diff options
author | John MacFarlane <jgm@berkeley.edu> | 2017-05-24 22:41:47 +0200 |
---|---|---|
committer | John MacFarlane <jgm@berkeley.edu> | 2017-05-24 22:41:47 +0200 |
commit | bc6aac7b474495c4433c31bcd4a3570057edb850 (patch) | |
tree | d2656a12fac472f359f38cfc4af50ab18d25b777 /src | |
parent | d4ccd1e001bce84f1bae338cd07d4e8e2e7875cb (diff) | |
download | pandoc-bc6aac7b474495c4433c31bcd4a3570057edb850.tar.gz |
Parsing: Provide parseFromString'.
This is a verison of parseFromString specialied to
ParserState, which resets stateLastStrPos at the end.
This is almost always what we want.
This fixes a bug where `_hi_` wasn't treated as emphasis in
the following, because pandoc got confused about the
position of the last word:
- [o] _hi_
Closes #3690.
Diffstat (limited to 'src')
-rw-r--r-- | src/Text/Pandoc/Parsing.hs | 18 | ||||
-rw-r--r-- | src/Text/Pandoc/Readers/LaTeX.hs | 12 | ||||
-rw-r--r-- | src/Text/Pandoc/Readers/Markdown.hs | 40 | ||||
-rw-r--r-- | src/Text/Pandoc/Readers/RST.hs | 50 | ||||
-rw-r--r-- | src/Text/Pandoc/Readers/TWiki.hs | 8 | ||||
-rw-r--r-- | src/Text/Pandoc/Readers/Textile.hs | 8 | ||||
-rw-r--r-- | src/Text/Pandoc/Readers/Txt2Tags.hs | 6 |
7 files changed, 80 insertions, 62 deletions
diff --git a/src/Text/Pandoc/Parsing.hs b/src/Text/Pandoc/Parsing.hs index c6be48d19..e6157dde3 100644 --- a/src/Text/Pandoc/Parsing.hs +++ b/src/Text/Pandoc/Parsing.hs @@ -50,6 +50,7 @@ module Text.Pandoc.Parsing ( anyLine, enclosed, stringAnyCase, parseFromString, + parseFromString', lineClump, charsInBalanced, romanNumeral, @@ -358,7 +359,10 @@ stringAnyCase (x:xs) = do return (firstChar:rest) -- | Parse contents of 'str' using 'parser' and return result. -parseFromString :: Monad m => ParserT String st m a -> String -> ParserT String st m a +parseFromString :: Monad m + => ParserT String st m a + -> String + -> ParserT String st m a parseFromString parser str = do oldPos <- getPosition oldInput <- getInput @@ -370,6 +374,18 @@ parseFromString parser str = do setPosition oldPos return result +-- | Like 'parseFromString' but specialized for 'ParserState'. +-- This resets 'stateLastStrPos', which is almost always what we want. +parseFromString' :: Monad m + => ParserT String ParserState m a + -> String + -> ParserT String ParserState m a +parseFromString' parser str = do + oldStrPos <- stateLastStrPos <$> getState + res <- parseFromString parser str + updateState $ \st -> st{ stateLastStrPos = oldStrPos } + return res + -- | Parse raw line block up to and including blank lines. lineClump :: Stream [Char] m Char => ParserT [Char] st m String lineClump = blanklines diff --git a/src/Text/Pandoc/Readers/LaTeX.hs b/src/Text/Pandoc/Readers/LaTeX.hs index af7c1d9b7..88be40e3e 100644 --- a/src/Text/Pandoc/Readers/LaTeX.hs +++ b/src/Text/Pandoc/Readers/LaTeX.hs @@ -304,8 +304,8 @@ blockCommand = try $ do rawcommand <- getRawCommand name' transformed <- applyMacros' rawcommand guard $ transformed /= rawcommand - notFollowedBy $ parseFromString inlines transformed - parseFromString blocks transformed + notFollowedBy $ parseFromString' inlines transformed + parseFromString' blocks transformed lookupListDefault raw [name',name] blockCommands inBrackets :: Inlines -> Inlines @@ -475,7 +475,7 @@ inlineCommand = try $ do transformed <- applyMacros' rawcommand exts <- getOption readerExtensions