5104c21 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

MediaWiki reader: preformatted blocks and tests.

5104c21John MacFarlane14 years ago

src/Text/Pandoc/Readers/MediaWiki.hs+28−11

compare with master

unifiedsplit

@@ -1,6 +1,6 @@

1{-# LANGUAGE RelaxedPolyRec #-} -- needed for inlinesBetween on GHC < 7

2{-

5This program is free software; you can redistribute it and/or modify

6it under the terms of the GNU General Public License as published by

@@ -33,8 +33,6 @@ TODO:

33_ support HTML lists

34_ support list style attributes and start values in ol lists, also

35 value attribute on li

36−_ support preformatted text (lines starting with space)

37−_ support preformatted text blocks

38_ support internal links http://www.mediawiki.org/wiki/Help:Links

36_ support internal links http://www.mediawiki.org/wiki/Help:Links

39_ support external links (partially implemented)

37_ support external links (partially implemented)

40_ support images http://www.mediawiki.org/wiki/Help:Images

38_ support images http://www.mediawiki.org/wiki/Help:Images

@@ -50,13 +48,13 @@ import Text.Pandoc.Definition

50import qualified Text.Pandoc.Builder as B

48import qualified Text.Pandoc.Builder as B

51import Text.Pandoc.Builder (Inlines, Blocks, trimInlines, (<>))

49import Text.Pandoc.Builder (Inlines, Blocks, trimInlines, (<>))

52import Text.Pandoc.Options

50import Text.Pandoc.Options

53−import Text.Pandoc.Readers.HTML ( htmlTag, htmlInBalanced,

51+import Text.Pandoc.Readers.HTML ( htmlTag, isInlineTag,

54− isInlineTag, isBlockTag, isTextTag, isCommentTag )

52+ isBlockTag, isCommentTag )

55import Text.Pandoc.XML ( fromEntities )

53import Text.Pandoc.XML ( fromEntities )

56import Text.Pandoc.Parsing

54import Text.Pandoc.Parsing

55+import Text.Pandoc.Generic ( bottomUp )

57import Text.Pandoc.Shared ( stripTrailingNewlines )

56import Text.Pandoc.Shared ( stripTrailingNewlines )

58import Data.Monoid (mconcat, mempty)

57import Data.Monoid (mconcat, mempty)

59−import qualified Data.Foldable as F

60import Control.Applicative ((<$>), (<*), (*>), (<$))

58import Control.Applicative ((<$>), (<*), (*>), (<$))

61import Control.Monad

59import Control.Monad

62import Data.List (intersperse)

60import Data.List (intersperse)

@@ -124,12 +122,13 @@ block = header

124 <|> bulletList

122 <|> bulletList

125 <|> orderedList

123 <|> orderedList

126 <|> definitionList

124 <|> definitionList

125+ <|> mempty <$ try (spaces *> htmlComment)

126+ <|> preformatted

127 <|> blockquote

128 <|> codeblock

129 <|> syntaxhighlight

130 <|> haskell

131 <|> mempty <$ skipMany1 blankline

132− <|> mempty <$ try (spaces *> htmlComment)

133 <|> pTag

132 <|> pTag

134 <|> blockHtml

133 <|> blockHtml

135 <|> para

134 <|> para

@@ -147,6 +146,22 @@ blockHtml = (B.rawBlock "html" . snd <$> htmlTag isBlockTag)

147hrule :: MWParser Blocks

146hrule :: MWParser Blocks

148hrule = B.horizontalRule <$ try (string "----" *> many (char '-') *> newline)

147hrule = B.horizontalRule <$ try (string "----" *> many (char '-') *> newline)

149

148

149+preformatted :: MWParser Blocks

150+preformatted = do

151+ char ' '

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

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

154+ let spToNbsp ' ' = '\160'

155+ spToNbsp x = x

156+ let nowiki' = mconcat . intersperse B.linebreak . map B.str .

157+ lines . fromEntities . map spToNbsp <$> try

158+ (htmlTag (~== TagOpen "nowiki" []) *>

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

160+ let inline' = whitespace' <|> endline' <|> nowiki' <|> inline

161+ let strToCode (Str s) = Code ("",[],[]) s

162+ strToCode x = x

163+ B.para . bottomUp strToCode . mconcat <$> many1 inline'

164+

150blockquote :: MWParser Blocks

165blockquote :: MWParser Blocks

151blockquote = B.blockQuote <$> blocksInTags "blockquote"

166blockquote = B.blockQuote <$> blocksInTags "blockquote"

152

167

@@ -159,8 +174,8 @@ trimCode xs = stripTrailingNewlines xs

159

174

160syntaxhighlight :: MWParser Blocks

175syntaxhighlight :: MWParser Blocks

161syntaxhighlight = try $ do

176syntaxhighlight = try $ do

162− (tag@(TagOpen _ attrs), _) <- lookAhead

177+ (TagOpen _ attrs, _) <- lookAhead

163− $ htmlTag (~== TagOpen "syntaxhighlight" [])

178+ $ htmlTag (~== TagOpen "syntaxhighlight" [])

164 let mblang = lookup "lang" attrs

179 let mblang = lookup "lang" attrs

165 let mbstart = lookup "start" attrs

180 let mbstart = lookup "start" attrs

166 let mbline = lookup "line" attrs

181 let mbline = lookup "line" attrs

@@ -220,6 +235,7 @@ listItem c = try $ do

220 if null extras

235 if null extras

221 then listItem' c

236 then listItem' c

222 else do

237 else do

238+ skipMany spaceChar

223 first <- manyTill anyChar newline

239 first <- manyTill anyChar newline

224 rest <- many (try $ string extras *> manyTill anyChar newline)

