12a5bd3 · typeable/pandoc

git.iko.soy/typeable/pandoc

Entity handling fixes:

- Text.Pandoc.XML.fromEntities: handle entities without a semicolon. Always lookup character references with the trailing ';', even if it wasn't present. And never add it when looking up numerical entities. (This is what tagsoup seems to require.) - Text.Pandoc.Parsing.characterReference: Always lookup character references with the trailing ';', and leave off the ';' when looking up numerical entities. This fixes a regression for e.g. `&lang;`.

commit12a5bd3c8d34eddbabee0dc54fd7ce6d9539c9d4 authorJohn MacFarlane <jgm@berkeley.edu> dateFri 8 Jan 2016, 17:08 (11 years ago) parent52d95dd

2 files changed+10−3

compare with master

unifiedsplit

src/Text/Pandoc/Parsing.hs+4−1 src/Text/Pandoc/XML.hs+6−2

src/Text/Pandoc/Parsing.hs+4−1

@@ -573,7 +573,10 @@ characterReference :: Stream s m Char => ParserT s st m Char

573573characterReference = try $ do

574574 char '&'

575575 ent <- many1Till nonspaceChar (char ';')

576− case lookupEntity ent of

576+ let ent' = case ent of

577+ '#':_ -> ent

578+ _ -> ent ++ ";"

579+ case lookupEntity ent' of

577580 Just c -> return c

578581 Nothing -> fail "entity not found"

579582

src/Text/Pandoc/XML.hs+6−2

@@ -100,11 +100,15 @@ toEntities (c:cs)

100100-- Unescapes XML entities

101101fromEntities :: String -> String

102102fromEntities ('&':xs) =

103− case lookupEntity ent of

103+ case lookupEntity ent' of

104104 Just c -> c : fromEntities rest

105105 Nothing -> '&' : fromEntities xs

106106 where (ent, rest) = case break (\c -> isSpace c || c == ';') xs of

107107 (zs,';':ys) -> (zs,ys)

108− _ -> ("",xs)

108+ (zs, ys) -> (zs,ys)

109+ ent' = case ent of

110+ '#':_ -> ent

111+ _ -> ent ++ ";"

112+

109113fromEntities (x:xs) = x : fromEntities xs

110114fromEntities [] = []