12045d8 · typeable/pandoc

git.iko.soy/typeable/pandoc

Revert "More intelligent handling of text encodings."

This reverts commit 7272735b3d413a644fd9ab01eeae8ae9cd5a925b.

commit12045d84b6f20c9d959f61d376bd89e722f7f50c authorJohn MacFarlane <fiddlosopher@gmail.com> dateSun 23 Sep 2012, 22:53 (14 years ago) parent7272735

5 files changed+20−30

compare with master

unifiedsplit

src/Text/Pandoc/Parsing.hs+2−1 src/Text/Pandoc/Shared.hs+3−3 src/Text/Pandoc/UTF8.hs+4−16 src/Text/Pandoc/Writers/Docx.hs+3−2 src/pandoc.hs+8−8

src/Text/Pandoc/Parsing.hs+2−1

@@ -146,6 +146,7 @@ where

146146import Text.Pandoc.Definition

147147import Text.Pandoc.Options

148148import Text.Pandoc.Builder (Blocks)

149+import qualified Text.Pandoc.UTF8 as UTF8 (putStrLn)

149150import Text.Parsec

150151import Text.Parsec.Pos (newPos)

151152import Data.Char ( toLower, toUpper, ord, isAscii, isAlphaNum, isDigit, isPunctuation )

@@ -707,7 +708,7 @@ readWith parser state input =

707708testStringWith :: (Show a) => Parser [Char] ParserState a

708709 -> String

709710 -> IO ()

710−testStringWith parser str = putStrLn $ show $

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

711712 readWith parser defaultParserState str

712713

713714-- | Parsing options.

src/Text/Pandoc/Shared.hs+3−3

@@ -90,7 +90,7 @@ import Paths_pandoc (getDataFileName)

9090import Text.Pandoc.Pretty (charWidth)

9191import System.Locale (defaultTimeLocale)

9292import Data.Time

93−import System.IO (stderr, hPutStrLn)

93+import System.IO (stderr)

9494import Text.HTML.TagSoup (renderTagsOptions, RenderOptions(..), Tag(..),

9595 renderOptions)

9696

@@ -503,14 +503,14 @@ readDataFile userDir fname = findDataFile userDir fname >>= UTF8.readFile

503503err :: Int -> String -> IO a

504504err exitCode msg = do

505505 name <- getProgName

506− hPutStrLn stderr $ name ++ ": " ++ msg

506+ UTF8.hPutStrLn stderr $ name ++ ": " ++ msg

507507 exitWith $ ExitFailure exitCode

508508 return undefined

509509

510510warn :: String -> IO ()

511511warn msg = do

512512 name <- getProgName

513− hPutStrLn stderr $ name ++ ": " ++ msg

513+ UTF8.hPutStrLn stderr $ name ++ ": " ++ msg

514514

515515--

516516-- Safe read

src/Text/Pandoc/UTF8.hs+4−16

@@ -25,11 +25,7 @@ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA

2525 Stability : alpha

2626 Portability : portable

2727

28−UTF-8 aware string IO functions that will work with GHC 6.12 or 7.

29−The reading functions first attempt to read UTF-8; if an encoding

30−error is encountered, the local encoding is used instead. This

31−should work well in practice because text in other encodings

32−is usually not valid UTF-8.

28+UTF-8 aware string IO functions that will work with GHC 6.10, 6.12, or 7.

3329-}

3430module Text.Pandoc.UTF8 ( readFile

3531 , writeFile

@@ -49,11 +45,10 @@ where

4945#else

5046import Codec.Binary.UTF8.String (encodeString, decodeString)

5147#endif

52−import Control.Exception (catch, throwIO)

53−import GHC.IO.Exception (IOException(..), IOErrorType(..))

