diff --git a/NOTES b/NOTES new file mode 100644 --- /dev/null +++ b/NOTES @@ -0,0 +1,7 @@ +This is the same as package version 0.94.4.5.8.31 + +PS. JeffShaw, next time when you upload a package, set yourself as the maintainer or at least leave some note with your contact! This is very important, as you broke the package, and then didn't even leave a change log as to why you uploaded the new version! + +http://stackoverflow.com/questions/17361047/compiling-libpandoc-and-undefined-reference-to-utf8-table4-in-regex-pcre-builtin + +I am Timothy Hobbs. timothyhobbs@seznam.cz diff --git a/Text/Regex/PCRE/Wrap.hsc b/Text/Regex/PCRE/Wrap.hsc --- a/Text/Regex/PCRE/Wrap.hsc +++ b/Text/Regex/PCRE/Wrap.hsc @@ -73,8 +73,8 @@ import Control.Monad(when) import Data.Array(Array,accumArray) import Data.Bits(Bits((.|.))) -- ((.&.),(.|.),complement)) -import System.IO.Unsafe(unsafePerformIO) -import Foreign(Ptr,ForeignPtr,FinalizerPtr -- ,FunPtr +import Foreign(unsafePerformIO + ,Ptr,ForeignPtr,FinalizerPtr -- ,FunPtr ,alloca,allocaBytes,nullPtr ,peek,peekElemOff ,newForeignPtr,withForeignPtr) diff --git a/pcre/config.h b/pcre/config.h --- a/pcre/config.h +++ b/pcre/config.h @@ -1,70 +1,62 @@ /* config.h. Generated from config.h.in by configure. */ /* config.h.in. Generated from configure.ac by autoheader. */ -/* PCRE is written in Standard C, but there are a few non-standard things it -can cope with, allowing it to run on SunOS4 and other "close to standard" -systems. -In environments that support the GNU autotools, config.h.in is converted into -config.h by the "configure" script. In environments that use CMake, -config-cmake.in is converted into config.h. If you are going to build PCRE "by -hand" without using "configure" or CMake, you should copy the distributed -config.h.generic to config.h, and edit the macro definitions to be the way you -need them. You must then add -DHAVE_CONFIG_H to all of your compile commands, -so that config.h is included at the start of every source. +/* On Unix-like systems config.h.in is converted by "configure" into config.h. +Some other environments also support the use of "configure". PCRE is written in +Standard C, but there are a few non-standard things it can cope with, allowing +it to run on SunOS4 and other "close to standard" systems. -Alternatively, you can avoid editing by using -D on the compiler command line -to set the macro values. In this case, you do not have to set -DHAVE_CONFIG_H, -but if you do, default values will be taken from config.h for non-boolean -macros that are not defined on the command line. +If you are going to build PCRE "by hand" on a system without "configure" you +should copy the distributed config.h.generic to config.h, and then set up the +macro definitions the way you need them. You must then add -DHAVE_CONFIG_H to +all of your compile commands, so that config.h is included at the start of +every source. -Boolean macros such as HAVE_STDLIB_H and SUPPORT_PCRE8 should either be defined -(conventionally to 1) for TRUE, and not defined at all for FALSE. All such -macros are listed as a commented #undef in config.h.generic. Macros such as -MATCH_LIMIT, whose actual value is relevant, have defaults defined, but are -surrounded by #ifndef/#endif lines so that the value can be overridden by -D. +Alternatively, you can avoid editing by using -D on the compiler command line +to set the macro values. In this case, you do not have to set -DHAVE_CONFIG_H. -PCRE uses memmove() if HAVE_MEMMOVE is defined; otherwise it uses bcopy() if -HAVE_BCOPY is defined. If your system has neither bcopy() nor memmove(), make -sure both macros are undefined; an emulation function will then be used. */ +PCRE uses memmove() if HAVE_MEMMOVE is set to 1; otherwise it uses bcopy() if +HAVE_BCOPY is set to 1. If your system has neither bcopy() nor memmove(), set +them both to 0; an emulation function will be used. */ /* By default, the \R escape sequence matches any Unicode line ending - character or sequence of characters. If BSR_ANYCRLF is defined (to any - value), this is changed so that backslash-R matches only CR, LF, or CRLF. - The build-time default can be overridden by the user of PCRE at runtime. */ + character or sequence of characters. If BSR_ANYCRLF is defined, this is + changed so that backslash-R matches only CR, LF, or CRLF. The build- time + default can be overridden by the user of PCRE at runtime. On systems that + support it, "configure" can be used to override the default. */ /* #undef BSR_ANYCRLF */ /* If you are compiling for a system that uses EBCDIC instead of ASCII - character codes, define this macro to any value. You must also edit the - NEWLINE macro below to set a suitable EBCDIC newline, commonly 21 (0x15). - On systems that can use "configure" or CMake to set EBCDIC, NEWLINE is - automatically adjusted. When EBCDIC is set, PCRE assumes that all input - strings are in EBCDIC. If you do not define this macro, PCRE will assume - input strings are ASCII or UTF-8/16/32 Unicode. It is not possible to build - a version of PCRE that supports both EBCDIC and UTF-8/16/32. */ + character codes, define this macro as 1. On systems that can use + "configure", this can be done via --enable-ebcdic. PCRE will then assume + that all input strings are in EBCDIC. If you do not define this macro, PCRE + will assume input strings are ASCII or UTF-8/16 Unicode. It is not possible + to build a version of PCRE that supports both EBCDIC and UTF-8/16. */ /* #undef EBCDIC */ -/* In an EBCDIC environment, define this macro to any value to arrange for the - NL character to be 0x25 instead of the default 0x15. NL plays the role that - LF does in an ASCII/Unicode environment. The value must also be set in the - NEWLINE macro below. On systems that can use "configure" or CMake to set - EBCDIC_NL25, the adjustment of NEWLINE is automatic. */ -/* #undef EBCDIC_NL25 */ - /* Define to 1 if you have the `bcopy' function. */ -/* #undef HAVE_BCOPY */ +#ifndef HAVE_BCOPY +#define HAVE_BCOPY 1 +#endif /* Define to 1 if you have the header file. */ /* #undef HAVE_BITS_TYPE_TRAITS_H */ /* Define to 1 if you have the header file. */ -/* #undef HAVE_BZLIB_H */ +#ifndef HAVE_BZLIB_H +#define HAVE_BZLIB_H 1 +#endif /* Define to 1 if you have the header file. */ -/* #undef HAVE_DIRENT_H */ +#ifndef HAVE_DIRENT_H +#define HAVE_DIRENT_H 1 +#endif /* Define to 1 if you have the header file. */ -/* #undef HAVE_DLFCN_H */ +#ifndef HAVE_DLFCN_H +#define HAVE_DLFCN_H 1 +#endif /* Define to 1 if you have the header file. */ /* #undef HAVE_EDITLINE_READLINE_H */ @@ -73,25 +65,29 @@ /* #undef HAVE_EDIT_READLINE_READLINE_H */ /* Define to 1 if you have the header file. */ -/* #undef HAVE_INTTYPES_H */ +#ifndef HAVE_INTTYPES_H +#define HAVE_INTTYPES_H 1 +#endif /* Define to 1 if you have the header file. */ -/* #undef HAVE_LIMITS_H */ +#ifndef HAVE_LIMITS_H +#define HAVE_LIMITS_H 1 +#endif /* Define to 1 if the system has the type `long long'. */ -/* #undef HAVE_LONG_LONG */ +#ifndef HAVE_LONG_LONG +#define HAVE_LONG_LONG 1 +#endif /* Define to 1 if you have the `memmove' function. */ -/* #undef HAVE_MEMMOVE */ +#ifndef HAVE_MEMMOVE +#define HAVE_MEMMOVE 1 +#endif /* Define to 1 if you have the header file. */ -/* #undef HAVE_MEMORY_H */ - -/* Define if you have POSIX threads libraries and header files. */ -/* #undef HAVE_PTHREAD */ - -/* Have PTHREAD_PRIO_INHERIT. */ -/* #undef HAVE_PTHREAD_PRIO_INHERIT */ +#ifndef HAVE_MEMORY_H +#define HAVE_MEMORY_H 1 +#endif /* Define to 1 if you have the header file. */ /* #undef HAVE_READLINE_HISTORY_H */ @@ -100,22 +96,34 @@ /* #undef HAVE_READLINE_READLINE_H */ /* Define to 1 if you have the header file. */ -/* #undef HAVE_STDINT_H */ +#ifndef HAVE_STDINT_H +#define HAVE_STDINT_H 1 +#endif /* Define to 1 if you have the header file. */ -/* #undef HAVE_STDLIB_H */ +#ifndef HAVE_STDLIB_H +#define HAVE_STDLIB_H 1 +#endif /* Define to 1 if you have the `strerror' function. */ -/* #undef HAVE_STRERROR */ +#ifndef HAVE_STRERROR +#define HAVE_STRERROR 1 +#endif /* Define to 1 if you have the header file. */ -/* #undef HAVE_STRING */ +#ifndef HAVE_STRING +#define HAVE_STRING 1 +#endif /* Define to 1 if you have the header file. */ -/* #undef HAVE_STRINGS_H */ +#ifndef HAVE_STRINGS_H +#define HAVE_STRINGS_H 1 +#endif /* Define to 1 if you have the header file. */ -/* #undef HAVE_STRING_H */ +#ifndef HAVE_STRING_H +#define HAVE_STRING_H 1 +#endif /* Define to 1 if you have `strtoimax'. */ /* #undef HAVE_STRTOIMAX */ @@ -124,31 +132,40 @@ /* #undef HAVE_STRTOLL */ /* Define to 1 if you have `strtoq'. */ -/* #undef HAVE_STRTOQ */ +#ifndef HAVE_STRTOQ +#define HAVE_STRTOQ 1 +#endif /* Define to 1 if you have the header file. */ -/* #undef HAVE_SYS_STAT_H */ +#ifndef HAVE_SYS_STAT_H +#define HAVE_SYS_STAT_H 1 +#endif /* Define to 1 if you have the header file. */ -/* #undef HAVE_SYS_TYPES_H */ +#ifndef HAVE_SYS_TYPES_H +#define HAVE_SYS_TYPES_H 1 +#endif /* Define to 1 if you have the header file. */ /* #undef HAVE_TYPE_TRAITS_H */ /* Define to 1 if you have the header file. */ -/* #undef HAVE_UNISTD_H */ +#ifndef HAVE_UNISTD_H +#define HAVE_UNISTD_H 1 +#endif /* Define to 1 if the system has the type `unsigned long long'. */ -/* #undef HAVE_UNSIGNED_LONG_LONG */ - -/* Define to 1 if the compiler supports simple visibility declarations. */ -/* #undef HAVE_VISIBILITY */ +#ifndef HAVE_UNSIGNED_LONG_LONG +#define HAVE_UNSIGNED_LONG_LONG 1 +#endif /* Define to 1 if you have the header file. */ /* #undef HAVE_WINDOWS_H */ /* Define to 1 if you have the header file. */ -/* #undef HAVE_ZLIB_H */ +#ifndef HAVE_ZLIB_H +#define HAVE_ZLIB_H 1 +#endif /* Define to 1 if you have `_strtoi64'. */ /* #undef HAVE__STRTOI64 */ @@ -157,14 +174,14 @@ as offsets within the compiled regex. The default is 2, which allows for compiled patterns up to 64K long. This covers the vast majority of cases. However, PCRE can also be compiled to use 3 or 4 bytes instead. This allows - for longer patterns in extreme cases. */ + for longer patterns in extreme cases. On systems that support it, + "configure" can be used to override this default. */ #ifndef LINK_SIZE #define LINK_SIZE 2 #endif /* Define to the sub-directory in which libtool stores uninstalled libraries. */ -/* This is ignored unless you are using libtool. */ #ifndef LT_OBJDIR #define LT_OBJDIR ".libs/" #endif @@ -174,7 +191,8 @@ pcre_exec(). There is a runtime interface for setting a different limit. The limit exists in order to catch runaway regular expressions that take for ever to determine that they do not match. The default is set very large - so that it does not accidentally catch legitimate cases. */ + so that it does not accidentally catch legitimate cases. On systems that + support it, "configure" can be used to override this default default. */ #ifndef MATCH_LIMIT #define MATCH_LIMIT 10000000 #endif @@ -186,7 +204,8 @@ used. The value of MATCH_LIMIT_RECURSION applies only to recursive calls of match(). To have any useful effect, it must be less than the value of MATCH_LIMIT. The default is to use the same value as MATCH_LIMIT. There is - a runtime method for setting a different limit. */ + a runtime method for setting a different limit. On systems that support it, + "configure" can be used to override the default. */ #ifndef MATCH_LIMIT_RECURSION #define MATCH_LIMIT_RECURSION MATCH_LIMIT #endif @@ -205,25 +224,22 @@ #define MAX_NAME_SIZE 32 #endif -/* The value of NEWLINE determines the default newline character sequence. - PCRE client programs can override this by selecting other values at run - time. In ASCII environments, the value can be 10 (LF), 13 (CR), or 3338 - (CRLF); in EBCDIC environments the value can be 21 or 37 (LF), 13 (CR), or - 3349 or 3365 (CRLF) because there are two alternative codepoints (0x15 and - 0x25) that are used as the NL line terminator that is equivalent to ASCII - LF. In both ASCII and EBCDIC environments the value can also be -1 (ANY), - or -2 (ANYCRLF). */ +/* The value of NEWLINE determines the newline character sequence. On systems + that support it, "configure" can be used to override the default, which is + 10. The possible values are 10 (LF), 13 (CR), 3338 (CRLF), -1 (ANY), or -2 + (ANYCRLF). */ #ifndef NEWLINE #define NEWLINE 10 #endif /* PCRE uses recursive function calls to handle backtracking while matching. This can sometimes be a problem on systems that have stacks of limited - size. Define NO_RECURSE to any value to get a version that doesn't use - recursion in the match() function; instead it creates its own stack by - steam using pcre_recurse_malloc() to obtain memory from the heap. For more - detail, see the comments and other stuff just above the match() function. - */ + size. Define NO_RECURSE to get a version that doesn't use recursion in the + match() function; instead it creates its own stack by steam using + pcre_recurse_malloc() to obtain memory from the heap. For more detail, see + the comments and other stuff just above the match() function. On systems + that support it, "configure" can be used to set this in the Makefile (use + --disable-stack-for-recursion). */ /* #undef NO_RECURSE */ /* Name of package */ @@ -236,7 +252,7 @@ #define PACKAGE_NAME "PCRE" /* Define to the full name and version of this package. */ -#define PACKAGE_STRING "PCRE 8.34" +#define PACKAGE_STRING "PCRE 8.31" /* Define to the one symbol short name of this package. */ #define PACKAGE_TARNAME "pcre" @@ -245,36 +261,31 @@ #define PACKAGE_URL "" /* Define to the version of this package. */ -#define PACKAGE_VERSION "8.34" - -/* The value of PARENS_NEST_LIMIT specifies the maximum depth of nested - parentheses (of any kind) in a pattern. This limits the amount of system - stack that is used while compiling a pattern. */ -#ifndef PARENS_NEST_LIMIT -#define PARENS_NEST_LIMIT 250 -#endif +#define PACKAGE_VERSION "8.31" /* The value of PCREGREP_BUFSIZE determines the size of buffer used by - pcregrep to hold parts of the file it is searching. This is also the - minimum value. The actual amount of memory used by pcregrep is three times - this number, because it allows for the buffering of "before" and "after" - lines. */ + pcregrep to hold parts of the file it is searching. On systems that support + it, "configure" can be used to override the default, which is 8192. This is + also the minimum value. The actual amount of memory used by pcregrep is + three times this number, because it allows for the buffering of "before" + and "after" lines. */ #ifndef PCREGREP_BUFSIZE #define PCREGREP_BUFSIZE 20480 #endif + /* If you are compiling for a system other than a Unix-like system or Win32, and it needs some magic to be inserted before the definition of a function that is exported by the library, define this macro to - contain the relevant magic. If you do not define this macro, a suitable - __declspec value is used for Windows systems; in other environments - "extern" is used for a C compiler and "extern C" for a C++ compiler. - This macro apears at the start of every exported function that is part - of the external API. It does not appear on functions that are "external" - in the C sense, but which are internal to the library. */ + contain the relevant magic. If you do not define this macro, it + defaults to "extern" for a C compiler and "extern C" for a C++ + compiler on non-Win32 systems. This macro apears at the start of + every exported function that is part of the external API. It does + not appear on functions that are "external" in the C sense, but + which are internal to the library. */ /* #undef PCRE_EXP_DEFN */ -/* Define to any value if linking statically (TODO: make nice with Libtool) */ +/* Define if linking statically (TODO: make nice with Libtool) */ /* #undef PCRE_STATIC */ /* When calling PCRE via the POSIX interface, additional working storage is @@ -283,61 +294,58 @@ only two. If the number of expected substrings is small, the wrapper function uses space on the stack, because this is faster than using malloc() for each call. The threshold above which the stack is no longer - used is defined by POSIX_MALLOC_THRESHOLD. */ + used is defined by POSIX_MALLOC_THRESHOLD. On systems that support it, + "configure" can be used to override this default. */ #ifndef POSIX_MALLOC_THRESHOLD #define POSIX_MALLOC_THRESHOLD 10 #endif -/* Define to necessary symbol if this constant uses a non-standard name on - your system. */ -/* #undef PTHREAD_CREATE_JOINABLE */ - /* Define to 1 if you have the ANSI C header files. */ -/* #undef STDC_HEADERS */ +#ifndef STDC_HEADERS +#define STDC_HEADERS 1 +#endif -/* Define to any value to enable support for Just-In-Time compiling. */ +/* Define to enable support for Just-In-Time compiling. */ /* #undef SUPPORT_JIT */ -/* Define to any value to allow pcregrep to be linked with libbz2, so that it - is able to handle .bz2 files. */ +/* Define to allow pcregrep to be linked with libbz2, so that it is able to + handle .bz2 files. */ /* #undef SUPPORT_LIBBZ2 */ -/* Define to any value to allow pcretest to be linked with libedit. */ +/* Define to allow pcretest to be linked with libedit. */ /* #undef SUPPORT_LIBEDIT */ -/* Define to any value to allow pcretest to be linked with libreadline. */ +/* Define to allow pcretest to be linked with libreadline. */ /* #undef SUPPORT_LIBREADLINE */ -/* Define to any value to allow pcregrep to be linked with libz, so that it is - able to handle .gz files. */ +/* Define to allow pcregrep to be linked with libz, so that it is able to + handle .gz files. */ /* #undef SUPPORT_LIBZ */ -/* Define to any value to enable the 16 bit PCRE library. */ +/* Define to enable the 16 bit PCRE library. */ /* #undef SUPPORT_PCRE16 */ -/* Define to any value to enable the 32 bit PCRE library. */ -/* #undef SUPPORT_PCRE32 */ - -/* Define to any value to enable the 8 bit PCRE library. */ -/* #undef SUPPORT_PCRE8 */ +/* Define to enable the 8 bit PCRE library. */ +#ifndef SUPPORT_PCRE8 +#define SUPPORT_PCRE8 /**/ +#endif -/* Define to any value to enable JIT support in pcregrep. */ +/* Define to enable JIT support in pcregrep. */ /* #undef SUPPORT_PCREGREP_JIT */ -/* Define to any value to enable support for Unicode properties. */ +/* Define to enable support for Unicode properties. */ /* #undef SUPPORT_UCP */ -/* Define to any value to enable support for the UTF-8/16/32 Unicode encoding. - This will work even in an EBCDIC environment, but it is incompatible with - the EBCDIC macro. That is, PCRE can support *either* EBCDIC code *or* - ASCII/UTF-8/16/32, but not both at once. */ +/* Define to enable support for the UTF-8/16 Unicode encoding. This will work + even in an EBCDIC environment, but it is incompatible with the EBCDIC + macro. That is, PCRE can support *either* EBCDIC code *or* ASCII/UTF-8/16, + but not both at once. */ /* #undef SUPPORT_UTF */ -/* Define to any value for valgrind support to find invalid memory reads. */ -/* #undef SUPPORT_VALGRIND */ - /* Version number of package */ -#define VERSION "8.34" +#ifndef VERSION +#define VERSION "8.31" +#endif /* Define to empty if `const' does not conform to ANSI C. */ /* #undef const */ diff --git a/pcre/pcre.h b/pcre/pcre.h --- a/pcre/pcre.h +++ b/pcre/pcre.h @@ -5,7 +5,7 @@ /* This is the public header file for the PCRE library, to be #included by applications that call the PCRE functions. - Copyright (c) 1997-2013 University of Cambridge + Copyright (c) 1997-2012 University of Cambridge ----------------------------------------------------------------------------- Redistribution and use in source and binary forms, with or without @@ -42,9 +42,9 @@ /* The current PCRE version information. */ #define PCRE_MAJOR 8 -#define PCRE_MINOR 34 +#define PCRE_MINOR 31 #define PCRE_PRERELEASE -#define PCRE_DATE 2013-12-15 +#define PCRE_DATE 2012-07-06 /* When an application links to a PCRE DLL in Windows, the symbols that are imported have to be identified as such. When building PCRE, the appropriate @@ -95,80 +95,54 @@ extern "C" { #endif -/* Public options. Some are compile-time only, some are run-time only, and some -are both. Most of the compile-time options are saved with the compiled regex so -that they can be inspected during studying (and therefore JIT compiling). Note -that pcre_study() has its own set of options. Originally, all the options -defined here used distinct bits. However, almost all the bits in a 32-bit word -are now used, so in order to conserve them, option bits that were previously -only recognized at matching time (i.e. by pcre_exec() or pcre_dfa_exec()) may -also be used for compile-time options that affect only compiling and are not -relevant for studying or JIT compiling. - -Some options for pcre_compile() change its behaviour but do not affect the -behaviour of the execution functions. Other options are passed through to the -execution functions and affect their behaviour, with or without affecting the -behaviour of pcre_compile(). - -Options that can be passed to pcre_compile() are tagged Cx below, with these -variants: - -C1 Affects compile only -C2 Does not affect compile; affects exec, dfa_exec -C3 Affects compile, exec, dfa_exec -C4 Affects compile, exec, dfa_exec, study -C5 Affects compile, exec, study - -Options that can be set for pcre_exec() and/or pcre_dfa_exec() are flagged with -E and D, respectively. They take precedence over C3, C4, and C5 settings passed -from pcre_compile(). Those that are compatible with JIT execution are flagged -with J. */ - -#define PCRE_CASELESS 0x00000001 /* C1 */ -#define PCRE_MULTILINE 0x00000002 /* C1 */ -#define PCRE_DOTALL 0x00000004 /* C1 */ -#define PCRE_EXTENDED 0x00000008 /* C1 */ -#define PCRE_ANCHORED 0x00000010 /* C4 E D */ -#define PCRE_DOLLAR_ENDONLY 0x00000020 /* C2 */ -#define PCRE_EXTRA 0x00000040 /* C1 */ -#define PCRE_NOTBOL 0x00000080 /* E D J */ -#define PCRE_NOTEOL 0x00000100 /* E D J */ -#define PCRE_UNGREEDY 0x00000200 /* C1 */ -#define PCRE_NOTEMPTY 0x00000400 /* E D J */ -#define PCRE_UTF8 0x00000800 /* C4 ) */ -#define PCRE_UTF16 0x00000800 /* C4 ) Synonyms */ -#define PCRE_UTF32 0x00000800 /* C4 ) */ -#define PCRE_NO_AUTO_CAPTURE 0x00001000 /* C1 */ -#define PCRE_NO_UTF8_CHECK 0x00002000 /* C1 E D J ) */ -#define PCRE_NO_UTF16_CHECK 0x00002000 /* C1 E D J ) Synonyms */ -#define PCRE_NO_UTF32_CHECK 0x00002000 /* C1 E D J ) */ -#define PCRE_AUTO_CALLOUT 0x00004000 /* C1 */ -#define PCRE_PARTIAL_SOFT 0x00008000 /* E D J ) Synonyms */ -#define PCRE_PARTIAL 0x00008000 /* E D J ) */ - -/* This pair use the same bit. */ -#define PCRE_NEVER_UTF 0x00010000 /* C1 ) Overlaid */ -#define PCRE_DFA_SHORTEST 0x00010000 /* D ) Overlaid */ - -/* This pair use the same bit. */ -#define PCRE_NO_AUTO_POSSESS 0x00020000 /* C1 ) Overlaid */ -#define PCRE_DFA_RESTART 0x00020000 /* D ) Overlaid */ +/* Options. Some are compile-time only, some are run-time only, and some are +both, so we keep them all distinct. However, almost all the bits in the options +word are now used. In the long run, we may have to re-use some of the +compile-time only bits for runtime options, or vice versa. In the comments +below, "compile", "exec", and "DFA exec" mean that the option is permitted to +be set for those functions; "used in" means that an option may be set only for +compile, but is subsequently referenced in exec and/or DFA exec. Any of the +compile-time options may be inspected during studying (and therefore JIT +compiling). */ -#define PCRE_FIRSTLINE 0x00040000 /* C3 */ -#define PCRE_DUPNAMES 0x00080000 /* C1 */ -#define PCRE_NEWLINE_CR 0x00100000 /* C3 E D */ -#define PCRE_NEWLINE_LF 0x00200000 /* C3 E D */ -#define PCRE_NEWLINE_CRLF 0x00300000 /* C3 E D */ -#define PCRE_NEWLINE_ANY 0x00400000 /* C3 E D */ -#define PCRE_NEWLINE_ANYCRLF 0x00500000 /* C3 E D */ -#define PCRE_BSR_ANYCRLF 0x00800000 /* C3 E D */ -#define PCRE_BSR_UNICODE 0x01000000 /* C3 E D */ -#define PCRE_JAVASCRIPT_COMPAT 0x02000000 /* C5 */ -#define PCRE_NO_START_OPTIMIZE 0x04000000 /* C2 E D ) Synonyms */ -#define PCRE_NO_START_OPTIMISE 0x04000000 /* C2 E D ) */ -#define PCRE_PARTIAL_HARD 0x08000000 /* E D J */ -#define PCRE_NOTEMPTY_ATSTART 0x10000000 /* E D J */ -#define PCRE_UCP 0x20000000 /* C3 */ +#define PCRE_CASELESS 0x00000001 /* Compile */ +#define PCRE_MULTILINE 0x00000002 /* Compile */ +#define PCRE_DOTALL 0x00000004 /* Compile */ +#define PCRE_EXTENDED 0x00000008 /* Compile */ +#define PCRE_ANCHORED 0x00000010 /* Compile, exec, DFA exec */ +#define PCRE_DOLLAR_ENDONLY 0x00000020 /* Compile, used in exec, DFA exec */ +#define PCRE_EXTRA 0x00000040 /* Compile */ +#define PCRE_NOTBOL 0x00000080 /* Exec, DFA exec */ +#define PCRE_NOTEOL 0x00000100 /* Exec, DFA exec */ +#define PCRE_UNGREEDY 0x00000200 /* Compile */ +#define PCRE_NOTEMPTY 0x00000400 /* Exec, DFA exec */ +/* The next two are also used in exec and DFA exec */ +#define PCRE_UTF8 0x00000800 /* Compile (same as PCRE_UTF16) */ +#define PCRE_UTF16 0x00000800 /* Compile (same as PCRE_UTF8) */ +#define PCRE_NO_AUTO_CAPTURE 0x00001000 /* Compile */ +/* The next two are also used in exec and DFA exec */ +#define PCRE_NO_UTF8_CHECK 0x00002000 /* Compile (same as PCRE_NO_UTF16_CHECK) */ +#define PCRE_NO_UTF16_CHECK 0x00002000 /* Compile (same as PCRE_NO_UTF8_CHECK) */ +#define PCRE_AUTO_CALLOUT 0x00004000 /* Compile */ +#define PCRE_PARTIAL_SOFT 0x00008000 /* Exec, DFA exec */ +#define PCRE_PARTIAL 0x00008000 /* Backwards compatible synonym */ +#define PCRE_DFA_SHORTEST 0x00010000 /* DFA exec */ +#define PCRE_DFA_RESTART 0x00020000 /* DFA exec */ +#define PCRE_FIRSTLINE 0x00040000 /* Compile, used in exec, DFA exec */ +#define PCRE_DUPNAMES 0x00080000 /* Compile */ +#define PCRE_NEWLINE_CR 0x00100000 /* Compile, exec, DFA exec */ +#define PCRE_NEWLINE_LF 0x00200000 /* Compile, exec, DFA exec */ +#define PCRE_NEWLINE_CRLF 0x00300000 /* Compile, exec, DFA exec */ +#define PCRE_NEWLINE_ANY 0x00400000 /* Compile, exec, DFA exec */ +#define PCRE_NEWLINE_ANYCRLF 0x00500000 /* Compile, exec, DFA exec */ +#define PCRE_BSR_ANYCRLF 0x00800000 /* Compile, exec, DFA exec */ +#define PCRE_BSR_UNICODE 0x01000000 /* Compile, exec, DFA exec */ +#define PCRE_JAVASCRIPT_COMPAT 0x02000000 /* Compile, used in exec */ +#define PCRE_NO_START_OPTIMIZE 0x04000000 /* Compile, exec, DFA exec */ +#define PCRE_NO_START_OPTIMISE 0x04000000 /* Synonym */ +#define PCRE_PARTIAL_HARD 0x08000000 /* Exec, DFA exec */ +#define PCRE_NOTEMPTY_ATSTART 0x10000000 /* Exec, DFA exec */ +#define PCRE_UCP 0x20000000 /* Compile, used in exec, DFA exec */ /* Exec-time and get/set-time error codes */ @@ -182,9 +156,8 @@ #define PCRE_ERROR_NOSUBSTRING (-7) #define PCRE_ERROR_MATCHLIMIT (-8) #define PCRE_ERROR_CALLOUT (-9) /* Never used by PCRE itself */ -#define PCRE_ERROR_BADUTF8 (-10) /* Same for 8/16/32 */ -#define PCRE_ERROR_BADUTF16 (-10) /* Same for 8/16/32 */ -#define PCRE_ERROR_BADUTF32 (-10) /* Same for 8/16/32 */ +#define PCRE_ERROR_BADUTF8 (-10) /* Same for 8/16 */ +#define PCRE_ERROR_BADUTF16 (-10) /* Same for 8/16 */ #define PCRE_ERROR_BADUTF8_OFFSET (-11) /* Same for 8/16 */ #define PCRE_ERROR_BADUTF16_OFFSET (-11) /* Same for 8/16 */ #define PCRE_ERROR_PARTIAL (-12) @@ -207,9 +180,6 @@ #define PCRE_ERROR_BADMODE (-28) #define PCRE_ERROR_BADENDIANNESS (-29) #define PCRE_ERROR_DFA_BADRESTART (-30) -#define PCRE_ERROR_JIT_BADOPTION (-31) -#define PCRE_ERROR_BADLENGTH (-32) -#define PCRE_ERROR_UNSET (-33) /* Specific error codes for UTF-8 validity checks */ @@ -235,7 +205,6 @@ #define PCRE_UTF8_ERR19 19 #define PCRE_UTF8_ERR20 20 #define PCRE_UTF8_ERR21 21 -#define PCRE_UTF8_ERR22 22 /* Unused (was non-character) */ /* Specific error codes for UTF-16 validity checks */ @@ -243,14 +212,7 @@ #define PCRE_UTF16_ERR1 1 #define PCRE_UTF16_ERR2 2 #define PCRE_UTF16_ERR3 3 -#define PCRE_UTF16_ERR4 4 /* Unused (was non-character) */ - -/* Specific error codes for UTF-32 validity checks */ - -#define PCRE_UTF32_ERR0 0 -#define PCRE_UTF32_ERR1 1 -#define PCRE_UTF32_ERR2 2 /* Unused (was non-character) */ -#define PCRE_UTF32_ERR3 3 +#define PCRE_UTF16_ERR4 4 /* Request types for pcre_fullinfo() */ @@ -274,13 +236,6 @@ #define PCRE_INFO_JIT 16 #define PCRE_INFO_JITSIZE 17 #define PCRE_INFO_MAXLOOKBEHIND 18 -#define PCRE_INFO_FIRSTCHARACTER 19 -#define PCRE_INFO_FIRSTCHARACTERFLAGS 20 -#define PCRE_INFO_REQUIREDCHAR 21 -#define PCRE_INFO_REQUIREDCHARFLAGS 22 -#define PCRE_INFO_MATCHLIMIT 23 -#define PCRE_INFO_RECURSIONLIMIT 24 -#define PCRE_INFO_MATCH_EMPTY 25 /* Request types for pcre_config(). Do not re-arrange, in order to remain compatible. */ @@ -297,8 +252,6 @@ #define PCRE_CONFIG_JIT 9 #define PCRE_CONFIG_UTF16 10 #define PCRE_CONFIG_JITTARGET 11 -#define PCRE_CONFIG_UTF32 12 -#define PCRE_CONFIG_PARENS_LIMIT 13 /* Request types for pcre_study(). Do not re-arrange, in order to remain compatible. */ @@ -306,9 +259,8 @@ #define PCRE_STUDY_JIT_COMPILE 0x0001 #define PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE 0x0002 #define PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE 0x0004 -#define PCRE_STUDY_EXTRA_NEEDED 0x0008 -/* Bit flags for the pcre[16|32]_extra structure. Do not re-arrange or redefine +/* Bit flags for the pcre[16]_extra structure. Do not re-arrange or redefine these bits, just add new ones on the end, in order to remain compatible. */ #define PCRE_EXTRA_STUDY_DATA 0x0001 @@ -327,18 +279,12 @@ struct real_pcre16; /* declaration; the definition is private */ typedef struct real_pcre16 pcre16; -struct real_pcre32; /* declaration; the definition is private */ -typedef struct real_pcre32 pcre32; - struct real_pcre_jit_stack; /* declaration; the definition is private */ typedef struct real_pcre_jit_stack pcre_jit_stack; struct real_pcre16_jit_stack; /* declaration; the definition is private */ typedef struct real_pcre16_jit_stack pcre16_jit_stack; -struct real_pcre32_jit_stack; /* declaration; the definition is private */ -typedef struct real_pcre32_jit_stack pcre32_jit_stack; - /* If PCRE is compiled with 16 bit character support, PCRE_UCHAR16 must contain a 16 bit wide signed data type. Otherwise it can be a dummy data type since pcre16 functions are not implemented. There is a check for this in pcre_internal.h. */ @@ -350,17 +296,6 @@ #define PCRE_SPTR16 const PCRE_UCHAR16 * #endif -/* If PCRE is compiled with 32 bit character support, PCRE_UCHAR32 must contain -a 32 bit wide signed data type. Otherwise it can be a dummy data type since -pcre32 functions are not implemented. There is a check for this in pcre_internal.h. */ -#ifndef PCRE_UCHAR32 -#define PCRE_UCHAR32 unsigned int -#endif - -#ifndef PCRE_SPTR32 -#define PCRE_SPTR32 const PCRE_UCHAR32 * -#endif - /* When PCRE is compiled as a C++ library, the subject pointer type can be replaced with a custom type. For conventional use, the public interface is a const char *. */ @@ -397,19 +332,6 @@ void *executable_jit; /* Contains a pointer to a compiled jit code */ } pcre16_extra; -/* Same structure as above, but with 32 bit char pointers. */ - -typedef struct pcre32_extra { - unsigned long int flags; /* Bits for which fields are set */ - void *study_data; /* Opaque data from pcre_study() */ - unsigned long int match_limit; /* Maximum number of calls to match() */ - void *callout_data; /* Data passed back in callouts */ - const unsigned char *tables; /* Pointer to character tables */ - unsigned long int match_limit_recursion; /* Max recursive calls to match() */ - PCRE_UCHAR32 **mark; /* For passing back a mark pointer */ - void *executable_jit; /* Contains a pointer to a compiled jit code */ -} pcre32_extra; - /* The structure for passing out data via the pcre_callout_function. We use a structure so that new fields can be added on the end in future versions, without changing the API of the function, thereby allowing old clients to work @@ -457,28 +379,6 @@ /* ------------------------------------------------------------------ */ } pcre16_callout_block; -/* Same structure as above, but with 32 bit char pointers. */ - -typedef struct pcre32_callout_block { - int version; /* Identifies version of block */ - /* ------------------------ Version 0 ------------------------------- */ - int callout_number; /* Number compiled into pattern */ - int *offset_vector; /* The offset vector */ - PCRE_SPTR32 subject; /* The subject being matched */ - int subject_length; /* The length of the subject */ - int start_match; /* Offset to start of this match attempt */ - int current_position; /* Where we currently are in the subject */ - int capture_top; /* Max current capture */ - int capture_last; /* Most recently closed capture */ - void *callout_data; /* Data passed in with the call */ - /* ------------------- Added for Version 1 -------------------------- */ - int pattern_position; /* Offset to next item in the pattern */ - int next_item_length; /* Length of next item in the pattern */ - /* ------------------- Added for Version 2 -------------------------- */ - const PCRE_UCHAR32 *mark; /* Pointer to current mark or NULL */ - /* ------------------------------------------------------------------ */ -} pcre32_callout_block; - /* Indirection for store get and free functions. These can be set to alternative malloc/free functions if required. Special ones are used in the non-recursive case for "frames". There is also an optional callout function @@ -497,12 +397,6 @@ PCRE_EXP_DECL void *(*pcre16_stack_malloc)(size_t); PCRE_EXP_DECL void (*pcre16_stack_free)(void *); PCRE_EXP_DECL int (*pcre16_callout)(pcre16_callout_block *); - -PCRE_EXP_DECL void *(*pcre32_malloc)(size_t); -PCRE_EXP_DECL void (*pcre32_free)(void *); -PCRE_EXP_DECL void *(*pcre32_stack_malloc)(size_t); -PCRE_EXP_DECL void (*pcre32_stack_free)(void *); -PCRE_EXP_DECL int (*pcre32_callout)(pcre32_callout_block *); #else /* VPCOMPAT */ PCRE_EXP_DECL void *pcre_malloc(size_t); PCRE_EXP_DECL void pcre_free(void *); @@ -515,19 +409,12 @@ PCRE_EXP_DECL void *pcre16_stack_malloc(size_t); PCRE_EXP_DECL void pcre16_stack_free(void *); PCRE_EXP_DECL int pcre16_callout(pcre16_callout_block *); - -PCRE_EXP_DECL void *pcre32_malloc(size_t); -PCRE_EXP_DECL void pcre32_free(void *); -PCRE_EXP_DECL void *pcre32_stack_malloc(size_t); -PCRE_EXP_DECL void pcre32_stack_free(void *); -PCRE_EXP_DECL int pcre32_callout(pcre32_callout_block *); #endif /* VPCOMPAT */ /* User defined callback which provides a stack just before the match starts. */ typedef pcre_jit_stack *(*pcre_jit_callback)(void *); typedef pcre16_jit_stack *(*pcre16_jit_callback)(void *); -typedef pcre32_jit_stack *(*pcre32_jit_callback)(void *); /* Exported PCRE functions */ @@ -535,134 +422,83 @@ const unsigned char *); PCRE_EXP_DECL pcre16 *pcre16_compile(PCRE_SPTR16, int, const char **, int *, const unsigned char *); -PCRE_EXP_DECL pcre32 *pcre32_compile(PCRE_SPTR32, int, const char **, int *, - const unsigned char *); PCRE_EXP_DECL pcre *pcre_compile2(const char *, int, int *, const char **, int *, const unsigned char *); PCRE_EXP_DECL pcre16 *pcre16_compile2(PCRE_SPTR16, int, int *, const char **, int *, const unsigned char *); -PCRE_EXP_DECL pcre32 *pcre32_compile2(PCRE_SPTR32, int, int *, const char **, - int *, const unsigned char *); PCRE_EXP_DECL int pcre_config(int, void *); PCRE_EXP_DECL int pcre16_config(int, void *); -PCRE_EXP_DECL int pcre32_config(int, void *); PCRE_EXP_DECL int pcre_copy_named_substring(const pcre *, const char *, int *, int, const char *, char *, int); PCRE_EXP_DECL int pcre16_copy_named_substring(const pcre16 *, PCRE_SPTR16, int *, int, PCRE_SPTR16, PCRE_UCHAR16 *, int); -PCRE_EXP_DECL int pcre32_copy_named_substring(const pcre32 *, PCRE_SPTR32, - int *, int, PCRE_SPTR32, PCRE_UCHAR32 *, int); PCRE_EXP_DECL int pcre_copy_substring(const char *, int *, int, int, char *, int); PCRE_EXP_DECL int pcre16_copy_substring(PCRE_SPTR16, int *, int, int, PCRE_UCHAR16 *, int); -PCRE_EXP_DECL int pcre32_copy_substring(PCRE_SPTR32, int *, int, int, - PCRE_UCHAR32 *, int); PCRE_EXP_DECL int pcre_dfa_exec(const pcre *, const pcre_extra *, const char *, int, int, int, int *, int , int *, int); PCRE_EXP_DECL int pcre16_dfa_exec(const pcre16 *, const pcre16_extra *, PCRE_SPTR16, int, int, int, int *, int , int *, int); -PCRE_EXP_DECL int pcre32_dfa_exec(const pcre32 *, const pcre32_extra *, - PCRE_SPTR32, int, int, int, int *, int , int *, int); PCRE_EXP_DECL int pcre_exec(const pcre *, const pcre_extra *, PCRE_SPTR, int, int, int, int *, int); PCRE_EXP_DECL int pcre16_exec(const pcre16 *, const pcre16_extra *, PCRE_SPTR16, int, int, int, int *, int); -PCRE_EXP_DECL int pcre32_exec(const pcre32 *, const pcre32_extra *, - PCRE_SPTR32, int, int, int, int *, int); -PCRE_EXP_DECL int pcre_jit_exec(const pcre *, const pcre_extra *, - PCRE_SPTR, int, int, int, int *, int, - pcre_jit_stack *); -PCRE_EXP_DECL int pcre16_jit_exec(const pcre16 *, const pcre16_extra *, - PCRE_SPTR16, int, int, int, int *, int, - pcre16_jit_stack *); -PCRE_EXP_DECL int pcre32_jit_exec(const pcre32 *, const pcre32_extra *, - PCRE_SPTR32, int, int, int, int *, int, - pcre32_jit_stack *); PCRE_EXP_DECL void pcre_free_substring(const char *); PCRE_EXP_DECL void pcre16_free_substring(PCRE_SPTR16); -PCRE_EXP_DECL void pcre32_free_substring(PCRE_SPTR32); PCRE_EXP_DECL void pcre_free_substring_list(const char **); PCRE_EXP_DECL void pcre16_free_substring_list(PCRE_SPTR16 *); -PCRE_EXP_DECL void pcre32_free_substring_list(PCRE_SPTR32 *); PCRE_EXP_DECL int pcre_fullinfo(const pcre *, const pcre_extra *, int, void *); PCRE_EXP_DECL int pcre16_fullinfo(const pcre16 *, const pcre16_extra *, int, void *); -PCRE_EXP_DECL int pcre32_fullinfo(const pcre32 *, const pcre32_extra *, int, - void *); PCRE_EXP_DECL int pcre_get_named_substring(const pcre *, const char *, int *, int, const char *, const char **); PCRE_EXP_DECL int pcre16_get_named_substring(const pcre16 *, PCRE_SPTR16, int *, int, PCRE_SPTR16, PCRE_SPTR16 *); -PCRE_EXP_DECL int pcre32_get_named_substring(const pcre32 *, PCRE_SPTR32, - int *, int, PCRE_SPTR32, PCRE_SPTR32 *); PCRE_EXP_DECL int pcre_get_stringnumber(const pcre *, const char *); PCRE_EXP_DECL int pcre16_get_stringnumber(const pcre16 *, PCRE_SPTR16); -PCRE_EXP_DECL int pcre32_get_stringnumber(const pcre32 *, PCRE_SPTR32); PCRE_EXP_DECL int pcre_get_stringtable_entries(const pcre *, const char *, char **, char **); PCRE_EXP_DECL int pcre16_get_stringtable_entries(const pcre16 *, PCRE_SPTR16, PCRE_UCHAR16 **, PCRE_UCHAR16 **); -PCRE_EXP_DECL int pcre32_get_stringtable_entries(const pcre32 *, PCRE_SPTR32, - PCRE_UCHAR32 **, PCRE_UCHAR32 **); PCRE_EXP_DECL int pcre_get_substring(const char *, int *, int, int, const char **); PCRE_EXP_DECL int pcre16_get_substring(PCRE_SPTR16, int *, int, int, PCRE_SPTR16 *); -PCRE_EXP_DECL int pcre32_get_substring(PCRE_SPTR32, int *, int, int, - PCRE_SPTR32 *); PCRE_EXP_DECL int pcre_get_substring_list(const char *, int *, int, const char ***); PCRE_EXP_DECL int pcre16_get_substring_list(PCRE_SPTR16, int *, int, PCRE_SPTR16 **); -PCRE_EXP_DECL int pcre32_get_substring_list(PCRE_SPTR32, int *, int, - PCRE_SPTR32 **); PCRE_EXP_DECL const unsigned char *pcre_maketables(void); PCRE_EXP_DECL const unsigned char *pcre16_maketables(void); -PCRE_EXP_DECL const unsigned char *pcre32_maketables(void); PCRE_EXP_DECL int pcre_refcount(pcre *, int); PCRE_EXP_DECL int pcre16_refcount(pcre16 *, int); -PCRE_EXP_DECL int pcre32_refcount(pcre32 *, int); PCRE_EXP_DECL pcre_extra *pcre_study(const pcre *, int, const char **); PCRE_EXP_DECL pcre16_extra *pcre16_study(const pcre16 *, int, const char **); -PCRE_EXP_DECL pcre32_extra *pcre32_study(const pcre32 *, int, const char **); PCRE_EXP_DECL void pcre_free_study(pcre_extra *); PCRE_EXP_DECL void pcre16_free_study(pcre16_extra *); -PCRE_EXP_DECL void pcre32_free_study(pcre32_extra *); PCRE_EXP_DECL const char *pcre_version(void); PCRE_EXP_DECL const char *pcre16_version(void); -PCRE_EXP_DECL const char *pcre32_version(void); /* Utility functions for byte order swaps. */ PCRE_EXP_DECL int pcre_pattern_to_host_byte_order(pcre *, pcre_extra *, const unsigned char *); PCRE_EXP_DECL int pcre16_pattern_to_host_byte_order(pcre16 *, pcre16_extra *, const unsigned char *); -PCRE_EXP_DECL int pcre32_pattern_to_host_byte_order(pcre32 *, pcre32_extra *, - const unsigned char *); PCRE_EXP_DECL int pcre16_utf16_to_host_byte_order(PCRE_UCHAR16 *, PCRE_SPTR16, int, int *, int); -PCRE_EXP_DECL int pcre32_utf32_to_host_byte_order(PCRE_UCHAR32 *, - PCRE_SPTR32, int, int *, int); /* JIT compiler related functions. */ PCRE_EXP_DECL pcre_jit_stack *pcre_jit_stack_alloc(int, int); PCRE_EXP_DECL pcre16_jit_stack *pcre16_jit_stack_alloc(int, int); -PCRE_EXP_DECL pcre32_jit_stack *pcre32_jit_stack_alloc(int, int); PCRE_EXP_DECL void pcre_jit_stack_free(pcre_jit_stack *); PCRE_EXP_DECL void pcre16_jit_stack_free(pcre16_jit_stack *); -PCRE_EXP_DECL void pcre32_jit_stack_free(pcre32_jit_stack *); PCRE_EXP_DECL void pcre_assign_jit_stack(pcre_extra *, pcre_jit_callback, void *); PCRE_EXP_DECL void pcre16_assign_jit_stack(pcre16_extra *, pcre16_jit_callback, void *); -PCRE_EXP_DECL void pcre32_assign_jit_stack(pcre32_extra *, - pcre32_jit_callback, void *); -PCRE_EXP_DECL void pcre_jit_free_unused_memory(void); -PCRE_EXP_DECL void pcre16_jit_free_unused_memory(void); -PCRE_EXP_DECL void pcre32_jit_free_unused_memory(void); #ifdef __cplusplus } /* extern "C" */ diff --git a/pcre/pcre_byte_order.c b/pcre/pcre_byte_order.c --- a/pcre/pcre_byte_order.c +++ b/pcre/pcre_byte_order.c @@ -6,7 +6,7 @@ and semantics are as close as possible to those of the Perl 5 language. Written by Philip Hazel - Copyright (c) 1997-2013 University of Cambridge + Copyright (c) 1997-2012 University of Cambridge ----------------------------------------------------------------------------- Redistribution and use in source and binary forms, with or without @@ -95,15 +95,12 @@ Returns: 0 if the swap is successful, negative on error */ -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 PCRE_EXP_DECL int pcre_pattern_to_host_byte_order(pcre *argument_re, pcre_extra *extra_data, const unsigned char *tables) -#elif defined COMPILE_PCRE16 +#else PCRE_EXP_DECL int pcre16_pattern_to_host_byte_order(pcre16 *argument_re, pcre16_extra *extra_data, const unsigned char *tables) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DECL int pcre32_pattern_to_host_byte_order(pcre32 *argument_re, - pcre32_extra *extra_data, const unsigned char *tables) #endif { REAL_PCRE *re = (REAL_PCRE *)argument_re; @@ -111,10 +108,10 @@ #ifndef COMPILE_PCRE8 pcre_uchar *ptr; int length; -#if defined SUPPORT_UTF && defined COMPILE_PCRE16 +#ifdef SUPPORT_UTF BOOL utf; BOOL utf16_char; -#endif /* SUPPORT_UTF && COMPILE_PCRE16 */ +#endif /* SUPPORT_UTF */ #endif /* !COMPILE_PCRE8 */ if (re == NULL) return PCRE_ERROR_NULL; @@ -126,26 +123,16 @@ } if (re->magic_number != REVERSED_MAGIC_NUMBER) return PCRE_ERROR_BADMAGIC; -if ((swap_uint32(re->flags) & PCRE_MODE) == 0) return PCRE_ERROR_BADMODE; +if ((swap_uint16(re->flags) & PCRE_MODE) == 0) return PCRE_ERROR_BADMODE; re->magic_number = MAGIC_NUMBER; re->size = swap_uint32(re->size); re->options = swap_uint32(re->options); -re->flags = swap_uint32(re->flags); -re->limit_match = swap_uint32(re->limit_match); -re->limit_recursion = swap_uint32(re->limit_recursion); - -#if defined COMPILE_PCRE8 || defined COMPILE_PCRE16 -re->first_char = swap_uint16(re->first_char); -re->req_char = swap_uint16(re->req_char); -#elif defined COMPILE_PCRE32 -re->first_char = swap_uint32(re->first_char); -re->req_char = swap_uint32(re->req_char); -#endif - -re->max_lookbehind = swap_uint16(re->max_lookbehind); +re->flags = swap_uint16(re->flags); re->top_bracket = swap_uint16(re->top_bracket); re->top_backref = swap_uint16(re->top_backref); +re->first_char = swap_uint16(re->first_char); +re->req_char = swap_uint16(re->req_char); re->name_table_offset = swap_uint16(re->name_table_offset); re->name_entry_size = swap_uint16(re->name_entry_size); re->name_count = swap_uint16(re->name_count); @@ -163,24 +150,20 @@ #ifndef COMPILE_PCRE8 ptr = (pcre_uchar *)re + re->name_table_offset; length = re->name_count * re->name_entry_size; -#if defined SUPPORT_UTF && defined COMPILE_PCRE16 +#ifdef SUPPORT_UTF utf = (re->options & PCRE_UTF16) != 0; utf16_char = FALSE; -#endif /* SUPPORT_UTF && COMPILE_PCRE16 */ +#endif while(TRUE) { /* Swap previous characters. */ while (length-- > 0) { -#if defined COMPILE_PCRE16 *ptr = swap_uint16(*ptr); -#elif defined COMPILE_PCRE32 - *ptr = swap_uint32(*ptr); -#endif ptr++; } -#if defined SUPPORT_UTF && defined COMPILE_PCRE16 +#ifdef SUPPORT_UTF if (utf16_char) { if (HAS_EXTRALEN(ptr[-1])) @@ -195,17 +178,13 @@ /* Get next opcode. */ length = 0; -#if defined COMPILE_PCRE16 *ptr = swap_uint16(*ptr); -#elif defined COMPILE_PCRE32 - *ptr = swap_uint32(*ptr); -#endif switch (*ptr) { case OP_END: return 0; -#if defined SUPPORT_UTF && defined COMPILE_PCRE16 +#ifdef SUPPORT_UTF case OP_CHAR: case OP_CHARI: case OP_NOT: @@ -280,26 +259,16 @@ case OP_XCLASS: /* Reverse the size of the XCLASS instance. */ ptr++; -#if defined COMPILE_PCRE16 *ptr = swap_uint16(*ptr); -#elif defined COMPILE_PCRE32 - *ptr = swap_uint32(*ptr); -#endif -#ifndef COMPILE_PCRE32 if (LINK_SIZE > 1) { /* LINK_SIZE can be 1 or 2 in 16 bit mode. */ ptr++; *ptr = swap_uint16(*ptr); } -#endif ptr++; length = (GET(ptr, -LINK_SIZE)) - (1 + LINK_SIZE + 1); -#if defined COMPILE_PCRE16 *ptr = swap_uint16(*ptr); -#elif defined COMPILE_PCRE32 - *ptr = swap_uint32(*ptr); -#endif if ((*ptr & XCL_MAP) != 0) { /* Skip the character bit map. */ @@ -310,7 +279,7 @@ } ptr++; } -/* Control should never reach here in 16/32 bit mode. */ +/* Control should never reach here in 16 bit mode. */ #endif /* !COMPILE_PCRE8 */ return 0; diff --git a/pcre/pcre_compile.c b/pcre/pcre_compile.c --- a/pcre/pcre_compile.c +++ b/pcre/pcre_compile.c @@ -6,9428 +6,8204 @@ and semantics are as close as possible to those of the Perl 5 language. Written by Philip Hazel - Copyright (c) 1997-2013 University of Cambridge - ------------------------------------------------------------------------------ -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. - - * Neither the name of the University of Cambridge 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 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. ------------------------------------------------------------------------------ -*/ - - -/* This module contains the external function pcre_compile(), along with -supporting internal functions that are not used by other modules. */ - - -#ifdef HAVE_CONFIG_H -#include "config.h" -#endif - -#define NLBLOCK cd /* Block containing newline information */ -#define PSSTART start_pattern /* Field containing processed string start */ -#define PSEND end_pattern /* Field containing processed string end */ - -#include "pcre_internal.h" - - -/* When PCRE_DEBUG is defined, we need the pcre(16|32)_printint() function, which -is also used by pcretest. PCRE_DEBUG is not defined when building a production -library. We do not need to select pcre16_printint.c specially, because the -COMPILE_PCREx macro will already be appropriately set. */ - -#ifdef PCRE_DEBUG -/* pcre_printint.c should not include any headers */ -#define PCRE_INCLUDED -#include "pcre_printint.c" -#undef PCRE_INCLUDED -#endif - - -/* Macro for setting individual bits in class bitmaps. */ - -#define SETBIT(a,b) a[(b)/8] |= (1 << ((b)&7)) - -/* Maximum length value to check against when making sure that the integer that -holds the compiled pattern length does not overflow. We make it a bit less than -INT_MAX to allow for adding in group terminating bytes, so that we don't have -to check them every time. */ - -#define OFLOW_MAX (INT_MAX - 20) - -/* Definitions to allow mutual recursion */ - -static int - add_list_to_class(pcre_uint8 *, pcre_uchar **, int, compile_data *, - const pcre_uint32 *, unsigned int); - -static BOOL - compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL, int, int, - pcre_uint32 *, pcre_int32 *, pcre_uint32 *, pcre_int32 *, branch_chain *, - compile_data *, int *); - - - -/************************************************* -* Code parameters and static tables * -*************************************************/ - -/* This value specifies the size of stack workspace that is used during the -first pre-compile phase that determines how much memory is required. The regex -is partly compiled into this space, but the compiled parts are discarded as -soon as they can be, so that hopefully there will never be an overrun. The code -does, however, check for an overrun. The largest amount I've seen used is 218, -so this number is very generous. - -The same workspace is used during the second, actual compile phase for -remembering forward references to groups so that they can be filled in at the -end. Each entry in this list occupies LINK_SIZE bytes, so even when LINK_SIZE -is 4 there is plenty of room for most patterns. However, the memory can get -filled up by repetitions of forward references, for example patterns like -/(?1){0,1999}(b)/, and one user did hit the limit. The code has been changed so -that the workspace is expanded using malloc() in this situation. The value -below is therefore a minimum, and we put a maximum on it for safety. The -minimum is now also defined in terms of LINK_SIZE so that the use of malloc() -kicks in at the same number of forward references in all cases. */ - -#define COMPILE_WORK_SIZE (2048*LINK_SIZE) -#define COMPILE_WORK_SIZE_MAX (100*COMPILE_WORK_SIZE) - -/* This value determines the size of the initial vector that is used for -remembering named groups during the pre-compile. It is allocated on the stack, -but if it is too small, it is expanded using malloc(), in a similar way to the -workspace. The value is the number of slots in the list. */ - -#define NAMED_GROUP_LIST_SIZE 20 - -/* The overrun tests check for a slightly smaller size so that they detect the -overrun before it actually does run off the end of the data block. */ - -#define WORK_SIZE_SAFETY_MARGIN (100) - -/* Private flags added to firstchar and reqchar. */ - -#define REQ_CASELESS (1 << 0) /* Indicates caselessness */ -#define REQ_VARY (1 << 1) /* Reqchar followed non-literal item */ -/* Negative values for the firstchar and reqchar flags */ -#define REQ_UNSET (-2) -#define REQ_NONE (-1) - -/* Repeated character flags. */ - -#define UTF_LENGTH 0x10000000l /* The char contains its length. */ - -/* Table for handling escaped characters in the range '0'-'z'. Positive returns -are simple data values; negative values are for special things like \d and so -on. Zero means further processing is needed (for things like \x), or the escape -is invalid. */ - -#ifndef EBCDIC - -/* This is the "normal" table for ASCII systems or for EBCDIC systems running -in UTF-8 mode. */ - -static const short int escapes[] = { - 0, 0, - 0, 0, - 0, 0, - 0, 0, - 0, 0, - CHAR_COLON, CHAR_SEMICOLON, - CHAR_LESS_THAN_SIGN, CHAR_EQUALS_SIGN, - CHAR_GREATER_THAN_SIGN, CHAR_QUESTION_MARK, - CHAR_COMMERCIAL_AT, -ESC_A, - -ESC_B, -ESC_C, - -ESC_D, -ESC_E, - 0, -ESC_G, - -ESC_H, 0, - 0, -ESC_K, - 0, 0, - -ESC_N, 0, - -ESC_P, -ESC_Q, - -ESC_R, -ESC_S, - 0, 0, - -ESC_V, -ESC_W, - -ESC_X, 0, - -ESC_Z, CHAR_LEFT_SQUARE_BRACKET, - CHAR_BACKSLASH, CHAR_RIGHT_SQUARE_BRACKET, - CHAR_CIRCUMFLEX_ACCENT, CHAR_UNDERSCORE, - CHAR_GRAVE_ACCENT, 7, - -ESC_b, 0, - -ESC_d, ESC_e, - ESC_f, 0, - -ESC_h, 0, - 0, -ESC_k, - 0, 0, - ESC_n, 0, - -ESC_p, 0, - ESC_r, -ESC_s, - ESC_tee, 0, - -ESC_v, -ESC_w, - 0, 0, - -ESC_z -}; - -#else - -/* This is the "abnormal" table for EBCDIC systems without UTF-8 support. */ - -static const short int escapes[] = { -/* 48 */ 0, 0, 0, '.', '<', '(', '+', '|', -/* 50 */ '&', 0, 0, 0, 0, 0, 0, 0, -/* 58 */ 0, 0, '!', '$', '*', ')', ';', '~', -/* 60 */ '-', '/', 0, 0, 0, 0, 0, 0, -/* 68 */ 0, 0, '|', ',', '%', '_', '>', '?', -/* 70 */ 0, 0, 0, 0, 0, 0, 0, 0, -/* 78 */ 0, '`', ':', '#', '@', '\'', '=', '"', -/* 80 */ 0, 7, -ESC_b, 0, -ESC_d, ESC_e, ESC_f, 0, -/* 88 */-ESC_h, 0, 0, '{', 0, 0, 0, 0, -/* 90 */ 0, 0, -ESC_k, 'l', 0, ESC_n, 0, -ESC_p, -/* 98 */ 0, ESC_r, 0, '}', 0, 0, 0, 0, -/* A0 */ 0, '~', -ESC_s, ESC_tee, 0,-ESC_v, -ESC_w, 0, -/* A8 */ 0,-ESC_z, 0, 0, 0, '[', 0, 0, -/* B0 */ 0, 0, 0, 0, 0, 0, 0, 0, -/* B8 */ 0, 0, 0, 0, 0, ']', '=', '-', -/* C0 */ '{',-ESC_A, -ESC_B, -ESC_C, -ESC_D,-ESC_E, 0, -ESC_G, -/* C8 */-ESC_H, 0, 0, 0, 0, 0, 0, 0, -/* D0 */ '}', 0, -ESC_K, 0, 0,-ESC_N, 0, -ESC_P, -/* D8 */-ESC_Q,-ESC_R, 0, 0, 0, 0, 0, 0, -/* E0 */ '\\', 0, -ESC_S, 0, 0,-ESC_V, -ESC_W, -ESC_X, -/* E8 */ 0,-ESC_Z, 0, 0, 0, 0, 0, 0, -/* F0 */ 0, 0, 0, 0, 0, 0, 0, 0, -/* F8 */ 0, 0, 0, 0, 0, 0, 0, 0 -}; -#endif - - -/* Table of special "verbs" like (*PRUNE). This is a short table, so it is -searched linearly. Put all the names into a single string, in order to reduce -the number of relocations when a shared library is dynamically linked. The -string is built from string macros so that it works in UTF-8 mode on EBCDIC -platforms. */ - -typedef struct verbitem { - int len; /* Length of verb name */ - int op; /* Op when no arg, or -1 if arg mandatory */ - int op_arg; /* Op when arg present, or -1 if not allowed */ -} verbitem; - -static const char verbnames[] = - "\0" /* Empty name is a shorthand for MARK */ - STRING_MARK0 - STRING_ACCEPT0 - STRING_COMMIT0 - STRING_F0 - STRING_FAIL0 - STRING_PRUNE0 - STRING_SKIP0 - STRING_THEN; - -static const verbitem verbs[] = { - { 0, -1, OP_MARK }, - { 4, -1, OP_MARK }, - { 6, OP_ACCEPT, -1 }, - { 6, OP_COMMIT, -1 }, - { 1, OP_FAIL, -1 }, - { 4, OP_FAIL, -1 }, - { 5, OP_PRUNE, OP_PRUNE_ARG }, - { 4, OP_SKIP, OP_SKIP_ARG }, - { 4, OP_THEN, OP_THEN_ARG } -}; - -static const int verbcount = sizeof(verbs)/sizeof(verbitem); - - -/* Substitutes for [[:<:]] and [[:>:]], which mean start and end of word in -another regex library. */ - -static const pcre_uchar sub_start_of_word[] = { - CHAR_BACKSLASH, CHAR_b, CHAR_LEFT_PARENTHESIS, CHAR_QUESTION_MARK, - CHAR_EQUALS_SIGN, CHAR_BACKSLASH, CHAR_w, CHAR_RIGHT_PARENTHESIS, '\0' }; - -static const pcre_uchar sub_end_of_word[] = { - CHAR_BACKSLASH, CHAR_b, CHAR_LEFT_PARENTHESIS, CHAR_QUESTION_MARK, - CHAR_LESS_THAN_SIGN, CHAR_EQUALS_SIGN, CHAR_BACKSLASH, CHAR_w, - CHAR_RIGHT_PARENTHESIS, '\0' }; - - -/* Tables of names of POSIX character classes and their lengths. The names are -now all in a single string, to reduce the number of relocations when a shared -library is dynamically loaded. The list of lengths is terminated by a zero -length entry. The first three must be alpha, lower, upper, as this is assumed -for handling case independence. The indices for graph, print, and punct are -needed, so identify them. */ - -static const char posix_names[] = - STRING_alpha0 STRING_lower0 STRING_upper0 STRING_alnum0 - STRING_ascii0 STRING_blank0 STRING_cntrl0 STRING_digit0 - STRING_graph0 STRING_print0 STRING_punct0 STRING_space0 - STRING_word0 STRING_xdigit; - -static const pcre_uint8 posix_name_lengths[] = { - 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 }; - -#define PC_GRAPH 8 -#define PC_PRINT 9 -#define PC_PUNCT 10 - - -/* Table of class bit maps for each POSIX class. Each class is formed from a -base map, with an optional addition or removal of another map. Then, for some -classes, there is some additional tweaking: for [:blank:] the vertical space -characters are removed, and for [:alpha:] and [:alnum:] the underscore -character is removed. The triples in the table consist of the base map offset, -second map offset or -1 if no second map, and a non-negative value for map -addition or a negative value for map subtraction (if there are two maps). The -absolute value of the third field has these meanings: 0 => no tweaking, 1 => -remove vertical space characters, 2 => remove underscore. */ - -static const int posix_class_maps[] = { - cbit_word, cbit_digit, -2, /* alpha */ - cbit_lower, -1, 0, /* lower */ - cbit_upper, -1, 0, /* upper */ - cbit_word, -1, 2, /* alnum - word without underscore */ - cbit_print, cbit_cntrl, 0, /* ascii */ - cbit_space, -1, 1, /* blank - a GNU extension */ - cbit_cntrl, -1, 0, /* cntrl */ - cbit_digit, -1, 0, /* digit */ - cbit_graph, -1, 0, /* graph */ - cbit_print, -1, 0, /* print */ - cbit_punct, -1, 0, /* punct */ - cbit_space, -1, 0, /* space */ - cbit_word, -1, 0, /* word - a Perl extension */ - cbit_xdigit,-1, 0 /* xdigit */ -}; - -/* Table of substitutes for \d etc when PCRE_UCP is set. They are replaced by -Unicode property escapes. */ - -#ifdef SUPPORT_UCP -static const pcre_uchar string_PNd[] = { - CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, - CHAR_N, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' }; -static const pcre_uchar string_pNd[] = { - CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, - CHAR_N, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' }; -static const pcre_uchar string_PXsp[] = { - CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, - CHAR_X, CHAR_s, CHAR_p, CHAR_RIGHT_CURLY_BRACKET, '\0' }; -static const pcre_uchar string_pXsp[] = { - CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, - CHAR_X, CHAR_s, CHAR_p, CHAR_RIGHT_CURLY_BRACKET, '\0' }; -static const pcre_uchar string_PXwd[] = { - CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, - CHAR_X, CHAR_w, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' }; -static const pcre_uchar string_pXwd[] = { - CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, - CHAR_X, CHAR_w, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' }; - -static const pcre_uchar *substitutes[] = { - string_PNd, /* \D */ - string_pNd, /* \d */ - string_PXsp, /* \S */ /* Xsp is Perl space, but from 8.34, Perl */ - string_pXsp, /* \s */ /* space and POSIX space are the same. */ - string_PXwd, /* \W */ - string_pXwd /* \w */ -}; - -/* The POSIX class substitutes must be in the order of the POSIX class names, -defined above, and there are both positive and negative cases. NULL means no -general substitute of a Unicode property escape (\p or \P). However, for some -POSIX classes (e.g. graph, print, punct) a special property code is compiled -directly. */ - -static const pcre_uchar string_pL[] = { - CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, - CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' }; -static const pcre_uchar string_pLl[] = { - CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, - CHAR_L, CHAR_l, CHAR_RIGHT_CURLY_BRACKET, '\0' }; -static const pcre_uchar string_pLu[] = { - CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, - CHAR_L, CHAR_u, CHAR_RIGHT_CURLY_BRACKET, '\0' }; -static const pcre_uchar string_pXan[] = { - CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, - CHAR_X, CHAR_a, CHAR_n, CHAR_RIGHT_CURLY_BRACKET, '\0' }; -static const pcre_uchar string_h[] = { - CHAR_BACKSLASH, CHAR_h, '\0' }; -static const pcre_uchar string_pXps[] = { - CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, - CHAR_X, CHAR_p, CHAR_s, CHAR_RIGHT_CURLY_BRACKET, '\0' }; -static const pcre_uchar string_PL[] = { - CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, - CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' }; -static const pcre_uchar string_PLl[] = { - CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, - CHAR_L, CHAR_l, CHAR_RIGHT_CURLY_BRACKET, '\0' }; -static const pcre_uchar string_PLu[] = { - CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, - CHAR_L, CHAR_u, CHAR_RIGHT_CURLY_BRACKET, '\0' }; -static const pcre_uchar string_PXan[] = { - CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, - CHAR_X, CHAR_a, CHAR_n, CHAR_RIGHT_CURLY_BRACKET, '\0' }; -static const pcre_uchar string_H[] = { - CHAR_BACKSLASH, CHAR_H, '\0' }; -static const pcre_uchar string_PXps[] = { - CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, - CHAR_X, CHAR_p, CHAR_s, CHAR_RIGHT_CURLY_BRACKET, '\0' }; - -static const pcre_uchar *posix_substitutes[] = { - string_pL, /* alpha */ - string_pLl, /* lower */ - string_pLu, /* upper */ - string_pXan, /* alnum */ - NULL, /* ascii */ - string_h, /* blank */ - NULL, /* cntrl */ - string_pNd, /* digit */ - NULL, /* graph */ - NULL, /* print */ - NULL, /* punct */ - string_pXps, /* space */ /* Xps is POSIX space, but from 8.34 */ - string_pXwd, /* word */ /* Perl and POSIX space are the same */ - NULL, /* xdigit */ - /* Negated cases */ - string_PL, /* ^alpha */ - string_PLl, /* ^lower */ - string_PLu, /* ^upper */ - string_PXan, /* ^alnum */ - NULL, /* ^ascii */ - string_H, /* ^blank */ - NULL, /* ^cntrl */ - string_PNd, /* ^digit */ - NULL, /* ^graph */ - NULL, /* ^print */ - NULL, /* ^punct */ - string_PXps, /* ^space */ /* Xps is POSIX space, but from 8.34 */ - string_PXwd, /* ^word */ /* Perl and POSIX space are the same */ - NULL /* ^xdigit */ -}; -#define POSIX_SUBSIZE (sizeof(posix_substitutes) / sizeof(pcre_uchar *)) -#endif - -#define STRING(a) # a -#define XSTRING(s) STRING(s) - -/* The texts of compile-time error messages. These are "char *" because they -are passed to the outside world. Do not ever re-use any error number, because -they are documented. Always add a new error instead. Messages marked DEAD below -are no longer used. This used to be a table of strings, but in order to reduce -the number of relocations needed when a shared library is loaded dynamically, -it is now one long string. We cannot use a table of offsets, because the -lengths of inserts such as XSTRING(MAX_NAME_SIZE) are not known. Instead, we -simply count through to the one we want - this isn't a performance issue -because these strings are used only when there is a compilation error. - -Each substring ends with \0 to insert a null character. This includes the final -substring, so that the whole string ends with \0\0, which can be detected when -counting through. */ - -static const char error_texts[] = - "no error\0" - "\\ at end of pattern\0" - "\\c at end of pattern\0" - "unrecognized character follows \\\0" - "numbers out of order in {} quantifier\0" - /* 5 */ - "number too big in {} quantifier\0" - "missing terminating ] for character class\0" - "invalid escape sequence in character class\0" - "range out of order in character class\0" - "nothing to repeat\0" - /* 10 */ - "operand of unlimited repeat could match the empty string\0" /** DEAD **/ - "internal error: unexpected repeat\0" - "unrecognized character after (? or (?-\0" - "POSIX named classes are supported only within a class\0" - "missing )\0" - /* 15 */ - "reference to non-existent subpattern\0" - "erroffset passed as NULL\0" - "unknown option bit(s) set\0" - "missing ) after comment\0" - "parentheses nested too deeply\0" /** DEAD **/ - /* 20 */ - "regular expression is too large\0" - "failed to get memory\0" - "unmatched parentheses\0" - "internal error: code overflow\0" - "unrecognized character after (?<\0" - /* 25 */ - "lookbehind assertion is not fixed length\0" - "malformed number or name after (?(\0" - "conditional group contains more than two branches\0" - "assertion expected after (?(\0" - "(?R or (?[+-]digits must be followed by )\0" - /* 30 */ - "unknown POSIX class name\0" - "POSIX collating elements are not supported\0" - "this version of PCRE is compiled without UTF support\0" - "spare error\0" /** DEAD **/ - "character value in \\x{} or \\o{} is too large\0" - /* 35 */ - "invalid condition (?(0)\0" - "\\C not allowed in lookbehind assertion\0" - "PCRE does not support \\L, \\l, \\N{name}, \\U, or \\u\0" - "number after (?C is > 255\0" - "closing ) for (?C expected\0" - /* 40 */ - "recursive call could loop indefinitely\0" - "unrecognized character after (?P\0" - "syntax error in subpattern name (missing terminator)\0" - "two named subpatterns have the same name\0" - "invalid UTF-8 string\0" - /* 45 */ - "support for \\P, \\p, and \\X has not been compiled\0" - "malformed \\P or \\p sequence\0" - "unknown property name after \\P or \\p\0" - "subpattern name is too long (maximum " XSTRING(MAX_NAME_SIZE) " characters)\0" - "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0" - /* 50 */ - "repeated subpattern is too long\0" /** DEAD **/ - "octal value is greater than \\377 in 8-bit non-UTF-8 mode\0" - "internal error: overran compiling workspace\0" - "internal error: previously-checked referenced subpattern not found\0" - "DEFINE group contains more than one branch\0" - /* 55 */ - "repeating a DEFINE group is not allowed\0" /** DEAD **/ - "inconsistent NEWLINE options\0" - "\\g is not followed by a braced, angle-bracketed, or quoted name/number or by a plain number\0" - "a numbered reference must not be zero\0" - "an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)\0" - /* 60 */ - "(*VERB) not recognized or malformed\0" - "number is too big\0" - "subpattern name expected\0" - "digit expected after (?+\0" - "] is an invalid data character in JavaScript compatibility mode\0" - /* 65 */ - "different names for subpatterns of the same number are not allowed\0" - "(*MARK) must have an argument\0" - "this version of PCRE is not compiled with Unicode property support\0" - "\\c must be followed by an ASCII character\0" - "\\k is not followed by a braced, angle-bracketed, or quoted name\0" - /* 70 */ - "internal error: unknown opcode in find_fixedlength()\0" - "\\N is not supported in a class\0" - "too many forward references\0" - "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0" - "invalid UTF-16 string\0" - /* 75 */ - "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0" - "character value in \\u.... sequence is too large\0" - "invalid UTF-32 string\0" - "setting UTF is disabled by the application\0" - "non-hex character in \\x{} (closing brace missing?)\0" - /* 80 */ - "non-octal character in \\o{} (closing brace missing?)\0" - "missing opening brace after \\o\0" - "parentheses are too deeply nested\0" - "invalid range in character class\0" - "group name must start with a non-digit\0" - ; - -/* Table to identify digits and hex digits. This is used when compiling -patterns. Note that the tables in chartables are dependent on the locale, and -may mark arbitrary characters as digits - but the PCRE compiling code expects -to handle only 0-9, a-z, and A-Z as digits when compiling. That is why we have -a private table here. It costs 256 bytes, but it is a lot faster than doing -character value tests (at least in some simple cases I timed), and in some -applications one wants PCRE to compile efficiently as well as match -efficiently. - -For convenience, we use the same bit definitions as in chartables: - - 0x04 decimal digit - 0x08 hexadecimal digit - -Then we can use ctype_digit and ctype_xdigit in the code. */ - -/* Using a simple comparison for decimal numbers rather than a memory read -is much faster, and the resulting code is simpler (the compiler turns it -into a subtraction and unsigned comparison). */ - -#define IS_DIGIT(x) ((x) >= CHAR_0 && (x) <= CHAR_9) - -#ifndef EBCDIC - -/* This is the "normal" case, for ASCII systems, and EBCDIC systems running in -UTF-8 mode. */ - -static const pcre_uint8 digitab[] = - { - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 0- 7 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 8- 15 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 16- 23 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 24- 31 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* - ' */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* ( - / */ - 0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /* 0 - 7 */ - 0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00, /* 8 - ? */ - 0x00,0x08,0x08,0x08,0x08,0x08,0x08,0x00, /* @ - G */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* H - O */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* P - W */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* X - _ */ - 0x00,0x08,0x08,0x08,0x08,0x08,0x08,0x00, /* ` - g */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* h - o */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* p - w */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* x -127 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 128-135 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 136-143 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 144-151 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 152-159 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 160-167 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 168-175 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 176-183 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 184-191 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 192-199 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 200-207 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 208-215 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 216-223 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 224-231 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 232-239 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 240-247 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00};/* 248-255 */ - -#else - -/* This is the "abnormal" case, for EBCDIC systems not running in UTF-8 mode. */ - -static const pcre_uint8 digitab[] = - { - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 0- 7 0 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 8- 15 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 16- 23 10 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 24- 31 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 32- 39 20 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 40- 47 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 48- 55 30 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 56- 63 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* - 71 40 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 72- | */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* & - 87 50 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 88- 95 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* - -103 60 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 104- ? */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 112-119 70 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 120- " */ - 0x00,0x08,0x08,0x08,0x08,0x08,0x08,0x00, /* 128- g 80 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* h -143 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 144- p 90 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* q -159 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 160- x A0 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* y -175 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* ^ -183 B0 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 184-191 */ - 0x00,0x08,0x08,0x08,0x08,0x08,0x08,0x00, /* { - G C0 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* H -207 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* } - P D0 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* Q -223 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* \ - X E0 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* Y -239 */ - 0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /* 0 - 7 F0 */ - 0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/* 8 -255 */ - -static const pcre_uint8 ebcdic_chartab[] = { /* chartable partial dup */ - 0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /* 0- 7 */ - 0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /* 8- 15 */ - 0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /* 16- 23 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 24- 31 */ - 0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /* 32- 39 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 40- 47 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 48- 55 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 56- 63 */ - 0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* - 71 */ - 0x00,0x00,0x00,0x80,0x00,0x80,0x80,0x80, /* 72- | */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* & - 87 */ - 0x00,0x00,0x00,0x80,0x80,0x80,0x00,0x00, /* 88- 95 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* - -103 */ - 0x00,0x00,0x00,0x00,0x00,0x10,0x00,0x80, /* 104- ? */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 112-119 */ - 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 120- " */ - 0x00,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x12, /* 128- g */ - 0x12,0x12,0x00,0x00,0x00,0x00,0x00,0x00, /* h -143 */ - 0x00,0x12,0x12,0x12,0x12,0x12,0x12,0x12, /* 144- p */ - 0x12,0x12,0x00,0x00,0x00,0x00,0x00,0x00, /* q -159 */ - 0x00,0x00,0x12,0x12,0x12,0x12,0x12,0x12, /* 160- x */ - 0x12,0x12,0x00,0x00,0x00,0x00,0x00,0x00, /* y -175 */ - 0x80,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* ^ -183 */ - 0x00,0x00,0x80,0x00,0x00,0x00,0x00,0x00, /* 184-191 */ - 0x80,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x12, /* { - G */ - 0x12,0x12,0x00,0x00,0x00,0x00,0x00,0x00, /* H -207 */ - 0x00,0x12,0x12,0x12,0x12,0x12,0x12,0x12, /* } - P */ - 0x12,0x12,0x00,0x00,0x00,0x00,0x00,0x00, /* Q -223 */ - 0x00,0x00,0x12,0x12,0x12,0x12,0x12,0x12, /* \ - X */ - 0x12,0x12,0x00,0x00,0x00,0x00,0x00,0x00, /* Y -239 */ - 0x1c,0x1c,0x1c,0x1c,0x1c,0x1c,0x1c,0x1c, /* 0 - 7 */ - 0x1c,0x1c,0x00,0x00,0x00,0x00,0x00,0x00};/* 8 -255 */ -#endif - - -/* This table is used to check whether auto-possessification is possible -between adjacent character-type opcodes. The left-hand (repeated) opcode is -used to select the row, and the right-hand opcode is use to select the column. -A value of 1 means that auto-possessification is OK. For example, the second -value in the first row means that \D+\d can be turned into \D++\d. - -The Unicode property types (\P and \p) have to be present to fill out the table -because of what their opcode values are, but the table values should always be -zero because property types are handled separately in the code. The last four -columns apply to items that cannot be repeated, so there is no need to have -rows for them. Note that OP_DIGIT etc. are generated only when PCRE_UCP is -*not* set. When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */ - -#define APTROWS (LAST_AUTOTAB_LEFT_OP - FIRST_AUTOTAB_OP + 1) -#define APTCOLS (LAST_AUTOTAB_RIGHT_OP - FIRST_AUTOTAB_OP + 1) - -static const pcre_uint8 autoposstab[APTROWS][APTCOLS] = { -/* \D \d \S \s \W \w . .+ \C \P \p \R \H \h \V \v \X \Z \z $ $M */ - { 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 }, /* \D */ - { 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 1 }, /* \d */ - { 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 1 }, /* \S */ - { 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 }, /* \s */ - { 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 }, /* \W */ - { 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 1 }, /* \w */ - { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 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, 1, 0, 0 }, /* .+ */ - { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 }, /* \C */ - { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* \P */ - { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* \p */ - { 0, 1, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0 }, /* \R */ - { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0 }, /* \H */ - { 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1, 0, 0, 1, 0, 0 }, /* \h */ - { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 1, 0, 0 }, /* \V */ - { 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 0 }, /* \v */ - { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 } /* \X */ -}; - - -/* This table is used to check whether auto-possessification is possible -between adjacent Unicode property opcodes (OP_PROP and OP_NOTPROP). The -left-hand (repeated) opcode is used to select the row, and the right-hand -opcode is used to select the column. The values are as follows: - - 0 Always return FALSE (never auto-possessify) - 1 Character groups are distinct (possessify if both are OP_PROP) - 2 Check character categories in the same group (general or particular) - 3 TRUE if the two opcodes are not the same (PROP vs NOTPROP) - - 4 Check left general category vs right particular category - 5 Check right general category vs left particular category - - 6 Left alphanum vs right general category - 7 Left space vs right general category - 8 Left word vs right general category - - 9 Right alphanum vs left general category - 10 Right space vs left general category - 11 Right word vs left general category - - 12 Left alphanum vs right particular category - 13 Left space vs right particular category - 14 Left word vs right particular category - - 15 Right alphanum vs left particular category - 16 Right space vs left particular category - 17 Right word vs left particular category -*/ - -static const pcre_uint8 propposstab[PT_TABSIZE][PT_TABSIZE] = { -/* ANY LAMP GC PC SC ALNUM SPACE PXSPACE WORD CLIST UCNC */ - { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* PT_ANY */ - { 0, 3, 0, 0, 0, 3, 1, 1, 0, 0, 0 }, /* PT_LAMP */ - { 0, 0, 2, 4, 0, 9, 10, 10, 11, 0, 0 }, /* PT_GC */ - { 0, 0, 5, 2, 0, 15, 16, 16, 17, 0, 0 }, /* PT_PC */ - { 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0 }, /* PT_SC */ - { 0, 3, 6, 12, 0, 3, 1, 1, 0, 0, 0 }, /* PT_ALNUM */ - { 0, 1, 7, 13, 0, 1, 3, 3, 1, 0, 0 }, /* PT_SPACE */ - { 0, 1, 7, 13, 0, 1, 3, 3, 1, 0, 0 }, /* PT_PXSPACE */ - { 0, 0, 8, 14, 0, 0, 1, 1, 3, 0, 0 }, /* PT_WORD */ - { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* PT_CLIST */ - { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3 } /* PT_UCNC */ -}; - -/* This table is used to check whether auto-possessification is possible -between adjacent Unicode property opcodes (OP_PROP and OP_NOTPROP) when one -specifies a general category and the other specifies a particular category. The -row is selected by the general category and the column by the particular -category. The value is 1 if the particular category is not part of the general -category. */ - -static const pcre_uint8 catposstab[7][30] = { -/* Cc Cf Cn Co Cs Ll Lm Lo Lt Lu Mc Me Mn Nd Nl No Pc Pd Pe Pf Pi Po Ps Sc Sk Sm So Zl Zp Zs */ - { 0, 0, 0, 0, 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 }, /* C */ - { 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 }, /* L */ - { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 }, /* M */ - { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 }, /* N */ - { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1 }, /* P */ - { 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, 1, 1, 1 }, /* S */ - { 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 } /* Z */ -}; - -/* This table is used when checking ALNUM, (PX)SPACE, SPACE, and WORD against -a general or particular category. The properties in each row are those -that apply to the character set in question. Duplication means that a little -unnecessary work is done when checking, but this keeps things much simpler -because they can all use the same code. For more details see the comment where -this table is used. - -Note: SPACE and PXSPACE used to be different because Perl excluded VT from -"space", but from Perl 5.18 it's included, so both categories are treated the -same here. */ - -static const pcre_uint8 posspropstab[3][4] = { - { ucp_L, ucp_N, ucp_N, ucp_Nl }, /* ALNUM, 3rd and 4th values redundant */ - { ucp_Z, ucp_Z, ucp_C, ucp_Cc }, /* SPACE and PXSPACE, 2nd value redundant */ - { ucp_L, ucp_N, ucp_P, ucp_Po } /* WORD */ -}; - -/* This table is used when converting repeating opcodes into possessified -versions as a result of an explicit possessive quantifier such as ++. A zero -value means there is no possessified version - in those cases the item in -question must be wrapped in ONCE brackets. The table is truncated at OP_CALLOUT -because all relevant opcodes are less than that. */ - -static const pcre_uint8 opcode_possessify[] = { - 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 0 - 15 */ - 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 16 - 31 */ - - 0, /* NOTI */ - OP_POSSTAR, 0, /* STAR, MINSTAR */ - OP_POSPLUS, 0, /* PLUS, MINPLUS */ - OP_POSQUERY, 0, /* QUERY, MINQUERY */ - OP_POSUPTO, 0, /* UPTO, MINUPTO */ - 0, /* EXACT */ - 0, 0, 0, 0, /* POS{STAR,PLUS,QUERY,UPTO} */ - - OP_POSSTARI, 0, /* STARI, MINSTARI */ - OP_POSPLUSI, 0, /* PLUSI, MINPLUSI */ - OP_POSQUERYI, 0, /* QUERYI, MINQUERYI */ - OP_POSUPTOI, 0, /* UPTOI, MINUPTOI */ - 0, /* EXACTI */ - 0, 0, 0, 0, /* POS{STARI,PLUSI,QUERYI,UPTOI} */ - - OP_NOTPOSSTAR, 0, /* NOTSTAR, NOTMINSTAR */ - OP_NOTPOSPLUS, 0, /* NOTPLUS, NOTMINPLUS */ - OP_NOTPOSQUERY, 0, /* NOTQUERY, NOTMINQUERY */ - OP_NOTPOSUPTO, 0, /* NOTUPTO, NOTMINUPTO */ - 0, /* NOTEXACT */ - 0, 0, 0, 0, /* NOTPOS{STAR,PLUS,QUERY,UPTO} */ - - OP_NOTPOSSTARI, 0, /* NOTSTARI, NOTMINSTARI */ - OP_NOTPOSPLUSI, 0, /* NOTPLUSI, NOTMINPLUSI */ - OP_NOTPOSQUERYI, 0, /* NOTQUERYI, NOTMINQUERYI */ - OP_NOTPOSUPTOI, 0, /* NOTUPTOI, NOTMINUPTOI */ - 0, /* NOTEXACTI */ - 0, 0, 0, 0, /* NOTPOS{STARI,PLUSI,QUERYI,UPTOI} */ - - OP_TYPEPOSSTAR, 0, /* TYPESTAR, TYPEMINSTAR */ - OP_TYPEPOSPLUS, 0, /* TYPEPLUS, TYPEMINPLUS */ - OP_TYPEPOSQUERY, 0, /* TYPEQUERY, TYPEMINQUERY */ - OP_TYPEPOSUPTO, 0, /* TYPEUPTO, TYPEMINUPTO */ - 0, /* TYPEEXACT */ - 0, 0, 0, 0, /* TYPEPOS{STAR,PLUS,QUERY,UPTO} */ - - OP_CRPOSSTAR, 0, /* CRSTAR, CRMINSTAR */ - OP_CRPOSPLUS, 0, /* CRPLUS, CRMINPLUS */ - OP_CRPOSQUERY, 0, /* CRQUERY, CRMINQUERY */ - OP_CRPOSRANGE, 0, /* CRRANGE, CRMINRANGE */ - 0, 0, 0, 0, /* CRPOS{STAR,PLUS,QUERY,RANGE} */ - - 0, 0, 0, /* CLASS, NCLASS, XCLASS */ - 0, 0, /* REF, REFI */ - 0, 0, /* DNREF, DNREFI */ - 0, 0 /* RECURSE, CALLOUT */ -}; - - - -/************************************************* -* Find an error text * -*************************************************/ - -/* The error texts are now all in one long string, to save on relocations. As -some of the text is of unknown length, we can't use a table of offsets. -Instead, just count through the strings. This is not a performance issue -because it happens only when there has been a compilation error. - -Argument: the error number -Returns: pointer to the error string -*/ - -static const char * -find_error_text(int n) -{ -const char *s = error_texts; -for (; n > 0; n--) - { - while (*s++ != CHAR_NULL) {}; - if (*s == CHAR_NULL) return "Error text not found (please report)"; - } -return s; -} - - - -/************************************************* -* Expand the workspace * -*************************************************/ - -/* This function is called during the second compiling phase, if the number of -forward references fills the existing workspace, which is originally a block on -the stack. A larger block is obtained from malloc() unless the ultimate limit -has been reached or the increase will be rather small. - -Argument: pointer to the compile data block -Returns: 0 if all went well, else an error number -*/ - -static int -expand_workspace(compile_data *cd) -{ -pcre_uchar *newspace; -int newsize = cd->workspace_size * 2; - -if (newsize > COMPILE_WORK_SIZE_MAX) newsize = COMPILE_WORK_SIZE_MAX; -if (cd->workspace_size >= COMPILE_WORK_SIZE_MAX || - newsize - cd->workspace_size < WORK_SIZE_SAFETY_MARGIN) - return ERR72; - -newspace = (PUBL(malloc))(IN_UCHARS(newsize)); -if (newspace == NULL) return ERR21; -memcpy(newspace, cd->start_workspace, cd->workspace_size * sizeof(pcre_uchar)); -cd->hwm = (pcre_uchar *)newspace + (cd->hwm - cd->start_workspace); -if (cd->workspace_size > COMPILE_WORK_SIZE) - (PUBL(free))((void *)cd->start_workspace); -cd->start_workspace = newspace; -cd->workspace_size = newsize; -return 0; -} - - - -/************************************************* -* Check for counted repeat * -*************************************************/ - -/* This function is called when a '{' is encountered in a place where it might -start a quantifier. It looks ahead to see if it really is a quantifier or not. -It is only a quantifier if it is one of the forms {ddd} {ddd,} or {ddd,ddd} -where the ddds are digits. - -Arguments: - p pointer to the first char after '{' - -Returns: TRUE or FALSE -*/ - -static BOOL -is_counted_repeat(const pcre_uchar *p) -{ -if (!IS_DIGIT(*p)) return FALSE; -p++; -while (IS_DIGIT(*p)) p++; -if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE; - -if (*p++ != CHAR_COMMA) return FALSE; -if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE; - -if (!IS_DIGIT(*p)) return FALSE; -p++; -while (IS_DIGIT(*p)) p++; - -return (*p == CHAR_RIGHT_CURLY_BRACKET); -} - - - -/************************************************* -* Handle escapes * -*************************************************/ - -/* This function is called when a \ has been encountered. It either returns a -positive value for a simple escape such as \n, or 0 for a data character which -will be placed in chptr. A backreference to group n is returned as negative n. -When UTF-8 is enabled, a positive value greater than 255 may be returned in -chptr. On entry, ptr is pointing at the \. On exit, it is on the final -character of the escape sequence. - -Arguments: - ptrptr points to the pattern position pointer - chptr points to a returned data character - errorcodeptr points to the errorcode variable - bracount number of previous extracting brackets - options the options bits - isclass TRUE if inside a character class - -Returns: zero => a data character - positive => a special escape sequence - negative => a back reference - on error, errorcodeptr is set -*/ - -static int -check_escape(const pcre_uchar **ptrptr, pcre_uint32 *chptr, int *errorcodeptr, - int bracount, int options, BOOL isclass) -{ -/* PCRE_UTF16 has the same value as PCRE_UTF8. */ -BOOL utf = (options & PCRE_UTF8) != 0; -const pcre_uchar *ptr = *ptrptr + 1; -pcre_uint32 c; -int escape = 0; -int i; - -GETCHARINCTEST(c, ptr); /* Get character value, increment pointer */ -ptr--; /* Set pointer back to the last byte */ - -/* If backslash is at the end of the pattern, it's an error. */ - -if (c == CHAR_NULL) *errorcodeptr = ERR1; - -/* Non-alphanumerics are literals. For digits or letters, do an initial lookup -in a table. A non-zero result is something that can be returned immediately. -Otherwise further processing may be required. */ - -#ifndef EBCDIC /* ASCII/UTF-8 coding */ -/* Not alphanumeric */ -else if (c < CHAR_0 || c > CHAR_z) {} -else if ((i = escapes[c - CHAR_0]) != 0) - { if (i > 0) c = (pcre_uint32)i; else escape = -i; } - -#else /* EBCDIC coding */ -/* Not alphanumeric */ -else if (c < CHAR_a || (!MAX_255(c) || (ebcdic_chartab[c] & 0x0E) == 0)) {} -else if ((i = escapes[c - 0x48]) != 0) { if (i > 0) c = (pcre_uint32)i; else escape = -i; } -#endif - -/* Escapes that need further processing, or are illegal. */ - -else - { - const pcre_uchar *oldptr; - BOOL braced, negated, overflow; - int s; - - switch (c) - { - /* A number of Perl escapes are not handled by PCRE. We give an explicit - error. */ - - case CHAR_l: - case CHAR_L: - *errorcodeptr = ERR37; - break; - - case CHAR_u: - if ((options & PCRE_JAVASCRIPT_COMPAT) != 0) - { - /* In JavaScript, \u must be followed by four hexadecimal numbers. - Otherwise it is a lowercase u letter. */ - if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0 - && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0 - && MAX_255(ptr[3]) && (digitab[ptr[3]] & ctype_xdigit) != 0 - && MAX_255(ptr[4]) && (digitab[ptr[4]] & ctype_xdigit) != 0) - { - c = 0; - for (i = 0; i < 4; ++i) - { - register pcre_uint32 cc = *(++ptr); -#ifndef EBCDIC /* ASCII/UTF-8 coding */ - if (cc >= CHAR_a) cc -= 32; /* Convert to upper case */ - c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10)); -#else /* EBCDIC coding */ - if (cc >= CHAR_a && cc <= CHAR_z) cc += 64; /* Convert to upper case */ - c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10)); -#endif - } - -#if defined COMPILE_PCRE8 - if (c > (utf ? 0x10ffffU : 0xffU)) -#elif defined COMPILE_PCRE16 - if (c > (utf ? 0x10ffffU : 0xffffU)) -#elif defined COMPILE_PCRE32 - if (utf && c > 0x10ffffU) -#endif - { - *errorcodeptr = ERR76; - } - else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73; - } - } - else - *errorcodeptr = ERR37; - break; - - case CHAR_U: - /* In JavaScript, \U is an uppercase U letter. */ - if ((options & PCRE_JAVASCRIPT_COMPAT) == 0) *errorcodeptr = ERR37; - break; - - /* In a character class, \g is just a literal "g". Outside a character - class, \g must be followed by one of a number of specific things: - - (1) A number, either plain or braced. If positive, it is an absolute - backreference. If negative, it is a relative backreference. This is a Perl - 5.10 feature. - - (2) Perl 5.10 also supports \g{name} as a reference to a named group. This - is part of Perl's movement towards a unified syntax for back references. As - this is synonymous with \k{name}, we fudge it up by pretending it really - was \k. - - (3) For Oniguruma compatibility we also support \g followed by a name or a - number either in angle brackets or in single quotes. However, these are - (possibly recursive) subroutine calls, _not_ backreferences. Just return - the ESC_g code (cf \k). */ - - case CHAR_g: - if (isclass) break; - if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE) - { - escape = ESC_g; - break; - } - - /* Handle the Perl-compatible cases */ - - if (ptr[1] == CHAR_LEFT_CURLY_BRACKET) - { - const pcre_uchar *p; - for (p = ptr+2; *p != CHAR_NULL && *p != CHAR_RIGHT_CURLY_BRACKET; p++) - if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break; - if (*p != CHAR_NULL && *p != CHAR_RIGHT_CURLY_BRACKET) - { - escape = ESC_k; - break; - } - braced = TRUE; - ptr++; - } - else braced = FALSE; - - if (ptr[1] == CHAR_MINUS) - { - negated = TRUE; - ptr++; - } - else negated = FALSE; - - /* The integer range is limited by the machine's int representation. */ - s = 0; - overflow = FALSE; - while (IS_DIGIT(ptr[1])) - { - if (s > INT_MAX / 10 - 1) /* Integer overflow */ - { - overflow = TRUE; - break; - } - s = s * 10 + (int)(*(++ptr) - CHAR_0); - } - if (overflow) /* Integer overflow */ - { - while (IS_DIGIT(ptr[1])) - ptr++; - *errorcodeptr = ERR61; - break; - } - - if (braced && *(++ptr) != CHAR_RIGHT_CURLY_BRACKET) - { - *errorcodeptr = ERR57; - break; - } - - if (s == 0) - { - *errorcodeptr = ERR58; - break; - } - - if (negated) - { - if (s > bracount) - { - *errorcodeptr = ERR15; - break; - } - s = bracount - (s - 1); - } - - escape = -s; - break; - - /* The handling of escape sequences consisting of a string of digits - starting with one that is not zero is not straightforward. Perl has changed - over the years. Nowadays \g{} for backreferences and \o{} for octal are - recommended to avoid the ambiguities in the old syntax. - - Outside a character class, the digits are read as a decimal number. If the - number is less than 8 (used to be 10), or if there are that many previous - extracting left brackets, then it is a back reference. Otherwise, up to - three octal digits are read to form an escaped byte. Thus \123 is likely to - be octal 123 (cf \0123, which is octal 012 followed by the literal 3). If - the octal value is greater than 377, the least significant 8 bits are - taken. \8 and \9 are treated as the literal characters 8 and 9. - - Inside a character class, \ followed by a digit is always either a literal - 8 or 9 or an octal number. */ - - case CHAR_1: case CHAR_2: case CHAR_3: case CHAR_4: case CHAR_5: - case CHAR_6: case CHAR_7: case CHAR_8: case CHAR_9: - - if (!isclass) - { - oldptr = ptr; - /* The integer range is limited by the machine's int representation. */ - s = (int)(c -CHAR_0); - overflow = FALSE; - while (IS_DIGIT(ptr[1])) - { - if (s > INT_MAX / 10 - 1) /* Integer overflow */ - { - overflow = TRUE; - break; - } - s = s * 10 + (int)(*(++ptr) - CHAR_0); - } - if (overflow) /* Integer overflow */ - { - while (IS_DIGIT(ptr[1])) - ptr++; - *errorcodeptr = ERR61; - break; - } - if (s < 8 || s <= bracount) /* Check for back reference */ - { - escape = -s; - break; - } - ptr = oldptr; /* Put the pointer back and fall through */ - } - - /* Handle a digit following \ when the number is not a back reference. If - the first digit is 8 or 9, Perl used to generate a binary zero byte and - then treat the digit as a following literal. At least by Perl 5.18 this - changed so as not to insert the binary zero. */ - - if ((c = *ptr) >= CHAR_8) break; - - /* Fall through with a digit less than 8 */ - - /* \0 always starts an octal number, but we may drop through to here with a - larger first octal digit. The original code used just to take the least - significant 8 bits of octal numbers (I think this is what early Perls used - to do). Nowadays we allow for larger numbers in UTF-8 mode and 16-bit mode, - but no more than 3 octal digits. */ - - case CHAR_0: - c -= CHAR_0; - while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7) - c = c * 8 + *(++ptr) - CHAR_0; -#ifdef COMPILE_PCRE8 - if (!utf && c > 0xff) *errorcodeptr = ERR51; -#endif - break; - - /* \o is a relatively new Perl feature, supporting a more general way of - specifying character codes in octal. The only supported form is \o{ddd}. */ - - case CHAR_o: - if (ptr[1] != CHAR_LEFT_CURLY_BRACKET) *errorcodeptr = ERR81; else - { - ptr += 2; - c = 0; - overflow = FALSE; - while (*ptr >= CHAR_0 && *ptr <= CHAR_7) - { - register pcre_uint32 cc = *ptr++; - if (c == 0 && cc == CHAR_0) continue; /* Leading zeroes */ -#ifdef COMPILE_PCRE32 - if (c >= 0x20000000l) { overflow = TRUE; break; } -#endif - c = (c << 3) + cc - CHAR_0 ; -#if defined COMPILE_PCRE8 - if (c > (utf ? 0x10ffffU : 0xffU)) { overflow = TRUE; break; } -#elif defined COMPILE_PCRE16 - if (c > (utf ? 0x10ffffU : 0xffffU)) { overflow = TRUE; break; } -#elif defined COMPILE_PCRE32 - if (utf && c > 0x10ffffU) { overflow = TRUE; break; } -#endif - } - if (overflow) - { - while (*ptr >= CHAR_0 && *ptr <= CHAR_7) ptr++; - *errorcodeptr = ERR34; - } - else if (*ptr == CHAR_RIGHT_CURLY_BRACKET) - { - if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73; - } - else *errorcodeptr = ERR80; - } - break; - - /* \x is complicated. In JavaScript, \x must be followed by two hexadecimal - numbers. Otherwise it is a lowercase x letter. */ - - case CHAR_x: - if ((options & PCRE_JAVASCRIPT_COMPAT) != 0) - { - if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0 - && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0) - { - c = 0; - for (i = 0; i < 2; ++i) - { - register pcre_uint32 cc = *(++ptr); -#ifndef EBCDIC /* ASCII/UTF-8 coding */ - if (cc >= CHAR_a) cc -= 32; /* Convert to upper case */ - c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10)); -#else /* EBCDIC coding */ - if (cc >= CHAR_a && cc <= CHAR_z) cc += 64; /* Convert to upper case */ - c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10)); -#endif - } - } - } /* End JavaScript handling */ - - /* Handle \x in Perl's style. \x{ddd} is a character number which can be - greater than 0xff in utf or non-8bit mode, but only if the ddd are hex - digits. If not, { used to be treated as a data character. However, Perl - seems to read hex digits up to the first non-such, and ignore the rest, so - that, for example \x{zz} matches a binary zero. This seems crazy, so PCRE - now gives an error. */ - - else - { - if (ptr[1] == CHAR_LEFT_CURLY_BRACKET) - { - ptr += 2; - c = 0; - overflow = FALSE; - while (MAX_255(*ptr) && (digitab[*ptr] & ctype_xdigit) != 0) - { - register pcre_uint32 cc = *ptr++; - if (c == 0 && cc == CHAR_0) continue; /* Leading zeroes */ - -#ifdef COMPILE_PCRE32 - if (c >= 0x10000000l) { overflow = TRUE; break; } -#endif - -#ifndef EBCDIC /* ASCII/UTF-8 coding */ - if (cc >= CHAR_a) cc -= 32; /* Convert to upper case */ - c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10)); -#else /* EBCDIC coding */ - if (cc >= CHAR_a && cc <= CHAR_z) cc += 64; /* Convert to upper case */ - c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10)); -#endif - -#if defined COMPILE_PCRE8 - if (c > (utf ? 0x10ffffU : 0xffU)) { overflow = TRUE; break; } -#elif defined COMPILE_PCRE16 - if (c > (utf ? 0x10ffffU : 0xffffU)) { overflow = TRUE; break; } -#elif defined COMPILE_PCRE32 - if (utf && c > 0x10ffffU) { overflow = TRUE; break; } -#endif - } - - if (overflow) - { - while (MAX_255(*ptr) && (digitab[*ptr] & ctype_xdigit) != 0) ptr++; - *errorcodeptr = ERR34; - } - - else if (*ptr == CHAR_RIGHT_CURLY_BRACKET) - { - if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73; - } - - /* If the sequence of hex digits does not end with '}', give an error. - We used just to recognize this construct and fall through to the normal - \x handling, but nowadays Perl gives an error, which seems much more - sensible, so we do too. */ - - else *errorcodeptr = ERR79; - } /* End of \x{} processing */ - - /* Read a single-byte hex-defined char (up to two hex digits after \x) */ - - else - { - c = 0; - while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0) - { - pcre_uint32 cc; /* Some compilers don't like */ - cc = *(++ptr); /* ++ in initializers */ -#ifndef EBCDIC /* ASCII/UTF-8 coding */ - if (cc >= CHAR_a) cc -= 32; /* Convert to upper case */ - c = c * 16 + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10)); -#else /* EBCDIC coding */ - if (cc <= CHAR_z) cc += 64; /* Convert to upper case */ - c = c * 16 + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10)); -#endif - } - } /* End of \xdd handling */ - } /* End of Perl-style \x handling */ - break; - - /* For \c, a following letter is upper-cased; then the 0x40 bit is flipped. - An error is given if the byte following \c is not an ASCII character. This - coding is ASCII-specific, but then the whole concept of \cx is - ASCII-specific. (However, an EBCDIC equivalent has now been added.) */ - - case CHAR_c: - c = *(++ptr); - if (c == CHAR_NULL) - { - *errorcodeptr = ERR2; - break; - } -#ifndef EBCDIC /* ASCII/UTF-8 coding */ - if (c > 127) /* Excludes all non-ASCII in either mode */ - { - *errorcodeptr = ERR68; - break; - } - if (c >= CHAR_a && c <= CHAR_z) c -= 32; - c ^= 0x40; -#else /* EBCDIC coding */ - if (c >= CHAR_a && c <= CHAR_z) c += 64; - c ^= 0xC0; -#endif - break; - - /* PCRE_EXTRA enables extensions to Perl in the matter of escapes. Any - other alphanumeric following \ is an error if PCRE_EXTRA was set; - otherwise, for Perl compatibility, it is a literal. This code looks a bit - odd, but there used to be some cases other than the default, and there may - be again in future, so I haven't "optimized" it. */ - - default: - if ((options & PCRE_EXTRA) != 0) switch(c) - { - default: - *errorcodeptr = ERR3; - break; - } - break; - } - } - -/* Perl supports \N{name} for character names, as well as plain \N for "not -newline". PCRE does not support \N{name}. However, it does support -quantification such as \N{2,3}. */ - -if (escape == ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET && - !is_counted_repeat(ptr+2)) - *errorcodeptr = ERR37; - -/* If PCRE_UCP is set, we change the values for \d etc. */ - -if ((options & PCRE_UCP) != 0 && escape >= ESC_D && escape <= ESC_w) - escape += (ESC_DU - ESC_D); - -/* Set the pointer to the final character before returning. */ - -*ptrptr = ptr; -*chptr = c; -return escape; -} - - - -#ifdef SUPPORT_UCP -/************************************************* -* Handle \P and \p * -*************************************************/ - -/* This function is called after \P or \p has been encountered, provided that -PCRE is compiled with support for Unicode properties. On entry, ptrptr is -pointing at the P or p. On exit, it is pointing at the final character of the -escape sequence. - -Argument: - ptrptr points to the pattern position pointer - negptr points to a boolean that is set TRUE for negation else FALSE - ptypeptr points to an unsigned int that is set to the type value - pdataptr points to an unsigned int that is set to the detailed property value - errorcodeptr points to the error code variable - -Returns: TRUE if the type value was found, or FALSE for an invalid type -*/ - -static BOOL -get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, unsigned int *ptypeptr, - unsigned int *pdataptr, int *errorcodeptr) -{ -pcre_uchar c; -int i, bot, top; -const pcre_uchar *ptr = *ptrptr; -pcre_uchar name[32]; - -c = *(++ptr); -if (c == CHAR_NULL) goto ERROR_RETURN; - -*negptr = FALSE; - -/* \P or \p can be followed by a name in {}, optionally preceded by ^ for -negation. */ - -if (c == CHAR_LEFT_CURLY_BRACKET) - { - if (ptr[1] == CHAR_CIRCUMFLEX_ACCENT) - { - *negptr = TRUE; - ptr++; - } - for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++) - { - c = *(++ptr); - if (c == CHAR_NULL) goto ERROR_RETURN; - if (c == CHAR_RIGHT_CURLY_BRACKET) break; - name[i] = c; - } - if (c != CHAR_RIGHT_CURLY_BRACKET) goto ERROR_RETURN; - name[i] = 0; - } - -/* Otherwise there is just one following character */ - -else - { - name[0] = c; - name[1] = 0; - } - -*ptrptr = ptr; - -/* Search for a recognized property name using binary chop */ - -bot = 0; -top = PRIV(utt_size); - -while (bot < top) - { - int r; - i = (bot + top) >> 1; - r = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset); - if (r == 0) - { - *ptypeptr = PRIV(utt)[i].type; - *pdataptr = PRIV(utt)[i].value; - return TRUE; - } - if (r > 0) bot = i + 1; else top = i; - } - -*errorcodeptr = ERR47; -*ptrptr = ptr; -return FALSE; - -ERROR_RETURN: -*errorcodeptr = ERR46; -*ptrptr = ptr; -return FALSE; -} -#endif - - - -/************************************************* -* Read repeat counts * -*************************************************/ - -/* Read an item of the form {n,m} and return the values. This is called only -after is_counted_repeat() has confirmed that a repeat-count quantifier exists, -so the syntax is guaranteed to be correct, but we need to check the values. - -Arguments: - p pointer to first char after '{' - minp pointer to int for min - maxp pointer to int for max - returned as -1 if no max - errorcodeptr points to error code variable - -Returns: pointer to '}' on success; - current ptr on error, with errorcodeptr set non-zero -*/ - -static const pcre_uchar * -read_repeat_counts(const pcre_uchar *p, int *minp, int *maxp, int *errorcodeptr) -{ -int min = 0; -int max = -1; - -/* Read the minimum value and do a paranoid check: a negative value indicates -an integer overflow. */ - -while (IS_DIGIT(*p)) min = min * 10 + (int)(*p++ - CHAR_0); -if (min < 0 || min > 65535) - { - *errorcodeptr = ERR5; - return p; - } - -/* Read the maximum value if there is one, and again do a paranoid on its size. -Also, max must not be less than min. */ - -if (*p == CHAR_RIGHT_CURLY_BRACKET) max = min; else - { - if (*(++p) != CHAR_RIGHT_CURLY_BRACKET) - { - max = 0; - while(IS_DIGIT(*p)) max = max * 10 + (int)(*p++ - CHAR_0); - if (max < 0 || max > 65535) - { - *errorcodeptr = ERR5; - return p; - } - if (max < min) - { - *errorcodeptr = ERR4; - return p; - } - } - } - -/* Fill in the required variables, and pass back the pointer to the terminating -'}'. */ - -*minp = min; -*maxp = max; -return p; -} - - - -/************************************************* -* Find first significant op code * -*************************************************/ - -/* This is called by several functions that scan a compiled expression looking -for a fixed first character, or an anchoring op code etc. It skips over things -that do not influence this. For some calls, it makes sense to skip negative -forward and all backward assertions, and also the \b assertion; for others it -does not. - -Arguments: - code pointer to the start of the group - skipassert TRUE if certain assertions are to be skipped - -Returns: pointer to the first significant opcode -*/ - -static const pcre_uchar* -first_significant_code(const pcre_uchar *code, BOOL skipassert) -{ -for (;;) - { - switch ((int)*code) - { - case OP_ASSERT_NOT: - case OP_ASSERTBACK: - case OP_ASSERTBACK_NOT: - if (!skipassert) return code; - do code += GET(code, 1); while (*code == OP_ALT); - code += PRIV(OP_lengths)[*code]; - break; - - case OP_WORD_BOUNDARY: - case OP_NOT_WORD_BOUNDARY: - if (!skipassert) return code; - /* Fall through */ - - case OP_CALLOUT: - case OP_CREF: - case OP_DNCREF: - case OP_RREF: - case OP_DNRREF: - case OP_DEF: - code += PRIV(OP_lengths)[*code]; - break; - - default: - return code; - } - } -/* Control never reaches here */ -} - - - -/************************************************* -* Find the fixed length of a branch * -*************************************************/ - -/* Scan a branch and compute the fixed length of subject that will match it, -if the length is fixed. This is needed for dealing with backward assertions. -In UTF8 mode, the result is in characters rather than bytes. The branch is -temporarily terminated with OP_END when this function is called. - -This function is called when a backward assertion is encountered, so that if it -fails, the error message can point to the correct place in the pattern. -However, we cannot do this when the assertion contains subroutine calls, -because they can be forward references. We solve this by remembering this case -and doing the check at the end; a flag specifies which mode we are running in. - -Arguments: - code points to the start of the pattern (the bracket) - utf TRUE in UTF-8 / UTF-16 / UTF-32 mode - atend TRUE if called when the pattern is complete - cd the "compile data" structure - -Returns: the fixed length, - or -1 if there is no fixed length, - or -2 if \C was encountered (in UTF-8 mode only) - or -3 if an OP_RECURSE item was encountered and atend is FALSE - or -4 if an unknown opcode was encountered (internal error) -*/ - -static int -find_fixedlength(pcre_uchar *code, BOOL utf, BOOL atend, compile_data *cd) -{ -int length = -1; - -register int branchlength = 0; -register pcre_uchar *cc = code + 1 + LINK_SIZE; - -/* Scan along the opcodes for this branch. If we get to the end of the -branch, check the length against that of the other branches. */ - -for (;;) - { - int d; - pcre_uchar *ce, *cs; - register pcre_uchar op = *cc; - - switch (op) - { - /* We only need to continue for OP_CBRA (normal capturing bracket) and - OP_BRA (normal non-capturing bracket) because the other variants of these - opcodes are all concerned with unlimited repeated groups, which of course - are not of fixed length. */ - - case OP_CBRA: - case OP_BRA: - case OP_ONCE: - case OP_ONCE_NC: - case OP_COND: - d = find_fixedlength(cc + ((op == OP_CBRA)? IMM2_SIZE : 0), utf, atend, cd); - if (d < 0) return d; - branchlength += d; - do cc += GET(cc, 1); while (*cc == OP_ALT); - cc += 1 + LINK_SIZE; - break; - - /* Reached end of a branch; if it's a ket it is the end of a nested call. - If it's ALT it is an alternation in a nested call. An ACCEPT is effectively - an ALT. If it is END it's the end of the outer call. All can be handled by - the same code. Note that we must not include the OP_KETRxxx opcodes here, - because they all imply an unlimited repeat. */ - - case OP_ALT: - case OP_KET: - case OP_END: - case OP_ACCEPT: - case OP_ASSERT_ACCEPT: - if (length < 0) length = branchlength; - else if (length != branchlength) return -1; - if (*cc != OP_ALT) return length; - cc += 1 + LINK_SIZE; - branchlength = 0; - break; - - /* A true recursion implies not fixed length, but a subroutine call may - be OK. If the subroutine is a forward reference, we can't deal with - it until the end of the pattern, so return -3. */ - - case OP_RECURSE: - if (!atend) return -3; - cs = ce = (pcre_uchar *)cd->start_code + GET(cc, 1); /* Start subpattern */ - do ce += GET(ce, 1); while (*ce == OP_ALT); /* End subpattern */ - if (cc > cs && cc < ce) return -1; /* Recursion */ - d = find_fixedlength(cs + IMM2_SIZE, utf, atend, cd); - if (d < 0) return d; - branchlength += d; - cc += 1 + LINK_SIZE; - break; - - /* Skip over assertive subpatterns */ - - case OP_ASSERT: - case OP_ASSERT_NOT: - case OP_ASSERTBACK: - case OP_ASSERTBACK_NOT: - do cc += GET(cc, 1); while (*cc == OP_ALT); - cc += PRIV(OP_lengths)[*cc]; - break; - - /* Skip over things that don't match chars */ - - case OP_MARK: - case OP_PRUNE_ARG: - case OP_SKIP_ARG: - case OP_THEN_ARG: - cc += cc[1] + PRIV(OP_lengths)[*cc]; - break; - - case OP_CALLOUT: - case OP_CIRC: - case OP_CIRCM: - case OP_CLOSE: - case OP_COMMIT: - case OP_CREF: - case OP_DEF: - case OP_DNCREF: - case OP_DNRREF: - case OP_DOLL: - case OP_DOLLM: - case OP_EOD: - case OP_EODN: - case OP_FAIL: - case OP_NOT_WORD_BOUNDARY: - case OP_PRUNE: - case OP_REVERSE: - case OP_RREF: - case OP_SET_SOM: - case OP_SKIP: - case OP_SOD: - case OP_SOM: - case OP_THEN: - case OP_WORD_BOUNDARY: - cc += PRIV(OP_lengths)[*cc]; - break; - - /* Handle literal characters */ - - case OP_CHAR: - case OP_CHARI: - case OP_NOT: - case OP_NOTI: - branchlength++; - cc += 2; -#ifdef SUPPORT_UTF - if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); -#endif - break; - - /* Handle exact repetitions. The count is already in characters, but we - need to skip over a multibyte character in UTF8 mode. */ - - case OP_EXACT: - case OP_EXACTI: - case OP_NOTEXACT: - case OP_NOTEXACTI: - branchlength += (int)GET2(cc,1); - cc += 2 + IMM2_SIZE; -#ifdef SUPPORT_UTF - if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); -#endif - break; - - case OP_TYPEEXACT: - branchlength += GET2(cc,1); - if (cc[1 + IMM2_SIZE] == OP_PROP || cc[1 + IMM2_SIZE] == OP_NOTPROP) - cc += 2; - cc += 1 + IMM2_SIZE + 1; - break; - - /* Handle single-char matchers */ - - case OP_PROP: - case OP_NOTPROP: - cc += 2; - /* Fall through */ - - case OP_HSPACE: - case OP_VSPACE: - case OP_NOT_HSPACE: - case OP_NOT_VSPACE: - case OP_NOT_DIGIT: - case OP_DIGIT: - case OP_NOT_WHITESPACE: - case OP_WHITESPACE: - case OP_NOT_WORDCHAR: - case OP_WORDCHAR: - case OP_ANY: - case OP_ALLANY: - branchlength++; - cc++; - break; - - /* The single-byte matcher isn't allowed. This only happens in UTF-8 mode; - otherwise \C is coded as OP_ALLANY. */ - - case OP_ANYBYTE: - return -2; - - /* Check a class for variable quantification */ - - case OP_CLASS: - case OP_NCLASS: -#if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - case OP_XCLASS: - /* The original code caused an unsigned overflow in 64 bit systems, - so now we use a conditional statement. */ - if (op == OP_XCLASS) - cc += GET(cc, 1); - else - cc += PRIV(OP_lengths)[OP_CLASS]; -#else - cc += PRIV(OP_lengths)[OP_CLASS]; -#endif - - switch (*cc) - { - case OP_CRSTAR: - case OP_CRMINSTAR: - case OP_CRPLUS: - case OP_CRMINPLUS: - case OP_CRQUERY: - case OP_CRMINQUERY: - case OP_CRPOSSTAR: - case OP_CRPOSPLUS: - case OP_CRPOSQUERY: - return -1; - - case OP_CRRANGE: - case OP_CRMINRANGE: - case OP_CRPOSRANGE: - if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1; - branchlength += (int)GET2(cc,1); - cc += 1 + 2 * IMM2_SIZE; - break; - - default: - branchlength++; - } - break; - - /* Anything else is variable length */ - - case OP_ANYNL: - case OP_BRAMINZERO: - case OP_BRAPOS: - case OP_BRAPOSZERO: - case OP_BRAZERO: - case OP_CBRAPOS: - case OP_EXTUNI: - case OP_KETRMAX: - case OP_KETRMIN: - case OP_KETRPOS: - case OP_MINPLUS: - case OP_MINPLUSI: - case OP_MINQUERY: - case OP_MINQUERYI: - case OP_MINSTAR: - case OP_MINSTARI: - case OP_MINUPTO: - case OP_MINUPTOI: - case OP_NOTMINPLUS: - case OP_NOTMINPLUSI: - case OP_NOTMINQUERY: - case OP_NOTMINQUERYI: - case OP_NOTMINSTAR: - case OP_NOTMINSTARI: - case OP_NOTMINUPTO: - case OP_NOTMINUPTOI: - case OP_NOTPLUS: - case OP_NOTPLUSI: - case OP_NOTPOSPLUS: - case OP_NOTPOSPLUSI: - case OP_NOTPOSQUERY: - case OP_NOTPOSQUERYI: - case OP_NOTPOSSTAR: - case OP_NOTPOSSTARI: - case OP_NOTPOSUPTO: - case OP_NOTPOSUPTOI: - case OP_NOTQUERY: - case OP_NOTQUERYI: - case OP_NOTSTAR: - case OP_NOTSTARI: - case OP_NOTUPTO: - case OP_NOTUPTOI: - case OP_PLUS: - case OP_PLUSI: - case OP_POSPLUS: - case OP_POSPLUSI: - case OP_POSQUERY: - case OP_POSQUERYI: - case OP_POSSTAR: - case OP_POSSTARI: - case OP_POSUPTO: - case OP_POSUPTOI: - case OP_QUERY: - case OP_QUERYI: - case OP_REF: - case OP_REFI: - case OP_DNREF: - case OP_DNREFI: - case OP_SBRA: - case OP_SBRAPOS: - case OP_SCBRA: - case OP_SCBRAPOS: - case OP_SCOND: - case OP_SKIPZERO: - case OP_STAR: - case OP_STARI: - case OP_TYPEMINPLUS: - case OP_TYPEMINQUERY: - case OP_TYPEMINSTAR: - case OP_TYPEMINUPTO: - case OP_TYPEPLUS: - case OP_TYPEPOSPLUS: - case OP_TYPEPOSQUERY: - case OP_TYPEPOSSTAR: - case OP_TYPEPOSUPTO: - case OP_TYPEQUERY: - case OP_TYPESTAR: - case OP_TYPEUPTO: - case OP_UPTO: - case OP_UPTOI: - return -1; - - /* Catch unrecognized opcodes so that when new ones are added they - are not forgotten, as has happened in the past. */ - - default: - return -4; - } - } -/* Control never gets here */ -} - - - -/************************************************* -* Scan compiled regex for specific bracket * -*************************************************/ - -/* This little function scans through a compiled pattern until it finds a -capturing bracket with the given number, or, if the number is negative, an -instance of OP_REVERSE for a lookbehind. The function is global in the C sense -so that it can be called from pcre_study() when finding the minimum matching -length. - -Arguments: - code points to start of expression - utf TRUE in UTF-8 / UTF-16 / UTF-32 mode - number the required bracket number or negative to find a lookbehind - -Returns: pointer to the opcode for the bracket, or NULL if not found -*/ - -const pcre_uchar * -PRIV(find_bracket)(const pcre_uchar *code, BOOL utf, int number) -{ -for (;;) - { - register pcre_uchar c = *code; - - if (c == OP_END) return NULL; - - /* XCLASS is used for classes that cannot be represented just by a bit - map. This includes negated single high-valued characters. The length in - the table is zero; the actual length is stored in the compiled code. */ - - if (c == OP_XCLASS) code += GET(code, 1); - - /* Handle recursion */ - - else if (c == OP_REVERSE) - { - if (number < 0) return (pcre_uchar *)code; - code += PRIV(OP_lengths)[c]; - } - - /* Handle capturing bracket */ - - else if (c == OP_CBRA || c == OP_SCBRA || - c == OP_CBRAPOS || c == OP_SCBRAPOS) - { - int n = (int)GET2(code, 1+LINK_SIZE); - if (n == number) return (pcre_uchar *)code; - code += PRIV(OP_lengths)[c]; - } - - /* Otherwise, we can get the item's length from the table, except that for - repeated character types, we have to test for \p and \P, which have an extra - two bytes of parameters, and for MARK/PRUNE/SKIP/THEN with an argument, we - must add in its length. */ - - else - { - switch(c) - { - case OP_TYPESTAR: - case OP_TYPEMINSTAR: - case OP_TYPEPLUS: - case OP_TYPEMINPLUS: - case OP_TYPEQUERY: - case OP_TYPEMINQUERY: - case OP_TYPEPOSSTAR: - case OP_TYPEPOSPLUS: - case OP_TYPEPOSQUERY: - if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2; - break; - - case OP_TYPEUPTO: - case OP_TYPEMINUPTO: - case OP_TYPEEXACT: - case OP_TYPEPOSUPTO: - if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP) - code += 2; - break; - - case OP_MARK: - case OP_PRUNE_ARG: - case OP_SKIP_ARG: - case OP_THEN_ARG: - code += code[1]; - break; - } - - /* Add in the fixed length from the table */ - - code += PRIV(OP_lengths)[c]; - - /* In UTF-8 mode, opcodes that are followed by a character may be followed by - a multi-byte character. The length in the table is a minimum, so we have to - arrange to skip the extra bytes. */ - -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 - if (utf) switch(c) - { - case OP_CHAR: - case OP_CHARI: - case OP_EXACT: - case OP_EXACTI: - case OP_UPTO: - case OP_UPTOI: - case OP_MINUPTO: - case OP_MINUPTOI: - case OP_POSUPTO: - case OP_POSUPTOI: - case OP_STAR: - case OP_STARI: - case OP_MINSTAR: - case OP_MINSTARI: - case OP_POSSTAR: - case OP_POSSTARI: - case OP_PLUS: - case OP_PLUSI: - case OP_MINPLUS: - case OP_MINPLUSI: - case OP_POSPLUS: - case OP_POSPLUSI: - case OP_QUERY: - case OP_QUERYI: - case OP_MINQUERY: - case OP_MINQUERYI: - case OP_POSQUERY: - case OP_POSQUERYI: - if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]); - break; - } -#else - (void)(utf); /* Keep compiler happy by referencing function argument */ -#endif - } - } -} - - - -/************************************************* -* Scan compiled regex for recursion reference * -*************************************************/ - -/* This little function scans through a compiled pattern until it finds an -instance of OP_RECURSE. - -Arguments: - code points to start of expression - utf TRUE in UTF-8 / UTF-16 / UTF-32 mode - -Returns: pointer to the opcode for OP_RECURSE, or NULL if not found -*/ - -static const pcre_uchar * -find_recurse(const pcre_uchar *code, BOOL utf) -{ -for (;;) - { - register pcre_uchar c = *code; - if (c == OP_END) return NULL; - if (c == OP_RECURSE) return code; - - /* XCLASS is used for classes that cannot be represented just by a bit - map. This includes negated single high-valued characters. The length in - the table is zero; the actual length is stored in the compiled code. */ - - if (c == OP_XCLASS) code += GET(code, 1); - - /* Otherwise, we can get the item's length from the table, except that for - repeated character types, we have to test for \p and \P, which have an extra - two bytes of parameters, and for MARK/PRUNE/SKIP/THEN with an argument, we - must add in its length. */ - - else - { - switch(c) - { - case OP_TYPESTAR: - case OP_TYPEMINSTAR: - case OP_TYPEPLUS: - case OP_TYPEMINPLUS: - case OP_TYPEQUERY: - case OP_TYPEMINQUERY: - case OP_TYPEPOSSTAR: - case OP_TYPEPOSPLUS: - case OP_TYPEPOSQUERY: - if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2; - break; - - case OP_TYPEPOSUPTO: - case OP_TYPEUPTO: - case OP_TYPEMINUPTO: - case OP_TYPEEXACT: - if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP) - code += 2; - break; - - case OP_MARK: - case OP_PRUNE_ARG: - case OP_SKIP_ARG: - case OP_THEN_ARG: - code += code[1]; - break; - } - - /* Add in the fixed length from the table */ - - code += PRIV(OP_lengths)[c]; - - /* In UTF-8 mode, opcodes that are followed by a character may be followed - by a multi-byte character. The length in the table is a minimum, so we have - to arrange to skip the extra bytes. */ - -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 - if (utf) switch(c) - { - case OP_CHAR: - case OP_CHARI: - case OP_NOT: - case OP_NOTI: - case OP_EXACT: - case OP_EXACTI: - case OP_NOTEXACT: - case OP_NOTEXACTI: - case OP_UPTO: - case OP_UPTOI: - case OP_NOTUPTO: - case OP_NOTUPTOI: - case OP_MINUPTO: - case OP_MINUPTOI: - case OP_NOTMINUPTO: - case OP_NOTMINUPTOI: - case OP_POSUPTO: - case OP_POSUPTOI: - case OP_NOTPOSUPTO: - case OP_NOTPOSUPTOI: - case OP_STAR: - case OP_STARI: - case OP_NOTSTAR: - case OP_NOTSTARI: - case OP_MINSTAR: - case OP_MINSTARI: - case OP_NOTMINSTAR: - case OP_NOTMINSTARI: - case OP_POSSTAR: - case OP_POSSTARI: - case OP_NOTPOSSTAR: - case OP_NOTPOSSTARI: - case OP_PLUS: - case OP_PLUSI: - case OP_NOTPLUS: - case OP_NOTPLUSI: - case OP_MINPLUS: - case OP_MINPLUSI: - case OP_NOTMINPLUS: - case OP_NOTMINPLUSI: - case OP_POSPLUS: - case OP_POSPLUSI: - case OP_NOTPOSPLUS: - case OP_NOTPOSPLUSI: - case OP_QUERY: - case OP_QUERYI: - case OP_NOTQUERY: - case OP_NOTQUERYI: - case OP_MINQUERY: - case OP_MINQUERYI: - case OP_NOTMINQUERY: - case OP_NOTMINQUERYI: - case OP_POSQUERY: - case OP_POSQUERYI: - case OP_NOTPOSQUERY: - case OP_NOTPOSQUERYI: - if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]); - break; - } -#else - (void)(utf); /* Keep compiler happy by referencing function argument */ -#endif - } - } -} - - - -/************************************************* -* Scan compiled branch for non-emptiness * -*************************************************/ - -/* This function scans through a branch of a compiled pattern to see whether it -can match the empty string or not. It is called from could_be_empty() -below and from compile_branch() when checking for an unlimited repeat of a -group that can match nothing. Note that first_significant_code() skips over -backward and negative forward assertions when its final argument is TRUE. If we -hit an unclosed bracket, we return "empty" - this means we've struck an inner -bracket whose current branch will already have been scanned. - -Arguments: - code points to start of search - endcode points to where to stop - utf TRUE if in UTF-8 / UTF-16 / UTF-32 mode - cd contains pointers to tables etc. - recurses chain of recurse_check to catch mutual recursion - -Returns: TRUE if what is matched could be empty -*/ - -typedef struct recurse_check { - struct recurse_check *prev; - const pcre_uchar *group; -} recurse_check; - -static BOOL -could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode, - BOOL utf, compile_data *cd, recurse_check *recurses) -{ -register pcre_uchar c; -recurse_check this_recurse; - -for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE); - code < endcode; - code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE)) - { - const pcre_uchar *ccode; - - c = *code; - - /* Skip over forward assertions; the other assertions are skipped by - first_significant_code() with a TRUE final argument. */ - - if (c == OP_ASSERT) - { - do code += GET(code, 1); while (*code == OP_ALT); - c = *code; - continue; - } - - /* For a recursion/subroutine call, if its end has been reached, which - implies a backward reference subroutine call, we can scan it. If it's a - forward reference subroutine call, we can't. To detect forward reference - we have to scan up the list that is kept in the workspace. This function is - called only when doing the real compile, not during the pre-compile that - measures the size of the compiled pattern. */ - - if (c == OP_RECURSE) - { - const pcre_uchar *scode = cd->start_code + GET(code, 1); - BOOL empty_branch; - - /* Test for forward reference or uncompleted reference. This is disabled - when called to scan a completed pattern by setting cd->start_workspace to - NULL. */ - - if (cd->start_workspace != NULL) - { - const pcre_uchar *tcode; - for (tcode = cd->start_workspace; tcode < cd->hwm; tcode += LINK_SIZE) - if ((int)GET(tcode, 0) == (int)(code + 1 - cd->start_code)) return TRUE; - if (GET(scode, 1) == 0) return TRUE; /* Unclosed */ - } - - /* If we are scanning a completed pattern, there are no forward references - and all groups are complete. We need to detect whether this is a recursive - call, as otherwise there will be an infinite loop. If it is a recursion, - just skip over it. Simple recursions are easily detected. For mutual - recursions we keep a chain on the stack. */ - - else - { - recurse_check *r = recurses; - const pcre_uchar *endgroup = scode; - - do endgroup += GET(endgroup, 1); while (*endgroup == OP_ALT); - if (code >= scode && code <= endgroup) continue; /* Simple recursion */ - - for (r = recurses; r != NULL; r = r->prev) - if (r->group == scode) break; - if (r != NULL) continue; /* Mutual recursion */ - } - - /* Completed reference; scan the referenced group, remembering it on the - stack chain to detect mutual recursions. */ - - empty_branch = FALSE; - this_recurse.prev = recurses; - this_recurse.group = scode; - - do - { - if (could_be_empty_branch(scode, endcode, utf, cd, &this_recurse)) - { - empty_branch = TRUE; - break; - } - scode += GET(scode, 1); - } - while (*scode == OP_ALT); - - if (!empty_branch) return FALSE; /* All branches are non-empty */ - continue; - } - - /* Groups with zero repeats can of course be empty; skip them. */ - - if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO || - c == OP_BRAPOSZERO) - { - code += PRIV(OP_lengths)[c]; - do code += GET(code, 1); while (*code == OP_ALT); - c = *code; - continue; - } - - /* A nested group that is already marked as "could be empty" can just be - skipped. */ - - if (c == OP_SBRA || c == OP_SBRAPOS || - c == OP_SCBRA || c == OP_SCBRAPOS) - { - do code += GET(code, 1); while (*code == OP_ALT); - c = *code; - continue; - } - - /* For other groups, scan the branches. */ - - if (c == OP_BRA || c == OP_BRAPOS || - c == OP_CBRA || c == OP_CBRAPOS || - c == OP_ONCE || c == OP_ONCE_NC || - c == OP_COND) - { - BOOL empty_branch; - if (GET(code, 1) == 0) return TRUE; /* Hit unclosed bracket */ - - /* If a conditional group has only one branch, there is a second, implied, - empty branch, so just skip over the conditional, because it could be empty. - Otherwise, scan the individual branches of the group. */ - - if (c == OP_COND && code[GET(code, 1)] != OP_ALT) - code += GET(code, 1); - else - { - empty_branch = FALSE; - do - { - if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd, NULL)) - empty_branch = TRUE; - code += GET(code, 1); - } - while (*code == OP_ALT); - if (!empty_branch) return FALSE; /* All branches are non-empty */ - } - - c = *code; - continue; - } - - /* Handle the other opcodes */ - - switch (c) - { - /* Check for quantifiers after a class. XCLASS is used for classes that - cannot be represented just by a bit map. This includes negated single - high-valued characters. The length in PRIV(OP_lengths)[] is zero; the - actual length is stored in the compiled code, so we must update "code" - here. */ - -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - case OP_XCLASS: - ccode = code += GET(code, 1); - goto CHECK_CLASS_REPEAT; -#endif - - case OP_CLASS: - case OP_NCLASS: - ccode = code + PRIV(OP_lengths)[OP_CLASS]; - -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - CHECK_CLASS_REPEAT: -#endif - - switch (*ccode) - { - case OP_CRSTAR: /* These could be empty; continue */ - case OP_CRMINSTAR: - case OP_CRQUERY: - case OP_CRMINQUERY: - case OP_CRPOSSTAR: - case OP_CRPOSQUERY: - break; - - default: /* Non-repeat => class must match */ - case OP_CRPLUS: /* These repeats aren't empty */ - case OP_CRMINPLUS: - case OP_CRPOSPLUS: - return FALSE; - - case OP_CRRANGE: - case OP_CRMINRANGE: - case OP_CRPOSRANGE: - if (GET2(ccode, 1) > 0) return FALSE; /* Minimum > 0 */ - break; - } - break; - - /* Opcodes that must match a character */ - - case OP_ANY: - case OP_ALLANY: - case OP_ANYBYTE: - - case OP_PROP: - case OP_NOTPROP: - case OP_ANYNL: - - case OP_NOT_HSPACE: - case OP_HSPACE: - case OP_NOT_VSPACE: - case OP_VSPACE: - case OP_EXTUNI: - - case OP_NOT_DIGIT: - case OP_DIGIT: - case OP_NOT_WHITESPACE: - case OP_WHITESPACE: - case OP_NOT_WORDCHAR: - case OP_WORDCHAR: - - case OP_CHAR: - case OP_CHARI: - case OP_NOT: - case OP_NOTI: - - case OP_PLUS: - case OP_PLUSI: - case OP_MINPLUS: - case OP_MINPLUSI: - - case OP_NOTPLUS: - case OP_NOTPLUSI: - case OP_NOTMINPLUS: - case OP_NOTMINPLUSI: - - case OP_POSPLUS: - case OP_POSPLUSI: - case OP_NOTPOSPLUS: - case OP_NOTPOSPLUSI: - - case OP_EXACT: - case OP_EXACTI: - case OP_NOTEXACT: - case OP_NOTEXACTI: - - case OP_TYPEPLUS: - case OP_TYPEMINPLUS: - case OP_TYPEPOSPLUS: - case OP_TYPEEXACT: - - return FALSE; - - /* These are going to continue, as they may be empty, but we have to - fudge the length for the \p and \P cases. */ - - case OP_TYPESTAR: - case OP_TYPEMINSTAR: - case OP_TYPEPOSSTAR: - case OP_TYPEQUERY: - case OP_TYPEMINQUERY: - case OP_TYPEPOSQUERY: - if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2; - break; - - /* Same for these */ - - case OP_TYPEUPTO: - case OP_TYPEMINUPTO: - case OP_TYPEPOSUPTO: - if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP) - code += 2; - break; - - /* End of branch */ - - case OP_KET: - case OP_KETRMAX: - case OP_KETRMIN: - case OP_KETRPOS: - case OP_ALT: - return TRUE; - - /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO, - MINUPTO, and POSUPTO and their caseless and negative versions may be - followed by a multibyte character. */ - -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 - case OP_STAR: - case OP_STARI: - case OP_NOTSTAR: - case OP_NOTSTARI: - - case OP_MINSTAR: - case OP_MINSTARI: - case OP_NOTMINSTAR: - case OP_NOTMINSTARI: - - case OP_POSSTAR: - case OP_POSSTARI: - case OP_NOTPOSSTAR: - case OP_NOTPOSSTARI: - - case OP_QUERY: - case OP_QUERYI: - case OP_NOTQUERY: - case OP_NOTQUERYI: - - case OP_MINQUERY: - case OP_MINQUERYI: - case OP_NOTMINQUERY: - case OP_NOTMINQUERYI: - - case OP_POSQUERY: - case OP_POSQUERYI: - case OP_NOTPOSQUERY: - case OP_NOTPOSQUERYI: - - if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]); - break; - - case OP_UPTO: - case OP_UPTOI: - case OP_NOTUPTO: - case OP_NOTUPTOI: - - case OP_MINUPTO: - case OP_MINUPTOI: - case OP_NOTMINUPTO: - case OP_NOTMINUPTOI: - - case OP_POSUPTO: - case OP_POSUPTOI: - case OP_NOTPOSUPTO: - case OP_NOTPOSUPTOI: - - if (utf && HAS_EXTRALEN(code[1 + IMM2_SIZE])) code += GET_EXTRALEN(code[1 + IMM2_SIZE]); - break; -#endif - - /* MARK, and PRUNE/SKIP/THEN with an argument must skip over the argument - string. */ - - case OP_MARK: - case OP_PRUNE_ARG: - case OP_SKIP_ARG: - case OP_THEN_ARG: - code += code[1]; - break; - - /* None of the remaining opcodes are required to match a character. */ - - default: - break; - } - } - -return TRUE; -} - - - -/************************************************* -* Scan compiled regex for non-emptiness * -*************************************************/ - -/* This function is called to check for left recursive calls. We want to check -the current branch of the current pattern to see if it could match the empty -string. If it could, we must look outwards for branches at other levels, -stopping when we pass beyond the bracket which is the subject of the recursion. -This function is called only during the real compile, not during the -pre-compile. - -Arguments: - code points to start of the recursion - endcode points to where to stop (current RECURSE item) - bcptr points to the chain of current (unclosed) branch starts - utf TRUE if in UTF-8 / UTF-16 / UTF-32 mode - cd pointers to tables etc - -Returns: TRUE if what is matched could be empty -*/ - -static BOOL -could_be_empty(const pcre_uchar *code, const pcre_uchar *endcode, - branch_chain *bcptr, BOOL utf, compile_data *cd) -{ -while (bcptr != NULL && bcptr->current_branch >= code) - { - if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd, NULL)) - return FALSE; - bcptr = bcptr->outer; - } -return TRUE; -} - - - -/************************************************* -* Base opcode of repeated opcodes * -*************************************************/ - -/* Returns the base opcode for repeated single character type opcodes. If the -opcode is not a repeated character type, it returns with the original value. - -Arguments: c opcode -Returns: base opcode for the type -*/ - -static pcre_uchar -get_repeat_base(pcre_uchar c) -{ -return (c > OP_TYPEPOSUPTO)? c : - (c >= OP_TYPESTAR)? OP_TYPESTAR : - (c >= OP_NOTSTARI)? OP_NOTSTARI : - (c >= OP_NOTSTAR)? OP_NOTSTAR : - (c >= OP_STARI)? OP_STARI : - OP_STAR; -} - - - -#ifdef SUPPORT_UCP -/************************************************* -* Check a character and a property * -*************************************************/ - -/* This function is called by check_auto_possessive() when a property item -is adjacent to a fixed character. - -Arguments: - c the character - ptype the property type - pdata the data for the type - negated TRUE if it's a negated property (\P or \p{^) - -Returns: TRUE if auto-possessifying is OK -*/ - -static BOOL -check_char_prop(pcre_uint32 c, unsigned int ptype, unsigned int pdata, - BOOL negated) -{ -const pcre_uint32 *p; -const ucd_record *prop = GET_UCD(c); - -switch(ptype) - { - case PT_LAMP: - return (prop->chartype == ucp_Lu || - prop->chartype == ucp_Ll || - prop->chartype == ucp_Lt) == negated; - - case PT_GC: - return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated; - - case PT_PC: - return (pdata == prop->chartype) == negated; - - case PT_SC: - return (pdata == prop->script) == negated; - - /* These are specials */ - - case PT_ALNUM: - return (PRIV(ucp_gentype)[prop->chartype] == ucp_L || - PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated; - - /* Perl space used to exclude VT, but from Perl 5.18 it is included, which - means that Perl space and POSIX space are now identical. PCRE was changed - at release 8.34. */ - - case PT_SPACE: /* Perl space */ - case PT_PXSPACE: /* POSIX space */ - switch(c) - { - HSPACE_CASES: - VSPACE_CASES: - return negated; - - default: - return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z) == negated; - } - break; /* Control never reaches here */ - - case PT_WORD: - return (PRIV(ucp_gentype)[prop->chartype] == ucp_L || - PRIV(ucp_gentype)[prop->chartype] == ucp_N || - c == CHAR_UNDERSCORE) == negated; - - case PT_CLIST: - p = PRIV(ucd_caseless_sets) + prop->caseset; - for (;;) - { - if (c < *p) return !negated; - if (c == *p++) return negated; - } - break; /* Control never reaches here */ - } - -return FALSE; -} -#endif /* SUPPORT_UCP */ - - - -/************************************************* -* Fill the character property list * -*************************************************/ - -/* Checks whether the code points to an opcode that can take part in auto- -possessification, and if so, fills a list with its properties. - -Arguments: - code points to start of expression - utf TRUE if in UTF-8 / UTF-16 / UTF-32 mode - fcc points to case-flipping table - list points to output list - list[0] will be filled with the opcode - list[1] will be non-zero if this opcode - can match an empty character string - list[2..7] depends on the opcode - -Returns: points to the start of the next opcode if *code is accepted - NULL if *code is not accepted -*/ - -static const pcre_uchar * -get_chr_property_list(const pcre_uchar *code, BOOL utf, - const pcre_uint8 *fcc, pcre_uint32 *list) -{ -pcre_uchar c = *code; -pcre_uchar base; -const pcre_uchar *end; -pcre_uint32 chr; - -#ifdef SUPPORT_UCP -pcre_uint32 *clist_dest; -const pcre_uint32 *clist_src; -#else -utf = utf; /* Suppress "unused parameter" compiler warning */ -#endif - -list[0] = c; -list[1] = FALSE; -code++; - -if (c >= OP_STAR && c <= OP_TYPEPOSUPTO) - { - base = get_repeat_base(c); - c -= (base - OP_STAR); - - if (c == OP_UPTO || c == OP_MINUPTO || c == OP_EXACT || c == OP_POSUPTO) - code += IMM2_SIZE; - - list[1] = (c != OP_PLUS && c != OP_MINPLUS && c != OP_EXACT && c != OP_POSPLUS); - - switch(base) - { - case OP_STAR: - list[0] = OP_CHAR; - break; - - case OP_STARI: - list[0] = OP_CHARI; - break; - - case OP_NOTSTAR: - list[0] = OP_NOT; - break; - - case OP_NOTSTARI: - list[0] = OP_NOTI; - break; - - case OP_TYPESTAR: - list[0] = *code; - code++; - break; - } - c = list[0]; - } - -switch(c) - { - case OP_NOT_DIGIT: - case OP_DIGIT: - case OP_NOT_WHITESPACE: - case OP_WHITESPACE: - case OP_NOT_WORDCHAR: - case OP_WORDCHAR: - case OP_ANY: - case OP_ALLANY: - case OP_ANYNL: - case OP_NOT_HSPACE: - case OP_HSPACE: - case OP_NOT_VSPACE: - case OP_VSPACE: - case OP_EXTUNI: - case OP_EODN: - case OP_EOD: - case OP_DOLL: - case OP_DOLLM: - return code; - - case OP_CHAR: - case OP_NOT: - GETCHARINCTEST(chr, code); - list[2] = chr; - list[3] = NOTACHAR; - return code; - - case OP_CHARI: - case OP_NOTI: - list[0] = (c == OP_CHARI) ? OP_CHAR : OP_NOT; - GETCHARINCTEST(chr, code); - list[2] = chr; - -#ifdef SUPPORT_UCP - if (chr < 128 || (chr < 256 && !utf)) - list[3] = fcc[chr]; - else - list[3] = UCD_OTHERCASE(chr); -#elif defined SUPPORT_UTF || !defined COMPILE_PCRE8 - list[3] = (chr < 256) ? fcc[chr] : chr; -#else - list[3] = fcc[chr]; -#endif - - /* The othercase might be the same value. */ - - if (chr == list[3]) - list[3] = NOTACHAR; - else - list[4] = NOTACHAR; - return code; - -#ifdef SUPPORT_UCP - case OP_PROP: - case OP_NOTPROP: - if (code[0] != PT_CLIST) - { - list[2] = code[0]; - list[3] = code[1]; - return code + 2; - } - - /* Convert only if we have enough space. */ - - clist_src = PRIV(ucd_caseless_sets) + code[1]; - clist_dest = list + 2; - code += 2; - - do { - if (clist_dest >= list + 8) - { - /* Early return if there is not enough space. This should never - happen, since all clists are shorter than 5 character now. */ - list[2] = code[0]; - list[3] = code[1]; - return code; - } - *clist_dest++ = *clist_src; - } - while(*clist_src++ != NOTACHAR); - - /* All characters are stored. The terminating NOTACHAR - is copied form the clist itself. */ - - list[0] = (c == OP_PROP) ? OP_CHAR : OP_NOT; - return code; -#endif - - case OP_NCLASS: - case OP_CLASS: -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - case OP_XCLASS: - if (c == OP_XCLASS) - end = code + GET(code, 0) - 1; - else -#endif - end = code + 32 / sizeof(pcre_uchar); - - switch(*end) - { - case OP_CRSTAR: - case OP_CRMINSTAR: - case OP_CRQUERY: - case OP_CRMINQUERY: - case OP_CRPOSSTAR: - case OP_CRPOSQUERY: - list[1] = TRUE; - end++; - break; - - case OP_CRPLUS: - case OP_CRMINPLUS: - case OP_CRPOSPLUS: - end++; - break; - - case OP_CRRANGE: - case OP_CRMINRANGE: - case OP_CRPOSRANGE: - list[1] = (GET2(end, 1) == 0); - end += 1 + 2 * IMM2_SIZE; - break; - } - list[2] = end - code; - return end; - } -return NULL; /* Opcode not accepted */ -} - - - -/************************************************* -* Scan further character sets for match * -*************************************************/ - -/* Checks whether the base and the current opcode have a common character, in -which case the base cannot be possessified. - -Arguments: - code points to the byte code - utf TRUE in UTF-8 / UTF-16 / UTF-32 mode - cd static compile data - base_list the data list of the base opcode - -Returns: TRUE if the auto-possessification is possible -*/ - -static BOOL -compare_opcodes(const pcre_uchar *code, BOOL utf, const compile_data *cd, - const pcre_uint32 *base_list, const pcre_uchar *base_end) -{ -pcre_uchar c; -pcre_uint32 list[8]; -const pcre_uint32 *chr_ptr; -const pcre_uint32 *ochr_ptr; -const pcre_uint32 *list_ptr; -const pcre_uchar *next_code; -const pcre_uint8 *class_bitset; -const pcre_uint32 *set1, *set2, *set_end; -pcre_uint32 chr; -BOOL accepted, invert_bits; - -/* Note: the base_list[1] contains whether the current opcode has greedy -(represented by a non-zero value) quantifier. This is a different from -other character type lists, which stores here that the character iterator -matches to an empty string (also represented by a non-zero value). */ - -for(;;) - { - /* All operations move the code pointer forward. - Therefore infinite recursions are not possible. */ - - c = *code; - - /* Skip over callouts */ - - if (c == OP_CALLOUT) - { - code += PRIV(OP_lengths)[c]; - continue; - } - - if (c == OP_ALT) - { - do code += GET(code, 1); while (*code == OP_ALT); - c = *code; - } - - switch(c) - { - case OP_END: - case OP_KETRPOS: - /* TRUE only in greedy case. The non-greedy case could be replaced by - an OP_EXACT, but it is probably not worth it. (And note that OP_EXACT - uses more memory, which we cannot get at this stage.) */ - - return base_list[1] != 0; - - case OP_KET: - /* If the bracket is capturing, and referenced by an OP_RECURSE, or - it is an atomic sub-pattern (assert, once, etc.) the non-greedy case - cannot be converted to a possessive form. */ - - if (base_list[1] == 0) return FALSE; - - switch(*(code - GET(code, 1))) - { - case OP_ASSERT: - case OP_ASSERT_NOT: - case OP_ASSERTBACK: - case OP_ASSERTBACK_NOT: - case OP_ONCE: - case OP_ONCE_NC: - /* Atomic sub-patterns and assertions can always auto-possessify their - last iterator. */ - return TRUE; - } - - code += PRIV(OP_lengths)[c]; - continue; - - case OP_ONCE: - case OP_ONCE_NC: - case OP_BRA: - case OP_CBRA: - next_code = code + GET(code, 1); - code += PRIV(OP_lengths)[c]; - - while (*next_code == OP_ALT) - { - if (!compare_opcodes(code, utf, cd, base_list, base_end)) return FALSE; - code = next_code + 1 + LINK_SIZE; - next_code += GET(next_code, 1); - } - continue; - - case OP_BRAZERO: - case OP_BRAMINZERO: - - next_code = code + 1; - if (*next_code != OP_BRA && *next_code != OP_CBRA - && *next_code != OP_ONCE && *next_code != OP_ONCE_NC) return FALSE; - - do next_code += GET(next_code, 1); while (*next_code == OP_ALT); - - /* The bracket content will be checked by the - OP_BRA/OP_CBRA case above. */ - next_code += 1 + LINK_SIZE; - if (!compare_opcodes(next_code, utf, cd, base_list, base_end)) - return FALSE; - - code += PRIV(OP_lengths)[c]; - continue; - } - - /* Check for a supported opcode, and load its properties. */ - - code = get_chr_property_list(code, utf, cd->fcc, list); - if (code == NULL) return FALSE; /* Unsupported */ - - /* If either opcode is a small character list, set pointers for comparing - characters from that list with another list, or with a property. */ - - if (base_list[0] == OP_CHAR) - { - chr_ptr = base_list + 2; - list_ptr = list; - } - else if (list[0] == OP_CHAR) - { - chr_ptr = list + 2; - list_ptr = base_list; - } - - /* Character bitsets can also be compared to certain opcodes. */ - - else if (base_list[0] == OP_CLASS || list[0] == OP_CLASS -#ifdef COMPILE_PCRE8 - /* In 8 bit, non-UTF mode, OP_CLASS and OP_NCLASS are the same. */ - || (!utf && (base_list[0] == OP_NCLASS || list[0] == OP_NCLASS)) -#endif - ) - { -#ifdef COMPILE_PCRE8 - if (base_list[0] == OP_CLASS || (!utf && base_list[0] == OP_NCLASS)) -#else - if (base_list[0] == OP_CLASS) -#endif - { - set1 = (pcre_uint32 *)(base_end - base_list[2]); - list_ptr = list; - } - else - { - set1 = (pcre_uint32 *)(code - list[2]); - list_ptr = base_list; - } - - invert_bits = FALSE; - switch(list_ptr[0]) - { - case OP_CLASS: - case OP_NCLASS: - set2 = (pcre_uint32 *) - ((list_ptr == list ? code : base_end) - list_ptr[2]); - break; - - /* OP_XCLASS cannot be supported here, because its bitset - is not necessarily complete. E.g: [a-\0x{200}] is stored - as a character range, and the appropriate bits are not set. */ - - case OP_NOT_DIGIT: - invert_bits = TRUE; - /* Fall through */ - case OP_DIGIT: - set2 = (pcre_uint32 *)(cd->cbits + cbit_digit); - break; - - case OP_NOT_WHITESPACE: - invert_bits = TRUE; - /* Fall through */ - case OP_WHITESPACE: - set2 = (pcre_uint32 *)(cd->cbits + cbit_space); - break; - - case OP_NOT_WORDCHAR: - invert_bits = TRUE; - /* Fall through */ - case OP_WORDCHAR: - set2 = (pcre_uint32 *)(cd->cbits + cbit_word); - break; - - default: - return FALSE; - } - - /* Compare 4 bytes to improve speed. */ - set_end = set1 + (32 / 4); - if (invert_bits) - { - do - { - if ((*set1++ & ~(*set2++)) != 0) return FALSE; - } - while (set1 < set_end); - } - else - { - do - { - if ((*set1++ & *set2++) != 0) return FALSE; - } - while (set1 < set_end); - } - - if (list[1] == 0) return TRUE; - /* Might be an empty repeat. */ - continue; - } - - /* Some property combinations also acceptable. Unicode property opcodes are - processed specially; the rest can be handled with a lookup table. */ - - else - { - pcre_uint32 leftop, rightop; - - leftop = base_list[0]; - rightop = list[0]; - -#ifdef SUPPORT_UCP - accepted = FALSE; /* Always set in non-unicode case. */ - if (leftop == OP_PROP || leftop == OP_NOTPROP) - { - if (rightop == OP_EOD) - accepted = TRUE; - else if (rightop == OP_PROP || rightop == OP_NOTPROP) - { - int n; - const pcre_uint8 *p; - BOOL same = leftop == rightop; - BOOL lisprop = leftop == OP_PROP; - BOOL risprop = rightop == OP_PROP; - BOOL bothprop = lisprop && risprop; - - /* There's a table that specifies how each combination is to be - processed: - 0 Always return FALSE (never auto-possessify) - 1 Character groups are distinct (possessify if both are OP_PROP) - 2 Check character categories in the same group (general or particular) - 3 Return TRUE if the two opcodes are not the same - ... see comments below - */ - - n = propposstab[base_list[2]][list[2]]; - switch(n) - { - case 0: break; - case 1: accepted = bothprop; break; - case 2: accepted = (base_list[3] == list[3]) != same; break; - case 3: accepted = !same; break; - - case 4: /* Left general category, right particular category */ - accepted = risprop && catposstab[base_list[3]][list[3]] == same; - break; - - case 5: /* Right general category, left particular category */ - accepted = lisprop && catposstab[list[3]][base_list[3]] == same; - break; - - /* This code is logically tricky. Think hard before fiddling with it. - The posspropstab table has four entries per row. Each row relates to - one of PCRE's special properties such as ALNUM or SPACE or WORD. - Only WORD actually needs all four entries, but using repeats for the - others means they can all use the same code below. - - The first two entries in each row are Unicode general categories, and - apply always, because all the characters they include are part of the - PCRE character set. The third and fourth entries are a general and a - particular category, respectively, that include one or more relevant - characters. One or the other is used, depending on whether the check - is for a general or a particular category. However, in both cases the - category contains more characters than the specials that are defined - for the property being tested against. Therefore, it cannot be used - in a NOTPROP case. - - Example: the row for WORD contains ucp_L, ucp_N, ucp_P, ucp_Po. - Underscore is covered by ucp_P or ucp_Po. */ - - case 6: /* Left alphanum vs right general category */ - case 7: /* Left space vs right general category */ - case 8: /* Left word vs right general category */ - p = posspropstab[n-6]; - accepted = risprop && lisprop == - (list[3] != p[0] && - list[3] != p[1] && - (list[3] != p[2] || !lisprop)); - break; - - case 9: /* Right alphanum vs left general category */ - case 10: /* Right space vs left general category */ - case 11: /* Right word vs left general category */ - p = posspropstab[n-9]; - accepted = lisprop && risprop == - (base_list[3] != p[0] && - base_list[3] != p[1] && - (base_list[3] != p[2] || !risprop)); - break; - - case 12: /* Left alphanum vs right particular category */ - case 13: /* Left space vs right particular category */ - case 14: /* Left word vs right particular category */ - p = posspropstab[n-12]; - accepted = risprop && lisprop == - (catposstab[p[0]][list[3]] && - catposstab[p[1]][list[3]] && - (list[3] != p[3] || !lisprop)); - break; - - case 15: /* Right alphanum vs left particular category */ - case 16: /* Right space vs left particular category */ - case 17: /* Right word vs left particular category */ - p = posspropstab[n-15]; - accepted = lisprop && risprop == - (catposstab[p[0]][base_list[3]] && - catposstab[p[1]][base_list[3]] && - (base_list[3] != p[3] || !risprop)); - break; - } - } - } - - else -#endif /* SUPPORT_UCP */ - - accepted = leftop >= FIRST_AUTOTAB_OP && leftop <= LAST_AUTOTAB_LEFT_OP && - rightop >= FIRST_AUTOTAB_OP && rightop <= LAST_AUTOTAB_RIGHT_OP && - autoposstab[leftop - FIRST_AUTOTAB_OP][rightop - FIRST_AUTOTAB_OP]; - - if (!accepted) - return FALSE; - - if (list[1] == 0) return TRUE; - /* Might be an empty repeat. */ - continue; - } - - /* Control reaches here only if one of the items is a small character list. - All characters are checked against the other side. */ - - do - { - chr = *chr_ptr; - - switch(list_ptr[0]) - { - case OP_CHAR: - ochr_ptr = list_ptr + 2; - do - { - if (chr == *ochr_ptr) return FALSE; - ochr_ptr++; - } - while(*ochr_ptr != NOTACHAR); - break; - - case OP_NOT: - ochr_ptr = list_ptr + 2; - do - { - if (chr == *ochr_ptr) - break; - ochr_ptr++; - } - while(*ochr_ptr != NOTACHAR); - if (*ochr_ptr == NOTACHAR) return FALSE; /* Not found */ - break; - - /* Note that OP_DIGIT etc. are generated only when PCRE_UCP is *not* - set. When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */ - - case OP_DIGIT: - if (chr < 256 && (cd->ctypes[chr] & ctype_digit) != 0) return FALSE; - break; - - case OP_NOT_DIGIT: - if (chr > 255 || (cd->ctypes[chr] & ctype_digit) == 0) return FALSE; - break; - - case OP_WHITESPACE: - if (chr < 256 && (cd->ctypes[chr] & ctype_space) != 0) return FALSE; - break; - - case OP_NOT_WHITESPACE: - if (chr > 255 || (cd->ctypes[chr] & ctype_space) == 0) return FALSE; - break; - - case OP_WORDCHAR: - if (chr < 255 && (cd->ctypes[chr] & ctype_word) != 0) return FALSE; - break; - - case OP_NOT_WORDCHAR: - if (chr > 255 || (cd->ctypes[chr] & ctype_word) == 0) return FALSE; - break; - - case OP_HSPACE: - switch(chr) - { - HSPACE_CASES: return FALSE; - default: break; - } - break; - - case OP_NOT_HSPACE: - switch(chr) - { - HSPACE_CASES: break; - default: return FALSE; - } - break; - - case OP_ANYNL: - case OP_VSPACE: - switch(chr) - { - VSPACE_CASES: return FALSE; - default: break; - } - break; - - case OP_NOT_VSPACE: - switch(chr) - { - VSPACE_CASES: break; - default: return FALSE; - } - break; - - case OP_DOLL: - case OP_EODN: - switch (chr) - { - case CHAR_CR: - case CHAR_LF: - case CHAR_VT: - case CHAR_FF: - case CHAR_NEL: -#ifndef EBCDIC - case 0x2028: - case 0x2029: -#endif /* Not EBCDIC */ - return FALSE; - } - break; - - case OP_EOD: /* Can always possessify before \z */ - break; - -#ifdef SUPPORT_UCP - case OP_PROP: - case OP_NOTPROP: - if (!check_char_prop(chr, list_ptr[2], list_ptr[3], - list_ptr[0] == OP_NOTPROP)) - return FALSE; - break; -#endif - - case OP_NCLASS: - if (chr > 255) return FALSE; - /* Fall through */ - - case OP_CLASS: - if (chr > 255) break; - class_bitset = (pcre_uint8 *) - ((list_ptr == list ? code : base_end) - list_ptr[2]); - if ((class_bitset[chr >> 3] & (1 << (chr & 7))) != 0) return FALSE; - break; - -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - case OP_XCLASS: - if (PRIV(xclass)(chr, (list_ptr == list ? code : base_end) - - list_ptr[2] + LINK_SIZE, utf)) return FALSE; - break; -#endif - - default: - return FALSE; - } - - chr_ptr++; - } - while(*chr_ptr != NOTACHAR); - - /* At least one character must be matched from this opcode. */ - - if (list[1] == 0) return TRUE; - } - -return FALSE; -} - - - -/************************************************* -* Scan compiled regex for auto-possession * -*************************************************/ - -/* Replaces single character iterations with their possessive alternatives -if appropriate. This function modifies the compiled opcode! - -Arguments: - code points to start of the byte code - utf TRUE in UTF-8 / UTF-16 / UTF-32 mode - cd static compile data - -Returns: nothing -*/ - -static void -auto_possessify(pcre_uchar *code, BOOL utf, const compile_data *cd) -{ -register pcre_uchar c; -const pcre_uchar *end; -pcre_uchar *repeat_opcode; -pcre_uint32 list[8]; - -for (;;) - { - c = *code; - - if (c >= OP_STAR && c <= OP_TYPEPOSUPTO) - { - c -= get_repeat_base(c) - OP_STAR; - end = (c <= OP_MINUPTO) ? - get_chr_property_list(code, utf, cd->fcc, list) : NULL; - list[1] = c == OP_STAR || c == OP_PLUS || c == OP_QUERY || c == OP_UPTO; - - if (end != NULL && compare_opcodes(end, utf, cd, list, end)) - { - switch(c) - { - case OP_STAR: - *code += OP_POSSTAR - OP_STAR; - break; - - case OP_MINSTAR: - *code += OP_POSSTAR - OP_MINSTAR; - break; - - case OP_PLUS: - *code += OP_POSPLUS - OP_PLUS; - break; - - case OP_MINPLUS: - *code += OP_POSPLUS - OP_MINPLUS; - break; - - case OP_QUERY: - *code += OP_POSQUERY - OP_QUERY; - break; - - case OP_MINQUERY: - *code += OP_POSQUERY - OP_MINQUERY; - break; - - case OP_UPTO: - *code += OP_POSUPTO - OP_UPTO; - break; - - case OP_MINUPTO: - *code += OP_MINUPTO - OP_UPTO; - break; - } - } - c = *code; - } - else if (c == OP_CLASS || c == OP_NCLASS || c == OP_XCLASS) - { -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - if (c == OP_XCLASS) - repeat_opcode = code + GET(code, 1); - else -#endif - repeat_opcode = code + 1 + (32 / sizeof(pcre_uchar)); - - c = *repeat_opcode; - if (c >= OP_CRSTAR && c <= OP_CRMINRANGE) - { - /* end must not be NULL. */ - end = get_chr_property_list(code, utf, cd->fcc, list); - - list[1] = (c & 1) == 0; - - if (compare_opcodes(end, utf, cd, list, end)) - { - switch (c) - { - case OP_CRSTAR: - case OP_CRMINSTAR: - *repeat_opcode = OP_CRPOSSTAR; - break; - - case OP_CRPLUS: - case OP_CRMINPLUS: - *repeat_opcode = OP_CRPOSPLUS; - break; - - case OP_CRQUERY: - case OP_CRMINQUERY: - *repeat_opcode = OP_CRPOSQUERY; - break; - - case OP_CRRANGE: - case OP_CRMINRANGE: - *repeat_opcode = OP_CRPOSRANGE; - break; - } - } - } - c = *code; - } - - switch(c) - { - case OP_END: - return; - - case OP_TYPESTAR: - case OP_TYPEMINSTAR: - case OP_TYPEPLUS: - case OP_TYPEMINPLUS: - case OP_TYPEQUERY: - case OP_TYPEMINQUERY: - case OP_TYPEPOSSTAR: - case OP_TYPEPOSPLUS: - case OP_TYPEPOSQUERY: - if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2; - break; - - case OP_TYPEUPTO: - case OP_TYPEMINUPTO: - case OP_TYPEEXACT: - case OP_TYPEPOSUPTO: - if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP) - code += 2; - break; - -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - case OP_XCLASS: - code += GET(code, 1); - break; -#endif - - case OP_MARK: - case OP_PRUNE_ARG: - case OP_SKIP_ARG: - case OP_THEN_ARG: - code += code[1]; - break; - } - - /* Add in the fixed length from the table */ - - code += PRIV(OP_lengths)[c]; - - /* In UTF-8 mode, opcodes that are followed by a character may be followed by - a multi-byte character. The length in the table is a minimum, so we have to - arrange to skip the extra bytes. */ - -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 - if (utf) switch(c) - { - case OP_CHAR: - case OP_CHARI: - case OP_NOT: - case OP_NOTI: - case OP_STAR: - case OP_MINSTAR: - case OP_PLUS: - case OP_MINPLUS: - case OP_QUERY: - case OP_MINQUERY: - case OP_UPTO: - case OP_MINUPTO: - case OP_EXACT: - case OP_POSSTAR: - case OP_POSPLUS: - case OP_POSQUERY: - case OP_POSUPTO: - case OP_STARI: - case OP_MINSTARI: - case OP_PLUSI: - case OP_MINPLUSI: - case OP_QUERYI: - case OP_MINQUERYI: - case OP_UPTOI: - case OP_MINUPTOI: - case OP_EXACTI: - case OP_POSSTARI: - case OP_POSPLUSI: - case OP_POSQUERYI: - case OP_POSUPTOI: - case OP_NOTSTAR: - case OP_NOTMINSTAR: - case OP_NOTPLUS: - case OP_NOTMINPLUS: - case OP_NOTQUERY: - case OP_NOTMINQUERY: - case OP_NOTUPTO: - case OP_NOTMINUPTO: - case OP_NOTEXACT: - case OP_NOTPOSSTAR: - case OP_NOTPOSPLUS: - case OP_NOTPOSQUERY: - case OP_NOTPOSUPTO: - case OP_NOTSTARI: - case OP_NOTMINSTARI: - case OP_NOTPLUSI: - case OP_NOTMINPLUSI: - case OP_NOTQUERYI: - case OP_NOTMINQUERYI: - case OP_NOTUPTOI: - case OP_NOTMINUPTOI: - case OP_NOTEXACTI: - case OP_NOTPOSSTARI: - case OP_NOTPOSPLUSI: - case OP_NOTPOSQUERYI: - case OP_NOTPOSUPTOI: - if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]); - break; - } -#else - (void)(utf); /* Keep compiler happy by referencing function argument */ -#endif - } -} - - - -/************************************************* -* Check for POSIX class syntax * -*************************************************/ - -/* This function is called when the sequence "[:" or "[." or "[=" is -encountered in a character class. It checks whether this is followed by a -sequence of characters terminated by a matching ":]" or ".]" or "=]". If we -reach an unescaped ']' without the special preceding character, return FALSE. - -Originally, this function only recognized a sequence of letters between the -terminators, but it seems that Perl recognizes any sequence of characters, -though of course unknown POSIX names are subsequently rejected. Perl gives an -"Unknown POSIX class" error for [:f\oo:] for example, where previously PCRE -didn't consider this to be a POSIX class. Likewise for [:1234:]. - -The problem in trying to be exactly like Perl is in the handling of escapes. We -have to be sure that [abc[:x\]pqr] is *not* treated as containing a POSIX -class, but [abc[:x\]pqr:]] is (so that an error can be generated). The code -below handles the special case of \], but does not try to do any other escape -processing. This makes it different from Perl for cases such as [:l\ower:] -where Perl recognizes it as the POSIX class "lower" but PCRE does not recognize -"l\ower". This is a lesser evil than not diagnosing bad classes when Perl does, -I think. - -A user pointed out that PCRE was rejecting [:a[:digit:]] whereas Perl was not. -It seems that the appearance of a nested POSIX class supersedes an apparent -external class. For example, [:a[:digit:]b:] matches "a", "b", ":", or -a digit. - -In Perl, unescaped square brackets may also appear as part of class names. For -example, [:a[:abc]b:] gives unknown POSIX class "[:abc]b:]". However, for -[:a[:abc]b][b:] it gives unknown POSIX class "[:abc]b][b:]", which does not -seem right at all. PCRE does not allow closing square brackets in POSIX class -names. - -Arguments: - ptr pointer to the initial [ - endptr where to return the end pointer - -Returns: TRUE or FALSE -*/ - -static BOOL -check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr) -{ -pcre_uchar terminator; /* Don't combine these lines; the Solaris cc */ -terminator = *(++ptr); /* compiler warns about "non-constant" initializer. */ -for (++ptr; *ptr != CHAR_NULL; ptr++) - { - if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) - ptr++; - else if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE; - else - { - if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) - { - *endptr = ptr; - return TRUE; - } - if (*ptr == CHAR_LEFT_SQUARE_BRACKET && - (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT || - ptr[1] == CHAR_EQUALS_SIGN) && - check_posix_syntax(ptr, endptr)) - return FALSE; - } - } -return FALSE; -} - - - - -/************************************************* -* Check POSIX class name * -*************************************************/ - -/* This function is called to check the name given in a POSIX-style class entry -such as [:alnum:]. - -Arguments: - ptr points to the first letter - len the length of the name - -Returns: a value representing the name, or -1 if unknown -*/ - -static int -check_posix_name(const pcre_uchar *ptr, int len) -{ -const char *pn = posix_names; -register int yield = 0; -while (posix_name_lengths[yield] != 0) - { - if (len == posix_name_lengths[yield] && - STRNCMP_UC_C8(ptr, pn, (unsigned int)len) == 0) return yield; - pn += posix_name_lengths[yield] + 1; - yield++; - } -return -1; -} - - -/************************************************* -* Adjust OP_RECURSE items in repeated group * -*************************************************/ - -/* OP_RECURSE items contain an offset from the start of the regex to the group -that is referenced. This means that groups can be replicated for fixed -repetition simply by copying (because the recursion is allowed to refer to -earlier groups that are outside the current group). However, when a group is -optional (i.e. the minimum quantifier is zero), OP_BRAZERO or OP_SKIPZERO is -inserted before it, after it has been compiled. This means that any OP_RECURSE -items within it that refer to the group itself or any contained groups have to -have their offsets adjusted. That one of the jobs of this function. Before it -is called, the partially compiled regex must be temporarily terminated with -OP_END. - -This function has been extended with the possibility of forward references for -recursions and subroutine calls. It must also check the list of such references -for the group we are dealing with. If it finds that one of the recursions in -the current group is on this list, it adjusts the offset in the list, not the -value in the reference (which is a group number). - -Arguments: - group points to the start of the group - adjust the amount by which the group is to be moved - utf TRUE in UTF-8 / UTF-16 / UTF-32 mode - cd contains pointers to tables etc. - save_hwm the hwm forward reference pointer at the start of the group - -Returns: nothing -*/ - -static void -adjust_recurse(pcre_uchar *group, int adjust, BOOL utf, compile_data *cd, - pcre_uchar *save_hwm) -{ -pcre_uchar *ptr = group; - -while ((ptr = (pcre_uchar *)find_recurse(ptr, utf)) != NULL) - { - int offset; - pcre_uchar *hc; - - /* See if this recursion is on the forward reference list. If so, adjust the - reference. */ - - for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE) - { - offset = (int)GET(hc, 0); - if (cd->start_code + offset == ptr + 1) - { - PUT(hc, 0, offset + adjust); - break; - } - } - - /* Otherwise, adjust the recursion offset if it's after the start of this - group. */ - - if (hc >= cd->hwm) - { - offset = (int)GET(ptr, 1); - if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust); - } - - ptr += 1 + LINK_SIZE; - } -} - - - -/************************************************* -* Insert an automatic callout point * -*************************************************/ - -/* This function is called when the PCRE_AUTO_CALLOUT option is set, to insert -callout points before each pattern item. - -Arguments: - code current code pointer - ptr current pattern pointer - cd pointers to tables etc - -Returns: new code pointer -*/ - -static pcre_uchar * -auto_callout(pcre_uchar *code, const pcre_uchar *ptr, compile_data *cd) -{ -*code++ = OP_CALLOUT; -*code++ = 255; -PUT(code, 0, (int)(ptr - cd->start_pattern)); /* Pattern offset */ -PUT(code, LINK_SIZE, 0); /* Default length */ -return code + 2 * LINK_SIZE; -} - - - -/************************************************* -* Complete a callout item * -*************************************************/ - -/* A callout item contains the length of the next item in the pattern, which -we can't fill in till after we have reached the relevant point. This is used -for both automatic and manual callouts. - -Arguments: - previous_callout points to previous callout item - ptr current pattern pointer - cd pointers to tables etc - -Returns: nothing -*/ - -static void -complete_callout(pcre_uchar *previous_callout, const pcre_uchar *ptr, compile_data *cd) -{ -int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2)); -PUT(previous_callout, 2 + LINK_SIZE, length); -} - - - -#ifdef SUPPORT_UCP -/************************************************* -* Get othercase range * -*************************************************/ - -/* This function is passed the start and end of a class range, in UTF-8 mode -with UCP support. It searches up the characters, looking for ranges of -characters in the "other" case. Each call returns the next one, updating the -start address. A character with multiple other cases is returned on its own -with a special return value. - -Arguments: - cptr points to starting character value; updated - d end value - ocptr where to put start of othercase range - odptr where to put end of othercase range - -Yield: -1 when no more - 0 when a range is returned - >0 the CASESET offset for char with multiple other cases - in this case, ocptr contains the original -*/ - -static int -get_othercase_range(pcre_uint32 *cptr, pcre_uint32 d, pcre_uint32 *ocptr, - pcre_uint32 *odptr) -{ -pcre_uint32 c, othercase, next; -unsigned int co; - -/* Find the first character that has an other case. If it has multiple other -cases, return its case offset value. */ - -for (c = *cptr; c <= d; c++) - { - if ((co = UCD_CASESET(c)) != 0) - { - *ocptr = c++; /* Character that has the set */ - *cptr = c; /* Rest of input range */ - return (int)co; - } - if ((othercase = UCD_OTHERCASE(c)) != c) break; - } - -if (c > d) return -1; /* Reached end of range */ - -*ocptr = othercase; -next = othercase + 1; - -for (++c; c <= d; c++) - { - if (UCD_OTHERCASE(c) != next) break; - next++; - } - -*odptr = next - 1; /* End of othercase range */ -*cptr = c; /* Rest of input range */ -return 0; -} -#endif /* SUPPORT_UCP */ - - - -/************************************************* -* Add a character or range to a class * -*************************************************/ - -/* This function packages up the logic of adding a character or range of -characters to a class. The character values in the arguments will be within the -valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is -mutually recursive with the function immediately below. - -Arguments: - classbits the bit map for characters < 256 - uchardptr points to the pointer for extra data - options the options word - cd contains pointers to tables etc. - start start of range character - end end of range character - -Returns: the number of < 256 characters added - the pointer to extra data is updated -*/ - -static int -add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options, - compile_data *cd, pcre_uint32 start, pcre_uint32 end) -{ -pcre_uint32 c; -int n8 = 0; - -/* If caseless matching is required, scan the range and process alternate -cases. In Unicode, there are 8-bit characters that have alternate cases that -are greater than 255 and vice-versa. Sometimes we can just extend the original -range. */ - -if ((options & PCRE_CASELESS) != 0) - { -#ifdef SUPPORT_UCP - if ((options & PCRE_UTF8) != 0) - { - int rc; - pcre_uint32 oc, od; - - options &= ~PCRE_CASELESS; /* Remove for recursive calls */ - c = start; - - while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0) - { - /* Handle a single character that has more than one other case. */ - - if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd, - PRIV(ucd_caseless_sets) + rc, oc); - - /* Do nothing if the other case range is within the original range. */ - - else if (oc >= start && od <= end) continue; - - /* Extend the original range if there is overlap, noting that if oc < c, we - can't have od > end because a subrange is always shorter than the basic - range. Otherwise, use a recursive call to add the additional range. */ - - else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */ - else if (od > end && oc <= end + 1) end = od; /* Extend upwards */ - else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od); - } - } - else -#endif /* SUPPORT_UCP */ - - /* Not UTF-mode, or no UCP */ - - for (c = start; c <= end && c < 256; c++) - { - SETBIT(classbits, cd->fcc[c]); - n8++; - } - } - -/* Now handle the original range. Adjust the final value according to the bit -length - this means that the same lists of (e.g.) horizontal spaces can be used -in all cases. */ - -#if defined COMPILE_PCRE8 -#ifdef SUPPORT_UTF - if ((options & PCRE_UTF8) == 0) -#endif - if (end > 0xff) end = 0xff; - -#elif defined COMPILE_PCRE16 -#ifdef SUPPORT_UTF - if ((options & PCRE_UTF16) == 0) -#endif - if (end > 0xffff) end = 0xffff; - -#endif /* COMPILE_PCRE[8|16] */ - -/* If all characters are less than 256, use the bit map. Otherwise use extra -data. */ - -if (end < 0x100) - { - for (c = start; c <= end; c++) - { - n8++; - SETBIT(classbits, c); - } - } - -else - { - pcre_uchar *uchardata = *uchardptr; - -#ifdef SUPPORT_UTF - if ((options & PCRE_UTF8) != 0) /* All UTFs use the same flag bit */ - { - if (start < end) - { - *uchardata++ = XCL_RANGE; - uchardata += PRIV(ord2utf)(start, uchardata); - uchardata += PRIV(ord2utf)(end, uchardata); - } - else if (start == end) - { - *uchardata++ = XCL_SINGLE; - uchardata += PRIV(ord2utf)(start, uchardata); - } - } - else -#endif /* SUPPORT_UTF */ - - /* Without UTF support, character values are constrained by the bit length, - and can only be > 256 for 16-bit and 32-bit libraries. */ - -#ifdef COMPILE_PCRE8 - {} -#else - if (start < end) - { - *uchardata++ = XCL_RANGE; - *uchardata++ = start; - *uchardata++ = end; - } - else if (start == end) - { - *uchardata++ = XCL_SINGLE; - *uchardata++ = start; - } -#endif - - *uchardptr = uchardata; /* Updata extra data pointer */ - } - -return n8; /* Number of 8-bit characters */ -} - - - - -/************************************************* -* Add a list of characters to a class * -*************************************************/ - -/* This function is used for adding a list of case-equivalent characters to a -class, and also for adding a list of horizontal or vertical whitespace. If the -list is in order (which it should be), ranges of characters are detected and -handled appropriately. This function is mutually recursive with the function -above. - -Arguments: - classbits the bit map for characters < 256 - uchardptr points to the pointer for extra data - options the options word - cd contains pointers to tables etc. - p points to row of 32-bit values, terminated by NOTACHAR - except character to omit; this is used when adding lists of - case-equivalent characters to avoid including the one we - already know about - -Returns: the number of < 256 characters added - the pointer to extra data is updated -*/ - -static int -add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options, - compile_data *cd, const pcre_uint32 *p, unsigned int except) -{ -int n8 = 0; -while (p[0] < NOTACHAR) - { - int n = 0; - if (p[0] != except) - { - while(p[n+1] == p[0] + n + 1) n++; - n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]); - } - p += n + 1; - } -return n8; -} - - - -/************************************************* -* Add characters not in a list to a class * -*************************************************/ - -/* This function is used for adding the complement of a list of horizontal or -vertical whitespace to a class. The list must be in order. - -Arguments: - classbits the bit map for characters < 256 - uchardptr points to the pointer for extra data - options the options word - cd contains pointers to tables etc. - p points to row of 32-bit values, terminated by NOTACHAR - -Returns: the number of < 256 characters added - the pointer to extra data is updated -*/ - -static int -add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, - int options, compile_data *cd, const pcre_uint32 *p) -{ -BOOL utf = (options & PCRE_UTF8) != 0; -int n8 = 0; -if (p[0] > 0) - n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1); -while (p[0] < NOTACHAR) - { - while (p[1] == p[0] + 1) p++; - n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1, - (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1); - p++; - } -return n8; -} - - - -/************************************************* -* Compile one branch * -*************************************************/ - -/* Scan the pattern, compiling it into the a vector. If the options are -changed during the branch, the pointer is used to change the external options -bits. This function is used during the pre-compile phase when we are trying -to find out the amount of memory needed, as well as during the real compile -phase. The value of lengthptr distinguishes the two phases. - -Arguments: - optionsptr pointer to the option bits - codeptr points to the pointer to the current code point - ptrptr points to the current pattern pointer - errorcodeptr points to error code variable - firstcharptr place to put the first required character - firstcharflagsptr place to put the first character flags, or a negative number - reqcharptr place to put the last required character - reqcharflagsptr place to put the last required character flags, or a negative number - bcptr points to current branch chain - cond_depth conditional nesting depth - cd contains pointers to tables etc. - lengthptr NULL during the real compile phase - points to length accumulator during pre-compile phase - -Returns: TRUE on success - FALSE, with *errorcodeptr set non-zero on error -*/ - -static BOOL -compile_branch(int *optionsptr, pcre_uchar **codeptr, - const pcre_uchar **ptrptr, int *errorcodeptr, - pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr, - pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr, - branch_chain *bcptr, int cond_depth, - compile_data *cd, int *lengthptr) -{ -int repeat_type, op_type; -int repeat_min = 0, repeat_max = 0; /* To please picky compilers */ -int bravalue = 0; -int greedy_default, greedy_non_default; -pcre_uint32 firstchar, reqchar; -pcre_int32 firstcharflags, reqcharflags; -pcre_uint32 zeroreqchar, zerofirstchar; -pcre_int32 zeroreqcharflags, zerofirstcharflags; -pcre_int32 req_caseopt, reqvary, tempreqvary; -int options = *optionsptr; /* May change dynamically */ -int after_manual_callout = 0; -int length_prevgroup = 0; -register pcre_uint32 c; -int escape; -register pcre_uchar *code = *codeptr; -pcre_uchar *last_code = code; -pcre_uchar *orig_code = code; -pcre_uchar *tempcode; -BOOL inescq = FALSE; -BOOL groupsetfirstchar = FALSE; -const pcre_uchar *ptr = *ptrptr; -const pcre_uchar *tempptr; -const pcre_uchar *nestptr = NULL; -pcre_uchar *previous = NULL; -pcre_uchar *previous_callout = NULL; -pcre_uchar *save_hwm = NULL; -pcre_uint8 classbits[32]; - -/* We can fish out the UTF-8 setting once and for all into a BOOL, but we -must not do this for other options (e.g. PCRE_EXTENDED) because they may change -dynamically as we process the pattern. */ - -#ifdef SUPPORT_UTF -/* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */ -BOOL utf = (options & PCRE_UTF8) != 0; -#ifndef COMPILE_PCRE32 -pcre_uchar utf_chars[6]; -#endif -#else -BOOL utf = FALSE; -#endif - -/* Helper variables for OP_XCLASS opcode (for characters > 255). We define -class_uchardata always so that it can be passed to add_to_class() always, -though it will not be used in non-UTF 8-bit cases. This avoids having to supply -alternative calls for the different cases. */ - -pcre_uchar *class_uchardata; -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 -BOOL xclass; -pcre_uchar *class_uchardata_base; -#endif - -#ifdef PCRE_DEBUG -if (lengthptr != NULL) DPRINTF((">> start branch\n")); -#endif - -/* Set up the default and non-default settings for greediness */ - -greedy_default = ((options & PCRE_UNGREEDY) != 0); -greedy_non_default = greedy_default ^ 1; - -/* Initialize no first byte, no required byte. REQ_UNSET means "no char -matching encountered yet". It gets changed to REQ_NONE if we hit something that -matches a non-fixed char first char; reqchar just remains unset if we never -find one. - -When we hit a repeat whose minimum is zero, we may have to adjust these values -to take the zero repeat into account. This is implemented by setting them to -zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual -item types that can be repeated set these backoff variables appropriately. */ - -firstchar = reqchar = zerofirstchar = zeroreqchar = 0; -firstcharflags = reqcharflags = zerofirstcharflags = zeroreqcharflags = REQ_UNSET; - -/* The variable req_caseopt contains either the REQ_CASELESS value -or zero, according to the current setting of the caseless flag. The -REQ_CASELESS leaves the lower 28 bit empty. It is added into the -firstchar or reqchar variables to record the case status of the -value. This is used only for ASCII characters. */ - -req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS:0; - -/* Switch on next character until the end of the branch */ - -for (;; ptr++) - { - BOOL negate_class; - BOOL should_flip_negation; - BOOL possessive_quantifier; - BOOL is_quantifier; - BOOL is_recurse; - BOOL reset_bracount; - int class_has_8bitchar; - int class_one_char; - int newoptions; - int recno; - int refsign; - int skipbytes; - pcre_uint32 subreqchar, subfirstchar; - pcre_int32 subreqcharflags, subfirstcharflags; - int terminator; - unsigned int mclength; - unsigned int tempbracount; - pcre_uint32 ec; - pcre_uchar mcbuffer[8]; - - /* Get next character in the pattern */ - - c = *ptr; - - /* If we are at the end of a nested substitution, revert to the outer level - string. Nesting only happens one level deep. */ - - if (c == CHAR_NULL && nestptr != NULL) - { - ptr = nestptr; - nestptr = NULL; - c = *ptr; - } - - /* If we are in the pre-compile phase, accumulate the length used for the - previous cycle of this loop. */ - - if (lengthptr != NULL) - { -#ifdef PCRE_DEBUG - if (code > cd->hwm) cd->hwm = code; /* High water info */ -#endif - if (code > cd->start_workspace + cd->workspace_size - - WORK_SIZE_SAFETY_MARGIN) /* Check for overrun */ - { - *errorcodeptr = ERR52; - goto FAILED; - } - - /* There is at least one situation where code goes backwards: this is the - case of a zero quantifier after a class (e.g. [ab]{0}). At compile time, - the class is simply eliminated. However, it is created first, so we have to - allow memory for it. Therefore, don't ever reduce the length at this point. - */ - - if (code < last_code) code = last_code; - - /* Paranoid check for integer overflow */ - - if (OFLOW_MAX - *lengthptr < code - last_code) - { - *errorcodeptr = ERR20; - goto FAILED; - } - - *lengthptr += (int)(code - last_code); - DPRINTF(("length=%d added %d c=%c (0x%x)\n", *lengthptr, - (int)(code - last_code), c, c)); - - /* If "previous" is set and it is not at the start of the work space, move - it back to there, in order to avoid filling up the work space. Otherwise, - if "previous" is NULL, reset the current code pointer to the start. */ - - if (previous != NULL) - { - if (previous > orig_code) - { - memmove(orig_code, previous, IN_UCHARS(code - previous)); - code -= previous - orig_code; - previous = orig_code; - } - } - else code = orig_code; - - /* Remember where this code item starts so we can pick up the length - next time round. */ - - last_code = code; - } - - /* In the real compile phase, just check the workspace used by the forward - reference list. */ - - else if (cd->hwm > cd->start_workspace + cd->workspace_size - - WORK_SIZE_SAFETY_MARGIN) - { - *errorcodeptr = ERR52; - goto FAILED; - } - - /* If in \Q...\E, check for the end; if not, we have a literal */ - - if (inescq && c != CHAR_NULL) - { - if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E) - { - inescq = FALSE; - ptr++; - continue; - } - else - { - if (previous_callout != NULL) - { - if (lengthptr == NULL) /* Don't attempt in pre-compile phase */ - complete_callout(previous_callout, ptr, cd); - previous_callout = NULL; - } - if ((options & PCRE_AUTO_CALLOUT) != 0) - { - previous_callout = code; - code = auto_callout(code, ptr, cd); - } - goto NORMAL_CHAR; - } - /* Control does not reach here. */ - } - - /* In extended mode, skip white space and comments. We need a loop in order - to check for more white space and more comments after a comment. */ - - if ((options & PCRE_EXTENDED) != 0) - { - for (;;) - { - while (MAX_255(c) && (cd->ctypes[c] & ctype_space) != 0) c = *(++ptr); - if (c != CHAR_NUMBER_SIGN) break; - ptr++; - while (*ptr != CHAR_NULL) - { - if (IS_NEWLINE(ptr)) /* For non-fixed-length newline cases, */ - { /* IS_NEWLINE sets cd->nllen. */ - ptr += cd->nllen; - break; - } - ptr++; -#ifdef SUPPORT_UTF - if (utf) FORWARDCHAR(ptr); -#endif - } - c = *ptr; /* Either NULL or the char after a newline */ - } - } - - /* See if the next thing is a quantifier. */ - - is_quantifier = - c == CHAR_ASTERISK || c == CHAR_PLUS || c == CHAR_QUESTION_MARK || - (c == CHAR_LEFT_CURLY_BRACKET && is_counted_repeat(ptr+1)); - - /* Fill in length of a previous callout, except when the next thing is a - quantifier or when processing a property substitution string in UCP mode. */ - - if (!is_quantifier && previous_callout != NULL && nestptr == NULL && - after_manual_callout-- <= 0) - { - if (lengthptr == NULL) /* Don't attempt in pre-compile phase */ - complete_callout(previous_callout, ptr, cd); - previous_callout = NULL; - } - - /* Create auto callout, except for quantifiers, or while processing property - strings that are substituted for \w etc in UCP mode. */ - - if ((options & PCRE_AUTO_CALLOUT) != 0 && !is_quantifier && nestptr == NULL) - { - previous_callout = code; - code = auto_callout(code, ptr, cd); - } - - /* Process the next pattern item. */ - - switch(c) - { - /* ===================================================================*/ - case CHAR_NULL: /* The branch terminates at string end */ - case CHAR_VERTICAL_LINE: /* or | or ) */ - case CHAR_RIGHT_PARENTHESIS: - *firstcharptr = firstchar; - *firstcharflagsptr = firstcharflags; - *reqcharptr = reqchar; - *reqcharflagsptr = reqcharflags; - *codeptr = code; - *ptrptr = ptr; - if (lengthptr != NULL) - { - if (OFLOW_MAX - *lengthptr < code - last_code) - { - *errorcodeptr = ERR20; - goto FAILED; - } - *lengthptr += (int)(code - last_code); /* To include callout length */ - DPRINTF((">> end branch\n")); - } - return TRUE; - - - /* ===================================================================*/ - /* Handle single-character metacharacters. In multiline mode, ^ disables - the setting of any following char as a first character. */ - - case CHAR_CIRCUMFLEX_ACCENT: - previous = NULL; - if ((options & PCRE_MULTILINE) != 0) - { - if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE; - *code++ = OP_CIRCM; - } - else *code++ = OP_CIRC; - break; - - case CHAR_DOLLAR_SIGN: - previous = NULL; - *code++ = ((options & PCRE_MULTILINE) != 0)? OP_DOLLM : OP_DOLL; - break; - - /* There can never be a first char if '.' is first, whatever happens about - repeats. The value of reqchar doesn't change either. */ - - case CHAR_DOT: - if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE; - zerofirstchar = firstchar; - zerofirstcharflags = firstcharflags; - zeroreqchar = reqchar; - zeroreqcharflags = reqcharflags; - previous = code; - *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY; - break; - - - /* ===================================================================*/ - /* Character classes. If the included characters are all < 256, we build a - 32-byte bitmap of the permitted characters, except in the special case - where there is only one such character. For negated classes, we build the - map as usual, then invert it at the end. However, we use a different opcode - so that data characters > 255 can be handled correctly. - - If the class contains characters outside the 0-255 range, a different - opcode is compiled. It may optionally have a bit map for characters < 256, - but those above are are explicitly listed afterwards. A flag byte tells - whether the bitmap is present, and whether this is a negated class or not. - - In JavaScript compatibility mode, an isolated ']' causes an error. In - default (Perl) mode, it is treated as a data character. */ - - case CHAR_RIGHT_SQUARE_BRACKET: - if ((cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0) - { - *errorcodeptr = ERR64; - goto FAILED; - } - goto NORMAL_CHAR; - - /* In another (POSIX) regex library, the ugly syntax [[:<:]] and [[:>:]] is - used for "start of word" and "end of word". As these are otherwise illegal - sequences, we don't break anything by recognizing them. They are replaced - by \b(?=\w) and \b(?<=\w) respectively. Sequences like [a[:<:]] are - erroneous and are handled by the normal code below. */ - - case CHAR_LEFT_SQUARE_BRACKET: - if (STRNCMP_UC_C8(ptr+1, STRING_WEIRD_STARTWORD, 6) == 0) - { - nestptr = ptr + 7; - ptr = sub_start_of_word - 1; - continue; - } - - if (STRNCMP_UC_C8(ptr+1, STRING_WEIRD_ENDWORD, 6) == 0) - { - nestptr = ptr + 7; - ptr = sub_end_of_word - 1; - continue; - } - - /* Handle a real character class. */ - - previous = code; - - /* PCRE supports POSIX class stuff inside a class. Perl gives an error if - they are encountered at the top level, so we'll do that too. */ - - if ((ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT || - ptr[1] == CHAR_EQUALS_SIGN) && - check_posix_syntax(ptr, &tempptr)) - { - *errorcodeptr = (ptr[1] == CHAR_COLON)? ERR13 : ERR31; - goto FAILED; - } - - /* If the first character is '^', set the negation flag and skip it. Also, - if the first few characters (either before or after ^) are \Q\E or \E we - skip them too. This makes for compatibility with Perl. */ - - negate_class = FALSE; - for (;;) - { - c = *(++ptr); - if (c == CHAR_BACKSLASH) - { - if (ptr[1] == CHAR_E) - ptr++; - else if (STRNCMP_UC_C8(ptr + 1, STR_Q STR_BACKSLASH STR_E, 3) == 0) - ptr += 3; - else - break; - } - else if (!negate_class && c == CHAR_CIRCUMFLEX_ACCENT) - negate_class = TRUE; - else break; - } - - /* Empty classes are allowed in JavaScript compatibility mode. Otherwise, - an initial ']' is taken as a data character -- the code below handles - that. In JS mode, [] must always fail, so generate OP_FAIL, whereas - [^] must match any character, so generate OP_ALLANY. */ - - if (c == CHAR_RIGHT_SQUARE_BRACKET && - (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0) - { - *code++ = negate_class? OP_ALLANY : OP_FAIL; - if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE; - zerofirstchar = firstchar; - zerofirstcharflags = firstcharflags; - break; - } - - /* If a class contains a negative special such as \S, we need to flip the - negation flag at the end, so that support for characters > 255 works - correctly (they are all included in the class). */ - - should_flip_negation = FALSE; - - /* For optimization purposes, we track some properties of the class: - class_has_8bitchar will be non-zero if the class contains at least one < - 256 character; class_one_char will be 1 if the class contains just one - character. */ - - class_has_8bitchar = 0; - class_one_char = 0; - - /* Initialize the 32-char bit map to all zeros. We build the map in a - temporary bit of memory, in case the class contains fewer than two - 8-bit characters because in that case the compiled code doesn't use the bit - map. */ - - memset(classbits, 0, 32 * sizeof(pcre_uint8)); - -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - xclass = FALSE; - class_uchardata = code + LINK_SIZE + 2; /* For XCLASS items */ - class_uchardata_base = class_uchardata; /* Save the start */ -#endif - - /* Process characters until ] is reached. By writing this as a "do" it - means that an initial ] is taken as a data character. At the start of the - loop, c contains the first byte of the character. */ - - if (c != CHAR_NULL) do - { - const pcre_uchar *oldptr; - -#ifdef SUPPORT_UTF - if (utf && HAS_EXTRALEN(c)) - { /* Braces are required because the */ - GETCHARLEN(c, ptr, ptr); /* macro generates multiple statements */ - } -#endif - -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - /* In the pre-compile phase, accumulate the length of any extra - data and reset the pointer. This is so that very large classes that - contain a zillion > 255 characters no longer overwrite the work space - (which is on the stack). We have to remember that there was XCLASS data, - however. */ - - if (lengthptr != NULL && class_uchardata > class_uchardata_base) - { - xclass = TRUE; - *lengthptr += class_uchardata - class_uchardata_base; - class_uchardata = class_uchardata_base; - } -#endif - - /* Inside \Q...\E everything is literal except \E */ - - if (inescq) - { - if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E) /* If we are at \E */ - { - inescq = FALSE; /* Reset literal state */ - ptr++; /* Skip the 'E' */ - continue; /* Carry on with next */ - } - goto CHECK_RANGE; /* Could be range if \E follows */ - } - - /* Handle POSIX class names. Perl allows a negation extension of the - form [:^name:]. A square bracket that doesn't match the syntax is - treated as a literal. We also recognize the POSIX constructions - [.ch.] and [=ch=] ("collating elements") and fault them, as Perl - 5.6 and 5.8 do. */ - - if (c == CHAR_LEFT_SQUARE_BRACKET && - (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT || - ptr[1] == CHAR_EQUALS_SIGN) && check_posix_syntax(ptr, &tempptr)) - { - BOOL local_negate = FALSE; - int posix_class, taboffset, tabopt; - register const pcre_uint8 *cbits = cd->cbits; - pcre_uint8 pbits[32]; - - if (ptr[1] != CHAR_COLON) - { - *errorcodeptr = ERR31; - goto FAILED; - } - - ptr += 2; - if (*ptr == CHAR_CIRCUMFLEX_ACCENT) - { - local_negate = TRUE; - should_flip_negation = TRUE; /* Note negative special */ - ptr++; - } - - posix_class = check_posix_name(ptr, (int)(tempptr - ptr)); - if (posix_class < 0) - { - *errorcodeptr = ERR30; - goto FAILED; - } - - /* If matching is caseless, upper and lower are converted to - alpha. This relies on the fact that the class table starts with - alpha, lower, upper as the first 3 entries. */ - - if ((options & PCRE_CASELESS) != 0 && posix_class <= 2) - posix_class = 0; - - /* When PCRE_UCP is set, some of the POSIX classes are converted to - different escape sequences that use Unicode properties \p or \P. Others - that are not available via \p or \P generate XCL_PROP/XCL_NOTPROP - directly. */ - -#ifdef SUPPORT_UCP - if ((options & PCRE_UCP) != 0) - { - unsigned int ptype = 0; - int pc = posix_class + ((local_negate)? POSIX_SUBSIZE/2 : 0); - - /* The posix_substitutes table specifies which POSIX classes can be - converted to \p or \P items. */ - - if (posix_substitutes[pc] != NULL) - { - nestptr = tempptr + 1; - ptr = posix_substitutes[pc] - 1; - continue; - } - - /* There are three other classes that generate special property calls - that are recognized only in an XCLASS. */ - - else switch(posix_class) - { - case PC_GRAPH: - ptype = PT_PXGRAPH; - /* Fall through */ - case PC_PRINT: - if (ptype == 0) ptype = PT_PXPRINT; - /* Fall through */ - case PC_PUNCT: - if (ptype == 0) ptype = PT_PXPUNCT; - *class_uchardata++ = local_negate? XCL_NOTPROP : XCL_PROP; - *class_uchardata++ = ptype; - *class_uchardata++ = 0; - ptr = tempptr + 1; - continue; - - /* For all other POSIX classes, no special action is taken in UCP - mode. Fall through to the non_UCP case. */ - - default: - break; - } - } -#endif - /* In the non-UCP case, or when UCP makes no difference, we build the - bit map for the POSIX class in a chunk of local store because we may be - adding and subtracting from it, and we don't want to subtract bits that - may be in the main map already. At the end we or the result into the - bit map that is being built. */ - - posix_class *= 3; - - /* Copy in the first table (always present) */ - - memcpy(pbits, cbits + posix_class_maps[posix_class], - 32 * sizeof(pcre_uint8)); - - /* If there is a second table, add or remove it as required. */ - - taboffset = posix_class_maps[posix_class + 1]; - tabopt = posix_class_maps[posix_class + 2]; - - if (taboffset >= 0) - { - if (tabopt >= 0) - for (c = 0; c < 32; c++) pbits[c] |= cbits[c + taboffset]; - else - for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset]; - } - - /* Now see if we need to remove any special characters. An option - value of 1 removes vertical space and 2 removes underscore. */ - - if (tabopt < 0) tabopt = -tabopt; - if (tabopt == 1) pbits[1] &= ~0x3c; - else if (tabopt == 2) pbits[11] &= 0x7f; - - /* Add the POSIX table or its complement into the main table that is - being built and we are done. */ - - if (local_negate) - for (c = 0; c < 32; c++) classbits[c] |= ~pbits[c]; - else - for (c = 0; c < 32; c++) classbits[c] |= pbits[c]; - - ptr = tempptr + 1; - /* Every class contains at least one < 256 character. */ - class_has_8bitchar = 1; - /* Every class contains at least two characters. */ - class_one_char = 2; - continue; /* End of POSIX syntax handling */ - } - - /* Backslash may introduce a single character, or it may introduce one - of the specials, which just set a flag. The sequence \b is a special - case. Inside a class (and only there) it is treated as backspace. We - assume that other escapes have more than one character in them, so - speculatively set both class_has_8bitchar and class_one_char bigger - than one. Unrecognized escapes fall through and are either treated - as literal characters (by default), or are faulted if - PCRE_EXTRA is set. */ - - if (c == CHAR_BACKSLASH) - { - escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, - TRUE); - if (*errorcodeptr != 0) goto FAILED; - if (escape == 0) c = ec; - else if (escape == ESC_b) c = CHAR_BS; /* \b is backspace in a class */ - else if (escape == ESC_N) /* \N is not supported in a class */ - { - *errorcodeptr = ERR71; - goto FAILED; - } - else if (escape == ESC_Q) /* Handle start of quoted string */ - { - if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E) - { - ptr += 2; /* avoid empty string */ - } - else inescq = TRUE; - continue; - } - else if (escape == ESC_E) continue; /* Ignore orphan \E */ - - else - { - register const pcre_uint8 *cbits = cd->cbits; - /* Every class contains at least two < 256 characters. */ - class_has_8bitchar++; - /* Every class contains at least two characters. */ - class_one_char += 2; - - switch (escape) - { -#ifdef SUPPORT_UCP - case ESC_du: /* These are the values given for \d etc */ - case ESC_DU: /* when PCRE_UCP is set. We replace the */ - case ESC_wu: /* escape sequence with an appropriate \p */ - case ESC_WU: /* or \P to test Unicode properties instead */ - case ESC_su: /* of the default ASCII testing. */ - case ESC_SU: - nestptr = ptr; - ptr = substitutes[escape - ESC_DU] - 1; /* Just before substitute */ - class_has_8bitchar--; /* Undo! */ - continue; -#endif - case ESC_d: - for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_digit]; - continue; - - case ESC_D: - should_flip_negation = TRUE; - for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit]; - continue; - - case ESC_w: - for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_word]; - continue; - - case ESC_W: - should_flip_negation = TRUE; - for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word]; - continue; - - /* Perl 5.004 onwards omitted VT from \s, but restored it at Perl - 5.18. Before PCRE 8.34, we had to preserve the VT bit if it was - previously set by something earlier in the character class. - Luckily, the value of CHAR_VT is 0x0b in both ASCII and EBCDIC, so - we could just adjust the appropriate bit. From PCRE 8.34 we no - longer treat \s and \S specially. */ - - case ESC_s: - for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_space]; - continue; - - case ESC_S: - should_flip_negation = TRUE; - for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space]; - continue; - - /* The rest apply in both UCP and non-UCP cases. */ - - case ESC_h: - (void)add_list_to_class(classbits, &class_uchardata, options, cd, - PRIV(hspace_list), NOTACHAR); - continue; - - case ESC_H: - (void)add_not_list_to_class(classbits, &class_uchardata, options, - cd, PRIV(hspace_list)); - continue; - - case ESC_v: - (void)add_list_to_class(classbits, &class_uchardata, options, cd, - PRIV(vspace_list), NOTACHAR); - continue; - - case ESC_V: - (void)add_not_list_to_class(classbits, &class_uchardata, options, - cd, PRIV(vspace_list)); - continue; - -#ifdef SUPPORT_UCP - case ESC_p: - case ESC_P: - { - BOOL negated; - unsigned int ptype = 0, pdata = 0; - if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr)) - goto FAILED; - *class_uchardata++ = ((escape == ESC_p) != negated)? - XCL_PROP : XCL_NOTPROP; - *class_uchardata++ = ptype; - *class_uchardata++ = pdata; - class_has_8bitchar--; /* Undo! */ - continue; - } -#endif - /* Unrecognized escapes are faulted if PCRE is running in its - strict mode. By default, for compatibility with Perl, they are - treated as literals. */ - - default: - if ((options & PCRE_EXTRA) != 0) - { - *errorcodeptr = ERR7; - goto FAILED; - } - class_has_8bitchar--; /* Undo the speculative increase. */ - class_one_char -= 2; /* Undo the speculative increase. */ - c = *ptr; /* Get the final character and fall through */ - break; - } - } - - /* Fall through if the escape just defined a single character (c >= 0). - This may be greater than 256. */ - - escape = 0; - - } /* End of backslash handling */ - - /* A character may be followed by '-' to form a range. However, Perl does - not permit ']' to be the end of the range. A '-' character at the end is - treated as a literal. Perl ignores orphaned \E sequences entirely. The - code for handling \Q and \E is messy. */ - - CHECK_RANGE: - while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E) - { - inescq = FALSE; - ptr += 2; - } - oldptr = ptr; - - /* Remember if \r or \n were explicitly used */ - - if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF; - - /* Check for range */ - - if (!inescq && ptr[1] == CHAR_MINUS) - { - pcre_uint32 d; - ptr += 2; - while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2; - - /* If we hit \Q (not followed by \E) at this point, go into escaped - mode. */ - - while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_Q) - { - ptr += 2; - if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) - { ptr += 2; continue; } - inescq = TRUE; - break; - } - - /* Minus (hyphen) at the end of a class is treated as a literal, so put - back the pointer and jump to handle the character that preceded it. */ - - if (*ptr == CHAR_NULL || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET)) - { - ptr = oldptr; - goto CLASS_SINGLE_CHARACTER; - } - - /* Otherwise, we have a potential range; pick up the next character */ - -#ifdef SUPPORT_UTF - if (utf) - { /* Braces are required because the */ - GETCHARLEN(d, ptr, ptr); /* macro generates multiple statements */ - } - else -#endif - d = *ptr; /* Not UTF-8 mode */ - - /* The second part of a range can be a single-character escape - sequence, but not any of the other escapes. Perl treats a hyphen as a - literal in such circumstances. However, in Perl's warning mode, a - warning is given, so PCRE now faults it as it is almost certainly a - mistake on the user's part. */ - - if (!inescq) - { - if (d == CHAR_BACKSLASH) - { - int descape; - descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE); - if (*errorcodeptr != 0) goto FAILED; - - /* 0 means a character was put into d; \b is backspace; any other - special causes an error. */ - - if (descape != 0) - { - if (descape == ESC_b) d = CHAR_BS; else - { - *errorcodeptr = ERR83; - goto FAILED; - } - } - } - - /* A hyphen followed by a POSIX class is treated in the same way. */ - - else if (d == CHAR_LEFT_SQUARE_BRACKET && - (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT || - ptr[1] == CHAR_EQUALS_SIGN) && - check_posix_syntax(ptr, &tempptr)) - { - *errorcodeptr = ERR83; - goto FAILED; - } - } - - /* Check that the two values are in the correct order. Optimize - one-character ranges. */ - - if (d < c) - { - *errorcodeptr = ERR8; - goto FAILED; - } - if (d == c) goto CLASS_SINGLE_CHARACTER; /* A few lines below */ - - /* We have found a character range, so single character optimizations - cannot be done anymore. Any value greater than 1 indicates that there - is more than one character. */ - - class_one_char = 2; - - /* Remember an explicit \r or \n, and add the range to the class. */ - - if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF; - - class_has_8bitchar += - add_to_class(classbits, &class_uchardata, options, cd, c, d); - - continue; /* Go get the next char in the class */ - } - - /* Handle a single character - we can get here for a normal non-escape - char, or after \ that introduces a single character or for an apparent - range that isn't. Only the value 1 matters for class_one_char, so don't - increase it if it is already 2 or more ... just in case there's a class - with a zillion characters in it. */ - - CLASS_SINGLE_CHARACTER: - if (class_one_char < 2) class_one_char++; - - /* If class_one_char is 1, we have the first single character in the - class, and there have been no prior ranges, or XCLASS items generated by - escapes. If this is the final character in the class, we can optimize by - turning the item into a 1-character OP_CHAR[I] if it's positive, or - OP_NOT[I] if it's negative. In the positive case, it can cause firstchar - to be set. Otherwise, there can be no first char if this item is first, - whatever repeat count may follow. In the case of reqchar, save the - previous value for reinstating. */ - - if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) - { - ptr++; - zeroreqchar = reqchar; - zeroreqcharflags = reqcharflags; - - if (negate_class) - { -#ifdef SUPPORT_UCP - int d; -#endif - if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE; - zerofirstchar = firstchar; - zerofirstcharflags = firstcharflags; - - /* For caseless UTF-8 mode when UCP support is available, check - whether this character has more than one other case. If so, generate - a special OP_NOTPROP item instead of OP_NOTI. */ - -#ifdef SUPPORT_UCP - if (utf && (options & PCRE_CASELESS) != 0 && - (d = UCD_CASESET(c)) != 0) - { - *code++ = OP_NOTPROP; - *code++ = PT_CLIST; - *code++ = d; - } - else -#endif - /* Char has only one other case, or UCP not available */ - - { - *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT; -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 - if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR) - code += PRIV(ord2utf)(c, code); - else -#endif - *code++ = c; - } - - /* We are finished with this character class */ - - goto END_CLASS; - } - - /* For a single, positive character, get the value into mcbuffer, and - then we can handle this with the normal one-character code. */ - -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 - if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR) - mclength = PRIV(ord2utf)(c, mcbuffer); - else -#endif - { - mcbuffer[0] = c; - mclength = 1; - } - goto ONE_CHAR; - } /* End of 1-char optimization */ - - /* There is more than one character in the class, or an XCLASS item - has been generated. Add this character to the class. */ - - class_has_8bitchar += - add_to_class(classbits, &class_uchardata, options, cd, c, c); - } - - /* Loop until ']' reached. This "while" is the end of the "do" far above. - If we are at the end of an internal nested string, revert to the outer - string. */ - - while (((c = *(++ptr)) != CHAR_NULL || - (nestptr != NULL && - (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != CHAR_NULL)) && - (c != CHAR_RIGHT_SQUARE_BRACKET || inescq)); - - /* Check for missing terminating ']' */ - - if (c == CHAR_NULL) - { - *errorcodeptr = ERR6; - goto FAILED; - } - - /* We will need an XCLASS if data has been placed in class_uchardata. In - the second phase this is a sufficient test. However, in the pre-compile - phase, class_uchardata gets emptied to prevent workspace overflow, so it - only if the very last character in the class needs XCLASS will it contain - anything at this point. For this reason, xclass gets set TRUE above when - uchar_classdata is emptied, and that's why this code is the way it is here - instead of just doing a test on class_uchardata below. */ - -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - if (class_uchardata > class_uchardata_base) xclass = TRUE; -#endif - - /* If this is the first thing in the branch, there can be no first char - setting, whatever the repeat count. Any reqchar setting must remain - unchanged after any kind of repeat. */ - - if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE; - zerofirstchar = firstchar; - zerofirstcharflags = firstcharflags; - zeroreqchar = reqchar; - zeroreqcharflags = reqcharflags; - - /* If there are characters with values > 255, we have to compile an - extended class, with its own opcode, unless there was a negated special - such as \S in the class, and PCRE_UCP is not set, because in that case all - characters > 255 are in the class, so any that were explicitly given as - well can be ignored. If (when there are explicit characters > 255 that must - be listed) there are no characters < 256, we can omit the bitmap in the - actual compiled code. */ - -#ifdef SUPPORT_UTF - if (xclass && (!should_flip_negation || (options & PCRE_UCP) != 0)) -#elif !defined COMPILE_PCRE8 - if (xclass && !should_flip_negation) -#endif -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - { - *class_uchardata++ = XCL_END; /* Marks the end of extra data */ - *code++ = OP_XCLASS; - code += LINK_SIZE; - *code = negate_class? XCL_NOT:0; - - /* If the map is required, move up the extra data to make room for it; - otherwise just move the code pointer to the end of the extra data. */ - - if (class_has_8bitchar > 0) - { - *code++ |= XCL_MAP; - memmove(code + (32 / sizeof(pcre_uchar)), code, - IN_UCHARS(class_uchardata - code)); - memcpy(code, classbits, 32); - code = class_uchardata + (32 / sizeof(pcre_uchar)); - } - else code = class_uchardata; - - /* Now fill in the complete length of the item */ - - PUT(previous, 1, (int)(code - previous)); - break; /* End of class handling */ - } -#endif - - /* If there are no characters > 255, or they are all to be included or - excluded, set the opcode to OP_CLASS or OP_NCLASS, depending on whether the - whole class was negated and whether there were negative specials such as \S - (non-UCP) in the class. Then copy the 32-byte map into the code vector, - negating it if necessary. */ - - *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS; - if (lengthptr == NULL) /* Save time in the pre-compile phase */ - { - if (negate_class) - for (c = 0; c < 32; c++) classbits[c] = ~classbits[c]; - memcpy(code, classbits, 32); - } - code += 32 / sizeof(pcre_uchar); - - END_CLASS: - break; - - - /* ===================================================================*/ - /* Various kinds of repeat; '{' is not necessarily a quantifier, but this - has been tested above. */ - - case CHAR_LEFT_CURLY_BRACKET: - if (!is_quantifier) goto NORMAL_CHAR; - ptr = read_repeat_counts(ptr+1, &repeat_min, &repeat_max, errorcodeptr); - if (*errorcodeptr != 0) goto FAILED; - goto REPEAT; - - case CHAR_ASTERISK: - repeat_min = 0; - repeat_max = -1; - goto REPEAT; - - case CHAR_PLUS: - repeat_min = 1; - repeat_max = -1; - goto REPEAT; - - case CHAR_QUESTION_MARK: - repeat_min = 0; - repeat_max = 1; - - REPEAT: - if (previous == NULL) - { - *errorcodeptr = ERR9; - goto FAILED; - } - - if (repeat_min == 0) - { - firstchar = zerofirstchar; /* Adjust for zero repeat */ - firstcharflags = zerofirstcharflags; - reqchar = zeroreqchar; /* Ditto */ - reqcharflags = zeroreqcharflags; - } - - /* Remember whether this is a variable length repeat */ - - reqvary = (repeat_min == repeat_max)? 0 : REQ_VARY; - - op_type = 0; /* Default single-char op codes */ - possessive_quantifier = FALSE; /* Default not possessive quantifier */ - - /* Save start of previous item, in case we have to move it up in order to - insert something before it. */ - - tempcode = previous; - - /* Before checking for a possessive quantifier, we must skip over - whitespace and comments in extended mode because Perl allows white space at - this point. */ - - if ((options & PCRE_EXTENDED) != 0) - { - const pcre_uchar *p = ptr + 1; - for (;;) - { - while (MAX_255(*p) && (cd->ctypes[*p] & ctype_space) != 0) p++; - if (*p != CHAR_NUMBER_SIGN) break; - p++; - while (*p != CHAR_NULL) - { - if (IS_NEWLINE(p)) /* For non-fixed-length newline cases, */ - { /* IS_NEWLINE sets cd->nllen. */ - p += cd->nllen; - break; - } - p++; -#ifdef SUPPORT_UTF - if (utf) FORWARDCHAR(p); -#endif - } /* Loop for comment characters */ - } /* Loop for multiple comments */ - ptr = p - 1; /* Character before the next significant one. */ - } - - /* If the next character is '+', we have a possessive quantifier. This - implies greediness, whatever the setting of the PCRE_UNGREEDY option. - If the next character is '?' this is a minimizing repeat, by default, - but if PCRE_UNGREEDY is set, it works the other way round. We change the - repeat type to the non-default. */ - - if (ptr[1] == CHAR_PLUS) - { - repeat_type = 0; /* Force greedy */ - possessive_quantifier = TRUE; - ptr++; - } - else if (ptr[1] == CHAR_QUESTION_MARK) - { - repeat_type = greedy_non_default; - ptr++; - } - else repeat_type = greedy_default; - - /* If previous was a recursion call, wrap it in atomic brackets so that - previous becomes the atomic group. All recursions were so wrapped in the - past, but it no longer happens for non-repeated recursions. In fact, the - repeated ones could be re-implemented independently so as not to need this, - but for the moment we rely on the code for repeating groups. */ - - if (*previous == OP_RECURSE) - { - memmove(previous + 1 + LINK_SIZE, previous, IN_UCHARS(1 + LINK_SIZE)); - *previous = OP_ONCE; - PUT(previous, 1, 2 + 2*LINK_SIZE); - previous[2 + 2*LINK_SIZE] = OP_KET; - PUT(previous, 3 + 2*LINK_SIZE, 2 + 2*LINK_SIZE); - code += 2 + 2 * LINK_SIZE; - length_prevgroup = 3 + 3*LINK_SIZE; - - /* When actually compiling, we need to check whether this was a forward - reference, and if so, adjust the offset. */ - - if (lengthptr == NULL && cd->hwm >= cd->start_workspace + LINK_SIZE) - { - int offset = GET(cd->hwm, -LINK_SIZE); - if (offset == previous + 1 - cd->start_code) - PUT(cd->hwm, -LINK_SIZE, offset + 1 + LINK_SIZE); - } - } - - /* Now handle repetition for the different types of item. */ - - /* If previous was a character or negated character match, abolish the item - and generate a repeat item instead. If a char item has a minimum of more - than one, ensure that it is set in reqchar - it might not be if a sequence - such as x{3} is the first thing in a branch because the x will have gone - into firstchar instead. */ - - if (*previous == OP_CHAR || *previous == OP_CHARI - || *previous == OP_NOT || *previous == OP_NOTI) - { - switch (*previous) - { - default: /* Make compiler happy. */ - case OP_CHAR: op_type = OP_STAR - OP_STAR; break; - case OP_CHARI: op_type = OP_STARI - OP_STAR; break; - case OP_NOT: op_type = OP_NOTSTAR - OP_STAR; break; - case OP_NOTI: op_type = OP_NOTSTARI - OP_STAR; break; - } - - /* Deal with UTF characters that take up more than one character. It's - easier to write this out separately than try to macrify it. Use c to - hold the length of the character in bytes, plus UTF_LENGTH to flag that - it's a length rather than a small character. */ - -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 - if (utf && NOT_FIRSTCHAR(code[-1])) - { - pcre_uchar *lastchar = code - 1; - BACKCHAR(lastchar); - c = (int)(code - lastchar); /* Length of UTF-8 character */ - memcpy(utf_chars, lastchar, IN_UCHARS(c)); /* Save the char */ - c |= UTF_LENGTH; /* Flag c as a length */ - } - else -#endif /* SUPPORT_UTF */ - - /* Handle the case of a single charater - either with no UTF support, or - with UTF disabled, or for a single character UTF character. */ - { - c = code[-1]; - if (*previous <= OP_CHARI && repeat_min > 1) - { - reqchar = c; - reqcharflags = req_caseopt | cd->req_varyopt; - } - } - - goto OUTPUT_SINGLE_REPEAT; /* Code shared with single character types */ - } - - /* If previous was a character type match (\d or similar), abolish it and - create a suitable repeat item. The code is shared with single-character - repeats by setting op_type to add a suitable offset into repeat_type. Note - the the Unicode property types will be present only when SUPPORT_UCP is - defined, but we don't wrap the little bits of code here because it just - makes it horribly messy. */ - - else if (*previous < OP_EODN) - { - pcre_uchar *oldcode; - int prop_type, prop_value; - op_type = OP_TYPESTAR - OP_STAR; /* Use type opcodes */ - c = *previous; - - OUTPUT_SINGLE_REPEAT: - if (*previous == OP_PROP || *previous == OP_NOTPROP) - { - prop_type = previous[1]; - prop_value = previous[2]; - } - else prop_type = prop_value = -1; - - oldcode = code; - code = previous; /* Usually overwrite previous item */ - - /* If the maximum is zero then the minimum must also be zero; Perl allows - this case, so we do too - by simply omitting the item altogether. */ - - if (repeat_max == 0) goto END_REPEAT; - - /* Combine the op_type with the repeat_type */ - - repeat_type += op_type; - - /* A minimum of zero is handled either as the special case * or ?, or as - an UPTO, with the maximum given. */ - - if (repeat_min == 0) - { - if (repeat_max == -1) *code++ = OP_STAR + repeat_type; - else if (repeat_max == 1) *code++ = OP_QUERY + repeat_type; - else - { - *code++ = OP_UPTO + repeat_type; - PUT2INC(code, 0, repeat_max); - } - } - - /* A repeat minimum of 1 is optimized into some special cases. If the - maximum is unlimited, we use OP_PLUS. Otherwise, the original item is - left in place and, if the maximum is greater than 1, we use OP_UPTO with - one less than the maximum. */ - - else if (repeat_min == 1) - { - if (repeat_max == -1) - *code++ = OP_PLUS + repeat_type; - else - { - code = oldcode; /* leave previous item in place */ - if (repeat_max == 1) goto END_REPEAT; - *code++ = OP_UPTO + repeat_type; - PUT2INC(code, 0, repeat_max - 1); - } - } - - /* The case {n,n} is just an EXACT, while the general case {n,m} is - handled as an EXACT followed by an UPTO. */ - - else - { - *code++ = OP_EXACT + op_type; /* NB EXACT doesn't have repeat_type */ - PUT2INC(code, 0, repeat_min); - - /* If the maximum is unlimited, insert an OP_STAR. Before doing so, - we have to insert the character for the previous code. For a repeated - Unicode property match, there are two extra bytes that define the - required property. In UTF-8 mode, long characters have their length in - c, with the UTF_LENGTH bit as a flag. */ - - if (repeat_max < 0) - { -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 - if (utf && (c & UTF_LENGTH) != 0) - { - memcpy(code, utf_chars, IN_UCHARS(c & 7)); - code += c & 7; - } - else -#endif - { - *code++ = c; - if (prop_type >= 0) - { - *code++ = prop_type; - *code++ = prop_value; - } - } - *code++ = OP_STAR + repeat_type; - } - - /* Else insert an UPTO if the max is greater than the min, again - preceded by the character, for the previously inserted code. If the - UPTO is just for 1 instance, we can use QUERY instead. */ - - else if (repeat_max != repeat_min) - { -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 - if (utf && (c & UTF_LENGTH) != 0) - { - memcpy(code, utf_chars, IN_UCHARS(c & 7)); - code += c & 7; - } - else -#endif - *code++ = c; - if (prop_type >= 0) - { - *code++ = prop_type; - *code++ = prop_value; - } - repeat_max -= repeat_min; - - if (repeat_max == 1) - { - *code++ = OP_QUERY + repeat_type; - } - else - { - *code++ = OP_UPTO + repeat_type; - PUT2INC(code, 0, repeat_max); - } - } - } - - /* The character or character type itself comes last in all cases. */ - -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 - if (utf && (c & UTF_LENGTH) != 0) - { - memcpy(code, utf_chars, IN_UCHARS(c & 7)); - code += c & 7; - } - else -#endif - *code++ = c; - - /* For a repeated Unicode property match, there are two extra bytes that - define the required property. */ - -#ifdef SUPPORT_UCP - if (prop_type >= 0) - { - *code++ = prop_type; - *code++ = prop_value; - } -#endif - } - - /* If previous was a character class or a back reference, we put the repeat - stuff after it, but just skip the item if the repeat was {0,0}. */ - - else if (*previous == OP_CLASS || *previous == OP_NCLASS || -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - *previous == OP_XCLASS || -#endif - *previous == OP_REF || *previous == OP_REFI || - *previous == OP_DNREF || *previous == OP_DNREFI) - { - if (repeat_max == 0) - { - code = previous; - goto END_REPEAT; - } - - if (repeat_min == 0 && repeat_max == -1) - *code++ = OP_CRSTAR + repeat_type; - else if (repeat_min == 1 && repeat_max == -1) - *code++ = OP_CRPLUS + repeat_type; - else if (repeat_min == 0 && repeat_max == 1) - *code++ = OP_CRQUERY + repeat_type; - else - { - *code++ = OP_CRRANGE + repeat_type; - PUT2INC(code, 0, repeat_min); - if (repeat_max == -1) repeat_max = 0; /* 2-byte encoding for max */ - PUT2INC(code, 0, repeat_max); - } - } - - /* If previous was a bracket group, we may have to replicate it in certain - cases. Note that at this point we can encounter only the "basic" bracket - opcodes such as BRA and CBRA, as this is the place where they get converted - into the more special varieties such as BRAPOS and SBRA. A test for >= - OP_ASSERT and <= OP_COND includes ASSERT, ASSERT_NOT, ASSERTBACK, - ASSERTBACK_NOT, ONCE, ONCE_NC, BRA, BRAPOS, CBRA, CBRAPOS, and COND. - Originally, PCRE did not allow repetition of assertions, but now it does, - for Perl compatibility. */ - - else if (*previous >= OP_ASSERT && *previous <= OP_COND) - { - register int i; - int len = (int)(code - previous); - pcre_uchar *bralink = NULL; - pcre_uchar *brazeroptr = NULL; - - /* Repeating a DEFINE group is pointless, but Perl allows the syntax, so - we just ignore the repeat. */ - - if (*previous == OP_COND && previous[LINK_SIZE+1] == OP_DEF) - goto END_REPEAT; - - /* There is no sense in actually repeating assertions. The only potential - use of repetition is in cases when the assertion is optional. Therefore, - if the minimum is greater than zero, just ignore the repeat. If the - maximum is not zero or one, set it to 1. */ - - if (*previous < OP_ONCE) /* Assertion */ - { - if (repeat_min > 0) goto END_REPEAT; - if (repeat_max < 0 || repeat_max > 1) repeat_max = 1; - } - - /* The case of a zero minimum is special because of the need to stick - OP_BRAZERO in front of it, and because the group appears once in the - data, whereas in other cases it appears the minimum number of times. For - this reason, it is simplest to treat this case separately, as otherwise - the code gets far too messy. There are several special subcases when the - minimum is zero. */ - - if (repeat_min == 0) - { - /* If the maximum is also zero, we used to just omit the group from the - output altogether, like this: - - ** if (repeat_max == 0) - ** { - ** code = previous; - ** goto END_REPEAT; - ** } - - However, that fails when a group or a subgroup within it is referenced - as a subroutine from elsewhere in the pattern, so now we stick in - OP_SKIPZERO in front of it so that it is skipped on execution. As we - don't have a list of which groups are referenced, we cannot do this - selectively. - - If the maximum is 1 or unlimited, we just have to stick in the BRAZERO - and do no more at this point. However, we do need to adjust any - OP_RECURSE calls inside the group that refer to the group itself or any - internal or forward referenced group, because the offset is from the - start of the whole regex. Temporarily terminate the pattern while doing - this. */ - - if (repeat_max <= 1) /* Covers 0, 1, and unlimited */ - { - *code = OP_END; - adjust_recurse(previous, 1, utf, cd, save_hwm); - memmove(previous + 1, previous, IN_UCHARS(len)); - code++; - if (repeat_max == 0) - { - *previous++ = OP_SKIPZERO; - goto END_REPEAT; - } - brazeroptr = previous; /* Save for possessive optimizing */ - *previous++ = OP_BRAZERO + repeat_type; - } - - /* If the maximum is greater than 1 and limited, we have to replicate - in a nested fashion, sticking OP_BRAZERO before each set of brackets. - The first one has to be handled carefully because it's the original - copy, which has to be moved up. The remainder can be handled by code - that is common with the non-zero minimum case below. We have to - adjust the value or repeat_max, since one less copy is required. Once - again, we may have to adjust any OP_RECURSE calls inside the group. */ - - else - { - int offset; - *code = OP_END; - adjust_recurse(previous, 2 + LINK_SIZE, utf, cd, save_hwm); - memmove(previous + 2 + LINK_SIZE, previous, IN_UCHARS(len)); - code += 2 + LINK_SIZE; - *previous++ = OP_BRAZERO + repeat_type; - *previous++ = OP_BRA; - - /* We chain together the bracket offset fields that have to be - filled in later when the ends of the brackets are reached. */ - - offset = (bralink == NULL)? 