dce360e · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

Docx Reader: Introduce link rewriting.

dce360eJesse Rosenthal12 years ago

src/Text/Pandoc/Readers/Docx.hs+30−7

compare with master

unifiedsplit

@@ -104,7 +104,7 @@ readDocx opts bytes =

104 Nothing -> error $ "couldn't parse docx file"

105

106

107−data DState = DState { docxHeaderAnchors :: M.Map String String }

107+data DState = DState { docxAnchorMap :: M.Map String String }

108

109data DEnv = DEnv { docxOptions :: ReaderOptions

110 , docxDocument :: Docx}

@@ -276,7 +276,22 @@ parPartToInlines (Deletion _ author date runs) = do

276 ("", ["deletion"], [("author", author), ("date", date)])

277 ils]

278parPartToInlines (BookMark _ anchor) | anchor `elem` dummyAnchors = return []

279−parPartToInlines (BookMark _ anchor) = return [Span (anchor, ["anchor"], []) []]

279+parPartToInlines (BookMark _ anchor) =

280+ -- We record these, so we can make sure not to overwrite

281+ -- user-defined anchor links with header auto ids.

282+ do

283+ -- Get the anchor map.

284+ anchorMap <- gets docxAnchorMap

285+ -- Check to see if the id is already in there. Rewrite if

286+ -- necessary. This will have the possible effect of rewriting

287+ -- user-defined anchor links. However, since these are not defined

288+ -- in pandoc, it seems like a necessary evil to avoid an extra

289+ -- pass.

290+ let newAnchor = case anchor `elem` (M.elems anchorMap) of

291+ True -> uniqueIdent [Str anchor] (M.elems anchorMap)

292+ False -> anchor

293+ put DState{ docxAnchorMap = M.insert anchor newAnchor anchorMap}

294+ return [Span (anchor, ["anchor"], []) []]

280parPartToInlines (Drawing relid) = do

295parPartToInlines (Drawing relid) = do

281 (Docx _ _ _ rels _) <- asks docxDocument

296 (Docx _ _ _ rels _) <- asks docxDocument

282 return $ case lookupRelationship relid rels of

297 return $ case lookupRelationship relid rels of

@@ -311,9 +326,9 @@ makeHeaderAnchor (Header n (_, classes, kvs) ils)

311 , (Span (ident, _, _) _) <- x

326 , (Span (ident, _, _) _) <- x

312 , notElem ident dummyAnchors =

327 , notElem ident dummyAnchors =

313 do

328 do

314− hdrIDMap <- gets docxHeaderAnchors

329+ hdrIDMap <- gets docxAnchorMap

315 let newIdent = uniqueIdent ils (M.elems hdrIDMap)

330 let newIdent = uniqueIdent ils (M.elems hdrIDMap)

316− put DState{docxHeaderAnchors = M.insert ident newIdent hdrIDMap}

331+ put DState{docxAnchorMap = M.insert ident newIdent hdrIDMap}

317 return $ Header n (newIdent, classes, kvs) (ils \\ (x:xs))

332 return $ Header n (newIdent, classes, kvs) (ils \\ (x:xs))

318makeHeaderAnchor blk = return blk

333makeHeaderAnchor blk = return blk

319

334

@@ -411,6 +426,14 @@ bodyPartToBlocks (Tbl cap _ look (r:rs)) = do

411

426

412 return [Table caption alignments widths hdrCells cells]

427 return [Table caption alignments widths hdrCells cells]

413

428

429+-- replace targets with generated anchors.

430+rewriteLink :: Inline -> DocxContext Inline

431+rewriteLink l@(Link ils ('#':target, title)) = do

432+ anchorMap <- gets docxAnchorMap

433+ return $ case M.lookup target anchorMap of

434+ Just newTarget -> (Link ils ('#':newTarget, title))

435+ Nothing -> l

436+rewriteLink il = return il

414

437

415makeImagesSelfContained :: Inline -> DocxContext Inline

438makeImagesSelfContained :: Inline -> DocxContext Inline

416makeImagesSelfContained i@(Image alt (uri, title)) = do

439makeImagesSelfContained i@(Image alt (uri, title)) = do

@@ -429,14 +452,15 @@ makeImagesSelfContained inline = return inline

429

452

430bodyToBlocks :: Body -> DocxContext [Block]

453bodyToBlocks :: Body -> DocxContext [Block]

431bodyToBlocks (Body bps) = do

454bodyToBlocks (Body bps) = do

432− blks <- concatMapM bodyPartToBlocks bps

455+ blks <- concatMapM bodyPartToBlocks bps >>=

456+ walkM rewriteLink

433 return $

457 return $

434 blocksToDefinitions $

458 blocksToDefinitions $

435 blocksToBullets $ blks

459 blocksToBullets $ blks

436

460

437docxToBlocks :: ReaderOptions -> Docx -> [Block]

461docxToBlocks :: ReaderOptions -> Docx -> [Block]

438docxToBlocks opts d@(Docx (Document _ body) _ _ _ _) =

462docxToBlocks opts d@(Docx (Document _ body) _ _ _ _) =

439− let dState = DState { docxHeaderAnchors = M.empty }

463+ let dState = DState { docxAnchorMap = M.empty }

440 dEnv = DEnv { docxOptions = opts

464 dEnv = DEnv { docxOptions = opts

441 , docxDocument = d}

465 , docxDocument = d}

442 in

466 in

@@ -447,7 +471,6 @@ ilToCode (Str s) = s

447ilToCode Space = " "

471ilToCode Space = " "

448ilToCode _ = ""

472ilToCode _ = ""

449

473

450−

451isHeaderClass :: String -> Maybe Int

474isHeaderClass :: String -> Maybe Int

452isHeaderClass s | "Heading" `isPrefixOf` s =

475isHeaderClass s | "Heading" `isPrefixOf` s =

453 case reads (drop (length "Heading") s) :: [(Int, String)] of

476 case reads (drop (length "Heading") s) :: [(Int, String)] of