37931cb · typeable/pandoc

git.iko.soy/typeable/pandoc

Docx reader: image attributes

commit37931cb0c56c8ae770f696a31f7beba1d6cb0f08 authormb21 <mb21@users.noreply.github.com> dateSun 6 Dec 2015, 00:41 (11 years ago) parent2060f5f

6 files changed+38−16

compare with master

unifiedsplit

src/Text/Pandoc/Readers/Docx.hs+11−4 src/Text/Pandoc/Readers/Docx/Parse.hs+21−6 tests/docx/image_no_embed.native+1−1 tests/docx/image_no_embed_writer.native+1−1 tests/docx/inline_images.native+2−2 tests/docx/inline_images_writer.native+2−2

src/Text/Pandoc/Readers/Docx.hs+11−4

@@ -298,10 +298,17 @@ runToInlines (Footnote bps) = do

298298runToInlines (Endnote bps) = do

299299 blksList <- concatReduce <$> (mapM bodyPartToBlocks bps)

300300 return $ note blksList

301−runToInlines (InlineDrawing fp bs) = do

301+runToInlines (InlineDrawing fp bs ext) = do

302302 mediaBag <- gets docxMediaBag

303303 modify $ \s -> s { docxMediaBag = insertMedia fp Nothing bs mediaBag }

304− return $ image fp "" ""

304+ return $ imageWith (extentToAttr ext) fp "" ""

305+

306+extentToAttr :: Extent -> Attr

307+extentToAttr (Just (w, h)) =

308+ ("", [], [("width", showDim w), ("height", showDim h)] )

309+ where

310+ showDim d = show (d / 914400) ++ "in"

311+extentToAttr _ = nullAttr

305312

306313parPartToInlines :: ParPart -> DocxContext Inlines

307314parPartToInlines (PlainRun r) = runToInlines r

@@ -348,10 +355,10 @@ parPartToInlines (BookMark _ anchor) =

348355 unless inHdrBool

349356 (modify $ \s -> s { docxAnchorMap = M.insert anchor newAnchor anchorMap})

350357 return $ spanWith (newAnchor, ["anchor"], []) mempty

351−parPartToInlines (Drawing fp bs) = do

358+parPartToInlines (Drawing fp bs ext) = do

352359 mediaBag <- gets docxMediaBag

353360 modify $ \s -> s { docxMediaBag = insertMedia fp Nothing bs mediaBag }

354− return $ image fp "" ""

361+ return $ imageWith (extentToAttr ext) fp "" ""

355362parPartToInlines (InternalHyperLink anchor runs) = do

356363 ils <- concatReduce <$> mapM runToInlines runs

357364 return $ link ('#' : anchor) "" ils

src/Text/Pandoc/Readers/Docx/Parse.hs+21−6

@@ -35,6 +35,7 @@ module Text.Pandoc.Readers.Docx.Parse ( Docx(..)

3535 , Body(..)

3636 , BodyPart(..)

3737 , TblLook(..)

38+ , Extent

3839 , ParPart(..)

3940 , Run(..)

4041 , RunElem(..)

@@ -62,6 +63,7 @@ import Control.Monad.Reader

6263import Control.Applicative ((<|>))

6364import qualified Data.Map as M

6465import Text.Pandoc.Compat.Except

66+import Text.Pandoc.Shared (safeRead)

6567import Text.TeXMath.Readers.OMML (readOMML)

6668import Text.Pandoc.Readers.Docx.Fonts (getUnicode, Font(..))

6769import Text.TeXMath (Exp)

@@ -196,20 +198,23 @@ data Row = Row [Cell]

196198data Cell = Cell [BodyPart]

197199 deriving Show

198200

201+-- (width, height) in EMUs

202+type Extent = Maybe (Double, Double)

203+

199204data ParPart = PlainRun Run

200205 | Insertion ChangeId Author ChangeDate [Run]

201206 | Deletion ChangeId Author ChangeDate [Run]

202207 | BookMark BookMarkId Anchor

203208 | InternalHyperLink Anchor [Run]

204209 | ExternalHyperLink URL [Run]

205− | Drawing FilePath B.ByteString

210+ | Drawing FilePath B.ByteString Extent

206211 | PlainOMath [Exp]

207212 deriving Show

208213

209214data Run = Run RunStyle [RunElem]

210215 | Footnote [BodyPart]

211216 | Endnote [BodyPart]

212− | InlineDrawing FilePath B.ByteString

217+ | InlineDrawing FilePath B.ByteString Extent

213218 deriving Show

214219

215220data RunElem = TextRun String | LnBrk | Tab | SoftHyphen | NoBreakHyphen

@@ -619,13 +624,13 @@ expandDrawingId s = do

619624elemToParPart :: NameSpaces -> Element -> D ParPart

620625elemToParPart ns element

621626 | isElem ns "w" "r" element

622− , Just _ <- findChild (elemName ns "w" "drawing") element =

627+ , Just drawingElem <- findChild (elemName ns "w" "drawing") element =

623628 let a_ns = "http://schemas.openxmlformats.org/drawingml/2006/main"

624629 drawing = findElement (QName "blip" (Just a_ns) (Just "a")) element

625630 >>= findAttr (QName "embed" (lookup "r" ns) (Just "r"))

626631 in

627632 case drawing of

628− Just s -> expandDrawingId s >>= (\(fp, bs) -> return $ Drawing fp bs)

633+ Just s -> expandDrawingId s >>= (\(fp, bs) -> return $ Drawing fp bs $ elemToExtent drawingElem)

629634 Nothing -> throwError WrongElem

630635-- The below is an attempt to deal with images in deprecated vml format.

