eda5540 · typeable/pandoc

git.iko.soy/typeable/pandoc

DOCX reader: Allow empty dates in comments and tracked changes (#6726)

For security reasons, some legal firms delete the date from comments and tracked changes. * Make date optional (Maybe) in tracked changes and comments datatypes * Add tests

commiteda5540719e9771b48f15aa7f431033163e1f161 authorDiego Balseiro <dbalseiro@stackbuilders.com> dateTue 6 Oct 2020, 23:03 (6 years ago) parenta27a0b5

8 files changed+42−23

compare with master

unifiedsplit

src/Text/Pandoc/Readers/Docx.hs+10−5 src/Text/Pandoc/Readers/Docx/Parse.hs+7−7 src/Text/Pandoc/Writers/Docx.hs+7−11 test/Tests/Readers/Docx.hs+4 test/Tests/Writers/Docx.hs+5 test/docx/golden/track_changes_scrubbed_metadata.docx test/docx/track_changes_scrubbed_metadata.docx test/docx/track_changes_scrubbed_metadata.native+9

src/Text/Pandoc/Readers/Docx.hs+10−5

@@ -365,7 +365,7 @@ parPartToInlines' (ChangedRuns (TrackedChange Insertion (ChangeInfo _ author dat

365365 RejectChanges -> return mempty

366366 AllChanges -> do

367367 ils <- smushInlines <$> mapM runToInlines runs

368− let attr = ("", ["insertion"], [("author", author), ("date", date)])

368+ let attr = ("", ["insertion"], addAuthorAndDate author date)

369369 return $ spanWith attr ils

370370parPartToInlines' (ChangedRuns (TrackedChange Deletion (ChangeInfo _ author date)) runs) = do

371371 opts <- asks docxOptions

@@ -374,7 +374,7 @@ parPartToInlines' (ChangedRuns (TrackedChange Deletion (ChangeInfo _ author date

374374 RejectChanges -> smushInlines <$> mapM runToInlines runs

375375 AllChanges -> do

376376 ils <- smushInlines <$> mapM runToInlines runs

377− let attr = ("", ["deletion"], [("author", author), ("date", date)])

377+ let attr = ("", ["deletion"], addAuthorAndDate author date)

378378 return $ spanWith attr ils

379379parPartToInlines' (CommentStart cmtId author date bodyParts) = do

380380 opts <- asks docxOptions

@@ -382,7 +382,7 @@ parPartToInlines' (CommentStart cmtId author date bodyParts) = do

382382 AllChanges -> do

383383 blks <- smushBlocks <$> mapM bodyPartToBlocks bodyParts

384384 ils <- blocksToInlinesWarn cmtId blks

385− let attr = ("", ["comment-start"], [("id", cmtId), ("author", author), ("date", date)])

385+ let attr = ("", ["comment-start"], ("id", cmtId) : addAuthorAndDate author date)

386386 return $ spanWith attr ils

387387 _ -> return mempty

388388parPartToInlines' (CommentEnd cmtId) = do

@@ -593,7 +593,7 @@ bodyPartToBlocks (Paragraph pPr parparts)

593593 return mempty

594594 (Just (TrackedChange Insertion (ChangeInfo _ cAuthor cDate))

595595 , AllChanges) -> do

596− let attr = ("", ["paragraph-insertion"], [("author", cAuthor), ("date", cDate)])

596+ let attr = ("", ["paragraph-insertion"], addAuthorAndDate cAuthor cDate)

597597 insertMark = spanWith attr mempty

598598 transform <- parStyleToTransform pPr'

599599 return $ transform $

@@ -605,7 +605,7 @@ bodyPartToBlocks (Paragraph pPr parparts)

605605 handleInsertion

606606 (Just (TrackedChange Deletion (ChangeInfo _ cAuthor cDate))

607607 , AllChanges) -> do

608− let attr = ("", ["paragraph-deletion"], [("author", cAuthor), ("date", cDate)])

