), pathname); +#if defined(_WIN32) || defined(__CYGWIN__) + /* + * Convert a path-separator from '\' to '/' + */ + if (cleanup_backslash_1(file->parentdir.s) != 0) { + const wchar_t *wp = archive_entry_pathname_w(file->entry); + struct archive_wstring ws; + + if (wp != NULL) { + int r; + archive_string_init(&ws); + archive_wstrcpy(&ws, wp); + cleanup_backslash_2(ws.s); + archive_string_empty(&(file->parentdir)); + r = archive_string_append_from_wcs(&(file->parentdir), + ws.s, ws.length); + archive_wstring_free(&ws); + if (r < 0 && errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + } + } +#endif + + len = file->parentdir.length; + p = dirname = file->parentdir.s; + + /* + * Remove leading '/', '../' and './' elements + */ + while (*p) { + if (p[0] == '/') { + p++; + len--; + } else if (p[0] != '.') + break; + else if (p[1] == '.' && p[2] == '/') { + p += 3; + len -= 3; + } else if (p[1] == '/' || (p[1] == '.' && p[2] == '\0')) { + p += 2; + len -= 2; + } else if (p[1] == '\0') { + p++; + len--; + } else + break; + } + if (p != dirname) { + memmove(dirname, p, len+1); + p = dirname; + } + /* + * Remove "/","/." and "/.." elements from tail. + */ + while (len > 0) { + size_t ll = len; + + if (len > 0 && p[len-1] == '/') { + p[len-1] = '\0'; + len--; + } + if (len > 1 && p[len-2] == '/' && p[len-1] == '.') { + p[len-2] = '\0'; + len -= 2; + } + if (len > 2 && p[len-3] == '/' && p[len-2] == '.' && + p[len-1] == '.') { + p[len-3] = '\0'; + len -= 3; + } + if (ll == len) + break; + } + while (*p) { + if (p[0] == '/') { + if (p[1] == '/') + /* Convert '//' --> '/' */ + memmove(p, p+1, strlen(p+1) + 1); + else if (p[1] == '.' && p[2] == '/') + /* Convert '/./' --> '/' */ + memmove(p, p+2, strlen(p+2) + 1); + else if (p[1] == '.' && p[2] == '.' && p[3] == '/') { + /* Convert 'dir/dir1/../dir2/' + * --> 'dir/dir2/' + */ + char *rp = p -1; + while (rp >= dirname) { + if (*rp == '/') + break; + --rp; + } + if (rp > dirname) { + strcpy(rp, p+3); + p = rp; + } else { + strcpy(dirname, p+4); + p = dirname; + } + } else + p++; + } else + p++; + } + p = dirname; + len = strlen(p); + + if (archive_entry_filetype(file->entry) == AE_IFLNK) { + /* Convert symlink name too. */ + pathname = archive_entry_symlink(file->entry); + archive_strcpy(&(file->symlink), pathname); +#if defined(_WIN32) || defined(__CYGWIN__) + /* + * Convert a path-separator from '\' to '/' + */ + if (archive_strlen(&(file->symlink)) > 0 && + cleanup_backslash_1(file->symlink.s) != 0) { + const wchar_t *wp = + archive_entry_symlink_w(file->entry); + struct archive_wstring ws; + + if (wp != NULL) { + int r; + archive_string_init(&ws); + archive_wstrcpy(&ws, wp); + cleanup_backslash_2(ws.s); + archive_string_empty(&(file->symlink)); + r = archive_string_append_from_wcs( + &(file->symlink), + ws.s, ws.length); + archive_wstring_free(&ws); + if (r < 0 && errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + } + } +#endif + } + /* + * - Count up directory elements. + * - Find out the position which points the last position of + * path separator('/'). + */ + slash = NULL; + file->dircnt = 0; + for (; *p != '\0'; p++) + if (*p == '/') { + slash = p; + file->dircnt++; + } + if (slash == NULL) { + /* The pathname doesn't have a parent directory. */ + file->parentdir.length = len; + archive_string_copy(&(file->basename), &(file->parentdir)); + archive_string_empty(&(file->parentdir)); + *file->parentdir.s = '\0'; + return (ret); + } + + /* Make a basename from dirname and slash */ + *slash = '\0'; + file->parentdir.length = slash - dirname; + archive_strcpy(&(file->basename), slash + 1); + if (archive_entry_filetype(file->entry) == AE_IFDIR) + file->dircnt ++; + return (ret); +} + +/* + * Register a entry to get a hardlink target. + */ +static int +isofile_register_hardlink(struct archive_write *a, struct isofile *file) +{ + struct iso9660 *iso9660 = a->format_data; + struct hardlink *hl; + const char *pathname; + + archive_entry_set_nlink(file->entry, 1); + pathname = archive_entry_hardlink(file->entry); + if (pathname == NULL) { + /* This `file` is a hardlink target. */ + hl = malloc(sizeof(*hl)); + if (hl == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + hl->nlink = 1; + /* A hardlink target must be the first position. */ + file->hlnext = NULL; + hl->file_list.first = file; + hl->file_list.last = &(file->hlnext); + __archive_rb_tree_insert_node(&(iso9660->hardlink_rbtree), + (struct archive_rb_node *)hl); + } else { + hl = (struct hardlink *)__archive_rb_tree_find_node( + &(iso9660->hardlink_rbtree), pathname); + if (hl != NULL) { + /* Insert `file` entry into the tail. */ + file->hlnext = NULL; + *hl->file_list.last = file; + hl->file_list.last = &(file->hlnext); + hl->nlink++; + } + archive_entry_unset_size(file->entry); + } + + return (ARCHIVE_OK); +} + +/* + * Hardlinked files have to have the same location of extent. + * We have to find out hardlink target entries for the entries + * which have a hardlink target name. + */ +static void +isofile_connect_hardlink_files(struct iso9660 *iso9660) +{ + struct archive_rb_node *n; + struct hardlink *hl; + struct isofile *target, *nf; + + ARCHIVE_RB_TREE_FOREACH(n, &(iso9660->hardlink_rbtree)) { + hl = (struct hardlink *)n; + + /* The first entry must be a hardlink target. */ + target = hl->file_list.first; + archive_entry_set_nlink(target->entry, hl->nlink); + /* Set a hardlink target to reference entries. */ + for (nf = target->hlnext; + nf != NULL; nf = nf->hlnext) { + nf->hardlink_target = target; + archive_entry_set_nlink(nf->entry, hl->nlink); + } + } +} + +static int +isofile_hd_cmp_node(const struct archive_rb_node *n1, + const struct archive_rb_node *n2) +{ + const struct hardlink *h1 = (const struct hardlink *)n1; + const struct hardlink *h2 = (const struct hardlink *)n2; + + return (strcmp(archive_entry_pathname(h1->file_list.first->entry), + archive_entry_pathname(h2->file_list.first->entry))); +} + +static int +isofile_hd_cmp_key(const struct archive_rb_node *n, const void *key) +{ + const struct hardlink *h = (const struct hardlink *)n; + + return (strcmp(archive_entry_pathname(h->file_list.first->entry), + (const char *)key)); +} + +static void +isofile_init_hardlinks(struct iso9660 *iso9660) +{ + static const struct archive_rb_tree_ops rb_ops = { + isofile_hd_cmp_node, isofile_hd_cmp_key, + }; + + __archive_rb_tree_init(&(iso9660->hardlink_rbtree), &rb_ops); +} + +static void +isofile_free_hardlinks(struct iso9660 *iso9660) +{ + struct archive_rb_node *n, *tmp; + + ARCHIVE_RB_TREE_FOREACH_SAFE(n, &(iso9660->hardlink_rbtree), tmp) { + __archive_rb_tree_remove_node(&(iso9660->hardlink_rbtree), n); + free(n); + } +} + +static struct isoent * +isoent_new(struct isofile *file) +{ + struct isoent *isoent; + static const struct archive_rb_tree_ops rb_ops = { + isoent_cmp_node, isoent_cmp_key, + }; + + isoent = calloc(1, sizeof(*isoent)); + if (isoent == NULL) + return (NULL); + isoent->file = file; + isoent->children.first = NULL; + isoent->children.last = &(isoent->children.first); + __archive_rb_tree_init(&(isoent->rbtree), &rb_ops); + isoent->subdirs.first = NULL; + isoent->subdirs.last = &(isoent->subdirs.first); + isoent->extr_rec_list.first = NULL; + isoent->extr_rec_list.last = &(isoent->extr_rec_list.first); + isoent->extr_rec_list.current = NULL; + if (archive_entry_filetype(file->entry) == AE_IFDIR) + isoent->dir = 1; + + return (isoent); +} + +static inline struct isoent * +isoent_clone(struct isoent *src) +{ + return (isoent_new(src->file)); +} + +static void +_isoent_free(struct isoent *isoent) +{ + struct extr_rec *er, *er_next; + + free(isoent->children_sorted); + free(isoent->identifier); + er = isoent->extr_rec_list.first; + while (er != NULL) { + er_next = er->next; + free(er); + er = er_next; + } + free(isoent); +} + +static void +isoent_free_all(struct isoent *isoent) +{ + struct isoent *np, *np_temp; + + if (isoent == NULL) + return; + np = isoent; + for (;;) { + if (np->dir) { + if (np->children.first != NULL) { + /* Enter to sub directories. */ + np = np->children.first; + continue; + } + } + for (;;) { + np_temp = np; + if (np->chnext == NULL) { + /* Return to the parent directory. */ + np = np->parent; + _isoent_free(np_temp); + if (np == np_temp) + return; + } else { + np = np->chnext; + _isoent_free(np_temp); + break; + } + } + } +} + +static struct isoent * +isoent_create_virtual_dir(struct archive_write *a, struct iso9660 *iso9660, const char *pathname) +{ + struct isofile *file; + struct isoent *isoent; + + file = isofile_new(a, NULL); + if (file == NULL) + return (NULL); + archive_entry_set_pathname(file->entry, pathname); + archive_entry_unset_mtime(file->entry); + archive_entry_unset_atime(file->entry); + archive_entry_unset_ctime(file->entry); + archive_entry_set_uid(file->entry, getuid()); + archive_entry_set_gid(file->entry, getgid()); + archive_entry_set_mode(file->entry, 0555 | AE_IFDIR); + archive_entry_set_nlink(file->entry, 2); + if (isofile_gen_utility_names(a, file) < ARCHIVE_WARN) { + isofile_free(file); + return (NULL); + } + isofile_add_entry(iso9660, file); + + isoent = isoent_new(file); + if (isoent == NULL) + return (NULL); + isoent->dir = 1; + isoent->virtual = 1; + + return (isoent); +} + +static int +isoent_cmp_node(const struct archive_rb_node *n1, + const struct archive_rb_node *n2) +{ + const struct isoent *e1 = (const struct isoent *)n1; + const struct isoent *e2 = (const struct isoent *)n2; + + return (strcmp(e1->file->basename.s, e2->file->basename.s)); +} + +static int +isoent_cmp_key(const struct archive_rb_node *n, const void *key) +{ + const struct isoent *e = (const struct isoent *)n; + + return (strcmp(e->file->basename.s, (const char *)key)); +} + +static int +isoent_add_child_head(struct isoent *parent, struct isoent *child) +{ + + if (!__archive_rb_tree_insert_node( + &(parent->rbtree), (struct archive_rb_node *)child)) + return (0); + if ((child->chnext = parent->children.first) == NULL) + parent->children.last = &(child->chnext); + parent->children.first = child; + parent->children.cnt++; + child->parent = parent; + + /* Add a child to a sub-directory chain */ + if (child->dir) { + if ((child->drnext = parent->subdirs.first) == NULL) + parent->subdirs.last = &(child->drnext); + parent->subdirs.first = child; + parent->subdirs.cnt++; + child->parent = parent; + } else + child->drnext = NULL; + return (1); +} + +static int +isoent_add_child_tail(struct isoent *parent, struct isoent *child) +{ + + if (!__archive_rb_tree_insert_node( + &(parent->rbtree), (struct archive_rb_node *)child)) + return (0); + child->chnext = NULL; + *parent->children.last = child; + parent->children.last = &(child->chnext); + parent->children.cnt++; + child->parent = parent; + + /* Add a child to a sub-directory chain */ + child->drnext = NULL; + if (child->dir) { + *parent->subdirs.last = child; + parent->subdirs.last = &(child->drnext); + parent->subdirs.cnt++; + child->parent = parent; + } + return (1); +} + +static void +isoent_remove_child(struct isoent *parent, struct isoent *child) +{ + struct isoent *ent; + + /* Remove a child entry from children chain. */ + ent = parent->children.first; + while (ent->chnext != child) + ent = ent->chnext; + if ((ent->chnext = ent->chnext->chnext) == NULL) + parent->children.last = &(ent->chnext); + parent->children.cnt--; + + if (child->dir) { + /* Remove a child entry from sub-directory chain. */ + ent = parent->subdirs.first; + while (ent->drnext != child) + ent = ent->drnext; + if ((ent->drnext = ent->drnext->drnext) == NULL) + parent->subdirs.last = &(ent->drnext); + parent->subdirs.cnt--; + } + + __archive_rb_tree_remove_node(&(parent->rbtree), + (struct archive_rb_node *)child); +} + +static int +isoent_clone_tree(struct archive_write *a, struct isoent **nroot, + struct isoent *root) +{ + struct isoent *np, *xroot, *newent; + + np = root; + xroot = NULL; + do { + newent = isoent_clone(np); + if (newent == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + if (xroot == NULL) { + *nroot = xroot = newent; + newent->parent = xroot; + } else + isoent_add_child_tail(xroot, newent); + if (np->dir && np->children.first != NULL) { + /* Enter to sub directories. */ + np = np->children.first; + xroot = newent; + continue; + } + while (np != np->parent) { + if (np->chnext == NULL) { + /* Return to the parent directory. */ + np = np->parent; + xroot = xroot->parent; + } else { + np = np->chnext; + break; + } + } + } while (np != np->parent); + + return (ARCHIVE_OK); +} + +/* + * Setup directory locations. + */ +static void +isoent_setup_directory_location(struct iso9660 *iso9660, int location, + struct vdd *vdd) +{ + struct isoent *np; + int depth; + + vdd->total_dir_block = 0; + depth = 0; + np = vdd->rootent; + do { + int block; + + np->dir_block = calculate_directory_descriptors( + iso9660, vdd, np, depth); + vdd->total_dir_block += np->dir_block; + np->dir_location = location; + location += np->dir_block; + block = extra_setup_location(np, location); + vdd->total_dir_block += block; + location += block; + + if (np->subdirs.first != NULL && depth + 1 < vdd->max_depth) { + /* Enter to sub directories. */ + np = np->subdirs.first; + depth++; + continue; + } + while (np != np->parent) { + if (np->drnext == NULL) { + /* Return to the parent directory. */ + np = np->parent; + depth--; + } else { + np = np->drnext; + break; + } + } + } while (np != np->parent); +} + +static void +_isoent_file_location(struct iso9660 *iso9660, struct isoent *isoent, + int *symlocation) +{ + struct isoent **children; + int n; + + if (isoent->children.cnt == 0) + return; + + children = isoent->children_sorted; + for (n = 0; n < isoent->children.cnt; n++) { + struct isoent *np; + struct isofile *file; + + np = children[n]; + if (np->dir) + continue; + if (np == iso9660->el_torito.boot) + continue; + file = np->file; + if (file->boot || file->hardlink_target != NULL) + continue; + if (archive_entry_filetype(file->entry) == AE_IFLNK || + file->content.size == 0) { + /* + * Do not point a valid location. + * Make sure entry is not hardlink file. + */ + file->content.location = (*symlocation)--; + continue; + } + + file->write_content = 1; + } +} + +/* + * Setup file locations. + */ +static void +isoent_setup_file_location(struct iso9660 *iso9660, int location) +{ + struct isoent *isoent; + struct isoent *np; + struct isofile *file; + size_t size; + int block; + int depth; + int joliet; + int symlocation; + int total_block; + + iso9660->total_file_block = 0; + if ((isoent = iso9660->el_torito.catalog) != NULL) { + isoent->file->content.location = location; + block = (int)((archive_entry_size(isoent->file->entry) + + LOGICAL_BLOCK_SIZE -1) >> LOGICAL_BLOCK_BITS); + location += block; + iso9660->total_file_block += block; + } + if ((isoent = iso9660->el_torito.boot) != NULL) { + isoent->file->content.location = location; + size = fd_boot_image_size(iso9660->el_torito.media_type); + if (size == 0) + size = (size_t)archive_entry_size(isoent->file->entry); + block = ((int)size + LOGICAL_BLOCK_SIZE -1) + >> LOGICAL_BLOCK_BITS; + location += block; + iso9660->total_file_block += block; + isoent->file->content.blocks = block; + } + + depth = 0; + symlocation = -16; + if (!iso9660->opt.rr && iso9660->opt.joliet) { + joliet = 1; + np = iso9660->joliet.rootent; + } else { + joliet = 0; + np = iso9660->primary.rootent; + } + do { + _isoent_file_location(iso9660, np, &symlocation); + + if (np->subdirs.first != NULL && + (joliet || + ((iso9660->opt.rr == OPT_RR_DISABLED && + depth + 2 < iso9660->primary.max_depth) || + (iso9660->opt.rr && + depth + 1 < iso9660->primary.max_depth)))) { + /* Enter to sub directories. */ + np = np->subdirs.first; + depth++; + continue; + } + while (np != np->parent) { + if (np->drnext == NULL) { + /* Return to the parent directory. */ + np = np->parent; + depth--; + } else { + np = np->drnext; + break; + } + } + } while (np != np->parent); + + total_block = 0; + for (file = iso9660->data_file_list.first; + file != NULL; file = file->datanext) { + + if (!file->write_content) + continue; + + file->cur_content = &(file->content); + do { + file->cur_content->location = location; + location += file->cur_content->blocks; + total_block += file->cur_content->blocks; + /* Next fragment */ + file->cur_content = file->cur_content->next; + } while (file->cur_content != NULL); + } + iso9660->total_file_block += total_block; +} + +static int +get_path_component(char *name, size_t n, const char *fn) +{ + char *p; + size_t l; + + p = strchr(fn, '/'); + if (p == NULL) { + if ((l = strlen(fn)) == 0) + return (0); + } else + l = p - fn; + if (l > n -1) + return (-1); + memcpy(name, fn, l); + name[l] = '\0'; + + return ((int)l); +} + +/* + * Add a new entry into the tree. + */ +static int +isoent_tree(struct archive_write *a, struct isoent **isoentpp) +{ +#if defined(_WIN32) && !defined(__CYGWIN__) + char name[_MAX_FNAME];/* Included null terminator size. */ +#elif defined(NAME_MAX) && NAME_MAX >= 255 + char name[NAME_MAX+1]; +#else + char name[256]; +#endif + struct iso9660 *iso9660 = a->format_data; + struct isoent *dent, *isoent, *np; + struct isofile *f1, *f2; + const char *fn, *p; + int l; + + isoent = *isoentpp; + dent = iso9660->primary.rootent; + if (isoent->file->parentdir.length > 0) + fn = p = isoent->file->parentdir.s; + else + fn = p = ""; + + /* + * If the path of the parent directory of `isoent' entry is + * the same as the path of `cur_dirent', add isoent to + * `cur_dirent'. + */ + if (archive_strlen(&(iso9660->cur_dirstr)) + == archive_strlen(&(isoent->file->parentdir)) && + strcmp(iso9660->cur_dirstr.s, fn) == 0) { + if (!isoent_add_child_tail(iso9660->cur_dirent, isoent)) { + np = (struct isoent *)__archive_rb_tree_find_node( + &(iso9660->cur_dirent->rbtree), + isoent->file->basename.s); + goto same_entry; + } + return (ARCHIVE_OK); + } + + for (;;) { + l = get_path_component(name, sizeof(name), fn); + if (l == 0) { + np = NULL; + break; + } + if (l < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "A name buffer is too small"); + _isoent_free(isoent); + return (ARCHIVE_FATAL); + } + + np = isoent_find_child(dent, name); + if (np == NULL || fn[0] == '\0') + break; + + /* Find next subdirectory. */ + if (!np->dir) { + /* NOT Directory! */ + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "`%s' is not directory, we cannot insert `%s' ", + archive_entry_pathname(np->file->entry), + archive_entry_pathname(isoent->file->entry)); + _isoent_free(isoent); + *isoentpp = NULL; + return (ARCHIVE_FAILED); + } + fn += l; + if (fn[0] == '/') + fn++; + dent = np; + } + if (np == NULL) { + /* + * Create virtual parent directories. + */ + while (fn[0] != '\0') { + struct isoent *vp; + struct archive_string as; + + archive_string_init(&as); + archive_strncat(&as, p, fn - p + l); + if (as.s[as.length-1] == '/') { + as.s[as.length-1] = '\0'; + as.length--; + } + vp = isoent_create_virtual_dir(a, iso9660, as.s); + if (vp == NULL) { + archive_string_free(&as); + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + _isoent_free(isoent); + *isoentpp = NULL; + return (ARCHIVE_FATAL); + } + archive_string_free(&as); + + if (vp->file->dircnt > iso9660->dircnt_max) + iso9660->dircnt_max = vp->file->dircnt; + isoent_add_child_tail(dent, vp); + np = vp; + + fn += l; + if (fn[0] == '/') + fn++; + l = get_path_component(name, sizeof(name), fn); + if (l < 0) { + archive_string_free(&as); + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "A name buffer is too small"); + _isoent_free(isoent); + *isoentpp = NULL; + return (ARCHIVE_FATAL); + } + dent = np; + } + + /* Found out the parent directory where isoent can be + * inserted. */ + iso9660->cur_dirent = dent; + archive_string_empty(&(iso9660->cur_dirstr)); + archive_string_ensure(&(iso9660->cur_dirstr), + archive_strlen(&(dent->file->parentdir)) + + archive_strlen(&(dent->file->basename)) + 2); + if (archive_strlen(&(dent->file->parentdir)) + + archive_strlen(&(dent->file->basename)) == 0) + iso9660->cur_dirstr.s[0] = 0; + else { + if (archive_strlen(&(dent->file->parentdir)) > 0) { + archive_string_copy(&(iso9660->cur_dirstr), + &(dent->file->parentdir)); + archive_strappend_char(&(iso9660->cur_dirstr), '/'); + } + archive_string_concat(&(iso9660->cur_dirstr), + &(dent->file->basename)); + } + + if (!isoent_add_child_tail(dent, isoent)) { + np = (struct isoent *)__archive_rb_tree_find_node( + &(dent->rbtree), isoent->file->basename.s); + goto same_entry; + } + return (ARCHIVE_OK); + } + +same_entry: + /* + * We have already has the entry the filename of which is + * the same. + */ + f1 = np->file; + f2 = isoent->file; + + /* If the file type of entries is different, + * we cannot handle it. */ + if (archive_entry_filetype(f1->entry) != + archive_entry_filetype(f2->entry)) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Found duplicate entries `%s' and its file type is " + "different", + archive_entry_pathname(f1->entry)); + _isoent_free(isoent); + *isoentpp = NULL; + return (ARCHIVE_FAILED); + } + + /* Swap file entries. */ + np->file = f2; + isoent->file = f1; + np->virtual = 0; + + _isoent_free(isoent); + *isoentpp = np; + return (ARCHIVE_OK); +} + +/* + * Find a entry from `isoent' + */ +static struct isoent * +isoent_find_child(struct isoent *isoent, const char *child_name) +{ + struct isoent *np; + + np = (struct isoent *)__archive_rb_tree_find_node( + &(isoent->rbtree), child_name); + return (np); +} + +/* + * Find a entry full-path of which is specified by `fn' parameter, + * in the tree. + */ +static struct isoent * +isoent_find_entry(struct isoent *rootent, const char *fn) +{ +#if defined(_WIN32) && !defined(__CYGWIN__) + char name[_MAX_FNAME];/* Included null terminator size. */ +#elif defined(NAME_MAX) && NAME_MAX >= 255 + char name[NAME_MAX+1]; +#else + char name[256]; +#endif + struct isoent *isoent, *np; + int l; + + isoent = rootent; + np = NULL; + for (;;) { + l = get_path_component(name, sizeof(name), fn); + if (l == 0) + break; + fn += l; + if (fn[0] == '/') + fn++; + + np = isoent_find_child(isoent, name); + if (np == NULL) + break; + if (fn[0] == '\0') + break;/* We found out the entry */ + + /* Try sub directory. */ + isoent = np; + np = NULL; + if (!isoent->dir) + break;/* Not directory */ + } + + return (np); +} + +/* + * Following idr_* functions are used for resolving duplicated filenames + * and unreceivable filenames to generate ISO9660/Joliet Identifiers. + */ + +static void +idr_relaxed_filenames(char *map) +{ + int i; + + for (i = 0x21; i <= 0x2F; i++) + map[i] = 1; + for (i = 0x3A; i <= 0x41; i++) + map[i] = 1; + for (i = 0x5B; i <= 0x5E; i++) + map[i] = 1; + map[0x60] = 1; + for (i = 0x7B; i <= 0x7E; i++) + map[i] = 1; +} + +static void +idr_init(struct iso9660 *iso9660, struct vdd *vdd, struct idr *idr) +{ + + idr->idrent_pool = NULL; + idr->pool_size = 0; + if (vdd->vdd_type != VDD_JOLIET) { + if (iso9660->opt.iso_level <= 3) { + memcpy(idr->char_map, d_characters_map, + sizeof(idr->char_map)); + } else { + memcpy(idr->char_map, d1_characters_map, + sizeof(idr->char_map)); + idr_relaxed_filenames(idr->char_map); + } + } +} + +static void +idr_cleanup(struct idr *idr) +{ + free(idr->idrent_pool); +} + +static int +idr_ensure_poolsize(struct archive_write *a, struct idr *idr, + int cnt) +{ + + if (idr->pool_size < cnt) { + void *p; + const int bk = (1 << 7) - 1; + int psize; + + psize = (cnt + bk) & ~bk; + p = realloc(idr->idrent_pool, sizeof(struct idrent) * psize); + if (p == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + idr->idrent_pool = (struct idrent *)p; + idr->pool_size = psize; + } + return (ARCHIVE_OK); +} + +static int +idr_start(struct archive_write *a, struct idr *idr, int cnt, int ffmax, + int num_size, int null_size, const struct archive_rb_tree_ops *rbt_ops) +{ + int r; + + (void)ffmax; /* UNUSED */ + + r = idr_ensure_poolsize(a, idr, cnt); + if (r != ARCHIVE_OK) + return (r); + __archive_rb_tree_init(&(idr->rbtree), rbt_ops); + idr->wait_list.first = NULL; + idr->wait_list.last = &(idr->wait_list.first); + idr->pool_idx = 0; + idr->num_size = num_size; + idr->null_size = null_size; + return (ARCHIVE_OK); +} + +static void +idr_register(struct idr *idr, struct isoent *isoent, int weight, int noff) +{ + struct idrent *idrent, *n; + + idrent = &(idr->idrent_pool[idr->pool_idx++]); + idrent->wnext = idrent->avail = NULL; + idrent->isoent = isoent; + idrent->weight = weight; + idrent->noff = noff; + idrent->rename_num = 0; + + if (!__archive_rb_tree_insert_node(&(idr->rbtree), &(idrent->rbnode))) { + n = (struct idrent *)__archive_rb_tree_find_node( + &(idr->rbtree), idrent->isoent); + if (n != NULL) { + /* this `idrent' needs to rename. */ + idrent->avail = n; + *idr->wait_list.last = idrent; + idr->wait_list.last = &(idrent->wnext); + } + } +} + +static void +idr_extend_identifier(struct idrent *wnp, int numsize, int nullsize) +{ + unsigned char *p; + int wnp_ext_off; + + wnp_ext_off = wnp->isoent->ext_off; + if (wnp->noff + numsize != wnp_ext_off) { + p = (unsigned char *)wnp->isoent->identifier; + /* Extend the filename; foo.c --> foo___.c */ + memmove(p + wnp->noff + numsize, p + wnp_ext_off, + wnp->isoent->ext_len + nullsize); + wnp->isoent->ext_off = wnp_ext_off = wnp->noff + numsize; + wnp->isoent->id_len = wnp_ext_off + wnp->isoent->ext_len; + } +} + +static void +idr_resolve(struct idr *idr, void (*fsetnum)(unsigned char *p, int num)) +{ + struct idrent *n; + unsigned char *p; + + for (n = idr->wait_list.first; n != NULL; n = n->wnext) { + idr_extend_identifier(n, idr->num_size, idr->null_size); + p = (unsigned char *)n->isoent->identifier + n->noff; + do { + fsetnum(p, n->avail->rename_num++); + } while (!__archive_rb_tree_insert_node( + &(idr->rbtree), &(n->rbnode))); + } +} + +static void +idr_set_num(unsigned char *p, int num) +{ + static const char xdig[] = { + '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', + 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', + 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 'S', 'T', + 'U', 'V', 'W', 'X', 'Y', 'Z' + }; + + num %= sizeof(xdig) * sizeof(xdig) * sizeof(xdig); + p[0] = xdig[(num / (sizeof(xdig) * sizeof(xdig)))]; + num %= sizeof(xdig) * sizeof(xdig); + p[1] = xdig[ (num / sizeof(xdig))]; + num %= sizeof(xdig); + p[2] = xdig[num]; +} + +static void +idr_set_num_beutf16(unsigned char *p, int num) +{ + static const uint16_t xdig[] = { + 0x0030, 0x0031, 0x0032, 0x0033, 0x0034, 0x0035, + 0x0036, 0x0037, 0x0038, 0x0039, + 0x0041, 0x0042, 0x0043, 0x0044, 0x0045, 0x0046, + 0x0047, 0x0048, 0x0049, 0x004A, 0x004B, 0x004C, + 0x004D, 0x004E, 0x004F, 0x0050, 0x0051, 0x0052, + 0x0053, 0x0054, 0x0055, 0x0056, 0x0057, 0x0058, + 0x0059, 0x005A + }; +#define XDIG_CNT (sizeof(xdig)/sizeof(xdig[0])) + + num %= XDIG_CNT * XDIG_CNT * XDIG_CNT; + archive_be16enc(p, xdig[(num / (XDIG_CNT * XDIG_CNT))]); + num %= XDIG_CNT * XDIG_CNT; + archive_be16enc(p+2, xdig[ (num / XDIG_CNT)]); + num %= XDIG_CNT; + archive_be16enc(p+4, xdig[num]); +} + +/* + * Generate ISO9660 Identifier. + */ +static int +isoent_gen_iso9660_identifier(struct archive_write *a, struct isoent *isoent, + struct idr *idr) +{ + struct iso9660 *iso9660; + struct isoent *np; + char *p; + int l, r; + const char *char_map; + char allow_ldots, allow_multidot, allow_period, allow_vernum; + int fnmax, ffmax, dnmax; + static const struct archive_rb_tree_ops rb_ops = { + isoent_cmp_node_iso9660, isoent_cmp_key_iso9660 + }; + + if (isoent->children.cnt == 0) + return (0); + + iso9660 = a->format_data; + char_map = idr->char_map; + if (iso9660->opt.iso_level <= 3) { + allow_ldots = 0; + allow_multidot = 0; + allow_period = 1; + allow_vernum = iso9660->opt.allow_vernum; + if (iso9660->opt.iso_level == 1) { + fnmax = 8; + ffmax = 12;/* fnmax + '.' + 3 */ + dnmax = 8; + } else { + fnmax = 30; + ffmax = 31; + dnmax = 31; + } + } else { + allow_ldots = allow_multidot = 1; + allow_period = allow_vernum = 0; + if (iso9660->opt.rr) + /* + * MDR : The maximum size of Directory Record(254). + * DRL : A Directory Record Length(33). + * CE : A size of SUSP CE System Use Entry(28). + * MDR - DRL - CE = 254 - 33 - 28 = 193. + */ + fnmax = ffmax = dnmax = 193; + else + /* + * XA : CD-ROM XA System Use Extension + * Information(14). + * MDR - DRL - XA = 254 - 33 -14 = 207. + */ + fnmax = ffmax = dnmax = 207; + } + + r = idr_start(a, idr, isoent->children.cnt, ffmax, 3, 1, &rb_ops); + if (r < 0) + return (r); + + for (np = isoent->children.first; np != NULL; np = np->chnext) { + char *dot, *xdot; + int ext_off, noff, weight; + + l = (int)np->file->basename.length; + p = malloc(l+31+2+1); + if (p == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + memcpy(p, np->file->basename.s, l); + p[l] = '\0'; + np->identifier = p; + + dot = xdot = NULL; + if (!allow_ldots) { + /* + * If there is a '.' character at the first byte, + * it has to be replaced by '_' character. + */ + if (*p == '.') + *p++ = '_'; + } + for (;*p; p++) { + if (*p & 0x80) { + *p = '_'; + continue; + } + if (char_map[(unsigned char)*p]) { + /* if iso-level is '4', a character '.' is + * allowed by char_map. */ + if (*p == '.') { + xdot = dot; + dot = p; + } + continue; + } + if (*p >= 'a' && *p <= 'z') { + *p -= 'a' - 'A'; + continue; + } + if (*p == '.') { + xdot = dot; + dot = p; + if (allow_multidot) + continue; + } + *p = '_'; + } + p = np->identifier; + weight = -1; + if (dot == NULL) { + int nammax; + + if (np->dir) + nammax = dnmax; + else + nammax = fnmax; + + if (l > nammax) { + p[nammax] = '\0'; + weight = nammax; + ext_off = nammax; + } else + ext_off = l; + } else { + *dot = '.'; + ext_off = (int)(dot - p); + + if (iso9660->opt.iso_level == 1) { + if (dot - p <= 8) { + if (strlen(dot) > 4) { + /* A length of a file extension + * must be less than 4 */ + dot[4] = '\0'; + weight = 0; + } + } else { + p[8] = dot[0]; + p[9] = dot[1]; + p[10] = dot[2]; + p[11] = dot[3]; + p[12] = '\0'; + weight = 8; + ext_off = 8; + } + } else if (np->dir) { + if (l > dnmax) { + p[dnmax] = '\0'; + weight = dnmax; + if (ext_off > dnmax) + ext_off = dnmax; + } + } else if (l > ffmax) { + int extlen = (int)strlen(dot); + int xdoff; + + if (xdot != NULL) + xdoff = (int)(xdot - p); + else + xdoff = 0; + + if (extlen > 1 && xdoff < fnmax-1) { + int off; + + if (extlen > ffmax) + extlen = ffmax; + off = ffmax - extlen; + if (off == 0) { + /* A dot('.') character + * doesn't place to the first + * byte of identifier. */ + off ++; + extlen --; + } + memmove(p+off, dot, extlen); + p[ffmax] = '\0'; + ext_off = off; + weight = off; +#ifdef COMPAT_MKISOFS + } else if (xdoff >= fnmax-1) { + /* Simulate a bug(?) of mkisofs. */ + p[fnmax-1] = '\0'; + ext_off = fnmax-1; + weight = fnmax-1; +#endif + } else { + p[fnmax] = '\0'; + ext_off = fnmax; + weight = fnmax; + } + } + } + /* Save an offset of a file name extension to sort files. */ + np->ext_off = ext_off; + np->ext_len = (int)strlen(&p[ext_off]); + np->id_len = l = ext_off + np->ext_len; + + /* Make an offset of the number which is used to be set + * hexadecimal number to avoid duplicate identifier. */ + if (iso9660->opt.iso_level == 1) { + if (ext_off >= 5) + noff = 5; + else + noff = ext_off; + } else { + if (l == ffmax) + noff = ext_off - 3; + else if (l == ffmax-1) + noff = ext_off - 2; + else if (l == ffmax-2) + noff = ext_off - 1; + else + noff = ext_off; + } + /* Register entry to the identifier resolver. */ + idr_register(idr, np, weight, noff); + } + + /* Resolve duplicate identifier. */ + idr_resolve(idr, idr_set_num); + + /* Add a period and a version number to identifiers. */ + for (np = isoent->children.first; np != NULL; np = np->chnext) { + if (!np->dir && np->rr_child == NULL) { + p = np->identifier + np->ext_off + np->ext_len; + if (np->ext_len == 0 && allow_period) { + *p++ = '.'; + np->ext_len = 1; + } + if (np->ext_len == 1 && !allow_period) { + *--p = '\0'; + np->ext_len = 0; + } + np->id_len = np->ext_off + np->ext_len; + if (allow_vernum) { + *p++ = ';'; + *p++ = '1'; + np->id_len += 2; + } + *p = '\0'; + } else + np->id_len = np->ext_off + np->ext_len; + np->mb_len = np->id_len; + } + return (ARCHIVE_OK); +} + +/* + * Generate Joliet Identifier. + */ +static int +isoent_gen_joliet_identifier(struct archive_write *a, struct isoent *isoent, + struct idr *idr) +{ + struct iso9660 *iso9660; + struct isoent *np; + unsigned char *p; + size_t l; + int r; + size_t ffmax, parent_len; + static const struct archive_rb_tree_ops rb_ops = { + isoent_cmp_node_joliet, isoent_cmp_key_joliet + }; + + if (isoent->children.cnt == 0) + return (0); + + iso9660 = a->format_data; + if (iso9660->opt.joliet == OPT_JOLIET_LONGNAME) + ffmax = 206; + else + ffmax = 128; + + r = idr_start(a, idr, isoent->children.cnt, (int)ffmax, 6, 2, &rb_ops); + if (r < 0) + return (r); + + parent_len = 1; + for (np = isoent; np->parent != np; np = np->parent) + parent_len += np->mb_len + 1; + + for (np = isoent->children.first; np != NULL; np = np->chnext) { + unsigned char *dot; + int ext_off, noff, weight; + size_t lt; + + if ((l = np->file->basename_utf16.length) > ffmax) + l = ffmax; + + p = malloc((l+1)*2); + if (p == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + memcpy(p, np->file->basename_utf16.s, l); + p[l] = 0; + p[l+1] = 0; + + np->identifier = (char *)p; + lt = l; + dot = p + l; + weight = 0; + while (lt > 0) { + if (!joliet_allowed_char(p[0], p[1])) + archive_be16enc(p, 0x005F); /* '_' */ + else if (p[0] == 0 && p[1] == 0x2E) /* '.' */ + dot = p; + p += 2; + lt -= 2; + } + ext_off = (int)(dot - (unsigned char *)np->identifier); + np->ext_off = ext_off; + np->ext_len = (int)l - ext_off; + np->id_len = (int)l; + + /* + * Get a length of MBS of a full-pathname. + */ + if (np->file->basename_utf16.length > ffmax) { + if (archive_strncpy_l(&iso9660->mbs, + (const char *)np->identifier, l, + iso9660->sconv_from_utf16be) != 0 && + errno == ENOMEM) { + archive_set_error(&a->archive, errno, + "No memory"); + return (ARCHIVE_FATAL); + } + np->mb_len = (int)iso9660->mbs.length; + if (np->mb_len != (int)np->file->basename.length) + weight = np->mb_len; + } else + np->mb_len = (int)np->file->basename.length; + + /* If a length of full-pathname is longer than 240 bytes, + * it violates Joliet extensions regulation. */ + if (parent_len > 240 + || np->mb_len > 240 + || parent_len + np->mb_len > 240) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "The regulation of Joliet extensions;" + " A length of a full-pathname of `%s' is " + "longer than 240 bytes, (p=%d, b=%d)", + archive_entry_pathname(np->file->entry), + (int)parent_len, (int)np->mb_len); + return (ARCHIVE_FATAL); + } + + /* Make an offset of the number which is used to be set + * hexadecimal number to avoid duplicate identifier. */ + if (l == ffmax) + noff = ext_off - 6; + else if (l == ffmax-2) + noff = ext_off - 4; + else if (l == ffmax-4) + noff = ext_off - 2; + else + noff = ext_off; + /* Register entry to the identifier resolver. */ + idr_register(idr, np, weight, noff); + } + + /* Resolve duplicate identifier with Joliet Volume. */ + idr_resolve(idr, idr_set_num_beutf16); + + return (ARCHIVE_OK); +} + +/* + * This comparing rule is according to ISO9660 Standard 9.3 + */ +static int +isoent_cmp_iso9660_identifier(const struct isoent *p1, const struct isoent *p2) +{ + const char *s1, *s2; + int cmp; + int l; + + s1 = p1->identifier; + s2 = p2->identifier; + + /* Compare File Name */ + l = p1->ext_off; + if (l > p2->ext_off) + l = p2->ext_off; + cmp = memcmp(s1, s2, l); + if (cmp != 0) + return (cmp); + if (p1->ext_off < p2->ext_off) { + s2 += l; + l = p2->ext_off - p1->ext_off; + while (l--) + if (0x20 != *s2++) + return (0x20 + - *(const unsigned char *)(s2 - 1)); + } else if (p1->ext_off > p2->ext_off) { + s1 += l; + l = p1->ext_off - p2->ext_off; + while (l--) + if (0x20 != *s1++) + return (*(const unsigned char *)(s1 - 1) + - 0x20); + } + /* Compare File Name Extension */ + if (p1->ext_len == 0 && p2->ext_len == 0) + return (0); + if (p1->ext_len == 1 && p2->ext_len == 1) + return (0); + if (p1->ext_len <= 1) + return (-1); + if (p2->ext_len <= 1) + return (1); + l = p1->ext_len; + if (l > p2->ext_len) + l = p2->ext_len; + s1 = p1->identifier + p1->ext_off; + s2 = p2->identifier + p2->ext_off; + if (l > 1) { + cmp = memcmp(s1, s2, l); + if (cmp != 0) + return (cmp); + } + if (p1->ext_len < p2->ext_len) { + s2 += l; + l = p2->ext_len - p1->ext_len; + while (l--) + if (0x20 != *s2++) + return (0x20 + - *(const unsigned char *)(s2 - 1)); + } else if (p1->ext_len > p2->ext_len) { + s1 += l; + l = p1->ext_len - p2->ext_len; + while (l--) + if (0x20 != *s1++) + return (*(const unsigned char *)(s1 - 1) + - 0x20); + } + /* Compare File Version Number */ + /* No operation. The File Version Number is always one. */ + + return (cmp); +} + +static int +isoent_cmp_node_iso9660(const struct archive_rb_node *n1, + const struct archive_rb_node *n2) +{ + const struct idrent *e1 = (const struct idrent *)n1; + const struct idrent *e2 = (const struct idrent *)n2; + + return (isoent_cmp_iso9660_identifier(e2->isoent, e1->isoent)); +} + +static int +isoent_cmp_key_iso9660(const struct archive_rb_node *node, const void *key) +{ + const struct isoent *isoent = (const struct isoent *)key; + const struct idrent *idrent = (const struct idrent *)node; + + return (isoent_cmp_iso9660_identifier(isoent, idrent->isoent)); +} + +static int +isoent_cmp_joliet_identifier(const struct isoent *p1, const struct isoent *p2) +{ + const unsigned char *s1, *s2; + int cmp; + int l; + + s1 = (const unsigned char *)p1->identifier; + s2 = (const unsigned char *)p2->identifier; + + /* Compare File Name */ + l = p1->ext_off; + if (l > p2->ext_off) + l = p2->ext_off; + cmp = memcmp(s1, s2, l); + if (cmp != 0) + return (cmp); + if (p1->ext_off < p2->ext_off) { + s2 += l; + l = p2->ext_off - p1->ext_off; + while (l--) + if (0 != *s2++) + return (- *(const unsigned char *)(s2 - 1)); + } else if (p1->ext_off > p2->ext_off) { + s1 += l; + l = p1->ext_off - p2->ext_off; + while (l--) + if (0 != *s1++) + return (*(const unsigned char *)(s1 - 1)); + } + /* Compare File Name Extension */ + if (p1->ext_len == 0 && p2->ext_len == 0) + return (0); + if (p1->ext_len == 2 && p2->ext_len == 2) + return (0); + if (p1->ext_len <= 2) + return (-1); + if (p2->ext_len <= 2) + return (1); + l = p1->ext_len; + if (l > p2->ext_len) + l = p2->ext_len; + s1 = (unsigned char *)(p1->identifier + p1->ext_off); + s2 = (unsigned char *)(p2->identifier + p2->ext_off); + if (l > 1) { + cmp = memcmp(s1, s2, l); + if (cmp != 0) + return (cmp); + } + if (p1->ext_len < p2->ext_len) { + s2 += l; + l = p2->ext_len - p1->ext_len; + while (l--) + if (0 != *s2++) + return (- *(const unsigned char *)(s2 - 1)); + } else if (p1->ext_len > p2->ext_len) { + s1 += l; + l = p1->ext_len - p2->ext_len; + while (l--) + if (0 != *s1++) + return (*(const unsigned char *)(s1 - 1)); + } + /* Compare File Version Number */ + /* No operation. The File Version Number is always one. */ + + return (cmp); +} + +static int +isoent_cmp_node_joliet(const struct archive_rb_node *n1, + const struct archive_rb_node *n2) +{ + const struct idrent *e1 = (const struct idrent *)n1; + const struct idrent *e2 = (const struct idrent *)n2; + + return (isoent_cmp_joliet_identifier(e2->isoent, e1->isoent)); +} + +static int +isoent_cmp_key_joliet(const struct archive_rb_node *node, const void *key) +{ + const struct isoent *isoent = (const struct isoent *)key; + const struct idrent *idrent = (const struct idrent *)node; + + return (isoent_cmp_joliet_identifier(isoent, idrent->isoent)); +} + +static int +isoent_make_sorted_files(struct archive_write *a, struct isoent *isoent, + struct idr *idr) +{ + struct archive_rb_node *rn; + struct isoent **children; + + children = malloc(isoent->children.cnt * sizeof(struct isoent *)); + if (children == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + isoent->children_sorted = children; + + ARCHIVE_RB_TREE_FOREACH(rn, &(idr->rbtree)) { + struct idrent *idrent = (struct idrent *)rn; + *children ++ = idrent->isoent; + } + return (ARCHIVE_OK); +} + +/* + * - Generate ISO9660 and Joliet identifiers from basenames. + * - Sort files by each directory. + */ +static int +isoent_traverse_tree(struct archive_write *a, struct vdd* vdd) +{ + struct iso9660 *iso9660 = a->format_data; + struct isoent *np; + struct idr idr; + int depth; + int r; + int (*genid)(struct archive_write *, struct isoent *, struct idr *); + + idr_init(iso9660, vdd, &idr); + np = vdd->rootent; + depth = 0; + if (vdd->vdd_type == VDD_JOLIET) + genid = isoent_gen_joliet_identifier; + else + genid = isoent_gen_iso9660_identifier; + do { + if (np->virtual && + !archive_entry_mtime_is_set(np->file->entry)) { + /* Set properly times to virtual directory */ + archive_entry_set_mtime(np->file->entry, + iso9660->birth_time, 0); + archive_entry_set_atime(np->file->entry, + iso9660->birth_time, 0); + archive_entry_set_ctime(np->file->entry, + iso9660->birth_time, 0); + } + if (np->children.first != NULL) { + if (vdd->vdd_type != VDD_JOLIET && + !iso9660->opt.rr && depth + 1 >= vdd->max_depth) { + if (np->children.cnt > 0) + iso9660->directories_too_deep = np; + } else { + /* Generate Identifier */ + r = genid(a, np, &idr); + if (r < 0) + goto exit_traverse_tree; + r = isoent_make_sorted_files(a, np, &idr); + if (r < 0) + goto exit_traverse_tree; + + if (np->subdirs.first != NULL && + depth + 1 < vdd->max_depth) { + /* Enter to sub directories. */ + np = np->subdirs.first; + depth++; + continue; + } + } + } + while (np != np->parent) { + if (np->drnext == NULL) { + /* Return to the parent directory. */ + np = np->parent; + depth--; + } else { + np = np->drnext; + break; + } + } + } while (np != np->parent); + + r = ARCHIVE_OK; +exit_traverse_tree: + idr_cleanup(&idr); + + return (r); +} + +/* + * Collect directory entries into path_table by a directory depth. + */ +static int +isoent_collect_dirs(struct vdd *vdd, struct isoent *rootent, int depth) +{ + struct isoent *np; + + if (rootent == NULL) + rootent = vdd->rootent; + np = rootent; + do { + /* Register current directory to pathtable. */ + path_table_add_entry(&(vdd->pathtbl[depth]), np); + + if (np->subdirs.first != NULL && depth + 1 < vdd->max_depth) { + /* Enter to sub directories. */ + np = np->subdirs.first; + depth++; + continue; + } + while (np != rootent) { + if (np->drnext == NULL) { + /* Return to the parent directory. */ + np = np->parent; + depth--; + } else { + np = np->drnext; + break; + } + } + } while (np != rootent); + + return (ARCHIVE_OK); +} + +/* + * The entry whose number of levels in a directory hierarchy is + * large than eight relocate to rr_move directory. + */ +static int +isoent_rr_move_dir(struct archive_write *a, struct isoent **rr_moved, + struct isoent *curent, struct isoent **newent) +{ + struct iso9660 *iso9660 = a->format_data; + struct isoent *rrmoved, *mvent, *np; + + if ((rrmoved = *rr_moved) == NULL) { + struct isoent *rootent = iso9660->primary.rootent; + /* There isn't rr_move entry. + * Create rr_move entry and insert it into the root entry. + */ + rrmoved = isoent_create_virtual_dir(a, iso9660, "rr_moved"); + if (rrmoved == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + /* Add "rr_moved" entry to the root entry. */ + isoent_add_child_head(rootent, rrmoved); + archive_entry_set_nlink(rootent->file->entry, + archive_entry_nlink(rootent->file->entry) + 1); + /* Register "rr_moved" entry to second level pathtable. */ + path_table_add_entry(&(iso9660->primary.pathtbl[1]), rrmoved); + /* Save rr_moved. */ + *rr_moved = rrmoved; + } + /* + * Make a clone of curent which is going to be relocated + * to rr_moved. + */ + mvent = isoent_clone(curent); + if (mvent == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + /* linking.. and use for creating "CL", "PL" and "RE" */ + mvent->rr_parent = curent->parent; + curent->rr_child = mvent; + /* + * Move subdirectories from the curent to mvent + */ + if (curent->children.first != NULL) { + *mvent->children.last = curent->children.first; + mvent->children.last = curent->children.last; + } + for (np = mvent->children.first; np != NULL; np = np->chnext) + np->parent = mvent; + mvent->children.cnt = curent->children.cnt; + curent->children.cnt = 0; + curent->children.first = NULL; + curent->children.last = &curent->children.first; + + if (curent->subdirs.first != NULL) { + *mvent->subdirs.last = curent->subdirs.first; + mvent->subdirs.last = curent->subdirs.last; + } + mvent->subdirs.cnt = curent->subdirs.cnt; + curent->subdirs.cnt = 0; + curent->subdirs.first = NULL; + curent->subdirs.last = &curent->subdirs.first; + + /* + * The mvent becomes a child of the rr_moved entry. + */ + isoent_add_child_tail(rrmoved, mvent); + archive_entry_set_nlink(rrmoved->file->entry, + archive_entry_nlink(rrmoved->file->entry) + 1); + /* + * This entry which relocated to the rr_moved directory + * has to set the flag as a file. + * See also RRIP 4.1.5.1 Description of the "CL" System Use Entry. + */ + curent->dir = 0; + + *newent = mvent; + + return (ARCHIVE_OK); +} + +static int +isoent_rr_move(struct archive_write *a) +{ + struct iso9660 *iso9660 = a->format_data; + struct path_table *pt; + struct isoent *rootent, *rr_moved; + struct isoent *np, *last; + int r; + + pt = &(iso9660->primary.pathtbl[MAX_DEPTH-1]); + /* There aren't level 8 directories reaching a deeper level. */ + if (pt->cnt == 0) + return (ARCHIVE_OK); + + rootent = iso9660->primary.rootent; + /* If "rr_moved" directory is already existing, + * we have to use it. */ + rr_moved = isoent_find_child(rootent, "rr_moved"); + if (rr_moved != NULL && + rr_moved != rootent->children.first) { + /* + * It's necessary that rr_move is the first entry + * of the root. + */ + /* Remove "rr_moved" entry from children chain. */ + isoent_remove_child(rootent, rr_moved); + + /* Add "rr_moved" entry into the head of children chain. */ + isoent_add_child_head(rootent, rr_moved); + } + + /* + * Check level 8 path_table. + * If find out sub directory entries, that entries move to rr_move. + */ + np = pt->first; + while (np != NULL) { + last = path_table_last_entry(pt); + for (; np != NULL; np = np->ptnext) { + struct isoent *mvent; + struct isoent *newent; + + if (!np->dir) + continue; + for (mvent = np->subdirs.first; + mvent != NULL; mvent = mvent->drnext) { + r = isoent_rr_move_dir(a, &rr_moved, + mvent, &newent); + if (r < 0) + return (r); + isoent_collect_dirs(&(iso9660->primary), + newent, 2); + } + } + /* If new entries are added to level 8 path_talbe, + * its sub directory entries move to rr_move too. + */ + np = last->ptnext; + } + + return (ARCHIVE_OK); +} + +/* + * This comparing rule is according to ISO9660 Standard 6.9.1 + */ +static int +_compare_path_table(const void *v1, const void *v2) +{ + const struct isoent *p1, *p2; + const char *s1, *s2; + int cmp, l; + + p1 = *((const struct isoent **)(uintptr_t)v1); + p2 = *((const struct isoent **)(uintptr_t)v2); + + /* Compare parent directory number */ + cmp = p1->parent->dir_number - p2->parent->dir_number; + if (cmp != 0) + return (cmp); + + /* Compare identifier */ + s1 = p1->identifier; + s2 = p2->identifier; + l = p1->ext_off; + if (l > p2->ext_off) + l = p2->ext_off; + cmp = strncmp(s1, s2, l); + if (cmp != 0) + return (cmp); + if (p1->ext_off < p2->ext_off) { + s2 += l; + l = p2->ext_off - p1->ext_off; + while (l--) + if (0x20 != *s2++) + return (0x20 + - *(const unsigned char *)(s2 - 1)); + } else if (p1->ext_off > p2->ext_off) { + s1 += l; + l = p1->ext_off - p2->ext_off; + while (l--) + if (0x20 != *s1++) + return (*(const unsigned char *)(s1 - 1) + - 0x20); + } + return (0); +} + +static int +_compare_path_table_joliet(const void *v1, const void *v2) +{ + const struct isoent *p1, *p2; + const unsigned char *s1, *s2; + int cmp, l; + + p1 = *((const struct isoent **)(uintptr_t)v1); + p2 = *((const struct isoent **)(uintptr_t)v2); + + /* Compare parent directory number */ + cmp = p1->parent->dir_number - p2->parent->dir_number; + if (cmp != 0) + return (cmp); + + /* Compare identifier */ + s1 = (const unsigned char *)p1->identifier; + s2 = (const unsigned char *)p2->identifier; + l = p1->ext_off; + if (l > p2->ext_off) + l = p2->ext_off; + cmp = memcmp(s1, s2, l); + if (cmp != 0) + return (cmp); + if (p1->ext_off < p2->ext_off) { + s2 += l; + l = p2->ext_off - p1->ext_off; + while (l--) + if (0 != *s2++) + return (- *(const unsigned char *)(s2 - 1)); + } else if (p1->ext_off > p2->ext_off) { + s1 += l; + l = p1->ext_off - p2->ext_off; + while (l--) + if (0 != *s1++) + return (*(const unsigned char *)(s1 - 1)); + } + return (0); +} + +static inline void +path_table_add_entry(struct path_table *pathtbl, struct isoent *ent) +{ + ent->ptnext = NULL; + *pathtbl->last = ent; + pathtbl->last = &(ent->ptnext); + pathtbl->cnt ++; +} + +static inline struct isoent * +path_table_last_entry(struct path_table *pathtbl) +{ + if (pathtbl->first == NULL) + return (NULL); + return (((struct isoent *)(void *) + ((char *)(pathtbl->last) - offsetof(struct isoent, ptnext)))); +} + +/* + * Sort directory entries in path_table + * and assign directory number to each entries. + */ +static int +isoent_make_path_table_2(struct archive_write *a, struct vdd *vdd, + int depth, int *dir_number) +{ + struct isoent *np; + struct isoent **enttbl; + struct path_table *pt; + int i; + + pt = &vdd->pathtbl[depth]; + if (pt->cnt == 0) { + pt->sorted = NULL; + return (ARCHIVE_OK); + } + enttbl = malloc(pt->cnt * sizeof(struct isoent *)); + if (enttbl == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + pt->sorted = enttbl; + for (np = pt->first; np != NULL; np = np->ptnext) + *enttbl ++ = np; + enttbl = pt->sorted; + + switch (vdd->vdd_type) { + case VDD_PRIMARY: + case VDD_ENHANCED: +#ifdef __COMPAR_FN_T + qsort(enttbl, pt->cnt, sizeof(struct isoent *), + (__compar_fn_t)_compare_path_table); +#else + qsort(enttbl, pt->cnt, sizeof(struct isoent *), + _compare_path_table); +#endif + break; + case VDD_JOLIET: +#ifdef __COMPAR_FN_T + qsort(enttbl, pt->cnt, sizeof(struct isoent *), + (__compar_fn_t)_compare_path_table_joliet); +#else + qsort(enttbl, pt->cnt, sizeof(struct isoent *), + _compare_path_table_joliet); +#endif + break; + } + for (i = 0; i < pt->cnt; i++) + enttbl[i]->dir_number = (*dir_number)++; + + return (ARCHIVE_OK); +} + +static int +isoent_alloc_path_table(struct archive_write *a, struct vdd *vdd, + int max_depth) +{ + int i; + + vdd->max_depth = max_depth; + vdd->pathtbl = malloc(sizeof(*vdd->pathtbl) * vdd->max_depth); + if (vdd->pathtbl == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + for (i = 0; i < vdd->max_depth; i++) { + vdd->pathtbl[i].first = NULL; + vdd->pathtbl[i].last = &(vdd->pathtbl[i].first); + vdd->pathtbl[i].sorted = NULL; + vdd->pathtbl[i].cnt = 0; + } + return (ARCHIVE_OK); +} + +/* + * Make Path Tables + */ +static int +isoent_make_path_table(struct archive_write *a) +{ + struct iso9660 *iso9660 = a->format_data; + int depth, r; + int dir_number; + + /* + * Init Path Table. + */ + if (iso9660->dircnt_max >= MAX_DEPTH && + (!iso9660->opt.limit_depth || iso9660->opt.iso_level == 4)) + r = isoent_alloc_path_table(a, &(iso9660->primary), + iso9660->dircnt_max + 1); + else + /* The number of levels in the hierarchy cannot exceed + * eight. */ + r = isoent_alloc_path_table(a, &(iso9660->primary), + MAX_DEPTH); + if (r < 0) + return (r); + if (iso9660->opt.joliet) { + r = isoent_alloc_path_table(a, &(iso9660->joliet), + iso9660->dircnt_max + 1); + if (r < 0) + return (r); + } + + /* Step 0. + * - Collect directories for primary and joliet. + */ + isoent_collect_dirs(&(iso9660->primary), NULL, 0); + if (iso9660->opt.joliet) + isoent_collect_dirs(&(iso9660->joliet), NULL, 0); + /* + * Rockridge; move deeper depth directories to rr_moved. + */ + if (iso9660->opt.rr) { + r = isoent_rr_move(a); + if (r < 0) + return (r); + } + + /* Update nlink. */ + isofile_connect_hardlink_files(iso9660); + + /* Step 1. + * - Renew a value of the depth of that directories. + * - Resolve hardlinks. + * - Convert pathnames to ISO9660 name or UCS2(joliet). + * - Sort files by each directory. + */ + r = isoent_traverse_tree(a, &(iso9660->primary)); + if (r < 0) + return (r); + if (iso9660->opt.joliet) { + r = isoent_traverse_tree(a, &(iso9660->joliet)); + if (r < 0) + return (r); + } + + /* Step 2. + * - Sort directories. + * - Assign all directory number. + */ + dir_number = 1; + for (depth = 0; depth < iso9660->primary.max_depth; depth++) { + r = isoent_make_path_table_2(a, &(iso9660->primary), + depth, &dir_number); + if (r < 0) + return (r); + } + if (iso9660->opt.joliet) { + dir_number = 1; + for (depth = 0; depth < iso9660->joliet.max_depth; depth++) { + r = isoent_make_path_table_2(a, &(iso9660->joliet), + depth, &dir_number); + if (r < 0) + return (r); + } + } + if (iso9660->opt.limit_dirs && dir_number > 0xffff) { + /* + * Maximum number of directories is 65535(0xffff) + * doe to size(16bit) of Parent Directory Number of + * the Path Table. + * See also ISO9660 Standard 9.4. + */ + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Too many directories(%d) over 65535.", dir_number); + return (ARCHIVE_FATAL); + } + + /* Get the size of the Path Table. */ + calculate_path_table_size(&(iso9660->primary)); + if (iso9660->opt.joliet) + calculate_path_table_size(&(iso9660->joliet)); + + return (ARCHIVE_OK); +} + +static int +isoent_find_out_boot_file(struct archive_write *a, struct isoent *rootent) +{ + struct iso9660 *iso9660 = a->format_data; + + /* Find a isoent of the boot file. */ + iso9660->el_torito.boot = isoent_find_entry(rootent, + iso9660->el_torito.boot_filename.s); + if (iso9660->el_torito.boot == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Can't find the boot image file ``%s''", + iso9660->el_torito.boot_filename.s); + return (ARCHIVE_FATAL); + } + iso9660->el_torito.boot->file->boot = BOOT_IMAGE; + return (ARCHIVE_OK); +} + +static int +isoent_create_boot_catalog(struct archive_write *a, struct isoent *rootent) +{ + struct iso9660 *iso9660 = a->format_data; + struct isofile *file; + struct isoent *isoent; + struct archive_entry *entry; + + (void)rootent; /* UNUSED */ + /* + * Create the entry which is the "boot.catalog" file. + */ + file = isofile_new(a, NULL); + if (file == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + archive_entry_set_pathname(file->entry, + iso9660->el_torito.catalog_filename.s); + archive_entry_set_size(file->entry, LOGICAL_BLOCK_SIZE); + archive_entry_set_mtime(file->entry, iso9660->birth_time, 0); + archive_entry_set_atime(file->entry, iso9660->birth_time, 0); + archive_entry_set_ctime(file->entry, iso9660->birth_time, 0); + archive_entry_set_uid(file->entry, getuid()); + archive_entry_set_gid(file->entry, getgid()); + archive_entry_set_mode(file->entry, AE_IFREG | 0444); + archive_entry_set_nlink(file->entry, 1); + + if (isofile_gen_utility_names(a, file) < ARCHIVE_WARN) { + isofile_free(file); + return (ARCHIVE_FATAL); + } + file->boot = BOOT_CATALOG; + file->content.size = LOGICAL_BLOCK_SIZE; + isofile_add_entry(iso9660, file); + + isoent = isoent_new(file); + if (isoent == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + isoent->virtual = 1; + + /* Add the "boot.catalog" entry into tree */ + if (isoent_tree(a, &isoent) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + iso9660->el_torito.catalog = isoent; + /* + * Get a boot media type. + */ + switch (iso9660->opt.boot_type) { + default: + case OPT_BOOT_TYPE_AUTO: + /* Try detecting a media type of the boot image. */ + entry = iso9660->el_torito.boot->file->entry; + if (archive_entry_size(entry) == FD_1_2M_SIZE) + iso9660->el_torito.media_type = + BOOT_MEDIA_1_2M_DISKETTE; + else if (archive_entry_size(entry) == FD_1_44M_SIZE) + iso9660->el_torito.media_type = + BOOT_MEDIA_1_44M_DISKETTE; + else if (archive_entry_size(entry) == FD_2_88M_SIZE) + iso9660->el_torito.media_type = + BOOT_MEDIA_2_88M_DISKETTE; + else + /* We cannot decide whether the boot image is + * hard-disk. */ + iso9660->el_torito.media_type = + BOOT_MEDIA_NO_EMULATION; + break; + case OPT_BOOT_TYPE_NO_EMU: + iso9660->el_torito.media_type = BOOT_MEDIA_NO_EMULATION; + break; + case OPT_BOOT_TYPE_HARD_DISK: + iso9660->el_torito.media_type = BOOT_MEDIA_HARD_DISK; + break; + case OPT_BOOT_TYPE_FD: + entry = iso9660->el_torito.boot->file->entry; + if (archive_entry_size(entry) <= FD_1_2M_SIZE) + iso9660->el_torito.media_type = + BOOT_MEDIA_1_2M_DISKETTE; + else if (archive_entry_size(entry) <= FD_1_44M_SIZE) + iso9660->el_torito.media_type = + BOOT_MEDIA_1_44M_DISKETTE; + else if (archive_entry_size(entry) <= FD_2_88M_SIZE) + iso9660->el_torito.media_type = + BOOT_MEDIA_2_88M_DISKETTE; + else { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Boot image file(``%s'') size is too big " + "for fd type.", + iso9660->el_torito.boot_filename.s); + return (ARCHIVE_FATAL); + } + break; + } + + /* + * Get a system type. + * TODO: `El Torito' specification says "A copy of byte 5 from the + * Partition Table found in the boot image". + */ + iso9660->el_torito.system_type = 0; + + /* + * Get an ID. + */ + if (iso9660->opt.publisher) + archive_string_copy(&(iso9660->el_torito.id), + &(iso9660->publisher_identifier)); + + + return (ARCHIVE_OK); +} + +/* + * If a media type is floppy, return its image size. + * otherwise return 0. + */ +static size_t +fd_boot_image_size(int media_type) +{ + switch (media_type) { + case BOOT_MEDIA_1_2M_DISKETTE: + return (FD_1_2M_SIZE); + case BOOT_MEDIA_1_44M_DISKETTE: + return (FD_1_44M_SIZE); + case BOOT_MEDIA_2_88M_DISKETTE: + return (FD_2_88M_SIZE); + default: + return (0); + } +} + +/* + * Make a boot catalog image data. + */ +static int +make_boot_catalog(struct archive_write *a) +{ + struct iso9660 *iso9660 = a->format_data; + unsigned char *block; + unsigned char *p; + uint16_t sum, *wp; + + block = wb_buffptr(a); + memset(block, 0, LOGICAL_BLOCK_SIZE); + p = block; + /* + * Validation Entry + */ + /* Header ID */ + p[0] = 1; + /* Platform ID */ + p[1] = iso9660->el_torito.platform_id; + /* Reserved */ + p[2] = p[3] = 0; + /* ID */ + if (archive_strlen(&(iso9660->el_torito.id)) > 0) + strncpy((char *)p+4, iso9660->el_torito.id.s, 23); + p[27] = 0; + /* Checksum */ + p[28] = p[29] = 0; + /* Key */ + p[30] = 0x55; + p[31] = 0xAA; + + sum = 0; + wp = (uint16_t *)block; + while (wp < (uint16_t *)&block[32]) + sum += archive_le16dec(wp++); + set_num_721(&block[28], (~sum) + 1); + + /* + * Initial/Default Entry + */ + p = &block[32]; + /* Boot Indicator */ + p[0] = 0x88; + /* Boot media type */ + p[1] = iso9660->el_torito.media_type; + /* Load Segment */ + if (iso9660->el_torito.media_type == BOOT_MEDIA_NO_EMULATION) + set_num_721(&p[2], iso9660->el_torito.boot_load_seg); + else + set_num_721(&p[2], 0); + /* System Type */ + p[4] = iso9660->el_torito.system_type; + /* Unused */ + p[5] = 0; + /* Sector Count */ + if (iso9660->el_torito.media_type == BOOT_MEDIA_NO_EMULATION) + set_num_721(&p[6], iso9660->el_torito.boot_load_size); + else + set_num_721(&p[6], 1); + /* Load RBA */ + set_num_731(&p[8], + iso9660->el_torito.boot->file->content.location); + /* Unused */ + memset(&p[12], 0, 20); + + return (wb_consume(a, LOGICAL_BLOCK_SIZE)); +} + +static int +setup_boot_information(struct archive_write *a) +{ + struct iso9660 *iso9660 = a->format_data; + struct isoent *np; + int64_t size; + uint32_t sum; + unsigned char buff[4096]; + + np = iso9660->el_torito.boot; + lseek(iso9660->temp_fd, + np->file->content.offset_of_temp + 64, SEEK_SET); + size = archive_entry_size(np->file->entry) - 64; + if (size <= 0) { + archive_set_error(&a->archive, errno, + "Boot file(%jd) is too small", (intmax_t)size + 64); + return (ARCHIVE_FATAL); + } + sum = 0; + while (size > 0) { + size_t rsize; + ssize_t i, rs; + + if (size > (int64_t)sizeof(buff)) + rsize = sizeof(buff); + else + rsize = (size_t)size; + + rs = read(iso9660->temp_fd, buff, rsize); + if (rs <= 0) { + archive_set_error(&a->archive, errno, + "Can't read temporary file(%jd)", + (intmax_t)rs); + return (ARCHIVE_FATAL); + } + for (i = 0; i < rs; i += 4) + sum += archive_le32dec(buff + i); + size -= rs; + } + /* Set the location of Primary Volume Descriptor. */ + set_num_731(buff, SYSTEM_AREA_BLOCK); + /* Set the location of the boot file. */ + set_num_731(buff+4, np->file->content.location); + /* Set the size of the boot file. */ + size = fd_boot_image_size(iso9660->el_torito.media_type); + if (size == 0) + size = archive_entry_size(np->file->entry); + set_num_731(buff+8, (uint32_t)size); + /* Set the sum of the boot file. */ + set_num_731(buff+12, sum); + /* Clear reserved bytes. */ + memset(buff+16, 0, 40); + + /* Overwrite the boot file. */ + lseek(iso9660->temp_fd, + np->file->content.offset_of_temp + 8, SEEK_SET); + return (write_to_temp(a, buff, 56)); +} + +#ifdef HAVE_ZLIB_H + +static int +zisofs_init_zstream(struct archive_write *a) +{ + struct iso9660 *iso9660 = a->format_data; + int r; + + iso9660->zisofs.stream.next_in = NULL; + iso9660->zisofs.stream.avail_in = 0; + iso9660->zisofs.stream.total_in = 0; + iso9660->zisofs.stream.total_out = 0; + if (iso9660->zisofs.stream_valid) + r = deflateReset(&(iso9660->zisofs.stream)); + else { + r = deflateInit(&(iso9660->zisofs.stream), + iso9660->zisofs.compression_level); + iso9660->zisofs.stream_valid = 1; + } + switch (r) { + case Z_OK: + break; + default: + case Z_STREAM_ERROR: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Internal error initializing " + "compression library: invalid setup parameter"); + return (ARCHIVE_FATAL); + case Z_MEM_ERROR: + archive_set_error(&a->archive, ENOMEM, + "Internal error initializing " + "compression library"); + return (ARCHIVE_FATAL); + case Z_VERSION_ERROR: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Internal error initializing " + "compression library: invalid library version"); + return (ARCHIVE_FATAL); + } + return (ARCHIVE_OK); +} + +#endif /* HAVE_ZLIB_H */ + +static int +zisofs_init(struct archive_write *a, struct isofile *file) +{ + struct iso9660 *iso9660 = a->format_data; +#ifdef HAVE_ZLIB_H + uint64_t tsize; + size_t _ceil, bpsize; + int r; +#endif + + iso9660->zisofs.detect_magic = 0; + iso9660->zisofs.making = 0; + + if (!iso9660->opt.rr || !iso9660->opt.zisofs) + return (ARCHIVE_OK); + + if (archive_entry_size(file->entry) >= 24 && + archive_entry_size(file->entry) < MULTI_EXTENT_SIZE) { + /* Acceptable file size for zisofs. */ + iso9660->zisofs.detect_magic = 1; + iso9660->zisofs.magic_cnt = 0; + } + if (!iso9660->zisofs.detect_magic) + return (ARCHIVE_OK); + +#ifdef HAVE_ZLIB_H + /* The number of Logical Blocks which uncompressed data + * will use in iso-image file is the same as the number of + * Logical Blocks which zisofs(compressed) data will use + * in ISO-image file. It won't reduce iso-image file size. */ + if (archive_entry_size(file->entry) <= LOGICAL_BLOCK_SIZE) + return (ARCHIVE_OK); + + /* Initialize compression library */ + r = zisofs_init_zstream(a); + if (r != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + /* Mark file->zisofs to create RRIP 'ZF' Use Entry. */ + file->zisofs.header_size = ZF_HEADER_SIZE >> 2; + file->zisofs.log2_bs = ZF_LOG2_BS; + file->zisofs.uncompressed_size = + (uint32_t)archive_entry_size(file->entry); + + /* Calculate a size of Block Pointers of zisofs. */ + _ceil = (file->zisofs.uncompressed_size + ZF_BLOCK_SIZE -1) + >> file->zisofs.log2_bs; + iso9660->zisofs.block_pointers_cnt = (int)_ceil + 1; + iso9660->zisofs.block_pointers_idx = 0; + + /* Ensure a buffer size used for Block Pointers */ + bpsize = iso9660->zisofs.block_pointers_cnt * + sizeof(iso9660->zisofs.block_pointers[0]); + if (iso9660->zisofs.block_pointers_allocated < bpsize) { + free(iso9660->zisofs.block_pointers); + iso9660->zisofs.block_pointers = malloc(bpsize); + if (iso9660->zisofs.block_pointers == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate data"); + return (ARCHIVE_FATAL); + } + iso9660->zisofs.block_pointers_allocated = bpsize; + } + + /* + * Skip zisofs header and Block Pointers, which we will write + * after all compressed data of a file written to the temporary + * file. + */ + tsize = ZF_HEADER_SIZE + bpsize; + if (write_null(a, (size_t)tsize) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + /* + * Initialize some variables to make zisofs. + */ + archive_le32enc(&(iso9660->zisofs.block_pointers[0]), + (uint32_t)tsize); + iso9660->zisofs.remaining = file->zisofs.uncompressed_size; + iso9660->zisofs.making = 1; + iso9660->zisofs.allzero = 1; + iso9660->zisofs.block_offset = tsize; + iso9660->zisofs.total_size = tsize; + iso9660->cur_file->cur_content->size = tsize; +#endif + + return (ARCHIVE_OK); +} + +static void +zisofs_detect_magic(struct archive_write *a, const void *buff, size_t s) +{ + struct iso9660 *iso9660 = a->format_data; + struct isofile *file = iso9660->cur_file; + const unsigned char *p, *endp; + const unsigned char *magic_buff; + uint32_t uncompressed_size; + unsigned char header_size; + unsigned char log2_bs; + size_t _ceil, doff; + uint32_t bst, bed; + int magic_max; + int64_t entry_size; + + entry_size = archive_entry_size(file->entry); + if ((int64_t)sizeof(iso9660->zisofs.magic_buffer) > entry_size) + magic_max = (int)entry_size; + else + magic_max = sizeof(iso9660->zisofs.magic_buffer); + + if (iso9660->zisofs.magic_cnt == 0 && s >= (size_t)magic_max) + /* It's unnecessary we copy buffer. */ + magic_buff = buff; + else { + if (iso9660->zisofs.magic_cnt < magic_max) { + size_t l; + + l = sizeof(iso9660->zisofs.magic_buffer) + - iso9660->zisofs.magic_cnt; + if (l > s) + l = s; + memcpy(iso9660->zisofs.magic_buffer + + iso9660->zisofs.magic_cnt, buff, l); + iso9660->zisofs.magic_cnt += (int)l; + if (iso9660->zisofs.magic_cnt < magic_max) + return; + } + magic_buff = iso9660->zisofs.magic_buffer; + } + iso9660->zisofs.detect_magic = 0; + p = magic_buff; + + /* Check the magic code of zisofs. */ + if (memcmp(p, zisofs_magic, sizeof(zisofs_magic)) != 0) + /* This is not zisofs file which made by mkzftree. */ + return; + p += sizeof(zisofs_magic); + + /* Read a zisofs header. */ + uncompressed_size = archive_le32dec(p); + header_size = p[4]; + log2_bs = p[5]; + if (uncompressed_size < 24 || header_size != 4 || + log2_bs > 30 || log2_bs < 7) + return;/* Invalid or not supported header. */ + + /* Calculate a size of Block Pointers of zisofs. */ + _ceil = (uncompressed_size + + (ARCHIVE_LITERAL_LL(1) << log2_bs) -1) >> log2_bs; + doff = (_ceil + 1) * 4 + 16; + if (entry_size < (int64_t)doff) + return;/* Invalid data. */ + + /* Check every Block Pointer has valid value. */ + p = magic_buff + 16; + endp = magic_buff + magic_max; + while (_ceil && p + 8 <= endp) { + bst = archive_le32dec(p); + if (bst != doff) + return;/* Invalid data. */ + p += 4; + bed = archive_le32dec(p); + if (bed < bst || bed > entry_size) + return;/* Invalid data. */ + doff += bed - bst; + _ceil--; + } + + file->zisofs.uncompressed_size = uncompressed_size; + file->zisofs.header_size = header_size; + file->zisofs.log2_bs = log2_bs; + + /* Disable making a zisofs image. */ + iso9660->zisofs.making = 0; +} + +#ifdef HAVE_ZLIB_H + +/* + * Compress data and write it to a temporary file. + */ +static int +zisofs_write_to_temp(struct archive_write *a, const void *buff, size_t s) +{ + struct iso9660 *iso9660 = a->format_data; + struct isofile *file = iso9660->cur_file; + const unsigned char *b; + z_stream *zstrm; + size_t avail, csize; + int flush, r; + + zstrm = &(iso9660->zisofs.stream); + zstrm->next_out = wb_buffptr(a); + zstrm->avail_out = (uInt)wb_remaining(a); + b = (const unsigned char *)buff; + do { + avail = ZF_BLOCK_SIZE - zstrm->total_in; + if (s < avail) { + avail = s; + flush = Z_NO_FLUSH; + } else + flush = Z_FINISH; + iso9660->zisofs.remaining -= avail; + if (iso9660->zisofs.remaining <= 0) + flush = Z_FINISH; + + zstrm->next_in = (Bytef *)(uintptr_t)(const void *)b; + zstrm->avail_in = (uInt)avail; + + /* + * Check if current data block are all zero. + */ + if (iso9660->zisofs.allzero) { + const unsigned char *nonzero = b; + const unsigned char *nonzeroend = b + avail; + + while (nonzero < nonzeroend) + if (*nonzero++) { + iso9660->zisofs.allzero = 0; + break; + } + } + b += avail; + s -= avail; + + /* + * If current data block are all zero, we do not use + * compressed data. + */ + if (flush == Z_FINISH && iso9660->zisofs.allzero && + avail + zstrm->total_in == ZF_BLOCK_SIZE) { + if (iso9660->zisofs.block_offset != + file->cur_content->size) { + int64_t diff; + + r = wb_set_offset(a, + file->cur_content->offset_of_temp + + iso9660->zisofs.block_offset); + if (r != ARCHIVE_OK) + return (r); + diff = file->cur_content->size - + iso9660->zisofs.block_offset; + file->cur_content->size -= diff; + iso9660->zisofs.total_size -= diff; + } + zstrm->avail_in = 0; + } + + /* + * Compress file data. + */ + while (zstrm->avail_in > 0) { + csize = zstrm->total_out; + r = deflate(zstrm, flush); + switch (r) { + case Z_OK: + case Z_STREAM_END: + csize = zstrm->total_out - csize; + if (wb_consume(a, csize) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + iso9660->zisofs.total_size += csize; + iso9660->cur_file->cur_content->size += csize; + zstrm->next_out = wb_buffptr(a); + zstrm->avail_out = (uInt)wb_remaining(a); + break; + default: + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Compression failed:" + " deflate() call returned status %d", + r); + return (ARCHIVE_FATAL); + } + } + + if (flush == Z_FINISH) { + /* + * Save the information of one zisofs block. + */ + iso9660->zisofs.block_pointers_idx ++; + archive_le32enc(&(iso9660->zisofs.block_pointers[ + iso9660->zisofs.block_pointers_idx]), + (uint32_t)iso9660->zisofs.total_size); + r = zisofs_init_zstream(a); + if (r != ARCHIVE_OK) + return (ARCHIVE_FATAL); + iso9660->zisofs.allzero = 1; + iso9660->zisofs.block_offset = file->cur_content->size; + } + } while (s); + + return (ARCHIVE_OK); +} + +static int +zisofs_finish_entry(struct archive_write *a) +{ + struct iso9660 *iso9660 = a->format_data; + struct isofile *file = iso9660->cur_file; + unsigned char buff[16]; + size_t s; + int64_t tail; + + /* Direct temp file stream to zisofs temp file stream. */ + archive_entry_set_size(file->entry, iso9660->zisofs.total_size); + + /* + * Save a file pointer which points the end of current zisofs data. + */ + tail = wb_offset(a); + + /* + * Make a header. + * + * +-----------------+----------------+-----------------+ + * | Header 16 bytes | Block Pointers | Compressed data | + * +-----------------+----------------+-----------------+ + * 0 16 +X + * Block Pointers : + * 4 * (((Uncompressed file size + block_size -1) / block_size) + 1) + * + * Write zisofs header. + * Magic number + * +----+----+----+----+----+----+----+----+ + * | 37 | E4 | 53 | 96 | C9 | DB | D6 | 07 | + * +----+----+----+----+----+----+----+----+ + * 0 1 2 3 4 5 6 7 8 + * + * +------------------------+------------------+ + * | Uncompressed file size | header_size >> 2 | + * +------------------------+------------------+ + * 8 12 13 + * + * +-----------------+----------------+ + * | log2 block_size | Reserved(0000) | + * +-----------------+----------------+ + * 13 14 16 + */ + memcpy(buff, zisofs_magic, 8); + set_num_731(buff+8, file->zisofs.uncompressed_size); + buff[12] = file->zisofs.header_size; + buff[13] = file->zisofs.log2_bs; + buff[14] = buff[15] = 0;/* Reserved */ + + /* Move to the right position to write the header. */ + wb_set_offset(a, file->content.offset_of_temp); + + /* Write the header. */ + if (wb_write_to_temp(a, buff, 16) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + /* + * Write zisofs Block Pointers. + */ + s = iso9660->zisofs.block_pointers_cnt * + sizeof(iso9660->zisofs.block_pointers[0]); + if (wb_write_to_temp(a, iso9660->zisofs.block_pointers, s) + != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + /* Set a file pointer back to the end of the temporary file. */ + wb_set_offset(a, tail); + + return (ARCHIVE_OK); +} + +static int +zisofs_free(struct archive_write *a) +{ + struct iso9660 *iso9660 = a->format_data; + int ret = ARCHIVE_OK; + + free(iso9660->zisofs.block_pointers); + if (iso9660->zisofs.stream_valid && + deflateEnd(&(iso9660->zisofs.stream)) != Z_OK) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Failed to clean up compressor"); + ret = ARCHIVE_FATAL; + } + iso9660->zisofs.block_pointers = NULL; + iso9660->zisofs.stream_valid = 0; + return (ret); +} + +struct zisofs_extract { + int pz_log2_bs; /* Log2 of block size */ + uint64_t pz_uncompressed_size; + size_t uncompressed_buffer_size; + + signed int initialized:1; + signed int header_passed:1; + + uint32_t pz_offset; + unsigned char *block_pointers; + size_t block_pointers_size; + size_t block_pointers_avail; + size_t block_off; + uint32_t block_avail; + + z_stream stream; + int stream_valid; +}; + +static ssize_t +zisofs_extract_init(struct archive_write *a, struct zisofs_extract *zisofs, + const unsigned char *p, size_t bytes) +{ + size_t avail = bytes; + size_t _ceil, xsize; + + /* Allocate block pointers buffer. */ + _ceil = (size_t)((zisofs->pz_uncompressed_size + + (((int64_t)1) << zisofs->pz_log2_bs) - 1) + >> zisofs->pz_log2_bs); + xsize = (_ceil + 1) * 4; + if (zisofs->block_pointers == NULL) { + size_t alloc = ((xsize >> 10) + 1) << 10; + zisofs->block_pointers = malloc(alloc); + if (zisofs->block_pointers == NULL) { + archive_set_error(&a->archive, ENOMEM, + "No memory for zisofs decompression"); + return (ARCHIVE_FATAL); + } + } + zisofs->block_pointers_size = xsize; + + /* Allocate uncompressed data buffer. */ + zisofs->uncompressed_buffer_size = (size_t)1UL << zisofs->pz_log2_bs; + + /* + * Read the file header, and check the magic code of zisofs. + */ + if (!zisofs->header_passed) { + int err = 0; + if (avail < 16) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Illegal zisofs file body"); + return (ARCHIVE_FATAL); + } + + if (memcmp(p, zisofs_magic, sizeof(zisofs_magic)) != 0) + err = 1; + else if (archive_le32dec(p + 8) != zisofs->pz_uncompressed_size) + err = 1; + else if (p[12] != 4 || p[13] != zisofs->pz_log2_bs) + err = 1; + if (err) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Illegal zisofs file body"); + return (ARCHIVE_FATAL); + } + avail -= 16; + p += 16; + zisofs->header_passed = 1; + } + + /* + * Read block pointers. + */ + if (zisofs->header_passed && + zisofs->block_pointers_avail < zisofs->block_pointers_size) { + xsize = zisofs->block_pointers_size + - zisofs->block_pointers_avail; + if (avail < xsize) + xsize = avail; + memcpy(zisofs->block_pointers + + zisofs->block_pointers_avail, p, xsize); + zisofs->block_pointers_avail += xsize; + avail -= xsize; + if (zisofs->block_pointers_avail + == zisofs->block_pointers_size) { + /* We've got all block pointers and initialize + * related variables. */ + zisofs->block_off = 0; + zisofs->block_avail = 0; + /* Complete a initialization */ + zisofs->initialized = 1; + } + } + return ((ssize_t)avail); +} + +static ssize_t +zisofs_extract(struct archive_write *a, struct zisofs_extract *zisofs, + const unsigned char *p, size_t bytes) +{ + size_t avail; + int r; + + if (!zisofs->initialized) { + ssize_t rs = zisofs_extract_init(a, zisofs, p, bytes); + if (rs < 0) + return (rs); + if (!zisofs->initialized) { + /* We need more data. */ + zisofs->pz_offset += (uint32_t)bytes; + return (bytes); + } + avail = rs; + p += bytes - avail; + } else + avail = bytes; + + /* + * Get block offsets from block pointers. + */ + if (zisofs->block_avail == 0) { + uint32_t bst, bed; + + if (zisofs->block_off + 4 >= zisofs->block_pointers_size) { + /* There isn't a pair of offsets. */ + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Illegal zisofs block pointers"); + return (ARCHIVE_FATAL); + } + bst = archive_le32dec( + zisofs->block_pointers + zisofs->block_off); + if (bst != zisofs->pz_offset + (bytes - avail)) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Illegal zisofs block pointers(cannot seek)"); + return (ARCHIVE_FATAL); + } + bed = archive_le32dec( + zisofs->block_pointers + zisofs->block_off + 4); + if (bed < bst) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Illegal zisofs block pointers"); + return (ARCHIVE_FATAL); + } + zisofs->block_avail = bed - bst; + zisofs->block_off += 4; + + /* Initialize compression library for new block. */ + if (zisofs->stream_valid) + r = inflateReset(&zisofs->stream); + else + r = inflateInit(&zisofs->stream); + if (r != Z_OK) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Can't initialize zisofs decompression."); + return (ARCHIVE_FATAL); + } + zisofs->stream_valid = 1; + zisofs->stream.total_in = 0; + zisofs->stream.total_out = 0; + } + + /* + * Make uncompressed data. + */ + if (zisofs->block_avail == 0) { + /* + * It's basically 32K bytes NUL data. + */ + unsigned char *wb; + size_t size, wsize; + + size = zisofs->uncompressed_buffer_size; + while (size) { + wb = wb_buffptr(a); + if (size > wb_remaining(a)) + wsize = wb_remaining(a); + else + wsize = size; + memset(wb, 0, wsize); + r = wb_consume(a, wsize); + if (r < 0) + return (r); + size -= wsize; + } + } else { + zisofs->stream.next_in = (Bytef *)(uintptr_t)(const void *)p; + if (avail > zisofs->block_avail) + zisofs->stream.avail_in = zisofs->block_avail; + else + zisofs->stream.avail_in = (uInt)avail; + zisofs->stream.next_out = wb_buffptr(a); + zisofs->stream.avail_out = (uInt)wb_remaining(a); + + r = inflate(&zisofs->stream, 0); + switch (r) { + case Z_OK: /* Decompressor made some progress.*/ + case Z_STREAM_END: /* Found end of stream. */ + break; + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "zisofs decompression failed (%d)", r); + return (ARCHIVE_FATAL); + } + avail -= zisofs->stream.next_in - p; + zisofs->block_avail -= (uint32_t)(zisofs->stream.next_in - p); + r = wb_consume(a, wb_remaining(a) - zisofs->stream.avail_out); + if (r < 0) + return (r); + } + zisofs->pz_offset += (uint32_t)bytes; + return (bytes - avail); +} + +static int +zisofs_rewind_boot_file(struct archive_write *a) +{ + struct iso9660 *iso9660 = a->format_data; + struct isofile *file; + unsigned char *rbuff; + ssize_t r; + size_t remaining, rbuff_size; + struct zisofs_extract zext; + int64_t read_offset, write_offset, new_offset; + int fd, ret = ARCHIVE_OK; + + file = iso9660->el_torito.boot->file; + /* + * There is nothing to do if this boot file does not have + * zisofs header. + */ + if (file->zisofs.header_size == 0) + return (ARCHIVE_OK); + + /* + * Uncompress the zisofs'ed file contents. + */ + memset(&zext, 0, sizeof(zext)); + zext.pz_uncompressed_size = file->zisofs.uncompressed_size; + zext.pz_log2_bs = file->zisofs.log2_bs; + + fd = iso9660->temp_fd; + new_offset = wb_offset(a); + read_offset = file->content.offset_of_temp; + remaining = (size_t)file->content.size; + if (remaining > 1024 * 32) + rbuff_size = 1024 * 32; + else + rbuff_size = remaining; + + rbuff = malloc(rbuff_size); + if (rbuff == NULL) { + archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + while (remaining) { + size_t rsize; + ssize_t rs; + + /* Get the current file pointer. */ + write_offset = lseek(fd, 0, SEEK_CUR); + + /* Change the file pointer to read. */ + lseek(fd, read_offset, SEEK_SET); + + rsize = rbuff_size; + if (rsize > remaining) + rsize = remaining; + rs = read(iso9660->temp_fd, rbuff, rsize); + if (rs <= 0) { + archive_set_error(&a->archive, errno, + "Can't read temporary file(%jd)", (intmax_t)rs); + ret = ARCHIVE_FATAL; + break; + } + remaining -= rs; + read_offset += rs; + + /* Put the file pointer back to write. */ + lseek(fd, write_offset, SEEK_SET); + + r = zisofs_extract(a, &zext, rbuff, rs); + if (r < 0) { + ret = (int)r; + break; + } + } + + if (ret == ARCHIVE_OK) { + /* + * Change the boot file content from zisofs'ed data + * to plain data. + */ + file->content.offset_of_temp = new_offset; + file->content.size = file->zisofs.uncompressed_size; + archive_entry_set_size(file->entry, file->content.size); + /* Set to be no zisofs. */ + file->zisofs.header_size = 0; + file->zisofs.log2_bs = 0; + file->zisofs.uncompressed_size = 0; + r = wb_write_padding_to_temp(a, file->content.size); + if (r < 0) + ret = ARCHIVE_FATAL; + } + + /* + * Free the resource we used in this function only. + */ + free(rbuff); + free(zext.block_pointers); + if (zext.stream_valid && inflateEnd(&(zext.stream)) != Z_OK) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Failed to clean up compressor"); + ret = ARCHIVE_FATAL; + } + + return (ret); +} + +#else + +static int +zisofs_write_to_temp(struct archive_write *a, const void *buff, size_t s) +{ + (void)buff; /* UNUSED */ + (void)s; /* UNUSED */ + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Programming error"); + return (ARCHIVE_FATAL); +} + +static int +zisofs_rewind_boot_file(struct archive_write *a) +{ + struct iso9660 *iso9660 = a->format_data; + + if (iso9660->el_torito.boot->file->zisofs.header_size != 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "We cannot extract the zisofs imaged boot file;" + " this may not boot in being zisofs imaged"); + return (ARCHIVE_FAILED); + } + return (ARCHIVE_OK); +} + +static int +zisofs_finish_entry(struct archive_write *a) +{ + (void)a; /* UNUSED */ + return (ARCHIVE_OK); +} + +static int +zisofs_free(struct archive_write *a) +{ + (void)a; /* UNUSED */ + return (ARCHIVE_OK); +} + +#endif /* HAVE_ZLIB_H */ + diff --git a/c/archive_write_set_format_mtree.c b/c/archive_write_set_format_mtree.c new file mode 100644 --- /dev/null +++ b/c/archive_write_set_format_mtree.c @@ -0,0 +1,2228 @@ +/*- + * Copyright (c) 2008 Joerg Sonnenberger + * Copyright (c) 2009-2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_format_mtree.c 201171 2009-12-29 06:39:07Z kientzle $"); + +#ifdef HAVE_SYS_TYPES_H +#include +#endif +#include +#include +#include + +#include "archive.h" +#include "archive_digest_private.h" +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_rb.h" +#include "archive_string.h" +#include "archive_write_private.h" + +#define INDENTNAMELEN 15 +#define MAXLINELEN 80 +#define SET_KEYS \ + (F_FLAGS | F_GID | F_GNAME | F_MODE | F_TYPE | F_UID | F_UNAME) + +struct attr_counter { + struct attr_counter *prev; + struct attr_counter *next; + struct mtree_entry *m_entry; + int count; +}; + +struct att_counter_set { + struct attr_counter *uid_list; + struct attr_counter *gid_list; + struct attr_counter *mode_list; + struct attr_counter *flags_list; +}; + +struct mtree_chain { + struct mtree_entry *first; + struct mtree_entry **last; +}; + +/* + * The Data only for a directory file. + */ +struct dir_info { + struct archive_rb_tree rbtree; + struct mtree_chain children; + struct mtree_entry *chnext; + int virtual; +}; + +/* + * The Data only for a regular file. + */ +struct reg_info { + int compute_sum; + uint32_t crc; +#ifdef ARCHIVE_HAS_MD5 + unsigned char buf_md5[16]; +#endif +#ifdef ARCHIVE_HAS_RMD160 + unsigned char buf_rmd160[20]; +#endif +#ifdef ARCHIVE_HAS_SHA1 + unsigned char buf_sha1[20]; +#endif +#ifdef ARCHIVE_HAS_SHA256 + unsigned char buf_sha256[32]; +#endif +#ifdef ARCHIVE_HAS_SHA384 + unsigned char buf_sha384[48]; +#endif +#ifdef ARCHIVE_HAS_SHA512 + unsigned char buf_sha512[64]; +#endif +}; + +struct mtree_entry { + struct archive_rb_node rbnode; + struct mtree_entry *next; + struct mtree_entry *parent; + struct dir_info *dir_info; + struct reg_info *reg_info; + + struct archive_string parentdir; + struct archive_string basename; + struct archive_string pathname; + struct archive_string symlink; + struct archive_string uname; + struct archive_string gname; + struct archive_string fflags_text; + unsigned int nlink; + mode_t filetype; + mode_t mode; + int64_t size; + int64_t uid; + int64_t gid; + time_t mtime; + long mtime_nsec; + unsigned long fflags_set; + unsigned long fflags_clear; + dev_t rdevmajor; + dev_t rdevminor; + dev_t devmajor; + dev_t devminor; + int64_t ino; +}; + +struct mtree_writer { + struct mtree_entry *mtree_entry; + struct mtree_entry *root; + struct mtree_entry *cur_dirent; + struct archive_string cur_dirstr; + struct mtree_chain file_list; + + struct archive_string ebuf; + struct archive_string buf; + int first; + uint64_t entry_bytes_remaining; + + /* + * Set global value. + */ + struct { + int processing; + mode_t type; + int keys; + int64_t uid; + int64_t gid; + mode_t mode; + unsigned long fflags_set; + unsigned long fflags_clear; + } set; + struct att_counter_set acs; + int classic; + int depth; + + /* check sum */ + int compute_sum; + uint32_t crc; + uint64_t crc_len; +#ifdef ARCHIVE_HAS_MD5 + archive_md5_ctx md5ctx; +#endif +#ifdef ARCHIVE_HAS_RMD160 + archive_rmd160_ctx rmd160ctx; +#endif +#ifdef ARCHIVE_HAS_SHA1 + archive_sha1_ctx sha1ctx; +#endif +#ifdef ARCHIVE_HAS_SHA256 + archive_sha256_ctx sha256ctx; +#endif +#ifdef ARCHIVE_HAS_SHA384 + archive_sha384_ctx sha384ctx; +#endif +#ifdef ARCHIVE_HAS_SHA512 + archive_sha512_ctx sha512ctx; +#endif + /* Keyword options */ + int keys; +#define F_CKSUM 0x00000001 /* checksum */ +#define F_DEV 0x00000002 /* device type */ +#define F_DONE 0x00000004 /* directory done */ +#define F_FLAGS 0x00000008 /* file flags */ +#define F_GID 0x00000010 /* gid */ +#define F_GNAME 0x00000020 /* group name */ +#define F_IGN 0x00000040 /* ignore */ +#define F_MAGIC 0x00000080 /* name has magic chars */ +#define F_MD5 0x00000100 /* MD5 digest */ +#define F_MODE 0x00000200 /* mode */ +#define F_NLINK 0x00000400 /* number of links */ +#define F_NOCHANGE 0x00000800 /* If owner/mode "wrong", do + * not change */ +#define F_OPT 0x00001000 /* existence optional */ +#define F_RMD160 0x00002000 /* RIPEMD160 digest */ +#define F_SHA1 0x00004000 /* SHA-1 digest */ +#define F_SIZE 0x00008000 /* size */ +#define F_SLINK 0x00010000 /* symbolic link */ +#define F_TAGS 0x00020000 /* tags */ +#define F_TIME 0x00040000 /* modification time */ +#define F_TYPE 0x00080000 /* file type */ +#define F_UID 0x00100000 /* uid */ +#define F_UNAME 0x00200000 /* user name */ +#define F_VISIT 0x00400000 /* file visited */ +#define F_SHA256 0x00800000 /* SHA-256 digest */ +#define F_SHA384 0x01000000 /* SHA-384 digest */ +#define F_SHA512 0x02000000 /* SHA-512 digest */ +#define F_INO 0x04000000 /* inode number */ +#define F_RESDEV 0x08000000 /* device ID on which the + * entry resides */ + + /* Options */ + int dironly; /* If it is set, ignore all files except + * directory files, like mtree(8) -d option. */ + int indent; /* If it is set, indent output data. */ + int output_global_set; /* If it is set, use /set keyword to set + * global values. When generating mtree + * classic format, it is set by default. */ +}; + +#define DEFAULT_KEYS (F_DEV | F_FLAGS | F_GID | F_GNAME | F_SLINK | F_MODE\ + | F_NLINK | F_SIZE | F_TIME | F_TYPE | F_UID\ + | F_UNAME) +#define attr_counter_set_reset attr_counter_set_free + +static void attr_counter_free(struct attr_counter **); +static int attr_counter_inc(struct attr_counter **, struct attr_counter *, + struct attr_counter *, struct mtree_entry *); +static struct attr_counter * attr_counter_new(struct mtree_entry *, + struct attr_counter *); +static int attr_counter_set_collect(struct mtree_writer *, + struct mtree_entry *); +static void attr_counter_set_free(struct mtree_writer *); +static int get_global_set_keys(struct mtree_writer *, struct mtree_entry *); +static int mtree_entry_add_child_tail(struct mtree_entry *, + struct mtree_entry *); +static int mtree_entry_create_virtual_dir(struct archive_write *, const char *, + struct mtree_entry **); +static int mtree_entry_cmp_node(const struct archive_rb_node *, + const struct archive_rb_node *); +static int mtree_entry_cmp_key(const struct archive_rb_node *, const void *); +static int mtree_entry_exchange_same_entry(struct archive_write *, + struct mtree_entry *, struct mtree_entry *); +static void mtree_entry_free(struct mtree_entry *); +static int mtree_entry_new(struct archive_write *, struct archive_entry *, + struct mtree_entry **); +static void mtree_entry_register_free(struct mtree_writer *); +static void mtree_entry_register_init(struct mtree_writer *); +static int mtree_entry_setup_filenames(struct archive_write *, + struct mtree_entry *, struct archive_entry *); +static int mtree_entry_tree_add(struct archive_write *, struct mtree_entry **); +static void sum_init(struct mtree_writer *); +static void sum_update(struct mtree_writer *, const void *, size_t); +static void sum_final(struct mtree_writer *, struct reg_info *); +static void sum_write(struct archive_string *, struct reg_info *); +static int write_mtree_entry(struct archive_write *, struct mtree_entry *); +static int write_dot_dot_entry(struct archive_write *, struct mtree_entry *); + +#define COMPUTE_CRC(var, ch) (var) = (var) << 8 ^ crctab[(var) >> 24 ^ (ch)] +static const uint32_t crctab[] = { + 0x0, + 0x04c11db7, 0x09823b6e, 0x0d4326d9, 0x130476dc, 0x17c56b6b, + 0x1a864db2, 0x1e475005, 0x2608edb8, 0x22c9f00f, 0x2f8ad6d6, + 0x2b4bcb61, 0x350c9b64, 0x31cd86d3, 0x3c8ea00a, 0x384fbdbd, + 0x4c11db70, 0x48d0c6c7, 0x4593e01e, 0x4152fda9, 0x5f15adac, + 0x5bd4b01b, 0x569796c2, 0x52568b75, 0x6a1936c8, 0x6ed82b7f, + 0x639b0da6, 0x675a1011, 0x791d4014, 0x7ddc5da3, 0x709f7b7a, + 0x745e66cd, 0x9823b6e0, 0x9ce2ab57, 0x91a18d8e, 0x95609039, + 0x8b27c03c, 0x8fe6dd8b, 0x82a5fb52, 0x8664e6e5, 0xbe2b5b58, + 0xbaea46ef, 0xb7a96036, 0xb3687d81, 0xad2f2d84, 0xa9ee3033, + 0xa4ad16ea, 0xa06c0b5d, 0xd4326d90, 0xd0f37027, 0xddb056fe, + 0xd9714b49, 0xc7361b4c, 0xc3f706fb, 0xceb42022, 0xca753d95, + 0xf23a8028, 0xf6fb9d9f, 0xfbb8bb46, 0xff79a6f1, 0xe13ef6f4, + 0xe5ffeb43, 0xe8bccd9a, 0xec7dd02d, 0x34867077, 0x30476dc0, + 0x3d044b19, 0x39c556ae, 0x278206ab, 0x23431b1c, 0x2e003dc5, + 0x2ac12072, 0x128e9dcf, 0x164f8078, 0x1b0ca6a1, 0x1fcdbb16, + 0x018aeb13, 0x054bf6a4, 0x0808d07d, 0x0cc9cdca, 0x7897ab07, + 0x7c56b6b0, 0x71159069, 0x75d48dde, 0x6b93dddb, 0x6f52c06c, + 0x6211e6b5, 0x66d0fb02, 0x5e9f46bf, 0x5a5e5b08, 0x571d7dd1, + 0x53dc6066, 0x4d9b3063, 0x495a2dd4, 0x44190b0d, 0x40d816ba, + 0xaca5c697, 0xa864db20, 0xa527fdf9, 0xa1e6e04e, 0xbfa1b04b, + 0xbb60adfc, 0xb6238b25, 0xb2e29692, 0x8aad2b2f, 0x8e6c3698, + 0x832f1041, 0x87ee0df6, 0x99a95df3, 0x9d684044, 0x902b669d, + 0x94ea7b2a, 0xe0b41de7, 0xe4750050, 0xe9362689, 0xedf73b3e, + 0xf3b06b3b, 0xf771768c, 0xfa325055, 0xfef34de2, 0xc6bcf05f, + 0xc27dede8, 0xcf3ecb31, 0xcbffd686, 0xd5b88683, 0xd1799b34, + 0xdc3abded, 0xd8fba05a, 0x690ce0ee, 0x6dcdfd59, 0x608edb80, + 0x644fc637, 0x7a089632, 0x7ec98b85, 0x738aad5c, 0x774bb0eb, + 0x4f040d56, 0x4bc510e1, 0x46863638, 0x42472b8f, 0x5c007b8a, + 0x58c1663d, 0x558240e4, 0x51435d53, 0x251d3b9e, 0x21dc2629, + 0x2c9f00f0, 0x285e1d47, 0x36194d42, 0x32d850f5, 0x3f9b762c, + 0x3b5a6b9b, 0x0315d626, 0x07d4cb91, 0x0a97ed48, 0x0e56f0ff, + 0x1011a0fa, 0x14d0bd4d, 0x19939b94, 0x1d528623, 0xf12f560e, + 0xf5ee4bb9, 0xf8ad6d60, 0xfc6c70d7, 0xe22b20d2, 0xe6ea3d65, + 0xeba91bbc, 0xef68060b, 0xd727bbb6, 0xd3e6a601, 0xdea580d8, + 0xda649d6f, 0xc423cd6a, 0xc0e2d0dd, 0xcda1f604, 0xc960ebb3, + 0xbd3e8d7e, 0xb9ff90c9, 0xb4bcb610, 0xb07daba7, 0xae3afba2, + 0xaafbe615, 0xa7b8c0cc, 0xa379dd7b, 0x9b3660c6, 0x9ff77d71, + 0x92b45ba8, 0x9675461f, 0x8832161a, 0x8cf30bad, 0x81b02d74, + 0x857130c3, 0x5d8a9099, 0x594b8d2e, 0x5408abf7, 0x50c9b640, + 0x4e8ee645, 0x4a4ffbf2, 0x470cdd2b, 0x43cdc09c, 0x7b827d21, + 0x7f436096, 0x7200464f, 0x76c15bf8, 0x68860bfd, 0x6c47164a, + 0x61043093, 0x65c52d24, 0x119b4be9, 0x155a565e, 0x18197087, + 0x1cd86d30, 0x029f3d35, 0x065e2082, 0x0b1d065b, 0x0fdc1bec, + 0x3793a651, 0x3352bbe6, 0x3e119d3f, 0x3ad08088, 0x2497d08d, + 0x2056cd3a, 0x2d15ebe3, 0x29d4f654, 0xc5a92679, 0xc1683bce, + 0xcc2b1d17, 0xc8ea00a0, 0xd6ad50a5, 0xd26c4d12, 0xdf2f6bcb, + 0xdbee767c, 0xe3a1cbc1, 0xe760d676, 0xea23f0af, 0xeee2ed18, + 0xf0a5bd1d, 0xf464a0aa, 0xf9278673, 0xfde69bc4, 0x89b8fd09, + 0x8d79e0be, 0x803ac667, 0x84fbdbd0, 0x9abc8bd5, 0x9e7d9662, + 0x933eb0bb, 0x97ffad0c, 0xafb010b1, 0xab710d06, 0xa6322bdf, + 0xa2f33668, 0xbcb4666d, 0xb8757bda, 0xb5365d03, 0xb1f740b4 +}; + +static const unsigned char safe_char[256] = { + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 00 - 0F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 10 - 1F */ + /* !"$%&'()*+,-./ EXCLUSION:0x20( ) 0x23(#) */ + 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 20 - 2F */ + /* 0123456789:;<>? EXCLUSION:0x3d(=) */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, /* 30 - 3F */ + /* @ABCDEFGHIJKLMNO */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 40 - 4F */ + /* PQRSTUVWXYZ[]^_ EXCLUSION:0x5c(\) */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, /* 50 - 5F */ + /* `abcdefghijklmno */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* 60 - 6F */ + /* pqrstuvwxyz{|}~ */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, /* 70 - 7F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 80 - 8F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 90 - 9F */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* A0 - AF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* B0 - BF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* C0 - CF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* D0 - DF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* E0 - EF */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* F0 - FF */ +}; + +static void +mtree_quote(struct archive_string *s, const char *str) +{ + const char *start; + char buf[4]; + unsigned char c; + + for (start = str; *str != '\0'; ++str) { + if (safe_char[*(const unsigned char *)str]) + continue; + if (start != str) + archive_strncat(s, start, str - start); + c = (unsigned char)*str; + buf[0] = '\\'; + buf[1] = (c / 64) + '0'; + buf[2] = (c / 8 % 8) + '0'; + buf[3] = (c % 8) + '0'; + archive_strncat(s, buf, 4); + start = str + 1; + } + + if (start != str) + archive_strncat(s, start, str - start); +} + +/* + * Indent a line as the mtree utility does so it is readable for people. + */ +static void +mtree_indent(struct mtree_writer *mtree) +{ + int i, fn, nd, pd; + const char *r, *s, *x; + + if (mtree->classic) { + if (mtree->indent) { + nd = 0; + pd = mtree->depth * 4; + } else { + nd = mtree->depth?4:0; + pd = 0; + } + } else + nd = pd = 0; + fn = 1; + s = r = mtree->ebuf.s; + x = NULL; + while (*r == ' ') + r++; + while ((r = strchr(r, ' ')) != NULL) { + if (fn) { + fn = 0; + for (i = 0; i < nd + pd; i++) + archive_strappend_char(&mtree->buf, ' '); + archive_strncat(&mtree->buf, s, r - s); + if (nd + (r -s) > INDENTNAMELEN) { + archive_strncat(&mtree->buf, " \\\n", 3); + for (i = 0; i < (INDENTNAMELEN + 1 + pd); i++) + archive_strappend_char(&mtree->buf, ' '); + } else { + for (i = (int)(r -s + nd); + i < (INDENTNAMELEN + 1); i++) + archive_strappend_char(&mtree->buf, ' '); + } + s = ++r; + x = NULL; + continue; + } + if (pd + (r - s) <= MAXLINELEN - 3 - INDENTNAMELEN) + x = r++; + else { + if (x == NULL) + x = r; + archive_strncat(&mtree->buf, s, x - s); + archive_strncat(&mtree->buf, " \\\n", 3); + for (i = 0; i < (INDENTNAMELEN + 1 + pd); i++) + archive_strappend_char(&mtree->buf, ' '); + s = r = ++x; + x = NULL; + } + } + if (fn) { + for (i = 0; i < nd + pd; i++) + archive_strappend_char(&mtree->buf, ' '); + archive_strcat(&mtree->buf, s); + s += strlen(s); + } + if (x != NULL && pd + strlen(s) > MAXLINELEN - 3 - INDENTNAMELEN) { + /* Last keyword is longer. */ + archive_strncat(&mtree->buf, s, x - s); + archive_strncat(&mtree->buf, " \\\n", 3); + for (i = 0; i < (INDENTNAMELEN + 1 + pd); i++) + archive_strappend_char(&mtree->buf, ' '); + s = ++x; + } + archive_strcat(&mtree->buf, s); + archive_string_empty(&mtree->ebuf); +} + +/* + * Write /set keyword. + * Set the most used value of uid, gid, mode and fflags, which are + * collected by the attr_counter_set_collect() function. + */ +static void +write_global(struct mtree_writer *mtree) +{ + struct archive_string setstr; + struct archive_string unsetstr; + struct att_counter_set *acs; + int keys, oldkeys, effkeys; + + archive_string_init(&setstr); + archive_string_init(&unsetstr); + keys = mtree->keys & SET_KEYS; + oldkeys = mtree->set.keys; + effkeys = keys; + acs = &mtree->acs; + if (mtree->set.processing) { + /* + * Check if the global data needs updating. + */ + effkeys &= ~F_TYPE; + if (acs->uid_list == NULL) + effkeys &= ~(F_UNAME | F_UID); + else if (oldkeys & (F_UNAME | F_UID)) { + if (acs->uid_list->count < 2 || + mtree->set.uid == acs->uid_list->m_entry->uid) + effkeys &= ~(F_UNAME | F_UID); + } + if (acs->gid_list == NULL) + effkeys &= ~(F_GNAME | F_GID); + else if (oldkeys & (F_GNAME | F_GID)) { + if (acs->gid_list->count < 2 || + mtree->set.gid == acs->gid_list->m_entry->gid) + effkeys &= ~(F_GNAME | F_GID); + } + if (acs->mode_list == NULL) + effkeys &= ~F_MODE; + else if (oldkeys & F_MODE) { + if (acs->mode_list->count < 2 || + mtree->set.mode == acs->mode_list->m_entry->mode) + effkeys &= ~F_MODE; + } + if (acs->flags_list == NULL) + effkeys &= ~F_FLAGS; + else if ((oldkeys & F_FLAGS) != 0) { + if (acs->flags_list->count < 2 || + (acs->flags_list->m_entry->fflags_set == + mtree->set.fflags_set && + acs->flags_list->m_entry->fflags_clear == + mtree->set.fflags_clear)) + effkeys &= ~F_FLAGS; + } + } else { + if (acs->uid_list == NULL) + keys &= ~(F_UNAME | F_UID); + if (acs->gid_list == NULL) + keys &= ~(F_GNAME | F_GID); + if (acs->mode_list == NULL) + keys &= ~F_MODE; + if (acs->flags_list == NULL) + keys &= ~F_FLAGS; + } + if ((keys & effkeys & F_TYPE) != 0) { + if (mtree->dironly) { + archive_strcat(&setstr, " type=dir"); + mtree->set.type = AE_IFDIR; + } else { + archive_strcat(&setstr, " type=file"); + mtree->set.type = AE_IFREG; + } + } + if ((keys & effkeys & F_UNAME) != 0) { + if (archive_strlen(&(acs->uid_list->m_entry->uname)) > 0) { + archive_strcat(&setstr, " uname="); + mtree_quote(&setstr, acs->uid_list->m_entry->uname.s); + } else { + keys &= ~F_UNAME; + if ((oldkeys & F_UNAME) != 0) + archive_strcat(&unsetstr, " uname"); + } + } + if ((keys & effkeys & F_UID) != 0) { + mtree->set.uid = acs->uid_list->m_entry->uid; + archive_string_sprintf(&setstr, " uid=%jd", + (intmax_t)mtree->set.uid); + } + if ((keys & effkeys & F_GNAME) != 0) { + if (archive_strlen(&(acs->gid_list->m_entry->gname)) > 0) { + archive_strcat(&setstr, " gname="); + mtree_quote(&setstr, acs->gid_list->m_entry->gname.s); + } else { + keys &= ~F_GNAME; + if ((oldkeys & F_GNAME) != 0) + archive_strcat(&unsetstr, " gname"); + } + } + if ((keys & effkeys & F_GID) != 0) { + mtree->set.gid = acs->gid_list->m_entry->gid; + archive_string_sprintf(&setstr, " gid=%jd", + (intmax_t)mtree->set.gid); + } + if ((keys & effkeys & F_MODE) != 0) { + mtree->set.mode = acs->mode_list->m_entry->mode; + archive_string_sprintf(&setstr, " mode=%o", + (unsigned int)mtree->set.mode); + } + if ((keys & effkeys & F_FLAGS) != 0) { + if (archive_strlen( + &(acs->flags_list->m_entry->fflags_text)) > 0) { + archive_strcat(&setstr, " flags="); + mtree_quote(&setstr, + acs->flags_list->m_entry->fflags_text.s); + mtree->set.fflags_set = + acs->flags_list->m_entry->fflags_set; + mtree->set.fflags_clear = + acs->flags_list->m_entry->fflags_clear; + } else { + keys &= ~F_FLAGS; + if ((oldkeys & F_FLAGS) != 0) + archive_strcat(&unsetstr, " flags"); + } + } + if (unsetstr.length > 0) + archive_string_sprintf(&mtree->buf, "/unset%s\n", unsetstr.s); + archive_string_free(&unsetstr); + if (setstr.length > 0) + archive_string_sprintf(&mtree->buf, "/set%s\n", setstr.s); + archive_string_free(&setstr); + mtree->set.keys = keys; + mtree->set.processing = 1; +} + +static struct attr_counter * +attr_counter_new(struct mtree_entry *me, struct attr_counter *prev) +{ + struct attr_counter *ac; + + ac = malloc(sizeof(*ac)); + if (ac != NULL) { + ac->prev = prev; + ac->next = NULL; + ac->count = 1; + ac->m_entry = me; + } + return (ac); +} + +static void +attr_counter_free(struct attr_counter **top) +{ + struct attr_counter *ac, *tac; + + if (*top == NULL) + return; + ac = *top; + while (ac != NULL) { + tac = ac->next; + free(ac); + ac = tac; + } + *top = NULL; +} + +static int +attr_counter_inc(struct attr_counter **top, struct attr_counter *ac, + struct attr_counter *last, struct mtree_entry *me) +{ + struct attr_counter *pac; + + if (ac != NULL) { + ac->count++; + if (*top == ac || ac->prev->count >= ac->count) + return (0); + for (pac = ac->prev; pac; pac = pac->prev) { + if (pac->count >= ac->count) + break; + } + ac->prev->next = ac->next; + if (ac->next != NULL) + ac->next->prev = ac->prev; + if (pac != NULL) { + ac->prev = pac; + ac->next = pac->next; + pac->next = ac; + if (ac->next != NULL) + ac->next->prev = ac; + } else { + ac->prev = NULL; + ac->next = *top; + *top = ac; + ac->next->prev = ac; + } + } else if (last != NULL) { + ac = attr_counter_new(me, last); + if (ac == NULL) + return (-1); + last->next = ac; + } + return (0); +} + +/* + * Tabulate uid, gid, mode and fflags of a entry in order to be used for /set. + */ +static int +attr_counter_set_collect(struct mtree_writer *mtree, struct mtree_entry *me) +{ + struct attr_counter *ac, *last; + struct att_counter_set *acs = &mtree->acs; + int keys = mtree->keys; + + if (keys & (F_UNAME | F_UID)) { + if (acs->uid_list == NULL) { + acs->uid_list = attr_counter_new(me, NULL); + if (acs->uid_list == NULL) + return (-1); + } else { + last = NULL; + for (ac = acs->uid_list; ac; ac = ac->next) { + if (ac->m_entry->uid == me->uid) + break; + last = ac; + } + if (attr_counter_inc(&acs->uid_list, ac, last, me) < 0) + return (-1); + } + } + if (keys & (F_GNAME | F_GID)) { + if (acs->gid_list == NULL) { + acs->gid_list = attr_counter_new(me, NULL); + if (acs->gid_list == NULL) + return (-1); + } else { + last = NULL; + for (ac = acs->gid_list; ac; ac = ac->next) { + if (ac->m_entry->gid == me->gid) + break; + last = ac; + } + if (attr_counter_inc(&acs->gid_list, ac, last, me) < 0) + return (-1); + } + } + if (keys & F_MODE) { + if (acs->mode_list == NULL) { + acs->mode_list = attr_counter_new(me, NULL); + if (acs->mode_list == NULL) + return (-1); + } else { + last = NULL; + for (ac = acs->mode_list; ac; ac = ac->next) { + if (ac->m_entry->mode == me->mode) + break; + last = ac; + } + if (attr_counter_inc(&acs->mode_list, ac, last, me) < 0) + return (-1); + } + } + if (keys & F_FLAGS) { + if (acs->flags_list == NULL) { + acs->flags_list = attr_counter_new(me, NULL); + if (acs->flags_list == NULL) + return (-1); + } else { + last = NULL; + for (ac = acs->flags_list; ac; ac = ac->next) { + if (ac->m_entry->fflags_set == me->fflags_set && + ac->m_entry->fflags_clear == + me->fflags_clear) + break; + last = ac; + } + if (attr_counter_inc(&acs->flags_list, ac, last, me) < 0) + return (-1); + } + } + + return (0); +} + +static void +attr_counter_set_free(struct mtree_writer *mtree) +{ + struct att_counter_set *acs = &mtree->acs; + + attr_counter_free(&acs->uid_list); + attr_counter_free(&acs->gid_list); + attr_counter_free(&acs->mode_list); + attr_counter_free(&acs->flags_list); +} + +static int +get_global_set_keys(struct mtree_writer *mtree, struct mtree_entry *me) +{ + int keys; + + keys = mtree->keys; + + /* + * If a keyword has been set by /set, we do not need to + * output it. + */ + if (mtree->set.keys == 0) + return (keys);/* /set is not used. */ + + if ((mtree->set.keys & (F_GNAME | F_GID)) != 0 && + mtree->set.gid == me->gid) + keys &= ~(F_GNAME | F_GID); + if ((mtree->set.keys & (F_UNAME | F_UID)) != 0 && + mtree->set.uid == me->uid) + keys &= ~(F_UNAME | F_UID); + if (mtree->set.keys & F_FLAGS) { + if (mtree->set.fflags_set == me->fflags_set && + mtree->set.fflags_clear == me->fflags_clear) + keys &= ~F_FLAGS; + } + if ((mtree->set.keys & F_MODE) != 0 && mtree->set.mode == me->mode) + keys &= ~F_MODE; + + switch (me->filetype) { + case AE_IFLNK: case AE_IFSOCK: case AE_IFCHR: + case AE_IFBLK: case AE_IFIFO: + break; + case AE_IFDIR: + if ((mtree->set.keys & F_TYPE) != 0 && + mtree->set.type == AE_IFDIR) + keys &= ~F_TYPE; + break; + case AE_IFREG: + default: /* Handle unknown file types as regular files. */ + if ((mtree->set.keys & F_TYPE) != 0 && + mtree->set.type == AE_IFREG) + keys &= ~F_TYPE; + break; + } + + return (keys); +} + +static int +mtree_entry_new(struct archive_write *a, struct archive_entry *entry, + struct mtree_entry **m_entry) +{ + struct mtree_entry *me; + const char *s; + int r; + static const struct archive_rb_tree_ops rb_ops = { + mtree_entry_cmp_node, mtree_entry_cmp_key + }; + + me = calloc(1, sizeof(*me)); + if (me == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for a mtree entry"); + *m_entry = NULL; + return (ARCHIVE_FATAL); + } + + r = mtree_entry_setup_filenames(a, me, entry); + if (r < ARCHIVE_WARN) { + mtree_entry_free(me); + *m_entry = NULL; + return (r); + } + + if ((s = archive_entry_symlink(entry)) != NULL) + archive_strcpy(&me->symlink, s); + me->nlink = archive_entry_nlink(entry); + me->filetype = archive_entry_filetype(entry); + me->mode = archive_entry_mode(entry) & 07777; + me->uid = archive_entry_uid(entry); + me->gid = archive_entry_gid(entry); + if ((s = archive_entry_uname(entry)) != NULL) + archive_strcpy(&me->uname, s); + if ((s = archive_entry_gname(entry)) != NULL) + archive_strcpy(&me->gname, s); + if ((s = archive_entry_fflags_text(entry)) != NULL) + archive_strcpy(&me->fflags_text, s); + archive_entry_fflags(entry, &me->fflags_set, &me->fflags_clear); + me->mtime = archive_entry_mtime(entry); + me->mtime_nsec = archive_entry_mtime_nsec(entry); + me->rdevmajor = archive_entry_rdevmajor(entry); + me->rdevminor = archive_entry_rdevminor(entry); + me->devmajor = archive_entry_devmajor(entry); + me->devminor = archive_entry_devminor(entry); + me->ino = archive_entry_ino(entry); + me->size = archive_entry_size(entry); + if (me->filetype == AE_IFDIR) { + me->dir_info = calloc(1, sizeof(*me->dir_info)); + if (me->dir_info == NULL) { + mtree_entry_free(me); + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for a mtree entry"); + *m_entry = NULL; + return (ARCHIVE_FATAL); + } + __archive_rb_tree_init(&me->dir_info->rbtree, &rb_ops); + me->dir_info->children.first = NULL; + me->dir_info->children.last = &(me->dir_info->children.first); + me->dir_info->chnext = NULL; + } else if (me->filetype == AE_IFREG) { + me->reg_info = calloc(1, sizeof(*me->reg_info)); + if (me->reg_info == NULL) { + mtree_entry_free(me); + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for a mtree entry"); + *m_entry = NULL; + return (ARCHIVE_FATAL); + } + me->reg_info->compute_sum = 0; + } + + *m_entry = me; + return (ARCHIVE_OK); +} + +static void +mtree_entry_free(struct mtree_entry *me) +{ + archive_string_free(&me->parentdir); + archive_string_free(&me->basename); + archive_string_free(&me->pathname); + archive_string_free(&me->symlink); + archive_string_free(&me->uname); + archive_string_free(&me->gname); + archive_string_free(&me->fflags_text); + free(me->dir_info); + free(me->reg_info); + free(me); +} + +static int +archive_write_mtree_header(struct archive_write *a, + struct archive_entry *entry) +{ + struct mtree_writer *mtree= a->format_data; + struct mtree_entry *mtree_entry; + int r, r2; + + if (mtree->first) { + mtree->first = 0; + archive_strcat(&mtree->buf, "#mtree\n"); + if ((mtree->keys & SET_KEYS) == 0) + mtree->output_global_set = 0;/* Disabled. */ + } + + mtree->entry_bytes_remaining = archive_entry_size(entry); + + /* While directory only mode, we do not handle non directory files. */ + if (mtree->dironly && archive_entry_filetype(entry) != AE_IFDIR) + return (ARCHIVE_OK); + + r2 = mtree_entry_new(a, entry, &mtree_entry); + if (r2 < ARCHIVE_WARN) + return (r2); + r = mtree_entry_tree_add(a, &mtree_entry); + if (r < ARCHIVE_WARN) { + mtree_entry_free(mtree_entry); + return (r); + } + mtree->mtree_entry = mtree_entry; + + /* If the current file is a regular file, we have to + * compute the sum of its content. + * Initialize a bunch of checksum context. */ + if (mtree_entry->reg_info) + sum_init(mtree); + + return (r2); +} + +static int +write_mtree_entry(struct archive_write *a, struct mtree_entry *me) +{ + struct mtree_writer *mtree = a->format_data; + struct archive_string *str; + int keys, ret; + + if (me->dir_info) { + if (mtree->classic) { + /* + * Output a comment line to describe the full + * pathname of the entry as mtree utility does + * while generating classic format. + */ + if (!mtree->dironly) + archive_strappend_char(&mtree->buf, '\n'); + if (me->parentdir.s) + archive_string_sprintf(&mtree->buf, + "# %s/%s\n", + me->parentdir.s, me->basename.s); + else + archive_string_sprintf(&mtree->buf, + "# %s\n", + me->basename.s); + } + if (mtree->output_global_set) + write_global(mtree); + } + archive_string_empty(&mtree->ebuf); + str = (mtree->indent || mtree->classic)? &mtree->ebuf : &mtree->buf; + + if (!mtree->classic && me->parentdir.s) { + /* + * If generating format is not classic one(v1), output + * a full pathname. + */ + mtree_quote(str, me->parentdir.s); + archive_strappend_char(str, '/'); + } + mtree_quote(str, me->basename.s); + + keys = get_global_set_keys(mtree, me); + if ((keys & F_NLINK) != 0 && + me->nlink != 1 && me->filetype != AE_IFDIR) + archive_string_sprintf(str, " nlink=%u", me->nlink); + + if ((keys & F_GNAME) != 0 && archive_strlen(&me->gname) > 0) { + archive_strcat(str, " gname="); + mtree_quote(str, me->gname.s); + } + if ((keys & F_UNAME) != 0 && archive_strlen(&me->uname) > 0) { + archive_strcat(str, " uname="); + mtree_quote(str, me->uname.s); + } + if ((keys & F_FLAGS) != 0) { + if (archive_strlen(&me->fflags_text) > 0) { + archive_strcat(str, " flags="); + mtree_quote(str, me->fflags_text.s); + } else if (mtree->set.processing && + (mtree->set.keys & F_FLAGS) != 0) + /* Overwrite the global parameter. */ + archive_strcat(str, " flags=none"); + } + if ((keys & F_TIME) != 0) + archive_string_sprintf(str, " time=%jd.%jd", + (intmax_t)me->mtime, (intmax_t)me->mtime_nsec); + if ((keys & F_MODE) != 0) + archive_string_sprintf(str, " mode=%o", (unsigned int)me->mode); + if ((keys & F_GID) != 0) + archive_string_sprintf(str, " gid=%jd", (intmax_t)me->gid); + if ((keys & F_UID) != 0) + archive_string_sprintf(str, " uid=%jd", (intmax_t)me->uid); + + if ((keys & F_INO) != 0) + archive_string_sprintf(str, " inode=%jd", (intmax_t)me->ino); + if ((keys & F_RESDEV) != 0) { + archive_string_sprintf(str, + " resdevice=native,%ju,%ju", + (uintmax_t)me->devmajor, + (uintmax_t)me->devminor); + } + + switch (me->filetype) { + case AE_IFLNK: + if ((keys & F_TYPE) != 0) + archive_strcat(str, " type=link"); + if ((keys & F_SLINK) != 0) { + archive_strcat(str, " link="); + mtree_quote(str, me->symlink.s); + } + break; + case AE_IFSOCK: + if ((keys & F_TYPE) != 0) + archive_strcat(str, " type=socket"); + break; + case AE_IFCHR: + if ((keys & F_TYPE) != 0) + archive_strcat(str, " type=char"); + if ((keys & F_DEV) != 0) { + archive_string_sprintf(str, + " device=native,%ju,%ju", + (uintmax_t)me->rdevmajor, + (uintmax_t)me->rdevminor); + } + break; + case AE_IFBLK: + if ((keys & F_TYPE) != 0) + archive_strcat(str, " type=block"); + if ((keys & F_DEV) != 0) { + archive_string_sprintf(str, + " device=native,%ju,%ju", + (uintmax_t)me->rdevmajor, + (uintmax_t)me->rdevminor); + } + break; + case AE_IFDIR: + if ((keys & F_TYPE) != 0) + archive_strcat(str, " type=dir"); + break; + case AE_IFIFO: + if ((keys & F_TYPE) != 0) + archive_strcat(str, " type=fifo"); + break; + case AE_IFREG: + default: /* Handle unknown file types as regular files. */ + if ((keys & F_TYPE) != 0) + archive_strcat(str, " type=file"); + if ((keys & F_SIZE) != 0) + archive_string_sprintf(str, " size=%jd", + (intmax_t)me->size); + break; + } + + /* Write a bunch of sum. */ + if (me->reg_info) + sum_write(str, me->reg_info); + + archive_strappend_char(str, '\n'); + if (mtree->indent || mtree->classic) + mtree_indent(mtree); + + if (mtree->buf.length > 32768) { + ret = __archive_write_output( + a, mtree->buf.s, mtree->buf.length); + archive_string_empty(&mtree->buf); + } else + ret = ARCHIVE_OK; + return (ret); +} + +static int +write_dot_dot_entry(struct archive_write *a, struct mtree_entry *n) +{ + struct mtree_writer *mtree = a->format_data; + int ret; + + if (n->parentdir.s) { + if (mtree->indent) { + int i, pd = mtree->depth * 4; + for (i = 0; i < pd; i++) + archive_strappend_char(&mtree->buf, ' '); + } + archive_string_sprintf(&mtree->buf, "# %s/%s\n", + n->parentdir.s, n->basename.s); + } + + if (mtree->indent) { + archive_string_empty(&mtree->ebuf); + archive_strncat(&mtree->ebuf, "..\n\n", (mtree->dironly)?3:4); + mtree_indent(mtree); + } else + archive_strncat(&mtree->buf, "..\n\n", (mtree->dironly)?3:4); + + if (mtree->buf.length > 32768) { + ret = __archive_write_output( + a, mtree->buf.s, mtree->buf.length); + archive_string_empty(&mtree->buf); + } else + ret = ARCHIVE_OK; + return (ret); +} + +/* + * Write mtree entries saved at attr_counter_set_collect() function. + */ +static int +write_mtree_entry_tree(struct archive_write *a) +{ + struct mtree_writer *mtree = a->format_data; + struct mtree_entry *np = mtree->root; + struct archive_rb_node *n; + int ret; + + do { + if (mtree->output_global_set) { + /* + * Collect attribute information to know which value + * is frequently used among the children. + */ + attr_counter_set_reset(mtree); + ARCHIVE_RB_TREE_FOREACH(n, &(np->dir_info->rbtree)) { + struct mtree_entry *e = (struct mtree_entry *)n; + if (attr_counter_set_collect(mtree, e) < 0) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + } + } + if (!np->dir_info->virtual || mtree->classic) { + ret = write_mtree_entry(a, np); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } else { + /* Whenever output_global_set is enabled + * output global value(/set keywords) + * even if the directory entry is not allowed + * to be written because the global values + * can be used for the children. */ + if (mtree->output_global_set) + write_global(mtree); + } + /* + * Output the attribute of all files except directory files. + */ + mtree->depth++; + ARCHIVE_RB_TREE_FOREACH(n, &(np->dir_info->rbtree)) { + struct mtree_entry *e = (struct mtree_entry *)n; + + if (e->dir_info) + mtree_entry_add_child_tail(np, e); + else { + ret = write_mtree_entry(a, e); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + } + mtree->depth--; + + if (np->dir_info->children.first != NULL) { + /* + * Descend the tree. + */ + np = np->dir_info->children.first; + if (mtree->indent) + mtree->depth++; + continue; + } else if (mtree->classic) { + /* + * While printing mtree classic, if there are not + * any directory files(except "." and "..") in the + * directory, output two dots ".." as returning + * the parent directory. + */ + ret = write_dot_dot_entry(a, np); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + + while (np != np->parent) { + if (np->dir_info->chnext == NULL) { + /* + * Ascend the tree; go back to the parent. + */ + if (mtree->indent) + mtree->depth--; + if (mtree->classic) { + ret = write_dot_dot_entry(a, + np->parent); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + np = np->parent; + } else { + /* + * Switch to next mtree entry in the directory. + */ + np = np->dir_info->chnext; + break; + } + } + } while (np != np->parent); + + return (ARCHIVE_OK); +} + +static int +archive_write_mtree_finish_entry(struct archive_write *a) +{ + struct mtree_writer *mtree = a->format_data; + struct mtree_entry *me; + + if ((me = mtree->mtree_entry) == NULL) + return (ARCHIVE_OK); + mtree->mtree_entry = NULL; + + if (me->reg_info) + sum_final(mtree, me->reg_info); + + return (ARCHIVE_OK); +} + +static int +archive_write_mtree_close(struct archive_write *a) +{ + struct mtree_writer *mtree= a->format_data; + int ret; + + if (mtree->root != NULL) { + ret = write_mtree_entry_tree(a); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + + archive_write_set_bytes_in_last_block(&a->archive, 1); + + return __archive_write_output(a, mtree->buf.s, mtree->buf.length); +} + +static ssize_t +archive_write_mtree_data(struct archive_write *a, const void *buff, size_t n) +{ + struct mtree_writer *mtree= a->format_data; + + if (n > mtree->entry_bytes_remaining) + n = (size_t)mtree->entry_bytes_remaining; + mtree->entry_bytes_remaining -= n; + + /* We don't need to compute a regular file sum */ + if (mtree->mtree_entry == NULL) + return (n); + + if (mtree->mtree_entry->filetype == AE_IFREG) + sum_update(mtree, buff, n); + + return (n); +} + +static int +archive_write_mtree_free(struct archive_write *a) +{ + struct mtree_writer *mtree= a->format_data; + + if (mtree == NULL) + return (ARCHIVE_OK); + + /* Make sure we do not leave any entries. */ + mtree_entry_register_free(mtree); + archive_string_free(&mtree->cur_dirstr); + archive_string_free(&mtree->ebuf); + archive_string_free(&mtree->buf); + attr_counter_set_free(mtree); + free(mtree); + a->format_data = NULL; + return (ARCHIVE_OK); +} + +static int +archive_write_mtree_options(struct archive_write *a, const char *key, + const char *value) +{ + struct mtree_writer *mtree= a->format_data; + int keybit = 0; + + switch (key[0]) { + case 'a': + if (strcmp(key, "all") == 0) + keybit = ~0; + break; + case 'c': + if (strcmp(key, "cksum") == 0) + keybit = F_CKSUM; + break; + case 'd': + if (strcmp(key, "device") == 0) + keybit = F_DEV; + else if (strcmp(key, "dironly") == 0) { + mtree->dironly = (value != NULL)? 1: 0; + return (ARCHIVE_OK); + } + break; + case 'f': + if (strcmp(key, "flags") == 0) + keybit = F_FLAGS; + break; + case 'g': + if (strcmp(key, "gid") == 0) + keybit = F_GID; + else if (strcmp(key, "gname") == 0) + keybit = F_GNAME; + break; + case 'i': + if (strcmp(key, "indent") == 0) { + mtree->indent = (value != NULL)? 1: 0; + return (ARCHIVE_OK); + } else if (strcmp(key, "inode") == 0) { + keybit = F_INO; + } + break; + case 'l': + if (strcmp(key, "link") == 0) + keybit = F_SLINK; + break; + case 'm': + if (strcmp(key, "md5") == 0 || + strcmp(key, "md5digest") == 0) + keybit = F_MD5; + if (strcmp(key, "mode") == 0) + keybit = F_MODE; + break; + case 'n': + if (strcmp(key, "nlink") == 0) + keybit = F_NLINK; + break; + case 'r': + if (strcmp(key, "resdevice") == 0) { + keybit = F_RESDEV; + } else if (strcmp(key, "ripemd160digest") == 0 || + strcmp(key, "rmd160") == 0 || + strcmp(key, "rmd160digest") == 0) + keybit = F_RMD160; + break; + case 's': + if (strcmp(key, "sha1") == 0 || + strcmp(key, "sha1digest") == 0) + keybit = F_SHA1; + if (strcmp(key, "sha256") == 0 || + strcmp(key, "sha256digest") == 0) + keybit = F_SHA256; + if (strcmp(key, "sha384") == 0 || + strcmp(key, "sha384digest") == 0) + keybit = F_SHA384; + if (strcmp(key, "sha512") == 0 || + strcmp(key, "sha512digest") == 0) + keybit = F_SHA512; + if (strcmp(key, "size") == 0) + keybit = F_SIZE; + break; + case 't': + if (strcmp(key, "time") == 0) + keybit = F_TIME; + else if (strcmp(key, "type") == 0) + keybit = F_TYPE; + break; + case 'u': + if (strcmp(key, "uid") == 0) + keybit = F_UID; + else if (strcmp(key, "uname") == 0) + keybit = F_UNAME; + else if (strcmp(key, "use-set") == 0) { + mtree->output_global_set = (value != NULL)? 1: 0; + return (ARCHIVE_OK); + } + break; + } + if (keybit != 0) { + if (value != NULL) + mtree->keys |= keybit; + else + mtree->keys &= ~keybit; + return (ARCHIVE_OK); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +static int +archive_write_set_format_mtree_default(struct archive *_a, const char *fn) +{ + struct archive_write *a = (struct archive_write *)_a; + struct mtree_writer *mtree; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, fn); + + if (a->format_free != NULL) + (a->format_free)(a); + + if ((mtree = calloc(1, sizeof(*mtree))) == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate mtree data"); + return (ARCHIVE_FATAL); + } + + mtree->mtree_entry = NULL; + mtree->first = 1; + memset(&(mtree->set), 0, sizeof(mtree->set)); + mtree->keys = DEFAULT_KEYS; + mtree->dironly = 0; + mtree->indent = 0; + archive_string_init(&mtree->ebuf); + archive_string_init(&mtree->buf); + mtree_entry_register_init(mtree); + a->format_data = mtree; + a->format_free = archive_write_mtree_free; + a->format_name = "mtree"; + a->format_options = archive_write_mtree_options; + a->format_write_header = archive_write_mtree_header; + a->format_close = archive_write_mtree_close; + a->format_write_data = archive_write_mtree_data; + a->format_finish_entry = archive_write_mtree_finish_entry; + a->archive.archive_format = ARCHIVE_FORMAT_MTREE; + a->archive.archive_format_name = "mtree"; + + return (ARCHIVE_OK); +} + +int +archive_write_set_format_mtree(struct archive *_a) +{ + return archive_write_set_format_mtree_default(_a, + "archive_write_set_format_mtree"); +} + +int +archive_write_set_format_mtree_classic(struct archive *_a) +{ + int r; + + r = archive_write_set_format_mtree_default(_a, + "archive_write_set_format_mtree_classic"); + if (r == ARCHIVE_OK) { + struct archive_write *a = (struct archive_write *)_a; + struct mtree_writer *mtree; + + mtree = (struct mtree_writer *)a->format_data; + + /* Set to output a mtree archive in classic format. */ + mtree->classic = 1; + /* Basically, mtree classic format uses '/set' global + * value. */ + mtree->output_global_set = 1; + } + return (r); +} + +static void +sum_init(struct mtree_writer *mtree) +{ + + mtree->compute_sum = 0; + + if (mtree->keys & F_CKSUM) { + mtree->compute_sum |= F_CKSUM; + mtree->crc = 0; + mtree->crc_len = 0; + } +#ifdef ARCHIVE_HAS_MD5 + if (mtree->keys & F_MD5) { + if (archive_md5_init(&mtree->md5ctx) == ARCHIVE_OK) + mtree->compute_sum |= F_MD5; + else + mtree->keys &= ~F_MD5;/* Not supported. */ + } +#endif +#ifdef ARCHIVE_HAS_RMD160 + if (mtree->keys & F_RMD160) { + if (archive_rmd160_init(&mtree->rmd160ctx) == ARCHIVE_OK) + mtree->compute_sum |= F_RMD160; + else + mtree->keys &= ~F_RMD160;/* Not supported. */ + } +#endif +#ifdef ARCHIVE_HAS_SHA1 + if (mtree->keys & F_SHA1) { + if (archive_sha1_init(&mtree->sha1ctx) == ARCHIVE_OK) + mtree->compute_sum |= F_SHA1; + else + mtree->keys &= ~F_SHA1;/* Not supported. */ + } +#endif +#ifdef ARCHIVE_HAS_SHA256 + if (mtree->keys & F_SHA256) { + if (archive_sha256_init(&mtree->sha256ctx) == ARCHIVE_OK) + mtree->compute_sum |= F_SHA256; + else + mtree->keys &= ~F_SHA256;/* Not supported. */ + } +#endif +#ifdef ARCHIVE_HAS_SHA384 + if (mtree->keys & F_SHA384) { + if (archive_sha384_init(&mtree->sha384ctx) == ARCHIVE_OK) + mtree->compute_sum |= F_SHA384; + else + mtree->keys &= ~F_SHA384;/* Not supported. */ + } +#endif +#ifdef ARCHIVE_HAS_SHA512 + if (mtree->keys & F_SHA512) { + if (archive_sha512_init(&mtree->sha512ctx) == ARCHIVE_OK) + mtree->compute_sum |= F_SHA512; + else + mtree->keys &= ~F_SHA512;/* Not supported. */ + } +#endif +} + +static void +sum_update(struct mtree_writer *mtree, const void *buff, size_t n) +{ + if (mtree->compute_sum & F_CKSUM) { + /* + * Compute a POSIX 1003.2 checksum + */ + const unsigned char *p; + size_t nn; + + for (nn = n, p = buff; nn--; ++p) + COMPUTE_CRC(mtree->crc, *p); + mtree->crc_len += n; + } +#ifdef ARCHIVE_HAS_MD5 + if (mtree->compute_sum & F_MD5) + archive_md5_update(&mtree->md5ctx, buff, n); +#endif +#ifdef ARCHIVE_HAS_RMD160 + if (mtree->compute_sum & F_RMD160) + archive_rmd160_update(&mtree->rmd160ctx, buff, n); +#endif +#ifdef ARCHIVE_HAS_SHA1 + if (mtree->compute_sum & F_SHA1) + archive_sha1_update(&mtree->sha1ctx, buff, n); +#endif +#ifdef ARCHIVE_HAS_SHA256 + if (mtree->compute_sum & F_SHA256) + archive_sha256_update(&mtree->sha256ctx, buff, n); +#endif +#ifdef ARCHIVE_HAS_SHA384 + if (mtree->compute_sum & F_SHA384) + archive_sha384_update(&mtree->sha384ctx, buff, n); +#endif +#ifdef ARCHIVE_HAS_SHA512 + if (mtree->compute_sum & F_SHA512) + archive_sha512_update(&mtree->sha512ctx, buff, n); +#endif +} + +static void +sum_final(struct mtree_writer *mtree, struct reg_info *reg) +{ + + if (mtree->compute_sum & F_CKSUM) { + uint64_t len; + /* Include the length of the file. */ + for (len = mtree->crc_len; len != 0; len >>= 8) + COMPUTE_CRC(mtree->crc, len & 0xff); + reg->crc = ~mtree->crc; + } +#ifdef ARCHIVE_HAS_MD5 + if (mtree->compute_sum & F_MD5) + archive_md5_final(&mtree->md5ctx, reg->buf_md5); +#endif +#ifdef ARCHIVE_HAS_RMD160 + if (mtree->compute_sum & F_RMD160) + archive_rmd160_final(&mtree->rmd160ctx, reg->buf_rmd160); +#endif +#ifdef ARCHIVE_HAS_SHA1 + if (mtree->compute_sum & F_SHA1) + archive_sha1_final(&mtree->sha1ctx, reg->buf_sha1); +#endif +#ifdef ARCHIVE_HAS_SHA256 + if (mtree->compute_sum & F_SHA256) + archive_sha256_final(&mtree->sha256ctx, reg->buf_sha256); +#endif +#ifdef ARCHIVE_HAS_SHA384 + if (mtree->compute_sum & F_SHA384) + archive_sha384_final(&mtree->sha384ctx, reg->buf_sha384); +#endif +#ifdef ARCHIVE_HAS_SHA512 + if (mtree->compute_sum & F_SHA512) + archive_sha512_final(&mtree->sha512ctx, reg->buf_sha512); +#endif + /* Save what types of sum are computed. */ + reg->compute_sum = mtree->compute_sum; +} + +#if defined(ARCHIVE_HAS_MD5) || defined(ARCHIVE_HAS_RMD160) || \ + defined(ARCHIVE_HAS_SHA1) || defined(ARCHIVE_HAS_SHA256) || \ + defined(ARCHIVE_HAS_SHA384) || defined(ARCHIVE_HAS_SHA512) +static void +strappend_bin(struct archive_string *s, const unsigned char *bin, int n) +{ + static const char hex[] = "0123456789abcdef"; + int i; + + for (i = 0; i < n; i++) { + archive_strappend_char(s, hex[bin[i] >> 4]); + archive_strappend_char(s, hex[bin[i] & 0x0f]); + } +} +#endif + +static void +sum_write(struct archive_string *str, struct reg_info *reg) +{ + + if (reg->compute_sum & F_CKSUM) { + archive_string_sprintf(str, " cksum=%ju", + (uintmax_t)reg->crc); + } +#ifdef ARCHIVE_HAS_MD5 + if (reg->compute_sum & F_MD5) { + archive_strcat(str, " md5digest="); + strappend_bin(str, reg->buf_md5, sizeof(reg->buf_md5)); + } +#endif +#ifdef ARCHIVE_HAS_RMD160 + if (reg->compute_sum & F_RMD160) { + archive_strcat(str, " rmd160digest="); + strappend_bin(str, reg->buf_rmd160, sizeof(reg->buf_rmd160)); + } +#endif +#ifdef ARCHIVE_HAS_SHA1 + if (reg->compute_sum & F_SHA1) { + archive_strcat(str, " sha1digest="); + strappend_bin(str, reg->buf_sha1, sizeof(reg->buf_sha1)); + } +#endif +#ifdef ARCHIVE_HAS_SHA256 + if (reg->compute_sum & F_SHA256) { + archive_strcat(str, " sha256digest="); + strappend_bin(str, reg->buf_sha256, sizeof(reg->buf_sha256)); + } +#endif +#ifdef ARCHIVE_HAS_SHA384 + if (reg->compute_sum & F_SHA384) { + archive_strcat(str, " sha384digest="); + strappend_bin(str, reg->buf_sha384, sizeof(reg->buf_sha384)); + } +#endif +#ifdef ARCHIVE_HAS_SHA512 + if (reg->compute_sum & F_SHA512) { + archive_strcat(str, " sha512digest="); + strappend_bin(str, reg->buf_sha512, sizeof(reg->buf_sha512)); + } +#endif +} + +static int +mtree_entry_cmp_node(const struct archive_rb_node *n1, + const struct archive_rb_node *n2) +{ + const struct mtree_entry *e1 = (const struct mtree_entry *)n1; + const struct mtree_entry *e2 = (const struct mtree_entry *)n2; + + return (strcmp(e2->basename.s, e1->basename.s)); +} + +static int +mtree_entry_cmp_key(const struct archive_rb_node *n, const void *key) +{ + const struct mtree_entry *e = (const struct mtree_entry *)n; + + return (strcmp((const char *)key, e->basename.s)); +} + +#if defined(_WIN32) || defined(__CYGWIN__) +static int +cleanup_backslash_1(char *p) +{ + int mb, dos; + + mb = dos = 0; + while (*p) { + if (*(unsigned char *)p > 127) + mb = 1; + if (*p == '\\') { + /* If we have not met any multi-byte characters, + * we can replace '\' with '/'. */ + if (!mb) + *p = '/'; + dos = 1; + } + p++; + } + if (!mb || !dos) + return (0); + return (-1); +} + +static void +cleanup_backslash_2(wchar_t *p) +{ + + /* Convert a path-separator from '\' to '/' */ + while (*p != L'\0') { + if (*p == L'\\') + *p = L'/'; + p++; + } +} +#endif + +/* + * Generate a parent directory name and a base name from a pathname. + */ +static int +mtree_entry_setup_filenames(struct archive_write *a, struct mtree_entry *file, + struct archive_entry *entry) +{ + const char *pathname; + char *p, *dirname, *slash; + size_t len; + int ret = ARCHIVE_OK; + + archive_strcpy(&file->pathname, archive_entry_pathname(entry)); +#if defined(_WIN32) || defined(__CYGWIN__) + /* + * Convert a path-separator from '\' to '/' + */ + if (cleanup_backslash_1(file->pathname.s) != 0) { + const wchar_t *wp = archive_entry_pathname_w(entry); + struct archive_wstring ws; + + if (wp != NULL) { + int r; + archive_string_init(&ws); + archive_wstrcpy(&ws, wp); + cleanup_backslash_2(ws.s); + archive_string_empty(&(file->pathname)); + r = archive_string_append_from_wcs(&(file->pathname), + ws.s, ws.length); + archive_wstring_free(&ws); + if (r < 0 && errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + } + } +#else + (void)a; /* UNUSED */ +#endif + pathname = file->pathname.s; + if (strcmp(pathname, ".") == 0) { + archive_strcpy(&file->basename, "."); + return (ARCHIVE_OK); + } + + archive_strcpy(&(file->parentdir), pathname); + + len = file->parentdir.length; + p = dirname = file->parentdir.s; + + /* + * Remove leading '/' and '../' elements + */ + while (*p) { + if (p[0] == '/') { + p++; + len--; + } else if (p[0] != '.') + break; + else if (p[1] == '.' && p[2] == '/') { + p += 3; + len -= 3; + } else + break; + } + if (p != dirname) { + memmove(dirname, p, len+1); + p = dirname; + } + /* + * Remove "/","/." and "/.." elements from tail. + */ + while (len > 0) { + size_t ll = len; + + if (len > 0 && p[len-1] == '/') { + p[len-1] = '\0'; + len--; + } + if (len > 1 && p[len-2] == '/' && p[len-1] == '.') { + p[len-2] = '\0'; + len -= 2; + } + if (len > 2 && p[len-3] == '/' && p[len-2] == '.' && + p[len-1] == '.') { + p[len-3] = '\0'; + len -= 3; + } + if (ll == len) + break; + } + while (*p) { + if (p[0] == '/') { + if (p[1] == '/') + /* Convert '//' --> '/' */ + memmove(p, p+1, strlen(p+1) + 1); + else if (p[1] == '.' && p[2] == '/') + /* Convert '/./' --> '/' */ + memmove(p, p+2, strlen(p+2) + 1); + else if (p[1] == '.' && p[2] == '.' && p[3] == '/') { + /* Convert 'dir/dir1/../dir2/' + * --> 'dir/dir2/' + */ + char *rp = p -1; + while (rp >= dirname) { + if (*rp == '/') + break; + --rp; + } + if (rp > dirname) { + strcpy(rp, p+3); + p = rp; + } else { + strcpy(dirname, p+4); + p = dirname; + } + } else + p++; + } else + p++; + } + p = dirname; + len = strlen(p); + + /* + * Add "./" prefix. + * NOTE: If the pathname does not have a path separator, we have + * to add "./" to the head of the pathname because mtree reader + * will suppose that it is v1(a.k.a classic) mtree format and + * change the directory unexpectedly and so it will make a wrong + * path. + */ + if (strcmp(p, ".") != 0 && strncmp(p, "./", 2) != 0) { + struct archive_string as; + archive_string_init(&as); + archive_strcpy(&as, "./"); + archive_strncat(&as, p, len); + archive_string_empty(&file->parentdir); + archive_string_concat(&file->parentdir, &as); + archive_string_free(&as); + p = file->parentdir.s; + len = archive_strlen(&file->parentdir); + } + + /* + * Find out the position which points the last position of + * path separator('/'). + */ + slash = NULL; + for (; *p != '\0'; p++) { + if (*p == '/') + slash = p; + } + if (slash == NULL) { + /* The pathname doesn't have a parent directory. */ + file->parentdir.length = len; + archive_string_copy(&(file->basename), &(file->parentdir)); + archive_string_empty(&(file->parentdir)); + *file->parentdir.s = '\0'; + return (ret); + } + + /* Make a basename from file->parentdir.s and slash */ + *slash = '\0'; + file->parentdir.length = slash - file->parentdir.s; + archive_strcpy(&(file->basename), slash + 1); + return (ret); +} + +static int +mtree_entry_create_virtual_dir(struct archive_write *a, const char *pathname, + struct mtree_entry **m_entry) +{ + struct archive_entry *entry; + struct mtree_entry *file; + int r; + + entry = archive_entry_new(); + if (entry == NULL) { + *m_entry = NULL; + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + archive_entry_copy_pathname(entry, pathname); + archive_entry_set_mode(entry, AE_IFDIR | 0755); + archive_entry_set_mtime(entry, time(NULL), 0); + + r = mtree_entry_new(a, entry, &file); + archive_entry_free(entry); + if (r < ARCHIVE_WARN) { + *m_entry = NULL; + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + + file->dir_info->virtual = 1; + + *m_entry = file; + return (ARCHIVE_OK); +} + +static void +mtree_entry_register_add(struct mtree_writer *mtree, struct mtree_entry *file) +{ + file->next = NULL; + *mtree->file_list.last = file; + mtree->file_list.last = &(file->next); +} + +static void +mtree_entry_register_init(struct mtree_writer *mtree) +{ + mtree->file_list.first = NULL; + mtree->file_list.last = &(mtree->file_list.first); +} + +static void +mtree_entry_register_free(struct mtree_writer *mtree) +{ + struct mtree_entry *file, *file_next; + + file = mtree->file_list.first; + while (file != NULL) { + file_next = file->next; + mtree_entry_free(file); + file = file_next; + } +} + +static int +mtree_entry_add_child_tail(struct mtree_entry *parent, + struct mtree_entry *child) +{ + child->dir_info->chnext = NULL; + *parent->dir_info->children.last = child; + parent->dir_info->children.last = &(child->dir_info->chnext); + return (1); +} + +/* + * Find a entry from a parent entry with the name. + */ +static struct mtree_entry * +mtree_entry_find_child(struct mtree_entry *parent, const char *child_name) +{ + struct mtree_entry *np; + + if (parent == NULL) + return (NULL); + np = (struct mtree_entry *)__archive_rb_tree_find_node( + &(parent->dir_info->rbtree), child_name); + return (np); +} + +static int +get_path_component(char *name, size_t n, const char *fn) +{ + char *p; + size_t l; + + p = strchr(fn, '/'); + if (p == NULL) { + if ((l = strlen(fn)) == 0) + return (0); + } else + l = p - fn; + if (l > n -1) + return (-1); + memcpy(name, fn, l); + name[l] = '\0'; + + return ((int)l); +} + +/* + * Add a new entry into the tree. + */ +static int +mtree_entry_tree_add(struct archive_write *a, struct mtree_entry **filep) +{ +#if defined(_WIN32) && !defined(__CYGWIN__) + char name[_MAX_FNAME];/* Included null terminator size. */ +#elif defined(NAME_MAX) && NAME_MAX >= 255 + char name[NAME_MAX+1]; +#else + char name[256]; +#endif + struct mtree_writer *mtree = (struct mtree_writer *)a->format_data; + struct mtree_entry *dent, *file, *np; + const char *fn, *p; + int l, r; + + file = *filep; + if (file->parentdir.length == 0 && file->basename.length == 1 && + file->basename.s[0] == '.') { + file->parent = file; + if (mtree->root != NULL) { + np = mtree->root; + goto same_entry; + } + mtree->root = file; + mtree_entry_register_add(mtree, file); + return (ARCHIVE_OK); + } + + if (file->parentdir.length == 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Internal programming error " + "in generating canonical name for %s", + file->pathname.s); + return (ARCHIVE_FAILED); + } + + fn = p = file->parentdir.s; + + /* + * If the path of the parent directory of `file' entry is + * the same as the path of `cur_dirent', add `file' entry to + * `cur_dirent'. + */ + if (archive_strlen(&(mtree->cur_dirstr)) + == archive_strlen(&(file->parentdir)) && + strcmp(mtree->cur_dirstr.s, fn) == 0) { + if (!__archive_rb_tree_insert_node( + &(mtree->cur_dirent->dir_info->rbtree), + (struct archive_rb_node *)file)) { + /* There is the same name in the tree. */ + np = (struct mtree_entry *)__archive_rb_tree_find_node( + &(mtree->cur_dirent->dir_info->rbtree), + file->basename.s); + goto same_entry; + } + file->parent = mtree->cur_dirent; + mtree_entry_register_add(mtree, file); + return (ARCHIVE_OK); + } + + dent = mtree->root; + for (;;) { + l = get_path_component(name, sizeof(name), fn); + if (l == 0) { + np = NULL; + break; + } + if (l < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "A name buffer is too small"); + return (ARCHIVE_FATAL); + } + if (l == 1 && name[0] == '.' && dent != NULL && + dent == mtree->root) { + fn += l; + if (fn[0] == '/') + fn++; + continue; + } + + np = mtree_entry_find_child(dent, name); + if (np == NULL || fn[0] == '\0') + break; + + /* Find next sub directory. */ + if (!np->dir_info) { + /* NOT Directory! */ + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "`%s' is not directory, we cannot insert `%s' ", + np->pathname.s, file->pathname.s); + return (ARCHIVE_FAILED); + } + fn += l; + if (fn[0] == '/') + fn++; + dent = np; + } + if (np == NULL) { + /* + * Create virtual parent directories. + */ + while (fn[0] != '\0') { + struct mtree_entry *vp; + struct archive_string as; + + archive_string_init(&as); + archive_strncat(&as, p, fn - p + l); + if (as.s[as.length-1] == '/') { + as.s[as.length-1] = '\0'; + as.length--; + } + r = mtree_entry_create_virtual_dir(a, as.s, &vp); + archive_string_free(&as); + if (r < ARCHIVE_WARN) + return (r); + + if (strcmp(vp->pathname.s, ".") == 0) { + vp->parent = vp; + mtree->root = vp; + } else { + __archive_rb_tree_insert_node( + &(dent->dir_info->rbtree), + (struct archive_rb_node *)vp); + vp->parent = dent; + } + mtree_entry_register_add(mtree, vp); + np = vp; + + fn += l; + if (fn[0] == '/') + fn++; + l = get_path_component(name, sizeof(name), fn); + if (l < 0) { + archive_string_free(&as); + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "A name buffer is too small"); + return (ARCHIVE_FATAL); + } + dent = np; + } + + /* Found out the parent directory where `file' can be + * inserted. */ + mtree->cur_dirent = dent; + archive_string_empty(&(mtree->cur_dirstr)); + archive_string_ensure(&(mtree->cur_dirstr), + archive_strlen(&(dent->parentdir)) + + archive_strlen(&(dent->basename)) + 2); + if (archive_strlen(&(dent->parentdir)) + + archive_strlen(&(dent->basename)) == 0) + mtree->cur_dirstr.s[0] = 0; + else { + if (archive_strlen(&(dent->parentdir)) > 0) { + archive_string_copy(&(mtree->cur_dirstr), + &(dent->parentdir)); + archive_strappend_char( + &(mtree->cur_dirstr), '/'); + } + archive_string_concat(&(mtree->cur_dirstr), + &(dent->basename)); + } + + if (!