diff --git a/README.md b/README.md --- a/README.md +++ b/README.md @@ -70,9 +70,9 @@ ## Getting source - Download the cabalized source package [from Hackage](http://hackage.haskell.org/package/epub-tools) -- Download the cabalized source tarball [from here](http://ui3.info/d/proj/epub-tools/epub-tools-2.4.tar.gz) +- Download the cabalized source tarball [from here](http://ui3.info/d/proj/epub-tools/epub-tools-2.5.tar.gz) - epub-tools is available for Arch Linux [from the AUR](http://aur.archlinux.org/packages/epub-tools/) -- Download [binaries for Windows](http://ui3.info/d/proj/epub-tools/epub-tools-2.4-win.zip) +- Download [binaries for Windows](http://ui3.info/d/proj/epub-tools/epub-tools-2.5-win.zip) - Get the source with darcs: `$ darcs get http://ui3.info/darcs/epub-tools` - If you're just looking, [browse the source](http://ui3.info/darcs/epub-tools) diff --git a/TODO b/TODO --- a/TODO +++ b/TODO @@ -3,8 +3,6 @@ - Some notes in README about installing epubcheck if possible -- Add more (all?) of the darcs-tracked files to cabal so they're part of the sdist - - Would like to way to be able to insert a publishing year tag into the OPF data like these: 2011 2011 diff --git a/changelog.md b/changelog.md --- a/changelog.md +++ b/changelog.md @@ -1,3 +1,17 @@ +2.5 (2014-04-04) + + * Additions and modifications to the stock rules to both support + more books and also use more generic rules than before + + * Added code to set proper case for Roman numerals in titles + + * Added code to handle file-as names with parenthesized name info + + * Simplified and consolidated special-character filtering code + + * Added support for multiline fields in the Metadata + + 2.4 (2014-03-19) * Fixed problem in Windows cmd shell with missing UNIX HOME diff --git a/epub-tools.cabal b/epub-tools.cabal --- a/epub-tools.cabal +++ b/epub-tools.cabal @@ -1,6 +1,6 @@ name: epub-tools cabal-version: >= 1.8 -version: 2.4 +version: 2.5 build-type: Simple license: BSD3 license-file: LICENSE diff --git a/src/EpubTools/EpubMeta/Opts.hs b/src/EpubTools/EpubMeta/Opts.hs --- a/src/EpubTools/EpubMeta/Opts.hs +++ b/src/EpubTools/EpubMeta/Opts.hs @@ -105,5 +105,5 @@ , " epub3: http://www.idpf.org/epub/30/spec/epub30-publications.html" , "" , "" - , "Version 2.4 Dino Morelli " + , "Version 2.5 Dino Morelli " ] diff --git a/src/EpubTools/EpubName/Doc/Rules.hs b/src/EpubTools/EpubName/Doc/Rules.hs --- a/src/EpubTools/EpubName/Doc/Rules.hs +++ b/src/EpubTools/EpubName/Doc/Rules.hs @@ -63,13 +63,9 @@ , " titlePat \".* Magazine Issue ([0-9]+) ::.*\"" , " name \"GUDMagazine(pad 2 1)\"" , "" - , "magInterzone_old" - , " titlePat \"^(Interzone)[^0-9]*([0-9]+)$\"" - , " name \"(1)SFF(pad 3 2)\"" - , "" , "magInterzone" - , " titlePat \"^(Interzone) ([0-9]+).*$\"" - , " name \"(1)SFF(pad 3 2)\"" + , " titlePat \"^(Interzone|INTERZONE)[^0-9]+([0-9]+).*\"" + , " name \"(scrub 1)SFF(pad 3 2)\"" , "" , "magLightspeed_date" , " titlePat \"(Lightspeed) Magazine, (.*) (.