diff options
| author | John MacFarlane <jgm@berkeley.edu> | 2010-12-22 17:05:17 -0800 |
|---|---|---|
| committer | John MacFarlane <jgm@berkeley.edu> | 2010-12-22 17:05:17 -0800 |
| commit | 4bfe140ed1cc7a07d762e4920e7365427d0b9618 (patch) | |
| tree | 02cd044515f5bac5dce17d3a6344e3377ef4c107 /src/Text | |
| parent | 63bf227e0414495a7966edd07d5faf6ca4a2f077 (diff) | |
Made --smart work with HTML reader.
It did not work before, because - and quotes were gobbled
up by the str parser.
Diffstat (limited to 'src/Text')
| -rw-r--r-- | src/Text/Pandoc/Readers/HTML.hs | 17 |
1 files changed, 13 insertions, 4 deletions
diff --git a/src/Text/Pandoc/Readers/HTML.hs b/src/Text/Pandoc/Readers/HTML.hs index ade15349e..1bbb11e62 100644 --- a/src/Text/Pandoc/Readers/HTML.hs +++ b/src/Text/Pandoc/Readers/HTML.hs @@ -587,8 +587,7 @@ plain = many1 inline >>= return . Plain . normalizeSpaces -- inline :: GenParser Char ParserState Inline -inline = choice [ smartPunctuation inline - , str +inline = choice [ str , strong , emph , superscript @@ -600,9 +599,10 @@ inline = choice [ smartPunctuation inline , whitespace , link , image + , smartPunctuation inline , charRef , rawHtmlInline - , char '&' >> return (Str "&") -- common HTML error + , symbol ] <?> "inline" code :: GenParser Char ParserState Inline @@ -620,6 +620,12 @@ rawHtmlInline = do state <- getState if stateParseRaw state then return (HtmlInline result) else return (Str "") +symbol :: GenParser Char ParserState Inline +symbol = do + notFollowedBy (char '<') + c <- oneOf specialChars + return $ Str [c] + betweenTags :: [Char] -> GenParser Char ParserState [Inline] betweenTags tag = try $ htmlOpenTag tag >> inlinesTilEnd tag >>= return . normalizeSpaces @@ -657,7 +663,10 @@ linebreak :: GenParser Char ParserState Inline linebreak = htmlSelfClosingTag "br" >> optional newline >> return LineBreak str :: GenParser Char st Inline -str = many1 (noneOf "< \t\n&") >>= return . Str +str = many1 (noneOf $ specialChars ++ " \t\n") >>= return . Str + +specialChars :: [Char] +specialChars = "<&-\"'.\8216\8217\8220\8221" -- -- links and images |