631636elemToParPart ns element

@@ -635,7 +640,7 @@ elemToParPart ns element

635640 >>= findAttr (elemName ns "r" "id")

636641 in

637642 case drawing of

638− Just s -> expandDrawingId s >>= (\(fp, bs) -> return $ Drawing fp bs)

643+ Just s -> expandDrawingId s >>= (\(fp, bs) -> return $ Drawing fp bs Nothing)

639644 Nothing -> throwError WrongElem

640645elemToParPart ns element

641646 | isElem ns "w" "r" element =

@@ -687,6 +692,16 @@ lookupFootnote s (Notes _ fns _) = fns >>= (M.lookup s)

687692lookupEndnote :: String -> Notes -> Maybe Element

688693lookupEndnote s (Notes _ _ ens) = ens >>= (M.lookup s)

689694

695+elemToExtent :: Element -> Extent

696+elemToExtent drawingElem =

697+ case (getDim "cx", getDim "cy") of

698+ (Just w, Just h) -> Just (w, h)

699+ _ -> Nothing

700+ where

701+ wp_ns = "http://schemas.openxmlformats.org/drawingml/2006/wordprocessingDrawing"

702+ getDim at = findElement (QName "extent" (Just wp_ns) (Just "wp")) drawingElem

703+ >>= findAttr (QName at Nothing Nothing) >>= safeRead

704+

690705elemToRun :: NameSpaces -> Element -> D Run

691706elemToRun ns element

692707 | isElem ns "w" "r" element

@@ -697,7 +712,7 @@ elemToRun ns element

697712 in

698713 case drawing of

699714 Just s -> expandDrawingId s >>=

700− (\(fp, bs) -> return $ InlineDrawing fp bs)

715+ (\(fp, bs) -> return $ InlineDrawing fp bs $ elemToExtent drawingElem)

701716 Nothing -> throwError WrongElem

702717elemToRun ns element

703718 | isElem ns "w" "r" element

tests/docx/image_no_embed.native+1−1

@@ -1,2 +1,2 @@

11[Para [Str "An",Space,Str "image:"]

2−,Para [Image ("",[],[]) [] ("media/image1.jpg","")]]

2+,Para [Image ("",[],[("width","6.5in"),("height","5.508333333333334in")]) [] ("media/image1.jpg","")]]

tests/docx/image_no_embed_writer.native+1−1

@@ -1,2 +1,2 @@

11[Para [Str "An",Space,Str "image:"]

2−,Para [Image ("",[],[]) [] ("media/rId25.jpg","")]]

2+,Para [Image ("",[],[("width","0.4166666666666667in"),("height","0.4166666666666667in")]) [] ("media/rId25.jpg","")]]

tests/docx/inline_images.native+2−2

@@ -1,2 +1,2 @@

1−[Para [Str "This",Space,Str "picture",Space,Image ("",[],[]) [] ("media/image1.jpg",""),Space,Str "is",Space,Str "an",Space,Str "identicon."]

2−,Para [Str "Here",Space,Str "is",Space,Link ("",[],[]) [Str "one",Space,Image ("",[],[]) [] ("media/image2.jpg",""),Space,Str "that"] ("http://www.google.com",""),Space,Str "links."]]

1+[Para [Str "This",Space,Str "picture",Space,Image ("",[],[("width","0.8888888888888888in"),("height","0.8888888888888888in")]) [] ("media/image1.jpg",""),Space,Str "is",Space,Str "an",Space,Str "identicon."]

2+,Para [Str "Here",Space,Str "is",Space,Link ("",[],[]) [Str "one",Space,Image ("",[],[("width","0.8888888888888888in"),("height","0.8888888888888888in")]) [] ("media/image2.jpg",""),Space,Str "that"] ("http://www.google.com",""),Space,Str "links."]]

tests/docx/inline_images_writer.native+2−2

@@ -1,2 +1,2 @@

1−[Para [Str "This",Space,Str "picture",Space,Image ("",[],[]) [] ("media/rId26.jpg",""),Space,Str "is",Space,Str "an",Space,Str "identicon."]

2−,Para [Str "Here",Space,Str "is",Space,Link ("",[],[]) [Str "one",Space,Image ("",[],[]) [] ("media/rId27.jpg",""),Space,Str "that"] ("http://www.google.com",""),Space,Str "links."]]

1+[Para [Str "This",Space,Str "picture",Space,Image ("",[],[("width","0.4166666666666667in"),("height","0.4166666666666667in")]) [] ("media/rId26.jpg",""),Space,Str "is",Space,Str "an",Space,Str "identicon."]

2+,Para [Str "Here",Space,Str "is",Space,Link ("",[],[]) [Str "one",Space,Image ("",[],[("width","0.4166666666666667in"),("height","0.4166666666666667in")]) [] ("media/rId26.jpg",""),Space,Str "that"] ("http://www.google.com",""),Space,Str "links."]]