8cc5625 · typeable/pandoc

git.iko.soy/typeable/pandoc

MediaWiki reader: Use MWState instead of ParserState.

commit8cc5625bd6ef2bda13b156b381d238f21beb789e authorJohn MacFarlane <jgm@berkeley.edu> dateThu 13 Sep 2012, 15:10 (14 years ago) parent5b29f0f

2 files changed+25−6

compare with master

unifiedsplit

src/Text/Pandoc/Readers/HTML.hs+1−1 src/Text/Pandoc/Readers/MediaWiki.hs+24−5

src/Text/Pandoc/Readers/HTML.hs+1−1

@@ -597,7 +597,7 @@ htmlInBalanced f = try $ do

597597 return $ tag ++ concat contents ++ endtag

598598

599599-- | Matches a tag meeting a certain condition.

600−htmlTag :: (Tag String -> Bool) -> Parser [Char] ParserState (Tag String, String)

600+htmlTag :: (Tag String -> Bool) -> Parser [Char] st (Tag String, String)

601601htmlTag f = try $ do

602602 lookAhead (char '<')

603603 (next : _) <- getInput >>= return . canonicalizeTags . parseTags

src/Text/Pandoc/Readers/MediaWiki.hs+24−5

@@ -34,6 +34,7 @@ _ support internal links http://www.mediawiki.org/wiki/Help:Links

3434_ support external links (partially implemented)

3535_ support images http://www.mediawiki.org/wiki/Help:Images

3636_ support tables http://www.mediawiki.org/wiki/Help:Tables

37+- footnotes?

3738-}

3839module Text.Pandoc.Readers.MediaWiki ( readMediaWiki ) where

3940

@@ -44,7 +45,7 @@ import Text.Pandoc.Options

4445import Text.Pandoc.Readers.HTML ( htmlTag, isInlineTag,

4546 isBlockTag, isCommentTag )

4647import Text.Pandoc.XML ( fromEntities )

47−import Text.Pandoc.Parsing

48+import Text.Pandoc.Parsing hiding ( nested )

4849import Text.Pandoc.Generic ( bottomUp )

4950import Text.Pandoc.Shared ( stripTrailingNewlines, safeRead )

5051import Data.Monoid (mconcat, mempty)

@@ -56,17 +57,35 @@ import Data.Sequence (viewl, ViewL(..), (<|))

5657

5758-- | Read mediawiki from an input string and return a Pandoc document.

5859readMediaWiki :: ReaderOptions -- ^ Reader options

59− -> String -- ^ String to parse (assuming @'\n'@ line endings)

60− -> Pandoc

60+ -> String -- ^ String to parse (assuming @'\n'@ line endings)

61+ -> Pandoc

6162readMediaWiki opts s =

62− (readWith parseMediaWiki) def{ stateOptions = opts } (s ++ "\n\n")

63+ case runParser parseMediaWiki MWState{ mwOptions = opts, mwMaxNestingLevel = 4 }

64+ "source" (s ++ "\n") of

65+ Left err' -> error $ "\nError:\n" ++ show err'

66+ Right result -> result

6367

64−type MWParser = Parser [Char] ParserState

68+data MWState = MWState { mwOptions :: ReaderOptions

69+ , mwMaxNestingLevel :: Int

70+ }

71+

72+type MWParser = Parser [Char] MWState

6573

6674--

6775-- auxiliary functions

6876--

6977

78+-- This is used to prevent exponential blowups for things like:

79+-- ''a'''a''a'''a''a'''a''a'''a

80+nested :: MWParser a -> MWParser a

81+nested p = do

82+ nestlevel <- mwMaxNestingLevel `fmap` getState

83+ guard $ nestlevel > 0

84+ updateState $ \st -> st{ mwMaxNestingLevel = mwMaxNestingLevel st - 1 }

85+ res <- p

86+ updateState $ \st -> st{ mwMaxNestingLevel = nestlevel }

87+ return res

88+

7089specialChars :: [Char]

7190specialChars = "'[]<=&*{}"

7291