74212eb · typeable/pandoc

git.iko.soy/typeable/pandoc

Added support for translations (localization) (see #3559).

* readDataFile, readDefaultDataFile, getReferenceDocx, getReferenceODT have been removed from Shared and moved into Class. They are now defined in terms of PandocMonad primitives, rather than being primitve methods of the class. * toLang has been moved from BCP47 to Class. * NoTranslation and CouldNotLoudTranslations have been added to LogMessage. * New module, Text.Pandoc.Translations, exporting Term, Translations, readTranslations. * New functions in Class: translateTerm, setTranslations. Note that nothing is loaded from data files until translateTerm is used; setTranslation just sets the language to be used. * Added two translation data files in data/translations. * LaTeX reader: Support `\setmainlanguage` or `\setdefaultlanguage` (polyglossia) and `\figurename`.

commit74212eb1b0e1757fc0ac3e5d45b0ee18bac491e5 authorJohn MacFarlane <jgm@berkeley.edu> dateFri 11 Aug 2017, 11:56 (9 years ago) parent7892dcd

16 files changed+682−358

compare with master

unifiedsplit

data/translations/en.trans+19 data/translations/fr.trans+19 pandoc.cabal+4−1 src/Text/Pandoc/App.hs+175−166 src/Text/Pandoc/BCP47.hs−14 src/Text/Pandoc/Class.hs+212−46 src/Text/Pandoc/Logging.hs+14 src/Text/Pandoc/Lua/PandocModule.hs+3−3 src/Text/Pandoc/Readers/LaTeX.hs+133−2 src/Text/Pandoc/Shared.hs+1−118 src/Text/Pandoc/Templates.hs+1−1 src/Text/Pandoc/Translations.hs+94 src/Text/Pandoc/Writers/ConTeXt.hs+1−1 src/Text/Pandoc/Writers/Docx.hs+2−2 src/Text/Pandoc/Writers/LaTeX.hs+2−2 src/Text/Pandoc/Writers/ODT.hs+2−2

data/translations/en.transnew file+19

@@ -0,0 +1,19 @@

1+Preface: Preface

2+References: References

3+Abstract: Abstract

4+Bibliography: Bibliography

5+Chapter: Chapter

6+Appendix: Appendix

7+Contents: Contents

8+ListOfFigures: ListOfFigures

9+ListOfTables: ListOfTables

10+Index: Index

11+Figure: Figure

12+Table: Table

13+Part: Part

14+Page: page

15+Proof: Proof

16+See: see

17+SeeAlso: see also

18+Cc: cc

19+To: To

data/translations/fr.transnew file+19

@@ -0,0 +1,19 @@

1+Preface: Préface

2+References: Références

3+Abstract: Résumé

4+Bibliography: Bibliographie

5+Chapter: Chaptire

6+Appendix: Annexe

7+Contents: Table des matières

8+ListOfFigures: Table des figures

9+ListOfTables: Liste des tableaux

10+Index: Index

11+Figure: Fig.

12+Table: Tab.

13+Part: partie

14+Page: page

15+Proof: Démonstration

16+See: voir

17+SeeAlso: voir aussi

18+Cc: Copie à

19+To:

pandoc.cabal+4−1

@@ -72,6 +72,8 @@ Data-Files:

7272 data/templates/default.org

7373 data/templates/default.epub2

7474 data/templates/default.epub3

75+ -- translations

76+ data/translations/*.trans

7577 -- source files for reference.docx

7678 data/docx/[Content_Types].xml

7779 data/docx/_rels/.rels

@@ -469,7 +471,8 @@ Library

469471 Text.Pandoc.CSS,

470472 Text.Pandoc.CSV,

471473 Text.Pandoc.UUID,

472− Text.Pandoc.BCP47

474+ Text.Pandoc.BCP47,

475+ Text.Pandoc.Translations,

473476 Text.Pandoc.Slides,

474477 Text.Pandoc.Compat.Time,

475478 Paths_pandoc

src/Text/Pandoc/App.hs+175−166

@@ -41,7 +41,7 @@ module Text.Pandoc.App (

4141import Control.Applicative ((<|>))

4242import qualified Control.Exception as E

4343import Control.Monad

44−import Control.Monad.Except (throwError)

44+import Control.Monad.Except (throwError, catchError)

4545import Control.Monad.Trans

4646import Data.Monoid

4747import Data.Aeson (FromJSON (..), ToJSON (..), defaultOptions, eitherDecode',

@@ -71,21 +71,22 @@ import System.Environment (getArgs, getEnvironment, getProgName)

7171import System.Exit (ExitCode (..), exitSuccess)

7272import System.FilePath

7373import System.IO (nativeNewline, stdout)

74−import qualified System.IO as IO (Newline (..))

7574import System.IO.Error (isDoesNotExistError)

75+import qualified System.IO as IO (Newline (..))

7676import Text.Pandoc

7777import Text.Pandoc.Builder (setMeta)

7878import Text.Pandoc.Class (PandocIO, extractMedia, fillMediaBag, getLog,

7979 setResourcePath, getMediaBag, setTrace, report,

80− setUserDataDir)

80+ setUserDataDir, readFileStrict, readDataFile,

81+ readDefaultDataFile, setTranslations)

8182import Text.Pandoc.Highlighting (highlightingStyles)

83+import Text.Pandoc.BCP47 (parseBCP47, Lang(..))

8284import Text.Pandoc.Lua (runLuaFilter, LuaException(..))

8385import Text.Pandoc.Writers.Math (defaultMathJaxURL, defaultKaTeXURL)

8486import Text.Pandoc.PDF (makePDF)

8587import Text.Pandoc.Process (pipeProcess)

8688import Text.Pandoc.SelfContained (makeDataURI, makeSelfContained)

87−import Text.Pandoc.Shared (headerShift, isURI, openURL, readDataFile,

88− readDataFileUTF8, readDefaultDataFile,

89+import Text.Pandoc.Shared (headerShift, isURI, openURL,

8990 safeRead, tabFilter, eastAsianLineBreakFilter)

9091import qualified Text.Pandoc.UTF8 as UTF8

9192import Text.Pandoc.XML (toEntities)

@@ -215,84 +216,6 @@ convertWithOpts opts = do

215216 _ -> e

216217

217218 let standalone = optStandalone opts || not (isTextFormat format) || pdfOutput

218−

219− templ <- case optTemplate opts of

220− _ | not standalone -> return Nothing

221− Nothing -> do

222− deftemp <- runIO $ do

223− setUserDataDir datadir

224− getDefaultTemplate format

225− case deftemp of

226− Left e -> E.throwIO e

227− Right t -> return (Just t)

228− Just tp -> do

229− -- strip off extensions

230− let tp' = case takeExtension tp of

231− "" -> tp <.> format

232− _ -> tp

233− Just <$> E.catch (UTF8.readFile tp')

234− (\e -> if isDoesNotExistError e

235− then E.catch

236− (readDataFileUTF8 datadir

237− ("templates" </> tp'))

238− (\e' -> let _ = (e' :: E.SomeException)

239− in E.throwIO e')

240− else E.throwIO e)

241−

242− let addStringAsVariable varname s vars = return $ (varname, s) : vars

243−

244− let addContentsAsVariable varname fp vars = do

245− s <- UTF8.readFile fp

246− return $ (varname, s) : vars

247−

248− -- note: this reverses the list constructed in option parsing,

249− -- which in turn was reversed from the command-line order,

250− -- so we end up with the correct order in the variable list:

251− let withList _ [] vars = return vars

252− withList f (x:xs) vars = f x vars >>= withList f xs

253−

254− variables <-

255−

256− withList (addStringAsVariable "sourcefile")

257− (reverse $ optInputFiles opts) (("outputfile", optOutputFile opts) : optVariables opts)

258− -- we reverse this list because, unlike

259− -- the other option lists here, it is

260− -- not reversed when parsed from CLI arguments.

261− -- See withList, above.