__archive_rb_tree_insert_node( + &(dent->dir_info->rbtree), + (struct archive_rb_node *)file)) { + np = (struct mtree_entry *)__archive_rb_tree_find_node( + &(dent->dir_info->rbtree), file->basename.s); + goto same_entry; + } + file->parent = dent; + mtree_entry_register_add(mtree, file); + return (ARCHIVE_OK); + } + +same_entry: + /* + * We have already has the entry the filename of which is + * the same. + */ + r = mtree_entry_exchange_same_entry(a, np, file); + if (r < ARCHIVE_WARN) + return (r); + if (np->dir_info) + np->dir_info->virtual = 0; + *filep = np; + mtree_entry_free(file); + return (ARCHIVE_WARN); +} + +static int +mtree_entry_exchange_same_entry(struct archive_write *a, struct mtree_entry *np, + struct mtree_entry *file) +{ + + if ((np->mode & AE_IFMT) != (file->mode & AE_IFMT)) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Found duplicate entries `%s' and its file type is " + "different", + np->pathname.s); + return (ARCHIVE_FAILED); + } + + /* Update the existent mtree entry's attributes by the new one's. */ + archive_string_empty(&np->symlink); + archive_string_concat(&np->symlink, &file->symlink); + archive_string_empty(&np->uname); + archive_string_concat(&np->uname, &file->uname); + archive_string_empty(&np->gname); + archive_string_concat(&np->gname, &file->gname); + archive_string_empty(&np->fflags_text); + archive_string_concat(&np->fflags_text, &file->fflags_text); + np->nlink = file->nlink; + np->filetype = file->filetype; + np->mode = file->mode; + np->size = file->size; + np->uid = file->uid; + np->gid = file->gid; + np->fflags_set = file->fflags_set; + np->fflags_clear = file->fflags_clear; + np->mtime = file->mtime; + np->mtime_nsec = file->mtime_nsec; + np->rdevmajor = file->rdevmajor; + np->rdevminor = file->rdevminor; + np->devmajor = file->devmajor; + np->devminor = file->devminor; + np->ino = file->ino; + + return (ARCHIVE_WARN); +} diff --git a/c/archive_write_set_format_pax.c b/c/archive_write_set_format_pax.c new file mode 100644 --- /dev/null +++ b/c/archive_write_set_format_pax.c @@ -0,0 +1,2044 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * Copyright (c) 2010-2012 Michihiro NAKAJIMA + * Copyright (c) 2016 Martin Matuska + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_format_pax.c 201162 2009-12-29 05:47:46Z kientzle $"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_private.h" +#include "archive_write_private.h" +#include "archive_write_set_format_private.h" + +struct sparse_block { + struct sparse_block *next; + int is_hole; + uint64_t offset; + uint64_t remaining; +}; + +struct pax { + uint64_t entry_bytes_remaining; + uint64_t entry_padding; + struct archive_string l_url_encoded_name; + struct archive_string pax_header; + struct archive_string sparse_map; + size_t sparse_map_padding; + struct sparse_block *sparse_list; + struct sparse_block *sparse_tail; + struct archive_string_conv *sconv_utf8; + int opt_binary; + + unsigned flags; +#define WRITE_SCHILY_XATTR (1 << 0) +#define WRITE_LIBARCHIVE_XATTR (1 << 1) +}; + +static void add_pax_attr(struct archive_string *, const char *key, + const char *value); +static void add_pax_attr_binary(struct archive_string *, + const char *key, + const char *value, size_t value_len); +static void add_pax_attr_int(struct archive_string *, + const char *key, int64_t value); +static void add_pax_attr_time(struct archive_string *, + const char *key, int64_t sec, + unsigned long nanos); +static int add_pax_acl(struct archive_write *, + struct archive_entry *, struct pax *, int); +static ssize_t archive_write_pax_data(struct archive_write *, + const void *, size_t); +static int archive_write_pax_close(struct archive_write *); +static int archive_write_pax_free(struct archive_write *); +static int archive_write_pax_finish_entry(struct archive_write *); +static int archive_write_pax_header(struct archive_write *, + struct archive_entry *); +static int archive_write_pax_options(struct archive_write *, + const char *, const char *); +static char *base64_encode(const char *src, size_t len); +static char *build_gnu_sparse_name(char *dest, const char *src); +static char *build_pax_attribute_name(char *dest, const char *src); +static char *build_ustar_entry_name(char *dest, const char *src, + size_t src_length, const char *insert); +static char *format_int(char *dest, int64_t); +static int has_non_ASCII(const char *); +static void sparse_list_clear(struct pax *); +static int sparse_list_add(struct pax *, int64_t, int64_t); +static char *url_encode(const char *in); + +/* + * Set output format to 'restricted pax' format. + * + * This is the same as normal 'pax', but tries to suppress + * the pax header whenever possible. This is the default for + * bsdtar, for instance. + */ +int +archive_write_set_format_pax_restricted(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + int r; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_set_format_pax_restricted"); + + r = archive_write_set_format_pax(&a->archive); + a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_RESTRICTED; + a->archive.archive_format_name = "restricted POSIX pax interchange"; + return (r); +} + +/* + * Set output format to 'pax' format. + */ +int +archive_write_set_format_pax(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct pax *pax; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_set_format_pax"); + + if (a->format_free != NULL) + (a->format_free)(a); + + pax = (struct pax *)calloc(1, sizeof(*pax)); + if (pax == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate pax data"); + return (ARCHIVE_FATAL); + } + pax->flags = WRITE_LIBARCHIVE_XATTR | WRITE_SCHILY_XATTR; + + a->format_data = pax; + a->format_name = "pax"; + a->format_options = archive_write_pax_options; + a->format_write_header = archive_write_pax_header; + a->format_write_data = archive_write_pax_data; + a->format_close = archive_write_pax_close; + a->format_free = archive_write_pax_free; + a->format_finish_entry = archive_write_pax_finish_entry; + a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE; + a->archive.archive_format_name = "POSIX pax interchange"; + return (ARCHIVE_OK); +} + +static int +archive_write_pax_options(struct archive_write *a, const char *key, + const char *val) +{ + struct pax *pax = (struct pax *)a->format_data; + int ret = ARCHIVE_FAILED; + + if (strcmp(key, "hdrcharset") == 0) { + /* + * The character-set we can use are defined in + * IEEE Std 1003.1-2001 + */ + if (val == NULL || val[0] == 0) + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "pax: hdrcharset option needs a character-set name"); + else if (strcmp(val, "BINARY") == 0 || + strcmp(val, "binary") == 0) { + /* + * Specify binary mode. We will not convert + * filenames, uname and gname to any charsets. + */ + pax->opt_binary = 1; + ret = ARCHIVE_OK; + } else if (strcmp(val, "UTF-8") == 0) { + /* + * Specify UTF-8 character-set to be used for + * filenames. This is almost the test that + * running platform supports the string conversion. + * Especially libarchive_test needs this trick for + * its test. + */ + pax->sconv_utf8 = archive_string_conversion_to_charset( + &(a->archive), "UTF-8", 0); + if (pax->sconv_utf8 == NULL) + ret = ARCHIVE_FATAL; + else + ret = ARCHIVE_OK; + } else + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "pax: invalid charset name"); + return (ret); + } else if (strcmp(key, "xattrheader") == 0) { + if (val == NULL || val[0] == 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "pax: xattrheader requires a value"); + } else if (strcmp(val, "ALL") == 0 || + strcmp(val, "all") == 0) { + pax->flags |= WRITE_LIBARCHIVE_XATTR | WRITE_SCHILY_XATTR; + ret = ARCHIVE_OK; + } else if (strcmp(val, "SCHILY") == 0 || + strcmp(val, "schily") == 0) { + pax->flags |= WRITE_SCHILY_XATTR; + pax->flags &= ~WRITE_LIBARCHIVE_XATTR; + ret = ARCHIVE_OK; + } else if (strcmp(val, "LIBARCHIVE") == 0 || + strcmp(val, "libarchive") == 0) { + pax->flags |= WRITE_LIBARCHIVE_XATTR; + pax->flags &= ~WRITE_SCHILY_XATTR; + ret = ARCHIVE_OK; + } else + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "pax: invalid xattr header name"); + return (ret); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +/* + * Note: This code assumes that 'nanos' has the same sign as 'sec', + * which implies that sec=-1, nanos=200000000 represents -1.2 seconds + * and not -0.8 seconds. This is a pretty pedantic point, as we're + * unlikely to encounter many real files created before Jan 1, 1970, + * much less ones with timestamps recorded to sub-second resolution. + */ +static void +add_pax_attr_time(struct archive_string *as, const char *key, + int64_t sec, unsigned long nanos) +{ + int digit, i; + char *t; + /* + * Note that each byte contributes fewer than 3 base-10 + * digits, so this will always be big enough. + */ + char tmp[1 + 3*sizeof(sec) + 1 + 3*sizeof(nanos)]; + + tmp[sizeof(tmp) - 1] = 0; + t = tmp + sizeof(tmp) - 1; + + /* Skip trailing zeros in the fractional part. */ + for (digit = 0, i = 10; i > 0 && digit == 0; i--) { + digit = nanos % 10; + nanos /= 10; + } + + /* Only format the fraction if it's non-zero. */ + if (i > 0) { + while (i > 0) { + *--t = "0123456789"[digit]; + digit = nanos % 10; + nanos /= 10; + i--; + } + *--t = '.'; + } + t = format_int(t, sec); + + add_pax_attr(as, key, t); +} + +static char * +format_int(char *t, int64_t i) +{ + uint64_t ui; + + if (i < 0) + ui = (i == INT64_MIN) ? (uint64_t)(INT64_MAX) + 1 : (uint64_t)(-i); + else + ui = i; + + do { + *--t = "0123456789"[ui % 10]; + } while (ui /= 10); + if (i < 0) + *--t = '-'; + return (t); +} + +static void +add_pax_attr_int(struct archive_string *as, const char *key, int64_t value) +{ + char tmp[1 + 3 * sizeof(value)]; + + tmp[sizeof(tmp) - 1] = 0; + add_pax_attr(as, key, format_int(tmp + sizeof(tmp) - 1, value)); +} + +/* + * Add a key/value attribute to the pax header. This function handles + * the length field and various other syntactic requirements. + */ +static void +add_pax_attr(struct archive_string *as, const char *key, const char *value) +{ + add_pax_attr_binary(as, key, value, strlen(value)); +} + +/* + * Add a key/value attribute to the pax header. This function handles + * binary values. + */ +static void +add_pax_attr_binary(struct archive_string *as, const char *key, + const char *value, size_t value_len) +{ + int digits, i, len, next_ten; + char tmp[1 + 3 * sizeof(int)]; /* < 3 base-10 digits per byte */ + + /*- + * PAX attributes have the following layout: + * <=> + */ + len = 1 + (int)strlen(key) + 1 + (int)value_len + 1; + + /* + * The field includes the length of the field, so + * computing the correct length is tricky. I start by + * counting the number of base-10 digits in 'len' and + * computing the next higher power of 10. + */ + next_ten = 1; + digits = 0; + i = len; + while (i > 0) { + i = i / 10; + digits++; + next_ten = next_ten * 10; + } + /* + * For example, if string without the length field is 99 + * chars, then adding the 2 digit length "99" will force the + * total length past 100, requiring an extra digit. The next + * statement adjusts for this effect. + */ + if (len + digits >= next_ten) + digits++; + + /* Now, we have the right length so we can build the line. */ + tmp[sizeof(tmp) - 1] = 0; /* Null-terminate the work area. */ + archive_strcat(as, format_int(tmp + sizeof(tmp) - 1, len + digits)); + archive_strappend_char(as, ' '); + archive_strcat(as, key); + archive_strappend_char(as, '='); + archive_array_append(as, value, value_len); + archive_strappend_char(as, '\n'); +} + +static void +archive_write_pax_header_xattr(struct pax *pax, const char *encoded_name, + const void *value, size_t value_len) +{ + struct archive_string s; + char *encoded_value; + + if (pax->flags & WRITE_LIBARCHIVE_XATTR) { + encoded_value = base64_encode((const char *)value, value_len); + + if (encoded_name != NULL && encoded_value != NULL) { + archive_string_init(&s); + archive_strcpy(&s, "LIBARCHIVE.xattr."); + archive_strcat(&s, encoded_name); + add_pax_attr(&(pax->pax_header), s.s, encoded_value); + archive_string_free(&s); + } + free(encoded_value); + } + if (pax->flags & WRITE_SCHILY_XATTR) { + archive_string_init(&s); + archive_strcpy(&s, "SCHILY.xattr."); + archive_strcat(&s, encoded_name); + add_pax_attr_binary(&(pax->pax_header), s.s, value, value_len); + archive_string_free(&s); + } +} + +static int +archive_write_pax_header_xattrs(struct archive_write *a, + struct pax *pax, struct archive_entry *entry) +{ + int i = archive_entry_xattr_reset(entry); + + while (i--) { + const char *name; + const void *value; + char *url_encoded_name = NULL, *encoded_name = NULL; + size_t size; + int r; + + archive_entry_xattr_next(entry, &name, &value, &size); + url_encoded_name = url_encode(name); + if (url_encoded_name != NULL) { + /* Convert narrow-character to UTF-8. */ + r = archive_strcpy_l(&(pax->l_url_encoded_name), + url_encoded_name, pax->sconv_utf8); + free(url_encoded_name); /* Done with this. */ + if (r == 0) + encoded_name = pax->l_url_encoded_name.s; + else if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Linkname"); + return (ARCHIVE_FATAL); + } + } + + archive_write_pax_header_xattr(pax, encoded_name, + value, size); + + } + return (ARCHIVE_OK); +} + +static int +get_entry_hardlink(struct archive_write *a, struct archive_entry *entry, + const char **name, size_t *length, struct archive_string_conv *sc) +{ + int r; + + r = archive_entry_hardlink_l(entry, name, length, sc); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Linkname"); + return (ARCHIVE_FATAL); + } + return (ARCHIVE_WARN); + } + return (ARCHIVE_OK); +} + +static int +get_entry_pathname(struct archive_write *a, struct archive_entry *entry, + const char **name, size_t *length, struct archive_string_conv *sc) +{ + int r; + + r = archive_entry_pathname_l(entry, name, length, sc); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Pathname"); + return (ARCHIVE_FATAL); + } + return (ARCHIVE_WARN); + } + return (ARCHIVE_OK); +} + +static int +get_entry_uname(struct archive_write *a, struct archive_entry *entry, + const char **name, size_t *length, struct archive_string_conv *sc) +{ + int r; + + r = archive_entry_uname_l(entry, name, length, sc); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Uname"); + return (ARCHIVE_FATAL); + } + return (ARCHIVE_WARN); + } + return (ARCHIVE_OK); +} + +static int +get_entry_gname(struct archive_write *a, struct archive_entry *entry, + const char **name, size_t *length, struct archive_string_conv *sc) +{ + int r; + + r = archive_entry_gname_l(entry, name, length, sc); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Gname"); + return (ARCHIVE_FATAL); + } + return (ARCHIVE_WARN); + } + return (ARCHIVE_OK); +} + +static int +get_entry_symlink(struct archive_write *a, struct archive_entry *entry, + const char **name, size_t *length, struct archive_string_conv *sc) +{ + int r; + + r = archive_entry_symlink_l(entry, name, length, sc); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Linkname"); + return (ARCHIVE_FATAL); + } + return (ARCHIVE_WARN); + } + return (ARCHIVE_OK); +} + +/* Add ACL to pax header */ +static int +add_pax_acl(struct archive_write *a, + struct archive_entry *entry, struct pax *pax, int flags) +{ + char *p; + const char *attr; + int acl_types; + + acl_types = archive_entry_acl_types(entry); + + if ((acl_types & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) + attr = "SCHILY.acl.ace"; + else if ((flags & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) + attr = "SCHILY.acl.access"; + else if ((flags & ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) != 0) + attr = "SCHILY.acl.default"; + else + return (ARCHIVE_FATAL); + + p = archive_entry_acl_to_text_l(entry, NULL, flags, pax->sconv_utf8); + if (p == NULL) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, "%s %s", + "Can't allocate memory for ", attr); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, "%s %s %s", + "Can't translate ", attr, " to UTF-8"); + return(ARCHIVE_WARN); + } + + if (*p != '\0') { + add_pax_attr(&(pax->pax_header), + attr, p); + } + free(p); + return(ARCHIVE_OK); +} + +/* + * TODO: Consider adding 'comment' and 'charset' fields to + * archive_entry so that clients can specify them. Also, consider + * adding generic key/value tags so clients can add arbitrary + * key/value data. + * + * TODO: Break up this 700-line function!!!! Yowza! + */ +static int +archive_write_pax_header(struct archive_write *a, + struct archive_entry *entry_original) +{ + struct archive_entry *entry_main; + const char *p; + const char *suffix; + int need_extension, r, ret; + int acl_types; + int sparse_count; + uint64_t sparse_total, real_size; + struct pax *pax; + const char *hardlink; + const char *path = NULL, *linkpath = NULL; + const char *uname = NULL, *gname = NULL; + const void *mac_metadata; + size_t mac_metadata_size; + struct archive_string_conv *sconv; + size_t hardlink_length, path_length, linkpath_length; + size_t uname_length, gname_length; + + char paxbuff[512]; + char ustarbuff[512]; + char ustar_entry_name[256]; + char pax_entry_name[256]; + char gnu_sparse_name[256]; + struct archive_string entry_name; + + ret = ARCHIVE_OK; + need_extension = 0; + pax = (struct pax *)a->format_data; + + /* Sanity check. */ + if (archive_entry_pathname(entry_original) == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Can't record entry in tar file without pathname"); + return (ARCHIVE_FAILED); + } + + /* + * Choose a header encoding. + */ + if (pax->opt_binary) + sconv = NULL;/* Binary mode. */ + else { + /* Header encoding is UTF-8. */ + if (pax->sconv_utf8 == NULL) { + /* Initialize the string conversion object + * we must need */ + pax->sconv_utf8 = archive_string_conversion_to_charset( + &(a->archive), "UTF-8", 1); + if (pax->sconv_utf8 == NULL) + /* Couldn't allocate memory */ + return (ARCHIVE_FAILED); + } + sconv = pax->sconv_utf8; + } + + r = get_entry_hardlink(a, entry_original, &hardlink, + &hardlink_length, sconv); + if (r == ARCHIVE_FATAL) + return (r); + else if (r != ARCHIVE_OK) { + r = get_entry_hardlink(a, entry_original, &hardlink, + &hardlink_length, NULL); + if (r == ARCHIVE_FATAL) + return (r); + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate linkname '%s' to %s", hardlink, + archive_string_conversion_charset_name(sconv)); + ret = ARCHIVE_WARN; + sconv = NULL;/* The header charset switches to binary mode. */ + } + + /* Make sure this is a type of entry that we can handle here */ + if (hardlink == NULL) { + switch (archive_entry_filetype(entry_original)) { + case AE_IFBLK: + case AE_IFCHR: + case AE_IFIFO: + case AE_IFLNK: + case AE_IFREG: + break; + case AE_IFDIR: + { + /* + * Ensure a trailing '/'. Modify the original + * entry so the client sees the change. + */ +#if defined(_WIN32) && !defined(__CYGWIN__) + const wchar_t *wp; + + wp = archive_entry_pathname_w(entry_original); + if (wp != NULL && wp[wcslen(wp) -1] != L'/') { + struct archive_wstring ws; + + archive_string_init(&ws); + path_length = wcslen(wp); + if (archive_wstring_ensure(&ws, + path_length + 2) == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate pax data"); + archive_wstring_free(&ws); + return(ARCHIVE_FATAL); + } + /* Should we keep '\' ? */ + if (wp[path_length -1] == L'\\') + path_length--; + archive_wstrncpy(&ws, wp, path_length); + archive_wstrappend_wchar(&ws, L'/'); + archive_entry_copy_pathname_w( + entry_original, ws.s); + archive_wstring_free(&ws); + p = NULL; + } else +#endif + p = archive_entry_pathname(entry_original); + /* + * On Windows, this is a backup operation just in + * case getting WCS failed. On POSIX, this is a + * normal operation. + */ + if (p != NULL && p[0] != '\0' && p[strlen(p) - 1] != '/') { + struct archive_string as; + + archive_string_init(&as); + path_length = strlen(p); + if (archive_string_ensure(&as, + path_length + 2) == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate pax data"); + archive_string_free(&as); + return(ARCHIVE_FATAL); + } +#if defined(_WIN32) && !defined(__CYGWIN__) + /* NOTE: This might break the pathname + * if the current code page is CP932 and + * the pathname includes a character '\' + * as a part of its multibyte pathname. */ + if (p[strlen(p) -1] == '\\') + path_length--; + else +#endif + archive_strncpy(&as, p, path_length); + archive_strappend_char(&as, '/'); + archive_entry_copy_pathname( + entry_original, as.s); + archive_string_free(&as); + } + break; + } + default: /* AE_IFSOCK and unknown */ + __archive_write_entry_filetype_unsupported( + &a->archive, entry_original, "pax"); + return (ARCHIVE_FAILED); + } + } + + /* + * If Mac OS metadata blob is here, recurse to write that + * as a separate entry. This is really a pretty poor design: + * In particular, it doubles the overhead for long filenames. + * TODO: Help Apple folks design something better and figure + * out how to transition from this legacy format. + * + * Note that this code is present on every platform; clients + * on non-Mac are unlikely to ever provide this data, but + * applications that copy entries from one archive to another + * should not lose data just because the local filesystem + * can't store it. + */ + mac_metadata = + archive_entry_mac_metadata(entry_original, &mac_metadata_size); + if (mac_metadata != NULL) { + const char *oname; + char *name, *bname; + size_t name_length; + struct archive_entry *extra = archive_entry_new2(&a->archive); + + oname = archive_entry_pathname(entry_original); + name_length = strlen(oname); + name = malloc(name_length + 3); + if (name == NULL || extra == NULL) { + /* XXX error message */ + archive_entry_free(extra); + free(name); + return (ARCHIVE_FAILED); + } + strcpy(name, oname); + /* Find last '/'; strip trailing '/' characters */ + bname = strrchr(name, '/'); + while (bname != NULL && bname[1] == '\0') { + *bname = '\0'; + bname = strrchr(name, '/'); + } + if (bname == NULL) { + memmove(name + 2, name, name_length + 1); + memmove(name, "._", 2); + } else { + bname += 1; + memmove(bname + 2, bname, strlen(bname) + 1); + memmove(bname, "._", 2); + } + archive_entry_copy_pathname(extra, name); + free(name); + + archive_entry_set_size(extra, mac_metadata_size); + archive_entry_set_filetype(extra, AE_IFREG); + archive_entry_set_perm(extra, + archive_entry_perm(entry_original)); + archive_entry_set_mtime(extra, + archive_entry_mtime(entry_original), + archive_entry_mtime_nsec(entry_original)); + archive_entry_set_gid(extra, + archive_entry_gid(entry_original)); + archive_entry_set_gname(extra, + archive_entry_gname(entry_original)); + archive_entry_set_uid(extra, + archive_entry_uid(entry_original)); + archive_entry_set_uname(extra, + archive_entry_uname(entry_original)); + + /* Recurse to write the special copyfile entry. */ + r = archive_write_pax_header(a, extra); + archive_entry_free(extra); + if (r < ARCHIVE_WARN) + return (r); + if (r < ret) + ret = r; + r = (int)archive_write_pax_data(a, mac_metadata, + mac_metadata_size); + if (r < ARCHIVE_WARN) + return (r); + if (r < ret) + ret = r; + r = archive_write_pax_finish_entry(a); + if (r < ARCHIVE_WARN) + return (r); + if (r < ret) + ret = r; + } + + /* Copy entry so we can modify it as needed. */ +#if defined(_WIN32) && !defined(__CYGWIN__) + /* Make sure the path separators in pathname, hardlink and symlink + * are all slash '/', not the Windows path separator '\'. */ + entry_main = __la_win_entry_in_posix_pathseparator(entry_original); + if (entry_main == entry_original) + entry_main = archive_entry_clone(entry_original); +#else + entry_main = archive_entry_clone(entry_original); +#endif + if (entry_main == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate pax data"); + return(ARCHIVE_FATAL); + } + archive_string_empty(&(pax->pax_header)); /* Blank our work area. */ + archive_string_empty(&(pax->sparse_map)); + sparse_total = 0; + sparse_list_clear(pax); + + if (hardlink == NULL && + archive_entry_filetype(entry_main) == AE_IFREG) + sparse_count = archive_entry_sparse_reset(entry_main); + else + sparse_count = 0; + if (sparse_count) { + int64_t offset, length, last_offset = 0; + /* Get the last entry of sparse block. */ + while (archive_entry_sparse_next( + entry_main, &offset, &length) == ARCHIVE_OK) + last_offset = offset + length; + + /* If the last sparse block does not reach the end of file, + * We have to add a empty sparse block as the last entry to + * manage storing file data. */ + if (last_offset < archive_entry_size(entry_main)) + archive_entry_sparse_add_entry(entry_main, + archive_entry_size(entry_main), 0); + sparse_count = archive_entry_sparse_reset(entry_main); + } + + /* + * First, check the name fields and see if any of them + * require binary coding. If any of them does, then all of + * them do. + */ + r = get_entry_pathname(a, entry_main, &path, &path_length, sconv); + if (r == ARCHIVE_FATAL) { + archive_entry_free(entry_main); + return (r); + } else if (r != ARCHIVE_OK) { + r = get_entry_pathname(a, entry_main, &path, + &path_length, NULL); + if (r == ARCHIVE_FATAL) { + archive_entry_free(entry_main); + return (r); + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate pathname '%s' to %s", path, + archive_string_conversion_charset_name(sconv)); + ret = ARCHIVE_WARN; + sconv = NULL;/* The header charset switches to binary mode. */ + } + r = get_entry_uname(a, entry_main, &uname, &uname_length, sconv); + if (r == ARCHIVE_FATAL) { + archive_entry_free(entry_main); + return (r); + } else if (r != ARCHIVE_OK) { + r = get_entry_uname(a, entry_main, &uname, &uname_length, NULL); + if (r == ARCHIVE_FATAL) { + archive_entry_free(entry_main); + return (r); + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate uname '%s' to %s", uname, + archive_string_conversion_charset_name(sconv)); + ret = ARCHIVE_WARN; + sconv = NULL;/* The header charset switches to binary mode. */ + } + r = get_entry_gname(a, entry_main, &gname, &gname_length, sconv); + if (r == ARCHIVE_FATAL) { + archive_entry_free(entry_main); + return (r); + } else if (r != ARCHIVE_OK) { + r = get_entry_gname(a, entry_main, &gname, &gname_length, NULL); + if (r == ARCHIVE_FATAL) { + archive_entry_free(entry_main); + return (r); + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate gname '%s' to %s", gname, + archive_string_conversion_charset_name(sconv)); + ret = ARCHIVE_WARN; + sconv = NULL;/* The header charset switches to binary mode. */ + } + linkpath = hardlink; + linkpath_length = hardlink_length; + if (linkpath == NULL) { + r = get_entry_symlink(a, entry_main, &linkpath, + &linkpath_length, sconv); + if (r == ARCHIVE_FATAL) { + archive_entry_free(entry_main); + return (r); + } else if (r != ARCHIVE_OK) { + r = get_entry_symlink(a, entry_main, &linkpath, + &linkpath_length, NULL); + if (r == ARCHIVE_FATAL) { + archive_entry_free(entry_main); + return (r); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate linkname '%s' to %s", linkpath, + archive_string_conversion_charset_name(sconv)); + ret = ARCHIVE_WARN; + sconv = NULL; + } + } + + /* If any string conversions failed, get all attributes + * in binary-mode. */ + if (sconv == NULL && !pax->opt_binary) { + if (hardlink != NULL) { + r = get_entry_hardlink(a, entry_main, &hardlink, + &hardlink_length, NULL); + if (r == ARCHIVE_FATAL) { + archive_entry_free(entry_main); + return (r); + } + linkpath = hardlink; + linkpath_length = hardlink_length; + } + r = get_entry_pathname(a, entry_main, &path, + &path_length, NULL); + if (r == ARCHIVE_FATAL) { + archive_entry_free(entry_main); + return (r); + } + r = get_entry_uname(a, entry_main, &uname, &uname_length, NULL); + if (r == ARCHIVE_FATAL) { + archive_entry_free(entry_main); + return (r); + } + r = get_entry_gname(a, entry_main, &gname, &gname_length, NULL); + if (r == ARCHIVE_FATAL) { + archive_entry_free(entry_main); + return (r); + } + } + + /* Store the header encoding first, to be nice to readers. */ + if (sconv == NULL) + add_pax_attr(&(pax->pax_header), "hdrcharset", "BINARY"); + + + /* + * If name is too long, or has non-ASCII characters, add + * 'path' to pax extended attrs. (Note that an unconvertible + * name must have non-ASCII characters.) + */ + if (has_non_ASCII(path)) { + /* We have non-ASCII characters. */ + add_pax_attr(&(pax->pax_header), "path", path); + archive_entry_set_pathname(entry_main, + build_ustar_entry_name(ustar_entry_name, + path, path_length, NULL)); + need_extension = 1; + } else { + /* We have an all-ASCII path; we'd like to just store + * it in the ustar header if it will fit. Yes, this + * duplicates some of the logic in + * archive_write_set_format_ustar.c + */ + if (path_length <= 100) { + /* Fits in the old 100-char tar name field. */ + } else { + /* Find largest suffix that will fit. */ + /* Note: strlen() > 100, so strlen() - 100 - 1 >= 0 */ + suffix = strchr(path + path_length - 100 - 1, '/'); + /* Don't attempt an empty prefix. */ + if (suffix == path) + suffix = strchr(suffix + 1, '/'); + /* We can put it in the ustar header if it's + * all ASCII and it's either <= 100 characters + * or can be split at a '/' into a prefix <= + * 155 chars and a suffix <= 100 chars. (Note + * the strchr() above will return NULL exactly + * when the path can't be split.) + */ + if (suffix == NULL /* Suffix > 100 chars. */ + || suffix[1] == '\0' /* empty suffix */ + || suffix - path > 155) /* Prefix > 155 chars */ + { + add_pax_attr(&(pax->pax_header), "path", path); + archive_entry_set_pathname(entry_main, + build_ustar_entry_name(ustar_entry_name, + path, path_length, NULL)); + need_extension = 1; + } + } + } + + if (linkpath != NULL) { + /* If link name is too long or has non-ASCII characters, add + * 'linkpath' to pax extended attrs. */ + if (linkpath_length > 100 || has_non_ASCII(linkpath)) { + add_pax_attr(&(pax->pax_header), "linkpath", linkpath); + if (linkpath_length > 100) { + if (hardlink != NULL) + archive_entry_set_hardlink(entry_main, + "././@LongHardLink"); + else + archive_entry_set_symlink(entry_main, + "././@LongSymLink"); + } + need_extension = 1; + } + } + /* Save a pathname since it will be renamed if `entry_main` has + * sparse blocks. */ + archive_string_init(&entry_name); + archive_strcpy(&entry_name, archive_entry_pathname(entry_main)); + + /* If file size is too large, add 'size' to pax extended attrs. */ + if (archive_entry_size(entry_main) >= (((int64_t)1) << 33)) { + add_pax_attr_int(&(pax->pax_header), "size", + archive_entry_size(entry_main)); + need_extension = 1; + } + + /* If numeric GID is too large, add 'gid' to pax extended attrs. */ + if ((unsigned int)archive_entry_gid(entry_main) >= (1 << 18)) { + add_pax_attr_int(&(pax->pax_header), "gid", + archive_entry_gid(entry_main)); + need_extension = 1; + } + + /* If group name is too large or has non-ASCII characters, add + * 'gname' to pax extended attrs. */ + if (gname != NULL) { + if (gname_length > 31 || has_non_ASCII(gname)) { + add_pax_attr(&(pax->pax_header), "gname", gname); + need_extension = 1; + } + } + + /* If numeric UID is too large, add 'uid' to pax extended attrs. */ + if ((unsigned int)archive_entry_uid(entry_main) >= (1 << 18)) { + add_pax_attr_int(&(pax->pax_header), "uid", + archive_entry_uid(entry_main)); + need_extension = 1; + } + + /* Add 'uname' to pax extended attrs if necessary. */ + if (uname != NULL) { + if (uname_length > 31 || has_non_ASCII(uname)) { + add_pax_attr(&(pax->pax_header), "uname", uname); + need_extension = 1; + } + } + + /* + * POSIX/SUSv3 doesn't provide a standard key for large device + * numbers. I use the same keys here that Joerg Schilling + * used for 'star.' (Which, somewhat confusingly, are called + * "devXXX" even though they code "rdev" values.) No doubt, + * other implementations use other keys. Note that there's no + * reason we can't write the same information into a number of + * different keys. + * + * Of course, this is only needed for block or char device entries. + */ + if (archive_entry_filetype(entry_main) == AE_IFBLK + || archive_entry_filetype(entry_main) == AE_IFCHR) { + /* + * If rdevmajor is too large, add 'SCHILY.devmajor' to + * extended attributes. + */ + int rdevmajor, rdevminor; + rdevmajor = archive_entry_rdevmajor(entry_main); + rdevminor = archive_entry_rdevminor(entry_main); + if (rdevmajor >= (1 << 18)) { + add_pax_attr_int(&(pax->pax_header), "SCHILY.devmajor", + rdevmajor); + /* + * Non-strict formatting below means we don't + * have to truncate here. Not truncating improves + * the chance that some more modern tar archivers + * (such as GNU tar 1.13) can restore the full + * value even if they don't understand the pax + * extended attributes. See my rant below about + * file size fields for additional details. + */ + /* archive_entry_set_rdevmajor(entry_main, + rdevmajor & ((1 << 18) - 1)); */ + need_extension = 1; + } + + /* + * If devminor is too large, add 'SCHILY.devminor' to + * extended attributes. + */ + if (rdevminor >= (1 << 18)) { + add_pax_attr_int(&(pax->pax_header), "SCHILY.devminor", + rdevminor); + /* Truncation is not necessary here, either. */ + /* archive_entry_set_rdevminor(entry_main, + rdevminor & ((1 << 18) - 1)); */ + need_extension = 1; + } + } + + /* + * Technically, the mtime field in the ustar header can + * support 33 bits, but many platforms use signed 32-bit time + * values. The cutoff of 0x7fffffff here is a compromise. + * Yes, this check is duplicated just below; this helps to + * avoid writing an mtime attribute just to handle a + * high-resolution timestamp in "restricted pax" mode. + */ + if (!need_extension && + ((archive_entry_mtime(entry_main) < 0) + || (archive_entry_mtime(entry_main) >= 0x7fffffff))) + need_extension = 1; + + /* I use a star-compatible file flag attribute. */ + p = archive_entry_fflags_text(entry_main); + if (!need_extension && p != NULL && *p != '\0') + need_extension = 1; + + /* If there are extended attributes, we need an extension */ + if (!need_extension && archive_entry_xattr_count(entry_original) > 0) + need_extension = 1; + + /* If there are sparse info, we need an extension */ + if (!need_extension && sparse_count > 0) + need_extension = 1; + + acl_types = archive_entry_acl_types(entry_original); + + /* If there are any ACL entries, we need an extension */ + if (!need_extension && acl_types != 0) + need_extension = 1; + + /* If the symlink type is defined, we need an extension */ + if (!need_extension && archive_entry_symlink_type(entry_main) > 0) + need_extension = 1; + + /* + * Libarchive used to include these in extended headers for + * restricted pax format, but that confused people who + * expected ustar-like time semantics. So now we only include + * them in full pax format. + */ + if (a->archive.archive_format != ARCHIVE_FORMAT_TAR_PAX_RESTRICTED) { + if (archive_entry_ctime(entry_main) != 0 || + archive_entry_ctime_nsec(entry_main) != 0) + add_pax_attr_time(&(pax->pax_header), "ctime", + archive_entry_ctime(entry_main), + archive_entry_ctime_nsec(entry_main)); + + if (archive_entry_atime(entry_main) != 0 || + archive_entry_atime_nsec(entry_main) != 0) + add_pax_attr_time(&(pax->pax_header), "atime", + archive_entry_atime(entry_main), + archive_entry_atime_nsec(entry_main)); + + /* Store birth/creationtime only if it's earlier than mtime */ + if (archive_entry_birthtime_is_set(entry_main) && + archive_entry_birthtime(entry_main) + < archive_entry_mtime(entry_main)) + add_pax_attr_time(&(pax->pax_header), + "LIBARCHIVE.creationtime", + archive_entry_birthtime(entry_main), + archive_entry_birthtime_nsec(entry_main)); + } + + /* + * The following items are handled differently in "pax + * restricted" format. In particular, in "pax restricted" + * format they won't be added unless need_extension is + * already set (we're already generating an extended header, so + * may as well include these). + */ + if (a->archive.archive_format != ARCHIVE_FORMAT_TAR_PAX_RESTRICTED || + need_extension) { + if (archive_entry_mtime(entry_main) < 0 || + archive_entry_mtime(entry_main) >= 0x7fffffff || + archive_entry_mtime_nsec(entry_main) != 0) + add_pax_attr_time(&(pax->pax_header), "mtime", + archive_entry_mtime(entry_main), + archive_entry_mtime_nsec(entry_main)); + + /* I use a star-compatible file flag attribute. */ + p = archive_entry_fflags_text(entry_main); + if (p != NULL && *p != '\0') + add_pax_attr(&(pax->pax_header), "SCHILY.fflags", p); + + /* I use star-compatible ACL attributes. */ + if ((acl_types & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { + ret = add_pax_acl(a, entry_original, pax, + ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID | + ARCHIVE_ENTRY_ACL_STYLE_SEPARATOR_COMMA | + ARCHIVE_ENTRY_ACL_STYLE_COMPACT); + if (ret == ARCHIVE_FATAL) { + archive_entry_free(entry_main); + archive_string_free(&entry_name); + return (ARCHIVE_FATAL); + } + } + if (acl_types & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) { + ret = add_pax_acl(a, entry_original, pax, + ARCHIVE_ENTRY_ACL_TYPE_ACCESS | + ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID | + ARCHIVE_ENTRY_ACL_STYLE_SEPARATOR_COMMA); + if (ret == ARCHIVE_FATAL) { + archive_entry_free(entry_main); + archive_string_free(&entry_name); + return (ARCHIVE_FATAL); + } + } + if (acl_types & ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) { + ret = add_pax_acl(a, entry_original, pax, + ARCHIVE_ENTRY_ACL_TYPE_DEFAULT | + ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID | + ARCHIVE_ENTRY_ACL_STYLE_SEPARATOR_COMMA); + if (ret == ARCHIVE_FATAL) { + archive_entry_free(entry_main); + archive_string_free(&entry_name); + return (ARCHIVE_FATAL); + } + } + + /* We use GNU-tar-compatible sparse attributes. */ + if (sparse_count > 0) { + int64_t soffset, slength; + + add_pax_attr_int(&(pax->pax_header), + "GNU.sparse.major", 1); + add_pax_attr_int(&(pax->pax_header), + "GNU.sparse.minor", 0); + /* + * Make sure to store the original path, since + * truncation to ustar limit happened already. + */ + add_pax_attr(&(pax->pax_header), + "GNU.sparse.name", path); + add_pax_attr_int(&(pax->pax_header), + "GNU.sparse.realsize", + archive_entry_size(entry_main)); + + /* Rename the file name which will be used for + * ustar header to a special name, which GNU + * PAX Format 1.0 requires */ + archive_entry_set_pathname(entry_main, + build_gnu_sparse_name(gnu_sparse_name, + entry_name.s)); + + /* + * - Make a sparse map, which will precede a file data. + * - Get the total size of available data of sparse. + */ + archive_string_sprintf(&(pax->sparse_map), "%d\n", + sparse_count); + while (archive_entry_sparse_next(entry_main, + &soffset, &slength) == ARCHIVE_OK) { + archive_string_sprintf(&(pax->sparse_map), + "%jd\n%jd\n", + (intmax_t)soffset, + (intmax_t)slength); + sparse_total += slength; + if (sparse_list_add(pax, soffset, slength) + != ARCHIVE_OK) { + archive_set_error(&a->archive, + ENOMEM, + "Can't allocate memory"); + archive_entry_free(entry_main); + archive_string_free(&entry_name); + return (ARCHIVE_FATAL); + } + } + } + + /* Store extended attributes */ + if (archive_write_pax_header_xattrs(a, pax, entry_original) + == ARCHIVE_FATAL) { + archive_entry_free(entry_main); + archive_string_free(&entry_name); + return (ARCHIVE_FATAL); + } + + /* Store extended symlink information */ + if (archive_entry_symlink_type(entry_main) == + AE_SYMLINK_TYPE_FILE) { + add_pax_attr(&(pax->pax_header), + "LIBARCHIVE.symlinktype", "file"); + } else if (archive_entry_symlink_type(entry_main) == + AE_SYMLINK_TYPE_DIRECTORY) { + add_pax_attr(&(pax->pax_header), + "LIBARCHIVE.symlinktype", "dir"); + } + } + + /* Only regular files have data. */ + if (archive_entry_filetype(entry_main) != AE_IFREG) + archive_entry_set_size(entry_main, 0); + + /* + * Pax-restricted does not store data for hardlinks, in order + * to improve compatibility with ustar. + */ + if (a->archive.archive_format != ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE && + hardlink != NULL) + archive_entry_set_size(entry_main, 0); + + /* + * XXX Full pax interchange format does permit a hardlink + * entry to have data associated with it. I'm not supporting + * that here because the client expects me to tell them whether + * or not this format expects data for hardlinks. If I + * don't check here, then every pax archive will end up with + * duplicated data for hardlinks. Someday, there may be + * need to select this behavior, in which case the following + * will need to be revisited. XXX + */ + if (hardlink != NULL) + archive_entry_set_size(entry_main, 0); + + /* Save a real file size. */ + real_size = archive_entry_size(entry_main); + /* + * Overwrite a file size by the total size of sparse blocks and + * the size of sparse map info. That file size is the length of + * the data, which we will exactly store into an archive file. + */ + if (archive_strlen(&(pax->sparse_map))) { + size_t mapsize = archive_strlen(&(pax->sparse_map)); + pax->sparse_map_padding = 0x1ff & (-(ssize_t)mapsize); + archive_entry_set_size(entry_main, + mapsize + pax->sparse_map_padding + sparse_total); + } + + /* Format 'ustar' header for main entry. + * + * The trouble with file size: If the reader can't understand + * the file size, they may not be able to locate the next + * entry and the rest of the archive is toast. Pax-compliant + * readers are supposed to ignore the file size in the main + * header, so the question becomes how to maximize portability + * for readers that don't support pax attribute extensions. + * For maximum compatibility, I permit numeric extensions in + * the main header so that the file size stored will always be + * correct, even if it's in a format that only some + * implementations understand. The technique used here is: + * + * a) If possible, follow the standard exactly. This handles + * files up to 8 gigabytes minus 1. + * + * b) If that fails, try octal but omit the field terminator. + * That handles files up to 64 gigabytes minus 1. + * + * c) Otherwise, use base-256 extensions. That handles files + * up to 2^63 in this implementation, with the potential to + * go up to 2^94. That should hold us for a while. ;-) + * + * The non-strict formatter uses similar logic for other + * numeric fields, though they're less critical. + */ + if (__archive_write_format_header_ustar(a, ustarbuff, entry_main, -1, 0, + NULL) == ARCHIVE_FATAL) { + archive_entry_free(entry_main); + archive_string_free(&entry_name); + return (ARCHIVE_FATAL); + } + + /* If we built any extended attributes, write that entry first. */ + if (archive_strlen(&(pax->pax_header)) > 0) { + struct archive_entry *pax_attr_entry; + time_t s; + int64_t uid, gid; + int mode; + + pax_attr_entry = archive_entry_new2(&a->archive); + p = entry_name.s; + archive_entry_set_pathname(pax_attr_entry, + build_pax_attribute_name(pax_entry_name, p)); + archive_entry_set_size(pax_attr_entry, + archive_strlen(&(pax->pax_header))); + /* Copy uid/gid (but clip to ustar limits). */ + uid = archive_entry_uid(entry_main); + if (uid >= 1 << 18) + uid = (1 << 18) - 1; + archive_entry_set_uid(pax_attr_entry, uid); + gid = archive_entry_gid(entry_main); + if (gid >= 1 << 18) + gid = (1 << 18) - 1; + archive_entry_set_gid(pax_attr_entry, gid); + /* Copy mode over (but not setuid/setgid bits) */ + mode = archive_entry_mode(entry_main); +#ifdef S_ISUID + mode &= ~S_ISUID; +#endif +#ifdef S_ISGID + mode &= ~S_ISGID; +#endif +#ifdef S_ISVTX + mode &= ~S_ISVTX; +#endif + archive_entry_set_mode(pax_attr_entry, mode); + + /* Copy uname/gname. */ + archive_entry_set_uname(pax_attr_entry, + archive_entry_uname(entry_main)); + archive_entry_set_gname(pax_attr_entry, + archive_entry_gname(entry_main)); + + /* Copy mtime, but clip to ustar limits. */ + s = archive_entry_mtime(entry_main); + if (s < 0) { s = 0; } + if (s >= 0x7fffffff) { s = 0x7fffffff; } + archive_entry_set_mtime(pax_attr_entry, s, 0); + + /* Standard ustar doesn't support atime. */ + archive_entry_set_atime(pax_attr_entry, 0, 0); + + /* Standard ustar doesn't support ctime. */ + archive_entry_set_ctime(pax_attr_entry, 0, 0); + + r = __archive_write_format_header_ustar(a, paxbuff, + pax_attr_entry, 'x', 1, NULL); + + archive_entry_free(pax_attr_entry); + + /* Note that the 'x' header shouldn't ever fail to format */ + if (r < ARCHIVE_WARN) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "archive_write_pax_header: " + "'x' header failed?! This can't happen.\n"); + archive_entry_free(entry_main); + archive_string_free(&entry_name); + return (ARCHIVE_FATAL); + } else if (r < ret) + ret = r; + r = __archive_write_output(a, paxbuff, 512); + if (r != ARCHIVE_OK) { + sparse_list_clear(pax); + pax->entry_bytes_remaining = 0; + pax->entry_padding = 0; + archive_entry_free(entry_main); + archive_string_free(&entry_name); + return (ARCHIVE_FATAL); + } + + pax->entry_bytes_remaining = archive_strlen(&(pax->pax_header)); + pax->entry_padding = + 0x1ff & (-(int64_t)pax->entry_bytes_remaining); + + r = __archive_write_output(a, pax->pax_header.s, + archive_strlen(&(pax->pax_header))); + if (r != ARCHIVE_OK) { + /* If a write fails, we're pretty much toast. */ + archive_entry_free(entry_main); + archive_string_free(&entry_name); + return (ARCHIVE_FATAL); + } + /* Pad out the end of the entry. */ + r = __archive_write_nulls(a, (size_t)pax->entry_padding); + if (r != ARCHIVE_OK) { + /* If a write fails, we're pretty much toast. */ + archive_entry_free(entry_main); + archive_string_free(&entry_name); + return (ARCHIVE_FATAL); + } + pax->entry_bytes_remaining = pax->entry_padding = 0; + } + + /* Write the header for main entry. */ + r = __archive_write_output(a, ustarbuff, 512); + if (r != ARCHIVE_OK) { + archive_entry_free(entry_main); + archive_string_free(&entry_name); + return (r); + } + + /* + * Inform the client of the on-disk size we're using, so + * they can avoid unnecessarily writing a body for something + * that we're just going to ignore. + */ + archive_entry_set_size(entry_original, real_size); + if (pax->sparse_list == NULL && real_size > 0) { + /* This is not a sparse file but we handle its data as + * a sparse block. */ + sparse_list_add(pax, 0, real_size); + sparse_total = real_size; + } + pax->entry_padding = 0x1ff & (-(int64_t)sparse_total); + archive_entry_free(entry_main); + archive_string_free(&entry_name); + + return (ret); +} + +/* + * We need a valid name for the regular 'ustar' entry. This routine + * tries to hack something more-or-less reasonable. + * + * The approach here tries to preserve leading dir names. We do so by + * working with four sections: + * 1) "prefix" directory names, + * 2) "suffix" directory names, + * 3) inserted dir name (optional), + * 4) filename. + * + * These sections must satisfy the following requirements: + * * Parts 1 & 2 together form an initial portion of the dir name. + * * Part 3 is specified by the caller. (It should not contain a leading + * or trailing '/'.) + * * Part 4 forms an initial portion of the base filename. + * * The filename must be <= 99 chars to fit the ustar 'name' field. + * * Parts 2, 3, 4 together must be <= 99 chars to fit the ustar 'name' fld. + * * Part 1 must be <= 155 chars to fit the ustar 'prefix' field. + * * If the original name ends in a '/', the new name must also end in a '/' + * * Trailing '/.' sequences may be stripped. + * + * Note: Recall that the ustar format does not store the '/' separating + * parts 1 & 2, but does store the '/' separating parts 2 & 3. + */ +static char * +build_ustar_entry_name(char *dest, const char *src, size_t src_length, + const char *insert) +{ + const char *prefix, *prefix_end; + const char *suffix, *suffix_end; + const char *filename, *filename_end; + char *p; + int need_slash = 0; /* Was there a trailing slash? */ + size_t suffix_length = 99; + size_t insert_length; + + /* Length of additional dir element to be added. */ + if (insert == NULL) + insert_length = 0; + else + /* +2 here allows for '/' before and after the insert. */ + insert_length = strlen(insert) + 2; + + /* Step 0: Quick bailout in a common case. */ + if (src_length < 100 && insert == NULL) { + strncpy(dest, src, src_length); + dest[src_length] = '\0'; + return (dest); + } + + /* Step 1: Locate filename and enforce the length restriction. */ + filename_end = src + src_length; + /* Remove trailing '/' chars and '/.' pairs. */ + for (;;) { + if (filename_end > src && filename_end[-1] == '/') { + filename_end --; + need_slash = 1; /* Remember to restore trailing '/'. */ + continue; + } + if (filename_end > src + 1 && filename_end[-1] == '.' + && filename_end[-2] == '/') { + filename_end -= 2; + need_slash = 1; /* "foo/." will become "foo/" */ + continue; + } + break; + } + if (need_slash) + suffix_length--; + /* Find start of filename. */ + filename = filename_end - 1; + while ((filename > src) && (*filename != '/')) + filename --; + if ((*filename == '/') && (filename < filename_end - 1)) + filename ++; + /* Adjust filename_end so that filename + insert fits in 99 chars. */ + suffix_length -= insert_length; + if (filename_end > filename + suffix_length) + filename_end = filename + suffix_length; + /* Calculate max size for "suffix" section (#3 above). */ + suffix_length -= filename_end - filename; + + /* Step 2: Locate the "prefix" section of the dirname, including + * trailing '/'. */ + prefix = src; + prefix_end = prefix + 155; + if (prefix_end > filename) + prefix_end = filename; + while (prefix_end > prefix && *prefix_end != '/') + prefix_end--; + if ((prefix_end < filename) && (*prefix_end == '/')) + prefix_end++; + + /* Step 3: Locate the "suffix" section of the dirname, + * including trailing '/'. */ + suffix = prefix_end; + suffix_end = suffix + suffix_length; /* Enforce limit. */ + if (suffix_end > filename) + suffix_end = filename; + if (suffix_end < suffix) + suffix_end = suffix; + while (suffix_end > suffix && *suffix_end != '/') + suffix_end--; + if ((suffix_end < filename) && (*suffix_end == '/')) + suffix_end++; + + /* Step 4: Build the new name. */ + /* The OpenBSD strlcpy function is safer, but less portable. */ + /* Rather than maintain two versions, just use the strncpy version. */ + p = dest; + if (prefix_end > prefix) { + strncpy(p, prefix, prefix_end - prefix); + p += prefix_end - prefix; + } + if (suffix_end > suffix) { + strncpy(p, suffix, suffix_end - suffix); + p += suffix_end - suffix; + } + if (insert != NULL) { + /* Note: assume insert does not have leading or trailing '/' */ + strcpy(p, insert); + p += strlen(insert); + *p++ = '/'; + } + strncpy(p, filename, filename_end - filename); + p += filename_end - filename; + if (need_slash) + *p++ = '/'; + *p = '\0'; + + return (dest); +} + +/* + * The ustar header for the pax extended attributes must have a + * reasonable name: SUSv3 requires 'dirname'/PaxHeader.'pid'/'filename' + * where 'pid' is the PID of the archiving process. Unfortunately, + * that makes testing a pain since the output varies for each run, + * so I'm sticking with the simpler 'dirname'/PaxHeader/'filename' + * for now. (Someday, I'll make this settable. Then I can use the + * SUS recommendation as default and test harnesses can override it + * to get predictable results.) + * + * Joerg Schilling has argued that this is unnecessary because, in + * practice, if the pax extended attributes get extracted as regular + * files, no one is going to bother reading those attributes to + * manually restore them. Based on this, 'star' uses + * /tmp/PaxHeader/'basename' as the ustar header name. This is a + * tempting argument, in part because it's simpler than the SUSv3 + * recommendation, but I'm not entirely convinced. I'm also + * uncomfortable with the fact that "/tmp" is a Unix-ism. + * + * The following routine leverages build_ustar_entry_name() above and + * so is simpler than you might think. It just needs to provide the + * additional path element and handle a few pathological cases). + */ +static char * +build_pax_attribute_name(char *dest, const char *src) +{ + char buff[64]; + const char *p; + + /* Handle the null filename case. */ + if (src == NULL || *src == '\0') { + strcpy(dest, "PaxHeader/blank"); + return (dest); + } + + /* Prune final '/' and other unwanted final elements. */ + p = src + strlen(src); + for (;;) { + /* Ends in "/", remove the '/' */ + if (p > src && p[-1] == '/') { + --p; + continue; + } + /* Ends in "/.", remove the '.' */ + if (p > src + 1 && p[-1] == '.' + && p[-2] == '/') { + --p; + continue; + } + break; + } + + /* Pathological case: After above, there was nothing left. + * This includes "/." "/./." "/.//./." etc. */ + if (p == src) { + strcpy(dest, "/PaxHeader/rootdir"); + return (dest); + } + + /* Convert unadorned "." into a suitable filename. */ + if (*src == '.' && p == src + 1) { + strcpy(dest, "PaxHeader/currentdir"); + return (dest); + } + + /* + * TODO: Push this string into the 'pax' structure to avoid + * recomputing it every time. That will also open the door + * to having clients override it. + */ +#if HAVE_GETPID && 0 /* Disable this for now; see above comment. */ + sprintf(buff, "PaxHeader.%d", getpid()); +#else + /* If the platform can't fetch the pid, don't include it. */ + strcpy(buff, "PaxHeader"); +#endif + /* General case: build a ustar-compatible name adding + * "/PaxHeader/". */ + build_ustar_entry_name(dest, src, p - src, buff); + + return (dest); +} + +/* + * GNU PAX Format 1.0 requires the special name, which pattern is: + * /GNUSparseFile./ + * + * Since reproducible archives are more important, use 0 as pid. + * + * This function is used for only Sparse file, a file type of which + * is regular file. + */ +static char * +build_gnu_sparse_name(char *dest, const char *src) +{ + const char *p; + + /* Handle the null filename case. */ + if (src == NULL || *src == '\0') { + strcpy(dest, "GNUSparseFile/blank"); + return (dest); + } + + /* Prune final '/' and other unwanted final elements. */ + p = src + strlen(src); + for (;;) { + /* Ends in "/", remove the '/' */ + if (p > src && p[-1] == '/') { + --p; + continue; + } + /* Ends in "/.", remove the '.' */ + if (p > src + 1 && p[-1] == '.' + && p[-2] == '/') { + --p; + continue; + } + break; + } + + /* General case: build a ustar-compatible name adding + * "/GNUSparseFile/". */ + build_ustar_entry_name(dest, src, p - src, "GNUSparseFile.0"); + + return (dest); +} + +/* Write two null blocks for the end of archive */ +static int +archive_write_pax_close(struct archive_write *a) +{ + return (__archive_write_nulls(a, 512 * 2)); +} + +static int +archive_write_pax_free(struct archive_write *a) +{ + struct pax *pax; + + pax = (struct pax *)a->format_data; + if (pax == NULL) + return (ARCHIVE_OK); + + archive_string_free(&pax->pax_header); + archive_string_free(&pax->sparse_map); + archive_string_free(&pax->l_url_encoded_name); + sparse_list_clear(pax); + free(pax); + a->format_data = NULL; + return (ARCHIVE_OK); +} + +static int +archive_write_pax_finish_entry(struct archive_write *a) +{ + struct pax *pax; + uint64_t remaining; + int ret; + + pax = (struct pax *)a->format_data; + remaining = pax->entry_bytes_remaining; + if (remaining == 0) { + while (pax->sparse_list) { + struct sparse_block *sb; + if (!pax->sparse_list->is_hole) + remaining += pax->sparse_list->remaining; + sb = pax->sparse_list->next; + free(pax->sparse_list); + pax->sparse_list = sb; + } + } + ret = __archive_write_nulls(a, (size_t)(remaining + pax->entry_padding)); + pax->entry_bytes_remaining = pax->entry_padding = 0; + return (ret); +} + +static ssize_t +archive_write_pax_data(struct archive_write *a, const void *buff, size_t s) +{ + struct pax *pax; + size_t ws; + size_t total; + int ret; + + pax = (struct pax *)a->format_data; + + /* + * According to GNU PAX format 1.0, write a sparse map + * before the body. + */ + if (archive_strlen(&(pax->sparse_map))) { + ret = __archive_write_output(a, pax->sparse_map.s, + archive_strlen(&(pax->sparse_map))); + if (ret != ARCHIVE_OK) + return (ret); + ret = __archive_write_nulls(a, pax->sparse_map_padding); + if (ret != ARCHIVE_OK) + return (ret); + archive_string_empty(&(pax->sparse_map)); + } + + total = 0; + while (total < s) { + const unsigned char *p; + + while (pax->sparse_list != NULL && + pax->sparse_list->remaining == 0) { + struct sparse_block *sb = pax->sparse_list->next; + free(pax->sparse_list); + pax->sparse_list = sb; + } + + if (pax->sparse_list == NULL) + return (total); + + p = ((const unsigned char *)buff) + total; + ws = s - total; + if (ws > pax->sparse_list->remaining) + ws = (size_t)pax->sparse_list->remaining; + + if (pax->sparse_list->is_hole) { + /* Current block is hole thus we do not write + * the body. */ + pax->sparse_list->remaining -= ws; + total += ws; + continue; + } + + ret = __archive_write_output(a, p, ws); + pax->sparse_list->remaining -= ws; + total += ws; + if (ret != ARCHIVE_OK) + return (ret); + } + return (total); +} + +static int +has_non_ASCII(const char *_p) +{ + const unsigned char *p = (const unsigned char *)_p; + + if (p == NULL) + return (1); + while (*p != '\0' && *p < 128) + p++; + return (*p != '\0'); +} + +/* + * Used by extended attribute support; encodes the name + * so that there will be no '=' characters in the result. + */ +static char * +url_encode(const char *in) +{ + const char *s; + char *d; + int out_len = 0; + char *out; + + for (s = in; *s != '\0'; s++) { + if (*s < 33 || *s > 126 || *s == '%' || *s == '=') + out_len += 3; + else + out_len++; + } + + out = (char *)malloc(out_len + 1); + if (out == NULL) + return (NULL); + + for (s = in, d = out; *s != '\0'; s++) { + /* encode any non-printable ASCII character or '%' or '=' */ + if (*s < 33 || *s > 126 || *s == '%' || *s == '=') { + /* URL encoding is '%' followed by two hex digits */ + *d++ = '%'; + *d++ = "0123456789ABCDEF"[0x0f & (*s >> 4)]; + *d++ = "0123456789ABCDEF"[0x0f & *s]; + } else { + *d++ = *s; + } + } + *d = '\0'; + return (out); +} + +/* + * Encode a sequence of bytes into a C string using base-64 encoding. + * + * Returns a null-terminated C string allocated with malloc(); caller + * is responsible for freeing the result. + */ +static char * +base64_encode(const char *s, size_t len) +{ + static const char digits[64] = + { 'A','B','C','D','E','F','G','H','I','J','K','L','M','N','O', + 'P','Q','R','S','T','U','V','W','X','Y','Z','a','b','c','d', + 'e','f','g','h','i','j','k','l','m','n','o','p','q','r','s', + 't','u','v','w','x','y','z','0','1','2','3','4','5','6','7', + '8','9','+','/' }; + int v; + char *d, *out; + + /* 3 bytes becomes 4 chars, but round up and allow for trailing NUL */ + out = (char *)malloc((len * 4 + 2) / 3 + 1); + if (out == NULL) + return (NULL); + d = out; + + /* Convert each group of 3 bytes into 4 characters. */ + while (len >= 3) { + v = (((int)s[0] << 16) & 0xff0000) + | (((int)s[1] << 8) & 0xff00) + | (((int)s[2]) & 0x00ff); + s += 3; + len -= 3; + *d++ = digits[(v >> 18) & 0x3f]; + *d++ = digits[(v >> 12) & 0x3f]; + *d++ = digits[(v >> 6) & 0x3f]; + *d++ = digits[(v) & 0x3f]; + } + /* Handle final group of 1 byte (2 chars) or 2 bytes (3 chars). */ + switch (len) { + case 0: break; + case 1: + v = (((int)s[0] << 16) & 0xff0000); + *d++ = digits[(v >> 18) & 0x3f]; + *d++ = digits[(v >> 12) & 0x3f]; + break; + case 2: + v = (((int)s[0] << 16) & 0xff0000) + | (((int)s[1] << 8) & 0xff00); + *d++ = digits[(v >> 18) & 0x3f]; + *d++ = digits[(v >> 12) & 0x3f]; + *d++ = digits[(v >> 6) & 0x3f]; + break; + } + /* Add trailing NUL character so output is a valid C string. */ + *d = '\0'; + return (out); +} + +static void +sparse_list_clear(struct pax *pax) +{ + while (pax->sparse_list != NULL) { + struct sparse_block *sb = pax->sparse_list; + pax->sparse_list = sb->next; + free(sb); + } + pax->sparse_tail = NULL; +} + +static int +_sparse_list_add_block(struct pax *pax, int64_t offset, int64_t length, + int is_hole) +{ + struct sparse_block *sb; + + sb = (struct sparse_block *)malloc(sizeof(*sb)); + if (sb == NULL) + return (ARCHIVE_FATAL); + sb->next = NULL; + sb->is_hole = is_hole; + sb->offset = offset; + sb->remaining = length; + if (pax->sparse_list == NULL || pax->sparse_tail == NULL) + pax->sparse_list = pax->sparse_tail = sb; + else { + pax->sparse_tail->next = sb; + pax->sparse_tail = sb; + } + return (ARCHIVE_OK); +} + +static int +sparse_list_add(struct pax *pax, int64_t offset, int64_t length) +{ + int64_t last_offset; + int r; + + if (pax->sparse_tail == NULL) + last_offset = 0; + else { + last_offset = pax->sparse_tail->offset + + pax->sparse_tail->remaining; + } + if (last_offset < offset) { + /* Add a hole block. */ + r = _sparse_list_add_block(pax, last_offset, + offset - last_offset, 1); + if (r != ARCHIVE_OK) + return (r); + } + /* Add data block. */ + return (_sparse_list_add_block(pax, offset, length, 0)); +} + diff --git a/c/archive_write_set_format_private.h b/c/archive_write_set_format_private.h new file mode 100644 --- /dev/null +++ b/c/archive_write_set_format_private.h @@ -0,0 +1,42 @@ +/*- + * Copyright (c) 2020 Martin Matuska + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD$ + */ + +#ifndef ARCHIVE_WRITE_SET_FORMAT_PRIVATE_H_INCLUDED +#define ARCHIVE_WRITE_SET_FORMAT_PRIVATE_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#ifndef __LIBARCHIVE_TEST +#error This header is only to be used internally to libarchive. +#endif +#endif + +#include "archive.h" +#include "archive_entry.h" + +void __archive_write_entry_filetype_unsupported(struct archive *a, + struct archive_entry *entry, const char *format); +#endif diff --git a/c/archive_write_set_format_raw.c b/c/archive_write_set_format_raw.c new file mode 100644 --- /dev/null +++ b/c/archive_write_set_format_raw.c @@ -0,0 +1,125 @@ +/*- + * Copyright (c) 2013 Marek Kubica + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +#ifdef HAVE_ERRNO_H +#include +#endif + +#include "archive_entry.h" +#include "archive_write_private.h" + +static ssize_t archive_write_raw_data(struct archive_write *, + const void *buff, size_t s); +static int archive_write_raw_free(struct archive_write *); +static int archive_write_raw_header(struct archive_write *, + struct archive_entry *); + +struct raw { + int entries_written; +}; + +/* + * Set output format to 'raw' format. + */ +int +archive_write_set_format_raw(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct raw *raw; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_set_format_raw"); + + /* If someone else was already registered, unregister them. */ + if (a->format_free != NULL) + (a->format_free)(a); + + raw = (struct raw *)calloc(1, sizeof(*raw)); + if (raw == NULL) { + archive_set_error(&a->archive, ENOMEM, "Can't allocate raw data"); + return (ARCHIVE_FATAL); + } + raw->entries_written = 0; + a->format_data = raw; + a->format_name = "raw"; + /* no options exist for this format */ + a->format_options = NULL; + a->format_write_header = archive_write_raw_header; + a->format_write_data = archive_write_raw_data; + a->format_finish_entry = NULL; + /* nothing needs to be done on closing */ + a->format_close = NULL; + a->format_free = archive_write_raw_free; + a->archive.archive_format = ARCHIVE_FORMAT_RAW; + a->archive.archive_format_name = "RAW"; + return (ARCHIVE_OK); +} + +static int +archive_write_raw_header(struct archive_write *a, struct archive_entry *entry) +{ + struct raw *raw = (struct raw *)a->format_data; + + if (archive_entry_filetype(entry) != AE_IFREG) { + archive_set_error(&a->archive, ERANGE, + "Raw format only supports filetype AE_IFREG"); + return (ARCHIVE_FATAL); + } + + + if (raw->entries_written > 0) { + archive_set_error(&a->archive, ERANGE, + "Raw format only supports one entry per archive"); + return (ARCHIVE_FATAL); + } + raw->entries_written++; + + return (ARCHIVE_OK); +} + +static ssize_t +archive_write_raw_data(struct archive_write *a, const void *buff, size_t s) +{ + int ret; + + ret = __archive_write_output(a, buff, s); + if (ret >= 0) + return (s); + else + return (ret); +} + +static int +archive_write_raw_free(struct archive_write *a) +{ + struct raw *raw; + + raw = (struct raw *)a->format_data; + free(raw); + a->format_data = NULL; + return (ARCHIVE_OK); +} diff --git a/c/archive_write_set_format_shar.c b/c/archive_write_set_format_shar.c new file mode 100644 --- /dev/null +++ b/c/archive_write_set_format_shar.c @@ -0,0 +1,641 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * Copyright (c) 2008 Joerg Sonnenberger + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_format_shar.c 189438 2009-03-06 05:58:56Z kientzle $"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_private.h" +#include "archive_write_private.h" +#include "archive_write_set_format_private.h" + +struct shar { + int dump; + int end_of_line; + struct archive_entry *entry; + int has_data; + char *last_dir; + + /* Line buffer for uuencoded dump format */ + char outbuff[45]; + size_t outpos; + + int wrote_header; + struct archive_string work; + struct archive_string quoted_name; +}; + +static int archive_write_shar_close(struct archive_write *); +static int archive_write_shar_free(struct archive_write *); +static int archive_write_shar_header(struct archive_write *, + struct archive_entry *); +static ssize_t archive_write_shar_data_sed(struct archive_write *, + const void * buff, size_t); +static ssize_t archive_write_shar_data_uuencode(struct archive_write *, + const void * buff, size_t); +static int archive_write_shar_finish_entry(struct archive_write *); + +/* + * Copy the given string to the buffer, quoting all shell meta characters + * found. + */ +static void +shar_quote(struct archive_string *buf, const char *str, int in_shell) +{ + static const char meta[] = "\n \t'`\";&<>()|*?{}[]\\$!#^~"; + size_t len; + + while (*str != '\0') { + if ((len = strcspn(str, meta)) != 0) { + archive_strncat(buf, str, len); + str += len; + } else if (*str == '\n') { + if (in_shell) + archive_strcat(buf, "\"\n\""); + else + archive_strcat(buf, "\\n"); + ++str; + } else { + archive_strappend_char(buf, '\\'); + archive_strappend_char(buf, *str); + ++str; + } + } +} + +/* + * Set output format to 'shar' format. + */ +int +archive_write_set_format_shar(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct shar *shar; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_set_format_shar"); + + /* If someone else was already registered, unregister them. */ + if (a->format_free != NULL) + (a->format_free)(a); + + shar = (struct shar *)calloc(1, sizeof(*shar)); + if (shar == NULL) { + archive_set_error(&a->archive, ENOMEM, "Can't allocate shar data"); + return (ARCHIVE_FATAL); + } + archive_string_init(&shar->work); + archive_string_init(&shar->quoted_name); + a->format_data = shar; + a->format_name = "shar"; + a->format_write_header = archive_write_shar_header; + a->format_close = archive_write_shar_close; + a->format_free = archive_write_shar_free; + a->format_write_data = archive_write_shar_data_sed; + a->format_finish_entry = archive_write_shar_finish_entry; + a->archive.archive_format = ARCHIVE_FORMAT_SHAR_BASE; + a->archive.archive_format_name = "shar"; + return (ARCHIVE_OK); +} + +/* + * An alternate 'shar' that uses uudecode instead of 'sed' to encode + * file contents and can therefore be used to archive binary files. + * In addition, this variant also attempts to restore ownership, file modes, + * and other extended file information. + */ +int +archive_write_set_format_shar_dump(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct shar *shar; + + archive_write_set_format_shar(&a->archive); + shar = (struct shar *)a->format_data; + shar->dump = 1; + a->format_write_data = archive_write_shar_data_uuencode; + a->archive.archive_format = ARCHIVE_FORMAT_SHAR_DUMP; + a->archive.archive_format_name = "shar dump"; + return (ARCHIVE_OK); +} + +static int +archive_write_shar_header(struct archive_write *a, struct archive_entry *entry) +{ + const char *linkname; + const char *name; + char *p, *pp; + struct shar *shar; + + shar = (struct shar *)a->format_data; + if (!shar->wrote_header) { + archive_strcat(&shar->work, "#!/bin/sh\n"); + archive_strcat(&shar->work, "# This is a shell archive\n"); + shar->wrote_header = 1; + } + + /* Save the entry for the closing. */ + archive_entry_free(shar->entry); + shar->entry = archive_entry_clone(entry); + name = archive_entry_pathname(entry); + + /* Handle some preparatory issues. */ + switch(archive_entry_filetype(entry)) { + case AE_IFREG: + /* Only regular files have non-zero size. */ + break; + case AE_IFDIR: + archive_entry_set_size(entry, 0); + /* Don't bother trying to recreate '.' */ + if (strcmp(name, ".") == 0 || strcmp(name, "./") == 0) + return (ARCHIVE_OK); + break; + case AE_IFIFO: + case AE_IFCHR: + case AE_IFBLK: + /* All other file types have zero size in the archive. */ + archive_entry_set_size(entry, 0); + break; + default: + archive_entry_set_size(entry, 0); + if (archive_entry_hardlink(entry) == NULL && + archive_entry_symlink(entry) == NULL) { + __archive_write_entry_filetype_unsupported( + &a->archive, entry, "shar"); + return (ARCHIVE_WARN); + } + } + + archive_string_empty(&shar->quoted_name); + shar_quote(&shar->quoted_name, name, 1); + + /* Stock preparation for all file types. */ + archive_string_sprintf(&shar->work, "echo x %s\n", shar->quoted_name.s); + + if (archive_entry_filetype(entry) != AE_IFDIR) { + /* Try to create the dir. */ + p = strdup(name); + pp = strrchr(p, '/'); + /* If there is a / character, try to create the dir. */ + if (pp != NULL) { + *pp = '\0'; + + /* Try to avoid a lot of redundant mkdir commands. */ + if (strcmp(p, ".") == 0) { + /* Don't try to "mkdir ." */ + free(p); + } else if (shar->last_dir == NULL) { + archive_strcat(&shar->work, "mkdir -p "); + shar_quote(&shar->work, p, 1); + archive_strcat(&shar->work, + " > /dev/null 2>&1\n"); + shar->last_dir = p; + } else if (strcmp(p, shar->last_dir) == 0) { + /* We've already created this exact dir. */ + free(p); + } else if (strlen(p) < strlen(shar->last_dir) && + strncmp(p, shar->last_dir, strlen(p)) == 0) { + /* We've already created a subdir. */ + free(p); + } else { + archive_strcat(&shar->work, "mkdir -p "); + shar_quote(&shar->work, p, 1); + archive_strcat(&shar->work, + " > /dev/null 2>&1\n"); + shar->last_dir = p; + } + } else { + free(p); + } + } + + /* Handle file-type specific issues. */ + shar->has_data = 0; + if ((linkname = archive_entry_hardlink(entry)) != NULL) { + archive_strcat(&shar->work, "ln -f "); + shar_quote(&shar->work, linkname, 1); + archive_string_sprintf(&shar->work, " %s\n", + shar->quoted_name.s); + } else if ((linkname = archive_entry_symlink(entry)) != NULL) { + archive_strcat(&shar->work, "ln -fs "); + shar_quote(&shar->work, linkname, 1); + archive_string_sprintf(&shar->work, " %s\n", + shar->quoted_name.s); + } else { + switch(archive_entry_filetype(entry)) { + case AE_IFREG: + if (archive_entry_size(entry) == 0) { + /* More portable than "touch." */ + archive_string_sprintf(&shar->work, + "test -e \"%s\" || :> \"%s\"\n", + shar->quoted_name.s, shar->quoted_name.s); + } else { + if (shar->dump) { + unsigned int mode = archive_entry_mode(entry) & 0777; + archive_string_sprintf(&shar->work, + "uudecode -p > %s << 'SHAR_END'\n", + shar->quoted_name.s); + archive_string_sprintf(&shar->work, + "begin %o ", mode); + shar_quote(&shar->work, name, 0); + archive_strcat(&shar->work, "\n"); + } else { + archive_string_sprintf(&shar->work, + "sed 's/^X//' > %s << 'SHAR_END'\n", + shar->quoted_name.s); + } + shar->has_data = 1; + shar->end_of_line = 1; + shar->outpos = 0; + } + break; + case AE_IFDIR: + archive_string_sprintf(&shar->work, + "mkdir -p %s > /dev/null 2>&1\n", + shar->quoted_name.s); + /* Record that we just created this directory. */ + free(shar->last_dir); + + shar->last_dir = strdup(name); + /* Trim a trailing '/'. */ + pp = strrchr(shar->last_dir, '/'); + if (pp != NULL && pp[1] == '\0') + *pp = '\0'; + /* + * TODO: Put dir name/mode on a list to be fixed + * up at end of archive. + */ + break; + case AE_IFIFO: + archive_string_sprintf(&shar->work, + "mkfifo %s\n", shar->quoted_name.s); + break; + case AE_IFCHR: + archive_string_sprintf(&shar->work, + "mknod %s c %ju %ju\n", shar->quoted_name.s, + (uintmax_t)archive_entry_rdevmajor(entry), + (uintmax_t)archive_entry_rdevminor(entry)); + break; + case AE_IFBLK: + archive_string_sprintf(&shar->work, + "mknod %s b %ju %ju\n", shar->quoted_name.s, + (uintmax_t)archive_entry_rdevmajor(entry), + (uintmax_t)archive_entry_rdevminor(entry)); + break; + default: + return (ARCHIVE_WARN); + } + } + + return (ARCHIVE_OK); +} + +static ssize_t +archive_write_shar_data_sed(struct archive_write *a, const void *buff, size_t n) +{ + static const size_t ensured = 65533; + struct shar *shar; + const char *src; + char *buf, *buf_end; + int ret; + size_t written = n; + + shar = (struct shar *)a->format_data; + if (!shar->has_data || n == 0) + return (0); + + src = (const char *)buff; + + /* + * ensure is the number of bytes in buffer before expanding the + * current character. Each operation writes the current character + * and optionally the start-of-new-line marker. This can happen + * twice before entering the loop, so make sure three additional + * bytes can be written. + */ + if (archive_string_ensure(&shar->work, ensured + 3) == NULL) { + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + + if (shar->work.length > ensured) { + ret = __archive_write_output(a, shar->work.s, + shar->work.length); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + archive_string_empty(&shar->work); + } + buf = shar->work.s + shar->work.length; + buf_end = shar->work.s + ensured; + + if (shar->end_of_line) { + *buf++ = 'X'; + shar->end_of_line = 0; + } + + while (n-- != 0) { + if ((*buf++ = *src++) == '\n') { + if (n == 0) + shar->end_of_line = 1; + else + *buf++ = 'X'; + } + + if (buf >= buf_end) { + shar->work.length = buf - shar->work.s; + ret = __archive_write_output(a, shar->work.s, + shar->work.length); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + archive_string_empty(&shar->work); + buf = shar->work.s; + } + } + + shar->work.length = buf - shar->work.s; + + return (written); +} + +#define UUENC(c) (((c)!