e690fe4 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

Merge pull request #1516 from mpickering/epubmetadata

e690fe4John MacFarlane12 years ago

README+18

compare with master

unifiedsplit

@@ -2972,43 +2972,60 @@ The following fields are recognized:

2972 `GTIN-13`, `UPC`, `ISMN-10`, `DOI`, `LCCN`, `GTIN-14`,

2973 `ISBN-13`, `Legal deposit number`, `URN`, `OCLC`,

2974 `ISMN-13`, `ISBN-A`, `JP`, `OLCC`.

2975+

2975`title`

2976`title`

2976 ~ Either a string value, or an object with fields `file-as` and

2977 ~ Either a string value, or an object with fields `file-as` and

2977 `type`, or a list of such objects. Valid values for `type` are

2978 `type`, or a list of such objects. Valid values for `type` are

2978 `main`, `subtitle`, `short`, `collection`, `edition`, `extended`.

2979 `main`, `subtitle`, `short`, `collection`, `edition`, `extended`.

2980+

2979`creator`

2981`creator`

2980 ~ Either a string value, or an object with fields `role`, `file-as`,

2982 ~ Either a string value, or an object with fields `role`, `file-as`,

2981 and `text`, or a list of such objects. Valid values for `role` are

2983 and `text`, or a list of such objects. Valid values for `role` are

2982 [marc relators](http://www.loc.gov/marc/relators/relaterm.html), but

2984 [marc relators](http://www.loc.gov/marc/relators/relaterm.html), but

2983 pandoc will attempt to translate the human-readable versions

2985 pandoc will attempt to translate the human-readable versions

2984 (like "author" and "editor") to the appropriate marc relators.

2986 (like "author" and "editor") to the appropriate marc relators.

2987+

2985`contributor`

2988`contributor`

2986 ~ Same format as `creator`.

2989 ~ Same format as `creator`.

2990+

2987`date`

2991`date`

2988 ~ A string value in `YYYY-MM-DD` format. (Only the year is necessary.)

2992 ~ A string value in `YYYY-MM-DD` format. (Only the year is necessary.)

2989 Pandoc will attempt to convert other common date formats.

2993 Pandoc will attempt to convert other common date formats.

2994+

2990`language`

2995`language`

2991 ~ A string value in [RFC5646] format. Pandoc will default to the local

2996 ~ A string value in [RFC5646] format. Pandoc will default to the local

2992 language if nothing is specified.

2997 language if nothing is specified.

2998+

2993`subject`

2999`subject`

2994 ~ A string value or a list of such values.

3000 ~ A string value or a list of such values.

3001+

2995`description`

3002`description`

2996 ~ A string value.

3003 ~ A string value.

3004+

2997`type`

3005`type`

2998 ~ A string value.

3006 ~ A string value.

3007+

2999`format`

3008`format`

3000 ~ A string value.

3009 ~ A string value.

3010+

3001`relation`

3011`relation`

3002 ~ A string value.

3012 ~ A string value.

3013+

3003`coverage`

3014`coverage`

3004 ~ A string value.

3015 ~ A string value.

3016+

3005`rights`

3017`rights`

3006 ~ A string value.

3018 ~ A string value.

3019+

3007`cover-image`

3020`cover-image`

3008 ~ A string value (path to cover image).

3021 ~ A string value (path to cover image).

3022+

3009`stylesheet`

3023`stylesheet`

3010 ~ A string value (path to CSS stylesheet).

3024 ~ A string value (path to CSS stylesheet).

3011

3025

3026+`page-progression-direction`

3027+ ~ Either `ltr` or `rtl`. Specifies the synonymous spine [attribute][EPUBspine].

3028+

3012Literate Haskell support

3029Literate Haskell support

3013========================

3030========================

3014

3031

@@ -3126,3 +3143,4 @@ Rosenthal.

