diff --git a/README b/README --- a/README +++ b/README @@ -70,22 +70,6 @@ projects at some point in some way that makes sense. - - zip-archive on Hackage - - I started out using zip-archive for this work but soon found that - it has problems with some poorly-made ePub files. Nearly 300 or 40% - of the books I have are unreadable at this time with zip-archive. - - As ugly as it sounds, I found the most reliable solution right - now is to use the unzip shell utility to extract the relevant XML - documents from ePub books. This is why the dependency on HSH. I'm - not thrilled with this situation and would like to find time to - dig into the zip file spec and submit patches to zip-archive. - - But also note that epub-meta is obscenely fast even with invoking - a shell for unzipping. Haskell FTW! - - ----- A word about the version numbering scheme: diff --git a/Setup.hs b/Setup.hs --- a/Setup.hs +++ b/Setup.hs @@ -22,7 +22,7 @@ -- convenience customPostBuild _ _ _ _ = do let binDir = "bin" - let binName = "epub-meta" + let binName = "epubmeta" let destPath = binDir binName createDirectoryIfMissing True binDir diff --git a/epub-metadata.cabal b/epub-metadata.cabal --- a/epub-metadata.cabal +++ b/epub-metadata.cabal @@ -1,5 +1,5 @@ name: epub-metadata -version: 2.0.1 +version: 2.0.2 cabal-version: >= 1.2 build-type: Simple license: BSD3 @@ -35,12 +35,12 @@ Codec.Epub.Opf.Package.Spine, Codec.Epub.Opf.Parse hs-source-dirs: src - build-depends: base >= 3 && < 5, containers, hxt >= 9, mtl, - process, regex-compat + build-depends: base >= 3 && < 5, containers, hxt >= 9, LibZip, + mtl, regex-compat ghc-options: -Wall -executable epub-meta - main-is: Codec/Epub/Opf/Cli/epub-meta.hs +executable epubmeta + main-is: Codec/Epub/Opf/Cli/epubmeta.hs hs-source-dirs: src build-depends: base >= 3 && < 5 ghc-options: -Wall diff --git a/src/Codec/Epub/IO.hs b/src/Codec/Epub/IO.hs --- a/src/Codec/Epub/IO.hs +++ b/src/Codec/Epub/IO.hs @@ -13,11 +13,9 @@ ( extractFileFromZip, opfPath ) where +import Codec.Archive.LibZip import Control.Arrow.ListArrows ( (>>>), deep ) import Control.Monad.Error -import System.Exit -import System.Process -import Text.Printf import Text.Regex import Text.XML.HXT.Arrow.XmlArrow ( getAttrValue, hasName, isElem ) import Text.XML.HXT.Arrow.XmlState ( no, runX, withValidate ) @@ -36,36 +34,18 @@ (mkRegexWithOpts "]*>" False True)) "" -{- | GNU unzip has annoying non-zero exit codes that aren't fatal - so we need to check for those special. --} -handleEC :: (MonadIO m, MonadError String m) - => String -> ExitCode -> m () -handleEC msg (ExitFailure c) - | c > 2 = throwError $ printf "%s status: %s]\n" msg (show c) - | otherwise = return () -handleEC _ ExitSuccess = return () - - {- | Extract a file from a zipfile. - This is here because ePub files are really just zip files. - - Yep, you saw right sports fans. This code is using the command- - line unzip utility. In the future I'd like to make it use a - library. -} extractFileFromZip :: (MonadIO m, MonadError String m) => FilePath -- ^ path to zip file -> FilePath -- ^ path within zip file to extract -> m String -- ^ contents of expected file extractFileFromZip zipPath filePath = do - let dearchiver = "unzip" - - (ec, output, _) <- liftIO $ readProcessWithExitCode - dearchiver ["-p", zipPath, filePath] "" + result <- liftIO $ catchZipError + (fmap Right $ withArchive [] zipPath $ fileContents [] filePath) + (return . Left) - handleEC (printf "[ERROR %s zip file: %s path in zip: %s" - dearchiver zipPath filePath) ec + output <- either (throwError . show) return result return . removeEncoding . removeDoctype $ output diff --git a/src/Codec/Epub/Opf/Cli/Opts.hs b/src/Codec/Epub/Opf/Cli/Opts.hs --- a/src/Codec/Epub/Opf/Cli/Opts.hs +++ b/src/Codec/Epub/Opf/Cli/Opts.hs @@ -46,11 +46,11 @@ usageText = (usageInfo header options) ++ "\n" ++ footer where header = init $ unlines - [ "Usage: epub-meta [OPTIONS] EPUBFILE" - , "Examine ePub OPF package data" + [ "Usage: epubmeta [OPTIONS] EPUBFILE" + , "Examine EPUB OPF package data" , "" , "Options:" ] footer = init $ unlines - [ "Version 2.0.1 Dino Morelli " + [ "Version 2.0.2 Dino Morelli " ] diff --git a/src/Codec/Epub/Opf/Cli/epub-meta.hs b/src/Codec/Epub/Opf/Cli/epub-meta.hs deleted file mode 100644 --- a/src/Codec/Epub/Opf/Cli/epub-meta.hs +++ /dev/null @@ -1,25 +0,0 @@ --- Copyright: 2010, 2011 Dino Morelli --- License: BSD3 (see LICENSE) --- Author: Dino Morelli - -import Control.Monad.Error -import System.Environment ( getArgs ) -import System.Exit - -import Codec.Epub.Opf.Cli.Opts -import Codec.Epub.Opf.Format.Package -import Codec.Epub.Opf.Parse - - -main :: IO () -main = do - (opts, paths) <- getArgs >>= parseOpts - - when ((optHelp opts) || (null paths)) $ do - putStrLn usageText - exitWith $ ExitFailure 1 - - let zipPath = head paths - result <- runErrorT $ parseEpubOpf zipPath - - putStr $ either id (formatPackage (optVerbose opts)) result diff --git a/src/Codec/Epub/Opf/Cli/epubmeta.hs b/src/Codec/Epub/Opf/Cli/epubmeta.hs new file mode 100644 --- /dev/null +++ b/src/Codec/Epub/Opf/Cli/epubmeta.hs @@ -0,0 +1,25 @@ +-- Copyright: 2010, 2011 Dino Morelli +-- License: BSD3 (see LICENSE) +-- Author: Dino Morelli + +import Control.Monad.Error +import System.Environment ( getArgs ) +import System.Exit + +import Codec.Epub.Opf.Cli.Opts +import Codec.Epub.Opf.Format.Package +import Codec.Epub.Opf.Parse + + +main :: IO () +main = do + (opts, paths) <- getArgs >>= parseOpts + + when ((optHelp opts) || (null paths)) $ do + putStrLn usageText + exitWith $ ExitFailure 1 + + let zipPath = head paths + result <- runErrorT $ parseEpubOpf zipPath + + putStr $ either id (formatPackage (optVerbose opts)) result