2e1e763 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

Merge pull request #2000 from bosmacs/org-smart

2e1e763John MacFarlane12 years ago

src/Text/Pandoc/Readers/Org.hs+50−8

compare with master

unifiedsplit

@@ -1,6 +1,6 @@

1{-# LANGUAGE OverloadedStrings #-}

2{-# LANGUAGE GeneralizedNewtypeDeriving #-}

3−{-# LANGUAGE FlexibleContexts #-}

3+{-# LANGUAGE MultiParamTypeClasses, FlexibleContexts, FlexibleInstances #-}

4{-

@@ -49,7 +49,7 @@ import Control.Applicative ( Applicative, pure

49 , (<$>), (<$), (<*>), (<*), (*>) )

50import Control.Arrow (first)

51import Control.Monad (foldM, guard, liftM, liftM2, mplus, mzero, when)

52−import Control.Monad.Reader (Reader, runReader, ask, asks)

52+import Control.Monad.Reader (Reader, runReader, ask, asks, local)

53import Data.Char (isAlphaNum, toLower)

54import Data.Default

55import Data.List (intersperse, isPrefixOf, isSuffixOf)

@@ -62,9 +62,11 @@ import Network.HTTP (urlEncode)

62readOrg :: ReaderOptions -- ^ Reader options

63 -> String -- ^ String to parse (assuming @'\n'@ line endings)

64 -> Pandoc

65−readOrg opts s = readWith parseOrg def{ orgStateOptions = opts } (s ++ "\n\n")

65+readOrg opts s = flip runReader def $ readWithM parseOrg def{ orgStateOptions = opts } (s ++ "\n\n")

67+data OrgParserLocal = OrgParserLocal { orgLocalQuoteContext :: QuoteContext }

68+

67−type OrgParser = Parser [Char] OrgParserState

69+type OrgParser = ParserT [Char] OrgParserState (Reader OrgParserLocal)

69parseOrg :: OrgParser Pandoc

71parseOrg :: OrgParser Pandoc

70parseOrg = do

72parseOrg = do

@@ -125,6 +127,9 @@ data OrgParserState = OrgParserState

125 , orgStateNotes' :: OrgNoteTable

127 , orgStateNotes' :: OrgNoteTable

126 }

128 }

127

129

130+instance Default OrgParserLocal where

131+ def = OrgParserLocal NoQuote

132+

128instance HasReaderOptions OrgParserState where

133instance HasReaderOptions OrgParserState where

129 extractReaderOptions = orgStateOptions

134 extractReaderOptions = orgStateOptions

130

135

@@ -138,6 +143,10 @@ instance HasLastStrPosition OrgParserState where

138 getLastStrPos = orgStateLastStrPos

143 getLastStrPos = orgStateLastStrPos

139 setLastStrPos pos st = st{ orgStateLastStrPos = Just pos }

144 setLastStrPos pos st = st{ orgStateLastStrPos = Just pos }

140

145

146+instance HasQuoteContext st (Reader OrgParserLocal) where

147+ getQuoteContext = asks orgLocalQuoteContext

148+ withQuoteContext q = local (\s -> s{orgLocalQuoteContext = q})

149+

141instance Default OrgParserState where

150instance Default OrgParserState where

142 def = defaultOrgParserState

151 def = defaultOrgParserState

143

152

@@ -964,6 +973,7 @@ inline =

964 , subscript

973 , subscript

965 , superscript

974 , superscript

966 , inlineLaTeX

975 , inlineLaTeX

976+ , smart

967 , symbol

977 , symbol

968 ] <* (guard =<< newlinesCountWithinLimits)

978 ] <* (guard =<< newlinesCountWithinLimits)

969 <?> "inline"

979 <?> "inline"

@@ -1270,12 +1280,16 @@ displayMath :: OrgParser (F Inlines)

1270displayMath = return . B.displayMath <$> choice [ rawMathBetween "\\[" "\\]"

1280displayMath = return . B.displayMath <$> choice [ rawMathBetween "\\[" "\\]"

1271 , rawMathBetween "$$" "$$"

1281 , rawMathBetween "$$" "$$"

1272 ]

1282 ]

1283+

1284+updatePositions :: Char

1285+ -> OrgParser (Char)