3126[InDesign ICML]: https://www.adobe.com/content/dam/Adobe/en/devnet/indesign/cs55-docs/IDML/idml-specification.pdf

3143[InDesign ICML]: https://www.adobe.com/content/dam/Adobe/en/devnet/indesign/cs55-docs/IDML/idml-specification.pdf

3127[txt2tags]: http://txt2tags.org/

3144[txt2tags]: http://txt2tags.org/

3128[EPUB]: http://idpf.org/epub

3145[EPUB]: http://idpf.org/epub

3146+[EPUBspine]: http://www.idpf.org/epub/301/spec/epub-publications.html#sec-spine-elem

src/Text/Pandoc/Readers/EPUB.hs+7−6

@@ -14,7 +14,7 @@ import Text.Pandoc.Walk (walk, query)

14import Text.Pandoc.Generic(bottomUp)

15import Text.Pandoc.Readers.HTML (readHtml)

16import Text.Pandoc.Options ( ReaderOptions(..), readerTrace)

17−import Text.Pandoc.Shared (escapeURI, collapseFilePath)

17+import Text.Pandoc.Shared (escapeURI, collapseFilePath, addMetaField)

18import Text.Pandoc.MediaBag (MediaBag, insertMedia)

19import Text.Pandoc.Compat.Except (MonadError, throwError, runExcept, Except)

20import qualified Text.Pandoc.Builder as B

@@ -155,7 +155,7 @@ parseMeta content = do

155-- http://www.idpf.org/epub/30/spec/epub30-publications.html#sec-metadata-elem

156parseMetaItem :: Element -> Meta -> Meta

157parseMetaItem e@(stripNamespace . elName -> field) meta =

158− B.setMeta (renameMeta field) (B.str $ strContent e) meta

158+ addMetaField (renameMeta field) (B.str $ strContent e) meta

159

160renameMeta :: String -> String

161renameMeta "creator" = "author"

@@ -169,10 +169,10 @@ getManifest archive = do

169 ns <- mkE "xmlns not in namespaces" (lookup "xmlns" namespaces)

170 as <- liftM ((map attrToPair) . elAttribs)

171 (findElementE (QName "rootfile" (Just ns) Nothing) docElem)

172− root <- mkE "Root not found" (lookup "full-path" as)

172+ manifestFile <- mkE "Root not found" (lookup "full-path" as)

173− let rootdir = dropFileName root

173+ let rootdir = dropFileName manifestFile

174 --mime <- lookup "media-type" as

175− manifest <- findEntryByPathE root archive

175+ manifest <- findEntryByPathE manifestFile archive

176 liftM ((,) rootdir) (parseXMLDocE . UTF8.toStringLazy . fromEntry $ manifest)

177

178-- Fixup

@@ -272,7 +272,8 @@ findAttrE :: MonadError String m => QName -> Element -> m String

272findAttrE q e = mkE "findAttr" $ findAttr q e

273

274findEntryByPathE :: MonadError String m => FilePath -> Archive -> m Entry

275+findEntryByPathE (normalise -> path) a =

275−findEntryByPathE path a = mkE ("No entry on path: " ++ path) $ findEntryByPath path a

276+ mkE ("No entry on path: " ++ path) $ findEntryByPath path a

276

277

277parseXMLDocE :: MonadError String m => String -> m Element

278parseXMLDocE :: MonadError String m => String -> m Element

278parseXMLDocE doc = mkE "Unable to parse XML doc" $ parseXMLDoc doc

279parseXMLDocE doc = mkE "Unable to parse XML doc" $ parseXMLDoc doc

src/Text/Pandoc/Writers/EPUB.hs+47−20

@@ -1,4 +1,4 @@

1−{-# LANGUAGE PatternGuards, CPP, ScopedTypeVariables #-}

1+{-# LANGUAGE PatternGuards, CPP, ScopedTypeVariables, ViewPatterns #-}

2{-

@@ -29,7 +29,7 @@ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA

29Conversion of 'Pandoc' documents to EPUB.

30-}

31module Text.Pandoc.Writers.EPUB ( writeEPUB ) where

32−import Data.IORef

32+import Data.IORef ( IORef, newIORef, readIORef, modifyIORef )

33import qualified Data.Map as M

34import Data.Maybe ( fromMaybe )

35import Data.List ( isPrefixOf, isInfixOf, intercalate )

@@ -40,27 +40,35 @@ import qualified Data.ByteString.Lazy as B

40import qualified Data.ByteString.Lazy.Char8 as B8

41import qualified Text.Pandoc.UTF8 as UTF8

42import Text.Pandoc.SelfContained ( makeSelfContained )

43−import Codec.Archive.Zip

43+import Codec.Archive.Zip ( emptyArchive, addEntryToArchive, eRelativePath, fromEntry , Entry, toEntry, fromArchive)

44import Control.Applicative ((<$>))

45−import Data.Time.Clock.POSIX

45+import Data.Time.Clock.POSIX ( getPOSIXTime )

46−import Data.Time

46+import Data.Time (getCurrentTime,UTCTime, formatTime)

47−import System.Locale

47+import System.Locale ( defaultTimeLocale )

48−import Text.Pandoc.Shared hiding ( Element )

48+import Text.Pandoc.Shared ( trimr, renderTags', safeRead, uniqueIdent, trim

49+ , normalizeDate, readDataFile, stringify, warn

50+ , hierarchicalize, fetchItem' )

49−import qualified Text.Pandoc.Shared as Shared

51+import qualified Text.Pandoc.Shared as S (Element(..))