262− >>=

263− withList (addContentsAsVariable "include-before")

264− (optIncludeBeforeBody opts)

265− >>=

266− withList (addContentsAsVariable "include-after")

267− (optIncludeAfterBody opts)

268− >>=

269− withList (addContentsAsVariable "header-includes")

270− (optIncludeInHeader opts)

271− >>=

272− withList (addStringAsVariable "css") (optCss opts)

273− >>=

274− maybe return (addStringAsVariable "title-prefix") (optTitlePrefix opts)

275− >>=

276− maybe return (addStringAsVariable "epub-cover-image")

277− (optEpubCoverImage opts)

278− >>=

279− (\vars -> case mathMethod of

280− LaTeXMathML Nothing -> do

281− s <- readDataFileUTF8 datadir "LaTeXMathML.js"

282− return $ ("mathml-script", s) : vars

283− _ -> return vars)

284− >>=

285− (\vars -> if format == "dzslides"

286− then do

287− dztempl <- readDataFileUTF8 datadir

288− ("dzslides" </> "template.html")

289− let dzline = "<!-- {{{{ dzslides core"

290− let dzcore = unlines

291− $ dropWhile (not . (dzline `isPrefixOf`))

292− $ lines dztempl

293− return $ ("dzslides-core", dzcore) : vars

294− else return vars)

295−

296219 let sourceURL = case sources of

297220 [] -> Nothing

298221 (x:_) -> case parseURI x of

@@ -302,21 +225,7 @@ convertWithOpts opts = do

302225 uriFragment = "" }

303226 _ -> Nothing

304227

305− abbrevs <- (Set.fromList . filter (not . null) . lines) <$>

306− case optAbbreviations opts of

307− Nothing -> readDataFileUTF8 datadir "abbreviations"

308− Just f -> UTF8.readFile f

309−

310− let readerOpts = def{ readerStandalone = standalone

311− , readerColumns = optColumns opts

312− , readerTabStop = optTabStop opts

313− , readerIndentedCodeClasses = optIndentedCodeClasses opts

314− , readerDefaultImageExtension =

315− optDefaultImageExtension opts

316− , readerTrackChanges = optTrackChanges opts

317− , readerAbbreviations = abbrevs

318− , readerExtensions = readerExts

319− }

228+ let addStringAsVariable varname s vars = return $ (varname, s) : vars

320229

321230 highlightStyle <- lookupHighlightStyle $ optHighlightStyle opts

322231 let addSyntaxMap existingmap f = do

@@ -336,40 +245,6 @@ convertWithOpts opts = do

336245 (\(syn,dep) -> (T.unpack syn ++ " requires " ++

337246 T.unpack dep ++ " through IncludeRules.")) xs)

338247

339− let writerOptions = def { writerTemplate = templ,

340− writerVariables = variables,

341− writerTabStop = optTabStop opts,

342− writerTableOfContents = optTableOfContents opts,

343− writerHTMLMathMethod = mathMethod,

344− writerIncremental = optIncremental opts,

345− writerCiteMethod = optCiteMethod opts,

346− writerNumberSections = optNumberSections opts,

347− writerNumberOffset = optNumberOffset opts,

348− writerSectionDivs = optSectionDivs opts,

349− writerExtensions = writerExts,

350− writerReferenceLinks = optReferenceLinks opts,

351− writerReferenceLocation = optReferenceLocation opts,

352− writerDpi = optDpi opts,

353− writerWrapText = optWrapText opts,

354− writerColumns = optColumns opts,

355− writerEmailObfuscation = optEmailObfuscation opts,

356− writerIdentifierPrefix = optIdentifierPrefix opts,

357− writerSourceURL = sourceURL,

358− writerHtmlQTags = optHtmlQTags opts,

359− writerTopLevelDivision = optTopLevelDivision opts,

360− writerListings = optListings opts,

361− writerSlideLevel = optSlideLevel opts,

362− writerHighlightStyle = highlightStyle,

363− writerSetextHeaders = optSetextHeaders opts,

364− writerEpubSubdirectory = optEpubSubdirectory opts,

365− writerEpubMetadata = epubMetadata,

366− writerEpubFonts = optEpubFonts opts,

367− writerEpubChapterLevel = optEpubChapterLevel opts,

368− writerTOCDepth = optTOCDepth opts,

369− writerReferenceDoc = optReferenceDoc opts,

370− writerLaTeXArgs = optLaTeXEngineArgs opts,

371− writerSyntaxMap = syntaxMap

372− }

373248

374249

375250#ifdef _WINDOWS

@@ -383,18 +258,6 @@ convertWithOpts opts = do

383258 "Specify an output file using the -o option."

384259

385260

386− let transforms = (case optBaseHeaderLevel opts of

387− x | x > 1 -> (headerShift (x - 1) :)

388− | otherwise -> id) $

389− (if extensionEnabled Ext_east_asian_line_breaks

390− readerExts &&

391− not (extensionEnabled Ext_east_asian_line_breaks

392− writerExts &&

393− writerWrapText writerOptions == WrapPreserve)

394− then (eastAsianLineBreakFilter :)

395− else id)

396− []

397−

398261 let convertTabs = tabFilter (if optPreserveTabs opts || readerName == "t2t"

399262 then 0

400263 else optTabStop opts)

@@ -419,32 +282,175 @@ convertWithOpts opts = do

419282 E.throwIO PandocFailOnWarningError

420283 return res

421284

422− let sourceToDoc :: [FilePath] -> PandocIO Pandoc

423− sourceToDoc sources' =

424− case reader of

425− TextReader r

426− | optFileScope opts || readerName == "json" ->

427− mconcat <$> mapM (readSource >=> r readerOpts) sources

428− | otherwise ->

429− readSources sources' >>= r readerOpts

430− ByteStringReader r ->

431− mconcat <$> mapM (readFile' >=> r readerOpts) sources

432−

433− metadata <- if format == "jats" &&

434− isNothing (lookup "csl" (optMetadata opts)) &&

435− isNothing (lookup "citation-style" (optMetadata opts))

436− then do

437− jatsCSL <- readDataFile datadir "jats.csl"

438− let jatsEncoded = makeDataURI ("application/xml", jatsCSL)

439− return $ ("csl", jatsEncoded) : optMetadata opts

440− else return $ optMetadata opts

441−

442285 let eol = case optEol opts of

443286 CRLF -> IO.CRLF

444287 LF -> IO.LF

445288 Native -> nativeNewline

446289

290+ -- note: this reverses the list constructed in option parsing,

291+ -- which in turn was reversed from the command-line order,

292+ -- so we end up with the correct order in the variable list:

293+ let withList _ [] vars = return vars

294+ withList f (x:xs) vars = f x vars >>= withList f xs

295+

296+ let addContentsAsVariable varname fp vars = do

297+ s <- UTF8.toString <$> readFileStrict fp

298+ return $ (varname, s) : vars

299+

447300 runIO' $ do

301+ variables <-

302+ withList (addStringAsVariable "sourcefile")

303+ (reverse $ optInputFiles opts)

304+ (("outputfile", optOutputFile opts) : optVariables opts)

305+ -- we reverse this list because, unlike

306+ -- the other option lists here, it is

307+ -- not reversed when parsed from CLI arguments.

308+ -- See withList, above.

309+ >>=

310+ withList (addContentsAsVariable "include-before")

311+ (optIncludeBeforeBody opts)

312+ >>=

313+ withList (addContentsAsVariable "include-after")

314+ (optIncludeAfterBody opts)

315+ >>=

316+ withList (addContentsAsVariable "header-includes")

317+ (optIncludeInHeader opts)

318+ >>=

319+ withList (addStringAsVariable "css") (optCss opts)

320+ >>=

321+ maybe return (addStringAsVariable "title-prefix")

322+ (optTitlePrefix opts)

323+ >>=

324+ maybe return (addStringAsVariable "epub-cover-image")

325+ (optEpubCoverImage opts)

326+ >>=

327+ (\vars -> case mathMethod of

328+ LaTeXMathML Nothing -> do

329+ s <- UTF8.toString <$> readDataFile "LaTeXMathML.js"

330+ return $ ("mathml-script", s) : vars

331+ _ -> return vars)

332+ >>=

333+ (\vars -> if format == "dzslides"

334+ then do

335+ dztempl <- UTF8.toString <$> readDataFile

336+ ("dzslides" </> "template.html")

337+ let dzline = "<!-- {{{{ dzslides core"

338+ let dzcore = unlines

339+ $ dropWhile (not . (dzline `isPrefixOf`))

340+ $ lines dztempl

341+ return $ ("dzslides-core", dzcore) : vars

342+ else return vars)

343+

344+ abbrevs <- (Set.fromList . filter (not . null) . lines) <$>

345+ case optAbbreviations opts of

346+ Nothing -> UTF8.toString <$> readDataFile "abbreviations"

347+ Just f -> UTF8.toString <$> readFileStrict f

348+

349+ templ <- case optTemplate opts of

350+ _ | not standalone -> return Nothing

351+ Nothing -> Just <$> getDefaultTemplate format

352+ Just tp -> do

353+ -- strip off extensions

354+ let tp' = case takeExtension tp of

355+ "" -> tp <.> format

356+ _ -> tp

357+ Just . UTF8.toString <$>

358+ (readFileStrict tp' `catchError`

359+ (\e ->

360+ case e of

361+ PandocIOError _ e' |

362+ isDoesNotExistError e' ->

363+ readDataFile ("templates" </> tp')

364+ _ -> throwError e))

365+

366+ metadata <- if format == "jats" &&

367+ isNothing (lookup "csl" (optMetadata opts)) &&

368+ isNothing (lookup "citation-style" (optMetadata opts))

369+ then do

370+ jatsCSL <- readDataFile "jats.csl"

371+ let jatsEncoded = makeDataURI

372+ ("application/xml", jatsCSL)

373+ return $ ("csl", jatsEncoded) : optMetadata opts

374+ else return $ optMetadata opts

375+

376+ case lookup "lang" (optMetadata opts) of

377+ Just l -> case parseBCP47 l of

378+ Left _ -> return ()

379+ Right l' -> setTranslations l'

380+ Nothing -> setTranslations $ Lang "en" "" "US" []