*)\"" @@ -91,10 +87,6 @@ , " titlePat \"^(Something Wicked)[^0-9]*([0-9]+)\"" , " name \"(scrub 1)(pad 3 2)\"" , "" - , "magUniverse" - , " titlePat \"^(Jim Baen's Universe)-Vol ([^ ]+) Num ([^ ]+)\"" - , " name \"(scrub 1)Vol(pad 2 2)Num(pad 2 3)\"" - , "" , "magWeirdTales" , " titlePat \"^(Weird Tales)[^0-9]*([0-9]+)$\"" , " name \"(scrub 1)(pad 3 2)\"" @@ -106,6 +98,24 @@ , "magGenericWithIssue" , " titlePat \"(.* Magazine):? Issue ([0-9]+).*\"" , " name \"(scrub 1)(pad 2 2)\"" + , "" + , "magGenericSubjWithIssue" + , " subjectMatch \"magazine\"" + , " titlePat \"(.*) Issue ([0-9]+).*\"" + , " name \"(scrub 1)Magazine(pad 2 2)\"" + , "" + , "magGenericVolNo3" + , " titlePat \"(.*)Vol(ume)? ([0-9]{1,2}),? (No|Num)[^0-9]+([0-9]{3}).*\"" + , " name \"(scrub 1)Vol(pad 2 3)No(pad 3 5)\"" + , "" + , "magGenericVolNo2" + , " titlePat \"(.*)Vol(ume)? ([0-9]{1,2}),? (No|Num)[^0-9]+([0-9]{1,2})[^0-9]?.*\"" + , " name \"(scrub 1)Vol(pad 2 3)No(pad 2 5)\"" + , "" + , "nonficWomensInstituteLibrary" + , " authorMatch \"Woman's Institute\"" + , " titlePat \"(.*)\"" + , " name \"(scrub 1)\"" , "" , "anthology" , " subjectMatch \"anthology\"" diff --git a/src/EpubTools/EpubName/Format/Author.hs b/src/EpubTools/EpubName/Format/Author.hs --- a/src/EpubTools/EpubName/Format/Author.hs +++ b/src/EpubTools/EpubName/Format/Author.hs @@ -92,7 +92,8 @@ authorSingle :: [String] -> String authorSingle (last:rest:_) = printf "%s%s-" (scrubString last) (scrubString rest) -authorSingle [n] = printf "%s-" $ scrubString n +authorSingle [n] = + printf "%s-" $ scrubString $ takeWhile (/= '(') n authorSingle _ = undefined diff --git a/src/EpubTools/EpubName/Format/Format.hs b/src/EpubTools/EpubName/Format/Format.hs --- a/src/EpubTools/EpubName/Format/Format.hs +++ b/src/EpubTools/EpubName/Format/Format.hs @@ -161,7 +161,7 @@ [] -> throwError "format failed, no title present" ts -> return . head $ ts - case matchRegex (mkRegex re) oldTitle of + case matchRegex (mkRegexWithOpts re False True) oldTitle of Just matches -> return matches Nothing -> throwError $ printf "extract title failed: %s" re diff --git a/src/EpubTools/EpubName/Format/Util.hs b/src/EpubTools/EpubName/Format/Util.hs --- a/src/EpubTools/EpubName/Format/Util.hs +++ b/src/EpubTools/EpubName/Format/Util.hs @@ -49,6 +49,9 @@ camelCase :: String -> String camelCase s = concat $ map capFirst $ words s where + capFirst s'@('I':'I':_) = map toUpper s' + capFirst s'@('I':'V':_) = map toUpper s' + capFirst s'@('I':'X':_) = map toUpper s' capFirst (first:rest) = (toUpper first) : (map toLower rest) capFirst _ = undefined @@ -58,18 +61,16 @@ -} aggressiveFilters :: [(String -> String)] aggressiveFilters = - [ repl "[',\\?();#’]" "" - , repl "\\." " " - , repl ":" "_" + [ repl "[',\\?