50import Text.Pandoc.Builder (fromList, setMeta)

52import Text.Pandoc.Builder (fromList, setMeta)

51−import Text.Pandoc.Options

53+import Text.Pandoc.Options ( WriterOptions(..)

54+ , HTMLMathMethod(..)

55+ , EPUBVersion(..)

56+ , ObfuscationMethod(NoObfuscation) )

52import Text.Pandoc.Definition

57import Text.Pandoc.Definition

53−import Text.Pandoc.Walk

58+import Text.Pandoc.Walk (walk, walkM)

54−import Control.Monad.State

59+import Control.Monad.State (modify, get, execState, State, put, evalState)

60+import Control.Monad (foldM, when, mplus, liftM)

55−import Text.XML.Light hiding (ppTopElement)

61+import Text.XML.Light ( unode, Element(..), unqual, Attr(..), add_attrs

62+ , strContent, lookupAttr, Node(..), QName(..), parseXML

63+ , onlyElems, node, ppElement)

56−import Text.Pandoc.UUID

64+import Text.Pandoc.UUID (getRandomUUID)

57−import Text.Pandoc.Writers.HTML

65+import Text.Pandoc.Writers.HTML (writeHtmlString, writeHtml)

58import Data.Char ( toLower, isDigit, isAlphaNum )

66import Data.Char ( toLower, isDigit, isAlphaNum )

59import Network.URI ( unEscapeString )

67import Network.URI ( unEscapeString )

60import Text.Pandoc.MIME (getMimeType)

68import Text.Pandoc.MIME (getMimeType)

61import qualified Control.Exception as E

69import qualified Control.Exception as E

62import Text.Blaze.Html.Renderer.Utf8 (renderHtml)

70import Text.Blaze.Html.Renderer.Utf8 (renderHtml)

63−import Text.HTML.TagSoup

71+import Text.HTML.TagSoup (Tag(TagOpen), fromAttrib, parseTags)

65-- A Chapter includes a list of blocks and maybe a section

73-- A Chapter includes a list of blocks and maybe a section

66-- number offset. Note, some chapters are unnumbered. The section

74-- number offset. Note, some chapters are unnumbered. The section

@@ -86,6 +94,7 @@ data EPUBMetadata = EPUBMetadata{

86 , epubRights :: Maybe String

94 , epubRights :: Maybe String

87 , epubCoverImage :: Maybe String

95 , epubCoverImage :: Maybe String

88 , epubStylesheet :: Maybe Stylesheet

96 , epubStylesheet :: Maybe Stylesheet

97+ , epubPageDirection :: ProgressionDirection

89 } deriving Show

98 } deriving Show

91data Stylesheet = StylesheetPath FilePath

100data Stylesheet = StylesheetPath FilePath

@@ -114,6 +123,8 @@ data Title = Title{

114 , titleType :: Maybe String

123 , titleType :: Maybe String

115 } deriving Show

124 } deriving Show

116

125

126+data ProgressionDirection = LTR | RTL | Default deriving Show

127+

117dcName :: String -> QName

128dcName :: String -> QName

118dcName n = QName n Nothing (Just "dc")

129dcName n = QName n Nothing (Just "dc")

119

130

@@ -288,6 +299,7 @@ metadataFromMeta opts meta = EPUBMetadata{

288 , epubRights = rights

299 , epubRights = rights

289 , epubCoverImage = coverImage

300 , epubCoverImage = coverImage

290 , epubStylesheet = stylesheet

301 , epubStylesheet = stylesheet

302+ , epubPageDirection = pageDirection

291 }

303 }

292 where identifiers = getIdentifier meta

304 where identifiers = getIdentifier meta

293 titles = getTitle meta

305 titles = getTitle meta

@@ -310,6 +322,14 @@ metadataFromMeta opts meta = EPUBMetadata{

310 stylesheet = (StylesheetContents <$> writerEpubStylesheet opts) `mplus`

322 stylesheet = (StylesheetContents <$> writerEpubStylesheet opts) `mplus`

311 ((StylesheetPath . metaValueToString) <$>

323 ((StylesheetPath . metaValueToString) <$>

312 lookupMeta "stylesheet" meta)

324 lookupMeta "stylesheet" meta)

325+ pageDirection = maybe Default stringToPageDirection

326+ (lookupMeta "page-progression-direction" meta)

327+ stringToPageDirection (metaValueToString -> s) =

328+ case s of

329+ "ltr" -> LTR

330+ "rtl" -> RTL

331+ _ -> Default

332+

313

333

314-- | Produce an EPUB file from a Pandoc document.

