diff --git a/LICENSE b/LICENSE --- a/LICENSE +++ b/LICENSE @@ -1,4 +1,4 @@ -Copyright Dino Morelli 2010 +Copyright Dino Morelli 2010, 2011 All rights reserved. diff --git a/Setup.hs b/Setup.hs --- a/Setup.hs +++ b/Setup.hs @@ -1,6 +1,6 @@ #! /usr/bin/env runhaskell --- Copyright: 2010 Dino Morelli +-- Copyright: 2010, 2011 Dino Morelli -- License: BSD3 (see LICENSE) -- Author: Dino Morelli diff --git a/TODO b/TODO --- a/TODO +++ b/TODO @@ -1,1 +1,3 @@ - Better errors on parse failure, this will emerge along with more unit testing + +- More unit testing diff --git a/epub-metadata.cabal b/epub-metadata.cabal --- a/epub-metadata.cabal +++ b/epub-metadata.cabal @@ -1,36 +1,45 @@ name: epub-metadata -version: 1.0.2 +version: 2.0.0 cabal-version: >= 1.2 build-type: Simple license: BSD3 license-file: LICENSE -copyright: 2010 Dino Morelli +copyright: 2010, 2011 Dino Morelli author: Dino Morelli maintainer: Dino Morelli stability: experimental homepage: http://ui3.info/d/proj/epub-metadata.html -synopsis: Library and utility for parsing and manipulating - ePub metadata -description: Library for parsing and manipulating ePub OPF - metadata. An attempt has been made here to very - thoroughly implement the metadata portion of the OPF - Package Document specification. Also included is a - command-line utility to dump ePub metadata to stdout - in a human-readable form. +synopsis: Library and utility for parsing and manipulating ePub OPF package data +description: Library and utility for parsing and manipulating ePub + OPF package data. An attempt has been made here to + very thoroughly implement the OPF Package Document + specification. Also included is a command-line + utility to dump OPF package data to stdout in a + human-readable form. category: Codec, Text tested-with: GHC>=6.12.1 library exposed-modules: Codec.Epub.IO, - Codec.Epub.Opf.Metadata, - Codec.Epub.Opf.Metadata.Format, - Codec.Epub.Opf.Metadata.Parse + Codec.Epub.Opf.Common, + Codec.Epub.Opf.Format.Guide, + Codec.Epub.Opf.Format.Manifest, + Codec.Epub.Opf.Format.Metadata, + Codec.Epub.Opf.Format.Package, + Codec.Epub.Opf.Format.Spine, + Codec.Epub.Opf.Format.Util, + Codec.Epub.Opf.Package, + Codec.Epub.Opf.Package.Guide, + Codec.Epub.Opf.Package.Manifest, + Codec.Epub.Opf.Package.Metadata, + Codec.Epub.Opf.Package.Spine, + Codec.Epub.Opf.Parse hs-source-dirs: src - build-depends: base >= 3 && < 5, HSH, hxt, mtl + build-depends: base >= 3 && < 5, containers, HSH, hxt >= 9, mtl ghc-options: -Wall executable epub-meta - main-is: Codec/Epub/Opf/epub-meta.hs + main-is: Codec/Epub/Opf/Cli/epub-meta.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 @@ -1,4 +1,4 @@ --- Copyright: 2010 Dino Morelli +-- Copyright: 2010, 2011 Dino Morelli -- License: BSD3 (see LICENSE) -- Author: Dino Morelli @@ -13,10 +13,13 @@ ( extractFileFromZip, opfPath ) where +import Control.Arrow.ListArrows ( (>>>), deep ) import Control.Monad.Error import HSH.Command import Text.Printf -import Text.XML.HXT.Arrow +import Text.XML.HXT.Arrow.XmlArrow ( getAttrValue, hasName, isElem ) +import Text.XML.HXT.Arrow.XmlState ( no, runX, withValidate ) +import Text.XML.HXT.Arrow.ReadDocument ( readString ) {- | Extract a file from a zipfile. @@ -46,7 +49,7 @@ "META-INF/container.xml" result <- liftIO $ runX ( - readString [(a_validate, v_0)] containerContents + readString [withValidate no] containerContents >>> deep (isElem >>> hasName "rootfile") >>> getAttrValue "full-path" ) diff --git a/src/Codec/Epub/Opf/Cli/Opts.hs b/src/Codec/Epub/Opf/Cli/Opts.hs new file mode 100644 --- /dev/null +++ b/src/Codec/Epub/Opf/Cli/Opts.hs @@ -0,0 +1,56 @@ +-- Copyright: 2010, 2011 Dino Morelli +-- License: BSD3 (see LICENSE) +-- Author: Dino Morelli + +module Codec.Epub.Opf.Cli.Opts + ( Options (..) + , parseOpts, usageText + ) + where + +import System.Console.GetOpt + + +data Options = Options + { optHelp :: Bool + , optVerbose :: Bool + } + + +defaultOptions :: Options +defaultOptions = Options + { optHelp = False + , optVerbose = False + } + + +options :: [OptDescr (Options -> Options)] +options = + [ Option ['h'] ["help"] + (NoArg (\opts -> opts { optHelp = True } )) + "This help text" + , Option ['v'] ["verbose"] + (NoArg (\opts -> opts { optVerbose = True } )) + "Display all OPF package info, including manifest, spine and guide" + ] + + +parseOpts :: [String] -> IO (Options, [String]) +parseOpts argv = + case getOpt Permute options argv of + (o,n,[] ) -> return (foldl (flip id) defaultOptions o, n) + (_,_,errs) -> ioError $ userError (concat errs ++ usageText) + + +usageText :: String +usageText = (usageInfo header options) ++ "\n" ++ footer + where + header = init $ unlines + [ "Usage: epub-meta [OPTIONS] EPUBFILE" + , "Examine ePub OPF package data" + , "" + , "Options:" + ] + footer = init $ unlines + [ "Version 2.0.0 Dino Morelli " + ] diff --git a/src/Codec/Epub/Opf/Cli/epub-meta.hs b/src/Codec/Epub/Opf/Cli/epub-meta.hs new file mode 100644 --- /dev/null +++ b/src/Codec/Epub/Opf/Cli/epub-meta.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 diff --git a/src/Codec/Epub/Opf/Common.hs b/src/Codec/Epub/Opf/Common.hs new file mode 100644 --- /dev/null +++ b/src/Codec/Epub/Opf/Common.hs @@ -0,0 +1,17 @@ +-- Copyright: 2010, 2011 Dino Morelli +-- License: BSD3 (see LICENSE) +-- Author: Dino Morelli + +{- | Data types for working with the metadata of ePub documents + + These data types were constructed by studying the IDPF OPF + specification for ePub documents found here: + + +-} +module Codec.Epub.Opf.Common + where + + +-- | Attribute value used to relate things to each other in the OPF data +type MFItemId = String diff --git a/src/Codec/Epub/Opf/Format/Guide.hs b/src/Codec/Epub/Opf/Format/Guide.hs new file mode 100644 --- /dev/null +++ b/src/Codec/Epub/Opf/Format/Guide.hs @@ -0,0 +1,31 @@ +-- Copyright: 2010, 2011 Dino Morelli +-- License: BSD3 (see LICENSE) +-- Author: Dino Morelli + +{-# LANGUAGE FlexibleContexts #-} + +-- | Module for pretty-printing ePub metadata info +module Codec.Epub.Opf.Format.Guide + where + +import Control.Monad.Writer.Lazy +import Text.Printf + +import Codec.Epub.Opf.Format.Util +import Codec.Epub.Opf.Package.Guide + + +tellGuideRef :: MonadWriter (Seq