6e9351b · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

MediaWiki reader: Support for basic mediawiki tables.

6e9351bJohn MacFarlane14 years ago

src/Text/Pandoc/Readers/MediaWiki.hs+52−5

compare with master

unifiedsplit

@@ -30,7 +30,6 @@ Conversion of mediawiki text to 'Pandoc' document.

30-}

31{-

32TODO:

33−_ support images http://www.mediawiki.org/wiki/Help:Images

34_ support tables http://www.mediawiki.org/wiki/Help:Tables

33_ support tables http://www.mediawiki.org/wiki/Help:Tables

35- footnotes?

34- footnotes?

36-}

35-}

@@ -147,6 +146,7 @@ parseMediaWiki = do

147

146

148block :: MWParser Blocks

147block :: MWParser Blocks

149block = mempty <$ skipMany1 blankline

148block = mempty <$ skipMany1 blankline

149+ <|> table

150 <|> header

151 <|> hrule

152 <|> orderedList

@@ -161,6 +161,48 @@ block = mempty <$ skipMany1 blankline

161para :: MWParser Blocks

162para = B.para . trimInlines . mconcat <$> many1 inline

163

164+table :: MWParser Blocks

165+table = do

166+ tableStart

167+ caption <- option mempty tableCaption

168+ optional rowsep

169+ hasheader <- option False $ True <$ (lookAhead (char '!'))

170+ hdr <- tableRow

171+ rows' <- many $ try $ rowsep *> tableRow

172+ tableEnd

173+ -- TODO handle cellspecs from styles and aligns...

174+ let cols = length $ head rows'

175+ let (headers,rows) = if hasheader

176+ then (hdr, rows')

177+ else (replicate cols mempty, hdr:rows')

178+ let cellspecs = replicate cols (AlignDefault, 0.0)

179+ return $ B.table caption cellspecs headers rows

180+

181+tableStart :: MWParser ()

182+tableStart = try $ guardColumnOne *> sym "{|" <* blanklines

183+

184+tableEnd :: MWParser ()

185+tableEnd = try $ guardColumnOne *> sym "|}" <* blanklines

186+

187+rowsep :: MWParser ()

188+rowsep = try $ guardColumnOne *> sym "|-" <* blanklines

189+

190+-- TODO add something like 'guard inTable' since this is used in endline

191+cellsep :: MWParser ()

192+cellsep = try $ guardColumnOne <*

193+ (char '!' <|> (char '|' <* notFollowedBy (oneOf "-}+")))

194+

195+tableCaption :: MWParser Inlines

196+tableCaption = try $ guardColumnOne *> sym "|+" *> skipMany spaceChar *>

197+ (trimInlines . mconcat <$> (many inline)) <* skipMany blankline

198+

199+tableRow :: MWParser [Blocks]

200+tableRow = try $ many tableCell <* skipMany blankline

201+

202+tableCell :: MWParser Blocks

203+tableCell =

204+ try $ cellsep *> skipMany spaceChar *> (mconcat <$> (many block))

205+

164template :: MWParser Blocks

206template :: MWParser Blocks

165template = B.rawBlock "mediawiki" <$> doublebrackets

207template = B.rawBlock "mediawiki" <$> doublebrackets

166 where doublebrackets = try $ do

208 where doublebrackets = try $ do

@@ -200,9 +242,12 @@ syntaxhighlight attrs = try $ do

200hrule :: MWParser Blocks

242hrule :: MWParser Blocks

201hrule = B.horizontalRule <$ try (string "----" *> many (char '-') *> newline)

243hrule = B.horizontalRule <$ try (string "----" *> many (char '-') *> newline)

202

244

245+guardColumnOne :: MWParser ()

246+guardColumnOne = getPosition >>= \pos -> guard (sourceColumn pos == 1)

247+

203preformatted :: MWParser Blocks

248preformatted :: MWParser Blocks

204preformatted = try $ do

249preformatted = try $ do

205− getPosition >>= \pos -> guard (sourceColumn pos == 1)

250+ guardColumnOne

206 char ' '

251 char ' '

207 let endline' = B.linebreak <$ (try $ newline <* char ' ')

252 let endline' = B.linebreak <$ (try $ newline <* char ' ')

208 let whitespace' = B.str <$> many1 ('\160' <$ spaceChar)

253 let whitespace' = B.str <$> many1 ('\160' <$ spaceChar)

@@ -219,8 +264,7 @@ preformatted = try $ do

219

264

220header :: MWParser Blocks

265header :: MWParser Blocks

221header = try $ do

266header = try $ do

222− col <- sourceColumn <$> getPosition

267+ guardColumnOne

223− guard $ col == 1 -- header must be at beginning of line

224 eqs <- many1 (char '=')

268 eqs <- many1 (char '=')

225 let lev = length eqs

269 let lev = length eqs

226 guard $ lev <= 6

270 guard $ lev <= 6

@@ -361,7 +405,7 @@ inlineTag = do

361

405

362special :: MWParser Inlines

406special :: MWParser Inlines

363special = B.str <$> count 1 (notFollowedBy' (htmlTag isBlockTag') *>

407special = B.str <$> count 1 (notFollowedBy' (htmlTag isBlockTag') *>

364− oneOf specialChars)

408+ notFollowedBy (char '|') *> oneOf specialChars)

365

409

366inlineHtml :: MWParser Inlines

410inlineHtml :: MWParser Inlines

367inlineHtml = B.rawInline "html" . snd <$> htmlTag isInlineTag

411inlineHtml = B.rawInline "html" . snd <$> htmlTag isInlineTag

@@ -373,6 +417,9 @@ endline :: MWParser ()

373endline = () <$ try (newline <*

417endline = () <$ try (newline <*

374 notFollowedBy blankline <*

418 notFollowedBy blankline <*

375 notFollowedBy' hrule <*

419 notFollowedBy' hrule <*

420+ notFollowedBy tableStart <*

421+ notFollowedBy' template <*

422+ notFollowedBy cellsep <*

376 notFollowedBy anyListStart)

423 notFollowedBy anyListStart)

377

424

378image :: MWParser Inlines

425image :: MWParser Inlines