334-- | Produce an EPUB file from a Pandoc document.

315writeEPUB :: WriterOptions -- ^ Writer options

335writeEPUB :: WriterOptions -- ^ Writer options

@@ -374,6 +394,12 @@ writeEPUB opts doc@(Pandoc meta _) = do

374 let mkFontEntry f = mkEntry (takeFileName f) `fmap` B.readFile f

394 let mkFontEntry f = mkEntry (takeFileName f) `fmap` B.readFile f

375 fontEntries <- mapM mkFontEntry $ writerEpubFonts opts'

395 fontEntries <- mapM mkFontEntry $ writerEpubFonts opts'

376

396

397+ -- set page progression direction

398+ let progressionDirection = case epubPageDirection metadata of

399+ LTR -> "ltr"

400+ RTL -> "rtl"

401+ Default -> "default"

402+

377 -- body pages

403 -- body pages

378

404

379 -- add level 1 header to beginning if none there

405 -- add level 1 header to beginning if none there

@@ -493,7 +519,8 @@ writeEPUB opts doc@(Pandoc meta _) = do

493 (pictureNode x)]) ++

519 (pictureNode x)]) ++

494 map pictureNode picEntries ++

520 map pictureNode picEntries ++

495 map fontNode fontEntries

521 map fontNode fontEntries

496− , unode "spine" ! [("toc","ncx")] $

522+ , unode "spine" ! [("toc","ncx")

523+ ,("page-progression-direction", progressionDirection)] $

497 case epubCoverImage metadata of

524 case epubCoverImage metadata of

498 Nothing -> []

525 Nothing -> []

499 Just _ -> [ unode "itemref" !

526 Just _ -> [ unode "itemref" !

@@ -524,8 +551,8 @@ writeEPUB opts doc@(Pandoc meta _) = do

524 let tocLevel = writerTOCDepth opts

551 let tocLevel = writerTOCDepth opts

525

552

526 let navPointNode :: (Int -> String -> String -> [Element] -> Element)

553 let navPointNode :: (Int -> String -> String -> [Element] -> Element)

527− -> Shared.Element -> State Int Element

554+ -> S.Element -> State Int Element

528− navPointNode formatter (Sec _ nums (ident,_,_) ils children) = do

555+ navPointNode formatter (S.Sec _ nums (ident,_,_) ils children) = do

529 n <- get

556 n <- get

530 modify (+1)

557 modify (+1)

531 let showNums :: [Int] -> String

558 let showNums :: [Int] -> String

@@ -537,12 +564,12 @@ writeEPUB opts doc@(Pandoc meta _) = do

537 let src = case lookup ident reftable of

564 let src = case lookup ident reftable of

538 Just x -> x

565 Just x -> x

539 Nothing -> error (ident ++ " not found in reftable")

566 Nothing -> error (ident ++ " not found in reftable")

540− let isSec (Sec lev _ _ _ _) = lev <= tocLevel

567+ let isSec (S.Sec lev _ _ _ _) = lev <= tocLevel

541 isSec _ = False

568 isSec _ = False

542 let subsecs = filter isSec children

569 let subsecs = filter isSec children

543 subs <- mapM (navPointNode formatter) subsecs

570 subs <- mapM (navPointNode formatter) subsecs

544 return $ formatter n tit src subs

571 return $ formatter n tit src subs

545− navPointNode _ (Blk _) = error "navPointNode encountered Blk"

572+ navPointNode _ (S.Blk _) = error "navPointNode encountered Blk"

546

573

547 let navMapFormatter :: Int -> String -> String -> [Element] -> Element

574 let navMapFormatter :: Int -> String -> String -> [Element] -> Element

548 navMapFormatter n tit src subs = unode "navPoint" !

575 navMapFormatter n tit src subs = unode "navPoint" !