if transformed /= rawcommand - then parseFromString inlines transformed + then parseFromString' inlines transformed else if extensionEnabled Ext_raw_tex exts then return $ rawInline "latex" rawcommand else ignore rawcommand @@ -1021,7 +1021,7 @@ rawEnv name = do (bs, raw) <- withRaw $ env name blocks raw' <- applyMacros' $ beginCommand ++ raw if raw' /= beginCommand ++ raw - then parseFromString blocks raw' + then parseFromString' blocks raw' else if parseRaw then return $ rawBlock "latex" $ beginCommand ++ raw' else do @@ -1119,7 +1119,7 @@ keyvals :: PandocMonad m => LP m [(String, String)] keyvals = try $ char '[' *> manyTill keyval (char ']') alltt :: PandocMonad m => String -> LP m Blocks -alltt t = walk strToCode <$> parseFromString blocks +alltt t = walk strToCode <$> parseFromString' blocks (substitute " " "\\ " $ substitute "%" "\\%" $ intercalate "\\\\\n" $ lines t) where strToCode (Str s) = Code nullAttr s @@ -1503,7 +1503,7 @@ parseTableRow cols prefixes suffixes = try $ do guard $ length rawcells == cols let rawcells' = zipWith3 (\c p s -> p ++ trim c ++ s) rawcells prefixes suffixes - cells' <- mapM (parseFromString tableCell) rawcells' + cells' <- mapM (parseFromString' tableCell) rawcells' let numcells = length cells' guard $ numcells <= cols && numcells >= 1 guard $ cells' /= [mempty] diff --git a/src/Text/Pandoc/Readers/Markdown.hs b/src/Text/Pandoc/Readers/Markdown.hs index af7588562..17a7184c0 100644 --- a/src/Text/Pandoc/Readers/Markdown.hs +++ b/src/Text/Pandoc/Readers/Markdown.hs @@ -155,9 +155,11 @@ litChar = escapedChar' inlinesInBalancedBrackets :: PandocMonad m => MarkdownParser m (F Inlines) inlinesInBalancedBrackets = do char '[' + pos <- getPosition (_, raw) <- withRaw $ charsInBalancedBrackets 1 guard $ not $ null raw - parseFromString (trimInlinesF . mconcat <$> many inline) (init raw) + parseFromString' (setPosition pos >> + trimInlinesF . mconcat <$> many inline) (init raw) charsInBalancedBrackets :: PandocMonad m => Int -> MarkdownParser m () charsInBalancedBrackets 0 = return () @@ -189,7 +191,7 @@ rawTitleBlockLine = do titleLine :: PandocMonad m => MarkdownParser m (F Inlines) titleLine = try $ do raw <- rawTitleBlockLine - res <- parseFromString (many inline) raw + res <- parseFromString' (many inline) raw return $ trimInlinesF $ mconcat res authorsLine :: PandocMonad m => MarkdownParser m (F [Inlines]) @@ -200,12 +202,12 @@ authorsLine = try $ do (trimInlinesF . mconcat <$> many (try $ notFollowedBy sep >> inline)) sep - sequence <$> parseFromString pAuthors raw + sequence <$> parseFromString' pAuthors raw dateLine :: PandocMonad m => MarkdownParser m (F Inlines) dateLine = try $ do raw <- rawTitleBlockLine - res <- parseFromString (many inline) raw + res <- parseFromString' (many inline) raw return $ trimInlinesF $ mconcat res titleBlock :: PandocMonad m => MarkdownParser m () @@ -290,7 +292,7 @@ ignorable t = (T.pack "_") `T.isSuffixOf` t toMetaValue :: PandocMonad m => Text -> MarkdownParser m (F MetaValue) -toMetaValue x = toMeta <$> parseFromString parseBlocks (T.unpack x) +toMetaValue x = toMeta <$> parseFromString' parseBlocks (T.unpack x) where toMeta p = do p' <- p @@ -466,7 +468,7 @@ noteBlock = try $ do rest <- many $ try $ blanklines >> indentSpaces >> rawLines let raw = unlines (first:rest) ++ "\n" optional blanklines - parsed <- parseFromString parseBlocks raw + parsed <- parseFromString' parseBlocks raw let newnote = (ref, parsed) oldnotes <- stateNotes' <$> getState case lookup ref oldnotes of @@ -774,7 +776,7 @@ blockQuote :: PandocMonad m => MarkdownParser m (F Blocks) blockQuote = do raw <- emailBlockQuote -- parse the extracted block, which may contain various block elements: - contents <- parseFromString parseBlocks $ (intercalate "\n" raw) ++ "\n\n" + contents <- parseFromString' parseBlocks $ (intercalate "\n" raw) ++ "\n\n" return $ B.blockQuote <$> contents -- @@ -887,7 +889,7 @@ listItem start = try $ do setState $ state {stateParserContext = ListItemState} -- parse the extracted block, which may contain various block elements: let raw = concat (first:continuations) - contents <- parseFromString parseBlocks raw + contents <- parseFromString' parseBlocks raw updateState (\st -> st {stateParserContext = oldContext}) return contents @@ -934,8 +936,8 @@ definitionListItem :: PandocMonad m => Bool -> MarkdownParser m (F (Inlines, [Bl definitionListItem compact = try $ do rawLine' <- anyLine raw <- many1 $ defRawBlock compact - term <- parseFromString (trimInlinesF . mconcat <$> many inline) rawLine' - contents <- mapM (parseFromString parseBlocks . (++"\n")) raw + term <- parseFromString' (trimInlinesF . mconcat <$> many inline) rawLine' + contents <- mapM (parseFromString' parseBlocks . (++"\n")) raw optional blanklines return $ liftM2 (,) term (sequence contents) @@ -1127,7 +1129,7 @@ lineBlock :: PandocMonad m => MarkdownParser m (F Blocks) lineBlock = try $ do guardEnabled Ext_line_blocks lines' <- lineBlockLines >>= - mapM (parseFromString (trimInlinesF . mconcat <$> many inline)) + mapM (parseFromString' (trimInlinesF . mconcat <$> many inline)) return $ B.lineBlock <$> sequence lines' -- @@ -1170,7 +1172,7 @@ simpleTableHeader headless = try $ do then replicate (length dashes) "" else rawHeads heads <- fmap sequence - $ mapM (parseFromString (mconcat <$> many plain)) + $ mapM (parseFromString' (mconcat <$> many plain)) $ map trim rawHeads' return (heads, aligns, indices) @@ -1216,7 +1218,7 @@ tableLine :: PandocMonad m => [Int] -> MarkdownParser m (F [Blocks]) tableLine indices = rawTableLine indices >>= - fmap sequence . mapM (parseFromString (mconcat <$> many plain)) + fmap sequence . mapM (parseFromString' (mconcat <$> many plain)) -- Parse a multiline table row and return a list of blocks (columns). multilineRow :: PandocMonad m @@ -1225,7 +1227,7 @@ multilineRow :: PandocMonad m multilineRow indices = do colLines <- many1 (rawTableLine indices) let cols = map unlines $ transpose colLines - fmap sequence $ mapM (parseFromString (mconcat <$> many plain)) cols + fmap sequence $ mapM (parseFromString' (mconcat <$> many plain)) cols -- Parses a table caption: inlines beginning with 'Table:' -- and followed by blank lines. @@ -1283,7 +1285,7 @@ multilineTableHeader headless = try $ do then replicate (length dashes) "" else map (unlines . map trim) rawHeadsList heads <- fmap sequence $ - mapM (parseFromString (mconcat <$> many plain)) $ + mapM (parseFromString' (mconcat <$> many plain)) $ map trim rawHeads return (heads, aligns, indices) @@ -1340,7 +1342,7 @@ pipeTableRow = try $ do let chunk = void (code <|> math <|> rawHtmlInline <|> escapedChar <|> rawLaTeXInline') <|> void (noneOf "|\n\r") let