diff options
Diffstat (limited to 'src/Text/Pandoc/Readers')
-rw-r--r-- | src/Text/Pandoc/Readers/Markdown.hs | 23 | ||||
-rw-r--r-- | src/Text/Pandoc/Readers/Org.hs | 66 |
2 files changed, 59 insertions, 30 deletions
diff --git a/src/Text/Pandoc/Readers/Markdown.hs b/src/Text/Pandoc/Readers/Markdown.hs index d1637b701..5129bc2e3 100644 --- a/src/Text/Pandoc/Readers/Markdown.hs +++ b/src/Text/Pandoc/Readers/Markdown.hs @@ -1474,9 +1474,7 @@ strongOrEmph = enclosure '*' <|> (checkIntraword >> enclosure '_') where checkIntraword = do exts <- getOption readerExtensions when (Ext_intraword_underscores `Set.member` exts) $ do - pos <- getPosition - lastStrPos <- stateLastStrPos <$> getState - guard $ lastStrPos /= Just pos + guard =<< notAfterString -- | Parses a list of inlines between start and end delimiters. inlinesBetween :: (Show b) @@ -1518,8 +1516,7 @@ nonEndline = satisfy (/='\n') str :: MarkdownParser (F Inlines) str = do result <- many1 alphaNum - pos <- getPosition - updateState $ \s -> s{ stateLastStrPos = Just pos } + updateLastStrPos let spacesToNbr = map (\c -> if c == ' ' then '\160' else c) isSmart <- getOption readerSmart if isSmart @@ -1817,22 +1814,6 @@ normalCite = try $ do char ']' return citations -citeKey :: MarkdownParser (Bool, String) -citeKey = try $ do - -- make sure we're not right after an alphanumeric, - -- since foo@bar.baz is probably an email address - lastStrPos <- stateLastStrPos <$> getState - pos <- getPosition - guard $ lastStrPos /= Just pos - suppress_author <- option False (char '-' >> return True) - char '@' - first <- letter <|> char '_' - let regchar = satisfy (\c -> isAlphaNum c || c == '_') - let internal p = try $ p >>~ lookAhead regchar - rest <- many $ regchar <|> internal (oneOf ":.#$%&-+?<>~/") - let key = first:rest - return (suppress_author, key) - suffix :: MarkdownParser (F Inlines) suffix = try $ do hasSpace <- option False (notFollowedBy nonspaceChar >> return True) diff --git a/src/Text/Pandoc/Readers/Org.hs b/src/Text/Pandoc/Readers/Org.hs index 2e4a29beb..86dda2732 100644 --- a/src/Text/Pandoc/Readers/Org.hs +++ b/src/Text/Pandoc/Readers/Org.hs @@ -105,6 +105,10 @@ instance HasMeta OrgParserState where deleteMeta field st = st{ orgStateMeta = deleteMeta field $ orgStateMeta st } +instance HasLastStrPosition OrgParserState where + getLastStrPos = orgStateLastStrPos + setLastStrPos pos st = st{ orgStateLastStrPos = Just pos } + instance Default OrgParserState where def = defaultOrgParserState @@ -865,6 +869,7 @@ inline :: OrgParser (F Inlines) inline = choice [ whitespace , linebreak + , cite , footnote , linkOrImage , anchor @@ -929,6 +934,51 @@ endline = try $ do updateLastPreCharPos return . return $ B.space +cite :: OrgParser (F Inlines) +cite = try $ do + guardEnabled Ext_citations + (cs, raw) <- withRaw normalCite + return $ (flip B.cite (B.text raw)) <$> cs + +normalCite :: OrgParser (F [Citation]) +normalCite = try $ char '[' + *> skipSpaces + *> citeList + <* skipSpaces + <* char ']' + +citeList :: OrgParser (F [Citation]) +citeList = sequence <$> sepBy1 citation (try $ char ';' *> skipSpaces) + +citation :: OrgParser (F Citation) +citation = try $ do + pref <- prefix + (suppress_author, key) <- citeKey + suff <- suffix + return $ do + x <- pref + y <- suff + return $ Citation{ citationId = key + , citationPrefix = B.toList x + , citationSuffix = B.toList y + , citationMode = if suppress_author + then SuppressAuthor + else NormalCitation + , citationNoteNum = 0 + , citationHash = 0 + } + where + prefix = trimInlinesF . mconcat <$> + manyTill inline (char ']' <|> (']' <$ lookAhead citeKey)) + suffix = try $ do + hasSpace <- option False (notFollowedBy nonspaceChar >> return True) + skipSpaces + rest <- trimInlinesF . mconcat <$> + many (notFollowedBy (oneOf ";]") *> inline) + return $ if hasSpace + then (B.space <>) <$> rest + else rest + footnote :: OrgParser (F Inlines) footnote = try $ inlineNote <|> referencedNote @@ -1003,7 +1053,7 @@ selfTarget :: OrgParser String selfTarget = try $ char '[' *> linkTarget <* char ']' linkTarget :: OrgParser String -linkTarget = enclosed (char '[') (char ']') (noneOf "\n\r[]") +linkTarget = enclosedByPair '[' ']' (noneOf "\n\r[]") applyCustomLinkFormat :: String -> OrgParser (F String) applyCustomLinkFormat link = do @@ -1079,7 +1129,12 @@ inlineCodeBlock = try $ do let attrClasses = [translateLang lang, rundocBlockClass] let attrKeyVal = map toRundocAttrib (("language", lang) : opts) returnF $ B.codeWith ("", attrClasses, attrKeyVal) inlineCode - where enclosedByPair s e p = char s *> many1Till p (char e) + +enclosedByPair :: Char -- ^ opening char + -> Char -- ^ closing char + -> OrgParser a -- ^ parser + -> OrgParser [a] +enclosedByPair s e p = char s *> many1Till p (char e) emph :: OrgParser (F Inlines) emph = fmap B.emph <$> emphasisBetween '/' @@ -1274,13 +1329,6 @@ afterEmphasisPreChar = do lastPrePos <- orgStateLastPreCharPos <$> getState return . fromMaybe True $ (== pos) <$> lastPrePos --- | Whether we are right after the end of a string -notAfterString :: OrgParser Bool -notAfterString = do - pos <- getPosition - lastStrPos <- orgStateLastStrPos <$> getState - return $ lastStrPos /= Just pos - -- | Whether the parser is right after a forbidden border char notAfterForbiddenBorderChar :: OrgParser Bool notAfterForbiddenBorderChar = do |