1cc11e6 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

Use new UTF8 module in Shared, ODT, and the executables.

1cc11e6John MacFarlane16 years ago

src/Text/Pandoc/ODT.hs−1

compare with master

unifiedsplit

@@ -32,7 +32,6 @@ import Data.List ( find )

32import System.FilePath ( (</>), takeFileName )

33import qualified Data.ByteString.Lazy as B

34import Data.ByteString.Lazy.UTF8 ( fromString )

35−import Prelude hiding ( writeFile, readFile )

36import Codec.Archive.Zip

35import Codec.Archive.Zip

37import Control.Applicative ( (<$>) )

36import Control.Applicative ( (<$>) )

38import Text.ParserCombinators.Parsec

37import Text.ParserCombinators.Parsec

src/Text/Pandoc/Shared.hs+5−11

@@ -112,6 +112,7 @@ module Text.Pandoc.Shared (

112 ) where

113

114import Text.Pandoc.Definition

115+import qualified Text.Pandoc.UTF8 as UTF8 (readFile, putStrLn)

115import Text.ParserCombinators.Parsec

116import Text.ParserCombinators.Parsec

116import Text.PrettyPrint.HughesPJ ( Doc, fsep, ($$), (<>), empty, isEmpty, text, nest )

117import Text.PrettyPrint.HughesPJ ( Doc, fsep, ($$), (<>), empty, isEmpty, text, nest )

117import qualified Text.PrettyPrint.HughesPJ as PP

118import qualified Text.PrettyPrint.HughesPJ as PP

@@ -123,13 +124,6 @@ import Network.URI ( parseURI, URI (..), isAllowedInURI, escapeURIString, unEsca

123import Codec.Binary.UTF8.String ( encodeString, decodeString )

124import Codec.Binary.UTF8.String ( encodeString, decodeString )

124import System.Directory

125import System.Directory

125import System.FilePath ( (</>) )

126import System.FilePath ( (</>) )

126−-- Note: ghc >= 6.12 (base >=4.2) supports unicode through iconv

127−-- So we use System.IO.UTF8 only if we have an earlier version

128−#if MIN_VERSION_base(4,2,0)

129−#else

130−import Prelude hiding ( putStr, putStrLn, writeFile, readFile, getContents )

131−import System.IO.UTF8

132−#endif

133import Data.Generics

127import Data.Generics

134import qualified Control.Monad.State as S

128import qualified Control.Monad.State as S

135import Control.Monad (join)

129import Control.Monad (join)

@@ -676,7 +670,7 @@ readWith parser state input =

676testStringWith :: (Show a) => GenParser Char ParserState a

670testStringWith :: (Show a) => GenParser Char ParserState a

677 -> String

671 -> String

678 -> IO ()

672 -> IO ()

679−testStringWith parser str = putStrLn $ show $

673+testStringWith parser str = UTF8.putStrLn $ show $

680 readWith parser defaultParserState str

674 readWith parser defaultParserState str

681

675

682-- | Parsing options.

676-- | Parsing options.

@@ -1074,6 +1068,6 @@ inDirectory path action = do

1074readDataFile :: Maybe FilePath -> FilePath -> IO String

1068readDataFile :: Maybe FilePath -> FilePath -> IO String

1075readDataFile userDir fname =

1069readDataFile userDir fname =

1076 case userDir of

1070 case userDir of

1077− Nothing -> getDataFileName fname >>= readFile

1071+ Nothing -> getDataFileName fname >>= UTF8.readFile

1078− Just u -> catch (readFile $ u </> fname)

1072+ Just u -> catch (UTF8.readFile $ u </> fname)

1079− (\_ -> getDataFileName fname >>= readFile)

1073+ (\_ -> getDataFileName fname >>= UTF8.readFile)

src/markdown2pdf.hs+8−15

@@ -9,14 +9,7 @@ import Control.Exception (tryJust, bracket)

9import System.IO (stderr)

10import System.IO.Error (isDoesNotExistError)

