diff options
author | Vaclav Zeman <vhaisman@gmail.com> | 2014-02-09 13:16:39 +0100 |
---|---|---|
committer | Vaclav Zeman <vhaisman@gmail.com> | 2014-02-09 13:16:39 +0100 |
commit | 1ba8066f67d1f8624e679e28fd800e6ade2a9511 (patch) | |
tree | 4bdd9c34936f7be77b95aaf4df723980aa675a25 /src/Text/Pandoc/Readers/MediaWiki.hs | |
parent | 3f0fe345f9aa69d1faf36e6a6f913013f21b3749 (diff) | |
parent | 927b51630852b56f6c8e0aa4a7763108fdb8ec89 (diff) | |
download | pandoc-1ba8066f67d1f8624e679e28fd800e6ade2a9511.tar.gz |
Merge remote-tracking branch 'origin/master' into en-dash-ligature-avoidance.
Diffstat (limited to 'src/Text/Pandoc/Readers/MediaWiki.hs')
-rw-r--r-- | src/Text/Pandoc/Readers/MediaWiki.hs | 52 |
1 files changed, 38 insertions, 14 deletions
diff --git a/src/Text/Pandoc/Readers/MediaWiki.hs b/src/Text/Pandoc/Readers/MediaWiki.hs index 136701bd0..794890eb6 100644 --- a/src/Text/Pandoc/Readers/MediaWiki.hs +++ b/src/Text/Pandoc/Readers/MediaWiki.hs @@ -1,4 +1,5 @@ -{-# LANGUAGE RelaxedPolyRec #-} -- needed for inlinesBetween on GHC < 7 +{-# LANGUAGE RelaxedPolyRec, FlexibleInstances, TypeSynonymInstances #-} +-- RelaxedPolyRec needed for inlinesBetween on GHC < 7 {- Copyright (C) 2012 John MacFarlane <jgm@berkeley.edu> @@ -51,7 +52,9 @@ import Data.List (intersperse, intercalate, isPrefixOf ) import Text.HTML.TagSoup import Data.Sequence (viewl, ViewL(..), (<|)) import qualified Data.Foldable as F +import qualified Data.Map as M import Data.Char (isDigit, isSpace) +import Data.Maybe (fromMaybe) -- | Read mediawiki from an input string and return a Pandoc document. readMediaWiki :: ReaderOptions -- ^ Reader options @@ -62,6 +65,8 @@ readMediaWiki opts s = , mwMaxNestingLevel = 4 , mwNextLinkNumber = 1 , mwCategoryLinks = [] + , mwHeaderMap = M.empty + , mwIdentifierList = [] } "source" (s ++ "\n") of Left err' -> error $ "\nError:\n" ++ show err' @@ -71,10 +76,23 @@ data MWState = MWState { mwOptions :: ReaderOptions , mwMaxNestingLevel :: Int , mwNextLinkNumber :: Int , mwCategoryLinks :: [Inlines] + , mwHeaderMap :: M.Map Inlines String + , mwIdentifierList :: [String] } type MWParser = Parser [Char] MWState +instance HasReaderOptions MWParser where + askReaderOption f = (f . mwOptions) `fmap` getState + +instance HasHeaderMap MWParser where + getHeaderMap = fmap mwHeaderMap getState + putHeaderMap hm = updateState $ \st -> st{ mwHeaderMap = hm } + +instance HasIdentifierList MWParser where + getIdentifierList = fmap mwIdentifierList getState + putIdentifierList l = updateState $ \st -> st{ mwIdentifierList = l } + -- -- auxiliary functions -- @@ -131,9 +149,16 @@ inlinesInTags tag = try $ do blocksInTags :: String -> MWParser Blocks blocksInTags tag = try $ do (_,raw) <- htmlTag (~== TagOpen tag []) + let closer = if tag == "li" + then htmlTag (~== TagClose "li") + <|> lookAhead ( + htmlTag (~== TagOpen "li" []) + <|> htmlTag (~== TagClose "ol") + <|> htmlTag (~== TagClose "ul")) + else htmlTag (~== TagClose tag) if '/' `elem` raw -- self-closing tag then return mempty - else mconcat <$> manyTill block (htmlTag (~== TagClose tag)) + else mconcat <$> manyTill block closer charsInTags :: String -> MWParser [Char] charsInTags tag = try $ do @@ -187,7 +212,7 @@ table = do tableStart styles <- option [] parseAttrs <* blankline let tableWidth = case lookup "width" styles of - Just w -> maybe 1.0 id $ parseWidth w + Just w -> fromMaybe 1.0 $ parseWidth w Nothing -> 1.0 caption <- option mempty tableCaption optional rowsep @@ -268,7 +293,7 @@ tableCell = try $ do Just "center" -> AlignCenter _ -> AlignDefault let width = case lookup "width" attrs of - Just xs -> maybe 0.0 id $ parseWidth xs + Just xs -> fromMaybe 0.0 $ parseWidth xs Nothing -> 0.0 return ((align, width), bs) @@ -351,7 +376,8 @@ header = try $ do let lev = length eqs guard $ lev <= 6 contents <- trimInlines . mconcat <$> manyTill inline (count lev $ char '=') - return $ B.header lev contents + attr <- registerHeader nullAttr contents + return $ B.headerWith attr lev contents bulletList :: MWParser Blocks bulletList = B.bulletList <$> @@ -362,15 +388,13 @@ bulletList = B.bulletList <$> orderedList :: MWParser Blocks orderedList = (B.orderedList <$> many1 (listItem '#')) - <|> (B.orderedList <$> (htmlTag (~== TagOpen "ul" []) *> spaces *> - many (listItem '#' <|> li) <* - optional (htmlTag (~== TagClose "ul")))) - <|> do (tag,_) <- htmlTag (~== TagOpen "ol" []) - spaces - items <- many (listItem '#' <|> li) - optional (htmlTag (~== TagClose "ol")) - let start = maybe 1 id $ safeRead $ fromAttrib "start" tag - return $ B.orderedListWith (start, DefaultStyle, DefaultDelim) items + <|> try + (do (tag,_) <- htmlTag (~== TagOpen "ol" []) + spaces + items <- many (listItem '#' <|> li) + optional (htmlTag (~== TagClose "ol")) + let start = fromMaybe 1 $ safeRead $ fromAttrib "start" tag + return $ B.orderedListWith (start, DefaultStyle, DefaultDelim) items) definitionList :: MWParser Blocks definitionList = B.definitionList <$> many1 defListItem |