1286+updatePositions c = do

1287+ when (c `elem` emphasisPreChars) updateLastPreCharPos

1288+ when (c `elem` emphasisForbiddenBorderChars) updateLastForbiddenCharPos

1289+ return c

1290+

1273symbol :: OrgParser (F Inlines)

1291symbol :: OrgParser (F Inlines)

1274symbol = return . B.str . (: "") <$> (oneOf specialChars >>= updatePositions)

1292symbol = return . B.str . (: "") <$> (oneOf specialChars >>= updatePositions)

1275− where updatePositions c = do

1276− when (c `elem` emphasisPreChars) updateLastPreCharPos

1277− when (c `elem` emphasisForbiddenBorderChars) updateLastForbiddenCharPos

1278− return c

1279

1293

1280emphasisBetween :: Char

1294emphasisBetween :: Char

1281 -> OrgParser (F Inlines)

1295 -> OrgParser (F Inlines)

@@ -1486,3 +1500,31 @@ inlineLaTeXCommand = try $ do

1486 count len anyChar

1500 count len anyChar

1487 return cs

1501 return cs

1488 _ -> mzero

1502 _ -> mzero

1503+

1504+smart :: OrgParser (F Inlines)

1505+smart = do

1506+ getOption readerSmart >>= guard

1507+ doubleQuoted <|> singleQuoted <|>

1508+ choice (map (return <$>) [orgApostrophe, dash, ellipses])

1509+ where orgApostrophe =

1510+ (char '\'' <|> char '\8217') <* updateLastPreCharPos

1511+ <* updateLastForbiddenCharPos

1512+ *> return (B.str "\x2019")

1513+

1514+singleQuoted :: OrgParser (F Inlines)

1515+singleQuoted = try $ do

1516+ singleQuoteStart

1517+ withQuoteContext InSingleQuote $

1518+ fmap B.singleQuoted . trimInlinesF . mconcat <$>

1519+ many1Till inline singleQuoteEnd

1520+

1521+-- doubleQuoted will handle regular double-quoted sections, as well

1522+-- as dialogues with an open double-quote without a close double-quote

1523+-- in the same paragraph.

1524+doubleQuoted :: OrgParser (F Inlines)

1525+doubleQuoted = try $ do

1526+ doubleQuoteStart

1527+ contents <- mconcat <$> many (try $ notFollowedBy doubleQuoteEnd >> inline)

1528+ (withQuoteContext InDoubleQuote $ (doubleQuoteEnd <* updateLastForbiddenCharPos) >> return

1529+ (fmap B.doubleQuoted . trimInlinesF $ contents))

1530+ <|> (return $ return (B.str "\8220") <> contents)

tests/Tests/Readers/Org.hs+24

@@ -12,6 +12,9 @@ import Data.Monoid (mempty, mappend, mconcat)

12org :: String -> Pandoc

13org = readOrg def

15+orgSmart :: String -> Pandoc

16+orgSmart = readOrg def { readerSmart = True }

17+

15infix 4 =:

18infix 4 =:

16(=:) :: ToString c

19(=:) :: ToString c

17 => String -> (String, c) -> Test

20 => String -> (String, c) -> Test

@@ -1152,4 +1155,25 @@ tests =

1152 ]

1155 ]

1153 in codeBlockWith ( "", classes, params) "code body\n"

1156 in codeBlockWith ( "", classes, params) "code body\n"

1154 ]

1157 ]

1158+ , testGroup "Smart punctuation"

1159+ [ test orgSmart "quote before ellipses"

1160+ ("'...hi'"

1161+ =?> para (singleQuoted "…hi"))

1162+

1163+ , test orgSmart "apostrophe before emph"

1164+ ("D'oh! A l'/aide/!"

1165+ =?> para ("D’oh! A l’" <> emph "aide" <> "!"))

1166+

1167+ , test orgSmart "apostrophe in French"

1168+ ("À l'arrivée de la guerre, le thème de l'«impossibilité du socialisme»"

1169+ =?> para "À l’arrivée de la guerre, le thème de l’«impossibilité du socialisme»")

1170+

1171+ , test orgSmart "Quotes cannot occur at the end of emphasized text"

1172+ ("/say \"yes\"/" =?>

1173+ para ("/say" <> space <> doubleQuoted "yes" <> "/"))

1174+

1175+ , test orgSmart "Dashes are allowed at the borders of emphasis'"

1176+ ("/foo---/" =?>

1177+ para (emph "foo—"))

1178+ ]

1155 ]

1179 ]