11import System.Environment ( getArgs, getProgName )

12−-- Note: ghc >= 6.12 (base >=4.2) supports unicode through iconv

12+import qualified Text.Pandoc.UTF8 as UTF8

13−-- So we use System.IO.UTF8 only if we have an earlier version

14−#if MIN_VERSION_base(4,2,0)

15−import System.IO (hPutStrLn)

16−#else

17−import Prelude hiding ( putStr, putStrLn, writeFile, readFile, getContents )

18−import System.IO.UTF8

19−#endif

20import System.Exit (ExitCode (..), exitWith)

13import System.Exit (ExitCode (..), exitWith)

21import System.FilePath

14import System.FilePath

22import System.Directory

15import System.Directory

@@ -57,7 +50,7 @@ runLatexRaw latexProgram file = do

57 takeDirectory file, dropExtension file] >> return ()

50 takeDirectory file, dropExtension file] >> return ()

58 let pdfFile = replaceExtension file "pdf"

51 let pdfFile = replaceExtension file "pdf"

59 let logFile = replaceExtension file "log"

52 let logFile = replaceExtension file "log"

60− txt <- tryJust (guard . isDoesNotExistError) (readFile logFile)

53+ txt <- tryJust (guard . isDoesNotExistError) (UTF8.readFile logFile)

61 let checks = checkLatex $ either (const "") id txt

54 let checks = checkLatex $ either (const "") id txt

62 case checks of

55 case checks of

63 -- err , bib , ref , msg

56 -- err , bib , ref , msg

@@ -122,13 +115,13 @@ runBibtex file = do

122exit :: String -> IO a

115exit :: String -> IO a

123exit x = do

116exit x = do

124 progName <- getProgName

117 progName <- getProgName

125− hPutStrLn stderr $ progName ++ ": " ++ x

118+ UTF8.hPutStrLn stderr $ progName ++ ": " ++ x

126 exitWith $ ExitFailure 1

119 exitWith $ ExitFailure 1

127

120

128saveStdin :: FilePath -> IO (Either String FilePath)

121saveStdin :: FilePath -> IO (Either String FilePath)

129saveStdin file = do

122saveStdin file = do

130− text <- getContents

123+ text <- UTF8.getContents

131− writeFile file text

124+ UTF8.writeFile file text

132 fileExist <- doesFileExist file

125 fileExist <- doesFileExist file

133 case fileExist of

126 case fileExist of

134 False -> return $ Left $! "Could not create " ++ file

127 False -> return $ Left $! "Could not create " ++ file

@@ -137,7 +130,7 @@ saveStdin file = do

137saveOutput :: FilePath -> FilePath -> IO ()

130saveOutput :: FilePath -> FilePath -> IO ()

138saveOutput input output = do

131saveOutput input output = do

139 copyFile input output

132 copyFile input output

140− hPutStrLn stderr $! "Created " ++ output

133+ UTF8.hPutStrLn stderr $! "Created " ++ output

141

134

142main :: IO ()

135main :: IO ()

143main = bracket

136main = bracket

@@ -170,8 +163,8 @@ main = bracket

170 any (\o -> (o ++ "=") `isPrefixOf` x) goodoptslong

163 any (\o -> (o ++ "=") `isPrefixOf` x) goodoptslong

171 unless (all isGoodopt opts) $ do

164 unless (all isGoodopt opts) $ do

172 (code, out, _err) <- readProcessWithExitCode "pandoc" ["--help"] ""

165 (code, out, _err) <- readProcessWithExitCode "pandoc" ["--help"] ""

173− putStrLn "markdown2pdf [OPTIONS] [FILES]\nOptions:"

166+ UTF8.putStrLn "markdown2pdf [OPTIONS] [FILES]\nOptions:"

