Currently, when exporting to odt, docutils SVN generates 'code' directive contents twice.
The attached patch does something horrible:
- it only uses the pygmented stuff coming from the odf_odt transformer
- while fixing it (it generated xml with an 'ns0' namespace)
Instead of using the already parsed stuff.
So yes, it's possibly the wrong way of doing it.
But I had troubles with the other one.
Maybe someone else will fix it in a prettier fashion...
The attached patch allows better handling of image dimension specifications:
- image widths in % (of text width) are handled.
Usually specifying an image dimension in % is good because the image
will be displayed in the same way in a browser and on a page.
- incomplete dimension specifications are better handled.
The aspect ratio is kept.
I made several tests with OOo / libreoffice.
Originally I thought it would figure out unspecified dimensions, but... no.
As collateral damage:
- page geometry is rendered a little more flexible.
- minor edits
PS: please also consider the previous patch, indispensable for 'code' directives