240 rest <- many (try $ string extras *> manyTill anyChar newline)

225 contents <- parseFromString (many1 $ listItem' c)

241 contents <- parseFromString (many1 $ listItem' c)

@@ -233,14 +249,15 @@ listItem c = try $ do

233listItem' :: Char -> MWParser Blocks

249listItem' :: Char -> MWParser Blocks

234listItem' c = try $ do

250listItem' c = try $ do

235 listStart c

251 listStart c

252+ skipMany spaceChar

236 first <- manyTill anyChar newline

253 first <- manyTill anyChar newline

237 rest <- many (try $ char c *> lookAhead listStartChar *>

254 rest <- many (try $ char c *> lookAhead listStartChar *>

238 manyTill anyChar newline)

255 manyTill anyChar newline)

239 contents <- parseFromString (mconcat <$> many1 block)

256 contents <- parseFromString (mconcat <$> many1 block)

240 $ unlines $ first : rest

257 $ unlines $ first : rest

241 case viewl (B.unMany contents) of

258 case viewl (B.unMany contents) of

242− (Para xs) :< rest -> return $ B.Many $ (Plain xs) <| rest

259+ (Para xs) :< ys -> return $ B.Many $ (Plain xs) <| ys

243− _ -> return contents

260+ _ -> return contents

244

261

245--

262--

246-- inline parsers

263-- inline parsers

tests/mediawiki-reader.native+6−2

@@ -6,7 +6,7 @@ Pandoc (Meta {docTitle = [], docAuthors = [], docDate = []})

6,Header 5 [Str "header",Space,Str "level",Space,Str "5"]

7,Header 6 [Str "header",Space,Str "level",Space,Str "6"]

8,Para [Str "=======",Space,Str "not",Space,Str "a",Space,Str "header",Space,Str "========"]

9−,Para [Str "==",Space,Str "not",Space,Str "a",Space,Str "header",Space,Str "=="]

9+,Para [Code ("",[],[]) "==\160not\160a\160header\160=="]

10,Header 2 [Str "emph",Space,Str "and",Space,Str "strong"]

11,Para [Emph [Str "emph"],Space,Strong [Str "strong"]]

12,Para [Strong [Emph [Str "strong",Space,Str "and",Space,Str "emph"]]]

@@ -131,4 +131,8 @@ Pandoc (Meta {docTitle = [], docAuthors = [], docDate = []})

131 [[Plain [Str "five",Space,Str "sub",Space,Str "1",Space,Str "sub",Space,Str "1"]]]]

132 ,[Plain [Str "five",Space,Str "sub",Space,Str "2"]]]]]

133,Header 2 [Str "math"]

134−,Para [Str "Here",Space,Str "is",Space,Str "some",Space,Math InlineMath "x=\\frac{y^\\pi}{z}",Str "."]]

134+,Para [Str "Here",Space,Str "is",Space,Str "some",Space,Math InlineMath "x=\\frac{y^\\pi}{z}",Str "."]

135+,Header 2 [Str "preformatted",Space,Str "blocks"]

136+,Para [Code ("",[],[]) "Start\160each\160line\160with\160a\160space.",LineBreak,Code ("",[],[]) "Text\160is\160",Strong [Code ("",[],[]) "preformatted"],Code ("",[],[]) "\160and",LineBreak,Emph [Code ("",[],[]) "markups"],Code ("",[],[]) "\160",Strong [Emph [Code ("",[],[]) "can"]],Code ("",[],[]) "\160be\160done."]

137+,Para [Code ("",[],[]) "\160hell\160\160\160\160\160\160yeah"]

138+,Para [Code ("",[],[]) "Start\160with\160a\160space\160in\160the\160first\160column,",LineBreak,Code ("",[],[]) "(before\160the\160<nowiki>).",LineBreak,Code ("",[],[]) "",LineBreak,Code ("",[],[]) "Then\160your\160block\160format\160will\160be",LineBreak,Code ("",[],[]) "\160\160\160\160maintained.",LineBreak,Code ("",[],[]) "",LineBreak,Code ("",[],[]) "This\160is\160good\160for\160copying\160in\160code\160blocks:",LineBreak,Code ("",[],[]) "",LineBreak,Code ("",[],[]) "def\160function():",LineBreak,Code ("",[],[]) "\160\160\160\160\"\"\"documentation\160string\"\"\"",LineBreak,Code ("",[],[]) "",LineBreak,Code ("",[],[]) "\160\160\160\160if\160True:",LineBreak,Code ("",[],[]) "\160\160\160\160\160\160\160\160print\160True",LineBreak,Code ("",[],[]) "\160\160\160\160else:",LineBreak,Code ("",[],[]) "\160\160\160\160\160\160\160\160print\160False"]]

tests/mediawiki-reader.wiki+25

@@ -193,3 +193,28 @@ ends the list.

193

194Here is some <math>x=\frac{y^\pi}{z}</math>.

195

196+== preformatted blocks ==

197+

198+ Start each line with a space.

199+ Text is '''preformatted''' and

200+ ''markups'' '''''can''''' be done.

201+

202+ hell yeah

203+

204+ <nowiki>Start with a space in the first column,

205+(before the <nowiki>).

206+

207+Then your block format will be

208+ maintained.

209+

210+This is good for copying in code blocks:

211+

212+def function():

213+ """documentation string"""

214+

215+ if True:

216+ print True

217+ else:

218+ print False</nowiki>

219+

220+