9857aa8 · typeable/pandoc

git.iko.soy/typeable/pandoc

HTML reader: Fixed detection of self-closing tags.

Earlier versions had a bug and would wrongly think opening tags containing attributes with slashes in them were self-closing. Closes #2146.

commit9857aa866af243d3fe89b5a951a4463e825e6a0c authorJohn MacFarlane <jgm@berkeley.edu> dateMon 11 May 2015, 16:17 (11 years ago) parent4b251e9

2 files changed+3−3

unifiedsplit

src/Text/Pandoc/Parsing.hs+1−1

@@ -1206,7 +1206,7 @@ citeKey = try $ do

12061206 guard =<< notAfterString

12071207 suppress_author <- option False (char '-' *> return True)

12081208 char '@'

1209− firstChar <- alphaNum <|> char '_'

1209+ firstChar <- alphaNum <|> char '_' <|> char '*' -- @* for wildcard in nocite

12101210 let regchar = satisfy (\c -> isAlphaNum c || c == '_')

12111211 let internal p = try $ p <* lookAhead regchar

12121212 rest <- many $ regchar <|> internal (oneOf ":.#$%&-+?<>~/")

src/Text/Pandoc/Readers/HTML.hs+2−2

@@ -51,7 +51,7 @@ import Text.Pandoc.Options (ReaderOptions(readerParseRaw, readerTrace)

5151import Text.Pandoc.Parsing hiding ((<|>))

5252import Text.Pandoc.Walk

5353import Data.Maybe ( fromMaybe, isJust)

54−import Data.List ( intercalate, isInfixOf, isPrefixOf )

54+import Data.List ( intercalate, isInfixOf, isPrefixOf, isSuffixOf )

5555import Data.Char ( isDigit )

5656import Control.Monad ( liftM, guard, when, mzero, void, unless )

5757import Control.Arrow ((***))

@@ -874,7 +874,7 @@ htmlInBalanced :: (Monad m)

874874 -> ParserT String st m String

875875htmlInBalanced f = try $ do

876876 (TagOpen t _, tag) <- htmlTag f

877− guard $ '/' `notElem` tag -- not a self-closing tag

877+ guard $ not $ "/>" `isSuffixOf` tag -- not a self-closing tag

878878 let stopper = htmlTag (~== TagClose t)

879879 let anytag = snd <$> htmlTag (const True)

880880 contents <- many $ notFollowedBy' stopper >>