0 : (int)(previous - bralink); - bralink = previous; - PUTINC(previous, 0, offset); - } - - repeat_max--; - } - - /* If the minimum is greater than zero, replicate the group as many - times as necessary, and adjust the maximum to the number of subsequent - copies that we need. If we set a first char from the group, and didn't - set a required char, copy the latter from the former. If there are any - forward reference subroutine calls in the group, there will be entries on - the workspace list; replicate these with an appropriate increment. */ - - else - { - if (repeat_min > 1) - { - /* In the pre-compile phase, we don't actually do the replication. We - just adjust the length as if we had. Do some paranoid checks for - potential integer overflow. The INT64_OR_DOUBLE type is a 64-bit - integer type when available, otherwise double. */ - - if (lengthptr != NULL) - { - int delta = (repeat_min - 1)*length_prevgroup; - if ((INT64_OR_DOUBLE)(repeat_min - 1)* - (INT64_OR_DOUBLE)length_prevgroup > - (INT64_OR_DOUBLE)INT_MAX || - OFLOW_MAX - *lengthptr < delta) - { - *errorcodeptr = ERR20; - goto FAILED; - } - *lengthptr += delta; - } - - /* This is compiling for real. If there is a set first byte for - the group, and we have not yet set a "required byte", set it. Make - sure there is enough workspace for copying forward references before - doing the copy. */ - - else - { - if (groupsetfirstchar && reqcharflags < 0) - { - reqchar = firstchar; - reqcharflags = firstcharflags; - } - - for (i = 1; i < repeat_min; i++) - { - pcre_uchar *hc; - pcre_uchar *this_hwm = cd->hwm; - memcpy(code, previous, IN_UCHARS(len)); - - while (cd->hwm > cd->start_workspace + cd->workspace_size - - WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm)) - { - int save_offset = save_hwm - cd->start_workspace; - int this_offset = this_hwm - cd->start_workspace; - *errorcodeptr = expand_workspace(cd); - if (*errorcodeptr != 0) goto FAILED; - save_hwm = (pcre_uchar *)cd->start_workspace + save_offset; - this_hwm = (pcre_uchar *)cd->start_workspace + this_offset; - } - - for (hc = save_hwm; hc < this_hwm; hc += LINK_SIZE) - { - PUT(cd->hwm, 0, GET(hc, 0) + len); - cd->hwm += LINK_SIZE; - } - save_hwm = this_hwm; - code += len; - } - } - } - - if (repeat_max > 0) repeat_max -= repeat_min; - } - - /* This code is common to both the zero and non-zero minimum cases. If - the maximum is limited, it replicates the group in a nested fashion, - remembering the bracket starts on a stack. In the case of a zero minimum, - the first one was set up above. In all cases the repeat_max now specifies - the number of additional copies needed. Again, we must remember to - replicate entries on the forward reference list. */ - - if (repeat_max >= 0) - { - /* In the pre-compile phase, we don't actually do the replication. We - just adjust the length as if we had. For each repetition we must add 1 - to the length for BRAZERO and for all but the last repetition we must - add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some - paranoid checks to avoid integer overflow. The INT64_OR_DOUBLE type is - a 64-bit integer type when available, otherwise double. */ - - if (lengthptr != NULL && repeat_max > 0) - { - int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) - - 2 - 2*LINK_SIZE; /* Last one doesn't nest */ - if ((INT64_OR_DOUBLE)repeat_max * - (INT64_OR_DOUBLE)(length_prevgroup + 1 + 2 + 2*LINK_SIZE) - > (INT64_OR_DOUBLE)INT_MAX || - OFLOW_MAX - *lengthptr < delta) - { - *errorcodeptr = ERR20; - goto FAILED; - } - *lengthptr += delta; - } - - /* This is compiling for real */ - - else for (i = repeat_max - 1; i >= 0; i--) - { - pcre_uchar *hc; - pcre_uchar *this_hwm = cd->hwm; - - *code++ = OP_BRAZERO + repeat_type; - - /* All but the final copy start a new nesting, maintaining the - chain of brackets outstanding. */ - - if (i != 0) - { - int offset; - *code++ = OP_BRA; - offset = (bralink == NULL)? 0 : (int)(code - bralink); - bralink = code; - PUTINC(code, 0, offset); - } - - memcpy(code, previous, IN_UCHARS(len)); - - /* Ensure there is enough workspace for forward references before - copying them. */ - - while (cd->hwm > cd->start_workspace + cd->workspace_size - - WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm)) - { - int save_offset = save_hwm - cd->start_workspace; - int this_offset = this_hwm - cd->start_workspace; - *errorcodeptr = expand_workspace(cd); - if (*errorcodeptr != 0) goto FAILED; - save_hwm = (pcre_uchar *)cd->start_workspace + save_offset; - this_hwm = (pcre_uchar *)cd->start_workspace + this_offset; - } - - for (hc = save_hwm; hc < this_hwm; hc += LINK_SIZE) - { - PUT(cd->hwm, 0, GET(hc, 0) + len + ((i != 0)? 2+LINK_SIZE : 1)); - cd->hwm += LINK_SIZE; - } - save_hwm = this_hwm; - code += len; - } - - /* Now chain through the pending brackets, and fill in their length - fields (which are holding the chain links pro tem). */ - - while (bralink != NULL) - { - int oldlinkoffset; - int offset = (int)(code - bralink + 1); - pcre_uchar *bra = code - offset; - oldlinkoffset = GET(bra, 1); - bralink = (oldlinkoffset == 0)? NULL : bralink - oldlinkoffset; - *code++ = OP_KET; - PUTINC(code, 0, offset); - PUT(bra, 1, offset); - } - } - - /* If the maximum is unlimited, set a repeater in the final copy. For - ONCE brackets, that's all we need to do. However, possessively repeated - ONCE brackets can be converted into non-capturing brackets, as the - behaviour of (?:xx)++ is the same as (?>xx)++ and this saves having to - deal with possessive ONCEs specially. - - Otherwise, when we are doing the actual compile phase, check to see - whether this group is one that could match an empty string. If so, - convert the initial operator to the S form (e.g. OP_BRA -> OP_SBRA) so - that runtime checking can be done. [This check is also applied to ONCE - groups at runtime, but in a different way.] - - Then, if the quantifier was possessive and the bracket is not a - conditional, we convert the BRA code to the POS form, and the KET code to - KETRPOS. (It turns out to be convenient at runtime to detect this kind of - subpattern at both the start and at the end.) The use of special opcodes - makes it possible to reduce greatly the stack usage in pcre_exec(). If - the group is preceded by OP_BRAZERO, convert this to OP_BRAPOSZERO. - - Then, if the minimum number of matches is 1 or 0, cancel the possessive - flag so that the default action below, of wrapping everything inside - atomic brackets, does not happen. When the minimum is greater than 1, - there will be earlier copies of the group, and so we still have to wrap - the whole thing. */ - - else - { - pcre_uchar *ketcode = code - 1 - LINK_SIZE; - pcre_uchar *bracode = ketcode - GET(ketcode, 1); - - /* Convert possessive ONCE brackets to non-capturing */ - - if ((*bracode == OP_ONCE || *bracode == OP_ONCE_NC) && - possessive_quantifier) *bracode = OP_BRA; - - /* For non-possessive ONCE brackets, all we need to do is to - set the KET. */ - - if (*bracode == OP_ONCE || *bracode == OP_ONCE_NC) - *ketcode = OP_KETRMAX + repeat_type; - - /* Handle non-ONCE brackets and possessive ONCEs (which have been - converted to non-capturing above). */ - - else - { - /* In the compile phase, check for empty string matching. */ - - if (lengthptr == NULL) - { - pcre_uchar *scode = bracode; - do - { - if (could_be_empty_branch(scode, ketcode, utf, cd, NULL)) - { - *bracode += OP_SBRA - OP_BRA; - break; - } - scode += GET(scode, 1); - } - while (*scode == OP_ALT); - } - - /* Handle possessive quantifiers. */ - - if (possessive_quantifier) - { - /* For COND brackets, we wrap the whole thing in a possessively - repeated non-capturing bracket, because we have not invented POS - versions of the COND opcodes. Because we are moving code along, we - must ensure that any pending recursive references are updated. */ - - if (*bracode == OP_COND || *bracode == OP_SCOND) - { - int nlen = (int)(code - bracode); - *code = OP_END; - adjust_recurse(bracode, 1 + LINK_SIZE, utf, cd, save_hwm); - memmove(bracode + 1 + LINK_SIZE, bracode, IN_UCHARS(nlen)); - code += 1 + LINK_SIZE; - nlen += 1 + LINK_SIZE; - *bracode = OP_BRAPOS; - *code++ = OP_KETRPOS; - PUTINC(code, 0, nlen); - PUT(bracode, 1, nlen); - } - - /* For non-COND brackets, we modify the BRA code and use KETRPOS. */ - - else - { - *bracode += 1; /* Switch to xxxPOS opcodes */ - *ketcode = OP_KETRPOS; - } - - /* If the minimum is zero, mark it as possessive, then unset the - possessive flag when the minimum is 0 or 1. */ - - if (brazeroptr != NULL) *brazeroptr = OP_BRAPOSZERO; - if (repeat_min < 2) possessive_quantifier = FALSE; - } - - /* Non-possessive quantifier */ - - else *ketcode = OP_KETRMAX + repeat_type; - } - } - } - - /* If previous is OP_FAIL, it was generated by an empty class [] in - JavaScript mode. The other ways in which OP_FAIL can be generated, that is - by (*FAIL) or (?!) set previous to NULL, which gives a "nothing to repeat" - error above. We can just ignore the repeat in JS case. */ - - else if (*previous == OP_FAIL) goto END_REPEAT; - - /* Else there's some kind of shambles */ - - else - { - *errorcodeptr = ERR11; - goto FAILED; - } - - /* If the character following a repeat is '+', possessive_quantifier is - TRUE. For some opcodes, there are special alternative opcodes for this - case. For anything else, we wrap the entire repeated item inside OP_ONCE - brackets. Logically, the '+' notation is just syntactic sugar, taken from - Sun's Java package, but the special opcodes can optimize it. - - Some (but not all) possessively repeated subpatterns have already been - completely handled in the code just above. For them, possessive_quantifier - is always FALSE at this stage. Note that the repeated item starts at - tempcode, not at previous, which might be the first part of a string whose - (former) last char we repeated. */ - - if (possessive_quantifier) - { - int len; - - /* Possessifying an EXACT quantifier has no effect, so we can ignore it. - However, QUERY, STAR, or UPTO may follow (for quantifiers such as {5,6}, - {5,}, or {5,10}). We skip over an EXACT item; if the length of what - remains is greater than zero, there's a further opcode that can be - handled. If not, do nothing, leaving the EXACT alone. */ - - switch(*tempcode) - { - case OP_TYPEEXACT: - tempcode += PRIV(OP_lengths)[*tempcode] + - ((tempcode[1 + IMM2_SIZE] == OP_PROP - || tempcode[1 + IMM2_SIZE] == OP_NOTPROP)? 2 : 0); - break; - - /* CHAR opcodes are used for exacts whose count is 1. */ - - case OP_CHAR: - case OP_CHARI: - case OP_NOT: - case OP_NOTI: - case OP_EXACT: - case OP_EXACTI: - case OP_NOTEXACT: - case OP_NOTEXACTI: - tempcode += PRIV(OP_lengths)[*tempcode]; -#ifdef SUPPORT_UTF - if (utf && HAS_EXTRALEN(tempcode[-1])) - tempcode += GET_EXTRALEN(tempcode[-1]); -#endif - break; - - /* For the class opcodes, the repeat operator appears at the end; - adjust tempcode to point to it. */ - - case OP_CLASS: - case OP_NCLASS: - tempcode += 1 + 32/sizeof(pcre_uchar); - break; - -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - case OP_XCLASS: - tempcode += GET(tempcode, 1); - break; -#endif - } - - /* If tempcode is equal to code (which points to the end of the repeated - item), it means we have skipped an EXACT item but there is no following - QUERY, STAR, or UPTO; the value of len will be 0, and we do nothing. In - all other cases, tempcode will be pointing to the repeat opcode, and will - be less than code, so the value of len will be greater than 0. */ - - len = (int)(code - tempcode); - if (len > 0) - { - unsigned int repcode = *tempcode; - - /* There is a table for possessifying opcodes, all of which are less - than OP_CALLOUT. A zero entry means there is no possessified version. - */ - - if (repcode < OP_CALLOUT && opcode_possessify[repcode] > 0) - *tempcode = opcode_possessify[repcode]; - - /* For opcode without a special possessified version, wrap the item in - ONCE brackets. Because we are moving code along, we must ensure that any - pending recursive references are updated. */ - - else - { - *code = OP_END; - adjust_recurse(tempcode, 1 + LINK_SIZE, utf, cd, save_hwm); - memmove(tempcode + 1 + LINK_SIZE, tempcode, IN_UCHARS(len)); - code += 1 + LINK_SIZE; - len += 1 + LINK_SIZE; - tempcode[0] = OP_ONCE; - *code++ = OP_KET; - PUTINC(code, 0, len); - PUT(tempcode, 1, len); - } - } - -#ifdef NEVER - if (len > 0) switch (*tempcode) - { - case OP_STAR: *tempcode = OP_POSSTAR; break; - case OP_PLUS: *tempcode = OP_POSPLUS; break; - case OP_QUERY: *tempcode = OP_POSQUERY; break; - case OP_UPTO: *tempcode = OP_POSUPTO; break; - - case OP_STARI: *tempcode = OP_POSSTARI; break; - case OP_PLUSI: *tempcode = OP_POSPLUSI; break; - case OP_QUERYI: *tempcode = OP_POSQUERYI; break; - case OP_UPTOI: *tempcode = OP_POSUPTOI; break; - - case OP_NOTSTAR: *tempcode = OP_NOTPOSSTAR; break; - case OP_NOTPLUS: *tempcode = OP_NOTPOSPLUS; break; - case OP_NOTQUERY: *tempcode = OP_NOTPOSQUERY; break; - case OP_NOTUPTO: *tempcode = OP_NOTPOSUPTO; break; - - case OP_NOTSTARI: *tempcode = OP_NOTPOSSTARI; break; - case OP_NOTPLUSI: *tempcode = OP_NOTPOSPLUSI; break; - case OP_NOTQUERYI: *tempcode = OP_NOTPOSQUERYI; break; - case OP_NOTUPTOI: *tempcode = OP_NOTPOSUPTOI; break; - - case OP_TYPESTAR: *tempcode = OP_TYPEPOSSTAR; break; - case OP_TYPEPLUS: *tempcode = OP_TYPEPOSPLUS; break; - case OP_TYPEQUERY: *tempcode = OP_TYPEPOSQUERY; break; - case OP_TYPEUPTO: *tempcode = OP_TYPEPOSUPTO; break; - - case OP_CRSTAR: *tempcode = OP_CRPOSSTAR; break; - case OP_CRPLUS: *tempcode = OP_CRPOSPLUS; break; - case OP_CRQUERY: *tempcode = OP_CRPOSQUERY; break; - case OP_CRRANGE: *tempcode = OP_CRPOSRANGE; break; - - /* Because we are moving code along, we must ensure that any - pending recursive references are updated. */ - - default: - *code = OP_END; - adjust_recurse(tempcode, 1 + LINK_SIZE, utf, cd, save_hwm); - memmove(tempcode + 1 + LINK_SIZE, tempcode, IN_UCHARS(len)); - code += 1 + LINK_SIZE; - len += 1 + LINK_SIZE; - tempcode[0] = OP_ONCE; - *code++ = OP_KET; - PUTINC(code, 0, len); - PUT(tempcode, 1, len); - break; - } -#endif - } - - /* In all case we no longer have a previous item. We also set the - "follows varying string" flag for subsequently encountered reqchars if - it isn't already set and we have just passed a varying length item. */ - - END_REPEAT: - previous = NULL; - cd->req_varyopt |= reqvary; - break; - - - /* ===================================================================*/ - /* Start of nested parenthesized sub-expression, or comment or lookahead or - lookbehind or option setting or condition or all the other extended - parenthesis forms. */ - - case CHAR_LEFT_PARENTHESIS: - newoptions = options; - skipbytes = 0; - bravalue = OP_CBRA; - save_hwm = cd->hwm; - reset_bracount = FALSE; - - /* First deal with various "verbs" that can be introduced by '*'. */ - - ptr++; - if (ptr[0] == CHAR_ASTERISK && (ptr[1] == ':' - || (MAX_255(ptr[1]) && ((cd->ctypes[ptr[1]] & ctype_letter) != 0)))) - { - int i, namelen; - int arglen = 0; - const char *vn = verbnames; - const pcre_uchar *name = ptr + 1; - const pcre_uchar *arg = NULL; - previous = NULL; - ptr++; - while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_letter) != 0) ptr++; - namelen = (int)(ptr - name); - - /* It appears that Perl allows any characters whatsoever, other than - a closing parenthesis, to appear in arguments, so we no longer insist on - letters, digits, and underscores. */ - - if (*ptr == CHAR_COLON) - { - arg = ++ptr; - while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++; - arglen = (int)(ptr - arg); - if ((unsigned int)arglen > MAX_MARK) - { - *errorcodeptr = ERR75; - goto FAILED; - } - } - - if (*ptr != CHAR_RIGHT_PARENTHESIS) - { - *errorcodeptr = ERR60; - goto FAILED; - } - - /* Scan the table of verb names */ - - for (i = 0; i < verbcount; i++) - { - if (namelen == verbs[i].len && - STRNCMP_UC_C8(name, vn, namelen) == 0) - { - int setverb; - - /* Check for open captures before ACCEPT and convert it to - ASSERT_ACCEPT if in an assertion. */ - - if (verbs[i].op == OP_ACCEPT) - { - open_capitem *oc; - if (arglen != 0) - { - *errorcodeptr = ERR59; - goto FAILED; - } - cd->had_accept = TRUE; - for (oc = cd->open_caps; oc != NULL; oc = oc->next) - { - *code++ = OP_CLOSE; - PUT2INC(code, 0, oc->number); - } - setverb = *code++ = - (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT; - - /* Do not set firstchar after *ACCEPT */ - if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE; - } - - /* Handle other cases with/without an argument */ - - else if (arglen == 0) - { - if (verbs[i].op < 0) /* Argument is mandatory */ - { - *errorcodeptr = ERR66; - goto FAILED; - } - setverb = *code++ = verbs[i].op; - } - - else - { - if (verbs[i].op_arg < 0) /* Argument is forbidden */ - { - *errorcodeptr = ERR59; - goto FAILED; - } - setverb = *code++ = verbs[i].op_arg; - *code++ = arglen; - memcpy(code, arg, IN_UCHARS(arglen)); - code += arglen; - *code++ = 0; - } - - switch (setverb) - { - case OP_THEN: - case OP_THEN_ARG: - cd->external_flags |= PCRE_HASTHEN; - break; - - case OP_PRUNE: - case OP_PRUNE_ARG: - case OP_SKIP: - case OP_SKIP_ARG: - cd->had_pruneorskip = TRUE; - break; - } - - break; /* Found verb, exit loop */ - } - - vn += verbs[i].len + 1; - } - - if (i < verbcount) continue; /* Successfully handled a verb */ - *errorcodeptr = ERR60; /* Verb not recognized */ - goto FAILED; - } - - /* Deal with the extended parentheses; all are introduced by '?', and the - appearance of any of them means that this is not a capturing group. */ - - else if (*ptr == CHAR_QUESTION_MARK) - { - int i, set, unset, namelen; - int *optset; - const pcre_uchar *name; - pcre_uchar *slot; - - switch (*(++ptr)) - { - case CHAR_NUMBER_SIGN: /* Comment; skip to ket */ - ptr++; - while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++; - if (*ptr == CHAR_NULL) - { - *errorcodeptr = ERR18; - goto FAILED; - } - continue; - - - /* ------------------------------------------------------------ */ - case CHAR_VERTICAL_LINE: /* Reset capture count for each branch */ - reset_bracount = TRUE; - /* Fall through */ - - /* ------------------------------------------------------------ */ - case CHAR_COLON: /* Non-capturing bracket */ - bravalue = OP_BRA; - ptr++; - break; - - - /* ------------------------------------------------------------ */ - case CHAR_LEFT_PARENTHESIS: - bravalue = OP_COND; /* Conditional group */ - tempptr = ptr; - - /* A condition can be an assertion, a number (referring to a numbered - group's having been set), a name (referring to a named group), or 'R', - referring to recursion. R and R&name are also permitted for - recursion tests. - - There are ways of testing a named group: (?(name)) is used by Python; - Perl 5.10 onwards uses (?() or (?('name')). - - There is one unfortunate ambiguity, caused by history. 'R' can be the - recursive thing or the name 'R' (and similarly for 'R' followed by - digits). We look for a name first; if not found, we try the other case. - - For compatibility with auto-callouts, we allow a callout to be - specified before a condition that is an assertion. First, check for the - syntax of a callout; if found, adjust the temporary pointer that is - used to check for an assertion condition. That's all that is needed! */ - - if (ptr[1] == CHAR_QUESTION_MARK && ptr[2] == CHAR_C) - { - for (i = 3;; i++) if (!IS_DIGIT(ptr[i])) break; - if (ptr[i] == CHAR_RIGHT_PARENTHESIS) - tempptr += i + 1; - } - - /* For conditions that are assertions, check the syntax, and then exit - the switch. This will take control down to where bracketed groups, - including assertions, are processed. */ - - if (tempptr[1] == CHAR_QUESTION_MARK && - (tempptr[2] == CHAR_EQUALS_SIGN || - tempptr[2] == CHAR_EXCLAMATION_MARK || - tempptr[2] == CHAR_LESS_THAN_SIGN)) - break; - - /* Other conditions use OP_CREF/OP_DNCREF/OP_RREF/OP_DNRREF, and all - need to skip at least 1+IMM2_SIZE bytes at the start of the group. */ - - code[1+LINK_SIZE] = OP_CREF; - skipbytes = 1+IMM2_SIZE; - refsign = -1; - - /* Check for a test for recursion in a named group. */ - - ptr++; - if (*ptr == CHAR_R && ptr[1] == CHAR_AMPERSAND) - { - terminator = -1; - ptr += 2; - code[1+LINK_SIZE] = OP_RREF; /* Change the type of test */ - } - - /* Check for a test for a named group's having been set, using the Perl - syntax (?() or (?('name'), and also allow for the original PCRE - syntax of (?(name) or for (?(+n), (?(-n), and just (?(n). */ - - else if (*ptr == CHAR_LESS_THAN_SIGN) - { - terminator = CHAR_GREATER_THAN_SIGN; - ptr++; - } - else if (*ptr == CHAR_APOSTROPHE) - { - terminator = CHAR_APOSTROPHE; - ptr++; - } - else - { - terminator = CHAR_NULL; - if (*ptr == CHAR_MINUS || *ptr == CHAR_PLUS) refsign = *ptr++; - else if (IS_DIGIT(*ptr)) refsign = 0; - } - - /* Handle a number */ - - if (refsign >= 0) - { - recno = 0; - while (IS_DIGIT(*ptr)) - { - recno = recno * 10 + (int)(*ptr - CHAR_0); - ptr++; - } - } - - /* Otherwise we expect to read a name; anything else is an error. When - a name is one of a number of duplicates, a different opcode is used and - it needs more memory. Unfortunately we cannot tell whether a name is a - duplicate in the first pass, so we have to allow for more memory. */ - - else - { - if (IS_DIGIT(*ptr)) - { - *errorcodeptr = ERR84; - goto FAILED; - } - if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_word) == 0) - { - *errorcodeptr = ERR28; /* Assertion expected */ - goto FAILED; - } - name = ptr++; - while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) - { - ptr++; - } - namelen = (int)(ptr - name); - if (lengthptr != NULL) *lengthptr += IMM2_SIZE; - } - - /* Check the terminator */ - - if ((terminator > 0 && *ptr++ != (pcre_uchar)terminator) || - *ptr++ != CHAR_RIGHT_PARENTHESIS) - { - ptr--; /* Error offset */ - *errorcodeptr = ERR26; /* Malformed number or name */ - goto FAILED; - } - - /* Do no further checking in the pre-compile phase. */ - - if (lengthptr != NULL) break; - - /* In the real compile we do the work of looking for the actual - reference. If refsign is not negative, it means we have a number in - recno. */ - - if (refsign >= 0) - { - if (recno <= 0) - { - *errorcodeptr = ERR35; - goto FAILED; - } - if (refsign != 0) recno = (refsign == CHAR_MINUS)? - cd->bracount - recno + 1 : recno + cd->bracount; - if (recno <= 0 || recno > cd->final_bracount) - { - *errorcodeptr = ERR15; - goto FAILED; - } - PUT2(code, 2+LINK_SIZE, recno); - break; - } - - /* Otherwise look for the name. */ - - slot = cd->name_table; - for (i = 0; i < cd->names_found; i++) - { - if (STRNCMP_UC_UC(name, slot+IMM2_SIZE, namelen) == 0) break; - slot += cd->name_entry_size; - } - - /* Found the named subpattern. If the name is duplicated, add one to - the opcode to change CREF/RREF into DNCREF/DNRREF and insert - appropriate data values. Otherwise, just insert the unique subpattern - number. */ - - if (i < cd->names_found) - { - int offset = i++; - int count = 1; - recno = GET2(slot, 0); /* Number from first found */ - for (; i < cd->names_found; i++) - { - slot += cd->name_entry_size; - if (STRNCMP_UC_UC(name, slot+IMM2_SIZE, namelen) != 0) break; - count++; - } - if (count > 1) - { - PUT2(code, 2+LINK_SIZE, offset); - PUT2(code, 2+LINK_SIZE+IMM2_SIZE, count); - skipbytes += IMM2_SIZE; - code[1+LINK_SIZE]++; - } - else /* Not a duplicated name */ - { - PUT2(code, 2+LINK_SIZE, recno); - } - } - - /* If terminator == CHAR_NULL it means that the name followed directly - after the opening parenthesis [e.g. (?(abc)...] and in this case there - are some further alternatives to try. For the cases where terminator != - CHAR_NULL [things like (?(... or (?('name')... or (?(R&name)... ] - we have now checked all the possibilities, so give an error. */ - - else if (terminator != CHAR_NULL) - { - *errorcodeptr = ERR15; - goto FAILED; - } - - /* Check for (?(R) for recursion. Allow digits after R to specify a - specific group number. */ - - else if (*name == CHAR_R) - { - recno = 0; - for (i = 1; i < namelen; i++) - { - if (!IS_DIGIT(name[i])) - { - *errorcodeptr = ERR15; - goto FAILED; - } - recno = recno * 10 + name[i] - CHAR_0; - } - if (recno == 0) recno = RREF_ANY; - code[1+LINK_SIZE] = OP_RREF; /* Change test type */ - PUT2(code, 2+LINK_SIZE, recno); - } - - /* Similarly, check for the (?(DEFINE) "condition", which is always - false. */ - - else if (namelen == 6 && STRNCMP_UC_C8(name, STRING_DEFINE, 6) == 0) - { - code[1+LINK_SIZE] = OP_DEF; - skipbytes = 1; - } - - /* Reference to an unidentified subpattern. */ - - else - { - *errorcodeptr = ERR15; - goto FAILED; - } - break; - - - /* ------------------------------------------------------------ */ - case CHAR_EQUALS_SIGN: /* Positive lookahead */ - bravalue = OP_ASSERT; - cd->assert_depth += 1; - ptr++; - break; - - /* Optimize (?!) to (*FAIL) unless it is quantified - which is a weird - thing to do, but Perl allows all assertions to be quantified, and when - they contain capturing parentheses there may be a potential use for - this feature. Not that that applies to a quantified (?!) but we allow - it for uniformity. */ - - /* ------------------------------------------------------------ */ - case CHAR_EXCLAMATION_MARK: /* Negative lookahead */ - ptr++; - if (*ptr == CHAR_RIGHT_PARENTHESIS && ptr[1] != CHAR_ASTERISK && - ptr[1] != CHAR_PLUS && ptr[1] != CHAR_QUESTION_MARK && - (ptr[1] != CHAR_LEFT_CURLY_BRACKET || !is_counted_repeat(ptr+2))) - { - *code++ = OP_FAIL; - previous = NULL; - continue; - } - bravalue = OP_ASSERT_NOT; - cd->assert_depth += 1; - break; - - - /* ------------------------------------------------------------ */ - case CHAR_LESS_THAN_SIGN: /* Lookbehind or named define */ - switch (ptr[1]) - { - case CHAR_EQUALS_SIGN: /* Positive lookbehind */ - bravalue = OP_ASSERTBACK; - cd->assert_depth += 1; - ptr += 2; - break; - - case CHAR_EXCLAMATION_MARK: /* Negative lookbehind */ - bravalue = OP_ASSERTBACK_NOT; - cd->assert_depth += 1; - ptr += 2; - break; - - default: /* Could be name define, else bad */ - if (MAX_255(ptr[1]) && (cd->ctypes[ptr[1]] & ctype_word) != 0) - goto DEFINE_NAME; - ptr++; /* Correct offset for error */ - *errorcodeptr = ERR24; - goto FAILED; - } - break; - - - /* ------------------------------------------------------------ */ - case CHAR_GREATER_THAN_SIGN: /* One-time brackets */ - bravalue = OP_ONCE; - ptr++; - break; - - - /* ------------------------------------------------------------ */ - case CHAR_C: /* Callout - may be followed by digits; */ - previous_callout = code; /* Save for later completion */ - after_manual_callout = 1; /* Skip one item before completing */ - *code++ = OP_CALLOUT; - { - int n = 0; - ptr++; - while(IS_DIGIT(*ptr)) - n = n * 10 + *ptr++ - CHAR_0; - if (*ptr != CHAR_RIGHT_PARENTHESIS) - { - *errorcodeptr = ERR39; - goto FAILED; - } - if (n > 255) - { - *errorcodeptr = ERR38; - goto FAILED; - } - *code++ = n; - PUT(code, 0, (int)(ptr - cd->start_pattern + 1)); /* Pattern offset */ - PUT(code, LINK_SIZE, 0); /* Default length */ - code += 2 * LINK_SIZE; - } - previous = NULL; - continue; - - - /* ------------------------------------------------------------ */ - case CHAR_P: /* Python-style named subpattern handling */ - if (*(++ptr) == CHAR_EQUALS_SIGN || - *ptr == CHAR_GREATER_THAN_SIGN) /* Reference or recursion */ - { - is_recurse = *ptr == CHAR_GREATER_THAN_SIGN; - terminator = CHAR_RIGHT_PARENTHESIS; - goto NAMED_REF_OR_RECURSE; - } - else if (*ptr != CHAR_LESS_THAN_SIGN) /* Test for Python-style defn */ - { - *errorcodeptr = ERR41; - goto FAILED; - } - /* Fall through to handle (?P< as (?< is handled */ - - - /* ------------------------------------------------------------ */ - DEFINE_NAME: /* Come here from (?< handling */ - case CHAR_APOSTROPHE: - terminator = (*ptr == CHAR_LESS_THAN_SIGN)? - CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE; - name = ++ptr; - if (IS_DIGIT(*ptr)) - { - *errorcodeptr = ERR84; /* Group name must start with non-digit */ - goto FAILED; - } - while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) ptr++; - namelen = (int)(ptr - name); - - /* In the pre-compile phase, do a syntax check, remember the longest - name, and then remember the group in a vector, expanding it if - necessary. Duplicates for the same number are skipped; other duplicates - are checked for validity. In the actual compile, there is nothing to - do. */ - - if (lengthptr != NULL) - { - named_group *ng; - pcre_uint32 number = cd->bracount + 1; - - if (*ptr != (pcre_uchar)terminator) - { - *errorcodeptr = ERR42; - goto FAILED; - } - - if (cd->names_found >= MAX_NAME_COUNT) - { - *errorcodeptr = ERR49; - goto FAILED; - } - - if (namelen + IMM2_SIZE + 1 > cd->name_entry_size) - { - cd->name_entry_size = namelen + IMM2_SIZE + 1; - if (namelen > MAX_NAME_SIZE) - { - *errorcodeptr = ERR48; - goto FAILED; - } - } - - /* Scan the list to check for duplicates. For duplicate names, if the - number is the same, break the loop, which causes the name to be - discarded; otherwise, if DUPNAMES is not set, give an error. - If it is set, allow the name with a different number, but continue - scanning in case this is a duplicate with the same number. For - non-duplicate names, give an error if the number is duplicated. */ - - ng = cd->named_groups; - for (i = 0; i < cd->names_found; i++, ng++) - { - if (namelen == ng->length && - STRNCMP_UC_UC(name, ng->name, namelen) == 0) - { - if (ng->number == number) break; - if ((options & PCRE_DUPNAMES) == 0) - { - *errorcodeptr = ERR43; - goto FAILED; - } - cd->dupnames = TRUE; /* Duplicate names exist */ - } - else if (ng->number == number) - { - *errorcodeptr = ERR65; - goto FAILED; - } - } - - if (i >= cd->names_found) /* Not a duplicate with same number */ - { - /* Increase the list size if necessary */ - - if (cd->names_found >= cd->named_group_list_size) - { - int newsize = cd->named_group_list_size * 2; - named_group *newspace = (PUBL(malloc)) - (newsize * sizeof(named_group)); - - if (newspace == NULL) - { - *errorcodeptr = ERR21; - goto FAILED; - } - - memcpy(newspace, cd->named_groups, - cd->named_group_list_size * sizeof(named_group)); - if (cd->named_group_list_size > NAMED_GROUP_LIST_SIZE) - (PUBL(free))((void *)cd->named_groups); - cd->named_groups = newspace; - cd->named_group_list_size = newsize; - } - - cd->named_groups[cd->names_found].name = name; - cd->named_groups[cd->names_found].length = namelen; - cd->named_groups[cd->names_found].number = number; - cd->names_found++; - } - } - - ptr++; /* Move past > or ' in both passes. */ - goto NUMBERED_GROUP; - - - /* ------------------------------------------------------------ */ - case CHAR_AMPERSAND: /* Perl recursion/subroutine syntax */ - terminator = CHAR_RIGHT_PARENTHESIS; - is_recurse = TRUE; - /* Fall through */ - - /* We come here from the Python syntax above that handles both - references (?P=name) and recursion (?P>name), as well as falling - through from the Perl recursion syntax (?&name). We also come here from - the Perl \k or \k'name' back reference syntax and the \k{name} - .NET syntax, and the Oniguruma \g<...> and \g'...' subroutine syntax. */ - - NAMED_REF_OR_RECURSE: - name = ++ptr; - if (IS_DIGIT(*ptr)) - { - *errorcodeptr = ERR84; /* Group name must start with non-digit */ - goto FAILED; - } - while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) ptr++; - namelen = (int)(ptr - name); - - /* In the pre-compile phase, do a syntax check. We used to just set - a dummy reference number, because it was not used in the first pass. - However, with the change of recursive back references to be atomic, - we have to look for the number so that this state can be identified, as - otherwise the incorrect length is computed. If it's not a backwards - reference, the dummy number will do. */ - - if (lengthptr != NULL) - { - named_group *ng; - - if (namelen == 0) - { - *errorcodeptr = ERR62; - goto FAILED; - } - if (*ptr != (pcre_uchar)terminator) - { - *errorcodeptr = ERR42; - goto FAILED; - } - if (namelen > MAX_NAME_SIZE) - { - *errorcodeptr = ERR48; - goto FAILED; - } - - /* The name table does not exist in the first pass; instead we must - scan the list of names encountered so far in order to get the - number. If the name is not found, set the value to 0 for a forward - reference. */ - - ng = cd->named_groups; - for (i = 0; i < cd->names_found; i++, ng++) - { - if (namelen == ng->length && - STRNCMP_UC_UC(name, ng->name, namelen) == 0) - break; - } - recno = (i < cd->names_found)? ng->number : 0; - - /* Count named back references. */ - - if (!is_recurse) cd->namedrefcount++; - } - - /* In the real compile, search the name table. We check the name - first, and then check that we have reached the end of the name in the - table. That way, if the name is longer than any in the table, the - comparison will fail without reading beyond the table entry. */ - - else - { - slot = cd->name_table; - for (i = 0; i < cd->names_found; i++) - { - if (STRNCMP_UC_UC(name, slot+IMM2_SIZE, namelen) == 0 && - slot[IMM2_SIZE+namelen] == 0) - break; - slot += cd->name_entry_size; - } - - if (i < cd->names_found) - { - recno = GET2(slot, 0); - } - else - { - *errorcodeptr = ERR15; - goto FAILED; - } - } - - /* In both phases, for recursions, we can now go to the code than - handles numerical recursion. */ - - if (is_recurse) goto HANDLE_RECURSION; - - /* In the second pass we must see if the name is duplicated. If so, we - generate a different opcode. */ - - if (lengthptr == NULL && cd->dupnames) - { - int count = 1; - unsigned int index = i; - pcre_uchar *cslot = slot + cd->name_entry_size; - - for (i++; i < cd->names_found; i++) - { - if (STRCMP_UC_UC(slot + IMM2_SIZE, cslot + IMM2_SIZE) != 0) break; - count++; - cslot += cd->name_entry_size; - } - - if (count > 1) - { - if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE; - previous = code; - *code++ = ((options & PCRE_CASELESS) != 0)? OP_DNREFI : OP_DNREF; - PUT2INC(code, 0, index); - PUT2INC(code, 0, count); - - /* Process each potentially referenced group. */ - - for (; slot < cslot; slot += cd->name_entry_size) - { - open_capitem *oc; - recno = GET2(slot, 0); - cd->backref_map |= (recno < 32)? (1 << recno) : 1; - if (recno > cd->top_backref) cd->top_backref = recno; - - /* Check to see if this back reference is recursive, that it, it - is inside the group that it references. A flag is set so that the - group can be made atomic. */ - - for (oc = cd->open_caps; oc != NULL; oc = oc->next) - { - if (oc->number == recno) - { - oc->flag = TRUE; - break; - } - } - } - - continue; /* End of back ref handling */ - } - } - - /* First pass, or a non-duplicated name. */ - - goto HANDLE_REFERENCE; - - - /* ------------------------------------------------------------ */ - case CHAR_R: /* Recursion */ - ptr++; /* Same as (?0) */ - /* Fall through */ - - - /* ------------------------------------------------------------ */ - case CHAR_MINUS: case CHAR_PLUS: /* Recursion or subroutine */ - case CHAR_0: case CHAR_1: case CHAR_2: case CHAR_3: case CHAR_4: - case CHAR_5: case CHAR_6: case CHAR_7: case CHAR_8: case CHAR_9: - { - const pcre_uchar *called; - terminator = CHAR_RIGHT_PARENTHESIS; - - /* Come here from the \g<...> and \g'...' code (Oniguruma - compatibility). However, the syntax has been checked to ensure that - the ... are a (signed) number, so that neither ERR63 nor ERR29 will - be called on this path, nor with the jump to OTHER_CHAR_AFTER_QUERY - ever be taken. */ - - HANDLE_NUMERICAL_RECURSION: - - if ((refsign = *ptr) == CHAR_PLUS) - { - ptr++; - if (!IS_DIGIT(*ptr)) - { - *errorcodeptr = ERR63; - goto FAILED; - } - } - else if (refsign == CHAR_MINUS) - { - if (!IS_DIGIT(ptr[1])) - goto OTHER_CHAR_AFTER_QUERY; - ptr++; - } - - recno = 0; - while(IS_DIGIT(*ptr)) - recno = recno * 10 + *ptr++ - CHAR_0; - - if (*ptr != (pcre_uchar)terminator) - { - *errorcodeptr = ERR29; - goto FAILED; - } - - if (refsign == CHAR_MINUS) - { - if (recno == 0) - { - *errorcodeptr = ERR58; - goto FAILED; - } - recno = cd->bracount - recno + 1; - if (recno <= 0) - { - *errorcodeptr = ERR15; - goto FAILED; - } - } - else if (refsign == CHAR_PLUS) - { - if (recno == 0) - { - *errorcodeptr = ERR58; - goto FAILED; - } - recno += cd->bracount; - } - - /* Come here from code above that handles a named recursion */ - - HANDLE_RECURSION: - - previous = code; - called = cd->start_code; - - /* When we are actually compiling, find the bracket that is being - referenced. Temporarily end the regex in case it doesn't exist before - this point. If we end up with a forward reference, first check that - the bracket does occur later so we can give the error (and position) - now. Then remember this forward reference in the workspace so it can - be filled in at the end. */ - - if (lengthptr == NULL) - { - *code = OP_END; - if (recno != 0) - called = PRIV(find_bracket)(cd->start_code, utf, recno); - - /* Forward reference */ - - if (called == NULL) - { - if (recno > cd->final_bracount) - { - *errorcodeptr = ERR15; - goto FAILED; - } - - /* Fudge the value of "called" so that when it is inserted as an - offset below, what it actually inserted is the reference number - of the group. Then remember the forward reference. */ - - called = cd->start_code + recno; - if (cd->hwm >= cd->start_workspace + cd->workspace_size - - WORK_SIZE_SAFETY_MARGIN) - { - *errorcodeptr = expand_workspace(cd); - if (*errorcodeptr != 0) goto FAILED; - } - PUTINC(cd->hwm, 0, (int)(code + 1 - cd->start_code)); - } - - /* If not a forward reference, and the subpattern is still open, - this is a recursive call. We check to see if this is a left - recursion that could loop for ever, and diagnose that case. We - must not, however, do this check if we are in a conditional - subpattern because the condition might be testing for recursion in - a pattern such as /(?(R)a+|(?R)b)/, which is perfectly valid. - Forever loops are also detected at runtime, so those that occur in - conditional subpatterns will be picked up then. */ - - else if (GET(called, 1) == 0 && cond_depth <= 0 && - could_be_empty(called, code, bcptr, utf, cd)) - { - *errorcodeptr = ERR40; - goto FAILED; - } - } - - /* Insert the recursion/subroutine item. It does not have a set first - character (relevant if it is repeated, because it will then be - wrapped with ONCE brackets). */ - - *code = OP_RECURSE; - PUT(code, 1, (int)(called - cd->start_code)); - code += 1 + LINK_SIZE; - groupsetfirstchar = FALSE; - } - - /* Can't determine a first byte now */ - - if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE; - continue; - - - /* ------------------------------------------------------------ */ - default: /* Other characters: check option setting */ - OTHER_CHAR_AFTER_QUERY: - set = unset = 0; - optset = &set; - - while (*ptr != CHAR_RIGHT_PARENTHESIS && *ptr != CHAR_COLON) - { - switch (*ptr++) - { - case CHAR_MINUS: optset = &unset; break; - - case CHAR_J: /* Record that it changed in the external options */ - *optset |= PCRE_DUPNAMES; - cd->external_flags |= PCRE_JCHANGED; - break; - - case CHAR_i: *optset |= PCRE_CASELESS; break; - case CHAR_m: *optset |= PCRE_MULTILINE; break; - case CHAR_s: *optset |= PCRE_DOTALL; break; - case CHAR_x: *optset |= PCRE_EXTENDED; break; - case CHAR_U: *optset |= PCRE_UNGREEDY; break; - case CHAR_X: *optset |= PCRE_EXTRA; break; - - default: *errorcodeptr = ERR12; - ptr--; /* Correct the offset */ - goto FAILED; - } - } - - /* Set up the changed option bits, but don't change anything yet. */ - - newoptions = (options | set) & (~unset); - - /* If the options ended with ')' this is not the start of a nested - group with option changes, so the options change at this level. If this - item is right at the start of the pattern, the options can be - abstracted and made external in the pre-compile phase, and ignored in - the compile phase. This can be helpful when matching -- for instance in - caseless checking of required bytes. - - If the code pointer is not (cd->start_code + 1 + LINK_SIZE), we are - definitely *not* at the start of the pattern because something has been - compiled. In the pre-compile phase, however, the code pointer can have - that value after the start, because it gets reset as code is discarded - during the pre-compile. However, this can happen only at top level - if - we are within parentheses, the starting BRA will still be present. At - any parenthesis level, the length value can be used to test if anything - has been compiled at that level. Thus, a test for both these conditions - is necessary to ensure we correctly detect the start of the pattern in - both phases. - - If we are not at the pattern start, reset the greedy defaults and the - case value for firstchar and reqchar. */ - - if (*ptr == CHAR_RIGHT_PARENTHESIS) - { - if (code == cd->start_code + 1 + LINK_SIZE && - (lengthptr == NULL || *lengthptr == 2 + 2*LINK_SIZE)) - { - cd->external_options = newoptions; - } - else - { - greedy_default = ((newoptions & PCRE_UNGREEDY) != 0); - greedy_non_default = greedy_default ^ 1; - req_caseopt = ((newoptions & PCRE_CASELESS) != 0)? REQ_CASELESS:0; - } - - /* Change options at this level, and pass them back for use - in subsequent branches. */ - - *optionsptr = options = newoptions; - previous = NULL; /* This item can't be repeated */ - continue; /* It is complete */ - } - - /* If the options ended with ':' we are heading into a nested group - with possible change of options. Such groups are non-capturing and are - not assertions of any kind. All we need to do is skip over the ':'; - the newoptions value is handled below. */ - - bravalue = OP_BRA; - ptr++; - } /* End of switch for character following (? */ - } /* End of (? handling */ - - /* Opening parenthesis not followed by '*' or '?'. If PCRE_NO_AUTO_CAPTURE - is set, all unadorned brackets become non-capturing and behave like (?:...) - brackets. */ - - else if ((options & PCRE_NO_AUTO_CAPTURE) != 0) - { - bravalue = OP_BRA; - } - - /* Else we have a capturing group. */ - - else - { - NUMBERED_GROUP: - cd->bracount += 1; - PUT2(code, 1+LINK_SIZE, cd->bracount); - skipbytes = IMM2_SIZE; - } - - /* Process nested bracketed regex. First check for parentheses nested too - deeply. */ - - if ((cd->parens_depth += 1) > PARENS_NEST_LIMIT) - { - *errorcodeptr = ERR82; - goto FAILED; - } - - /* Assertions used not to be repeatable, but this was changed for Perl - compatibility, so all kinds can now be repeated. We copy code into a - non-register variable (tempcode) in order to be able to pass its address - because some compilers complain otherwise. */ - - previous = code; /* For handling repetition */ - *code = bravalue; - tempcode = code; - tempreqvary = cd->req_varyopt; /* Save value before bracket */ - tempbracount = cd->bracount; /* Save value before bracket */ - length_prevgroup = 0; /* Initialize for pre-compile phase */ - - if (!compile_regex( - newoptions, /* The complete new option state */ - &tempcode, /* Where to put code (updated) */ - &ptr, /* Input pointer (updated) */ - errorcodeptr, /* Where to put an error message */ - (bravalue == OP_ASSERTBACK || - bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */ - reset_bracount, /* True if (?| group */ - skipbytes, /* Skip over bracket number */ - cond_depth + - ((bravalue == OP_COND)?1:0), /* Depth of condition subpatterns */ - &subfirstchar, /* For possible first char */ - &subfirstcharflags, - &subreqchar, /* For possible last char */ - &subreqcharflags, - bcptr, /* Current branch chain */ - cd, /* Tables block */ - (lengthptr == NULL)? NULL : /* Actual compile phase */ - &length_prevgroup /* Pre-compile phase */ - )) - goto FAILED; - - cd->parens_depth -= 1; - - /* If this was an atomic group and there are no capturing groups within it, - generate OP_ONCE_NC instead of OP_ONCE. */ - - if (bravalue == OP_ONCE && cd->bracount <= tempbracount) - *code = OP_ONCE_NC; - - if (bravalue >= OP_ASSERT && bravalue <= OP_ASSERTBACK_NOT) - cd->assert_depth -= 1; - - /* At the end of compiling, code is still pointing to the start of the - group, while tempcode has been updated to point past the end of the group. - The pattern pointer (ptr) is on the bracket. - - If this is a conditional bracket, check that there are no more than - two branches in the group, or just one if it's a DEFINE group. We do this - in the real compile phase, not in the pre-pass, where the whole group may - not be available. */ - - if (bravalue == OP_COND && lengthptr == NULL) - { - pcre_uchar *tc = code; - int condcount = 0; - - do { - condcount++; - tc += GET(tc,1); - } - while (*tc != OP_KET); - - /* A DEFINE group is never obeyed inline (the "condition" is always - false). It must have only one branch. */ - - if (code[LINK_SIZE+1] == OP_DEF) - { - if (condcount > 1) - { - *errorcodeptr = ERR54; - goto FAILED; - } - bravalue = OP_DEF; /* Just a flag to suppress char handling below */ - } - - /* A "normal" conditional group. If there is just one branch, we must not - make use of its firstchar or reqchar, because this is equivalent to an - empty second branch. */ - - else - { - if (condcount > 2) - { - *errorcodeptr = ERR27; - goto FAILED; - } - if (condcount == 1) subfirstcharflags = subreqcharflags = REQ_NONE; - } - } - - /* Error if hit end of pattern */ - - if (*ptr != CHAR_RIGHT_PARENTHESIS) - { - *errorcodeptr = ERR14; - goto FAILED; - } - - /* In the pre-compile phase, update the length by the length of the group, - less the brackets at either end. Then reduce the compiled code to just a - set of non-capturing brackets so that it doesn't use much memory if it is - duplicated by a quantifier.*/ - - if (lengthptr != NULL) - { - if (OFLOW_MAX - *lengthptr < length_prevgroup - 2 - 2*LINK_SIZE) - { - *errorcodeptr = ERR20; - goto FAILED; - } - *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE; - code++; /* This already contains bravalue */ - PUTINC(code, 0, 1 + LINK_SIZE); - *code++ = OP_KET; - PUTINC(code, 0, 1 + LINK_SIZE); - break; /* No need to waste time with special character handling */ - } - - /* Otherwise update the main code pointer to the end of the group. */ - - code = tempcode; - - /* For a DEFINE group, required and first character settings are not - relevant. */ - - if (bravalue == OP_DEF) break; - - /* Handle updating of the required and first characters for other types of - group. Update for normal brackets of all kinds, and conditions with two - branches (see code above). If the bracket is followed by a quantifier with - zero repeat, we have to back off. Hence the definition of zeroreqchar and - zerofirstchar outside the main loop so that they can be accessed for the - back off. */ - - zeroreqchar = reqchar; - zeroreqcharflags = reqcharflags; - zerofirstchar = firstchar; - zerofirstcharflags = firstcharflags; - groupsetfirstchar = FALSE; - - if (bravalue >= OP_ONCE) - { - /* If we have not yet set a firstchar in this branch, take it from the - subpattern, remembering that it was set here so that a repeat of more - than one can replicate it as reqchar if necessary. If the subpattern has - no firstchar, set "none" for the whole branch. In both cases, a zero - repeat forces firstchar to "none". */ - - if (firstcharflags == REQ_UNSET) - { - if (subfirstcharflags >= 0) - { - firstchar = subfirstchar; - firstcharflags = subfirstcharflags; - groupsetfirstchar = TRUE; - } - else firstcharflags = REQ_NONE; - zerofirstcharflags = REQ_NONE; - } - - /* If firstchar was previously set, convert the subpattern's firstchar - into reqchar if there wasn't one, using the vary flag that was in - existence beforehand. */ - - else if (subfirstcharflags >= 0 && subreqcharflags < 0) - { - subreqchar = subfirstchar; - subreqcharflags = subfirstcharflags | tempreqvary; - } - - /* If the subpattern set a required byte (or set a first byte that isn't - really the first byte - see above), set it. */ - - if (subreqcharflags >= 0) - { - reqchar = subreqchar; - reqcharflags = subreqcharflags; - } - } - - /* For a forward assertion, we take the reqchar, if set. This can be - helpful if the pattern that follows the assertion doesn't set a different - char. For example, it's useful for /(?=abcde).+/. We can't set firstchar - for an assertion, however because it leads to incorrect effect for patterns - such as /(?=a)a.+/ when the "real" "a" would then become a reqchar instead - of a firstchar. This is overcome by a scan at the end if there's no - firstchar, looking for an asserted first char. */ - - else if (bravalue == OP_ASSERT && subreqcharflags >= 0) - { - reqchar = subreqchar; - reqcharflags = subreqcharflags; - } - break; /* End of processing '(' */ - - - /* ===================================================================*/ - /* Handle metasequences introduced by \. For ones like \d, the ESC_ values - are arranged to be the negation of the corresponding OP_values in the - default case when PCRE_UCP is not set. For the back references, the values - are negative the reference number. Only back references and those types - that consume a character may be repeated. We can test for values between - ESC_b and ESC_Z for the latter; this may have to change if any new ones are - ever created. */ - - case CHAR_BACKSLASH: - tempptr = ptr; - escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, FALSE); - if (*errorcodeptr != 0) goto FAILED; - - if (escape == 0) /* The escape coded a single character */ - c = ec; - else - { - if (escape == ESC_Q) /* Handle start of quoted string */ - { - if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E) - ptr += 2; /* avoid empty string */ - else inescq = TRUE; - continue; - } - - if (escape == ESC_E) continue; /* Perl ignores an orphan \E */ - - /* For metasequences that actually match a character, we disable the - setting of a first character if it hasn't already been set. */ - - if (firstcharflags == REQ_UNSET && escape > ESC_b && escape < ESC_Z) - firstcharflags = REQ_NONE; - - /* Set values to reset to if this is followed by a zero repeat. */ - - zerofirstchar = firstchar; - zerofirstcharflags = firstcharflags; - zeroreqchar = reqchar; - zeroreqcharflags = reqcharflags; - - /* \g or \g'name' is a subroutine call by name and \g or \g'n' - is a subroutine call by number (Oniguruma syntax). In fact, the value - ESC_g is returned only for these cases. So we don't need to check for < - or ' if the value is ESC_g. For the Perl syntax \g{n} the value is - -n, and for the Perl syntax \g{name} the result is ESC_k (as - that is a synonym for a named back reference). */ - - if (escape == ESC_g) - { - const pcre_uchar *p; - pcre_uint32 cf; - - save_hwm = cd->hwm; /* Normally this is set when '(' is read */ - terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)? - CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE; - - /* These two statements stop the compiler for warning about possibly - unset variables caused by the jump to HANDLE_NUMERICAL_RECURSION. In - fact, because we do the check for a number below, the paths that - would actually be in error are never taken. */ - - skipbytes = 0; - reset_bracount = FALSE; - - /* If it's not a signed or unsigned number, treat it as a name. */ - - cf = ptr[1]; - if (cf != CHAR_PLUS && cf != CHAR_MINUS && !IS_DIGIT(cf)) - { - is_recurse = TRUE; - goto NAMED_REF_OR_RECURSE; - } - - /* Signed or unsigned number (cf = ptr[1]) is known to be plus or minus - or a digit. */ - - p = ptr + 2; - while (IS_DIGIT(*p)) p++; - if (*p != (pcre_uchar)terminator) - { - *errorcodeptr = ERR57; - break; - } - ptr++; - goto HANDLE_NUMERICAL_RECURSION; - } - - /* \k or \k'name' is a back reference by name (Perl syntax). - We also support \k{name} (.NET syntax). */ - - if (escape == ESC_k) - { - if ((ptr[1] != CHAR_LESS_THAN_SIGN && - ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET)) - { - *errorcodeptr = ERR69; - break; - } - is_recurse = FALSE; - terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)? - CHAR_GREATER_THAN_SIGN : (*ptr == CHAR_APOSTROPHE)? - CHAR_APOSTROPHE : CHAR_RIGHT_CURLY_BRACKET; - goto NAMED_REF_OR_RECURSE; - } - - /* Back references are handled specially; must disable firstchar if - not set to cope with cases like (?=(\w+))\1: which would otherwise set - ':' later. */ - - if (escape < 0) - { - open_capitem *oc; - recno = -escape; - - /* Come here from named backref handling when the reference is to a - single group (i.e. not to a duplicated name. */ - - HANDLE_REFERENCE: - if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE; - previous = code; - *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF; - PUT2INC(code, 0, recno); - cd->backref_map |= (recno < 32)? (1 << recno) : 1; - if (recno > cd->top_backref) cd->top_backref = recno; - - /* Check to see if this back reference is recursive, that it, it - is inside the group that it references. A flag is set so that the - group can be made atomic. */ - - for (oc = cd->open_caps; oc != NULL; oc = oc->next) - { - if (oc->number == recno) - { - oc->flag = TRUE; - break; - } - } - } - - /* So are Unicode property matches, if supported. */ - -#ifdef SUPPORT_UCP - else if (escape == ESC_P || escape == ESC_p) - { - BOOL negated; - unsigned int ptype = 0, pdata = 0; - if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr)) - goto FAILED; - previous = code; - *code++ = ((escape == ESC_p) != negated)? OP_PROP : OP_NOTPROP; - *code++ = ptype; - *code++ = pdata; - } -#else - - /* If Unicode properties are not supported, \X, \P, and \p are not - allowed. */ - - else if (escape == ESC_X || escape == ESC_P || escape == ESC_p) - { - *errorcodeptr = ERR45; - goto FAILED; - } -#endif - - /* For the rest (including \X when Unicode properties are supported), we - can obtain the OP value by negating the escape value in the default - situation when PCRE_UCP is not set. When it *is* set, we substitute - Unicode property tests. Note that \b and \B do a one-character - lookbehind, and \A also behaves as if it does. */ - - else - { - if ((escape == ESC_b || escape == ESC_B || escape == ESC_A) && - cd->max_lookbehind == 0) - cd->max_lookbehind = 1; -#ifdef SUPPORT_UCP - if (escape >= ESC_DU && escape <= ESC_wu) - { - nestptr = ptr + 1; /* Where to resume */ - ptr = substitutes[escape - ESC_DU] - 1; /* Just before substitute */ - } - else -#endif - /* In non-UTF-8 mode, we turn \C into OP_ALLANY instead of OP_ANYBYTE - so that it works in DFA mode and in lookbehinds. */ - - { - previous = (escape > ESC_b && escape < ESC_Z)? code : NULL; - *code++ = (!utf && escape == ESC_C)? OP_ALLANY : escape; - } - } - continue; - } - - /* We have a data character whose value is in c. In UTF-8 mode it may have - a value > 127. We set its representation in the length/buffer, and then - handle it as a data character. */ - -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 - if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR) - mclength = PRIV(ord2utf)(c, mcbuffer); - else -#endif - - { - mcbuffer[0] = c; - mclength = 1; - } - goto ONE_CHAR; - - - /* ===================================================================*/ - /* Handle a literal character. It is guaranteed not to be whitespace or # - when the extended flag is set. If we are in a UTF mode, it may be a - multi-unit literal character. */ - - default: - NORMAL_CHAR: - mclength = 1; - mcbuffer[0] = c; - -#ifdef SUPPORT_UTF - if (utf && HAS_EXTRALEN(c)) - ACROSSCHAR(TRUE, ptr[1], mcbuffer[mclength++] = *(++ptr)); -#endif - - /* At this point we have the character's bytes in mcbuffer, and the length - in mclength. When not in UTF-8 mode, the length is always 1. */ - - ONE_CHAR: - previous = code; - - /* For caseless UTF-8 mode when UCP support is available, check whether - this character has more than one other case. If so, generate a special - OP_PROP item instead of OP_CHARI. */ - -#ifdef SUPPORT_UCP - if (utf && (options & PCRE_CASELESS) != 0) - { - GETCHAR(c, mcbuffer); - if ((c = UCD_CASESET(c)) != 0) - { - *code++ = OP_PROP; - *code++ = PT_CLIST; - *code++ = c; - if (firstcharflags == REQ_UNSET) - firstcharflags = zerofirstcharflags = REQ_NONE; - break; - } - } -#endif - - /* Caseful matches, or not one of the multicase characters. */ - - *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR; - for (c = 0; c < mclength; c++) *code++ = mcbuffer[c]; - - /* Remember if \r or \n were seen */ - - if (mcbuffer[0] == CHAR_CR || mcbuffer[0] == CHAR_NL) - cd->external_flags |= PCRE_HASCRORLF; - - /* Set the first and required bytes appropriately. If no previous first - byte, set it from this character, but revert to none on a zero repeat. - Otherwise, leave the firstchar value alone, and don't change it on a zero - repeat. */ - - if (firstcharflags == REQ_UNSET) - { - zerofirstcharflags = REQ_NONE; - zeroreqchar = reqchar; - zeroreqcharflags = reqcharflags; - - /* If the character is more than one byte long, we can set firstchar - only if it is not to be matched caselessly. */ - - if (mclength == 1 || req_caseopt == 0) - { - firstchar = mcbuffer[0] | req_caseopt; - firstchar = mcbuffer[0]; - firstcharflags = req_caseopt; - - if (mclength != 1) - { - reqchar = code[-1]; - reqcharflags = cd->req_varyopt; - } - } - else firstcharflags = reqcharflags = REQ_NONE; - } - - /* firstchar was previously set; we can set reqchar only if the length is - 1 or the matching is caseful. */ - - else - { - zerofirstchar = firstchar; - zerofirstcharflags = firstcharflags; - zeroreqchar = reqchar; - zeroreqcharflags = reqcharflags; - if (mclength == 1 || req_caseopt == 0) - { - reqchar = code[-1]; - reqcharflags = req_caseopt | cd->req_varyopt; - } - } - - break; /* End of literal character handling */ - } - } /* end of big loop */ - - -/* Control never reaches here by falling through, only by a goto for all the -error states. Pass back the position in the pattern so that it can be displayed -to the user for diagnosing the error. */ - -FAILED: -*ptrptr = ptr; -return FALSE; -} - - - -/************************************************* -* Compile sequence of alternatives * -*************************************************/ - -/* On entry, ptr is pointing past the bracket character, but on return it -points to the closing bracket, or vertical bar, or end of string. The code -variable is pointing at the byte into which the BRA operator has been stored. -This function is used during the pre-compile phase when we are trying to find -out the amount of memory needed, as well as during the real compile phase. The -value of lengthptr distinguishes the two phases. - -Arguments: - options option bits, including any changes for this subpattern - codeptr -> the address of the current code pointer - ptrptr -> the address of the current pattern pointer - errorcodeptr -> pointer to error code variable - lookbehind TRUE if this is a lookbehind assertion - reset_bracount TRUE to reset the count for each branch - skipbytes skip this many bytes at start (for brackets and OP_COND) - cond_depth depth of nesting for conditional subpatterns - firstcharptr place to put the first required character - firstcharflagsptr place to put the first character flags, or a negative number - reqcharptr place to put the last required character - reqcharflagsptr place to put the last required character flags, or a negative number - bcptr pointer to the chain of currently open branches - cd points to the data block with tables pointers etc. - lengthptr NULL during the real compile phase - points to length accumulator during pre-compile phase - -Returns: TRUE on success -*/ - -static BOOL -compile_regex(int options, pcre_uchar **codeptr, const pcre_uchar **ptrptr, - int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes, - int cond_depth, - pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr, - pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr, - branch_chain *bcptr, compile_data *cd, int *lengthptr) -{ -const pcre_uchar *ptr = *ptrptr; -pcre_uchar *code = *codeptr; -pcre_uchar *last_branch = code; -pcre_uchar *start_bracket = code; -pcre_uchar *reverse_count = NULL; -open_capitem capitem; -int capnumber = 0; -pcre_uint32 firstchar, reqchar; -pcre_int32 firstcharflags, reqcharflags; -pcre_uint32 branchfirstchar, branchreqchar; -pcre_int32 branchfirstcharflags, branchreqcharflags; -int length; -unsigned int orig_bracount; -unsigned int max_bracount; -branch_chain bc; - -bc.outer = bcptr; -bc.current_branch = code; - -firstchar = reqchar = 0; -firstcharflags = reqcharflags = REQ_UNSET; - -/* Accumulate the length for use in the pre-compile phase. Start with the -length of the BRA and KET and any extra bytes that are required at the -beginning. We accumulate in a local variable to save frequent testing of -lenthptr for NULL. We cannot do this by looking at the value of code at the -start and end of each alternative, because compiled items are discarded during -the pre-compile phase so that the work space is not exceeded. */ - -length = 2 + 2*LINK_SIZE + skipbytes; - -/* WARNING: If the above line is changed for any reason, you must also change -the code that abstracts option settings at the start of the pattern and makes -them global. It tests the value of length for (2 + 2*LINK_SIZE) in the -pre-compile phase to find out whether anything has yet been compiled or not. */ - -/* If this is a capturing subpattern, add to the chain of open capturing items -so that we can detect them if (*ACCEPT) is encountered. This is also used to -detect groups that contain recursive back references to themselves. Note that -only OP_CBRA need be tested here; changing this opcode to one of its variants, -e.g. OP_SCBRAPOS, happens later, after the group has been compiled. */ - -if (*code == OP_CBRA) - { - capnumber = GET2(code, 1 + LINK_SIZE); - capitem.number = capnumber; - capitem.next = cd->open_caps; - capitem.flag = FALSE; - cd->open_caps = &capitem; - } - -/* Offset is set zero to mark that this bracket is still open */ - -PUT(code, 1, 0); -code += 1 + LINK_SIZE + skipbytes; - -/* Loop for each alternative branch */ - -orig_bracount = max_bracount = cd->bracount; -for (;;) - { - /* For a (?| group, reset the capturing bracket count so that each branch - uses the same numbers. */ - - if (reset_bracount) cd->bracount = orig_bracount; - - /* Set up dummy OP_REVERSE if lookbehind assertion */ - - if (lookbehind) - { - *code++ = OP_REVERSE; - reverse_count = code; - PUTINC(code, 0, 0); - length += 1 + LINK_SIZE; - } - - /* Now compile the branch; in the pre-compile phase its length gets added - into the length. */ - - if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstchar, - &branchfirstcharflags, &branchreqchar, &branchreqcharflags, &bc, - cond_depth, cd, (lengthptr == NULL)? NULL : &length)) - { - *ptrptr = ptr; - return FALSE; - } - - /* Keep the highest bracket count in case (?| was used and some branch - has fewer than the rest. */ - - if (cd->bracount > max_bracount) max_bracount = cd->bracount; - - /* In the real compile phase, there is some post-processing to be done. */ - - if (lengthptr == NULL) - { - /* If this is the first branch, the firstchar and reqchar values for the - branch become the values for the regex. */ - - if (*last_branch != OP_ALT) - { - firstchar = branchfirstchar; - firstcharflags = branchfirstcharflags; - reqchar = branchreqchar; - reqcharflags = branchreqcharflags; - } - - /* If this is not the first branch, the first char and reqchar have to - match the values from all the previous branches, except that if the - previous value for reqchar didn't have REQ_VARY set, it can still match, - and we set REQ_VARY for the regex. */ - - else - { - /* If we previously had a firstchar, but it doesn't match the new branch, - we have to abandon the firstchar for the regex, but if there was - previously no reqchar, it takes on the value of the old firstchar. */ - - if (firstcharflags >= 0 && - (firstcharflags != branchfirstcharflags || firstchar != branchfirstchar)) - { - if (reqcharflags < 0) - { - reqchar = firstchar; - reqcharflags = firstcharflags; - } - firstcharflags = REQ_NONE; - } - - /* If we (now or from before) have no firstchar, a firstchar from the - branch becomes a reqchar if there isn't a branch reqchar. */ - - if (firstcharflags < 0 && branchfirstcharflags >= 0 && branchreqcharflags < 0) - { - branchreqchar = branchfirstchar; - branchreqcharflags = branchfirstcharflags; - } - - /* Now ensure that the reqchars match */ - - if (((reqcharflags & ~REQ_VARY) != (branchreqcharflags & ~REQ_VARY)) || - reqchar != branchreqchar) - reqcharflags = REQ_NONE; - else - { - reqchar = branchreqchar; - reqcharflags |= branchreqcharflags; /* To "or" REQ_VARY */ - } - } - - /* If lookbehind, check that this branch matches a fixed-length string, and - put the length into the OP_REVERSE item. Temporarily mark the end of the - branch with OP_END. If the branch contains OP_RECURSE, the result is -3 - because there may be forward references that we can't check here. Set a - flag to cause another lookbehind check at the end. Why not do it all at the - end? Because common, erroneous checks are picked up here and the offset of - the problem can be shown. */ - - if (lookbehind) - { - int fixed_length; - *code = OP_END; - fixed_length = find_fixedlength(last_branch, (options & PCRE_UTF8) != 0, - FALSE, cd); - DPRINTF(("fixed length = %d\n", fixed_length)); - if (fixed_length == -3) - { - cd->check_lookbehind = TRUE; - } - else if (fixed_length < 0) - { - *errorcodeptr = (fixed_length == -2)? ERR36 : - (fixed_length == -4)? ERR70: ERR25; - *ptrptr = ptr; - return FALSE; - } - else - { - if (fixed_length > cd->max_lookbehind) - cd->max_lookbehind = fixed_length; - PUT(reverse_count, 0, fixed_length); - } - } - } - - /* Reached end of expression, either ')' or end of pattern. In the real - compile phase, go back through the alternative branches and reverse the chain - of offsets, with the field in the BRA item now becoming an offset to the - first alternative. If there are no alternatives, it points to the end of the - group. The length in the terminating ket is always the length of the whole - bracketed item. Return leaving the pointer at the terminating char. */ - - if (*ptr != CHAR_VERTICAL_LINE) - { - if (lengthptr == NULL) - { - int branch_length = (int)(code - last_branch); - do - { - int prev_length = GET(last_branch, 1); - PUT(last_branch, 1, branch_length); - branch_length = prev_length; - last_branch -= branch_length; - } - while (branch_length > 0); - } - - /* Fill in the ket */ - - *code = OP_KET; - PUT(code, 1, (int)(code - start_bracket)); - code += 1 + LINK_SIZE; - - /* If it was a capturing subpattern, check to see if it contained any - recursive back references. If so, we must wrap it in atomic brackets. - In any event, remove the block from the chain. */ - - if (capnumber > 0) - { - if (cd->open_caps->flag) - { - memmove(start_bracket + 1 + LINK_SIZE, start_bracket, - IN_UCHARS(code - start_bracket)); - *start_bracket = OP_ONCE; - code += 1 + LINK_SIZE; - PUT(start_bracket, 1, (int)(code - start_bracket)); - *code = OP_KET; - PUT(code, 1, (int)(code - start_bracket)); - code += 1 + LINK_SIZE; - length += 2 + 2*LINK_SIZE; - } - cd->open_caps = cd->open_caps->next; - } - - /* Retain the highest bracket number, in case resetting was used. */ - - cd->bracount = max_bracount; - - /* Set values to pass back */ - - *codeptr = code; - *ptrptr = ptr; - *firstcharptr = firstchar; - *firstcharflagsptr = firstcharflags; - *reqcharptr = reqchar; - *reqcharflagsptr = reqcharflags; - if (lengthptr != NULL) - { - if (OFLOW_MAX - *lengthptr < length) - { - *errorcodeptr = ERR20; - return FALSE; - } - *lengthptr += length; - } - return TRUE; - } - - /* Another branch follows. In the pre-compile phase, we can move the code - pointer back to where it was for the start of the first branch. (That is, - pretend that each branch is the only one.) - - In the real compile phase, insert an ALT node. Its length field points back - to the previous branch while the bracket remains open. At the end the chain - is reversed. It's done like this so that the start of the bracket has a - zero offset until it is closed, making it possible to detect recursion. */ - - if (lengthptr != NULL) - { - code = *codeptr + 1 + LINK_SIZE + skipbytes; - length += 1 + LINK_SIZE; - } - else - { - *code = OP_ALT; - PUT(code, 1, (int)(code - last_branch)); - bc.current_branch = last_branch = code; - code += 1 + LINK_SIZE; - } - - ptr++; - } -/* Control never reaches here */ -} - - - - -/************************************************* -* Check for anchored expression * -*************************************************/ - -/* Try to find out if this is an anchored regular expression. Consider each -alternative branch. If they all start with OP_SOD or OP_CIRC, or with a bracket -all of whose alternatives start with OP_SOD or OP_CIRC (recurse ad lib), then -it's anchored. However, if this is a multiline pattern, then only OP_SOD will -be found, because ^ generates OP_CIRCM in that mode. - -We can also consider a regex to be anchored if OP_SOM starts all its branches. -This is the code for \G, which means "match at start of match position, taking -into account the match offset". - -A branch is also implicitly anchored if it starts with .* and DOTALL is set, -because that will try the rest of the pattern at all possible matching points, -so there is no point trying again.... er .... - -.... except when the .* appears inside capturing parentheses, and there is a -subsequent back reference to those parentheses. We haven't enough information -to catch that case precisely. - -At first, the best we could do was to detect when .* was in capturing brackets -and the highest back reference was greater than or equal to that level. -However, by keeping a bitmap of the first 31 back references, we can catch some -of the more common cases more precisely. - -... A second exception is when the .* appears inside an atomic group, because -this prevents the number of characters it matches from being adjusted. - -Arguments: - code points to start of expression (the bracket) - bracket_map a bitmap of which brackets we are inside while testing; this - handles up to substring 31; after that we just have to take - the less precise approach - cd points to the compile data block - atomcount atomic group level - -Returns: TRUE or FALSE -*/ - -static BOOL -is_anchored(register const pcre_uchar *code, unsigned int bracket_map, - compile_data *cd, int atomcount) -{ -do { - const pcre_uchar *scode = first_significant_code( - code + PRIV(OP_lengths)[*code], FALSE); - register int op = *scode; - - /* Non-capturing brackets */ - - if (op == OP_BRA || op == OP_BRAPOS || - op == OP_SBRA || op == OP_SBRAPOS) - { - if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE; - } - - /* Capturing brackets */ - - else if (op == OP_CBRA || op == OP_CBRAPOS || - op == OP_SCBRA || op == OP_SCBRAPOS) - { - int n = GET2(scode, 1+LINK_SIZE); - int new_map = bracket_map | ((n < 32)? (1 << n) : 1); - if (!is_anchored(scode, new_map, cd, atomcount)) return FALSE; - } - - /* Positive forward assertions and conditions */ - - else if (op == OP_ASSERT || op == OP_COND) - { - if (!is_anchored(scode, bracket_map, cd, atomcount)) return FALSE; - } - - /* Atomic groups */ - - else if (op == OP_ONCE || op == OP_ONCE_NC) - { - if (!is_anchored(scode, bracket_map, cd, atomcount + 1)) - return FALSE; - } - - /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and - it isn't in brackets that are or may be referenced or inside an atomic - group. */ - - else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR || - op == OP_TYPEPOSSTAR)) - { - if (scode[1] != OP_ALLANY || (bracket_map & cd->backref_map) != 0 || - atomcount > 0 || cd->had_pruneorskip) - return FALSE; - } - - /* Check for explicit anchoring */ - - else if (op != OP_SOD && op != OP_SOM && op != OP_CIRC) return FALSE; - - code += GET(code, 1); - } -while (*code == OP_ALT); /* Loop for each alternative */ -return TRUE; -} - - - -/************************************************* -* Check for starting with ^ or .* * -*************************************************/ - -/* This is called to find out if every branch starts with ^ or .* so that -"first char" processing can be done to speed things up in multiline -matching and for non-DOTALL patterns that start with .* (which must start at -the beginning or after \n). As in the case of is_anchored() (see above), we -have to take account of back references to capturing brackets that contain .* -because in that case we can't make the assumption. Also, the appearance of .* -inside atomic brackets or in a pattern that contains *PRUNE or *SKIP does not -count, because once again the assumption no longer holds. - -Arguments: - code points to start of expression (the bracket) - bracket_map a bitmap of which brackets we are inside while testing; this - handles up to substring 31; after that we just have to take - the less precise approach - cd points to the compile data - atomcount atomic group level - -Returns: TRUE or FALSE -*/ - -static BOOL -is_startline(const pcre_uchar *code, unsigned int bracket_map, - compile_data *cd, int atomcount) -{ -do { - const pcre_uchar *scode = first_significant_code( - code + PRIV(OP_lengths)[*code], FALSE); - register int op = *scode; - - /* If we are at the start of a conditional assertion group, *both* the - conditional assertion *and* what follows the condition must satisfy the test - for start of line. Other kinds of condition fail. Note that there may be an - auto-callout at the start of a condition. */ - - if (op == OP_COND) - { - scode += 1 + LINK_SIZE; - if (*scode == OP_CALLOUT) scode += PRIV(OP_lengths)[OP_CALLOUT]; - switch (*scode) - { - case OP_CREF: - case OP_DNCREF: - case OP_RREF: - case OP_DNRREF: - case OP_DEF: - return FALSE; - - default: /* Assertion */ - if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE; - do scode += GET(scode, 1); while (*scode == OP_ALT); - scode += 1 + LINK_SIZE; - break; - } - scode = first_significant_code(scode, FALSE); - op = *scode; - } - - /* Non-capturing brackets */ - - if (op == OP_BRA || op == OP_BRAPOS || - op == OP_SBRA || op == OP_SBRAPOS) - { - if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE; - } - - /* Capturing brackets */ - - else if (op == OP_CBRA || op == OP_CBRAPOS || - op == OP_SCBRA || op == OP_SCBRAPOS) - { - int n = GET2(scode, 1+LINK_SIZE); - int new_map = bracket_map | ((n < 32)? (1 << n) : 1); - if (!is_startline(scode, new_map, cd, atomcount)) return FALSE; - } - - /* Positive forward assertions */ - - else if (op == OP_ASSERT) - { - if (!is_startline(scode, bracket_map, cd, atomcount)) return FALSE; - } - - /* Atomic brackets */ - - else if (op == OP_ONCE || op == OP_ONCE_NC) - { - if (!is_startline(scode, bracket_map, cd, atomcount + 1)) return FALSE; - } - - /* .* means "start at start or after \n" if it isn't in atomic brackets or - brackets that may be referenced, as long as the pattern does not contain - *PRUNE or *SKIP, because these break the feature. Consider, for example, - /.*?a(*PRUNE)b/ with the subject "aab", which matches "ab", i.e. not at the - start of a line. */ - - else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR) - { - if (scode[1] != OP_ANY || (bracket_map & cd->backref_map) != 0 || - atomcount > 0 || cd->had_pruneorskip) - return FALSE; - } - - /* Check for explicit circumflex; anything else gives a FALSE result. Note - in particular that this includes atomic brackets OP_ONCE and OP_ONCE_NC - because the number of characters matched by .* cannot be adjusted inside - them. */ - - else if (op != OP_CIRC && op != OP_CIRCM) return FALSE; - - /* Move on to the next alternative */ - - code += GET(code, 1); - } -while (*code == OP_ALT); /* Loop for each alternative */ -return TRUE; -} - - - -/************************************************* -* Check for asserted fixed first char * -*************************************************/ - -/* During compilation, the "first char" settings from forward assertions are -discarded, because they can cause conflicts with actual literals that follow. -However, if we end up without a first char setting for an unanchored pattern, -it is worth scanning the regex to see if there is an initial asserted first -char. If all branches start with the same asserted char, or with a -non-conditional bracket all of whose alternatives start with the same asserted -char (recurse ad lib), then we return that char, with the flags set to zero or -REQ_CASELESS; otherwise return zero with REQ_NONE in the flags. - -Arguments: - code points to start of expression (the bracket) - flags points to the first char flags, or to REQ_NONE - inassert TRUE if in an assertion - -Returns: the fixed first char, or 0 with REQ_NONE in flags -*/ - -static pcre_uint32 -find_firstassertedchar(const pcre_uchar *code, pcre_int32 *flags, - BOOL inassert) -{ -register pcre_uint32 c = 0; -int cflags = REQ_NONE; - -*flags = REQ_NONE; -do { - pcre_uint32 d; - int dflags; - int xl = (*code == OP_CBRA || *code == OP_SCBRA || - *code == OP_CBRAPOS || *code == OP_SCBRAPOS)? IMM2_SIZE:0; - const pcre_uchar *scode = first_significant_code(code + 1+LINK_SIZE + xl, - TRUE); - register pcre_uchar op = *scode; - - switch(op) - { - default: - return 0; - - case OP_BRA: - case OP_BRAPOS: - case OP_CBRA: - case OP_SCBRA: - case OP_CBRAPOS: - case OP_SCBRAPOS: - case OP_ASSERT: - case OP_ONCE: - case OP_ONCE_NC: - d = find_firstassertedchar(scode, &dflags, op == OP_ASSERT); - if (dflags < 0) - return 0; - if (cflags < 0) { c = d; cflags = dflags; } else if (c != d || cflags != dflags) return 0; - break; - - case OP_EXACT: - scode += IMM2_SIZE; - /* Fall through */ - - case OP_CHAR: - case OP_PLUS: - case OP_MINPLUS: - case OP_POSPLUS: - if (!inassert) return 0; - if (cflags < 0) { c = scode[1]; cflags = 0; } - else if (c != scode[1]) return 0; - break; - - case OP_EXACTI: - scode += IMM2_SIZE; - /* Fall through */ - - case OP_CHARI: - case OP_PLUSI: - case OP_MINPLUSI: - case OP_POSPLUSI: - if (!inassert) return 0; - if (cflags < 0) { c = scode[1]; cflags = REQ_CASELESS; } - else if (c != scode[1]) return 0; - break; - } - - code += GET(code, 1); - } -while (*code == OP_ALT); - -*flags = cflags; -return c; -} - - - -/************************************************* -* Add an entry to the name/number table * -*************************************************/ - -/* This function is called between compiling passes to add an entry to the -name/number table, maintaining alphabetical order. Checking for permitted -and forbidden duplicates has already been done. - -Arguments: - cd the compile data block - name the name to add - length the length of the name - groupno the group number - -Returns: nothing -*/ - -static void -add_name(compile_data *cd, const pcre_uchar *name, int length, - unsigned int groupno) -{ -int i; -pcre_uchar *slot = cd->name_table; - -for (i = 0; i < cd->names_found; i++) - { - int crc = memcmp(name, slot+IMM2_SIZE, IN_UCHARS(length)); - if (crc == 0 && slot[IMM2_SIZE+length] != 0) - crc = -1; /* Current name is a substring */ - - /* Make space in the table and break the loop for an earlier name. For a - duplicate or later name, carry on. We do this for duplicates so that in the - simple case (when ?(| is not used) they are in order of their numbers. In all - cases they are in the order in which they appear in the pattern. */ - - if (crc < 0) - { - memmove(slot + cd->name_entry_size, slot, - IN_UCHARS((cd->names_found - i) * cd->name_entry_size)); - break; - } - - /* Continue the loop for a later or duplicate name */ - - slot += cd->name_entry_size; - } - -PUT2(slot, 0, groupno); -memcpy(slot + IMM2_SIZE, name, IN_UCHARS(length)); -slot[IMM2_SIZE + length] = 0; -cd->names_found++; -} - - - -/************************************************* -* Compile a Regular Expression * -*************************************************/ - -/* This function takes a string and returns a pointer to a block of store -holding a compiled version of the expression. The original API for this -function had no error code return variable; it is retained for backwards -compatibility. The new function is given a new name. - -Arguments: - pattern the regular expression - options various option bits - errorcodeptr pointer to error code variable (pcre_compile2() only) - can be NULL if you don't want a code value - errorptr pointer to pointer to error text - erroroffset ptr offset in pattern where error was detected - tables pointer to character tables or NULL - -Returns: pointer to compiled data block, or NULL on error, - with errorptr and erroroffset set -*/ - -#if defined COMPILE_PCRE8 -PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION -pcre_compile(const char *pattern, int options, const char **errorptr, - int *erroroffset, const unsigned char *tables) -#elif defined COMPILE_PCRE16 -PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION -pcre16_compile(PCRE_SPTR16 pattern, int options, const char **errorptr, - int *erroroffset, const unsigned char *tables) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DEFN pcre32 * PCRE_CALL_CONVENTION -pcre32_compile(PCRE_SPTR32 pattern, int options, const char **errorptr, - int *erroroffset, const unsigned char *tables) -#endif -{ -#if defined COMPILE_PCRE8 -return pcre_compile2(pattern, options, NULL, errorptr, erroroffset, tables); -#elif defined COMPILE_PCRE16 -return pcre16_compile2(pattern, options, NULL, errorptr, erroroffset, tables); -#elif defined COMPILE_PCRE32 -return pcre32_compile2(pattern, options, NULL, errorptr, erroroffset, tables); -#endif -} - - -#if defined COMPILE_PCRE8 -PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION -pcre_compile2(const char *pattern, int options, int *errorcodeptr, - const char **errorptr, int *erroroffset, const unsigned char *tables) -#elif defined COMPILE_PCRE16 -PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION -pcre16_compile2(PCRE_SPTR16 pattern, int options, int *errorcodeptr, - const char **errorptr, int *erroroffset, const unsigned char *tables) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DEFN pcre32 * PCRE_CALL_CONVENTION -pcre32_compile2(PCRE_SPTR32 pattern, int options, int *errorcodeptr, - const char **errorptr, int *erroroffset, const unsigned char *tables) -#endif -{ -REAL_PCRE *re; -int length = 1; /* For final END opcode */ -pcre_int32 firstcharflags, reqcharflags; -pcre_uint32 firstchar, reqchar; -pcre_uint32 limit_match = PCRE_UINT32_MAX; -pcre_uint32 limit_recursion = PCRE_UINT32_MAX; -int newline; -int errorcode = 0; -int skipatstart = 0; -BOOL utf; -BOOL never_utf = FALSE; -size_t size; -pcre_uchar *code; -const pcre_uchar *codestart; -const pcre_uchar *ptr; -compile_data compile_block; -compile_data *cd = &compile_block; - -/* This space is used for "compiling" into during the first phase, when we are -computing the amount of memory that is needed. Compiled items are thrown away -as soon as possible, so that a fairly large buffer should be sufficient for -this purpose. The same space is used in the second phase for remembering where -to fill in forward references to subpatterns. That may overflow, in which case -new memory is obtained from malloc(). */ - -pcre_uchar cworkspace[COMPILE_WORK_SIZE]; - -/* This vector is used for remembering name groups during the pre-compile. In a -similar way to cworkspace, it can be expanded using malloc() if necessary. */ - -named_group named_groups[NAMED_GROUP_LIST_SIZE]; - -/* Set this early so that early errors get offset 0. */ - -ptr = (const pcre_uchar *)pattern; - -/* We can't pass back an error message if errorptr is NULL; I guess the best we -can do is just return NULL, but we can set a code value if there is a code -pointer. */ - -if (errorptr == NULL) - { - if (errorcodeptr != NULL) *errorcodeptr = 99; - return NULL; - } - -*errorptr = NULL; -if (errorcodeptr != NULL) *errorcodeptr = ERR0; - -/* However, we can give a message for this error */ - -if (erroroffset == NULL) - { - errorcode = ERR16; - goto PCRE_EARLY_ERROR_RETURN2; - } - -*erroroffset = 0; - -/* Set up pointers to the individual character tables */ - -if (tables == NULL) tables = PRIV(default_tables); -cd->lcc = tables + lcc_offset; -cd->fcc = tables + fcc_offset; -cd->cbits = tables + cbits_offset; -cd->ctypes = tables + ctypes_offset; - -/* Check that all undefined public option bits are zero */ - -if ((options & ~PUBLIC_COMPILE_OPTIONS) != 0) - { - errorcode = ERR17; - goto PCRE_EARLY_ERROR_RETURN; - } - -/* If PCRE_NEVER_UTF is set, remember it. */ - -if ((options & PCRE_NEVER_UTF) != 0) never_utf = TRUE; - -/* Check for global one-time settings at the start of the pattern, and remember -the offset for later. */ - -cd->external_flags = 0; /* Initialize here for LIMIT_MATCH/RECURSION */ - -while (ptr[skipatstart] == CHAR_LEFT_PARENTHESIS && - ptr[skipatstart+1] == CHAR_ASTERISK) - { - int newnl = 0; - int newbsr = 0; - -/* For completeness and backward compatibility, (*UTFn) is supported in the -relevant libraries, but (*UTF) is generic and always supported. Note that -PCRE_UTF8 == PCRE_UTF16 == PCRE_UTF32. */ - -#ifdef COMPILE_PCRE8 - if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF8_RIGHTPAR, 5) == 0) - { skipatstart += 7; options |= PCRE_UTF8; continue; } -#endif -#ifdef COMPILE_PCRE16 - if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF16_RIGHTPAR, 6) == 0) - { skipatstart += 8; options |= PCRE_UTF16; continue; } -#endif -#ifdef COMPILE_PCRE32 - if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF32_RIGHTPAR, 6) == 0) - { skipatstart += 8; options |= PCRE_UTF32; continue; } -#endif - - else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 4) == 0) - { skipatstart += 6; options |= PCRE_UTF8; continue; } - else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UCP_RIGHTPAR, 4) == 0) - { skipatstart += 6; options |= PCRE_UCP; continue; } - else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_NO_AUTO_POSSESS_RIGHTPAR, 16) == 0) - { skipatstart += 18; options |= PCRE_NO_AUTO_POSSESS; continue; } - else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_NO_START_OPT_RIGHTPAR, 13) == 0) - { skipatstart += 15; options |= PCRE_NO_START_OPTIMIZE; continue; } - - else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_LIMIT_MATCH_EQ, 12) == 0) - { - pcre_uint32 c = 0; - int p = skipatstart + 14; - while (isdigit(ptr[p])) - { - if (c > PCRE_UINT32_MAX / 10 - 1) break; /* Integer overflow */ - c = c*10 + ptr[p++] - CHAR_0; - } - if (ptr[p++] != CHAR_RIGHT_PARENTHESIS) break; - if (c < limit_match) - { - limit_match = c; - cd->external_flags |= PCRE_MLSET; - } - skipatstart = p; - continue; - } - - else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_LIMIT_RECURSION_EQ, 16) == 0) - { - pcre_uint32 c = 0; - int p = skipatstart + 18; - while (isdigit(ptr[p])) - { - if (c > PCRE_UINT32_MAX / 10 - 1) break; /* Integer overflow check */ - c = c*10 + ptr[p++] - CHAR_0; - } - if (ptr[p++] != CHAR_RIGHT_PARENTHESIS) break; - if (c < limit_recursion) - { - limit_recursion = c; - cd->external_flags |= PCRE_RLSET; - } - skipatstart = p; - continue; - } - - if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_CR_RIGHTPAR, 3) == 0) - { skipatstart += 5; newnl = PCRE_NEWLINE_CR; } - else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_LF_RIGHTPAR, 3) == 0) - { skipatstart += 5; newnl = PCRE_NEWLINE_LF; } - else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_CRLF_RIGHTPAR, 5) == 0) - { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; } - else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_ANY_RIGHTPAR, 4) == 0) - { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; } - else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_ANYCRLF_RIGHTPAR, 8) == 0) - { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; } - - else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_BSR_ANYCRLF_RIGHTPAR, 12) == 0) - { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; } - else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_BSR_UNICODE_RIGHTPAR, 12) == 0) - { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; } - - if (newnl != 0) - options = (options & ~PCRE_NEWLINE_BITS) | newnl; - else if (newbsr != 0) - options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr; - else break; - } - -/* PCRE_UTF(16|32) have the same value as PCRE_UTF8. */ -utf = (options & PCRE_UTF8) != 0; -if (utf && never_utf) - { - errorcode = ERR78; - goto PCRE_EARLY_ERROR_RETURN2; - } - -/* Can't support UTF unless PCRE has been compiled to include the code. The -return of an error code from PRIV(valid_utf)() is a new feature, introduced in -release 8.13. It is passed back from pcre_[dfa_]exec(), but at the moment is -not used here. */ - -#ifdef SUPPORT_UTF -if (utf && (options & PCRE_NO_UTF8_CHECK) == 0 && - (errorcode = PRIV(valid_utf)((PCRE_PUCHAR)pattern, -1, erroroffset)) != 0) - { -#if defined COMPILE_PCRE8 - errorcode = ERR44; -#elif defined COMPILE_PCRE16 - errorcode = ERR74; -#elif defined COMPILE_PCRE32 - errorcode = ERR77; -#endif - goto PCRE_EARLY_ERROR_RETURN2; - } -#else -if (utf) - { - errorcode = ERR32; - goto PCRE_EARLY_ERROR_RETURN; - } -#endif - -/* Can't support UCP unless PCRE has been compiled to include the code. */ - -#ifndef SUPPORT_UCP -if ((options & PCRE_UCP) != 0) - { - errorcode = ERR67; - goto PCRE_EARLY_ERROR_RETURN; - } -#endif - -/* Check validity of \R options. */ - -if ((options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) == - (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) - { - errorcode = ERR56; - goto PCRE_EARLY_ERROR_RETURN; - } - -/* Handle different types of newline. The three bits give seven cases. The -current code allows for fixed one- or two-byte sequences, plus "any" and -"anycrlf". */ - -switch (options & PCRE_NEWLINE_BITS) - { - case 0: newline = NEWLINE; break; /* Build-time default */ - case PCRE_NEWLINE_CR: newline = CHAR_CR; break; - case PCRE_NEWLINE_LF: newline = CHAR_NL; break; - case PCRE_NEWLINE_CR+ - PCRE_NEWLINE_LF: newline = (CHAR_CR << 8) | CHAR_NL; break; - case PCRE_NEWLINE_ANY: newline = -1; break; - case PCRE_NEWLINE_ANYCRLF: newline = -2; break; - default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN; - } - -if (newline == -2) - { - cd->nltype = NLTYPE_ANYCRLF; - } -else if (newline < 0) - { - cd->nltype = NLTYPE_ANY; - } -else - { - cd->nltype = NLTYPE_FIXED; - if (newline > 255) - { - cd->nllen = 2; - cd->nl[0] = (newline >> 8) & 255; - cd->nl[1] = newline & 255; - } - else - { - cd->nllen = 1; - cd->nl[0] = newline; - } - } - -/* Maximum back reference and backref bitmap. The bitmap records up to 31 back -references to help in deciding whether (.*) can be treated as anchored or not. -*/ - -cd->top_backref = 0; -cd->backref_map = 0; - -/* Reflect pattern for debugging output */ - -DPRINTF(("------------------------------------------------------------------\n")); -#ifdef PCRE_DEBUG -print_puchar(stdout, (PCRE_PUCHAR)pattern); -#endif -DPRINTF(("\n")); - -/* Pretend to compile the pattern while actually just accumulating the length -of memory required. This behaviour is triggered by passing a non-NULL final -argument to compile_regex(). We pass a block of workspace (cworkspace) for it -to compile parts of the pattern into; the compiled code is discarded when it is -no longer needed, so hopefully this workspace will never overflow, though there -is a test for its doing so. */ - -cd->bracount = cd->final_bracount = 0; -cd->names_found = 0; -cd->name_entry_size = 0; -cd->name_table = NULL; -cd->dupnames = FALSE; -cd->namedrefcount = 0; -cd->start_code = cworkspace; -cd->hwm = cworkspace; -cd->start_workspace = cworkspace; -cd->workspace_size = COMPILE_WORK_SIZE; -cd->named_groups = named_groups; -cd->named_group_list_size = NAMED_GROUP_LIST_SIZE; -cd->start_pattern = (const pcre_uchar *)pattern; -cd->end_pattern = (const pcre_uchar *)(pattern + STRLEN_UC((const pcre_uchar *)pattern)); -cd->req_varyopt = 0; -cd->parens_depth = 0; -cd->assert_depth = 0; -cd->max_lookbehind = 0; -cd->external_options = options; -cd->open_caps = NULL; - -/* Now do the pre-compile. On error, errorcode will be set non-zero, so we -don't need to look at the result of the function here. The initial options have -been put into the cd block so that they can be changed if an option setting is -found within the regex right at the beginning. Bringing initial option settings -outside can help speed up starting point checks. */ - -ptr += skipatstart; -code = cworkspace; -*code = OP_BRA; - -(void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE, - FALSE, 0, 0, &firstchar, &firstcharflags, &reqchar, &reqcharflags, NULL, - cd, &length); -if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN; - -DPRINTF(("end pre-compile: length=%d workspace=%d\n", length, - (int)(cd->hwm - cworkspace))); - -if (length > MAX_PATTERN_SIZE) - { - errorcode = ERR20; - goto PCRE_EARLY_ERROR_RETURN; - } - -/* If there are groups with duplicate names and there are also references by -name, we must allow for the possibility of named references to duplicated -groups. These require an extra data item each. */ - -if (cd->dupnames && cd->namedrefcount > 0) - length += cd->namedrefcount * IMM2_SIZE * sizeof(pcre_uchar); - -/* Compute the size of the data block for storing the compiled pattern. Integer -overflow should no longer be possible because nowadays we limit the maximum -value of cd->names_found and cd->name_entry_size. */ - -size = sizeof(REAL_PCRE) + - (length + cd->names_found * cd->name_entry_size) * sizeof(pcre_uchar); - -/* Get the memory. */ - -re = (REAL_PCRE *)(PUBL(malloc))(size); -if (re == NULL) - { - errorcode = ERR21; - goto PCRE_EARLY_ERROR_RETURN; - } - -/* Put in the magic number, and save the sizes, initial options, internal -flags, and character table pointer. NULL is used for the default character -tables. The nullpad field is at the end; it's there to help in the case when a -regex compiled on a system with 4-byte pointers is run on another with 8-byte -pointers. */ - -re->magic_number = MAGIC_NUMBER; -re->size = (int)size; -re->options = cd->external_options; -re->flags = cd->external_flags; -re->limit_match = limit_match; -re->limit_recursion = limit_recursion; -re->first_char = 0; -re->req_char = 0; -re->name_table_offset = sizeof(REAL_PCRE) / sizeof(pcre_uchar); -re->name_entry_size = cd->name_entry_size; -re->name_count = cd->names_found; -re->ref_count = 0; -re->tables = (tables == PRIV(default_tables))? NULL : tables; -re->nullpad = NULL; -#ifdef COMPILE_PCRE32 -re->dummy = 0; -#else -re->dummy1 = re->dummy2 = re->dummy3 = 0; -#endif - -/* The starting points of the name/number translation table and of the code are -passed around in the compile data block. The start/end pattern and initial -options are already set from the pre-compile phase, as is the name_entry_size -field. Reset the bracket count and the names_found field. Also reset the hwm -field; this time it's used for remembering forward references to subpatterns. -*/ - -cd->final_bracount = cd->bracount; /* Save for checking forward references */ -cd->parens_depth = 0; -cd->assert_depth = 0; -cd->bracount = 0; -cd->max_lookbehind = 0; -cd->name_table = (pcre_uchar *)re + re->name_table_offset; -codestart = cd->name_table + re->name_entry_size * re->name_count; -cd->start_code = codestart; -cd->hwm = (pcre_uchar *)(cd->start_workspace); -cd->req_varyopt = 0; -cd->had_accept = FALSE; -cd->had_pruneorskip = FALSE; -cd->check_lookbehind = FALSE; -cd->open_caps = NULL; - -/* If any named groups were found, create the name/number table from the list -created in the first pass. */ - -if (cd->names_found > 0) - { - int i = cd->names_found; - named_group *ng = cd->named_groups; - cd->names_found = 0; - for (; i > 0; i--, ng++) - add_name(cd, ng->name, ng->length, ng->number); - if (cd->named_group_list_size > NAMED_GROUP_LIST_SIZE) - (PUBL(free))((void *)cd->named_groups); - } - -/* Set up a starting, non-extracting bracket, then compile the expression. On -error, errorcode will be set non-zero, so we don't need to look at the result -of the function here. */ - -ptr = (const pcre_uchar *)pattern + skipatstart; -code = (pcre_uchar *)codestart; -*code = OP_BRA; -(void)compile_regex(re->options, &code, &ptr, &errorcode, FALSE, FALSE, 0, 0, - &firstchar, &firstcharflags, &reqchar, &reqcharflags, NULL, cd, NULL); -re->top_bracket = cd->bracount; -re->top_backref = cd->top_backref; -re->max_lookbehind = cd->max_lookbehind; -re->flags = cd->external_flags | PCRE_MODE; - -if (cd->had_accept) - { - reqchar = 0; /* Must disable after (*ACCEPT) */ - reqcharflags = REQ_NONE; - } - -/* If not reached end of pattern on success, there's an excess bracket. */ - -if (errorcode == 0 && *ptr != CHAR_NULL) errorcode = ERR22; - -/* Fill in the terminating state and check for disastrous overflow, but -if debugging, leave the test till after things are printed out. */ - -*code++ = OP_END; - -#ifndef PCRE_DEBUG -if (code - codestart > length) errorcode = ERR23; -#endif - -#ifdef SUPPORT_VALGRIND -/* If the estimated length exceeds the really used length, mark the extra -allocated memory as unaddressable, so that any out-of-bound reads can be -detected. */ -VALGRIND_MAKE_MEM_NOACCESS(code, (length - (code - codestart)) * sizeof(pcre_uchar)); -#endif - -/* Fill in any forward references that are required. There may be repeated -references; optimize for them, as searching a large regex takes time. */ - -if (cd->hwm > cd->start_workspace) - { - int prev_recno = -1; - const pcre_uchar *groupptr = NULL; - while (errorcode == 0 && cd->hwm > cd->start_workspace) - { - int offset, recno; - cd->hwm -= LINK_SIZE; - offset = GET(cd->hwm, 0); - recno = GET(codestart, offset); - if (recno != prev_recno) - { - groupptr = PRIV(find_bracket)(codestart, utf, recno); - prev_recno = recno; - } - if (groupptr == NULL) errorcode = ERR53; - else PUT(((pcre_uchar *)codestart), offset, (int)(groupptr - codestart)); - } - } - -/* If the workspace had to be expanded, free the new memory. Set the pointer to -NULL to indicate that forward references have been filled in. */ - -if (cd->workspace_size > COMPILE_WORK_SIZE) - (PUBL(free))((void *)cd->start_workspace); -cd->start_workspace = NULL; - -/* Give an error if there's back reference to a non-existent capturing -subpattern. */ - -if (errorcode == 0 && re->top_backref > re->top_bracket) errorcode = ERR15; - -/* Unless disabled, check whether single character iterators can be -auto-possessified. The function overwrites the appropriate opcode values. */ - -if ((options & PCRE_NO_AUTO_POSSESS) == 0) - auto_possessify((pcre_uchar *)codestart, utf, cd); - -/* If there were any lookbehind assertions that contained OP_RECURSE -(recursions or subroutine calls), a flag is set for them to be checked here, -because they may contain forward references. Actual recursions cannot be fixed -length, but subroutine calls can. It is done like this so that those without -OP_RECURSE that are not fixed length get a diagnosic with a useful offset. The -exceptional ones forgo this. We scan the pattern to check that they are fixed -length, and set their lengths. */ - -if (cd->check_lookbehind) - { - pcre_uchar *cc = (pcre_uchar *)codestart; - - /* Loop, searching for OP_REVERSE items, and process those that do not have - their length set. (Actually, it will also re-process any that have a length - of zero, but that is a pathological case, and it does no harm.) When we find - one, we temporarily terminate the branch it is in while we scan it. */ - - for (cc = (pcre_uchar *)PRIV(find_bracket)(codestart, utf, -1); - cc != NULL; - cc = (pcre_uchar *)PRIV(find_bracket)(cc, utf, -1)) - { - if (GET(cc, 1) == 0) - { - int fixed_length; - pcre_uchar *be = cc - 1 - LINK_SIZE + GET(cc, -LINK_SIZE); - int end_op = *be; - *be = OP_END; - fixed_length = find_fixedlength(cc, (re->options & PCRE_UTF8) != 0, TRUE, - cd); - *be = end_op; - DPRINTF(("fixed length = %d\n", fixed_length)); - if (fixed_length < 0) - { - errorcode = (fixed_length == -2)? ERR36 : - (fixed_length == -4)? ERR70 : ERR25; - break; - } - if (fixed_length > cd->max_lookbehind) cd->max_lookbehind = fixed_length; - PUT(cc, 1, fixed_length); - } - cc += 1 + LINK_SIZE; - } - } - -/* Failed to compile, or error while post-processing */ - -if (errorcode != 0) - { - (PUBL(free))(re); - PCRE_EARLY_ERROR_RETURN: - *erroroffset = (int)(ptr - (const pcre_uchar *)pattern); - PCRE_EARLY_ERROR_RETURN2: - *errorptr = find_error_text(errorcode); - if (errorcodeptr != NULL) *errorcodeptr = errorcode; - return NULL; - } - -/* If the anchored option was not passed, set the flag if we can determine that -the pattern is anchored by virtue of ^ characters or \A or anything else, such -as starting with non-atomic .* when DOTALL is set and there are no occurrences -of *PRUNE or *SKIP. - -Otherwise, if we know what the first byte has to be, save it, because that -speeds up unanchored matches no end. If not, see if we can set the -PCRE_STARTLINE flag. This is helpful for multiline matches when all branches -start with ^. and also when all branches start with non-atomic .* for -non-DOTALL matches when *PRUNE and SKIP are not present. */ - -if ((re->options & PCRE_ANCHORED) == 0) - { - if (is_anchored(codestart, 0, cd, 0)) re->options |= PCRE_ANCHORED; - else - { - if (firstcharflags < 0) - firstchar = find_firstassertedchar(codestart, &firstcharflags, FALSE); - if (firstcharflags >= 0) /* Remove caseless flag for non-caseable chars */ - { -#if defined COMPILE_PCRE8 - re->first_char = firstchar & 0xff; -#elif defined COMPILE_PCRE16 - re->first_char = firstchar & 0xffff; -#elif defined COMPILE_PCRE32 - re->first_char = firstchar; -#endif - if ((firstcharflags & REQ_CASELESS) != 0) - { -#if defined SUPPORT_UCP && !(defined COMPILE_PCRE8) - /* We ignore non-ASCII first chars in 8 bit mode. */ - if (utf) - { - if (re->first_char < 128) - { - if (cd->fcc[re->first_char] != re->first_char) - re->flags |= PCRE_FCH_CASELESS; - } - else if (UCD_OTHERCASE(re->first_char) != re->first_char) - re->flags |= PCRE_FCH_CASELESS; - } - else -#endif - if (MAX_255(re->first_char) - && cd->fcc[re->first_char] != re->first_char) - re->flags |= PCRE_FCH_CASELESS; - } - - re->flags |= PCRE_FIRSTSET; - } - - else if (is_startline(codestart, 0, cd, 0)) re->flags |= PCRE_STARTLINE; - } - } - -/* For an anchored pattern, we use the "required byte" only if it follows a -variable length item in the regex. Remove the caseless flag for non-caseable -bytes. */ - -if (reqcharflags >= 0 && - ((re->options & PCRE_ANCHORED) == 0 || (reqcharflags & REQ_VARY) != 0)) - { -#if defined COMPILE_PCRE8 - re->req_char = reqchar & 0xff; -#elif defined COMPILE_PCRE16 - re->req_char = reqchar & 0xffff; -#elif defined COMPILE_PCRE32 - re->req_char = reqchar; -#endif - if ((reqcharflags & REQ_CASELESS) != 0) - { -#if defined SUPPORT_UCP && !(defined COMPILE_PCRE8) - /* We ignore non-ASCII first chars in 8 bit mode. */ - if (utf) - { - if (re->req_char < 128) - { - if (cd->fcc[re->req_char] != re->req_char) - re->flags |= PCRE_RCH_CASELESS; - } - else if (UCD_OTHERCASE(re->req_char) != re->req_char) - re->flags |= PCRE_RCH_CASELESS; - } - else -#endif - if (MAX_255(re->req_char) && cd->fcc[re->req_char] != re->req_char) - re->flags |= PCRE_RCH_CASELESS; - } - - re->flags |= PCRE_REQCHSET; - } - -/* Print out the compiled data if debugging is enabled. This is never the -case when building a production library. */ - -#ifdef PCRE_DEBUG -printf("Length = %d top_bracket = %d top_backref = %d\n", - length, re->top_bracket, re->top_backref); - -printf("Options=%08x\n", re->options); - -if ((re->flags & PCRE_FIRSTSET) != 0) - { - pcre_uchar ch = re->first_char; - const char *caseless = - ((re->flags & PCRE_FCH_CASELESS) == 0)? "" : " (caseless)"; - if (PRINTABLE(ch)) printf("First char = %c%s\n", ch, caseless); - else printf("First char = \\x%02x%s\n", ch, caseless); - } - -if ((re->flags & PCRE_REQCHSET) != 0) - { - pcre_uchar ch = re->req_char; - const char *caseless = - ((re->flags & PCRE_RCH_CASELESS) == 0)? "" : " (caseless)"; - if (PRINTABLE(ch)) printf("Req char = %c%s\n", ch, caseless); - else printf("Req char = \\x%02x%s\n", ch, caseless); - } - -#if defined COMPILE_PCRE8 -pcre_printint((pcre *)re, stdout, TRUE); -#elif defined COMPILE_PCRE16 -pcre16_printint((pcre *)re, stdout, TRUE); -#elif defined COMPILE_PCRE32 -pcre32_printint((pcre *)re, stdout, TRUE); -#endif - -/* This check is done here in the debugging case so that the code that -was compiled can be seen. */ - -if (code - codestart > length) - { - (PUBL(free))(re); - *errorptr = find_error_text(ERR23); - *erroroffset = ptr - (pcre_uchar *)pattern; - if (errorcodeptr != NULL) *errorcodeptr = ERR23; - return NULL; - } -#endif /* PCRE_DEBUG */ - -/* Check for a pattern than can match an empty string, so that this information -can be provided to applications. */ - -do - { - if (could_be_empty_branch(codestart, code, utf, cd, NULL)) - { - re->flags |= PCRE_MATCH_EMPTY; - break; - } - codestart += GET(codestart, 1); - } -while (*codestart == OP_ALT); - -#if defined COMPILE_PCRE8 -return (pcre *)re; -#elif defined COMPILE_PCRE16 -return (pcre16 *)re; -#elif defined COMPILE_PCRE32 -return (pcre32 *)re; -#endif -} - -/* End of pcre_compile.c */ - + Copyright (c) 1997-2012 University of Cambridge + +----------------------------------------------------------------------------- +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. + + * Neither the name of the University of Cambridge 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 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. +----------------------------------------------------------------------------- +*/ + + +/* This module contains the external function pcre_compile(), along with +supporting internal functions that are not used by other modules. */ + + +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + +#define NLBLOCK cd /* Block containing newline information */ +#define PSSTART start_pattern /* Field containing processed string start */ +#define PSEND end_pattern /* Field containing processed string end */ + +#include "pcre_internal.h" + + +/* When PCRE_DEBUG is defined, we need the pcre(16)_printint() function, which +is also used by pcretest. PCRE_DEBUG is not defined when building a production +library. We do not need to select pcre16_printint.c specially, because the +COMPILE_PCREx macro will already be appropriately set. */ + +#ifdef PCRE_DEBUG +/* pcre_printint.c should not include any headers */ +#define PCRE_INCLUDED +#include "pcre_printint.c" +#undef PCRE_INCLUDED +#endif + + +/* Macro for setting individual bits in class bitmaps. */ + +#define SETBIT(a,b) a[b/8] |= (1 << (b%8)) + +/* Maximum length value to check against when making sure that the integer that +holds the compiled pattern length does not overflow. We make it a bit less than +INT_MAX to allow for adding in group terminating bytes, so that we don't have +to check them every time. */ + +#define OFLOW_MAX (INT_MAX - 20) + + +/************************************************* +* Code parameters and static tables * +*************************************************/ + +/* This value specifies the size of stack workspace that is used during the +first pre-compile phase that determines how much memory is required. The regex +is partly compiled into this space, but the compiled parts are discarded as +soon as they can be, so that hopefully there will never be an overrun. The code +does, however, check for an overrun. The largest amount I've seen used is 218, +so this number is very generous. + +The same workspace is used during the second, actual compile phase for +remembering forward references to groups so that they can be filled in at the +end. Each entry in this list occupies LINK_SIZE bytes, so even when LINK_SIZE +is 4 there is plenty of room for most patterns. However, the memory can get +filled up by repetitions of forward references, for example patterns like +/(?1){0,1999}(b)/, and one user did hit the limit. The code has been changed so +that the workspace is expanded using malloc() in this situation. The value +below is therefore a minimum, and we put a maximum on it for safety. The +minimum is now also defined in terms of LINK_SIZE so that the use of malloc() +kicks in at the same number of forward references in all cases. */ + +#define COMPILE_WORK_SIZE (2048*LINK_SIZE) +#define COMPILE_WORK_SIZE_MAX (100*COMPILE_WORK_SIZE) + +/* The overrun tests check for a slightly smaller size so that they detect the +overrun before it actually does run off the end of the data block. */ + +#define WORK_SIZE_SAFETY_MARGIN (100) + +/* Private flags added to firstchar and reqchar. */ + +#define REQ_CASELESS 0x10000000l /* Indicates caselessness */ +#define REQ_VARY 0x20000000l /* Reqchar followed non-literal item */ + +/* Repeated character flags. */ + +#define UTF_LENGTH 0x10000000l /* The char contains its length. */ + +/* Table for handling escaped characters in the range '0'-'z'. Positive returns +are simple data values; negative values are for special things like \d and so +on. Zero means further processing is needed (for things like \x), or the escape +is invalid. */ + +#ifndef EBCDIC + +/* This is the "normal" table for ASCII systems or for EBCDIC systems running +in UTF-8 mode. */ + +static const short int escapes[] = { + 0, 0, + 0, 0, + 0, 0, + 0, 0, + 0, 0, + CHAR_COLON, CHAR_SEMICOLON, + CHAR_LESS_THAN_SIGN, CHAR_EQUALS_SIGN, + CHAR_GREATER_THAN_SIGN, CHAR_QUESTION_MARK, + CHAR_COMMERCIAL_AT, -ESC_A, + -ESC_B, -ESC_C, + -ESC_D, -ESC_E, + 0, -ESC_G, + -ESC_H, 0, + 0, -ESC_K, + 0, 0, + -ESC_N, 0, + -ESC_P, -ESC_Q, + -ESC_R, -ESC_S, + 0, 0, + -ESC_V, -ESC_W, + -ESC_X, 0, + -ESC_Z, CHAR_LEFT_SQUARE_BRACKET, + CHAR_BACKSLASH, CHAR_RIGHT_SQUARE_BRACKET, + CHAR_CIRCUMFLEX_ACCENT, CHAR_UNDERSCORE, + CHAR_GRAVE_ACCENT, 7, + -ESC_b, 0, + -ESC_d, ESC_e, + ESC_f, 0, + -ESC_h, 0, + 0, -ESC_k, + 0, 0, + ESC_n, 0, + -ESC_p, 0, + ESC_r, -ESC_s, + ESC_tee, 0, + -ESC_v, -ESC_w, + 0, 0, + -ESC_z +}; + +#else + +/* This is the "abnormal" table for EBCDIC systems without UTF-8 support. */ + +static const short int escapes[] = { +/* 48 */ 0, 0, 0, '.', '<', '(', '+', '|', +/* 50 */ '&', 0, 0, 0, 0, 0, 0, 0, +/* 58 */ 0, 0, '!', '$', '*', ')', ';', '~', +/* 60 */ '-', '/', 0, 0, 0, 0, 0, 0, +/* 68 */ 0, 0, '|', ',', '%', '_', '>', '?', +/* 70 */ 0, 0, 0, 0, 0, 0, 0, 0, +/* 78 */ 0, '`', ':', '#', '@', '\'', '=', '"', +/* 80 */ 0, 7, -ESC_b, 0, -ESC_d, ESC_e, ESC_f, 0, +/* 88 */-ESC_h, 0, 0, '{', 0, 0, 0, 0, +/* 90 */ 0, 0, -ESC_k, 'l', 0, ESC_n, 0, -ESC_p, +/* 98 */ 0, ESC_r, 0, '}', 0, 0, 0, 0, +/* A0 */ 0, '~', -ESC_s, ESC_tee, 0,-ESC_v, -ESC_w, 0, +/* A8 */ 0,-ESC_z, 0, 0, 0, '[', 0, 0, +/* B0 */ 0, 0, 0, 0, 0, 0, 0, 0, +/* B8 */ 0, 0, 0, 0, 0, ']', '=', '-', +/* C0 */ '{',-ESC_A, -ESC_B, -ESC_C, -ESC_D,-ESC_E, 0, -ESC_G, +/* C8 */-ESC_H, 0, 0, 0, 0, 0, 0, 0, +/* D0 */ '}', 0, -ESC_K, 0, 0,-ESC_N, 0, -ESC_P, +/* D8 */-ESC_Q,-ESC_R, 0, 0, 0, 0, 0, 0, +/* E0 */ '\\', 0, -ESC_S, 0, 0,-ESC_V, -ESC_W, -ESC_X, +/* E8 */ 0,-ESC_Z, 0, 0, 0, 0, 0, 0, +/* F0 */ 0, 0, 0, 0, 0, 0, 0, 0, +/* F8 */ 0, 0, 0, 0, 0, 0, 0, 0 +}; +#endif + + +/* Table of special "verbs" like (*PRUNE). This is a short table, so it is +searched linearly. Put all the names into a single string, in order to reduce +the number of relocations when a shared library is dynamically linked. The +string is built from string macros so that it works in UTF-8 mode on EBCDIC +platforms. */ + +typedef struct verbitem { + int len; /* Length of verb name */ + int op; /* Op when no arg, or -1 if arg mandatory */ + int op_arg; /* Op when arg present, or -1 if not allowed */ +} verbitem; + +static const char verbnames[] = + "\0" /* Empty name is a shorthand for MARK */ + STRING_MARK0 + STRING_ACCEPT0 + STRING_COMMIT0 + STRING_F0 + STRING_FAIL0 + STRING_PRUNE0 + STRING_SKIP0 + STRING_THEN; + +static const verbitem verbs[] = { + { 0, -1, OP_MARK }, + { 4, -1, OP_MARK }, + { 6, OP_ACCEPT, -1 }, + { 6, OP_COMMIT, -1 }, + { 1, OP_FAIL, -1 }, + { 4, OP_FAIL, -1 }, + { 5, OP_PRUNE, OP_PRUNE_ARG }, + { 4, OP_SKIP, OP_SKIP_ARG }, + { 4, OP_THEN, OP_THEN_ARG } +}; + +static const int verbcount = sizeof(verbs)/sizeof(verbitem); + + +/* Tables of names of POSIX character classes and their lengths. The names are +now all in a single string, to reduce the number of relocations when a shared +library is dynamically loaded. The list of lengths is terminated by a zero +length entry. The first three must be alpha, lower, upper, as this is assumed +for handling case independence. */ + +static const char posix_names[] = + STRING_alpha0 STRING_lower0 STRING_upper0 STRING_alnum0 + STRING_ascii0 STRING_blank0 STRING_cntrl0 STRING_digit0 + STRING_graph0 STRING_print0 STRING_punct0 STRING_space0 + STRING_word0 STRING_xdigit; + +static const pcre_uint8 posix_name_lengths[] = { + 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 }; + +/* Table of class bit maps for each POSIX class. Each class is formed from a +base map, with an optional addition or removal of another map. Then, for some +classes, there is some additional tweaking: for [:blank:] the vertical space +characters are removed, and for [:alpha:] and [:alnum:] the underscore +character is removed. The triples in the table consist of the base map offset, +second map offset or -1 if no second map, and a non-negative value for map +addition or a negative value for map subtraction (if there are two maps). The +absolute value of the third field has these meanings: 0 => no tweaking, 1 => +remove vertical space characters, 2 => remove underscore. */ + +static const int posix_class_maps[] = { + cbit_word, cbit_digit, -2, /* alpha */ + cbit_lower, -1, 0, /* lower */ + cbit_upper, -1, 0, /* upper */ + cbit_word, -1, 2, /* alnum - word without underscore */ + cbit_print, cbit_cntrl, 0, /* ascii */ + cbit_space, -1, 1, /* blank - a GNU extension */ + cbit_cntrl, -1, 0, /* cntrl */ + cbit_digit, -1, 0, /* digit */ + cbit_graph, -1, 0, /* graph */ + cbit_print, -1, 0, /* print */ + cbit_punct, -1, 0, /* punct */ + cbit_space, -1, 0, /* space */ + cbit_word, -1, 0, /* word - a Perl extension */ + cbit_xdigit,-1, 0 /* xdigit */ +}; + +/* Table of substitutes for \d etc when PCRE_UCP is set. The POSIX class +substitutes must be in the order of the names, defined above, and there are +both positive and negative cases. NULL means no substitute. */ + +#ifdef SUPPORT_UCP +static const pcre_uchar string_PNd[] = { + CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, + CHAR_N, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_pNd[] = { + CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, + CHAR_N, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_PXsp[] = { + CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, + CHAR_X, CHAR_s, CHAR_p, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_pXsp[] = { + CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, + CHAR_X, CHAR_s, CHAR_p, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_PXwd[] = { + CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, + CHAR_X, CHAR_w, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_pXwd[] = { + CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, + CHAR_X, CHAR_w, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' }; + +static const pcre_uchar *substitutes[] = { + string_PNd, /* \D */ + string_pNd, /* \d */ + string_PXsp, /* \S */ /* NOTE: Xsp is Perl space */ + string_pXsp, /* \s */ + string_PXwd, /* \W */ + string_pXwd /* \w */ +}; + +static const pcre_uchar string_pL[] = { + CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, + CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_pLl[] = { + CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, + CHAR_L, CHAR_l, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_pLu[] = { + CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, + CHAR_L, CHAR_u, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_pXan[] = { + CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, + CHAR_X, CHAR_a, CHAR_n, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_h[] = { + CHAR_BACKSLASH, CHAR_h, '\0' }; +static const pcre_uchar string_pXps[] = { + CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET, + CHAR_X, CHAR_p, CHAR_s, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_PL[] = { + CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, + CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_PLl[] = { + CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, + CHAR_L, CHAR_l, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_PLu[] = { + CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, + CHAR_L, CHAR_u, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_PXan[] = { + CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, + CHAR_X, CHAR_a, CHAR_n, CHAR_RIGHT_CURLY_BRACKET, '\0' }; +static const pcre_uchar string_H[] = { + CHAR_BACKSLASH, CHAR_H, '\0' }; +static const pcre_uchar string_PXps[] = { + CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET, + CHAR_X, CHAR_p, CHAR_s, CHAR_RIGHT_CURLY_BRACKET, '\0' }; + +static const pcre_uchar *posix_substitutes[] = { + string_pL, /* alpha */ + string_pLl, /* lower */ + string_pLu, /* upper */ + string_pXan, /* alnum */ + NULL, /* ascii */ + string_h, /* blank */ + NULL, /* cntrl */ + string_pNd, /* digit */ + NULL, /* graph */ + NULL, /* print */ + NULL, /* punct */ + string_pXps, /* space */ /* NOTE: Xps is POSIX space */ + string_pXwd, /* word */ + NULL, /* xdigit */ + /* Negated cases */ + string_PL, /* ^alpha */ + string_PLl, /* ^lower */ + string_PLu, /* ^upper */ + string_PXan, /* ^alnum */ + NULL, /* ^ascii */ + string_H, /* ^blank */ + NULL, /* ^cntrl */ + string_PNd, /* ^digit */ + NULL, /* ^graph */ + NULL, /* ^print */ + NULL, /* ^punct */ + string_PXps, /* ^space */ /* NOTE: Xps is POSIX space */ + string_PXwd, /* ^word */ + NULL /* ^xdigit */ +}; +#define POSIX_SUBSIZE (sizeof(posix_substitutes) / sizeof(pcre_uchar *)) +#endif + +#define STRING(a) # a +#define XSTRING(s) STRING(s) + +/* The texts of compile-time error messages. These are "char *" because they +are passed to the outside world. Do not ever re-use any error number, because +they are documented. Always add a new error instead. Messages marked DEAD below +are no longer used. This used to be a table of strings, but in order to reduce +the number of relocations needed when a shared library is loaded dynamically, +it is now one long string. We cannot use a table of offsets, because the +lengths of inserts such as XSTRING(MAX_NAME_SIZE) are not known. Instead, we +simply count through to the one we want - this isn't a performance issue +because these strings are used only when there is a compilation error. + +Each substring ends with \0 to insert a null character. This includes the final +substring, so that the whole string ends with \0\0, which can be detected when +counting through. */ + +static const char error_texts[] = + "no error\0" + "\\ at end of pattern\0" + "\\c at end of pattern\0" + "unrecognized character follows \\\0" + "numbers out of order in {} quantifier\0" + /* 5 */ + "number too big in {} quantifier\0" + "missing terminating ] for character class\0" + "invalid escape sequence in character class\0" + "range out of order in character class\0" + "nothing to repeat\0" + /* 10 */ + "operand of unlimited repeat could match the empty string\0" /** DEAD **/ + "internal error: unexpected repeat\0" + "unrecognized character after (? or (?-\0" + "POSIX named classes are supported only within a class\0" + "missing )\0" + /* 15 */ + "reference to non-existent subpattern\0" + "erroffset passed as NULL\0" + "unknown option bit(s) set\0" + "missing ) after comment\0" + "parentheses nested too deeply\0" /** DEAD **/ + /* 20 */ + "regular expression is too large\0" + "failed to get memory\0" + "unmatched parentheses\0" + "internal error: code overflow\0" + "unrecognized character after (?<\0" + /* 25 */ + "lookbehind assertion is not fixed length\0" + "malformed number or name after (?(\0" + "conditional group contains more than two branches\0" + "assertion expected after (?(\0" + "(?R or (?[+-]digits must be followed by )\0" + /* 30 */ + "unknown POSIX class name\0" + "POSIX collating elements are not supported\0" + "this version of PCRE is compiled without UTF support\0" + "spare error\0" /** DEAD **/ + "character value in \\x{...} sequence is too large\0" + /* 35 */ + "invalid condition (?(0)\0" + "\\C not allowed in lookbehind assertion\0" + "PCRE does not support \\L, \\l, \\N{name}, \\U, or \\u\0" + "number after (?C is > 255\0" + "closing ) for (?C expected\0" + /* 40 */ + "recursive call could loop indefinitely\0" + "unrecognized character after (?P\0" + "syntax error in subpattern name (missing terminator)\0" + "two named subpatterns have the same name\0" + "invalid UTF-8 string\0" + /* 45 */ + "support for \\P, \\p, and \\X has not been compiled\0" + "malformed \\P or \\p sequence\0" + "unknown property name after \\P or \\p\0" + "subpattern name is too long (maximum " XSTRING(MAX_NAME_SIZE) " characters)\0" + "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0" + /* 50 */ + "repeated subpattern is too long\0" /** DEAD **/ + "octal value is greater than \\377 in 8-bit non-UTF-8 mode\0" + "internal error: overran compiling workspace\0" + "internal error: previously-checked referenced subpattern not found\0" + "DEFINE group contains more than one branch\0" + /* 55 */ + "repeating a DEFINE group is not allowed\0" /** DEAD **/ + "inconsistent NEWLINE options\0" + "\\g is not followed by a braced, angle-bracketed, or quoted name/number or by a plain number\0" + "a numbered reference must not be zero\0" + "an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)\0" + /* 60 */ + "(*VERB) not recognized\0" + "number is too big\0" + "subpattern name expected\0" + "digit expected after (?+\0" + "] is an invalid data character in JavaScript compatibility mode\0" + /* 65 */ + "different names for subpatterns of the same number are not allowed\0" + "(*MARK) must have an argument\0" + "this version of PCRE is not compiled with Unicode property support\0" + "\\c must be followed by an ASCII character\0" + "\\k is not followed by a braced, angle-bracketed, or quoted name\0" + /* 70 */ + "internal error: unknown opcode in find_fixedlength()\0" + "\\N is not supported in a class\0" + "too many forward references\0" + "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0" + "invalid UTF-16 string\0" + /* 75 */ + "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0" + "character value in \\u.... sequence is too large\0" + ; + +/* Table to identify digits and hex digits. This is used when compiling +patterns. Note that the tables in chartables are dependent on the locale, and +may mark arbitrary characters as digits - but the PCRE compiling code expects +to handle only 0-9, a-z, and A-Z as digits when compiling. That is why we have +a private table here. It costs 256 bytes, but it is a lot faster than doing +character value tests (at least in some simple cases I timed), and in some +applications one wants PCRE to compile efficiently as well as match +efficiently. + +For convenience, we use the same bit definitions as in chartables: + + 0x04 decimal digit + 0x08 hexadecimal digit + +Then we can use ctype_digit and ctype_xdigit in the code. */ + +/* Using a simple comparison for decimal numbers rather than a memory read +is much faster, and the resulting code is simpler (the compiler turns it +into a subtraction and unsigned comparison). */ + +#define IS_DIGIT(x) ((x) >= CHAR_0 && (x) <= CHAR_9) + +#ifndef EBCDIC + +/* This is the "normal" case, for ASCII systems, and EBCDIC systems running in +UTF-8 mode. */ + +static const pcre_uint8 digitab[] = + { + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 0- 7 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 8- 15 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 16- 23 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 24- 31 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* - ' */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* ( - / */ + 0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /* 0 - 7 */ + 0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00, /* 8 - ? */ + 0x00,0x08,0x08,0x08,0x08,0x08,0x08,0x00, /* @ - G */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* H - O */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* P - W */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* X - _ */ + 0x00,0x08,0x08,0x08,0x08,0x08,0x08,0x00, /* ` - g */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* h - o */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* p - w */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* x -127 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 128-135 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 136-143 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 144-151 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 152-159 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 160-167 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 168-175 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 176-183 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 184-191 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 192-199 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 200-207 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 208-215 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 216-223 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 224-231 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 232-239 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 240-247 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00};/* 248-255 */ + +#else + +/* This is the "abnormal" case, for EBCDIC systems not running in UTF-8 mode. */ + +static const pcre_uint8 digitab[] = + { + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 0- 7 0 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 8- 15 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 16- 23 10 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 24- 31 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 32- 39 20 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 40- 47 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 48- 55 30 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 56- 63 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* - 71 40 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 72- | */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* & - 87 50 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 88- 95 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* - -103 60 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 104- ? */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 112-119 70 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 120- " */ + 0x00,0x08,0x08,0x08,0x08,0x08,0x08,0x00, /* 128- g 80 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* h -143 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 144- p 90 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* q -159 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 160- x A0 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* y -175 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* ^ -183 B0 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 184-191 */ + 0x00,0x08,0x08,0x08,0x08,0x08,0x08,0x00, /* { - G C0 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* H -207 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* } - P D0 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* Q -223 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* \ - X E0 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* Y -239 */ + 0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /* 0 - 7 F0 */ + 0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/* 8 -255 */ + +static const pcre_uint8 ebcdic_chartab[] = { /* chartable partial dup */ + 0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /* 0- 7 */ + 0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /* 8- 15 */ + 0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /* 16- 23 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 24- 31 */ + 0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /* 32- 39 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 40- 47 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 48- 55 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 56- 63 */ + 0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* - 71 */ + 0x00,0x00,0x00,0x80,0x00,0x80,0x80,0x80, /* 72- | */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* & - 87 */ + 0x00,0x00,0x00,0x80,0x80,0x80,0x00,0x00, /* 88- 95 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* - -103 */ + 0x00,0x00,0x00,0x00,0x00,0x10,0x00,0x80, /* 104- ? */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 112-119 */ + 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 120- " */ + 0x00,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x12, /* 128- g */ + 0x12,0x12,0x00,0x00,0x00,0x00,0x00,0x00, /* h -143 */ + 0x00,0x12,0x12,0x12,0x12,0x12,0x12,0x12, /* 144- p */ + 0x12,0x12,0x00,0x00,0x00,0x00,0x00,0x00, /* q -159 */ + 0x00,0x00,0x12,0x12,0x12,0x12,0x12,0x12, /* 160- x */ + 0x12,0x12,0x00,0x00,0x00,0x00,0x00,0x00, /* y -175 */ + 0x80,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* ^ -183 */ + 0x00,0x00,0x80,0x00,0x00,0x00,0x00,0x00, /* 184-191 */ + 0x80,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x12, /* { - G */ + 0x12,0x12,0x00,0x00,0x00,0x00,0x00,0x00, /* H -207 */ + 0x00,0x12,0x12,0x12,0x12,0x12,0x12,0x12, /* } - P */ + 0x12,0x12,0x00,0x00,0x00,0x00,0x00,0x00, /* Q -223 */ + 0x00,0x00,0x12,0x12,0x12,0x12,0x12,0x12, /* \ - X */ + 0x12,0x12,0x00,0x00,0x00,0x00,0x00,0x00, /* Y -239 */ + 0x1c,0x1c,0x1c,0x1c,0x1c,0x1c,0x1c,0x1c, /* 0 - 7 */ + 0x1c,0x1c,0x00,0x00,0x00,0x00,0x00,0x00};/* 8 -255 */ +#endif + + +/* Definition to allow mutual recursion */ + +static BOOL + compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL, int, int, + int *, int *, branch_chain *, compile_data *, int *); + + + +/************************************************* +* Find an error text * +*************************************************/ + +/* The error texts are now all in one long string, to save on relocations. As +some of the text is of unknown length, we can't use a table of offsets. +Instead, just count through the strings. This is not a performance issue +because it happens only when there has been a compilation error. + +Argument: the error number +Returns: pointer to the error string +*/ + +static const char * +find_error_text(int n) +{ +const char *s = error_texts; +for (; n > 0; n--) + { + while (*s++ != 0) {}; + if (*s == 0) return "Error text not found (please report)"; + } +return s; +} + + +/************************************************* +* Expand the workspace * +*************************************************/ + +/* This function is called during the second compiling phase, if the number of +forward references fills the existing workspace, which is originally a block on +the stack. A larger block is obtained from malloc() unless the ultimate limit +has been reached or the increase will be rather small. + +Argument: pointer to the compile data block +Returns: 0 if all went well, else an error number +*/ + +static int +expand_workspace(compile_data *cd) +{ +pcre_uchar *newspace; +int newsize = cd->workspace_size * 2; + +if (newsize > COMPILE_WORK_SIZE_MAX) newsize = COMPILE_WORK_SIZE_MAX; +if (cd->workspace_size >= COMPILE_WORK_SIZE_MAX || + newsize - cd->workspace_size < WORK_SIZE_SAFETY_MARGIN) + return ERR72; + +newspace = (PUBL(malloc))(IN_UCHARS(newsize)); +if (newspace == NULL) return ERR21; +memcpy(newspace, cd->start_workspace, cd->workspace_size * sizeof(pcre_uchar)); +cd->hwm = (pcre_uchar *)newspace + (cd->hwm - cd->start_workspace); +if (cd->workspace_size > COMPILE_WORK_SIZE) + (PUBL(free))((void *)cd->start_workspace); +cd->start_workspace = newspace; +cd->workspace_size = newsize; +return 0; +} + + + +/************************************************* +* Check for counted repeat * +*************************************************/ + +/* This function is called when a '{' is encountered in a place where it might +start a quantifier. It looks ahead to see if it really is a quantifier or not. +It is only a quantifier if it is one of the forms {ddd} {ddd,} or {ddd,ddd} +where the ddds are digits. + +Arguments: + p pointer to the first char after '{' + +Returns: TRUE or FALSE +*/ + +static BOOL +is_counted_repeat(const pcre_uchar *p) +{ +if (!IS_DIGIT(*p)) return FALSE; +p++; +while (IS_DIGIT(*p)) p++; +if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE; + +if (*p++ != CHAR_COMMA) return FALSE; +if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE; + +if (!IS_DIGIT(*p)) return FALSE; +p++; +while (IS_DIGIT(*p)) p++; + +return (*p == CHAR_RIGHT_CURLY_BRACKET); +} + + + +/************************************************* +* Handle escapes * +*************************************************/ + +/* This function is called when a \ has been encountered. It either returns a +positive value for a simple escape such as \n, or a negative value which +encodes one of the more complicated things such as \d. A backreference to group +n is returned as -(ESC_REF + n); ESC_REF is the highest ESC_xxx macro. When +UTF-8 is enabled, a positive value greater than 255 may be returned. On entry, +ptr is pointing at the \. On exit, it is on the final character of the escape +sequence. + +Arguments: + ptrptr points to the pattern position pointer + errorcodeptr points to the errorcode variable + bracount number of previous extracting brackets + options the options bits + isclass TRUE if inside a character class + +Returns: zero or positive => a data character + negative => a special escape sequence + on error, errorcodeptr is set +*/ + +static int +check_escape(const pcre_uchar **ptrptr, int *errorcodeptr, int bracount, + int options, BOOL isclass) +{ +/* PCRE_UTF16 has the same value as PCRE_UTF8. */ +BOOL utf = (options & PCRE_UTF8) != 0; +const pcre_uchar *ptr = *ptrptr + 1; +pcre_int32 c; +int i; + +GETCHARINCTEST(c, ptr); /* Get character value, increment pointer */ +ptr--; /* Set pointer back to the last byte */ + +/* If backslash is at the end of the pattern, it's an error. */ + +if (c == 0) *errorcodeptr = ERR1; + +/* Non-alphanumerics are literals. For digits or letters, do an initial lookup +in a table. A non-zero result is something that can be returned immediately. +Otherwise further processing may be required. */ + +#ifndef EBCDIC /* ASCII/UTF-8 coding */ +/* Not alphanumeric */ +else if (c < CHAR_0 || c > CHAR_z) {} +else if ((i = escapes[c - CHAR_0]) != 0) c = i; + +#else /* EBCDIC coding */ +/* Not alphanumeric */ +else if (c < 'a' || (!MAX_255(c) || (ebcdic_chartab[c] & 0x0E) == 0)) {} +else if ((i = escapes[c - 0x48]) != 0) c = i; +#endif + +/* Escapes that need further processing, or are illegal. */ + +else + { + const pcre_uchar *oldptr; + BOOL braced, negated; + + switch (c) + { + /* A number of Perl escapes are not handled by PCRE. We give an explicit + error. */ + + case CHAR_l: + case CHAR_L: + *errorcodeptr = ERR37; + break; + + case CHAR_u: + if ((options & PCRE_JAVASCRIPT_COMPAT) != 0) + { + /* In JavaScript, \u must be followed by four hexadecimal numbers. + Otherwise it is a lowercase u letter. */ + if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0 + && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0 + && MAX_255(ptr[3]) && (digitab[ptr[3]] & ctype_xdigit) != 0 + && MAX_255(ptr[4]) && (digitab[ptr[4]] & ctype_xdigit) != 0) + { + c = 0; + for (i = 0; i < 4; ++i) + { + register int cc = *(++ptr); +#ifndef EBCDIC /* ASCII/UTF-8 coding */ + if (cc >= CHAR_a) cc -= 32; /* Convert to upper case */ + c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10)); +#else /* EBCDIC coding */ + if (cc >= CHAR_a && cc <= CHAR_z) cc += 64; /* Convert to upper case */ + c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10)); +#endif + } + +#ifdef COMPILE_PCRE8 + if (c > (utf ? 0x10ffff : 0xff)) +#else +#ifdef COMPILE_PCRE16 + if (c > (utf ? 0x10ffff : 0xffff)) +#endif +#endif + { + *errorcodeptr = ERR76; + } + else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73; + } + } + else + *errorcodeptr = ERR37; + break; + + case CHAR_U: + /* In JavaScript, \U is an uppercase U letter. */ + if ((options & PCRE_JAVASCRIPT_COMPAT) == 0) *errorcodeptr = ERR37; + break; + + /* In a character class, \g is just a literal "g". Outside a character + class, \g must be followed by one of a number of specific things: + + (1) A number, either plain or braced. If positive, it is an absolute + backreference. If negative, it is a relative backreference. This is a Perl + 5.10 feature. + + (2) Perl 5.10 also supports \g{name} as a reference to a named group. This + is part of Perl's movement towards a unified syntax for back references. As + this is synonymous with \k{name}, we fudge it up by pretending it really + was \k. + + (3) For Oniguruma compatibility we also support \g followed by a name or a + number either in angle brackets or in single quotes. However, these are + (possibly recursive) subroutine calls, _not_ backreferences. Just return + the -ESC_g code (cf \k). */ + + case CHAR_g: + if (isclass) break; + if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE) + { + c = -ESC_g; + break; + } + + /* Handle the Perl-compatible cases */ + + if (ptr[1] == CHAR_LEFT_CURLY_BRACKET) + { + const pcre_uchar *p; + for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++) + if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break; + if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET) + { + c = -ESC_k; + break; + } + braced = TRUE; + ptr++; + } + else braced = FALSE; + + if (ptr[1] == CHAR_MINUS) + { + negated = TRUE; + ptr++; + } + else negated = FALSE; + + /* The integer range is limited by the machine's int representation. */ + c = 0; + while (IS_DIGIT(ptr[1])) + { + if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */ + { + c = -1; + break; + } + c = c * 10 + *(++ptr) - CHAR_0; + } + if (((unsigned int)c) > INT_MAX) /* Integer overflow */ + { + while (IS_DIGIT(ptr[1])) + ptr++; + *errorcodeptr = ERR61; + break; + } + + if (braced && *(++ptr) != CHAR_RIGHT_CURLY_BRACKET) + { + *errorcodeptr = ERR57; + break; + } + + if (c == 0) + { + *errorcodeptr = ERR58; + break; + } + + if (negated) + { + if (c > bracount) + { + *errorcodeptr = ERR15; + break; + } + c = bracount - (c - 1); + } + + c = -(ESC_REF + c); + break; + + /* The handling of escape sequences consisting of a string of digits + starting with one that is not zero is not straightforward. By experiment, + the way Perl works seems to be as follows: + + Outside a character class, the digits are read as a decimal number. If the + number is less than 10, or if there are that many previous extracting + left brackets, then it is a back reference. Otherwise, up to three octal + digits are read to form an escaped byte. Thus \123 is likely to be octal + 123 (cf \0123, which is octal 012 followed by the literal 3). If the octal + value is greater than 377, the least significant 8 bits are taken. Inside a + character class, \ followed by a digit is always an octal number. */ + + case CHAR_1: case CHAR_2: case CHAR_3: case CHAR_4: case CHAR_5: + case CHAR_6: case CHAR_7: case CHAR_8: case CHAR_9: + + if (!isclass) + { + oldptr = ptr; + /* The integer range is limited by the machine's int representation. */ + c -= CHAR_0; + while (IS_DIGIT(ptr[1])) + { + if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */ + { + c = -1; + break; + } + c = c * 10 + *(++ptr) - CHAR_0; + } + if (((unsigned int)c) > INT_MAX) /* Integer overflow */ + { + while (IS_DIGIT(ptr[1])) + ptr++; + *errorcodeptr = ERR61; + break; + } + if (c < 10 || c <= bracount) + { + c = -(ESC_REF + c); + break; + } + ptr = oldptr; /* Put the pointer back and fall through */ + } + + /* Handle an octal number following \. If the first digit is 8 or 9, Perl + generates a binary zero byte and treats the digit as a following literal. + Thus we have to pull back the pointer by one. */ + + if ((c = *ptr) >= CHAR_8) + { + ptr--; + c = 0; + break; + } + + /* \0 always starts an octal number, but we may drop through to here with a + larger first octal digit. The original code used just to take the least + significant 8 bits of octal numbers (I think this is what early Perls used + to do). Nowadays we allow for larger numbers in UTF-8 mode and 16-bit mode, + but no more than 3 octal digits. */ + + case CHAR_0: + c -= CHAR_0; + while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7) + c = c * 8 + *(++ptr) - CHAR_0; +#ifdef COMPILE_PCRE8 + if (!utf && c > 0xff) *errorcodeptr = ERR51; +#endif + break; + + /* \x is complicated. \x{ddd} is a character number which can be greater + than 0xff in utf or non-8bit mode, but only if the ddd are hex digits. + If not, { is treated as a data character. */ + + case CHAR_x: + if ((options & PCRE_JAVASCRIPT_COMPAT) != 0) + { + /* In JavaScript, \x must be followed by two hexadecimal numbers. + Otherwise it is a lowercase x letter. */ + if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0 + && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0) + { + c = 0; + for (i = 0; i < 2; ++i) + { + register int cc = *(++ptr); +#ifndef EBCDIC /* ASCII/UTF-8 coding */ + if (cc >= CHAR_a) cc -= 32; /* Convert to upper case */ + c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10)); +#else /* EBCDIC coding */ + if (cc >= CHAR_a && cc <= CHAR_z) cc += 64; /* Convert to upper case */ + c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10)); +#endif + } + } + break; + } + + if (ptr[1] == CHAR_LEFT_CURLY_BRACKET) + { + const pcre_uchar *pt = ptr + 2; + + c = 0; + while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) + { + register int cc = *pt++; + if (c == 0 && cc == CHAR_0) continue; /* Leading zeroes */ + +#ifndef EBCDIC /* ASCII/UTF-8 coding */ + if (cc >= CHAR_a) cc -= 32; /* Convert to upper case */ + c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10)); +#else /* EBCDIC coding */ + if (cc >= CHAR_a && cc <= CHAR_z) cc += 64; /* Convert to upper case */ + c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10)); +#endif + +#ifdef COMPILE_PCRE8 + if (c > (utf ? 0x10ffff : 0xff)) { c = -1; break; } +#else +#ifdef COMPILE_PCRE16 + if (c > (utf ? 0x10ffff : 0xffff)) { c = -1; break; } +#endif +#endif + } + + if (c < 0) + { + while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++; + *errorcodeptr = ERR34; + } + + if (*pt == CHAR_RIGHT_CURLY_BRACKET) + { + if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73; + ptr = pt; + break; + } + + /* If the sequence of hex digits does not end with '}', then we don't + recognize this construct; fall through to the normal \x handling. */ + } + + /* Read just a single-byte hex-defined char */ + + c = 0; + while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0) + { + int cc; /* Some compilers don't like */ + cc = *(++ptr); /* ++ in initializers */ +#ifndef EBCDIC /* ASCII/UTF-8 coding */ + if (cc >= CHAR_a) cc -= 32; /* Convert to upper case */ + c = c * 16 + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10)); +#else /* EBCDIC coding */ + if (cc <= CHAR_z) cc += 64; /* Convert to upper case */ + c = c * 16 + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10)); +#endif + } + break; + + /* For \c, a following letter is upper-cased; then the 0x40 bit is flipped. + An error is given if the byte following \c is not an ASCII character. This + coding is ASCII-specific, but then the whole concept of \cx is + ASCII-specific. (However, an EBCDIC equivalent has now been added.) */ + + case CHAR_c: + c = *(++ptr); + if (c == 0) + { + *errorcodeptr = ERR2; + break; + } +#ifndef EBCDIC /* ASCII/UTF-8 coding */ + if (c > 127) /* Excludes all non-ASCII in either mode */ + { + *errorcodeptr = ERR68; + break; + } + if (c >= CHAR_a && c <= CHAR_z) c -= 32; + c ^= 0x40; +#else /* EBCDIC coding */ + if (c >= CHAR_a && c <= CHAR_z) c += 64; + c ^= 0xC0; +#endif + break; + + /* PCRE_EXTRA enables extensions to Perl in the matter of escapes. Any + other alphanumeric following \ is an error if PCRE_EXTRA was set; + otherwise, for Perl compatibility, it is a literal. This code looks a bit + odd, but there used to be some cases other than the default, and there may + be again in future, so I haven't "optimized" it. */ + + default: + if ((options & PCRE_EXTRA) != 0) switch(c) + { + default: + *errorcodeptr = ERR3; + break; + } + break; + } + } + +/* Perl supports \N{name} for character names, as well as plain \N for "not +newline". PCRE does not support \N{name}. However, it does support +quantification such as \N{2,3}. */ + +if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET && + !is_counted_repeat(ptr+2)) + *errorcodeptr = ERR37; + +/* If PCRE_UCP is set, we change the values for \d etc. */ + +if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w) + c -= (ESC_DU - ESC_D); + +/* Set the pointer to the final character before returning. */ + +*ptrptr = ptr; +return c; +} + + + +#ifdef SUPPORT_UCP +/************************************************* +* Handle \P and \p * +*************************************************/ + +/* This function is called after \P or \p has been encountered, provided that +PCRE is compiled with support for Unicode properties. On entry, ptrptr is +pointing at the P or p. On exit, it is pointing at the final character of the +escape sequence. + +Argument: + ptrptr points to the pattern position pointer + negptr points to a boolean that is set TRUE for negation else FALSE + dptr points to an int that is set to the detailed property value + errorcodeptr points to the error code variable + +Returns: type value from ucp_type_table, or -1 for an invalid type +*/ + +static int +get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr) +{ +int c, i, bot, top; +const pcre_uchar *ptr = *ptrptr; +pcre_uchar name[32]; + +c = *(++ptr); +if (c == 0) goto ERROR_RETURN; + +*negptr = FALSE; + +/* \P or \p can be followed by a name in {}, optionally preceded by ^ for +negation. */ + +if (c == CHAR_LEFT_CURLY_BRACKET) + { + if (ptr[1] == CHAR_CIRCUMFLEX_ACCENT) + { + *negptr = TRUE; + ptr++; + } + for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++) + { + c = *(++ptr); + if (c == 0) goto ERROR_RETURN; + if (c == CHAR_RIGHT_CURLY_BRACKET) break; + name[i] = c; + } + if (c != CHAR_RIGHT_CURLY_BRACKET) goto ERROR_RETURN; + name[i] = 0; + } + +/* Otherwise there is just one following character */ + +else + { + name[0] = c; + name[1] = 0; + } + +*ptrptr = ptr; + +/* Search for a recognized property name using binary chop */ + +bot = 0; +top = PRIV(utt_size); + +while (bot < top) + { + i = (bot + top) >> 1; + c = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset); + if (c == 0) + { + *dptr = PRIV(utt)[i].value; + return PRIV(utt)[i].type; + } + if (c > 0) bot = i + 1; else top = i; + } + +*errorcodeptr = ERR47; +*ptrptr = ptr; +return -1; + +ERROR_RETURN: +*errorcodeptr = ERR46; +*ptrptr = ptr; +return -1; +} +#endif + + + + +/************************************************* +* Read repeat counts * +*************************************************/ + +/* Read an item of the form {n,m} and return the values. This is called only +after is_counted_repeat() has confirmed that a repeat-count quantifier exists, +so the syntax is guaranteed to be correct, but we need to check the values. + +Arguments: + p pointer to first char after '{' + minp pointer to int for min + maxp pointer to int for max + returned as -1 if no max + errorcodeptr points to error code variable + +Returns: pointer to '}' on success; + current ptr on error, with errorcodeptr set non-zero +*/ + +static const pcre_uchar * +read_repeat_counts(const pcre_uchar *p, int *minp, int *maxp, int *errorcodeptr) +{ +int min = 0; +int max = -1; + +/* Read the minimum value and do a paranoid check: a negative value indicates +an integer overflow. */ + +while (IS_DIGIT(*p)) min = min * 10 + *p++ - CHAR_0; +if (min < 0 || min > 65535) + { + *errorcodeptr = ERR5; + return p; + } + +/* Read the maximum value if there is one, and again do a paranoid on its size. +Also, max must not be less than min. */ + +if (*p == CHAR_RIGHT_CURLY_BRACKET) max = min; else + { + if (*(++p) != CHAR_RIGHT_CURLY_BRACKET) + { + max = 0; + while(IS_DIGIT(*p)) max = max * 10 + *p++ - CHAR_0; + if (max < 0 || max > 65535) + { + *errorcodeptr = ERR5; + return p; + } + if (max < min) + { + *errorcodeptr = ERR4; + return p; + } + } + } + +/* Fill in the required variables, and pass back the pointer to the terminating +'}'. */ + +*minp = min; +*maxp = max; +return p; +} + + + +/************************************************* +* Subroutine for finding forward reference * +*************************************************/ + +/* This recursive function is called only from find_parens() below. The +top-level call starts at the beginning of the pattern. All other calls must +start at a parenthesis. It scans along a pattern's text looking for capturing +subpatterns, and counting them. If it finds a named pattern that matches the +name it is given, it returns its number. Alternatively, if the name is NULL, it +returns when it reaches a given numbered subpattern. Recursion is used to keep +track of subpatterns that reset the capturing group numbers - the (?| feature. + +This function was originally called only from the second pass, in which we know +that if (?< or (?' or (?P< is encountered, the name will be correctly +terminated because that is checked in the first pass. There is now one call to +this function in the first pass, to check for a recursive back reference by +name (so that we can make the whole group atomic). In this case, we need check +only up to the current position in the pattern, and that is still OK because +and previous occurrences will have been checked. To make this work, the test +for "end of pattern" is a check against cd->end_pattern in the main loop, +instead of looking for a binary zero. This means that the special first-pass +call can adjust cd->end_pattern temporarily. (Checks for binary zero while +processing items within the loop are OK, because afterwards the main loop will +terminate.) + +Arguments: + ptrptr address of the current character pointer (updated) + cd compile background data + name name to seek, or NULL if seeking a numbered subpattern + lorn name length, or subpattern number if name is NULL + xmode TRUE if we are in /x mode + utf TRUE if we are in UTF-8 / UTF-16 mode + count pointer to the current capturing subpattern number (updated) + +Returns: the number of the named subpattern, or -1 if not found +*/ + +static int +find_parens_sub(pcre_uchar **ptrptr, compile_data *cd, const pcre_uchar *name, int lorn, + BOOL xmode, BOOL utf, int *count) +{ +pcre_uchar *ptr = *ptrptr; +int start_count = *count; +int hwm_count = start_count; +BOOL dup_parens = FALSE; + +/* If the first character is a parenthesis, check on the type of group we are +dealing with. The very first call may not start with a parenthesis. */ + +if (ptr[0] == CHAR_LEFT_PARENTHESIS) + { + /* Handle specials such as (*SKIP) or (*UTF8) etc. */ + + if (ptr[1] == CHAR_ASTERISK) ptr += 2; + + /* Handle a normal, unnamed capturing parenthesis. */ + + else if (ptr[1] != CHAR_QUESTION_MARK) + { + *count += 1; + if (name == NULL && *count == lorn) return *count; + ptr++; + } + + /* All cases now have (? at the start. Remember when we are in a group + where the parenthesis numbers are duplicated. */ + + else if (ptr[2] == CHAR_VERTICAL_LINE) + { + ptr += 3; + dup_parens = TRUE; + } + + /* Handle comments; all characters are allowed until a ket is reached. */ + + else if (ptr[2] == CHAR_NUMBER_SIGN) + { + for (ptr += 3; *ptr != 0; ptr++) if (*ptr == CHAR_RIGHT_PARENTHESIS) break; + goto FAIL_EXIT; + } + + /* Handle a condition. If it is an assertion, just carry on so that it + is processed as normal. If not, skip to the closing parenthesis of the + condition (there can't be any nested parens). */ + + else if (ptr[2] == CHAR_LEFT_PARENTHESIS) + { + ptr += 2; + if (ptr[1] != CHAR_QUESTION_MARK) + { + while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++; + if (*ptr != 0) ptr++; + } + } + + /* Start with (? but not a condition. */ + + else + { + ptr += 2; + if (*ptr == CHAR_P) ptr++; /* Allow optional P */ + + /* We have to disambiguate (? for named groups */ + + if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK && + ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE) + { + int term; + const pcre_uchar *thisname; + *count += 1; + if (name == NULL && *count == lorn) return *count; + term = *ptr++; + if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN; + thisname = ptr; + while (*ptr != term) ptr++; + if (name != NULL && lorn == ptr - thisname && + STRNCMP_UC_UC(name, thisname, lorn) == 0) + return *count; + term++; + } + } + } + +/* Past any initial parenthesis handling, scan for parentheses or vertical +bars. Stop if we get to cd->end_pattern. Note that this is important for the +first-pass call when this value is temporarily adjusted to stop at the current +position. So DO NOT change this to a test for binary zero. */ + +for (; ptr < cd->end_pattern; ptr++) + { + /* Skip over backslashed characters and also entire \Q...\E */ + + if (*ptr == CHAR_BACKSLASH) + { + if (*(++ptr) == 0) goto FAIL_EXIT; + if (*ptr == CHAR_Q) for (;;) + { + while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {}; + if (*ptr == 0) goto FAIL_EXIT; + if (*(++ptr) == CHAR_E) break; + } + continue; + } + + /* Skip over character classes; this logic must be similar to the way they + are handled for real. If the first character is '^', skip it. Also, if the + first few characters (either before or after ^) are \Q\E or \E we skip them + too. This makes for compatibility with Perl. Note the use of STR macros to + encode "Q\\E" so that it works in UTF-8 on EBCDIC platforms. */ + + if (*ptr == CHAR_LEFT_SQUARE_BRACKET) + { + BOOL negate_class = FALSE; + for (;;) + { + if (ptr[1] == CHAR_BACKSLASH) + { + if (ptr[2] == CHAR_E) + ptr+= 2; + else if (STRNCMP_UC_C8(ptr + 2, + STR_Q STR_BACKSLASH STR_E, 3) == 0) + ptr += 4; + else + break; + } + else if (!negate_class && ptr[1] == CHAR_CIRCUMFLEX_ACCENT) + { + negate_class = TRUE; + ptr++; + } + else break; + } + + /* If the next character is ']', it is a data character that must be + skipped, except in JavaScript compatibility mode. */ + + if (ptr[1] == CHAR_RIGHT_SQUARE_BRACKET && + (cd->external_options & PCRE_JAVASCRIPT_COMPAT) == 0) + ptr++; + + while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET) + { + if (*ptr == 0) return -1; + if (*ptr == CHAR_BACKSLASH) + { + if (*(++ptr) == 0) goto FAIL_EXIT; + if (*ptr == CHAR_Q) for (;;) + { + while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {}; + if (*ptr == 0) goto FAIL_EXIT; + if (*(++ptr) == CHAR_E) break; + } + continue; + } + } + continue; + } + + /* Skip comments in /x mode */ + + if (xmode && *ptr == CHAR_NUMBER_SIGN) + { + ptr++; + while (*ptr != 0) + { + if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; } + ptr++; +#ifdef SUPPORT_UTF + if (utf) FORWARDCHAR(ptr); +#endif + } + if (*ptr == 0) goto FAIL_EXIT; + continue; + } + + /* Check for the special metacharacters */ + + if (*ptr == CHAR_LEFT_PARENTHESIS) + { + int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count); + if (rc > 0) return rc; + if (*ptr == 0) goto FAIL_EXIT; + } + + else if (*ptr == CHAR_RIGHT_PARENTHESIS) + { + if (dup_parens && *count < hwm_count) *count = hwm_count; + goto FAIL_EXIT; + } + + else if (*ptr == CHAR_VERTICAL_LINE && dup_parens) + { + if (*count > hwm_count) hwm_count = *count; + *count = start_count; + } + } + +FAIL_EXIT: +*ptrptr = ptr; +return -1; +} + + + + +/************************************************* +* Find forward referenced subpattern * +*************************************************/ + +/* This function scans along a pattern's text looking for capturing +subpatterns, and counting them. If it finds a named pattern that matches the +name it is given, it returns its number. Alternatively, if the name is NULL, it +returns when it reaches a given numbered subpattern. This is used for forward +references to subpatterns. We used to be able to start this scan from the +current compiling point, using the current count value from cd->bracount, and +do it all in a single loop, but the addition of the possibility of duplicate +subpattern numbers means that we have to scan from the very start, in order to +take account of such duplicates, and to use a recursive function to keep track +of the different types of group. + +Arguments: + cd compile background data + name name to seek, or NULL if seeking a numbered subpattern + lorn name length, or subpattern number if name is NULL + xmode TRUE if we are in /x mode + utf TRUE if we are in UTF-8 / UTF-16 mode + +Returns: the number of the found subpattern, or -1 if not found +*/ + +static int +find_parens(compile_data *cd, const pcre_uchar *name, int lorn, BOOL xmode, + BOOL utf) +{ +pcre_uchar *ptr = (pcre_uchar *)cd->start_pattern; +int count = 0; +int rc; + +/* If the pattern does not start with an opening parenthesis, the first call +to find_parens_sub() will scan right to the end (if necessary). However, if it +does start with a parenthesis, find_parens_sub() will return when it hits the +matching closing parens. That is why we have to have a loop. */ + +for (;;) + { + rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count); + if (rc > 0 || *ptr++ == 0) break; + } + +return rc; +} + + + + +/************************************************* +* Find first significant op code * +*************************************************/ + +/* This is called by several functions that scan a compiled expression looking +for a fixed first character, or an anchoring op code etc. It skips over things +that do not influence this. For some calls, it makes sense to skip negative +forward and all backward assertions, and also the \b assertion; for others it +does not. + +Arguments: + code pointer to the start of the group + skipassert TRUE if certain assertions are to be skipped + +Returns: pointer to the first significant opcode +*/ + +static const pcre_uchar* +first_significant_code(const pcre_uchar *code, BOOL skipassert) +{ +for (;;) + { + switch ((int)*code) + { + case OP_ASSERT_NOT: + case OP_ASSERTBACK: + case OP_ASSERTBACK_NOT: + if (!skipassert) return code; + do code += GET(code, 1); while (*code == OP_ALT); + code += PRIV(OP_lengths)[*code]; + break; + + case OP_WORD_BOUNDARY: + case OP_NOT_WORD_BOUNDARY: + if (!skipassert) return code; + /* Fall through */ + + case OP_CALLOUT: + case OP_CREF: + case OP_NCREF: + case OP_RREF: + case OP_NRREF: + case OP_DEF: + code += PRIV(OP_lengths)[*code]; + break; + + default: + return code; + } + } +/* Control never reaches here */ +} + + + + +/************************************************* +* Find the fixed length of a branch * +*************************************************/ + +/* Scan a branch and compute the fixed length of subject that will match it, +if the length is fixed. This is needed for dealing with backward assertions. +In UTF8 mode, the result is in characters rather than bytes. The branch is +temporarily terminated with OP_END when this function is called. + +This function is called when a backward assertion is encountered, so that if it +fails, the error message can point to the correct place in the pattern. +However, we cannot do this when the assertion contains subroutine calls, +because they can be forward references. We solve this by remembering this case +and doing the check at the end; a flag specifies which mode we are running in. + +Arguments: + code points to the start of the pattern (the bracket) + utf TRUE in UTF-8 / UTF-16 mode + atend TRUE if called when the pattern is complete + cd the "compile data" structure + +Returns: the fixed length, + or -1 if there is no fixed length, + or -2 if \C was encountered (in UTF-8 mode only) + or -3 if an OP_RECURSE item was encountered and atend is FALSE + or -4 if an unknown opcode was encountered (internal error) +*/ + +static int +find_fixedlength(pcre_uchar *code, BOOL utf, BOOL atend, compile_data *cd) +{ +int length = -1; + +register int branchlength = 0; +register pcre_uchar *cc = code + 1 + LINK_SIZE; + +/* Scan along the opcodes for this branch. If we get to the end of the +branch, check the length against that of the other branches. */ + +for (;;) + { + int d; + pcre_uchar *ce, *cs; + register int op = *cc; + + switch (op) + { + /* We only need to continue for OP_CBRA (normal capturing bracket) and + OP_BRA (normal non-capturing bracket) because the other variants of these + opcodes are all concerned with unlimited repeated groups, which of course + are not of fixed length. */ + + case OP_CBRA: + case OP_BRA: + case OP_ONCE: + case OP_ONCE_NC: + case OP_COND: + d = find_fixedlength(cc + ((op == OP_CBRA)? IMM2_SIZE : 0), utf, atend, cd); + if (d < 0) return d; + branchlength += d; + do cc += GET(cc, 1); while (*cc == OP_ALT); + cc += 1 + LINK_SIZE; + break; + + /* Reached end of a branch; if it's a ket it is the end of a nested call. + If it's ALT it is an alternation in a nested call. An ACCEPT is effectively + an ALT. If it is END it's the end of the outer call. All can be handled by + the same code. Note that we must not include the OP_KETRxxx opcodes here, + because they all imply an unlimited repeat. */ + + case OP_ALT: + case OP_KET: + case OP_END: + case OP_ACCEPT: + case OP_ASSERT_ACCEPT: + if (length < 0) length = branchlength; + else if (length != branchlength) return -1; + if (*cc != OP_ALT) return length; + cc += 1 + LINK_SIZE; + branchlength = 0; + break; + + /* A true recursion implies not fixed length, but a subroutine call may + be OK. If the subroutine is a forward reference, we can't deal with + it until the end of the pattern, so return -3. */ + + case OP_RECURSE: + if (!atend) return -3; + cs = ce = (pcre_uchar *)cd->start_code + GET(cc, 1); /* Start subpattern */ + do ce += GET(ce, 1); while (*ce == OP_ALT); /* End subpattern */ + if (cc > cs && cc < ce) return -1; /* Recursion */ + d = find_fixedlength(cs + IMM2_SIZE, utf, atend, cd); + if (d < 0) return d; + branchlength += d; + cc += 1 + LINK_SIZE; + break; + + /* Skip over assertive subpatterns */ + + case OP_ASSERT: + case OP_ASSERT_NOT: + case OP_ASSERTBACK: + case OP_ASSERTBACK_NOT: + do cc += GET(cc, 1); while (*cc == OP_ALT); + cc += PRIV(OP_lengths)[*cc]; + break; + + /* Skip over things that don't match chars */ + + case OP_MARK: + case OP_PRUNE_ARG: + case OP_SKIP_ARG: + case OP_THEN_ARG: + cc += cc[1] + PRIV(OP_lengths)[*cc]; + break; + + case OP_CALLOUT: + case OP_CIRC: + case OP_CIRCM: + case OP_CLOSE: + case OP_COMMIT: + case OP_CREF: + case OP_DEF: + case OP_DOLL: + case OP_DOLLM: + case OP_EOD: + case OP_EODN: + case OP_FAIL: + case OP_NCREF: + case OP_NRREF: + case OP_NOT_WORD_BOUNDARY: + case OP_PRUNE: + case OP_REVERSE: + case OP_RREF: + case OP_SET_SOM: + case OP_SKIP: + case OP_SOD: + case OP_SOM: + case OP_THEN: + case OP_WORD_BOUNDARY: + cc += PRIV(OP_lengths)[*cc]; + break; + + /* Handle literal characters */ + + case OP_CHAR: + case OP_CHARI: + case OP_NOT: + case OP_NOTI: + branchlength++; + cc += 2; +#ifdef SUPPORT_UTF + if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); +#endif + break; + + /* Handle exact repetitions. The count is already in characters, but we + need to skip over a multibyte character in UTF8 mode. */ + + case OP_EXACT: + case OP_EXACTI: + case OP_NOTEXACT: + case OP_NOTEXACTI: + branchlength += GET2(cc,1); + cc += 2 + IMM2_SIZE; +#ifdef SUPPORT_UTF + if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); +#endif + break; + + case OP_TYPEEXACT: + branchlength += GET2(cc,1); + if (cc[1 + IMM2_SIZE] == OP_PROP || cc[1 + IMM2_SIZE] == OP_NOTPROP) cc += 2; + cc += 1 + IMM2_SIZE + 1; + break; + + /* Handle single-char matchers */ + + case OP_PROP: + case OP_NOTPROP: + cc += 2; + /* Fall through */ + + case OP_HSPACE: + case OP_VSPACE: + case OP_NOT_HSPACE: + case OP_NOT_VSPACE: + case OP_NOT_DIGIT: + case OP_DIGIT: + case OP_NOT_WHITESPACE: + case OP_WHITESPACE: + case OP_NOT_WORDCHAR: + case OP_WORDCHAR: + case OP_ANY: + case OP_ALLANY: + branchlength++; + cc++; + break; + + /* The single-byte matcher isn't allowed. This only happens in UTF-8 mode; + otherwise \C is coded as OP_ALLANY. */ + + case OP_ANYBYTE: + return -2; + + /* Check a class for variable quantification */ + +#if defined SUPPORT_UTF || defined COMPILE_PCRE16 + case OP_XCLASS: + cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS]; + /* Fall through */ +#endif + + case OP_CLASS: + case OP_NCLASS: + cc += PRIV(OP_lengths)[OP_CLASS]; + + switch (*cc) + { + case OP_CRPLUS: + case OP_CRMINPLUS: + case OP_CRSTAR: + case OP_CRMINSTAR: + case OP_CRQUERY: + case OP_CRMINQUERY: + return -1; + + case OP_CRRANGE: + case OP_CRMINRANGE: + if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1; + branchlength += GET2(cc,1); + cc += 1 + 2 * IMM2_SIZE; + break; + + default: + branchlength++; + } + break; + + /* Anything else is variable length */ + + case OP_ANYNL: + case OP_BRAMINZERO: + case OP_BRAPOS: + case OP_BRAPOSZERO: + case OP_BRAZERO: + case OP_CBRAPOS: + case OP_EXTUNI: + case OP_KETRMAX: + case OP_KETRMIN: + case OP_KETRPOS: + case OP_MINPLUS: + case OP_MINPLUSI: + case OP_MINQUERY: + case OP_MINQUERYI: + case OP_MINSTAR: + case OP_MINSTARI: + case OP_MINUPTO: + case OP_MINUPTOI: + case OP_NOTMINPLUS: + case OP_NOTMINPLUSI: + case OP_NOTMINQUERY: + case OP_NOTMINQUERYI: + case OP_NOTMINSTAR: + case OP_NOTMINSTARI: + case OP_NOTMINUPTO: + case OP_NOTMINUPTOI: + case OP_NOTPLUS: + case OP_NOTPLUSI: + case OP_NOTPOSPLUS: + case OP_NOTPOSPLUSI: + case OP_NOTPOSQUERY: + case OP_NOTPOSQUERYI: + case OP_NOTPOSSTAR: + case OP_NOTPOSSTARI: + case OP_NOTPOSUPTO: + case OP_NOTPOSUPTOI: + case OP_NOTQUERY: + case OP_NOTQUERYI: + case OP_NOTSTAR: + case OP_NOTSTARI: + case OP_NOTUPTO: + case OP_NOTUPTOI: + case OP_PLUS: + case OP_PLUSI: + case OP_POSPLUS: + case OP_POSPLUSI: + case OP_POSQUERY: + case OP_POSQUERYI: + case OP_POSSTAR: + case OP_POSSTARI: + case OP_POSUPTO: + case OP_POSUPTOI: + case OP_QUERY: + case OP_QUERYI: + case OP_REF: + case OP_REFI: + case OP_SBRA: + case OP_SBRAPOS: + case OP_SCBRA: + case OP_SCBRAPOS: + case OP_SCOND: + case OP_SKIPZERO: + case OP_STAR: + case OP_STARI: + case OP_TYPEMINPLUS: + case OP_TYPEMINQUERY: + case OP_TYPEMINSTAR: + case OP_TYPEMINUPTO: + case OP_TYPEPLUS: + case OP_TYPEPOSPLUS: + case OP_TYPEPOSQUERY: + case OP_TYPEPOSSTAR: + case OP_TYPEPOSUPTO: + case OP_TYPEQUERY: + case OP_TYPESTAR: + case OP_TYPEUPTO: + case OP_UPTO: + case OP_UPTOI: + return -1; + + /* Catch unrecognized opcodes so that when new ones are added they + are not forgotten, as has happened in the past. */ + + default: + return -4; + } + } +/* Control never gets here */ +} + + + + +/************************************************* +* Scan compiled regex for specific bracket * +*************************************************/ + +/* This little function scans through a compiled pattern until it finds a +capturing bracket with the given number, or, if the number is negative, an +instance of OP_REVERSE for a lookbehind. The function is global in the C sense +so that it can be called from pcre_study() when finding the minimum matching +length. + +Arguments: + code points to start of expression + utf TRUE in UTF-8 / UTF-16 mode + number the required bracket number or negative to find a lookbehind + +Returns: pointer to the opcode for the bracket, or NULL if not found +*/ + +const pcre_uchar * +PRIV(find_bracket)(const pcre_uchar *code, BOOL utf, int number) +{ +for (;;) + { + register int c = *code; + + if (c == OP_END) return NULL; + + /* XCLASS is used for classes that cannot be represented just by a bit + map. This includes negated single high-valued characters. The length in + the table is zero; the actual length is stored in the compiled code. */ + + if (c == OP_XCLASS) code += GET(code, 1); + + /* Handle recursion */ + + else if (c == OP_REVERSE) + { + if (number < 0) return (pcre_uchar *)code; + code += PRIV(OP_lengths)[c]; + } + + /* Handle capturing bracket */ + + else if (c == OP_CBRA || c == OP_SCBRA || + c == OP_CBRAPOS || c == OP_SCBRAPOS) + { + int n = GET2(code, 1+LINK_SIZE); + if (n == number) return (pcre_uchar *)code; + code += PRIV(OP_lengths)[c]; + } + + /* Otherwise, we can get the item's length from the table, except that for + repeated character types, we have to test for \p and \P, which have an extra + two bytes of parameters, and for MARK/PRUNE/SKIP/THEN with an argument, we + must add in its length. */ + + else + { + switch(c) + { + case OP_TYPESTAR: + case OP_TYPEMINSTAR: + case OP_TYPEPLUS: + case OP_TYPEMINPLUS: + case OP_TYPEQUERY: + case OP_TYPEMINQUERY: + case OP_TYPEPOSSTAR: + case OP_TYPEPOSPLUS: + case OP_TYPEPOSQUERY: + if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2; + break; + + case OP_TYPEUPTO: + case OP_TYPEMINUPTO: + case OP_TYPEEXACT: + case OP_TYPEPOSUPTO: + if (code[1 + IMM2_SIZE] == OP_PROP + || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2; + break; + + case OP_MARK: + case OP_PRUNE_ARG: + case OP_SKIP_ARG: + code += code[1]; + break; + + case OP_THEN_ARG: + code += code[1]; + break; + } + + /* Add in the fixed length from the table */ + + code += PRIV(OP_lengths)[c]; + + /* In UTF-8 mode, opcodes that are followed by a character may be followed by + a multi-byte character. The length in the table is a minimum, so we have to + arrange to skip the extra bytes. */ + +#ifdef SUPPORT_UTF + if (utf) switch(c) + { + case OP_CHAR: + case OP_CHARI: + case OP_EXACT: + case OP_EXACTI: + case OP_UPTO: + case OP_UPTOI: + case OP_MINUPTO: + case OP_MINUPTOI: + case OP_POSUPTO: + case OP_POSUPTOI: + case OP_STAR: + case OP_STARI: + case OP_MINSTAR: + case OP_MINSTARI: + case OP_POSSTAR: + case OP_POSSTARI: + case OP_PLUS: + case OP_PLUSI: + case OP_MINPLUS: + case OP_MINPLUSI: + case OP_POSPLUS: + case OP_POSPLUSI: + case OP_QUERY: + case OP_QUERYI: + case OP_MINQUERY: + case OP_MINQUERYI: + case OP_POSQUERY: + case OP_POSQUERYI: + if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]); + break; + } +#else + (void)(utf); /* Keep compiler happy by referencing function argument */ +#endif + } + } +} + + + +/************************************************* +* Scan compiled regex for recursion reference * +*************************************************/ + +/* This little function scans through a compiled pattern until it finds an +instance of OP_RECURSE. + +Arguments: + code points to start of expression + utf TRUE in UTF-8 / UTF-16 mode + +Returns: pointer to the opcode for OP_RECURSE, or NULL if not found +*/ + +static const pcre_uchar * +find_recurse(const pcre_uchar *code, BOOL utf) +{ +for (;;) + { + register int c = *code; + if (c == OP_END) return NULL; + if (c == OP_RECURSE) return code; + + /* XCLASS is used for classes that cannot be represented just by a bit + map. This includes negated single high-valued characters. The length in + the table is zero; the actual length is stored in the compiled code. */ + + if (c == OP_XCLASS) code += GET(code, 1); + + /* Otherwise, we can get the item's length from the table, except that for + repeated character types, we have to test for \p and \P, which have an extra + two bytes of parameters, and for MARK/PRUNE/SKIP/THEN with an argument, we + must add in its length. */ + + else + { + switch(c) + { + case OP_TYPESTAR: + case OP_TYPEMINSTAR: + case OP_TYPEPLUS: + case OP_TYPEMINPLUS: + case OP_TYPEQUERY: + case OP_TYPEMINQUERY: + case OP_TYPEPOSSTAR: + case OP_TYPEPOSPLUS: + case OP_TYPEPOSQUERY: + if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2; + break; + + case OP_TYPEPOSUPTO: + case OP_TYPEUPTO: + case OP_TYPEMINUPTO: + case OP_TYPEEXACT: + if (code[1 + IMM2_SIZE] == OP_PROP + || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2; + break; + + case OP_MARK: + case OP_PRUNE_ARG: + case OP_SKIP_ARG: + code += code[1]; + break; + + case OP_THEN_ARG: + code += code[1]; + break; + } + + /* Add in the fixed length from the table */ + + code += PRIV(OP_lengths)[c]; + + /* In UTF-8 mode, opcodes that are followed by a character may be followed + by a multi-byte character. The length in the table is a minimum, so we have + to arrange to skip the extra bytes. */ + +#ifdef SUPPORT_UTF + if (utf) switch(c) + { + case OP_CHAR: + case OP_CHARI: + case OP_NOT: + case OP_NOTI: + case OP_EXACT: + case OP_EXACTI: + case OP_NOTEXACT: + case OP_NOTEXACTI: + case OP_UPTO: + case OP_UPTOI: + case OP_NOTUPTO: + case OP_NOTUPTOI: + case OP_MINUPTO: + case OP_MINUPTOI: + case OP_NOTMINUPTO: + case OP_NOTMINUPTOI: + case OP_POSUPTO: + case OP_POSUPTOI: + case OP_NOTPOSUPTO: + case OP_NOTPOSUPTOI: + case OP_STAR: + case OP_STARI: + case OP_NOTSTAR: + case OP_NOTSTARI: + case OP_MINSTAR: + case OP_MINSTARI: + case OP_NOTMINSTAR: + case OP_NOTMINSTARI: + case OP_POSSTAR: + case OP_POSSTARI: + case OP_NOTPOSSTAR: + case OP_NOTPOSSTARI: + case OP_PLUS: + case OP_PLUSI: + case OP_NOTPLUS: + case OP_NOTPLUSI: + case OP_MINPLUS: + case OP_MINPLUSI: + case OP_NOTMINPLUS: + case OP_NOTMINPLUSI: + case OP_POSPLUS: + case OP_POSPLUSI: + case OP_NOTPOSPLUS: + case OP_NOTPOSPLUSI: + case OP_QUERY: + case OP_QUERYI: + case OP_NOTQUERY: + case OP_NOTQUERYI: + case OP_MINQUERY: + case OP_MINQUERYI: + case OP_NOTMINQUERY: + case OP_NOTMINQUERYI: + case OP_POSQUERY: + case OP_POSQUERYI: + case OP_NOTPOSQUERY: + case OP_NOTPOSQUERYI: + if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]); + break; + } +#else + (void)(utf); /* Keep compiler happy by referencing function argument */ +#endif + } + } +} + + + +/************************************************* +* Scan compiled branch for non-emptiness * +*************************************************/ + +/* This function scans through a branch of a compiled pattern to see whether it +can match the empty string or not. It is called from could_be_empty() +below and from compile_branch() when checking for an unlimited repeat of a +group that can match nothing. Note that first_significant_code() skips over +backward and negative forward assertions when its final argument is TRUE. If we +hit an unclosed bracket, we return "empty" - this means we've struck an inner +bracket whose current branch will already have been scanned. + +Arguments: + code points to start of search + endcode points to where to stop + utf TRUE if in UTF-8 / UTF-16 mode + cd contains pointers to tables etc. + +Returns: TRUE if what is matched could be empty +*/ + +static BOOL +could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode, + BOOL utf, compile_data *cd) +{ +register int c; +for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE); + code < endcode; + code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE)) + { + const pcre_uchar *ccode; + + c = *code; + + /* Skip over forward assertions; the other assertions are skipped by + first_significant_code() with a TRUE final argument. */ + + if (c == OP_ASSERT) + { + do code += GET(code, 1); while (*code == OP_ALT); + c = *code; + continue; + } + + /* For a recursion/subroutine call, if its end has been reached, which + implies a backward reference subroutine call, we can scan it. If it's a + forward reference subroutine call, we can't. To detect forward reference + we have to scan up the list that is kept in the workspace. This function is + called only when doing the real compile, not during the pre-compile that + measures the size of the compiled pattern. */ + + if (c == OP_RECURSE) + { + const pcre_uchar *scode; + BOOL empty_branch; + + /* Test for forward reference */ + + for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE) + if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE; + + /* Not a forward reference, test for completed backward reference */ + + empty_branch = FALSE; + scode = cd->start_code + GET(code, 1); + if (GET(scode, 1) == 0) return TRUE; /* Unclosed */ + + /* Completed backwards reference */ + + do + { + if (could_be_empty_branch(scode, endcode, utf, cd)) + { + empty_branch = TRUE; + break; + } + scode += GET(scode, 1); + } + while (*scode == OP_ALT); + + if (!empty_branch) return FALSE; /* All branches are non-empty */ + continue; + } + + /* Groups with zero repeats can of course be empty; skip them. */ + + if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO || + c == OP_BRAPOSZERO) + { + code += PRIV(OP_lengths)[c]; + do code += GET(code, 1); while (*code == OP_ALT); + c = *code; + continue; + } + + /* A nested group that is already marked as "could be empty" can just be + skipped. */ + + if (c == OP_SBRA || c == OP_SBRAPOS || + c == OP_SCBRA || c == OP_SCBRAPOS) + { + do code += GET(code, 1); while (*code == OP_ALT); + c = *code; + continue; + } + + /* For other groups, scan the branches. */ + + if (c == OP_BRA || c == OP_BRAPOS || + c == OP_CBRA || c == OP_CBRAPOS || + c == OP_ONCE || c == OP_ONCE_NC || + c == OP_COND) + { + BOOL empty_branch; + if (GET(code, 1) == 0) return TRUE; /* Hit unclosed bracket */ + + /* If a conditional group has only one branch, there is a second, implied, + empty branch, so just skip over the conditional, because it could be empty. + Otherwise, scan the individual branches of the group. */ + + if (c == OP_COND && code[GET(code, 1)] != OP_ALT) + code += GET(code, 1); + else + { + empty_branch = FALSE; + do + { + if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd)) + empty_branch = TRUE; + code += GET(code, 1); + } + while (*code == OP_ALT); + if (!empty_branch) return FALSE; /* All branches are non-empty */ + } + + c = *code; + continue; + } + + /* Handle the other opcodes */ + + switch (c) + { + /* Check for quantifiers after a class. XCLASS is used for classes that + cannot be represented just by a bit map. This includes negated single + high-valued characters. The length in PRIV(OP_lengths)[] is zero; the + actual length is stored in the compiled code, so we must update "code" + here. */ + +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + case OP_XCLASS: + ccode = code += GET(code, 1); + goto CHECK_CLASS_REPEAT; +#endif + + case OP_CLASS: + case OP_NCLASS: + ccode = code + PRIV(OP_lengths)[OP_CLASS]; + +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + CHECK_CLASS_REPEAT: +#endif + + switch (*ccode) + { + case OP_CRSTAR: /* These could be empty; continue */ + case OP_CRMINSTAR: + case OP_CRQUERY: + case OP_CRMINQUERY: + break; + + default: /* Non-repeat => class must match */ + case OP_CRPLUS: /* These repeats aren't empty */ + case OP_CRMINPLUS: + return FALSE; + + case OP_CRRANGE: + case OP_CRMINRANGE: + if (GET2(ccode, 1) > 0) return FALSE; /* Minimum > 0 */ + break; + } + break; + + /* Opcodes that must match a character */ + + case OP_PROP: + case OP_NOTPROP: + case OP_EXTUNI: + case OP_NOT_DIGIT: + case OP_DIGIT: + case OP_NOT_WHITESPACE: + case OP_WHITESPACE: + case OP_NOT_WORDCHAR: + case OP_WORDCHAR: + case OP_ANY: + case OP_ALLANY: + case OP_ANYBYTE: + case OP_CHAR: + case OP_CHARI: + case OP_NOT: + case OP_NOTI: + case OP_PLUS: + case OP_MINPLUS: + case OP_POSPLUS: + case OP_EXACT: + case OP_NOTPLUS: + case OP_NOTMINPLUS: + case OP_NOTPOSPLUS: + case OP_NOTEXACT: + case OP_TYPEPLUS: + case OP_TYPEMINPLUS: + case OP_TYPEPOSPLUS: + case OP_TYPEEXACT: + return FALSE; + + /* These are going to continue, as they may be empty, but we have to + fudge the length for the \p and \P cases. */ + + case OP_TYPESTAR: + case OP_TYPEMINSTAR: + case OP_TYPEPOSSTAR: + case OP_TYPEQUERY: + case OP_TYPEMINQUERY: + case OP_TYPEPOSQUERY: + if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2; + break; + + /* Same for these */ + + case OP_TYPEUPTO: + case OP_TYPEMINUPTO: + case OP_TYPEPOSUPTO: + if (code[1 + IMM2_SIZE] == OP_PROP + || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2; + break; + + /* End of branch */ + + case OP_KET: + case OP_KETRMAX: + case OP_KETRMIN: + case OP_KETRPOS: + case OP_ALT: + return TRUE; + + /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO, + MINUPTO, and POSUPTO may be followed by a multibyte character */ + +#ifdef SUPPORT_UTF + case OP_STAR: + case OP_STARI: + case OP_MINSTAR: + case OP_MINSTARI: + case OP_POSSTAR: + case OP_POSSTARI: + case OP_QUERY: + case OP_QUERYI: + case OP_MINQUERY: + case OP_MINQUERYI: + case OP_POSQUERY: + case OP_POSQUERYI: + if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]); + break; + + case OP_UPTO: + case OP_UPTOI: + case OP_MINUPTO: + case OP_MINUPTOI: + case OP_POSUPTO: + case OP_POSUPTOI: + if (utf && HAS_EXTRALEN(code[1 + IMM2_SIZE])) code += GET_EXTRALEN(code[1 + IMM2_SIZE]); + break; +#endif + + /* MARK, and PRUNE/SKIP/THEN with an argument must skip over the argument + string. */ + + case OP_MARK: + case OP_PRUNE_ARG: + case OP_SKIP_ARG: + code += code[1]; + break; + + case OP_THEN_ARG: + code += code[1]; + break; + + /* None of the remaining opcodes are required to match a character. */ + + default: + break; + } + } + +return TRUE; +} + + + +/************************************************* +* Scan compiled regex for non-emptiness * +*************************************************/ + +/* This function is called to check for left recursive calls. We want to check +the current branch of the current pattern to see if it could match the empty +string. If it could, we must look outwards for branches at other levels, +stopping when we pass beyond the bracket which is the subject of the recursion. +This function is called only during the real compile, not during the +pre-compile. + +Arguments: + code points to start of the recursion + endcode points to where to stop (current RECURSE item) + bcptr points to the chain of current (unclosed) branch starts + utf TRUE if in UTF-8 / UTF-16 mode + cd pointers to tables etc + +Returns: TRUE if what is matched could be empty +*/ + +static BOOL +could_be_empty(const pcre_uchar *code, const pcre_uchar *endcode, + branch_chain *bcptr, BOOL utf, compile_data *cd) +{ +while (bcptr != NULL && bcptr->current_branch >= code) + { + if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd)) + return FALSE; + bcptr = bcptr->outer; + } +return TRUE; +} + + + +/************************************************* +* Check for POSIX class syntax * +*************************************************/ + +/* This function is called when the sequence "[:" or "[." or "[=" is +encountered in a character class. It checks whether this is followed by a +sequence of characters terminated by a matching ":]" or ".]" or "=]". If we +reach an unescaped ']' without the special preceding character, return FALSE. + +Originally, this function only recognized a sequence of letters between the +terminators, but it seems that Perl recognizes any sequence of characters, +though of course unknown POSIX names are subsequently rejected. Perl gives an +"Unknown POSIX class" error for [:f\oo:] for example, where previously PCRE +didn't consider this to be a POSIX class. Likewise for [:1234:]. + +The problem in trying to be exactly like Perl is in the handling of escapes. We +have to be sure that [abc[:x\]pqr] is *not* treated as containing a POSIX +class, but [abc[:x\]pqr:]] is (so that an error can be generated). The code +below handles the special case of \], but does not try to do any other escape +processing. This makes it different from Perl for cases such as [:l\ower:] +where Perl recognizes it as the POSIX class "lower" but PCRE does not recognize +"l\ower". This is a lesser evil that not diagnosing bad classes when Perl does, +I think. + +A user pointed out that PCRE was rejecting [:a[:digit:]] whereas Perl was not. +It seems that the appearance of a nested POSIX class supersedes an apparent +external class. For example, [:a[:digit:]b:] matches "a", "b", ":", or +a digit. + +In Perl, unescaped square brackets may also appear as part of class names. For +example, [:a[:abc]b:] gives unknown POSIX class "[:abc]b:]". However, for +[:a[:abc]b][b:] it gives unknown POSIX class "[:abc]b][b:]", which does not +seem right at all. PCRE does not allow closing square brackets in POSIX class +names. + +Arguments: + ptr pointer to the initial [ + endptr where to return the end pointer + +Returns: TRUE or FALSE +*/ + +static BOOL +check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr) +{ +int terminator; /* Don't combine these lines; the Solaris cc */ +terminator = *(++ptr); /* compiler warns about "non-constant" initializer. */ +for (++ptr; *ptr != 0; ptr++) + { + if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) + ptr++; + else if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE; + else + { + if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) + { + *endptr = ptr; + return TRUE; + } + if (*ptr == CHAR_LEFT_SQUARE_BRACKET && + (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT || + ptr[1] == CHAR_EQUALS_SIGN) && + check_posix_syntax(ptr, endptr)) + return FALSE; + } + } +return FALSE; +} + + + + +/************************************************* +* Check POSIX class name * +*************************************************/ + +/* This function is called to check the name given in a POSIX-style class entry +such as [:alnum:]. + +Arguments: + ptr points to the first letter + len the length of the name + +Returns: a value representing the name, or -1 if unknown +*/ + +static int +check_posix_name(const pcre_uchar *ptr, int len) +{ +const char *pn = posix_names; +register int yield = 0; +while (posix_name_lengths[yield] != 0) + { + if (len == posix_name_lengths[yield] && + STRNCMP_UC_C8(ptr, pn, len) == 0) return yield; + pn += posix_name_lengths[yield] + 1; + yield++; + } +return -1; +} + + +/************************************************* +* Adjust OP_RECURSE items in repeated group * +*************************************************/ + +/* OP_RECURSE items contain an offset from the start of the regex to the group +that is referenced. This means that groups can be replicated for fixed +repetition simply by copying (because the recursion is allowed to refer to +earlier groups that are outside the current group). However, when a group is +optional (i.e. the minimum quantifier is zero), OP_BRAZERO or OP_SKIPZERO is +inserted before it, after it has been compiled. This means that any OP_RECURSE +items within it that refer to the group itself or any contained groups have to +have their offsets adjusted. That one of the jobs of this function. Before it +is called, the partially compiled regex must be temporarily terminated with +OP_END. + +This function has been extended with the possibility of forward references for +recursions and subroutine calls. It must also check the list of such references +for the group we are dealing with. If it finds that one of the recursions in +the current group is on this list, it adjusts the offset in the list, not the +value in the reference (which is a group number). + +Arguments: + group points to the start of the group + adjust the amount by which the group is to be moved + utf TRUE in UTF-8 / UTF-16 mode + cd contains pointers to tables etc. + save_hwm the hwm forward reference pointer at the start of the group + +Returns: nothing +*/ + +static void +adjust_recurse(pcre_uchar *group, int adjust, BOOL utf, compile_data *cd, + pcre_uchar *save_hwm) +{ +pcre_uchar *ptr = group; + +while ((ptr = (pcre_uchar *)find_recurse(ptr, utf)) != NULL) + { + int offset; + pcre_uchar *hc; + + /* See if this recursion is on the forward reference list. If so, adjust the + reference. */ + + for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE) + { + offset = GET(hc, 0); + if (cd->start_code + offset == ptr + 1) + { + PUT(hc, 0, offset + adjust); + break; + } + } + + /* Otherwise, adjust the recursion offset if it's after the start of this + group. */ + + if (hc >= cd->hwm) + { + offset = GET(ptr, 1); + if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust); + } + + ptr += 1 + LINK_SIZE; + } +} + + + +/************************************************* +* Insert an automatic callout point * +*************************************************/ + +/* This function is called when the PCRE_AUTO_CALLOUT option is set, to insert +callout points before each pattern item. + +Arguments: + code current code pointer + ptr current pattern pointer + cd pointers to tables etc + +Returns: new code pointer +*/ + +static pcre_uchar * +auto_callout(pcre_uchar *code, const pcre_uchar *ptr, compile_data *cd) +{ +*code++ = OP_CALLOUT; +*code++ = 255; +PUT(code, 0, (int)(ptr - cd->start_pattern)); /* Pattern offset */ +PUT(code, LINK_SIZE, 0); /* Default length */ +return code + 2 * LINK_SIZE; +} + + + +/************************************************* +* Complete a callout item * +*************************************************/ + +/* A callout item contains the length of the next item in the pattern, which +we can't fill in till after we have reached the relevant point. This is used +for both automatic and manual callouts. + +Arguments: + previous_callout points to previous callout item + ptr current pattern pointer + cd pointers to tables etc + +Returns: nothing +*/ + +static void +complete_callout(pcre_uchar *previous_callout, const pcre_uchar *ptr, compile_data *cd) +{ +int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2)); +PUT(previous_callout, 2 + LINK_SIZE, length); +} + + + +#ifdef SUPPORT_UCP +/************************************************* +* Get othercase range * +*************************************************/ + +/* This function is passed the start and end of a class range, in UTF-8 mode +with UCP support. It searches up the characters, looking for internal ranges of +characters in the "other" case. Each call returns the next one, updating the +start address. + +Arguments: + cptr points to starting character value; updated + d end value + ocptr where to put start of othercase range + odptr where to put end of othercase range + +Yield: TRUE when range returned; FALSE when no more +*/ + +static BOOL +get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr, + unsigned int *odptr) +{ +unsigned int c, othercase, next; + +for (c = *cptr; c <= d; c++) + { if ((othercase = UCD_OTHERCASE(c)) != c) break; } + +if (c > d) return FALSE; + +*ocptr = othercase; +next = othercase + 1; + +for (++c; c <= d; c++) + { + if (UCD_OTHERCASE(c) != next) break; + next++; + } + +*odptr = next - 1; +*cptr = c; + +return TRUE; +} + + + +/************************************************* +* Check a character and a property * +*************************************************/ + +/* This function is called by check_auto_possessive() when a property item +is adjacent to a fixed character. + +Arguments: + c the character + ptype the property type + pdata the data for the type + negated TRUE if it's a negated property (\P or \p{^) + +Returns: TRUE if auto-possessifying is OK +*/ + +static BOOL +check_char_prop(int c, int ptype, int pdata, BOOL negated) +{ +const ucd_record *prop = GET_UCD(c); +switch(ptype) + { + case PT_LAMP: + return (prop->chartype == ucp_Lu || + prop->chartype == ucp_Ll || + prop->chartype == ucp_Lt) == negated; + + case PT_GC: + return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated; + + case PT_PC: + return (pdata == prop->chartype) == negated; + + case PT_SC: + return (pdata == prop->script) == negated; + + /* These are specials */ + + case PT_ALNUM: + return (PRIV(ucp_gentype)[prop->chartype] == ucp_L || + PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated; + + case PT_SPACE: /* Perl space */ + return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z || + c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR) + == negated; + + case PT_PXSPACE: /* POSIX space */ + return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z || + c == CHAR_HT || c == CHAR_NL || c == CHAR_VT || + c == CHAR_FF || c == CHAR_CR) + == negated; + + case PT_WORD: + return (PRIV(ucp_gentype)[prop->chartype] == ucp_L || + PRIV(ucp_gentype)[prop->chartype] == ucp_N || + c == CHAR_UNDERSCORE) == negated; + } +return FALSE; +} +#endif /* SUPPORT_UCP */ + + + +/************************************************* +* Check if auto-possessifying is possible * +*************************************************/ + +/* This function is called for unlimited repeats of certain items, to see +whether the next thing could possibly match the repeated item. If not, it makes +sense to automatically possessify the repeated item. + +Arguments: + previous pointer to the repeated opcode + utf TRUE in UTF-8 / UTF-16 mode + ptr next character in pattern + options options bits + cd contains pointers to tables etc. + +Returns: TRUE if possessifying is wanted +*/ + +static BOOL +check_auto_possessive(const pcre_uchar *previous, BOOL utf, + const pcre_uchar *ptr, int options, compile_data *cd) +{ +pcre_int32 c, next; +int op_code = *previous++; + +/* Skip whitespace and comments in extended mode */ + +if ((options & PCRE_EXTENDED) != 0) + { + for (;;) + { + while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++; + if (*ptr == CHAR_NUMBER_SIGN) + { + ptr++; + while (*ptr != 0) + { + if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; } + ptr++; +#ifdef SUPPORT_UTF + if (utf) FORWARDCHAR(ptr); +#endif + } + } + else break; + } + } + +/* If the next item is one that we can handle, get its value. A non-negative +value is a character, a negative value is an escape value. */ + +if (*ptr == CHAR_BACKSLASH) + { + int temperrorcode = 0; + next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE); + if (temperrorcode != 0) return FALSE; + ptr++; /* Point after the escape sequence */ + } +else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0) + { +#ifdef SUPPORT_UTF + if (utf) { GETCHARINC(next, ptr); } else +#endif + next = *ptr++; + } +else return FALSE; + +/* Skip whitespace and comments in extended mode */ + +if ((options & PCRE_EXTENDED) != 0) + { + for (;;) + { + while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++; + if (*ptr == CHAR_NUMBER_SIGN) + { + ptr++; + while (*ptr != 0) + { + if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; } + ptr++; +#ifdef SUPPORT_UTF + if (utf) FORWARDCHAR(ptr); +#endif + } + } + else break; + } + } + +/* If the next thing is itself optional, we have to give up. */ + +if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK || + STRNCMP_UC_C8(ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0) + return FALSE; + +/* Now compare the next item with the previous opcode. First, handle cases when +the next item is a character. */ + +if (next >= 0) switch(op_code) + { + case OP_CHAR: +#ifdef SUPPORT_UTF + GETCHARTEST(c, previous); +#else + c = *previous; +#endif + return c != next; + + /* For CHARI (caseless character) we must check the other case. If we have + Unicode property support, we can use it to test the other case of + high-valued characters. */ + + case OP_CHARI: +#ifdef SUPPORT_UTF + GETCHARTEST(c, previous); +#else + c = *previous; +#endif + if (c == next) return FALSE; +#ifdef SUPPORT_UTF + if (utf) + { + unsigned int othercase; + if (next < 128) othercase = cd->fcc[next]; else +#ifdef SUPPORT_UCP + othercase = UCD_OTHERCASE((unsigned int)next); +#else + othercase = NOTACHAR; +#endif + return (unsigned int)c != othercase; + } + else +#endif /* SUPPORT_UTF */ + return (c != TABLE_GET((unsigned int)next, cd->fcc, next)); /* Non-UTF-8 mode */ + + case OP_NOT: +#ifdef SUPPORT_UTF + GETCHARTEST(c, previous); +#else + c = *previous; +#endif + return c == next; + + case OP_NOTI: +#ifdef SUPPORT_UTF + GETCHARTEST(c, previous); +#else + c = *previous; +#endif + if (c == next) return TRUE; +#ifdef SUPPORT_UTF + if (utf) + { + unsigned int othercase; + if (next < 128) othercase = cd->fcc[next]; else +#ifdef SUPPORT_UCP + othercase = UCD_OTHERCASE((unsigned int)next); +#else + othercase = NOTACHAR; +#endif + return (unsigned int)c == othercase; + } + else +#endif /* SUPPORT_UTF */ + return (c == TABLE_GET((unsigned int)next, cd->fcc, next)); /* Non-UTF-8 mode */ + + /* Note that OP_DIGIT etc. are generated only when PCRE_UCP is *not* set. + When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */ + + case OP_DIGIT: + return next > 255 || (cd->ctypes[next] & ctype_digit) == 0; + + case OP_NOT_DIGIT: + return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0; + + case OP_WHITESPACE: + return next > 255 || (cd->ctypes[next] & ctype_space) == 0; + + case OP_NOT_WHITESPACE: + return next <= 255 && (cd->ctypes[next] & ctype_space) != 0; + + case OP_WORDCHAR: + return next > 255 || (cd->ctypes[next] & ctype_word) == 0; + + case OP_NOT_WORDCHAR: + return next <= 255 && (cd->ctypes[next] & ctype_word) != 0; + + case OP_HSPACE: + case OP_NOT_HSPACE: + switch(next) + { + case 0x09: + case 0x20: + case 0xa0: + case 0x1680: + case 0x180e: + case 0x2000: + case 0x2001: + case 0x2002: + case 0x2003: + case 0x2004: + case 0x2005: + case 0x2006: + case 0x2007: + case 0x2008: + case 0x2009: + case 0x200A: + case 0x202f: + case 0x205f: + case 0x3000: + return op_code == OP_NOT_HSPACE; + default: + return op_code != OP_NOT_HSPACE; + } + + case OP_ANYNL: + case OP_VSPACE: + case OP_NOT_VSPACE: + switch(next) + { + case 0x0a: + case 0x0b: + case 0x0c: + case 0x0d: + case 0x85: + case 0x2028: + case 0x2029: + return op_code == OP_NOT_VSPACE; + default: + return op_code != OP_NOT_VSPACE; + } + +#ifdef SUPPORT_UCP + case OP_PROP: + return check_char_prop(next, previous[0], previous[1], FALSE); + + case OP_NOTPROP: + return check_char_prop(next, previous[0], previous[1], TRUE); +#endif + + default: + return FALSE; + } + + +/* Handle the case when the next item is \d, \s, etc. Note that when PCRE_UCP +is set, \d turns into ESC_du rather than ESC_d, etc., so ESC_d etc. are +generated only when PCRE_UCP is *not* set, that is, when only ASCII +characteristics are recognized. Similarly, the opcodes OP_DIGIT etc. are +replaced by OP_PROP codes when PCRE_UCP is set. */ + +switch(op_code) + { + case OP_CHAR: + case OP_CHARI: +#ifdef SUPPORT_UTF + GETCHARTEST(c, previous); +#else + c = *previous; +#endif + switch(-next) + { + case ESC_d: + return c > 255 || (cd->ctypes[c] & ctype_digit) == 0; + + case ESC_D: + return c <= 255 && (cd->ctypes[c] & ctype_digit) != 0; + + case ESC_s: + return c > 255 || (cd->ctypes[c] & ctype_space) == 0; + + case ESC_S: + return c <= 255 && (cd->ctypes[c] & ctype_space) != 0; + + case ESC_w: + return c > 255 || (cd->ctypes[c] & ctype_word) == 0; + + case ESC_W: + return c <= 255 && (cd->ctypes[c] & ctype_word) != 0; + + case ESC_h: + case ESC_H: + switch(c) + { + case 0x09: + case 0x20: + case 0xa0: + case 0x1680: + case 0x180e: + case 0x2000: + case 0x2001: + case 0x2002: + case 0x2003: + case 0x2004: + case 0x2005: + case 0x2006: + case 0x2007: + case 0x2008: + case 0x2009: + case 0x200A: + case 0x202f: + case 0x205f: + case 0x3000: + return -next != ESC_h; + default: + return -next == ESC_h; + } + + case ESC_v: + case ESC_V: + switch(c) + { + case 0x0a: + case 0x0b: + case 0x0c: + case 0x0d: + case 0x85: + case 0x2028: + case 0x2029: + return -next != ESC_v; + default: + return -next == ESC_v; + } + + /* When PCRE_UCP is set, these values get generated for \d etc. Find + their substitutions and process them. The result will always be either + -ESC_p or -ESC_P. Then fall through to process those values. */ + +#ifdef SUPPORT_UCP + case ESC_du: + case ESC_DU: + case ESC_wu: + case ESC_WU: + case ESC_su: + case ESC_SU: + { + int temperrorcode = 0; + ptr = substitutes[-next - ESC_DU]; + next = check_escape(&ptr, &temperrorcode, 0, options, FALSE); + if (temperrorcode != 0) return FALSE; + ptr++; /* For compatibility */ + } + /* Fall through */ + + case ESC_p: + case ESC_P: + { + int ptype, pdata, errorcodeptr; + BOOL negated; + + ptr--; /* Make ptr point at the p or P */ + ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr); + if (ptype < 0) return FALSE; + ptr++; /* Point past the final curly ket */ + + /* If the property item is optional, we have to give up. (When generated + from \d etc by PCRE_UCP, this test will have been applied much earlier, + to the original \d etc. At this point, ptr will point to a zero byte. */ + + if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK || + STRNCMP_UC_C8(ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0) + return FALSE; + + /* Do the property check. */ + + return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated); + } +#endif + + default: + return FALSE; + } + + /* In principle, support for Unicode properties should be integrated here as + well. It means re-organizing the above code so as to get hold of the property + values before switching on the op-code. However, I wonder how many patterns + combine ASCII \d etc with Unicode properties? (Note that if PCRE_UCP is set, + these op-codes are never generated.) */ + + case OP_DIGIT: + return next == -ESC_D || next == -ESC_s || next == -ESC_W || + next == -ESC_h || next == -ESC_v || next == -ESC_R; + + case OP_NOT_DIGIT: + return next == -ESC_d; + + case OP_WHITESPACE: + return next == -ESC_S || next == -ESC_d || next == -ESC_w; + + case OP_NOT_WHITESPACE: + return next == -ESC_s || next == -ESC_h || next == -ESC_v || next == -ESC_R; + + case OP_HSPACE: + return next == -ESC_S || next == -ESC_H || next == -ESC_d || + next == -ESC_w || next == -ESC_v || next == -ESC_R; + + case OP_NOT_HSPACE: + return next == -ESC_h; + + /* Can't have \S in here because VT matches \S (Perl anomaly) */ + case OP_ANYNL: + case OP_VSPACE: + return next == -ESC_V || next == -ESC_d || next == -ESC_w; + + case OP_NOT_VSPACE: + return next == -ESC_v || next == -ESC_R; + + case OP_WORDCHAR: + return next == -ESC_W || next == -ESC_s || next == -ESC_h || + next == -ESC_v || next == -ESC_R; + + case OP_NOT_WORDCHAR: + return next == -ESC_w || next == -ESC_d; + + default: + return FALSE; + } + +/* Control does not reach here */ +} + + + +/************************************************* +* Compile one branch * +*************************************************/ + +/* Scan the pattern, compiling it into the a vector. If the options are +changed during the branch, the pointer is used to change the external options +bits. This function is used during the pre-compile phase when we are trying +to find out the amount of memory needed, as well as during the real compile +phase. The value of lengthptr distinguishes the two phases. + +Arguments: + optionsptr pointer to the option bits + codeptr points to the pointer to the current code point + ptrptr points to the current pattern pointer + errorcodeptr points to error code variable + firstcharptr set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE) + reqcharptr set to the last literal character required, else < 0 + bcptr points to current branch chain + cond_depth conditional nesting depth + cd contains pointers to tables etc. + lengthptr NULL during the real compile phase + points to length accumulator during pre-compile phase + +Returns: TRUE on success + FALSE, with *errorcodeptr set non-zero on error +*/ + +static BOOL +compile_branch(int *optionsptr, pcre_uchar **codeptr, + const pcre_uchar **ptrptr, int *errorcodeptr, pcre_int32 *firstcharptr, + pcre_int32 *reqcharptr, branch_chain *bcptr, int cond_depth, + compile_data *cd, int *lengthptr) +{ +int repeat_type, op_type; +int repeat_min = 0, repeat_max = 0; /* To please picky compilers */ +int bravalue = 0; +int greedy_default, greedy_non_default; +pcre_int32 firstchar, reqchar; +pcre_int32 zeroreqchar, zerofirstchar; +pcre_int32 req_caseopt, reqvary, tempreqvary; +int options = *optionsptr; /* May change dynamically */ +int after_manual_callout = 0; +int length_prevgroup = 0; +register int c; +register pcre_uchar *code = *codeptr; +pcre_uchar *last_code = code; +pcre_uchar *orig_code = code; +pcre_uchar *tempcode; +BOOL inescq = FALSE; +BOOL groupsetfirstchar = FALSE; +const pcre_uchar *ptr = *ptrptr; +const pcre_uchar *tempptr; +const pcre_uchar *nestptr = NULL; +pcre_uchar *previous = NULL; +pcre_uchar *previous_callout = NULL; +pcre_uchar *save_hwm = NULL; +pcre_uint8 classbits[32]; + +/* We can fish out the UTF-8 setting once and for all into a BOOL, but we +must not do this for other options (e.g. PCRE_EXTENDED) because they may change +dynamically as we process the pattern. */ + +#ifdef SUPPORT_UTF +/* PCRE_UTF16 has the same value as PCRE_UTF8. */ +BOOL utf = (options & PCRE_UTF8) != 0; +pcre_uchar utf_chars[6]; +#else +BOOL utf = FALSE; +#endif + +/* Helper variables for OP_XCLASS opcode (for characters > 255). */ + +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 +BOOL xclass; +pcre_uchar *class_uchardata; +pcre_uchar *class_uchardata_base; +#endif + +#ifdef PCRE_DEBUG +if (lengthptr != NULL) DPRINTF((">> start branch\n")); +#endif + +/* Set up the default and non-default settings for greediness */ + +greedy_default = ((options & PCRE_UNGREEDY) != 0); +greedy_non_default = greedy_default ^ 1; + +/* Initialize no first byte, no required byte. REQ_UNSET means "no char +matching encountered yet". It gets changed to REQ_NONE if we hit something that +matches a non-fixed char first char; reqchar just remains unset if we never +find one. + +When we hit a repeat whose minimum is zero, we may have to adjust these values +to take the zero repeat into account. This is implemented by setting them to +zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual +item types that can be repeated set these backoff variables appropriately. */ + +firstchar = reqchar = zerofirstchar = zeroreqchar = REQ_UNSET; + +/* The variable req_caseopt contains either the REQ_CASELESS value +or zero, according to the current setting of the caseless flag. The +REQ_CASELESS leaves the lower 28 bit empty. It is added into the +firstchar or reqchar variables to record the case status of the +value. This is used only for ASCII characters. */ + +req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS:0; + +/* Switch on next character until the end of the branch */ + +for (;; ptr++) + { + BOOL negate_class; + BOOL should_flip_negation; + BOOL possessive_quantifier; + BOOL is_quantifier; + BOOL is_recurse; + BOOL reset_bracount; + int class_has_8bitchar; + int class_single_char; + int newoptions; + int recno; + int refsign; + int skipbytes; + int subreqchar; + int subfirstchar; + int terminator; + int mclength; + int tempbracount; + pcre_uchar mcbuffer[8]; + + /* Get next character in the pattern */ + + c = *ptr; + + /* If we are at the end of a nested substitution, revert to the outer level + string. Nesting only happens one level deep. */ + + if (c == 0 && nestptr != NULL) + { + ptr = nestptr; + nestptr = NULL; + c = *ptr; + } + + /* If we are in the pre-compile phase, accumulate the length used for the + previous cycle of this loop. */ + + if (lengthptr != NULL) + { +#ifdef PCRE_DEBUG + if (code > cd->hwm) cd->hwm = code; /* High water info */ +#endif + if (code > cd->start_workspace + cd->workspace_size - + WORK_SIZE_SAFETY_MARGIN) /* Check for overrun */ + { + *errorcodeptr = ERR52; + goto FAILED; + } + + /* There is at least one situation where code goes backwards: this is the + case of a zero quantifier after a class (e.g. [ab]{0}). At compile time, + the class is simply eliminated. However, it is created first, so we have to + allow memory for it. Therefore, don't ever reduce the length at this point. + */ + + if (code < last_code) code = last_code; + + /* Paranoid check for integer overflow */ + + if (OFLOW_MAX - *lengthptr < code - last_code) + { + *errorcodeptr = ERR20; + goto FAILED; + } + + *lengthptr += (int)(code - last_code); + DPRINTF(("length=%d added %d c=%c (0x%x)\n", *lengthptr, + (int)(code - last_code), c, c)); + + /* If "previous" is set and it is not at the start of the work space, move + it back to there, in order to avoid filling up the work space. Otherwise, + if "previous" is NULL, reset the current code pointer to the start. */ + + if (previous != NULL) + { + if (previous > orig_code) + { + memmove(orig_code, previous, IN_UCHARS(code - previous)); + code -= previous - orig_code; + previous = orig_code; + } + } + else code = orig_code; + + /* Remember where this code item starts so we can pick up the length + next time round. */ + + last_code = code; + } + + /* In the real compile phase, just check the workspace used by the forward + reference list. */ + + else if (cd->hwm > cd->start_workspace + cd->workspace_size - + WORK_SIZE_SAFETY_MARGIN) + { + *errorcodeptr = ERR52; + goto FAILED; + } + + /* If in \Q...\E, check for the end; if not, we have a literal */ + + if (inescq && c != 0) + { + if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E) + { + inescq = FALSE; + ptr++; + continue; + } + else + { + if (previous_callout != NULL) + { + if (lengthptr == NULL) /* Don't attempt in pre-compile phase */ + complete_callout(previous_callout, ptr, cd); + previous_callout = NULL; + } + if ((options & PCRE_AUTO_CALLOUT) != 0) + { + previous_callout = code; + code = auto_callout(code, ptr, cd); + } + goto NORMAL_CHAR; + } + } + + /* Fill in length of a previous callout, except when the next thing is + a quantifier. */ + + is_quantifier = + c == CHAR_ASTERISK || c == CHAR_PLUS || c == CHAR_QUESTION_MARK || + (c == CHAR_LEFT_CURLY_BRACKET && is_counted_repeat(ptr+1)); + + if (!is_quantifier && previous_callout != NULL && + after_manual_callout-- <= 0) + { + if (lengthptr == NULL) /* Don't attempt in pre-compile phase */ + complete_callout(previous_callout, ptr, cd); + previous_callout = NULL; + } + + /* In extended mode, skip white space and comments. */ + + if ((options & PCRE_EXTENDED) != 0) + { + if (MAX_255(*ptr) && (cd->ctypes[c] & ctype_space) != 0) continue; + if (c == CHAR_NUMBER_SIGN) + { + ptr++; + while (*ptr != 0) + { + if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; } + ptr++; +#ifdef SUPPORT_UTF + if (utf) FORWARDCHAR(ptr); +#endif + } + if (*ptr != 0) continue; + + /* Else fall through to handle end of string */ + c = 0; + } + } + + /* No auto callout for quantifiers. */ + + if ((options & PCRE_AUTO_CALLOUT) != 0 && !is_quantifier) + { + previous_callout = code; + code = auto_callout(code, ptr, cd); + } + + switch(c) + { + /* ===================================================================*/ + case 0: /* The branch terminates at string end */ + case CHAR_VERTICAL_LINE: /* or | or ) */ + case CHAR_RIGHT_PARENTHESIS: + *firstcharptr = firstchar; + *reqcharptr = reqchar; + *codeptr = code; + *ptrptr = ptr; + if (lengthptr != NULL) + { + if (OFLOW_MAX - *lengthptr < code - last_code) + { + *errorcodeptr = ERR20; + goto FAILED; + } + *lengthptr += (int)(code - last_code); /* To include callout length */ + DPRINTF((">> end branch\n")); + } + return TRUE; + + + /* ===================================================================*/ + /* Handle single-character metacharacters. In multiline mode, ^ disables + the setting of any following char as a first character. */ + + case CHAR_CIRCUMFLEX_ACCENT: + previous = NULL; + if ((options & PCRE_MULTILINE) != 0) + { + if (firstchar == REQ_UNSET) firstchar = REQ_NONE; + *code++ = OP_CIRCM; + } + else *code++ = OP_CIRC; + break; + + case CHAR_DOLLAR_SIGN: + previous = NULL; + *code++ = ((options & PCRE_MULTILINE) != 0)? OP_DOLLM : OP_DOLL; + break; + + /* There can never be a first char if '.' is first, whatever happens about + repeats. The value of reqchar doesn't change either. */ + + case CHAR_DOT: + if (firstchar == REQ_UNSET) firstchar = REQ_NONE; + zerofirstchar = firstchar; + zeroreqchar = reqchar; + previous = code; + *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY; + break; + + + /* ===================================================================*/ + /* Character classes. If the included characters are all < 256, we build a + 32-byte bitmap of the permitted characters, except in the special case + where there is only one such character. For negated classes, we build the + map as usual, then invert it at the end. However, we use a different opcode + so that data characters > 255 can be handled correctly. + + If the class contains characters outside the 0-255 range, a different + opcode is compiled. It may optionally have a bit map for characters < 256, + but those above are are explicitly listed afterwards. A flag byte tells + whether the bitmap is present, and whether this is a negated class or not. + + In JavaScript compatibility mode, an isolated ']' causes an error. In + default (Perl) mode, it is treated as a data character. */ + + case CHAR_RIGHT_SQUARE_BRACKET: + if ((cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0) + { + *errorcodeptr = ERR64; + goto FAILED; + } + goto NORMAL_CHAR; + + case CHAR_LEFT_SQUARE_BRACKET: + previous = code; + + /* PCRE supports POSIX class stuff inside a class. Perl gives an error if + they are encountered at the top level, so we'll do that too. */ + + if ((ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT || + ptr[1] == CHAR_EQUALS_SIGN) && + check_posix_syntax(ptr, &tempptr)) + { + *errorcodeptr = (ptr[1] == CHAR_COLON)? ERR13 : ERR31; + goto FAILED; + } + + /* If the first character is '^', set the negation flag and skip it. Also, + if the first few characters (either before or after ^) are \Q\E or \E we + skip them too. This makes for compatibility with Perl. */ + + negate_class = FALSE; + for (;;) + { + c = *(++ptr); + if (c == CHAR_BACKSLASH) + { + if (ptr[1] == CHAR_E) + ptr++; + else if (STRNCMP_UC_C8(ptr + 1, STR_Q STR_BACKSLASH STR_E, 3) == 0) + ptr += 3; + else + break; + } + else if (!negate_class && c == CHAR_CIRCUMFLEX_ACCENT) + negate_class = TRUE; + else break; + } + + /* Empty classes are allowed in JavaScript compatibility mode. Otherwise, + an initial ']' is taken as a data character -- the code below handles + that. In JS mode, [] must always fail, so generate OP_FAIL, whereas + [^] must match any character, so generate OP_ALLANY. */ + + if (c == CHAR_RIGHT_SQUARE_BRACKET && + (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0) + { + *code++ = negate_class? OP_ALLANY : OP_FAIL; + if (firstchar == REQ_UNSET) firstchar = REQ_NONE; + zerofirstchar = firstchar; + break; + } + + /* If a class contains a negative special such as \S, we need to flip the + negation flag at the end, so that support for characters > 255 works + correctly (they are all included in the class). */ + + should_flip_negation = FALSE; + + /* For optimization purposes, we track some properties of the class. + class_has_8bitchar will be non-zero, if the class contains at least one + < 256 character. class_single_char will be 1 if the class contains only + a single character. */ + + class_has_8bitchar = 0; + class_single_char = 0; + + /* Initialize the 32-char bit map to all zeros. We build the map in a + temporary bit of memory, in case the class contains only 1 character (less + than 256), because in that case the compiled code doesn't use the bit map. + */ + + memset(classbits, 0, 32 * sizeof(pcre_uint8)); + +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + xclass = FALSE; /* No chars >= 256 */ + class_uchardata = code + LINK_SIZE + 2; /* For UTF-8 items */ + class_uchardata_base = class_uchardata; /* For resetting in pass 1 */ +#endif + + /* Process characters until ] is reached. By writing this as a "do" it + means that an initial ] is taken as a data character. At the start of the + loop, c contains the first byte of the character. */ + + if (c != 0) do + { + const pcre_uchar *oldptr; + +#ifdef SUPPORT_UTF + if (utf && HAS_EXTRALEN(c)) + { /* Braces are required because the */ + GETCHARLEN(c, ptr, ptr); /* macro generates multiple statements */ + } +#endif + +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + /* In the pre-compile phase, accumulate the length of any extra + data and reset the pointer. This is so that very large classes that + contain a zillion > 255 characters no longer overwrite the work space + (which is on the stack). */ + + if (lengthptr != NULL) + { + *lengthptr += class_uchardata - class_uchardata_base; + class_uchardata = class_uchardata_base; + } +#endif + + /* Inside \Q...