Char) m => GuideRef -> m () +tellGuideRef (GuideRef grty title href) = + tellSeq $ printf " type: %s%s, href: %s\n" + grty (titleToString title) href + + where + titleToString = maybe "" (printf ", title: %s") + + +tellGuide :: MonadWriter (Seq Char) m => [GuideRef] -> m () +tellGuide [] = return () +tellGuide grs = do + tellSeq "guide items:\n" + mapM_ tellGuideRef grs diff --git a/src/Codec/Epub/Opf/Format/Manifest.hs b/src/Codec/Epub/Opf/Format/Manifest.hs new file mode 100644 --- /dev/null +++ b/src/Codec/Epub/Opf/Format/Manifest.hs @@ -0,0 +1,27 @@ +-- Copyright: 2010, 2011 Dino Morelli +-- License: BSD3 (see LICENSE) +-- Author: Dino Morelli + +{-# LANGUAGE FlexibleContexts #-} + +-- | Module for pretty-printing ePub metadata info +module Codec.Epub.Opf.Format.Manifest + where + +import Control.Monad.Writer.Lazy +import Text.Printf + +import Codec.Epub.Opf.Format.Util +import Codec.Epub.Opf.Package.Manifest + + +tellManifestItem :: MonadWriter (Seq Char) m => ManifestItem -> m () +tellManifestItem (ManifestItem mfId href mediaType) = + tellSeq $ printf " id: %s, href: %s, media-type: %s\n" + mfId href mediaType + + +tellManifest :: MonadWriter (Seq Char) m => [ManifestItem] -> m () +tellManifest mas = do + tellSeq "manifest items:\n" + mapM_ tellManifestItem mas diff --git a/src/Codec/Epub/Opf/Format/Metadata.hs b/src/Codec/Epub/Opf/Format/Metadata.hs new file mode 100644 --- /dev/null +++ b/src/Codec/Epub/Opf/Format/Metadata.hs @@ -0,0 +1,120 @@ +-- Copyright: 2010, 2011 Dino Morelli +-- License: BSD3 (see LICENSE) +-- Author: Dino Morelli + +{-# LANGUAGE FlexibleContexts #-} + +-- | Module for pretty-printing ePub metadata info +module Codec.Epub.Opf.Format.Metadata + where + +import Control.Monad.Writer.Lazy +import Text.Printf + +import Codec.Epub.Opf.Format.Util +import Codec.Epub.Opf.Package.Metadata + + +tellTitle :: MonadWriter (Seq Char) m => MetaTitle -> m () +tellTitle (MetaTitle Nothing title) = tellSeq $ printf "title: %s\n" title +tellTitle (MetaTitle lang title) = + tellSeq $ printf "title\n%s%s" (formatSubline "lang" lang) + (formatSubline "title" (Just title)) + + +tellCreator :: MonadWriter (Seq Char) m => MetaCreator -> m () +tellCreator (MetaCreator Nothing Nothing creator) = + tellSeq $ printf "creator: %s\n" creator +tellCreator (MetaCreator role fileAs creator) = + tellSeq $ printf "creator\n%s%s%s" + (formatSubline "role" role) + (formatSubline "file-as" fileAs) + (formatSubline "creator" (Just creator)) + + +tellContributor :: MonadWriter (Seq Char) m => MetaCreator -> m () +tellContributor (MetaCreator Nothing Nothing contributor) = + tellSeq $ printf "contributor: %s\n" contributor +tellContributor (MetaCreator role fileAs contributor) = + tellSeq $ printf "contributor\n%s%s%s" + (formatSubline "role" role) + (formatSubline "file-as" fileAs) + (formatSubline "creator" (Just contributor)) + + +tellDate :: MonadWriter (Seq Char) m => MetaDate -> m () +tellDate (MetaDate Nothing date) = + tellSeq $ printf "date: %s\n" date +tellDate (MetaDate event date) = + tellSeq $ printf "date\n%s%s" + (formatSubline "event" event) + (formatSubline "date" (Just date)) + + +tellType :: MonadWriter (Seq Char) m => Maybe String -> m () +tellType = maybe (return ()) (tellSeq . (printf "type: %s\n")) + + +tellFormat :: MonadWriter (Seq Char) m => Maybe String -> m () +tellFormat = maybe (return ()) (tellSeq . (printf "format: %s\n")) + + +tellId :: MonadWriter (Seq Char) m => MetaId -> m () +tellId (MetaId idVal scheme content) = + tellSeq $ printf "identifier\n%s%s%s" + (formatSubline "id" (Just idVal)) + (formatSubline "scheme" scheme) + (formatSubline "identifier" (Just content)) + + +tellSource :: MonadWriter (Seq Char) m => Maybe String -> m () +tellSource = maybe (return ()) (tellSeq . (printf "source: %s\n")) + + +tellSubject :: MonadWriter (Seq Char) m => String -> m () +tellSubject = tellSeq . (printf "subject: %s\n") + + +tellDescription :: MonadWriter (Seq Char) m => Maybe String -> m () +tellDescription = maybe (return ()) (tellSeq . (printf "description: %s\n")) + + +tellPublisher :: MonadWriter (Seq Char) m => Maybe String -> m () +tellPublisher = maybe (return ()) (tellSeq . (printf "publisher: %s\n")) + + +tellLang :: MonadWriter (Seq Char) m => String -> m () +tellLang = tellSeq . (printf "language: %s\n") + + +tellRelation :: MonadWriter (Seq Char) m => Maybe String -> m () +tellRelation = maybe (return ()) (tellSeq . (printf "relation: %s\n")) + + +tellCoverage :: MonadWriter (Seq Char) m => Maybe String -> m () +tellCoverage = maybe (return ()) (tellSeq . (printf "coverage: %s\n")) + + +tellRights :: MonadWriter (Seq Char) m => Maybe String -> m () +tellRights = maybe (return ()) (tellSeq . (printf "rights: %s\n")) + + +tellMetadata :: MonadWriter (Seq Char) m => Metadata -> m () +tellMetadata (Metadata titles creators contributors subjects desc + publisher dates mType format ids source langs relation + coverage rights) = do + mapM_ tellTitle titles + mapM_ tellCreator creators + mapM_ tellContributor contributors + mapM_ tellDate dates + tellType mType + tellFormat format + mapM_ tellId ids + tellSource source + mapM_ tellSubject subjects + tellDescription desc + tellPublisher publisher + mapM_ tellLang langs + tellRelation relation + tellCoverage coverage + tellRights rights diff --git a/src/Codec/Epub/Opf/Format/Package.hs b/src/Codec/Epub/Opf/Format/Package.hs new file mode 100644 --- /dev/null +++ b/src/Codec/Epub/Opf/Format/Package.hs @@ -0,0 +1,40 @@ +-- Copyright: 2010, 2011 Dino Morelli +-- License: BSD3 (see LICENSE) +-- Author: Dino Morelli + +{-# LANGUAGE FlexibleContexts #-} + +-- | Module for pretty-printing OPF package data +module Codec.Epub.Opf.Format.Package + ( formatPackage + ) + where + +import Control.Monad.Writer.Lazy +import Data.Foldable ( toList ) + +import Codec.Epub.Opf.Format.Guide +import Codec.Epub.Opf.Format.Manifest +import Codec.Epub.Opf.Format.Metadata +import Codec.Epub.Opf.Format.Spine +import Codec.Epub.Opf.Format.Util +import Codec.Epub.Opf.Package + + +tellPackage :: MonadWriter (Seq Char) m => (String, String) -> m () +tellPackage (version, uniqueId) = do + tellSeq "package\n" + tellSeq $ formatSubline "version" (Just