cellContents = ((trim . snd) <$> withRaw (many chunk)) >>= - parseFromString pipeTableCell + parseFromString' pipeTableCell cells <- cellContents `sepEndBy1` (char '|') -- surrounding pipes needed for a one-column table: guard $ not (length cells == 1 && not openPipe) @@ -1747,8 +1749,8 @@ referenceLink constructor (lab, raw) = do when (raw' == "") $ guardEnabled Ext_shortcut_reference_links let labIsRef = raw' == "" || raw' == "[]" let key = toKey $ if labIsRef then raw else raw' - parsedRaw <- parseFromString (mconcat <$> many inline) raw' - fallback <- parseFromString (mconcat <$> many inline) $ dropBrackets raw + parsedRaw <- parseFromString' (mconcat <$> many inline) raw' + fallback <- parseFromString' (mconcat <$> many inline) $ dropBrackets raw implicitHeaderRefs <- option False $ True <$ guardEnabled Ext_implicit_header_references let makeFallback = do @@ -1954,7 +1956,7 @@ textualCite = try $ do let (spaces',raw') = span isSpace raw spc | null spaces' = mempty | otherwise = B.space - lab <- parseFromString (mconcat <$> many inline) $ dropBrackets raw' + lab <- parseFromString' (mconcat <$> many inline) $ dropBrackets raw' fallback <- referenceLink B.linkWith (lab,raw') return $ do fallback' <- fallback diff --git a/src/Text/Pandoc/Readers/RST.hs b/src/Text/Pandoc/Readers/RST.hs index e3780f89b..1ea142112 100644 --- a/src/Text/Pandoc/Readers/RST.hs +++ b/src/Text/Pandoc/Readers/RST.hs @@ -196,7 +196,7 @@ parseRST = do parseCitation :: PandocMonad m => (String, String) -> RSTParser m (Inlines, [Blocks]) parseCitation (ref, raw) = do - contents <- parseFromString parseBlocks raw + contents <- parseFromString' parseBlocks raw return $ (B.spanWith (ref, ["citation-label"], []) (B.str ref), [contents]) @@ -246,7 +246,7 @@ fieldListItem :: PandocMonad m => Int -> RSTParser m (Inlines, [Blocks]) fieldListItem minIndent = try $ do (name, raw) <- rawFieldListItem minIndent term <- parseInlineFromString name - contents <- parseFromString parseBlocks raw + contents <- parseFromString' parseBlocks raw optional blanklines return (term, [contents]) @@ -445,7 +445,7 @@ blockQuote :: PandocMonad m => RSTParser m Blocks blockQuote = do raw <- indentedBlock -- parse the extracted block, which may contain various block elements: - contents <- parseFromString parseBlocks $ raw ++ "\n\n" + contents <- parseFromString' parseBlocks $ raw ++ "\n\n" return $ B.blockQuote contents {- @@ -533,7 +533,7 @@ definitionListItem = try $ do term <- trimInlines . mconcat <$> many1Till inline endline raw <- indentedBlock -- parse the extracted block, which may contain various block elements: - contents <- parseFromString parseBlocks $ raw ++ "\n" + contents <- parseFromString' parseBlocks $ raw ++ "\n" return (term, [contents]) definitionList :: PandocMonad m => RSTParser m Blocks @@ -595,7 +595,7 @@ listItem start = try $ do let oldContext = stateParserContext state setState $ state {stateParserContext = ListItemState} -- parse the extracted block, which may itself contain block elements - parsed <- parseFromString parseBlocks $ concat (first:rest) ++ "\n" + parsed <- parseFromString' parseBlocks $ concat (first:rest) ++ "\n" updateState (\st -> st {stateParserContext = oldContext}) return $ case B.toList parsed of [Para xs] -> @@ -686,19 +686,19 @@ directive' = do "line-block" -> lineBlockDirective body' "raw" -> return $ B.rawBlock (trim top) (stripTrailingNewlines body) "role" -> addNewRole top $ map (\(k,v) -> (k, trim v)) fields - "container" -> parseFromString parseBlocks body' + "container" -> parseFromString' parseBlocks body' "replace" -> B.para <$> -- consumed by substKey parseInlineFromString (trim top) "unicode" -> B.para <$> -- consumed by substKey parseInlineFromString (trim $ unicodeTransform top) - "compound" -> parseFromString parseBlocks body' - "pull-quote" -> B.blockQuote <$> parseFromString parseBlocks body' - "epigraph" -> B.blockQuote <$> parseFromString parseBlocks body' - "highlights" -> B.blockQuote <$> parseFromString parseBlocks body' + "compound" -> parseFromString' parseBlocks body' + "pull-quote" -> B.blockQuote <$> parseFromString' parseBlocks body' + "epigraph" -> B.blockQuote <$> parseFromString' parseBlocks body' + "highlights" -> B.blockQuote <$> parseFromString' parseBlocks body' "rubric" -> B.para . B.strong <$> parseInlineFromString top _ | label `elem` ["attention","caution","danger","error","hint", "important","note","tip","warning","admonition"] -> - do bod <- parseFromString parseBlocks $ top ++ "\n\n" ++ body' + do bod <- parseFromString' parseBlocks $ top ++ "\n\n" ++ body' let lab = case label of "admonition" -> mempty (l:ls) -> B.divWith ("",["admonition-title"],[]) @@ -711,11 +711,11 @@ directive' = do (trim top ++ if null subtit then "" else (": " ++ subtit)) - bod <- parseFromString parseBlocks body' + bod <- parseFromString' parseBlocks body' return $ B.divWith ("",["sidebar"],[]) $ tit <> bod "topic" -> do tit <- B.para . B.strong <$> parseInlineFromString top - bod <- parseFromString parseBlocks body' + bod <- parseFromString' parseBlocks body' return $ B.divWith ("",["topic"],[]) $ tit <> bod "default-role" -> mempty <$ updateState (\s -> s { stateRstDefaultRole = @@ -731,7 +731,7 @@ directive' = do "math" -> return $ B.para $ mconcat $ map B.displayMath $ toChunks $ top ++ "\n\n" ++ body "figure" -> do - (caption, legend) <- parseFromString extractCaption body' + (caption, legend) <- parseFromString' extractCaption body' let src = escapeURI $ trim top return $ B.para (B.imageWith (imgAttr "figclass") src "fig:" caption) <> legend @@ -750,21 +750,21 @@ directive' = do -- directive content or the first immediately following element children <- case body of "" -> block - _ -> parseFromString parseBlocks body' + _ -> parseFromString' parseBlocks body' return $ B.divWith attrs children other -> do pos <- getPosition logMessage $ SkippedContent (".. " ++ other) pos - bod <- parseFromString parseBlocks $ top ++ "\n\n" ++ body' + bod <- parseFromString' parseBlocks $ top ++ "\n\n" ++ body' return $ B.divWith ("",[other],[]) bod tableDirective :: PandocMonad m => String -> [(String, String)] -> String -> RSTParser m Blocks tableDirective top _fields body = do - bs <- parseFromString parseBlocks body + bs <- parseFromString' parseBlocks body case B.toList bs of [Table _ aligns' widths' header' rows'] -> do - title <- parseFromString (trimInlines . mconcat <$> many inline) top + title <- parseFromString' (trimInlines . mconcat <$> many inline) top -- TODO widths -- align is not applicable since we can't represent whole table align return $ B.singleton $ Table (B.toList title) @@ -780,8 +780,8 @@ listTableDirective :: PandocMonad m => String -> [(String, String)] -> String -> RSTParser m Blocks listTableDirective top fields body = do - bs <- parseFromString parseBlocks body - title <- parseFromString (trimInlines . mconcat <$> many inline) top + bs <- parseFromString' parseBlocks body + title <- parseFromString' (trimInlines . mconcat <$> many inline) top let rows = takeRows $ B.toList bs headerRowsNum = fromMaybe (0 :: Int) $ lookup "header-rows" fields >>= safeRead @@ -812,7 +812,7 @@ addNewRole :: PandocMonad m => String -> [(String, String)] -> RSTParser m Blocks addNewRole roleString fields = do pos <- getPosition - (role, parentRole) <- parseFromString inheritedRole roleString + (role, parentRole) <- parseFromString' inheritedRole roleString customRoles <- stateRstCustomRoles <$> getState let getBaseRole (r, f, a) roles = case M.lookup r roles of @@ -1127,7 +1127,7 @@ simpleTableRow indices = do let cols = map unlines . transpose $ firstLine : conLines ++ [replicate (length indices) "" | not (null conLines)] - mapM (parseFromString parseBlocks) cols + mapM (parseFromString' parseBlocks) cols simpleTableSplitLine :: [Int] -> String -> [String] simpleTableSplitLine indices line = @@ -1150,7 +1150,7 @@ simpleTableHeader headless = try $ do let rawHeads = if headless then replicate (length dashes) "" else simpleTableSplitLine indices rawContent - heads <- mapM (parseFromString (mconcat <$> many plain)) $ + heads <- mapM (parseFromString' (mconcat <$> many plain)) $ map trim rawHeads return (heads, aligns, indices) @@ -1206,7 +1206,7 @@ inline = choice [ note -- can start with whitespace, so try before ws , symbol ] <?> "inline" parseInlineFromString :: PandocMonad m => String -> RSTParser m Inlines -parseInlineFromString = parseFromString (trimInlines . mconcat <$> many inline) +parseInlineFromString = parseFromString' (trimInlines . mconcat <$> many inline) hyphens :: Monad m => RSTParser m Inlines hyphens = do @@ -1470,7 +1470,7 @@ note = try $ do -- Note references inside other notes are allowed in reST, but -- not yet in this implementation. updateState $ \st -> st{ stateNotes = [] } - contents <- parseFromString parseBlocks raw + contents <- parseFromString' parseBlocks raw let newnotes = if (ref == "*" || ref == "#") -- auto-numbered -- delete the note so the next auto-numbered note -- doesn't get the same contents: diff --git a/src/Text/Pandoc/Readers/TWiki.hs b/src/Text/Pandoc/Readers/TWiki.hs index ecb609ae9..aea55b7a9 100644 --- a/src/Text/Pandoc/Readers/TWiki.hs +++ b/src/Text/Pandoc/Readers/TWiki.hs @@ -106,7 +106,7 @@ parseHtmlContentWithAttrs tag parser = do parsedContent <- try $ parseContent content return (attr, parsedContent) where - parseContent = parseFromString $ nested $ manyTill parser endOfContent + parseContent = parseFromString' $ nested $ manyTill parser endOfContent endOfContent = try $ skipMany blankline >> skipSpaces >> eof parseHtmlContent :: PandocMonad m => String -> TWParser m a -> TWParser m [a] @@ -233,7 +233,7 @@ listItemLine prefix marker = lineContent >>= parseContent >>= return . mconcat filterSpaces = reverse . dropWhile (== ' ') . reverse listContinuation = notFollowedBy (string prefix >> marker) >> string " " >> lineContent - parseContent = parseFromString $ many1 $ nestedList <|> parseInline + parseContent = parseFromString' $ many1 $ nestedList <|> parseInline parseInline = many1Till inline (lastNewline <|> newlineBeforeNestedList) >>= return . B.plain . mconcat nestedList = list prefix @@ -297,7 +297,7 @@ noautolink = do setState $ st{ stateAllowLinks = True } return $ mconcat blocks where - parseContent = parseFromString $ many $ block + parseContent = parseFromString' $ many $ block para :: PandocMonad m => TWParser m B.Blocks para = many1Till inline endOfParaElement >>= return . result . mconcat @@ -525,4 +525,4 @@ linkText = do return (url, "", content) where linkContent = (char '[') >> many1Till anyChar (char ']') >>= parseLinkContent - parseLinkContent = parseFromString $ many1 inline + parseLinkContent = parseFromString' $ many1 inline diff --git a/src/Text/Pandoc/Readers/Textile.hs b/src/Text/Pandoc/Readers/Textile.hs index abf8be452..52f4f2493 100644 --- a/src/Text/Pandoc/Readers/Textile.hs +++ b/src/Text/Pandoc/Readers/Textile.hs @@ -315,7 +315,7 @@ definitionListItem = try $ do optional whitespace >> newline s <- many1Till anyChar (try (string "=:" >> newline)) -- this ++ "\n\n" does not look very good - ds <- parseFromString parseBlocks (s ++ "\n\n") + ds <- parseFromString' parseBlocks (s ++ "\n\n") return [ds] -- raw content @@ -367,7 +367,7 @@ tableCell = try $ do notFollowedBy blankline raw <- trim <$> many (noneOf "|\n" <|> try (char '\n' <* notFollowedBy blankline)) - content <- mconcat <$> parseFromString (many inline) raw + content <- mconcat <$> parseFromString' (many inline) raw return ((isHeader, alignment), B.plain content) -- | A table row is made of many table cells @@ -389,7 +389,7 @@ table = try $ do _ <- attributes char '.' rawcapt <- trim <$> anyLine - parseFromString (mconcat <$> many inline) rawcapt + parseFromString' (mconcat <$> many inline) rawcapt rawrows <- many1 $ (skipMany ignorableRow) >> tableRow skipMany ignorableRow blanklines @@ -507,7 +507,7 @@ note = try $ do notes <- stateNotes <$> getState case lookup ref notes of Nothing -> fail "note not found" - Just raw -> B.note <$> parseFromString parseBlocks raw + Just raw -> B.note <$> parseFromString' parseBlocks raw -- | Special chars markupChars :: [Char] diff --git a/src/Text/Pandoc/Readers/Txt2Tags.hs b/src/Text/Pandoc/Readers/Txt2Tags.hs index 05c6c9a69..d8791869d 100644 --- a/src/Text/Pandoc/Readers/Txt2Tags.hs +++ b/src/Text/Pandoc/Readers/Txt2Tags.hs @@ -212,7 +212,7 @@ quote :: T2T Blocks quote = try $ do lookAhead tab rawQuote <- many1 (tab *> optional spaces *> anyLine) - contents <- parseFromString parseBlocks (intercalate "\n" rawQuote ++ "\n\n") + contents <- parseFromString' parseBlocks (intercalate "\n" rawQuote ++ "\n\n") return $ B.blockQuote contents commentLine :: T2T Inlines @@ -264,7 +264,7 @@ listItem start end = try $ do firstLine <- anyLineNewline blank <- option "" ("\n" <$ blankline) rest <- concat <$> many (listContinuation markerLength) - parseFromString end $ firstLine ++ blank ++ rest + parseFromString' end $ firstLine ++ blank ++ rest -- continuation of a list item - indented and separated by blankline or endline. -- Note: nested lists are parsed as continuations. @@ -439,7 +439,7 @@ inlineMarkup p f c special = try $ do Just middle -> do lastChar <- anyChar end <- many1 (char c) - let parser inp = parseFromString (mconcat <$> many p) inp + let parser inp = parseFromString' (mconcat <$> many p) inp let start' = case drop 2 start of "" -> mempty xs -> special xs |