48b23d4 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

MediaWiki reader: Correctly handle indented preformatted text

48b23d4John MacFarlane14 years ago

src/Text/Pandoc/Readers/MediaWiki.hs+17−5

compare with master

unifiedsplit

@@ -50,7 +50,8 @@ import Control.Monad

50import Data.List (intersperse, intercalate, isPrefixOf )

51import Text.HTML.TagSoup

52import Data.Sequence (viewl, ViewL(..), (<|))

53+import qualified Data.Foldable as F

53−import Data.Char (isDigit)

54+import Data.Char (isDigit, isSpace)

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

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

56readMediaWiki :: ReaderOptions -- ^ Reader options

57readMediaWiki :: ReaderOptions -- ^ Reader options

@@ -175,7 +176,11 @@ block = mempty <$ skipMany1 blankline

175 <|> para

176 <|> para

176

177

177para :: MWParser Blocks

178para :: MWParser Blocks

179+para = do

178−para = B.para . trimInlines . mconcat <$> many1 inline

180+ contents <- trimInlines . mconcat <$> many1 inline

181+ if F.all (==Space) contents

182+ then return mempty

183+ else return $ B.para contents

179

184

180table :: MWParser Blocks

185table :: MWParser Blocks

181table = do

186table = do

@@ -330,10 +335,16 @@ preformatted = try $ do

330 lines . fromEntities . map spToNbsp <$> try

335 lines . fromEntities . map spToNbsp <$> try

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

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

332 manyTill anyChar (htmlTag (~== TagClose "nowiki")))

337 manyTill anyChar (htmlTag (~== TagClose "nowiki")))

333− let inline' = whitespace' <|> endline' <|> nowiki' <|> inline

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

339+ <|> (notFollowedBy newline *> inline)

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

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

335 strToCode x = x

341 strToCode x = x

336− B.para . bottomUp strToCode . mconcat <$> many1 inline'

342+ contents <- mconcat <$> many1 inline'

343+ let spacesStr (Str xs) = all isSpace xs

344+ spacesStr _ = False

345+ if F.all spacesStr contents

346+ then return mempty

347+ else return $ B.para $ bottomUp strToCode contents

337

348

338header :: MWParser Blocks

349header :: MWParser Blocks

339header = try $ do

350header = try $ do

@@ -504,7 +515,8 @@ whitespace = B.space <$ (skipMany1 spaceChar <|> endline <|> htmlComment)

504

515

505endline :: MWParser ()

516endline :: MWParser ()

506endline = () <$ try (newline <*

517endline = () <$ try (newline <*

507− notFollowedBy blankline <*

518+ notFollowedBy spaceChar <*

519+ notFollowedBy newline <*

508 notFollowedBy' hrule <*

520 notFollowedBy' hrule <*

509 notFollowedBy tableStart <*

521 notFollowedBy tableStart <*

510 notFollowedBy' header <*

522 notFollowedBy' header <*

tests/mediawiki-reader.native+3

@@ -170,6 +170,9 @@ Pandoc (Meta {docTitle = [], docAuthors = [], docDate = []})

170,Para [Str "Not"]

171,RawBlock "html" "<hr/>"

172,Para [Str "preformatted"]

173+,Para [Str "Don't",Space,Str "need"]

174+,Para [Code ("",[],[]) "a\160blank\160line"]

175+,Para [Str "around",Space,Str "a",Space,Str "preformatted",Space,Str "block."]

173,Header 2 ("",[],[]) [Str "templates"]

176,Header 2 ("",[],[]) [Str "templates"]

174,RawBlock "mediawiki" "{{Welcome}}"

177,RawBlock "mediawiki" "{{Welcome}}"

175,RawBlock "mediawiki" "{{Foo:Bar}}"

178,RawBlock "mediawiki" "{{Foo:Bar}}"

tests/mediawiki-reader.wiki+4

@@ -274,6 +274,10 @@ def function():

274

275Not<hr/> preformatted

276

277+Don't need

278+ a blank line

279+around a preformatted block.

280+

277== templates ==

281== templates ==

278

282

279{{Welcome}}

283{{Welcome}}