381+

382+ let writerOptions = def {

383+ writerTemplate = templ

384+ , writerVariables = variables

385+ , writerTabStop = optTabStop opts

386+ , writerTableOfContents = optTableOfContents opts

387+ , writerHTMLMathMethod = mathMethod

388+ , writerIncremental = optIncremental opts

389+ , writerCiteMethod = optCiteMethod opts

390+ , writerNumberSections = optNumberSections opts

391+ , writerNumberOffset = optNumberOffset opts

392+ , writerSectionDivs = optSectionDivs opts

393+ , writerExtensions = writerExts

394+ , writerReferenceLinks = optReferenceLinks opts

395+ , writerReferenceLocation = optReferenceLocation opts

396+ , writerDpi = optDpi opts

397+ , writerWrapText = optWrapText opts

398+ , writerColumns = optColumns opts

399+ , writerEmailObfuscation = optEmailObfuscation opts

400+ , writerIdentifierPrefix = optIdentifierPrefix opts

401+ , writerSourceURL = sourceURL

402+ , writerHtmlQTags = optHtmlQTags opts

403+ , writerTopLevelDivision = optTopLevelDivision opts

404+ , writerListings = optListings opts

405+ , writerSlideLevel = optSlideLevel opts

406+ , writerHighlightStyle = highlightStyle

407+ , writerSetextHeaders = optSetextHeaders opts

408+ , writerEpubSubdirectory = optEpubSubdirectory opts

409+ , writerEpubMetadata = epubMetadata

410+ , writerEpubFonts = optEpubFonts opts

411+ , writerEpubChapterLevel = optEpubChapterLevel opts

412+ , writerTOCDepth = optTOCDepth opts

413+ , writerReferenceDoc = optReferenceDoc opts

414+ , writerLaTeXArgs = optLaTeXEngineArgs opts

415+ , writerSyntaxMap = syntaxMap

416+ }

417+

418+ let readerOpts = def{

419+ readerStandalone = standalone

420+ , readerColumns = optColumns opts

421+ , readerTabStop = optTabStop opts

422+ , readerIndentedCodeClasses = optIndentedCodeClasses opts

423+ , readerDefaultImageExtension =

424+ optDefaultImageExtension opts

425+ , readerTrackChanges = optTrackChanges opts

426+ , readerAbbreviations = abbrevs

427+ , readerExtensions = readerExts

428+ }

429+

430+ let transforms = (case optBaseHeaderLevel opts of

431+ x | x > 1 -> (headerShift (x - 1) :)

432+ | otherwise -> id) $

433+ (if extensionEnabled Ext_east_asian_line_breaks

434+ readerExts &&

435+ not (extensionEnabled Ext_east_asian_line_breaks

436+ writerExts &&

437+ writerWrapText writerOptions == WrapPreserve)

438+ then (eastAsianLineBreakFilter :)

439+ else id)

440+ []

441+

442+ let sourceToDoc :: [FilePath] -> PandocIO Pandoc

443+ sourceToDoc sources' =

444+ case reader of

445+ TextReader r

446+ | optFileScope opts || readerName == "json" ->

447+ mconcat <$> mapM (readSource >=> r readerOpts) sources

448+ | otherwise ->

449+ readSources sources' >>= r readerOpts

450+ ByteStringReader r ->

451+ mconcat <$> mapM (readFile' >=> r readerOpts) sources

452+

453+

448454 setUserDataDir datadir

449455 when (readerName == "markdown_github" ||

450456 writerName == "markdown_github") $

@@ -1011,7 +1017,8 @@ options =

10111017 , Option "" ["print-default-data-file"]

10121018 (ReqArg

10131019 (\arg _ -> do

1014− readDefaultDataFile arg >>= BS.hPutStr stdout

1020+ runIOorExplode $

1021+ readDefaultDataFile arg >>= liftIO . BS.hPutStr stdout

10151022 exitSuccess)

10161023 "FILE")

10171024 "" -- "Print default data file"

@@ -1469,7 +1476,9 @@ options =

14691476 (NoArg

14701477 (\_ -> do

14711478 ddir <- getDataDir

1472− tpl <- readDataFileUTF8 Nothing "bash_completion.tpl"

1479+ tpl <- runIOorExplode $

1480+ UTF8.toString <$>

1481+ readDefaultDataFile "bash_completion.tpl"

14731482 let optnames (Option shorts longs _ _) =

14741483 map (\c -> ['-',c]) shorts ++

14751484 map ("--" ++) longs

src/Text/Pandoc/BCP47.hs−14

@@ -29,7 +29,6 @@ Functions for parsing and rendering BCP47 language identifiers.

2929-}

3030module Text.Pandoc.BCP47 (

3131 getLang

32− , toLang

3332 , parseBCP47

3433 , Lang(..)

3534 , renderLang

@@ -40,8 +39,6 @@ import Data.Char (isAscii, isLetter, isUpper, isLower, toUpper, toLower,

4039 isAlphaNum)

4140import Data.List (intercalate)

4241import Text.Pandoc.Definition

43−import Text.Pandoc.Class (PandocMonad, report)

44−import Text.Pandoc.Logging

4542import Text.Pandoc.Options

4643import qualified Text.Parsec as P

4744

@@ -68,17 +65,6 @@ getLang opts meta =

6865 Just (MetaString s) -> Just s

6966 _ -> Nothing

7067

71−-- | Convert BCP47 string to a Lang, issuing warning

72−-- if there are problems.

73−toLang :: PandocMonad m => Maybe String -> m (Maybe Lang)

74−toLang Nothing = return Nothing

75−toLang (Just s) =

76− case parseBCP47 s of

77− Left _ -> do

78− report $ InvalidLang s

79− return Nothing

80− Right l -> return (Just l)

81−

8268-- | Parse a BCP 47 string as a Lang. Currently we parse

8369-- extensions and private-use fields as "variants," even

8470-- though officially they aren't.

src/Text/Pandoc/Class.hs+212−46

@@ -1,3 +1,4 @@

1+{-# LANGUAGE CPP #-}

12{-# LANGUAGE DeriveFunctor #-}

23{-# LANGUAGE TypeSynonymInstances #-}

34{-# LANGUAGE FlexibleInstances #-}

@@ -71,8 +72,15 @@ module Text.Pandoc.Class ( PandocMonad(..)

7172 , runIO

7273 , runIOorExplode

7374 , runPure

75+ , readDefaultDataFile

76+ , readDataFile

7477 , fillMediaBag

7578 , extractMedia

79+ , toLang

80+ , setTranslations

81+ , translateTerm

82+ , Translations(..)

83+ , Term(..)

7684 ) where

7785

7886import Prelude hiding (readFile)

@@ -81,10 +89,9 @@ import qualified System.Random as IO (newStdGen)

8189import Codec.Archive.Zip (Archive, fromArchive, emptyArchive)

8290import Data.Unique (hashUnique)

8391import qualified Data.Unique as IO (newUnique)

84−import qualified Text.Pandoc.Shared as IO ( readDataFile

85− , readDefaultDataFile

86− , openURL )

92+import qualified Text.Pandoc.Shared as Shared

8793import qualified Text.Pandoc.UTF8 as UTF8

94+import qualified System.Directory as Directory

8895import Text.Pandoc.Compat.Time (UTCTime)

8996import Text.Pandoc.Logging

9097import Text.Parsec (ParsecT, getPosition, sourceLine, sourceName)

@@ -98,6 +105,7 @@ import Data.Time.Clock.POSIX ( utcTimeToPOSIXSeconds

98105 , posixSecondsToUTCTime

99106 , POSIXTime )

100107import Data.Time.LocalTime (TimeZone, ZonedTime, utcToZonedTime, utc)

108+import qualified System.FilePath.Posix as Posix

101109import Network.URI ( escapeURIString, nonStrictRelativeTo,

102110 unEscapeString, parseURIReference, isAllowedInURI,

103111 parseURI, URI(..) )

@@ -111,7 +119,7 @@ import qualified System.Environment as IO (lookupEnv)

111119import System.FilePath.Glob (match, compile)

112120import System.Directory (createDirectoryIfMissing, getDirectoryContents,

113121 doesDirectoryExist)

114−import System.FilePath ((</>), (<.>), takeDirectory,

122+import System.FilePath ((</>), (<.>), takeDirectory, splitDirectories,

115123 takeExtension, dropExtension, isRelative, normalise)

116124import qualified System.FilePath.Glob as IO (glob)

117125import qualified System.Directory as IO (getModificationTime)

@@ -121,13 +129,21 @@ import Control.Monad.State.Strict

121129import Control.Monad.Except

122130import Control.Monad.Writer (WriterT)

123131import Control.Monad.RWS (RWST)

132+import Codec.Archive.Zip

124133import Data.Word (Word8)

125134import Data.Default

126135import System.IO.Error

127136import System.IO (stderr)

128137import qualified Data.Map as M

129138import Text.Pandoc.Error

139+import Text.Pandoc.BCP47 (Lang(..), parseBCP47, renderLang)

140+import Text.Pandoc.Translations (Term(..), Translations(..), readTranslations)

130141import qualified Debug.Trace

142+#ifdef EMBED_DATA_FILES

143+import Text.Pandoc.Data (dataFiles)

144+#else

145+import Paths_pandoc (getDataFileName)

146+#endif

131147

132148-- | The PandocMonad typeclass contains all the potentially

133149-- IO-related functions used in pandoc's readers and writers.