\E everything is literal except \E */ + + if (inescq) + { + if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E) /* If we are at \E */ + { + inescq = FALSE; /* Reset literal state */ + ptr++; /* Skip the 'E' */ + continue; /* Carry on with next */ + } + goto CHECK_RANGE; /* Could be range if \E follows */ + } + + /* Handle POSIX class names. Perl allows a negation extension of the + form [:^name:]. A square bracket that doesn't match the syntax is + treated as a literal. We also recognize the POSIX constructions + [.ch.] and [=ch=] ("collating elements") and fault them, as Perl + 5.6 and 5.8 do. */ + + if (c == CHAR_LEFT_SQUARE_BRACKET && + (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT || + ptr[1] == CHAR_EQUALS_SIGN) && check_posix_syntax(ptr, &tempptr)) + { + BOOL local_negate = FALSE; + int posix_class, taboffset, tabopt; + register const pcre_uint8 *cbits = cd->cbits; + pcre_uint8 pbits[32]; + + if (ptr[1] != CHAR_COLON) + { + *errorcodeptr = ERR31; + goto FAILED; + } + + ptr += 2; + if (*ptr == CHAR_CIRCUMFLEX_ACCENT) + { + local_negate = TRUE; + should_flip_negation = TRUE; /* Note negative special */ + ptr++; + } + + posix_class = check_posix_name(ptr, (int)(tempptr - ptr)); + if (posix_class < 0) + { + *errorcodeptr = ERR30; + goto FAILED; + } + + /* If matching is caseless, upper and lower are converted to + alpha. This relies on the fact that the class table starts with + alpha, lower, upper as the first 3 entries. */ + + if ((options & PCRE_CASELESS) != 0 && posix_class <= 2) + posix_class = 0; + + /* When PCRE_UCP is set, some of the POSIX classes are converted to + different escape sequences that use Unicode properties. */ + +#ifdef SUPPORT_UCP + if ((options & PCRE_UCP) != 0) + { + int pc = posix_class + ((local_negate)? POSIX_SUBSIZE/2 : 0); + if (posix_substitutes[pc] != NULL) + { + nestptr = tempptr + 1; + ptr = posix_substitutes[pc] - 1; + continue; + } + } +#endif + /* In the non-UCP case, we build the bit map for the POSIX class in a + chunk of local store because we may be adding and subtracting from it, + and we don't want to subtract bits that may be in the main map already. + At the end we or the result into the bit map that is being built. */ + + posix_class *= 3; + + /* Copy in the first table (always present) */ + + memcpy(pbits, cbits + posix_class_maps[posix_class], + 32 * sizeof(pcre_uint8)); + + /* If there is a second table, add or remove it as required. */ + + taboffset = posix_class_maps[posix_class + 1]; + tabopt = posix_class_maps[posix_class + 2]; + + if (taboffset >= 0) + { + if (tabopt >= 0) + for (c = 0; c < 32; c++) pbits[c] |= cbits[c + taboffset]; + else + for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset]; + } + + /* Not see if we need to remove any special characters. An option + value of 1 removes vertical space and 2 removes underscore. */ + + if (tabopt < 0) tabopt = -tabopt; + if (tabopt == 1) pbits[1] &= ~0x3c; + else if (tabopt == 2) pbits[11] &= 0x7f; + + /* Add the POSIX table or its complement into the main table that is + being built and we are done. */ + + if (local_negate) + for (c = 0; c < 32; c++) classbits[c] |= ~pbits[c]; + else + for (c = 0; c < 32; c++) classbits[c] |= pbits[c]; + + ptr = tempptr + 1; + /* Every class contains at least one < 256 characters. */ + class_has_8bitchar = 1; + /* Every class contains at least two characters. */ + class_single_char = 2; + continue; /* End of POSIX syntax handling */ + } + + /* Backslash may introduce a single character, or it may introduce one + of the specials, which just set a flag. The sequence \b is a special + case. Inside a class (and only there) it is treated as backspace. We + assume that other escapes have more than one character in them, so + speculatively set both class_has_8bitchar and class_single_char bigger + than one. Unrecognized escapes fall through and are either treated + as literal characters (by default), or are faulted if + PCRE_EXTRA is set. */ + + if (c == CHAR_BACKSLASH) + { + c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE); + if (*errorcodeptr != 0) goto FAILED; + + if (-c == ESC_b) c = CHAR_BS; /* \b is backspace in a class */ + else if (-c == ESC_N) /* \N is not supported in a class */ + { + *errorcodeptr = ERR71; + goto FAILED; + } + else if (-c == ESC_Q) /* Handle start of quoted string */ + { + if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E) + { + ptr += 2; /* avoid empty string */ + } + else inescq = TRUE; + continue; + } + else if (-c == ESC_E) continue; /* Ignore orphan \E */ + + if (c < 0) + { + register const pcre_uint8 *cbits = cd->cbits; + /* Every class contains at least two < 256 characters. */ + class_has_8bitchar++; + /* Every class contains at least two characters. */ + class_single_char += 2; + + switch (-c) + { +#ifdef SUPPORT_UCP + case ESC_du: /* These are the values given for \d etc */ + case ESC_DU: /* when PCRE_UCP is set. We replace the */ + case ESC_wu: /* escape sequence with an appropriate \p */ + case ESC_WU: /* or \P to test Unicode properties instead */ + case ESC_su: /* of the default ASCII testing. */ + case ESC_SU: + nestptr = ptr; + ptr = substitutes[-c - ESC_DU] - 1; /* Just before substitute */ + class_has_8bitchar--; /* Undo! */ + continue; +#endif + case ESC_d: + for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_digit]; + continue; + + case ESC_D: + should_flip_negation = TRUE; + for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit]; + continue; + + case ESC_w: + for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_word]; + continue; + + case ESC_W: + should_flip_negation = TRUE; + for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word]; + continue; + + /* Perl 5.004 onwards omits VT from \s, but we must preserve it + if it was previously set by something earlier in the character + class. */ + + case ESC_s: + classbits[0] |= cbits[cbit_space]; + classbits[1] |= cbits[cbit_space+1] & ~0x08; + for (c = 2; c < 32; c++) classbits[c] |= cbits[c+cbit_space]; + continue; + + case ESC_S: + should_flip_negation = TRUE; + for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space]; + classbits[1] |= 0x08; /* Perl 5.004 onwards omits VT from \s */ + continue; + + case ESC_h: + SETBIT(classbits, 0x09); /* VT */ + SETBIT(classbits, 0x20); /* SPACE */ + SETBIT(classbits, 0xa0); /* NSBP */ +#ifndef COMPILE_PCRE8 + xclass = TRUE; + *class_uchardata++ = XCL_SINGLE; + *class_uchardata++ = 0x1680; + *class_uchardata++ = XCL_SINGLE; + *class_uchardata++ = 0x180e; + *class_uchardata++ = XCL_RANGE; + *class_uchardata++ = 0x2000; + *class_uchardata++ = 0x200a; + *class_uchardata++ = XCL_SINGLE; + *class_uchardata++ = 0x202f; + *class_uchardata++ = XCL_SINGLE; + *class_uchardata++ = 0x205f; + *class_uchardata++ = XCL_SINGLE; + *class_uchardata++ = 0x3000; +#elif defined SUPPORT_UTF + if (utf) + { + xclass = TRUE; + *class_uchardata++ = XCL_SINGLE; + class_uchardata += PRIV(ord2utf)(0x1680, class_uchardata); + *class_uchardata++ = XCL_SINGLE; + class_uchardata += PRIV(ord2utf)(0x180e, class_uchardata); + *class_uchardata++ = XCL_RANGE; + class_uchardata += PRIV(ord2utf)(0x2000, class_uchardata); + class_uchardata += PRIV(ord2utf)(0x200a, class_uchardata); + *class_uchardata++ = XCL_SINGLE; + class_uchardata += PRIV(ord2utf)(0x202f, class_uchardata); + *class_uchardata++ = XCL_SINGLE; + class_uchardata += PRIV(ord2utf)(0x205f, class_uchardata); + *class_uchardata++ = XCL_SINGLE; + class_uchardata += PRIV(ord2utf)(0x3000, class_uchardata); + } +#endif + continue; + + case ESC_H: + for (c = 0; c < 32; c++) + { + int x = 0xff; + switch (c) + { + case 0x09/8: x ^= 1 << (0x09%8); break; + case 0x20/8: x ^= 1 << (0x20%8); break; + case 0xa0/8: x ^= 1 << (0xa0%8); break; + default: break; + } + classbits[c] |= x; + } +#ifndef COMPILE_PCRE8 + xclass = TRUE; + *class_uchardata++ = XCL_RANGE; + *class_uchardata++ = 0x0100; + *class_uchardata++ = 0x167f; + *class_uchardata++ = XCL_RANGE; + *class_uchardata++ = 0x1681; + *class_uchardata++ = 0x180d; + *class_uchardata++ = XCL_RANGE; + *class_uchardata++ = 0x180f; + *class_uchardata++ = 0x1fff; + *class_uchardata++ = XCL_RANGE; + *class_uchardata++ = 0x200b; + *class_uchardata++ = 0x202e; + *class_uchardata++ = XCL_RANGE; + *class_uchardata++ = 0x2030; + *class_uchardata++ = 0x205e; + *class_uchardata++ = XCL_RANGE; + *class_uchardata++ = 0x2060; + *class_uchardata++ = 0x2fff; + *class_uchardata++ = XCL_RANGE; + *class_uchardata++ = 0x3001; +#ifdef SUPPORT_UTF + if (utf) + class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata); + else +#endif + *class_uchardata++ = 0xffff; +#elif defined SUPPORT_UTF + if (utf) + { + xclass = TRUE; + *class_uchardata++ = XCL_RANGE; + class_uchardata += PRIV(ord2utf)(0x0100, class_uchardata); + class_uchardata += PRIV(ord2utf)(0x167f, class_uchardata); + *class_uchardata++ = XCL_RANGE; + class_uchardata += PRIV(ord2utf)(0x1681, class_uchardata); + class_uchardata += PRIV(ord2utf)(0x180d, class_uchardata); + *class_uchardata++ = XCL_RANGE; + class_uchardata += PRIV(ord2utf)(0x180f, class_uchardata); + class_uchardata += PRIV(ord2utf)(0x1fff, class_uchardata); + *class_uchardata++ = XCL_RANGE; + class_uchardata += PRIV(ord2utf)(0x200b, class_uchardata); + class_uchardata += PRIV(ord2utf)(0x202e, class_uchardata); + *class_uchardata++ = XCL_RANGE; + class_uchardata += PRIV(ord2utf)(0x2030, class_uchardata); + class_uchardata += PRIV(ord2utf)(0x205e, class_uchardata); + *class_uchardata++ = XCL_RANGE; + class_uchardata += PRIV(ord2utf)(0x2060, class_uchardata); + class_uchardata += PRIV(ord2utf)(0x2fff, class_uchardata); + *class_uchardata++ = XCL_RANGE; + class_uchardata += PRIV(ord2utf)(0x3001, class_uchardata); + class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata); + } +#endif + continue; + + case ESC_v: + SETBIT(classbits, 0x0a); /* LF */ + SETBIT(classbits, 0x0b); /* VT */ + SETBIT(classbits, 0x0c); /* FF */ + SETBIT(classbits, 0x0d); /* CR */ + SETBIT(classbits, 0x85); /* NEL */ +#ifndef COMPILE_PCRE8 + xclass = TRUE; + *class_uchardata++ = XCL_RANGE; + *class_uchardata++ = 0x2028; + *class_uchardata++ = 0x2029; +#elif defined SUPPORT_UTF + if (utf) + { + xclass = TRUE; + *class_uchardata++ = XCL_RANGE; + class_uchardata += PRIV(ord2utf)(0x2028, class_uchardata); + class_uchardata += PRIV(ord2utf)(0x2029, class_uchardata); + } +#endif + continue; + + case ESC_V: + for (c = 0; c < 32; c++) + { + int x = 0xff; + switch (c) + { + case 0x0a/8: x ^= 1 << (0x0a%8); + x ^= 1 << (0x0b%8); + x ^= 1 << (0x0c%8); + x ^= 1 << (0x0d%8); + break; + case 0x85/8: x ^= 1 << (0x85%8); break; + default: break; + } + classbits[c] |= x; + } + +#ifndef COMPILE_PCRE8 + xclass = TRUE; + *class_uchardata++ = XCL_RANGE; + *class_uchardata++ = 0x0100; + *class_uchardata++ = 0x2027; + *class_uchardata++ = XCL_RANGE; + *class_uchardata++ = 0x202a; +#ifdef SUPPORT_UTF + if (utf) + class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata); + else +#endif + *class_uchardata++ = 0xffff; +#elif defined SUPPORT_UTF + if (utf) + { + xclass = TRUE; + *class_uchardata++ = XCL_RANGE; + class_uchardata += PRIV(ord2utf)(0x0100, class_uchardata); + class_uchardata += PRIV(ord2utf)(0x2027, class_uchardata); + *class_uchardata++ = XCL_RANGE; + class_uchardata += PRIV(ord2utf)(0x202a, class_uchardata); + class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata); + } +#endif + continue; + +#ifdef SUPPORT_UCP + case ESC_p: + case ESC_P: + { + BOOL negated; + int pdata; + int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr); + if (ptype < 0) goto FAILED; + xclass = TRUE; + *class_uchardata++ = ((-c == ESC_p) != negated)? + XCL_PROP : XCL_NOTPROP; + *class_uchardata++ = ptype; + *class_uchardata++ = pdata; + class_has_8bitchar--; /* Undo! */ + continue; + } +#endif + /* Unrecognized escapes are faulted if PCRE is running in its + strict mode. By default, for compatibility with Perl, they are + treated as literals. */ + + default: + if ((options & PCRE_EXTRA) != 0) + { + *errorcodeptr = ERR7; + goto FAILED; + } + class_has_8bitchar--; /* Undo the speculative increase. */ + class_single_char -= 2; /* Undo the speculative increase. */ + c = *ptr; /* Get the final character and fall through */ + break; + } + } + + /* Fall through if we have a single character (c >= 0). This may be + greater than 256. */ + + } /* End of backslash handling */ + + /* A single character may be followed by '-' to form a range. However, + Perl does not permit ']' to be the end of the range. A '-' character + at the end is treated as a literal. Perl ignores orphaned \E sequences + entirely. The code for handling \Q and \E is messy. */ + + CHECK_RANGE: + while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E) + { + inescq = FALSE; + ptr += 2; + } + + oldptr = ptr; + + /* Remember \r or \n */ + + if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF; + + /* Check for range */ + + if (!inescq && ptr[1] == CHAR_MINUS) + { + int d; + ptr += 2; + while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2; + + /* If we hit \Q (not followed by \E) at this point, go into escaped + mode. */ + + while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_Q) + { + ptr += 2; + if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) + { ptr += 2; continue; } + inescq = TRUE; + break; + } + + if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET)) + { + ptr = oldptr; + goto LONE_SINGLE_CHARACTER; + } + +#ifdef SUPPORT_UTF + if (utf) + { /* Braces are required because the */ + GETCHARLEN(d, ptr, ptr); /* macro generates multiple statements */ + } + else +#endif + d = *ptr; /* Not UTF-8 mode */ + + /* The second part of a range can be a single-character escape, but + not any of the other escapes. Perl 5.6 treats a hyphen as a literal + in such circumstances. */ + + if (!inescq && d == CHAR_BACKSLASH) + { + d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE); + if (*errorcodeptr != 0) goto FAILED; + + /* \b is backspace; any other special means the '-' was literal */ + + if (d < 0) + { + if (d == -ESC_b) d = CHAR_BS; else + { + ptr = oldptr; + goto LONE_SINGLE_CHARACTER; /* A few lines below */ + } + } + } + + /* Check that the two values are in the correct order. Optimize + one-character ranges */ + + if (d < c) + { + *errorcodeptr = ERR8; + goto FAILED; + } + + if (d == c) goto LONE_SINGLE_CHARACTER; /* A few lines below */ + + /* Remember \r or \n */ + + if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF; + + /* Since we found a character range, single character optimizations + cannot be done anymore. */ + class_single_char = 2; + + /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless + matching, we have to use an XCLASS with extra data items. Caseless + matching for characters > 127 is available only if UCP support is + available. */ + +#if defined SUPPORT_UTF && !(defined COMPILE_PCRE8) + if ((d > 255) || (utf && ((options & PCRE_CASELESS) != 0 && d > 127))) +#elif defined SUPPORT_UTF + if (utf && (d > 255 || ((options & PCRE_CASELESS) != 0 && d > 127))) +#elif !(defined COMPILE_PCRE8) + if (d > 255) +#endif +#if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) + { + xclass = TRUE; + + /* With UCP support, we can find the other case equivalents of + the relevant characters. There may be several ranges. Optimize how + they fit with the basic range. */ + +#ifdef SUPPORT_UCP +#ifndef COMPILE_PCRE8 + if (utf && (options & PCRE_CASELESS) != 0) +#else + if ((options & PCRE_CASELESS) != 0) +#endif + { + unsigned int occ, ocd; + unsigned int cc = c; + unsigned int origd = d; + while (get_othercase_range(&cc, origd, &occ, &ocd)) + { + if (occ >= (unsigned int)c && + ocd <= (unsigned int)d) + continue; /* Skip embedded ranges */ + + if (occ < (unsigned int)c && + ocd >= (unsigned int)c - 1) /* Extend the basic range */ + { /* if there is overlap, */ + c = occ; /* noting that if occ < c */ + continue; /* we can't have ocd > d */ + } /* because a subrange is */ + if (ocd > (unsigned int)d && + occ <= (unsigned int)d + 1) /* always shorter than */ + { /* the basic range. */ + d = ocd; + continue; + } + + if (occ == ocd) + { + *class_uchardata++ = XCL_SINGLE; + } + else + { + *class_uchardata++ = XCL_RANGE; + class_uchardata += PRIV(ord2utf)(occ, class_uchardata); + } + class_uchardata += PRIV(ord2utf)(ocd, class_uchardata); + } + } +#endif /* SUPPORT_UCP */ + + /* Now record the original range, possibly modified for UCP caseless + overlapping ranges. */ + + *class_uchardata++ = XCL_RANGE; +#ifdef SUPPORT_UTF +#ifndef COMPILE_PCRE8 + if (utf) + { + class_uchardata += PRIV(ord2utf)(c, class_uchardata); + class_uchardata += PRIV(ord2utf)(d, class_uchardata); + } + else + { + *class_uchardata++ = c; + *class_uchardata++ = d; + } +#else + class_uchardata += PRIV(ord2utf)(c, class_uchardata); + class_uchardata += PRIV(ord2utf)(d, class_uchardata); +#endif +#else /* SUPPORT_UTF */ + *class_uchardata++ = c; + *class_uchardata++ = d; +#endif /* SUPPORT_UTF */ + + /* With UCP support, we are done. Without UCP support, there is no + caseless matching for UTF characters > 127; we can use the bit map + for the smaller ones. As for 16 bit characters without UTF, we + can still use */ + +#ifdef SUPPORT_UCP +#ifndef COMPILE_PCRE8 + if (utf) +#endif + continue; /* With next character in the class */ +#endif /* SUPPORT_UCP */ + +#if defined SUPPORT_UTF && !defined(SUPPORT_UCP) && !(defined COMPILE_PCRE8) + if (utf) + { + if ((options & PCRE_CASELESS) == 0 || c > 127) continue; + /* Adjust upper limit and fall through to set up the map */ + d = 127; + } + else + { + if (c > 255) continue; + /* Adjust upper limit and fall through to set up the map */ + d = 255; + } +#elif defined SUPPORT_UTF && !defined(SUPPORT_UCP) + if ((options & PCRE_CASELESS) == 0 || c > 127) continue; + /* Adjust upper limit and fall through to set up the map */ + d = 127; +#else + if (c > 255) continue; + /* Adjust upper limit and fall through to set up the map */ + d = 255; +#endif /* SUPPORT_UTF && !SUPPORT_UCP && !COMPILE_PCRE8 */ + } +#endif /* SUPPORT_UTF || !COMPILE_PCRE8 */ + + /* We use the bit map for 8 bit mode, or when the characters fall + partially or entirely to [0-255] ([0-127] for UCP) ranges. */ + + class_has_8bitchar = 1; + + /* We can save a bit of time by skipping this in the pre-compile. */ + + if (lengthptr == NULL) for (; c <= d; c++) + { + classbits[c/8] |= (1 << (c&7)); + if ((options & PCRE_CASELESS) != 0) + { + int uc = cd->fcc[c]; /* flip case */ + classbits[uc/8] |= (1 << (uc&7)); + } + } + + continue; /* Go get the next char in the class */ + } + + /* Handle a lone single character - we can get here for a normal + non-escape char, or after \ that introduces a single character or for an + apparent range that isn't. */ + + LONE_SINGLE_CHARACTER: + + /* Only the value of 1 matters for class_single_char. */ + + if (class_single_char < 2) class_single_char++; + + /* If class_charcount is 1, we saw precisely one character. As long as + there was no use of \p or \P, in other words, no use of any XCLASS + features, we can optimize. + + The optimization throws away the bit map. We turn the item into a + 1-character OP_CHAR[I] if it's positive, or OP_NOT[I] if it's negative. + In the positive case, it can cause firstchar to be set. Otherwise, there + can be no first char if this item is first, whatever repeat count may + follow. In the case of reqchar, save the previous value for reinstating. */ + + if (class_single_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) + { + ptr++; + zeroreqchar = reqchar; + + if (negate_class) + { + if (firstchar == REQ_UNSET) firstchar = REQ_NONE; + zerofirstchar = firstchar; + *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT; +#ifdef SUPPORT_UTF + if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR) + code += PRIV(ord2utf)(c, code); + else +#endif + *code++ = c; + goto NOT_CHAR; + } + + /* For a single, positive character, get the value into mcbuffer, and + then we can handle this with the normal one-character code. */ + +#ifdef SUPPORT_UTF + if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR) + mclength = PRIV(ord2utf)(c, mcbuffer); + else +#endif + { + mcbuffer[0] = c; + mclength = 1; + } + goto ONE_CHAR; + } /* End of 1-char optimization */ + + /* Handle a character that cannot go in the bit map. */ + +#if defined SUPPORT_UTF && !(defined COMPILE_PCRE8) + if ((c > 255) || (utf && ((options & PCRE_CASELESS) != 0 && c > 127))) +#elif defined SUPPORT_UTF + if (utf && (c > 255 || ((options & PCRE_CASELESS) != 0 && c > 127))) +#elif !(defined COMPILE_PCRE8) + if (c > 255) +#endif + +#if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) + { + xclass = TRUE; + *class_uchardata++ = XCL_SINGLE; +#ifdef SUPPORT_UTF +#ifndef COMPILE_PCRE8 + /* In non 8 bit mode, we can get here even if we are not in UTF mode. */ + if (!utf) + *class_uchardata++ = c; + else +#endif + class_uchardata += PRIV(ord2utf)(c, class_uchardata); +#else /* SUPPORT_UTF */ + *class_uchardata++ = c; +#endif /* SUPPORT_UTF */ + +#ifdef SUPPORT_UCP +#ifdef COMPILE_PCRE8 + if ((options & PCRE_CASELESS) != 0) +#else + /* In non 8 bit mode, we can get here even if we are not in UTF mode. */ + if (utf && (options & PCRE_CASELESS) != 0) +#endif + { + unsigned int othercase; + if ((int)(othercase = UCD_OTHERCASE(c)) != c) + { + *class_uchardata++ = XCL_SINGLE; + class_uchardata += PRIV(ord2utf)(othercase, class_uchardata); + } + } +#endif /* SUPPORT_UCP */ + + } + else +#endif /* SUPPORT_UTF || COMPILE_PCRE16 */ + + /* Handle a single-byte character */ + { + class_has_8bitchar = 1; + classbits[c/8] |= (1 << (c&7)); + if ((options & PCRE_CASELESS) != 0) + { + c = cd->fcc[c]; /* flip case */ + classbits[c/8] |= (1 << (c&7)); + } + } + } + + /* Loop until ']' reached. This "while" is the end of the "do" far above. + If we are at the end of an internal nested string, revert to the outer + string. */ + + while (((c = *(++ptr)) != 0 || + (nestptr != NULL && + (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != 0)) && + (c != CHAR_RIGHT_SQUARE_BRACKET || inescq)); + + /* Check for missing terminating ']' */ + + if (c == 0) + { + *errorcodeptr = ERR6; + goto FAILED; + } + + /* If this is the first thing in the branch, there can be no first char + setting, whatever the repeat count. Any reqchar setting must remain + unchanged after any kind of repeat. */ + + if (firstchar == REQ_UNSET) firstchar = REQ_NONE; + zerofirstchar = firstchar; + zeroreqchar = reqchar; + + /* If there are characters with values > 255, we have to compile an + extended class, with its own opcode, unless there was a negated special + such as \S in the class, and PCRE_UCP is not set, because in that case all + characters > 255 are in the class, so any that were explicitly given as + well can be ignored. If (when there are explicit characters > 255 that must + be listed) there are no characters < 256, we can omit the bitmap in the + actual compiled code. */ + +#ifdef SUPPORT_UTF + if (xclass && (!should_flip_negation || (options & PCRE_UCP) != 0)) +#elif !defined COMPILE_PCRE8 + if (xclass && !should_flip_negation) +#endif +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + { + *class_uchardata++ = XCL_END; /* Marks the end of extra data */ + *code++ = OP_XCLASS; + code += LINK_SIZE; + *code = negate_class? XCL_NOT:0; + + /* If the map is required, move up the extra data to make room for it; + otherwise just move the code pointer to the end of the extra data. */ + + if (class_has_8bitchar > 0) + { + *code++ |= XCL_MAP; + memmove(code + (32 / sizeof(pcre_uchar)), code, + IN_UCHARS(class_uchardata - code)); + memcpy(code, classbits, 32); + code = class_uchardata + (32 / sizeof(pcre_uchar)); + } + else code = class_uchardata; + + /* Now fill in the complete length of the item */ + + PUT(previous, 1, (int)(code - previous)); + break; /* End of class handling */ + } +#endif + + /* If there are no characters > 255, or they are all to be included or + excluded, set the opcode to OP_CLASS or OP_NCLASS, depending on whether the + whole class was negated and whether there were negative specials such as \S + (non-UCP) in the class. Then copy the 32-byte map into the code vector, + negating it if necessary. */ + + *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS; + if (lengthptr == NULL) /* Save time in the pre-compile phase */ + { + if (negate_class) + for (c = 0; c < 32; c++) classbits[c] = ~classbits[c]; + memcpy(code, classbits, 32); + } + code += 32 / sizeof(pcre_uchar); + NOT_CHAR: + break; + + + /* ===================================================================*/ + /* Various kinds of repeat; '{' is not necessarily a quantifier, but this + has been tested above. */ + + case CHAR_LEFT_CURLY_BRACKET: + if (!is_quantifier) goto NORMAL_CHAR; + ptr = read_repeat_counts(ptr+1, &repeat_min, &repeat_max, errorcodeptr); + if (*errorcodeptr != 0) goto FAILED; + goto REPEAT; + + case CHAR_ASTERISK: + repeat_min = 0; + repeat_max = -1; + goto REPEAT; + + case CHAR_PLUS: + repeat_min = 1; + repeat_max = -1; + goto REPEAT; + + case CHAR_QUESTION_MARK: + repeat_min = 0; + repeat_max = 1; + + REPEAT: + if (previous == NULL) + { + *errorcodeptr = ERR9; + goto FAILED; + } + + if (repeat_min == 0) + { + firstchar = zerofirstchar; /* Adjust for zero repeat */ + reqchar = zeroreqchar; /* Ditto */ + } + + /* Remember whether this is a variable length repeat */ + + reqvary = (repeat_min == repeat_max)? 0 : REQ_VARY; + + op_type = 0; /* Default single-char op codes */ + possessive_quantifier = FALSE; /* Default not possessive quantifier */ + + /* Save start of previous item, in case we have to move it up in order to + insert something before it. */ + + tempcode = previous; + + /* If the next character is '+', we have a possessive quantifier. This + implies greediness, whatever the setting of the PCRE_UNGREEDY option. + If the next character is '?' this is a minimizing repeat, by default, + but if PCRE_UNGREEDY is set, it works the other way round. We change the + repeat type to the non-default. */ + + if (ptr[1] == CHAR_PLUS) + { + repeat_type = 0; /* Force greedy */ + possessive_quantifier = TRUE; + ptr++; + } + else if (ptr[1] == CHAR_QUESTION_MARK) + { + repeat_type = greedy_non_default; + ptr++; + } + else repeat_type = greedy_default; + + /* If previous was a recursion call, wrap it in atomic brackets so that + previous becomes the atomic group. All recursions were so wrapped in the + past, but it no longer happens for non-repeated recursions. In fact, the + repeated ones could be re-implemented independently so as not to need this, + but for the moment we rely on the code for repeating groups. */ + + if (*previous == OP_RECURSE) + { + memmove(previous + 1 + LINK_SIZE, previous, IN_UCHARS(1 + LINK_SIZE)); + *previous = OP_ONCE; + PUT(previous, 1, 2 + 2*LINK_SIZE); + previous[2 + 2*LINK_SIZE] = OP_KET; + PUT(previous, 3 + 2*LINK_SIZE, 2 + 2*LINK_SIZE); + code += 2 + 2 * LINK_SIZE; + length_prevgroup = 3 + 3*LINK_SIZE; + + /* When actually compiling, we need to check whether this was a forward + reference, and if so, adjust the offset. */ + + if (lengthptr == NULL && cd->hwm >= cd->start_workspace + LINK_SIZE) + { + int offset = GET(cd->hwm, -LINK_SIZE); + if (offset == previous + 1 - cd->start_code) + PUT(cd->hwm, -LINK_SIZE, offset + 1 + LINK_SIZE); + } + } + + /* Now handle repetition for the different types of item. */ + + /* If previous was a character or negated character match, abolish the item + and generate a repeat item instead. If a char item has a minimum of more + than one, ensure that it is set in reqchar - it might not be if a sequence + such as x{3} is the first thing in a branch because the x will have gone + into firstchar instead. */ + + if (*previous == OP_CHAR || *previous == OP_CHARI + || *previous == OP_NOT || *previous == OP_NOTI) + { + switch (*previous) + { + default: /* Make compiler happy. */ + case OP_CHAR: op_type = OP_STAR - OP_STAR; break; + case OP_CHARI: op_type = OP_STARI - OP_STAR; break; + case OP_NOT: op_type = OP_NOTSTAR - OP_STAR; break; + case OP_NOTI: op_type = OP_NOTSTARI - OP_STAR; break; + } + + /* Deal with UTF characters that take up more than one character. It's + easier to write this out separately than try to macrify it. Use c to + hold the length of the character in bytes, plus UTF_LENGTH to flag that + it's a length rather than a small character. */ + +#ifdef SUPPORT_UTF + if (utf && NOT_FIRSTCHAR(code[-1])) + { + pcre_uchar *lastchar = code - 1; + BACKCHAR(lastchar); + c = (int)(code - lastchar); /* Length of UTF-8 character */ + memcpy(utf_chars, lastchar, IN_UCHARS(c)); /* Save the char */ + c |= UTF_LENGTH; /* Flag c as a length */ + } + else +#endif /* SUPPORT_UTF */ + + /* Handle the case of a single charater - either with no UTF support, or + with UTF disabled, or for a single character UTF character. */ + { + c = code[-1]; + if (*previous <= OP_CHARI && repeat_min > 1) + reqchar = c | req_caseopt | cd->req_varyopt; + } + + /* If the repetition is unlimited, it pays to see if the next thing on + the line is something that cannot possibly match this character. If so, + automatically possessifying this item gains some performance in the case + where the match fails. */ + + if (!possessive_quantifier && + repeat_max < 0 && + check_auto_possessive(previous, utf, ptr + 1, options, cd)) + { + repeat_type = 0; /* Force greedy */ + possessive_quantifier = TRUE; + } + + goto OUTPUT_SINGLE_REPEAT; /* Code shared with single character types */ + } + + /* If previous was a character type match (\d or similar), abolish it and + create a suitable repeat item. The code is shared with single-character + repeats by setting op_type to add a suitable offset into repeat_type. Note + the the Unicode property types will be present only when SUPPORT_UCP is + defined, but we don't wrap the little bits of code here because it just + makes it horribly messy. */ + + else if (*previous < OP_EODN) + { + pcre_uchar *oldcode; + int prop_type, prop_value; + op_type = OP_TYPESTAR - OP_STAR; /* Use type opcodes */ + c = *previous; + + if (!possessive_quantifier && + repeat_max < 0 && + check_auto_possessive(previous, utf, ptr + 1, options, cd)) + { + repeat_type = 0; /* Force greedy */ + possessive_quantifier = TRUE; + } + + OUTPUT_SINGLE_REPEAT: + if (*previous == OP_PROP || *previous == OP_NOTPROP) + { + prop_type = previous[1]; + prop_value = previous[2]; + } + else prop_type = prop_value = -1; + + oldcode = code; + code = previous; /* Usually overwrite previous item */ + + /* If the maximum is zero then the minimum must also be zero; Perl allows + this case, so we do too - by simply omitting the item altogether. */ + + if (repeat_max == 0) goto END_REPEAT; + + /*--------------------------------------------------------------------*/ + /* This code is obsolete from release 8.00; the restriction was finally + removed: */ + + /* All real repeats make it impossible to handle partial matching (maybe + one day we will be able to remove this restriction). */ + + /* if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL; */ + /*--------------------------------------------------------------------*/ + + /* Combine the op_type with the repeat_type */ + + repeat_type += op_type; + + /* A minimum of zero is handled either as the special case * or ?, or as + an UPTO, with the maximum given. */ + + if (repeat_min == 0) + { + if (repeat_max == -1) *code++ = OP_STAR + repeat_type; + else if (repeat_max == 1) *code++ = OP_QUERY + repeat_type; + else + { + *code++ = OP_UPTO + repeat_type; + PUT2INC(code, 0, repeat_max); + } + } + + /* A repeat minimum of 1 is optimized into some special cases. If the + maximum is unlimited, we use OP_PLUS. Otherwise, the original item is + left in place and, if the maximum is greater than 1, we use OP_UPTO with + one less than the maximum. */ + + else if (repeat_min == 1) + { + if (repeat_max == -1) + *code++ = OP_PLUS + repeat_type; + else + { + code = oldcode; /* leave previous item in place */ + if (repeat_max == 1) goto END_REPEAT; + *code++ = OP_UPTO + repeat_type; + PUT2INC(code, 0, repeat_max - 1); + } + } + + /* The case {n,n} is just an EXACT, while the general case {n,m} is + handled as an EXACT followed by an UPTO. */ + + else + { + *code++ = OP_EXACT + op_type; /* NB EXACT doesn't have repeat_type */ + PUT2INC(code, 0, repeat_min); + + /* If the maximum is unlimited, insert an OP_STAR. Before doing so, + we have to insert the character for the previous code. For a repeated + Unicode property match, there are two extra bytes that define the + required property. In UTF-8 mode, long characters have their length in + c, with the UTF_LENGTH bit as a flag. */ + + if (repeat_max < 0) + { +#ifdef SUPPORT_UTF + if (utf && (c & UTF_LENGTH) != 0) + { + memcpy(code, utf_chars, IN_UCHARS(c & 7)); + code += c & 7; + } + else +#endif + { + *code++ = c; + if (prop_type >= 0) + { + *code++ = prop_type; + *code++ = prop_value; + } + } + *code++ = OP_STAR + repeat_type; + } + + /* Else insert an UPTO if the max is greater than the min, again + preceded by the character, for the previously inserted code. If the + UPTO is just for 1 instance, we can use QUERY instead. */ + + else if (repeat_max != repeat_min) + { +#ifdef SUPPORT_UTF + if (utf && (c & UTF_LENGTH) != 0) + { + memcpy(code, utf_chars, IN_UCHARS(c & 7)); + code += c & 7; + } + else +#endif + *code++ = c; + if (prop_type >= 0) + { + *code++ = prop_type; + *code++ = prop_value; + } + repeat_max -= repeat_min; + + if (repeat_max == 1) + { + *code++ = OP_QUERY + repeat_type; + } + else + { + *code++ = OP_UPTO + repeat_type; + PUT2INC(code, 0, repeat_max); + } + } + } + + /* The character or character type itself comes last in all cases. */ + +#ifdef SUPPORT_UTF + if (utf && (c & UTF_LENGTH) != 0) + { + memcpy(code, utf_chars, IN_UCHARS(c & 7)); + code += c & 7; + } + else +#endif + *code++ = c; + + /* For a repeated Unicode property match, there are two extra bytes that + define the required property. */ + +#ifdef SUPPORT_UCP + if (prop_type >= 0) + { + *code++ = prop_type; + *code++ = prop_value; + } +#endif + } + + /* If previous was a character class or a back reference, we put the repeat + stuff after it, but just skip the item if the repeat was {0,0}. */ + + else if (*previous == OP_CLASS || + *previous == OP_NCLASS || +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + *previous == OP_XCLASS || +#endif + *previous == OP_REF || + *previous == OP_REFI) + { + if (repeat_max == 0) + { + code = previous; + goto END_REPEAT; + } + + /*--------------------------------------------------------------------*/ + /* This code is obsolete from release 8.00; the restriction was finally + removed: */ + + /* All real repeats make it impossible to handle partial matching (maybe + one day we will be able to remove this restriction). */ + + /* if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL; */ + /*--------------------------------------------------------------------*/ + + if (repeat_min == 0 && repeat_max == -1) + *code++ = OP_CRSTAR + repeat_type; + else if (repeat_min == 1 && repeat_max == -1) + *code++ = OP_CRPLUS + repeat_type; + else if (repeat_min == 0 && repeat_max == 1) + *code++ = OP_CRQUERY + repeat_type; + else + { + *code++ = OP_CRRANGE + repeat_type; + PUT2INC(code, 0, repeat_min); + if (repeat_max == -1) repeat_max = 0; /* 2-byte encoding for max */ + PUT2INC(code, 0, repeat_max); + } + } + + /* If previous was a bracket group, we may have to replicate it in certain + cases. Note that at this point we can encounter only the "basic" bracket + opcodes such as BRA and CBRA, as this is the place where they get converted + into the more special varieties such as BRAPOS and SBRA. A test for >= + OP_ASSERT and <= OP_COND includes ASSERT, ASSERT_NOT, ASSERTBACK, + ASSERTBACK_NOT, ONCE, BRA, CBRA, and COND. Originally, PCRE did not allow + repetition of assertions, but now it does, for Perl compatibility. */ + + else if (*previous >= OP_ASSERT && *previous <= OP_COND) + { + register int i; + int len = (int)(code - previous); + pcre_uchar *bralink = NULL; + pcre_uchar *brazeroptr = NULL; + + /* Repeating a DEFINE group is pointless, but Perl allows the syntax, so + we just ignore the repeat. */ + + if (*previous == OP_COND && previous[LINK_SIZE+1] == OP_DEF) + goto END_REPEAT; + + /* There is no sense in actually repeating assertions. The only potential + use of repetition is in cases when the assertion is optional. Therefore, + if the minimum is greater than zero, just ignore the repeat. If the + maximum is not not zero or one, set it to 1. */ + + if (*previous < OP_ONCE) /* Assertion */ + { + if (repeat_min > 0) goto END_REPEAT; + if (repeat_max < 0 || repeat_max > 1) repeat_max = 1; + } + + /* The case of a zero minimum is special because of the need to stick + OP_BRAZERO in front of it, and because the group appears once in the + data, whereas in other cases it appears the minimum number of times. For + this reason, it is simplest to treat this case separately, as otherwise + the code gets far too messy. There are several special subcases when the + minimum is zero. */ + + if (repeat_min == 0) + { + /* If the maximum is also zero, we used to just omit the group from the + output altogether, like this: + + ** if (repeat_max == 0) + ** { + ** code = previous; + ** goto END_REPEAT; + ** } + + However, that fails when a group or a subgroup within it is referenced + as a subroutine from elsewhere in the pattern, so now we stick in + OP_SKIPZERO in front of it so that it is skipped on execution. As we + don't have a list of which groups are referenced, we cannot do this + selectively. + + If the maximum is 1 or unlimited, we just have to stick in the BRAZERO + and do no more at this point. However, we do need to adjust any + OP_RECURSE calls inside the group that refer to the group itself or any + internal or forward referenced group, because the offset is from the + start of the whole regex. Temporarily terminate the pattern while doing + this. */ + + if (repeat_max <= 1) /* Covers 0, 1, and unlimited */ + { + *code = OP_END; + adjust_recurse(previous, 1, utf, cd, save_hwm); + memmove(previous + 1, previous, IN_UCHARS(len)); + code++; + if (repeat_max == 0) + { + *previous++ = OP_SKIPZERO; + goto END_REPEAT; + } + brazeroptr = previous; /* Save for possessive optimizing */ + *previous++ = OP_BRAZERO + repeat_type; + } + + /* If the maximum is greater than 1 and limited, we have to replicate + in a nested fashion, sticking OP_BRAZERO before each set of brackets. + The first one has to be handled carefully because it's the original + copy, which has to be moved up. The remainder can be handled by code + that is common with the non-zero minimum case below. We have to + adjust the value or repeat_max, since one less copy is required. Once + again, we may have to adjust any OP_RECURSE calls inside the group. */ + + else + { + int offset; + *code = OP_END; + adjust_recurse(previous, 2 + LINK_SIZE, utf, cd, save_hwm); + memmove(previous + 2 + LINK_SIZE, previous, IN_UCHARS(len)); + code += 2 + LINK_SIZE; + *previous++ = OP_BRAZERO + repeat_type; + *previous++ = OP_BRA; + + /* We chain together the bracket offset fields that have to be + filled in later when the ends of the brackets are reached. */ + + offset = (bralink == NULL)? 0 : (int)(previous - bralink); + bralink = previous; + PUTINC(previous, 0, offset); + } + + repeat_max--; + } + + /* If the minimum is greater than zero, replicate the group as many + times as necessary, and adjust the maximum to the number of subsequent + copies that we need. If we set a first char from the group, and didn't + set a required char, copy the latter from the former. If there are any + forward reference subroutine calls in the group, there will be entries on + the workspace list; replicate these with an appropriate increment. */ + + else + { + if (repeat_min > 1) + { + /* In the pre-compile phase, we don't actually do the replication. We + just adjust the length as if we had. Do some paranoid checks for + potential integer overflow. The INT64_OR_DOUBLE type is a 64-bit + integer type when available, otherwise double. */ + + if (lengthptr != NULL) + { + int delta = (repeat_min - 1)*length_prevgroup; + if ((INT64_OR_DOUBLE)(repeat_min - 1)* + (INT64_OR_DOUBLE)length_prevgroup > + (INT64_OR_DOUBLE)INT_MAX || + OFLOW_MAX - *lengthptr < delta) + { + *errorcodeptr = ERR20; + goto FAILED; + } + *lengthptr += delta; + } + + /* This is compiling for real. If there is a set first byte for + the group, and we have not yet set a "required byte", set it. Make + sure there is enough workspace for copying forward references before + doing the copy. */ + + else + { + if (groupsetfirstchar && reqchar < 0) reqchar = firstchar; + + for (i = 1; i < repeat_min; i++) + { + pcre_uchar *hc; + pcre_uchar *this_hwm = cd->hwm; + memcpy(code, previous, IN_UCHARS(len)); + + while (cd->hwm > cd->start_workspace + cd->workspace_size - + WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm)) + { + int save_offset = save_hwm - cd->start_workspace; + int this_offset = this_hwm - cd->start_workspace; + *errorcodeptr = expand_workspace(cd); + if (*errorcodeptr != 0) goto FAILED; + save_hwm = (pcre_uchar *)cd->start_workspace + save_offset; + this_hwm = (pcre_uchar *)cd->start_workspace + this_offset; + } + + for (hc = save_hwm; hc < this_hwm; hc += LINK_SIZE) + { + PUT(cd->hwm, 0, GET(hc, 0) + len); + cd->hwm += LINK_SIZE; + } + save_hwm = this_hwm; + code += len; + } + } + } + + if (repeat_max > 0) repeat_max -= repeat_min; + } + + /* This code is common to both the zero and non-zero minimum cases. If + the maximum is limited, it replicates the group in a nested fashion, + remembering the bracket starts on a stack. In the case of a zero minimum, + the first one was set up above. In all cases the repeat_max now specifies + the number of additional copies needed. Again, we must remember to + replicate entries on the forward reference list. */ + + if (repeat_max >= 0) + { + /* In the pre-compile phase, we don't actually do the replication. We + just adjust the length as if we had. For each repetition we must add 1 + to the length for BRAZERO and for all but the last repetition we must + add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some + paranoid checks to avoid integer overflow. The INT64_OR_DOUBLE type is + a 64-bit integer type when available, otherwise double. */ + + if (lengthptr != NULL && repeat_max > 0) + { + int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) - + 2 - 2*LINK_SIZE; /* Last one doesn't nest */ + if ((INT64_OR_DOUBLE)repeat_max * + (INT64_OR_DOUBLE)(length_prevgroup + 1 + 2 + 2*LINK_SIZE) + > (INT64_OR_DOUBLE)INT_MAX || + OFLOW_MAX - *lengthptr < delta) + { + *errorcodeptr = ERR20; + goto FAILED; + } + *lengthptr += delta; + } + + /* This is compiling for real */ + + else for (i = repeat_max - 1; i >= 0; i--) + { + pcre_uchar *hc; + pcre_uchar *this_hwm = cd->hwm; + + *code++ = OP_BRAZERO + repeat_type; + + /* All but the final copy start a new nesting, maintaining the + chain of brackets outstanding. */ + + if (i != 0) + { + int offset; + *code++ = OP_BRA; + offset = (bralink == NULL)? 0 : (int)(code - bralink); + bralink = code; + PUTINC(code, 0, offset); + } + + memcpy(code, previous, IN_UCHARS(len)); + + /* Ensure there is enough workspace for forward references before + copying them. */ + + while (cd->hwm > cd->start_workspace + cd->workspace_size - + WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm)) + { + int save_offset = save_hwm - cd->start_workspace; + int this_offset = this_hwm - cd->start_workspace; + *errorcodeptr = expand_workspace(cd); + if (*errorcodeptr != 0) goto FAILED; + save_hwm = (pcre_uchar *)cd->start_workspace + save_offset; + this_hwm = (pcre_uchar *)cd->start_workspace + this_offset; + } + + for (hc = save_hwm; hc < this_hwm; hc += LINK_SIZE) + { + PUT(cd->hwm, 0, GET(hc, 0) + len + ((i != 0)? 2+LINK_SIZE : 1)); + cd->hwm += LINK_SIZE; + } + save_hwm = this_hwm; + code += len; + } + + /* Now chain through the pending brackets, and fill in their length + fields (which are holding the chain links pro tem). */ + + while (bralink != NULL) + { + int oldlinkoffset; + int offset = (int)(code - bralink + 1); + pcre_uchar *bra = code - offset; + oldlinkoffset = GET(bra, 1); + bralink = (oldlinkoffset == 0)? NULL : bralink - oldlinkoffset; + *code++ = OP_KET; + PUTINC(code, 0, offset); + PUT(bra, 1, offset); + } + } + + /* If the maximum is unlimited, set a repeater in the final copy. For + ONCE brackets, that's all we need to do. However, possessively repeated + ONCE brackets can be converted into non-capturing brackets, as the + behaviour of (?:xx)++ is the same as (?>xx)++ and this saves having to + deal with possessive ONCEs specially. + + Otherwise, when we are doing the actual compile phase, check to see + whether this group is one that could match an empty string. If so, + convert the initial operator to the S form (e.g. OP_BRA -> OP_SBRA) so + that runtime checking can be done. [This check is also applied to ONCE + groups at runtime, but in a different way.] + + Then, if the quantifier was possessive and the bracket is not a + conditional, we convert the BRA code to the POS form, and the KET code to + KETRPOS. (It turns out to be convenient at runtime to detect this kind of + subpattern at both the start and at the end.) The use of special opcodes + makes it possible to reduce greatly the stack usage in pcre_exec(). If + the group is preceded by OP_BRAZERO, convert this to OP_BRAPOSZERO. + + Then, if the minimum number of matches is 1 or 0, cancel the possessive + flag so that the default action below, of wrapping everything inside + atomic brackets, does not happen. When the minimum is greater than 1, + there will be earlier copies of the group, and so we still have to wrap + the whole thing. */ + + else + { + pcre_uchar *ketcode = code - 1 - LINK_SIZE; + pcre_uchar *bracode = ketcode - GET(ketcode, 1); + + /* Convert possessive ONCE brackets to non-capturing */ + + if ((*bracode == OP_ONCE || *bracode == OP_ONCE_NC) && + possessive_quantifier) *bracode = OP_BRA; + + /* For non-possessive ONCE brackets, all we need to do is to + set the KET. */ + + if (*bracode == OP_ONCE || *bracode == OP_ONCE_NC) + *ketcode = OP_KETRMAX + repeat_type; + + /* Handle non-ONCE brackets and possessive ONCEs (which have been + converted to non-capturing above). */ + + else + { + /* In the compile phase, check for empty string matching. */ + + if (lengthptr == NULL) + { + pcre_uchar *scode = bracode; + do + { + if (could_be_empty_branch(scode, ketcode, utf, cd)) + { + *bracode += OP_SBRA - OP_BRA; + break; + } + scode += GET(scode, 1); + } + while (*scode == OP_ALT); + } + + /* Handle possessive quantifiers. */ + + if (possessive_quantifier) + { + /* For COND brackets, we wrap the whole thing in a possessively + repeated non-capturing bracket, because we have not invented POS + versions of the COND opcodes. Because we are moving code along, we + must ensure that any pending recursive references are updated. */ + + if (*bracode == OP_COND || *bracode == OP_SCOND) + { + int nlen = (int)(code - bracode); + *code = OP_END; + adjust_recurse(bracode, 1 + LINK_SIZE, utf, cd, save_hwm); + memmove(bracode + 1 + LINK_SIZE, bracode, IN_UCHARS(nlen)); + code += 1 + LINK_SIZE; + nlen += 1 + LINK_SIZE; + *bracode = OP_BRAPOS; + *code++ = OP_KETRPOS; + PUTINC(code, 0, nlen); + PUT(bracode, 1, nlen); + } + + /* For non-COND brackets, we modify the BRA code and use KETRPOS. */ + + else + { + *bracode += 1; /* Switch to xxxPOS opcodes */ + *ketcode = OP_KETRPOS; + } + + /* If the minimum is zero, mark it as possessive, then unset the + possessive flag when the minimum is 0 or 1. */ + + if (brazeroptr != NULL) *brazeroptr = OP_BRAPOSZERO; + if (repeat_min < 2) possessive_quantifier = FALSE; + } + + /* Non-possessive quantifier */ + + else *ketcode = OP_KETRMAX + repeat_type; + } + } + } + + /* If previous is OP_FAIL, it was generated by an empty class [] in + JavaScript mode. The other ways in which OP_FAIL can be generated, that is + by (*FAIL) or (?!) set previous to NULL, which gives a "nothing to repeat" + error above. We can just ignore the repeat in JS case. */ + + else if (*previous == OP_FAIL) goto END_REPEAT; + + /* Else there's some kind of shambles */ + + else + { + *errorcodeptr = ERR11; + goto FAILED; + } + + /* If the character following a repeat is '+', or if certain optimization + tests above succeeded, possessive_quantifier is TRUE. For some opcodes, + there are special alternative opcodes for this case. For anything else, we + wrap the entire repeated item inside OP_ONCE brackets. Logically, the '+' + notation is just syntactic sugar, taken from Sun's Java package, but the + special opcodes can optimize it. + + Some (but not all) possessively repeated subpatterns have already been + completely handled in the code just above. For them, possessive_quantifier + is always FALSE at this stage. + + Note that the repeated item starts at tempcode, not at previous, which + might be the first part of a string whose (former) last char we repeated. + + Possessifying an 'exact' quantifier has no effect, so we can ignore it. But + an 'upto' may follow. We skip over an 'exact' item, and then test the + length of what remains before proceeding. */ + + if (possessive_quantifier) + { + int len; + + if (*tempcode == OP_TYPEEXACT) + tempcode += PRIV(OP_lengths)[*tempcode] + + ((tempcode[1 + IMM2_SIZE] == OP_PROP + || tempcode[1 + IMM2_SIZE] == OP_NOTPROP)? 2 : 0); + + else if (*tempcode == OP_EXACT || *tempcode == OP_NOTEXACT) + { + tempcode += PRIV(OP_lengths)[*tempcode]; +#ifdef SUPPORT_UTF + if (utf && HAS_EXTRALEN(tempcode[-1])) + tempcode += GET_EXTRALEN(tempcode[-1]); +#endif + } + + len = (int)(code - tempcode); + if (len > 0) switch (*tempcode) + { + case OP_STAR: *tempcode = OP_POSSTAR; break; + case OP_PLUS: *tempcode = OP_POSPLUS; break; + case OP_QUERY: *tempcode = OP_POSQUERY; break; + case OP_UPTO: *tempcode = OP_POSUPTO; break; + + case OP_STARI: *tempcode = OP_POSSTARI; break; + case OP_PLUSI: *tempcode = OP_POSPLUSI; break; + case OP_QUERYI: *tempcode = OP_POSQUERYI; break; + case OP_UPTOI: *tempcode = OP_POSUPTOI; break; + + case OP_NOTSTAR: *tempcode = OP_NOTPOSSTAR; break; + case OP_NOTPLUS: *tempcode = OP_NOTPOSPLUS; break; + case OP_NOTQUERY: *tempcode = OP_NOTPOSQUERY; break; + case OP_NOTUPTO: *tempcode = OP_NOTPOSUPTO; break; + + case OP_NOTSTARI: *tempcode = OP_NOTPOSSTARI; break; + case OP_NOTPLUSI: *tempcode = OP_NOTPOSPLUSI; break; + case OP_NOTQUERYI: *tempcode = OP_NOTPOSQUERYI; break; + case OP_NOTUPTOI: *tempcode = OP_NOTPOSUPTOI; break; + + case OP_TYPESTAR: *tempcode = OP_TYPEPOSSTAR; break; + case OP_TYPEPLUS: *tempcode = OP_TYPEPOSPLUS; break; + case OP_TYPEQUERY: *tempcode = OP_TYPEPOSQUERY; break; + case OP_TYPEUPTO: *tempcode = OP_TYPEPOSUPTO; break; + + /* Because we are moving code along, we must ensure that any + pending recursive references are updated. */ + + default: + *code = OP_END; + adjust_recurse(tempcode, 1 + LINK_SIZE, utf, cd, save_hwm); + memmove(tempcode + 1 + LINK_SIZE, tempcode, IN_UCHARS(len)); + code += 1 + LINK_SIZE; + len += 1 + LINK_SIZE; + tempcode[0] = OP_ONCE; + *code++ = OP_KET; + PUTINC(code, 0, len); + PUT(tempcode, 1, len); + break; + } + } + + /* In all case we no longer have a previous item. We also set the + "follows varying string" flag for subsequently encountered reqchars if + it isn't already set and we have just passed a varying length item. */ + + END_REPEAT: + previous = NULL; + cd->req_varyopt |= reqvary; + break; + + + /* ===================================================================*/ + /* Start of nested parenthesized sub-expression, or comment or lookahead or + lookbehind or option setting or condition or all the other extended + parenthesis forms. */ + + case CHAR_LEFT_PARENTHESIS: + newoptions = options; + skipbytes = 0; + bravalue = OP_CBRA; + save_hwm = cd->hwm; + reset_bracount = FALSE; + + /* First deal with various "verbs" that can be introduced by '*'. */ + + ptr++; + if (ptr[0] == CHAR_ASTERISK && (ptr[1] == ':' + || (MAX_255(ptr[1]) && ((cd->ctypes[ptr[1]] & ctype_letter) != 0)))) + { + int i, namelen; + int arglen = 0; + const char *vn = verbnames; + const pcre_uchar *name = ptr + 1; + const pcre_uchar *arg = NULL; + previous = NULL; + ptr++; + while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_letter) != 0) ptr++; + namelen = (int)(ptr - name); + + /* It appears that Perl allows any characters whatsoever, other than + a closing parenthesis, to appear in arguments, so we no longer insist on + letters, digits, and underscores. */ + + if (*ptr == CHAR_COLON) + { + arg = ++ptr; + while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++; + arglen = (int)(ptr - arg); + if (arglen > (int)MAX_MARK) + { + *errorcodeptr = ERR75; + goto FAILED; + } + } + + if (*ptr != CHAR_RIGHT_PARENTHESIS) + { + *errorcodeptr = ERR60; + goto FAILED; + } + + /* Scan the table of verb names */ + + for (i = 0; i < verbcount; i++) + { + if (namelen == verbs[i].len && + STRNCMP_UC_C8(name, vn, namelen) == 0) + { + /* Check for open captures before ACCEPT and convert it to + ASSERT_ACCEPT if in an assertion. */ + + if (verbs[i].op == OP_ACCEPT) + { + open_capitem *oc; + if (arglen != 0) + { + *errorcodeptr = ERR59; + goto FAILED; + } + cd->had_accept = TRUE; + for (oc = cd->open_caps; oc != NULL; oc = oc->next) + { + *code++ = OP_CLOSE; + PUT2INC(code, 0, oc->number); + } + *code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT; + + /* Do not set firstchar after *ACCEPT */ + if (firstchar == REQ_UNSET) firstchar = REQ_NONE; + } + + /* Handle other cases with/without an argument */ + + else if (arglen == 0) + { + if (verbs[i].op < 0) /* Argument is mandatory */ + { + *errorcodeptr = ERR66; + goto FAILED; + } + *code = verbs[i].op; + if (*code++ == OP_THEN) cd->external_flags |= PCRE_HASTHEN; + } + + else + { + if (verbs[i].op_arg < 0) /* Argument is forbidden */ + { + *errorcodeptr = ERR59; + goto FAILED; + } + *code = verbs[i].op_arg; + if (*code++ == OP_THEN_ARG) cd->external_flags |= PCRE_HASTHEN; + *code++ = arglen; + memcpy(code, arg, IN_UCHARS(arglen)); + code += arglen; + *code++ = 0; + } + + break; /* Found verb, exit loop */ + } + + vn += verbs[i].len + 1; + } + + if (i < verbcount) continue; /* Successfully handled a verb */ + *errorcodeptr = ERR60; /* Verb not recognized */ + goto FAILED; + } + + /* Deal with the extended parentheses; all are introduced by '?', and the + appearance of any of them means that this is not a capturing group. */ + + else if (*ptr == CHAR_QUESTION_MARK) + { + int i, set, unset, namelen; + int *optset; + const pcre_uchar *name; + pcre_uchar *slot; + + switch (*(++ptr)) + { + case CHAR_NUMBER_SIGN: /* Comment; skip to ket */ + ptr++; + while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++; + if (*ptr == 0) + { + *errorcodeptr = ERR18; + goto FAILED; + } + continue; + + + /* ------------------------------------------------------------ */ + case CHAR_VERTICAL_LINE: /* Reset capture count for each branch */ + reset_bracount = TRUE; + /* Fall through */ + + /* ------------------------------------------------------------ */ + case CHAR_COLON: /* Non-capturing bracket */ + bravalue = OP_BRA; + ptr++; + break; + + + /* ------------------------------------------------------------ */ + case CHAR_LEFT_PARENTHESIS: + bravalue = OP_COND; /* Conditional group */ + + /* A condition can be an assertion, a number (referring to a numbered + group), a name (referring to a named group), or 'R', referring to + recursion. R and R&name are also permitted for recursion tests. + + There are several syntaxes for testing a named group: (?(name)) is used + by Python; Perl 5.10 onwards uses (?() or (?('name')). + + There are two unfortunate ambiguities, caused by history. (a) 'R' can + be the recursive thing or the name 'R' (and similarly for 'R' followed + by digits), and (b) a number could be a name that consists of digits. + In both cases, we look for a name first; if not found, we try the other + cases. */ + + /* For conditions that are assertions, check the syntax, and then exit + the switch. This will take control down to where bracketed groups, + including assertions, are processed. */ + + if (ptr[1] == CHAR_QUESTION_MARK && (ptr[2] == CHAR_EQUALS_SIGN || + ptr[2] == CHAR_EXCLAMATION_MARK || ptr[2] == CHAR_LESS_THAN_SIGN)) + break; + + /* Most other conditions use OP_CREF (a couple change to OP_RREF + below), and all need to skip 1+IMM2_SIZE bytes at the start of the group. */ + + code[1+LINK_SIZE] = OP_CREF; + skipbytes = 1+IMM2_SIZE; + refsign = -1; + + /* Check for a test for recursion in a named group. */ + + if (ptr[1] == CHAR_R && ptr[2] == CHAR_AMPERSAND) + { + terminator = -1; + ptr += 2; + code[1+LINK_SIZE] = OP_RREF; /* Change the type of test */ + } + + /* Check for a test for a named group's having been set, using the Perl + syntax (?() or (?('name') */ + + else if (ptr[1] == CHAR_LESS_THAN_SIGN) + { + terminator = CHAR_GREATER_THAN_SIGN; + ptr++; + } + else if (ptr[1] == CHAR_APOSTROPHE) + { + terminator = CHAR_APOSTROPHE; + ptr++; + } + else + { + terminator = 0; + if (ptr[1] == CHAR_MINUS || ptr[1] == CHAR_PLUS) refsign = *(++ptr); + } + + /* We now expect to read a name; any thing else is an error */ + + if (!MAX_255(ptr[1]) || (cd->ctypes[ptr[1]] & ctype_word) == 0) + { + ptr += 1; /* To get the right offset */ + *errorcodeptr = ERR28; + goto FAILED; + } + + /* Read the name, but also get it as a number if it's all digits */ + + recno = 0; + name = ++ptr; + while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) + { + if (recno >= 0) + recno = (IS_DIGIT(*ptr))? recno * 10 + *ptr - CHAR_0 : -1; + ptr++; + } + namelen = (int)(ptr - name); + + if ((terminator > 0 && *ptr++ != terminator) || + *ptr++ != CHAR_RIGHT_PARENTHESIS) + { + ptr--; /* Error offset */ + *errorcodeptr = ERR26; + goto FAILED; + } + + /* Do no further checking in the pre-compile phase. */ + + if (lengthptr != NULL) break; + + /* In the real compile we do the work of looking for the actual + reference. If the string started with "+" or "-" we require the rest to + be digits, in which case recno will be set. */ + + if (refsign > 0) + { + if (recno <= 0) + { + *errorcodeptr = ERR58; + goto FAILED; + } + recno = (refsign == CHAR_MINUS)? + cd->bracount - recno + 1 : recno +cd->bracount; + if (recno <= 0 || recno > cd->final_bracount) + { + *errorcodeptr = ERR15; + goto FAILED; + } + PUT2(code, 2+LINK_SIZE, recno); + break; + } + + /* Otherwise (did not start with "+" or "-"), start by looking for the + name. If we find a name, add one to the opcode to change OP_CREF or + OP_RREF into OP_NCREF or OP_NRREF. These behave exactly the same, + except they record that the reference was originally to a name. The + information is used to check duplicate names. */ + + slot = cd->name_table; + for (i = 0; i < cd->names_found; i++) + { + if (STRNCMP_UC_UC(name, slot+IMM2_SIZE, namelen) == 0) break; + slot += cd->name_entry_size; + } + + /* Found a previous named subpattern */ + + if (i < cd->names_found) + { + recno = GET2(slot, 0); + PUT2(code, 2+LINK_SIZE, recno); + code[1+LINK_SIZE]++; + } + + /* Search the pattern for a forward reference */ + + else if ((i = find_parens(cd, name, namelen, + (options & PCRE_EXTENDED) != 0, utf)) > 0) + { + PUT2(code, 2+LINK_SIZE, i); + code[1+LINK_SIZE]++; + } + + /* If terminator == 0 it means that the name followed directly after + the opening parenthesis [e.g. (?(abc)...] and in this case there are + some further alternatives to try. For the cases where terminator != 0 + [things like (?(... or (?('name')... or (?(R&name)... ] we have + now checked all the possibilities, so give an error. */ + + else if (terminator != 0) + { + *errorcodeptr = ERR15; + goto FAILED; + } + + /* Check for (?(R) for recursion. Allow digits after R to specify a + specific group number. */ + + else if (*name == CHAR_R) + { + recno = 0; + for (i = 1; i < namelen; i++) + { + if (!IS_DIGIT(name[i])) + { + *errorcodeptr = ERR15; + goto FAILED; + } + recno = recno * 10 + name[i] - CHAR_0; + } + if (recno == 0) recno = RREF_ANY; + code[1+LINK_SIZE] = OP_RREF; /* Change test type */ + PUT2(code, 2+LINK_SIZE, recno); + } + + /* Similarly, check for the (?(DEFINE) "condition", which is always + false. */ + + else if (namelen == 6 && STRNCMP_UC_C8(name, STRING_DEFINE, 6) == 0) + { + code[1+LINK_SIZE] = OP_DEF; + skipbytes = 1; + } + + /* Check for the "name" actually being a subpattern number. We are + in the second pass here, so final_bracount is set. */ + + else if (recno > 0 && recno <= cd->final_bracount) + { + PUT2(code, 2+LINK_SIZE, recno); + } + + /* Either an unidentified subpattern, or a reference to (?(0) */ + + else + { + *errorcodeptr = (recno == 0)? ERR35: ERR15; + goto FAILED; + } + break; + + + /* ------------------------------------------------------------ */ + case CHAR_EQUALS_SIGN: /* Positive lookahead */ + bravalue = OP_ASSERT; + cd->assert_depth += 1; + ptr++; + break; + + + /* ------------------------------------------------------------ */ + case CHAR_EXCLAMATION_MARK: /* Negative lookahead */ + ptr++; + if (*ptr == CHAR_RIGHT_PARENTHESIS) /* Optimize (?!) */ + { + *code++ = OP_FAIL; + previous = NULL; + continue; + } + bravalue = OP_ASSERT_NOT; + cd->assert_depth += 1; + break; + + + /* ------------------------------------------------------------ */ + case CHAR_LESS_THAN_SIGN: /* Lookbehind or named define */ + switch (ptr[1]) + { + case CHAR_EQUALS_SIGN: /* Positive lookbehind */ + bravalue = OP_ASSERTBACK; + cd->assert_depth += 1; + ptr += 2; + break; + + case CHAR_EXCLAMATION_MARK: /* Negative lookbehind */ + bravalue = OP_ASSERTBACK_NOT; + cd->assert_depth += 1; + ptr += 2; + break; + + default: /* Could be name define, else bad */ + if (MAX_255(ptr[1]) && (cd->ctypes[ptr[1]] & ctype_word) != 0) + goto DEFINE_NAME; + ptr++; /* Correct offset for error */ + *errorcodeptr = ERR24; + goto FAILED; + } + break; + + + /* ------------------------------------------------------------ */ + case CHAR_GREATER_THAN_SIGN: /* One-time brackets */ + bravalue = OP_ONCE; + ptr++; + break; + + + /* ------------------------------------------------------------ */ + case CHAR_C: /* Callout - may be followed by digits; */ + previous_callout = code; /* Save for later completion */ + after_manual_callout = 1; /* Skip one item before completing */ + *code++ = OP_CALLOUT; + { + int n = 0; + ptr++; + while(IS_DIGIT(*ptr)) + n = n * 10 + *ptr++ - CHAR_0; + if (*ptr != CHAR_RIGHT_PARENTHESIS) + { + *errorcodeptr = ERR39; + goto FAILED; + } + if (n > 255) + { + *errorcodeptr = ERR38; + goto FAILED; + } + *code++ = n; + PUT(code, 0, (int)(ptr - cd->start_pattern + 1)); /* Pattern offset */ + PUT(code, LINK_SIZE, 0); /* Default length */ + code += 2 * LINK_SIZE; + } + previous = NULL; + continue; + + + /* ------------------------------------------------------------ */ + case CHAR_P: /* Python-style named subpattern handling */ + if (*(++ptr) == CHAR_EQUALS_SIGN || + *ptr == CHAR_GREATER_THAN_SIGN) /* Reference or recursion */ + { + is_recurse = *ptr == CHAR_GREATER_THAN_SIGN; + terminator = CHAR_RIGHT_PARENTHESIS; + goto NAMED_REF_OR_RECURSE; + } + else if (*ptr != CHAR_LESS_THAN_SIGN) /* Test for Python-style defn */ + { + *errorcodeptr = ERR41; + goto FAILED; + } + /* Fall through to handle (?P< as (?< is handled */ + + + /* ------------------------------------------------------------ */ + DEFINE_NAME: /* Come here from (?< handling */ + case CHAR_APOSTROPHE: + { + terminator = (*ptr == CHAR_LESS_THAN_SIGN)? + CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE; + name = ++ptr; + + while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) ptr++; + namelen = (int)(ptr - name); + + /* In the pre-compile phase, just do a syntax check. */ + + if (lengthptr != NULL) + { + if (*ptr != terminator) + { + *errorcodeptr = ERR42; + goto FAILED; + } + if (cd->names_found >= MAX_NAME_COUNT) + { + *errorcodeptr = ERR49; + goto FAILED; + } + if (namelen + IMM2_SIZE + 1 > cd->name_entry_size) + { + cd->name_entry_size = namelen + IMM2_SIZE + 1; + if (namelen > MAX_NAME_SIZE) + { + *errorcodeptr = ERR48; + goto FAILED; + } + } + } + + /* In the real compile, create the entry in the table, maintaining + alphabetical order. Duplicate names for different numbers are + permitted only if PCRE_DUPNAMES is set. Duplicate names for the same + number are always OK. (An existing number can be re-used if (?| + appears in the pattern.) In either event, a duplicate name results in + a duplicate entry in the table, even if the number is the same. This + is because the number of names, and hence the table size, is computed + in the pre-compile, and it affects various numbers and pointers which + would all have to be modified, and the compiled code moved down, if + duplicates with the same number were omitted from the table. This + doesn't seem worth the hassle. However, *different* names for the + same number are not permitted. */ + + else + { + BOOL dupname = FALSE; + slot = cd->name_table; + + for (i = 0; i < cd->names_found; i++) + { + int crc = memcmp(name, slot+IMM2_SIZE, IN_UCHARS(namelen)); + if (crc == 0) + { + if (slot[IMM2_SIZE+namelen] == 0) + { + if (GET2(slot, 0) != cd->bracount + 1 && + (options & PCRE_DUPNAMES) == 0) + { + *errorcodeptr = ERR43; + goto FAILED; + } + else dupname = TRUE; + } + else crc = -1; /* Current name is a substring */ + } + + /* Make space in the table and break the loop for an earlier + name. For a duplicate or later name, carry on. We do this for + duplicates so that in the simple case (when ?(| is not used) they + are in order of their numbers. */ + + if (crc < 0) + { + memmove(slot + cd->name_entry_size, slot, + IN_UCHARS((cd->names_found - i) * cd->name_entry_size)); + break; + } + + /* Continue the loop for a later or duplicate name */ + + slot += cd->name_entry_size; + } + + /* For non-duplicate names, check for a duplicate number before + adding the new name. */ + + if (!dupname) + { + pcre_uchar *cslot = cd->name_table; + for (i = 0; i < cd->names_found; i++) + { + if (cslot != slot) + { + if (GET2(cslot, 0) == cd->bracount + 1) + { + *errorcodeptr = ERR65; + goto FAILED; + } + } + else i--; + cslot += cd->name_entry_size; + } + } + + PUT2(slot, 0, cd->bracount + 1); + memcpy(slot + IMM2_SIZE, name, IN_UCHARS(namelen)); + slot[IMM2_SIZE + namelen] = 0; + } + } + + /* In both pre-compile and compile, count the number of names we've + encountered. */ + + cd->names_found++; + ptr++; /* Move past > or ' */ + goto NUMBERED_GROUP; + + + /* ------------------------------------------------------------ */ + case CHAR_AMPERSAND: /* Perl recursion/subroutine syntax */ + terminator = CHAR_RIGHT_PARENTHESIS; + is_recurse = TRUE; + /* Fall through */ + + /* We come here from the Python syntax above that handles both + references (?P=name) and recursion (?P>name), as well as falling + through from the Perl recursion syntax (?&name). We also come here from + the Perl \k or \k'name' back reference syntax and the \k{name} + .NET syntax, and the Oniguruma \g<...> and \g'...' subroutine syntax. */ + + NAMED_REF_OR_RECURSE: + name = ++ptr; + while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) ptr++; + namelen = (int)(ptr - name); + + /* In the pre-compile phase, do a syntax check. We used to just set + a dummy reference number, because it was not used in the first pass. + However, with the change of recursive back references to be atomic, + we have to look for the number so that this state can be identified, as + otherwise the incorrect length is computed. If it's not a backwards + reference, the dummy number will do. */ + + if (lengthptr != NULL) + { + const pcre_uchar *temp; + + if (namelen == 0) + { + *errorcodeptr = ERR62; + goto FAILED; + } + if (*ptr != terminator) + { + *errorcodeptr = ERR42; + goto FAILED; + } + if (namelen > MAX_NAME_SIZE) + { + *errorcodeptr = ERR48; + goto FAILED; + } + + /* The name table does not exist in the first pass, so we cannot + do a simple search as in the code below. Instead, we have to scan the + pattern to find the number. It is important that we scan it only as + far as we have got because the syntax of named subpatterns has not + been checked for the rest of the pattern, and find_parens() assumes + correct syntax. In any case, it's a waste of resources to scan + further. We stop the scan at the current point by temporarily + adjusting the value of cd->endpattern. */ + + temp = cd->end_pattern; + cd->end_pattern = ptr; + recno = find_parens(cd, name, namelen, + (options & PCRE_EXTENDED) != 0, utf); + cd->end_pattern = temp; + if (recno < 0) recno = 0; /* Forward ref; set dummy number */ + } + + /* In the real compile, seek the name in the table. We check the name + first, and then check that we have reached the end of the name in the + table. That way, if the name that is longer than any in the table, + the comparison will fail without reading beyond the table entry. */ + + else + { + slot = cd->name_table; + for (i = 0; i < cd->names_found; i++) + { + if (STRNCMP_UC_UC(name, slot+IMM2_SIZE, namelen) == 0 && + slot[IMM2_SIZE+namelen] == 0) + break; + slot += cd->name_entry_size; + } + + if (i < cd->names_found) /* Back reference */ + { + recno = GET2(slot, 0); + } + else if ((recno = /* Forward back reference */ + find_parens(cd, name, namelen, + (options & PCRE_EXTENDED) != 0, utf)) <= 0) + { + *errorcodeptr = ERR15; + goto FAILED; + } + } + + /* In both phases, we can now go to the code than handles numerical + recursion or backreferences. */ + + if (is_recurse) goto HANDLE_RECURSION; + else goto HANDLE_REFERENCE; + + + /* ------------------------------------------------------------ */ + case CHAR_R: /* Recursion */ + ptr++; /* Same as (?0) */ + /* Fall through */ + + + /* ------------------------------------------------------------ */ + case CHAR_MINUS: case CHAR_PLUS: /* Recursion or subroutine */ + case CHAR_0: case CHAR_1: case CHAR_2: case CHAR_3: case CHAR_4: + case CHAR_5: case CHAR_6: case CHAR_7: case CHAR_8: case CHAR_9: + { + const pcre_uchar *called; + terminator = CHAR_RIGHT_PARENTHESIS; + + /* Come here from the \g<...> and \g'...' code (Oniguruma + compatibility). However, the syntax has been checked to ensure that + the ... are a (signed) number, so that neither ERR63 nor ERR29 will + be called on this path, nor with the jump to OTHER_CHAR_AFTER_QUERY + ever be taken. */ + + HANDLE_NUMERICAL_RECURSION: + + if ((refsign = *ptr) == CHAR_PLUS) + { + ptr++; + if (!IS_DIGIT(*ptr)) + { + *errorcodeptr = ERR63; + goto FAILED; + } + } + else if (refsign == CHAR_MINUS) + { + if (!IS_DIGIT(ptr[1])) + goto OTHER_CHAR_AFTER_QUERY; + ptr++; + } + + recno = 0; + while(IS_DIGIT(*ptr)) + recno = recno * 10 + *ptr++ - CHAR_0; + + if (*ptr != terminator) + { + *errorcodeptr = ERR29; + goto FAILED; + } + + if (refsign == CHAR_MINUS) + { + if (recno == 0) + { + *errorcodeptr = ERR58; + goto FAILED; + } + recno = cd->bracount - recno + 1; + if (recno <= 0) + { + *errorcodeptr = ERR15; + goto FAILED; + } + } + else if (refsign == CHAR_PLUS) + { + if (recno == 0) + { + *errorcodeptr = ERR58; + goto FAILED; + } + recno += cd->bracount; + } + + /* Come here from code above that handles a named recursion */ + + HANDLE_RECURSION: + + previous = code; + called = cd->start_code; + + /* When we are actually compiling, find the bracket that is being + referenced. Temporarily end the regex in case it doesn't exist before + this point. If we end up with a forward reference, first check that + the bracket does occur later so we can give the error (and position) + now. Then remember this forward reference in the workspace so it can + be filled in at the end. */ + + if (lengthptr == NULL) + { + *code = OP_END; + if (recno != 0) + called = PRIV(find_bracket)(cd->start_code, utf, recno); + + /* Forward reference */ + + if (called == NULL) + { + if (find_parens(cd, NULL, recno, + (options & PCRE_EXTENDED) != 0, utf) < 0) + { + *errorcodeptr = ERR15; + goto FAILED; + } + + /* Fudge the value of "called" so that when it is inserted as an + offset below, what it actually inserted is the reference number + of the group. Then remember the forward reference. */ + + called = cd->start_code + recno; + if (cd->hwm >= cd->start_workspace + cd->workspace_size - + WORK_SIZE_SAFETY_MARGIN) + { + *errorcodeptr = expand_workspace(cd); + if (*errorcodeptr != 0) goto FAILED; + } + PUTINC(cd->hwm, 0, (int)(code + 1 - cd->start_code)); + } + + /* If not a forward reference, and the subpattern is still open, + this is a recursive call. We check to see if this is a left + recursion that could loop for ever, and diagnose that case. We + must not, however, do this check if we are in a conditional + subpattern because the condition might be testing for recursion in + a pattern such as /(?(R)a+|(?R)b)/, which is perfectly valid. + Forever loops are also detected at runtime, so those that occur in + conditional subpatterns will be picked up then. */ + + else if (GET(called, 1) == 0 && cond_depth <= 0 && + could_be_empty(called, code, bcptr, utf, cd)) + { + *errorcodeptr = ERR40; + goto FAILED; + } + } + + /* Insert the recursion/subroutine item. It does not have a set first + character (relevant if it is repeated, because it will then be + wrapped with ONCE brackets). */ + + *code = OP_RECURSE; + PUT(code, 1, (int)(called - cd->start_code)); + code += 1 + LINK_SIZE; + groupsetfirstchar = FALSE; + } + + /* Can't determine a first byte now */ + + if (firstchar == REQ_UNSET) firstchar = REQ_NONE; + continue; + + + /* ------------------------------------------------------------ */ + default: /* Other characters: check option setting */ + OTHER_CHAR_AFTER_QUERY: + set = unset = 0; + optset = &set; + + while (*ptr != CHAR_RIGHT_PARENTHESIS && *ptr != CHAR_COLON) + { + switch (*ptr++) + { + case CHAR_MINUS: optset = &unset; break; + + case CHAR_J: /* Record that it changed in the external options */ + *optset |= PCRE_DUPNAMES; + cd->external_flags |= PCRE_JCHANGED; + break; + + case CHAR_i: *optset |= PCRE_CASELESS; break; + case CHAR_m: *optset |= PCRE_MULTILINE; break; + case CHAR_s: *optset |= PCRE_DOTALL; break; + case CHAR_x: *optset |= PCRE_EXTENDED; break; + case CHAR_U: *optset |= PCRE_UNGREEDY; break; + case CHAR_X: *optset |= PCRE_EXTRA; break; + + default: *errorcodeptr = ERR12; + ptr--; /* Correct the offset */ + goto FAILED; + } + } + + /* Set up the changed option bits, but don't change anything yet. */ + + newoptions = (options | set) & (~unset); + + /* If the options ended with ')' this is not the start of a nested + group with option changes, so the options change at this level. If this + item is right at the start of the pattern, the options can be + abstracted and made external in the pre-compile phase, and ignored in + the compile phase. This can be helpful when matching -- for instance in + caseless checking of required bytes. + + If the code pointer is not (cd->start_code + 1 + LINK_SIZE), we are + definitely *not* at the start of the pattern because something has been + compiled. In the pre-compile phase, however, the code pointer can have + that value after the start, because it gets reset as code is discarded + during the pre-compile. However, this can happen only at top level - if + we are within parentheses, the starting BRA will still be present. At + any parenthesis level, the length value can be used to test if anything + has been compiled at that level. Thus, a test for both these conditions + is necessary to ensure we correctly detect the start of the pattern in + both phases. + + If we are not at the pattern start, reset the greedy defaults and the + case value for firstchar and reqchar. */ + + if (*ptr == CHAR_RIGHT_PARENTHESIS) + { + if (code == cd->start_code + 1 + LINK_SIZE && + (lengthptr == NULL || *lengthptr == 2 + 2*LINK_SIZE)) + { + cd->external_options = newoptions; + } + else + { + greedy_default = ((newoptions & PCRE_UNGREEDY) != 0); + greedy_non_default = greedy_default ^ 1; + req_caseopt = ((newoptions & PCRE_CASELESS) != 0)? REQ_CASELESS:0; + } + + /* Change options at this level, and pass them back for use + in subsequent branches. */ + + *optionsptr = options = newoptions; + previous = NULL; /* This item can't be repeated */ + continue; /* It is complete */ + } + + /* If the options ended with ':' we are heading into a nested group + with possible change of options. Such groups are non-capturing and are + not assertions of any kind. All we need to do is skip over the ':'; + the newoptions value is handled below. */ + + bravalue = OP_BRA; + ptr++; + } /* End of switch for character following (? */ + } /* End of (? handling */ + + /* Opening parenthesis not followed by '*' or '?'. If PCRE_NO_AUTO_CAPTURE + is set, all unadorned brackets become non-capturing and behave like (?:...) + brackets. */ + + else if ((options & PCRE_NO_AUTO_CAPTURE) != 0) + { + bravalue = OP_BRA; + } + + /* Else we have a capturing group. */ + + else + { + NUMBERED_GROUP: + cd->bracount += 1; + PUT2(code, 1+LINK_SIZE, cd->bracount); + skipbytes = IMM2_SIZE; + } + + /* Process nested bracketed regex. Assertions used not to be repeatable, + but this was changed for Perl compatibility, so all kinds can now be + repeated. We copy code into a non-register variable (tempcode) in order to + be able to pass its address because some compilers complain otherwise. */ + + previous = code; /* For handling repetition */ + *code = bravalue; + tempcode = code; + tempreqvary = cd->req_varyopt; /* Save value before bracket */ + tempbracount = cd->bracount; /* Save value before bracket */ + length_prevgroup = 0; /* Initialize for pre-compile phase */ + + if (!compile_regex( + newoptions, /* The complete new option state */ + &tempcode, /* Where to put code (updated) */ + &ptr, /* Input pointer (updated) */ + errorcodeptr, /* Where to put an error message */ + (bravalue == OP_ASSERTBACK || + bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */ + reset_bracount, /* True if (?| group */ + skipbytes, /* Skip over bracket number */ + cond_depth + + ((bravalue == OP_COND)?1:0), /* Depth of condition subpatterns */ + &subfirstchar, /* For possible first char */ + &subreqchar, /* For possible last char */ + bcptr, /* Current branch chain */ + cd, /* Tables block */ + (lengthptr == NULL)? NULL : /* Actual compile phase */ + &length_prevgroup /* Pre-compile phase */ + )) + goto FAILED; + + /* If this was an atomic group and there are no capturing groups within it, + generate OP_ONCE_NC instead of OP_ONCE. */ + + if (bravalue == OP_ONCE && cd->bracount <= tempbracount) + *code = OP_ONCE_NC; + + if (bravalue >= OP_ASSERT && bravalue <= OP_ASSERTBACK_NOT) + cd->assert_depth -= 1; + + /* At the end of compiling, code is still pointing to the start of the + group, while tempcode has been updated to point past the end of the group. + The pattern pointer (ptr) is on the bracket. + + If this is a conditional bracket, check that there are no more than + two branches in the group, or just one if it's a DEFINE group. We do this + in the real compile phase, not in the pre-pass, where the whole group may + not be available. */ + + if (bravalue == OP_COND && lengthptr == NULL) + { + pcre_uchar *tc = code; + int condcount = 0; + + do { + condcount++; + tc += GET(tc,1); + } + while (*tc != OP_KET); + + /* A DEFINE group is never obeyed inline (the "condition" is always + false). It must have only one branch. */ + + if (code[LINK_SIZE+1] == OP_DEF) + { + if (condcount > 1) + { + *errorcodeptr = ERR54; + goto FAILED; + } + bravalue = OP_DEF; /* Just a flag to suppress char handling below */ + } + + /* A "normal" conditional group. If there is just one branch, we must not + make use of its firstchar or reqchar, because this is equivalent to an + empty second branch. */ + + else + { + if (condcount > 2) + { + *errorcodeptr = ERR27; + goto FAILED; + } + if (condcount == 1) subfirstchar = subreqchar = REQ_NONE; + } + } + + /* Error if hit end of pattern */ + + if (*ptr != CHAR_RIGHT_PARENTHESIS) + { + *errorcodeptr = ERR14; + goto FAILED; + } + + /* In the pre-compile phase, update the length by the length of the group, + less the brackets at either end. Then reduce the compiled code to just a + set of non-capturing brackets so that it doesn't use much memory if it is + duplicated by a quantifier.*/ + + if (lengthptr != NULL) + { + if (OFLOW_MAX - *lengthptr < length_prevgroup - 2 - 2*LINK_SIZE) + { + *errorcodeptr = ERR20; + goto FAILED; + } + *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE; + code++; /* This already contains bravalue */ + PUTINC(code, 0, 1 + LINK_SIZE); + *code++ = OP_KET; + PUTINC(code, 0, 1 + LINK_SIZE); + break; /* No need to waste time with special character handling */ + } + + /* Otherwise update the main code pointer to the end of the group. */ + + code = tempcode; + + /* For a DEFINE group, required and first character settings are not + relevant. */ + + if (bravalue == OP_DEF) break; + + /* Handle updating of the required and first characters for other types of + group. Update for normal brackets of all kinds, and conditions with two + branches (see code above). If the bracket is followed by a quantifier with + zero repeat, we have to back off. Hence the definition of zeroreqchar and + zerofirstchar outside the main loop so that they can be accessed for the + back off. */ + + zeroreqchar = reqchar; + zerofirstchar = firstchar; + groupsetfirstchar = FALSE; + + if (bravalue >= OP_ONCE) + { + /* If we have not yet set a firstchar in this branch, take it from the + subpattern, remembering that it was set here so that a repeat of more + than one can replicate it as reqchar if necessary. If the subpattern has + no firstchar, set "none" for the whole branch. In both cases, a zero + repeat forces firstchar to "none". */ + + if (firstchar == REQ_UNSET) + { + if (subfirstchar >= 0) + { + firstchar = subfirstchar; + groupsetfirstchar = TRUE; + } + else firstchar = REQ_NONE; + zerofirstchar = REQ_NONE; + } + + /* If firstchar was previously set, convert the subpattern's firstchar + into reqchar if there wasn't one, using the vary flag that was in + existence beforehand. */ + + else if (subfirstchar >= 0 && subreqchar < 0) + subreqchar = subfirstchar | tempreqvary; + + /* If the subpattern set a required byte (or set a first byte that isn't + really the first byte - see above), set it. */ + + if (subreqchar >= 0) reqchar = subreqchar; + } + + /* For a forward assertion, we take the reqchar, if set. This can be + helpful if the pattern that follows the assertion doesn't set a different + char. For example, it's useful for /(?=abcde).+/. We can't set firstchar + for an assertion, however because it leads to incorrect effect for patterns + such as /(?=a)a.+/ when the "real" "a" would then become a reqchar instead + of a firstchar. This is overcome by a scan at the end if there's no + firstchar, looking for an asserted first char. */ + + else if (bravalue == OP_ASSERT && subreqchar >= 0) reqchar = subreqchar; + break; /* End of processing '(' */ + + + /* ===================================================================*/ + /* Handle metasequences introduced by \. For ones like \d, the ESC_ values + are arranged to be the negation of the corresponding OP_values in the + default case when PCRE_UCP is not set. For the back references, the values + are ESC_REF plus the reference number. Only back references and those types + that consume a character may be repeated. We can test for values between + ESC_b and ESC_Z for the latter; this may have to change if any new ones are + ever created. */ + + case CHAR_BACKSLASH: + tempptr = ptr; + c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE); + if (*errorcodeptr != 0) goto FAILED; + + if (c < 0) + { + if (-c == ESC_Q) /* Handle start of quoted string */ + { + if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E) + ptr += 2; /* avoid empty string */ + else inescq = TRUE; + continue; + } + + if (-c == ESC_E) continue; /* Perl ignores an orphan \E */ + + /* For metasequences that actually match a character, we disable the + setting of a first character if it hasn't already been set. */ + + if (firstchar == REQ_UNSET && -c > ESC_b && -c < ESC_Z) + firstchar = REQ_NONE; + + /* Set values to reset to if this is followed by a zero repeat. */ + + zerofirstchar = firstchar; + zeroreqchar = reqchar; + + /* \g or \g'name' is a subroutine call by name and \g or \g'n' + is a subroutine call by number (Oniguruma syntax). In fact, the value + -ESC_g is returned only for these cases. So we don't need to check for < + or ' if the value is -ESC_g. For the Perl syntax \g{n} the value is + -ESC_REF+n, and for the Perl syntax \g{name} the result is -ESC_k (as + that is a synonym for a named back reference). */ + + if (-c == ESC_g) + { + const pcre_uchar *p; + save_hwm = cd->hwm; /* Normally this is set when '(' is read */ + terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)? + CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE; + + /* These two statements stop the compiler for warning about possibly + unset variables caused by the jump to HANDLE_NUMERICAL_RECURSION. In + fact, because we actually check for a number below, the paths that + would actually be in error are never taken. */ + + skipbytes = 0; + reset_bracount = FALSE; + + /* Test for a name */ + + if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS) + { + BOOL is_a_number = TRUE; + for (p = ptr + 1; *p != 0 && *p != terminator; p++) + { + if (!MAX_255(*p)) { is_a_number = FALSE; break; } + if ((cd->ctypes[*p] & ctype_digit) == 0) is_a_number = FALSE; + if ((cd->ctypes[*p] & ctype_word) == 0) break; + } + if (*p != terminator) + { + *errorcodeptr = ERR57; + break; + } + if (is_a_number) + { + ptr++; + goto HANDLE_NUMERICAL_RECURSION; + } + is_recurse = TRUE; + goto NAMED_REF_OR_RECURSE; + } + + /* Test a signed number in angle brackets or quotes. */ + + p = ptr + 2; + while (IS_DIGIT(*p)) p++; + if (*p != terminator) + { + *errorcodeptr = ERR57; + break; + } + ptr++; + goto HANDLE_NUMERICAL_RECURSION; + } + + /* \k or \k'name' is a back reference by name (Perl syntax). + We also support \k{name} (.NET syntax). */ + + if (-c == ESC_k) + { + if ((ptr[1] != CHAR_LESS_THAN_SIGN && + ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET)) + { + *errorcodeptr = ERR69; + break; + } + is_recurse = FALSE; + terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)? + CHAR_GREATER_THAN_SIGN : (*ptr == CHAR_APOSTROPHE)? + CHAR_APOSTROPHE : CHAR_RIGHT_CURLY_BRACKET; + goto NAMED_REF_OR_RECURSE; + } + + /* Back references are handled specially; must disable firstchar if + not set to cope with cases like (?=(\w+))\1: which would otherwise set + ':' later. */ + + if (-c >= ESC_REF) + { + open_capitem *oc; + recno = -c - ESC_REF; + + HANDLE_REFERENCE: /* Come here from named backref handling */ + if (firstchar == REQ_UNSET) firstchar = REQ_NONE; + previous = code; + *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF; + PUT2INC(code, 0, recno); + cd->backref_map |= (recno < 32)? (1 << recno) : 1; + if (recno > cd->top_backref) cd->top_backref = recno; + + /* Check to see if this back reference is recursive, that it, it + is inside the group that it references. A flag is set so that the + group can be made atomic. */ + + for (oc = cd->open_caps; oc != NULL; oc = oc->next) + { + if (oc->number == recno) + { + oc->flag = TRUE; + break; + } + } + } + + /* So are Unicode property matches, if supported. */ + +#ifdef SUPPORT_UCP + else if (-c == ESC_P || -c == ESC_p) + { + BOOL negated; + int pdata; + int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr); + if (ptype < 0) goto FAILED; + previous = code; + *code++ = ((-c == ESC_p) != negated)? OP_PROP : OP_NOTPROP; + *code++ = ptype; + *code++ = pdata; + } +#else + + /* If Unicode properties are not supported, \X, \P, and \p are not + allowed. */ + + else if (-c == ESC_X || -c == ESC_P || -c == ESC_p) + { + *errorcodeptr = ERR45; + goto FAILED; + } +#endif + + /* For the rest (including \X when Unicode properties are supported), we + can obtain the OP value by negating the escape value in the default + situation when PCRE_UCP is not set. When it *is* set, we substitute + Unicode property tests. Note that \b and \B do a one-character + lookbehind. */ + + else + { + if ((-c == ESC_b || -c == ESC_B) && cd->max_lookbehind == 0) + cd->max_lookbehind = 1; +#ifdef SUPPORT_UCP + if (-c >= ESC_DU && -c <= ESC_wu) + { + nestptr = ptr + 1; /* Where to resume */ + ptr = substitutes[-c - ESC_DU] - 1; /* Just before substitute */ + } + else +#endif + /* In non-UTF-8 mode, we turn \C into OP_ALLANY instead of OP_ANYBYTE + so that it works in DFA mode and in lookbehinds. */ + + { + previous = (-c > ESC_b && -c < ESC_Z)? code : NULL; + *code++ = (!utf && c == -ESC_C)? OP_ALLANY : -c; + } + } + continue; + } + + /* We have a data character whose value is in c. In UTF-8 mode it may have + a value > 127. We set its representation in the length/buffer, and then + handle it as a data character. */ + +#ifdef SUPPORT_UTF + if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR) + mclength = PRIV(ord2utf)(c, mcbuffer); + else +#endif + + { + mcbuffer[0] = c; + mclength = 1; + } + goto ONE_CHAR; + + + /* ===================================================================*/ + /* Handle a literal character. It is guaranteed not to be whitespace or # + when the extended flag is set. If we are in UTF-8 mode, it may be a + multi-byte literal character. */ + + default: + NORMAL_CHAR: + mclength = 1; + mcbuffer[0] = c; + +#ifdef SUPPORT_UTF + if (utf && HAS_EXTRALEN(c)) + ACROSSCHAR(TRUE, ptr[1], mcbuffer[mclength++] = *(++ptr)); +#endif + + /* At this point we have the character's bytes in mcbuffer, and the length + in mclength. When not in UTF-8 mode, the length is always 1. */ + + ONE_CHAR: + previous = code; + *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARI : OP_CHAR; + for (c = 0; c < mclength; c++) *code++ = mcbuffer[c]; + + /* Remember if \r or \n were seen */ + + if (mcbuffer[0] == CHAR_CR || mcbuffer[0] == CHAR_NL) + cd->external_flags |= PCRE_HASCRORLF; + + /* Set the first and required bytes appropriately. If no previous first + byte, set it from this character, but revert to none on a zero repeat. + Otherwise, leave the firstchar value alone, and don't change it on a zero + repeat. */ + + if (firstchar == REQ_UNSET) + { + zerofirstchar = REQ_NONE; + zeroreqchar = reqchar; + + /* If the character is more than one byte long, we can set firstchar + only if it is not to be matched caselessly. */ + + if (mclength == 1 || req_caseopt == 0) + { + firstchar = mcbuffer[0] | req_caseopt; + if (mclength != 1) reqchar = code[-1] | cd->req_varyopt; + } + else firstchar = reqchar = REQ_NONE; + } + + /* firstchar was previously set; we can set reqchar only if the length is + 1 or the matching is caseful. */ + + else + { + zerofirstchar = firstchar; + zeroreqchar = reqchar; + if (mclength == 1 || req_caseopt == 0) + reqchar = code[-1] | req_caseopt | cd->req_varyopt; + } + + break; /* End of literal character handling */ + } + } /* end of big loop */ + + +/* Control never reaches here by falling through, only by a goto for all the +error states. Pass back the position in the pattern so that it can be displayed +to the user for diagnosing the error. */ + +FAILED: +*ptrptr = ptr; +return FALSE; +} + + + + +/************************************************* +* Compile sequence of alternatives * +*************************************************/ + +/* On entry, ptr is pointing past the bracket character, but on return it +points to the closing bracket, or vertical bar, or end of string. The code +variable is pointing at the byte into which the BRA operator has been stored. +This function is used during the pre-compile phase when we are trying to find +out the amount of memory needed, as well as during the real compile phase. The +value of lengthptr distinguishes the two phases. + +Arguments: + options option bits, including any changes for this subpattern + codeptr -> the address of the current code pointer + ptrptr -> the address of the current pattern pointer + errorcodeptr -> pointer to error code variable + lookbehind TRUE if this is a lookbehind assertion + reset_bracount TRUE to reset the count for each branch + skipbytes skip this many bytes at start (for brackets and OP_COND) + cond_depth depth of nesting for conditional subpatterns + firstcharptr place to put the first required character, or a negative number + reqcharptr place to put the last required character, or a negative number + bcptr pointer to the chain of currently open branches + cd points to the data block with tables pointers etc. + lengthptr NULL during the real compile phase + points to length accumulator during pre-compile phase + +Returns: TRUE on success +*/ + +static BOOL +compile_regex(int options, pcre_uchar **codeptr, const pcre_uchar **ptrptr, + int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes, + int cond_depth, pcre_int32 *firstcharptr, pcre_int32 *reqcharptr, + branch_chain *bcptr, compile_data *cd, int *lengthptr) +{ +const pcre_uchar *ptr = *ptrptr; +pcre_uchar *code = *codeptr; +pcre_uchar *last_branch = code; +pcre_uchar *start_bracket = code; +pcre_uchar *reverse_count = NULL; +open_capitem capitem; +int capnumber = 0; +pcre_int32 firstchar, reqchar; +pcre_int32 branchfirstchar, branchreqchar; +int length; +int orig_bracount; +int max_bracount; +branch_chain bc; + +bc.outer = bcptr; +bc.current_branch = code; + +firstchar = reqchar = REQ_UNSET; + +/* Accumulate the length for use in the pre-compile phase. Start with the +length of the BRA and KET and any extra bytes that are required at the +beginning. We accumulate in a local variable to save frequent testing of +lenthptr for NULL. We cannot do this by looking at the value of code at the +start and end of each alternative, because compiled items are discarded during +the pre-compile phase so that the work space is not exceeded. */ + +length = 2 + 2*LINK_SIZE + skipbytes; + +/* WARNING: If the above line is changed for any reason, you must also change +the code that abstracts option settings at the start of the pattern and makes +them global. It tests the value of length for (2 + 2*LINK_SIZE) in the +pre-compile phase to find out whether anything has yet been compiled or not. */ + +/* If this is a capturing subpattern, add to the chain of open capturing items +so that we can detect them if (*ACCEPT) is encountered. This is also used to +detect groups that contain recursive back references to themselves. Note that +only OP_CBRA need be tested here; changing this opcode to one of its variants, +e.g. OP_SCBRAPOS, happens later, after the group has been compiled. */ + +if (*code == OP_CBRA) + { + capnumber = GET2(code, 1 + LINK_SIZE); + capitem.number = capnumber; + capitem.next = cd->open_caps; + capitem.flag = FALSE; + cd->open_caps = &capitem; + } + +/* Offset is set zero to mark that this bracket is still open */ + +PUT(code, 1, 0); +code += 1 + LINK_SIZE + skipbytes; + +/* Loop for each alternative branch */ + +orig_bracount = max_bracount = cd->bracount; +for (;;) + { + /* For a (?| group, reset the capturing bracket count so that each branch + uses the same numbers. */ + + if (reset_bracount) cd->bracount = orig_bracount; + + /* Set up dummy OP_REVERSE if lookbehind assertion */ + + if (lookbehind) + { + *code++ = OP_REVERSE; + reverse_count = code; + PUTINC(code, 0, 0); + length += 1 + LINK_SIZE; + } + + /* Now compile the branch; in the pre-compile phase its length gets added + into the length. */ + + if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstchar, + &branchreqchar, &bc, cond_depth, cd, + (lengthptr == NULL)? NULL : &length)) + { + *ptrptr = ptr; + return FALSE; + } + + /* Keep the highest bracket count in case (?| was used and some branch + has fewer than the rest. */ + + if (cd->bracount > max_bracount) max_bracount = cd->bracount; + + /* In the real compile phase, there is some post-processing to be done. */ + + if (lengthptr == NULL) + { + /* If this is the first branch, the firstchar and reqchar values for the + branch become the values for the regex. */ + + if (*last_branch != OP_ALT) + { + firstchar = branchfirstchar; + reqchar = branchreqchar; + } + + /* If this is not the first branch, the first char and reqchar have to + match the values from all the previous branches, except that if the + previous value for reqchar didn't have REQ_VARY set, it can still match, + and we set REQ_VARY for the regex. */ + + else + { + /* If we previously had a firstchar, but it doesn't match the new branch, + we have to abandon the firstchar for the regex, but if there was + previously no reqchar, it takes on the value of the old firstchar. */ + + if (firstchar >= 0 && firstchar != branchfirstchar) + { + if (reqchar < 0) reqchar = firstchar; + firstchar = REQ_NONE; + } + + /* If we (now or from before) have no firstchar, a firstchar from the + branch becomes a reqchar if there isn't a branch reqchar. */ + + if (firstchar < 0 && branchfirstchar >= 0 && branchreqchar < 0) + branchreqchar = branchfirstchar; + + /* Now ensure that the reqchars match */ + + if ((reqchar & ~REQ_VARY) != (branchreqchar & ~REQ_VARY)) + reqchar = REQ_NONE; + else reqchar |= branchreqchar; /* To "or" REQ_VARY */ + } + + /* If lookbehind, check that this branch matches a fixed-length string, and + put the length into the OP_REVERSE item. Temporarily mark the end of the + branch with OP_END. If the branch contains OP_RECURSE, the result is -3 + because there may be forward references that we can't check here. Set a + flag to cause another lookbehind check at the end. Why not do it all at the + end? Because common, erroneous checks are picked up here and the offset of + the problem can be shown. */ + + if (lookbehind) + { + int fixed_length; + *code = OP_END; + fixed_length = find_fixedlength(last_branch, (options & PCRE_UTF8) != 0, + FALSE, cd); + DPRINTF(("fixed length = %d\n", fixed_length)); + if (fixed_length == -3) + { + cd->check_lookbehind = TRUE; + } + else if (fixed_length < 0) + { + *errorcodeptr = (fixed_length == -2)? ERR36 : + (fixed_length == -4)? ERR70: ERR25; + *ptrptr = ptr; + return FALSE; + } + else + { + if (fixed_length > cd->max_lookbehind) + cd->max_lookbehind = fixed_length; + PUT(reverse_count, 0, fixed_length); + } + } + } + + /* Reached end of expression, either ')' or end of pattern. In the real + compile phase, go back through the alternative branches and reverse the chain + of offsets, with the field in the BRA item now becoming an offset to the + first alternative. If there are no alternatives, it points to the end of the + group. The length in the terminating ket is always the length of the whole + bracketed item. Return leaving the pointer at the terminating char. */ + + if (*ptr != CHAR_VERTICAL_LINE) + { + if (lengthptr == NULL) + { + int branch_length = (int)(code - last_branch); + do + { + int prev_length = GET(last_branch, 1); + PUT(last_branch, 1, branch_length); + branch_length = prev_length; + last_branch -= branch_length; + } + while (branch_length > 0); + } + + /* Fill in the ket */ + + *code = OP_KET; + PUT(code, 1, (int)(code - start_bracket)); + code += 1 + LINK_SIZE; + + /* If it was a capturing subpattern, check to see if it contained any + recursive back references. If so, we must wrap it in atomic brackets. + In any event, remove the block from the chain. */ + + if (capnumber > 0) + { + if (cd->open_caps->flag) + { + memmove(start_bracket + 1 + LINK_SIZE, start_bracket, + IN_UCHARS(code - start_bracket)); + *start_bracket = OP_ONCE; + code += 1 + LINK_SIZE; + PUT(start_bracket, 1, (int)(code - start_bracket)); + *code = OP_KET; + PUT(code, 1, (int)(code - start_bracket)); + code += 1 + LINK_SIZE; + length += 2 + 2*LINK_SIZE; + } + cd->open_caps = cd->open_caps->next; + } + + /* Retain the highest bracket number, in case resetting was used. */ + + cd->bracount = max_bracount; + + /* Set values to pass back */ + + *codeptr = code; + *ptrptr = ptr; + *firstcharptr = firstchar; + *reqcharptr = reqchar; + if (lengthptr != NULL) + { + if (OFLOW_MAX - *lengthptr < length) + { + *errorcodeptr = ERR20; + return FALSE; + } + *lengthptr += length; + } + return TRUE; + } + + /* Another branch follows. In the pre-compile phase, we can move the code + pointer back to where it was for the start of the first branch. (That is, + pretend that each branch is the only one.) + + In the real compile phase, insert an ALT node. Its length field points back + to the previous branch while the bracket remains open. At the end the chain + is reversed. It's done like this so that the start of the bracket has a + zero offset until it is closed, making it possible to detect recursion. */ + + if (lengthptr != NULL) + { + code = *codeptr + 1 + LINK_SIZE + skipbytes; + length += 1 + LINK_SIZE; + } + else + { + *code = OP_ALT; + PUT(code, 1, (int)(code - last_branch)); + bc.current_branch = last_branch = code; + code += 1 + LINK_SIZE; + } + + ptr++; + } +/* Control never reaches here */ +} + + + + +/************************************************* +* Check for anchored expression * +*************************************************/ + +/* Try to find out if this is an anchored regular expression. Consider each +alternative branch. If they all start with OP_SOD or OP_CIRC, or with a bracket +all of whose alternatives start with OP_SOD or OP_CIRC (recurse ad lib), then +it's anchored. However, if this is a multiline pattern, then only OP_SOD will +be found, because ^ generates OP_CIRCM in that mode. + +We can also consider a regex to be anchored if OP_SOM starts all its branches. +This is the code for \G, which means "match at start of match position, taking +into account the match offset". + +A branch is also implicitly anchored if it starts with .* and DOTALL is set, +because that will try the rest of the pattern at all possible matching points, +so there is no point trying again.... er .... + +.... except when the .* appears inside capturing parentheses, and there is a +subsequent back reference to those parentheses. We haven't enough information +to catch that case precisely. + +At first, the best we could do was to detect when .* was in capturing brackets +and the highest back reference was greater than or equal to that level. +However, by keeping a bitmap of the first 31 back references, we can catch some +of the more common cases more precisely. + +Arguments: + code points to start of expression (the bracket) + bracket_map a bitmap of which brackets we are inside while testing; this + handles up to substring 31; after that we just have to take + the less precise approach + backref_map the back reference bitmap + +Returns: TRUE or FALSE +*/ + +static BOOL +is_anchored(register const pcre_uchar *code, unsigned int bracket_map, + unsigned int backref_map) +{ +do { + const pcre_uchar *scode = first_significant_code( + code + PRIV(OP_lengths)[*code], FALSE); + register int op = *scode; + + /* Non-capturing brackets */ + + if (op == OP_BRA || op == OP_BRAPOS || + op == OP_SBRA || op == OP_SBRAPOS) + { + if (!is_anchored(scode, bracket_map, backref_map)) return FALSE; + } + + /* Capturing brackets */ + + else if (op == OP_CBRA || op == OP_CBRAPOS || + op == OP_SCBRA || op == OP_SCBRAPOS) + { + int n = GET2(scode, 1+LINK_SIZE); + int new_map = bracket_map | ((n < 32)? (1 << n) : 1); + if (!is_anchored(scode, new_map, backref_map)) return FALSE; + } + + /* Other brackets */ + + else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC || + op == OP_COND) + { + if (!is_anchored(scode, bracket_map, backref_map)) return FALSE; + } + + /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and + it isn't in brackets that are or may be referenced. */ + + else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR || + op == OP_TYPEPOSSTAR)) + { + if (scode[1] != OP_ALLANY || (bracket_map & backref_map) != 0) + return FALSE; + } + + /* Check for explicit anchoring */ + + else if (op != OP_SOD && op != OP_SOM && op != OP_CIRC) return FALSE; + code += GET(code, 1); + } +while (*code == OP_ALT); /* Loop for each alternative */ +return TRUE; +} + + + +/************************************************* +* Check for starting with ^ or .* * +*************************************************/ + +/* This is called to find out if every branch starts with ^ or .* so that +"first char" processing can be done to speed things up in multiline +matching and for non-DOTALL patterns that start with .* (which must start at +the beginning or after \n). As in the case of is_anchored() (see above), we +have to take account of back references to capturing brackets that contain .* +because in that case we can't make the assumption. + +Arguments: + code points to start of expression (the bracket) + bracket_map a bitmap of which brackets we are inside while testing; this + handles up to substring 31; after that we just have to take + the less precise approach + backref_map the back reference bitmap + +Returns: TRUE or FALSE +*/ + +static BOOL +is_startline(const pcre_uchar *code, unsigned int bracket_map, + unsigned int backref_map) +{ +do { + const pcre_uchar *scode = first_significant_code( + code + PRIV(OP_lengths)[*code], FALSE); + register int op = *scode; + + /* If we are at the start of a conditional assertion group, *both* the + conditional assertion *and* what follows the condition must satisfy the test + for start of line. Other kinds of condition fail. Note that there may be an + auto-callout at the start of a condition. */ + + if (op == OP_COND) + { + scode += 1 + LINK_SIZE; + if (*scode == OP_CALLOUT) scode += PRIV(OP_lengths)[OP_CALLOUT]; + switch (*scode) + { + case OP_CREF: + case OP_NCREF: + case OP_RREF: + case OP_NRREF: + case OP_DEF: + return FALSE; + + default: /* Assertion */ + if (!is_startline(scode, bracket_map, backref_map)) return FALSE; + do scode += GET(scode, 1); while (*scode == OP_ALT); + scode += 1 + LINK_SIZE; + break; + } + scode = first_significant_code(scode, FALSE); + op = *scode; + } + + /* Non-capturing brackets */ + + if (op == OP_BRA || op == OP_BRAPOS || + op == OP_SBRA || op == OP_SBRAPOS) + { + if (!is_startline(scode, bracket_map, backref_map)) return FALSE; + } + + /* Capturing brackets */ + + else if (op == OP_CBRA || op == OP_CBRAPOS || + op == OP_SCBRA || op == OP_SCBRAPOS) + { + int n = GET2(scode, 1+LINK_SIZE); + int new_map = bracket_map | ((n < 32)? (1 << n) : 1); + if (!is_startline(scode, new_map, backref_map)) return FALSE; + } + + /* Other brackets */ + + else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC) + { + if (!is_startline(scode, bracket_map, backref_map)) return FALSE; + } + + /* .* means "start at start or after \n" if it isn't in brackets that + may be referenced. */ + + else if (op == OP_TYPESTAR || op == OP_TYPEMINSTAR || op == OP_TYPEPOSSTAR) + { + if (scode[1] != OP_ANY || (bracket_map & backref_map) != 0) return FALSE; + } + + /* Check for explicit circumflex */ + + else if (op != OP_CIRC && op != OP_CIRCM) return FALSE; + + /* Move on to the next alternative */ + + code += GET(code, 1); + } +while (*code == OP_ALT); /* Loop for each alternative */ +return TRUE; +} + + + +/************************************************* +* Check for asserted fixed first char * +*************************************************/ + +/* During compilation, the "first char" settings from forward assertions are +discarded, because they can cause conflicts with actual literals that follow. +However, if we end up without a first char setting for an unanchored pattern, +it is worth scanning the regex to see if there is an initial asserted first +char. If all branches start with the same asserted char, or with a bracket all +of whose alternatives start with the same asserted char (recurse ad lib), then +we return that char, otherwise -1. + +Arguments: + code points to start of expression (the bracket) + inassert TRUE if in an assertion + +Returns: -1 or the fixed first char +*/ + +static int +find_firstassertedchar(const pcre_uchar *code, BOOL inassert) +{ +register int c = -1; +do { + int d; + int xl = (*code == OP_CBRA || *code == OP_SCBRA || + *code == OP_CBRAPOS || *code == OP_SCBRAPOS)? IMM2_SIZE:0; + const pcre_uchar *scode = first_significant_code(code + 1+LINK_SIZE + xl, + TRUE); + register int op = *scode; + + switch(op) + { + default: + return -1; + + case OP_BRA: + case OP_BRAPOS: + case OP_CBRA: + case OP_SCBRA: + case OP_CBRAPOS: + case OP_SCBRAPOS: + case OP_ASSERT: + case OP_ONCE: + case OP_ONCE_NC: + case OP_COND: + if ((d = find_firstassertedchar(scode, op == OP_ASSERT)) < 0) + return -1; + if (c < 0) c = d; else if (c != d) return -1; + break; + + case OP_EXACT: + scode += IMM2_SIZE; + /* Fall through */ + + case OP_CHAR: + case OP_PLUS: + case OP_MINPLUS: + case OP_POSPLUS: + if (!inassert) return -1; + if (c < 0) c = scode[1]; + else if (c != scode[1]) return -1; + break; + + case OP_EXACTI: + scode += IMM2_SIZE; + /* Fall through */ + + case OP_CHARI: + case OP_PLUSI: + case OP_MINPLUSI: + case OP_POSPLUSI: + if (!inassert) return -1; + if (c < 0) c = scode[1] | REQ_CASELESS; + else if (c != scode[1]) return -1; + break; + } + + code += GET(code, 1); + } +while (*code == OP_ALT); +return c; +} + + + +/************************************************* +* Compile a Regular Expression * +*************************************************/ + +/* This function takes a string and returns a pointer to a block of store +holding a compiled version of the expression. The original API for this +function had no error code return variable; it is retained for backwards +compatibility. The new function is given a new name. + +Arguments: + pattern the regular expression + options various option bits + errorcodeptr pointer to error code variable (pcre_compile2() only) + can be NULL if you don't want a code value + errorptr pointer to pointer to error text + erroroffset ptr offset in pattern where error was detected + tables pointer to character tables or NULL + +Returns: pointer to compiled data block, or NULL on error, + with errorptr and erroroffset set +*/ + +#ifdef COMPILE_PCRE8 +PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION +pcre_compile(const char *pattern, int options, const char **errorptr, + int *erroroffset, const unsigned char *tables) +#else +PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION +pcre16_compile(PCRE_SPTR16 pattern, int options, const char **errorptr, + int *erroroffset, const unsigned char *tables) +#endif +{ +#ifdef COMPILE_PCRE8 +return pcre_compile2(pattern, options, NULL, errorptr, erroroffset, tables); +#else +return pcre16_compile2(pattern, options, NULL, errorptr, erroroffset, tables); +#endif +} + + +#ifdef COMPILE_PCRE8 +PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION +pcre_compile2(const char *pattern, int options, int *errorcodeptr, + const char **errorptr, int *erroroffset, const unsigned char *tables) +#else +PCRE_EXP_DEFN pcre16 * PCRE_CALL_CONVENTION +pcre16_compile2(PCRE_SPTR16 pattern, int options, int *errorcodeptr, + const char **errorptr, int *erroroffset, const unsigned char *tables) +#endif +{ +REAL_PCRE *re; +int length = 1; /* For final END opcode */ +pcre_int32 firstchar, reqchar; +int newline; +int errorcode = 0; +int skipatstart = 0; +BOOL utf; +size_t size; +pcre_uchar *code; +const pcre_uchar *codestart; +const pcre_uchar *ptr; +compile_data compile_block; +compile_data *cd = &compile_block; + +/* This space is used for "compiling" into during the first phase, when we are +computing the amount of memory that is needed. Compiled items are thrown away +as soon as possible, so that a fairly large buffer should be sufficient for +this purpose. The same space is used in the second phase for remembering where +to fill in forward references to subpatterns. That may overflow, in which case +new memory is obtained from malloc(). */ + +pcre_uchar cworkspace[COMPILE_WORK_SIZE]; + +/* Set this early so that early errors get offset 0. */ + +ptr = (const pcre_uchar *)pattern; + +/* We can't pass back an error message if errorptr is NULL; I guess the best we +can do is just return NULL, but we can set a code value if there is a code +pointer. */ + +if (errorptr == NULL) + { + if (errorcodeptr != NULL) *errorcodeptr = 99; + return NULL; + } + +*errorptr = NULL; +if (errorcodeptr != NULL) *errorcodeptr = ERR0; + +/* However, we can give a message for this error */ + +if (erroroffset == NULL) + { + errorcode = ERR16; + goto PCRE_EARLY_ERROR_RETURN2; + } + +*erroroffset = 0; + +/* Set up pointers to the individual character tables */ + +if (tables == NULL) tables = PRIV(default_tables); +cd->lcc = tables + lcc_offset; +cd->fcc = tables + fcc_offset; +cd->cbits = tables + cbits_offset; +cd->ctypes = tables + ctypes_offset; + +/* Check that all undefined public option bits are zero */ + +if ((options & ~PUBLIC_COMPILE_OPTIONS) != 0) + { + errorcode = ERR17; + goto PCRE_EARLY_ERROR_RETURN; + } + +/* Check for global one-time settings at the start of the pattern, and remember +the offset for later. */ + +while (ptr[skipatstart] == CHAR_LEFT_PARENTHESIS && + ptr[skipatstart+1] == CHAR_ASTERISK) + { + int newnl = 0; + int newbsr = 0; + +#ifdef COMPILE_PCRE8 + if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 5) == 0) + { skipatstart += 7; options |= PCRE_UTF8; continue; } +#endif +#ifdef COMPILE_PCRE16 + if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UTF_RIGHTPAR, 6) == 0) + { skipatstart += 8; options |= PCRE_UTF16; continue; } +#endif + else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_UCP_RIGHTPAR, 4) == 0) + { skipatstart += 6; options |= PCRE_UCP; continue; } + else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_NO_START_OPT_RIGHTPAR, 13) == 0) + { skipatstart += 15; options |= PCRE_NO_START_OPTIMIZE; continue; } + + if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_CR_RIGHTPAR, 3) == 0) + { skipatstart += 5; newnl = PCRE_NEWLINE_CR; } + else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_LF_RIGHTPAR, 3) == 0) + { skipatstart += 5; newnl = PCRE_NEWLINE_LF; } + else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_CRLF_RIGHTPAR, 5) == 0) + { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; } + else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_ANY_RIGHTPAR, 4) == 0) + { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; } + else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_ANYCRLF_RIGHTPAR, 8) == 0) + { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; } + + else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_BSR_ANYCRLF_RIGHTPAR, 12) == 0) + { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; } + else if (STRNCMP_UC_C8(ptr+skipatstart+2, STRING_BSR_UNICODE_RIGHTPAR, 12) == 0) + { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; } + + if (newnl != 0) + options = (options & ~PCRE_NEWLINE_BITS) | newnl; + else if (newbsr != 0) + options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr; + else break; + } + +/* PCRE_UTF16 has the same value as PCRE_UTF8. */ +utf = (options & PCRE_UTF8) != 0; + +/* Can't support UTF unless PCRE has been compiled to include the code. The +return of an error code from PRIV(valid_utf)() is a new feature, introduced in +release 8.13. It is passed back from pcre_[dfa_]exec(), but at the moment is +not used here. */ + +#ifdef SUPPORT_UTF +if (utf && (options & PCRE_NO_UTF8_CHECK) == 0 && + (errorcode = PRIV(valid_utf)((PCRE_PUCHAR)pattern, -1, erroroffset)) != 0) + { +#ifdef COMPILE_PCRE8 + errorcode = ERR44; +#else + errorcode = ERR74; +#endif + goto PCRE_EARLY_ERROR_RETURN2; + } +#else +if (utf) + { + errorcode = ERR32; + goto PCRE_EARLY_ERROR_RETURN; + } +#endif + +/* Can't support UCP unless PCRE has been compiled to include the code. */ + +#ifndef SUPPORT_UCP +if ((options & PCRE_UCP) != 0) + { + errorcode = ERR67; + goto PCRE_EARLY_ERROR_RETURN; + } +#endif + +/* Check validity of \R options. */ + +if ((options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) == + (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) + { + errorcode = ERR56; + goto PCRE_EARLY_ERROR_RETURN; + } + +/* Handle different types of newline. The three bits give seven cases. The +current code allows for fixed one- or two-byte sequences, plus "any" and +"anycrlf". */ + +switch (options & PCRE_NEWLINE_BITS) + { + case 0: newline = NEWLINE; break; /* Build-time default */ + case PCRE_NEWLINE_CR: newline = CHAR_CR; break; + case PCRE_NEWLINE_LF: newline = CHAR_NL; break; + case PCRE_NEWLINE_CR+ + PCRE_NEWLINE_LF: newline = (CHAR_CR << 8) | CHAR_NL; break; + case PCRE_NEWLINE_ANY: newline = -1; break; + case PCRE_NEWLINE_ANYCRLF: newline = -2; break; + default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN; + } + +if (newline == -2) + { + cd->nltype = NLTYPE_ANYCRLF; + } +else if (newline < 0) + { + cd->nltype = NLTYPE_ANY; + } +else + { + cd->nltype = NLTYPE_FIXED; + if (newline > 255) + { + cd->nllen = 2; + cd->nl[0] = (newline >> 8) & 255; + cd->nl[1] = newline & 255; + } + else + { + cd->nllen = 1; + cd->nl[0] = newline; + } + } + +/* Maximum back reference and backref bitmap. The bitmap records up to 31 back +references to help in deciding whether (.*) can be treated as anchored or not. +*/ + +cd->top_backref = 0; +cd->backref_map = 0; + +/* Reflect pattern for debugging output */ + +DPRINTF(("------------------------------------------------------------------\n")); +#ifdef PCRE_DEBUG +print_puchar(stdout, (PCRE_PUCHAR)pattern); +#endif +DPRINTF(("\n")); + +/* Pretend to compile the pattern while actually just accumulating the length +of memory required. This behaviour is triggered by passing a non-NULL final +argument to compile_regex(). We pass a block of workspace (cworkspace) for it +to compile parts of the pattern into; the compiled code is discarded when it is +no longer needed, so hopefully this workspace will never overflow, though there +is a test for its doing so. */ + +cd->bracount = cd->final_bracount = 0; +cd->names_found = 0; +cd->name_entry_size = 0; +cd->name_table = NULL; +cd->start_code = cworkspace; +cd->hwm = cworkspace; +cd->start_workspace = cworkspace; +cd->workspace_size = COMPILE_WORK_SIZE; +cd->start_pattern = (const pcre_uchar *)pattern; +cd->end_pattern = (const pcre_uchar *)(pattern + STRLEN_UC((const pcre_uchar *)pattern)); +cd->req_varyopt = 0; +cd->assert_depth = 0; +cd->max_lookbehind = 0; +cd->external_options = options; +cd->external_flags = 0; +cd->open_caps = NULL; + +/* Now do the pre-compile. On error, errorcode will be set non-zero, so we +don't need to look at the result of the function here. The initial options have +been put into the cd block so that they can be changed if an option setting is +found within the regex right at the beginning. Bringing initial option settings +outside can help speed up starting point checks. */ + +ptr += skipatstart; +code = cworkspace; +*code = OP_BRA; +(void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE, + FALSE, 0, 0, &firstchar, &reqchar, NULL, cd, &length); +if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN; + +DPRINTF(("end pre-compile: length=%d workspace=%d\n", length, + (int)(cd->hwm - cworkspace))); + +if (length > MAX_PATTERN_SIZE) + { + errorcode = ERR20; + goto PCRE_EARLY_ERROR_RETURN; + } + +/* Compute the size of data block needed and get it, either from malloc or +externally provided function. Integer overflow should no longer be possible +because nowadays we limit the maximum value of cd->names_found and +cd->name_entry_size. */ + +size = sizeof(REAL_PCRE) + (length + cd->names_found * cd->name_entry_size) * sizeof(pcre_uchar); +re = (REAL_PCRE *)(PUBL(malloc))(size); + +if (re == NULL) + { + errorcode = ERR21; + goto PCRE_EARLY_ERROR_RETURN; + } + +/* Put in the magic number, and save the sizes, initial options, internal +flags, and character table pointer. NULL is used for the default character +tables. The nullpad field is at the end; it's there to help in the case when a +regex compiled on a system with 4-byte pointers is run on another with 8-byte +pointers. */ + +re->magic_number = MAGIC_NUMBER; +re->size = (int)size; +re->options = cd->external_options; +re->flags = cd->external_flags; +re->first_char = 0; +re->req_char = 0; +re->name_table_offset = sizeof(REAL_PCRE) / sizeof(pcre_uchar); +re->name_entry_size = cd->name_entry_size; +re->name_count = cd->names_found; +re->ref_count = 0; +re->tables = (tables == PRIV(default_tables))? NULL : tables; +re->nullpad = NULL; + +/* The starting points of the name/number translation table and of the code are +passed around in the compile data block. The start/end pattern and initial +options are already set from the pre-compile phase, as is the name_entry_size +field. Reset the bracket count and the names_found field. Also reset the hwm +field; this time it's used for remembering forward references to subpatterns. +*/ + +cd->final_bracount = cd->bracount; /* Save for checking forward references */ +cd->assert_depth = 0; +cd->bracount = 0; +cd->max_lookbehind = 0; +cd->names_found = 0; +cd->name_table = (pcre_uchar *)re + re->name_table_offset; +codestart = cd->name_table + re->name_entry_size * re->name_count; +cd->start_code = codestart; +cd->hwm = (pcre_uchar *)(cd->start_workspace); +cd->req_varyopt = 0; +cd->had_accept = FALSE; +cd->check_lookbehind = FALSE; +cd->open_caps = NULL; + +/* Set up a starting, non-extracting bracket, then compile the expression. On +error, errorcode will be set non-zero, so we don't need to look at the result +of the function here. */ + +ptr = (const pcre_uchar *)pattern + skipatstart; +code = (pcre_uchar *)codestart; +*code = OP_BRA; +(void)compile_regex(re->options, &code, &ptr, &errorcode, FALSE, FALSE, 0, 0, + &firstchar, &reqchar, NULL, cd, NULL); +re->top_bracket = cd->bracount; +re->top_backref = cd->top_backref; +re->max_lookbehind = cd->max_lookbehind; +re->flags = cd->external_flags | PCRE_MODE; + +if (cd->had_accept) reqchar = REQ_NONE; /* Must disable after (*ACCEPT) */ + +/* If not reached end of pattern on success, there's an excess bracket. */ + +if (errorcode == 0 && *ptr != 0) errorcode = ERR22; + +/* Fill in the terminating state and check for disastrous overflow, but +if debugging, leave the test till after things are printed out. */ + +*code++ = OP_END; + +#ifndef PCRE_DEBUG +if (code - codestart > length) errorcode = ERR23; +#endif + +/* Fill in any forward references that are required. There may be repeated +references; optimize for them, as searching a large regex takes time. */ + +if (cd->hwm > cd->start_workspace) + { + int prev_recno = -1; + const pcre_uchar *groupptr = NULL; + while (errorcode == 0 && cd->hwm > cd->start_workspace) + { + int offset, recno; + cd->hwm -= LINK_SIZE; + offset = GET(cd->hwm, 0); + recno = GET(codestart, offset); + if (recno != prev_recno) + { + groupptr = PRIV(find_bracket)(codestart, utf, recno); + prev_recno = recno; + } + if (groupptr == NULL) errorcode = ERR53; + else PUT(((pcre_uchar *)codestart), offset, (int)(groupptr - codestart)); + } + } + +/* If the workspace had to be expanded, free the new memory. */ + +if (cd->workspace_size > COMPILE_WORK_SIZE) + (PUBL(free))((void *)cd->start_workspace); + +/* Give an error if there's back reference to a non-existent capturing +subpattern. */ + +if (errorcode == 0 && re->top_backref > re->top_bracket) errorcode = ERR15; + +/* If there were any lookbehind assertions that contained OP_RECURSE +(recursions or subroutine calls), a flag is set for them to be checked here, +because they may contain forward references. Actual recursions can't be fixed +length, but subroutine calls can. It is done like this so that those without +OP_RECURSE that are not fixed length get a diagnosic with a useful offset. The +exceptional ones forgo this. We scan the pattern to check that they are fixed +length, and set their lengths. */ + +if (cd->check_lookbehind) + { + pcre_uchar *cc = (pcre_uchar *)codestart; + + /* Loop, searching for OP_REVERSE items, and process those that do not have + their length set. (Actually, it will also re-process any that have a length + of zero, but that is a pathological case, and it does no harm.) When we find + one, we temporarily terminate the branch it is in while we scan it. */ + + for (cc = (pcre_uchar *)PRIV(find_bracket)(codestart, utf, -1); + cc != NULL; + cc = (pcre_uchar *)PRIV(find_bracket)(cc, utf, -1)) + { + if (GET(cc, 1) == 0) + { + int fixed_length; + pcre_uchar *be = cc - 1 - LINK_SIZE + GET(cc, -LINK_SIZE); + int end_op = *be; + *be = OP_END; + fixed_length = find_fixedlength(cc, (re->options & PCRE_UTF8) != 0, TRUE, + cd); + *be = end_op; + DPRINTF(("fixed length = %d\n", fixed_length)); + if (fixed_length < 0) + { + errorcode = (fixed_length == -2)? ERR36 : + (fixed_length == -4)? ERR70 : ERR25; + break; + } + if (fixed_length > cd->max_lookbehind) cd->max_lookbehind = fixed_length; + PUT(cc, 1, fixed_length); + } + cc += 1 + LINK_SIZE; + } + } + +/* Failed to compile, or error while post-processing */ + +if (errorcode != 0) + { + (PUBL(free))(re); + PCRE_EARLY_ERROR_RETURN: + *erroroffset = (int)(ptr - (const pcre_uchar *)pattern); + PCRE_EARLY_ERROR_RETURN2: + *errorptr = find_error_text(errorcode); + if (errorcodeptr != NULL) *errorcodeptr = errorcode; + return NULL; + } + +/* If the anchored option was not passed, set the flag if we can determine that +the pattern is anchored by virtue of ^ characters or \A or anything else (such +as starting with .* when DOTALL is set). + +Otherwise, if we know what the first byte has to be, save it, because that +speeds up unanchored matches no end. If not, see if we can set the +PCRE_STARTLINE flag. This is helpful for multiline matches when all branches +start with ^. and also when all branches start with .* for non-DOTALL matches. +*/ + +if ((re->options & PCRE_ANCHORED) == 0) + { + if (is_anchored(codestart, 0, cd->backref_map)) + re->options |= PCRE_ANCHORED; + else + { + if (firstchar < 0) + firstchar = find_firstassertedchar(codestart, FALSE); + if (firstchar >= 0) /* Remove caseless flag for non-caseable chars */ + { +#ifdef COMPILE_PCRE8 + re->first_char = firstchar & 0xff; +#else +#ifdef COMPILE_PCRE16 + re->first_char = firstchar & 0xffff; +#endif +#endif + if ((firstchar & REQ_CASELESS) != 0) + { +#if defined SUPPORT_UCP && !(defined COMPILE_PCRE8) + /* We ignore non-ASCII first chars in 8 bit mode. */ + if (utf) + { + if (re->first_char < 128) + { + if (cd->fcc[re->first_char] != re->first_char) + re->flags |= PCRE_FCH_CASELESS; + } + else if (UCD_OTHERCASE(re->first_char) != re->first_char) + re->flags |= PCRE_FCH_CASELESS; + } + else +#endif + if (MAX_255(re->first_char) + && cd->fcc[re->first_char] != re->first_char) + re->flags |= PCRE_FCH_CASELESS; + } + + re->flags |= PCRE_FIRSTSET; + } + else if (is_startline(codestart, 0, cd->backref_map)) + re->flags |= PCRE_STARTLINE; + } + } + +/* For an anchored pattern, we use the "required byte" only if it follows a +variable length item in the regex. Remove the caseless flag for non-caseable +bytes. */ + +if (reqchar >= 0 && + ((re->options & PCRE_ANCHORED) == 0 || (reqchar & REQ_VARY) != 0)) + { +#ifdef COMPILE_PCRE8 + re->req_char = reqchar & 0xff; +#else +#ifdef COMPILE_PCRE16 + re->req_char = reqchar & 0xffff; +#endif +#endif + if ((reqchar & REQ_CASELESS) != 0) + { +#if defined SUPPORT_UCP && !(defined COMPILE_PCRE8) + /* We ignore non-ASCII first chars in 8 bit mode. */ + if (utf) + { + if (re->req_char < 128) + { + if (cd->fcc[re->req_char] != re->req_char) + re->flags |= PCRE_RCH_CASELESS; + } + else if (UCD_OTHERCASE(re->req_char) != re->req_char) + re->flags |= PCRE_RCH_CASELESS; + } + else +#endif + if (MAX_255(re->req_char) && cd->fcc[re->req_char] != re->req_char) + re->flags |= PCRE_RCH_CASELESS; + } + + re->flags |= PCRE_REQCHSET; + } + +/* Print out the compiled data if debugging is enabled. This is never the +case when building a production library. */ + +#ifdef PCRE_DEBUG +printf("Length = %d top_bracket = %d top_backref = %d\n", + length, re->top_bracket, re->top_backref); + +printf("Options=%08x\n", re->options); + +if ((re->flags & PCRE_FIRSTSET) != 0) + { + pcre_uchar ch = re->first_char; + const char *caseless = + ((re->flags & PCRE_FCH_CASELESS) == 0)? "" : " (caseless)"; + if (PRINTABLE(ch)) printf("First char = %c%s\n", ch, caseless); + else printf("First char = \\x%02x%s\n", ch, caseless); + } + +if ((re->flags & PCRE_REQCHSET) != 0) + { + pcre_uchar ch = re->req_char; + const char *caseless = + ((re->flags & PCRE_RCH_CASELESS) == 0)? "" : " (caseless)"; + if (PRINTABLE(ch)) printf("Req char = %c%s\n", ch, caseless); + else printf("Req char = \\x%02x%s\n", ch, caseless); + } + +#ifdef COMPILE_PCRE8 +pcre_printint((pcre *)re, stdout, TRUE); +#else +pcre16_printint((pcre *)re, stdout, TRUE); +#endif + +/* This check is done here in the debugging case so that the code that +was compiled can be seen. */ + +if (code - codestart > length) + { + (PUBL(free))(re); + *errorptr = find_error_text(ERR23); + *erroroffset = ptr - (pcre_uchar *)pattern; + if (errorcodeptr != NULL) *errorcodeptr = ERR23; + return NULL; + } +#endif /* PCRE_DEBUG */ + +#ifdef COMPILE_PCRE8 +return (pcre *)re; +#else +return (pcre16 *)re; +#endif +} + +/* End of pcre_compile.c */ diff --git a/pcre/pcre_config.c b/pcre/pcre_config.c --- a/pcre/pcre_config.c +++ b/pcre/pcre_config.c @@ -65,21 +65,18 @@ Returns: 0 if data returned, negative on error */ -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre_config(int what, void *where) -#elif defined COMPILE_PCRE16 +#else PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre16_config(int what, void *where) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DEFN int PCRE_CALL_CONVENTION -pcre32_config(int what, void *where) #endif { switch (what) { case PCRE_CONFIG_UTF8: -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 +#if defined COMPILE_PCRE16 *((int *)where) = 0; return PCRE_ERROR_BADOPTION; #else @@ -92,20 +89,7 @@ #endif case PCRE_CONFIG_UTF16: -#if defined COMPILE_PCRE8 || defined COMPILE_PCRE32 - *((int *)where) = 0; - return PCRE_ERROR_BADOPTION; -#else -#if defined SUPPORT_UTF - *((int *)where) = 1; -#else - *((int *)where) = 0; -#endif - break; -#endif - - case PCRE_CONFIG_UTF32: -#if defined COMPILE_PCRE8 || defined COMPILE_PCRE16 +#if defined COMPILE_PCRE8 *((int *)where) = 0; return PCRE_ERROR_BADOPTION; #else @@ -159,10 +143,6 @@ case PCRE_CONFIG_POSIX_MALLOC_THRESHOLD: *((int *)where) = POSIX_MALLOC_THRESHOLD; - break; - - case PCRE_CONFIG_PARENS_LIMIT: - *((unsigned long int *)where) = PARENS_NEST_LIMIT; break; case PCRE_CONFIG_MATCH_LIMIT: diff --git a/pcre/pcre_dfa_exec.c b/pcre/pcre_dfa_exec.c --- a/pcre/pcre_dfa_exec.c +++ b/pcre/pcre_dfa_exec.c @@ -7,7 +7,7 @@ below for why this module is different). Written by Philip Hazel - Copyright (c) 1997-2013 University of Cambridge + Copyright (c) 1997-2012 University of Cambridge ----------------------------------------------------------------------------- Redistribution and use in source and binary forms, with or without @@ -120,7 +120,7 @@ 0, 0, /* \P, \p */ 0, 0, 0, 0, 0, /* \R, \H, \h, \V, \v */ 0, /* \X */ - 0, 0, 0, 0, 0, 0, /* \Z, \z, $, $M, ^, ^M */ + 0, 0, 0, 0, 0, 0, /* \Z, \z, ^, ^M, $, $M */ 1, /* Char */ 1, /* Chari */ 1, /* not */ @@ -151,14 +151,11 @@ /* Character class & ref repeats */ 0, 0, 0, 0, 0, 0, /* *, *?, +, +?, ?, ?? */ 0, 0, /* CRRANGE, CRMINRANGE */ - 0, 0, 0, 0, /* Possessive *+, ++, ?+, CRPOSRANGE */ 0, /* CLASS */ 0, /* NCLASS */ 0, /* XCLASS - variable length */ 0, /* REF */ 0, /* REFI */ - 0, /* DNREF */ - 0, /* DNREFI */ 0, /* RECURSE */ 0, /* CALLOUT */ 0, /* Alt */ @@ -174,8 +171,8 @@ 0, 0, /* ONCE, ONCE_NC */ 0, 0, 0, 0, 0, /* BRA, BRAPOS, CBRA, CBRAPOS, COND */ 0, 0, 0, 0, 0, /* SBRA, SBRAPOS, SCBRA, SCBRAPOS, SCOND */ - 0, 0, /* CREF, DNCREF */ - 0, 0, /* RREF, DNRREF */ + 0, 0, /* CREF, NCREF */ + 0, 0, /* RREF, NRREF */ 0, /* DEF */ 0, 0, 0, /* BRAZERO, BRAMINZERO, BRAPOSZERO */ 0, 0, 0, /* MARK, PRUNE, PRUNE_ARG */ @@ -197,7 +194,7 @@ 1, 1, /* \P, \p */ 1, 1, 1, 1, 1, /* \R, \H, \h, \V, \v */ 1, /* \X */ - 0, 0, 0, 0, 0, 0, /* \Z, \z, $, $M, ^, ^M */ + 0, 0, 0, 0, 0, 0, /* \Z, \z, ^, ^M, $, $M */ 1, /* Char */ 1, /* Chari */ 1, /* not */ @@ -223,14 +220,11 @@ /* Character class & ref repeats */ 1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */ 1, 1, /* CRRANGE, CRMINRANGE */ - 1, 1, 1, 1, /* Possessive *+, ++, ?+, CRPOSRANGE */ 1, /* CLASS */ 1, /* NCLASS */ 1, /* XCLASS - variable length */ 0, /* REF */ 0, /* REFI */ - 0, /* DNREF */ - 0, /* DNREFI */ 0, /* RECURSE */ 0, /* CALLOUT */ 0, /* Alt */ @@ -246,8 +240,8 @@ 0, 0, /* ONCE, ONCE_NC */ 0, 0, 0, 0, 0, /* BRA, BRAPOS, CBRA, CBRAPOS, COND */ 0, 0, 0, 0, 0, /* SBRA, SBRAPOS, SCBRA, SCBRAPOS, SCOND */ - 0, 0, /* CREF, DNCREF */ - 0, 0, /* RREF, DNRREF */ + 0, 0, /* CREF, NCREF */ + 0, 0, /* RREF, NRREF */ 0, /* DEF */ 0, 0, 0, /* BRAZERO, BRAMINZERO, BRAPOSZERO */ 0, 0, 0, /* MARK, PRUNE, PRUNE_ARG */ @@ -308,13 +302,13 @@ static void pchars(const pcre_uchar *p, int length, FILE *f) { -pcre_uint32 c; +int c; while (length-- > 0) { if (isprint(c = *(p++))) fprintf(f, "%c", c); else - fprintf(f, "\\x{%02x}", c); + fprintf(f, "\\x%02x", c); } } #endif @@ -577,7 +571,7 @@ { int i, j; int clen, dlen; - pcre_uint32 c, d; + unsigned int c, d; int forced_fail = 0; BOOL partial_newline = FALSE; BOOL could_continue = reset_could_continue; @@ -619,10 +613,9 @@ { clen = 1; /* Number of data items in the character */ #ifdef SUPPORT_UTF - GETCHARLENTEST(c, ptr, clen); -#else - c = *ptr; + if (utf) { GETCHARLEN(c, ptr, clen); } else #endif /* SUPPORT_UTF */ + c = *ptr; } else { @@ -641,8 +634,7 @@ BOOL caseless = FALSE; const pcre_uchar *code; int state_offset = current_state->offset; - int codevalue, rrc; - int count; + int count, codevalue, rrc; #ifdef PCRE_DEBUG printf ("%.*sProcessing state %d c=", rlevel*2-2, SP, state_offset); @@ -1015,7 +1007,7 @@ { const pcre_uchar *temp = ptr - 1; if (temp < md->start_used_ptr) md->start_used_ptr = temp; -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 +#ifdef SUPPORT_UTF if (utf) { BACKCHAR(temp); } #endif GETCHARTEST(d, temp); @@ -1068,7 +1060,6 @@ if (clen > 0) { BOOL OK; - const pcre_uint32 *cp; const ucd_record * prop = GET_UCD(c); switch(code[1]) { @@ -1100,23 +1091,15 @@ PRIV(ucp_gentype)[prop->chartype] == ucp_N; break; - /* Perl space used to exclude VT, but from Perl 5.18 it is included, - which means that Perl space and POSIX space are now identical. PCRE - was changed at release 8.34. */ - case PT_SPACE: /* Perl space */ - case PT_PXSPACE: /* POSIX space */ - switch(c) - { - HSPACE_CASES: - VSPACE_CASES: - OK = TRUE; - break; + OK = PRIV(ucp_gentype)[prop->chartype] == ucp_Z || + c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR; + break; - default: - OK = PRIV(ucp_gentype)[prop->chartype] == ucp_Z; - break; - } + case PT_PXSPACE: /* POSIX space */ + OK = PRIV(ucp_gentype)[prop->chartype] == ucp_Z || + c == CHAR_HT || c == CHAR_NL || c == CHAR_VT || + c == CHAR_FF || c == CHAR_CR; break; case PT_WORD: @@ -1125,21 +1108,6 @@ c == CHAR_UNDERSCORE; break; - case PT_CLIST: - cp = PRIV(ucd_caseless_sets) + code[2]; - for (;;) - { - if (c < *cp) { OK = FALSE; break; } - if (c == *cp++) { OK = TRUE; break; } - } - break; - - case PT_UCNC: - OK = c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT || - c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) || - c >= 0xe000; - break; - /* Should never occur, but keep compilers from grumbling. */ default: @@ -1269,7 +1237,7 @@ (d != OP_ANY || !IS_NEWLINE(ptr)) && ((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0)) { - if (++count >= (int)GET2(code, 1)) + if (++count >= GET2(code, 1)) { ADD_NEW(state_offset + 1 + IMM2_SIZE + 1, 0); } else { ADD_NEW(state_offset, count); } @@ -1303,7 +1271,7 @@ active_count--; /* Remove non-match possibility */ next_active_state--; } - if (++count >= (int)GET2(code, 1)) + if (++count >= GET2(code, 1)) { ADD_NEW(state_offset + 2 + IMM2_SIZE, 0); } else { ADD_NEW(state_offset, count); } @@ -1326,7 +1294,6 @@ if (clen > 0) { BOOL OK; - const pcre_uint32 *cp; const ucd_record * prop = GET_UCD(c); switch(code[2]) { @@ -1358,23 +1325,15 @@ PRIV(ucp_gentype)[prop->chartype] == ucp_N; break; - /* Perl space used to exclude VT, but from Perl 5.18 it is included, - which means that Perl space and POSIX space are now identical. PCRE - was changed at release 8.34. */ - case PT_SPACE: /* Perl space */ - case PT_PXSPACE: /* POSIX space */ - switch(c) - { - HSPACE_CASES: - VSPACE_CASES: - OK = TRUE; - break; + OK = PRIV(ucp_gentype)[prop->chartype] == ucp_Z || + c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR; + break; - default: - OK = PRIV(ucp_gentype)[prop->chartype] == ucp_Z; - break; - } + case PT_PXSPACE: /* POSIX space */ + OK = PRIV(ucp_gentype)[prop->chartype] == ucp_Z || + c == CHAR_HT || c == CHAR_NL || c == CHAR_VT || + c == CHAR_FF || c == CHAR_CR; break; case PT_WORD: @@ -1383,21 +1342,6 @@ c == CHAR_UNDERSCORE; break; - case PT_CLIST: - cp = PRIV(ucd_caseless_sets) + code[3]; - for (;;) - { - if (c < *cp) { OK = FALSE; break; } - if (c == *cp++) { OK = TRUE; break; } - } - break; - - case PT_UCNC: - OK = c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT || - c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) || - c >= 0xe000; - break; - /* Should never occur, but keep compilers from grumbling. */ default: @@ -1424,9 +1368,8 @@ case OP_EXTUNI_EXTRA + OP_TYPEPOSPLUS: count = current_state->count; /* Already matched */ if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); } - if (clen > 0) + if (clen > 0 && UCD_CATEGORY(c) != ucp_M) { - int lgb, rgb; const pcre_uchar *nptr = ptr + clen; int ncount = 0; if (count > 0 && codevalue == OP_EXTUNI_EXTRA + OP_TYPEPOSPLUS) @@ -1434,16 +1377,14 @@ active_count--; /* Remove non-match possibility */ next_active_state--; } - lgb = UCD_GRAPHBREAK(c); while (nptr < end_subject) { - dlen = 1; - if (!utf) d = *nptr; else { GETCHARLEN(d, nptr, dlen); } - rgb = UCD_GRAPHBREAK(d); - if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break; + int nd; + int ndlen = 1; + GETCHARLEN(nd, nptr, ndlen); + if (UCD_CATEGORY(nd) != ucp_M) break; ncount++; - lgb = rgb; - nptr += dlen; + nptr += ndlen; } count++; ADD_NEW_DATA(-state_offset, count, ncount); @@ -1462,22 +1403,20 @@ int ncount = 0; switch (c) { - case CHAR_VT: - case CHAR_FF: - case CHAR_NEL: -#ifndef EBCDIC + case 0x000b: + case 0x000c: + case 0x0085: case 0x2028: case 0x2029: -#endif /* Not EBCDIC */ if ((md->moptions & PCRE_BSR_ANYCRLF) != 0) break; goto ANYNL01; - case CHAR_CR: - if (ptr + 1 < end_subject && RAWUCHARTEST(ptr + 1) == CHAR_LF) ncount = 1; + case 0x000d: + if (ptr + 1 < end_subject && ptr[1] == 0x0a) ncount = 1; /* Fall through */ ANYNL01: - case CHAR_LF: + case 0x000a: if (count > 0 && codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSPLUS) { active_count--; /* Remove non-match possibility */ @@ -1504,7 +1443,13 @@ BOOL OK; switch (c) { - VSPACE_CASES: + case 0x000a: + case 0x000b: + case 0x000c: + case 0x000d: + case 0x0085: + case 0x2028: + case 0x2029: OK = TRUE; break; @@ -1537,7 +1482,25 @@ BOOL OK; switch (c) { - HSPACE_CASES: + case 0x09: /* HT */ + case 0x20: /* SPACE */ + case 0xa0: /* NBSP */ + case 0x1680: /* OGHAM SPACE MARK */ + case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ + case 0x2000: /* EN QUAD */ + case 0x2001: /* EM QUAD */ + case 0x2002: /* EN SPACE */ + case 0x2003: /* EM SPACE */ + case 0x2004: /* THREE-PER-EM SPACE */ + case 0x2005: /* FOUR-PER-EM SPACE */ + case 0x2006: /* SIX-PER-EM SPACE */ + case 0x2007: /* FIGURE SPACE */ + case 0x2008: /* PUNCTUATION SPACE */ + case 0x2009: /* THIN SPACE */ + case 0x200A: /* HAIR SPACE */ + case 0x202f: /* NARROW NO-BREAK SPACE */ + case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ + case 0x3000: /* IDEOGRAPHIC SPACE */ OK = TRUE; break; @@ -1578,7 +1541,6 @@ if (clen > 0) { BOOL OK; - const pcre_uint32 *cp; const ucd_record * prop = GET_UCD(c); switch(code[2]) { @@ -1610,23 +1572,15 @@ PRIV(ucp_gentype)[prop->chartype] == ucp_N; break; - /* Perl space used to exclude VT, but from Perl 5.18 it is included, - which means that Perl space and POSIX space are now identical. PCRE - was changed at release 8.34. */ - case PT_SPACE: /* Perl space */ - case PT_PXSPACE: /* POSIX space */ - switch(c) - { - HSPACE_CASES: - VSPACE_CASES: - OK = TRUE; - break; + OK = PRIV(ucp_gentype)[prop->chartype] == ucp_Z || + c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR; + break; - default: - OK = PRIV(ucp_gentype)[prop->chartype] == ucp_Z; - break; - } + case PT_PXSPACE: /* POSIX space */ + OK = PRIV(ucp_gentype)[prop->chartype] == ucp_Z || + c == CHAR_HT || c == CHAR_NL || c == CHAR_VT || + c == CHAR_FF || c == CHAR_CR; break; case PT_WORD: @@ -1635,21 +1589,6 @@ c == CHAR_UNDERSCORE; break; - case PT_CLIST: - cp = PRIV(ucd_caseless_sets) + code[3]; - for (;;) - { - if (c < *cp) { OK = FALSE; break; } - if (c == *cp++) { OK = TRUE; break; } - } - break; - - case PT_UCNC: - OK = c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT || - c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) || - c >= 0xe000; - break; - /* Should never occur, but keep compilers from grumbling. */ default: @@ -1685,9 +1624,8 @@ QS2: ADD_ACTIVE(state_offset + 2, 0); - if (clen > 0) + if (clen > 0 && UCD_CATEGORY(c) != ucp_M) { - int lgb, rgb; const pcre_uchar *nptr = ptr + clen; int ncount = 0; if (codevalue == OP_EXTUNI_EXTRA + OP_TYPEPOSSTAR || @@ -1696,16 +1634,14 @@ active_count--; /* Remove non-match possibility */ next_active_state--; } - lgb = UCD_GRAPHBREAK(c); while (nptr < end_subject) { - dlen = 1; - if (!utf) d = *nptr; else { GETCHARLEN(d, nptr, dlen); } - rgb = UCD_GRAPHBREAK(d); - if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break; + int nd; + int ndlen = 1; + GETCHARLEN(nd, nptr, ndlen); + if (UCD_CATEGORY(nd) != ucp_M) break; ncount++; - lgb = rgb; - nptr += dlen; + nptr += ndlen; } ADD_NEW_DATA(-(state_offset + count), 0, ncount); } @@ -1731,29 +1667,27 @@ int ncount = 0; switch (c) { - case CHAR_VT: - case CHAR_FF: - case CHAR_NEL: -#ifndef EBCDIC + case 0x000b: + case 0x000c: + case 0x0085: case 0x2028: case 0x2029: -#endif /* Not EBCDIC */ if ((md->moptions & PCRE_BSR_ANYCRLF) != 0) break; goto ANYNL02; - case CHAR_CR: - if (ptr + 1 < end_subject && RAWUCHARTEST(ptr + 1) == CHAR_LF) ncount = 1; + case 0x000d: + if (ptr + 1 < end_subject && ptr[1] == 0x0a) ncount = 1; /* Fall through */ ANYNL02: - case CHAR_LF: + case 0x000a: if (codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSSTAR || codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSQUERY) { active_count--; /* Remove non-match possibility */ next_active_state--; } - ADD_NEW_DATA(-(state_offset + (int)count), 0, ncount); + ADD_NEW_DATA(-(state_offset + count), 0, ncount); break; default: @@ -1781,7 +1715,13 @@ BOOL OK; switch (c) { - VSPACE_CASES: + case 0x000a: + case 0x000b: + case 0x000c: + case 0x000d: + case 0x0085: + case 0x2028: + case 0x2029: OK = TRUE; break; @@ -1797,7 +1737,7 @@ active_count--; /* Remove non-match possibility */ next_active_state--; } - ADD_NEW_DATA(-(state_offset + (int)count), 0, 0); + ADD_NEW_DATA(-(state_offset + count), 0, 0); } } break; @@ -1821,7 +1761,25 @@ BOOL OK; switch (c) { - HSPACE_CASES: + case 0x09: /* HT */ + case 0x20: /* SPACE */ + case 0xa0: /* NBSP */ + case 0x1680: /* OGHAM SPACE MARK */ + case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ + case 0x2000: /* EN QUAD */ + case 0x2001: /* EM QUAD */ + case 0x2002: /* EN SPACE */ + case 0x2003: /* EM SPACE */ + case 0x2004: /* THREE-PER-EM SPACE */ + case 0x2005: /* FOUR-PER-EM SPACE */ + case 0x2006: /* SIX-PER-EM SPACE */ + case 0x2007: /* FIGURE SPACE */ + case 0x2008: /* PUNCTUATION SPACE */ + case 0x2009: /* THIN SPACE */ + case 0x200A: /* HAIR SPACE */ + case 0x202f: /* NARROW NO-BREAK SPACE */ + case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ + case 0x3000: /* IDEOGRAPHIC SPACE */ OK = TRUE; break; @@ -1838,7 +1796,7 @@ active_count--; /* Remove non-match possibility */ next_active_state--; } - ADD_NEW_DATA(-(state_offset + (int)count), 0, 0); + ADD_NEW_DATA(-(state_offset + count), 0, 0); } } break; @@ -1855,7 +1813,6 @@ if (clen > 0) { BOOL OK; - const pcre_uint32 *cp; const ucd_record * prop = GET_UCD(c); switch(code[1 + IMM2_SIZE + 1]) { @@ -1887,23 +1844,15 @@ PRIV(ucp_gentype)[prop->chartype] == ucp_N; break; - /* Perl space used to exclude VT, but from Perl 5.18 it is included, - which means that Perl space and POSIX space are now identical. PCRE - was changed at release 8.34. */ - case PT_SPACE: /* Perl space */ - case PT_PXSPACE: /* POSIX space */ - switch(c) - { - HSPACE_CASES: - VSPACE_CASES: - OK = TRUE; - break; + OK = PRIV(ucp_gentype)[prop->chartype] == ucp_Z || + c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR; + break; - default: - OK = PRIV(ucp_gentype)[prop->chartype] == ucp_Z; - break; - } + case PT_PXSPACE: /* POSIX space */ + OK = PRIV(ucp_gentype)[prop->chartype] == ucp_Z || + c == CHAR_HT || c == CHAR_NL || c == CHAR_VT || + c == CHAR_FF || c == CHAR_CR; break; case PT_WORD: @@ -1912,21 +1861,6 @@ c == CHAR_UNDERSCORE; break; - case PT_CLIST: - cp = PRIV(ucd_caseless_sets) + code[1 + IMM2_SIZE + 2]; - for (;;) - { - if (c < *cp) { OK = FALSE; break; } - if (c == *cp++) { OK = TRUE; break; } - } - break; - - case PT_UCNC: - OK = c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT || - c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) || - c >= 0xe000; - break; - /* Should never occur, but keep compilers from grumbling. */ default: @@ -1941,7 +1875,7 @@ active_count--; /* Remove non-match possibility */ next_active_state--; } - if (++count >= (int)GET2(code, 1)) + if (++count >= GET2(code, 1)) { ADD_NEW(state_offset + 1 + IMM2_SIZE + 3, 0); } else { ADD_NEW(state_offset, count); } @@ -1957,9 +1891,8 @@ if (codevalue != OP_EXTUNI_EXTRA + OP_TYPEEXACT) { ADD_ACTIVE(state_offset + 2 + IMM2_SIZE, 0); } count = current_state->count; /* Number already matched */ - if (clen > 0) + if (clen > 0 && UCD_CATEGORY(c) != ucp_M) { - int lgb, rgb; const pcre_uchar *nptr = ptr + clen; int ncount = 0; if (codevalue == OP_EXTUNI_EXTRA + OP_TYPEPOSUPTO) @@ -1967,20 +1900,18 @@ active_count--; /* Remove non-match possibility */ next_active_state--; } - lgb = UCD_GRAPHBREAK(c); while (nptr < end_subject) { - dlen = 1; - if (!utf) d = *nptr; else { GETCHARLEN(d, nptr, dlen); } - rgb = UCD_GRAPHBREAK(d); - if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break; + int nd; + int ndlen = 1; + GETCHARLEN(nd, nptr, ndlen); + if (UCD_CATEGORY(nd) != ucp_M) break; ncount++; - lgb = rgb; - nptr += dlen; + nptr += ndlen; } if (nptr >= end_subject && (md->moptions & PCRE_PARTIAL_HARD) != 0) reset_could_continue = TRUE; - if (++count >= (int)GET2(code, 1)) + if (++count >= GET2(code, 1)) { ADD_NEW_DATA(-(state_offset + 2 + IMM2_SIZE), 0, ncount); } else { ADD_NEW_DATA(-state_offset, count, ncount); } @@ -2001,28 +1932,26 @@ int ncount = 0; switch (c) { - case CHAR_VT: - case CHAR_FF: - case CHAR_NEL: -#ifndef EBCDIC + case 0x000b: + case 0x000c: + case 0x0085: case 0x2028: case 0x2029: -#endif /* Not EBCDIC */ if ((md->moptions & PCRE_BSR_ANYCRLF) != 0) break; goto ANYNL03; - case CHAR_CR: - if (ptr + 1 < end_subject && RAWUCHARTEST(ptr + 1) == CHAR_LF) ncount = 1; + case 0x000d: + if (ptr + 1 < end_subject && ptr[1] == 0x0a) ncount = 1; /* Fall through */ ANYNL03: - case CHAR_LF: + case 0x000a: if (codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSUPTO) { active_count--; /* Remove non-match possibility */ next_active_state--; } - if (++count >= (int)GET2(code, 1)) + if (++count >= GET2(code, 1)) { ADD_NEW_DATA(-(state_offset + 2 + IMM2_SIZE), 0, ncount); } else { ADD_NEW_DATA(-state_offset, count, ncount); } @@ -2047,7 +1976,13 @@ BOOL OK; switch (c) { - VSPACE_CASES: + case 0x000a: + case 0x000b: + case 0x000c: + case 0x000d: + case 0x0085: + case 0x2028: + case 0x2029: OK = TRUE; break; @@ -2062,7 +1997,7 @@ active_count--; /* Remove non-match possibility */ next_active_state--; } - if (++count >= (int)GET2(code, 1)) + if (++count >= GET2(code, 1)) { ADD_NEW_DATA(-(state_offset + 2 + IMM2_SIZE), 0, 0); } else { ADD_NEW_DATA(-state_offset, count, 0); } @@ -2083,7 +2018,25 @@ BOOL OK; switch (c) { - HSPACE_CASES: + case 0x09: /* HT */ + case 0x20: /* SPACE */ + case 0xa0: /* NBSP */ + case 0x1680: /* OGHAM SPACE MARK */ + case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ + case 0x2000: /* EN QUAD */ + case 0x2001: /* EM QUAD */ + case 0x2002: /* EN SPACE */ + case 0x2003: /* EM SPACE */ + case 0x2004: /* THREE-PER-EM SPACE */ + case 0x2005: /* FOUR-PER-EM SPACE */ + case 0x2006: /* SIX-PER-EM SPACE */ + case 0x2007: /* FIGURE SPACE */ + case 0x2008: /* PUNCTUATION SPACE */ + case 0x2009: /* THIN SPACE */ + case 0x200A: /* HAIR SPACE */ + case 0x202f: /* NARROW NO-BREAK SPACE */ + case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ + case 0x3000: /* IDEOGRAPHIC SPACE */ OK = TRUE; break; @@ -2099,7 +2052,7 @@ active_count--; /* Remove non-match possibility */ next_active_state--; } - if (++count >= (int)GET2(code, 1)) + if (++count >= GET2(code, 1)) { ADD_NEW_DATA(-(state_offset + 2 + IMM2_SIZE), 0, 0); } else { ADD_NEW_DATA(-state_offset, count, 0); } @@ -2159,21 +2112,17 @@ to wait for them to pass before continuing. */ case OP_EXTUNI: - if (clen > 0) + if (clen > 0 && UCD_CATEGORY(c) != ucp_M) { - int lgb, rgb; const pcre_uchar *nptr = ptr + clen; int ncount = 0; - lgb = UCD_GRAPHBREAK(c); while (nptr < end_subject) { - dlen = 1; - if (!utf) d = *nptr; else { GETCHARLEN(d, nptr, dlen); } - rgb = UCD_GRAPHBREAK(d); - if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break; + int nclen = 1; + GETCHARLEN(c, nptr, nclen); + if (UCD_CATEGORY(c) != ucp_M) break; ncount++; - lgb = rgb; - nptr += dlen; + nptr += nclen; } if (nptr >= end_subject && (md->moptions & PCRE_PARTIAL_HARD) != 0) reset_could_continue = TRUE; @@ -2190,27 +2139,25 @@ case OP_ANYNL: if (clen > 0) switch(c) { - case CHAR_VT: - case CHAR_FF: - case CHAR_NEL: -#ifndef EBCDIC + case 0x000b: + case 0x000c: + case 0x0085: case 0x2028: case 0x2029: -#endif /* Not EBCDIC */ if ((md->moptions & PCRE_BSR_ANYCRLF) != 0) break; - case CHAR_LF: + case 0x000a: ADD_NEW(state_offset + 1, 0); break; - case CHAR_CR: + case 0x000d: if (ptr + 1 >= end_subject) { ADD_NEW(state_offset + 1, 0); if ((md->moptions & PCRE_PARTIAL_HARD) != 0) reset_could_continue = TRUE; } - else if (RAWUCHARTEST(ptr + 1) == CHAR_LF) + else if (ptr[1] == 0x0a) { ADD_NEW_DATA(-(state_offset + 1), 0, 1); } @@ -2226,7 +2173,13 @@ case OP_NOT_VSPACE: if (clen > 0) switch(c) { - VSPACE_CASES: + case 0x000a: + case 0x000b: + case 0x000c: + case 0x000d: + case 0x0085: + case 0x2028: + case 0x2029: break; default: @@ -2239,12 +2192,17 @@ case OP_VSPACE: if (clen > 0) switch(c) { - VSPACE_CASES: + case 0x000a: + case 0x000b: + case 0x000c: + case 0x000d: + case 0x0085: + case 0x2028: + case 0x2029: ADD_NEW(state_offset + 1, 0); break; - default: - break; + default: break; } break; @@ -2252,7 +2210,25 @@ case OP_NOT_HSPACE: if (clen > 0) switch(c) { - HSPACE_CASES: + case 0x09: /* HT */ + case 0x20: /* SPACE */ + case 0xa0: /* NBSP */ + case 0x1680: /* OGHAM SPACE MARK */ + case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ + case 0x2000: /* EN QUAD */ + case 0x2001: /* EM QUAD */ + case 0x2002: /* EN SPACE */ + case 0x2003: /* EM SPACE */ + case 0x2004: /* THREE-PER-EM SPACE */ + case 0x2005: /* FOUR-PER-EM SPACE */ + case 0x2006: /* SIX-PER-EM SPACE */ + case 0x2007: /* FIGURE SPACE */ + case 0x2008: /* PUNCTUATION SPACE */ + case 0x2009: /* THIN SPACE */ + case 0x200A: /* HAIR SPACE */ + case 0x202f: /* NARROW NO-BREAK SPACE */ + case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ + case 0x3000: /* IDEOGRAPHIC SPACE */ break; default: @@ -2265,12 +2241,27 @@ case OP_HSPACE: if (clen > 0) switch(c) { - HSPACE_CASES: + case 0x09: /* HT */ + case 0x20: /* SPACE */ + case 0xa0: /* NBSP */ + case 0x1680: /* OGHAM SPACE MARK */ + case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ + case 0x2000: /* EN QUAD */ + case 0x2001: /* EM QUAD */ + case 0x2002: /* EN SPACE */ + case 0x2003: /* EM SPACE */ + case 0x2004: /* THREE-PER-EM SPACE */ + case 0x2005: /* FOUR-PER-EM SPACE */ + case 0x2006: /* SIX-PER-EM SPACE */ + case 0x2007: /* FIGURE SPACE */ + case 0x2008: /* PUNCTUATION SPACE */ + case 0x2009: /* THIN SPACE */ + case 0x200A: /* HAIR SPACE */ + case 0x202f: /* NARROW NO-BREAK SPACE */ + case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ + case 0x3000: /* IDEOGRAPHIC SPACE */ ADD_NEW(state_offset + 1, 0); break; - - default: - break; } break; @@ -2324,7 +2315,7 @@ if (count > 0) { ADD_ACTIVE(state_offset + dlen + 1, 0); } if (clen > 0) { - pcre_uint32 otherd = NOTACHAR; + unsigned int otherd = NOTACHAR; if (caseless) { #ifdef SUPPORT_UTF @@ -2371,7 +2362,7 @@ ADD_ACTIVE(state_offset + dlen + 1, 0); if (clen > 0) { - pcre_uint32 otherd = NOTACHAR; + unsigned int otherd = NOTACHAR; if (caseless) { #ifdef SUPPORT_UTF @@ -2416,7 +2407,7 @@ ADD_ACTIVE(state_offset + dlen + 1, 0); if (clen > 0) { - pcre_uint32 otherd = NOTACHAR; + unsigned int otherd = NOTACHAR; if (caseless) { #ifdef SUPPORT_UTF @@ -2453,7 +2444,7 @@ count = current_state->count; /* Number already matched */ if (clen > 0) { - pcre_uint32 otherd = NOTACHAR; + unsigned int otherd = NOTACHAR; if (caseless) { #ifdef SUPPORT_UTF @@ -2469,7 +2460,7 @@ } if ((c == d || c == otherd) == (codevalue < OP_NOTSTAR)) { - if (++count >= (int)GET2(code, 1)) + if (++count >= GET2(code, 1)) { ADD_NEW(state_offset + dlen + 1 + IMM2_SIZE, 0); } else { ADD_NEW(state_offset, count); } @@ -2497,7 +2488,7 @@ count = current_state->count; /* Number already matched */ if (clen > 0) { - pcre_uint32 otherd = NOTACHAR; + unsigned int otherd = NOTACHAR; if (caseless) { #ifdef SUPPORT_UTF @@ -2518,7 +2509,7 @@ active_count--; /* Remove non-match possibility */ next_active_state--; } - if (++count >= (int)GET2(code, 1)) + if (++count >= GET2(code, 1)) { ADD_NEW(state_offset + dlen + 1 + IMM2_SIZE, 0); } else { ADD_NEW(state_offset, count); } @@ -2571,65 +2562,31 @@ { case OP_CRSTAR: case OP_CRMINSTAR: - case OP_CRPOSSTAR: ADD_ACTIVE(next_state_offset + 1, 0); - if (isinclass) - { - if (*ecode == OP_CRPOSSTAR) - { - active_count--; /* Remove non-match possibility */ - next_active_state--; - } - ADD_NEW(state_offset, 0); - } + if (isinclass) { ADD_NEW(state_offset, 0); } break; case OP_CRPLUS: case OP_CRMINPLUS: - case OP_CRPOSPLUS: count = current_state->count; /* Already matched */ if (count > 0) { ADD_ACTIVE(next_state_offset + 1, 0); } - if (isinclass) - { - if (count > 0 && *ecode == OP_CRPOSPLUS) - { - active_count--; /* Remove non-match possibility */ - next_active_state--; - } - count++; - ADD_NEW(state_offset, count); - } + if (isinclass) { count++; ADD_NEW(state_offset, count); } break; case OP_CRQUERY: case OP_CRMINQUERY: - case OP_CRPOSQUERY: ADD_ACTIVE(next_state_offset + 1, 0); - if (isinclass) - { - if (*ecode == OP_CRPOSQUERY) - { - active_count--; /* Remove non-match possibility */ - next_active_state--; - } - ADD_NEW(next_state_offset + 1, 0); - } + if (isinclass) { ADD_NEW(next_state_offset + 1, 0); } break; case OP_CRRANGE: case OP_CRMINRANGE: - case OP_CRPOSRANGE: count = current_state->count; /* Already matched */ - if (count >= (int)GET2(ecode, 1)) + if (count >= GET2(ecode, 1)) { ADD_ACTIVE(next_state_offset + 1 + 2 * IMM2_SIZE, 0); } if (isinclass) { - int max = (int)GET2(ecode, 1 + IMM2_SIZE); - if (*ecode == OP_CRPOSRANGE) - { - active_count--; /* Remove non-match possibility */ - next_active_state--; - } + int max = GET2(ecode, 1 + IMM2_SIZE); if (++count >= max && max != 0) /* Max 0 => no limit */ { ADD_NEW(next_state_offset + 1 + 2 * IMM2_SIZE, 0); } else @@ -2705,12 +2662,10 @@ cb.version = 1; /* Version 1 of the callout block */ cb.callout_number = code[LINK_SIZE+2]; cb.offset_vector = offsets; -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 cb.subject = (PCRE_SPTR)start_subject; -#elif defined COMPILE_PCRE16 +#else cb.subject = (PCRE_SPTR16)start_subject; -#elif defined COMPILE_PCRE32 - cb.subject = (PCRE_SPTR32)start_subject; #endif cb.subject_length = (int)(end_subject - start_subject); cb.start_match = (int)(current_subject - start_subject); @@ -2729,11 +2684,9 @@ condcode = code[LINK_SIZE+1]; - /* Back reference conditions and duplicate named recursion conditions - are not supported */ + /* Back reference conditions are not supported */ - if (condcode == OP_CREF || condcode == OP_DNCREF || - condcode == OP_DNRREF) + if (condcode == OP_CREF || condcode == OP_NCREF) return PCRE_ERROR_DFA_UCOND; /* The DEFINE condition is always false */ @@ -2745,7 +2698,7 @@ which means "test if in any recursion". We can't test for specifically recursed groups. */ - else if (condcode == OP_RREF) + else if (condcode == OP_RREF || condcode == OP_NRREF) { int value = GET2(code, LINK_SIZE + 2); if (value != RREF_ANY) return PCRE_ERROR_DFA_UCOND; @@ -2843,7 +2796,7 @@ for (rc = rc*2 - 2; rc >= 0; rc -= 2) { int charcount = local_offsets[rc+1] - local_offsets[rc]; -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 +#ifdef SUPPORT_UTF if (utf) { const pcre_uchar *p = start_subject + local_offsets[rc]; @@ -2947,7 +2900,7 @@ const pcre_uchar *p = ptr; const pcre_uchar *pp = local_ptr; charcount = (int)(pp - p); -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 +#ifdef SUPPORT_UTF if (utf) while (p < pp) if (NOT_FIRSTCHAR(*p++)) charcount--; #endif ADD_NEW_DATA(-next_state_offset, 0, (charcount - 1)); @@ -3029,7 +2982,7 @@ } else { -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 +#ifdef SUPPORT_UTF if (utf) { const pcre_uchar *p = start_subject + local_offsets[0]; @@ -3058,12 +3011,10 @@ cb.version = 1; /* Version 1 of the callout block */ cb.callout_number = code[1]; cb.offset_vector = offsets; -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 cb.subject = (PCRE_SPTR)start_subject; -#elif defined COMPILE_PCRE16 +#else cb.subject = (PCRE_SPTR16)start_subject; -#elif defined COMPILE_PCRE32 - cb.subject = (PCRE_SPTR32)start_subject; #endif cb.subject_length = (int)(end_subject - start_subject); cb.start_match = (int)(current_subject - start_subject); @@ -3121,7 +3072,15 @@ ptr > md->start_used_ptr) /* Inspected non-empty string */ ) ) + { + if (offsetcount >= 2) + { + offsets[0] = (int)(md->start_used_ptr - start_subject); + offsets[1] = (int)(end_subject - start_subject); + } match_count = PCRE_ERROR_PARTIAL; + } + DPRINTF(("%.*sEnd of internal_dfa_exec %d: returning %d\n" "%.*s---------------------\n\n", rlevel*2-2, SP, rlevel, match_count, rlevel*2-2, SP)); @@ -3171,21 +3130,16 @@ < -1 => some kind of unexpected problem */ -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre_dfa_exec(const pcre *argument_re, const pcre_extra *extra_data, const char *subject, int length, int start_offset, int options, int *offsets, int offsetcount, int *workspace, int wscount) -#elif defined COMPILE_PCRE16 +#else PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre16_dfa_exec(const pcre16 *argument_re, const pcre16_extra *extra_data, PCRE_SPTR16 subject, int length, int start_offset, int options, int *offsets, int offsetcount, int *workspace, int wscount) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DEFN int PCRE_CALL_CONVENTION -pcre32_dfa_exec(const pcre32 *argument_re, const pcre32_extra *extra_data, - PCRE_SPTR32 subject, int length, int start_offset, int options, int *offsets, - int offsetcount, int *workspace, int wscount) #endif { REAL_PCRE *re = (REAL_PCRE *)argument_re; @@ -3212,7 +3166,6 @@ (offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL; if (offsetcount < 0) return PCRE_ERROR_BADCOUNT; if (wscount < 20) return PCRE_ERROR_DFA_WSSIZE; -if (length < 0) return PCRE_ERROR_BADLENGTH; if (start_offset < 0 || start_offset > length) return PCRE_ERROR_BADOFFSET; /* Check that the first field in the block is the magic number. If it is not, @@ -3261,7 +3214,7 @@ req_char_ptr = current_subject - 1; #ifdef SUPPORT_UTF -/* PCRE_UTF(16|32) have the same value as PCRE_UTF8. */ +/* PCRE_UTF16 has the same value as PCRE_UTF8. */ utf = (re->options & PCRE_UTF8) != 0; #else utf = FALSE; @@ -3347,21 +3300,12 @@ offsets[0] = erroroffset; offsets[1] = errorcode; } -#if defined COMPILE_PCRE8 - return (errorcode <= PCRE_UTF8_ERR5 && (options & PCRE_PARTIAL_HARD) != 0) ? + return (errorcode <= PCRE_UTF8_ERR5 && (options & PCRE_PARTIAL_HARD) != 0)? PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8; -#elif defined COMPILE_PCRE16 - return (errorcode <= PCRE_UTF16_ERR1 && (options & PCRE_PARTIAL_HARD) != 0) ? - PCRE_ERROR_SHORTUTF16 : PCRE_ERROR_BADUTF16; -#elif defined COMPILE_PCRE32 - return PCRE_ERROR_BADUTF32; -#endif } -#if defined COMPILE_PCRE8 || defined COMPILE_PCRE16 if (start_offset > 0 && start_offset < length && NOT_FIRSTCHAR(((PCRE_PUCHAR)subject)[start_offset])) return PCRE_ERROR_BADUTF8_OFFSET; -#endif } #endif @@ -3471,15 +3415,12 @@ if (has_first_char) { if (first_char != first_char2) - { - pcre_uchar csc; while (current_subject < end_subject && - (csc = RAWUCHARTEST(current_subject)) != first_char && csc != first_char2) + *current_subject != first_char && *current_subject != first_char2) current_subject++; - } else while (current_subject < end_subject && - RAWUCHARTEST(current_subject) != first_char) + *current_subject != first_char) current_subject++; } @@ -3509,10 +3450,10 @@ ANYCRLF, and we are now at a LF, advance the match position by one more character. */ - if (RAWUCHARTEST(current_subject - 1) == CHAR_CR && + if (current_subject[-1] == CHAR_CR && (md->nltype == NLTYPE_ANY || md->nltype == NLTYPE_ANYCRLF) && current_subject < end_subject && - RAWUCHARTEST(current_subject) == CHAR_NL) + *current_subject == CHAR_NL) current_subject++; } } @@ -3523,7 +3464,7 @@ { while (current_subject < end_subject) { - register pcre_uint32 c = RAWUCHARTEST(current_subject); + register unsigned int c = *current_subject; #ifndef COMPILE_PCRE8 if (c > 255) c = 255; #endif @@ -3589,7 +3530,7 @@ { while (p < end_subject) { - register pcre_uint32 pp = RAWUCHARINCTEST(p); + register int pp = *p++; if (pp == req_char || pp == req_char2) { p--; break; } } } @@ -3597,7 +3538,7 @@ { while (p < end_subject) { - if (RAWUCHARINCTEST(p) == req_char) { p--; break; } + if (*p++ == req_char) { p--; break; } } } @@ -3635,17 +3576,7 @@ /* Anything other than "no match" means we are done, always; otherwise, carry on only if not anchored. */ - if (rc != PCRE_ERROR_NOMATCH || anchored) - { - if (rc == PCRE_ERROR_PARTIAL && offsetcount >= 2) - { - offsets[0] = (int)(md->start_used_ptr - (PCRE_PUCHAR)subject); - offsets[1] = (int)(end_subject - (PCRE_PUCHAR)subject); - if (offsetcount > 2) - offsets[2] = (int)(current_subject - (PCRE_PUCHAR)subject); - } - return rc; - } + if (rc != PCRE_ERROR_NOMATCH || anchored) return rc; /* Advance to the next subject character unless we are at the end of a line and firstline is set. */ @@ -3665,9 +3596,9 @@ not contain any explicit matches for \r or \n, and the newline option is CRLF or ANY or ANYCRLF, advance the match position by one more character. */ - if (RAWUCHARTEST(current_subject - 1) == CHAR_CR && + if (current_subject[-1] == CHAR_CR && current_subject < end_subject && - RAWUCHARTEST(current_subject) == CHAR_NL && + *current_subject == CHAR_NL && (re->flags & PCRE_HASCRORLF) == 0 && (md->nltype == NLTYPE_ANY || md->nltype == NLTYPE_ANYCRLF || diff --git a/pcre/pcre_exec.c b/pcre/pcre_exec.c --- a/pcre/pcre_exec.c +++ b/pcre/pcre_exec.c @@ -6,7157 +6,7124 @@ and semantics are as close as possible to those of the Perl 5 language. Written by Philip Hazel - Copyright (c) 1997-2013 University of Cambridge - ------------------------------------------------------------------------------ -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. - - * Neither the name of the University of Cambridge 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 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. ------------------------------------------------------------------------------ -*/ - -/* This module contains pcre_exec(), the externally visible function that does -pattern matching using an NFA algorithm, trying to mimic Perl as closely as -possible. There are also some static supporting functions. */ - -#ifdef HAVE_CONFIG_H -#include "config.h" -#endif - -#define NLBLOCK md /* Block containing newline information */ -#define PSSTART start_subject /* Field containing processed string start */ -#define PSEND end_subject /* Field containing processed string end */ - -#include "pcre_internal.h" - -/* Undefine some potentially clashing cpp symbols */ - -#undef min -#undef max - -/* The md->capture_last field uses the lower 16 bits for the last captured -substring (which can never be greater than 65535) and a bit in the top half -to mean "capture vector overflowed". This odd way of doing things was -implemented when it was realized that preserving and restoring the overflow bit -whenever the last capture number was saved/restored made for a neater -interface, and doing it this way saved on (a) another variable, which would -have increased the stack frame size (a big NO-NO in PCRE) and (b) another -separate set of save/restore instructions. The following defines are used in -implementing this. */ - -#define CAPLMASK 0x0000ffff /* The bits used for last_capture */ -#define OVFLMASK 0xffff0000 /* The bits used for the overflow flag */ -#define OVFLBIT 0x00010000 /* The bit that is set for overflow */ - -/* Values for setting in md->match_function_type to indicate two special types -of call to match(). We do it this way to save on using another stack variable, -as stack usage is to be discouraged. */ - -#define MATCH_CONDASSERT 1 /* Called to check a condition assertion */ -#define MATCH_CBEGROUP 2 /* Could-be-empty unlimited repeat group */ - -/* Non-error returns from the match() function. Error returns are externally -defined PCRE_ERROR_xxx codes, which are all negative. */ - -#define MATCH_MATCH 1 -#define MATCH_NOMATCH 0 - -/* Special internal returns from the match() function. Make them sufficiently -negative to avoid the external error codes. */ - -#define MATCH_ACCEPT (-999) -#define MATCH_KETRPOS (-998) -#define MATCH_ONCE (-997) -/* The next 5 must be kept together and in sequence so that a test that checks -for any one of them can use a range. */ -#define MATCH_COMMIT (-996) -#define MATCH_PRUNE (-995) -#define MATCH_SKIP (-994) -#define MATCH_SKIP_ARG (-993) -#define MATCH_THEN (-992) -#define MATCH_BACKTRACK_MAX MATCH_THEN -#define MATCH_BACKTRACK_MIN MATCH_COMMIT - -/* Maximum number of ints of offset to save on the stack for recursive calls. -If the offset vector is bigger, malloc is used. This should be a multiple of 3, -because the offset vector is always a multiple of 3 long. */ - -#define REC_STACK_SAVE_MAX 30 - -/* Min and max values for the common repeats; for the maxima, 0 => infinity */ - -static const char rep_min[] = { 0, 0, 1, 1, 0, 0, 0, 0, 0, 1, 0, }; -static const char rep_max[] = { 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 1, }; - -#ifdef PCRE_DEBUG -/************************************************* -* Debugging function to print chars * -*************************************************/ - -/* Print a sequence of chars in printable format, stopping at the end of the -subject if the requested. - -Arguments: - p points to characters - length number to print - is_subject TRUE if printing from within md->start_subject - md pointer to matching data block, if is_subject is TRUE - -Returns: nothing -*/ - -static void -pchars(const pcre_uchar *p, int length, BOOL is_subject, match_data *md) -{ -pcre_uint32 c; -BOOL utf = md->utf; -if (is_subject && length > md->end_subject - p) length = md->end_subject - p; -while (length-- > 0) - if (isprint(c = RAWUCHARINCTEST(p))) printf("%c", (char)c); else printf("\\x{%02x}", c); -} -#endif - - - -/************************************************* -* Match a back-reference * -*************************************************/ - -/* Normally, if a back reference hasn't been set, the length that is passed is -negative, so the match always fails. However, in JavaScript compatibility mode, -the length passed is zero. Note that in caseless UTF-8 mode, the number of -subject bytes matched may be different to the number of reference bytes. - -Arguments: - offset index into the offset vector - eptr pointer into the subject - length length of reference to be matched (number of bytes) - md points to match data block - caseless TRUE if caseless - -Returns: >= 0 the number of subject bytes matched - -1 no match - -2 partial match; always given if at end subject -*/ - -static int -match_ref(int offset, register PCRE_PUCHAR eptr, int length, match_data *md, - BOOL caseless) -{ -PCRE_PUCHAR eptr_start = eptr; -register PCRE_PUCHAR p = md->start_subject + md->offset_vector[offset]; -#if defined SUPPORT_UTF && defined SUPPORT_UCP -BOOL utf = md->utf; -#endif - -#ifdef PCRE_DEBUG -if (eptr >= md->end_subject) - printf("matching subject "); -else - { - printf("matching subject "); - pchars(eptr, length, TRUE, md); - } -printf(" against backref "); -pchars(p, length, FALSE, md); -printf("\n"); -#endif - -/* Always fail if reference not set (and not JavaScript compatible - in that -case the length is passed as zero). */ - -if (length < 0) return -1; - -/* Separate the caseless case for speed. In UTF-8 mode we can only do this -properly if Unicode properties are supported. Otherwise, we can check only -ASCII characters. */ - -if (caseless) - { -#if defined SUPPORT_UTF && defined SUPPORT_UCP - if (utf) - { - /* Match characters up to the end of the reference. NOTE: the number of - data units matched may differ, because in UTF-8 there are some characters - whose upper and lower case versions code have different numbers of bytes. - For example, U+023A (2 bytes in UTF-8) is the upper case version of U+2C65 - (3 bytes in UTF-8); a sequence of 3 of the former uses 6 bytes, as does a - sequence of two of the latter. It is important, therefore, to check the - length along the reference, not along the subject (earlier code did this - wrong). */ - - PCRE_PUCHAR endptr = p + length; - while (p < endptr) - { - pcre_uint32 c, d; - const ucd_record *ur; - if (eptr >= md->end_subject) return -2; /* Partial match */ - GETCHARINC(c, eptr); - GETCHARINC(d, p); - ur = GET_UCD(d); - if (c != d && c != d + ur->other_case) - { - const pcre_uint32 *pp = PRIV(ucd_caseless_sets) + ur->caseset; - for (;;) - { - if (c < *pp) return -1; - if (c == *pp++) break; - } - } - } - } - else -#endif - - /* The same code works when not in UTF-8 mode and in UTF-8 mode when there - is no UCP support. */ - { - while (length-- > 0) - { - pcre_uint32 cc, cp; - if (eptr >= md->end_subject) return -2; /* Partial match */ - cc = RAWUCHARTEST(eptr); - cp = RAWUCHARTEST(p); - if (TABLE_GET(cp, md->lcc, cp) != TABLE_GET(cc, md->lcc, cc)) return -1; - p++; - eptr++; - } - } - } - -/* In the caseful case, we can just compare the bytes, whether or not we -are in UTF-8 mode. */ - -else - { - while (length-- > 0) - { - if (eptr >= md->end_subject) return -2; /* Partial match */ - if (RAWUCHARINCTEST(p) != RAWUCHARINCTEST(eptr)) return -1; - } - } - -return (int)(eptr - eptr_start); -} - - - -/*************************************************************************** -**************************************************************************** - RECURSION IN THE match() FUNCTION - -The match() function is highly recursive, though not every recursive call -increases the recursive depth. Nevertheless, some regular expressions can cause -it to recurse to a great depth. I was writing for Unix, so I just let it call -itself recursively. This uses the stack for saving everything that has to be -saved for a recursive call. On Unix, the stack can be large, and this works -fine. - -It turns out that on some non-Unix-like systems there are problems with -programs that use a lot of stack. (This despite the fact that every last chip -has oodles of memory these days, and techniques for extending the stack have -been known for decades.) So.... - -There is a fudge, triggered by defining NO_RECURSE, which avoids recursive -calls by keeping local variables that need to be preserved in blocks of memory -obtained from malloc() instead instead of on the stack. Macros are used to -achieve this so that the actual code doesn't look very different to what it -always used to. - -The original heap-recursive code used longjmp(). However, it seems that this -can be very slow on some operating systems. Following a suggestion from Stan -Switzer, the use of longjmp() has been abolished, at the cost of having to -provide a unique number for each call to RMATCH. There is no way of generating -a sequence of numbers at compile time in C. I have given them names, to make -them stand out more clearly. - -Crude tests on x86 Linux show a small speedup of around 5-8%. However, on -FreeBSD, avoiding longjmp() more than halves the time taken to run the standard -tests. Furthermore, not using longjmp() means that local dynamic variables -don't have indeterminate values; this has meant that the frame size can be -reduced because the result can be "passed back" by straight setting of the -variable instead of being passed in the frame. -**************************************************************************** -***************************************************************************/ - -/* Numbers for RMATCH calls. When this list is changed, the code at HEAP_RETURN -below must be updated in sync. */ - -enum { RM1=1, RM2, RM3, RM4, RM5, RM6, RM7, RM8, RM9, RM10, - RM11, RM12, RM13, RM14, RM15, RM16, RM17, RM18, RM19, RM20, - RM21, RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30, - RM31, RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40, - RM41, RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50, - RM51, RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60, - RM61, RM62, RM63, RM64, RM65, RM66, RM67 }; - -/* These versions of the macros use the stack, as normal. There are debugging -versions and production versions. Note that the "rw" argument of RMATCH isn't -actually used in this definition. */ - -#ifndef NO_RECURSE -#define REGISTER register - -#ifdef PCRE_DEBUG -#define RMATCH(ra,rb,rc,rd,re,rw) \ - { \ - printf("match() called in line %d\n", __LINE__); \ - rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1); \ - printf("to line %d\n", __LINE__); \ - } -#define RRETURN(ra) \ - { \ - printf("match() returned %d from line %d\n", ra, __LINE__); \ - return ra; \ - } -#else -#define RMATCH(ra,rb,rc,rd,re,rw) \ - rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1) -#define RRETURN(ra) return ra -#endif - -#else - - -/* These versions of the macros manage a private stack on the heap. Note that -the "rd" argument of RMATCH isn't actually used in this definition. It's the md -argument of match(), which never changes. */ - -#define REGISTER - -#define RMATCH(ra,rb,rc,rd,re,rw)\ - {\ - heapframe *newframe = frame->Xnextframe;\ - if (newframe == NULL)\ - {\ - newframe = (heapframe *)(PUBL(stack_malloc))(sizeof(heapframe));\ - if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\ - newframe->Xnextframe = NULL;\ - frame->Xnextframe = newframe;\ - }\ - frame->Xwhere = rw;\ - newframe->Xeptr = ra;\ - newframe->Xecode = rb;\ - newframe->Xmstart = mstart;\ - newframe->Xoffset_top = rc;\ - newframe->Xeptrb = re;\ - newframe->Xrdepth = frame->Xrdepth + 1;\ - newframe->Xprevframe = frame;\ - frame = newframe;\ - DPRINTF(("restarting from line %d\n", __LINE__));\ - goto HEAP_RECURSE;\ - L_##rw:\ - DPRINTF(("jumped back to line %d\n", __LINE__));\ - } - -#define RRETURN(ra)\ - {\ - heapframe *oldframe = frame;\ - frame = oldframe->Xprevframe;\ - if (frame != NULL)\ - {\ - rrc = ra;\ - goto HEAP_RETURN;\ - }\ - return ra;\ - } - - -/* Structure for remembering the local variables in a private frame */ - -typedef struct heapframe { - struct heapframe *Xprevframe; - struct heapframe *Xnextframe; - - /* Function arguments that may change */ - - PCRE_PUCHAR Xeptr; - const pcre_uchar *Xecode; - PCRE_PUCHAR Xmstart; - int Xoffset_top; - eptrblock *Xeptrb; - unsigned int Xrdepth; - - /* Function local variables */ - - PCRE_PUCHAR Xcallpat; -#ifdef SUPPORT_UTF - PCRE_PUCHAR Xcharptr; -#endif - PCRE_PUCHAR Xdata; - PCRE_PUCHAR Xnext; - PCRE_PUCHAR Xpp; - PCRE_PUCHAR Xprev; - PCRE_PUCHAR Xsaved_eptr; - - recursion_info Xnew_recursive; - - BOOL Xcur_is_word; - BOOL Xcondition; - BOOL Xprev_is_word; - -#ifdef SUPPORT_UCP - int Xprop_type; - unsigned int Xprop_value; - int Xprop_fail_result; - int Xoclength; - pcre_uchar Xocchars[6]; -#endif - - int Xcodelink; - int Xctype; - unsigned int Xfc; - int Xfi; - int Xlength; - int Xmax; - int Xmin; - unsigned int Xnumber; - int Xoffset; - unsigned int Xop; - pcre_int32 Xsave_capture_last; - int Xsave_offset1, Xsave_offset2, Xsave_offset3; - int Xstacksave[REC_STACK_SAVE_MAX]; - - eptrblock Xnewptrb; - - /* Where to jump back to */ - - int Xwhere; - -} heapframe; - -#endif - - -/*************************************************************************** -***************************************************************************/ - - - -/************************************************* -* Match from current position * -*************************************************/ - -/* This function is called recursively in many circumstances. Whenever it -returns a negative (error) response, the outer incarnation must also return the -same response. */ - -/* These macros pack up tests that are used for partial matching, and which -appear several times in the code. We set the "hit end" flag if the pointer is -at the end of the subject and also past the start of the subject (i.e. -something has been matched). For hard partial matching, we then return -immediately. The second one is used when we already know we are past the end of -the subject. */ - -#define CHECK_PARTIAL()\ - if (md->partial != 0 && eptr >= md->end_subject && \ - eptr > md->start_used_ptr) \ - { \ - md->hitend = TRUE; \ - if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \ - } - -#define SCHECK_PARTIAL()\ - if (md->partial != 0 && eptr > md->start_used_ptr) \ - { \ - md->hitend = TRUE; \ - if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \ - } - - -/* Performance note: It might be tempting to extract commonly used fields from -the md structure (e.g. utf, end_subject) into individual variables to improve -performance. Tests using gcc on a SPARC disproved this; in the first case, it -made performance worse. - -Arguments: - eptr pointer to current character in subject - ecode pointer to current position in compiled code - mstart pointer to the current match start position (can be modified - by encountering \K) - offset_top current top pointer - md pointer to "static" info for the match - eptrb pointer to chain of blocks containing eptr at start of - brackets - for testing for empty matches - rdepth the recursion depth - -Returns: MATCH_MATCH if matched ) these values are >= 0 - MATCH_NOMATCH if failed to match ) - a negative MATCH_xxx value for PRUNE, SKIP, etc - a negative PCRE_ERROR_xxx value if aborted by an error condition - (e.g. stopped by repeated call or recursion limit) -*/ - -static int -match(REGISTER PCRE_PUCHAR eptr, REGISTER const pcre_uchar *ecode, - PCRE_PUCHAR mstart, int offset_top, match_data *md, eptrblock *eptrb, - unsigned int rdepth) -{ -/* These variables do not need to be preserved over recursion in this function, -so they can be ordinary variables in all cases. Mark some of them with -"register" because they are used a lot in loops. */ - -register int rrc; /* Returns from recursive calls */ -register int i; /* Used for loops not involving calls to RMATCH() */ -register pcre_uint32 c; /* Character values not kept over RMATCH() calls */ -register BOOL utf; /* Local copy of UTF flag for speed */ - -BOOL minimize, possessive; /* Quantifier options */ -BOOL caseless; -int condcode; - -/* When recursion is not being used, all "local" variables that have to be -preserved over calls to RMATCH() are part of a "frame". We set up the top-level -frame on the stack here; subsequent instantiations are obtained from the heap -whenever RMATCH() does a "recursion". See the macro definitions above. Putting -the top-level on the stack rather than malloc-ing them all gives a performance -boost in many cases where there is not much "recursion". */ - -#ifdef NO_RECURSE -heapframe *frame = (heapframe *)md->match_frames_base; - -/* Copy in the original argument variables */ - -frame->Xeptr = eptr; -frame->Xecode = ecode; -frame->Xmstart = mstart; -frame->Xoffset_top = offset_top; -frame->Xeptrb = eptrb; -frame->Xrdepth = rdepth; - -/* This is where control jumps back to to effect "recursion" */ - -HEAP_RECURSE: - -/* Macros make the argument variables come from the current frame */ - -#define eptr frame->Xeptr -#define ecode frame->Xecode -#define mstart frame->Xmstart -#define offset_top frame->Xoffset_top -#define eptrb frame->Xeptrb -#define rdepth frame->Xrdepth - -/* Ditto for the local variables */ - -#ifdef SUPPORT_UTF -#define charptr frame->Xcharptr -#endif -#define callpat frame->Xcallpat -#define codelink frame->Xcodelink -#define data frame->Xdata -#define next frame->Xnext -#define pp frame->Xpp -#define prev frame->Xprev -#define saved_eptr frame->Xsaved_eptr - -#define new_recursive frame->Xnew_recursive - -#define cur_is_word frame->Xcur_is_word -#define condition frame->Xcondition -#define prev_is_word frame->Xprev_is_word - -#ifdef SUPPORT_UCP -#define prop_type frame->Xprop_type -#define prop_value frame->Xprop_value -#define prop_fail_result frame->Xprop_fail_result -#define oclength frame->Xoclength -#define occhars frame->Xocchars -#endif - -#define ctype frame->Xctype -#define fc frame->Xfc -#define fi frame->Xfi -#define length frame->Xlength -#define max frame->Xmax -#define min frame->Xmin -#define number frame->Xnumber -#define offset frame->Xoffset -#define op frame->Xop -#define save_capture_last frame->Xsave_capture_last -#define save_offset1 frame->Xsave_offset1 -#define save_offset2 frame->Xsave_offset2 -#define save_offset3 frame->Xsave_offset3 -#define stacksave frame->Xstacksave - -#define newptrb frame->Xnewptrb - -/* When recursion is being used, local variables are allocated on the stack and -get preserved during recursion in the normal way. In this environment, fi and -i, and fc and c, can be the same variables. */ - -#else /* NO_RECURSE not defined */ -#define fi i -#define fc c - -/* Many of the following variables are used only in small blocks of the code. -My normal style of coding would have declared them within each of those blocks. -However, in order to accommodate the version of this code that uses an external -"stack" implemented on the heap, it is easier to declare them all here, so the -declarations can be cut out in a block. The only declarations within blocks -below are for variables that do not have to be preserved over a recursive call -to RMATCH(). */ - -#ifdef SUPPORT_UTF -const pcre_uchar *charptr; -#endif -const pcre_uchar *callpat; -const pcre_uchar *data; -const pcre_uchar *next; -PCRE_PUCHAR pp; -const pcre_uchar *prev; -PCRE_PUCHAR saved_eptr; - -recursion_info new_recursive; - -BOOL cur_is_word; -BOOL condition; -BOOL prev_is_word; - -#ifdef SUPPORT_UCP -int prop_type; -unsigned int prop_value; -int prop_fail_result; -int oclength; -pcre_uchar occhars[6]; -#endif - -int codelink; -int ctype; -int length; -int max; -int min; -unsigned int number; -int offset; -unsigned int op; -pcre_int32 save_capture_last; -int save_offset1, save_offset2, save_offset3; -int stacksave[REC_STACK_SAVE_MAX]; - -eptrblock newptrb; - -/* There is a special fudge for calling match() in a way that causes it to -measure the size of its basic stack frame when the stack is being used for -recursion. The second argument (ecode) being NULL triggers this behaviour. It -cannot normally ever be NULL. The return is the negated value of the frame -size. */ - -if (ecode == NULL) - { - if (rdepth == 0) - return match((PCRE_PUCHAR)&rdepth, NULL, NULL, 0, NULL, NULL, 1); - else - { - int len = (char *)&rdepth - (char *)eptr; - return (len > 0)? -len : len; - } - } -#endif /* NO_RECURSE */ - -/* To save space on the stack and in the heap frame, I have doubled up on some -of the local variables that are used only in localised parts of the code, but -still need to be preserved over recursive calls of match(). These macros define -the alternative names that are used. */ - -#define allow_zero cur_is_word -#define cbegroup condition -#define code_offset codelink -#define condassert condition -#define matched_once prev_is_word -#define foc number -#define save_mark data - -/* These statements are here to stop the compiler complaining about unitialized -variables. */ - -#ifdef SUPPORT_UCP -prop_value = 0; -prop_fail_result = 0; -#endif - - -/* This label is used for tail recursion, which is used in a few cases even -when NO_RECURSE is not defined, in order to reduce the amount of stack that is -used. Thanks to Ian Taylor for noticing this possibility and sending the -original patch. */ - -TAIL_RECURSE: - -/* OK, now we can get on with the real code of the function. Recursive calls -are specified by the macro RMATCH and RRETURN is used to return. When -NO_RECURSE is *not* defined, these just turn into a recursive call to match() -and a "return", respectively (possibly with some debugging if PCRE_DEBUG is -defined). However, RMATCH isn't like a function call because it's quite a -complicated macro. It has to be used in one particular way. This shouldn't, -however, impact performance when true recursion is being used. */ - -#ifdef SUPPORT_UTF -utf = md->utf; /* Local copy of the flag */ -#else -utf = FALSE; -#endif - -/* First check that we haven't called match() too many times, or that we -haven't exceeded the recursive call limit. */ - -if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT); -if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT); - -/* At the start of a group with an unlimited repeat that may match an empty -string, the variable md->match_function_type is set to MATCH_CBEGROUP. It is -done this way to save having to use another function argument, which would take -up space on the stack. See also MATCH_CONDASSERT below. - -When MATCH_CBEGROUP is set, add the current subject pointer to the chain of -such remembered pointers, to be checked when we hit the closing ket, in order -to break infinite loops that match no characters. When match() is called in -other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must -NOT be used with tail recursion, because the memory block that is used is on -the stack, so a new one may be required for each match(). */ - -if (md->match_function_type == MATCH_CBEGROUP) - { - newptrb.epb_saved_eptr = eptr; - newptrb.epb_prev = eptrb; - eptrb = &newptrb; - md->match_function_type = 0; - } - -/* Now start processing the opcodes. */ - -for (;;) - { - minimize = possessive = FALSE; - op = *ecode; - - switch(op) - { - case OP_MARK: - md->nomatch_mark = ecode + 2; - md->mark = NULL; /* In case previously set by assertion */ - RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md, - eptrb, RM55); - if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) && - md->mark == NULL) md->mark = ecode + 2; - - /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an - argument, and we must check whether that argument matches this MARK's - argument. It is passed back in md->start_match_ptr (an overloading of that - variable). If it does match, we reset that variable to the current subject - position and return MATCH_SKIP. Otherwise, pass back the return code - unaltered. */ - - else if (rrc == MATCH_SKIP_ARG && - STRCMP_UC_UC_TEST(ecode + 2, md->start_match_ptr) == 0) - { - md->start_match_ptr = eptr; - RRETURN(MATCH_SKIP); - } - RRETURN(rrc); - - case OP_FAIL: - RRETURN(MATCH_NOMATCH); - - case OP_COMMIT: - RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, - eptrb, RM52); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - RRETURN(MATCH_COMMIT); - - case OP_PRUNE: - RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, - eptrb, RM51); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - RRETURN(MATCH_PRUNE); - - case OP_PRUNE_ARG: - md->nomatch_mark = ecode + 2; - md->mark = NULL; /* In case previously set by assertion */ - RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md, - eptrb, RM56); - if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) && - md->mark == NULL) md->mark = ecode + 2; - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - RRETURN(MATCH_PRUNE); - - case OP_SKIP: - RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, - eptrb, RM53); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - md->start_match_ptr = eptr; /* Pass back current position */ - RRETURN(MATCH_SKIP); - - /* Note that, for Perl compatibility, SKIP with an argument does NOT set - nomatch_mark. When a pattern match ends with a SKIP_ARG for which there was - not a matching mark, we have to re-run the match, ignoring the SKIP_ARG - that failed and any that precede it (either they also failed, or were not - triggered). To do this, we maintain a count of executed SKIP_ARGs. If a - SKIP_ARG gets to top level, the match is re-run with md->ignore_skip_arg - set to the count of the one that failed. */ - - case OP_SKIP_ARG: - md->skip_arg_count++; - if (md->skip_arg_count <= md->ignore_skip_arg) - { - ecode += PRIV(OP_lengths)[*ecode] + ecode[1]; - break; - } - RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md, - eptrb, RM57); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - - /* Pass back the current skip name by overloading md->start_match_ptr and - returning the special MATCH_SKIP_ARG return code. This will either be - caught by a matching MARK, or get to the top, where it causes a rematch - with md->ignore_skip_arg set to the value of md->skip_arg_count. */ - - md->start_match_ptr = ecode + 2; - RRETURN(MATCH_SKIP_ARG); - - /* For THEN (and THEN_ARG) we pass back the address of the opcode, so that - the branch in which it occurs can be determined. Overload the start of - match pointer to do this. */ - - case OP_THEN: - RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, - eptrb, RM54); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - md->start_match_ptr = ecode; - RRETURN(MATCH_THEN); - - case OP_THEN_ARG: - md->nomatch_mark = ecode + 2; - md->mark = NULL; /* In case previously set by assertion */ - RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, - md, eptrb, RM58); - if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) && - md->mark == NULL) md->mark = ecode + 2; - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - md->start_match_ptr = ecode; - RRETURN(MATCH_THEN); - - /* Handle an atomic group that does not contain any capturing parentheses. - This can be handled like an assertion. Prior to 8.13, all atomic groups - were handled this way. In 8.13, the code was changed as below for ONCE, so - that backups pass through the group and thereby reset captured values. - However, this uses a lot more stack, so in 8.20, atomic groups that do not - contain any captures generate OP_ONCE_NC, which can be handled in the old, - less stack intensive way. - - Check the alternative branches in turn - the matching won't pass the KET - for this kind of subpattern. If any one branch matches, we carry on as at - the end of a normal bracket, leaving the subject pointer, but resetting - the start-of-match value in case it was changed by \K. */ - - case OP_ONCE_NC: - prev = ecode; - saved_eptr = eptr; - save_mark = md->mark; - do - { - RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64); - if (rrc == MATCH_MATCH) /* Note: _not_ MATCH_ACCEPT */ - { - mstart = md->start_match_ptr; - break; - } - if (rrc == MATCH_THEN) - { - next = ecode + GET(ecode,1); - if (md->start_match_ptr < next && - (*ecode == OP_ALT || *next == OP_ALT)) - rrc = MATCH_NOMATCH; - } - - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - ecode += GET(ecode,1); - md->mark = save_mark; - } - while (*ecode == OP_ALT); - - /* If hit the end of the group (which could be repeated), fail */ - - if (*ecode != OP_ONCE_NC && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH); - - /* Continue as from after the group, updating the offsets high water - mark, since extracts may have been taken. */ - - do ecode += GET(ecode, 1); while (*ecode == OP_ALT); - - offset_top = md->end_offset_top; - eptr = md->end_match_ptr; - - /* For a non-repeating ket, just continue at this level. This also - happens for a repeating ket if no characters were matched in the group. - This is the forcible breaking of infinite loops as implemented in Perl - 5.005. */ - - if (*ecode == OP_KET || eptr == saved_eptr) - { - ecode += 1+LINK_SIZE; - break; - } - - /* The repeating kets try the rest of the pattern or restart from the - preceding bracket, in the appropriate order. The second "call" of match() - uses tail recursion, to avoid using another stack frame. */ - - if (*ecode == OP_KETRMIN) - { - RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - ecode = prev; - goto TAIL_RECURSE; - } - else /* OP_KETRMAX */ - { - RMATCH(eptr, prev, offset_top, md, eptrb, RM66); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - ecode += 1 + LINK_SIZE; - goto TAIL_RECURSE; - } - /* Control never gets here */ - - /* Handle a capturing bracket, other than those that are possessive with an - unlimited repeat. If there is space in the offset vector, save the current - subject position in the working slot at the top of the vector. We mustn't - change the current values of the data slot, because they may be set from a - previous iteration of this group, and be referred to by a reference inside - the group. A failure to match might occur after the group has succeeded, - if something later on doesn't match. For this reason, we need to restore - the working value and also the values of the final offsets, in case they - were set by a previous iteration of the same bracket. - - If there isn't enough space in the offset vector, treat this as if it were - a non-capturing bracket. Don't worry about setting the flag for the error - case here; that is handled in the code for KET. */ - - case OP_CBRA: - case OP_SCBRA: - number = GET2(ecode, 1+LINK_SIZE); - offset = number << 1; - -#ifdef PCRE_DEBUG - printf("start bracket %d\n", number); - printf("subject="); - pchars(eptr, 16, TRUE, md); - printf("\n"); -#endif - - if (offset < md->offset_max) - { - save_offset1 = md->offset_vector[offset]; - save_offset2 = md->offset_vector[offset+1]; - save_offset3 = md->offset_vector[md->offset_end - number]; - save_capture_last = md->capture_last; - save_mark = md->mark; - - DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3)); - md->offset_vector[md->offset_end - number] = - (int)(eptr - md->start_subject); - - for (;;) - { - if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; - RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, - eptrb, RM1); - if (rrc == MATCH_ONCE) break; /* Backing up through an atomic group */ - - /* If we backed up to a THEN, check whether it is within the current - branch by comparing the address of the THEN that is passed back with - the end of the branch. If it is within the current branch, and the - branch is one of two or more alternatives (it either starts or ends - with OP_ALT), we have reached the limit of THEN's action, so convert - the return code to NOMATCH, which will cause normal backtracking to - happen from now on. Otherwise, THEN is passed back to an outer - alternative. This implements Perl's treatment of parenthesized groups, - where a group not containing | does not affect the current alternative, - that is, (X) is NOT the same as (X|(*F)). */ - - if (rrc == MATCH_THEN) - { - next = ecode + GET(ecode,1); - if (md->start_match_ptr < next && - (*ecode == OP_ALT || *next == OP_ALT)) - rrc = MATCH_NOMATCH; - } - - /* Anything other than NOMATCH is passed back. */ - - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - md->capture_last = save_capture_last; - ecode += GET(ecode, 1); - md->mark = save_mark; - if (*ecode != OP_ALT) break; - } - - DPRINTF(("bracket %d failed\n", number)); - md->offset_vector[offset] = save_offset1; - md->offset_vector[offset+1] = save_offset2; - md->offset_vector[md->offset_end - number] = save_offset3; - - /* At this point, rrc will be one of MATCH_ONCE or MATCH_NOMATCH. */ - - RRETURN(rrc); - } - - /* FALL THROUGH ... Insufficient room for saving captured contents. Treat - as a non-capturing bracket. */ - - /* VVVVVVVVVVVVVVVVVVVVVVVVV */ - /* VVVVVVVVVVVVVVVVVVVVVVVVV */ - - DPRINTF(("insufficient capture room: treat as non-capturing\n")); - - /* VVVVVVVVVVVVVVVVVVVVVVVVV */ - /* VVVVVVVVVVVVVVVVVVVVVVVVV */ - - /* Non-capturing or atomic group, except for possessive with unlimited - repeat and ONCE group with no captures. Loop for all the alternatives. - - When we get to the final alternative within the brackets, we used to return - the result of a recursive call to match() whatever happened so it was - possible to reduce stack usage by turning this into a tail recursion, - except in the case of a possibly empty group. However, now that there is - the possiblity of (*THEN) occurring in the final alternative, this - optimization is no longer always possible. - - We can optimize if we know there are no (*THEN)s in the pattern; at present - this is the best that can be done. - - MATCH_ONCE is returned when the end of an atomic group is successfully - reached, but subsequent matching fails. It passes back up the tree (causing - captured values to be reset) until the original atomic group level is - reached. This is tested by comparing md->once_target with the start of the - group. At this point, the return is converted into MATCH_NOMATCH so that - previous backup points can be taken. */ - - case OP_ONCE: - case OP_BRA: - case OP_SBRA: - DPRINTF(("start non-capturing bracket\n")); - - for (;;) - { - if (op >= OP_SBRA || op == OP_ONCE) - md->match_function_type = MATCH_CBEGROUP; - - /* If this is not a possibly empty group, and there are no (*THEN)s in - the pattern, and this is the final alternative, optimize as described - above. */ - - else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT) - { - ecode += PRIV(OP_lengths)[*ecode]; - goto TAIL_RECURSE; - } - - /* In all other cases, we have to make another call to match(). */ - - save_mark = md->mark; - save_capture_last = md->capture_last; - RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, eptrb, - RM2); - - /* See comment in the code for capturing groups above about handling - THEN. */ - - if (rrc == MATCH_THEN) - { - next = ecode + GET(ecode,1); - if (md->start_match_ptr < next && - (*ecode == OP_ALT || *next == OP_ALT)) - rrc = MATCH_NOMATCH; - } - - if (rrc != MATCH_NOMATCH) - { - if (rrc == MATCH_ONCE) - { - const pcre_uchar *scode = ecode; - if (*scode != OP_ONCE) /* If not at start, find it */ - { - while (*scode == OP_ALT) scode += GET(scode, 1); - scode -= GET(scode, 1); - } - if (md->once_target == scode) rrc = MATCH_NOMATCH; - } - RRETURN(rrc); - } - ecode += GET(ecode, 1); - md->mark = save_mark; - if (*ecode != OP_ALT) break; - md->capture_last = save_capture_last; - } - - RRETURN(MATCH_NOMATCH); - - /* Handle possessive capturing brackets with an unlimited repeat. We come - here from BRAZERO with allow_zero set TRUE. The offset_vector values are - handled similarly to the normal case above. However, the matching is - different. The end of these brackets will always be OP_KETRPOS, which - returns MATCH_KETRPOS without going further in the pattern. By this means - we can handle the group by iteration rather than recursion, thereby - reducing the amount of stack needed. */ - - case OP_CBRAPOS: - case OP_SCBRAPOS: - allow_zero = FALSE; - - POSSESSIVE_CAPTURE: - number = GET2(ecode, 1+LINK_SIZE); - offset = number << 1; - -#ifdef PCRE_DEBUG - printf("start possessive bracket %d\n", number); - printf("subject="); - pchars(eptr, 16, TRUE, md); - printf("\n"); -#endif - - if (offset < md->offset_max) - { - matched_once = FALSE; - code_offset = (int)(ecode - md->start_code); - - save_offset1 = md->offset_vector[offset]; - save_offset2 = md->offset_vector[offset+1]; - save_offset3 = md->offset_vector[md->offset_end - number]; - save_capture_last = md->capture_last; - - DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3)); - - /* Each time round the loop, save the current subject position for use - when the group matches. For MATCH_MATCH, the group has matched, so we - restart it with a new subject starting position, remembering that we had - at least one match. For MATCH_NOMATCH, carry on with the alternatives, as - usual. If we haven't matched any alternatives in any iteration, check to - see if a previous iteration matched. If so, the group has matched; - continue from afterwards. Otherwise it has failed; restore the previous - capture values before returning NOMATCH. */ - - for (;;) - { - md->offset_vector[md->offset_end - number] = - (int)(eptr - md->start_subject); - if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; - RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, - eptrb, RM63); - if (rrc == MATCH_KETRPOS) - { - offset_top = md->end_offset_top; - eptr = md->end_match_ptr; - ecode = md->start_code + code_offset; - save_capture_last = md->capture_last; - matched_once = TRUE; - mstart = md->start_match_ptr; /* In case \K changed it */ - continue; - } - - /* See comment in the code for capturing groups above about handling - THEN. */ - - if (rrc == MATCH_THEN) - { - next = ecode + GET(ecode,1); - if (md->start_match_ptr < next && - (*ecode == OP_ALT || *next == OP_ALT)) - rrc = MATCH_NOMATCH; - } - - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - md->capture_last = save_capture_last; - ecode += GET(ecode, 1); - if (*ecode != OP_ALT) break; - } - - if (!matched_once) - { - md->offset_vector[offset] = save_offset1; - md->offset_vector[offset+1] = save_offset2; - md->offset_vector[md->offset_end - number] = save_offset3; - } - - if (allow_zero || matched_once) - { - ecode += 1 + LINK_SIZE; - break; - } - - RRETURN(MATCH_NOMATCH); - } - - /* FALL THROUGH ... Insufficient room for saving captured contents. Treat - as a non-capturing bracket. */ - - /* VVVVVVVVVVVVVVVVVVVVVVVVV */ - /* VVVVVVVVVVVVVVVVVVVVVVVVV */ - - DPRINTF(("insufficient capture room: treat as non-capturing\n")); - - /* VVVVVVVVVVVVVVVVVVVVVVVVV */ - /* VVVVVVVVVVVVVVVVVVVVVVVVV */ - - /* Non-capturing possessive bracket with unlimited repeat. We come here - from BRAZERO with allow_zero = TRUE. The code is similar to the above, - without the capturing complication. It is written out separately for speed - and cleanliness. */ - - case OP_BRAPOS: - case OP_SBRAPOS: - allow_zero = FALSE; - - POSSESSIVE_NON_CAPTURE: - matched_once = FALSE; - code_offset = (int)(ecode - md->start_code); - save_capture_last = md->capture_last; - - for (;;) - { - if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; - RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, - eptrb, RM48); - if (rrc == MATCH_KETRPOS) - { - offset_top = md->end_offset_top; - eptr = md->end_match_ptr; - ecode = md->start_code + code_offset; - matched_once = TRUE; - mstart = md->start_match_ptr; /* In case \K reset it */ - continue; - } - - /* See comment in the code for capturing groups above about handling - THEN. */ - - if (rrc == MATCH_THEN) - { - next = ecode + GET(ecode,1); - if (md->start_match_ptr < next && - (*ecode == OP_ALT || *next == OP_ALT)) - rrc = MATCH_NOMATCH; - } - - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - ecode += GET(ecode, 1); - if (*ecode != OP_ALT) break; - md->capture_last = save_capture_last; - } - - if (matched_once || allow_zero) - { - ecode += 1 + LINK_SIZE; - break; - } - RRETURN(MATCH_NOMATCH); - - /* Control never reaches here. */ - - /* Conditional group: compilation checked that there are no more than two - branches. If the condition is false, skipping the first branch takes us - past the end of the item if there is only one branch, but that's exactly - what we want. */ - - case OP_COND: - case OP_SCOND: - - /* The variable codelink will be added to ecode when the condition is - false, to get to the second branch. Setting it to the offset to the ALT - or KET, then incrementing ecode achieves this effect. We now have ecode - pointing to the condition or callout. */ - - codelink = GET(ecode, 1); /* Offset to the second branch */ - ecode += 1 + LINK_SIZE; /* From this opcode */ - - /* Because of the way auto-callout works during compile, a callout item is - inserted between OP_COND and an assertion condition. */ - - if (*ecode == OP_CALLOUT) - { - if (PUBL(callout) != NULL) - { - PUBL(callout_block) cb; - cb.version = 2; /* Version 1 of the callout block */ - cb.callout_number = ecode[1]; - cb.offset_vector = md->offset_vector; -#if defined COMPILE_PCRE8 - cb.subject = (PCRE_SPTR)md->start_subject; -#elif defined COMPILE_PCRE16 - cb.subject = (PCRE_SPTR16)md->start_subject; -#elif defined COMPILE_PCRE32 - cb.subject = (PCRE_SPTR32)md->start_subject; -#endif - cb.subject_length = (int)(md->end_subject - md->start_subject); - cb.start_match = (int)(mstart - md->start_subject); - cb.current_position = (int)(eptr - md->start_subject); - cb.pattern_position = GET(ecode, 2); - cb.next_item_length = GET(ecode, 2 + LINK_SIZE); - cb.capture_top = offset_top/2; - cb.capture_last = md->capture_last & CAPLMASK; - /* Internal change requires this for API compatibility. */ - if (cb.capture_last == 0) cb.capture_last = -1; - cb.callout_data = md->callout_data; - cb.mark = md->nomatch_mark; - if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH); - if (rrc < 0) RRETURN(rrc); - } - - /* Advance ecode past the callout, so it now points to the condition. We - must adjust codelink so that the value of ecode+codelink is unchanged. */ - - ecode += PRIV(OP_lengths)[OP_CALLOUT]; - codelink -= PRIV(OP_lengths)[OP_CALLOUT]; - } - - /* Test the various possible conditions */ - - condition = FALSE; - switch(condcode = *ecode) - { - case OP_RREF: /* Numbered group recursion test */ - if (md->recursive != NULL) /* Not recursing => FALSE */ - { - unsigned int recno = GET2(ecode, 1); /* Recursion group number*/ - condition = (recno == RREF_ANY || recno == md->recursive->group_num); - } - break; - - case OP_DNRREF: /* Duplicate named group recursion test */ - if (md->recursive != NULL) - { - int count = GET2(ecode, 1 + IMM2_SIZE); - pcre_uchar *slot = md->name_table + GET2(ecode, 1) * md->name_entry_size; - while (count-- > 0) - { - unsigned int recno = GET2(slot, 0); - condition = recno == md->recursive->group_num; - if (condition) break; - slot += md->name_entry_size; - } - } - break; - - case OP_CREF: /* Numbered group used test */ - offset = GET2(ecode, 1) << 1; /* Doubled ref number */ - condition = offset < offset_top && md->offset_vector[offset] >= 0; - break; - - case OP_DNCREF: /* Duplicate named group used test */ - { - int count = GET2(ecode, 1 + IMM2_SIZE); - pcre_uchar *slot = md->name_table + GET2(ecode, 1) * md->name_entry_size; - while (count-- > 0) - { - offset = GET2(slot, 0) << 1; - condition = offset < offset_top && md->offset_vector[offset] >= 0; - if (condition) break; - slot += md->name_entry_size; - } - } - break; - - case OP_DEF: /* DEFINE - always false */ - break; - - /* The condition is an assertion. Call match() to evaluate it - setting - md->match_function_type to MATCH_CONDASSERT causes it to stop at the end - of an assertion. */ - - default: - md->match_function_type = MATCH_CONDASSERT; - RMATCH(eptr, ecode, offset_top, md, NULL, RM3); - if (rrc == MATCH_MATCH) - { - if (md->end_offset_top > offset_top) - offset_top = md->end_offset_top; /* Captures may have happened */ - condition = TRUE; - - /* Advance ecode past the assertion to the start of the first branch, - but adjust it so that the general choosing code below works. */ - - ecode += GET(ecode, 1); - while (*ecode == OP_ALT) ecode += GET(ecode, 1); - ecode += 1 + LINK_SIZE - PRIV(OP_lengths)[condcode]; - } - - /* PCRE doesn't allow the effect of (*THEN) to escape beyond an - assertion; it is therefore treated as NOMATCH. Any other return is an - error. */ - - else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) - { - RRETURN(rrc); /* Need braces because of following else */ - } - break; - } - - /* Choose branch according to the condition */ - - ecode += condition? PRIV(OP_lengths)[condcode] : codelink; - - /* We are now at the branch that is to be obeyed. As there is only one, we - can use tail recursion to avoid using another stack frame, except when - there is unlimited repeat of a possibly empty group. In the latter case, a - recursive call to match() is always required, unless the second alternative - doesn't exist, in which case we can just plough on. Note that, for - compatibility with Perl, the | in a conditional group is NOT treated as - creating two alternatives. If a THEN is encountered in the branch, it - propagates out to the enclosing alternative (unless nested in a deeper set - of alternatives, of course). */ - - if (condition || ecode[-(1+LINK_SIZE)] == OP_ALT) - { - if (op != OP_SCOND) - { - goto TAIL_RECURSE; - } - - md->match_function_type = MATCH_CBEGROUP; - RMATCH(eptr, ecode, offset_top, md, eptrb, RM49); - RRETURN(rrc); - } - - /* Condition false & no alternative; continue after the group. */ - - else - { - } - break; - - - /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes, - to close any currently open capturing brackets. */ - - case OP_CLOSE: - number = GET2(ecode, 1); /* Must be less than 65536 */ - offset = number << 1; - -#ifdef PCRE_DEBUG - printf("end bracket %d at *ACCEPT", number); - printf("\n"); -#endif - - md->capture_last = (md->capture_last & OVFLMASK) | number; - if (offset >= md->offset_max) md->capture_last |= OVFLBIT; else - { - md->offset_vector[offset] = - md->offset_vector[md->offset_end - number]; - md->offset_vector[offset+1] = (int)(eptr - md->start_subject); - if (offset_top <= offset) offset_top = offset + 2; - } - ecode += 1 + IMM2_SIZE; - break; - - - /* End of the pattern, either real or forced. */ - - case OP_END: - case OP_ACCEPT: - case OP_ASSERT_ACCEPT: - - /* If we have matched an empty string, fail if not in an assertion and not - in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART - is set and we have matched at the start of the subject. In both cases, - backtracking will then try other alternatives, if any. */ - - if (eptr == mstart && op != OP_ASSERT_ACCEPT && - md->recursive == NULL && - (md->notempty || - (md->notempty_atstart && - mstart == md->start_subject + md->start_offset))) - RRETURN(MATCH_NOMATCH); - - /* Otherwise, we have a match. */ - - md->end_match_ptr = eptr; /* Record where we ended */ - md->end_offset_top = offset_top; /* and how many extracts were taken */ - md->start_match_ptr = mstart; /* and the start (\K can modify) */ - - /* For some reason, the macros don't work properly if an expression is - given as the argument to RRETURN when the heap is in use. */ - - rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT; - RRETURN(rrc); - - /* Assertion brackets. Check the alternative branches in turn - the - matching won't pass the KET for an assertion. If any one branch matches, - the assertion is true. Lookbehind assertions have an OP_REVERSE item at the - start of each branch to move the current point backwards, so the code at - this level is identical to the lookahead case. When the assertion is part - of a condition, we want to return immediately afterwards. The caller of - this incarnation of the match() function will have set MATCH_CONDASSERT in - md->match_function type, and one of these opcodes will be the first opcode - that is processed. We use a local variable that is preserved over calls to - match() to remember this case. */ - - case OP_ASSERT: - case OP_ASSERTBACK: - save_mark = md->mark; - if (md->match_function_type == MATCH_CONDASSERT) - { - condassert = TRUE; - md->match_function_type = 0; - } - else condassert = FALSE; - - /* Loop for each branch */ - - do - { - RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4); - - /* A match means that the assertion is true; break out of the loop - that matches its alternatives. */ - - if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) - { - mstart = md->start_match_ptr; /* In case \K reset it */ - break; - } - - /* If not matched, restore the previous mark setting. */ - - md->mark = save_mark; - - /* See comment in the code for capturing groups above about handling - THEN. */ - - if (rrc == MATCH_THEN) - { - next = ecode + GET(ecode,1); - if (md->start_match_ptr < next && - (*ecode == OP_ALT || *next == OP_ALT)) - rrc = MATCH_NOMATCH; - } - - /* Anything other than NOMATCH causes the entire assertion to fail, - passing back the return code. This includes COMMIT, SKIP, PRUNE and an - uncaptured THEN, which means they take their normal effect. This - consistent approach does not always have exactly the same effect as in - Perl. */ - - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - ecode += GET(ecode, 1); - } - while (*ecode == OP_ALT); /* Continue for next alternative */ - - /* If we have tried all the alternative branches, the assertion has - failed. If not, we broke out after a match. */ - - if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH); - - /* If checking an assertion for a condition, return MATCH_MATCH. */ - - if (condassert) RRETURN(MATCH_MATCH); - - /* Continue from after a successful assertion, updating the offsets high - water mark, since extracts may have been taken during the assertion. */ - - do ecode += GET(ecode,1); while (*ecode == OP_ALT); - ecode += 1 + LINK_SIZE; - offset_top = md->end_offset_top; - continue; - - /* Negative assertion: all branches must fail to match for the assertion to - succeed. */ - - case OP_ASSERT_NOT: - case OP_ASSERTBACK_NOT: - save_mark = md->mark; - if (md->match_function_type == MATCH_CONDASSERT) - { - condassert = TRUE; - md->match_function_type = 0; - } - else condassert = FALSE; - - /* Loop for each alternative branch. */ - - do - { - RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5); - md->mark = save_mark; /* Always restore the mark setting */ - - switch(rrc) - { - case MATCH_MATCH: /* A successful match means */ - case MATCH_ACCEPT: /* the assertion has failed. */ - RRETURN(MATCH_NOMATCH); - - case MATCH_NOMATCH: /* Carry on with next branch */ - break; - - /* See comment in the code for capturing groups above about handling - THEN. */ - - case MATCH_THEN: - next = ecode + GET(ecode,1); - if (md->start_match_ptr < next && - (*ecode == OP_ALT || *next == OP_ALT)) - { - rrc = MATCH_NOMATCH; - break; - } - /* Otherwise fall through. */ - - /* COMMIT, SKIP, PRUNE, and an uncaptured THEN cause the whole - assertion to fail to match, without considering any more alternatives. - Failing to match means the assertion is true. This is a consistent - approach, but does not always have the same effect as in Perl. */ - - case MATCH_COMMIT: - case MATCH_SKIP: - case MATCH_SKIP_ARG: - case MATCH_PRUNE: - do ecode += GET(ecode,1); while (*ecode == OP_ALT); - goto NEG_ASSERT_TRUE; /* Break out of alternation loop */ - - /* Anything else is an error */ - - default: - RRETURN(rrc); - } - - /* Continue with next branch */ - - ecode += GET(ecode,1); - } - while (*ecode == OP_ALT); - - /* All branches in the assertion failed to match. */ - - NEG_ASSERT_TRUE: - if (condassert) RRETURN(MATCH_MATCH); /* Condition assertion */ - ecode += 1 + LINK_SIZE; /* Continue with current branch */ - continue; - - /* Move the subject pointer back. This occurs only at the start of - each branch of a lookbehind assertion. If we are too close to the start to - move back, this match function fails. When working with UTF-8 we move - back a number of characters, not bytes. */ - - case OP_REVERSE: -#ifdef SUPPORT_UTF - if (utf) - { - i = GET(ecode, 1); - while (i-- > 0) - { - eptr--; - if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); - BACKCHAR(eptr); - } - } - else -#endif - - /* No UTF-8 support, or not in UTF-8 mode: count is byte count */ - - { - eptr -= GET(ecode, 1); - if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); - } - - /* Save the earliest consulted character, then skip to next op code */ - - if (eptr < md->start_used_ptr) md->start_used_ptr = eptr; - ecode += 1 + LINK_SIZE; - break; - - /* The callout item calls an external function, if one is provided, passing - details of the match so far. This is mainly for debugging, though the - function is able to force a failure. */ - - case OP_CALLOUT: - if (PUBL(callout) != NULL) - { - PUBL(callout_block) cb; - cb.version = 2; /* Version 1 of the callout block */ - cb.callout_number = ecode[1]; - cb.offset_vector = md->offset_vector; -#if defined COMPILE_PCRE8 - cb.subject = (PCRE_SPTR)md->start_subject; -#elif defined COMPILE_PCRE16 - cb.subject = (PCRE_SPTR16)md->start_subject; -#elif defined COMPILE_PCRE32 - cb.subject = (PCRE_SPTR32)md->start_subject; -#endif - cb.subject_length = (int)(md->end_subject - md->start_subject); - cb.start_match = (int)(mstart - md->start_subject); - cb.current_position = (int)(eptr - md->start_subject); - cb.pattern_position = GET(ecode, 2); - cb.next_item_length = GET(ecode, 2 + LINK_SIZE); - cb.capture_top = offset_top/2; - cb.capture_last = md->capture_last & CAPLMASK; - /* Internal change requires this for API compatibility. */ - if (cb.capture_last == 0) cb.capture_last = -1; - cb.callout_data = md->callout_data; - cb.mark = md->nomatch_mark; - if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH); - if (rrc < 0) RRETURN(rrc); - } - ecode += 2 + 2*LINK_SIZE; - break; - - /* Recursion either matches the current regex, or some subexpression. The - offset data is the offset to the starting bracket from the start of the - whole pattern. (This is so that it works from duplicated subpatterns.) - - The state of the capturing groups is preserved over recursion, and - re-instated afterwards. We don't know how many are started and not yet - finished (offset_top records the completed total) so we just have to save - all the potential data. There may be up to 65535 such values, which is too - large to put on the stack, but using malloc for small numbers seems - expensive. As a compromise, the stack is used when there are no more than - REC_STACK_SAVE_MAX values to store; otherwise malloc is used. - - There are also other values that have to be saved. We use a chained - sequence of blocks that actually live on the stack. Thanks to Robin Houston - for the original version of this logic. It has, however, been hacked around - a lot, so he is not to blame for the current way it works. */ - - case OP_RECURSE: - { - recursion_info *ri; - unsigned int recno; - - callpat = md->start_code + GET(ecode, 1); - recno = (callpat == md->start_code)? 