diff options
| -rw-r--r-- | pandoc.cabal | 2 | ||||
| -rw-r--r-- | src/Text/Pandoc/Readers/Docx.hs | 58 | ||||
| -rw-r--r-- | tests/Tests/Readers/Docx.hs | 4 | ||||
| -rw-r--r-- | tests/docx.already_auto_ident.docx | bin | 0 -> 8463 bytes | |||
| -rw-r--r-- | tests/docx.already_auto_ident.native | 2 |
5 files changed, 39 insertions, 27 deletions
diff --git a/pandoc.cabal b/pandoc.cabal index 052f2632a..d9132c568 100644 --- a/pandoc.cabal +++ b/pandoc.cabal @@ -188,6 +188,8 @@ Extra-Source-Files: tests/fb2.math.fb2, tests/fb2.test-small.png, tests/fb2.test.jpg, + tests/docx.already_auto_ident.native, + tests/docx.already_auto_ident.docx, tests/docx.block_quotes.docx, tests/docx.block_quotes_parse_indent.native, tests/docx.headers.docx, diff --git a/src/Text/Pandoc/Readers/Docx.hs b/src/Text/Pandoc/Readers/Docx.hs index 882e8d7d8..196a3cec5 100644 --- a/src/Text/Pandoc/Readers/Docx.hs +++ b/src/Text/Pandoc/Readers/Docx.hs @@ -105,6 +105,7 @@ readDocx opts bytes = Left _ -> error $ "couldn't parse docx file" data DState = DState { docxAnchorMap :: M.Map String String + , docxInHeaderBlock :: Bool , docxInTexSubscript :: Bool } data DEnv = DEnv { docxOptions :: ReaderOptions @@ -112,18 +113,13 @@ data DEnv = DEnv { docxOptions :: ReaderOptions type DocxContext = ReaderT DEnv (State DState) -updateDState :: (DState -> DState) -> DocxContext () -updateDState f = do - st <- get - put $ f st - -withDState :: DState -> DocxContext a -> DocxContext a -withDState ds dctx = do - ds' <- get - updateDState (\_ -> ds) - dctx' <- dctx - put ds' - return dctx' +withDState :: (DState -> DState) -> DocxContext a -> DocxContext a +withDState f dctx = do + ds <- get + modify f + ctx' <- dctx + put ds + return ctx' evalDocxContext :: DocxContext a -> DEnv -> DState -> a evalDocxContext ctx env st = evalState (runReaderT ctx env) st @@ -297,18 +293,24 @@ parPartToInlines (BookMark _ anchor) = -- We record these, so we can make sure not to overwrite -- user-defined anchor links with header auto ids. do + -- get whether we're in a header. + inHdrBool <- gets docxInHeaderBlock -- Get the anchor map. anchorMap <- gets docxAnchorMap - -- Check to see if the id is already in there. Rewrite if - -- necessary. This will have the possible effect of rewriting - -- user-defined anchor links. However, since these are not defined - -- in pandoc, it seems like a necessary evil to avoid an extra - -- pass. - let newAnchor = case anchor `elem` (M.elems anchorMap) of - True -> uniqueIdent [Str anchor] (M.elems anchorMap) - False -> anchor - updateDState $ \s -> s { docxAnchorMap = M.insert anchor newAnchor anchorMap} - return [Span (anchor, ["anchor"], []) []] + -- We don't want to rewrite if we're in a header, since we'll take + -- care of that later, when we make the header anchor. If the + -- bookmark were already in uniqueIdent form, this would lead to a + -- duplication. Otherwise, we check to see if the id is already in + -- there. Rewrite if necessary. This will have the possible effect + -- of rewriting user-defined anchor links. However, since these + -- are not defined in pandoc, it seems like a necessary evil to + -- avoid an extra pass. + let newAnchor = + if not inHdrBool && anchor `elem` (M.elems anchorMap) + then uniqueIdent [Str anchor] (M.elems anchorMap) + else anchor + modify $ \s -> s { docxAnchorMap = M.insert anchor newAnchor anchorMap} + return [Span (newAnchor, ["anchor"], []) []] parPartToInlines (Drawing fp bs) = do return $ case True of -- TODO: add self-contained images True -> [Image [] (fp, "")] @@ -427,8 +429,8 @@ oMathElemToTexString (Matrix bases) = do s <- liftM (intercalate " \\\\\n")(mapM rowString bases) return $ printf "\\begin{matrix}\n%s\n\\end{matrix}" s oMathElemToTexString (NAry style sub sup base) | Just c <- nAryChar style = do - ds <- gets (\s -> s{docxInTexSubscript = True}) - subString <- withDState ds $ concatMapM oMathElemToTexString sub + subString <- withDState (\s -> s{docxInTexSubscript = True}) $ + concatMapM oMathElemToTexString sub supString <- concatMapM oMathElemToTexString sup baseString <- baseToTexString base return $ case M.lookup c uniconvMap of @@ -497,7 +499,7 @@ makeHeaderAnchor (Header n (_, classes, kvs) ils) do hdrIDMap <- gets docxAnchorMap let newIdent = uniqueIdent ils (M.elems hdrIDMap) - updateDState $ \s -> s {docxAnchorMap = M.insert ident newIdent hdrIDMap} + modify $ \s -> s {docxAnchorMap = M.insert ident newIdent hdrIDMap} return $ Header n (newIdent, classes, kvs) (ils \\ (x:xs)) -- Otherwise we just give it a name, and register that name (associate -- it with itself.) @@ -505,7 +507,7 @@ makeHeaderAnchor (Header n (_, classes, kvs) ils) = do hdrIDMap <- gets docxAnchorMap let newIdent = uniqueIdent ils (M.elems hdrIDMap) - updateDState $ \s -> s {docxAnchorMap = M.insert newIdent newIdent hdrIDMap} + modify $ \s -> s {docxAnchorMap = M.insert newIdent newIdent hdrIDMap} return $ Header n (newIdent, classes, kvs) ils makeHeaderAnchor blk = return blk @@ -541,7 +543,8 @@ bodyPartToBlocks (Paragraph pPr parparts) [CodeBlock ("", [], []) (concatMap parPartToString parparts)] bodyPartToBlocks (Paragraph pPr parparts) | any isHeaderContainer (parStyleToContainers pPr) = do - ils <- parPartsToInlines parparts >>= (return . normalizeSpaces) + ils <-withDState (\s -> s{docxInHeaderBlock = True}) $ + parPartsToInlines parparts >>= (return . normalizeSpaces) let (Container hdrFun) = head $ filter isHeaderContainer (parStyleToContainers pPr) Header n attr _ = hdrFun [] hdr <- makeHeaderAnchor $ Header n attr ils @@ -624,6 +627,7 @@ bodyToBlocks (Body bps) = do docxToBlocks :: ReaderOptions -> Docx -> [Block] docxToBlocks opts d@(Docx (Document _ body)) = let dState = DState { docxAnchorMap = M.empty + , docxInHeaderBlock = False , docxInTexSubscript = False} dEnv = DEnv { docxOptions = opts , docxDocument = d} diff --git a/tests/Tests/Readers/Docx.hs b/tests/Tests/Readers/Docx.hs index a379bbf23..7b737f95a 100644 --- a/tests/Tests/Readers/Docx.hs +++ b/tests/Tests/Readers/Docx.hs @@ -97,6 +97,10 @@ tests = [ testGroup "inlines" "docx.headers.docx" "docx.headers.native" , testCompare + "headers already having auto identifiers" + "docx.already_auto_ident.docx" + "docx.already_auto_ident.native" + , testCompare "lists" "docx.lists.docx" "docx.lists.native" diff --git a/tests/docx.already_auto_ident.docx b/tests/docx.already_auto_ident.docx Binary files differnew file mode 100644 index 000000000..ec2b348d7 --- /dev/null +++ b/tests/docx.already_auto_ident.docx diff --git a/tests/docx.already_auto_ident.native b/tests/docx.already_auto_ident.native new file mode 100644 index 000000000..054bfe34a --- /dev/null +++ b/tests/docx.already_auto_ident.native @@ -0,0 +1,2 @@ +[Header 1 ("anchor-header",[],[]) [Str "Anchor",Space,Str "Header"] +,Para [Str "A",Space,Link [Str "link"] ("#anchor-header","")]] |