fc2f7a4 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

MediaWiki reader: Implemented basic internal links.

fc2f7a4John MacFarlane14 years ago

src/Text/Pandoc/Readers/MediaWiki.hs+62−41

compare with master

unifiedsplit

@@ -110,16 +110,27 @@ htmlComment :: MWParser ()

110htmlComment = () <$ htmlTag isCommentTag

111

112inlinesInTags :: String -> MWParser Inlines

113+inlinesInTags tag = try $ do

114+ (_,raw) <- htmlTag (~== TagOpen tag [])

115+ if '/' `elem` raw -- self-closing tag

116+ then return mempty

113−inlinesInTags tag = trimInlines . mconcat <$> try

117+ else trimInlines . mconcat <$>

114− (manyTill inline (htmlTag (~== TagClose tag)))

118+ manyTill inline (htmlTag (~== TagClose tag))

115

119

116blocksInTags :: String -> MWParser Blocks

120blocksInTags :: String -> MWParser Blocks

117−blocksInTags tag = mconcat <$> try

121+blocksInTags tag = try $ do

122+ (_,raw) <- htmlTag (~== TagOpen tag [])

123+ if '/' `elem` raw -- self-closing tag

124+ then return mempty

118− (manyTill block (htmlTag (~== TagClose tag)))

125+ else mconcat <$> manyTill block (htmlTag (~== TagClose tag))

119

126

120charsInTags :: String -> MWParser [Char]

127charsInTags :: String -> MWParser [Char]

128+charsInTags tag = try $ do

129+ (_,raw) <- htmlTag (~== TagOpen tag [])

130+ if '/' `elem` raw -- self-closing tag

131+ then return ""

121−charsInTags tag = innerText . parseTags <$> try

132+ else innerText . parseTags <$>

122− (manyTill anyChar (htmlTag (~== TagClose tag)))

133+ manyTill anyChar (htmlTag (~== TagClose tag))

123

134

124--

135--

125-- main parser

136-- main parser

@@ -146,8 +157,6 @@ block = mempty <$ skipMany1 blankline

146 <|> mempty <$ try (spaces *> htmlComment)

157 <|> mempty <$ try (spaces *> htmlComment)

147 <|> preformatted

158 <|> preformatted

148 <|> blockTag

159 <|> blockTag

149− <|> pTag

150− <|> blockHtml

151 <|> template

160 <|> template

152 <|> para

161 <|> para

153

162

@@ -164,16 +173,17 @@ template = B.rawBlock "mediawiki" <$> doublebrackets

164

173

165blockTag :: MWParser Blocks

174blockTag :: MWParser Blocks

166blockTag = do

175blockTag = do

167− (TagOpen t attrs, raw) <- htmlTag (\x -> isBlockTag' x && isTagOpen x)

176+ (tag, _) <- lookAhead $ htmlTag isBlockTag'

168− case t of

177+ case tag of

169− "blockquote" -> B.blockQuote <$> blocksInTags "blockquote"

178+ TagOpen "blockquote" _ -> B.blockQuote <$> blocksInTags "blockquote"

170− "pre" -> B.codeBlock . trimCode <$> charsInTags "pre"

179+ TagOpen "pre" _ -> B.codeBlock . trimCode <$> charsInTags "pre"

171− "syntaxhighlight" -> syntaxhighlight attrs

180+ TagOpen "syntaxhighlight" attrs -> syntaxhighlight attrs

172− "haskell" -> B.codeBlockWith ("",["haskell"],[]) . trimCode <$>

181+ TagOpen "haskell" _ -> B.codeBlockWith ("",["haskell"],[]) . trimCode <$>

173− charsInTags "haskell"

182+ charsInTags "haskell"

174− "gallery" -> blocksInTags "gallery"

183+ TagOpen "gallery" _ -> blocksInTags "gallery"

175− "p" -> return mempty

184+ TagOpen "p" _ -> mempty <$ htmlTag (~== tag)

185+ TagClose "p" -> mempty <$ htmlTag (~== tag)

