Hi,
Currently, when parsing a docx, a v:imagedata tag inside a w:r/w:object/v:shape
is ignored. But it could be treated the same as a v:imagedata inside a
w:r/w:pict, both represent images linked by the r:id attribute.
In fact, we now use a pre-processing to transform a
w:r/w:object/v:shape/v:imagedata to a w:r/w:pict/v:imagedata before transforming
with Pandoc and we have tested this on many documents.
So we would like to implement an improvement to treat both the same. Do you have
any objections or feedback? We'll make a pull request soon.