@@ -155,15 +171,12 @@ class (Functor m, Applicative m, Monad m, MonadError PandocError m)

155171 -- | Read the strict ByteString contents from a file path,

156172 -- raising an error on failure.

157173 readFileStrict :: FilePath -> m B.ByteString

158− -- | Read file from from Cabal data directory.

159− readDefaultDataFile :: FilePath -> m B.ByteString

160− -- | Read file from user data directory or,

161− -- if not found there, from Cabal data directory.

162− readDataFile :: FilePath -> m B.ByteString

163174 -- | Return a list of paths that match a glob, relative to

164175 -- the working directory. See 'System.FilePath.Glob' for

165176 -- the glob syntax.

166177 glob :: String -> m [FilePath]

178+ -- | Returns True if file exists.

179+ fileExists :: FilePath -> m Bool

167180 -- | Return the modification time of a file.

168181 getModificationTime :: FilePath -> m UTCTime

169182 -- | Get the value of the 'CommonState' used by all instances

@@ -272,6 +285,9 @@ data CommonState = CommonState { stLog :: [LogMessage]

272285 -- ^ Directory to search for data files

273286 , stMediaBag :: MediaBag

274287 -- ^ Media parsed from binary containers

288+ , stTranslations :: Maybe

289+ (Lang, Maybe Translations)

290+ -- ^ Translations for localization

275291 , stInputFiles :: Maybe [FilePath]

276292 -- ^ List of input files from command line

277293 , stOutputFile :: Maybe FilePath

@@ -290,6 +306,7 @@ instance Default CommonState where

290306 def = CommonState { stLog = []

291307 , stUserDataDir = Nothing

292308 , stMediaBag = mempty

309+ , stTranslations = Nothing

293310 , stInputFiles = Nothing

294311 , stOutputFile = Nothing

295312 , stResourcePath = ["."]

@@ -297,6 +314,71 @@ instance Default CommonState where

297314 , stTrace = False

298315 }

299316

317+-- | Convert BCP47 string to a Lang, issuing warning

318+-- if there are problems.

319+toLang :: PandocMonad m => Maybe String -> m (Maybe Lang)

320+toLang Nothing = return Nothing

321+toLang (Just s) =

322+ case parseBCP47 s of

323+ Left _ -> do

324+ report $ InvalidLang s

325+ return Nothing

326+ Right l -> return (Just l)

327+

328+-- | Select the language to use with 'translateTerm'.

329+-- Note that this does not read a translation file;

330+-- that is only done the first time 'translateTerm' is

331+-- used.

332+setTranslations :: PandocMonad m => Lang -> m ()

333+setTranslations lang =

334+ modifyCommonState $ \st -> st{ stTranslations = Just (lang, Nothing) }

335+

336+-- | Load term map.

337+getTranslations :: PandocMonad m => m Translations

338+getTranslations = do

339+ mbtrans <- getsCommonState stTranslations

340+ case mbtrans of

341+ Nothing -> return mempty -- no language defined

342+ Just (_, Just t) -> return t

343+ Just (lang, Nothing) -> do -- read from file

344+ let translationFile = "translations/" ++ renderLang lang ++ ".trans"

345+ let fallbackFile = "translations/" ++ langLanguage lang ++ ".trans"

346+ let getTrans bs =

347+ case readTranslations (UTF8.toString bs) of

348+ Left e -> do

349+ report $ CouldNotLoadTranslations (renderLang lang) e

350+ -- make sure we don't try again...

351+ modifyCommonState $ \st ->

352+ st{ stTranslations = Nothing }

353+ return mempty

354+ Right t -> do

355+ modifyCommonState $ \st ->

356+ st{ stTranslations = Just (lang, Just t) }

357+ return t

358+ catchError (readDataFile translationFile >>= getTrans)

359+ (\_ ->

360+ catchError (readDataFile fallbackFile >>= getTrans)

361+ (\e -> do

362+ report $ CouldNotLoadTranslations (renderLang lang)

363+ $ case e of

364+ PandocCouldNotFindDataFileError _ ->

365+ ("data file " ++ fallbackFile ++ " not found")

366+ _ -> ""

367+ -- make sure we don't try again...

368+ modifyCommonState $ \st -> st{ stTranslations = Nothing }

369+ return mempty))

370+

371+-- | Get a translation from the current term map.

372+-- Issue a warning if the term is not defined.

373+translateTerm :: PandocMonad m => Term -> m String

374+translateTerm term = do

375+ Translations termMap <- getTranslations

376+ case M.lookup term termMap of

377+ Just s -> return s

378+ Nothing -> do

379+ report $ NoTranslation (show term)

380+ return ""

381+

300382-- | Evaluate a 'PandocIO' operation.

301383runIO :: PandocIO a -> IO (Either PandocError a)

302384runIO ma = flip evalStateT def $ runExceptT $ unPandocIO ma

@@ -317,7 +399,7 @@ liftIOError :: (String -> IO a) -> String -> PandocIO a

317399liftIOError f u = do

318400 res <- liftIO $ tryIOError $ f u

319401 case res of

320− Left e -> throwError $ PandocIOError u e

402+ Left e -> throwError $ PandocIOError u e

321403 Right r -> return r

322404

323405instance PandocMonad PandocIO where

@@ -328,17 +410,15 @@ instance PandocMonad PandocIO where

328410 newUniqueHash = hashUnique <$> liftIO IO.newUnique

329411 openURL u = do

330412 report $ Fetching u

331− res <- liftIO (IO.openURL u)

413+ res <- liftIOError Shared.openURL u

332414 case res of

333415 Right r -> return r

334416 Left e -> throwError $ PandocHttpError u e

335417 readFileLazy s = liftIOError BL.readFile s

336418 readFileStrict s = liftIOError B.readFile s

337− readDefaultDataFile fname = liftIOError IO.readDefaultDataFile fname

338− readDataFile fname = do

339− datadir <- getUserDataDir

340− liftIOError (IO.readDataFile datadir) fname

341− glob = liftIO . IO.glob

419+

420+ glob = liftIOError IO.glob

421+ fileExists = liftIOError Directory.doesFileExist

342422 getModificationTime fp = liftIOError IO.getModificationTime fp

343423 getCommonState = PandocIO $ lift get

344424 putCommonState x = PandocIO $ lift $ put x

@@ -432,6 +512,109 @@ downloadOrRead sourceURL s =

432512 convertSlash '\\' = '/'

433513 convertSlash x = x

434514

515+getDefaultReferenceDocx :: PandocMonad m => m Archive

516+getDefaultReferenceDocx = do

517+ let paths = ["[Content_Types].xml",

518+ "_rels/.rels",

519+ "docProps/app.xml",

520+ "docProps/core.xml",

521+ "word/document.xml",

522+ "word/fontTable.xml",

523+ "word/footnotes.xml",

524+ "word/numbering.xml",

525+ "word/settings.xml",

526+ "word/webSettings.xml",

527+ "word/styles.xml",

528+ "word/_rels/document.xml.rels",

529+ "word/_rels/footnotes.xml.rels",

530+ "word/theme/theme1.xml"]

531+ let toLazy = BL.fromChunks . (:[])

532+ let pathToEntry path = do

533+ epochtime <- (floor . utcTimeToPOSIXSeconds) <$> getCurrentTime

534+ contents <- toLazy <$> readDataFile ("docx/" ++ path)

535+ return $ toEntry path epochtime contents

536+ datadir <- getUserDataDir

537+ mbArchive <- case datadir of

538+ Nothing -> return Nothing

539+ Just d -> do

540+ exists <- fileExists (d </> "reference.docx")

541+ if exists

542+ then return (Just (d </> "reference.docx"))

543+ else return Nothing

544+ case mbArchive of

545+ Just arch -> toArchive <$> readFileLazy arch

546+ Nothing -> foldr addEntryToArchive emptyArchive <$>

547+ mapM pathToEntry paths

548+

549+getDefaultReferenceODT :: PandocMonad m => m Archive

550+getDefaultReferenceODT = do

551+ let paths = ["mimetype",

552+ "manifest.rdf",

553+ "styles.xml",

554+ "content.xml",

555+ "meta.xml",

556+ "settings.xml",

557+ "Configurations2/accelerator/current.xml",

558+ "Thumbnails/thumbnail.png",

559+ "META-INF/manifest.xml"]

560+ let pathToEntry path = do epochtime <- floor `fmap` getPOSIXTime

561+ contents <- (BL.fromChunks . (:[])) `fmap`

562+ readDataFile ("odt/" ++ path)

563+ return $ toEntry path epochtime contents

564+ datadir <- getUserDataDir

565+ mbArchive <- case datadir of

566+ Nothing -> return Nothing

567+ Just d -> do

568+ exists <- fileExists (d </> "reference.odt")

569+ if exists

570+ then return (Just (d </> "reference.odt"))

571+ else return Nothing

572+ case mbArchive of

573+ Just arch -> toArchive <$> readFileLazy arch

574+ Nothing -> foldr addEntryToArchive emptyArchive <$>

575+ mapM pathToEntry paths

576+

577+-- | Read file from user data directory or,

578+-- if not found there, from Cabal data directory.

