47a5f04 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

Parsing: Simplified dash and ellipsis.

47a5f04John MacFarlane12 years ago

src/Text/Pandoc/Parsing.hs+13−40

compare with master

unifiedsplit

@@ -169,7 +169,7 @@ import Text.Pandoc.XML (fromEntities)

169import qualified Text.Pandoc.UTF8 as UTF8 (putStrLn)

170import Text.Parsec

171import Text.Parsec.Pos (newPos)

172−import Data.Char ( toLower, toUpper, ord, chr, isAscii, isAlphaNum, isDigit,

172+import Data.Char ( toLower, toUpper, ord, chr, isAscii, isAlphaNum,

173 isHexDigit, isSpace )

174import Data.List ( intercalate, transpose )

175import Text.Pandoc.Shared

@@ -1124,48 +1124,21 @@ doubleQuoteEnd = void (charOrRef "\"\8221\148")

1124

1125ellipses :: Stream s m Char

1126 => ParserT s st m Inlines

1127−ellipses = do

1128− try (charOrRef "\8230\133") <|> try (string "..." >> return '…')

1127+ellipses = try (string "..." >> return (B.str "\8230"))

1129− return (B.str "\8230")

1130

1128

1129+dash :: (HasReaderOptions st, Stream s m Char)

1131−dash :: Stream s m Char => ParserT s ParserState m Inlines

1130+ => ParserT s st m Inlines

1132−dash = do

1131+dash = try $ do

1133 oldDashes <- getOption readerOldDashes

1132 oldDashes <- getOption readerOldDashes

1134 if oldDashes

1133 if oldDashes

1135− then emDashOld <|> enDashOld

1134+ then do

1136− else B.str <$> (hyphenDash <|> emDash <|> enDash)

1135+ char '-'

1137−

1138−-- Two hyphens = en-dash, three = em-dash

1139−hyphenDash :: Stream s m Char

1140− => ParserT s st m String

1141−hyphenDash = do

1142− try $ string "--"

1143− option "\8211" (char '-' >> return "\8212")

1136+ (char '-' >> return (B.str "\8212"))

1144−

1145−emDash :: Stream s m Char

1146− => ParserT s st m String

1147−emDash = do

1148− try (charOrRef "\8212\151")

1149− return "\8212"

1150−

1151−enDash :: Stream s m Char

1152− => ParserT s st m String

1153−enDash = do

1154− try (charOrRef "\8212\151")

1155− return "\8211"

1156−

1157−enDashOld :: Stream s m Char

1158− => ParserT s st m Inlines

1159−enDashOld = do

1160− try (charOrRef "\8211\150") <|>

1161− try (char '-' >> lookAhead (satisfy isDigit) >> return '–')

1162− return (B.str "\8211")

1137+ <|> (lookAhead digit >> return (B.str "\8211"))

1163−

1138+ else do

1164−emDashOld :: Stream s m Char

1139+ string "--"

1165− => ParserT s st m Inlines

1140+ (char '-' >> return (B.str "\8212"))

1166−emDashOld = do

1167− try (charOrRef "\8212\151") <|> (try $ string "--" >> optional (char '-') >> return '-')

1168− return (B.str "\8212")

1141+ <|> return (B.str "\8211")

1169

1142

1170-- This is used to prevent exponential blowups for things like:

1143-- This is used to prevent exponential blowups for things like:

1171-- a**a*a**a*a**a*a**a*a**a*a**a*a**

1144-- a**a*a**a*a**a*a**a*a**a*a**a*a**