diff options
Diffstat (limited to 'src/Text/Pandoc/Readers')
| -rw-r--r-- | src/Text/Pandoc/Readers/LaTeX.hs | 32 | ||||
| -rw-r--r-- | src/Text/Pandoc/Readers/Markdown.hs | 15 | ||||
| -rw-r--r-- | src/Text/Pandoc/Readers/MediaWiki.hs | 10 | ||||
| -rw-r--r-- | src/Text/Pandoc/Readers/Org.hs | 58 |
4 files changed, 83 insertions, 32 deletions
diff --git a/src/Text/Pandoc/Readers/LaTeX.hs b/src/Text/Pandoc/Readers/LaTeX.hs index 942b9f3b3..31ac37bf1 100644 --- a/src/Text/Pandoc/Readers/LaTeX.hs +++ b/src/Text/Pandoc/Readers/LaTeX.hs @@ -271,7 +271,9 @@ ignoreBlocks name = (name, doraw <|> (mempty <$ optargs)) blockCommands :: M.Map String (LP Blocks) blockCommands = M.fromList $ [ ("par", mempty <$ skipopts) - , ("title", mempty <$ (skipopts *> tok >>= addMeta "title")) + , ("title", mempty <$ (skipopts *> + (grouped inline >>= addMeta "title") + <|> (grouped block >>= addMeta "title"))) , ("subtitle", mempty <$ (skipopts *> tok >>= addMeta "subtitle")) , ("author", mempty <$ (skipopts *> authors)) -- -- in letter class, temp. store address & sig as title, author @@ -341,7 +343,7 @@ setCaption :: LP Blocks setCaption = do ils <- tok mblabel <- option Nothing $ - try $ spaces >> controlSeq "label" >> (Just <$> tok) + try $ spaces' >> controlSeq "label" >> (Just <$> tok) let ils' = case mblabel of Just lab -> ils <> spanWith ("",[],[("data-label", stringify lab)]) mempty @@ -369,7 +371,7 @@ section (ident, classes, kvs) lvl = do let lvl' = if hasChapters then lvl + 1 else lvl skipopts contents <- grouped inline - lab <- option ident $ try (spaces >> controlSeq "label" >> spaces >> braced) + lab <- option ident $ try (spaces' >> controlSeq "label" >> spaces' >> braced) attr' <- registerHeader (lab, classes, kvs) contents return $ headerWith attr' lvl' contents @@ -495,7 +497,7 @@ inlineCommands = M.fromList $ , ("v", option (str "v") $ try $ tok >>= accent hacek) , ("u", option (str "u") $ try $ tok >>= accent breve) , ("i", lit "i") - , ("\\", linebreak <$ (optional (bracketed inline) *> optional sp)) + , ("\\", linebreak <$ (optional (bracketed inline) *> spaces')) , (",", pure mempty) , ("@", pure mempty) , (" ", lit "\160") @@ -508,7 +510,7 @@ inlineCommands = M.fromList $ , ("thanks", (note . mconcat) <$> (char '{' *> manyTill block (char '}'))) , ("footnote", (note . mconcat) <$> (char '{' *> manyTill block (char '}'))) , ("verb", doverb) - , ("lstinline", doverb) + , ("lstinline", skipopts *> doverb) , ("Verb", doverb) , ("texttt", (code . stringify . toList) <$> tok) , ("url", (unescapeURL <$> braced) >>= \url -> @@ -1275,7 +1277,7 @@ complexNatbibCitation mode = try $ do parseAligns :: LP [Alignment] parseAligns = try $ do char '{' - let maybeBar = skipMany $ sp <|> () <$ char '|' <|> () <$ try (string "@{}") + let maybeBar = skipMany $ sp <|> () <$ char '|' <|> () <$ (char '@' >> braced) maybeBar let cAlign = AlignCenter <$ char 'c' let lAlign = AlignLeft <$ char 'l' @@ -1289,13 +1291,13 @@ parseAligns = try $ do return aligns' hline :: LP () -hline = () <$ (try $ spaces >> controlSeq "hline") +hline = () <$ (try $ spaces' *> controlSeq "hline" <* spaces') lbreak :: LP () -lbreak = () <$ (try $ spaces *> controlSeq "\\") +lbreak = () <$ (try $ spaces' *> controlSeq "\\" <* spaces') amp :: LP () -amp = () <$ (try $ spaces *> char '&') +amp = () <$ (try $ spaces' *> char '&') parseTableRow :: Int -- ^ number of columns -> LP [Blocks] @@ -1308,20 +1310,22 @@ parseTableRow cols = try $ do guard $ cells' /= [mempty] -- note: a & b in a three-column table leaves an empty 3rd cell: let cells'' = cells' ++ replicate (cols - numcells) mempty - spaces + spaces' return cells'' +spaces' :: LP () +spaces' = spaces *> skipMany (comment *> spaces) + simpTable :: Bool -> LP Blocks simpTable hasWidthParameter = try $ do - when hasWidthParameter $ () <$ (spaces >> tok) - spaces + when hasWidthParameter $ () <$ (spaces' >> tok) + skipopts aligns <- parseAligns let cols = length aligns optional hline header' <- option [] $ try (parseTableRow cols <* lbreak <* hline) rows <- sepEndBy (parseTableRow cols) (lbreak <* optional hline) - spaces - skipMany (comment *> spaces) + spaces' let header'' = if null header' then replicate cols mempty else header' diff --git a/src/Text/Pandoc/Readers/Markdown.hs b/src/Text/Pandoc/Readers/Markdown.hs index 187b479c3..a36c2acde 100644 --- a/src/Text/Pandoc/Readers/Markdown.hs +++ b/src/Text/Pandoc/Readers/Markdown.hs @@ -448,18 +448,18 @@ block = do res <- choice [ mempty <$ blanklines , codeBlockFenced , yamlMetaBlock - , guardEnabled Ext_latex_macros *> (macro >>= return . return) -- note: bulletList needs to be before header because of -- the possibility of empty list items: - , bulletList , header , lhsCodeBlock - , rawTeXBlock , divHtml , htmlBlock , table - , lineBlock , codeBlockIndented + , guardEnabled Ext_latex_macros *> (macro >>= return . return) + , rawTeXBlock + , lineBlock , blockQuote , hrule , orderedList @@ -1287,11 +1287,9 @@ pipeBreak = try $ do pipeTable :: MarkdownParser ([Alignment], [Double], F [Blocks], F [[Blocks]]) pipeTable = try $ do - (heads,aligns) <- try ( pipeBreak >>= \als -> - return (return $ replicate (length als) mempty, als)) - <|> ( pipeTableRow >>= \row -> pipeBreak >>= \als -> - - return (row, als) ) + (heads,aligns) <- pipeTableRow >>= \row -> + pipeBreak >>= \als -> + return (row, als) lines' <- sequence <$> many1 pipeTableRow let widths = replicate (length aligns) 0.0 return $ (aligns, widths, heads, lines') @@ -1679,6 +1677,7 @@ referenceLink constructor (lab, raw) = do lookAhead (try (spnl >> normalCite >> return (mempty, ""))) <|> try (spnl >> reference) + when (raw' == "") $ guardEnabled Ext_shortcut_reference_links let labIsRef = raw' == "" || raw' == "[]" let key = toKey $ if labIsRef then raw else raw' parsedRaw <- parseFromString (mconcat <$> many inline) raw' diff --git a/src/Text/Pandoc/Readers/MediaWiki.hs b/src/Text/Pandoc/Readers/MediaWiki.hs index e43b8a86c..d1ba35ba0 100644 --- a/src/Text/Pandoc/Readers/MediaWiki.hs +++ b/src/Text/Pandoc/Readers/MediaWiki.hs @@ -593,11 +593,17 @@ imageOption = <|> try (many1 (oneOf "x0123456789") <* string "px") <|> try (oneOfStrings ["link=","alt=","page=","class="] <* many (noneOf "|]")) +collapseUnderscores :: String -> String +collapseUnderscores [] = [] +collapseUnderscores ('_':'_':xs) = collapseUnderscores ('_':xs) +collapseUnderscores (x:xs) = x : collapseUnderscores xs + +addUnderscores :: String -> String +addUnderscores = collapseUnderscores . intercalate "_" . words + internalLink :: MWParser Inlines internalLink = try $ do sym "[[" - let addUnderscores x = let (pref,suff) = break (=='#') x - in pref ++ intercalate "_" (words suff) pagename <- unwords . words <$> many (noneOf "|]") label <- option (B.text pagename) $ char '|' *> ( (mconcat <$> many1 (notFollowedBy (char ']') *> inline)) diff --git a/src/Text/Pandoc/Readers/Org.hs b/src/Text/Pandoc/Readers/Org.hs index f16aed48d..4a523657c 100644 --- a/src/Text/Pandoc/Readers/Org.hs +++ b/src/Text/Pandoc/Readers/Org.hs @@ -1,6 +1,6 @@ {-# LANGUAGE OverloadedStrings #-} {-# LANGUAGE GeneralizedNewtypeDeriving #-} -{-# LANGUAGE FlexibleContexts #-} +{-# LANGUAGE MultiParamTypeClasses, FlexibleContexts, FlexibleInstances #-} {- Copyright (C) 2014 Albert Krewinkel <tarleb@moltkeplatz.de> @@ -49,7 +49,7 @@ import