579+readDataFile :: PandocMonad m => FilePath -> m B.ByteString

580+readDataFile fname = do

581+ datadir <- getUserDataDir

582+ case datadir of

583+ Nothing -> readDefaultDataFile fname

584+ Just userDir -> do

585+ exists <- fileExists (userDir </> fname)

586+ if exists

587+ then readFileStrict (userDir </> fname)

588+ else readDefaultDataFile fname

589+

590+-- | Read file from from Cabal data directory.

591+readDefaultDataFile :: PandocMonad m => FilePath -> m B.ByteString

592+readDefaultDataFile "reference.docx" =

593+ (B.concat . BL.toChunks . fromArchive) <$> getDefaultReferenceDocx

594+readDefaultDataFile "reference.odt" =

595+ (B.concat . BL.toChunks . fromArchive) <$> getDefaultReferenceODT

596+readDefaultDataFile fname =

597+#ifdef EMBED_DATA_FILES

598+ case lookup (makeCanonical fname) dataFiles of

599+ Nothing -> throwError $ PandocCouldNotFindDataFileError fname

600+ Just contents -> return contents

601+ where makeCanonical = Posix.joinPath . transformPathParts . splitDirectories

602+ transformPathParts = reverse . foldl go []

603+ go as "." = as

604+ go (_:as) ".." = as

605+ go as x = x : as

606+#else

607+ getDataFileName fname' >>= checkExistence >>= readFileStrict

608+ where fname' = if fname == "MANUAL.txt" then fname else "data" </> fname

609+

610+checkExistence :: PandocMonad m => FilePath -> m FilePath

611+checkExistence fn = do

612+ exists <- fileExists fn

613+ if exists

614+ then return fn

615+ else throwError $ PandocCouldNotFindDataFileError fn

616+#endif

617+

435618withPaths :: PandocMonad m => [FilePath] -> (FilePath -> m a) -> FilePath -> m a

436619withPaths [] _ fp = throwError $ PandocResourceNotFound fp

437620withPaths (p:ps) action fp =

@@ -491,9 +674,8 @@ writeMedia dir mediabag subpath = do

491674 Nothing -> throwError $ PandocResourceNotFound subpath

492675 Just (_, bs) -> do

493676 report $ Extracting fullpath

494− liftIO $ do

495− createDirectoryIfMissing True $ takeDirectory fullpath

496− BL.writeFile fullpath bs

677+ liftIOError (createDirectoryIfMissing True) (takeDirectory fullpath)

678+ liftIOError (\p -> BL.writeFile p bs) fullpath

497679

498680adjustImagePath :: FilePath -> [FilePath] -> Inline -> Inline

499681adjustImagePath dir paths (Image attr lab (src, tit))

@@ -624,28 +806,17 @@ instance PandocMonad PandocPure where

624806 case infoFileContents <$> getFileInfo fp fps of

625807 Just bs -> return bs

626808 Nothing -> throwError $ PandocResourceNotFound fp

627− readDefaultDataFile "reference.docx" =

628− (B.concat . BL.toChunks . fromArchive) <$> getsPureState stReferenceDocx

629− readDefaultDataFile "reference.odt" =

630− (B.concat . BL.toChunks . fromArchive) <$> getsPureState stReferenceODT

631− readDefaultDataFile fname = do

632− let fname' = if fname == "MANUAL.txt" then fname else "data" </> fname

633− readFileStrict fname'

634− readDataFile fname = do

635− datadir <- getUserDataDir

636− case datadir of

637− Just userDir -> do

638− userDirFiles <- getsPureState stUserDataFiles

639− case infoFileContents <$> getFileInfo (userDir </> fname)

640− userDirFiles of

641− Just bs -> return bs

642− Nothing -> readDefaultDataFile fname

643− Nothing -> readDefaultDataFile fname

644809

645810 glob s = do

646811 FileTree ftmap <- getsPureState stFiles

647812 return $ filter (match (compile s)) $ M.keys ftmap

648813

814+ fileExists fp = do

815+ fps <- getsPureState stFiles

816+ case getFileInfo fp fps of

817+ Nothing -> return False

818+ Just _ -> return True

819+

649820 getModificationTime fp = do

650821 fps <- getsPureState stFiles

651822 case infoFileMTime <$> getFileInfo fp fps of

@@ -667,9 +838,8 @@ instance PandocMonad m => PandocMonad (ParsecT s st m) where

667838 openURL = lift . openURL

668839 readFileLazy = lift . readFileLazy

669840 readFileStrict = lift . readFileStrict

670− readDefaultDataFile = lift . readDefaultDataFile

671− readDataFile = lift . readDataFile

672841 glob = lift . glob

842+ fileExists = lift . fileExists

673843 getModificationTime = lift . getModificationTime

674844 getCommonState = lift getCommonState

675845 putCommonState = lift . putCommonState

@@ -696,9 +866,8 @@ instance PandocMonad m => PandocMonad (ReaderT r m) where

696866 openURL = lift . openURL

697867 readFileLazy = lift . readFileLazy

698868 readFileStrict = lift . readFileStrict

699− readDefaultDataFile = lift . readDefaultDataFile

700− readDataFile = lift . readDataFile

701869 glob = lift . glob

870+ fileExists = lift . fileExists

702871 getModificationTime = lift . getModificationTime

703872 getCommonState = lift getCommonState

704873 putCommonState = lift . putCommonState

@@ -713,9 +882,8 @@ instance (PandocMonad m, Monoid w) => PandocMonad (WriterT w m) where

713882 openURL = lift . openURL

714883 readFileLazy = lift . readFileLazy

715884 readFileStrict = lift . readFileStrict

716− readDefaultDataFile = lift . readDefaultDataFile

717− readDataFile = lift . readDataFile

718885 glob = lift . glob

886+ fileExists = lift . fileExists

719887 getModificationTime = lift . getModificationTime

720888 getCommonState = lift getCommonState

721889 putCommonState = lift . putCommonState

@@ -730,9 +898,8 @@ instance (PandocMonad m, Monoid w) => PandocMonad (RWST r w st m) where

730898 openURL = lift . openURL

731899 readFileLazy = lift . readFileLazy

732900 readFileStrict = lift . readFileStrict

733− readDefaultDataFile = lift . readDefaultDataFile

734− readDataFile = lift . readDataFile

735901 glob = lift . glob

902+ fileExists = lift . fileExists

736903 getModificationTime = lift . getModificationTime

737904 getCommonState = lift getCommonState

738905 putCommonState = lift . putCommonState

@@ -747,9 +914,8 @@ instance PandocMonad m => PandocMonad (StateT st m) where

747914 openURL = lift . openURL

748915 readFileLazy = lift . readFileLazy

749916 readFileStrict = lift . readFileStrict

750− readDefaultDataFile = lift . readDefaultDataFile

751− readDataFile = lift . readDataFile

752917 glob = lift . glob

918+ fileExists = lift . fileExists

753919 getModificationTime = lift . getModificationTime

754920 getCommonState = lift getCommonState

755921 putCommonState = lift . putCommonState

src/Text/Pandoc/Logging.hs+14

@@ -95,6 +95,8 @@ data LogMessage =

9595 | CouldNotHighlight String

9696 | MissingCharacter String

9797 | Deprecated String String

98+ | NoTranslation String

99+ | CouldNotLoadTranslations String String

98100 deriving (Show, Eq, Data, Ord, Typeable, Generic)

99101

100102instance ToJSON LogMessage where

@@ -195,6 +197,11 @@ instance ToJSON LogMessage where

195197 Deprecated thing msg ->

196198 ["thing" .= Text.pack thing,

197199 "message" .= Text.pack msg]

200+ NoTranslation term ->

201+ ["term" .= Text.pack term]

202+ CouldNotLoadTranslations lang msg ->

203+ ["lang" .= Text.pack lang,

204+ "message" .= Text.pack msg]

198205

199206

200207showPos :: SourcePos -> String

@@ -282,6 +289,11 @@ showLogMessage msg =

282289 if null m

283290 then ""

284291 else ". " ++ m

292+ NoTranslation t ->

293+ "The term " ++ t ++ " has no translation defined."

294+ CouldNotLoadTranslations lang m ->

295+ "Could not load translations for " ++ lang ++

296+ if null m then "" else ('\n':m)

285297

286298messageVerbosity:: LogMessage -> Verbosity

287299messageVerbosity msg =

@@ -314,3 +326,5 @@ messageVerbosity msg =

314326 CouldNotHighlight{} -> WARNING

315327 MissingCharacter{} -> WARNING

316328 Deprecated{} -> WARNING

329+ NoTranslation{} -> WARNING

330+ CouldNotLoadTranslations{} -> WARNING

src/Text/Pandoc/Lua/PandocModule.hs+3−3

@@ -32,13 +32,12 @@ import Data.ByteString.Char8 (unpack)

3232import Data.Default (Default (..))