();#’\\*!]" "" + , repl "]" "" -- Tricky to put this in [ ] + , repl "[./]" " " + , repl ":" "_" + , repl "[\\[\\\n]" "_ " + , repl "-" "- " , filter (/= '"') - , repl "]" "" - , repl "\\*" "" - , repl "!" "" - , repl "[-/]" " " - , repl "\\[" "_ " -- Decided that I like the article included in titles - --, repl "^The " "" - , repl "&" " And " + --, repl "^The " "" + , repl "&" " And " , camelCase ] diff --git a/src/EpubTools/EpubName/Opts.hs b/src/EpubTools/EpubName/Opts.hs --- a/src/EpubTools/EpubName/Opts.hs +++ b/src/EpubTools/EpubName/Opts.hs @@ -219,5 +219,5 @@ , " epub3: http://www.idpf.org/epub/30/spec/epub30-publications.html" , "" , "" - , "Version 2.4 Dino Morelli " + , "Version 2.5 Dino Morelli " ] diff --git a/src/EpubTools/EpubZip/Opts.hs b/src/EpubTools/EpubZip/Opts.hs --- a/src/EpubTools/EpubZip/Opts.hs +++ b/src/EpubTools/EpubZip/Opts.hs @@ -70,5 +70,5 @@ , " epub3: http://www.idpf.org/epub/30/spec/epub30-publications.html" , "" , "" - , "Version 2.4 Dino Morelli " + , "Version 2.5 Dino Morelli " ] diff --git a/testsuite/EpubTools/Test/EpubName/Format.hs b/testsuite/EpubTools/Test/EpubName/Format.hs --- a/testsuite/EpubTools/Test/EpubName/Format.hs +++ b/testsuite/EpubTools/Test/EpubName/Format.hs @@ -29,17 +29,22 @@ , testAuthorOneName , testAuthorRole , testAuthorFileas + , testAuthorFileasParens , testAuthorFull , testMultiAutCreators , testMultiAutOneString , testNoAuthor + , testColon , testCreatorsNoAuthor , testCreatorsNoAuthorPubDate , testCreatorLastFirst - , testCapsTitle - , testColon - , testBracketTitle - , testNoTitle + , testTitleCaps + , testTitleBracket + , testTitleNone + , testTitleMultiline + , testTitleHyphen + , testTitleHyphenDates + , testTitleRomanNum , testAllPunctuation , testPubYearNoDatesPresent , testPubYearEpub3 @@ -66,9 +71,10 @@ , testGudShort , testGudLong , testGudVeryLong - , testInterzoneOldShort - , testInterzoneOldLong , testInterzone + , testInterzoneCaps + , testInterzoneOldLong + , testInterzoneOldShort , testNemesisShort , testNemesisLong , testMagSomethingWicked @@ -93,6 +99,10 @@ , testMagGalaxysEdge , testMagPlasmaFreq , testMagPlasmaFreqMonth + , testMagPunchinello1800s + , testMagGenericVolNo3 + , testMagSubjWithIssue + , testWomensInstituteLib , testAnthology ] @@ -168,6 +178,24 @@ ) +testAuthorFileasParens :: (Globals, [Formatter]) -> Test +testAuthorFileasParens (gs, fs) = TestCase $ + assertNewName gs { gMetadata = meta } fs + "author with file-as and pesky parenthesized full names" expected + where + meta = emptyMetadata + { metaCreators = [Creator Nothing + (Just "Fletcher, J. S. (Joseph Smith)") + Nothing + "J. S. Fletcher"] + , metaTitles = [Title Nothing Nothing Nothing "The Middle of Things"] + } + expected = + ( "ordinary_book" + , "FletcherJS-TheMiddleOfThings.epub" + ) + + testAuthorFull :: (Globals, [Formatter]) -> Test testAuthorFull (gs, fs) = TestCase $ assertNewName gs { gMetadata = meta } fs "author fully filled out" expected @@ -258,6 +286,23 @@ ) +testColon :: (Globals, [Formatter]) -> Test +testColon (gs, fs) = TestCase $ + assertNewName gs { gMetadata = meta } fs + "colon becomes underscore" expected + where + meta = emptyMetadata + { metaCreators = [Creator Nothing Nothing Nothing + "Ed Howdershelt"] + , metaTitles = [Title Nothing Nothing Nothing + "Book 1: 3rd World Products, Inc."] + } + expected = + ( "ordinary_book" + , "HowdersheltEd-Book1_3rdWorldProductsInc.epub" + ) + + testCreatorsNoAuthorPubDate :: (Globals, [Formatter]) -> Test testCreatorsNoAuthorPubDate (gs, fs) = TestCase $ assertNewName gs { gMetadata = meta } fs @@ -300,8 +345,8 @@ ) -testCapsTitle :: (Globals, [Formatter]) -> Test -testCapsTitle (gs, fs) = TestCase $ +testTitleCaps :: (Globals, [Formatter]) -> Test +testTitleCaps (gs, fs) = TestCase $ assertNewName gs { gMetadata = meta } fs "title all caps" expected where meta = emptyMetadata @@ -314,25 +359,8 @@ ) -testColon :: (Globals, [Formatter]) -> Test -testColon (gs, fs) = TestCase $ - assertNewName gs { gMetadata = meta } fs - "colon becomes underscore" expected - where - meta = emptyMetadata - { metaCreators = [Creator Nothing Nothing Nothing - "Ed Howdershelt"] - , metaTitles = [Title Nothing Nothing Nothing - "Book 1: 3rd World Products, Inc."] - } - expected = - ( "ordinary_book" - , "HowdersheltEd-Book1_3rdWorldProductsInc.epub" - ) - - -testBracketTitle :: (Globals, [Formatter]) -> Test -testBracketTitle (gs, fs) = TestCase $ +testTitleBracket :: (Globals, [Formatter]) -> Test +testTitleBracket (gs, fs) = TestCase $ assertNewName gs { gMetadata = meta } fs "title with brackets" expected where @@ -347,8 +375,8 @@ ) -testNoTitle :: (Globals, [Formatter]) -> Test -testNoTitle (gs, fs) = TestCase $ +testTitleNone :: (Globals, [Formatter]) -> Test +testTitleNone (gs, fs) = TestCase $ assertNewName gs { gMetadata = meta } fs "missing title" expected where meta = emptyMetadata @@ -361,6 +389,85 @@ ) +testTitleMultiline :: (Globals, [Formatter]) -> Test +testTitleMultiline (gs, fs) = TestCase $ + assertNewName gs { gMetadata = meta } fs + "multiline title" expected + where + meta = emptyMetadata + { metaCreators = [Creator + Nothing + (Just "Castleman, Virginia Carter") + Nothing + "Virginia Carter Castleman"] + , metaTitles = [Title Nothing Nothing Nothing + "Pocahontas.\nA Poem"] + } + expected = + ( "ordinary_book" + , "CastlemanVirginiaCarter-Pocahontas_APoem.epub" + ) + + +testTitleHyphen :: (Globals, [Formatter]) -> Test +testTitleHyphen (gs, fs) = TestCase $ + assertNewName gs { gMetadata = meta } fs + "title with hyphen" expected + where + meta = emptyMetadata + { metaCreators = [Creator + Nothing + (Just "Spinoza, Benedictus de") + Nothing + "Benedictus de Spinoza"] + , metaTitles = [Title Nothing Nothing Nothing + "Theologico-Political Treatise"] + } + expected = + ( "ordinary_book" + , "SpinozaBenedictusDe-Theologico-PoliticalTreatise.epub" + ) + + +testTitleHyphenDates :: (Globals, [Formatter]) -> Test +testTitleHyphenDates (gs, fs) = TestCase $ + assertNewName gs { gMetadata = meta } fs + "title with hyphen separating dates" expected + where + meta = emptyMetadata + { metaCreators = [Creator + Nothing + (Just "Anonymous") + Nothing + "Anonymous"] + , metaTitles = [Title Nothing Nothing Nothing + "Queen Victoria\nStory of Her Life and Reign, 1819-1901"] + } + expected = + ( "ordinary_book" + , "Anonymous-QueenVictoria_StoryOfHerLifeAndReign1819-1901.epub" + ) + +testTitleRomanNum :: (Globals, [Formatter]) -> Test +testTitleRomanNum (gs, fs) = TestCase $ + assertNewName gs { gMetadata = meta } fs + "title with Roman numerals" expected + where + meta = emptyMetadata + { metaCreators = [Creator + Nothing + (Just "Hauptmann, Gerhart") + Nothing + "Gerhart Hauptmann"] + , metaTitles = [Title Nothing Nothing Nothing + "The Dramatic Works of Gerhart Hauptmann\nVolume III"] + } + expected = + ( "ordinary_book" + , "HauptmannGerhart-TheDramaticWorksOfGerhartHauptmann_VolumeIII.epub" + ) + + testAllPunctuation :: (Globals, [Formatter]) -> Test testAllPunctuation (gs, fs) = TestCase $ assertNewName gs { gMetadata = meta } fs "big test of all punctuation" expected @@ -373,7 +480,7 @@ } expected = ( "ordinary_book" - , "MorelliDino-TheCrazy_SandBoxOfSmedleysDiscontentLoathingFearAndMalnourishmentMaybeNot_Part2.epub" + , "MorelliDino-TheCrazy_Sand-BoxOfSmedleysDiscontentLoathingFearAndMalnourishmentMaybeNot_Part2.epub" ) @@ -806,23 +913,40 @@ ) -testInterzoneOldShort :: (Globals, [Formatter]) -> Test -testInterzoneOldShort (gs, fs) = TestCase $ +testInterzone :: (Globals, [Formatter]) -> Test +testInterzone (gs, fs) = TestCase $ assertNewName gs { gMetadata = meta } fs - "Interzone Magazine, old style, short" expected + "Interzone Magazine, Smashwords style" expected where meta = emptyMetadata { metaCreators = [Creator Nothing Nothing - Nothing "TTA Press Authors"] + Nothing "TTA Press"] , metaTitles = [Title Nothing Nothing Nothing - "Interzone SFF #212"] + "Interzone 233 Mar - Apr 2011"] } expected = - ( "magInterzone_old" - , "InterzoneSFF212.epub" + ( "magInterzone" + , "InterzoneSFF233.epub" ) +testInterzoneCaps :: (Globals, [Formatter]) -> Test +testInterzoneCaps (gs, fs) = TestCase $ + assertNewName gs { gMetadata = meta } fs + "Interzone Magazine, started in 2014" expected + where + meta = emptyMetadata + { metaCreators = [Creator Nothing (Just "Interzone 249") + Nothing "TTA Press"] + , metaTitles = [Title Nothing Nothing Nothing + "INTERZONE #249 SCIENCE FICTION & FANTASY MAGAZINE (NOVâ\x80\x93DEC 2013)"] + } + expected = + ( "magInterzone" + , "InterzoneSFF249.epub" + ) + + testInterzoneOldLong :: (Globals, [Formatter]) -> Test testInterzoneOldLong (gs, fs) = TestCase $ assertNewName gs { gMetadata = meta } fs @@ -835,25 +959,25 @@ "Interzone Science Fiction and Fantasy Magazine #216"] } expected = - ( "magInterzone_old" + ( "magInterzone" , "InterzoneSFF216.epub" ) -testInterzone :: (Globals, [Formatter]) -> Test -testInterzone (gs, fs) = TestCase $ +testInterzoneOldShort :: (Globals, [Formatter]) -> Test +testInterzoneOldShort (gs, fs) = TestCase $ assertNewName gs { gMetadata = meta } fs - "Interzone Magazine, Smashwords