0 : - GET2(callpat, 1 + LINK_SIZE); - - /* Check for repeating a recursion without advancing the subject pointer. - This should catch convoluted mutual recursions. (Some simple cases are - caught at compile time.) */ - - for (ri = md->recursive; ri != NULL; ri = ri->prevrec) - if (recno == ri->group_num && eptr == ri->subject_position) - RRETURN(PCRE_ERROR_RECURSELOOP); - - /* Add to "recursing stack" */ - - new_recursive.group_num = recno; - new_recursive.saved_capture_last = md->capture_last; - new_recursive.subject_position = eptr; - new_recursive.prevrec = md->recursive; - md->recursive = &new_recursive; - - /* Where to continue from afterwards */ - - ecode += 1 + LINK_SIZE; - - /* Now save the offset data */ - - new_recursive.saved_max = md->offset_end; - if (new_recursive.saved_max <= REC_STACK_SAVE_MAX) - new_recursive.offset_save = stacksave; - else - { - new_recursive.offset_save = - (int *)(PUBL(malloc))(new_recursive.saved_max * sizeof(int)); - if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY); - } - memcpy(new_recursive.offset_save, md->offset_vector, - new_recursive.saved_max * sizeof(int)); - - /* OK, now we can do the recursion. After processing each alternative, - restore the offset data and the last captured value. If there were nested - recursions, md->recursive might be changed, so reset it before looping. - */ - - DPRINTF(("Recursing into group %d\n", new_recursive.group_num)); - cbegroup = (*callpat >= OP_SBRA); - do - { - if (cbegroup) md->match_function_type = MATCH_CBEGROUP; - RMATCH(eptr, callpat + PRIV(OP_lengths)[*callpat], offset_top, - md, eptrb, RM6); - memcpy(md->offset_vector, new_recursive.offset_save, - new_recursive.saved_max * sizeof(int)); - md->capture_last = new_recursive.saved_capture_last; - md->recursive = new_recursive.prevrec; - if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) - { - DPRINTF(("Recursion matched\n")); - if (new_recursive.offset_save != stacksave) - (PUBL(free))(new_recursive.offset_save); - - /* Set where we got to in the subject, and reset the start in case - it was changed by \K. This *is* propagated back out of a recursion, - for Perl compatibility. */ - - eptr = md->end_match_ptr; - mstart = md->start_match_ptr; - goto RECURSION_MATCHED; /* Exit loop; end processing */ - } - - /* PCRE does not allow THEN, SKIP, PRUNE or COMMIT to escape beyond a - recursion; they cause a NOMATCH for the entire recursion. These codes - are defined in a range that can be tested for. */ - - if (rrc >= MATCH_BACKTRACK_MIN && rrc <= MATCH_BACKTRACK_MAX) - RRETURN(MATCH_NOMATCH); - - /* Any return code other than NOMATCH is an error. */ - - if (rrc != MATCH_NOMATCH) - { - DPRINTF(("Recursion gave error %d\n", rrc)); - if (new_recursive.offset_save != stacksave) - (PUBL(free))(new_recursive.offset_save); - RRETURN(rrc); - } - - md->recursive = &new_recursive; - callpat += GET(callpat, 1); - } - while (*callpat == OP_ALT); - - DPRINTF(("Recursion didn't match\n")); - md->recursive = new_recursive.prevrec; - if (new_recursive.offset_save != stacksave) - (PUBL(free))(new_recursive.offset_save); - RRETURN(MATCH_NOMATCH); - } - - RECURSION_MATCHED: - break; - - /* An alternation is the end of a branch; scan along to find the end of the - bracketed group and go to there. */ - - case OP_ALT: - do ecode += GET(ecode,1); while (*ecode == OP_ALT); - break; - - /* BRAZERO, BRAMINZERO and SKIPZERO occur just before a bracket group, - indicating that it may occur zero times. It may repeat infinitely, or not - at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets - with fixed upper repeat limits are compiled as a number of copies, with the - optional ones preceded by BRAZERO or BRAMINZERO. */ - - case OP_BRAZERO: - next = ecode + 1; - RMATCH(eptr, next, offset_top, md, eptrb, RM10); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - do next += GET(next, 1); while (*next == OP_ALT); - ecode = next + 1 + LINK_SIZE; - break; - - case OP_BRAMINZERO: - next = ecode + 1; - do next += GET(next, 1); while (*next == OP_ALT); - RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - ecode++; - break; - - case OP_SKIPZERO: - next = ecode+1; - do next += GET(next,1); while (*next == OP_ALT); - ecode = next + 1 + LINK_SIZE; - break; - - /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything - here; just jump to the group, with allow_zero set TRUE. */ - - case OP_BRAPOSZERO: - op = *(++ecode); - allow_zero = TRUE; - if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE; - goto POSSESSIVE_NON_CAPTURE; - - /* End of a group, repeated or non-repeating. */ - - case OP_KET: - case OP_KETRMIN: - case OP_KETRMAX: - case OP_KETRPOS: - prev = ecode - GET(ecode, 1); - - /* If this was a group that remembered the subject start, in order to break - infinite repeats of empty string matches, retrieve the subject start from - the chain. Otherwise, set it NULL. */ - - if (*prev >= OP_SBRA || *prev == OP_ONCE) - { - saved_eptr = eptrb->epb_saved_eptr; /* Value at start of group */ - eptrb = eptrb->epb_prev; /* Backup to previous group */ - } - else saved_eptr = NULL; - - /* If we are at the end of an assertion group or a non-capturing atomic - group, stop matching and return MATCH_MATCH, but record the current high - water mark for use by positive assertions. We also need to record the match - start in case it was changed by \K. */ - - if ((*prev >= OP_ASSERT && *prev <= OP_ASSERTBACK_NOT) || - *prev == OP_ONCE_NC) - { - md->end_match_ptr = eptr; /* For ONCE_NC */ - md->end_offset_top = offset_top; - md->start_match_ptr = mstart; - RRETURN(MATCH_MATCH); /* Sets md->mark */ - } - - /* For capturing groups we have to check the group number back at the start - and if necessary complete handling an extraction by setting the offsets and - bumping the high water mark. Whole-pattern recursion is coded as a recurse - into group 0, so it won't be picked up here. Instead, we catch it when the - OP_END is reached. Other recursion is handled here. We just have to record - the current subject position and start match pointer and give a MATCH - return. */ - - if (*prev == OP_CBRA || *prev == OP_SCBRA || - *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS) - { - number = GET2(prev, 1+LINK_SIZE); - offset = number << 1; - -#ifdef PCRE_DEBUG - printf("end bracket %d", number); - printf("\n"); -#endif - - /* Handle a recursively called group. */ - - if (md->recursive != NULL && md->recursive->group_num == number) - { - md->end_match_ptr = eptr; - md->start_match_ptr = mstart; - RRETURN(MATCH_MATCH); - } - - /* Deal with capturing */ - - md->capture_last = (md->capture_last & OVFLMASK) | number; - if (offset >= md->offset_max) md->capture_last |= OVFLBIT; else - { - /* If offset is greater than offset_top, it means that we are - "skipping" a capturing group, and that group's offsets must be marked - unset. In earlier versions of PCRE, all the offsets were unset at the - start of matching, but this doesn't work because atomic groups and - assertions can cause a value to be set that should later be unset. - Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as - part of the atomic group, but this is not on the final matching path, - so must be unset when 2 is set. (If there is no group 2, there is no - problem, because offset_top will then be 2, indicating no capture.) */ - - if (offset > offset_top) - { - register int *iptr = md->offset_vector + offset_top; - register int *iend = md->offset_vector + offset; - while (iptr < iend) *iptr++ = -1; - } - - /* Now make the extraction */ - - md->offset_vector[offset] = - md->offset_vector[md->offset_end - number]; - md->offset_vector[offset+1] = (int)(eptr - md->start_subject); - if (offset_top <= offset) offset_top = offset + 2; - } - } - - /* For an ordinary non-repeating ket, just continue at this level. This - also happens for a repeating ket if no characters were matched in the - group. This is the forcible breaking of infinite loops as implemented in - Perl 5.005. For a non-repeating atomic group that includes captures, - establish a backup point by processing the rest of the pattern at a lower - level. If this results in a NOMATCH return, pass MATCH_ONCE back to the - original OP_ONCE level, thereby bypassing intermediate backup points, but - resetting any captures that happened along the way. */ - - if (*ecode == OP_KET || eptr == saved_eptr) - { - if (*prev == OP_ONCE) - { - RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - md->once_target = prev; /* Level at which to change to MATCH_NOMATCH */ - RRETURN(MATCH_ONCE); - } - ecode += 1 + LINK_SIZE; /* Carry on at this level */ - break; - } - - /* OP_KETRPOS is a possessive repeating ket. Remember the current position, - and return the MATCH_KETRPOS. This makes it possible to do the repeats one - at a time from the outer level, thus saving stack. */ - - if (*ecode == OP_KETRPOS) - { - md->start_match_ptr = mstart; /* In case \K reset it */ - md->end_match_ptr = eptr; - md->end_offset_top = offset_top; - RRETURN(MATCH_KETRPOS); - } - - /* The normal repeating kets try the rest of the pattern or restart from - the preceding bracket, in the appropriate order. In the second case, we can - use tail recursion to avoid using another stack frame, unless we have an - an atomic group or an unlimited repeat of a group that can match an empty - string. */ - - if (*ecode == OP_KETRMIN) - { - RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (*prev == OP_ONCE) - { - RMATCH(eptr, prev, offset_top, md, eptrb, RM8); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - md->once_target = prev; /* Level at which to change to MATCH_NOMATCH */ - RRETURN(MATCH_ONCE); - } - if (*prev >= OP_SBRA) /* Could match an empty string */ - { - RMATCH(eptr, prev, offset_top, md, eptrb, RM50); - RRETURN(rrc); - } - ecode = prev; - goto TAIL_RECURSE; - } - else /* OP_KETRMAX */ - { - RMATCH(eptr, prev, offset_top, md, eptrb, RM13); - if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH; - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (*prev == OP_ONCE) - { - RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - md->once_target = prev; - RRETURN(MATCH_ONCE); - } - ecode += 1 + LINK_SIZE; - goto TAIL_RECURSE; - } - /* Control never gets here */ - - /* Not multiline mode: start of subject assertion, unless notbol. */ - - case OP_CIRC: - if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH); - - /* Start of subject assertion */ - - case OP_SOD: - if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH); - ecode++; - break; - - /* Multiline mode: start of subject unless notbol, or after any newline. */ - - case OP_CIRCM: - if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH); - if (eptr != md->start_subject && - (eptr == md->end_subject || !WAS_NEWLINE(eptr))) - RRETURN(MATCH_NOMATCH); - ecode++; - break; - - /* Start of match assertion */ - - case OP_SOM: - if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH); - ecode++; - break; - - /* Reset the start of match point */ - - case OP_SET_SOM: - mstart = eptr; - ecode++; - break; - - /* Multiline mode: assert before any newline, or before end of subject - unless noteol is set. */ - - case OP_DOLLM: - if (eptr < md->end_subject) - { - if (!IS_NEWLINE(eptr)) - { - if (md->partial != 0 && - eptr + 1 >= md->end_subject && - NLBLOCK->nltype == NLTYPE_FIXED && - NLBLOCK->nllen == 2 && - RAWUCHARTEST(eptr) == NLBLOCK->nl[0]) - { - md->hitend = TRUE; - if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); - } - RRETURN(MATCH_NOMATCH); - } - } - else - { - if (md->noteol) RRETURN(MATCH_NOMATCH); - SCHECK_PARTIAL(); - } - ecode++; - break; - - /* Not multiline mode: assert before a terminating newline or before end of - subject unless noteol is set. */ - - case OP_DOLL: - if (md->noteol) RRETURN(MATCH_NOMATCH); - if (!md->endonly) goto ASSERT_NL_OR_EOS; - - /* ... else fall through for endonly */ - - /* End of subject assertion (\z) */ - - case OP_EOD: - if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH); - SCHECK_PARTIAL(); - ecode++; - break; - - /* End of subject or ending \n assertion (\Z) */ - - case OP_EODN: - ASSERT_NL_OR_EOS: - if (eptr < md->end_subject && - (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) - { - if (md->partial != 0 && - eptr + 1 >= md->end_subject && - NLBLOCK->nltype == NLTYPE_FIXED && - NLBLOCK->nllen == 2 && - RAWUCHARTEST(eptr) == NLBLOCK->nl[0]) - { - md->hitend = TRUE; - if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); - } - RRETURN(MATCH_NOMATCH); - } - - /* Either at end of string or \n before end. */ - - SCHECK_PARTIAL(); - ecode++; - break; - - /* Word boundary assertions */ - - case OP_NOT_WORD_BOUNDARY: - case OP_WORD_BOUNDARY: - { - - /* Find out if the previous and current characters are "word" characters. - It takes a bit more work in UTF-8 mode. Characters > 255 are assumed to - be "non-word" characters. Remember the earliest consulted character for - partial matching. */ - -#ifdef SUPPORT_UTF - if (utf) - { - /* Get status of previous character */ - - if (eptr == md->start_subject) prev_is_word = FALSE; else - { - PCRE_PUCHAR lastptr = eptr - 1; - BACKCHAR(lastptr); - if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr; - GETCHAR(c, lastptr); -#ifdef SUPPORT_UCP - if (md->use_ucp) - { - if (c == '_') prev_is_word = TRUE; else - { - int cat = UCD_CATEGORY(c); - prev_is_word = (cat == ucp_L || cat == ucp_N); - } - } - else -#endif - prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; - } - - /* Get status of next character */ - - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - cur_is_word = FALSE; - } - else - { - GETCHAR(c, eptr); -#ifdef SUPPORT_UCP - if (md->use_ucp) - { - if (c == '_') cur_is_word = TRUE; else - { - int cat = UCD_CATEGORY(c); - cur_is_word = (cat == ucp_L || cat == ucp_N); - } - } - else -#endif - cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; - } - } - else -#endif - - /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for - consistency with the behaviour of \w we do use it in this case. */ - - { - /* Get status of previous character */ - - if (eptr == md->start_subject) prev_is_word = FALSE; else - { - if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1; -#ifdef SUPPORT_UCP - if (md->use_ucp) - { - c = eptr[-1]; - if (c == '_') prev_is_word = TRUE; else - { - int cat = UCD_CATEGORY(c); - prev_is_word = (cat == ucp_L || cat == ucp_N); - } - } - else -#endif - prev_is_word = MAX_255(eptr[-1]) - && ((md->ctypes[eptr[-1]] & ctype_word) != 0); - } - - /* Get status of next character */ - - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - cur_is_word = FALSE; - } - else -#ifdef SUPPORT_UCP - if (md->use_ucp) - { - c = *eptr; - if (c == '_') cur_is_word = TRUE; else - { - int cat = UCD_CATEGORY(c); - cur_is_word = (cat == ucp_L || cat == ucp_N); - } - } - else -#endif - cur_is_word = MAX_255(*eptr) - && ((md->ctypes[*eptr] & ctype_word) != 0); - } - - /* Now see if the situation is what we want */ - - if ((*ecode++ == OP_WORD_BOUNDARY)? - cur_is_word == prev_is_word : cur_is_word != prev_is_word) - RRETURN(MATCH_NOMATCH); - } - break; - - /* Match any single character type except newline; have to take care with - CRLF newlines and partial matching. */ - - case OP_ANY: - if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); - if (md->partial != 0 && - eptr + 1 >= md->end_subject && - NLBLOCK->nltype == NLTYPE_FIXED && - NLBLOCK->nllen == 2 && - RAWUCHARTEST(eptr) == NLBLOCK->nl[0]) - { - md->hitend = TRUE; - if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); - } - - /* Fall through */ - - /* Match any single character whatsoever. */ - - case OP_ALLANY: - if (eptr >= md->end_subject) /* DO NOT merge the eptr++ here; it must */ - { /* not be updated before SCHECK_PARTIAL. */ - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - eptr++; -#ifdef SUPPORT_UTF - if (utf) ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); -#endif - ecode++; - break; - - /* Match a single byte, even in UTF-8 mode. This opcode really does match - any byte, even newline, independent of the setting of PCRE_DOTALL. */ - - case OP_ANYBYTE: - if (eptr >= md->end_subject) /* DO NOT merge the eptr++ here; it must */ - { /* not be updated before SCHECK_PARTIAL. */ - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - eptr++; - ecode++; - break; - - case OP_NOT_DIGIT: - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - if ( -#if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) - c < 256 && -#endif - (md->ctypes[c] & ctype_digit) != 0 - ) - RRETURN(MATCH_NOMATCH); - ecode++; - break; - - case OP_DIGIT: - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - if ( -#if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) - c > 255 || -#endif - (md->ctypes[c] & ctype_digit) == 0 - ) - RRETURN(MATCH_NOMATCH); - ecode++; - break; - - case OP_NOT_WHITESPACE: - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - if ( -#if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) - c < 256 && -#endif - (md->ctypes[c] & ctype_space) != 0 - ) - RRETURN(MATCH_NOMATCH); - ecode++; - break; - - case OP_WHITESPACE: - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - if ( -#if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) - c > 255 || -#endif - (md->ctypes[c] & ctype_space) == 0 - ) - RRETURN(MATCH_NOMATCH); - ecode++; - break; - - case OP_NOT_WORDCHAR: - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - if ( -#if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) - c < 256 && -#endif - (md->ctypes[c] & ctype_word) != 0 - ) - RRETURN(MATCH_NOMATCH); - ecode++; - break; - - case OP_WORDCHAR: - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - if ( -#if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) - c > 255 || -#endif - (md->ctypes[c] & ctype_word) == 0 - ) - RRETURN(MATCH_NOMATCH); - ecode++; - break; - - case OP_ANYNL: - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - switch(c) - { - default: RRETURN(MATCH_NOMATCH); - - case CHAR_CR: - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - } - else if (RAWUCHARTEST(eptr) == CHAR_LF) eptr++; - break; - - case CHAR_LF: - break; - - case CHAR_VT: - case CHAR_FF: - case CHAR_NEL: -#ifndef EBCDIC - case 0x2028: - case 0x2029: -#endif /* Not EBCDIC */ - if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); - break; - } - ecode++; - break; - - case OP_NOT_HSPACE: - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - switch(c) - { - HSPACE_CASES: RRETURN(MATCH_NOMATCH); /* Byte and multibyte cases */ - default: break; - } - ecode++; - break; - - case OP_HSPACE: - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - switch(c) - { - HSPACE_CASES: break; /* Byte and multibyte cases */ - default: RRETURN(MATCH_NOMATCH); - } - ecode++; - break; - - case OP_NOT_VSPACE: - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - switch(c) - { - VSPACE_CASES: RRETURN(MATCH_NOMATCH); - default: break; - } - ecode++; - break; - - case OP_VSPACE: - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - switch(c) - { - VSPACE_CASES: break; - default: RRETURN(MATCH_NOMATCH); - } - ecode++; - break; - -#ifdef SUPPORT_UCP - /* Check the next character by Unicode property. We will get here only - if the support is in the binary; otherwise a compile-time error occurs. */ - - case OP_PROP: - case OP_NOTPROP: - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - { - const pcre_uint32 *cp; - const ucd_record *prop = GET_UCD(c); - - switch(ecode[1]) - { - case PT_ANY: - if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH); - break; - - case PT_LAMP: - if ((prop->chartype == ucp_Lu || - prop->chartype == ucp_Ll || - prop->chartype == ucp_Lt) == (op == OP_NOTPROP)) - RRETURN(MATCH_NOMATCH); - break; - - case PT_GC: - if ((ecode[2] != PRIV(ucp_gentype)[prop->chartype]) == (op == OP_PROP)) - RRETURN(MATCH_NOMATCH); - break; - - case PT_PC: - if ((ecode[2] != prop->chartype) == (op == OP_PROP)) - RRETURN(MATCH_NOMATCH); - break; - - case PT_SC: - if ((ecode[2] != prop->script) == (op == OP_PROP)) - RRETURN(MATCH_NOMATCH); - break; - - /* These are specials */ - - case PT_ALNUM: - if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L || - PRIV(ucp_gentype)[prop->chartype] == ucp_N) == (op == OP_NOTPROP)) - RRETURN(MATCH_NOMATCH); - break; - - /* Perl space used to exclude VT, but from Perl 5.18 it is included, - which means that Perl space and POSIX space are now identical. PCRE - was changed at release 8.34. */ - - case PT_SPACE: /* Perl space */ - case PT_PXSPACE: /* POSIX space */ - switch(c) - { - HSPACE_CASES: - VSPACE_CASES: - if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH); - break; - - default: - if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z) == - (op == OP_NOTPROP)) RRETURN(MATCH_NOMATCH); - break; - } - break; - - case PT_WORD: - if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L || - PRIV(ucp_gentype)[prop->chartype] == ucp_N || - c == CHAR_UNDERSCORE) == (op == OP_NOTPROP)) - RRETURN(MATCH_NOMATCH); - break; - - case PT_CLIST: - cp = PRIV(ucd_caseless_sets) + ecode[2]; - for (;;) - { - if (c < *cp) - { if (op == OP_PROP) { RRETURN(MATCH_NOMATCH); } else break; } - if (c == *cp++) - { if (op == OP_PROP) break; else { RRETURN(MATCH_NOMATCH); } } - } - break; - - case PT_UCNC: - if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT || - c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) || - c >= 0xe000) == (op == OP_NOTPROP)) - RRETURN(MATCH_NOMATCH); - break; - - /* This should never occur */ - - default: - RRETURN(PCRE_ERROR_INTERNAL); - } - - ecode += 3; - } - break; - - /* Match an extended Unicode sequence. We will get here only if the support - is in the binary; otherwise a compile-time error occurs. */ - - case OP_EXTUNI: - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - else - { - int lgb, rgb; - GETCHARINCTEST(c, eptr); - lgb = UCD_GRAPHBREAK(c); - while (eptr < md->end_subject) - { - int len = 1; - if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } - rgb = UCD_GRAPHBREAK(c); - if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break; - lgb = rgb; - eptr += len; - } - } - CHECK_PARTIAL(); - ecode++; - break; -#endif /* SUPPORT_UCP */ - - - /* Match a back reference, possibly repeatedly. Look past the end of the - item to see if there is repeat information following. The code is similar - to that for character classes, but repeated for efficiency. Then obey - similar code to character type repeats - written out again for speed. - However, if the referenced string is the empty string, always treat - it as matched, any number of times (otherwise there could be infinite - loops). If the reference is unset, there are two possibilities: - - (a) In the default, Perl-compatible state, set the length negative; - this ensures that every attempt at a match fails. We can't just fail - here, because of the possibility of quantifiers with zero minima. - - (b) If the JavaScript compatibility flag is set, set the length to zero - so that the back reference matches an empty string. - - Otherwise, set the length to the length of what was matched by the - referenced subpattern. - - The OP_REF and OP_REFI opcodes are used for a reference to a numbered group - or to a non-duplicated named group. For a duplicated named group, OP_DNREF - and OP_DNREFI are used. In this case we must scan the list of groups to - which the name refers, and use the first one that is set. */ - - case OP_DNREF: - case OP_DNREFI: - caseless = op == OP_DNREFI; - { - int count = GET2(ecode, 1+IMM2_SIZE); - pcre_uchar *slot = md->name_table + GET2(ecode, 1) * md->name_entry_size; - ecode += 1 + 2*IMM2_SIZE; - - while (count-- > 0) - { - offset = GET2(slot, 0) << 1; - if (offset < offset_top && md->offset_vector[offset] >= 0) break; - slot += md->name_entry_size; - } - if (count < 0) - length = (md->jscript_compat)? 0 : -1; - else - length = md->offset_vector[offset+1] - md->offset_vector[offset]; - } - goto REF_REPEAT; - - case OP_REF: - case OP_REFI: - caseless = op == OP_REFI; - offset = GET2(ecode, 1) << 1; /* Doubled ref number */ - ecode += 1 + IMM2_SIZE; - if (offset >= offset_top || md->offset_vector[offset] < 0) - length = (md->jscript_compat)? 0 : -1; - else - length = md->offset_vector[offset+1] - md->offset_vector[offset]; - - /* Set up for repetition, or handle the non-repeated case */ - - REF_REPEAT: - switch (*ecode) - { - case OP_CRSTAR: - case OP_CRMINSTAR: - case OP_CRPLUS: - case OP_CRMINPLUS: - case OP_CRQUERY: - case OP_CRMINQUERY: - c = *ecode++ - OP_CRSTAR; - minimize = (c & 1) != 0; - min = rep_min[c]; /* Pick up values from tables; */ - max = rep_max[c]; /* zero for max => infinity */ - if (max == 0) max = INT_MAX; - break; - - case OP_CRRANGE: - case OP_CRMINRANGE: - minimize = (*ecode == OP_CRMINRANGE); - min = GET2(ecode, 1); - max = GET2(ecode, 1 + IMM2_SIZE); - if (max == 0) max = INT_MAX; - ecode += 1 + 2 * IMM2_SIZE; - break; - - default: /* No repeat follows */ - if ((length = match_ref(offset, eptr, length, md, caseless)) < 0) - { - if (length == -2) eptr = md->end_subject; /* Partial match */ - CHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - eptr += length; - continue; /* With the main loop */ - } - - /* Handle repeated back references. If the length of the reference is - zero, just continue with the main loop. If the length is negative, it - means the reference is unset in non-Java-compatible mode. If the minimum is - zero, we can continue at the same level without recursion. For any other - minimum, carrying on will result in NOMATCH. */ - - if (length == 0) continue; - if (length < 0 && min == 0) continue; - - /* First, ensure the minimum number of matches are present. We get back - the length of the reference string explicitly rather than passing the - address of eptr, so that eptr can be a register variable. */ - - for (i = 1; i <= min; i++) - { - int slength; - if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) - { - if (slength == -2) eptr = md->end_subject; /* Partial match */ - CHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - eptr += slength; - } - - /* If min = max, continue at the same level without recursion. - They are not both allowed to be zero. */ - - if (min == max) continue; - - /* If minimizing, keep trying and advancing the pointer */ - - if (minimize) - { - for (fi = min;; fi++) - { - int slength; - RMATCH(eptr, ecode, offset_top, md, eptrb, RM14); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) - { - if (slength == -2) eptr = md->end_subject; /* Partial match */ - CHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - eptr += slength; - } - /* Control never gets here */ - } - - /* If maximizing, find the longest string and work backwards */ - - else - { - pp = eptr; - for (i = min; i < max; i++) - { - int slength; - if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) - { - /* Can't use CHECK_PARTIAL because we don't want to update eptr in - the soft partial matching case. */ - - if (slength == -2 && md->partial != 0 && - md->end_subject > md->start_used_ptr) - { - md->hitend = TRUE; - if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); - } - break; - } - eptr += slength; - } - - while (eptr >= pp) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM15); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - eptr -= length; - } - RRETURN(MATCH_NOMATCH); - } - /* Control never gets here */ - - /* Match a bit-mapped character class, possibly repeatedly. This op code is - used when all the characters in the class have values in the range 0-255, - and either the matching is caseful, or the characters are in the range - 0-127 when UTF-8 processing is enabled. The only difference between - OP_CLASS and OP_NCLASS occurs when a data character outside the range is - encountered. - - First, look past the end of the item to see if there is repeat information - following. Then obey similar code to character type repeats - written out - again for speed. */ - - case OP_NCLASS: - case OP_CLASS: - { - /* The data variable is saved across frames, so the byte map needs to - be stored there. */ -#define BYTE_MAP ((pcre_uint8 *)data) - data = ecode + 1; /* Save for matching */ - ecode += 1 + (32 / sizeof(pcre_uchar)); /* Advance past the item */ - - switch (*ecode) - { - case OP_CRSTAR: - case OP_CRMINSTAR: - case OP_CRPLUS: - case OP_CRMINPLUS: - case OP_CRQUERY: - case OP_CRMINQUERY: - case OP_CRPOSSTAR: - case OP_CRPOSPLUS: - case OP_CRPOSQUERY: - c = *ecode++ - OP_CRSTAR; - if (c < OP_CRPOSSTAR - OP_CRSTAR) minimize = (c & 1) != 0; - else possessive = TRUE; - min = rep_min[c]; /* Pick up values from tables; */ - max = rep_max[c]; /* zero for max => infinity */ - if (max == 0) max = INT_MAX; - break; - - case OP_CRRANGE: - case OP_CRMINRANGE: - case OP_CRPOSRANGE: - minimize = (*ecode == OP_CRMINRANGE); - possessive = (*ecode == OP_CRPOSRANGE); - min = GET2(ecode, 1); - max = GET2(ecode, 1 + IMM2_SIZE); - if (max == 0) max = INT_MAX; - ecode += 1 + 2 * IMM2_SIZE; - break; - - default: /* No repeat follows */ - min = max = 1; - break; - } - - /* First, ensure the minimum number of matches are present. */ - -#ifdef SUPPORT_UTF - if (utf) - { - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINC(c, eptr); - if (c > 255) - { - if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); - } - else - if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); - } - } - else -#endif - /* Not UTF mode */ - { - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - c = *eptr++; -#ifndef COMPILE_PCRE8 - if (c > 255) - { - if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); - } - else -#endif - if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); - } - } - - /* If max == min we can continue with the main loop without the - need to recurse. */ - - if (min == max) continue; - - /* If minimizing, keep testing the rest of the expression and advancing - the pointer while it matches the class. */ - - if (minimize) - { -#ifdef SUPPORT_UTF - if (utf) - { - for (fi = min;; fi++) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM16); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINC(c, eptr); - if (c > 255) - { - if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); - } - else - if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); - } - } - else -#endif - /* Not UTF mode */ - { - for (fi = min;; fi++) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM17); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - c = *eptr++; -#ifndef COMPILE_PCRE8 - if (c > 255) - { - if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); - } - else -#endif - if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); - } - } - /* Control never gets here */ - } - - /* If maximizing, find the longest possible run, then work backwards. */ - - else - { - pp = eptr; - -#ifdef SUPPORT_UTF - if (utf) - { - for (i = min; i < max; i++) - { - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLEN(c, eptr, len); - if (c > 255) - { - if (op == OP_CLASS) break; - } - else - if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break; - eptr += len; - } - - if (possessive) continue; /* No backtracking */ - - for (;;) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM18); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (eptr-- == pp) break; /* Stop if tried at original pos */ - BACKCHAR(eptr); - } - } - else -#endif - /* Not UTF mode */ - { - for (i = min; i < max; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - c = *eptr; -#ifndef COMPILE_PCRE8 - if (c > 255) - { - if (op == OP_CLASS) break; - } - else -#endif - if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break; - eptr++; - } - - if (possessive) continue; /* No backtracking */ - - while (eptr >= pp) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM19); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - eptr--; - } - } - - RRETURN(MATCH_NOMATCH); - } -#undef BYTE_MAP - } - /* Control never gets here */ - - - /* Match an extended character class. In the 8-bit library, this opcode is - encountered only when UTF-8 mode mode is supported. In the 16-bit and - 32-bit libraries, codepoints greater than 255 may be encountered even when - UTF is not supported. */ - -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - case OP_XCLASS: - { - data = ecode + 1 + LINK_SIZE; /* Save for matching */ - ecode += GET(ecode, 1); /* Advance past the item */ - - switch (*ecode) - { - case OP_CRSTAR: - case OP_CRMINSTAR: - case OP_CRPLUS: - case OP_CRMINPLUS: - case OP_CRQUERY: - case OP_CRMINQUERY: - case OP_CRPOSSTAR: - case OP_CRPOSPLUS: - case OP_CRPOSQUERY: - c = *ecode++ - OP_CRSTAR; - if (c < OP_CRPOSSTAR - OP_CRSTAR) minimize = (c & 1) != 0; - else possessive = TRUE; - min = rep_min[c]; /* Pick up values from tables; */ - max = rep_max[c]; /* zero for max => infinity */ - if (max == 0) max = INT_MAX; - break; - - case OP_CRRANGE: - case OP_CRMINRANGE: - case OP_CRPOSRANGE: - minimize = (*ecode == OP_CRMINRANGE); - possessive = (*ecode == OP_CRPOSRANGE); - min = GET2(ecode, 1); - max = GET2(ecode, 1 + IMM2_SIZE); - if (max == 0) max = INT_MAX; - ecode += 1 + 2 * IMM2_SIZE; - break; - - default: /* No repeat follows */ - min = max = 1; - break; - } - - /* First, ensure the minimum number of matches are present. */ - - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH); - } - - /* If max == min we can continue with the main loop without the - need to recurse. */ - - if (min == max) continue; - - /* If minimizing, keep testing the rest of the expression and advancing - the pointer while it matches the class. */ - - if (minimize) - { - for (fi = min;; fi++) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM20); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH); - } - /* Control never gets here */ - } - - /* If maximizing, find the longest possible run, then work backwards. */ - - else - { - pp = eptr; - for (i = min; i < max; i++) - { - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } -#ifdef SUPPORT_UTF - GETCHARLENTEST(c, eptr, len); -#else - c = *eptr; -#endif - if (!PRIV(xclass)(c, data, utf)) break; - eptr += len; - } - - if (possessive) continue; /* No backtracking */ - - for(;;) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM21); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (eptr-- == pp) break; /* Stop if tried at original pos */ -#ifdef SUPPORT_UTF - if (utf) BACKCHAR(eptr); -#endif - } - RRETURN(MATCH_NOMATCH); - } - - /* Control never gets here */ - } -#endif /* End of XCLASS */ - - /* Match a single character, casefully */ - - case OP_CHAR: -#ifdef SUPPORT_UTF - if (utf) - { - length = 1; - ecode++; - GETCHARLEN(fc, ecode, length); - if (length > md->end_subject - eptr) - { - CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ - RRETURN(MATCH_NOMATCH); - } - while (length-- > 0) if (*ecode++ != RAWUCHARINC(eptr)) RRETURN(MATCH_NOMATCH); - } - else -#endif - /* Not UTF mode */ - { - if (md->end_subject - eptr < 1) - { - SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ - RRETURN(MATCH_NOMATCH); - } - if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH); - ecode += 2; - } - break; - - /* Match a single character, caselessly. If we are at the end of the - subject, give up immediately. */ - - case OP_CHARI: - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - -#ifdef SUPPORT_UTF - if (utf) - { - length = 1; - ecode++; - GETCHARLEN(fc, ecode, length); - - /* If the pattern character's value is < 128, we have only one byte, and - we know that its other case must also be one byte long, so we can use the - fast lookup table. We know that there is at least one byte left in the - subject. */ - - if (fc < 128) - { - pcre_uint32 cc = RAWUCHAR(eptr); - if (md->lcc[fc] != TABLE_GET(cc, md->lcc, cc)) RRETURN(MATCH_NOMATCH); - ecode++; - eptr++; - } - - /* Otherwise we must pick up the subject character. Note that we cannot - use the value of "length" to check for sufficient bytes left, because the - other case of the character may have more or fewer bytes. */ - - else - { - pcre_uint32 dc; - GETCHARINC(dc, eptr); - ecode += length; - - /* If we have Unicode property support, we can use it to test the other - case of the character, if there is one. */ - - if (fc != dc) - { -#ifdef SUPPORT_UCP - if (dc != UCD_OTHERCASE(fc)) -#endif - RRETURN(MATCH_NOMATCH); - } - } - } - else -#endif /* SUPPORT_UTF */ - - /* Not UTF mode */ - { - if (TABLE_GET(ecode[1], md->lcc, ecode[1]) - != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH); - eptr++; - ecode += 2; - } - break; - - /* Match a single character repeatedly. */ - - case OP_EXACT: - case OP_EXACTI: - min = max = GET2(ecode, 1); - ecode += 1 + IMM2_SIZE; - goto REPEATCHAR; - - case OP_POSUPTO: - case OP_POSUPTOI: - possessive = TRUE; - /* Fall through */ - - case OP_UPTO: - case OP_UPTOI: - case OP_MINUPTO: - case OP_MINUPTOI: - min = 0; - max = GET2(ecode, 1); - minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI; - ecode += 1 + IMM2_SIZE; - goto REPEATCHAR; - - case OP_POSSTAR: - case OP_POSSTARI: - possessive = TRUE; - min = 0; - max = INT_MAX; - ecode++; - goto REPEATCHAR; - - case OP_POSPLUS: - case OP_POSPLUSI: - possessive = TRUE; - min = 1; - max = INT_MAX; - ecode++; - goto REPEATCHAR; - - case OP_POSQUERY: - case OP_POSQUERYI: - possessive = TRUE; - min = 0; - max = 1; - ecode++; - goto REPEATCHAR; - - case OP_STAR: - case OP_STARI: - case OP_MINSTAR: - case OP_MINSTARI: - case OP_PLUS: - case OP_PLUSI: - case OP_MINPLUS: - case OP_MINPLUSI: - case OP_QUERY: - case OP_QUERYI: - case OP_MINQUERY: - case OP_MINQUERYI: - c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI); - minimize = (c & 1) != 0; - min = rep_min[c]; /* Pick up values from tables; */ - max = rep_max[c]; /* zero for max => infinity */ - if (max == 0) max = INT_MAX; - - /* Common code for all repeated single-character matches. We first check - for the minimum number of characters. If the minimum equals the maximum, we - are done. Otherwise, if minimizing, check the rest of the pattern for a - match; if there isn't one, advance up to the maximum, one character at a - time. - - If maximizing, advance up to the maximum number of matching characters, - until eptr is past the end of the maximum run. If possessive, we are - then done (no backing up). Otherwise, match at this position; anything - other than no match is immediately returned. For nomatch, back up one - character, unless we are matching \R and the last thing matched was - \r\n, in which case, back up two bytes. When we reach the first optional - character position, we can save stack by doing a tail recurse. - - The various UTF/non-UTF and caseful/caseless cases are handled separately, - for speed. */ - - REPEATCHAR: -#ifdef SUPPORT_UTF - if (utf) - { - length = 1; - charptr = ecode; - GETCHARLEN(fc, ecode, length); - ecode += length; - - /* Handle multibyte character matching specially here. There is - support for caseless matching if UCP support is present. */ - - if (length > 1) - { -#ifdef SUPPORT_UCP - pcre_uint32 othercase; - if (op >= OP_STARI && /* Caseless */ - (othercase = UCD_OTHERCASE(fc)) != fc) - oclength = PRIV(ord2utf)(othercase, occhars); - else oclength = 0; -#endif /* SUPPORT_UCP */ - - for (i = 1; i <= min; i++) - { - if (eptr <= md->end_subject - length && - memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length; -#ifdef SUPPORT_UCP - else if (oclength > 0 && - eptr <= md->end_subject - oclength && - memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength; -#endif /* SUPPORT_UCP */ - else - { - CHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - } - - if (min == max) continue; - - if (minimize) - { - for (fi = min;; fi++) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM22); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr <= md->end_subject - length && - memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length; -#ifdef SUPPORT_UCP - else if (oclength > 0 && - eptr <= md->end_subject - oclength && - memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength; -#endif /* SUPPORT_UCP */ - else - { - CHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - } - /* Control never gets here */ - } - - else /* Maximize */ - { - pp = eptr; - for (i = min; i < max; i++) - { - if (eptr <= md->end_subject - length && - memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length; -#ifdef SUPPORT_UCP - else if (oclength > 0 && - eptr <= md->end_subject - oclength && - memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength; -#endif /* SUPPORT_UCP */ - else - { - CHECK_PARTIAL(); - break; - } - } - - if (possessive) continue; /* No backtracking */ - for(;;) - { - if (eptr == pp) goto TAIL_RECURSE; - RMATCH(eptr, ecode, offset_top, md, eptrb, RM23); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); -#ifdef SUPPORT_UCP - eptr--; - BACKCHAR(eptr); -#else /* without SUPPORT_UCP */ - eptr -= length; -#endif /* SUPPORT_UCP */ - } - } - /* Control never gets here */ - } - - /* If the length of a UTF-8 character is 1, we fall through here, and - obey the code as for non-UTF-8 characters below, though in this case the - value of fc will always be < 128. */ - } - else -#endif /* SUPPORT_UTF */ - /* When not in UTF-8 mode, load a single-byte character. */ - fc = *ecode++; - - /* The value of fc at this point is always one character, though we may - or may not be in UTF mode. The code is duplicated for the caseless and - caseful cases, for speed, since matching characters is likely to be quite - common. First, ensure the minimum number of matches are present. If min = - max, continue at the same level without recursing. Otherwise, if - minimizing, keep trying the rest of the expression and advancing one - matching character if failing, up to the maximum. Alternatively, if - maximizing, find the maximum number of characters and work backwards. */ - - DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max, - max, (char *)eptr)); - - if (op >= OP_STARI) /* Caseless */ - { -#ifdef COMPILE_PCRE8 - /* fc must be < 128 if UTF is enabled. */ - foc = md->fcc[fc]; -#else -#ifdef SUPPORT_UTF -#ifdef SUPPORT_UCP - if (utf && fc > 127) - foc = UCD_OTHERCASE(fc); -#else - if (utf && fc > 127) - foc = fc; -#endif /* SUPPORT_UCP */ - else -#endif /* SUPPORT_UTF */ - foc = TABLE_GET(fc, md->fcc, fc); -#endif /* COMPILE_PCRE8 */ - - for (i = 1; i <= min; i++) - { - pcre_uint32 cc; /* Faster than pcre_uchar */ - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - cc = RAWUCHARTEST(eptr); - if (fc != cc && foc != cc) RRETURN(MATCH_NOMATCH); - eptr++; - } - if (min == max) continue; - if (minimize) - { - for (fi = min;; fi++) - { - pcre_uint32 cc; /* Faster than pcre_uchar */ - RMATCH(eptr, ecode, offset_top, md, eptrb, RM24); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - cc = RAWUCHARTEST(eptr); - if (fc != cc && foc != cc) RRETURN(MATCH_NOMATCH); - eptr++; - } - /* Control never gets here */ - } - else /* Maximize */ - { - pp = eptr; - for (i = min; i < max; i++) - { - pcre_uint32 cc; /* Faster than pcre_uchar */ - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - cc = RAWUCHARTEST(eptr); - if (fc != cc && foc != cc) break; - eptr++; - } - if (possessive) continue; /* No backtracking */ - for (;;) - { - if (eptr == pp) goto TAIL_RECURSE; - RMATCH(eptr, ecode, offset_top, md, eptrb, RM25); - eptr--; - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - } - /* Control never gets here */ - } - } - - /* Caseful comparisons (includes all multi-byte characters) */ - - else - { - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - if (fc != RAWUCHARINCTEST(eptr)) RRETURN(MATCH_NOMATCH); - } - - if (min == max) continue; - - if (minimize) - { - for (fi = min;; fi++) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM26); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - if (fc != RAWUCHARINCTEST(eptr)) RRETURN(MATCH_NOMATCH); - } - /* Control never gets here */ - } - else /* Maximize */ - { - pp = eptr; - for (i = min; i < max; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - if (fc != RAWUCHARTEST(eptr)) break; - eptr++; - } - if (possessive) continue; /* No backtracking */ - for (;;) - { - if (eptr == pp) goto TAIL_RECURSE; - RMATCH(eptr, ecode, offset_top, md, eptrb, RM27); - eptr--; - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - } - /* Control never gets here */ - } - } - /* Control never gets here */ - - /* Match a negated single one-byte character. The character we are - checking can be multibyte. */ - - case OP_NOT: - case OP_NOTI: - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } -#ifdef SUPPORT_UTF - if (utf) - { - register pcre_uint32 ch, och; - - ecode++; - GETCHARINC(ch, ecode); - GETCHARINC(c, eptr); - - if (op == OP_NOT) - { - if (ch == c) RRETURN(MATCH_NOMATCH); - } - else - { -#ifdef SUPPORT_UCP - if (ch > 127) - och = UCD_OTHERCASE(ch); -#else - if (ch > 127) - och = ch; -#endif /* SUPPORT_UCP */ - else - och = TABLE_GET(ch, md->fcc, ch); - if (ch == c || och == c) RRETURN(MATCH_NOMATCH); - } - } - else -#endif - { - register pcre_uint32 ch = ecode[1]; - c = *eptr++; - if (ch == c || (op == OP_NOTI && TABLE_GET(ch, md->fcc, ch) == c)) - RRETURN(MATCH_NOMATCH); - ecode += 2; - } - break; - - /* Match a negated single one-byte character repeatedly. This is almost a - repeat of the code for a repeated single character, but I haven't found a - nice way of commoning these up that doesn't require a test of the - positive/negative option for each character match. Maybe that wouldn't add - very much to the time taken, but character matching *is* what this is all - about... */ - - case OP_NOTEXACT: - case OP_NOTEXACTI: - min = max = GET2(ecode, 1); - ecode += 1 + IMM2_SIZE; - goto REPEATNOTCHAR; - - case OP_NOTUPTO: - case OP_NOTUPTOI: - case OP_NOTMINUPTO: - case OP_NOTMINUPTOI: - min = 0; - max = GET2(ecode, 1); - minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI; - ecode += 1 + IMM2_SIZE; - goto REPEATNOTCHAR; - - case OP_NOTPOSSTAR: - case OP_NOTPOSSTARI: - possessive = TRUE; - min = 0; - max = INT_MAX; - ecode++; - goto REPEATNOTCHAR; - - case OP_NOTPOSPLUS: - case OP_NOTPOSPLUSI: - possessive = TRUE; - min = 1; - max = INT_MAX; - ecode++; - goto REPEATNOTCHAR; - - case OP_NOTPOSQUERY: - case OP_NOTPOSQUERYI: - possessive = TRUE; - min = 0; - max = 1; - ecode++; - goto REPEATNOTCHAR; - - case OP_NOTPOSUPTO: - case OP_NOTPOSUPTOI: - possessive = TRUE; - min = 0; - max = GET2(ecode, 1); - ecode += 1 + IMM2_SIZE; - goto REPEATNOTCHAR; - - case OP_NOTSTAR: - case OP_NOTSTARI: - case OP_NOTMINSTAR: - case OP_NOTMINSTARI: - case OP_NOTPLUS: - case OP_NOTPLUSI: - case OP_NOTMINPLUS: - case OP_NOTMINPLUSI: - case OP_NOTQUERY: - case OP_NOTQUERYI: - case OP_NOTMINQUERY: - case OP_NOTMINQUERYI: - c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR); - minimize = (c & 1) != 0; - min = rep_min[c]; /* Pick up values from tables; */ - max = rep_max[c]; /* zero for max => infinity */ - if (max == 0) max = INT_MAX; - - /* Common code for all repeated single-byte matches. */ - - REPEATNOTCHAR: - GETCHARINCTEST(fc, ecode); - - /* The code is duplicated for the caseless and caseful cases, for speed, - since matching characters is likely to be quite common. First, ensure the - minimum number of matches are present. If min = max, continue at the same - level without recursing. Otherwise, if minimizing, keep trying the rest of - the expression and advancing one matching character if failing, up to the - maximum. Alternatively, if maximizing, find the maximum number of - characters and work backwards. */ - - DPRINTF(("negative matching %c{%d,%d} against subject %.*s\n", fc, min, max, - max, (char *)eptr)); - - if (op >= OP_NOTSTARI) /* Caseless */ - { -#ifdef SUPPORT_UTF -#ifdef SUPPORT_UCP - if (utf && fc > 127) - foc = UCD_OTHERCASE(fc); -#else - if (utf && fc > 127) - foc = fc; -#endif /* SUPPORT_UCP */ - else -#endif /* SUPPORT_UTF */ - foc = TABLE_GET(fc, md->fcc, fc); - -#ifdef SUPPORT_UTF - if (utf) - { - register pcre_uint32 d; - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINC(d, eptr); - if (fc == d || (unsigned int)foc == d) RRETURN(MATCH_NOMATCH); - } - } - else -#endif /* SUPPORT_UTF */ - /* Not UTF mode */ - { - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH); - eptr++; - } - } - - if (min == max) continue; - - if (minimize) - { -#ifdef SUPPORT_UTF - if (utf) - { - register pcre_uint32 d; - for (fi = min;; fi++) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM28); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINC(d, eptr); - if (fc == d || (unsigned int)foc == d) RRETURN(MATCH_NOMATCH); - } - } - else -#endif /*SUPPORT_UTF */ - /* Not UTF mode */ - { - for (fi = min;; fi++) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM29); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH); - eptr++; - } - } - /* Control never gets here */ - } - - /* Maximize case */ - - else - { - pp = eptr; - -#ifdef SUPPORT_UTF - if (utf) - { - register pcre_uint32 d; - for (i = min; i < max; i++) - { - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLEN(d, eptr, len); - if (fc == d || (unsigned int)foc == d) break; - eptr += len; - } - if (possessive) continue; /* No backtracking */ - for(;;) - { - if (eptr == pp) goto TAIL_RECURSE; - RMATCH(eptr, ecode, offset_top, md, eptrb, RM30); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - eptr--; - BACKCHAR(eptr); - } - } - else -#endif /* SUPPORT_UTF */ - /* Not UTF mode */ - { - for (i = min; i < max; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - if (fc == *eptr || foc == *eptr) break; - eptr++; - } - if (possessive) continue; /* No backtracking */ - for (;;) - { - if (eptr == pp) goto TAIL_RECURSE; - RMATCH(eptr, ecode, offset_top, md, eptrb, RM31); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - eptr--; - } - } - /* Control never gets here */ - } - } - - /* Caseful comparisons */ - - else - { -#ifdef SUPPORT_UTF - if (utf) - { - register pcre_uint32 d; - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINC(d, eptr); - if (fc == d) RRETURN(MATCH_NOMATCH); - } - } - else -#endif - /* Not UTF mode */ - { - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - if (fc == *eptr++) RRETURN(MATCH_NOMATCH); - } - } - - if (min == max) continue; - - if (minimize) - { -#ifdef SUPPORT_UTF - if (utf) - { - register pcre_uint32 d; - for (fi = min;; fi++) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM32); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINC(d, eptr); - if (fc == d) RRETURN(MATCH_NOMATCH); - } - } - else -#endif - /* Not UTF mode */ - { - for (fi = min;; fi++) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM33); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - if (fc == *eptr++) RRETURN(MATCH_NOMATCH); - } - } - /* Control never gets here */ - } - - /* Maximize case */ - - else - { - pp = eptr; - -#ifdef SUPPORT_UTF - if (utf) - { - register pcre_uint32 d; - for (i = min; i < max; i++) - { - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLEN(d, eptr, len); - if (fc == d) break; - eptr += len; - } - if (possessive) continue; /* No backtracking */ - for(;;) - { - if (eptr == pp) goto TAIL_RECURSE; - RMATCH(eptr, ecode, offset_top, md, eptrb, RM34); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - eptr--; - BACKCHAR(eptr); - } - } - else -#endif - /* Not UTF mode */ - { - for (i = min; i < max; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - if (fc == *eptr) break; - eptr++; - } - if (possessive) continue; /* No backtracking */ - for (;;) - { - if (eptr == pp) goto TAIL_RECURSE; - RMATCH(eptr, ecode, offset_top, md, eptrb, RM35); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - eptr--; - } - } - /* Control never gets here */ - } - } - /* Control never gets here */ - - /* Match a single character type repeatedly; several different opcodes - share code. This is very similar to the code for single characters, but we - repeat it in the interests of efficiency. */ - - case OP_TYPEEXACT: - min = max = GET2(ecode, 1); - minimize = TRUE; - ecode += 1 + IMM2_SIZE; - goto REPEATTYPE; - - case OP_TYPEUPTO: - case OP_TYPEMINUPTO: - min = 0; - max = GET2(ecode, 1); - minimize = *ecode == OP_TYPEMINUPTO; - ecode += 1 + IMM2_SIZE; - goto REPEATTYPE; - - case OP_TYPEPOSSTAR: - possessive = TRUE; - min = 0; - max = INT_MAX; - ecode++; - goto REPEATTYPE; - - case OP_TYPEPOSPLUS: - possessive = TRUE; - min = 1; - max = INT_MAX; - ecode++; - goto REPEATTYPE; - - case OP_TYPEPOSQUERY: - possessive = TRUE; - min = 0; - max = 1; - ecode++; - goto REPEATTYPE; - - case OP_TYPEPOSUPTO: - possessive = TRUE; - min = 0; - max = GET2(ecode, 1); - ecode += 1 + IMM2_SIZE; - goto REPEATTYPE; - - case OP_TYPESTAR: - case OP_TYPEMINSTAR: - case OP_TYPEPLUS: - case OP_TYPEMINPLUS: - case OP_TYPEQUERY: - case OP_TYPEMINQUERY: - c = *ecode++ - OP_TYPESTAR; - minimize = (c & 1) != 0; - min = rep_min[c]; /* Pick up values from tables; */ - max = rep_max[c]; /* zero for max => infinity */ - if (max == 0) max = INT_MAX; - - /* Common code for all repeated single character type matches. Note that - in UTF-8 mode, '.' matches a character of any length, but for the other - character types, the valid characters are all one-byte long. */ - - REPEATTYPE: - ctype = *ecode++; /* Code for the character type */ - -#ifdef SUPPORT_UCP - if (ctype == OP_PROP || ctype == OP_NOTPROP) - { - prop_fail_result = ctype == OP_NOTPROP; - prop_type = *ecode++; - prop_value = *ecode++; - } - else prop_type = -1; -#endif - - /* First, ensure the minimum number of matches are present. Use inline - code for maximizing the speed, and do the type test once at the start - (i.e. keep it out of the loop). Separate the UTF-8 code completely as that - is tidier. Also separate the UCP code, which can be the same for both UTF-8 - and single-bytes. */ - - if (min > 0) - { -#ifdef SUPPORT_UCP - if (prop_type >= 0) - { - switch(prop_type) - { - case PT_ANY: - if (prop_fail_result) RRETURN(MATCH_NOMATCH); - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - } - break; - - case PT_LAMP: - for (i = 1; i <= min; i++) - { - int chartype; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - chartype = UCD_CHARTYPE(c); - if ((chartype == ucp_Lu || - chartype == ucp_Ll || - chartype == ucp_Lt) == prop_fail_result) - RRETURN(MATCH_NOMATCH); - } - break; - - case PT_GC: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) - RRETURN(MATCH_NOMATCH); - } - break; - - case PT_PC: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) - RRETURN(MATCH_NOMATCH); - } - break; - - case PT_SC: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) - RRETURN(MATCH_NOMATCH); - } - break; - - case PT_ALNUM: - for (i = 1; i <= min; i++) - { - int category; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - category = UCD_CATEGORY(c); - if ((category == ucp_L || category == ucp_N) == prop_fail_result) - RRETURN(MATCH_NOMATCH); - } - break; - - /* Perl space used to exclude VT, but from Perl 5.18 it is included, - which means that Perl space and POSIX space are now identical. PCRE - was changed at release 8.34. */ - - case PT_SPACE: /* Perl space */ - case PT_PXSPACE: /* POSIX space */ - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - switch(c) - { - HSPACE_CASES: - VSPACE_CASES: - if (prop_fail_result) RRETURN(MATCH_NOMATCH); - break; - - default: - if ((UCD_CATEGORY(c) == ucp_Z) == prop_fail_result) - RRETURN(MATCH_NOMATCH); - break; - } - } - break; - - case PT_WORD: - for (i = 1; i <= min; i++) - { - int category; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - category = UCD_CATEGORY(c); - if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE) - == prop_fail_result) - RRETURN(MATCH_NOMATCH); - } - break; - - case PT_CLIST: - for (i = 1; i <= min; i++) - { - const pcre_uint32 *cp; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - cp = PRIV(ucd_caseless_sets) + prop_value; - for (;;) - { - if (c < *cp) - { if (prop_fail_result) break; else { RRETURN(MATCH_NOMATCH); } } - if (c == *cp++) - { if (prop_fail_result) { RRETURN(MATCH_NOMATCH); } else break; } - } - } - break; - - case PT_UCNC: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT || - c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) || - c >= 0xe000) == prop_fail_result) - RRETURN(MATCH_NOMATCH); - } - break; - - /* This should not occur */ - - default: - RRETURN(PCRE_ERROR_INTERNAL); - } - } - - /* Match extended Unicode sequences. We will get here only if the - support is in the binary; otherwise a compile-time error occurs. */ - - else if (ctype == OP_EXTUNI) - { - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - else - { - int lgb, rgb; - GETCHARINCTEST(c, eptr); - lgb = UCD_GRAPHBREAK(c); - while (eptr < md->end_subject) - { - int len = 1; - if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } - rgb = UCD_GRAPHBREAK(c); - if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break; - lgb = rgb; - eptr += len; - } - } - CHECK_PARTIAL(); - } - } - - else -#endif /* SUPPORT_UCP */ - -/* Handle all other cases when the coding is UTF-8 */ - -#ifdef SUPPORT_UTF - if (utf) switch(ctype) - { - case OP_ANY: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); - if (md->partial != 0 && - eptr + 1 >= md->end_subject && - NLBLOCK->nltype == NLTYPE_FIXED && - NLBLOCK->nllen == 2 && - RAWUCHAR(eptr) == NLBLOCK->nl[0]) - { - md->hitend = TRUE; - if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); - } - eptr++; - ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); - } - break; - - case OP_ALLANY: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - eptr++; - ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); - } - break; - - case OP_ANYBYTE: - if (eptr > md->end_subject - min) RRETURN(MATCH_NOMATCH); - eptr += min; - break; - - case OP_ANYNL: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINC(c, eptr); - switch(c) - { - default: RRETURN(MATCH_NOMATCH); - - case CHAR_CR: - if (eptr < md->end_subject && RAWUCHAR(eptr) == CHAR_LF) eptr++; - break; - - case CHAR_LF: - break; - - case CHAR_VT: - case CHAR_FF: - case CHAR_NEL: -#ifndef EBCDIC - case 0x2028: - case 0x2029: -#endif /* Not EBCDIC */ - if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); - break; - } - } - break; - - case OP_NOT_HSPACE: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINC(c, eptr); - switch(c) - { - HSPACE_CASES: RRETURN(MATCH_NOMATCH); /* Byte and multibyte cases */ - default: break; - } - } - break; - - case OP_HSPACE: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINC(c, eptr); - switch(c) - { - HSPACE_CASES: break; /* Byte and multibyte cases */ - default: RRETURN(MATCH_NOMATCH); - } - } - break; - - case OP_NOT_VSPACE: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINC(c, eptr); - switch(c) - { - VSPACE_CASES: RRETURN(MATCH_NOMATCH); - default: break; - } - } - break; - - case OP_VSPACE: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINC(c, eptr); - switch(c) - { - VSPACE_CASES: break; - default: RRETURN(MATCH_NOMATCH); - } - } - break; - - case OP_NOT_DIGIT: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINC(c, eptr); - if (c < 128 && (md->ctypes[c] & ctype_digit) != 0) - RRETURN(MATCH_NOMATCH); - } - break; - - case OP_DIGIT: - for (i = 1; i <= min; i++) - { - pcre_uint32 cc; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - cc = RAWUCHAR(eptr); - if (cc >= 128 || (md->ctypes[cc] & ctype_digit) == 0) - RRETURN(MATCH_NOMATCH); - eptr++; - /* No need to skip more bytes - we know it's a 1-byte character */ - } - break; - - case OP_NOT_WHITESPACE: - for (i = 1; i <= min; i++) - { - pcre_uint32 cc; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - cc = RAWUCHAR(eptr); - if (cc < 128 && (md->ctypes[cc] & ctype_space) != 0) - RRETURN(MATCH_NOMATCH); - eptr++; - ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); - } - break; - - case OP_WHITESPACE: - for (i = 1; i <= min; i++) - { - pcre_uint32 cc; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - cc = RAWUCHAR(eptr); - if (cc >= 128 || (md->ctypes[cc] & ctype_space) == 0) - RRETURN(MATCH_NOMATCH); - eptr++; - /* No need to skip more bytes - we know it's a 1-byte character */ - } - break; - - case OP_NOT_WORDCHAR: - for (i = 1; i <= min; i++) - { - pcre_uint32 cc; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - cc = RAWUCHAR(eptr); - if (cc < 128 && (md->ctypes[cc] & ctype_word) != 0) - RRETURN(MATCH_NOMATCH); - eptr++; - ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); - } - break; - - case OP_WORDCHAR: - for (i = 1; i <= min; i++) - { - pcre_uint32 cc; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - cc = RAWUCHAR(eptr); - if (cc >= 128 || (md->ctypes[cc] & ctype_word) == 0) - RRETURN(MATCH_NOMATCH); - eptr++; - /* No need to skip more bytes - we know it's a 1-byte character */ - } - break; - - default: - RRETURN(PCRE_ERROR_INTERNAL); - } /* End switch(ctype) */ - - else -#endif /* SUPPORT_UTF */ - - /* Code for the non-UTF-8 case for minimum matching of operators other - than OP_PROP and OP_NOTPROP. */ - - switch(ctype) - { - case OP_ANY: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); - if (md->partial != 0 && - eptr + 1 >= md->end_subject && - NLBLOCK->nltype == NLTYPE_FIXED && - NLBLOCK->nllen == 2 && - *eptr == NLBLOCK->nl[0]) - { - md->hitend = TRUE; - if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); - } - eptr++; - } - break; - - case OP_ALLANY: - if (eptr > md->end_subject - min) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - eptr += min; - break; - - case OP_ANYBYTE: - if (eptr > md->end_subject - min) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - eptr += min; - break; - - case OP_ANYNL: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - switch(*eptr++) - { - default: RRETURN(MATCH_NOMATCH); - - case CHAR_CR: - if (eptr < md->end_subject && *eptr == CHAR_LF) eptr++; - break; - - case CHAR_LF: - break; - - case CHAR_VT: - case CHAR_FF: - case CHAR_NEL: -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - case 0x2028: - case 0x2029: -#endif - if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); - break; - } - } - break; - - case OP_NOT_HSPACE: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - switch(*eptr++) - { - default: break; - HSPACE_BYTE_CASES: -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - HSPACE_MULTIBYTE_CASES: -#endif - RRETURN(MATCH_NOMATCH); - } - } - break; - - case OP_HSPACE: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - switch(*eptr++) - { - default: RRETURN(MATCH_NOMATCH); - HSPACE_BYTE_CASES: -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - HSPACE_MULTIBYTE_CASES: -#endif - break; - } - } - break; - - case OP_NOT_VSPACE: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - switch(*eptr++) - { - VSPACE_BYTE_CASES: -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - VSPACE_MULTIBYTE_CASES: -#endif - RRETURN(MATCH_NOMATCH); - default: break; - } - } - break; - - case OP_VSPACE: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - switch(*eptr++) - { - default: RRETURN(MATCH_NOMATCH); - VSPACE_BYTE_CASES: -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - VSPACE_MULTIBYTE_CASES: -#endif - break; - } - } - break; - - case OP_NOT_DIGIT: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_digit) != 0) - RRETURN(MATCH_NOMATCH); - eptr++; - } - break; - - case OP_DIGIT: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_digit) == 0) - RRETURN(MATCH_NOMATCH); - eptr++; - } - break; - - case OP_NOT_WHITESPACE: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_space) != 0) - RRETURN(MATCH_NOMATCH); - eptr++; - } - break; - - case OP_WHITESPACE: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_space) == 0) - RRETURN(MATCH_NOMATCH); - eptr++; - } - break; - - case OP_NOT_WORDCHAR: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_word) != 0) - RRETURN(MATCH_NOMATCH); - eptr++; - } - break; - - case OP_WORDCHAR: - for (i = 1; i <= min; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_word) == 0) - RRETURN(MATCH_NOMATCH); - eptr++; - } - break; - - default: - RRETURN(PCRE_ERROR_INTERNAL); - } - } - - /* If min = max, continue at the same level without recursing */ - - if (min == max) continue; - - /* If minimizing, we have to test the rest of the pattern before each - subsequent match. Again, separate the UTF-8 case for speed, and also - separate the UCP cases. */ - - if (minimize) - { -#ifdef SUPPORT_UCP - if (prop_type >= 0) - { - switch(prop_type) - { - case PT_ANY: - for (fi = min;; fi++) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM36); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - if (prop_fail_result) RRETURN(MATCH_NOMATCH); - } - /* Control never gets here */ - - case PT_LAMP: - for (fi = min;; fi++) - { - int chartype; - RMATCH(eptr, ecode, offset_top, md, eptrb, RM37); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - chartype = UCD_CHARTYPE(c); - if ((chartype == ucp_Lu || - chartype == ucp_Ll || - chartype == ucp_Lt) == prop_fail_result) - RRETURN(MATCH_NOMATCH); - } - /* Control never gets here */ - - case PT_GC: - for (fi = min;; fi++) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM38); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) - RRETURN(MATCH_NOMATCH); - } - /* Control never gets here */ - - case PT_PC: - for (fi = min;; fi++) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM39); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) - RRETURN(MATCH_NOMATCH); - } - /* Control never gets here */ - - case PT_SC: - for (fi = min;; fi++) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM40); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) - RRETURN(MATCH_NOMATCH); - } - /* Control never gets here */ - - case PT_ALNUM: - for (fi = min;; fi++) - { - int category; - RMATCH(eptr, ecode, offset_top, md, eptrb, RM59); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - category = UCD_CATEGORY(c); - if ((category == ucp_L || category == ucp_N) == prop_fail_result) - RRETURN(MATCH_NOMATCH); - } - /* Control never gets here */ - - /* Perl space used to exclude VT, but from Perl 5.18 it is included, - which means that Perl space and POSIX space are now identical. PCRE - was changed at release 8.34. */ - - case PT_SPACE: /* Perl space */ - case PT_PXSPACE: /* POSIX space */ - for (fi = min;; fi++) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM61); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - switch(c) - { - HSPACE_CASES: - VSPACE_CASES: - if (prop_fail_result) RRETURN(MATCH_NOMATCH); - break; - - default: - if ((UCD_CATEGORY(c) == ucp_Z) == prop_fail_result) - RRETURN(MATCH_NOMATCH); - break; - } - } - /* Control never gets here */ - - case PT_WORD: - for (fi = min;; fi++) - { - int category; - RMATCH(eptr, ecode, offset_top, md, eptrb, RM62); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - category = UCD_CATEGORY(c); - if ((category == ucp_L || - category == ucp_N || - c == CHAR_UNDERSCORE) - == prop_fail_result) - RRETURN(MATCH_NOMATCH); - } - /* Control never gets here */ - - case PT_CLIST: - for (fi = min;; fi++) - { - const pcre_uint32 *cp; - RMATCH(eptr, ecode, offset_top, md, eptrb, RM67); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - cp = PRIV(ucd_caseless_sets) + prop_value; - for (;;) - { - if (c < *cp) - { if (prop_fail_result) break; else { RRETURN(MATCH_NOMATCH); } } - if (c == *cp++) - { if (prop_fail_result) { RRETURN(MATCH_NOMATCH); } else break; } - } - } - /* Control never gets here */ - - case PT_UCNC: - for (fi = min;; fi++) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM60); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - GETCHARINCTEST(c, eptr); - if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT || - c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) || - c >= 0xe000) == prop_fail_result) - RRETURN(MATCH_NOMATCH); - } - /* Control never gets here */ - - /* This should never occur */ - default: - RRETURN(PCRE_ERROR_INTERNAL); - } - } - - /* Match extended Unicode sequences. We will get here only if the - support is in the binary; otherwise a compile-time error occurs. */ - - else if (ctype == OP_EXTUNI) - { - for (fi = min;; fi++) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM41); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - else - { - int lgb, rgb; - GETCHARINCTEST(c, eptr); - lgb = UCD_GRAPHBREAK(c); - while (eptr < md->end_subject) - { - int len = 1; - if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } - rgb = UCD_GRAPHBREAK(c); - if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break; - lgb = rgb; - eptr += len; - } - } - CHECK_PARTIAL(); - } - } - else -#endif /* SUPPORT_UCP */ - -#ifdef SUPPORT_UTF - if (utf) - { - for (fi = min;; fi++) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM42); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - if (ctype == OP_ANY && IS_NEWLINE(eptr)) - RRETURN(MATCH_NOMATCH); - GETCHARINC(c, eptr); - switch(ctype) - { - case OP_ANY: /* This is the non-NL case */ - if (md->partial != 0 && /* Take care with CRLF partial */ - eptr >= md->end_subject && - NLBLOCK->nltype == NLTYPE_FIXED && - NLBLOCK->nllen == 2 && - c == NLBLOCK->nl[0]) - { - md->hitend = TRUE; - if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); - } - break; - - case OP_ALLANY: - case OP_ANYBYTE: - break; - - case OP_ANYNL: - switch(c) - { - default: RRETURN(MATCH_NOMATCH); - case CHAR_CR: - if (eptr < md->end_subject && RAWUCHAR(eptr) == CHAR_LF) eptr++; - break; - - case CHAR_LF: - break; - - case CHAR_VT: - case CHAR_FF: - case CHAR_NEL: -#ifndef EBCDIC - case 0x2028: - case 0x2029: -#endif /* Not EBCDIC */ - if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); - break; - } - break; - - case OP_NOT_HSPACE: - switch(c) - { - HSPACE_CASES: RRETURN(MATCH_NOMATCH); - default: break; - } - break; - - case OP_HSPACE: - switch(c) - { - HSPACE_CASES: break; - default: RRETURN(MATCH_NOMATCH); - } - break; - - case OP_NOT_VSPACE: - switch(c) - { - VSPACE_CASES: RRETURN(MATCH_NOMATCH); - default: break; - } - break; - - case OP_VSPACE: - switch(c) - { - VSPACE_CASES: break; - default: RRETURN(MATCH_NOMATCH); - } - break; - - case OP_NOT_DIGIT: - if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) - RRETURN(MATCH_NOMATCH); - break; - - case OP_DIGIT: - if (c >= 256 || (md->ctypes[c] & ctype_digit) == 0) - RRETURN(MATCH_NOMATCH); - break; - - case OP_NOT_WHITESPACE: - if (c < 256 && (md->ctypes[c] & ctype_space) != 0) - RRETURN(MATCH_NOMATCH); - break; - - case OP_WHITESPACE: - if (c >= 256 || (md->ctypes[c] & ctype_space) == 0) - RRETURN(MATCH_NOMATCH); - break; - - case OP_NOT_WORDCHAR: - if (c < 256 && (md->ctypes[c] & ctype_word) != 0) - RRETURN(MATCH_NOMATCH); - break; - - case OP_WORDCHAR: - if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) - RRETURN(MATCH_NOMATCH); - break; - - default: - RRETURN(PCRE_ERROR_INTERNAL); - } - } - } - else -#endif - /* Not UTF mode */ - { - for (fi = min;; fi++) - { - RMATCH(eptr, ecode, offset_top, md, eptrb, RM43); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - if (fi >= max) RRETURN(MATCH_NOMATCH); - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - RRETURN(MATCH_NOMATCH); - } - if (ctype == OP_ANY && IS_NEWLINE(eptr)) - RRETURN(MATCH_NOMATCH); - c = *eptr++; - switch(ctype) - { - case OP_ANY: /* This is the non-NL case */ - if (md->partial != 0 && /* Take care with CRLF partial */ - eptr >= md->end_subject && - NLBLOCK->nltype == NLTYPE_FIXED && - NLBLOCK->nllen == 2 && - c == NLBLOCK->nl[0]) - { - md->hitend = TRUE; - if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); - } - break; - - case OP_ALLANY: - case OP_ANYBYTE: - break; - - case OP_ANYNL: - switch(c) - { - default: RRETURN(MATCH_NOMATCH); - case CHAR_CR: - if (eptr < md->end_subject && *eptr == CHAR_LF) eptr++; - break; - - case CHAR_LF: - break; - - case CHAR_VT: - case CHAR_FF: - case CHAR_NEL: -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - case 0x2028: - case 0x2029: -#endif - if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); - break; - } - break; - - case OP_NOT_HSPACE: - switch(c) - { - default: break; - HSPACE_BYTE_CASES: -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - HSPACE_MULTIBYTE_CASES: -#endif - RRETURN(MATCH_NOMATCH); - } - break; - - case OP_HSPACE: - switch(c) - { - default: RRETURN(MATCH_NOMATCH); - HSPACE_BYTE_CASES: -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - HSPACE_MULTIBYTE_CASES: -#endif - break; - } - break; - - case OP_NOT_VSPACE: - switch(c) - { - default: break; - VSPACE_BYTE_CASES: -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - VSPACE_MULTIBYTE_CASES: -#endif - RRETURN(MATCH_NOMATCH); - } - break; - - case OP_VSPACE: - switch(c) - { - default: RRETURN(MATCH_NOMATCH); - VSPACE_BYTE_CASES: -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - VSPACE_MULTIBYTE_CASES: -#endif - break; - } - break; - - case OP_NOT_DIGIT: - if (MAX_255(c) && (md->ctypes[c] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH); - break; - - case OP_DIGIT: - if (!MAX_255(c) || (md->ctypes[c] & ctype_digit) == 0) RRETURN(MATCH_NOMATCH); - break; - - case OP_NOT_WHITESPACE: - if (MAX_255(c) && (md->ctypes[c] & ctype_space) != 0) RRETURN(MATCH_NOMATCH); - break; - - case OP_WHITESPACE: - if (!MAX_255(c) || (md->ctypes[c] & ctype_space) == 0) RRETURN(MATCH_NOMATCH); - break; - - case OP_NOT_WORDCHAR: - if (MAX_255(c) && (md->ctypes[c] & ctype_word) != 0) RRETURN(MATCH_NOMATCH); - break; - - case OP_WORDCHAR: - if (!MAX_255(c) || (md->ctypes[c] & ctype_word) == 0) RRETURN(MATCH_NOMATCH); - break; - - default: - RRETURN(PCRE_ERROR_INTERNAL); - } - } - } - /* Control never gets here */ - } - - /* If maximizing, it is worth using inline code for speed, doing the type - test once at the start (i.e. keep it out of the loop). Again, keep the - UTF-8 and UCP stuff separate. */ - - else - { - pp = eptr; /* Remember where we started */ - -#ifdef SUPPORT_UCP - if (prop_type >= 0) - { - switch(prop_type) - { - case PT_ANY: - for (i = min; i < max; i++) - { - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLENTEST(c, eptr, len); - if (prop_fail_result) break; - eptr+= len; - } - break; - - case PT_LAMP: - for (i = min; i < max; i++) - { - int chartype; - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLENTEST(c, eptr, len); - chartype = UCD_CHARTYPE(c); - if ((chartype == ucp_Lu || - chartype == ucp_Ll || - chartype == ucp_Lt) == prop_fail_result) - break; - eptr+= len; - } - break; - - case PT_GC: - for (i = min; i < max; i++) - { - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLENTEST(c, eptr, len); - if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) break; - eptr+= len; - } - break; - - case PT_PC: - for (i = min; i < max; i++) - { - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLENTEST(c, eptr, len); - if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) break; - eptr+= len; - } - break; - - case PT_SC: - for (i = min; i < max; i++) - { - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLENTEST(c, eptr, len); - if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) break; - eptr+= len; - } - break; - - case PT_ALNUM: - for (i = min; i < max; i++) - { - int category; - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLENTEST(c, eptr, len); - category = UCD_CATEGORY(c); - if ((category == ucp_L || category == ucp_N) == prop_fail_result) - break; - eptr+= len; - } - break; - - /* Perl space used to exclude VT, but from Perl 5.18 it is included, - which means that Perl space and POSIX space are now identical. PCRE - was changed at release 8.34. */ - - case PT_SPACE: /* Perl space */ - case PT_PXSPACE: /* POSIX space */ - for (i = min; i < max; i++) - { - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLENTEST(c, eptr, len); - switch(c) - { - HSPACE_CASES: - VSPACE_CASES: - if (prop_fail_result) goto ENDLOOP99; /* Break the loop */ - break; - - default: - if ((UCD_CATEGORY(c) == ucp_Z) == prop_fail_result) - goto ENDLOOP99; /* Break the loop */ - break; - } - eptr+= len; - } - ENDLOOP99: - break; - - case PT_WORD: - for (i = min; i < max; i++) - { - int category; - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLENTEST(c, eptr, len); - category = UCD_CATEGORY(c); - if ((category == ucp_L || category == ucp_N || - c == CHAR_UNDERSCORE) == prop_fail_result) - break; - eptr+= len; - } - break; - - case PT_CLIST: - for (i = min; i < max; i++) - { - const pcre_uint32 *cp; - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLENTEST(c, eptr, len); - cp = PRIV(ucd_caseless_sets) + prop_value; - for (;;) - { - if (c < *cp) - { if (prop_fail_result) break; else goto GOT_MAX; } - if (c == *cp++) - { if (prop_fail_result) goto GOT_MAX; else break; } - } - eptr += len; - } - GOT_MAX: - break; - - case PT_UCNC: - for (i = min; i < max; i++) - { - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLENTEST(c, eptr, len); - if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT || - c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) || - c >= 0xe000) == prop_fail_result) - break; - eptr += len; - } - break; - - default: - RRETURN(PCRE_ERROR_INTERNAL); - } - - /* eptr is now past the end of the maximum run */ - - if (possessive) continue; /* No backtracking */ - for(;;) - { - if (eptr == pp) goto TAIL_RECURSE; - RMATCH(eptr, ecode, offset_top, md, eptrb, RM44); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - eptr--; - if (utf) BACKCHAR(eptr); - } - } - - /* Match extended Unicode grapheme clusters. We will get here only if the - support is in the binary; otherwise a compile-time error occurs. */ - - else if (ctype == OP_EXTUNI) - { - for (i = min; i < max; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - else - { - int lgb, rgb; - GETCHARINCTEST(c, eptr); - lgb = UCD_GRAPHBREAK(c); - while (eptr < md->end_subject) - { - int len = 1; - if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } - rgb = UCD_GRAPHBREAK(c); - if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break; - lgb = rgb; - eptr += len; - } - } - CHECK_PARTIAL(); - } - - /* eptr is now past the end of the maximum run */ - - if (possessive) continue; /* No backtracking */ - - for(;;) - { - int lgb, rgb; - PCRE_PUCHAR fptr; - - if (eptr == pp) goto TAIL_RECURSE; /* At start of char run */ - RMATCH(eptr, ecode, offset_top, md, eptrb, RM45); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - - /* Backtracking over an extended grapheme cluster involves inspecting - the previous two characters (if present) to see if a break is - permitted between them. */ - - eptr--; - if (!utf) c = *eptr; else - { - BACKCHAR(eptr); - GETCHAR(c, eptr); - } - rgb = UCD_GRAPHBREAK(c); - - for (;;) - { - if (eptr == pp) goto TAIL_RECURSE; /* At start of char run */ - fptr = eptr - 1; - if (!utf) c = *fptr; else - { - BACKCHAR(fptr); - GETCHAR(c, fptr); - } - lgb = UCD_GRAPHBREAK(c); - if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break; - eptr = fptr; - rgb = lgb; - } - } - } - - else -#endif /* SUPPORT_UCP */ - -#ifdef SUPPORT_UTF - if (utf) - { - switch(ctype) - { - case OP_ANY: - if (max < INT_MAX) - { - for (i = min; i < max; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - if (IS_NEWLINE(eptr)) break; - if (md->partial != 0 && /* Take care with CRLF partial */ - eptr + 1 >= md->end_subject && - NLBLOCK->nltype == NLTYPE_FIXED && - NLBLOCK->nllen == 2 && - RAWUCHAR(eptr) == NLBLOCK->nl[0]) - { - md->hitend = TRUE; - if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); - } - eptr++; - ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); - } - } - - /* Handle unlimited UTF-8 repeat */ - - else - { - for (i = min; i < max; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - if (IS_NEWLINE(eptr)) break; - if (md->partial != 0 && /* Take care with CRLF partial */ - eptr + 1 >= md->end_subject && - NLBLOCK->nltype == NLTYPE_FIXED && - NLBLOCK->nllen == 2 && - RAWUCHAR(eptr) == NLBLOCK->nl[0]) - { - md->hitend = TRUE; - if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); - } - eptr++; - ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); - } - } - break; - - case OP_ALLANY: - if (max < INT_MAX) - { - for (i = min; i < max; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - eptr++; - ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); - } - } - else - { - eptr = md->end_subject; /* Unlimited UTF-8 repeat */ - SCHECK_PARTIAL(); - } - break; - - /* The byte case is the same as non-UTF8 */ - - case OP_ANYBYTE: - c = max - min; - if (c > (unsigned int)(md->end_subject - eptr)) - { - eptr = md->end_subject; - SCHECK_PARTIAL(); - } - else eptr += c; - break; - - case OP_ANYNL: - for (i = min; i < max; i++) - { - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLEN(c, eptr, len); - if (c == CHAR_CR) - { - if (++eptr >= md->end_subject) break; - if (RAWUCHAR(eptr) == CHAR_LF) eptr++; - } - else - { - if (c != CHAR_LF && - (md->bsr_anycrlf || - (c != CHAR_VT && c != CHAR_FF && c != CHAR_NEL -#ifndef EBCDIC - && c != 0x2028 && c != 0x2029 -#endif /* Not EBCDIC */ - ))) - break; - eptr += len; - } - } - break; - - case OP_NOT_HSPACE: - case OP_HSPACE: - for (i = min; i < max; i++) - { - BOOL gotspace; - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLEN(c, eptr, len); - switch(c) - { - HSPACE_CASES: gotspace = TRUE; break; - default: gotspace = FALSE; break; - } - if (gotspace == (ctype == OP_NOT_HSPACE)) break; - eptr += len; - } - break; - - case OP_NOT_VSPACE: - case OP_VSPACE: - for (i = min; i < max; i++) - { - BOOL gotspace; - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLEN(c, eptr, len); - switch(c) - { - VSPACE_CASES: gotspace = TRUE; break; - default: gotspace = FALSE; break; - } - if (gotspace == (ctype == OP_NOT_VSPACE)) break; - eptr += len; - } - break; - - case OP_NOT_DIGIT: - for (i = min; i < max; i++) - { - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLEN(c, eptr, len); - if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) break; - eptr+= len; - } - break; - - case OP_DIGIT: - for (i = min; i < max; i++) - { - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLEN(c, eptr, len); - if (c >= 256 ||(md->ctypes[c] & ctype_digit) == 0) break; - eptr+= len; - } - break; - - case OP_NOT_WHITESPACE: - for (i = min; i < max; i++) - { - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLEN(c, eptr, len); - if (c < 256 && (md->ctypes[c] & ctype_space) != 0) break; - eptr+= len; - } - break; - - case OP_WHITESPACE: - for (i = min; i < max; i++) - { - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLEN(c, eptr, len); - if (c >= 256 ||(md->ctypes[c] & ctype_space) == 0) break; - eptr+= len; - } - break; - - case OP_NOT_WORDCHAR: - for (i = min; i < max; i++) - { - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLEN(c, eptr, len); - if (c < 256 && (md->ctypes[c] & ctype_word) != 0) break; - eptr+= len; - } - break; - - case OP_WORDCHAR: - for (i = min; i < max; i++) - { - int len = 1; - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - GETCHARLEN(c, eptr, len); - if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) break; - eptr+= len; - } - break; - - default: - RRETURN(PCRE_ERROR_INTERNAL); - } - - if (possessive) continue; /* No backtracking */ - for(;;) - { - if (eptr == pp) goto TAIL_RECURSE; - RMATCH(eptr, ecode, offset_top, md, eptrb, RM46); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - eptr--; - BACKCHAR(eptr); - if (ctype == OP_ANYNL && eptr > pp && RAWUCHAR(eptr) == CHAR_NL && - RAWUCHAR(eptr - 1) == CHAR_CR) eptr--; - } - } - else -#endif /* SUPPORT_UTF */ - /* Not UTF mode */ - { - switch(ctype) - { - case OP_ANY: - for (i = min; i < max; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - if (IS_NEWLINE(eptr)) break; - if (md->partial != 0 && /* Take care with CRLF partial */ - eptr + 1 >= md->end_subject && - NLBLOCK->nltype == NLTYPE_FIXED && - NLBLOCK->nllen == 2 && - *eptr == NLBLOCK->nl[0]) - { - md->hitend = TRUE; - if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); - } - eptr++; - } - break; - - case OP_ALLANY: - case OP_ANYBYTE: - c = max - min; - if (c > (unsigned int)(md->end_subject - eptr)) - { - eptr = md->end_subject; - SCHECK_PARTIAL(); - } - else eptr += c; - break; - - case OP_ANYNL: - for (i = min; i < max; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - c = *eptr; - if (c == CHAR_CR) - { - if (++eptr >= md->end_subject) break; - if (*eptr == CHAR_LF) eptr++; - } - else - { - if (c != CHAR_LF && (md->bsr_anycrlf || - (c != CHAR_VT && c != CHAR_FF && c != CHAR_NEL -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - && c != 0x2028 && c != 0x2029 -#endif - ))) break; - eptr++; - } - } - break; - - case OP_NOT_HSPACE: - for (i = min; i < max; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - switch(*eptr) - { - default: eptr++; break; - HSPACE_BYTE_CASES: -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - HSPACE_MULTIBYTE_CASES: -#endif - goto ENDLOOP00; - } - } - ENDLOOP00: - break; - - case OP_HSPACE: - for (i = min; i < max; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - switch(*eptr) - { - default: goto ENDLOOP01; - HSPACE_BYTE_CASES: -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - HSPACE_MULTIBYTE_CASES: -#endif - eptr++; break; - } - } - ENDLOOP01: - break; - - case OP_NOT_VSPACE: - for (i = min; i < max; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - switch(*eptr) - { - default: eptr++; break; - VSPACE_BYTE_CASES: -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - VSPACE_MULTIBYTE_CASES: -#endif - goto ENDLOOP02; - } - } - ENDLOOP02: - break; - - case OP_VSPACE: - for (i = min; i < max; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - switch(*eptr) - { - default: goto ENDLOOP03; - VSPACE_BYTE_CASES: -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - VSPACE_MULTIBYTE_CASES: -#endif - eptr++; break; - } - } - ENDLOOP03: - break; - - case OP_NOT_DIGIT: - for (i = min; i < max; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_digit) != 0) break; - eptr++; - } - break; - - case OP_DIGIT: - for (i = min; i < max; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_digit) == 0) break; - eptr++; - } - break; - - case OP_NOT_WHITESPACE: - for (i = min; i < max; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_space) != 0) break; - eptr++; - } - break; - - case OP_WHITESPACE: - for (i = min; i < max; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_space) == 0) break; - eptr++; - } - break; - - case OP_NOT_WORDCHAR: - for (i = min; i < max; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_word) != 0) break; - eptr++; - } - break; - - case OP_WORDCHAR: - for (i = min; i < max; i++) - { - if (eptr >= md->end_subject) - { - SCHECK_PARTIAL(); - break; - } - if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_word) == 0) break; - eptr++; - } - break; - - default: - RRETURN(PCRE_ERROR_INTERNAL); - } - - if (possessive) continue; /* No backtracking */ - for (;;) - { - if (eptr == pp) goto TAIL_RECURSE; - RMATCH(eptr, ecode, offset_top, md, eptrb, RM47); - if (rrc != MATCH_NOMATCH) RRETURN(rrc); - eptr--; - if (ctype == OP_ANYNL && eptr > pp && *eptr == CHAR_LF && - eptr[-1] == CHAR_CR) eptr--; - } - } - - /* Control never gets here */ - } - - /* There's been some horrible disaster. Arrival here can only mean there is - something seriously wrong in the code above or the OP_xxx definitions. */ - - default: - DPRINTF(("Unknown opcode %d\n", *ecode)); - RRETURN(PCRE_ERROR_UNKNOWN_OPCODE); - } - - /* Do not stick any code in here without much thought; it is assumed - that "continue" in the code above comes out to here to repeat the main - loop. */ - - } /* End of main loop */ -/* Control never reaches here */ - - -/* When compiling to use the heap rather than the stack for recursive calls to -match(), the RRETURN() macro jumps here. The number that is saved in -frame->Xwhere indicates which label we actually want to return to. */ - -#ifdef NO_RECURSE -#define LBL(val) case val: goto L_RM##val; -HEAP_RETURN: -switch (frame->Xwhere) - { - LBL( 1) LBL( 2) LBL( 3) LBL( 4) LBL( 5) LBL( 6) LBL( 7) LBL( 8) - LBL( 9) LBL(10) LBL(11) LBL(12) LBL(13) LBL(14) LBL(15) LBL(17) - LBL(19) LBL(24) LBL(25) LBL(26) LBL(27) LBL(29) LBL(31) LBL(33) - LBL(35) LBL(43) LBL(47) LBL(48) LBL(49) LBL(50) LBL(51) LBL(52) - LBL(53) LBL(54) LBL(55) LBL(56) LBL(57) LBL(58) LBL(63) LBL(64) - LBL(65) LBL(66) -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - LBL(20) LBL(21) -#endif -#ifdef SUPPORT_UTF - LBL(16) LBL(18) - LBL(22) LBL(23) LBL(28) LBL(30) - LBL(32) LBL(34) LBL(42) LBL(46) -#ifdef SUPPORT_UCP - LBL(36) LBL(37) LBL(38) LBL(39) LBL(40) LBL(41) LBL(44) LBL(45) - LBL(59) LBL(60) LBL(61) LBL(62) LBL(67) -#endif /* SUPPORT_UCP */ -#endif /* SUPPORT_UTF */ - default: - DPRINTF(("jump error in pcre match: label %d non-existent\n", frame->Xwhere)); - return PCRE_ERROR_INTERNAL; - } -#undef LBL -#endif /* NO_RECURSE */ -} - - -/*************************************************************************** -**************************************************************************** - RECURSION IN THE match() FUNCTION - -Undefine all the macros that were defined above to handle this. */ - -#ifdef NO_RECURSE -#undef eptr -#undef ecode -#undef mstart -#undef offset_top -#undef eptrb -#undef flags - -#undef callpat -#undef charptr -#undef data -#undef next -#undef pp -#undef prev -#undef saved_eptr - -#undef new_recursive - -#undef cur_is_word -#undef condition -#undef prev_is_word - -#undef ctype -#undef length -#undef max -#undef min -#undef number -#undef offset -#undef op -#undef save_capture_last -#undef save_offset1 -#undef save_offset2 -#undef save_offset3 -#undef stacksave - -#undef newptrb - -#endif - -/* These two are defined as macros in both cases */ - -#undef fc -#undef fi - -/*************************************************************************** -***************************************************************************/ - - -#ifdef NO_RECURSE -/************************************************* -* Release allocated heap frames * -*************************************************/ - -/* This function releases all the allocated frames. The base frame is on the -machine stack, and so must not be freed. - -Argument: the address of the base frame -Returns: nothing -*/ - -static void -release_match_heapframes (heapframe *frame_base) -{ -heapframe *nextframe = frame_base->Xnextframe; -while (nextframe != NULL) - { - heapframe *oldframe = nextframe; - nextframe = nextframe->Xnextframe; - (PUBL(stack_free))(oldframe); - } -} -#endif - - -/************************************************* -* Execute a Regular Expression * -*************************************************/ - -/* This function applies a compiled re to a subject string and picks out -portions of the string if it matches. Two elements in the vector are set for -each substring: the offsets to the start and end of the substring. - -Arguments: - argument_re points to the compiled expression - extra_data points to extra data or is NULL - subject points to the subject string - length length of subject string (may contain binary zeros) - start_offset where to start in the subject string - options option bits - offsets points to a vector of ints to be filled in with offsets - offsetcount the number of elements in the vector - -Returns: > 0 => success; value is the number of elements filled in - = 0 => success, but offsets is not big enough - -1 => failed to match - < -1 => some kind of unexpected problem -*/ - -#if defined COMPILE_PCRE8 -PCRE_EXP_DEFN int PCRE_CALL_CONVENTION -pcre_exec(const pcre *argument_re, const pcre_extra *extra_data, - PCRE_SPTR subject, int length, int start_offset, int options, int *offsets, - int offsetcount) -#elif defined COMPILE_PCRE16 -PCRE_EXP_DEFN int PCRE_CALL_CONVENTION -pcre16_exec(const pcre16 *argument_re, const pcre16_extra *extra_data, - PCRE_SPTR16 subject, int length, int start_offset, int options, int *offsets, - int offsetcount) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DEFN int PCRE_CALL_CONVENTION -pcre32_exec(const pcre32 *argument_re, const pcre32_extra *extra_data, - PCRE_SPTR32 subject, int length, int start_offset, int options, int *offsets, - int offsetcount) -#endif -{ -int rc, ocount, arg_offset_max; -int newline; -BOOL using_temporary_offsets = FALSE; -BOOL anchored; -BOOL startline; -BOOL firstline; -BOOL utf; -BOOL has_first_char = FALSE; -BOOL has_req_char = FALSE; -pcre_uchar first_char = 0; -pcre_uchar first_char2 = 0; -pcre_uchar req_char = 0; -pcre_uchar req_char2 = 0; -match_data match_block; -match_data *md = &match_block; -const pcre_uint8 *tables; -const pcre_uint8 *start_bits = NULL; -PCRE_PUCHAR start_match = (PCRE_PUCHAR)subject + start_offset; -PCRE_PUCHAR end_subject; -PCRE_PUCHAR start_partial = NULL; -PCRE_PUCHAR match_partial = NULL; -PCRE_PUCHAR req_char_ptr = start_match - 1; - -const pcre_study_data *study; -const REAL_PCRE *re = (const REAL_PCRE *)argument_re; - -#ifdef NO_RECURSE -heapframe frame_zero; -frame_zero.Xprevframe = NULL; /* Marks the top level */ -frame_zero.Xnextframe = NULL; /* None are allocated yet */ -md->match_frames_base = &frame_zero; -#endif - -/* Check for the special magic call that measures the size of the stack used -per recursive call of match(). Without the funny casting for sizeof, a Windows -compiler gave this error: "unary minus operator applied to unsigned type, -result still unsigned". Hopefully the cast fixes that. */ - -if (re == NULL && extra_data == NULL && subject == NULL && length == -999 && - start_offset == -999) -#ifdef NO_RECURSE - return -((int)sizeof(heapframe)); -#else - return match(NULL, NULL, NULL, 0, NULL, NULL, 0); -#endif - -/* Plausibility checks */ - -if ((options & ~PUBLIC_EXEC_OPTIONS) != 0) return PCRE_ERROR_BADOPTION; -if (re == NULL || subject == NULL || (offsets == NULL && offsetcount > 0)) - return PCRE_ERROR_NULL; -if (offsetcount < 0) return PCRE_ERROR_BADCOUNT; -if (length < 0) return PCRE_ERROR_BADLENGTH; -if (start_offset < 0 || start_offset > length) return PCRE_ERROR_BADOFFSET; - -/* Check that the first field in the block is the magic number. If it is not, -return with PCRE_ERROR_BADMAGIC. However, if the magic number is equal to -REVERSED_MAGIC_NUMBER we return with PCRE_ERROR_BADENDIANNESS, which -means that the pattern is likely compiled with different endianness. */ - -if (re->magic_number != MAGIC_NUMBER) - return re->magic_number == REVERSED_MAGIC_NUMBER? - PCRE_ERROR_BADENDIANNESS:PCRE_ERROR_BADMAGIC; -if ((re->flags & PCRE_MODE) == 0) return PCRE_ERROR_BADMODE; - -/* These two settings are used in the code for checking a UTF-8 string that -follows immediately afterwards. Other values in the md block are used only -during "normal" pcre_exec() processing, not when the JIT support is in use, -so they are set up later. */ - -/* PCRE_UTF16 has the same value as PCRE_UTF8. */ -utf = md->utf = (re->options & PCRE_UTF8) != 0; -md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 : - ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0; - -/* Check a UTF-8 string if required. Pass back the character offset and error -code for an invalid string if a results vector is available. */ - -#ifdef SUPPORT_UTF -if (utf && (options & PCRE_NO_UTF8_CHECK) == 0) - { - int erroroffset; - int errorcode = PRIV(valid_utf)((PCRE_PUCHAR)subject, length, &erroroffset); - if (errorcode != 0) - { - if (offsetcount >= 2) - { - offsets[0] = erroroffset; - offsets[1] = errorcode; - } -#if defined COMPILE_PCRE8 - return (errorcode <= PCRE_UTF8_ERR5 && md->partial > 1)? - PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8; -#elif defined COMPILE_PCRE16 - return (errorcode <= PCRE_UTF16_ERR1 && md->partial > 1)? - PCRE_ERROR_SHORTUTF16 : PCRE_ERROR_BADUTF16; -#elif defined COMPILE_PCRE32 - return PCRE_ERROR_BADUTF32; -#endif - } -#if defined COMPILE_PCRE8 || defined COMPILE_PCRE16 - /* Check that a start_offset points to the start of a UTF character. */ - if (start_offset > 0 && start_offset < length && - NOT_FIRSTCHAR(((PCRE_PUCHAR)subject)[start_offset])) - return PCRE_ERROR_BADUTF8_OFFSET; -#endif - } -#endif - -/* If the pattern was successfully studied with JIT support, run the JIT -executable instead of the rest of this function. Most options must be set at -compile time for the JIT code to be usable. Fallback to the normal code path if -an unsupported flag is set. */ - -#ifdef SUPPORT_JIT -if (extra_data != NULL - && (extra_data->flags & (PCRE_EXTRA_EXECUTABLE_JIT | - PCRE_EXTRA_TABLES)) == PCRE_EXTRA_EXECUTABLE_JIT - && extra_data->executable_jit != NULL - && (options & ~PUBLIC_JIT_EXEC_OPTIONS) == 0) - { - rc = PRIV(jit_exec)(extra_data, (const pcre_uchar *)subject, length, - start_offset, options, offsets, offsetcount); - - /* PCRE_ERROR_NULL means that the selected normal or partial matching - mode is not compiled. In this case we simply fallback to interpreter. */ - - if (rc != PCRE_ERROR_JIT_BADOPTION) return rc; - } -#endif - -/* Carry on with non-JIT matching. This information is for finding all the -numbers associated with a given name, for condition testing. */ - -md->name_table = (pcre_uchar *)re + re->name_table_offset; -md->name_count = re->name_count; -md->name_entry_size = re->name_entry_size; - -/* Fish out the optional data from the extra_data structure, first setting -the default values. */ - -study = NULL; -md->match_limit = MATCH_LIMIT; -md->match_limit_recursion = MATCH_LIMIT_RECURSION; -md->callout_data = NULL; - -/* The table pointer is always in native byte order. */ - -tables = re->tables; - -/* The two limit values override the defaults, whatever their value. */ - -if (extra_data != NULL) - { - register unsigned int flags = extra_data->flags; - if ((flags & PCRE_EXTRA_STUDY_DATA) != 0) - study = (const pcre_study_data *)extra_data->study_data; - if ((flags & PCRE_EXTRA_MATCH_LIMIT) != 0) - md->match_limit = extra_data->match_limit; - if ((flags & PCRE_EXTRA_MATCH_LIMIT_RECURSION) != 0) - md->match_limit_recursion = extra_data->match_limit_recursion; - if ((flags & PCRE_EXTRA_CALLOUT_DATA) != 0) - md->callout_data = extra_data->callout_data; - if ((flags & PCRE_EXTRA_TABLES) != 0) tables = extra_data->tables; - } - -/* Limits in the regex override only if they are smaller. */ - -if ((re->flags & PCRE_MLSET) != 0 && re->limit_match < md->match_limit) - md->match_limit = re->limit_match; - -if ((re->flags & PCRE_RLSET) != 0 && - re->limit_recursion < md->match_limit_recursion) - md->match_limit_recursion = re->limit_recursion; - -/* If the exec call supplied NULL for tables, use the inbuilt ones. This -is a feature that makes it possible to save compiled regex and re-use them -in other programs later. */ - -if (tables == NULL) tables = PRIV(default_tables); - -/* Set up other data */ - -anchored = ((re->options | options) & PCRE_ANCHORED) != 0; -startline = (re->flags & PCRE_STARTLINE) != 0; -firstline = (re->options & PCRE_FIRSTLINE) != 0; - -/* The code starts after the real_pcre block and the capture name table. */ - -md->start_code = (const pcre_uchar *)re + re->name_table_offset + - re->name_count * re->name_entry_size; - -md->start_subject = (PCRE_PUCHAR)subject; -md->start_offset = start_offset; -md->end_subject = md->start_subject + length; -end_subject = md->end_subject; - -md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0; -md->use_ucp = (re->options & PCRE_UCP) != 0; -md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0; -md->ignore_skip_arg = 0; - -/* Some options are unpacked into BOOL variables in the hope that testing -them will be faster than individual option bits. */ - -md->notbol = (options & PCRE_NOTBOL) != 0; -md->noteol = (options & PCRE_NOTEOL) != 0; -md->notempty = (options & PCRE_NOTEMPTY) != 0; -md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0; - -md->hitend = FALSE; -md->mark = md->nomatch_mark = NULL; /* In case never set */ - -md->recursive = NULL; /* No recursion at top level */ -md->hasthen = (re->flags & PCRE_HASTHEN) != 0; - -md->lcc = tables + lcc_offset; -md->fcc = tables + fcc_offset; -md->ctypes = tables + ctypes_offset; - -/* Handle different \R options. */ - -switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) - { - case 0: - if ((re->options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) != 0) - md->bsr_anycrlf = (re->options & PCRE_BSR_ANYCRLF) != 0; - else -#ifdef BSR_ANYCRLF - md->bsr_anycrlf = TRUE; -#else - md->bsr_anycrlf = FALSE; -#endif - break; - - case PCRE_BSR_ANYCRLF: - md->bsr_anycrlf = TRUE; - break; - - case PCRE_BSR_UNICODE: - md->bsr_anycrlf = FALSE; - break; - - default: return PCRE_ERROR_BADNEWLINE; - } - -/* Handle different types of newline. The three bits give eight cases. If -nothing is set at run time, whatever was used at compile time applies. */ - -switch ((((options & PCRE_NEWLINE_BITS) == 0)? re->options : - (pcre_uint32)options) & PCRE_NEWLINE_BITS) - { - case 0: newline = NEWLINE; break; /* Compile-time default */ - case PCRE_NEWLINE_CR: newline = CHAR_CR; break; - case PCRE_NEWLINE_LF: newline = CHAR_NL; break; - case PCRE_NEWLINE_CR+ - PCRE_NEWLINE_LF: newline = (CHAR_CR << 8) | CHAR_NL; break; - case PCRE_NEWLINE_ANY: newline = -1; break; - case PCRE_NEWLINE_ANYCRLF: newline = -2; break; - default: return PCRE_ERROR_BADNEWLINE; - } - -if (newline == -2) - { - md->nltype = NLTYPE_ANYCRLF; - } -else if (newline < 0) - { - md->nltype = NLTYPE_ANY; - } -else - { - md->nltype = NLTYPE_FIXED; - if (newline > 255) - { - md->nllen = 2; - md->nl[0] = (newline >> 8) & 255; - md->nl[1] = newline & 255; - } - else - { - md->nllen = 1; - md->nl[0] = newline; - } - } - -/* Partial matching was originally supported only for a restricted set of -regexes; from release 8.00 there are no restrictions, but the bits are still -defined (though never set). So there's no harm in leaving this code. */ - -if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0) - return PCRE_ERROR_BADPARTIAL; - -/* If the expression has got more back references than the offsets supplied can -hold, we get a temporary chunk of working store to use during the matching. -Otherwise, we can use the vector supplied, rounding down its size to a multiple -of 3. */ - -ocount = offsetcount - (offsetcount % 3); -arg_offset_max = (2*ocount)/3; - -if (re->top_backref > 0 && re->top_backref >= ocount/3) - { - ocount = re->top_backref * 3 + 3; - md->offset_vector = (int *)(PUBL(malloc))(ocount * sizeof(int)); - if (md->offset_vector == NULL) return PCRE_ERROR_NOMEMORY; - using_temporary_offsets = TRUE; - DPRINTF(("Got memory to hold back references\n")); - } -else md->offset_vector = offsets; -md->offset_end = ocount; -md->offset_max = (2*ocount)/3; -md->capture_last = 0; - -/* Reset the working variable associated with each extraction. These should -never be used unless previously set, but they get saved and restored, and so we -initialize them to avoid reading uninitialized locations. Also, unset the -offsets for the matched string. This is really just for tidiness with callouts, -in case they inspect these fields. */ - -if (md->offset_vector != NULL) - { - register int *iptr = md->offset_vector + ocount; - register int *iend = iptr - re->top_bracket; - if (iend < md->offset_vector + 2) iend = md->offset_vector + 2; - while (--iptr >= iend) *iptr = -1; - md->offset_vector[0] = md->offset_vector[1] = -1; - } - -/* Set up the first character to match, if available. The first_char value is -never set for an anchored regular expression, but the anchoring may be forced -at run time, so we have to test for anchoring. The first char may be unset for -an unanchored pattern, of course. If there's no first char and the pattern was -studied, there may be a bitmap of possible first characters. */ - -if (!anchored) - { - if ((re->flags & PCRE_FIRSTSET) != 0) - { - has_first_char = TRUE; - first_char = first_char2 = (pcre_uchar)(re->first_char); - if ((re->flags & PCRE_FCH_CASELESS) != 0) - { - first_char2 = TABLE_GET(first_char, md->fcc, first_char); -#if defined SUPPORT_UCP && !(defined COMPILE_PCRE8) - if (utf && first_char > 127) - first_char2 = UCD_OTHERCASE(first_char); -#endif - } - } - else - if (!startline && study != NULL && - (study->flags & PCRE_STUDY_MAPPED) != 0) - start_bits = study->start_bits; - } - -/* For anchored or unanchored matches, there may be a "last known required -character" set. */ - -if ((re->flags & PCRE_REQCHSET) != 0) - { - has_req_char = TRUE; - req_char = req_char2 = (pcre_uchar)(re->req_char); - if ((re->flags & PCRE_RCH_CASELESS) != 0) - { - req_char2 = TABLE_GET(req_char, md->fcc, req_char); -#if defined SUPPORT_UCP && !(defined COMPILE_PCRE8) - if (utf && req_char > 127) - req_char2 = UCD_OTHERCASE(req_char); -#endif - } - } - - -/* ==========================================================================*/ - -/* Loop for handling unanchored repeated matching attempts; for anchored regexs -the loop runs just once. */ - -for(;;) - { - PCRE_PUCHAR save_end_subject = end_subject; - PCRE_PUCHAR new_start_match; - - /* If firstline is TRUE, the start of the match is constrained to the first - line of a multiline string. That is, the match must be before or at the first - newline. Implement this by temporarily adjusting end_subject so that we stop - scanning at a newline. If the match fails at the newline, later code breaks - this loop. */ - - if (firstline) - { - PCRE_PUCHAR t = start_match; -#ifdef SUPPORT_UTF - if (utf) - { - while (t < md->end_subject && !IS_NEWLINE(t)) - { - t++; - ACROSSCHAR(t < end_subject, *t, t++); - } - } - else -#endif - while (t < md->end_subject && !IS_NEWLINE(t)) t++; - end_subject = t; - } - - /* There are some optimizations that avoid running the match if a known - starting point is not found, or if a known later character is not present. - However, there is an option that disables these, for testing and for ensuring - that all callouts do actually occur. The option can be set in the regex by - (*NO_START_OPT) or passed in match-time options. */ - - if (((options | re->options) & PCRE_NO_START_OPTIMIZE) == 0) - { - /* Advance to a unique first char if there is one. */ - - if (has_first_char) - { - pcre_uchar smc; - - if (first_char != first_char2) - while (start_match < end_subject && - (smc = RAWUCHARTEST(start_match)) != first_char && smc != first_char2) - start_match++; - else - while (start_match < end_subject && RAWUCHARTEST(start_match) != first_char) - start_match++; - } - - /* Or to just after a linebreak for a multiline match */ - - else if (startline) - { - if (start_match > md->start_subject + start_offset) - { -#ifdef SUPPORT_UTF - if (utf) - { - while (start_match < end_subject && !WAS_NEWLINE(start_match)) - { - start_match++; - ACROSSCHAR(start_match < end_subject, *start_match, - start_match++); - } - } - else -#endif - while (start_match < end_subject && !WAS_NEWLINE(start_match)) - start_match++; - - /* If we have just passed a CR and the newline option is ANY or ANYCRLF, - and we are now at a LF, advance the match position by one more character. - */ - - if (start_match[-1] == CHAR_CR && - (md->nltype == NLTYPE_ANY || md->nltype == NLTYPE_ANYCRLF) && - start_match < end_subject && - RAWUCHARTEST(start_match) == CHAR_NL) - start_match++; - } - } - - /* Or to a non-unique first byte after study */ - - else if (start_bits != NULL) - { - while (start_match < end_subject) - { - register pcre_uint32 c = RAWUCHARTEST(start_match); -#ifndef COMPILE_PCRE8 - if (c > 255) c = 255; -#endif - if ((start_bits[c/8] & (1 << (c&7))) == 0) - { - start_match++; -#if defined SUPPORT_UTF && defined COMPILE_PCRE8 - /* In non 8-bit mode, the iteration will stop for - characters > 255 at the beginning or not stop at all. */ - if (utf) - ACROSSCHAR(start_match < end_subject, *start_match, - start_match++); -#endif - } - else break; - } - } - } /* Starting optimizations */ - - /* Restore fudged end_subject */ - - end_subject = save_end_subject; - - /* The following two optimizations are disabled for partial matching or if - disabling is explicitly requested. */ - - if (((options | re->options) & PCRE_NO_START_OPTIMIZE) == 0 && !md->partial) - { - /* If the pattern was studied, a minimum subject length may be set. This is - a lower bound; no actual string of that length may actually match the - pattern. Although the value is, strictly, in characters, we treat it as - bytes to avoid spending too much time in this optimization. */ - - if (study != NULL && (study->flags & PCRE_STUDY_MINLEN) != 0 && - (pcre_uint32)(end_subject - start_match) < study->minlength) - { - rc = MATCH_NOMATCH; - break; - } - - /* If req_char is set, we know that that character must appear in the - subject for the match to succeed. If the first character is set, req_char - must be later in the subject; otherwise the test starts at the match point. - This optimization can save a huge amount of backtracking in patterns with - nested unlimited repeats that aren't going to match. Writing separate code - for cased/caseless versions makes it go faster, as does using an - autoincrement and backing off on a match. - - HOWEVER: when the subject string is very, very long, searching to its end - can take a long time, and give bad performance on quite ordinary patterns. - This showed up when somebody was matching something like /^\d+C/ on a - 32-megabyte string... so we don't do this when the string is sufficiently - long. */ - - if (has_req_char && end_subject - start_match < REQ_BYTE_MAX) - { - register PCRE_PUCHAR p = start_match + (has_first_char? 1:0); - - /* We don't need to repeat the search if we haven't yet reached the - place we found it at last time. */ - - if (p > req_char_ptr) - { - if (req_char != req_char2) - { - while (p < end_subject) - { - register pcre_uint32 pp = RAWUCHARINCTEST(p); - if (pp == req_char || pp == req_char2) { p--; break; } - } - } - else - { - while (p < end_subject) - { - if (RAWUCHARINCTEST(p) == req_char) { p--; break; } - } - } - - /* If we can't find the required character, break the matching loop, - forcing a match failure. */ - - if (p >= end_subject) - { - rc = MATCH_NOMATCH; - break; - } - - /* If we have found the required character, save the point where we - found it, so that we don't search again next time round the loop if - the start hasn't passed this character yet. */ - - req_char_ptr = p; - } - } - } - -#ifdef PCRE_DEBUG /* Sigh. Some compilers never learn. */ - printf(">>>> Match against: "); - pchars(start_match, end_subject - start_match, TRUE, md); - printf("\n"); -#endif - - /* OK, we can now run the match. If "hitend" is set afterwards, remember the - first starting point for which a partial match was found. */ - - md->start_match_ptr = start_match; - md->start_used_ptr = start_match; - md->match_call_count = 0; - md->match_function_type = 0; - md->end_offset_top = 0; - md->skip_arg_count = 0; - rc = match(start_match, md->start_code, start_match, 2, md, NULL, 0); - if (md->hitend && start_partial == NULL) - { - start_partial = md->start_used_ptr; - match_partial = start_match; - } - - switch(rc) - { - /* If MATCH_SKIP_ARG reaches this level it means that a MARK that matched - the SKIP's arg was not found. In this circumstance, Perl ignores the SKIP - entirely. The only way we can do that is to re-do the match at the same - point, with a flag to force SKIP with an argument to be ignored. Just - treating this case as NOMATCH does not work because it does not check other - alternatives in patterns such as A(*SKIP:A)B|AC when the subject is AC. */ - - case MATCH_SKIP_ARG: - new_start_match = start_match; - md->ignore_skip_arg = md->skip_arg_count; - break; - - /* SKIP passes back the next starting point explicitly, but if it is no - greater than the match we have just done, treat it as NOMATCH. */ - - case MATCH_SKIP: - if (md->start_match_ptr > start_match) - { - new_start_match = md->start_match_ptr; - break; - } - /* Fall through */ - - /* NOMATCH and PRUNE advance by one character. THEN at this level acts - exactly like PRUNE. Unset ignore SKIP-with-argument. */ - - case MATCH_NOMATCH: - case MATCH_PRUNE: - case MATCH_THEN: - md->ignore_skip_arg = 0; - new_start_match = start_match + 1; -#ifdef SUPPORT_UTF - if (utf) - ACROSSCHAR(new_start_match < end_subject, *new_start_match, - new_start_match++); -#endif - break; - - /* COMMIT disables the bumpalong, but otherwise behaves as NOMATCH. */ - - case MATCH_COMMIT: - rc = MATCH_NOMATCH; - goto ENDLOOP; - - /* Any other return is either a match, or some kind of error. */ - - default: - goto ENDLOOP; - } - - /* Control reaches here for the various types of "no match at this point" - result. Reset the code to MATCH_NOMATCH for subsequent checking. */ - - rc = MATCH_NOMATCH; - - /* If PCRE_FIRSTLINE is set, the match must happen before or at the first - newline in the subject (though it may continue over the newline). Therefore, - if we have just failed to match, starting at a newline, do not continue. */ - - if (firstline && IS_NEWLINE(start_match)) break; - - /* Advance to new matching position */ - - start_match = new_start_match; - - /* Break the loop if the pattern is anchored or if we have passed the end of - the subject. */ - - if (anchored || start_match > end_subject) break; - - /* If we have just passed a CR and we are now at a LF, and the pattern does - not contain any explicit matches for \r or \n, and the newline option is CRLF - or ANY or ANYCRLF, advance the match position by one more character. In - normal matching start_match will aways be greater than the first position at - this stage, but a failed *SKIP can cause a return at the same point, which is - why the first test exists. */ - - if (start_match > (PCRE_PUCHAR)subject + start_offset && - start_match[-1] == CHAR_CR && - start_match < end_subject && - *start_match == CHAR_NL && - (re->flags & PCRE_HASCRORLF) == 0 && - (md->nltype == NLTYPE_ANY || - md->nltype == NLTYPE_ANYCRLF || - md->nllen == 2)) - start_match++; - - md->mark = NULL; /* Reset for start of next match attempt */ - } /* End of for(;;) "bumpalong" loop */ - -/* ==========================================================================*/ - -/* We reach here when rc is not MATCH_NOMATCH, or if one of the stopping -conditions is true: - -(1) The pattern is anchored or the match was failed by (*COMMIT); - -(2) We are past the end of the subject; - -(3) PCRE_FIRSTLINE is set and we have failed to match at a newline, because - this option requests that a match occur at or before the first newline in - the subject. - -When we have a match and the offset vector is big enough to deal with any -backreferences, captured substring offsets will already be set up. In the case -where we had to get some local store to hold offsets for backreference -processing, copy those that we can. In this case there need not be overflow if -certain parts of the pattern were not used, even though there are more -capturing parentheses than vector slots. */ - -ENDLOOP: - -if (rc == MATCH_MATCH || rc == MATCH_ACCEPT) - { - if (using_temporary_offsets) - { - if (arg_offset_max >= 4) - { - memcpy(offsets + 2, md->offset_vector + 2, - (arg_offset_max - 2) * sizeof(int)); - DPRINTF(("Copied offsets from temporary memory\n")); - } - if (md->end_offset_top > arg_offset_max) md->capture_last |= OVFLBIT; - DPRINTF(("Freeing temporary memory\n")); - (PUBL(free))(md->offset_vector); - } - - /* Set the return code to the number of captured strings, or 0 if there were - too many to fit into the vector. */ - - rc = ((md->capture_last & OVFLBIT) != 0 && - md->end_offset_top >= arg_offset_max)? - 0 : md->end_offset_top/2; - - /* If there is space in the offset vector, set any unused pairs at the end of - the pattern to -1 for backwards compatibility. It is documented that this - happens. In earlier versions, the whole set of potential capturing offsets - was set to -1 each time round the loop, but this is handled differently now. - "Gaps" are set to -1 dynamically instead (this fixes a bug). Thus, it is only - those at the end that need unsetting here. We can't just unset them all at - the start of the whole thing because they may get set in one branch that is - not the final matching branch. */ - - if (md->end_offset_top/2 <= re->top_bracket && offsets != NULL) - { - register int *iptr, *iend; - int resetcount = 2 + re->top_bracket * 2; - if (resetcount > offsetcount) resetcount = offsetcount; - iptr = offsets + md->end_offset_top; - iend = offsets + resetcount; - while (iptr < iend) *iptr++ = -1; - } - - /* If there is space, set up the whole thing as substring 0. The value of - md->start_match_ptr might be modified if \K was encountered on the success - matching path. */ - - if (offsetcount < 2) rc = 0; else - { - offsets[0] = (int)(md->start_match_ptr - md->start_subject); - offsets[1] = (int)(md->end_match_ptr - md->start_subject); - } - - /* Return MARK data if requested */ - - if (extra_data != NULL && (extra_data->flags & PCRE_EXTRA_MARK) != 0) - *(extra_data->mark) = (pcre_uchar *)md->mark; - DPRINTF((">>>> returning %d\n", rc)); -#ifdef NO_RECURSE - release_match_heapframes(&frame_zero); -#endif - return rc; - } - -/* Control gets here if there has been an error, or if the overall match -attempt has failed at all permitted starting positions. */ - -if (using_temporary_offsets) - { - DPRINTF(("Freeing temporary memory\n")); - (PUBL(free))(md->offset_vector); - } - -/* For anything other than nomatch or partial match, just return the code. */ - -if (rc != MATCH_NOMATCH && rc != PCRE_ERROR_PARTIAL) - { - DPRINTF((">>>> error: returning %d\n", rc)); -#ifdef NO_RECURSE - release_match_heapframes(&frame_zero); -#endif - return rc; - } - -/* Handle partial matches - disable any mark data */ - -if (match_partial != NULL) - { - DPRINTF((">>>> returning PCRE_ERROR_PARTIAL\n")); - md->mark = NULL; - if (offsetcount > 1) - { - offsets[0] = (int)(start_partial - (PCRE_PUCHAR)subject); - offsets[1] = (int)(end_subject - (PCRE_PUCHAR)subject); - if (offsetcount > 2) - offsets[2] = (int)(match_partial - (PCRE_PUCHAR)subject); + Copyright (c) 1997-2012 University of Cambridge + +----------------------------------------------------------------------------- +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. + + * Neither the name of the University of Cambridge 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 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. +----------------------------------------------------------------------------- +*/ + +/* This module contains pcre_exec(), the externally visible function that does +pattern matching using an NFA algorithm, trying to mimic Perl as closely as +possible. There are also some static supporting functions. */ + +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + +#define NLBLOCK md /* Block containing newline information */ +#define PSSTART start_subject /* Field containing processed string start */ +#define PSEND end_subject /* Field containing processed string end */ + +#include "pcre_internal.h" + +/* Undefine some potentially clashing cpp symbols */ + +#undef min +#undef max + +/* Values for setting in md->match_function_type to indicate two special types +of call to match(). We do it this way to save on using another stack variable, +as stack usage is to be discouraged. */ + +#define MATCH_CONDASSERT 1 /* Called to check a condition assertion */ +#define MATCH_CBEGROUP 2 /* Could-be-empty unlimited repeat group */ + +/* Non-error returns from the match() function. Error returns are externally +defined PCRE_ERROR_xxx codes, which are all negative. */ + +#define MATCH_MATCH 1 +#define MATCH_NOMATCH 0 + +/* Special internal returns from the match() function. Make them sufficiently +negative to avoid the external error codes. */ + +#define MATCH_ACCEPT (-999) +#define MATCH_COMMIT (-998) +#define MATCH_KETRPOS (-997) +#define MATCH_ONCE (-996) +#define MATCH_PRUNE (-995) +#define MATCH_SKIP (-994) +#define MATCH_SKIP_ARG (-993) +#define MATCH_THEN (-992) + +/* Maximum number of ints of offset to save on the stack for recursive calls. +If the offset vector is bigger, malloc is used. This should be a multiple of 3, +because the offset vector is always a multiple of 3 long. */ + +#define REC_STACK_SAVE_MAX 30 + +/* Min and max values for the common repeats; for the maxima, 0 => infinity */ + +static const char rep_min[] = { 0, 0, 1, 1, 0, 0 }; +static const char rep_max[] = { 0, 0, 0, 0, 1, 1 }; + + + +#ifdef PCRE_DEBUG +/************************************************* +* Debugging function to print chars * +*************************************************/ + +/* Print a sequence of chars in printable format, stopping at the end of the +subject if the requested. + +Arguments: + p points to characters + length number to print + is_subject TRUE if printing from within md->start_subject + md pointer to matching data block, if is_subject is TRUE + +Returns: nothing +*/ + +static void +pchars(const pcre_uchar *p, int length, BOOL is_subject, match_data *md) +{ +unsigned int c; +if (is_subject && length > md->end_subject - p) length = md->end_subject - p; +while (length-- > 0) + if (isprint(c = *(p++))) printf("%c", c); else printf("\\x%02x", c); +} +#endif + + + +/************************************************* +* Match a back-reference * +*************************************************/ + +/* Normally, if a back reference hasn't been set, the length that is passed is +negative, so the match always fails. However, in JavaScript compatibility mode, +the length passed is zero. Note that in caseless UTF-8 mode, the number of +subject bytes matched may be different to the number of reference bytes. + +Arguments: + offset index into the offset vector + eptr pointer into the subject + length length of reference to be matched (number of bytes) + md points to match data block + caseless TRUE if caseless + +Returns: >= 0 the number of subject bytes matched + -1 no match + -2 partial match; always given if at end subject +*/ + +static int +match_ref(int offset, register PCRE_PUCHAR eptr, int length, match_data *md, + BOOL caseless) +{ +PCRE_PUCHAR eptr_start = eptr; +register PCRE_PUCHAR p = md->start_subject + md->offset_vector[offset]; + +#ifdef PCRE_DEBUG +if (eptr >= md->end_subject) + printf("matching subject "); +else + { + printf("matching subject "); + pchars(eptr, length, TRUE, md); + } +printf(" against backref "); +pchars(p, length, FALSE, md); +printf("\n"); +#endif + +/* Always fail if reference not set (and not JavaScript compatible - in that +case the length is passed as zero). */ + +if (length < 0) return -1; + +/* Separate the caseless case for speed. In UTF-8 mode we can only do this +properly if Unicode properties are supported. Otherwise, we can check only +ASCII characters. */ + +if (caseless) + { +#ifdef SUPPORT_UTF +#ifdef SUPPORT_UCP + if (md->utf) + { + /* Match characters up to the end of the reference. NOTE: the number of + bytes matched may differ, because there are some characters whose upper and + lower case versions code as different numbers of bytes. For example, U+023A + (2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8); + a sequence of 3 of the former uses 6 bytes, as does a sequence of two of + the latter. It is important, therefore, to check the length along the + reference, not along the subject (earlier code did this wrong). */ + + PCRE_PUCHAR endptr = p + length; + while (p < endptr) + { + int c, d; + if (eptr >= md->end_subject) return -2; /* Partial match */ + GETCHARINC(c, eptr); + GETCHARINC(d, p); + if (c != d && c != UCD_OTHERCASE(d)) return -1; + } + } + else +#endif +#endif + + /* The same code works when not in UTF-8 mode and in UTF-8 mode when there + is no UCP support. */ + { + while (length-- > 0) + { + if (eptr >= md->end_subject) return -2; /* Partial match */ + if (TABLE_GET(*p, md->lcc, *p) != TABLE_GET(*eptr, md->lcc, *eptr)) return -1; + p++; + eptr++; + } + } + } + +/* In the caseful case, we can just compare the bytes, whether or not we +are in UTF-8 mode. */ + +else + { + while (length-- > 0) + { + if (eptr >= md->end_subject) return -2; /* Partial match */ + if (*p++ != *eptr++) return -1; + } + } + +return (int)(eptr - eptr_start); +} + + + +/*************************************************************************** +**************************************************************************** + RECURSION IN THE match() FUNCTION + +The match() function is highly recursive, though not every recursive call +increases the recursive depth. Nevertheless, some regular expressions can cause +it to recurse to a great depth. I was writing for Unix, so I just let it call +itself recursively. This uses the stack for saving everything that has to be +saved for a recursive call. On Unix, the stack can be large, and this works +fine. + +It turns out that on some non-Unix-like systems there are problems with +programs that use a lot of stack. (This despite the fact that every last chip +has oodles of memory these days, and techniques for extending the stack have +been known for decades.) So.... + +There is a fudge, triggered by defining NO_RECURSE, which avoids recursive +calls by keeping local variables that need to be preserved in blocks of memory +obtained from malloc() instead instead of on the stack. Macros are used to +achieve this so that the actual code doesn't look very different to what it +always used to. + +The original heap-recursive code used longjmp(). However, it seems that this +can be very slow on some operating systems. Following a suggestion from Stan +Switzer, the use of longjmp() has been abolished, at the cost of having to +provide a unique number for each call to RMATCH. There is no way of generating +a sequence of numbers at compile time in C. I have given them names, to make +them stand out more clearly. + +Crude tests on x86 Linux show a small speedup of around 5-8%. However, on +FreeBSD, avoiding longjmp() more than halves the time taken to run the standard +tests. Furthermore, not using longjmp() means that local dynamic variables +don't have indeterminate values; this has meant that the frame size can be +reduced because the result can be "passed back" by straight setting of the +variable instead of being passed in the frame. +**************************************************************************** +***************************************************************************/ + +/* Numbers for RMATCH calls. When this list is changed, the code at HEAP_RETURN +below must be updated in sync. */ + +enum { RM1=1, RM2, RM3, RM4, RM5, RM6, RM7, RM8, RM9, RM10, + RM11, RM12, RM13, RM14, RM15, RM16, RM17, RM18, RM19, RM20, + RM21, RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30, + RM31, RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40, + RM41, RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50, + RM51, RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60, + RM61, RM62, RM63, RM64, RM65, RM66 }; + +/* These versions of the macros use the stack, as normal. There are debugging +versions and production versions. Note that the "rw" argument of RMATCH isn't +actually used in this definition. */ + +#ifndef NO_RECURSE +#define REGISTER register + +#ifdef PCRE_DEBUG +#define RMATCH(ra,rb,rc,rd,re,rw) \ + { \ + printf("match() called in line %d\n", __LINE__); \ + rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1); \ + printf("to line %d\n", __LINE__); \ + } +#define RRETURN(ra) \ + { \ + printf("match() returned %d from line %d ", ra, __LINE__); \ + return ra; \ + } +#else +#define RMATCH(ra,rb,rc,rd,re,rw) \ + rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1) +#define RRETURN(ra) return ra +#endif + +#else + + +/* These versions of the macros manage a private stack on the heap. Note that +the "rd" argument of RMATCH isn't actually used in this definition. It's the md +argument of match(), which never changes. */ + +#define REGISTER + +#define RMATCH(ra,rb,rc,rd,re,rw)\ + {\ + heapframe *newframe = frame->Xnextframe;\ + if (newframe == NULL)\ + {\ + newframe = (heapframe *)(PUBL(stack_malloc))(sizeof(heapframe));\ + if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\ + newframe->Xnextframe = NULL;\ + frame->Xnextframe = newframe;\ + }\ + frame->Xwhere = rw;\ + newframe->Xeptr = ra;\ + newframe->Xecode = rb;\ + newframe->Xmstart = mstart;\ + newframe->Xoffset_top = rc;\ + newframe->Xeptrb = re;\ + newframe->Xrdepth = frame->Xrdepth + 1;\ + newframe->Xprevframe = frame;\ + frame = newframe;\ + DPRINTF(("restarting from line %d\n", __LINE__));\ + goto HEAP_RECURSE;\ + L_##rw:\ + DPRINTF(("jumped back to line %d\n", __LINE__));\ + } + +#define RRETURN(ra)\ + {\ + heapframe *oldframe = frame;\ + frame = oldframe->Xprevframe;\ + if (frame != NULL)\ + {\ + rrc = ra;\ + goto HEAP_RETURN;\ + }\ + return ra;\ + } + + +/* Structure for remembering the local variables in a private frame */ + +typedef struct heapframe { + struct heapframe *Xprevframe; + struct heapframe *Xnextframe; + + /* Function arguments that may change */ + + PCRE_PUCHAR Xeptr; + const pcre_uchar *Xecode; + PCRE_PUCHAR Xmstart; + int Xoffset_top; + eptrblock *Xeptrb; + unsigned int Xrdepth; + + /* Function local variables */ + + PCRE_PUCHAR Xcallpat; +#ifdef SUPPORT_UTF + PCRE_PUCHAR Xcharptr; +#endif + PCRE_PUCHAR Xdata; + PCRE_PUCHAR Xnext; + PCRE_PUCHAR Xpp; + PCRE_PUCHAR Xprev; + PCRE_PUCHAR Xsaved_eptr; + + recursion_info Xnew_recursive; + + BOOL Xcur_is_word; + BOOL Xcondition; + BOOL Xprev_is_word; + +#ifdef SUPPORT_UCP + int Xprop_type; + int Xprop_value; + int Xprop_fail_result; + int Xoclength; + pcre_uchar Xocchars[6]; +#endif + + int Xcodelink; + int Xctype; + unsigned int Xfc; + int Xfi; + int Xlength; + int Xmax; + int Xmin; + int Xnumber; + int Xoffset; + int Xop; + int Xsave_capture_last; + int Xsave_offset1, Xsave_offset2, Xsave_offset3; + int Xstacksave[REC_STACK_SAVE_MAX]; + + eptrblock Xnewptrb; + + /* Where to jump back to */ + + int Xwhere; + +} heapframe; + +#endif + + +/*************************************************************************** +***************************************************************************/ + + + +/************************************************* +* Match from current position * +*************************************************/ + +/* This function is called recursively in many circumstances. Whenever it +returns a negative (error) response, the outer incarnation must also return the +same response. */ + +/* These macros pack up tests that are used for partial matching, and which +appear several times in the code. We set the "hit end" flag if the pointer is +at the end of the subject and also past the start of the subject (i.e. +something has been matched). For hard partial matching, we then return +immediately. The second one is used when we already know we are past the end of +the subject. */ + +#define CHECK_PARTIAL()\ + if (md->partial != 0 && eptr >= md->end_subject && \ + eptr > md->start_used_ptr) \ + { \ + md->hitend = TRUE; \ + if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \ + } + +#define SCHECK_PARTIAL()\ + if (md->partial != 0 && eptr > md->start_used_ptr) \ + { \ + md->hitend = TRUE; \ + if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \ + } + + +/* Performance note: It might be tempting to extract commonly used fields from +the md structure (e.g. utf, end_subject) into individual variables to improve +performance. Tests using gcc on a SPARC disproved this; in the first case, it +made performance worse. + +Arguments: + eptr pointer to current character in subject + ecode pointer to current position in compiled code + mstart pointer to the current match start position (can be modified + by encountering \K) + offset_top current top pointer + md pointer to "static" info for the match + eptrb pointer to chain of blocks containing eptr at start of + brackets - for testing for empty matches + rdepth the recursion depth + +Returns: MATCH_MATCH if matched ) these values are >= 0 + MATCH_NOMATCH if failed to match ) + a negative MATCH_xxx value for PRUNE, SKIP, etc + a negative PCRE_ERROR_xxx value if aborted by an error condition + (e.g. stopped by repeated call or recursion limit) +*/ + +static int +match(REGISTER PCRE_PUCHAR eptr, REGISTER const pcre_uchar *ecode, + PCRE_PUCHAR mstart, int offset_top, match_data *md, eptrblock *eptrb, + unsigned int rdepth) +{ +/* These variables do not need to be preserved over recursion in this function, +so they can be ordinary variables in all cases. Mark some of them with +"register" because they are used a lot in loops. */ + +register int rrc; /* Returns from recursive calls */ +register int i; /* Used for loops not involving calls to RMATCH() */ +register unsigned int c; /* Character values not kept over RMATCH() calls */ +register BOOL utf; /* Local copy of UTF flag for speed */ + +BOOL minimize, possessive; /* Quantifier options */ +BOOL caseless; +int condcode; + +/* When recursion is not being used, all "local" variables that have to be +preserved over calls to RMATCH() are part of a "frame". We set up the top-level +frame on the stack here; subsequent instantiations are obtained from the heap +whenever RMATCH() does a "recursion". See the macro definitions above. Putting +the top-level on the stack rather than malloc-ing them all gives a performance +boost in many cases where there is not much "recursion". */ + +#ifdef NO_RECURSE +heapframe *frame = (heapframe *)md->match_frames_base; + +/* Copy in the original argument variables */ + +frame->Xeptr = eptr; +frame->Xecode = ecode; +frame->Xmstart = mstart; +frame->Xoffset_top = offset_top; +frame->Xeptrb = eptrb; +frame->Xrdepth = rdepth; + +/* This is where control jumps back to to effect "recursion" */ + +HEAP_RECURSE: + +/* Macros make the argument variables come from the current frame */ + +#define eptr frame->Xeptr +#define ecode frame->Xecode +#define mstart frame->Xmstart +#define offset_top frame->Xoffset_top +#define eptrb frame->Xeptrb +#define rdepth frame->Xrdepth + +/* Ditto for the local variables */ + +#ifdef SUPPORT_UTF +#define charptr frame->Xcharptr +#endif +#define callpat frame->Xcallpat +#define codelink frame->Xcodelink +#define data frame->Xdata +#define next frame->Xnext +#define pp frame->Xpp +#define prev frame->Xprev +#define saved_eptr frame->Xsaved_eptr + +#define new_recursive frame->Xnew_recursive + +#define cur_is_word frame->Xcur_is_word +#define condition frame->Xcondition +#define prev_is_word frame->Xprev_is_word + +#ifdef SUPPORT_UCP +#define prop_type frame->Xprop_type +#define prop_value frame->Xprop_value +#define prop_fail_result frame->Xprop_fail_result +#define oclength frame->Xoclength +#define occhars frame->Xocchars +#endif + +#define ctype frame->Xctype +#define fc frame->Xfc +#define fi frame->Xfi +#define length frame->Xlength +#define max frame->Xmax +#define min frame->Xmin +#define number frame->Xnumber +#define offset frame->Xoffset +#define op frame->Xop +#define save_capture_last frame->Xsave_capture_last +#define save_offset1 frame->Xsave_offset1 +#define save_offset2 frame->Xsave_offset2 +#define save_offset3 frame->Xsave_offset3 +#define stacksave frame->Xstacksave + +#define newptrb frame->Xnewptrb + +/* When recursion is being used, local variables are allocated on the stack and +get preserved during recursion in the normal way. In this environment, fi and +i, and fc and c, can be the same variables. */ + +#else /* NO_RECURSE not defined */ +#define fi i +#define fc c + +/* Many of the following variables are used only in small blocks of the code. +My normal style of coding would have declared them within each of those blocks. +However, in order to accommodate the version of this code that uses an external +"stack" implemented on the heap, it is easier to declare them all here, so the +declarations can be cut out in a block. The only declarations within blocks +below are for variables that do not have to be preserved over a recursive call +to RMATCH(). */ + +#ifdef SUPPORT_UTF +const pcre_uchar *charptr; +#endif +const pcre_uchar *callpat; +const pcre_uchar *data; +const pcre_uchar *next; +PCRE_PUCHAR pp; +const pcre_uchar *prev; +PCRE_PUCHAR saved_eptr; + +recursion_info new_recursive; + +BOOL cur_is_word; +BOOL condition; +BOOL prev_is_word; + +#ifdef SUPPORT_UCP +int prop_type; +int prop_value; +int prop_fail_result; +int oclength; +pcre_uchar occhars[6]; +#endif + +int codelink; +int ctype; +int length; +int max; +int min; +int number; +int offset; +int op; +int save_capture_last; +int save_offset1, save_offset2, save_offset3; +int stacksave[REC_STACK_SAVE_MAX]; + +eptrblock newptrb; + +/* There is a special fudge for calling match() in a way that causes it to +measure the size of its basic stack frame when the stack is being used for +recursion. The second argument (ecode) being NULL triggers this behaviour. It +cannot normally ever be NULL. The return is the negated value of the frame +size. */ + +if (ecode == NULL) + { + if (rdepth == 0) + return match((PCRE_PUCHAR)&rdepth, NULL, NULL, 0, NULL, NULL, 1); + else + { + int len = (char *)&rdepth - (char *)eptr; + return (len > 0)? -len : len; + } + } +#endif /* NO_RECURSE */ + +/* To save space on the stack and in the heap frame, I have doubled up on some +of the local variables that are used only in localised parts of the code, but +still need to be preserved over recursive calls of match(). These macros define +the alternative names that are used. */ + +#define allow_zero cur_is_word +#define cbegroup condition +#define code_offset codelink +#define condassert condition +#define matched_once prev_is_word +#define foc number +#define save_mark data + +/* These statements are here to stop the compiler complaining about unitialized +variables. */ + +#ifdef SUPPORT_UCP +prop_value = 0; +prop_fail_result = 0; +#endif + + +/* This label is used for tail recursion, which is used in a few cases even +when NO_RECURSE is not defined, in order to reduce the amount of stack that is +used. Thanks to Ian Taylor for noticing this possibility and sending the +original patch. */ + +TAIL_RECURSE: + +/* OK, now we can get on with the real code of the function. Recursive calls +are specified by the macro RMATCH and RRETURN is used to return. When +NO_RECURSE is *not* defined, these just turn into a recursive call to match() +and a "return", respectively (possibly with some debugging if PCRE_DEBUG is +defined). However, RMATCH isn't like a function call because it's quite a +complicated macro. It has to be used in one particular way. This shouldn't, +however, impact performance when true recursion is being used. */ + +#ifdef SUPPORT_UTF +utf = md->utf; /* Local copy of the flag */ +#else +utf = FALSE; +#endif + +/* First check that we haven't called match() too many times, or that we +haven't exceeded the recursive call limit. */ + +if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT); +if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT); + +/* At the start of a group with an unlimited repeat that may match an empty +string, the variable md->match_function_type is set to MATCH_CBEGROUP. It is +done this way to save having to use another function argument, which would take +up space on the stack. See also MATCH_CONDASSERT below. + +When MATCH_CBEGROUP is set, add the current subject pointer to the chain of +such remembered pointers, to be checked when we hit the closing ket, in order +to break infinite loops that match no characters. When match() is called in +other circumstances, don't add to the chain. The MATCH_CBEGROUP feature must +NOT be used with tail recursion, because the memory block that is used is on +the stack, so a new one may be required for each match(). */ + +if (md->match_function_type == MATCH_CBEGROUP) + { + newptrb.epb_saved_eptr = eptr; + newptrb.epb_prev = eptrb; + eptrb = &newptrb; + md->match_function_type = 0; + } + +/* Now start processing the opcodes. */ + +for (;;) + { + minimize = possessive = FALSE; + op = *ecode; + + switch(op) + { + case OP_MARK: + md->nomatch_mark = ecode + 2; + md->mark = NULL; /* In case previously set by assertion */ + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md, + eptrb, RM55); + if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) && + md->mark == NULL) md->mark = ecode + 2; + + /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an + argument, and we must check whether that argument matches this MARK's + argument. It is passed back in md->start_match_ptr (an overloading of that + variable). If it does match, we reset that variable to the current subject + position and return MATCH_SKIP. Otherwise, pass back the return code + unaltered. */ + + else if (rrc == MATCH_SKIP_ARG && + STRCMP_UC_UC(ecode + 2, md->start_match_ptr) == 0) + { + md->start_match_ptr = eptr; + RRETURN(MATCH_SKIP); + } + RRETURN(rrc); + + case OP_FAIL: + RRETURN(MATCH_NOMATCH); + + /* COMMIT overrides PRUNE, SKIP, and THEN */ + + case OP_COMMIT: + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, + eptrb, RM52); + if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && + rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG && + rrc != MATCH_THEN) + RRETURN(rrc); + RRETURN(MATCH_COMMIT); + + /* PRUNE overrides THEN */ + + case OP_PRUNE: + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, + eptrb, RM51); + if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); + RRETURN(MATCH_PRUNE); + + case OP_PRUNE_ARG: + md->nomatch_mark = ecode + 2; + md->mark = NULL; /* In case previously set by assertion */ + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md, + eptrb, RM56); + if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) && + md->mark == NULL) md->mark = ecode + 2; + if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); + RRETURN(MATCH_PRUNE); + + /* SKIP overrides PRUNE and THEN */ + + case OP_SKIP: + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, + eptrb, RM53); + if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN) + RRETURN(rrc); + md->start_match_ptr = eptr; /* Pass back current position */ + RRETURN(MATCH_SKIP); + + /* Note that, for Perl compatibility, SKIP with an argument does NOT set + nomatch_mark. There is a flag that disables this opcode when re-matching a + pattern that ended with a SKIP for which there was not a matching MARK. */ + + case OP_SKIP_ARG: + if (md->ignore_skip_arg) + { + ecode += PRIV(OP_lengths)[*ecode] + ecode[1]; + break; + } + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md, + eptrb, RM57); + if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN) + RRETURN(rrc); + + /* Pass back the current skip name by overloading md->start_match_ptr and + returning the special MATCH_SKIP_ARG return code. This will either be + caught by a matching MARK, or get to the top, where it causes a rematch + with the md->ignore_skip_arg flag set. */ + + md->start_match_ptr = ecode + 2; + RRETURN(MATCH_SKIP_ARG); + + /* For THEN (and THEN_ARG) we pass back the address of the opcode, so that + the branch in which it occurs can be determined. Overload the start of + match pointer to do this. */ + + case OP_THEN: + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, + eptrb, RM54); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + md->start_match_ptr = ecode; + RRETURN(MATCH_THEN); + + case OP_THEN_ARG: + md->nomatch_mark = ecode + 2; + md->mark = NULL; /* In case previously set by assertion */ + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, + md, eptrb, RM58); + if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) && + md->mark == NULL) md->mark = ecode + 2; + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + md->start_match_ptr = ecode; + RRETURN(MATCH_THEN); + + /* Handle an atomic group that does not contain any capturing parentheses. + This can be handled like an assertion. Prior to 8.13, all atomic groups + were handled this way. In 8.13, the code was changed as below for ONCE, so + that backups pass through the group and thereby reset captured values. + However, this uses a lot more stack, so in 8.20, atomic groups that do not + contain any captures generate OP_ONCE_NC, which can be handled in the old, + less stack intensive way. + + Check the alternative branches in turn - the matching won't pass the KET + for this kind of subpattern. If any one branch matches, we carry on as at + the end of a normal bracket, leaving the subject pointer, but resetting + the start-of-match value in case it was changed by \K. */ + + case OP_ONCE_NC: + prev = ecode; + saved_eptr = eptr; + save_mark = md->mark; + do + { + RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64); + if (rrc == MATCH_MATCH) /* Note: _not_ MATCH_ACCEPT */ + { + mstart = md->start_match_ptr; + break; + } + if (rrc == MATCH_THEN) + { + next = ecode + GET(ecode,1); + if (md->start_match_ptr < next && + (*ecode == OP_ALT || *next == OP_ALT)) + rrc = MATCH_NOMATCH; + } + + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + ecode += GET(ecode,1); + md->mark = save_mark; + } + while (*ecode == OP_ALT); + + /* If hit the end of the group (which could be repeated), fail */ + + if (*ecode != OP_ONCE_NC && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH); + + /* Continue as from after the group, updating the offsets high water + mark, since extracts may have been taken. */ + + do ecode += GET(ecode, 1); while (*ecode == OP_ALT); + + offset_top = md->end_offset_top; + eptr = md->end_match_ptr; + + /* For a non-repeating ket, just continue at this level. This also + happens for a repeating ket if no characters were matched in the group. + This is the forcible breaking of infinite loops as implemented in Perl + 5.005. */ + + if (*ecode == OP_KET || eptr == saved_eptr) + { + ecode += 1+LINK_SIZE; + break; + } + + /* The repeating kets try the rest of the pattern or restart from the + preceding bracket, in the appropriate order. The second "call" of match() + uses tail recursion, to avoid using another stack frame. */ + + if (*ecode == OP_KETRMIN) + { + RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + ecode = prev; + goto TAIL_RECURSE; + } + else /* OP_KETRMAX */ + { + RMATCH(eptr, prev, offset_top, md, eptrb, RM66); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + ecode += 1 + LINK_SIZE; + goto TAIL_RECURSE; + } + /* Control never gets here */ + + /* Handle a capturing bracket, other than those that are possessive with an + unlimited repeat. If there is space in the offset vector, save the current + subject position in the working slot at the top of the vector. We mustn't + change the current values of the data slot, because they may be set from a + previous iteration of this group, and be referred to by a reference inside + the group. A failure to match might occur after the group has succeeded, + if something later on doesn't match. For this reason, we need to restore + the working value and also the values of the final offsets, in case they + were set by a previous iteration of the same bracket. + + If there isn't enough space in the offset vector, treat this as if it were + a non-capturing bracket. Don't worry about setting the flag for the error + case here; that is handled in the code for KET. */ + + case OP_CBRA: + case OP_SCBRA: + number = GET2(ecode, 1+LINK_SIZE); + offset = number << 1; + +#ifdef PCRE_DEBUG + printf("start bracket %d\n", number); + printf("subject="); + pchars(eptr, 16, TRUE, md); + printf("\n"); +#endif + + if (offset < md->offset_max) + { + save_offset1 = md->offset_vector[offset]; + save_offset2 = md->offset_vector[offset+1]; + save_offset3 = md->offset_vector[md->offset_end - number]; + save_capture_last = md->capture_last; + save_mark = md->mark; + + DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3)); + md->offset_vector[md->offset_end - number] = + (int)(eptr - md->start_subject); + + for (;;) + { + if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, + eptrb, RM1); + if (rrc == MATCH_ONCE) break; /* Backing up through an atomic group */ + + /* If we backed up to a THEN, check whether it is within the current + branch by comparing the address of the THEN that is passed back with + the end of the branch. If it is within the current branch, and the + branch is one of two or more alternatives (it either starts or ends + with OP_ALT), we have reached the limit of THEN's action, so convert + the return code to NOMATCH, which will cause normal backtracking to + happen from now on. Otherwise, THEN is passed back to an outer + alternative. This implements Perl's treatment of parenthesized groups, + where a group not containing | does not affect the current alternative, + that is, (X) is NOT the same as (X|(*F)). */ + + if (rrc == MATCH_THEN) + { + next = ecode + GET(ecode,1); + if (md->start_match_ptr < next && + (*ecode == OP_ALT || *next == OP_ALT)) + rrc = MATCH_NOMATCH; + } + + /* Anything other than NOMATCH is passed back. */ + + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + md->capture_last = save_capture_last; + ecode += GET(ecode, 1); + md->mark = save_mark; + if (*ecode != OP_ALT) break; + } + + DPRINTF(("bracket %d failed\n", number)); + md->offset_vector[offset] = save_offset1; + md->offset_vector[offset+1] = save_offset2; + md->offset_vector[md->offset_end - number] = save_offset3; + + /* At this point, rrc will be one of MATCH_ONCE or MATCH_NOMATCH. */ + + RRETURN(rrc); + } + + /* FALL THROUGH ... Insufficient room for saving captured contents. Treat + as a non-capturing bracket. */ + + /* VVVVVVVVVVVVVVVVVVVVVVVVV */ + /* VVVVVVVVVVVVVVVVVVVVVVVVV */ + + DPRINTF(("insufficient capture room: treat as non-capturing\n")); + + /* VVVVVVVVVVVVVVVVVVVVVVVVV */ + /* VVVVVVVVVVVVVVVVVVVVVVVVV */ + + /* Non-capturing or atomic group, except for possessive with unlimited + repeat and ONCE group with no captures. Loop for all the alternatives. + + When we get to the final alternative within the brackets, we used to return + the result of a recursive call to match() whatever happened so it was + possible to reduce stack usage by turning this into a tail recursion, + except in the case of a possibly empty group. However, now that there is + the possiblity of (*THEN) occurring in the final alternative, this + optimization is no longer always possible. + + We can optimize if we know there are no (*THEN)s in the pattern; at present + this is the best that can be done. + + MATCH_ONCE is returned when the end of an atomic group is successfully + reached, but subsequent matching fails. It passes back up the tree (causing + captured values to be reset) until the original atomic group level is + reached. This is tested by comparing md->once_target with the start of the + group. At this point, the return is converted into MATCH_NOMATCH so that + previous backup points can be taken. */ + + case OP_ONCE: + case OP_BRA: + case OP_SBRA: + DPRINTF(("start non-capturing bracket\n")); + + for (;;) + { + if (op >= OP_SBRA || op == OP_ONCE) + md->match_function_type = MATCH_CBEGROUP; + + /* If this is not a possibly empty group, and there are no (*THEN)s in + the pattern, and this is the final alternative, optimize as described + above. */ + + else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT) + { + ecode += PRIV(OP_lengths)[*ecode]; + goto TAIL_RECURSE; + } + + /* In all other cases, we have to make another call to match(). */ + + save_mark = md->mark; + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, eptrb, + RM2); + + /* See comment in the code for capturing groups above about handling + THEN. */ + + if (rrc == MATCH_THEN) + { + next = ecode + GET(ecode,1); + if (md->start_match_ptr < next && + (*ecode == OP_ALT || *next == OP_ALT)) + rrc = MATCH_NOMATCH; + } + + if (rrc != MATCH_NOMATCH) + { + if (rrc == MATCH_ONCE) + { + const pcre_uchar *scode = ecode; + if (*scode != OP_ONCE) /* If not at start, find it */ + { + while (*scode == OP_ALT) scode += GET(scode, 1); + scode -= GET(scode, 1); + } + if (md->once_target == scode) rrc = MATCH_NOMATCH; + } + RRETURN(rrc); + } + ecode += GET(ecode, 1); + md->mark = save_mark; + if (*ecode != OP_ALT) break; + } + + RRETURN(MATCH_NOMATCH); + + /* Handle possessive capturing brackets with an unlimited repeat. We come + here from BRAZERO with allow_zero set TRUE. The offset_vector values are + handled similarly to the normal case above. However, the matching is + different. The end of these brackets will always be OP_KETRPOS, which + returns MATCH_KETRPOS without going further in the pattern. By this means + we can handle the group by iteration rather than recursion, thereby + reducing the amount of stack needed. */ + + case OP_CBRAPOS: + case OP_SCBRAPOS: + allow_zero = FALSE; + + POSSESSIVE_CAPTURE: + number = GET2(ecode, 1+LINK_SIZE); + offset = number << 1; + +#ifdef PCRE_DEBUG + printf("start possessive bracket %d\n", number); + printf("subject="); + pchars(eptr, 16, TRUE, md); + printf("\n"); +#endif + + if (offset < md->offset_max) + { + matched_once = FALSE; + code_offset = (int)(ecode - md->start_code); + + save_offset1 = md->offset_vector[offset]; + save_offset2 = md->offset_vector[offset+1]; + save_offset3 = md->offset_vector[md->offset_end - number]; + save_capture_last = md->capture_last; + + DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3)); + + /* Each time round the loop, save the current subject position for use + when the group matches. For MATCH_MATCH, the group has matched, so we + restart it with a new subject starting position, remembering that we had + at least one match. For MATCH_NOMATCH, carry on with the alternatives, as + usual. If we haven't matched any alternatives in any iteration, check to + see if a previous iteration matched. If so, the group has matched; + continue from afterwards. Otherwise it has failed; restore the previous + capture values before returning NOMATCH. */ + + for (;;) + { + md->offset_vector[md->offset_end - number] = + (int)(eptr - md->start_subject); + if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, + eptrb, RM63); + if (rrc == MATCH_KETRPOS) + { + offset_top = md->end_offset_top; + eptr = md->end_match_ptr; + ecode = md->start_code + code_offset; + save_capture_last = md->capture_last; + matched_once = TRUE; + continue; + } + + /* See comment in the code for capturing groups above about handling + THEN. */ + + if (rrc == MATCH_THEN) + { + next = ecode + GET(ecode,1); + if (md->start_match_ptr < next && + (*ecode == OP_ALT || *next == OP_ALT)) + rrc = MATCH_NOMATCH; + } + + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + md->capture_last = save_capture_last; + ecode += GET(ecode, 1); + if (*ecode != OP_ALT) break; + } + + if (!matched_once) + { + md->offset_vector[offset] = save_offset1; + md->offset_vector[offset+1] = save_offset2; + md->offset_vector[md->offset_end - number] = save_offset3; + } + + if (allow_zero || matched_once) + { + ecode += 1 + LINK_SIZE; + break; + } + + RRETURN(MATCH_NOMATCH); + } + + /* FALL THROUGH ... Insufficient room for saving captured contents. Treat + as a non-capturing bracket. */ + + /* VVVVVVVVVVVVVVVVVVVVVVVVV */ + /* VVVVVVVVVVVVVVVVVVVVVVVVV */ + + DPRINTF(("insufficient capture room: treat as non-capturing\n")); + + /* VVVVVVVVVVVVVVVVVVVVVVVVV */ + /* VVVVVVVVVVVVVVVVVVVVVVVVV */ + + /* Non-capturing possessive bracket with unlimited repeat. We come here + from BRAZERO with allow_zero = TRUE. The code is similar to the above, + without the capturing complication. It is written out separately for speed + and cleanliness. */ + + case OP_BRAPOS: + case OP_SBRAPOS: + allow_zero = FALSE; + + POSSESSIVE_NON_CAPTURE: + matched_once = FALSE; + code_offset = (int)(ecode - md->start_code); + + for (;;) + { + if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; + RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, + eptrb, RM48); + if (rrc == MATCH_KETRPOS) + { + offset_top = md->end_offset_top; + eptr = md->end_match_ptr; + ecode = md->start_code + code_offset; + matched_once = TRUE; + continue; + } + + /* See comment in the code for capturing groups above about handling + THEN. */ + + if (rrc == MATCH_THEN) + { + next = ecode + GET(ecode,1); + if (md->start_match_ptr < next && + (*ecode == OP_ALT || *next == OP_ALT)) + rrc = MATCH_NOMATCH; + } + + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + ecode += GET(ecode, 1); + if (*ecode != OP_ALT) break; + } + + if (matched_once || allow_zero) + { + ecode += 1 + LINK_SIZE; + break; + } + RRETURN(MATCH_NOMATCH); + + /* Control never reaches here. */ + + /* Conditional group: compilation checked that there are no more than + two branches. If the condition is false, skipping the first branch takes us + past the end if there is only one branch, but that's OK because that is + exactly what going to the ket would do. */ + + case OP_COND: + case OP_SCOND: + codelink = GET(ecode, 1); + + /* Because of the way auto-callout works during compile, a callout item is + inserted between OP_COND and an assertion condition. */ + + if (ecode[LINK_SIZE+1] == OP_CALLOUT) + { + if (PUBL(callout) != NULL) + { + PUBL(callout_block) cb; + cb.version = 2; /* Version 1 of the callout block */ + cb.callout_number = ecode[LINK_SIZE+2]; + cb.offset_vector = md->offset_vector; +#ifdef COMPILE_PCRE8 + cb.subject = (PCRE_SPTR)md->start_subject; +#else + cb.subject = (PCRE_SPTR16)md->start_subject; +#endif + cb.subject_length = (int)(md->end_subject - md->start_subject); + cb.start_match = (int)(mstart - md->start_subject); + cb.current_position = (int)(eptr - md->start_subject); + cb.pattern_position = GET(ecode, LINK_SIZE + 3); + cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE); + cb.capture_top = offset_top/2; + cb.capture_last = md->capture_last; + cb.callout_data = md->callout_data; + cb.mark = md->nomatch_mark; + if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH); + if (rrc < 0) RRETURN(rrc); + } + ecode += PRIV(OP_lengths)[OP_CALLOUT]; + } + + condcode = ecode[LINK_SIZE+1]; + + /* Now see what the actual condition is */ + + if (condcode == OP_RREF || condcode == OP_NRREF) /* Recursion test */ + { + if (md->recursive == NULL) /* Not recursing => FALSE */ + { + condition = FALSE; + ecode += GET(ecode, 1); + } + else + { + int recno = GET2(ecode, LINK_SIZE + 2); /* Recursion group number*/ + condition = (recno == RREF_ANY || recno == md->recursive->group_num); + + /* If the test is for recursion into a specific subpattern, and it is + false, but the test was set up by name, scan the table to see if the + name refers to any other numbers, and test them. The condition is true + if any one is set. */ + + if (!condition && condcode == OP_NRREF) + { + pcre_uchar *slotA = md->name_table; + for (i = 0; i < md->name_count; i++) + { + if (GET2(slotA, 0) == recno) break; + slotA += md->name_entry_size; + } + + /* Found a name for the number - there can be only one; duplicate + names for different numbers are allowed, but not vice versa. First + scan down for duplicates. */ + + if (i < md->name_count) + { + pcre_uchar *slotB = slotA; + while (slotB > md->name_table) + { + slotB -= md->name_entry_size; + if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) + { + condition = GET2(slotB, 0) == md->recursive->group_num; + if (condition) break; + } + else break; + } + + /* Scan up for duplicates */ + + if (!condition) + { + slotB = slotA; + for (i++; i < md->name_count; i++) + { + slotB += md->name_entry_size; + if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) + { + condition = GET2(slotB, 0) == md->recursive->group_num; + if (condition) break; + } + else break; + } + } + } + } + + /* Chose branch according to the condition */ + + ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1); + } + } + + else if (condcode == OP_CREF || condcode == OP_NCREF) /* Group used test */ + { + offset = GET2(ecode, LINK_SIZE+2) << 1; /* Doubled ref number */ + condition = offset < offset_top && md->offset_vector[offset] >= 0; + + /* If the numbered capture is unset, but the reference was by name, + scan the table to see if the name refers to any other numbers, and test + them. The condition is true if any one is set. This is tediously similar + to the code above, but not close enough to try to amalgamate. */ + + if (!condition && condcode == OP_NCREF) + { + int refno = offset >> 1; + pcre_uchar *slotA = md->name_table; + + for (i = 0; i < md->name_count; i++) + { + if (GET2(slotA, 0) == refno) break; + slotA += md->name_entry_size; + } + + /* Found a name for the number - there can be only one; duplicate names + for different numbers are allowed, but not vice versa. First scan down + for duplicates. */ + + if (i < md->name_count) + { + pcre_uchar *slotB = slotA; + while (slotB > md->name_table) + { + slotB -= md->name_entry_size; + if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) + { + offset = GET2(slotB, 0) << 1; + condition = offset < offset_top && + md->offset_vector[offset] >= 0; + if (condition) break; + } + else break; + } + + /* Scan up for duplicates */ + + if (!condition) + { + slotB = slotA; + for (i++; i < md->name_count; i++) + { + slotB += md->name_entry_size; + if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) + { + offset = GET2(slotB, 0) << 1; + condition = offset < offset_top && + md->offset_vector[offset] >= 0; + if (condition) break; + } + else break; + } + } + } + } + + /* Chose branch according to the condition */ + + ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1); + } + + else if (condcode == OP_DEF) /* DEFINE - always false */ + { + condition = FALSE; + ecode += GET(ecode, 1); + } + + /* The condition is an assertion. Call match() to evaluate it - setting + md->match_function_type to MATCH_CONDASSERT causes it to stop at the end of + an assertion. */ + + else + { + md->match_function_type = MATCH_CONDASSERT; + RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM3); + if (rrc == MATCH_MATCH) + { + if (md->end_offset_top > offset_top) + offset_top = md->end_offset_top; /* Captures may have happened */ + condition = TRUE; + ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2); + while (*ecode == OP_ALT) ecode += GET(ecode, 1); + } + + /* PCRE doesn't allow the effect of (*THEN) to escape beyond an + assertion; it is therefore treated as NOMATCH. */ + + else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) + { + RRETURN(rrc); /* Need braces because of following else */ + } + else + { + condition = FALSE; + ecode += codelink; + } + } + + /* We are now at the branch that is to be obeyed. As there is only one, can + use tail recursion to avoid using another stack frame, except when there is + unlimited repeat of a possibly empty group. In the latter case, a recursive + call to match() is always required, unless the second alternative doesn't + exist, in which case we can just plough on. Note that, for compatibility + with Perl, the | in a conditional group is NOT treated as creating two + alternatives. If a THEN is encountered in the branch, it propagates out to + the enclosing alternative (unless nested in a deeper set of alternatives, + of course). */ + + if (condition || *ecode == OP_ALT) + { + if (op != OP_SCOND) + { + ecode += 1 + LINK_SIZE; + goto TAIL_RECURSE; + } + + md->match_function_type = MATCH_CBEGROUP; + RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49); + RRETURN(rrc); + } + + /* Condition false & no alternative; continue after the group. */ + + else + { + ecode += 1 + LINK_SIZE; + } + break; + + + /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes, + to close any currently open capturing brackets. */ + + case OP_CLOSE: + number = GET2(ecode, 1); + offset = number << 1; + +#ifdef PCRE_DEBUG + printf("end bracket %d at *ACCEPT", number); + printf("\n"); +#endif + + md->capture_last = number; + if (offset >= md->offset_max) md->offset_overflow = TRUE; else + { + md->offset_vector[offset] = + md->offset_vector[md->offset_end - number]; + md->offset_vector[offset+1] = (int)(eptr - md->start_subject); + if (offset_top <= offset) offset_top = offset + 2; + } + ecode += 1 + IMM2_SIZE; + break; + + + /* End of the pattern, either real or forced. */ + + case OP_END: + case OP_ACCEPT: + case OP_ASSERT_ACCEPT: + + /* If we have matched an empty string, fail if not in an assertion and not + in a recursion if either PCRE_NOTEMPTY is set, or if PCRE_NOTEMPTY_ATSTART + is set and we have matched at the start of the subject. In both cases, + backtracking will then try other alternatives, if any. */ + + if (eptr == mstart && op != OP_ASSERT_ACCEPT && + md->recursive == NULL && + (md->notempty || + (md->notempty_atstart && + mstart == md->start_subject + md->start_offset))) + RRETURN(MATCH_NOMATCH); + + /* Otherwise, we have a match. */ + + md->end_match_ptr = eptr; /* Record where we ended */ + md->end_offset_top = offset_top; /* and how many extracts were taken */ + md->start_match_ptr = mstart; /* and the start (\K can modify) */ + + /* For some reason, the macros don't work properly if an expression is + given as the argument to RRETURN when the heap is in use. */ + + rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT; + RRETURN(rrc); + + /* Assertion brackets. Check the alternative branches in turn - the + matching won't pass the KET for an assertion. If any one branch matches, + the assertion is true. Lookbehind assertions have an OP_REVERSE item at the + start of each branch to move the current point backwards, so the code at + this level is identical to the lookahead case. When the assertion is part + of a condition, we want to return immediately afterwards. The caller of + this incarnation of the match() function will have set MATCH_CONDASSERT in + md->match_function type, and one of these opcodes will be the first opcode + that is processed. We use a local variable that is preserved over calls to + match() to remember this case. */ + + case OP_ASSERT: + case OP_ASSERTBACK: + save_mark = md->mark; + if (md->match_function_type == MATCH_CONDASSERT) + { + condassert = TRUE; + md->match_function_type = 0; + } + else condassert = FALSE; + + do + { + RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM4); + if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) + { + mstart = md->start_match_ptr; /* In case \K reset it */ + break; + } + md->mark = save_mark; + + /* A COMMIT failure must fail the entire assertion, without trying any + subsequent branches. */ + + if (rrc == MATCH_COMMIT) RRETURN(MATCH_NOMATCH); + + /* PCRE does not allow THEN to escape beyond an assertion; it + is treated as NOMATCH. */ + + if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); + ecode += GET(ecode, 1); + } + while (*ecode == OP_ALT); + + if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH); + + /* If checking an assertion for a condition, return MATCH_MATCH. */ + + if (condassert) RRETURN(MATCH_MATCH); + + /* Continue from after the assertion, updating the offsets high water + mark, since extracts may have been taken during the assertion. */ + + do ecode += GET(ecode,1); while (*ecode == OP_ALT); + ecode += 1 + LINK_SIZE; + offset_top = md->end_offset_top; + continue; + + /* Negative assertion: all branches must fail to match. Encountering SKIP, + PRUNE, or COMMIT means we must assume failure without checking subsequent + branches. */ + + case OP_ASSERT_NOT: + case OP_ASSERTBACK_NOT: + save_mark = md->mark; + if (md->match_function_type == MATCH_CONDASSERT) + { + condassert = TRUE; + md->match_function_type = 0; + } + else condassert = FALSE; + + do + { + RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5); + md->mark = save_mark; + if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) RRETURN(MATCH_NOMATCH); + if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT) + { + do ecode += GET(ecode,1); while (*ecode == OP_ALT); + break; + } + + /* PCRE does not allow THEN to escape beyond an assertion; it is treated + as NOMATCH. */ + + if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); + ecode += GET(ecode,1); + } + while (*ecode == OP_ALT); + + if (condassert) RRETURN(MATCH_MATCH); /* Condition assertion */ + + ecode += 1 + LINK_SIZE; + continue; + + /* Move the subject pointer back. This occurs only at the start of + each branch of a lookbehind assertion. If we are too close to the start to + move back, this match function fails. When working with UTF-8 we move + back a number of characters, not bytes. */ + + case OP_REVERSE: +#ifdef SUPPORT_UTF + if (utf) + { + i = GET(ecode, 1); + while (i-- > 0) + { + eptr--; + if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); + BACKCHAR(eptr); + } + } + else +#endif + + /* No UTF-8 support, or not in UTF-8 mode: count is byte count */ + + { + eptr -= GET(ecode, 1); + if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); + } + + /* Save the earliest consulted character, then skip to next op code */ + + if (eptr < md->start_used_ptr) md->start_used_ptr = eptr; + ecode += 1 + LINK_SIZE; + break; + + /* The callout item calls an external function, if one is provided, passing + details of the match so far. This is mainly for debugging, though the + function is able to force a failure. */ + + case OP_CALLOUT: + if (PUBL(callout) != NULL) + { + PUBL(callout_block) cb; + cb.version = 2; /* Version 1 of the callout block */ + cb.callout_number = ecode[1]; + cb.offset_vector = md->offset_vector; +#ifdef COMPILE_PCRE8 + cb.subject = (PCRE_SPTR)md->start_subject; +#else + cb.subject = (PCRE_SPTR16)md->start_subject; +#endif + cb.subject_length = (int)(md->end_subject - md->start_subject); + cb.start_match = (int)(mstart - md->start_subject); + cb.current_position = (int)(eptr - md->start_subject); + cb.pattern_position = GET(ecode, 2); + cb.next_item_length = GET(ecode, 2 + LINK_SIZE); + cb.capture_top = offset_top/2; + cb.capture_last = md->capture_last; + cb.callout_data = md->callout_data; + cb.mark = md->nomatch_mark; + if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH); + if (rrc < 0) RRETURN(rrc); + } + ecode += 2 + 2*LINK_SIZE; + break; + + /* Recursion either matches the current regex, or some subexpression. The + offset data is the offset to the starting bracket from the start of the + whole pattern. (This is so that it works from duplicated subpatterns.) + + The state of the capturing groups is preserved over recursion, and + re-instated afterwards. We don't know how many are started and not yet + finished (offset_top records the completed total) so we just have to save + all the potential data. There may be up to 65535 such values, which is too + large to put on the stack, but using malloc for small numbers seems + expensive. As a compromise, the stack is used when there are no more than + REC_STACK_SAVE_MAX values to store; otherwise malloc is used. + + There are also other values that have to be saved. We use a chained + sequence of blocks that actually live on the stack. Thanks to Robin Houston + for the original version of this logic. It has, however, been hacked around + a lot, so he is not to blame for the current way it works. */ + + case OP_RECURSE: + { + recursion_info *ri; + int recno; + + callpat = md->start_code + GET(ecode, 1); + recno = (callpat == md->start_code)? 0 : + GET2(callpat, 1 + LINK_SIZE); + + /* Check for repeating a recursion without advancing the subject pointer. + This should catch convoluted mutual recursions. (Some simple cases are + caught at compile time.) */ + + for (ri = md->recursive; ri != NULL; ri = ri->prevrec) + if (recno == ri->group_num && eptr == ri->subject_position) + RRETURN(PCRE_ERROR_RECURSELOOP); + + /* Add to "recursing stack" */ + + new_recursive.group_num = recno; + new_recursive.subject_position = eptr; + new_recursive.prevrec = md->recursive; + md->recursive = &new_recursive; + + /* Where to continue from afterwards */ + + ecode += 1 + LINK_SIZE; + + /* Now save the offset data */ + + new_recursive.saved_max = md->offset_end; + if (new_recursive.saved_max <= REC_STACK_SAVE_MAX) + new_recursive.offset_save = stacksave; + else + { + new_recursive.offset_save = + (int *)(PUBL(malloc))(new_recursive.saved_max * sizeof(int)); + if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY); + } + memcpy(new_recursive.offset_save, md->offset_vector, + new_recursive.saved_max * sizeof(int)); + + /* OK, now we can do the recursion. After processing each alternative, + restore the offset data. If there were nested recursions, md->recursive + might be changed, so reset it before looping. */ + + DPRINTF(("Recursing into group %d\n", new_recursive.group_num)); + cbegroup = (*callpat >= OP_SBRA); + do + { + if (cbegroup) md->match_function_type = MATCH_CBEGROUP; + RMATCH(eptr, callpat + PRIV(OP_lengths)[*callpat], offset_top, + md, eptrb, RM6); + memcpy(md->offset_vector, new_recursive.offset_save, + new_recursive.saved_max * sizeof(int)); + md->recursive = new_recursive.prevrec; + if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) + { + DPRINTF(("Recursion matched\n")); + if (new_recursive.offset_save != stacksave) + (PUBL(free))(new_recursive.offset_save); + + /* Set where we got to in the subject, and reset the start in case + it was changed by \K. This *is* propagated back out of a recursion, + for Perl compatibility. */ + + eptr = md->end_match_ptr; + mstart = md->start_match_ptr; + goto RECURSION_MATCHED; /* Exit loop; end processing */ + } + + /* PCRE does not allow THEN or COMMIT to escape beyond a recursion; it + is treated as NOMATCH. */ + + else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN && + rrc != MATCH_COMMIT) + { + DPRINTF(("Recursion gave error %d\n", rrc)); + if (new_recursive.offset_save != stacksave) + (PUBL(free))(new_recursive.offset_save); + RRETURN(rrc); + } + + md->recursive = &new_recursive; + callpat += GET(callpat, 1); + } + while (*callpat == OP_ALT); + + DPRINTF(("Recursion didn't match\n")); + md->recursive = new_recursive.prevrec; + if (new_recursive.offset_save != stacksave) + (PUBL(free))(new_recursive.offset_save); + RRETURN(MATCH_NOMATCH); + } + + RECURSION_MATCHED: + break; + + /* An alternation is the end of a branch; scan along to find the end of the + bracketed group and go to there. */ + + case OP_ALT: + do ecode += GET(ecode,1); while (*ecode == OP_ALT); + break; + + /* BRAZERO, BRAMINZERO and SKIPZERO occur just before a bracket group, + indicating that it may occur zero times. It may repeat infinitely, or not + at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets + with fixed upper repeat limits are compiled as a number of copies, with the + optional ones preceded by BRAZERO or BRAMINZERO. */ + + case OP_BRAZERO: + next = ecode + 1; + RMATCH(eptr, next, offset_top, md, eptrb, RM10); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + do next += GET(next, 1); while (*next == OP_ALT); + ecode = next + 1 + LINK_SIZE; + break; + + case OP_BRAMINZERO: + next = ecode + 1; + do next += GET(next, 1); while (*next == OP_ALT); + RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, eptrb, RM11); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + ecode++; + break; + + case OP_SKIPZERO: + next = ecode+1; + do next += GET(next,1); while (*next == OP_ALT); + ecode = next + 1 + LINK_SIZE; + break; + + /* BRAPOSZERO occurs before a possessive bracket group. Don't do anything + here; just jump to the group, with allow_zero set TRUE. */ + + case OP_BRAPOSZERO: + op = *(++ecode); + allow_zero = TRUE; + if (op == OP_CBRAPOS || op == OP_SCBRAPOS) goto POSSESSIVE_CAPTURE; + goto POSSESSIVE_NON_CAPTURE; + + /* End of a group, repeated or non-repeating. */ + + case OP_KET: + case OP_KETRMIN: + case OP_KETRMAX: + case OP_KETRPOS: + prev = ecode - GET(ecode, 1); + + /* If this was a group that remembered the subject start, in order to break + infinite repeats of empty string matches, retrieve the subject start from + the chain. Otherwise, set it NULL. */ + + if (*prev >= OP_SBRA || *prev == OP_ONCE) + { + saved_eptr = eptrb->epb_saved_eptr; /* Value at start of group */ + eptrb = eptrb->epb_prev; /* Backup to previous group */ + } + else saved_eptr = NULL; + + /* If we are at the end of an assertion group or a non-capturing atomic + group, stop matching and return MATCH_MATCH, but record the current high + water mark for use by positive assertions. We also need to record the match + start in case it was changed by \K. */ + + if ((*prev >= OP_ASSERT && *prev <= OP_ASSERTBACK_NOT) || + *prev == OP_ONCE_NC) + { + md->end_match_ptr = eptr; /* For ONCE_NC */ + md->end_offset_top = offset_top; + md->start_match_ptr = mstart; + RRETURN(MATCH_MATCH); /* Sets md->mark */ + } + + /* For capturing groups we have to check the group number back at the start + and if necessary complete handling an extraction by setting the offsets and + bumping the high water mark. Whole-pattern recursion is coded as a recurse + into group 0, so it won't be picked up here. Instead, we catch it when the + OP_END is reached. Other recursion is handled here. We just have to record + the current subject position and start match pointer and give a MATCH + return. */ + + if (*prev == OP_CBRA || *prev == OP_SCBRA || + *prev == OP_CBRAPOS || *prev == OP_SCBRAPOS) + { + number = GET2(prev, 1+LINK_SIZE); + offset = number << 1; + +#ifdef PCRE_DEBUG + printf("end bracket %d", number); + printf("\n"); +#endif + + /* Handle a recursively called group. */ + + if (md->recursive != NULL && md->recursive->group_num == number) + { + md->end_match_ptr = eptr; + md->start_match_ptr = mstart; + RRETURN(MATCH_MATCH); + } + + /* Deal with capturing */ + + md->capture_last = number; + if (offset >= md->offset_max) md->offset_overflow = TRUE; else + { + /* If offset is greater than offset_top, it means that we are + "skipping" a capturing group, and that group's offsets must be marked + unset. In earlier versions of PCRE, all the offsets were unset at the + start of matching, but this doesn't work because atomic groups and + assertions can cause a value to be set that should later be unset. + Example: matching /(?>(a))b|(a)c/ against "ac". This sets group 1 as + part of the atomic group, but this is not on the final matching path, + so must be unset when 2 is set. (If there is no group 2, there is no + problem, because offset_top will then be 2, indicating no capture.) */ + + if (offset > offset_top) + { + register int *iptr = md->offset_vector + offset_top; + register int *iend = md->offset_vector + offset; + while (iptr < iend) *iptr++ = -1; + } + + /* Now make the extraction */ + + md->offset_vector[offset] = + md->offset_vector[md->offset_end - number]; + md->offset_vector[offset+1] = (int)(eptr - md->start_subject); + if (offset_top <= offset) offset_top = offset + 2; + } + } + + /* For an ordinary non-repeating ket, just continue at this level. This + also happens for a repeating ket if no characters were matched in the + group. This is the forcible breaking of infinite loops as implemented in + Perl 5.005. For a non-repeating atomic group that includes captures, + establish a backup point by processing the rest of the pattern at a lower + level. If this results in a NOMATCH return, pass MATCH_ONCE back to the + original OP_ONCE level, thereby bypassing intermediate backup points, but + resetting any captures that happened along the way. */ + + if (*ecode == OP_KET || eptr == saved_eptr) + { + if (*prev == OP_ONCE) + { + RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM12); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + md->once_target = prev; /* Level at which to change to MATCH_NOMATCH */ + RRETURN(MATCH_ONCE); + } + ecode += 1 + LINK_SIZE; /* Carry on at this level */ + break; + } + + /* OP_KETRPOS is a possessive repeating ket. Remember the current position, + and return the MATCH_KETRPOS. This makes it possible to do the repeats one + at a time from the outer level, thus saving stack. */ + + if (*ecode == OP_KETRPOS) + { + md->end_match_ptr = eptr; + md->end_offset_top = offset_top; + RRETURN(MATCH_KETRPOS); + } + + /* The normal repeating kets try the rest of the pattern or restart from + the preceding bracket, in the appropriate order. In the second case, we can + use tail recursion to avoid using another stack frame, unless we have an + an atomic group or an unlimited repeat of a group that can match an empty + string. */ + + if (*ecode == OP_KETRMIN) + { + RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM7); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (*prev == OP_ONCE) + { + RMATCH(eptr, prev, offset_top, md, eptrb, RM8); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + md->once_target = prev; /* Level at which to change to MATCH_NOMATCH */ + RRETURN(MATCH_ONCE); + } + if (*prev >= OP_SBRA) /* Could match an empty string */ + { + RMATCH(eptr, prev, offset_top, md, eptrb, RM50); + RRETURN(rrc); + } + ecode = prev; + goto TAIL_RECURSE; + } + else /* OP_KETRMAX */ + { + RMATCH(eptr, prev, offset_top, md, eptrb, RM13); + if (rrc == MATCH_ONCE && md->once_target == prev) rrc = MATCH_NOMATCH; + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (*prev == OP_ONCE) + { + RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM9); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + md->once_target = prev; + RRETURN(MATCH_ONCE); + } + ecode += 1 + LINK_SIZE; + goto TAIL_RECURSE; + } + /* Control never gets here */ + + /* Not multiline mode: start of subject assertion, unless notbol. */ + + case OP_CIRC: + if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH); + + /* Start of subject assertion */ + + case OP_SOD: + if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH); + ecode++; + break; + + /* Multiline mode: start of subject unless notbol, or after any newline. */ + + case OP_CIRCM: + if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH); + if (eptr != md->start_subject && + (eptr == md->end_subject || !WAS_NEWLINE(eptr))) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + + /* Start of match assertion */ + + case OP_SOM: + if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH); + ecode++; + break; + + /* Reset the start of match point */ + + case OP_SET_SOM: + mstart = eptr; + ecode++; + break; + + /* Multiline mode: assert before any newline, or before end of subject + unless noteol is set. */ + + case OP_DOLLM: + if (eptr < md->end_subject) + { + if (!IS_NEWLINE(eptr)) + { + if (md->partial != 0 && + eptr + 1 >= md->end_subject && + NLBLOCK->nltype == NLTYPE_FIXED && + NLBLOCK->nllen == 2 && + *eptr == NLBLOCK->nl[0]) + { + md->hitend = TRUE; + if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); + } + RRETURN(MATCH_NOMATCH); + } + } + else + { + if (md->noteol) RRETURN(MATCH_NOMATCH); + SCHECK_PARTIAL(); + } + ecode++; + break; + + /* Not multiline mode: assert before a terminating newline or before end of + subject unless noteol is set. */ + + case OP_DOLL: + if (md->noteol) RRETURN(MATCH_NOMATCH); + if (!md->endonly) goto ASSERT_NL_OR_EOS; + + /* ... else fall through for endonly */ + + /* End of subject assertion (\z) */ + + case OP_EOD: + if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH); + SCHECK_PARTIAL(); + ecode++; + break; + + /* End of subject or ending \n assertion (\Z) */ + + case OP_EODN: + ASSERT_NL_OR_EOS: + if (eptr < md->end_subject && + (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) + { + if (md->partial != 0 && + eptr + 1 >= md->end_subject && + NLBLOCK->nltype == NLTYPE_FIXED && + NLBLOCK->nllen == 2 && + *eptr == NLBLOCK->nl[0]) + { + md->hitend = TRUE; + if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); + } + RRETURN(MATCH_NOMATCH); + } + + /* Either at end of string or \n before end. */ + + SCHECK_PARTIAL(); + ecode++; + break; + + /* Word boundary assertions */ + + case OP_NOT_WORD_BOUNDARY: + case OP_WORD_BOUNDARY: + { + + /* Find out if the previous and current characters are "word" characters. + It takes a bit more work in UTF-8 mode. Characters > 255 are assumed to + be "non-word" characters. Remember the earliest consulted character for + partial matching. */ + +#ifdef SUPPORT_UTF + if (utf) + { + /* Get status of previous character */ + + if (eptr == md->start_subject) prev_is_word = FALSE; else + { + PCRE_PUCHAR lastptr = eptr - 1; + BACKCHAR(lastptr); + if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr; + GETCHAR(c, lastptr); +#ifdef SUPPORT_UCP + if (md->use_ucp) + { + if (c == '_') prev_is_word = TRUE; else + { + int cat = UCD_CATEGORY(c); + prev_is_word = (cat == ucp_L || cat == ucp_N); + } + } + else +#endif + prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; + } + + /* Get status of next character */ + + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + cur_is_word = FALSE; + } + else + { + GETCHAR(c, eptr); +#ifdef SUPPORT_UCP + if (md->use_ucp) + { + if (c == '_') cur_is_word = TRUE; else + { + int cat = UCD_CATEGORY(c); + cur_is_word = (cat == ucp_L || cat == ucp_N); + } + } + else +#endif + cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; + } + } + else +#endif + + /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for + consistency with the behaviour of \w we do use it in this case. */ + + { + /* Get status of previous character */ + + if (eptr == md->start_subject) prev_is_word = FALSE; else + { + if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1; +#ifdef SUPPORT_UCP + if (md->use_ucp) + { + c = eptr[-1]; + if (c == '_') prev_is_word = TRUE; else + { + int cat = UCD_CATEGORY(c); + prev_is_word = (cat == ucp_L || cat == ucp_N); + } + } + else +#endif + prev_is_word = MAX_255(eptr[-1]) + && ((md->ctypes[eptr[-1]] & ctype_word) != 0); + } + + /* Get status of next character */ + + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + cur_is_word = FALSE; + } + else +#ifdef SUPPORT_UCP + if (md->use_ucp) + { + c = *eptr; + if (c == '_') cur_is_word = TRUE; else + { + int cat = UCD_CATEGORY(c); + cur_is_word = (cat == ucp_L || cat == ucp_N); + } + } + else +#endif + cur_is_word = MAX_255(*eptr) + && ((md->ctypes[*eptr] & ctype_word) != 0); + } + + /* Now see if the situation is what we want */ + + if ((*ecode++ == OP_WORD_BOUNDARY)? + cur_is_word == prev_is_word : cur_is_word != prev_is_word) + RRETURN(MATCH_NOMATCH); + } + break; + + /* Match any single character type except newline; have to take care with + CRLF newlines and partial matching. */ + + case OP_ANY: + if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); + if (md->partial != 0 && + eptr + 1 >= md->end_subject && + NLBLOCK->nltype == NLTYPE_FIXED && + NLBLOCK->nllen == 2 && + *eptr == NLBLOCK->nl[0]) + { + md->hitend = TRUE; + if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); + } + + /* Fall through */ + + /* Match any single character whatsoever. */ + + case OP_ALLANY: + if (eptr >= md->end_subject) /* DO NOT merge the eptr++ here; it must */ + { /* not be updated before SCHECK_PARTIAL. */ + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + eptr++; +#ifdef SUPPORT_UTF + if (utf) ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); +#endif + ecode++; + break; + + /* Match a single byte, even in UTF-8 mode. This opcode really does match + any byte, even newline, independent of the setting of PCRE_DOTALL. */ + + case OP_ANYBYTE: + if (eptr >= md->end_subject) /* DO NOT merge the eptr++ here; it must */ + { /* not be updated before SCHECK_PARTIAL. */ + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + eptr++; + ecode++; + break; + + case OP_NOT_DIGIT: + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ( +#if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) + c < 256 && +#endif + (md->ctypes[c] & ctype_digit) != 0 + ) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + + case OP_DIGIT: + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ( +#if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) + c > 255 || +#endif + (md->ctypes[c] & ctype_digit) == 0 + ) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + + case OP_NOT_WHITESPACE: + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ( +#if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) + c < 256 && +#endif + (md->ctypes[c] & ctype_space) != 0 + ) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + + case OP_WHITESPACE: + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ( +#if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) + c > 255 || +#endif + (md->ctypes[c] & ctype_space) == 0 + ) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + + case OP_NOT_WORDCHAR: + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ( +#if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) + c < 256 && +#endif + (md->ctypes[c] & ctype_word) != 0 + ) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + + case OP_WORDCHAR: + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ( +#if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) + c > 255 || +#endif + (md->ctypes[c] & ctype_word) == 0 + ) + RRETURN(MATCH_NOMATCH); + ecode++; + break; + + case OP_ANYNL: + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + switch(c) + { + default: RRETURN(MATCH_NOMATCH); + + case 0x000d: + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + } + else if (*eptr == 0x0a) eptr++; + break; + + case 0x000a: + break; + + case 0x000b: + case 0x000c: + case 0x0085: + case 0x2028: + case 0x2029: + if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); + break; + } + ecode++; + break; + + case OP_NOT_HSPACE: + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + switch(c) + { + default: break; + case 0x09: /* HT */ + case 0x20: /* SPACE */ + case 0xa0: /* NBSP */ + case 0x1680: /* OGHAM SPACE MARK */ + case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ + case 0x2000: /* EN QUAD */ + case 0x2001: /* EM QUAD */ + case 0x2002: /* EN SPACE */ + case 0x2003: /* EM SPACE */ + case 0x2004: /* THREE-PER-EM SPACE */ + case 0x2005: /* FOUR-PER-EM SPACE */ + case 0x2006: /* SIX-PER-EM SPACE */ + case 0x2007: /* FIGURE SPACE */ + case 0x2008: /* PUNCTUATION SPACE */ + case 0x2009: /* THIN SPACE */ + case 0x200A: /* HAIR SPACE */ + case 0x202f: /* NARROW NO-BREAK SPACE */ + case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ + case 0x3000: /* IDEOGRAPHIC SPACE */ + RRETURN(MATCH_NOMATCH); + } + ecode++; + break; + + case OP_HSPACE: + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + switch(c) + { + default: RRETURN(MATCH_NOMATCH); + case 0x09: /* HT */ + case 0x20: /* SPACE */ + case 0xa0: /* NBSP */ + case 0x1680: /* OGHAM SPACE MARK */ + case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ + case 0x2000: /* EN QUAD */ + case 0x2001: /* EM QUAD */ + case 0x2002: /* EN SPACE */ + case 0x2003: /* EM SPACE */ + case 0x2004: /* THREE-PER-EM SPACE */ + case 0x2005: /* FOUR-PER-EM SPACE */ + case 0x2006: /* SIX-PER-EM SPACE */ + case 0x2007: /* FIGURE SPACE */ + case 0x2008: /* PUNCTUATION SPACE */ + case 0x2009: /* THIN SPACE */ + case 0x200A: /* HAIR SPACE */ + case 0x202f: /* NARROW NO-BREAK SPACE */ + case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ + case 0x3000: /* IDEOGRAPHIC SPACE */ + break; + } + ecode++; + break; + + case OP_NOT_VSPACE: + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + switch(c) + { + default: break; + case 0x0a: /* LF */ + case 0x0b: /* VT */ + case 0x0c: /* FF */ + case 0x0d: /* CR */ + case 0x85: /* NEL */ + case 0x2028: /* LINE SEPARATOR */ + case 0x2029: /* PARAGRAPH SEPARATOR */ + RRETURN(MATCH_NOMATCH); + } + ecode++; + break; + + case OP_VSPACE: + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + switch(c) + { + default: RRETURN(MATCH_NOMATCH); + case 0x0a: /* LF */ + case 0x0b: /* VT */ + case 0x0c: /* FF */ + case 0x0d: /* CR */ + case 0x85: /* NEL */ + case 0x2028: /* LINE SEPARATOR */ + case 0x2029: /* PARAGRAPH SEPARATOR */ + break; + } + ecode++; + break; + +#ifdef SUPPORT_UCP + /* Check the next character by Unicode property. We will get here only + if the support is in the binary; otherwise a compile-time error occurs. */ + + case OP_PROP: + case OP_NOTPROP: + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + { + const ucd_record *prop = GET_UCD(c); + + switch(ecode[1]) + { + case PT_ANY: + if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH); + break; + + case PT_LAMP: + if ((prop->chartype == ucp_Lu || + prop->chartype == ucp_Ll || + prop->chartype == ucp_Lt) == (op == OP_NOTPROP)) + RRETURN(MATCH_NOMATCH); + break; + + case PT_GC: + if ((ecode[2] != PRIV(ucp_gentype)[prop->chartype]) == (op == OP_PROP)) + RRETURN(MATCH_NOMATCH); + break; + + case PT_PC: + if ((ecode[2] != prop->chartype) == (op == OP_PROP)) + RRETURN(MATCH_NOMATCH); + break; + + case PT_SC: + if ((ecode[2] != prop->script) == (op == OP_PROP)) + RRETURN(MATCH_NOMATCH); + break; + + /* These are specials */ + + case PT_ALNUM: + if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L || + PRIV(ucp_gentype)[prop->chartype] == ucp_N) == (op == OP_NOTPROP)) + RRETURN(MATCH_NOMATCH); + break; + + case PT_SPACE: /* Perl space */ + if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z || + c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR) + == (op == OP_NOTPROP)) + RRETURN(MATCH_NOMATCH); + break; + + case PT_PXSPACE: /* POSIX space */ + if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z || + c == CHAR_HT || c == CHAR_NL || c == CHAR_VT || + c == CHAR_FF || c == CHAR_CR) + == (op == OP_NOTPROP)) + RRETURN(MATCH_NOMATCH); + break; + + case PT_WORD: + if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L || + PRIV(ucp_gentype)[prop->chartype] == ucp_N || + c == CHAR_UNDERSCORE) == (op == OP_NOTPROP)) + RRETURN(MATCH_NOMATCH); + break; + + /* This should never occur */ + + default: + RRETURN(PCRE_ERROR_INTERNAL); + } + + ecode += 3; + } + break; + + /* Match an extended Unicode sequence. We will get here only if the support + is in the binary; otherwise a compile-time error occurs. */ + + case OP_EXTUNI: + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if (UCD_CATEGORY(c) == ucp_M) RRETURN(MATCH_NOMATCH); + while (eptr < md->end_subject) + { + int len = 1; + if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } + if (UCD_CATEGORY(c) != ucp_M) break; + eptr += len; + } + CHECK_PARTIAL(); + ecode++; + break; +#endif + + + /* Match a back reference, possibly repeatedly. Look past the end of the + item to see if there is repeat information following. The code is similar + to that for character classes, but repeated for efficiency. Then obey + similar code to character type repeats - written out again for speed. + However, if the referenced string is the empty string, always treat + it as matched, any number of times (otherwise there could be infinite + loops). */ + + case OP_REF: + case OP_REFI: + caseless = op == OP_REFI; + offset = GET2(ecode, 1) << 1; /* Doubled ref number */ + ecode += 1 + IMM2_SIZE; + + /* If the reference is unset, there are two possibilities: + + (a) In the default, Perl-compatible state, set the length negative; + this ensures that every attempt at a match fails. We can't just fail + here, because of the possibility of quantifiers with zero minima. + + (b) If the JavaScript compatibility flag is set, set the length to zero + so that the back reference matches an empty string. + + Otherwise, set the length to the length of what was matched by the + referenced subpattern. */ + + if (offset >= offset_top || md->offset_vector[offset] < 0) + length = (md->jscript_compat)? 0 : -1; + else + length = md->offset_vector[offset+1] - md->offset_vector[offset]; + + /* Set up for repetition, or handle the non-repeated case */ + + switch (*ecode) + { + case OP_CRSTAR: + case OP_CRMINSTAR: + case OP_CRPLUS: + case OP_CRMINPLUS: + case OP_CRQUERY: + case OP_CRMINQUERY: + c = *ecode++ - OP_CRSTAR; + minimize = (c & 1) != 0; + min = rep_min[c]; /* Pick up values from tables; */ + max = rep_max[c]; /* zero for max => infinity */ + if (max == 0) max = INT_MAX; + break; + + case OP_CRRANGE: + case OP_CRMINRANGE: + minimize = (*ecode == OP_CRMINRANGE); + min = GET2(ecode, 1); + max = GET2(ecode, 1 + IMM2_SIZE); + if (max == 0) max = INT_MAX; + ecode += 1 + 2 * IMM2_SIZE; + break; + + default: /* No repeat follows */ + if ((length = match_ref(offset, eptr, length, md, caseless)) < 0) + { + if (length == -2) eptr = md->end_subject; /* Partial match */ + CHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + eptr += length; + continue; /* With the main loop */ + } + + /* Handle repeated back references. If the length of the reference is + zero, just continue with the main loop. If the length is negative, it + means the reference is unset in non-Java-compatible mode. If the minimum is + zero, we can continue at the same level without recursion. For any other + minimum, carrying on will result in NOMATCH. */ + + if (length == 0) continue; + if (length < 0 && min == 0) continue; + + /* First, ensure the minimum number of matches are present. We get back + the length of the reference string explicitly rather than passing the + address of eptr, so that eptr can be a register variable. */ + + for (i = 1; i <= min; i++) + { + int slength; + if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) + { + if (slength == -2) eptr = md->end_subject; /* Partial match */ + CHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + eptr += slength; + } + + /* If min = max, continue at the same level without recursion. + They are not both allowed to be zero. */ + + if (min == max) continue; + + /* If minimizing, keep trying and advancing the pointer */ + + if (minimize) + { + for (fi = min;; fi++) + { + int slength; + RMATCH(eptr, ecode, offset_top, md, eptrb, RM14); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) + { + if (slength == -2) eptr = md->end_subject; /* Partial match */ + CHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + eptr += slength; + } + /* Control never gets here */ + } + + /* If maximizing, find the longest string and work backwards */ + + else + { + pp = eptr; + for (i = min; i < max; i++) + { + int slength; + if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) + { + /* Can't use CHECK_PARTIAL because we don't want to update eptr in + the soft partial matching case. */ + + if (slength == -2 && md->partial != 0 && + md->end_subject > md->start_used_ptr) + { + md->hitend = TRUE; + if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); + } + break; + } + eptr += slength; + } + + while (eptr >= pp) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM15); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + eptr -= length; + } + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + /* Match a bit-mapped character class, possibly repeatedly. This op code is + used when all the characters in the class have values in the range 0-255, + and either the matching is caseful, or the characters are in the range + 0-127 when UTF-8 processing is enabled. The only difference between + OP_CLASS and OP_NCLASS occurs when a data character outside the range is + encountered. + + First, look past the end of the item to see if there is repeat information + following. Then obey similar code to character type repeats - written out + again for speed. */ + + case OP_NCLASS: + case OP_CLASS: + { + /* The data variable is saved across frames, so the byte map needs to + be stored there. */ +#define BYTE_MAP ((pcre_uint8 *)data) + data = ecode + 1; /* Save for matching */ + ecode += 1 + (32 / sizeof(pcre_uchar)); /* Advance past the item */ + + switch (*ecode) + { + case OP_CRSTAR: + case OP_CRMINSTAR: + case OP_CRPLUS: + case OP_CRMINPLUS: + case OP_CRQUERY: + case OP_CRMINQUERY: + c = *ecode++ - OP_CRSTAR; + minimize = (c & 1) != 0; + min = rep_min[c]; /* Pick up values from tables; */ + max = rep_max[c]; /* zero for max => infinity */ + if (max == 0) max = INT_MAX; + break; + + case OP_CRRANGE: + case OP_CRMINRANGE: + minimize = (*ecode == OP_CRMINRANGE); + min = GET2(ecode, 1); + max = GET2(ecode, 1 + IMM2_SIZE); + if (max == 0) max = INT_MAX; + ecode += 1 + 2 * IMM2_SIZE; + break; + + default: /* No repeat follows */ + min = max = 1; + break; + } + + /* First, ensure the minimum number of matches are present. */ + +#ifdef SUPPORT_UTF + if (utf) + { + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(c, eptr); + if (c > 255) + { + if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); + } + else + if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); + } + } + else +#endif + /* Not UTF mode */ + { + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + c = *eptr++; +#ifndef COMPILE_PCRE8 + if (c > 255) + { + if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); + } + else +#endif + if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); + } + } + + /* If max == min we can continue with the main loop without the + need to recurse. */ + + if (min == max) continue; + + /* If minimizing, keep testing the rest of the expression and advancing + the pointer while it matches the class. */ + + if (minimize) + { +#ifdef SUPPORT_UTF + if (utf) + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM16); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(c, eptr); + if (c > 255) + { + if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); + } + else + if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); + } + } + else +#endif + /* Not UTF mode */ + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM17); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + c = *eptr++; +#ifndef COMPILE_PCRE8 + if (c > 255) + { + if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); + } + else +#endif + if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); + } + } + /* Control never gets here */ + } + + /* If maximizing, find the longest possible run, then work backwards. */ + + else + { + pp = eptr; + +#ifdef SUPPORT_UTF + if (utf) + { + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(c, eptr, len); + if (c > 255) + { + if (op == OP_CLASS) break; + } + else + if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break; + eptr += len; + } + for (;;) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM18); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (eptr-- == pp) break; /* Stop if tried at original pos */ + BACKCHAR(eptr); + } + } + else +#endif + /* Not UTF mode */ + { + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + c = *eptr; +#ifndef COMPILE_PCRE8 + if (c > 255) + { + if (op == OP_CLASS) break; + } + else +#endif + if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break; + eptr++; + } + while (eptr >= pp) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM19); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + eptr--; + } + } + + RRETURN(MATCH_NOMATCH); + } +#undef BYTE_MAP + } + /* Control never gets here */ + + + /* Match an extended character class. This opcode is encountered only + when UTF-8 mode mode is supported. Nevertheless, we may not be in UTF-8 + mode, because Unicode properties are supported in non-UTF-8 mode. */ + +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + case OP_XCLASS: + { + data = ecode + 1 + LINK_SIZE; /* Save for matching */ + ecode += GET(ecode, 1); /* Advance past the item */ + + switch (*ecode) + { + case OP_CRSTAR: + case OP_CRMINSTAR: + case OP_CRPLUS: + case OP_CRMINPLUS: + case OP_CRQUERY: + case OP_CRMINQUERY: + c = *ecode++ - OP_CRSTAR; + minimize = (c & 1) != 0; + min = rep_min[c]; /* Pick up values from tables; */ + max = rep_max[c]; /* zero for max => infinity */ + if (max == 0) max = INT_MAX; + break; + + case OP_CRRANGE: + case OP_CRMINRANGE: + minimize = (*ecode == OP_CRMINRANGE); + min = GET2(ecode, 1); + max = GET2(ecode, 1 + IMM2_SIZE); + if (max == 0) max = INT_MAX; + ecode += 1 + 2 * IMM2_SIZE; + break; + + default: /* No repeat follows */ + min = max = 1; + break; + } + + /* First, ensure the minimum number of matches are present. */ + + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH); + } + + /* If max == min we can continue with the main loop without the + need to recurse. */ + + if (min == max) continue; + + /* If minimizing, keep testing the rest of the expression and advancing + the pointer while it matches the class. */ + + if (minimize) + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM20); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + } + + /* If maximizing, find the longest possible run, then work backwards. */ + + else + { + pp = eptr; + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } +#ifdef SUPPORT_UTF + GETCHARLENTEST(c, eptr, len); +#else + c = *eptr; +#endif + if (!PRIV(xclass)(c, data, utf)) break; + eptr += len; + } + for(;;) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM21); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (eptr-- == pp) break; /* Stop if tried at original pos */ +#ifdef SUPPORT_UTF + if (utf) BACKCHAR(eptr); +#endif + } + RRETURN(MATCH_NOMATCH); + } + + /* Control never gets here */ + } +#endif /* End of XCLASS */ + + /* Match a single character, casefully */ + + case OP_CHAR: +#ifdef SUPPORT_UTF + if (utf) + { + length = 1; + ecode++; + GETCHARLEN(fc, ecode, length); + if (length > md->end_subject - eptr) + { + CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ + RRETURN(MATCH_NOMATCH); + } + while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH); + } + else +#endif + /* Not UTF mode */ + { + if (md->end_subject - eptr < 1) + { + SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ + RRETURN(MATCH_NOMATCH); + } + if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH); + ecode += 2; + } + break; + + /* Match a single character, caselessly. If we are at the end of the + subject, give up immediately. */ + + case OP_CHARI: + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + +#ifdef SUPPORT_UTF + if (utf) + { + length = 1; + ecode++; + GETCHARLEN(fc, ecode, length); + + /* If the pattern character's value is < 128, we have only one byte, and + we know that its other case must also be one byte long, so we can use the + fast lookup table. We know that there is at least one byte left in the + subject. */ + + if (fc < 128) + { + if (md->lcc[fc] + != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH); + ecode++; + eptr++; + } + + /* Otherwise we must pick up the subject character. Note that we cannot + use the value of "length" to check for sufficient bytes left, because the + other case of the character may have more or fewer bytes. */ + + else + { + unsigned int dc; + GETCHARINC(dc, eptr); + ecode += length; + + /* If we have Unicode property support, we can use it to test the other + case of the character, if there is one. */ + + if (fc != dc) + { +#ifdef SUPPORT_UCP + if (dc != UCD_OTHERCASE(fc)) +#endif + RRETURN(MATCH_NOMATCH); + } + } + } + else +#endif /* SUPPORT_UTF */ + + /* Not UTF mode */ + { + if (TABLE_GET(ecode[1], md->lcc, ecode[1]) + != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH); + eptr++; + ecode += 2; + } + break; + + /* Match a single character repeatedly. */ + + case OP_EXACT: + case OP_EXACTI: + min = max = GET2(ecode, 1); + ecode += 1 + IMM2_SIZE; + goto REPEATCHAR; + + case OP_POSUPTO: + case OP_POSUPTOI: + possessive = TRUE; + /* Fall through */ + + case OP_UPTO: + case OP_UPTOI: + case OP_MINUPTO: + case OP_MINUPTOI: + min = 0; + max = GET2(ecode, 1); + minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI; + ecode += 1 + IMM2_SIZE; + goto REPEATCHAR; + + case OP_POSSTAR: + case OP_POSSTARI: + possessive = TRUE; + min = 0; + max = INT_MAX; + ecode++; + goto REPEATCHAR; + + case OP_POSPLUS: + case OP_POSPLUSI: + possessive = TRUE; + min = 1; + max = INT_MAX; + ecode++; + goto REPEATCHAR; + + case OP_POSQUERY: + case OP_POSQUERYI: + possessive = TRUE; + min = 0; + max = 1; + ecode++; + goto REPEATCHAR; + + case OP_STAR: + case OP_STARI: + case OP_MINSTAR: + case OP_MINSTARI: + case OP_PLUS: + case OP_PLUSI: + case OP_MINPLUS: + case OP_MINPLUSI: + case OP_QUERY: + case OP_QUERYI: + case OP_MINQUERY: + case OP_MINQUERYI: + c = *ecode++ - ((op < OP_STARI)? OP_STAR : OP_STARI); + minimize = (c & 1) != 0; + min = rep_min[c]; /* Pick up values from tables; */ + max = rep_max[c]; /* zero for max => infinity */ + if (max == 0) max = INT_MAX; + + /* Common code for all repeated single-character matches. */ + + REPEATCHAR: +#ifdef SUPPORT_UTF + if (utf) + { + length = 1; + charptr = ecode; + GETCHARLEN(fc, ecode, length); + ecode += length; + + /* Handle multibyte character matching specially here. There is + support for caseless matching if UCP support is present. */ + + if (length > 1) + { +#ifdef SUPPORT_UCP + unsigned int othercase; + if (op >= OP_STARI && /* Caseless */ + (othercase = UCD_OTHERCASE(fc)) != fc) + oclength = PRIV(ord2utf)(othercase, occhars); + else oclength = 0; +#endif /* SUPPORT_UCP */ + + for (i = 1; i <= min; i++) + { + if (eptr <= md->end_subject - length && + memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length; +#ifdef SUPPORT_UCP + else if (oclength > 0 && + eptr <= md->end_subject - oclength && + memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength; +#endif /* SUPPORT_UCP */ + else + { + CHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + } + + if (min == max) continue; + + if (minimize) + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM22); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr <= md->end_subject - length && + memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length; +#ifdef SUPPORT_UCP + else if (oclength > 0 && + eptr <= md->end_subject - oclength && + memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength; +#endif /* SUPPORT_UCP */ + else + { + CHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + } + /* Control never gets here */ + } + + else /* Maximize */ + { + pp = eptr; + for (i = min; i < max; i++) + { + if (eptr <= md->end_subject - length && + memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length; +#ifdef SUPPORT_UCP + else if (oclength > 0 && + eptr <= md->end_subject - oclength && + memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength; +#endif /* SUPPORT_UCP */ + else + { + CHECK_PARTIAL(); + break; + } + } + + if (possessive) continue; + + for(;;) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM23); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (eptr == pp) { RRETURN(MATCH_NOMATCH); } +#ifdef SUPPORT_UCP + eptr--; + BACKCHAR(eptr); +#else /* without SUPPORT_UCP */ + eptr -= length; +#endif /* SUPPORT_UCP */ + } + } + /* Control never gets here */ + } + + /* If the length of a UTF-8 character is 1, we fall through here, and + obey the code as for non-UTF-8 characters below, though in this case the + value of fc will always be < 128. */ + } + else +#endif /* SUPPORT_UTF */ + /* When not in UTF-8 mode, load a single-byte character. */ + fc = *ecode++; + + /* The value of fc at this point is always one character, though we may + or may not be in UTF mode. The code is duplicated for the caseless and + caseful cases, for speed, since matching characters is likely to be quite + common. First, ensure the minimum number of matches are present. If min = + max, continue at the same level without recursing. Otherwise, if + minimizing, keep trying the rest of the expression and advancing one + matching character if failing, up to the maximum. Alternatively, if + maximizing, find the maximum number of characters and work backwards. */ + + DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max, + max, (char *)eptr)); + + if (op >= OP_STARI) /* Caseless */ + { +#ifdef COMPILE_PCRE8 + /* fc must be < 128 if UTF is enabled. */ + foc = md->fcc[fc]; +#else +#ifdef SUPPORT_UTF +#ifdef SUPPORT_UCP + if (utf && fc > 127) + foc = UCD_OTHERCASE(fc); +#else + if (utf && fc > 127) + foc = fc; +#endif /* SUPPORT_UCP */ + else +#endif /* SUPPORT_UTF */ + foc = TABLE_GET(fc, md->fcc, fc); +#endif /* COMPILE_PCRE8 */ + + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH); + eptr++; + } + if (min == max) continue; + if (minimize) + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM24); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH); + eptr++; + } + /* Control never gets here */ + } + else /* Maximize */ + { + pp = eptr; + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (fc != *eptr && foc != *eptr) break; + eptr++; + } + + if (possessive) continue; + + while (eptr >= pp) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM25); + eptr--; + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + } + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + } + + /* Caseful comparisons (includes all multi-byte characters) */ + + else + { + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (fc != *eptr++) RRETURN(MATCH_NOMATCH); + } + + if (min == max) continue; + + if (minimize) + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM26); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (fc != *eptr++) RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + } + else /* Maximize */ + { + pp = eptr; + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (fc != *eptr) break; + eptr++; + } + if (possessive) continue; + + while (eptr >= pp) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM27); + eptr--; + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + } + RRETURN(MATCH_NOMATCH); + } + } + /* Control never gets here */ + + /* Match a negated single one-byte character. The character we are + checking can be multibyte. */ + + case OP_NOT: + case OP_NOTI: + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } +#ifdef SUPPORT_UTF + if (utf) + { + register unsigned int ch, och; + + ecode++; + GETCHARINC(ch, ecode); + GETCHARINC(c, eptr); + + if (op == OP_NOT) + { + if (ch == c) RRETURN(MATCH_NOMATCH); + } + else + { +#ifdef SUPPORT_UCP + if (ch > 127) + och = UCD_OTHERCASE(ch); +#else + if (ch > 127) + och = ch; +#endif /* SUPPORT_UCP */ + else + och = TABLE_GET(ch, md->fcc, ch); + if (ch == c || och == c) RRETURN(MATCH_NOMATCH); + } + } + else +#endif + { + register unsigned int ch = ecode[1]; + c = *eptr++; + if (ch == c || (op == OP_NOTI && TABLE_GET(ch, md->fcc, ch) == c)) + RRETURN(MATCH_NOMATCH); + ecode += 2; + } + break; + + /* Match a negated single one-byte character repeatedly. This is almost a + repeat of the code for a repeated single character, but I haven't found a + nice way of commoning these up that doesn't require a test of the + positive/negative option for each character match. Maybe that wouldn't add + very much to the time taken, but character matching *is* what this is all + about... */ + + case OP_NOTEXACT: + case OP_NOTEXACTI: + min = max = GET2(ecode, 1); + ecode += 1 + IMM2_SIZE; + goto REPEATNOTCHAR; + + case OP_NOTUPTO: + case OP_NOTUPTOI: + case OP_NOTMINUPTO: + case OP_NOTMINUPTOI: + min = 0; + max = GET2(ecode, 1); + minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI; + ecode += 1 + IMM2_SIZE; + goto REPEATNOTCHAR; + + case OP_NOTPOSSTAR: + case OP_NOTPOSSTARI: + possessive = TRUE; + min = 0; + max = INT_MAX; + ecode++; + goto REPEATNOTCHAR; + + case OP_NOTPOSPLUS: + case OP_NOTPOSPLUSI: + possessive = TRUE; + min = 1; + max = INT_MAX; + ecode++; + goto REPEATNOTCHAR; + + case OP_NOTPOSQUERY: + case OP_NOTPOSQUERYI: + possessive = TRUE; + min = 0; + max = 1; + ecode++; + goto REPEATNOTCHAR; + + case OP_NOTPOSUPTO: + case OP_NOTPOSUPTOI: + possessive = TRUE; + min = 0; + max = GET2(ecode, 1); + ecode += 1 + IMM2_SIZE; + goto REPEATNOTCHAR; + + case OP_NOTSTAR: + case OP_NOTSTARI: + case OP_NOTMINSTAR: + case OP_NOTMINSTARI: + case OP_NOTPLUS: + case OP_NOTPLUSI: + case OP_NOTMINPLUS: + case OP_NOTMINPLUSI: + case OP_NOTQUERY: + case OP_NOTQUERYI: + case OP_NOTMINQUERY: + case OP_NOTMINQUERYI: + c = *ecode++ - ((op >= OP_NOTSTARI)? OP_NOTSTARI: OP_NOTSTAR); + minimize = (c & 1) != 0; + min = rep_min[c]; /* Pick up values from tables; */ + max = rep_max[c]; /* zero for max => infinity */ + if (max == 0) max = INT_MAX; + + /* Common code for all repeated single-byte matches. */ + + REPEATNOTCHAR: + GETCHARINCTEST(fc, ecode); + + /* The code is duplicated for the caseless and caseful cases, for speed, + since matching characters is likely to be quite common. First, ensure the + minimum number of matches are present. If min = max, continue at the same + level without recursing. Otherwise, if minimizing, keep trying the rest of + the expression and advancing one matching character if failing, up to the + maximum. Alternatively, if maximizing, find the maximum number of + characters and work backwards. */ + + DPRINTF(("negative matching %c{%d,%d} against subject %.