b94f1e2 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

RST reader: more accurate parsing of references.

b94f1e2John MacFarlane9 years ago

src/Text/Pandoc/Readers/RST.hs+24−36

compare with master

unifiedsplit

@@ -1030,7 +1030,7 @@ noteBlock' marker = try $ do

1030citationMarker :: Monad m => RSTParser m [Char]

1031citationMarker = do

1032 char '['

1033− res <- simpleReferenceName'

1033+ res <- simpleReferenceName

1034 char ']'

1035 return res

1036

@@ -1039,7 +1039,7 @@ noteMarker = do

1039 char '['

1040 res <- many1 digit

1041 <|>

1042− try (char '#' >> liftM ('#':) simpleReferenceName')

1042+ try (char '#' >> liftM ('#':) simpleReferenceName)

1043 <|> count 1 (oneOf "#*")

1044 char ']'

1045 return res

@@ -1048,34 +1048,24 @@ noteMarker = do

1048-- reference key

1049--

1050

1051−quotedReferenceName :: PandocMonad m => RSTParser m Inlines

1051+quotedReferenceName :: PandocMonad m => RSTParser m String

1052quotedReferenceName = try $ do

1053 char '`' >> notFollowedBy (char '`') -- `` means inline code!

1054− trimInlines . mconcat <$> many1Till inline (char '`')

1054+ manyTill anyChar (char '`')

1055−

1056−unquotedReferenceName :: PandocMonad m => RSTParser m Inlines

1057−unquotedReferenceName = try $ do -- `` means inline code!

1058− trimInlines . mconcat <$> many1Till inline (lookAhead $ char ':')

1059

1055

1060-- Simple reference names are single words consisting of alphanumerics

1056-- Simple reference names are single words consisting of alphanumerics

1061-- plus isolated (no two adjacent) internal hyphens, underscores,

1057-- plus isolated (no two adjacent) internal hyphens, underscores,

1062-- periods, colons and plus signs; no whitespace or other characters

1058-- periods, colons and plus signs; no whitespace or other characters

1063-- are allowed.

1059-- are allowed.

1064−simpleReferenceName' :: Monad m => ParserT [Char] st m String

1060+simpleReferenceName :: Monad m => ParserT [Char] st m String

1065−simpleReferenceName' = do

1061+simpleReferenceName = do

1066 x <- alphaNum

1062 x <- alphaNum

1067 xs <- many $ alphaNum

1063 xs <- many $ alphaNum

1068− <|>

1069− try (oneOf "-_:+." <* lookAhead alphaNum)

1064+ <|> try (oneOf "-_:+." <* lookAhead alphaNum)

1070 return (x:xs)

1065 return (x:xs)

1071

1066

1072−simpleReferenceName :: Monad m => ParserT [Char] st m Inlines

1073−simpleReferenceName = B.str <$> simpleReferenceName'

1074−

1075−referenceName :: PandocMonad m => RSTParser m Inlines

1067+referenceName :: PandocMonad m => RSTParser m String

1076−referenceName = quotedReferenceName <|>

1068+referenceName = quotedReferenceName <|> simpleReferenceName

1077− try (simpleReferenceName <* lookAhead (char ':')) <|>

1078− unquotedReferenceName

1079

1069

1080referenceKey :: PandocMonad m => RSTParser m [Char]

1070referenceKey :: PandocMonad m => RSTParser m [Char]

1081referenceKey = do

1071referenceKey = do

@@ -1123,16 +1113,17 @@ anonymousKey = try $ do

1123 updateState $ \s -> s { stateKeys = M.insert key ((src,""), nullAttr) $

1113 updateState $ \s -> s { stateKeys = M.insert key ((src,""), nullAttr) $

1124 stateKeys s }

1114 stateKeys s }

1125

1115

1126−stripTicks :: String -> String

1127−stripTicks = reverse . stripTick . reverse . stripTick

1128− where stripTick ('`':xs) = xs

1129− stripTick xs = xs

1130−

1131referenceNames :: PandocMonad m => RSTParser m [String]

1116referenceNames :: PandocMonad m => RSTParser m [String]

1132referenceNames = do

1117referenceNames = do

1133 let rn = try $ do

1118 let rn = try $ do

1134 string ".. _"

1119 string ".. _"

1135− (_, ref) <- withRaw referenceName

1120+ ref <- quotedReferenceName

1121+ <|> many ( noneOf ":\n"

1122+ <|> try (char '\n' <*

1123+ string " " <*

1124+ notFollowedBy blankline)