3333import Data.Text (pack)

3434import Scripting.Lua (LuaState, call, push, pushhsfunction, rawset)

35−import Text.Pandoc.Class hiding (readDataFile)

35+import Text.Pandoc.Class

3636import Text.Pandoc.Definition (Pandoc)

3737import Text.Pandoc.Options (ReaderOptions(readerExtensions))

3838import Text.Pandoc.Lua.Compat (loadstring)

3939import Text.Pandoc.Lua.StackInstances ()

4040import Text.Pandoc.Readers (Reader (..), getReader)

41−import Text.Pandoc.Shared (readDataFile)

4241

4342-- | Push the "pandoc" on the lua stack.

4443pushPandocModule :: Maybe FilePath -> LuaState -> IO ()

@@ -52,7 +51,8 @@ pushPandocModule datadir lua = do

5251

5352-- | Get the string representation of the pandoc module

5453pandocModuleScript :: Maybe FilePath -> IO String

55−pandocModuleScript datadir = unpack <$> readDataFile datadir "pandoc.lua"

54+pandocModuleScript datadir = unpack <$>

55+ runIOorExplode (setUserDataDir datadir >> readDataFile "pandoc.lua")

5656

5757read_doc :: String -> String -> IO (Either String Pandoc)

5858read_doc formatSpec content = do

src/Text/Pandoc/Readers/LaTeX.hs+133−2

@@ -55,8 +55,11 @@ import Data.Maybe (fromMaybe, maybeToList)

5555import Safe (minimumDef)

5656import System.FilePath (addExtension, replaceExtension, takeExtension)

5757import Text.Pandoc.Builder

58−import Text.Pandoc.Class (PandocMonad, PandocPure, lookupEnv, readFileFromDirs,

59− report, setResourcePath, getResourcePath)

58+import Text.Pandoc.Class (PandocMonad, PandocPure, lookupEnv,

59+ readFileFromDirs, report, setResourcePath,

60+ getResourcePath, setTranslations, translateTerm)

61+import qualified Text.Pandoc.Translations as Translations

62+import Text.Pandoc.BCP47 (Lang(..))

6063import Text.Pandoc.Highlighting (fromListingsLanguage, languagesByExtension)

6164import Text.Pandoc.ImageSize (numUnit, showFl)

6265import Text.Pandoc.Logging

@@ -1247,6 +1250,7 @@ inlineCommands = M.fromList $

12471250 removeDoubleQuotes . untokenize <$> braced

12481251 mkImage options src)

12491252 , ("enquote", enquote)

1253+ , ("figurename", doTerm Translations.Figure)

12501254 , ("cite", citation "cite" NormalCitation False)

12511255 , ("Cite", citation "Cite" NormalCitation False)

12521256 , ("citep", citation "citep" NormalCitation False)

@@ -1326,6 +1330,12 @@ inlineCommands = M.fromList $

13261330 , ("ifstrequal", ifstrequal)

13271331 ]

13281332

1333+doTerm :: PandocMonad m => Translations.Term -> LP m Inlines

1334+doTerm term = do

1335+ s <- (symbol '~' >> return (str "\160")) <|> return space

1336+ t <- translateTerm term

1337+ return (str t <> s)

1338+

13291339ifstrequal :: PandocMonad m => LP m Inlines

13301340ifstrequal = do

13311341 str1 <- tok

@@ -1759,6 +1769,9 @@ blockCommands = M.fromList $

17591769 -- includes

17601770 , ("lstinputlisting", inputListing)

17611771 , ("graphicspath", graphicsPath)

1772+ -- polyglossia

1773+ , ("setdefaultlanguage", setDefaultLanguage)

1774+ , ("setmainlanguage", setDefaultLanguage)

17621775 -- hyperlink

17631776 , ("hypertarget", try $ braced >> grouped block)

17641777 -- LaTeX colors

@@ -2206,3 +2219,121 @@ block = (mempty <$ spaces1)

22062219blocks :: PandocMonad m => LP m Blocks

22072220blocks = mconcat <$> many block

22082221

2222+setDefaultLanguage :: PandocMonad m => LP m Blocks

2223+setDefaultLanguage = do

2224+ o <- option "" $ (T.unpack . T.filter (\c -> c /= '[' && c /= ']'))

2225+ <$> rawopt

2226+ polylang <- toksToString <$> braced

2227+ case polyglossiaLangToBCP47 polylang o of

2228+ Nothing -> return () -- TODO mzero? warning?

2229+ Just l -> setTranslations l

2230+ return mempty

2231+

2232+polyglossiaLangToBCP47 :: String -> String -> Maybe Lang

2233+polyglossiaLangToBCP47 s o =

2234+ case (s, filter (/=' ') o) of

2235+ ("arabic", "locale=algeria") -> Just $ Lang "ar" "" "DZ" []

2236+ ("arabic", "locale=mashriq") -> Just $ Lang "ar" "" "SY" []

2237+ ("arabic", "locale=libya") -> Just $ Lang "ar" "" "LY" []

2238+ ("arabic", "locale=morocco") -> Just $ Lang "ar" "" "MA" []

2239+ ("arabic", "locale=mauritania") -> Just $ Lang "ar" "" "MR" []

2240+ ("arabic", "locale=tunisia") -> Just $ Lang "ar" "" "TN" []

2241+ ("german", "spelling=old") -> Just $ Lang "de" "" "DE" ["1901"]

2242+ ("german", "variant=austrian,spelling=old")

2243+ -> Just $ Lang "de" "" "AT" ["1901"]

2244+ ("german", "variant=austrian") -> Just $ Lang "de" "" "AT" []

2245+ ("german", "variant=swiss,spelling=old")

2246+ -> Just $ Lang "de" "" "CH" ["1901"]

2247+ ("german", "variant=swiss") -> Just $ Lang "de" "" "CH" []

2248+ ("german", _) -> Just $ Lang "de" "" "" []

2249+ ("lsorbian", _) -> Just $ Lang "dsb" "" "" []

2250+ ("greek", "variant=poly") -> Just $ Lang "el" "" "polyton" []

2251+ ("english", "variant=australian") -> Just $ Lang "en" "" "AU" []

2252+ ("english", "variant=canadian") -> Just $ Lang "en" "" "CA" []

2253+ ("english", "variant=british") -> Just $ Lang "en" "" "GB" []

2254+ ("english", "variant=newzealand") -> Just $ Lang "en" "" "NZ" []

2255+ ("english", "variant=american") -> Just $ Lang "en" "" "US" []

2256+ ("greek", "variant=ancient") -> Just $ Lang "grc" "" "" []

2257+ ("usorbian", _) -> Just $ Lang "hsb" "" "" []

2258+ ("latin", "variant=classic") -> Just $ Lang "la" "" "" ["x-classic"]

2259+ ("slovenian", _) -> Just $ Lang "sl" "" "" []

2260+ ("serbianc", _) -> Just $ Lang "sr" "cyrl" "" []

2261+ ("pinyin", _) -> Just $ Lang "zh" "Latn" "" ["pinyin"]

2262+ ("afrikaans", _) -> Just $ Lang "af" "" "" []

2263+ ("amharic", _) -> Just $ Lang "am" "" "" []

2264+ ("arabic", _) -> Just $ Lang "ar" "" "" []

2265+ ("assamese", _) -> Just $ Lang "as" "" "" []

2266+ ("asturian", _) -> Just $ Lang "ast" "" "" []

2267+ ("bulgarian", _) -> Just $ Lang "bg" "" "" []

2268+ ("bengali", _) -> Just $ Lang "bn" "" "" []

2269+ ("tibetan", _) -> Just $ Lang "bo" "" "" []

2270+ ("breton", _) -> Just $ Lang "br" "" "" []

2271+ ("catalan", _) -> Just $ Lang "ca" "" "" []

2272+ ("welsh", _) -> Just $ Lang "cy" "" "" []

2273+ ("czech", _) -> Just $ Lang "cs" "" "" []

2274+ ("coptic", _) -> Just $ Lang "cop" "" "" []

2275+ ("danish", _) -> Just $ Lang "da" "" "" []

2276+ ("divehi", _) -> Just $ Lang "dv" "" "" []

2277+ ("greek", _) -> Just $ Lang "el" "" "" []

2278+ ("english", _) -> Just $ Lang "en" "" "" []

2279+ ("esperanto", _) -> Just $ Lang "eo" "" "" []

2280+ ("spanish", _) -> Just $ Lang "es" "" "" []

2281+ ("estonian", _) -> Just $ Lang "et" "" "" []

2282+ ("basque", _) -> Just $ Lang "eu" "" "" []

2283+ ("farsi", _) -> Just $ Lang "fa" "" "" []

2284+ ("finnish", _) -> Just $ Lang "fi" "" "" []

2285+ ("french", _) -> Just $ Lang "fr" "" "" []

2286+ ("friulan", _) -> Just $ Lang "fur" "" "" []

