2746f73 · typeable/pandoc

git.iko.soy/typeable/pandoc

Docx reader: Parse track changes info into paragraph props.

This will tell us whether a paragraph break was inserted or deleted. We add a generalized track-changes parsing function, and use it in `elemToParPart` as well.

commit2746f73093dcf126977ae3cfa89dda1297d0dcec authorJesse Rosenthal <jrosenthal@jhu.edu> dateTue 2 Jan 2018, 08:09 (9 years ago) parentf6fde0a

1 file changed+27−15

compare with master

unifiedsplit

src/Text/Pandoc/Readers/Docx/Parse.hs+27−15

src/Text/Pandoc/Readers/Docx/Parse.hs+27−15

@@ -216,6 +216,7 @@ data ParagraphStyle = ParagraphStyle { pStyle :: [String]

216216 , pHeading :: Maybe (String, Int)

217217 , pNumInfo :: Maybe (String, String)

218218 , pBlockQuote :: Maybe Bool

219+ , pChange :: Maybe TrackedChange

219220 }

220221 deriving Show

221222

@@ -226,6 +227,7 @@ defaultParagraphStyle = ParagraphStyle { pStyle = []

226227 , pHeading = Nothing

227228 , pNumInfo = Nothing

228229 , pBlockQuote = Nothing

230+ , pChange = Nothing

229231 }

230232

231233

@@ -738,21 +740,9 @@ elemToParPart ns element

738740 | isElem ns "w" "r" element =

739741 elemToRun ns element >>= (\r -> return $ PlainRun r)

740742elemToParPart ns element

741− | isElem ns "w" "ins" element || isElem ns "w" "moveTo" element

742− , Just cId <- findAttrByName ns "w" "id" element

743− , Just cAuthor <- findAttrByName ns "w" "author" element

744− , Just cDate <- findAttrByName ns "w" "date" element = do

745− runs <- mapD (elemToRun ns) (elChildren element)

746− return $ ChangedRuns

747− (TrackedChange Insertion (ChangeInfo cId cAuthor cDate)) runs

748−elemToParPart ns element

749− | isElem ns "w" "del" element || isElem ns "w" "moveFrom" element

750− , Just cId <- findAttrByName ns "w" "id" element

751− , Just cAuthor <- findAttrByName ns "w" "author" element

752− , Just cDate <- findAttrByName ns "w" "date" element = do

753− runs <- mapD (elemToRun ns) (elChildren element)

754− return $ ChangedRuns

755− (TrackedChange Deletion (ChangeInfo cId cAuthor cDate)) runs

743+ | Just change <- getTrackedChange ns element = do

744+ runs <- mapD (elemToRun ns) (elChildren element)

745+ return $ ChangedRuns change runs

756746elemToParPart ns element

757747 | isElem ns "w" "smartTag" element = do

758748 runs <- mapD (elemToRun ns) (elChildren element)

@@ -903,6 +893,21 @@ getParStyleField field stylemap styles

903893 = Just y

904894getParStyleField _ _ _ = Nothing

905895

896+getTrackedChange :: NameSpaces -> Element -> Maybe TrackedChange

897+getTrackedChange ns element

898+ | isElem ns "w" "ins" element || isElem ns "w" "moveTo" element

899+ , Just cId <- findAttrByName ns "w" "id" element

900+ , Just cAuthor <- findAttrByName ns "w" "author" element

901+ , Just cDate <- findAttrByName ns "w" "date" element =

902+ Just $ TrackedChange Insertion (ChangeInfo cId cAuthor cDate)

903+getTrackedChange ns element

904+ | isElem ns "w" "del" element || isElem ns "w" "moveFrom" element

905+ , Just cId <- findAttrByName ns "w" "id" element

906+ , Just cAuthor <- findAttrByName ns "w" "author" element

907+ , Just cDate <- findAttrByName ns "w" "date" element =

908+ Just $ TrackedChange Deletion (ChangeInfo cId cAuthor cDate)

909+getTrackedChange _ _ = Nothing

910+

906911elemToParagraphStyle :: NameSpaces -> Element -> ParStyleMap -> ParagraphStyle

907912elemToParagraphStyle ns element sty

908913 | Just pPr <- findChildByName ns "w" "pPr" element =

@@ -926,6 +931,13 @@ elemToParagraphStyle ns element sty

926931 , pHeading = getParStyleField headingLev sty style

927932 , pNumInfo = getParStyleField numInfo sty style

928933 , pBlockQuote = getParStyleField isBlockQuote sty style

934+ , pChange = findChildByName ns "w" "rPr" pPr >>=

935+ filterChild (\e -> isElem ns "w" "ins" e ||

936+ isElem ns "w" "moveTo" e ||

937+ isElem ns "w" "del" e ||

938+ isElem ns "w" "moveFrom" e

939+ ) >>=

940+ getTrackedChange ns

929941 }

930942elemToParagraphStyle _ _ _ = defaultParagraphStyle

931943