174− putStr $ unlines $

167+ UTF8.putStr $ unlines $

175 filter (\l -> any (`isInfixOf` l) goodoptslong) $ lines out

168 filter (\l -> any (`isInfixOf` l) goodoptslong) $ lines out

176 exitWith code

169 exitWith code

177

170

src/pandoc.hs+22−29

@@ -45,14 +45,7 @@ import Data.Char ( toLower, isDigit )

45import Data.List ( intercalate, isSuffixOf )

46import System.Directory ( getAppUserDataDirectory )

47import System.IO ( stdout, stderr )

48−-- Note: ghc >= 6.12 (base >=4.2) supports unicode through iconv

48+import qualified Text.Pandoc.UTF8 as UTF8

49−-- So we use System.IO.UTF8 only if we have an earlier version

50−#if MIN_VERSION_base(4,2,0)

51−import System.IO ( hPutStr, hPutStrLn )

52−#else

53−import Prelude hiding ( putStr, putStrLn, writeFile, readFile, getContents )

54−import System.IO.UTF8

55−#endif

56#ifdef _CITEPROC

49#ifdef _CITEPROC

57import Text.CSL

50import Text.CSL

58import Text.Pandoc.Biblio

51import Text.Pandoc.Biblio

@@ -344,7 +337,7 @@ options =

344 "references" -> return ReferenceObfuscation

337 "references" -> return ReferenceObfuscation

345 "javascript" -> return JavascriptObfuscation

338 "javascript" -> return JavascriptObfuscation

346 "none" -> return NoObfuscation

339 "none" -> return NoObfuscation

347− _ -> hPutStrLn stderr ("Error: Unknown obfuscation method: " ++ arg) >>

340+ _ -> UTF8.hPutStrLn stderr ("Error: Unknown obfuscation method: " ++ arg) >>

348 exitWith (ExitFailure 6)

341 exitWith (ExitFailure 6)

349 return opt { optEmailObfuscation = method })

342 return opt { optEmailObfuscation = method })

350 "none|javascript|references")

343 "none|javascript|references")

@@ -378,7 +371,7 @@ options =

378 return opt{ optTransforms =

371 return opt{ optTransforms =

379 headerShift shift : oldTransforms }

372 headerShift shift : oldTransforms }

380 else do

373 else do

381− hPutStrLn stderr $ "base-header-level must be a number >= 1"

374+ UTF8.hPutStrLn stderr $ "base-header-level must be a number >= 1"

382 exitWith $ ExitFailure 19)

375 exitWith $ ExitFailure 19)

383 "LEVEL")

376 "LEVEL")

384 "" -- "Headers base level"

377 "" -- "Headers base level"

@@ -386,7 +379,7 @@ options =

386 , Option "" ["template"]

379 , Option "" ["template"]

387 (ReqArg

380 (ReqArg

388 (\arg opt -> do

381 (\arg opt -> do

389− text <- readFile arg

382+ text <- UTF8.readFile arg

390 return opt{ optTemplate = text,

383 return opt{ optTemplate = text,

391 optStandalone = True })

384 optStandalone = True })

392 "FILENAME")

385 "FILENAME")

@@ -400,7 +393,7 @@ options =

400 let newvars = optVariables opt ++ [(k,v)]

393 let newvars = optVariables opt ++ [(k,v)]

401 return opt{ optVariables = newvars }

394 return opt{ optVariables = newvars }

402 _ -> do

395 _ -> do

403− hPutStrLn stderr $ "Could not parse `" ++ arg ++ "' as a key/value pair (k=v or k:v)"

396+ UTF8.hPutStrLn stderr $ "Could not parse `" ++ arg ++ "' as a key/value pair (k=v or k:v)"

404 exitWith $ ExitFailure 17)

397 exitWith $ ExitFailure 17)

405 "FILENAME")

398 "FILENAME")

406 "" -- "Use custom template"

399 "" -- "Use custom template"

@@ -418,7 +411,7 @@ options =

418 , Option "H" ["include-in-header"]

411 , Option "H" ["include-in-header"]