48+

5449import System.IO hiding (readFile, writeFile, getContents,

5550 putStr, putStrLn, hPutStr, hPutStrLn, hGetContents)

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

51+import Prelude hiding (readFile, writeFile, getContents, putStr, putStrLn )

5752import qualified System.IO as IO

5853

5954readFile :: FilePath -> IO String

@@ -80,14 +75,7 @@ hPutStrLn :: Handle -> String -> IO ()

8075hPutStrLn h s = hSetEncoding h utf8 >> IO.hPutStrLn h s

8176

8277hGetContents :: Handle -> IO String

83−hGetContents h = do

84− hSetEncoding h utf8_bom

85− catch (IO.hGetContents h) $ \e ->

86− case ioe_type e of

87− InvalidArgument -> do

88− hSetEncoding h localeEncoding

89− IO.hGetContents h

90− _ -> throwIO e

78+hGetContents h = hSetEncoding h utf8_bom >> IO.hGetContents h

9179

9280encodePath :: FilePath -> FilePath

9381decodeArg :: String -> String

src/Text/Pandoc/Writers/Docx.hs+3−2

@@ -33,7 +33,8 @@ import System.FilePath ( (</>) )

3333import qualified Data.ByteString.Lazy as B

3434import qualified Data.Map as M

3535import Data.ByteString.Lazy.UTF8 ( fromString, toString )

36−import System.IO ( stderr, hPutStrLn )

36+import Text.Pandoc.UTF8 as UTF8

37+import System.IO ( stderr )

3738import Codec.Archive.Zip

3839import Data.Time.Clock.POSIX

3940import Paths_pandoc ( getDataFileName )

@@ -662,7 +663,7 @@ inlineToOpenXML opts (Image alt (src, tit)) = do

662663 , mknode "wp:docPr" [("descr",tit),("id","1"),("name","Picture")] ()

663664 , graphic ] ]

664665 else do

665− liftIO $ hPutStrLn stderr $

666+ liftIO $ UTF8.hPutStrLn stderr $

666667 "Could not find image `" ++ src ++ "', skipping..."

667668 inlinesToOpenXML opts alt

668669

src/pandoc.hs+8−8

@@ -46,7 +46,7 @@ import System.Console.GetOpt

4646import Data.Char ( toLower )

4747import Data.List ( intercalate, isPrefixOf )

4848import System.Directory ( getAppUserDataDirectory, doesFileExist, findExecutable )

49−import System.IO ( stdout, hPutStr, hPutStrLn )

49+import System.IO ( stdout )

5050import System.IO.Error ( isDoesNotExistError )

5151import qualified Control.Exception as E

5252import Control.Exception.Extensible ( throwIO )

@@ -312,7 +312,7 @@ options =

312312 (\arg _ -> do

313313 templ <- getDefaultTemplate Nothing arg

314314 case templ of

315− Right t -> hPutStr stdout t

315+ Right t -> UTF8.hPutStr stdout t

316316 Left e -> error $ show e

317317 exitWith ExitSuccess)

318318 "FORMAT")

@@ -663,7 +663,7 @@ options =

663663 (NoArg

664664 (\_ -> do

665665 prg <- getProgName

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

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

667667 copyrightMessage)

668668 exitWith ExitSuccess ))

669669 "" -- "Print version"

@@ -672,7 +672,7 @@ options =

672672 (NoArg

673673 (\_ -> do

674674 prg <- getProgName

675− hPutStr stdout (usageMessage prg options)

675+ UTF8.hPutStr stdout (usageMessage prg options)

676676 exitWith ExitSuccess ))

677677 "" -- "Show help"

678678

@@ -827,8 +827,8 @@ main = do

827827 } = opts

828828

829829 when dumpArgs $

830− do hPutStrLn stdout outputFile

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

830+ do UTF8.hPutStrLn stdout outputFile

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

832832 exitWith ExitSuccess

833833

834834 let sources = if ignoreArgs then [] else args

@@ -1026,8 +1026,8 @@ main = do

10261026 writeBinary = B.writeFile (UTF8.encodePath outputFile)

10271027

10281028 let writerFn :: FilePath -> String -> IO ()

1029− writerFn "-" = putStr

1030− writerFn f = writeFile f

1029+ writerFn "-" = UTF8.putStr

1030+ writerFn f = UTF8.writeFile f

10311031

10321032 case getWriter writerName' of

10331033 Left e -> err 9 e