*s\n", fc, min, max, + max, (char *)eptr)); + + if (op >= OP_NOTSTARI) /* Caseless */ + { +#ifdef SUPPORT_UTF +#ifdef SUPPORT_UCP + if (utf && fc > 127) + foc = UCD_OTHERCASE(fc); +#else + if (utf && fc > 127) + foc = fc; +#endif /* SUPPORT_UCP */ + else +#endif /* SUPPORT_UTF */ + foc = TABLE_GET(fc, md->fcc, fc); + +#ifdef SUPPORT_UTF + if (utf) + { + register unsigned int d; + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(d, eptr); + if (fc == d || (unsigned int)foc == d) RRETURN(MATCH_NOMATCH); + } + } + else +#endif + /* Not UTF mode */ + { + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH); + eptr++; + } + } + + if (min == max) continue; + + if (minimize) + { +#ifdef SUPPORT_UTF + if (utf) + { + register unsigned int d; + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM28); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(d, eptr); + if (fc == d || (unsigned int)foc == d) RRETURN(MATCH_NOMATCH); + } + } + else +#endif + /* Not UTF mode */ + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM29); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH); + eptr++; + } + } + /* Control never gets here */ + } + + /* Maximize case */ + + else + { + pp = eptr; + +#ifdef SUPPORT_UTF + if (utf) + { + register unsigned int d; + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(d, eptr, len); + if (fc == d || (unsigned int)foc == d) break; + eptr += len; + } + if (possessive) continue; + for(;;) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM30); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (eptr-- == pp) break; /* Stop if tried at original pos */ + BACKCHAR(eptr); + } + } + else +#endif + /* Not UTF mode */ + { + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (fc == *eptr || foc == *eptr) break; + eptr++; + } + if (possessive) continue; + while (eptr >= pp) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM31); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + eptr--; + } + } + + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + } + + /* Caseful comparisons */ + + else + { +#ifdef SUPPORT_UTF + if (utf) + { + register unsigned int d; + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(d, eptr); + if (fc == d) RRETURN(MATCH_NOMATCH); + } + } + else +#endif + /* Not UTF mode */ + { + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (fc == *eptr++) RRETURN(MATCH_NOMATCH); + } + } + + if (min == max) continue; + + if (minimize) + { +#ifdef SUPPORT_UTF + if (utf) + { + register unsigned int d; + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM32); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(d, eptr); + if (fc == d) RRETURN(MATCH_NOMATCH); + } + } + else +#endif + /* Not UTF mode */ + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM33); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (fc == *eptr++) RRETURN(MATCH_NOMATCH); + } + } + /* Control never gets here */ + } + + /* Maximize case */ + + else + { + pp = eptr; + +#ifdef SUPPORT_UTF + if (utf) + { + register unsigned int d; + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(d, eptr, len); + if (fc == d) break; + eptr += len; + } + if (possessive) continue; + for(;;) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM34); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (eptr-- == pp) break; /* Stop if tried at original pos */ + BACKCHAR(eptr); + } + } + else +#endif + /* Not UTF mode */ + { + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (fc == *eptr) break; + eptr++; + } + if (possessive) continue; + while (eptr >= pp) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM35); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + eptr--; + } + } + + RRETURN(MATCH_NOMATCH); + } + } + /* Control never gets here */ + + /* Match a single character type repeatedly; several different opcodes + share code. This is very similar to the code for single characters, but we + repeat it in the interests of efficiency. */ + + case OP_TYPEEXACT: + min = max = GET2(ecode, 1); + minimize = TRUE; + ecode += 1 + IMM2_SIZE; + goto REPEATTYPE; + + case OP_TYPEUPTO: + case OP_TYPEMINUPTO: + min = 0; + max = GET2(ecode, 1); + minimize = *ecode == OP_TYPEMINUPTO; + ecode += 1 + IMM2_SIZE; + goto REPEATTYPE; + + case OP_TYPEPOSSTAR: + possessive = TRUE; + min = 0; + max = INT_MAX; + ecode++; + goto REPEATTYPE; + + case OP_TYPEPOSPLUS: + possessive = TRUE; + min = 1; + max = INT_MAX; + ecode++; + goto REPEATTYPE; + + case OP_TYPEPOSQUERY: + possessive = TRUE; + min = 0; + max = 1; + ecode++; + goto REPEATTYPE; + + case OP_TYPEPOSUPTO: + possessive = TRUE; + min = 0; + max = GET2(ecode, 1); + ecode += 1 + IMM2_SIZE; + goto REPEATTYPE; + + case OP_TYPESTAR: + case OP_TYPEMINSTAR: + case OP_TYPEPLUS: + case OP_TYPEMINPLUS: + case OP_TYPEQUERY: + case OP_TYPEMINQUERY: + c = *ecode++ - OP_TYPESTAR; + minimize = (c & 1) != 0; + min = rep_min[c]; /* Pick up values from tables; */ + max = rep_max[c]; /* zero for max => infinity */ + if (max == 0) max = INT_MAX; + + /* Common code for all repeated single character type matches. Note that + in UTF-8 mode, '.' matches a character of any length, but for the other + character types, the valid characters are all one-byte long. */ + + REPEATTYPE: + ctype = *ecode++; /* Code for the character type */ + +#ifdef SUPPORT_UCP + if (ctype == OP_PROP || ctype == OP_NOTPROP) + { + prop_fail_result = ctype == OP_NOTPROP; + prop_type = *ecode++; + prop_value = *ecode++; + } + else prop_type = -1; +#endif + + /* First, ensure the minimum number of matches are present. Use inline + code for maximizing the speed, and do the type test once at the start + (i.e. keep it out of the loop). Separate the UTF-8 code completely as that + is tidier. Also separate the UCP code, which can be the same for both UTF-8 + and single-bytes. */ + + if (min > 0) + { +#ifdef SUPPORT_UCP + if (prop_type >= 0) + { + switch(prop_type) + { + case PT_ANY: + if (prop_fail_result) RRETURN(MATCH_NOMATCH); + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + } + break; + + case PT_LAMP: + for (i = 1; i <= min; i++) + { + int chartype; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + chartype = UCD_CHARTYPE(c); + if ((chartype == ucp_Lu || + chartype == ucp_Ll || + chartype == ucp_Lt) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + case PT_GC: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + case PT_PC: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + case PT_SC: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + case PT_ALNUM: + for (i = 1; i <= min; i++) + { + int category; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + category = UCD_CATEGORY(c); + if ((category == ucp_L || category == ucp_N) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + case PT_SPACE: /* Perl space */ + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || + c == CHAR_FF || c == CHAR_CR) + == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + case PT_PXSPACE: /* POSIX space */ + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || + c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) + == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + case PT_WORD: + for (i = 1; i <= min; i++) + { + int category; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + category = UCD_CATEGORY(c); + if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE) + == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + break; + + /* This should not occur */ + + default: + RRETURN(PCRE_ERROR_INTERNAL); + } + } + + /* Match extended Unicode sequences. We will get here only if the + support is in the binary; otherwise a compile-time error occurs. */ + + else if (ctype == OP_EXTUNI) + { + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if (UCD_CATEGORY(c) == ucp_M) RRETURN(MATCH_NOMATCH); + while (eptr < md->end_subject) + { + int len = 1; + if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } + if (UCD_CATEGORY(c) != ucp_M) break; + eptr += len; + } + CHECK_PARTIAL(); + } + } + + else +#endif /* SUPPORT_UCP */ + +/* Handle all other cases when the coding is UTF-8 */ + +#ifdef SUPPORT_UTF + if (utf) switch(ctype) + { + case OP_ANY: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); + if (md->partial != 0 && + eptr + 1 >= md->end_subject && + NLBLOCK->nltype == NLTYPE_FIXED && + NLBLOCK->nllen == 2 && + *eptr == NLBLOCK->nl[0]) + { + md->hitend = TRUE; + if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); + } + eptr++; + ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); + } + break; + + case OP_ALLANY: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + eptr++; + ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); + } + break; + + case OP_ANYBYTE: + if (eptr > md->end_subject - min) RRETURN(MATCH_NOMATCH); + eptr += min; + break; + + case OP_ANYNL: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(c, eptr); + switch(c) + { + default: RRETURN(MATCH_NOMATCH); + + case 0x000d: + if (eptr < md->end_subject && *eptr == 0x0a) eptr++; + break; + + case 0x000a: + break; + + case 0x000b: + case 0x000c: + case 0x0085: + case 0x2028: + case 0x2029: + if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); + break; + } + } + break; + + case OP_NOT_HSPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(c, eptr); + switch(c) + { + default: break; + case 0x09: /* HT */ + case 0x20: /* SPACE */ + case 0xa0: /* NBSP */ + case 0x1680: /* OGHAM SPACE MARK */ + case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ + case 0x2000: /* EN QUAD */ + case 0x2001: /* EM QUAD */ + case 0x2002: /* EN SPACE */ + case 0x2003: /* EM SPACE */ + case 0x2004: /* THREE-PER-EM SPACE */ + case 0x2005: /* FOUR-PER-EM SPACE */ + case 0x2006: /* SIX-PER-EM SPACE */ + case 0x2007: /* FIGURE SPACE */ + case 0x2008: /* PUNCTUATION SPACE */ + case 0x2009: /* THIN SPACE */ + case 0x200A: /* HAIR SPACE */ + case 0x202f: /* NARROW NO-BREAK SPACE */ + case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ + case 0x3000: /* IDEOGRAPHIC SPACE */ + RRETURN(MATCH_NOMATCH); + } + } + break; + + case OP_HSPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(c, eptr); + switch(c) + { + default: RRETURN(MATCH_NOMATCH); + case 0x09: /* HT */ + case 0x20: /* SPACE */ + case 0xa0: /* NBSP */ + case 0x1680: /* OGHAM SPACE MARK */ + case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ + case 0x2000: /* EN QUAD */ + case 0x2001: /* EM QUAD */ + case 0x2002: /* EN SPACE */ + case 0x2003: /* EM SPACE */ + case 0x2004: /* THREE-PER-EM SPACE */ + case 0x2005: /* FOUR-PER-EM SPACE */ + case 0x2006: /* SIX-PER-EM SPACE */ + case 0x2007: /* FIGURE SPACE */ + case 0x2008: /* PUNCTUATION SPACE */ + case 0x2009: /* THIN SPACE */ + case 0x200A: /* HAIR SPACE */ + case 0x202f: /* NARROW NO-BREAK SPACE */ + case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ + case 0x3000: /* IDEOGRAPHIC SPACE */ + break; + } + } + break; + + case OP_NOT_VSPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(c, eptr); + switch(c) + { + default: break; + case 0x0a: /* LF */ + case 0x0b: /* VT */ + case 0x0c: /* FF */ + case 0x0d: /* CR */ + case 0x85: /* NEL */ + case 0x2028: /* LINE SEPARATOR */ + case 0x2029: /* PARAGRAPH SEPARATOR */ + RRETURN(MATCH_NOMATCH); + } + } + break; + + case OP_VSPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(c, eptr); + switch(c) + { + default: RRETURN(MATCH_NOMATCH); + case 0x0a: /* LF */ + case 0x0b: /* VT */ + case 0x0c: /* FF */ + case 0x0d: /* CR */ + case 0x85: /* NEL */ + case 0x2028: /* LINE SEPARATOR */ + case 0x2029: /* PARAGRAPH SEPARATOR */ + break; + } + } + break; + + case OP_NOT_DIGIT: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINC(c, eptr); + if (c < 128 && (md->ctypes[c] & ctype_digit) != 0) + RRETURN(MATCH_NOMATCH); + } + break; + + case OP_DIGIT: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (*eptr >= 128 || (md->ctypes[*eptr] & ctype_digit) == 0) + RRETURN(MATCH_NOMATCH); + eptr++; + /* No need to skip more bytes - we know it's a 1-byte character */ + } + break; + + case OP_NOT_WHITESPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0) + RRETURN(MATCH_NOMATCH); + eptr++; + ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); + } + break; + + case OP_WHITESPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (*eptr >= 128 || (md->ctypes[*eptr] & ctype_space) == 0) + RRETURN(MATCH_NOMATCH); + eptr++; + /* No need to skip more bytes - we know it's a 1-byte character */ + } + break; + + case OP_NOT_WORDCHAR: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0) + RRETURN(MATCH_NOMATCH); + eptr++; + ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); + } + break; + + case OP_WORDCHAR: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (*eptr >= 128 || (md->ctypes[*eptr] & ctype_word) == 0) + RRETURN(MATCH_NOMATCH); + eptr++; + /* No need to skip more bytes - we know it's a 1-byte character */ + } + break; + + default: + RRETURN(PCRE_ERROR_INTERNAL); + } /* End switch(ctype) */ + + else +#endif /* SUPPORT_UTF */ + + /* Code for the non-UTF-8 case for minimum matching of operators other + than OP_PROP and OP_NOTPROP. */ + + switch(ctype) + { + case OP_ANY: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); + if (md->partial != 0 && + eptr + 1 >= md->end_subject && + NLBLOCK->nltype == NLTYPE_FIXED && + NLBLOCK->nllen == 2 && + *eptr == NLBLOCK->nl[0]) + { + md->hitend = TRUE; + if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); + } + eptr++; + } + break; + + case OP_ALLANY: + if (eptr > md->end_subject - min) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + eptr += min; + break; + + case OP_ANYBYTE: + if (eptr > md->end_subject - min) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + eptr += min; + break; + + case OP_ANYNL: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + switch(*eptr++) + { + default: RRETURN(MATCH_NOMATCH); + + case 0x000d: + if (eptr < md->end_subject && *eptr == 0x0a) eptr++; + break; + + case 0x000a: + break; + + case 0x000b: + case 0x000c: + case 0x0085: +#ifdef COMPILE_PCRE16 + case 0x2028: + case 0x2029: +#endif + if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); + break; + } + } + break; + + case OP_NOT_HSPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + switch(*eptr++) + { + default: break; + case 0x09: /* HT */ + case 0x20: /* SPACE */ + case 0xa0: /* NBSP */ +#ifdef COMPILE_PCRE16 + case 0x1680: /* OGHAM SPACE MARK */ + case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ + case 0x2000: /* EN QUAD */ + case 0x2001: /* EM QUAD */ + case 0x2002: /* EN SPACE */ + case 0x2003: /* EM SPACE */ + case 0x2004: /* THREE-PER-EM SPACE */ + case 0x2005: /* FOUR-PER-EM SPACE */ + case 0x2006: /* SIX-PER-EM SPACE */ + case 0x2007: /* FIGURE SPACE */ + case 0x2008: /* PUNCTUATION SPACE */ + case 0x2009: /* THIN SPACE */ + case 0x200A: /* HAIR SPACE */ + case 0x202f: /* NARROW NO-BREAK SPACE */ + case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ + case 0x3000: /* IDEOGRAPHIC SPACE */ +#endif + RRETURN(MATCH_NOMATCH); + } + } + break; + + case OP_HSPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + switch(*eptr++) + { + default: RRETURN(MATCH_NOMATCH); + case 0x09: /* HT */ + case 0x20: /* SPACE */ + case 0xa0: /* NBSP */ +#ifdef COMPILE_PCRE16 + case 0x1680: /* OGHAM SPACE MARK */ + case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ + case 0x2000: /* EN QUAD */ + case 0x2001: /* EM QUAD */ + case 0x2002: /* EN SPACE */ + case 0x2003: /* EM SPACE */ + case 0x2004: /* THREE-PER-EM SPACE */ + case 0x2005: /* FOUR-PER-EM SPACE */ + case 0x2006: /* SIX-PER-EM SPACE */ + case 0x2007: /* FIGURE SPACE */ + case 0x2008: /* PUNCTUATION SPACE */ + case 0x2009: /* THIN SPACE */ + case 0x200A: /* HAIR SPACE */ + case 0x202f: /* NARROW NO-BREAK SPACE */ + case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ + case 0x3000: /* IDEOGRAPHIC SPACE */ +#endif + break; + } + } + break; + + case OP_NOT_VSPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + switch(*eptr++) + { + default: break; + case 0x0a: /* LF */ + case 0x0b: /* VT */ + case 0x0c: /* FF */ + case 0x0d: /* CR */ + case 0x85: /* NEL */ +#ifdef COMPILE_PCRE16 + case 0x2028: /* LINE SEPARATOR */ + case 0x2029: /* PARAGRAPH SEPARATOR */ +#endif + RRETURN(MATCH_NOMATCH); + } + } + break; + + case OP_VSPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + switch(*eptr++) + { + default: RRETURN(MATCH_NOMATCH); + case 0x0a: /* LF */ + case 0x0b: /* VT */ + case 0x0c: /* FF */ + case 0x0d: /* CR */ + case 0x85: /* NEL */ +#ifdef COMPILE_PCRE16 + case 0x2028: /* LINE SEPARATOR */ + case 0x2029: /* PARAGRAPH SEPARATOR */ +#endif + break; + } + } + break; + + case OP_NOT_DIGIT: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_digit) != 0) + RRETURN(MATCH_NOMATCH); + eptr++; + } + break; + + case OP_DIGIT: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_digit) == 0) + RRETURN(MATCH_NOMATCH); + eptr++; + } + break; + + case OP_NOT_WHITESPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_space) != 0) + RRETURN(MATCH_NOMATCH); + eptr++; + } + break; + + case OP_WHITESPACE: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_space) == 0) + RRETURN(MATCH_NOMATCH); + eptr++; + } + break; + + case OP_NOT_WORDCHAR: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_word) != 0) + RRETURN(MATCH_NOMATCH); + eptr++; + } + break; + + case OP_WORDCHAR: + for (i = 1; i <= min; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_word) == 0) + RRETURN(MATCH_NOMATCH); + eptr++; + } + break; + + default: + RRETURN(PCRE_ERROR_INTERNAL); + } + } + + /* If min = max, continue at the same level without recursing */ + + if (min == max) continue; + + /* If minimizing, we have to test the rest of the pattern before each + subsequent match. Again, separate the UTF-8 case for speed, and also + separate the UCP cases. */ + + if (minimize) + { +#ifdef SUPPORT_UCP + if (prop_type >= 0) + { + switch(prop_type) + { + case PT_ANY: + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM36); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if (prop_fail_result) RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + case PT_LAMP: + for (fi = min;; fi++) + { + int chartype; + RMATCH(eptr, ecode, offset_top, md, eptrb, RM37); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + chartype = UCD_CHARTYPE(c); + if ((chartype == ucp_Lu || + chartype == ucp_Ll || + chartype == ucp_Lt) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + case PT_GC: + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM38); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + case PT_PC: + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM39); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + case PT_SC: + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM40); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + case PT_ALNUM: + for (fi = min;; fi++) + { + int category; + RMATCH(eptr, ecode, offset_top, md, eptrb, RM59); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + category = UCD_CATEGORY(c); + if ((category == ucp_L || category == ucp_N) == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + case PT_SPACE: /* Perl space */ + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM60); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || + c == CHAR_FF || c == CHAR_CR) + == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + case PT_PXSPACE: /* POSIX space */ + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM61); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || + c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) + == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + case PT_WORD: + for (fi = min;; fi++) + { + int category; + RMATCH(eptr, ecode, offset_top, md, eptrb, RM62); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + category = UCD_CATEGORY(c); + if ((category == ucp_L || + category == ucp_N || + c == CHAR_UNDERSCORE) + == prop_fail_result) + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + /* This should never occur */ + + default: + RRETURN(PCRE_ERROR_INTERNAL); + } + } + + /* Match extended Unicode sequences. We will get here only if the + support is in the binary; otherwise a compile-time error occurs. */ + + else if (ctype == OP_EXTUNI) + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM41); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + GETCHARINCTEST(c, eptr); + if (UCD_CATEGORY(c) == ucp_M) RRETURN(MATCH_NOMATCH); + while (eptr < md->end_subject) + { + int len = 1; + if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } + if (UCD_CATEGORY(c) != ucp_M) break; + eptr += len; + } + CHECK_PARTIAL(); + } + } + else +#endif /* SUPPORT_UCP */ + +#ifdef SUPPORT_UTF + if (utf) + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM42); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (ctype == OP_ANY && IS_NEWLINE(eptr)) + RRETURN(MATCH_NOMATCH); + GETCHARINC(c, eptr); + switch(ctype) + { + case OP_ANY: /* This is the non-NL case */ + if (md->partial != 0 && /* Take care with CRLF partial */ + eptr >= md->end_subject && + NLBLOCK->nltype == NLTYPE_FIXED && + NLBLOCK->nllen == 2 && + c == NLBLOCK->nl[0]) + { + md->hitend = TRUE; + if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); + } + break; + + case OP_ALLANY: + case OP_ANYBYTE: + break; + + case OP_ANYNL: + switch(c) + { + default: RRETURN(MATCH_NOMATCH); + case 0x000d: + if (eptr < md->end_subject && *eptr == 0x0a) eptr++; + break; + case 0x000a: + break; + + case 0x000b: + case 0x000c: + case 0x0085: + case 0x2028: + case 0x2029: + if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); + break; + } + break; + + case OP_NOT_HSPACE: + switch(c) + { + default: break; + case 0x09: /* HT */ + case 0x20: /* SPACE */ + case 0xa0: /* NBSP */ + case 0x1680: /* OGHAM SPACE MARK */ + case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ + case 0x2000: /* EN QUAD */ + case 0x2001: /* EM QUAD */ + case 0x2002: /* EN SPACE */ + case 0x2003: /* EM SPACE */ + case 0x2004: /* THREE-PER-EM SPACE */ + case 0x2005: /* FOUR-PER-EM SPACE */ + case 0x2006: /* SIX-PER-EM SPACE */ + case 0x2007: /* FIGURE SPACE */ + case 0x2008: /* PUNCTUATION SPACE */ + case 0x2009: /* THIN SPACE */ + case 0x200A: /* HAIR SPACE */ + case 0x202f: /* NARROW NO-BREAK SPACE */ + case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ + case 0x3000: /* IDEOGRAPHIC SPACE */ + RRETURN(MATCH_NOMATCH); + } + break; + + case OP_HSPACE: + switch(c) + { + default: RRETURN(MATCH_NOMATCH); + case 0x09: /* HT */ + case 0x20: /* SPACE */ + case 0xa0: /* NBSP */ + case 0x1680: /* OGHAM SPACE MARK */ + case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ + case 0x2000: /* EN QUAD */ + case 0x2001: /* EM QUAD */ + case 0x2002: /* EN SPACE */ + case 0x2003: /* EM SPACE */ + case 0x2004: /* THREE-PER-EM SPACE */ + case 0x2005: /* FOUR-PER-EM SPACE */ + case 0x2006: /* SIX-PER-EM SPACE */ + case 0x2007: /* FIGURE SPACE */ + case 0x2008: /* PUNCTUATION SPACE */ + case 0x2009: /* THIN SPACE */ + case 0x200A: /* HAIR SPACE */ + case 0x202f: /* NARROW NO-BREAK SPACE */ + case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ + case 0x3000: /* IDEOGRAPHIC SPACE */ + break; + } + break; + + case OP_NOT_VSPACE: + switch(c) + { + default: break; + case 0x0a: /* LF */ + case 0x0b: /* VT */ + case 0x0c: /* FF */ + case 0x0d: /* CR */ + case 0x85: /* NEL */ + case 0x2028: /* LINE SEPARATOR */ + case 0x2029: /* PARAGRAPH SEPARATOR */ + RRETURN(MATCH_NOMATCH); + } + break; + + case OP_VSPACE: + switch(c) + { + default: RRETURN(MATCH_NOMATCH); + case 0x0a: /* LF */ + case 0x0b: /* VT */ + case 0x0c: /* FF */ + case 0x0d: /* CR */ + case 0x85: /* NEL */ + case 0x2028: /* LINE SEPARATOR */ + case 0x2029: /* PARAGRAPH SEPARATOR */ + break; + } + break; + + case OP_NOT_DIGIT: + if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) + RRETURN(MATCH_NOMATCH); + break; + + case OP_DIGIT: + if (c >= 256 || (md->ctypes[c] & ctype_digit) == 0) + RRETURN(MATCH_NOMATCH); + break; + + case OP_NOT_WHITESPACE: + if (c < 256 && (md->ctypes[c] & ctype_space) != 0) + RRETURN(MATCH_NOMATCH); + break; + + case OP_WHITESPACE: + if (c >= 256 || (md->ctypes[c] & ctype_space) == 0) + RRETURN(MATCH_NOMATCH); + break; + + case OP_NOT_WORDCHAR: + if (c < 256 && (md->ctypes[c] & ctype_word) != 0) + RRETURN(MATCH_NOMATCH); + break; + + case OP_WORDCHAR: + if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) + RRETURN(MATCH_NOMATCH); + break; + + default: + RRETURN(PCRE_ERROR_INTERNAL); + } + } + } + else +#endif + /* Not UTF mode */ + { + for (fi = min;; fi++) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM43); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (fi >= max) RRETURN(MATCH_NOMATCH); + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + RRETURN(MATCH_NOMATCH); + } + if (ctype == OP_ANY && IS_NEWLINE(eptr)) + RRETURN(MATCH_NOMATCH); + c = *eptr++; + switch(ctype) + { + case OP_ANY: /* This is the non-NL case */ + if (md->partial != 0 && /* Take care with CRLF partial */ + eptr >= md->end_subject && + NLBLOCK->nltype == NLTYPE_FIXED && + NLBLOCK->nllen == 2 && + c == NLBLOCK->nl[0]) + { + md->hitend = TRUE; + if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); + } + break; + + case OP_ALLANY: + case OP_ANYBYTE: + break; + + case OP_ANYNL: + switch(c) + { + default: RRETURN(MATCH_NOMATCH); + case 0x000d: + if (eptr < md->end_subject && *eptr == 0x0a) eptr++; + break; + + case 0x000a: + break; + + case 0x000b: + case 0x000c: + case 0x0085: +#ifdef COMPILE_PCRE16 + case 0x2028: + case 0x2029: +#endif + if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); + break; + } + break; + + case OP_NOT_HSPACE: + switch(c) + { + default: break; + case 0x09: /* HT */ + case 0x20: /* SPACE */ + case 0xa0: /* NBSP */ +#ifdef COMPILE_PCRE16 + case 0x1680: /* OGHAM SPACE MARK */ + case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ + case 0x2000: /* EN QUAD */ + case 0x2001: /* EM QUAD */ + case 0x2002: /* EN SPACE */ + case 0x2003: /* EM SPACE */ + case 0x2004: /* THREE-PER-EM SPACE */ + case 0x2005: /* FOUR-PER-EM SPACE */ + case 0x2006: /* SIX-PER-EM SPACE */ + case 0x2007: /* FIGURE SPACE */ + case 0x2008: /* PUNCTUATION SPACE */ + case 0x2009: /* THIN SPACE */ + case 0x200A: /* HAIR SPACE */ + case 0x202f: /* NARROW NO-BREAK SPACE */ + case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ + case 0x3000: /* IDEOGRAPHIC SPACE */ +#endif + RRETURN(MATCH_NOMATCH); + } + break; + + case OP_HSPACE: + switch(c) + { + default: RRETURN(MATCH_NOMATCH); + case 0x09: /* HT */ + case 0x20: /* SPACE */ + case 0xa0: /* NBSP */ +#ifdef COMPILE_PCRE16 + case 0x1680: /* OGHAM SPACE MARK */ + case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ + case 0x2000: /* EN QUAD */ + case 0x2001: /* EM QUAD */ + case 0x2002: /* EN SPACE */ + case 0x2003: /* EM SPACE */ + case 0x2004: /* THREE-PER-EM SPACE */ + case 0x2005: /* FOUR-PER-EM SPACE */ + case 0x2006: /* SIX-PER-EM SPACE */ + case 0x2007: /* FIGURE SPACE */ + case 0x2008: /* PUNCTUATION SPACE */ + case 0x2009: /* THIN SPACE */ + case 0x200A: /* HAIR SPACE */ + case 0x202f: /* NARROW NO-BREAK SPACE */ + case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ + case 0x3000: /* IDEOGRAPHIC SPACE */ +#endif + break; + } + break; + + case OP_NOT_VSPACE: + switch(c) + { + default: break; + case 0x0a: /* LF */ + case 0x0b: /* VT */ + case 0x0c: /* FF */ + case 0x0d: /* CR */ + case 0x85: /* NEL */ +#ifdef COMPILE_PCRE16 + case 0x2028: /* LINE SEPARATOR */ + case 0x2029: /* PARAGRAPH SEPARATOR */ +#endif + RRETURN(MATCH_NOMATCH); + } + break; + + case OP_VSPACE: + switch(c) + { + default: RRETURN(MATCH_NOMATCH); + case 0x0a: /* LF */ + case 0x0b: /* VT */ + case 0x0c: /* FF */ + case 0x0d: /* CR */ + case 0x85: /* NEL */ +#ifdef COMPILE_PCRE16 + case 0x2028: /* LINE SEPARATOR */ + case 0x2029: /* PARAGRAPH SEPARATOR */ +#endif + break; + } + break; + + case OP_NOT_DIGIT: + if (MAX_255(c) && (md->ctypes[c] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH); + break; + + case OP_DIGIT: + if (!MAX_255(c) || (md->ctypes[c] & ctype_digit) == 0) RRETURN(MATCH_NOMATCH); + break; + + case OP_NOT_WHITESPACE: + if (MAX_255(c) && (md->ctypes[c] & ctype_space) != 0) RRETURN(MATCH_NOMATCH); + break; + + case OP_WHITESPACE: + if (!MAX_255(c) || (md->ctypes[c] & ctype_space) == 0) RRETURN(MATCH_NOMATCH); + break; + + case OP_NOT_WORDCHAR: + if (MAX_255(c) && (md->ctypes[c] & ctype_word) != 0) RRETURN(MATCH_NOMATCH); + break; + + case OP_WORDCHAR: + if (!MAX_255(c) || (md->ctypes[c] & ctype_word) == 0) RRETURN(MATCH_NOMATCH); + break; + + default: + RRETURN(PCRE_ERROR_INTERNAL); + } + } + } + /* Control never gets here */ + } + + /* If maximizing, it is worth using inline code for speed, doing the type + test once at the start (i.e. keep it out of the loop). Again, keep the + UTF-8 and UCP stuff separate. */ + + else + { + pp = eptr; /* Remember where we started */ + +#ifdef SUPPORT_UCP + if (prop_type >= 0) + { + switch(prop_type) + { + case PT_ANY: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLENTEST(c, eptr, len); + if (prop_fail_result) break; + eptr+= len; + } + break; + + case PT_LAMP: + for (i = min; i < max; i++) + { + int chartype; + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLENTEST(c, eptr, len); + chartype = UCD_CHARTYPE(c); + if ((chartype == ucp_Lu || + chartype == ucp_Ll || + chartype == ucp_Lt) == prop_fail_result) + break; + eptr+= len; + } + break; + + case PT_GC: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLENTEST(c, eptr, len); + if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) break; + eptr+= len; + } + break; + + case PT_PC: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLENTEST(c, eptr, len); + if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) break; + eptr+= len; + } + break; + + case PT_SC: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLENTEST(c, eptr, len); + if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) break; + eptr+= len; + } + break; + + case PT_ALNUM: + for (i = min; i < max; i++) + { + int category; + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLENTEST(c, eptr, len); + category = UCD_CATEGORY(c); + if ((category == ucp_L || category == ucp_N) == prop_fail_result) + break; + eptr+= len; + } + break; + + case PT_SPACE: /* Perl space */ + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLENTEST(c, eptr, len); + if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || + c == CHAR_FF || c == CHAR_CR) + == prop_fail_result) + break; + eptr+= len; + } + break; + + case PT_PXSPACE: /* POSIX space */ + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLENTEST(c, eptr, len); + if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || + c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) + == prop_fail_result) + break; + eptr+= len; + } + break; + + case PT_WORD: + for (i = min; i < max; i++) + { + int category; + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLENTEST(c, eptr, len); + category = UCD_CATEGORY(c); + if ((category == ucp_L || category == ucp_N || + c == CHAR_UNDERSCORE) == prop_fail_result) + break; + eptr+= len; + } + break; + + default: + RRETURN(PCRE_ERROR_INTERNAL); + } + + /* eptr is now past the end of the maximum run */ + + if (possessive) continue; + for(;;) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM44); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (eptr-- == pp) break; /* Stop if tried at original pos */ + if (utf) BACKCHAR(eptr); + } + } + + /* Match extended Unicode sequences. We will get here only if the + support is in the binary; otherwise a compile-time error occurs. */ + + else if (ctype == OP_EXTUNI) + { + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } + if (UCD_CATEGORY(c) == ucp_M) break; + eptr += len; + while (eptr < md->end_subject) + { + len = 1; + if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } + if (UCD_CATEGORY(c) != ucp_M) break; + eptr += len; + } + CHECK_PARTIAL(); + } + + /* eptr is now past the end of the maximum run */ + + if (possessive) continue; + + for(;;) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM45); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (eptr-- == pp) break; /* Stop if tried at original pos */ + for (;;) /* Move back over one extended */ + { + if (!utf) c = *eptr; else + { + BACKCHAR(eptr); + GETCHAR(c, eptr); + } + if (UCD_CATEGORY(c) != ucp_M) break; + eptr--; + } + } + } + + else +#endif /* SUPPORT_UCP */ + +#ifdef SUPPORT_UTF + if (utf) + { + switch(ctype) + { + case OP_ANY: + if (max < INT_MAX) + { + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (IS_NEWLINE(eptr)) break; + if (md->partial != 0 && /* Take care with CRLF partial */ + eptr + 1 >= md->end_subject && + NLBLOCK->nltype == NLTYPE_FIXED && + NLBLOCK->nllen == 2 && + *eptr == NLBLOCK->nl[0]) + { + md->hitend = TRUE; + if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); + } + eptr++; + ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); + } + } + + /* Handle unlimited UTF-8 repeat */ + + else + { + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (IS_NEWLINE(eptr)) break; + if (md->partial != 0 && /* Take care with CRLF partial */ + eptr + 1 >= md->end_subject && + NLBLOCK->nltype == NLTYPE_FIXED && + NLBLOCK->nllen == 2 && + *eptr == NLBLOCK->nl[0]) + { + md->hitend = TRUE; + if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); + } + eptr++; + ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); + } + } + break; + + case OP_ALLANY: + if (max < INT_MAX) + { + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + eptr++; + ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); + } + } + else + { + eptr = md->end_subject; /* Unlimited UTF-8 repeat */ + SCHECK_PARTIAL(); + } + break; + + /* The byte case is the same as non-UTF8 */ + + case OP_ANYBYTE: + c = max - min; + if (c > (unsigned int)(md->end_subject - eptr)) + { + eptr = md->end_subject; + SCHECK_PARTIAL(); + } + else eptr += c; + break; + + case OP_ANYNL: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(c, eptr, len); + if (c == 0x000d) + { + if (++eptr >= md->end_subject) break; + if (*eptr == 0x000a) eptr++; + } + else + { + if (c != 0x000a && + (md->bsr_anycrlf || + (c != 0x000b && c != 0x000c && + c != 0x0085 && c != 0x2028 && c != 0x2029))) + break; + eptr += len; + } + } + break; + + case OP_NOT_HSPACE: + case OP_HSPACE: + for (i = min; i < max; i++) + { + BOOL gotspace; + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(c, eptr, len); + switch(c) + { + default: gotspace = FALSE; break; + case 0x09: /* HT */ + case 0x20: /* SPACE */ + case 0xa0: /* NBSP */ + case 0x1680: /* OGHAM SPACE MARK */ + case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ + case 0x2000: /* EN QUAD */ + case 0x2001: /* EM QUAD */ + case 0x2002: /* EN SPACE */ + case 0x2003: /* EM SPACE */ + case 0x2004: /* THREE-PER-EM SPACE */ + case 0x2005: /* FOUR-PER-EM SPACE */ + case 0x2006: /* SIX-PER-EM SPACE */ + case 0x2007: /* FIGURE SPACE */ + case 0x2008: /* PUNCTUATION SPACE */ + case 0x2009: /* THIN SPACE */ + case 0x200A: /* HAIR SPACE */ + case 0x202f: /* NARROW NO-BREAK SPACE */ + case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ + case 0x3000: /* IDEOGRAPHIC SPACE */ + gotspace = TRUE; + break; + } + if (gotspace == (ctype == OP_NOT_HSPACE)) break; + eptr += len; + } + break; + + case OP_NOT_VSPACE: + case OP_VSPACE: + for (i = min; i < max; i++) + { + BOOL gotspace; + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(c, eptr, len); + switch(c) + { + default: gotspace = FALSE; break; + case 0x0a: /* LF */ + case 0x0b: /* VT */ + case 0x0c: /* FF */ + case 0x0d: /* CR */ + case 0x85: /* NEL */ + case 0x2028: /* LINE SEPARATOR */ + case 0x2029: /* PARAGRAPH SEPARATOR */ + gotspace = TRUE; + break; + } + if (gotspace == (ctype == OP_NOT_VSPACE)) break; + eptr += len; + } + break; + + case OP_NOT_DIGIT: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(c, eptr, len); + if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) break; + eptr+= len; + } + break; + + case OP_DIGIT: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(c, eptr, len); + if (c >= 256 ||(md->ctypes[c] & ctype_digit) == 0) break; + eptr+= len; + } + break; + + case OP_NOT_WHITESPACE: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(c, eptr, len); + if (c < 256 && (md->ctypes[c] & ctype_space) != 0) break; + eptr+= len; + } + break; + + case OP_WHITESPACE: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(c, eptr, len); + if (c >= 256 ||(md->ctypes[c] & ctype_space) == 0) break; + eptr+= len; + } + break; + + case OP_NOT_WORDCHAR: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(c, eptr, len); + if (c < 256 && (md->ctypes[c] & ctype_word) != 0) break; + eptr+= len; + } + break; + + case OP_WORDCHAR: + for (i = min; i < max; i++) + { + int len = 1; + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + GETCHARLEN(c, eptr, len); + if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) break; + eptr+= len; + } + break; + + default: + RRETURN(PCRE_ERROR_INTERNAL); + } + + /* eptr is now past the end of the maximum run. If possessive, we are + done (no backing up). Otherwise, match at this position; anything other + than no match is immediately returned. For nomatch, back up one + character, unless we are matching \R and the last thing matched was + \r\n, in which case, back up two bytes. */ + + if (possessive) continue; + for(;;) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM46); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + if (eptr-- == pp) break; /* Stop if tried at original pos */ + BACKCHAR(eptr); + if (ctype == OP_ANYNL && eptr > pp && *eptr == '\n' && + eptr[-1] == '\r') eptr--; + } + } + else +#endif /* SUPPORT_UTF */ + /* Not UTF mode */ + { + switch(ctype) + { + case OP_ANY: + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (IS_NEWLINE(eptr)) break; + if (md->partial != 0 && /* Take care with CRLF partial */ + eptr + 1 >= md->end_subject && + NLBLOCK->nltype == NLTYPE_FIXED && + NLBLOCK->nllen == 2 && + *eptr == NLBLOCK->nl[0]) + { + md->hitend = TRUE; + if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); + } + eptr++; + } + break; + + case OP_ALLANY: + case OP_ANYBYTE: + c = max - min; + if (c > (unsigned int)(md->end_subject - eptr)) + { + eptr = md->end_subject; + SCHECK_PARTIAL(); + } + else eptr += c; + break; + + case OP_ANYNL: + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + c = *eptr; + if (c == 0x000d) + { + if (++eptr >= md->end_subject) break; + if (*eptr == 0x000a) eptr++; + } + else + { + if (c != 0x000a && (md->bsr_anycrlf || + (c != 0x000b && c != 0x000c && c != 0x0085 +#ifdef COMPILE_PCRE16 + && c != 0x2028 && c != 0x2029 +#endif + ))) break; + eptr++; + } + } + break; + + case OP_NOT_HSPACE: + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + c = *eptr; + if (c == 0x09 || c == 0x20 || c == 0xa0 +#ifdef COMPILE_PCRE16 + || c == 0x1680 || c == 0x180e || (c >= 0x2000 && c <= 0x200A) + || c == 0x202f || c == 0x205f || c == 0x3000 +#endif + ) break; + eptr++; + } + break; + + case OP_HSPACE: + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + c = *eptr; + if (c != 0x09 && c != 0x20 && c != 0xa0 +#ifdef COMPILE_PCRE16 + && c != 0x1680 && c != 0x180e && (c < 0x2000 || c > 0x200A) + && c != 0x202f && c != 0x205f && c != 0x3000 +#endif + ) break; + eptr++; + } + break; + + case OP_NOT_VSPACE: + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + c = *eptr; + if (c == 0x0a || c == 0x0b || c == 0x0c || c == 0x0d || c == 0x85 +#ifdef COMPILE_PCRE16 + || c == 0x2028 || c == 0x2029 +#endif + ) break; + eptr++; + } + break; + + case OP_VSPACE: + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + c = *eptr; + if (c != 0x0a && c != 0x0b && c != 0x0c && c != 0x0d && c != 0x85 +#ifdef COMPILE_PCRE16 + && c != 0x2028 && c != 0x2029 +#endif + ) break; + eptr++; + } + break; + + case OP_NOT_DIGIT: + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_digit) != 0) break; + eptr++; + } + break; + + case OP_DIGIT: + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_digit) == 0) break; + eptr++; + } + break; + + case OP_NOT_WHITESPACE: + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_space) != 0) break; + eptr++; + } + break; + + case OP_WHITESPACE: + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_space) == 0) break; + eptr++; + } + break; + + case OP_NOT_WORDCHAR: + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_word) != 0) break; + eptr++; + } + break; + + case OP_WORDCHAR: + for (i = min; i < max; i++) + { + if (eptr >= md->end_subject) + { + SCHECK_PARTIAL(); + break; + } + if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_word) == 0) break; + eptr++; + } + break; + + default: + RRETURN(PCRE_ERROR_INTERNAL); + } + + /* eptr is now past the end of the maximum run. If possessive, we are + done (no backing up). Otherwise, match at this position; anything other + than no match is immediately returned. For nomatch, back up one + character (byte), unless we are matching \R and the last thing matched + was \r\n, in which case, back up two bytes. */ + + if (possessive) continue; + while (eptr >= pp) + { + RMATCH(eptr, ecode, offset_top, md, eptrb, RM47); + if (rrc != MATCH_NOMATCH) RRETURN(rrc); + eptr--; + if (ctype == OP_ANYNL && eptr > pp && *eptr == '\n' && + eptr[-1] == '\r') eptr--; + } + } + + /* Get here if we can't make it match with any permitted repetitions */ + + RRETURN(MATCH_NOMATCH); + } + /* Control never gets here */ + + /* There's been some horrible disaster. Arrival here can only mean there is + something seriously wrong in the code above or the OP_xxx definitions. */ + + default: + DPRINTF(("Unknown opcode %d\n", *ecode)); + RRETURN(PCRE_ERROR_UNKNOWN_OPCODE); + } + + /* Do not stick any code in here without much thought; it is assumed + that "continue" in the code above comes out to here to repeat the main + loop. */ + + } /* End of main loop */ +/* Control never reaches here */ + + +/* When compiling to use the heap rather than the stack for recursive calls to +match(), the RRETURN() macro jumps here. The number that is saved in +frame->Xwhere indicates which label we actually want to return to. */ + +#ifdef NO_RECURSE +#define LBL(val) case val: goto L_RM##val; +HEAP_RETURN: +switch (frame->Xwhere) + { + LBL( 1) LBL( 2) LBL( 3) LBL( 4) LBL( 5) LBL( 6) LBL( 7) LBL( 8) + LBL( 9) LBL(10) LBL(11) LBL(12) LBL(13) LBL(14) LBL(15) LBL(17) + LBL(19) LBL(24) LBL(25) LBL(26) LBL(27) LBL(29) LBL(31) LBL(33) + LBL(35) LBL(43) LBL(47) LBL(48) LBL(49) LBL(50) LBL(51) LBL(52) + LBL(53) LBL(54) LBL(55) LBL(56) LBL(57) LBL(58) LBL(63) LBL(64) + LBL(65) LBL(66) +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + LBL(21) +#endif +#ifdef SUPPORT_UTF + LBL(16) LBL(18) LBL(20) + LBL(22) LBL(23) LBL(28) LBL(30) + LBL(32) LBL(34) LBL(42) LBL(46) +#ifdef SUPPORT_UCP + LBL(36) LBL(37) LBL(38) LBL(39) LBL(40) LBL(41) LBL(44) LBL(45) + LBL(59) LBL(60) LBL(61) LBL(62) +#endif /* SUPPORT_UCP */ +#endif /* SUPPORT_UTF */ + default: + DPRINTF(("jump error in pcre match: label %d non-existent\n", frame->Xwhere)); + +printf("+++jump error in pcre match: label %d non-existent\n", frame->Xwhere); + + return PCRE_ERROR_INTERNAL; + } +#undef LBL +#endif /* NO_RECURSE */ +} + + +/*************************************************************************** +**************************************************************************** + RECURSION IN THE match() FUNCTION + +Undefine all the macros that were defined above to handle this. */ + +#ifdef NO_RECURSE +#undef eptr +#undef ecode +#undef mstart +#undef offset_top +#undef eptrb +#undef flags + +#undef callpat +#undef charptr +#undef data +#undef next +#undef pp +#undef prev +#undef saved_eptr + +#undef new_recursive + +#undef cur_is_word +#undef condition +#undef prev_is_word + +#undef ctype +#undef length +#undef max +#undef min +#undef number +#undef offset +#undef op +#undef save_capture_last +#undef save_offset1 +#undef save_offset2 +#undef save_offset3 +#undef stacksave + +#undef newptrb + +#endif + +/* These two are defined as macros in both cases */ + +#undef fc +#undef fi + +/*************************************************************************** +***************************************************************************/ + + +#ifdef NO_RECURSE +/************************************************* +* Release allocated heap frames * +*************************************************/ + +/* This function releases all the allocated frames. The base frame is on the +machine stack, and so must not be freed. + +Argument: the address of the base frame +Returns: nothing +*/ + +static void +release_match_heapframes (heapframe *frame_base) +{ +heapframe *nextframe = frame_base->Xnextframe; +while (nextframe != NULL) + { + heapframe *oldframe = nextframe; + nextframe = nextframe->Xnextframe; + (PUBL(stack_free))(oldframe); + } +} +#endif + + +/************************************************* +* Execute a Regular Expression * +*************************************************/ + +/* This function applies a compiled re to a subject string and picks out +portions of the string if it matches. Two elements in the vector are set for +each substring: the offsets to the start and end of the substring. + +Arguments: + argument_re points to the compiled expression + extra_data points to extra data or is NULL + subject points to the subject string + length length of subject string (may contain binary zeros) + start_offset where to start in the subject string + options option bits + offsets points to a vector of ints to be filled in with offsets + offsetcount the number of elements in the vector + +Returns: > 0 => success; value is the number of elements filled in + = 0 => success, but offsets is not big enough + -1 => failed to match + < -1 => some kind of unexpected problem +*/ + +#ifdef COMPILE_PCRE8 +PCRE_EXP_DEFN int PCRE_CALL_CONVENTION +pcre_exec(const pcre *argument_re, const pcre_extra *extra_data, + PCRE_SPTR subject, int length, int start_offset, int options, int *offsets, + int offsetcount) +#else +PCRE_EXP_DEFN int PCRE_CALL_CONVENTION +pcre16_exec(const pcre16 *argument_re, const pcre16_extra *extra_data, + PCRE_SPTR16 subject, int length, int start_offset, int options, int *offsets, + int offsetcount) +#endif +{ +int rc, ocount, arg_offset_max; +int newline; +BOOL using_temporary_offsets = FALSE; +BOOL anchored; +BOOL startline; +BOOL firstline; +BOOL utf; +BOOL has_first_char = FALSE; +BOOL has_req_char = FALSE; +pcre_uchar first_char = 0; +pcre_uchar first_char2 = 0; +pcre_uchar req_char = 0; +pcre_uchar req_char2 = 0; +match_data match_block; +match_data *md = &match_block; +const pcre_uint8 *tables; +const pcre_uint8 *start_bits = NULL; +PCRE_PUCHAR start_match = (PCRE_PUCHAR)subject + start_offset; +PCRE_PUCHAR end_subject; +PCRE_PUCHAR start_partial = NULL; +PCRE_PUCHAR req_char_ptr = start_match - 1; + +const pcre_study_data *study; +const REAL_PCRE *re = (const REAL_PCRE *)argument_re; + +#ifdef NO_RECURSE +heapframe frame_zero; +frame_zero.Xprevframe = NULL; /* Marks the top level */ +frame_zero.Xnextframe = NULL; /* None are allocated yet */ +md->match_frames_base = &frame_zero; +#endif + +/* Check for the special magic call that measures the size of the stack used +per recursive call of match(). Without the funny casting for sizeof, a Windows +compiler gave this error: "unary minus operator applied to unsigned type, +result still unsigned". Hopefully the cast fixes that. */ + +if (re == NULL && extra_data == NULL && subject == NULL && length == -999 && + start_offset == -999) +#ifdef NO_RECURSE + return -((int)sizeof(heapframe)); +#else + return match(NULL, NULL, NULL, 0, NULL, NULL, 0); +#endif + +/* Plausibility checks */ + +if ((options & ~PUBLIC_EXEC_OPTIONS) != 0) return PCRE_ERROR_BADOPTION; +if (re == NULL || subject == NULL || (offsets == NULL && offsetcount > 0)) + return PCRE_ERROR_NULL; +if (offsetcount < 0) return PCRE_ERROR_BADCOUNT; +if (start_offset < 0 || start_offset > length) return PCRE_ERROR_BADOFFSET; + +/* Check that the first field in the block is the magic number. If it is not, +return with PCRE_ERROR_BADMAGIC. However, if the magic number is equal to +REVERSED_MAGIC_NUMBER we return with PCRE_ERROR_BADENDIANNESS, which +means that the pattern is likely compiled with different endianness. */ + +if (re->magic_number != MAGIC_NUMBER) + return re->magic_number == REVERSED_MAGIC_NUMBER? + PCRE_ERROR_BADENDIANNESS:PCRE_ERROR_BADMAGIC; +if ((re->flags & PCRE_MODE) == 0) return PCRE_ERROR_BADMODE; + +/* These two settings are used in the code for checking a UTF-8 string that +follows immediately afterwards. Other values in the md block are used only +during "normal" pcre_exec() processing, not when the JIT support is in use, +so they are set up later. */ + +/* PCRE_UTF16 has the same value as PCRE_UTF8. */ +utf = md->utf = (re->options & PCRE_UTF8) != 0; +md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 : + ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0; + +/* Check a UTF-8 string if required. Pass back the character offset and error +code for an invalid string if a results vector is available. */ + +#ifdef SUPPORT_UTF +if (utf && (options & PCRE_NO_UTF8_CHECK) == 0) + { + int erroroffset; + int errorcode = PRIV(valid_utf)((PCRE_PUCHAR)subject, length, &erroroffset); + if (errorcode != 0) + { + if (offsetcount >= 2) + { + offsets[0] = erroroffset; + offsets[1] = errorcode; + } +#ifdef COMPILE_PCRE16 + return (errorcode <= PCRE_UTF16_ERR1 && md->partial > 1)? + PCRE_ERROR_SHORTUTF16 : PCRE_ERROR_BADUTF16; +#else + return (errorcode <= PCRE_UTF8_ERR5 && md->partial > 1)? + PCRE_ERROR_SHORTUTF8 : PCRE_ERROR_BADUTF8; +#endif + } + + /* Check that a start_offset points to the start of a UTF character. */ + if (start_offset > 0 && start_offset < length && + NOT_FIRSTCHAR(((PCRE_PUCHAR)subject)[start_offset])) + return PCRE_ERROR_BADUTF8_OFFSET; + } +#endif + +/* If the pattern was successfully studied with JIT support, run the JIT +executable instead of the rest of this function. Most options must be set at +compile time for the JIT code to be usable. Fallback to the normal code path if +an unsupported flag is set. */ + +#ifdef SUPPORT_JIT +if (extra_data != NULL + && (extra_data->flags & (PCRE_EXTRA_EXECUTABLE_JIT | + PCRE_EXTRA_TABLES)) == PCRE_EXTRA_EXECUTABLE_JIT + && extra_data->executable_jit != NULL + && (options & ~(PCRE_NO_UTF8_CHECK | PCRE_NOTBOL | PCRE_NOTEOL | + PCRE_NOTEMPTY | PCRE_NOTEMPTY_ATSTART | + PCRE_PARTIAL_SOFT | PCRE_PARTIAL_HARD)) == 0) + { + rc = PRIV(jit_exec)(re, extra_data, (const pcre_uchar *)subject, length, + start_offset, options, offsets, offsetcount); + + /* PCRE_ERROR_NULL means that the selected normal or partial matching + mode is not compiled. In this case we simply fallback to interpreter. */ + + if (rc != PCRE_ERROR_NULL) return rc; + } +#endif + +/* Carry on with non-JIT matching. This information is for finding all the +numbers associated with a given name, for condition testing. */ + +md->name_table = (pcre_uchar *)re + re->name_table_offset; +md->name_count = re->name_count; +md->name_entry_size = re->name_entry_size; + +/* Fish out the optional data from the extra_data structure, first setting +the default values. */ + +study = NULL; +md->match_limit = MATCH_LIMIT; +md->match_limit_recursion = MATCH_LIMIT_RECURSION; +md->callout_data = NULL; + +/* The table pointer is always in native byte order. */ + +tables = re->tables; + +if (extra_data != NULL) + { + register unsigned int flags = extra_data->flags; + if ((flags & PCRE_EXTRA_STUDY_DATA) != 0) + study = (const pcre_study_data *)extra_data->study_data; + if ((flags & PCRE_EXTRA_MATCH_LIMIT) != 0) + md->match_limit = extra_data->match_limit; + if ((flags & PCRE_EXTRA_MATCH_LIMIT_RECURSION) != 0) + md->match_limit_recursion = extra_data->match_limit_recursion; + if ((flags & PCRE_EXTRA_CALLOUT_DATA) != 0) + md->callout_data = extra_data->callout_data; + if ((flags & PCRE_EXTRA_TABLES) != 0) tables = extra_data->tables; + } + +/* If the exec call supplied NULL for tables, use the inbuilt ones. This +is a feature that makes it possible to save compiled regex and re-use them +in other programs later. */ + +if (tables == NULL) tables = PRIV(default_tables); + +/* Set up other data */ + +anchored = ((re->options | options) & PCRE_ANCHORED) != 0; +startline = (re->flags & PCRE_STARTLINE) != 0; +firstline = (re->options & PCRE_FIRSTLINE) != 0; + +/* The code starts after the real_pcre block and the capture name table. */ + +md->start_code = (const pcre_uchar *)re + re->name_table_offset + + re->name_count * re->name_entry_size; + +md->start_subject = (PCRE_PUCHAR)subject; +md->start_offset = start_offset; +md->end_subject = md->start_subject + length; +end_subject = md->end_subject; + +md->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0; +md->use_ucp = (re->options & PCRE_UCP) != 0; +md->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0; +md->ignore_skip_arg = FALSE; + +/* Some options are unpacked into BOOL variables in the hope that testing +them will be faster than individual option bits. */ + +md->notbol = (options & PCRE_NOTBOL) != 0; +md->noteol = (options & PCRE_NOTEOL) != 0; +md->notempty = (options & PCRE_NOTEMPTY) != 0; +md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0; + +md->hitend = FALSE; +md->mark = md->nomatch_mark = NULL; /* In case never set */ + +md->recursive = NULL; /* No recursion at top level */ +md->hasthen = (re->flags & PCRE_HASTHEN) != 0; + +md->lcc = tables + lcc_offset; +md->fcc = tables + fcc_offset; +md->ctypes = tables + ctypes_offset; + +/* Handle different \R options. */ + +switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) + { + case 0: + if ((re->options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) != 0) + md->bsr_anycrlf = (re->options & PCRE_BSR_ANYCRLF) != 0; + else +#ifdef BSR_ANYCRLF + md->bsr_anycrlf = TRUE; +#else + md->bsr_anycrlf = FALSE; +#endif + break; + + case PCRE_BSR_ANYCRLF: + md->bsr_anycrlf = TRUE; + break; + + case PCRE_BSR_UNICODE: + md->bsr_anycrlf = FALSE; + break; + + default: return PCRE_ERROR_BADNEWLINE; + } + +/* Handle different types of newline. The three bits give eight cases. If +nothing is set at run time, whatever was used at compile time applies. */ + +switch ((((options & PCRE_NEWLINE_BITS) == 0)? re->options : + (pcre_uint32)options) & PCRE_NEWLINE_BITS) + { + case 0: newline = NEWLINE; break; /* Compile-time default */ + case PCRE_NEWLINE_CR: newline = CHAR_CR; break; + case PCRE_NEWLINE_LF: newline = CHAR_NL; break; + case PCRE_NEWLINE_CR+ + PCRE_NEWLINE_LF: newline = (CHAR_CR << 8) | CHAR_NL; break; + case PCRE_NEWLINE_ANY: newline = -1; break; + case PCRE_NEWLINE_ANYCRLF: newline = -2; break; + default: return PCRE_ERROR_BADNEWLINE; + } + +if (newline == -2) + { + md->nltype = NLTYPE_ANYCRLF; + } +else if (newline < 0) + { + md->nltype = NLTYPE_ANY; + } +else + { + md->nltype = NLTYPE_FIXED; + if (newline > 255) + { + md->nllen = 2; + md->nl[0] = (newline >> 8) & 255; + md->nl[1] = newline & 255; + } + else + { + md->nllen = 1; + md->nl[0] = newline; + } + } + +/* Partial matching was originally supported only for a restricted set of +regexes; from release 8.00 there are no restrictions, but the bits are still +defined (though never set). So there's no harm in leaving this code. */ + +if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0) + return PCRE_ERROR_BADPARTIAL; + +/* If the expression has got more back references than the offsets supplied can +hold, we get a temporary chunk of working store to use during the matching. +Otherwise, we can use the vector supplied, rounding down its size to a multiple +of 3. */ + +ocount = offsetcount - (offsetcount % 3); +arg_offset_max = (2*ocount)/3; + +if (re->top_backref > 0 && re->top_backref >= ocount/3) + { + ocount = re->top_backref * 3 + 3; + md->offset_vector = (int *)(PUBL(malloc))(ocount * sizeof(int)); + if (md->offset_vector == NULL) return PCRE_ERROR_NOMEMORY; + using_temporary_offsets = TRUE; + DPRINTF(("Got memory to hold back references\n")); + } +else md->offset_vector = offsets; + +md->offset_end = ocount; +md->offset_max = (2*ocount)/3; +md->offset_overflow = FALSE; +md->capture_last = -1; + +/* Reset the working variable associated with each extraction. These should +never be used unless previously set, but they get saved and restored, and so we +initialize them to avoid reading uninitialized locations. Also, unset the +offsets for the matched string. This is really just for tidiness with callouts, +in case they inspect these fields. */ + +if (md->offset_vector != NULL) + { + register int *iptr = md->offset_vector + ocount; + register int *iend = iptr - re->top_bracket; + if (iend < md->offset_vector + 2) iend = md->offset_vector + 2; + while (--iptr >= iend) *iptr = -1; + md->offset_vector[0] = md->offset_vector[1] = -1; + } + +/* Set up the first character to match, if available. The first_char value is +never set for an anchored regular expression, but the anchoring may be forced +at run time, so we have to test for anchoring. The first char may be unset for +an unanchored pattern, of course. If there's no first char and the pattern was +studied, there may be a bitmap of possible first characters. */ + +if (!anchored) + { + if ((re->flags & PCRE_FIRSTSET) != 0) + { + has_first_char = TRUE; + first_char = first_char2 = (pcre_uchar)(re->first_char); + if ((re->flags & PCRE_FCH_CASELESS) != 0) + { + first_char2 = TABLE_GET(first_char, md->fcc, first_char); +#if defined SUPPORT_UCP && !(defined COMPILE_PCRE8) + if (utf && first_char > 127) + first_char2 = UCD_OTHERCASE(first_char); +#endif + } + } + else + if (!startline && study != NULL && + (study->flags & PCRE_STUDY_MAPPED) != 0) + start_bits = study->start_bits; + } + +/* For anchored or unanchored matches, there may be a "last known required +character" set. */ + +if ((re->flags & PCRE_REQCHSET) != 0) + { + has_req_char = TRUE; + req_char = req_char2 = (pcre_uchar)(re->req_char); + if ((re->flags & PCRE_RCH_CASELESS) != 0) + { + req_char2 = TABLE_GET(req_char, md->fcc, req_char); +#if defined SUPPORT_UCP && !(defined COMPILE_PCRE8) + if (utf && req_char > 127) + req_char2 = UCD_OTHERCASE(req_char); +#endif + } + } + + +/* ==========================================================================*/ + +/* Loop for handling unanchored repeated matching attempts; for anchored regexs +the loop runs just once. */ + +for(;;) + { + PCRE_PUCHAR save_end_subject = end_subject; + PCRE_PUCHAR new_start_match; + + /* If firstline is TRUE, the start of the match is constrained to the first + line of a multiline string. That is, the match must be before or at the first + newline. Implement this by temporarily adjusting end_subject so that we stop + scanning at a newline. If the match fails at the newline, later code breaks + this loop. */ + + if (firstline) + { + PCRE_PUCHAR t = start_match; +#ifdef SUPPORT_UTF + if (utf) + { + while (t < md->end_subject && !IS_NEWLINE(t)) + { + t++; + ACROSSCHAR(t < end_subject, *t, t++); + } + } + else +#endif + while (t < md->end_subject && !IS_NEWLINE(t)) t++; + end_subject = t; + } + + /* There are some optimizations that avoid running the match if a known + starting point is not found, or if a known later character is not present. + However, there is an option that disables these, for testing and for ensuring + that all callouts do actually occur. The option can be set in the regex by + (*NO_START_OPT) or passed in match-time options. */ + + if (((options | re->options) & PCRE_NO_START_OPTIMIZE) == 0) + { + /* Advance to a unique first char if there is one. */ + + if (has_first_char) + { + if (first_char != first_char2) + while (start_match < end_subject && + *start_match != first_char && *start_match != first_char2) + start_match++; + else + while (start_match < end_subject && *start_match != first_char) + start_match++; + } + + /* Or to just after a linebreak for a multiline match */ + + else if (startline) + { + if (start_match > md->start_subject + start_offset) + { +#ifdef SUPPORT_UTF + if (utf) + { + while (start_match < end_subject && !WAS_NEWLINE(start_match)) + { + start_match++; + ACROSSCHAR(start_match < end_subject, *start_match, + start_match++); + } + } + else +#endif + while (start_match < end_subject && !WAS_NEWLINE(start_match)) + start_match++; + + /* If we have just passed a CR and the newline option is ANY or ANYCRLF, + and we are now at a LF, advance the match position by one more character. + */ + + if (start_match[-1] == CHAR_CR && + (md->nltype == NLTYPE_ANY || md->nltype == NLTYPE_ANYCRLF) && + start_match < end_subject && + *start_match == CHAR_NL) + start_match++; + } + } + + /* Or to a non-unique first byte after study */ + + else if (start_bits != NULL) + { + while (start_match < end_subject) + { + register unsigned int c = *start_match; +#ifndef COMPILE_PCRE8 + if (c > 255) c = 255; +#endif + if ((start_bits[c/8] & (1 << (c&7))) == 0) + { + start_match++; +#if defined SUPPORT_UTF && defined COMPILE_PCRE8 + /* In non 8-bit mode, the iteration will stop for + characters > 255 at the beginning or not stop at all. */ + if (utf) + ACROSSCHAR(start_match < end_subject, *start_match, + start_match++); +#endif + } + else break; + } + } + } /* Starting optimizations */ + + /* Restore fudged end_subject */ + + end_subject = save_end_subject; + + /* The following two optimizations are disabled for partial matching or if + disabling is explicitly requested. */ + + if (((options | re->options) & PCRE_NO_START_OPTIMIZE) == 0 && !md->partial) + { + /* If the pattern was studied, a minimum subject length may be set. This is + a lower bound; no actual string of that length may actually match the + pattern. Although the value is, strictly, in characters, we treat it as + bytes to avoid spending too much time in this optimization. */ + + if (study != NULL && (study->flags & PCRE_STUDY_MINLEN) != 0 && + (pcre_uint32)(end_subject - start_match) < study->minlength) + { + rc = MATCH_NOMATCH; + break; + } + + /* If req_char is set, we know that that character must appear in the + subject for the match to succeed. If the first character is set, req_char + must be later in the subject; otherwise the test starts at the match point. + This optimization can save a huge amount of backtracking in patterns with + nested unlimited repeats that aren't going to match. Writing separate code + for cased/caseless versions makes it go faster, as does using an + autoincrement and backing off on a match. + + HOWEVER: when the subject string is very, very long, searching to its end + can take a long time, and give bad performance on quite ordinary patterns. + This showed up when somebody was matching something like /^\d+C/ on a + 32-megabyte string... so we don't do this when the string is sufficiently + long. */ + + if (has_req_char && end_subject - start_match < REQ_BYTE_MAX) + { + register PCRE_PUCHAR p = start_match + (has_first_char? 1:0); + + /* We don't need to repeat the search if we haven't yet reached the + place we found it at last time. */ + + if (p > req_char_ptr) + { + if (req_char != req_char2) + { + while (p < end_subject) + { + register int pp = *p++; + if (pp == req_char || pp == req_char2) { p--; break; } + } + } + else + { + while (p < end_subject) + { + if (*p++ == req_char) { p--; break; } + } + } + + /* If we can't find the required character, break the matching loop, + forcing a match failure. */ + + if (p >= end_subject) + { + rc = MATCH_NOMATCH; + break; + } + + /* If we have found the required character, save the point where we + found it, so that we don't search again next time round the loop if + the start hasn't passed this character yet. */ + + req_char_ptr = p; + } + } + } + +#ifdef PCRE_DEBUG /* Sigh. Some compilers never learn. */ + printf(">>>> Match against: "); + pchars(start_match, end_subject - start_match, TRUE, md); + printf("\n"); +#endif + + /* OK, we can now run the match. If "hitend" is set afterwards, remember the + first starting point for which a partial match was found. */ + + md->start_match_ptr = start_match; + md->start_used_ptr = start_match; + md->match_call_count = 0; + md->match_function_type = 0; + md->end_offset_top = 0; + rc = match(start_match, md->start_code, start_match, 2, md, NULL, 0); + if (md->hitend && start_partial == NULL) start_partial = md->start_used_ptr; + + switch(rc) + { + /* If MATCH_SKIP_ARG reaches this level it means that a MARK that matched + the SKIP's arg was not found. In this circumstance, Perl ignores the SKIP + entirely. The only way we can do that is to re-do the match at the same + point, with a flag to force SKIP with an argument to be ignored. Just + treating this case as NOMATCH does not work because it does not check other + alternatives in patterns such as A(*SKIP:A)B|AC when the subject is AC. */ + + case MATCH_SKIP_ARG: + new_start_match = start_match; + md->ignore_skip_arg = TRUE; + break; + + /* SKIP passes back the next starting point explicitly, but if it is the + same as the match we have just done, treat it as NOMATCH. */ + + case MATCH_SKIP: + if (md->start_match_ptr != start_match) + { + new_start_match = md->start_match_ptr; + break; + } + /* Fall through */ + + /* NOMATCH and PRUNE advance by one character. THEN at this level acts + exactly like PRUNE. Unset the ignore SKIP-with-argument flag. */ + + case MATCH_NOMATCH: + case MATCH_PRUNE: + case MATCH_THEN: + md->ignore_skip_arg = FALSE; + new_start_match = start_match + 1; +#ifdef SUPPORT_UTF + if (utf) + ACROSSCHAR(new_start_match < end_subject, *new_start_match, + new_start_match++); +#endif + break; + + /* COMMIT disables the bumpalong, but otherwise behaves as NOMATCH. */ + + case MATCH_COMMIT: + rc = MATCH_NOMATCH; + goto ENDLOOP; + + /* Any other return is either a match, or some kind of error. */ + + default: + goto ENDLOOP; + } + + /* Control reaches here for the various types of "no match at this point" + result. Reset the code to MATCH_NOMATCH for subsequent checking. */ + + rc = MATCH_NOMATCH; + + /* If PCRE_FIRSTLINE is set, the match must happen before or at the first + newline in the subject (though it may continue over the newline). Therefore, + if we have just failed to match, starting at a newline, do not continue. */ + + if (firstline && IS_NEWLINE(start_match)) break; + + /* Advance to new matching position */ + + start_match = new_start_match; + + /* Break the loop if the pattern is anchored or if we have passed the end of + the subject. */ + + if (anchored || start_match > end_subject) break; + + /* If we have just passed a CR and we are now at a LF, and the pattern does + not contain any explicit matches for \r or \n, and the newline option is CRLF + or ANY or ANYCRLF, advance the match position by one more character. In + normal matching start_match will aways be greater than the first position at + this stage, but a failed *SKIP can cause a return at the same point, which is + why the first test exists. */ + + if (start_match > (PCRE_PUCHAR)subject + start_offset && + start_match[-1] == CHAR_CR && + start_match < end_subject && + *start_match == CHAR_NL && + (re->flags & PCRE_HASCRORLF) == 0 && + (md->nltype == NLTYPE_ANY || + md->nltype == NLTYPE_ANYCRLF || + md->nllen == 2)) + start_match++; + + md->mark = NULL; /* Reset for start of next match attempt */ + } /* End of for(;;) "bumpalong" loop */ + +/* ==========================================================================*/ + +/* We reach here when rc is not MATCH_NOMATCH, or if one of the stopping +conditions is true: + +(1) The pattern is anchored or the match was failed by (*COMMIT); + +(2) We are past the end of the subject; + +(3) PCRE_FIRSTLINE is set and we have failed to match at a newline, because + this option requests that a match occur at or before the first newline in + the subject. + +When we have a match and the offset vector is big enough to deal with any +backreferences, captured substring offsets will already be set up. In the case +where we had to get some local store to hold offsets for backreference +processing, copy those that we can. In this case there need not be overflow if +certain parts of the pattern were not used, even though there are more +capturing parentheses than vector slots. */ + +ENDLOOP: + +if (rc == MATCH_MATCH || rc == MATCH_ACCEPT) + { + if (using_temporary_offsets) + { + if (arg_offset_max >= 4) + { + memcpy(offsets + 2, md->offset_vector + 2, + (arg_offset_max - 2) * sizeof(int)); + DPRINTF(("Copied offsets from temporary memory\n")); + } + if (md->end_offset_top > arg_offset_max) md->offset_overflow = TRUE; + DPRINTF(("Freeing temporary memory\n")); + (PUBL(free))(md->offset_vector); + } + + /* Set the return code to the number of captured strings, or 0 if there were + too many to fit into the vector. */ + + rc = (md->offset_overflow && md->end_offset_top >= arg_offset_max)? + 0 : md->end_offset_top/2; + + /* If there is space in the offset vector, set any unused pairs at the end of + the pattern to -1 for backwards compatibility. It is documented that this + happens. In earlier versions, the whole set of potential capturing offsets + was set to -1 each time round the loop, but this is handled differently now. + "Gaps" are set to -1 dynamically instead (this fixes a bug). Thus, it is only + those at the end that need unsetting here. We can't just unset them all at + the start of the whole thing because they may get set in one branch that is + not the final matching branch. */ + + if (md->end_offset_top/2 <= re->top_bracket && offsets != NULL) + { + register int *iptr, *iend; + int resetcount = 2 + re->top_bracket * 2; + if (resetcount > offsetcount) resetcount = offsetcount; + iptr = offsets + md->end_offset_top; + iend = offsets + resetcount; + while (iptr < iend) *iptr++ = -1; + } + + /* If there is space, set up the whole thing as substring 0. The value of + md->start_match_ptr might be modified if \K was encountered on the success + matching path. */ + + if (offsetcount < 2) rc = 0; else + { + offsets[0] = (int)(md->start_match_ptr - md->start_subject); + offsets[1] = (int)(md->end_match_ptr - md->start_subject); + } + + /* Return MARK data if requested */ + + if (extra_data != NULL && (extra_data->flags & PCRE_EXTRA_MARK) != 0) + *(extra_data->mark) = (pcre_uchar *)md->mark; + DPRINTF((">>>> returning %d\n", rc)); +#ifdef NO_RECURSE + release_match_heapframes(&frame_zero); +#endif + return rc; + } + +/* Control gets here if there has been an error, or if the overall match +attempt has failed at all permitted starting positions. */ + +if (using_temporary_offsets) + { + DPRINTF(("Freeing temporary memory\n")); + (PUBL(free))(md->offset_vector); + } + +/* For anything other than nomatch or partial match, just return the code. */ + +if (rc != MATCH_NOMATCH && rc != PCRE_ERROR_PARTIAL) + { + DPRINTF((">>>> error: returning %d\n", rc)); +#ifdef NO_RECURSE + release_match_heapframes(&frame_zero); +#endif + return rc; + } + +/* Handle partial matches - disable any mark data */ + +if (start_partial != NULL) + { + DPRINTF((">>>> returning PCRE_ERROR_PARTIAL\n")); + md->mark = NULL; + if (offsetcount > 1) + { + offsets[0] = (int)(start_partial - (PCRE_PUCHAR)subject); + offsets[1] = (int)(end_subject - (PCRE_PUCHAR)subject); } rc = PCRE_ERROR_PARTIAL; } diff --git a/pcre/pcre_fullinfo.c b/pcre/pcre_fullinfo.c --- a/pcre/pcre_fullinfo.c +++ b/pcre/pcre_fullinfo.c @@ -6,7 +6,7 @@ and semantics are as close as possible to those of the Perl 5 language. Written by Philip Hazel - Copyright (c) 1997-2013 University of Cambridge + Copyright (c) 1997-2012 University of Cambridge ----------------------------------------------------------------------------- Redistribution and use in source and binary forms, with or without @@ -65,18 +65,14 @@ Returns: 0 if data returned, negative on error */ -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre_fullinfo(const pcre *argument_re, const pcre_extra *extra_data, int what, void *where) -#elif defined COMPILE_PCRE16 +#else PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre16_fullinfo(const pcre16 *argument_re, const pcre16_extra *extra_data, int what, void *where) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DEFN int PCRE_CALL_CONVENTION -pcre32_fullinfo(const pcre32 *argument_re, const pcre32_extra *extra_data, - int what, void *where) #endif { const REAL_PCRE *re = (const REAL_PCRE *)argument_re; @@ -136,21 +132,10 @@ case PCRE_INFO_FIRSTBYTE: *((int *)where) = - ((re->flags & PCRE_FIRSTSET) != 0)? (int)re->first_char : + ((re->flags & PCRE_FIRSTSET) != 0)? re->first_char : ((re->flags & PCRE_STARTLINE) != 0)? -1 : -2; break; - case PCRE_INFO_FIRSTCHARACTER: - *((pcre_uint32 *)where) = - (re->flags & PCRE_FIRSTSET) != 0 ? re->first_char : 0; - break; - - case PCRE_INFO_FIRSTCHARACTERFLAGS: - *((int *)where) = - ((re->flags & PCRE_FIRSTSET) != 0) ? 1 : - ((re->flags & PCRE_STARTLINE) != 0) ? 2 : 0; - break; - /* Make sure we pass back the pointer to the bit vector in the external block, not the internal copy (with flipped integer fields). */ @@ -174,19 +159,9 @@ case PCRE_INFO_LASTLITERAL: *((int *)where) = - ((re->flags & PCRE_REQCHSET) != 0)? (int)re->req_char : -1; + ((re->flags & PCRE_REQCHSET) != 0)? re->req_char : -1; break; - case PCRE_INFO_REQUIREDCHAR: - *((pcre_uint32 *)where) = - ((re->flags & PCRE_REQCHSET) != 0) ? re->req_char : 0; - break; - - case PCRE_INFO_REQUIREDCHARFLAGS: - *((int *)where) = - ((re->flags & PCRE_REQCHSET) != 0); - break; - case PCRE_INFO_NAMEENTRYSIZE: *((int *)where) = re->name_entry_size; break; @@ -220,20 +195,6 @@ case PCRE_INFO_MAXLOOKBEHIND: *((int *)where) = re->max_lookbehind; - break; - - case PCRE_INFO_MATCHLIMIT: - if ((re->flags & PCRE_MLSET) == 0) return PCRE_ERROR_UNSET; - *((pcre_uint32 *)where) = re->limit_match; - break; - - case PCRE_INFO_RECURSIONLIMIT: - if ((re->flags & PCRE_RLSET) == 0) return PCRE_ERROR_UNSET; - *((pcre_uint32 *)where) = re->limit_recursion; - break; - - case PCRE_INFO_MATCH_EMPTY: - *((int *)where) = (re->flags & PCRE_MATCH_EMPTY) != 0; break; default: return PCRE_ERROR_BADOPTION; diff --git a/pcre/pcre_get.c b/pcre/pcre_get.c --- a/pcre/pcre_get.c +++ b/pcre/pcre_get.c @@ -65,15 +65,12 @@ (PCRE_ERROR_NOSUBSTRING) if not found */ -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre_get_stringnumber(const pcre *code, const char *stringname) -#elif defined COMPILE_PCRE16 +#else PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre16_get_stringnumber(const pcre16 *code, PCRE_SPTR16 stringname) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DEFN int PCRE_CALL_CONVENTION -pcre32_get_stringnumber(const pcre32 *code, PCRE_SPTR32 stringname) #endif { int rc; @@ -101,17 +98,7 @@ if ((rc = pcre16_fullinfo(code, NULL, PCRE_INFO_NAMETABLE, &nametable)) != 0) return rc; #endif -#ifdef COMPILE_PCRE32 -if ((rc = pcre32_fullinfo(code, NULL, PCRE_INFO_NAMECOUNT, &top)) != 0) - return rc; -if (top <= 0) return PCRE_ERROR_NOSUBSTRING; -if ((rc = pcre32_fullinfo(code, NULL, PCRE_INFO_NAMEENTRYSIZE, &entrysize)) != 0) - return rc; -if ((rc = pcre32_fullinfo(code, NULL, PCRE_INFO_NAMETABLE, &nametable)) != 0) - return rc; -#endif - bot = 0; while (top > bot) { @@ -145,18 +132,14 @@ (PCRE_ERROR_NOSUBSTRING) if not found */ -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre_get_stringtable_entries(const pcre *code, const char *stringname, char **firstptr, char **lastptr) -#elif defined COMPILE_PCRE16 +#else PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre16_get_stringtable_entries(const pcre16 *code, PCRE_SPTR16 stringname, PCRE_UCHAR16 **firstptr, PCRE_UCHAR16 **lastptr) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DEFN int PCRE_CALL_CONVENTION -pcre32_get_stringtable_entries(const pcre32 *code, PCRE_SPTR32 stringname, - PCRE_UCHAR32 **firstptr, PCRE_UCHAR32 **lastptr) #endif { int rc; @@ -184,17 +167,7 @@ if ((rc = pcre16_fullinfo(code, NULL, PCRE_INFO_NAMETABLE, &nametable)) != 0) return rc; #endif -#ifdef COMPILE_PCRE32 -if ((rc = pcre32_fullinfo(code, NULL, PCRE_INFO_NAMECOUNT, &top)) != 0) - return rc; -if (top <= 0) return PCRE_ERROR_NOSUBSTRING; -if ((rc = pcre32_fullinfo(code, NULL, PCRE_INFO_NAMEENTRYSIZE, &entrysize)) != 0) - return rc; -if ((rc = pcre32_fullinfo(code, NULL, PCRE_INFO_NAMETABLE, &nametable)) != 0) - return rc; -#endif - lastentry = nametable + entrysize * (top - 1); bot = 0; while (top > bot) @@ -219,15 +192,12 @@ (pcre_uchar *)(last + entrysize + IMM2_SIZE)) != 0) break; last += entrysize; } -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 *firstptr = (char *)first; *lastptr = (char *)last; -#elif defined COMPILE_PCRE16 +#else *firstptr = (PCRE_UCHAR16 *)first; *lastptr = (PCRE_UCHAR16 *)last; -#elif defined COMPILE_PCRE32 - *firstptr = (PCRE_UCHAR32 *)first; - *lastptr = (PCRE_UCHAR32 *)last; #endif return entrysize; } @@ -256,40 +226,31 @@ or a negative number on error */ -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 static int get_first_set(const pcre *code, const char *stringname, int *ovector) -#elif defined COMPILE_PCRE16 +#else static int get_first_set(const pcre16 *code, PCRE_SPTR16 stringname, int *ovector) -#elif defined COMPILE_PCRE32 -static int -get_first_set(const pcre32 *code, PCRE_SPTR32 stringname, int *ovector) #endif { const REAL_PCRE *re = (const REAL_PCRE *)code; int entrysize; pcre_uchar *entry; -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 char *first, *last; -#elif defined COMPILE_PCRE16 +#else PCRE_UCHAR16 *first, *last; -#elif defined COMPILE_PCRE32 -PCRE_UCHAR32 *first, *last; #endif -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 if ((re->options & PCRE_DUPNAMES) == 0 && (re->flags & PCRE_JCHANGED) == 0) return pcre_get_stringnumber(code, stringname); entrysize = pcre_get_stringtable_entries(code, stringname, &first, &last); -#elif defined COMPILE_PCRE16 +#else if ((re->options & PCRE_DUPNAMES) == 0 && (re->flags & PCRE_JCHANGED) == 0) return pcre16_get_stringnumber(code, stringname); entrysize = pcre16_get_stringtable_entries(code, stringname, &first, &last); -#elif defined COMPILE_PCRE32 -if ((re->options & PCRE_DUPNAMES) == 0 && (re->flags & PCRE_JCHANGED) == 0) - return pcre32_get_stringnumber(code, stringname); -entrysize = pcre32_get_stringtable_entries(code, stringname, &first, &last); #endif if (entrysize <= 0) return entrysize; for (entry = (pcre_uchar *)first; entry <= (pcre_uchar *)last; entry += entrysize) @@ -330,18 +291,14 @@ PCRE_ERROR_NOSUBSTRING (-7) no such captured substring */ -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre_copy_substring(const char *subject, int *ovector, int stringcount, int stringnumber, char *buffer, int size) -#elif defined COMPILE_PCRE16 +#else PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre16_copy_substring(PCRE_SPTR16 subject, int *ovector, int stringcount, int stringnumber, PCRE_UCHAR16 *buffer, int size) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DEFN int PCRE_CALL_CONVENTION -pcre32_copy_substring(PCRE_SPTR32 subject, int *ovector, int stringcount, - int stringnumber, PCRE_UCHAR32 *buffer, int size) #endif { int yield; @@ -385,31 +342,24 @@ PCRE_ERROR_NOSUBSTRING (-7) no such captured substring */ -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre_copy_named_substring(const pcre *code, const char *subject, int *ovector, int stringcount, const char *stringname, char *buffer, int size) -#elif defined COMPILE_PCRE16 +#else PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre16_copy_named_substring(const pcre16 *code, PCRE_SPTR16 subject, int *ovector, int stringcount, PCRE_SPTR16 stringname, PCRE_UCHAR16 *buffer, int size) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DEFN int PCRE_CALL_CONVENTION -pcre32_copy_named_substring(const pcre32 *code, PCRE_SPTR32 subject, - int *ovector, int stringcount, PCRE_SPTR32 stringname, - PCRE_UCHAR32 *buffer, int size) #endif { int n = get_first_set(code, stringname, ovector); if (n <= 0) return n; -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 return pcre_copy_substring(subject, ovector, stringcount, n, buffer, size); -#elif defined COMPILE_PCRE16 +#else return pcre16_copy_substring(subject, ovector, stringcount, n, buffer, size); -#elif defined COMPILE_PCRE32 -return pcre32_copy_substring(subject, ovector, stringcount, n, buffer, size); #endif } @@ -436,18 +386,14 @@ PCRE_ERROR_NOMEMORY (-6) failed to get store */ -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre_get_substring_list(const char *subject, int *ovector, int stringcount, const char ***listptr) -#elif defined COMPILE_PCRE16 +#else PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre16_get_substring_list(PCRE_SPTR16 subject, int *ovector, int stringcount, PCRE_SPTR16 **listptr) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DEFN int PCRE_CALL_CONVENTION -pcre32_get_substring_list(PCRE_SPTR32 subject, int *ovector, int stringcount, - PCRE_SPTR32 **listptr) #endif { int i; @@ -462,12 +408,10 @@ stringlist = (pcre_uchar **)(PUBL(malloc))(size); if (stringlist == NULL) return PCRE_ERROR_NOMEMORY; -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 *listptr = (const char **)stringlist; -#elif defined COMPILE_PCRE16 +#else *listptr = (PCRE_SPTR16 *)stringlist; -#elif defined COMPILE_PCRE32 -*listptr = (PCRE_SPTR32 *)stringlist; #endif p = (pcre_uchar *)(stringlist + stringcount + 1); @@ -498,15 +442,12 @@ Returns: nothing */ -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 PCRE_EXP_DEFN void PCRE_CALL_CONVENTION pcre_free_substring_list(const char **pointer) -#elif defined COMPILE_PCRE16 +#else PCRE_EXP_DEFN void PCRE_CALL_CONVENTION pcre16_free_substring_list(PCRE_SPTR16 *pointer) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DEFN void PCRE_CALL_CONVENTION -pcre32_free_substring_list(PCRE_SPTR32 *pointer) #endif { (PUBL(free))((void *)pointer); @@ -539,18 +480,14 @@ PCRE_ERROR_NOSUBSTRING (-7) substring not present */ -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre_get_substring(const char *subject, int *ovector, int stringcount, int stringnumber, const char **stringptr) -#elif defined COMPILE_PCRE16 +#else PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre16_get_substring(PCRE_SPTR16 subject, int *ovector, int stringcount, int stringnumber, PCRE_SPTR16 *stringptr) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DEFN int PCRE_CALL_CONVENTION -pcre32_get_substring(PCRE_SPTR32 subject, int *ovector, int stringcount, - int stringnumber, PCRE_SPTR32 *stringptr) #endif { int yield; @@ -563,12 +500,10 @@ if (substring == NULL) return PCRE_ERROR_NOMEMORY; memcpy(substring, subject + ovector[stringnumber], IN_UCHARS(yield)); substring[yield] = 0; -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 *stringptr = (const char *)substring; -#elif defined COMPILE_PCRE16 +#else *stringptr = (PCRE_SPTR16)substring; -#elif defined COMPILE_PCRE32 -*stringptr = (PCRE_SPTR32)substring; #endif return yield; } @@ -602,31 +537,24 @@ PCRE_ERROR_NOSUBSTRING (-7) no such captured substring */ -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre_get_named_substring(const pcre *code, const char *subject, int *ovector, int stringcount, const char *stringname, const char **stringptr) -#elif defined COMPILE_PCRE16 +#else PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre16_get_named_substring(const pcre16 *code, PCRE_SPTR16 subject, int *ovector, int stringcount, PCRE_SPTR16 stringname, PCRE_SPTR16 *stringptr) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DEFN int PCRE_CALL_CONVENTION -pcre32_get_named_substring(const pcre32 *code, PCRE_SPTR32 subject, - int *ovector, int stringcount, PCRE_SPTR32 stringname, - PCRE_SPTR32 *stringptr) #endif { int n = get_first_set(code, stringname, ovector); if (n <= 0) return n; -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 return pcre_get_substring(subject, ovector, stringcount, n, stringptr); -#elif defined COMPILE_PCRE16 +#else return pcre16_get_substring(subject, ovector, stringcount, n, stringptr); -#elif defined COMPILE_PCRE32 -return pcre32_get_substring(subject, ovector, stringcount, n, stringptr); #endif } @@ -645,15 +573,12 @@ Returns: nothing */ -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 PCRE_EXP_DEFN void PCRE_CALL_CONVENTION pcre_free_substring(const char *pointer) -#elif defined COMPILE_PCRE16 +#else PCRE_EXP_DEFN void PCRE_CALL_CONVENTION pcre16_free_substring(PCRE_SPTR16 pointer) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DEFN void PCRE_CALL_CONVENTION -pcre32_free_substring(PCRE_SPTR32 pointer) #endif { (PUBL(free))((void *)pointer); diff --git a/pcre/pcre_internal.h b/pcre/pcre_internal.h --- a/pcre/pcre_internal.h +++ b/pcre/pcre_internal.h @@ -7,2832 +7,2337 @@ and semantics are as close as possible to those of the Perl 5 language. Written by Philip Hazel - Copyright (c) 1997-2013 University of Cambridge - ------------------------------------------------------------------------------ -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. - - * Neither the name of the University of Cambridge 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 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. ------------------------------------------------------------------------------ -*/ - -/* This header contains definitions that are shared between the different -modules, but which are not relevant to the exported API. This includes some -functions whose names all begin with "_pcre_", "_pcre16_" or "_pcre32_" -depending on the PRIV macro. */ - -#ifndef PCRE_INTERNAL_H -#define PCRE_INTERNAL_H - -/* Define PCRE_DEBUG to get debugging output on stdout. */ - -#if 0 -#define PCRE_DEBUG -#endif - -/* PCRE is compiled as an 8 bit library if it is not requested otherwise. */ - -#if !defined COMPILE_PCRE16 && !defined COMPILE_PCRE32 -#define COMPILE_PCRE8 -#endif - -/* If SUPPORT_UCP is defined, SUPPORT_UTF must also be defined. The -"configure" script ensures this, but not everybody uses "configure". */ - -#if defined SUPPORT_UCP && !(defined SUPPORT_UTF) -#define SUPPORT_UTF 1 -#endif - -/* We define SUPPORT_UTF if SUPPORT_UTF8 is enabled for compatibility -reasons with existing code. */ - -#if defined SUPPORT_UTF8 && !(defined SUPPORT_UTF) -#define SUPPORT_UTF 1 -#endif - -/* Fixme: SUPPORT_UTF8 should be eventually disappear from the code. -Until then we define it if SUPPORT_UTF is defined. */ - -#if defined SUPPORT_UTF && !(defined SUPPORT_UTF8) -#define SUPPORT_UTF8 1 -#endif - -/* We do not support both EBCDIC and UTF-8/16/32 at the same time. The "configure" -script prevents both being selected, but not everybody uses "configure". */ - -#if defined EBCDIC && defined SUPPORT_UTF -#error The use of both EBCDIC and SUPPORT_UTF is not supported. -#endif - -/* Use a macro for debugging printing, 'cause that eliminates the use of #ifdef -inline, and there are *still* stupid compilers about that don't like indented -pre-processor statements, or at least there were when I first wrote this. After -all, it had only been about 10 years then... - -It turns out that the Mac Debugging.h header also defines the macro DPRINTF, so -be absolutely sure we get our version. */ - -#undef DPRINTF -#ifdef PCRE_DEBUG -#define DPRINTF(p) printf p -#else -#define DPRINTF(p) /* Nothing */ -#endif - - -/* Standard C headers plus the external interface definition. The only time -setjmp and stdarg are used is when NO_RECURSE is set. */ - -#include -#include -#include -#include -#include -#include - -/* Valgrind (memcheck) support */ - -#ifdef SUPPORT_VALGRIND -#include -#endif - -/* When compiling a DLL for Windows, the exported symbols have to be declared -using some MS magic. I found some useful information on this web page: -http://msdn2.microsoft.com/en-us/library/y4h7bcy6(VS.80).aspx. According to the -information there, using __declspec(dllexport) without "extern" we have a -definition; with "extern" we have a declaration. The settings here override the -setting in pcre.h (which is included below); it defines only PCRE_EXP_DECL, -which is all that is needed for applications (they just import the symbols). We -use: - - PCRE_EXP_DECL for declarations - PCRE_EXP_DEFN for definitions of exported functions - PCRE_EXP_DATA_DEFN for definitions of exported variables - -The reason for the two DEFN macros is that in non-Windows environments, one -does not want to have "extern" before variable definitions because it leads to -compiler warnings. So we distinguish between functions and variables. In -Windows, the two should always be the same. - -The reason for wrapping this in #ifndef PCRE_EXP_DECL is so that pcretest, -which is an application, but needs to import this file in order to "peek" at -internals, can #include pcre.h first to get an application's-eye view. - -In principle, people compiling for non-Windows, non-Unix-like (i.e. uncommon, -special-purpose environments) might want to stick other stuff in front of -exported symbols. That's why, in the non-Windows case, we set PCRE_EXP_DEFN and -PCRE_EXP_DATA_DEFN only if they are not already set. */ - -#ifndef PCRE_EXP_DECL -# ifdef _WIN32 -# ifndef PCRE_STATIC -# define PCRE_EXP_DECL extern __declspec(dllexport) -# define PCRE_EXP_DEFN __declspec(dllexport) -# define PCRE_EXP_DATA_DEFN __declspec(dllexport) -# else -# define PCRE_EXP_DECL extern -# define PCRE_EXP_DEFN -# define PCRE_EXP_DATA_DEFN -# endif -# else -# ifdef __cplusplus -# define PCRE_EXP_DECL extern "C" -# else -# define PCRE_EXP_DECL extern -# endif -# ifndef PCRE_EXP_DEFN -# define PCRE_EXP_DEFN PCRE_EXP_DECL -# endif -# ifndef PCRE_EXP_DATA_DEFN -# define PCRE_EXP_DATA_DEFN -# endif -# endif -#endif - -/* When compiling with the MSVC compiler, it is sometimes necessary to include -a "calling convention" before exported function names. (This is secondhand -information; I know nothing about MSVC myself). For example, something like - - void __cdecl function(....) - -might be needed. In order so make this easy, all the exported functions have -PCRE_CALL_CONVENTION just before their names. It is rarely needed; if not -set, we ensure here that it has no effect. */ - -#ifndef PCRE_CALL_CONVENTION -#define PCRE_CALL_CONVENTION -#endif - -/* We need to have types that specify unsigned 8, 16 and 32-bit integers. We -cannot determine these outside the compilation (e.g. by running a program as -part of "configure") because PCRE is often cross-compiled for use on other -systems. Instead we make use of the maximum sizes that are available at -preprocessor time in standard C environments. */ - -typedef unsigned char pcre_uint8; - -#if USHRT_MAX == 65535 -typedef unsigned short pcre_uint16; -typedef short pcre_int16; -#define PCRE_UINT16_MAX USHRT_MAX -#define PCRE_INT16_MAX SHRT_MAX -#elif UINT_MAX == 65535 -typedef unsigned int pcre_uint16; -typedef int pcre_int16; -#define PCRE_UINT16_MAX UINT_MAX -#define PCRE_INT16_MAX INT_MAX -#else -#error Cannot determine a type for 16-bit integers -#endif - -#if UINT_MAX == 4294967295U -typedef unsigned int pcre_uint32; -typedef int pcre_int32; -#define PCRE_UINT32_MAX UINT_MAX -#define PCRE_INT32_MAX INT_MAX -#elif ULONG_MAX == 4294967295UL -typedef unsigned long int pcre_uint32; -typedef long int pcre_int32; -#define PCRE_UINT32_MAX ULONG_MAX -#define PCRE_INT32_MAX LONG_MAX -#else -#error Cannot determine a type for 32-bit integers -#endif - -/* When checking for integer overflow in pcre_compile(), we need to handle -large integers. If a 64-bit integer type is available, we can use that. -Otherwise we have to cast to double, which of course requires floating point -arithmetic. Handle this by defining a macro for the appropriate type. If -stdint.h is available, include it; it may define INT64_MAX. Systems that do not -have stdint.h (e.g. Solaris) may have inttypes.h. The macro int64_t may be set -by "configure". */ - -#if defined HAVE_STDINT_H -#include -#elif defined HAVE_INTTYPES_H -#include -#endif - -#if defined INT64_MAX || defined int64_t -#define INT64_OR_DOUBLE int64_t -#else -#define INT64_OR_DOUBLE double -#endif - -/* All character handling must be done as unsigned characters. Otherwise there -are problems with top-bit-set characters and functions such as isspace(). -However, we leave the interface to the outside world as char * or short *, -because that should make things easier for callers. This character type is -called pcre_uchar. - -The IN_UCHARS macro multiply its argument with the byte size of the current -pcre_uchar type. Useful for memcpy and such operations, whose require the -byte size of their input/output buffers. - -The MAX_255 macro checks whether its pcre_uchar input is less than 256. - -The TABLE_GET macro is designed for accessing elements of tables whose contain -exactly 256 items. When the character is able to contain more than 256 -items, some check is needed before accessing these tables. -*/ - -#if defined COMPILE_PCRE8 - -typedef unsigned char pcre_uchar; -#define IN_UCHARS(x) (x) -#define MAX_255(c) 1 -#define TABLE_GET(c, table, default) ((table)[c]) - -#elif defined COMPILE_PCRE16 - -#if USHRT_MAX != 65535 -/* This is a warning message. Change PCRE_UCHAR16 to a 16 bit data type in -pcre.h(.in) and disable (comment out) this message. */ -#error Warning: PCRE_UCHAR16 is not a 16 bit data type. -#endif - -typedef pcre_uint16 pcre_uchar; -#define UCHAR_SHIFT (1) -#define IN_UCHARS(x) ((x) << UCHAR_SHIFT) -#define MAX_255(c) ((c) <= 255u) -#define TABLE_GET(c, table, default) (MAX_255(c)? ((table)[c]):(default)) - -#elif defined COMPILE_PCRE32 - -typedef pcre_uint32 pcre_uchar; -#define UCHAR_SHIFT (2) -#define IN_UCHARS(x) ((x) << UCHAR_SHIFT) -#define MAX_255(c) ((c) <= 255u) -#define TABLE_GET(c, table, default) (MAX_255(c)? ((table)[c]):(default)) - -#else -#error Unsupported compiling mode -#endif /* COMPILE_PCRE[8|16|32] */ - -/* This is an unsigned int value that no character can ever have. UTF-8 -characters only go up to 0x7fffffff (though Unicode doesn't go beyond -0x0010ffff). */ - -#define NOTACHAR 0xffffffff - -/* PCRE is able to support several different kinds of newline (CR, LF, CRLF, -"any" and "anycrlf" at present). The following macros are used to package up -testing for newlines. NLBLOCK, PSSTART, and PSEND are defined in the various -modules to indicate in which datablock the parameters exist, and what the -start/end of string field names are. */ - -#define NLTYPE_FIXED 0 /* Newline is a fixed length string */ -#define NLTYPE_ANY 1 /* Newline is any Unicode line ending */ -#define NLTYPE_ANYCRLF 2 /* Newline is CR, LF, or CRLF */ - -/* This macro checks for a newline at the given position */ - -#define IS_NEWLINE(p) \ - ((NLBLOCK->nltype != NLTYPE_FIXED)? \ - ((p) < NLBLOCK->PSEND && \ - PRIV(is_newline)((p), NLBLOCK->nltype, NLBLOCK->PSEND, \ - &(NLBLOCK->nllen), utf)) \ - : \ - ((p) <= NLBLOCK->PSEND - NLBLOCK->nllen && \ - RAWUCHARTEST(p) == NLBLOCK->nl[0] && \ - (NLBLOCK->nllen == 1 || RAWUCHARTEST(p+1) == NLBLOCK->nl[1]) \ - ) \ - ) - -/* This macro checks for a newline immediately preceding the given position */ - -#define WAS_NEWLINE(p) \ - ((NLBLOCK->nltype != NLTYPE_FIXED)? \ - ((p) > NLBLOCK->PSSTART && \ - PRIV(was_newline)((p), NLBLOCK->nltype, NLBLOCK->PSSTART, \ - &(NLBLOCK->nllen), utf)) \ - : \ - ((p) >= NLBLOCK->PSSTART + NLBLOCK->nllen && \ - RAWUCHARTEST(p - NLBLOCK->nllen) == NLBLOCK->nl[0] && \ - (NLBLOCK->nllen == 1 || RAWUCHARTEST(p - NLBLOCK->nllen + 1) == NLBLOCK->nl[1]) \ - ) \ - ) - -/* When PCRE is compiled as a C++ library, the subject pointer can be replaced -with a custom type. This makes it possible, for example, to allow pcre_exec() -to process subject strings that are discontinuous by using a smart pointer -class. It must always be possible to inspect all of the subject string in -pcre_exec() because of the way it backtracks. Two macros are required in the -normal case, for sign-unspecified and unsigned char pointers. The former is -used for the external interface and appears in pcre.h, which is why its name -must begin with PCRE_. */ - -#ifdef CUSTOM_SUBJECT_PTR -#define PCRE_PUCHAR CUSTOM_SUBJECT_PTR -#else -#define PCRE_PUCHAR const pcre_uchar * -#endif - -/* Include the public PCRE header and the definitions of UCP character property -values. */ - -#include "pcre.h" -#include "ucp.h" - -#ifdef COMPILE_PCRE32 -/* Assert that the public PCRE_UCHAR32 is a 32-bit type */ -typedef int __assert_pcre_uchar32_size[sizeof(PCRE_UCHAR32) == 4 ? 1 : -1]; -#endif - -/* When compiling for use with the Virtual Pascal compiler, these functions -need to have their names changed. PCRE must be compiled with the -DVPCOMPAT -option on the command line. */ - -#ifdef VPCOMPAT -#define strlen(s) _strlen(s) -#define strncmp(s1,s2,m) _strncmp(s1,s2,m) -#define memcmp(s,c,n) _memcmp(s,c,n) -#define memcpy(d,s,n) _memcpy(d,s,n) -#define memmove(d,s,n) _memmove(d,s,n) -#define memset(s,c,n) _memset(s,c,n) -#else /* VPCOMPAT */ - -/* To cope with SunOS4 and other systems that lack memmove() but have bcopy(), -define a macro for memmove() if HAVE_MEMMOVE is false, provided that HAVE_BCOPY -is set. Otherwise, include an emulating function for those systems that have -neither (there some non-Unix environments where this is the case). */ - -#ifndef HAVE_MEMMOVE -#undef memmove /* some systems may have a macro */ -#ifdef HAVE_BCOPY -#define memmove(a, b, c) bcopy(b, a, c) -#else /* HAVE_BCOPY */ -static void * -pcre_memmove(void *d, const void *s, size_t n) -{ -size_t i; -unsigned char *dest = (unsigned char *)d; -const unsigned char *src = (const unsigned char *)s; -if (dest > src) - { - dest += n; - src += n; - for (i = 0; i < n; ++i) *(--dest) = *(--src); - return (void *)dest; - } -else - { - for (i = 0; i < n; ++i) *dest++ = *src++; - return (void *)(dest - n); - } -} -#define memmove(a, b, c) pcre_memmove(a, b, c) -#endif /* not HAVE_BCOPY */ -#endif /* not HAVE_MEMMOVE */ -#endif /* not VPCOMPAT */ - - -/* PCRE keeps offsets in its compiled code as 2-byte quantities (always stored -in big-endian order) by default. These are used, for example, to link from the -start of a subpattern to its alternatives and its end. The use of 2 bytes per -offset limits the size of the compiled regex to around 64K, which is big enough -for almost everybody. However, I received a request for an even bigger limit. -For this reason, and also to make the code easier to maintain, the storing and -loading of offsets from the byte string is now handled by the macros that are -defined here. - -The macros are controlled by the value of LINK_SIZE. This defaults to 2 in -the config.h file, but can be overridden by using -D on the command line. This -is automated on Unix systems via the "configure" command. */ - -#if defined COMPILE_PCRE8 - -#if LINK_SIZE == 2 - -#define PUT(a,n,d) \ - (a[n] = (d) >> 8), \ - (a[(n)+1] = (d) & 255) - -#define GET(a,n) \ - (((a)[n] << 8) | (a)[(n)+1]) - -#define MAX_PATTERN_SIZE (1 << 16) - - -#elif LINK_SIZE == 3 - -#define PUT(a,n,d) \ - (a[n] = (d) >> 16), \ - (a[(n)+1] = (d) >> 8), \ - (a[(n)+2] = (d) & 255) - -#define GET(a,n) \ - (((a)[n] << 16) | ((a)[(n)+1] << 8) | (a)[(n)+2]) - -#define MAX_PATTERN_SIZE (1 << 24) - - -#elif LINK_SIZE == 4 - -#define PUT(a,n,d) \ - (a[n] = (d) >> 24), \ - (a[(n)+1] = (d) >> 16), \ - (a[(n)+2] = (d) >> 8), \ - (a[(n)+3] = (d) & 255) - -#define GET(a,n) \ - (((a)[n] << 24) | ((a)[(n)+1] << 16) | ((a)[(n)+2] << 8) | (a)[(n)+3]) - -/* Keep it positive */ -#define MAX_PATTERN_SIZE (1 << 30) - -#else -#error LINK_SIZE must be either 2, 3, or 4 -#endif - -#elif defined COMPILE_PCRE16 - -#if LINK_SIZE == 2 - -/* Redefine LINK_SIZE as a multiple of sizeof(pcre_uchar) */ -#undef LINK_SIZE -#define LINK_SIZE 1 - -#define PUT(a,n,d) \ - (a[n] = (d)) - -#define GET(a,n) \ - (a[n]) - -#define MAX_PATTERN_SIZE (1 << 16) - -#elif LINK_SIZE == 3 || LINK_SIZE == 4 - -/* Redefine LINK_SIZE as a multiple of sizeof(pcre_uchar) */ -#undef LINK_SIZE -#define LINK_SIZE 2 - -#define PUT(a,n,d) \ - (a[n] = (d) >> 16), \ - (a[(n)+1] = (d) & 65535) - -#define GET(a,n) \ - (((a)[n] << 16) | (a)[(n)+1]) - -/* Keep it positive */ -#define MAX_PATTERN_SIZE (1 << 30) - -#else -#error LINK_SIZE must be either 2, 3, or 4 -#endif - -#elif defined COMPILE_PCRE32 - -/* Only supported LINK_SIZE is 4 */ -/* Redefine LINK_SIZE as a multiple of sizeof(pcre_uchar) */ -#undef LINK_SIZE -#define LINK_SIZE 1 - -#define PUT(a,n,d) \ - (a[n] = (d)) - -#define GET(a,n) \ - (a[n]) - -/* Keep it positive */ -#define MAX_PATTERN_SIZE (1 << 30) - -#else -#error Unsupported compiling mode -#endif /* COMPILE_PCRE[8|16|32] */ - -/* Convenience macro defined in terms of the others */ - -#define PUTINC(a,n,d) PUT(a,n,d), a += LINK_SIZE - - -/* PCRE uses some other 2-byte quantities that do not change when the size of -offsets changes. There are used for repeat counts and for other things such as -capturing parenthesis numbers in back references. */ - -#if defined COMPILE_PCRE8 - -#define IMM2_SIZE 2 - -#define PUT2(a,n,d) \ - a[n] = (d) >> 8; \ - a[(n)+1] = (d) & 255 - -/* For reasons that I do not understand, the expression in this GET2 macro is -treated by gcc as a signed expression, even when a is declared as unsigned. It -seems that any kind of arithmetic results in a signed value. */ - -#define GET2(a,n) \ - (unsigned int)(((a)[n] << 8) | (a)[(n)+1]) - -#elif defined COMPILE_PCRE16 - -#define IMM2_SIZE 1 - -#define PUT2(a,n,d) \ - a[n] = d - -#define GET2(a,n) \ - a[n] - -#elif defined COMPILE_PCRE32 - -#define IMM2_SIZE 1 - -#define PUT2(a,n,d) \ - a[n] = d - -#define GET2(a,n) \ - a[n] - -#else -#error Unsupported compiling mode -#endif /* COMPILE_PCRE[8|16|32] */ - -#define PUT2INC(a,n,d) PUT2(a,n,d), a += IMM2_SIZE - -/* The maximum length of a MARK name is currently one data unit; it may be -changed in future to be a fixed number of bytes or to depend on LINK_SIZE. */ - -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 -#define MAX_MARK ((1u << 16) - 1) -#else -#define MAX_MARK ((1u << 8) - 1) -#endif - -/* When UTF encoding is being used, a character is no longer just a single -byte. The macros for character handling generate simple sequences when used in -character-mode, and more complicated ones for UTF characters. GETCHARLENTEST -and other macros are not used when UTF is not supported, so they are not -defined. To make sure they can never even appear when UTF support is omitted, -we don't even define them. */ - -#ifndef SUPPORT_UTF - -/* #define MAX_VALUE_FOR_SINGLE_CHAR */ -/* #define HAS_EXTRALEN(c) */ -/* #define GET_EXTRALEN(c) */ -/* #define NOT_FIRSTCHAR(c) */ -#define GETCHAR(c, eptr) c = *eptr; -#define GETCHARTEST(c, eptr) c = *eptr; -#define GETCHARINC(c, eptr) c = *eptr++; -#define GETCHARINCTEST(c, eptr) c = *eptr++; -#define GETCHARLEN(c, eptr, len) c = *eptr; -#define RAWUCHAR(eptr) (*(eptr)) -#define RAWUCHARINC(eptr) (*(eptr)++) -#define RAWUCHARTEST(eptr) (*(eptr)) -#define RAWUCHARINCTEST(eptr) (*(eptr)++) -/* #define GETCHARLENTEST(c, eptr, len) */ -/* #define BACKCHAR(eptr) */ -/* #define FORWARDCHAR(eptr) */ -/* #define ACROSSCHAR(condition, eptr, action) */ - -#else /* SUPPORT_UTF */ - -/* Tests whether the code point needs extra characters to decode. */ - -#define HASUTF8EXTRALEN(c) ((c) >= 0xc0) - -/* Base macro to pick up the remaining bytes of a UTF-8 character, not -advancing the pointer. */ - -#define GETUTF8(c, eptr) \ - { \ - if ((c & 0x20) == 0) \ - c = ((c & 0x1f) << 6) | (eptr[1] & 0x3f); \ - else if ((c & 0x10) == 0) \ - c = ((c & 0x0f) << 12) | ((eptr[1] & 0x3f) << 6) | (eptr[2] & 0x3f); \ - else if ((c & 0x08) == 0) \ - c = ((c & 0x07) << 18) | ((eptr[1] & 0x3f) << 12) | \ - ((eptr[2] & 0x3f) << 6) | (eptr[3] & 0x3f); \ - else if ((c & 0x04) == 0) \ - c = ((c & 0x03) << 24) | ((eptr[1] & 0x3f) << 18) | \ - ((eptr[2] & 0x3f) << 12) | ((eptr[3] & 0x3f) << 6) | \ - (eptr[4] & 0x3f); \ - else \ - c = ((c & 0x01) << 30) | ((eptr[1] & 0x3f) << 24) | \ - ((eptr[2] & 0x3f) << 18) | ((eptr[3] & 0x3f) << 12) | \ - ((eptr[4] & 0x3f) << 6) | (eptr[5] & 0x3f); \ - } - -/* Base macro to pick up the remaining bytes of a UTF-8 character, advancing -the pointer. */ - -#define GETUTF8INC(c, eptr) \ - { \ - if ((c & 0x20) == 0) \ - c = ((c & 0x1f) << 6) | (*eptr++ & 0x3f); \ - else if ((c & 0x10) == 0) \ - { \ - c = ((c & 0x0f) << 12) | ((*eptr & 0x3f) << 6) | (eptr[1] & 0x3f); \ - eptr += 2; \ - } \ - else if ((c & 0x08) == 0) \ - { \ - c = ((c & 0x07) << 18) | ((*eptr & 0x3f) << 12) | \ - ((eptr[1] & 0x3f) << 6) | (eptr[2] & 0x3f); \ - eptr += 3; \ - } \ - else if ((c & 0x04) == 0) \ - { \ - c = ((c & 0x03) << 24) | ((*eptr & 0x3f) << 18) | \ - ((eptr[1] & 0x3f) << 12) | ((eptr[2] & 0x3f) << 6) | \ - (eptr[3] & 0x3f); \ - eptr += 4; \ - } \ - else \ - { \ - c = ((c & 0x01) << 30) | ((*eptr & 0x3f) << 24) | \ - ((eptr[1] & 0x3f) << 18) | ((eptr[2] & 0x3f) << 12) | \ - ((eptr[3] & 0x3f) << 6) | (eptr[4] & 0x3f); \ - eptr += 5; \ - } \ - } - -#if defined COMPILE_PCRE8 - -/* These macros were originally written in the form of loops that used data -from the tables whose names start with PRIV(utf8_table). They were rewritten by -a user so as not to use loops, because in some environments this gives a -significant performance advantage, and it seems never to do any harm. */ - -/* Tells the biggest code point which can be encoded as a single character. */ - -#define MAX_VALUE_FOR_SINGLE_CHAR 127 - -/* Tests whether the code point needs extra characters to decode. */ - -#define HAS_EXTRALEN(c) ((c) >= 0xc0) - -/* Returns with the additional number of characters if IS_MULTICHAR(c) is TRUE. -Otherwise it has an undefined behaviour. */ - -#define GET_EXTRALEN(c) (PRIV(utf8_table4)[(c) & 0x3f]) - -/* Returns TRUE, if the given character is not the first character -of a UTF sequence. */ - -#define NOT_FIRSTCHAR(c) (((c) & 0xc0) == 0x80) - -/* Get the next UTF-8 character, not advancing the pointer. This is called when -we know we are in UTF-8 mode. */ - -#define GETCHAR(c, eptr) \ - c = *eptr; \ - if (c >= 0xc0) GETUTF8(c, eptr); - -/* Get the next UTF-8 character, testing for UTF-8 mode, and not advancing the -pointer. */ - -#define GETCHARTEST(c, eptr) \ - c = *eptr; \ - if (utf && c >= 0xc0) GETUTF8(c, eptr); - -/* Get the next UTF-8 character, advancing the pointer. This is called when we -know we are in UTF-8 mode. */ - -#define GETCHARINC(c, eptr) \ - c = *eptr++; \ - if (c >= 0xc0) GETUTF8INC(c, eptr); - -/* Get the next character, testing for UTF-8 mode, and advancing the pointer. -This is called when we don't know if we are in UTF-8 mode. */ - -#define GETCHARINCTEST(c, eptr) \ - c = *eptr++; \ - if (utf && c >= 0xc0) GETUTF8INC(c, eptr); - -/* Base macro to pick up the remaining bytes of a UTF-8 character, not -advancing the pointer, incrementing the length. */ - -#define GETUTF8LEN(c, eptr, len) \ - { \ - if ((c & 0x20) == 0) \ - { \ - c = ((c & 0x1f) << 6) | (eptr[1] & 0x3f); \ - len++; \ - } \ - else if ((c & 0x10) == 0) \ - { \ - c = ((c & 0x0f) << 12) | ((eptr[1] & 0x3f) << 6) | (eptr[2] & 0x3f); \ - len += 2; \ - } \ - else if ((c & 0x08) == 0) \ - {\ - c = ((c & 0x07) << 18) | ((eptr[1] & 0x3f) << 12) | \ - ((eptr[2] & 0x3f) << 6) | (eptr[3] & 0x3f); \ - len += 3; \ - } \ - else if ((c & 0x04) == 0) \ - { \ - c = ((c & 0x03) << 24) | ((eptr[1] & 0x3f) << 18) | \ - ((eptr[2] & 0x3f) << 12) | ((eptr[3] & 0x3f) << 6) | \ - (eptr[4] & 0x3f); \ - len += 4; \ - } \ - else \ - {\ - c = ((c & 0x01) << 30) | ((eptr[1] & 0x3f) << 24) | \ - ((eptr[2] & 0x3f) << 18) | ((eptr[3] & 0x3f) << 12) | \ - ((eptr[4] & 0x3f) << 6) | (eptr[5] & 0x3f); \ - len += 5; \ - } \ - } - -/* Get the next UTF-8 character, not advancing the pointer, incrementing length -if there are extra bytes. This is called when we know we are in UTF-8 mode. */ - -#define GETCHARLEN(c, eptr, len) \ - c = *eptr; \ - if (c >= 0xc0) GETUTF8LEN(c, eptr, len); - -/* Get the next UTF-8 character, testing for UTF-8 mode, not advancing the -pointer, incrementing length if there are extra bytes. This is called when we -do not know if we are in UTF-8 mode. */ - -#define GETCHARLENTEST(c, eptr, len) \ - c = *eptr; \ - if (utf && c >= 0xc0) GETUTF8LEN(c, eptr, len); - -/* Returns the next uchar, not advancing the pointer. This is called when -we know we are in UTF mode. */ - -#define RAWUCHAR(eptr) \ - (*(eptr)) - -/* Returns the next uchar, advancing the pointer. This is called when -we know we are in UTF mode. */ - -#define RAWUCHARINC(eptr) \ - (*((eptr)++)) - -/* Returns the next uchar, testing for UTF mode, and not advancing the -pointer. */ - -#define RAWUCHARTEST(eptr) \ - (*(eptr)) - -/* Returns the next uchar, testing for UTF mode, advancing the -pointer. */ - -#define RAWUCHARINCTEST(eptr) \ - (*((eptr)++)) - -/* If the pointer is not at the start of a character, move it back until -it is. This is called only in UTF-8 mode - we don't put a test within the macro -because almost all calls are already within a block of UTF-8 only code. */ - -#define BACKCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr-- - -/* Same as above, just in the other direction. */ -#define FORWARDCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr++ - -/* Same as above, but it allows a fully customizable form. */ -#define ACROSSCHAR(condition, eptr, action) \ - while((condition) && ((eptr) & 0xc0) == 0x80) action - -#elif defined COMPILE_PCRE16 - -/* Tells the biggest code point which can be encoded as a single character. */ - -#define MAX_VALUE_FOR_SINGLE_CHAR 65535 - -/* Tests whether the code point needs extra characters to decode. */ - -#define HAS_EXTRALEN(c) (((c) & 0xfc00) == 0xd800) - -/* Returns with the additional number of characters if IS_MULTICHAR(c) is TRUE. -Otherwise it has an undefined behaviour. */ - -#define GET_EXTRALEN(c) 1 - -/* Returns TRUE, if the given character is not the first character -of a UTF sequence. */ - -#define NOT_FIRSTCHAR(c) (((c) & 0xfc00) == 0xdc00) - -/* Base macro to pick up the low surrogate of a UTF-16 character, not -advancing the pointer. */ - -#define GETUTF16(c, eptr) \ - { c = (((c & 0x3ff) << 10) | (eptr[1] & 0x3ff)) + 0x10000; } - -/* Get the next UTF-16 character, not advancing the pointer. This is called when -we know we are in UTF-16 mode. */ - -#define GETCHAR(c, eptr) \ - c = *eptr; \ - if ((c & 0xfc00) == 0xd800) GETUTF16(c, eptr); - -/* Get the next UTF-16 character, testing for UTF-16 mode, and not advancing the -pointer. */ - -#define GETCHARTEST(c, eptr) \ - c = *eptr; \ - if (utf && (c & 0xfc00) == 0xd800) GETUTF16(c, eptr); - -/* Base macro to pick up the low surrogate of a UTF-16 character, advancing -the pointer. */ - -#define GETUTF16INC(c, eptr) \ - { c = (((c & 0x3ff) << 10) | (*eptr++ & 0x3ff)) + 0x10000; } - -/* Get the next UTF-16 character, advancing the pointer. This is called when we -know we are in UTF-16 mode. */ - -#define GETCHARINC(c, eptr) \ - c = *eptr++; \ - if ((c & 0xfc00) == 0xd800) GETUTF16INC(c, eptr); - -/* Get the next character, testing for UTF-16 mode, and advancing the pointer. -This is called when we don't know if we are in UTF-16 mode. */ - -#define GETCHARINCTEST(c, eptr) \ - c = *eptr++; \ - if (utf && (c & 0xfc00) == 0xd800) GETUTF16INC(c, eptr); - -/* Base macro to pick up the low surrogate of a UTF-16 character, not -advancing the pointer, incrementing the length. */ - -#define GETUTF16LEN(c, eptr, len) \ - { c = (((c & 0x3ff) << 10) | (eptr[1] & 0x3ff)) + 0x10000; len++; } - -/* Get the next UTF-16 character, not advancing the pointer, incrementing -length if there is a low surrogate. This is called when we know we are in -UTF-16 mode. */ - -#define GETCHARLEN(c, eptr, len) \ - c = *eptr; \ - if ((c & 0xfc00) == 0xd800) GETUTF16LEN(c, eptr, len); - -/* Get the next UTF-816character, testing for UTF-16 mode, not advancing the -pointer, incrementing length if there is a low surrogate. This is called when -we do not know if we are in UTF-16 mode. */ - -#define GETCHARLENTEST(c, eptr, len) \ - c = *eptr; \ - if (utf && (c & 0xfc00) == 0xd800) GETUTF16LEN(c, eptr, len); - -/* Returns the next uchar, not advancing the pointer. This is called when -we know we are in UTF mode. */ - -#define RAWUCHAR(eptr) \ - (*(eptr)) - -/* Returns the next uchar, advancing the pointer. This is called when -we know we are in UTF mode. */ - -#define RAWUCHARINC(eptr) \ - (*((eptr)++)) - -/* Returns the next uchar, testing for UTF mode, and not advancing the -pointer. */ - -#define RAWUCHARTEST(eptr) \ - (*(eptr)) - -/* Returns the next uchar, testing for UTF mode, advancing the -pointer. */ - -#define RAWUCHARINCTEST(eptr) \ - (*((eptr)++)) - -/* If the pointer is not at the start of a character, move it back until -it is. This is called only in UTF-16 mode - we don't put a test within the -macro because almost all calls are already within a block of UTF-16 only -code. */ - -#define BACKCHAR(eptr) if ((*eptr & 0xfc00) == 0xdc00) eptr-- - -/* Same as above, just in the other direction. */ -#define FORWARDCHAR(eptr) if ((*eptr & 0xfc00) == 0xdc00) eptr++ - -/* Same as above, but it allows a fully customizable form. */ -#define ACROSSCHAR(condition, eptr, action) \ - if ((condition) && ((eptr) & 0xfc00) == 0xdc00) action - -#elif defined COMPILE_PCRE32 - -/* These are trivial for the 32-bit library, since all UTF-32 characters fit -into one pcre_uchar unit. */ -#define MAX_VALUE_FOR_SINGLE_CHAR (0x10ffffu) -#define HAS_EXTRALEN(c) (0) -#define GET_EXTRALEN(c) (0) -#define NOT_FIRSTCHAR(c) (0) - -/* Get the next UTF-32 character, not advancing the pointer. This is called when -we know we are in UTF-32 mode. */ - -#define GETCHAR(c, eptr) \ - c = *(eptr); - -/* Get the next UTF-32 character, testing for UTF-32 mode, and not advancing the -pointer. */ - -#define GETCHARTEST(c, eptr) \ - c = *(eptr); - -/* Get the next UTF-32 character, advancing the pointer. This is called when we -know we are in UTF-32 mode. */ - -#define GETCHARINC(c, eptr) \ - c = *((eptr)++); - -/* Get the next character, testing for UTF-32 mode, and advancing the pointer. -This is called when we don't know if we are in UTF-32 mode. */ - -#define GETCHARINCTEST(c, eptr) \ - c = *((eptr)++); - -/* Get the next UTF-32 character, not advancing the pointer, not incrementing -length (since all UTF-32 is of length 1). This is called when we know we are in -UTF-32 mode. */ - -#define GETCHARLEN(c, eptr, len) \ - GETCHAR(c, eptr) - -/* Get the next UTF-32character, testing for UTF-32 mode, not advancing the -pointer, not incrementing the length (since all UTF-32 is of length 1). -This is called when we do not know if we are in UTF-32 mode. */ - -#define GETCHARLENTEST(c, eptr, len) \ - GETCHARTEST(c, eptr) - -/* Returns the next uchar, not advancing the pointer. This is called when -we know we are in UTF mode. */ - -#define RAWUCHAR(eptr) \ - (*(eptr)) - -/* Returns the next uchar, advancing the pointer. This is called when -we know we are in UTF mode. */ - -#define RAWUCHARINC(eptr) \ - (*((eptr)++)) - -/* Returns the next uchar, testing for UTF mode, and not advancing the -pointer. */ - -#define RAWUCHARTEST(eptr) \ - (*(eptr)) - -/* Returns the next uchar, testing for UTF mode, advancing the -pointer. */ - -#define RAWUCHARINCTEST(eptr) \ - (*((eptr)++)) - -/* If the pointer is not at the start of a character, move it back until -it is. This is called only in UTF-32 mode - we don't put a test within the -macro because almost all calls are already within a block of UTF-32 only -code. -These are all no-ops since all UTF-32 characters fit into one pcre_uchar. */ - -#define BACKCHAR(eptr) do { } while (0) - -/* Same as above, just in the other direction. */ -#define FORWARDCHAR(eptr) do { } while (0) - -/* Same as above, but it allows a fully customizable form. */ -#define ACROSSCHAR(condition, eptr, action) do { } while (0) - -#else -#error Unsupported compiling mode -#endif /* COMPILE_PCRE[8|16|32] */ - -#endif /* SUPPORT_UTF */ - -/* Tests for Unicode horizontal and vertical whitespace characters must check a -number of different values. Using a switch statement for this generates the -fastest code (no loop, no memory access), and there are several places in the -interpreter code where this happens. In order to ensure that all the case lists -remain in step, we use macros so that there is only one place where the lists -are defined. - -These values are also required as lists in pcre_compile.c when processing \h, -\H, \v and \V in a character class. The lists are defined in pcre_tables.c, but -macros that define the values are here so that all the definitions are -together. The lists must be in ascending character order, terminated by -NOTACHAR (which is 0xffffffff). - -Any changes should ensure that the various macros are kept in step with each -other. NOTE: The values also appear in pcre_jit_compile.c. */ - -/* ------ ASCII/Unicode environments ------ */ - -#ifndef EBCDIC - -#define HSPACE_LIST \ - CHAR_HT, CHAR_SPACE, 0xa0, \ - 0x1680, 0x180e, 0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005, \ - 0x2006, 0x2007, 0x2008, 0x2009, 0x200A, 0x202f, 0x205f, 0x3000, \ - NOTACHAR - -#define HSPACE_MULTIBYTE_CASES \ - case 0x1680: /* OGHAM SPACE MARK */ \ - case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ \ - case 0x2000: /* EN QUAD */ \ - case 0x2001: /* EM QUAD */ \ - case 0x2002: /* EN SPACE */ \ - case 0x2003: /* EM SPACE */ \ - case 0x2004: /* THREE-PER-EM SPACE */ \ - case 0x2005: /* FOUR-PER-EM SPACE */ \ - case 0x2006: /* SIX-PER-EM SPACE */ \ - case 0x2007: /* FIGURE SPACE */ \ - case 0x2008: /* PUNCTUATION SPACE */ \ - case 0x2009: /* THIN SPACE */ \ - case 0x200A: /* HAIR SPACE */ \ - case 0x202f: /* NARROW NO-BREAK SPACE */ \ - case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ \ - case 0x3000 /* IDEOGRAPHIC SPACE */ - -#define HSPACE_BYTE_CASES \ - case CHAR_HT: \ - case CHAR_SPACE: \ - case 0xa0 /* NBSP */ - -#define HSPACE_CASES \ - HSPACE_BYTE_CASES: \ - HSPACE_MULTIBYTE_CASES - -#define VSPACE_LIST \ - CHAR_LF, CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, 0x2028, 0x2029, NOTACHAR - -#define VSPACE_MULTIBYTE_CASES \ - case 0x2028: /* LINE SEPARATOR */ \ - case 0x2029 /* PARAGRAPH SEPARATOR */ - -#define VSPACE_BYTE_CASES \ - case CHAR_LF: \ - case CHAR_VT: \ - case CHAR_FF: \ - case CHAR_CR: \ - case CHAR_NEL - -#define VSPACE_CASES \ - VSPACE_BYTE_CASES: \ - VSPACE_MULTIBYTE_CASES - -/* ------ EBCDIC environments ------ */ - -#else -#define HSPACE_LIST CHAR_HT, CHAR_SPACE - -#define HSPACE_BYTE_CASES \ - case CHAR_HT: \ - case CHAR_SPACE - -#define HSPACE_CASES HSPACE_BYTE_CASES - -#ifdef EBCDIC_NL25 -#define VSPACE_LIST \ - CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, CHAR_LF, NOTACHAR -#else -#define VSPACE_LIST \ - CHAR_VT, CHAR_FF, CHAR_CR, CHAR_LF, CHAR_NEL, NOTACHAR -#endif - -#define VSPACE_BYTE_CASES \ - case CHAR_LF: \ - case CHAR_VT: \ - case CHAR_FF: \ - case CHAR_CR: \ - case CHAR_NEL - -#define VSPACE_CASES VSPACE_BYTE_CASES -#endif /* EBCDIC */ - -/* ------ End of whitespace macros ------ */ - - - -/* Private flags containing information about the compiled regex. They used to -live at the top end of the options word, but that got almost full, so they were -moved to a 16-bit flags word - which got almost full, so now they are in a -32-bit flags word. From release 8.00, PCRE_NOPARTIAL is unused, as the -restrictions on partial matching have been lifted. It remains for backwards -compatibility. */ - -#define PCRE_MODE8 0x00000001 /* compiled in 8 bit mode */ -#define PCRE_MODE16 0x00000002 /* compiled in 16 bit mode */ -#define PCRE_MODE32 0x00000004 /* compiled in 32 bit mode */ -#define PCRE_FIRSTSET 0x00000010 /* first_char is set */ -#define PCRE_FCH_CASELESS 0x00000020 /* caseless first char */ -#define PCRE_REQCHSET 0x00000040 /* req_byte is set */ -#define PCRE_RCH_CASELESS 0x00000080 /* caseless requested char */ -#define PCRE_STARTLINE 0x00000100 /* start after \n for multiline */ -#define PCRE_NOPARTIAL 0x00000200 /* can't use partial with this regex */ -#define PCRE_JCHANGED 0x00000400 /* j option used in regex */ -#define PCRE_HASCRORLF 0x00000800 /* explicit \r or \n in pattern */ -#define PCRE_HASTHEN 0x00001000 /* pattern contains (*THEN) */ -#define PCRE_MLSET 0x00002000 /* match limit set by regex */ -#define PCRE_RLSET 0x00004000 /* recursion limit set by regex */ -#define PCRE_MATCH_EMPTY 0x00008000 /* pattern can match empty string */ - -#if defined COMPILE_PCRE8 -#define PCRE_MODE PCRE_MODE8 -#elif defined COMPILE_PCRE16 -#define PCRE_MODE PCRE_MODE16 -#elif defined COMPILE_PCRE32 -#define PCRE_MODE PCRE_MODE32 -#endif -#define PCRE_MODE_MASK (PCRE_MODE8 | PCRE_MODE16 | PCRE_MODE32) - -/* Flags for the "extra" block produced by pcre_study(). */ - -#define PCRE_STUDY_MAPPED 0x0001 /* a map of starting chars exists */ -#define PCRE_STUDY_MINLEN 0x0002 /* a minimum length field exists */ - -/* Masks for identifying the public options that are permitted at compile -time, run time, or study time, respectively. */ - -#define PCRE_NEWLINE_BITS (PCRE_NEWLINE_CR|PCRE_NEWLINE_LF|PCRE_NEWLINE_ANY| \ - PCRE_NEWLINE_ANYCRLF) - -#define PUBLIC_COMPILE_OPTIONS \ - (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \ - PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \ - PCRE_NO_AUTO_CAPTURE|PCRE_NO_AUTO_POSSESS| \ - PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE| \ - PCRE_DUPNAMES|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \ - PCRE_JAVASCRIPT_COMPAT|PCRE_UCP|PCRE_NO_START_OPTIMIZE|PCRE_NEVER_UTF) - -#define PUBLIC_EXEC_OPTIONS \ - (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NOTEMPTY_ATSTART| \ - PCRE_NO_UTF8_CHECK|PCRE_PARTIAL_HARD|PCRE_PARTIAL_SOFT|PCRE_NEWLINE_BITS| \ - PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE|PCRE_NO_START_OPTIMIZE) - -#define PUBLIC_DFA_EXEC_OPTIONS \ - (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NOTEMPTY_ATSTART| \ - PCRE_NO_UTF8_CHECK|PCRE_PARTIAL_HARD|PCRE_PARTIAL_SOFT|PCRE_DFA_SHORTEST| \ - PCRE_DFA_RESTART|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \ - PCRE_NO_START_OPTIMIZE) - -#define PUBLIC_STUDY_OPTIONS \ - (PCRE_STUDY_JIT_COMPILE|PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE| \ - PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE|PCRE_STUDY_EXTRA_NEEDED) - -#define PUBLIC_JIT_EXEC_OPTIONS \ - (PCRE_NO_UTF8_CHECK|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|\ - PCRE_NOTEMPTY_ATSTART|PCRE_PARTIAL_SOFT|PCRE_PARTIAL_HARD) - -/* Magic number to provide a small check against being handed junk. */ - -#define MAGIC_NUMBER 0x50435245UL /* 'PCRE' */ - -/* This variable is used to detect a loaded regular expression -in different endianness. */ - -#define REVERSED_MAGIC_NUMBER 0x45524350UL /* 'ERCP' */ - -/* The maximum remaining length of subject we are prepared to search for a -req_byte match. */ - -#define REQ_BYTE_MAX 1000 - -/* Miscellaneous definitions. The #ifndef is to pacify compiler warnings in -environments where these macros are defined elsewhere. Unfortunately, there -is no way to do the same for the typedef. */ - -typedef int BOOL; - -#ifndef FALSE -#define FALSE 0 -#define TRUE 1 -#endif - -/* If PCRE is to support UTF-8 on EBCDIC platforms, we cannot use normal -character constants like '*' because the compiler would emit their EBCDIC code, -which is different from their ASCII/UTF-8 code. Instead we define macros for -the characters so that they always use the ASCII/UTF-8 code when UTF-8 support -is enabled. When UTF-8 support is not enabled, the definitions use character -literals. Both character and string versions of each character are needed, and -there are some longer strings as well. - -This means that, on EBCDIC platforms, the PCRE library can handle either -EBCDIC, or UTF-8, but not both. To support both in the same compiled library -would need different lookups depending on whether PCRE_UTF8 was set or not. -This would make it impossible to use characters in switch/case statements, -which would reduce performance. For a theoretical use (which nobody has asked -for) in a minority area (EBCDIC platforms), this is not sensible. Any -application that did need both could compile two versions of the library, using -macros to give the functions distinct names. */ - -#ifndef SUPPORT_UTF - -/* UTF-8 support is not enabled; use the platform-dependent character literals -so that PCRE works in both ASCII and EBCDIC environments, but only in non-UTF -mode. Newline characters are problematic in EBCDIC. Though it has CR and LF -characters, a common practice has been to use its NL (0x15) character as the -line terminator in C-like processing environments. However, sometimes the LF -(0x25) character is used instead, according to this Unicode document: - -http://unicode.org/standard/reports/tr13/tr13-5.html - -PCRE defaults EBCDIC NL to 0x15, but has a build-time option to select 0x25 -instead. Whichever is *not* chosen is defined as NEL. - -In both ASCII and EBCDIC environments, CHAR_NL and CHAR_LF are synonyms for the -same code point. */ - -#ifdef EBCDIC - -#ifndef EBCDIC_NL25 -#define CHAR_NL '\x15' -#define CHAR_NEL '\x25' -#define STR_NL "\x15" -#define STR_NEL "\x25" -#else -#define CHAR_NL '\x25' -#define CHAR_NEL '\x15' -#define STR_NL "\x25" -#define STR_NEL "\x15" -#endif - -#define CHAR_LF CHAR_NL -#define STR_LF STR_NL - -#define CHAR_ESC '\047' -#define CHAR_DEL '\007' -#define STR_ESC "\047" -#define STR_DEL "\007" - -#else /* Not EBCDIC */ - -/* In ASCII/Unicode, linefeed is '\n' and we equate this to NL for -compatibility. NEL is the Unicode newline character; make sure it is -a positive value. */ - -#define CHAR_LF '\n' -#define CHAR_NL CHAR_LF -#define CHAR_NEL ((unsigned char)'\x85') -#define CHAR_ESC '\033' -#define CHAR_DEL '\177' - -#define STR_LF "\n" -#define STR_NL STR_LF -#define STR_NEL "\x85" -#define STR_ESC "\033" -#define STR_DEL "\177" - -#endif /* EBCDIC */ - -/* The remaining definitions work in both environments. */ - -#define CHAR_NULL '\0' -#define CHAR_HT '\t' -#define CHAR_VT '\v' -#define CHAR_FF '\f' -#define CHAR_CR '\r' -#define CHAR_BS '\b' -#define CHAR_BEL '\a' - -#define CHAR_SPACE ' ' -#define CHAR_EXCLAMATION_MARK '!' -#define CHAR_QUOTATION_MARK '"' -#define CHAR_NUMBER_SIGN '#' -#define CHAR_DOLLAR_SIGN '$' -#define CHAR_PERCENT_SIGN '%' -#define CHAR_AMPERSAND '&' -#define CHAR_APOSTROPHE '\'' -#define CHAR_LEFT_PARENTHESIS '(' -#define CHAR_RIGHT_PARENTHESIS ')' -#define CHAR_ASTERISK '*' -#define CHAR_PLUS '+' -#define CHAR_COMMA ',' -#define CHAR_MINUS '-' -#define CHAR_DOT '.' -#define CHAR_SLASH '/' -#define CHAR_0 '0' -#define CHAR_1 '1' -#define CHAR_2 '2' -#define CHAR_3 '3' -#define CHAR_4 '4' -#define CHAR_5 '5' -#define CHAR_6 '6' -#define CHAR_7 '7' -#define CHAR_8 '8' -#define CHAR_9 '9' -#define CHAR_COLON ':' -#define CHAR_SEMICOLON ';' -#define CHAR_LESS_THAN_SIGN '<' -#define CHAR_EQUALS_SIGN '=' -#define CHAR_GREATER_THAN_SIGN '>' -#define CHAR_QUESTION_MARK '?' -#define CHAR_COMMERCIAL_AT '@' -#define CHAR_A 'A' -#define CHAR_B 'B' -#define CHAR_C 'C' -#define CHAR_D 'D' -#define CHAR_E 'E' -#define CHAR_F 'F' -#define CHAR_G 'G' -#define CHAR_H 'H' -#define CHAR_I 'I' -#define CHAR_J 'J' -#define CHAR_K 'K' -#define CHAR_L 'L' -#define CHAR_M 'M' -#define CHAR_N 'N' -#define CHAR_O 'O' -#define CHAR_P 'P' -#define CHAR_Q 'Q' -#define CHAR_R 'R' -#define CHAR_S 'S' -#define CHAR_T 'T' -#define CHAR_U 'U' -#define CHAR_V 'V' -#define CHAR_W 'W' -#define CHAR_X 'X' -#define CHAR_Y 'Y' -#define CHAR_Z 'Z' -#define CHAR_LEFT_SQUARE_BRACKET '[' -#define CHAR_BACKSLASH '\\' -#define CHAR_RIGHT_SQUARE_BRACKET ']' -#define CHAR_CIRCUMFLEX_ACCENT '^' -#define CHAR_UNDERSCORE '_' -#define CHAR_GRAVE_ACCENT '`' -#define CHAR_a 'a' -#define CHAR_b 'b' -#define CHAR_c 'c' -#define CHAR_d 'd' -#define CHAR_e 'e' -#define CHAR_f 'f' -#define CHAR_g 'g' -#define CHAR_h 'h' -#define CHAR_i 'i' -#define CHAR_j 'j' -#define CHAR_k 'k' -#define CHAR_l 'l' -#define CHAR_m 'm' -#define CHAR_n 'n' -#define CHAR_o 'o' -#define CHAR_p 'p' -#define CHAR_q 'q' -#define CHAR_r 'r' -#define CHAR_s 's' -#define CHAR_t 't' -#define CHAR_u 'u' -#define CHAR_v 'v' -#define CHAR_w 'w' -#define CHAR_x 'x' -#define CHAR_y 'y' -#define CHAR_z 'z' -#define CHAR_LEFT_CURLY_BRACKET '{' -#define CHAR_VERTICAL_LINE '|' -#define CHAR_RIGHT_CURLY_BRACKET '}' -#define CHAR_TILDE '~' - -#define STR_HT "\t" -#define STR_VT "\v" -#define STR_FF "\f" -#define STR_CR "\r" -#define STR_BS "\b" -#define STR_BEL "\a" - -#define STR_SPACE " " -#define STR_EXCLAMATION_MARK "!" -#define STR_QUOTATION_MARK "\"" -#define STR_NUMBER_SIGN "#" -#define STR_DOLLAR_SIGN "$" -#define STR_PERCENT_SIGN "%" -#define STR_AMPERSAND "&" -#define STR_APOSTROPHE "'" -#define STR_LEFT_PARENTHESIS "(" -#define STR_RIGHT_PARENTHESIS ")" -#define STR_ASTERISK "*" -#define STR_PLUS "+" -#define STR_COMMA "," -#define STR_MINUS "-" -#define STR_DOT "." -#define STR_SLASH "/" -#define STR_0 "0" -#define STR_1 "1" -#define STR_2 "2" -#define STR_3 "3" -#define STR_4 "4" -#define STR_5 "5" -#define STR_6 "6" -#define STR_7 "7" -#define STR_8 "8" -#define STR_9 "9" -#define STR_COLON ":" -#define STR_SEMICOLON ";" -#define STR_LESS_THAN_SIGN "<" -#define STR_EQUALS_SIGN "=" -#define STR_GREATER_THAN_SIGN ">" -#define STR_QUESTION_MARK "?" -#define STR_COMMERCIAL_AT "@" -#define STR_A "A" -#define STR_B "B" -#define STR_C "C" -#define STR_D "D" -#define STR_E "E" -#define STR_F "F" -#define STR_G "G" -#define STR_H "H" -#define STR_I "I" -#define STR_J "J" -#define STR_K "K" -#define STR_L "L" -#define STR_M "M" -#define STR_N "N" -#define STR_O "O" -#define STR_P "P" -#define STR_Q "Q" -#define STR_R "R" -#define STR_S "S" -#define STR_T "T" -#define STR_U "U" -#define STR_V "V" -#define STR_W "W" -#define STR_X "X" -#define STR_Y "Y" -#define STR_Z "Z" -#define STR_LEFT_SQUARE_BRACKET "[" -#define STR_BACKSLASH "\\" -#define STR_RIGHT_SQUARE_BRACKET "]" -#define STR_CIRCUMFLEX_ACCENT "^" -#define STR_UNDERSCORE "_" -#define STR_GRAVE_ACCENT "`" -#define STR_a "a" -#define STR_b "b" -#define STR_c "c" -#define STR_d "d" -#define STR_e "e" -#define STR_f "f" -#define STR_g "g" -#define STR_h "h" -#define STR_i "i" -#define STR_j "j" -#define STR_k "k" -#define STR_l "l" -#define STR_m "m" -#define STR_n "n" -#define STR_o "o" -#define STR_p "p" -#define STR_q "q" -#define STR_r "r" -#define STR_s "s" -#define STR_t "t" -#define STR_u "u" -#define STR_v "v" -#define STR_w "w" -#define STR_x "x" -#define STR_y "y" -#define STR_z "z" -#define STR_LEFT_CURLY_BRACKET "{" -#define STR_VERTICAL_LINE "|" -#define STR_RIGHT_CURLY_BRACKET "}" -#define STR_TILDE "~" - -#define STRING_ACCEPT0 "ACCEPT\0" -#define STRING_COMMIT0 "COMMIT\0" -#define STRING_F0 "F\0" -#define STRING_FAIL0 "FAIL\0" -#define STRING_MARK0 "MARK\0" -#define STRING_PRUNE0 "PRUNE\0" -#define STRING_SKIP0 "SKIP\0" -#define STRING_THEN "THEN" - -#define STRING_alpha0 "alpha\0" -#define STRING_lower0 "lower\0" -#define STRING_upper0 "upper\0" -#define STRING_alnum0 "alnum\0" -#define STRING_ascii0 "ascii\0" -#define STRING_blank0 "blank\0" -#define STRING_cntrl0 "cntrl\0" -#define STRING_digit0 "digit\0" -#define STRING_graph0 "graph\0" -#define STRING_print0 "print\0" -#define STRING_punct0 "punct\0" -#define STRING_space0 "space\0" -#define STRING_word0 "word\0" -#define STRING_xdigit "xdigit" - -#define STRING_DEFINE "DEFINE" -#define STRING_WEIRD_STARTWORD "[:<:]]" -#define STRING_WEIRD_ENDWORD "[:>:]]" - -#define STRING_CR_RIGHTPAR "CR)" -#define STRING_LF_RIGHTPAR "LF)" -#define STRING_CRLF_RIGHTPAR "CRLF)" -#define STRING_ANY_RIGHTPAR "ANY)" -#define STRING_ANYCRLF_RIGHTPAR "ANYCRLF)" -#define STRING_BSR_ANYCRLF_RIGHTPAR "BSR_ANYCRLF)" -#define STRING_BSR_UNICODE_RIGHTPAR "BSR_UNICODE)" -#define STRING_UTF8_RIGHTPAR "UTF8)" -#define STRING_UTF16_RIGHTPAR "UTF16)" -#define STRING_UTF32_RIGHTPAR "UTF32)" -#define STRING_UTF_RIGHTPAR "UTF)" -#define STRING_UCP_RIGHTPAR "UCP)" -#define STRING_NO_AUTO_POSSESS_RIGHTPAR "NO_AUTO_POSSESS)" -#define STRING_NO_START_OPT_RIGHTPAR "NO_START_OPT)" -#define STRING_LIMIT_MATCH_EQ "LIMIT_MATCH=" -#define STRING_LIMIT_RECURSION_EQ "LIMIT_RECURSION=" - -#else /* SUPPORT_UTF */ - -/* UTF-8 support is enabled; always use UTF-8 (=ASCII) character codes. This -works in both modes non-EBCDIC platforms, and on EBCDIC platforms in UTF-8 mode -only. */ - -#define CHAR_HT '\011' -#define CHAR_VT '\013' -#define CHAR_FF '\014' -#define CHAR_CR '\015' -#define CHAR_LF '\012' -#define CHAR_NL CHAR_LF -#define CHAR_NEL ((unsigned char)'\x85') -#define CHAR_BS '\010' -#define CHAR_BEL '\007' -#define CHAR_ESC '\033' -#define CHAR_DEL '\177' - -#define CHAR_NULL '\0' -#define CHAR_SPACE '\040' -#define CHAR_EXCLAMATION_MARK '\041' -#define CHAR_QUOTATION_MARK '\042' -#define CHAR_NUMBER_SIGN '\043' -#define CHAR_DOLLAR_SIGN '\044' -#define CHAR_PERCENT_SIGN '\045' -#define CHAR_AMPERSAND '\046' -#define CHAR_APOSTROPHE '\047' -#define CHAR_LEFT_PARENTHESIS '\050' -#define CHAR_RIGHT_PARENTHESIS '\051' -#define CHAR_ASTERISK '\052' -#define CHAR_PLUS '\053' -#define CHAR_COMMA '\054' -#define CHAR_MINUS '\055' -#define CHAR_DOT '\056' -#define CHAR_SLASH '\057' -#define CHAR_0 '\060' -#define CHAR_1 '\061' -#define CHAR_2 '\062' -#define CHAR_3 '\063' -#define CHAR_4 '\064' -#define CHAR_5 '\065' -#define CHAR_6 '\066' -#define CHAR_7 '\067' -#define CHAR_8 '\070' -#define CHAR_9 '\071' -#define CHAR_COLON '\072' -#define CHAR_SEMICOLON '\073' -#define CHAR_LESS_THAN_SIGN '\074' -#define CHAR_EQUALS_SIGN '\075' -#define CHAR_GREATER_THAN_SIGN '\076' -#define CHAR_QUESTION_MARK '\077' -#define CHAR_COMMERCIAL_AT '\100' -#define CHAR_A '\101' -#define CHAR_B '\102' -#define CHAR_C '\103' -#define CHAR_D '\104' -#define CHAR_E '\105' -#define CHAR_F '\106' -#define CHAR_G '\107' -#define CHAR_H '\110' -#define CHAR_I '\111' -#define CHAR_J '\112' -#define CHAR_K '\113' -#define CHAR_L '\114' -#define CHAR_M '\115' -#define CHAR_N '\116' -#define CHAR_O '\117' -#define CHAR_P '\120' -#define CHAR_Q '\121' -#define CHAR_R '\122' -#define CHAR_S '\123' -#define CHAR_T '\124' -#define CHAR_U '\125' -#define CHAR_V '\126' -#define CHAR_W '\127' -#define CHAR_X '\130' -#define CHAR_Y '\131' -#define CHAR_Z '\132' -#define CHAR_LEFT_SQUARE_BRACKET '\133' -#define CHAR_BACKSLASH '\134' -#define CHAR_RIGHT_SQUARE_BRACKET '\135' -#define CHAR_CIRCUMFLEX_ACCENT '\136' -#define CHAR_UNDERSCORE '\137' -#define CHAR_GRAVE_ACCENT '\140' -#define CHAR_a '\141' -#define CHAR_b '\142' -#define CHAR_c '\143' -#define CHAR_d '\144' -#define CHAR_e '\145' -#define CHAR_f '\146' -#define CHAR_g '\147' -#define CHAR_h '\150' -#define CHAR_i '\151' -#define CHAR_j '\152' -#define CHAR_k '\153' -#define CHAR_l '\154' -#define CHAR_m '\155' -#define CHAR_n '\156' -#define CHAR_o '\157' -#define CHAR_p '\160' -#define CHAR_q '\161' -#define CHAR_r '\162' -#define CHAR_s '\163' -#define CHAR_t '\164' -#define CHAR_u '\165' -#define CHAR_v '\166' -#define CHAR_w '\167' -#define CHAR_x '\170' -#define CHAR_y '\171' -#define CHAR_z '\172' -#define CHAR_LEFT_CURLY_BRACKET '\173' -#define CHAR_VERTICAL_LINE '\174' -#define CHAR_RIGHT_CURLY_BRACKET '\175' -#define CHAR_TILDE '\176' - -#define STR_HT "\011" -#define STR_VT "\013" -#define STR_FF "\014" -#define STR_CR "\015" -#define STR_NL "\012" -#define STR_BS "\010" -#define STR_BEL "\007" -#define STR_ESC "\033" -#define STR_DEL "\177" - -#define STR_SPACE "\040" -#define STR_EXCLAMATION_MARK "\041" -#define STR_QUOTATION_MARK "\042" -#define STR_NUMBER_SIGN "\043" -#define STR_DOLLAR_SIGN "\044" -#define STR_PERCENT_SIGN "\045" -#define STR_AMPERSAND "\046" -#define STR_APOSTROPHE "\047" -#define STR_LEFT_PARENTHESIS "\050" -#define STR_RIGHT_PARENTHESIS "\051" -#define STR_ASTERISK "\052" -#define STR_PLUS "\053" -#define STR_COMMA "\054" -#define STR_MINUS "\055" -#define STR_DOT "\056" -#define STR_SLASH "\057" -#define STR_0 "\060" -#define STR_1 "\061" -#define STR_2 "\062" -#define STR_3 "\063" -#define STR_4 "\064" -#define STR_5 "\065" -#define STR_6 "\066" -#define STR_7 "\067" -#define STR_8 "\070" -#define STR_9 "\071" -#define STR_COLON "\072" -#define STR_SEMICOLON "\073" -#define STR_LESS_THAN_SIGN "\074" -#define STR_EQUALS_SIGN "\075" -#define STR_GREATER_THAN_SIGN "\076" -#define STR_QUESTION_MARK "\077" -#define STR_COMMERCIAL_AT "\100" -#define STR_A "\101" -#define STR_B "\102" -#define STR_C "\103" -#define STR_D "\104" -#define STR_E "\105" -#define STR_F "\106" -#define STR_G "\107" -#define STR_H "\110" -#define STR_I "\111" -#define STR_J "\112" -#define STR_K "\113" -#define STR_L "\114" -#define STR_M "\115" -#define STR_N "\116" -#define STR_O "\117" -#define STR_P "\120" -#define STR_Q "\121" -#define STR_R "\122" -#define STR_S "\123" -#define STR_T "\124" -#define STR_U "\125" -#define STR_V "\126" -#define STR_W "\127" -#define STR_X "\130" -#define STR_Y "\131" -#define STR_Z "\132" -#define STR_LEFT_SQUARE_BRACKET "\133" -#define STR_BACKSLASH "\134" -#define STR_RIGHT_SQUARE_BRACKET "\135" -#define STR_CIRCUMFLEX_ACCENT "\136" -#define STR_UNDERSCORE "\137" -#define STR_GRAVE_ACCENT "\140" -#define STR_a "\141" -#define STR_b "\142" -#define STR_c "\143" -#define STR_d "\144" -#define STR_e "\145" -#define STR_f "\146" -#define STR_g "\147" -#define STR_h "\150" -#define STR_i "\151" -#define STR_j "\152" -#define STR_k "\153" -#define STR_l "\154" -#define STR_m "\155" -#define STR_n "\156" -#define STR_o "\157" -#define STR_p "\160" -#define STR_q "\161" -#define STR_r "\162" -#define STR_s "\163" -#define STR_t "\164" -#define STR_u "\165" -#define STR_v "\166" -#define STR_w "\167" -#define STR_x "\170" -#define STR_y "\171" -#define STR_z "\172" -#define STR_LEFT_CURLY_BRACKET "\173" -#define STR_VERTICAL_LINE "\174" -#define STR_RIGHT_CURLY_BRACKET "\175" -#define STR_TILDE "\176" - -#define STRING_ACCEPT0 STR_A STR_C STR_C STR_E STR_P STR_T "\0" -#define STRING_COMMIT0 STR_C STR_O STR_M STR_M STR_I STR_T "\0" -#define STRING_F0 STR_F "\0" -#define STRING_FAIL0 STR_F STR_A STR_I STR_L "\0" -#define STRING_MARK0 STR_M STR_A STR_R STR_K "\0" -#define STRING_PRUNE0 STR_P STR_R STR_U STR_N STR_E "\0" -#define STRING_SKIP0 STR_S STR_K STR_I STR_P "\0" -#define STRING_THEN STR_T STR_H STR_E STR_N - -#define STRING_alpha0 STR_a STR_l STR_p STR_h STR_a "\0" -#define STRING_lower0 STR_l STR_o STR_w STR_e STR_r "\0" -#define STRING_upper0 STR_u STR_p STR_p STR_e STR_r "\0" -#define STRING_alnum0 STR_a STR_l STR_n STR_u STR_m "\0" -#define STRING_ascii0 STR_a STR_s STR_c STR_i STR_i "\0" -#define STRING_blank0 STR_b STR_l STR_a STR_n STR_k "\0" -#define STRING_cntrl0 STR_c STR_n STR_t STR_r STR_l "\0" -#define STRING_digit0 STR_d STR_i STR_g STR_i STR_t "\0" -#define STRING_graph0 STR_g STR_r STR_a STR_p STR_h "\0" -#define STRING_print0 STR_p STR_r STR_i STR_n STR_t "\0" -#define STRING_punct0 STR_p STR_u STR_n STR_c STR_t "\0" -#define STRING_space0 STR_s STR_p STR_a STR_c STR_e "\0" -#define STRING_word0 STR_w STR_o STR_r STR_d "\0" -#define STRING_xdigit STR_x STR_d STR_i STR_g STR_i STR_t - -#define STRING_DEFINE STR_D STR_E STR_F STR_I STR_N STR_E -#define STRING_WEIRD_STARTWORD STR_LEFT_SQUARE_BRACKET STR_COLON STR_LESS_THAN_SIGN STR_COLON STR_RIGHT_SQUARE_BRACKET STR_RIGHT_SQUARE_BRACKET -#define STRING_WEIRD_ENDWORD STR_LEFT_SQUARE_BRACKET STR_COLON STR_GREATER_THAN_SIGN STR_COLON STR_RIGHT_SQUARE_BRACKET STR_RIGHT_SQUARE_BRACKET - -#define STRING_CR_RIGHTPAR STR_C STR_R STR_RIGHT_PARENTHESIS -#define STRING_LF_RIGHTPAR STR_L STR_F STR_RIGHT_PARENTHESIS -#define STRING_CRLF_RIGHTPAR STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS -#define STRING_ANY_RIGHTPAR STR_A STR_N STR_Y STR_RIGHT_PARENTHESIS -#define STRING_ANYCRLF_RIGHTPAR STR_A STR_N STR_Y STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS -#define STRING_BSR_ANYCRLF_RIGHTPAR STR_B STR_S STR_R STR_UNDERSCORE STR_A STR_N STR_Y STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS -#define STRING_BSR_UNICODE_RIGHTPAR STR_B STR_S STR_R STR_UNDERSCORE STR_U STR_N STR_I STR_C STR_O STR_D STR_E STR_RIGHT_PARENTHESIS -#define STRING_UTF8_RIGHTPAR STR_U STR_T STR_F STR_8 STR_RIGHT_PARENTHESIS -#define STRING_UTF16_RIGHTPAR STR_U STR_T STR_F STR_1 STR_6 STR_RIGHT_PARENTHESIS -#define STRING_UTF32_RIGHTPAR STR_U STR_T STR_F STR_3 STR_2 STR_RIGHT_PARENTHESIS -#define STRING_UTF_RIGHTPAR STR_U STR_T STR_F STR_RIGHT_PARENTHESIS -#define STRING_UCP_RIGHTPAR STR_U STR_C STR_P STR_RIGHT_PARENTHESIS -#define STRING_NO_AUTO_POSSESS_RIGHTPAR STR_N STR_O STR_UNDERSCORE STR_A STR_U STR_T STR_O STR_UNDERSCORE STR_P STR_O STR_S STR_S STR_E STR_S STR_S STR_RIGHT_PARENTHESIS -#define STRING_NO_START_OPT_RIGHTPAR STR_N STR_O STR_UNDERSCORE STR_S STR_T STR_A STR_R STR_T STR_UNDERSCORE STR_O STR_P STR_T STR_RIGHT_PARENTHESIS -#define STRING_LIMIT_MATCH_EQ STR_L STR_I STR_M STR_I STR_T STR_UNDERSCORE STR_M STR_A STR_T STR_C STR_H STR_EQUALS_SIGN -#define STRING_LIMIT_RECURSION_EQ STR_L STR_I STR_M STR_I STR_T STR_UNDERSCORE STR_R STR_E STR_C STR_U STR_R STR_S STR_I STR_O STR_N STR_EQUALS_SIGN - -#endif /* SUPPORT_UTF */ - -/* Escape items that are just an encoding of a particular data value. */ - -#ifndef ESC_e -#define ESC_e CHAR_ESC -#endif - -#ifndef ESC_f -#define ESC_f CHAR_FF -#endif - -#ifndef ESC_n -#define ESC_n CHAR_LF -#endif - -#ifndef ESC_r -#define ESC_r CHAR_CR -#endif - -/* We can't officially use ESC_t because it is a POSIX reserved identifier -(presumably because of all the others like size_t). */ - -#ifndef ESC_tee -#define ESC_tee CHAR_HT -#endif - -/* Codes for different types of Unicode property */ - -#define PT_ANY 0 /* Any property - matches all chars */ -#define PT_LAMP 1 /* L& - the union of Lu, Ll, Lt */ -#define PT_GC 2 /* Specified general characteristic (e.g. L) */ -#define PT_PC 3 /* Specified particular characteristic (e.g. Lu) */ -#define PT_SC 4 /* Script (e.g. Han) */ -#define PT_ALNUM 5 /* Alphanumeric - the union of L and N */ -#define PT_SPACE 6 /* Perl space - Z plus 9,10,12,13 */ -#define PT_PXSPACE 7 /* POSIX space - Z plus 9,10,11,12,13 */ -#define PT_WORD 8 /* Word - L plus N plus underscore */ -#define PT_CLIST 9 /* Pseudo-property: match character list */ -#define PT_UCNC 10 /* Universal Character nameable character */ -#define PT_TABSIZE 11 /* Size of square table for autopossessify tests */ - -/* The following special properties are used only in XCLASS items, when POSIX -classes are specified and PCRE_UCP is set - in other words, for Unicode -handling of these classes. They are not available via the \p or \P escapes like -those in the above list, and so they do not take part in the autopossessifying -table. */ - -#define PT_PXGRAPH 11 /* [:graph:] - characters that mark the paper */ -#define PT_PXPRINT 12 /* [:print:] - [:graph:] plus non-control spaces */ -#define PT_PXPUNCT 13 /* [:punct:] - punctuation characters */ - -/* Flag bits and data types for the extended class (OP_XCLASS) for classes that -contain characters with values greater than 255. */ - -#define XCL_NOT 0x01 /* Flag: this is a negative class */ -#define XCL_MAP 0x02 /* Flag: a 32-byte map is present */ - -#define XCL_END 0 /* Marks end of individual items */ -#define XCL_SINGLE 1 /* Single item (one multibyte char) follows */ -#define XCL_RANGE 2 /* A range (two multibyte chars) follows */ -#define XCL_PROP 3 /* Unicode property (2-byte property code follows) */ -#define XCL_NOTPROP 4 /* Unicode inverted property (ditto) */ - -/* These are escaped items that aren't just an encoding of a particular data -value such as \n. They must have non-zero values, as check_escape() returns 0 -for a data character. Also, they must appear in the same order as in the -opcode definitions below, up to ESC_z. There's a dummy for OP_ALLANY because it -corresponds to "." in DOTALL mode rather than an escape sequence. It is also -used for [^] in JavaScript compatibility mode, and for \C in non-utf mode. In -non-DOTALL mode, "." behaves like \N. - -The special values ESC_DU, ESC_du, etc. are used instead of ESC_D, ESC_d, etc. -when PCRE_UCP is set and replacement of \d etc by \p sequences is required. -They must be contiguous, and remain in order so that the replacements can be -looked up from a table. - -Negative numbers are used to encode a backreference (\1, \2, \3, etc.) in -check_escape(). There are two tests in the code for an escape -greater than ESC_b and less than ESC_Z to detect the types that may be -repeated. These are the types that consume characters. If any new escapes are -put in between that don't consume a character, that code will have to change. -*/ - -enum { ESC_A = 1, ESC_G, ESC_K, ESC_B, ESC_b, ESC_D, ESC_d, ESC_S, ESC_s, - ESC_W, ESC_w, ESC_N, ESC_dum, ESC_C, ESC_P, ESC_p, ESC_R, ESC_H, - ESC_h, ESC_V, ESC_v, ESC_X, ESC_Z, ESC_z, - ESC_E, ESC_Q, ESC_g, ESC_k, - ESC_DU, ESC_du, ESC_SU, ESC_su, ESC_WU, ESC_wu }; - - -/********************** Opcode definitions ******************/ - -/****** NOTE NOTE NOTE ****** - -Starting from 1 (i.e. after OP_END), the values up to OP_EOD must correspond in -order to the list of escapes immediately above. Furthermore, values up to -OP_DOLLM must not be changed without adjusting the table called autoposstab in -pcre_compile.c - -Whenever this list is updated, the two macro definitions that follow must be -updated to match. The possessification table called "opcode_possessify" in -pcre_compile.c must also be updated, and also the tables called "coptable" -and "poptable" in pcre_dfa_exec.c. - -****** NOTE NOTE NOTE ******/ - - -/* The values between FIRST_AUTOTAB_OP and LAST_AUTOTAB_RIGHT_OP, inclusive, -are used in a table for deciding whether a repeated character type can be -auto-possessified. */ - -#define FIRST_AUTOTAB_OP OP_NOT_DIGIT -#define LAST_AUTOTAB_LEFT_OP OP_EXTUNI -#define LAST_AUTOTAB_RIGHT_OP OP_DOLLM - -enum { - OP_END, /* 0 End of pattern */ - - /* Values corresponding to backslashed metacharacters */ - - OP_SOD, /* 1 Start of data: \A */ - OP_SOM, /* 2 Start of match (subject + offset): \G */ - OP_SET_SOM, /* 3 Set start of match (\K) */ - OP_NOT_WORD_BOUNDARY, /* 4 \B */ - OP_WORD_BOUNDARY, /* 5 \b */ - OP_NOT_DIGIT, /* 6 \D */ - OP_DIGIT, /* 7 \d */ - OP_NOT_WHITESPACE, /* 8 \S */ - OP_WHITESPACE, /* 9 \s */ - OP_NOT_WORDCHAR, /* 10 \W */ - OP_WORDCHAR, /* 11 \w */ - - OP_ANY, /* 12 Match any character except newline (\N) */ - OP_ALLANY, /* 13 Match any character */ - OP_ANYBYTE, /* 14 Match any byte (\C); different to OP_ANY for UTF-8 */ - OP_NOTPROP, /* 15 \P (not Unicode property) */ - OP_PROP, /* 16 \p (Unicode property) */ - OP_ANYNL, /* 17 \R (any newline sequence) */ - OP_NOT_HSPACE, /* 18 \H (not horizontal whitespace) */ - OP_HSPACE, /* 19 \h (horizontal whitespace) */ - OP_NOT_VSPACE, /* 20 \V (not vertical whitespace) */ - OP_VSPACE, /* 21 \v (vertical whitespace) */ - OP_EXTUNI, /* 22 \X (extended Unicode sequence */ - OP_EODN, /* 23 End of data or \n at end of data (\Z) */ - OP_EOD, /* 24 End of data (\z) */ - - /* Line end assertions */ - - OP_DOLL, /* 25 End of line - not multiline */ - OP_DOLLM, /* 26 End of line - multiline */ - OP_CIRC, /* 27 Start of line - not multiline */ - OP_CIRCM, /* 28 Start of line - multiline */ - - /* Single characters; caseful must precede the caseless ones */ - - OP_CHAR, /* 29 Match one character, casefully */ - OP_CHARI, /* 30 Match one character, caselessly */ - OP_NOT, /* 31 Match one character, not the given one, casefully */ - OP_NOTI, /* 32 Match one character, not the given one, caselessly */ - - /* The following sets of 13 opcodes must always be kept in step because - the offset from the first one is used to generate the others. */ - - /* Repeated characters; caseful must precede the caseless ones */ - - OP_STAR, /* 33 The maximizing and minimizing versions of */ - OP_MINSTAR, /* 34 these six opcodes must come in pairs, with */ - OP_PLUS, /* 35 the minimizing one second. */ - OP_MINPLUS, /* 36 */ - OP_QUERY, /* 37 */ - OP_MINQUERY, /* 38 */ - - OP_UPTO, /* 39 From 0 to n matches of one character, caseful*/ - OP_MINUPTO, /* 40 */ - OP_EXACT, /* 41 Exactly n matches */ - - OP_POSSTAR, /* 42 Possessified star, caseful */ - OP_POSPLUS, /* 43 Possessified plus, caseful */ - OP_POSQUERY, /* 44 Posesssified query, caseful */ - OP_POSUPTO, /* 45 Possessified upto, caseful */ - - /* Repeated characters; caseless must follow the caseful ones */ - - OP_STARI, /* 46 */ - OP_MINSTARI, /* 47 */ - OP_PLUSI, /* 48 */ - OP_MINPLUSI, /* 49 */ - OP_QUERYI, /* 50 */ - OP_MINQUERYI, /* 51 */ - - OP_UPTOI, /* 52 From 0 to n matches of one character, caseless */ - OP_MINUPTOI, /* 53 */ - OP_EXACTI, /* 54 */ - - OP_POSSTARI, /* 55 Possessified star, caseless */ - OP_POSPLUSI, /* 56 Possessified plus, caseless */ - OP_POSQUERYI, /* 57 Posesssified query, caseless */ - OP_POSUPTOI, /* 58 Possessified upto, caseless */ - - /* The negated ones must follow the non-negated ones, and match them */ - /* Negated repeated character, caseful; must precede the caseless ones */ - - OP_NOTSTAR, /* 59 The maximizing and minimizing versions of */ - OP_NOTMINSTAR, /* 60 these six opcodes must come in pairs, with */ - OP_NOTPLUS, /* 61 the minimizing one second. They must be in */ - OP_NOTMINPLUS, /* 62 exactly the same order as those above. */ - OP_NOTQUERY, /* 63 */ - OP_NOTMINQUERY, /* 64 */ - - OP_NOTUPTO, /* 65 From 0 to n matches, caseful */ - OP_NOTMINUPTO, /* 66 */ - OP_NOTEXACT, /* 67 Exactly n matches */ - - OP_NOTPOSSTAR, /* 68 Possessified versions, caseful */ - OP_NOTPOSPLUS, /* 69 */ - OP_NOTPOSQUERY, /* 70 */ - OP_NOTPOSUPTO, /* 71 */ - - /* Negated repeated character, caseless; must follow the caseful ones */ - - OP_NOTSTARI, /* 72 */ - OP_NOTMINSTARI, /* 73 */ - OP_NOTPLUSI, /* 74 */ - OP_NOTMINPLUSI, /* 75 */ - OP_NOTQUERYI, /* 76 */ - OP_NOTMINQUERYI, /* 77 */ - - OP_NOTUPTOI, /* 78 From 0 to n matches, caseless */ - OP_NOTMINUPTOI, /* 79 */ - OP_NOTEXACTI, /* 80 Exactly n matches */ - - OP_NOTPOSSTARI, /* 81 Possessified versions, caseless */ - OP_NOTPOSPLUSI, /* 82 */ - OP_NOTPOSQUERYI, /* 83 */ - OP_NOTPOSUPTOI, /* 84 */ - - /* Character types */ - - OP_TYPESTAR, /* 85 The maximizing and minimizing versions of */ - OP_TYPEMINSTAR, /* 86 these six opcodes must come in pairs, with */ - OP_TYPEPLUS, /* 87 the minimizing one second. These codes must */ - OP_TYPEMINPLUS, /* 88 be in exactly the same order as those above. */ - OP_TYPEQUERY, /* 89 */ - OP_TYPEMINQUERY, /* 90 */ - - OP_TYPEUPTO, /* 91 From 0 to n matches */ - OP_TYPEMINUPTO, /* 92 */ - OP_TYPEEXACT, /* 93 Exactly n matches */ - - OP_TYPEPOSSTAR, /* 94 Possessified versions */ - OP_TYPEPOSPLUS, /* 95 */ - OP_TYPEPOSQUERY, /* 96 */ - OP_TYPEPOSUPTO, /* 97 */ - - /* These are used for character classes and back references; only the - first six are the same as the sets above. */ - - OP_CRSTAR, /* 98 The maximizing and minimizing versions of */ - OP_CRMINSTAR, /* 99 all these opcodes must come in pairs, with */ - OP_CRPLUS, /* 100 the minimizing one second. These codes must */ - OP_CRMINPLUS, /* 101 be in exactly the same order as those above. */ - OP_CRQUERY, /* 102 */ - OP_CRMINQUERY, /* 103 */ - - OP_CRRANGE, /* 104 These are different to the three sets above. */ - OP_CRMINRANGE, /* 105 */ - - OP_CRPOSSTAR, /* 106 Possessified versions */ - OP_CRPOSPLUS, /* 107 */ - OP_CRPOSQUERY, /* 108 */ - OP_CRPOSRANGE, /* 109 */ - - /* End of quantifier opcodes */ - - OP_CLASS, /* 110 Match a character class, chars < 256 only */ - OP_NCLASS, /* 111 Same, but the bitmap was created from a negative - class - the difference is relevant only when a - character > 255 is encountered. */ - OP_XCLASS, /* 112 Extended class for handling > 255 chars within the - class. This does both positive and negative. */ - OP_REF, /* 113 Match a back reference, casefully */ - OP_REFI, /* 114 Match a back reference, caselessly */ - OP_DNREF, /* 115 Match a duplicate name backref, casefully */ - OP_DNREFI, /* 116 Match a duplicate name backref, caselessly */ - OP_RECURSE, /* 117 Match a numbered subpattern (possibly recursive) */ - OP_CALLOUT, /* 118 Call out to external function if provided */ - - OP_ALT, /* 119 Start of alternation */ - OP_KET, /* 120 End of group that doesn't have an unbounded repeat */ - OP_KETRMAX, /* 121 These two must remain together and in this */ - OP_KETRMIN, /* 122 order. They are for groups the repeat for ever. */ - OP_KETRPOS, /* 123 Possessive unlimited repeat. */ - - /* The assertions must come before BRA, CBRA, ONCE, and COND, and the four - asserts must remain in order. */ - - OP_REVERSE, /* 124 Move pointer back - used in lookbehind assertions */ - OP_ASSERT, /* 125 Positive lookahead */ - OP_ASSERT_NOT, /* 126 Negative lookahead */ - OP_ASSERTBACK, /* 127 Positive lookbehind */ - OP_ASSERTBACK_NOT, /* 128 Negative lookbehind */ - - /* ONCE, ONCE_NC, BRA, BRAPOS, CBRA, CBRAPOS, and COND must come immediately - after the assertions, with ONCE first, as there's a test for >= ONCE for a - subpattern that isn't an assertion. The POS versions must immediately follow - the non-POS versions in each case. */ - - OP_ONCE, /* 129 Atomic group, contains captures */ - OP_ONCE_NC, /* 130 Atomic group containing no captures */ - OP_BRA, /* 131 Start of non-capturing bracket */ - OP_BRAPOS, /* 132 Ditto, with unlimited, possessive repeat */ - OP_CBRA, /* 133 Start of capturing bracket */ - OP_CBRAPOS, /* 134 Ditto, with unlimited, possessive repeat */ - OP_COND, /* 135 Conditional group */ - - /* These five must follow the previous five, in the same order. There's a - check for >= SBRA to distinguish the two sets. */ - - OP_SBRA, /* 136 Start of non-capturing bracket, check empty */ - OP_SBRAPOS, /* 137 Ditto, with unlimited, possessive repeat */ - OP_SCBRA, /* 138 Start of capturing bracket, check empty */ - OP_SCBRAPOS, /* 139 Ditto, with unlimited, possessive repeat */ - OP_SCOND, /* 140 Conditional group, check empty */ - - /* The next two pairs must (respectively) be kept together. */ - - OP_CREF, /* 141 Used to hold a capture number as condition */ - OP_DNCREF, /* 142 Used to point to duplicate names as a condition */ - OP_RREF, /* 143 Used to hold a recursion number as condition */ - OP_DNRREF, /* 144 Used to point to duplicate names as a condition */ - OP_DEF, /* 145 The DEFINE condition */ - - OP_BRAZERO, /* 146 These two must remain together and in this */ - OP_BRAMINZERO, /* 147 order. */ - OP_BRAPOSZERO, /* 148 */ - - /* These are backtracking control verbs */ - - OP_MARK, /* 149 always has an argument */ - OP_PRUNE, /* 150 */ - OP_PRUNE_ARG, /* 151 same, but with argument */ - OP_SKIP, /* 152 */ - OP_SKIP_ARG, /* 153 same, but with argument */ - OP_THEN, /* 154 */ - OP_THEN_ARG, /* 155 same, but with argument */ - OP_COMMIT, /* 156 */ - - /* These are forced failure and success verbs */ - - OP_FAIL, /* 157 */ - OP_ACCEPT, /* 158 */ - OP_ASSERT_ACCEPT, /* 159 Used inside assertions */ - OP_CLOSE, /* 160 Used before OP_ACCEPT to close open captures */ - - /* This is used to skip a subpattern with a {0} quantifier */ - - OP_SKIPZERO, /* 161 */ - - /* This is not an opcode, but is used to check that tables indexed by opcode - are the correct length, in order to catch updating errors - there have been - some in the past. */ - - OP_TABLE_LENGTH -}; - -/* *** NOTE NOTE NOTE *** Whenever the list above is updated, the two macro -definitions that follow must also be updated to match. There are also tables -called "opcode_possessify" in pcre_compile.c and "coptable" and "poptable" in -pcre_dfa_exec.c that must be updated. */ - - -/* This macro defines textual names for all the opcodes. These are used only -for debugging, and some of them are only partial names. The macro is referenced -only in pcre_printint.c, which fills out the full names in many cases (and in -some cases doesn't actually use these names at all). */ - -#define OP_NAME_LIST \ - "End", "\\A", "\\G", "\\K", "\\B", "\\b", "\\D", "\\d", \ - "\\S", "\\s", "\\W", "\\w", "Any", "AllAny", "Anybyte", \ - "notprop", "prop", "\\R", "\\H", "\\h", "\\V", "\\v", \ - "extuni", "\\Z", "\\z", \ - "$", "$", "^", "^", "char", "chari", "not", "noti", \ - "*", "*?", "+", "+?", "?", "??", \ - "{", "{", "{", \ - "*+","++", "?+", "{", \ - "*", "*?", "+", "+?", "?", "??", \ - "{", "{", "{", \ - "*+","++", "?+", "{", \ - "*", "*?", "+", "+?", "?", "??", \ - "{", "{", "{", \ - "*+","++", "?+", "{", \ - "*", "*?", "+", "+?", "?", "??", \ - "{", "{", "{", \ - "*+","++", "?+", "{", \ - "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \ - "*+","++", "?+", "{", \ - "*", "*?", "+", "+?", "?", "??", "{", "{", \ - "*+","++", "?+", "{", \ - "class", "nclass", "xclass", "Ref", "Refi", "DnRef", "DnRefi", \ - "Recurse", "Callout", \ - "Alt", "Ket", "KetRmax", "KetRmin", "KetRpos", \ - "Reverse", "Assert", "Assert not", "AssertB", "AssertB not", \ - "Once", "Once_NC", \ - "Bra", "BraPos", "CBra", "CBraPos", \ - "Cond", \ - "SBra", "SBraPos", "SCBra", "SCBraPos", \ - "SCond", \ - "Cond ref", "Cond dnref", "Cond rec", "Cond dnrec", "Cond def", \ - "Brazero", "Braminzero", "Braposzero", \ - "*MARK", "*PRUNE", "*PRUNE", "*SKIP", "*SKIP", \ - "*THEN", "*THEN", "*COMMIT", "*FAIL", \ - "*ACCEPT", "*ASSERT_ACCEPT", \ - "Close", "Skip zero" - - -/* This macro defines the length of fixed length operations in the compiled -regex. The lengths are used when searching for specific things, and also in the -debugging printing of a compiled regex. We use a macro so that it can be -defined close to the definitions of the opcodes themselves. - -As things have been extended, some of these are no longer fixed lenths, but are -minima instead. For example, the length of a single-character repeat may vary -in UTF-8 mode. The code that uses this table must know about such things. */ - -#define OP_LENGTHS \ - 1, /* End */ \ - 1, 1, 1, 1, 1, /* \A, \G, \K, \B, \b */ \ - 1, 1, 1, 1, 1, 1, /* \D, \d, \S, \s, \W, \w */ \ - 1, 1, 1, /* Any, AllAny, Anybyte */ \ - 3, 3, /* \P, \p */ \ - 1, 1, 1, 1, 1, /* \R, \H, \h, \V, \v */ \ - 1, /* \X */ \ - 1, 1, 1, 1, 1, 1, /* \Z, \z, $, $M ^, ^M */ \ - 2, /* Char - the minimum length */ \ - 2, /* Chari - the minimum length */ \ - 2, /* not */ \ - 2, /* noti */ \ - /* Positive single-char repeats ** These are */ \ - 2, 2, 2, 2, 2, 2, /* *, *?, +, +?, ?, ?? ** minima in */ \ - 2+IMM2_SIZE, 2+IMM2_SIZE, /* upto, minupto ** mode */ \ - 2+IMM2_SIZE, /* exact */ \ - 2, 2, 2, 2+IMM2_SIZE, /* *+, ++, ?+, upto+ */ \ - 2, 2, 2, 2, 2, 2, /* *I, *?I, +I, +?I, ?I, ??I ** UTF-8 */ \ - 2+IMM2_SIZE, 2+IMM2_SIZE, /* upto I, minupto I */ \ - 2+IMM2_SIZE, /* exact I */ \ - 2, 2, 2, 2+IMM2_SIZE, /* *+I, ++I, ?+I, upto+I */ \ - /* Negative single-char repeats - only for chars < 256 */ \ - 2, 2, 2, 2, 2, 2, /* NOT *, *?, +, +?, ?, ?? */ \ - 2+IMM2_SIZE, 2+IMM2_SIZE, /* NOT upto, minupto */ \ - 2+IMM2_SIZE, /* NOT exact */ \ - 2, 2, 2, 2+IMM2_SIZE, /* Possessive NOT *, +, ?, upto */ \ - 2, 2, 2, 2, 2, 2, /* NOT *I, *?I, +I, +?I, ?I, ??I */ \ - 2+IMM2_SIZE, 2+IMM2_SIZE, /* NOT upto I, minupto I */ \ - 2+IMM2_SIZE, /* NOT exact I */ \ - 2, 2, 2, 2+IMM2_SIZE, /* Possessive NOT *I, +I, ?I, upto I */ \ - /* Positive type repeats */ \ - 2, 2, 2, 2, 2, 2, /* Type *, *?, +, +?, ?, ?? */ \ - 2+IMM2_SIZE, 2+IMM2_SIZE, /* Type upto, minupto */ \ - 2+IMM2_SIZE, /* Type exact */ \ - 2, 2, 2, 2+IMM2_SIZE, /* Possessive *+, ++, ?+, upto+ */ \ - /* Character class & ref repeats */ \ - 1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */ \ - 1+2*IMM2_SIZE, 1+2*IMM2_SIZE, /* CRRANGE, CRMINRANGE */ \ - 1, 1, 1, 1+2*IMM2_SIZE, /* Possessive *+, ++, ?+, CRPOSRANGE */ \ - 1+(32/sizeof(pcre_uchar)), /* CLASS */ \ - 1+(32/sizeof(pcre_uchar)), /* NCLASS */ \ - 0, /* XCLASS - variable length */ \ - 1+IMM2_SIZE, /* REF */ \ - 1+IMM2_SIZE, /* REFI */ \ - 1+2*IMM2_SIZE, /* DNREF */ \ - 1+2*IMM2_SIZE, /* DNREFI */ \ - 1+LINK_SIZE, /* RECURSE */ \ - 2+2*LINK_SIZE, /* CALLOUT */ \ - 1+LINK_SIZE, /* Alt */ \ - 1+LINK_SIZE, /* Ket */ \ - 1+LINK_SIZE, /* KetRmax */ \ - 1+LINK_SIZE, /* KetRmin */ \ - 1+LINK_SIZE, /* KetRpos */ \ - 1+LINK_SIZE, /* Reverse */ \ - 1+LINK_SIZE, /* Assert */ \ - 1+LINK_SIZE, /* Assert not */ \ - 1+LINK_SIZE, /* Assert behind */ \ - 1+LINK_SIZE, /* Assert behind not */ \ - 1+LINK_SIZE, /* ONCE */ \ - 1+LINK_SIZE, /* ONCE_NC */ \ - 1+LINK_SIZE, /* BRA */ \ - 1+LINK_SIZE, /* BRAPOS */ \ - 1+LINK_SIZE+IMM2_SIZE, /* CBRA */ \ - 1+LINK_SIZE+IMM2_SIZE, /* CBRAPOS */ \ - 1+LINK_SIZE, /* COND */ \ - 1+LINK_SIZE, /* SBRA */ \ - 1+LINK_SIZE, /* SBRAPOS */ \ - 1+LINK_SIZE+IMM2_SIZE, /* SCBRA */ \ - 1+LINK_SIZE+IMM2_SIZE, /* SCBRAPOS */ \ - 1+LINK_SIZE, /* SCOND */ \ - 1+IMM2_SIZE, 1+2*IMM2_SIZE, /* CREF, DNCREF */ \ - 1+IMM2_SIZE, 1+2*IMM2_SIZE, /* RREF, DNRREF */ \ - 1, /* DEF */ \ - 1, 1, 1, /* BRAZERO, BRAMINZERO, BRAPOSZERO */ \ - 3, 1, 3, /* MARK, PRUNE, PRUNE_ARG */ \ - 1, 3, /* SKIP, SKIP_ARG */ \ - 1, 3, /* THEN, THEN_ARG */ \ - 1, 1, 1, 1, /* COMMIT, FAIL, ACCEPT, ASSERT_ACCEPT */ \ - 1+IMM2_SIZE, 1 /* CLOSE, SKIPZERO */ - -/* A magic value for OP_RREF to indicate the "any recursion" condition. */ - -#define RREF_ANY 0xffff - -/* Compile time error code numbers. They are given names so that they can more -easily be tracked. When a new number is added, the table called eint in -pcreposix.c must be updated. */ - -enum { ERR0, ERR1, ERR2, ERR3, ERR4, ERR5, ERR6, ERR7, ERR8, ERR9, - ERR10, ERR11, ERR12, ERR13, ERR14, ERR15, ERR16, ERR17, ERR18, ERR19, - ERR20, ERR21, ERR22, ERR23, ERR24, ERR25, ERR26, ERR27, ERR28, ERR29, - ERR30, ERR31, ERR32, ERR33, ERR34, ERR35, ERR36, ERR37, ERR38, ERR39, - ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49, - ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59, - ERR60, ERR61, ERR62, ERR63, ERR64, ERR65, ERR66, ERR67, ERR68, ERR69, - ERR70, ERR71, ERR72, ERR73, ERR74, ERR75, ERR76, ERR77, ERR78, ERR79, - ERR80, ERR81, ERR82, ERR83, ERR84, ERRCOUNT }; - -/* JIT compiling modes. The function list is indexed by them. */ - -enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COMPILE, JIT_PARTIAL_HARD_COMPILE, - JIT_NUMBER_OF_COMPILE_MODES }; - -/* The real format of the start of the pcre block; the index of names and the -code vector run on as long as necessary after the end. We store an explicit -offset to the name table so that if a regex is compiled on one host, saved, and -then run on another where the size of pointers is different, all might still -be well. - -The size of the structure must be a multiple of 8 bytes. For the case of -compiled-on-4 and run-on-8, we include an extra pointer that is always NULL so -that there are an even number of pointers which therefore are a multiple of 8 -bytes. - -It is necessary to fork the struct for the 32 bit library, since it needs to -use pcre_uint32 for first_char and req_char. We can't put an ifdef inside the -typedef because pcretest needs access to the struct of the 8-, 16- and 32-bit -variants. - -*** WARNING *** -When new fields are added to these structures, remember to adjust the code in -pcre_byte_order.c that is concerned with swapping the byte order of the fields -when a compiled regex is reloaded on a host with different endianness. -*** WARNING *** -There is also similar byte-flipping code in pcretest.c, which is used for -testing the byte-flipping features. It must also be kept in step. -*** WARNING *** -*/ - -typedef struct real_pcre8_or_16 { - pcre_uint32 magic_number; - pcre_uint32 size; /* Total that was malloced */ - pcre_uint32 options; /* Public options */ - pcre_uint32 flags; /* Private flags */ - pcre_uint32 limit_match; /* Limit set from regex */ - pcre_uint32 limit_recursion; /* Limit set from regex */ - pcre_uint16 first_char; /* Starting character */ - pcre_uint16 req_char; /* This character must be seen */ - pcre_uint16 max_lookbehind; /* Longest lookbehind (characters) */ - pcre_uint16 top_bracket; /* Highest numbered group */ - pcre_uint16 top_backref; /* Highest numbered back reference */ - pcre_uint16 name_table_offset; /* Offset to name table that follows */ - pcre_uint16 name_entry_size; /* Size of any name items */ - pcre_uint16 name_count; /* Number of name items */ - pcre_uint16 ref_count; /* Reference count */ - pcre_uint16 dummy1; /* To ensure size is a multiple of 8 */ - pcre_uint16 dummy2; /* To ensure size is a multiple of 8 */ - pcre_uint16 dummy3; /* To ensure size is a multiple of 8 */ - const pcre_uint8 *tables; /* Pointer to tables or NULL for std */ - void *nullpad; /* NULL padding */ -} real_pcre8_or_16; - -typedef struct real_pcre8_or_16 real_pcre; -typedef struct real_pcre8_or_16 real_pcre16; - -typedef struct real_pcre32 { - pcre_uint32 magic_number; - pcre_uint32 size; /* Total that was malloced */ - pcre_uint32 options; /* Public options */ - pcre_uint32 flags; /* Private flags */ - pcre_uint32 limit_match; /* Limit set from regex */ - pcre_uint32 limit_recursion; /* Limit set from regex */ - pcre_uint32 first_char; /* Starting character */ - pcre_uint32 req_char; /* This character must be seen */ - pcre_uint16 max_lookbehind; /* Longest lookbehind (characters) */ - pcre_uint16 top_bracket; /* Highest numbered group */ - pcre_uint16 top_backref; /* Highest numbered back reference */ - pcre_uint16 name_table_offset; /* Offset to name table that follows */ - pcre_uint16 name_entry_size; /* Size of any name items */ - pcre_uint16 name_count; /* Number of name items */ - pcre_uint16 ref_count; /* Reference count */ - pcre_uint16 dummy; /* To ensure size is a multiple of 8 */ - const pcre_uint8 *tables; /* Pointer to tables or NULL for std */ - void *nullpad; /* NULL padding */ -} real_pcre32; - -#if defined COMPILE_PCRE8 -#define REAL_PCRE real_pcre -#elif defined COMPILE_PCRE16 -#define REAL_PCRE real_pcre16 -#elif defined COMPILE_PCRE32 -#define REAL_PCRE real_pcre32 -#endif - -/* Assert that the size of REAL_PCRE is divisible by 8 */ -typedef int __assert_real_pcre_size_divisible_8[(sizeof(REAL_PCRE) % 8) == 0 ? 1 : -1]; - -/* Needed in pcretest to access some fields in the real_pcre* structures - * directly. They're unified for 8/16/32 bits since the structs only differ - * after these fields; if that ever changes, need to fork those defines into - * 8/16 and 32 bit versions. */ -#define REAL_PCRE_MAGIC(re) (((REAL_PCRE*)re)->magic_number) -#define REAL_PCRE_SIZE(re) (((REAL_PCRE*)re)->size) -#define REAL_PCRE_OPTIONS(re) (((REAL_PCRE*)re)->options) -#define REAL_PCRE_FLAGS(re) (((REAL_PCRE*)re)->flags) - -/* The format of the block used to store data from pcre_study(). The same -remark (see NOTE above) about extending this structure applies. */ - -typedef struct pcre_study_data { - pcre_uint32 size; /* Total that was malloced */ - pcre_uint32 flags; /* Private flags */ - pcre_uint8 start_bits[32]; /* Starting char bits */ - pcre_uint32 minlength; /* Minimum subject length */ -} pcre_study_data; - -/* Structure for building a chain of open capturing subpatterns during -compiling, so that instructions to close them can be compiled when (*ACCEPT) is -encountered. This is also used to identify subpatterns that contain recursive -back references to themselves, so that they can be made atomic. */ - -typedef struct open_capitem { - struct open_capitem *next; /* Chain link */ - pcre_uint16 number; /* Capture number */ - pcre_uint16 flag; /* Set TRUE if recursive back ref */ -} open_capitem; - -/* Structure for building a list of named groups during the first pass of -compiling. */ - -typedef struct named_group { - const pcre_uchar *name; /* Points to the name in the pattern */ - int length; /* Length of the name */ - pcre_uint32 number; /* Group number */ -} named_group; - -/* Structure for passing "static" information around between the functions -doing the compiling, so that they are thread-safe. */ - -typedef struct compile_data { - const pcre_uint8 *lcc; /* Points to lower casing table */ - const pcre_uint8 *fcc; /* Points to case-flipping table */ - const pcre_uint8 *cbits; /* Points to character type table */ - const pcre_uint8 *ctypes; /* Points to table of type maps */ - const pcre_uchar *start_workspace;/* The start of working space */ - const pcre_uchar *start_code; /* The start of the compiled code */ - const pcre_uchar *start_pattern; /* The start of the pattern */ - const pcre_uchar *end_pattern; /* The end of the pattern */ - pcre_uchar *hwm; /* High watermark of workspace */ - open_capitem *open_caps; /* Chain of open capture items */ - named_group *named_groups; /* Points to vector in pre-compile */ - pcre_uchar *name_table; /* The name/number table */ - int names_found; /* Number of entries so far */ - int name_entry_size; /* Size of each entry */ - int named_group_list_size; /* Number of entries in the list */ - int workspace_size; /* Size of workspace */ - unsigned int bracount; /* Count of capturing parens as we compile */ - int final_bracount; /* Saved value after first pass */ - int max_lookbehind; /* Maximum lookbehind (characters) */ - int top_backref; /* Maximum back reference */ - unsigned int backref_map; /* Bitmap of low back refs */ - unsigned int namedrefcount; /* Number of backreferences by name */ - int parens_depth; /* Depth of nested parentheses */ - int assert_depth; /* Depth of nested assertions */ - pcre_uint32 external_options; /* External (initial) options */ - pcre_uint32 external_flags; /* External flag bits to be set */ - int req_varyopt; /* "After variable item" flag for reqbyte */ - BOOL had_accept; /* (*ACCEPT) encountered */ - BOOL had_pruneorskip; /* (*PRUNE) or (*SKIP) encountered */ - BOOL check_lookbehind; /* Lookbehinds need later checking */ - BOOL dupnames; /* Duplicate names exist */ - int nltype; /* Newline type */ - int nllen; /* Newline string length */ - pcre_uchar nl[4]; /* Newline string when fixed length */ -} compile_data; - -/* Structure for maintaining a chain of pointers to the currently incomplete -branches, for testing for left recursion while compiling. */ - -typedef struct branch_chain { - struct branch_chain *outer; - pcre_uchar *current_branch; -} branch_chain; - -/* Structure for items in a linked list that represents an explicit recursive -call within the pattern; used by pcre_exec(). */ - -typedef struct recursion_info { - struct recursion_info *prevrec; /* Previous recursion record (or NULL) */ - unsigned int group_num; /* Number of group that was called */ - int *offset_save; /* Pointer to start of saved offsets */ - int saved_max; /* Number of saved offsets */ - int saved_capture_last; /* Last capture number */ - PCRE_PUCHAR subject_position; /* Position at start of recursion */ -} recursion_info; - -/* A similar structure for pcre_dfa_exec(). */ - -typedef struct dfa_recursion_info { - struct dfa_recursion_info *prevrec; - int group_num; - PCRE_PUCHAR subject_position; -} dfa_recursion_info; - -/* Structure for building a chain of data for holding the values of the subject -pointer at the start of each subpattern, so as to detect when an empty string -has been matched by a subpattern - to break infinite loops; used by -pcre_exec(). */ - -typedef struct eptrblock { - struct eptrblock *epb_prev; - PCRE_PUCHAR epb_saved_eptr; -} eptrblock; - - -/* Structure for passing "static" information around between the functions -doing traditional NFA matching, so that they are thread-safe. */ - -typedef struct match_data { - unsigned long int match_call_count; /* As it says */ - unsigned long int match_limit; /* As it says */ - unsigned long int match_limit_recursion; /* As it says */ - int *offset_vector; /* Offset vector */ - int offset_end; /* One past the end */ - int offset_max; /* The maximum usable for return data */ - int nltype; /* Newline type */ - int nllen; /* Newline string length */ - int name_count; /* Number of names in name table */ - int name_entry_size; /* Size of entry in names table */ - unsigned int skip_arg_count; /* For counting SKIP_ARGs */ - unsigned int ignore_skip_arg; /* For re-run when SKIP arg name not found */ - pcre_uchar *name_table; /* Table of names */ - pcre_uchar nl[4]; /* Newline string when fixed */ - const pcre_uint8 *lcc; /* Points to lower casing table */ - const pcre_uint8 *fcc; /* Points to case-flipping table */ - const pcre_uint8 *ctypes; /* Points to table of type maps */ - BOOL notbol; /* NOTBOL flag */ - BOOL noteol; /* NOTEOL flag */ - BOOL utf; /* UTF-8 / UTF-16 flag */ - BOOL jscript_compat; /* JAVASCRIPT_COMPAT flag */ - BOOL use_ucp; /* PCRE_UCP flag */ - BOOL endonly; /* Dollar not before final \n */ - BOOL notempty; /* Empty string match not wanted */ - BOOL notempty_atstart; /* Empty string match at start not wanted */ - BOOL hitend; /* Hit the end of the subject at some point */ - BOOL bsr_anycrlf; /* \R is just any CRLF, not full Unicode */ - BOOL hasthen; /* Pattern contains (*THEN) */ - const pcre_uchar *start_code; /* For use when recursing */ - PCRE_PUCHAR start_subject; /* Start of the subject string */ - PCRE_PUCHAR end_subject; /* End of the subject string */ - PCRE_PUCHAR start_match_ptr; /* Start of matched string */ - PCRE_PUCHAR end_match_ptr; /* Subject position at end match */ - PCRE_PUCHAR start_used_ptr; /* Earliest consulted character */ - int partial; /* PARTIAL options */ - int end_offset_top; /* Highwater mark at end of match */ - pcre_int32 capture_last; /* Most recent capture number + overflow flag */ - int start_offset; /* The start offset value */ - int match_function_type; /* Set for certain special calls of MATCH() */ - eptrblock *eptrchain; /* Chain of eptrblocks for tail recursions */ - int eptrn; /* Next free eptrblock */ - recursion_info *recursive; /* Linked list of recursion data */ - void *callout_data; /* To pass back to callouts */ - const pcre_uchar *mark; /* Mark pointer to pass back on success */ - const pcre_uchar *nomatch_mark;/* Mark pointer to pass back on failure */ - const pcre_uchar *once_target; /* Where to back up to for atomic groups */ -#ifdef NO_RECURSE - void *match_frames_base; /* For remembering malloc'd frames */ -#endif -} match_data; - -/* A similar structure is used for the same purpose by the DFA matching -functions. */ - -typedef struct dfa_match_data { - const pcre_uchar *start_code; /* Start of the compiled pattern */ - const pcre_uchar *start_subject ; /* Start of the subject string */ - const pcre_uchar *end_subject; /* End of subject string */ - const pcre_uchar *start_used_ptr; /* Earliest consulted character */ - const pcre_uint8 *tables; /* Character tables */ - int start_offset; /* The start offset value */ - int moptions; /* Match options */ - int poptions; /* Pattern options */ - int nltype; /* Newline type */ - int nllen; /* Newline string length */ - pcre_uchar nl[4]; /* Newline string when fixed */ - void *callout_data; /* To pass back to callouts */ - dfa_recursion_info *recursive; /* Linked list of recursion data */ -} dfa_match_data; - -/* Bit definitions for entries in the pcre_ctypes table. */ - -#define ctype_space 0x01 -#define ctype_letter 0x02 -#define ctype_digit 0x04 -#define ctype_xdigit 0x08 -#define ctype_word 0x10 /* alphanumeric or '_' */ -#define ctype_meta 0x80 /* regexp meta char or zero (end pattern) */ - -/* Offsets for the bitmap tables in pcre_cbits. Each table contains a set -of bits for a class map. Some classes are built by combining these tables. */ - -#define cbit_space 0 /* [:space:] or \s */ -#define cbit_xdigit 32 /* [:xdigit:] */ -#define cbit_digit 64 /* [:digit:] or \d */ -#define cbit_upper 96 /* [:upper:] */ -#define cbit_lower 128 /* [:lower:] */ -#define cbit_word 160 /* [:word:] or \w */ -#define cbit_graph 192 /* [:graph:] */ -#define cbit_print 224 /* [:print:] */ -#define cbit_punct 256 /* [:punct:] */ -#define cbit_cntrl 288 /* [:cntrl:] */ -#define cbit_length 320 /* Length of the cbits table */ - -/* Offsets of the various tables from the base tables pointer, and -total length. */ - -#define lcc_offset 0 -#define fcc_offset 256 -#define cbits_offset 512 -#define ctypes_offset (cbits_offset + cbit_length) -#define tables_length (ctypes_offset + 256) - -/* Internal function and data prefixes. */ - -#if defined COMPILE_PCRE8 -#ifndef PUBL -#define PUBL(name) pcre_##name -#endif -#ifndef PRIV -#define PRIV(name) _pcre_##name -#endif -#elif defined COMPILE_PCRE16 -#ifndef PUBL -#define PUBL(name) pcre16_##name -#endif -#ifndef PRIV -#define PRIV(name) _pcre16_##name -#endif -#elif defined COMPILE_PCRE32 -#ifndef PUBL -#define PUBL(name) pcre32_##name -#endif -#ifndef PRIV -#define PRIV(name) _pcre32_##name -#endif -#else -#error Unsupported compiling mode -#endif /* COMPILE_PCRE[8|16|32] */ - -/* Layout of the UCP type table that translates property names into types and -codes. Each entry used to point directly to a name, but to reduce the number of -relocations in shared libraries, it now has an offset into a single string -instead. */ - -typedef struct { - pcre_uint16 name_offset; - pcre_uint16 type; - pcre_uint16 value; -} ucp_type_table; - - -/* Internal shared data tables. These are tables that are used by more than one -of the exported public functions. They have to be "external" in the C sense, -but are not part of the PCRE public API. The data for these tables is in the -pcre_tables.c module. */ - -#ifdef COMPILE_PCRE8 -extern const int PRIV(utf8_table1)[]; -extern const int PRIV(utf8_table1_size); -extern const int PRIV(utf8_table2)[]; -extern const int PRIV(utf8_table3)[]; -extern const pcre_uint8 PRIV(utf8_table4)[]; -#endif /* COMPILE_PCRE8 */ - -extern const char PRIV(utt_names)[]; -extern const ucp_type_table PRIV(utt)[]; -extern const int PRIV(utt_size); - -extern const pcre_uint8 PRIV(OP_lengths)[]; -extern const pcre_uint8 PRIV(default_tables)[]; - -extern const pcre_uint32 PRIV(hspace_list)[]; -extern const pcre_uint32 PRIV(vspace_list)[]; - - -/* Internal shared functions. These are functions that are used by more than -one of the exported public functions. They have to be "external" in the C -sense, but are not part of the PCRE public API. */ - -/* String comparison functions. */ -#if defined COMPILE_PCRE8 - -#define STRCMP_UC_UC(str1, str2) \ - strcmp((char *)(str1), (char *)(str2)) -#define STRCMP_UC_C8(str1, str2) \ - strcmp((char *)(str1), (str2)) -#define STRNCMP_UC_UC(str1, str2, num) \ - strncmp((char *)(str1), (char *)(str2), (num)) -#define STRNCMP_UC_C8(str1, str2, num) \ - strncmp((char *)(str1), (str2), (num)) -#define STRLEN_UC(str) strlen((const char *)str) - -#elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - -extern int PRIV(strcmp_uc_uc)(const pcre_uchar *, - const pcre_uchar *); -extern int PRIV(strcmp_uc_c8)(const pcre_uchar *, - const char *); -extern int PRIV(strncmp_uc_uc)(const pcre_uchar *, - const pcre_uchar *, unsigned int num); -extern int PRIV(strncmp_uc_c8)(const pcre_uchar *, - const char *, unsigned int num); -extern unsigned int PRIV(strlen_uc)(const pcre_uchar *str); - -#define STRCMP_UC_UC(str1, str2) \ - PRIV(strcmp_uc_uc)((str1), (str2)) -#define STRCMP_UC_C8(str1, str2) \ - PRIV(strcmp_uc_c8)((str1), (str2)) -#define STRNCMP_UC_UC(str1, str2, num) \ - PRIV(strncmp_uc_uc)((str1), (str2), (num)) -#define STRNCMP_UC_C8(str1, str2, num) \ - PRIV(strncmp_uc_c8)((str1), (str2), (num)) -#define STRLEN_UC(str) PRIV(strlen_uc)(str) - -#endif /* COMPILE_PCRE[8|16|32] */ - -#if defined COMPILE_PCRE8 || defined COMPILE_PCRE16 - -#define STRCMP_UC_UC_TEST(str1, str2) STRCMP_UC_UC(str1, str2) -#define STRCMP_UC_C8_TEST(str1, str2) STRCMP_UC_C8(str1, str2) - -#elif defined COMPILE_PCRE32 - -extern int PRIV(strcmp_uc_uc_utf)(const pcre_uchar *, - const pcre_uchar *); -extern int PRIV(strcmp_uc_c8_utf)(const pcre_uchar *, - const char *); - -#define STRCMP_UC_UC_TEST(str1, str2) \ - (utf ? PRIV(strcmp_uc_uc_utf)((str1), (str2)) : PRIV(strcmp_uc_uc)((str1), (str2))) -#define STRCMP_UC_C8_TEST(str1, str2) \ - (utf ? PRIV(strcmp_uc_c8_utf)((str1), (str2)) : PRIV(strcmp_uc_c8)((str1), (str2))) - -#endif /* COMPILE_PCRE[8|16|32] */ - -extern const pcre_uchar *PRIV(find_bracket)(const pcre_uchar *, BOOL, int); -extern BOOL PRIV(is_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR, - int *, BOOL); -extern unsigned int PRIV(ord2utf)(pcre_uint32, pcre_uchar *); -extern int PRIV(valid_utf)(PCRE_PUCHAR, int, int *); -extern BOOL PRIV(was_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR, - int *, BOOL); -extern BOOL PRIV(xclass)(pcre_uint32, const pcre_uchar *, BOOL); - -#ifdef SUPPORT_JIT -extern void PRIV(jit_compile)(const REAL_PCRE *, - PUBL(extra) *, int); -extern int PRIV(jit_exec)(const PUBL(extra) *, - const pcre_uchar *, int, int, int, int *, int); -extern void PRIV(jit_free)(void *); -extern int PRIV(jit_get_size)(void *); -extern const char* PRIV(jit_get_target)(void); -#endif - -/* Unicode character database (UCD) */ - -typedef struct { - pcre_uint8 script; /* ucp_Arabic, etc. */ - pcre_uint8 chartype; /* ucp_Cc, etc. (general categories) */ - pcre_uint8 gbprop; /* ucp_gbControl, etc. (grapheme break property) */ - pcre_uint8 caseset; /* offset to multichar other cases or zero */ - pcre_int32 other_case; /* offset to other case, or zero if none */ -} ucd_record; - -extern const pcre_uint32 PRIV(ucd_caseless_sets)[]; -extern const ucd_record PRIV(ucd_records)[]; -extern const pcre_uint8 PRIV(ucd_stage1)[]; -extern const pcre_uint16 PRIV(ucd_stage2)[]; -extern const pcre_uint32 PRIV(ucp_gentype)[]; -extern const pcre_uint32 PRIV(ucp_gbtable)[]; -#ifdef SUPPORT_JIT -extern const int PRIV(ucp_typerange)[]; -#endif - -#ifdef SUPPORT_UCP -/* UCD access macros */ - -#define UCD_BLOCK_SIZE 128 -#define GET_UCD(ch) (PRIV(ucd_records) + \ - PRIV(ucd_stage2)[PRIV(ucd_stage1)[(int)(ch) / UCD_BLOCK_SIZE] * \ - UCD_BLOCK_SIZE + (int)(ch) % UCD_BLOCK_SIZE]) - -#define UCD_CHARTYPE(ch) GET_UCD(ch)->chartype -#define UCD_SCRIPT(ch) GET_UCD(ch)->script -#define UCD_CATEGORY(ch) PRIV(ucp_gentype)[UCD_CHARTYPE(ch)] -#define UCD_GRAPHBREAK(ch) GET_UCD(ch)->gbprop -#define UCD_CASESET(ch) GET_UCD(ch)->caseset -#define UCD_OTHERCASE(ch) ((pcre_uint32)((int)ch + (int)(GET_UCD(ch)->other_case))) + Copyright (c) 1997-2012 University of Cambridge + +----------------------------------------------------------------------------- +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. + + * Neither the name of the University of Cambridge 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 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. +----------------------------------------------------------------------------- +*/ + +/* This header contains definitions that are shared between the different +modules, but which are not relevant to the exported API. This includes some +functions whose names all begin with "_pcre_" or "_pcre16_" depending on +the PRIV macro. */ + +#ifndef PCRE_INTERNAL_H +#define PCRE_INTERNAL_H + +/* Define PCRE_DEBUG to get debugging output on stdout. */ + +#if 0 +#define PCRE_DEBUG +#endif + +/* PCRE is compiled as an 8 bit library if it is not requested otherwise. */ +#ifndef COMPILE_PCRE16 +#define COMPILE_PCRE8 +#endif + +/* If SUPPORT_UCP is defined, SUPPORT_UTF must also be defined. The +"configure" script ensures this, but not everybody uses "configure". */ + +#if defined SUPPORT_UCP && !(defined SUPPORT_UTF) +#define SUPPORT_UTF 1 +#endif + +/* We define SUPPORT_UTF if SUPPORT_UTF8 is enabled for compatibility +reasons with existing code. */ + +#if defined SUPPORT_UTF8 && !(defined SUPPORT_UTF) +#define SUPPORT_UTF 1 +#endif + +/* Fixme: SUPPORT_UTF8 should be eventually disappear from the code. +Until then we define it if SUPPORT_UTF is defined. */ + +#if defined SUPPORT_UTF && !(defined SUPPORT_UTF8) +#define SUPPORT_UTF8 1 +#endif + +/* We do not support both EBCDIC and UTF-8/16 at the same time. The "configure" +script prevents both being selected, but not everybody uses "configure". */ + +#if defined EBCDIC && defined SUPPORT_UTF +#error The use of both EBCDIC and SUPPORT_UTF8/16 is not supported. +#endif + +/* Use a macro for debugging printing, 'cause that eliminates the use of #ifdef +inline, and there are *still* stupid compilers about that don't like indented +pre-processor statements, or at least there were when I first wrote this. After +all, it had only been about 10 years then... + +It turns out that the Mac Debugging.h header also defines the macro DPRINTF, so +be absolutely sure we get our version. */ + +#undef DPRINTF +#ifdef PCRE_DEBUG +#define DPRINTF(p) printf p +#else +#define DPRINTF(p) /* Nothing */ +#endif + + +/* Standard C headers plus the external interface definition. The only time +setjmp and stdarg are used is when NO_RECURSE is set. */ + +#include +#include +#include +#include +#include +#include + +/* When compiling a DLL for Windows, the exported symbols have to be declared +using some MS magic. I found some useful information on this web page: +http://msdn2.microsoft.com/en-us/library/y4h7bcy6(VS.80).aspx. According to the +information there, using __declspec(dllexport) without "extern" we have a +definition; with "extern" we have a declaration. The settings here override the +setting in pcre.h (which is included below); it defines only PCRE_EXP_DECL, +which is all that is needed for applications (they just import the symbols). We +use: + + PCRE_EXP_DECL for declarations + PCRE_EXP_DEFN for definitions of exported functions + PCRE_EXP_DATA_DEFN for definitions of exported variables + +The reason for the two DEFN macros is that in non-Windows environments, one +does not want to have "extern" before variable definitions because it leads to +compiler warnings. So we distinguish between functions and variables. In +Windows, the two should always be the same. + +The reason for wrapping this in #ifndef PCRE_EXP_DECL is so that pcretest, +which is an application, but needs to import this file in order to "peek" at +internals, can #include pcre.h first to get an application's-eye view. + +In principle, people compiling for non-Windows, non-Unix-like (i.e. uncommon, +special-purpose environments) might want to stick other stuff in front of +exported symbols. That's why, in the non-Windows case, we set PCRE_EXP_DEFN and +PCRE_EXP_DATA_DEFN only if they are not already set. */ + +#ifndef PCRE_EXP_DECL +# ifdef _WIN32 +# ifndef PCRE_STATIC +# define PCRE_EXP_DECL extern __declspec(dllexport) +# define PCRE_EXP_DEFN __declspec(dllexport) +# define PCRE_EXP_DATA_DEFN __declspec(dllexport) +# else +# define PCRE_EXP_DECL extern +# define PCRE_EXP_DEFN +# define PCRE_EXP_DATA_DEFN +# endif +# else +# ifdef __cplusplus +# define PCRE_EXP_DECL extern "C" +# else +# define PCRE_EXP_DECL extern +# endif +# ifndef PCRE_EXP_DEFN +# define PCRE_EXP_DEFN PCRE_EXP_DECL +# endif +# ifndef PCRE_EXP_DATA_DEFN +# define PCRE_EXP_DATA_DEFN +# endif +# endif +#endif + +/* When compiling with the MSVC compiler, it is sometimes necessary to include +a "calling convention" before exported function names. (This is secondhand +information; I know nothing about MSVC myself). For example, something like + + void __cdecl function(....) + +might be needed. In order so make this easy, all the exported functions have +PCRE_CALL_CONVENTION just before their names. It is rarely needed; if not +set, we ensure here that it has no effect. */ + +#ifndef PCRE_CALL_CONVENTION +#define PCRE_CALL_CONVENTION +#endif + +/* We need to have types that specify unsigned 8, 16 and 32-bit integers. We +cannot determine these outside the compilation (e.g. by running a program as +part of "configure") because PCRE is often cross-compiled for use on other +systems. Instead we make use of the maximum sizes that are available at +preprocessor time in standard C environments. */ + +typedef unsigned char pcre_uint8; + +#if USHRT_MAX == 65535 + typedef unsigned short pcre_uint16; + typedef short pcre_int16; +#elif UINT_MAX == 65535 + typedef unsigned int pcre_uint16; + typedef int pcre_int16; +#else + #error Cannot determine a type for 16-bit unsigned integers +#endif + +#if UINT_MAX == 4294967295 + typedef unsigned int pcre_uint32; + typedef int pcre_int32; +#elif ULONG_MAX == 4294967295 + typedef unsigned long int pcre_uint32; + typedef long int pcre_int32; +#else + #error Cannot determine a type for 32-bit unsigned integers +#endif + +/* When checking for integer overflow in pcre_compile(), we need to handle +large integers. If a 64-bit integer type is available, we can use that. +Otherwise we have to cast to double, which of course requires floating point +arithmetic. Handle this by defining a macro for the appropriate type. If +stdint.h is available, include it; it may define INT64_MAX. Systems that do not +have stdint.h (e.g. Solaris) may have inttypes.h. The macro int64_t may be set +by "configure". */ + +#if HAVE_STDINT_H +#include +#elif HAVE_INTTYPES_H +#include +#endif + +#if defined INT64_MAX || defined int64_t +#define INT64_OR_DOUBLE int64_t +#else +#define INT64_OR_DOUBLE double +#endif + +/* All character handling must be done as unsigned characters. Otherwise there +are problems with top-bit-set characters and functions such as isspace(). +However, we leave the interface to the outside world as char * or short *, +because that should make things easier for callers. This character type is +called pcre_uchar. + +The IN_UCHARS macro multiply its argument with the byte size of the current +pcre_uchar type. Useful for memcpy and such operations, whose require the +byte size of their input/output buffers. + +The MAX_255 macro checks whether its pcre_uchar input is less than 256. + +The TABLE_GET macro is designed for accessing elements of tables whose contain +exactly 256 items. When the character is able to contain more than 256 +items, some check is needed before accessing these tables. +*/ + +#ifdef COMPILE_PCRE8 + +typedef unsigned char pcre_uchar; +#define IN_UCHARS(x) (x) +#define MAX_255(c) 1 +#define TABLE_GET(c, table, default) ((table)[c]) + +#else + +#ifdef COMPILE_PCRE16 +#if USHRT_MAX != 65535 +/* This is a warning message. Change PCRE_UCHAR16 to a 16 bit data type in +pcre.h(.in) and disable (comment out) this message. */ +#error Warning: PCRE_UCHAR16 is not a 16 bit data type. +#endif + +typedef pcre_uint16 pcre_uchar; +#define IN_UCHARS(x) ((x) << 1) +#define MAX_255(c) ((c) <= 255u) +#define TABLE_GET(c, table, default) (MAX_255(c)? ((table)[c]):(default)) + +#else +#error Unsupported compiling mode +#endif /* COMPILE_PCRE16 */ + +#endif /* COMPILE_PCRE8 */ + +/* This is an unsigned int value that no character can ever have. UTF-8 +characters only go up to 0x7fffffff (though Unicode doesn't go beyond +0x0010ffff). */ + +#define NOTACHAR 0xffffffff + +/* PCRE is able to support several different kinds of newline (CR, LF, CRLF, +"any" and "anycrlf" at present). The following macros are used to package up +testing for newlines. NLBLOCK, PSSTART, and PSEND are defined in the various +modules to indicate in which datablock the parameters exist, and what the +start/end of string field names are. */ + +#define NLTYPE_FIXED 0 /* Newline is a fixed length string */ +#define NLTYPE_ANY 1 /* Newline is any Unicode line ending */ +#define NLTYPE_ANYCRLF 2 /* Newline is CR, LF, or CRLF */ + +/* This macro checks for a newline at the given position */ + +#define IS_NEWLINE(p) \ + ((NLBLOCK->nltype != NLTYPE_FIXED)? \ + ((p) < NLBLOCK->PSEND && \ + PRIV(is_newline)((p), NLBLOCK->nltype, NLBLOCK->PSEND, \ + &(NLBLOCK->nllen), utf)) \ + : \ + ((p) <= NLBLOCK->PSEND - NLBLOCK->nllen && \ + (p)[0] == NLBLOCK->nl[0] && \ + (NLBLOCK->nllen == 1 || (p)[1] == NLBLOCK->nl[1]) \ + ) \ + ) + +/* This macro checks for a newline immediately preceding the given position */ + +#define WAS_NEWLINE(p) \ + ((NLBLOCK->nltype != NLTYPE_FIXED)? \ + ((p) > NLBLOCK->PSSTART && \ + PRIV(was_newline)((p), NLBLOCK->nltype, NLBLOCK->PSSTART, \ + &(NLBLOCK->nllen), utf)) \ + : \ + ((p) >= NLBLOCK->PSSTART + NLBLOCK->nllen && \ + (p)[-NLBLOCK->nllen] == NLBLOCK->nl[0] && \ + (NLBLOCK->nllen == 1 || (p)[-NLBLOCK->nllen+1] == NLBLOCK->nl[1]) \ + ) \ + ) + +/* When PCRE is compiled as a C++ library, the subject pointer can be replaced +with a custom type. This makes it possible, for example, to allow pcre_exec() +to process subject strings that are discontinuous by using a smart pointer +class. It must always be possible to inspect all of the subject string in +pcre_exec() because of the way it backtracks. Two macros are required in the +normal case, for sign-unspecified and unsigned char pointers. The former is +used for the external interface and appears in pcre.h, which is why its name +must begin with PCRE_. */ + +#ifdef CUSTOM_SUBJECT_PTR +#define PCRE_PUCHAR CUSTOM_SUBJECT_PTR +#else +#define PCRE_PUCHAR const pcre_uchar * +#endif + +/* Include the public PCRE header and the definitions of UCP character property +values. */ + +#include "pcre.h" +#include "ucp.h" + +/* When compiling for use with the Virtual Pascal compiler, these functions +need to have their names changed. PCRE must be compiled with the -DVPCOMPAT +option on the command line. */ + +#ifdef VPCOMPAT +#define strlen(s) _strlen(s) +#define strncmp(s1,s2,m) _strncmp(s1,s2,m) +#define memcmp(s,c,n) _memcmp(s,c,n) +#define memcpy(d,s,n) _memcpy(d,s,n) +#define memmove(d,s,n) _memmove(d,s,n) +#define memset(s,c,n) _memset(s,c,n) +#else /* VPCOMPAT */ + +/* To cope with SunOS4 and other systems that lack memmove() but have bcopy(), +define a macro for memmove() if HAVE_MEMMOVE is false, provided that HAVE_BCOPY +is set. Otherwise, include an emulating function for those systems that have +neither (there some non-Unix environments where this is the case). */ + +#ifndef HAVE_MEMMOVE +#undef memmove /* some systems may have a macro */ +#ifdef HAVE_BCOPY +#define memmove(a, b, c) bcopy(b, a, c) +#else /* HAVE_BCOPY */ +static void * +pcre_memmove(void *d, const void *s, size_t n) +{ +size_t i; +unsigned char *dest = (unsigned char *)d; +const unsigned char *src = (const unsigned char *)s; +if (dest > src) + { + dest += n; + src += n; + for (i = 0; i < n; ++i) *(--dest) = *(--src); + return (void *)dest; + } +else + { + for (i = 0; i < n; ++i) *dest++ = *src++; + return (void *)(dest - n); + } +} +#define memmove(a, b, c) pcre_memmove(a, b, c) +#endif /* not HAVE_BCOPY */ +#endif /* not HAVE_MEMMOVE */ +#endif /* not VPCOMPAT */ + + +/* PCRE keeps offsets in its compiled code as 2-byte quantities (always stored +in big-endian order) by default. These are used, for example, to link from the +start of a subpattern to its alternatives and its end. The use of 2 bytes per +offset limits the size of the compiled regex to around 64K, which is big enough +for almost everybody. However, I received a request for an even bigger limit. +For this reason, and also to make the code easier to maintain, the storing and +loading of offsets from the byte string is now handled by the macros that are +defined here. + +The macros are controlled by the value of LINK_SIZE. This defaults to 2 in +the config.h file, but can be overridden by using -D on the command line. This +is automated on Unix systems via the "configure" command. */ + +#ifdef COMPILE_PCRE8 + +#if LINK_SIZE == 2 + +#define PUT(a,n,d) \ + (a[n] = (d) >> 8), \ + (a[(n)+1] = (d) & 255) + +#define GET(a,n) \ + (((a)[n] << 8) | (a)[(n)+1]) + +#define MAX_PATTERN_SIZE (1 << 16) + + +#elif LINK_SIZE == 3 + +#define PUT(a,n,d) \ + (a[n] = (d) >> 16), \ + (a[(n)+1] = (d) >> 8), \ + (a[(n)+2] = (d) & 255) + +#define GET(a,n) \ + (((a)[n] << 16) | ((a)[(n)+1] << 8) | (a)[(n)+2]) + +#define MAX_PATTERN_SIZE (1 << 24) + + +#elif LINK_SIZE == 4 + +#define PUT(a,n,d) \ + (a[n] = (d) >> 24), \ + (a[(n)+1] = (d) >> 16), \ + (a[(n)+2] = (d) >> 8), \ + (a[(n)+3] = (d) & 255) + +#define GET(a,n) \ + (((a)[n] << 24) | ((a)[(n)+1] << 16) | ((a)[(n)+2] << 8) | (a)[(n)+3]) + +/* Keep it positive */ +#define MAX_PATTERN_SIZE (1 << 30) + +#else +#error LINK_SIZE must be either 2, 3, or 4 +#endif + +#else /* COMPILE_PCRE8 */ + +#ifdef COMPILE_PCRE16 + +#if LINK_SIZE == 2 + +#undef LINK_SIZE +#define LINK_SIZE 1 + +#define PUT(a,n,d) \ + (a[n] = (d)) + +#define GET(a,n) \ + (a[n]) + +#define MAX_PATTERN_SIZE (1 << 16) + +#elif LINK_SIZE == 3 || LINK_SIZE == 4 + +#undef LINK_SIZE +#define LINK_SIZE 2 + +#define PUT(a,n,d) \ + (a[n] = (d) >> 16), \ + (a[(n)+1] = (d) & 65535) + +#define GET(a,n) \ + (((a)[n] << 16) | (a)[(n)+1]) + +/* Keep it positive */ +#define MAX_PATTERN_SIZE (1 << 30) + +#else +#error LINK_SIZE must be either 2, 3, or 4 +#endif + +#else +#error Unsupported compiling mode +#endif /* COMPILE_PCRE16 */ + +#endif /* COMPILE_PCRE8 */ + +/* Convenience macro defined in terms of the others */ + +#define PUTINC(a,n,d) PUT(a,n,d), a += LINK_SIZE + + +/* PCRE uses some other 2-byte quantities that do not change when the size of +offsets changes. There are used for repeat counts and for other things such as +capturing parenthesis numbers in back references. */ + +#ifdef COMPILE_PCRE8 + +#define IMM2_SIZE 2 + +#define PUT2(a,n,d) \ + a[n] = (d) >> 8; \ + a[(n)+1] = (d) & 255 + +#define GET2(a,n) \ + (((a)[n] << 8) | (a)[(n)+1]) + +#else /* COMPILE_PCRE8 */ + +#ifdef COMPILE_PCRE16 + +#define IMM2_SIZE 1 + +#define PUT2(a,n,d) \ + a[n] = d + +#define GET2(a,n) \ + a[n] + +#else +#error Unsupported compiling mode +#endif /* COMPILE_PCRE16 */ + +#endif /* COMPILE_PCRE8 */ + +#define PUT2INC(a,n,d) PUT2(a,n,d), a += IMM2_SIZE + +/* The maximum length of a MARK name is currently one data unit; it may be +changed in future to be a fixed number of bytes or to depend on LINK_SIZE. */ + +#define MAX_MARK ((1 << (sizeof(pcre_uchar)*8)) - 1) + +/* When UTF encoding is being used, a character is no longer just a single +character. The macros for character handling generate simple sequences when +used in character-mode, and more complicated ones for UTF characters. +GETCHARLENTEST and other macros are not used when UTF is not supported, +so they are not defined. To make sure they can never even appear when +UTF support is omitted, we don't even define them. */ + +#ifndef SUPPORT_UTF + +/* #define MAX_VALUE_FOR_SINGLE_CHAR */ +/* #define HAS_EXTRALEN(c) */ +/* #define GET_EXTRALEN(c) */ +/* #define NOT_FIRSTCHAR(c) */ +#define GETCHAR(c, eptr) c = *eptr; +#define GETCHARTEST(c, eptr) c = *eptr; +#define GETCHARINC(c, eptr) c = *eptr++; +#define GETCHARINCTEST(c, eptr) c = *eptr++; +#define GETCHARLEN(c, eptr, len) c = *eptr; +/* #define GETCHARLENTEST(c, eptr, len) */ +/* #define BACKCHAR(eptr) */ +/* #define FORWARDCHAR(eptr) */ +/* #define ACROSSCHAR(condition, eptr, action) */ + +#else /* SUPPORT_UTF */ + +#ifdef COMPILE_PCRE8 + +/* These macros were originally written in the form of loops that used data +from the tables whose names start with PRIV(utf8_table). They were rewritten by +a user so as not to use loops, because in some environments this gives a +significant performance advantage, and it seems never to do any harm. */ + +/* Tells the biggest code point which can be encoded as a single character. */ + +#define MAX_VALUE_FOR_SINGLE_CHAR 127 + +/* Tests whether the code point needs extra characters to decode. */ + +#define HAS_EXTRALEN(c) ((c) >= 0xc0) + +/* Returns with the additional number of characters if IS_MULTICHAR(c) is TRUE. +Otherwise it has an undefined behaviour. */ + +#define GET_EXTRALEN(c) (PRIV(utf8_table4)[(c) & 0x3f]) + +/* Returns TRUE, if the given character is not the first character +of a UTF sequence. */ + +#define NOT_FIRSTCHAR(c) (((c) & 0xc0) == 0x80) + +/* Base macro to pick up the remaining bytes of a UTF-8 character, not +advancing the pointer. */ + +#define GETUTF8(c, eptr) \ + { \ + if ((c & 0x20) == 0) \ + c = ((c & 0x1f) << 6) | (eptr[1] & 0x3f); \ + else if ((c & 0x10) == 0) \ + c = ((c & 0x0f) << 12) | ((eptr[1] & 0x3f) << 6) | (eptr[2] & 0x3f); \ + else if ((c & 0x08) == 0) \ + c = ((c & 0x07) << 18) | ((eptr[1] & 0x3f) << 12) | \ + ((eptr[2] & 0x3f) << 6) | (eptr[3] & 0x3f); \ + else if ((c & 0x04) == 0) \ + c = ((c & 0x03) << 24) | ((eptr[1] & 0x3f) << 18) | \ + ((eptr[2] & 0x3f) << 12) | ((eptr[3] & 0x3f) << 6) | \ + (eptr[4] & 0x3f); \ + else \ + c = ((c & 0x01) << 30) | ((eptr[1] & 0x3f) << 24) | \ + ((eptr[2] & 0x3f) << 18) | ((eptr[3] & 0x3f) << 12) | \ + ((eptr[4] & 0x3f) << 6) | (eptr[5] & 0x3f); \ + } + +/* Get the next UTF-8 character, not advancing the pointer. This is called when +we know we are in UTF-8 mode. */ + +#define GETCHAR(c, eptr) \ + c = *eptr; \ + if (c >= 0xc0) GETUTF8(c, eptr); + +/* Get the next UTF-8 character, testing for UTF-8 mode, and not advancing the +pointer. */ + +#define GETCHARTEST(c, eptr) \ + c = *eptr; \ + if (utf && c >= 0xc0) GETUTF8(c, eptr); + +/* Base macro to pick up the remaining bytes of a UTF-8 character, advancing +the pointer. */ + +#define GETUTF8INC(c, eptr) \ + { \ + if ((c & 0x20) == 0) \ + c = ((c & 0x1f) << 6) | (*eptr++ & 0x3f); \ + else if ((c & 0x10) == 0) \ + { \ + c = ((c & 0x0f) << 12) | ((*eptr & 0x3f) << 6) | (eptr[1] & 0x3f); \ + eptr += 2; \ + } \ + else if ((c & 0x08) == 0) \ + { \ + c = ((c & 0x07) << 18) | ((*eptr & 0x3f) << 12) | \ + ((eptr[1] & 0x3f) << 6) | (eptr[2] & 0x3f); \ + eptr += 3; \ + } \ + else if ((c & 0x04) == 0) \ + { \ + c = ((c & 0x03) << 24) | ((*eptr & 0x3f) << 18) | \ + ((eptr[1] & 0x3f) << 12) | ((eptr[2] & 0x3f) << 6) | \ + (eptr[3] & 0x3f); \ + eptr += 4; \ + } \ + else \ + { \ + c = ((c & 0x01) << 30) | ((*eptr & 0x3f) << 24) | \ + ((eptr[1] & 0x3f) << 18) | ((eptr[2] & 0x3f) << 12) | \ + ((eptr[3] & 0x3f) << 6) | (eptr[4] & 0x3f); \ + eptr += 5; \ + } \ + } + +/* Get the next UTF-8 character, advancing the pointer. This is called when we +know we are in UTF-8 mode. */ + +#define GETCHARINC(c, eptr) \ + c = *eptr++; \ + if (c >= 0xc0) GETUTF8INC(c, eptr); + +/* Get the next character, testing for UTF-8 mode, and advancing the pointer. +This is called when we don't know if we are in UTF-8 mode. */ + +#define GETCHARINCTEST(c, eptr) \ + c = *eptr++; \ + if (utf && c >= 0xc0) GETUTF8INC(c, eptr); + +/* Base macro to pick up the remaining bytes of a UTF-8 character, not +advancing the pointer, incrementing the length. */ + +#define GETUTF8LEN(c, eptr, len) \ + { \ + if ((c & 0x20) == 0) \ + { \ + c = ((c & 0x1f) << 6) | (eptr[1] & 0x3f); \ + len++; \ + } \ + else if ((c & 0x10) == 0) \ + { \ + c = ((c & 0x0f) << 12) | ((eptr[1] & 0x3f) << 6) | (eptr[2] & 0x3f); \ + len += 2; \ + } \ + else if ((c & 0x08) == 0) \ + {\ + c = ((c & 0x07) << 18) | ((eptr[1] & 0x3f) << 12) | \ + ((eptr[2] & 0x3f) << 6) | (eptr[3] & 0x3f); \ + len += 3; \ + } \ + else if ((c & 0x04) == 0) \ + { \ + c = ((c & 0x03) << 24) | ((eptr[1] & 0x3f) << 18) | \ + ((eptr[2] & 0x3f) << 12) | ((eptr[3] & 0x3f) << 6) | \ + (eptr[4] & 0x3f); \ + len += 4; \ + } \ + else \ + {\ + c = ((c & 0x01) << 30) | ((eptr[1] & 0x3f) << 24) | \ + ((eptr[2] & 0x3f) << 18) | ((eptr[3] & 0x3f) << 12) | \ + ((eptr[4] & 0x3f) << 6) | (eptr[5] & 0x3f); \ + len += 5; \ + } \ + } + +/* Get the next UTF-8 character, not advancing the pointer, incrementing length +if there are extra bytes. This is called when we know we are in UTF-8 mode. */ + +#define GETCHARLEN(c, eptr, len) \ + c = *eptr; \ + if (c >= 0xc0) GETUTF8LEN(c, eptr, len); + +/* Get the next UTF-8 character, testing for UTF-8 mode, not advancing the +pointer, incrementing length if there are extra bytes. This is called when we +do not know if we are in UTF-8 mode. */ + +#define GETCHARLENTEST(c, eptr, len) \ + c = *eptr; \ + if (utf && c >= 0xc0) GETUTF8LEN(c, eptr, len); + +/* If the pointer is not at the start of a character, move it back until +it is. This is called only in UTF-8 mode - we don't put a test within the macro +because almost all calls are already within a block of UTF-8 only code. */ + +#define BACKCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr-- + +/* Same as above, just in the other direction. */ +#define FORWARDCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr++ + +/* Same as above, but it allows a fully customizable form. */ +#define ACROSSCHAR(condition, eptr, action) \ + while((condition) && ((eptr) & 0xc0) == 0x80) action + +#else /* COMPILE_PCRE8 */ + +#ifdef COMPILE_PCRE16 + +/* Tells the biggest code point which can be encoded as a single character. */ + +#define MAX_VALUE_FOR_SINGLE_CHAR 65535 + +/* Tests whether the code point needs extra characters to decode. */ + +#define HAS_EXTRALEN(c) (((c) & 0xfc00) == 0xd800) + +/* Returns with the additional number of characters if IS_MULTICHAR(c) is TRUE. +Otherwise it has an undefined behaviour. */ + +#define GET_EXTRALEN(c) 1 + +/* Returns TRUE, if the given character is not the first character +of a UTF sequence. */ + +#define NOT_FIRSTCHAR(c) (((c) & 0xfc00) == 0xdc00) + +/* Base macro to pick up the low surrogate of a UTF-16 character, not +advancing the pointer. */ + +#define GETUTF16(c, eptr) \ + { c = (((c & 0x3ff) << 10) | (eptr[1] & 0x3ff)) + 0x10000; } + +/* Get the next UTF-16 character, not advancing the pointer. This is called when +we know we are in UTF-16 mode. */ + +#define GETCHAR(c, eptr) \ + c = *eptr; \ + if ((c & 0xfc00) == 0xd800) GETUTF16(c, eptr); + +/* Get the next UTF-16 character, testing for UTF-16 mode, and not advancing the +pointer. */ + +#define GETCHARTEST(c, eptr) \ + c = *eptr; \ + if (utf && (c & 0xfc00) == 0xd800) GETUTF16(c, eptr); + +/* Base macro to pick up the low surrogate of a UTF-16 character, advancing +the pointer. */ + +#define GETUTF16INC(c, eptr) \ + { c = (((c & 0x3ff) << 10) | (*eptr++ & 0x3ff)) + 0x10000; } + +/* Get the next UTF-16 character, advancing the pointer. This is called when we +know we are in UTF-16 mode. */ + +#define GETCHARINC(c, eptr) \ + c = *eptr++; \ + if ((c & 0xfc00) == 0xd800) GETUTF16INC(c, eptr); + +/* Get the next character, testing for UTF-16 mode, and advancing the pointer. +This is called when we don't know if we are in UTF-16 mode. */ + +#define GETCHARINCTEST(c, eptr) \ + c = *eptr++; \ + if (utf && (c & 0xfc00) == 0xd800) GETUTF16INC(c, eptr); + +/* Base macro to pick up the low surrogate of a UTF-16 character, not +advancing the pointer, incrementing the length. */ + +#define GETUTF16LEN(c, eptr, len) \ + { c = (((c & 0x3ff) << 10) | (eptr[1] & 0x3ff)) + 0x10000; len++; } + +/* Get the next UTF-16 character, not advancing the pointer, incrementing +length if there is a low surrogate. This is called when we know we are in +UTF-16 mode. */ + +#define GETCHARLEN(c, eptr, len) \ + c = *eptr; \ + if ((c & 0xfc00) == 0xd800) GETUTF16LEN(c, eptr, len); + +/* Get the next UTF-816character, testing for UTF-16 mode, not advancing the +pointer, incrementing length if there is a low surrogate. This is called when +we do not know if we are in UTF-16 mode. */ + +#define GETCHARLENTEST(c, eptr, len) \ + c = *eptr; \ + if (utf && (c & 0xfc00) == 0xd800) GETUTF16LEN(c, eptr, len); + +/* If the pointer is not at the start of a character, move it back until +it is. This is called only in UTF-16 mode - we don't put a test within the +macro because almost all calls are already within a block of UTF-16 only +code. */ + +#define BACKCHAR(eptr) if ((*eptr & 0xfc00) == 0xdc00) eptr-- + +/* Same as above, just in the other direction. */ +#define FORWARDCHAR(eptr) if ((*eptr & 0xfc00) == 0xdc00) eptr++ + +/* Same as above, but it allows a fully customizable form. */ +#define ACROSSCHAR(condition, eptr, action) \ + if ((condition) && ((eptr) & 0xfc00) == 0xdc00) action + +#endif + +#endif /* COMPILE_PCRE8 */ + +#endif /* SUPPORT_UTF */ + + +/* In case there is no definition of offsetof() provided - though any proper +Standard C system should have one. */ + +#ifndef offsetof +#define offsetof(p_type,field) ((size_t)&(((p_type *)0)->field)) +#endif + + +/* Private flags containing information about the compiled regex. They used to +live at the top end of the options word, but that got almost full, so now they +are in a 16-bit flags word. From release 8.00, PCRE_NOPARTIAL is unused, as +the restrictions on partial matching have been lifted. It remains for backwards +compatibility. */ + +#ifdef COMPILE_PCRE8 +#define PCRE_MODE 0x0001 /* compiled in 8 bit mode */ +#endif +#ifdef COMPILE_PCRE16 +#define PCRE_MODE 0x0002 /* compiled in 16 bit mode */ +#endif +#define PCRE_FIRSTSET 0x0010 /* first_char is set */ +#define PCRE_FCH_CASELESS 0x0020 /* caseless first char */ +#define PCRE_REQCHSET 0x0040 /* req_byte is set */ +#define PCRE_RCH_CASELESS 0x0080 /* caseless requested char */ +#define PCRE_STARTLINE 0x0100 /* start after \n for multiline */ +#define PCRE_NOPARTIAL 0x0200 /* can't use partial with this regex */ +#define PCRE_JCHANGED 0x0400 /* j option used in regex */ +#define PCRE_HASCRORLF 0x0800 /* explicit \r or \n in pattern */ +#define PCRE_HASTHEN 0x1000 /* pattern contains (*THEN) */ + +/* Flags for the "extra" block produced by pcre_study(). */ + +#define PCRE_STUDY_MAPPED 0x0001 /* a map of starting chars exists */ +#define PCRE_STUDY_MINLEN 0x0002 /* a minimum length field exists */ + +/* Masks for identifying the public options that are permitted at compile +time, run time, or study time, respectively. */ + +#define PCRE_NEWLINE_BITS (PCRE_NEWLINE_CR|PCRE_NEWLINE_LF|PCRE_NEWLINE_ANY| \ + PCRE_NEWLINE_ANYCRLF) + +#define PUBLIC_COMPILE_OPTIONS \ + (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \ + PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \ + PCRE_NO_AUTO_CAPTURE|PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE| \ + PCRE_DUPNAMES|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \ + PCRE_JAVASCRIPT_COMPAT|PCRE_UCP|PCRE_NO_START_OPTIMIZE) + +#define PUBLIC_EXEC_OPTIONS \ + (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NOTEMPTY_ATSTART| \ + PCRE_NO_UTF8_CHECK|PCRE_PARTIAL_HARD|PCRE_PARTIAL_SOFT|PCRE_NEWLINE_BITS| \ + PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE|PCRE_NO_START_OPTIMIZE) + +#define PUBLIC_DFA_EXEC_OPTIONS \ + (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NOTEMPTY_ATSTART| \ + PCRE_NO_UTF8_CHECK|PCRE_PARTIAL_HARD|PCRE_PARTIAL_SOFT|PCRE_DFA_SHORTEST| \ + PCRE_DFA_RESTART|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \ + PCRE_NO_START_OPTIMIZE) + +#define PUBLIC_STUDY_OPTIONS \ + (PCRE_STUDY_JIT_COMPILE|PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE| \ + PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE) + +/* Magic number to provide a small check against being handed junk. */ + +#define MAGIC_NUMBER 0x50435245UL /* 'PCRE' */ + +/* This variable is used to detect a loaded regular expression +in different endianness. */ + +#define REVERSED_MAGIC_NUMBER 0x45524350UL /* 'ERCP' */ + +/* Negative values for the firstchar and reqchar variables */ + +#define REQ_UNSET (-2) +#define REQ_NONE (-1) + +/* The maximum remaining length of subject we are prepared to search for a +req_byte match. */ + +#define REQ_BYTE_MAX 1000 + +/* Miscellaneous definitions. The #ifndef is to pacify compiler warnings in +environments where these macros are defined elsewhere. Unfortunately, there +is no way to do the same for the typedef. */ + +typedef int BOOL; + +#ifndef FALSE +#define FALSE 0 +#define TRUE 1 +#endif + +/* If PCRE is to support UTF-8 on EBCDIC platforms, we cannot use normal +character constants like '*' because the compiler would emit their EBCDIC code, +which is different from their ASCII/UTF-8 code. Instead we define macros for +the characters so that they always use the ASCII/UTF-8 code when UTF-8 support +is enabled. When UTF-8 support is not enabled, the definitions use character +literals. Both character and string versions of each character are needed, and +there are some longer strings as well. + +This means that, on EBCDIC platforms, the PCRE library can handle either +EBCDIC, or UTF-8, but not both. To support both in the same compiled library +would need different lookups depending on whether PCRE_UTF8 was set or not. +This would make it impossible to use characters in switch/case statements, +which would reduce performance. For a theoretical use (which nobody has asked +for) in a minority area (EBCDIC platforms), this is not sensible. Any +application that did need both could compile two versions of the library, using +macros to give the functions distinct names. */ + +#ifndef SUPPORT_UTF + +/* UTF-8 support is not enabled; use the platform-dependent character literals +so that PCRE works on both ASCII and EBCDIC platforms, in non-UTF-mode only. */ + +#define CHAR_HT '\t' +#define CHAR_VT '\v' +#define CHAR_FF '\f' +#define CHAR_CR '\r' +#define CHAR_NL '\n' +#define CHAR_BS '\b' +#define CHAR_BEL '\a' +#ifdef EBCDIC +#define CHAR_ESC '\047' +#define CHAR_DEL '\007' +#else +#define CHAR_ESC '\033' +#define CHAR_DEL '\177' +#endif + +#define CHAR_SPACE ' ' +#define CHAR_EXCLAMATION_MARK '!' +#define CHAR_QUOTATION_MARK '"' +#define CHAR_NUMBER_SIGN '#' +#define CHAR_DOLLAR_SIGN '$' +#define CHAR_PERCENT_SIGN '%' +#define CHAR_AMPERSAND '&' +#define CHAR_APOSTROPHE '\'' +#define CHAR_LEFT_PARENTHESIS '(' +#define CHAR_RIGHT_PARENTHESIS ')' +#define CHAR_ASTERISK '*' +#define CHAR_PLUS '+' +#define CHAR_COMMA ',' +#define CHAR_MINUS '-' +#define CHAR_DOT '.' +#define CHAR_SLASH '/' +#define CHAR_0 '0' +#define CHAR_1 '1' +#define CHAR_2 '2' +#define CHAR_3 '3' +#define CHAR_4 '4' +#define CHAR_5 '5' +#define CHAR_6 '6' +#define CHAR_7 '7' +#define CHAR_8 '8' +#define CHAR_9 '9' +#define CHAR_COLON ':' +#define CHAR_SEMICOLON ';' +#define CHAR_LESS_THAN_SIGN '<' +#define CHAR_EQUALS_SIGN '=' +#define CHAR_GREATER_THAN_SIGN '>' +#define CHAR_QUESTION_MARK '?' +#define CHAR_COMMERCIAL_AT '@' +#define CHAR_A 'A' +#define CHAR_B 'B' +#define CHAR_C 'C' +#define CHAR_D 'D' +#define CHAR_E 'E' +#define CHAR_F 'F' +#define CHAR_G 'G' +#define CHAR_H 'H' +#define CHAR_I 'I' +#define CHAR_J 'J' +#define CHAR_K 'K' +#define CHAR_L 'L' +#define CHAR_M 'M' +#define CHAR_N 'N' +#define CHAR_O 'O' +#define CHAR_P 'P' +#define CHAR_Q 'Q' +#define CHAR_R 'R' +#define CHAR_S 'S' +#define CHAR_T 'T' +#define CHAR_U 'U' +#define CHAR_V 'V' +#define CHAR_W 'W' +#define CHAR_X 'X' +#define CHAR_Y 'Y' +#define CHAR_Z 'Z' +#define CHAR_LEFT_SQUARE_BRACKET '[' +#define CHAR_BACKSLASH '\\' +#define CHAR_RIGHT_SQUARE_BRACKET ']' +#define CHAR_CIRCUMFLEX_ACCENT '^' +#define CHAR_UNDERSCORE '_' +#define CHAR_GRAVE_ACCENT '`' +#define CHAR_a 'a' +#define CHAR_b 'b' +#define CHAR_c 'c' +#define CHAR_d 'd' +#define CHAR_e 'e' +#define CHAR_f 'f' +#define CHAR_g 'g' +#define CHAR_h 'h' +#define CHAR_i 'i' +#define CHAR_j 'j' +#define CHAR_k 'k' +#define CHAR_l 'l' +#define CHAR_m 'm' +#define CHAR_n 'n' +#define CHAR_o 'o' +#define CHAR_p 'p' +#define CHAR_q 'q' +#define CHAR_r 'r' +#define CHAR_s 's' +#define CHAR_t 't' +#define CHAR_u 'u' +#define CHAR_v 'v' +#define CHAR_w 'w' +#define CHAR_x 'x' +#define CHAR_y 'y' +#define CHAR_z 'z' +#define CHAR_LEFT_CURLY_BRACKET '{' +#define CHAR_VERTICAL_LINE '|' +#define CHAR_RIGHT_CURLY_BRACKET '}' +#define CHAR_TILDE '~' + +#define STR_HT "\t" +#define STR_VT "\v" +#define STR_FF "\f" +#define STR_CR "\r" +#define STR_NL "\n" +#define STR_BS "\b" +#define STR_BEL "\a" +#ifdef EBCDIC +#define STR_ESC "\047" +#define STR_DEL "\007" +#else +#define STR_ESC "\033" +#define STR_DEL "\177" +#endif + +#define STR_SPACE " " +#define STR_EXCLAMATION_MARK "!" +#define STR_QUOTATION_MARK "\"" +#define STR_NUMBER_SIGN "#" +#define STR_DOLLAR_SIGN "$" +#define STR_PERCENT_SIGN "%" +#define STR_AMPERSAND "&" +#define STR_APOSTROPHE "'" +#define STR_LEFT_PARENTHESIS "(" +#define STR_RIGHT_PARENTHESIS ")" +#define STR_ASTERISK "*" +#define STR_PLUS "+" +#define STR_COMMA "," +#define STR_MINUS "-" +#define STR_DOT "." +#define STR_SLASH "/" +#define STR_0 "0" +#define STR_1 "1" +#define STR_2 "2" +#define STR_3 "3" +#define STR_4 "4" +#define STR_5 "5" +#define STR_6 "6" +#define STR_7 "7" +#define STR_8 "8" +#define STR_9 "9" +#define STR_COLON ":" +#define STR_SEMICOLON ";" +#define STR_LESS_THAN_SIGN "<" +#define STR_EQUALS_SIGN "=" +#define STR_GREATER_THAN_SIGN ">" +#define STR_QUESTION_MARK "?" +#define STR_COMMERCIAL_AT "@" +#define STR_A "A" +#define STR_B "B" +#define STR_C "C" +#define STR_D "D" +#define STR_E "E" +#define STR_F "F" +#define STR_G "G" +#define STR_H "H" +#define STR_I "I" +#define STR_J "J" +#define STR_K "K" +#define STR_L "L" +#define STR_M "M" +#define STR_N "N" +#define STR_O "O" +#define STR_P "P" +#define STR_Q "Q" +#define STR_R "R" +#define STR_S "S" +#define STR_T "T" +#define STR_U "U" +#define STR_V "V" +#define STR_W "W" +#define STR_X "X" +#define STR_Y "Y" +#define STR_Z "Z" +#define STR_LEFT_SQUARE_BRACKET "[" +#define STR_BACKSLASH "\\" +#define STR_RIGHT_SQUARE_BRACKET "]" +#define STR_CIRCUMFLEX_ACCENT "^" +#define STR_UNDERSCORE "_" +#define STR_GRAVE_ACCENT "`" +#define STR_a "a" +#define STR_b "b" +#define STR_c "c" +#define STR_d "d" +#define STR_e "e" +#define STR_f "f" +#define STR_g "g" +#define STR_h "h" +#define STR_i "i" +#define STR_j "j" +#define STR_k "k" +#define STR_l "l" +#define STR_m "m" +#define STR_n "n" +#define STR_o "o" +#define STR_p "p" +#define STR_q "q" +#define STR_r "r" +#define STR_s "s" +#define STR_t "t" +#define STR_u "u" +#define STR_v "v" +#define STR_w "w" +#define STR_x "x" +#define STR_y "y" +#define STR_z "z" +#define STR_LEFT_CURLY_BRACKET "{" +#define STR_VERTICAL_LINE "|" +#define STR_RIGHT_CURLY_BRACKET "}" +#define STR_TILDE "~" + +#define STRING_ACCEPT0 "ACCEPT\0" +#define STRING_COMMIT0 "COMMIT\0" +#define STRING_F0 "F\0" +#define STRING_FAIL0 "FAIL\0" +#define STRING_MARK0 "MARK\0" +#define STRING_PRUNE0 "PRUNE\0" +#define STRING_SKIP0 "SKIP\0" +#define STRING_THEN "THEN" + +#define STRING_alpha0 "alpha\0" +#define STRING_lower0 "lower\0" +#define STRING_upper0 "upper\0" +#define STRING_alnum0 "alnum\0" +#define STRING_ascii0 "ascii\0" +#define STRING_blank0 "blank\0" +#define STRING_cntrl0 "cntrl\0" +#define STRING_digit0 "digit\0" +#define STRING_graph0 "graph\0" +#define STRING_print0 "print\0" +#define STRING_punct0 "punct\0" +#define STRING_space0 "space\0" +#define STRING_word0 "word\0" +#define STRING_xdigit "xdigit" + +#define STRING_DEFINE "DEFINE" + +#define STRING_CR_RIGHTPAR "CR)" +#define STRING_LF_RIGHTPAR "LF)" +#define STRING_CRLF_RIGHTPAR "CRLF)" +#define STRING_ANY_RIGHTPAR "ANY)" +#define STRING_ANYCRLF_RIGHTPAR "ANYCRLF)" +#define STRING_BSR_ANYCRLF_RIGHTPAR "BSR_ANYCRLF)" +#define STRING_BSR_UNICODE_RIGHTPAR "BSR_UNICODE)" +#ifdef COMPILE_PCRE8 +#define STRING_UTF_RIGHTPAR "UTF8)" +#endif +#ifdef COMPILE_PCRE16 +#define STRING_UTF_RIGHTPAR "UTF16)" +#endif +#define STRING_UCP_RIGHTPAR "UCP)" +#define STRING_NO_START_OPT_RIGHTPAR "NO_START_OPT)" + +#else /* SUPPORT_UTF */ + +/* UTF-8 support is enabled; always use UTF-8 (=ASCII) character codes. This +works in both modes non-EBCDIC platforms, and on EBCDIC platforms in UTF-8 mode +only. */ + +#define CHAR_HT '\011' +#define CHAR_VT '\013' +#define CHAR_FF '\014' +#define CHAR_CR '\015' +#define CHAR_NL '\012' +#define CHAR_BS '\010' +#define CHAR_BEL '\007' +#define CHAR_ESC '\033' +#define CHAR_DEL '\177' + +#define CHAR_SPACE '\040' +#define CHAR_EXCLAMATION_MARK '\041' +#define CHAR_QUOTATION_MARK '\042' +#define CHAR_NUMBER_SIGN '\043' +#define CHAR_DOLLAR_SIGN '\044' +#define CHAR_PERCENT_SIGN '\045' +#define CHAR_AMPERSAND '\046' +#define CHAR_APOSTROPHE '\047' +#define CHAR_LEFT_PARENTHESIS '\050' +#define CHAR_RIGHT_PARENTHESIS '\051' +#define CHAR_ASTERISK '\052' +#define CHAR_PLUS '\053' +#define CHAR_COMMA '\054' +#define CHAR_MINUS '\055' +#define CHAR_DOT '\056' +#define CHAR_SLASH '\057' +#define CHAR_0 '\060' +#define CHAR_1 '\061' +#define CHAR_2 '\062' +#define CHAR_3 '\063' +#define CHAR_4 '\064' +#define CHAR_5 '\065' +#define CHAR_6 '\066' +#define CHAR_7 '\067' +#define CHAR_8 '\070' +#define CHAR_9 '\071' +#define CHAR_COLON '\072' +#define CHAR_SEMICOLON '\073' +#define CHAR_LESS_THAN_SIGN '\074' +#define CHAR_EQUALS_SIGN '\075' +#define CHAR_GREATER_THAN_SIGN '\076' +#define CHAR_QUESTION_MARK '\077' +#define CHAR_COMMERCIAL_AT '\100' +#define CHAR_A '\101' +#define CHAR_B '\102' +#define CHAR_C '\103' +#define CHAR_D '\104' +#define CHAR_E '\105' +#define CHAR_F '\106' +#define CHAR_G '\107' +#define CHAR_H '\110' +#define CHAR_I '\111' +#define CHAR_J '\112' +#define CHAR_K '\113' +#define CHAR_L '\114' +#define CHAR_M '\115' +#define CHAR_N '\116' +#define CHAR_O '\117' +#define CHAR_P '\120' +#define CHAR_Q '\121' +#define CHAR_R '\122' +#define CHAR_S '\123' +#define CHAR_T '\124' +#define CHAR_U '\125' +#define CHAR_V '\126' +#define CHAR_W '\127' +#define CHAR_X '\130' +#define CHAR_Y '\131' +#define CHAR_Z '\132' +#define CHAR_LEFT_SQUARE_BRACKET '\133' +#define CHAR_BACKSLASH '\134' +#define CHAR_RIGHT_SQUARE_BRACKET '\135' +#define CHAR_CIRCUMFLEX_ACCENT '\136' +#define CHAR_UNDERSCORE '\137' +#define CHAR_GRAVE_ACCENT '\140' +#define CHAR_a '\141' +#define CHAR_b '\142' +#define CHAR_c '\143' +#define CHAR_d '\144' +#define CHAR_e '\145' +#define CHAR_f '\146' +#define CHAR_g '\147' +#define CHAR_h '\150' +#define CHAR_i '\151' +#define CHAR_j '\152' +#define CHAR_k '\153' +#define CHAR_l '\154' +#define CHAR_m '\155' +#define CHAR_n '\156' +#define CHAR_o '\157' +#define CHAR_p '\160' +#define CHAR_q '\161' +#define CHAR_r '\162' +#define CHAR_s '\163' +#define CHAR_t '\164' +#define CHAR_u '\165' +#define CHAR_v '\166' +#define CHAR_w '\167' +#define CHAR_x '\170' +#define CHAR_y '\171' +#define CHAR_z '\172' +#define CHAR_LEFT_CURLY_BRACKET '\173' +#define CHAR_VERTICAL_LINE '\174' +#define CHAR_RIGHT_CURLY_BRACKET '\175' +#define CHAR_TILDE '\176' + +#define STR_HT "\011" +#define STR_VT "\013" +#define STR_FF "\014" +#define STR_CR "\015" +#define STR_NL "\012" +#define STR_BS "\010" +#define STR_BEL "\007" +#define STR_ESC "\033" +#define STR_DEL "\177" + +#define STR_SPACE "\040" +#define STR_EXCLAMATION_MARK "\041" +#define STR_QUOTATION_MARK "\042" +#define STR_NUMBER_SIGN "\043" +#define STR_DOLLAR_SIGN "\044" +#define STR_PERCENT_SIGN "\045" +#define STR_AMPERSAND "\046" +#define STR_APOSTROPHE "\047" +#define STR_LEFT_PARENTHESIS "\050" +#define STR_RIGHT_PARENTHESIS "\051" +#define STR_ASTERISK "\052" +#define STR_PLUS "\053" +#define STR_COMMA "\054" +#define STR_MINUS "\055" +#define STR_DOT "\056" +#define STR_SLASH "\057" +#define STR_0 "\060" +#define STR_1 "\061" +#define STR_2 "\062" +#define STR_3 "\063" +#define STR_4 "\064" +#define STR_5 "\065" +#define STR_6 "\066" +#define STR_7 "\067" +#define STR_8 "\070" +#define STR_9 "\071" +#define STR_COLON "\072" +#define STR_SEMICOLON "\073" +#define STR_LESS_THAN_SIGN "\074" +#define STR_EQUALS_SIGN "\075" +#define STR_GREATER_THAN_SIGN "\076" +#define STR_QUESTION_MARK "\077" +#define STR_COMMERCIAL_AT "\100" +#define STR_A "\101" +#define STR_B "\102" +#define STR_C "\103" +#define STR_D "\104" +#define STR_E "\105" +#define STR_F "\106" +#define STR_G "\107" +#define STR_H "\110" +#define STR_I "\111" +#define STR_J "\112" +#define STR_K "\113" +#define STR_L "\114" +#define STR_M "\115" +#define STR_N "\116" +#define STR_O "\117" +#define STR_P "\120" +#define STR_Q "\121" +#define STR_R "\122" +#define STR_S "\123" +#define STR_T "\124" +#define STR_U "\125" +#define STR_V "\126" +#define STR_W "\127" +#define STR_X "\130" +#define STR_Y "\131" +#define STR_Z "\132" +#define STR_LEFT_SQUARE_BRACKET "\133" +#define STR_BACKSLASH "\134" +#define STR_RIGHT_SQUARE_BRACKET "\135" +#define STR_CIRCUMFLEX_ACCENT "\136" +#define STR_UNDERSCORE "\137" +#define STR_GRAVE_ACCENT "\140" +#define STR_a "\141" +#define STR_b "\142" +#define STR_c "\143" +#define STR_d "\144" +#define STR_e "\145" +#define STR_f "\146" +#define STR_g "\147" +#define STR_h "\150" +#define STR_i "\151" +#define STR_j "\152" +#define STR_k "\153" +#define STR_l "\154" +#define STR_m "\155" +#define STR_n "\156" +#define STR_o "\157" +#define STR_p "\160" +#define STR_q "\161" +#define STR_r "\162" +#define STR_s "\163" +#define STR_t "\164" +#define STR_u "\165" +#define STR_v "\166" +#define STR_w "\167" +#define STR_x "\170" +#define STR_y "\171" +#define STR_z "\172" +#define STR_LEFT_CURLY_BRACKET "\173" +#define STR_VERTICAL_LINE "\174" +#define STR_RIGHT_CURLY_BRACKET "\175" +#define STR_TILDE "\176" + +#define STRING_ACCEPT0 STR_A STR_C STR_C STR_E STR_P STR_T "\0" +#define STRING_COMMIT0 STR_C STR_O STR_M STR_M STR_I STR_T "\0" +#define STRING_F0 STR_F "\0" +#define STRING_FAIL0 STR_F STR_A STR_I STR_L "\0" +#define STRING_MARK0 STR_M STR_A STR_R STR_K "\0" +#define STRING_PRUNE0 STR_P STR_R STR_U STR_N STR_E "\0" +#define STRING_SKIP0 STR_S STR_K STR_I STR_P "\0" +#define STRING_THEN STR_T STR_H STR_E STR_N + +#define STRING_alpha0 STR_a STR_l STR_p STR_h STR_a "\0" +#define STRING_lower0 STR_l STR_o STR_w STR_e STR_r "\0" +#define STRING_upper0 STR_u STR_p STR_p STR_e STR_r "\0" +#define STRING_alnum0 STR_a STR_l STR_n STR_u STR_m "\0" +#define STRING_ascii0 STR_a STR_s STR_c STR_i STR_i "\0" +#define STRING_blank0 STR_b STR_l STR_a STR_n STR_k "\0" +#define STRING_cntrl0 STR_c STR_n STR_t STR_r STR_l "\0" +#define STRING_digit0 STR_d STR_i STR_g STR_i STR_t "\0" +#define STRING_graph0 STR_g STR_r STR_a STR_p STR_h "\0" +#define STRING_print0 STR_p STR_r STR_i STR_n STR_t "\0" +#define STRING_punct0 STR_p STR_u STR_n STR_c STR_t "\0" +#define STRING_space0 STR_s STR_p STR_a STR_c STR_e "\0" +#define STRING_word0 STR_w STR_o STR_r STR_d "\0" +#define STRING_xdigit STR_x STR_d STR_i STR_g STR_i STR_t + +#define STRING_DEFINE STR_D STR_E STR_F STR_I STR_N STR_E + +#define STRING_CR_RIGHTPAR STR_C STR_R STR_RIGHT_PARENTHESIS +#define STRING_LF_RIGHTPAR STR_L STR_F STR_RIGHT_PARENTHESIS +#define STRING_CRLF_RIGHTPAR STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS +#define STRING_ANY_RIGHTPAR STR_A STR_N STR_Y STR_RIGHT_PARENTHESIS +#define STRING_ANYCRLF_RIGHTPAR STR_A STR_N STR_Y STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS +#define STRING_BSR_ANYCRLF_RIGHTPAR STR_B STR_S STR_R STR_UNDERSCORE STR_A STR_N STR_Y STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS +#define STRING_BSR_UNICODE_RIGHTPAR STR_B STR_S STR_R STR_UNDERSCORE STR_U STR_N STR_I STR_C STR_O STR_D STR_E STR_RIGHT_PARENTHESIS +#ifdef COMPILE_PCRE8 +#define STRING_UTF_RIGHTPAR STR_U STR_T STR_F STR_8 STR_RIGHT_PARENTHESIS +#endif +#ifdef COMPILE_PCRE16 +#define STRING_UTF_RIGHTPAR STR_U STR_T STR_F STR_1 STR_6 STR_RIGHT_PARENTHESIS +#endif +#define STRING_UCP_RIGHTPAR STR_U STR_C STR_P STR_RIGHT_PARENTHESIS +#define STRING_NO_START_OPT_RIGHTPAR STR_N STR_O STR_UNDERSCORE STR_S STR_T STR_A STR_R STR_T STR_UNDERSCORE STR_O STR_P STR_T STR_RIGHT_PARENTHESIS + +#endif /* SUPPORT_UTF */ + +/* Escape items that are just an encoding of a particular data value. */ + +#ifndef ESC_e +#define ESC_e CHAR_ESC +#endif + +#ifndef ESC_f +#define ESC_f CHAR_FF +#endif + +#ifndef ESC_n +#define ESC_n CHAR_NL +#endif + +#ifndef ESC_r +#define ESC_r CHAR_CR +#endif + +/* We can't officially use ESC_t because it is a POSIX reserved identifier +(presumably because of all the others like size_t). */ + +#ifndef ESC_tee +#define ESC_tee CHAR_HT +#endif + +/* Codes for different types of Unicode property */ + +#define PT_ANY 0 /* Any property - matches all chars */ +#define PT_LAMP 1 /* L& - the union of Lu, Ll, Lt */ +#define PT_GC 2 /* Specified general characteristic (e.g. L) */ +#define PT_PC 3 /* Specified particular characteristic (e.g. Lu) */ +#define PT_SC 4 /* Script (e.g. Han) */ +#define PT_ALNUM 5 /* Alphanumeric - the union of L and N */ +#define PT_SPACE 6 /* Perl space - Z plus 9,10,12,13 */ +#define PT_PXSPACE 7 /* POSIX space - Z plus 9,10,11,12,13 */ +#define PT_WORD 8 /* Word - L plus N plus underscore */ + +/* Flag bits and data types for the extended class (OP_XCLASS) for classes that +contain characters with values greater than 255. */ + +#define XCL_NOT 0x01 /* Flag: this is a negative class */ +#define XCL_MAP 0x02 /* Flag: a 32-byte map is present */ + +#define XCL_END 0 /* Marks end of individual items */ +#define XCL_SINGLE 1 /* Single item (one multibyte char) follows */ +#define XCL_RANGE 2 /* A range (two multibyte chars) follows */ +#define XCL_PROP 3 /* Unicode property (2-byte property code follows) */ +#define XCL_NOTPROP 4 /* Unicode inverted property (ditto) */ + +/* These are escaped items that aren't just an encoding of a particular data +value such as \n. They must have non-zero values, as check_escape() returns +their negation. Also, they must appear in the same order as in the opcode +definitions below, up to ESC_z. There's a dummy for OP_ALLANY because it +corresponds to "." in DOTALL mode rather than an escape sequence. It is also +used for [^] in JavaScript compatibility mode, and for \C in non-utf mode. In +non-DOTALL mode, "." behaves like \N. + +The special values ESC_DU, ESC_du, etc. are used instead of ESC_D, ESC_d, etc. +when PCRE_UCP is set, when replacement of \d etc by \p sequences is required. +They must be contiguous, and remain in order so that the replacements can be +looked up from a table. + +The final escape must be ESC_REF as subsequent values are used for +backreferences (\1, \2, \3, etc). There are two tests in the code for an escape +greater than ESC_b and less than ESC_Z to detect the types that may be +repeated. These are the types that consume characters. If any new escapes are +put in between that don't consume a character, that code will have to change. +*/ + +enum { ESC_A = 1, ESC_G, ESC_K, ESC_B, ESC_b, ESC_D, ESC_d, ESC_S, ESC_s, + ESC_W, ESC_w, ESC_N, ESC_dum, ESC_C, ESC_P, ESC_p, ESC_R, ESC_H, + ESC_h, ESC_V, ESC_v, ESC_X, ESC_Z, ESC_z, + ESC_E, ESC_Q, ESC_g, ESC_k, + ESC_DU, ESC_du, ESC_SU, ESC_su, ESC_WU, ESC_wu, + ESC_REF }; + +/* Opcode table: Starting from 1 (i.e. after OP_END), the values up to +OP_EOD must correspond in order to the list of escapes immediately above. + +*** NOTE NOTE NOTE *** Whenever this list is updated, the two macro definitions +that follow must also be updated to match. There are also tables called +"coptable" and "poptable" in pcre_dfa_exec.c that must be updated. */ + +enum { + OP_END, /* 0 End of pattern */ + + /* Values corresponding to backslashed metacharacters */ + + OP_SOD, /* 1 Start of data: \A */ + OP_SOM, /* 2 Start of match (subject + offset): \G */ + OP_SET_SOM, /* 3 Set start of match (\K) */ + OP_NOT_WORD_BOUNDARY, /* 4 \B */ + OP_WORD_BOUNDARY, /* 5 \b */ + OP_NOT_DIGIT, /* 6 \D */ + OP_DIGIT, /* 7 \d */ + OP_NOT_WHITESPACE, /* 8 \S */ + OP_WHITESPACE, /* 9 \s */ + OP_NOT_WORDCHAR, /* 10 \W */ + OP_WORDCHAR, /* 11 \w */ + + OP_ANY, /* 12 Match any character except newline */ + OP_ALLANY, /* 13 Match any character */ + OP_ANYBYTE, /* 14 Match any byte (\C); different to OP_ANY for UTF-8 */ + OP_NOTPROP, /* 15 \P (not Unicode property) */ + OP_PROP, /* 16 \p (Unicode property) */ + OP_ANYNL, /* 17 \R (any newline sequence) */ + OP_NOT_HSPACE, /* 18 \H (not horizontal whitespace) */ + OP_HSPACE, /* 19 \h (horizontal whitespace) */ + OP_NOT_VSPACE, /* 20 \V (not vertical whitespace) */ + OP_VSPACE, /* 21 \v (vertical whitespace) */ + OP_EXTUNI, /* 22 \X (extended Unicode sequence */ + OP_EODN, /* 23 End of data or \n at end of data: \Z. */ + OP_EOD, /* 24 End of data: \z */ + + OP_CIRC, /* 25 Start of line - not multiline */ + OP_CIRCM, /* 26 Start of line - multiline */ + OP_DOLL, /* 27 End of line - not multiline */ + OP_DOLLM, /* 28 End of line - multiline */ + OP_CHAR, /* 29 Match one character, casefully */ + OP_CHARI, /* 30 Match one character, caselessly */ + OP_NOT, /* 31 Match one character, not the given one, casefully */ + OP_NOTI, /* 32 Match one character, not the given one, caselessly */ + + /* The following sets of 13 opcodes must always be kept in step because + the offset from the first one is used to generate the others. */ + + /**** Single characters, caseful, must precede the caseless ones ****/ + + OP_STAR, /* 33 The maximizing and minimizing versions of */ + OP_MINSTAR, /* 34 these six opcodes must come in pairs, with */ + OP_PLUS, /* 35 the minimizing one second. */ + OP_MINPLUS, /* 36 */ + OP_QUERY, /* 37 */ + OP_MINQUERY, /* 38 */ + + OP_UPTO, /* 39 From 0 to n matches of one character, caseful*/ + OP_MINUPTO, /* 40 */ + OP_EXACT, /* 41 Exactly n matches */ + + OP_POSSTAR, /* 42 Possessified star, caseful */ + OP_POSPLUS, /* 43 Possessified plus, caseful */ + OP_POSQUERY, /* 44 Posesssified query, caseful */ + OP_POSUPTO, /* 45 Possessified upto, caseful */ + + /**** Single characters, caseless, must follow the caseful ones */ + + OP_STARI, /* 46 */ + OP_MINSTARI, /* 47 */ + OP_PLUSI, /* 48 */ + OP_MINPLUSI, /* 49 */ + OP_QUERYI, /* 50 */ + OP_MINQUERYI, /* 51 */ + + OP_UPTOI, /* 52 From 0 to n matches of one character, caseless */ + OP_MINUPTOI, /* 53 */ + OP_EXACTI, /* 54 */ + + OP_POSSTARI, /* 55 Possessified star, caseless */ + OP_POSPLUSI, /* 56 Possessified plus, caseless */ + OP_POSQUERYI, /* 57 Posesssified query, caseless */ + OP_POSUPTOI, /* 58 Possessified upto, caseless */ + + /**** The negated ones must follow the non-negated ones, and match them ****/ + /**** Negated single character, caseful; must precede the caseless ones ****/ + + OP_NOTSTAR, /* 59 The maximizing and minimizing versions of */ + OP_NOTMINSTAR, /* 60 these six opcodes must come in pairs, with */ + OP_NOTPLUS, /* 61 the minimizing one second. They must be in */ + OP_NOTMINPLUS, /* 62 exactly the same order as those above. */ + OP_NOTQUERY, /* 63 */ + OP_NOTMINQUERY, /* 64 */ + + OP_NOTUPTO, /* 65 From 0 to n matches, caseful */ + OP_NOTMINUPTO, /* 66 */ + OP_NOTEXACT, /* 67 Exactly n matches */ + + OP_NOTPOSSTAR, /* 68 Possessified versions, caseful */ + OP_NOTPOSPLUS, /* 69 */ + OP_NOTPOSQUERY, /* 70 */ + OP_NOTPOSUPTO, /* 71 */ + + /**** Negated single character, caseless; must follow the caseful ones ****/ + + OP_NOTSTARI, /* 72 */ + OP_NOTMINSTARI, /* 73 */ + OP_NOTPLUSI, /* 74 */ + OP_NOTMINPLUSI, /* 75 */ + OP_NOTQUERYI, /* 76 */ + OP_NOTMINQUERYI, /* 77 */ + + OP_NOTUPTOI, /* 78 From 0 to n matches, caseless */ + OP_NOTMINUPTOI, /* 79 */ + OP_NOTEXACTI, /* 80 Exactly n matches */ + + OP_NOTPOSSTARI, /* 81 Possessified versions, caseless */ + OP_NOTPOSPLUSI, /* 82 */ + OP_NOTPOSQUERYI, /* 83 */ + OP_NOTPOSUPTOI, /* 84 */ + + /**** Character types ****/ + + OP_TYPESTAR, /* 85 The maximizing and minimizing versions of */ + OP_TYPEMINSTAR, /* 86 these six opcodes must come in pairs, with */ + OP_TYPEPLUS, /* 87 the minimizing one second. These codes must */ + OP_TYPEMINPLUS, /* 88 be in exactly the same order as those above. */ + OP_TYPEQUERY, /* 89 */ + OP_TYPEMINQUERY, /* 90 */ + + OP_TYPEUPTO, /* 91 From 0 to n matches */ + OP_TYPEMINUPTO, /* 92 */ + OP_TYPEEXACT, /* 93 Exactly n matches */ + + OP_TYPEPOSSTAR, /* 94 Possessified versions */ + OP_TYPEPOSPLUS, /* 95 */ + OP_TYPEPOSQUERY, /* 96 */ + OP_TYPEPOSUPTO, /* 97 */ + + /* These are used for character classes and back references; only the + first six are the same as the sets above. */ + + OP_CRSTAR, /* 98 The maximizing and minimizing versions of */ + OP_CRMINSTAR, /* 99 all these opcodes must come in pairs, with */ + OP_CRPLUS, /* 100 the minimizing one second. These codes must */ + OP_CRMINPLUS, /* 101 be in exactly the same order as those above. */ + OP_CRQUERY, /* 102 */ + OP_CRMINQUERY, /* 103 */ + + OP_CRRANGE, /* 104 These are different to the three sets above. */ + OP_CRMINRANGE, /* 105 */ + + /* End of quantifier opcodes */ + + OP_CLASS, /* 106 Match a character class, chars < 256 only */ + OP_NCLASS, /* 107 Same, but the bitmap was created from a negative + class - the difference is relevant only when a + character > 255 is encountered. */ + OP_XCLASS, /* 108 Extended class for handling > 255 chars within the + class. This does both positive and negative. */ + OP_REF, /* 109 Match a back reference, casefully */ + OP_REFI, /* 110 Match a back reference, caselessly */ + OP_RECURSE, /* 111 Match a numbered subpattern (possibly recursive) */ + OP_CALLOUT, /* 112 Call out to external function if provided */ + + OP_ALT, /* 113 Start of alternation */ + OP_KET, /* 114 End of group that doesn't have an unbounded repeat */ + OP_KETRMAX, /* 115 These two must remain together and in this */ + OP_KETRMIN, /* 116 order. They are for groups the repeat for ever. */ + OP_KETRPOS, /* 117 Possessive unlimited repeat. */ + + /* The assertions must come before BRA, CBRA, ONCE, and COND, and the four + asserts must remain in order. */ + + OP_REVERSE, /* 118 Move pointer back - used in lookbehind assertions */ + OP_ASSERT, /* 119 Positive lookahead */ + OP_ASSERT_NOT, /* 120 Negative lookahead */ + OP_ASSERTBACK, /* 121 Positive lookbehind */ + OP_ASSERTBACK_NOT, /* 122 Negative lookbehind */ + + /* ONCE, ONCE_NC, BRA, BRAPOS, CBRA, CBRAPOS, and COND must come immediately + after the assertions, with ONCE first, as there's a test for >= ONCE for a + subpattern that isn't an assertion. The POS versions must immediately follow + the non-POS versions in each case. */ + + OP_ONCE, /* 123 Atomic group, contains captures */ + OP_ONCE_NC, /* 124 Atomic group containing no captures */ + OP_BRA, /* 125 Start of non-capturing bracket */ + OP_BRAPOS, /* 126 Ditto, with unlimited, possessive repeat */ + OP_CBRA, /* 127 Start of capturing bracket */ + OP_CBRAPOS, /* 128 Ditto, with unlimited, possessive repeat */ + OP_COND, /* 129 Conditional group */ + + /* These five must follow the previous five, in the same order. There's a + check for >= SBRA to distinguish the two sets. */ + + OP_SBRA, /* 130 Start of non-capturing bracket, check empty */ + OP_SBRAPOS, /* 131 Ditto, with unlimited, possessive repeat */ + OP_SCBRA, /* 132 Start of capturing bracket, check empty */ + OP_SCBRAPOS, /* 133 Ditto, with unlimited, possessive repeat */ + OP_SCOND, /* 134 Conditional group, check empty */ + + /* The next two pairs must (respectively) be kept together. */ + + OP_CREF, /* 135 Used to hold a capture number as condition */ + OP_NCREF, /* 136 Same, but generated by a name reference*/ + OP_RREF, /* 137 Used to hold a recursion number as condition */ + OP_NRREF, /* 138 Same, but generated by a name reference*/ + OP_DEF, /* 139 The DEFINE condition */ + + OP_BRAZERO, /* 140 These two must remain together and in this */ + OP_BRAMINZERO, /* 141 order. */ + OP_BRAPOSZERO, /* 142 */ + + /* These are backtracking control verbs */ + + OP_MARK, /* 143 always has an argument */ + OP_PRUNE, /* 144 */ + OP_PRUNE_ARG, /* 145 same, but with argument */ + OP_SKIP, /* 146 */ + OP_SKIP_ARG, /* 147 same, but with argument */ + OP_THEN, /* 148 */ + OP_THEN_ARG, /* 149 same, but with argument */ + OP_COMMIT, /* 150 */ + + /* These are forced failure and success verbs */ + + OP_FAIL, /* 151 */ + OP_ACCEPT, /* 152 */ + OP_ASSERT_ACCEPT, /* 153 Used inside assertions */ + OP_CLOSE, /* 154 Used before OP_ACCEPT to close open captures */ + + /* This is used to skip a subpattern with a {0} quantifier */ + + OP_SKIPZERO, /* 155 */ + + /* This is not an opcode, but is used to check that tables indexed by opcode + are the correct length, in order to catch updating errors - there have been + some in the past. */ + + OP_TABLE_LENGTH +}; + +/* *** NOTE NOTE NOTE *** Whenever the list above is updated, the two macro +definitions that follow must also be updated to match. There are also tables +called "coptable" and "poptable" in pcre_dfa_exec.c that must be updated. */ + + +/* This macro defines textual names for all the opcodes. These are used only +for debugging, and some of them are only partial names. The macro is referenced +only in pcre_printint.c, which fills out the full names in many cases (and in +some cases doesn't actually use these names at all). */ + +#define OP_NAME_LIST \ + "End", "\\A", "\\G", "\\K", "\\B", "\\b", "\\D", "\\d", \ + "\\S", "\\s", "\\W", "\\w", "Any", "AllAny", "Anybyte", \ + "notprop", "prop", "\\R", "\\H", "\\h", "\\V", "\\v", \ + "extuni", "\\Z", "\\z", \ + "^", "^", "$", "$", "char", "chari", "not", "noti", \ + "*", "*?", "+", "+?", "?", "??", \ + "{", "{", "{", \ + "*+","++", "?