2287+ ("irish", _) -> Just $ Lang "ga" "" "" []

2288+ ("scottish", _) -> Just $ Lang "gd" "" "" []

2289+ ("ethiopic", _) -> Just $ Lang "gez" "" "" []

2290+ ("galician", _) -> Just $ Lang "gl" "" "" []

2291+ ("hebrew", _) -> Just $ Lang "he" "" "" []

2292+ ("hindi", _) -> Just $ Lang "hi" "" "" []

2293+ ("croatian", _) -> Just $ Lang "hr" "" "" []

2294+ ("magyar", _) -> Just $ Lang "hu" "" "" []

2295+ ("armenian", _) -> Just $ Lang "hy" "" "" []

2296+ ("interlingua", _) -> Just $ Lang "ia" "" "" []

2297+ ("indonesian", _) -> Just $ Lang "id" "" "" []

2298+ ("icelandic", _) -> Just $ Lang "is" "" "" []

2299+ ("italian", _) -> Just $ Lang "it" "" "" []

2300+ ("japanese", _) -> Just $ Lang "jp" "" "" []

2301+ ("khmer", _) -> Just $ Lang "km" "" "" []

2302+ ("kurmanji", _) -> Just $ Lang "kmr" "" "" []

2303+ ("kannada", _) -> Just $ Lang "kn" "" "" []

2304+ ("korean", _) -> Just $ Lang "ko" "" "" []

2305+ ("latin", _) -> Just $ Lang "la" "" "" []

2306+ ("lao", _) -> Just $ Lang "lo" "" "" []

2307+ ("lithuanian", _) -> Just $ Lang "lt" "" "" []

2308+ ("latvian", _) -> Just $ Lang "lv" "" "" []

2309+ ("malayalam", _) -> Just $ Lang "ml" "" "" []

2310+ ("mongolian", _) -> Just $ Lang "mn" "" "" []

2311+ ("marathi", _) -> Just $ Lang "mr" "" "" []

2312+ ("dutch", _) -> Just $ Lang "nl" "" "" []

2313+ ("nynorsk", _) -> Just $ Lang "nn" "" "" []

2314+ ("norsk", _) -> Just $ Lang "no" "" "" []

2315+ ("nko", _) -> Just $ Lang "nqo" "" "" []

2316+ ("occitan", _) -> Just $ Lang "oc" "" "" []

2317+ ("panjabi", _) -> Just $ Lang "pa" "" "" []

2318+ ("polish", _) -> Just $ Lang "pl" "" "" []

2319+ ("piedmontese", _) -> Just $ Lang "pms" "" "" []

2320+ ("portuguese", _) -> Just $ Lang "pt" "" "" []

2321+ ("romansh", _) -> Just $ Lang "rm" "" "" []

2322+ ("romanian", _) -> Just $ Lang "ro" "" "" []

2323+ ("russian", _) -> Just $ Lang "ru" "" "" []

2324+ ("sanskrit", _) -> Just $ Lang "sa" "" "" []

2325+ ("samin", _) -> Just $ Lang "se" "" "" []

2326+ ("slovak", _) -> Just $ Lang "sk" "" "" []

2327+ ("albanian", _) -> Just $ Lang "sq" "" "" []

2328+ ("serbian", _) -> Just $ Lang "sr" "" "" []

2329+ ("swedish", _) -> Just $ Lang "sv" "" "" []

2330+ ("syriac", _) -> Just $ Lang "syr" "" "" []

2331+ ("tamil", _) -> Just $ Lang "ta" "" "" []

2332+ ("telugu", _) -> Just $ Lang "te" "" "" []

2333+ ("thai", _) -> Just $ Lang "th" "" "" []

2334+ ("turkmen", _) -> Just $ Lang "tk" "" "" []

2335+ ("turkish", _) -> Just $ Lang "tr" "" "" []

2336+ ("ukrainian", _) -> Just $ Lang "uk" "" "" []

2337+ ("urdu", _) -> Just $ Lang "ur" "" "" []

2338+ ("vietnamese", _) -> Just $ Lang "vi" "" "" []

2339+ _ -> Nothing

src/Text/Pandoc/Shared.hs+1−118

@@ -76,11 +76,6 @@ module Text.Pandoc.Shared (

7676 renderTags',

7777 -- * File handling

7878 inDirectory,

79− getDefaultReferenceDocx,

80− getDefaultReferenceODT,

81− readDefaultDataFile,

82− readDataFile,

83− readDataFileUTF8,

8479 openURL,

8580 collapseFilePath,

8681 filteredFilesFromArchive,

@@ -116,8 +111,6 @@ import System.Directory

116111import System.FilePath (splitDirectories, isPathSeparator)

117112import qualified System.FilePath.Posix as Posix

118113import Text.Pandoc.MIME (MimeType)

119−import Text.Pandoc.Error (PandocError(..))

120−import System.FilePath ( (</>) )

121114import Data.Generics (Typeable, Data)

122115import qualified Control.Monad.State.Strict as S

123116import qualified Control.Exception as E

@@ -125,7 +118,6 @@ import Control.Monad (msum, unless, MonadPlus(..))

125118import Text.Pandoc.Pretty (charWidth)

126119import Text.Pandoc.Generic (bottomUp)

127120import Text.Pandoc.Compat.Time

128−import Data.Time.Clock.POSIX

129121import System.IO.Error

130122import System.IO.Temp

131123import Text.HTML.TagSoup (renderTagsOptions, RenderOptions(..), Tag(..),

@@ -136,17 +128,12 @@ import qualified Data.ByteString.Char8 as B8

136128import Data.ByteString.Base64 (decodeLenient)

137129import Data.Sequence (ViewR(..), ViewL(..), viewl, viewr)

138130import qualified Data.Text as T

139−import Data.ByteString.Lazy (toChunks, fromChunks)

131+import Data.ByteString.Lazy (toChunks)

140132import qualified Data.ByteString.Lazy as BL

141133import Paths_pandoc (version)

142134

143135import Codec.Archive.Zip

144136

145−#ifdef EMBED_DATA_FILES

146−import Text.Pandoc.Data (dataFiles)

147−#else

148−import Paths_pandoc (getDataFileName)

149−#endif

150137import Network.HTTP.Client (httpLbs, responseBody, responseHeaders,

151138 Request(port,host,requestHeaders),

152139 HttpException)

@@ -612,110 +599,6 @@ inDirectory path action = E.bracket

612599 setCurrentDirectory

613600 (const $ setCurrentDirectory path >> action)

614601

615−getDefaultReferenceDocx :: Maybe FilePath -> IO Archive

616−getDefaultReferenceDocx datadir = do

617− let paths = ["[Content_Types].xml",

618− "_rels/.rels",

619− "docProps/app.xml",

620− "docProps/core.xml",

621− "word/document.xml",

622− "word/fontTable.xml",

623− "word/footnotes.xml",

624− "word/numbering.xml",

625− "word/settings.xml",

626− "word/webSettings.xml",

627− "word/styles.xml",

628− "word/_rels/document.xml.rels",

629− "word/_rels/footnotes.xml.rels",

630− "word/theme/theme1.xml"]

631− let toLazy = fromChunks . (:[])

632− let pathToEntry path = do epochtime <- (floor . utcTimeToPOSIXSeconds) <$>

633− getCurrentTime

634− contents <- toLazy <$> readDataFile datadir

635− ("docx/" ++ path)

636− return $ toEntry path epochtime contents

637− mbArchive <- case datadir of

638− Nothing -> return Nothing

639− Just d -> do

640− exists <- doesFileExist (d </> "reference.docx")

641− if exists

642− then return (Just (d </> "reference.docx"))

643− else return Nothing

644− case mbArchive of

645− Just arch -> toArchive <$> BL.readFile arch

646− Nothing -> foldr addEntryToArchive emptyArchive <$>

647− mapM pathToEntry paths

648−

649−getDefaultReferenceODT :: Maybe FilePath -> IO Archive

650−getDefaultReferenceODT datadir = do

651− let paths = ["mimetype",

652− "manifest.rdf",

653− "styles.xml",

654− "content.xml",

655− "meta.xml",

656− "settings.xml",

657− "Configurations2/accelerator/current.xml",

658− "Thumbnails/thumbnail.png",

659− "META-INF/manifest.xml"]

660− let pathToEntry path = do epochtime <- floor `fmap` getPOSIXTime

661− contents <- (fromChunks . (:[])) `fmap`

662− readDataFile datadir ("odt/" ++ path)

663− return $ toEntry path epochtime contents

664− mbArchive <- case datadir of

665− Nothing -> return Nothing

666− Just d -> do

667− exists <- doesFileExist (d </> "reference.odt")

668− if exists

669− then return (Just (d </> "reference.odt"))

670− else return Nothing

671− case mbArchive of

672− Just arch -> toArchive <$> BL.readFile arch

673− Nothing -> foldr addEntryToArchive emptyArchive <$>

674− mapM pathToEntry paths

675−

676−

677−readDefaultDataFile :: FilePath -> IO BS.ByteString

678−readDefaultDataFile "reference.docx" =

679− (BS.concat . toChunks . fromArchive) <$> getDefaultReferenceDocx Nothing

680−readDefaultDataFile "reference.odt" =

681− (BS.concat . toChunks . fromArchive) <$> getDefaultReferenceODT Nothing

682−readDefaultDataFile fname =

683−#ifdef EMBED_DATA_FILES

684− case lookup (makeCanonical fname) dataFiles of

685− Nothing -> E.throwIO $ PandocCouldNotFindDataFileError fname

686− Just contents -> return contents

687− where makeCanonical = Posix.joinPath . transformPathParts . splitDirectories

688− transformPathParts = reverse . foldl go []

689− go as "." = as

690− go (_:as) ".." = as

691− go as x = x : as

692−#else

693− getDataFileName fname' >>= checkExistence >>= BS.readFile

694− where fname' = if fname == "MANUAL.txt" then fname else "data" </> fname

695−

696−checkExistence :: FilePath -> IO FilePath

697−checkExistence fn = do

698− exists <- doesFileExist fn

699− if exists

700− then return fn

701− else E.throwIO $ PandocCouldNotFindDataFileError fn

702−#endif

703−

704−-- | Read file from specified user data directory or, if not found there, from

705−-- Cabal data directory.