version) + tellSeq $ formatSubline "unique-identifier" (Just uniqueId) + + +formatPackage :: Bool -> Package -> String +formatPackage showAll (Package v u meta ma sp gu) = + toList . execWriter $ do + tellPackage (v, u) + tellMetadata meta + + when showAll $ do + tellManifest ma + tellSpine sp + tellGuide gu diff --git a/src/Codec/Epub/Opf/Format/Spine.hs b/src/Codec/Epub/Opf/Format/Spine.hs new file mode 100644 --- /dev/null +++ b/src/Codec/Epub/Opf/Format/Spine.hs @@ -0,0 +1,33 @@ +-- Copyright: 2010, 2011 Dino Morelli +-- License: BSD3 (see LICENSE) +-- Author: Dino Morelli + +{-# LANGUAGE FlexibleContexts #-} + +-- | Module for pretty-printing ePub metadata info +module Codec.Epub.Opf.Format.Spine + where + +import Control.Monad.Writer.Lazy +import Text.Printf + +import Codec.Epub.Opf.Format.Util +import Codec.Epub.Opf.Package.Spine + + +tellSpineItemref :: MonadWriter (Seq Char) m => SpineItemref -> m () +tellSpineItemref (SpineItemref idref linear) = + tellSeq $ printf " idref: %s%s\n" idref (linearToString linear) + + where + boolToYn True = "yes" + boolToYn False = "no" + + linearToString Nothing = "" + linearToString (Just l) = printf ", linear: %s" (boolToYn l) + + +tellSpine :: MonadWriter (Seq Char) m => Spine -> m () +tellSpine (Spine toc itemRefs) = do + tellSeq $ printf "spine toc: %s, itemrefs:\n" toc + mapM_ tellSpineItemref itemRefs diff --git a/src/Codec/Epub/Opf/Format/Util.hs b/src/Codec/Epub/Opf/Format/Util.hs new file mode 100644 --- /dev/null +++ b/src/Codec/Epub/Opf/Format/Util.hs @@ -0,0 +1,26 @@ +-- Copyright: 2010, 2011 Dino Morelli +-- License: BSD3 (see LICENSE) +-- Author: Dino Morelli + +{-# LANGUAGE FlexibleContexts #-} + +-- | Functions shared by several formatting modules +module Codec.Epub.Opf.Format.Util + ( formatSubline + , tellSeq + , Seq + ) + where + +import Control.Monad.Writer.Lazy +import Data.Sequence ( Seq, fromList ) +import Text.Printf + + +formatSubline :: String -> Maybe String -> String +formatSubline _ Nothing = "" +formatSubline key (Just value) = printf " %s: %s\n" key value + + +tellSeq :: MonadWriter (Seq a) m => [a] -> m () +tellSeq = tell . fromList diff --git a/src/Codec/Epub/Opf/Metadata.hs b/src/Codec/Epub/Opf/Metadata.hs deleted file mode 100644 --- a/src/Codec/Epub/Opf/Metadata.hs +++ /dev/null @@ -1,92 +0,0 @@ --- Copyright: 2010 Dino Morelli --- License: BSD3 (see LICENSE) --- Author: Dino Morelli - -{- | Data types for working with the metadata of ePub documents - - These data types were constructed by studying the IDPF OPF - specification for ePub documents found here: - - --} -module Codec.Epub.Opf.Metadata - ( OPFPackage (..) - , EMTitle (..) - , EMCreator (..) - , EMDate (..) - , EMId (..) - , EpubMeta (..) - , emptyEpubMeta - ) - where - - -{- | opf:package tag - - Note that we are not yet storing the data that comes after - \/package\/metadata in an OPF Package Document. But that may - be added at a later time. --} -data OPFPackage = OPFPackage - { opVersion :: String -- ^ version attr - , opUniqueId :: String -- ^ unique-identifier attr - , opMeta :: EpubMeta -- ^ metadata child element contents - } - deriving (Eq, Show) - - --- | dc:title tag, xml:lang attr, content -data EMTitle = EMTitle (Maybe String) String - deriving (Eq, Show) - --- | dc:creator tag, opf:role attr, opf:file-as attr, content -data EMCreator = EMCreator (Maybe String) (Maybe String) String - deriving (Eq, Show) - --- | dc:date tag, opf:event attr, content -data EMDate = EMDate (Maybe String) String - deriving (Eq, Show) - --- | dc:identifier tag, id attr, opf:scheme attr, content -data EMId = EMId String (Maybe String) String - deriving (Eq, Show) - --- | opf:metadata tag -data EpubMeta = EpubMeta - { emTitles :: [EMTitle] -- ^ at least one required - , emCreators :: [EMCreator] - , emContributors :: [EMCreator] - , emSubjects :: [String] -- ^ dc:subject tags - , emDescription :: Maybe String -- ^ dc:description tags - , emPublisher :: Maybe String -- ^ dc:publisher tag - , emDates :: [EMDate] - , emType :: Maybe String -- ^ dc:type tag - , emFormat :: Maybe String -- ^ dc:format tag - , emIds :: [EMId] -- ^ at least one required - , emSource :: Maybe String -- ^ dc:source tag - , emLangs :: [String] -- ^ dc:language tags, at least one required - , emRelation :: Maybe String -- ^ dc:relation tag - , emCoverage :: Maybe String -- ^ dc:coverage tag - , emRights :: Maybe String -- ^ dc:rights tag - } - deriving (Eq, Show) - --- | Note: This isn't valid as-is, some required values are empty lists! -emptyEpubMeta :: EpubMeta -emptyEpubMeta = EpubMeta - { emTitles = [] -- one required - , emCreators = [] - , emContributors = [] - , emSubjects = [] - , emDescription = Nothing - , emPublisher = Nothing - , emDates = [] - , emType = Nothing - , emFormat = Nothing - , emIds = [] -- one required - , emSource = Nothing - , emLangs = [] -- one required - , emRelation = Nothing - , emCoverage = Nothing - , emRights = Nothing - } diff --git a/src/Codec/Epub/Opf/Metadata/Format.hs b/src/Codec/Epub/Opf/Metadata/Format.hs deleted file mode 100644 --- a/src/Codec/Epub/Opf/Metadata/Format.hs +++ /dev/null @@ -1,131 +0,0 @@ --- Copyright: 2010 Dino Morelli --- License: BSD3 (see LICENSE) --- Author: Dino Morelli - --- | Module for pretty-printing ePub metadata info -module Codec.Epub.Opf.Metadata.Format - ( opfToString - ) - where - -import Text.Printf - -import Codec.Epub.Opf.Metadata - - -formatSubline :: String -> Maybe String -> String -formatSubline _ Nothing = "" -formatSubline key (Just value) = printf " %s: %s\n" key value - - -packageToString :: (String, String) -> String -packageToString (version, uniqueId) = - "package\n" ++ - (formatSubline "version" (Just version)) ++ - (formatSubline "unique-identifier" (Just uniqueId)) - - -titleToString :: EMTitle -> String -titleToString (EMTitle Nothing title) = printf "title: %s\n" title -titleToString (EMTitle lang title) = - "title\n" ++ - (formatSubline "lang" lang) ++ - (formatSubline "title" (Just title)) - - -creatorToString :: EMCreator -> String -creatorToString (EMCreator Nothing Nothing creator) = - printf "creator: %s\n" creator -creatorToString (EMCreator role fileAs creator) = - "creator\n" ++ - (formatSubline "role" role) ++ - (formatSubline "file-as" fileAs) ++ - (formatSubline "creator" (Just creator)) - - -contributorToString :: EMCreator -> String -contributorToString (EMCreator Nothing Nothing contributor) = - printf "contributor: %s\n" contributor -contributorToString (EMCreator role fileAs contributor) = - "contributor\n" ++ - (formatSubline "role" role) ++ - (formatSubline "file-as" fileAs) ++ - (formatSubline "creator" (Just contributor)) - - -subjectToString :: String -> String -subjectToString = printf "subject: %s\n" - - -descriptionToString :: Maybe String -> String -descriptionToString = maybe "" (printf "description: %s\n") - - -publisherToString :: Maybe String -> String -publisherToString = maybe "" (printf "publisher: %s\n") - - -dateToString :: EMDate -> String -dateToString (EMDate Nothing date) = - printf "date: %s\n" date -dateToString (EMDate event date) = - "date\n" ++ - (formatSubline "event" event) ++ - (formatSubline "date" (Just date)) - - -typeToString :: Maybe String -> String -typeToString = maybe "" (printf "type: %s\n") - - -formatToString :: Maybe String -> String -formatToString = maybe "" (printf "format: %s\n") - - -idToString :: EMId -> String -idToString (EMId idVal scheme content) = - "identifier\n" ++ - (formatSubline "id" (Just idVal)) ++ - (formatSubline "scheme" scheme) ++ - (formatSubline "identifier" (Just content)) - - -sourceToString :: Maybe String -> String -sourceToString = maybe "" (printf "source: %s\n") - - -langToString :: String -> String -langToString = printf "language: %s\n" - - -relationToString :: Maybe String -> String -relationToString = maybe "" (printf "relation: %s\n") - - -coverageToString :: Maybe String -> String -coverageToString = maybe "" (printf "coverage: %s\n") - - -rightsToString :: Maybe String -> String -rightsToString = maybe "" (printf "rights: %s\n") - - --- | Format an ePub metadata into a String -opfToString :: OPFPackage -> String -opfToString (OPFPackage v u em) = concat $ - [packageToString (v, u)] ++ - (map titleToString $ emTitles em) ++ - (map creatorToString $ emCreators em) ++ - (map contributorToString $ emContributors em) ++ - (map dateToString $ emDates em) ++ - [typeToString . emType $ em] ++ - [formatToString . emFormat $ em] ++ - (map idToString $ emIds em) ++ - [sourceToString . emSource $ em] ++ - (map subjectToString $ emSubjects em) ++ - [descriptionToString . emDescription $ em] ++ - [publisherToString . emPublisher $ em] ++ - (map langToString $ emLangs em) ++ - [relationToString . emRelation $ em] ++ - [coverageToString . emCoverage $ em] ++ - [rightsToString . emRights $ em] diff --git a/src/Codec/Epub/Opf/Metadata/Parse.hs b/src/Codec/Epub/Opf/Metadata/Parse.hs deleted file mode 100644 --- a/src/Codec/Epub/Opf/Metadata/Parse.hs +++ /dev/null @@ -1,210 +0,0 @@ --- Copyright: 2010 Dino Morelli --- License: BSD3 (see LICENSE) --- Author: Dino Morelli - -{-# LANGUAGE Arrows #-} -{-# LANGUAGE FlexibleContexts #-} - --- | Module for extracting the metadata from an ePub file -module Codec.Epub.Opf.Metadata.Parse - ( parseXmlToOpf - , parseEpubOpf - ) - where - -import Control.Applicative -import Control.Monad.Error -import Data.Tree.NTree.TypeDefs ( NTree ) -import Prelude hiding ( cos ) -import Text.XML.HXT.Arrow - -import Codec.Epub.IO -import Codec.Epub.Opf.Metadata - - --- HXT helpers - -atTag :: (ArrowXml a) => String -> a (NTree XNode) XmlTree -atTag tag = deep (isElem >>> hasName tag) - -atQTag :: (ArrowXml a) => QName -> a (NTree XNode) XmlTree -atQTag tag = deep (isElem >>> hasQName tag) - -text :: (ArrowXml a) => a (NTree XNode) String -text = getChildren >>> getText - -notNullA :: (ArrowList a) => a [b] [b] -notNullA = isA $ not . null - - -mbQTagText :: (ArrowXml a) => QName -> a (NTree XNode) (Maybe String) -mbQTagText tag = - ( atQTag tag >>> - text >>> notNullA >>> arr Just ) - `orElse` - (constA Nothing) - - -mbGetAttrValue :: (ArrowXml a) => - String -> a XmlTree (Maybe String) -mbGetAttrValue n = - (getAttrValue n >>> notNullA >>> arr Just) - `orElse` (constA Nothing) - -mbGetQAttrValue :: (ArrowXml a) => - QName -> a XmlTree (Maybe String) -mbGetQAttrValue qn = - (getQAttrValue qn >>> notNullA >>> arr Just) - `orElse` (constA Nothing) - - -{- ePub parsing helpers - - Note that these URIs could conceivably change in the future - Is it ok that they're hardcoded like this? - - Well, ok, the xml namespace URI will probably never change. --} - -dcName, opfName, xmlName :: String -> QName -dcName local = mkQName "dc" local "http://purl.org/dc/elements/1.1/" -opfName local = mkQName "opf" local "http://www.idpf.org/2007/opf" -xmlName local = mkQName "xml" local "http://www.w3.org/XML/1998/namespace" - - -getPackage :: (ArrowXml a) => a (NTree XNode) (String, String) -getPackage = atTag "package" >>> - proc x -> do - v <- getAttrValue "version" -< x - u <- getAttrValue "unique-identifier" -< x - returnA -< (v, u) - - -getTitle :: (ArrowXml a) => a (NTree XNode) EMTitle -getTitle = atQTag (dcName "title") >>> - proc x -> do - l <- mbGetQAttrValue (xmlName "lang") -< x - c <- text -< x - returnA -< EMTitle l c - - -{- Since creators and contributors have the same exact XML structure, - this arrow is used to get either of them --} -getCreator :: (ArrowXml a) => String -> a (NTree XNode) EMCreator -getCreator tag = atQTag (dcName tag) >>> ( unwrapArrow $ EMCreator - <$> (WrapArrow $ mbGetQAttrValue (opfName "role")) - <*> (WrapArrow $ mbGetQAttrValue (opfName "file-as")) - <*> (WrapArrow $ text) - ) - - -getSubject :: (ArrowXml a) => a (NTree XNode) String -getSubject = atQTag (dcName "subject") >>> text - - -getDescription :: (ArrowXml a) => a (NTree XNode) (Maybe String) -getDescription = mbQTagText $ dcName "description" - - -getPublisher :: (ArrowXml a) => a (NTree XNode) (Maybe String) -getPublisher = mbQTagText $ dcName "publisher" - - -getDate :: (ArrowXml a) => a (NTree XNode) EMDate -getDate = atQTag (dcName "date") >>> - proc x -> do - e <- mbGetQAttrValue (opfName "event") -< x - c <- text -< x - returnA -< EMDate e c - - -getType :: (ArrowXml a) => a (NTree XNode) (Maybe String) -getType = mbQTagText $ dcName "type" - - -getFormat :: (ArrowXml a) => a (NTree