=0) ? ((c) & 077) + ' ': '`') + +static void +uuencode_group(const char _in[3], char out[4]) +{ + const unsigned char *in = (const unsigned char *)_in; + int t; + + t = (in[0] << 16) | (in[1] << 8) | in[2]; + out[0] = UUENC( 0x3f & (t >> 18) ); + out[1] = UUENC( 0x3f & (t >> 12) ); + out[2] = UUENC( 0x3f & (t >> 6) ); + out[3] = UUENC( 0x3f & t ); +} + +static int +_uuencode_line(struct archive_write *a, struct shar *shar, const char *inbuf, size_t len) +{ + char *buf; + size_t alloc_len; + + /* len <= 45 -> expanded to 60 + len byte + new line */ + alloc_len = shar->work.length + 62; + if (archive_string_ensure(&shar->work, alloc_len) == NULL) { + archive_set_error(&a->archive, ENOMEM, "Out of memory"); + return (ARCHIVE_FATAL); + } + + buf = shar->work.s + shar->work.length; + *buf++ = UUENC(len); + while (len >= 3) { + uuencode_group(inbuf, buf); + len -= 3; + inbuf += 3; + buf += 4; + } + if (len != 0) { + char tmp_buf[3]; + tmp_buf[0] = inbuf[0]; + if (len == 1) + tmp_buf[1] = '\0'; + else + tmp_buf[1] = inbuf[1]; + tmp_buf[2] = '\0'; + uuencode_group(tmp_buf, buf); + buf += 4; + } + *buf++ = '\n'; + if ((buf - shar->work.s) > (ptrdiff_t)(shar->work.length + 62)) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, "Buffer overflow"); + return (ARCHIVE_FATAL); + } + shar->work.length = buf - shar->work.s; + return (ARCHIVE_OK); +} + +#define uuencode_line(__a, __shar, __inbuf, __len) \ + do { \ + int r = _uuencode_line(__a, __shar, __inbuf, __len); \ + if (r != ARCHIVE_OK) \ + return (ARCHIVE_FATAL); \ + } while (0) + +static ssize_t +archive_write_shar_data_uuencode(struct archive_write *a, const void *buff, + size_t length) +{ + struct shar *shar; + const char *src; + size_t n; + int ret; + + shar = (struct shar *)a->format_data; + if (!shar->has_data) + return (ARCHIVE_OK); + src = (const char *)buff; + + if (shar->outpos != 0) { + n = 45 - shar->outpos; + if (n > length) + n = length; + memcpy(shar->outbuff + shar->outpos, src, n); + if (shar->outpos + n < 45) { + shar->outpos += n; + return length; + } + uuencode_line(a, shar, shar->outbuff, 45); + src += n; + n = length - n; + } else { + n = length; + } + + while (n >= 45) { + uuencode_line(a, shar, src, 45); + src += 45; + n -= 45; + + if (shar->work.length < 65536) + continue; + ret = __archive_write_output(a, shar->work.s, + shar->work.length); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + archive_string_empty(&shar->work); + } + if (n != 0) { + memcpy(shar->outbuff, src, n); + shar->outpos = n; + } + return (length); +} + +static int +archive_write_shar_finish_entry(struct archive_write *a) +{ + const char *g, *p, *u; + struct shar *shar; + int ret; + + shar = (struct shar *)a->format_data; + if (shar->entry == NULL) + return (0); + + if (shar->dump) { + /* Finish uuencoded data. */ + if (shar->has_data) { + if (shar->outpos > 0) + uuencode_line(a, shar, shar->outbuff, + shar->outpos); + archive_strcat(&shar->work, "`\nend\n"); + archive_strcat(&shar->work, "SHAR_END\n"); + } + /* Restore file mode, owner, flags. */ + /* + * TODO: Don't immediately restore mode for + * directories; defer that to end of script. + */ + archive_string_sprintf(&shar->work, "chmod %o ", + (unsigned int)(archive_entry_mode(shar->entry) & 07777)); + shar_quote(&shar->work, archive_entry_pathname(shar->entry), 1); + archive_strcat(&shar->work, "\n"); + + u = archive_entry_uname(shar->entry); + g = archive_entry_gname(shar->entry); + if (u != NULL || g != NULL) { + archive_strcat(&shar->work, "chown "); + if (u != NULL) + shar_quote(&shar->work, u, 1); + if (g != NULL) { + archive_strcat(&shar->work, ":"); + shar_quote(&shar->work, g, 1); + } + archive_strcat(&shar->work, " "); + shar_quote(&shar->work, + archive_entry_pathname(shar->entry), 1); + archive_strcat(&shar->work, "\n"); + } + + if ((p = archive_entry_fflags_text(shar->entry)) != NULL) { + archive_string_sprintf(&shar->work, "chflags %s ", p); + shar_quote(&shar->work, + archive_entry_pathname(shar->entry), 1); + archive_strcat(&shar->work, "\n"); + } + + /* TODO: restore ACLs */ + + } else { + if (shar->has_data) { + /* Finish sed-encoded data: ensure last line ends. */ + if (!shar->end_of_line) + archive_strappend_char(&shar->work, '\n'); + archive_strcat(&shar->work, "SHAR_END\n"); + } + } + + archive_entry_free(shar->entry); + shar->entry = NULL; + + if (shar->work.length < 65536) + return (ARCHIVE_OK); + + ret = __archive_write_output(a, shar->work.s, shar->work.length); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + archive_string_empty(&shar->work); + + return (ARCHIVE_OK); +} + +static int +archive_write_shar_close(struct archive_write *a) +{ + struct shar *shar; + int ret; + + /* + * TODO: Accumulate list of directory names/modes and + * fix them all up at end-of-archive. + */ + + shar = (struct shar *)a->format_data; + + /* + * Only write the end-of-archive markers if the archive was + * actually started. This avoids problems if someone sets + * shar format, then sets another format (which would invoke + * shar_finish to free the format-specific data). + */ + if (shar->wrote_header == 0) + return (ARCHIVE_OK); + + archive_strcat(&shar->work, "exit\n"); + + ret = __archive_write_output(a, shar->work.s, shar->work.length); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + + /* Shar output is never padded. */ + archive_write_set_bytes_in_last_block(&a->archive, 1); + /* + * TODO: shar should also suppress padding of + * uncompressed data within gzip/bzip2 streams. + */ + + return (ARCHIVE_OK); +} + +static int +archive_write_shar_free(struct archive_write *a) +{ + struct shar *shar; + + shar = (struct shar *)a->format_data; + if (shar == NULL) + return (ARCHIVE_OK); + + archive_entry_free(shar->entry); + free(shar->last_dir); + archive_string_free(&(shar->work)); + archive_string_free(&(shar->quoted_name)); + free(shar); + a->format_data = NULL; + return (ARCHIVE_OK); +} diff --git a/c/archive_write_set_format_ustar.c b/c/archive_write_set_format_ustar.c new file mode 100644 --- /dev/null +++ b/c/archive_write_set_format_ustar.c @@ -0,0 +1,758 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * Copyright (c) 2011-2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_format_ustar.c 191579 2009-04-27 18:35:03Z kientzle $"); + + +#ifdef HAVE_ERRNO_H +#include +#endif +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_private.h" +#include "archive_write_private.h" +#include "archive_write_set_format_private.h" + +struct ustar { + uint64_t entry_bytes_remaining; + uint64_t entry_padding; + + struct archive_string_conv *opt_sconv; + struct archive_string_conv *sconv_default; + int init_default_conversion; +}; + +/* + * Define structure of POSIX 'ustar' tar header. + */ +#define USTAR_name_offset 0 +#define USTAR_name_size 100 +#define USTAR_mode_offset 100 +#define USTAR_mode_size 6 +#define USTAR_mode_max_size 8 +#define USTAR_uid_offset 108 +#define USTAR_uid_size 6 +#define USTAR_uid_max_size 8 +#define USTAR_gid_offset 116 +#define USTAR_gid_size 6 +#define USTAR_gid_max_size 8 +#define USTAR_size_offset 124 +#define USTAR_size_size 11 +#define USTAR_size_max_size 12 +#define USTAR_mtime_offset 136 +#define USTAR_mtime_size 11 +#define USTAR_mtime_max_size 11 +#define USTAR_checksum_offset 148 +#define USTAR_checksum_size 8 +#define USTAR_typeflag_offset 156 +#define USTAR_typeflag_size 1 +#define USTAR_linkname_offset 157 +#define USTAR_linkname_size 100 +#define USTAR_magic_offset 257 +#define USTAR_magic_size 6 +#define USTAR_version_offset 263 +#define USTAR_version_size 2 +#define USTAR_uname_offset 265 +#define USTAR_uname_size 32 +#define USTAR_gname_offset 297 +#define USTAR_gname_size 32 +#define USTAR_rdevmajor_offset 329 +#define USTAR_rdevmajor_size 6 +#define USTAR_rdevmajor_max_size 8 +#define USTAR_rdevminor_offset 337 +#define USTAR_rdevminor_size 6 +#define USTAR_rdevminor_max_size 8 +#define USTAR_prefix_offset 345 +#define USTAR_prefix_size 155 +#define USTAR_padding_offset 500 +#define USTAR_padding_size 12 + +/* + * A filled-in copy of the header for initialization. + */ +static const char template_header[] = { + /* name: 100 bytes */ + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0, + /* Mode, space-null termination: 8 bytes */ + '0','0','0','0','0','0', ' ','\0', + /* uid, space-null termination: 8 bytes */ + '0','0','0','0','0','0', ' ','\0', + /* gid, space-null termination: 8 bytes */ + '0','0','0','0','0','0', ' ','\0', + /* size, space termination: 12 bytes */ + '0','0','0','0','0','0','0','0','0','0','0', ' ', + /* mtime, space termination: 12 bytes */ + '0','0','0','0','0','0','0','0','0','0','0', ' ', + /* Initial checksum value: 8 spaces */ + ' ',' ',' ',' ',' ',' ',' ',' ', + /* Typeflag: 1 byte */ + '0', /* '0' = regular file */ + /* Linkname: 100 bytes */ + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0, + /* Magic: 6 bytes, Version: 2 bytes */ + 'u','s','t','a','r','\0', '0','0', + /* Uname: 32 bytes */ + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + /* Gname: 32 bytes */ + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + /* rdevmajor + space/null padding: 8 bytes */ + '0','0','0','0','0','0', ' ','\0', + /* rdevminor + space/null padding: 8 bytes */ + '0','0','0','0','0','0', ' ','\0', + /* Prefix: 155 bytes */ + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0, + /* Padding: 12 bytes */ + 0,0,0,0,0,0,0,0, 0,0,0,0 +}; + +static ssize_t archive_write_ustar_data(struct archive_write *a, const void *buff, + size_t s); +static int archive_write_ustar_free(struct archive_write *); +static int archive_write_ustar_close(struct archive_write *); +static int archive_write_ustar_finish_entry(struct archive_write *); +static int archive_write_ustar_header(struct archive_write *, + struct archive_entry *entry); +static int archive_write_ustar_options(struct archive_write *, + const char *, const char *); +static int format_256(int64_t, char *, int); +static int format_number(int64_t, char *, int size, int max, int strict); +static int format_octal(int64_t, char *, int); + +/* + * Set output format to 'ustar' format. + */ +int +archive_write_set_format_ustar(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct ustar *ustar; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_set_format_ustar"); + + /* If someone else was already registered, unregister them. */ + if (a->format_free != NULL) + (a->format_free)(a); + + /* Basic internal sanity test. */ + if (sizeof(template_header) != 512) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Internal: template_header wrong size: %zu should be 512", + sizeof(template_header)); + return (ARCHIVE_FATAL); + } + + ustar = (struct ustar *)calloc(1, sizeof(*ustar)); + if (ustar == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate ustar data"); + return (ARCHIVE_FATAL); + } + a->format_data = ustar; + a->format_name = "ustar"; + a->format_options = archive_write_ustar_options; + a->format_write_header = archive_write_ustar_header; + a->format_write_data = archive_write_ustar_data; + a->format_close = archive_write_ustar_close; + a->format_free = archive_write_ustar_free; + a->format_finish_entry = archive_write_ustar_finish_entry; + a->archive.archive_format = ARCHIVE_FORMAT_TAR_USTAR; + a->archive.archive_format_name = "POSIX ustar"; + return (ARCHIVE_OK); +} + +static int +archive_write_ustar_options(struct archive_write *a, const char *key, + const char *val) +{ + struct ustar *ustar = (struct ustar *)a->format_data; + int ret = ARCHIVE_FAILED; + + if (strcmp(key, "hdrcharset") == 0) { + if (val == NULL || val[0] == 0) + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "%s: hdrcharset option needs a character-set name", + a->format_name); + else { + ustar->opt_sconv = archive_string_conversion_to_charset( + &a->archive, val, 0); + if (ustar->opt_sconv != NULL) + ret = ARCHIVE_OK; + else + ret = ARCHIVE_FATAL; + } + return (ret); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +static int +archive_write_ustar_header(struct archive_write *a, struct archive_entry *entry) +{ + char buff[512]; + int ret, ret2; + struct ustar *ustar; + struct archive_entry *entry_main; + struct archive_string_conv *sconv; + + ustar = (struct ustar *)a->format_data; + + /* Setup default string conversion. */ + if (ustar->opt_sconv == NULL) { + if (!ustar->init_default_conversion) { + ustar->sconv_default = + archive_string_default_conversion_for_write(&(a->archive)); + ustar->init_default_conversion = 1; + } + sconv = ustar->sconv_default; + } else + sconv = ustar->opt_sconv; + + /* Sanity check. */ + if (archive_entry_pathname(entry) == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Can't record entry in tar file without pathname"); + return (ARCHIVE_FAILED); + } + + /* Only regular files (not hardlinks) have data. */ + if (archive_entry_hardlink(entry) != NULL || + archive_entry_symlink(entry) != NULL || + !(archive_entry_filetype(entry) == AE_IFREG)) + archive_entry_set_size(entry, 0); + + if (AE_IFDIR == archive_entry_filetype(entry)) { + const char *p; + size_t path_length; + /* + * Ensure a trailing '/'. Modify the entry so + * the client sees the change. + */ +#if defined(_WIN32) && !defined(__CYGWIN__) + const wchar_t *wp; + + wp = archive_entry_pathname_w(entry); + if (wp != NULL && wp[wcslen(wp) -1] != L'/') { + struct archive_wstring ws; + + archive_string_init(&ws); + path_length = wcslen(wp); + if (archive_wstring_ensure(&ws, + path_length + 2) == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate ustar data"); + archive_wstring_free(&ws); + return(ARCHIVE_FATAL); + } + /* Should we keep '\' ? */ + if (wp[path_length -1] == L'\\') + path_length--; + archive_wstrncpy(&ws, wp, path_length); + archive_wstrappend_wchar(&ws, L'/'); + archive_entry_copy_pathname_w(entry, ws.s); + archive_wstring_free(&ws); + p = NULL; + } else +#endif + p = archive_entry_pathname(entry); + /* + * On Windows, this is a backup operation just in + * case getting WCS failed. On POSIX, this is a + * normal operation. + */ + if (p != NULL && p[0] != '\0' && p[strlen(p) - 1] != '/') { + struct archive_string as; + + archive_string_init(&as); + path_length = strlen(p); + if (archive_string_ensure(&as, + path_length + 2) == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate ustar data"); + archive_string_free(&as); + return(ARCHIVE_FATAL); + } +#if defined(_WIN32) && !defined(__CYGWIN__) + /* NOTE: This might break the pathname + * if the current code page is CP932 and + * the pathname includes a character '\' + * as a part of its multibyte pathname. */ + if (p[strlen(p) -1] == '\\') + path_length--; + else +#endif + archive_strncpy(&as, p, path_length); + archive_strappend_char(&as, '/'); + archive_entry_copy_pathname(entry, as.s); + archive_string_free(&as); + } + } + +#if defined(_WIN32) && !defined(__CYGWIN__) + /* Make sure the path separators in pathname, hardlink and symlink + * are all slash '/', not the Windows path separator '\'. */ + entry_main = __la_win_entry_in_posix_pathseparator(entry); + if (entry_main == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate ustar data"); + return(ARCHIVE_FATAL); + } + if (entry != entry_main) + entry = entry_main; + else + entry_main = NULL; +#else + entry_main = NULL; +#endif + ret = __archive_write_format_header_ustar(a, buff, entry, -1, 1, sconv); + if (ret < ARCHIVE_WARN) { + archive_entry_free(entry_main); + return (ret); + } + ret2 = __archive_write_output(a, buff, 512); + if (ret2 < ARCHIVE_WARN) { + archive_entry_free(entry_main); + return (ret2); + } + if (ret2 < ret) + ret = ret2; + + ustar->entry_bytes_remaining = archive_entry_size(entry); + ustar->entry_padding = 0x1ff & (-(int64_t)ustar->entry_bytes_remaining); + archive_entry_free(entry_main); + return (ret); +} + +/* + * Format a basic 512-byte "ustar" header. + * + * Returns -1 if format failed (due to field overflow). + * Note that this always formats as much of the header as possible. + * If "strict" is set to zero, it will extend numeric fields as + * necessary (overwriting terminators or using base-256 extensions). + * + * This is exported so that other 'tar' formats can use it. + */ +int +__archive_write_format_header_ustar(struct archive_write *a, char h[512], + struct archive_entry *entry, int tartype, int strict, + struct archive_string_conv *sconv) +{ + unsigned int checksum; + int i, r, ret; + size_t copy_length; + const char *p, *pp; + int mytartype; + + ret = 0; + mytartype = -1; + /* + * The "template header" already includes the "ustar" + * signature, various end-of-field markers and other required + * elements. + */ + memcpy(h, &template_header, 512); + + /* + * Because the block is already null-filled, and strings + * are allowed to exactly fill their destination (without null), + * I use memcpy(dest, src, strlen()) here a lot to copy strings. + */ + r = archive_entry_pathname_l(entry, &pp, ©_length, sconv); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Pathname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate pathname '%s' to %s", + pp, archive_string_conversion_charset_name(sconv)); + ret = ARCHIVE_WARN; + } + if (copy_length <= USTAR_name_size) + memcpy(h + USTAR_name_offset, pp, copy_length); + else { + /* Store in two pieces, splitting at a '/'. */ + p = strchr(pp + copy_length - USTAR_name_size - 1, '/'); + /* + * Look for the next '/' if we chose the first character + * as the separator. (ustar format doesn't permit + * an empty prefix.) + */ + if (p == pp) + p = strchr(p + 1, '/'); + /* Fail if the name won't fit. */ + if (!p) { + /* No separator. */ + archive_set_error(&a->archive, ENAMETOOLONG, + "Pathname too long"); + ret = ARCHIVE_FAILED; + } else if (p[1] == '\0') { + /* + * The only feasible separator is a final '/'; + * this would result in a non-empty prefix and + * an empty name, which POSIX doesn't + * explicitly forbid, but it just feels wrong. + */ + archive_set_error(&a->archive, ENAMETOOLONG, + "Pathname too long"); + ret = ARCHIVE_FAILED; + } else if (p > pp + USTAR_prefix_size) { + /* Prefix is too long. */ + archive_set_error(&a->archive, ENAMETOOLONG, + "Pathname too long"); + ret = ARCHIVE_FAILED; + } else { + /* Copy prefix and remainder to appropriate places */ + memcpy(h + USTAR_prefix_offset, pp, p - pp); + memcpy(h + USTAR_name_offset, p + 1, + pp + copy_length - p - 1); + } + } + + r = archive_entry_hardlink_l(entry, &p, ©_length, sconv); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Linkname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate linkname '%s' to %s", + p, archive_string_conversion_charset_name(sconv)); + ret = ARCHIVE_WARN; + } + if (copy_length > 0) + mytartype = '1'; + else { + r = archive_entry_symlink_l(entry, &p, ©_length, sconv); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Linkname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate linkname '%s' to %s", + p, archive_string_conversion_charset_name(sconv)); + ret = ARCHIVE_WARN; + } + } + if (copy_length > 0) { + if (copy_length > USTAR_linkname_size) { + archive_set_error(&a->archive, ENAMETOOLONG, + "Link contents too long"); + ret = ARCHIVE_FAILED; + copy_length = USTAR_linkname_size; + } + memcpy(h + USTAR_linkname_offset, p, copy_length); + } + + r = archive_entry_uname_l(entry, &p, ©_length, sconv); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Uname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate uname '%s' to %s", + p, archive_string_conversion_charset_name(sconv)); + ret = ARCHIVE_WARN; + } + if (copy_length > 0) { + if (copy_length > USTAR_uname_size) { + if (tartype != 'x') { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, "Username too long"); + ret = ARCHIVE_FAILED; + } + copy_length = USTAR_uname_size; + } + memcpy(h + USTAR_uname_offset, p, copy_length); + } + + r = archive_entry_gname_l(entry, &p, ©_length, sconv); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Gname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate gname '%s' to %s", + p, archive_string_conversion_charset_name(sconv)); + ret = ARCHIVE_WARN; + } + if (copy_length > 0) { + if (strlen(p) > USTAR_gname_size) { + if (tartype != 'x') { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, "Group name too long"); + ret = ARCHIVE_FAILED; + } + copy_length = USTAR_gname_size; + } + memcpy(h + USTAR_gname_offset, p, copy_length); + } + + if (format_number(archive_entry_mode(entry) & 07777, + h + USTAR_mode_offset, USTAR_mode_size, USTAR_mode_max_size, strict)) { + archive_set_error(&a->archive, ERANGE, + "Numeric mode too large"); + ret = ARCHIVE_FAILED; + } + + if (format_number(archive_entry_uid(entry), + h + USTAR_uid_offset, USTAR_uid_size, USTAR_uid_max_size, strict)) { + archive_set_error(&a->archive, ERANGE, + "Numeric user ID too large"); + ret = ARCHIVE_FAILED; + } + + if (format_number(archive_entry_gid(entry), + h + USTAR_gid_offset, USTAR_gid_size, USTAR_gid_max_size, strict)) { + archive_set_error(&a->archive, ERANGE, + "Numeric group ID too large"); + ret = ARCHIVE_FAILED; + } + + if (format_number(archive_entry_size(entry), + h + USTAR_size_offset, USTAR_size_size, USTAR_size_max_size, strict)) { + archive_set_error(&a->archive, ERANGE, + "File size out of range"); + ret = ARCHIVE_FAILED; + } + + if (format_number(archive_entry_mtime(entry), + h + USTAR_mtime_offset, USTAR_mtime_size, USTAR_mtime_max_size, strict)) { + archive_set_error(&a->archive, ERANGE, + "File modification time too large"); + ret = ARCHIVE_FAILED; + } + + if (archive_entry_filetype(entry) == AE_IFBLK + || archive_entry_filetype(entry) == AE_IFCHR) { + if (format_number(archive_entry_rdevmajor(entry), + h + USTAR_rdevmajor_offset, USTAR_rdevmajor_size, + USTAR_rdevmajor_max_size, strict)) { + archive_set_error(&a->archive, ERANGE, + "Major device number too large"); + ret = ARCHIVE_FAILED; + } + + if (format_number(archive_entry_rdevminor(entry), + h + USTAR_rdevminor_offset, USTAR_rdevminor_size, + USTAR_rdevminor_max_size, strict)) { + archive_set_error(&a->archive, ERANGE, + "Minor device number too large"); + ret = ARCHIVE_FAILED; + } + } + + if (tartype >= 0) { + h[USTAR_typeflag_offset] = tartype; + } else if (mytartype >= 0) { + h[USTAR_typeflag_offset] = mytartype; + } else { + switch (archive_entry_filetype(entry)) { + case AE_IFREG: h[USTAR_typeflag_offset] = '0' ; break; + case AE_IFLNK: h[USTAR_typeflag_offset] = '2' ; break; + case AE_IFCHR: h[USTAR_typeflag_offset] = '3' ; break; + case AE_IFBLK: h[USTAR_typeflag_offset] = '4' ; break; + case AE_IFDIR: h[USTAR_typeflag_offset] = '5' ; break; + case AE_IFIFO: h[USTAR_typeflag_offset] = '6' ; break; + default: /* AE_IFSOCK and unknown */ + __archive_write_entry_filetype_unsupported( + &a->archive, entry, "ustar"); + ret = ARCHIVE_FAILED; + } + } + + checksum = 0; + for (i = 0; i < 512; i++) + checksum += 255 & (unsigned int)h[i]; + h[USTAR_checksum_offset + 6] = '\0'; /* Can't be pre-set in the template. */ + /* h[USTAR_checksum_offset + 7] = ' '; */ /* This is pre-set in the template. */ + format_octal(checksum, h + USTAR_checksum_offset, 6); + return (ret); +} + +/* + * Format a number into a field, with some intelligence. + */ +static int +format_number(int64_t v, char *p, int s, int maxsize, int strict) +{ + int64_t limit; + + limit = ((int64_t)1 << (s*3)); + + /* "Strict" only permits octal values with proper termination. */ + if (strict) + return (format_octal(v, p, s)); + + /* + * In non-strict mode, we allow the number to overwrite one or + * more bytes of the field termination. Even old tar + * implementations should be able to handle this with no + * problem. + */ + if (v >= 0) { + while (s <= maxsize) { + if (v < limit) + return (format_octal(v, p, s)); + s++; + limit <<= 3; + } + } + + /* Base-256 can handle any number, positive or negative. */ + return (format_256(v, p, maxsize)); +} + +/* + * Format a number into the specified field using base-256. + */ +static int +format_256(int64_t v, char *p, int s) +{ + p += s; + while (s-- > 0) { + *--p = (char)(v & 0xff); + v >>= 8; + } + *p |= 0x80; /* Set the base-256 marker bit. */ + return (0); +} + +/* + * Format a number into the specified field. + */ +static int +format_octal(int64_t v, char *p, int s) +{ + int len; + + len = s; + + /* Octal values can't be negative, so use 0. */ + if (v < 0) { + while (len-- > 0) + *p++ = '0'; + return (-1); + } + + p += s; /* Start at the end and work backwards. */ + while (s-- > 0) { + *--p = (char)('0' + (v & 7)); + v >>= 3; + } + + if (v == 0) + return (0); + + /* If it overflowed, fill field with max value. */ + while (len-- > 0) + *p++ = '7'; + + return (-1); +} + +static int +archive_write_ustar_close(struct archive_write *a) +{ + return (__archive_write_nulls(a, 512*2)); +} + +static int +archive_write_ustar_free(struct archive_write *a) +{ + struct ustar *ustar; + + ustar = (struct ustar *)a->format_data; + free(ustar); + a->format_data = NULL; + return (ARCHIVE_OK); +} + +static int +archive_write_ustar_finish_entry(struct archive_write *a) +{ + struct ustar *ustar; + int ret; + + ustar = (struct ustar *)a->format_data; + ret = __archive_write_nulls(a, + (size_t)(ustar->entry_bytes_remaining + ustar->entry_padding)); + ustar->entry_bytes_remaining = ustar->entry_padding = 0; + return (ret); +} + +static ssize_t +archive_write_ustar_data(struct archive_write *a, const void *buff, size_t s) +{ + struct ustar *ustar; + int ret; + + ustar = (struct ustar *)a->format_data; + if (s > ustar->entry_bytes_remaining) + s = (size_t)ustar->entry_bytes_remaining; + ret = __archive_write_output(a, buff, s); + ustar->entry_bytes_remaining -= s; + if (ret != ARCHIVE_OK) + return (ret); + return (s); +} diff --git a/c/archive_write_set_format_v7tar.c b/c/archive_write_set_format_v7tar.c new file mode 100644 --- /dev/null +++ b/c/archive_write_set_format_v7tar.c @@ -0,0 +1,638 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * Copyright (c) 2011-2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + + +#ifdef HAVE_ERRNO_H +#include +#endif +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_private.h" +#include "archive_write_private.h" +#include "archive_write_set_format_private.h" + +struct v7tar { + uint64_t entry_bytes_remaining; + uint64_t entry_padding; + + struct archive_string_conv *opt_sconv; + struct archive_string_conv *sconv_default; + int init_default_conversion; +}; + +/* + * Define structure of POSIX 'v7tar' tar header. + */ +#define V7TAR_name_offset 0 +#define V7TAR_name_size 100 +#define V7TAR_mode_offset 100 +#define V7TAR_mode_size 6 +#define V7TAR_mode_max_size 8 +#define V7TAR_uid_offset 108 +#define V7TAR_uid_size 6 +#define V7TAR_uid_max_size 8 +#define V7TAR_gid_offset 116 +#define V7TAR_gid_size 6 +#define V7TAR_gid_max_size 8 +#define V7TAR_size_offset 124 +#define V7TAR_size_size 11 +#define V7TAR_size_max_size 12 +#define V7TAR_mtime_offset 136 +#define V7TAR_mtime_size 11 +#define V7TAR_mtime_max_size 12 +#define V7TAR_checksum_offset 148 +#define V7TAR_checksum_size 8 +#define V7TAR_typeflag_offset 156 +#define V7TAR_typeflag_size 1 +#define V7TAR_linkname_offset 157 +#define V7TAR_linkname_size 100 +#define V7TAR_padding_offset 257 +#define V7TAR_padding_size 255 + +/* + * A filled-in copy of the header for initialization. + */ +static const char template_header[] = { + /* name: 100 bytes */ + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0, + /* Mode, space-null termination: 8 bytes */ + '0','0','0','0','0','0', ' ','\0', + /* uid, space-null termination: 8 bytes */ + '0','0','0','0','0','0', ' ','\0', + /* gid, space-null termination: 8 bytes */ + '0','0','0','0','0','0', ' ','\0', + /* size, space termination: 12 bytes */ + '0','0','0','0','0','0','0','0','0','0','0', ' ', + /* mtime, space termination: 12 bytes */ + '0','0','0','0','0','0','0','0','0','0','0', ' ', + /* Initial checksum value: 8 spaces */ + ' ',' ',' ',' ',' ',' ',' ',' ', + /* Typeflag: 1 byte */ + 0, + /* Linkname: 100 bytes */ + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0, + /* Padding: 255 bytes */ + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0 +}; + +static ssize_t archive_write_v7tar_data(struct archive_write *a, const void *buff, + size_t s); +static int archive_write_v7tar_free(struct archive_write *); +static int archive_write_v7tar_close(struct archive_write *); +static int archive_write_v7tar_finish_entry(struct archive_write *); +static int archive_write_v7tar_header(struct archive_write *, + struct archive_entry *entry); +static int archive_write_v7tar_options(struct archive_write *, + const char *, const char *); +static int format_256(int64_t, char *, int); +static int format_number(int64_t, char *, int size, int max, int strict); +static int format_octal(int64_t, char *, int); +static int format_header_v7tar(struct archive_write *, char h[512], + struct archive_entry *, int, struct archive_string_conv *); + +/* + * Set output format to 'v7tar' format. + */ +int +archive_write_set_format_v7tar(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct v7tar *v7tar; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_set_format_v7tar"); + + /* If someone else was already registered, unregister them. */ + if (a->format_free != NULL) + (a->format_free)(a); + + /* Basic internal sanity test. */ + if (sizeof(template_header) != 512) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Internal: template_header wrong size: %zu should be 512", + sizeof(template_header)); + return (ARCHIVE_FATAL); + } + + v7tar = (struct v7tar *)calloc(1, sizeof(*v7tar)); + if (v7tar == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate v7tar data"); + return (ARCHIVE_FATAL); + } + a->format_data = v7tar; + a->format_name = "tar (non-POSIX)"; + a->format_options = archive_write_v7tar_options; + a->format_write_header = archive_write_v7tar_header; + a->format_write_data = archive_write_v7tar_data; + a->format_close = archive_write_v7tar_close; + a->format_free = archive_write_v7tar_free; + a->format_finish_entry = archive_write_v7tar_finish_entry; + a->archive.archive_format = ARCHIVE_FORMAT_TAR; + a->archive.archive_format_name = "tar (non-POSIX)"; + return (ARCHIVE_OK); +} + +static int +archive_write_v7tar_options(struct archive_write *a, const char *key, + const char *val) +{ + struct v7tar *v7tar = (struct v7tar *)a->format_data; + int ret = ARCHIVE_FAILED; + + if (strcmp(key, "hdrcharset") == 0) { + if (val == NULL || val[0] == 0) + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "%s: hdrcharset option needs a character-set name", + a->format_name); + else { + v7tar->opt_sconv = archive_string_conversion_to_charset( + &a->archive, val, 0); + if (v7tar->opt_sconv != NULL) + ret = ARCHIVE_OK; + else + ret = ARCHIVE_FATAL; + } + return (ret); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +static int +archive_write_v7tar_header(struct archive_write *a, struct archive_entry *entry) +{ + char buff[512]; + int ret, ret2; + struct v7tar *v7tar; + struct archive_entry *entry_main; + struct archive_string_conv *sconv; + + v7tar = (struct v7tar *)a->format_data; + + /* Setup default string conversion. */ + if (v7tar->opt_sconv == NULL) { + if (!v7tar->init_default_conversion) { + v7tar->sconv_default = + archive_string_default_conversion_for_write( + &(a->archive)); + v7tar->init_default_conversion = 1; + } + sconv = v7tar->sconv_default; + } else + sconv = v7tar->opt_sconv; + + /* Sanity check. */ + if (archive_entry_pathname(entry) == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Can't record entry in tar file without pathname"); + return (ARCHIVE_FAILED); + } + + /* Only regular files (not hardlinks) have data. */ + if (archive_entry_hardlink(entry) != NULL || + archive_entry_symlink(entry) != NULL || + !(archive_entry_filetype(entry) == AE_IFREG)) + archive_entry_set_size(entry, 0); + + if (AE_IFDIR == archive_entry_filetype(entry)) { + const char *p; + size_t path_length; + /* + * Ensure a trailing '/'. Modify the entry so + * the client sees the change. + */ +#if defined(_WIN32) && !defined(__CYGWIN__) + const wchar_t *wp; + + wp = archive_entry_pathname_w(entry); + if (wp != NULL && wp[wcslen(wp) -1] != L'/') { + struct archive_wstring ws; + + archive_string_init(&ws); + path_length = wcslen(wp); + if (archive_wstring_ensure(&ws, + path_length + 2) == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate v7tar data"); + archive_wstring_free(&ws); + return(ARCHIVE_FATAL); + } + /* Should we keep '\' ? */ + if (wp[path_length -1] == L'\\') + path_length--; + archive_wstrncpy(&ws, wp, path_length); + archive_wstrappend_wchar(&ws, L'/'); + archive_entry_copy_pathname_w(entry, ws.s); + archive_wstring_free(&ws); + p = NULL; + } else +#endif + p = archive_entry_pathname(entry); + /* + * On Windows, this is a backup operation just in + * case getting WCS failed. On POSIX, this is a + * normal operation. + */ + if (p != NULL && p[0] != '\0' && p[strlen(p) - 1] != '/') { + struct archive_string as; + + archive_string_init(&as); + path_length = strlen(p); + if (archive_string_ensure(&as, + path_length + 2) == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate v7tar data"); + archive_string_free(&as); + return(ARCHIVE_FATAL); + } +#if defined(_WIN32) && !defined(__CYGWIN__) + /* NOTE: This might break the pathname + * if the current code page is CP932 and + * the pathname includes a character '\' + * as a part of its multibyte pathname. */ + if (p[strlen(p) -1] == '\\') + path_length--; + else +#endif + archive_strncpy(&as, p, path_length); + archive_strappend_char(&as, '/'); + archive_entry_copy_pathname(entry, as.s); + archive_string_free(&as); + } + } + +#if defined(_WIN32) && !defined(__CYGWIN__) + /* Make sure the path separators in pathname, hardlink and symlink + * are all slash '/', not the Windows path separator '\'. */ + entry_main = __la_win_entry_in_posix_pathseparator(entry); + if (entry_main == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate v7tar data"); + return(ARCHIVE_FATAL); + } + if (entry != entry_main) + entry = entry_main; + else + entry_main = NULL; +#else + entry_main = NULL; +#endif + ret = format_header_v7tar(a, buff, entry, 1, sconv); + if (ret < ARCHIVE_WARN) { + archive_entry_free(entry_main); + return (ret); + } + ret2 = __archive_write_output(a, buff, 512); + if (ret2 < ARCHIVE_WARN) { + archive_entry_free(entry_main); + return (ret2); + } + if (ret2 < ret) + ret = ret2; + + v7tar->entry_bytes_remaining = archive_entry_size(entry); + v7tar->entry_padding = 0x1ff & (-(int64_t)v7tar->entry_bytes_remaining); + archive_entry_free(entry_main); + return (ret); +} + +/* + * Format a basic 512-byte "v7tar" header. + * + * Returns -1 if format failed (due to field overflow). + * Note that this always formats as much of the header as possible. + * If "strict" is set to zero, it will extend numeric fields as + * necessary (overwriting terminators or using base-256 extensions). + * + */ +static int +format_header_v7tar(struct archive_write *a, char h[512], + struct archive_entry *entry, int strict, + struct archive_string_conv *sconv) +{ + unsigned int checksum; + int i, r, ret; + size_t copy_length; + const char *p, *pp; + int mytartype; + + ret = 0; + mytartype = -1; + /* + * The "template header" already includes the "v7tar" + * signature, various end-of-field markers and other required + * elements. + */ + memcpy(h, &template_header, 512); + + /* + * Because the block is already null-filled, and strings + * are allowed to exactly fill their destination (without null), + * I use memcpy(dest, src, strlen()) here a lot to copy strings. + */ + r = archive_entry_pathname_l(entry, &pp, ©_length, sconv); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Pathname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate pathname '%s' to %s", + pp, archive_string_conversion_charset_name(sconv)); + ret = ARCHIVE_WARN; + } + if (strict && copy_length < V7TAR_name_size) + memcpy(h + V7TAR_name_offset, pp, copy_length); + else if (!strict && copy_length <= V7TAR_name_size) + memcpy(h + V7TAR_name_offset, pp, copy_length); + else { + /* Prefix is too long. */ + archive_set_error(&a->archive, ENAMETOOLONG, + "Pathname too long"); + ret = ARCHIVE_FAILED; + } + + r = archive_entry_hardlink_l(entry, &p, ©_length, sconv); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Linkname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate linkname '%s' to %s", + p, archive_string_conversion_charset_name(sconv)); + ret = ARCHIVE_WARN; + } + if (copy_length > 0) + mytartype = '1'; + else { + r = archive_entry_symlink_l(entry, &p, ©_length, sconv); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Linkname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate linkname '%s' to %s", + p, archive_string_conversion_charset_name(sconv)); + ret = ARCHIVE_WARN; + } + } + if (copy_length > 0) { + if (copy_length >= V7TAR_linkname_size) { + archive_set_error(&a->archive, ENAMETOOLONG, + "Link contents too long"); + ret = ARCHIVE_FAILED; + copy_length = V7TAR_linkname_size; + } + memcpy(h + V7TAR_linkname_offset, p, copy_length); + } + + if (format_number(archive_entry_mode(entry) & 07777, + h + V7TAR_mode_offset, V7TAR_mode_size, + V7TAR_mode_max_size, strict)) { + archive_set_error(&a->archive, ERANGE, + "Numeric mode too large"); + ret = ARCHIVE_FAILED; + } + + if (format_number(archive_entry_uid(entry), + h + V7TAR_uid_offset, V7TAR_uid_size, V7TAR_uid_max_size, strict)) { + archive_set_error(&a->archive, ERANGE, + "Numeric user ID too large"); + ret = ARCHIVE_FAILED; + } + + if (format_number(archive_entry_gid(entry), + h + V7TAR_gid_offset, V7TAR_gid_size, V7TAR_gid_max_size, strict)) { + archive_set_error(&a->archive, ERANGE, + "Numeric group ID too large"); + ret = ARCHIVE_FAILED; + } + + if (format_number(archive_entry_size(entry), + h + V7TAR_size_offset, V7TAR_size_size, + V7TAR_size_max_size, strict)) { + archive_set_error(&a->archive, ERANGE, + "File size out of range"); + ret = ARCHIVE_FAILED; + } + + if (format_number(archive_entry_mtime(entry), + h + V7TAR_mtime_offset, V7TAR_mtime_size, + V7TAR_mtime_max_size, strict)) { + archive_set_error(&a->archive, ERANGE, + "File modification time too large"); + ret = ARCHIVE_FAILED; + } + + if (mytartype >= 0) { + h[V7TAR_typeflag_offset] = mytartype; + } else { + switch (archive_entry_filetype(entry)) { + case AE_IFREG: case AE_IFDIR: + break; + case AE_IFLNK: + h[V7TAR_typeflag_offset] = '2'; + break; + default: + /* AE_IFBLK, AE_IFCHR, AE_IFIFO, AE_IFSOCK + * and unknown */ + __archive_write_entry_filetype_unsupported( + &a->archive, entry, "v7tar"); + ret = ARCHIVE_FAILED; + } + } + + checksum = 0; + for (i = 0; i < 512; i++) + checksum += 255 & (unsigned int)h[i]; + format_octal(checksum, h + V7TAR_checksum_offset, 6); + /* Can't be pre-set in the template. */ + h[V7TAR_checksum_offset + 6] = '\0'; + return (ret); +} + +/* + * Format a number into a field, with some intelligence. + */ +static int +format_number(int64_t v, char *p, int s, int maxsize, int strict) +{ + int64_t limit; + + limit = ((int64_t)1 << (s*3)); + + /* "Strict" only permits octal values with proper termination. */ + if (strict) + return (format_octal(v, p, s)); + + /* + * In non-strict mode, we allow the number to overwrite one or + * more bytes of the field termination. Even old tar + * implementations should be able to handle this with no + * problem. + */ + if (v >= 0) { + while (s <= maxsize) { + if (v < limit) + return (format_octal(v, p, s)); + s++; + limit <<= 3; + } + } + + /* Base-256 can handle any number, positive or negative. */ + return (format_256(v, p, maxsize)); +} + +/* + * Format a number into the specified field using base-256. + */ +static int +format_256(int64_t v, char *p, int s) +{ + p += s; + while (s-- > 0) { + *--p = (char)(v & 0xff); + v >>= 8; + } + *p |= 0x80; /* Set the base-256 marker bit. */ + return (0); +} + +/* + * Format a number into the specified field. + */ +static int +format_octal(int64_t v, char *p, int s) +{ + int len; + + len = s; + + /* Octal values can't be negative, so use 0. */ + if (v < 0) { + while (len-- > 0) + *p++ = '0'; + return (-1); + } + + p += s; /* Start at the end and work backwards. */ + while (s-- > 0) { + *--p = (char)('0' + (v & 7)); + v >>= 3; + } + + if (v == 0) + return (0); + + /* If it overflowed, fill field with max value. */ + while (len-- > 0) + *p++ = '7'; + + return (-1); +} + +static int +archive_write_v7tar_close(struct archive_write *a) +{ + return (__archive_write_nulls(a, 512*2)); +} + +static int +archive_write_v7tar_free(struct archive_write *a) +{ + struct v7tar *v7tar; + + v7tar = (struct v7tar *)a->format_data; + free(v7tar); + a->format_data = NULL; + return (ARCHIVE_OK); +} + +static int +archive_write_v7tar_finish_entry(struct archive_write *a) +{ + struct v7tar *v7tar; + int ret; + + v7tar = (struct v7tar *)a->format_data; + ret = __archive_write_nulls(a, + (size_t)(v7tar->entry_bytes_remaining + v7tar->entry_padding)); + v7tar->entry_bytes_remaining = v7tar->entry_padding = 0; + return (ret); +} + +static ssize_t +archive_write_v7tar_data(struct archive_write *a, const void *buff, size_t s) +{ + struct v7tar *v7tar; + int ret; + + v7tar = (struct v7tar *)a->format_data; + if (s > v7tar->entry_bytes_remaining) + s = (size_t)v7tar->entry_bytes_remaining; + ret = __archive_write_output(a, buff, s); + v7tar->entry_bytes_remaining -= s; + if (ret != ARCHIVE_OK) + return (ret); + return (s); +} diff --git a/c/archive_write_set_format_warc.c b/c/archive_write_set_format_warc.c new file mode 100644 --- /dev/null +++ b/c/archive_write_set_format_warc.c @@ -0,0 +1,453 @@ +/*- + * Copyright (c) 2014 Sebastian Freundt + * Author: Sebastian Freundt + * + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#include +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_TIME_H +#include +#endif + +#include "archive.h" +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_private.h" +#include "archive_random_private.h" +#include "archive_write_private.h" +#include "archive_write_set_format_private.h" + +struct warc_s { + unsigned int omit_warcinfo:1; + + time_t now; + mode_t typ; + unsigned int rng; + /* populated size */ + uint64_t populz; +}; + +static const char warcinfo[] = + "software: libarchive/" ARCHIVE_VERSION_ONLY_STRING "\r\n" + "format: WARC file version 1.0\r\n"; + +typedef enum { + WT_NONE, + /* warcinfo */ + WT_INFO, + /* metadata */ + WT_META, + /* resource */ + WT_RSRC, + /* request, unsupported */ + WT_REQ, + /* response, unsupported */ + WT_RSP, + /* revisit, unsupported */ + WT_RVIS, + /* conversion, unsupported */ + WT_CONV, + /* continuation, unsupported at the moment */ + WT_CONT, + /* invalid type */ + LAST_WT +} warc_type_t; + +typedef struct { + warc_type_t type; + const char *tgturi; + const char *recid; + time_t rtime; + time_t mtime; + const char *cnttyp; + uint64_t cntlen; +} warc_essential_hdr_t; + +typedef struct { + unsigned int u[4U]; +} warc_uuid_t; + +static int _warc_options(struct archive_write*, const char *key, const char *v); +static int _warc_header(struct archive_write *a, struct archive_entry *entry); +static ssize_t _warc_data(struct archive_write *a, const void *buf, size_t sz); +static int _warc_finish_entry(struct archive_write *a); +static int _warc_close(struct archive_write *a); +static int _warc_free(struct archive_write *a); + +/* private routines */ +static ssize_t _popul_ehdr(struct archive_string *t, size_t z, warc_essential_hdr_t); +static int _gen_uuid(warc_uuid_t *tgt); + + +/* + * Set output format to ISO 28500 (aka WARC) format. + */ +int +archive_write_set_format_warc(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct warc_s *w; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_set_format_warc"); + + /* If another format was already registered, unregister it. */ + if (a->format_free != NULL) { + (a->format_free)(a); + } + + w = malloc(sizeof(*w)); + if (w == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate warc data"); + return (ARCHIVE_FATAL); + } + /* by default we're emitting a file wide header */ + w->omit_warcinfo = 0U; + /* obtain current time for date fields */ + w->now = time(NULL); + /* reset file type info */ + w->typ = 0; + /* also initialise our rng */ + w->rng = (unsigned int)w->now; + + a->format_data = w; + a->format_name = "WARC/1.0"; + a->format_options = _warc_options; + a->format_write_header = _warc_header; + a->format_write_data = _warc_data; + a->format_close = _warc_close; + a->format_free = _warc_free; + a->format_finish_entry = _warc_finish_entry; + a->archive.archive_format = ARCHIVE_FORMAT_WARC; + a->archive.archive_format_name = "WARC/1.0"; + return (ARCHIVE_OK); +} + + +/* archive methods */ +static int +_warc_options(struct archive_write *a, const char *key, const char *val) +{ + struct warc_s *w = a->format_data; + + if (strcmp(key, "omit-warcinfo") == 0) { + if (val == NULL || strcmp(val, "true") == 0) { + /* great */ + w->omit_warcinfo = 1U; + return (ARCHIVE_OK); + } + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +static int +_warc_header(struct archive_write *a, struct archive_entry *entry) +{ + struct warc_s *w = a->format_data; + struct archive_string hdr; +#define MAX_HDR_SIZE 512 + + /* check whether warcinfo record needs outputting */ + if (!w->omit_warcinfo) { + ssize_t r; + warc_essential_hdr_t wi = { + WT_INFO, + /*uri*/NULL, + /*urn*/NULL, + /*rtm*/0, + /*mtm*/0, + /*cty*/"application/warc-fields", + /*len*/sizeof(warcinfo) - 1U, + }; + wi.rtime = w->now; + wi.mtime = w->now; + + archive_string_init(&hdr); + r = _popul_ehdr(&hdr, MAX_HDR_SIZE, wi); + if (r >= 0) { + /* jackpot! */ + /* now also use HDR buffer for the actual warcinfo */ + archive_strncat(&hdr, warcinfo, sizeof(warcinfo) -1); + + /* append end-of-record indicator */ + archive_strncat(&hdr, "\r\n\r\n", 4); + + /* write to output stream */ + __archive_write_output(a, hdr.s, archive_strlen(&hdr)); + } + /* indicate we're done with file header writing */ + w->omit_warcinfo = 1U; + archive_string_free(&hdr); + } + + if (archive_entry_pathname(entry) == NULL) { + archive_set_error(&a->archive, EINVAL, + "Invalid filename"); + return (ARCHIVE_WARN); + } + + w->typ = archive_entry_filetype(entry); + w->populz = 0U; + if (w->typ == AE_IFREG) { + warc_essential_hdr_t rh = { + WT_RSRC, + /*uri*/NULL, + /*urn*/NULL, + /*rtm*/0, + /*mtm*/0, + /*cty*/NULL, + /*len*/0, + }; + ssize_t r; + rh.tgturi = archive_entry_pathname(entry); + rh.rtime = w->now; + rh.mtime = archive_entry_mtime(entry); + rh.cntlen = (size_t)archive_entry_size(entry); + + archive_string_init(&hdr); + r = _popul_ehdr(&hdr, MAX_HDR_SIZE, rh); + if (r < 0) { + /* don't bother */ + archive_set_error( + &a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "cannot archive file"); + return (ARCHIVE_WARN); + } + /* otherwise append to output stream */ + __archive_write_output(a, hdr.s, r); + /* and let subsequent calls to _data() know about the size */ + w->populz = rh.cntlen; + archive_string_free(&hdr); + return (ARCHIVE_OK); + } + /* just resort to erroring as per Tim's advice */ + __archive_write_entry_filetype_unsupported( + &a->archive, entry, "WARC"); + return (ARCHIVE_FAILED); +} + +static ssize_t +_warc_data(struct archive_write *a, const void *buf, size_t len) +{ + struct warc_s *w = a->format_data; + + if (w->typ == AE_IFREG) { + int rc; + + /* never write more bytes than announced */ + if (len > w->populz) { + len = (size_t)w->populz; + } + + /* now then, out we put the whole shebang */ + rc = __archive_write_output(a, buf, len); + if (rc != ARCHIVE_OK) { + return rc; + } + } + return len; +} + +static int +_warc_finish_entry(struct archive_write *a) +{ + static const char _eor[] = "\r\n\r\n"; + struct warc_s *w = a->format_data; + + if (w->typ == AE_IFREG) { + int rc = __archive_write_output(a, _eor, sizeof(_eor) - 1U); + + if (rc != ARCHIVE_OK) { + return rc; + } + } + /* reset type info */ + w->typ = 0; + return (ARCHIVE_OK); +} + +static int +_warc_close(struct archive_write *a) +{ + (void)a; /* UNUSED */ + return (ARCHIVE_OK); +} + +static int +_warc_free(struct archive_write *a) +{ + struct warc_s *w = a->format_data; + + free(w); + a->format_data = NULL; + return (ARCHIVE_OK); +} + + +/* private routines */ +static void +xstrftime(struct archive_string *as, const char *fmt, time_t t) +{ +/** like strftime(3) but for time_t objects */ + struct tm *rt; +#if defined(HAVE_GMTIME_R) || defined(HAVE__GMTIME64_S) + struct tm timeHere; +#endif +#if defined(HAVE__GMTIME64_S) + errno_t terr; + __time64_t tmptime; +#endif + char strtime[100]; + size_t len; + +#ifdef HAVE_GMTIME_R + if ((rt = gmtime_r(&t, &timeHere)) == NULL) + return; +#elif defined(HAVE__GMTIME64_S) + tmptime = t; + terr = _gmtime64_s(&timeHere, &tmptime); + if (terr) + rt = NULL; + else + rt = &timeHere; +#else + if ((rt = gmtime(&t)) == NULL) + return; +#endif + /* leave the hard yacker to our role model strftime() */ + len = strftime(strtime, sizeof(strtime)-1, fmt, rt); + archive_strncat(as, strtime, len); +} + +static ssize_t +_popul_ehdr(struct archive_string *tgt, size_t tsz, warc_essential_hdr_t hdr) +{ + static const char _ver[] = "WARC/1.0\r\n"; + static const char * const _typ[LAST_WT] = { + NULL, "warcinfo", "metadata", "resource", NULL + }; + char std_uuid[48U]; + + if (hdr.type == WT_NONE || hdr.type > WT_RSRC) { + /* brilliant, how exactly did we get here? */ + return -1; + } + + archive_strcpy(tgt, _ver); + + archive_string_sprintf(tgt, "WARC-Type: %s\r\n", _typ[hdr.type]); + + if (hdr.tgturi != NULL) { + /* check if there's a xyz:// */ + static const char _uri[] = ""; + static const char _fil[] = "file://"; + const char *u; + char *chk = strchr(hdr.tgturi, ':'); + + if (chk != NULL && chk[1U] == '/' && chk[2U] == '/') { + /* yep, it's definitely a URI */ + u = _uri; + } else { + /* hm, best to prepend file:// then */ + u = _fil; + } + archive_string_sprintf(tgt, + "WARC-Target-URI: %s%s\r\n", u, hdr.tgturi); + } + + /* record time is usually when the http is sent off, + * just treat the archive writing as such for a moment */ + xstrftime(tgt, "WARC-Date: %Y-%m-%dT%H:%M:%SZ\r\n", hdr.rtime); + + /* while we're at it, record the mtime */ + xstrftime(tgt, "Last-Modified: %Y-%m-%dT%H:%M:%SZ\r\n", hdr.mtime); + + if (hdr.recid == NULL) { + /* generate one, grrrr */ + warc_uuid_t u; + + _gen_uuid(&u); + /* Unfortunately, archive_string_sprintf does not + * handle the minimum number following '%'. + * So we have to use snprintf function here instead + * of archive_string_snprintf function. */ +#if defined(_WIN32) && !defined(__CYGWIN__) && !( defined(_MSC_VER) && _MSC_VER >= 1900) +#define snprintf _snprintf +#endif + snprintf( + std_uuid, sizeof(std_uuid), + "", + u.u[0U], + u.u[1U] >> 16U, u.u[1U] & 0xffffU, + u.u[2U] >> 16U, u.u[2U] & 0xffffU, + u.u[3U]); + hdr.recid = std_uuid; + } + + /* record-id is mandatory, fingers crossed we won't fail */ + archive_string_sprintf(tgt, "WARC-Record-ID: %s\r\n", hdr.recid); + + if (hdr.cnttyp != NULL) { + archive_string_sprintf(tgt, "Content-Type: %s\r\n", hdr.cnttyp); + } + + /* next one is mandatory */ + archive_string_sprintf(tgt, "Content-Length: %ju\r\n", (uintmax_t)hdr.cntlen); + /**/ + archive_strncat(tgt, "\r\n", 2); + + return (archive_strlen(tgt) >= tsz)? -1: (ssize_t)archive_strlen(tgt); +} + +static int +_gen_uuid(warc_uuid_t *tgt) +{ + archive_random(tgt->u, sizeof(tgt->u)); + /* obey uuid version 4 rules */ + tgt->u[1U] &= 0xffff0fffU; + tgt->u[1U] |= 0x4000U; + tgt->u[2U] &= 0x3fffffffU; + tgt->u[2U] |= 0x80000000U; + return 0; +} + +/* archive_write_set_format_warc.c ends here */ diff --git a/c/archive_write_set_format_xar.c b/c/archive_write_set_format_xar.c new file mode 100644 --- /dev/null +++ b/c/archive_write_set_format_xar.c @@ -0,0 +1,3259 @@ +/*- + * Copyright (c) 2010-2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_LIMITS_H +#include +#endif +#include +#if HAVE_LIBXML_XMLWRITER_H +#include +#endif +#ifdef HAVE_BZLIB_H +#include +#endif +#if HAVE_LZMA_H +#include +#endif +#ifdef HAVE_ZLIB_H +#include +#endif + +#include "archive.h" +#include "archive_digest_private.h" +#include "archive_endian.h" +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_private.h" +#include "archive_rb.h" +#include "archive_string.h" +#include "archive_write_private.h" + +/* + * Differences to xar utility. + * - Subdocument is not supported yet. + * - ACL is not supported yet. + * - When writing an XML element , + * which is a file type a symbolic link is referencing is always marked + * as "broken". Xar utility uses stat(2) to get the file type, but, in + * libarchive format writer, we should not use it; if it is needed, we + * should get about it at archive_read_disk.c. + * - It is possible to appear both and elements. + * Xar utility generates on BSD platform and on Linux + * platform. + * + */ + +#if !(defined(HAVE_LIBXML_XMLWRITER_H) && defined(LIBXML_VERSION) &&\ + LIBXML_VERSION >= 20703) ||\ + !defined(HAVE_ZLIB_H) || \ + !defined(ARCHIVE_HAS_MD5) || !defined(ARCHIVE_HAS_SHA1) +/* + * xar needs several external libraries. + * o libxml2 + * o openssl or MD5/SHA1 hash function + * o zlib + * o bzlib2 (option) + * o liblzma (option) + */ +int +archive_write_set_format_xar(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Xar not supported on this platform"); + return (ARCHIVE_WARN); +} + +#else /* Support xar format */ + +/*#define DEBUG_PRINT_TOC 1 */ + +#define BAD_CAST_CONST (const xmlChar *) + +#define HEADER_MAGIC 0x78617221 +#define HEADER_SIZE 28 +#define HEADER_VERSION 1 + +enum sumalg { + CKSUM_NONE = 0, + CKSUM_SHA1 = 1, + CKSUM_MD5 = 2 +}; + +#define MD5_SIZE 16 +#define SHA1_SIZE 20 +#define MAX_SUM_SIZE 20 +#define MD5_NAME "md5" +#define SHA1_NAME "sha1" + +enum enctype { + NONE, + GZIP, + BZIP2, + LZMA, + XZ, +}; + +struct chksumwork { + enum sumalg alg; +#ifdef ARCHIVE_HAS_MD5 + archive_md5_ctx md5ctx; +#endif +#ifdef ARCHIVE_HAS_SHA1 + archive_sha1_ctx sha1ctx; +#endif +}; + +enum la_zaction { + ARCHIVE_Z_FINISH, + ARCHIVE_Z_RUN +}; + +/* + * Universal zstream. + */ +struct la_zstream { + const unsigned char *next_in; + size_t avail_in; + uint64_t total_in; + + unsigned char *next_out; + size_t avail_out; + uint64_t total_out; + + int valid; + void *real_stream; + int (*code) (struct archive *a, + struct la_zstream *lastrm, + enum la_zaction action); + int (*end)(struct archive *a, + struct la_zstream *lastrm); +}; + +struct chksumval { + enum sumalg alg; + size_t len; + unsigned char val[MAX_SUM_SIZE]; +}; + +struct heap_data { + int id; + struct heap_data *next; + uint64_t temp_offset; + uint64_t length; /* archived size. */ + uint64_t size; /* extracted size. */ + enum enctype compression; + struct chksumval a_sum; /* archived checksum. */ + struct chksumval e_sum; /* extracted checksum. */ +}; + +struct file { + struct archive_rb_node rbnode; + + int id; + struct archive_entry *entry; + + struct archive_rb_tree rbtree; + struct file *next; + struct file *chnext; + struct file *hlnext; + /* For hardlinked files. + * Use only when archive_entry_nlink() > 1 */ + struct file *hardlink_target; + struct file *parent; /* parent directory entry */ + /* + * To manage sub directory files. + * We use 'chnext' (a member of struct file) to chain. + */ + struct { + struct file *first; + struct file **last; + } children; + + /* For making a directory tree. */ + struct archive_string parentdir; + struct archive_string basename; + struct archive_string symlink; + + int ea_idx; + struct { + struct heap_data *first; + struct heap_data **last; + } xattr; + struct heap_data data; + struct archive_string script; + + signed int virtual:1; + signed int dir:1; +}; + +struct hardlink { + struct archive_rb_node rbnode; + int nlink; + struct { + struct file *first; + struct file **last; + } file_list; +}; + +struct xar { + int temp_fd; + uint64_t temp_offset; + + int file_idx; + struct file *root; + struct file *cur_dirent; + struct archive_string cur_dirstr; + struct file *cur_file; + uint64_t bytes_remaining; + struct archive_string tstr; + struct archive_string vstr; + + enum sumalg opt_toc_sumalg; + enum sumalg opt_sumalg; + enum enctype opt_compression; + int opt_compression_level; + uint32_t opt_threads; + + struct chksumwork a_sumwrk; /* archived checksum. */ + struct chksumwork e_sumwrk; /* extracted checksum. */ + struct la_zstream stream; + struct archive_string_conv *sconv; + /* + * Compressed data buffer. + */ + unsigned char wbuff[1024 * 64]; + size_t wbuff_remaining; + + struct heap_data toc; + /* + * The list of all file entries is used to manage struct file + * objects. + * We use 'next' (a member of struct file) to chain. + */ + struct { + struct file *first; + struct file **last; + } file_list; + /* + * The list of hard-linked file entries. + * We use 'hlnext' (a member of struct file) to chain. + */ + struct archive_rb_tree hardlink_rbtree; +}; + +static int xar_options(struct archive_write *, + const char *, const char *); +static int xar_write_header(struct archive_write *, + struct archive_entry *); +static ssize_t xar_write_data(struct archive_write *, + const void *, size_t); +static int xar_finish_entry(struct archive_write *); +static int xar_close(struct archive_write *); +static int xar_free(struct archive_write *); + +static struct file *file_new(struct archive_write *a, struct archive_entry *); +static void file_free(struct file *); +static struct file *file_create_virtual_dir(struct archive_write *a, struct xar *, + const char *); +static int file_add_child_tail(struct file *, struct file *); +static struct file *file_find_child(struct file *, const char *); +static int file_gen_utility_names(struct archive_write *, + struct file *); +static int get_path_component(char *, int, const char *); +static int file_tree(struct archive_write *, struct file **); +static void file_register(struct xar *, struct file *); +static void file_init_register(struct xar *); +static void file_free_register(struct xar *); +static int file_register_hardlink(struct archive_write *, + struct file *); +static void file_connect_hardlink_files(struct xar *); +static void file_init_hardlinks(struct xar *); +static void file_free_hardlinks(struct xar *); + +static void checksum_init(struct chksumwork *, enum sumalg); +static void checksum_update(struct chksumwork *, const void *, size_t); +static void checksum_final(struct chksumwork *, struct chksumval *); +static int compression_init_encoder_gzip(struct archive *, + struct la_zstream *, int, int); +static int compression_code_gzip(struct archive *, + struct la_zstream *, enum la_zaction); +static int compression_end_gzip(struct archive *, struct la_zstream *); +static int compression_init_encoder_bzip2(struct archive *, + struct la_zstream *, int); +#if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) +static int compression_code_bzip2(struct archive *, + struct la_zstream *, enum la_zaction); +static int compression_end_bzip2(struct archive *, struct la_zstream *); +#endif +static int compression_init_encoder_lzma(struct archive *, + struct la_zstream *, int); +static int compression_init_encoder_xz(struct archive *, + struct la_zstream *, int, int); +#if defined(HAVE_LZMA_H) +static int compression_code_lzma(struct archive *, + struct la_zstream *, enum la_zaction); +static int compression_end_lzma(struct archive *, struct la_zstream *); +#endif +static int xar_compression_init_encoder(struct archive_write *); +static int compression_code(struct archive *, + struct la_zstream *, enum la_zaction); +static int compression_end(struct archive *, + struct la_zstream *); +static int save_xattrs(struct archive_write *, struct file *); +static int getalgsize(enum sumalg); +static const char *getalgname(enum sumalg); + +int +archive_write_set_format_xar(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct xar *xar; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_set_format_xar"); + + /* If another format was already registered, unregister it. */ + if (a->format_free != NULL) + (a->format_free)(a); + + xar = calloc(1, sizeof(*xar)); + if (xar == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate xar data"); + return (ARCHIVE_FATAL); + } + xar->temp_fd = -1; + file_init_register(xar); + file_init_hardlinks(xar); + archive_string_init(&(xar->tstr)); + archive_string_init(&(xar->vstr)); + + /* + * Create the root directory. + */ + xar->root = file_create_virtual_dir(a, xar, ""); + if (xar->root == NULL) { + free(xar); + archive_set_error(&a->archive, ENOMEM, + "Can't allocate xar data"); + return (ARCHIVE_FATAL); + } + xar->root->parent = xar->root; + file_register(xar, xar->root); + xar->cur_dirent = xar->root; + archive_string_init(&(xar->cur_dirstr)); + archive_string_ensure(&(xar->cur_dirstr), 1); + xar->cur_dirstr.s[0] = 0; + + /* + * Initialize option. + */ + /* Set default checksum type. */ + xar->opt_toc_sumalg = CKSUM_SHA1; + xar->opt_sumalg = CKSUM_SHA1; + /* Set default compression type, level, and number of threads. */ + xar->opt_compression = GZIP; + xar->opt_compression_level = 6; + xar->opt_threads = 1; + + a->format_data = xar; + + a->format_name = "xar"; + a->format_options = xar_options; + a->format_write_header = xar_write_header; + a->format_write_data = xar_write_data; + a->format_finish_entry = xar_finish_entry; + a->format_close = xar_close; + a->format_free = xar_free; + a->archive.archive_format = ARCHIVE_FORMAT_XAR; + a->archive.archive_format_name = "xar"; + + return (ARCHIVE_OK); +} + +static int +xar_options(struct archive_write *a, const char *key, const char *value) +{ + struct xar *xar; + + xar = (struct xar *)a->format_data; + + if (strcmp(key, "checksum") == 0) { + if (value == NULL) + xar->opt_sumalg = CKSUM_NONE; + else if (strcmp(value, "none") == 0) + xar->opt_sumalg = CKSUM_NONE; + else if (strcmp(value, "sha1") == 0) + xar->opt_sumalg = CKSUM_SHA1; + else if (strcmp(value, "md5") == 0) + xar->opt_sumalg = CKSUM_MD5; + else { + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "Unknown checksum name: `%s'", + value); + return (ARCHIVE_FAILED); + } + return (ARCHIVE_OK); + } + if (strcmp(key, "compression") == 0) { + const char *name = NULL; + + if (value == NULL) + xar->opt_compression = NONE; + else if (strcmp(value, "none") == 0) + xar->opt_compression = NONE; + else if (strcmp(value, "gzip") == 0) + xar->opt_compression = GZIP; + else if (strcmp(value, "bzip2") == 0) +#if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) + xar->opt_compression = BZIP2; +#else + name = "bzip2"; +#endif + else if (strcmp(value, "lzma") == 0) +#if HAVE_LZMA_H + xar->opt_compression = LZMA; +#else + name = "lzma"; +#endif + else if (strcmp(value, "xz") == 0) +#if HAVE_LZMA_H + xar->opt_compression = XZ; +#else + name = "xz"; +#endif + else { + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "Unknown compression name: `%s'", + value); + return (ARCHIVE_FAILED); + } + if (name != NULL) { + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "`%s' compression not supported " + "on this platform", + name); + return (ARCHIVE_FAILED); + } + return (ARCHIVE_OK); + } + if (strcmp(key, "compression-level") == 0) { + if (value == NULL || + !(value[0] >= '0' && value[0] <= '9') || + value[1] != '\0') { + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "Illegal value `%s'", + value); + return (ARCHIVE_FAILED); + } + xar->opt_compression_level = value[0] - '0'; + return (ARCHIVE_OK); + } + if (strcmp(key, "toc-checksum") == 0) { + if (value == NULL) + xar->opt_toc_sumalg = CKSUM_NONE; + else if (strcmp(value, "none") == 0) + xar->opt_toc_sumalg = CKSUM_NONE; + else if (strcmp(value, "sha1") == 0) + xar->opt_toc_sumalg = CKSUM_SHA1; + else if (strcmp(value, "md5") == 0) + xar->opt_toc_sumalg = CKSUM_MD5; + else { + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "Unknown checksum name: `%s'", + value); + return (ARCHIVE_FAILED); + } + return (ARCHIVE_OK); + } + if (strcmp(key, "threads") == 0) { + char *endptr; + + if (value == NULL) + return (ARCHIVE_FAILED); + errno = 0; + xar->opt_threads = (int)strtoul(value, &endptr, 10); + if (errno != 0 || *endptr != '\0') { + xar->opt_threads = 1; + archive_set_error(&(a->archive), + ARCHIVE_ERRNO_MISC, + "Illegal value `%s'", + value); + return (ARCHIVE_FAILED); + } + if (xar->opt_threads == 0) { +#ifdef HAVE_LZMA_STREAM_ENCODER_MT + xar->opt_threads = lzma_cputhreads(); +#else + xar->opt_threads = 1; +#endif + } + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +static int +xar_write_header(struct archive_write *a, struct archive_entry *entry) +{ + struct xar *xar; + struct file *file; + struct archive_entry *file_entry; + int r, r2; + + xar = (struct xar *)a->format_data; + xar->cur_file = NULL; + xar->bytes_remaining = 0; + + if (xar->sconv == NULL) { + xar->sconv = archive_string_conversion_to_charset( + &a->archive, "UTF-8", 1); + if (xar->sconv == NULL) + return (ARCHIVE_FATAL); + } + + file = file_new(a, entry); + if (file == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate data"); + return (ARCHIVE_FATAL); + } + r2 = file_gen_utility_names(a, file); + if (r2 < ARCHIVE_WARN) + return (r2); + + /* + * Ignore a path which looks like the top of directory name + * since we have already made the root directory of an Xar archive. + */ + if (archive_strlen(&(file->parentdir)) == 0 && + archive_strlen(&(file->basename)) == 0) { + file_free(file); + return (r2); + } + + /* Add entry into tree */ + file_entry = file->entry; + r = file_tree(a, &file); + if (r != ARCHIVE_OK) + return (r); + /* There is the same file in tree and + * the current file is older than the file in tree. + * So we don't need the current file data anymore. */ + if (file->entry != file_entry) + return (r2); + if (file->id == 0) + file_register(xar, file); + + /* A virtual file, which is a directory, does not have + * any contents and we won't store it into a archive + * file other than its name. */ + if (file->virtual) + return (r2); + + /* + * Prepare to save the contents of the file. + */ + if (xar->temp_fd == -1) { + int algsize; + xar->temp_offset = 0; + xar->temp_fd = __archive_mktemp(NULL); + if (xar->temp_fd < 0) { + archive_set_error(&a->archive, errno, + "Couldn't create temporary file"); + return (ARCHIVE_FATAL); + } + algsize = getalgsize(xar->opt_toc_sumalg); + if (algsize > 0) { + if (lseek(xar->temp_fd, algsize, SEEK_SET) < 0) { + archive_set_error(&(a->archive), errno, + "lseek failed"); + return (ARCHIVE_FATAL); + } + xar->temp_offset = algsize; + } + } + + if (archive_entry_hardlink(file->entry) == NULL) { + r = save_xattrs(a, file); + if (r != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + + /* Non regular files contents are unneeded to be saved to + * a temporary file. */ + if (archive_entry_filetype(file->entry) != AE_IFREG) + return (r2); + + /* + * Set the current file to cur_file to read its contents. + */ + xar->cur_file = file; + + if (archive_entry_nlink(file->entry) > 1) { + r = file_register_hardlink(a, file); + if (r != ARCHIVE_OK) + return (r); + if (archive_entry_hardlink(file->entry) != NULL) { + archive_entry_unset_size(file->entry); + return (r2); + } + } + + /* Save a offset of current file in temporary file. */ + file->data.temp_offset = xar->temp_offset; + file->data.size = archive_entry_size(file->entry); + file->data.compression = xar->opt_compression; + xar->bytes_remaining = archive_entry_size(file->entry); + checksum_init(&(xar->a_sumwrk), xar->opt_sumalg); + checksum_init(&(xar->e_sumwrk), xar->opt_sumalg); + r = xar_compression_init_encoder(a); + + if (r != ARCHIVE_OK) + return (r); + else + return (r2); +} + +static int +write_to_temp(struct archive_write *a, const void *buff, size_t s) +{ + struct xar *xar; + const unsigned char *p; + ssize_t ws; + + xar = (struct xar *)a->format_data; + p = (const unsigned char *)buff; + while (s) { + ws = write(xar->temp_fd, p, s); + if (ws < 0) { + archive_set_error(&(a->archive), errno, + "fwrite function failed"); + return (ARCHIVE_FATAL); + } + s -= ws; + p += ws; + xar->temp_offset += ws; + } + return (ARCHIVE_OK); +} + +static ssize_t +xar_write_data(struct archive_write *a, const void *buff, size_t s) +{ + struct xar *xar; + enum la_zaction run; + size_t size = 0; + size_t rsize; + int r; + + xar = (struct xar *)a->format_data; + + if (s > xar->bytes_remaining) + s = (size_t)xar->bytes_remaining; + if (s == 0 || xar->cur_file == NULL) + return (0); + if (xar->cur_file->data.compression == NONE) { + checksum_update(&(xar->e_sumwrk), buff, s); + checksum_update(&(xar->a_sumwrk), buff, s); + size = rsize = s; + } else { + xar->stream.next_in = (const unsigned char *)buff; + xar->stream.avail_in = s; + if (xar->bytes_remaining > s) + run = ARCHIVE_Z_RUN; + else + run = ARCHIVE_Z_FINISH; + /* Compress file data. */ + for (;;) { + r = compression_code(&(a->archive), &(xar->stream), + run); + if (r != ARCHIVE_OK && r != ARCHIVE_EOF) + return (ARCHIVE_FATAL); + if (xar->stream.avail_out == 0 || + run == ARCHIVE_Z_FINISH) { + size = sizeof(xar->wbuff) - + xar->stream.avail_out; + checksum_update(&(xar->a_sumwrk), xar->wbuff, + size); + xar->cur_file->data.length += size; + if (write_to_temp(a, xar->wbuff, + size) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + if (r == ARCHIVE_OK) { + /* Output buffer was full */ + xar->stream.next_out = xar->wbuff; + xar->stream.avail_out = + sizeof(xar->wbuff); + } else { + /* ARCHIVE_EOF - We are done */ + break; + } + } else { + /* Compressor wants more input */ + break; + } + } + rsize = s - xar->stream.avail_in; + checksum_update(&(xar->e_sumwrk), buff, rsize); + } +#if !defined(_WIN32) || defined(__CYGWIN__) + if (xar->bytes_remaining == + (uint64_t)archive_entry_size(xar->cur_file->entry)) { + /* + * Get the path of a shell script if so. + */ + const unsigned char *b = (const unsigned char *)buff; + + archive_string_empty(&(xar->cur_file->script)); + if (rsize > 2 && b[0] == '#' && b[1] == '!') { + size_t i, end, off; + + off = 2; + if (b[off] == ' ') + off++; +#ifdef PATH_MAX + if ((rsize - off) > PATH_MAX) + end = off + PATH_MAX; + else +#endif + end = rsize; + /* Find the end of a script path. */ + for (i = off; i < end && b[i] != '\0' && + b[i] != '\n' && b[i] != '\r' && + b[i] != ' ' && b[i] != '\t'; i++) + ; + archive_strncpy(&(xar->cur_file->script), b + off, + i - off); + } + } +#endif + + if (xar->cur_file->data.compression == NONE) { + if (write_to_temp(a, buff, size) != ARCHIVE_OK) + return (ARCHIVE_FATAL); + xar->cur_file->data.length += size; + } + xar->bytes_remaining -= rsize; + + return (rsize); +} + +static int +xar_finish_entry(struct archive_write *a) +{ + struct xar *xar; + struct file *file; + size_t s; + ssize_t w; + + xar = (struct xar *)a->format_data; + if (xar->cur_file == NULL) + return (ARCHIVE_OK); + + while (xar->bytes_remaining > 0) { + s = (size_t)xar->bytes_remaining; + if (s > a->null_length) + s = a->null_length; + w = xar_write_data(a, a->nulls, s); + if (w > 0) + xar->bytes_remaining -= w; + else + return (w); + } + file = xar->cur_file; + checksum_final(&(xar->e_sumwrk), &(file->data.e_sum)); + checksum_final(&(xar->a_sumwrk), &(file->data.a_sum)); + xar->cur_file = NULL; + + return (ARCHIVE_OK); +} + +static int +xmlwrite_string_attr(struct archive_write *a, xmlTextWriterPtr writer, + const char *key, const char *value, + const char *attrkey, const char *attrvalue) +{ + int r; + + r = xmlTextWriterStartElement(writer, BAD_CAST_CONST(key)); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterStartElement() failed: %d", r); + return (ARCHIVE_FATAL); + } + if (attrkey != NULL && attrvalue != NULL) { + r = xmlTextWriterWriteAttribute(writer, + BAD_CAST_CONST(attrkey), BAD_CAST_CONST(attrvalue)); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterWriteAttribute() failed: %d", r); + return (ARCHIVE_FATAL); + } + } + if (value != NULL) { + r = xmlTextWriterWriteString(writer, BAD_CAST_CONST(value)); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterWriteString() failed: %d", r); + return (ARCHIVE_FATAL); + } + } + r = xmlTextWriterEndElement(writer); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterEndElement() failed: %d", r); + return (ARCHIVE_FATAL); + } + return (ARCHIVE_OK); +} + +static int +xmlwrite_string(struct archive_write *a, xmlTextWriterPtr writer, + const char *key, const char *value) +{ + int r; + + if (value == NULL) + return (ARCHIVE_OK); + + r = xmlTextWriterStartElement(writer, BAD_CAST_CONST(key)); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterStartElement() failed: %d", r); + return (ARCHIVE_FATAL); + } + if (value != NULL) { + r = xmlTextWriterWriteString(writer, BAD_CAST_CONST(value)); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterWriteString() failed: %d", r); + return (ARCHIVE_FATAL); + } + } + r = xmlTextWriterEndElement(writer); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterEndElement() failed: %d", r); + return (ARCHIVE_FATAL); + } + return (ARCHIVE_OK); +} + +static int +xmlwrite_fstring(struct archive_write *a, xmlTextWriterPtr writer, + const char *key, const char *fmt, ...) +{ + struct xar *xar; + va_list ap; + + xar = (struct xar *)a->format_data; + va_start(ap, fmt); + archive_string_empty(&xar->vstr); + archive_string_vsprintf(&xar->vstr, fmt, ap); + va_end(ap); + return (xmlwrite_string(a, writer, key, xar->vstr.s)); +} + +static int +xmlwrite_time(struct archive_write *a, xmlTextWriterPtr writer, + const char *key, time_t t, int z) +{ + char timestr[100]; + struct tm tm; +#if defined(HAVE__GMTIME64_S) + __time64_t tmptime; +#endif + +#if defined(HAVE_GMTIME_R) + gmtime_r(&t, &tm); +#elif defined(HAVE__GMTIME64_S) + tmptime = t; + _gmtime64_s(&tm, &tmptime); +#else + memcpy(&tm, gmtime(&t), sizeof(tm)); +#endif + memset(×tr, 0, sizeof(timestr)); + /* Do not use %F and %T for portability. */ + strftime(timestr, sizeof(timestr), "%Y-%m-%dT%H:%M:%S", &tm); + if (z) + strcat(timestr, "Z"); + return (xmlwrite_string(a, writer, key, timestr)); +} + +static int +xmlwrite_mode(struct archive_write *a, xmlTextWriterPtr writer, + const char *key, mode_t mode) +{ + char ms[5]; + + ms[0] = '0'; + ms[1] = '0' + ((mode >> 6) & 07); + ms[2] = '0' + ((mode >> 3) & 07); + ms[3] = '0' + (mode & 07); + ms[4] = '\0'; + + return (xmlwrite_string(a, writer, key, ms)); +} + +static int +xmlwrite_sum(struct archive_write *a, xmlTextWriterPtr writer, + const char *key, struct chksumval *sum) +{ + const char *algname; + int algsize; + char buff[MAX_SUM_SIZE*2 + 1]; + char *p; + unsigned char *s; + int i, r; + + if (sum->len > 0) { + algname = getalgname(sum->alg); + algsize = getalgsize(sum->alg); + if (algname != NULL) { + const char *hex = "0123456789abcdef"; + p = buff; + s = sum->val; + for (i = 0; i < algsize; i++) { + *p++ = hex[(*s >> 4)]; + *p++ = hex[(*s & 0x0f)]; + s++; + } + *p = '\0'; + r = xmlwrite_string_attr(a, writer, + key, buff, + "style", algname); + if (r < 0) + return (ARCHIVE_FATAL); + } + } + return (ARCHIVE_OK); +} + +static int +xmlwrite_heap(struct archive_write *a, xmlTextWriterPtr writer, + struct heap_data *heap) +{ + const char *encname; + int r; + + r = xmlwrite_fstring(a, writer, "length", "%ju", heap->length); + if (r < 0) + return (ARCHIVE_FATAL); + r = xmlwrite_fstring(a, writer, "offset", "%ju", heap->temp_offset); + if (r < 0) + return (ARCHIVE_FATAL); + r = xmlwrite_fstring(a, writer, "size", "%ju", heap->size); + if (r < 0) + return (ARCHIVE_FATAL); + switch (heap->compression) { + case GZIP: + encname = "application/x-gzip"; break; + case BZIP2: + encname = "application/x-bzip2"; break; + case LZMA: + encname = "application/x-lzma"; break; + case XZ: + encname = "application/x-xz"; break; + default: + encname = "application/octet-stream"; break; + } + r = xmlwrite_string_attr(a, writer, "encoding", NULL, + "style", encname); + if (r < 0) + return (ARCHIVE_FATAL); + r = xmlwrite_sum(a, writer, "archived-checksum", &(heap->a_sum)); + if (r < 0) + return (ARCHIVE_FATAL); + r = xmlwrite_sum(a, writer, "extracted-checksum", &(heap->e_sum)); + if (r < 0) + return (ARCHIVE_FATAL); + return (ARCHIVE_OK); +} + +/* + * xar utility records fflags as following xml elements: + * + * + * ..... + * + * or + * + * + * ..... + * + * If xar is running on BSD platform, records ..; + * if xar is running on linux platform, records ..; + * otherwise does not record. + * + * Our implements records both and if it's necessary. + */ +static int +make_fflags_entry(struct archive_write *a, xmlTextWriterPtr writer, + const char *element, const char *fflags_text) +{ + static const struct flagentry { + const char *name; + const char *xarname; + } + flagbsd[] = { + { "sappnd", "SystemAppend"}, + { "sappend", "SystemAppend"}, + { "arch", "SystemArchived"}, + { "archived", "SystemArchived"}, + { "schg", "SystemImmutable"}, + { "schange", "SystemImmutable"}, + { "simmutable", "SystemImmutable"}, + { "nosunlnk", "SystemNoUnlink"}, + { "nosunlink", "SystemNoUnlink"}, + { "snapshot", "SystemSnapshot"}, + { "uappnd", "UserAppend"}, + { "uappend", "UserAppend"}, + { "uchg", "UserImmutable"}, + { "uchange", "UserImmutable"}, + { "uimmutable", "UserImmutable"}, + { "nodump", "UserNoDump"}, + { "noopaque", "UserOpaque"}, + { "nouunlnk", "UserNoUnlink"}, + { "nouunlink", "UserNoUnlink"}, + { NULL, NULL} + }, + flagext2[] = { + { "sappnd", "AppendOnly"}, + { "sappend", "AppendOnly"}, + { "schg", "Immutable"}, + { "schange", "Immutable"}, + { "simmutable", "Immutable"}, + { "nodump", "NoDump"}, + { "nouunlnk", "Undelete"}, + { "nouunlink", "Undelete"}, + { "btree", "BTree"}, + { "comperr", "CompError"}, + { "compress", "Compress"}, + { "noatime", "NoAtime"}, + { "compdirty", "CompDirty"}, + { "comprblk", "CompBlock"}, + { "dirsync", "DirSync"}, + { "hashidx", "HashIndexed"}, + { "imagic", "iMagic"}, + { "journal", "Journaled"}, + { "securedeletion", "SecureDeletion"}, + { "sync", "Synchronous"}, + { "notail", "NoTail"}, + { "topdir", "TopDir"}, + { "reserved", "Reserved"}, + { NULL, NULL} + }; + const struct flagentry *fe, *flagentry; +#define FLAGENTRY_MAXSIZE ((sizeof(flagbsd)+sizeof(flagext2))/sizeof(flagbsd)) + const struct flagentry *avail[FLAGENTRY_MAXSIZE]; + const char *p; + int i, n, r; + + if (strcmp(element, "ext2") == 0) + flagentry = flagext2; + else + flagentry = flagbsd; + n = 0; + p = fflags_text; + do { + const char *cp; + + cp = strchr(p, ','); + if (cp == NULL) + cp = p + strlen(p); + + for (fe = flagentry; fe->name != NULL; fe++) { + if (fe->name[cp - p] != '\0' + || p[0] != fe->name[0]) + continue; + if (strncmp(p, fe->name, cp - p) == 0) { + avail[n++] = fe; + break; + } + } + if (*cp == ',') + p = cp + 1; + else + p = NULL; + } while (p != NULL); + + if (n > 0) { + r = xmlTextWriterStartElement(writer, BAD_CAST_CONST(element)); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterStartElement() failed: %d", r); + return (ARCHIVE_FATAL); + } + for (i = 0; i < n; i++) { + r = xmlwrite_string(a, writer, + avail[i]->xarname, NULL); + if (r != ARCHIVE_OK) + return (r); + } + + r = xmlTextWriterEndElement(writer); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterEndElement() failed: %d", r); + return (ARCHIVE_FATAL); + } + } + return (ARCHIVE_OK); +} + +static int +make_file_entry(struct archive_write *a, xmlTextWriterPtr writer, + struct file *file) +{ + struct xar *xar; + const char *filetype, *filelink, *fflags; + struct archive_string linkto; + struct heap_data *heap; + unsigned char *tmp; + const char *p; + size_t len; + int r, r2, l, ll; + + xar = (struct xar *)a->format_data; + r2 = ARCHIVE_OK; + + /* + * Make a file name entry, "". + */ + l = ll = archive_strlen(&(file->basename)); + tmp = malloc(l); + if (tmp == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + r = UTF8Toisolat1(tmp, &l, BAD_CAST(file->basename.s), &ll); + free(tmp); + if (r < 0) { + r = xmlTextWriterStartElement(writer, BAD_CAST("name")); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterStartElement() failed: %d", r); + return (ARCHIVE_FATAL); + } + r = xmlTextWriterWriteAttribute(writer, + BAD_CAST("enctype"), BAD_CAST("base64")); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterWriteAttribute() failed: %d", r); + return (ARCHIVE_FATAL); + } + r = xmlTextWriterWriteBase64(writer, file->basename.s, + 0, archive_strlen(&(file->basename))); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterWriteBase64() failed: %d", r); + return (ARCHIVE_FATAL); + } + r = xmlTextWriterEndElement(writer); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterEndElement() failed: %d", r); + return (ARCHIVE_FATAL); + } + } else { + r = xmlwrite_string(a, writer, "name", file->basename.s); + if (r < 0) + return (ARCHIVE_FATAL); + } + + /* + * Make a file type entry, "". + */ + filelink = NULL; + archive_string_init(&linkto); + switch (archive_entry_filetype(file->entry)) { + case AE_IFDIR: + filetype = "directory"; break; + case AE_IFLNK: + filetype = "symlink"; break; + case AE_IFCHR: + filetype = "character special"; break; + case AE_IFBLK: + filetype = "block special"; break; + case AE_IFSOCK: + filetype = "socket"; break; + case AE_IFIFO: + filetype = "fifo"; break; + case AE_IFREG: + default: + if (file->hardlink_target != NULL) { + filetype = "hardlink"; + filelink = "link"; + if (file->hardlink_target == file) + archive_strcpy(&linkto, "original"); + else + archive_string_sprintf(&linkto, "%d", + file->hardlink_target->id); + } else + filetype = "file"; + break; + } + r = xmlwrite_string_attr(a, writer, "type", filetype, + filelink, linkto.s); + archive_string_free(&linkto); + if (r < 0) + return (ARCHIVE_FATAL); + + /* + * On a virtual directory, we record "name" and "type" only. + */ + if (file->virtual) + return (ARCHIVE_OK); + + switch (archive_entry_filetype(file->entry)) { + case AE_IFLNK: + /* + * xar utility has checked a file type, which + * a symbolic-link file has referenced. + * For example: + * ../ref/ + * The symlink target file is "../ref/" and its + * file type is a directory. + * + * ../f + * The symlink target file is "../f" and its + * file type is a regular file. + * + * But our implementation cannot do it, and then we + * always record that a attribute "type" is "broken", + * for example: + * foo/bar + * It means "foo/bar" is not reachable. + */ + r = xmlwrite_string_attr(a, writer, "link", + file->symlink.s, + "type", "broken"); + if (r < 0) + return (ARCHIVE_FATAL); + break; + case AE_IFCHR: + case AE_IFBLK: + r = xmlTextWriterStartElement(writer, BAD_CAST("device")); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterStartElement() failed: %d", r); + return (ARCHIVE_FATAL); + } + r = xmlwrite_fstring(a, writer, "major", + "%d", archive_entry_rdevmajor(file->entry)); + if (r < 0) + return (ARCHIVE_FATAL); + r = xmlwrite_fstring(a, writer, "minor", + "%d", archive_entry_rdevminor(file->entry)); + if (r < 0) + return (ARCHIVE_FATAL); + r = xmlTextWriterEndElement(writer); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterEndElement() failed: %d", r); + return (ARCHIVE_FATAL); + } + break; + default: + break; + } + + /* + * Make a inode entry, "". + */ + r = xmlwrite_fstring(a, writer, "inode", + "%jd", archive_entry_ino64(file->entry)); + if (r < 0) + return (ARCHIVE_FATAL); + if (archive_entry_dev(file->entry) != 0) { + r = xmlwrite_fstring(a, writer, "deviceno", + "%d", archive_entry_dev(file->entry)); + if (r < 0) + return (ARCHIVE_FATAL); + } + + /* + * Make a file mode entry, "". + */ + r = xmlwrite_mode(a, writer, "mode", + archive_entry_mode(file->entry)); + if (r < 0) + return (ARCHIVE_FATAL); + + /* + * Make a user entry, "" and ". + */ + r = xmlwrite_fstring(a, writer, "uid", + "%d", archive_entry_uid(file->entry)); + if (r < 0) + return (ARCHIVE_FATAL); + r = archive_entry_uname_l(file->entry, &p, &len, xar->sconv); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Uname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate uname '%s' to UTF-8", + archive_entry_uname(file->entry)); + r2 = ARCHIVE_WARN; + } + if (len > 0) { + r = xmlwrite_string(a, writer, "user", p); + if (r < 0) + return (ARCHIVE_FATAL); + } + + /* + * Make a group entry, "" and ". + */ + r = xmlwrite_fstring(a, writer, "gid", + "%d", archive_entry_gid(file->entry)); + if (r < 0) + return (ARCHIVE_FATAL); + r = archive_entry_gname_l(file->entry, &p, &len, xar->sconv); + if (r != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Gname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate gname '%s' to UTF-8", + archive_entry_gname(file->entry)); + r2 = ARCHIVE_WARN; + } + if (len > 0) { + r = xmlwrite_string(a, writer, "group", p); + if (r < 0) + return (ARCHIVE_FATAL); + } + + /* + * Make a ctime entry, "". + */ + if (archive_entry_ctime_is_set(file->entry)) { + r = xmlwrite_time(a, writer, "ctime", + archive_entry_ctime(file->entry), 1); + if (r < 0) + return (ARCHIVE_FATAL); + } + + /* + * Make a mtime entry, "". + */ + if (archive_entry_mtime_is_set(file->entry)) { + r = xmlwrite_time(a, writer, "mtime", + archive_entry_mtime(file->entry), 1); + if (r < 0) + return (ARCHIVE_FATAL); + } + + /* + * Make a atime entry, "". + */ + if (archive_entry_atime_is_set(file->entry)) { + r = xmlwrite_time(a, writer, "atime", + archive_entry_atime(file->entry), 1); + if (r < 0) + return (ARCHIVE_FATAL); + } + + /* + * Make fflags entries, "" and "". + */ + fflags = archive_entry_fflags_text(file->entry); + if (fflags != NULL) { + r = make_fflags_entry(a, writer, "flags", fflags); + if (r < 0) + return (r); + r = make_fflags_entry(a, writer, "ext2", fflags); + if (r < 0) + return (r); + } + + /* + * Make extended attribute entries, "". + */ + archive_entry_xattr_reset(file->entry); + for (heap = file->xattr.first; heap != NULL; heap = heap->next) { + const char *name; + const void *value; + size_t size; + + archive_entry_xattr_next(file->entry, + &name, &value, &size); + r = xmlTextWriterStartElement(writer, BAD_CAST("ea")); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterStartElement() failed: %d", r); + return (ARCHIVE_FATAL); + } + r = xmlTextWriterWriteFormatAttribute(writer, + BAD_CAST("id"), "%d", heap->id); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterWriteAttribute() failed: %d", r); + return (ARCHIVE_FATAL); + } + r = xmlwrite_heap(a, writer, heap); + if (r < 0) + return (ARCHIVE_FATAL); + r = xmlwrite_string(a, writer, "name", name); + if (r < 0) + return (ARCHIVE_FATAL); + + r = xmlTextWriterEndElement(writer); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterEndElement() failed: %d", r); + return (ARCHIVE_FATAL); + } + } + + /* + * Make a file data entry, "". + */ + if (file->data.length > 0) { + r = xmlTextWriterStartElement(writer, BAD_CAST("data")); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterStartElement() failed: %d", r); + return (ARCHIVE_FATAL); + } + + r = xmlwrite_heap(a, writer, &(file->data)); + if (r < 0) + return (ARCHIVE_FATAL); + + r = xmlTextWriterEndElement(writer); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterEndElement() failed: %d", r); + return (ARCHIVE_FATAL); + } + } + + if (archive_strlen(&file->script) > 0) { + r = xmlTextWriterStartElement(writer, BAD_CAST("content")); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterStartElement() failed: %d", r); + return (ARCHIVE_FATAL); + } + + r = xmlwrite_string(a, writer, + "interpreter", file->script.s); + if (r < 0) + return (ARCHIVE_FATAL); + + r = xmlwrite_string(a, writer, "type", "script"); + if (r < 0) + return (ARCHIVE_FATAL); + + r = xmlTextWriterEndElement(writer); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterEndElement() failed: %d", r); + return (ARCHIVE_FATAL); + } + } + + return (r2); +} + +/* + * Make the TOC + */ +static int +make_toc(struct archive_write *a) +{ + struct xar *xar; + struct file *np; + xmlBufferPtr bp; + xmlTextWriterPtr writer; + int algsize; + int r, ret; + + xar = (struct xar *)a->format_data; + + ret = ARCHIVE_FATAL; + + /* + * Initialize xml writer. + */ + writer = NULL; + bp = xmlBufferCreate(); + if (bp == NULL) { + archive_set_error(&a->archive, ENOMEM, + "xmlBufferCreate() " + "couldn't create xml buffer"); + goto exit_toc; + } + writer = xmlNewTextWriterMemory(bp, 0); + if (writer == NULL) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlNewTextWriterMemory() " + "couldn't create xml writer"); + goto exit_toc; + } + r = xmlTextWriterStartDocument(writer, "1.0", "UTF-8", NULL); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterStartDocument() failed: %d", r); + goto exit_toc; + } + r = xmlTextWriterSetIndent(writer, 4); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterSetIndent() failed: %d", r); + goto exit_toc; + } + + /* + * Start recording TOC + */ + r = xmlTextWriterStartElement(writer, BAD_CAST("xar")); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterStartElement() failed: %d", r); + goto exit_toc; + } + r = xmlTextWriterStartElement(writer, BAD_CAST("toc")); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterStartDocument() failed: %d", r); + goto exit_toc; + } + + /* + * Record the creation time of the archive file. + */ + r = xmlwrite_time(a, writer, "creation-time", time(NULL), 0); + if (r < 0) + goto exit_toc; + + /* + * Record the checksum value of TOC + */ + algsize = getalgsize(xar->opt_toc_sumalg); + if (algsize) { + /* + * Record TOC checksum + */ + r = xmlTextWriterStartElement(writer, BAD_CAST("checksum")); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterStartElement() failed: %d", r); + goto exit_toc; + } + r = xmlTextWriterWriteAttribute(writer, BAD_CAST("style"), + BAD_CAST_CONST(getalgname(xar->opt_toc_sumalg))); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterWriteAttribute() failed: %d", r); + goto exit_toc; + } + + /* + * Record the offset of the value of checksum of TOC + */ + r = xmlwrite_string(a, writer, "offset", "0"); + if (r < 0) + goto exit_toc; + + /* + * Record the size of the value of checksum of TOC + */ + r = xmlwrite_fstring(a, writer, "size", "%d", algsize); + if (r < 0) + goto exit_toc; + + r = xmlTextWriterEndElement(writer); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterEndElement() failed: %d", r); + goto exit_toc; + } + } + + np = xar->root; + do { + if (np != np->parent) { + r = make_file_entry(a, writer, np); + if (r != ARCHIVE_OK) + goto exit_toc; + } + + if (np->dir && np->children.first != NULL) { + /* Enter to sub directories. */ + np = np->children.first; + r = xmlTextWriterStartElement(writer, + BAD_CAST("file")); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterStartElement() " + "failed: %d", r); + goto exit_toc; + } + r = xmlTextWriterWriteFormatAttribute( + writer, BAD_CAST("id"), "%d", np->id); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterWriteAttribute() " + "failed: %d", r); + goto exit_toc; + } + continue; + } + while (np != np->parent) { + r = xmlTextWriterEndElement(writer); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterEndElement() " + "failed: %d", r); + goto exit_toc; + } + if (np->chnext == NULL) { + /* Return to the parent directory. */ + np = np->parent; + } else { + np = np->chnext; + r = xmlTextWriterStartElement(writer, + BAD_CAST("file")); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterStartElement() " + "failed: %d", r); + goto exit_toc; + } + r = xmlTextWriterWriteFormatAttribute( + writer, BAD_CAST("id"), "%d", np->id); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterWriteAttribute() " + "failed: %d", r); + goto exit_toc; + } + break; + } + } + } while (np != np->parent); + + r = xmlTextWriterEndDocument(writer); + if (r < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "xmlTextWriterEndDocument() failed: %d", r); + goto exit_toc; + } +#if DEBUG_PRINT_TOC + fprintf(stderr, "\n---TOC-- %d bytes --\n%s\n", + strlen((const char *)bp->content), bp->content); +#endif + + /* + * Compress the TOC and calculate the sum of the TOC. + */ + xar->toc.temp_offset = xar->temp_offset; + xar->toc.size = bp->use; + checksum_init(&(xar->a_sumwrk), xar->opt_toc_sumalg); + + r = compression_init_encoder_gzip(&(a->archive), + &(xar->stream), 6, 1); + if (r != ARCHIVE_OK) + goto exit_toc; + xar->stream.next_in = bp->content; + xar->stream.avail_in = bp->use; + xar->stream.total_in = 0; + xar->stream.next_out = xar->wbuff; + xar->stream.avail_out = sizeof(xar->wbuff); + xar->stream.total_out = 0; + for (;;) { + size_t size; + + r = compression_code(&(a->archive), + &(xar->stream), ARCHIVE_Z_FINISH); + if (r != ARCHIVE_OK && r != ARCHIVE_EOF) + goto exit_toc; + size = sizeof(xar->wbuff) - xar->stream.avail_out; + checksum_update(&(xar->a_sumwrk), xar->wbuff, size); + if (write_to_temp(a, xar->wbuff, size) != ARCHIVE_OK) + goto exit_toc; + if (r == ARCHIVE_EOF) + break; + xar->stream.next_out = xar->wbuff; + xar->stream.avail_out = sizeof(xar->wbuff); + } + r = compression_end(&(a->archive), &(xar->stream)); + if (r != ARCHIVE_OK) + goto exit_toc; + xar->toc.length = xar->stream.total_out; + xar->toc.compression = GZIP; + checksum_final(&(xar->a_sumwrk), &(xar->toc.a_sum)); + + ret = ARCHIVE_OK; +exit_toc: + if (writer) + xmlFreeTextWriter(writer); + if (bp) + xmlBufferFree(bp); + + return (ret); +} + +static int +flush_wbuff(struct archive_write *a) +{ + struct xar *xar; + int r; + size_t s; + + xar = (struct xar *)a->format_data; + s = sizeof(xar->wbuff) - xar->wbuff_remaining; + r = __archive_write_output(a, xar->wbuff, s); + if (r != ARCHIVE_OK) + return (r); + xar->wbuff_remaining = sizeof(xar->wbuff); + return (r); +} + +static int +copy_out(struct archive_write *a, uint64_t offset, uint64_t length) +{ + struct xar *xar; + int r; + + xar = (struct xar *)a->format_data; + if (lseek(xar->temp_fd, offset, SEEK_SET) < 0) { + archive_set_error(&(a->archive), errno, "lseek failed"); + return (ARCHIVE_FATAL); + } + while (length) { + size_t rsize; + ssize_t rs; + unsigned char *wb; + + if (length > xar->wbuff_remaining) + rsize = xar->wbuff_remaining; + else + rsize = (size_t)length; + wb = xar->wbuff + (sizeof(xar->wbuff) - xar->wbuff_remaining); + rs = read(xar->temp_fd, wb, rsize); + if (rs < 0) { + archive_set_error(&(a->archive), errno, + "Can't read temporary file(%jd)", + (intmax_t)rs); + return (ARCHIVE_FATAL); + } + if (rs == 0) { + archive_set_error(&(a->archive), 0, + "Truncated xar archive"); + return (ARCHIVE_FATAL); + } + xar->wbuff_remaining -= rs; + length -= rs; + if (xar->wbuff_remaining == 0) { + r = flush_wbuff(a); + if (r != ARCHIVE_OK) + return (r); + } + } + return (ARCHIVE_OK); +} + +static int +xar_close(struct archive_write *a) +{ + struct xar *xar; + unsigned char *wb; + uint64_t length; + int r; + + xar = (struct xar *)a->format_data; + + /* Empty! */ + if (xar->root->children.first == NULL) + return (ARCHIVE_OK); + + /* Save the length of all file extended attributes and contents. */ + length = xar->temp_offset; + + /* Connect hardlinked files */ + file_connect_hardlink_files(xar); + + /* Make the TOC */ + r = make_toc(a); + if (r != ARCHIVE_OK) + return (r); + /* + * Make the xar header on wbuff(write buffer). + */ + wb = xar->wbuff; + xar->wbuff_remaining = sizeof(xar->wbuff); + archive_be32enc(&wb[0], HEADER_MAGIC); + archive_be16enc(&wb[4], HEADER_SIZE); + archive_be16enc(&wb[6], HEADER_VERSION); + archive_be64enc(&wb[8], xar->toc.length); + archive_be64enc(&wb[16], xar->toc.size); + archive_be32enc(&wb[24], xar->toc.a_sum.alg); + xar->wbuff_remaining -= HEADER_SIZE; + + /* + * Write the TOC + */ + r = copy_out(a, xar->toc.temp_offset, xar->toc.length); + if (r != ARCHIVE_OK) + return (r); + + /* Write the checksum value of the TOC. */ + if (xar->toc.a_sum.len) { + if (xar->wbuff_remaining < xar->toc.a_sum.len) { + r = flush_wbuff(a); + if (r != ARCHIVE_OK) + return (r); + } + wb = xar->wbuff + (sizeof(xar->wbuff) - xar->wbuff_remaining); + memcpy(wb, xar->toc.a_sum.val, xar->toc.a_sum.len); + xar->wbuff_remaining -= xar->toc.a_sum.len; + } + + /* + * Write all file extended attributes and contents. + */ + r = copy_out(a, xar->toc.a_sum.len, length); + if (r != ARCHIVE_OK) + return (r); + r = flush_wbuff(a); + return (r); +} + +static int +xar_free(struct archive_write *a) +{ + struct xar *xar; + + xar = (struct xar *)a->format_data; + + /* Close the temporary file. */ + if (xar->temp_fd >= 0) + close(xar->temp_fd); + + archive_string_free(&(xar->cur_dirstr)); + archive_string_free(&(xar->tstr)); + archive_string_free(&(xar->vstr)); + file_free_hardlinks(xar); + file_free_register(xar); + compression_end(&(a->archive), &(xar->stream)); + free(xar); + + return (ARCHIVE_OK); +} + +static int +file_cmp_node(const struct archive_rb_node *n1, + const struct archive_rb_node *n2) +{ + const struct file *f1 = (const struct file *)n1; + const struct file *f2 = (const struct file *)n2; + + return (strcmp(f1->basename.s, f2->basename.s)); +} + +static int +file_cmp_key(const struct archive_rb_node *n, const void *key) +{ + const struct file *f = (const struct file *)n; + + return (strcmp(f->basename.s, (const char *)key)); +} + +static struct file * +file_new(struct archive_write *a, struct archive_entry *entry) +{ + struct file *file; + static const struct archive_rb_tree_ops rb_ops = { + file_cmp_node, file_cmp_key + }; + + file = calloc(1, sizeof(*file)); + if (file == NULL) + return (NULL); + + if (entry != NULL) + file->entry = archive_entry_clone(entry); + else + file->entry = archive_entry_new2(&a->archive); + if (file->entry == NULL) { + free(file); + return (NULL); + } + __archive_rb_tree_init(&(file->rbtree), &rb_ops); + file->children.first = NULL; + file->children.last = &(file->children.first); + file->xattr.first = NULL; + file->xattr.last = &(file->xattr.first); + archive_string_init(&(file->parentdir)); + archive_string_init(&(file->basename)); + archive_string_init(&(file->symlink)); + archive_string_init(&(file->script)); + if (entry != NULL && archive_entry_filetype(entry) == AE_IFDIR) + file->dir = 1; + + return (file); +} + +static void +file_free(struct file *file) +{ + struct heap_data *heap, *next_heap; + + heap = file->xattr.first; + while (heap != NULL) { + next_heap = heap->next; + free(heap); + heap = next_heap; + } + archive_string_free(&(file->parentdir)); + archive_string_free(&(file->basename)); + archive_string_free(&(file->symlink)); + archive_string_free(&(file->script)); + archive_entry_free(file->entry); + free(file); +} + +static struct file * +file_create_virtual_dir(struct archive_write *a, struct xar *xar, + const char *pathname) +{ + struct file *file; + + (void)xar; /* UNUSED */ + + file = file_new(a, NULL); + if (file == NULL) + return (NULL); + archive_entry_set_pathname(file->entry, pathname); + archive_entry_set_mode(file->entry, 0555 | AE_IFDIR); + + file->dir = 1; + file->virtual = 1; + + return (file); +} + +static int +file_add_child_tail(struct file *parent, struct file *child) +{ + if (!__archive_rb_tree_insert_node( + &(parent->rbtree), (struct archive_rb_node *)child)) + return (0); + child->chnext = NULL; + *parent->children.last = child; + parent->children.last = &(child->chnext); + child->parent = parent; + return (1); +} + +/* + * Find a entry from `parent' + */ +static struct file * +file_find_child(struct file *parent, const char *child_name) +{ + struct file *np; + + np = (struct file *)__archive_rb_tree_find_node( + &(parent->rbtree), child_name); + return (np); +} + +#if defined(_WIN32) || defined(__CYGWIN__) +static void +cleanup_backslash(char *utf8, size_t len) +{ + + /* Convert a path-separator from '\' to '/' */ + while (*utf8 != '\0' && len) { + if (*utf8 == '\\') + *utf8 = '/'; + ++utf8; + --len; + } +} +#else +#define cleanup_backslash(p, len) /* nop */ +#endif + +/* + * Generate a parent directory name and a base name from a pathname. + */ +static int +file_gen_utility_names(struct archive_write *a, struct file *file) +{ + struct xar *xar; + const char *pp; + char *p, *dirname, *slash; + size_t len; + int r = ARCHIVE_OK; + + xar = (struct xar *)a->format_data; + archive_string_empty(&(file->parentdir)); + archive_string_empty(&(file->basename)); + archive_string_empty(&(file->symlink)); + + if (file->parent == file)/* virtual root */ + return (ARCHIVE_OK); + + if (archive_entry_pathname_l(file->entry, &pp, &len, xar->sconv) + != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Pathname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate pathname '%s' to UTF-8", + archive_entry_pathname(file->entry)); + r = ARCHIVE_WARN; + } + archive_strncpy(&(file->parentdir), pp, len); + len = file->parentdir.length; + p = dirname = file->parentdir.s; + /* + * Convert a path-separator from '\' to '/' + */ + cleanup_backslash(p, len); + + /* + * Remove leading '/', '../' and './' elements + */ + while (*p) { + if (p[0] == '/') { + p++; + len--; + } else if (p[0] != '.') + break; + else if (p[1] == '.' && p[2] == '/') { + p += 3; + len -= 3; + } else if (p[1] == '/' || (p[1] == '.' && p[2] == '\0')) { + p += 2; + len -= 2; + } else if (p[1] == '\0') { + p++; + len--; + } else + break; + } + if (p != dirname) { + memmove(dirname, p, len+1); + p = dirname; + } + /* + * Remove "/","/." and "/.." elements from tail. + */ + while (len > 0) { + size_t ll = len; + + if (p[len-1] == '/') { + p[len-1] = '\0'; + len--; + } + if (len > 1 && p[len-2] == '/' && p[len-1] == '.') { + p[len-2] = '\0'; + len -= 2; + } + if (len > 2 && p[len-3] == '/' && p[len-2] == '.' && + p[len-1] == '.') { + p[len-3] = '\0'; + len -= 3; + } + if (ll == len) + break; + } + while (*p) { + if (p[0] == '/') { + if (p[1] == '/') + /* Convert '//' --> '/' */ + memmove(p, p+1, strlen(p+1) + 1); + else if (p[1] == '.' && p[2] == '/') + /* Convert '/./' --> '/' */ + memmove(p, p+2, strlen(p+2) + 1); + else if (p[1] == '.' && p[2] == '.' && p[3] == '/') { + /* Convert 'dir/dir1/../dir2/' + * --> 'dir/dir2/' + */ + char *rp = p -1; + while (rp >= dirname) { + if (*rp == '/') + break; + --rp; + } + if (rp > dirname) { + strcpy(rp, p+3); + p = rp; + } else { + strcpy(dirname, p+4); + p = dirname; + } + } else + p++; + } else + p++; + } + p = dirname; + len = strlen(p); + + if (archive_entry_filetype(file->entry) == AE_IFLNK) { + size_t len2; + /* Convert symlink name too. */ + if (archive_entry_symlink_l(file->entry, &pp, &len2, + xar->sconv) != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Linkname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate symlink '%s' to UTF-8", + archive_entry_symlink(file->entry)); + r = ARCHIVE_WARN; + } + archive_strncpy(&(file->symlink), pp, len2); + cleanup_backslash(file->symlink.s, file->symlink.length); + } + /* + * - Count up directory elements. + * - Find out the position which points the last position of + * path separator('/'). + */ + slash = NULL; + for (; *p != '\0'; p++) + if (*p == '/') + slash = p; + if (slash == NULL) { + /* The pathname doesn't have a parent directory. */ + file->parentdir.length = len; + archive_string_copy(&(file->basename), &(file->parentdir)); + archive_string_empty(&(file->parentdir)); + *file->parentdir.s = '\0'; + return (r); + } + + /* Make a basename from dirname and slash */ + *slash = '\0'; + file->parentdir.length = slash - dirname; + archive_strcpy(&(file->basename), slash + 1); + return (r); +} + +static int +get_path_component(char *name, int n, const char *fn) +{ + char *p; + int l; + + p = strchr(fn, '/'); + if (p == NULL) { + if ((l = strlen(fn)) == 0) + return (0); + } else + l = p - fn; + if (l > n -1) + return (-1); + memcpy(name, fn, l); + name[l] = '\0'; + + return (l); +} + +/* + * Add a new entry into the tree. + */ +static int +file_tree(struct archive_write *a, struct file **filepp) +{ +#if defined(_WIN32) && !defined(__CYGWIN__) + char name[_MAX_FNAME];/* Included null terminator size. */ +#elif defined(NAME_MAX) && NAME_MAX >= 255 + char name[NAME_MAX+1]; +#else + char name[256]; +#endif + struct xar *xar = (struct xar *)a->format_data; + struct file *dent, *file, *np; + struct archive_entry *ent; + const char *fn, *p; + int l; + + file = *filepp; + dent = xar->root; + if (file->parentdir.length > 0) + fn = p = file->parentdir.s; + else + fn = p = ""; + + /* + * If the path of the parent directory of `file' entry is + * the same as the path of `cur_dirent', add isoent to + * `cur_dirent'. + */ + if (archive_strlen(&(xar->cur_dirstr)) + == archive_strlen(&(file->parentdir)) && + strcmp(xar->cur_dirstr.s, fn) == 0) { + if (!file_add_child_tail(xar->cur_dirent, file)) { + np = (struct file *)__archive_rb_tree_find_node( + &(xar->cur_dirent->rbtree), + file->basename.s); + goto same_entry; + } + return (ARCHIVE_OK); + } + + for (;;) { + l = get_path_component(name, sizeof(name), fn); + if (l == 0) { + np = NULL; + break; + } + if (l < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "A name buffer is too small"); + file_free(file); + *filepp = NULL; + return (ARCHIVE_FATAL); + } + + np = file_find_child(dent, name); + if (np == NULL || fn[0] == '\0') + break; + + /* Find next subdirectory. */ + if (!np->dir) { + /* NOT Directory! */ + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "`%s' is not directory, we cannot insert `%s' ", + archive_entry_pathname(np->entry), + archive_entry_pathname(file->entry)); + file_free(file); + *filepp = NULL; + return (ARCHIVE_FAILED); + } + fn += l; + if (fn[0] == '/') + fn++; + dent = np; + } + if (np == NULL) { + /* + * Create virtual parent directories. + */ + while (fn[0] != '\0') { + struct file *vp; + struct archive_string as; + + archive_string_init(&as); + archive_strncat(&as, p, fn - p + l); + if (as.s[as.length-1] == '/') { + as.s[as.length-1] = '\0'; + as.length--; + } + vp = file_create_virtual_dir(a, xar, as.s); + if (vp == NULL) { + archive_string_free(&as); + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + file_free(file); + *filepp = NULL; + return (ARCHIVE_FATAL); + } + archive_string_free(&as); + if (file_gen_utility_names(a, vp) <= ARCHIVE_FAILED) + return (ARCHIVE_FATAL); + file_add_child_tail(dent, vp); + file_register(xar, vp); + np = vp; + + fn += l; + if (fn[0] == '/') + fn++; + l = get_path_component(name, sizeof(name), fn); + if (l < 0) { + archive_string_free(&as); + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "A name buffer is too small"); + file_free(file); + *filepp = NULL; + return (ARCHIVE_FATAL); + } + dent = np; + } + + /* Found out the parent directory where isoent can be + * inserted. */ + xar->cur_dirent = dent; + archive_string_empty(&(xar->cur_dirstr)); + archive_string_ensure(&(xar->cur_dirstr), + archive_strlen(&(dent->parentdir)) + + archive_strlen(&(dent->basename)) + 2); + if (archive_strlen(&(dent->parentdir)) + + archive_strlen(&(dent->basename)) == 0) + xar->cur_dirstr.s[0] = 0; + else { + if (archive_strlen(&(dent->parentdir)) > 0) { + archive_string_copy(&(xar->cur_dirstr), + &(dent->parentdir)); + archive_strappend_char(&(xar->cur_dirstr), '/'); + } + archive_string_concat(&(xar->cur_dirstr), + &(dent->basename)); + } + + if (!file_add_child_tail(dent, file)) { + np = (struct file *)__archive_rb_tree_find_node( + &(dent->rbtree), file->basename.s); + goto same_entry; + } + return (ARCHIVE_OK); + } + +same_entry: + /* + * We have already has the entry the filename of which is + * the same. + */ + if (archive_entry_filetype(np->entry) != + archive_entry_filetype(file->entry)) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Found duplicate entries `%s' and its file type is " + "different", + archive_entry_pathname(np->entry)); + file_free(file); + *filepp = NULL; + return (ARCHIVE_FAILED); + } + + /* Swap files. */ + ent = np->entry; + np->entry = file->entry; + file->entry = ent; + np->virtual = 0; + + file_free(file); + *filepp = np; + return (ARCHIVE_OK); +} + +static void +file_register(struct xar *xar, struct file *file) +{ + file->id = xar->file_idx++; + file->next = NULL; + *xar->file_list.last = file; + xar->file_list.last = &(file->next); +} + +static void +file_init_register(struct xar *xar) +{ + xar->file_list.first = NULL; + xar->file_list.last = &(xar->file_list.first); +} + +static void +file_free_register(struct xar *xar) +{ + struct file *file, *file_next; + + file = xar->file_list.first; + while (file != NULL) { + file_next = file->next; + file_free(file); + file = file_next; + } +} + +/* + * Register entry to get a hardlink target. + */ +static int +file_register_hardlink(struct archive_write *a, struct file *file) +{ + struct xar *xar = (struct xar *)a->format_data; + struct hardlink *hl; + const char *pathname; + + archive_entry_set_nlink(file->entry, 1); + pathname = archive_entry_hardlink(file->entry); + if (pathname == NULL) { + /* This `file` is a hardlink target. */ + hl = malloc(sizeof(*hl)); + if (hl == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory"); + return (ARCHIVE_FATAL); + } + hl->nlink = 1; + /* A hardlink target must be the first position. */ + file->hlnext = NULL; + hl->file_list.first = file; + hl->file_list.last = &(file->hlnext); + __archive_rb_tree_insert_node(&(xar->hardlink_rbtree), + (struct archive_rb_node *)hl); + } else { + hl = (struct hardlink *)__archive_rb_tree_find_node( + &(xar->hardlink_rbtree), pathname); + if (hl != NULL) { + /* Insert `file` entry into the tail. */ + file->hlnext = NULL; + *hl->file_list.last = file; + hl->file_list.last = &(file->hlnext); + hl->nlink++; + } + archive_entry_unset_size(file->entry); + } + + return (ARCHIVE_OK); +} + +/* + * Hardlinked files have to have the same location of extent. + * We have to find out hardlink target entries for entries which + * have a hardlink target name. + */ +static void +file_connect_hardlink_files(struct xar *xar) +{ + struct archive_rb_node *n; + struct hardlink *hl; + struct file *target, *nf; + + ARCHIVE_RB_TREE_FOREACH(n, &(xar->hardlink_rbtree)) { + hl = (struct hardlink *)n; + + /* The first entry must be a hardlink target. */ + target = hl->file_list.first; + archive_entry_set_nlink(target->entry, hl->nlink); + if (hl->nlink > 1) + /* It means this file is a hardlink + * target itself. */ + target->hardlink_target = target; + for (nf = target->hlnext; + nf != NULL; nf = nf->hlnext) { + nf->hardlink_target = target; + archive_entry_set_nlink(nf->entry, hl->nlink); + } + } +} + +static int +file_hd_cmp_node(const struct archive_rb_node *n1, + const struct archive_rb_node *n2) +{ + const struct hardlink *h1 = (const struct hardlink *)n1; + const struct hardlink *h2 = (const struct hardlink *)n2; + + return (strcmp(archive_entry_pathname(h1->file_list.first->entry), + archive_entry_pathname(h2->file_list.first->entry))); +} + +static int +file_hd_cmp_key(const struct archive_rb_node *n, const void *key) +{ + const struct hardlink *h = (const struct hardlink *)n; + + return (strcmp(archive_entry_pathname(h->file_list.first->entry), + (const char *)key)); +} + + +static void +file_init_hardlinks(struct xar *xar) +{ + static const struct archive_rb_tree_ops rb_ops = { + file_hd_cmp_node, file_hd_cmp_key, + }; + + __archive_rb_tree_init(&(xar->hardlink_rbtree), &rb_ops); +} + +static void +file_free_hardlinks(struct xar *xar) +{ + struct archive_rb_node *n, *tmp; + + ARCHIVE_RB_TREE_FOREACH_SAFE(n, &(xar->hardlink_rbtree), tmp) { + __archive_rb_tree_remove_node(&(xar->hardlink_rbtree), n); + free(n); + } +} + +static void +checksum_init(struct chksumwork *sumwrk, enum sumalg sum_alg) +{ + sumwrk->alg = sum_alg; + switch (sum_alg) { + case CKSUM_NONE: + break; + case CKSUM_SHA1: + archive_sha1_init(&(sumwrk->sha1ctx)); + break; + case CKSUM_MD5: + archive_md5_init(&(sumwrk->md5ctx)); + break; + } +} + +static void +checksum_update(struct chksumwork *sumwrk, const void *buff, size_t size) +{ + + switch (sumwrk->alg) { + case CKSUM_NONE: + break; + case CKSUM_SHA1: + archive_sha1_update(&(sumwrk->sha1ctx), buff, size); + break; + case CKSUM_MD5: + archive_md5_update(&(sumwrk->md5ctx), buff, size); + break; + } +} + +static void +checksum_final(struct chksumwork *sumwrk, struct chksumval *sumval) +{ + + switch (sumwrk->alg) { + case CKSUM_NONE: + sumval->len = 0; + break; + case CKSUM_SHA1: + archive_sha1_final(&(sumwrk->sha1ctx), sumval->val); + sumval->len = SHA1_SIZE; + break; + case CKSUM_MD5: + archive_md5_final(&(sumwrk->md5ctx), sumval->val); + sumval->len = MD5_SIZE; + break; + } + sumval->alg = sumwrk->alg; +} + +#if !defined(HAVE_BZLIB_H) || !defined(BZ_CONFIG_ERROR) || !defined(HAVE_LZMA_H) +static int +compression_unsupported_encoder(struct archive *a, + struct la_zstream *lastrm, const char *name) +{ + + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "%s compression not supported on this platform", name); + lastrm->valid = 0; + lastrm->real_stream = NULL; + return (ARCHIVE_FAILED); +} +#endif + +static int +compression_init_encoder_gzip(struct archive *a, + struct la_zstream *lastrm, int level, int withheader) +{ + z_stream *strm; + + if (lastrm->valid) + compression_end(a, lastrm); + strm = calloc(1, sizeof(*strm)); + if (strm == NULL) { + archive_set_error(a, ENOMEM, + "Can't allocate memory for gzip stream"); + return (ARCHIVE_FATAL); + } + /* zlib.h is not const-correct, so we need this one bit + * of ugly hackery to convert a const * pointer to + * a non-const pointer. */ + strm->next_in = (Bytef *)(uintptr_t)(const void *)lastrm->next_in; + strm->avail_in = lastrm->avail_in; + strm->total_in = (uLong)lastrm->total_in; + strm->next_out = lastrm->next_out; + strm->avail_out = lastrm->avail_out; + strm->total_out = (uLong)lastrm->total_out; + if (deflateInit2(strm, level, Z_DEFLATED, + (withheader)?15:-15, + 8, Z_DEFAULT_STRATEGY) != Z_OK) { + free(strm); + lastrm->real_stream = NULL; + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Internal error initializing compression library"); + return (ARCHIVE_FATAL); + } + lastrm->real_stream = strm; + lastrm->valid = 1; + lastrm->code = compression_code_gzip; + lastrm->end = compression_end_gzip; + return (ARCHIVE_OK); +} + +static int +compression_code_gzip(struct archive *a, + struct la_zstream *lastrm, enum la_zaction action) +{ + z_stream *strm; + int r; + + strm = (z_stream *)lastrm->real_stream; + /* zlib.h is not const-correct, so we need this one bit + * of ugly hackery to convert a const * pointer to + * a non-const pointer. */ + strm->next_in = (Bytef *)(uintptr_t)(const void *)lastrm->next_in; + strm->avail_in = lastrm->avail_in; + strm->total_in = (uLong)lastrm->total_in; + strm->next_out = lastrm->next_out; + strm->avail_out = lastrm->avail_out; + strm->total_out = (uLong)lastrm->total_out; + r = deflate(strm, + (action == ARCHIVE_Z_FINISH)? Z_FINISH: Z_NO_FLUSH); + lastrm->next_in = strm->next_in; + lastrm->avail_in = strm->avail_in; + lastrm->total_in = strm->total_in; + lastrm->next_out = strm->next_out; + lastrm->avail_out = strm->avail_out; + lastrm->total_out = strm->total_out; + switch (r) { + case Z_OK: + return (ARCHIVE_OK); + case Z_STREAM_END: + return (ARCHIVE_EOF); + default: + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "GZip compression failed:" + " deflate() call returned status %d", r); + return (ARCHIVE_FATAL); + } +} + +static int +compression_end_gzip(struct archive *a, struct la_zstream *lastrm) +{ + z_stream *strm; + int r; + + strm = (z_stream *)lastrm->real_stream; + r = deflateEnd(strm); + free(strm); + lastrm->real_stream = NULL; + lastrm->valid = 0; + if (r != Z_OK) { + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Failed to clean up compressor"); + return (ARCHIVE_FATAL); + } + return (ARCHIVE_OK); +} + +#if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) +static int +compression_init_encoder_bzip2(struct archive *a, + struct la_zstream *lastrm, int level) +{ + bz_stream *strm; + + if (lastrm->valid) + compression_end(a, lastrm); + strm = calloc(1, sizeof(*strm)); + if (strm == NULL) { + archive_set_error(a, ENOMEM, + "Can't allocate memory for bzip2 stream"); + return (ARCHIVE_FATAL); + } + /* bzlib.h is not const-correct, so we need this one bit + * of ugly hackery to convert a const * pointer to + * a non-const pointer. */ + strm->next_in = (char *)(uintptr_t)(const void *)lastrm->next_in; + strm->avail_in = lastrm->avail_in; + strm->total_in_lo32 = (uint32_t)(lastrm->total_in & 0xffffffff); + strm->total_in_hi32 = (uint32_t)(lastrm->total_in >> 32); + strm->next_out = (char *)lastrm->next_out; + strm->avail_out = lastrm->avail_out; + strm->total_out_lo32 = (uint32_t)(lastrm->total_out & 0xffffffff); + strm->total_out_hi32 = (uint32_t)(lastrm->total_out >> 32); + if (BZ2_bzCompressInit(strm, level, 0, 30) != BZ_OK) { + free(strm); + lastrm->real_stream = NULL; + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Internal error initializing compression library"); + return (ARCHIVE_FATAL); + } + lastrm->real_stream = strm; + lastrm->valid = 1; + lastrm->code = compression_code_bzip2; + lastrm->end = compression_end_bzip2; + return (ARCHIVE_OK); +} + +static int +compression_code_bzip2(struct archive *a, + struct la_zstream *lastrm, enum la_zaction action) +{ + bz_stream *strm; + int r; + + strm = (bz_stream *)lastrm->real_stream; + /* bzlib.h is not const-correct, so we need this one bit + * of ugly hackery to convert a const * pointer to + * a non-const pointer. */ + strm->next_in = (char *)(uintptr_t)(const void *)lastrm->next_in; + strm->avail_in = lastrm->avail_in; + strm->total_in_lo32 = (uint32_t)(lastrm->total_in & 0xffffffff); + strm->total_in_hi32 = (uint32_t)(lastrm->total_in >> 32); + strm->next_out = (char *)lastrm->next_out; + strm->avail_out = lastrm->avail_out; + strm->total_out_lo32 = (uint32_t)(lastrm->total_out & 0xffffffff); + strm->total_out_hi32 = (uint32_t)(lastrm->total_out >> 32); + r = BZ2_bzCompress(strm, + (action == ARCHIVE_Z_FINISH)? BZ_FINISH: BZ_RUN); + lastrm->next_in = (const unsigned char *)strm->next_in; + lastrm->avail_in = strm->avail_in; + lastrm->total_in = + (((uint64_t)(uint32_t)strm->total_in_hi32) << 32) + + (uint64_t)(uint32_t)strm->total_in_lo32; + lastrm->next_out = (unsigned char *)strm->next_out; + lastrm->avail_out = strm->avail_out; + lastrm->total_out = + (((uint64_t)(uint32_t)strm->total_out_hi32) << 32) + + (uint64_t)(uint32_t)strm->total_out_lo32; + switch (r) { + case BZ_RUN_OK: /* Non-finishing */ + case BZ_FINISH_OK: /* Finishing: There's more work to do */ + return (ARCHIVE_OK); + case BZ_STREAM_END: /* Finishing: all done */ + /* Only occurs in finishing case */ + return (ARCHIVE_EOF); + default: + /* Any other return value indicates an error */ + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Bzip2 compression failed:" + " BZ2_bzCompress() call returned status %d", r); + return (ARCHIVE_FATAL); + } +} + +static int +compression_end_bzip2(struct archive *a, struct la_zstream *lastrm) +{ + bz_stream *strm; + int r; + + strm = (bz_stream *)lastrm->real_stream; + r = BZ2_bzCompressEnd(strm); + free(strm); + lastrm->real_stream = NULL; + lastrm->valid = 0; + if (r != BZ_OK) { + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Failed to clean up compressor"); + return (ARCHIVE_FATAL); + } + return (ARCHIVE_OK); +} + +#else +static int +compression_init_encoder_bzip2(struct archive *a, + struct la_zstream *lastrm, int level) +{ + + (void) level; /* UNUSED */ + if (lastrm->valid) + compression_end(a, lastrm); + return (compression_unsupported_encoder(a, lastrm, "bzip2")); +} +#endif + +#if defined(HAVE_LZMA_H) +static int +compression_init_encoder_lzma(struct archive *a, + struct la_zstream *lastrm, int level) +{ + static const lzma_stream lzma_init_data = LZMA_STREAM_INIT; + lzma_stream *strm; + lzma_options_lzma lzma_opt; + int r; + + if (lastrm->valid) + compression_end(a, lastrm); + if (lzma_lzma_preset(&lzma_opt, level)) { + lastrm->real_stream = NULL; + archive_set_error(a, ENOMEM, + "Internal error initializing compression library"); + return (ARCHIVE_FATAL); + } + strm = calloc(1, sizeof(*strm)); + if (strm == NULL) { + archive_set_error(a, ENOMEM, + "Can't allocate memory for lzma stream"); + return (ARCHIVE_FATAL); + } + *strm = lzma_init_data; + r = lzma_alone_encoder(strm, &lzma_opt); + switch (r) { + case LZMA_OK: + lastrm->real_stream = strm; + lastrm->valid = 1; + lastrm->code = compression_code_lzma; + lastrm->end = compression_end_lzma; + r = ARCHIVE_OK; + break; + case LZMA_MEM_ERROR: + free(strm); + lastrm->real_stream = NULL; + archive_set_error(a, ENOMEM, + "Internal error initializing compression library: " + "Cannot allocate memory"); + r = ARCHIVE_FATAL; + break; + default: + free(strm); + lastrm->real_stream = NULL; + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Internal error initializing compression library: " + "It's a bug in liblzma"); + r = ARCHIVE_FATAL; + break; + } + return (r); +} + +static int +compression_init_encoder_xz(struct archive *a, + struct la_zstream *lastrm, int level, int threads) +{ + static const lzma_stream lzma_init_data = LZMA_STREAM_INIT; + lzma_stream *strm; + lzma_filter *lzmafilters; + lzma_options_lzma lzma_opt; + int r; +#ifdef HAVE_LZMA_STREAM_ENCODER_MT + lzma_mt mt_options; +#endif + + (void)threads; /* UNUSED (if multi-threaded LZMA library not avail) */ + + if (lastrm->valid) + compression_end(a, lastrm); + strm = calloc(1, sizeof(*strm) + sizeof(*lzmafilters) * 2); + if (strm == NULL) { + archive_set_error(a, ENOMEM, + "Can't allocate memory for xz stream"); + return (ARCHIVE_FATAL); + } + lzmafilters = (lzma_filter *)(strm+1); + if (level > 6) + level = 6; + if (lzma_lzma_preset(&lzma_opt, level)) { + free(strm); + lastrm->real_stream = NULL; + archive_set_error(a, ENOMEM, + "Internal error initializing compression library"); + return (ARCHIVE_FATAL); + } + lzmafilters[0].id = LZMA_FILTER_LZMA2; + lzmafilters[0].options = &lzma_opt; + lzmafilters[1].id = LZMA_VLI_UNKNOWN;/* Terminate */ + + *strm = lzma_init_data; +#ifdef HAVE_LZMA_STREAM_ENCODER_MT + if (threads > 1) { + memset(&mt_options, 0, sizeof(mt_options)); + mt_options.threads = threads; + mt_options.timeout = 300; + mt_options.filters = lzmafilters; + mt_options.check = LZMA_CHECK_CRC64; + r = lzma_stream_encoder_mt(strm, &mt_options); + } else +#endif + r = lzma_stream_encoder(strm, lzmafilters, LZMA_CHECK_CRC64); + switch (r) { + case LZMA_OK: + lastrm->real_stream = strm; + lastrm->valid = 1; + lastrm->code = compression_code_lzma; + lastrm->end = compression_end_lzma; + r = ARCHIVE_OK; + break; + case LZMA_MEM_ERROR: + free(strm); + lastrm->real_stream = NULL; + archive_set_error(a, ENOMEM, + "Internal error initializing compression library: " + "Cannot allocate memory"); + r = ARCHIVE_FATAL; + break; + default: + free(strm); + lastrm->real_stream = NULL; + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "Internal error initializing compression library: " + "It's a bug in liblzma"); + r = ARCHIVE_FATAL; + break; + } + return (r); +} + +static int +compression_code_lzma(struct archive *a, + struct la_zstream *lastrm, enum la_zaction action) +{ + lzma_stream *strm; + int r; + + strm = (lzma_stream *)lastrm->real_stream; + strm->next_in = lastrm->next_in; + strm->avail_in = lastrm->avail_in; + strm->total_in = lastrm->total_in; + strm->next_out = lastrm->next_out; + strm->avail_out = lastrm->avail_out; + strm->total_out = lastrm->total_out; + r = lzma_code(strm, + (action == ARCHIVE_Z_FINISH)? LZMA_FINISH: LZMA_RUN); + lastrm->next_in = strm->next_in; + lastrm->avail_in = strm->avail_in; + lastrm->total_in = strm->total_in; + lastrm->next_out = strm->next_out; + lastrm->avail_out = strm->avail_out; + lastrm->total_out = strm->total_out; + switch (r) { + case LZMA_OK: + /* Non-finishing case */ + return (ARCHIVE_OK); + case LZMA_STREAM_END: + /* This return can only occur in finishing case. */ + return (ARCHIVE_EOF); + case LZMA_MEMLIMIT_ERROR: + archive_set_error(a, ENOMEM, + "lzma compression error:" + " %ju MiB would have been needed", + (uintmax_t)((lzma_memusage(strm) + 1024 * 1024 -1) + / (1024 * 1024))); + return (ARCHIVE_FATAL); + default: + /* Any other return value indicates an error */ + archive_set_error(a, ARCHIVE_ERRNO_MISC, + "lzma compression failed:" + " lzma_code() call returned status %d", r); + return (ARCHIVE_FATAL); + } +} + +static int +compression_end_lzma(struct archive *a, struct la_zstream *lastrm) +{ + lzma_stream *strm; + + (void)a; /* UNUSED */ + strm = (lzma_stream *)lastrm->real_stream; + lzma_end(strm); + free(strm); + lastrm->valid = 0; + lastrm->real_stream = NULL; + return (ARCHIVE_OK); +} +#else +static int +compression_init_encoder_lzma(struct archive *a, + struct la_zstream *lastrm, int level) +{ + + (void) level; /* UNUSED */ + if (lastrm->valid) + compression_end(a, lastrm); + return (compression_unsupported_encoder(a, lastrm, "lzma")); +} +static int +compression_init_encoder_xz(struct archive *a, + struct la_zstream *lastrm, int level, int threads) +{ + + (void) level; /* UNUSED */ + (void) threads; /* UNUSED */ + if (lastrm->valid) + compression_end(a, lastrm); + return (compression_unsupported_encoder(a, lastrm, "xz")); +} +#endif + +static int +xar_compression_init_encoder(struct archive_write *a) +{ + struct xar *xar; + int r; + + xar = (struct xar *)a->format_data; + switch (xar->opt_compression) { + case GZIP: + r = compression_init_encoder_gzip( + &(a->archive), &(xar->stream), + xar->opt_compression_level, 1); + break; + case BZIP2: + r = compression_init_encoder_bzip2( + &(a->archive), &(xar->stream), + xar->opt_compression_level); + break; + case LZMA: + r = compression_init_encoder_lzma( + &(a->archive), &(xar->stream), + xar->opt_compression_level); + break; + case XZ: + r = compression_init_encoder_xz( + &(a->archive), &(xar->stream), + xar->opt_compression_level, xar->opt_threads); + break; + default: + r = ARCHIVE_OK; + break; + } + if (r == ARCHIVE_OK) { + xar->stream.total_in = 0; + xar->stream.next_out = xar->wbuff; + xar->stream.avail_out = sizeof(xar->wbuff); + xar->stream.total_out = 0; + } + + return (r); +} + +static int +compression_code(struct archive *a, struct la_zstream *lastrm, + enum la_zaction action) +{ + if (lastrm->valid) + return (lastrm->code(a, lastrm, action)); + return (ARCHIVE_OK); +} + +static int +compression_end(struct archive *a, struct la_zstream *lastrm) +{ + if (lastrm->valid) + return (lastrm->end(a, lastrm)); + return (ARCHIVE_OK); +} + + +static int +save_xattrs(struct archive_write *a, struct file *file) +{ + struct xar *xar; + const char *name; + const void *value; + struct heap_data *heap; + size_t size; + int count, r; + + xar = (struct xar *)a->format_data; + count = archive_entry_xattr_reset(file->entry); + if (count == 0) + return (ARCHIVE_OK); + while (count--) { + archive_entry_xattr_next(file->entry, + &name, &value, &size); + checksum_init(&(xar->a_sumwrk), xar->opt_sumalg); + checksum_init(&(xar->e_sumwrk), xar->opt_sumalg); + + heap = calloc(1, sizeof(*heap)); + if (heap == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for xattr"); + return (ARCHIVE_FATAL); + } + heap->id = file->ea_idx++; + heap->temp_offset = xar->temp_offset; + heap->size = size;/* save a extracted size */ + heap->compression = xar->opt_compression; + /* Get a extracted sumcheck value. */ + checksum_update(&(xar->e_sumwrk), value, size); + checksum_final(&(xar->e_sumwrk), &(heap->e_sum)); + + /* + * Not compression to xattr is simple way. + */ + if (heap->compression == NONE) { + checksum_update(&(xar->a_sumwrk), value, size); + checksum_final(&(xar->a_sumwrk), &(heap->a_sum)); + if (write_to_temp(a, value, size) + != ARCHIVE_OK) { + free(heap); + return (ARCHIVE_FATAL); + } + heap->length = size; + /* Add heap to the tail of file->xattr. */ + heap->next = NULL; + *file->xattr.last = heap; + file->xattr.last = &(heap->next); + /* Next xattr */ + continue; + } + + /* + * Init compression library. + */ + r = xar_compression_init_encoder(a); + if (r != ARCHIVE_OK) { + free(heap); + return (ARCHIVE_FATAL); + } + + xar->stream.next_in = (const unsigned char *)value; + xar->stream.avail_in = size; + for (;;) { + r = compression_code(&(a->archive), + &(xar->stream), ARCHIVE_Z_FINISH); + if (r != ARCHIVE_OK && r != ARCHIVE_EOF) { + free(heap); + return (ARCHIVE_FATAL); + } + size = sizeof(xar->wbuff) - xar->stream.avail_out; + checksum_update(&(xar->a_sumwrk), + xar->wbuff, size); + if (write_to_temp(a, xar->wbuff, size) + != ARCHIVE_OK) { + free(heap); + return (ARCHIVE_FATAL); + } + if (r == ARCHIVE_OK) { + xar->stream.next_out = xar->wbuff; + xar->stream.avail_out = sizeof(xar->wbuff); + } else { + checksum_final(&(xar->a_sumwrk), + &(heap->a_sum)); + heap->length = xar->stream.total_out; + /* Add heap to the tail of file->xattr. */ + heap->next = NULL; + *file->xattr.last = heap; + file->xattr.last = &(heap->next); + break; + } + } + /* Clean up compression library. */ + r = compression_end(&(a->archive), &(xar->stream)); + if (r != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + return (ARCHIVE_OK); +} + +static int +getalgsize(enum sumalg sumalg) +{ + switch (sumalg) { + default: + case CKSUM_NONE: + return (0); + case CKSUM_SHA1: + return (SHA1_SIZE); + case CKSUM_MD5: + return (MD5_SIZE); + } +} + +static const char * +getalgname(enum sumalg sumalg) +{ + switch (sumalg) { + default: + case CKSUM_NONE: + return (NULL); + case CKSUM_SHA1: + return (SHA1_NAME); + case CKSUM_MD5: + return (MD5_NAME); + } +} + +#endif /* Support xar format */ diff --git a/c/archive_write_set_format_zip.c b/c/archive_write_set_format_zip.c new file mode 100644 --- /dev/null +++ b/c/archive_write_set_format_zip.c @@ -0,0 +1,1695 @@ +/*- + * Copyright (c) 2008 Anselm Strauss + * Copyright (c) 2009 Joerg Sonnenberger + * Copyright (c) 2011-2012,2014 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +/* + * Development supported by Google Summer of Code 2008. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_format_zip.c 201168 2009-12-29 06:15:32Z kientzle $"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#ifdef HAVE_LANGINFO_H +#include +#endif +#ifdef HAVE_STDLIB_H +#include +#endif +#ifdef HAVE_STRING_H +#include +#endif +#ifdef HAVE_ZLIB_H +#include +#endif + +#include "archive.h" +#include "archive_cryptor_private.h" +#include "archive_endian.h" +#include "archive_entry.h" +#include "archive_entry_locale.h" +#include "archive_hmac_private.h" +#include "archive_private.h" +#include "archive_random_private.h" +#include "archive_write_private.h" +#include "archive_write_set_format_private.h" + +#ifndef HAVE_ZLIB_H +#include "archive_crc32.h" +#endif + +#define ZIP_ENTRY_FLAG_ENCRYPTED (1<<0) +#define ZIP_ENTRY_FLAG_LENGTH_AT_END (1<<3) +#define ZIP_ENTRY_FLAG_UTF8_NAME (1 << 11) + +#define ZIP_4GB_MAX ARCHIVE_LITERAL_LL(0xffffffff) +#define ZIP_4GB_MAX_UNCOMPRESSED ARCHIVE_LITERAL_LL(0xff000000) + +enum compression { + COMPRESSION_UNSPECIFIED = -1, + COMPRESSION_STORE = 0, + COMPRESSION_DEFLATE = 8 +}; + +#ifdef HAVE_ZLIB_H +#define COMPRESSION_DEFAULT COMPRESSION_DEFLATE +#else +#define COMPRESSION_DEFAULT COMPRESSION_STORE +#endif + +enum encryption { + ENCRYPTION_NONE = 0, + ENCRYPTION_TRADITIONAL, /* Traditional PKWARE encryption. */ + ENCRYPTION_WINZIP_AES128, /* WinZIP AES-128 encryption. */ + ENCRYPTION_WINZIP_AES256, /* WinZIP AES-256 encryption. */ +}; + +#define TRAD_HEADER_SIZE 12 +/* + * See "WinZip - AES Encryption Information" + * http://www.winzip.com/aes_info.htm + */ +/* Value used in compression method. */ +#define WINZIP_AES_ENCRYPTION 99 +/* A WinZip AES header size which is stored at the beginning of + * file contents. */ +#define WINZIP_AES128_HEADER_SIZE (8 + 2) +#define WINZIP_AES256_HEADER_SIZE (16 + 2) +/* AES vendor version. */ +#define AES_VENDOR_AE_1 0x0001 +#define AES_VENDOR_AE_2 0x0002 +/* Authentication code size. */ +#define AUTH_CODE_SIZE 10 +/**/ +#define MAX_DERIVED_KEY_BUF_SIZE (AES_MAX_KEY_SIZE * 2 + 2) + +struct cd_segment { + struct cd_segment *next; + size_t buff_size; + unsigned char *buff; + unsigned char *p; +}; + +struct trad_enc_ctx { + uint32_t keys[3]; +}; + +struct zip { + + int64_t entry_offset; + int64_t entry_compressed_size; + int64_t entry_uncompressed_size; + int64_t entry_compressed_written; + int64_t entry_uncompressed_written; + int64_t entry_uncompressed_limit; + struct archive_entry *entry; + uint32_t entry_crc32; + enum compression entry_compression; + enum encryption entry_encryption; + int entry_flags; + int entry_uses_zip64; + int experiments; + struct trad_enc_ctx tctx; + char tctx_valid; + unsigned char trad_chkdat; + unsigned aes_vendor; + archive_crypto_ctx cctx; + char cctx_valid; + archive_hmac_sha1_ctx hctx; + char hctx_valid; + + unsigned char *file_header; + size_t file_header_extra_offset; + unsigned long (*crc32func)(unsigned long crc, const void *buff, size_t len); + + struct cd_segment *central_directory; + struct cd_segment *central_directory_last; + size_t central_directory_bytes; + size_t central_directory_entries; + + int64_t written_bytes; /* Overall position in file. */ + + struct archive_string_conv *opt_sconv; + struct archive_string_conv *sconv_default; + enum compression requested_compression; + int deflate_compression_level; + int init_default_conversion; + enum encryption encryption_type; + +#define ZIP_FLAG_AVOID_ZIP64 1 +#define ZIP_FLAG_FORCE_ZIP64 2 +#define ZIP_FLAG_EXPERIMENT_xl 4 + int flags; + +#ifdef HAVE_ZLIB_H + z_stream stream; +#endif + size_t len_buf; + unsigned char *buf; +}; + +/* Don't call this min or MIN, since those are already defined + on lots of platforms (but not all). */ +#define zipmin(a, b) ((a) > (b) ? (b) : (a)) + +static ssize_t archive_write_zip_data(struct archive_write *, + const void *buff, size_t s); +static int archive_write_zip_close(struct archive_write *); +static int archive_write_zip_free(struct archive_write *); +static int archive_write_zip_finish_entry(struct archive_write *); +static int archive_write_zip_header(struct archive_write *, + struct archive_entry *); +static int archive_write_zip_options(struct archive_write *, + const char *, const char *); +static unsigned int dos_time(const time_t); +static size_t path_length(struct archive_entry *); +static int write_path(struct archive_entry *, struct archive_write *); +static void copy_path(struct archive_entry *, unsigned char *); +static struct archive_string_conv *get_sconv(struct archive_write *, struct zip *); +static int trad_enc_init(struct trad_enc_ctx *, const char *, size_t); +static unsigned trad_enc_encrypt_update(struct trad_enc_ctx *, const uint8_t *, + size_t, uint8_t *, size_t); +static int init_traditional_pkware_encryption(struct archive_write *); +static int is_traditional_pkware_encryption_supported(void); +static int init_winzip_aes_encryption(struct archive_write *); +static int is_winzip_aes_encryption_supported(int encryption); + +static unsigned char * +cd_alloc(struct zip *zip, size_t length) +{ + unsigned char *p; + + if (zip->central_directory == NULL + || (zip->central_directory_last->p + length + > zip->central_directory_last->buff + zip->central_directory_last->buff_size)) { + struct cd_segment *segment = calloc(1, sizeof(*segment)); + if (segment == NULL) + return NULL; + segment->buff_size = 64 * 1024; + segment->buff = malloc(segment->buff_size); + if (segment->buff == NULL) { + free(segment); + return NULL; + } + segment->p = segment->buff; + + if (zip->central_directory == NULL) { + zip->central_directory + = zip->central_directory_last + = segment; + } else { + zip->central_directory_last->next = segment; + zip->central_directory_last = segment; + } + } + + p = zip->central_directory_last->p; + zip->central_directory_last->p += length; + zip->central_directory_bytes += length; + return (p); +} + +static unsigned long +real_crc32(unsigned long crc, const void *buff, size_t len) +{ + return crc32(crc, buff, (unsigned int)len); +} + +static unsigned long +fake_crc32(unsigned long crc, const void *buff, size_t len) +{ + (void)crc; /* UNUSED */ + (void)buff; /* UNUSED */ + (void)len; /* UNUSED */ + return 0; +} + +static int +archive_write_zip_options(struct archive_write *a, const char *key, + const char *val) +{ + struct zip *zip = a->format_data; + int ret = ARCHIVE_FAILED; + + if (strcmp(key, "compression") == 0) { + /* + * Set compression to use on all future entries. + * This only affects regular files. + */ + if (val == NULL || val[0] == 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "%s: compression option needs a compression name", + a->format_name); + } else if (strcmp(val, "deflate") == 0) { +#ifdef HAVE_ZLIB_H + zip->requested_compression = COMPRESSION_DEFLATE; + ret = ARCHIVE_OK; +#else + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "deflate compression not supported"); +#endif + } else if (strcmp(val, "store") == 0) { + zip->requested_compression = COMPRESSION_STORE; + ret = ARCHIVE_OK; + } + return (ret); + } else if (strcmp(key, "compression-level") == 0) { + if (val == NULL || !