1125+ <|> try (char ':' <* lookAhead alphaNum)

1126+ )

1136 char ':'

1127 char ':'

1137 return ref

1128 return ref

1138 first <- rn

1129 first <- rn

@@ -1147,16 +1138,15 @@ regularKey = try $ do

1147 refs <- referenceNames

1138 refs <- referenceNames

1148 src <- targetURI

1139 src <- targetURI

1149 guard $ not (null src)

1140 guard $ not (null src)

1150− let keys = map (toKey . stripTicks) refs

1141+ let keys = map toKey refs

1151 forM_ keys $ \key ->

1142 forM_ keys $ \key ->

1152 updateState $ \s -> s { stateKeys = M.insert key ((src,""), nullAttr) $

1143 updateState $ \s -> s { stateKeys = M.insert key ((src,""), nullAttr) $

1153 stateKeys s }

1144 stateKeys s }

1154

1145

1155anchorDef :: PandocMonad m => RSTParser m [Char]

1146anchorDef :: PandocMonad m => RSTParser m [Char]

1156anchorDef = try $ do

1147anchorDef = try $ do

1157− (refs, raw) <- withRaw (try (referenceNames <* blanklines))

1148+ (refs, raw) <- withRaw $ try (referenceNames <* blanklines)

1158− let keys = map stripTicks refs

1159− forM_ keys $ \rawkey ->

1149+ forM_ refs $ \rawkey ->

1160 updateState $ \s -> s { stateKeys =

1150 updateState $ \s -> s { stateKeys =

1161 M.insert (toKey rawkey) (('#':rawkey,""), nullAttr) $ stateKeys s }

1151 M.insert (toKey rawkey) (('#':rawkey,""), nullAttr) $ stateKeys s }

1162 -- keep this for 2nd round of parsing, where we'll add the divs (anchor)

1152 -- keep this for 2nd round of parsing, where we'll add the divs (anchor)

@@ -1479,22 +1469,20 @@ explicitLink = try $ do

1479 _ -> return ((src, ""), nullAttr)

1469 _ -> return ((src, ""), nullAttr)

1480 return $ B.linkWith attr (escapeURI src') tit label''

1470 return $ B.linkWith attr (escapeURI src') tit label''

1481

1471

1482−citationName :: PandocMonad m => RSTParser m Inlines

1472+citationName :: PandocMonad m => RSTParser m String

1483citationName = do

1473citationName = do

1484 raw <- citationMarker

1474 raw <- citationMarker

1485− return $ B.str $ "[" ++ raw ++ "]"

1475+ return $ "[" ++ raw ++ "]"

1486

1476

1487referenceLink :: PandocMonad m => RSTParser m Inlines

1477referenceLink :: PandocMonad m => RSTParser m Inlines

1488referenceLink = try $ do

1478referenceLink = try $ do

1489− (label',ref) <- withRaw (quotedReferenceName

1490− <|> simpleReferenceName

1491− <|> citationName) <*

1479+ ref <- (referenceName <|> citationName) <* char '_'

1492− char '_'

1480+ let label' = B.text ref

1493 let isAnonKey (Key ('_':_)) = True

1481 let isAnonKey (Key ('_':_)) = True

1494 isAnonKey _ = False

1482 isAnonKey _ = False

1495 state <- getState

1483 state <- getState

1496 let keyTable = stateKeys state

1484 let keyTable = stateKeys state

1497− key <- option (toKey $ stripTicks ref) $

1485+ key <- option (toKey ref) $

1498 do char '_'

1486 do char '_'

1499 let anonKeys = sort $ filter isAnonKey $ M.keys keyTable

1487 let anonKeys = sort $ filter isAnonKey $ M.keys keyTable

1500 case anonKeys of

1488 case anonKeys of

test/command/4156.mdnew file+10

@@ -0,0 +1,10 @@

1+```

2+% pandoc -f rst

3+.. _`SOMEID`:

5+foo

6+^D

7+<div id="SOMEID">

8+<p>foo</p>

9+</div>

10+```

test/command/rst-links.mdnew file+18

@@ -0,0 +1,18 @@

1+```

2+% pandoc -f rst

3+`*ab*`_

5+.. _`*ab*`: foo

6+^D

7+<p><a href="foo">*ab*</a></p>

8+```

10+```

11+% pandoc -f rst

12+`A B

13+c`_

14+

15+.. _A B C: foo

16+^D

17+<p><a href="foo">A B c</a></p>

18+```