176− _ -> return $ B.rawBlock "html" raw

186+ _ -> B.rawBlock "html" . snd <$> htmlTag (~== tag)

177

187

178trimCode :: String -> String

188trimCode :: String -> String

179trimCode ('\n':xs) = stripTrailingNewlines xs

189trimCode ('\n':xs) = stripTrailingNewlines xs

@@ -189,13 +199,6 @@ syntaxhighlight attrs = try $ do

189 contents <- charsInTags "syntaxhighlight"

199 contents <- charsInTags "syntaxhighlight"

190 return $ B.codeBlockWith ("",classes,kvs) $ trimCode contents

200 return $ B.codeBlockWith ("",classes,kvs) $ trimCode contents

191

201

192−-- We can just skip pTags, as contents will be treated as paragraphs

193−pTag :: MWParser Blocks

194−pTag = mempty <$ (htmlTag (\t -> t ~== TagOpen "p" [] || t ~== TagClose "p"))

195−

196−blockHtml :: MWParser Blocks

197−blockHtml = (B.rawBlock "html" . snd <$> htmlTag isBlockTag)

198−

199hrule :: MWParser Blocks

202hrule :: MWParser Blocks

200hrule = B.horizontalRule <$ try (string "----" *> many (char '-') *> newline)

203hrule = B.horizontalRule <$ try (string "----" *> many (char '-') *> newline)

201

204

@@ -271,7 +274,7 @@ anyListStart = char '*'

271 <|> char ';'

274 <|> char ';'

272

275

273li :: MWParser Blocks

276li :: MWParser Blocks

274−li = htmlTag (~== TagOpen "li" []) *>

277+li = lookAhead (htmlTag (~== TagOpen "li" [])) *>

275 (firstParaToPlain <$> blocksInTags "li") <* spaces

278 (firstParaToPlain <$> blocksInTags "li") <* spaces

276

279

277listItem :: Char -> MWParser Blocks

280listItem :: Char -> MWParser Blocks

@@ -317,6 +320,7 @@ inline = whitespace

317 <|> str

320 <|> str

318 <|> strong

321 <|> strong

319 <|> emph

322 <|> emph

323+ <|> internalLink

320 <|> externalLink

324 <|> externalLink

321 <|> inlineTag

325 <|> inlineTag

322 <|> B.singleton <$> charRef

326 <|> B.singleton <$> charRef

@@ -336,21 +340,25 @@ variable = B.rawInline "mediawiki" <$> triplebrackets

336

340

337inlineTag :: MWParser Inlines

341inlineTag :: MWParser Inlines

338inlineTag = do

342inlineTag = do

339− (TagOpen t _, raw) <- htmlTag (\x -> isInlineTag x && isTagOpen x)

343+ (tag, _) <- lookAhead $ htmlTag isInlineTag

340− case t of

344+ case tag of

345+ TagOpen "nowiki" _ -> try $ do

346+ (_,raw) <- htmlTag (~== tag)

347+ if '/' `elem` raw

348+ then return mempty

341− "nowiki" -> B.text . fromEntities <$> try

349+ else B.text . fromEntities <$>

342− (manyTill anyChar (htmlTag (~== TagClose "nowiki")))

350+ manyTill anyChar (htmlTag (~== TagClose "nowiki"))

343− "br" -> B.linebreak <$

351+ TagOpen "br" _ -> B.linebreak <$ (htmlTag (~== TagOpen "br" []) -- will get /> too

344− (optional (htmlTag (~== TagClose "br")) *> optional blankline)

352+ *> optional blankline)

345− "strike" -> B.strikeout <$> inlinesInTags "strike"

353+ TagOpen "strike" _ -> B.strikeout <$> inlinesInTags "strike"

346− "del" -> B.strikeout <$> inlinesInTags "del"

354+ TagOpen "del" _ -> B.strikeout <$> inlinesInTags "del"

347− "sub" -> B.subscript <$> inlinesInTags "sub"

355+ TagOpen "sub" _ -> B.subscript <$> inlinesInTags "sub"