style" expected + "Interzone Magazine, old style, short" expected where meta = emptyMetadata { metaCreators = [Creator Nothing Nothing - Nothing "TTA Press"] + Nothing "TTA Press Authors"] , metaTitles = [Title Nothing Nothing Nothing - "Interzone 233 Mar - Apr 2011"] + "Interzone SFF #212"] } expected = ( "magInterzone" - , "InterzoneSFF233.epub" + , "InterzoneSFF212.epub" ) @@ -1151,8 +1275,8 @@ "Jim Baen's Universe-Vol 4 Num 6"] } expected = - ( "magUniverse" - , "JimBaensUniverseVol04Num06.epub" + ( "magGenericVolNo2" + , "JimBaensUniverse-Vol04No06.epub" ) @@ -1285,6 +1409,81 @@ expected = ( "magGenericWithIssue" , "PlasmaFrequencyMagazine08.epub" + ) + + +testMagPunchinello1800s :: (Globals, [Formatter]) -> Test +testMagPunchinello1800s (gs, fs) = TestCase $ + assertNewName gs { gMetadata = meta } fs + "Punchinello magazine from the 1800s" expected + where + meta = emptyMetadata + { metaTitles = [Title Nothing Nothing Nothing + "Punchinello, Volume 1, No. 06, May 7, 1870"] + } + expected = + ( "magGenericVolNo2" + , "PunchinelloVol01No06.epub" + ) + + +testMagGenericVolNo3 :: (Globals, [Formatter]) -> Test +testMagGenericVolNo3 (gs, fs) = TestCase $ + assertNewName gs { gMetadata = meta } fs + "generic magazine with volume and 3-digit number" expected + where + meta = emptyMetadata + { metaTitles = [Title Nothing Nothing Nothing + "Blackwood's Edinburgh Magazine Volume 53, No. 327, January, 1843"] + } + expected = + ( "magGenericVolNo3" + , "BlackwoodsEdinburghMagazineVol53No327.epub" + ) + + +testMagSubjWithIssue :: (Globals, [Formatter]) -> Test +testMagSubjWithIssue (gs, fs) = TestCase $ + assertNewName gs { gMetadata = meta } fs + "with magazine subject and issue" expected + where + meta = emptyMetadata + { metaTitles = [Title Nothing Nothing Nothing "The Dark Issue 1"] + , metaCreators = + [ Creator (Just "aut") (Just "Okorafor, Nnedi & Swirsky, Rachel & Slatter, Angela & Hannett, Lisa L.") Nothing + "Nnedi Okorafor" + , Creator (Just "aut") Nothing Nothing + "Rachel Swirsky" + ] + , metaDates = [Date (Just "publication") "2013-10-01"] + , metaSubjects = + [ "magazine" + , "horror" + , "dark fantasy" + ] + } + expected = + ( "magGenericSubjWithIssue" + , "TheDarkMagazine01.epub" + ) + + +testWomensInstituteLib :: (Globals, [Formatter]) -> Test +testWomensInstituteLib (gs, fs) = TestCase $ + assertNewName gs { gOpts = testOpts, gMetadata = meta } fs + "Women's Institute Library" expected + where + testOpts = (gOpts gs) { optPublisher = True } + meta = emptyMetadata + { metaCreators = [Creator Nothing + (Just "Woman's Institute of Domestic Arts and Sciences") + Nothing "Woman's Institute of Domestic Arts and Sciences"] + , metaTitles = [Title Nothing Nothing Nothing + "Woman's Institute Library of Cookery\nVolume 1: Essentials of Cookery; Cereals; Bread; Hot Breads"] + } + expected = + ( "nonficWomensInstituteLibrary" + , "WomansInstituteLibraryOfCookery_Volume1_EssentialsOfCookeryCerealsBreadHotBreads.epub" )