XNode) (Maybe String) -getFormat = mbQTagText $ dcName "format" - - -getId :: (ArrowXml a) => a (NTree XNode) EMId -getId = atQTag (dcName "identifier") >>> - proc x -> do - mbi <- mbGetAttrValue "id" -< x - s <- mbGetQAttrValue (opfName "scheme") -< x - c <- text -< x - let i = maybe "[WARNING: missing required id attribute]" id mbi - returnA -< EMId i s c - - -getSource :: (ArrowXml a) => a (NTree XNode) (Maybe String) -getSource = mbQTagText $ dcName "source" - - -getLang :: (ArrowXml a) => a (NTree XNode) String -getLang = atQTag (dcName "language") >>> text - - -getRelation :: (ArrowXml a) => a (NTree XNode) (Maybe String) -getRelation = mbQTagText $ dcName "relation" - - -getCoverage :: (ArrowXml a) => a (NTree XNode) (Maybe String) -getCoverage = mbQTagText $ dcName "coverage" - - -getRights :: (ArrowXml a) => a (NTree XNode) (Maybe String) -getRights = mbQTagText $ dcName "rights" - - -getMeta :: (ArrowXml a) => a (NTree XNode) EpubMeta -getMeta = atTag "metadata" >>> ( unwrapArrow $ EpubMeta - <$> (WrapArrow $ listA getTitle) - <*> (WrapArrow $ listA $ getCreator "creator") - <*> (WrapArrow $ listA $ getCreator "contributor") - <*> (WrapArrow $ listA getSubject) - <*> (WrapArrow $ getDescription) - <*> (WrapArrow $ getPublisher) - <*> (WrapArrow $ listA getDate) - <*> (WrapArrow $ getType) - <*> (WrapArrow $ getFormat) - <*> (WrapArrow $ listA getId) - <*> (WrapArrow $ getSource) - <*> (WrapArrow $ listA getLang) - <*> (WrapArrow $ getRelation) - <*> (WrapArrow $ getCoverage) - <*> (WrapArrow $ getRights) - ) - - -getBookData :: (ArrowXml a) => a (NTree XNode) OPFPackage -getBookData = - proc x -> do - (v, u) <- getPackage -< x - m <- getMeta -< x - returnA -< OPFPackage v u m - - -{- | Extract the ePub metadata contained in the OPF Package Document - contained in the supplied string --} -parseXmlToOpf :: (MonadIO m) => String -> m [OPFPackage] -parseXmlToOpf opfContents = - liftIO $ runX ( - readString [(a_validate, v_0)] opfContents - >>> propagateNamespaces - >>> getBookData - ) - - --- | Given the path to an ePub file, extract the metadata -parseEpubOpf :: (MonadIO m, MonadError String m) => - FilePath -> m OPFPackage -parseEpubOpf zipPath = do - opfContents <- extractFileFromZip zipPath =<< opfPath zipPath - result <- parseXmlToOpf opfContents - - case result of - (em : []) -> return em - _ -> throwError - "ERROR: we didn't come up with a single EpubMeta" diff --git a/src/Codec/Epub/Opf/Package.hs b/src/Codec/Epub/Opf/Package.hs new file mode 100644 --- /dev/null +++ b/src/Codec/Epub/Opf/Package.hs @@ -0,0 +1,43 @@ +-- Copyright: 2010, 2011 Dino Morelli +-- License: BSD3 (see LICENSE) +-- Author: Dino Morelli + +{- | Data types for working with the metadata of ePub documents + + These data types were constructed by studying the IDPF OPF + specification for ePub documents found here: + + +-} +module Codec.Epub.Opf.Package + ( Package (..) + , MetaTitle (..) + , MetaCreator (..) + , MetaDate (..) + , MetaId (..) + , Metadata (..) + , ManifestItem (..) + , SpineItemref (..) + , Spine (..) + , GuideRef (..) + , emptyMetadata + ) + where + +import Codec.Epub.Opf.Package.Guide +import Codec.Epub.Opf.Package.Manifest +import Codec.Epub.Opf.Package.Metadata +import Codec.Epub.Opf.Package.Spine + + +{- | package tag +-} +data Package = Package + { opVersion :: String -- ^ version attr + , opUniqueId :: String -- ^ unique-identifier attr + , opMeta :: Metadata -- ^ metadata child element contents + , opManifest :: [ManifestItem] -- ^ manifest child element contents. One required + , opSpine :: Spine -- ^ spine child element contents + , opGuide :: [GuideRef] -- ^ guide child element contents + } + deriving (Eq, Show) diff --git a/src/Codec/Epub/Opf/Package/Guide.hs b/src/Codec/Epub/Opf/Package/Guide.hs new file mode 100644 --- /dev/null +++ b/src/Codec/Epub/Opf/Package/Guide.hs @@ -0,0 +1,24 @@ +-- Copyright: 2010, 2011 Dino Morelli +-- License: BSD3 (see LICENSE) +-- Author: Dino Morelli + +{- | Data types for working with the metadata of ePub documents + + These data types were constructed by studying the IDPF OPF + specification for ePub documents found here: + + +-} +module Codec.Epub.Opf.Package.Guide + ( GuideRef (..) + ) + where + + +-- | package\/guide\/reference tag +data GuideRef = GuideRef + { grType :: String -- ^ type attr. Must follow 13th edition of the Chicago Manual of Style + , grTitle :: Maybe String -- ^ title attr + , grHref :: String -- ^ href attr. Must reference item in manifest + } + deriving (Eq, Show) diff --git a/src/Codec/Epub/Opf/Package/Manifest.hs b/src/Codec/Epub/Opf/Package/Manifest.hs new file mode 100644 --- /dev/null +++ b/src/Codec/Epub/Opf/Package/Manifest.hs @@ -0,0 +1,31 @@ +-- Copyright: 2010, 2011 Dino Morelli +-- License: BSD3 (see LICENSE) +-- Author: Dino Morelli + +{- | Data types for working with the metadata of ePub documents + + These data types were constructed by studying the IDPF OPF + specification for ePub documents found here: + + +-} +module Codec.Epub.Opf.Package.Manifest + ( ManifestItem (..) + ) + where + +import Codec.Epub.Opf.Common + + +-- | manifest attribute values +type MFItemHref = String +type MFItemMediaType = String + + +-- | package\/manifest\/item tag +data ManifestItem = ManifestItem + { mfiId :: MFItemId -- ^ id attr + , mfiHref :: MFItemHref -- ^ href attr + , mfiMediaType :: MFItemMediaType -- ^ media-type attr + } + deriving (Eq, Show) diff --git a/src/Codec/Epub/Opf/Package/Metadata.hs b/src/Codec/Epub/Opf/Package/Metadata.hs new file mode 100644 --- /dev/null +++ b/src/Codec/Epub/Opf/Package/Metadata.hs @@ -0,0 +1,81 @@ +-- Copyright: 2010, 2011 Dino Morelli +-- License: BSD3 (see LICENSE) +-- Author: Dino Morelli + +{- | Data types for working with the metadata of ePub documents + + These data types were constructed by studying the IDPF OPF + specification for ePub documents found here: + + +-} +module Codec.Epub.Opf.Package.Metadata + ( MetaTitle (..) + , MetaCreator (..) + , MetaDate (..) + , MetaId (..) + , Metadata (..) + , emptyMetadata + ) + where + + +-- | package\/metadata\/dc:title tag, xml:lang attr, content +data MetaTitle = MetaTitle (Maybe String) String + deriving (Eq, Show) + +{- | package\/metadata\/dc:creator