(val[0] >= '0' && val[0] <= '9') || val[1] != '\0') { + return ARCHIVE_WARN; + } + + if (val[0] == '0') { + zip->requested_compression = COMPRESSION_STORE; + return ARCHIVE_OK; + } else { +#ifdef HAVE_ZLIB_H + zip->requested_compression = COMPRESSION_DEFLATE; + zip->deflate_compression_level = val[0] - '0'; + return ARCHIVE_OK; +#else + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "deflate compression not supported"); +#endif + } + } else if (strcmp(key, "encryption") == 0) { + if (val == NULL) { + zip->encryption_type = ENCRYPTION_NONE; + ret = ARCHIVE_OK; + } else if (val[0] == '1' || strcmp(val, "traditional") == 0 + || strcmp(val, "zipcrypt") == 0 + || strcmp(val, "ZipCrypt") == 0) { + if (is_traditional_pkware_encryption_supported()) { + zip->encryption_type = ENCRYPTION_TRADITIONAL; + ret = ARCHIVE_OK; + } else { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "encryption not supported"); + } + } else if (strcmp(val, "aes128") == 0) { + if (is_winzip_aes_encryption_supported( + ENCRYPTION_WINZIP_AES128)) { + zip->encryption_type = ENCRYPTION_WINZIP_AES128; + ret = ARCHIVE_OK; + } else { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "encryption not supported"); + } + } else if (strcmp(val, "aes256") == 0) { + if (is_winzip_aes_encryption_supported( + ENCRYPTION_WINZIP_AES256)) { + zip->encryption_type = ENCRYPTION_WINZIP_AES256; + ret = ARCHIVE_OK; + } else { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "encryption not supported"); + } + } else { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "%s: unknown encryption '%s'", + a->format_name, val); + } + return (ret); + } else if (strcmp(key, "experimental") == 0) { + if (val == NULL || val[0] == 0) { + zip->flags &= ~ ZIP_FLAG_EXPERIMENT_xl; + } else { + zip->flags |= ZIP_FLAG_EXPERIMENT_xl; + } + return (ARCHIVE_OK); + } else if (strcmp(key, "fakecrc32") == 0) { + /* + * FOR TESTING ONLY: disable CRC calculation to speed up + * certain complex tests. + */ + if (val == NULL || val[0] == 0) { + zip->crc32func = real_crc32; + } else { + zip->crc32func = fake_crc32; + } + return (ARCHIVE_OK); + } else if (strcmp(key, "hdrcharset") == 0) { + /* + * Set the character set used in translating filenames. + */ + if (val == NULL || val[0] == 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "%s: hdrcharset option needs a character-set name", + a->format_name); + } else { + zip->opt_sconv = archive_string_conversion_to_charset( + &a->archive, val, 0); + if (zip->opt_sconv != NULL) + ret = ARCHIVE_OK; + else + ret = ARCHIVE_FATAL; + } + return (ret); + } else if (strcmp(key, "zip64") == 0) { + /* + * Bias decisions about Zip64: force them to be + * generated in certain cases where they are not + * forbidden or avoid them in certain cases where they + * are not strictly required. + */ + if (val != NULL && *val != '\0') { + zip->flags |= ZIP_FLAG_FORCE_ZIP64; + zip->flags &= ~ZIP_FLAG_AVOID_ZIP64; + } else { + zip->flags &= ~ZIP_FLAG_FORCE_ZIP64; + zip->flags |= ZIP_FLAG_AVOID_ZIP64; + } + return (ARCHIVE_OK); + } + + /* Note: The "warn" return is just to inform the options + * supervisor that we didn't handle it. It will generate + * a suitable error if no one used this option. */ + return (ARCHIVE_WARN); +} + +int +archive_write_zip_set_compression_deflate(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + int ret = ARCHIVE_FAILED; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW | ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, + "archive_write_zip_set_compression_deflate"); + if (a->archive.archive_format != ARCHIVE_FORMAT_ZIP) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Can only use archive_write_zip_set_compression_deflate" + " with zip format"); + ret = ARCHIVE_FATAL; + } else { +#ifdef HAVE_ZLIB_H + struct zip *zip = a->format_data; + zip->requested_compression = COMPRESSION_DEFLATE; + ret = ARCHIVE_OK; +#else + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "deflate compression not supported"); + ret = ARCHIVE_FAILED; +#endif + } + return (ret); +} + +int +archive_write_zip_set_compression_store(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct zip *zip = a->format_data; + int ret = ARCHIVE_FAILED; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW | ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, + "archive_write_zip_set_compression_deflate"); + if (a->archive.archive_format != ARCHIVE_FORMAT_ZIP) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Can only use archive_write_zip_set_compression_store" + " with zip format"); + ret = ARCHIVE_FATAL; + } else { + zip->requested_compression = COMPRESSION_STORE; + ret = ARCHIVE_OK; + } + return (ret); +} + +int +archive_write_set_format_zip(struct archive *_a) +{ + struct archive_write *a = (struct archive_write *)_a; + struct zip *zip; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, + ARCHIVE_STATE_NEW, "archive_write_set_format_zip"); + + /* If another format was already registered, unregister it. */ + if (a->format_free != NULL) + (a->format_free)(a); + + zip = (struct zip *) calloc(1, sizeof(*zip)); + if (zip == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate zip data"); + return (ARCHIVE_FATAL); + } + + /* "Unspecified" lets us choose the appropriate compression. */ + zip->requested_compression = COMPRESSION_UNSPECIFIED; +#ifdef HAVE_ZLIB_H + zip->deflate_compression_level = Z_DEFAULT_COMPRESSION; +#endif + zip->crc32func = real_crc32; + + /* A buffer used for both compression and encryption. */ + zip->len_buf = 65536; + zip->buf = malloc(zip->len_buf); + if (zip->buf == NULL) { + free(zip); + archive_set_error(&a->archive, ENOMEM, + "Can't allocate compression buffer"); + return (ARCHIVE_FATAL); + } + + a->format_data = zip; + a->format_name = "zip"; + a->format_options = archive_write_zip_options; + a->format_write_header = archive_write_zip_header; + a->format_write_data = archive_write_zip_data; + a->format_finish_entry = archive_write_zip_finish_entry; + a->format_close = archive_write_zip_close; + a->format_free = archive_write_zip_free; + a->archive.archive_format = ARCHIVE_FORMAT_ZIP; + a->archive.archive_format_name = "ZIP"; + + return (ARCHIVE_OK); +} + +static int +is_all_ascii(const char *p) +{ + const unsigned char *pp = (const unsigned char *)p; + + while (*pp) { + if (*pp++ > 127) + return (0); + } + return (1); +} + +static int +archive_write_zip_header(struct archive_write *a, struct archive_entry *entry) +{ + unsigned char local_header[32]; + unsigned char local_extra[144]; + struct zip *zip = a->format_data; + unsigned char *e; + unsigned char *cd_extra; + size_t filename_length; + const char *slink = NULL; + size_t slink_size = 0; + struct archive_string_conv *sconv = get_sconv(a, zip); + int ret, ret2 = ARCHIVE_OK; + mode_t type; + int version_needed = 10; + + /* Ignore types of entries that we don't support. */ + type = archive_entry_filetype(entry); + if (type != AE_IFREG && type != AE_IFDIR && type != AE_IFLNK) { + __archive_write_entry_filetype_unsupported( + &a->archive, entry, "zip"); + return ARCHIVE_FAILED; + }; + + /* If we're not using Zip64, reject large files. */ + if (zip->flags & ZIP_FLAG_AVOID_ZIP64) { + /* Reject entries over 4GB. */ + if (archive_entry_size_is_set(entry) + && (archive_entry_size(entry) > ZIP_4GB_MAX)) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Files > 4GB require Zip64 extensions"); + return ARCHIVE_FAILED; + } + /* Reject entries if archive is > 4GB. */ + if (zip->written_bytes > ZIP_4GB_MAX) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Archives > 4GB require Zip64 extensions"); + return ARCHIVE_FAILED; + } + } + + /* Only regular files can have size > 0. */ + if (type != AE_IFREG) + archive_entry_set_size(entry, 0); + + + /* Reset information from last entry. */ + zip->entry_offset = zip->written_bytes; + zip->entry_uncompressed_limit = INT64_MAX; + zip->entry_compressed_size = 0; + zip->entry_uncompressed_size = 0; + zip->entry_compressed_written = 0; + zip->entry_uncompressed_written = 0; + zip->entry_flags = 0; + zip->entry_uses_zip64 = 0; + zip->entry_crc32 = zip->crc32func(0, NULL, 0); + zip->entry_encryption = 0; + archive_entry_free(zip->entry); + zip->entry = NULL; + + if (zip->cctx_valid) + archive_encrypto_aes_ctr_release(&zip->cctx); + if (zip->hctx_valid) + archive_hmac_sha1_cleanup(&zip->hctx); + zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; + + if (type == AE_IFREG + &&(!archive_entry_size_is_set(entry) + || archive_entry_size(entry) > 0)) { + switch (zip->encryption_type) { + case ENCRYPTION_TRADITIONAL: + case ENCRYPTION_WINZIP_AES128: + case ENCRYPTION_WINZIP_AES256: + zip->entry_flags |= ZIP_ENTRY_FLAG_ENCRYPTED; + zip->entry_encryption = zip->encryption_type; + break; + default: + break; + } + } + + +#if defined(_WIN32) && !defined(__CYGWIN__) + /* Make sure the path separators in pathname, hardlink and symlink + * are all slash '/', not the Windows path separator '\'. */ + zip->entry = __la_win_entry_in_posix_pathseparator(entry); + if (zip->entry == entry) + zip->entry = archive_entry_clone(entry); +#else + zip->entry = archive_entry_clone(entry); +#endif + if (zip->entry == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate zip header data"); + return (ARCHIVE_FATAL); + } + + if (sconv != NULL) { + const char *p; + size_t len; + + if (archive_entry_pathname_l(entry, &p, &len, sconv) != 0) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory for Pathname"); + return (ARCHIVE_FATAL); + } + archive_set_error(&a->archive, + ARCHIVE_ERRNO_FILE_FORMAT, + "Can't translate Pathname '%s' to %s", + archive_entry_pathname(entry), + archive_string_conversion_charset_name(sconv)); + ret2 = ARCHIVE_WARN; + } + if (len > 0) + archive_entry_set_pathname(zip->entry, p); + + /* + * There is no standard for symlink handling; we convert + * it using the same character-set translation that we use + * for filename. + */ + if (type == AE_IFLNK) { + if (archive_entry_symlink_l(entry, &p, &len, sconv)) { + if (errno == ENOMEM) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate memory " + " for Symlink"); + return (ARCHIVE_FATAL); + } + /* No error if we can't convert. */ + } else if (len > 0) + archive_entry_set_symlink(zip->entry, p); + } + } + + /* If filename isn't ASCII and we can use UTF-8, set the UTF-8 flag. */ + if (!is_all_ascii(archive_entry_pathname(zip->entry))) { + if (zip->opt_sconv != NULL) { + if (strcmp(archive_string_conversion_charset_name( + zip->opt_sconv), "UTF-8") == 0) + zip->entry_flags |= ZIP_ENTRY_FLAG_UTF8_NAME; +#if HAVE_NL_LANGINFO + } else if (strcmp(nl_langinfo(CODESET), "UTF-8") == 0) { + zip->entry_flags |= ZIP_ENTRY_FLAG_UTF8_NAME; +#endif + } + } + filename_length = path_length(zip->entry); + + /* Determine appropriate compression and size for this entry. */ + if (type == AE_IFLNK) { + slink = archive_entry_symlink(zip->entry); + if (slink != NULL) + slink_size = strlen(slink); + else + slink_size = 0; + zip->entry_uncompressed_limit = slink_size; + zip->entry_compressed_size = slink_size; + zip->entry_uncompressed_size = slink_size; + zip->entry_crc32 = zip->crc32func(zip->entry_crc32, + (const unsigned char *)slink, slink_size); + zip->entry_compression = COMPRESSION_STORE; + version_needed = 20; + } else if (type != AE_IFREG) { + zip->entry_compression = COMPRESSION_STORE; + zip->entry_uncompressed_limit = 0; + version_needed = 20; + } else if (archive_entry_size_is_set(zip->entry)) { + int64_t size = archive_entry_size(zip->entry); + int64_t additional_size = 0; + + zip->entry_uncompressed_limit = size; + zip->entry_compression = zip->requested_compression; + if (zip->entry_compression == COMPRESSION_UNSPECIFIED) { + zip->entry_compression = COMPRESSION_DEFAULT; + } + if (zip->entry_compression == COMPRESSION_STORE) { + zip->entry_compressed_size = size; + zip->entry_uncompressed_size = size; + version_needed = 10; + } else { + zip->entry_uncompressed_size = size; + version_needed = 20; + } + + if (zip->entry_flags & ZIP_ENTRY_FLAG_ENCRYPTED) { + switch (zip->entry_encryption) { + case ENCRYPTION_TRADITIONAL: + additional_size = TRAD_HEADER_SIZE; + version_needed = 20; + break; + case ENCRYPTION_WINZIP_AES128: + additional_size = WINZIP_AES128_HEADER_SIZE + + AUTH_CODE_SIZE; + version_needed = 20; + break; + case ENCRYPTION_WINZIP_AES256: + additional_size = WINZIP_AES256_HEADER_SIZE + + AUTH_CODE_SIZE; + version_needed = 20; + break; + default: + break; + } + if (zip->entry_compression == COMPRESSION_STORE) + zip->entry_compressed_size += additional_size; + } + + /* + * Set Zip64 extension in any of the following cases + * (this was suggested by discussion on info-zip-dev + * mailing list): + * = Zip64 is being forced by user + * = File is over 4GiB uncompressed + * (including encryption header, if any) + * = File is close to 4GiB and is being compressed + * (compression might make file larger) + */ + if ((zip->flags & ZIP_FLAG_FORCE_ZIP64) + || (zip->entry_uncompressed_size + additional_size > ZIP_4GB_MAX) + || (zip->entry_uncompressed_size > ZIP_4GB_MAX_UNCOMPRESSED + && zip->entry_compression != COMPRESSION_STORE)) { + zip->entry_uses_zip64 = 1; + version_needed = 45; + } + + /* We may know the size, but never the CRC. */ + zip->entry_flags |= ZIP_ENTRY_FLAG_LENGTH_AT_END; + } else { + /* We don't know the size. In this case, we prefer + * deflate (it has a clear end-of-data marker which + * makes length-at-end more reliable) and will + * enable Zip64 extensions unless we're told not to. + */ + zip->entry_compression = COMPRESSION_DEFAULT; + zip->entry_flags |= ZIP_ENTRY_FLAG_LENGTH_AT_END; + if ((zip->flags & ZIP_FLAG_AVOID_ZIP64) == 0) { + zip->entry_uses_zip64 = 1; + version_needed = 45; + } else if (zip->entry_compression == COMPRESSION_STORE) { + version_needed = 10; + } else { + version_needed = 20; + } + + if (zip->entry_flags & ZIP_ENTRY_FLAG_ENCRYPTED) { + switch (zip->entry_encryption) { + case ENCRYPTION_TRADITIONAL: + case ENCRYPTION_WINZIP_AES128: + case ENCRYPTION_WINZIP_AES256: + if (version_needed < 20) + version_needed = 20; + break; + default: + break; + } + } + } + + /* Format the local header. */ + memset(local_header, 0, sizeof(local_header)); + memcpy(local_header, "PK\003\004", 4); + archive_le16enc(local_header + 4, version_needed); + archive_le16enc(local_header + 6, zip->entry_flags); + if (zip->entry_encryption == ENCRYPTION_WINZIP_AES128 + || zip->entry_encryption == ENCRYPTION_WINZIP_AES256) + archive_le16enc(local_header + 8, WINZIP_AES_ENCRYPTION); + else + archive_le16enc(local_header + 8, zip->entry_compression); + archive_le32enc(local_header + 10, + dos_time(archive_entry_mtime(zip->entry))); + archive_le32enc(local_header + 14, zip->entry_crc32); + if (zip->entry_uses_zip64) { + /* Zip64 data in the local header "must" include both + * compressed and uncompressed sizes AND those fields + * are included only if these are 0xffffffff; + * THEREFORE these must be set this way, even if we + * know one of them is smaller. */ + archive_le32enc(local_header + 18, ZIP_4GB_MAX); + archive_le32enc(local_header + 22, ZIP_4GB_MAX); + } else { + archive_le32enc(local_header + 18, (uint32_t)zip->entry_compressed_size); + archive_le32enc(local_header + 22, (uint32_t)zip->entry_uncompressed_size); + } + archive_le16enc(local_header + 26, (uint16_t)filename_length); + + if (zip->entry_encryption == ENCRYPTION_TRADITIONAL) { + if (zip->entry_flags & ZIP_ENTRY_FLAG_LENGTH_AT_END) + zip->trad_chkdat = local_header[11]; + else + zip->trad_chkdat = local_header[17]; + } + + /* Format as much of central directory file header as we can: */ + zip->file_header = cd_alloc(zip, 46); + /* If (zip->file_header == NULL) XXXX */ + ++zip->central_directory_entries; + memset(zip->file_header, 0, 46); + memcpy(zip->file_header, "PK\001\002", 4); + /* "Made by PKZip 2.0 on Unix." */ + archive_le16enc(zip->file_header + 4, 3 * 256 + version_needed); + archive_le16enc(zip->file_header + 6, version_needed); + archive_le16enc(zip->file_header + 8, zip->entry_flags); + if (zip->entry_encryption == ENCRYPTION_WINZIP_AES128 + || zip->entry_encryption == ENCRYPTION_WINZIP_AES256) + archive_le16enc(zip->file_header + 10, WINZIP_AES_ENCRYPTION); + else + archive_le16enc(zip->file_header + 10, zip->entry_compression); + archive_le32enc(zip->file_header + 12, + dos_time(archive_entry_mtime(zip->entry))); + archive_le16enc(zip->file_header + 28, (uint16_t)filename_length); + /* Following Info-Zip, store mode in the "external attributes" field. */ + archive_le32enc(zip->file_header + 38, + ((uint32_t)archive_entry_mode(zip->entry)) << 16); + e = cd_alloc(zip, filename_length); + /* If (e == NULL) XXXX */ + copy_path(zip->entry, e); + + /* Format extra data. */ + memset(local_extra, 0, sizeof(local_extra)); + e = local_extra; + + /* First, extra blocks that are the same between + * the local file header and the central directory. + * We format them once and then duplicate them. */ + + /* UT timestamp, length depends on what timestamps are set. */ + memcpy(e, "UT", 2); + archive_le16enc(e + 2, + 1 + + (archive_entry_mtime_is_set(entry) ? 4 : 0) + + (archive_entry_atime_is_set(entry) ? 4 : 0) + + (archive_entry_ctime_is_set(entry) ? 4 : 0)); + e += 4; + *e++ = + (archive_entry_mtime_is_set(entry) ? 1 : 0) + | (archive_entry_atime_is_set(entry) ? 2 : 0) + | (archive_entry_ctime_is_set(entry) ? 4 : 0); + if (archive_entry_mtime_is_set(entry)) { + archive_le32enc(e, (uint32_t)archive_entry_mtime(entry)); + e += 4; + } + if (archive_entry_atime_is_set(entry)) { + archive_le32enc(e, (uint32_t)archive_entry_atime(entry)); + e += 4; + } + if (archive_entry_ctime_is_set(entry)) { + archive_le32enc(e, (uint32_t)archive_entry_ctime(entry)); + e += 4; + } + + /* ux Unix extra data, length 11, version 1 */ + /* TODO: If uid < 64k, use 2 bytes, ditto for gid. */ + memcpy(e, "ux\013\000\001", 5); + e += 5; + *e++ = 4; /* Length of following UID */ + archive_le32enc(e, (uint32_t)archive_entry_uid(entry)); + e += 4; + *e++ = 4; /* Length of following GID */ + archive_le32enc(e, (uint32_t)archive_entry_gid(entry)); + e += 4; + + /* AES extra data field: WinZIP AES information, ID=0x9901 */ + if ((zip->entry_flags & ZIP_ENTRY_FLAG_ENCRYPTED) + && (zip->entry_encryption == ENCRYPTION_WINZIP_AES128 + || zip->entry_encryption == ENCRYPTION_WINZIP_AES256)) { + + memcpy(e, "\001\231\007\000\001\000AE", 8); + /* AES vendor version AE-2 does not store a CRC. + * WinZip 11 uses AE-1, which does store the CRC, + * but it does not store the CRC when the file size + * is less than 20 bytes. So we simulate what + * WinZip 11 does. + * NOTE: WinZip 9.0 and 10.0 uses AE-2 by default. */ + if (archive_entry_size_is_set(zip->entry) + && archive_entry_size(zip->entry) < 20) { + archive_le16enc(e+4, AES_VENDOR_AE_2); + zip->aes_vendor = AES_VENDOR_AE_2;/* no CRC. */ + } else + zip->aes_vendor = AES_VENDOR_AE_1; + e += 8; + /* AES encryption strength. */ + *e++ = (zip->entry_encryption == ENCRYPTION_WINZIP_AES128)?1:3; + /* Actual compression method. */ + archive_le16enc(e, zip->entry_compression); + e += 2; + } + + /* Copy UT ,ux, and AES-extra into central directory as well. */ + zip->file_header_extra_offset = zip->central_directory_bytes; + cd_extra = cd_alloc(zip, e - local_extra); + memcpy(cd_extra, local_extra, e - local_extra); + + /* + * Following extra blocks vary between local header and + * central directory. These are the local header versions. + * Central directory versions get formatted in + * archive_write_zip_finish_entry() below. + */ + + /* "[Zip64 entry] in the local header MUST include BOTH + * original [uncompressed] and compressed size fields." */ + if (zip->entry_uses_zip64) { + unsigned char *zip64_start = e; + memcpy(e, "\001\000\020\000", 4); + e += 4; + archive_le64enc(e, zip->entry_uncompressed_size); + e += 8; + archive_le64enc(e, zip->entry_compressed_size); + e += 8; + archive_le16enc(zip64_start + 2, (uint16_t)(e - (zip64_start + 4))); + } + + if (zip->flags & ZIP_FLAG_EXPERIMENT_xl) { + /* Experimental 'xl' extension to improve streaming. */ + unsigned char *external_info = e; + int included = 7; + memcpy(e, "xl\000\000", 4); // 0x6c65 + 2-byte length + e += 4; + e[0] = included; /* bitmap of included fields */ + e += 1; + if (included & 1) { + archive_le16enc(e, /* "Version created by" */ + 3 * 256 + version_needed); + e += 2; + } + if (included & 2) { + archive_le16enc(e, 0); /* internal file attributes */ + e += 2; + } + if (included & 4) { + archive_le32enc(e, /* external file attributes */ + ((uint32_t)archive_entry_mode(zip->entry)) << 16); + e += 4; + } + if (included & 8) { + // Libarchive does not currently support file comments. + } + archive_le16enc(external_info + 2, (uint16_t)(e - (external_info + 4))); + } + + /* Update local header with size of extra data and write it all out: */ + archive_le16enc(local_header + 28, (uint16_t)(e - local_extra)); + + ret = __archive_write_output(a, local_header, 30); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + zip->written_bytes += 30; + + ret = write_path(zip->entry, a); + if (ret <= ARCHIVE_OK) + return (ARCHIVE_FATAL); + zip->written_bytes += ret; + + ret = __archive_write_output(a, local_extra, e - local_extra); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + zip->written_bytes += e - local_extra; + + /* For symlinks, write the body now. */ + if (slink != NULL) { + ret = __archive_write_output(a, slink, slink_size); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + zip->entry_compressed_written += slink_size; + zip->entry_uncompressed_written += slink_size; + zip->written_bytes += slink_size; + } + +#ifdef HAVE_ZLIB_H + if (zip->entry_compression == COMPRESSION_DEFLATE) { + zip->stream.zalloc = Z_NULL; + zip->stream.zfree = Z_NULL; + zip->stream.opaque = Z_NULL; + zip->stream.next_out = zip->buf; + zip->stream.avail_out = (uInt)zip->len_buf; + if (deflateInit2(&zip->stream, zip->deflate_compression_level, + Z_DEFLATED, -15, 8, Z_DEFAULT_STRATEGY) != Z_OK) { + archive_set_error(&a->archive, ENOMEM, + "Can't init deflate compressor"); + return (ARCHIVE_FATAL); + } + } +#endif + + return (ret2); +} + +static ssize_t +archive_write_zip_data(struct archive_write *a, const void *buff, size_t s) +{ + int ret; + struct zip *zip = a->format_data; + + if ((int64_t)s > zip->entry_uncompressed_limit) + s = (size_t)zip->entry_uncompressed_limit; + zip->entry_uncompressed_written += s; + + if (s == 0) return 0; + + if (zip->entry_flags & ZIP_ENTRY_FLAG_ENCRYPTED) { + switch (zip->entry_encryption) { + case ENCRYPTION_TRADITIONAL: + /* Initialize traditional PKWARE encryption context. */ + if (!zip->tctx_valid) { + ret = init_traditional_pkware_encryption(a); + if (ret != ARCHIVE_OK) + return (ret); + zip->tctx_valid = 1; + } + break; + case ENCRYPTION_WINZIP_AES128: + case ENCRYPTION_WINZIP_AES256: + if (!zip->cctx_valid) { + ret = init_winzip_aes_encryption(a); + if (ret != ARCHIVE_OK) + return (ret); + zip->cctx_valid = zip->hctx_valid = 1; + } + break; + default: + break; + } + } + + switch (zip->entry_compression) { + case COMPRESSION_STORE: + if (zip->tctx_valid || zip->cctx_valid) { + const uint8_t *rb = (const uint8_t *)buff; + const uint8_t * const re = rb + s; + + while (rb < re) { + size_t l; + + if (zip->tctx_valid) { + l = trad_enc_encrypt_update(&zip->tctx, + rb, re - rb, + zip->buf, zip->len_buf); + } else { + l = zip->len_buf; + ret = archive_encrypto_aes_ctr_update( + &zip->cctx, + rb, re - rb, zip->buf, &l); + if (ret < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Failed to encrypt file"); + return (ARCHIVE_FAILED); + } + archive_hmac_sha1_update(&zip->hctx, + zip->buf, l); + } + ret = __archive_write_output(a, zip->buf, l); + if (ret != ARCHIVE_OK) + return (ret); + zip->entry_compressed_written += l; + zip->written_bytes += l; + rb += l; + } + } else { + ret = __archive_write_output(a, buff, s); + if (ret != ARCHIVE_OK) + return (ret); + zip->written_bytes += s; + zip->entry_compressed_written += s; + } + break; +#if HAVE_ZLIB_H + case COMPRESSION_DEFLATE: + zip->stream.next_in = (unsigned char*)(uintptr_t)buff; + zip->stream.avail_in = (uInt)s; + do { + ret = deflate(&zip->stream, Z_NO_FLUSH); + if (ret == Z_STREAM_ERROR) + return (ARCHIVE_FATAL); + if (zip->stream.avail_out == 0) { + if (zip->tctx_valid) { + trad_enc_encrypt_update(&zip->tctx, + zip->buf, zip->len_buf, + zip->buf, zip->len_buf); + } else if (zip->cctx_valid) { + size_t outl = zip->len_buf; + ret = archive_encrypto_aes_ctr_update( + &zip->cctx, + zip->buf, zip->len_buf, + zip->buf, &outl); + if (ret < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Failed to encrypt file"); + return (ARCHIVE_FAILED); + } + archive_hmac_sha1_update(&zip->hctx, + zip->buf, zip->len_buf); + } + ret = __archive_write_output(a, zip->buf, + zip->len_buf); + if (ret != ARCHIVE_OK) + return (ret); + zip->entry_compressed_written += zip->len_buf; + zip->written_bytes += zip->len_buf; + zip->stream.next_out = zip->buf; + zip->stream.avail_out = (uInt)zip->len_buf; + } + } while (zip->stream.avail_in != 0); + break; +#endif + + default: + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Invalid ZIP compression type"); + return ARCHIVE_FATAL; + } + + zip->entry_uncompressed_limit -= s; + if (!zip->cctx_valid || zip->aes_vendor != AES_VENDOR_AE_2) + zip->entry_crc32 = + zip->crc32func(zip->entry_crc32, buff, (unsigned)s); + return (s); + +} + +static int +archive_write_zip_finish_entry(struct archive_write *a) +{ + struct zip *zip = a->format_data; + int ret; + +#if HAVE_ZLIB_H + if (zip->entry_compression == COMPRESSION_DEFLATE) { + for (;;) { + size_t remainder; + + ret = deflate(&zip->stream, Z_FINISH); + if (ret == Z_STREAM_ERROR) + return (ARCHIVE_FATAL); + remainder = zip->len_buf - zip->stream.avail_out; + if (zip->tctx_valid) { + trad_enc_encrypt_update(&zip->tctx, + zip->buf, remainder, zip->buf, remainder); + } else if (zip->cctx_valid) { + size_t outl = remainder; + ret = archive_encrypto_aes_ctr_update( + &zip->cctx, zip->buf, remainder, + zip->buf, &outl); + if (ret < 0) { + archive_set_error(&a->archive, + ARCHIVE_ERRNO_MISC, + "Failed to encrypt file"); + return (ARCHIVE_FAILED); + } + archive_hmac_sha1_update(&zip->hctx, + zip->buf, remainder); + } + ret = __archive_write_output(a, zip->buf, remainder); + if (ret != ARCHIVE_OK) + return (ret); + zip->entry_compressed_written += remainder; + zip->written_bytes += remainder; + zip->stream.next_out = zip->buf; + if (zip->stream.avail_out != 0) + break; + zip->stream.avail_out = (uInt)zip->len_buf; + } + deflateEnd(&zip->stream); + } +#endif + if (zip->hctx_valid) { + uint8_t hmac[20]; + size_t hmac_len = 20; + + archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len); + ret = __archive_write_output(a, hmac, AUTH_CODE_SIZE); + if (ret != ARCHIVE_OK) + return (ret); + zip->entry_compressed_written += AUTH_CODE_SIZE; + zip->written_bytes += AUTH_CODE_SIZE; + } + + /* Write trailing data descriptor. */ + if ((zip->entry_flags & ZIP_ENTRY_FLAG_LENGTH_AT_END) != 0) { + char d[24]; + memcpy(d, "PK\007\010", 4); + if (zip->cctx_valid && zip->aes_vendor == AES_VENDOR_AE_2) + archive_le32enc(d + 4, 0);/* no CRC.*/ + else + archive_le32enc(d + 4, zip->entry_crc32); + if (zip->entry_uses_zip64) { + archive_le64enc(d + 8, + (uint64_t)zip->entry_compressed_written); + archive_le64enc(d + 16, + (uint64_t)zip->entry_uncompressed_written); + ret = __archive_write_output(a, d, 24); + zip->written_bytes += 24; + } else { + archive_le32enc(d + 8, + (uint32_t)zip->entry_compressed_written); + archive_le32enc(d + 12, + (uint32_t)zip->entry_uncompressed_written); + ret = __archive_write_output(a, d, 16); + zip->written_bytes += 16; + } + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + } + + /* Append Zip64 extra data to central directory information. */ + if (zip->entry_compressed_written > ZIP_4GB_MAX + || zip->entry_uncompressed_written > ZIP_4GB_MAX + || zip->entry_offset > ZIP_4GB_MAX) { + unsigned char zip64[32]; + unsigned char *z = zip64, *zd; + memcpy(z, "\001\000\000\000", 4); + z += 4; + if (zip->entry_uncompressed_written >= ZIP_4GB_MAX) { + archive_le64enc(z, zip->entry_uncompressed_written); + z += 8; + } + if (zip->entry_compressed_written >= ZIP_4GB_MAX) { + archive_le64enc(z, zip->entry_compressed_written); + z += 8; + } + if (zip->entry_offset >= ZIP_4GB_MAX) { + archive_le64enc(z, zip->entry_offset); + z += 8; + } + archive_le16enc(zip64 + 2, (uint16_t)(z - (zip64 + 4))); + zd = cd_alloc(zip, z - zip64); + if (zd == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate zip data"); + return (ARCHIVE_FATAL); + } + memcpy(zd, zip64, z - zip64); + /* Zip64 means version needs to be set to at least 4.5 */ + if (archive_le16dec(zip->file_header + 6) < 45) + archive_le16enc(zip->file_header + 6, 45); + } + + /* Fix up central directory file header. */ + if (zip->cctx_valid && zip->aes_vendor == AES_VENDOR_AE_2) + archive_le32enc(zip->file_header + 16, 0);/* no CRC.*/ + else + archive_le32enc(zip->file_header + 16, zip->entry_crc32); + archive_le32enc(zip->file_header + 20, + (uint32_t)zipmin(zip->entry_compressed_written, + ZIP_4GB_MAX)); + archive_le32enc(zip->file_header + 24, + (uint32_t)zipmin(zip->entry_uncompressed_written, + ZIP_4GB_MAX)); + archive_le16enc(zip->file_header + 30, + (uint16_t)(zip->central_directory_bytes - zip->file_header_extra_offset)); + archive_le32enc(zip->file_header + 42, + (uint32_t)zipmin(zip->entry_offset, + ZIP_4GB_MAX)); + + return (ARCHIVE_OK); +} + +static int +archive_write_zip_close(struct archive_write *a) +{ + uint8_t buff[64]; + int64_t offset_start, offset_end; + struct zip *zip = a->format_data; + struct cd_segment *segment; + int ret; + + offset_start = zip->written_bytes; + segment = zip->central_directory; + while (segment != NULL) { + ret = __archive_write_output(a, + segment->buff, segment->p - segment->buff); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + zip->written_bytes += segment->p - segment->buff; + segment = segment->next; + } + offset_end = zip->written_bytes; + + /* If central dir info is too large, write Zip64 end-of-cd */ + if (offset_end - offset_start > ZIP_4GB_MAX + || offset_start > ZIP_4GB_MAX + || zip->central_directory_entries > 0xffffUL + || (zip->flags & ZIP_FLAG_FORCE_ZIP64)) { + /* Zip64 end-of-cd record */ + memset(buff, 0, 56); + memcpy(buff, "PK\006\006", 4); + archive_le64enc(buff + 4, 44); + archive_le16enc(buff + 12, 45); + archive_le16enc(buff + 14, 45); + /* This is disk 0 of 0. */ + archive_le64enc(buff + 24, zip->central_directory_entries); + archive_le64enc(buff + 32, zip->central_directory_entries); + archive_le64enc(buff + 40, offset_end - offset_start); + archive_le64enc(buff + 48, offset_start); + ret = __archive_write_output(a, buff, 56); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + zip->written_bytes += 56; + + /* Zip64 end-of-cd locator record. */ + memset(buff, 0, 20); + memcpy(buff, "PK\006\007", 4); + archive_le32enc(buff + 4, 0); + archive_le64enc(buff + 8, offset_end); + archive_le32enc(buff + 16, 1); + ret = __archive_write_output(a, buff, 20); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + zip->written_bytes += 20; + + } + + /* Format and write end of central directory. */ + memset(buff, 0, sizeof(buff)); + memcpy(buff, "PK\005\006", 4); + archive_le16enc(buff + 8, (uint16_t)zipmin(0xffffU, + zip->central_directory_entries)); + archive_le16enc(buff + 10, (uint16_t)zipmin(0xffffU, + zip->central_directory_entries)); + archive_le32enc(buff + 12, + (uint32_t)zipmin(ZIP_4GB_MAX, (offset_end - offset_start))); + archive_le32enc(buff + 16, + (uint32_t)zipmin(ZIP_4GB_MAX, offset_start)); + ret = __archive_write_output(a, buff, 22); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + zip->written_bytes += 22; + return (ARCHIVE_OK); +} + +static int +archive_write_zip_free(struct archive_write *a) +{ + struct zip *zip; + struct cd_segment *segment; + + zip = a->format_data; + while (zip->central_directory != NULL) { + segment = zip->central_directory; + zip->central_directory = segment->next; + free(segment->buff); + free(segment); + } + free(zip->buf); + archive_entry_free(zip->entry); + if (zip->cctx_valid) + archive_encrypto_aes_ctr_release(&zip->cctx); + if (zip->hctx_valid) + archive_hmac_sha1_cleanup(&zip->hctx); + /* TODO: Free opt_sconv, sconv_default */ + + free(zip); + a->format_data = NULL; + return (ARCHIVE_OK); +} + +/* Convert into MSDOS-style date/time. */ +static unsigned int +dos_time(const time_t unix_time) +{ + struct tm *t; + unsigned int dt; +#if defined(HAVE_LOCALTIME_R) || defined(HAVE__LOCALTIME64_S) + struct tm tmbuf; +#endif +#if defined(HAVE__LOCALTIME64_S) + errno_t terr; + __time64_t tmptime; +#endif + + /* This will not preserve time when creating/extracting the archive + * on two systems with different time zones. */ +#if defined(HAVE_LOCALTIME_R) + t = localtime_r(&unix_time, &tmbuf); +#elif defined(HAVE__LOCALTIME64_S) + tmptime = unix_time; + terr = _localtime64_s(&tmbuf, &tmptime); + if (terr) + t = NULL; + else + t = &tmbuf; +#else + t = localtime(&unix_time); +#endif + + /* MSDOS-style date/time is only between 1980-01-01 and 2107-12-31 */ + if (t->tm_year < 1980 - 1900) + /* Set minimum date/time '1980-01-01 00:00:00'. */ + dt = 0x00210000U; + else if (t->tm_year > 2107 - 1900) + /* Set maximum date/time '2107-12-31 23:59:58'. */ + dt = 0xff9fbf7dU; + else { + dt = 0; + dt += ((t->tm_year - 80) & 0x7f) << 9; + dt += ((t->tm_mon + 1) & 0x0f) << 5; + dt += (t->tm_mday & 0x1f); + dt <<= 16; + dt += (t->tm_hour & 0x1f) << 11; + dt += (t->tm_min & 0x3f) << 5; + dt += (t->tm_sec & 0x3e) >> 1; /* Only counting every 2 seconds. */ + } + return dt; +} + +static size_t +path_length(struct archive_entry *entry) +{ + mode_t type; + const char *path; + size_t len; + + type = archive_entry_filetype(entry); + path = archive_entry_pathname(entry); + + if (path == NULL) + return (0); + len = strlen(path); + if (type == AE_IFDIR && (path[0] == '\0' || path[len - 1] != '/')) + ++len; /* Space for the trailing / */ + return len; +} + +static int +write_path(struct archive_entry *entry, struct archive_write *archive) +{ + int ret; + const char *path; + mode_t type; + size_t written_bytes; + + path = archive_entry_pathname(entry); + type = archive_entry_filetype(entry); + written_bytes = 0; + + if (path == NULL) + return (ARCHIVE_FATAL); + + ret = __archive_write_output(archive, path, strlen(path)); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + written_bytes += strlen(path); + + /* Folders are recognized by a trailing slash. */ + if ((type == AE_IFDIR) & (path[strlen(path) - 1] != '/')) { + ret = __archive_write_output(archive, "/", 1); + if (ret != ARCHIVE_OK) + return (ARCHIVE_FATAL); + written_bytes += 1; + } + + return ((int)written_bytes); +} + +static void +copy_path(struct archive_entry *entry, unsigned char *p) +{ + const char *path; + size_t pathlen; + mode_t type; + + path = archive_entry_pathname(entry); + pathlen = strlen(path); + type = archive_entry_filetype(entry); + + memcpy(p, path, pathlen); + + /* Folders are recognized by a trailing slash. */ + if ((type == AE_IFDIR) && (path[pathlen - 1] != '/')) + p[pathlen] = '/'; +} + + +static struct archive_string_conv * +get_sconv(struct archive_write *a, struct zip *zip) +{ + if (zip->opt_sconv != NULL) + return (zip->opt_sconv); + + if (!zip->init_default_conversion) { + zip->sconv_default = + archive_string_default_conversion_for_write(&(a->archive)); + zip->init_default_conversion = 1; + } + return (zip->sconv_default); +} + +/* + Traditional PKWARE Decryption functions. + */ + +static void +trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c) +{ + uint8_t t; +#define CRC32(c, b) (crc32(c ^ 0xffffffffUL, &b, 1) ^ 0xffffffffUL) + + ctx->keys[0] = CRC32(ctx->keys[0], c); + ctx->keys[1] = (ctx->keys[1] + (ctx->keys[0] & 0xff)) * 134775813L + 1; + t = (ctx->keys[1] >> 24) & 0xff; + ctx->keys[2] = CRC32(ctx->keys[2], t); +#undef CRC32 +} + +static uint8_t +trad_enc_decrypt_byte(struct trad_enc_ctx *ctx) +{ + unsigned temp = ctx->keys[2] | 2; + return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff; +} + +static unsigned +trad_enc_encrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in, + size_t in_len, uint8_t *out, size_t out_len) +{ + unsigned i, max; + + max = (unsigned)((in_len < out_len)? in_len: out_len); + + for (i = 0; i < max; i++) { + uint8_t t = in[i]; + out[i] = t ^ trad_enc_decrypt_byte(ctx); + trad_enc_update_keys(ctx, t); + } + return i; +} + +static int +trad_enc_init(struct trad_enc_ctx *ctx, const char *pw, size_t pw_len) +{ + + ctx->keys[0] = 305419896L; + ctx->keys[1] = 591751049L; + ctx->keys[2] = 878082192L; + + for (;pw_len; --pw_len) + trad_enc_update_keys(ctx, *pw++); + return 0; +} + +static int +is_traditional_pkware_encryption_supported(void) +{ + uint8_t key[TRAD_HEADER_SIZE]; + + if (archive_random(key, sizeof(key)-1) != ARCHIVE_OK) + return (0); + return (1); +} + +static int +init_traditional_pkware_encryption(struct archive_write *a) +{ + struct zip *zip = a->format_data; + const char *passphrase; + uint8_t key[TRAD_HEADER_SIZE]; + uint8_t key_encrypted[TRAD_HEADER_SIZE]; + int ret; + + passphrase = __archive_write_get_passphrase(a); + if (passphrase == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Encryption needs passphrase"); + return ARCHIVE_FAILED; + } + if (archive_random(key, sizeof(key)-1) != ARCHIVE_OK) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Can't generate random number for encryption"); + return ARCHIVE_FATAL; + } + trad_enc_init(&zip->tctx, passphrase, strlen(passphrase)); + /* Set the last key code which will be used as a check code + * for verifying passphrase in decryption. */ + key[TRAD_HEADER_SIZE-1] = zip->trad_chkdat; + trad_enc_encrypt_update(&zip->tctx, key, TRAD_HEADER_SIZE, + key_encrypted, TRAD_HEADER_SIZE); + /* Write encrypted keys in the top of the file content. */ + ret = __archive_write_output(a, key_encrypted, TRAD_HEADER_SIZE); + if (ret != ARCHIVE_OK) + return (ret); + zip->written_bytes += TRAD_HEADER_SIZE; + zip->entry_compressed_written += TRAD_HEADER_SIZE; + return (ret); +} + +static int +init_winzip_aes_encryption(struct archive_write *a) +{ + struct zip *zip = a->format_data; + const char *passphrase; + size_t key_len, salt_len; + uint8_t salt[16 + 2]; + uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; + int ret; + + passphrase = __archive_write_get_passphrase(a); + if (passphrase == NULL) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Encryption needs passphrase"); + return (ARCHIVE_FAILED); + } + if (zip->entry_encryption == ENCRYPTION_WINZIP_AES128) { + salt_len = 8; + key_len = 16; + } else { + /* AES 256 */ + salt_len = 16; + key_len = 32; + } + if (archive_random(salt, salt_len) != ARCHIVE_OK) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Can't generate random number for encryption"); + return (ARCHIVE_FATAL); + } + archive_pbkdf2_sha1(passphrase, strlen(passphrase), + salt, salt_len, 1000, derived_key, key_len * 2 + 2); + + ret = archive_encrypto_aes_ctr_init(&zip->cctx, derived_key, key_len); + if (ret != 0) { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Decryption is unsupported due to lack of crypto library"); + return (ARCHIVE_FAILED); + } + ret = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, + key_len); + if (ret != 0) { + archive_encrypto_aes_ctr_release(&zip->cctx); + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Failed to initialize HMAC-SHA1"); + return (ARCHIVE_FAILED); + } + + /* Set a password verification value after the 'salt'. */ + salt[salt_len] = derived_key[key_len * 2]; + salt[salt_len + 1] = derived_key[key_len * 2 + 1]; + + /* Write encrypted keys in the top of the file content. */ + ret = __archive_write_output(a, salt, salt_len + 2); + if (ret != ARCHIVE_OK) + return (ret); + zip->written_bytes += salt_len + 2; + zip->entry_compressed_written += salt_len + 2; + + return (ARCHIVE_OK); +} + +static int +is_winzip_aes_encryption_supported(int encryption) +{ + size_t key_len, salt_len; + uint8_t salt[16 + 2]; + uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; + archive_crypto_ctx cctx; + archive_hmac_sha1_ctx hctx; + int ret; + + if (encryption == ENCRYPTION_WINZIP_AES128) { + salt_len = 8; + key_len = 16; + } else { + /* AES 256 */ + salt_len = 16; + key_len = 32; + } + if (archive_random(salt, salt_len) != ARCHIVE_OK) + return (0); + ret = archive_pbkdf2_sha1("p", 1, salt, salt_len, 1000, + derived_key, key_len * 2 + 2); + if (ret != 0) + return (0); + + ret = archive_encrypto_aes_ctr_init(&cctx, derived_key, key_len); + if (ret != 0) + return (0); + ret = archive_hmac_sha1_init(&hctx, derived_key + key_len, + key_len); + archive_encrypto_aes_ctr_release(&cctx); + if (ret != 0) + return (0); + archive_hmac_sha1_cleanup(&hctx); + return (1); +} diff --git a/c/archive_write_set_options.c b/c/archive_write_set_options.c new file mode 100644 --- /dev/null +++ b/c/archive_write_set_options.c @@ -0,0 +1,130 @@ +/*- + * Copyright (c) 2003-2010 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#include "archive_write_private.h" +#include "archive_options_private.h" + +static int archive_set_format_option(struct archive *a, + const char *m, const char *o, const char *v); +static int archive_set_filter_option(struct archive *a, + const char *m, const char *o, const char *v); +static int archive_set_option(struct archive *a, + const char *m, const char *o, const char *v); + +int +archive_write_set_format_option(struct archive *a, const char *m, const char *o, + const char *v) +{ + return _archive_set_option(a, m, o, v, + ARCHIVE_WRITE_MAGIC, "archive_write_set_format_option", + archive_set_format_option); +} + +int +archive_write_set_filter_option(struct archive *a, const char *m, const char *o, + const char *v) +{ + return _archive_set_option(a, m, o, v, + ARCHIVE_WRITE_MAGIC, "archive_write_set_filter_option", + archive_set_filter_option); +} + +int +archive_write_set_option(struct archive *a, const char *m, const char *o, + const char *v) +{ + return _archive_set_option(a, m, o, v, + ARCHIVE_WRITE_MAGIC, "archive_write_set_option", + archive_set_option); +} + +int +archive_write_set_options(struct archive *a, const char *options) +{ + return _archive_set_options(a, options, + ARCHIVE_WRITE_MAGIC, "archive_write_set_options", + archive_set_option); +} + +static int +archive_set_format_option(struct archive *_a, const char *m, const char *o, + const char *v) +{ + struct archive_write *a = (struct archive_write *)_a; + + if (a->format_name == NULL) + return (m == NULL)?ARCHIVE_FAILED:ARCHIVE_WARN - 1; + /* If the format name didn't match, return a special code for + * _archive_set_option[s]. */ + if (m != NULL && strcmp(m, a->format_name) != 0) + return (ARCHIVE_WARN - 1); + if (a->format_options == NULL) + return (ARCHIVE_WARN); + return a->format_options(a, o, v); +} + +static int +archive_set_filter_option(struct archive *_a, const char *m, const char *o, + const char *v) +{ + struct archive_write *a = (struct archive_write *)_a; + struct archive_write_filter *filter; + int r, rv = ARCHIVE_WARN; + + for (filter = a->filter_first; filter != NULL; filter = filter->next_filter) { + if (filter->options == NULL) + continue; + if (m != NULL && strcmp(filter->name, m) != 0) + continue; + + r = filter->options(filter, o, v); + + if (r == ARCHIVE_FATAL) + return (ARCHIVE_FATAL); + + if (m != NULL) + return (r); + + if (r == ARCHIVE_OK) + rv = ARCHIVE_OK; + } + /* If the filter name didn't match, return a special code for + * _archive_set_option[s]. */ + if (rv == ARCHIVE_WARN && m != NULL) + rv = ARCHIVE_WARN - 1; + return (rv); +} + +static int +archive_set_option(struct archive *a, const char *m, const char *o, + const char *v) +{ + return _archive_set_either_option(a, m, o, v, + archive_set_format_option, + archive_set_filter_option); +} diff --git a/c/archive_write_set_passphrase.c b/c/archive_write_set_passphrase.c new file mode 100644 --- /dev/null +++ b/c/archive_write_set_passphrase.c @@ -0,0 +1,95 @@ +/*- + * Copyright (c) 2014 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" +__FBSDID("$FreeBSD$"); + +#ifdef HAVE_ERRNO_H +#include +#endif +#include "archive_write_private.h" + +int +archive_write_set_passphrase(struct archive *_a, const char *p) +{ + struct archive_write *a = (struct archive_write *)_a; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, + "archive_write_set_passphrase"); + + if (p == NULL || p[0] == '\0') { + archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, + "Empty passphrase is unacceptable"); + return (ARCHIVE_FAILED); + } + free(a->passphrase); + a->passphrase = strdup(p); + if (a->passphrase == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate data for passphrase"); + return (ARCHIVE_FATAL); + } + return (ARCHIVE_OK); +} + + +int +archive_write_set_passphrase_callback(struct archive *_a, void *client_data, + archive_passphrase_callback *cb) +{ + struct archive_write *a = (struct archive_write *)_a; + + archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, + "archive_write_set_passphrase_callback"); + + a->passphrase_callback = cb; + a->passphrase_client_data = client_data; + return (ARCHIVE_OK); +} + + +const char * +__archive_write_get_passphrase(struct archive_write *a) +{ + + if (a->passphrase != NULL) + return (a->passphrase); + + if (a->passphrase_callback != NULL) { + const char *p; + p = a->passphrase_callback(&a->archive, + a->passphrase_client_data); + if (p != NULL) { + a->passphrase = strdup(p); + if (a->passphrase == NULL) { + archive_set_error(&a->archive, ENOMEM, + "Can't allocate data for passphrase"); + return (NULL); + } + return (a->passphrase); + } + } + return (NULL); +} diff --git a/c/archive_xxhash.h b/c/archive_xxhash.h new file mode 100644 --- /dev/null +++ b/c/archive_xxhash.h @@ -0,0 +1,48 @@ +/*- + * Copyright (c) 2014 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + */ + +#ifndef ARCHIVE_XXHASH_H_INCLUDED +#define ARCHIVE_XXHASH_H_INCLUDED + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif + + +typedef enum { XXH_OK=0, XXH_ERROR } XXH_errorcode; + +struct archive_xxhash { + unsigned int (*XXH32)(const void* input, unsigned int len, + unsigned int seed); + void* (*XXH32_init)(unsigned int seed); + XXH_errorcode (*XXH32_update)(void* state, const void* input, + unsigned int len); + unsigned int (*XXH32_digest)(void* state); +}; + +extern const struct archive_xxhash __archive_xxhash; + +#endif diff --git a/c/autoconf-darwin/config.h b/c/autoconf-darwin/config.h new file mode 100644 --- /dev/null +++ b/c/autoconf-darwin/config.h @@ -0,0 +1,1428 @@ +/* config.h. Generated from config.h.in by configure. */ +/* config.h.in. Generated from configure.ac by autoheader. */ + +/* Darwin ACL support */ +/* #undef ARCHIVE_ACL_DARWIN */ + +/* FreeBSD ACL support */ +/* #undef ARCHIVE_ACL_FREEBSD */ + +/* FreeBSD NFSv4 ACL support */ +/* #undef ARCHIVE_ACL_FREEBSD_NFS4 */ + +/* Linux POSIX.1e ACL support via libacl */ +/* #undef ARCHIVE_ACL_LIBACL */ + +/* Linux NFSv4 ACL support via librichacl */ +/* #undef ARCHIVE_ACL_LIBRICHACL */ + +/* Solaris ACL support */ +/* #undef ARCHIVE_ACL_SUNOS */ + +/* Solaris NFSv4 ACL support */ +/* #undef ARCHIVE_ACL_SUNOS_NFS4 */ + +/* MD5 via ARCHIVE_CRYPTO_MD5_LIBC supported. */ +/* #undef ARCHIVE_CRYPTO_MD5_LIBC */ + +/* MD5 via ARCHIVE_CRYPTO_MD5_LIBMD supported. */ +/* #undef ARCHIVE_CRYPTO_MD5_LIBMD */ + +/* MD5 via ARCHIVE_CRYPTO_MD5_LIBSYSTEM supported. */ +#define ARCHIVE_CRYPTO_MD5_LIBSYSTEM 1 + +/* MD5 via ARCHIVE_CRYPTO_MD5_MBEDTLS supported. */ +/* #undef ARCHIVE_CRYPTO_MD5_MBEDTLS */ + +/* MD5 via ARCHIVE_CRYPTO_MD5_NETTLE supported. */ +/* #undef ARCHIVE_CRYPTO_MD5_NETTLE */ + +/* MD5 via ARCHIVE_CRYPTO_MD5_OPENSSL supported. */ +/* #undef ARCHIVE_CRYPTO_MD5_OPENSSL */ + +/* MD5 via ARCHIVE_CRYPTO_MD5_WIN supported. */ +/* #undef ARCHIVE_CRYPTO_MD5_WIN */ + +/* RMD160 via ARCHIVE_CRYPTO_RMD160_LIBC supported. */ +/* #undef ARCHIVE_CRYPTO_RMD160_LIBC */ + +/* RMD160 via ARCHIVE_CRYPTO_RMD160_LIBMD supported. */ +/* #undef ARCHIVE_CRYPTO_RMD160_LIBMD */ + +/* RMD160 via ARCHIVE_CRYPTO_RMD160_MBEDTLS supported. */ +/* #undef ARCHIVE_CRYPTO_RMD160_MBEDTLS */ + +/* RMD160 via ARCHIVE_CRYPTO_RMD160_NETTLE supported. */ +/* #undef ARCHIVE_CRYPTO_RMD160_NETTLE */ + +/* RMD160 via ARCHIVE_CRYPTO_RMD160_OPENSSL supported. */ +/* #undef ARCHIVE_CRYPTO_RMD160_OPENSSL */ + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_LIBC supported. */ +/* #undef ARCHIVE_CRYPTO_SHA1_LIBC */ + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_LIBMD supported. */ +/* #undef ARCHIVE_CRYPTO_SHA1_LIBMD */ + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_LIBSYSTEM supported. */ +#define ARCHIVE_CRYPTO_SHA1_LIBSYSTEM 1 + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_MBEDTLS supported. */ +/* #undef ARCHIVE_CRYPTO_SHA1_MBEDTLS */ + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_NETTLE supported. */ +/* #undef ARCHIVE_CRYPTO_SHA1_NETTLE */ + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_OPENSSL supported. */ +/* #undef ARCHIVE_CRYPTO_SHA1_OPENSSL */ + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_WIN supported. */ +/* #undef ARCHIVE_CRYPTO_SHA1_WIN */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_LIBC supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_LIBC */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_LIBC2 supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_LIBC2 */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_LIBC3 supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_LIBC3 */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_LIBMD supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_LIBMD */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_LIBSYSTEM supported. */ +#define ARCHIVE_CRYPTO_SHA256_LIBSYSTEM 1 + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_MBEDTLS supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_MBEDTLS */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_NETTLE supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_NETTLE */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_OPENSSL supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_OPENSSL */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_WIN supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_WIN */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_LIBC supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_LIBC */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_LIBC2 supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_LIBC2 */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_LIBC3 supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_LIBC3 */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_LIBSYSTEM supported. */ +#define ARCHIVE_CRYPTO_SHA384_LIBSYSTEM 1 + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_MBEDTLS supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_MBEDTLS */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_NETTLE supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_NETTLE */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_OPENSSL supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_OPENSSL */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_WIN supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_WIN */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_LIBC supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_LIBC */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_LIBC2 supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_LIBC2 */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_LIBC3 supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_LIBC3 */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_LIBMD supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_LIBMD */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_LIBSYSTEM supported. */ +#define ARCHIVE_CRYPTO_SHA512_LIBSYSTEM 1 + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_MBEDTLS supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_MBEDTLS */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_NETTLE supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_NETTLE */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_OPENSSL supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_OPENSSL */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_WIN supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_WIN */ + +/* AIX xattr support */ +/* #undef ARCHIVE_XATTR_AIX */ + +/* Darwin xattr support */ +/* #undef ARCHIVE_XATTR_DARWIN */ + +/* FreeBSD xattr support */ +/* #undef ARCHIVE_XATTR_FREEBSD */ + +/* Linux xattr support */ +/* #undef ARCHIVE_XATTR_LINUX */ + +/* Version number of bsdcat */ +#define BSDCAT_VERSION_STRING "3.4.3" + +/* Version number of bsdcpio */ +#define BSDCPIO_VERSION_STRING "3.4.3" + +/* Version number of bsdtar */ +#define BSDTAR_VERSION_STRING "3.4.3" + +/* Define to 1 if the system has the type `ace_t'. */ +/* #undef HAVE_ACE_T */ + +/* Define to 1 if you have the `acl' function. */ +/* #undef HAVE_ACL */ + +/* Define to 1 if the system has the type `aclent_t'. */ +/* #undef HAVE_ACLENT_T */ + +/* Define to 1 if you have the `acl_add_flag_np' function. */ +/* #undef HAVE_ACL_ADD_FLAG_NP */ + +/* Define to 1 if you have the `acl_add_perm' function. */ +/* #undef HAVE_ACL_ADD_PERM */ + +/* Define to 1 if you have the `acl_clear_flags_np' function. */ +/* #undef HAVE_ACL_CLEAR_FLAGS_NP */ + +/* Define to 1 if you have the `acl_clear_perms' function. */ +/* #undef HAVE_ACL_CLEAR_PERMS */ + +/* Define to 1 if you have the `acl_create_entry' function. */ +/* #undef HAVE_ACL_CREATE_ENTRY */ + +/* Define to 1 if you have the `acl_delete_def_file' function. */ +/* #undef HAVE_ACL_DELETE_DEF_FILE */ + +/* Define to 1 if the system has the type `acl_entry_t'. */ +/* #undef HAVE_ACL_ENTRY_T */ + +/* Define to 1 if you have the `acl_free' function. */ +/* #undef HAVE_ACL_FREE */ + +/* Define to 1 if you have the `acl_get_brand_np' function. */ +/* #undef HAVE_ACL_GET_BRAND_NP */ + +/* Define to 1 if you have the `acl_get_entry' function. */ +/* #undef HAVE_ACL_GET_ENTRY */ + +/* Define to 1 if you have the `acl_get_entry_type_np' function. */ +/* #undef HAVE_ACL_GET_ENTRY_TYPE_NP */ + +/* Define to 1 if you have the `acl_get_fd' function. */ +/* #undef HAVE_ACL_GET_FD */ + +/* Define to 1 if you have the `acl_get_fd_np' function. */ +/* #undef HAVE_ACL_GET_FD_NP */ + +/* Define to 1 if you have the `acl_get_file' function. */ +/* #undef HAVE_ACL_GET_FILE */ + +/* Define to 1 if you have the `acl_get_flagset_np' function. */ +/* #undef HAVE_ACL_GET_FLAGSET_NP */ + +/* Define to 1 if you have the `acl_get_flag_np' function. */ +/* #undef HAVE_ACL_GET_FLAG_NP */ + +/* Define to 1 if you have the `acl_get_link_np' function. */ +/* #undef HAVE_ACL_GET_LINK_NP */ + +/* Define to 1 if you have the `acl_get_perm' function. */ +/* #undef HAVE_ACL_GET_PERM */ + +/* Define to 1 if you have the `acl_get_permset' function. */ +/* #undef HAVE_ACL_GET_PERMSET */ + +/* Define to 1 if you have the `acl_get_perm_np' function. */ +/* #undef HAVE_ACL_GET_PERM_NP */ + +/* Define to 1 if you have the `acl_get_qualifier' function. */ +/* #undef HAVE_ACL_GET_QUALIFIER */ + +/* Define to 1 if you have the `acl_get_tag_type' function. */ +/* #undef HAVE_ACL_GET_TAG_TYPE */ + +/* Define to 1 if you have the `acl_init' function. */ +/* #undef HAVE_ACL_INIT */ + +/* Define to 1 if you have the `acl_is_trivial_np' function. */ +/* #undef HAVE_ACL_IS_TRIVIAL_NP */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_ACL_LIBACL_H */ + +/* Define to 1 if the system has the type `acl_permset_t'. */ +/* #undef HAVE_ACL_PERMSET_T */ + +/* Define to 1 if you have the `acl_set_entry_type_np' function. */ +/* #undef HAVE_ACL_SET_ENTRY_TYPE_NP */ + +/* Define to 1 if you have the `acl_set_fd' function. */ +/* #undef HAVE_ACL_SET_FD */ + +/* Define to 1 if you have the `acl_set_fd_np' function. */ +/* #undef HAVE_ACL_SET_FD_NP */ + +/* Define to 1 if you have the `acl_set_file' function. */ +/* #undef HAVE_ACL_SET_FILE */ + +/* Define to 1 if you have the `acl_set_link_np' function. */ +/* #undef HAVE_ACL_SET_LINK_NP */ + +/* Define to 1 if you have the `acl_set_qualifier' function. */ +/* #undef HAVE_ACL_SET_QUALIFIER */ + +/* Define to 1 if you have the `acl_set_tag_type' function. */ +/* #undef HAVE_ACL_SET_TAG_TYPE */ + +/* Define to 1 if the system has the type `acl_t'. */ +/* #undef HAVE_ACL_T */ + +/* Define to 1 if the system has the type `acl_tag_t'. */ +/* #undef HAVE_ACL_TAG_T */ + +/* Define to 1 if you have the `arc4random_buf' function. */ +#define HAVE_ARC4RANDOM_BUF 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_ATTR_XATTR_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_BCRYPT_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_BLAKE2_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_BZLIB_H */ + +/* Define to 1 if you have the `chflags' function. */ +#define HAVE_CHFLAGS 1 + +/* Define to 1 if you have the `chown' function. */ +#define HAVE_CHOWN 1 + +/* Define to 1 if you have the `chroot' function. */ +#define HAVE_CHROOT 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_COPYFILE_H 1 + +/* Define to 1 if you have the `ctime_r' function. */ +#define HAVE_CTIME_R 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_CTYPE_H 1 + +/* Define to 1 if you have the `cygwin_conv_path' function. */ +/* #undef HAVE_CYGWIN_CONV_PATH */ + +/* Define to 1 if you have the declaration of `ACE_GETACL', and to 0 if you + don't. */ +/* #undef HAVE_DECL_ACE_GETACL */ + +/* Define to 1 if you have the declaration of `ACE_GETACLCNT', and to 0 if you + don't. */ +/* #undef HAVE_DECL_ACE_GETACLCNT */ + +/* Define to 1 if you have the declaration of `ACE_SETACL', and to 0 if you + don't. */ +/* #undef HAVE_DECL_ACE_SETACL */ + +/* Define to 1 if you have the declaration of `ACL_SYNCHRONIZE', and to 0 if + you don't. */ +/* #undef HAVE_DECL_ACL_SYNCHRONIZE */ + +/* Define to 1 if you have the declaration of `ACL_TYPE_EXTENDED', and to 0 if + you don't. */ +/* #undef HAVE_DECL_ACL_TYPE_EXTENDED */ + +/* Define to 1 if you have the declaration of `ACL_TYPE_NFS4', and to 0 if you + don't. */ +/* #undef HAVE_DECL_ACL_TYPE_NFS4 */ + +/* Define to 1 if you have the declaration of `ACL_USER', and to 0 if you + don't. */ +/* #undef HAVE_DECL_ACL_USER */ + +/* Define to 1 if you have the declaration of `EXTATTR_NAMESPACE_USER', and to + 0 if you don't. */ +/* #undef HAVE_DECL_EXTATTR_NAMESPACE_USER */ + +/* Define to 1 if you have the declaration of `GETACL', and to 0 if you don't. + */ +/* #undef HAVE_DECL_GETACL */ + +/* Define to 1 if you have the declaration of `GETACLCNT', and to 0 if you + don't. */ +/* #undef HAVE_DECL_GETACLCNT */ + +/* Define to 1 if you have the declaration of `INT32_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_INT32_MAX 1 + +/* Define to 1 if you have the declaration of `INT32_MIN', and to 0 if you + don't. */ +#define HAVE_DECL_INT32_MIN 1 + +/* Define to 1 if you have the declaration of `INT64_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_INT64_MAX 1 + +/* Define to 1 if you have the declaration of `INT64_MIN', and to 0 if you + don't. */ +#define HAVE_DECL_INT64_MIN 1 + +/* Define to 1 if you have the declaration of `INTMAX_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_INTMAX_MAX 1 + +/* Define to 1 if you have the declaration of `INTMAX_MIN', and to 0 if you + don't. */ +#define HAVE_DECL_INTMAX_MIN 1 + +/* Define to 1 if you have the declaration of `SETACL', and to 0 if you don't. + */ +/* #undef HAVE_DECL_SETACL */ + +/* Define to 1 if you have the declaration of `SIZE_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_SIZE_MAX 1 + +/* Define to 1 if you have the declaration of `SSIZE_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_SSIZE_MAX 1 + +/* Define to 1 if you have the declaration of `strerror_r', and to 0 if you + don't. */ +#define HAVE_DECL_STRERROR_R 1 + +/* Define to 1 if you have the declaration of `UINT32_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_UINT32_MAX 1 + +/* Define to 1 if you have the declaration of `UINT64_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_UINT64_MAX 1 + +/* Define to 1 if you have the declaration of `UINTMAX_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_UINTMAX_MAX 1 + +/* Define to 1 if you have the declaration of `XATTR_NOFOLLOW', and to 0 if + you don't. */ +/* #undef HAVE_DECL_XATTR_NOFOLLOW */ + +/* Define to 1 if you have the header file, and it defines `DIR'. + */ +#define HAVE_DIRENT_H 1 + +/* Define to 1 if you have a dirfd function or macro */ +#define HAVE_DIRFD 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_DLFCN_H 1 + +/* Define to 1 if you don't have `vprintf' but do have `_doprnt.' */ +/* #undef HAVE_DOPRNT */ + +/* Define to 1 if nl_langinfo supports D_MD_ORDER */ +#define HAVE_D_MD_ORDER 1 + +/* A possible errno value for invalid file format errors */ +#define HAVE_EFTYPE 1 + +/* A possible errno value for invalid file format errors */ +#define HAVE_EILSEQ 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_ERRNO_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_EXPAT_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_EXT2FS_EXT2_FS_H */ + +/* Define to 1 if you have the `extattr_get_fd' function. */ +/* #undef HAVE_EXTATTR_GET_FD */ + +/* Define to 1 if you have the `extattr_get_file' function. */ +/* #undef HAVE_EXTATTR_GET_FILE */ + +/* Define to 1 if you have the `extattr_get_link' function. */ +/* #undef HAVE_EXTATTR_GET_LINK */ + +/* Define to 1 if you have the `extattr_list_fd' function. */ +/* #undef HAVE_EXTATTR_LIST_FD */ + +/* Define to 1 if you have the `extattr_list_file' function. */ +/* #undef HAVE_EXTATTR_LIST_FILE */ + +/* Define to 1 if you have the `extattr_list_link' function. */ +/* #undef HAVE_EXTATTR_LIST_LINK */ + +/* Define to 1 if you have the `extattr_set_fd' function. */ +/* #undef HAVE_EXTATTR_SET_FD */ + +/* Define to 1 if you have the `extattr_set_link' function. */ +/* #undef HAVE_EXTATTR_SET_LINK */ + +/* Define to 1 if you have the `facl' function. */ +/* #undef HAVE_FACL */ + +/* Define to 1 if you have the `fchdir' function. */ +#define HAVE_FCHDIR 1 + +/* Define to 1 if you have the `fchflags' function. */ +#define HAVE_FCHFLAGS 1 + +/* Define to 1 if you have the `fchmod' function. */ +#define HAVE_FCHMOD 1 + +/* Define to 1 if you have the `fchown' function. */ +#define HAVE_FCHOWN 1 + +/* Define to 1 if you have the `fcntl' function. */ +#define HAVE_FCNTL 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_FCNTL_H 1 + +/* Define to 1 if you have the `fdopendir' function. */ +#define HAVE_FDOPENDIR 1 + +/* Define to 1 if you have the `fgetea' function. */ +/* #undef HAVE_FGETEA */ + +/* Define to 1 if you have the `fgetxattr' function. */ +/* #undef HAVE_FGETXATTR */ + +/* Define to 1 if you have the `flistea' function. */ +/* #undef HAVE_FLISTEA */ + +/* Define to 1 if you have the `flistxattr' function. */ +/* #undef HAVE_FLISTXATTR */ + +/* Define to 1 if you have the `fork' function. */ +#define HAVE_FORK 1 + +/* Define to 1 if fseeko (and presumably ftello) exists and is declared. */ +#define HAVE_FSEEKO 1 + +/* Define to 1 if you have the `fsetea' function. */ +/* #undef HAVE_FSETEA */ + +/* Define to 1 if you have the `fsetxattr' function. */ +/* #undef HAVE_FSETXATTR */ + +/* Define to 1 if you have the `fstat' function. */ +#define HAVE_FSTAT 1 + +/* Define to 1 if you have the `fstatat' function. */ +#define HAVE_FSTATAT 1 + +/* Define to 1 if you have the `fstatfs' function. */ +#define HAVE_FSTATFS 1 + +/* Define to 1 if you have the `fstatvfs' function. */ +#define HAVE_FSTATVFS 1 + +/* Define to 1 if you have the `ftruncate' function. */ +#define HAVE_FTRUNCATE 1 + +/* Define to 1 if you have the `futimens' function. */ +#define HAVE_FUTIMENS 1 + +/* Define to 1 if you have the `futimes' function. */ +#define HAVE_FUTIMES 1 + +/* Define to 1 if you have the `futimesat' function. */ +/* #undef HAVE_FUTIMESAT */ + +/* Define to 1 if you have the `getea' function. */ +/* #undef HAVE_GETEA */ + +/* Define to 1 if you have the `geteuid' function. */ +#define HAVE_GETEUID 1 + +/* Define to 1 if you have the `getgrgid_r' function. */ +#define HAVE_GETGRGID_R 1 + +/* Define to 1 if you have the `getgrnam_r' function. */ +#define HAVE_GETGRNAM_R 1 + +/* Define to 1 if you have the `getpid' function. */ +#define HAVE_GETPID 1 + +/* Define to 1 if you have the `getpwnam_r' function. */ +#define HAVE_GETPWNAM_R 1 + +/* Define to 1 if you have the `getpwuid_r' function. */ +#define HAVE_GETPWUID_R 1 + +/* Define to 1 if you have the `getvfsbyname' function. */ +#define HAVE_GETVFSBYNAME 1 + +/* Define to 1 if you have the `getxattr' function. */ +/* #undef HAVE_GETXATTR */ + +/* Define to 1 if you have the `gmtime_r' function. */ +#define HAVE_GMTIME_R 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_GRP_H 1 + +/* Define if you have the iconv() function and it works. */ +/* #undef HAVE_ICONV */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_ICONV_H */ + +/* Define to 1 if the system has the type `intmax_t'. */ +#define HAVE_INTMAX_T 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_INTTYPES_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_IO_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_LANGINFO_H 1 + +/* Define to 1 if you have the `lchflags' function. */ +#define HAVE_LCHFLAGS 1 + +/* Define to 1 if you have the `lchmod' function. */ +#define HAVE_LCHMOD 1 + +/* Define to 1 if you have the `lchown' function. */ +#define HAVE_LCHOWN 1 + +/* Define to 1 if you have the `lgetea' function. */ +/* #undef HAVE_LGETEA */ + +/* Define to 1 if you have the `lgetxattr' function. */ +/* #undef HAVE_LGETXATTR */ + +/* Define to 1 if you have the `acl' library (-lacl). */ +/* #undef HAVE_LIBACL */ + +/* Define to 1 if you have the `b2' library (-lb2). */ +/* #undef HAVE_LIBB2 */ + +/* Define to 1 if you have the `bz2' library (-lbz2). */ +/* #undef HAVE_LIBBZ2 */ + +/* Define to 1 if you have the `charset' library (-lcharset). */ +/* #undef HAVE_LIBCHARSET */ + +/* Define to 1 if you have the `crypto' library (-lcrypto). */ +/* #undef HAVE_LIBCRYPTO */ + +/* Define to 1 if you have the `eay32' library (-leay32). */ +/* #undef HAVE_LIBEAY32 */ + +/* Define to 1 if you have the `eay64' library (-leay64). */ +/* #undef HAVE_LIBEAY64 */ + +/* Define to 1 if you have the `expat' library (-lexpat). */ +/* #undef HAVE_LIBEXPAT */ + +/* Define to 1 if you have the `lz4' library (-llz4). */ +/* #undef HAVE_LIBLZ4 */ + +/* Define to 1 if you have the `lzma' library (-llzma). */ +/* #undef HAVE_LIBLZMA */ + +/* Define to 1 if you have the `lzo2' library (-llzo2). */ +/* #undef HAVE_LIBLZO2 */ + +/* Define to 1 if you have the `mbedcrypto' library (-lmbedcrypto). */ +/* #undef HAVE_LIBMBEDCRYPTO */ + +/* Define to 1 if you have the `md' library (-lmd). */ +/* #undef HAVE_LIBMD */ + +/* Define to 1 if you have the `nettle' library (-lnettle). */ +/* #undef HAVE_LIBNETTLE */ + +/* Define to 1 if you have the `pcre' library (-lpcre). */ +/* #undef HAVE_LIBPCRE */ + +/* Define to 1 if you have the `pcreposix' library (-lpcreposix). */ +/* #undef HAVE_LIBPCREPOSIX */ + +/* Define to 1 if you have the `regex' library (-lregex). */ +/* #undef HAVE_LIBREGEX */ + +/* Define to 1 if you have the `richacl' library (-lrichacl). */ +/* #undef HAVE_LIBRICHACL */ + +/* Define to 1 if you have the `xml2' library (-lxml2). */ +/* #undef HAVE_LIBXML2 */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LIBXML_XMLREADER_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LIBXML_XMLWRITER_H */ + +/* Define to 1 if you have the `z' library (-lz). */ +/* #undef HAVE_LIBZ */ + +/* Define to 1 if you have the `zstd' library (-lzstd). */ +/* #undef HAVE_LIBZSTD */ + +/* Define to 1 if you have the header file. */ +#define HAVE_LIMITS_H 1 + +/* Define to 1 if you have the `link' function. */ +#define HAVE_LINK 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LINUX_FIEMAP_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LINUX_FS_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LINUX_MAGIC_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LINUX_TYPES_H */ + +/* Define to 1 if you have the `listea' function. */ +/* #undef HAVE_LISTEA */ + +/* Define to 1 if you have the `listxattr' function. */ +/* #undef HAVE_LISTXATTR */ + +/* Define to 1 if you have the `llistea' function. */ +/* #undef HAVE_LLISTEA */ + +/* Define to 1 if you have the `llistxattr' function. */ +/* #undef HAVE_LLISTXATTR */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LOCALCHARSET_H */ + +/* Define to 1 if you have the `locale_charset' function. */ +/* #undef HAVE_LOCALE_CHARSET */ + +/* Define to 1 if you have the header file. */ +#define HAVE_LOCALE_H 1 + +/* Define to 1 if you have the `localtime_r' function. */ +#define HAVE_LOCALTIME_R 1 + +/* Define to 1 if the system has the type `long long int'. */ +#define HAVE_LONG_LONG_INT 1 + +/* Define to 1 if you have the `lsetea' function. */ +/* #undef HAVE_LSETEA */ + +/* Define to 1 if you have the `lsetxattr' function. */ +/* #undef HAVE_LSETXATTR */ + +/* Define to 1 if you have the `lstat' function. */ +#define HAVE_LSTAT 1 + +/* Define to 1 if `lstat' has the bug that it succeeds when given the + zero-length file name argument. */ +/* #undef HAVE_LSTAT_EMPTY_STRING_BUG */ + +/* Define to 1 if you have the `lutimes' function. */ +#define HAVE_LUTIMES 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LZ4HC_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LZ4_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LZMA_H */ + +/* Define to 1 if you have the `lzma_stream_encoder_mt' function. */ +/* #undef HAVE_LZMA_STREAM_ENCODER_MT */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LZO_LZO1X_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LZO_LZOCONF_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_MBEDTLS_AES_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_MBEDTLS_MD_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_MBEDTLS_PKCS5_H */ + +/* Define to 1 if you have the `mbrtowc' function. */ +#define HAVE_MBRTOWC 1 + +/* Define to 1 if you have the `mbr_gid_to_uuid' function. */ +/* #undef HAVE_MBR_GID_TO_UUID */ + +/* Define to 1 if you have the `mbr_uid_to_uuid' function. */ +/* #undef HAVE_MBR_UID_TO_UUID */ + +/* Define to 1 if you have the `mbr_uuid_to_id' function. */ +/* #undef HAVE_MBR_UUID_TO_ID */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_MD5_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_MEMBERSHIP_H 1 + +/* Define to 1 if you have the `memmove' function. */ +#define HAVE_MEMMOVE 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_MEMORY_H 1 + +/* Define to 1 if you have the `memset' function. */ +#define HAVE_MEMSET 1 + +/* Define to 1 if you have the `mkdir' function. */ +#define HAVE_MKDIR 1 + +/* Define to 1 if you have the `mkfifo' function. */ +#define HAVE_MKFIFO 1 + +/* Define to 1 if you have the `mknod' function. */ +#define HAVE_MKNOD 1 + +/* Define to 1 if you have the `mkstemp' function. */ +#define HAVE_MKSTEMP 1 + +/* Define to 1 if you have the header file, and it defines `DIR'. */ +/* #undef HAVE_NDIR_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NETTLE_AES_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NETTLE_HMAC_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NETTLE_MD5_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NETTLE_PBKDF2_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NETTLE_RIPEMD160_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NETTLE_SHA_H */ + +/* Define to 1 if you have the `nl_langinfo' function. */ +#define HAVE_NL_LANGINFO 1 + +/* Define to 1 if you have the `openat' function. */ +#define HAVE_OPENAT 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_OPENSSL_EVP_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_PATHS_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_PCREPOSIX_H */ + +/* Define to 1 if you have the `pipe' function. */ +#define HAVE_PIPE 1 + +/* Define to 1 if you have the `PKCS5_PBKDF2_HMAC_SHA1' function. */ +/* #undef HAVE_PKCS5_PBKDF2_HMAC_SHA1 */ + +/* Define to 1 if you have the `poll' function. */ +#define HAVE_POLL 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_POLL_H 1 + +/* Define to 1 if you have the `posix_spawnp' function. */ +#define HAVE_POSIX_SPAWNP 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_PTHREAD_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_PWD_H 1 + +/* Define to 1 if you have a POSIX compatible readdir_r */ +#define HAVE_READDIR_R 1 + +/* Define to 1 if you have the `readlink' function. */ +#define HAVE_READLINK 1 + +/* Define to 1 if you have the `readlinkat' function. */ +#define HAVE_READLINKAT 1 + +/* Define to 1 if you have the `readpassphrase' function. */ +#define HAVE_READPASSPHRASE 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_READPASSPHRASE_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_REGEX_H */ + +/* Define to 1 if you have the `richacl_alloc' function. */ +/* #undef HAVE_RICHACL_ALLOC */ + +/* Define to 1 if you have the `richacl_equiv_mode' function. */ +/* #undef HAVE_RICHACL_EQUIV_MODE */ + +/* Define to 1 if you have the `richacl_free' function. */ +/* #undef HAVE_RICHACL_FREE */ + +/* Define to 1 if you have the `richacl_get_fd' function. */ +/* #undef HAVE_RICHACL_GET_FD */ + +/* Define to 1 if you have the `richacl_get_file' function. */ +/* #undef HAVE_RICHACL_GET_FILE */ + +/* Define to 1 if you have the `richacl_set_fd' function. */ +/* #undef HAVE_RICHACL_SET_FD */ + +/* Define to 1 if you have the `richacl_set_file' function. */ +/* #undef HAVE_RICHACL_SET_FILE */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_RIPEMD_H */ + +/* Define to 1 if you have the `select' function. */ +#define HAVE_SELECT 1 + +/* Define to 1 if you have the `setenv' function. */ +#define HAVE_SETENV 1 + +/* Define to 1 if you have the `setlocale' function. */ +#define HAVE_SETLOCALE 1 + +/* Define to 1 if you have the `setxattr' function. */ +/* #undef HAVE_SETXATTR */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SHA256_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SHA512_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SHA_H */ + +/* Define to 1 if you have the `sigaction' function. */ +#define HAVE_SIGACTION 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SIGNAL_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SPAWN_H 1 + +/* Define to 1 if you have the `statfs' function. */ +#define HAVE_STATFS 1 + +/* Define to 1 if you have the `statvfs' function. */ +#define HAVE_STATVFS 1 + +/* Define to 1 if `stat' has the bug that it succeeds when given the + zero-length file name argument. */ +/* #undef HAVE_STAT_EMPTY_STRING_BUG */ + +/* Define to 1 if you have the header file. */ +#define HAVE_STDARG_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_STDINT_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_STDLIB_H 1 + +/* Define to 1 if you have the `strchr' function. */ +#define HAVE_STRCHR 1 + +/* Define to 1 if you have the `strdup' function. */ +#define HAVE_STRDUP 1 + +/* Define to 1 if you have the `strerror' function. */ +#define HAVE_STRERROR 1 + +/* Define to 1 if you have the `strerror_r' function. */ +#define HAVE_STRERROR_R 1 + +/* Define to 1 if you have the `strftime' function. */ +#define HAVE_STRFTIME 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_STRINGS_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_STRING_H 1 + +/* Define to 1 if you have the `strncpy_s' function. */ +/* #undef HAVE_STRNCPY_S */ + +/* Define to 1 if you have the `strrchr' function. */ +#define HAVE_STRRCHR 1 + +/* Define to 1 if the system has the type `struct richace'. */ +/* #undef HAVE_STRUCT_RICHACE */ + +/* Define to 1 if the system has the type `struct richacl'. */ +/* #undef HAVE_STRUCT_RICHACL */ + +/* Define to 1 if `f_namemax' is a member of `struct statfs'. */ +/* #undef HAVE_STRUCT_STATFS_F_NAMEMAX */ + +/* Define to 1 if `f_iosize' is a member of `struct statvfs'. */ +/* #undef HAVE_STRUCT_STATVFS_F_IOSIZE */ + +/* Define to 1 if `st_birthtime' is a member of `struct stat'. */ +#define HAVE_STRUCT_STAT_ST_BIRTHTIME 1 + +/* Define to 1 if `st_birthtimespec.tv_nsec' is a member of `struct stat'. */ +#define HAVE_STRUCT_STAT_ST_BIRTHTIMESPEC_TV_NSEC 1 + +/* Define to 1 if `st_blksize' is a member of `struct stat'. */ +#define HAVE_STRUCT_STAT_ST_BLKSIZE 1 + +/* Define to 1 if `st_flags' is a member of `struct stat'. */ +#define HAVE_STRUCT_STAT_ST_FLAGS 1 + +/* Define to 1 if `st_mtimespec.tv_nsec' is a member of `struct stat'. */ +#define HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC 1 + +/* Define to 1 if `st_mtime_n' is a member of `struct stat'. */ +/* #undef HAVE_STRUCT_STAT_ST_MTIME_N */ + +/* Define to 1 if `st_mtime_usec' is a member of `struct stat'. */ +/* #undef HAVE_STRUCT_STAT_ST_MTIME_USEC */ + +/* Define to 1 if `st_mtim.tv_nsec' is a member of `struct stat'. */ +/* #undef HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC */ + +/* Define to 1 if `st_umtime' is a member of `struct stat'. */ +/* #undef HAVE_STRUCT_STAT_ST_UMTIME */ + +/* Define to 1 if `tm_gmtoff' is a member of `struct tm'. */ +#define HAVE_STRUCT_TM_TM_GMTOFF 1 + +/* Define to 1 if `__tm_gmtoff' is a member of `struct tm'. */ +/* #undef HAVE_STRUCT_TM___TM_GMTOFF */ + +/* Define to 1 if the system has the type `struct vfsconf'. */ +#define HAVE_STRUCT_VFSCONF 1 + +/* Define to 1 if the system has the type `struct xvfsconf'. */ +/* #undef HAVE_STRUCT_XVFSCONF */ + +/* Define to 1 if you have the `symlink' function. */ +#define HAVE_SYMLINK 1 + +/* Define to 1 if you have the header file. */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_CDEFS_H 1 + +/* Define to 1 if you have the header file, and it defines `DIR'. + */ +/* #undef HAVE_SYS_DIR_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_EA_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_EXTATTR_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_IOCTL_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_MKDEV_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_MOUNT_H 1 + +/* Define to 1 if you have the header file, and it defines `DIR'. + */ +/* #undef HAVE_SYS_NDIR_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_PARAM_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_POLL_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_RICHACL_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_SELECT_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_STATFS_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_STATVFS_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_STAT_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_SYSMACROS_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_TIME_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_TYPES_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_UTIME_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_UTSNAME_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_VFS_H */ + +/* Define to 1 if you have that is POSIX.1 compatible. */ +#define HAVE_SYS_WAIT_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_XATTR_H 1 + +/* Define to 1 if you have the `timegm' function. */ +#define HAVE_TIMEGM 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_TIME_H 1 + +/* Define to 1 if you have the `tzset' function. */ +#define HAVE_TZSET 1 + +/* Define to 1 if the system has the type `uintmax_t'. */ +#define HAVE_UINTMAX_T 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_UNISTD_H 1 + +/* Define to 1 if you have the `unlinkat' function. */ +#define HAVE_UNLINKAT 1 + +/* Define to 1 if you have the `unsetenv' function. */ +#define HAVE_UNSETENV 1 + +/* Define to 1 if the system has the type `unsigned long long'. */ +#define HAVE_UNSIGNED_LONG_LONG 1 + +/* Define to 1 if the system has the type `unsigned long long int'. */ +#define HAVE_UNSIGNED_LONG_LONG_INT 1 + +/* Define to 1 if you have the `utime' function. */ +#define HAVE_UTIME 1 + +/* Define to 1 if you have the `utimensat' function. */ +#define HAVE_UTIMENSAT 1 + +/* Define to 1 if you have the `utimes' function. */ +#define HAVE_UTIMES 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_UTIME_H 1 + +/* Define to 1 if you have the `vfork' function. */ +#define HAVE_VFORK 1 + +/* Define to 1 if you have the `vprintf' function. */ +#define HAVE_VPRINTF 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_WCHAR_H 1 + +/* Define to 1 if the system has the type `wchar_t'. */ +#define HAVE_WCHAR_T 1 + +/* Define to 1 if you have the `wcrtomb' function. */ +#define HAVE_WCRTOMB 1 + +/* Define to 1 if you have the `wcscmp' function. */ +#define HAVE_WCSCMP 1 + +/* Define to 1 if you have the `wcscpy' function. */ +#define HAVE_WCSCPY 1 + +/* Define to 1 if you have the `wcslen' function. */ +#define HAVE_WCSLEN 1 + +/* Define to 1 if you have the `wctomb' function. */ +#define HAVE_WCTOMB 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_WCTYPE_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_WINCRYPT_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_WINDOWS_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_WINIOCTL_H */ + +/* Define to 1 if you have the `wmemcmp' function. */ +#define HAVE_WMEMCMP 1 + +/* Define to 1 if you have the `wmemcpy' function. */ +#define HAVE_WMEMCPY 1 + +/* Define to 1 if you have the `wmemmove' function. */ +#define HAVE_WMEMMOVE 1 + +/* Define to 1 if you have a working EXT2_IOC_GETFLAGS */ +/* #undef HAVE_WORKING_EXT2_IOC_GETFLAGS */ + +/* Define to 1 if you have a working FS_IOC_GETFLAGS */ +/* #undef HAVE_WORKING_FS_IOC_GETFLAGS */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_ZLIB_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_ZSTD_H */ + +/* Define to 1 if you have the `_ctime64_s' function. */ +/* #undef HAVE__CTIME64_S */ + +/* Define to 1 if you have the `_fseeki64' function. */ +/* #undef HAVE__FSEEKI64 */ + +/* Define to 1 if you have the `_get_timezone' function. */ +/* #undef HAVE__GET_TIMEZONE */ + +/* Define to 1 if you have the `_gmtime64_s' function. */ +/* #undef HAVE__GMTIME64_S */ + +/* Define to 1 if you have the `_localtime64_s' function. */ +/* #undef HAVE__LOCALTIME64_S */ + +/* Define to 1 if you have the `_mkgmtime64' function. */ +/* #undef HAVE__MKGMTIME64 */ + +/* Define as const if the declaration of iconv() needs const. */ +/* #undef ICONV_CONST */ + +/* Version number of libarchive as a single integer */ +#define LIBARCHIVE_VERSION_NUMBER "3004003" + +/* Version number of libarchive */ +#define LIBARCHIVE_VERSION_STRING "3.4.3" + +/* Define to 1 if `lstat' dereferences a symlink specified with a trailing + slash. */ +/* #undef LSTAT_FOLLOWS_SLASHED_SYMLINK */ + +/* Define to the sub-directory where libtool stores uninstalled libraries. */ +#define LT_OBJDIR ".libs/" + +/* Define to 1 if `major', `minor', and `makedev' are declared in . + */ +/* #undef MAJOR_IN_MKDEV */ + +/* Define to 1 if `major', `minor', and `makedev' are declared in + . */ +/* #undef MAJOR_IN_SYSMACROS */ + +/* Define to '0x05020000' for Windows Server 2003 APIs. */ +/* #undef NTDDI_VERSION */ + +/* Name of package */ +#define PACKAGE "libarchive" + +/* Define to the address where bug reports for this package should be sent. */ +#define PACKAGE_BUGREPORT "libarchive-discuss@googlegroups.com" + +/* Define to the full name of this package. */ +#define PACKAGE_NAME "libarchive" + +/* Define to the full name and version of this package. */ +#define PACKAGE_STRING "libarchive 3.4.3" + +/* Define to the one symbol short name of this package. */ +#define PACKAGE_TARNAME "libarchive" + +/* Define to the home page for this package. */ +#define PACKAGE_URL "" + +/* Define to the version of this package. */ +#define PACKAGE_VERSION "3.4.3" + +/* Define to 1 if PCRE_STATIC needs to be defined. */ +/* #undef PCRE_STATIC */ + +/* The size of `wchar_t', as computed by sizeof. */ +#define SIZEOF_WCHAR_T 4 + +/* Define to 1 if you have the ANSI C header files. */ +#define STDC_HEADERS 1 + +/* Define to 1 if strerror_r returns char *. */ +/* #undef STRERROR_R_CHAR_P */ + +/* Define to 1 if you can safely include both and . */ +#define TIME_WITH_SYS_TIME 1 + +/* Enable extensions on AIX 3, Interix. */ +#ifndef _ALL_SOURCE +# define _ALL_SOURCE 1 +#endif +/* Enable GNU extensions on systems that have them. */ +#ifndef _GNU_SOURCE +# define _GNU_SOURCE 1 +#endif +/* Enable threading extensions on Solaris. */ +#ifndef _POSIX_PTHREAD_SEMANTICS +# define _POSIX_PTHREAD_SEMANTICS 1 +#endif +/* Enable extensions on HP NonStop. */ +#ifndef _TANDEM_SOURCE +# define _TANDEM_SOURCE 1 +#endif +/* Enable general extensions on Solaris. */ +#ifndef __EXTENSIONS__ +# define __EXTENSIONS__ 1 +#endif + + +/* Version number of package */ +#define VERSION "3.4.3" + +/* Define to '0x0502' for Windows Server 2003 APIs. */ +/* #undef WINVER */ + +/* Enable large inode numbers on Mac OS X 10.5. */ +#ifndef _DARWIN_USE_64_BIT_INODE +# define _DARWIN_USE_64_BIT_INODE 1 +#endif + +/* Number of bits in a file offset, on hosts where this is settable. */ +/* #undef _FILE_OFFSET_BITS */ + +/* Define to 1 to make fseeko visible on some hosts (e.g. glibc 2.2). */ +/* #undef _LARGEFILE_SOURCE */ + +/* Define for large files, on AIX-style hosts. */ +/* #undef _LARGE_FILES */ + +/* Define to 1 if on MINIX. */ +/* #undef _MINIX */ + +/* Define to 2 if the system does not provide POSIX.1 features except with + this defined. */ +/* #undef _POSIX_1_SOURCE */ + +/* Define to 1 if you need to in order for `stat' and other things to work. */ +/* #undef _POSIX_SOURCE */ + +/* Define for Solaris 2.5.1 so the uint32_t typedef from , + , or is not used. If the typedef were allowed, the + #define below would cause a syntax error. */ +/* #undef _UINT32_T */ + +/* Define for Solaris 2.5.1 so the uint64_t typedef from , + , or is not used. If the typedef were allowed, the + #define below would cause a syntax error. */ +/* #undef _UINT64_T */ + +/* Define for Solaris 2.5.1 so the uint8_t typedef from , + , or is not used. If the typedef were allowed, the + #define below would cause a syntax error. */ +/* #undef _UINT8_T */ + +/* Define to '0x0502' for Windows Server 2003 APIs. */ +/* #undef _WIN32_WINNT */ + +/* Define to empty if `const' does not conform to ANSI C. */ +/* #undef const */ + +/* Define to match typeof st_gid field of struct stat if doesn't + define. */ +/* #undef gid_t */ + +/* Define to `unsigned long' if does not define. */ +/* #undef id_t */ + +/* Define to the type of a signed integer type of width exactly 16 bits if + such a type exists and the standard includes do not define it. */ +/* #undef int16_t */ + +/* Define to the type of a signed integer type of width exactly 32 bits if + such a type exists and the standard includes do not define it. */ +/* #undef int32_t */ + +/* Define to the type of a signed integer type of width exactly 64 bits if + such a type exists and the standard includes do not define it. */ +/* #undef int64_t */ + +/* Define to the widest signed integer type if and do + not define. */ +/* #undef intmax_t */ + +/* Define to `int' if does not define. */ +/* #undef mode_t */ + +/* Define to `long long' if does not define. */ +/* #undef off_t */ + +/* Define to `unsigned int' if does not define. */ +/* #undef size_t */ + +/* Define to match typeof st_uid field of struct stat if doesn't + define. */ +/* #undef uid_t */ + +/* Define to the type of an unsigned integer type of width exactly 16 bits if + such a type exists and the standard includes do not define it. */ +/* #undef uint16_t */ + +/* Define to the type of an unsigned integer type of width exactly 32 bits if + such a type exists and the standard includes do not define it. */ +/* #undef uint32_t */ + +/* Define to the type of an unsigned integer type of width exactly 64 bits if + such a type exists and the standard includes do not define it. */ +/* #undef uint64_t */ + +/* Define to the type of an unsigned integer type of width exactly 8 bits if + such a type exists and the standard includes do not define it. */ +/* #undef uint8_t */ + +/* Define to the widest unsigned integer type if and + do not define. */ +/* #undef uintmax_t */ + +/* Define to `unsigned int' if does not define. */ +/* #undef uintptr_t */ diff --git a/c/autoconf-freebsd/config.h b/c/autoconf-freebsd/config.h new file mode 100644 --- /dev/null +++ b/c/autoconf-freebsd/config.h @@ -0,0 +1,1417 @@ +/* config.h. Generated from config.h.in by configure. */ +/* config.h.in. Generated from configure.ac by autoheader. */ + +/* Darwin ACL support */ +/* #undef ARCHIVE_ACL_DARWIN */ + +/* FreeBSD ACL support */ +/* #undef ARCHIVE_ACL_FREEBSD */ + +/* FreeBSD NFSv4 ACL support */ +/* #undef ARCHIVE_ACL_FREEBSD_NFS4 */ + +/* Linux POSIX.1e ACL support via libacl */ +/* #undef ARCHIVE_ACL_LIBACL */ + +/* Linux NFSv4 ACL support via librichacl */ +/* #undef ARCHIVE_ACL_LIBRICHACL */ + +/* Solaris ACL support */ +/* #undef ARCHIVE_ACL_SUNOS */ + +/* Solaris NFSv4 ACL support */ +/* #undef ARCHIVE_ACL_SUNOS_NFS4 */ + +/* MD5 via ARCHIVE_CRYPTO_MD5_LIBC supported. */ +/* #undef ARCHIVE_CRYPTO_MD5_LIBC */ + +/* MD5 via ARCHIVE_CRYPTO_MD5_LIBMD supported. */ + +/* MD5 via ARCHIVE_CRYPTO_MD5_LIBSYSTEM supported. */ +/* #undef ARCHIVE_CRYPTO_MD5_LIBSYSTEM */ + +/* MD5 via ARCHIVE_CRYPTO_MD5_MBEDTLS supported. */ +/* #undef ARCHIVE_CRYPTO_MD5_MBEDTLS */ + +/* MD5 via ARCHIVE_CRYPTO_MD5_NETTLE supported. */ +/* #undef ARCHIVE_CRYPTO_MD5_NETTLE */ + +/* MD5 via ARCHIVE_CRYPTO_MD5_OPENSSL supported. */ +/* #undef ARCHIVE_CRYPTO_MD5_OPENSSL */ + +/* MD5 via ARCHIVE_CRYPTO_MD5_WIN supported. */ +/* #undef ARCHIVE_CRYPTO_MD5_WIN */ + +/* RMD160 via ARCHIVE_CRYPTO_RMD160_LIBC supported. */ +/* #undef ARCHIVE_CRYPTO_RMD160_LIBC */ + +/* RMD160 via ARCHIVE_CRYPTO_RMD160_LIBMD supported. */ + +/* RMD160 via ARCHIVE_CRYPTO_RMD160_MBEDTLS supported. */ +/* #undef ARCHIVE_CRYPTO_RMD160_MBEDTLS */ + +/* RMD160 via ARCHIVE_CRYPTO_RMD160_NETTLE supported. */ +/* #undef ARCHIVE_CRYPTO_RMD160_NETTLE */ + +/* RMD160 via ARCHIVE_CRYPTO_RMD160_OPENSSL supported. */ +/* #undef ARCHIVE_CRYPTO_RMD160_OPENSSL */ + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_LIBC supported. */ +/* #undef ARCHIVE_CRYPTO_SHA1_LIBC */ + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_LIBMD supported. */ + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_LIBSYSTEM supported. */ +/* #undef ARCHIVE_CRYPTO_SHA1_LIBSYSTEM */ + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_MBEDTLS supported. */ +/* #undef ARCHIVE_CRYPTO_SHA1_MBEDTLS */ + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_NETTLE supported. */ +/* #undef ARCHIVE_CRYPTO_SHA1_NETTLE */ + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_OPENSSL supported. */ +/* #undef ARCHIVE_CRYPTO_SHA1_OPENSSL */ + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_WIN supported. */ +/* #undef ARCHIVE_CRYPTO_SHA1_WIN */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_LIBC supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_LIBC */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_LIBC2 supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_LIBC2 */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_LIBC3 supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_LIBC3 */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_LIBMD supported. */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_LIBSYSTEM supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_LIBSYSTEM */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_MBEDTLS supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_MBEDTLS */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_NETTLE supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_NETTLE */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_OPENSSL supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_OPENSSL */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_WIN supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_WIN */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_LIBC supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_LIBC */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_LIBC2 supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_LIBC2 */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_LIBC3 supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_LIBC3 */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_LIBSYSTEM supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_LIBSYSTEM */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_MBEDTLS supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_MBEDTLS */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_NETTLE supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_NETTLE */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_OPENSSL supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_OPENSSL */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_WIN supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_WIN */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_LIBC supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_LIBC */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_LIBC2 supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_LIBC2 */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_LIBC3 supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_LIBC3 */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_LIBMD supported. */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_LIBSYSTEM supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_LIBSYSTEM */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_MBEDTLS supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_MBEDTLS */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_NETTLE supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_NETTLE */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_OPENSSL supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_OPENSSL */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_WIN supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_WIN */ + +/* AIX xattr support */ +/* #undef ARCHIVE_XATTR_AIX */ + +/* Darwin xattr support */ +/* #undef ARCHIVE_XATTR_DARWIN */ + +/* FreeBSD xattr support */ +/* #undef ARCHIVE_XATTR_FREEBSD */ + +/* Linux xattr support */ +/* #undef ARCHIVE_XATTR_LINUX */ + +/* Version number of bsdcat */ +#define BSDCAT_VERSION_STRING "3.4.3" + +/* Version number of bsdcpio */ +#define BSDCPIO_VERSION_STRING "3.4.3" + +/* Version number of bsdtar */ +#define BSDTAR_VERSION_STRING "3.4.3" + +/* Define to 1 if the system has the type `ace_t'. */ +/* #undef HAVE_ACE_T */ + +/* Define to 1 if you have the `acl' function. */ +/* #undef HAVE_ACL */ + +/* Define to 1 if the system has the type `aclent_t'. */ +/* #undef HAVE_ACLENT_T */ + +/* Define to 1 if you have the `acl_add_flag_np' function. */ +/* #undef HAVE_ACL_ADD_FLAG_NP */ + +/* Define to 1 if you have the `acl_add_perm' function. */ +/* #undef HAVE_ACL_ADD_PERM */ + +/* Define to 1 if you have the `acl_clear_flags_np' function. */ +/* #undef HAVE_ACL_CLEAR_FLAGS_NP */ + +/* Define to 1 if you have the `acl_clear_perms' function. */ +/* #undef HAVE_ACL_CLEAR_PERMS */ + +/* Define to 1 if you have the `acl_create_entry' function. */ +/* #undef HAVE_ACL_CREATE_ENTRY */ + +/* Define to 1 if you have the `acl_delete_def_file' function. */ +/* #undef HAVE_ACL_DELETE_DEF_FILE */ + +/* Define to 1 if the system has the type `acl_entry_t'. */ +/* #undef HAVE_ACL_ENTRY_T */ + +/* Define to 1 if you have the `acl_free' function. */ +/* #undef HAVE_ACL_FREE */ + +/* Define to 1 if you have the `acl_get_brand_np' function. */ +/* #undef HAVE_ACL_GET_BRAND_NP */ + +/* Define to 1 if you have the `acl_get_entry' function. */ +/* #undef HAVE_ACL_GET_ENTRY */ + +/* Define to 1 if you have the `acl_get_entry_type_np' function. */ +/* #undef HAVE_ACL_GET_ENTRY_TYPE_NP */ + +/* Define to 1 if you have the `acl_get_fd' function. */ +/* #undef HAVE_ACL_GET_FD */ + +/* Define to 1 if you have the `acl_get_fd_np' function. */ +/* #undef HAVE_ACL_GET_FD_NP */ + +/* Define to 1 if you have the `acl_get_file' function. */ +/* #undef HAVE_ACL_GET_FILE */ + +/* Define to 1 if you have the `acl_get_flagset_np' function. */ +/* #undef HAVE_ACL_GET_FLAGSET_NP */ + +/* Define to 1 if you have the `acl_get_flag_np' function. */ +/* #undef HAVE_ACL_GET_FLAG_NP */ + +/* Define to 1 if you have the `acl_get_link_np' function. */ +/* #undef HAVE_ACL_GET_LINK_NP */ + +/* Define to 1 if you have the `acl_get_perm' function. */ +/* #undef HAVE_ACL_GET_PERM */ + +/* Define to 1 if you have the `acl_get_permset' function. */ +/* #undef HAVE_ACL_GET_PERMSET */ + +/* Define to 1 if you have the `acl_get_perm_np' function. */ +/* #undef HAVE_ACL_GET_PERM_NP */ + +/* Define to 1 if you have the `acl_get_qualifier' function. */ +/* #undef HAVE_ACL_GET_QUALIFIER */ + +/* Define to 1 if you have the `acl_get_tag_type' function. */ +/* #undef HAVE_ACL_GET_TAG_TYPE */ + +/* Define to 1 if you have the `acl_init' function. */ +/* #undef HAVE_ACL_INIT */ + +/* Define to 1 if you have the `acl_is_trivial_np' function. */ +/* #undef HAVE_ACL_IS_TRIVIAL_NP */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_ACL_LIBACL_H */ + +/* Define to 1 if the system has the type `acl_permset_t'. */ +/* #undef HAVE_ACL_PERMSET_T */ + +/* Define to 1 if you have the `acl_set_entry_type_np' function. */ +/* #undef HAVE_ACL_SET_ENTRY_TYPE_NP */ + +/* Define to 1 if you have the `acl_set_fd' function. */ +/* #undef HAVE_ACL_SET_FD */ + +/* Define to 1 if you have the `acl_set_fd_np' function. */ +/* #undef HAVE_ACL_SET_FD_NP */ + +/* Define to 1 if you have the `acl_set_file' function. */ +/* #undef HAVE_ACL_SET_FILE */ + +/* Define to 1 if you have the `acl_set_link_np' function. */ +/* #undef HAVE_ACL_SET_LINK_NP */ + +/* Define to 1 if you have the `acl_set_qualifier' function. */ +/* #undef HAVE_ACL_SET_QUALIFIER */ + +/* Define to 1 if you have the `acl_set_tag_type' function. */ +/* #undef HAVE_ACL_SET_TAG_TYPE */ + +/* Define to 1 if the system has the type `acl_t'. */ +/* #undef HAVE_ACL_T */ + +/* Define to 1 if the system has the type `acl_tag_t'. */ +/* #undef HAVE_ACL_TAG_T */ + +/* Define to 1 if you have the `arc4random_buf' function. */ +#define HAVE_ARC4RANDOM_BUF 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_ATTR_XATTR_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_BCRYPT_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_BLAKE2_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_BZLIB_H */ + +/* Define to 1 if you have the `chflags' function. */ +#define HAVE_CHFLAGS 1 + +/* Define to 1 if you have the `chown' function. */ +#define HAVE_CHOWN 1 + +/* Define to 1 if you have the `chroot' function. */ +#define HAVE_CHROOT 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_COPYFILE_H */ + +/* Define to 1 if you have the `ctime_r' function. */ +#define HAVE_CTIME_R 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_CTYPE_H 1 + +/* Define to 1 if you have the `cygwin_conv_path' function. */ +/* #undef HAVE_CYGWIN_CONV_PATH */ + +/* Define to 1 if you have the declaration of `ACE_GETACL', and to 0 if you + don't. */ +/* #undef HAVE_DECL_ACE_GETACL */ + +/* Define to 1 if you have the declaration of `ACE_GETACLCNT', and to 0 if you + don't. */ +/* #undef HAVE_DECL_ACE_GETACLCNT */ + +/* Define to 1 if you have the declaration of `ACE_SETACL', and to 0 if you + don't. */ +/* #undef HAVE_DECL_ACE_SETACL */ + +/* Define to 1 if you have the declaration of `ACL_SYNCHRONIZE', and to 0 if + you don't. */ +/* #undef HAVE_DECL_ACL_SYNCHRONIZE */ + +/* Define to 1 if you have the declaration of `ACL_TYPE_EXTENDED', and to 0 if + you don't. */ +/* #undef HAVE_DECL_ACL_TYPE_EXTENDED */ + +/* Define to 1 if you have the declaration of `ACL_TYPE_NFS4', and to 0 if you + don't. */ +/* #undef HAVE_DECL_ACL_TYPE_NFS4 */ + +/* Define to 1 if you have the declaration of `ACL_USER', and to 0 if you + don't. */ +/* #undef HAVE_DECL_ACL_USER */ + +/* Define to 1 if you have the declaration of `EXTATTR_NAMESPACE_USER', and to + 0 if you don't. */ +/* #undef HAVE_DECL_EXTATTR_NAMESPACE_USER */ + +/* Define to 1 if you have the declaration of `GETACL', and to 0 if you don't. + */ +/* #undef HAVE_DECL_GETACL */ + +/* Define to 1 if you have the declaration of `GETACLCNT', and to 0 if you + don't. */ +/* #undef HAVE_DECL_GETACLCNT */ + +/* Define to 1 if you have the declaration of `INT32_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_INT32_MAX 1 + +/* Define to 1 if you have the declaration of `INT32_MIN', and to 0 if you + don't. */ +#define HAVE_DECL_INT32_MIN 1 + +/* Define to 1 if you have the declaration of `INT64_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_INT64_MAX 1 + +/* Define to 1 if you have the declaration of `INT64_MIN', and to 0 if you + don't. */ +#define HAVE_DECL_INT64_MIN 1 + +/* Define to 1 if you have the declaration of `INTMAX_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_INTMAX_MAX 1 + +/* Define to 1 if you have the declaration of `INTMAX_MIN', and to 0 if you + don't. */ +#define HAVE_DECL_INTMAX_MIN 1 + +/* Define to 1 if you have the declaration of `SETACL', and to 0 if you don't. + */ +/* #undef HAVE_DECL_SETACL */ + +/* Define to 1 if you have the declaration of `SIZE_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_SIZE_MAX 1 + +/* Define to 1 if you have the declaration of `SSIZE_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_SSIZE_MAX 1 + +/* Define to 1 if you have the declaration of `strerror_r', and to 0 if you + don't. */ +#define HAVE_DECL_STRERROR_R 1 + +/* Define to 1 if you have the declaration of `UINT32_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_UINT32_MAX 1 + +/* Define to 1 if you have the declaration of `UINT64_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_UINT64_MAX 1 + +/* Define to 1 if you have the declaration of `UINTMAX_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_UINTMAX_MAX 1 + +/* Define to 1 if you have the declaration of `XATTR_NOFOLLOW', and to 0 if + you don't. */ +/* #undef HAVE_DECL_XATTR_NOFOLLOW */ + +/* Define to 1 if you have the header file, and it defines `DIR'. + */ +#define HAVE_DIRENT_H 1 + +/* Define to 1 if you have a dirfd function or macro */ +#define HAVE_DIRFD 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_DLFCN_H 1 + +/* Define to 1 if you don't have `vprintf' but do have `_doprnt.' */ +/* #undef HAVE_DOPRNT */ + +/* Define to 1 if nl_langinfo supports D_MD_ORDER */ +#define HAVE_D_MD_ORDER 1 + +/* A possible errno value for invalid file format errors */ +#define HAVE_EFTYPE 1 + +/* A possible errno value for invalid file format errors */ +#define HAVE_EILSEQ 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_ERRNO_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_EXPAT_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_EXT2FS_EXT2_FS_H */ + +/* Define to 1 if you have the `extattr_get_fd' function. */ +/* #undef HAVE_EXTATTR_GET_FD */ + +/* Define to 1 if you have the `extattr_get_file' function. */ +/* #undef HAVE_EXTATTR_GET_FILE */ + +/* Define to 1 if you have the `extattr_get_link' function. */ +/* #undef HAVE_EXTATTR_GET_LINK */ + +/* Define to 1 if you have the `extattr_list_fd' function. */ +/* #undef HAVE_EXTATTR_LIST_FD */ + +/* Define to 1 if you have the `extattr_list_file' function. */ +/* #undef HAVE_EXTATTR_LIST_FILE */ + +/* Define to 1 if you have the `extattr_list_link' function. */ +/* #undef HAVE_EXTATTR_LIST_LINK */ + +/* Define to 1 if you have the `extattr_set_fd' function. */ +/* #undef HAVE_EXTATTR_SET_FD */ + +/* Define to 1 if you have the `extattr_set_link' function. */ +/* #undef HAVE_EXTATTR_SET_LINK */ + +/* Define to 1 if you have the `facl' function. */ +/* #undef HAVE_FACL */ + +/* Define to 1 if you have the `fchdir' function. */ +#define HAVE_FCHDIR 1 + +/* Define to 1 if you have the `fchflags' function. */ +#define HAVE_FCHFLAGS 1 + +/* Define to 1 if you have the `fchmod' function. */ +#define HAVE_FCHMOD 1 + +/* Define to 1 if you have the `fchown' function. */ +#define HAVE_FCHOWN 1 + +/* Define to 1 if you have the `fcntl' function. */ +#define HAVE_FCNTL 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_FCNTL_H 1 + +/* Define to 1 if you have the `fdopendir' function. */ +#define HAVE_FDOPENDIR 1 + +/* Define to 1 if you have the `fgetea' function. */ +/* #undef HAVE_FGETEA */ + +/* Define to 1 if you have the `fgetxattr' function. */ +/* #undef HAVE_FGETXATTR */ + +/* Define to 1 if you have the `flistea' function. */ +/* #undef HAVE_FLISTEA */ + +/* Define to 1 if you have the `flistxattr' function. */ +/* #undef HAVE_FLISTXATTR */ + +/* Define to 1 if you have the `fork' function. */ +#define HAVE_FORK 1 + +/* Define to 1 if fseeko (and presumably ftello) exists and is declared. */ +#define HAVE_FSEEKO 1 + +/* Define to 1 if you have the `fsetea' function. */ +/* #undef HAVE_FSETEA */ + +/* Define to 1 if you have the `fsetxattr' function. */ +/* #undef HAVE_FSETXATTR */ + +/* Define to 1 if you have the `fstat' function. */ +#define HAVE_FSTAT 1 + +/* Define to 1 if you have the `fstatat' function. */ +#define HAVE_FSTATAT 1 + +/* Define to 1 if you have the `fstatfs' function. */ +#define HAVE_FSTATFS 1 + +/* Define to 1 if you have the `fstatvfs' function. */ +#define HAVE_FSTATVFS 1 + +/* Define to 1 if you have the `ftruncate' function. */ +#define HAVE_FTRUNCATE 1 + +/* Define to 1 if you have the `futimens' function. */ +#define HAVE_FUTIMENS 1 + +/* Define to 1 if you have the `futimes' function. */ +#define HAVE_FUTIMES 1 + +/* Define to 1 if you have the `futimesat' function. */ +#define HAVE_FUTIMESAT 1 + +/* Define to 1 if you have the `getea' function. */ +/* #undef HAVE_GETEA */ + +/* Define to 1 if you have the `geteuid' function. */ +#define HAVE_GETEUID 1 + +/* Define to 1 if you have the `getgrgid_r' function. */ +#define HAVE_GETGRGID_R 1 + +/* Define to 1 if you have the `getgrnam_r' function. */ +#define HAVE_GETGRNAM_R 1 + +/* Define to 1 if you have the `getpid' function. */ +#define HAVE_GETPID 1 + +/* Define to 1 if you have the `getpwnam_r' function. */ +#define HAVE_GETPWNAM_R 1 + +/* Define to 1 if you have the `getpwuid_r' function. */ +#define HAVE_GETPWUID_R 1 + +/* Define to 1 if you have the `getvfsbyname' function. */ +#define HAVE_GETVFSBYNAME 1 + +/* Define to 1 if you have the `getxattr' function. */ +/* #undef HAVE_GETXATTR */ + +/* Define to 1 if you have the `gmtime_r' function. */ +#define HAVE_GMTIME_R 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_GRP_H 1 + +/* Define if you have the iconv() function and it works. */ +/* #undef HAVE_ICONV */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_ICONV_H */ + +/* Define to 1 if the system has the type `intmax_t'. */ +#define HAVE_INTMAX_T 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_INTTYPES_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_IO_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_LANGINFO_H 1 + +/* Define to 1 if you have the `lchflags' function. */ +#define HAVE_LCHFLAGS 1 + +/* Define to 1 if you have the `lchmod' function. */ +#define HAVE_LCHMOD 1 + +/* Define to 1 if you have the `lchown' function. */ +#define HAVE_LCHOWN 1 + +/* Define to 1 if you have the `lgetea' function. */ +/* #undef HAVE_LGETEA */ + +/* Define to 1 if you have the `lgetxattr' function. */ +/* #undef HAVE_LGETXATTR */ + +/* Define to 1 if you have the `acl' library (-lacl). */ +/* #undef HAVE_LIBACL */ + +/* Define to 1 if you have the `b2' library (-lb2). */ +/* #undef HAVE_LIBB2 */ + +/* Define to 1 if you have the `bz2' library (-lbz2). */ +/* #undef HAVE_LIBBZ2 */ + +/* Define to 1 if you have the `charset' library (-lcharset). */ +/* #undef HAVE_LIBCHARSET */ + +/* Define to 1 if you have the `crypto' library (-lcrypto). */ +/* #undef HAVE_LIBCRYPTO */ + +/* Define to 1 if you have the `eay32' library (-leay32). */ +/* #undef HAVE_LIBEAY32 */ + +/* Define to 1 if you have the `eay64' library (-leay64). */ +/* #undef HAVE_LIBEAY64 */ + +/* Define to 1 if you have the `expat' library (-lexpat). */ +/* #undef HAVE_LIBEXPAT */ + +/* Define to 1 if you have the `lz4' library (-llz4). */ +/* #undef HAVE_LIBLZ4 */ + +/* Define to 1 if you have the `lzma' library (-llzma). */ +/* #undef HAVE_LIBLZMA */ + +/* Define to 1 if you have the `lzo2' library (-llzo2). */ +/* #undef HAVE_LIBLZO2 */ + +/* Define to 1 if you have the `mbedcrypto' library (-lmbedcrypto). */ +/* #undef HAVE_LIBMBEDCRYPTO */ + +/* Define to 1 if you have the `md' library (-lmd). */ + +/* Define to 1 if you have the `nettle' library (-lnettle). */ +/* #undef HAVE_LIBNETTLE */ + +/* Define to 1 if you have the `pcre' library (-lpcre). */ +/* #undef HAVE_LIBPCRE */ + +/* Define to 1 if you have the `pcreposix' library (-lpcreposix). */ +/* #undef HAVE_LIBPCREPOSIX */ + +/* Define to 1 if you have the `regex' library (-lregex). */ +/* #undef HAVE_LIBREGEX */ + +/* Define to 1 if you have the `richacl' library (-lrichacl). */ +/* #undef HAVE_LIBRICHACL */ + +/* Define to 1 if you have the `xml2' library (-lxml2). */ +/* #undef HAVE_LIBXML2 */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LIBXML_XMLREADER_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LIBXML_XMLWRITER_H */ + +/* Define to 1 if you have the `z' library (-lz). */ +/* #undef HAVE_LIBZ */ + +/* Define to 1 if you have the `zstd' library (-lzstd). */ +/* #undef HAVE_LIBZSTD */ + +/* Define to 1 if you have the header file. */ +#define HAVE_LIMITS_H 1 + +/* Define to 1 if you have the `link' function. */ +#define HAVE_LINK 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LINUX_FIEMAP_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LINUX_FS_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LINUX_MAGIC_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LINUX_TYPES_H */ + +/* Define to 1 if you have the `listea' function. */ +/* #undef HAVE_LISTEA */ + +/* Define to 1 if you have the `listxattr' function. */ +/* #undef HAVE_LISTXATTR */ + +/* Define to 1 if you have the `llistea' function. */ +/* #undef HAVE_LLISTEA */ + +/* Define to 1 if you have the `llistxattr' function. */ +/* #undef HAVE_LLISTXATTR */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LOCALCHARSET_H */ + +/* Define to 1 if you have the `locale_charset' function. */ +/* #undef HAVE_LOCALE_CHARSET */ + +/* Define to 1 if you have the header file. */ +#define HAVE_LOCALE_H 1 + +/* Define to 1 if you have the `localtime_r' function. */ +#define HAVE_LOCALTIME_R 1 + +/* Define to 1 if the system has the type `long long int'. */ +#define HAVE_LONG_LONG_INT 1 + +/* Define to 1 if you have the `lsetea' function. */ +/* #undef HAVE_LSETEA */ + +/* Define to 1 if you have the `lsetxattr' function. */ +/* #undef HAVE_LSETXATTR */ + +/* Define to 1 if you have the `lstat' function. */ +#define HAVE_LSTAT 1 + +/* Define to 1 if `lstat' has the bug that it succeeds when given the + zero-length file name argument. */ +/* #undef HAVE_LSTAT_EMPTY_STRING_BUG */ + +/* Define to 1 if you have the `lutimes' function. */ +#define HAVE_LUTIMES 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LZ4HC_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LZ4_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LZMA_H */ + +/* Define to 1 if you have the `lzma_stream_encoder_mt' function. */ +/* #undef HAVE_LZMA_STREAM_ENCODER_MT */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LZO_LZO1X_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LZO_LZOCONF_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_MBEDTLS_AES_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_MBEDTLS_MD_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_MBEDTLS_PKCS5_H */ + +/* Define to 1 if you have the `mbrtowc' function. */ +#define HAVE_MBRTOWC 1 + +/* Define to 1 if you have the `mbr_gid_to_uuid' function. */ +/* #undef HAVE_MBR_GID_TO_UUID */ + +/* Define to 1 if you have the `mbr_uid_to_uuid' function. */ +/* #undef HAVE_MBR_UID_TO_UUID */ + +/* Define to 1 if you have the `mbr_uuid_to_id' function. */ +/* #undef HAVE_MBR_UUID_TO_ID */ + +/* Define to 1 if you have the header file. */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_MEMBERSHIP_H */ + +/* Define to 1 if you have the `memmove' function. */ +#define HAVE_MEMMOVE 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_MEMORY_H 1 + +/* Define to 1 if you have the `memset' function. */ +#define HAVE_MEMSET 1 + +/* Define to 1 if you have the `mkdir' function. */ +#define HAVE_MKDIR 1 + +/* Define to 1 if you have the `mkfifo' function. */ +#define HAVE_MKFIFO 1 + +/* Define to 1 if you have the `mknod' function. */ +#define HAVE_MKNOD 1 + +/* Define to 1 if you have the `mkstemp' function. */ +#define HAVE_MKSTEMP 1 + +/* Define to 1 if you have the header file, and it defines `DIR'. */ +/* #undef HAVE_NDIR_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NETTLE_AES_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NETTLE_HMAC_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NETTLE_MD5_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NETTLE_PBKDF2_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NETTLE_RIPEMD160_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NETTLE_SHA_H */ + +/* Define to 1 if you have the `nl_langinfo' function. */ +#define HAVE_NL_LANGINFO 1 + +/* Define to 1 if you have the `openat' function. */ +#define HAVE_OPENAT 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_OPENSSL_EVP_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_PATHS_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_PCREPOSIX_H */ + +/* Define to 1 if you have the `pipe' function. */ +#define HAVE_PIPE 1 + +/* Define to 1 if you have the `PKCS5_PBKDF2_HMAC_SHA1' function. */ +/* #undef HAVE_PKCS5_PBKDF2_HMAC_SHA1 */ + +/* Define to 1 if you have the `poll' function. */ +#define HAVE_POLL 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_POLL_H 1 + +/* Define to 1 if you have the `posix_spawnp' function. */ +#define HAVE_POSIX_SPAWNP 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_PTHREAD_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_PWD_H 1 + +/* Define to 1 if you have a POSIX compatible readdir_r */ +#define HAVE_READDIR_R 1 + +/* Define to 1 if you have the `readlink' function. */ +#define HAVE_READLINK 1 + +/* Define to 1 if you have the `readlinkat' function. */ +#define HAVE_READLINKAT 1 + +/* Define to 1 if you have the `readpassphrase' function. */ +#define HAVE_READPASSPHRASE 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_READPASSPHRASE_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_REGEX_H */ + +/* Define to 1 if you have the `richacl_alloc' function. */ +/* #undef HAVE_RICHACL_ALLOC */ + +/* Define to 1 if you have the `richacl_equiv_mode' function. */ +/* #undef HAVE_RICHACL_EQUIV_MODE */ + +/* Define to 1 if you have the `richacl_free' function. */ +/* #undef HAVE_RICHACL_FREE */ + +/* Define to 1 if you have the `richacl_get_fd' function. */ +/* #undef HAVE_RICHACL_GET_FD */ + +/* Define to 1 if you have the `richacl_get_file' function. */ +/* #undef HAVE_RICHACL_GET_FILE */ + +/* Define to 1 if you have the `richacl_set_fd' function. */ +/* #undef HAVE_RICHACL_SET_FD */ + +/* Define to 1 if you have the `richacl_set_file' function. */ +/* #undef HAVE_RICHACL_SET_FILE */ + +/* Define to 1 if you have the header file. */ + +/* Define to 1 if you have the `select' function. */ +#define HAVE_SELECT 1 + +/* Define to 1 if you have the `setenv' function. */ +#define HAVE_SETENV 1 + +/* Define to 1 if you have the `setlocale' function. */ +#define HAVE_SETLOCALE 1 + +/* Define to 1 if you have the `setxattr' function. */ +/* #undef HAVE_SETXATTR */ + +/* Define to 1 if you have the header file. */ + +/* Define to 1 if you have the header file. */ + +/* Define to 1 if you have the header file. */ + +/* Define to 1 if you have the `sigaction' function. */ +#define HAVE_SIGACTION 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SIGNAL_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SPAWN_H 1 + +/* Define to 1 if you have the `statfs' function. */ +#define HAVE_STATFS 1 + +/* Define to 1 if you have the `statvfs' function. */ +#define HAVE_STATVFS 1 + +/* Define to 1 if `stat' has the bug that it succeeds when given the + zero-length file name argument. */ +/* #undef HAVE_STAT_EMPTY_STRING_BUG */ + +/* Define to 1 if you have the header file. */ +#define HAVE_STDARG_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_STDINT_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_STDLIB_H 1 + +/* Define to 1 if you have the `strchr' function. */ +#define HAVE_STRCHR 1 + +/* Define to 1 if you have the `strdup' function. */ +#define HAVE_STRDUP 1 + +/* Define to 1 if you have the `strerror' function. */ +#define HAVE_STRERROR 1 + +/* Define to 1 if you have the `strerror_r' function. */ +#define HAVE_STRERROR_R 1 + +/* Define to 1 if you have the `strftime' function. */ +#define HAVE_STRFTIME 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_STRINGS_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_STRING_H 1 + +/* Define to 1 if you have the `strncpy_s' function. */ +/* #undef HAVE_STRNCPY_S */ + +/* Define to 1 if you have the `strrchr' function. */ +#define HAVE_STRRCHR 1 + +/* Define to 1 if the system has the type `struct richace'. */ +/* #undef HAVE_STRUCT_RICHACE */ + +/* Define to 1 if the system has the type `struct richacl'. */ +/* #undef HAVE_STRUCT_RICHACL */ + +/* Define to 1 if `f_namemax' is a member of `struct statfs'. */ +#define HAVE_STRUCT_STATFS_F_NAMEMAX 1 + +/* Define to 1 if `f_iosize' is a member of `struct statvfs'. */ +/* #undef HAVE_STRUCT_STATVFS_F_IOSIZE */ + +/* Define to 1 if `st_birthtime' is a member of `struct stat'. */ +#define HAVE_STRUCT_STAT_ST_BIRTHTIME 1 + +/* Define to 1 if `st_birthtimespec.tv_nsec' is a member of `struct stat'. */ +#define HAVE_STRUCT_STAT_ST_BIRTHTIMESPEC_TV_NSEC 1 + +/* Define to 1 if `st_blksize' is a member of `struct stat'. */ +#define HAVE_STRUCT_STAT_ST_BLKSIZE 1 + +/* Define to 1 if `st_flags' is a member of `struct stat'. */ +#define HAVE_STRUCT_STAT_ST_FLAGS 1 + +/* Define to 1 if `st_mtimespec.tv_nsec' is a member of `struct stat'. */ +#define HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC 1 + +/* Define to 1 if `st_mtime_n' is a member of `struct stat'. */ +/* #undef HAVE_STRUCT_STAT_ST_MTIME_N */ + +/* Define to 1 if `st_mtime_usec' is a member of `struct stat'. */ +/* #undef HAVE_STRUCT_STAT_ST_MTIME_USEC */ + +/* Define to 1 if `st_mtim.tv_nsec' is a member of `struct stat'. */ +#define HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC 1 + +/* Define to 1 if `st_umtime' is a member of `struct stat'. */ +/* #undef HAVE_STRUCT_STAT_ST_UMTIME */ + +/* Define to 1 if `tm_gmtoff' is a member of `struct tm'. */ +#define HAVE_STRUCT_TM_TM_GMTOFF 1 + +/* Define to 1 if `__tm_gmtoff' is a member of `struct tm'. */ +/* #undef HAVE_STRUCT_TM___TM_GMTOFF */ + +/* Define to 1 if the system has the type `struct vfsconf'. */ +#define HAVE_STRUCT_VFSCONF 1 + +/* Define to 1 if the system has the type `struct xvfsconf'. */ +#define HAVE_STRUCT_XVFSCONF 1 + +/* Define to 1 if you have the `symlink' function. */ +#define HAVE_SYMLINK 1 + +/* Define to 1 if you have the header file. */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_CDEFS_H 1 + +/* Define to 1 if you have the header file, and it defines `DIR'. + */ +/* #undef HAVE_SYS_DIR_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_EA_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_EXTATTR_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_IOCTL_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_MKDEV_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_MOUNT_H 1 + +/* Define to 1 if you have the header file, and it defines `DIR'. + */ +/* #undef HAVE_SYS_NDIR_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_PARAM_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_POLL_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_RICHACL_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_SELECT_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_STATFS_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_STATVFS_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_STAT_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_SYSMACROS_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_TIME_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_TYPES_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_UTIME_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_UTSNAME_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_VFS_H */ + +/* Define to 1 if you have that is POSIX.1 compatible. */ +#define HAVE_SYS_WAIT_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_XATTR_H */ + +/* Define to 1 if you have the `timegm' function. */ +#define HAVE_TIMEGM 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_TIME_H 1 + +/* Define to 1 if you have the `tzset' function. */ +#define HAVE_TZSET 1 + +/* Define to 1 if the system has the type `uintmax_t'. */ +#define HAVE_UINTMAX_T 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_UNISTD_H 1 + +/* Define to 1 if you have the `unlinkat' function. */ +#define HAVE_UNLINKAT 1 + +/* Define to 1 if you have the `unsetenv' function. */ +#define HAVE_UNSETENV 1 + +/* Define to 1 if the system has the type `unsigned long long'. */ +#define HAVE_UNSIGNED_LONG_LONG 1 + +/* Define to 1 if the system has the type `unsigned long long int'. */ +#define HAVE_UNSIGNED_LONG_LONG_INT 1 + +/* Define to 1 if you have the `utime' function. */ +#define HAVE_UTIME 1 + +/* Define to 1 if you have the `utimensat' function. */ +#define HAVE_UTIMENSAT 1 + +/* Define to 1 if you have the `utimes' function. */ +#define HAVE_UTIMES 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_UTIME_H 1 + +/* Define to 1 if you have the `vfork' function. */ +#define HAVE_VFORK 1 + +/* Define to 1 if you have the `vprintf' function. */ +#define HAVE_VPRINTF 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_WCHAR_H 1 + +/* Define to 1 if the system has the type `wchar_t'. */ +#define HAVE_WCHAR_T 1 + +/* Define to 1 if you have the `wcrtomb' function. */ +#define HAVE_WCRTOMB 1 + +/* Define to 1 if you have the `wcscmp' function. */ +#define HAVE_WCSCMP 1 + +/* Define to 1 if you have the `wcscpy' function. */ +#define HAVE_WCSCPY 1 + +/* Define to 1 if you have the `wcslen' function. */ +#define HAVE_WCSLEN 1 + +/* Define to 1 if you have the `wctomb' function. */ +#define HAVE_WCTOMB 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_WCTYPE_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_WINCRYPT_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_WINDOWS_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_WINIOCTL_H */ + +/* Define to 1 if you have the `wmemcmp' function. */ +#define HAVE_WMEMCMP 1 + +/* Define to 1 if you have the `wmemcpy' function. */ +#define HAVE_WMEMCPY 1 + +/* Define to 1 if you have the `wmemmove' function. */ +#define HAVE_WMEMMOVE 1 + +/* Define to 1 if you have a working EXT2_IOC_GETFLAGS */ +/* #undef HAVE_WORKING_EXT2_IOC_GETFLAGS */ + +/* Define to 1 if you have a working FS_IOC_GETFLAGS */ +/* #undef HAVE_WORKING_FS_IOC_GETFLAGS */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_ZLIB_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_ZSTD_H */ + +/* Define to 1 if you have the `_ctime64_s' function. */ +/* #undef HAVE__CTIME64_S */ + +/* Define to 1 if you have the `_fseeki64' function. */ +/* #undef HAVE__FSEEKI64 */ + +/* Define to 1 if you have the `_get_timezone' function. */ +/* #undef HAVE__GET_TIMEZONE */ + +/* Define to 1 if you have the `_gmtime64_s' function. */ +/* #undef HAVE__GMTIME64_S */ + +/* Define to 1 if you have the `_localtime64_s' function. */ +/* #undef HAVE__LOCALTIME64_S */ + +/* Define to 1 if you have the `_mkgmtime64' function. */ +/* #undef HAVE__MKGMTIME64 */ + +/* Define as const if the declaration of iconv() needs const. */ +/* #undef ICONV_CONST */ + +/* Version number of libarchive as a single integer */ +#define LIBARCHIVE_VERSION_NUMBER "3004003" + +/* Version number of libarchive */ +#define LIBARCHIVE_VERSION_STRING "3.4.3" + +/* Define to 1 if `lstat' dereferences a symlink specified with a trailing + slash. */ +#define LSTAT_FOLLOWS_SLASHED_SYMLINK 1 + +/* Define to the sub-directory where libtool stores uninstalled libraries. */ +#define LT_OBJDIR ".libs/" + +/* Define to 1 if `major', `minor', and `makedev' are declared in . + */ +/* #undef MAJOR_IN_MKDEV */ + +/* Define to 1 if `major', `minor', and `makedev' are declared in + . */ +/* #undef MAJOR_IN_SYSMACROS */ + +/* Define to '0x05020000' for Windows Server 2003 APIs. */ +/* #undef NTDDI_VERSION */ + +/* Name of package */ +#define PACKAGE "libarchive" + +/* Define to the address where bug reports for this package should be sent. */ +#define PACKAGE_BUGREPORT "libarchive-discuss@googlegroups.com" + +/* Define to the full name of this package. */ +#define PACKAGE_NAME "libarchive" + +/* Define to the full name and version of this package. */ +#define PACKAGE_STRING "libarchive 3.4.3" + +/* Define to the one symbol short name of this package. */ +#define PACKAGE_TARNAME "libarchive" + +/* Define to the home page for this package. */ +#define PACKAGE_URL "" + +/* Define to the version of this package. */ +#define PACKAGE_VERSION "3.4.3" + +/* Define to 1 if PCRE_STATIC needs to be defined. */ +/* #undef PCRE_STATIC */ + +/* The size of `wchar_t', as computed by sizeof. */ +#define SIZEOF_WCHAR_T 4 + +/* Define to 1 if you have the ANSI C header files. */ +#define STDC_HEADERS 1 + +/* Define to 1 if strerror_r returns char *. */ +/* #undef STRERROR_R_CHAR_P */ + +/* Define to 1 if you can safely include both and . */ +#define TIME_WITH_SYS_TIME 1 + +/* Enable extensions on AIX 3, Interix. */ +#ifndef _ALL_SOURCE +# define _ALL_SOURCE 1 +#endif +/* Enable GNU extensions on systems that have them. */ +#ifndef _GNU_SOURCE +# define _GNU_SOURCE 1 +#endif +/* Enable threading extensions on Solaris. */ +#ifndef _POSIX_PTHREAD_SEMANTICS +# define _POSIX_PTHREAD_SEMANTICS 1 +#endif +/* Enable extensions on HP NonStop. */ +#ifndef _TANDEM_SOURCE +# define _TANDEM_SOURCE 1 +#endif +/* Enable general extensions on Solaris. */ +#ifndef __EXTENSIONS__ +# define __EXTENSIONS__ 1 +#endif + + +/* Version number of package */ +#define VERSION "3.4.3" + +/* Define to '0x0502' for Windows Server 2003 APIs. */ +/* #undef WINVER */ + +/* Enable large inode numbers on Mac OS X 10.5. */ +#ifndef _DARWIN_USE_64_BIT_INODE +# define _DARWIN_USE_64_BIT_INODE 1 +#endif + +/* Number of bits in a file offset, on hosts where this is settable. */ +/* #undef _FILE_OFFSET_BITS */ + +/* Define to 1 to make fseeko visible on some hosts (e.g. glibc 2.2). */ +/* #undef _LARGEFILE_SOURCE */ + +/* Define for large files, on AIX-style hosts. */ +/* #undef _LARGE_FILES */ + +/* Define to 1 if on MINIX. */ +/* #undef _MINIX */ + +/* Define to 2 if the system does not provide POSIX.1 features except with + this defined. */ +/* #undef _POSIX_1_SOURCE */ + +/* Define to 1 if you need to in order for `stat' and other things to work. */ +/* #undef _POSIX_SOURCE */ + +/* Define for Solaris 2.5.1 so the uint32_t typedef from , + , or is not used. If the typedef were allowed, the + #define below would cause a syntax error. */ +/* #undef _UINT32_T */ + +/* Define for Solaris 2.5.1 so the uint64_t typedef from , + , or is not used. If the typedef were allowed, the + #define below would cause a syntax error. */ +/* #undef _UINT64_T */ + +/* Define for Solaris 2.5.1 so the uint8_t typedef from , + , or is not used. If the typedef were allowed, the + #define below would cause a syntax error. */ +/* #undef _UINT8_T */ + +/* Define to '0x0502' for Windows Server 2003 APIs. */ +/* #undef _WIN32_WINNT */ + +/* Define to empty if `const' does not conform to ANSI C. */ +/* #undef const */ + +/* Define to match typeof st_gid field of struct stat if doesn't + define. */ +/* #undef gid_t */ + +/* Define to `unsigned long' if does not define. */ +/* #undef id_t */ + +/* Define to the type of a signed integer type of width exactly 16 bits if + such a type exists and the standard includes do not define it. */ +/* #undef int16_t */ + +/* Define to the type of a signed integer type of width exactly 32 bits if + such a type exists and the standard includes do not define it. */ +/* #undef int32_t */ + +/* Define to the type of a signed integer type of width exactly 64 bits if + such a type exists and the standard includes do not define it. */ +/* #undef int64_t */ + +/* Define to the widest signed integer type if and do + not define. */ +/* #undef intmax_t */ + +/* Define to `int' if does not define. */ +/* #undef mode_t */ + +/* Define to `long long' if does not define. */ +/* #undef off_t */ + +/* Define to `unsigned int' if does not define. */ +/* #undef size_t */ + +/* Define to match typeof st_uid field of struct stat if doesn't + define. */ +/* #undef uid_t */ + +/* Define to the type of an unsigned integer type of width exactly 16 bits if + such a type exists and the standard includes do not define it. */ +/* #undef uint16_t */ + +/* Define to the type of an unsigned integer type of width exactly 32 bits if + such a type exists and the standard includes do not define it. */ +/* #undef uint32_t */ + +/* Define to the type of an unsigned integer type of width exactly 64 bits if + such a type exists and the standard includes do not define it. */ +/* #undef uint64_t */ + +/* Define to the type of an unsigned integer type of width exactly 8 bits if + such a type exists and the standard includes do not define it. */ +/* #undef uint8_t */ + +/* Define to the widest unsigned integer type if and + do not define. */ +/* #undef uintmax_t */ + +/* Define to `unsigned int' if does not define. */ +/* #undef uintptr_t */ diff --git a/c/autoconf-linux/config.h b/c/autoconf-linux/config.h new file mode 100644 --- /dev/null +++ b/c/autoconf-linux/config.h @@ -0,0 +1,1427 @@ +/* config.h. Generated from config.h.in by configure. */ +/* config.h.in. Generated from configure.ac by autoheader. */ + +/* Darwin ACL support */ +/* #undef ARCHIVE_ACL_DARWIN */ + +/* FreeBSD ACL support */ +/* #undef ARCHIVE_ACL_FREEBSD */ + +/* FreeBSD NFSv4 ACL support */ +/* #undef ARCHIVE_ACL_FREEBSD_NFS4 */ + +/* Linux POSIX.1e ACL support via libacl */ +/* #undef ARCHIVE_ACL_LIBACL */ + +/* Linux NFSv4 ACL support via librichacl */ +/* #undef ARCHIVE_ACL_LIBRICHACL */ + +/* Solaris ACL support */ +/* #undef ARCHIVE_ACL_SUNOS */ + +/* Solaris NFSv4 ACL support */ +/* #undef ARCHIVE_ACL_SUNOS_NFS4 */ + +/* MD5 via ARCHIVE_CRYPTO_MD5_LIBC supported. */ +/* #undef ARCHIVE_CRYPTO_MD5_LIBC */ + +/* MD5 via ARCHIVE_CRYPTO_MD5_LIBMD supported. */ +/* #undef ARCHIVE_CRYPTO_MD5_LIBMD */ + +/* MD5 via ARCHIVE_CRYPTO_MD5_LIBSYSTEM supported. */ +/* #undef ARCHIVE_CRYPTO_MD5_LIBSYSTEM */ + +/* MD5 via ARCHIVE_CRYPTO_MD5_MBEDTLS supported. */ +/* #undef ARCHIVE_CRYPTO_MD5_MBEDTLS */ + +/* MD5 via ARCHIVE_CRYPTO_MD5_NETTLE supported. */ +/* #undef ARCHIVE_CRYPTO_MD5_NETTLE */ + +/* MD5 via ARCHIVE_CRYPTO_MD5_OPENSSL supported. */ +/* #undef ARCHIVE_CRYPTO_MD5_OPENSSL */ + +/* MD5 via ARCHIVE_CRYPTO_MD5_WIN supported. */ +/* #undef ARCHIVE_CRYPTO_MD5_WIN */ + +/* RMD160 via ARCHIVE_CRYPTO_RMD160_LIBC supported. */ +/* #undef ARCHIVE_CRYPTO_RMD160_LIBC */ + +/* RMD160 via ARCHIVE_CRYPTO_RMD160_LIBMD supported. */ +/* #undef ARCHIVE_CRYPTO_RMD160_LIBMD */ + +/* RMD160 via ARCHIVE_CRYPTO_RMD160_MBEDTLS supported. */ +/* #undef ARCHIVE_CRYPTO_RMD160_MBEDTLS */ + +/* RMD160 via ARCHIVE_CRYPTO_RMD160_NETTLE supported. */ +/* #undef ARCHIVE_CRYPTO_RMD160_NETTLE */ + +/* RMD160 via ARCHIVE_CRYPTO_RMD160_OPENSSL supported. */ +/* #undef ARCHIVE_CRYPTO_RMD160_OPENSSL */ + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_LIBC supported. */ +/* #undef ARCHIVE_CRYPTO_SHA1_LIBC */ + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_LIBMD supported. */ +/* #undef ARCHIVE_CRYPTO_SHA1_LIBMD */ + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_LIBSYSTEM supported. */ +/* #undef ARCHIVE_CRYPTO_SHA1_LIBSYSTEM */ + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_MBEDTLS supported. */ +/* #undef ARCHIVE_CRYPTO_SHA1_MBEDTLS */ + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_NETTLE supported. */ +/* #undef ARCHIVE_CRYPTO_SHA1_NETTLE */ + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_OPENSSL supported. */ +/* #undef ARCHIVE_CRYPTO_SHA1_OPENSSL */ + +/* SHA1 via ARCHIVE_CRYPTO_SHA1_WIN supported. */ +/* #undef ARCHIVE_CRYPTO_SHA1_WIN */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_LIBC supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_LIBC */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_LIBC2 supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_LIBC2 */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_LIBC3 supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_LIBC3 */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_LIBMD supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_LIBMD */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_LIBSYSTEM supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_LIBSYSTEM */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_MBEDTLS supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_MBEDTLS */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_NETTLE supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_NETTLE */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_OPENSSL supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_OPENSSL */ + +/* SHA256 via ARCHIVE_CRYPTO_SHA256_WIN supported. */ +/* #undef ARCHIVE_CRYPTO_SHA256_WIN */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_LIBC supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_LIBC */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_LIBC2 supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_LIBC2 */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_LIBC3 supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_LIBC3 */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_LIBSYSTEM supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_LIBSYSTEM */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_MBEDTLS supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_MBEDTLS */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_NETTLE supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_NETTLE */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_OPENSSL supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_OPENSSL */ + +/* SHA384 via ARCHIVE_CRYPTO_SHA384_WIN supported. */ +/* #undef ARCHIVE_CRYPTO_SHA384_WIN */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_LIBC supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_LIBC */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_LIBC2 supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_LIBC2 */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_LIBC3 supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_LIBC3 */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_LIBMD supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_LIBMD */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_LIBSYSTEM supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_LIBSYSTEM */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_MBEDTLS supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_MBEDTLS */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_NETTLE supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_NETTLE */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_OPENSSL supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_OPENSSL */ + +/* SHA512 via ARCHIVE_CRYPTO_SHA512_WIN supported. */ +/* #undef ARCHIVE_CRYPTO_SHA512_WIN */ + +/* AIX xattr support */ +/* #undef ARCHIVE_XATTR_AIX */ + +/* Darwin xattr support */ +/* #undef ARCHIVE_XATTR_DARWIN */ + +/* FreeBSD xattr support */ +/* #undef ARCHIVE_XATTR_FREEBSD */ + +/* Linux xattr support */ +/* #undef ARCHIVE_XATTR_LINUX */ + +/* Version number of bsdcat */ +#define BSDCAT_VERSION_STRING "3.4.3" + +/* Version number of bsdcpio */ +#define BSDCPIO_VERSION_STRING "3.4.3" + +/* Version number of bsdtar */ +#define BSDTAR_VERSION_STRING "3.4.3" + +/* Define to 1 if the system has the type `ace_t'. */ +/* #undef HAVE_ACE_T */ + +/* Define to 1 if you have the `acl' function. */ +/* #undef HAVE_ACL */ + +/* Define to 1 if the system has the type `aclent_t'. */ +/* #undef HAVE_ACLENT_T */ + +/* Define to 1 if you have the `acl_add_flag_np' function. */ +/* #undef HAVE_ACL_ADD_FLAG_NP */ + +/* Define to 1 if you have the `acl_add_perm' function. */ +/* #undef HAVE_ACL_ADD_PERM */ + +/* Define to 1 if you have the `acl_clear_flags_np' function. */ +/* #undef HAVE_ACL_CLEAR_FLAGS_NP */ + +/* Define to 1 if you have the `acl_clear_perms' function. */ +/* #undef HAVE_ACL_CLEAR_PERMS */ + +/* Define to 1 if you have the `acl_create_entry' function. */ +/* #undef HAVE_ACL_CREATE_ENTRY */ + +/* Define to 1 if you have the `acl_delete_def_file' function. */ +/* #undef HAVE_ACL_DELETE_DEF_FILE */ + +/* Define to 1 if the system has the type `acl_entry_t'. */ +/* #undef HAVE_ACL_ENTRY_T */ + +/* Define to 1 if you have the `acl_free' function. */ +/* #undef HAVE_ACL_FREE */ + +/* Define to 1 if you have the `acl_get_brand_np' function. */ +/* #undef HAVE_ACL_GET_BRAND_NP */ + +/* Define to 1 if you have the `acl_get_entry' function. */ +/* #undef HAVE_ACL_GET_ENTRY */ + +/* Define to 1 if you have the `acl_get_entry_type_np' function. */ +/* #undef HAVE_ACL_GET_ENTRY_TYPE_NP */ + +/* Define to 1 if you have the `acl_get_fd' function. */ +/* #undef HAVE_ACL_GET_FD */ + +/* Define to 1 if you have the `acl_get_fd_np' function. */ +/* #undef HAVE_ACL_GET_FD_NP */ + +/* Define to 1 if you have the `acl_get_file' function. */ +/* #undef HAVE_ACL_GET_FILE */ + +/* Define to 1 if you have the `acl_get_flagset_np' function. */ +/* #undef HAVE_ACL_GET_FLAGSET_NP */ + +/* Define to 1 if you have the `acl_get_flag_np' function. */ +/* #undef HAVE_ACL_GET_FLAG_NP */ + +/* Define to 1 if you have the `acl_get_link_np' function. */ +/* #undef HAVE_ACL_GET_LINK_NP */ + +/* Define to 1 if you have the `acl_get_perm' function. */ +/* #undef HAVE_ACL_GET_PERM */ + +/* Define to 1 if you have the `acl_get_permset' function. */ +/* #undef HAVE_ACL_GET_PERMSET */ + +/* Define to 1 if you have the `acl_get_perm_np' function. */ +/* #undef HAVE_ACL_GET_PERM_NP */ + +/* Define to 1 if you have the `acl_get_qualifier' function. */ +/* #undef HAVE_ACL_GET_QUALIFIER */ + +/* Define to 1 if you have the `acl_get_tag_type' function. */ +/* #undef HAVE_ACL_GET_TAG_TYPE */ + +/* Define to 1 if you have the `acl_init' function. */ +/* #undef HAVE_ACL_INIT */ + +/* Define to 1 if you have the `acl_is_trivial_np' function. */ +/* #undef HAVE_ACL_IS_TRIVIAL_NP */ + +/* Define to 1 if you have the header file. */ +#define HAVE_ACL_LIBACL_H 1 + +/* Define to 1 if the system has the type `acl_permset_t'. */ +/* #undef HAVE_ACL_PERMSET_T */ + +/* Define to 1 if you have the `acl_set_entry_type_np' function. */ +/* #undef HAVE_ACL_SET_ENTRY_TYPE_NP */ + +/* Define to 1 if you have the `acl_set_fd' function. */ +/* #undef HAVE_ACL_SET_FD */ + +/* Define to 1 if you have the `acl_set_fd_np' function. */ +/* #undef HAVE_ACL_SET_FD_NP */ + +/* Define to 1 if you have the `acl_set_file' function. */ +/* #undef HAVE_ACL_SET_FILE */ + +/* Define to 1 if you have the `acl_set_link_np' function. */ +/* #undef HAVE_ACL_SET_LINK_NP */ + +/* Define to 1 if you have the `acl_set_qualifier' function. */ +/* #undef HAVE_ACL_SET_QUALIFIER */ + +/* Define to 1 if you have the `acl_set_tag_type' function. */ +/* #undef HAVE_ACL_SET_TAG_TYPE */ + +/* Define to 1 if the system has the type `acl_t'. */ +/* #undef HAVE_ACL_T */ + +/* Define to 1 if the system has the type `acl_tag_t'. */ +/* #undef HAVE_ACL_TAG_T */ + +/* Define to 1 if you have the `arc4random_buf' function. */ +/* #undef HAVE_ARC4RANDOM_BUF */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_ATTR_XATTR_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_BCRYPT_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_BLAKE2_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_BZLIB_H */ + +/* Define to 1 if you have the `chflags' function. */ +/* #undef HAVE_CHFLAGS */ + +/* Define to 1 if you have the `chown' function. */ +#define HAVE_CHOWN 1 + +/* Define to 1 if you have the `chroot' function. */ +#define HAVE_CHROOT 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_COPYFILE_H */ + +/* Define to 1 if you have the `ctime_r' function. */ +#define HAVE_CTIME_R 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_CTYPE_H 1 + +/* Define to 1 if you have the `cygwin_conv_path' function. */ +/* #undef HAVE_CYGWIN_CONV_PATH */ + +/* Define to 1 if you have the declaration of `ACE_GETACL', and to 0 if you + don't. */ +/* #undef HAVE_DECL_ACE_GETACL */ + +/* Define to 1 if you have the declaration of `ACE_GETACLCNT', and to 0 if you + don't. */ +/* #undef HAVE_DECL_ACE_GETACLCNT */ + +/* Define to 1 if you have the declaration of `ACE_SETACL', and to 0 if you + don't. */ +/* #undef HAVE_DECL_ACE_SETACL */ + +/* Define to 1 if you have the declaration of `ACL_SYNCHRONIZE', and to 0 if + you don't. */ +/* #undef HAVE_DECL_ACL_SYNCHRONIZE */ + +/* Define to 1 if you have the declaration of `ACL_TYPE_EXTENDED', and to 0 if + you don't. */ +/* #undef HAVE_DECL_ACL_TYPE_EXTENDED */ + +/* Define to 1 if you have the declaration of `ACL_TYPE_NFS4', and to 0 if