+", "{", \ + "*", "*?", "+", "+?", "?", "??", \ + "{", "{", "{", \ + "*+","++", "?+", "{", \ + "*", "*?", "+", "+?", "?", "??", \ + "{", "{", "{", \ + "*+","++", "?+", "{", \ + "*", "*?", "+", "+?", "?", "??", \ + "{", "{", "{", \ + "*+","++", "?+", "{", \ + "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \ + "*+","++", "?+", "{", \ + "*", "*?", "+", "+?", "?", "??", "{", "{", \ + "class", "nclass", "xclass", "Ref", "Refi", \ + "Recurse", "Callout", \ + "Alt", "Ket", "KetRmax", "KetRmin", "KetRpos", \ + "Reverse", "Assert", "Assert not", "AssertB", "AssertB not", \ + "Once", "Once_NC", \ + "Bra", "BraPos", "CBra", "CBraPos", \ + "Cond", \ + "SBra", "SBraPos", "SCBra", "SCBraPos", \ + "SCond", \ + "Cond ref", "Cond nref", "Cond rec", "Cond nrec", "Cond def", \ + "Brazero", "Braminzero", "Braposzero", \ + "*MARK", "*PRUNE", "*PRUNE", "*SKIP", "*SKIP", \ + "*THEN", "*THEN", "*COMMIT", "*FAIL", \ + "*ACCEPT", "*ASSERT_ACCEPT", \ + "Close", "Skip zero" + + +/* This macro defines the length of fixed length operations in the compiled +regex. The lengths are used when searching for specific things, and also in the +debugging printing of a compiled regex. We use a macro so that it can be +defined close to the definitions of the opcodes themselves. + +As things have been extended, some of these are no longer fixed lenths, but are +minima instead. For example, the length of a single-character repeat may vary +in UTF-8 mode. The code that uses this table must know about such things. */ + +#define OP_LENGTHS \ + 1, /* End */ \ + 1, 1, 1, 1, 1, /* \A, \G, \K, \B, \b */ \ + 1, 1, 1, 1, 1, 1, /* \D, \d, \S, \s, \W, \w */ \ + 1, 1, 1, /* Any, AllAny, Anybyte */ \ + 3, 3, /* \P, \p */ \ + 1, 1, 1, 1, 1, /* \R, \H, \h, \V, \v */ \ + 1, /* \X */ \ + 1, 1, 1, 1, 1, 1, /* \Z, \z, ^, ^M, $, $M */ \ + 2, /* Char - the minimum length */ \ + 2, /* Chari - the minimum length */ \ + 2, /* not */ \ + 2, /* noti */ \ + /* Positive single-char repeats ** These are */ \ + 2, 2, 2, 2, 2, 2, /* *, *?, +, +?, ?, ?? ** minima in */ \ + 2+IMM2_SIZE, 2+IMM2_SIZE, /* upto, minupto ** mode */ \ + 2+IMM2_SIZE, /* exact */ \ + 2, 2, 2, 2+IMM2_SIZE, /* *+, ++, ?+, upto+ */ \ + 2, 2, 2, 2, 2, 2, /* *I, *?I, +I, +?I, ?I, ??I ** UTF-8 */ \ + 2+IMM2_SIZE, 2+IMM2_SIZE, /* upto I, minupto I */ \ + 2+IMM2_SIZE, /* exact I */ \ + 2, 2, 2, 2+IMM2_SIZE, /* *+I, ++I, ?+I, upto+I */ \ + /* Negative single-char repeats - only for chars < 256 */ \ + 2, 2, 2, 2, 2, 2, /* NOT *, *?, +, +?, ?, ?? */ \ + 2+IMM2_SIZE, 2+IMM2_SIZE, /* NOT upto, minupto */ \ + 2+IMM2_SIZE, /* NOT exact */ \ + 2, 2, 2, 2+IMM2_SIZE, /* Possessive NOT *, +, ?, upto */ \ + 2, 2, 2, 2, 2, 2, /* NOT *I, *?I, +I, +?I, ?I, ??I */ \ + 2+IMM2_SIZE, 2+IMM2_SIZE, /* NOT upto I, minupto I */ \ + 2+IMM2_SIZE, /* NOT exact I */ \ + 2, 2, 2, 2+IMM2_SIZE, /* Possessive NOT *I, +I, ?I, upto I */ \ + /* Positive type repeats */ \ + 2, 2, 2, 2, 2, 2, /* Type *, *?, +, +?, ?, ?? */ \ + 2+IMM2_SIZE, 2+IMM2_SIZE, /* Type upto, minupto */ \ + 2+IMM2_SIZE, /* Type exact */ \ + 2, 2, 2, 2+IMM2_SIZE, /* Possessive *+, ++, ?+, upto+ */ \ + /* Character class & ref repeats */ \ + 1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */ \ + 1+2*IMM2_SIZE, 1+2*IMM2_SIZE, /* CRRANGE, CRMINRANGE */ \ + 1+(32/sizeof(pcre_uchar)), /* CLASS */ \ + 1+(32/sizeof(pcre_uchar)), /* NCLASS */ \ + 0, /* XCLASS - variable length */ \ + 1+IMM2_SIZE, /* REF */ \ + 1+IMM2_SIZE, /* REFI */ \ + 1+LINK_SIZE, /* RECURSE */ \ + 2+2*LINK_SIZE, /* CALLOUT */ \ + 1+LINK_SIZE, /* Alt */ \ + 1+LINK_SIZE, /* Ket */ \ + 1+LINK_SIZE, /* KetRmax */ \ + 1+LINK_SIZE, /* KetRmin */ \ + 1+LINK_SIZE, /* KetRpos */ \ + 1+LINK_SIZE, /* Reverse */ \ + 1+LINK_SIZE, /* Assert */ \ + 1+LINK_SIZE, /* Assert not */ \ + 1+LINK_SIZE, /* Assert behind */ \ + 1+LINK_SIZE, /* Assert behind not */ \ + 1+LINK_SIZE, /* ONCE */ \ + 1+LINK_SIZE, /* ONCE_NC */ \ + 1+LINK_SIZE, /* BRA */ \ + 1+LINK_SIZE, /* BRAPOS */ \ + 1+LINK_SIZE+IMM2_SIZE, /* CBRA */ \ + 1+LINK_SIZE+IMM2_SIZE, /* CBRAPOS */ \ + 1+LINK_SIZE, /* COND */ \ + 1+LINK_SIZE, /* SBRA */ \ + 1+LINK_SIZE, /* SBRAPOS */ \ + 1+LINK_SIZE+IMM2_SIZE, /* SCBRA */ \ + 1+LINK_SIZE+IMM2_SIZE, /* SCBRAPOS */ \ + 1+LINK_SIZE, /* SCOND */ \ + 1+IMM2_SIZE, 1+IMM2_SIZE, /* CREF, NCREF */ \ + 1+IMM2_SIZE, 1+IMM2_SIZE, /* RREF, NRREF */ \ + 1, /* DEF */ \ + 1, 1, 1, /* BRAZERO, BRAMINZERO, BRAPOSZERO */ \ + 3, 1, 3, /* MARK, PRUNE, PRUNE_ARG */ \ + 1, 3, /* SKIP, SKIP_ARG */ \ + 1, 3, /* THEN, THEN_ARG */ \ + 1, 1, 1, 1, /* COMMIT, FAIL, ACCEPT, ASSERT_ACCEPT */ \ + 1+IMM2_SIZE, 1 /* CLOSE, SKIPZERO */ + +/* A magic value for OP_RREF and OP_NRREF to indicate the "any recursion" +condition. */ + +#define RREF_ANY 0xffff + +/* Compile time error code numbers. They are given names so that they can more +easily be tracked. When a new number is added, the table called eint in +pcreposix.c must be updated. */ + +enum { ERR0, ERR1, ERR2, ERR3, ERR4, ERR5, ERR6, ERR7, ERR8, ERR9, + ERR10, ERR11, ERR12, ERR13, ERR14, ERR15, ERR16, ERR17, ERR18, ERR19, + ERR20, ERR21, ERR22, ERR23, ERR24, ERR25, ERR26, ERR27, ERR28, ERR29, + ERR30, ERR31, ERR32, ERR33, ERR34, ERR35, ERR36, ERR37, ERR38, ERR39, + ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49, + ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59, + ERR60, ERR61, ERR62, ERR63, ERR64, ERR65, ERR66, ERR67, ERR68, ERR69, + ERR70, ERR71, ERR72, ERR73, ERR74, ERR75, ERR76, ERRCOUNT }; + +/* JIT compiling modes. The function list is indexed by them. */ +enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COMPILE, JIT_PARTIAL_HARD_COMPILE, + JIT_NUMBER_OF_COMPILE_MODES }; + +/* The real format of the start of the pcre block; the index of names and the +code vector run on as long as necessary after the end. We store an explicit +offset to the name table so that if a regex is compiled on one host, saved, and +then run on another where the size of pointers is different, all might still +be well. For the case of compiled-on-4 and run-on-8, we include an extra +pointer that is always NULL. For future-proofing, a few dummy fields were +originally included - even though you can never get this planning right - but +there is only one left now. + +NOTE NOTE NOTE: +Because people can now save and re-use compiled patterns, any additions to this +structure should be made at the end, and something earlier (e.g. a new +flag in the options or one of the dummy fields) should indicate that the new +fields are present. Currently PCRE always sets the dummy fields to zero. +NOTE NOTE NOTE +*/ + +#ifdef COMPILE_PCRE8 +#define REAL_PCRE real_pcre +#else +#define REAL_PCRE real_pcre16 +#endif + +typedef struct REAL_PCRE { + pcre_uint32 magic_number; + pcre_uint32 size; /* Total that was malloced */ + pcre_uint32 options; /* Public options */ + pcre_uint16 flags; /* Private flags */ + pcre_uint16 max_lookbehind; /* Longest lookbehind (characters) */ + pcre_uint16 top_bracket; /* Highest numbered group */ + pcre_uint16 top_backref; /* Highest numbered back reference */ + pcre_uint16 first_char; /* Starting character */ + pcre_uint16 req_char; /* This character must be seen */ + pcre_uint16 name_table_offset; /* Offset to name table that follows */ + pcre_uint16 name_entry_size; /* Size of any name items */ + pcre_uint16 name_count; /* Number of name items */ + pcre_uint16 ref_count; /* Reference count */ + const pcre_uint8 *tables; /* Pointer to tables or NULL for std */ + const pcre_uint8 *nullpad; /* NULL padding */ +} REAL_PCRE; + +/* The format of the block used to store data from pcre_study(). The same +remark (see NOTE above) about extending this structure applies. */ + +typedef struct pcre_study_data { + pcre_uint32 size; /* Total that was malloced */ + pcre_uint32 flags; /* Private flags */ + pcre_uint8 start_bits[32]; /* Starting char bits */ + pcre_uint32 minlength; /* Minimum subject length */ +} pcre_study_data; + +/* Structure for building a chain of open capturing subpatterns during +compiling, so that instructions to close them can be compiled when (*ACCEPT) is +encountered. This is also used to identify subpatterns that contain recursive +back references to themselves, so that they can be made atomic. */ + +typedef struct open_capitem { + struct open_capitem *next; /* Chain link */ + pcre_uint16 number; /* Capture number */ + pcre_uint16 flag; /* Set TRUE if recursive back ref */ +} open_capitem; + +/* Structure for passing "static" information around between the functions +doing the compiling, so that they are thread-safe. */ + +typedef struct compile_data { + const pcre_uint8 *lcc; /* Points to lower casing table */ + const pcre_uint8 *fcc; /* Points to case-flipping table */ + const pcre_uint8 *cbits; /* Points to character type table */ + const pcre_uint8 *ctypes; /* Points to table of type maps */ + const pcre_uchar *start_workspace;/* The start of working space */ + const pcre_uchar *start_code; /* The start of the compiled code */ + const pcre_uchar *start_pattern; /* The start of the pattern */ + const pcre_uchar *end_pattern; /* The end of the pattern */ + open_capitem *open_caps; /* Chain of open capture items */ + pcre_uchar *hwm; /* High watermark of workspace */ + pcre_uchar *name_table; /* The name/number table */ + int names_found; /* Number of entries so far */ + int name_entry_size; /* Size of each entry */ + int workspace_size; /* Size of workspace */ + int bracount; /* Count of capturing parens as we compile */ + int final_bracount; /* Saved value after first pass */ + int max_lookbehind; /* Maximum lookbehind (characters) */ + int top_backref; /* Maximum back reference */ + unsigned int backref_map; /* Bitmap of low back refs */ + int assert_depth; /* Depth of nested assertions */ + int external_options; /* External (initial) options */ + int external_flags; /* External flag bits to be set */ + int req_varyopt; /* "After variable item" flag for reqbyte */ + BOOL had_accept; /* (*ACCEPT) encountered */ + BOOL check_lookbehind; /* Lookbehinds need later checking */ + int nltype; /* Newline type */ + int nllen; /* Newline string length */ + pcre_uchar nl[4]; /* Newline string when fixed length */ +} compile_data; + +/* Structure for maintaining a chain of pointers to the currently incomplete +branches, for testing for left recursion while compiling. */ + +typedef struct branch_chain { + struct branch_chain *outer; + pcre_uchar *current_branch; +} branch_chain; + +/* Structure for items in a linked list that represents an explicit recursive +call within the pattern; used by pcre_exec(). */ + +typedef struct recursion_info { + struct recursion_info *prevrec; /* Previous recursion record (or NULL) */ + int group_num; /* Number of group that was called */ + int *offset_save; /* Pointer to start of saved offsets */ + int saved_max; /* Number of saved offsets */ + PCRE_PUCHAR subject_position; /* Position at start of recursion */ +} recursion_info; + +/* A similar structure for pcre_dfa_exec(). */ + +typedef struct dfa_recursion_info { + struct dfa_recursion_info *prevrec; + int group_num; + PCRE_PUCHAR subject_position; +} dfa_recursion_info; + +/* Structure for building a chain of data for holding the values of the subject +pointer at the start of each subpattern, so as to detect when an empty string +has been matched by a subpattern - to break infinite loops; used by +pcre_exec(). */ + +typedef struct eptrblock { + struct eptrblock *epb_prev; + PCRE_PUCHAR epb_saved_eptr; +} eptrblock; + + +/* Structure for passing "static" information around between the functions +doing traditional NFA matching, so that they are thread-safe. */ + +typedef struct match_data { + unsigned long int match_call_count; /* As it says */ + unsigned long int match_limit; /* As it says */ + unsigned long int match_limit_recursion; /* As it says */ + int *offset_vector; /* Offset vector */ + int offset_end; /* One past the end */ + int offset_max; /* The maximum usable for return data */ + int nltype; /* Newline type */ + int nllen; /* Newline string length */ + int name_count; /* Number of names in name table */ + int name_entry_size; /* Size of entry in names table */ + pcre_uchar *name_table; /* Table of names */ + pcre_uchar nl[4]; /* Newline string when fixed */ + const pcre_uint8 *lcc; /* Points to lower casing table */ + const pcre_uint8 *fcc; /* Points to case-flipping table */ + const pcre_uint8 *ctypes; /* Points to table of type maps */ + BOOL offset_overflow; /* Set if too many extractions */ + BOOL notbol; /* NOTBOL flag */ + BOOL noteol; /* NOTEOL flag */ + BOOL utf; /* UTF-8 / UTF-16 flag */ + BOOL jscript_compat; /* JAVASCRIPT_COMPAT flag */ + BOOL use_ucp; /* PCRE_UCP flag */ + BOOL endonly; /* Dollar not before final \n */ + BOOL notempty; /* Empty string match not wanted */ + BOOL notempty_atstart; /* Empty string match at start not wanted */ + BOOL hitend; /* Hit the end of the subject at some point */ + BOOL bsr_anycrlf; /* \R is just any CRLF, not full Unicode */ + BOOL hasthen; /* Pattern contains (*THEN) */ + BOOL ignore_skip_arg; /* For re-run when SKIP name not found */ + const pcre_uchar *start_code; /* For use when recursing */ + PCRE_PUCHAR start_subject; /* Start of the subject string */ + PCRE_PUCHAR end_subject; /* End of the subject string */ + PCRE_PUCHAR start_match_ptr; /* Start of matched string */ + PCRE_PUCHAR end_match_ptr; /* Subject position at end match */ + PCRE_PUCHAR start_used_ptr; /* Earliest consulted character */ + int partial; /* PARTIAL options */ + int end_offset_top; /* Highwater mark at end of match */ + int capture_last; /* Most recent capture number */ + int start_offset; /* The start offset value */ + int match_function_type; /* Set for certain special calls of MATCH() */ + eptrblock *eptrchain; /* Chain of eptrblocks for tail recursions */ + int eptrn; /* Next free eptrblock */ + recursion_info *recursive; /* Linked list of recursion data */ + void *callout_data; /* To pass back to callouts */ + const pcre_uchar *mark; /* Mark pointer to pass back on success */ + const pcre_uchar *nomatch_mark;/* Mark pointer to pass back on failure */ + const pcre_uchar *once_target; /* Where to back up to for atomic groups */ +#ifdef NO_RECURSE + void *match_frames_base; /* For remembering malloc'd frames */ +#endif +} match_data; + +/* A similar structure is used for the same purpose by the DFA matching +functions. */ + +typedef struct dfa_match_data { + const pcre_uchar *start_code; /* Start of the compiled pattern */ + const pcre_uchar *start_subject ; /* Start of the subject string */ + const pcre_uchar *end_subject; /* End of subject string */ + const pcre_uchar *start_used_ptr; /* Earliest consulted character */ + const pcre_uint8 *tables; /* Character tables */ + int start_offset; /* The start offset value */ + int moptions; /* Match options */ + int poptions; /* Pattern options */ + int nltype; /* Newline type */ + int nllen; /* Newline string length */ + pcre_uchar nl[4]; /* Newline string when fixed */ + void *callout_data; /* To pass back to callouts */ + dfa_recursion_info *recursive; /* Linked list of recursion data */ +} dfa_match_data; + +/* Bit definitions for entries in the pcre_ctypes table. */ + +#define ctype_space 0x01 +#define ctype_letter 0x02 +#define ctype_digit 0x04 +#define ctype_xdigit 0x08 +#define ctype_word 0x10 /* alphanumeric or '_' */ +#define ctype_meta 0x80 /* regexp meta char or zero (end pattern) */ + +/* Offsets for the bitmap tables in pcre_cbits. Each table contains a set +of bits for a class map. Some classes are built by combining these tables. */ + +#define cbit_space 0 /* [:space:] or \s */ +#define cbit_xdigit 32 /* [:xdigit:] */ +#define cbit_digit 64 /* [:digit:] or \d */ +#define cbit_upper 96 /* [:upper:] */ +#define cbit_lower 128 /* [:lower:] */ +#define cbit_word 160 /* [:word:] or \w */ +#define cbit_graph 192 /* [:graph:] */ +#define cbit_print 224 /* [:print:] */ +#define cbit_punct 256 /* [:punct:] */ +#define cbit_cntrl 288 /* [:cntrl:] */ +#define cbit_length 320 /* Length of the cbits table */ + +/* Offsets of the various tables from the base tables pointer, and +total length. */ + +#define lcc_offset 0 +#define fcc_offset 256 +#define cbits_offset 512 +#define ctypes_offset (cbits_offset + cbit_length) +#define tables_length (ctypes_offset + 256) + +/* Internal function and data prefixes. */ + +#ifdef COMPILE_PCRE8 +#ifndef PUBL +#define PUBL(name) pcre_##name +#endif +#ifndef PRIV +#define PRIV(name) _pcre_##name +#endif +#else /* COMPILE_PCRE8 */ +#ifdef COMPILE_PCRE16 +#ifndef PUBL +#define PUBL(name) pcre16_##name +#endif +#ifndef PRIV +#define PRIV(name) _pcre16_##name +#endif +#else +#error Unsupported compiling mode +#endif /* COMPILE_PCRE16 */ +#endif /* COMPILE_PCRE8 */ + +/* Layout of the UCP type table that translates property names into types and +codes. Each entry used to point directly to a name, but to reduce the number of +relocations in shared libraries, it now has an offset into a single string +instead. */ + +typedef struct { + pcre_uint16 name_offset; + pcre_uint16 type; + pcre_uint16 value; +} ucp_type_table; + + +/* Internal shared data tables. These are tables that are used by more than one +of the exported public functions. They have to be "external" in the C sense, +but are not part of the PCRE public API. The data for these tables is in the +pcre_tables.c module. */ + +#ifdef COMPILE_PCRE8 + +extern const int PRIV(utf8_table1)[]; +extern const int PRIV(utf8_table1_size); +extern const int PRIV(utf8_table2)[]; +extern const int PRIV(utf8_table3)[]; +extern const pcre_uint8 PRIV(utf8_table4)[]; + +#endif /* COMPILE_PCRE8 */ + +extern const char PRIV(utt_names)[]; +extern const ucp_type_table PRIV(utt)[]; +extern const int PRIV(utt_size); + +extern const pcre_uint8 PRIV(default_tables)[]; + +extern const pcre_uint8 PRIV(OP_lengths)[]; + + +/* Internal shared functions. These are functions that are used by more than +one of the exported public functions. They have to be "external" in the C +sense, but are not part of the PCRE public API. */ + +/* String comparison functions. */ +#ifdef COMPILE_PCRE8 + +#define STRCMP_UC_UC(str1, str2) \ + strcmp((char *)(str1), (char *)(str2)) +#define STRCMP_UC_C8(str1, str2) \ + strcmp((char *)(str1), (str2)) +#define STRNCMP_UC_UC(str1, str2, num) \ + strncmp((char *)(str1), (char *)(str2), (num)) +#define STRNCMP_UC_C8(str1, str2, num) \ + strncmp((char *)(str1), (str2), (num)) +#define STRLEN_UC(str) strlen((const char *)str) + +#else + +extern int PRIV(strcmp_uc_uc)(const pcre_uchar *, + const pcre_uchar *); +extern int PRIV(strcmp_uc_c8)(const pcre_uchar *, + const char *); +extern int PRIV(strncmp_uc_uc)(const pcre_uchar *, + const pcre_uchar *, unsigned int num); +extern int PRIV(strncmp_uc_c8)(const pcre_uchar *, + const char *, unsigned int num); +extern unsigned int PRIV(strlen_uc)(const pcre_uchar *str); + +#define STRCMP_UC_UC(str1, str2) \ + PRIV(strcmp_uc_uc)((str1), (str2)) +#define STRCMP_UC_C8(str1, str2) \ + PRIV(strcmp_uc_c8)((str1), (str2)) +#define STRNCMP_UC_UC(str1, str2, num) \ + PRIV(strncmp_uc_uc)((str1), (str2), (num)) +#define STRNCMP_UC_C8(str1, str2, num) \ + PRIV(strncmp_uc_c8)((str1), (str2), (num)) +#define STRLEN_UC(str) PRIV(strlen_uc)(str) + +#endif /* COMPILE_PCRE8 */ + +extern const pcre_uchar *PRIV(find_bracket)(const pcre_uchar *, BOOL, int); +extern BOOL PRIV(is_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR, + int *, BOOL); +extern int PRIV(ord2utf)(pcre_uint32, pcre_uchar *); +extern int PRIV(valid_utf)(PCRE_PUCHAR, int, int *); +extern BOOL PRIV(was_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR, + int *, BOOL); +extern BOOL PRIV(xclass)(int, const pcre_uchar *, BOOL); + +#ifdef SUPPORT_JIT +extern void PRIV(jit_compile)(const REAL_PCRE *, + PUBL(extra) *, int); +extern int PRIV(jit_exec)(const REAL_PCRE *, const PUBL(extra) *, + const pcre_uchar *, int, int, int, int *, int); +extern void PRIV(jit_free)(void *); +extern int PRIV(jit_get_size)(void *); +extern const char* PRIV(jit_get_target)(void); +#endif + +/* Unicode character database (UCD) */ + +typedef struct { + pcre_uint8 script; + pcre_uint8 chartype; + pcre_int32 other_case; +} ucd_record; + +extern const ucd_record PRIV(ucd_records)[]; +extern const pcre_uint8 PRIV(ucd_stage1)[]; +extern const pcre_uint16 PRIV(ucd_stage2)[]; +extern const int PRIV(ucp_gentype)[]; +#ifdef SUPPORT_JIT +extern const int PRIV(ucp_typerange)[]; +#endif + +#ifdef SUPPORT_UCP +/* UCD access macros */ + +#define UCD_BLOCK_SIZE 128 +#define GET_UCD(ch) (PRIV(ucd_records) + \ + PRIV(ucd_stage2)[PRIV(ucd_stage1)[(ch) / UCD_BLOCK_SIZE] * \ + UCD_BLOCK_SIZE + (ch) % UCD_BLOCK_SIZE]) + +#define UCD_CHARTYPE(ch) GET_UCD(ch)->chartype +#define UCD_SCRIPT(ch) GET_UCD(ch)->script +#define UCD_CATEGORY(ch) PRIV(ucp_gentype)[UCD_CHARTYPE(ch)] +#define UCD_OTHERCASE(ch) (ch + GET_UCD(ch)->other_case) #endif /* SUPPORT_UCP */ diff --git a/pcre/pcre_jit_compile.c b/pcre/pcre_jit_compile.c --- a/pcre/pcre_jit_compile.c +++ b/pcre/pcre_jit_compile.c @@ -6,9859 +6,7447 @@ and semantics are as close as possible to those of the Perl 5 language. Written by Philip Hazel - Copyright (c) 1997-2013 University of Cambridge - - The machine code generator part (this module) was written by Zoltan Herczeg - Copyright (c) 2010-2013 - ------------------------------------------------------------------------------ -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. - - * Neither the name of the University of Cambridge 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 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. ------------------------------------------------------------------------------ -*/ - -#ifdef HAVE_CONFIG_H -#include "config.h" -#endif - -#include "pcre_internal.h" - -#if defined SUPPORT_JIT - -/* All-in-one: Since we use the JIT compiler only from here, -we just include it. This way we don't need to touch the build -system files. */ - -#define SLJIT_MALLOC(size) (PUBL(malloc))(size) -#define SLJIT_FREE(ptr) (PUBL(free))(ptr) -#define SLJIT_CONFIG_AUTO 1 -#define SLJIT_CONFIG_STATIC 1 -#define SLJIT_VERBOSE 0 -#define SLJIT_DEBUG 0 - -#include "sljit/sljitLir.c" - -#if defined SLJIT_CONFIG_UNSUPPORTED && SLJIT_CONFIG_UNSUPPORTED -#error Unsupported architecture -#endif - -/* Defines for debugging purposes. */ - -/* 1 - Use unoptimized capturing brackets. - 2 - Enable capture_last_ptr (includes option 1). */ -/* #define DEBUG_FORCE_UNOPTIMIZED_CBRAS 2 */ - -/* 1 - Always have a control head. */ -/* #define DEBUG_FORCE_CONTROL_HEAD 1 */ - -/* Allocate memory for the regex stack on the real machine stack. -Fast, but limited size. */ -#define MACHINE_STACK_SIZE 32768 - -/* Growth rate for stack allocated by the OS. Should be the multiply -of page size. */ -#define STACK_GROWTH_RATE 8192 - -/* Enable to check that the allocation could destroy temporaries. */ -#if defined SLJIT_DEBUG && SLJIT_DEBUG -#define DESTROY_REGISTERS 1 -#endif - -/* -Short summary about the backtracking mechanism empolyed by the jit code generator: - -The code generator follows the recursive nature of the PERL compatible regular -expressions. The basic blocks of regular expressions are condition checkers -whose execute different commands depending on the result of the condition check. -The relationship between the operators can be horizontal (concatenation) and -vertical (sub-expression) (See struct backtrack_common for more details). - - 'ab' - 'a' and 'b' regexps are concatenated - 'a+' - 'a' is the sub-expression of the '+' operator - -The condition checkers are boolean (true/false) checkers. Machine code is generated -for the checker itself and for the actions depending on the result of the checker. -The 'true' case is called as the matching path (expected path), and the other is called as -the 'backtrack' path. Branch instructions are expesive for all CPUs, so we avoid taken -branches on the matching path. - - Greedy star operator (*) : - Matching path: match happens. - Backtrack path: match failed. - Non-greedy star operator (*?) : - Matching path: no need to perform a match. - Backtrack path: match is required. - -The following example shows how the code generated for a capturing bracket -with two alternatives. Let A, B, C, D are arbirary regular expressions, and -we have the following regular expression: - - A(B|C)D - -The generated code will be the following: - - A matching path - '(' matching path (pushing arguments to the stack) - B matching path - ')' matching path (pushing arguments to the stack) - D matching path - return with successful match - - D backtrack path - ')' backtrack path (If we arrived from "C" jump to the backtrack of "C") - B backtrack path - C expected path - jump to D matching path - C backtrack path - A backtrack path - - Notice, that the order of backtrack code paths are the opposite of the fast - code paths. In this way the topmost value on the stack is always belong - to the current backtrack code path. The backtrack path must check - whether there is a next alternative. If so, it needs to jump back to - the matching path eventually. Otherwise it needs to clear out its own stack - frame and continue the execution on the backtrack code paths. -*/ - -/* -Saved stack frames: - -Atomic blocks and asserts require reloading the values of private data -when the backtrack mechanism performed. Because of OP_RECURSE, the data -are not necessarly known in compile time, thus we need a dynamic restore -mechanism. - -The stack frames are stored in a chain list, and have the following format: -([ capturing bracket offset ][ start value ][ end value ])+ ... [ 0 ] [ previous head ] - -Thus we can restore the private data to a particular point in the stack. -*/ - -typedef struct jit_arguments { - /* Pointers first. */ - struct sljit_stack *stack; - const pcre_uchar *str; - const pcre_uchar *begin; - const pcre_uchar *end; - int *offsets; - pcre_uchar *uchar_ptr; - pcre_uchar *mark_ptr; - void *callout_data; - /* Everything else after. */ - pcre_uint32 limit_match; - int real_offset_count; - int offset_count; - pcre_uint8 notbol; - pcre_uint8 noteol; - pcre_uint8 notempty; - pcre_uint8 notempty_atstart; -} jit_arguments; - -typedef struct executable_functions { - void *executable_funcs[JIT_NUMBER_OF_COMPILE_MODES]; - PUBL(jit_callback) callback; - void *userdata; - pcre_uint32 top_bracket; - pcre_uint32 limit_match; - sljit_uw executable_sizes[JIT_NUMBER_OF_COMPILE_MODES]; -} executable_functions; - -typedef struct jump_list { - struct sljit_jump *jump; - struct jump_list *next; -} jump_list; - -typedef struct stub_list { - struct sljit_jump *start; - struct sljit_label *quit; - struct stub_list *next; -} stub_list; - -enum frame_types { - no_frame = -1, - no_stack = -2 -}; - -enum control_types { - type_mark = 0, - type_then_trap = 1 -}; - -typedef int (SLJIT_CALL *jit_function)(jit_arguments *args); - -/* The following structure is the key data type for the recursive -code generator. It is allocated by compile_matchingpath, and contains -the arguments for compile_backtrackingpath. Must be the first member -of its descendants. */ -typedef struct backtrack_common { - /* Concatenation stack. */ - struct backtrack_common *prev; - jump_list *nextbacktracks; - /* Internal stack (for component operators). */ - struct backtrack_common *top; - jump_list *topbacktracks; - /* Opcode pointer. */ - pcre_uchar *cc; -} backtrack_common; - -typedef struct assert_backtrack { - backtrack_common common; - jump_list *condfailed; - /* Less than 0 if a frame is not needed. */ - int framesize; - /* Points to our private memory word on the stack. */ - int private_data_ptr; - /* For iterators. */ - struct sljit_label *matchingpath; -} assert_backtrack; - -typedef struct bracket_backtrack { - backtrack_common common; - /* Where to coninue if an alternative is successfully matched. */ - struct sljit_label *alternative_matchingpath; - /* For rmin and rmax iterators. */ - struct sljit_label *recursive_matchingpath; - /* For greedy ? operator. */ - struct sljit_label *zero_matchingpath; - /* Contains the branches of a failed condition. */ - union { - /* Both for OP_COND, OP_SCOND. */ - jump_list *condfailed; - assert_backtrack *assert; - /* For OP_ONCE. Less than 0 if not needed. */ - int framesize; - } u; - /* Points to our private memory word on the stack. */ - int private_data_ptr; -} bracket_backtrack; - -typedef struct bracketpos_backtrack { - backtrack_common common; - /* Points to our private memory word on the stack. */ - int private_data_ptr; - /* Reverting stack is needed. */ - int framesize; - /* Allocated stack size. */ - int stacksize; -} bracketpos_backtrack; - -typedef struct braminzero_backtrack { - backtrack_common common; - struct sljit_label *matchingpath; -} braminzero_backtrack; - -typedef struct iterator_backtrack { - backtrack_common common; - /* Next iteration. */ - struct sljit_label *matchingpath; -} iterator_backtrack; - -typedef struct recurse_entry { - struct recurse_entry *next; - /* Contains the function entry. */ - struct sljit_label *entry; - /* Collects the calls until the function is not created. */ - jump_list *calls; - /* Points to the starting opcode. */ - sljit_sw start; -} recurse_entry; - -typedef struct recurse_backtrack { - backtrack_common common; - BOOL inlined_pattern; -} recurse_backtrack; - -#define OP_THEN_TRAP OP_TABLE_LENGTH - -typedef struct then_trap_backtrack { - backtrack_common common; - /* If then_trap is not NULL, this structure contains the real - then_trap for the backtracking path. */ - struct then_trap_backtrack *then_trap; - /* Points to the starting opcode. */ - sljit_sw start; - /* Exit point for the then opcodes of this alternative. */ - jump_list *quit; - /* Frame size of the current alternative. */ - int framesize; -} then_trap_backtrack; - -#define MAX_RANGE_SIZE 6 - -typedef struct compiler_common { - /* The sljit ceneric compiler. */ - struct sljit_compiler *compiler; - /* First byte code. */ - pcre_uchar *start; - /* Maps private data offset to each opcode. */ - sljit_si *private_data_ptrs; - /* Tells whether the capturing bracket is optimized. */ - pcre_uint8 *optimized_cbracket; - /* Tells whether the starting offset is a target of then. */ - pcre_uint8 *then_offsets; - /* Current position where a THEN must jump. */ - then_trap_backtrack *then_trap; - /* Starting offset of private data for capturing brackets. */ - int cbra_ptr; - /* Output vector starting point. Must be divisible by 2. */ - int ovector_start; - /* Last known position of the requested byte. */ - int req_char_ptr; - /* Head of the last recursion. */ - int recursive_head_ptr; - /* First inspected character for partial matching. */ - int start_used_ptr; - /* Starting pointer for partial soft matches. */ - int hit_start; - /* End pointer of the first line. */ - int first_line_end; - /* Points to the marked string. */ - int mark_ptr; - /* Recursive control verb management chain. */ - int control_head_ptr; - /* Points to the last matched capture block index. */ - int capture_last_ptr; - /* Points to the starting position of the current match. */ - int start_ptr; - - /* Flipped and lower case tables. */ - const pcre_uint8 *fcc; - sljit_sw lcc; - /* Mode can be PCRE_STUDY_JIT_COMPILE and others. */ - int mode; - /* \K is found in the pattern. */ - BOOL has_set_som; - /* (*SKIP:arg) is found in the pattern. */ - BOOL has_skip_arg; - /* (*THEN) is found in the pattern. */ - BOOL has_then; - /* Needs to know the start position anytime. */ - BOOL needs_start_ptr; - /* Currently in recurse or negative assert. */ - BOOL local_exit; - /* Currently in a positive assert. */ - BOOL positive_assert; - /* Newline control. */ - int nltype; - int newline; - int bsr_nltype; - /* Dollar endonly. */ - int endonly; - /* Tables. */ - sljit_sw ctypes; - int digits[2 + MAX_RANGE_SIZE]; - /* Named capturing brackets. */ - pcre_uchar *name_table; - sljit_sw name_count; - sljit_sw name_entry_size; - - /* Labels and jump lists. */ - struct sljit_label *partialmatchlabel; - struct sljit_label *quit_label; - struct sljit_label *forced_quit_label; - struct sljit_label *accept_label; - stub_list *stubs; - recurse_entry *entries; - recurse_entry *currententry; - jump_list *partialmatch; - jump_list *quit; - jump_list *positive_assert_quit; - jump_list *forced_quit; - jump_list *accept; - jump_list *calllimit; - jump_list *stackalloc; - jump_list *revertframes; - jump_list *wordboundary; - jump_list *anynewline; - jump_list *hspace; - jump_list *vspace; - jump_list *casefulcmp; - jump_list *caselesscmp; - jump_list *reset_match; - BOOL jscript_compat; -#ifdef SUPPORT_UTF - BOOL utf; -#ifdef SUPPORT_UCP - BOOL use_ucp; -#endif -#ifndef COMPILE_PCRE32 - jump_list *utfreadchar; -#endif -#ifdef COMPILE_PCRE8 - jump_list *utfreadtype8; -#endif -#endif /* SUPPORT_UTF */ -#ifdef SUPPORT_UCP - jump_list *getucd; -#endif -} compiler_common; - -/* For byte_sequence_compare. */ - -typedef struct compare_context { - int length; - int sourcereg; -#if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED - int ucharptr; - union { - sljit_si asint; - sljit_uh asushort; -#if defined COMPILE_PCRE8 - sljit_ub asbyte; - sljit_ub asuchars[4]; -#elif defined COMPILE_PCRE16 - sljit_uh asuchars[2]; -#elif defined COMPILE_PCRE32 - sljit_ui asuchars[1]; -#endif - } c; - union { - sljit_si asint; - sljit_uh asushort; -#if defined COMPILE_PCRE8 - sljit_ub asbyte; - sljit_ub asuchars[4]; -#elif defined COMPILE_PCRE16 - sljit_uh asuchars[2]; -#elif defined COMPILE_PCRE32 - sljit_ui asuchars[1]; -#endif - } oc; -#endif -} compare_context; - -/* Undefine sljit macros. */ -#undef CMP - -/* Used for accessing the elements of the stack. */ -#define STACK(i) ((-(i) - 1) * (int)sizeof(sljit_sw)) - -#define TMP1 SLJIT_SCRATCH_REG1 -#define TMP2 SLJIT_SCRATCH_REG3 -#define TMP3 SLJIT_TEMPORARY_EREG2 -#define STR_PTR SLJIT_SAVED_REG1 -#define STR_END SLJIT_SAVED_REG2 -#define STACK_TOP SLJIT_SCRATCH_REG2 -#define STACK_LIMIT SLJIT_SAVED_REG3 -#define ARGUMENTS SLJIT_SAVED_EREG1 -#define COUNT_MATCH SLJIT_SAVED_EREG2 -#define RETURN_ADDR SLJIT_TEMPORARY_EREG1 - -/* Local space layout. */ -/* These two locals can be used by the current opcode. */ -#define LOCALS0 (0 * sizeof(sljit_sw)) -#define LOCALS1 (1 * sizeof(sljit_sw)) -/* Two local variables for possessive quantifiers (char1 cannot use them). */ -#define POSSESSIVE0 (2 * sizeof(sljit_sw)) -#define POSSESSIVE1 (3 * sizeof(sljit_sw)) -/* Max limit of recursions. */ -#define LIMIT_MATCH (4 * sizeof(sljit_sw)) -/* The output vector is stored on the stack, and contains pointers -to characters. The vector data is divided into two groups: the first -group contains the start / end character pointers, and the second is -the start pointers when the end of the capturing group has not yet reached. */ -#define OVECTOR_START (common->ovector_start) -#define OVECTOR(i) (OVECTOR_START + (i) * (sljit_sw)sizeof(sljit_sw)) -#define OVECTOR_PRIV(i) (common->cbra_ptr + (i) * (sljit_sw)sizeof(sljit_sw)) -#define PRIVATE_DATA(cc) (common->private_data_ptrs[(cc) - common->start]) - -#if defined COMPILE_PCRE8 -#define MOV_UCHAR SLJIT_MOV_UB -#define MOVU_UCHAR SLJIT_MOVU_UB -#elif defined COMPILE_PCRE16 -#define MOV_UCHAR SLJIT_MOV_UH -#define MOVU_UCHAR SLJIT_MOVU_UH -#elif defined COMPILE_PCRE32 -#define MOV_UCHAR SLJIT_MOV_UI -#define MOVU_UCHAR SLJIT_MOVU_UI -#else -#error Unsupported compiling mode -#endif - -/* Shortcuts. */ -#define DEFINE_COMPILER \ - struct sljit_compiler *compiler = common->compiler -#define OP1(op, dst, dstw, src, srcw) \ - sljit_emit_op1(compiler, (op), (dst), (dstw), (src), (srcw)) -#define OP2(op, dst, dstw, src1, src1w, src2, src2w) \ - sljit_emit_op2(compiler, (op), (dst), (dstw), (src1), (src1w), (src2), (src2w)) -#define LABEL() \ - sljit_emit_label(compiler) -#define JUMP(type) \ - sljit_emit_jump(compiler, (type)) -#define JUMPTO(type, label) \ - sljit_set_label(sljit_emit_jump(compiler, (type)), (label)) -#define JUMPHERE(jump) \ - sljit_set_label((jump), sljit_emit_label(compiler)) -#define SET_LABEL(jump, label) \ - sljit_set_label((jump), (label)) -#define CMP(type, src1, src1w, src2, src2w) \ - sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w)) -#define CMPTO(type, src1, src1w, src2, src2w, label) \ - sljit_set_label(sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w)), (label)) -#define OP_FLAGS(op, dst, dstw, src, srcw, type) \ - sljit_emit_op_flags(compiler, (op), (dst), (dstw), (src), (srcw), (type)) -#define GET_LOCAL_BASE(dst, dstw, offset) \ - sljit_get_local_base(compiler, (dst), (dstw), (offset)) - -static pcre_uchar* bracketend(pcre_uchar* cc) -{ -SLJIT_ASSERT((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND)); -do cc += GET(cc, 1); while (*cc == OP_ALT); -SLJIT_ASSERT(*cc >= OP_KET && *cc <= OP_KETRPOS); -cc += 1 + LINK_SIZE; -return cc; -} - -/* Functions whose might need modification for all new supported opcodes: - next_opcode - check_opcode_types - set_private_data_ptrs - get_framesize - init_frame - get_private_data_copy_length - copy_private_data - compile_matchingpath - compile_backtrackingpath -*/ - -static pcre_uchar *next_opcode(compiler_common *common, pcre_uchar *cc) -{ -SLJIT_UNUSED_ARG(common); -switch(*cc) - { - case OP_SOD: - case OP_SOM: - case OP_SET_SOM: - case OP_NOT_WORD_BOUNDARY: - case OP_WORD_BOUNDARY: - case OP_NOT_DIGIT: - case OP_DIGIT: - case OP_NOT_WHITESPACE: - case OP_WHITESPACE: - case OP_NOT_WORDCHAR: - case OP_WORDCHAR: - case OP_ANY: - case OP_ALLANY: - case OP_NOTPROP: - case OP_PROP: - case OP_ANYNL: - case OP_NOT_HSPACE: - case OP_HSPACE: - case OP_NOT_VSPACE: - case OP_VSPACE: - case OP_EXTUNI: - case OP_EODN: - case OP_EOD: - case OP_CIRC: - case OP_CIRCM: - case OP_DOLL: - case OP_DOLLM: - case OP_CRSTAR: - case OP_CRMINSTAR: - case OP_CRPLUS: - case OP_CRMINPLUS: - case OP_CRQUERY: - case OP_CRMINQUERY: - case OP_CRRANGE: - case OP_CRMINRANGE: - case OP_CRPOSSTAR: - case OP_CRPOSPLUS: - case OP_CRPOSQUERY: - case OP_CRPOSRANGE: - case OP_CLASS: - case OP_NCLASS: - case OP_REF: - case OP_REFI: - case OP_DNREF: - case OP_DNREFI: - case OP_RECURSE: - case OP_CALLOUT: - case OP_ALT: - case OP_KET: - case OP_KETRMAX: - case OP_KETRMIN: - case OP_KETRPOS: - case OP_REVERSE: - case OP_ASSERT: - case OP_ASSERT_NOT: - case OP_ASSERTBACK: - case OP_ASSERTBACK_NOT: - case OP_ONCE: - case OP_ONCE_NC: - case OP_BRA: - case OP_BRAPOS: - case OP_CBRA: - case OP_CBRAPOS: - case OP_COND: - case OP_SBRA: - case OP_SBRAPOS: - case OP_SCBRA: - case OP_SCBRAPOS: - case OP_SCOND: - case OP_CREF: - case OP_DNCREF: - case OP_RREF: - case OP_DNRREF: - case OP_DEF: - case OP_BRAZERO: - case OP_BRAMINZERO: - case OP_BRAPOSZERO: - case OP_PRUNE: - case OP_SKIP: - case OP_THEN: - case OP_COMMIT: - case OP_FAIL: - case OP_ACCEPT: - case OP_ASSERT_ACCEPT: - case OP_CLOSE: - case OP_SKIPZERO: - return cc + PRIV(OP_lengths)[*cc]; - - case OP_CHAR: - case OP_CHARI: - case OP_NOT: - case OP_NOTI: - case OP_STAR: - case OP_MINSTAR: - case OP_PLUS: - case OP_MINPLUS: - case OP_QUERY: - case OP_MINQUERY: - case OP_UPTO: - case OP_MINUPTO: - case OP_EXACT: - case OP_POSSTAR: - case OP_POSPLUS: - case OP_POSQUERY: - case OP_POSUPTO: - case OP_STARI: - case OP_MINSTARI: - case OP_PLUSI: - case OP_MINPLUSI: - case OP_QUERYI: - case OP_MINQUERYI: - case OP_UPTOI: - case OP_MINUPTOI: - case OP_EXACTI: - case OP_POSSTARI: - case OP_POSPLUSI: - case OP_POSQUERYI: - case OP_POSUPTOI: - case OP_NOTSTAR: - case OP_NOTMINSTAR: - case OP_NOTPLUS: - case OP_NOTMINPLUS: - case OP_NOTQUERY: - case OP_NOTMINQUERY: - case OP_NOTUPTO: - case OP_NOTMINUPTO: - case OP_NOTEXACT: - case OP_NOTPOSSTAR: - case OP_NOTPOSPLUS: - case OP_NOTPOSQUERY: - case OP_NOTPOSUPTO: - case OP_NOTSTARI: - case OP_NOTMINSTARI: - case OP_NOTPLUSI: - case OP_NOTMINPLUSI: - case OP_NOTQUERYI: - case OP_NOTMINQUERYI: - case OP_NOTUPTOI: - case OP_NOTMINUPTOI: - case OP_NOTEXACTI: - case OP_NOTPOSSTARI: - case OP_NOTPOSPLUSI: - case OP_NOTPOSQUERYI: - case OP_NOTPOSUPTOI: - cc += PRIV(OP_lengths)[*cc]; -#ifdef SUPPORT_UTF - if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); -#endif - return cc; - - /* Special cases. */ - case OP_TYPESTAR: - case OP_TYPEMINSTAR: - case OP_TYPEPLUS: - case OP_TYPEMINPLUS: - case OP_TYPEQUERY: - case OP_TYPEMINQUERY: - case OP_TYPEUPTO: - case OP_TYPEMINUPTO: - case OP_TYPEEXACT: - case OP_TYPEPOSSTAR: - case OP_TYPEPOSPLUS: - case OP_TYPEPOSQUERY: - case OP_TYPEPOSUPTO: - return cc + PRIV(OP_lengths)[*cc] - 1; - - case OP_ANYBYTE: -#ifdef SUPPORT_UTF - if (common->utf) return NULL; -#endif - return cc + 1; - -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - case OP_XCLASS: - return cc + GET(cc, 1); -#endif - - case OP_MARK: - case OP_PRUNE_ARG: - case OP_SKIP_ARG: - case OP_THEN_ARG: - return cc + 1 + 2 + cc[1]; - - default: - /* All opcodes are supported now! */ - SLJIT_ASSERT_STOP(); - return NULL; - } -} - -static BOOL check_opcode_types(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) -{ -int count; -pcre_uchar *slot; - -/* Calculate important variables (like stack size) and checks whether all opcodes are supported. */ -while (cc < ccend) - { - switch(*cc) - { - case OP_SET_SOM: - common->has_set_som = TRUE; - cc += 1; - break; - - case OP_REF: - case OP_REFI: - common->optimized_cbracket[GET2(cc, 1)] = 0; - cc += 1 + IMM2_SIZE; - break; - - case OP_CBRAPOS: - case OP_SCBRAPOS: - common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] = 0; - cc += 1 + LINK_SIZE + IMM2_SIZE; - break; - - case OP_COND: - case OP_SCOND: - /* Only AUTO_CALLOUT can insert this opcode. We do - not intend to support this case. */ - if (cc[1 + LINK_SIZE] == OP_CALLOUT) - return FALSE; - cc += 1 + LINK_SIZE; - break; - - case OP_CREF: - common->optimized_cbracket[GET2(cc, 1)] = 0; - cc += 1 + IMM2_SIZE; - break; - - case OP_DNREF: - case OP_DNREFI: - case OP_DNCREF: - count = GET2(cc, 1 + IMM2_SIZE); - slot = common->name_table + GET2(cc, 1) * common->name_entry_size; - while (count-- > 0) - { - common->optimized_cbracket[GET2(slot, 0)] = 0; - slot += common->name_entry_size; - } - cc += 1 + 2 * IMM2_SIZE; - break; - - case OP_RECURSE: - /* Set its value only once. */ - if (common->recursive_head_ptr == 0) - { - common->recursive_head_ptr = common->ovector_start; - common->ovector_start += sizeof(sljit_sw); - } - cc += 1 + LINK_SIZE; - break; - - case OP_CALLOUT: - if (common->capture_last_ptr == 0) - { - common->capture_last_ptr = common->ovector_start; - common->ovector_start += sizeof(sljit_sw); - } - cc += 2 + 2 * LINK_SIZE; - break; - - case OP_THEN_ARG: - common->has_then = TRUE; - common->control_head_ptr = 1; - /* Fall through. */ - - case OP_PRUNE_ARG: - common->needs_start_ptr = TRUE; - /* Fall through. */ - - case OP_MARK: - if (common->mark_ptr == 0) - { - common->mark_ptr = common->ovector_start; - common->ovector_start += sizeof(sljit_sw); - } - cc += 1 + 2 + cc[1]; - break; - - case OP_THEN: - common->has_then = TRUE; - common->control_head_ptr = 1; - /* Fall through. */ - - case OP_PRUNE: - case OP_SKIP: - common->needs_start_ptr = TRUE; - cc += 1; - break; - - case OP_SKIP_ARG: - common->control_head_ptr = 1; - common->has_skip_arg = TRUE; - cc += 1 + 2 + cc[1]; - break; - - default: - cc = next_opcode(common, cc); - if (cc == NULL) - return FALSE; - break; - } - } -return TRUE; -} - -static int get_class_iterator_size(pcre_uchar *cc) -{ -switch(*cc) - { - case OP_CRSTAR: - case OP_CRPLUS: - return 2; - - case OP_CRMINSTAR: - case OP_CRMINPLUS: - case OP_CRQUERY: - case OP_CRMINQUERY: - return 1; - - case OP_CRRANGE: - case OP_CRMINRANGE: - if (GET2(cc, 1) == GET2(cc, 1 + IMM2_SIZE)) - return 0; - return 2; - - default: - return 0; - } -} - -static BOOL detect_repeat(compiler_common *common, pcre_uchar *begin) -{ -pcre_uchar *end = bracketend(begin); -pcre_uchar *next; -pcre_uchar *next_end; -pcre_uchar *max_end; -pcre_uchar type; -sljit_sw length = end - begin; -int min, max, i; - -/* Detect fixed iterations first. */ -if (end[-(1 + LINK_SIZE)] != OP_KET) - return FALSE; - -/* Already detected repeat. */ -if (common->private_data_ptrs[end - common->start - LINK_SIZE] != 0) - return TRUE; - -next = end; -min = 1; -while (1) - { - if (*next != *begin) - break; - next_end = bracketend(next); - if (next_end - next != length || memcmp(begin, next, IN_UCHARS(length)) != 0) - break; - next = next_end; - min++; - } - -if (min == 2) - return FALSE; - -max = 0; -max_end = next; -if (*next == OP_BRAZERO || *next == OP_BRAMINZERO) - { - type = *next; - while (1) - { - if (next[0] != type || next[1] != OP_BRA || next[2 + LINK_SIZE] != *begin) - break; - next_end = bracketend(next + 2 + LINK_SIZE); - if (next_end - next != (length + 2 + LINK_SIZE) || memcmp(begin, next + 2 + LINK_SIZE, IN_UCHARS(length)) != 0) - break; - next = next_end; - max++; - } - - if (next[0] == type && next[1] == *begin && max >= 1) - { - next_end = bracketend(next + 1); - if (next_end - next == (length + 1) && memcmp(begin, next + 1, IN_UCHARS(length)) == 0) - { - for (i = 0; i < max; i++, next_end += 1 + LINK_SIZE) - if (*next_end != OP_KET) - break; - - if (i == max) - { - common->private_data_ptrs[max_end - common->start - LINK_SIZE] = next_end - max_end; - common->private_data_ptrs[max_end - common->start - LINK_SIZE + 1] = (type == OP_BRAZERO) ? OP_UPTO : OP_MINUPTO; - /* +2 the original and the last. */ - common->private_data_ptrs[max_end - common->start - LINK_SIZE + 2] = max + 2; - if (min == 1) - return TRUE; - min--; - max_end -= (1 + LINK_SIZE) + GET(max_end, -LINK_SIZE); - } - } - } - } - -if (min >= 3) - { - common->private_data_ptrs[end - common->start - LINK_SIZE] = max_end - end; - common->private_data_ptrs[end - common->start - LINK_SIZE + 1] = OP_EXACT; - common->private_data_ptrs[end - common->start - LINK_SIZE + 2] = min; - return TRUE; - } - -return FALSE; -} - -#define CASE_ITERATOR_PRIVATE_DATA_1 \ - case OP_MINSTAR: \ - case OP_MINPLUS: \ - case OP_QUERY: \ - case OP_MINQUERY: \ - case OP_MINSTARI: \ - case OP_MINPLUSI: \ - case OP_QUERYI: \ - case OP_MINQUERYI: \ - case OP_NOTMINSTAR: \ - case OP_NOTMINPLUS: \ - case OP_NOTQUERY: \ - case OP_NOTMINQUERY: \ - case OP_NOTMINSTARI: \ - case OP_NOTMINPLUSI: \ - case OP_NOTQUERYI: \ - case OP_NOTMINQUERYI: - -#define CASE_ITERATOR_PRIVATE_DATA_2A \ - case OP_STAR: \ - case OP_PLUS: \ - case OP_STARI: \ - case OP_PLUSI: \ - case OP_NOTSTAR: \ - case OP_NOTPLUS: \ - case OP_NOTSTARI: \ - case OP_NOTPLUSI: - -#define CASE_ITERATOR_PRIVATE_DATA_2B \ - case OP_UPTO: \ - case OP_MINUPTO: \ - case OP_UPTOI: \ - case OP_MINUPTOI: \ - case OP_NOTUPTO: \ - case OP_NOTMINUPTO: \ - case OP_NOTUPTOI: \ - case OP_NOTMINUPTOI: - -#define CASE_ITERATOR_TYPE_PRIVATE_DATA_1 \ - case OP_TYPEMINSTAR: \ - case OP_TYPEMINPLUS: \ - case OP_TYPEQUERY: \ - case OP_TYPEMINQUERY: - -#define CASE_ITERATOR_TYPE_PRIVATE_DATA_2A \ - case OP_TYPESTAR: \ - case OP_TYPEPLUS: - -#define CASE_ITERATOR_TYPE_PRIVATE_DATA_2B \ - case OP_TYPEUPTO: \ - case OP_TYPEMINUPTO: - -static void set_private_data_ptrs(compiler_common *common, int *private_data_start, pcre_uchar *ccend) -{ -pcre_uchar *cc = common->start; -pcre_uchar *alternative; -pcre_uchar *end = NULL; -int private_data_ptr = *private_data_start; -int space, size, bracketlen; - -while (cc < ccend) - { - space = 0; - size = 0; - bracketlen = 0; - if (private_data_ptr > SLJIT_MAX_LOCAL_SIZE) - return; - - if (*cc == OP_ONCE || *cc == OP_ONCE_NC || *cc == OP_BRA || *cc == OP_CBRA || *cc == OP_COND) - if (detect_repeat(common, cc)) - { - /* These brackets are converted to repeats, so no global - based single character repeat is allowed. */ - if (cc >= end) - end = bracketend(cc); - } - - switch(*cc) - { - case OP_KET: - if (common->private_data_ptrs[cc + 1 - common->start] != 0) - { - common->private_data_ptrs[cc - common->start] = private_data_ptr; - private_data_ptr += sizeof(sljit_sw); - cc += common->private_data_ptrs[cc + 1 - common->start]; - } - cc += 1 + LINK_SIZE; - break; - - case OP_ASSERT: - case OP_ASSERT_NOT: - case OP_ASSERTBACK: - case OP_ASSERTBACK_NOT: - case OP_ONCE: - case OP_ONCE_NC: - case OP_BRAPOS: - case OP_SBRA: - case OP_SBRAPOS: - case OP_SCOND: - common->private_data_ptrs[cc - common->start] = private_data_ptr; - private_data_ptr += sizeof(sljit_sw); - bracketlen = 1 + LINK_SIZE; - break; - - case OP_CBRAPOS: - case OP_SCBRAPOS: - common->private_data_ptrs[cc - common->start] = private_data_ptr; - private_data_ptr += sizeof(sljit_sw); - bracketlen = 1 + LINK_SIZE + IMM2_SIZE; - break; - - case OP_COND: - /* Might be a hidden SCOND. */ - alternative = cc + GET(cc, 1); - if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) - { - common->private_data_ptrs[cc - common->start] = private_data_ptr; - private_data_ptr += sizeof(sljit_sw); - } - bracketlen = 1 + LINK_SIZE; - break; - - case OP_BRA: - bracketlen = 1 + LINK_SIZE; - break; - - case OP_CBRA: - case OP_SCBRA: - bracketlen = 1 + LINK_SIZE + IMM2_SIZE; - break; - - CASE_ITERATOR_PRIVATE_DATA_1 - space = 1; - size = -2; - break; - - CASE_ITERATOR_PRIVATE_DATA_2A - space = 2; - size = -2; - break; - - CASE_ITERATOR_PRIVATE_DATA_2B - space = 2; - size = -(2 + IMM2_SIZE); - break; - - CASE_ITERATOR_TYPE_PRIVATE_DATA_1 - space = 1; - size = 1; - break; - - CASE_ITERATOR_TYPE_PRIVATE_DATA_2A - if (cc[1] != OP_ANYNL && cc[1] != OP_EXTUNI) - space = 2; - size = 1; - break; - - CASE_ITERATOR_TYPE_PRIVATE_DATA_2B - if (cc[1 + IMM2_SIZE] != OP_ANYNL && cc[1 + IMM2_SIZE] != OP_EXTUNI) - space = 2; - size = 1 + IMM2_SIZE; - break; - - case OP_CLASS: - case OP_NCLASS: - size += 1 + 32 / sizeof(pcre_uchar); - space = get_class_iterator_size(cc + size); - break; - -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - case OP_XCLASS: - size = GET(cc, 1); - space = get_class_iterator_size(cc + size); - break; -#endif - - default: - cc = next_opcode(common, cc); - SLJIT_ASSERT(cc != NULL); - break; - } - - /* Character iterators, which are not inside a repeated bracket, - gets a private slot instead of allocating it on the stack. */ - if (space > 0 && cc >= end) - { - common->private_data_ptrs[cc - common->start] = private_data_ptr; - private_data_ptr += sizeof(sljit_sw) * space; - } - - if (size != 0) - { - if (size < 0) - { - cc += -size; -#ifdef SUPPORT_UTF - if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); -#endif - } - else - cc += size; - } - - if (bracketlen > 0) - { - if (cc >= end) - { - end = bracketend(cc); - if (end[-1 - LINK_SIZE] == OP_KET) - end = NULL; - } - cc += bracketlen; - } - } -*private_data_start = private_data_ptr; -} - -/* Returns with a frame_types (always < 0) if no need for frame. */ -static int get_framesize(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, BOOL recursive, BOOL* needs_control_head) -{ -int length = 0; -int possessive = 0; -BOOL stack_restore = FALSE; -BOOL setsom_found = recursive; -BOOL setmark_found = recursive; -/* The last capture is a local variable even for recursions. */ -BOOL capture_last_found = FALSE; - -#if defined DEBUG_FORCE_CONTROL_HEAD && DEBUG_FORCE_CONTROL_HEAD -SLJIT_ASSERT(common->control_head_ptr != 0); -*needs_control_head = TRUE; -#else -*needs_control_head = FALSE; -#endif - -if (ccend == NULL) - { - ccend = bracketend(cc) - (1 + LINK_SIZE); - if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS)) - { - possessive = length = (common->capture_last_ptr != 0) ? 5 : 3; - /* This is correct regardless of common->capture_last_ptr. */ - capture_last_found = TRUE; - } - cc = next_opcode(common, cc); - } - -SLJIT_ASSERT(cc != NULL); -while (cc < ccend) - switch(*cc) - { - case OP_SET_SOM: - SLJIT_ASSERT(common->has_set_som); - stack_restore = TRUE; - if (!setsom_found) - { - length += 2; - setsom_found = TRUE; - } - cc += 1; - break; - - case OP_MARK: - case OP_PRUNE_ARG: - case OP_THEN_ARG: - SLJIT_ASSERT(common->mark_ptr != 0); - stack_restore = TRUE; - if (!setmark_found) - { - length += 2; - setmark_found = TRUE; - } - if (common->control_head_ptr != 0) - *needs_control_head = TRUE; - cc += 1 + 2 + cc[1]; - break; - - case OP_RECURSE: - stack_restore = TRUE; - if (common->has_set_som && !setsom_found) - { - length += 2; - setsom_found = TRUE; - } - if (common->mark_ptr != 0 && !setmark_found) - { - length += 2; - setmark_found = TRUE; - } - if (common->capture_last_ptr != 0 && !capture_last_found) - { - length += 2; - capture_last_found = TRUE; - } - cc += 1 + LINK_SIZE; - break; - - case OP_CBRA: - case OP_CBRAPOS: - case OP_SCBRA: - case OP_SCBRAPOS: - stack_restore = TRUE; - if (common->capture_last_ptr != 0 && !capture_last_found) - { - length += 2; - capture_last_found = TRUE; - } - length += 3; - cc += 1 + LINK_SIZE + IMM2_SIZE; - break; - - default: - stack_restore = TRUE; - /* Fall through. */ - - case OP_NOT_WORD_BOUNDARY: - case OP_WORD_BOUNDARY: - case OP_NOT_DIGIT: - case OP_DIGIT: - case OP_NOT_WHITESPACE: - case OP_WHITESPACE: - case OP_NOT_WORDCHAR: - case OP_WORDCHAR: - case OP_ANY: - case OP_ALLANY: - case OP_ANYBYTE: - case OP_NOTPROP: - case OP_PROP: - case OP_ANYNL: - case OP_NOT_HSPACE: - case OP_HSPACE: - case OP_NOT_VSPACE: - case OP_VSPACE: - case OP_EXTUNI: - case OP_EODN: - case OP_EOD: - case OP_CIRC: - case OP_CIRCM: - case OP_DOLL: - case OP_DOLLM: - case OP_CHAR: - case OP_CHARI: - case OP_NOT: - case OP_NOTI: - - case OP_EXACT: - case OP_POSSTAR: - case OP_POSPLUS: - case OP_POSQUERY: - case OP_POSUPTO: - - case OP_EXACTI: - case OP_POSSTARI: - case OP_POSPLUSI: - case OP_POSQUERYI: - case OP_POSUPTOI: - - case OP_NOTEXACT: - case OP_NOTPOSSTAR: - case OP_NOTPOSPLUS: - case OP_NOTPOSQUERY: - case OP_NOTPOSUPTO: - - case OP_NOTEXACTI: - case OP_NOTPOSSTARI: - case OP_NOTPOSPLUSI: - case OP_NOTPOSQUERYI: - case OP_NOTPOSUPTOI: - - case OP_TYPEEXACT: - case OP_TYPEPOSSTAR: - case OP_TYPEPOSPLUS: - case OP_TYPEPOSQUERY: - case OP_TYPEPOSUPTO: - - case OP_CLASS: - case OP_NCLASS: - case OP_XCLASS: - - cc = next_opcode(common, cc); - SLJIT_ASSERT(cc != NULL); - break; - } - -/* Possessive quantifiers can use a special case. */ -if (SLJIT_UNLIKELY(possessive == length)) - return stack_restore ? no_frame : no_stack; - -if (length > 0) - return length + 1; -return stack_restore ? no_frame : no_stack; -} - -static void init_frame(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, int stackpos, int stacktop, BOOL recursive) -{ -DEFINE_COMPILER; -BOOL setsom_found = recursive; -BOOL setmark_found = recursive; -/* The last capture is a local variable even for recursions. */ -BOOL capture_last_found = FALSE; -int offset; - -/* >= 1 + shortest item size (2) */ -SLJIT_UNUSED_ARG(stacktop); -SLJIT_ASSERT(stackpos >= stacktop + 2); - -stackpos = STACK(stackpos); -if (ccend == NULL) - { - ccend = bracketend(cc) - (1 + LINK_SIZE); - if (recursive || (*cc != OP_CBRAPOS && *cc != OP_SCBRAPOS)) - cc = next_opcode(common, cc); - } - -SLJIT_ASSERT(cc != NULL); -while (cc < ccend) - switch(*cc) - { - case OP_SET_SOM: - SLJIT_ASSERT(common->has_set_som); - if (!setsom_found) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -OVECTOR(0)); - stackpos += (int)sizeof(sljit_sw); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); - stackpos += (int)sizeof(sljit_sw); - setsom_found = TRUE; - } - cc += 1; - break; - - case OP_MARK: - case OP_PRUNE_ARG: - case OP_THEN_ARG: - SLJIT_ASSERT(common->mark_ptr != 0); - if (!setmark_found) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->mark_ptr); - stackpos += (int)sizeof(sljit_sw); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); - stackpos += (int)sizeof(sljit_sw); - setmark_found = TRUE; - } - cc += 1 + 2 + cc[1]; - break; - - case OP_RECURSE: - if (common->has_set_som && !setsom_found) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -OVECTOR(0)); - stackpos += (int)sizeof(sljit_sw); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); - stackpos += (int)sizeof(sljit_sw); - setsom_found = TRUE; - } - if (common->mark_ptr != 0 && !setmark_found) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->mark_ptr); - stackpos += (int)sizeof(sljit_sw); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); - stackpos += (int)sizeof(sljit_sw); - setmark_found = TRUE; - } - if (common->capture_last_ptr != 0 && !capture_last_found) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->capture_last_ptr); - stackpos += (int)sizeof(sljit_sw); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); - stackpos += (int)sizeof(sljit_sw); - capture_last_found = TRUE; - } - cc += 1 + LINK_SIZE; - break; - - case OP_CBRA: - case OP_CBRAPOS: - case OP_SCBRA: - case OP_SCBRAPOS: - if (common->capture_last_ptr != 0 && !capture_last_found) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->capture_last_ptr); - stackpos += (int)sizeof(sljit_sw); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); - stackpos += (int)sizeof(sljit_sw); - capture_last_found = TRUE; - } - offset = (GET2(cc, 1 + LINK_SIZE)) << 1; - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, OVECTOR(offset)); - stackpos += (int)sizeof(sljit_sw); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); - stackpos += (int)sizeof(sljit_sw); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP2, 0); - stackpos += (int)sizeof(sljit_sw); - - cc += 1 + LINK_SIZE + IMM2_SIZE; - break; - - default: - cc = next_opcode(common, cc); - SLJIT_ASSERT(cc != NULL); - break; - } - -OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, 0); -SLJIT_ASSERT(stackpos == STACK(stacktop)); -} - -static SLJIT_INLINE int get_private_data_copy_length(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, BOOL needs_control_head) -{ -int private_data_length = needs_control_head ? 3 : 2; -int size; -pcre_uchar *alternative; -/* Calculate the sum of the private machine words. */ -while (cc < ccend) - { - size = 0; - switch(*cc) - { - case OP_KET: - if (PRIVATE_DATA(cc) != 0) - private_data_length++; - cc += 1 + LINK_SIZE; - break; - - case OP_ASSERT: - case OP_ASSERT_NOT: - case OP_ASSERTBACK: - case OP_ASSERTBACK_NOT: - case OP_ONCE: - case OP_ONCE_NC: - case OP_BRAPOS: - case OP_SBRA: - case OP_SBRAPOS: - case OP_SCOND: - private_data_length++; - cc += 1 + LINK_SIZE; - break; - - case OP_CBRA: - case OP_SCBRA: - if (common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] == 0) - private_data_length++; - cc += 1 + LINK_SIZE + IMM2_SIZE; - break; - - case OP_CBRAPOS: - case OP_SCBRAPOS: - private_data_length += 2; - cc += 1 + LINK_SIZE + IMM2_SIZE; - break; - - case OP_COND: - /* Might be a hidden SCOND. */ - alternative = cc + GET(cc, 1); - if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) - private_data_length++; - cc += 1 + LINK_SIZE; - break; - - CASE_ITERATOR_PRIVATE_DATA_1 - if (PRIVATE_DATA(cc)) - private_data_length++; - cc += 2; -#ifdef SUPPORT_UTF - if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); -#endif - break; - - CASE_ITERATOR_PRIVATE_DATA_2A - if (PRIVATE_DATA(cc)) - private_data_length += 2; - cc += 2; -#ifdef SUPPORT_UTF - if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); -#endif - break; - - CASE_ITERATOR_PRIVATE_DATA_2B - if (PRIVATE_DATA(cc)) - private_data_length += 2; - cc += 2 + IMM2_SIZE; -#ifdef SUPPORT_UTF - if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); -#endif - break; - - CASE_ITERATOR_TYPE_PRIVATE_DATA_1 - if (PRIVATE_DATA(cc)) - private_data_length++; - cc += 1; - break; - - CASE_ITERATOR_TYPE_PRIVATE_DATA_2A - if (PRIVATE_DATA(cc)) - private_data_length += 2; - cc += 1; - break; - - CASE_ITERATOR_TYPE_PRIVATE_DATA_2B - if (PRIVATE_DATA(cc)) - private_data_length += 2; - cc += 1 + IMM2_SIZE; - break; - - case OP_CLASS: - case OP_NCLASS: -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - case OP_XCLASS: - size = (*cc == OP_XCLASS) ? GET(cc, 1) : 1 + 32 / (int)sizeof(pcre_uchar); -#else - size = 1 + 32 / (int)sizeof(pcre_uchar); -#endif - if (PRIVATE_DATA(cc)) - private_data_length += get_class_iterator_size(cc + size); - cc += size; - break; - - default: - cc = next_opcode(common, cc); - SLJIT_ASSERT(cc != NULL); - break; - } - } -SLJIT_ASSERT(cc == ccend); -return private_data_length; -} - -static void copy_private_data(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, - BOOL save, int stackptr, int stacktop, BOOL needs_control_head) -{ -DEFINE_COMPILER; -int srcw[2]; -int count, size; -BOOL tmp1next = TRUE; -BOOL tmp1empty = TRUE; -BOOL tmp2empty = TRUE; -pcre_uchar *alternative; -enum { - start, - loop, - end -} status; - -status = save ? start : loop; -stackptr = STACK(stackptr - 2); -stacktop = STACK(stacktop - 1); - -if (!save) - { - stackptr += (needs_control_head ? 2 : 1) * sizeof(sljit_sw); - if (stackptr < stacktop) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr); - stackptr += sizeof(sljit_sw); - tmp1empty = FALSE; - } - if (stackptr < stacktop) - { - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), stackptr); - stackptr += sizeof(sljit_sw); - tmp2empty = FALSE; - } - /* The tmp1next must be TRUE in either way. */ - } - -do - { - count = 0; - switch(status) - { - case start: - SLJIT_ASSERT(save && common->recursive_head_ptr != 0); - count = 1; - srcw[0] = common->recursive_head_ptr; - if (needs_control_head) - { - SLJIT_ASSERT(common->control_head_ptr != 0); - count = 2; - srcw[1] = common->control_head_ptr; - } - status = loop; - break; - - case loop: - if (cc >= ccend) - { - status = end; - break; - } - - switch(*cc) - { - case OP_KET: - if (PRIVATE_DATA(cc) != 0) - { - count = 1; - srcw[0] = PRIVATE_DATA(cc); - } - cc += 1 + LINK_SIZE; - break; - - case OP_ASSERT: - case OP_ASSERT_NOT: - case OP_ASSERTBACK: - case OP_ASSERTBACK_NOT: - case OP_ONCE: - case OP_ONCE_NC: - case OP_BRAPOS: - case OP_SBRA: - case OP_SBRAPOS: - case OP_SCOND: - count = 1; - srcw[0] = PRIVATE_DATA(cc); - SLJIT_ASSERT(srcw[0] != 0); - cc += 1 + LINK_SIZE; - break; - - case OP_CBRA: - case OP_SCBRA: - if (common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] == 0) - { - count = 1; - srcw[0] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); - } - cc += 1 + LINK_SIZE + IMM2_SIZE; - break; - - case OP_CBRAPOS: - case OP_SCBRAPOS: - count = 2; - srcw[0] = PRIVATE_DATA(cc); - srcw[1] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); - SLJIT_ASSERT(srcw[0] != 0 && srcw[1] != 0); - cc += 1 + LINK_SIZE + IMM2_SIZE; - break; - - case OP_COND: - /* Might be a hidden SCOND. */ - alternative = cc + GET(cc, 1); - if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) - { - count = 1; - srcw[0] = PRIVATE_DATA(cc); - SLJIT_ASSERT(srcw[0] != 0); - } - cc += 1 + LINK_SIZE; - break; - - CASE_ITERATOR_PRIVATE_DATA_1 - if (PRIVATE_DATA(cc)) - { - count = 1; - srcw[0] = PRIVATE_DATA(cc); - } - cc += 2; -#ifdef SUPPORT_UTF - if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); -#endif - break; - - CASE_ITERATOR_PRIVATE_DATA_2A - if (PRIVATE_DATA(cc)) - { - count = 2; - srcw[0] = PRIVATE_DATA(cc); - srcw[1] = PRIVATE_DATA(cc) + sizeof(sljit_sw); - } - cc += 2; -#ifdef SUPPORT_UTF - if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); -#endif - break; - - CASE_ITERATOR_PRIVATE_DATA_2B - if (PRIVATE_DATA(cc)) - { - count = 2; - srcw[0] = PRIVATE_DATA(cc); - srcw[1] = PRIVATE_DATA(cc) + sizeof(sljit_sw); - } - cc += 2 + IMM2_SIZE; -#ifdef SUPPORT_UTF - if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); -#endif - break; - - CASE_ITERATOR_TYPE_PRIVATE_DATA_1 - if (PRIVATE_DATA(cc)) - { - count = 1; - srcw[0] = PRIVATE_DATA(cc); - } - cc += 1; - break; - - CASE_ITERATOR_TYPE_PRIVATE_DATA_2A - if (PRIVATE_DATA(cc)) - { - count = 2; - srcw[0] = PRIVATE_DATA(cc); - srcw[1] = srcw[0] + sizeof(sljit_sw); - } - cc += 1; - break; - - CASE_ITERATOR_TYPE_PRIVATE_DATA_2B - if (PRIVATE_DATA(cc)) - { - count = 2; - srcw[0] = PRIVATE_DATA(cc); - srcw[1] = srcw[0] + sizeof(sljit_sw); - } - cc += 1 + IMM2_SIZE; - break; - - case OP_CLASS: - case OP_NCLASS: -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - case OP_XCLASS: - size = (*cc == OP_XCLASS) ? GET(cc, 1) : 1 + 32 / (int)sizeof(pcre_uchar); -#else - size = 1 + 32 / (int)sizeof(pcre_uchar); -#endif - if (PRIVATE_DATA(cc)) - switch(get_class_iterator_size(cc + size)) - { - case 1: - count = 1; - srcw[0] = PRIVATE_DATA(cc); - break; - - case 2: - count = 2; - srcw[0] = PRIVATE_DATA(cc); - srcw[1] = srcw[0] + sizeof(sljit_sw); - break; - - default: - SLJIT_ASSERT_STOP(); - break; - } - cc += size; - break; - - default: - cc = next_opcode(common, cc); - SLJIT_ASSERT(cc != NULL); - break; - } - break; - - case end: - SLJIT_ASSERT_STOP(); - break; - } - - while (count > 0) - { - count--; - if (save) - { - if (tmp1next) - { - if (!tmp1empty) - { - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); - stackptr += sizeof(sljit_sw); - } - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]); - tmp1empty = FALSE; - tmp1next = FALSE; - } - else - { - if (!tmp2empty) - { - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); - stackptr += sizeof(sljit_sw); - } - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]); - tmp2empty = FALSE; - tmp1next = TRUE; - } - } - else - { - if (tmp1next) - { - SLJIT_ASSERT(!tmp1empty); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count], TMP1, 0); - tmp1empty = stackptr >= stacktop; - if (!tmp1empty) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr); - stackptr += sizeof(sljit_sw); - } - tmp1next = FALSE; - } - else - { - SLJIT_ASSERT(!tmp2empty); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count], TMP2, 0); - tmp2empty = stackptr >= stacktop; - if (!tmp2empty) - { - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), stackptr); - stackptr += sizeof(sljit_sw); - } - tmp1next = TRUE; - } - } - } - } -while (status != end); - -if (save) - { - if (tmp1next) - { - if (!tmp1empty) - { - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); - stackptr += sizeof(sljit_sw); - } - if (!tmp2empty) - { - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); - stackptr += sizeof(sljit_sw); - } - } - else - { - if (!tmp2empty) - { - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); - stackptr += sizeof(sljit_sw); - } - if (!tmp1empty) - { - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); - stackptr += sizeof(sljit_sw); - } - } - } -SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty))); -} - -static SLJIT_INLINE pcre_uchar *set_then_offsets(compiler_common *common, pcre_uchar *cc, pcre_uint8 *current_offset) -{ -pcre_uchar *end = bracketend(cc); -BOOL has_alternatives = cc[GET(cc, 1)] == OP_ALT; - -/* Assert captures then. */ -if (*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) - current_offset = NULL; -/* Conditional block does not. */ -if (*cc == OP_COND || *cc == OP_SCOND) - has_alternatives = FALSE; - -cc = next_opcode(common, cc); -if (has_alternatives) - current_offset = common->then_offsets + (cc - common->start); - -while (cc < end) - { - if ((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND)) - cc = set_then_offsets(common, cc, current_offset); - else - { - if (*cc == OP_ALT && has_alternatives) - current_offset = common->then_offsets + (cc + 1 + LINK_SIZE - common->start); - if (*cc >= OP_THEN && *cc <= OP_THEN_ARG && current_offset != NULL) - *current_offset = 1; - cc = next_opcode(common, cc); - } - } - -return end; -} - -#undef CASE_ITERATOR_PRIVATE_DATA_1 -#undef CASE_ITERATOR_PRIVATE_DATA_2A -#undef CASE_ITERATOR_PRIVATE_DATA_2B -#undef CASE_ITERATOR_TYPE_PRIVATE_DATA_1 -#undef CASE_ITERATOR_TYPE_PRIVATE_DATA_2A -#undef CASE_ITERATOR_TYPE_PRIVATE_DATA_2B - -static SLJIT_INLINE BOOL is_powerof2(unsigned int value) -{ -return (value & (value - 1)) == 0; -} - -static SLJIT_INLINE void set_jumps(jump_list *list, struct sljit_label *label) -{ -while (list) - { - /* sljit_set_label is clever enough to do nothing - if either the jump or the label is NULL. */ - SET_LABEL(list->jump, label); - list = list->next; - } -} - -static SLJIT_INLINE void add_jump(struct sljit_compiler *compiler, jump_list **list, struct sljit_jump* jump) -{ -jump_list *list_item = sljit_alloc_memory(compiler, sizeof(jump_list)); -if (list_item) - { - list_item->next = *list; - list_item->jump = jump; - *list = list_item; - } -} - -static void add_stub(compiler_common *common, struct sljit_jump *start) -{ -DEFINE_COMPILER; -stub_list* list_item = sljit_alloc_memory(compiler, sizeof(stub_list)); - -if (list_item) - { - list_item->start = start; - list_item->quit = LABEL(); - list_item->next = common->stubs; - common->stubs = list_item; - } -} - -static void flush_stubs(compiler_common *common) -{ -DEFINE_COMPILER; -stub_list* list_item = common->stubs; - -while (list_item) - { - JUMPHERE(list_item->start); - add_jump(compiler, &common->stackalloc, JUMP(SLJIT_FAST_CALL)); - JUMPTO(SLJIT_JUMP, list_item->quit); - list_item = list_item->next; - } -common->stubs = NULL; -} - -static SLJIT_INLINE void count_match(compiler_common *common) -{ -DEFINE_COMPILER; - -OP2(SLJIT_SUB | SLJIT_SET_E, COUNT_MATCH, 0, COUNT_MATCH, 0, SLJIT_IMM, 1); -add_jump(compiler, &common->calllimit, JUMP(SLJIT_C_ZERO)); -} - -static SLJIT_INLINE void allocate_stack(compiler_common *common, int size) -{ -/* May destroy all locals and registers except TMP2. */ -DEFINE_COMPILER; - -OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, size * sizeof(sljit_sw)); -#ifdef DESTROY_REGISTERS -OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 12345); -OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); -OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP1, 0); -OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, TMP1, 0); -OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP1, 0); -#endif -add_stub(common, CMP(SLJIT_C_GREATER, STACK_TOP, 0, STACK_LIMIT, 0)); -} - -static SLJIT_INLINE void free_stack(compiler_common *common, int size) -{ -DEFINE_COMPILER; -OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, size * sizeof(sljit_sw)); -} - -static SLJIT_INLINE void reset_ovector(compiler_common *common, int length) -{ -DEFINE_COMPILER; -struct sljit_label *loop; -int i; - -/* At this point we can freely use all temporary registers. */ -SLJIT_ASSERT(length > 1); -/* TMP1 returns with begin - 1. */ -OP2(SLJIT_SUB, SLJIT_SCRATCH_REG1, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), SLJIT_OFFSETOF(jit_arguments, begin), SLJIT_IMM, IN_UCHARS(1)); -if (length < 8) - { - for (i = 1; i < length; i++) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(i), SLJIT_SCRATCH_REG1, 0); - } -else - { - GET_LOCAL_BASE(SLJIT_SCRATCH_REG2, 0, OVECTOR_START); - OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, length - 1); - loop = LABEL(); - OP1(SLJIT_MOVU, SLJIT_MEM1(SLJIT_SCRATCH_REG2), sizeof(sljit_sw), SLJIT_SCRATCH_REG1, 0); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 1); - JUMPTO(SLJIT_C_NOT_ZERO, loop); - } -} - -static SLJIT_INLINE void do_reset_match(compiler_common *common, int length) -{ -DEFINE_COMPILER; -struct sljit_label *loop; -int i; - -SLJIT_ASSERT(length > 1); -/* OVECTOR(1) contains the "string begin - 1" constant. */ -if (length > 2) - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)); -if (length < 8) - { - for (i = 2; i < length; i++) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(i), TMP1, 0); - } -else - { - GET_LOCAL_BASE(TMP2, 0, OVECTOR_START + sizeof(sljit_sw)); - OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_IMM, length - 2); - loop = LABEL(); - OP1(SLJIT_MOVU, SLJIT_MEM1(TMP2), sizeof(sljit_sw), TMP1, 0); - OP2(SLJIT_SUB | SLJIT_SET_E, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, 1); - JUMPTO(SLJIT_C_NOT_ZERO, loop); - } - -OP1(SLJIT_MOV, STACK_TOP, 0, ARGUMENTS, 0); -if (common->mark_ptr != 0) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, SLJIT_IMM, 0); -if (common->control_head_ptr != 0) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, SLJIT_IMM, 0); -OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), SLJIT_OFFSETOF(jit_arguments, stack)); -OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_ptr); -OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), SLJIT_OFFSETOF(struct sljit_stack, base)); -} - -static sljit_sw SLJIT_CALL do_search_mark(sljit_sw *current, const pcre_uchar *skip_arg) -{ -while (current != NULL) - { - switch (current[-2]) - { - case type_then_trap: - break; - - case type_mark: - if (STRCMP_UC_UC(skip_arg, (pcre_uchar *)current[-3]) == 0) - return current[-4]; - break; - - default: - SLJIT_ASSERT_STOP(); - break; - } - current = (sljit_sw*)current[-1]; - } -return -1; -} - -static SLJIT_INLINE void copy_ovector(compiler_common *common, int topbracket) -{ -DEFINE_COMPILER; -struct sljit_label *loop; -struct sljit_jump *early_quit; - -/* At this point we can freely use all registers. */ -OP1(SLJIT_MOV, SLJIT_SAVED_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)); -OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1), STR_PTR, 0); - -OP1(SLJIT_MOV, SLJIT_SCRATCH_REG1, 0, ARGUMENTS, 0); -if (common->mark_ptr != 0) - OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); -OP1(SLJIT_MOV_SI, SLJIT_SCRATCH_REG2, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG1), SLJIT_OFFSETOF(jit_arguments, offset_count)); -if (common->mark_ptr != 0) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SCRATCH_REG1), SLJIT_OFFSETOF(jit_arguments, mark_ptr), SLJIT_SCRATCH_REG3, 0); -OP2(SLJIT_SUB, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG1), SLJIT_OFFSETOF(jit_arguments, offsets), SLJIT_IMM, sizeof(int)); -OP1(SLJIT_MOV, SLJIT_SCRATCH_REG1, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG1), SLJIT_OFFSETOF(jit_arguments, begin)); -GET_LOCAL_BASE(SLJIT_SAVED_REG1, 0, OVECTOR_START); -/* Unlikely, but possible */ -early_quit = CMP(SLJIT_C_EQUAL, SLJIT_SCRATCH_REG2, 0, SLJIT_IMM, 0); -loop = LABEL(); -OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), 0, SLJIT_SCRATCH_REG1, 0); -OP2(SLJIT_ADD, SLJIT_SAVED_REG1, 0, SLJIT_SAVED_REG1, 0, SLJIT_IMM, sizeof(sljit_sw)); -/* Copy the integer value to the output buffer */ -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 -OP2(SLJIT_ASHR, SLJIT_SAVED_REG2, 0, SLJIT_SAVED_REG2, 0, SLJIT_IMM, UCHAR_SHIFT); -#endif -OP1(SLJIT_MOVU_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG3), sizeof(int), SLJIT_SAVED_REG2, 0); -OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_SCRATCH_REG2, 0, SLJIT_SCRATCH_REG2, 0, SLJIT_IMM, 1); -JUMPTO(SLJIT_C_NOT_ZERO, loop); -JUMPHERE(early_quit); - -/* Calculate the return value, which is the maximum ovector value. */ -if (topbracket > 1) - { - GET_LOCAL_BASE(SLJIT_SCRATCH_REG1, 0, OVECTOR_START + topbracket * 2 * sizeof(sljit_sw)); - OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, SLJIT_IMM, topbracket + 1); - - /* OVECTOR(0) is never equal to SLJIT_SAVED_REG3. */ - loop = LABEL(); - OP1(SLJIT_MOVU, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG1), -(2 * (sljit_sw)sizeof(sljit_sw))); - OP2(SLJIT_SUB, SLJIT_SCRATCH_REG2, 0, SLJIT_SCRATCH_REG2, 0, SLJIT_IMM, 1); - CMPTO(SLJIT_C_EQUAL, SLJIT_SCRATCH_REG3, 0, SLJIT_SAVED_REG3, 0, loop); - OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_SCRATCH_REG2, 0); - } -else - OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1); -} - -static SLJIT_INLINE void return_with_partial_match(compiler_common *common, struct sljit_label *quit) -{ -DEFINE_COMPILER; -struct sljit_jump *jump; - -SLJIT_COMPILE_ASSERT(STR_END == SLJIT_SAVED_REG2, str_end_must_be_saved_reg2); -SLJIT_ASSERT(common->start_used_ptr != 0 && common->start_ptr != 0 - && (common->mode == JIT_PARTIAL_SOFT_COMPILE ? common->hit_start != 0 : common->hit_start == 0)); - -OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, ARGUMENTS, 0); -OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_PARTIAL); -OP1(SLJIT_MOV_SI, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, real_offset_count)); -CMPTO(SLJIT_C_SIG_LESS, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 2, quit); - -/* Store match begin and end. */ -OP1(SLJIT_MOV, SLJIT_SAVED_REG1, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, begin)); -OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, offsets)); - -jump = CMP(SLJIT_C_SIG_LESS, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 3); -OP2(SLJIT_SUB, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_ptr : (common->hit_start + (int)sizeof(sljit_sw)), SLJIT_SAVED_REG1, 0); -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 -OP2(SLJIT_ASHR, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, UCHAR_SHIFT); -#endif -OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG2), 2 * sizeof(int), SLJIT_SCRATCH_REG3, 0); -JUMPHERE(jump); - -OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_used_ptr : common->hit_start); -OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, STR_END, 0, SLJIT_SAVED_REG1, 0); -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 -OP2(SLJIT_ASHR, SLJIT_SAVED_REG2, 0, SLJIT_SAVED_REG2, 0, SLJIT_IMM, UCHAR_SHIFT); -#endif -OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG2), sizeof(int), SLJIT_SAVED_REG2, 0); - -OP2(SLJIT_SUB, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_SAVED_REG1, 0); -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 -OP2(SLJIT_ASHR, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, UCHAR_SHIFT); -#endif -OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG2), 0, SLJIT_SCRATCH_REG3, 0); - -JUMPTO(SLJIT_JUMP, quit); -} - -static SLJIT_INLINE void check_start_used_ptr(compiler_common *common) -{ -/* May destroy TMP1. */ -DEFINE_COMPILER; -struct sljit_jump *jump; - -if (common->mode == JIT_PARTIAL_SOFT_COMPILE) - { - /* The value of -1 must be kept for start_used_ptr! */ - OP2(SLJIT_ADD, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, 1); - /* Jumps if start_used_ptr < STR_PTR, or start_used_ptr == -1. Although overwriting - is not necessary if start_used_ptr == STR_PTR, it does not hurt as well. */ - jump = CMP(SLJIT_C_LESS_EQUAL, TMP1, 0, STR_PTR, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); - JUMPHERE(jump); - } -else if (common->mode == JIT_PARTIAL_HARD_COMPILE) - { - jump = CMP(SLJIT_C_LESS_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); - JUMPHERE(jump); - } -} - -static SLJIT_INLINE BOOL char_has_othercase(compiler_common *common, pcre_uchar* cc) -{ -/* Detects if the character has an othercase. */ -unsigned int c; - -#ifdef SUPPORT_UTF -if (common->utf) - { - GETCHAR(c, cc); - if (c > 127) - { -#ifdef SUPPORT_UCP - return c != UCD_OTHERCASE(c); -#else - return FALSE; -#endif - } -#ifndef COMPILE_PCRE8 - return common->fcc[c] != c; -#endif - } -else -#endif - c = *cc; -return MAX_255(c) ? common->fcc[c] != c : FALSE; -} - -static SLJIT_INLINE unsigned int char_othercase(compiler_common *common, unsigned int c) -{ -/* Returns with the othercase. */ -#ifdef SUPPORT_UTF -if (common->utf && c > 127) - { -#ifdef SUPPORT_UCP - return UCD_OTHERCASE(c); -#else - return c; -#endif - } -#endif -return TABLE_GET(c, common->fcc, c); -} - -static unsigned int char_get_othercase_bit(compiler_common *common, pcre_uchar* cc) -{ -/* Detects if the character and its othercase has only 1 bit difference. */ -unsigned int c, oc, bit; -#if defined SUPPORT_UTF && defined COMPILE_PCRE8 -int n; -#endif - -#ifdef SUPPORT_UTF -if (common->utf) - { - GETCHAR(c, cc); - if (c <= 127) - oc = common->fcc[c]; - else - { -#ifdef SUPPORT_UCP - oc = UCD_OTHERCASE(c); -#else - oc = c; -#endif - } - } -else - { - c = *cc; - oc = TABLE_GET(c, common->fcc, c); - } -#else -c = *cc; -oc = TABLE_GET(c, common->fcc, c); -#endif - -SLJIT_ASSERT(c != oc); - -bit = c ^ oc; -/* Optimized for English alphabet. */ -if (c <= 127 && bit == 0x20) - return (0 << 8) | 0x20; - -/* Since c != oc, they must have at least 1 bit difference. */ -if (!is_powerof2(bit)) - return 0; - -#if defined COMPILE_PCRE8 - -#ifdef SUPPORT_UTF -if (common->utf && c > 127) - { - n = GET_EXTRALEN(*cc); - while ((bit & 0x3f) == 0) - { - n--; - bit >>= 6; - } - return (n << 8) | bit; - } -#endif /* SUPPORT_UTF */ -return (0 << 8) | bit; - -#elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - -#ifdef SUPPORT_UTF -if (common->utf && c > 65535) - { - if (bit >= (1 << 10)) - bit >>= 10; - else - return (bit < 256) ? ((2 << 8) | bit) : ((3 << 8) | (bit >> 8)); - } -#endif /* SUPPORT_UTF */ -return (bit < 256) ? ((0 << 8) | bit) : ((1 << 8) | (bit >> 8)); - -#endif /* COMPILE_PCRE[8|16|32] */ -} - -static void check_partial(compiler_common *common, BOOL force) -{ -/* Checks whether a partial matching is occurred. Does not modify registers. */ -DEFINE_COMPILER; -struct sljit_jump *jump = NULL; - -SLJIT_ASSERT(!force || common->mode != JIT_COMPILE); - -if (common->mode == JIT_COMPILE) - return; - -if (!force) - jump = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); -else if (common->mode == JIT_PARTIAL_SOFT_COMPILE) - jump = CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, -1); - -if (common->mode == JIT_PARTIAL_SOFT_COMPILE) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, 0); -else - { - if (common->partialmatchlabel != NULL) - JUMPTO(SLJIT_JUMP, common->partialmatchlabel); - else - add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP)); - } - -if (jump != NULL) - JUMPHERE(jump); -} - -static void check_str_end(compiler_common *common, jump_list **end_reached) -{ -/* Does not affect registers. Usually used in a tight spot. */ -DEFINE_COMPILER; -struct sljit_jump *jump; - -if (common->mode == JIT_COMPILE) - { - add_jump(compiler, end_reached, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); - return; - } - -jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); -if (common->mode == JIT_PARTIAL_SOFT_COMPILE) - { - add_jump(compiler, end_reached, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0)); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, 0); - add_jump(compiler, end_reached, JUMP(SLJIT_JUMP)); - } -else - { - add_jump(compiler, end_reached, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0)); - if (common->partialmatchlabel != NULL) - JUMPTO(SLJIT_JUMP, common->partialmatchlabel); - else - add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP)); - } -JUMPHERE(jump); -} - -static void detect_partial_match(compiler_common *common, jump_list **backtracks) -{ -DEFINE_COMPILER; -struct sljit_jump *jump; - -if (common->mode == JIT_COMPILE) - { - add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); - return; - } - -/* Partial matching mode. */ -jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); -add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0)); -if (common->mode == JIT_PARTIAL_SOFT_COMPILE) - { - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, 0); - add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); - } -else - { - if (common->partialmatchlabel != NULL) - JUMPTO(SLJIT_JUMP, common->partialmatchlabel); - else - add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP)); - } -JUMPHERE(jump); -} - -static void read_char(compiler_common *common) -{ -/* Reads the character into TMP1, updates STR_PTR. -Does not check STR_END. TMP2 Destroyed. */ -DEFINE_COMPILER; -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 -struct sljit_jump *jump; -#endif - -OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 -if (common->utf) - { -#if defined COMPILE_PCRE8 - jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); -#elif defined COMPILE_PCRE16 - jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); -#endif /* COMPILE_PCRE[8|16] */ - add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); - JUMPHERE(jump); - } -#endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ -OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); -} - -static void peek_char(compiler_common *common) -{ -/* Reads the character into TMP1, keeps STR_PTR. -Does not check STR_END. TMP2 Destroyed. */ -DEFINE_COMPILER; -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 -struct sljit_jump *jump; -#endif - -OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 -if (common->utf) - { -#if defined COMPILE_PCRE8 - jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); -#elif defined COMPILE_PCRE16 - jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); -#endif /* COMPILE_PCRE[8|16] */ - add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); - OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); - JUMPHERE(jump); - } -#endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ -} - -static void read_char8_type(compiler_common *common) -{ -/* Reads the character type into TMP1, updates STR_PTR. Does not check STR_END. */ -DEFINE_COMPILER; -#if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 -struct sljit_jump *jump; -#endif - -#ifdef SUPPORT_UTF -if (common->utf) - { - OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); -#if defined COMPILE_PCRE8 - /* This can be an extra read in some situations, but hopefully - it is needed in most cases. */ - OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); - jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0); - add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL)); - JUMPHERE(jump); -#elif defined COMPILE_PCRE16 - OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); - jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); - OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); - JUMPHERE(jump); - /* Skip low surrogate if necessary. */ - OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xfc00); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0xd800); - OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); - OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); -#elif defined COMPILE_PCRE32 - OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); - jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); - OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); - JUMPHERE(jump); -#endif /* COMPILE_PCRE[8|16|32] */ - return; - } -#endif /* SUPPORT_UTF */ -OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); -OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 -/* The ctypes array contains only 256 values. */ -OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); -jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); -#endif -OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 -JUMPHERE(jump); -#endif -} - -static void skip_char_back(compiler_common *common) -{ -/* Goes one character back. Affects STR_PTR and TMP1. Does not check begin. */ -DEFINE_COMPILER; -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 -#if defined COMPILE_PCRE8 -struct sljit_label *label; - -if (common->utf) - { - label = LABEL(); - OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -IN_UCHARS(1)); - OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); - OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xc0); - CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0x80, label); - return; - } -#elif defined COMPILE_PCRE16 -if (common->utf) - { - OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -IN_UCHARS(1)); - OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); - /* Skip low surrogate if necessary. */ - OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xdc00); - OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); - OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); - OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0); - return; - } -#endif /* COMPILE_PCRE[8|16] */ -#endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ -OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); -} - -static void check_newlinechar(compiler_common *common, int nltype, jump_list **backtracks, BOOL jumpiftrue) -{ -/* Character comes in TMP1. Checks if it is a newline. TMP2 may be destroyed. */ -DEFINE_COMPILER; - -if (nltype == NLTYPE_ANY) - { - add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL)); - add_jump(compiler, backtracks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); - } -else if (nltype == NLTYPE_ANYCRLF) - { - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_CR); - OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_NL); - OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - add_jump(compiler, backtracks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); - } -else - { - SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256); - add_jump(compiler, backtracks, CMP(jumpiftrue ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline)); - } -} - -#ifdef SUPPORT_UTF - -#if defined COMPILE_PCRE8 -static void do_utfreadchar(compiler_common *common) -{ -/* Fast decoding a UTF-8 character. TMP1 contains the first byte -of the character (>= 0xc0). Return char value in TMP1, length - 1 in TMP2. */ -DEFINE_COMPILER; -struct sljit_jump *jump; - -sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); -/* Searching for the first zero. */ -OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x20); -jump = JUMP(SLJIT_C_NOT_ZERO); -/* Two byte sequence. */ -OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); -OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); -OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1f); -OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); -OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); -OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); -OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(1)); -sljit_emit_fast_return(compiler, RETURN_ADDR, 0); -JUMPHERE(jump); - -OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x10); -jump = JUMP(SLJIT_C_NOT_ZERO); -/* Three byte sequence. */ -OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); -OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0f); -OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 12); -OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); -OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); -OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); -OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2)); -OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); -OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); -OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); -OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(2)); -sljit_emit_fast_return(compiler, RETURN_ADDR, 0); -JUMPHERE(jump); - -/* Four byte sequence. */ -OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); -OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x07); -OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 18); -OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); -OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 12); -OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); -OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2)); -OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); -OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); -OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); -OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(3)); -OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3)); -OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); -OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); -OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(3)); -sljit_emit_fast_return(compiler, RETURN_ADDR, 0); -} - -static void do_utfreadtype8(compiler_common *common) -{ -/* Fast decoding a UTF-8 character type. TMP2 contains the first byte -of the character (>= 0xc0). Return value in TMP1. */ -DEFINE_COMPILER; -struct sljit_jump *jump; -struct sljit_jump *compare; - -sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); - -OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0x20); -jump = JUMP(SLJIT_C_NOT_ZERO); -/* Two byte sequence. */ -OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); -OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); -OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x1f); -OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); -OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f); -OP2(SLJIT_OR, TMP2, 0, TMP2, 0, TMP1, 0); -compare = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); -OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); -sljit_emit_fast_return(compiler, RETURN_ADDR, 0); - -JUMPHERE(compare); -OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); -sljit_emit_fast_return(compiler, RETURN_ADDR, 0); -JUMPHERE(jump); - -/* We only have types for characters less than 256. */ -OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(utf8_table4) - 0xc0); -OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); -OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); -sljit_emit_fast_return(compiler, RETURN_ADDR, 0); -} - -#elif defined COMPILE_PCRE16 - -static void do_utfreadchar(compiler_common *common) -{ -/* Fast decoding a UTF-16 character. TMP1 contains the first 16 bit char -of the character (>= 0xd800). Return char value in TMP1, length - 1 in TMP2. */ -DEFINE_COMPILER; -struct sljit_jump *jump; - -sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); -jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xdc00); -/* Do nothing, only return. */ -sljit_emit_fast_return(compiler, RETURN_ADDR, 0); - -JUMPHERE(jump); -/* Combine two 16 bit characters. */ -OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); -OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); -OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3ff); -OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 10); -OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3ff); -OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); -OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(1)); -OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000); -sljit_emit_fast_return(compiler, RETURN_ADDR, 0); -} - -#endif /* COMPILE_PCRE[8|16] */ - -#endif /* SUPPORT_UTF */ - -#ifdef SUPPORT_UCP - -/* UCD_BLOCK_SIZE must be 128 (see the assert below). */ -#define UCD_BLOCK_MASK 127 -#define UCD_BLOCK_SHIFT 7 - -static void do_getucd(compiler_common *common) -{ -/* Search the UCD record for the character comes in TMP1. -Returns chartype in TMP1 and UCD offset in TMP2. */ -DEFINE_COMPILER; - -SLJIT_ASSERT(UCD_BLOCK_SIZE == 128 && sizeof(ucd_record) == 8); - -sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); -OP2(SLJIT_LSHR, TMP2, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_SHIFT); -OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(ucd_stage1)); -OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_MASK); -OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCD_BLOCK_SHIFT); -OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0); -OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_stage2)); -OP1(SLJIT_MOV_UH, TMP2, 0, SLJIT_MEM2(TMP2, TMP1), 1); -OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, chartype)); -OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM2(TMP1, TMP2), 3); -sljit_emit_fast_return(compiler, RETURN_ADDR, 0); -} -#endif - -static SLJIT_INLINE struct sljit_label *mainloop_entry(compiler_common *common, BOOL hascrorlf, BOOL firstline) -{ -DEFINE_COMPILER; -struct sljit_label *mainloop; -struct sljit_label *newlinelabel = NULL; -struct sljit_jump *start; -struct sljit_jump *end = NULL; -struct sljit_jump *nl = NULL; -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 -struct sljit_jump *singlechar; -#endif -jump_list *newline = NULL; -BOOL newlinecheck = FALSE; -BOOL readuchar = FALSE; - -if (!(hascrorlf || firstline) && (common->nltype == NLTYPE_ANY || - common->nltype == NLTYPE_ANYCRLF || common->newline > 255)) - newlinecheck = TRUE; - -if (firstline) - { - /* Search for the end of the first line. */ - SLJIT_ASSERT(common->first_line_end != 0); - OP1(SLJIT_MOV, TMP3, 0, STR_PTR, 0); - - if (common->nltype == NLTYPE_FIXED && common->newline > 255) - { - mainloop = LABEL(); - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); - end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); - OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); - OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); - CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, mainloop); - CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, mainloop); - JUMPHERE(end); - OP2(SLJIT_SUB, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); - } - else - { - end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); - mainloop = LABEL(); - /* Continual stores does not cause data dependency. */ - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0); - read_char(common); - check_newlinechar(common, common->nltype, &newline, TRUE); - CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop); - JUMPHERE(end); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0); - set_jumps(newline, LABEL()); - } - - OP1(SLJIT_MOV, STR_PTR, 0, TMP3, 0); - } - -start = JUMP(SLJIT_JUMP); - -if (newlinecheck) - { - newlinelabel = LABEL(); - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); - end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); - OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, common->newline & 0xff); - OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); -#endif - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); - nl = JUMP(SLJIT_JUMP); - } - -mainloop = LABEL(); - -/* Increasing the STR_PTR here requires one less jump in the most common case. */ -#ifdef SUPPORT_UTF -if (common->utf) readuchar = TRUE; -#endif -if (newlinecheck) readuchar = TRUE; - -if (readuchar) - OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); - -if (newlinecheck) - CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, newlinelabel); - -OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); -#if defined SUPPORT_UTF && !defined COMPILE_PCRE32 -#if defined COMPILE_PCRE8 -if (common->utf) - { - singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); - OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0); - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); - JUMPHERE(singlechar); - } -#elif defined COMPILE_PCRE16 -if (common->utf) - { - singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); - OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); - OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); - OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); - JUMPHERE(singlechar); - } -#endif /* COMPILE_PCRE[8|16] */ -#endif /* SUPPORT_UTF && !COMPILE_PCRE32 */ -JUMPHERE(start); - -if (newlinecheck) - { - JUMPHERE(end); - JUMPHERE(nl); - } - -return mainloop; -} - -#define MAX_N_CHARS 3 - -static SLJIT_INLINE BOOL fast_forward_first_n_chars(compiler_common *common, BOOL firstline) -{ -DEFINE_COMPILER; -struct sljit_label *start; -struct sljit_jump *quit; -pcre_uint32 chars[MAX_N_CHARS * 2]; -pcre_uchar *cc = common->start + 1 + LINK_SIZE; -int location = 0; -pcre_int32 len, c, bit, caseless; -int must_stop; - -/* We do not support alternatives now. */ -if (*(common->start + GET(common->start, 1)) == OP_ALT) - return FALSE; - -while (TRUE) - { - caseless = 0; - must_stop = 1; - switch(*cc) - { - case OP_CHAR: - must_stop = 0; - cc++; - break; - - case OP_CHARI: - caseless = 1; - must_stop = 0; - cc++; - break; - - case OP_SOD: - case OP_SOM: - case OP_SET_SOM: - case OP_NOT_WORD_BOUNDARY: - case OP_WORD_BOUNDARY: - case OP_EODN: - case OP_EOD: - case OP_CIRC: - case OP_CIRCM: - case OP_DOLL: - case OP_DOLLM: - /* Zero width assertions. */ - cc++; - continue; - - case OP_PLUS: - case OP_MINPLUS: - case OP_POSPLUS: - cc++; - break; - - case OP_EXACT: - cc += 1 + IMM2_SIZE; - break; - - case OP_PLUSI: - case OP_MINPLUSI: - case OP_POSPLUSI: - caseless = 1; - cc++; - break; - - case OP_EXACTI: - caseless = 1; - cc += 1 + IMM2_SIZE; - break; - - default: - must_stop = 2; - break; - } - - if (must_stop == 2) - break; - - len = 1; -#ifdef SUPPORT_UTF - if (common->utf && HAS_EXTRALEN(cc[0])) len += GET_EXTRALEN(cc[0]); -#endif - - if (caseless && char_has_othercase(common, cc)) - { - caseless = char_get_othercase_bit(common, cc); - if (caseless == 0) - return FALSE; -#ifdef COMPILE_PCRE8 - caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 8)); -#else - if ((caseless & 0x100) != 0) - caseless = ((caseless & 0xff) << 16) | (len - (caseless >> 9)); - else - caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 9)); -#endif - } - else - caseless = 0; - - while (len > 0 && location < MAX_N_CHARS * 2) - { - c = *cc; - bit = 0; - if (len == (caseless & 0xff)) - { - bit = caseless >> 8; - c |= bit; - } - - chars[location] = c; - chars[location + 1] = bit; - - len--; - location += 2; - cc++; - } - - if (location >= MAX_N_CHARS * 2 || must_stop != 0) - break; - } - -/* At least two characters are required. */ -if (location < 2 * 2) - return FALSE; - -if (firstline) - { - SLJIT_ASSERT(common->first_line_end != 0); - OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); - OP2(SLJIT_SUB, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, SLJIT_IMM, IN_UCHARS((location >> 1) - 1)); - } -else - OP2(SLJIT_SUB, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS((location >> 1) - 1)); - -start = LABEL(); -quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); - -OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); -OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); -OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); -if (chars[1] != 0) - OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[1]); -CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[0], start); -if (location > 2 * 2) - OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); -if (chars[3] != 0) - OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, chars[3]); -CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, chars[2], start); -if (location > 2 * 2) - { - if (chars[5] != 0) - OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[5]); - CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[4], start); - } -OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); - -JUMPHERE(quit); - -if (firstline) - OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); -else - OP2(SLJIT_ADD, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS((location >> 1) - 1)); -return TRUE; -} - -#undef MAX_N_CHARS - -static SLJIT_INLINE void fast_forward_first_char(compiler_common *common, pcre_uchar first_char, BOOL caseless, BOOL firstline) -{ -DEFINE_COMPILER; -struct sljit_label *start; -struct sljit_jump *quit; -struct sljit_jump *found; -pcre_uchar oc, bit; - -if (firstline) - { - SLJIT_ASSERT(common->first_line_end != 0); - OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); - OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); - } - -start = LABEL(); -quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); -OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); - -oc = first_char; -if (caseless) - { - oc = TABLE_GET(first_char, common->fcc, first_char); -#if defined SUPPORT_UCP && !(defined COMPILE_PCRE8) - if (first_char > 127 && common->utf) - oc = UCD_OTHERCASE(first_char); -#endif - } -if (first_char == oc) - found = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, first_char); -else - { - bit = first_char ^ oc; - if (is_powerof2(bit)) - { - OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, bit); - found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, first_char | bit); - } - else - { - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, first_char); - OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, oc); - OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - found = JUMP(SLJIT_C_NOT_ZERO); - } - } - -OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); -JUMPTO(SLJIT_JUMP, start); -JUMPHERE(found); -JUMPHERE(quit); - -if (firstline) - OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); -} - -static SLJIT_INLINE void fast_forward_newline(compiler_common *common, BOOL firstline) -{ -DEFINE_COMPILER; -struct sljit_label *loop; -struct sljit_jump *lastchar; -struct sljit_jump *firstchar; -struct sljit_jump *quit; -struct sljit_jump *foundcr = NULL; -struct sljit_jump *notfoundnl; -jump_list *newline = NULL; - -if (firstline) - { - SLJIT_ASSERT(common->first_line_end != 0); - OP1(SLJIT_MOV, TMP3, 0, STR_END, 0); - OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); - } - -if (common->nltype == NLTYPE_FIXED && common->newline > 255) - { - lastchar = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); - OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); - firstchar = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP2, 0); - - OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2)); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, STR_PTR, 0, TMP1, 0); - OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER_EQUAL); -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCHAR_SHIFT); -#endif - OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); - - loop = LABEL(); - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); - quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); - OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); - OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); - CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, loop); - CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, loop); - - JUMPHERE(quit); - JUMPHERE(firstchar); - JUMPHERE(lastchar); - - if (firstline) - OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); - return; - } - -OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); -OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); -firstchar = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP2, 0); -skip_char_back(common); - -loop = LABEL(); -read_char(common); -lastchar = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); -if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF) - foundcr = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR); -check_newlinechar(common, common->nltype, &newline, FALSE); -set_jumps(newline, loop); - -if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF) - { - quit = JUMP(SLJIT_JUMP); - JUMPHERE(foundcr); - notfoundnl = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); - OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_NL); - OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT); -#endif - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); - JUMPHERE(notfoundnl); - JUMPHERE(quit); - } -JUMPHERE(lastchar); -JUMPHERE(firstchar); - -if (firstline) - OP1(SLJIT_MOV, STR_END, 0, TMP3, 0); -} - -static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, jump_list **backtracks); - -static SLJIT_INLINE void fast_forward_start_bits(compiler_common *common, sljit_uw start_bits, BOOL firstline) -{ -DEFINE_COMPILER; -struct sljit_label *start; -struct sljit_jump *quit; -struct sljit_jump *found = NULL; -jump_list *matches = NULL; -pcre_uint8 inverted_start_bits[32]; -int i; -#ifndef COMPILE_PCRE8 -struct sljit_jump *jump; -#endif - -for (i = 0; i < 32; ++i) - inverted_start_bits[i] = ~(((pcre_uint8*)start_bits)[i]); - -if (firstline) - { - SLJIT_ASSERT(common->first_line_end != 0); - OP1(SLJIT_MOV, RETURN_ADDR, 0, STR_END, 0); - OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); - } - -start = LABEL(); -quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); -OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); -#ifdef SUPPORT_UTF -if (common->utf) - OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); -#endif - -if (!check_class_ranges(common, inverted_start_bits, (inverted_start_bits[31] & 0x80) != 0, &matches)) - { -#ifndef COMPILE_PCRE8 - jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 255); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 255); - JUMPHERE(jump); -#endif - OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); - OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); - OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), start_bits); - OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); - OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); - found = JUMP(SLJIT_C_NOT_ZERO); - } - -#ifdef SUPPORT_UTF -if (common->utf) - OP1(SLJIT_MOV, TMP1, 0, TMP3, 0); -#endif -OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); -#ifdef SUPPORT_UTF -#if defined COMPILE_PCRE8 -if (common->utf) - { - CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0, start); - OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0); - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); - } -#elif defined COMPILE_PCRE16 -if (common->utf) - { - CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start); - OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); - OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); - OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); - } -#endif /* COMPILE_PCRE[8|16] */ -#endif /* SUPPORT_UTF */ -JUMPTO(SLJIT_JUMP, start); -if (found != NULL) - JUMPHERE(found); -if (matches != NULL) - set_jumps(matches, LABEL()); -JUMPHERE(quit); - -if (firstline) - OP1(SLJIT_MOV, STR_END, 0, RETURN_ADDR, 0); -} - -static SLJIT_INLINE struct sljit_jump *search_requested_char(compiler_common *common, pcre_uchar req_char, BOOL caseless, BOOL has_firstchar) -{ -DEFINE_COMPILER; -struct sljit_label *loop; -struct sljit_jump *toolong; -struct sljit_jump *alreadyfound; -struct sljit_jump *found; -struct sljit_jump *foundoc = NULL; -struct sljit_jump *notfound; -pcre_uint32 oc, bit; - -SLJIT_ASSERT(common->req_char_ptr != 0); -OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->req_char_ptr); -OP2(SLJIT_ADD, TMP1, 0, STR_PTR, 0, SLJIT_IMM, REQ_BYTE_MAX); -toolong = CMP(SLJIT_C_LESS, TMP1, 0, STR_END, 0); -alreadyfound = CMP(SLJIT_C_LESS, STR_PTR, 0, TMP2, 0); - -if (has_firstchar) - OP2(SLJIT_ADD, TMP1, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); -else - OP1(SLJIT_MOV, TMP1, 0, STR_PTR, 0); - -loop = LABEL(); -notfound = CMP(SLJIT_C_GREATER_EQUAL, TMP1, 0, STR_END, 0); - -OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(TMP1), 0); -oc = req_char; -if (caseless) - { - oc = TABLE_GET(req_char, common->fcc, req_char); -#if defined SUPPORT_UCP && !(defined COMPILE_PCRE8) - if (req_char > 127 && common->utf) - oc = UCD_OTHERCASE(req_char); -#endif - } -if (req_char == oc) - found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, req_char); -else - { - bit = req_char ^ oc; - if (is_powerof2(bit)) - { - OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, bit); - found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, req_char | bit); - } - else - { - found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, req_char); - foundoc = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, oc); - } - } -OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(1)); -JUMPTO(SLJIT_JUMP, loop); - -JUMPHERE(found); -if (foundoc) - JUMPHERE(foundoc); -OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->req_char_ptr, TMP1, 0); -JUMPHERE(alreadyfound); -JUMPHERE(toolong); -return notfound; -} - -static void do_revertframes(compiler_common *common) -{ -DEFINE_COMPILER; -struct sljit_jump *jump; -struct sljit_label *mainloop; - -sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); -OP1(SLJIT_MOV, TMP1, 0, STACK_TOP, 0); -GET_LOCAL_BASE(TMP3, 0, 0); - -/* Drop frames until we reach STACK_TOP. */ -mainloop = LABEL(); -OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), 0); -OP2(SLJIT_SUB | SLJIT_SET_S, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0); -jump = JUMP(SLJIT_C_SIG_LESS_EQUAL); - -OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP3, 0); -OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), 0, SLJIT_MEM1(TMP1), sizeof(sljit_sw)); -OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), sizeof(sljit_sw), SLJIT_MEM1(TMP1), 2 * sizeof(sljit_sw)); -OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 3 * sizeof(sljit_sw)); -JUMPTO(SLJIT_JUMP, mainloop); - -JUMPHERE(jump); -jump = JUMP(SLJIT_C_SIG_LESS); -/* End of dropping frames. */ -sljit_emit_fast_return(compiler, RETURN_ADDR, 0); - -JUMPHERE(jump); -OP1(SLJIT_NEG, TMP2, 0, TMP2, 0); -OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP3, 0); -OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), 0, SLJIT_MEM1(TMP1), sizeof(sljit_sw)); -OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_sw)); -JUMPTO(SLJIT_JUMP, mainloop); -} - -static void check_wordboundary(compiler_common *common) -{ -DEFINE_COMPILER; -struct sljit_jump *skipread; -jump_list *skipread_list = NULL; -#if !(defined COMPILE_PCRE8) || defined SUPPORT_UTF -struct sljit_jump *jump; -#endif - -SLJIT_COMPILE_ASSERT(ctype_word == 0x10, ctype_word_must_be_16); - -sljit_emit_fast_enter(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); -/* Get type of the previous char, and put it to LOCALS1. */ -OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); -OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); -OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, SLJIT_IMM, 0); -skipread = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP1, 0); -skip_char_back(common); -check_start_used_ptr(common); -read_char(common); - -/* Testing char type. */ -#ifdef SUPPORT_UCP -if (common->use_ucp) - { - OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 1); - jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE); - add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); - OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); - OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); - OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Nd - ucp_Ll); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_No - ucp_Nd); - OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); - JUMPHERE(jump); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP2, 0); - } -else -#endif - { -#ifndef COMPILE_PCRE8 - jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); -#elif defined SUPPORT_UTF - /* Here LOCALS1 has already been zeroed. */ - jump = NULL; - if (common->utf) - jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); -#endif /* COMPILE_PCRE8 */ - OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), common->ctypes); - OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 4 /* ctype_word */); - OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP1, 0); -#ifndef COMPILE_PCRE8 - JUMPHERE(jump); -#elif defined SUPPORT_UTF - if (jump != NULL) - JUMPHERE(jump); -#endif /* COMPILE_PCRE8 */ - } -JUMPHERE(skipread); - -OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); -check_str_end(common, &skipread_list); -peek_char(common); - -/* Testing char type. This is a code duplication. */ -#ifdef SUPPORT_UCP -if (common->use_ucp) - { - OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 1); - jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE); - add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); - OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); - OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); - OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Nd - ucp_Ll); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_No - ucp_Nd); - OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); - JUMPHERE(jump); - } -else -#endif - { -#ifndef COMPILE_PCRE8 - /* TMP2 may be destroyed by peek_char. */ - OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); - jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); -#elif defined SUPPORT_UTF - OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); - jump = NULL; - if (common->utf) - jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); -#endif - OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), common->ctypes); - OP2(SLJIT_LSHR, TMP2, 0, TMP2, 0, SLJIT_IMM, 4 /* ctype_word */); - OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); -#ifndef COMPILE_PCRE8 - JUMPHERE(jump); -#elif defined SUPPORT_UTF - if (jump != NULL) - JUMPHERE(jump); -#endif /* COMPILE_PCRE8 */ - } -set_jumps(skipread_list, LABEL()); - -OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); -sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); -} - -/* - range format: - - ranges[0] = length of the range (max MAX_RANGE_SIZE, -1 means invalid range). - ranges[1] = first bit (0 or 1) - ranges[2-length] = position of the bit change (when the current bit is not equal to the previous) -*/ - -static BOOL check_ranges(compiler_common *common, int *ranges, jump_list **backtracks, BOOL readch) -{ -DEFINE_COMPILER; -struct sljit_jump *jump; - -if (ranges[0] < 0) - return FALSE; - -switch(ranges[0]) - { - case 1: - if (readch) - read_char(common); - add_jump(compiler, backtracks, CMP(ranges[1] == 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2])); - return TRUE; - - case 2: - if (readch) - read_char(common); - OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2]); - add_jump(compiler, backtracks, CMP(ranges[1] != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2])); - return TRUE; - - case 4: - if (ranges[2] + 1 == ranges[3] && ranges[4] + 1 == ranges[5]) - { - if (readch) - read_char(common); - if (ranges[1] != 0) - { - add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2])); - add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[4])); - } - else - { - jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[4])); - JUMPHERE(jump); - } - return TRUE; - } - if ((ranges[3] - ranges[2]) == (ranges[5] - ranges[4]) && is_powerof2(ranges[4] - ranges[2])) - { - if (readch) - read_char(common); - OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[4] - ranges[2]); - OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[4]); - add_jump(compiler, backtracks, CMP(ranges[1] != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[5] - ranges[4])); - return TRUE; - } - return FALSE; - - default: - return FALSE; - } -} - -static void get_ctype_ranges(compiler_common *common, int flag, int *ranges) -{ -int i, bit, length; -const pcre_uint8 *ctypes = (const pcre_uint8*)common->ctypes; - -bit = ctypes[0] & flag; -ranges[0] = -1; -ranges[1] = bit != 0 ? 1 : 0; -length = 0; - -for (i = 1; i < 256; i++) - if ((ctypes[i] & flag) != bit) - { - if (length >= MAX_RANGE_SIZE) - return; - ranges[2 + length] = i; - length++; - bit ^= flag; - } - -if (bit != 0) - { - if (length >= MAX_RANGE_SIZE) - return; - ranges[2 + length] = 256; - length++; - } -ranges[0] = length; -} - -static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, jump_list **backtracks) -{ -int ranges[2 + MAX_RANGE_SIZE]; -pcre_uint8 bit, cbit, all; -int i, byte, length = 0; - -bit = bits[0] & 0x1; -ranges[1] = bit; -/* Can be 0 or 255. */ -all = -bit; - -for (i = 0; i < 256; ) - { - byte = i >> 3; - if ((i & 0x7) == 0 && bits[byte] == all) - i += 8; - else - { - cbit = (bits[byte] >> (i & 0x7)) & 0x1; - if (cbit != bit) - { - if (length >= MAX_RANGE_SIZE) - return FALSE; - ranges[2 + length] = i; - length++; - bit = cbit; - all = -cbit; - } - i++; - } - } - -if (((bit == 0) && nclass) || ((bit == 1) && !nclass)) - { - if (length >= MAX_RANGE_SIZE) - return FALSE; - ranges[2 + length] = 256; - length++; - } -ranges[0] = length; - -return check_ranges(common, ranges, backtracks, FALSE); -} - -static void check_anynewline(compiler_common *common) -{ -/* Check whether TMP1 contains a newline character. TMP2 destroyed. */ -DEFINE_COMPILER; - -sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); - -OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a); -OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x0d - 0x0a); -OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); -OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x0a); -#if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 -#ifdef COMPILE_PCRE8 -if (common->utf) - { -#endif - OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2029 - 0x0a); -#ifdef COMPILE_PCRE8 - } -#endif -#endif /* SUPPORT_UTF || COMPILE_PCRE16 || COMPILE_PCRE32 */ -OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); -sljit_emit_fast_return(compiler, RETURN_ADDR, 0); -} - -static void check_hspace(compiler_common *common) -{ -/* Check whether TMP1 contains a newline character. TMP2 destroyed. */ -DEFINE_COMPILER; - -sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); - -OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x09); -OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); -OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x20); -OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); -OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xa0); -#if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 -#ifdef COMPILE_PCRE8 -if (common->utf) - { -#endif - OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x1680); - OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e); - OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x2000); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x200A - 0x2000); - OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x202f - 0x2000); - OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x205f - 0x2000); - OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x3000 - 0x2000); -#ifdef COMPILE_PCRE8 - } -#endif -#endif /* SUPPORT_UTF || COMPILE_PCRE16 || COMPILE_PCRE32 */ -OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - -sljit_emit_fast_return(compiler, RETURN_ADDR, 0); -} - -static void check_vspace(compiler_common *common) -{ -/* Check whether TMP1 contains a newline character. TMP2 destroyed. */ -DEFINE_COMPILER; - -sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); - -OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a); -OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x0d - 0x0a); -OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); -OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x0a); -#if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 -#ifdef COMPILE_PCRE8 -if (common->utf) - { -#endif - OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2029 - 0x0a); -#ifdef COMPILE_PCRE8 - } -#endif -#endif /* SUPPORT_UTF || COMPILE_PCRE16 || COMPILE_PCRE32 */ -OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - -sljit_emit_fast_return(compiler, RETURN_ADDR, 0); -} - -#define CHAR1 STR_END -#define CHAR2 STACK_TOP - -static void do_casefulcmp(compiler_common *common) -{ -DEFINE_COMPILER; -struct sljit_jump *jump; -struct sljit_label *label; - -sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); -OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); -OP1(SLJIT_MOV, TMP3, 0, CHAR1, 0); -OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, CHAR2, 0); -OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(1)); -OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); - -label = LABEL(); -OP1(MOVU_UCHAR, CHAR1, 0, SLJIT_MEM1(TMP1), IN_UCHARS(1)); -OP1(MOVU_UCHAR, CHAR2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); -jump = CMP(SLJIT_C_NOT_EQUAL, CHAR1, 0, CHAR2, 0); -OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_IMM, IN_UCHARS(1)); -JUMPTO(SLJIT_C_NOT_ZERO, label); - -JUMPHERE(jump); -OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); -OP1(SLJIT_MOV, CHAR1, 0, TMP3, 0); -OP1(SLJIT_MOV, CHAR2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); -sljit_emit_fast_return(compiler, RETURN_ADDR, 0); -} - -#define LCC_TABLE STACK_LIMIT - -static void do_caselesscmp(compiler_common *common) -{ -DEFINE_COMPILER; -struct sljit_jump *jump; -struct sljit_label *label; - -sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); -OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); - -OP1(SLJIT_MOV, TMP3, 0, LCC_TABLE, 0); -OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, CHAR1, 0); -OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, CHAR2, 0); -OP1(SLJIT_MOV, LCC_TABLE, 0, SLJIT_IMM, common->lcc); -OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(1)); -OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); - -label = LABEL(); -OP1(MOVU_UCHAR, CHAR1, 0, SLJIT_MEM1(TMP1), IN_UCHARS(1)); -OP1(MOVU_UCHAR, CHAR2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); -#ifndef COMPILE_PCRE8 -jump = CMP(SLJIT_C_GREATER, CHAR1, 0, SLJIT_IMM, 255); -#endif -OP1(SLJIT_MOV_UB, CHAR1, 0, SLJIT_MEM2(LCC_TABLE, CHAR1), 0); -#ifndef COMPILE_PCRE8 -JUMPHERE(jump); -jump = CMP(SLJIT_C_GREATER, CHAR2, 0, SLJIT_IMM, 255); -#endif -OP1(SLJIT_MOV_UB, CHAR2, 0, SLJIT_MEM2(LCC_TABLE, CHAR2), 0); -#ifndef COMPILE_PCRE8 -JUMPHERE(jump); -#endif -jump = CMP(SLJIT_C_NOT_EQUAL, CHAR1, 0, CHAR2, 0); -OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_IMM, IN_UCHARS(1)); -JUMPTO(SLJIT_C_NOT_ZERO, label); - -JUMPHERE(jump); -OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); -OP1(SLJIT_MOV, LCC_TABLE, 0, TMP3, 0); -OP1(SLJIT_MOV, CHAR1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); -OP1(SLJIT_MOV, CHAR2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); -sljit_emit_fast_return(compiler, RETURN_ADDR, 0); -} - -#undef LCC_TABLE -#undef CHAR1 -#undef CHAR2 - -#if defined SUPPORT_UTF && defined SUPPORT_UCP - -static const pcre_uchar * SLJIT_CALL do_utf_caselesscmp(pcre_uchar *src1, jit_arguments *args, pcre_uchar *end1) -{ -/* This function would be ineffective to do in JIT level. */ -pcre_uint32 c1, c2; -const pcre_uchar *src2 = args->uchar_ptr; -const pcre_uchar *end2 = args->end; -const ucd_record *ur; -const pcre_uint32 *pp; - -while (src1 < end1) - { - if (src2 >= end2) - return (pcre_uchar*)1; - GETCHARINC(c1, src1); - GETCHARINC(c2, src2); - ur = GET_UCD(c2); - if (c1 != c2 && c1 != c2 + ur->other_case) - { - pp = PRIV(ucd_caseless_sets) + ur->caseset; - for (;;) - { - if (c1 < *pp) return NULL; - if (c1 == *pp++) break; - } - } - } -return src2; -} - -#endif /* SUPPORT_UTF && SUPPORT_UCP */ - -static pcre_uchar *byte_sequence_compare(compiler_common *common, BOOL caseless, pcre_uchar *cc, - compare_context* context, jump_list **backtracks) -{ -DEFINE_COMPILER; -unsigned int othercasebit = 0; -pcre_uchar *othercasechar = NULL; -#ifdef SUPPORT_UTF -int utflength; -#endif - -if (caseless && char_has_othercase(common, cc)) - { - othercasebit = char_get_othercase_bit(common, cc); - SLJIT_ASSERT(othercasebit); - /* Extracting bit difference info. */ -#if defined COMPILE_PCRE8 - othercasechar = cc + (othercasebit >> 8); - othercasebit &= 0xff; -#elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - /* Note that this code only handles characters in the BMP. If there - ever are characters outside the BMP whose othercase differs in only one - bit from itself (there currently are none), this code will need to be - revised for COMPILE_PCRE32. */ - othercasechar = cc + (othercasebit >> 9); - if ((othercasebit & 0x100) != 0) - othercasebit = (othercasebit & 0xff) << 8; - else - othercasebit &= 0xff; -#endif /* COMPILE_PCRE[8|16|32] */ - } - -if (context->sourcereg == -1) - { -#if defined COMPILE_PCRE8 -#if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED - if (context->length >= 4) - OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); - else if (context->length >= 2) - OP1(SLJIT_MOV_UH, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); - else -#endif - OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); -#elif defined COMPILE_PCRE16 -#if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED - if (context->length >= 4) - OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); - else -#endif - OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); -#elif defined COMPILE_PCRE32 - OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); -#endif /* COMPILE_PCRE[8|16|32] */ - context->sourcereg = TMP2; - } - -#ifdef SUPPORT_UTF -utflength = 1; -if (common->utf && HAS_EXTRALEN(*cc)) - utflength += GET_EXTRALEN(*cc); - -do - { -#endif - - context->length -= IN_UCHARS(1); -#if (defined SLJIT_UNALIGNED && SLJIT_UNALIGNED) && (defined COMPILE_PCRE8 || defined COMPILE_PCRE16) - - /* Unaligned read is supported. */ - if (othercasebit != 0 && othercasechar == cc) - { - context->c.asuchars[context->ucharptr] = *cc | othercasebit; - context->oc.asuchars[context->ucharptr] = othercasebit; - } - else - { - context->c.asuchars[context->ucharptr] = *cc; - context->oc.asuchars[context->ucharptr] = 0; - } - context->ucharptr++; - -#if defined COMPILE_PCRE8 - if (context->ucharptr >= 4 || context->length == 0 || (context->ucharptr == 2 && context->length == 1)) -#else - if (context->ucharptr >= 2 || context->length == 0) -#endif - { - if (context->length >= 4) - OP1(SLJIT_MOV_SI, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); - else if (context->length >= 2) - OP1(SLJIT_MOV_UH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); -#if defined COMPILE_PCRE8 - else if (context->length >= 1) - OP1(SLJIT_MOV_UB, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); -#endif /* COMPILE_PCRE8 */ - context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; - - switch(context->ucharptr) - { - case 4 / sizeof(pcre_uchar): - if (context->oc.asint != 0) - OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asint); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asint | context->oc.asint)); - break; - - case 2 / sizeof(pcre_uchar): - if (context->oc.asushort != 0) - OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asushort); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asushort | context->oc.asushort)); - break; - -#ifdef COMPILE_PCRE8 - case 1: - if (context->oc.asbyte != 0) - OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asbyte); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asbyte | context->oc.asbyte)); - break; -#endif - - default: - SLJIT_ASSERT_STOP(); - break; - } - context->ucharptr = 0; - } - -#else - - /* Unaligned read is unsupported or in 32 bit mode. */ - if (context->length >= 1) - OP1(MOV_UCHAR, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); - - context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; - - if (othercasebit != 0 && othercasechar == cc) - { - OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, othercasebit); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, *cc | othercasebit)); - } - else - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, *cc)); - -#endif - - cc++; -#ifdef SUPPORT_UTF - utflength--; - } -while (utflength > 0); -#endif - -return cc; -} - -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - -#define SET_TYPE_OFFSET(value) \ - if ((value) != typeoffset) \ - { \ - if ((value) > typeoffset) \ - OP2(SLJIT_SUB, typereg, 0, typereg, 0, SLJIT_IMM, (value) - typeoffset); \ - else \ - OP2(SLJIT_ADD, typereg, 0, typereg, 0, SLJIT_IMM, typeoffset - (value)); \ - } \ - typeoffset = (value); - -#define SET_CHAR_OFFSET(value) \ - if ((value) != charoffset) \ - { \ - if ((value) > charoffset) \ - OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, (value) - charoffset); \ - else \ - OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, charoffset - (value)); \ - } \ - charoffset = (value); - -static void compile_xclass_matchingpath(compiler_common *common, pcre_uchar *cc, jump_list **backtracks) -{ -DEFINE_COMPILER; -jump_list *found = NULL; -jump_list **list = (*cc & XCL_NOT) == 0 ? &found : backtracks; -pcre_int32 c, charoffset; -struct sljit_jump *jump = NULL; -pcre_uchar *ccbegin; -int compares, invertcmp, numberofcmps; - -#ifdef SUPPORT_UCP -BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE; -BOOL charsaved = FALSE; -int typereg = TMP1, scriptreg = TMP1; -const pcre_uint32 *other_cases; -pcre_int32 typeoffset; -#endif - -/* Although SUPPORT_UTF must be defined, we are - not necessary in utf mode even in 8 bit mode. */ -detect_partial_match(common, backtracks); -read_char(common); - -if ((*cc++ & XCL_MAP) != 0) - { - OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); -#ifndef COMPILE_PCRE8 - jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); -#elif defined SUPPORT_UTF - if (common->utf) - jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); -#endif - - if (!check_class_ranges(common, (const pcre_uint8 *)cc, TRUE, list)) - { - OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); - OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); - OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc); - OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); - OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); - add_jump(compiler, list, JUMP(SLJIT_C_NOT_ZERO)); - } - -#ifndef COMPILE_PCRE8 - JUMPHERE(jump); -#elif defined SUPPORT_UTF - if (common->utf) - JUMPHERE(jump); -#endif - OP1(SLJIT_MOV, TMP1, 0, TMP3, 0); -#ifdef SUPPORT_UCP - charsaved = TRUE; -#endif - cc += 32 / sizeof(pcre_uchar); - } - -/* Scanning the necessary info. */ -ccbegin = cc; -compares = 0; -while (*cc != XCL_END) - { - compares++; - if (*cc == XCL_SINGLE) - { - cc += 2; -#ifdef SUPPORT_UTF - if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); -#endif -#ifdef SUPPORT_UCP - needschar = TRUE; -#endif - } - else if (*cc == XCL_RANGE) - { - cc += 2; -#ifdef SUPPORT_UTF - if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); -#endif - cc++; -#ifdef SUPPORT_UTF - if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); -#endif -#ifdef SUPPORT_UCP - needschar = TRUE; -#endif - } -#ifdef SUPPORT_UCP - else - { - SLJIT_ASSERT(*cc == XCL_PROP || *cc == XCL_NOTPROP); - cc++; - switch(*cc) - { - case PT_ANY: - break; - - case PT_LAMP: - case PT_GC: - case PT_PC: - case PT_ALNUM: - needstype = TRUE; - break; - - case PT_SC: - needsscript = TRUE; - break; - - case PT_SPACE: - case PT_PXSPACE: - case PT_WORD: - case PT_PXGRAPH: - case PT_PXPRINT: - case PT_PXPUNCT: - needstype = TRUE; - needschar = TRUE; - break; - - case PT_CLIST: - case PT_UCNC: - needschar = TRUE; - break; - - default: - SLJIT_ASSERT_STOP(); - break; - } - cc += 2; - } -#endif - } - -#ifdef SUPPORT_UCP -/* Simple register allocation. TMP1 is preferred if possible. */ -if (needstype || needsscript) - { - if (needschar && !charsaved) - OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); - add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); - if (needschar) - { - if (needstype) - { - OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP1, 0); - typereg = RETURN_ADDR; - } - - if (needsscript) - scriptreg = TMP3; - OP1(SLJIT_MOV, TMP1, 0, TMP3, 0); - } - else if (needstype && needsscript) - scriptreg = TMP3; - /* In all other cases only one of them was specified, and that can goes to TMP1. */ - - if (needsscript) - { - if (scriptreg == TMP1) - { - OP1(SLJIT_MOV, scriptreg, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, script)); - OP1(SLJIT_MOV_UB, scriptreg, 0, SLJIT_MEM2(scriptreg, TMP2), 3); - } - else - { - OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 3); - OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, script)); - OP1(SLJIT_MOV_UB, scriptreg, 0, SLJIT_MEM1(TMP2), 0); - } - } - } -#endif - -/* Generating code. */ -cc = ccbegin; -charoffset = 0; -numberofcmps = 0; -#ifdef SUPPORT_UCP -typeoffset = 0; -#endif - -while (*cc != XCL_END) - { - compares--; - invertcmp = (compares == 0 && list != backtracks); - jump = NULL; - - if (*cc == XCL_SINGLE) - { - cc ++; -#ifdef SUPPORT_UTF - if (common->utf) - { - GETCHARINC(c, cc); - } - else -#endif - c = *cc++; - - if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE)) - { - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset); - OP_FLAGS(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, numberofcmps == 0 ? SLJIT_UNUSED : TMP2, 0, SLJIT_C_EQUAL); - numberofcmps++; - } - else if (numberofcmps > 0) - { - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset); - OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); - numberofcmps = 0; - } - else - { - jump = CMP(SLJIT_C_EQUAL ^ invertcmp, TMP1, 0, SLJIT_IMM, c - charoffset); - numberofcmps = 0; - } - } - else if (*cc == XCL_RANGE) - { - cc ++; -#ifdef SUPPORT_UTF - if (common->utf) - { - GETCHARINC(c, cc); - } - else -#endif - c = *cc++; - SET_CHAR_OFFSET(c); -#ifdef SUPPORT_UTF - if (common->utf) - { - GETCHARINC(c, cc); - } - else -#endif - c = *cc++; - if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE)) - { - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset); - OP_FLAGS(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, numberofcmps == 0 ? SLJIT_UNUSED : TMP2, 0, SLJIT_C_LESS_EQUAL); - numberofcmps++; - } - else if (numberofcmps > 0) - { - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset); - OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); - jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); - numberofcmps = 0; - } - else - { - jump = CMP(SLJIT_C_LESS_EQUAL ^ invertcmp, TMP1, 0, SLJIT_IMM, c - charoffset); - numberofcmps = 0; - } - } -#ifdef SUPPORT_UCP - else - { - if (*cc == XCL_NOTPROP) - invertcmp ^= 0x1; - cc++; - switch(*cc) - { - case PT_ANY: - if (list != backtracks) - { - if ((cc[-1] == XCL_NOTPROP && compares > 0) || (cc[-1] == XCL_PROP && compares == 0)) - continue; - } - else if (cc[-1] == XCL_NOTPROP) - continue; - jump = JUMP(SLJIT_JUMP); - break; - - case PT_LAMP: - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lu - typeoffset); - OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Ll - typeoffset); - OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lt - typeoffset); - OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); - break; - - case PT_GC: - c = PRIV(ucp_typerange)[(int)cc[1] * 2]; - SET_TYPE_OFFSET(c); - jump = CMP(SLJIT_C_LESS_EQUAL ^ invertcmp, typereg, 0, SLJIT_IMM, PRIV(ucp_typerange)[(int)cc[1] * 2 + 1] - c); - break; - - case PT_PC: - jump = CMP(SLJIT_C_EQUAL ^ invertcmp, typereg, 0, SLJIT_IMM, (int)cc[1] - typeoffset); - break; - - case PT_SC: - jump = CMP(SLJIT_C_EQUAL ^ invertcmp, scriptreg, 0, SLJIT_IMM, (int)cc[1]); - break; - - case PT_SPACE: - case PT_PXSPACE: - SET_CHAR_OFFSET(9); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd - 0x9); - OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); - - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x9); - OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e - 0x9); - OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - - SET_TYPE_OFFSET(ucp_Zl); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Zl); - OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); - jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); - break; - - case PT_WORD: - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE - charoffset); - OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); - /* Fall through. */ - - case PT_ALNUM: - SET_TYPE_OFFSET(ucp_Ll); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); - OP_FLAGS((*cc == PT_ALNUM) ? SLJIT_MOV : SLJIT_OR, TMP2, 0, (*cc == PT_ALNUM) ? SLJIT_UNUSED : TMP2, 0, SLJIT_C_LESS_EQUAL); - SET_TYPE_OFFSET(ucp_Nd); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_No - ucp_Nd); - OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); - jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); - break; - - case PT_CLIST: - other_cases = PRIV(ucd_caseless_sets) + cc[1]; - - /* At least three characters are required. - Otherwise this case would be handled by the normal code path. */ - SLJIT_ASSERT(other_cases[0] != NOTACHAR && other_cases[1] != NOTACHAR && other_cases[2] != NOTACHAR); - SLJIT_ASSERT(other_cases[0] < other_cases[1] && other_cases[1] < other_cases[2]); - - /* Optimizing character pairs, if their difference is power of 2. */ - if (is_powerof2(other_cases[1] ^ other_cases[0])) - { - if (charoffset == 0) - OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, other_cases[1] ^ other_cases[0]); - else - { - OP2(SLJIT_ADD, TMP2, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)charoffset); - OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, other_cases[1] ^ other_cases[0]); - } - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, other_cases[1]); - OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); - other_cases += 2; - } - else if (is_powerof2(other_cases[2] ^ other_cases[1])) - { - if (charoffset == 0) - OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, other_cases[2] ^ other_cases[1]); - else - { - OP2(SLJIT_ADD, TMP2, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)charoffset); - OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, other_cases[1] ^ other_cases[0]); - } - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, other_cases[2]); - OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); - - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, other_cases[0] - charoffset); - OP_FLAGS(SLJIT_OR | ((other_cases[3] == NOTACHAR) ? SLJIT_SET_E : 0), TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - - other_cases += 3; - } - else - { - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, *other_cases++ - charoffset); - OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); - } - - while (*other_cases != NOTACHAR) - { - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, *other_cases++ - charoffset); - OP_FLAGS(SLJIT_OR | ((*other_cases == NOTACHAR) ? SLJIT_SET_E : 0), TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - } - jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); - break; - - case PT_UCNC: - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_DOLLAR_SIGN - charoffset); - OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_COMMERCIAL_AT - charoffset); - OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_GRAVE_ACCENT - charoffset); - OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - - SET_CHAR_OFFSET(0xa0); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd7ff - charoffset); - OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); - SET_CHAR_OFFSET(0); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xe000 - 0); - OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_GREATER_EQUAL); - jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); - break; - - case PT_PXGRAPH: - /* C and Z groups are the farthest two groups. */ - SET_TYPE_OFFSET(ucp_Ll); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Ll); - OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER); - - jump = CMP(SLJIT_C_NOT_EQUAL, typereg, 0, SLJIT_IMM, ucp_Cf - ucp_Ll); - - /* In case of ucp_Cf, we overwrite the result. */ - SET_CHAR_OFFSET(0x2066); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2069 - 0x2066); - OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); - - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x061c - 0x2066); - OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e - 0x2066); - OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - - JUMPHERE(jump); - jump = CMP(SLJIT_C_ZERO ^ invertcmp, TMP2, 0, SLJIT_IMM, 0); - break; - - case PT_PXPRINT: - /* C and Z groups are the farthest two groups. */ - SET_TYPE_OFFSET(ucp_Ll); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Ll); - OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER); - - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Ll); - OP_FLAGS(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_C_NOT_EQUAL); - - jump = CMP(SLJIT_C_NOT_EQUAL, typereg, 0, SLJIT_IMM, ucp_Cf - ucp_Ll); - - /* In case of ucp_Cf, we overwrite the result. */ - SET_CHAR_OFFSET(0x2066); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2069 - 0x2066); - OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); - - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x061c - 0x2066); - OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - - JUMPHERE(jump); - jump = CMP(SLJIT_C_ZERO ^ invertcmp, TMP2, 0, SLJIT_IMM, 0); - break; - - case PT_PXPUNCT: - SET_TYPE_OFFSET(ucp_Sc); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Sc); - OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL); - - SET_CHAR_OFFSET(0); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xff); - OP_FLAGS(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); - - SET_TYPE_OFFSET(ucp_Pc); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Ps - ucp_Pc); - OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL); - jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); - break; - } - cc += 2; - } -#endif - - if (jump != NULL) - add_jump(compiler, compares > 0 ? list : backtracks, jump); - } - -if (found != NULL) - set_jumps(found, LABEL()); -} - -#undef SET_TYPE_OFFSET -#undef SET_CHAR_OFFSET - -#endif - -static pcre_uchar *compile_char1_matchingpath(compiler_common *common, pcre_uchar type, pcre_uchar *cc, jump_list **backtracks) -{ -DEFINE_COMPILER; -int length; -unsigned int c, oc, bit; -compare_context context; -struct sljit_jump *jump[4]; -jump_list *end_list; -#ifdef SUPPORT_UTF -struct sljit_label *label; -#ifdef SUPPORT_UCP -pcre_uchar propdata[5]; -#endif -#endif - -switch(type) - { - case OP_SOD: - OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, TMP1, 0)); - return cc; - - case OP_SOM: - OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, TMP1, 0)); - return cc; - - case OP_NOT_WORD_BOUNDARY: - case OP_WORD_BOUNDARY: - add_jump(compiler, &common->wordboundary, JUMP(SLJIT_FAST_CALL)); - add_jump(compiler, backtracks, JUMP(type == OP_NOT_WORD_BOUNDARY ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); - return cc; - - case OP_NOT_DIGIT: - case OP_DIGIT: - /* Digits are usually 0-9, so it is worth to optimize them. */ - if (common->digits[0] == -2) - get_ctype_ranges(common, ctype_digit, common->digits); - detect_partial_match(common, backtracks); - /* Flip the starting bit in the negative case. */ - if (type == OP_NOT_DIGIT) - common->digits[1] ^= 1; - if (!check_ranges(common, common->digits, backtracks, TRUE)) - { - read_char8_type(common); - OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_digit); - add_jump(compiler, backtracks, JUMP(type == OP_DIGIT ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); - } - if (type == OP_NOT_DIGIT) - common->digits[1] ^= 1; - return cc; - - case OP_NOT_WHITESPACE: - case OP_WHITESPACE: - detect_partial_match(common, backtracks); - read_char8_type(common); - OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_space); - add_jump(compiler, backtracks, JUMP(type == OP_WHITESPACE ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); - return cc; - - case OP_NOT_WORDCHAR: - case OP_WORDCHAR: - detect_partial_match(common, backtracks); - read_char8_type(common); - OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_word); - add_jump(compiler, backtracks, JUMP(type == OP_WORDCHAR ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); - return cc; - - case OP_ANY: - detect_partial_match(common, backtracks); - read_char(common); - if (common->nltype == NLTYPE_FIXED && common->newline > 255) - { - jump[0] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff); - end_list = NULL; - if (common->mode != JIT_PARTIAL_HARD_COMPILE) - add_jump(compiler, &end_list, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); - else - check_str_end(common, &end_list); - - OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); - add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, common->newline & 0xff)); - set_jumps(end_list, LABEL()); - JUMPHERE(jump[0]); - } - else - check_newlinechar(common, common->nltype, backtracks, TRUE); - return cc; - - case OP_ALLANY: - detect_partial_match(common, backtracks); -#ifdef SUPPORT_UTF - if (common->utf) - { - OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); -#if defined COMPILE_PCRE8 || defined COMPILE_PCRE16 -#if defined COMPILE_PCRE8 - jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); - OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0); - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); -#elif defined COMPILE_PCRE16 - jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); - OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); - OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); - OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); -#endif - JUMPHERE(jump[0]); -#endif /* COMPILE_PCRE[8|16] */ - return cc; - } -#endif - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); - return cc; - - case OP_ANYBYTE: - detect_partial_match(common, backtracks); - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); - return cc; - -#ifdef SUPPORT_UTF -#ifdef SUPPORT_UCP - case OP_NOTPROP: - case OP_PROP: - propdata[0] = 0; - propdata[1] = type == OP_NOTPROP ? XCL_NOTPROP : XCL_PROP; - propdata[2] = cc[0]; - propdata[3] = cc[1]; - propdata[4] = XCL_END; - compile_xclass_matchingpath(common, propdata, backtracks); - return cc + 2; -#endif -#endif - - case OP_ANYNL: - detect_partial_match(common, backtracks); - read_char(common); - jump[0] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR); - /* We don't need to handle soft partial matching case. */ - end_list = NULL; - if (common->mode != JIT_PARTIAL_HARD_COMPILE) - add_jump(compiler, &end_list, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); - else - check_str_end(common, &end_list); - OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); - jump[1] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL); - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); - jump[2] = JUMP(SLJIT_JUMP); - JUMPHERE(jump[0]); - check_newlinechar(common, common->bsr_nltype, backtracks, FALSE); - set_jumps(end_list, LABEL()); - JUMPHERE(jump[1]); - JUMPHERE(jump[2]); - return cc; - - case OP_NOT_HSPACE: - case OP_HSPACE: - detect_partial_match(common, backtracks); - read_char(common); - add_jump(compiler, &common->hspace, JUMP(SLJIT_FAST_CALL)); - add_jump(compiler, backtracks, JUMP(type == OP_NOT_HSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); - return cc; - - case OP_NOT_VSPACE: - case OP_VSPACE: - detect_partial_match(common, backtracks); - read_char(common); - add_jump(compiler, &common->vspace, JUMP(SLJIT_FAST_CALL)); - add_jump(compiler, backtracks, JUMP(type == OP_NOT_VSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); - return cc; - -#ifdef SUPPORT_UCP - case OP_EXTUNI: - detect_partial_match(common, backtracks); - read_char(common); - add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, gbprop)); - /* Optimize register allocation: use a real register. */ - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, STACK_TOP, 0); - OP1(SLJIT_MOV_UB, STACK_TOP, 0, SLJIT_MEM2(TMP1, TMP2), 3); - - label = LABEL(); - jump[0] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); - OP1(SLJIT_MOV, TMP3, 0, STR_PTR, 0); - read_char(common); - add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, gbprop)); - OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM2(TMP1, TMP2), 3); - - OP2(SLJIT_SHL, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, 2); - OP1(SLJIT_MOV_UI, TMP1, 0, SLJIT_MEM1(STACK_TOP), (sljit_sw)PRIV(ucp_gbtable)); - OP1(SLJIT_MOV, STACK_TOP, 0, TMP2, 0); - OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); - OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); - JUMPTO(SLJIT_C_NOT_ZERO, label); - - OP1(SLJIT_MOV, STR_PTR, 0, TMP3, 0); - JUMPHERE(jump[0]); - OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); - - if (common->mode == JIT_PARTIAL_HARD_COMPILE) - { - jump[0] = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); - /* Since we successfully read a char above, partial matching must occure. */ - check_partial(common, TRUE); - JUMPHERE(jump[0]); - } - return cc; -#endif - - case OP_EODN: - /* Requires rather complex checks. */ - jump[0] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); - if (common->nltype == NLTYPE_FIXED && common->newline > 255) - { - OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); - OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); - if (common->mode == JIT_COMPILE) - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_END, 0)); - else - { - jump[1] = CMP(SLJIT_C_EQUAL, TMP2, 0, STR_END, 0); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP2, 0, STR_END, 0); - OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff); - OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_NOT_EQUAL); - add_jump(compiler, backtracks, JUMP(SLJIT_C_NOT_EQUAL)); - check_partial(common, TRUE); - add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); - JUMPHERE(jump[1]); - } - OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff)); - } - else if (common->nltype == NLTYPE_FIXED) - { - OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); - OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_END, 0)); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline)); - } - else - { - OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); - jump[1] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR); - OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); - OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP2, 0, STR_END, 0); - jump[2] = JUMP(SLJIT_C_GREATER); - add_jump(compiler, backtracks, JUMP(SLJIT_C_LESS)); - /* Equal. */ - OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); - jump[3] = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL); - add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); - - JUMPHERE(jump[1]); - if (common->nltype == NLTYPE_ANYCRLF) - { - OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); - add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP2, 0, STR_END, 0)); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL)); - } - else - { - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, STR_PTR, 0); - read_char(common); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, STR_END, 0)); - add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL)); - add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO)); - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); - } - JUMPHERE(jump[2]); - JUMPHERE(jump[3]); - } - JUMPHERE(jump[0]); - check_partial(common, FALSE); - return cc; - - case OP_EOD: - add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0)); - check_partial(common, FALSE); - return cc; - - case OP_CIRC: - OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, begin)); - add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER, STR_PTR, 0, TMP1, 0)); - OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, notbol)); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); - return cc; - - case OP_CIRCM: - OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, begin)); - jump[1] = CMP(SLJIT_C_GREATER, STR_PTR, 0, TMP1, 0); - OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, notbol)); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); - jump[0] = JUMP(SLJIT_JUMP); - JUMPHERE(jump[1]); - - add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); - if (common->nltype == NLTYPE_FIXED && common->newline > 255) - { - OP2(SLJIT_SUB, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); - add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP2, 0, TMP1, 0)); - OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); - OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff)); - } - else - { - skip_char_back(common); - read_char(common); - check_newlinechar(common, common->nltype, backtracks, FALSE); - } - JUMPHERE(jump[0]); - return cc; - - case OP_DOLL: - OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); - OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, noteol)); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); - - if (!common->endonly) - compile_char1_matchingpath(common, OP_EODN, cc, backtracks); - else - { - add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0)); - check_partial(common, FALSE); - } - return cc; - - case OP_DOLLM: - jump[1] = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); - OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); - OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, noteol)); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); - check_partial(common, FALSE); - jump[0] = JUMP(SLJIT_JUMP); - JUMPHERE(jump[1]); - - if (common->nltype == NLTYPE_FIXED && common->newline > 255) - { - OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); - OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); - if (common->mode == JIT_COMPILE) - add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER, TMP2, 0, STR_END, 0)); - else - { - jump[1] = CMP(SLJIT_C_LESS_EQUAL, TMP2, 0, STR_END, 0); - /* STR_PTR = STR_END - IN_UCHARS(1) */ - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); - check_partial(common, TRUE); - add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); - JUMPHERE(jump[1]); - } - - OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff)); - } - else - { - peek_char(common); - check_newlinechar(common, common->nltype, backtracks, FALSE); - } - JUMPHERE(jump[0]); - return cc; - - case OP_CHAR: - case OP_CHARI: - length = 1; -#ifdef SUPPORT_UTF - if (common->utf && HAS_EXTRALEN(*cc)) length += GET_EXTRALEN(*cc); -#endif - if (common->mode == JIT_COMPILE && (type == OP_CHAR || !char_has_othercase(common, cc) || char_get_othercase_bit(common, cc) != 0)) - { - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(length)); - add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER, STR_PTR, 0, STR_END, 0)); - - context.length = IN_UCHARS(length); - context.sourcereg = -1; -#if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED - context.ucharptr = 0; -#endif - return byte_sequence_compare(common, type == OP_CHARI, cc, &context, backtracks); - } - detect_partial_match(common, backtracks); - read_char(common); -#ifdef SUPPORT_UTF - if (common->utf) - { - GETCHAR(c, cc); - } - else -#endif - c = *cc; - if (type == OP_CHAR || !char_has_othercase(common, cc)) - { - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c)); - return cc + length; - } - oc = char_othercase(common, c); - bit = c ^ oc; - if (is_powerof2(bit)) - { - OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, bit); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c | bit)); - return cc + length; - } - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c); - OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, oc); - OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL); - add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO)); - return cc + length; - - case OP_NOT: - case OP_NOTI: - detect_partial_match(common, backtracks); - length = 1; -#ifdef SUPPORT_UTF - if (common->utf) - { -#ifdef COMPILE_PCRE8 - c = *cc; - if (c < 128) - { - OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); - if (type == OP_NOT || !char_has_othercase(common, cc)) - add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c)); - else - { - /* Since UTF8 code page is fixed, we know that c is in [a-z] or [A-Z] range. */ - OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x20); - add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, c | 0x20)); - } - /* Skip the variable-length character. */ - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); - jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); - OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0); - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); - JUMPHERE(jump[0]); - return cc + 1; - } - else -#endif /* COMPILE_PCRE8 */ - { - GETCHARLEN(c, cc, length); - read_char(common); - } - } - else -#endif /* SUPPORT_UTF */ - { - read_char(common); - c = *cc; - } - - if (type == OP_NOT || !char_has_othercase(common, cc)) - add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c)); - else - { - oc = char_othercase(common, c); - bit = c ^ oc; - if (is_powerof2(bit)) - { - OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, bit); - add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c | bit)); - } - else - { - add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c)); - add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, oc)); - } - } - return cc + length; - - case OP_CLASS: - case OP_NCLASS: - detect_partial_match(common, backtracks); - read_char(common); - if (check_class_ranges(common, (const pcre_uint8 *)cc, type == OP_NCLASS, backtracks)) - return cc + 32 / sizeof(pcre_uchar); - -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - jump[0] = NULL; -#ifdef COMPILE_PCRE8 - /* This check only affects 8 bit mode. In other modes, we - always need to compare the value with 255. */ - if (common->utf) -#endif /* COMPILE_PCRE8 */ - { - jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); - if (type == OP_CLASS) - { - add_jump(compiler, backtracks, jump[0]); - jump[0] = NULL; - } - } -#endif /* SUPPORT_UTF || !COMPILE_PCRE8 */ - OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); - OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); - OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc); - OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); - OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); - add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO)); -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - if (jump[0] != NULL) - JUMPHERE(jump[0]); -#endif /* SUPPORT_UTF || !COMPILE_PCRE8 */ - return cc + 32 / sizeof(pcre_uchar); - -#if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - case OP_XCLASS: - compile_xclass_matchingpath(common, cc + LINK_SIZE, backtracks); - return cc + GET(cc, 0) - 1; -#endif - - case OP_REVERSE: - length = GET(cc, 0); - if (length == 0) - return cc + LINK_SIZE; - OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); -#ifdef SUPPORT_UTF - if (common->utf) - { - OP1(SLJIT_MOV, TMP3, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); - OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, length); - label = LABEL(); - add_jump(compiler, backtracks, CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP3, 0)); - skip_char_back(common); - OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); - JUMPTO(SLJIT_C_NOT_ZERO, label); - } - else -#endif - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); - OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(length)); - add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, STR_PTR, 0, TMP1, 0)); - } - check_start_used_ptr(common); - return cc + LINK_SIZE; - } -SLJIT_ASSERT_STOP(); -return cc; -} - -static SLJIT_INLINE pcre_uchar *compile_charn_matchingpath(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, jump_list **backtracks) -{ -/* This function consumes at least one input character. */ -/* To decrease the number of length checks, we try to concatenate the fixed length character sequences. */ -DEFINE_COMPILER; -pcre_uchar *ccbegin = cc; -compare_context context; -int size; - -context.length = 0; -do - { - if (cc >= ccend) - break; - - if (*cc == OP_CHAR) - { - size = 1; -#ifdef SUPPORT_UTF - if (common->utf && HAS_EXTRALEN(cc[1])) - size += GET_EXTRALEN(cc[1]); -#endif - } - else if (*cc == OP_CHARI) - { - size = 1; -#ifdef SUPPORT_UTF - if (common->utf) - { - if (char_has_othercase(common, cc + 1) && char_get_othercase_bit(common, cc + 1) == 0) - size = 0; - else if (HAS_EXTRALEN(cc[1])) - size += GET_EXTRALEN(cc[1]); - } - else -#endif - if (char_has_othercase(common, cc + 1) && char_get_othercase_bit(common, cc + 1) == 0) - size = 0; - } - else - size = 0; - - cc += 1 + size; - context.length += IN_UCHARS(size); - } -while (size > 0 && context.length <= 128); - -cc = ccbegin; -if (context.length > 0) - { - /* We have a fixed-length byte sequence. */ - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, context.length); - add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER, STR_PTR, 0, STR_END, 0)); - - context.sourcereg = -1; -#if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED - context.ucharptr = 0; -#endif - do cc = byte_sequence_compare(common, *cc == OP_CHARI, cc + 1, &context, backtracks); while (context.length > 0); - return cc; - } - -/* A non-fixed length character will be checked if length == 0. */ -return compile_char1_matchingpath(common, *cc, cc + 1, backtracks); -} - -/* Forward definitions. */ -static void compile_matchingpath(compiler_common *, pcre_uchar *, pcre_uchar *, backtrack_common *); -static void compile_backtrackingpath(compiler_common *, struct backtrack_common *); - -#define PUSH_BACKTRACK(size, ccstart, error) \ - do \ - { \ - backtrack = sljit_alloc_memory(compiler, (size)); \ - if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) \ - return error; \ - memset(backtrack, 0, size); \ - backtrack->prev = parent->top; \ - backtrack->cc = (ccstart); \ - parent->top = backtrack; \ - } \ - while (0) - -#define PUSH_BACKTRACK_NOVALUE(size, ccstart) \ - do \ - { \ - backtrack = sljit_alloc_memory(compiler, (size)); \ - if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) \ - return; \ - memset(backtrack, 0, size); \ - backtrack->prev = parent->top; \ - backtrack->cc = (ccstart); \ - parent->top = backtrack; \ - } \ - while (0) - -#define BACKTRACK_AS(type) ((type *)backtrack) - -static void compile_dnref_search(compiler_common *common, pcre_uchar *cc, jump_list **backtracks) -{ -/* The OVECTOR offset goes to TMP2. */ -DEFINE_COMPILER; -int count = GET2(cc, 1 + IMM2_SIZE); -pcre_uchar *slot = common->name_table + GET2(cc, 1) * common->name_entry_size; -unsigned int offset; -jump_list *found = NULL; - -SLJIT_ASSERT(*cc == OP_DNREF || *cc == OP_DNREFI); - -OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)); - -count--; -while (count-- > 0) - { - offset = GET2(slot, 0) << 1; - GET_LOCAL_BASE(TMP2, 0, OVECTOR(offset)); - add_jump(compiler, &found, CMP(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0)); - slot += common->name_entry_size; - } - -offset = GET2(slot, 0) << 1; -GET_LOCAL_BASE(TMP2, 0, OVECTOR(offset)); -if (backtracks != NULL && !common->jscript_compat) - add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0)); - -set_jumps(found, LABEL()); -} - -static void compile_ref_matchingpath(compiler_common *common, pcre_uchar *cc, jump_list **backtracks, BOOL withchecks, BOOL emptyfail) -{ -DEFINE_COMPILER; -BOOL ref = (*cc == OP_REF || *cc == OP_REFI); -int offset = 0; -struct sljit_jump *jump = NULL; -struct sljit_jump *partial; -struct sljit_jump *nopartial; - -if (ref) - { - offset = GET2(cc, 1) << 1; - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); - /* OVECTOR(1) contains the "string begin - 1" constant. */ - if (withchecks && !common->jscript_compat) - add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1))); - } -else - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0); - -#if defined SUPPORT_UTF && defined SUPPORT_UCP -if (common->utf && *cc == OP_REFI) - { - SLJIT_ASSERT(TMP1 == SLJIT_SCRATCH_REG1 && STACK_TOP == SLJIT_SCRATCH_REG2 && TMP2 == SLJIT_SCRATCH_REG3); - if (ref) - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); - else - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw)); - - if (withchecks) - jump = CMP(SLJIT_C_EQUAL, TMP1, 0, TMP2, 0); - - /* Needed to save important temporary registers. */ - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, STACK_TOP, 0); - OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, ARGUMENTS, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, uchar_ptr), STR_PTR, 0); - sljit_emit_ijump(compiler, SLJIT_CALL3, SLJIT_IMM, SLJIT_FUNC_OFFSET(do_utf_caselesscmp)); - OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); - if (common->mode == JIT_COMPILE) - add_jump(compiler, backtracks, CMP(SLJIT_C_LESS_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1)); - else - { - add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 0)); - nopartial = CMP(SLJIT_C_NOT_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1); - check_partial(common, FALSE); - add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); - JUMPHERE(nopartial); - } - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_RETURN_REG, 0); - } -else -#endif /* SUPPORT_UTF && SUPPORT_UCP */ - { - if (ref) - OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), TMP1, 0); - else - OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw), TMP1, 0); - - if (withchecks) - jump = JUMP(SLJIT_C_ZERO); - - OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); - partial = CMP(SLJIT_C_GREATER, STR_PTR, 0, STR_END, 0); - if (common->mode == JIT_COMPILE) - add_jump(compiler, backtracks, partial); - - add_jump(compiler, *cc == OP_REF ? &common->casefulcmp : &common->caselesscmp, JUMP(SLJIT_FAST_CALL)); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); - - if (common->mode != JIT_COMPILE) - { - nopartial = JUMP(SLJIT_JUMP); - JUMPHERE(partial); - /* TMP2 -= STR_END - STR_PTR */ - OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, STR_PTR, 0); - OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, STR_END, 0); - partial = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, 0); - OP1(SLJIT_MOV, STR_PTR, 0, STR_END, 0); - add_jump(compiler, *cc == OP_REF ? &common->casefulcmp : &common->caselesscmp, JUMP(SLJIT_FAST_CALL)); - add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); - JUMPHERE(partial); - check_partial(common, FALSE); - add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); - JUMPHERE(nopartial); - } - } - -if (jump != NULL) - { - if (emptyfail) - add_jump(compiler, backtracks, jump); - else - JUMPHERE(jump); - } -} - -static SLJIT_INLINE pcre_uchar *compile_ref_iterator_matchingpath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent) -{ -DEFINE_COMPILER; -BOOL ref = (*cc == OP_REF || *cc == OP_REFI); -backtrack_common *backtrack; -pcre_uchar type; -int offset = 0; -struct sljit_label *label; -struct sljit_jump *zerolength; -struct sljit_jump *jump = NULL; -pcre_uchar *ccbegin = cc; -int min = 0, max = 0; -BOOL minimize; - -PUSH_BACKTRACK(sizeof(iterator_backtrack), cc, NULL); - -if (ref) - offset = GET2(cc, 1) << 1; -else - cc += IMM2_SIZE; -type = cc[1 + IMM2_SIZE]; - -SLJIT_COMPILE_ASSERT((OP_CRSTAR & 0x1) == 0, crstar_opcode_must_be_even); -minimize = (type & 0x1) != 0; -switch(type) - { - case OP_CRSTAR: - case OP_CRMINSTAR: - min = 0; - max = 0; - cc += 1 + IMM2_SIZE + 1; - break; - case OP_CRPLUS: - case OP_CRMINPLUS: - min = 1; - max = 0; - cc += 1 + IMM2_SIZE + 1; - break; - case OP_CRQUERY: - case OP_CRMINQUERY: - min = 0; - max = 1; - cc += 1 + IMM2_SIZE + 1; - break; - case OP_CRRANGE: - case OP_CRMINRANGE: - min = GET2(cc, 1 + IMM2_SIZE + 1); - max = GET2(cc, 1 + IMM2_SIZE + 1 + IMM2_SIZE); - cc += 1 + IMM2_SIZE + 1 + 2 * IMM2_SIZE; - break; - default: - SLJIT_ASSERT_STOP(); - break; - } - -if (!minimize) - { - if (min == 0) - { - allocate_stack(common, 2); - if (ref) - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0); - /* Temporary release of STR_PTR. */ - OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_sw)); - /* Handles both invalid and empty cases. Since the minimum repeat, - is zero the invalid case is basically the same as an empty case. */ - if (ref) - zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); - else - { - compile_dnref_search(common, ccbegin, NULL); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, TMP2, 0); - zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw)); - } - /* Restore if not zero length. */ - OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_sw)); - } - else - { - allocate_stack(common, 1); - if (ref) - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); - if (ref) - { - add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1))); - zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); - } - else - { - compile_dnref_search(common, ccbegin, &backtrack->topbacktracks); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, TMP2, 0); - zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw)); - } - } - - if (min > 1 || max > 1) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, SLJIT_IMM, 0); - - label = LABEL(); - if (!ref) - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1); - compile_ref_matchingpath(common, ccbegin, &backtrack->topbacktracks, FALSE, FALSE); - - if (min > 1 || max > 1) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); - OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, TMP1, 0); - if (min > 1) - CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, min, label); - if (max > 1) - { - jump = CMP(SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, max); - allocate_stack(common, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); - JUMPTO(SLJIT_JUMP, label); - JUMPHERE(jump); - } - } - - if (max == 0) - { - /* Includes min > 1 case as well. */ - allocate_stack(common, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); - JUMPTO(SLJIT_JUMP, label); - } - - JUMPHERE(zerolength); - BACKTRACK_AS(iterator_backtrack)->matchingpath = LABEL(); - - count_match(common); - return cc; - } - -allocate_stack(common, ref ? 2 : 3); -if (ref) - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); -OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); -if (type != OP_CRMINSTAR) - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0); - -if (min == 0) - { - /* Handles both invalid and empty cases. Since the minimum repeat, - is zero the invalid case is basically the same as an empty case. */ - if (ref) - zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); - else - { - compile_dnref_search(common, ccbegin, NULL); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), TMP2, 0); - zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw)); - } - /* Length is non-zero, we can match real repeats. */ - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); - jump = JUMP(SLJIT_JUMP); - } -else - { - if (ref) - { - add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1))); - zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); - } - else - { - compile_dnref_search(common, ccbegin, &backtrack->topbacktracks); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), TMP2, 0); - zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw)); - } - } - -BACKTRACK_AS(iterator_backtrack)->matchingpath = LABEL(); -if (max > 0) - add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, max)); - -if (!ref) - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(2)); -compile_ref_matchingpath(common, ccbegin, &backtrack->topbacktracks, TRUE, TRUE); -OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); - -if (min > 1) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); - OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0); - CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, min, BACKTRACK_AS(iterator_backtrack)->matchingpath); - } -else if (max > 0) - OP2(SLJIT_ADD, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 1); - -if (jump != NULL) - JUMPHERE(jump); -JUMPHERE(zerolength); - -count_match(common); -return cc; -} - -static SLJIT_INLINE pcre_uchar *compile_recurse_matchingpath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent) -{ -DEFINE_COMPILER; -backtrack_common *backtrack; -recurse_entry *entry = common->entries; -recurse_entry *prev = NULL; -sljit_sw start = GET(cc, 1); -pcre_uchar *start_cc; -BOOL needs_control_head; - -PUSH_BACKTRACK(sizeof(recurse_backtrack), cc, NULL); - -/* Inlining simple patterns. */ -if (get_framesize(common, common->start + start, NULL, TRUE, &needs_control_head) == no_stack) - { - start_cc = common->start + start; - compile_matchingpath(common, next_opcode(common, start_cc), bracketend(start_cc) - (1 + LINK_SIZE), backtrack); - BACKTRACK_AS(recurse_backtrack)->inlined_pattern = TRUE; - return cc + 1 + LINK_SIZE; - } - -while (entry != NULL) - { - if (entry->start == start) - break; - prev = entry; - entry = entry->next; - } - -if (entry == NULL) - { - entry = sljit_alloc_memory(compiler, sizeof(recurse_entry)); - if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) - return NULL; - entry->next = NULL; - entry->entry = NULL; - entry->calls = NULL; - entry->start = start; - - if (prev != NULL) - prev->next = entry; - else - common->entries = entry; - } - -if (common->has_set_som && common->mark_ptr != 0) - { - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); - allocate_stack(common, 2); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0); - } -else if (common->has_set_som || common->mark_ptr != 0) - { - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->has_set_som ? (int)(OVECTOR(0)) : common->mark_ptr); - allocate_stack(common, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0); - } - -if (entry->entry == NULL) - add_jump(compiler, &entry->calls, JUMP(SLJIT_FAST_CALL)); -else - JUMPTO(SLJIT_FAST_CALL, entry->entry); -/* Leave if the match is failed. */ -add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0)); -return cc + 1 + LINK_SIZE; -} - -static int SLJIT_CALL do_callout(struct jit_arguments* arguments, PUBL(callout_block) *callout_block, pcre_uchar **jit_ovector) -{ -const pcre_uchar *begin = arguments->begin; -int *offset_vector = arguments->offsets; -int offset_count = arguments->offset_count; -int i; - -if (PUBL(callout) == NULL) - return 0; - -callout_block->version = 2; -callout_block->callout_data = arguments->callout_data; - -/* Offsets in subject. */ -callout_block->subject_length = arguments->end - arguments->begin; -callout_block->start_match = (pcre_uchar*)callout_block->subject - arguments->begin; -callout_block->current_position = (pcre_uchar*)callout_block->offset_vector - arguments->begin; -#if defined COMPILE_PCRE8 -callout_block->subject = (PCRE_SPTR)begin; -#elif defined COMPILE_PCRE16 -callout_block->subject = (PCRE_SPTR16)begin; -#elif defined COMPILE_PCRE32 -callout_block->subject = (PCRE_SPTR32)begin; -#endif - -/* Convert and copy the JIT offset vector to the offset_vector array. */ -callout_block->capture_top = 0; -callout_block->offset_vector = offset_vector; -for (i = 2; i < offset_count; i += 2) - { - offset_vector[i] = jit_ovector[i] - begin; - offset_vector[i + 1] = jit_ovector[i + 1] - begin; - if (jit_ovector[i] >= begin) - callout_block->capture_top = i; - } - -callout_block->capture_top = (callout_block->capture_top >> 1) + 1; -if (offset_count > 0) - offset_vector[0] = -1; -if (offset_count > 1) - offset_vector[1] = -1; -return (*PUBL(callout))(callout_block); -} - -/* Aligning to 8 byte. */ -#define CALLOUT_ARG_SIZE \ - (((int)sizeof(PUBL(callout_block)) + 7) & ~7) - -#define CALLOUT_ARG_OFFSET(arg) \ - (-CALLOUT_ARG_SIZE + SLJIT_OFFSETOF(PUBL(callout_block), arg)) - -static SLJIT_INLINE pcre_uchar *compile_callout_matchingpath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent) -{ -DEFINE_COMPILER; -backtrack_common *backtrack; - -PUSH_BACKTRACK(sizeof(backtrack_common), cc, NULL); - -allocate_stack(common, CALLOUT_ARG_SIZE / sizeof(sljit_sw)); - -OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr); -OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); -SLJIT_ASSERT(common->capture_last_ptr != 0); -OP1(SLJIT_MOV_SI, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(callout_number), SLJIT_IMM, cc[1]); -OP1(SLJIT_MOV_SI, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(capture_last), TMP2, 0); - -/* These pointer sized fields temporarly stores internal variables. */ -OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); -OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(offset_vector), STR_PTR, 0); -OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(subject), TMP2, 0); - -if (common->mark_ptr != 0) - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, mark_ptr)); -OP1(SLJIT_MOV_SI, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(pattern_position), SLJIT_IMM, GET(cc, 2)); -OP1(SLJIT_MOV_SI, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(next_item_length), SLJIT_IMM, GET(cc, 2 + LINK_SIZE)); -OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(mark), (common->mark_ptr != 0) ? TMP2 : SLJIT_IMM, 0); - -/* Needed to save important temporary registers. */ -OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, STACK_TOP, 0); -OP2(SLJIT_SUB, SLJIT_SCRATCH_REG2, 0, STACK_TOP, 0, SLJIT_IMM, CALLOUT_ARG_SIZE); -GET_LOCAL_BASE(SLJIT_SCRATCH_REG3, 0, OVECTOR_START); -sljit_emit_ijump(compiler, SLJIT_CALL3, SLJIT_IMM, SLJIT_FUNC_OFFSET(do_callout)); -OP1(SLJIT_MOV_SI, SLJIT_RETURN_REG, 0, SLJIT_RETURN_REG, 0); -OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); -free_stack(common, CALLOUT_ARG_SIZE / sizeof(sljit_sw)); - -/* Check return value. */ -OP2(SLJIT_SUB | SLJIT_SET_S, SLJIT_UNUSED, 0, SLJIT_RETURN_REG, 0, SLJIT_IMM, 0); -add_jump(compiler, &backtrack->topbacktracks, JUMP(SLJIT_C_SIG_GREATER)); -if (common->forced_quit_label == NULL) - add_jump(compiler, &common->forced_quit, JUMP(SLJIT_C_SIG_LESS)); -else - JUMPTO(SLJIT_C_SIG_LESS, common->forced_quit_label); -return cc + 2 + 2 * LINK_SIZE; -} - -#undef CALLOUT_ARG_SIZE -#undef CALLOUT_ARG_OFFSET - -static pcre_uchar *compile_assert_matchingpath(compiler_common *common, pcre_uchar *cc, assert_backtrack *backtrack, BOOL conditional) -{ -DEFINE_COMPILER; -int framesize; -int extrasize; -BOOL needs_control_head; -int private_data_ptr; -backtrack_common altbacktrack; -pcre_uchar *ccbegin; -pcre_uchar opcode; -pcre_uchar bra = OP_BRA; -jump_list *tmp = NULL; -jump_list **target = (conditional) ? &backtrack->condfailed : &backtrack->common.topbacktracks; -jump_list **found; -/* Saving previous accept variables. */ -BOOL save_local_exit = common->local_exit; -BOOL save_positive_assert = common->positive_assert; -then_trap_backtrack *save_then_trap = common->then_trap; -struct sljit_label *save_quit_label = common->quit_label; -struct sljit_label *save_accept_label = common->accept_label; -jump_list *save_quit = common->quit; -jump_list *save_positive_assert_quit = common->positive_assert_quit; -jump_list *save_accept = common->accept; -struct sljit_jump *jump; -struct sljit_jump *brajump = NULL; - -/* Assert captures then. */ -common->then_trap = NULL; - -if (*cc == OP_BRAZERO || *cc == OP_BRAMINZERO) - { - SLJIT_ASSERT(!conditional); - bra = *cc; - cc++; - } -private_data_ptr = PRIVATE_DATA(cc); -SLJIT_ASSERT(private_data_ptr != 0); -framesize = get_framesize(common, cc, NULL, FALSE, &needs_control_head); -backtrack->framesize = framesize; -backtrack->private_data_ptr = private_data_ptr; -opcode = *cc; -SLJIT_ASSERT(opcode >= OP_ASSERT && opcode <= OP_ASSERTBACK_NOT); -found = (opcode == OP_ASSERT || opcode == OP_ASSERTBACK) ? &tmp : target; -ccbegin = cc; -cc += GET(cc, 1); - -if (bra == OP_BRAMINZERO) - { - /* This is a braminzero backtrack path. */ - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - free_stack(common, 1); - brajump = CMP(SLJIT_C_EQUAL, STR_PTR, 0, SLJIT_IMM, 0); - } - -if (framesize < 0) - { - extrasize = needs_control_head ? 2 : 1; - if (framesize == no_frame) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, STACK_TOP, 0); - allocate_stack(common, extrasize); - if (needs_control_head) - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); - if (needs_control_head) - { - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, SLJIT_IMM, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0); - } - } -else - { - extrasize = needs_control_head ? 3 : 2; - allocate_stack(common, framesize + extrasize); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - OP2(SLJIT_SUB, TMP2, 0, STACK_TOP, 0, SLJIT_IMM, (framesize + extrasize) * sizeof(sljit_sw)); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, TMP2, 0); - if (needs_control_head) - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); - if (needs_control_head) - { - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), TMP1, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP2, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, SLJIT_IMM, 0); - } - else - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0); - init_frame(common, ccbegin, NULL, framesize + extrasize - 1, extrasize, FALSE); - } - -memset(&altbacktrack, 0, sizeof(backtrack_common)); -if (opcode == OP_ASSERT_NOT || opcode == OP_ASSERTBACK_NOT) - { - /* Negative assert is stronger than positive assert. */ - common->local_exit = TRUE; - common->quit_label = NULL; - common->quit = NULL; - common->positive_assert = FALSE; - } -else - common->positive_assert = TRUE; -common->positive_assert_quit = NULL; - -while (1) - { - common->accept_label = NULL; - common->accept = NULL; - altbacktrack.top = NULL; - altbacktrack.topbacktracks = NULL; - - if (*ccbegin == OP_ALT) - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - - altbacktrack.cc = ccbegin; - compile_matchingpath(common, ccbegin + 1 + LINK_SIZE, cc, &altbacktrack); - if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) - { - if (opcode == OP_ASSERT_NOT || opcode == OP_ASSERTBACK_NOT) - { - common->local_exit = save_local_exit; - common->quit_label = save_quit_label; - common->quit = save_quit; - } - common->positive_assert = save_positive_assert; - common->then_trap = save_then_trap; - common->accept_label = save_accept_label; - common->positive_assert_quit = save_positive_assert_quit; - common->accept = save_accept; - return NULL; - } - common->accept_label = LABEL(); - if (common->accept != NULL) - set_jumps(common->accept, common->accept_label); - - /* Reset stack. */ - if (framesize < 0) - { - if (framesize == no_frame) - OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - else - free_stack(common, extrasize); - if (needs_control_head) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, SLJIT_MEM1(STACK_TOP), 0); - } - else - { - if ((opcode != OP_ASSERT_NOT && opcode != OP_ASSERTBACK_NOT) || conditional) - { - /* We don't need to keep the STR_PTR, only the previous private_data_ptr. */ - OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, SLJIT_IMM, (framesize + 1) * sizeof(sljit_sw)); - if (needs_control_head) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, SLJIT_MEM1(STACK_TOP), 0); - } - else - { - OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - if (needs_control_head) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, SLJIT_MEM1(STACK_TOP), (framesize + 1) * sizeof(sljit_sw)); - add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); - } - } - - if (opcode == OP_ASSERT_NOT || opcode == OP_ASSERTBACK_NOT) - { - /* We know that STR_PTR was stored on the top of the stack. */ - if (conditional) - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), needs_control_head ? sizeof(sljit_sw) : 0); - else if (bra == OP_BRAZERO) - { - if (framesize < 0) - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), (extrasize - 1) * sizeof(sljit_sw)); - else - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), framesize * sizeof(sljit_sw)); - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), (framesize + extrasize - 1) * sizeof(sljit_sw)); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, TMP1, 0); - } - OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_sw)); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); - } - else if (framesize >= 0) - { - /* For OP_BRA and OP_BRAMINZERO. */ - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, SLJIT_MEM1(STACK_TOP), framesize * sizeof(sljit_sw)); - } - } - add_jump(compiler, found, JUMP(SLJIT_JUMP)); - - compile_backtrackingpath(common, altbacktrack.top); - if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) - { - if (opcode == OP_ASSERT_NOT || opcode == OP_ASSERTBACK_NOT) - { - common->local_exit = save_local_exit; - common->quit_label = save_quit_label; - common->quit = save_quit; - } - common->positive_assert = save_positive_assert; - common->then_trap = save_then_trap; - common->accept_label = save_accept_label; - common->positive_assert_quit = save_positive_assert_quit; - common->accept = save_accept; - return NULL; - } - set_jumps(altbacktrack.topbacktracks, LABEL()); - - if (*cc != OP_ALT) - break; - - ccbegin = cc; - cc += GET(cc, 1); - } - -if (opcode == OP_ASSERT_NOT || opcode == OP_ASSERTBACK_NOT) - { - SLJIT_ASSERT(common->positive_assert_quit == NULL); - /* Makes the check less complicated below. */ - common->positive_assert_quit = common->quit; - } - -/* None of them matched. */ -if (common->positive_assert_quit != NULL) - { - jump = JUMP(SLJIT_JUMP); - set_jumps(common->positive_assert_quit, LABEL()); - SLJIT_ASSERT(framesize != no_stack); - if (framesize < 0) - OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, SLJIT_IMM, extrasize * sizeof(sljit_sw)); - else - { - OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); - OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, (framesize + extrasize) * sizeof(sljit_sw)); - } - JUMPHERE(jump); - } - -if (needs_control_head) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, SLJIT_MEM1(STACK_TOP), STACK(1)); - -if (opcode == OP_ASSERT || opcode == OP_ASSERTBACK) - { - /* Assert is failed. */ - if (conditional || bra == OP_BRAZERO) - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - - if (framesize < 0) - { - /* The topmost item should be 0. */ - if (bra == OP_BRAZERO) - { - if (extrasize == 2) - free_stack(common, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); - } - else - free_stack(common, extrasize); - } - else - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(extrasize - 1)); - /* The topmost item should be 0. */ - if (bra == OP_BRAZERO) - { - free_stack(common, framesize + extrasize - 1); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); - } - else - free_stack(common, framesize + extrasize); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, TMP1, 0); - } - jump = JUMP(SLJIT_JUMP); - if (bra != OP_BRAZERO) - add_jump(compiler, target, jump); - - /* Assert is successful. */ - set_jumps(tmp, LABEL()); - if (framesize < 0) - { - /* We know that STR_PTR was stored on the top of the stack. */ - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), (extrasize - 1) * sizeof(sljit_sw)); - /* Keep the STR_PTR on the top of the stack. */ - if (bra == OP_BRAZERO) - { - OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_sw)); - if (extrasize == 2) - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); - } - else if (bra == OP_BRAMINZERO) - { - OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_sw)); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); - } - } - else - { - if (bra == OP_BRA) - { - /* We don't need to keep the STR_PTR, only the previous private_data_ptr. */ - OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, SLJIT_IMM, (framesize + 1) * sizeof(sljit_sw)); - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), (extrasize - 2) * sizeof(sljit_sw)); - } - else - { - /* We don't need to keep the STR_PTR, only the previous private_data_ptr. */ - OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, SLJIT_IMM, (framesize + 2) * sizeof(sljit_sw)); - if (extrasize == 2) - { - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - if (bra == OP_BRAMINZERO) - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); - } - else - { - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), 0); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), bra == OP_BRAZERO ? STR_PTR : SLJIT_IMM, 0); - } - } - } - - if (bra == OP_BRAZERO) - { - backtrack->matchingpath = LABEL(); - SET_LABEL(jump, backtrack->matchingpath); - } - else if (bra == OP_BRAMINZERO) - { - JUMPTO(SLJIT_JUMP, backtrack->matchingpath); - JUMPHERE(brajump); - if (framesize >= 0) - { - OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, SLJIT_MEM1(STACK_TOP), framesize * sizeof(sljit_sw)); - } - set_jumps(backtrack->common.topbacktracks, LABEL()); - } - } -else - { - /* AssertNot is successful. */ - if (framesize < 0) - { - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - if (bra != OP_BRA) - { - if (extrasize == 2) - free_stack(common, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); - } - else - free_stack(common, extrasize); - } - else - { - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(extrasize - 1)); - /* The topmost item should be 0. */ - if (bra != OP_BRA) - { - free_stack(common, framesize + extrasize - 1); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); - } - else - free_stack(common, framesize + extrasize); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, TMP1, 0); - } - - if (bra == OP_BRAZERO) - backtrack->matchingpath = LABEL(); - else if (bra == OP_BRAMINZERO) - { - JUMPTO(SLJIT_JUMP, backtrack->matchingpath); - JUMPHERE(brajump); - } - - if (bra != OP_BRA) - { - SLJIT_ASSERT(found == &backtrack->common.topbacktracks); - set_jumps(backtrack->common.topbacktracks, LABEL()); - backtrack->common.topbacktracks = NULL; - } - } - -if (opcode == OP_ASSERT_NOT || opcode == OP_ASSERTBACK_NOT) - { - common->local_exit = save_local_exit; - common->quit_label = save_quit_label; - common->quit = save_quit; - } -common->positive_assert = save_positive_assert; -common->then_trap = save_then_trap; -common->accept_label = save_accept_label; -common->positive_assert_quit = save_positive_assert_quit; -common->accept = save_accept; -return cc + 1 + LINK_SIZE; -} - -static SLJIT_INLINE void match_once_common(compiler_common *common, pcre_uchar ket, int framesize, int private_data_ptr, BOOL has_alternatives, BOOL needs_control_head) -{ -DEFINE_COMPILER; -int stacksize; - -if (framesize < 0) - { - if (framesize == no_frame) - OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - else - { - stacksize = needs_control_head ? 1 : 0; - if (ket != OP_KET || has_alternatives) - stacksize++; - free_stack(common, stacksize); - } - - if (needs_control_head) - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), (ket != OP_KET || has_alternatives) ? sizeof(sljit_sw) : 0); - - /* TMP2 which is set here used by OP_KETRMAX below. */ - if (ket == OP_KETRMAX) - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), 0); - else if (ket == OP_KETRMIN) - { - /* Move the STR_PTR to the private_data_ptr. */ - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, SLJIT_MEM1(STACK_TOP), 0); - } - } -else - { - stacksize = (ket != OP_KET || has_alternatives) ? 