diff --git a/LICENSE b/LICENSE --- a/LICENSE +++ b/LICENSE @@ -1,4 +1,4 @@ -Copyright Dino Morelli 2010-2012 +Copyright Dino Morelli 2010-2013 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: 2010-2012 Dino Morelli +-- Copyright: 2010-2013 Dino Morelli -- License: BSD3 (see LICENSE) -- Author: Dino Morelli diff --git a/epub-metadata.cabal b/epub-metadata.cabal --- a/epub-metadata.cabal +++ b/epub-metadata.cabal @@ -1,10 +1,10 @@ name: epub-metadata -version: 2.3.1 +version: 2.3.2 cabal-version: >= 1.8 build-type: Simple license: BSD3 license-file: LICENSE -copyright: 2010-2012 Dino Morelli +copyright: 2010-2013 Dino Morelli author: Dino Morelli maintainer: Dino Morelli stability: stable diff --git a/src/Codec/Epub/Archive.hs b/src/Codec/Epub/Archive.hs --- a/src/Codec/Epub/Archive.hs +++ b/src/Codec/Epub/Archive.hs @@ -1,4 +1,4 @@ --- Copyright: 2011-2012 Dino Morelli +-- Copyright: 2011-2013 Dino Morelli -- License: BSD3 (see LICENSE) -- Author: Dino Morelli 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: 2010-2012 Dino Morelli +-- Copyright: 2010-2013 Dino Morelli -- License: BSD3 (see LICENSE) -- Author: Dino Morelli @@ -10,6 +10,7 @@ ( opfContentsFromZip , opfContentsFromBS , opfContentsFromDir + , removeIllegalStartChars , removeEncoding , removeDoctype ) @@ -30,10 +31,17 @@ import Text.XML.HXT.Arrow.ReadDocument ( readString ) +-- | An evil hack to remove *ILLEGAL* characters before the XML +-- | declaration. Why do people write software that does this? +-- | Can't they follow directions? +removeIllegalStartChars :: String -> String +removeIllegalStartChars = dropWhile (/= '<') + + -- | An evil hack to remove encoding from the document removeEncoding :: String -> String removeEncoding = flip (subRegex - (mkRegexWithOpts " +encoding=\"UTF-8\"" False True)) "" + (mkRegexWithOpts " +encoding=\"UTF-8\"" False False)) "" -- | An evil hack to remove any from the document diff --git a/src/Codec/Epub/Opf/Common.hs b/src/Codec/Epub/Opf/Common.hs --- a/src/Codec/Epub/Opf/Common.hs +++ b/src/Codec/Epub/Opf/Common.hs @@ -1,4 +1,4 @@ --- Copyright: 2010-2012 Dino Morelli +-- Copyright: 2010-2013 Dino Morelli -- License: BSD3 (see LICENSE) -- Author: Dino Morelli diff --git a/src/Codec/Epub/Opf/Format/Guide.hs b/src/Codec/Epub/Opf/Format/Guide.hs --- a/src/Codec/Epub/Opf/Format/Guide.hs +++ b/src/Codec/Epub/Opf/Format/Guide.hs @@ -1,4 +1,4 @@ --- Copyright: 2010-2012 Dino Morelli +-- Copyright: 2010-2013 Dino Morelli -- License: BSD3 (see LICENSE) -- Author: Dino Morelli diff --git a/src/Codec/Epub/Opf/Format/Manifest.hs b/src/Codec/Epub/Opf/Format/Manifest.hs --- a/src/Codec/Epub/Opf/Format/Manifest.hs +++ b/src/Codec/Epub/Opf/Format/Manifest.hs @@ -1,4 +1,4 @@ --- Copyright: 2010-2012 Dino Morelli +-- Copyright: 2010-2013 Dino Morelli -- License: BSD3 (see LICENSE) -- Author: Dino Morelli diff --git a/src/Codec/Epub/Opf/Format/Metadata.hs b/src/Codec/Epub/Opf/Format/Metadata.hs --- a/src/Codec/Epub/Opf/Format/Metadata.hs +++ b/src/Codec/Epub/Opf/Format/Metadata.hs @@ -1,4 +1,4 @@ --- Copyright: 2010-2012 Dino Morelli +-- Copyright: 2010-2013 Dino Morelli -- License: BSD3 (see LICENSE) -- Author: Dino Morelli diff --git a/src/Codec/Epub/Opf/Format/Package.hs b/src/Codec/Epub/Opf/Format/Package.hs --- a/src/Codec/Epub/Opf/Format/Package.hs +++ b/src/Codec/Epub/Opf/Format/Package.hs @@ -1,4 +1,4 @@ --- Copyright: 2010-2012 Dino Morelli +-- Copyright: 2010-2013 Dino Morelli -- License: BSD3 (see LICENSE) -- Author: Dino Morelli diff --git a/src/Codec/Epub/Opf/Format/Spine.hs b/src/Codec/Epub/Opf/Format/Spine.hs --- a/src/Codec/Epub/Opf/Format/Spine.hs +++ b/src/Codec/Epub/Opf/Format/Spine.hs @@ -1,4 +1,4 @@ --- Copyright: 2010-2012 Dino Morelli +-- Copyright: 2010-2013 Dino Morelli -- License: BSD3 (see LICENSE) -- Author: Dino Morelli diff --git a/src/Codec/Epub/Opf/Format/Util.hs b/src/Codec/Epub/Opf/Format/Util.hs --- a/src/Codec/Epub/Opf/Format/Util.hs +++ b/src/Codec/Epub/Opf/Format/Util.hs @@ -1,4 +1,4 @@ --- Copyright: 2010-2012 Dino Morelli +-- Copyright: 2010-2013 Dino Morelli -- License: BSD3 (see LICENSE) -- Author: Dino Morelli diff --git a/src/Codec/Epub/Opf/Package.hs b/src/Codec/Epub/Opf/Package.hs --- a/src/Codec/Epub/Opf/Package.hs +++ b/src/Codec/Epub/Opf/Package.hs @@ -1,4 +1,4 @@ --- Copyright: 2010-2012 Dino Morelli +-- Copyright: 2010-2013 Dino Morelli -- License: BSD3 (see LICENSE) -- Author: Dino Morelli diff --git a/src/Codec/Epub/Opf/Package/Guide.hs b/src/Codec/Epub/Opf/Package/Guide.hs --- a/src/Codec/Epub/Opf/Package/Guide.hs +++ b/src/Codec/Epub/Opf/Package/Guide.hs @@ -1,4 +1,4 @@ --- Copyright: 2010-2012 Dino Morelli +-- Copyright: 2010-2013 Dino Morelli -- License: BSD3 (see LICENSE) -- Author: Dino Morelli diff --git a/src/Codec/Epub/Opf/Package/Manifest.hs b/src/Codec/Epub/Opf/Package/Manifest.hs --- a/src/Codec/Epub/Opf/Package/Manifest.hs +++ b/src/Codec/Epub/Opf/Package/Manifest.hs @@ -1,4 +1,4 @@ --- Copyright: 2010-2012 Dino Morelli +-- Copyright: 2010-2013 Dino Morelli -- License: BSD3 (see LICENSE) -- Author: Dino Morelli diff --git a/src/Codec/Epub/Opf/Package/Metadata.hs b/src/Codec/Epub/Opf/Package/Metadata.hs --- a/src/Codec/Epub/Opf/Package/Metadata.hs +++ b/src/Codec/Epub/Opf/Package/Metadata.hs @@ -1,4 +1,4 @@ --- Copyright: 2010-2012 Dino Morelli +-- Copyright: 2010-2013 Dino Morelli -- License: BSD3 (see LICENSE) -- Author: Dino Morelli diff --git a/src/Codec/Epub/Opf/Package/Spine.hs b/src/Codec/Epub/Opf/Package/Spine.hs --- a/src/Codec/Epub/Opf/Package/Spine.hs +++ b/src/Codec/Epub/Opf/Package/Spine.hs @@ -1,4 +1,4 @@ --- Copyright: 2010-2012 Dino Morelli +-- Copyright: 2010-2013 Dino Morelli -- License: BSD3 (see LICENSE) -- Author: Dino Morelli diff --git a/src/Codec/Epub/Opf/Parse.hs b/src/Codec/Epub/Opf/Parse.hs --- a/src/Codec/Epub/Opf/Parse.hs +++ b/src/Codec/Epub/Opf/Parse.hs @@ -1,4 +1,4 @@ --- Copyright: 2010-2012 Dino Morelli +-- Copyright: 2010-2013 Dino Morelli -- License: BSD3 (see LICENSE) -- Author: Dino Morelli @@ -255,7 +255,8 @@ parseXmlToOpf contents = do {- Improper encoding and schema declarations have been causing havok with this parse, cruelly strip them out. -} - let cleanedContents = removeEncoding . removeDoctype $ contents + let cleanedContents = removeIllegalStartChars . removeEncoding + . removeDoctype $ contents result <- liftIO $ runX ( readString [withValidate no] cleanedContents diff --git a/testsuite/opf-parse.hs b/testsuite/opf-parse.hs --- a/testsuite/opf-parse.hs +++ b/testsuite/opf-parse.hs @@ -1,4 +1,4 @@ --- Copyright: 2010-2012 Dino Morelli +-- Copyright: 2010-2013 Dino Morelli -- License: BSD3 (see LICENSE) -- Author: Dino Morelli @@ -33,6 +33,7 @@ , testMinimal , testMissingAll , testDamagedZip + , testIllegalCharsBeforeDecl ] @@ -206,3 +207,46 @@ actual <- runErrorT $ opfContentsFromZip $ "testsuite" "damagedZipCentralDir.epub" actual @?= Left "Did not find end of central directory signature. Failed reading at byte position 138" + + +{- Found books coming from Barnes & Noble (for their NOOK reader) to + contain illegal characters before the XML declaration. This is + strictly not allowed by the XML specification. I am very + disappointed with Barnes & Noble for selling garbage like this. +-} +testIllegalCharsBeforeDecl :: Test +testIllegalCharsBeforeDecl = TestCase $ do + xmlString <- readFile $ + "testsuite" "testIllegalCharsBeforeDecl.opf" + actual <- runErrorT $ parseXmlToOpf xmlString + let expected = + Right Package + { opVersion = "2.0" + , opUniqueId = "uuid_id" + , opMeta = Metadata + { metaTitles = [MetaTitle Nothing "Foo Bar Baz"] + , metaCreators = [] + , metaContributors = [] + , metaSubjects = [] + , metaDescription = Nothing + , metaPublisher = Nothing + , metaDates = [] + , metaType = Nothing + , metaFormat = Nothing + , metaIds = [MetaId "uuid_id" (Just "uuid") "1122334455"] + , metaSource = Nothing + , metaLangs = ["en"] + , 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 "illegal chars before XML declaration" expected actual diff --git a/testsuite/testIllegalCharsBeforeDecl.opf b/testsuite/testIllegalCharsBeforeDecl.opf new file mode 100644 --- /dev/null +++ b/testsuite/testIllegalCharsBeforeDecl.opf @@ -0,0 +1,18 @@ + + + + en + Foo Bar Baz + 1122334455 + + + + + + + + + +