you + don't. */ +/* #undef HAVE_DECL_ACL_TYPE_NFS4 */ + +/* Define to 1 if you have the declaration of `ACL_USER', and to 0 if you + don't. */ +/* #undef HAVE_DECL_ACL_USER */ + +/* Define to 1 if you have the declaration of `EXTATTR_NAMESPACE_USER', and to + 0 if you don't. */ +/* #undef HAVE_DECL_EXTATTR_NAMESPACE_USER */ + +/* Define to 1 if you have the declaration of `GETACL', and to 0 if you don't. + */ +/* #undef HAVE_DECL_GETACL */ + +/* Define to 1 if you have the declaration of `GETACLCNT', and to 0 if you + don't. */ +/* #undef HAVE_DECL_GETACLCNT */ + +/* Define to 1 if you have the declaration of `INT32_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_INT32_MAX 1 + +/* Define to 1 if you have the declaration of `INT32_MIN', and to 0 if you + don't. */ +#define HAVE_DECL_INT32_MIN 1 + +/* Define to 1 if you have the declaration of `INT64_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_INT64_MAX 1 + +/* Define to 1 if you have the declaration of `INT64_MIN', and to 0 if you + don't. */ +#define HAVE_DECL_INT64_MIN 1 + +/* Define to 1 if you have the declaration of `INTMAX_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_INTMAX_MAX 1 + +/* Define to 1 if you have the declaration of `INTMAX_MIN', and to 0 if you + don't. */ +#define HAVE_DECL_INTMAX_MIN 1 + +/* Define to 1 if you have the declaration of `SETACL', and to 0 if you don't. + */ +/* #undef HAVE_DECL_SETACL */ + +/* Define to 1 if you have the declaration of `SIZE_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_SIZE_MAX 1 + +/* Define to 1 if you have the declaration of `SSIZE_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_SSIZE_MAX 1 + +/* Define to 1 if you have the declaration of `strerror_r', and to 0 if you + don't. */ +#define HAVE_DECL_STRERROR_R 1 + +/* Define to 1 if you have the declaration of `UINT32_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_UINT32_MAX 1 + +/* Define to 1 if you have the declaration of `UINT64_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_UINT64_MAX 1 + +/* Define to 1 if you have the declaration of `UINTMAX_MAX', and to 0 if you + don't. */ +#define HAVE_DECL_UINTMAX_MAX 1 + +/* Define to 1 if you have the declaration of `XATTR_NOFOLLOW', and to 0 if + you don't. */ +/* #undef HAVE_DECL_XATTR_NOFOLLOW */ + +/* Define to 1 if you have the header file, and it defines `DIR'. + */ +#define HAVE_DIRENT_H 1 + +/* Define to 1 if you have a dirfd function or macro */ +#define HAVE_DIRFD 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_DLFCN_H 1 + +/* Define to 1 if you don't have `vprintf' but do have `_doprnt.' */ +/* #undef HAVE_DOPRNT */ + +/* Define to 1 if nl_langinfo supports D_MD_ORDER */ +/* #undef HAVE_D_MD_ORDER */ + +/* A possible errno value for invalid file format errors */ +/* #undef HAVE_EFTYPE */ + +/* A possible errno value for invalid file format errors */ +#define HAVE_EILSEQ 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_ERRNO_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_EXPAT_H */ + +/* Define to 1 if you have the header file. */ + +/* Define to 1 if you have the `extattr_get_fd' function. */ +/* #undef HAVE_EXTATTR_GET_FD */ + +/* Define to 1 if you have the `extattr_get_file' function. */ +/* #undef HAVE_EXTATTR_GET_FILE */ + +/* Define to 1 if you have the `extattr_get_link' function. */ +/* #undef HAVE_EXTATTR_GET_LINK */ + +/* Define to 1 if you have the `extattr_list_fd' function. */ +/* #undef HAVE_EXTATTR_LIST_FD */ + +/* Define to 1 if you have the `extattr_list_file' function. */ +/* #undef HAVE_EXTATTR_LIST_FILE */ + +/* Define to 1 if you have the `extattr_list_link' function. */ +/* #undef HAVE_EXTATTR_LIST_LINK */ + +/* Define to 1 if you have the `extattr_set_fd' function. */ +/* #undef HAVE_EXTATTR_SET_FD */ + +/* Define to 1 if you have the `extattr_set_link' function. */ +/* #undef HAVE_EXTATTR_SET_LINK */ + +/* Define to 1 if you have the `facl' function. */ +/* #undef HAVE_FACL */ + +/* Define to 1 if you have the `fchdir' function. */ +#define HAVE_FCHDIR 1 + +/* Define to 1 if you have the `fchflags' function. */ +/* #undef HAVE_FCHFLAGS */ + +/* Define to 1 if you have the `fchmod' function. */ +#define HAVE_FCHMOD 1 + +/* Define to 1 if you have the `fchown' function. */ +#define HAVE_FCHOWN 1 + +/* Define to 1 if you have the `fcntl' function. */ +#define HAVE_FCNTL 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_FCNTL_H 1 + +/* Define to 1 if you have the `fdopendir' function. */ +#define HAVE_FDOPENDIR 1 + +/* Define to 1 if you have the `fgetea' function. */ +/* #undef HAVE_FGETEA */ + +/* Define to 1 if you have the `fgetxattr' function. */ +/* #undef HAVE_FGETXATTR */ + +/* Define to 1 if you have the `flistea' function. */ +/* #undef HAVE_FLISTEA */ + +/* Define to 1 if you have the `flistxattr' function. */ +/* #undef HAVE_FLISTXATTR */ + +/* Define to 1 if you have the `fork' function. */ +#define HAVE_FORK 1 + +/* Define to 1 if fseeko (and presumably ftello) exists and is declared. */ +#define HAVE_FSEEKO 1 + +/* Define to 1 if you have the `fsetea' function. */ +/* #undef HAVE_FSETEA */ + +/* Define to 1 if you have the `fsetxattr' function. */ +/* #undef HAVE_FSETXATTR */ + +/* Define to 1 if you have the `fstat' function. */ +#define HAVE_FSTAT 1 + +/* Define to 1 if you have the `fstatat' function. */ +#define HAVE_FSTATAT 1 + +/* Define to 1 if you have the `fstatfs' function. */ +#define HAVE_FSTATFS 1 + +/* Define to 1 if you have the `fstatvfs' function. */ +#define HAVE_FSTATVFS 1 + +/* Define to 1 if you have the `ftruncate' function. */ +#define HAVE_FTRUNCATE 1 + +/* Define to 1 if you have the `futimens' function. */ +#define HAVE_FUTIMENS 1 + +/* Define to 1 if you have the `futimes' function. */ +#define HAVE_FUTIMES 1 + +/* Define to 1 if you have the `futimesat' function. */ +#define HAVE_FUTIMESAT 1 + +/* Define to 1 if you have the `getea' function. */ +/* #undef HAVE_GETEA */ + +/* Define to 1 if you have the `geteuid' function. */ +#define HAVE_GETEUID 1 + +/* Define to 1 if you have the `getgrgid_r' function. */ +#define HAVE_GETGRGID_R 1 + +/* Define to 1 if you have the `getgrnam_r' function. */ +#define HAVE_GETGRNAM_R 1 + +/* Define to 1 if you have the `getpid' function. */ +#define HAVE_GETPID 1 + +/* Define to 1 if you have the `getpwnam_r' function. */ +#define HAVE_GETPWNAM_R 1 + +/* Define to 1 if you have the `getpwuid_r' function. */ +#define HAVE_GETPWUID_R 1 + +/* Define to 1 if you have the `getvfsbyname' function. */ +/* #undef HAVE_GETVFSBYNAME */ + +/* Define to 1 if you have the `getxattr' function. */ +/* #undef HAVE_GETXATTR */ + +/* Define to 1 if you have the `gmtime_r' function. */ +#define HAVE_GMTIME_R 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_GRP_H 1 + +/* Define if you have the iconv() function and it works. */ +/* #undef HAVE_ICONV */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_ICONV_H */ + +/* Define to 1 if the system has the type `intmax_t'. */ +#define HAVE_INTMAX_T 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_INTTYPES_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_IO_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_LANGINFO_H 1 + +/* Define to 1 if you have the `lchflags' function. */ +/* #undef HAVE_LCHFLAGS */ + +/* Define to 1 if you have the `lchmod' function. */ +/* #undef HAVE_LCHMOD */ + +/* Define to 1 if you have the `lchown' function. */ +#define HAVE_LCHOWN 1 + +/* Define to 1 if you have the `lgetea' function. */ +/* #undef HAVE_LGETEA */ + +/* Define to 1 if you have the `lgetxattr' function. */ +/* #undef HAVE_LGETXATTR */ + +/* Define to 1 if you have the `acl' library (-lacl). */ +/* #undef HAVE_LIBACL */ + +/* Define to 1 if you have the `b2' library (-lb2). */ +/* #undef HAVE_LIBB2 */ + +/* Define to 1 if you have the `bz2' library (-lbz2). */ +/* #undef HAVE_LIBBZ2 */ + +/* Define to 1 if you have the `charset' library (-lcharset). */ +/* #undef HAVE_LIBCHARSET */ + +/* Define to 1 if you have the `crypto' library (-lcrypto). */ +/* #undef HAVE_LIBCRYPTO */ + +/* Define to 1 if you have the `eay32' library (-leay32). */ +/* #undef HAVE_LIBEAY32 */ + +/* Define to 1 if you have the `eay64' library (-leay64). */ +/* #undef HAVE_LIBEAY64 */ + +/* Define to 1 if you have the `expat' library (-lexpat). */ +/* #undef HAVE_LIBEXPAT */ + +/* Define to 1 if you have the `lz4' library (-llz4). */ +/* #undef HAVE_LIBLZ4 */ + +/* Define to 1 if you have the `lzma' library (-llzma). */ +/* #undef HAVE_LIBLZMA */ + +/* Define to 1 if you have the `lzo2' library (-llzo2). */ +/* #undef HAVE_LIBLZO2 */ + +/* Define to 1 if you have the `mbedcrypto' library (-lmbedcrypto). */ +/* #undef HAVE_LIBMBEDCRYPTO */ + +/* Define to 1 if you have the `md' library (-lmd). */ +/* #undef HAVE_LIBMD */ + +/* Define to 1 if you have the `nettle' library (-lnettle). */ +/* #undef HAVE_LIBNETTLE */ + +/* Define to 1 if you have the `pcre' library (-lpcre). */ +/* #undef HAVE_LIBPCRE */ + +/* Define to 1 if you have the `pcreposix' library (-lpcreposix). */ +/* #undef HAVE_LIBPCREPOSIX */ + +/* Define to 1 if you have the `regex' library (-lregex). */ +/* #undef HAVE_LIBREGEX */ + +/* Define to 1 if you have the `richacl' library (-lrichacl). */ +/* #undef HAVE_LIBRICHACL */ + +/* Define to 1 if you have the `xml2' library (-lxml2). */ +/* #undef HAVE_LIBXML2 */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LIBXML_XMLREADER_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LIBXML_XMLWRITER_H */ + +/* Define to 1 if you have the `z' library (-lz). */ +/* #undef HAVE_LIBZ */ + +/* Define to 1 if you have the `zstd' library (-lzstd). */ +/* #undef HAVE_LIBZSTD */ + +/* Define to 1 if you have the header file. */ +#define HAVE_LIMITS_H 1 + +/* Define to 1 if you have the `link' function. */ +#define HAVE_LINK 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_LINUX_FIEMAP_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_LINUX_FS_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_LINUX_MAGIC_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_LINUX_TYPES_H 1 + +/* Define to 1 if you have the `listea' function. */ +/* #undef HAVE_LISTEA */ + +/* Define to 1 if you have the `listxattr' function. */ +/* #undef HAVE_LISTXATTR */ + +/* Define to 1 if you have the `llistea' function. */ +/* #undef HAVE_LLISTEA */ + +/* Define to 1 if you have the `llistxattr' function. */ +/* #undef HAVE_LLISTXATTR */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LOCALCHARSET_H */ + +/* Define to 1 if you have the `locale_charset' function. */ +/* #undef HAVE_LOCALE_CHARSET */ + +/* Define to 1 if you have the header file. */ +#define HAVE_LOCALE_H 1 + +/* Define to 1 if you have the `localtime_r' function. */ +#define HAVE_LOCALTIME_R 1 + +/* Define to 1 if the system has the type `long long int'. */ +#define HAVE_LONG_LONG_INT 1 + +/* Define to 1 if you have the `lsetea' function. */ +/* #undef HAVE_LSETEA */ + +/* Define to 1 if you have the `lsetxattr' function. */ +/* #undef HAVE_LSETXATTR */ + +/* Define to 1 if you have the `lstat' function. */ +#define HAVE_LSTAT 1 + +/* Define to 1 if `lstat' has the bug that it succeeds when given the + zero-length file name argument. */ +/* #undef HAVE_LSTAT_EMPTY_STRING_BUG */ + +/* Define to 1 if you have the `lutimes' function. */ +#define HAVE_LUTIMES 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LZ4HC_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LZ4_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LZMA_H */ + +/* Define to 1 if you have the `lzma_stream_encoder_mt' function. */ +/* #undef HAVE_LZMA_STREAM_ENCODER_MT */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LZO_LZO1X_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LZO_LZOCONF_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_MBEDTLS_AES_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_MBEDTLS_MD_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_MBEDTLS_PKCS5_H */ + +/* Define to 1 if you have the `mbrtowc' function. */ +#define HAVE_MBRTOWC 1 + +/* Define to 1 if you have the `mbr_gid_to_uuid' function. */ +/* #undef HAVE_MBR_GID_TO_UUID */ + +/* Define to 1 if you have the `mbr_uid_to_uuid' function. */ +/* #undef HAVE_MBR_UID_TO_UUID */ + +/* Define to 1 if you have the `mbr_uuid_to_id' function. */ +/* #undef HAVE_MBR_UUID_TO_ID */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_MD5_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_MEMBERSHIP_H */ + +/* Define to 1 if you have the `memmove' function. */ +#define HAVE_MEMMOVE 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_MEMORY_H 1 + +/* Define to 1 if you have the `memset' function. */ +#define HAVE_MEMSET 1 + +/* Define to 1 if you have the `mkdir' function. */ +#define HAVE_MKDIR 1 + +/* Define to 1 if you have the `mkfifo' function. */ +#define HAVE_MKFIFO 1 + +/* Define to 1 if you have the `mknod' function. */ +#define HAVE_MKNOD 1 + +/* Define to 1 if you have the `mkstemp' function. */ +#define HAVE_MKSTEMP 1 + +/* Define to 1 if you have the header file, and it defines `DIR'. */ +/* #undef HAVE_NDIR_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NETTLE_AES_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NETTLE_HMAC_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NETTLE_MD5_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NETTLE_PBKDF2_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NETTLE_RIPEMD160_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NETTLE_SHA_H */ + +/* Define to 1 if you have the `nl_langinfo' function. */ +#define HAVE_NL_LANGINFO 1 + +/* Define to 1 if you have the `openat' function. */ +#define HAVE_OPENAT 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_OPENSSL_EVP_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_PATHS_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_PCREPOSIX_H */ + +/* Define to 1 if you have the `pipe' function. */ +#define HAVE_PIPE 1 + +/* Define to 1 if you have the `PKCS5_PBKDF2_HMAC_SHA1' function. */ +/* #undef HAVE_PKCS5_PBKDF2_HMAC_SHA1 */ + +/* Define to 1 if you have the `poll' function. */ +#define HAVE_POLL 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_POLL_H 1 + +/* Define to 1 if you have the `posix_spawnp' function. */ +#define HAVE_POSIX_SPAWNP 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_PTHREAD_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_PWD_H 1 + +/* Define to 1 if you have a POSIX compatible readdir_r */ +#define HAVE_READDIR_R 1 + +/* Define to 1 if you have the `readlink' function. */ +#define HAVE_READLINK 1 + +/* Define to 1 if you have the `readlinkat' function. */ +#define HAVE_READLINKAT 1 + +/* Define to 1 if you have the `readpassphrase' function. */ +/* #undef HAVE_READPASSPHRASE */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_READPASSPHRASE_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_REGEX_H */ + +/* Define to 1 if you have the `richacl_alloc' function. */ +/* #undef HAVE_RICHACL_ALLOC */ + +/* Define to 1 if you have the `richacl_equiv_mode' function. */ +/* #undef HAVE_RICHACL_EQUIV_MODE */ + +/* Define to 1 if you have the `richacl_free' function. */ +/* #undef HAVE_RICHACL_FREE */ + +/* Define to 1 if you have the `richacl_get_fd' function. */ +/* #undef HAVE_RICHACL_GET_FD */ + +/* Define to 1 if you have the `richacl_get_file' function. */ +/* #undef HAVE_RICHACL_GET_FILE */ + +/* Define to 1 if you have the `richacl_set_fd' function. */ +/* #undef HAVE_RICHACL_SET_FD */ + +/* Define to 1 if you have the `richacl_set_file' function. */ +/* #undef HAVE_RICHACL_SET_FILE */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_RIPEMD_H */ + +/* Define to 1 if you have the `select' function. */ +#define HAVE_SELECT 1 + +/* Define to 1 if you have the `setenv' function. */ +#define HAVE_SETENV 1 + +/* Define to 1 if you have the `setlocale' function. */ +#define HAVE_SETLOCALE 1 + +/* Define to 1 if you have the `setxattr' function. */ +/* #undef HAVE_SETXATTR */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SHA256_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SHA512_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SHA_H */ + +/* Define to 1 if you have the `sigaction' function. */ +#define HAVE_SIGACTION 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SIGNAL_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SPAWN_H 1 + +/* Define to 1 if you have the `statfs' function. */ +#define HAVE_STATFS 1 + +/* Define to 1 if you have the `statvfs' function. */ +#define HAVE_STATVFS 1 + +/* Define to 1 if `stat' has the bug that it succeeds when given the + zero-length file name argument. */ +/* #undef HAVE_STAT_EMPTY_STRING_BUG */ + +/* Define to 1 if you have the header file. */ +#define HAVE_STDARG_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_STDINT_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_STDLIB_H 1 + +/* Define to 1 if you have the `strchr' function. */ +#define HAVE_STRCHR 1 + +/* Define to 1 if you have the `strdup' function. */ +#define HAVE_STRDUP 1 + +/* Define to 1 if you have the `strerror' function. */ +#define HAVE_STRERROR 1 + +/* Define to 1 if you have the `strerror_r' function. */ +#define HAVE_STRERROR_R 1 + +/* Define to 1 if you have the `strftime' function. */ +#define HAVE_STRFTIME 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_STRINGS_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_STRING_H 1 + +/* Define to 1 if you have the `strncpy_s' function. */ +/* #undef HAVE_STRNCPY_S */ + +/* Define to 1 if you have the `strrchr' function. */ +#define HAVE_STRRCHR 1 + +/* Define to 1 if the system has the type `struct richace'. */ +/* #undef HAVE_STRUCT_RICHACE */ + +/* Define to 1 if the system has the type `struct richacl'. */ +/* #undef HAVE_STRUCT_RICHACL */ + +/* Define to 1 if `f_namemax' is a member of `struct statfs'. */ +/* #undef HAVE_STRUCT_STATFS_F_NAMEMAX */ + +/* Define to 1 if `f_iosize' is a member of `struct statvfs'. */ +/* #undef HAVE_STRUCT_STATVFS_F_IOSIZE */ + +/* Define to 1 if `st_birthtime' is a member of `struct stat'. */ +/* #undef HAVE_STRUCT_STAT_ST_BIRTHTIME */ + +/* Define to 1 if `st_birthtimespec.tv_nsec' is a member of `struct stat'. */ +/* #undef HAVE_STRUCT_STAT_ST_BIRTHTIMESPEC_TV_NSEC */ + +/* Define to 1 if `st_blksize' is a member of `struct stat'. */ +#define HAVE_STRUCT_STAT_ST_BLKSIZE 1 + +/* Define to 1 if `st_flags' is a member of `struct stat'. */ +/* #undef HAVE_STRUCT_STAT_ST_FLAGS */ + +/* Define to 1 if `st_mtimespec.tv_nsec' is a member of `struct stat'. */ +/* #undef HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC */ + +/* Define to 1 if `st_mtime_n' is a member of `struct stat'. */ +/* #undef HAVE_STRUCT_STAT_ST_MTIME_N */ + +/* Define to 1 if `st_mtime_usec' is a member of `struct stat'. */ +/* #undef HAVE_STRUCT_STAT_ST_MTIME_USEC */ + +/* Define to 1 if `st_mtim.tv_nsec' is a member of `struct stat'. */ +#define HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC 1 + +/* Define to 1 if `st_umtime' is a member of `struct stat'. */ +/* #undef HAVE_STRUCT_STAT_ST_UMTIME */ + +/* Define to 1 if `tm_gmtoff' is a member of `struct tm'. */ +#define HAVE_STRUCT_TM_TM_GMTOFF 1 + +/* Define to 1 if `__tm_gmtoff' is a member of `struct tm'. */ +/* #undef HAVE_STRUCT_TM___TM_GMTOFF */ + +/* Define to 1 if the system has the type `struct vfsconf'. */ +/* #undef HAVE_STRUCT_VFSCONF */ + +/* Define to 1 if the system has the type `struct xvfsconf'. */ +/* #undef HAVE_STRUCT_XVFSCONF */ + +/* Define to 1 if you have the `symlink' function. */ +#define HAVE_SYMLINK 1 + +/* Define to 1 if you have the header file. */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_CDEFS_H 1 + +/* Define to 1 if you have the header file, and it defines `DIR'. + */ +/* #undef HAVE_SYS_DIR_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_EA_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_EXTATTR_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_IOCTL_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_MKDEV_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_MOUNT_H 1 + +/* Define to 1 if you have the header file, and it defines `DIR'. + */ +/* #undef HAVE_SYS_NDIR_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_PARAM_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_POLL_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_RICHACL_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_SELECT_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_STATFS_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_STATVFS_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_STAT_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_SYSMACROS_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_TIME_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_TYPES_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_UTIME_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_UTSNAME_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_VFS_H 1 + +/* Define to 1 if you have that is POSIX.1 compatible. */ +#define HAVE_SYS_WAIT_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_XATTR_H 1 + +/* Define to 1 if you have the `timegm' function. */ +#define HAVE_TIMEGM 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_TIME_H 1 + +/* Define to 1 if you have the `tzset' function. */ +#define HAVE_TZSET 1 + +/* Define to 1 if the system has the type `uintmax_t'. */ +#define HAVE_UINTMAX_T 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_UNISTD_H 1 + +/* Define to 1 if you have the `unlinkat' function. */ +#define HAVE_UNLINKAT 1 + +/* Define to 1 if you have the `unsetenv' function. */ +#define HAVE_UNSETENV 1 + +/* Define to 1 if the system has the type `unsigned long long'. */ +#define HAVE_UNSIGNED_LONG_LONG 1 + +/* Define to 1 if the system has the type `unsigned long long int'. */ +#define HAVE_UNSIGNED_LONG_LONG_INT 1 + +/* Define to 1 if you have the `utime' function. */ +#define HAVE_UTIME 1 + +/* Define to 1 if you have the `utimensat' function. */ +#define HAVE_UTIMENSAT 1 + +/* Define to 1 if you have the `utimes' function. */ +#define HAVE_UTIMES 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_UTIME_H 1 + +/* Define to 1 if you have the `vfork' function. */ +#define HAVE_VFORK 1 + +/* Define to 1 if you have the `vprintf' function. */ +#define HAVE_VPRINTF 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_WCHAR_H 1 + +/* Define to 1 if the system has the type `wchar_t'. */ +#define HAVE_WCHAR_T 1 + +/* Define to 1 if you have the `wcrtomb' function. */ +#define HAVE_WCRTOMB 1 + +/* Define to 1 if you have the `wcscmp' function. */ +#define HAVE_WCSCMP 1 + +/* Define to 1 if you have the `wcscpy' function. */ +#define HAVE_WCSCPY 1 + +/* Define to 1 if you have the `wcslen' function. */ +#define HAVE_WCSLEN 1 + +/* Define to 1 if you have the `wctomb' function. */ +#define HAVE_WCTOMB 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_WCTYPE_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_WINCRYPT_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_WINDOWS_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_WINIOCTL_H */ + +/* Define to 1 if you have the `wmemcmp' function. */ +#define HAVE_WMEMCMP 1 + +/* Define to 1 if you have the `wmemcpy' function. */ +#define HAVE_WMEMCPY 1 + +/* Define to 1 if you have the `wmemmove' function. */ +#define HAVE_WMEMMOVE 1 + +/* Define to 1 if you have a working EXT2_IOC_GETFLAGS */ +#define HAVE_WORKING_EXT2_IOC_GETFLAGS 1 + +/* Define to 1 if you have a working FS_IOC_GETFLAGS */ +#define HAVE_WORKING_FS_IOC_GETFLAGS 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_ZLIB_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_ZSTD_H */ + +/* Define to 1 if you have the `_ctime64_s' function. */ +/* #undef HAVE__CTIME64_S */ + +/* Define to 1 if you have the `_fseeki64' function. */ +/* #undef HAVE__FSEEKI64 */ + +/* Define to 1 if you have the `_get_timezone' function. */ +/* #undef HAVE__GET_TIMEZONE */ + +/* Define to 1 if you have the `_gmtime64_s' function. */ +/* #undef HAVE__GMTIME64_S */ + +/* Define to 1 if you have the `_localtime64_s' function. */ +/* #undef HAVE__LOCALTIME64_S */ + +/* Define to 1 if you have the `_mkgmtime64' function. */ +/* #undef HAVE__MKGMTIME64 */ + +/* Define as const if the declaration of iconv() needs const. */ +/* #undef ICONV_CONST */ + +/* Version number of libarchive as a single integer */ +#define LIBARCHIVE_VERSION_NUMBER "3004003" + +/* Version number of libarchive */ +#define LIBARCHIVE_VERSION_STRING "3.4.3" + +/* Define to 1 if `lstat' dereferences a symlink specified with a trailing + slash. */ +#define LSTAT_FOLLOWS_SLASHED_SYMLINK 1 + +/* Define to the sub-directory where libtool stores uninstalled libraries. */ +#define LT_OBJDIR ".libs/" + +/* Define to 1 if `major', `minor', and `makedev' are declared in . + */ +/* #undef MAJOR_IN_MKDEV */ + +/* Define to 1 if `major', `minor', and `makedev' are declared in + . */ +#define MAJOR_IN_SYSMACROS 1 + +/* Define to '0x05020000' for Windows Server 2003 APIs. */ +/* #undef NTDDI_VERSION */ + +/* Name of package */ +#define PACKAGE "libarchive" + +/* Define to the address where bug reports for this package should be sent. */ +#define PACKAGE_BUGREPORT "libarchive-discuss@googlegroups.com" + +/* Define to the full name of this package. */ +#define PACKAGE_NAME "libarchive" + +/* Define to the full name and version of this package. */ +#define PACKAGE_STRING "libarchive 3.4.3" + +/* Define to the one symbol short name of this package. */ +#define PACKAGE_TARNAME "libarchive" + +/* Define to the home page for this package. */ +#define PACKAGE_URL "" + +/* Define to the version of this package. */ +#define PACKAGE_VERSION "3.4.3" + +/* Define to 1 if PCRE_STATIC needs to be defined. */ +/* #undef PCRE_STATIC */ + +/* The size of `wchar_t', as computed by sizeof. */ +#define SIZEOF_WCHAR_T 4 + +/* Define to 1 if you have the ANSI C header files. */ +#define STDC_HEADERS 1 + +/* Define to 1 if strerror_r returns char *. */ +#define STRERROR_R_CHAR_P 1 + +/* Define to 1 if you can safely include both and . */ +#define TIME_WITH_SYS_TIME 1 + +/* Enable extensions on AIX 3, Interix. */ +#ifndef _ALL_SOURCE +# define _ALL_SOURCE 1 +#endif +/* Enable GNU extensions on systems that have them. */ +#ifndef _GNU_SOURCE +# define _GNU_SOURCE 1 +#endif +/* Enable threading extensions on Solaris. */ +#ifndef _POSIX_PTHREAD_SEMANTICS +# define _POSIX_PTHREAD_SEMANTICS 1 +#endif +/* Enable extensions on HP NonStop. */ +#ifndef _TANDEM_SOURCE +# define _TANDEM_SOURCE 1 +#endif +/* Enable general extensions on Solaris. */ +#ifndef __EXTENSIONS__ +# define __EXTENSIONS__ 1 +#endif + + +/* Version number of package */ +#define VERSION "3.4.3" + +/* Define to '0x0502' for Windows Server 2003 APIs. */ +/* #undef WINVER */ + +/* Enable large inode numbers on Mac OS X 10.5. */ +#ifndef _DARWIN_USE_64_BIT_INODE +# define _DARWIN_USE_64_BIT_INODE 1 +#endif + +/* Number of bits in a file offset, on hosts where this is settable. */ +/* #undef _FILE_OFFSET_BITS */ + +/* Define to 1 to make fseeko visible on some hosts (e.g. glibc 2.2). */ +/* #undef _LARGEFILE_SOURCE */ + +/* Define for large files, on AIX-style hosts. */ +/* #undef _LARGE_FILES */ + +/* Define to 1 if on MINIX. */ +/* #undef _MINIX */ + +/* Define to 2 if the system does not provide POSIX.1 features except with + this defined. */ +/* #undef _POSIX_1_SOURCE */ + +/* Define to 1 if you need to in order for `stat' and other things to work. */ +/* #undef _POSIX_SOURCE */ + +/* Define for Solaris 2.5.1 so the uint32_t typedef from , + , or is not used. If the typedef were allowed, the + #define below would cause a syntax error. */ +/* #undef _UINT32_T */ + +/* Define for Solaris 2.5.1 so the uint64_t typedef from , + , or is not used. If the typedef were allowed, the + #define below would cause a syntax error. */ +/* #undef _UINT64_T */ + +/* Define for Solaris 2.5.1 so the uint8_t typedef from , + , or is not used. If the typedef were allowed, the + #define below would cause a syntax error. */ +/* #undef _UINT8_T */ + +/* Define to '0x0502' for Windows Server 2003 APIs. */ +/* #undef _WIN32_WINNT */ + +/* Define to empty if `const' does not conform to ANSI C. */ +/* #undef const */ + +/* Define to match typeof st_gid field of struct stat if doesn't + define. */ +/* #undef gid_t */ + +/* Define to `unsigned long' if does not define. */ +/* #undef id_t */ + +/* Define to the type of a signed integer type of width exactly 16 bits if + such a type exists and the standard includes do not define it. */ +/* #undef int16_t */ + +/* Define to the type of a signed integer type of width exactly 32 bits if + such a type exists and the standard includes do not define it. */ +/* #undef int32_t */ + +/* Define to the type of a signed integer type of width exactly 64 bits if + such a type exists and the standard includes do not define it. */ +/* #undef int64_t */ + +/* Define to the widest signed integer type if and do + not define. */ +/* #undef intmax_t */ + +/* Define to `int' if does not define. */ +/* #undef mode_t */ + +/* Define to `long long' if does not define. */ +/* #undef off_t */ + +/* Define to `unsigned int' if does not define. */ +/* #undef size_t */ + +/* Define to match typeof st_uid field of struct stat if doesn't + define. */ +/* #undef uid_t */ + +/* Define to the type of an unsigned integer type of width exactly 16 bits if + such a type exists and the standard includes do not define it. */ +/* #undef uint16_t */ + +/* Define to the type of an unsigned integer type of width exactly 32 bits if + such a type exists and the standard includes do not define it. */ +/* #undef uint32_t */ + +/* Define to the type of an unsigned integer type of width exactly 64 bits if + such a type exists and the standard includes do not define it. */ +/* #undef uint64_t */ + +/* Define to the type of an unsigned integer type of width exactly 8 bits if + such a type exists and the standard includes do not define it. */ +/* #undef uint8_t */ + +/* Define to the widest unsigned integer type if and + do not define. */ +/* #undef uintmax_t */ + +/* Define to `unsigned int' if does not define. */ +/* #undef uintptr_t */ diff --git a/c/config_freebsd.h b/c/config_freebsd.h new file mode 100644 --- /dev/null +++ b/c/config_freebsd.h @@ -0,0 +1,260 @@ +/*- + * Copyright (c) 2003-2007 Tim Kientzle + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD$ + */ + +#include + +/* FreeBSD 5.0 and later has ACL and extattr support. */ +#if __FreeBSD__ > 4 +#define ARCHIVE_ACL_FREEBSD 1 +#define HAVE_ACL_GET_PERM_NP 1 +#define HAVE_ARC4RANDOM_BUF 1 +#define HAVE_EXTATTR_GET_FILE 1 +#define HAVE_EXTATTR_LIST_FILE 1 +#define HAVE_EXTATTR_SET_FD 1 +#define HAVE_EXTATTR_SET_FILE 1 +#define HAVE_STRUCT_XVFSCONF 1 +#define HAVE_SYS_ACL_H 1 +#define HAVE_SYS_EXTATTR_H 1 +#if __FreeBSD__ > 7 +/* FreeBSD 8.0 and later has NFSv4 ACL support */ +#define ARCHIVE_ACL_FREEBSD_NFS4 1 +#define HAVE_ACL_GET_LINK_NP 1 +#define HAVE_ACL_IS_TRIVIAL_NP 1 +#define HAVE_ACL_SET_LINK_NP 1 +#endif /* __FreeBSD__ > 7 */ +#endif /* __FreeBSD__ > 4 */ + +#ifdef WITH_OPENSSL +#define HAVE_LIBCRYPTO 1 +#define HAVE_OPENSSL_EVP_H 1 +#define HAVE_OPENSSL_MD5_H 1 +#define HAVE_OPENSSL_RIPEMD_H 1 +#define HAVE_OPENSSL_SHA_H 1 +#define HAVE_OPENSSL_SHA256_INIT 1 +#define HAVE_OPENSSL_SHA384_INIT 1 +#define HAVE_OPENSSL_SHA512_INIT 1 +#define HAVE_PKCS5_PBKDF2_HMAC_SHA1 1 +#define HAVE_SHA256 1 +#define HAVE_SHA384 1 +#define HAVE_SHA512 1 +#else +#define HAVE_LIBMD 1 +#define HAVE_MD5_H 1 +#define HAVE_MD5INIT 1 +#define HAVE_RIPEMD_H 1 +#define HAVE_SHA_H 1 +#define HAVE_SHA1 1 +#define HAVE_SHA1_INIT 1 +#define HAVE_SHA256 1 +#define HAVE_SHA256_H 1 +#define HAVE_SHA256_INIT 1 +#define HAVE_SHA512 1 +#define HAVE_SHA512_H 1 +#define HAVE_SHA512_INIT 1 +#endif + +#define HAVE_BSDXML_H 1 +#define HAVE_BZLIB_H 1 +#define HAVE_CHFLAGS 1 +#define HAVE_CHOWN 1 +#define HAVE_CHROOT 1 +#define HAVE_CTIME_R 1 +#define HAVE_CTYPE_H 1 +#define HAVE_DECL_EXTATTR_NAMESPACE_USER 1 +#define HAVE_DECL_INT32_MAX 1 +#define HAVE_DECL_INT32_MIN 1 +#define HAVE_DECL_INT64_MAX 1 +#define HAVE_DECL_INT64_MIN 1 +#define HAVE_DECL_INTMAX_MAX 1 +#define HAVE_DECL_INTMAX_MIN 1 +#define HAVE_DECL_SIZE_MAX 1 +#define HAVE_DECL_SSIZE_MAX 1 +#define HAVE_DECL_STRERROR_R 1 +#define HAVE_DECL_UINT32_MAX 1 +#define HAVE_DECL_UINT64_MAX 1 +#define HAVE_DECL_UINTMAX_MAX 1 +#define HAVE_DIRENT_H 1 +#define HAVE_DLFCN_H 1 +#define HAVE_D_MD_ORDER 1 +#define HAVE_EFTYPE 1 +#define HAVE_EILSEQ 1 +#define HAVE_ERRNO_H 1 +#define HAVE_FCHDIR 1 +#define HAVE_FCHFLAGS 1 +#define HAVE_FCHMOD 1 +#define HAVE_FCHOWN 1 +#define HAVE_FCNTL 1 +#define HAVE_FCNTL_H 1 +#define HAVE_FDOPENDIR 1 +#define HAVE_FORK 1 +#define HAVE_FSEEKO 1 +#define HAVE_FSTAT 1 +#define HAVE_FSTATAT 1 +#define HAVE_FSTATFS 1 +#define HAVE_FSTATVFS 1 +#define HAVE_FTRUNCATE 1 +#define HAVE_FUTIMES 1 +#define HAVE_FUTIMESAT 1 +#define HAVE_GETEUID 1 +#define HAVE_GETGRGID_R 1 +#define HAVE_GETGRNAM_R 1 +#define HAVE_GETPID 1 +#define HAVE_GETPWNAM_R 1 +#define HAVE_GETPWUID_R 1 +#define HAVE_GETVFSBYNAME 1 +#define HAVE_GMTIME_R 1 +#define HAVE_GRP_H 1 +#define HAVE_INTMAX_T 1 +#define HAVE_INTTYPES_H 1 +#define HAVE_LANGINFO_H 1 +#define HAVE_LCHFLAGS 1 +#define HAVE_LCHMOD 1 +#define HAVE_LCHOWN 1 +#define HAVE_LIBZ 1 +#define HAVE_LIMITS_H 1 +#define HAVE_LINK 1 +#define HAVE_LOCALE_H 1 +#define HAVE_LOCALTIME_R 1 +#define HAVE_LONG_LONG_INT 1 +#define HAVE_LSTAT 1 +#define HAVE_LUTIMES 1 +#define HAVE_MBRTOWC 1 +#define HAVE_MEMMOVE 1 +#define HAVE_MEMORY_H 1 +#define HAVE_MEMSET 1 +#define HAVE_MKDIR 1 +#define HAVE_MKFIFO 1 +#define HAVE_MKNOD 1 +#define HAVE_MKSTEMP 1 +#define HAVE_NL_LANGINFO 1 +#define HAVE_OPENAT 1 +#define HAVE_PATHS_H 1 +#define HAVE_PIPE 1 +#define HAVE_POLL 1 +#define HAVE_POLL_H 1 +#define HAVE_POSIX_SPAWNP 1 +#define HAVE_PTHREAD_H 1 +#define HAVE_PWD_H 1 +#define HAVE_READDIR_R 1 +#define HAVE_READLINK 1 +#define HAVE_READLINKAT 1 +#define HAVE_READPASSPHRASE 1 +#define HAVE_READPASSPHRASE_H 1 +#define HAVE_REGEX_H 1 +#define HAVE_SELECT 1 +#define HAVE_SETENV 1 +#define HAVE_SETLOCALE 1 +#define HAVE_SIGACTION 1 +#define HAVE_SIGNAL_H 1 +#define HAVE_SPAWN_H 1 +#define HAVE_STATFS 1 +#define HAVE_STATVFS 1 +#define HAVE_STDARG_H 1 +#define HAVE_STDINT_H 1 +#define HAVE_STDLIB_H 1 +#define HAVE_STRCHR 1 +#define HAVE_STRDUP 1 +#define HAVE_STRERROR 1 +#define HAVE_STRERROR_R 1 +#define HAVE_STRFTIME 1 +#define HAVE_STRINGS_H 1 +#define HAVE_STRING_H 1 +#define HAVE_STRRCHR 1 +#define HAVE_STRUCT_STATFS_F_NAMEMAX 1 +#define HAVE_STRUCT_STAT_ST_BIRTHTIME 1 +#define HAVE_STRUCT_STAT_ST_BIRTHTIMESPEC_TV_NSEC 1 +#define HAVE_STRUCT_STAT_ST_BLKSIZE 1 +#define HAVE_STRUCT_STAT_ST_FLAGS 1 +#define HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC 1 +#define HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC 1 +#define HAVE_STRUCT_TM_TM_GMTOFF 1 +#define HAVE_SYMLINK 1 +#define HAVE_SYS_CDEFS_H 1 +#define HAVE_SYS_IOCTL_H 1 +#define HAVE_SYS_MOUNT_H 1 +#define HAVE_SYS_PARAM_H 1 +#define HAVE_SYS_POLL_H 1 +#define HAVE_SYS_SELECT_H 1 +#define HAVE_SYS_STATVFS_H 1 +#define HAVE_SYS_STAT_H 1 +#define HAVE_SYS_TIME_H 1 +#define HAVE_SYS_TYPES_H 1 +#define HAVE_SYS_UTSNAME_H 1 +#define HAVE_SYS_WAIT_H 1 +#define HAVE_TIMEGM 1 +#define HAVE_TIME_H 1 +#define HAVE_TZSET 1 +#define HAVE_UINTMAX_T 1 +#define HAVE_UNISTD_H 1 +#define HAVE_UNLINKAT 1 +#define HAVE_UNSETENV 1 +#define HAVE_UNSIGNED_LONG_LONG 1 +#define HAVE_UNSIGNED_LONG_LONG_INT 1 +#define HAVE_UTIME 1 +#define HAVE_UTIMES 1 +#define HAVE_UTIME_H 1 +#define HAVE_VFORK 1 +#define HAVE_VPRINTF 1 +#define HAVE_WCHAR_H 1 +#define HAVE_WCHAR_T 1 +#define HAVE_WCRTOMB 1 +#define HAVE_WCSCMP 1 +#define HAVE_WCSCPY 1 +#define HAVE_WCSLEN 1 +#define HAVE_WCTOMB 1 +#define HAVE_WCTYPE_H 1 +#define HAVE_WMEMCMP 1 +#define HAVE_WMEMCPY 1 +#define HAVE_WMEMMOVE 1 +#define HAVE_ZLIB_H 1 +#define TIME_WITH_SYS_TIME 1 + +#if __FreeBSD_version >= 1100056 +#define HAVE_FUTIMENS 1 +#define HAVE_UTIMENSAT 1 +#endif + +/* FreeBSD 4 and earlier lack intmax_t/uintmax_t */ +#if __FreeBSD__ < 5 +#define intmax_t int64_t +#define uintmax_t uint64_t +#endif + +/* FreeBSD defines for archive_hash.h */ +#ifdef WITH_OPENSSL +#define ARCHIVE_CRYPTO_MD5_OPENSSL 1 +#define ARCHIVE_CRYPTO_RMD160_OPENSSL 1 +#define ARCHIVE_CRYPTO_SHA1_OPENSSL +#define ARCHIVE_CRYPTO_SHA256_OPENSSL 1 +#define ARCHIVE_CRYPTO_SHA384_OPENSSL 1 +#define ARCHIVE_CRYPTO_SHA512_OPENSSL 1 +#else +#define ARCHIVE_CRYPTO_MD5_LIBMD 1 +#define ARCHIVE_CRYPTO_SHA1_LIBMD 1 +#define ARCHIVE_CRYPTO_SHA256_LIBMD 1 +#define ARCHIVE_CRYPTO_SHA512_LIBMD 1 +#endif diff --git a/c/filter_fork.h b/c/filter_fork.h new file mode 100644 --- /dev/null +++ b/c/filter_fork.h @@ -0,0 +1,46 @@ +/*- + * Copyright (c) 2007 Joerg Sonnenberger + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD: head/lib/libarchive/filter_fork.h 201087 2009-12-28 02:18:26Z kientzle $ + */ + +#ifndef FILTER_FORK_H +#define FILTER_FORK_H + +#ifndef __LIBARCHIVE_BUILD +#error This header is only to be used internally to libarchive. +#endif + +int +__archive_create_child(const char *cmd, int *child_stdin, int *child_stdout, +#if defined(_WIN32) && !defined(__CYGWIN__) + HANDLE *out_child); +#else + pid_t *out_child); +#endif + +void +__archive_check_child(int in, int out); + +#endif diff --git a/c/filter_fork_posix.c b/c/filter_fork_posix.c new file mode 100644 --- /dev/null +++ b/c/filter_fork_posix.c @@ -0,0 +1,240 @@ +/*- + * Copyright (c) 2007 Joerg Sonnenberger + * Copyright (c) 2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +/* This capability is only available on POSIX systems. */ +#if defined(HAVE_PIPE) && defined(HAVE_FCNTL) && \ + (defined(HAVE_FORK) || defined(HAVE_VFORK) || defined(HAVE_POSIX_SPAWNP)) + +__FBSDID("$FreeBSD: head/lib/libarchive/filter_fork.c 182958 2008-09-12 05:33:00Z kientzle $"); + +#if defined(HAVE_SYS_TYPES_H) +# include +#endif +#ifdef HAVE_ERRNO_H +# include +#endif +#ifdef HAVE_STRING_H +# include +#endif +#if defined(HAVE_POLL) && (defined(HAVE_POLL_H) || defined(HAVE_SYS_POLL_H)) +# if defined(HAVE_POLL_H) +# include +# elif defined(HAVE_SYS_POLL_H) +# include +# endif +#elif defined(HAVE_SELECT) +# if defined(HAVE_SYS_SELECT_H) +# include +# elif defined(HAVE_UNISTD_H) +# include +# endif +#endif +#ifdef HAVE_FCNTL_H +# include +#endif +#ifdef HAVE_SPAWN_H +# include +#endif +#ifdef HAVE_STDLIB_H +# include +#endif +#ifdef HAVE_UNISTD_H +# include +#endif + +#include "archive.h" +#include "archive_cmdline_private.h" + +#include "filter_fork.h" + +int +__archive_create_child(const char *cmd, int *child_stdin, int *child_stdout, + pid_t *out_child) +{ + pid_t child; + int stdin_pipe[2], stdout_pipe[2], tmp; +#if HAVE_POSIX_SPAWNP + posix_spawn_file_actions_t actions; + int r; +#endif + struct archive_cmdline *cmdline; + + cmdline = __archive_cmdline_allocate(); + if (cmdline == NULL) + goto state_allocated; + if (__archive_cmdline_parse(cmdline, cmd) != ARCHIVE_OK) + goto state_allocated; + + if (pipe(stdin_pipe) == -1) + goto state_allocated; + if (stdin_pipe[0] == 1 /* stdout */) { + if ((tmp = dup(stdin_pipe[0])) == -1) + goto stdin_opened; + close(stdin_pipe[0]); + stdin_pipe[0] = tmp; + } + if (pipe(stdout_pipe) == -1) + goto stdin_opened; + if (stdout_pipe[1] == 0 /* stdin */) { + if ((tmp = dup(stdout_pipe[1])) == -1) + goto stdout_opened; + close(stdout_pipe[1]); + stdout_pipe[1] = tmp; + } + +#if HAVE_POSIX_SPAWNP + + r = posix_spawn_file_actions_init(&actions); + if (r != 0) { + errno = r; + goto stdout_opened; + } + r = posix_spawn_file_actions_addclose(&actions, stdin_pipe[1]); + if (r != 0) + goto actions_inited; + r = posix_spawn_file_actions_addclose(&actions, stdout_pipe[0]); + if (r != 0) + goto actions_inited; + /* Setup for stdin. */ + r = posix_spawn_file_actions_adddup2(&actions, stdin_pipe[0], 0); + if (r != 0) + goto actions_inited; + if (stdin_pipe[0] != 0 /* stdin */) { + r = posix_spawn_file_actions_addclose(&actions, stdin_pipe[0]); + if (r != 0) + goto actions_inited; + } + /* Setup for stdout. */ + r = posix_spawn_file_actions_adddup2(&actions, stdout_pipe[1], 1); + if (r != 0) + goto actions_inited; + if (stdout_pipe[1] != 1 /* stdout */) { + r = posix_spawn_file_actions_addclose(&actions, stdout_pipe[1]); + if (r != 0) + goto actions_inited; + } + r = posix_spawnp(&child, cmdline->path, &actions, NULL, + cmdline->argv, NULL); + if (r != 0) + goto actions_inited; + posix_spawn_file_actions_destroy(&actions); + +#else /* HAVE_POSIX_SPAWNP */ + +#if HAVE_VFORK + child = vfork(); +#else + child = fork(); +#endif + if (child == -1) + goto stdout_opened; + if (child == 0) { + close(stdin_pipe[1]); + close(stdout_pipe[0]); + if (dup2(stdin_pipe[0], 0 /* stdin */) == -1) + _exit(254); + if (stdin_pipe[0] != 0 /* stdin */) + close(stdin_pipe[0]); + if (dup2(stdout_pipe[1], 1 /* stdout */) == -1) + _exit(254); + if (stdout_pipe[1] != 1 /* stdout */) + close(stdout_pipe[1]); + execvp(cmdline->path, cmdline->argv); + _exit(254); + } +#endif /* HAVE_POSIX_SPAWNP */ + + close(stdin_pipe[0]); + close(stdout_pipe[1]); + + *child_stdin = stdin_pipe[1]; + fcntl(*child_stdin, F_SETFL, O_NONBLOCK); + *child_stdout = stdout_pipe[0]; + fcntl(*child_stdout, F_SETFL, O_NONBLOCK); + __archive_cmdline_free(cmdline); + + *out_child = child; + return ARCHIVE_OK; + +#if HAVE_POSIX_SPAWNP +actions_inited: + errno = r; + posix_spawn_file_actions_destroy(&actions); +#endif +stdout_opened: + close(stdout_pipe[0]); + close(stdout_pipe[1]); +stdin_opened: + close(stdin_pipe[0]); + close(stdin_pipe[1]); +state_allocated: + __archive_cmdline_free(cmdline); + return ARCHIVE_FAILED; +} + +void +__archive_check_child(int in, int out) +{ +#if defined(HAVE_POLL) && (defined(HAVE_POLL_H) || defined(HAVE_SYS_POLL_H)) + struct pollfd fds[2]; + int idx; + + idx = 0; + if (in != -1) { + fds[idx].fd = in; + fds[idx].events = POLLOUT; + ++idx; + } + if (out != -1) { + fds[idx].fd = out; + fds[idx].events = POLLIN; + ++idx; + } + + poll(fds, idx, -1); /* -1 == INFTIM, wait forever */ +#elif defined(HAVE_SELECT) + fd_set fds_in, fds_out, fds_error; + + FD_ZERO(&fds_in); + FD_ZERO(&fds_out); + FD_ZERO(&fds_error); + if (out != -1) { + FD_SET(out, &fds_in); + FD_SET(out, &fds_error); + } + if (in != -1) { + FD_SET(in, &fds_out); + FD_SET(in, &fds_error); + } + select(in < out ? out + 1 : in + 1, &fds_in, &fds_out, &fds_error, NULL); +#else + sleep(1); +#endif +} + +#endif /* defined(HAVE_PIPE) && defined(HAVE_VFORK) && defined(HAVE_FCNTL) */ diff --git a/c/filter_fork_windows.c b/c/filter_fork_windows.c new file mode 100644 --- /dev/null +++ b/c/filter_fork_windows.c @@ -0,0 +1,199 @@ +/*- + * Copyright (c) 2009-2012 Michihiro NAKAJIMA + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include "archive_platform.h" + +#if defined(_WIN32) && !defined(__CYGWIN__) +#include "archive_cmdline_private.h" +#include "archive_string.h" + +#include "filter_fork.h" + +int +__archive_create_child(const char *cmd, int *child_stdin, int *child_stdout, + HANDLE *out_child) +{ + HANDLE childStdout[2], childStdin[2],childStderr; + SECURITY_ATTRIBUTES secAtts; + STARTUPINFOA staInfo; + PROCESS_INFORMATION childInfo; + struct archive_string cmdline; + struct archive_string fullpath; + struct archive_cmdline *acmd; + char *arg0, *ext; + int i, l; + DWORD fl, fl_old; + HANDLE child; + + childStdout[0] = childStdout[1] = INVALID_HANDLE_VALUE; + childStdin[0] = childStdin[1] = INVALID_HANDLE_VALUE; + childStderr = INVALID_HANDLE_VALUE; + archive_string_init(&cmdline); + archive_string_init(&fullpath); + + acmd = __archive_cmdline_allocate(); + if (acmd == NULL) + goto fail; + if (__archive_cmdline_parse(acmd, cmd) != ARCHIVE_OK) + goto fail; + + /* + * Search the full path of 'path'. + * NOTE: This does not need if we give CreateProcessA 'path' as + * a part of the cmdline and give CreateProcessA NULL as first + * parameter, but I do not like that way. + */ + ext = strrchr(acmd->path, '.'); + if (ext == NULL || strlen(ext) > 4) + /* 'path' does not have a proper extension, so we have to + * give SearchPath() ".exe" as the extension. */ + ext = ".exe"; + else + ext = NULL;/* 'path' has an extension. */ + + fl = MAX_PATH; + do { + if (archive_string_ensure(&fullpath, fl) == NULL) + goto fail; + fl_old = fl; + fl = SearchPathA(NULL, acmd->path, ext, fl, fullpath.s, + &arg0); + } while (fl != 0 && fl > fl_old); + if (fl == 0) + goto fail; + + /* + * Make a command line. + */ + for (l = 0, i = 0; acmd->argv[i] != NULL; i++) { + if (i == 0) + continue; + l += (int)strlen(acmd->argv[i]) + 1; + } + if (archive_string_ensure(&cmdline, l + 1) == NULL) + goto fail; + for (i = 0; acmd->argv[i] != NULL; i++) { + if (i == 0) { + const char *p, *sp; + + if ((p = strchr(acmd->argv[i], '/')) != NULL || + (p = strchr(acmd->argv[i], '\\')) != NULL) + p++; + else + p = acmd->argv[i]; + if ((sp = strchr(p, ' ')) != NULL) + archive_strappend_char(&cmdline, '"'); + archive_strcat(&cmdline, p); + if (sp != NULL) + archive_strappend_char(&cmdline, '"'); + } else { + archive_strappend_char(&cmdline, ' '); + archive_strcat(&cmdline, acmd->argv[i]); + } + } + if (i <= 1) { + const char *sp; + + if ((sp = strchr(arg0, ' ')) != NULL) + archive_strappend_char(&cmdline, '"'); + archive_strcat(&cmdline, arg0); + if (sp != NULL) + archive_strappend_char(&cmdline, '"'); + } + + secAtts.nLength = sizeof(SECURITY_ATTRIBUTES); + secAtts.bInheritHandle = TRUE; + secAtts.lpSecurityDescriptor = NULL; + if (CreatePipe(&childStdout[0], &childStdout[1], &secAtts, 0) == 0) + goto fail; + if (!SetHandleInformation(childStdout[0], HANDLE_FLAG_INHERIT, 0)) + goto fail; + if (CreatePipe(&childStdin[0], &childStdin[1], &secAtts, 0) == 0) + goto fail; + if (!SetHandleInformation(childStdin[1], HANDLE_FLAG_INHERIT, 0)) + goto fail; + if (DuplicateHandle(GetCurrentProcess(), GetStdHandle(STD_ERROR_HANDLE), + GetCurrentProcess(), &childStderr, 0, TRUE, + DUPLICATE_SAME_ACCESS) == 0) + goto fail; + + memset(&staInfo, 0, sizeof(staInfo)); + staInfo.cb = sizeof(staInfo); + staInfo.hStdError = childStderr; + staInfo.hStdOutput = childStdout[1]; + staInfo.hStdInput = childStdin[0]; + staInfo.wShowWindow = SW_HIDE; + staInfo.dwFlags = STARTF_USESTDHANDLES | STARTF_USESHOWWINDOW; + if (CreateProcessA(fullpath.s, cmdline.s, NULL, NULL, TRUE, 0, + NULL, NULL, &staInfo, &childInfo) == 0) + goto fail; + WaitForInputIdle(childInfo.hProcess, INFINITE); + CloseHandle(childInfo.hProcess); + CloseHandle(childInfo.hThread); + + *child_stdout = _open_osfhandle((intptr_t)childStdout[0], _O_RDONLY); + *child_stdin = _open_osfhandle((intptr_t)childStdin[1], _O_WRONLY); + + child = OpenProcess(PROCESS_QUERY_INFORMATION, FALSE, + childInfo.dwProcessId); + if (child == NULL) // INVALID_HANDLE_VALUE ? + goto fail; + + *out_child = child; + + CloseHandle(childStdout[1]); + CloseHandle(childStdin[0]); + + archive_string_free(&cmdline); + archive_string_free(&fullpath); + __archive_cmdline_free(acmd); + return ARCHIVE_OK; + +fail: + if (childStdout[0] != INVALID_HANDLE_VALUE) + CloseHandle(childStdout[0]); + if (childStdout[1] != INVALID_HANDLE_VALUE) + CloseHandle(childStdout[1]); + if (childStdin[0] != INVALID_HANDLE_VALUE) + CloseHandle(childStdin[0]); + if (childStdin[1] != INVALID_HANDLE_VALUE) + CloseHandle(childStdin[1]); + if (childStderr != INVALID_HANDLE_VALUE) + CloseHandle(childStderr); + archive_string_free(&cmdline); + archive_string_free(&fullpath); + __archive_cmdline_free(acmd); + return ARCHIVE_FAILED; +} + +void +__archive_check_child(int in, int out) +{ + (void)in; /* UNUSED */ + (void)out; /* UNUSED */ + Sleep(100); +} + +#endif /* _WIN32 && !__CYGWIN__ */ diff --git a/c/xxhash.c b/c/xxhash.c new file mode 100644 --- /dev/null +++ b/c/xxhash.c @@ -0,0 +1,521 @@ +/* +xxHash - Fast Hash algorithm +Copyright (C) 2012-2014, Yann Collet. +BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are +met: + +* Redistributions of source code must retain the above copyright +notice, this list of conditions and the following disclaimer. +* Redistributions in binary form must reproduce the above +copyright notice, this list of conditions and the following disclaimer +in the documentation and/or other materials provided with the +distribution. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +You can contact the author at : +- xxHash source repository : http://code.google.com/p/xxhash/ +*/ +#include "archive_platform.h" + +#include +#include + +#include "archive_xxhash.h" + +#ifdef HAVE_LIBLZ4 + +/*************************************** +** Tuning parameters +****************************************/ +/* Unaligned memory access is automatically enabled for "common" CPU, such as x86. +** For others CPU, the compiler will be more cautious, and insert extra code to ensure aligned access is respected. +** If you know your target CPU supports unaligned memory access, you want to force this option manually to improve performance. +** You can also enable this parameter if you know your input data will always be aligned (boundaries of 4, for U32). +*/ +#if defined(__ARM_FEATURE_UNALIGNED) || defined(__i386) || defined(_M_IX86) || defined(__x86_64__) || defined(_M_X64) +# define XXH_USE_UNALIGNED_ACCESS 1 +#endif + +/* XXH_ACCEPT_NULL_INPUT_POINTER : +** If the input pointer is a null pointer, xxHash default behavior is to trigger a memory access error, since it is a bad pointer. +** When this option is enabled, xxHash output for null input pointers will be the same as a null-length input. +** This option has a very small performance cost (only measurable on small inputs). +** By default, this option is disabled. To enable it, uncomment below define : +** #define XXH_ACCEPT_NULL_INPUT_POINTER 1 + +** XXH_FORCE_NATIVE_FORMAT : +** By default, xxHash library provides endian-independent Hash values, based on little-endian convention. +** Results are therefore identical for little-endian and big-endian CPU. +** This comes at a performance cost for big-endian CPU, since some swapping is required to emulate little-endian format. +** Should endian-independence be of no importance for your application, you may set the #define below to 1. +** It will improve speed for Big-endian CPU. +** This option has no impact on Little_Endian CPU. +*/ +#define XXH_FORCE_NATIVE_FORMAT 0 + +/*************************************** +** Compiler Specific Options +****************************************/ +/* Disable some Visual warning messages */ +#ifdef _MSC_VER /* Visual Studio */ +# pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */ +#endif + +#ifdef _MSC_VER /* Visual Studio */ +# define FORCE_INLINE __forceinline +#else +# ifdef __GNUC__ +# define FORCE_INLINE inline __attribute__((always_inline)) +# else +# define FORCE_INLINE inline +# endif +#endif + +/*************************************** +** Includes & Memory related functions +****************************************/ +#define XXH_malloc malloc +#define XXH_free free +#define XXH_memcpy memcpy + + +static unsigned int XXH32 (const void*, unsigned int, unsigned int); +static void* XXH32_init (unsigned int); +static XXH_errorcode XXH32_update (void*, const void*, unsigned int); +static unsigned int XXH32_digest (void*); +/*static int XXH32_sizeofState(void);*/ +static XXH_errorcode XXH32_resetState(void*, unsigned int); +#define XXH32_SIZEOFSTATE 48 +typedef struct { long long ll[(XXH32_SIZEOFSTATE+(sizeof(long long)-1))/sizeof(long long)]; } XXH32_stateSpace_t; +static unsigned int XXH32_intermediateDigest (void*); + +/*************************************** +** Basic Types +****************************************/ +#if defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L /* C99 */ +# include + typedef uint8_t BYTE; + typedef uint16_t U16; + typedef uint32_t U32; + typedef int32_t S32; + typedef uint64_t U64; +#else + typedef unsigned char BYTE; + typedef unsigned short U16; + typedef unsigned int U32; + typedef signed int S32; + typedef unsigned long long U64; +#endif + +#if defined(__GNUC__) && !defined(XXH_USE_UNALIGNED_ACCESS) +# define _PACKED __attribute__ ((packed)) +#else +# define _PACKED +#endif + +#if !defined(XXH_USE_UNALIGNED_ACCESS) && !defined(__GNUC__) +# ifdef __IBMC__ +# pragma pack(1) +# else +# pragma pack(push, 1) +# endif +#endif + +typedef struct _U32_S { U32 v; } _PACKED U32_S; + +#if !defined(XXH_USE_UNALIGNED_ACCESS) && !defined(__GNUC__) +# pragma pack(pop) +#endif + + +/**************************************** +** Compiler-specific Functions and Macros +*****************************************/ +#define GCC_VERSION ((__GNUC__-0) * 100 + (__GNUC_MINOR__ - 0)) + +#if GCC_VERSION >= 409 +__attribute__((__no_sanitize_undefined__)) +#endif +static inline U32 A32(const void * x) +{ + return (((const U32_S *)(x))->v); +} + +/* Note : although _rotl exists for minGW (GCC under windows), performance seems poor */ +#if defined(_MSC_VER) +# define XXH_rotl32(x,r) _rotl(x,r) +#else +# define XXH_rotl32(x,r) ((x << r) | (x >> (32 - r))) +#endif + +#if defined(_MSC_VER) /* Visual Studio */ +# define XXH_swap32 _byteswap_ulong +#elif GCC_VERSION >= 403 +# define XXH_swap32 __builtin_bswap32 +#else +static inline U32 XXH_swap32 (U32 x) { + return ((x << 24) & 0xff000000 ) | + ((x << 8) & 0x00ff0000 ) | + ((x >> 8) & 0x0000ff00 ) | + ((x >> 24) & 0x000000ff );} +#endif + + +/*************************************** +** Constants +****************************************/ +#define PRIME32_1 2654435761U +#define PRIME32_2 2246822519U +#define PRIME32_3 3266489917U +#define PRIME32_4 668265263U +#define PRIME32_5 374761393U + + +/*************************************** +** Architecture Macros +****************************************/ +typedef enum { XXH_bigEndian=0, XXH_littleEndian=1 } XXH_endianess; +#ifndef XXH_CPU_LITTLE_ENDIAN /* It is possible to define XXH_CPU_LITTLE_ENDIAN externally, for example using a compiler switch */ + static const int one = 1; +# define XXH_CPU_LITTLE_ENDIAN (*(const char*)(&one)) +#endif + + +/*************************************** +** Macros +****************************************/ +#define XXH_STATIC_ASSERT(c) { enum { XXH_static_assert = 1/(!!(c)) }; } /* use only *after* variable declarations */ + + +/***************************** +** Memory reads +******************************/ +typedef enum { XXH_aligned, XXH_unaligned } XXH_alignment; + +static +FORCE_INLINE U32 XXH_readLE32_align(const U32* ptr, XXH_endianess endian, XXH_alignment align) +{ + if (align==XXH_unaligned) + return endian==XXH_littleEndian ? A32(ptr) : XXH_swap32(A32(ptr)); + else + return endian==XXH_littleEndian ? *ptr : XXH_swap32(*ptr); +} + +static +FORCE_INLINE U32 XXH_readLE32(const U32* ptr, XXH_endianess endian) { return XXH_readLE32_align(ptr, endian, XXH_unaligned); } + + +/***************************** +** Simple Hash Functions +******************************/ +static +FORCE_INLINE U32 XXH32_endian_align(const void* input, unsigned int len, U32 seed, XXH_endianess endian, XXH_alignment align) +{ + const BYTE* p = (const BYTE*)input; + const BYTE* bEnd = p + len; + U32 h32; +#define XXH_get32bits(p) XXH_readLE32_align((const U32*)p, endian, align) + +#ifdef XXH_ACCEPT_NULL_INPUT_POINTER + if (p==NULL) { len=0; bEnd=p=(const BYTE*)(size_t)16; } +#endif + + if (len>=16) + { + const BYTE* const limit = bEnd - 16; + U32 v1 = seed + PRIME32_1 + PRIME32_2; + U32 v2 = seed + PRIME32_2; + U32 v3 = seed + 0; + U32 v4 = seed - PRIME32_1; + + do + { + v1 += XXH_get32bits(p) * PRIME32_2; v1 = XXH_rotl32(v1, 13); v1 *= PRIME32_1; p+=4; + v2 += XXH_get32bits(p) * PRIME32_2; v2 = XXH_rotl32(v2, 13); v2 *= PRIME32_1; p+=4; + v3 += XXH_get32bits(p) * PRIME32_2; v3 = XXH_rotl32(v3, 13); v3 *= PRIME32_1; p+=4; + v4 += XXH_get32bits(p) * PRIME32_2; v4 = XXH_rotl32(v4, 13); v4 *= PRIME32_1; p+=4; + } while (p<=limit); + + h32 = XXH_rotl32(v1, 1) + XXH_rotl32(v2, 7) + XXH_rotl32(v3, 12) + XXH_rotl32(v4, 18); + } + else + { + h32 = seed + PRIME32_5; + } + + h32 += (U32) len; + + while (p<=bEnd-4) + { + h32 += XXH_get32bits(p) * PRIME32_3; + h32 = XXH_rotl32(h32, 17) * PRIME32_4 ; + p+=4; + } + + while (p> 15; + h32 *= PRIME32_2; + h32 ^= h32 >> 13; + h32 *= PRIME32_3; + h32 ^= h32 >> 16; + + return h32; +} + + +U32 XXH32(const void* input, unsigned int len, U32 seed) +{ +#if 0 + // Simple version, good for code maintenance, but unfortunately slow for small inputs + void* state = XXH32_init(seed); + XXH32_update(state, input, len); + return XXH32_digest(state); +#else + XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN; + +# if !defined(XXH_USE_UNALIGNED_ACCESS) + if ((((size_t)input) & 3) == 0) /* Input is aligned, let's leverage the speed advantage */ + { + if ((endian_detected==XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) + return XXH32_endian_align(input, len, seed, XXH_littleEndian, XXH_aligned); + else + return XXH32_endian_align(input, len, seed, XXH_bigEndian, XXH_aligned); + } +# endif + + if ((endian_detected==XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) + return XXH32_endian_align(input, len, seed, XXH_littleEndian, XXH_unaligned); + else + return XXH32_endian_align(input, len, seed, XXH_bigEndian, XXH_unaligned); +#endif +} + +/***************************** +** Advanced Hash Functions +******************************/ + +struct XXH_state32_t +{ + U64 total_len; + U32 seed; + U32 v1; + U32 v2; + U32 v3; + U32 v4; + int memsize; + char memory[16]; +}; + +#if 0 +static +int XXH32_sizeofState(void) +{ + XXH_STATIC_ASSERT(XXH32_SIZEOFSTATE >= sizeof(struct XXH_state32_t)); /* A compilation error here means XXH32_SIZEOFSTATE is not large enough */ + return sizeof(struct XXH_state32_t); +} +#endif + +static +XXH_errorcode XXH32_resetState(void* state_in, U32 seed) +{ + struct XXH_state32_t * state = (struct XXH_state32_t *) state_in; + state->seed = seed; + state->v1 = seed + PRIME32_1 + PRIME32_2; + state->v2 = seed + PRIME32_2; + state->v3 = seed + 0; + state->v4 = seed - PRIME32_1; + state->total_len = 0; + state->memsize = 0; + return XXH_OK; +} + +static +void* XXH32_init (U32 seed) +{ + void* state = XXH_malloc (sizeof(struct XXH_state32_t)); + XXH32_resetState(state, seed); + return state; +} + +static +FORCE_INLINE XXH_errorcode XXH32_update_endian (void* state_in, const void* input, int len, XXH_endianess endian) +{ + struct XXH_state32_t * state = (struct XXH_state32_t *) state_in; + const BYTE* p = (const BYTE*)input; + const BYTE* const bEnd = p + len; + +#ifdef XXH_ACCEPT_NULL_INPUT_POINTER + if (input==NULL) return XXH_ERROR; +#endif + + state->total_len += len; + + if (state->memsize + len < 16) /* fill in tmp buffer */ + { + XXH_memcpy(state->memory + state->memsize, input, len); + state->memsize += len; + return XXH_OK; + } + + if (state->memsize) /* some data left from previous update */ + { + XXH_memcpy(state->memory + state->memsize, input, 16-state->memsize); + { + const U32* p32 = (const U32*)state->memory; + state->v1 += XXH_readLE32(p32, endian) * PRIME32_2; state->v1 = XXH_rotl32(state->v1, 13); state->v1 *= PRIME32_1; p32++; + state->v2 += XXH_readLE32(p32, endian) * PRIME32_2; state->v2 = XXH_rotl32(state->v2, 13); state->v2 *= PRIME32_1; p32++; + state->v3 += XXH_readLE32(p32, endian) * PRIME32_2; state->v3 = XXH_rotl32(state->v3, 13); state->v3 *= PRIME32_1; p32++; + state->v4 += XXH_readLE32(p32, endian) * PRIME32_2; state->v4 = XXH_rotl32(state->v4, 13); state->v4 *= PRIME32_1; p32++; + } + p += 16-state->memsize; + state->memsize = 0; + } + + if (p <= bEnd-16) + { + const BYTE* const limit = bEnd - 16; + U32 v1 = state->v1; + U32 v2 = state->v2; + U32 v3 = state->v3; + U32 v4 = state->v4; + + do + { + v1 += XXH_readLE32((const U32*)p, endian) * PRIME32_2; v1 = XXH_rotl32(v1, 13); v1 *= PRIME32_1; p+=4; + v2 += XXH_readLE32((const U32*)p, endian) * PRIME32_2; v2 = XXH_rotl32(v2, 13); v2 *= PRIME32_1; p+=4; + v3 += XXH_readLE32((const U32*)p, endian) * PRIME32_2; v3 = XXH_rotl32(v3, 13); v3 *= PRIME32_1; p+=4; + v4 += XXH_readLE32((const U32*)p, endian) * PRIME32_2; v4 = XXH_rotl32(v4, 13); v4 *= PRIME32_1; p+=4; + } while (p<=limit); + + state->v1 = v1; + state->v2 = v2; + state->v3 = v3; + state->v4 = v4; + } + + if (p < bEnd) + { + XXH_memcpy(state->memory, p, bEnd-p); + state->memsize = (int)(bEnd-p); + } + + return XXH_OK; +} + +static +XXH_errorcode XXH32_update (void* state_in, const void* input, unsigned int len) +{ + XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN; + + if ((endian_detected==XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) + return XXH32_update_endian(state_in, input, len, XXH_littleEndian); + else + return XXH32_update_endian(state_in, input, len, XXH_bigEndian); +} + + + +static +FORCE_INLINE U32 XXH32_intermediateDigest_endian (void* state_in, XXH_endianess endian) +{ + struct XXH_state32_t * state = (struct XXH_state32_t *) state_in; + const BYTE * p = (const BYTE*)state->memory; + BYTE* bEnd = (BYTE*)state->memory + state->memsize; + U32 h32; + + if (state->total_len >= 16) + { + h32 = XXH_rotl32(state->v1, 1) + XXH_rotl32(state->v2, 7) + XXH_rotl32(state->v3, 12) + XXH_rotl32(state->v4, 18); + } + else + { + h32 = state->seed + PRIME32_5; + } + + h32 += (U32) state->total_len; + + while (p<=bEnd-4) + { + h32 += XXH_readLE32((const U32*)p, endian) * PRIME32_3; + h32 = XXH_rotl32(h32, 17) * PRIME32_4; + p+=4; + } + + while (p> 15; + h32 *= PRIME32_2; + h32 ^= h32 >> 13; + h32 *= PRIME32_3; + h32 ^= h32 >> 16; + + return h32; +} + +static +U32 XXH32_intermediateDigest (void* state_in) +{ + XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN; + + if ((endian_detected==XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) + return XXH32_intermediateDigest_endian(state_in, XXH_littleEndian); + else + return XXH32_intermediateDigest_endian(state_in, XXH_bigEndian); +} + +static +U32 XXH32_digest (void* state_in) +{ + U32 h32 = XXH32_intermediateDigest(state_in); + + XXH_free(state_in); + + return h32; +} + +const +struct archive_xxhash __archive_xxhash = { + XXH32, + XXH32_init, + XXH32_update, + XXH32_digest +}; +#else + +/* + * Define an empty version of the struct if we aren't using the LZ4 library. + */ +const +struct archive_xxhash __archive_xxhash = { + NULL, + NULL, + NULL, + NULL +}; + +#endif /* HAVE_LIBLZ4 */ diff --git a/libarchive.cabal b/libarchive.cabal --- a/libarchive.cabal +++ b/libarchive.cabal @@ -1,19 +1,26 @@ -cabal-version: 1.18 -name: libarchive -version: 2.2.5.0 -license: BSD3 -license-file: LICENSE -copyright: Copyright: (c) 2018-2020 Vanessa McHale -maintainer: vamchale@gmail.com -author: Vanessa McHale -tested-with: ghc ==8.4.4 ghc ==8.6.5 ghc ==8.8.3 ghc ==8.10.1 -bug-reports: https://github.com/vmchale/libarchive/issues -synopsis: Haskell interface to libarchive +cabal-version: 3.0 +name: libarchive +version: 2.2.5.1 +license: BSD-3-Clause +license-file: LICENSE +copyright: Copyright: (c) 2018-2020 Vanessa McHale +maintainer: vamchale@gmail.com +author: Vanessa McHale +tested-with: ghc ==8.4.4 ghc ==8.6.5 ghc ==8.8.3 ghc ==8.10.1 +bug-reports: https://github.com/vmchale/libarchive/issues +synopsis: Haskell interface to libarchive description: Haskell bindings for [libarchive](https://www.libarchive.org/). Provides the ability to unpack archives, including the ability to unpack archives lazily. -category: Codec -build-type: Custom +category: Codec +build-type: Custom +extra-source-files: + c/autoconf-darwin/config.h + c/autoconf-linux/config.h + c/autoconf-freebsd/config.h + c/*.c + c/*.h + extra-doc-files: README.md CHANGELOG.md @@ -37,6 +44,13 @@ description: Run low-memory version of test suite default: False +flag static + description: + Use the bundled sources (this is default for windows and darwin) + + default: False + manual: True + library exposed-modules: Codec.Archive @@ -44,8 +58,7 @@ Codec.Archive.Foreign.Archive Codec.Archive.Foreign.ArchiveEntry - pkgconfig-depends: libarchive (==3.4.0 || >3.4.0) && <4.0 - hs-source-dirs: src + hs-source-dirs: src other-modules: Codec.Archive.Foreign.Archive.Macros Codec.Archive.Foreign.ArchiveEntry.Macros @@ -60,8 +73,8 @@ Codec.Archive.Common Codec.Archive.Monad - default-language: Haskell2010 - other-extensions: DeriveGeneric DeriveAnyClass + default-language: Haskell2010 + other-extensions: DeriveGeneric DeriveAnyClass ghc-options: -Wall -Wincomplete-uni-patterns -Wincomplete-record-updates -Wredundant-constraints @@ -84,6 +97,158 @@ if impl(ghc >=8.10) ghc-options: -Wunused-packages + + if flag(static) + cc-options: -std=gnu99 -DHAVE_CONFIG_H + c-sources: + c/archive_acl.c + c/archive_blake2sp_ref.c + c/archive_blake2s_ref.c + c/archive_check_magic.c + c/archive_cmdline.c + c/archive_cryptor.c + c/archive_digest.c + c/archive_disk_acl_darwin.c + c/archive_disk_acl_freebsd.c + c/archive_disk_acl_linux.c + c/archive_disk_acl_sunos.c + c/archive_entry.c + c/archive_entry_copy_bhfi.c + c/archive_entry_copy_stat.c + c/archive_entry_link_resolver.c + c/archive_entry_sparse.c + c/archive_entry_stat.c + c/archive_entry_strmode.c + c/archive_entry_xattr.c + c/archive_getdate.c + c/archive_hmac.c + c/archive_match.c + c/archive_options.c + c/archive_pack_dev.c + c/archive_pathmatch.c + c/archive_ppmd7.c + c/archive_ppmd8.c + c/archive_random.c + c/archive_rb.c + c/archive_read_add_passphrase.c + c/archive_read_append_filter.c + c/archive_read.c + c/archive_read_data_into_fd.c + c/archive_read_disk_entry_from_file.c + c/archive_read_disk_posix.c + c/archive_read_disk_set_standard_lookup.c + c/archive_read_disk_windows.c + c/archive_read_extract2.c + c/archive_read_extract.c + c/archive_read_open_fd.c + c/archive_read_open_file.c + c/archive_read_open_filename.c + c/archive_read_open_memory.c + c/archive_read_set_format.c + c/archive_read_set_options.c + c/archive_read_support_filter_all.c + c/archive_read_support_filter_bzip2.c + c/archive_read_support_filter_compress.c + c/archive_read_support_filter_grzip.c + c/archive_read_support_filter_gzip.c + c/archive_read_support_filter_lrzip.c + c/archive_read_support_filter_lz4.c + c/archive_read_support_filter_lzop.c + c/archive_read_support_filter_none.c + c/archive_read_support_filter_program.c + c/archive_read_support_filter_rpm.c + c/archive_read_support_filter_uu.c + c/archive_read_support_filter_xz.c + c/archive_read_support_filter_zstd.c + c/archive_read_support_format_7zip.c + c/archive_read_support_format_all.c + c/archive_read_support_format_ar.c + c/archive_read_support_format_by_code.c + c/archive_read_support_format_cab.c + c/archive_read_support_format_cpio.c + c/archive_read_support_format_empty.c + c/archive_read_support_format_iso9660.c + c/archive_read_support_format_lha.c + c/archive_read_support_format_mtree.c + c/archive_read_support_format_rar5.c + c/archive_read_support_format_rar.c + c/archive_read_support_format_raw.c + c/archive_read_support_format_tar.c + c/archive_read_support_format_warc.c + c/archive_read_support_format_xar.c + c/archive_read_support_format_zip.c + c/archive_string.c + c/archive_string_sprintf.c + c/archive_util.c + c/archive_version_details.c + c/archive_virtual.c + c/archive_windows.c + c/archive_write_add_filter_b64encode.c + c/archive_write_add_filter_by_name.c + c/archive_write_add_filter_bzip2.c + c/archive_write_add_filter.c + c/archive_write_add_filter_compress.c + c/archive_write_add_filter_grzip.c + c/archive_write_add_filter_gzip.c + c/archive_write_add_filter_lrzip.c + c/archive_write_add_filter_lz4.c + c/archive_write_add_filter_lzop.c + c/archive_write_add_filter_none.c + c/archive_write_add_filter_program.c + c/archive_write_add_filter_uuencode.c + c/archive_write_add_filter_xz.c + c/archive_write_add_filter_zstd.c + c/archive_write.c + c/archive_write_disk_posix.c + c/archive_write_disk_set_standard_lookup.c + c/archive_write_disk_windows.c + c/archive_write_open_fd.c + c/archive_write_open_file.c + c/archive_write_open_filename.c + c/archive_write_open_memory.c + c/archive_write_set_format_7zip.c + c/archive_write_set_format_ar.c + c/archive_write_set_format_by_name.c + c/archive_write_set_format.c + c/archive_write_set_format_cpio.c + c/archive_write_set_format_cpio_newc.c + c/archive_write_set_format_filter_by_ext.c + c/archive_write_set_format_gnutar.c + c/archive_write_set_format_iso9660.c + c/archive_write_set_format_mtree.c + c/archive_write_set_format_pax.c + c/archive_write_set_format_raw.c + c/archive_write_set_format_shar.c + c/archive_write_set_format_ustar.c + c/archive_write_set_format_v7tar.c + c/archive_write_set_format_warc.c + c/archive_write_set_format_xar.c + c/archive_write_set_format_zip.c + c/archive_write_set_options.c + c/archive_write_set_passphrase.c + c/filter_fork_posix.c + c/filter_fork_windows.c + c/xxhash.c + + include-dirs: c/ + + if os(osx) + include-dirs: c/autoconf-darwin + + else + if os(linux) + include-dirs: c/autoconf-linux + + else + if os(freebsd) + include-dirs: c/autoconf-freebsd + + else + buildable: False + build-depends: unbuildable <0 + + else + pkgconfig-depends: libarchive (==3.4.0 || >3.4.0) && <4.0 test-suite libarchive-test type: exitcode-stdio-1.0 diff --git a/mem/Mem.hs b/mem/Mem.hs --- a/mem/Mem.hs +++ b/mem/Mem.hs @@ -10,5 +10,5 @@ main = readArc readArc :: IO () -readArc = either throw forceList =<< runArchiveM +readArc = forceList =<< throwArchiveM (readArchiveFile "test/data/llvm-9.0.0.src.tar") diff --git a/src/Codec/Archive/Pack/Lazy.hs b/src/Codec/Archive/Pack/Lazy.hs --- a/src/Codec/Archive/Pack/Lazy.hs +++ b/src/Codec/Archive/Pack/Lazy.hs @@ -101,3 +101,4 @@ modifyIORef' bsRef (`DL.snoc` bsl) pure bytesRead doNothing _ _ = pure ArchiveOk + -- FIXME: this part isn't sufficiently lazy diff --git a/src/Codec/Archive/Types/Foreign.chs b/src/Codec/Archive/Types/Foreign.chs --- a/src/Codec/Archive/Types/Foreign.chs +++ b/src/Codec/Archive/Types/Foreign.chs @@ -35,6 +35,7 @@ import Control.DeepSeq (NFData) import Control.Exception (Exception) import Data.Bits (Bits (..)) +import Data.Semigroup (Semigroup (..)) import Foreign.C.String (CString) import Foreign.C.Types (CInt, CSize) import Foreign.Ptr (Ptr)