348− "sup" -> B.superscript <$> inlinesInTags "sup"

356+ TagOpen "sup" _ -> B.superscript <$> inlinesInTags "sup"

349− "math" -> B.math <$> charsInTags "math"

357+ TagOpen "math" _ -> B.math <$> charsInTags "math"

350− "code" -> B.code <$> charsInTags "code"

358+ TagOpen "code" _ -> B.code <$> charsInTags "code"

351− "tt" -> B.code <$> charsInTags "tt"

359+ TagOpen "tt" _ -> B.code <$> charsInTags "tt"

352− "hask" -> B.codeWith ("",["haskell"],[]) <$> charsInTags "hask"

360+ TagOpen "hask" _ -> B.codeWith ("",["haskell"],[]) <$> charsInTags "hask"

353− _ -> return $ B.rawInline "html" raw

361+ _ -> B.rawInline "html" . snd <$> htmlTag (~== tag)

354

362

355special :: MWParser Inlines

363special :: MWParser Inlines

356special = B.str <$> count 1 (notFollowedBy' (htmlTag isBlockTag') *>

364special = B.str <$> count 1 (notFollowedBy' (htmlTag isBlockTag') *>

@@ -368,6 +376,19 @@ endline = () <$ try (newline <*

368 notFollowedBy' hrule <*

376 notFollowedBy' hrule <*

369 notFollowedBy anyListStart)

377 notFollowedBy anyListStart)

370

378

379+internalLink :: MWParser Inlines

380+internalLink = try $ do

381+ string "[["

382+ pagename <- unwords . words <$> many (noneOf "|]")

383+ label <- option (B.text pagename) $ char '|' *>

384+ ( (mconcat <$> many1 (notFollowedBy (char ']') *> inline))

385+ -- the "pipe trick"

386+ -- [[Help:Contents|] -> "Contents"

387+ <|> (return $ B.text $ drop 1 $ dropWhile (/=':') pagename) )

388+ string "]]"

389+ linktrail <- B.text <$> many (char '\'' <|> letter)

390+ return $ B.link pagename "wikilink" (label <> linktrail)

391+

371externalLink :: MWParser Inlines

392externalLink :: MWParser Inlines

372externalLink = try $ do

393externalLink = try $ do

373 char '['

394 char '['

tests/mediawiki-reader.native+6

@@ -74,6 +74,12 @@ Pandoc (Meta {docTitle = [], docAuthors = [], docDate = []})

74,Para [Link [Str "http://johnmacfarlane.net/pandoc/"] ("http://johnmacfarlane.net/pandoc/","")]

75,Para [Link [Str "1"] ("http://google.com",""),Space,Link [Str "2"] ("http://yahoo.com","")]

76,Para [Link [Str "email",Space,Str "me"] ("mailto:info@example.org","")]

77+,Header 2 [Str "internal",Space,Str "links"]

78+,Para [Link [Str "Help"] ("Help","wikilink")]

79+,Para [Link [Str "the",Space,Str "help",Space,Str "page"] ("Help","wikilink")]

80+,Para [Link [Str "Helpers"] ("Help","wikilink")]

81+,Para [Link [Str "Help"] ("Help","wikilink"),Str "ers"]

82+,Para [Link [Str "Contents"] ("Help:Contents","wikilink")]

77,Header 2 [Str "lists"]

83,Header 2 [Str "lists"]

78,BulletList

84,BulletList

79 [[Plain [Str "Start",Space,Str "each",Space,Str "line"]]

85 [[Plain [Str "Start",Space,Str "each",Space,Str "line"]]

tests/mediawiki-reader.wiki+12

@@ -147,6 +147,18 @@ http://johnmacfarlane.net/pandoc/

147

148[mailto:info@example.org email me]

149

150+== internal links ==

151+

152+[[Help]]

153+

154+[[Help|the help page]]

155+

156+[[Help]]ers

157+

158+[[Help]]<nowiki/>ers

159+

160+[[Help:Contents|]]

161+

150== lists ==

162== lists ==

151

163

152* Start each line

164* Start each line