47a5f04 · typeable/pandoc

git.iko.soy/typeable/pandoc

Parsing: Simplified dash and ellipsis.

This originated with @dubiousjim's observation in #1419 that there was a typo in the definition of enDash. It returned an em dash character instead of an en dash. I thought about why this had not been noticed before, and realized that en dashes were just being parsed as regular symbols. That made me realize that, now that we no longer have dedicate EnDash, EmDash, and Ellipses inline elements, as we used to in pandoc, we no longer need to parse the unicode characters specially. This allowed a considerable simplification of the code. Partially resolves #1419.

commit47a5f04761d9921ee273e8264e26409c806a912d authorJohn MacFarlane <jgm@berkeley.edu> dateSat 12 Jul 2014, 22:59 (12 years ago) parent4676bfd

1 file changed+13−40

compare with master

unifiedsplit

src/Text/Pandoc/Parsing.hs+13−40

src/Text/Pandoc/Parsing.hs+13−40

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

169169import qualified Text.Pandoc.UTF8 as UTF8 (putStrLn)

170170import Text.Parsec

171171import 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,

173173 isHexDigit, isSpace )

174174import Data.List ( intercalate, transpose )

175175import Text.Pandoc.Shared

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

11241124

11251125ellipses :: Stream s m Char

11261126 => ParserT s st m Inlines

1127−ellipses = do

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

1129− return (B.str "\8230")

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

11301128

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

1132−dash = do

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

1130+ => ParserT s st m Inlines

1131+dash = try $ do

11331132 oldDashes <- getOption readerOldDashes

11341133 if oldDashes

1135− then emDashOld <|> enDashOld

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

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")

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")

1163−

1164−emDashOld :: Stream s m Char

1165− => ParserT s st m Inlines

1166−emDashOld = do

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

1168− return (B.str "\8212")

1134+ then do

1135+ char '-'

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

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

1138+ else do

1139+ string "--"

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

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

11691142

11701143-- This is used to prevent exponential blowups for things like:

11711144-- a**a*a**a*a**a*a**a*a**a*a**a*a**