2 : 1; - OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, SLJIT_IMM, (framesize + stacksize) * sizeof(sljit_sw)); - if (needs_control_head) - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), 0); - - if (ket == OP_KETRMAX) - { - /* TMP2 which is set here used by OP_KETRMAX below. */ - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - } - } -if (needs_control_head) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, TMP1, 0); -} - -static SLJIT_INLINE int match_capture_common(compiler_common *common, int stacksize, int offset, int private_data_ptr) -{ -DEFINE_COMPILER; - -if (common->capture_last_ptr != 0) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr, SLJIT_IMM, offset >> 1); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), TMP1, 0); - stacksize++; - } -if (common->optimized_cbracket[offset >> 1] == 0) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), TMP1, 0); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize + 1), TMP2, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), STR_PTR, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0); - stacksize += 2; - } -return stacksize; -} - -/* - Handling bracketed expressions is probably the most complex part. - - Stack layout naming characters: - S - Push the current STR_PTR - 0 - Push a 0 (NULL) - A - Push the current STR_PTR. Needed for restoring the STR_PTR - before the next alternative. Not pushed if there are no alternatives. - M - Any values pushed by the current alternative. Can be empty, or anything. - C - Push the previous OVECTOR(i), OVECTOR(i+1) and OVECTOR_PRIV(i) to the stack. - L - Push the previous local (pointed by localptr) to the stack - () - opional values stored on the stack - ()* - optonal, can be stored multiple times - - The following list shows the regular expression templates, their PCRE byte codes - and stack layout supported by pcre-sljit. - - (?:) OP_BRA | OP_KET A M - () OP_CBRA | OP_KET C M - (?:)+ OP_BRA | OP_KETRMAX 0 A M S ( A M S )* - OP_SBRA | OP_KETRMAX 0 L M S ( L M S )* - (?:)+? OP_BRA | OP_KETRMIN 0 A M S ( A M S )* - OP_SBRA | OP_KETRMIN 0 L M S ( L M S )* - ()+ OP_CBRA | OP_KETRMAX 0 C M S ( C M S )* - OP_SCBRA | OP_KETRMAX 0 C M S ( C M S )* - ()+? OP_CBRA | OP_KETRMIN 0 C M S ( C M S )* - OP_SCBRA | OP_KETRMIN 0 C M S ( C M S )* - (?:)? OP_BRAZERO | OP_BRA | OP_KET S ( A M 0 ) - (?:)?? OP_BRAMINZERO | OP_BRA | OP_KET S ( A M 0 ) - ()? OP_BRAZERO | OP_CBRA | OP_KET S ( C M 0 ) - ()?? OP_BRAMINZERO | OP_CBRA | OP_KET S ( C M 0 ) - (?:)* OP_BRAZERO | OP_BRA | OP_KETRMAX S 0 ( A M S )* - OP_BRAZERO | OP_SBRA | OP_KETRMAX S 0 ( L M S )* - (?:)*? OP_BRAMINZERO | OP_BRA | OP_KETRMIN S 0 ( A M S )* - OP_BRAMINZERO | OP_SBRA | OP_KETRMIN S 0 ( L M S )* - ()* OP_BRAZERO | OP_CBRA | OP_KETRMAX S 0 ( C M S )* - OP_BRAZERO | OP_SCBRA | OP_KETRMAX S 0 ( C M S )* - ()*? OP_BRAMINZERO | OP_CBRA | OP_KETRMIN S 0 ( C M S )* - OP_BRAMINZERO | OP_SCBRA | OP_KETRMIN S 0 ( C M S )* - - - Stack layout naming characters: - A - Push the alternative index (starting from 0) on the stack. - Not pushed if there is no alternatives. - M - Any values pushed by the current alternative. Can be empty, or anything. - - The next list shows the possible content of a bracket: - (|) OP_*BRA | OP_ALT ... M A - (?()|) OP_*COND | OP_ALT M A - (?>|) OP_ONCE | OP_ALT ... [stack trace] M A - (?>|) OP_ONCE_NC | OP_ALT ... [stack trace] M A - Or nothing, if trace is unnecessary -*/ - -static pcre_uchar *compile_bracket_matchingpath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent) -{ -DEFINE_COMPILER; -backtrack_common *backtrack; -pcre_uchar opcode; -int private_data_ptr = 0; -int offset = 0; -int i, stacksize; -int repeat_ptr = 0, repeat_length = 0; -int repeat_type = 0, repeat_count = 0; -pcre_uchar *ccbegin; -pcre_uchar *matchingpath; -pcre_uchar *slot; -pcre_uchar bra = OP_BRA; -pcre_uchar ket; -assert_backtrack *assert; -BOOL has_alternatives; -BOOL needs_control_head = FALSE; -struct sljit_jump *jump; -struct sljit_jump *skip; -struct sljit_label *rmax_label = NULL; -struct sljit_jump *braminzero = NULL; - -PUSH_BACKTRACK(sizeof(bracket_backtrack), cc, NULL); - -if (*cc == OP_BRAZERO || *cc == OP_BRAMINZERO) - { - bra = *cc; - cc++; - opcode = *cc; - } - -opcode = *cc; -ccbegin = cc; -matchingpath = bracketend(cc) - 1 - LINK_SIZE; -ket = *matchingpath; -if (ket == OP_KET && PRIVATE_DATA(matchingpath) != 0) - { - repeat_ptr = PRIVATE_DATA(matchingpath); - repeat_length = PRIVATE_DATA(matchingpath + 1); - repeat_type = PRIVATE_DATA(matchingpath + 2); - repeat_count = PRIVATE_DATA(matchingpath + 3); - SLJIT_ASSERT(repeat_length != 0 && repeat_type != 0 && repeat_count != 0); - if (repeat_type == OP_UPTO) - ket = OP_KETRMAX; - if (repeat_type == OP_MINUPTO) - ket = OP_KETRMIN; - } - -if ((opcode == OP_COND || opcode == OP_SCOND) && cc[1 + LINK_SIZE] == OP_DEF) - { - /* Drop this bracket_backtrack. */ - parent->top = backtrack->prev; - return matchingpath + 1 + LINK_SIZE + repeat_length; - } - -matchingpath = ccbegin + 1 + LINK_SIZE; -SLJIT_ASSERT(ket == OP_KET || ket == OP_KETRMAX || ket == OP_KETRMIN); -SLJIT_ASSERT(!((bra == OP_BRAZERO && ket == OP_KETRMIN) || (bra == OP_BRAMINZERO && ket == OP_KETRMAX))); -cc += GET(cc, 1); - -has_alternatives = *cc == OP_ALT; -if (SLJIT_UNLIKELY(opcode == OP_COND || opcode == OP_SCOND)) - has_alternatives = (*matchingpath == OP_RREF || *matchingpath == OP_DNRREF) ? FALSE : TRUE; - -if (SLJIT_UNLIKELY(opcode == OP_COND) && (*cc == OP_KETRMAX || *cc == OP_KETRMIN)) - opcode = OP_SCOND; -if (SLJIT_UNLIKELY(opcode == OP_ONCE_NC)) - opcode = OP_ONCE; - -if (opcode == OP_CBRA || opcode == OP_SCBRA) - { - /* Capturing brackets has a pre-allocated space. */ - offset = GET2(ccbegin, 1 + LINK_SIZE); - if (common->optimized_cbracket[offset] == 0) - { - private_data_ptr = OVECTOR_PRIV(offset); - offset <<= 1; - } - else - { - offset <<= 1; - private_data_ptr = OVECTOR(offset); - } - BACKTRACK_AS(bracket_backtrack)->private_data_ptr = private_data_ptr; - matchingpath += IMM2_SIZE; - } -else if (opcode == OP_ONCE || opcode == OP_SBRA || opcode == OP_SCOND) - { - /* Other brackets simply allocate the next entry. */ - private_data_ptr = PRIVATE_DATA(ccbegin); - SLJIT_ASSERT(private_data_ptr != 0); - BACKTRACK_AS(bracket_backtrack)->private_data_ptr = private_data_ptr; - if (opcode == OP_ONCE) - BACKTRACK_AS(bracket_backtrack)->u.framesize = get_framesize(common, ccbegin, NULL, FALSE, &needs_control_head); - } - -/* Instructions before the first alternative. */ -stacksize = 0; -if (ket == OP_KETRMAX || (ket == OP_KETRMIN && bra != OP_BRAMINZERO)) - stacksize++; -if (bra == OP_BRAZERO) - stacksize++; - -if (stacksize > 0) - allocate_stack(common, stacksize); - -stacksize = 0; -if (ket == OP_KETRMAX || (ket == OP_KETRMIN && bra != OP_BRAMINZERO)) - { - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), SLJIT_IMM, 0); - stacksize++; - } - -if (bra == OP_BRAZERO) - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), STR_PTR, 0); - -if (bra == OP_BRAMINZERO) - { - /* This is a backtrack path! (Since the try-path of OP_BRAMINZERO matches to the empty string) */ - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - if (ket != OP_KETRMIN) - { - free_stack(common, 1); - braminzero = CMP(SLJIT_C_EQUAL, STR_PTR, 0, SLJIT_IMM, 0); - } - else - { - if (opcode == OP_ONCE || opcode >= OP_SBRA) - { - jump = CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0); - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); - /* Nothing stored during the first run. */ - skip = JUMP(SLJIT_JUMP); - JUMPHERE(jump); - /* Checking zero-length iteration. */ - if (opcode != OP_ONCE || BACKTRACK_AS(bracket_backtrack)->u.framesize < 0) - { - /* When we come from outside, private_data_ptr contains the previous STR_PTR. */ - braminzero = CMP(SLJIT_C_EQUAL, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - } - else - { - /* Except when the whole stack frame must be saved. */ - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - braminzero = CMP(SLJIT_C_EQUAL, STR_PTR, 0, SLJIT_MEM1(TMP1), (BACKTRACK_AS(bracket_backtrack)->u.framesize + 1) * sizeof(sljit_sw)); - } - JUMPHERE(skip); - } - else - { - jump = CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0); - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); - JUMPHERE(jump); - } - } - } - -if (repeat_type != 0) - { - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), repeat_ptr, SLJIT_IMM, repeat_count); - if (repeat_type == OP_EXACT) - rmax_label = LABEL(); - } - -if (ket == OP_KETRMIN) - BACKTRACK_AS(bracket_backtrack)->recursive_matchingpath = LABEL(); - -if (ket == OP_KETRMAX) - { - rmax_label = LABEL(); - if (has_alternatives && opcode != OP_ONCE && opcode < OP_SBRA && repeat_type == 0) - BACKTRACK_AS(bracket_backtrack)->alternative_matchingpath = rmax_label; - } - -/* Handling capturing brackets and alternatives. */ -if (opcode == OP_ONCE) - { - stacksize = 0; - if (needs_control_head) - { - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr); - stacksize++; - } - - if (BACKTRACK_AS(bracket_backtrack)->u.framesize < 0) - { - /* Neither capturing brackets nor recursions are found in the block. */ - if (ket == OP_KETRMIN) - { - stacksize += 2; - if (!needs_control_head) - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - } - else - { - if (BACKTRACK_AS(bracket_backtrack)->u.framesize == no_frame) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, STACK_TOP, 0); - if (ket == OP_KETRMAX || has_alternatives) - stacksize++; - } - - if (stacksize > 0) - allocate_stack(common, stacksize); - - stacksize = 0; - if (needs_control_head) - { - stacksize++; - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0); - } - - if (ket == OP_KETRMIN) - { - if (needs_control_head) - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), STR_PTR, 0); - if (BACKTRACK_AS(bracket_backtrack)->u.framesize == no_frame) - OP2(SLJIT_SUB, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, STACK_TOP, 0, SLJIT_IMM, needs_control_head ? (2 * sizeof(sljit_sw)) : sizeof(sljit_sw)); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize + 1), TMP2, 0); - } - else if (ket == OP_KETRMAX || has_alternatives) - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), STR_PTR, 0); - } - else - { - if (ket != OP_KET || has_alternatives) - stacksize++; - - stacksize += BACKTRACK_AS(bracket_backtrack)->u.framesize + 1; - allocate_stack(common, stacksize); - - if (needs_control_head) - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0); - - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - OP2(SLJIT_SUB, TMP2, 0, STACK_TOP, 0, SLJIT_IMM, stacksize * sizeof(sljit_sw)); - - stacksize = needs_control_head ? 1 : 0; - if (ket != OP_KET || has_alternatives) - { - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), STR_PTR, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, TMP2, 0); - stacksize++; - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), TMP1, 0); - } - else - { - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, TMP2, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), TMP1, 0); - } - init_frame(common, ccbegin, NULL, BACKTRACK_AS(bracket_backtrack)->u.framesize + stacksize, stacksize + 1, FALSE); - } - } -else if (opcode == OP_CBRA || opcode == OP_SCBRA) - { - /* Saving the previous values. */ - if (common->optimized_cbracket[offset >> 1] != 0) - { - SLJIT_ASSERT(private_data_ptr == OVECTOR(offset)); - allocate_stack(common, 2); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr + sizeof(sljit_sw)); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, STR_PTR, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP1, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP2, 0); - } - else - { - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - allocate_stack(common, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, STR_PTR, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0); - } - } -else if (opcode == OP_SBRA || opcode == OP_SCOND) - { - /* Saving the previous value. */ - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - allocate_stack(common, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, STR_PTR, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0); - } -else if (has_alternatives) - { - /* Pushing the starting string pointer. */ - allocate_stack(common, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); - } - -/* Generating code for the first alternative. */ -if (opcode == OP_COND || opcode == OP_SCOND) - { - if (*matchingpath == OP_CREF) - { - SLJIT_ASSERT(has_alternatives); - add_jump(compiler, &(BACKTRACK_AS(bracket_backtrack)->u.condfailed), - CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(GET2(matchingpath, 1) << 1), SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1))); - matchingpath += 1 + IMM2_SIZE; - } - else if (*matchingpath == OP_DNCREF) - { - SLJIT_ASSERT(has_alternatives); - - i = GET2(matchingpath, 1 + IMM2_SIZE); - slot = common->name_table + GET2(matchingpath, 1) * common->name_entry_size; - OP1(SLJIT_MOV, TMP3, 0, STR_PTR, 0); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)); - OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(GET2(slot, 0) << 1), TMP1, 0); - slot += common->name_entry_size; - i--; - while (i-- > 0) - { - OP2(SLJIT_SUB, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(GET2(slot, 0) << 1), TMP1, 0); - OP2(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, STR_PTR, 0); - slot += common->name_entry_size; - } - OP1(SLJIT_MOV, STR_PTR, 0, TMP3, 0); - add_jump(compiler, &(BACKTRACK_AS(bracket_backtrack)->u.condfailed), JUMP(SLJIT_C_ZERO)); - matchingpath += 1 + 2 * IMM2_SIZE; - } - else if (*matchingpath == OP_RREF || *matchingpath == OP_DNRREF) - { - /* Never has other case. */ - BACKTRACK_AS(bracket_backtrack)->u.condfailed = NULL; - SLJIT_ASSERT(!has_alternatives); - - if (*matchingpath == OP_RREF) - { - stacksize = GET2(matchingpath, 1); - if (common->currententry == NULL) - stacksize = 0; - else if (stacksize == RREF_ANY) - stacksize = 1; - else if (common->currententry->start == 0) - stacksize = stacksize == 0; - else - stacksize = stacksize == (int)GET2(common->start, common->currententry->start + 1 + LINK_SIZE); - - if (stacksize != 0) - matchingpath += 1 + IMM2_SIZE; - } - else - { - if (common->currententry == NULL || common->currententry->start == 0) - stacksize = 0; - else - { - stacksize = GET2(matchingpath, 1 + IMM2_SIZE); - slot = common->name_table + GET2(matchingpath, 1) * common->name_entry_size; - i = (int)GET2(common->start, common->currententry->start + 1 + LINK_SIZE); - while (stacksize > 0) - { - if ((int)GET2(slot, 0) == i) - break; - slot += common->name_entry_size; - stacksize--; - } - } - - if (stacksize != 0) - matchingpath += 1 + 2 * IMM2_SIZE; - } - - /* The stacksize == 0 is a common "else" case. */ - if (stacksize == 0) - { - if (*cc == OP_ALT) - { - matchingpath = cc + 1 + LINK_SIZE; - cc += GET(cc, 1); - } - else - matchingpath = cc; - } - } - else - { - SLJIT_ASSERT(has_alternatives && *matchingpath >= OP_ASSERT && *matchingpath <= OP_ASSERTBACK_NOT); - /* Similar code as PUSH_BACKTRACK macro. */ - assert = sljit_alloc_memory(compiler, sizeof(assert_backtrack)); - if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) - return NULL; - memset(assert, 0, sizeof(assert_backtrack)); - assert->common.cc = matchingpath; - BACKTRACK_AS(bracket_backtrack)->u.assert = assert; - matchingpath = compile_assert_matchingpath(common, matchingpath, assert, TRUE); - } - } - -compile_matchingpath(common, matchingpath, cc, backtrack); -if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) - return NULL; - -if (opcode == OP_ONCE) - match_once_common(common, ket, BACKTRACK_AS(bracket_backtrack)->u.framesize, private_data_ptr, has_alternatives, needs_control_head); - -stacksize = 0; -if (repeat_type == OP_MINUPTO) - { - /* We need to preserve the counter. TMP2 will be used below. */ - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), repeat_ptr); - stacksize++; - } -if (ket != OP_KET || bra != OP_BRA) - stacksize++; -if (offset != 0) - { - if (common->capture_last_ptr != 0) - stacksize++; - if (common->optimized_cbracket[offset >> 1] == 0) - stacksize += 2; - } -if (has_alternatives && opcode != OP_ONCE) - stacksize++; - -if (stacksize > 0) - allocate_stack(common, stacksize); - -stacksize = 0; -if (repeat_type == OP_MINUPTO) - { - /* TMP2 was set above. */ - OP2(SLJIT_SUB, SLJIT_MEM1(STACK_TOP), STACK(stacksize), TMP2, 0, SLJIT_IMM, 1); - stacksize++; - } - -if (ket != OP_KET || bra != OP_BRA) - { - if (ket != OP_KET) - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), STR_PTR, 0); - else - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), SLJIT_IMM, 0); - stacksize++; - } - -if (offset != 0) - stacksize = match_capture_common(common, stacksize, offset, private_data_ptr); - -if (has_alternatives) - { - if (opcode != OP_ONCE) - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), SLJIT_IMM, 0); - if (ket != OP_KETRMAX) - BACKTRACK_AS(bracket_backtrack)->alternative_matchingpath = LABEL(); - } - -/* Must be after the matchingpath label. */ -if (offset != 0 && common->optimized_cbracket[offset >> 1] != 0) - { - SLJIT_ASSERT(private_data_ptr == OVECTOR(offset + 0)); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), STR_PTR, 0); - } - -if (ket == OP_KETRMAX) - { - if (repeat_type != 0) - { - if (has_alternatives) - BACKTRACK_AS(bracket_backtrack)->alternative_matchingpath = LABEL(); - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_MEM1(SLJIT_LOCALS_REG), repeat_ptr, SLJIT_MEM1(SLJIT_LOCALS_REG), repeat_ptr, SLJIT_IMM, 1); - JUMPTO(SLJIT_C_NOT_ZERO, rmax_label); - /* Drop STR_PTR for greedy plus quantifier. */ - if (opcode != OP_ONCE) - free_stack(common, 1); - } - else if (opcode == OP_ONCE || opcode >= OP_SBRA) - { - if (has_alternatives) - BACKTRACK_AS(bracket_backtrack)->alternative_matchingpath = LABEL(); - /* Checking zero-length iteration. */ - if (opcode != OP_ONCE) - { - CMPTO(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, STR_PTR, 0, rmax_label); - /* Drop STR_PTR for greedy plus quantifier. */ - if (bra != OP_BRAZERO) - free_stack(common, 1); - } - else - /* TMP2 must contain the starting STR_PTR. */ - CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_PTR, 0, rmax_label); - } - else - JUMPTO(SLJIT_JUMP, rmax_label); - BACKTRACK_AS(bracket_backtrack)->recursive_matchingpath = LABEL(); - } - -if (repeat_type == OP_EXACT) - { - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_MEM1(SLJIT_LOCALS_REG), repeat_ptr, SLJIT_MEM1(SLJIT_LOCALS_REG), repeat_ptr, SLJIT_IMM, 1); - JUMPTO(SLJIT_C_NOT_ZERO, rmax_label); - } -else if (repeat_type == OP_UPTO) - { - /* We need to preserve the counter. */ - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), repeat_ptr); - allocate_stack(common, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0); - } - -if (bra == OP_BRAZERO) - BACKTRACK_AS(bracket_backtrack)->zero_matchingpath = LABEL(); - -if (bra == OP_BRAMINZERO) - { - /* This is a backtrack path! (From the viewpoint of OP_BRAMINZERO) */ - JUMPTO(SLJIT_JUMP, ((braminzero_backtrack *)parent)->matchingpath); - if (braminzero != NULL) - { - JUMPHERE(braminzero); - /* We need to release the end pointer to perform the - backtrack for the zero-length iteration. When - framesize is < 0, OP_ONCE will do the release itself. */ - if (opcode == OP_ONCE && BACKTRACK_AS(bracket_backtrack)->u.framesize >= 0) - { - OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); - } - else if (ket == OP_KETRMIN && opcode != OP_ONCE) - free_stack(common, 1); - } - /* Continue to the normal backtrack. */ - } - -if ((ket != OP_KET && bra != OP_BRAMINZERO) || bra == OP_BRAZERO) - count_match(common); - -/* Skip the other alternatives. */ -while (*cc == OP_ALT) - cc += GET(cc, 1); -cc += 1 + LINK_SIZE; - -/* Temporarily encoding the needs_control_head in framesize. */ -if (opcode == OP_ONCE) - BACKTRACK_AS(bracket_backtrack)->u.framesize = (BACKTRACK_AS(bracket_backtrack)->u.framesize << 1) | (needs_control_head ? 1 : 0); -return cc + repeat_length; -} - -static pcre_uchar *compile_bracketpos_matchingpath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent) -{ -DEFINE_COMPILER; -backtrack_common *backtrack; -pcre_uchar opcode; -int private_data_ptr; -int cbraprivptr = 0; -BOOL needs_control_head; -int framesize; -int stacksize; -int offset = 0; -BOOL zero = FALSE; -pcre_uchar *ccbegin = NULL; -int stack; /* Also contains the offset of control head. */ -struct sljit_label *loop = NULL; -struct jump_list *emptymatch = NULL; - -PUSH_BACKTRACK(sizeof(bracketpos_backtrack), cc, NULL); -if (*cc == OP_BRAPOSZERO) - { - zero = TRUE; - cc++; - } - -opcode = *cc; -private_data_ptr = PRIVATE_DATA(cc); -SLJIT_ASSERT(private_data_ptr != 0); -BACKTRACK_AS(bracketpos_backtrack)->private_data_ptr = private_data_ptr; -switch(opcode) - { - case OP_BRAPOS: - case OP_SBRAPOS: - ccbegin = cc + 1 + LINK_SIZE; - break; - - case OP_CBRAPOS: - case OP_SCBRAPOS: - offset = GET2(cc, 1 + LINK_SIZE); - /* This case cannot be optimized in the same was as - normal capturing brackets. */ - SLJIT_ASSERT(common->optimized_cbracket[offset] == 0); - cbraprivptr = OVECTOR_PRIV(offset); - offset <<= 1; - ccbegin = cc + 1 + LINK_SIZE + IMM2_SIZE; - break; - - default: - SLJIT_ASSERT_STOP(); - break; - } - -framesize = get_framesize(common, cc, NULL, FALSE, &needs_control_head); -BACKTRACK_AS(bracketpos_backtrack)->framesize = framesize; -if (framesize < 0) - { - if (offset != 0) - { - stacksize = 2; - if (common->capture_last_ptr != 0) - stacksize++; - } - else - stacksize = 1; - - if (needs_control_head) - stacksize++; - if (!zero) - stacksize++; - - BACKTRACK_AS(bracketpos_backtrack)->stacksize = stacksize; - allocate_stack(common, stacksize); - if (framesize == no_frame) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, STACK_TOP, 0); - - stack = 0; - if (offset != 0) - { - stack = 2; - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP1, 0); - if (common->capture_last_ptr != 0) - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP2, 0); - if (needs_control_head) - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr); - if (common->capture_last_ptr != 0) - { - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), TMP1, 0); - stack = 3; - } - } - else - { - if (needs_control_head) - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); - stack = 1; - } - - if (needs_control_head) - stack++; - if (!zero) - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stack), SLJIT_IMM, 1); - if (needs_control_head) - { - stack--; - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stack), TMP2, 0); - } - } -else - { - stacksize = framesize + 1; - if (!zero) - stacksize++; - if (needs_control_head) - stacksize++; - if (offset == 0) - stacksize++; - BACKTRACK_AS(bracketpos_backtrack)->stacksize = stacksize; - - allocate_stack(common, stacksize); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - if (needs_control_head) - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr); - OP2(SLJIT_SUB, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, STACK_TOP, 0, SLJIT_IMM, -STACK(stacksize - 1)); - - stack = 0; - if (!zero) - { - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 1); - stack = 1; - } - if (needs_control_head) - { - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stack), TMP2, 0); - stack++; - } - if (offset == 0) - { - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stack), STR_PTR, 0); - stack++; - } - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stack), TMP1, 0); - init_frame(common, cc, NULL, stacksize - 1, stacksize - framesize, FALSE); - stack -= 1 + (offset == 0); - } - -if (offset != 0) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), cbraprivptr, STR_PTR, 0); - -loop = LABEL(); -while (*cc != OP_KETRPOS) - { - backtrack->top = NULL; - backtrack->topbacktracks = NULL; - cc += GET(cc, 1); - - compile_matchingpath(common, ccbegin, cc, backtrack); - if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) - return NULL; - - if (framesize < 0) - { - if (framesize == no_frame) - OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - - if (offset != 0) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), cbraprivptr); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), STR_PTR, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), cbraprivptr, STR_PTR, 0); - if (common->capture_last_ptr != 0) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr, SLJIT_IMM, offset >> 1); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0); - } - else - { - if (opcode == OP_SBRAPOS) - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); - } - - if (opcode == OP_SBRAPOS || opcode == OP_SCBRAPOS) - add_jump(compiler, &emptymatch, CMP(SLJIT_C_EQUAL, TMP1, 0, STR_PTR, 0)); - - if (!zero) - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize - 1), SLJIT_IMM, 0); - } - else - { - if (offset != 0) - { - OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, SLJIT_IMM, stacksize * sizeof(sljit_sw)); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), cbraprivptr); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), STR_PTR, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), cbraprivptr, STR_PTR, 0); - if (common->capture_last_ptr != 0) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr, SLJIT_IMM, offset >> 1); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0); - } - else - { - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - OP2(SLJIT_ADD, STACK_TOP, 0, TMP2, 0, SLJIT_IMM, stacksize * sizeof(sljit_sw)); - if (opcode == OP_SBRAPOS) - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), (framesize + 1) * sizeof(sljit_sw)); - OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), (framesize + 1) * sizeof(sljit_sw), STR_PTR, 0); - } - - if (opcode == OP_SBRAPOS || opcode == OP_SCBRAPOS) - add_jump(compiler, &emptymatch, CMP(SLJIT_C_EQUAL, TMP1, 0, STR_PTR, 0)); - - if (!zero) - { - if (framesize < 0) - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize - 1), SLJIT_IMM, 0); - else - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); - } - } - - if (needs_control_head) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, SLJIT_MEM1(STACK_TOP), STACK(stack)); - - JUMPTO(SLJIT_JUMP, loop); - flush_stubs(common); - - compile_backtrackingpath(common, backtrack->top); - if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) - return NULL; - set_jumps(backtrack->topbacktracks, LABEL()); - - if (framesize < 0) - { - if (offset != 0) - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), cbraprivptr); - else - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - } - else - { - if (offset != 0) - { - /* Last alternative. */ - if (*cc == OP_KETRPOS) - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), cbraprivptr); - } - else - { - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(TMP2), (framesize + 1) * sizeof(sljit_sw)); - } - } - - if (*cc == OP_KETRPOS) - break; - ccbegin = cc + 1 + LINK_SIZE; - } - -/* We don't have to restore the control head in case of a failed match. */ - -backtrack->topbacktracks = NULL; -if (!zero) - { - if (framesize < 0) - add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(STACK_TOP), STACK(stacksize - 1), SLJIT_IMM, 0)); - else /* TMP2 is set to [private_data_ptr] above. */ - add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(TMP2), (stacksize - 1) * sizeof(sljit_sw), SLJIT_IMM, 0)); - } - -/* None of them matched. */ -set_jumps(emptymatch, LABEL()); -count_match(common); -return cc + 1 + LINK_SIZE; -} - -static SLJIT_INLINE pcre_uchar *get_iterator_parameters(compiler_common *common, pcre_uchar *cc, pcre_uchar *opcode, pcre_uchar *type, int *max, int *min, pcre_uchar **end) -{ -int class_len; - -*opcode = *cc; -if (*opcode >= OP_STAR && *opcode <= OP_POSUPTO) - { - cc++; - *type = OP_CHAR; - } -else if (*opcode >= OP_STARI && *opcode <= OP_POSUPTOI) - { - cc++; - *type = OP_CHARI; - *opcode -= OP_STARI - OP_STAR; - } -else if (*opcode >= OP_NOTSTAR && *opcode <= OP_NOTPOSUPTO) - { - cc++; - *type = OP_NOT; - *opcode -= OP_NOTSTAR - OP_STAR; - } -else if (*opcode >= OP_NOTSTARI && *opcode <= OP_NOTPOSUPTOI) - { - cc++; - *type = OP_NOTI; - *opcode -= OP_NOTSTARI - OP_STAR; - } -else if (*opcode >= OP_TYPESTAR && *opcode <= OP_TYPEPOSUPTO) - { - cc++; - *opcode -= OP_TYPESTAR - OP_STAR; - *type = 0; - } -else - { - SLJIT_ASSERT(*opcode == OP_CLASS || *opcode == OP_NCLASS || *opcode == OP_XCLASS); - *type = *opcode; - cc++; - class_len = (*type < OP_XCLASS) ? (int)(1 + (32 / sizeof(pcre_uchar))) : GET(cc, 0); - *opcode = cc[class_len - 1]; - if (*opcode >= OP_CRSTAR && *opcode <= OP_CRMINQUERY) - { - *opcode -= OP_CRSTAR - OP_STAR; - if (end != NULL) - *end = cc + class_len; - } - else if (*opcode >= OP_CRPOSSTAR && *opcode <= OP_CRPOSQUERY) - { - *opcode -= OP_CRPOSSTAR - OP_POSSTAR; - if (end != NULL) - *end = cc + class_len; - } - else - { - SLJIT_ASSERT(*opcode == OP_CRRANGE || *opcode == OP_CRMINRANGE || *opcode == OP_CRPOSRANGE); - *max = GET2(cc, (class_len + IMM2_SIZE)); - *min = GET2(cc, class_len); - - if (*min == 0) - { - SLJIT_ASSERT(*max != 0); - *opcode = (*opcode == OP_CRRANGE) ? OP_UPTO : (*opcode == OP_CRMINRANGE ? OP_MINUPTO : OP_POSUPTO); - } - if (*max == *min) - *opcode = OP_EXACT; - - if (end != NULL) - *end = cc + class_len + 2 * IMM2_SIZE; - } - return cc; - } - -if (*opcode == OP_UPTO || *opcode == OP_MINUPTO || *opcode == OP_EXACT || *opcode == OP_POSUPTO) - { - *max = GET2(cc, 0); - cc += IMM2_SIZE; - } - -if (*type == 0) - { - *type = *cc; - if (end != NULL) - *end = next_opcode(common, cc); - cc++; - return cc; - } - -if (end != NULL) - { - *end = cc + 1; -#ifdef SUPPORT_UTF - if (common->utf && HAS_EXTRALEN(*cc)) *end += GET_EXTRALEN(*cc); -#endif - } -return cc; -} - -static pcre_uchar *compile_iterator_matchingpath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent) -{ -DEFINE_COMPILER; -backtrack_common *backtrack; -pcre_uchar opcode; -pcre_uchar type; -int max = -1, min = -1; -pcre_uchar* end; -jump_list *nomatch = NULL; -struct sljit_jump *jump = NULL; -struct sljit_label *label; -int private_data_ptr = PRIVATE_DATA(cc); -int base = (private_data_ptr == 0) ? SLJIT_MEM1(STACK_TOP) : SLJIT_MEM1(SLJIT_LOCALS_REG); -int offset0 = (private_data_ptr == 0) ? STACK(0) : private_data_ptr; -int offset1 = (private_data_ptr == 0) ? STACK(1) : private_data_ptr + (int)sizeof(sljit_sw); -int tmp_base, tmp_offset; - -PUSH_BACKTRACK(sizeof(iterator_backtrack), cc, NULL); - -cc = get_iterator_parameters(common, cc, &opcode, &type, &max, &min, &end); - -switch(type) - { - case OP_NOT_DIGIT: - case OP_DIGIT: - case OP_NOT_WHITESPACE: - case OP_WHITESPACE: - case OP_NOT_WORDCHAR: - case OP_WORDCHAR: - case OP_ANY: - case OP_ALLANY: - case OP_ANYBYTE: - case OP_ANYNL: - case OP_NOT_HSPACE: - case OP_HSPACE: - case OP_NOT_VSPACE: - case OP_VSPACE: - case OP_CHAR: - case OP_CHARI: - case OP_NOT: - case OP_NOTI: - case OP_CLASS: - case OP_NCLASS: - tmp_base = TMP3; - tmp_offset = 0; - break; - - default: - SLJIT_ASSERT_STOP(); - /* Fall through. */ - - case OP_EXTUNI: - case OP_XCLASS: - case OP_NOTPROP: - case OP_PROP: - tmp_base = SLJIT_MEM1(SLJIT_LOCALS_REG); - tmp_offset = POSSESSIVE0; - break; - } - -switch(opcode) - { - case OP_STAR: - case OP_PLUS: - case OP_UPTO: - case OP_CRRANGE: - if (type == OP_ANYNL || type == OP_EXTUNI) - { - SLJIT_ASSERT(private_data_ptr == 0); - if (opcode == OP_STAR || opcode == OP_UPTO) - { - allocate_stack(common, 2); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0); - } - else - { - allocate_stack(common, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); - } - - if (opcode == OP_UPTO || opcode == OP_CRRANGE) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, SLJIT_IMM, 0); - - label = LABEL(); - compile_char1_matchingpath(common, type, cc, &backtrack->topbacktracks); - if (opcode == OP_UPTO || opcode == OP_CRRANGE) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); - OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); - if (opcode == OP_CRRANGE && min > 0) - CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, min, label); - if (opcode == OP_UPTO || (opcode == OP_CRRANGE && max > 0)) - jump = CMP(SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, max); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, TMP1, 0); - } - - /* We cannot use TMP3 because of this allocate_stack. */ - allocate_stack(common, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); - JUMPTO(SLJIT_JUMP, label); - if (jump != NULL) - JUMPHERE(jump); - } - else - { - if (opcode == OP_PLUS) - compile_char1_matchingpath(common, type, cc, &backtrack->topbacktracks); - if (private_data_ptr == 0) - allocate_stack(common, 2); - OP1(SLJIT_MOV, base, offset0, STR_PTR, 0); - if (opcode <= OP_PLUS) - OP1(SLJIT_MOV, base, offset1, STR_PTR, 0); - else - OP1(SLJIT_MOV, base, offset1, SLJIT_IMM, 1); - label = LABEL(); - compile_char1_matchingpath(common, type, cc, &nomatch); - OP1(SLJIT_MOV, base, offset0, STR_PTR, 0); - if (opcode <= OP_PLUS) - JUMPTO(SLJIT_JUMP, label); - else if (opcode == OP_CRRANGE && max == 0) - { - OP2(SLJIT_ADD, base, offset1, base, offset1, SLJIT_IMM, 1); - JUMPTO(SLJIT_JUMP, label); - } - else - { - OP1(SLJIT_MOV, TMP1, 0, base, offset1); - OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); - OP1(SLJIT_MOV, base, offset1, TMP1, 0); - CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, max + 1, label); - } - set_jumps(nomatch, LABEL()); - if (opcode == OP_CRRANGE) - add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_LESS, base, offset1, SLJIT_IMM, min + 1)); - OP1(SLJIT_MOV, STR_PTR, 0, base, offset0); - } - BACKTRACK_AS(iterator_backtrack)->matchingpath = LABEL(); - break; - - case OP_MINSTAR: - case OP_MINPLUS: - if (opcode == OP_MINPLUS) - compile_char1_matchingpath(common, type, cc, &backtrack->topbacktracks); - if (private_data_ptr == 0) - allocate_stack(common, 1); - OP1(SLJIT_MOV, base, offset0, STR_PTR, 0); - BACKTRACK_AS(iterator_backtrack)->matchingpath = LABEL(); - break; - - case OP_MINUPTO: - case OP_CRMINRANGE: - if (private_data_ptr == 0) - allocate_stack(common, 2); - OP1(SLJIT_MOV, base, offset0, STR_PTR, 0); - OP1(SLJIT_MOV, base, offset1, SLJIT_IMM, 1); - if (opcode == OP_CRMINRANGE) - add_jump(compiler, &backtrack->topbacktracks, JUMP(SLJIT_JUMP)); - BACKTRACK_AS(iterator_backtrack)->matchingpath = LABEL(); - break; - - case OP_QUERY: - case OP_MINQUERY: - if (private_data_ptr == 0) - allocate_stack(common, 1); - OP1(SLJIT_MOV, base, offset0, STR_PTR, 0); - if (opcode == OP_QUERY) - compile_char1_matchingpath(common, type, cc, &backtrack->topbacktracks); - BACKTRACK_AS(iterator_backtrack)->matchingpath = LABEL(); - break; - - case OP_EXACT: - OP1(SLJIT_MOV, tmp_base, tmp_offset, SLJIT_IMM, max); - label = LABEL(); - compile_char1_matchingpath(common, type, cc, &backtrack->topbacktracks); - OP2(SLJIT_SUB | SLJIT_SET_E, tmp_base, tmp_offset, tmp_base, tmp_offset, SLJIT_IMM, 1); - JUMPTO(SLJIT_C_NOT_ZERO, label); - break; - - case OP_POSSTAR: - case OP_POSPLUS: - case OP_POSUPTO: - if (opcode == OP_POSPLUS) - compile_char1_matchingpath(common, type, cc, &backtrack->topbacktracks); - if (opcode == OP_POSUPTO) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, SLJIT_IMM, max); - OP1(SLJIT_MOV, tmp_base, tmp_offset, STR_PTR, 0); - label = LABEL(); - compile_char1_matchingpath(common, type, cc, &nomatch); - OP1(SLJIT_MOV, tmp_base, tmp_offset, STR_PTR, 0); - if (opcode != OP_POSUPTO) - JUMPTO(SLJIT_JUMP, label); - else - { - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, SLJIT_IMM, 1); - JUMPTO(SLJIT_C_NOT_ZERO, label); - } - set_jumps(nomatch, LABEL()); - OP1(SLJIT_MOV, STR_PTR, 0, tmp_base, tmp_offset); - break; - - case OP_POSQUERY: - OP1(SLJIT_MOV, tmp_base, tmp_offset, STR_PTR, 0); - compile_char1_matchingpath(common, type, cc, &nomatch); - OP1(SLJIT_MOV, tmp_base, tmp_offset, STR_PTR, 0); - set_jumps(nomatch, LABEL()); - OP1(SLJIT_MOV, STR_PTR, 0, tmp_base, tmp_offset); - break; - - case OP_CRPOSRANGE: - /* Combination of OP_EXACT and OP_POSSTAR or OP_POSUPTO */ - OP1(SLJIT_MOV, tmp_base, tmp_offset, SLJIT_IMM, min); - label = LABEL(); - compile_char1_matchingpath(common, type, cc, &backtrack->topbacktracks); - OP2(SLJIT_SUB | SLJIT_SET_E, tmp_base, tmp_offset, tmp_base, tmp_offset, SLJIT_IMM, 1); - JUMPTO(SLJIT_C_NOT_ZERO, label); - - if (max != 0) - { - SLJIT_ASSERT(max - min > 0); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, SLJIT_IMM, max - min); - } - OP1(SLJIT_MOV, tmp_base, tmp_offset, STR_PTR, 0); - label = LABEL(); - compile_char1_matchingpath(common, type, cc, &nomatch); - OP1(SLJIT_MOV, tmp_base, tmp_offset, STR_PTR, 0); - if (max == 0) - JUMPTO(SLJIT_JUMP, label); - else - { - OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, SLJIT_IMM, 1); - JUMPTO(SLJIT_C_NOT_ZERO, label); - } - set_jumps(nomatch, LABEL()); - OP1(SLJIT_MOV, STR_PTR, 0, tmp_base, tmp_offset); - break; - - default: - SLJIT_ASSERT_STOP(); - break; - } - -count_match(common); -return end; -} - -static SLJIT_INLINE pcre_uchar *compile_fail_accept_matchingpath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent) -{ -DEFINE_COMPILER; -backtrack_common *backtrack; - -PUSH_BACKTRACK(sizeof(backtrack_common), cc, NULL); - -if (*cc == OP_FAIL) - { - add_jump(compiler, &backtrack->topbacktracks, JUMP(SLJIT_JUMP)); - return cc + 1; - } - -if (*cc == OP_ASSERT_ACCEPT || common->currententry != NULL) - { - /* No need to check notempty conditions. */ - if (common->accept_label == NULL) - add_jump(compiler, &common->accept, JUMP(SLJIT_JUMP)); - else - JUMPTO(SLJIT_JUMP, common->accept_label); - return cc + 1; - } - -if (common->accept_label == NULL) - add_jump(compiler, &common->accept, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0))); -else - CMPTO(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0), common->accept_label); -OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); -OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, notempty)); -add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); -OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, notempty_atstart)); -if (common->accept_label == NULL) - add_jump(compiler, &common->accept, CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, 0)); -else - CMPTO(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, 0, common->accept_label); -OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); -if (common->accept_label == NULL) - add_jump(compiler, &common->accept, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_PTR, 0)); -else - CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_PTR, 0, common->accept_label); -add_jump(compiler, &backtrack->topbacktracks, JUMP(SLJIT_JUMP)); -return cc + 1; -} - -static SLJIT_INLINE pcre_uchar *compile_close_matchingpath(compiler_common *common, pcre_uchar *cc) -{ -DEFINE_COMPILER; -int offset = GET2(cc, 1); -BOOL optimized_cbracket = common->optimized_cbracket[offset] != 0; - -/* Data will be discarded anyway... */ -if (common->currententry != NULL) - return cc + 1 + IMM2_SIZE; - -if (!optimized_cbracket) - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR_PRIV(offset)); -offset <<= 1; -OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), STR_PTR, 0); -if (!optimized_cbracket) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0); -return cc + 1 + IMM2_SIZE; -} - -static SLJIT_INLINE pcre_uchar *compile_control_verb_matchingpath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent) -{ -DEFINE_COMPILER; -backtrack_common *backtrack; -pcre_uchar opcode = *cc; -pcre_uchar *ccend = cc + 1; - -if (opcode == OP_PRUNE_ARG || opcode == OP_SKIP_ARG || opcode == OP_THEN_ARG) - ccend += 2 + cc[1]; - -PUSH_BACKTRACK(sizeof(backtrack_common), cc, NULL); - -if (opcode == OP_SKIP) - { - allocate_stack(common, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); - return ccend; - } - -if (opcode == OP_PRUNE_ARG || opcode == OP_THEN_ARG) - { - OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); - OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, (sljit_sw)(cc + 2)); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, TMP2, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, mark_ptr), TMP2, 0); - } - -return ccend; -} - -static pcre_uchar then_trap_opcode[1] = { OP_THEN_TRAP }; - -static SLJIT_INLINE void compile_then_trap_matchingpath(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, backtrack_common *parent) -{ -DEFINE_COMPILER; -backtrack_common *backtrack; -BOOL needs_control_head; -int size; - -PUSH_BACKTRACK_NOVALUE(sizeof(then_trap_backtrack), cc); -common->then_trap = BACKTRACK_AS(then_trap_backtrack); -BACKTRACK_AS(then_trap_backtrack)->common.cc = then_trap_opcode; -BACKTRACK_AS(then_trap_backtrack)->start = (sljit_sw)(cc - common->start); -BACKTRACK_AS(then_trap_backtrack)->framesize = get_framesize(common, cc, ccend, FALSE, &needs_control_head); - -size = BACKTRACK_AS(then_trap_backtrack)->framesize; -size = 3 + (size < 0 ? 0 : size); - -OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr); -allocate_stack(common, size); -if (size > 3) - OP2(SLJIT_SUB, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, STACK_TOP, 0, SLJIT_IMM, (size - 3) * sizeof(sljit_sw)); -else - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, STACK_TOP, 0); -OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(size - 1), SLJIT_IMM, BACKTRACK_AS(then_trap_backtrack)->start); -OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(size - 2), SLJIT_IMM, type_then_trap); -OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(size - 3), TMP2, 0); - -size = BACKTRACK_AS(then_trap_backtrack)->framesize; -if (size >= 0) - init_frame(common, cc, ccend, size - 1, 0, FALSE); -} - -static void compile_matchingpath(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, backtrack_common *parent) -{ -DEFINE_COMPILER; -backtrack_common *backtrack; -BOOL has_then_trap = FALSE; -then_trap_backtrack *save_then_trap = NULL; - -SLJIT_ASSERT(*ccend == OP_END || (*ccend >= OP_ALT && *ccend <= OP_KETRPOS)); - -if (common->has_then && common->then_offsets[cc - common->start] != 0) - { - SLJIT_ASSERT(*ccend != OP_END && common->control_head_ptr != 0); - has_then_trap = TRUE; - save_then_trap = common->then_trap; - /* Tail item on backtrack. */ - compile_then_trap_matchingpath(common, cc, ccend, parent); - } - -while (cc < ccend) - { - switch(*cc) - { - case OP_SOD: - case OP_SOM: - case OP_NOT_WORD_BOUNDARY: - case OP_WORD_BOUNDARY: - case OP_NOT_DIGIT: - case OP_DIGIT: - case OP_NOT_WHITESPACE: - case OP_WHITESPACE: - case OP_NOT_WORDCHAR: - case OP_WORDCHAR: - case OP_ANY: - case OP_ALLANY: - case OP_ANYBYTE: - case OP_NOTPROP: - case OP_PROP: - case OP_ANYNL: - case OP_NOT_HSPACE: - case OP_HSPACE: - case OP_NOT_VSPACE: - case OP_VSPACE: - case OP_EXTUNI: - case OP_EODN: - case OP_EOD: - case OP_CIRC: - case OP_CIRCM: - case OP_DOLL: - case OP_DOLLM: - case OP_NOT: - case OP_NOTI: - case OP_REVERSE: - cc = compile_char1_matchingpath(common, *cc, cc + 1, parent->top != NULL ? &parent->top->nextbacktracks : &parent->topbacktracks); - break; - - case OP_SET_SOM: - PUSH_BACKTRACK_NOVALUE(sizeof(backtrack_common), cc); - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); - allocate_stack(common, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0), STR_PTR, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0); - cc++; - break; - - case OP_CHAR: - case OP_CHARI: - if (common->mode == JIT_COMPILE) - cc = compile_charn_matchingpath(common, cc, ccend, parent->top != NULL ? &parent->top->nextbacktracks : &parent->topbacktracks); - else - cc = compile_char1_matchingpath(common, *cc, cc + 1, parent->top != NULL ? &parent->top->nextbacktracks : &parent->topbacktracks); - break; - - case OP_STAR: - case OP_MINSTAR: - case OP_PLUS: - case OP_MINPLUS: - case OP_QUERY: - case OP_MINQUERY: - case OP_UPTO: - case OP_MINUPTO: - case OP_EXACT: - case OP_POSSTAR: - case OP_POSPLUS: - case OP_POSQUERY: - case OP_POSUPTO: - case OP_STARI: - case OP_MINSTARI: - case OP_PLUSI: - case OP_MINPLUSI: - case OP_QUERYI: - case OP_MINQUERYI: - case OP_UPTOI: - case OP_MINUPTOI: - case OP_EXACTI: - case OP_POSSTARI: - case OP_POSPLUSI: - case OP_POSQUERYI: - case OP_POSUPTOI: - case OP_NOTSTAR: - case OP_NOTMINSTAR: - case OP_NOTPLUS: - case OP_NOTMINPLUS: - case OP_NOTQUERY: - case OP_NOTMINQUERY: - case OP_NOTUPTO: - case OP_NOTMINUPTO: - case OP_NOTEXACT: - case OP_NOTPOSSTAR: - case OP_NOTPOSPLUS: - case OP_NOTPOSQUERY: - case OP_NOTPOSUPTO: - case OP_NOTSTARI: - case OP_NOTMINSTARI: - case OP_NOTPLUSI: - case OP_NOTMINPLUSI: - case OP_NOTQUERYI: - case OP_NOTMINQUERYI: - case OP_NOTUPTOI: - case OP_NOTMINUPTOI: - case OP_NOTEXACTI: - case OP_NOTPOSSTARI: - case OP_NOTPOSPLUSI: - case OP_NOTPOSQUERYI: - case OP_NOTPOSUPTOI: - case OP_TYPESTAR: - case OP_TYPEMINSTAR: - case OP_TYPEPLUS: - case OP_TYPEMINPLUS: - case OP_TYPEQUERY: - case OP_TYPEMINQUERY: - case OP_TYPEUPTO: - case OP_TYPEMINUPTO: - case OP_TYPEEXACT: - case OP_TYPEPOSSTAR: - case OP_TYPEPOSPLUS: - case OP_TYPEPOSQUERY: - case OP_TYPEPOSUPTO: - cc = compile_iterator_matchingpath(common, cc, parent); - break; - - case OP_CLASS: - case OP_NCLASS: - if (cc[1 + (32 / sizeof(pcre_uchar))] >= OP_CRSTAR && cc[1 + (32 / sizeof(pcre_uchar))] <= OP_CRPOSRANGE) - cc = compile_iterator_matchingpath(common, cc, parent); - else - cc = compile_char1_matchingpath(common, *cc, cc + 1, parent->top != NULL ? &parent->top->nextbacktracks : &parent->topbacktracks); - break; - -#if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - case OP_XCLASS: - if (*(cc + GET(cc, 1)) >= OP_CRSTAR && *(cc + GET(cc, 1)) <= OP_CRPOSRANGE) - cc = compile_iterator_matchingpath(common, cc, parent); - else - cc = compile_char1_matchingpath(common, *cc, cc + 1, parent->top != NULL ? &parent->top->nextbacktracks : &parent->topbacktracks); - break; -#endif - - case OP_REF: - case OP_REFI: - if (cc[1 + IMM2_SIZE] >= OP_CRSTAR && cc[1 + IMM2_SIZE] <= OP_CRPOSRANGE) - cc = compile_ref_iterator_matchingpath(common, cc, parent); - else - { - compile_ref_matchingpath(common, cc, parent->top != NULL ? &parent->top->nextbacktracks : &parent->topbacktracks, TRUE, FALSE); - cc += 1 + IMM2_SIZE; - } - break; - - case OP_DNREF: - case OP_DNREFI: - if (cc[1 + 2 * IMM2_SIZE] >= OP_CRSTAR && cc[1 + 2 * IMM2_SIZE] <= OP_CRPOSRANGE) - cc = compile_ref_iterator_matchingpath(common, cc, parent); - else - { - compile_dnref_search(common, cc, parent->top != NULL ? &parent->top->nextbacktracks : &parent->topbacktracks); - compile_ref_matchingpath(common, cc, parent->top != NULL ? &parent->top->nextbacktracks : &parent->topbacktracks, TRUE, FALSE); - cc += 1 + 2 * IMM2_SIZE; - } - break; - - case OP_RECURSE: - cc = compile_recurse_matchingpath(common, cc, parent); - break; - - case OP_CALLOUT: - cc = compile_callout_matchingpath(common, cc, parent); - break; - - case OP_ASSERT: - case OP_ASSERT_NOT: - case OP_ASSERTBACK: - case OP_ASSERTBACK_NOT: - PUSH_BACKTRACK_NOVALUE(sizeof(assert_backtrack), cc); - cc = compile_assert_matchingpath(common, cc, BACKTRACK_AS(assert_backtrack), FALSE); - break; - - case OP_BRAMINZERO: - PUSH_BACKTRACK_NOVALUE(sizeof(braminzero_backtrack), cc); - cc = bracketend(cc + 1); - if (*(cc - 1 - LINK_SIZE) != OP_KETRMIN) - { - allocate_stack(common, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); - } - else - { - allocate_stack(common, 2); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), STR_PTR, 0); - } - BACKTRACK_AS(braminzero_backtrack)->matchingpath = LABEL(); - if (cc[1] > OP_ASSERTBACK_NOT) - count_match(common); - break; - - case OP_ONCE: - case OP_ONCE_NC: - case OP_BRA: - case OP_CBRA: - case OP_COND: - case OP_SBRA: - case OP_SCBRA: - case OP_SCOND: - cc = compile_bracket_matchingpath(common, cc, parent); - break; - - case OP_BRAZERO: - if (cc[1] > OP_ASSERTBACK_NOT) - cc = compile_bracket_matchingpath(common, cc, parent); - else - { - PUSH_BACKTRACK_NOVALUE(sizeof(assert_backtrack), cc); - cc = compile_assert_matchingpath(common, cc, BACKTRACK_AS(assert_backtrack), FALSE); - } - break; - - case OP_BRAPOS: - case OP_CBRAPOS: - case OP_SBRAPOS: - case OP_SCBRAPOS: - case OP_BRAPOSZERO: - cc = compile_bracketpos_matchingpath(common, cc, parent); - break; - - case OP_MARK: - PUSH_BACKTRACK_NOVALUE(sizeof(backtrack_common), cc); - SLJIT_ASSERT(common->mark_ptr != 0); - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); - allocate_stack(common, common->has_skip_arg ? 5 : 1); - OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(common->has_skip_arg ? 4 : 0), TMP2, 0); - OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, (sljit_sw)(cc + 2)); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, TMP2, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, mark_ptr), TMP2, 0); - if (common->has_skip_arg) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, STACK_TOP, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, type_mark); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), SLJIT_IMM, (sljit_sw)(cc + 2)); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(3), STR_PTR, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP1, 0); - } - cc += 1 + 2 + cc[1]; - break; - - case OP_PRUNE: - case OP_PRUNE_ARG: - case OP_SKIP: - case OP_SKIP_ARG: - case OP_THEN: - case OP_THEN_ARG: - case OP_COMMIT: - cc = compile_control_verb_matchingpath(common, cc, parent); - break; - - case OP_FAIL: - case OP_ACCEPT: - case OP_ASSERT_ACCEPT: - cc = compile_fail_accept_matchingpath(common, cc, parent); - break; - - case OP_CLOSE: - cc = compile_close_matchingpath(common, cc); - break; - - case OP_SKIPZERO: - cc = bracketend(cc + 1); - break; - - default: - SLJIT_ASSERT_STOP(); - return; - } - if (cc == NULL) - return; - } - -if (has_then_trap) - { - /* Head item on backtrack. */ - PUSH_BACKTRACK_NOVALUE(sizeof(then_trap_backtrack), cc); - BACKTRACK_AS(then_trap_backtrack)->common.cc = then_trap_opcode; - BACKTRACK_AS(then_trap_backtrack)->then_trap = common->then_trap; - common->then_trap = save_then_trap; - } -SLJIT_ASSERT(cc == ccend); -} - -#undef PUSH_BACKTRACK -#undef PUSH_BACKTRACK_NOVALUE -#undef BACKTRACK_AS - -#define COMPILE_BACKTRACKINGPATH(current) \ - do \ - { \ - compile_backtrackingpath(common, (current)); \ - if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) \ - return; \ - } \ - while (0) - -#define CURRENT_AS(type) ((type *)current) - -static void compile_iterator_backtrackingpath(compiler_common *common, struct backtrack_common *current) -{ -DEFINE_COMPILER; -pcre_uchar *cc = current->cc; -pcre_uchar opcode; -pcre_uchar type; -int max = -1, min = -1; -struct sljit_label *label = NULL; -struct sljit_jump *jump = NULL; -jump_list *jumplist = NULL; -int private_data_ptr = PRIVATE_DATA(cc); -int base = (private_data_ptr == 0) ? SLJIT_MEM1(STACK_TOP) : SLJIT_MEM1(SLJIT_LOCALS_REG); -int offset0 = (private_data_ptr == 0) ? STACK(0) : private_data_ptr; -int offset1 = (private_data_ptr == 0) ? STACK(1) : private_data_ptr + (int)sizeof(sljit_sw); - -cc = get_iterator_parameters(common, cc, &opcode, &type, &max, &min, NULL); - -switch(opcode) - { - case OP_STAR: - case OP_PLUS: - case OP_UPTO: - case OP_CRRANGE: - if (type == OP_ANYNL || type == OP_EXTUNI) - { - SLJIT_ASSERT(private_data_ptr == 0); - set_jumps(current->topbacktracks, LABEL()); - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - free_stack(common, 1); - CMPTO(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(iterator_backtrack)->matchingpath); - } - else - { - if (opcode == OP_UPTO) - min = 0; - if (opcode <= OP_PLUS) - { - OP1(SLJIT_MOV, STR_PTR, 0, base, offset0); - jump = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, base, offset1); - } - else - { - OP1(SLJIT_MOV, TMP1, 0, base, offset1); - OP1(SLJIT_MOV, STR_PTR, 0, base, offset0); - jump = CMP(SLJIT_C_LESS_EQUAL, TMP1, 0, SLJIT_IMM, min + 1); - OP2(SLJIT_SUB, base, offset1, TMP1, 0, SLJIT_IMM, 1); - } - skip_char_back(common); - OP1(SLJIT_MOV, base, offset0, STR_PTR, 0); - JUMPTO(SLJIT_JUMP, CURRENT_AS(iterator_backtrack)->matchingpath); - if (opcode == OP_CRRANGE) - set_jumps(current->topbacktracks, LABEL()); - JUMPHERE(jump); - if (private_data_ptr == 0) - free_stack(common, 2); - if (opcode == OP_PLUS) - set_jumps(current->topbacktracks, LABEL()); - } - break; - - case OP_MINSTAR: - case OP_MINPLUS: - OP1(SLJIT_MOV, STR_PTR, 0, base, offset0); - compile_char1_matchingpath(common, type, cc, &jumplist); - OP1(SLJIT_MOV, base, offset0, STR_PTR, 0); - JUMPTO(SLJIT_JUMP, CURRENT_AS(iterator_backtrack)->matchingpath); - set_jumps(jumplist, LABEL()); - if (private_data_ptr == 0) - free_stack(common, 1); - if (opcode == OP_MINPLUS) - set_jumps(current->topbacktracks, LABEL()); - break; - - case OP_MINUPTO: - case OP_CRMINRANGE: - if (opcode == OP_CRMINRANGE) - { - label = LABEL(); - set_jumps(current->topbacktracks, label); - } - OP1(SLJIT_MOV, STR_PTR, 0, base, offset0); - compile_char1_matchingpath(common, type, cc, &jumplist); - - OP1(SLJIT_MOV, TMP1, 0, base, offset1); - OP1(SLJIT_MOV, base, offset0, STR_PTR, 0); - OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); - OP1(SLJIT_MOV, base, offset1, TMP1, 0); - - if (opcode == OP_CRMINRANGE) - CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, min + 1, label); - - if (opcode == OP_CRMINRANGE && max == 0) - JUMPTO(SLJIT_JUMP, CURRENT_AS(iterator_backtrack)->matchingpath); - else - CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, max + 2, CURRENT_AS(iterator_backtrack)->matchingpath); - - set_jumps(jumplist, LABEL()); - if (private_data_ptr == 0) - free_stack(common, 2); - break; - - case OP_QUERY: - OP1(SLJIT_MOV, STR_PTR, 0, base, offset0); - OP1(SLJIT_MOV, base, offset0, SLJIT_IMM, 0); - CMPTO(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(iterator_backtrack)->matchingpath); - jump = JUMP(SLJIT_JUMP); - set_jumps(current->topbacktracks, LABEL()); - OP1(SLJIT_MOV, STR_PTR, 0, base, offset0); - OP1(SLJIT_MOV, base, offset0, SLJIT_IMM, 0); - JUMPTO(SLJIT_JUMP, CURRENT_AS(iterator_backtrack)->matchingpath); - JUMPHERE(jump); - if (private_data_ptr == 0) - free_stack(common, 1); - break; - - case OP_MINQUERY: - OP1(SLJIT_MOV, STR_PTR, 0, base, offset0); - OP1(SLJIT_MOV, base, offset0, SLJIT_IMM, 0); - jump = CMP(SLJIT_C_EQUAL, STR_PTR, 0, SLJIT_IMM, 0); - compile_char1_matchingpath(common, type, cc, &jumplist); - JUMPTO(SLJIT_JUMP, CURRENT_AS(iterator_backtrack)->matchingpath); - set_jumps(jumplist, LABEL()); - JUMPHERE(jump); - if (private_data_ptr == 0) - free_stack(common, 1); - break; - - case OP_EXACT: - case OP_POSPLUS: - case OP_CRPOSRANGE: - set_jumps(current->topbacktracks, LABEL()); - break; - - case OP_POSSTAR: - case OP_POSQUERY: - case OP_POSUPTO: - break; - - default: - SLJIT_ASSERT_STOP(); - break; - } -} - -static SLJIT_INLINE void compile_ref_iterator_backtrackingpath(compiler_common *common, struct backtrack_common *current) -{ -DEFINE_COMPILER; -pcre_uchar *cc = current->cc; -BOOL ref = (*cc == OP_REF || *cc == OP_REFI); -pcre_uchar type; - -type = cc[ref ? 1 + IMM2_SIZE : 1 + 2 * IMM2_SIZE]; - -if ((type & 0x1) == 0) - { - /* Maximize case. */ - set_jumps(current->topbacktracks, LABEL()); - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - free_stack(common, 1); - CMPTO(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(iterator_backtrack)->matchingpath); - return; - } - -OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); -CMPTO(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(iterator_backtrack)->matchingpath); -set_jumps(current->topbacktracks, LABEL()); -free_stack(common, ref ? 2 : 3); -} - -static SLJIT_INLINE void compile_recurse_backtrackingpath(compiler_common *common, struct backtrack_common *current) -{ -DEFINE_COMPILER; - -if (CURRENT_AS(recurse_backtrack)->inlined_pattern) - compile_backtrackingpath(common, current->top); -set_jumps(current->topbacktracks, LABEL()); -if (CURRENT_AS(recurse_backtrack)->inlined_pattern) - return; - -if (common->has_set_som && common->mark_ptr != 0) - { - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); - free_stack(common, 2); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0), TMP2, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, TMP1, 0); - } -else if (common->has_set_som || common->mark_ptr != 0) - { - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - free_stack(common, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->has_set_som ? (int)(OVECTOR(0)) : common->mark_ptr, TMP2, 0); - } -} - -static void compile_assert_backtrackingpath(compiler_common *common, struct backtrack_common *current) -{ -DEFINE_COMPILER; -pcre_uchar *cc = current->cc; -pcre_uchar bra = OP_BRA; -struct sljit_jump *brajump = NULL; - -SLJIT_ASSERT(*cc != OP_BRAMINZERO); -if (*cc == OP_BRAZERO) - { - bra = *cc; - cc++; - } - -if (bra == OP_BRAZERO) - { - SLJIT_ASSERT(current->topbacktracks == NULL); - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - } - -if (CURRENT_AS(assert_backtrack)->framesize < 0) - { - set_jumps(current->topbacktracks, LABEL()); - - if (bra == OP_BRAZERO) - { - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); - CMPTO(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(assert_backtrack)->matchingpath); - free_stack(common, 1); - } - return; - } - -if (bra == OP_BRAZERO) - { - if (*cc == OP_ASSERT_NOT || *cc == OP_ASSERTBACK_NOT) - { - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); - CMPTO(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(assert_backtrack)->matchingpath); - free_stack(common, 1); - return; - } - free_stack(common, 1); - brajump = CMP(SLJIT_C_EQUAL, STR_PTR, 0, SLJIT_IMM, 0); - } - -if (*cc == OP_ASSERT || *cc == OP_ASSERTBACK) - { - OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), CURRENT_AS(assert_backtrack)->private_data_ptr); - add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), CURRENT_AS(assert_backtrack)->private_data_ptr, SLJIT_MEM1(STACK_TOP), CURRENT_AS(assert_backtrack)->framesize * sizeof(sljit_sw)); - - set_jumps(current->topbacktracks, LABEL()); - } -else - set_jumps(current->topbacktracks, LABEL()); - -if (bra == OP_BRAZERO) - { - /* We know there is enough place on the stack. */ - OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_sw)); - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); - JUMPTO(SLJIT_JUMP, CURRENT_AS(assert_backtrack)->matchingpath); - JUMPHERE(brajump); - } -} - -static void compile_bracket_backtrackingpath(compiler_common *common, struct backtrack_common *current) -{ -DEFINE_COMPILER; -int opcode, stacksize, count; -int offset = 0; -int private_data_ptr = CURRENT_AS(bracket_backtrack)->private_data_ptr; -int repeat_ptr = 0, repeat_type = 0, repeat_count = 0; -pcre_uchar *cc = current->cc; -pcre_uchar *ccbegin; -pcre_uchar *ccprev; -jump_list *jumplist = NULL; -jump_list *jumplistitem = NULL; -pcre_uchar bra = OP_BRA; -pcre_uchar ket; -assert_backtrack *assert; -BOOL has_alternatives; -BOOL needs_control_head = FALSE; -struct sljit_jump *brazero = NULL; -struct sljit_jump *once = NULL; -struct sljit_jump *cond = NULL; -struct sljit_label *rmin_label = NULL; -struct sljit_label *exact_label = NULL; - -if (*cc == OP_BRAZERO || *cc == OP_BRAMINZERO) - { - bra = *cc; - cc++; - } - -opcode = *cc; -ccbegin = bracketend(cc) - 1 - LINK_SIZE; -ket = *ccbegin; -if (ket == OP_KET && PRIVATE_DATA(ccbegin) != 0) - { - repeat_ptr = PRIVATE_DATA(ccbegin); - repeat_type = PRIVATE_DATA(ccbegin + 2); - repeat_count = PRIVATE_DATA(ccbegin + 3); - SLJIT_ASSERT(repeat_type != 0 && repeat_count != 0); - if (repeat_type == OP_UPTO) - ket = OP_KETRMAX; - if (repeat_type == OP_MINUPTO) - ket = OP_KETRMIN; - } -ccbegin = cc; -cc += GET(cc, 1); -has_alternatives = *cc == OP_ALT; -if (SLJIT_UNLIKELY(opcode == OP_COND) || SLJIT_UNLIKELY(opcode == OP_SCOND)) - has_alternatives = (ccbegin[1 + LINK_SIZE] >= OP_ASSERT && ccbegin[1 + LINK_SIZE] <= OP_ASSERTBACK_NOT) || CURRENT_AS(bracket_backtrack)->u.condfailed != NULL; -if (opcode == OP_CBRA || opcode == OP_SCBRA) - offset = (GET2(ccbegin, 1 + LINK_SIZE)) << 1; -if (SLJIT_UNLIKELY(opcode == OP_COND) && (*cc == OP_KETRMAX || *cc == OP_KETRMIN)) - opcode = OP_SCOND; -if (SLJIT_UNLIKELY(opcode == OP_ONCE_NC)) - opcode = OP_ONCE; - -/* Decoding the needs_control_head in framesize. */ -if (opcode == OP_ONCE) - { - needs_control_head = (CURRENT_AS(bracket_backtrack)->u.framesize & 0x1) != 0; - CURRENT_AS(bracket_backtrack)->u.framesize >>= 1; - } - -if (ket != OP_KET && repeat_type != 0) - { - /* TMP1 is used in OP_KETRMIN below. */ - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - free_stack(common, 1); - if (repeat_type == OP_UPTO) - OP2(SLJIT_ADD, SLJIT_MEM1(SLJIT_LOCALS_REG), repeat_ptr, TMP1, 0, SLJIT_IMM, 1); - else - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), repeat_ptr, TMP1, 0); - } - -if (ket == OP_KETRMAX) - { - if (bra == OP_BRAZERO) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - free_stack(common, 1); - brazero = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0); - } - } -else if (ket == OP_KETRMIN) - { - if (bra != OP_BRAMINZERO) - { - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - if (repeat_type != 0) - { - /* TMP1 was set a few lines above. */ - CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 0, CURRENT_AS(bracket_backtrack)->recursive_matchingpath); - /* Drop STR_PTR for non-greedy plus quantifier. */ - if (opcode != OP_ONCE) - free_stack(common, 1); - } - else if (opcode >= OP_SBRA || opcode == OP_ONCE) - { - /* Checking zero-length iteration. */ - if (opcode != OP_ONCE || CURRENT_AS(bracket_backtrack)->u.framesize < 0) - CMPTO(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, CURRENT_AS(bracket_backtrack)->recursive_matchingpath); - else - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - CMPTO(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_MEM1(TMP1), (CURRENT_AS(bracket_backtrack)->u.framesize + 1) * sizeof(sljit_sw), CURRENT_AS(bracket_backtrack)->recursive_matchingpath); - } - /* Drop STR_PTR for non-greedy plus quantifier. */ - if (opcode != OP_ONCE) - free_stack(common, 1); - } - else - JUMPTO(SLJIT_JUMP, CURRENT_AS(bracket_backtrack)->recursive_matchingpath); - } - rmin_label = LABEL(); - if (repeat_type != 0) - OP2(SLJIT_ADD, SLJIT_MEM1(SLJIT_LOCALS_REG), repeat_ptr, SLJIT_MEM1(SLJIT_LOCALS_REG), repeat_ptr, SLJIT_IMM, 1); - } -else if (bra == OP_BRAZERO) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - free_stack(common, 1); - brazero = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 0); - } -else if (repeat_type == OP_EXACT) - { - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), repeat_ptr, SLJIT_IMM, 1); - exact_label = LABEL(); - } - -if (offset != 0) - { - if (common->capture_last_ptr != 0) - { - SLJIT_ASSERT(common->optimized_cbracket[offset >> 1] == 0); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr, TMP1, 0); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(2)); - free_stack(common, 3); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP2, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), TMP1, 0); - } - else if (common->optimized_cbracket[offset >> 1] == 0) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); - free_stack(common, 2); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), TMP2, 0); - } - } - -if (SLJIT_UNLIKELY(opcode == OP_ONCE)) - { - if (CURRENT_AS(bracket_backtrack)->u.framesize >= 0) - { - OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); - } - once = JUMP(SLJIT_JUMP); - } -else if (SLJIT_UNLIKELY(opcode == OP_COND) || SLJIT_UNLIKELY(opcode == OP_SCOND)) - { - if (has_alternatives) - { - /* Always exactly one alternative. */ - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - free_stack(common, 1); - - jumplistitem = sljit_alloc_memory(compiler, sizeof(jump_list)); - if (SLJIT_UNLIKELY(!jumplistitem)) - return; - jumplist = jumplistitem; - jumplistitem->next = NULL; - jumplistitem->jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 1); - } - } -else if (*cc == OP_ALT) - { - /* Build a jump list. Get the last successfully matched branch index. */ - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - free_stack(common, 1); - count = 1; - do - { - /* Append as the last item. */ - if (jumplist != NULL) - { - jumplistitem->next = sljit_alloc_memory(compiler, sizeof(jump_list)); - jumplistitem = jumplistitem->next; - } - else - { - jumplistitem = sljit_alloc_memory(compiler, sizeof(jump_list)); - jumplist = jumplistitem; - } - - if (SLJIT_UNLIKELY(!jumplistitem)) - return; - - jumplistitem->next = NULL; - jumplistitem->jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, count++); - cc += GET(cc, 1); - } - while (*cc == OP_ALT); - - cc = ccbegin + GET(ccbegin, 1); - } - -COMPILE_BACKTRACKINGPATH(current->top); -if (current->topbacktracks) - set_jumps(current->topbacktracks, LABEL()); - -if (SLJIT_UNLIKELY(opcode == OP_COND) || SLJIT_UNLIKELY(opcode == OP_SCOND)) - { - /* Conditional block always has at most one alternative. */ - if (ccbegin[1 + LINK_SIZE] >= OP_ASSERT && ccbegin[1 + LINK_SIZE] <= OP_ASSERTBACK_NOT) - { - SLJIT_ASSERT(has_alternatives); - assert = CURRENT_AS(bracket_backtrack)->u.assert; - if (assert->framesize >= 0 && (ccbegin[1 + LINK_SIZE] == OP_ASSERT || ccbegin[1 + LINK_SIZE] == OP_ASSERTBACK)) - { - OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), assert->private_data_ptr); - add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), assert->private_data_ptr, SLJIT_MEM1(STACK_TOP), assert->framesize * sizeof(sljit_sw)); - } - cond = JUMP(SLJIT_JUMP); - set_jumps(CURRENT_AS(bracket_backtrack)->u.assert->condfailed, LABEL()); - } - else if (CURRENT_AS(bracket_backtrack)->u.condfailed != NULL) - { - SLJIT_ASSERT(has_alternatives); - cond = JUMP(SLJIT_JUMP); - set_jumps(CURRENT_AS(bracket_backtrack)->u.condfailed, LABEL()); - } - else - SLJIT_ASSERT(!has_alternatives); - } - -if (has_alternatives) - { - count = 1; - do - { - current->top = NULL; - current->topbacktracks = NULL; - current->nextbacktracks = NULL; - /* Conditional blocks always have an additional alternative, even if it is empty. */ - if (*cc == OP_ALT) - { - ccprev = cc + 1 + LINK_SIZE; - cc += GET(cc, 1); - if (opcode != OP_COND && opcode != OP_SCOND) - { - if (opcode != OP_ONCE) - { - if (private_data_ptr != 0) - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr); - else - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - } - else - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(needs_control_head ? 1 : 0)); - } - compile_matchingpath(common, ccprev, cc, current); - if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) - return; - } - - /* Instructions after the current alternative is successfully matched. */ - /* There is a similar code in compile_bracket_matchingpath. */ - if (opcode == OP_ONCE) - match_once_common(common, ket, CURRENT_AS(bracket_backtrack)->u.framesize, private_data_ptr, has_alternatives, needs_control_head); - - stacksize = 0; - if (repeat_type == OP_MINUPTO) - { - /* We need to preserve the counter. TMP2 will be used below. */ - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), repeat_ptr); - stacksize++; - } - if (ket != OP_KET || bra != OP_BRA) - stacksize++; - if (offset != 0) - { - if (common->capture_last_ptr != 0) - stacksize++; - if (common->optimized_cbracket[offset >> 1] == 0) - stacksize += 2; - } - if (opcode != OP_ONCE) - stacksize++; - - if (stacksize > 0) - allocate_stack(common, stacksize); - - stacksize = 0; - if (repeat_type == OP_MINUPTO) - { - /* TMP2 was set above. */ - OP2(SLJIT_SUB, SLJIT_MEM1(STACK_TOP), STACK(stacksize), TMP2, 0, SLJIT_IMM, 1); - stacksize++; - } - - if (ket != OP_KET || bra != OP_BRA) - { - if (ket != OP_KET) - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), STR_PTR, 0); - else - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), SLJIT_IMM, 0); - stacksize++; - } - - if (offset != 0) - stacksize = match_capture_common(common, stacksize, offset, private_data_ptr); - - if (opcode != OP_ONCE) - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), SLJIT_IMM, count++); - - if (offset != 0 && ket == OP_KETRMAX && common->optimized_cbracket[offset >> 1] != 0) - { - /* If ket is not OP_KETRMAX, this code path is executed after the jump to alternative_matchingpath. */ - SLJIT_ASSERT(private_data_ptr == OVECTOR(offset + 0)); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), STR_PTR, 0); - } - - JUMPTO(SLJIT_JUMP, CURRENT_AS(bracket_backtrack)->alternative_matchingpath); - - if (opcode != OP_ONCE) - { - SLJIT_ASSERT(jumplist); - JUMPHERE(jumplist->jump); - jumplist = jumplist->next; - } - - COMPILE_BACKTRACKINGPATH(current->top); - if (current->topbacktracks) - set_jumps(current->topbacktracks, LABEL()); - SLJIT_ASSERT(!current->nextbacktracks); - } - while (*cc == OP_ALT); - SLJIT_ASSERT(!jumplist); - - if (cond != NULL) - { - SLJIT_ASSERT(opcode == OP_COND || opcode == OP_SCOND); - assert = CURRENT_AS(bracket_backtrack)->u.assert; - if ((ccbegin[1 + LINK_SIZE] == OP_ASSERT_NOT || ccbegin[1 + LINK_SIZE] == OP_ASSERTBACK_NOT) && assert->framesize >= 0) - { - OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), assert->private_data_ptr); - add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), assert->private_data_ptr, SLJIT_MEM1(STACK_TOP), assert->framesize * sizeof(sljit_sw)); - } - JUMPHERE(cond); - } - - /* Free the STR_PTR. */ - if (private_data_ptr == 0) - free_stack(common, 1); - } - -if (offset != 0) - { - /* Using both tmp register is better for instruction scheduling. */ - if (common->optimized_cbracket[offset >> 1] != 0) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); - free_stack(common, 2); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), TMP2, 0); - } - else - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - free_stack(common, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, TMP1, 0); - } - } -else if (opcode == OP_SBRA || opcode == OP_SCOND) - { - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, SLJIT_MEM1(STACK_TOP), STACK(0)); - free_stack(common, 1); - } -else if (opcode == OP_ONCE) - { - cc = ccbegin + GET(ccbegin, 1); - stacksize = needs_control_head ? 1 : 0; - - if (CURRENT_AS(bracket_backtrack)->u.framesize >= 0) - { - /* Reset head and drop saved frame. */ - stacksize += CURRENT_AS(bracket_backtrack)->u.framesize + ((ket != OP_KET || *cc == OP_ALT) ? 2 : 1); - } - else if (ket == OP_KETRMAX || (*cc == OP_ALT && ket != OP_KETRMIN)) - { - /* The STR_PTR must be released. */ - stacksize++; - } - free_stack(common, stacksize); - - JUMPHERE(once); - /* Restore previous private_data_ptr */ - if (CURRENT_AS(bracket_backtrack)->u.framesize >= 0) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, SLJIT_MEM1(STACK_TOP), CURRENT_AS(bracket_backtrack)->u.framesize * sizeof(sljit_sw)); - else if (ket == OP_KETRMIN) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); - /* See the comment below. */ - free_stack(common, 2); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), private_data_ptr, TMP1, 0); - } - } - -if (repeat_type == OP_EXACT) - { - OP2(SLJIT_ADD, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), repeat_ptr, SLJIT_IMM, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), repeat_ptr, TMP1, 0); - CMPTO(SLJIT_C_LESS_EQUAL, TMP1, 0, SLJIT_IMM, repeat_count, exact_label); - } -else if (ket == OP_KETRMAX) - { - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - if (bra != OP_BRAZERO) - free_stack(common, 1); - - CMPTO(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(bracket_backtrack)->recursive_matchingpath); - if (bra == OP_BRAZERO) - { - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); - JUMPTO(SLJIT_JUMP, CURRENT_AS(bracket_backtrack)->zero_matchingpath); - JUMPHERE(brazero); - free_stack(common, 1); - } - } -else if (ket == OP_KETRMIN) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - - /* OP_ONCE removes everything in case of a backtrack, so we don't - need to explicitly release the STR_PTR. The extra release would - affect badly the free_stack(2) above. */ - if (opcode != OP_ONCE) - free_stack(common, 1); - CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 0, rmin_label); - if (opcode == OP_ONCE) - free_stack(common, bra == OP_BRAMINZERO ? 2 : 1); - else if (bra == OP_BRAMINZERO) - free_stack(common, 1); - } -else if (bra == OP_BRAZERO) - { - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - JUMPTO(SLJIT_JUMP, CURRENT_AS(bracket_backtrack)->zero_matchingpath); - JUMPHERE(brazero); - } -} - -static SLJIT_INLINE void compile_bracketpos_backtrackingpath(compiler_common *common, struct backtrack_common *current) -{ -DEFINE_COMPILER; -int offset; -struct sljit_jump *jump; - -if (CURRENT_AS(bracketpos_backtrack)->framesize < 0) - { - if (*current->cc == OP_CBRAPOS || *current->cc == OP_SCBRAPOS) - { - offset = (GET2(current->cc, 1 + LINK_SIZE)) << 1; - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0); - if (common->capture_last_ptr != 0) - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(2)); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), TMP2, 0); - if (common->capture_last_ptr != 0) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr, TMP1, 0); - } - set_jumps(current->topbacktracks, LABEL()); - free_stack(common, CURRENT_AS(bracketpos_backtrack)->stacksize); - return; - } - -OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), CURRENT_AS(bracketpos_backtrack)->private_data_ptr); -add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); - -if (current->topbacktracks) - { - jump = JUMP(SLJIT_JUMP); - set_jumps(current->topbacktracks, LABEL()); - /* Drop the stack frame. */ - free_stack(common, CURRENT_AS(bracketpos_backtrack)->stacksize); - JUMPHERE(jump); - } -OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), CURRENT_AS(bracketpos_backtrack)->private_data_ptr, SLJIT_MEM1(STACK_TOP), CURRENT_AS(bracketpos_backtrack)->framesize * sizeof(sljit_sw)); -} - -static SLJIT_INLINE void compile_braminzero_backtrackingpath(compiler_common *common, struct backtrack_common *current) -{ -assert_backtrack backtrack; - -current->top = NULL; -current->topbacktracks = NULL; -current->nextbacktracks = NULL; -if (current->cc[1] > OP_ASSERTBACK_NOT) - { - /* Manual call of compile_bracket_matchingpath and compile_bracket_backtrackingpath. */ - compile_bracket_matchingpath(common, current->cc, current); - compile_bracket_backtrackingpath(common, current->top); - } -else - { - memset(&backtrack, 0, sizeof(backtrack)); - backtrack.common.cc = current->cc; - backtrack.matchingpath = CURRENT_AS(braminzero_backtrack)->matchingpath; - /* Manual call of compile_assert_matchingpath. */ - compile_assert_matchingpath(common, current->cc, &backtrack, FALSE); - } -SLJIT_ASSERT(!current->nextbacktracks && !current->topbacktracks); -} - -static SLJIT_INLINE void compile_control_verb_backtrackingpath(compiler_common *common, struct backtrack_common *current) -{ -DEFINE_COMPILER; -pcre_uchar opcode = *current->cc; -struct sljit_label *loop; -struct sljit_jump *jump; - -if (opcode == OP_THEN || opcode == OP_THEN_ARG) - { - if (common->then_trap != NULL) - { - SLJIT_ASSERT(common->control_head_ptr != 0); - - OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, type_then_trap); - OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, common->then_trap->start); - jump = JUMP(SLJIT_JUMP); - - loop = LABEL(); - OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), -(int)sizeof(sljit_sw)); - JUMPHERE(jump); - CMPTO(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(STACK_TOP), -(int)(2 * sizeof(sljit_sw)), TMP1, 0, loop); - CMPTO(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(STACK_TOP), -(int)(3 * sizeof(sljit_sw)), TMP2, 0, loop); - add_jump(compiler, &common->then_trap->quit, JUMP(SLJIT_JUMP)); - return; - } - else if (common->positive_assert) - { - add_jump(compiler, &common->positive_assert_quit, JUMP(SLJIT_JUMP)); - return; - } - } - -if (common->local_exit) - { - if (common->quit_label == NULL) - add_jump(compiler, &common->quit, JUMP(SLJIT_JUMP)); - else - JUMPTO(SLJIT_JUMP, common->quit_label); - return; - } - -if (opcode == OP_SKIP_ARG) - { - SLJIT_ASSERT(common->control_head_ptr != 0); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, STACK_TOP, 0); - OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_IMM, (sljit_sw)(current->cc + 2)); - sljit_emit_ijump(compiler, SLJIT_CALL2, SLJIT_IMM, SLJIT_FUNC_OFFSET(do_search_mark)); - OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); - - OP1(SLJIT_MOV, STR_PTR, 0, TMP1, 0); - add_jump(compiler, &common->reset_match, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, -1)); - return; - } - -if (opcode == OP_SKIP) - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); -else - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_IMM, 0); -add_jump(compiler, &common->reset_match, JUMP(SLJIT_JUMP)); -} - -static SLJIT_INLINE void compile_then_trap_backtrackingpath(compiler_common *common, struct backtrack_common *current) -{ -DEFINE_COMPILER; -struct sljit_jump *jump; -int size; - -if (CURRENT_AS(then_trap_backtrack)->then_trap) - { - common->then_trap = CURRENT_AS(then_trap_backtrack)->then_trap; - return; - } - -size = CURRENT_AS(then_trap_backtrack)->framesize; -size = 3 + (size < 0 ? 0 : size); - -OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(size - 3)); -free_stack(common, size); -jump = JUMP(SLJIT_JUMP); - -set_jumps(CURRENT_AS(then_trap_backtrack)->quit, LABEL()); -/* STACK_TOP is set by THEN. */ -if (CURRENT_AS(then_trap_backtrack)->framesize >= 0) - add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); -OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); -free_stack(common, 3); - -JUMPHERE(jump); -OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, TMP1, 0); -} - -static void compile_backtrackingpath(compiler_common *common, struct backtrack_common *current) -{ -DEFINE_COMPILER; -then_trap_backtrack *save_then_trap = common->then_trap; - -while (current) - { - if (current->nextbacktracks != NULL) - set_jumps(current->nextbacktracks, LABEL()); - switch(*current->cc) - { - case OP_SET_SOM: - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - free_stack(common, 1); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0), TMP1, 0); - break; - - case OP_STAR: - case OP_MINSTAR: - case OP_PLUS: - case OP_MINPLUS: - case OP_QUERY: - case OP_MINQUERY: - case OP_UPTO: - case OP_MINUPTO: - case OP_EXACT: - case OP_POSSTAR: - case OP_POSPLUS: - case OP_POSQUERY: - case OP_POSUPTO: - case OP_STARI: - case OP_MINSTARI: - case OP_PLUSI: - case OP_MINPLUSI: - case OP_QUERYI: - case OP_MINQUERYI: - case OP_UPTOI: - case OP_MINUPTOI: - case OP_EXACTI: - case OP_POSSTARI: - case OP_POSPLUSI: - case OP_POSQUERYI: - case OP_POSUPTOI: - case OP_NOTSTAR: - case OP_NOTMINSTAR: - case OP_NOTPLUS: - case OP_NOTMINPLUS: - case OP_NOTQUERY: - case OP_NOTMINQUERY: - case OP_NOTUPTO: - case OP_NOTMINUPTO: - case OP_NOTEXACT: - case OP_NOTPOSSTAR: - case OP_NOTPOSPLUS: - case OP_NOTPOSQUERY: - case OP_NOTPOSUPTO: - case OP_NOTSTARI: - case OP_NOTMINSTARI: - case OP_NOTPLUSI: - case OP_NOTMINPLUSI: - case OP_NOTQUERYI: - case OP_NOTMINQUERYI: - case OP_NOTUPTOI: - case OP_NOTMINUPTOI: - case OP_NOTEXACTI: - case OP_NOTPOSSTARI: - case OP_NOTPOSPLUSI: - case OP_NOTPOSQUERYI: - case OP_NOTPOSUPTOI: - case OP_TYPESTAR: - case OP_TYPEMINSTAR: - case OP_TYPEPLUS: - case OP_TYPEMINPLUS: - case OP_TYPEQUERY: - case OP_TYPEMINQUERY: - case OP_TYPEUPTO: - case OP_TYPEMINUPTO: - case OP_TYPEEXACT: - case OP_TYPEPOSSTAR: - case OP_TYPEPOSPLUS: - case OP_TYPEPOSQUERY: - case OP_TYPEPOSUPTO: - case OP_CLASS: - case OP_NCLASS: -#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 - case OP_XCLASS: -#endif - compile_iterator_backtrackingpath(common, current); - break; - - case OP_REF: - case OP_REFI: - case OP_DNREF: - case OP_DNREFI: - compile_ref_iterator_backtrackingpath(common, current); - break; - - case OP_RECURSE: - compile_recurse_backtrackingpath(common, current); - break; - - case OP_ASSERT: - case OP_ASSERT_NOT: - case OP_ASSERTBACK: - case OP_ASSERTBACK_NOT: - compile_assert_backtrackingpath(common, current); - break; - - case OP_ONCE: - case OP_ONCE_NC: - case OP_BRA: - case OP_CBRA: - case OP_COND: - case OP_SBRA: - case OP_SCBRA: - case OP_SCOND: - compile_bracket_backtrackingpath(common, current); - break; - - case OP_BRAZERO: - if (current->cc[1] > OP_ASSERTBACK_NOT) - compile_bracket_backtrackingpath(common, current); - else - compile_assert_backtrackingpath(common, current); - break; - - case OP_BRAPOS: - case OP_CBRAPOS: - case OP_SBRAPOS: - case OP_SCBRAPOS: - case OP_BRAPOSZERO: - compile_bracketpos_backtrackingpath(common, current); - break; - - case OP_BRAMINZERO: - compile_braminzero_backtrackingpath(common, current); - break; - - case OP_MARK: - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(common->has_skip_arg ? 4 : 0)); - if (common->has_skip_arg) - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - free_stack(common, common->has_skip_arg ? 5 : 1); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, TMP1, 0); - if (common->has_skip_arg) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, TMP2, 0); - break; - - case OP_THEN: - case OP_THEN_ARG: - case OP_PRUNE: - case OP_PRUNE_ARG: - case OP_SKIP: - case OP_SKIP_ARG: - compile_control_verb_backtrackingpath(common, current); - break; - - case OP_COMMIT: - if (!common->local_exit) - OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_NOMATCH); - if (common->quit_label == NULL) - add_jump(compiler, &common->quit, JUMP(SLJIT_JUMP)); - else - JUMPTO(SLJIT_JUMP, common->quit_label); - break; - - case OP_CALLOUT: - case OP_FAIL: - case OP_ACCEPT: - case OP_ASSERT_ACCEPT: - set_jumps(current->topbacktracks, LABEL()); - break; - - case OP_THEN_TRAP: - /* A virtual opcode for then traps. */ - compile_then_trap_backtrackingpath(common, current); - break; - - default: - SLJIT_ASSERT_STOP(); - break; - } - current = current->prev; - } -common->then_trap = save_then_trap; -} - -static SLJIT_INLINE void compile_recurse(compiler_common *common) -{ -DEFINE_COMPILER; -pcre_uchar *cc = common->start + common->currententry->start; -pcre_uchar *ccbegin = cc + 1 + LINK_SIZE + (*cc == OP_BRA ? 0 : IMM2_SIZE); -pcre_uchar *ccend = bracketend(cc); -BOOL needs_control_head; -int framesize = get_framesize(common, cc, NULL, TRUE, &needs_control_head); -int private_data_size = get_private_data_copy_length(common, ccbegin, ccend, needs_control_head); -int alternativesize; -BOOL needs_frame; -backtrack_common altbacktrack; -struct sljit_jump *jump; - -/* Recurse captures then. */ -common->then_trap = NULL; - -SLJIT_ASSERT(*cc == OP_BRA || *cc == OP_CBRA || *cc == OP_CBRAPOS || *cc == OP_SCBRA || *cc == OP_SCBRAPOS); -needs_frame = framesize >= 0; -if (!needs_frame) - framesize = 0; -alternativesize = *(cc + GET(cc, 1)) == OP_ALT ? 1 : 0; - -SLJIT_ASSERT(common->currententry->entry == NULL && common->recursive_head_ptr != 0); -common->currententry->entry = LABEL(); -set_jumps(common->currententry->calls, common->currententry->entry); - -sljit_emit_fast_enter(compiler, TMP2, 0); -allocate_stack(common, private_data_size + framesize + alternativesize); -OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(private_data_size + framesize + alternativesize - 1), TMP2, 0); -copy_private_data(common, ccbegin, ccend, TRUE, private_data_size + framesize + alternativesize, framesize + alternativesize, needs_control_head); -if (needs_control_head) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, SLJIT_IMM, 0); -OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->recursive_head_ptr, STACK_TOP, 0); -if (needs_frame) - init_frame(common, cc, NULL, framesize + alternativesize - 1, alternativesize, TRUE); - -if (alternativesize > 0) - OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); - -memset(&altbacktrack, 0, sizeof(backtrack_common)); -common->quit_label = NULL; -common->accept_label = NULL; -common->quit = NULL; -common->accept = NULL; -altbacktrack.cc = ccbegin; -cc += GET(cc, 1); -while (1) - { - altbacktrack.top = NULL; - altbacktrack.topbacktracks = NULL; - - if (altbacktrack.cc != ccbegin) - OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); - - compile_matchingpath(common, altbacktrack.cc, cc, &altbacktrack); - if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) - return; - - add_jump(compiler, &common->accept, JUMP(SLJIT_JUMP)); - - compile_backtrackingpath(common, altbacktrack.top); - if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) - return; - set_jumps(altbacktrack.topbacktracks, LABEL()); - - if (*cc != OP_ALT) - break; - - altbacktrack.cc = cc + 1 + LINK_SIZE; - cc += GET(cc, 1); - } - -/* None of them matched. */ -OP1(SLJIT_MOV, TMP3, 0, SLJIT_IMM, 0); -jump = JUMP(SLJIT_JUMP); - -if (common->quit != NULL) - { - set_jumps(common->quit, LABEL()); - OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->recursive_head_ptr); - if (needs_frame) - { - OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, (framesize + alternativesize) * sizeof(sljit_sw)); - add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); - OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, (framesize + alternativesize) * sizeof(sljit_sw)); - } - OP1(SLJIT_MOV, TMP3, 0, SLJIT_IMM, 0); - common->quit = NULL; - add_jump(compiler, &common->quit, JUMP(SLJIT_JUMP)); - } - -set_jumps(common->accept, LABEL()); -OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->recursive_head_ptr); -if (needs_frame) - { - OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, (framesize + alternativesize) * sizeof(sljit_sw)); - add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); - OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, (framesize + alternativesize) * sizeof(sljit_sw)); - } -OP1(SLJIT_MOV, TMP3, 0, SLJIT_IMM, 1); - -JUMPHERE(jump); -if (common->quit != NULL) - set_jumps(common->quit, LABEL()); -copy_private_data(common, ccbegin, ccend, FALSE, private_data_size + framesize + alternativesize, framesize + alternativesize, needs_control_head); -free_stack(common, private_data_size + framesize + alternativesize); -if (needs_control_head) - { - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), 2 * sizeof(sljit_sw)); - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), sizeof(sljit_sw)); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->recursive_head_ptr, TMP1, 0); - OP1(SLJIT_MOV, TMP1, 0, TMP3, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, TMP2, 0); - } -else - { - OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), sizeof(sljit_sw)); - OP1(SLJIT_MOV, TMP1, 0, TMP3, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->recursive_head_ptr, TMP2, 0); - } -sljit_emit_fast_return(compiler, SLJIT_MEM1(STACK_TOP), 0); -} - -#undef COMPILE_BACKTRACKINGPATH -#undef CURRENT_AS - -void -PRIV(jit_compile)(const REAL_PCRE *re, PUBL(extra) *extra, int mode) -{ -struct sljit_compiler *compiler; -backtrack_common rootbacktrack; -compiler_common common_data; -compiler_common *common = &common_data; -const pcre_uint8 *tables = re->tables; -pcre_study_data *study; -int private_data_size; -pcre_uchar *ccend; -executable_functions *functions; -void *executable_func; -sljit_uw executable_size; -struct sljit_label *mainloop_label = NULL; -struct sljit_label *continue_match_label; -struct sljit_label *empty_match_found_label; -struct sljit_label *empty_match_backtrack_label; -struct sljit_label *reset_match_label; -struct sljit_jump *jump; -struct sljit_jump *minlength_check_failed = NULL; -struct sljit_jump *reqbyte_notfound = NULL; -struct sljit_jump *empty_match; -struct sljit_label *quit_label; - -SLJIT_ASSERT((extra->flags & PCRE_EXTRA_STUDY_DATA) != 0); -study = extra->study_data; - -if (!tables) - tables = PRIV(default_tables); - -memset(&rootbacktrack, 0, sizeof(backtrack_common)); -memset(common, 0, sizeof(compiler_common)); -rootbacktrack.cc = (pcre_uchar *)re + re->name_table_offset + re->name_count * re->name_entry_size; - -common->start = rootbacktrack.cc; -common->fcc = tables + fcc_offset; -common->lcc = (sljit_sw)(tables + lcc_offset); -common->mode = mode; -common->nltype = NLTYPE_FIXED; -switch(re->options & PCRE_NEWLINE_BITS) - { - case 0: - /* Compile-time default */ - switch(NEWLINE) - { - case -1: common->newline = (CHAR_CR << 8) | CHAR_NL; common->nltype = NLTYPE_ANY; break; - case -2: common->newline = (CHAR_CR << 8) | CHAR_NL; common->nltype = NLTYPE_ANYCRLF; break; - default: common->newline = NEWLINE; break; - } - break; - case PCRE_NEWLINE_CR: common->newline = CHAR_CR; break; - case PCRE_NEWLINE_LF: common->newline = CHAR_NL; break; - case PCRE_NEWLINE_CR+ - PCRE_NEWLINE_LF: common->newline = (CHAR_CR << 8) | CHAR_NL; break; - case PCRE_NEWLINE_ANY: common->newline = (CHAR_CR << 8) | CHAR_NL; common->nltype = NLTYPE_ANY; break; - case PCRE_NEWLINE_ANYCRLF: common->newline = (CHAR_CR << 8) | CHAR_NL; common->nltype = NLTYPE_ANYCRLF; break; - default: return; - } -if ((re->options & PCRE_BSR_ANYCRLF) != 0) - common->bsr_nltype = NLTYPE_ANYCRLF; -else if ((re->options & PCRE_BSR_UNICODE) != 0) - common->bsr_nltype = NLTYPE_ANY; -else - { -#ifdef BSR_ANYCRLF - common->bsr_nltype = NLTYPE_ANYCRLF; -#else - common->bsr_nltype = NLTYPE_ANY; -#endif - } -common->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0; -common->ctypes = (sljit_sw)(tables + ctypes_offset); -common->digits[0] = -2; -common->name_table = ((pcre_uchar *)re) + re->name_table_offset; -common->name_count = re->name_count; -common->name_entry_size = re->name_entry_size; -common->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0; -#ifdef SUPPORT_UTF -/* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */ -common->utf = (re->options & PCRE_UTF8) != 0; -#ifdef SUPPORT_UCP -common->use_ucp = (re->options & PCRE_UCP) != 0; -#endif -#endif /* SUPPORT_UTF */ -ccend = bracketend(rootbacktrack.cc); - -/* Calculate the local space size on the stack. */ -common->ovector_start = LIMIT_MATCH + sizeof(sljit_sw); -common->optimized_cbracket = (pcre_uint8 *)SLJIT_MALLOC(re->top_bracket + 1); -if (!common->optimized_cbracket) - return; -#if defined DEBUG_FORCE_UNOPTIMIZED_CBRAS && DEBUG_FORCE_UNOPTIMIZED_CBRAS == 1 -memset(common->optimized_cbracket, 0, re->top_bracket + 1); -#else -memset(common->optimized_cbracket, 1, re->top_bracket + 1); -#endif - -SLJIT_ASSERT(*rootbacktrack.cc == OP_BRA && ccend[-(1 + LINK_SIZE)] == OP_KET); -#if defined DEBUG_FORCE_UNOPTIMIZED_CBRAS && DEBUG_FORCE_UNOPTIMIZED_CBRAS == 2 -common->capture_last_ptr = common->ovector_start; -common->ovector_start += sizeof(sljit_sw); -#endif -if (!check_opcode_types(common, rootbacktrack.cc, ccend)) - { - SLJIT_FREE(common->optimized_cbracket); - return; - } - -/* Checking flags and updating ovector_start. */ -if (mode == JIT_COMPILE && (re->flags & PCRE_REQCHSET) != 0 && (re->options & PCRE_NO_START_OPTIMIZE) == 0) - { - common->req_char_ptr = common->ovector_start; - common->ovector_start += sizeof(sljit_sw); - } -if (mode != JIT_COMPILE) - { - common->start_used_ptr = common->ovector_start; - common->ovector_start += sizeof(sljit_sw); - if (mode == JIT_PARTIAL_SOFT_COMPILE) - { - common->hit_start = common->ovector_start; - common->ovector_start += 2 * sizeof(sljit_sw); - } - else - { - SLJIT_ASSERT(mode == JIT_PARTIAL_HARD_COMPILE); - common->needs_start_ptr = TRUE; - } - } -if ((re->options & PCRE_FIRSTLINE) != 0) - { - common->first_line_end = common->ovector_start; - common->ovector_start += sizeof(sljit_sw); - } -#if defined DEBUG_FORCE_CONTROL_HEAD && DEBUG_FORCE_CONTROL_HEAD -common->control_head_ptr = 1; -#endif -if (common->control_head_ptr != 0) - { - common->control_head_ptr = common->ovector_start; - common->ovector_start += sizeof(sljit_sw); - } -if (common->needs_start_ptr && common->has_set_som) - { - /* Saving the real start pointer is necessary. */ - common->start_ptr = common->ovector_start; - common->ovector_start += sizeof(sljit_sw); - } -else - common->needs_start_ptr = FALSE; - -/* Aligning ovector to even number of sljit words. */ -if ((common->ovector_start & sizeof(sljit_sw)) != 0) - common->ovector_start += sizeof(sljit_sw); - -if (common->start_ptr == 0) - common->start_ptr = OVECTOR(0); - -/* Capturing brackets cannot be optimized if callouts are allowed. */ -if (common->capture_last_ptr != 0) - memset(common->optimized_cbracket, 0, re->top_bracket + 1); - -SLJIT_ASSERT(!(common->req_char_ptr != 0 && common->start_used_ptr != 0)); -common->cbra_ptr = OVECTOR_START + (re->top_bracket + 1) * 2 * sizeof(sljit_sw); - -common->private_data_ptrs = (int *)SLJIT_MALLOC((ccend - rootbacktrack.cc) * sizeof(sljit_si)); -if (!common->private_data_ptrs) - { - SLJIT_FREE(common->optimized_cbracket); - return; - } -memset(common->private_data_ptrs, 0, (ccend - rootbacktrack.cc) * sizeof(int)); - -private_data_size = common->cbra_ptr + (re->top_bracket + 1) * sizeof(sljit_sw); -set_private_data_ptrs(common, &private_data_size, ccend); -if (private_data_size > SLJIT_MAX_LOCAL_SIZE) - { - SLJIT_FREE(common->private_data_ptrs); - SLJIT_FREE(common->optimized_cbracket); - return; - } - -if (common->has_then) - { - common->then_offsets = (pcre_uint8 *)SLJIT_MALLOC(ccend - rootbacktrack.cc); - if (!common->then_offsets) - { - SLJIT_FREE(common->optimized_cbracket); - SLJIT_FREE(common->private_data_ptrs); - return; - } - memset(common->then_offsets, 0, ccend - rootbacktrack.cc); - set_then_offsets(common, rootbacktrack.cc, NULL); - } - -compiler = sljit_create_compiler(); -if (!compiler) - { - SLJIT_FREE(common->optimized_cbracket); - SLJIT_FREE(common->private_data_ptrs); - if (common->has_then) - SLJIT_FREE(common->then_offsets); - return; - } -common->compiler = compiler; - -/* Main pcre_jit_exec entry. */ -sljit_emit_enter(compiler, 1, 5, 5, private_data_size); - -/* Register init. */ -reset_ovector(common, (re->top_bracket + 1) * 2); -if (common->req_char_ptr != 0) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->req_char_ptr, SLJIT_SCRATCH_REG1, 0); - -OP1(SLJIT_MOV, ARGUMENTS, 0, SLJIT_SAVED_REG1, 0); -OP1(SLJIT_MOV, TMP1, 0, SLJIT_SAVED_REG1, 0); -OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); -OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, end)); -OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, stack)); -OP1(SLJIT_MOV_UI, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, limit_match)); -OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(struct sljit_stack, base)); -OP1(SLJIT_MOV, STACK_LIMIT, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(struct sljit_stack, limit)); -OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LIMIT_MATCH, TMP1, 0); - -if (mode == JIT_PARTIAL_SOFT_COMPILE) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1); -if (common->mark_ptr != 0) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, SLJIT_IMM, 0); -if (common->control_head_ptr != 0) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, SLJIT_IMM, 0); - -/* Main part of the matching */ -if ((re->options & PCRE_ANCHORED) == 0) - { - mainloop_label = mainloop_entry(common, (re->flags & PCRE_HASCRORLF) != 0, (re->options & PCRE_FIRSTLINE) != 0); - continue_match_label = LABEL(); - /* Forward search if possible. */ - if ((re->options & PCRE_NO_START_OPTIMIZE) == 0) - { - if (mode == JIT_COMPILE && fast_forward_first_n_chars(common, (re->options & PCRE_FIRSTLINE) != 0)) - { /* Do nothing */ } - else if ((re->flags & PCRE_FIRSTSET) != 0) - fast_forward_first_char(common, (pcre_uchar)re->first_char, (re->flags & PCRE_FCH_CASELESS) != 0, (re->options & PCRE_FIRSTLINE) != 0); - else if ((re->flags & PCRE_STARTLINE) != 0) - fast_forward_newline(common, (re->options & PCRE_FIRSTLINE) != 0); - else if ((re->flags & PCRE_STARTLINE) == 0 && study != NULL && (study->flags & PCRE_STUDY_MAPPED) != 0) - fast_forward_start_bits(common, (sljit_uw)study->start_bits, (re->options & PCRE_FIRSTLINE) != 0); - } - } -else - continue_match_label = LABEL(); - -if (mode == JIT_COMPILE && study->minlength > 0 && (re->options & PCRE_NO_START_OPTIMIZE) == 0) - { - OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_NOMATCH); - OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(study->minlength)); - minlength_check_failed = CMP(SLJIT_C_GREATER, TMP2, 0, STR_END, 0); - } -if (common->req_char_ptr != 0) - reqbyte_notfound = search_requested_char(common, (pcre_uchar)re->req_char, (re->flags & PCRE_RCH_CASELESS) != 0, (re->flags & PCRE_FIRSTSET) != 0); - -/* Store the current STR_PTR in OVECTOR(0). */ -OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0), STR_PTR, 0); -/* Copy the limit of allowed recursions. */ -OP1(SLJIT_MOV, COUNT_MATCH, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LIMIT_MATCH); -if (common->capture_last_ptr != 0) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->capture_last_ptr, SLJIT_IMM, -1); - -if (common->needs_start_ptr) - { - SLJIT_ASSERT(common->start_ptr != OVECTOR(0)); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_ptr, STR_PTR, 0); - } -else - SLJIT_ASSERT(common->start_ptr == OVECTOR(0)); - -/* Copy the beginning of the string. */ -if (mode == JIT_PARTIAL_SOFT_COMPILE) - { - jump = CMP(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start + sizeof(sljit_sw), STR_PTR, 0); - JUMPHERE(jump); - } -else if (mode == JIT_PARTIAL_HARD_COMPILE) - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); - -compile_matchingpath(common, rootbacktrack.cc, ccend, &rootbacktrack); -if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) - { - sljit_free_compiler(compiler); - SLJIT_FREE(common->optimized_cbracket); - SLJIT_FREE(common->private_data_ptrs); - if (common->has_then) - SLJIT_FREE(common->then_offsets); - return; - } - -empty_match = CMP(SLJIT_C_EQUAL, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); -empty_match_found_label = LABEL(); - -common->accept_label = LABEL(); -if (common->accept != NULL) - set_jumps(common->accept, common->accept_label); - -/* This means we have a match. Update the ovector. */ -copy_ovector(common, re->top_bracket + 1); -common->quit_label = common->forced_quit_label = LABEL(); -if (common->quit != NULL) - set_jumps(common->quit, common->quit_label); -if (common->forced_quit != NULL) - set_jumps(common->forced_quit, common->forced_quit_label); -if (minlength_check_failed != NULL) - SET_LABEL(minlength_check_failed, common->forced_quit_label); -sljit_emit_return(compiler, SLJIT_MOV, SLJIT_RETURN_REG, 0); - -if (mode != JIT_COMPILE) - { - common->partialmatchlabel = LABEL(); - set_jumps(common->partialmatch, common->partialmatchlabel); - return_with_partial_match(common, common->quit_label); - } - -empty_match_backtrack_label = LABEL(); -compile_backtrackingpath(common, rootbacktrack.top); -if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) - { - sljit_free_compiler(compiler); - SLJIT_FREE(common->optimized_cbracket); - SLJIT_FREE(common->private_data_ptrs); - if (common->has_then) - SLJIT_FREE(common->then_offsets); - return; - } - -SLJIT_ASSERT(rootbacktrack.prev == NULL); -reset_match_label = LABEL(); - -if (mode == JIT_PARTIAL_SOFT_COMPILE) - { - /* Update hit_start only in the first time. */ - jump = CMP(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, 0); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, -1); - OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, TMP1, 0); - JUMPHERE(jump); - } - -/* Check we have remaining characters. */ -if ((re->options & PCRE_ANCHORED) == 0 && (re->options & PCRE_FIRSTLINE) != 0) - { - SLJIT_ASSERT(common->first_line_end != 0); - OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); - } - -OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_ptr); - -if ((re->options & PCRE_ANCHORED) == 0) - { - if ((re->options & PCRE_FIRSTLINE) == 0) - CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop_label); - else - CMPTO(SLJIT_C_LESS, STR_PTR, 0, TMP1, 0, mainloop_label); - } - -/* No more remaining characters. */ -if (reqbyte_notfound != NULL) - JUMPHERE(reqbyte_notfound); - -if (mode == JIT_PARTIAL_SOFT_COMPILE) - CMPTO(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1, common->partialmatchlabel); - -OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_NOMATCH); -JUMPTO(SLJIT_JUMP, common->quit_label); - -flush_stubs(common); - -JUMPHERE(empty_match); -OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); -OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, notempty)); -CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0, empty_match_backtrack_label); -OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, notempty_atstart)); -CMPTO(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, 0, empty_match_found_label); -OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); -CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_PTR, 0, empty_match_found_label); -JUMPTO(SLJIT_JUMP, empty_match_backtrack_label); - -common->currententry = common->entries; -common->local_exit = TRUE; -quit_label = common->quit_label; -while (common->currententry != NULL) - { - /* Might add new entries. */ - compile_recurse(common); - if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) - { - sljit_free_compiler(compiler); - SLJIT_FREE(common->optimized_cbracket); - SLJIT_FREE(common->private_data_ptrs); - if (common->has_then) - SLJIT_FREE(common->then_offsets); - return; - } - flush_stubs(common); - common->currententry = common->currententry->next; - } -common->local_exit = FALSE; -common->quit_label = quit_label; - -/* Allocating stack, returns with PCRE_ERROR_JIT_STACKLIMIT if fails. */ -/* This is a (really) rare case. */ -set_jumps(common->stackalloc, LABEL()); -/* RETURN_ADDR is not a saved register. */ -sljit_emit_fast_enter(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); -OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP2, 0); -OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); -OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, stack)); -OP1(SLJIT_MOV, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(struct sljit_stack, top), STACK_TOP, 0); -OP2(SLJIT_ADD, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(struct sljit_stack, limit), SLJIT_IMM, STACK_GROWTH_RATE); - -sljit_emit_ijump(compiler, SLJIT_CALL2, SLJIT_IMM, SLJIT_FUNC_OFFSET(sljit_stack_resize)); -jump = CMP(SLJIT_C_NOT_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 0); -OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); -OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, stack)); -OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(struct sljit_stack, top)); -OP1(SLJIT_MOV, STACK_LIMIT, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(struct sljit_stack, limit)); -OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); -sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); - -/* Allocation failed. */ -JUMPHERE(jump); -/* We break the return address cache here, but this is a really rare case. */ -OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_JIT_STACKLIMIT); -JUMPTO(SLJIT_JUMP, common->quit_label); - -/* Call limit reached. */ -set_jumps(common->calllimit, LABEL()); -OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_MATCHLIMIT); -JUMPTO(SLJIT_JUMP, common->quit_label); - -if (common->revertframes != NULL) - { - set_jumps(common->revertframes, LABEL()); - do_revertframes(common); - } -if (common->wordboundary != NULL) - { - set_jumps(common->wordboundary, LABEL()); - check_wordboundary(common); - } -if (common->anynewline != NULL) - { - set_jumps(common->anynewline, LABEL()); - check_anynewline(common); - } -if (common->hspace != NULL) - { - set_jumps(common->hspace, LABEL()); - check_hspace(common); - } -if (common->vspace != NULL) - { - set_jumps(common->vspace, LABEL()); - check_vspace(common); - } -if (common->casefulcmp != NULL) - { - set_jumps(common->casefulcmp, LABEL()); - do_casefulcmp(common); - } -if (common->caselesscmp != NULL) - { - set_jumps(common->caselesscmp, LABEL()); - do_caselesscmp(common); - } -if (common->reset_match != NULL) - { - set_jumps(common->reset_match, LABEL()); - do_reset_match(common, (re->top_bracket + 1) * 2); - CMPTO(SLJIT_C_GREATER, STR_PTR, 0, TMP1, 0, continue_match_label); - OP1(SLJIT_MOV, STR_PTR, 0, TMP1, 0); - JUMPTO(SLJIT_JUMP, reset_match_label); - } -#ifdef SUPPORT_UTF -#ifndef COMPILE_PCRE32 -if (common->utfreadchar != NULL) - { - set_jumps(common->utfreadchar, LABEL()); - do_utfreadchar(common); - } -#endif /* !COMPILE_PCRE32 */ -#ifdef COMPILE_PCRE8 -if (common->utfreadtype8 != NULL) - { - set_jumps(common->utfreadtype8, LABEL()); - do_utfreadtype8(common); - } -#endif /* COMPILE_PCRE8 */ -#endif /* SUPPORT_UTF */ -#ifdef SUPPORT_UCP -if (common->getucd != NULL) - { - set_jumps(common->getucd, LABEL()); - do_getucd(common); - } -#endif - -SLJIT_FREE(common->optimized_cbracket); -SLJIT_FREE(common->private_data_ptrs); -if (common->has_then) - SLJIT_FREE(common->then_offsets); - -executable_func = sljit_generate_code(compiler); -executable_size = sljit_get_generated_code_size(compiler); -sljit_free_compiler(compiler); -if (executable_func == NULL) - return; - -/* Reuse the function descriptor if possible. */ -if ((extra->flags & PCRE_EXTRA_EXECUTABLE_JIT) != 0 && extra->executable_jit != NULL) - functions = (executable_functions *)extra->executable_jit; -else - { - /* Note: If your memory-checker has flagged the allocation below as a - * memory leak, it is probably because you either forgot to call - * pcre_free_study() (or pcre16_free_study()) on the pcre_extra (or - * pcre16_extra) object, or you called said function after having - * cleared the PCRE_EXTRA_EXECUTABLE_JIT bit from the "flags" field - * of the object. (The function will only free the JIT data if the - * bit remains set, as the bit indicates that the pointer to the data - * is valid.) - */ - functions = SLJIT_MALLOC(sizeof(executable_functions)); - if (functions == NULL) - { - /* This case is highly unlikely since we just recently - freed a lot of memory. Although not impossible. */ - sljit_free_code(executable_func); - return; - } - memset(functions, 0, sizeof(executable_functions)); - functions->top_bracket = (re->top_bracket + 1) * 2; - functions->limit_match = (re->flags & PCRE_MLSET) != 0 ? re->limit_match : 0; - extra->executable_jit = functions; - extra->flags |= PCRE_EXTRA_EXECUTABLE_JIT; - } - -functions->executable_funcs[mode] = executable_func; -functions->executable_sizes[mode] = executable_size; -} - -static int jit_machine_stack_exec(jit_arguments *arguments, void* executable_func) -{ -union { - void* executable_func; - jit_function call_executable_func; -} convert_executable_func; -pcre_uint8 local_space[MACHINE_STACK_SIZE]; -struct sljit_stack local_stack; - -local_stack.top = (sljit_sw)&local_space; -local_stack.base = local_stack.top; -local_stack.limit = local_stack.base + MACHINE_STACK_SIZE; -local_stack.max_limit = local_stack.limit; -arguments->stack = &local_stack; -convert_executable_func.executable_func = executable_func; -return convert_executable_func.call_executable_func(arguments); -} - -int -PRIV(jit_exec)(const PUBL(extra) *extra_data, const pcre_uchar *subject, - int length, int start_offset, int options, int *offsets, int offset_count) -{ -executable_functions *functions = (executable_functions *)extra_data->executable_jit; -union { - void* executable_func; - jit_function call_executable_func; -} convert_executable_func; -jit_arguments arguments; -int max_offset_count; -int retval; -int mode = JIT_COMPILE; - -if ((options & PCRE_PARTIAL_HARD) != 0) - mode = JIT_PARTIAL_HARD_COMPILE; -else if ((options & PCRE_PARTIAL_SOFT) != 0) - mode = JIT_PARTIAL_SOFT_COMPILE; - -if (functions->executable_funcs[mode] == NULL) - return PCRE_ERROR_JIT_BADOPTION; - -/* Sanity checks should be handled by pcre_exec. */ -arguments.str = subject + start_offset; -arguments.begin = subject; -arguments.end = subject + length; -arguments.mark_ptr = NULL; -/* JIT decreases this value less frequently than the interpreter. */ -arguments.limit_match = ((extra_data->flags & PCRE_EXTRA_MATCH_LIMIT) == 0) ? MATCH_LIMIT : (pcre_uint32)(extra_data->match_limit); -if (functions->limit_match != 0 && functions->limit_match < arguments.limit_match) - arguments.limit_match = functions->limit_match; -arguments.notbol = (options & PCRE_NOTBOL) != 0; -arguments.noteol = (options & PCRE_NOTEOL) != 0; -arguments.notempty = (options & PCRE_NOTEMPTY) != 0; -arguments.notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0; -arguments.offsets = offsets; -arguments.callout_data = (extra_data->flags & PCRE_EXTRA_CALLOUT_DATA) != 0 ? extra_data->callout_data : NULL; -arguments.real_offset_count = offset_count; - -/* pcre_exec() rounds offset_count to a multiple of 3, and then uses only 2/3 of -the output vector for storing captured strings, with the remainder used as -workspace. We don't need the workspace here. For compatibility, we limit the -number of captured strings in the same way as pcre_exec(), so that the user -gets the same result with and without JIT. */ - -if (offset_count != 2) - offset_count = ((offset_count - (offset_count % 3)) * 2) / 3; -max_offset_count = functions->top_bracket; -if (offset_count > max_offset_count) - offset_count = max_offset_count; -arguments.offset_count = offset_count; - -if (functions->callback) - arguments.stack = (struct sljit_stack *)functions->callback(functions->userdata); -else - arguments.stack = (struct sljit_stack *)functions->userdata; - -if (arguments.stack == NULL) - retval = jit_machine_stack_exec(&arguments, functions->executable_funcs[mode]); -else - { - convert_executable_func.executable_func = functions->executable_funcs[mode]; - retval = convert_executable_func.call_executable_func(&arguments); - } - -if (retval * 2 > offset_count) - retval = 0; -if ((extra_data->flags & PCRE_EXTRA_MARK) != 0) - *(extra_data->mark) = arguments.mark_ptr; - -return retval; -} - -#if defined COMPILE_PCRE8 -PCRE_EXP_DEFN int PCRE_CALL_CONVENTION -pcre_jit_exec(const pcre *argument_re, const pcre_extra *extra_data, - PCRE_SPTR subject, int length, int start_offset, int options, - int *offsets, int offset_count, pcre_jit_stack *stack) -#elif defined COMPILE_PCRE16 -PCRE_EXP_DEFN int PCRE_CALL_CONVENTION -pcre16_jit_exec(const pcre16 *argument_re, const pcre16_extra *extra_data, - PCRE_SPTR16 subject, int length, int start_offset, int options, - int *offsets, int offset_count, pcre16_jit_stack *stack) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DEFN int PCRE_CALL_CONVENTION -pcre32_jit_exec(const pcre32 *argument_re, const pcre32_extra *extra_data, - PCRE_SPTR32 subject, int length, int start_offset, int options, - int *offsets, int offset_count, pcre32_jit_stack *stack) -#endif -{ -pcre_uchar *subject_ptr = (pcre_uchar *)subject; -executable_functions *functions = (executable_functions *)extra_data->executable_jit; -union { - void* executable_func; - jit_function call_executable_func; -} convert_executable_func; -jit_arguments arguments; -int max_offset_count; -int retval; -int mode = JIT_COMPILE; - -SLJIT_UNUSED_ARG(argument_re); - -/* Plausibility checks */ -if ((options & ~PUBLIC_JIT_EXEC_OPTIONS) != 0) return PCRE_ERROR_JIT_BADOPTION; - -if ((options & PCRE_PARTIAL_HARD) != 0) - mode = JIT_PARTIAL_HARD_COMPILE; -else if ((options & PCRE_PARTIAL_SOFT) != 0) - mode = JIT_PARTIAL_SOFT_COMPILE; - -if (functions->executable_funcs[mode] == NULL) - return PCRE_ERROR_JIT_BADOPTION; - -/* Sanity checks should be handled by pcre_exec. */ -arguments.stack = (struct sljit_stack *)stack; -arguments.str = subject_ptr + start_offset; -arguments.begin = subject_ptr; -arguments.end = subject_ptr + length; -arguments.mark_ptr = NULL; -/* JIT decreases this value less frequently than the interpreter. */ -arguments.limit_match = ((extra_data->flags & PCRE_EXTRA_MATCH_LIMIT) == 0) ? MATCH_LIMIT : (pcre_uint32)(extra_data->match_limit); -if (functions->limit_match != 0 && functions->limit_match < arguments.limit_match) - arguments.limit_match = functions->limit_match; -arguments.notbol = (options & PCRE_NOTBOL) != 0; -arguments.noteol = (options & PCRE_NOTEOL) != 0; -arguments.notempty = (options & PCRE_NOTEMPTY) != 0; -arguments.notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0; -arguments.offsets = offsets; -arguments.callout_data = (extra_data->flags & PCRE_EXTRA_CALLOUT_DATA) != 0 ? extra_data->callout_data : NULL; -arguments.real_offset_count = offset_count; - -/* pcre_exec() rounds offset_count to a multiple of 3, and then uses only 2/3 of -the output vector for storing captured strings, with the remainder used as -workspace. We don't need the workspace here. For compatibility, we limit the -number of captured strings in the same way as pcre_exec(), so that the user -gets the same result with and without JIT. */ - -if (offset_count != 2) - offset_count = ((offset_count - (offset_count % 3)) * 2) / 3; -max_offset_count = functions->top_bracket; -if (offset_count > max_offset_count) - offset_count = max_offset_count; -arguments.offset_count = offset_count; - -convert_executable_func.executable_func = functions->executable_funcs[mode]; -retval = convert_executable_func.call_executable_func(&arguments); - -if (retval * 2 > offset_count) - retval = 0; -if ((extra_data->flags & PCRE_EXTRA_MARK) != 0) - *(extra_data->mark) = arguments.mark_ptr; - -return retval; -} - -void -PRIV(jit_free)(void *executable_funcs) -{ -int i; -executable_functions *functions = (executable_functions *)executable_funcs; -for (i = 0; i < JIT_NUMBER_OF_COMPILE_MODES; i++) - { - if (functions->executable_funcs[i] != NULL) - sljit_free_code(functions->executable_funcs[i]); - } -SLJIT_FREE(functions); -} - -int -PRIV(jit_get_size)(void *executable_funcs) -{ -int i; -sljit_uw size = 0; -sljit_uw *executable_sizes = ((executable_functions *)executable_funcs)->executable_sizes; -for (i = 0; i < JIT_NUMBER_OF_COMPILE_MODES; i++) - size += executable_sizes[i]; -return (int)size; -} - -const char* -PRIV(jit_get_target)(void) -{ -return sljit_get_platform_name(); -} - -#if defined COMPILE_PCRE8 -PCRE_EXP_DECL pcre_jit_stack * -pcre_jit_stack_alloc(int startsize, int maxsize) -#elif defined COMPILE_PCRE16 -PCRE_EXP_DECL pcre16_jit_stack * -pcre16_jit_stack_alloc(int startsize, int maxsize) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DECL pcre32_jit_stack * -pcre32_jit_stack_alloc(int startsize, int maxsize) -#endif -{ -if (startsize < 1 || maxsize < 1) - return NULL; -if (startsize > maxsize) - startsize = maxsize; -startsize = (startsize + STACK_GROWTH_RATE - 1) & ~(STACK_GROWTH_RATE - 1); -maxsize = (maxsize + STACK_GROWTH_RATE - 1) & ~(STACK_GROWTH_RATE - 1); -return (PUBL(jit_stack)*)sljit_allocate_stack(startsize, maxsize); -} - -#if defined COMPILE_PCRE8 -PCRE_EXP_DECL void -pcre_jit_stack_free(pcre_jit_stack *stack) -#elif defined COMPILE_PCRE16 -PCRE_EXP_DECL void -pcre16_jit_stack_free(pcre16_jit_stack *stack) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DECL void -pcre32_jit_stack_free(pcre32_jit_stack *stack) -#endif -{ -sljit_free_stack((struct sljit_stack *)stack); -} - -#if defined COMPILE_PCRE8 -PCRE_EXP_DECL void -pcre_assign_jit_stack(pcre_extra *extra, pcre_jit_callback callback, void *userdata) -#elif defined COMPILE_PCRE16 -PCRE_EXP_DECL void -pcre16_assign_jit_stack(pcre16_extra *extra, pcre16_jit_callback callback, void *userdata) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DECL void -pcre32_assign_jit_stack(pcre32_extra *extra, pcre32_jit_callback callback, void *userdata) -#endif -{ -executable_functions *functions; -if (extra != NULL && - (extra->flags & PCRE_EXTRA_EXECUTABLE_JIT) != 0 && - extra->executable_jit != NULL) - { - functions = (executable_functions *)extra->executable_jit; - functions->callback = callback; - functions->userdata = userdata; - } -} - -#if defined COMPILE_PCRE8 -PCRE_EXP_DECL void -pcre_jit_free_unused_memory(void) -#elif defined COMPILE_PCRE16 -PCRE_EXP_DECL void -pcre16_jit_free_unused_memory(void) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DECL void -pcre32_jit_free_unused_memory(void) -#endif -{ -sljit_free_unused_memory_exec(); -} - -#else /* SUPPORT_JIT */ - -/* These are dummy functions to avoid linking errors when JIT support is not -being compiled. */ - -#if defined COMPILE_PCRE8 -PCRE_EXP_DECL pcre_jit_stack * -pcre_jit_stack_alloc(int startsize, int maxsize) -#elif defined COMPILE_PCRE16 -PCRE_EXP_DECL pcre16_jit_stack * -pcre16_jit_stack_alloc(int startsize, int maxsize) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DECL pcre32_jit_stack * -pcre32_jit_stack_alloc(int startsize, int maxsize) -#endif -{ -(void)startsize; -(void)maxsize; -return NULL; -} - -#if defined COMPILE_PCRE8 -PCRE_EXP_DECL void -pcre_jit_stack_free(pcre_jit_stack *stack) -#elif defined COMPILE_PCRE16 -PCRE_EXP_DECL void -pcre16_jit_stack_free(pcre16_jit_stack *stack) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DECL void -pcre32_jit_stack_free(pcre32_jit_stack *stack) -#endif -{ -(void)stack; -} - -#if defined COMPILE_PCRE8 -PCRE_EXP_DECL void -pcre_assign_jit_stack(pcre_extra *extra, pcre_jit_callback callback, void *userdata) -#elif defined COMPILE_PCRE16 -PCRE_EXP_DECL void -pcre16_assign_jit_stack(pcre16_extra *extra, pcre16_jit_callback callback, void *userdata) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DECL void -pcre32_assign_jit_stack(pcre32_extra *extra, pcre32_jit_callback callback, void *userdata) -#endif -{ -(void)extra; -(void)callback; -(void)userdata; -} - -#if defined COMPILE_PCRE8 -PCRE_EXP_DECL void -pcre_jit_free_unused_memory(void) -#elif defined COMPILE_PCRE16 -PCRE_EXP_DECL void -pcre16_jit_free_unused_memory(void) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DECL void -pcre32_jit_free_unused_memory(void) -#endif -{ + Copyright (c) 1997-2012 University of Cambridge + + The machine code generator part (this module) was written by Zoltan Herczeg + Copyright (c) 2010-2012 + +----------------------------------------------------------------------------- +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. + + * Neither the name of the University of Cambridge 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 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. +----------------------------------------------------------------------------- +*/ + +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + +#include "pcre_internal.h" + +#ifdef SUPPORT_JIT + +/* All-in-one: Since we use the JIT compiler only from here, +we just include it. This way we don't need to touch the build +system files. */ + +#define SLJIT_MALLOC(size) (PUBL(malloc))(size) +#define SLJIT_FREE(ptr) (PUBL(free))(ptr) +#define SLJIT_CONFIG_AUTO 1 +#define SLJIT_CONFIG_STATIC 1 +#define SLJIT_VERBOSE 0 +#define SLJIT_DEBUG 0 + +#include "sljit/sljitLir.c" + +#if defined SLJIT_CONFIG_UNSUPPORTED && SLJIT_CONFIG_UNSUPPORTED +#error Unsupported architecture +#endif + +/* Allocate memory on the stack. Fast, but limited size. */ +#define LOCAL_SPACE_SIZE 32768 + +#define STACK_GROWTH_RATE 8192 + +/* Enable to check that the allocation could destroy temporaries. */ +#if defined SLJIT_DEBUG && SLJIT_DEBUG +#define DESTROY_REGISTERS 1 +#endif + +/* +Short summary about the backtracking mechanism empolyed by the jit code generator: + +The code generator follows the recursive nature of the PERL compatible regular +expressions. The basic blocks of regular expressions are condition checkers +whose execute different commands depending on the result of the condition check. +The relationship between the operators can be horizontal (concatenation) and +vertical (sub-expression) (See struct backtrack_common for more details). + + 'ab' - 'a' and 'b' regexps are concatenated + 'a+' - 'a' is the sub-expression of the '+' operator + +The condition checkers are boolean (true/false) checkers. Machine code is generated +for the checker itself and for the actions depending on the result of the checker. +The 'true' case is called as the try path (expected path), and the other is called as +the 'backtrack' path. Branch instructions are expesive for all CPUs, so we avoid taken +branches on the try path. + + Greedy star operator (*) : + Try path: match happens. + Backtrack path: match failed. + Non-greedy star operator (*?) : + Try path: no need to perform a match. + Backtrack path: match is required. + +The following example shows how the code generated for a capturing bracket +with two alternatives. Let A, B, C, D are arbirary regular expressions, and +we have the following regular expression: + + A(B|C)D + +The generated code will be the following: + + A try path + '(' try path (pushing arguments to the stack) + B try path + ')' try path (pushing arguments to the stack) + D try path + return with successful match + + D backtrack path + ')' backtrack path (If we arrived from "C" jump to the backtrack of "C") + B backtrack path + C expected path + jump to D try path + C backtrack path + A backtrack path + + Notice, that the order of backtrack code paths are the opposite of the fast + code paths. In this way the topmost value on the stack is always belong + to the current backtrack code path. The backtrack path must check + whether there is a next alternative. If so, it needs to jump back to + the try path eventually. Otherwise it needs to clear out its own stack + frame and continue the execution on the backtrack code paths. +*/ + +/* +Saved stack frames: + +Atomic blocks and asserts require reloading the values of local variables +when the backtrack mechanism performed. Because of OP_RECURSE, the locals +are not necessarly known in compile time, thus we need a dynamic restore +mechanism. + +The stack frames are stored in a chain list, and have the following format: +([ capturing bracket offset ][ start value ][ end value ])+ ... [ 0 ] [ previous head ] + +Thus we can restore the locals to a particular point in the stack. +*/ + +typedef struct jit_arguments { + /* Pointers first. */ + struct sljit_stack *stack; + const pcre_uchar *str; + const pcre_uchar *begin; + const pcre_uchar *end; + int *offsets; + pcre_uchar *uchar_ptr; + pcre_uchar *mark_ptr; + /* Everything else after. */ + int offsetcount; + int calllimit; + pcre_uint8 notbol; + pcre_uint8 noteol; + pcre_uint8 notempty; + pcre_uint8 notempty_atstart; +} jit_arguments; + +typedef struct executable_functions { + void *executable_funcs[JIT_NUMBER_OF_COMPILE_MODES]; + PUBL(jit_callback) callback; + void *userdata; + sljit_uw executable_sizes[JIT_NUMBER_OF_COMPILE_MODES]; +} executable_functions; + +typedef struct jump_list { + struct sljit_jump *jump; + struct jump_list *next; +} jump_list; + +enum stub_types { stack_alloc }; + +typedef struct stub_list { + enum stub_types type; + int data; + struct sljit_jump *start; + struct sljit_label *leave; + struct stub_list *next; +} stub_list; + +typedef int (SLJIT_CALL *jit_function)(jit_arguments *args); + +/* The following structure is the key data type for the recursive +code generator. It is allocated by compile_trypath, and contains +the aguments for compile_backtrackpath. Must be the first member +of its descendants. */ +typedef struct backtrack_common { + /* Concatenation stack. */ + struct backtrack_common *prev; + jump_list *nextbacktracks; + /* Internal stack (for component operators). */ + struct backtrack_common *top; + jump_list *topbacktracks; + /* Opcode pointer. */ + pcre_uchar *cc; +} backtrack_common; + +typedef struct assert_backtrack { + backtrack_common common; + jump_list *condfailed; + /* Less than 0 (-1) if a frame is not needed. */ + int framesize; + /* Points to our private memory word on the stack. */ + int localptr; + /* For iterators. */ + struct sljit_label *trypath; +} assert_backtrack; + +typedef struct bracket_backtrack { + backtrack_common common; + /* Where to coninue if an alternative is successfully matched. */ + struct sljit_label *alttrypath; + /* For rmin and rmax iterators. */ + struct sljit_label *recursivetrypath; + /* For greedy ? operator. */ + struct sljit_label *zerotrypath; + /* Contains the branches of a failed condition. */ + union { + /* Both for OP_COND, OP_SCOND. */ + jump_list *condfailed; + assert_backtrack *assert; + /* For OP_ONCE. -1 if not needed. */ + int framesize; + } u; + /* Points to our private memory word on the stack. */ + int localptr; +} bracket_backtrack; + +typedef struct bracketpos_backtrack { + backtrack_common common; + /* Points to our private memory word on the stack. */ + int localptr; + /* Reverting stack is needed. */ + int framesize; + /* Allocated stack size. */ + int stacksize; +} bracketpos_backtrack; + +typedef struct braminzero_backtrack { + backtrack_common common; + struct sljit_label *trypath; +} braminzero_backtrack; + +typedef struct iterator_backtrack { + backtrack_common common; + /* Next iteration. */ + struct sljit_label *trypath; +} iterator_backtrack; + +typedef struct recurse_entry { + struct recurse_entry *next; + /* Contains the function entry. */ + struct sljit_label *entry; + /* Collects the calls until the function is not created. */ + jump_list *calls; + /* Points to the starting opcode. */ + int start; +} recurse_entry; + +typedef struct recurse_backtrack { + backtrack_common common; +} recurse_backtrack; + +typedef struct compiler_common { + struct sljit_compiler *compiler; + pcre_uchar *start; + + /* Opcode local area direct map. */ + int *localptrs; + int cbraptr; + /* OVector starting point. Must be divisible by 2. */ + int ovector_start; + /* Last known position of the requested byte. */ + int req_char_ptr; + /* Head of the last recursion. */ + int recursive_head; + /* First inspected character for partial matching. */ + int start_used_ptr; + /* Starting pointer for partial soft matches. */ + int hit_start; + /* End pointer of the first line. */ + int first_line_end; + /* Points to the marked string. */ + int mark_ptr; + + /* Other */ + const pcre_uint8 *fcc; + sljit_w lcc; + int mode; + int nltype; + int newline; + int bsr_nltype; + int endonly; + BOOL has_set_som; + sljit_w ctypes; + sljit_uw name_table; + sljit_w name_count; + sljit_w name_entry_size; + + /* Labels and jump lists. */ + struct sljit_label *partialmatchlabel; + struct sljit_label *leavelabel; + struct sljit_label *acceptlabel; + stub_list *stubs; + recurse_entry *entries; + recurse_entry *currententry; + jump_list *partialmatch; + jump_list *leave; + jump_list *accept; + jump_list *calllimit; + jump_list *stackalloc; + jump_list *revertframes; + jump_list *wordboundary; + jump_list *anynewline; + jump_list *hspace; + jump_list *vspace; + jump_list *casefulcmp; + jump_list *caselesscmp; + BOOL jscript_compat; +#ifdef SUPPORT_UTF + BOOL utf; +#ifdef SUPPORT_UCP + BOOL use_ucp; +#endif + jump_list *utfreadchar; +#ifdef COMPILE_PCRE8 + jump_list *utfreadtype8; +#endif +#endif /* SUPPORT_UTF */ +#ifdef SUPPORT_UCP + jump_list *getucd; +#endif +} compiler_common; + +/* For byte_sequence_compare. */ + +typedef struct compare_context { + int length; + int sourcereg; +#if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED + int ucharptr; + union { + sljit_i asint; + sljit_uh asushort; +#ifdef COMPILE_PCRE8 + sljit_ub asbyte; + sljit_ub asuchars[4]; +#else +#ifdef COMPILE_PCRE16 + sljit_uh asuchars[2]; +#endif +#endif + } c; + union { + sljit_i asint; + sljit_uh asushort; +#ifdef COMPILE_PCRE8 + sljit_ub asbyte; + sljit_ub asuchars[4]; +#else +#ifdef COMPILE_PCRE16 + sljit_uh asuchars[2]; +#endif +#endif + } oc; +#endif +} compare_context; + +enum { + frame_end = 0, + frame_setstrbegin = -1, + frame_setmark = -2 +}; + +/* Undefine sljit macros. */ +#undef CMP + +/* Used for accessing the elements of the stack. */ +#define STACK(i) ((-(i) - 1) * (int)sizeof(sljit_w)) + +#define TMP1 SLJIT_TEMPORARY_REG1 +#define TMP2 SLJIT_TEMPORARY_REG3 +#define TMP3 SLJIT_TEMPORARY_EREG2 +#define STR_PTR SLJIT_SAVED_REG1 +#define STR_END SLJIT_SAVED_REG2 +#define STACK_TOP SLJIT_TEMPORARY_REG2 +#define STACK_LIMIT SLJIT_SAVED_REG3 +#define ARGUMENTS SLJIT_SAVED_EREG1 +#define CALL_COUNT SLJIT_SAVED_EREG2 +#define RETURN_ADDR SLJIT_TEMPORARY_EREG1 + +/* Locals layout. */ +/* These two locals can be used by the current opcode. */ +#define LOCALS0 (0 * sizeof(sljit_w)) +#define LOCALS1 (1 * sizeof(sljit_w)) +/* Two local variables for possessive quantifiers (char1 cannot use them). */ +#define POSSESSIVE0 (2 * sizeof(sljit_w)) +#define POSSESSIVE1 (3 * sizeof(sljit_w)) +/* Max limit of recursions. */ +#define CALL_LIMIT (4 * sizeof(sljit_w)) +/* The output vector is stored on the stack, and contains pointers +to characters. The vector data is divided into two groups: the first +group contains the start / end character pointers, and the second is +the start pointers when the end of the capturing group has not yet reached. */ +#define OVECTOR_START (common->ovector_start) +#define OVECTOR(i) (OVECTOR_START + (i) * sizeof(sljit_w)) +#define OVECTOR_PRIV(i) (common->cbraptr + (i) * sizeof(sljit_w)) +#define PRIV_DATA(cc) (common->localptrs[(cc) - common->start]) + +#ifdef COMPILE_PCRE8 +#define MOV_UCHAR SLJIT_MOV_UB +#define MOVU_UCHAR SLJIT_MOVU_UB +#else +#ifdef COMPILE_PCRE16 +#define MOV_UCHAR SLJIT_MOV_UH +#define MOVU_UCHAR SLJIT_MOVU_UH +#else +#error Unsupported compiling mode +#endif +#endif + +/* Shortcuts. */ +#define DEFINE_COMPILER \ + struct sljit_compiler *compiler = common->compiler +#define OP1(op, dst, dstw, src, srcw) \ + sljit_emit_op1(compiler, (op), (dst), (dstw), (src), (srcw)) +#define OP2(op, dst, dstw, src1, src1w, src2, src2w) \ + sljit_emit_op2(compiler, (op), (dst), (dstw), (src1), (src1w), (src2), (src2w)) +#define LABEL() \ + sljit_emit_label(compiler) +#define JUMP(type) \ + sljit_emit_jump(compiler, (type)) +#define JUMPTO(type, label) \ + sljit_set_label(sljit_emit_jump(compiler, (type)), (label)) +#define JUMPHERE(jump) \ + sljit_set_label((jump), sljit_emit_label(compiler)) +#define CMP(type, src1, src1w, src2, src2w) \ + sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w)) +#define CMPTO(type, src1, src1w, src2, src2w, label) \ + sljit_set_label(sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w)), (label)) +#define COND_VALUE(op, dst, dstw, type) \ + sljit_emit_cond_value(compiler, (op), (dst), (dstw), (type)) +#define GET_LOCAL_BASE(dst, dstw, offset) \ + sljit_get_local_base(compiler, (dst), (dstw), (offset)) + +static pcre_uchar* bracketend(pcre_uchar* cc) +{ +SLJIT_ASSERT((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND)); +do cc += GET(cc, 1); while (*cc == OP_ALT); +SLJIT_ASSERT(*cc >= OP_KET && *cc <= OP_KETRPOS); +cc += 1 + LINK_SIZE; +return cc; +} + +/* Functions whose might need modification for all new supported opcodes: + next_opcode + get_localspace + set_localptrs + get_framesize + init_frame + get_localsize + copy_locals + compile_trypath + compile_backtrackpath +*/ + +static pcre_uchar *next_opcode(compiler_common *common, pcre_uchar *cc) +{ +SLJIT_UNUSED_ARG(common); +switch(*cc) + { + case OP_SOD: + case OP_SOM: + case OP_SET_SOM: + case OP_NOT_WORD_BOUNDARY: + case OP_WORD_BOUNDARY: + case OP_NOT_DIGIT: + case OP_DIGIT: + case OP_NOT_WHITESPACE: + case OP_WHITESPACE: + case OP_NOT_WORDCHAR: + case OP_WORDCHAR: + case OP_ANY: + case OP_ALLANY: + case OP_ANYNL: + case OP_NOT_HSPACE: + case OP_HSPACE: + case OP_NOT_VSPACE: + case OP_VSPACE: + case OP_EXTUNI: + case OP_EODN: + case OP_EOD: + case OP_CIRC: + case OP_CIRCM: + case OP_DOLL: + case OP_DOLLM: + case OP_TYPESTAR: + case OP_TYPEMINSTAR: + case OP_TYPEPLUS: + case OP_TYPEMINPLUS: + case OP_TYPEQUERY: + case OP_TYPEMINQUERY: + case OP_TYPEPOSSTAR: + case OP_TYPEPOSPLUS: + case OP_TYPEPOSQUERY: + case OP_CRSTAR: + case OP_CRMINSTAR: + case OP_CRPLUS: + case OP_CRMINPLUS: + case OP_CRQUERY: + case OP_CRMINQUERY: + case OP_DEF: + case OP_BRAZERO: + case OP_BRAMINZERO: + case OP_BRAPOSZERO: + case OP_COMMIT: + case OP_FAIL: + case OP_ACCEPT: + case OP_ASSERT_ACCEPT: + case OP_SKIPZERO: + return cc + 1; + + case OP_ANYBYTE: +#ifdef SUPPORT_UTF + if (common->utf) return NULL; +#endif + return cc + 1; + + case OP_CHAR: + case OP_CHARI: + case OP_NOT: + case OP_NOTI: + case OP_STAR: + case OP_MINSTAR: + case OP_PLUS: + case OP_MINPLUS: + case OP_QUERY: + case OP_MINQUERY: + case OP_POSSTAR: + case OP_POSPLUS: + case OP_POSQUERY: + case OP_STARI: + case OP_MINSTARI: + case OP_PLUSI: + case OP_MINPLUSI: + case OP_QUERYI: + case OP_MINQUERYI: + case OP_POSSTARI: + case OP_POSPLUSI: + case OP_POSQUERYI: + case OP_NOTSTAR: + case OP_NOTMINSTAR: + case OP_NOTPLUS: + case OP_NOTMINPLUS: + case OP_NOTQUERY: + case OP_NOTMINQUERY: + case OP_NOTPOSSTAR: + case OP_NOTPOSPLUS: + case OP_NOTPOSQUERY: + case OP_NOTSTARI: + case OP_NOTMINSTARI: + case OP_NOTPLUSI: + case OP_NOTMINPLUSI: + case OP_NOTQUERYI: + case OP_NOTMINQUERYI: + case OP_NOTPOSSTARI: + case OP_NOTPOSPLUSI: + case OP_NOTPOSQUERYI: + cc += 2; +#ifdef SUPPORT_UTF + if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); +#endif + return cc; + + case OP_UPTO: + case OP_MINUPTO: + case OP_EXACT: + case OP_POSUPTO: + case OP_UPTOI: + case OP_MINUPTOI: + case OP_EXACTI: + case OP_POSUPTOI: + case OP_NOTUPTO: + case OP_NOTMINUPTO: + case OP_NOTEXACT: + case OP_NOTPOSUPTO: + case OP_NOTUPTOI: + case OP_NOTMINUPTOI: + case OP_NOTEXACTI: + case OP_NOTPOSUPTOI: + cc += 2 + IMM2_SIZE; +#ifdef SUPPORT_UTF + if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); +#endif + return cc; + + case OP_NOTPROP: + case OP_PROP: + return cc + 1 + 2; + + case OP_TYPEUPTO: + case OP_TYPEMINUPTO: + case OP_TYPEEXACT: + case OP_TYPEPOSUPTO: + case OP_REF: + case OP_REFI: + case OP_CREF: + case OP_NCREF: + case OP_RREF: + case OP_NRREF: + case OP_CLOSE: + cc += 1 + IMM2_SIZE; + return cc; + + case OP_CRRANGE: + case OP_CRMINRANGE: + return cc + 1 + 2 * IMM2_SIZE; + + case OP_CLASS: + case OP_NCLASS: + return cc + 1 + 32 / sizeof(pcre_uchar); + +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + case OP_XCLASS: + return cc + GET(cc, 1); +#endif + + case OP_RECURSE: + case OP_ASSERT: + case OP_ASSERT_NOT: + case OP_ASSERTBACK: + case OP_ASSERTBACK_NOT: + case OP_REVERSE: + case OP_ONCE: + case OP_ONCE_NC: + case OP_BRA: + case OP_BRAPOS: + case OP_COND: + case OP_SBRA: + case OP_SBRAPOS: + case OP_SCOND: + case OP_ALT: + case OP_KET: + case OP_KETRMAX: + case OP_KETRMIN: + case OP_KETRPOS: + return cc + 1 + LINK_SIZE; + + case OP_CBRA: + case OP_CBRAPOS: + case OP_SCBRA: + case OP_SCBRAPOS: + return cc + 1 + LINK_SIZE + IMM2_SIZE; + + case OP_MARK: + return cc + 1 + 2 + cc[1]; + + default: + return NULL; + } +} + +static int get_localspace(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) +{ +int localspace = 0; +pcre_uchar *alternative; +/* Calculate important variables (like stack size) and checks whether all opcodes are supported. */ +while (cc < ccend) + { + switch(*cc) + { + case OP_SET_SOM: + common->has_set_som = TRUE; + cc += 1; + break; + + case OP_ASSERT: + case OP_ASSERT_NOT: + case OP_ASSERTBACK: + case OP_ASSERTBACK_NOT: + case OP_ONCE: + case OP_ONCE_NC: + case OP_BRAPOS: + case OP_SBRA: + case OP_SBRAPOS: + case OP_SCOND: + localspace += sizeof(sljit_w); + cc += 1 + LINK_SIZE; + break; + + case OP_CBRAPOS: + case OP_SCBRAPOS: + localspace += sizeof(sljit_w); + cc += 1 + LINK_SIZE + IMM2_SIZE; + break; + + case OP_COND: + /* Might be a hidden SCOND. */ + alternative = cc + GET(cc, 1); + if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) + localspace += sizeof(sljit_w); + cc += 1 + LINK_SIZE; + break; + + case OP_RECURSE: + /* Set its value only once. */ + if (common->recursive_head == 0) + { + common->recursive_head = common->ovector_start; + common->ovector_start += sizeof(sljit_w); + } + cc += 1 + LINK_SIZE; + break; + + case OP_MARK: + if (common->mark_ptr == 0) + { + common->mark_ptr = common->ovector_start; + common->ovector_start += sizeof(sljit_w); + } + cc += 1 + 2 + cc[1]; + break; + + default: + cc = next_opcode(common, cc); + if (cc == NULL) + return -1; + break; + } + } +return localspace; +} + +static void set_localptrs(compiler_common *common, int localptr, pcre_uchar *ccend) +{ +pcre_uchar *cc = common->start; +pcre_uchar *alternative; +while (cc < ccend) + { + switch(*cc) + { + case OP_ASSERT: + case OP_ASSERT_NOT: + case OP_ASSERTBACK: + case OP_ASSERTBACK_NOT: + case OP_ONCE: + case OP_ONCE_NC: + case OP_BRAPOS: + case OP_SBRA: + case OP_SBRAPOS: + case OP_SCOND: + common->localptrs[cc - common->start] = localptr; + localptr += sizeof(sljit_w); + cc += 1 + LINK_SIZE; + break; + + case OP_CBRAPOS: + case OP_SCBRAPOS: + common->localptrs[cc - common->start] = localptr; + localptr += sizeof(sljit_w); + cc += 1 + LINK_SIZE + IMM2_SIZE; + break; + + case OP_COND: + /* Might be a hidden SCOND. */ + alternative = cc + GET(cc, 1); + if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) + { + common->localptrs[cc - common->start] = localptr; + localptr += sizeof(sljit_w); + } + cc += 1 + LINK_SIZE; + break; + + default: + cc = next_opcode(common, cc); + SLJIT_ASSERT(cc != NULL); + break; + } + } +} + +/* Returns with -1 if no need for frame. */ +static int get_framesize(compiler_common *common, pcre_uchar *cc, BOOL recursive) +{ +pcre_uchar *ccend = bracketend(cc); +int length = 0; +BOOL possessive = FALSE; +BOOL setsom_found = recursive; +BOOL setmark_found = recursive; + +if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS)) + { + length = 3; + possessive = TRUE; + } + +cc = next_opcode(common, cc); +SLJIT_ASSERT(cc != NULL); +while (cc < ccend) + switch(*cc) + { + case OP_SET_SOM: + SLJIT_ASSERT(common->has_set_som); + if (!setsom_found) + { + length += 2; + setsom_found = TRUE; + } + cc += 1; + break; + + case OP_MARK: + SLJIT_ASSERT(common->mark_ptr != 0); + if (!setmark_found) + { + length += 2; + setmark_found = TRUE; + } + cc += 1 + 2 + cc[1]; + break; + + case OP_RECURSE: + if (common->has_set_som && !setsom_found) + { + length += 2; + setsom_found = TRUE; + } + if (common->mark_ptr != 0 && !setmark_found) + { + length += 2; + setmark_found = TRUE; + } + cc += 1 + LINK_SIZE; + break; + + case OP_CBRA: + case OP_CBRAPOS: + case OP_SCBRA: + case OP_SCBRAPOS: + length += 3; + cc += 1 + LINK_SIZE + IMM2_SIZE; + break; + + default: + cc = next_opcode(common, cc); + SLJIT_ASSERT(cc != NULL); + break; + } + +/* Possessive quantifiers can use a special case. */ +if (SLJIT_UNLIKELY(possessive) && length == 3) + return -1; + +if (length > 0) + return length + 1; +return -1; +} + +static void init_frame(compiler_common *common, pcre_uchar *cc, int stackpos, int stacktop, BOOL recursive) +{ +DEFINE_COMPILER; +pcre_uchar *ccend = bracketend(cc); +BOOL setsom_found = recursive; +BOOL setmark_found = recursive; +int offset; + +/* >= 1 + shortest item size (2) */ +SLJIT_UNUSED_ARG(stacktop); +SLJIT_ASSERT(stackpos >= stacktop + 2); + +stackpos = STACK(stackpos); +if (recursive || (*cc != OP_CBRAPOS && *cc != OP_SCBRAPOS)) + cc = next_opcode(common, cc); +SLJIT_ASSERT(cc != NULL); +while (cc < ccend) + switch(*cc) + { + case OP_SET_SOM: + SLJIT_ASSERT(common->has_set_som); + if (!setsom_found) + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setstrbegin); + stackpos += (int)sizeof(sljit_w); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); + stackpos += (int)sizeof(sljit_w); + setsom_found = TRUE; + } + cc += 1; + break; + + case OP_MARK: + SLJIT_ASSERT(common->mark_ptr != 0); + if (!setmark_found) + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setmark); + stackpos += (int)sizeof(sljit_w); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); + stackpos += (int)sizeof(sljit_w); + setmark_found = TRUE; + } + cc += 1 + 2 + cc[1]; + break; + + case OP_RECURSE: + if (common->has_set_som && !setsom_found) + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setstrbegin); + stackpos += (int)sizeof(sljit_w); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); + stackpos += (int)sizeof(sljit_w); + setsom_found = TRUE; + } + if (common->mark_ptr != 0 && !setmark_found) + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_setmark); + stackpos += (int)sizeof(sljit_w); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); + stackpos += (int)sizeof(sljit_w); + setmark_found = TRUE; + } + cc += 1 + LINK_SIZE; + break; + + case OP_CBRA: + case OP_CBRAPOS: + case OP_SCBRA: + case OP_SCBRAPOS: + offset = (GET2(cc, 1 + LINK_SIZE)) << 1; + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, OVECTOR(offset)); + stackpos += (int)sizeof(sljit_w); + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0); + stackpos += (int)sizeof(sljit_w); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP2, 0); + stackpos += (int)sizeof(sljit_w); + + cc += 1 + LINK_SIZE + IMM2_SIZE; + break; + + default: + cc = next_opcode(common, cc); + SLJIT_ASSERT(cc != NULL); + break; + } + +OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, frame_end); +SLJIT_ASSERT(stackpos == STACK(stacktop)); +} + +static SLJIT_INLINE int get_localsize(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend) +{ +int localsize = 2; +pcre_uchar *alternative; +/* Calculate the sum of the local variables. */ +while (cc < ccend) + { + switch(*cc) + { + case OP_ASSERT: + case OP_ASSERT_NOT: + case OP_ASSERTBACK: + case OP_ASSERTBACK_NOT: + case OP_ONCE: + case OP_ONCE_NC: + case OP_BRAPOS: + case OP_SBRA: + case OP_SBRAPOS: + case OP_SCOND: + localsize++; + cc += 1 + LINK_SIZE; + break; + + case OP_CBRA: + case OP_SCBRA: + localsize++; + cc += 1 + LINK_SIZE + IMM2_SIZE; + break; + + case OP_CBRAPOS: + case OP_SCBRAPOS: + localsize += 2; + cc += 1 + LINK_SIZE + IMM2_SIZE; + break; + + case OP_COND: + /* Might be a hidden SCOND. */ + alternative = cc + GET(cc, 1); + if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) + localsize++; + cc += 1 + LINK_SIZE; + break; + + default: + cc = next_opcode(common, cc); + SLJIT_ASSERT(cc != NULL); + break; + } + } +SLJIT_ASSERT(cc == ccend); +return localsize; +} + +static void copy_locals(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, + BOOL save, int stackptr, int stacktop) +{ +DEFINE_COMPILER; +int srcw[2]; +int count; +BOOL tmp1next = TRUE; +BOOL tmp1empty = TRUE; +BOOL tmp2empty = TRUE; +pcre_uchar *alternative; +enum { + start, + loop, + end +} status; + +status = save ? start : loop; +stackptr = STACK(stackptr - 2); +stacktop = STACK(stacktop - 1); + +if (!save) + { + stackptr += sizeof(sljit_w); + if (stackptr < stacktop) + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr); + stackptr += sizeof(sljit_w); + tmp1empty = FALSE; + } + if (stackptr < stacktop) + { + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), stackptr); + stackptr += sizeof(sljit_w); + tmp2empty = FALSE; + } + /* The tmp1next must be TRUE in either way. */ + } + +while (status != end) + { + count = 0; + switch(status) + { + case start: + SLJIT_ASSERT(save && common->recursive_head != 0); + count = 1; + srcw[0] = common->recursive_head; + status = loop; + break; + + case loop: + if (cc >= ccend) + { + status = end; + break; + } + + switch(*cc) + { + case OP_ASSERT: + case OP_ASSERT_NOT: + case OP_ASSERTBACK: + case OP_ASSERTBACK_NOT: + case OP_ONCE: + case OP_ONCE_NC: + case OP_BRAPOS: + case OP_SBRA: + case OP_SBRAPOS: + case OP_SCOND: + count = 1; + srcw[0] = PRIV_DATA(cc); + SLJIT_ASSERT(srcw[0] != 0); + cc += 1 + LINK_SIZE; + break; + + case OP_CBRA: + case OP_SCBRA: + count = 1; + srcw[0] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); + cc += 1 + LINK_SIZE + IMM2_SIZE; + break; + + case OP_CBRAPOS: + case OP_SCBRAPOS: + count = 2; + srcw[1] = OVECTOR_PRIV(GET2(cc, 1 + LINK_SIZE)); + srcw[0] = PRIV_DATA(cc); + SLJIT_ASSERT(srcw[0] != 0); + cc += 1 + LINK_SIZE + IMM2_SIZE; + break; + + case OP_COND: + /* Might be a hidden SCOND. */ + alternative = cc + GET(cc, 1); + if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) + { + count = 1; + srcw[0] = PRIV_DATA(cc); + SLJIT_ASSERT(srcw[0] != 0); + } + cc += 1 + LINK_SIZE; + break; + + default: + cc = next_opcode(common, cc); + SLJIT_ASSERT(cc != NULL); + break; + } + break; + + case end: + SLJIT_ASSERT_STOP(); + break; + } + + while (count > 0) + { + count--; + if (save) + { + if (tmp1next) + { + if (!tmp1empty) + { + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); + stackptr += sizeof(sljit_w); + } + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]); + tmp1empty = FALSE; + tmp1next = FALSE; + } + else + { + if (!tmp2empty) + { + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); + stackptr += sizeof(sljit_w); + } + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count]); + tmp2empty = FALSE; + tmp1next = TRUE; + } + } + else + { + if (tmp1next) + { + SLJIT_ASSERT(!tmp1empty); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count], TMP1, 0); + tmp1empty = stackptr >= stacktop; + if (!tmp1empty) + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr); + stackptr += sizeof(sljit_w); + } + tmp1next = FALSE; + } + else + { + SLJIT_ASSERT(!tmp2empty); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), srcw[count], TMP2, 0); + tmp2empty = stackptr >= stacktop; + if (!tmp2empty) + { + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), stackptr); + stackptr += sizeof(sljit_w); + } + tmp1next = TRUE; + } + } + } + } + +if (save) + { + if (tmp1next) + { + if (!tmp1empty) + { + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); + stackptr += sizeof(sljit_w); + } + if (!tmp2empty) + { + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); + stackptr += sizeof(sljit_w); + } + } + else + { + if (!tmp2empty) + { + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP2, 0); + stackptr += sizeof(sljit_w); + } + if (!tmp1empty) + { + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackptr, TMP1, 0); + stackptr += sizeof(sljit_w); + } + } + } +SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty))); +} + +static SLJIT_INLINE BOOL ispowerof2(unsigned int value) +{ +return (value & (value - 1)) == 0; +} + +static SLJIT_INLINE void set_jumps(jump_list *list, struct sljit_label *label) +{ +while (list) + { + /* sljit_set_label is clever enough to do nothing + if either the jump or the label is NULL */ + sljit_set_label(list->jump, label); + list = list->next; + } +} + +static SLJIT_INLINE void add_jump(struct sljit_compiler *compiler, jump_list **list, struct sljit_jump* jump) +{ +jump_list *list_item = sljit_alloc_memory(compiler, sizeof(jump_list)); +if (list_item) + { + list_item->next = *list; + list_item->jump = jump; + *list = list_item; + } +} + +static void add_stub(compiler_common *common, enum stub_types type, int data, struct sljit_jump *start) +{ +DEFINE_COMPILER; +stub_list* list_item = sljit_alloc_memory(compiler, sizeof(stub_list)); + +if (list_item) + { + list_item->type = type; + list_item->data = data; + list_item->start = start; + list_item->leave = LABEL(); + list_item->next = common->stubs; + common->stubs = list_item; + } +} + +static void flush_stubs(compiler_common *common) +{ +DEFINE_COMPILER; +stub_list* list_item = common->stubs; + +while (list_item) + { + JUMPHERE(list_item->start); + switch(list_item->type) + { + case stack_alloc: + add_jump(compiler, &common->stackalloc, JUMP(SLJIT_FAST_CALL)); + break; + } + JUMPTO(SLJIT_JUMP, list_item->leave); + list_item = list_item->next; + } +common->stubs = NULL; +} + +static SLJIT_INLINE void decrease_call_count(compiler_common *common) +{ +DEFINE_COMPILER; + +OP2(SLJIT_SUB | SLJIT_SET_E, CALL_COUNT, 0, CALL_COUNT, 0, SLJIT_IMM, 1); +add_jump(compiler, &common->calllimit, JUMP(SLJIT_C_ZERO)); +} + +static SLJIT_INLINE void allocate_stack(compiler_common *common, int size) +{ +/* May destroy all locals and registers except TMP2. */ +DEFINE_COMPILER; + +OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, size * sizeof(sljit_w)); +#ifdef DESTROY_REGISTERS +OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 12345); +OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); +OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP1, 0); +OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, TMP1, 0); +OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP1, 0); +#endif +add_stub(common, stack_alloc, 0, CMP(SLJIT_C_GREATER, STACK_TOP, 0, STACK_LIMIT, 0)); +} + +static SLJIT_INLINE void free_stack(compiler_common *common, int size) +{ +DEFINE_COMPILER; +OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, size * sizeof(sljit_w)); +} + +static SLJIT_INLINE void reset_ovector(compiler_common *common, int length) +{ +DEFINE_COMPILER; +struct sljit_label *loop; +int i; +/* At this point we can freely use all temporary registers. */ +/* TMP1 returns with begin - 1. */ +OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG1, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), SLJIT_OFFSETOF(jit_arguments, begin), SLJIT_IMM, IN_UCHARS(1)); +if (length < 8) + { + for (i = 0; i < length; i++) + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(i), SLJIT_TEMPORARY_REG1, 0); + } +else + { + GET_LOCAL_BASE(SLJIT_TEMPORARY_REG2, 0, OVECTOR_START - sizeof(sljit_w)); + OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, length); + loop = LABEL(); + OP1(SLJIT_MOVU, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), sizeof(sljit_w), SLJIT_TEMPORARY_REG1, 0); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 1); + JUMPTO(SLJIT_C_NOT_ZERO, loop); + } +} + +static SLJIT_INLINE void copy_ovector(compiler_common *common, int topbracket) +{ +DEFINE_COMPILER; +struct sljit_label *loop; +struct sljit_jump *earlyexit; + +/* At this point we can freely use all registers. */ +OP1(SLJIT_MOV, SLJIT_SAVED_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)); +OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1), STR_PTR, 0); + +OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG1, 0, ARGUMENTS, 0); +if (common->mark_ptr != 0) + OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); +OP1(SLJIT_MOV_SI, SLJIT_TEMPORARY_REG2, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, offsetcount)); +if (common->mark_ptr != 0) + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, mark_ptr), SLJIT_TEMPORARY_REG3, 0); +OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, offsets), SLJIT_IMM, sizeof(int)); +OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG1, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), SLJIT_OFFSETOF(jit_arguments, begin)); +GET_LOCAL_BASE(SLJIT_SAVED_REG1, 0, OVECTOR_START); +/* Unlikely, but possible */ +earlyexit = CMP(SLJIT_C_EQUAL, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 0); +loop = LABEL(); +OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), 0, SLJIT_TEMPORARY_REG1, 0); +OP2(SLJIT_ADD, SLJIT_SAVED_REG1, 0, SLJIT_SAVED_REG1, 0, SLJIT_IMM, sizeof(sljit_w)); +/* Copy the integer value to the output buffer */ +#ifdef COMPILE_PCRE16 +OP2(SLJIT_ASHR, SLJIT_SAVED_REG2, 0, SLJIT_SAVED_REG2, 0, SLJIT_IMM, 1); +#endif +OP1(SLJIT_MOVU_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG3), sizeof(int), SLJIT_SAVED_REG2, 0); +OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_TEMPORARY_REG2, 0, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 1); +JUMPTO(SLJIT_C_NOT_ZERO, loop); +JUMPHERE(earlyexit); + +/* Calculate the return value, which is the maximum ovector value. */ +if (topbracket > 1) + { + GET_LOCAL_BASE(SLJIT_TEMPORARY_REG1, 0, OVECTOR_START + topbracket * 2 * sizeof(sljit_w)); + OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, topbracket + 1); + + /* OVECTOR(0) is never equal to SLJIT_SAVED_REG3. */ + loop = LABEL(); + OP1(SLJIT_MOVU, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG1), -(2 * (sljit_w)sizeof(sljit_w))); + OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG2, 0, SLJIT_TEMPORARY_REG2, 0, SLJIT_IMM, 1); + CMPTO(SLJIT_C_EQUAL, SLJIT_TEMPORARY_REG3, 0, SLJIT_SAVED_REG3, 0, loop); + OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_TEMPORARY_REG2, 0); + } +else + OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1); +} + +static SLJIT_INLINE void return_with_partial_match(compiler_common *common, struct sljit_label *leave) +{ +DEFINE_COMPILER; + +SLJIT_COMPILE_ASSERT(STR_END == SLJIT_SAVED_REG2, str_end_must_be_saved_reg2); +SLJIT_ASSERT(common->start_used_ptr != 0 && (common->mode == JIT_PARTIAL_SOFT_COMPILE ? common->hit_start != 0 : common->hit_start == 0)); + +OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, ARGUMENTS, 0); +OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_PARTIAL); +OP1(SLJIT_MOV_SI, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, offsetcount)); +CMPTO(SLJIT_C_LESS, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 2, leave); + +/* Store match begin and end. */ +OP1(SLJIT_MOV, SLJIT_SAVED_REG1, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, begin)); +OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, offsets)); +OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_used_ptr : common->hit_start); +OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, STR_END, 0, SLJIT_SAVED_REG1, 0); +#ifdef COMPILE_PCRE16 +OP2(SLJIT_ASHR, SLJIT_SAVED_REG2, 0, SLJIT_SAVED_REG2, 0, SLJIT_IMM, 1); +#endif +OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), sizeof(int), SLJIT_SAVED_REG2, 0); + +OP2(SLJIT_SUB, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_SAVED_REG1, 0); +#ifdef COMPILE_PCRE16 +OP2(SLJIT_ASHR, SLJIT_TEMPORARY_REG3, 0, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, 1); +#endif +OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), 0, SLJIT_TEMPORARY_REG3, 0); + +JUMPTO(SLJIT_JUMP, leave); +} + +static SLJIT_INLINE void check_start_used_ptr(compiler_common *common) +{ +/* May destroy TMP1. */ +DEFINE_COMPILER; +struct sljit_jump *jump; + +if (common->mode == JIT_PARTIAL_SOFT_COMPILE) + { + /* The value of -1 must be kept for start_used_ptr! */ + OP2(SLJIT_ADD, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, 1); + /* Jumps if start_used_ptr < STR_PTR, or start_used_ptr == -1. Although overwriting + is not necessary if start_used_ptr == STR_PTR, it does not hurt as well. */ + jump = CMP(SLJIT_C_LESS_EQUAL, TMP1, 0, STR_PTR, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); + JUMPHERE(jump); + } +else if (common->mode == JIT_PARTIAL_HARD_COMPILE) + { + jump = CMP(SLJIT_C_LESS_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); + JUMPHERE(jump); + } +} + +static SLJIT_INLINE BOOL char_has_othercase(compiler_common *common, pcre_uchar* cc) +{ +/* Detects if the character has an othercase. */ +unsigned int c; + +#ifdef SUPPORT_UTF +if (common->utf) + { + GETCHAR(c, cc); + if (c > 127) + { +#ifdef SUPPORT_UCP + return c != UCD_OTHERCASE(c); +#else + return FALSE; +#endif + } +#ifndef COMPILE_PCRE8 + return common->fcc[c] != c; +#endif + } +else +#endif + c = *cc; +return MAX_255(c) ? common->fcc[c] != c : FALSE; +} + +static SLJIT_INLINE unsigned int char_othercase(compiler_common *common, unsigned int c) +{ +/* Returns with the othercase. */ +#ifdef SUPPORT_UTF +if (common->utf && c > 127) + { +#ifdef SUPPORT_UCP + return UCD_OTHERCASE(c); +#else + return c; +#endif + } +#endif +return TABLE_GET(c, common->fcc, c); +} + +static unsigned int char_get_othercase_bit(compiler_common *common, pcre_uchar* cc) +{ +/* Detects if the character and its othercase has only 1 bit difference. */ +unsigned int c, oc, bit; +#if defined SUPPORT_UTF && defined COMPILE_PCRE8 +int n; +#endif + +#ifdef SUPPORT_UTF +if (common->utf) + { + GETCHAR(c, cc); + if (c <= 127) + oc = common->fcc[c]; + else + { +#ifdef SUPPORT_UCP + oc = UCD_OTHERCASE(c); +#else + oc = c; +#endif + } + } +else + { + c = *cc; + oc = TABLE_GET(c, common->fcc, c); + } +#else +c = *cc; +oc = TABLE_GET(c, common->fcc, c); +#endif + +SLJIT_ASSERT(c != oc); + +bit = c ^ oc; +/* Optimized for English alphabet. */ +if (c <= 127 && bit == 0x20) + return (0 << 8) | 0x20; + +/* Since c != oc, they must have at least 1 bit difference. */ +if (!ispowerof2(bit)) + return 0; + +#ifdef COMPILE_PCRE8 + +#ifdef SUPPORT_UTF +if (common->utf && c > 127) + { + n = GET_EXTRALEN(*cc); + while ((bit & 0x3f) == 0) + { + n--; + bit >>= 6; + } + return (n << 8) | bit; + } +#endif /* SUPPORT_UTF */ +return (0 << 8) | bit; + +#else /* COMPILE_PCRE8 */ + +#ifdef COMPILE_PCRE16 +#ifdef SUPPORT_UTF +if (common->utf && c > 65535) + { + if (bit >= (1 << 10)) + bit >>= 10; + else + return (bit < 256) ? ((2 << 8) | bit) : ((3 << 8) | (bit >> 8)); + } +#endif /* SUPPORT_UTF */ +return (bit < 256) ? ((0 << 8) | bit) : ((1 << 8) | (bit >> 8)); +#endif /* COMPILE_PCRE16 */ + +#endif /* COMPILE_PCRE8 */ +} + +static void check_partial(compiler_common *common, BOOL force) +{ +/* Checks whether a partial matching is occured. Does not modify registers. */ +DEFINE_COMPILER; +struct sljit_jump *jump = NULL; + +SLJIT_ASSERT(!force || common->mode != JIT_COMPILE); + +if (common->mode == JIT_COMPILE) + return; + +if (!force) + jump = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); +else if (common->mode == JIT_PARTIAL_SOFT_COMPILE) + jump = CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, -1); + +if (common->mode == JIT_PARTIAL_SOFT_COMPILE) + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1); +else + { + if (common->partialmatchlabel != NULL) + JUMPTO(SLJIT_JUMP, common->partialmatchlabel); + else + add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP)); + } + +if (jump != NULL) + JUMPHERE(jump); +} + +static struct sljit_jump *check_str_end(compiler_common *common) +{ +/* Does not affect registers. Usually used in a tight spot. */ +DEFINE_COMPILER; +struct sljit_jump *jump; +struct sljit_jump *nohit; +struct sljit_jump *return_value; + +if (common->mode == JIT_COMPILE) + return CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); + +jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); +if (common->mode == JIT_PARTIAL_SOFT_COMPILE) + { + nohit = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1); + JUMPHERE(nohit); + return_value = JUMP(SLJIT_JUMP); + } +else + { + return_value = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); + if (common->partialmatchlabel != NULL) + JUMPTO(SLJIT_JUMP, common->partialmatchlabel); + else + add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP)); + } +JUMPHERE(jump); +return return_value; +} + +static void detect_partial_match(compiler_common *common, jump_list **backtracks) +{ +DEFINE_COMPILER; +struct sljit_jump *jump; + +if (common->mode == JIT_COMPILE) + { + add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); + return; + } + +/* Partial matching mode. */ +jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); +add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0)); +if (common->mode == JIT_PARTIAL_SOFT_COMPILE) + { + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1); + add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); + } +else + { + if (common->partialmatchlabel != NULL) + JUMPTO(SLJIT_JUMP, common->partialmatchlabel); + else + add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP)); + } +JUMPHERE(jump); +} + +static void read_char(compiler_common *common) +{ +/* Reads the character into TMP1, updates STR_PTR. +Does not check STR_END. TMP2 Destroyed. */ +DEFINE_COMPILER; +#ifdef SUPPORT_UTF +struct sljit_jump *jump; +#endif + +OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); +#ifdef SUPPORT_UTF +if (common->utf) + { +#ifdef COMPILE_PCRE8 + jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); +#else +#ifdef COMPILE_PCRE16 + jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); +#endif +#endif /* COMPILE_PCRE8 */ + add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); + JUMPHERE(jump); + } +#endif +OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); +} + +static void peek_char(compiler_common *common) +{ +/* Reads the character into TMP1, keeps STR_PTR. +Does not check STR_END. TMP2 Destroyed. */ +DEFINE_COMPILER; +#ifdef SUPPORT_UTF +struct sljit_jump *jump; +#endif + +OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); +#ifdef SUPPORT_UTF +if (common->utf) + { +#ifdef COMPILE_PCRE8 + jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); +#else +#ifdef COMPILE_PCRE16 + jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); +#endif +#endif /* COMPILE_PCRE8 */ + add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL)); + OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); + JUMPHERE(jump); + } +#endif +} + +static void read_char8_type(compiler_common *common) +{ +/* Reads the character type into TMP1, updates STR_PTR. Does not check STR_END. */ +DEFINE_COMPILER; +#if defined SUPPORT_UTF || defined COMPILE_PCRE16 +struct sljit_jump *jump; +#endif + +#ifdef SUPPORT_UTF +if (common->utf) + { + OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); +#ifdef COMPILE_PCRE8 + /* This can be an extra read in some situations, but hopefully + it is needed in most cases. */ + OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); + jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0); + add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL)); + JUMPHERE(jump); +#else +#ifdef COMPILE_PCRE16 + OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); + jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); + OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); + JUMPHERE(jump); + /* Skip low surrogate if necessary. */ + OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xfc00); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0xd800); + COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); + OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); +#endif +#endif /* COMPILE_PCRE8 */ + return; + } +#endif +OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0); +OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); +#ifdef COMPILE_PCRE16 +/* The ctypes array contains only 256 values. */ +OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); +jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); +#endif +OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); +#ifdef COMPILE_PCRE16 +JUMPHERE(jump); +#endif +} + +static void skip_char_back(compiler_common *common) +{ +/* Goes one character back. Affects STR_PTR and TMP1. Does not check begin. */ +DEFINE_COMPILER; +#if defined SUPPORT_UTF && defined COMPILE_PCRE8 +struct sljit_label *label; + +if (common->utf) + { + label = LABEL(); + OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -IN_UCHARS(1)); + OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); + OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xc0); + CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0x80, label); + return; + } +#endif +#if defined SUPPORT_UTF && defined COMPILE_PCRE16 +if (common->utf) + { + OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -IN_UCHARS(1)); + OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); + /* Skip low surrogate if necessary. */ + OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xdc00); + COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); + OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); + OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0); + return; + } +#endif +OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); +} + +static void check_newlinechar(compiler_common *common, int nltype, jump_list **backtracks, BOOL jumpiftrue) +{ +/* Character comes in TMP1. Checks if it is a newline. TMP2 may be destroyed. */ +DEFINE_COMPILER; + +if (nltype == NLTYPE_ANY) + { + add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL)); + add_jump(compiler, backtracks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); + } +else if (nltype == NLTYPE_ANYCRLF) + { + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_CR); + COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_NL); + COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); + add_jump(compiler, backtracks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); + } +else + { + SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256); + add_jump(compiler, backtracks, CMP(jumpiftrue ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline)); + } +} + +#ifdef SUPPORT_UTF + +#ifdef COMPILE_PCRE8 +static void do_utfreadchar(compiler_common *common) +{ +/* Fast decoding a UTF-8 character. TMP1 contains the first byte +of the character (>= 0xc0). Return char value in TMP1, length - 1 in TMP2. */ +DEFINE_COMPILER; +struct sljit_jump *jump; + +sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); +/* Searching for the first zero. */ +OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x20); +jump = JUMP(SLJIT_C_NOT_ZERO); +/* Two byte sequence. */ +OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); +OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); +OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1f); +OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6); +OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); +OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); +OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(1)); +sljit_emit_fast_return(compiler, RETURN_ADDR, 0); +JUMPHERE(jump); + +OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x10); +jump = JUMP(SLJIT_C_NOT_ZERO); +/* Three byte sequence. */ +OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); +OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0f); +OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 12); +OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); +OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); +OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); +OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2)); +OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); +OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); +OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); +OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(2)); +sljit_emit_fast_return(compiler, RETURN_ADDR, 0); +JUMPHERE(jump); + +/* Four byte sequence. */ +OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); +OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x07); +OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 18); +OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); +OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 12); +OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); +OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2)); +OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); +OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); +OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); +OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(3)); +OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3)); +OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f); +OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); +OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(3)); +sljit_emit_fast_return(compiler, RETURN_ADDR, 0); +} + +static void do_utfreadtype8(compiler_common *common) +{ +/* Fast decoding a UTF-8 character type. TMP2 contains the first byte +of the character (>= 0xc0). Return value in TMP1. */ +DEFINE_COMPILER; +struct sljit_jump *jump; +struct sljit_jump *compare; + +sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); + +OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0x20); +jump = JUMP(SLJIT_C_NOT_ZERO); +/* Two byte sequence. */ +OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); +OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); +OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x1f); +OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6); +OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f); +OP2(SLJIT_OR, TMP2, 0, TMP2, 0, TMP1, 0); +compare = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255); +OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes); +sljit_emit_fast_return(compiler, RETURN_ADDR, 0); + +JUMPHERE(compare); +OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); +sljit_emit_fast_return(compiler, RETURN_ADDR, 0); +JUMPHERE(jump); + +/* We only have types for characters less than 256. */ +OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), (sljit_w)PRIV(utf8_table4) - 0xc0); +OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); +OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0); +sljit_emit_fast_return(compiler, RETURN_ADDR, 0); +} + +#else /* COMPILE_PCRE8 */ + +#ifdef COMPILE_PCRE16 +static void do_utfreadchar(compiler_common *common) +{ +/* Fast decoding a UTF-16 character. TMP1 contains the first 16 bit char +of the character (>= 0xd800). Return char value in TMP1, length - 1 in TMP2. */ +DEFINE_COMPILER; +struct sljit_jump *jump; + +sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); +jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xdc00); +/* Do nothing, only return. */ +sljit_emit_fast_return(compiler, RETURN_ADDR, 0); + +JUMPHERE(jump); +/* Combine two 16 bit characters. */ +OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); +OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); +OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3ff); +OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 10); +OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3ff); +OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0); +OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(1)); +OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000); +sljit_emit_fast_return(compiler, RETURN_ADDR, 0); +} +#endif /* COMPILE_PCRE16 */ + +#endif /* COMPILE_PCRE8 */ + +#endif /* SUPPORT_UTF */ + +#ifdef SUPPORT_UCP + +/* UCD_BLOCK_SIZE must be 128 (see the assert below). */ +#define UCD_BLOCK_MASK 127 +#define UCD_BLOCK_SHIFT 7 + +static void do_getucd(compiler_common *common) +{ +/* Search the UCD record for the character comes in TMP1. +Returns chartype in TMP1 and UCD offset in TMP2. */ +DEFINE_COMPILER; + +SLJIT_ASSERT(UCD_BLOCK_SIZE == 128 && sizeof(ucd_record) == 8); + +sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); +OP2(SLJIT_LSHR, TMP2, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_SHIFT); +OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_w)PRIV(ucd_stage1)); +OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_MASK); +OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCD_BLOCK_SHIFT); +OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0); +OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_stage2)); +OP1(SLJIT_MOV_UH, TMP2, 0, SLJIT_MEM2(TMP2, TMP1), 1); +OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, chartype)); +OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM2(TMP1, TMP2), 3); +sljit_emit_fast_return(compiler, RETURN_ADDR, 0); +} +#endif + +static SLJIT_INLINE struct sljit_label *mainloop_entry(compiler_common *common, BOOL hascrorlf, BOOL firstline) +{ +DEFINE_COMPILER; +struct sljit_label *mainloop; +struct sljit_label *newlinelabel = NULL; +struct sljit_jump *start; +struct sljit_jump *end = NULL; +struct sljit_jump *nl = NULL; +#ifdef SUPPORT_UTF +struct sljit_jump *singlechar; +#endif +jump_list *newline = NULL; +BOOL newlinecheck = FALSE; +BOOL readuchar = FALSE; + +if (!(hascrorlf || firstline) && (common->nltype == NLTYPE_ANY || + common->nltype == NLTYPE_ANYCRLF || common->newline > 255)) + newlinecheck = TRUE; + +if (firstline) + { + /* Search for the end of the first line. */ + SLJIT_ASSERT(common->first_line_end != 0); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, STR_PTR, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_END, 0); + + if (common->nltype == NLTYPE_FIXED && common->newline > 255) + { + mainloop = LABEL(); + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); + end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); + OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); + OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); + CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, mainloop); + CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, mainloop); + OP2(SLJIT_SUB, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); + } + else + { + end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); + mainloop = LABEL(); + /* Continual stores does not cause data dependency. */ + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0); + read_char(common); + check_newlinechar(common, common->nltype, &newline, TRUE); + CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0); + set_jumps(newline, LABEL()); + } + + JUMPHERE(end); + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); + } + +start = JUMP(SLJIT_JUMP); + +if (newlinecheck) + { + newlinelabel = LABEL(); + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); + end = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); + OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, common->newline & 0xff); + COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); +#ifdef COMPILE_PCRE16 + OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); +#endif + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); + nl = JUMP(SLJIT_JUMP); + } + +mainloop = LABEL(); + +/* Increasing the STR_PTR here requires one less jump in the most common case. */ +#ifdef SUPPORT_UTF +if (common->utf) readuchar = TRUE; +#endif +if (newlinecheck) readuchar = TRUE; + +if (readuchar) + OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); + +if (newlinecheck) + CMPTO(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, newlinelabel); + +OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); +#if defined SUPPORT_UTF && defined COMPILE_PCRE8 +if (common->utf) + { + singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); + OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0); + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); + JUMPHERE(singlechar); + } +#endif +#if defined SUPPORT_UTF && defined COMPILE_PCRE16 +if (common->utf) + { + singlechar = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); + OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); + COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); + OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); + JUMPHERE(singlechar); + } +#endif +JUMPHERE(start); + +if (newlinecheck) + { + JUMPHERE(end); + JUMPHERE(nl); + } + +return mainloop; +} + +static SLJIT_INLINE void fast_forward_first_char(compiler_common *common, pcre_uchar first_char, BOOL caseless, BOOL firstline) +{ +DEFINE_COMPILER; +struct sljit_label *start; +struct sljit_jump *leave; +struct sljit_jump *found; +pcre_uchar oc, bit; + +if (firstline) + { + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); + OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); + } + +start = LABEL(); +leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); +OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); + +oc = first_char; +if (caseless) + { + oc = TABLE_GET(first_char, common->fcc, first_char); +#if defined SUPPORT_UCP && !(defined COMPILE_PCRE8) + if (first_char > 127 && common->utf) + oc = UCD_OTHERCASE(first_char); +#endif + } +if (first_char == oc) + found = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, first_char); +else + { + bit = first_char ^ oc; + if (ispowerof2(bit)) + { + OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, bit); + found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, first_char | bit); + } + else + { + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, first_char); + COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, oc); + COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); + found = JUMP(SLJIT_C_NOT_ZERO); + } + } + +OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); +#if defined SUPPORT_UTF && defined COMPILE_PCRE8 +if (common->utf) + { + CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0, start); + OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0); + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); + } +#endif +#if defined SUPPORT_UTF && defined COMPILE_PCRE16 +if (common->utf) + { + CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start); + OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); + COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); + OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); + } +#endif +JUMPTO(SLJIT_JUMP, start); +JUMPHERE(found); +JUMPHERE(leave); + +if (firstline) + OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); +} + +static SLJIT_INLINE void fast_forward_newline(compiler_common *common, BOOL firstline) +{ +DEFINE_COMPILER; +struct sljit_label *loop; +struct sljit_jump *lastchar; +struct sljit_jump *firstchar; +struct sljit_jump *leave; +struct sljit_jump *foundcr = NULL; +struct sljit_jump *notfoundnl; +jump_list *newline = NULL; + +if (firstline) + { + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); + OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); + } + +if (common->nltype == NLTYPE_FIXED && common->newline > 255) + { + lastchar = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); + OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); + firstchar = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP2, 0); + + OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2)); + OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, STR_PTR, 0, TMP1, 0); + COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_GREATER_EQUAL); +#ifdef COMPILE_PCRE16 + OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); +#endif + OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); + + loop = LABEL(); + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); + leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); + OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); + OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); + CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, loop); + CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, loop); + + JUMPHERE(leave); + JUMPHERE(firstchar); + JUMPHERE(lastchar); + + if (firstline) + OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); + return; + } + +OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); +OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); +firstchar = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP2, 0); +skip_char_back(common); + +loop = LABEL(); +read_char(common); +lastchar = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); +if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF) + foundcr = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR); +check_newlinechar(common, common->nltype, &newline, FALSE); +set_jumps(newline, loop); + +if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF) + { + leave = JUMP(SLJIT_JUMP); + JUMPHERE(foundcr); + notfoundnl = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); + OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_NL); + COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); +#ifdef COMPILE_PCRE16 + OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); +#endif + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); + JUMPHERE(notfoundnl); + JUMPHERE(leave); + } +JUMPHERE(lastchar); +JUMPHERE(firstchar); + +if (firstline) + OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); +} + +static SLJIT_INLINE void fast_forward_start_bits(compiler_common *common, sljit_uw start_bits, BOOL firstline) +{ +DEFINE_COMPILER; +struct sljit_label *start; +struct sljit_jump *leave; +struct sljit_jump *found; +#ifndef COMPILE_PCRE8 +struct sljit_jump *jump; +#endif + +if (firstline) + { + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, STR_END, 0); + OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); + } + +start = LABEL(); +leave = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); +OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); +#ifdef SUPPORT_UTF +if (common->utf) + OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); +#endif +#ifndef COMPILE_PCRE8 +jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 255); +OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 255); +JUMPHERE(jump); +#endif +OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); +OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); +OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), start_bits); +OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); +OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); +found = JUMP(SLJIT_C_NOT_ZERO); + +#ifdef SUPPORT_UTF +if (common->utf) + OP1(SLJIT_MOV, TMP1, 0, TMP3, 0); +#endif +OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); +#if defined SUPPORT_UTF && defined COMPILE_PCRE8 +if (common->utf) + { + CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0, start); + OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0); + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); + } +#endif +#if defined SUPPORT_UTF && defined COMPILE_PCRE16 +if (common->utf) + { + CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800, start); + OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); + COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); + OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); + } +#endif +JUMPTO(SLJIT_JUMP, start); +JUMPHERE(found); +JUMPHERE(leave); + +if (firstline) + OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); +} + +static SLJIT_INLINE struct sljit_jump *search_requested_char(compiler_common *common, pcre_uchar req_char, BOOL caseless, BOOL has_firstchar) +{ +DEFINE_COMPILER; +struct sljit_label *loop; +struct sljit_jump *toolong; +struct sljit_jump *alreadyfound; +struct sljit_jump *found; +struct sljit_jump *foundoc = NULL; +struct sljit_jump *notfound; +pcre_uchar oc, bit; + +SLJIT_ASSERT(common->req_char_ptr != 0); +OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->req_char_ptr); +OP2(SLJIT_ADD, TMP1, 0, STR_PTR, 0, SLJIT_IMM, REQ_BYTE_MAX); +toolong = CMP(SLJIT_C_LESS, TMP1, 0, STR_END, 0); +alreadyfound = CMP(SLJIT_C_LESS, STR_PTR, 0, TMP2, 0); + +if (has_firstchar) + OP2(SLJIT_ADD, TMP1, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); +else + OP1(SLJIT_MOV, TMP1, 0, STR_PTR, 0); + +loop = LABEL(); +notfound = CMP(SLJIT_C_GREATER_EQUAL, TMP1, 0, STR_END, 0); + +OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(TMP1), 0); +oc = req_char; +if (caseless) + { + oc = TABLE_GET(req_char, common->fcc, req_char); +#if defined SUPPORT_UCP && !(defined COMPILE_PCRE8) + if (req_char > 127 && common->utf) + oc = UCD_OTHERCASE(req_char); +#endif + } +if (req_char == oc) + found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, req_char); +else + { + bit = req_char ^ oc; + if (ispowerof2(bit)) + { + OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, bit); + found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, req_char | bit); + } + else + { + found = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, req_char); + foundoc = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, oc); + } + } +OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(1)); +JUMPTO(SLJIT_JUMP, loop); + +JUMPHERE(found); +if (foundoc) + JUMPHERE(foundoc); +OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->req_char_ptr, TMP1, 0); +JUMPHERE(alreadyfound); +JUMPHERE(toolong); +return notfound; +} + +static void do_revertframes(compiler_common *common) +{ +DEFINE_COMPILER; +struct sljit_jump *jump; +struct sljit_label *mainloop; + +sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); +OP1(SLJIT_MOV, TMP1, 0, STACK_TOP, 0); +GET_LOCAL_BASE(TMP3, 0, 0); + +/* Drop frames until we reach STACK_TOP. */ +mainloop = LABEL(); +OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), 0); +jump = CMP(SLJIT_C_SIG_LESS_EQUAL, TMP2, 0, SLJIT_IMM, frame_end); +OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP3, 0); +OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), 0, SLJIT_MEM1(TMP1), sizeof(sljit_w)); +OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), sizeof(sljit_w), SLJIT_MEM1(TMP1), 2 * sizeof(sljit_w)); +OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 3 * sizeof(sljit_w)); +JUMPTO(SLJIT_JUMP, mainloop); + +JUMPHERE(jump); +jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_end); +/* End of dropping frames. */ +sljit_emit_fast_return(compiler, RETURN_ADDR, 0); + +JUMPHERE(jump); +jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_setstrbegin); +/* Set string begin. */ +OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), sizeof(sljit_w)); +OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_w)); +OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0), TMP2, 0); +JUMPTO(SLJIT_JUMP, mainloop); + +JUMPHERE(jump); +if (common->mark_ptr != 0) + { + jump = CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, frame_setmark); + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), sizeof(sljit_w)); + OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_w)); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, TMP2, 0); + JUMPTO(SLJIT_JUMP, mainloop); + + JUMPHERE(jump); + } + +/* Unknown command. */ +OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 2 * sizeof(sljit_w)); +JUMPTO(SLJIT_JUMP, mainloop); +} + +static void check_wordboundary(compiler_common *common) +{ +DEFINE_COMPILER; +struct sljit_jump *skipread; +#if !(defined COMPILE_PCRE8) || defined SUPPORT_UTF +struct sljit_jump *jump; +#endif + +SLJIT_COMPILE_ASSERT(ctype_word == 0x10, ctype_word_must_be_16); + +sljit_emit_fast_enter(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); +/* Get type of the previous char, and put it to LOCALS1. */ +OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); +OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); +OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, SLJIT_IMM, 0); +skipread = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP1, 0); +skip_char_back(common); +check_start_used_ptr(common); +read_char(common); + +/* Testing char type. */ +#ifdef SUPPORT_UCP +if (common->use_ucp) + { + OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 1); + jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE); + add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); + OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); + OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); + COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); + OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Nd - ucp_Ll); + OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_No - ucp_Nd); + COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL); + JUMPHERE(jump); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP2, 0); + } +else +#endif + { +#ifndef COMPILE_PCRE8 + jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); +#elif defined SUPPORT_UTF + /* Here LOCALS1 has already been zeroed. */ + jump = NULL; + if (common->utf) + jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); +#endif /* COMPILE_PCRE8 */ + OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), common->ctypes); + OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 4 /* ctype_word */); + OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP1, 0); +#ifndef COMPILE_PCRE8 + JUMPHERE(jump); +#elif defined SUPPORT_UTF + if (jump != NULL) + JUMPHERE(jump); +#endif /* COMPILE_PCRE8 */ + } +JUMPHERE(skipread); + +OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); +skipread = check_str_end(common); +peek_char(common); + +/* Testing char type. This is a code duplication. */ +#ifdef SUPPORT_UCP +if (common->use_ucp) + { + OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 1); + jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE); + add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); + OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Ll); + OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); + COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); + OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Nd - ucp_Ll); + OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ucp_No - ucp_Nd); + COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL); + JUMPHERE(jump); + } +else +#endif + { +#ifndef COMPILE_PCRE8 + /* TMP2 may be destroyed by peek_char. */ + OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); + jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); +#elif defined SUPPORT_UTF + OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); + jump = NULL; + if (common->utf) + jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); +#endif + OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), common->ctypes); + OP2(SLJIT_LSHR, TMP2, 0, TMP2, 0, SLJIT_IMM, 4 /* ctype_word */); + OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); +#ifndef COMPILE_PCRE8 + JUMPHERE(jump); +#elif defined SUPPORT_UTF + if (jump != NULL) + JUMPHERE(jump); +#endif /* COMPILE_PCRE8 */ + } +JUMPHERE(skipread); + +OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); +sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); +} + +static void check_anynewline(compiler_common *common) +{ +/* Check whether TMP1 contains a newline character. TMP2 destroyed. */ +DEFINE_COMPILER; + +sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); + +OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a); +OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x0d - 0x0a); +COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); +OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x0a); +#if defined SUPPORT_UTF || defined COMPILE_PCRE16 +#ifdef COMPILE_PCRE8 +if (common->utf) + { +#endif + COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); + OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2029 - 0x0a); +#ifdef COMPILE_PCRE8 + } +#endif +#endif /* SUPPORT_UTF || COMPILE_PCRE16 */ +COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); +sljit_emit_fast_return(compiler, RETURN_ADDR, 0); +} + +static void check_hspace(compiler_common *common) +{ +/* Check whether TMP1 contains a newline character. TMP2 destroyed. */ +DEFINE_COMPILER; + +sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); + +OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x09); +COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); +OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x20); +COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); +OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xa0); +#if defined SUPPORT_UTF || defined COMPILE_PCRE16 +#ifdef COMPILE_PCRE8 +if (common->utf) + { +#endif + COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x1680); + COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e); + COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); + OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x2000); + OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x200A - 0x2000); + COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x202f - 0x2000); + COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x205f - 0x2000); + COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x3000 - 0x2000); +#ifdef COMPILE_PCRE8 + } +#endif +#endif /* SUPPORT_UTF || COMPILE_PCRE16 */ +COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); + +sljit_emit_fast_return(compiler, RETURN_ADDR, 0); +} + +static void check_vspace(compiler_common *common) +{ +/* Check whether TMP1 contains a newline character. TMP2 destroyed. */ +DEFINE_COMPILER; + +sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); + +OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a); +OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x0d - 0x0a); +COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); +OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x0a); +#if defined SUPPORT_UTF || defined COMPILE_PCRE16 +#ifdef COMPILE_PCRE8 +if (common->utf) + { +#endif + COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); + OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2029 - 0x0a); +#ifdef COMPILE_PCRE8 + } +#endif +#endif /* SUPPORT_UTF || COMPILE_PCRE16 */ +COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); + +sljit_emit_fast_return(compiler, RETURN_ADDR, 0); +} + +#define CHAR1 STR_END +#define CHAR2 STACK_TOP + +static void do_casefulcmp(compiler_common *common) +{ +DEFINE_COMPILER; +struct sljit_jump *jump; +struct sljit_label *label; + +sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); +OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); +OP1(SLJIT_MOV, TMP3, 0, CHAR1, 0); +OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, CHAR2, 0); +OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(1)); +OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); + +label = LABEL(); +OP1(MOVU_UCHAR, CHAR1, 0, SLJIT_MEM1(TMP1), IN_UCHARS(1)); +OP1(MOVU_UCHAR, CHAR2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); +jump = CMP(SLJIT_C_NOT_EQUAL, CHAR1, 0, CHAR2, 0); +OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_IMM, IN_UCHARS(1)); +JUMPTO(SLJIT_C_NOT_ZERO, label); + +JUMPHERE(jump); +OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); +OP1(SLJIT_MOV, CHAR1, 0, TMP3, 0); +OP1(SLJIT_MOV, CHAR2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); +sljit_emit_fast_return(compiler, RETURN_ADDR, 0); +} + +#define LCC_TABLE STACK_LIMIT + +static void do_caselesscmp(compiler_common *common) +{ +DEFINE_COMPILER; +struct sljit_jump *jump; +struct sljit_label *label; + +sljit_emit_fast_enter(compiler, RETURN_ADDR, 0); +OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0); + +OP1(SLJIT_MOV, TMP3, 0, LCC_TABLE, 0); +OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, CHAR1, 0); +OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, CHAR2, 0); +OP1(SLJIT_MOV, LCC_TABLE, 0, SLJIT_IMM, common->lcc); +OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(1)); +OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); + +label = LABEL(); +OP1(MOVU_UCHAR, CHAR1, 0, SLJIT_MEM1(TMP1), IN_UCHARS(1)); +OP1(MOVU_UCHAR, CHAR2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); +#ifndef COMPILE_PCRE8 +jump = CMP(SLJIT_C_GREATER, CHAR1, 0, SLJIT_IMM, 255); +#endif +OP1(SLJIT_MOV_UB, CHAR1, 0, SLJIT_MEM2(LCC_TABLE, CHAR1), 0); +#ifndef COMPILE_PCRE8 +JUMPHERE(jump); +jump = CMP(SLJIT_C_GREATER, CHAR2, 0, SLJIT_IMM, 255); +#endif +OP1(SLJIT_MOV_UB, CHAR2, 0, SLJIT_MEM2(LCC_TABLE, CHAR2), 0); +#ifndef COMPILE_PCRE8 +JUMPHERE(jump); +#endif +jump = CMP(SLJIT_C_NOT_EQUAL, CHAR1, 0, CHAR2, 0); +OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_IMM, IN_UCHARS(1)); +JUMPTO(SLJIT_C_NOT_ZERO, label); + +JUMPHERE(jump); +OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); +OP1(SLJIT_MOV, LCC_TABLE, 0, TMP3, 0); +OP1(SLJIT_MOV, CHAR1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); +OP1(SLJIT_MOV, CHAR2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); +sljit_emit_fast_return(compiler, RETURN_ADDR, 0); +} + +#undef LCC_TABLE +#undef CHAR1 +#undef CHAR2 + +#if defined SUPPORT_UTF && defined SUPPORT_UCP + +static const pcre_uchar *SLJIT_CALL do_utf_caselesscmp(pcre_uchar *src1, jit_arguments *args, pcre_uchar *end1) +{ +/* This function would be ineffective to do in JIT level. */ +int c1, c2; +const pcre_uchar *src2 = args->uchar_ptr; +const pcre_uchar *end2 = args->end; + +while (src1 < end1) + { + if (src2 >= end2) + return (pcre_uchar*)1; + GETCHARINC(c1, src1); + GETCHARINC(c2, src2); + if (c1 != c2 && c1 != UCD_OTHERCASE(c2)) return NULL; + } +return src2; +} + +#endif /* SUPPORT_UTF && SUPPORT_UCP */ + +static pcre_uchar *byte_sequence_compare(compiler_common *common, BOOL caseless, pcre_uchar *cc, + compare_context* context, jump_list **backtracks) +{ +DEFINE_COMPILER; +unsigned int othercasebit = 0; +pcre_uchar *othercasechar = NULL; +#ifdef SUPPORT_UTF +int utflength; +#endif + +if (caseless && char_has_othercase(common, cc)) + { + othercasebit = char_get_othercase_bit(common, cc); + SLJIT_ASSERT(othercasebit); + /* Extracting bit difference info. */ +#ifdef COMPILE_PCRE8 + othercasechar = cc + (othercasebit >> 8); + othercasebit &= 0xff; +#else +#ifdef COMPILE_PCRE16 + othercasechar = cc + (othercasebit >> 9); + if ((othercasebit & 0x100) != 0) + othercasebit = (othercasebit & 0xff) << 8; + else + othercasebit &= 0xff; +#endif +#endif + } + +if (context->sourcereg == -1) + { +#ifdef COMPILE_PCRE8 +#if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED + if (context->length >= 4) + OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); + else if (context->length >= 2) + OP1(SLJIT_MOV_UH, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); + else +#endif + OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); +#else +#ifdef COMPILE_PCRE16 +#if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED + if (context->length >= 4) + OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); + else +#endif + OP1(SLJIT_MOV_UH, TMP1, 0, SLJIT_MEM1(STR_PTR), -context->length); +#endif +#endif /* COMPILE_PCRE8 */ + context->sourcereg = TMP2; + } + +#ifdef SUPPORT_UTF +utflength = 1; +if (common->utf && HAS_EXTRALEN(*cc)) + utflength += GET_EXTRALEN(*cc); + +do + { +#endif + + context->length -= IN_UCHARS(1); +#if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED + + /* Unaligned read is supported. */ + if (othercasebit != 0 && othercasechar == cc) + { + context->c.asuchars[context->ucharptr] = *cc | othercasebit; + context->oc.asuchars[context->ucharptr] = othercasebit; + } + else + { + context->c.asuchars[context->ucharptr] = *cc; + context->oc.asuchars[context->ucharptr] = 0; + } + context->ucharptr++; + +#ifdef COMPILE_PCRE8 + if (context->ucharptr >= 4 || context->length == 0 || (context->ucharptr == 2 && context->length == 1)) +#else + if (context->ucharptr >= 2 || context->length == 0) +#endif + { + if (context->length >= 4) + OP1(SLJIT_MOV_SI, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); +#ifdef COMPILE_PCRE8 + else if (context->length >= 2) + OP1(SLJIT_MOV_UH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); + else if (context->length >= 1) + OP1(SLJIT_MOV_UB, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); +#else + else if (context->length >= 2) + OP1(SLJIT_MOV_UH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); +#endif + context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; + + switch(context->ucharptr) + { + case 4 / sizeof(pcre_uchar): + if (context->oc.asint != 0) + OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asint); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asint | context->oc.asint)); + break; + + case 2 / sizeof(pcre_uchar): + if (context->oc.asushort != 0) + OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asushort); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asushort | context->oc.asushort)); + break; + +#ifdef COMPILE_PCRE8 + case 1: + if (context->oc.asbyte != 0) + OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, context->oc.asbyte); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, context->c.asbyte | context->oc.asbyte)); + break; +#endif + + default: + SLJIT_ASSERT_STOP(); + break; + } + context->ucharptr = 0; + } + +#else + + /* Unaligned read is unsupported. */ +#ifdef COMPILE_PCRE8 + if (context->length > 0) + OP1(SLJIT_MOV_UB, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); +#else + if (context->length > 0) + OP1(SLJIT_MOV_UH, context->sourcereg, 0, SLJIT_MEM1(STR_PTR), -context->length); +#endif + context->sourcereg = context->sourcereg == TMP1 ? TMP2 : TMP1; + + if (othercasebit != 0 && othercasechar == cc) + { + OP2(SLJIT_OR, context->sourcereg, 0, context->sourcereg, 0, SLJIT_IMM, othercasebit); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, *cc | othercasebit)); + } + else + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, context->sourcereg, 0, SLJIT_IMM, *cc)); + +#endif + + cc++; +#ifdef SUPPORT_UTF + utflength--; + } +while (utflength > 0); +#endif + +return cc; +} + +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + +#define SET_TYPE_OFFSET(value) \ + if ((value) != typeoffset) \ + { \ + if ((value) > typeoffset) \ + OP2(SLJIT_SUB, typereg, 0, typereg, 0, SLJIT_IMM, (value) - typeoffset); \ + else \ + OP2(SLJIT_ADD, typereg, 0, typereg, 0, SLJIT_IMM, typeoffset - (value)); \ + } \ + typeoffset = (value); + +#define SET_CHAR_OFFSET(value) \ + if ((value) != charoffset) \ + { \ + if ((value) > charoffset) \ + OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, (value) - charoffset); \ + else \ + OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, charoffset - (value)); \ + } \ + charoffset = (value); + +static void compile_xclass_trypath(compiler_common *common, pcre_uchar *cc, jump_list **backtracks) +{ +DEFINE_COMPILER; +jump_list *found = NULL; +jump_list **list = (*cc & XCL_NOT) == 0 ? &found : backtracks; +unsigned int c; +int compares; +struct sljit_jump *jump = NULL; +pcre_uchar *ccbegin; +#ifdef SUPPORT_UCP +BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE; +BOOL charsaved = FALSE; +int typereg = TMP1, scriptreg = TMP1; +unsigned int typeoffset; +#endif +int invertcmp, numberofcmps; +unsigned int charoffset; + +/* Although SUPPORT_UTF must be defined, we are not necessary in utf mode. */ +detect_partial_match(common, backtracks); +read_char(common); + +if ((*cc++ & XCL_MAP) != 0) + { + OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); +#ifndef COMPILE_PCRE8 + jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); +#elif defined SUPPORT_UTF + if (common->utf) + jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); +#endif + + OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); + OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); + OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)cc); + OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); + OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); + add_jump(compiler, list, JUMP(SLJIT_C_NOT_ZERO)); + +#ifndef COMPILE_PCRE8 + JUMPHERE(jump); +#elif defined SUPPORT_UTF + if (common->utf) + JUMPHERE(jump); +#endif + OP1(SLJIT_MOV, TMP1, 0, TMP3, 0); +#ifdef SUPPORT_UCP + charsaved = TRUE; +#endif + cc += 32 / sizeof(pcre_uchar); + } + +/* Scanning the necessary info. */ +ccbegin = cc; +compares = 0; +while (*cc != XCL_END) + { + compares++; + if (*cc == XCL_SINGLE) + { + cc += 2; +#ifdef SUPPORT_UTF + if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); +#endif +#ifdef SUPPORT_UCP + needschar = TRUE; +#endif + } + else if (*cc == XCL_RANGE) + { + cc += 2; +#ifdef SUPPORT_UTF + if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); +#endif + cc++; +#ifdef SUPPORT_UTF + if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]); +#endif +#ifdef SUPPORT_UCP + needschar = TRUE; +#endif + } +#ifdef SUPPORT_UCP + else + { + SLJIT_ASSERT(*cc == XCL_PROP || *cc == XCL_NOTPROP); + cc++; + switch(*cc) + { + case PT_ANY: + break; + + case PT_LAMP: + case PT_GC: + case PT_PC: + case PT_ALNUM: + needstype = TRUE; + break; + + case PT_SC: + needsscript = TRUE; + break; + + case PT_SPACE: + case PT_PXSPACE: + case PT_WORD: + needstype = TRUE; + needschar = TRUE; + break; + + default: + SLJIT_ASSERT_STOP(); + break; + } + cc += 2; + } +#endif + } + +#ifdef SUPPORT_UCP +/* Simple register allocation. TMP1 is preferred if possible. */ +if (needstype || needsscript) + { + if (needschar && !charsaved) + OP1(SLJIT_MOV, TMP3, 0, TMP1, 0); + add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); + if (needschar) + { + if (needstype) + { + OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP1, 0); + typereg = RETURN_ADDR; + } + + if (needsscript) + scriptreg = TMP3; + OP1(SLJIT_MOV, TMP1, 0, TMP3, 0); + } + else if (needstype && needsscript) + scriptreg = TMP3; + /* In all other cases only one of them was specified, and that can goes to TMP1. */ + + if (needsscript) + { + if (scriptreg == TMP1) + { + OP1(SLJIT_MOV, scriptreg, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, script)); + OP1(SLJIT_MOV_UB, scriptreg, 0, SLJIT_MEM2(scriptreg, TMP2), 3); + } + else + { + OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 3); + OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, (sljit_w)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, script)); + OP1(SLJIT_MOV_UB, scriptreg, 0, SLJIT_MEM1(TMP2), 0); + } + } + } +#endif + +/* Generating code. */ +cc = ccbegin; +charoffset = 0; +numberofcmps = 0; +#ifdef SUPPORT_UCP +typeoffset = 0; +#endif + +while (*cc != XCL_END) + { + compares--; + invertcmp = (compares == 0 && list != backtracks); + jump = NULL; + + if (*cc == XCL_SINGLE) + { + cc ++; +#ifdef SUPPORT_UTF + if (common->utf) + { + GETCHARINC(c, cc); + } + else +#endif + c = *cc++; + + if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE)) + { + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset); + COND_VALUE(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); + numberofcmps++; + } + else if (numberofcmps > 0) + { + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset); + COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); + jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); + numberofcmps = 0; + } + else + { + jump = CMP(SLJIT_C_EQUAL ^ invertcmp, TMP1, 0, SLJIT_IMM, c - charoffset); + numberofcmps = 0; + } + } + else if (*cc == XCL_RANGE) + { + cc ++; +#ifdef SUPPORT_UTF + if (common->utf) + { + GETCHARINC(c, cc); + } + else +#endif + c = *cc++; + SET_CHAR_OFFSET(c); +#ifdef SUPPORT_UTF + if (common->utf) + { + GETCHARINC(c, cc); + } + else +#endif + c = *cc++; + if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE)) + { + OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset); + COND_VALUE(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL); + numberofcmps++; + } + else if (numberofcmps > 0) + { + OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset); + COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_LESS_EQUAL); + jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); + numberofcmps = 0; + } + else + { + jump = CMP(SLJIT_C_LESS_EQUAL ^ invertcmp, TMP1, 0, SLJIT_IMM, c - charoffset); + numberofcmps = 0; + } + } +#ifdef SUPPORT_UCP + else + { + if (*cc == XCL_NOTPROP) + invertcmp ^= 0x1; + cc++; + switch(*cc) + { + case PT_ANY: + if (list != backtracks) + { + if ((cc[-1] == XCL_NOTPROP && compares > 0) || (cc[-1] == XCL_PROP && compares == 0)) + continue; + } + else if (cc[-1] == XCL_NOTPROP) + continue; + jump = JUMP(SLJIT_JUMP); + break; + + case PT_LAMP: + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lu - typeoffset); + COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Ll - typeoffset); + COND_VALUE(SLJIT_OR, TMP2, 0, SLJIT_C_EQUAL); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lt - typeoffset); + COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); + jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); + break; + + case PT_GC: + c = PRIV(ucp_typerange)[(int)cc[1] * 2]; + SET_TYPE_OFFSET(c); + jump = CMP(SLJIT_C_LESS_EQUAL ^ invertcmp, typereg, 0, SLJIT_IMM, PRIV(ucp_typerange)[(int)cc[1] * 2 + 1] - c); + break; + + case PT_PC: + jump = CMP(SLJIT_C_EQUAL ^ invertcmp, typereg, 0, SLJIT_IMM, (int)cc[1] - typeoffset); + break; + + case PT_SC: + jump = CMP(SLJIT_C_EQUAL ^ invertcmp, scriptreg, 0, SLJIT_IMM, (int)cc[1]); + break; + + case PT_SPACE: + case PT_PXSPACE: + if (*cc == PT_SPACE) + { + OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0); + jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 11 - charoffset); + } + SET_CHAR_OFFSET(9); + OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 13 - 9); + COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS_EQUAL); + if (*cc == PT_SPACE) + JUMPHERE(jump); + + SET_TYPE_OFFSET(ucp_Zl); + OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Zl); + COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_LESS_EQUAL); + jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); + break; + + case PT_WORD: + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE - charoffset); + COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); + /* ... fall through */ + + case PT_ALNUM: + SET_TYPE_OFFSET(ucp_Ll); + OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Lu - ucp_Ll); + COND_VALUE((*cc == PT_ALNUM) ? SLJIT_MOV : SLJIT_OR, TMP2, 0, SLJIT_C_LESS_EQUAL); + SET_TYPE_OFFSET(ucp_Nd); + OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_No - ucp_Nd); + COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_LESS_EQUAL); + jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp); + break; + } + cc += 2; + } +#endif + + if (jump != NULL) + add_jump(compiler, compares > 0 ? list : backtracks, jump); + } + +if (found != NULL) + set_jumps(found, LABEL()); +} + +#undef SET_TYPE_OFFSET +#undef SET_CHAR_OFFSET + +#endif + +static pcre_uchar *compile_char1_trypath(compiler_common *common, pcre_uchar type, pcre_uchar *cc, jump_list **backtracks) +{ +DEFINE_COMPILER; +int length; +unsigned int c, oc, bit; +compare_context context; +struct sljit_jump *jump[4]; +#ifdef SUPPORT_UTF +struct sljit_label *label; +#ifdef SUPPORT_UCP +pcre_uchar propdata[5]; +#endif +#endif + +switch(type) + { + case OP_SOD: + OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, TMP1, 0)); + return cc; + + case OP_SOM: + OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, TMP1, 0)); + return cc; + + case OP_NOT_WORD_BOUNDARY: + case OP_WORD_BOUNDARY: + add_jump(compiler, &common->wordboundary, JUMP(SLJIT_FAST_CALL)); + add_jump(compiler, backtracks, JUMP(type == OP_NOT_WORD_BOUNDARY ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); + return cc; + + case OP_NOT_DIGIT: + case OP_DIGIT: + detect_partial_match(common, backtracks); + read_char8_type(common); + OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_digit); + add_jump(compiler, backtracks, JUMP(type == OP_DIGIT ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); + return cc; + + case OP_NOT_WHITESPACE: + case OP_WHITESPACE: + detect_partial_match(common, backtracks); + read_char8_type(common); + OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_space); + add_jump(compiler, backtracks, JUMP(type == OP_WHITESPACE ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); + return cc; + + case OP_NOT_WORDCHAR: + case OP_WORDCHAR: + detect_partial_match(common, backtracks); + read_char8_type(common); + OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_word); + add_jump(compiler, backtracks, JUMP(type == OP_WORDCHAR ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO)); + return cc; + + case OP_ANY: + detect_partial_match(common, backtracks); + read_char(common); + if (common->nltype == NLTYPE_FIXED && common->newline > 255) + { + jump[0] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff); + if (common->mode != JIT_PARTIAL_HARD_COMPILE) + jump[1] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); + else + jump[1] = check_str_end(common); + + OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); + add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, common->newline & 0xff)); + if (jump[1] != NULL) + JUMPHERE(jump[1]); + JUMPHERE(jump[0]); + } + else + check_newlinechar(common, common->nltype, backtracks, TRUE); + return cc; + + case OP_ALLANY: + detect_partial_match(common, backtracks); +#ifdef SUPPORT_UTF + if (common->utf) + { + OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); +#ifdef COMPILE_PCRE8 + jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); + OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0); + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); +#else /* COMPILE_PCRE8 */ +#ifdef COMPILE_PCRE16 + jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800); + OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd800); + COND_VALUE(SLJIT_MOV, TMP1, 0, SLJIT_C_EQUAL); + OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); +#endif /* COMPILE_PCRE16 */ +#endif /* COMPILE_PCRE8 */ + JUMPHERE(jump[0]); + return cc; + } +#endif + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); + return cc; + + case OP_ANYBYTE: + detect_partial_match(common, backtracks); + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); + return cc; + +#ifdef SUPPORT_UTF +#ifdef SUPPORT_UCP + case OP_NOTPROP: + case OP_PROP: + propdata[0] = 0; + propdata[1] = type == OP_NOTPROP ? XCL_NOTPROP : XCL_PROP; + propdata[2] = cc[0]; + propdata[3] = cc[1]; + propdata[4] = XCL_END; + compile_xclass_trypath(common, propdata, backtracks); + return cc + 2; +#endif +#endif + + case OP_ANYNL: + detect_partial_match(common, backtracks); + read_char(common); + jump[0] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR); + /* We don't need to handle soft partial matching case. */ + if (common->mode != JIT_PARTIAL_HARD_COMPILE) + jump[1] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); + else + jump[1] = check_str_end(common); + OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); + jump[2] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL); + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); + jump[3] = JUMP(SLJIT_JUMP); + JUMPHERE(jump[0]); + check_newlinechar(common, common->bsr_nltype, backtracks, FALSE); + JUMPHERE(jump[1]); + JUMPHERE(jump[2]); + JUMPHERE(jump[3]); + return cc; + + case OP_NOT_HSPACE: + case OP_HSPACE: + detect_partial_match(common, backtracks); + read_char(common); + add_jump(compiler, &common->hspace, JUMP(SLJIT_FAST_CALL)); + add_jump(compiler, backtracks, JUMP(type == OP_NOT_HSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); + return cc; + + case OP_NOT_VSPACE: + case OP_VSPACE: + detect_partial_match(common, backtracks); + read_char(common); + add_jump(compiler, &common->vspace, JUMP(SLJIT_FAST_CALL)); + add_jump(compiler, backtracks, JUMP(type == OP_NOT_VSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO)); + return cc; + +#ifdef SUPPORT_UCP + case OP_EXTUNI: + detect_partial_match(common, backtracks); + read_char(common); + add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); + OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Mc); + add_jump(compiler, backtracks, CMP(SLJIT_C_LESS_EQUAL, TMP1, 0, SLJIT_IMM, ucp_Mn - ucp_Mc)); + + label = LABEL(); + jump[0] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); + OP1(SLJIT_MOV, TMP3, 0, STR_PTR, 0); + read_char(common); + add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL)); + OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ucp_Mc); + CMPTO(SLJIT_C_LESS_EQUAL, TMP1, 0, SLJIT_IMM, ucp_Mn - ucp_Mc, label); + + OP1(SLJIT_MOV, STR_PTR, 0, TMP3, 0); + JUMPHERE(jump[0]); + if (common->mode == JIT_PARTIAL_HARD_COMPILE) + { + jump[0] = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); + /* Since we successfully read a char above, partial matching must occure. */ + check_partial(common, TRUE); + JUMPHERE(jump[0]); + } + return cc; +#endif + + case OP_EODN: + /* Requires rather complex checks. */ + jump[0] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0); + if (common->nltype == NLTYPE_FIXED && common->newline > 255) + { + OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); + OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); + if (common->mode == JIT_COMPILE) + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_END, 0)); + else + { + jump[1] = CMP(SLJIT_C_EQUAL, TMP2, 0, STR_END, 0); + OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP2, 0, STR_END, 0); + COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_LESS); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff); + COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_NOT_EQUAL); + add_jump(compiler, backtracks, JUMP(SLJIT_C_NOT_EQUAL)); + check_partial(common, TRUE); + add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); + JUMPHERE(jump[1]); + } + OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff)); + } + else if (common->nltype == NLTYPE_FIXED) + { + OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); + OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_END, 0)); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline)); + } + else + { + OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); + jump[1] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR); + OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); + OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP2, 0, STR_END, 0); + jump[2] = JUMP(SLJIT_C_GREATER); + add_jump(compiler, backtracks, JUMP(SLJIT_C_LESS)); + /* Equal. */ + OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); + jump[3] = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL); + add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); + + JUMPHERE(jump[1]); + if (common->nltype == NLTYPE_ANYCRLF) + { + OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); + add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP2, 0, STR_END, 0)); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL)); + } + else + { + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, STR_PTR, 0); + read_char(common); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, STR_END, 0)); + add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL)); + add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO)); + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); + } + JUMPHERE(jump[2]); + JUMPHERE(jump[3]); + } + JUMPHERE(jump[0]); + check_partial(common, FALSE); + return cc; + + case OP_EOD: + add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0)); + check_partial(common, FALSE); + return cc; + + case OP_CIRC: + OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, begin)); + add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER, STR_PTR, 0, TMP1, 0)); + OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, notbol)); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); + return cc; + + case OP_CIRCM: + OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, begin)); + jump[1] = CMP(SLJIT_C_GREATER, STR_PTR, 0, TMP1, 0); + OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, notbol)); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); + jump[0] = JUMP(SLJIT_JUMP); + JUMPHERE(jump[1]); + + add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0)); + if (common->nltype == NLTYPE_FIXED && common->newline > 255) + { + OP2(SLJIT_SUB, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); + add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP2, 0, TMP1, 0)); + OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2)); + OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1)); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff)); + } + else + { + skip_char_back(common); + read_char(common); + check_newlinechar(common, common->nltype, backtracks, FALSE); + } + JUMPHERE(jump[0]); + return cc; + + case OP_DOLL: + OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); + OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, noteol)); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); + + if (!common->endonly) + compile_char1_trypath(common, OP_EODN, cc, backtracks); + else + { + add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0)); + check_partial(common, FALSE); + } + return cc; + + case OP_DOLLM: + jump[1] = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0); + OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0); + OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, noteol)); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); + check_partial(common, FALSE); + jump[0] = JUMP(SLJIT_JUMP); + JUMPHERE(jump[1]); + + if (common->nltype == NLTYPE_FIXED && common->newline > 255) + { + OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2)); + OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0)); + if (common->mode == JIT_COMPILE) + add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER, TMP2, 0, STR_END, 0)); + else + { + jump[1] = CMP(SLJIT_C_LESS_EQUAL, TMP2, 0, STR_END, 0); + /* STR_PTR = STR_END - IN_UCHARS(1) */ + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); + check_partial(common, TRUE); + add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); + JUMPHERE(jump[1]); + } + + OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff)); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff)); + } + else + { + peek_char(common); + check_newlinechar(common, common->nltype, backtracks, FALSE); + } + JUMPHERE(jump[0]); + return cc; + + case OP_CHAR: + case OP_CHARI: + length = 1; +#ifdef SUPPORT_UTF + if (common->utf && HAS_EXTRALEN(*cc)) length += GET_EXTRALEN(*cc); +#endif + if (common->mode == JIT_COMPILE && (type == OP_CHAR || !char_has_othercase(common, cc) || char_get_othercase_bit(common, cc) != 0)) + { + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(length)); + add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER, STR_PTR, 0, STR_END, 0)); + + context.length = IN_UCHARS(length); + context.sourcereg = -1; +#if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED + context.ucharptr = 0; +#endif + return byte_sequence_compare(common, type == OP_CHARI, cc, &context, backtracks); + } + detect_partial_match(common, backtracks); + read_char(common); +#ifdef SUPPORT_UTF + if (common->utf) + { + GETCHAR(c, cc); + } + else +#endif + c = *cc; + if (type == OP_CHAR || !char_has_othercase(common, cc)) + { + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c)); + return cc + length; + } + oc = char_othercase(common, c); + bit = c ^ oc; + if (ispowerof2(bit)) + { + OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, bit); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c | bit)); + return cc + length; + } + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c); + COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, char_othercase(common, c)); + COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); + add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO)); + return cc + length; + + case OP_NOT: + case OP_NOTI: + detect_partial_match(common, backtracks); + length = 1; +#ifdef SUPPORT_UTF + if (common->utf) + { +#ifdef COMPILE_PCRE8 + c = *cc; + if (c < 128) + { + OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), 0); + if (type == OP_NOT || !char_has_othercase(common, cc)) + add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c)); + else + { + /* Since UTF8 code page is fixed, we know that c is in [a-z] or [A-Z] range. */ + OP2(SLJIT_OR, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x20); + add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, c | 0x20)); + } + /* Skip the variable-length character. */ + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1)); + jump[0] = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0); + OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)PRIV(utf8_table4) - 0xc0); + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0); + JUMPHERE(jump[0]); + return cc + 1; + } + else +#endif /* COMPILE_PCRE8 */ + { + GETCHARLEN(c, cc, length); + read_char(common); + } + } + else +#endif /* SUPPORT_UTF */ + { + read_char(common); + c = *cc; + } + + if (type == OP_NOT || !char_has_othercase(common, cc)) + add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c)); + else + { + oc = char_othercase(common, c); + bit = c ^ oc; + if (ispowerof2(bit)) + { + OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, bit); + add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c | bit)); + } + else + { + add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c)); + add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, oc)); + } + } + return cc + length; + + case OP_CLASS: + case OP_NCLASS: + detect_partial_match(common, backtracks); + read_char(common); +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + jump[0] = NULL; +#ifdef COMPILE_PCRE8 + /* This check only affects 8 bit mode. In other modes, we + always need to compare the value with 255. */ + if (common->utf) +#endif /* COMPILE_PCRE8 */ + { + jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255); + if (type == OP_CLASS) + { + add_jump(compiler, backtracks, jump[0]); + jump[0] = NULL; + } + } +#endif /* SUPPORT_UTF || !COMPILE_PCRE8 */ + OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7); + OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3); + OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_w)cc); + OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0); + OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0); + add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO)); +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + if (jump[0] != NULL) + JUMPHERE(jump[0]); +#endif /* SUPPORT_UTF || !COMPILE_PCRE8 */ + return cc + 32 / sizeof(pcre_uchar); + +#if defined SUPPORT_UTF || defined COMPILE_PCRE16 + case OP_XCLASS: + compile_xclass_trypath(common, cc + LINK_SIZE, backtracks); + return cc + GET(cc, 0) - 1; +#endif + + case OP_REVERSE: + length = GET(cc, 0); + if (length == 0) + return cc + LINK_SIZE; + OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); +#ifdef SUPPORT_UTF + if (common->utf) + { + OP1(SLJIT_MOV, TMP3, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); + OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, length); + label = LABEL(); + add_jump(compiler, backtracks, CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP3, 0)); + skip_char_back(common); + OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_IMM, 1); + JUMPTO(SLJIT_C_NOT_ZERO, label); + } + else +#endif + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin)); + OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(length)); + add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, STR_PTR, 0, TMP1, 0)); + } + check_start_used_ptr(common); + return cc + LINK_SIZE; + } +SLJIT_ASSERT_STOP(); +return cc; +} + +static SLJIT_INLINE pcre_uchar *compile_charn_trypath(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, jump_list **backtracks) +{ +/* This function consumes at least one input character. */ +/* To decrease the number of length checks, we try to concatenate the fixed length character sequences. */ +DEFINE_COMPILER; +pcre_uchar *ccbegin = cc; +compare_context context; +int size; + +context.length = 0; +do + { + if (cc >= ccend) + break; + + if (*cc == OP_CHAR) + { + size = 1; +#ifdef SUPPORT_UTF + if (common->utf && HAS_EXTRALEN(cc[1])) + size += GET_EXTRALEN(cc[1]); +#endif + } + else if (*cc == OP_CHARI) + { + size = 1; +#ifdef SUPPORT_UTF + if (common->utf) + { + if (char_has_othercase(common, cc + 1) && char_get_othercase_bit(common, cc + 1) == 0) + size = 0; + else if (HAS_EXTRALEN(cc[1])) + size += GET_EXTRALEN(cc[1]); + } + else +#endif + if (char_has_othercase(common, cc + 1) && char_get_othercase_bit(common, cc + 1) == 0) + size = 0; + } + else + size = 0; + + cc += 1 + size; + context.length += IN_UCHARS(size); + } +while (size > 0 && context.length <= 128); + +cc = ccbegin; +if (context.length > 0) + { + /* We have a fixed-length byte sequence. */ + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, context.length); + add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER, STR_PTR, 0, STR_END, 0)); + + context.sourcereg = -1; +#if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED + context.ucharptr = 0; +#endif + do cc = byte_sequence_compare(common, *cc == OP_CHARI, cc + 1, &context, backtracks); while (context.length > 0); + return cc; + } + +/* A non-fixed length character will be checked if length == 0. */ +return compile_char1_trypath(common, *cc, cc + 1, backtracks); +} + +static struct sljit_jump *compile_ref_checks(compiler_common *common, pcre_uchar *cc, jump_list **backtracks) +{ +DEFINE_COMPILER; +int offset = GET2(cc, 1) << 1; + +OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); +if (!common->jscript_compat) + { + if (backtracks == NULL) + { + /* OVECTOR(1) contains the "string begin - 1" constant. */ + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)); + COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_EQUAL); + OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); + COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_EQUAL); + return JUMP(SLJIT_C_NOT_ZERO); + } + add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1))); + } +return CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); +} + +/* Forward definitions. */ +static void compile_trypath(compiler_common *, pcre_uchar *, pcre_uchar *, backtrack_common *); +static void compile_backtrackpath(compiler_common *, struct backtrack_common *); + +#define PUSH_BACKTRACK(size, ccstart, error) \ + do \ + { \ + backtrack = sljit_alloc_memory(compiler, (size)); \ + if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) \ + return error; \ + memset(backtrack, 0, size); \ + backtrack->prev = parent->top; \ + backtrack->cc = (ccstart); \ + parent->top = backtrack; \ + } \ + while (0) + +#define PUSH_BACKTRACK_NOVALUE(size, ccstart) \ + do \ + { \ + backtrack = sljit_alloc_memory(compiler, (size)); \ + if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) \ + return; \ + memset(backtrack, 0, size); \ + backtrack->prev = parent->top; \ + backtrack->cc = (ccstart); \ + parent->top = backtrack; \ + } \ + while (0) + +#define BACKTRACK_AS(type) ((type *)backtrack) + +static pcre_uchar *compile_ref_trypath(compiler_common *common, pcre_uchar *cc, jump_list **backtracks, BOOL withchecks, BOOL emptyfail) +{ +DEFINE_COMPILER; +int offset = GET2(cc, 1) << 1; +struct sljit_jump *jump = NULL; +struct sljit_jump *partial; +struct sljit_jump *nopartial; + +OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); +/* OVECTOR(1) contains the "string begin - 1" constant. */ +if (withchecks && !common->jscript_compat) + add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1))); + +#if defined SUPPORT_UTF && defined SUPPORT_UCP +if (common->utf && *cc == OP_REFI) + { + SLJIT_ASSERT(TMP1 == SLJIT_TEMPORARY_REG1 && STACK_TOP == SLJIT_TEMPORARY_REG2 && TMP2 == SLJIT_TEMPORARY_REG3); + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); + if (withchecks) + jump = CMP(SLJIT_C_EQUAL, TMP1, 0, TMP2, 0); + + /* Needed to save important temporary registers. */ + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, STACK_TOP, 0); + OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG2, 0, ARGUMENTS, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_TEMPORARY_REG2), SLJIT_OFFSETOF(jit_arguments, uchar_ptr), STR_PTR, 0); + sljit_emit_ijump(compiler, SLJIT_CALL3, SLJIT_IMM, SLJIT_FUNC_OFFSET(do_utf_caselesscmp)); + OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); + if (common->mode == JIT_COMPILE) + add_jump(compiler, backtracks, CMP(SLJIT_C_LESS_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1)); + else + { + add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 0)); + nopartial = CMP(SLJIT_C_NOT_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1); + check_partial(common, FALSE); + add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); + JUMPHERE(nopartial); + } + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_RETURN_REG, 0); + } +else +#endif /* SUPPORT_UTF && SUPPORT_UCP */ + { + OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), TMP1, 0); + if (withchecks) + jump = JUMP(SLJIT_C_ZERO); + + OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0); + partial = CMP(SLJIT_C_GREATER, STR_PTR, 0, STR_END, 0); + if (common->mode == JIT_COMPILE) + add_jump(compiler, backtracks, partial); + + add_jump(compiler, *cc == OP_REF ? &common->casefulcmp : &common->caselesscmp, JUMP(SLJIT_FAST_CALL)); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); + + if (common->mode != JIT_COMPILE) + { + nopartial = JUMP(SLJIT_JUMP); + JUMPHERE(partial); + /* TMP2 -= STR_END - STR_PTR */ + OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, STR_PTR, 0); + OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, STR_END, 0); + partial = CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, 0); + OP1(SLJIT_MOV, STR_PTR, 0, STR_END, 0); + add_jump(compiler, *cc == OP_REF ? &common->casefulcmp : &common->caselesscmp, JUMP(SLJIT_FAST_CALL)); + add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); + JUMPHERE(partial); + check_partial(common, FALSE); + add_jump(compiler, backtracks, JUMP(SLJIT_JUMP)); + JUMPHERE(nopartial); + } + } + +if (jump != NULL) + { + if (emptyfail) + add_jump(compiler, backtracks, jump); + else + JUMPHERE(jump); + } +return cc + 1 + IMM2_SIZE; +} + +static SLJIT_INLINE pcre_uchar *compile_ref_iterator_trypath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent) +{ +DEFINE_COMPILER; +backtrack_common *backtrack; +pcre_uchar type; +struct sljit_label *label; +struct sljit_jump *zerolength; +struct sljit_jump *jump = NULL; +pcre_uchar *ccbegin = cc; +int min = 0, max = 0; +BOOL minimize; + +PUSH_BACKTRACK(sizeof(iterator_backtrack), cc, NULL); + +type = cc[1 + IMM2_SIZE]; +minimize = (type & 0x1) != 0; +switch(type) + { + case OP_CRSTAR: + case OP_CRMINSTAR: + min = 0; + max = 0; + cc += 1 + IMM2_SIZE + 1; + break; + case OP_CRPLUS: + case OP_CRMINPLUS: + min = 1; + max = 0; + cc += 1 + IMM2_SIZE + 1; + break; + case OP_CRQUERY: + case OP_CRMINQUERY: + min = 0; + max = 1; + cc += 1 + IMM2_SIZE + 1; + break; + case OP_CRRANGE: + case OP_CRMINRANGE: + min = GET2(cc, 1 + IMM2_SIZE + 1); + max = GET2(cc, 1 + IMM2_SIZE + 1 + IMM2_SIZE); + cc += 1 + IMM2_SIZE + 1 + 2 * IMM2_SIZE; + break; + default: + SLJIT_ASSERT_STOP(); + break; + } + +if (!minimize) + { + if (min == 0) + { + allocate_stack(common, 2); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0); + /* Temporary release of STR_PTR. */ + OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_w)); + zerolength = compile_ref_checks(common, ccbegin, NULL); + /* Restore if not zero length. */ + OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_w)); + } + else + { + allocate_stack(common, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); + zerolength = compile_ref_checks(common, ccbegin, &backtrack->topbacktracks); + } + + if (min > 1 || max > 1) + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, SLJIT_IMM, 0); + + label = LABEL(); + compile_ref_trypath(common, ccbegin, &backtrack->topbacktracks, FALSE, FALSE); + + if (min > 1 || max > 1) + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); + OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, TMP1, 0); + if (min > 1) + CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, min, label); + if (max > 1) + { + jump = CMP(SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, max); + allocate_stack(common, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + JUMPTO(SLJIT_JUMP, label); + JUMPHERE(jump); + } + } + + if (max == 0) + { + /* Includes min > 1 case as well. */ + allocate_stack(common, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + JUMPTO(SLJIT_JUMP, label); + } + + JUMPHERE(zerolength); + BACKTRACK_AS(iterator_backtrack)->trypath = LABEL(); + + decrease_call_count(common); + return cc; + } + +allocate_stack(common, 2); +OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); +if (type != OP_CRMINSTAR) + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0); + +if (min == 0) + { + zerolength = compile_ref_checks(common, ccbegin, NULL); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + jump = JUMP(SLJIT_JUMP); + } +else + zerolength = compile_ref_checks(common, ccbegin, &backtrack->topbacktracks); + +BACKTRACK_AS(iterator_backtrack)->trypath = LABEL(); +if (max > 0) + add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, max)); + +compile_ref_trypath(common, ccbegin, &backtrack->topbacktracks, TRUE, TRUE); +OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + +if (min > 1) + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); + OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0); + CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, min, BACKTRACK_AS(iterator_backtrack)->trypath); + } +else if (max > 0) + OP2(SLJIT_ADD, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 1); + +if (jump != NULL) + JUMPHERE(jump); +JUMPHERE(zerolength); + +decrease_call_count(common); +return cc; +} + +static SLJIT_INLINE pcre_uchar *compile_recurse_trypath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent) +{ +DEFINE_COMPILER; +backtrack_common *backtrack; +recurse_entry *entry = common->entries; +recurse_entry *prev = NULL; +int start = GET(cc, 1); + +PUSH_BACKTRACK(sizeof(recurse_backtrack), cc, NULL); +while (entry != NULL) + { + if (entry->start == start) + break; + prev = entry; + entry = entry->next; + } + +if (entry == NULL) + { + entry = sljit_alloc_memory(compiler, sizeof(recurse_entry)); + if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) + return NULL; + entry->next = NULL; + entry->entry = NULL; + entry->calls = NULL; + entry->start = start; + + if (prev != NULL) + prev->next = entry; + else + common->entries = entry; + } + +if (common->has_set_som && common->mark_ptr != 0) + { + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); + allocate_stack(common, 2); + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0); + } +else if (common->has_set_som || common->mark_ptr != 0) + { + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->has_set_som ? (int)(OVECTOR(0)) : common->mark_ptr); + allocate_stack(common, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0); + } + +if (entry->entry == NULL) + add_jump(compiler, &entry->calls, JUMP(SLJIT_FAST_CALL)); +else + JUMPTO(SLJIT_FAST_CALL, entry->entry); +/* Leave if the match is failed. */ +add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0)); +return cc + 1 + LINK_SIZE; +} + +static pcre_uchar *compile_assert_trypath(compiler_common *common, pcre_uchar *cc, assert_backtrack *backtrack, BOOL conditional) +{ +DEFINE_COMPILER; +int framesize; +int localptr; +backtrack_common altbacktrack; +pcre_uchar *ccbegin; +pcre_uchar opcode; +pcre_uchar bra = OP_BRA; +jump_list *tmp = NULL; +jump_list **target = (conditional) ? &backtrack->condfailed : &backtrack->common.topbacktracks; +jump_list **found; +/* Saving previous accept variables. */ +struct sljit_label *save_leavelabel = common->leavelabel; +struct sljit_label *save_acceptlabel = common->acceptlabel; +jump_list *save_leave = common->leave; +jump_list *save_accept = common->accept; +struct sljit_jump *jump; +struct sljit_jump *brajump = NULL; + +if (*cc == OP_BRAZERO || *cc == OP_BRAMINZERO) + { + SLJIT_ASSERT(!conditional); + bra = *cc; + cc++; + } +localptr = PRIV_DATA(cc); +SLJIT_ASSERT(localptr != 0); +framesize = get_framesize(common, cc, FALSE); +backtrack->framesize = framesize; +backtrack->localptr = localptr; +opcode = *cc; +SLJIT_ASSERT(opcode >= OP_ASSERT && opcode <= OP_ASSERTBACK_NOT); +found = (opcode == OP_ASSERT || opcode == OP_ASSERTBACK) ? &tmp : target; +ccbegin = cc; +cc += GET(cc, 1); + +if (bra == OP_BRAMINZERO) + { + /* This is a braminzero backtrack path. */ + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + free_stack(common, 1); + brajump = CMP(SLJIT_C_EQUAL, STR_PTR, 0, SLJIT_IMM, 0); + } + +if (framesize < 0) + { + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, STACK_TOP, 0); + allocate_stack(common, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + } +else + { + allocate_stack(common, framesize + 2); + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + OP2(SLJIT_SUB, TMP2, 0, STACK_TOP, 0, SLJIT_IMM, -STACK(framesize + 1)); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, TMP2, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0); + init_frame(common, ccbegin, framesize + 1, 2, FALSE); + } + +memset(&altbacktrack, 0, sizeof(backtrack_common)); +common->leavelabel = NULL; +common->leave = NULL; +while (1) + { + common->acceptlabel = NULL; + common->accept = NULL; + altbacktrack.top = NULL; + altbacktrack.topbacktracks = NULL; + + if (*ccbegin == OP_ALT) + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + + altbacktrack.cc = ccbegin; + compile_trypath(common, ccbegin + 1 + LINK_SIZE, cc, &altbacktrack); + if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) + { + common->leavelabel = save_leavelabel; + common->acceptlabel = save_acceptlabel; + common->leave = save_leave; + common->accept = save_accept; + return NULL; + } + common->acceptlabel = LABEL(); + if (common->accept != NULL) + set_jumps(common->accept, common->acceptlabel); + + /* Reset stack. */ + if (framesize < 0) + OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + else { + if ((opcode != OP_ASSERT_NOT && opcode != OP_ASSERTBACK_NOT) || conditional) + { + /* We don't need to keep the STR_PTR, only the previous localptr. */ + OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_IMM, (framesize + 1) * sizeof(sljit_w)); + } + else + { + OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); + } + } + + if (opcode == OP_ASSERT_NOT || opcode == OP_ASSERTBACK_NOT) + { + /* We know that STR_PTR was stored on the top of the stack. */ + if (conditional) + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), 0); + else if (bra == OP_BRAZERO) + { + if (framesize < 0) + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), 0); + else + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), framesize * sizeof(sljit_w)); + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), (framesize + 1) * sizeof(sljit_w)); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, TMP1, 0); + } + OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_w)); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); + } + else if (framesize >= 0) + { + /* For OP_BRA and OP_BRAMINZERO. */ + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_MEM1(STACK_TOP), framesize * sizeof(sljit_w)); + } + } + add_jump(compiler, found, JUMP(SLJIT_JUMP)); + + compile_backtrackpath(common, altbacktrack.top); + if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) + { + common->leavelabel = save_leavelabel; + common->acceptlabel = save_acceptlabel; + common->leave = save_leave; + common->accept = save_accept; + return NULL; + } + set_jumps(altbacktrack.topbacktracks, LABEL()); + + if (*cc != OP_ALT) + break; + + ccbegin = cc; + cc += GET(cc, 1); + } +/* None of them matched. */ +if (common->leave != NULL) + set_jumps(common->leave, LABEL()); + +if (opcode == OP_ASSERT || opcode == OP_ASSERTBACK) + { + /* Assert is failed. */ + if (conditional || bra == OP_BRAZERO) + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + + if (framesize < 0) + { + /* The topmost item should be 0. */ + if (bra == OP_BRAZERO) + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); + else + free_stack(common, 1); + } + else + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); + /* The topmost item should be 0. */ + if (bra == OP_BRAZERO) + { + free_stack(common, framesize + 1); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); + } + else + free_stack(common, framesize + 2); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, TMP1, 0); + } + jump = JUMP(SLJIT_JUMP); + if (bra != OP_BRAZERO) + add_jump(compiler, target, jump); + + /* Assert is successful. */ + set_jumps(tmp, LABEL()); + if (framesize < 0) + { + /* We know that STR_PTR was stored on the top of the stack. */ + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), 0); + /* Keep the STR_PTR on the top of the stack. */ + if (bra == OP_BRAZERO) + OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_w)); + else if (bra == OP_BRAMINZERO) + { + OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_w)); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); + } + } + else + { + if (bra == OP_BRA) + { + /* We don't need to keep the STR_PTR, only the previous localptr. */ + OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_IMM, (framesize + 1) * sizeof(sljit_w)); + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), 0); + } + else + { + /* We don't need to keep the STR_PTR, only the previous localptr. */ + OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_IMM, (framesize + 2) * sizeof(sljit_w)); + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), bra == OP_BRAZERO ? STR_PTR : SLJIT_IMM, 0); + } + } + + if (bra == OP_BRAZERO) + { + backtrack->trypath = LABEL(); + sljit_set_label(jump, backtrack->trypath); + } + else if (bra == OP_BRAMINZERO) + { + JUMPTO(SLJIT_JUMP, backtrack->trypath); + JUMPHERE(brajump); + if (framesize >= 0) + { + OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_MEM1(STACK_TOP), framesize * sizeof(sljit_w)); + } + set_jumps(backtrack->common.topbacktracks, LABEL()); + } + } +else + { + /* AssertNot is successful. */ + if (framesize < 0) + { + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + if (bra != OP_BRA) + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); + else + free_stack(common, 1); + } + else + { + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); + /* The topmost item should be 0. */ + if (bra != OP_BRA) + { + free_stack(common, framesize + 1); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); + } + else + free_stack(common, framesize + 2); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, TMP1, 0); + } + + if (bra == OP_BRAZERO) + backtrack->trypath = LABEL(); + else if (bra == OP_BRAMINZERO) + { + JUMPTO(SLJIT_JUMP, backtrack->trypath); + JUMPHERE(brajump); + } + + if (bra != OP_BRA) + { + SLJIT_ASSERT(found == &backtrack->common.topbacktracks); + set_jumps(backtrack->common.topbacktracks, LABEL()); + backtrack->common.topbacktracks = NULL; + } + } + +common->leavelabel = save_leavelabel; +common->acceptlabel = save_acceptlabel; +common->leave = save_leave; +common->accept = save_accept; +return cc + 1 + LINK_SIZE; +} + +static sljit_w SLJIT_CALL do_searchovector(sljit_w refno, sljit_w* locals, pcre_uchar *name_table) +{ +int condition = FALSE; +pcre_uchar *slotA = name_table; +pcre_uchar *slotB; +sljit_w name_count = locals[LOCALS0 / sizeof(sljit_w)]; +sljit_w name_entry_size = locals[LOCALS1 / sizeof(sljit_w)]; +sljit_w no_capture; +int i; + +locals += refno & 0xff; +refno >>= 8; +no_capture = locals[1]; + +for (i = 0; i < name_count; i++) + { + if (GET2(slotA, 0) == refno) break; + slotA += name_entry_size; + } + +if (i < name_count) + { + /* Found a name for the number - there can be only one; duplicate names + for different numbers are allowed, but not vice versa. First scan down + for duplicates. */ + + slotB = slotA; + while (slotB > name_table) + { + slotB -= name_entry_size; + if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) + { + condition = locals[GET2(slotB, 0) << 1] != no_capture; + if (condition) break; + } + else break; + } + + /* Scan up for duplicates */ + if (!condition) + { + slotB = slotA; + for (i++; i < name_count; i++) + { + slotB += name_entry_size; + if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) + { + condition = locals[GET2(slotB, 0) << 1] != no_capture; + if (condition) break; + } + else break; + } + } + } +return condition; +} + +static sljit_w SLJIT_CALL do_searchgroups(sljit_w recno, sljit_w* locals, pcre_uchar *name_table) +{ +int condition = FALSE; +pcre_uchar *slotA = name_table; +pcre_uchar *slotB; +sljit_w name_count = locals[LOCALS0 / sizeof(sljit_w)]; +sljit_w name_entry_size = locals[LOCALS1 / sizeof(sljit_w)]; +sljit_w group_num = locals[POSSESSIVE0 / sizeof(sljit_w)]; +int i; + +for (i = 0; i < name_count; i++) + { + if (GET2(slotA, 0) == recno) break; + slotA += name_entry_size; + } + +if (i < name_count) + { + /* Found a name for the number - there can be only one; duplicate + names for different numbers are allowed, but not vice versa. First + scan down for duplicates. */ + + slotB = slotA; + while (slotB > name_table) + { + slotB -= name_entry_size; + if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) + { + condition = GET2(slotB, 0) == group_num; + if (condition) break; + } + else break; + } + + /* Scan up for duplicates */ + if (!condition) + { + slotB = slotA; + for (i++; i < name_count; i++) + { + slotB += name_entry_size; + if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) + { + condition = GET2(slotB, 0) == group_num; + if (condition) break; + } + else break; + } + } + } +return condition; +} + +/* + Handling bracketed expressions is probably the most complex part. + + Stack layout naming characters: + S - Push the current STR_PTR + 0 - Push a 0 (NULL) + A - Push the current STR_PTR. Needed for restoring the STR_PTR + before the next alternative. Not pushed if there are no alternatives. + M - Any values pushed by the current alternative. Can be empty, or anything. + C - Push the previous OVECTOR(i), OVECTOR(i+1) and OVECTOR_PRIV(i) to the stack. + L - Push the previous local (pointed by localptr) to the stack + () - opional values stored on the stack + ()* - optonal, can be stored multiple times + + The following list shows the regular expression templates, their PCRE byte codes + and stack layout supported by pcre-sljit. + + (?:) OP_BRA | OP_KET A M + () OP_CBRA | OP_KET C M + (?:)+ OP_BRA | OP_KETRMAX 0 A M S ( A M S )* + OP_SBRA | OP_KETRMAX 0 L M S ( L M S )* + (?:)+? OP_BRA | OP_KETRMIN 0 A M S ( A M S )* + OP_SBRA | OP_KETRMIN 0 L M S ( L M S )* + ()+ OP_CBRA | OP_KETRMAX 0 C M S ( C M S )* + OP_SCBRA | OP_KETRMAX 0 C M S ( C M S )* + ()+? OP_CBRA | OP_KETRMIN 0 C M S ( C M S )* + OP_SCBRA | OP_KETRMIN 0 C M S ( C M S )* + (?:)? OP_BRAZERO | OP_BRA | OP_KET S ( A M 0 ) + (?:)?? OP_BRAMINZERO | OP_BRA | OP_KET S ( A M 0 ) + ()? OP_BRAZERO | OP_CBRA | OP_KET S ( C M 0 ) + ()?? OP_BRAMINZERO | OP_CBRA | OP_KET S ( C M 0 ) + (?:)* OP_BRAZERO | OP_BRA | OP_KETRMAX S 0 ( A M S )* + OP_BRAZERO | OP_SBRA | OP_KETRMAX S 0 ( L M S )* + (?:)*? OP_BRAMINZERO | OP_BRA | OP_KETRMIN S 0 ( A M S )* + OP_BRAMINZERO | OP_SBRA | OP_KETRMIN S 0 ( L M S )* + ()* OP_BRAZERO | OP_CBRA | OP_KETRMAX S 0 ( C M S )* + OP_BRAZERO | OP_SCBRA | OP_KETRMAX S 0 ( C M S )* + ()*? OP_BRAMINZERO | OP_CBRA | OP_KETRMIN S 0 ( C M S )* + OP_BRAMINZERO | OP_SCBRA | OP_KETRMIN S 0 ( C M S )* + + + Stack layout naming characters: + A - Push the alternative index (starting from 0) on the stack. + Not pushed if there is no alternatives. + M - Any values pushed by the current alternative. Can be empty, or anything. + + The next list shows the possible content of a bracket: + (|) OP_*BRA | OP_ALT ... M A + (?