0a2fb20 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

MediaWiki reader: Interpret width attribute on table columns.

0a2fb20John MacFarlane14 years ago

src/Text/Pandoc/Readers/MediaWiki.hs+33−14

compare with master

unifiedsplit

@@ -30,10 +30,8 @@ Conversion of mediawiki text to 'Pandoc' document.

30-}

31{-

32TODO:

33−_ tables - cell alignment and width

34_ wikipedia {{cite}} tags

33_ wikipedia {{cite}} tags

35_ <references /> {{Reflist}}

34_ <references /> {{Reflist}}

36−_ calculate cell widths when not given??? see html? latex? reader

37_ support tables http://www.mediawiki.org/wiki/Help:Tables

35_ support tables http://www.mediawiki.org/wiki/Help:Tables

38- footnotes?

36- footnotes?

39-}

37-}

@@ -54,6 +52,7 @@ import Control.Monad

54import Data.List (intersperse, intercalate, isPrefixOf )

52import Data.List (intersperse, intercalate, isPrefixOf )

55import Text.HTML.TagSoup

53import Text.HTML.TagSoup

56import Data.Sequence (viewl, ViewL(..), (<|))

54import Data.Sequence (viewl, ViewL(..), (<|))

55+import Data.Char (isDigit)

58-- | Read mediawiki from an input string and return a Pandoc document.

57-- | Read mediawiki from an input string and return a Pandoc document.

59readMediaWiki :: ReaderOptions -- ^ Reader options

58readMediaWiki :: ReaderOptions -- ^ Reader options

@@ -185,20 +184,27 @@ table :: MWParser Blocks

185table = do

184table = do

186 tableStart

185 tableStart

187 styles <- manyTill anyChar newline

186 styles <- manyTill anyChar newline

188− let tableWidth = maybe (1.0 :: Double) read

189− $ lookup "width" $ parseAttrs styles

187+ let tableWidth = case lookup "width" $ parseAttrs styles of

188+ Just w -> maybe 1.0 id $ parseWidth w

189+ Nothing -> 1.0

190 caption <- option mempty tableCaption

191 optional rowsep

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

193− (aligns,hdr) <- unzip <$> tableRow

193+ (cellspecs',hdr) <- unzip <$> tableRow

194+ let widths = map ((tableWidth *) . snd) cellspecs'

195+ let restwidth = tableWidth - sum widths

196+ let zerocols = length $ filter (==0.0) widths

197+ let defaultwidth = if zerocols == 0 || zerocols == length widths

198+ then 0.0

199+ else restwidth / fromIntegral zerocols

200+ let widths' = map (\w -> if w == 0 then defaultwidth else w) widths

201+ let cellspecs = zip (map fst cellspecs') widths'

194 rows' <- many $ try $ rowsep *> (map snd <$> tableRow)

202 rows' <- many $ try $ rowsep *> (map snd <$> tableRow)

195 tableEnd

203 tableEnd

196− -- TODO handle cellspecs from styles and aligns...

197 let cols = length hdr

204 let cols = length hdr

198 let (headers,rows) = if hasheader

205 let (headers,rows) = if hasheader

199 then (hdr, rows')

206 then (hdr, rows')

200 else (replicate cols mempty, hdr:rows')

207 else (replicate cols mempty, hdr:rows')

201− let cellspecs = zip aligns (repeat 0.0)

202 return $ B.table caption cellspecs headers rows

208 return $ B.table caption cellspecs headers rows

203

209

204parseAttrs :: String -> [(String,String)]

210parseAttrs :: String -> [(String,String)]

@@ -224,11 +230,15 @@ tableEnd = try $ guardColumnOne *> sym "|}" <* blanklines

224rowsep :: MWParser ()

230rowsep :: MWParser ()

225rowsep = try $ guardColumnOne *> sym "|-" <* blanklines

231rowsep = try $ guardColumnOne *> sym "|-" <* blanklines

226

232

227−-- TODO add something like 'guard inTable' since this is used in endline

228cellsep :: MWParser ()

233cellsep :: MWParser ()

234+cellsep = try $

229−cellsep = (try $ guardColumnOne <*

235+ (guardColumnOne <*

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

236+ ( (char '|' <* notFollowedBy (oneOf "-}+"))

237+ <|> (char '!')

238+ )

239+ )

231− <|> (() <$ try (string "||"))

240+ <|> (() <$ try (string "||"))

241+ <|> (() <$ try (string "!!"))

232

242

233tableCaption :: MWParser Inlines

243tableCaption :: MWParser Inlines

234tableCaption = try $ do

244tableCaption = try $ do

@@ -238,10 +248,10 @@ tableCaption = try $ do

238 res <- manyTill anyChar newline >>= parseFromString (many inline)

248 res <- manyTill anyChar newline >>= parseFromString (many inline)

239 return $ trimInlines $ mconcat res

249 return $ trimInlines $ mconcat res

240

250

241−tableRow :: MWParser [(Alignment, Blocks)]

251+tableRow :: MWParser [((Alignment, Double), Blocks)]

242tableRow = try $ many tableCell

252tableRow = try $ many tableCell

243

253

244−tableCell :: MWParser (Alignment, Blocks)

254+tableCell :: MWParser ((Alignment, Double), Blocks)

245tableCell = try $ do

255tableCell = try $ do

246 cellsep

256 cellsep

247 skipMany spaceChar

257 skipMany spaceChar

@@ -255,7 +265,16 @@ tableCell = try $ do

255 Just "right" -> AlignRight

265 Just "right" -> AlignRight

256 Just "center" -> AlignCenter

266 Just "center" -> AlignCenter

257 _ -> AlignDefault

267 _ -> AlignDefault

268+ let width = case lookup "width" attrs of

269+ Just xs -> maybe 0.0 id $ parseWidth xs

270+ Nothing -> 0.0

258− return (align, bs)

271+ return ((align, width), bs)

272+

273+parseWidth :: String -> Maybe Double

274+parseWidth s =

275+ case reverse s of

276+ ('%':ds) | all isDigit ds -> safeRead ('0':'.':reverse ds)

277+ _ -> Nothing

259

278

260template :: MWParser String

279template :: MWParser String

261template = try $ do

280template = try $ do

tests/mediawiki-reader.native+1−1

@@ -214,7 +214,7 @@ Pandoc (Meta {docTitle = [], docAuthors = [], docDate = []})

214 ,[[Para [Str "Butter"]]

215 ,[Para [Str "Ice",Space,Str "cream"]]

216 ,[Para [Str "and",Space,Str "more"]]]]

217−,Table [] [AlignLeft,AlignRight,AlignCenter] [0.0,0.0,0.0]

217+,Table [] [AlignLeft,AlignRight,AlignCenter] [0.25,0.125,0.125]

218 [[Para [Str "Left"]]

219 ,[Para [Str "Right"]]

220 ,[Para [Str "Center"]]]

tests/mediawiki-reader.wiki+2−2

@@ -330,8 +330,8 @@ and cheese

330| Butter || Ice cream || and more

331|}

332

333−{|

333+{|width="50%"

334−! align="left"| Left

334+! align="left" width="50%"| Left

335! align="right"|Right

336! align="center"|Center

337|-