608+ let attr = ("", ["paragraph-deletion"], addAuthorAndDate cAuthor cDate)

609609 insertMark = spanWith attr mempty

610610 transform <- parStyleToTransform pPr'

611611 return $ transform $

@@ -732,3 +732,8 @@ docxToOutput :: PandocMonad m

732732docxToOutput opts (Docx (Document _ body)) =

733733 let dEnv = def { docxOptions = opts} in

734734 evalDocxContext (bodyToOutput body) dEnv def

735+

736+addAuthorAndDate :: T.Text -> Maybe T.Text -> [(T.Text, T.Text)]

737+addAuthorAndDate author mdate =

738+ ("author", author) : maybe [] (\date -> [("date", date)]) mdate

739+

src/Text/Pandoc/Readers/Docx/Parse.hs+7−7

@@ -213,7 +213,7 @@ data ParIndentation = ParIndentation { leftParIndent :: Maybe Integer

213213data ChangeType = Insertion | Deletion

214214 deriving Show

215215

216−data ChangeInfo = ChangeInfo ChangeId Author ChangeDate

216+data ChangeInfo = ChangeInfo ChangeId Author (Maybe ChangeDate)

217217 deriving Show

218218

219219data TrackedChange = TrackedChange ChangeType ChangeInfo

@@ -276,7 +276,7 @@ type Extent = Maybe (Double, Double)

276276

277277data ParPart = PlainRun Run

278278 | ChangedRuns TrackedChange [Run]

279− | CommentStart CommentId Author CommentDate [BodyPart]

279+ | CommentStart CommentId Author (Maybe CommentDate) [BodyPart]

280280 | CommentEnd CommentId

281281 | BookMark BookMarkId Anchor

282282 | InternalHyperLink Anchor [Run]

@@ -852,7 +852,7 @@ elemToCommentStart ns element

852852 | isElem ns "w" "comment" element

853853 , Just cmtId <- findAttrTextByName ns "w" "id" element

854854 , Just cmtAuthor <- findAttrTextByName ns "w" "author" element

855− , Just cmtDate <- findAttrTextByName ns "w" "date" element = do

855+ , cmtDate <- findAttrTextByName ns "w" "date" element = do

856856 bps <- mapD (elemToBodyPart ns) (elChildren element)

857857 return $ CommentStart cmtId cmtAuthor cmtDate bps

858858elemToCommentStart _ _ = throwError WrongElem

@@ -958,14 +958,14 @@ getTrackedChange ns element

958958 | isElem ns "w" "ins" element || isElem ns "w" "moveTo" element

959959 , Just cId <- findAttrTextByName ns "w" "id" element

960960 , Just cAuthor <- findAttrTextByName ns "w" "author" element

961− , Just cDate <- findAttrTextByName ns "w" "date" element =

962− Just $ TrackedChange Insertion (ChangeInfo cId cAuthor cDate)

961+ , mcDate <- findAttrTextByName ns "w" "date" element =

962+ Just $ TrackedChange Insertion (ChangeInfo cId cAuthor mcDate)

963963getTrackedChange ns element

964964 | isElem ns "w" "del" element || isElem ns "w" "moveFrom" element

965965 , Just cId <- findAttrTextByName ns "w" "id" element

966966 , Just cAuthor <- findAttrTextByName ns "w" "author" element

967− , Just cDate <- findAttrTextByName ns "w" "date" element =

968− Just $ TrackedChange Deletion (ChangeInfo cId cAuthor cDate)

967+ , mcDate <- findAttrTextByName ns "w" "date" element =

968+ Just $ TrackedChange Deletion (ChangeInfo cId cAuthor mcDate)

969969getTrackedChange _ _ = Nothing

970970

971971elemToParagraphStyle :: NameSpaces -> Element -> ParStyleMap -> ParagraphStyle

src/Text/Pandoc/Writers/Docx.hs+7−11

@@ -1244,33 +1244,29 @@ inlineToOpenXML' opts (Span (ident,classes,kvs) ils) = do

12441244 else id)