419 (ReqArg

412 (ReqArg

420 (\arg opt -> do

413 (\arg opt -> do

421− text <- readFile arg

414+ text <- UTF8.readFile arg

422 -- add new ones to end, so they're included in order specified

415 -- add new ones to end, so they're included in order specified

423 let newvars = optVariables opt ++ [("header-includes",text)]

416 let newvars = optVariables opt ++ [("header-includes",text)]

424 return opt { optVariables = newvars,

417 return opt { optVariables = newvars,

@@ -429,7 +422,7 @@ options =

429 , Option "B" ["include-before-body"]

422 , Option "B" ["include-before-body"]

430 (ReqArg

423 (ReqArg

431 (\arg opt -> do

424 (\arg opt -> do

432− text <- readFile arg

425+ text <- UTF8.readFile arg

433 -- add new ones to end, so they're included in order specified

426 -- add new ones to end, so they're included in order specified

434 let newvars = optVariables opt ++ [("include-before",text)]

427 let newvars = optVariables opt ++ [("include-before",text)]

435 return opt { optVariables = newvars,

428 return opt { optVariables = newvars,

@@ -440,7 +433,7 @@ options =

440 , Option "A" ["include-after-body"]

433 , Option "A" ["include-after-body"]

441 (ReqArg

434 (ReqArg

442 (\arg opt -> do

435 (\arg opt -> do

443− text <- readFile arg

436+ text <- UTF8.readFile arg

444 -- add new ones to end, so they're included in order specified

437 -- add new ones to end, so they're included in order specified

445 let newvars = optVariables opt ++ [("include-after",text)]

438 let newvars = optVariables opt ++ [("include-after",text)]

446 return opt { optVariables = newvars,

439 return opt { optVariables = newvars,

@@ -451,7 +444,7 @@ options =

451 , Option "C" ["custom-header"]

444 , Option "C" ["custom-header"]

452 (ReqArg

445 (ReqArg

453 (\arg opt -> do

446 (\arg opt -> do

454− text <- readFile arg

447+ text <- UTF8.readFile arg

455 let newVars = ("legacy-header", text) : optVariables opt

448 let newVars = ("legacy-header", text) : optVariables opt

456 return opt { optVariables = newVars

449 return opt { optVariables = newVars

457 , optStandalone = True })

450 , optStandalone = True })

@@ -479,7 +472,7 @@ options =

479 (\arg _ -> do

472 (\arg _ -> do

480 templ <- getDefaultTemplate Nothing arg

473 templ <- getDefaultTemplate Nothing arg

481 case templ of

474 case templ of

482− Right t -> hPutStr stdout t

475+ Right t -> UTF8.hPutStr stdout t

483 Left e -> error $ show e

476 Left e -> error $ show e

484 exitWith ExitSuccess)

477 exitWith ExitSuccess)

485 "FORMAT")

478 "FORMAT")

@@ -521,7 +514,7 @@ options =

521 (NoArg

514 (NoArg

522 (\_ -> do

515 (\_ -> do

523 prg <- getProgName

516 prg <- getProgName

524− hPutStrLn stdout (prg ++ " " ++ pandocVersion ++ compileInfo ++

517+ UTF8.hPutStrLn stdout (prg ++ " " ++ pandocVersion ++ compileInfo ++

525 copyrightMessage)

518 copyrightMessage)

526 exitWith ExitSuccess ))

519 exitWith ExitSuccess ))

527 "" -- "Print version"

520 "" -- "Print version"

@@ -530,7 +523,7 @@ options =

530 (NoArg

523 (NoArg

531 (\_ -> do

524 (\_ -> do

532 prg <- getProgName

525 prg <- getProgName

533− hPutStr stdout (usageMessage prg options)

526+ UTF8.hPutStr stdout (usageMessage prg options)

534 exitWith ExitSuccess ))

527 exitWith ExitSuccess ))

535 "" -- "Show help"

528 "" -- "Show help"

536 ]

529 ]

@@ -603,8 +596,8 @@ main = do

603

596

604 unless (null errors) $

597 unless (null errors) $

605 do name <- getProgName

598 do name <- getProgName

606− mapM_ (\e -> hPutStr stderr (name ++ ": ") >> hPutStr stderr e) errors

599+ mapM_ (\e -> UTF8.hPutStr stderr (name ++ ": ") >> UTF8.hPutStr stderr e) errors

607− hPutStrLn stderr $ "Try " ++ name ++ " --help for more information."

