dd96267 · typeable/pandoc

git.iko.soy/typeable/pandoc

Modified str parser to capture apostrophes in smart mode.

This solves a problem stemming from the fact that a parser doesn't know what came *before* in the input stream. Previously pandoc would parse D'oh l'*aide* as containing a single quoted "oh l", when both `'`s should be apostrophes. (Issue #360.) There are two issues here. (a) It is obvious that the first `'` is not an open quote, becaues of the preceding `D`. This patch solves the problem. (b) It is obvious to us that the second `'` is not an open quote, because we see that *aide* is some text. But getting a good algorithm that has good performance is a bit tricky. You can't assume that `'` followed by `*` is always an apostrophe: *'this is quoted'* This patch does not fix (b).

commitdd96267626e6f52e49c912cfe90380de2df469e5 authorJohn MacFarlane <fiddlosopher@gmail.com> dateMon 26 Dec 2011, 23:04 (15 years ago) parent765a777

1 file changed+9−2

compare with master

unifiedsplit

src/Text/Pandoc/Readers/Markdown.hs+9−2

src/Text/Pandoc/Readers/Markdown.hs+9−2

@@ -44,7 +44,7 @@ import Text.Pandoc.Readers.HTML ( htmlTag, htmlInBalanced, isInlineTag, isBlockT

4444 isTextTag, isCommentTag )

4545import Text.Pandoc.CharacterReferences ( decodeCharacterReferences )

4646import Text.ParserCombinators.Parsec

47−import Control.Monad (when, liftM, guard)

47+import Control.Monad (when, liftM, guard, mzero)

4848import Text.HTML.TagSoup

4949import Text.HTML.TagSoup.Match (tagOpen)

5050

@@ -1087,8 +1087,15 @@ nonEndline = satisfy (/='\n')

10871087

10881088str :: GenParser Char ParserState Inline

10891089str = do

1090+ st <- getState

10901091 a <- alphaNum

1091− as <- many $ alphaNum <|> (try $ char '_' >>~ lookAhead alphaNum)

1092+ as <- many $ alphaNum

1093+ <|> (try $ char '_' >>~ lookAhead alphaNum)

1094+ <|> if stateStrict st

1095+ then mzero

1096+ else (try $ char '\'' >> lookAhead alphaNum >> return '\x2019')

1097+ -- for things like l'aide - would be better to return

1098+ -- an Apostrophe, but we can't in this context

10921099 let result = a:as

10931100 state <- getState

10941101 let spacesToNbr = map (\c -> if c == ' ' then '\160' else c)