12451245 getChangeAuthorDate = do

12461246 defaultAuthor <- asks envChangesAuthor

1247− defaultDate <- asks envChangesDate

12481247 let author = fromMaybe defaultAuthor (lookup "author" kvs)

1249− date = fromMaybe defaultDate (lookup "date" kvs)

1250− return (author, date)

1248+ let mdate = lookup "date" kvs

1249+ return $ ("w:author", T.unpack author) :

1250+ maybe [] (\date -> [("w:date", T.unpack date)]) mdate

12511251 insmod <- if "insertion" `elem` classes

12521252 then do

1253− (author, date) <- getChangeAuthorDate

1253+ changeAuthorDate <- getChangeAuthorDate

12541254 insId <- gets stInsId

12551255 modify $ \s -> s{stInsId = insId + 1}

12561256 return $ \f -> do

12571257 x <- f

12581258 return [ mknode "w:ins"

1259− [("w:id", show insId),

1260− ("w:author", T.unpack author),

1261− ("w:date", T.unpack date)] x ]

1259+ (("w:id", show insId) : changeAuthorDate) x]

12621260 else return id

12631261 delmod <- if "deletion" `elem` classes

12641262 then do

1265− (author, date) <- getChangeAuthorDate

1263+ changeAuthorDate <- getChangeAuthorDate

12661264 delId <- gets stDelId

12671265 modify $ \s -> s{stDelId = delId + 1}

12681266 return $ \f -> local (\env->env{envInDel=True}) $ do

12691267 x <- f

12701268 return [mknode "w:del"

1271− [("w:id", show delId),

1272− ("w:author", T.unpack author),

1273− ("w:date", T.unpack date)] x]

1269+ (("w:id", show delId) : changeAuthorDate) x]

12741270 else return id

12751271 contents <- insmod $ delmod $ dirmod $ stylemod $ pmod

12761272 $ inlinesToOpenXML opts ils

test/Tests/Readers/Docx.hs+4

@@ -405,6 +405,10 @@ tests = [ testGroup "document"

405405 "paragraph insertion/deletion (all)"

406406 "docx/paragraph_insertion_deletion.docx"

407407 "docx/paragraph_insertion_deletion_all.native"

408+ , testCompareWithOpts def{readerTrackChanges=AllChanges}

409+ "paragraph insertion/deletion (all)"

410+ "docx/track_changes_scrubbed_metadata.docx"

411+ "docx/track_changes_scrubbed_metadata.native"

408412 , testForWarningsWithOpts def{readerTrackChanges=AcceptChanges}

409413 "comment warnings (accept -- no warnings)"

410414 "docx/comments_warning.docx"

test/Tests/Writers/Docx.hs+5

@@ -150,6 +150,11 @@ tests = [ testGroup "inlines"

150150 def

151151 "docx/comments.native"

152152 "docx/golden/comments.docx"

153+ , docxTest

154+ "scrubbed metadata"

155+ def

156+ "docx/track_changes_scrubbed_metadata.native"

157+ "docx/golden/track_changes_scrubbed_metadata.docx"

153158 ]

154159 , testGroup "custom styles"

155160 [ docxTest "custom styles without reference.docx"

test/docx/golden/track_changes_scrubbed_metadata.docxnew file

Binary file not shown.

test/docx/track_changes_scrubbed_metadata.docxnew file

Binary file not shown.

test/docx/track_changes_scrubbed_metadata.nativenew file+9

@@ -0,0 +1,9 @@

1+[Para [ Str "Here", Space, Str "is", Space, Str "a", Space

2+ , Span ("",["deletion"],[("author","Author")]) [Str "dummy"]

3+ , Span ("",["insertion"],[("author","Author")]) [Str "test"]

4+ , Space

5+ , Span ("",["comment-start"],[("id","3"),("author","Author")])

6+ [Str "With",Space,Str "a",Space,Str "comment!"]

7+ , Str "document",Span ("",["comment-end"],[("id","3")]) [],Str "."

8+ ]

9+]