600+ UTF8.hPutStrLn stderr $ "Try " ++ name ++ " --help for more information."

608 exitWith $ ExitFailure 2

601 exitWith $ ExitFailure 2

609

602

610 let defaultOpts' = if compatMode

603 let defaultOpts' = if compatMode

@@ -653,13 +646,13 @@ main = do

653 } = opts

646 } = opts

654

647

655 when dumpArgs $

648 when dumpArgs $

656− do hPutStrLn stdout outputFile

649+ do UTF8.hPutStrLn stdout outputFile

657− mapM_ (\arg -> hPutStrLn stdout arg) args

650+ mapM_ (\arg -> UTF8.hPutStrLn stdout arg) args

658 exitWith ExitSuccess

651 exitWith ExitSuccess

659

652

660 -- warn about deprecated options

653 -- warn about deprecated options

661 case lookup "legacy-header" variables of

654 case lookup "legacy-header" variables of

662− Just _ -> hPutStrLn stderr $

655+ Just _ -> UTF8.hPutStrLn stderr $

663 "Warning: The -C/--custom-header is deprecated.\n" ++

656 "Warning: The -C/--custom-header is deprecated.\n" ++

664 "Please transition to using --template instead."

657 "Please transition to using --template instead."

665 Nothing -> return ()

658 Nothing -> return ()

@@ -765,7 +758,7 @@ main = do

765 writerIdentifierPrefix = idPrefix }

758 writerIdentifierPrefix = idPrefix }

766

759

767 when (isNonTextOutput writerName' && outputFile == "-") $

760 when (isNonTextOutput writerName' && outputFile == "-") $

768− do hPutStrLn stderr ("Error: Cannot write " ++ writerName ++ " output to stdout.\n" ++

761+ do UTF8.hPutStrLn stderr ("Error: Cannot write " ++ writerName ++ " output to stdout.\n" ++

769 "Specify an output file using the -o option.")

762 "Specify an output file using the -o option.")

770 exitWith $ ExitFailure 5

763 exitWith $ ExitFailure 5

771

764

@@ -775,10 +768,10 @@ main = do

775

768

776 let readSources [] = mapM readSource ["-"]

769 let readSources [] = mapM readSource ["-"]

777 readSources srcs = mapM readSource srcs

770 readSources srcs = mapM readSource srcs

778− readSource "-" = getContents

771+ readSource "-" = UTF8.getContents

779 readSource src = case parseURI src of

772 readSource src = case parseURI src of

780 Just u -> readURI u

773 Just u -> readURI u

781− Nothing -> readFile src

774+ Nothing -> UTF8.readFile src

782 readURI uri = simpleHTTP (mkRequest GET uri) >>= getResponseBody >>=

775 readURI uri = simpleHTTP (mkRequest GET uri) >>= getResponseBody >>=

783 return . toString -- treat all as UTF8

776 return . toString -- treat all as UTF8

784

777

@@ -800,5 +793,5 @@ main = do

800 case writerName' of

793 case writerName' of

801 "odt" -> saveOpenDocumentAsODT datadir outputFile sourceDirRelative referenceODT writerOutput

794 "odt" -> saveOpenDocumentAsODT datadir outputFile sourceDirRelative referenceODT writerOutput

802 _ -> if outputFile == "-"

795 _ -> if outputFile == "-"

803− then putStr writerOutput

796+ then UTF8.putStr writerOutput

804− else writeFile outputFile writerOutput

797+ else UTF8.writeFile outputFile writerOutput