732f6ab · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

HTML reader: Use pandoc Div and Span for raw "<div>", "<span>".

732f6abJohn MacFarlane13 years ago

src/Text/Pandoc/Readers/HTML.hs+25−10

compare with master

unifiedsplit

@@ -92,6 +92,7 @@ block = choice

92 , pHead

93 , pBody

94 , pPlain

95+ , pDiv

95 , pRawHtmlBlock

96 , pRawHtmlBlock

96 ]

97 ]

@@ -177,6 +178,13 @@ pRawTag = do

177 then return []

178 then return []

178 else return $ renderTags' [tag]

179 else return $ renderTags' [tag]

179

180

181+pDiv :: TagParser [Block]

182+pDiv = try $ do

183+ getOption readerParseRaw >>= guard

184+ TagOpen _ attr <- lookAhead $ pSatisfy $ tagOpen (=="div") (const True)

185+ contents <- pInTags "div" block

186+ return [Div (mkAttr attr) contents]

187+

180pRawHtmlBlock :: TagParser [Block]

188pRawHtmlBlock :: TagParser [Block]

181pRawHtmlBlock = do

189pRawHtmlBlock = do

182 raw <- pHtmlBlock "script" <|> pHtmlBlock "style" <|> pRawTag

190 raw <- pHtmlBlock "script" <|> pHtmlBlock "style" <|> pRawTag

@@ -295,11 +303,7 @@ pCodeBlock = try $ do

295 let result = case reverse result' of

303 let result = case reverse result' of

296 '\n':_ -> init result'

304 '\n':_ -> init result'

297 _ -> result'

305 _ -> result'

298− let attribsId = fromMaybe "" $ lookup "id" attr

299− let attribsClasses = words $ fromMaybe "" $ lookup "class" attr

300− let attribsKV = filter (\(k,_) -> k /= "class" && k /= "id") attr

301− let attribs = (attribsId, attribsClasses, attribsKV)

302− return [CodeBlock attribs result]

306+ return [CodeBlock (mkAttr attr) result]

303

307

304inline :: TagParser [Inline]

308inline :: TagParser [Inline]

305inline = choice

309inline = choice

@@ -314,6 +318,7 @@ inline = choice

314 , pLink

318 , pLink

315 , pImage

319 , pImage

316 , pCode

320 , pCode

321+ , pSpan

317 , pRawHtmlInline

322 , pRawHtmlInline

318 ]

323 ]

319

324

@@ -397,11 +402,14 @@ pCode :: TagParser [Inline]

397pCode = try $ do

402pCode = try $ do

398 (TagOpen open attr) <- pSatisfy $ tagOpen (`elem` ["code","tt"]) (const True)

403 (TagOpen open attr) <- pSatisfy $ tagOpen (`elem` ["code","tt"]) (const True)

399 result <- manyTill pAnyTag (pCloses open)

404 result <- manyTill pAnyTag (pCloses open)

400− let ident = fromMaybe "" $ lookup "id" attr

401− let classes = words $ fromMaybe [] $ lookup "class" attr

402− let rest = filter (\(x,_) -> x /= "id" && x /= "class") attr

403− return [Code (ident,classes,rest)

404− $ intercalate " " $ lines $ innerText result]

405+ return [Code (mkAttr attr) $ intercalate " " $ lines $ innerText result]

406+

407+pSpan :: TagParser [Inline]

408+pSpan = try $ do

409+ getOption readerParseRaw >>= guard

410+ TagOpen _ attr <- lookAhead $ pSatisfy $ tagOpen (=="span") (const True)

411+ contents <- pInTags "span" inline

412+ return [Span (mkAttr attr) contents]

405

413

406pRawHtmlInline :: TagParser [Inline]

414pRawHtmlInline :: TagParser [Inline]

407pRawHtmlInline = do

415pRawHtmlInline = do

@@ -648,3 +656,10 @@ htmlTag f = try $ do

648 _ -> do

656 _ -> do

649 rendered <- manyTill anyChar (char '>')

657 rendered <- manyTill anyChar (char '>')

650 return (next, rendered ++ ">")

658 return (next, rendered ++ ">")

659+

660+mkAttr :: [(String, String)] -> Attr

661+mkAttr attr = (attribsId, attribsClasses, attribsKV)

662+ where attribsId = fromMaybe "" $ lookup "id" attr

663+ attribsClasses = words $ fromMaybe "" $ lookup "class" attr

664+ attribsKV = filter (\(k,_) -> k /= "class" && k /= "id") attr

665+