tag, opf:role attr, opf:file-as attr, + content +-} +data MetaCreator = MetaCreator (Maybe String) (Maybe String) String + deriving (Eq, Show) + +-- | package\/metadata\/dc:date tag, opf:event attr, content +data MetaDate = MetaDate (Maybe String) String + deriving (Eq, Show) + +{- | package\/metadata\/dc:identifier tag, id attr, opf:scheme attr, + content +-} +data MetaId = MetaId String (Maybe String) String + deriving (Eq, Show) + +-- | package\/metadata tag +data Metadata = Metadata + { metaTitles :: [MetaTitle] -- ^ at least one required + , metaCreators :: [MetaCreator] + , metaContributors :: [MetaCreator] + , metaSubjects :: [String] -- ^ dc:subject tags + , metaDescription :: Maybe String -- ^ dc:description tags + , metaPublisher :: Maybe String -- ^ dc:publisher tag + , metaDates :: [MetaDate] + , metaType :: Maybe String -- ^ dc:type tag + , metaFormat :: Maybe String -- ^ dc:format tag + , metaIds :: [MetaId] -- ^ at least one required + , metaSource :: Maybe String -- ^ dc:source tag + , metaLangs :: [String] -- ^ dc:language tags, at least one required + , metaRelation :: Maybe String -- ^ dc:relation tag + , metaCoverage :: Maybe String -- ^ dc:coverage tag + , metaRights :: Maybe String -- ^ dc:rights tag + } + deriving (Eq, Show) + +-- | Note: This isn't valid as-is, some required values are empty lists! +emptyMetadata :: Metadata +emptyMetadata = Metadata + { metaTitles = [] -- one required + , metaCreators = [] + , metaContributors = [] + , metaSubjects = [] + , metaDescription = Nothing + , metaPublisher = Nothing + , metaDates = [] + , metaType = Nothing + , metaFormat = Nothing + , metaIds = [] -- one required + , metaSource = Nothing + , metaLangs = [] -- one required + , metaRelation = Nothing + , metaCoverage = Nothing + , metaRights = Nothing + } diff --git a/src/Codec/Epub/Opf/Package/Spine.hs b/src/Codec/Epub/Opf/Package/Spine.hs new file mode 100644 --- /dev/null +++ b/src/Codec/Epub/Opf/Package/Spine.hs @@ -0,0 +1,34 @@ +-- Copyright: 2010, 2011 Dino Morelli +-- License: BSD3 (see LICENSE) +-- Author: Dino Morelli + +{- | Data types for working with the metadata of ePub documents + + These data types were constructed by studying the IDPF OPF + specification for ePub documents found here: + + +-} +module Codec.Epub.Opf.Package.Spine + ( Spine (..) + , SpineItemref (..) + ) + where + +import Codec.Epub.Opf.Common + + +-- | package\/spine\/itemref tag +data SpineItemref = SpineItemref + { siIdRef :: MFItemId -- ^ idref attr. Must reference item in manifest + , siLinear :: Maybe Bool -- ^ linear attr + } + deriving (Eq, Show) + + +-- | package\/spine tag +data Spine = Spine + { spineToc :: MFItemId -- ^ toc attr. Must reference the NCX in the manifest + , spineItemrefs :: [ SpineItemref ] -- one required + } + deriving (Eq, Show) diff --git a/src/Codec/Epub/Opf/Parse.hs b/src/Codec/Epub/Opf/Parse.hs new file mode 100644 --- /dev/null +++ b/src/Codec/Epub/Opf/Parse.hs @@ -0,0 +1,270 @@ +-- Copyright: 2010, 2011 Dino Morelli +-- License: BSD3 (see LICENSE) +-- Author: Dino Morelli + +{-# LANGUAGE Arrows #-} +{-# LANGUAGE FlexibleContexts #-} + +-- | Module for extracting the metadata from an ePub file +module Codec.Epub.Opf.Parse + ( parseXmlToOpf + , parseEpubOpf + ) + where + +import Control.Applicative +import Control.Arrow.ListArrows +import Control.Monad.Error +import Data.Tree.NTree.TypeDefs ( NTree ) +import Text.XML.HXT.Arrow.Namespace ( propagateNamespaces ) +import Text.XML.HXT.Arrow.XmlArrow +import Text.XML.HXT.Arrow.XmlState ( no, runX, withValidate ) +import Text.XML.HXT.Arrow.ReadDocument ( readString ) +import Text.XML.HXT.DOM.TypeDefs + +import Codec.Epub.IO +import Codec.Epub.Opf.Package + + +-- HXT helpers + +atTag :: (ArrowXml a) => String -> a (NTree XNode) XmlTree +atTag tag = deep (isElem >>> hasName tag) + +atQTag :: (ArrowXml a) => QName -> a (NTree XNode) XmlTree +atQTag tag = deep (isElem >>> hasQName tag) + +text :: (ArrowXml a) => a (NTree XNode) String +text = getChildren >>> getText + +notNullA :: (ArrowList a) => a [b] [b] +notNullA = isA $ not . null + + +mbQTagText :: (ArrowXml a) => QName -> a (NTree XNode) (Maybe String) +mbQTagText tag = + ( atQTag tag >>> + text >>> notNullA >>> arr Just ) + `orElse` + (constA Nothing) + + +mbGetAttrValue :: (ArrowXml a) => + String -> a XmlTree (Maybe String) +mbGetAttrValue n = + (getAttrValue n >>> notNullA >>> arr Just) + `orElse` (constA Nothing) + +mbGetQAttrValue :: (ArrowXml a) => + QName -> a XmlTree (Maybe String) +mbGetQAttrValue qn = + (getQAttrValue qn >>> notNullA >>> arr Just) + `orElse` (constA Nothing) + + +{- ePub parsing helpers + + Note that these URIs could conceivably change in the future + Is it ok that they're hardcoded like this? + + Well, ok, the xml namespace URI will probably never change. +-} + +dcName, opfName, xmlName :: String -> QName +dcName local = mkQName "dc" local "http://purl.org/dc/elements/1.1/" +opfName local = mkQName "opf" local "http://www.idpf.org/2007/opf" +xmlName local = mkQName "xml" local "http://www.w3.org/XML/1998/namespace" + + +getPackage :: (ArrowXml a) => a (NTree XNode) (String, String) +getPackage = atTag "package" >>> + proc x -> do + v <- getAttrValue "version" -< x + u <- getAttrValue "unique-identifier" -< x + returnA -< (v, u) + + +getTitle :: (ArrowXml a) => a (NTree XNode) MetaTitle +getTitle = atQTag (dcName "title") >>> + proc x -> do + l <- mbGetQAttrValue (xmlName "lang") -< x + c <- text -< x + returnA -< MetaTitle l c + + +{- Since creators and contributors have the same exact XML structure, + this arrow is used to get either of them +-} +getCreator :: (ArrowXml a) => String -> a (NTree XNode) MetaCreator +getCreator tag = atQTag (dcName tag) >>> ( unwrapArrow $ MetaCreator + <$> (WrapArrow $ mbGetQAttrValue (opfName "role")) + <*> (WrapArrow $ mbGetQAttrValue (opfName "file-as")) + <*> (WrapArrow $ text) + ) + + +getSubject :: (ArrowXml a) => a (NTree XNode) String +getSubject = atQTag (dcName "subject") >>> text + + +getDescription :: (ArrowXml a) => a (NTree XNode) (Maybe String) +getDescription = mbQTagText $ dcName "description" + + +getPublisher :: (ArrowXml a) => a (NTree XNode) (Maybe String) +getPublisher = mbQTagText $ dcName "publisher" + + +getDate :: (ArrowXml a) => a (NTree XNode) MetaDate +getDate = atQTag (dcName "date") >>> + proc x -> do + e <- mbGetQAttrValue (opfName "event") -< x + c <- text -< x + returnA -< MetaDate e c + + +getType :: (ArrowXml a) => a (NTree XNode) (Maybe String) +getType = mbQTagText $ dcName "type" + + +getFormat :: (ArrowXml a) => a (NTree XNode) (Maybe String) +getFormat = mbQTagText $ dcName "format" + + +getId :: (ArrowXml a) => a (NTree XNode) MetaId +getId = atQTag (dcName "identifier") >>> + proc x -> do + mbi <- mbGetAttrValue "id" -< x + s <- mbGetQAttrValue (opfName "scheme") -< x + c <- text -< x + let i = maybe "[WARNING: missing required id attribute]" id mbi + returnA -< MetaId i s c + + +getSource :: (ArrowXml a) => a (NTree XNode) (Maybe String) +getSource = mbQTagText $ dcName "source" + + +getLang :: (ArrowXml a) => a (NTree XNode) String +getLang = atQTag (dcName "language") >>> text + + +getRelation :: (ArrowXml a) => a (NTree XNode) (Maybe String) +getRelation = mbQTagText $ dcName "relation" + + +getCoverage :: (ArrowXml a) => a (NTree XNode) (Maybe String) +getCoverage = mbQTagText $ dcName "coverage" + + +getRights :: (ArrowXml a) => a (NTree XNode) (Maybe String) +getRights = mbQTagText $ dcName "rights" + + +getMeta :: (ArrowXml a) => a (NTree XNode) Metadata +getMeta = atTag "metadata" >>> ( unwrapArrow $ Metadata + <$> (WrapArrow $ listA getTitle) + <*> (WrapArrow $ listA $ getCreator "creator") + <*> (WrapArrow $ listA $ getCreator "contributor") + <*> (WrapArrow $ listA getSubject) + <*> (WrapArrow $ getDescription) + <*> (WrapArrow $ getPublisher) + <*> (WrapArrow $ listA getDate) + <*> (WrapArrow $ getType) + <*> (WrapArrow $ getFormat) + <*> (WrapArrow $ listA getId) + <*> (WrapArrow $ getSource) + <*> (WrapArrow $ listA getLang) + <*> (WrapArrow $ getRelation) + <*> (WrapArrow $ getCoverage) + <*> (WrapArrow $ getRights) + ) + + +getManifestItem :: (ArrowXml a) => a (NTree XNode) ManifestItem +getManifestItem = atTag "item" >>> + proc x -> do + i <- getAttrValue "id" -< x + h <- getAttrValue "href" -< x + m <- getAttrValue "media-type" -< x + returnA -< ManifestItem i h m + + +getManifest :: (ArrowXml a) => a (NTree XNode) [ManifestItem] +getManifest = atTag "manifest" >>> + proc x -> do + l <- listA getManifestItem -< x + returnA -< l + + +getSpineItemref :: (ArrowXml a) => a (NTree XNode) SpineItemref +getSpineItemref = atTag "itemref" >>> + proc x -> do + i <- getAttrValue "idref" -< x + ml <- mbGetAttrValue "linear" -< x + let l = maybe Nothing (\v -> if v == "no" then Just False else Just True) ml + returnA -< SpineItemref i l + + +getSpine :: (ArrowXml a) => a (NTree XNode) Spine +getSpine = atTag "spine" >>> + proc x -> do + i <- getAttrValue "toc" -< x + l <- listA getSpineItemref -< x + returnA -< (Spine i l) + + +getGuideRef :: (ArrowXml a) => a (NTree XNode) GuideRef +getGuideRef = atTag "reference" >>> + proc x -> do + t <- getAttrValue "type" -< x + mt <- mbGetAttrValue "title" -< x + h <- getAttrValue "href" -< x + returnA -< GuideRef t mt h + + +getGuide :: (ArrowXml a) => a (NTree XNode) [GuideRef] +getGuide = atTag "guide" >>> + proc x -> do + l <- listA getGuideRef -< x + returnA -< l + + +getBookData :: (ArrowXml a) => a (NTree XNode) Package +getBookData = + proc x -> do + (v, u) <- getPackage -< x + m <- getMeta -< x + mf <- getManifest -< x + sp <- getSpine -< x + gl <- listA getGuide -< x + let g = case gl of + [] -> [] + [e] -> e + _ -> error "ERROR: more than one guide entries" + returnA -< (Package v u m mf sp g) + + +{- | Extract the ePub OPF Package data contained in the supplied + XML string +-} +parseXmlToOpf :: (MonadIO m) => String -> m [Package] +parseXmlToOpf opfContents = + liftIO $ runX ( + readString [withValidate no] opfContents + >>> propagateNamespaces + >>> getBookData + ) + + +-- | Given the path to an ePub file, extract the OPF Package data +parseEpubOpf :: (MonadIO m, MonadError String m) => + FilePath -> m Package +parseEpubOpf zipPath = do + opfContents <- extractFileFromZip zipPath =<< opfPath zipPath + result <- parseXmlToOpf opfContents + + case result of + (em : []) -> return em + _ -> throwError + "ERROR: Parse didn't result in a single document metadata" diff --git a/src/Codec/Epub/Opf/epub-meta.hs b/src/Codec/Epub/Opf/epub-meta.hs deleted file mode 100644 --- a/src/Codec/Epub/Opf/epub-meta.hs +++ /dev/null @@ -1,24 +0,0 @@ --- Copyright: 2010 Dino Morelli --- License: BSD3 (see LICENSE) --- Author: Dino Morelli - -import Control.Monad.Error -import System.Environment ( getArgs ) -import System.Exit - -import Codec.Epub.Opf.Metadata.Format -import Codec.Epub.Opf.Metadata.Parse - - -main :: IO () -main = do - as <- getArgs - - when (length as /= 1) $ do - putStrLn "usage: epub-meta EPUBFILE" - exitWith $ ExitFailure 1 - - let zipPath = head as - result <- runErrorT $ parseEpubOpf zipPath - - putStr $ either (++ "\n") opfToString result diff --git a/testsuite/runtests.hs b/testsuite/runtests.hs --- a/testsuite/runtests.hs +++ b/testsuite/runtests.hs @@ -1,4 +1,4 @@ --- Copyright: 2010 Dino Morelli +-- Copyright: 2010, 2011 Dino Morelli -- License: BSD3 (see LICENSE) -- Author: Dino Morelli @@ -6,8 +6,8 @@ import Test.HUnit ( Counts, Test (..), assertEqual, runTestTT ) import Test.HUnit.Base ( Assertion ) -import Codec.Epub.Opf.Metadata -import Codec.Epub.Opf.Metadata.Parse +import Codec.Epub.Opf.Package +import Codec.Epub.Opf.Parse main = runTestTT tests >> return () @@ -28,55 +28,97 @@ testFull = TestCase $ do xmlString <- readFile $ "testsuite" "testFull.opf" actual <- parseXmlToOpf xmlString - let expected = [ OPFPackage "2.0" "isbn" ( EpubMeta - { emTitles = - [ EMTitle Nothing "Title Of This Book" - , EMTitle (Just "fr") "Titre De Ce Livre" - ] - , emCreators = - [ EMCreator - (Just "aut") - (Just "Wiggins, Josephine B.") - "Josephine B. Wiggins" - , EMCreator - (Just "aut") - Nothing - "Horatio Cromwell" - , EMCreator - Nothing - Nothing - "Natalia Jenkins" - ] - , emContributors = - [ EMCreator - (Just "ill") - (Just "Knickerbocker, Reginald Q.") - "Reginald Q. Knickerbocker" - , EMCreator - (Just "edt") - Nothing - "Beverly Abercrombie" - ] - , emSubjects = ["Fiction", "Science Fiction"] - , emDescription = Just "This document is a stub used for unit testing. It is missing the rest of the tags that normally occur after metadata." - , emPublisher = Just "Fictional Books Ltd." - , emDates = - [ EMDate (Just "published") "2010" - , EMDate (Just "created") "2010-05-07" - , EMDate (Just "modified") "2010-05-08T10:20:57" - , EMDate Nothing "2009-08-03T16:22:20" - ] - , emType = Just "test OPF Package Document" - , emFormat = Just "ePub publication" - , emIds = - [ EMId "isbn" (Just "ISBN") "1-82057-821-9" - , EMId "other" Nothing "1386506873266"] - , emSource = Just "document source" - , emLangs = ["en-us"] - , emRelation = Just "document relation" - , emCoverage = Just "coverage information" - , emRights = Just "Copyright: 2010 Dino Morelli, License: BSD3" - } ) ] + let expected = + [ Package + { opVersion = "2.0" + , opUniqueId = "isbn" + , opMeta = Metadata + { metaTitles = + [ MetaTitle Nothing "Title Of This Book" + , MetaTitle (Just "fr") "Titre De Ce Livre" + ] + , metaCreators = + [ MetaCreator + (Just "aut") + (Just "Wiggins, Josephine B.") + "Josephine B. Wiggins" + , MetaCreator + (Just "aut") + Nothing + "Horatio Cromwell" + , MetaCreator + Nothing + Nothing + "Natalia Jenkins" + ] + , metaContributors = + [ MetaCreator + (Just "ill") + (Just "Knickerbocker, Reginald Q.") + "Reginald Q. Knickerbocker" + , MetaCreator + (Just "edt") + Nothing "Beverly Abercrombie" + ] + , metaSubjects = ["Fiction","Science Fiction"] + , metaDescription = Just "This document is a stub used for unit testing. It is missing the rest of the tags that normally occur after metadata." + , metaPublisher = Just "Fictional Books Ltd." + , metaDates = + [ MetaDate (Just "published") "2010" + , MetaDate (Just "created") "2010-05-07" + , MetaDate (Just "modified") "2010-05-08T10:20:57" + , MetaDate Nothing "2009" + ] + , metaType = Just "test OPF Package Document" + , metaFormat = Just "ePub publication" + , metaIds = + [ MetaId "isbn" (Just "ISBN") "1-82057-821-9" + , MetaId "other" Nothing "1386506873266" + ] + , metaSource = Just "document source" + , metaLangs = ["en-us"] + , metaRelation = Just "document relation" + , metaCoverage = Just "coverage information" + , metaRights = Just "Copyright: 2010 Dino Morelli, License: BSD3" + } + , opManifest = + [ ManifestItem + { mfiId = "ncx" + , mfiHref = "toc.ncx" + , mfiMediaType = "application/x-dtbncx+xml" + } + , ManifestItem + { mfiId = "titlePage" + , mfiHref = "content/titlePage.html" + , mfiMediaType = "application/xhtml+xml" + } + , ManifestItem + { mfiId = "someContent" + , mfiHref = "content/someContent.html" + , mfiMediaType = "application/xhtml+xml" + } + ] + , opSpine = Spine + { spineToc = "ncx" + , spineItemrefs = + [ SpineItemref {siIdRef = "titlePage", siLinear = Nothing} + , SpineItemref {siIdRef = "someContent", siLinear = Nothing} + ] + } + , opGuide = + [ GuideRef + { grType = "title-page" + , grTitle = Just "Title page" + , grHref = "content/titlePage.html" + } + , GuideRef + { grType = "text" + , grTitle = Just "Title Of This Book" + , grHref = "content/someContent.html" + } + ] + } + ] assertEqual "very full" expected actual @@ -87,23 +129,38 @@ testMinimal = TestCase $ do xmlString <- readFile $ "testsuite" "testMinimal.opf" actual <- parseXmlToOpf xmlString - let expected = [ OPFPackage "2.0" "isbn" ( EpubMeta - { emTitles = [EMTitle Nothing "Title Of This Book"] - , emCreators = [] - , emContributors = [] - , emSubjects = [] - , emDescription = Nothing - , emPublisher = Nothing - , emDates = [] - , emType = Nothing - , emFormat = Nothing - , emIds = [EMId "isbn" (Just "ISBN") "1-82057-821-9"] - , emSource = Nothing - , emLangs = ["en-us"] - , emRelation = Nothing - , emCoverage = Nothing - , emRights = Nothing - } ) ] + let expected = + [ Package + { opVersion = "2.0" + , opUniqueId = "isbn" + , opMeta = Metadata + { metaTitles = [MetaTitle Nothing "Title Of This Book"] + , metaCreators = [] + , metaContributors = [] + , metaSubjects = [] + , metaDescription = Nothing + , metaPublisher = Nothing + , metaDates = [] + , metaType = Nothing + , metaFormat = Nothing + , metaIds = [MetaId "isbn" (Just "ISBN") "1-82057-821-9"] + , metaSource = Nothing + , metaLangs = ["en-us"] + , metaRelation = Nothing + , metaCoverage = Nothing + , metaRights = Nothing + } + , opManifest = + [ ManifestItem + { mfiId = "ncx" + , mfiHref = "toc.ncx" + , mfiMediaType = "application/x-dtbncx+xml" + } + ] + , opSpine = Spine {spineToc = "ncx", spineItemrefs = []} + , opGuide = [] + } + ] assertEqual "minimal" expected actual @@ -114,5 +171,14 @@ testMissingAll = TestCase $ do xmlString <- readFile $ "testsuite" "testMissingAll.opf" actual <- parseXmlToOpf xmlString - let expected = [ OPFPackage "" "" emptyEpubMeta ] + let expected = + [ Package + { opVersion = "" + , opUniqueId = "" + , opMeta = emptyMetadata + , opManifest = [] + , opSpine = Spine {spineToc = "", spineItemrefs = []} + , opGuide = [] + } + ] assertEqual "missing all" expected actual diff --git a/testsuite/testFull.opf b/testsuite/testFull.opf --- a/testsuite/testFull.opf +++ b/testsuite/testFull.opf @@ -19,7 +19,7 @@ 2010 2010-05-07 2010-05-08T10:20:57 - 2009-08-03T16:22:20 + 2009 en-us Fictional Books Ltd. 1386506873266 @@ -31,9 +31,30 @@ document relation coverage information Copyright: 2010 Dino Morelli, License: BSD3 - - + + + + + + + + + + + + + + + + diff --git a/testsuite/testMinimal.opf b/testsuite/testMinimal.opf --- a/testsuite/testMinimal.opf +++ b/testsuite/testMinimal.opf @@ -12,6 +12,13 @@ en-us - + + + + + + + diff --git a/testsuite/testMissingAll.opf b/testsuite/testMissingAll.opf --- a/testsuite/testMissingAll.opf +++ b/testsuite/testMissingAll.opf @@ -6,6 +6,10 @@ xmlns:xml="http://www.w3.org/XML/1998/namespace"> - + + + + + + diff --git a/util/show-opf.hs b/util/show-opf.hs new file mode 100644 --- /dev/null +++ b/util/show-opf.hs @@ -0,0 +1,13 @@ +#! /usr/bin/runhaskell -isrc + +-- Copyright: 2010, 2011 Dino Morelli +-- License: BSD3 (see LICENSE) +-- Author: Dino Morelli + +import System.Environment ( getArgs ) + +import Codec.Epub.Opf.Parse + + +main :: IO () +main = (fmap head) getArgs >>= readFile >>= parseXmlToOpf >>= print