706−readDataFile :: Maybe FilePath -> FilePath -> IO BS.ByteString

707−readDataFile Nothing fname = readDefaultDataFile fname

708−readDataFile (Just userDir) fname = do

709− exists <- doesFileExist (userDir </> fname)

710− if exists

711− then BS.readFile (userDir </> fname)

712− else readDefaultDataFile fname

713−

714−-- | Same as 'readDataFile' but returns a String instead of a ByteString.

715−readDataFileUTF8 :: Maybe FilePath -> FilePath -> IO String

716−readDataFileUTF8 userDir fname =

717− UTF8.toString `fmap` readDataFile userDir fname

718−

719602-- | Read from a URL and return raw data and maybe mime type.

720603openURL :: String -> IO (Either HttpException (BS.ByteString, Maybe MimeType))

721604openURL u

src/Text/Pandoc/Templates.hs+1−1

@@ -44,7 +44,7 @@ import qualified Data.Text as T

4444import System.FilePath ((<.>), (</>))

4545import Text.DocTemplates (Template, TemplateTarget, applyTemplate,

4646 compileTemplate, renderTemplate, varListToJSON)

47−import Text.Pandoc.Class (PandocMonad(readDataFile))

47+import Text.Pandoc.Class (readDataFile, PandocMonad)

4848import Text.Pandoc.Error

4949import qualified Text.Pandoc.UTF8 as UTF8

5050

src/Text/Pandoc/Translations.hsnew file+94

@@ -0,0 +1,94 @@

1+{-# LANGUAGE DeriveGeneric #-}

2+{-# LANGUAGE GeneralizedNewtypeDeriving #-}

3+{-

6+This program is free software; you can redistribute it and/or modify

7+it under the terms of the GNU General Public License as published by

8+the Free Software Foundation; either version 2 of the License, or

9+(at your option) any later version.

10+

11+This program is distributed in the hope that it will be useful,

12+but WITHOUT ANY WARRANTY; without even the implied warranty of

13+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the

14+GNU General Public License for more details.

15+

16+You should have received a copy of the GNU General Public License

17+along with this program; if not, write to the Free Software

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

19+-}

20+

21+{- |

22+ Module : Text.Pandoc.Translations

24+ License : GNU GPL, version 2 or above

25+

26+ Maintainer : John MacFarlane <jgm@berkeley.edu>

27+ Stability : alpha

28+ Portability : portable

29+

30+Data types for localization.

31+

32+Translations are stored in @data/translations/langname.trans@,

33+where langname can be the full BCP47 language specifier, or

34+just the language part. File format is:

35+

36+> # A comment, ignored

37+> Figure: Figura

38+> Index: Indeksi

39+

40+-}

41+module Text.Pandoc.Translations (

42+ Term(..)

43+ , Translations(..)

44+ , readTranslations

45+ )

46+where

47+import qualified Data.Map as M

48+import GHC.Generics (Generic)

49+import Text.Pandoc.Shared (trim, safeRead)

50+

51+data Term =

52+ Preface

53+ | References

54+ | Abstract

55+ | Bibliography

56+ | Chapter

57+ | Appendix

58+ | Contents

59+ | ListOfFigures

60+ | ListOfTables

61+ | Index

62+ | Figure

63+ | Table

64+ | Part

65+ | Page

66+ | Proof

67+ | See

68+ | SeeAlso

69+ | Cc

70+ | To

71+ deriving (Show, Eq, Ord, Generic, Read)

72+

73+newtype Translations = Translations (M.Map Term String)

74+ deriving (Show, Eq, Ord, Generic, Monoid)

75+

76+readTranslations :: String -> Either String Translations

77+readTranslations = foldr parseLine (Right mempty) . lines

78+

79+parseLine :: String

80+ -> Either String Translations

81+ -> Either String Translations

82+parseLine _ (Left s) = Left s

83+parseLine ('#':_) x = x

84+parseLine [] x = x

85+parseLine t (Right (Translations tm)) =

86+ if null rest

87+ then Left $ "no colon in " ++ term

88+ else

89+ case safeRead term of

90+ Nothing -> Left $ term ++ " is not a recognized term name"

91+ Just term' -> Right (Translations $ (M.insert term' defn) tm)

92+ where (trm, rest) = break (\c -> c == ':') t

93+ defn = trim $ drop 1 rest

94+ term = trim trm

src/Text/Pandoc/Writers/ConTeXt.hs+1−1

@@ -37,7 +37,7 @@ import Data.Maybe (catMaybes)

3737import Data.Text (Text)

3838import Network.URI (unEscapeString)

3939import Text.Pandoc.BCP47

40−import Text.Pandoc.Class (PandocMonad, report)

40+import Text.Pandoc.Class (PandocMonad, report, toLang)

4141import Text.Pandoc.Logging

4242import Text.Pandoc.Definition

4343import Text.Pandoc.ImageSize

src/Text/Pandoc/Writers/Docx.hs+2−2

@@ -51,7 +51,7 @@ import qualified Data.Text as T

5151import Data.Time.Clock.POSIX

5252import Skylighting

5353import System.Random (randomR)

54−import Text.Pandoc.Class (PandocMonad, report)

54+import Text.Pandoc.Class (PandocMonad, report, toLang)

5555import qualified Text.Pandoc.Class as P

5656import Text.Pandoc.Compat.Time

5757import Text.Pandoc.Definition

@@ -68,7 +68,7 @@ import qualified Text.Pandoc.UTF8 as UTF8

6868import Text.Pandoc.Walk

6969import Text.Pandoc.Writers.Math

7070import Text.Pandoc.Writers.Shared (fixDisplayMath)

71−import Text.Pandoc.BCP47 (getLang, renderLang, toLang)

71+import Text.Pandoc.BCP47 (getLang, renderLang)

7272import Text.Printf (printf)

7373import Text.TeXMath

7474import Text.XML.Light as XML

src/Text/Pandoc/Writers/LaTeX.hs+2−2

@@ -45,8 +45,8 @@ import Data.Maybe (catMaybes, fromMaybe, isJust)

4545import Data.Text (Text)

4646import qualified Data.Text as T

4747import Network.URI (unEscapeString)

48−import Text.Pandoc.BCP47 (Lang(..), toLang, getLang, renderLang)

49−import Text.Pandoc.Class (PandocMonad, report)

48+import Text.Pandoc.BCP47 (Lang(..), getLang, renderLang)

49+import Text.Pandoc.Class (PandocMonad, report, toLang)

5050import Text.Pandoc.Definition

5151import Text.Pandoc.Highlighting (formatLaTeXBlock, formatLaTeXInline, highlight,

5252 styleToLaTeX, toListingsLanguage)

src/Text/Pandoc/Writers/ODT.hs+2−2

@@ -38,7 +38,7 @@ import Data.List (isPrefixOf)

3838import Data.Maybe (fromMaybe)

3939import qualified Data.Text.Lazy as TL

4040import System.FilePath (takeDirectory, takeExtension, (<.>))

41−import Text.Pandoc.Class (PandocMonad, report)

41+import Text.Pandoc.Class (PandocMonad, report, toLang)

4242import qualified Text.Pandoc.Class as P

4343import Text.Pandoc.Definition

4444import Text.Pandoc.ImageSize

@@ -51,7 +51,7 @@ import Text.Pandoc.UTF8 (fromStringLazy, fromTextLazy, toStringLazy)

5151import Text.Pandoc.Walk

5252import Text.Pandoc.Writers.OpenDocument (writeOpenDocument)

5353import Text.Pandoc.Writers.Shared (fixDisplayMath)

54−import Text.Pandoc.BCP47 (getLang, toLang, Lang(..), renderLang)

54+import Text.Pandoc.BCP47 (getLang, Lang(..), renderLang)

5555import Text.Pandoc.XML

5656import Text.TeXMath

5757import Text.XML.Light