Re: How do I use media bags?

[John MacFarlane <jgm-TVLZxgkOlNX2fBVCVOL8/A@public.gmane.org>]

Paul <freack1208-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> writes:

> Thanks for pointing that out! Now I got it to almost work. The epub is 
> being generated and it contains images, but it seems to like using images 
> where they shouldn't be (not like randomly adding images, but using the 
> wrong image). Are there collisions between the images? If so, is that 
> something one can fix?

The mediabag code was designed with the idea that you'd be extracting
media from one file and doing something with it.  Thus, insertMedia
just uses Data.Map.insert with the normalized file path. If two docx
files have different images with the same filename, and you read
both of them in, you'll get a collision in the media bag
and unwanted results.

Perhaps this is something that could be improved.  For example,
insertMedia could be modified to return a filename, perhaps
modified if there's already an item with that name.  This would
require some modifications across several readers.
@jkr @mpickering @tarleb - any thoughts?

In the mean time, your best bet would be to postprocess the Pandoc
structure after each docx is read using a filter, changing the
image names systematically in both the Image elements and the
MediaBag, to avoid collisions.