Control.Applicative ( Applicative, pure , (<$>), (<$), (<*>), (<*), (*>) ) import Control.Arrow (first) import Control.Monad (foldM, guard, liftM, liftM2, mplus, mzero, when) -import Control.Monad.Reader (Reader, runReader, ask, asks) +import Control.Monad.Reader (Reader, runReader, ask, asks, local) import Data.Char (isAlphaNum, toLower) import Data.Default import Data.List (intersperse, isPrefixOf, isSuffixOf) @@ -62,9 +62,11 @@ import Network.HTTP (urlEncode) readOrg :: ReaderOptions -- ^ Reader options -> String -- ^ String to parse (assuming @'\n'@ line endings) -> Pandoc -readOrg opts s = readWith parseOrg def{ orgStateOptions = opts } (s ++ "\n\n") +readOrg opts s = flip runReader def $ readWithM parseOrg def{ orgStateOptions = opts } (s ++ "\n\n") -type OrgParser = Parser [Char] OrgParserState +data OrgParserLocal = OrgParserLocal { orgLocalQuoteContext :: QuoteContext } + +type OrgParser = ParserT [Char] OrgParserState (Reader OrgParserLocal) parseOrg :: OrgParser Pandoc parseOrg = do @@ -125,6 +127,9 @@ data OrgParserState = OrgParserState , orgStateNotes' :: OrgNoteTable } +instance Default OrgParserLocal where + def = OrgParserLocal NoQuote + instance HasReaderOptions OrgParserState where extractReaderOptions = orgStateOptions @@ -138,6 +143,10 @@ instance HasLastStrPosition OrgParserState where getLastStrPos = orgStateLastStrPos setLastStrPos pos st = st{ orgStateLastStrPos = Just pos } +instance HasQuoteContext st (Reader OrgParserLocal) where + getQuoteContext = asks orgLocalQuoteContext + withQuoteContext q = local (\s -> s{orgLocalQuoteContext = q}) + instance Default OrgParserState where def = defaultOrgParserState @@ -964,6 +973,7 @@ inline = , subscript , superscript , inlineLaTeX + , smart , symbol ] <* (guard =<< newlinesCountWithinLimits) <?> "inline" @@ -1270,12 +1280,16 @@ displayMath :: OrgParser (F Inlines) displayMath = return . B.displayMath <$> choice [ rawMathBetween "\\[" "\\]" , rawMathBetween "$$" "$$" ] + +updatePositions :: Char + -> OrgParser (Char) +updatePositions c = do + when (c `elem` emphasisPreChars) updateLastPreCharPos + when (c `elem` emphasisForbiddenBorderChars) updateLastForbiddenCharPos + return c + symbol :: OrgParser (F Inlines) symbol = return . B.str . (: "") <$> (oneOf specialChars >>= updatePositions) - where updatePositions c = do - when (c `elem` emphasisPreChars) updateLastPreCharPos - when (c `elem` emphasisForbiddenBorderChars) updateLastForbiddenCharPos - return c emphasisBetween :: Char -> OrgParser (F Inlines) @@ -1486,3 +1500,31 @@ inlineLaTeXCommand = try $ do count len anyChar return cs _ -> mzero + +smart :: OrgParser (F Inlines) +smart = do + getOption readerSmart >>= guard + doubleQuoted <|> singleQuoted <|> + choice (map (return <$>) [orgApostrophe, dash, ellipses]) + where orgApostrophe = + (char '\'' <|> char '\8217') <* updateLastPreCharPos + <* updateLastForbiddenCharPos + *> return (B.str "\x2019") + +singleQuoted :: OrgParser (F Inlines) +singleQuoted = try $ do + singleQuoteStart + withQuoteContext InSingleQuote $ + fmap B.singleQuoted . trimInlinesF . mconcat <$> + many1Till inline singleQuoteEnd + +-- doubleQuoted will handle regular double-quoted sections, as well +-- as dialogues with an open double-quote without a close double-quote +-- in the same paragraph. +doubleQuoted :: OrgParser (F Inlines) +doubleQuoted = try $ do + doubleQuoteStart + contents <- mconcat <$> many (try $ notFollowedBy doubleQuoteEnd >> inline) + (withQuoteContext InDoubleQuote $ (doubleQuoteEnd <* updateLastForbiddenCharPos) >> return + (fmap B.doubleQuoted . trimInlinesF $ contents)) + <|> (return $ return (B.str "\8220") <> contents) |