Docx: Parse images in deprecated vml format.

Some older versions of word use vml (vector markup language) and put
their images in a "v:imagedata" tag inside a "w:pict". We read those as
we read the more modern "blip" inside a "w:drawing".

Note that this does not mean the reader knows anything about vml. It
just looks for a `v:imagdata`. It's possible that, with more complicated
uses of images in vml, it won't do the right thing.

1 files changed, 10 insertions, 0 deletions

diff --git a/src/Text/Pandoc/Readers/Docx/Parse.hs b/src/Text/Pandoc/Readers/Docx/Parse.hs
index 5fd6b7a81..671d2acf3 100644
--- a/src/Text/Pandoc/Readers/Docx/Parse.hs
+++ b/src/Text/Pandoc/Readers/Docx/Parse.hs
@@ -639,6 +639,16 @@ elemToParPart ns element
      case drawing of
        Just s -> expandDrawingId s >>= (\(fp, bs) -> return $ Drawing fp bs)
        Nothing -> throwError WrongElem
+-- The below is an attempt to deal with images in deprecated vml format.
+elemToParPart ns element
+  | isElem ns "w" "r" element
+  , Just _ <- findChild (elemName ns "w" "pict") element =
+    let drawing = findElement (elemName ns "v" "imagedata") element
+                  >>= findAttr (elemName ns "r" "id")
+    in
+     case drawing of
+       Just s -> expandDrawingId s >>= (\(fp, bs) -> return $ Drawing fp bs)
+       Nothing -> throwError WrongElem
 elemToParPart ns element
   | isElem ns "w" "r" element =
     elemToRun ns element >>= (\r -> return $ PlainRun r)