()|) OP_*COND | OP_ALT M A + (?>|) OP_ONCE | OP_ALT ... [stack trace] M A + (?>|) OP_ONCE_NC | OP_ALT ... [stack trace] M A + Or nothing, if trace is unnecessary +*/ + +static pcre_uchar *compile_bracket_trypath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent) +{ +DEFINE_COMPILER; +backtrack_common *backtrack; +pcre_uchar opcode; +int localptr = 0; +int offset = 0; +int stacksize; +pcre_uchar *ccbegin; +pcre_uchar *trypath; +pcre_uchar bra = OP_BRA; +pcre_uchar ket; +assert_backtrack *assert; +BOOL has_alternatives; +struct sljit_jump *jump; +struct sljit_jump *skip; +struct sljit_label *rmaxlabel = NULL; +struct sljit_jump *braminzerojump = NULL; + +PUSH_BACKTRACK(sizeof(bracket_backtrack), cc, NULL); + +if (*cc == OP_BRAZERO || *cc == OP_BRAMINZERO) + { + bra = *cc; + cc++; + opcode = *cc; + } + +opcode = *cc; +ccbegin = cc; +trypath = ccbegin + 1 + LINK_SIZE; + +if ((opcode == OP_COND || opcode == OP_SCOND) && cc[1 + LINK_SIZE] == OP_DEF) + { + /* Drop this bracket_backtrack. */ + parent->top = backtrack->prev; + return bracketend(cc); + } + +ket = *(bracketend(cc) - 1 - LINK_SIZE); +SLJIT_ASSERT(ket == OP_KET || ket == OP_KETRMAX || ket == OP_KETRMIN); +SLJIT_ASSERT(!((bra == OP_BRAZERO && ket == OP_KETRMIN) || (bra == OP_BRAMINZERO && ket == OP_KETRMAX))); +cc += GET(cc, 1); + +has_alternatives = *cc == OP_ALT; +if (SLJIT_UNLIKELY(opcode == OP_COND) || SLJIT_UNLIKELY(opcode == OP_SCOND)) + { + has_alternatives = (*trypath == OP_RREF) ? FALSE : TRUE; + if (*trypath == OP_NRREF) + { + stacksize = GET2(trypath, 1); + if (common->currententry == NULL || stacksize == RREF_ANY) + has_alternatives = FALSE; + else if (common->currententry->start == 0) + has_alternatives = stacksize != 0; + else + has_alternatives = stacksize != GET2(common->start, common->currententry->start + 1 + LINK_SIZE); + } + } + +if (SLJIT_UNLIKELY(opcode == OP_COND) && (*cc == OP_KETRMAX || *cc == OP_KETRMIN)) + opcode = OP_SCOND; +if (SLJIT_UNLIKELY(opcode == OP_ONCE_NC)) + opcode = OP_ONCE; + +if (opcode == OP_CBRA || opcode == OP_SCBRA) + { + /* Capturing brackets has a pre-allocated space. */ + offset = GET2(ccbegin, 1 + LINK_SIZE); + localptr = OVECTOR_PRIV(offset); + offset <<= 1; + BACKTRACK_AS(bracket_backtrack)->localptr = localptr; + trypath += IMM2_SIZE; + } +else if (opcode == OP_ONCE || opcode == OP_SBRA || opcode == OP_SCOND) + { + /* Other brackets simply allocate the next entry. */ + localptr = PRIV_DATA(ccbegin); + SLJIT_ASSERT(localptr != 0); + BACKTRACK_AS(bracket_backtrack)->localptr = localptr; + if (opcode == OP_ONCE) + BACKTRACK_AS(bracket_backtrack)->u.framesize = get_framesize(common, ccbegin, FALSE); + } + +/* Instructions before the first alternative. */ +stacksize = 0; +if ((ket == OP_KETRMAX) || (ket == OP_KETRMIN && bra != OP_BRAMINZERO)) + stacksize++; +if (bra == OP_BRAZERO) + stacksize++; + +if (stacksize > 0) + allocate_stack(common, stacksize); + +stacksize = 0; +if ((ket == OP_KETRMAX) || (ket == OP_KETRMIN && bra != OP_BRAMINZERO)) + { + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), SLJIT_IMM, 0); + stacksize++; + } + +if (bra == OP_BRAZERO) + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), STR_PTR, 0); + +if (bra == OP_BRAMINZERO) + { + /* This is a backtrack path! (Since the try-path of OP_BRAMINZERO matches to the empty string) */ + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + if (ket != OP_KETRMIN) + { + free_stack(common, 1); + braminzerojump = CMP(SLJIT_C_EQUAL, STR_PTR, 0, SLJIT_IMM, 0); + } + else + { + if (opcode == OP_ONCE || opcode >= OP_SBRA) + { + jump = CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0); + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); + /* Nothing stored during the first run. */ + skip = JUMP(SLJIT_JUMP); + JUMPHERE(jump); + /* Checking zero-length iteration. */ + if (opcode != OP_ONCE || BACKTRACK_AS(bracket_backtrack)->u.framesize < 0) + { + /* When we come from outside, localptr contains the previous STR_PTR. */ + braminzerojump = CMP(SLJIT_C_EQUAL, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + } + else + { + /* Except when the whole stack frame must be saved. */ + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + braminzerojump = CMP(SLJIT_C_EQUAL, STR_PTR, 0, SLJIT_MEM1(TMP1), (BACKTRACK_AS(bracket_backtrack)->u.framesize + 1) * sizeof(sljit_w)); + } + JUMPHERE(skip); + } + else + { + jump = CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0); + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); + JUMPHERE(jump); + } + } + } + +if (ket == OP_KETRMIN) + BACKTRACK_AS(bracket_backtrack)->recursivetrypath = LABEL(); + +if (ket == OP_KETRMAX) + { + rmaxlabel = LABEL(); + if (has_alternatives && opcode != OP_ONCE && opcode < OP_SBRA) + BACKTRACK_AS(bracket_backtrack)->alttrypath = rmaxlabel; + } + +/* Handling capturing brackets and alternatives. */ +if (opcode == OP_ONCE) + { + if (BACKTRACK_AS(bracket_backtrack)->u.framesize < 0) + { + /* Neither capturing brackets nor recursions are not found in the block. */ + if (ket == OP_KETRMIN) + { + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + allocate_stack(common, 2); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP2, 0); + OP2(SLJIT_SUB, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_w)); + } + else if (ket == OP_KETRMAX || has_alternatives) + { + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, STACK_TOP, 0); + allocate_stack(common, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + } + else + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, STACK_TOP, 0); + } + else + { + if (ket == OP_KETRMIN || ket == OP_KETRMAX || has_alternatives) + { + allocate_stack(common, BACKTRACK_AS(bracket_backtrack)->u.framesize + 2); + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + OP2(SLJIT_SUB, TMP2, 0, STACK_TOP, 0, SLJIT_IMM, -STACK(BACKTRACK_AS(bracket_backtrack)->u.framesize + 1)); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, TMP2, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0); + init_frame(common, ccbegin, BACKTRACK_AS(bracket_backtrack)->u.framesize + 1, 2, FALSE); + } + else + { + allocate_stack(common, BACKTRACK_AS(bracket_backtrack)->u.framesize + 1); + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + OP2(SLJIT_SUB, TMP2, 0, STACK_TOP, 0, SLJIT_IMM, -STACK(BACKTRACK_AS(bracket_backtrack)->u.framesize)); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, TMP2, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP1, 0); + init_frame(common, ccbegin, BACKTRACK_AS(bracket_backtrack)->u.framesize, 1, FALSE); + } + } + } +else if (opcode == OP_CBRA || opcode == OP_SCBRA) + { + /* Saving the previous values. */ + allocate_stack(common, 3); + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP1, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP2, 0); + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, STR_PTR, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), TMP1, 0); + } +else if (opcode == OP_SBRA || opcode == OP_SCOND) + { + /* Saving the previous value. */ + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + allocate_stack(common, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, STR_PTR, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0); + } +else if (has_alternatives) + { + /* Pushing the starting string pointer. */ + allocate_stack(common, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + } + +/* Generating code for the first alternative. */ +if (opcode == OP_COND || opcode == OP_SCOND) + { + if (*trypath == OP_CREF) + { + SLJIT_ASSERT(has_alternatives); + add_jump(compiler, &(BACKTRACK_AS(bracket_backtrack)->u.condfailed), + CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(GET2(trypath, 1) << 1), SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1))); + trypath += 1 + IMM2_SIZE; + } + else if (*trypath == OP_NCREF) + { + SLJIT_ASSERT(has_alternatives); + stacksize = GET2(trypath, 1); + jump = CMP(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(stacksize << 1), SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)); + + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, STACK_TOP, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, SLJIT_IMM, common->name_count); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, SLJIT_IMM, common->name_entry_size); + OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG1, 0, SLJIT_IMM, (stacksize << 8) | (common->ovector_start / sizeof(sljit_w))); + GET_LOCAL_BASE(SLJIT_TEMPORARY_REG2, 0, 0); + OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, common->name_table); + sljit_emit_ijump(compiler, SLJIT_CALL3, SLJIT_IMM, SLJIT_FUNC_OFFSET(do_searchovector)); + OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1); + add_jump(compiler, &(BACKTRACK_AS(bracket_backtrack)->u.condfailed), CMP(SLJIT_C_EQUAL, SLJIT_TEMPORARY_REG1, 0, SLJIT_IMM, 0)); + + JUMPHERE(jump); + trypath += 1 + IMM2_SIZE; + } + else if (*trypath == OP_RREF || *trypath == OP_NRREF) + { + /* Never has other case. */ + BACKTRACK_AS(bracket_backtrack)->u.condfailed = NULL; + + stacksize = GET2(trypath, 1); + if (common->currententry == NULL) + stacksize = 0; + else if (stacksize == RREF_ANY) + stacksize = 1; + else if (common->currententry->start == 0) + stacksize = stacksize == 0; + else + stacksize = stacksize == GET2(common->start, common->currententry->start + 1 + LINK_SIZE); + + if (*trypath == OP_RREF || stacksize || common->currententry == NULL) + { + SLJIT_ASSERT(!has_alternatives); + if (stacksize != 0) + trypath += 1 + IMM2_SIZE; + else + { + if (*cc == OP_ALT) + { + trypath = cc + 1 + LINK_SIZE; + cc += GET(cc, 1); + } + else + trypath = cc; + } + } + else + { + SLJIT_ASSERT(has_alternatives); + + stacksize = GET2(trypath, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, STACK_TOP, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0, SLJIT_IMM, common->name_count); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, SLJIT_IMM, common->name_entry_size); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, SLJIT_IMM, GET2(common->start, common->currententry->start + 1 + LINK_SIZE)); + OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG1, 0, SLJIT_IMM, stacksize); + GET_LOCAL_BASE(SLJIT_TEMPORARY_REG2, 0, 0); + OP1(SLJIT_MOV, SLJIT_TEMPORARY_REG3, 0, SLJIT_IMM, common->name_table); + sljit_emit_ijump(compiler, SLJIT_CALL3, SLJIT_IMM, SLJIT_FUNC_OFFSET(do_searchgroups)); + OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1); + add_jump(compiler, &(BACKTRACK_AS(bracket_backtrack)->u.condfailed), CMP(SLJIT_C_EQUAL, SLJIT_TEMPORARY_REG1, 0, SLJIT_IMM, 0)); + trypath += 1 + IMM2_SIZE; + } + } + else + { + SLJIT_ASSERT(has_alternatives && *trypath >= OP_ASSERT && *trypath <= OP_ASSERTBACK_NOT); + /* Similar code as PUSH_BACKTRACK macro. */ + assert = sljit_alloc_memory(compiler, sizeof(assert_backtrack)); + if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) + return NULL; + memset(assert, 0, sizeof(assert_backtrack)); + assert->common.cc = trypath; + BACKTRACK_AS(bracket_backtrack)->u.assert = assert; + trypath = compile_assert_trypath(common, trypath, assert, TRUE); + } + } + +compile_trypath(common, trypath, cc, backtrack); +if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) + return NULL; + +if (opcode == OP_ONCE) + { + if (BACKTRACK_AS(bracket_backtrack)->u.framesize < 0) + { + OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + /* TMP2 which is set here used by OP_KETRMAX below. */ + if (ket == OP_KETRMAX) + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), 0); + else if (ket == OP_KETRMIN) + { + /* Move the STR_PTR to the localptr. */ + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_MEM1(STACK_TOP), 0); + } + } + else + { + stacksize = (ket == OP_KETRMIN || ket == OP_KETRMAX || has_alternatives) ? 2 : 1; + OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_IMM, (BACKTRACK_AS(bracket_backtrack)->u.framesize + stacksize) * sizeof(sljit_w)); + if (ket == OP_KETRMAX) + { + /* TMP2 which is set here used by OP_KETRMAX below. */ + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + } + } + } + +stacksize = 0; +if (ket != OP_KET || bra != OP_BRA) + stacksize++; +if (has_alternatives && opcode != OP_ONCE) + stacksize++; + +if (stacksize > 0) + allocate_stack(common, stacksize); + +stacksize = 0; +if (ket != OP_KET) + { + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), STR_PTR, 0); + stacksize++; + } +else if (bra != OP_BRA) + { + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), SLJIT_IMM, 0); + stacksize++; + } + +if (has_alternatives) + { + if (opcode != OP_ONCE) + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), SLJIT_IMM, 0); + if (ket != OP_KETRMAX) + BACKTRACK_AS(bracket_backtrack)->alttrypath = LABEL(); + } + +/* Must be after the trypath label. */ +if (offset != 0) + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), STR_PTR, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 0), TMP1, 0); + } + +if (ket == OP_KETRMAX) + { + if (opcode == OP_ONCE || opcode >= OP_SBRA) + { + if (has_alternatives) + BACKTRACK_AS(bracket_backtrack)->alttrypath = LABEL(); + /* Checking zero-length iteration. */ + if (opcode != OP_ONCE) + { + CMPTO(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, STR_PTR, 0, rmaxlabel); + /* Drop STR_PTR for greedy plus quantifier. */ + if (bra != OP_BRAZERO) + free_stack(common, 1); + } + else + /* TMP2 must contain the starting STR_PTR. */ + CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_PTR, 0, rmaxlabel); + } + else + JUMPTO(SLJIT_JUMP, rmaxlabel); + BACKTRACK_AS(bracket_backtrack)->recursivetrypath = LABEL(); + } + +if (bra == OP_BRAZERO) + BACKTRACK_AS(bracket_backtrack)->zerotrypath = LABEL(); + +if (bra == OP_BRAMINZERO) + { + /* This is a backtrack path! (From the viewpoint of OP_BRAMINZERO) */ + JUMPTO(SLJIT_JUMP, ((braminzero_backtrack *)parent)->trypath); + if (braminzerojump != NULL) + { + JUMPHERE(braminzerojump); + /* We need to release the end pointer to perform the + backtrack for the zero-length iteration. When + framesize is < 0, OP_ONCE will do the release itself. */ + if (opcode == OP_ONCE && BACKTRACK_AS(bracket_backtrack)->u.framesize >= 0) + { + OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); + } + else if (ket == OP_KETRMIN && opcode != OP_ONCE) + free_stack(common, 1); + } + /* Continue to the normal backtrack. */ + } + +if ((ket != OP_KET && bra != OP_BRAMINZERO) || bra == OP_BRAZERO) + decrease_call_count(common); + +/* Skip the other alternatives. */ +while (*cc == OP_ALT) + cc += GET(cc, 1); +cc += 1 + LINK_SIZE; +return cc; +} + +static pcre_uchar *compile_bracketpos_trypath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent) +{ +DEFINE_COMPILER; +backtrack_common *backtrack; +pcre_uchar opcode; +int localptr; +int cbraprivptr = 0; +int framesize; +int stacksize; +int offset = 0; +BOOL zero = FALSE; +pcre_uchar *ccbegin = NULL; +int stack; +struct sljit_label *loop = NULL; +struct jump_list *emptymatch = NULL; + +PUSH_BACKTRACK(sizeof(bracketpos_backtrack), cc, NULL); +if (*cc == OP_BRAPOSZERO) + { + zero = TRUE; + cc++; + } + +opcode = *cc; +localptr = PRIV_DATA(cc); +SLJIT_ASSERT(localptr != 0); +BACKTRACK_AS(bracketpos_backtrack)->localptr = localptr; +switch(opcode) + { + case OP_BRAPOS: + case OP_SBRAPOS: + ccbegin = cc + 1 + LINK_SIZE; + break; + + case OP_CBRAPOS: + case OP_SCBRAPOS: + offset = GET2(cc, 1 + LINK_SIZE); + cbraprivptr = OVECTOR_PRIV(offset); + offset <<= 1; + ccbegin = cc + 1 + LINK_SIZE + IMM2_SIZE; + break; + + default: + SLJIT_ASSERT_STOP(); + break; + } + +framesize = get_framesize(common, cc, FALSE); +BACKTRACK_AS(bracketpos_backtrack)->framesize = framesize; +if (framesize < 0) + { + stacksize = (opcode == OP_CBRAPOS || opcode == OP_SCBRAPOS) ? 2 : 1; + if (!zero) + stacksize++; + BACKTRACK_AS(bracketpos_backtrack)->stacksize = stacksize; + allocate_stack(common, stacksize); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, STACK_TOP, 0); + + if (opcode == OP_CBRAPOS || opcode == OP_SCBRAPOS) + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset)); + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1)); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP1, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP2, 0); + } + else + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + + if (!zero) + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize - 1), SLJIT_IMM, 1); + } +else + { + stacksize = framesize + 1; + if (!zero) + stacksize++; + if (opcode == OP_BRAPOS || opcode == OP_SBRAPOS) + stacksize++; + BACKTRACK_AS(bracketpos_backtrack)->stacksize = stacksize; + allocate_stack(common, stacksize); + + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + OP2(SLJIT_SUB, TMP2, 0, STACK_TOP, 0, SLJIT_IMM, -STACK(stacksize - 1)); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, TMP2, 0); + stack = 0; + if (!zero) + { + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 1); + stack++; + } + if (opcode == OP_BRAPOS || opcode == OP_SBRAPOS) + { + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stack), STR_PTR, 0); + stack++; + } + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stack), TMP1, 0); + init_frame(common, cc, stacksize - 1, stacksize - framesize, FALSE); + } + +if (opcode == OP_CBRAPOS || opcode == OP_SCBRAPOS) + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), cbraprivptr, STR_PTR, 0); + +loop = LABEL(); +while (*cc != OP_KETRPOS) + { + backtrack->top = NULL; + backtrack->topbacktracks = NULL; + cc += GET(cc, 1); + + compile_trypath(common, ccbegin, cc, backtrack); + if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) + return NULL; + + if (framesize < 0) + { + OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + + if (opcode == OP_CBRAPOS || opcode == OP_SCBRAPOS) + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), cbraprivptr); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), STR_PTR, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), cbraprivptr, STR_PTR, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0); + } + else + { + if (opcode == OP_SBRAPOS) + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + } + + if (opcode == OP_SBRAPOS || opcode == OP_SCBRAPOS) + add_jump(compiler, &emptymatch, CMP(SLJIT_C_EQUAL, TMP1, 0, STR_PTR, 0)); + + if (!zero) + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize - 1), SLJIT_IMM, 0); + } + else + { + if (opcode == OP_CBRAPOS || opcode == OP_SCBRAPOS) + { + OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_IMM, stacksize * sizeof(sljit_w)); + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), cbraprivptr); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), STR_PTR, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), cbraprivptr, STR_PTR, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0); + } + else + { + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + OP2(SLJIT_ADD, STACK_TOP, 0, TMP2, 0, SLJIT_IMM, stacksize * sizeof(sljit_w)); + if (opcode == OP_SBRAPOS) + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), (framesize + 1) * sizeof(sljit_w)); + OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), (framesize + 1) * sizeof(sljit_w), STR_PTR, 0); + } + + if (opcode == OP_SBRAPOS || opcode == OP_SCBRAPOS) + add_jump(compiler, &emptymatch, CMP(SLJIT_C_EQUAL, TMP1, 0, STR_PTR, 0)); + + if (!zero) + { + if (framesize < 0) + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize - 1), SLJIT_IMM, 0); + else + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); + } + } + JUMPTO(SLJIT_JUMP, loop); + flush_stubs(common); + + compile_backtrackpath(common, backtrack->top); + if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) + return NULL; + set_jumps(backtrack->topbacktracks, LABEL()); + + if (framesize < 0) + { + if (opcode == OP_CBRAPOS || opcode == OP_SCBRAPOS) + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), cbraprivptr); + else + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + } + else + { + if (opcode == OP_CBRAPOS || opcode == OP_SCBRAPOS) + { + /* Last alternative. */ + if (*cc == OP_KETRPOS) + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), cbraprivptr); + } + else + { + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(TMP2), (framesize + 1) * sizeof(sljit_w)); + } + } + + if (*cc == OP_KETRPOS) + break; + ccbegin = cc + 1 + LINK_SIZE; + } + +backtrack->topbacktracks = NULL; +if (!zero) + { + if (framesize < 0) + add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(STACK_TOP), STACK(stacksize - 1), SLJIT_IMM, 0)); + else /* TMP2 is set to [localptr] above. */ + add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(TMP2), (stacksize - 1) * sizeof(sljit_w), SLJIT_IMM, 0)); + } + +/* None of them matched. */ +set_jumps(emptymatch, LABEL()); +decrease_call_count(common); +return cc + 1 + LINK_SIZE; +} + +static SLJIT_INLINE pcre_uchar *get_iterator_parameters(compiler_common *common, pcre_uchar *cc, pcre_uchar *opcode, pcre_uchar *type, int *arg1, int *arg2, pcre_uchar **end) +{ +int class_len; + +*opcode = *cc; +if (*opcode >= OP_STAR && *opcode <= OP_POSUPTO) + { + cc++; + *type = OP_CHAR; + } +else if (*opcode >= OP_STARI && *opcode <= OP_POSUPTOI) + { + cc++; + *type = OP_CHARI; + *opcode -= OP_STARI - OP_STAR; + } +else if (*opcode >= OP_NOTSTAR && *opcode <= OP_NOTPOSUPTO) + { + cc++; + *type = OP_NOT; + *opcode -= OP_NOTSTAR - OP_STAR; + } +else if (*opcode >= OP_NOTSTARI && *opcode <= OP_NOTPOSUPTOI) + { + cc++; + *type = OP_NOTI; + *opcode -= OP_NOTSTARI - OP_STAR; + } +else if (*opcode >= OP_TYPESTAR && *opcode <= OP_TYPEPOSUPTO) + { + cc++; + *opcode -= OP_TYPESTAR - OP_STAR; + *type = 0; + } +else + { + SLJIT_ASSERT(*opcode >= OP_CLASS || *opcode <= OP_XCLASS); + *type = *opcode; + cc++; + class_len = (*type < OP_XCLASS) ? (int)(1 + (32 / sizeof(pcre_uchar))) : GET(cc, 0); + *opcode = cc[class_len - 1]; + if (*opcode >= OP_CRSTAR && *opcode <= OP_CRMINQUERY) + { + *opcode -= OP_CRSTAR - OP_STAR; + if (end != NULL) + *end = cc + class_len; + } + else + { + SLJIT_ASSERT(*opcode == OP_CRRANGE || *opcode == OP_CRMINRANGE); + *arg1 = GET2(cc, (class_len + IMM2_SIZE)); + *arg2 = GET2(cc, class_len); + + if (*arg2 == 0) + { + SLJIT_ASSERT(*arg1 != 0); + *opcode = (*opcode == OP_CRRANGE) ? OP_UPTO : OP_MINUPTO; + } + if (*arg1 == *arg2) + *opcode = OP_EXACT; + + if (end != NULL) + *end = cc + class_len + 2 * IMM2_SIZE; + } + return cc; + } + +if (*opcode == OP_UPTO || *opcode == OP_MINUPTO || *opcode == OP_EXACT || *opcode == OP_POSUPTO) + { + *arg1 = GET2(cc, 0); + cc += IMM2_SIZE; + } + +if (*type == 0) + { + *type = *cc; + if (end != NULL) + *end = next_opcode(common, cc); + cc++; + return cc; + } + +if (end != NULL) + { + *end = cc + 1; +#ifdef SUPPORT_UTF + if (common->utf && HAS_EXTRALEN(*cc)) *end += GET_EXTRALEN(*cc); +#endif + } +return cc; +} + +static pcre_uchar *compile_iterator_trypath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent) +{ +DEFINE_COMPILER; +backtrack_common *backtrack; +pcre_uchar opcode; +pcre_uchar type; +int arg1 = -1, arg2 = -1; +pcre_uchar* end; +jump_list *nomatch = NULL; +struct sljit_jump *jump = NULL; +struct sljit_label *label; + +PUSH_BACKTRACK(sizeof(iterator_backtrack), cc, NULL); + +cc = get_iterator_parameters(common, cc, &opcode, &type, &arg1, &arg2, &end); + +switch(opcode) + { + case OP_STAR: + case OP_PLUS: + case OP_UPTO: + case OP_CRRANGE: + if (type == OP_ANYNL || type == OP_EXTUNI) + { + if (opcode == OP_STAR || opcode == OP_UPTO) + { + allocate_stack(common, 2); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0); + } + else + { + allocate_stack(common, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); + } + if (opcode == OP_UPTO || opcode == OP_CRRANGE) + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, SLJIT_IMM, 0); + + label = LABEL(); + compile_char1_trypath(common, type, cc, &backtrack->topbacktracks); + if (opcode == OP_UPTO || opcode == OP_CRRANGE) + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); + OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); + if (opcode == OP_CRRANGE && arg2 > 0) + CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, arg2, label); + if (opcode == OP_UPTO || (opcode == OP_CRRANGE && arg1 > 0)) + jump = CMP(SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, arg1); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, TMP1, 0); + } + + allocate_stack(common, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + JUMPTO(SLJIT_JUMP, label); + if (jump != NULL) + JUMPHERE(jump); + } + else + { + if (opcode == OP_PLUS) + compile_char1_trypath(common, type, cc, &backtrack->topbacktracks); + allocate_stack(common, 2); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 1); + label = LABEL(); + compile_char1_trypath(common, type, cc, &nomatch); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + if (opcode <= OP_PLUS || (opcode == OP_CRRANGE && arg1 == 0)) + { + OP2(SLJIT_ADD, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 1); + JUMPTO(SLJIT_JUMP, label); + } + else + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); + OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0); + CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, arg1 + 1, label); + } + set_jumps(nomatch, LABEL()); + if (opcode == OP_CRRANGE) + add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_LESS, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, arg2 + 1)); + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + } + BACKTRACK_AS(iterator_backtrack)->trypath = LABEL(); + break; + + case OP_MINSTAR: + case OP_MINPLUS: + if (opcode == OP_MINPLUS) + compile_char1_trypath(common, type, cc, &backtrack->topbacktracks); + allocate_stack(common, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + BACKTRACK_AS(iterator_backtrack)->trypath = LABEL(); + break; + + case OP_MINUPTO: + case OP_CRMINRANGE: + allocate_stack(common, 2); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 1); + if (opcode == OP_CRMINRANGE) + add_jump(compiler, &backtrack->topbacktracks, JUMP(SLJIT_JUMP)); + BACKTRACK_AS(iterator_backtrack)->trypath = LABEL(); + break; + + case OP_QUERY: + case OP_MINQUERY: + allocate_stack(common, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + if (opcode == OP_QUERY) + compile_char1_trypath(common, type, cc, &backtrack->topbacktracks); + BACKTRACK_AS(iterator_backtrack)->trypath = LABEL(); + break; + + case OP_EXACT: + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, SLJIT_IMM, 1); + label = LABEL(); + compile_char1_trypath(common, type, cc, &backtrack->topbacktracks); + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); + OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, TMP1, 0); + CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, arg1 + 1, label); + break; + + case OP_POSSTAR: + case OP_POSPLUS: + case OP_POSUPTO: + if (opcode != OP_POSSTAR) + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, SLJIT_IMM, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, STR_PTR, 0); + label = LABEL(); + compile_char1_trypath(common, type, cc, &nomatch); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, STR_PTR, 0); + if (opcode != OP_POSUPTO) + { + if (opcode == OP_POSPLUS) + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, SLJIT_IMM, 2); + JUMPTO(SLJIT_JUMP, label); + } + else + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0); + OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, TMP1, 0); + CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, arg1 + 1, label); + } + set_jumps(nomatch, LABEL()); + if (opcode == OP_POSPLUS) + add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_LESS, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0, SLJIT_IMM, 2)); + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1); + break; + + case OP_POSQUERY: + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, STR_PTR, 0); + compile_char1_trypath(common, type, cc, &nomatch); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, STR_PTR, 0); + set_jumps(nomatch, LABEL()); + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1); + break; + + default: + SLJIT_ASSERT_STOP(); + break; + } + +decrease_call_count(common); +return end; +} + +static SLJIT_INLINE pcre_uchar *compile_fail_accept_trypath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent) +{ +DEFINE_COMPILER; +backtrack_common *backtrack; + +PUSH_BACKTRACK(sizeof(bracket_backtrack), cc, NULL); + +if (*cc == OP_FAIL) + { + add_jump(compiler, &backtrack->topbacktracks, JUMP(SLJIT_JUMP)); + return cc + 1; + } + +if (*cc == OP_ASSERT_ACCEPT || common->currententry != NULL) + { + /* No need to check notempty conditions. */ + if (common->acceptlabel == NULL) + add_jump(compiler, &common->accept, JUMP(SLJIT_JUMP)); + else + JUMPTO(SLJIT_JUMP, common->acceptlabel); + return cc + 1; + } + +if (common->acceptlabel == NULL) + add_jump(compiler, &common->accept, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0))); +else + CMPTO(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0), common->acceptlabel); +OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); +OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, notempty)); +add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0)); +OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, notempty_atstart)); +if (common->acceptlabel == NULL) + add_jump(compiler, &common->accept, CMP(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, 0)); +else + CMPTO(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, 0, common->acceptlabel); +OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); +if (common->acceptlabel == NULL) + add_jump(compiler, &common->accept, CMP(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_PTR, 0)); +else + CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_PTR, 0, common->acceptlabel); +add_jump(compiler, &backtrack->topbacktracks, JUMP(SLJIT_JUMP)); +return cc + 1; +} + +static SLJIT_INLINE pcre_uchar *compile_close_trypath(compiler_common *common, pcre_uchar *cc) +{ +DEFINE_COMPILER; +int offset = GET2(cc, 1); + +/* Data will be discarded anyway... */ +if (common->currententry != NULL) + return cc + 1 + IMM2_SIZE; + +OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR_PRIV(offset)); +offset <<= 1; +OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), STR_PTR, 0); +OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0); +return cc + 1 + IMM2_SIZE; +} + +static void compile_trypath(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, backtrack_common *parent) +{ +DEFINE_COMPILER; +backtrack_common *backtrack; + +while (cc < ccend) + { + switch(*cc) + { + case OP_SOD: + case OP_SOM: + case OP_NOT_WORD_BOUNDARY: + case OP_WORD_BOUNDARY: + case OP_NOT_DIGIT: + case OP_DIGIT: + case OP_NOT_WHITESPACE: + case OP_WHITESPACE: + case OP_NOT_WORDCHAR: + case OP_WORDCHAR: + case OP_ANY: + case OP_ALLANY: + case OP_ANYBYTE: + case OP_NOTPROP: + case OP_PROP: + case OP_ANYNL: + case OP_NOT_HSPACE: + case OP_HSPACE: + case OP_NOT_VSPACE: + case OP_VSPACE: + case OP_EXTUNI: + case OP_EODN: + case OP_EOD: + case OP_CIRC: + case OP_CIRCM: + case OP_DOLL: + case OP_DOLLM: + case OP_NOT: + case OP_NOTI: + case OP_REVERSE: + cc = compile_char1_trypath(common, *cc, cc + 1, parent->top != NULL ? &parent->top->nextbacktracks : &parent->topbacktracks); + break; + + case OP_SET_SOM: + PUSH_BACKTRACK_NOVALUE(sizeof(backtrack_common), cc); + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); + allocate_stack(common, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0), STR_PTR, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0); + cc++; + break; + + case OP_CHAR: + case OP_CHARI: + if (common->mode == JIT_COMPILE) + cc = compile_charn_trypath(common, cc, ccend, parent->top != NULL ? &parent->top->nextbacktracks : &parent->topbacktracks); + else + cc = compile_char1_trypath(common, *cc, cc + 1, parent->top != NULL ? &parent->top->nextbacktracks : &parent->topbacktracks); + break; + + case OP_STAR: + case OP_MINSTAR: + case OP_PLUS: + case OP_MINPLUS: + case OP_QUERY: + case OP_MINQUERY: + case OP_UPTO: + case OP_MINUPTO: + case OP_EXACT: + case OP_POSSTAR: + case OP_POSPLUS: + case OP_POSQUERY: + case OP_POSUPTO: + case OP_STARI: + case OP_MINSTARI: + case OP_PLUSI: + case OP_MINPLUSI: + case OP_QUERYI: + case OP_MINQUERYI: + case OP_UPTOI: + case OP_MINUPTOI: + case OP_EXACTI: + case OP_POSSTARI: + case OP_POSPLUSI: + case OP_POSQUERYI: + case OP_POSUPTOI: + case OP_NOTSTAR: + case OP_NOTMINSTAR: + case OP_NOTPLUS: + case OP_NOTMINPLUS: + case OP_NOTQUERY: + case OP_NOTMINQUERY: + case OP_NOTUPTO: + case OP_NOTMINUPTO: + case OP_NOTEXACT: + case OP_NOTPOSSTAR: + case OP_NOTPOSPLUS: + case OP_NOTPOSQUERY: + case OP_NOTPOSUPTO: + case OP_NOTSTARI: + case OP_NOTMINSTARI: + case OP_NOTPLUSI: + case OP_NOTMINPLUSI: + case OP_NOTQUERYI: + case OP_NOTMINQUERYI: + case OP_NOTUPTOI: + case OP_NOTMINUPTOI: + case OP_NOTEXACTI: + case OP_NOTPOSSTARI: + case OP_NOTPOSPLUSI: + case OP_NOTPOSQUERYI: + case OP_NOTPOSUPTOI: + case OP_TYPESTAR: + case OP_TYPEMINSTAR: + case OP_TYPEPLUS: + case OP_TYPEMINPLUS: + case OP_TYPEQUERY: + case OP_TYPEMINQUERY: + case OP_TYPEUPTO: + case OP_TYPEMINUPTO: + case OP_TYPEEXACT: + case OP_TYPEPOSSTAR: + case OP_TYPEPOSPLUS: + case OP_TYPEPOSQUERY: + case OP_TYPEPOSUPTO: + cc = compile_iterator_trypath(common, cc, parent); + break; + + case OP_CLASS: + case OP_NCLASS: + if (cc[1 + (32 / sizeof(pcre_uchar))] >= OP_CRSTAR && cc[1 + (32 / sizeof(pcre_uchar))] <= OP_CRMINRANGE) + cc = compile_iterator_trypath(common, cc, parent); + else + cc = compile_char1_trypath(common, *cc, cc + 1, parent->top != NULL ? &parent->top->nextbacktracks : &parent->topbacktracks); + break; + +#if defined SUPPORT_UTF || defined COMPILE_PCRE16 + case OP_XCLASS: + if (*(cc + GET(cc, 1)) >= OP_CRSTAR && *(cc + GET(cc, 1)) <= OP_CRMINRANGE) + cc = compile_iterator_trypath(common, cc, parent); + else + cc = compile_char1_trypath(common, *cc, cc + 1, parent->top != NULL ? &parent->top->nextbacktracks : &parent->topbacktracks); + break; +#endif + + case OP_REF: + case OP_REFI: + if (cc[1 + IMM2_SIZE] >= OP_CRSTAR && cc[1 + IMM2_SIZE] <= OP_CRMINRANGE) + cc = compile_ref_iterator_trypath(common, cc, parent); + else + cc = compile_ref_trypath(common, cc, parent->top != NULL ? &parent->top->nextbacktracks : &parent->topbacktracks, TRUE, FALSE); + break; + + case OP_RECURSE: + cc = compile_recurse_trypath(common, cc, parent); + break; + + case OP_ASSERT: + case OP_ASSERT_NOT: + case OP_ASSERTBACK: + case OP_ASSERTBACK_NOT: + PUSH_BACKTRACK_NOVALUE(sizeof(assert_backtrack), cc); + cc = compile_assert_trypath(common, cc, BACKTRACK_AS(assert_backtrack), FALSE); + break; + + case OP_BRAMINZERO: + PUSH_BACKTRACK_NOVALUE(sizeof(braminzero_backtrack), cc); + cc = bracketend(cc + 1); + if (*(cc - 1 - LINK_SIZE) != OP_KETRMIN) + { + allocate_stack(common, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + } + else + { + allocate_stack(common, 2); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), STR_PTR, 0); + } + BACKTRACK_AS(braminzero_backtrack)->trypath = LABEL(); + if (cc[1] > OP_ASSERTBACK_NOT) + decrease_call_count(common); + break; + + case OP_ONCE: + case OP_ONCE_NC: + case OP_BRA: + case OP_CBRA: + case OP_COND: + case OP_SBRA: + case OP_SCBRA: + case OP_SCOND: + cc = compile_bracket_trypath(common, cc, parent); + break; + + case OP_BRAZERO: + if (cc[1] > OP_ASSERTBACK_NOT) + cc = compile_bracket_trypath(common, cc, parent); + else + { + PUSH_BACKTRACK_NOVALUE(sizeof(assert_backtrack), cc); + cc = compile_assert_trypath(common, cc, BACKTRACK_AS(assert_backtrack), FALSE); + } + break; + + case OP_BRAPOS: + case OP_CBRAPOS: + case OP_SBRAPOS: + case OP_SCBRAPOS: + case OP_BRAPOSZERO: + cc = compile_bracketpos_trypath(common, cc, parent); + break; + + case OP_MARK: + PUSH_BACKTRACK_NOVALUE(sizeof(backtrack_common), cc); + SLJIT_ASSERT(common->mark_ptr != 0); + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr); + allocate_stack(common, 1); + OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0); + OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, (sljit_w)(cc + 2)); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, TMP2, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, mark_ptr), TMP2, 0); + cc += 1 + 2 + cc[1]; + break; + + case OP_COMMIT: + PUSH_BACKTRACK_NOVALUE(sizeof(backtrack_common), cc); + cc += 1; + break; + + case OP_FAIL: + case OP_ACCEPT: + case OP_ASSERT_ACCEPT: + cc = compile_fail_accept_trypath(common, cc, parent); + break; + + case OP_CLOSE: + cc = compile_close_trypath(common, cc); + break; + + case OP_SKIPZERO: + cc = bracketend(cc + 1); + break; + + default: + SLJIT_ASSERT_STOP(); + return; + } + if (cc == NULL) + return; + } +SLJIT_ASSERT(cc == ccend); +} + +#undef PUSH_BACKTRACK +#undef PUSH_BACKTRACK_NOVALUE +#undef BACKTRACK_AS + +#define COMPILE_BACKTRACKPATH(current) \ + do \ + { \ + compile_backtrackpath(common, (current)); \ + if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) \ + return; \ + } \ + while (0) + +#define CURRENT_AS(type) ((type *)current) + +static void compile_iterator_backtrackpath(compiler_common *common, struct backtrack_common *current) +{ +DEFINE_COMPILER; +pcre_uchar *cc = current->cc; +pcre_uchar opcode; +pcre_uchar type; +int arg1 = -1, arg2 = -1; +struct sljit_label *label = NULL; +struct sljit_jump *jump = NULL; +jump_list *jumplist = NULL; + +cc = get_iterator_parameters(common, cc, &opcode, &type, &arg1, &arg2, NULL); + +switch(opcode) + { + case OP_STAR: + case OP_PLUS: + case OP_UPTO: + case OP_CRRANGE: + if (type == OP_ANYNL || type == OP_EXTUNI) + { + set_jumps(current->topbacktracks, LABEL()); + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + free_stack(common, 1); + CMPTO(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(iterator_backtrack)->trypath); + } + else + { + if (opcode <= OP_PLUS || opcode == OP_UPTO) + arg2 = 0; + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); + jump = CMP(SLJIT_C_LESS_EQUAL, TMP1, 0, SLJIT_IMM, arg2 + 1); + OP2(SLJIT_SUB, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0, SLJIT_IMM, 1); + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + skip_char_back(common); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + JUMPTO(SLJIT_JUMP, CURRENT_AS(iterator_backtrack)->trypath); + if (opcode == OP_CRRANGE) + set_jumps(current->topbacktracks, LABEL()); + JUMPHERE(jump); + free_stack(common, 2); + if (opcode == OP_PLUS) + set_jumps(current->topbacktracks, LABEL()); + } + break; + + case OP_MINSTAR: + case OP_MINPLUS: + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + compile_char1_trypath(common, type, cc, &jumplist); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + JUMPTO(SLJIT_JUMP, CURRENT_AS(iterator_backtrack)->trypath); + set_jumps(jumplist, LABEL()); + free_stack(common, 1); + if (opcode == OP_MINPLUS) + set_jumps(current->topbacktracks, LABEL()); + break; + + case OP_MINUPTO: + case OP_CRMINRANGE: + if (opcode == OP_CRMINRANGE) + { + label = LABEL(); + set_jumps(current->topbacktracks, label); + } + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + compile_char1_trypath(common, type, cc, &jumplist); + + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0); + + if (opcode == OP_CRMINRANGE) + CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, arg2 + 1, label); + + if (opcode == OP_CRMINRANGE && arg1 == 0) + JUMPTO(SLJIT_JUMP, CURRENT_AS(iterator_backtrack)->trypath); + else + CMPTO(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, arg1 + 2, CURRENT_AS(iterator_backtrack)->trypath); + + set_jumps(jumplist, LABEL()); + free_stack(common, 2); + break; + + case OP_QUERY: + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); + CMPTO(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(iterator_backtrack)->trypath); + jump = JUMP(SLJIT_JUMP); + set_jumps(current->topbacktracks, LABEL()); + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); + JUMPTO(SLJIT_JUMP, CURRENT_AS(iterator_backtrack)->trypath); + JUMPHERE(jump); + free_stack(common, 1); + break; + + case OP_MINQUERY: + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); + jump = CMP(SLJIT_C_EQUAL, STR_PTR, 0, SLJIT_IMM, 0); + compile_char1_trypath(common, type, cc, &jumplist); + JUMPTO(SLJIT_JUMP, CURRENT_AS(iterator_backtrack)->trypath); + set_jumps(jumplist, LABEL()); + JUMPHERE(jump); + free_stack(common, 1); + break; + + case OP_EXACT: + case OP_POSPLUS: + set_jumps(current->topbacktracks, LABEL()); + break; + + case OP_POSSTAR: + case OP_POSQUERY: + case OP_POSUPTO: + break; + + default: + SLJIT_ASSERT_STOP(); + break; + } +} + +static void compile_ref_iterator_backtrackpath(compiler_common *common, struct backtrack_common *current) +{ +DEFINE_COMPILER; +pcre_uchar *cc = current->cc; +pcre_uchar type; + +type = cc[1 + IMM2_SIZE]; +if ((type & 0x1) == 0) + { + set_jumps(current->topbacktracks, LABEL()); + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + free_stack(common, 1); + CMPTO(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(iterator_backtrack)->trypath); + return; + } + +OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); +CMPTO(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(iterator_backtrack)->trypath); +set_jumps(current->topbacktracks, LABEL()); +free_stack(common, 2); +} + +static void compile_recurse_backtrackpath(compiler_common *common, struct backtrack_common *current) +{ +DEFINE_COMPILER; + +set_jumps(current->topbacktracks, LABEL()); + +if (common->has_set_som && common->mark_ptr != 0) + { + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); + free_stack(common, 2); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0), TMP2, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, TMP1, 0); + } +else if (common->has_set_som || common->mark_ptr != 0) + { + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + free_stack(common, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->has_set_som ? (int)(OVECTOR(0)) : common->mark_ptr, TMP2, 0); + } +} + +static void compile_assert_backtrackpath(compiler_common *common, struct backtrack_common *current) +{ +DEFINE_COMPILER; +pcre_uchar *cc = current->cc; +pcre_uchar bra = OP_BRA; +struct sljit_jump *brajump = NULL; + +SLJIT_ASSERT(*cc != OP_BRAMINZERO); +if (*cc == OP_BRAZERO) + { + bra = *cc; + cc++; + } + +if (bra == OP_BRAZERO) + { + SLJIT_ASSERT(current->topbacktracks == NULL); + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + } + +if (CURRENT_AS(assert_backtrack)->framesize < 0) + { + set_jumps(current->topbacktracks, LABEL()); + + if (bra == OP_BRAZERO) + { + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); + CMPTO(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(assert_backtrack)->trypath); + free_stack(common, 1); + } + return; + } + +if (bra == OP_BRAZERO) + { + if (*cc == OP_ASSERT_NOT || *cc == OP_ASSERTBACK_NOT) + { + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); + CMPTO(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(assert_backtrack)->trypath); + free_stack(common, 1); + return; + } + free_stack(common, 1); + brajump = CMP(SLJIT_C_EQUAL, STR_PTR, 0, SLJIT_IMM, 0); + } + +if (*cc == OP_ASSERT || *cc == OP_ASSERTBACK) + { + OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), CURRENT_AS(assert_backtrack)->localptr); + add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), CURRENT_AS(assert_backtrack)->localptr, SLJIT_MEM1(STACK_TOP), CURRENT_AS(assert_backtrack)->framesize * sizeof(sljit_w)); + + set_jumps(current->topbacktracks, LABEL()); + } +else + set_jumps(current->topbacktracks, LABEL()); + +if (bra == OP_BRAZERO) + { + /* We know there is enough place on the stack. */ + OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_w)); + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0); + JUMPTO(SLJIT_JUMP, CURRENT_AS(assert_backtrack)->trypath); + JUMPHERE(brajump); + } +} + +static void compile_bracket_backtrackpath(compiler_common *common, struct backtrack_common *current) +{ +DEFINE_COMPILER; +int opcode; +int offset = 0; +int localptr = CURRENT_AS(bracket_backtrack)->localptr; +int stacksize; +int count; +pcre_uchar *cc = current->cc; +pcre_uchar *ccbegin; +pcre_uchar *ccprev; +jump_list *jumplist = NULL; +jump_list *jumplistitem = NULL; +pcre_uchar bra = OP_BRA; +pcre_uchar ket; +assert_backtrack *assert; +BOOL has_alternatives; +struct sljit_jump *brazero = NULL; +struct sljit_jump *once = NULL; +struct sljit_jump *cond = NULL; +struct sljit_label *rminlabel = NULL; + +if (*cc == OP_BRAZERO || *cc == OP_BRAMINZERO) + { + bra = *cc; + cc++; + } + +opcode = *cc; +ccbegin = cc; +ket = *(bracketend(ccbegin) - 1 - LINK_SIZE); +cc += GET(cc, 1); +has_alternatives = *cc == OP_ALT; +if (SLJIT_UNLIKELY(opcode == OP_COND) || SLJIT_UNLIKELY(opcode == OP_SCOND)) + has_alternatives = (ccbegin[1 + LINK_SIZE] >= OP_ASSERT && ccbegin[1 + LINK_SIZE] <= OP_ASSERTBACK_NOT) || CURRENT_AS(bracket_backtrack)->u.condfailed != NULL; +if (opcode == OP_CBRA || opcode == OP_SCBRA) + offset = (GET2(ccbegin, 1 + LINK_SIZE)) << 1; +if (SLJIT_UNLIKELY(opcode == OP_COND) && (*cc == OP_KETRMAX || *cc == OP_KETRMIN)) + opcode = OP_SCOND; +if (SLJIT_UNLIKELY(opcode == OP_ONCE_NC)) + opcode = OP_ONCE; + +if (ket == OP_KETRMAX) + { + if (bra == OP_BRAZERO) + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + free_stack(common, 1); + brazero = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 0); + } + } +else if (ket == OP_KETRMIN) + { + if (bra != OP_BRAMINZERO) + { + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + if (opcode >= OP_SBRA || opcode == OP_ONCE) + { + /* Checking zero-length iteration. */ + if (opcode != OP_ONCE || CURRENT_AS(bracket_backtrack)->u.framesize < 0) + CMPTO(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, CURRENT_AS(bracket_backtrack)->recursivetrypath); + else + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + CMPTO(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_MEM1(TMP1), (CURRENT_AS(bracket_backtrack)->u.framesize + 1) * sizeof(sljit_w), CURRENT_AS(bracket_backtrack)->recursivetrypath); + } + if (opcode != OP_ONCE) + free_stack(common, 1); + } + else + JUMPTO(SLJIT_JUMP, CURRENT_AS(bracket_backtrack)->recursivetrypath); + } + rminlabel = LABEL(); + } +else if (bra == OP_BRAZERO) + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + free_stack(common, 1); + brazero = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 0); + } + +if (SLJIT_UNLIKELY(opcode == OP_ONCE)) + { + if (CURRENT_AS(bracket_backtrack)->u.framesize >= 0) + { + OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); + } + once = JUMP(SLJIT_JUMP); + } +else if (SLJIT_UNLIKELY(opcode == OP_COND) || SLJIT_UNLIKELY(opcode == OP_SCOND)) + { + if (has_alternatives) + { + /* Always exactly one alternative. */ + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + free_stack(common, 1); + + jumplistitem = sljit_alloc_memory(compiler, sizeof(jump_list)); + if (SLJIT_UNLIKELY(!jumplistitem)) + return; + jumplist = jumplistitem; + jumplistitem->next = NULL; + jumplistitem->jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 1); + } + } +else if (*cc == OP_ALT) + { + /* Build a jump list. Get the last successfully matched branch index. */ + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + free_stack(common, 1); + count = 1; + do + { + /* Append as the last item. */ + if (jumplist != NULL) + { + jumplistitem->next = sljit_alloc_memory(compiler, sizeof(jump_list)); + jumplistitem = jumplistitem->next; + } + else + { + jumplistitem = sljit_alloc_memory(compiler, sizeof(jump_list)); + jumplist = jumplistitem; + } + + if (SLJIT_UNLIKELY(!jumplistitem)) + return; + + jumplistitem->next = NULL; + jumplistitem->jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, count++); + cc += GET(cc, 1); + } + while (*cc == OP_ALT); + + cc = ccbegin + GET(ccbegin, 1); + } + +COMPILE_BACKTRACKPATH(current->top); +if (current->topbacktracks) + set_jumps(current->topbacktracks, LABEL()); + +if (SLJIT_UNLIKELY(opcode == OP_COND) || SLJIT_UNLIKELY(opcode == OP_SCOND)) + { + /* Conditional block always has at most one alternative. */ + if (ccbegin[1 + LINK_SIZE] >= OP_ASSERT && ccbegin[1 + LINK_SIZE] <= OP_ASSERTBACK_NOT) + { + SLJIT_ASSERT(has_alternatives); + assert = CURRENT_AS(bracket_backtrack)->u.assert; + if (assert->framesize >= 0 && (ccbegin[1 + LINK_SIZE] == OP_ASSERT || ccbegin[1 + LINK_SIZE] == OP_ASSERTBACK)) + { + OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), assert->localptr); + add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), assert->localptr, SLJIT_MEM1(STACK_TOP), assert->framesize * sizeof(sljit_w)); + } + cond = JUMP(SLJIT_JUMP); + set_jumps(CURRENT_AS(bracket_backtrack)->u.assert->condfailed, LABEL()); + } + else if (CURRENT_AS(bracket_backtrack)->u.condfailed != NULL) + { + SLJIT_ASSERT(has_alternatives); + cond = JUMP(SLJIT_JUMP); + set_jumps(CURRENT_AS(bracket_backtrack)->u.condfailed, LABEL()); + } + else + SLJIT_ASSERT(!has_alternatives); + } + +if (has_alternatives) + { + count = 1; + do + { + current->top = NULL; + current->topbacktracks = NULL; + current->nextbacktracks = NULL; + if (*cc == OP_ALT) + { + ccprev = cc + 1 + LINK_SIZE; + cc += GET(cc, 1); + if (opcode != OP_COND && opcode != OP_SCOND) + { + if (localptr != 0 && opcode != OP_ONCE) + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + else + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + } + compile_trypath(common, ccprev, cc, current); + if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) + return; + } + + /* Instructions after the current alternative is succesfully matched. */ + /* There is a similar code in compile_bracket_trypath. */ + if (opcode == OP_ONCE) + { + if (CURRENT_AS(bracket_backtrack)->u.framesize < 0) + { + OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + /* TMP2 which is set here used by OP_KETRMAX below. */ + if (ket == OP_KETRMAX) + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), 0); + else if (ket == OP_KETRMIN) + { + /* Move the STR_PTR to the localptr. */ + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_MEM1(STACK_TOP), 0); + } + } + else + { + OP2(SLJIT_ADD, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_IMM, (CURRENT_AS(bracket_backtrack)->u.framesize + 2) * sizeof(sljit_w)); + if (ket == OP_KETRMAX) + { + /* TMP2 which is set here used by OP_KETRMAX below. */ + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + } + } + } + + stacksize = 0; + if (opcode != OP_ONCE) + stacksize++; + if (ket != OP_KET || bra != OP_BRA) + stacksize++; + + if (stacksize > 0) { + if (opcode != OP_ONCE || CURRENT_AS(bracket_backtrack)->u.framesize >= 0) + allocate_stack(common, stacksize); + else + { + /* We know we have place at least for one item on the top of the stack. */ + SLJIT_ASSERT(stacksize == 1); + OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_w)); + } + } + + stacksize = 0; + if (ket != OP_KET || bra != OP_BRA) + { + if (ket != OP_KET) + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), STR_PTR, 0); + else + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), SLJIT_IMM, 0); + stacksize++; + } + + if (opcode != OP_ONCE) + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), SLJIT_IMM, count++); + + if (offset != 0) + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), STR_PTR, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 0), TMP1, 0); + } + + JUMPTO(SLJIT_JUMP, CURRENT_AS(bracket_backtrack)->alttrypath); + + if (opcode != OP_ONCE) + { + SLJIT_ASSERT(jumplist); + JUMPHERE(jumplist->jump); + jumplist = jumplist->next; + } + + COMPILE_BACKTRACKPATH(current->top); + if (current->topbacktracks) + set_jumps(current->topbacktracks, LABEL()); + SLJIT_ASSERT(!current->nextbacktracks); + } + while (*cc == OP_ALT); + SLJIT_ASSERT(!jumplist); + + if (cond != NULL) + { + SLJIT_ASSERT(opcode == OP_COND || opcode == OP_SCOND); + assert = CURRENT_AS(bracket_backtrack)->u.assert; + if ((ccbegin[1 + LINK_SIZE] == OP_ASSERT_NOT || ccbegin[1 + LINK_SIZE] == OP_ASSERTBACK_NOT) && assert->framesize >= 0) + + { + OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), assert->localptr); + add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), assert->localptr, SLJIT_MEM1(STACK_TOP), assert->framesize * sizeof(sljit_w)); + } + JUMPHERE(cond); + } + + /* Free the STR_PTR. */ + if (localptr == 0) + free_stack(common, 1); + } + +if (offset != 0) + { + /* Using both tmp register is better for instruction scheduling. */ + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), TMP2, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_MEM1(STACK_TOP), STACK(2)); + free_stack(common, 3); + } +else if (opcode == OP_SBRA || opcode == OP_SCOND) + { + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_MEM1(STACK_TOP), STACK(0)); + free_stack(common, 1); + } +else if (opcode == OP_ONCE) + { + cc = ccbegin + GET(ccbegin, 1); + if (CURRENT_AS(bracket_backtrack)->u.framesize >= 0) + { + /* Reset head and drop saved frame. */ + stacksize = (ket == OP_KETRMAX || ket == OP_KETRMIN || *cc == OP_ALT) ? 2 : 1; + free_stack(common, CURRENT_AS(bracket_backtrack)->u.framesize + stacksize); + } + else if (ket == OP_KETRMAX || (*cc == OP_ALT && ket != OP_KETRMIN)) + { + /* The STR_PTR must be released. */ + free_stack(common, 1); + } + + JUMPHERE(once); + /* Restore previous localptr */ + if (CURRENT_AS(bracket_backtrack)->u.framesize >= 0) + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, SLJIT_MEM1(STACK_TOP), CURRENT_AS(bracket_backtrack)->u.framesize * sizeof(sljit_w)); + else if (ket == OP_KETRMIN) + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); + /* See the comment below. */ + free_stack(common, 2); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), localptr, TMP1, 0); + } + } + +if (ket == OP_KETRMAX) + { + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + if (bra != OP_BRAZERO) + free_stack(common, 1); + CMPTO(SLJIT_C_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(bracket_backtrack)->recursivetrypath); + if (bra == OP_BRAZERO) + { + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); + JUMPTO(SLJIT_JUMP, CURRENT_AS(bracket_backtrack)->zerotrypath); + JUMPHERE(brazero); + free_stack(common, 1); + } + } +else if (ket == OP_KETRMIN) + { + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + + /* OP_ONCE removes everything in case of a backtrack, so we don't + need to explicitly release the STR_PTR. The extra release would + affect badly the free_stack(2) above. */ + if (opcode != OP_ONCE) + free_stack(common, 1); + CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 0, rminlabel); + if (opcode == OP_ONCE) + free_stack(common, bra == OP_BRAMINZERO ? 2 : 1); + else if (bra == OP_BRAMINZERO) + free_stack(common, 1); + } +else if (bra == OP_BRAZERO) + { + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + JUMPTO(SLJIT_JUMP, CURRENT_AS(bracket_backtrack)->zerotrypath); + JUMPHERE(brazero); + } +} + +static void compile_bracketpos_backtrackpath(compiler_common *common, struct backtrack_common *current) +{ +DEFINE_COMPILER; +int offset; +struct sljit_jump *jump; + +if (CURRENT_AS(bracketpos_backtrack)->framesize < 0) + { + if (*current->cc == OP_CBRAPOS || *current->cc == OP_SCBRAPOS) + { + offset = (GET2(current->cc, 1 + LINK_SIZE)) << 1; + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(1)); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), TMP2, 0); + } + set_jumps(current->topbacktracks, LABEL()); + free_stack(common, CURRENT_AS(bracketpos_backtrack)->stacksize); + return; + } + +OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), CURRENT_AS(bracketpos_backtrack)->localptr); +add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); + +if (current->topbacktracks) + { + jump = JUMP(SLJIT_JUMP); + set_jumps(current->topbacktracks, LABEL()); + /* Drop the stack frame. */ + free_stack(common, CURRENT_AS(bracketpos_backtrack)->stacksize); + JUMPHERE(jump); + } +OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), CURRENT_AS(bracketpos_backtrack)->localptr, SLJIT_MEM1(STACK_TOP), CURRENT_AS(bracketpos_backtrack)->framesize * sizeof(sljit_w)); +} + +static void compile_braminzero_backtrackpath(compiler_common *common, struct backtrack_common *current) +{ +assert_backtrack backtrack; + +current->top = NULL; +current->topbacktracks = NULL; +current->nextbacktracks = NULL; +if (current->cc[1] > OP_ASSERTBACK_NOT) + { + /* Manual call of compile_bracket_trypath and compile_bracket_backtrackpath. */ + compile_bracket_trypath(common, current->cc, current); + compile_bracket_backtrackpath(common, current->top); + } +else + { + memset(&backtrack, 0, sizeof(backtrack)); + backtrack.common.cc = current->cc; + backtrack.trypath = CURRENT_AS(braminzero_backtrack)->trypath; + /* Manual call of compile_assert_trypath. */ + compile_assert_trypath(common, current->cc, &backtrack, FALSE); + } +SLJIT_ASSERT(!current->nextbacktracks && !current->topbacktracks); +} + +static void compile_backtrackpath(compiler_common *common, struct backtrack_common *current) +{ +DEFINE_COMPILER; + +while (current) + { + if (current->nextbacktracks != NULL) + set_jumps(current->nextbacktracks, LABEL()); + switch(*current->cc) + { + case OP_SET_SOM: + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + free_stack(common, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0), TMP1, 0); + break; + + case OP_STAR: + case OP_MINSTAR: + case OP_PLUS: + case OP_MINPLUS: + case OP_QUERY: + case OP_MINQUERY: + case OP_UPTO: + case OP_MINUPTO: + case OP_EXACT: + case OP_POSSTAR: + case OP_POSPLUS: + case OP_POSQUERY: + case OP_POSUPTO: + case OP_STARI: + case OP_MINSTARI: + case OP_PLUSI: + case OP_MINPLUSI: + case OP_QUERYI: + case OP_MINQUERYI: + case OP_UPTOI: + case OP_MINUPTOI: + case OP_EXACTI: + case OP_POSSTARI: + case OP_POSPLUSI: + case OP_POSQUERYI: + case OP_POSUPTOI: + case OP_NOTSTAR: + case OP_NOTMINSTAR: + case OP_NOTPLUS: + case OP_NOTMINPLUS: + case OP_NOTQUERY: + case OP_NOTMINQUERY: + case OP_NOTUPTO: + case OP_NOTMINUPTO: + case OP_NOTEXACT: + case OP_NOTPOSSTAR: + case OP_NOTPOSPLUS: + case OP_NOTPOSQUERY: + case OP_NOTPOSUPTO: + case OP_NOTSTARI: + case OP_NOTMINSTARI: + case OP_NOTPLUSI: + case OP_NOTMINPLUSI: + case OP_NOTQUERYI: + case OP_NOTMINQUERYI: + case OP_NOTUPTOI: + case OP_NOTMINUPTOI: + case OP_NOTEXACTI: + case OP_NOTPOSSTARI: + case OP_NOTPOSPLUSI: + case OP_NOTPOSQUERYI: + case OP_NOTPOSUPTOI: + case OP_TYPESTAR: + case OP_TYPEMINSTAR: + case OP_TYPEPLUS: + case OP_TYPEMINPLUS: + case OP_TYPEQUERY: + case OP_TYPEMINQUERY: + case OP_TYPEUPTO: + case OP_TYPEMINUPTO: + case OP_TYPEEXACT: + case OP_TYPEPOSSTAR: + case OP_TYPEPOSPLUS: + case OP_TYPEPOSQUERY: + case OP_TYPEPOSUPTO: + case OP_CLASS: + case OP_NCLASS: +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + case OP_XCLASS: +#endif + compile_iterator_backtrackpath(common, current); + break; + + case OP_REF: + case OP_REFI: + compile_ref_iterator_backtrackpath(common, current); + break; + + case OP_RECURSE: + compile_recurse_backtrackpath(common, current); + break; + + case OP_ASSERT: + case OP_ASSERT_NOT: + case OP_ASSERTBACK: + case OP_ASSERTBACK_NOT: + compile_assert_backtrackpath(common, current); + break; + + case OP_ONCE: + case OP_ONCE_NC: + case OP_BRA: + case OP_CBRA: + case OP_COND: + case OP_SBRA: + case OP_SCBRA: + case OP_SCOND: + compile_bracket_backtrackpath(common, current); + break; + + case OP_BRAZERO: + if (current->cc[1] > OP_ASSERTBACK_NOT) + compile_bracket_backtrackpath(common, current); + else + compile_assert_backtrackpath(common, current); + break; + + case OP_BRAPOS: + case OP_CBRAPOS: + case OP_SBRAPOS: + case OP_SCBRAPOS: + case OP_BRAPOSZERO: + compile_bracketpos_backtrackpath(common, current); + break; + + case OP_BRAMINZERO: + compile_braminzero_backtrackpath(common, current); + break; + + case OP_MARK: + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + free_stack(common, 1); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, TMP1, 0); + break; + + case OP_COMMIT: + OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_NOMATCH); + if (common->leavelabel == NULL) + add_jump(compiler, &common->leave, JUMP(SLJIT_JUMP)); + else + JUMPTO(SLJIT_JUMP, common->leavelabel); + break; + + case OP_FAIL: + case OP_ACCEPT: + case OP_ASSERT_ACCEPT: + set_jumps(current->topbacktracks, LABEL()); + break; + + default: + SLJIT_ASSERT_STOP(); + break; + } + current = current->prev; + } +} + +static SLJIT_INLINE void compile_recurse(compiler_common *common) +{ +DEFINE_COMPILER; +pcre_uchar *cc = common->start + common->currententry->start; +pcre_uchar *ccbegin = cc + 1 + LINK_SIZE + (*cc == OP_BRA ? 0 : IMM2_SIZE); +pcre_uchar *ccend = bracketend(cc); +int localsize = get_localsize(common, ccbegin, ccend); +int framesize = get_framesize(common, cc, TRUE); +int alternativesize; +BOOL needsframe; +backtrack_common altbacktrack; +struct sljit_label *save_leavelabel = common->leavelabel; +jump_list *save_leave = common->leave; +struct sljit_jump *jump; + +SLJIT_ASSERT(*cc == OP_BRA || *cc == OP_CBRA || *cc == OP_CBRAPOS || *cc == OP_SCBRA || *cc == OP_SCBRAPOS); +needsframe = framesize >= 0; +if (!needsframe) + framesize = 0; +alternativesize = *(cc + GET(cc, 1)) == OP_ALT ? 1 : 0; + +SLJIT_ASSERT(common->currententry->entry == NULL && common->recursive_head != 0); +common->currententry->entry = LABEL(); +set_jumps(common->currententry->calls, common->currententry->entry); + +sljit_emit_fast_enter(compiler, TMP2, 0); +allocate_stack(common, localsize + framesize + alternativesize); +OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(localsize + framesize + alternativesize - 1), TMP2, 0); +copy_locals(common, ccbegin, ccend, TRUE, localsize + framesize + alternativesize, framesize + alternativesize); +OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->recursive_head, STACK_TOP, 0); +if (needsframe) + init_frame(common, cc, framesize + alternativesize - 1, alternativesize, TRUE); + +if (alternativesize > 0) + OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0); + +memset(&altbacktrack, 0, sizeof(backtrack_common)); +common->leavelabel = NULL; +common->acceptlabel = NULL; +common->leave = NULL; +common->accept = NULL; +altbacktrack.cc = ccbegin; +cc += GET(cc, 1); +while (1) + { + altbacktrack.top = NULL; + altbacktrack.topbacktracks = NULL; + + if (altbacktrack.cc != ccbegin) + OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0)); + + compile_trypath(common, altbacktrack.cc, cc, &altbacktrack); + if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) + { + common->leavelabel = save_leavelabel; + common->leave = save_leave; + return; + } + + add_jump(compiler, &common->accept, JUMP(SLJIT_JUMP)); + + compile_backtrackpath(common, altbacktrack.top); + if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) + { + common->leavelabel = save_leavelabel; + common->leave = save_leave; + return; + } + set_jumps(altbacktrack.topbacktracks, LABEL()); + + if (*cc != OP_ALT) + break; + + altbacktrack.cc = cc + 1 + LINK_SIZE; + cc += GET(cc, 1); + } +/* None of them matched. */ +if (common->leave != NULL) + set_jumps(common->leave, LABEL()); + +OP1(SLJIT_MOV, TMP3, 0, SLJIT_IMM, 0); +jump = JUMP(SLJIT_JUMP); + +set_jumps(common->accept, LABEL()); +OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->recursive_head); +if (needsframe) + { + OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, (framesize + alternativesize) * sizeof(sljit_w)); + add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL)); + OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, (framesize + alternativesize) * sizeof(sljit_w)); + } +OP1(SLJIT_MOV, TMP3, 0, SLJIT_IMM, 1); + +JUMPHERE(jump); +copy_locals(common, ccbegin, ccend, FALSE, localsize + framesize + alternativesize, framesize + alternativesize); +free_stack(common, localsize + framesize + alternativesize); +OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), sizeof(sljit_w)); +OP1(SLJIT_MOV, TMP1, 0, TMP3, 0); +OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->recursive_head, TMP2, 0); +sljit_emit_fast_return(compiler, SLJIT_MEM1(STACK_TOP), 0); + +common->leavelabel = save_leavelabel; +common->leave = save_leave; +} + +#undef COMPILE_BACKTRACKPATH +#undef CURRENT_AS + +void +PRIV(jit_compile)(const REAL_PCRE *re, PUBL(extra) *extra, int mode) +{ +struct sljit_compiler *compiler; +backtrack_common rootbacktrack; +compiler_common common_data; +compiler_common *common = &common_data; +const pcre_uint8 *tables = re->tables; +pcre_study_data *study; +int localsize; +pcre_uchar *ccend; +executable_functions *functions; +void *executable_func; +sljit_uw executable_size; +struct sljit_label *mainloop = NULL; +struct sljit_label *empty_match_found; +struct sljit_label *empty_match_backtrack; +struct sljit_jump *jump; +struct sljit_jump *reqbyte_notfound = NULL; +struct sljit_jump *empty_match; + +SLJIT_ASSERT((extra->flags & PCRE_EXTRA_STUDY_DATA) != 0); +study = extra->study_data; + +if (!tables) + tables = PRIV(default_tables); + +memset(&rootbacktrack, 0, sizeof(backtrack_common)); +memset(common, 0, sizeof(compiler_common)); +rootbacktrack.cc = (pcre_uchar *)re + re->name_table_offset + re->name_count * re->name_entry_size; + +common->start = rootbacktrack.cc; +common->fcc = tables + fcc_offset; +common->lcc = (sljit_w)(tables + lcc_offset); +common->mode = mode; +common->nltype = NLTYPE_FIXED; +switch(re->options & PCRE_NEWLINE_BITS) + { + case 0: + /* Compile-time default */ + switch (NEWLINE) + { + case -1: common->newline = (CHAR_CR << 8) | CHAR_NL; common->nltype = NLTYPE_ANY; break; + case -2: common->newline = (CHAR_CR << 8) | CHAR_NL; common->nltype = NLTYPE_ANYCRLF; break; + default: common->newline = NEWLINE; break; + } + break; + case PCRE_NEWLINE_CR: common->newline = CHAR_CR; break; + case PCRE_NEWLINE_LF: common->newline = CHAR_NL; break; + case PCRE_NEWLINE_CR+ + PCRE_NEWLINE_LF: common->newline = (CHAR_CR << 8) | CHAR_NL; break; + case PCRE_NEWLINE_ANY: common->newline = (CHAR_CR << 8) | CHAR_NL; common->nltype = NLTYPE_ANY; break; + case PCRE_NEWLINE_ANYCRLF: common->newline = (CHAR_CR << 8) | CHAR_NL; common->nltype = NLTYPE_ANYCRLF; break; + default: return; + } +if ((re->options & PCRE_BSR_ANYCRLF) != 0) + common->bsr_nltype = NLTYPE_ANYCRLF; +else if ((re->options & PCRE_BSR_UNICODE) != 0) + common->bsr_nltype = NLTYPE_ANY; +else + { +#ifdef BSR_ANYCRLF + common->bsr_nltype = NLTYPE_ANYCRLF; +#else + common->bsr_nltype = NLTYPE_ANY; +#endif + } +common->endonly = (re->options & PCRE_DOLLAR_ENDONLY) != 0; +common->ctypes = (sljit_w)(tables + ctypes_offset); +common->name_table = (sljit_w)((pcre_uchar *)re + re->name_table_offset); +common->name_count = re->name_count; +common->name_entry_size = re->name_entry_size; +common->jscript_compat = (re->options & PCRE_JAVASCRIPT_COMPAT) != 0; +#ifdef SUPPORT_UTF +/* PCRE_UTF16 has the same value as PCRE_UTF8. */ +common->utf = (re->options & PCRE_UTF8) != 0; +#ifdef SUPPORT_UCP +common->use_ucp = (re->options & PCRE_UCP) != 0; +#endif +#endif /* SUPPORT_UTF */ +ccend = bracketend(rootbacktrack.cc); + +/* Calculate the local space size on the stack. */ +common->ovector_start = CALL_LIMIT + sizeof(sljit_w); + +SLJIT_ASSERT(*rootbacktrack.cc == OP_BRA && ccend[-(1 + LINK_SIZE)] == OP_KET); +localsize = get_localspace(common, rootbacktrack.cc, ccend); +if (localsize < 0) + return; + +/* Checking flags and updating ovector_start. */ +if (mode == JIT_COMPILE && (re->flags & PCRE_REQCHSET) != 0 && (re->options & PCRE_NO_START_OPTIMIZE) == 0) + { + common->req_char_ptr = common->ovector_start; + common->ovector_start += sizeof(sljit_w); + } +if (mode != JIT_COMPILE) + { + common->start_used_ptr = common->ovector_start; + common->ovector_start += sizeof(sljit_w); + if (mode == JIT_PARTIAL_SOFT_COMPILE) + { + common->hit_start = common->ovector_start; + common->ovector_start += sizeof(sljit_w); + } + } +if ((re->options & PCRE_FIRSTLINE) != 0) + { + common->first_line_end = common->ovector_start; + common->ovector_start += sizeof(sljit_w); + } + +/* Aligning ovector to even number of sljit words. */ +if ((common->ovector_start & sizeof(sljit_w)) != 0) + common->ovector_start += sizeof(sljit_w); + +SLJIT_ASSERT(!(common->req_char_ptr != 0 && common->start_used_ptr != 0)); +common->cbraptr = OVECTOR_START + (re->top_bracket + 1) * 2 * sizeof(sljit_w); +localsize += common->cbraptr + (re->top_bracket + 1) * sizeof(sljit_w); +if (localsize > SLJIT_MAX_LOCAL_SIZE) + return; +common->localptrs = (int *)SLJIT_MALLOC((ccend - rootbacktrack.cc) * sizeof(int)); +if (!common->localptrs) + return; +memset(common->localptrs, 0, (ccend - rootbacktrack.cc) * sizeof(int)); +set_localptrs(common, common->cbraptr + (re->top_bracket + 1) * sizeof(sljit_w), ccend); + +compiler = sljit_create_compiler(); +if (!compiler) + { + SLJIT_FREE(common->localptrs); + return; + } +common->compiler = compiler; + +/* Main pcre_jit_exec entry. */ +sljit_emit_enter(compiler, 1, 5, 5, localsize); + +/* Register init. */ +reset_ovector(common, (re->top_bracket + 1) * 2); +if (common->req_char_ptr != 0) + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->req_char_ptr, SLJIT_TEMPORARY_REG1, 0); + +OP1(SLJIT_MOV, ARGUMENTS, 0, SLJIT_SAVED_REG1, 0); +OP1(SLJIT_MOV, TMP1, 0, SLJIT_SAVED_REG1, 0); +OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); +OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, end)); +OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, stack)); +OP1(SLJIT_MOV_SI, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, calllimit)); +OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(struct sljit_stack, base)); +OP1(SLJIT_MOV, STACK_LIMIT, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(struct sljit_stack, limit)); +OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), CALL_LIMIT, TMP1, 0); + +if (mode == JIT_PARTIAL_SOFT_COMPILE) + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, 0); + +/* Main part of the matching */ +if ((re->options & PCRE_ANCHORED) == 0) + { + mainloop = mainloop_entry(common, (re->flags & PCRE_HASCRORLF) != 0, (re->options & PCRE_FIRSTLINE) != 0); + /* Forward search if possible. */ + if ((re->options & PCRE_NO_START_OPTIMIZE) == 0) + { + if ((re->flags & PCRE_FIRSTSET) != 0) + fast_forward_first_char(common, (pcre_uchar)re->first_char, (re->flags & PCRE_FCH_CASELESS) != 0, (re->options & PCRE_FIRSTLINE) != 0); + else if ((re->flags & PCRE_STARTLINE) != 0) + fast_forward_newline(common, (re->options & PCRE_FIRSTLINE) != 0); + else if ((re->flags & PCRE_STARTLINE) == 0 && study != NULL && (study->flags & PCRE_STUDY_MAPPED) != 0) + fast_forward_start_bits(common, (sljit_uw)study->start_bits, (re->options & PCRE_FIRSTLINE) != 0); + } + } +if (common->req_char_ptr != 0) + reqbyte_notfound = search_requested_char(common, (pcre_uchar)re->req_char, (re->flags & PCRE_RCH_CASELESS) != 0, (re->flags & PCRE_FIRSTSET) != 0); + +/* Store the current STR_PTR in OVECTOR(0). */ +OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0), STR_PTR, 0); +/* Copy the limit of allowed recursions. */ +OP1(SLJIT_MOV, CALL_COUNT, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), CALL_LIMIT); +if (common->mark_ptr != 0) + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, SLJIT_IMM, 0); +/* Copy the beginning of the string. */ +if (mode == JIT_PARTIAL_SOFT_COMPILE) + { + jump = CMP(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, 0); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); + JUMPHERE(jump); + } +else if (mode == JIT_PARTIAL_HARD_COMPILE) + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0); + +compile_trypath(common, rootbacktrack.cc, ccend, &rootbacktrack); +if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) + { + sljit_free_compiler(compiler); + SLJIT_FREE(common->localptrs); + return; + } + +empty_match = CMP(SLJIT_C_EQUAL, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); +empty_match_found = LABEL(); + +common->acceptlabel = LABEL(); +if (common->accept != NULL) + set_jumps(common->accept, common->acceptlabel); + +/* This means we have a match. Update the ovector. */ +copy_ovector(common, re->top_bracket + 1); +common->leavelabel = LABEL(); +if (common->leave != NULL) + set_jumps(common->leave, common->leavelabel); +sljit_emit_return(compiler, SLJIT_MOV, SLJIT_RETURN_REG, 0); + +if (mode != JIT_COMPILE) + { + common->partialmatchlabel = LABEL(); + set_jumps(common->partialmatch, common->partialmatchlabel); + return_with_partial_match(common, common->leavelabel); + } + +empty_match_backtrack = LABEL(); +compile_backtrackpath(common, rootbacktrack.top); +if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) + { + sljit_free_compiler(compiler); + SLJIT_FREE(common->localptrs); + return; + } + +SLJIT_ASSERT(rootbacktrack.prev == NULL); + +if (mode == JIT_PARTIAL_SOFT_COMPILE) + { + /* Update hit_start only in the first time. */ + jump = CMP(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1); + OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, -1); + OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, TMP1, 0); + JUMPHERE(jump); + } + +/* Check we have remaining characters. */ +OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(0)); + +if ((re->options & PCRE_ANCHORED) == 0) + { + if ((re->options & PCRE_FIRSTLINE) == 0) + { + if (mode == JIT_COMPILE && study != NULL && study->minlength > 1 && (re->options & PCRE_NO_START_OPTIMIZE) == 0) + { + OP2(SLJIT_ADD, TMP1, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(study->minlength + 1)); + CMPTO(SLJIT_C_LESS_EQUAL, TMP1, 0, STR_END, 0, mainloop); + } + else + CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop); + } + else + { + SLJIT_ASSERT(common->first_line_end != 0); + if (mode == JIT_COMPILE && study != NULL && study->minlength > 1 && (re->options & PCRE_NO_START_OPTIMIZE) == 0) + { + OP2(SLJIT_ADD, TMP1, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(study->minlength + 1)); + OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, STR_END, 0); + COND_VALUE(SLJIT_MOV, TMP2, 0, SLJIT_C_GREATER); + OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end); + COND_VALUE(SLJIT_OR | SLJIT_SET_E, TMP2, 0, SLJIT_C_GREATER_EQUAL); + JUMPTO(SLJIT_C_ZERO, mainloop); + } + else + CMPTO(SLJIT_C_LESS, STR_PTR, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, mainloop); + } + } + +/* No more remaining characters. */ +if (reqbyte_notfound != NULL) + JUMPHERE(reqbyte_notfound); + +if (mode == JIT_PARTIAL_SOFT_COMPILE) + CMPTO(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, 0, common->partialmatchlabel); + +OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_NOMATCH); +JUMPTO(SLJIT_JUMP, common->leavelabel); + +flush_stubs(common); + +JUMPHERE(empty_match); +OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); +OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, notempty)); +CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0, empty_match_backtrack); +OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, notempty_atstart)); +CMPTO(SLJIT_C_EQUAL, TMP2, 0, SLJIT_IMM, 0, empty_match_found); +OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str)); +CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, STR_PTR, 0, empty_match_found); +JUMPTO(SLJIT_JUMP, empty_match_backtrack); + +common->currententry = common->entries; +while (common->currententry != NULL) + { + /* Might add new entries. */ + compile_recurse(common); + if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) + { + sljit_free_compiler(compiler); + SLJIT_FREE(common->localptrs); + return; + } + flush_stubs(common); + common->currententry = common->currententry->next; + } + +/* Allocating stack, returns with PCRE_ERROR_JIT_STACKLIMIT if fails. */ +/* This is a (really) rare case. */ +set_jumps(common->stackalloc, LABEL()); +/* RETURN_ADDR is not a saved register. */ +sljit_emit_fast_enter(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); +OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, TMP2, 0); +OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); +OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, stack)); +OP1(SLJIT_MOV, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(struct sljit_stack, top), STACK_TOP, 0); +OP2(SLJIT_ADD, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(struct sljit_stack, limit), SLJIT_IMM, STACK_GROWTH_RATE); + +sljit_emit_ijump(compiler, SLJIT_CALL2, SLJIT_IMM, SLJIT_FUNC_OFFSET(sljit_stack_resize)); +jump = CMP(SLJIT_C_NOT_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 0); +OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0); +OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, stack)); +OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(struct sljit_stack, top)); +OP1(SLJIT_MOV, STACK_LIMIT, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(struct sljit_stack, limit)); +OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1); +sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0); + +/* Allocation failed. */ +JUMPHERE(jump); +/* We break the return address cache here, but this is a really rare case. */ +OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_JIT_STACKLIMIT); +JUMPTO(SLJIT_JUMP, common->leavelabel); + +/* Call limit reached. */ +set_jumps(common->calllimit, LABEL()); +OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_MATCHLIMIT); +JUMPTO(SLJIT_JUMP, common->leavelabel); + +if (common->revertframes != NULL) + { + set_jumps(common->revertframes, LABEL()); + do_revertframes(common); + } +if (common->wordboundary != NULL) + { + set_jumps(common->wordboundary, LABEL()); + check_wordboundary(common); + } +if (common->anynewline != NULL) + { + set_jumps(common->anynewline, LABEL()); + check_anynewline(common); + } +if (common->hspace != NULL) + { + set_jumps(common->hspace, LABEL()); + check_hspace(common); + } +if (common->vspace != NULL) + { + set_jumps(common->vspace, LABEL()); + check_vspace(common); + } +if (common->casefulcmp != NULL) + { + set_jumps(common->casefulcmp, LABEL()); + do_casefulcmp(common); + } +if (common->caselesscmp != NULL) + { + set_jumps(common->caselesscmp, LABEL()); + do_caselesscmp(common); + } +#ifdef SUPPORT_UTF +if (common->utfreadchar != NULL) + { + set_jumps(common->utfreadchar, LABEL()); + do_utfreadchar(common); + } +#ifdef COMPILE_PCRE8 +if (common->utfreadtype8 != NULL) + { + set_jumps(common->utfreadtype8, LABEL()); + do_utfreadtype8(common); + } +#endif +#endif /* COMPILE_PCRE8 */ +#ifdef SUPPORT_UCP +if (common->getucd != NULL) + { + set_jumps(common->getucd, LABEL()); + do_getucd(common); + } +#endif + +SLJIT_FREE(common->localptrs); +executable_func = sljit_generate_code(compiler); +executable_size = sljit_get_generated_code_size(compiler); +sljit_free_compiler(compiler); +if (executable_func == NULL) + return; + +/* Reuse the function descriptor if possible. */ +if ((extra->flags & PCRE_EXTRA_EXECUTABLE_JIT) != 0 && extra->executable_jit != NULL) + functions = (executable_functions *)extra->executable_jit; +else + { + functions = SLJIT_MALLOC(sizeof(executable_functions)); + if (functions == NULL) + { + /* This case is highly unlikely since we just recently + freed a lot of memory. Although not impossible. */ + sljit_free_code(executable_func); + return; + } + memset(functions, 0, sizeof(executable_functions)); + extra->executable_jit = functions; + extra->flags |= PCRE_EXTRA_EXECUTABLE_JIT; + } + +functions->executable_funcs[mode] = executable_func; +functions->executable_sizes[mode] = executable_size; +} + +static int jit_machine_stack_exec(jit_arguments *arguments, void* executable_func) +{ +union { + void* executable_func; + jit_function call_executable_func; +} convert_executable_func; +pcre_uint8 local_area[LOCAL_SPACE_SIZE]; +struct sljit_stack local_stack; + +local_stack.top = (sljit_w)&local_area; +local_stack.base = local_stack.top; +local_stack.limit = local_stack.base + LOCAL_SPACE_SIZE; +local_stack.max_limit = local_stack.limit; +arguments->stack = &local_stack; +convert_executable_func.executable_func = executable_func; +return convert_executable_func.call_executable_func(arguments); +} + +int +PRIV(jit_exec)(const REAL_PCRE *re, const PUBL(extra) *extra_data, const pcre_uchar *subject, + int length, int start_offset, int options, int *offsets, int offsetcount) +{ +executable_functions *functions = (executable_functions *)extra_data->executable_jit; +union { + void* executable_func; + jit_function call_executable_func; +} convert_executable_func; +jit_arguments arguments; +int maxoffsetcount; +int retval; +int mode = JIT_COMPILE; + +if ((options & PCRE_PARTIAL_HARD) != 0) + mode = JIT_PARTIAL_HARD_COMPILE; +else if ((options & PCRE_PARTIAL_SOFT) != 0) + mode = JIT_PARTIAL_SOFT_COMPILE; + +if (functions->executable_funcs[mode] == NULL) + return PCRE_ERROR_NULL; + +/* Sanity checks should be handled by pcre_exec. */ +arguments.stack = NULL; +arguments.str = subject + start_offset; +arguments.begin = subject; +arguments.end = subject + length; +arguments.mark_ptr = NULL; +/* JIT decreases this value less frequently than the interpreter. */ +arguments.calllimit = ((extra_data->flags & PCRE_EXTRA_MATCH_LIMIT) == 0) ? MATCH_LIMIT : extra_data->match_limit; +arguments.notbol = (options & PCRE_NOTBOL) != 0; +arguments.noteol = (options & PCRE_NOTEOL) != 0; +arguments.notempty = (options & PCRE_NOTEMPTY) != 0; +arguments.notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0; +arguments.offsets = offsets; + +/* pcre_exec() rounds offsetcount to a multiple of 3, and then uses only 2/3 of +the output vector for storing captured strings, with the remainder used as +workspace. We don't need the workspace here. For compatibility, we limit the +number of captured strings in the same way as pcre_exec(), so that the user +gets the same result with and without JIT. */ + +if (offsetcount != 2) + offsetcount = ((offsetcount - (offsetcount % 3)) * 2) / 3; +maxoffsetcount = (re->top_bracket + 1) * 2; +if (offsetcount > maxoffsetcount) + offsetcount = maxoffsetcount; +arguments.offsetcount = offsetcount; + +if (functions->callback) + arguments.stack = (struct sljit_stack *)functions->callback(functions->userdata); +else + arguments.stack = (struct sljit_stack *)functions->userdata; + +if (arguments.stack == NULL) + retval = jit_machine_stack_exec(&arguments, functions->executable_funcs[mode]); +else + { + convert_executable_func.executable_func = functions->executable_funcs[mode]; + retval = convert_executable_func.call_executable_func(&arguments); + } + +if (retval * 2 > offsetcount) + retval = 0; +if ((extra_data->flags & PCRE_EXTRA_MARK) != 0) + *(extra_data->mark) = arguments.mark_ptr; + +return retval; +} + +void +PRIV(jit_free)(void *executable_funcs) +{ +int i; +executable_functions *functions = (executable_functions *)executable_funcs; +for (i = 0; i < JIT_NUMBER_OF_COMPILE_MODES; i++) + { + if (functions->executable_funcs[i] != NULL) + sljit_free_code(functions->executable_funcs[i]); + } +SLJIT_FREE(functions); +} + +int +PRIV(jit_get_size)(void *executable_funcs) +{ +int i; +sljit_uw size = 0; +sljit_uw *executable_sizes = ((executable_functions *)executable_funcs)->executable_sizes; +for (i = 0; i < JIT_NUMBER_OF_COMPILE_MODES; i++) + size += executable_sizes[i]; +return (int)size; +} + +const char* +PRIV(jit_get_target)(void) +{ +return sljit_get_platform_name(); +} + +#ifdef COMPILE_PCRE8 +PCRE_EXP_DECL pcre_jit_stack * +pcre_jit_stack_alloc(int startsize, int maxsize) +#else +PCRE_EXP_DECL pcre16_jit_stack * +pcre16_jit_stack_alloc(int startsize, int maxsize) +#endif +{ +if (startsize < 1 || maxsize < 1) + return NULL; +if (startsize > maxsize) + startsize = maxsize; +startsize = (startsize + STACK_GROWTH_RATE - 1) & ~(STACK_GROWTH_RATE - 1); +maxsize = (maxsize + STACK_GROWTH_RATE - 1) & ~(STACK_GROWTH_RATE - 1); +return (PUBL(jit_stack)*)sljit_allocate_stack(startsize, maxsize); +} + +#ifdef COMPILE_PCRE8 +PCRE_EXP_DECL void +pcre_jit_stack_free(pcre_jit_stack *stack) +#else +PCRE_EXP_DECL void +pcre16_jit_stack_free(pcre16_jit_stack *stack) +#endif +{ +sljit_free_stack((struct sljit_stack *)stack); +} + +#ifdef COMPILE_PCRE8 +PCRE_EXP_DECL void +pcre_assign_jit_stack(pcre_extra *extra, pcre_jit_callback callback, void *userdata) +#else +PCRE_EXP_DECL void +pcre16_assign_jit_stack(pcre16_extra *extra, pcre16_jit_callback callback, void *userdata) +#endif +{ +executable_functions *functions; +if (extra != NULL && + (extra->flags & PCRE_EXTRA_EXECUTABLE_JIT) != 0 && + extra->executable_jit != NULL) + { + functions = (executable_functions *)extra->executable_jit; + functions->callback = callback; + functions->userdata = userdata; + } +} + +#else /* SUPPORT_JIT */ + +/* These are dummy functions to avoid linking errors when JIT support is not +being compiled. */ + +#ifdef COMPILE_PCRE8 +PCRE_EXP_DECL pcre_jit_stack * +pcre_jit_stack_alloc(int startsize, int maxsize) +#else +PCRE_EXP_DECL pcre16_jit_stack * +pcre16_jit_stack_alloc(int startsize, int maxsize) +#endif +{ +(void)startsize; +(void)maxsize; +return NULL; +} + +#ifdef COMPILE_PCRE8 +PCRE_EXP_DECL void +pcre_jit_stack_free(pcre_jit_stack *stack) +#else +PCRE_EXP_DECL void +pcre16_jit_stack_free(pcre16_jit_stack *stack) +#endif +{ +(void)stack; +} + +#ifdef COMPILE_PCRE8 +PCRE_EXP_DECL void +pcre_assign_jit_stack(pcre_extra *extra, pcre_jit_callback callback, void *userdata) +#else +PCRE_EXP_DECL void +pcre16_assign_jit_stack(pcre16_extra *extra, pcre16_jit_callback callback, void *userdata) +#endif +{ +(void)extra; +(void)callback; +(void)userdata; } #endif diff --git a/pcre/pcre_jit_test.c b/pcre/pcre_jit_test.c new file mode 100644 --- /dev/null +++ b/pcre/pcre_jit_test.c @@ -0,0 +1,1279 @@ +/************************************************* +* Perl-Compatible Regular Expressions * +*************************************************/ + +/* PCRE is a library of functions to support regular expressions whose syntax +and semantics are as close as possible to those of the Perl 5 language. + + Main Library written by Philip Hazel + Copyright (c) 1997-2012 University of Cambridge + + This JIT compiler regression test program was written by Zoltan Herczeg + Copyright (c) 2010-2012 + +----------------------------------------------------------------------------- +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. + + * Neither the name of the University of Cambridge 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 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. +----------------------------------------------------------------------------- +*/ + +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + +#include +#include +#include "pcre.h" + +#define PCRE_BUG 0x80000000 + +/* + Letter characters: + \xe6\x92\xad = 0x64ad = 25773 (kanji) + Non-letter characters: + \xc2\xa1 = 0xa1 = (Inverted Exclamation Mark) + \xf3\xa9\xb7\x80 = 0xe9dc0 = 957888 + \xed\xa0\x80 = 55296 = 0xd800 (Invalid UTF character) + \xed\xb0\x80 = 56320 = 0xdc00 (Invalid UTF character) + Newlines: + \xc2\x85 = 0x85 = 133 (NExt Line = NEL) + \xe2\x80\xa8 = 0x2028 = 8232 (Line Separator) + Othercase pairs: + \xc3\xa9 = 0xe9 = 233 (e') + \xc3\x89 = 0xc9 = 201 (E') + \xc3\xa1 = 0xe1 = 225 (a') + \xc3\x81 = 0xc1 = 193 (A') + \xc8\xba = 0x23a = 570 + \xe2\xb1\xa5 = 0x2c65 = 11365 + \xe1\xbd\xb8 = 0x1f78 = 8056 + \xe1\xbf\xb8 = 0x1ff8 = 8184 + \xf0\x90\x90\x80 = 0x10400 = 66560 + \xf0\x90\x90\xa8 = 0x10428 = 66600 + Mark property: + \xcc\x8d = 0x30d = 781 + Special: + \xdf\xbf = 0x7ff = 2047 (highest 2 byte character) + \xe0\xa0\x80 = 0x800 = 2048 (lowest 2 byte character) + \xef\xbf\xbf = 0xffff = 65535 (highest 3 byte character) + \xf0\x90\x80\x80 = 0x10000 = 65536 (lowest 4 byte character) + \xf4\x8f\xbf\xbf = 0x10ffff = 1114111 (highest allowed utf character) +*/ + +static int regression_tests(void); + +int main(void) +{ + int jit = 0; +#ifdef SUPPORT_PCRE8 + pcre_config(PCRE_CONFIG_JIT, &jit); +#else + pcre16_config(PCRE_CONFIG_JIT, &jit); +#endif + if (!jit) { + printf("JIT must be enabled to run pcre_jit_test\n"); + return 1; + } + return regression_tests(); +} + +/* --------------------------------------------------------------------------------------- */ + +#if !(defined SUPPORT_PCRE8) && !(defined SUPPORT_PCRE16) +#error SUPPORT_PCRE8 or SUPPORT_PCRE16 must be defined +#endif + +#define MUA (PCRE_MULTILINE | PCRE_UTF8 | PCRE_NEWLINE_ANYCRLF) +#define MUAP (PCRE_MULTILINE | PCRE_UTF8 | PCRE_NEWLINE_ANYCRLF | PCRE_UCP) +#define CMUA (PCRE_CASELESS | PCRE_MULTILINE | PCRE_UTF8 | PCRE_NEWLINE_ANYCRLF) +#define CMUAP (PCRE_CASELESS | PCRE_MULTILINE | PCRE_UTF8 | PCRE_NEWLINE_ANYCRLF | PCRE_UCP) +#define MA (PCRE_MULTILINE | PCRE_NEWLINE_ANYCRLF) +#define MAP (PCRE_MULTILINE | PCRE_NEWLINE_ANYCRLF | PCRE_UCP) +#define CMA (PCRE_CASELESS | PCRE_MULTILINE | PCRE_NEWLINE_ANYCRLF) + +#define OFFSET_MASK 0x00ffff +#define F_NO8 0x010000 +#define F_NO16 0x020000 +#define F_NOMATCH 0x040000 +#define F_DIFF 0x080000 +#define F_FORCECONV 0x100000 +#define F_PROPERTY 0x200000 + +struct regression_test_case { + int flags; + int start_offset; + const char *pattern; + const char *input; +}; + +static struct regression_test_case regression_test_cases[] = { + /* Constant strings. */ + { MUA, 0, "AbC", "AbAbC" }, + { MUA, 0, "ACCEPT", "AACACCACCEACCEPACCEPTACCEPTT" }, + { CMUA, 0, "aA#\xc3\xa9\xc3\x81", "aA#Aa#\xc3\x89\xc3\xa1" }, + { MA, 0, "[^a]", "aAbB" }, + { CMA, 0, "[^m]", "mMnN" }, + { MA, 0, "a[^b][^#]", "abacd" }, + { CMA, 0, "A[^B][^E]", "abacd" }, + { CMUA, 0, "[^x][^#]", "XxBll" }, + { MUA, 0, "[^a]", "aaa\xc3\xa1#Ab" }, + { CMUA, 0, "[^A]", "aA\xe6\x92\xad" }, + { MUA, 0, "\\W(\\W)?\\w", "\r\n+bc" }, + { MUA, 0, "\\W(\\W)?\\w", "\n\r+bc" }, + { MUA, 0, "\\W(\\W)?\\w", "\r\r+bc" }, + { MUA, 0, "\\W(\\W)?\\w", "\n\n+bc" }, + { MUA, 0, "[axd]", "sAXd" }, + { CMUA, 0, "[axd]", "sAXd" }, + { CMUA, 0 | F_NOMATCH, "[^axd]", "DxA" }, + { MUA, 0, "[a-dA-C]", "\xe6\x92\xad\xc3\xa9.B" }, + { MUA, 0, "[^a-dA-C]", "\xe6\x92\xad\xc3\xa9" }, + { CMUA, 0, "[^\xc3\xa9]", "\xc3\xa9\xc3\x89." }, + { MUA, 0, "[^\xc3\xa9]", "\xc3\xa9\xc3\x89." }, + { MUA, 0, "[^a]", "\xc2\x80[]" }, + { CMUA, 0, "\xf0\x90\x90\xa7", "\xf0\x90\x91\x8f" }, + { CMA, 0, "1a2b3c4", "1a2B3c51A2B3C4" }, + { PCRE_CASELESS, 0, "\xff#a", "\xff#\xff\xfe##\xff#A" }, + { PCRE_CASELESS, 0, "\xfe", "\xff\xfc#\xfe\xfe" }, + { PCRE_CASELESS, 0, "a1", "Aa1" }, + { MA, 0, "\\Ca", "cda" }, + { CMA, 0, "\\Ca", "CDA" }, + { MA, 0 | F_NOMATCH, "\\Cx", "cda" }, + { CMA, 0 | F_NOMATCH, "\\Cx", "CDA" }, + { CMUAP, 0, "\xf0\x90\x90\x80\xf0\x90\x90\xa8", "\xf0\x90\x90\xa8\xf0\x90\x90\x80" }, + { CMUAP, 0, "\xf0\x90\x90\x80{2}", "\xf0\x90\x90\x80#\xf0\x90\x90\xa8\xf0\x90\x90\x80" }, + { CMUAP, 0, "\xf0\x90\x90\xa8{2}", "\xf0\x90\x90\x80#\xf0\x90\x90\xa8\xf0\x90\x90\x80" }, + { CMUAP, 0, "\xe1\xbd\xb8\xe1\xbf\xb8", "\xe1\xbf\xb8\xe1\xbd\xb8" }, + + /* Assertions. */ + { MUA, 0, "\\b[^A]", "A_B#" }, + { MA, 0 | F_NOMATCH, "\\b\\W", "\n*" }, + { MUA, 0, "\\B[^,]\\b[^s]\\b", "#X" }, + { MAP, 0, "\\B", "_\xa1" }, + { MAP, 0, "\\b_\\b[,A]\\B", "_," }, + { MUAP, 0, "\\b", "\xe6\x92\xad!" }, + { MUAP, 0, "\\B", "_\xc2\xa1\xc3\xa1\xc2\x85" }, + { MUAP, 0, "\\b[^A]\\B[^c]\\b[^_]\\B", "_\xc3\xa1\xe2\x80\xa8" }, + { MUAP, 0, "\\b\\w+\\B", "\xc3\x89\xc2\xa1\xe6\x92\xad\xc3\x81\xc3\xa1" }, + { MUA, 0 | F_NOMATCH, "\\b.", "\xcd\xbe" }, + { CMUAP, 0, "\\By", "\xf0\x90\x90\xa8y" }, + { MA, 0 | F_NOMATCH, "\\R^", "\n" }, + { MA, 1 | F_NOMATCH, "^", "\n" }, + { 0, 0, "^ab", "ab" }, + { 0, 0 | F_NOMATCH, "^ab", "aab" }, + { PCRE_MULTILINE | PCRE_NEWLINE_CRLF, 0, "^a", "\r\raa\n\naa\r\naa" }, + { PCRE_MULTILINE | PCRE_UTF8 | PCRE_NEWLINE_ANYCRLF, 0, "^-", "\xe2\x80\xa8--\xc2\x85-\r\n-" }, + { PCRE_MULTILINE | PCRE_NEWLINE_ANY, 0, "^-", "a--b--\x85--" }, + { PCRE_MULTILINE | PCRE_UTF8 | PCRE_NEWLINE_ANY, 0, "^-", "a--\xe2\x80\xa8--" }, + { PCRE_MULTILINE | PCRE_UTF8 | PCRE_NEWLINE_ANY, 0, "^-", "a--\xc2\x85--" }, + { 0, 0, "ab$", "ab" }, + { 0, 0 | F_NOMATCH, "ab$", "abab\n\n" }, + { PCRE_DOLLAR_ENDONLY, 0 | F_NOMATCH, "ab$", "abab\r\n" }, + { PCRE_MULTILINE | PCRE_NEWLINE_CRLF, 0, "a$", "\r\raa\n\naa\r\naa" }, + { PCRE_MULTILINE | PCRE_NEWLINE_ANY, 0, "a$", "aaa" }, + { PCRE_MULTILINE | PCRE_UTF8 | PCRE_NEWLINE_ANYCRLF, 0, "#$", "#\xc2\x85###\r#" }, + { PCRE_MULTILINE | PCRE_UTF8 | PCRE_NEWLINE_ANY, 0, "#$", "#\xe2\x80\xa9" }, + { PCRE_NOTBOL | PCRE_NEWLINE_ANY, 0 | F_NOMATCH, "^a", "aa\naa" }, + { PCRE_NOTBOL | PCRE_MULTILINE | PCRE_NEWLINE_ANY, 0, "^a", "aa\naa" }, + { PCRE_NOTEOL | PCRE_NEWLINE_ANY, 0 | F_NOMATCH, "a$", "aa\naa" }, + { PCRE_NOTEOL | PCRE_NEWLINE_ANY, 0 | F_NOMATCH, "a$", "aa\r\n" }, + { PCRE_UTF8 | PCRE_DOLLAR_ENDONLY | PCRE_NEWLINE_ANY, 0 | F_PROPERTY, "\\p{Any}{2,}$", "aa\r\n" }, + { PCRE_NOTEOL | PCRE_MULTILINE | PCRE_NEWLINE_ANY, 0, "a$", "aa\naa" }, + { PCRE_NEWLINE_CR, 0, ".\\Z", "aaa" }, + { PCRE_NEWLINE_CR | PCRE_UTF8, 0, "a\\Z", "aaa\r" }, + { PCRE_NEWLINE_CR, 0, ".\\Z", "aaa\n" }, + { PCRE_NEWLINE_CRLF, 0, ".\\Z", "aaa\r" }, + { PCRE_NEWLINE_CRLF | PCRE_UTF8, 0, ".\\Z", "aaa\n" }, + { PCRE_NEWLINE_CRLF, 0, ".\\Z", "aaa\r\n" }, + { PCRE_NEWLINE_ANYCRLF | PCRE_UTF8, 0, ".\\Z", "aaa" }, + { PCRE_NEWLINE_ANYCRLF | PCRE_UTF8, 0, ".\\Z", "aaa\r" }, + { PCRE_NEWLINE_ANYCRLF | PCRE_UTF8, 0, ".\\Z", "aaa\n" }, + { PCRE_NEWLINE_ANYCRLF | PCRE_UTF8, 0, ".\\Z", "aaa\r\n" }, + { PCRE_NEWLINE_ANYCRLF | PCRE_UTF8, 0, ".\\Z", "aaa\xe2\x80\xa8" }, + { PCRE_NEWLINE_ANYCRLF | PCRE_UTF8, 0, ".\\Z", "aaa" }, + { PCRE_NEWLINE_ANYCRLF | PCRE_UTF8, 0, ".\\Z", "aaa\r" }, + { PCRE_NEWLINE_ANYCRLF | PCRE_UTF8, 0, ".\\Z", "aaa\n" }, + { PCRE_NEWLINE_ANYCRLF | PCRE_UTF8, 0, ".\\Z", "aaa\r\n" }, + { PCRE_NEWLINE_ANY | PCRE_UTF8, 0, ".\\Z", "aaa\xc2\x85" }, + { PCRE_NEWLINE_ANY | PCRE_UTF8, 0, ".\\Z", "aaa\xe2\x80\xa8" }, + { MA, 0, "\\Aa", "aaa" }, + { MA, 1 | F_NOMATCH, "\\Aa", "aaa" }, + { MA, 1, "\\Ga", "aaa" }, + { MA, 1 | F_NOMATCH, "\\Ga", "aba" }, + { MA, 0, "a\\z", "aaa" }, + { MA, 0 | F_NOMATCH, "a\\z", "aab" }, + + /* Brackets. */ + { MUA, 0, "(ab|bb|cd)", "bacde" }, + { MUA, 0, "(?:ab|a)(bc|c)", "ababc" }, + { MUA, 0, "((ab|(cc))|(bb)|(?:cd|efg))", "abac" }, + { CMUA, 0, "((aB|(Cc))|(bB)|(?:cd|EFg))", "AcCe" }, + { MUA, 0, "((ab|(cc))|(bb)|(?:cd|ebg))", "acebebg" }, + { MUA, 0, "(?:(a)|(?:b))(cc|(?:d|e))(a|b)k", "accabdbbccbk" }, + + /* Greedy and non-greedy ? operators. */ + { MUA, 0, "(?:a)?a", "laab" }, + { CMUA, 0, "(A)?A", "llaab" }, + { MUA, 0, "(a)?\?a", "aab" }, /* ?? is the prefix of trygraphs in GCC. */ + { MUA, 0, "(a)?a", "manm" }, + { CMUA, 0, "(a|b)?\?d((?:e)?)", "ABABdx" }, + { MUA, 0, "(a|b)?\?d((?:e)?)", "abcde" }, + { MUA, 0, "((?:ab)?\?g|b(?:g(nn|d)?\?)?)?\?(?:n)?m", "abgnbgnnbgdnmm" }, + + /* Greedy and non-greedy + operators */ + { MUA, 0, "(aa)+aa", "aaaaaaa" }, + { MUA, 0, "(aa)+?aa", "aaaaaaa" }, + { MUA, 0, "(?:aba|ab|a)+l", "ababamababal" }, + { MUA, 0, "(?:aba|ab|a)+?l", "ababamababal" }, + { MUA, 0, "(a(?:bc|cb|b|c)+?|ss)+e", "accssabccbcacbccbbXaccssabccbcacbccbbe" }, + { MUA, 0, "(a(?:bc|cb|b|c)+|ss)+?e", "accssabccbcacbccbbXaccssabccbcacbccbbe" }, + { MUA, 0, "(?:(b(c)+?)+)?\?(?:(bc)+|(cb)+)+(?:m)+", "bccbcccbcbccbcbPbccbcccbcbccbcbmmn" }, + + /* Greedy and non-greedy * operators */ + { CMUA, 0, "(?:AA)*AB", "aaaaaaamaaaaaaab" }, + { MUA, 0, "(?:aa)*?ab", "aaaaaaamaaaaaaab" }, + { MUA, 0, "(aa|ab)*ab", "aaabaaab" }, + { CMUA, 0, "(aa|Ab)*?aB", "aaabaaab" }, + { MUA, 0, "(a|b)*(?:a)*(?:b)*m", "abbbaaababanabbbaaababamm" }, + { MUA, 0, "(a|b)*?(?:a)*?(?:b)*?m", "abbbaaababanabbbaaababamm" }, + { MA, 0, "a(a(\\1*)a|(b)b+){0}a", "aa" }, + { MA, 0, "((?:a|)*){0}a", "a" }, + + /* Combining ? + * operators */ + { MUA, 0, "((bm)+)?\?(?:a)*(bm)+n|((am)+?)?(?:a)+(am)*n", "bmbmabmamaaamambmaman" }, + { MUA, 0, "(((ab)?cd)*ef)+g", "abcdcdefcdefefmabcdcdefcdefefgg" }, + { MUA, 0, "(((ab)?\?cd)*?ef)+?g", "abcdcdefcdefefmabcdcdefcdefefgg" }, + { MUA, 0, "(?:(ab)?c|(?:ab)+?d)*g", "ababcdccababddg" }, + { MUA, 0, "(?:(?:ab)?\?c|(ab)+d)*?g", "ababcdccababddg" }, + + /* Single character iterators. */ + { MUA, 0, "(a+aab)+aaaab", "aaaabcaaaabaabcaabcaaabaaaab" }, + { MUA, 0, "(a*a*aab)+x", "aaaaabaabaaabmaabx" }, + { MUA, 0, "(a*?(b|ab)a*?)+x", "aaaabcxbbaabaacbaaabaabax" }, + { MUA, 0, "(a+(ab|ad)a+)+x", "aaabaaaadaabaaabaaaadaaax" }, + { MUA, 0, "(a?(a)a?)+(aaa)", "abaaabaaaaaaaa" }, + { MUA, 0, "(a?\?(a)a?\?)+(b)", "aaaacaaacaacacbaaab" }, + { MUA, 0, "(a{0,4}(b))+d", "aaaaaabaabcaaaaabaaaaabd" }, + { MUA, 0, "(a{0,4}?[^b])+d+(a{0,4}[^b])d+", "aaaaadaaaacaadddaaddd" }, + { MUA, 0, "(ba{2})+c", "baabaaabacbaabaac" }, + { MUA, 0, "(a*+bc++)+", "aaabbcaaabcccab" }, + { MUA, 0, "(a?+[^b])+", "babaacacb" }, + { MUA, 0, "(a{0,3}+b)(a{0,3}+b)(a{0,3}+)[^c]", "abaabaaacbaabaaaac" }, + { CMUA, 0, "([a-c]+[d-f]+?)+?g", "aBdacdehAbDaFgA" }, + { CMUA, 0, "[c-f]+k", "DemmFke" }, + { MUA, 0, "([DGH]{0,4}M)+", "GGDGHDGMMHMDHHGHM" }, + { MUA, 0, "([a-c]{4,}s)+", "abasabbasbbaabsbba" }, + { CMUA, 0, "[ace]{3,7}", "AcbDAcEEcEd" }, + { CMUA, 0, "[ace]{3,7}?", "AcbDAcEEcEd" }, + { CMUA, 0, "[ace]{3,}", "AcbDAcEEcEd" }, + { CMUA, 0, "[ace]{3,}?", "AcbDAcEEcEd" }, + { MUA, 0, "[ckl]{2,}?g", "cdkkmlglglkcg" }, + { CMUA, 0, "[ace]{5}?", "AcCebDAcEEcEd" }, + { MUA, 0, "([AbC]{3,5}?d)+", "BACaAbbAEAACCbdCCbdCCAAbb" }, + { MUA, 0, "([^ab]{0,}s){2}", "abaabcdsABamsDDs" }, + { MUA, 0, "\\b\\w+\\B", "x,a_cd" }, + { MUAP, 0, "\\b[^\xc2\xa1]+\\B", "\xc3\x89\xc2\xa1\xe6\x92\xad\xc3\x81\xc3\xa1" }, + { CMUA, 0, "[^b]+(a*)([^c]?d{3})", "aaaaddd" }, + { CMUAP, 0, "\xe1\xbd\xb8{2}", "\xe1\xbf\xb8#\xe1\xbf\xb8\xe1\xbd\xb8" }, + { CMUA, 0, "[^\xf0\x90\x90\x80]{2,4}@", "\xf0\x90\x90\xa8\xf0\x90\x90\x80###\xf0\x90\x90\x80@@@" }, + { CMUA, 0, "[^\xe1\xbd\xb8][^\xc3\xa9]", "\xe1\xbd\xb8\xe1\xbf\xb8\xc3\xa9\xc3\x89#" }, + { MUA, 0, "[^\xe1\xbd\xb8][^\xc3\xa9]", "\xe1\xbd\xb8\xe1\xbf\xb8\xc3\xa9\xc3\x89#" }, + { MUA, 0, "[^\xe1\xbd\xb8]{3,}?", "##\xe1\xbd\xb8#\xe1\xbd\xb8#\xc3\x89#\xe1\xbd\xb8" }, + + /* Basic character sets. */ + { MUA, 0, "(?:\\s)+(?:\\S)+", "ab \t\xc3\xa9\xe6\x92\xad " }, + { MUA, 0, "(\\w)*(k)(\\W)?\?", "abcdef abck11" }, + { MUA, 0, "\\((\\d)+\\)\\D", "a() (83 (8)2 (9)ab" }, + { MUA, 0, "\\w(\\s|(?:\\d)*,)+\\w\\wb", "a 5, 4,, bb 5, 4,, aab" }, + { MUA, 0, "(\\v+)(\\V+)", "\x0e\xc2\x85\xe2\x80\xa8\x0b\x09\xe2\x80\xa9" }, + { MUA, 0, "(\\h+)(\\H+)", "\xe2\x80\xa8\xe2\x80\x80\x20\xe2\x80\x8a\xe2\x81\x9f\xe3\x80\x80\x09\x20\xc2\xa0\x0a" }, + + /* Unicode properties. */ + { MUAP, 0, "[1-5\xc3\xa9\\w]", "\xc3\xa1_" }, + { MUAP, 0 | F_PROPERTY, "[\xc3\x81\\p{Ll}]", "A_\xc3\x89\xc3\xa1" }, + { MUAP, 0, "[\\Wd-h_x-z]+", "a\xc2\xa1#_yhzdxi" }, + { MUAP, 0 | F_NOMATCH | F_PROPERTY, "[\\P{Any}]", "abc" }, + { MUAP, 0 | F_NOMATCH | F_PROPERTY, "[^\\p{Any}]", "abc" }, + { MUAP, 0 | F_NOMATCH | F_PROPERTY, "[\\P{Any}\xc3\xa1-\xc3\xa8]", "abc" }, + { MUAP, 0 | F_NOMATCH | F_PROPERTY, "[^\\p{Any}\xc3\xa1-\xc3\xa8]", "abc" }, + { MUAP, 0 | F_NOMATCH | F_PROPERTY, "[\xc3\xa1-\xc3\xa8\\P{Any}]", "abc" }, + { MUAP, 0 | F_NOMATCH | F_PROPERTY, "[^\xc3\xa1-\xc3\xa8\\p{Any}]", "abc" }, + { MUAP, 0 | F_PROPERTY, "[\xc3\xa1-\xc3\xa8\\p{Any}]", "abc" }, + { MUAP, 0 | F_PROPERTY, "[^\xc3\xa1-\xc3\xa8\\P{Any}]", "abc" }, + { MUAP, 0, "[b-\xc3\xa9\\s]", "a\xc\xe6\x92\xad" }, + { CMUAP, 0, "[\xc2\x85-\xc2\x89\xc3\x89]", "\xc2\x84\xc3\xa9" }, + { MUAP, 0, "[^b-d^&\\s]{3,}", "db^ !a\xe2\x80\xa8_ae" }, + { MUAP, 0 | F_PROPERTY, "[^\\S\\P{Any}][\\sN]{1,3}[\\P{N}]{4}", "\xe2\x80\xaa\xa N\x9\xc3\xa9_0" }, + { MUA, 0 | F_PROPERTY, "[^\\P{L}\x9!D-F\xa]{2,3}", "\x9,.DF\xa.CG\xc3\x81" }, + { CMUAP, 0, "[\xc3\xa1-\xc3\xa9_\xe2\x80\xa0-\xe2\x80\xaf]{1,5}[^\xe2\x80\xa0-\xe2\x80\xaf]", "\xc2\xa1\xc3\x89\xc3\x89\xe2\x80\xaf_\xe2\x80\xa0" }, + { MUAP, 0 | F_PROPERTY, "[\xc3\xa2-\xc3\xa6\xc3\x81-\xc3\x84\xe2\x80\xa8-\xe2\x80\xa9\xe6\x92\xad\\p{Zs}]{2,}", "\xe2\x80\xa7\xe2\x80\xa9\xe6\x92\xad \xe6\x92\xae" }, + { MUAP, 0 | F_PROPERTY, "[\\P{L&}]{2}[^\xc2\x85-\xc2\x89\\p{Ll}\\p{Lu}]{2}", "\xc3\xa9\xe6\x92\xad.a\xe6\x92\xad|\xc2\x8a#" }, + { PCRE_UCP, 0, "[a-b\\s]{2,5}[^a]", "AB baaa" }, + + /* Possible empty brackets. */ + { MUA, 0, "(?:|ab||bc|a)+d", "abcxabcabd" }, + { MUA, 0, "(|ab||bc|a)+d", "abcxabcabd" }, + { MUA, 0, "(?:|ab||bc|a)*d", "abcxabcabd" }, + { MUA, 0, "(|ab||bc|a)*d", "abcxabcabd" }, + { MUA, 0, "(?:|ab||bc|a)+?d", "abcxabcabd" }, + { MUA, 0, "(|ab||bc|a)+?d", "abcxabcabd" }, + { MUA, 0, "(?:|ab||bc|a)*?d", "abcxabcabd" }, + { MUA, 0, "(|ab||bc|a)*?d", "abcxabcabd" }, + { MUA, 0, "(((a)*?|(?:ba)+)+?|(?:|c|ca)*)*m", "abaacaccabacabalabaacaccabacabamm" }, + { MUA, 0, "(?:((?:a)*|(ba)+?)+|(|c|ca)*?)*?m", "abaacaccabacabalabaacaccabacabamm" }, + + /* Start offset. */ + { MUA, 3, "(\\d|(?:\\w)*\\w)+", "0ac01Hb" }, + { MUA, 4 | F_NOMATCH, "(\\w\\W\\w)+", "ab#d" }, + { MUA, 2 | F_NOMATCH, "(\\w\\W\\w)+", "ab#d" }, + { MUA, 1, "(\\w\\W\\w)+", "ab#d" }, + + /* Newline. */ + { PCRE_MULTILINE | PCRE_NEWLINE_CRLF, 0, "\\W{0,2}[^#]{3}", "\r\n#....." }, + { PCRE_MULTILINE | PCRE_NEWLINE_CR, 0, "\\W{0,2}[^#]{3}", "\r\n#....." }, + { PCRE_MULTILINE | PCRE_NEWLINE_CRLF, 0, "\\W{1,3}[^#]", "\r\n##...." }, + + /* Any character except newline or any newline. */ + { PCRE_NEWLINE_CRLF, 0, ".", "\r" }, + { PCRE_NEWLINE_CRLF | PCRE_UTF8, 0, ".(.).", "a\xc3\xa1\r\n\n\r\r" }, + { PCRE_NEWLINE_ANYCRLF, 0, ".(.)", "a\rb\nc\r\n\xc2\x85\xe2\x80\xa8" }, + { PCRE_NEWLINE_ANYCRLF | PCRE_UTF8, 0, ".(.)", "a\rb\nc\r\n\xc2\x85\xe2\x80\xa8" }, + { PCRE_NEWLINE_ANY | PCRE_UTF8, 0, "(.).", "a\rb\nc\r\n\xc2\x85\xe2\x80\xa9$de" }, + { PCRE_NEWLINE_ANYCRLF | PCRE_UTF8, 0 | F_NOMATCH, ".(.).", "\xe2\x80\xa8\nb\r" }, + { PCRE_NEWLINE_ANY, 0, "(.)(.)", "#\x85#\r#\n#\r\n#\x84" }, + { PCRE_NEWLINE_ANY | PCRE_UTF8, 0, "(.+)#", "#\rMn\xc2\x85#\n###" }, + { PCRE_BSR_ANYCRLF, 0, "\\R", "\r" }, + { PCRE_BSR_ANYCRLF, 0, "\\R", "\x85#\r\n#" }, + { PCRE_BSR_UNICODE | PCRE_UTF8, 0, "\\R", "ab\xe2\x80\xa8#c" }, + { PCRE_BSR_UNICODE | PCRE_UTF8, 0, "\\R", "ab\r\nc" }, + { PCRE_NEWLINE_CRLF | PCRE_BSR_UNICODE | PCRE_UTF8, 0, "(\\R.)+", "\xc2\x85\r\n#\xe2\x80\xa8\n\r\n\r" }, + { MUA, 0 | F_NOMATCH, "\\R+", "ab" }, + { MUA, 0, "\\R+", "ab\r\n\r" }, + { MUA, 0, "\\R*", "ab\r\n\r" }, + { MUA, 0, "\\R*", "\r\n\r" }, + { MUA, 0, "\\R{2,4}", "\r\nab\r\r" }, + { MUA, 0, "\\R{2,4}", "\r\nab\n\n\n\r\r\r" }, + { MUA, 0, "\\R{2,}", "\r\nab\n\n\n\r\r\r" }, + { MUA, 0, "\\R{0,3}", "\r\n\r\n\r\n\r\n\r\n" }, + { MUA, 0 | F_NOMATCH, "\\R+\\R\\R", "\r\n\r\n" }, + { MUA, 0, "\\R+\\R\\R", "\r\r\r" }, + { MUA, 0, "\\R*\\R\\R", "\n\r" }, + { MUA, 0 | F_NOMATCH, "\\R{2,4}\\R\\R", "\r\r\r" }, + { MUA, 0, "\\R{2,4}\\R\\R", "\r\r\r\r" }, + + /* Atomic groups (no fallback from "next" direction). */ + { MUA, 0 | F_NOMATCH, "(?>ab)ab", "bab" }, + { MUA, 0 | F_NOMATCH, "(?>(ab))ab", "bab" }, + { MUA, 0, "(?>ab)+abc(?>de)*def(?>gh)?ghe(?>ij)+?k(?>lm)*?n(?>op)?\?op", + "bababcdedefgheijijklmlmnop" }, + { MUA, 0, "(?>a(b)+a|(ab)?\?(b))an", "abban" }, + { MUA, 0, "(?>ab+a|(?:ab)?\?b)an", "abban" }, + { MUA, 0, "((?>ab|ad|)*?)(?>|c)*abad", "abababcababad" }, + { MUA, 0, "(?>(aa|b|)*+(?>(##)|###)*d|(aa)(?>(baa)?)m)", "aabaa#####da" }, + { MUA, 0, "((?>a|)+?)b", "aaacaaab" }, + { MUA, 0, "(?>x|)*$", "aaa" }, + { MUA, 0, "(?>(x)|)*$", "aaa" }, + { MUA, 0, "(?>x|())*$", "aaa" }, + { MUA, 0, "((?>[cxy]a|[a-d])*?)b", "aaa+ aaab" }, + { MUA, 0, "((?>[cxy](a)|[a-d])*?)b", "aaa+ aaab" }, + { MUA, 0, "(?>((?>(a+))))bab|(?>((?>(a+))))bb", "aaaabaaabaabab" }, + { MUA, 0, "(?>(?>a+))bab|(?>(?>a+))bb", "aaaabaaabaabab" }, + { MUA, 0, "(?>(a)c|(?>(c)|(a))a)b*?bab", "aaaabaaabaabab" }, + { MUA, 0, "(?>ac|(?>c|a)a)b*?bab", "aaaabaaabaabab" }, + { MUA, 0, "(?>(b)b|(a))*b(?>(c)|d)?x", "ababcaaabdbx" }, + { MUA, 0, "(?>bb|a)*b(?>c|d)?x", "ababcaaabdbx" }, + { MUA, 0, "(?>(bb)|a)*b(?>c|(d))?x", "ababcaaabdbx" }, + { MUA, 0, "(?>(a))*?(?>(a))+?(?>(a))??x", "aaaaaacccaaaaabax" }, + { MUA, 0, "(?>a)*?(?>a)+?(?>a)??x", "aaaaaacccaaaaabax" }, + { MUA, 0, "(?>(a)|)*?(?>(a)|)+?(?>(a)|)??x", "aaaaaacccaaaaabax" }, + { MUA, 0, "(?>a|)*?(?>a|)+?(?>a|)??x", "aaaaaacccaaaaabax" }, + { MUA, 0, "(?>a(?>(a{0,2}))*?b|aac)+b", "aaaaaaacaaaabaaaaacaaaabaacaaabb" }, + { CMA, 0, "(?>((?>a{32}|b+|(a*))?(?>c+|d*)?\?)+e)+?f", "aaccebbdde bbdaaaccebbdee bbdaaaccebbdeef" }, + { MUA, 0, "(?>(?:(?>aa|a||x)+?b|(?>aa|a||(x))+?c)?(?>[ad]{0,2})*?d)+d", "aaacdbaabdcabdbaaacd aacaabdbdcdcaaaadaabcbaadd" }, + { MUA, 0, "(?>(?:(?>aa|a||(x))+?b|(?>aa|a||x)+?c)?(?>[ad]{0,2})*?d)+d", "aaacdbaabdcabdbaaacd aacaabdbdcdcaaaadaabcbaadd" }, + { MUA, 0 | F_NOMATCH | F_PROPERTY, "\\X", "\xcc\x8d\xcc\x8d" }, + { MUA, 0 | F_PROPERTY, "\\X", "\xcc\x8d\xcc\x8d#\xcc\x8d\xcc\x8d" }, + { MUA, 0 | F_PROPERTY, "\\X+..", "\xcc\x8d#\xcc\x8d#\xcc\x8d\xcc\x8d" }, + { MUA, 0 | F_PROPERTY, "\\X{2,4}", "abcdef" }, + { MUA, 0 | F_PROPERTY, "\\X{2,4}?", "abcdef" }, + { MUA, 0 | F_NOMATCH | F_PROPERTY, "\\X{2,4}..", "#\xcc\x8d##" }, + { MUA, 0 | F_PROPERTY, "\\X{2,4}..", "#\xcc\x8d#\xcc\x8d##" }, + { MUA, 0, "(c(ab)?+ab)+", "cabcababcab" }, + { MUA, 0, "(?>(a+)b)+aabab", "aaaabaaabaabab" }, + + /* Possessive quantifiers. */ + { MUA, 0, "(?:a|b)++m", "mababbaaxababbaam" }, + { MUA, 0, "(?:a|b)*+m", "mababbaaxababbaam" }, + { MUA, 0, "(?:a|b)*+m", "ababbaaxababbaam" }, + { MUA, 0, "(a|b)++m", "mababbaaxababbaam" }, + { MUA, 0, "(a|b)*+m", "mababbaaxababbaam" }, + { MUA, 0, "(a|b)*+m", "ababbaaxababbaam" }, + { MUA, 0, "(a|b(*ACCEPT))++m", "maaxab" }, + { MUA, 0, "(?:b*)++m", "bxbbxbbbxm" }, + { MUA, 0, "(?:b*)++m", "bxbbxbbbxbbm" }, + { MUA, 0, "(?:b*)*+m", "bxbbxbbbxm" }, + { MUA, 0, "(?:b*)*+m", "bxbbxbbbxbbm" }, + { MUA, 0, "(b*)++m", "bxbbxbbbxm" }, + { MUA, 0, "(b*)++m", "bxbbxbbbxbbm" }, + { MUA, 0, "(b*)*+m", "bxbbxbbbxm" }, + { MUA, 0, "(b*)*+m", "bxbbxbbbxbbm" }, + { MUA, 0, "(?:a|(b))++m", "mababbaaxababbaam" }, + { MUA, 0, "(?:(a)|b)*+m", "mababbaaxababbaam" }, + { MUA, 0, "(?:(a)|(b))*+m", "ababbaaxababbaam" }, + { MUA, 0, "(a|(b))++m", "mababbaaxababbaam" }, + { MUA, 0, "((a)|b)*+m", "mababbaaxababbaam" }, + { MUA, 0, "((a)|(b))*+m", "ababbaaxababbaam" }, + { MUA, 0, "(a|(b)(*ACCEPT))++m", "maaxab" }, + { MUA, 0, "(?:(b*))++m", "bxbbxbbbxm" }, + { MUA, 0, "(?:(b*))++m", "bxbbxbbbxbbm" }, + { MUA, 0, "(?:(b*))*+m", "bxbbxbbbxm" }, + { MUA, 0, "(?:(b*))*+m", "bxbbxbbbxbbm" }, + { MUA, 0, "((b*))++m", "bxbbxbbbxm" }, + { MUA, 0, "((b*))++m", "bxbbxbbbxbbm" }, + { MUA, 0, "((b*))*+m", "bxbbxbbbxm" }, + { MUA, 0, "((b*))*+m", "bxbbxbbbxbbm" }, + { MUA, 0 | F_NOMATCH, "(?>(b{2,4}))(?:(?:(aa|c))++m|(?:(aa|c))+n)", "bbaacaaccaaaacxbbbmbn" }, + { MUA, 0, "((?:b)++a)+(cd)*+m", "bbababbacdcdnbbababbacdcdm" }, + { MUA, 0, "((?:(b))++a)+((c)d)*+m", "bbababbacdcdnbbababbacdcdm" }, + { MUA, 0, "(?:(?:(?:ab)*+k)++(?:n(?:cd)++)*+)*+m", "ababkkXababkkabkncXababkkabkncdcdncdXababkkabkncdcdncdkkabkncdXababkkabkncdcdncdkkabkncdm" }, + { MUA, 0, "(?:((ab)*+(k))++(n(?:c(d))++)*+)*+m", "ababkkXababkkabkncXababkkabkncdcdncdXababkkabkncdcdncdkkabkncdXababkkabkncdcdncdkkabkncdm" }, + + /* Back references. */ + { MUA, 0, "(aa|bb)(\\1*)(ll|)(\\3*)bbbbbbc", "aaaaaabbbbbbbbc" }, + { CMUA, 0, "(aa|bb)(\\1+)(ll|)(\\3+)bbbbbbc", "bBbbBbCbBbbbBbbcbbBbbbBBbbC" }, + { CMA, 0, "(a{2,4})\\1", "AaAaaAaA" }, + { MUA, 0, "(aa|bb)(\\1?)aa(\\1?)(ll|)(\\4+)bbc", "aaaaaaaabbaabbbbaabbbbc" }, + { MUA, 0, "(aa|bb)(\\1{0,5})(ll|)(\\3{0,5})cc", "bbxxbbbbxxaaaaaaaaaaaaaaaacc" }, + { MUA, 0, "(aa|bb)(\\1{3,5})(ll|)(\\3{3,5})cc", "bbbbbbbbbbbbaaaaaaccbbbbbbbbbbbbbbcc" }, + { MUA, 0, "(aa|bb)(\\1{3,})(ll|)(\\3{3,})cc", "bbbbbbbbbbbbaaaaaaccbbbbbbbbbbbbbbcc" }, + { MUA, 0, "(\\w+)b(\\1+)c", "GabGaGaDbGaDGaDc" }, + { MUA, 0, "(?:(aa)|b)\\1?b", "bb" }, + { CMUA, 0, "(aa|bb)(\\1*?)aa(\\1+?)", "bBBbaaAAaaAAaa" }, + { MUA, 0, "(aa|bb)(\\1*?)(dd|)cc(\\3+?)", "aaaaaccdd" }, + { CMUA, 0, "(?:(aa|bb)(\\1?\?)cc){2}(\\1?\?)", "aAaABBbbAAaAcCaAcCaA" }, + { MUA, 0, "(?:(aa|bb)(\\1{3,5}?)){2}(dd|)(\\3{3,5}?)", "aaaaaabbbbbbbbbbaaaaaaaaaaaaaa" }, + { CMA, 0, "(?:(aa|bb)(\\1{3,}?)){2}(dd|)(\\3{3,}?)", "aaaaaabbbbbbbbbbaaaaaaaaaaaaaa" }, + { MUA, 0, "(?:(aa|bb)(\\1{0,3}?)){2}(dd|)(\\3{0,3}?)b(\\1{0,3}?)(\\1{0,3})", "aaaaaaaaaaaaaaabaaaaa" }, + { MUA, 0, "(a(?:\\1|)a){3}b", "aaaaaaaaaaab" }, + { MA, 0, "(a?)b(\\1\\1*\\1+\\1?\\1*?\\1+?\\1??\\1*+\\1++\\1?+\\1{4}\\1{3,5}\\1{4,}\\1{0,5}\\1{3,5}?\\1{4,}?\\1{0,5}?\\1{3,5}+\\1{4,}+\\1{0,5}+#){2}d", "bb#b##d" }, + { MUAP, 0 | F_PROPERTY, "(\\P{N})\\1{2,}", ".www." }, + { MUAP, 0 | F_PROPERTY, "(\\P{N})\\1{0,2}", "wwwww." }, + { MUAP, 0 | F_PROPERTY, "(\\P{N})\\1{1,2}ww", "wwww" }, + { MUAP, 0 | F_PROPERTY, "(\\P{N})\\1{1,2}ww", "wwwww" }, + { PCRE_UCP, 0 | F_PROPERTY, "(\\P{N})\\1{2,}", ".www." }, + { CMUAP, 0, "(\xf0\x90\x90\x80)\\1", "\xf0\x90\x90\xa8\xf0\x90\x90\xa8" }, + + /* Assertions. */ + { MUA, 0, "(?=xx|yy|zz)\\w{4}", "abczzdefg" }, + { MUA, 0, "(?=((\\w+)b){3}|ab)", "dbbbb ab" }, + { MUA, 0, "(?!ab|bc|cd)[a-z]{2}", "Xabcdef" }, + { MUA, 0, "(?<=aaa|aa|a)a", "aaa" }, + { MUA, 2, "(?<=aaa|aa|a)a", "aaa" }, + { MA, 0, "(?<=aaa|aa|a)a", "aaa" }, + { MA, 2, "(?<=aaa|aa|a)a", "aaa" }, + { MUA, 0, "(\\d{2})(?!\\w+c|(((\\w?)m){2}n)+|\\1)", "x5656" }, + { MUA, 0, "((?=((\\d{2,6}\\w){2,}))\\w{5,20}K){2,}", "567v09708K12l00M00 567v09708K12l00M00K45K" }, + { MUA, 0, "(?=(?:(?=\\S+a)\\w*(b)){3})\\w+\\d", "bba bbab nbbkba nbbkba0kl" }, + { MUA, 0, "(?>a(?>(b+))a(?=(..)))*?k", "acabbcabbaabacabaabbakk" }, + { MUA, 0, "((?(?=(a))a)+k)", "bbak" }, + { MUA, 0, "((?(?=a)a)+k)", "bbak" }, + { MUA, 0 | F_NOMATCH, "(?=(?>(a))m)amk", "a k" }, + { MUA, 0 | F_NOMATCH, "(?!(?>(a))m)amk", "a k" }, + { MUA, 0 | F_NOMATCH, "(?>(?=(a))am)amk", "a k" }, + { MUA, 0, "(?=(?>a|(?=(?>(b+))a|c)[a-c]+)*?m)[a-cm]+k", "aaam bbam baaambaam abbabba baaambaamk" }, + { MUA, 0, "(?> ?\?\\b(?(?=\\w{1,4}(a))m)\\w{0,8}bc){2,}?", "bca ssbc mabd ssbc mabc" }, + { MUA, 0, "(?:(?=ab)?[^n][^n])+m", "ababcdabcdcdabnababcdabcdcdabm" }, + { MUA, 0, "(?:(?=a(b))?[^n][^n])+m", "ababcdabcdcdabnababcdabcdcdabm" }, + { MUA, 0, "(?:(?=.(.))??\\1.)+m", "aabbbcbacccanaabbbcbacccam" }, + { MUA, 0, "(?:(?=.)??[a-c])+m", "abacdcbacacdcaccam" }, + { MUA, 0, "((?!a)?(?!([^a]))?)+$", "acbab" }, + { MUA, 0, "((?!a)?\?(?!([^a]))?\?)+$", "acbab" }, + + /* Not empty, ACCEPT, FAIL */ + { MUA | PCRE_NOTEMPTY, 0 | F_NOMATCH, "a*", "bcx" }, + { MUA | PCRE_NOTEMPTY, 0, "a*", "bcaad" }, + { MUA | PCRE_NOTEMPTY, 0, "a*?", "bcaad" }, + { MUA | PCRE_NOTEMPTY_ATSTART, 0, "a*", "bcaad" }, + { MUA, 0, "a(*ACCEPT)b", "ab" }, + { MUA | PCRE_NOTEMPTY, 0 | F_NOMATCH, "a*(*ACCEPT)b", "bcx" }, + { MUA | PCRE_NOTEMPTY, 0, "a*(*ACCEPT)b", "bcaad" }, + { MUA | PCRE_NOTEMPTY, 0, "a*?(*ACCEPT)b", "bcaad" }, + { MUA | PCRE_NOTEMPTY, 0 | F_NOMATCH, "(?:z|a*(*ACCEPT)b)", "bcx" }, + { MUA | PCRE_NOTEMPTY, 0, "(?:z|a*(*ACCEPT)b)", "bcaad" }, + { MUA | PCRE_NOTEMPTY, 0, "(?:z|a*?(*ACCEPT)b)", "bcaad" }, + { MUA | PCRE_NOTEMPTY_ATSTART, 0, "a*(*ACCEPT)b", "bcx" }, + { MUA | PCRE_NOTEMPTY_ATSTART, 0 | F_NOMATCH, "a*(*ACCEPT)b", "" }, + { MUA, 0, "((a(*ACCEPT)b))", "ab" }, + { MUA, 0, "(a(*FAIL)a|a)", "aaa" }, + { MUA, 0, "(?=ab(*ACCEPT)b)a", "ab" }, + { MUA, 0, "(?=(?:x|ab(*ACCEPT)b))", "ab" }, + { MUA, 0, "(?=(a(b(*ACCEPT)b)))a", "ab" }, + { MUA | PCRE_NOTEMPTY, 0, "(?=a*(*ACCEPT))c", "c" }, + + /* Conditional blocks. */ + { MUA, 0, "(?(?=(a))a|b)+k", "ababbalbbadabak" }, + { MUA, 0, "(?(?!(b))a|b)+k", "ababbalbbadabak" }, + { MUA, 0, "(?(?=a)a|b)+k", "ababbalbbadabak" }, + { MUA, 0, "(?(?!b)a|b)+k", "ababbalbbadabak" }, + { MUA, 0, "(?(?=(a))a*|b*)+k", "ababbalbbadabak" }, + { MUA, 0, "(?(?!(b))a*|b*)+k", "ababbalbbadabak" }, + { MUA, 0, "(?(?!(b))(?:aaaaaa|a)|(?:bbbbbb|b))+aaaak", "aaaaaaaaaaaaaa bbbbbbbbbbbbbbb aaaaaaak" }, + { MUA, 0, "(?(?!b)(?:aaaaaa|a)|(?:bbbbbb|b))+aaaak", "aaaaaaaaaaaaaa bbbbbbbbbbbbbbb aaaaaaak" }, + { MUA, 0 | F_DIFF, "(?(?!(b))(?:aaaaaa|a)|(?:bbbbbb|b))+bbbbk", "aaaaaaaaaaaaaa bbbbbbbbbbbbbbb bbbbbbbk" }, + { MUA, 0, "(?(?!b)(?:aaaaaa|a)|(?:bbbbbb|b))+bbbbk", "aaaaaaaaaaaaaa bbbbbbbbbbbbbbb bbbbbbbk" }, + { MUA, 0, "(?(?=a)a*|b*)+k", "ababbalbbadabak" }, + { MUA, 0, "(?(?!b)a*|b*)+k", "ababbalbbadabak" }, + { MUA, 0, "(?(?=a)ab)", "a" }, + { MUA, 0, "(?(?a)?(?Pb)?(?(Name)c|d)*l", "bc ddd abccabccl" }, + { MUA, 0, "(?Pa)?(?Pb)?(?(Name)c|d)+?dd", "bcabcacdb bdddd" }, + { MUA, 0, "(?Pa)?(?Pb)?(?(Name)c|d)+l", "ababccddabdbccd abcccl" }, + + /* Set start of match. */ + { MUA, 0, "(?:\\Ka)*aaaab", "aaaaaaaa aaaaaaabb" }, + { MUA, 0, "(?>\\Ka\\Ka)*aaaab", "aaaaaaaa aaaaaaaaaabb" }, + { MUA, 0, "a+\\K(?<=\\Gaa)a", "aaaaaa" }, + { MUA | PCRE_NOTEMPTY, 0 | F_NOMATCH, "a\\K(*ACCEPT)b", "aa" }, + { MUA | PCRE_NOTEMPTY_ATSTART, 0, "a\\K(*ACCEPT)b", "aa" }, + + /* First line. */ + { MUA | PCRE_FIRSTLINE, 0 | F_PROPERTY, "\\p{Any}a", "bb\naaa" }, + { MUA | PCRE_FIRSTLINE, 0 | F_NOMATCH | F_PROPERTY, "\\p{Any}a", "bb\r\naaa" }, + { MUA | PCRE_FIRSTLINE, 0, "(?<=a)", "a" }, + { MUA | PCRE_FIRSTLINE, 0 | F_NOMATCH, "[^a][^b]", "ab" }, + { MUA | PCRE_FIRSTLINE, 0 | F_NOMATCH, "a", "\na" }, + { MUA | PCRE_FIRSTLINE, 0 | F_NOMATCH, "[abc]", "\na" }, + { MUA | PCRE_FIRSTLINE, 0 | F_NOMATCH, "^a", "\na" }, + { MUA | PCRE_FIRSTLINE, 0 | F_NOMATCH, "^(?<=\n)", "\na" }, + { PCRE_MULTILINE | PCRE_UTF8 | PCRE_NEWLINE_ANY | PCRE_FIRSTLINE, 0 | F_NOMATCH, "#", "\xc2\x85#" }, + { PCRE_MULTILINE | PCRE_NEWLINE_ANY | PCRE_FIRSTLINE, 0 | F_NOMATCH, "#", "\x85#" }, + { PCRE_MULTILINE | PCRE_UTF8 | PCRE_NEWLINE_ANY | PCRE_FIRSTLINE, 0 | F_NOMATCH, "^#", "\xe2\x80\xa8#" }, + { PCRE_MULTILINE | PCRE_UTF8 | PCRE_NEWLINE_CRLF | PCRE_FIRSTLINE, 0 | F_PROPERTY, "\\p{Any}", "\r\na" }, + { PCRE_MULTILINE | PCRE_UTF8 | PCRE_NEWLINE_CRLF | PCRE_FIRSTLINE, 0, ".", "\r" }, + { PCRE_MULTILINE | PCRE_UTF8 | PCRE_NEWLINE_CRLF | PCRE_FIRSTLINE, 0, "a", "\ra" }, + { PCRE_MULTILINE | PCRE_UTF8 | PCRE_NEWLINE_CRLF | PCRE_FIRSTLINE, 0 | F_NOMATCH, "ba", "bbb\r\nba" }, + { PCRE_MULTILINE | PCRE_UTF8 | PCRE_NEWLINE_CRLF | PCRE_FIRSTLINE, 0 | F_NOMATCH | F_PROPERTY, "\\p{Any}{4}|a", "\r\na" }, + { PCRE_MULTILINE | PCRE_UTF8 | PCRE_NEWLINE_CRLF | PCRE_FIRSTLINE, 1, ".", "\r\n" }, + + /* Recurse. */ + { MUA, 0, "(a)(?1)", "aa" }, + { MUA, 0, "((a))(?1)", "aa" }, + { MUA, 0, "(b|a)(?1)", "aa" }, + { MUA, 0, "(b|(a))(?1)", "aa" }, + { MUA, 0 | F_NOMATCH, "((a)(b)(?:a*))(?1)", "aba" }, + { MUA, 0, "((a)(b)(?:a*))(?1)", "abab" }, + { MUA, 0, "((a+)c(?2))b(?1)", "aacaabaca" }, + { MUA, 0, "((?2)b|(a)){2}(?1)", "aabab" }, + { MUA, 0, "(?1)(a)*+(?2)(b(?1))", "aababa" }, + { MUA, 0, "(?1)(((a(*ACCEPT)))b)", "axaa" }, + { MUA, 0, "(?1)(?(DEFINE) (((ac(*ACCEPT)))b) )", "akaac" }, + { MUA, 0, "(a+)b(?1)b\\1", "abaaabaaaaa" }, + { MUA, 0 | F_NOMATCH, "(?(DEFINE)(aa|a))(?1)ab", "aab" }, + { MUA, 0, "(?(DEFINE)(a\\Kb))(?1)+ababc", "abababxabababc" }, + { MUA, 0, "(a\\Kb)(?1)+ababc", "abababxababababc" }, + { MUA, 0 | F_NOMATCH, "(a\\Kb)(?1)+ababc", "abababxababababxc" }, + { MUA, 0, "b|<(?R)*>", "<" }, + { MUA, 0, "(a\\K){0}(?:(?1)b|ac)", "ac" }, + { MUA, 0, "(?(DEFINE)(a(?2)|b)(b(?1)|(a)))(?:(?1)|(?2))m", "ababababnababababaam" }, + { MUA, 0, "(a)((?(R)a|b))(?2)", "aabbabaa" }, + { MUA, 0, "(a)((?(R2)a|b))(?2)", "aabbabaa" }, + { MUA, 0, "(a)((?(R1)a|b))(?2)", "ababba" }, + { MUA, 0, "(?(R0)aa|bb(?R))", "abba aabb bbaa" }, + { MUA, 0, "((?(R)(?:aaaa|a)|(?:(aaaa)|(a)))+)(?1)$", "aaaaaaaaaa aaaa" }, + { MUA, 0, "(?Pa(?(R&Name)a|b))(?1)", "aab abb abaa" }, + + /* 16 bit specific tests. */ + { CMA, 0 | F_FORCECONV, "\xc3\xa1", "\xc3\x81\xc3\xa1" }, + { CMA, 0 | F_FORCECONV, "\xe1\xbd\xb8", "\xe1\xbf\xb8\xe1\xbd\xb8" }, + { CMA, 0 | F_FORCECONV, "[\xc3\xa1]", "\xc3\x81\xc3\xa1" }, + { CMA, 0 | F_FORCECONV, "[\xe1\xbd\xb8]", "\xe1\xbf\xb8\xe1\xbd\xb8" }, + { CMA, 0 | F_FORCECONV, "[a-\xed\xb0\x80]", "A" }, + { CMA, 0 | F_NO8 | F_FORCECONV, "[a-\\x{dc00}]", "B" }, + { CMA, 0 | F_NO8 | F_NOMATCH | F_FORCECONV, "[b-\\x{dc00}]", "a" }, + { CMA, 0 | F_NO8 | F_FORCECONV, "\xed\xa0\x80\\x{d800}\xed\xb0\x80\\x{dc00}", "\xed\xa0\x80\xed\xa0\x80\xed\xb0\x80\xed\xb0\x80" }, + { CMA, 0 | F_NO8 | F_FORCECONV, "[\xed\xa0\x80\\x{d800}]{1,2}?[\xed\xb0\x80\\x{dc00}]{1,2}?#", "\xed\xa0\x80\xed\xa0\x80\xed\xb0\x80\xed\xb0\x80#" }, + { CMA, 0 | F_FORCECONV, "[\xed\xa0\x80\xed\xb0\x80#]{0,3}(?<=\xed\xb0\x80.)", "\xed\xa0\x80#\xed\xa0\x80##\xed\xb0\x80\xed\xa0\x80" }, + { CMA, 0 | F_FORCECONV, "[\xed\xa0\x80-\xed\xb3\xbf]", "\xed\x9f\xbf\xed\xa0\x83" }, + { CMA, 0 | F_FORCECONV, "[\xed\xa0\x80-\xed\xb3\xbf]", "\xed\xb4\x80\xed\xb3\xb0" }, + { CMA, 0 | F_NO8 | F_FORCECONV, "[\\x{d800}-\\x{dcff}]", "\xed\x9f\xbf\xed\xa0\x83" }, + { CMA, 0 | F_NO8 | F_FORCECONV, "[\\x{d800}-\\x{dcff}]", "\xed\xb4\x80\xed\xb3\xb0" }, + { CMA, 0 | F_FORCECONV, "[\xed\xa0\x80-\xef\xbf\xbf]+[\x1-\xed\xb0\x80]+#", "\xed\xa0\x85\xc3\x81\xed\xa0\x85\xef\xbf\xb0\xc2\x85\xed\xa9\x89#" }, + { CMA, 0 | F_FORCECONV, "[\xed\xa0\x80][\xed\xb0\x80]{2,}", "\xed\xa0\x80\xed\xb0\x80\xed\xa0\x80\xed\xb0\x80\xed\xb0\x80\xed\xb0\x80" }, + { MA, 0 | F_FORCECONV, "[^\xed\xb0\x80]{3,}?", "##\xed\xb0\x80#\xed\xb0\x80#\xc3\x89#\xed\xb0\x80" }, + { MA, 0 | F_NO8 | F_FORCECONV, "[^\\x{dc00}]{3,}?", "##\xed\xb0\x80#\xed\xb0\x80#\xc3\x89#\xed\xb0\x80" }, + { CMA, 0 | F_FORCECONV, ".\\B.", "\xed\xa0\x80\xed\xb0\x80" }, + { CMA, 0 | F_FORCECONV, "\\D+(?:\\d+|.)\\S+(?:\\s+|.)\\W+(?:\\w+|.)\xed\xa0\x80\xed\xa0\x80", "\xed\xa0\x80\xed\xa0\x80\xed\xa0\x80\xed\xa0\x80\xed\xa0\x80\xed\xa0\x80\xed\xa0\x80\xed\xa0\x80" }, + { CMA, 0 | F_FORCECONV, "\\d*\\s*\\w*\xed\xa0\x80\xed\xa0\x80", "\xed\xa0\x80\xed\xa0\x80" }, + { CMA, 0 | F_FORCECONV | F_NOMATCH, "\\d*?\\D*?\\s*?\\S*?\\w*?\\W*?##", "\xed\xa0\x80\xed\xa0\x80\xed\xa0\x80\xed\xa0\x80#" }, + { CMA | PCRE_EXTENDED, 0 | F_FORCECONV, "\xed\xa0\x80 \xed\xb0\x80 !", "\xed\xa0\x80\xed\xb0\x80!" }, + { CMA, 0 | F_FORCECONV, "\xed\xa0\x80+#[^#]+\xed\xa0\x80", "\xed\xa0\x80#a\xed\xa0\x80" }, + { CMA, 0 | F_FORCECONV, "(\xed\xa0\x80+)#\\1", "\xed\xa0\x80\xed\xa0\x80#\xed\xa0\x80\xed\xa0\x80" }, + { PCRE_MULTILINE | PCRE_NEWLINE_ANY, 0 | F_NO8 | F_FORCECONV, "^-", "a--\xe2\x80\xa8--" }, + { PCRE_BSR_UNICODE, 0 | F_NO8 | F_FORCECONV, "\\R", "ab\xe2\x80\xa8" }, + { 0, 0 | F_NO8 | F_FORCECONV, "\\v", "ab\xe2\x80\xa9" }, + { 0, 0 | F_NO8 | F_FORCECONV, "\\h", "ab\xe1\xa0\x8e" }, + { 0, 0 | F_NO8 | F_FORCECONV, "\\v+?\\V+?#", "\xe2\x80\xa9\xe2\x80\xa9\xef\xbf\xbf\xef\xbf\xbf#" }, + { 0, 0 | F_NO8 | F_FORCECONV, "\\h+?\\H+?#", "\xe1\xa0\x8e\xe1\xa0\x8e\xef\xbf\xbf\xef\xbf\xbf#" }, + + /* Partial matching. */ + { MUA | PCRE_PARTIAL_SOFT, 0, "ab", "a" }, + { MUA | PCRE_PARTIAL_SOFT, 0, "ab|a", "a" }, + { MUA | PCRE_PARTIAL_HARD, 0, "ab|a", "a" }, + { MUA | PCRE_PARTIAL_SOFT, 0, "\\b#", "a" }, + { MUA | PCRE_PARTIAL_SOFT, 0, "(?<=a)b", "a" }, + { MUA | PCRE_PARTIAL_SOFT, 0, "abc|(?<=xxa)bc", "xxab" }, + { MUA | PCRE_PARTIAL_SOFT, 0, "a\\B", "a" }, + { MUA | PCRE_PARTIAL_HARD, 0, "a\\b", "a" }, + + /* (*MARK) verb. */ + { MUA, 0, "a(*MARK:aa)a", "ababaa" }, + { MUA, 0 | F_NOMATCH, "a(*:aa)a", "abab" }, + { MUA, 0, "a(*:aa)(b(*:bb)b|bc)", "abc" }, + { MUA, 0 | F_NOMATCH, "a(*:1)x|b(*:2)y", "abc" }, + { MUA, 0, "(?>a(*:aa))b|ac", "ac" }, + { MUA, 0, "(?(DEFINE)(a(*:aa)))(?1)", "a" }, + { MUA, 0 | F_NOMATCH, "(?(DEFINE)((a)(*:aa)))(?1)b", "aa" }, + { MUA, 0, "(?(DEFINE)(a(*:aa)))a(?1)b|aac", "aac" }, + { MUA, 0, "(a(*:aa)){0}(?:b(?1)b|c)+c", "babbab cc" }, + { MUA, 0, "(a(*:aa)){0}(?:b(?1)b)+", "babba" }, + { MUA, 0 | F_NOMATCH, "(a(*:aa)){0}(?:b(?1)b)+", "ba" }, + { MUA, 0, "(a\\K(*:aa)){0}(?:b(?1)b|c)+c", "babbab cc" }, + { MUA, 0, "(a\\K(*:aa)){0}(?:b(?1)b)+", "babba" }, + { MUA, 0 | F_NOMATCH, "(a\\K(*:aa)){0}(?:b(?1)b)+", "ba" }, + + /* (*COMMIT) verb. */ + { MUA, 0 | F_NOMATCH, "a(*COMMIT)b", "ac" }, + { MUA, 0, "aa(*COMMIT)b", "xaxaab" }, + { MUA, 0 | F_NOMATCH, "a(*COMMIT)(*:msg)b|ac", "ac" }, + { MUA, 0, "(?=a(*COMMIT)b|ac)ac|(*:m)(a)c", "ac" }, + { MUA, 0, "(?!a(*COMMIT)(*:msg)b)a(c)|cd", "acd" }, + + /* Deep recursion. */ + { MUA, 0, "((((?:(?:(?:\\w)+)?)*|(?>\\w)+?)+|(?>\\w)?\?)*)?\\s", "aaaaa+ " }, + { MUA, 0, "(?:((?:(?:(?:\\w*?)+)??|(?>\\w)?|\\w*+)*)+)+?\\s", "aa+ " }, + { MUA, 0, "((a?)+)+b", "aaaaaaaaaaaa b" }, + + /* Deep recursion: Stack limit reached. */ + { MA, 0 | F_NOMATCH, "a?a?a?a?a?a?a?a?a?a?a?a?a?a?a?a?a?a?a?a?a?a?a?aaaaaaaaaaaaaaaaaaaaaaa", "aaaaaaaaaaaaaaaaaaaaaaa" }, + { MA, 0 | F_NOMATCH, "(?:a+)+b", "aaaaaaaaaaaaaaaaaaaaaaaa b" }, + { MA, 0 | F_NOMATCH, "(?:a+?)+?b", "aaaaaaaaaaaaaaaaaaaaaaaa b" }, + { MA, 0 | F_NOMATCH, "(?:a*)*b", "aaaaaaaaaaaaaaaaaaaaaaaa b" }, + { MA, 0 | F_NOMATCH, "(?:a*?)*?b", "aaaaaaaaaaaaaaaaaaaaaaaa b" }, + + { 0, 0, NULL, NULL } +}; + +static const unsigned char *tables(int mode) +{ + /* The purpose of this function to allow valgrind + for reporting invalid reads and writes. */ + static unsigned char *tables_copy; + const char *errorptr; + int erroroffset; + unsigned char *default_tables; +#ifdef SUPPORT_PCRE8 + pcre *regex; + char null_str[1] = { 0 }; +#else + pcre16 *regex; + PCRE_UCHAR16 null_str[1] = { 0 }; +#endif + + if (mode) { + if (tables_copy) + free(tables_copy); + tables_copy = NULL; + return NULL; + } + + if (tables_copy) + return tables_copy; + + default_tables = NULL; +#ifdef SUPPORT_PCRE8 + regex = pcre_compile(null_str, 0, &errorptr, &erroroffset, NULL); + if (regex) { + pcre_fullinfo(regex, NULL, PCRE_INFO_DEFAULT_TABLES, &default_tables); + pcre_free(regex); + } +#else + regex = pcre16_compile(null_str, 0, &errorptr, &erroroffset, NULL); + if (regex) { + pcre16_fullinfo(regex, NULL, PCRE_INFO_DEFAULT_TABLES, &default_tables); + pcre16_free(regex); + } +#endif + /* Shouldn't ever happen. */ + if (!default_tables) + return NULL; + + /* Unfortunately this value cannot get from pcre_fullinfo. + Since this is a test program, this is acceptable at the moment. */ + tables_copy = (unsigned char *)malloc(1088); + if (!tables_copy) + return NULL; + + memcpy(tables_copy, default_tables, 1088); + return tables_copy; +} + +#ifdef SUPPORT_PCRE8 +static pcre_jit_stack* callback8(void *arg) +{ + return (pcre_jit_stack *)arg; +} +#endif + +#ifdef SUPPORT_PCRE16 +static pcre16_jit_stack* callback16(void *arg) +{ + return (pcre16_jit_stack *)arg; +} +#endif + +#ifdef SUPPORT_PCRE8 +static void setstack8(pcre_extra *extra) +{ + static pcre_jit_stack *stack; + + if (!extra) { + if (stack) + pcre_jit_stack_free(stack); + stack = NULL; + return; + } + + if (!stack) + stack = pcre_jit_stack_alloc(1, 1024 * 1024); + /* Extra can be NULL. */ + pcre_assign_jit_stack(extra, callback8, stack); +} +#endif /* SUPPORT_PCRE8 */ + +#ifdef SUPPORT_PCRE16 +static void setstack16(pcre16_extra *extra) +{ + static pcre16_jit_stack *stack; + + if (!extra) { + if (stack) + pcre16_jit_stack_free(stack); + stack = NULL; + return; + } + + if (!stack) + stack = pcre16_jit_stack_alloc(1, 1024 * 1024); + /* Extra can be NULL. */ + pcre16_assign_jit_stack(extra, callback16, stack); +} +#endif /* SUPPORT_PCRE8 */ + +#ifdef SUPPORT_PCRE16 + +static int convert_utf8_to_utf16(const char *input, PCRE_UCHAR16 *output, int *offsetmap, int max_length) +{ + unsigned char *iptr = (unsigned char*)input; + unsigned short *optr = (unsigned short *)output; + unsigned int c; + + if (max_length == 0) + return 0; + + while (*iptr && max_length > 1) { + c = 0; + if (offsetmap) + *offsetmap++ = (int)(iptr - (unsigned char*)input); + + if (!(*iptr & 0x80)) + c = *iptr++; + else if (!(*iptr & 0x20)) { + c = ((iptr[0] & 0x1f) << 6) | (iptr[1] & 0x3f); + iptr += 2; + } else if (!(*iptr & 0x10)) { + c = ((iptr[0] & 0x0f) << 12) | ((iptr[1] & 0x3f) << 6) | (iptr[2] & 0x3f); + iptr += 3; + } else if (!(*iptr & 0x08)) { + c = ((iptr[0] & 0x07) << 18) | ((iptr[1] & 0x3f) << 12) | ((iptr[2] & 0x3f) << 6) | (iptr[3] & 0x3f); + iptr += 4; + } + + if (c < 65536) { + *optr++ = c; + max_length--; + } else if (max_length <= 2) { + *optr = '\0'; + return (int)(optr - (unsigned short *)output); + } else { + c -= 0x10000; + *optr++ = 0xd800 | ((c >> 10) & 0x3ff); + *optr++ = 0xdc00 | (c & 0x3ff); + max_length -= 2; + if (offsetmap) + offsetmap++; + } + } + if (offsetmap) + *offsetmap = (int)(iptr - (unsigned char*)input); + *optr = '\0'; + return (int)(optr - (unsigned short *)output); +} + +static int copy_char8_to_char16(const char *input, PCRE_UCHAR16 *output, int max_length) +{ + unsigned char *iptr = (unsigned char*)input; + unsigned short *optr = (unsigned short *)output; + + if (max_length == 0) + return 0; + + while (*iptr && max_length > 1) { + *optr++ = *iptr++; + max_length--; + } + *optr = '\0'; + return (int)(optr - (unsigned short *)output); +} + +#define REGTEST_MAX_LENGTH 4096 +static PCRE_UCHAR16 regtest_buf[REGTEST_MAX_LENGTH]; +static int regtest_offsetmap[REGTEST_MAX_LENGTH]; + +#endif /* SUPPORT_PCRE16 */ + +static int check_ascii(const char *input) +{ + const unsigned char *ptr = (unsigned char *)input; + while (*ptr) { + if (*ptr > 127) + return 0; + ptr++; + } + return 1; +} + +static int regression_tests(void) +{ + struct regression_test_case *current = regression_test_cases; + const char *error; + char *cpu_info; + int i, err_offs; + int is_successful, is_ascii_pattern, is_ascii_input; + int total = 0; + int successful = 0; + int successful_row = 0; + int counter = 0; + int study_mode; +#ifdef SUPPORT_PCRE8 + pcre *re8; + pcre_extra *extra8; + pcre_extra dummy_extra8; + int ovector8_1[32]; + int ovector8_2[32]; + int return_value8_1, return_value8_2; + unsigned char *mark8_1, *mark8_2; + int utf8 = 0, ucp8 = 0; + int disabled_flags8 = 0; +#endif +#ifdef SUPPORT_PCRE16 + pcre16 *re16; + pcre16_extra *extra16; + pcre16_extra dummy_extra16; + int ovector16_1[32]; + int ovector16_2[32]; + int return_value16_1, return_value16_2; + PCRE_UCHAR16 *mark16_1, *mark16_2; + int utf16 = 0, ucp16 = 0; + int disabled_flags16 = 0; + int length16; +#endif + + /* This test compares the behaviour of interpreter and JIT. Although disabling + utf or ucp may make tests fail, if the pcre_exec result is the SAME, it is + still considered successful from pcre_jit_test point of view. */ + +#ifdef SUPPORT_PCRE8 + pcre_config(PCRE_CONFIG_JITTARGET, &cpu_info); +#else + pcre16_config(PCRE_CONFIG_JITTARGET, &cpu_info); +#endif + + printf("Running JIT regression tests\n"); + printf(" target CPU of SLJIT compiler: %s\n", cpu_info); + +#ifdef SUPPORT_PCRE8 + pcre_config(PCRE_CONFIG_UTF8, &utf8); + pcre_config(PCRE_CONFIG_UNICODE_PROPERTIES, &ucp8); + if (!utf8) + disabled_flags8 |= PCRE_UTF8; + if (!ucp8) + disabled_flags8 |= PCRE_UCP; + printf(" in 8 bit mode with utf8 %s and ucp %s:\n", utf8 ? "enabled" : "disabled", ucp8 ? "enabled" : "disabled"); +#endif +#ifdef SUPPORT_PCRE16 + pcre16_config(PCRE_CONFIG_UTF16, &utf16); + pcre16_config(PCRE_CONFIG_UNICODE_PROPERTIES, &ucp16); + if (!utf16) + disabled_flags16 |= PCRE_UTF8; + if (!ucp16) + disabled_flags16 |= PCRE_UCP; + printf(" in 16 bit mode with utf16 %s and ucp %s:\n", utf16 ? "enabled" : "disabled", ucp16 ? "enabled" : "disabled"); +#endif + + while (current->pattern) { + /* printf("\nPattern: %s :\n", current->pattern); */ + total++; + if (current->start_offset & F_PROPERTY) { + is_ascii_pattern = 0; + is_ascii_input = 0; + } else { + is_ascii_pattern = check_ascii(current->pattern); + is_ascii_input = check_ascii(current->input); + } + + if (current->flags & PCRE_PARTIAL_SOFT) + study_mode = PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE; + else if (current->flags & PCRE_PARTIAL_HARD) + study_mode = PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE; + else + study_mode = PCRE_STUDY_JIT_COMPILE; + error = NULL; +#ifdef SUPPORT_PCRE8 + re8 = NULL; + if (!(current->start_offset & F_NO8)) + re8 = pcre_compile(current->pattern, + current->flags & ~(PCRE_NOTBOL | PCRE_NOTEOL | PCRE_NOTEMPTY | PCRE_NOTEMPTY_ATSTART | PCRE_PARTIAL_SOFT | PCRE_PARTIAL_HARD | disabled_flags8), + &error, &err_offs, tables(0)); + + extra8 = NULL; + if (re8) { + error = NULL; + extra8 = pcre_study(re8, study_mode, &error); + if (!extra8) { + printf("\n8 bit: Cannot study pattern: %s\n", current->pattern); + pcre_free(re8); + re8 = NULL; + } + if (!(extra8->flags & PCRE_EXTRA_EXECUTABLE_JIT)) { + printf("\n8 bit: JIT compiler does not support: %s\n", current->pattern); + pcre_free_study(extra8); + pcre_free(re8); + re8 = NULL; + } + extra8->flags |= PCRE_EXTRA_MARK; + } else if (((utf8 && ucp8) || is_ascii_pattern) && !(current->start_offset & F_NO8)) + printf("\n8 bit: Cannot compile pattern: %s\n", current->pattern); +#endif +#ifdef SUPPORT_PCRE16 + if ((current->flags & PCRE_UTF8) || (current->start_offset & F_FORCECONV)) + convert_utf8_to_utf16(current->pattern, regtest_buf, NULL, REGTEST_MAX_LENGTH); + else + copy_char8_to_char16(current->pattern, regtest_buf, REGTEST_MAX_LENGTH); + + re16 = NULL; + if (!(current->start_offset & F_NO16)) + re16 = pcre16_compile(regtest_buf, + current->flags & ~(PCRE_NOTBOL | PCRE_NOTEOL | PCRE_NOTEMPTY | PCRE_NOTEMPTY_ATSTART | PCRE_PARTIAL_SOFT | PCRE_PARTIAL_HARD | disabled_flags16), + &error, &err_offs, tables(0)); + + extra16 = NULL; + if (re16) { + error = NULL; + extra16 = pcre16_study(re16, study_mode, &error); + if (!extra16) { + printf("\n16 bit: Cannot study pattern: %s\n", current->pattern); + pcre16_free(re16); + re16 = NULL; + } + if (!(extra16->flags & PCRE_EXTRA_EXECUTABLE_JIT)) { + printf("\n16 bit: JIT compiler does not support: %s\n", current->pattern); + pcre16_free_study(extra16); + pcre16_free(re16); + re16 = NULL; + } + extra16->flags |= PCRE_EXTRA_MARK; + } else if (((utf16 && ucp16) || is_ascii_pattern) && !(current->start_offset & F_NO16)) + printf("\n16 bit: Cannot compile pattern: %s\n", current->pattern); +#endif + + counter++; + if ((counter & 0x3) != 0) { +#ifdef SUPPORT_PCRE8 + setstack8(NULL); +#endif +#ifdef SUPPORT_PCRE16 + setstack16(NULL); +#endif + } + +#ifdef SUPPORT_PCRE8 + return_value8_1 = -1000; + return_value8_2 = -1000; + for (i = 0; i < 32; ++i) + ovector8_1[i] = -2; + for (i = 0; i < 32; ++i) + ovector8_2[i] = -2; + if (re8) { + mark8_1 = NULL; + mark8_2 = NULL; + setstack8(extra8); + extra8->mark = &mark8_1; + return_value8_1 = pcre_exec(re8, extra8, current->input, strlen(current->input), current->start_offset & OFFSET_MASK, + current->flags & (PCRE_NOTBOL | PCRE_NOTEOL | PCRE_NOTEMPTY | PCRE_NOTEMPTY_ATSTART | PCRE_PARTIAL_SOFT | PCRE_PARTIAL_HARD), ovector8_1, 32); + memset(&dummy_extra8, 0, sizeof(pcre_extra)); + dummy_extra8.flags = PCRE_EXTRA_MARK; + dummy_extra8.mark = &mark8_2; + return_value8_2 = pcre_exec(re8, &dummy_extra8, current->input, strlen(current->input), current->start_offset & OFFSET_MASK, + current->flags & (PCRE_NOTBOL | PCRE_NOTEOL | PCRE_NOTEMPTY | PCRE_NOTEMPTY_ATSTART | PCRE_PARTIAL_SOFT | PCRE_PARTIAL_HARD), ovector8_2, 32); + } +#endif + +#ifdef SUPPORT_PCRE16 + return_value16_1 = -1000; + return_value16_2 = -1000; + for (i = 0; i < 32; ++i) + ovector16_1[i] = -2; + for (i = 0; i < 32; ++i) + ovector16_2[i] = -2; + if (re16) { + mark16_1 = NULL; + mark16_2 = NULL; + setstack16(extra16); + if ((current->flags & PCRE_UTF8) || (current->start_offset & F_FORCECONV)) + length16 = convert_utf8_to_utf16(current->input, regtest_buf, regtest_offsetmap, REGTEST_MAX_LENGTH); + else + length16 = copy_char8_to_char16(current->input, regtest_buf, REGTEST_MAX_LENGTH); + extra16->mark = &mark16_1; + return_value16_1 = pcre16_exec(re16, extra16, regtest_buf, length16, current->start_offset & OFFSET_MASK, + current->flags & (PCRE_NOTBOL | PCRE_NOTEOL | PCRE_NOTEMPTY | PCRE_NOTEMPTY_ATSTART | PCRE_PARTIAL_SOFT | PCRE_PARTIAL_HARD), ovector16_1, 32); + memset(&dummy_extra16, 0, sizeof(pcre16_extra)); + dummy_extra16.flags = PCRE_EXTRA_MARK; + dummy_extra16.mark = &mark16_2; + return_value16_2 = pcre16_exec(re16, &dummy_extra16, regtest_buf, length16, current->start_offset & OFFSET_MASK, + current->flags & (PCRE_NOTBOL | PCRE_NOTEOL | PCRE_NOTEMPTY | PCRE_NOTEMPTY_ATSTART | PCRE_PARTIAL_SOFT | PCRE_PARTIAL_HARD), ovector16_2, 32); + } +#endif + + /* printf("[%d-%d|%d-%d|%d-%d]%s", return_value8_1, return_value16_1, ovector8_1[0], ovector8_1[1], ovector16_1[0], ovector16_1[1], (current->flags & PCRE_CASELESS) ? "C" : ""); */ + + /* If F_DIFF is set, just run the test, but do not compare the results. + Segfaults can still be captured. */ + + is_successful = 1; + if (!(current->start_offset & F_DIFF)) { +#if defined SUPPORT_PCRE8 && defined SUPPORT_PCRE16 + if (utf8 == utf16 && !(current->start_offset & F_FORCECONV)) { + /* All results must be the same. */ + if (return_value8_1 != return_value8_2 || return_value8_1 != return_value16_1 || return_value8_1 != return_value16_2) { + printf("\n8 and 16 bit: Return value differs(%d:%d:%d:%d): [%d] '%s' @ '%s'\n", + return_value8_1, return_value8_2, return_value16_1, return_value16_2, + total, current->pattern, current->input); + is_successful = 0; + } else if (return_value8_1 >= 0 || return_value8_1 == PCRE_ERROR_PARTIAL) { + if (return_value8_1 == PCRE_ERROR_PARTIAL) { + return_value8_1 = 2; + return_value16_1 = 2; + } else { + return_value8_1 *= 2; + return_value16_1 *= 2; + } + + /* Transform back the results. */ + if (current->flags & PCRE_UTF8) { + for (i = 0; i < return_value8_1; ++i) { + if (ovector16_1[i] >= 0) + ovector16_1[i] = regtest_offsetmap[ovector16_1[i]]; + if (ovector16_2[i] >= 0) + ovector16_2[i] = regtest_offsetmap[ovector16_2[i]]; + } + } + + for (i = 0; i < return_value8_1; ++i) + if (ovector8_1[i] != ovector8_2[i] || ovector8_1[i] != ovector16_1[i] || ovector8_1[i] != ovector16_2[i]) { + printf("\n8 and 16 bit: Ovector[%d] value differs(%d:%d:%d:%d): [%d] '%s' @ '%s' \n", + i, ovector8_1[i], ovector8_2[i], ovector16_1[i], ovector16_2[i], + total, current->pattern, current->input); + is_successful = 0; + } + } + } else { +#endif /* SUPPORT_PCRE8 && SUPPORT_PCRE16 */ + /* Only the 8 bit and 16 bit results must be equal. */ +#ifdef SUPPORT_PCRE8 + if (return_value8_1 != return_value8_2) { + printf("\n8 bit: Return value differs(%d:%d): [%d] '%s' @ '%s'\n", + return_value8_1, return_value8_2, total, current->pattern, current->input); + is_successful = 0; + } else if (return_value8_1 >= 0 || return_value8_1 == PCRE_ERROR_PARTIAL) { + if (return_value8_1 == PCRE_ERROR_PARTIAL) + return_value8_1 = 2; + else + return_value8_1 *= 2; + + for (i = 0; i < return_value8_1; ++i) + if (ovector8_1[i] != ovector8_2[i]) { + printf("\n8 bit: Ovector[%d] value differs(%d:%d): [%d] '%s' @ '%s'\n", + i, ovector8_1[i], ovector8_2[i], total, current->pattern, current->input); + is_successful = 0; + } + } +#endif + +#ifdef SUPPORT_PCRE16 + if (return_value16_1 != return_value16_2) { + printf("\n16 bit: Return value differs(%d:%d): [%d] '%s' @ '%s'\n", + return_value16_1, return_value16_2, total, current->pattern, current->input); + is_successful = 0; + } else if (return_value16_1 >= 0 || return_value16_1 == PCRE_ERROR_PARTIAL) { + if (return_value16_1 == PCRE_ERROR_PARTIAL) + return_value16_1 = 2; + else + return_value16_1 *= 2; + + for (i = 0; i < return_value16_1; ++i) + if (ovector16_1[i] != ovector16_2[i]) { + printf("\n16 bit: Ovector[%d] value differs(%d:%d): [%d] '%s' @ '%s'\n", + i, ovector16_1[i], ovector16_2[i], total, current->pattern, current->input); + is_successful = 0; + } + } +#endif + +#if defined SUPPORT_PCRE8 && defined SUPPORT_PCRE16 + } +#endif /* SUPPORT_PCRE8 && SUPPORT_PCRE16 */ + } + + if (is_successful) { +#ifdef SUPPORT_PCRE8 + if (!(current->start_offset & F_NO8) && ((utf8 && ucp8) || is_ascii_input)) { + if (return_value8_1 < 0 && !(current->start_offset & F_NOMATCH)) { + printf("8 bit: Test should match: [%d] '%s' @ '%s'\n", + total, current->pattern, current->input); + is_successful = 0; + } + + if (return_value8_1 >= 0 && (current->start_offset & F_NOMATCH)) { + printf("8 bit: Test should not match: [%d] '%s' @ '%s'\n", + total, current->pattern, current->input); + is_successful = 0; + } + } +#endif +#ifdef SUPPORT_PCRE16 + if (!(current->start_offset & F_NO16) && ((utf16 && ucp16) || is_ascii_input)) { + if (return_value16_1 < 0 && !(current->start_offset & F_NOMATCH)) { + printf("16 bit: Test should match: [%d] '%s' @ '%s'\n", + total, current->pattern, current->input); + is_successful = 0; + } + + if (return_value16_1 >= 0 && (current->start_offset & F_NOMATCH)) { + printf("16 bit: Test should not match: [%d] '%s' @ '%s'\n", + total, current->pattern, current->input); + is_successful = 0; + } + } +#endif + } + + if (is_successful) { +#ifdef SUPPORT_PCRE8 + if (mark8_1 != mark8_2) { + printf("8 bit: Mark value mismatch: [%d] '%s' @ '%s'\n", + total, current->pattern, current->input); + is_successful = 0; + } +#endif +#ifdef SUPPORT_PCRE16 + if (mark16_1 != mark16_2) { + printf("16 bit: Mark value mismatch: [%d] '%s' @ '%s'\n", + total, current->pattern, current->input); + is_successful = 0; + } +#endif + } + +#ifdef SUPPORT_PCRE8 + if (re8) { + pcre_free_study(extra8); + pcre_free(re8); + } +#endif +#ifdef SUPPORT_PCRE16 + if (re16) { + pcre16_free_study(extra16); + pcre16_free(re16); + } +#endif + + if (is_successful) { + successful++; + successful_row++; + printf("."); + if (successful_row >= 60) { + successful_row = 0; + printf("\n"); + } + } else + successful_row = 0; + + fflush(stdout); + current++; + } + tables(1); +#ifdef SUPPORT_PCRE8 + setstack8(NULL); +#endif +#ifdef SUPPORT_PCRE16 + setstack16(NULL); +#endif + + if (total == successful) { + printf("\nAll JIT regression tests are successfully passed.\n"); + return 0; + } else { + printf("\nSuccessful test ratio: %d%% (%d failed)\n", successful * 100 / total, total - successful); + return 1; + } +} + +/* End of pcre_jit_test.c */ diff --git a/pcre/pcre_maketables.c b/pcre/pcre_maketables.c --- a/pcre/pcre_maketables.c +++ b/pcre/pcre_maketables.c @@ -66,15 +66,12 @@ Returns: pointer to the contiguous block of data */ -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 const unsigned char * pcre_maketables(void) -#elif defined COMPILE_PCRE16 +#else const unsigned char * pcre16_maketables(void) -#elif defined COMPILE_PCRE32 -const unsigned char * -pcre32_maketables(void) #endif { unsigned char *yield, *p; @@ -98,17 +95,13 @@ for (i = 0; i < 256; i++) *p++ = islower(i)? toupper(i) : tolower(i); /* Then the character class tables. Don't try to be clever and save effort on -exclusive ones - in some locales things may be different. - -Note that the table for "space" includes everything "isspace" gives, including -VT in the default locale. This makes it work for the POSIX class [:space:]. -From release 8.34 is is also correct for Perl space, because Perl added VT at -release 5.18. - -Note also that it is possible for a character to be alnum or alpha without -being lower or upper, such as "male and female ordinals" (\xAA and \xBA) in the -fr_FR locale (at least under Debian Linux's locales as of 12/2005). So we must -test for alnum specially. */ +exclusive ones - in some locales things may be different. Note that the table +for "space" includes everything "isspace" gives, including VT in the default +locale. This makes it work for the POSIX class [:space:]. Note also that it is +possible for a character to be alnum or alpha without being lower or upper, +such as "male and female ordinals" (\xAA and \xBA) in the fr_FR locale (at +least under Debian Linux's locales as of 12/2005). So we must test for alnum +specially. */ memset(p, 0, cbit_length); for (i = 0; i < 256; i++) @@ -127,15 +120,14 @@ } p += cbit_length; -/* Finally, the character type table. In this, we used to exclude VT from the -white space chars, because Perl didn't recognize it as such for \s and for -comments within regexes. However, Perl changed at release 5.18, so PCRE changed -at release 8.34. */ +/* Finally, the character type table. In this, we exclude VT from the white +space chars, because Perl doesn't recognize it as such for \s and for comments +within regexes. */ for (i = 0; i < 256; i++) { int x = 0; - if (isspace(i)) x += ctype_space; + if (i != 0x0b && isspace(i)) x += ctype_space; if (isalpha(i)) x += ctype_letter; if (isdigit(i)) x += ctype_digit; if (isxdigit(i)) x += ctype_xdigit; diff --git a/pcre/pcre_newline.c b/pcre/pcre_newline.c --- a/pcre/pcre_newline.c +++ b/pcre/pcre_newline.c @@ -76,7 +76,7 @@ PRIV(is_newline)(PCRE_PUCHAR ptr, int type, PCRE_PUCHAR endptr, int *lenptr, BOOL utf) { -pcre_uint32 c; +int c; (void)utf; #ifdef SUPPORT_UTF if (utf) @@ -87,13 +87,11 @@ #endif /* SUPPORT_UTF */ c = *ptr; -/* Note that this function is called only for ANY or ANYCRLF. */ - if (type == NLTYPE_ANYCRLF) switch(c) { - case CHAR_LF: *lenptr = 1; return TRUE; - case CHAR_CR: *lenptr = (ptr < endptr - 1 && ptr[1] == CHAR_LF)? 2 : 1; - return TRUE; + case 0x000a: *lenptr = 1; return TRUE; /* LF */ + case 0x000d: *lenptr = (ptr < endptr - 1 && ptr[1] == 0x0a)? 2 : 1; + return TRUE; /* CR */ default: return FALSE; } @@ -101,29 +99,20 @@ else switch(c) { -#ifdef EBCDIC - case CHAR_NEL: -#endif - case CHAR_LF: - case CHAR_VT: - case CHAR_FF: *lenptr = 1; return TRUE; - - case CHAR_CR: - *lenptr = (ptr < endptr - 1 && ptr[1] == CHAR_LF)? 2 : 1; - return TRUE; - -#ifndef EBCDIC + case 0x000a: /* LF */ + case 0x000b: /* VT */ + case 0x000c: *lenptr = 1; return TRUE; /* FF */ + case 0x000d: *lenptr = (ptr < endptr - 1 && ptr[1] == 0x0a)? 2 : 1; + return TRUE; /* CR */ #ifdef COMPILE_PCRE8 - case CHAR_NEL: *lenptr = utf? 2 : 1; return TRUE; + case 0x0085: *lenptr = utf? 2 : 1; return TRUE; /* NEL */ case 0x2028: /* LS */ case 0x2029: *lenptr = 3; return TRUE; /* PS */ -#else /* COMPILE_PCRE16 || COMPILE_PCRE32 */ - case CHAR_NEL: +#else + case 0x0085: /* NEL */ case 0x2028: /* LS */ case 0x2029: *lenptr = 1; return TRUE; /* PS */ -#endif /* COMPILE_PCRE8 */ -#endif /* Not EBCDIC */ - +#endif /* COMPILE_PCRE8 */ default: return FALSE; } } @@ -151,7 +140,7 @@ PRIV(was_newline)(PCRE_PUCHAR ptr, int type, PCRE_PUCHAR startptr, int *lenptr, BOOL utf) { -pcre_uint32 c; +int c; (void)utf; ptr--; #ifdef SUPPORT_UTF @@ -164,45 +153,30 @@ #endif /* SUPPORT_UTF */ c = *ptr; -/* Note that this function is called only for ANY or ANYCRLF. */ - if (type == NLTYPE_ANYCRLF) switch(c) { - case CHAR_LF: - *lenptr = (ptr > startptr && ptr[-1] == CHAR_CR)? 2 : 1; - return TRUE; - - case CHAR_CR: *lenptr = 1; return TRUE; + case 0x000a: *lenptr = (ptr > startptr && ptr[-1] == 0x0d)? 2 : 1; + return TRUE; /* LF */ + case 0x000d: *lenptr = 1; return TRUE; /* CR */ default: return FALSE; } -/* NLTYPE_ANY */ - else switch(c) { - case CHAR_LF: - *lenptr = (ptr > startptr && ptr[-1] == CHAR_CR)? 2 : 1; - return TRUE; - -#ifdef EBCDIC - case CHAR_NEL: -#endif - case CHAR_VT: - case CHAR_FF: - case CHAR_CR: *lenptr = 1; return TRUE; - -#ifndef EBCDIC + case 0x000a: *lenptr = (ptr > startptr && ptr[-1] == 0x0d)? 2 : 1; + return TRUE; /* LF */ + case 0x000b: /* VT */ + case 0x000c: /* FF */ + case 0x000d: *lenptr = 1; return TRUE; /* CR */ #ifdef COMPILE_PCRE8 - case CHAR_NEL: *lenptr = utf? 2 : 1; return TRUE; - case 0x2028: /* LS */ - case 0x2029: *lenptr = 3; return TRUE; /* PS */ -#else /* COMPILE_PCRE16 || COMPILE_PCRE32 */ - case CHAR_NEL: + case 0x0085: *lenptr = utf? 2 : 1; return TRUE; /* NEL */ + case 0x2028: /* LS */ + case 0x2029: *lenptr = 3; return TRUE; /* PS */ +#else + case 0x0085: /* NEL */ case 0x2028: /* LS */ case 0x2029: *lenptr = 1; return TRUE; /* PS */ -#endif /* COMPILE_PCRE8 */ -#endif /* NotEBCDIC */ - +#endif /* COMPILE_PCRE8 */ default: return FALSE; } } diff --git a/pcre/pcre_ord2utf8.c b/pcre/pcre_ord2utf8.c --- a/pcre/pcre_ord2utf8.c +++ b/pcre/pcre_ord2utf8.c @@ -45,16 +45,15 @@ #include "config.h" #endif -#define COMPILE_PCRE8 - #include "pcre_internal.h" + /************************************************* * Convert character value to UTF-8 * *************************************************/ /* This function takes an integer value in the range 0 - 0x10ffff -and encodes it as a UTF-8 character in 1 to 4 pcre_uchars. +and encodes it as a UTF-8 character in 1 to 6 pcre_uchars. Arguments: cvalue the character value @@ -63,13 +62,17 @@ Returns: number of characters placed in the buffer */ -unsigned int PRIV(ord2utf)(pcre_uint32 cvalue, pcre_uchar *buffer) { #ifdef SUPPORT_UTF register int i, j; + +/* Checking invalid cvalue character, encoded as invalid UTF-16 character. +Should never happen in practice. */ +if ((cvalue & 0xf800) == 0xd800 || cvalue >= 0x110000) + cvalue = 0xfffe; for (i = 0; i < PRIV(utf8_table1_size); i++) if ((int)cvalue <= PRIV(utf8_table1)[i]) break; diff --git a/pcre/pcre_printint.c b/pcre/pcre_printint.c --- a/pcre/pcre_printint.c +++ b/pcre/pcre_printint.c @@ -78,13 +78,10 @@ #ifdef PCRE_INCLUDED static /* Keep the following function as private. */ #endif - -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 void pcre_printint(pcre *external_re, FILE *f, BOOL print_lengths); -#elif defined COMPILE_PCRE16 +#else void pcre16_printint(pcre *external_re, FILE *f, BOOL print_lengths); -#elif defined COMPILE_PCRE32 -void pcre32_printint(pcre *external_re, FILE *f, BOOL print_lengths); #endif /* Macro that decides whether a character should be output as a literal or in @@ -114,28 +111,26 @@ * Print single- or multi-byte character * *************************************************/ -static unsigned int +static int print_char(FILE *f, pcre_uchar *ptr, BOOL utf) { -pcre_uint32 c = *ptr; +int c = *ptr; #ifndef SUPPORT_UTF (void)utf; /* Avoid compiler warning */ -if (PRINTABLE(c)) fprintf(f, "%c", (char)c); -else if (c <= 0x80) fprintf(f, "\\x%02x", c); +if (PRINTABLE(c)) fprintf(f, "%c", c); +else if (c <= 0xff) fprintf(f, "\\x%02x", c); else fprintf(f, "\\x{%x}", c); return 0; #else -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 if (!utf || (c & 0xc0) != 0xc0) { - if (PRINTABLE(c)) fprintf(f, "%c", (char)c); - else if (c < 0x80) fprintf(f, "\\x%02x", c); - else fprintf(f, "\\x{%02x}", c); + if (PRINTABLE(c)) fprintf(f, "%c", c); else fprintf(f, "\\x%02x", c); return 0; } else @@ -165,13 +160,15 @@ return a; } -#elif defined COMPILE_PCRE16 +#else +#ifdef COMPILE_PCRE16 + if (!utf || (c & 0xfc00) != 0xd800) { - if (PRINTABLE(c)) fprintf(f, "%c", (char)c); - else if (c <= 0x80) fprintf(f, "\\x%02x", c); - else fprintf(f, "\\x{%02x}", c); + if (PRINTABLE(c)) fprintf(f, "%c", c); + else if (c <= 0xff) fprintf(f, "\\x%02x", c); + else fprintf(f, "\\x{%x}", c); return 0; } else @@ -191,25 +188,9 @@ return 1; } -#elif defined COMPILE_PCRE32 - -if (!utf || (c & 0xfffff800u) != 0xd800u) - { - if (PRINTABLE(c)) fprintf(f, "%c", (char)c); - else if (c <= 0x80) fprintf(f, "\\x%02x", c); - else fprintf(f, "\\x{%x}", c); - return 0; - } -else - { - /* This is a check for malformed UTF-32; it should only occur if the sanity - check has been turned off. Rather than swallow a surrogate, just stop if - we hit one. Print it with \X instead of \x as an indication. */ - fprintf(f, "\\X{%x}", c); - return 0; - } +#endif /* COMPILE_PCRE16 */ -#endif /* COMPILE_PCRE[8|16|32] */ +#endif /* COMPILE_PCRE8 */ #endif /* SUPPORT_UTF */ } @@ -223,7 +204,7 @@ { while (*ptr != '\0') { - register pcre_uint32 c = *ptr++; + register int c = *ptr++; if (PRINTABLE(c)) fprintf(f, "%c", c); else fprintf(f, "\\x{%x}", c); } } @@ -233,7 +214,7 @@ *************************************************/ static const char * -get_ucpname(unsigned int ptype, unsigned int pvalue) +get_ucpname(int ptype, int pvalue) { #ifdef SUPPORT_UCP int i; @@ -250,41 +231,7 @@ } -/************************************************* -* Print Unicode property value * -*************************************************/ -/* "Normal" properties can be printed from tables. The PT_CLIST property is a -pseudo-property that contains a pointer to a list of case-equivalent -characters. This is used only when UCP support is available and UTF mode is -selected. It should never occur otherwise, but just in case it does, have -something ready to print. */ - -static void -print_prop(FILE *f, pcre_uchar *code, const char *before, const char *after) -{ -if (code[1] != PT_CLIST) - { - fprintf(f, "%s%s %s%s", before, priv_OP_names[*code], get_ucpname(code[1], - code[2]), after); - } -else - { - const char *not = (*code == OP_PROP)? "" : "not "; -#ifndef SUPPORT_UCP - fprintf(f, "%s%sclist %d%s", before, not, code[2], after); -#else - const pcre_uint32 *p = PRIV(ucd_caseless_sets) + code[2]; - fprintf (f, "%s%sclist", before, not); - while (*p < NOTACHAR) fprintf(f, " %04x", *p++); - fprintf(f, "%s", after); -#endif - } -} - - - - /************************************************* * Print compiled regex * *************************************************/ @@ -298,15 +245,12 @@ #ifdef PCRE_INCLUDED static /* Keep the following function as private. */ #endif -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 void pcre_printint(pcre *external_re, FILE *f, BOOL print_lengths) -#elif defined COMPILE_PCRE16 +#else void pcre16_printint(pcre *external_re, FILE *f, BOOL print_lengths) -#elif defined COMPILE_PCRE32 -void -pcre32_printint(pcre *external_re, FILE *f, BOOL print_lengths) #endif { REAL_PCRE *re = (REAL_PCRE *)external_re; @@ -330,15 +274,15 @@ } code = codestart = (pcre_uchar *)re + offset + count * size; -/* PCRE_UTF(16|32) have the same value as PCRE_UTF8. */ +/* PCRE_UTF16 has the same value as PCRE_UTF8. */ utf = (options & PCRE_UTF8) != 0; for(;;) { pcre_uchar *ccode; const char *flag = " "; - pcre_uint32 c; - unsigned int extra = 0; + int c; + int extra = 0; if (print_lengths) fprintf(f, "%3d ", (int)(code - codestart)); @@ -425,19 +369,10 @@ break; case OP_CREF: + case OP_NCREF: fprintf(f, "%3d %s", GET2(code,1), priv_OP_names[*code]); break; - case OP_DNCREF: - { - pcre_uchar *entry = (pcre_uchar *)re + offset + (GET2(code, 1) * size) + - IMM2_SIZE; - fprintf(f, " %s Cond ref <", flag); - print_puchar(f, entry); - fprintf(f, ">%d", GET2(code, 1 + IMM2_SIZE)); - } - break; - case OP_RREF: c = GET2(code, 1); if (c == RREF_ANY) @@ -446,14 +381,12 @@ fprintf(f, " Cond recurse %d", c); break; - case OP_DNRREF: - { - pcre_uchar *entry = (pcre_uchar *)re + offset + (GET2(code, 1) * size) + - IMM2_SIZE; - fprintf(f, " %s Cond recurse <", flag); - print_puchar(f, entry); - fprintf(f, ">%d", GET2(code, 1 + IMM2_SIZE)); - } + case OP_NRREF: + c = GET2(code, 1); + if (c == RREF_ANY) + fprintf(f, " Cond nrecurse any"); + else + fprintf(f, " Cond nrecurse %d", c); break; case OP_DEF: @@ -492,12 +425,12 @@ fprintf(f, " %s ", flag); if (*code >= OP_TYPESTAR) { + fprintf(f, "%s", priv_OP_names[code[1]]); if (code[1] == OP_PROP || code[1] == OP_NOTPROP) { - print_prop(f, code + 1, "", " "); + fprintf(f, " %s ", get_ucpname(code[2], code[3])); extra = 2; } - else fprintf(f, "%s", priv_OP_names[code[1]]); } else extra = print_char(f, code+1, utf); fprintf(f, "%s", priv_OP_names[*code]); @@ -526,12 +459,13 @@ case OP_TYPEUPTO: case OP_TYPEMINUPTO: case OP_TYPEPOSUPTO: + fprintf(f, " %s", priv_OP_names[code[1 + IMM2_SIZE]]); if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP) { - print_prop(f, code + IMM2_SIZE + 1, " ", " "); + fprintf(f, " %s ", get_ucpname(code[1 + IMM2_SIZE + 1], + code[1 + IMM2_SIZE + 2])); extra = 2; } - else fprintf(f, " %s", priv_OP_names[code[1 + IMM2_SIZE]]); fprintf(f, "{"); if (*code != OP_TYPEEXACT) fprintf(f, "0,"); fprintf(f, "%d}", GET2(code,1)); @@ -609,20 +543,6 @@ ccode = code + priv_OP_lengths[*code]; goto CLASS_REF_REPEAT; - case OP_DNREFI: - flag = "/i"; - /* Fall through */ - case OP_DNREF: - { - pcre_uchar *entry = (pcre_uchar *)re + offset + (GET2(code, 1) * size) + - IMM2_SIZE; - fprintf(f, " %s \\k<", flag); - print_puchar(f, entry); - fprintf(f, ">%d", GET2(code, 1 + IMM2_SIZE)); - } - ccode = code + priv_OP_lengths[*code]; - goto CLASS_REF_REPEAT; - case OP_CALLOUT: fprintf(f, " %s %d %d %d", priv_OP_names[*code], code[1], GET(code,2), GET(code, 2 + LINK_SIZE)); @@ -630,19 +550,18 @@ case OP_PROP: case OP_NOTPROP: - print_prop(f, code, " ", ""); + fprintf(f, " %s %s", priv_OP_names[*code], get_ucpname(code[1], code[2])); break; - /* OP_XCLASS cannot occur in 8-bit, non-UTF mode. However, there's no harm - in having this code always here, and it makes it less messy without all - those #ifdefs. */ + /* OP_XCLASS can only occur in UTF or PCRE16 modes. However, there's no + harm in having this code always here, and it makes it less messy without + all those #ifdefs. */ case OP_CLASS: case OP_NCLASS: case OP_XCLASS: { - int i; - unsigned int min, max; + int i, min, max; BOOL printmap; pcre_uint8 *map; @@ -693,54 +612,29 @@ if (*code == OP_XCLASS) { - pcre_uchar ch; + int ch; while ((ch = *ccode++) != XCL_END) { - BOOL not = FALSE; - const char *notch = ""; - - switch(ch) + if (ch == XCL_PROP) { - case XCL_NOTPROP: - not = TRUE; - notch = "^"; - /* Fall through */ - - case XCL_PROP: - { - unsigned int ptype = *ccode++; - unsigned int pvalue = *ccode++; - - switch(ptype) - { - case PT_PXGRAPH: - fprintf(f, "[:%sgraph:]", notch); - break; - - case PT_PXPRINT: - fprintf(f, "[:%sprint:]", notch); - break; - - case PT_PXPUNCT: - fprintf(f, "[:%spunct:]", notch); - break; - - default: - fprintf(f, "\\%c{%s}", (not? 'P':'p'), - get_ucpname(ptype, pvalue)); - break; - } - } - break; - - default: + int ptype = *ccode++; + int pvalue = *ccode++; + fprintf(f, "\\p{%s}", get_ucpname(ptype, pvalue)); + } + else if (ch == XCL_NOTPROP) + { + int ptype = *ccode++; + int pvalue = *ccode++; + fprintf(f, "\\P{%s}", get_ucpname(ptype, pvalue)); + } + else + { ccode += 1 + print_char(f, ccode, utf); if (ch == XCL_RANGE) { fprintf(f, "-"); ccode += 1 + print_char(f, ccode, utf); } - break; } } } @@ -760,22 +654,17 @@ case OP_CRMINPLUS: case OP_CRQUERY: case OP_CRMINQUERY: - case OP_CRPOSSTAR: - case OP_CRPOSPLUS: - case OP_CRPOSQUERY: fprintf(f, "%s", priv_OP_names[*ccode]); extra += priv_OP_lengths[*ccode]; break; case OP_CRRANGE: case OP_CRMINRANGE: - case OP_CRPOSRANGE: min = GET2(ccode,1); max = GET2(ccode,1 + IMM2_SIZE); - if (max == 0) fprintf(f, "{%u,}", min); - else fprintf(f, "{%u,%u}", min, max); + if (max == 0) fprintf(f, "{%d,}", min); + else fprintf(f, "{%d,%d}", min, max); if (*ccode == OP_CRMINRANGE) fprintf(f, "?"); - else if (*ccode == OP_CRPOSRANGE) fprintf(f, "+"); extra += priv_OP_lengths[*ccode]; break; diff --git a/pcre/pcre_refcount.c b/pcre/pcre_refcount.c --- a/pcre/pcre_refcount.c +++ b/pcre/pcre_refcount.c @@ -68,15 +68,12 @@ a negative error number */ -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre_refcount(pcre *argument_re, int adjust) -#elif defined COMPILE_PCRE16 +#else PCRE_EXP_DEFN int PCRE_CALL_CONVENTION pcre16_refcount(pcre16 *argument_re, int adjust) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DEFN int PCRE_CALL_CONVENTION -pcre32_refcount(pcre32 *argument_re, int adjust) #endif { REAL_PCRE *re = (REAL_PCRE *)argument_re; diff --git a/pcre/pcre_string_utils.c b/pcre/pcre_string_utils.c --- a/pcre/pcre_string_utils.c +++ b/pcre/pcre_string_utils.c @@ -6,7 +6,7 @@ and semantics are as close as possible to those of the Perl 5 language. Written by Philip Hazel - Copyright (c) 1997-2013 University of Cambridge + Copyright (c) 1997-2012 University of Cambridge ----------------------------------------------------------------------------- Redistribution and use in source and binary forms, with or without @@ -38,8 +38,8 @@ */ -/* This module contains internal functions for comparing and finding the length -of strings for different data item sizes. */ +/* This module contains an internal function that is used to match an extended +class. It is used by both pcre_exec() and pcre_def_exec(). */ #ifdef HAVE_CONFIG_H @@ -54,7 +54,7 @@ * Compare string utilities * *************************************************/ -/* The following two functions compares two strings. Basically a strcmp +/* The following two functions compares two strings. Basically an strcmp for non 8 bit characters. Arguments: @@ -81,28 +81,7 @@ return 0; } -#ifdef COMPILE_PCRE32 - int -PRIV(strcmp_uc_uc_utf)(const pcre_uchar *str1, const pcre_uchar *str2) -{ -pcre_uchar c1; -pcre_uchar c2; - -while (*str1 != '\0' || *str2 != '\0') - { - c1 = RAWUCHARINC(str1); - c2 = RAWUCHARINC(str2); - if (c1 != c2) - return ((c1 > c2) << 1) - 1; - } -/* Both length and characters must be equal. */ -return 0; -} - -#endif /* COMPILE_PCRE32 */ - -int PRIV(strcmp_uc_c8)(const pcre_uchar *str1, const char *str2) { const pcre_uint8 *ustr2 = (pcre_uint8 *)str2; @@ -120,28 +99,6 @@ return 0; } -#ifdef COMPILE_PCRE32 - -int -PRIV(strcmp_uc_c8_utf)(const pcre_uchar *str1, const char *str2) -{ -const pcre_uint8 *ustr2 = (pcre_uint8 *)str2; -pcre_uchar c1; -pcre_uchar c2; - -while (*str1 != '\0' || *ustr2 != '\0') - { - c1 = RAWUCHARINC(str1); - c2 = (pcre_uchar)*ustr2++; - if (c1 != c2) - return ((c1 > c2) << 1) - 1; - } -/* Both length and characters must be equal. */ -return 0; -} - -#endif /* COMPILE_PCRE32 */ - /* The following two functions compares two, fixed length strings. Basically an strncmp for non 8 bit characters. @@ -206,6 +163,6 @@ return len; } -#endif /* !COMPILE_PCRE8 */ +#endif /* COMPILE_PCRE8 */ /* End of pcre_string_utils.c */ diff --git a/pcre/pcre_study.c b/pcre/pcre_study.c --- a/pcre/pcre_study.c +++ b/pcre/pcre_study.c @@ -66,9 +66,8 @@ rather than bytes. Arguments: - re compiled pattern block code pointer to start of group (the bracket) - startcode pointer to start of the whole pattern's code + startcode pointer to start of the whole pattern options the compiling options int RECURSE depth @@ -79,8 +78,8 @@ */ static int -find_minlength(const REAL_PCRE *re, const pcre_uchar *code, - const pcre_uchar *startcode, int options, int recurse_depth) +find_minlength(const pcre_uchar *code, const pcre_uchar *startcode, int options, + int recurse_depth) { int length = -1; /* PCRE_UTF16 has the same value as PCRE_UTF8. */ @@ -99,7 +98,7 @@ { int d, min; pcre_uchar *cs, *ce; - register pcre_uchar op = *cc; + register int op = *cc; switch (op) { @@ -130,7 +129,7 @@ case OP_SBRAPOS: case OP_ONCE: case OP_ONCE_NC: - d = find_minlength(re, cc, startcode, options, recurse_depth); + d = find_minlength(cc, startcode, options, recurse_depth); if (d < 0) return d; branchlength += d; do cc += GET(cc, 1); while (*cc == OP_ALT); @@ -176,9 +175,9 @@ case OP_REVERSE: case OP_CREF: - case OP_DNCREF: + case OP_NCREF: case OP_RREF: - case OP_DNRREF: + case OP_NRREF: case OP_DEF: case OP_CALLOUT: case OP_SOD: @@ -324,25 +323,20 @@ /* Check a class for variable quantification */ +#if defined SUPPORT_UTF || !defined COMPILE_PCRE8 + case OP_XCLASS: + cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS]; + /* Fall through */ +#endif + case OP_CLASS: case OP_NCLASS: -#if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - case OP_XCLASS: - /* The original code caused an unsigned overflow in 64 bit systems, - so now we use a conditional statement. */ - if (op == OP_XCLASS) - cc += GET(cc, 1); - else - cc += PRIV(OP_lengths)[OP_CLASS]; -#else cc += PRIV(OP_lengths)[OP_CLASS]; -#endif switch (*cc) { case OP_CRPLUS: case OP_CRMINPLUS: - case OP_CRPOSPLUS: branchlength++; /* Fall through */ @@ -350,14 +344,11 @@ case OP_CRMINSTAR: case OP_CRQUERY: case OP_CRMINQUERY: - case OP_CRPOSSTAR: - case OP_CRPOSQUERY: cc++; break; case OP_CRRANGE: case OP_CRMINRANGE: - case OP_CRPOSRANGE: branchlength += GET2(cc,1); cc += 1 + 2 * IMM2_SIZE; break; @@ -380,38 +371,7 @@ matches an empty string (by default it causes a matching failure), so in that case we must set the minimum length to zero. */ - case OP_DNREF: /* Duplicate named pattern back reference */ - case OP_DNREFI: - if ((options & PCRE_JAVASCRIPT_COMPAT) == 0) - { - int count = GET2(cc, 1+IMM2_SIZE); - pcre_uchar *slot = (pcre_uchar *)re + - re->name_table_offset + GET2(cc, 1) * re->name_entry_size; - d = INT_MAX; - while (count-- > 0) - { - ce = cs = (pcre_uchar *)PRIV(find_bracket)(startcode, utf, GET2(slot, 0)); - if (cs == NULL) return -2; - do ce += GET(ce, 1); while (*ce == OP_ALT); - if (cc > cs && cc < ce) - { - d = 0; - had_recurse = TRUE; - break; - } - else - { - int dd = find_minlength(re, cs, startcode, options, recurse_depth); - if (dd < d) d = dd; - } - slot += re->name_entry_size; - } - } - else d = 0; - cc += 1 + 2*IMM2_SIZE; - goto REPEAT_BACK_REFERENCE; - - case OP_REF: /* Single back reference */ + case OP_REF: case OP_REFI: if ((options & PCRE_JAVASCRIPT_COMPAT) == 0) { @@ -425,7 +385,7 @@ } else { - d = find_minlength(re, cs, startcode, options, recurse_depth); + d = find_minlength(cs, startcode, options, recurse_depth); } } else d = 0; @@ -433,29 +393,24 @@ /* Handle repeated back references */ - REPEAT_BACK_REFERENCE: switch (*cc) { case OP_CRSTAR: case OP_CRMINSTAR: case OP_CRQUERY: case OP_CRMINQUERY: - case OP_CRPOSSTAR: - case OP_CRPOSQUERY: min = 0; cc++; break; case OP_CRPLUS: case OP_CRMINPLUS: - case OP_CRPOSPLUS: min = 1; cc++; break; case OP_CRRANGE: case OP_CRMINRANGE: - case OP_CRPOSRANGE: min = GET2(cc, 1); cc += 1 + 2 * IMM2_SIZE; break; @@ -478,8 +433,7 @@ had_recurse = TRUE; else { - branchlength += find_minlength(re, cs, startcode, options, - recurse_depth + 1); + branchlength += find_minlength(cs, startcode, options, recurse_depth + 1); } cc += 1 + LINK_SIZE; break; @@ -584,7 +538,7 @@ p points to the character caseless the caseless flag cd the block with char table pointers - utf TRUE for UTF-8 / UTF-16 / UTF-32 mode + utf TRUE for UTF-8 / UTF-16 mode Returns: pointer after the character */ @@ -593,7 +547,7 @@ set_table_bit(pcre_uint8 *start_bits, const pcre_uchar *p, BOOL caseless, compile_data *cd, BOOL utf) { -pcre_uint32 c = *p; +unsigned int c = *p; #ifdef COMPILE_PCRE8 SET_BIT(c); @@ -610,20 +564,18 @@ (void)PRIV(ord2utf)(c, buff); SET_BIT(buff[0]); } -#endif /* Not SUPPORT_UCP */ +#endif return p; } -#else /* Not SUPPORT_UTF */ -(void)(utf); /* Stops warning for unused parameter */ -#endif /* SUPPORT_UTF */ +#endif /* Not UTF-8 mode, or character is less than 127. */ if (caseless && (cd->ctypes[c] & ctype_letter) != 0) SET_BIT(cd->fcc[c]); return p + 1; -#endif /* COMPILE_PCRE8 */ +#endif -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 +#ifdef COMPILE_PCRE16 if (c > 0xff) { c = 0xff; @@ -643,12 +595,10 @@ c = 0xff; SET_BIT(c); } -#endif /* SUPPORT_UCP */ +#endif return p; } -#else /* Not SUPPORT_UTF */ -(void)(utf); /* Stops warning for unused parameter */ -#endif /* SUPPORT_UTF */ +#endif if (caseless && (cd->ctypes[c] & ctype_letter) != 0) SET_BIT(cd->fcc[c]); return p + 1; @@ -678,10 +628,10 @@ */ static void -set_type_bits(pcre_uint8 *start_bits, int cbit_type, unsigned int table_limit, +set_type_bits(pcre_uint8 *start_bits, int cbit_type, int table_limit, compile_data *cd) { -register pcre_uint32 c; +register int c; for (c = 0; c < table_limit; c++) start_bits[c] |= cd->cbits[c+cbit_type]; #if defined SUPPORT_UTF && defined COMPILE_PCRE8 if (table_limit == 32) return; @@ -720,10 +670,10 @@ */ static void -set_nottype_bits(pcre_uint8 *start_bits, int cbit_type, unsigned int table_limit, +set_nottype_bits(pcre_uint8 *start_bits, int cbit_type, int table_limit, compile_data *cd) { -register pcre_uint32 c; +register int c; for (c = 0; c < table_limit; c++) start_bits[c] |= ~cd->cbits[c+cbit_type]; #if defined SUPPORT_UTF && defined COMPILE_PCRE8 if (table_limit != 32) for (c = 24; c < 32; c++) start_bits[c] = 0xff; @@ -747,7 +697,7 @@ Arguments: code points to an expression start_bits points to a 32-byte table, initialized to 0 - utf TRUE if in UTF-8 / UTF-16 / UTF-32 mode + utf TRUE if in UTF-8 / UTF-16 mode cd the block with char table pointers Returns: SSB_FAIL => Failed to find any starting bytes @@ -760,7 +710,7 @@ set_start_bits(const pcre_uchar *code, pcre_uint8 *start_bits, BOOL utf, compile_data *cd) { -register pcre_uint32 c; +register int c; int yield = SSB_DONE; #if defined SUPPORT_UTF && defined COMPILE_PCRE8 int table_limit = utf? 16:32; @@ -820,10 +770,6 @@ case OP_COND: case OP_CREF: case OP_DEF: - case OP_DNCREF: - case OP_DNREF: - case OP_DNREFI: - case OP_DNRREF: case OP_DOLL: case OP_DOLLM: case OP_END: @@ -832,6 +778,7 @@ case OP_EXTUNI: case OP_FAIL: case OP_MARK: + case OP_NCREF: case OP_NOT: case OP_NOTEXACT: case OP_NOTEXACTI: @@ -863,6 +810,7 @@ case OP_NOTUPTOI: case OP_NOT_HSPACE: case OP_NOT_VSPACE: + case OP_NRREF: case OP_PROP: case OP_PRUNE: case OP_PRUNE_ARG: @@ -1038,8 +986,8 @@ identical. */ case OP_HSPACE: - SET_BIT(CHAR_HT); - SET_BIT(CHAR_SPACE); + SET_BIT(0x09); + SET_BIT(0x20); #ifdef SUPPORT_UTF if (utf) { @@ -1048,46 +996,46 @@ SET_BIT(0xE1); /* For U+1680, U+180E */ SET_BIT(0xE2); /* For U+2000 - U+200A, U+202F, U+205F */ SET_BIT(0xE3); /* For U+3000 */ -#elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32 +#endif +#ifdef COMPILE_PCRE16 SET_BIT(0xA0); SET_BIT(0xFF); /* For characters > 255 */ -#endif /* COMPILE_PCRE[8|16|32] */ +#endif } else #endif /* SUPPORT_UTF */ { -#ifndef EBCDIC SET_BIT(0xA0); -#endif /* Not EBCDIC */ -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 +#ifdef COMPILE_PCRE16 SET_BIT(0xFF); /* For characters > 255 */ -#endif /* COMPILE_PCRE[16|32] */ +#endif } try_next = FALSE; break; case OP_ANYNL: case OP_VSPACE: - SET_BIT(CHAR_LF); - SET_BIT(CHAR_VT); - SET_BIT(CHAR_FF); - SET_BIT(CHAR_CR); + SET_BIT(0x0A); + SET_BIT(0x0B); + SET_BIT(0x0C); + SET_BIT(0x0D); #ifdef SUPPORT_UTF if (utf) { #ifdef COMPILE_PCRE8 SET_BIT(0xC2); /* For U+0085 */ SET_BIT(0xE2); /* For U+2028, U+2029 */ -#elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - SET_BIT(CHAR_NEL); +#endif +#ifdef COMPILE_PCRE16 + SET_BIT(0x85); SET_BIT(0xFF); /* For characters > 255 */ -#endif /* COMPILE_PCRE[8|16|32] */ +#endif } else #endif /* SUPPORT_UTF */ { - SET_BIT(CHAR_NEL); -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 + SET_BIT(0x85); +#ifdef COMPILE_PCRE16 SET_BIT(0xFF); /* For characters > 255 */ #endif } @@ -1110,8 +1058,7 @@ break; /* The cbit_space table has vertical tab as whitespace; we have to - ensure it is set as not whitespace. Luckily, the code value is the same - (0x0b) in ASCII and EBCDIC, so we can just adjust the appropriate bit. */ + ensure it is set as not whitespace. */ case OP_NOT_WHITESPACE: set_nottype_bits(start_bits, cbit_space, table_limit, cd); @@ -1119,9 +1066,8 @@ try_next = FALSE; break; - /* The cbit_space table has vertical tab as whitespace; we have to not - set it from the table. Luckily, the code value is the same (0x0b) in - ASCII and EBCDIC, so we can just adjust the appropriate bit. */ + /* The cbit_space table has vertical tab as whitespace; we have to + not set it from the table. */ case OP_WHITESPACE: c = start_bits[1]; /* Save in case it was already set */ @@ -1175,8 +1121,8 @@ return SSB_FAIL; case OP_HSPACE: - SET_BIT(CHAR_HT); - SET_BIT(CHAR_SPACE); + SET_BIT(0x09); + SET_BIT(0x20); #ifdef SUPPORT_UTF if (utf) { @@ -1185,38 +1131,38 @@ SET_BIT(0xE1); /* For U+1680, U+180E */ SET_BIT(0xE2); /* For U+2000 - U+200A, U+202F, U+205F */ SET_BIT(0xE3); /* For U+3000 */ -#elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32 +#endif +#ifdef COMPILE_PCRE16 SET_BIT(0xA0); SET_BIT(0xFF); /* For characters > 255 */ -#endif /* COMPILE_PCRE[8|16|32] */ +#endif } else #endif /* SUPPORT_UTF */ -#ifndef EBCDIC SET_BIT(0xA0); -#endif /* Not EBCDIC */ break; case OP_ANYNL: case OP_VSPACE: - SET_BIT(CHAR_LF); - SET_BIT(CHAR_VT); - SET_BIT(CHAR_FF); - SET_BIT(CHAR_CR); + SET_BIT(0x0A); + SET_BIT(0x0B); + SET_BIT(0x0C); + SET_BIT(0x0D); #ifdef SUPPORT_UTF if (utf) { #ifdef COMPILE_PCRE8 SET_BIT(0xC2); /* For U+0085 */ SET_BIT(0xE2); /* For U+2028, U+2029 */ -#elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32 - SET_BIT(CHAR_NEL); +#endif +#ifdef COMPILE_PCRE16 + SET_BIT(0x85); SET_BIT(0xFF); /* For characters > 255 */ -#endif /* COMPILE_PCRE16 */ +#endif } else #endif /* SUPPORT_UTF */ - SET_BIT(CHAR_NEL); + SET_BIT(0x85); break; case OP_NOT_DIGIT: @@ -1227,16 +1173,21 @@ set_type_bits(start_bits, cbit_digit, table_limit, cd); break; - /* The cbit_space table has vertical tab as whitespace; we no longer - have to play fancy tricks because Perl added VT to its whitespace at - release 5.18. PCRE added it at release 8.34. */ + /* The cbit_space table has vertical tab as whitespace; we have to + ensure it gets set as not whitespace. */ case OP_NOT_WHITESPACE: set_nottype_bits(start_bits, cbit_space, table_limit, cd); + start_bits[1] |= 0x08; break; + /* The cbit_space table has vertical tab as whitespace; we have to + avoid setting it. */ + case OP_WHITESPACE: + c = start_bits[1]; /* Save in case it was already set */ set_type_bits(start_bits, cbit_space, table_limit, cd); + start_bits[1] = (start_bits[1] & ~0x08) | c; break; case OP_NOT_WORDCHAR: @@ -1265,7 +1216,7 @@ memset(start_bits+25, 0xff, 7); /* Bits for 0xc9 - 0xff */ } #endif -#if defined COMPILE_PCRE16 || defined COMPILE_PCRE32 +#ifdef COMPILE_PCRE16 SET_BIT(0xFF); /* For characters > 255 */ #endif /* Fall through */ @@ -1313,14 +1264,11 @@ case OP_CRMINSTAR: case OP_CRQUERY: case OP_CRMINQUERY: - case OP_CRPOSSTAR: - case OP_CRPOSQUERY: tcode++; break; case OP_CRRANGE: case OP_CRMINRANGE: - case OP_CRPOSRANGE: if (GET2(tcode, 1) == 0) tcode += 1 + 2 * IMM2_SIZE; else try_next = FALSE; break; @@ -1364,15 +1312,12 @@ NULL on error or if no optimization possible */ -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 PCRE_EXP_DEFN pcre_extra * PCRE_CALL_CONVENTION pcre_study(const pcre *external_re, int options, const char **errorptr) -#elif defined COMPILE_PCRE16 +#else PCRE_EXP_DEFN pcre16_extra * PCRE_CALL_CONVENTION pcre16_study(const pcre16 *external_re, int options, const char **errorptr) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DEFN pcre32_extra * PCRE_CALL_CONVENTION -pcre32_study(const pcre32 *external_re, int options, const char **errorptr) #endif { int min; @@ -1385,7 +1330,6 @@ compile_data compile_block; const REAL_PCRE *re = (const REAL_PCRE *)external_re; - *errorptr = NULL; if (re == NULL || re->magic_number != MAGIC_NUMBER) @@ -1396,12 +1340,10 @@ if ((re->flags & PCRE_MODE) == 0) { -#if defined COMPILE_PCRE8 - *errorptr = "argument not compiled in 8 bit mode"; -#elif defined COMPILE_PCRE16 - *errorptr = "argument not compiled in 16 bit mode"; -#elif defined COMPILE_PCRE32 - *errorptr = "argument not compiled in 32 bit mode"; +#ifdef COMPILE_PCRE8 + *errorptr = "argument is compiled in 16 bit mode"; +#else + *errorptr = "argument is compiled in 8 bit mode"; #endif return NULL; } @@ -1428,18 +1370,14 @@ tables = re->tables; -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 if (tables == NULL) (void)pcre_fullinfo(external_re, NULL, PCRE_INFO_DEFAULT_TABLES, (void *)(&tables)); -#elif defined COMPILE_PCRE16 +#else if (tables == NULL) (void)pcre16_fullinfo(external_re, NULL, PCRE_INFO_DEFAULT_TABLES, (void *)(&tables)); -#elif defined COMPILE_PCRE32 - if (tables == NULL) - (void)pcre32_fullinfo(external_re, NULL, PCRE_INFO_DEFAULT_TABLES, - (void *)(&tables)); #endif compile_block.lcc = tables + lcc_offset; @@ -1462,7 +1400,7 @@ /* Find the minimum length of subject string. */ -switch(min = find_minlength(re, code, code, re->options, 0)) +switch(min = find_minlength(code, code, re->options, 0)) { case -2: *errorptr = "internal error: missing capturing bracket"; return NULL; case -3: *errorptr = "internal error: opcode not recognized"; return NULL; @@ -1470,20 +1408,20 @@ } /* If a set of starting bytes has been identified, or if the minimum length is -greater than zero, or if JIT optimization has been requested, or if -PCRE_STUDY_EXTRA_NEEDED is set, get a pcre[16]_extra block and a -pcre_study_data block. The study data is put in the latter, which is pointed to -by the former, which may also get additional data set later by the calling -program. At the moment, the size of pcre_study_data is fixed. We nevertheless -save it in a field for returning via the pcre_fullinfo() function so that if it -becomes variable in the future, we don't have to change that code. */ +greater than zero, or if JIT optimization has been requested, get a +pcre[16]_extra block and a pcre_study_data block. The study data is put in the +latter, which is pointed to by the former, which may also get additional data +set later by the calling program. At the moment, the size of pcre_study_data +is fixed. We nevertheless save it in a field for returning via the +pcre_fullinfo() function so that if it becomes variable in the future, +we don't have to change that code. */ -if (bits_set || min > 0 || (options & ( +if (bits_set || min > 0 #ifdef SUPPORT_JIT - PCRE_STUDY_JIT_COMPILE | PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE | - PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE | + || (options & (PCRE_STUDY_JIT_COMPILE | PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE + | PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE)) != 0 #endif - PCRE_STUDY_EXTRA_NEEDED)) != 0) + ) { extra = (PUBL(extra) *)(PUBL(malloc)) (sizeof(PUBL(extra)) + sizeof(pcre_study_data)); @@ -1537,8 +1475,7 @@ /* If JIT support was compiled and requested, attempt the JIT compilation. If no starting bytes were found, and the minimum length is zero, and JIT - compilation fails, abandon the extra block and return NULL, unless - PCRE_STUDY_EXTRA_NEEDED is set. */ + compilation fails, abandon the extra block and return NULL. */ #ifdef SUPPORT_JIT extra->executable_jit = NULL; @@ -1549,15 +1486,13 @@ if ((options & PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE) != 0) PRIV(jit_compile)(re, extra, JIT_PARTIAL_HARD_COMPILE); - if (study->flags == 0 && (extra->flags & PCRE_EXTRA_EXECUTABLE_JIT) == 0 && - (options & PCRE_STUDY_EXTRA_NEEDED) == 0) + if (study->flags == 0 && (extra->flags & PCRE_EXTRA_EXECUTABLE_JIT) == 0) { -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 pcre_free_study(extra); -#elif defined COMPILE_PCRE16 +#endif +#ifdef COMPILE_PCRE16 pcre16_free_study(extra); -#elif defined COMPILE_PCRE32 - pcre32_free_study(extra); #endif extra = NULL; } @@ -1578,15 +1513,12 @@ Returns: nothing */ -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 PCRE_EXP_DEFN void pcre_free_study(pcre_extra *extra) -#elif defined COMPILE_PCRE16 +#else PCRE_EXP_DEFN void pcre16_free_study(pcre16_extra *extra) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DEFN void -pcre32_free_study(pcre32_extra *extra) #endif { if (extra == NULL) diff --git a/pcre/pcre_tables.c b/pcre/pcre_tables.c --- a/pcre/pcre_tables.c +++ b/pcre/pcre_tables.c @@ -58,14 +58,8 @@ const pcre_uint8 PRIV(OP_lengths)[] = { OP_LENGTHS }; -/* Tables of horizontal and vertical whitespace characters, suitable for -adding to classes. */ -const pcre_uint32 PRIV(hspace_list)[] = { HSPACE_LIST }; -const pcre_uint32 PRIV(vspace_list)[] = { VSPACE_LIST }; - - /************************************************* * Tables for UTF-8 support * *************************************************/ @@ -74,9 +68,9 @@ character. */ #if (defined SUPPORT_UTF && defined COMPILE_PCRE8) \ - || (defined PCRE_INCLUDED && (defined SUPPORT_PCRE16 || defined SUPPORT_PCRE32)) + || (defined PCRE_INCLUDED && defined SUPPORT_PCRE16) -/* These tables are also required by pcretest in 16- or 32-bit mode. */ +/* These tables are also required by pcretest in 16 bit mode. */ const int PRIV(utf8_table1)[] = { 0x7f, 0x7ff, 0xffff, 0x1fffff, 0x3ffffff, 0x7fffffff}; @@ -98,13 +92,13 @@ 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, 3,3,3,3,3,3,3,3,4,4,4,4,5,5,5,5 }; -#endif /* (SUPPORT_UTF && COMPILE_PCRE8) || (PCRE_INCLUDED && SUPPORT_PCRE[16|32])*/ +#endif /* (SUPPORT_UTF && COMPILE_PCRE8) || (PCRE_INCLUDED && SUPPORT_PCRE16)*/ #ifdef SUPPORT_UTF /* Table to translate from particular type value to the general value. */ -const pcre_uint32 PRIV(ucp_gentype)[] = { +const int PRIV(ucp_gentype)[] = { ucp_C, ucp_C, ucp_C, ucp_C, ucp_C, /* Cc, Cf, Cn, Co, Cs */ ucp_L, ucp_L, ucp_L, ucp_L, ucp_L, /* Ll, Lu, Lm, Lo, Lt */ ucp_M, ucp_M, ucp_M, /* Mc, Me, Mn */ @@ -115,66 +109,6 @@ ucp_Z, ucp_Z, ucp_Z /* Zl, Zp, Zs */ }; -/* This table encodes the rules for finding the end of an extended grapheme -cluster. Every code point has a grapheme break property which is one of the -ucp_gbXX values defined in ucp.h. The 2-dimensional table is indexed by the -properties of two adjacent code points. The left property selects a word from -the table, and the right property selects a bit from that word like this: - - ucp_gbtable[left-property] & (1 << right-property) - -The value is non-zero if a grapheme break is NOT permitted between the relevant -two code points. The breaking rules are as follows: - -1. Break at the start and end of text (pretty obviously). - -2. Do not break between a CR and LF; otherwise, break before and after - controls. - -3. Do not break Hangul syllable sequences, the rules for which are: - - L may be followed by L, V, LV or LVT - LV or V may be followed by V or T - LVT or T may be followed by T - -4. Do not break before extending characters. - -The next two rules are only for extended grapheme clusters (but that's what we -are implementing). - -5. Do not break before SpacingMarks. - -6. Do not break after Prepend characters. - -7. Otherwise, break everywhere. -*/ - -const pcre_uint32 PRIV(ucp_gbtable[]) = { - (1< 0; p++) { - register pcre_uchar ab, c, d; + register int ab, c, d; c = *p; if (c < 128) continue; /* ASCII character */ @@ -289,10 +288,9 @@ } } -#else /* Not SUPPORT_UTF */ +#else /* SUPPORT_UTF */ (void)(string); /* Keep picky compilers happy */ (void)(length); -(void)(erroroffset); #endif return PCRE_UTF8_ERR0; /* This indicates success */ diff --git a/pcre/pcre_version.c b/pcre/pcre_version.c --- a/pcre/pcre_version.c +++ b/pcre/pcre_version.c @@ -79,15 +79,12 @@ pre-processor time. This hack uses a standard trick for avoiding calling the STRING macro with an empty argument when doing the test. */ -#if defined COMPILE_PCRE8 +#ifdef COMPILE_PCRE8 PCRE_EXP_DEFN const char * PCRE_CALL_CONVENTION pcre_version(void) -#elif defined COMPILE_PCRE16 +#else PCRE_EXP_DEFN const char * PCRE_CALL_CONVENTION pcre16_version(void) -#elif defined COMPILE_PCRE32 -PCRE_EXP_DEFN const char * PCRE_CALL_CONVENTION -pcre32_version(void) #endif { return (XSTRING(Z PCRE_PRERELEASE)[1] == 0)? diff --git a/pcre/pcre_xclass.c b/pcre/pcre_xclass.c --- a/pcre/pcre_xclass.c +++ b/pcre/pcre_xclass.c @@ -6,7 +6,7 @@ and semantics are as close as possible to those of the Perl 5 language. Written by Philip Hazel - Copyright (c) 1997-2013 University of Cambridge + Copyright (c) 1997-2012 University of Cambridge ----------------------------------------------------------------------------- Redistribution and use in source and binary forms, with or without @@ -64,9 +64,9 @@ */ BOOL -PRIV(xclass)(pcre_uint32 c, const pcre_uchar *data, BOOL utf) +PRIV(xclass)(int c, const pcre_uchar *data, BOOL utf) { -pcre_uchar t; +int t; BOOL negated = (*data & XCL_NOT) != 0; (void)utf; @@ -94,7 +94,7 @@ while ((t = *data++) != XCL_END) { - pcre_uint32 x, y; + int x, y; if (t == XCL_SINGLE) { #ifdef SUPPORT_UTF @@ -128,120 +128,55 @@ else /* XCL_PROP & XCL_NOTPROP */ { const ucd_record *prop = GET_UCD(c); - BOOL isprop = t == XCL_PROP; switch(*data) { case PT_ANY: - if (isprop) return !negated; + if (t == XCL_PROP) return !negated; break; case PT_LAMP: if ((prop->chartype == ucp_Lu || prop->chartype == ucp_Ll || - prop->chartype == ucp_Lt) == isprop) return !negated; + prop->chartype == ucp_Lt) == (t == XCL_PROP)) return !negated; break; case PT_GC: - if ((data[1] == PRIV(ucp_gentype)[prop->chartype]) == isprop) + if ((data[1] == PRIV(ucp_gentype)[prop->chartype]) == (t == XCL_PROP)) return !negated; break; case PT_PC: - if ((data[1] == prop->chartype) == isprop) return !negated; + if ((data[1] == prop->chartype) == (t == XCL_PROP)) return !negated; break; case PT_SC: - if ((data[1] == prop->script) == isprop) return !negated; + if ((data[1] == prop->script) == (t == XCL_PROP)) return !negated; break; case PT_ALNUM: if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L || - PRIV(ucp_gentype)[prop->chartype] == ucp_N) == isprop) + PRIV(ucp_gentype)[prop->chartype] == ucp_N) == (t == XCL_PROP)) return !negated; break; - /* Perl space used to exclude VT, but from Perl 5.18 it is included, - which means that Perl space and POSIX space are now identical. PCRE - was changed at release 8.34. */ - case PT_SPACE: /* Perl space */ - case PT_PXSPACE: /* POSIX space */ - switch(c) - { - HSPACE_CASES: - VSPACE_CASES: - if (isprop) return !negated; - break; - - default: - if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z) == isprop) - return !negated; - break; - } - break; - - case PT_WORD: - if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L || - PRIV(ucp_gentype)[prop->chartype] == ucp_N || c == CHAR_UNDERSCORE) - == isprop) - return !negated; - break; - - case PT_UCNC: - if (c < 0xa0) - { - if ((c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT || - c == CHAR_GRAVE_ACCENT) == isprop) - return !negated; - } - else - { - if ((c < 0xd800 || c > 0xdfff) == isprop) - return !negated; - } - break; - - /* The following three properties can occur only in an XCLASS, as there - is no \p or \P coding for them. */ - - /* Graphic character. Implement this as not Z (space or separator) and - not C (other), except for Cf (format) with a few exceptions. This seems - to be what Perl does. The exceptional characters are: - - U+061C Arabic Letter Mark - U+180E Mongolian Vowel Separator - U+2066 - U+2069 Various "isolate"s - */ - - case PT_PXGRAPH: - if ((PRIV(ucp_gentype)[prop->chartype] != ucp_Z && - (PRIV(ucp_gentype)[prop->chartype] != ucp_C || - (prop->chartype == ucp_Cf && - c != 0x061c && c != 0x180e && (c < 0x2066 || c > 0x2069)) - )) == isprop) + if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z || + c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR) + == (t == XCL_PROP)) return !negated; break; - /* Printable character: same as graphic, with the addition of Zs, i.e. - not Zl and not Zp, and U+180E. */ - - case PT_PXPRINT: - if ((prop->chartype != ucp_Zl && - prop->chartype != ucp_Zp && - (PRIV(ucp_gentype)[prop->chartype] != ucp_C || - (prop->chartype == ucp_Cf && - c != 0x061c && (c < 0x2066 || c > 0x2069)) - )) == isprop) + case PT_PXSPACE: /* POSIX space */ + if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z || + c == CHAR_HT || c == CHAR_NL || c == CHAR_VT || + c == CHAR_FF || c == CHAR_CR) == (t == XCL_PROP)) return !negated; break; - /* Punctuation: all Unicode punctuation, plus ASCII characters that - Unicode treats as symbols rather than punctuation, for Perl - compatibility (these are $+<=>^`|~). */ - - case PT_PXPUNCT: - if ((PRIV(ucp_gentype)[prop->chartype] == ucp_P || - (c < 256 && PRIV(ucp_gentype)[prop->chartype] == ucp_S)) == isprop) + case PT_WORD: + if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L || + PRIV(ucp_gentype)[prop->chartype] == ucp_N || c == CHAR_UNDERSCORE) + == (t == XCL_PROP)) return !negated; break; diff --git a/pcre/ucp.h b/pcre/ucp.h --- a/pcre/ucp.h +++ b/pcre/ucp.h @@ -7,14 +7,7 @@ /* This file contains definitions of the property values that are returned by the UCD access macros. New values that are added for new releases of Unicode -should always be at the end of each enum, for backwards compatibility. - -IMPORTANT: Note also that the specific numeric values of the enums have to be -the same as the values that are generated by the maint/MultiStage2.py script, -where the equivalent property descriptive names are listed in vectors. - -ALSO: The specific values of the first two enums are assumed for the table -called catposstab in pcre_compile.c. */ +should always be at the end of each enum, for backwards compatibility. */ /* These are the general character categories. */ @@ -28,7 +21,7 @@ ucp_Z /* Separator */ }; -/* These are the particular character categories. */ +/* These are the particular character types. */ enum { ucp_Cc, /* Control */ @@ -61,26 +54,6 @@ ucp_Zl, /* Line separator */ ucp_Zp, /* Paragraph separator */ ucp_Zs /* Space separator */ -}; - -/* These are grapheme break properties. Note that the code for processing them -assumes that the values are less than 16. If more values are added that take -the number to 16 or more, the code will have to be rewritten. */ - -enum { - ucp_gbCR, /* 0 */ - ucp_gbLF, /* 1 */ - ucp_gbControl, /* 2 */ - ucp_gbExtend, /* 3 */ - ucp_gbPrepend, /* 4 */ - ucp_gbSpacingMark, /* 5 */ - ucp_gbL, /* 6 Hangul syllable type L */ - ucp_gbV, /* 7 Hangul syllable type V */ - ucp_gbT, /* 8 Hangul syllable type T */ - ucp_gbLV, /* 9 Hangul syllable type LV */ - ucp_gbLVT, /* 10 Hangul syllable type LVT */ - ucp_gbRegionalIndicator, /* 11 */ - ucp_gbOther /* 12 */ }; /* These are the script identifications. */ diff --git a/regex-pcre-builtin.cabal b/regex-pcre-builtin.cabal --- a/regex-pcre-builtin.cabal +++ b/regex-pcre-builtin.cabal @@ -1,6 +1,6 @@ Name: regex-pcre-builtin -- Keep the Version below in sync with ./Text/Regex/PCRE.hs value getVersion_Text_Regex_PCRE :: Version -Version: 0.94.4.6.8.34 +Version: 0.94.4.7.8.31 Cabal-Version: >=1.2.3 License: BSD3 License-File: LICENSE @@ -16,7 +16,7 @@ Tested-With: GHC Build-Type: Simple Extra-source-files: - pcre/config.h pcre/pcre.h pcre/pcre_byte_order.c pcre/pcre_compile.c pcre/pcre_config.c pcre/pcre_chartables.c pcre/pcre_dfa_exec.c pcre/pcre_exec.c pcre/pcre_fullinfo.c pcre/pcre_get.c pcre/pcre_globals.c pcre/pcre_internal.h pcre/pcre_jit_compile.c pcre/pcre_maketables.c pcre/pcre_newline.c pcre/pcre_ord2utf8.c pcre/pcre_printint.c pcre/pcre_refcount.c pcre/pcre_scanner.h pcre/pcre_string_utils.c pcre/pcre_study.c pcre/pcre_tables.c pcre/pcre_ucd.c pcre/pcre_valid_utf8.c pcre/pcre_version.c pcre/pcre_xclass.c pcre/pcrecpp.h pcre/pcrecpp_internal.h pcre/pcreposix.h pcre/ucp.h + pcre/config.h pcre/pcre.h pcre/pcre_byte_order.c pcre/pcre_compile.c pcre/pcre_config.c pcre/pcre_chartables.c pcre/pcre_dfa_exec.c pcre/pcre_exec.c pcre/pcre_fullinfo.c pcre/pcre_get.c pcre/pcre_globals.c pcre/pcre_internal.h pcre/pcre_jit_compile.c pcre/pcre_jit_test.c pcre/pcre_maketables.c pcre/pcre_newline.c pcre/pcre_ord2utf8.c pcre/pcre_printint.c pcre/pcre_refcount.c pcre/pcre_scanner.h pcre/pcre_string_utils.c pcre/pcre_study.c pcre/pcre_tables.c pcre/pcre_ucd.c pcre/pcre_valid_utf8.c pcre/pcre_version.c pcre/pcre_xclass.c pcre/pcrecpp.h pcre/pcrecpp_internal.h pcre/pcreposix.h pcre/ucp.h NOTES flag newBase description: Choose base >= 4 @@ -65,10 +65,10 @@ -- The only reason to NOT define -DHAVE_PCRE_H is if you are on -- a plotform without a regex library but want to compile this package -- anyway. The resulting regex-posix will exist, but throw errors. - CC-Options: -DHAVE_PCRE_H -DHAVE_CONFIG_H -DPCRE_STATIC + CC-Options: -DHAVE_PCRE_H -DHAVE_CONFIG_H --Include-Dirs: /opt/local/include --Extra-Lib-Dirs: /opt/local/lib Include-dirs: pcre Includes: pcre.h config.h C-sources: - pcre/pcre_byte_order.c pcre/pcre_compile.c pcre/pcre_config.c pcre/pcre_chartables.c pcre/pcre_dfa_exec.c pcre/pcre_exec.c pcre/pcre_fullinfo.c pcre/pcre_get.c pcre/pcre_globals.c pcre/pcre_jit_compile.c pcre/pcre_maketables.c pcre/pcre_newline.c pcre/pcre_ord2utf8.c pcre/pcre_printint.c pcre/pcre_refcount.c pcre/pcre_string_utils.c pcre/pcre_study.c pcre/pcre_tables.c pcre/pcre_ucd.c pcre/pcre_valid_utf8.c pcre/pcre_version.c pcre/pcre_xclass.c + pcre/pcre_byte_order.c pcre/pcre_compile.c pcre/pcre_config.c pcre/pcre_chartables.c pcre/pcre_dfa_exec.c pcre/pcre_exec.c pcre/pcre_fullinfo.c pcre/pcre_get.c pcre/pcre_globals.c pcre/pcre_jit_compile.c pcre/pcre_jit_test.c pcre/pcre_maketables.c pcre/pcre_newline.c pcre/pcre_ord2utf8.c pcre/pcre_printint.c pcre/pcre_refcount.c pcre/pcre_string_utils.c pcre/pcre_study.c pcre/pcre_tables.c pcre/pcre_ucd.c pcre/pcre_valid_utf8.c pcre/pcre_version.c pcre/pcre_xclass.c