[skreutzer]

Thanks for your hint! Unfortunately, I think writer2epub and writer2latex aren't automatable, and for my purposes I would also like to be able to intervene intermediate processing results to do custom adjustments to it. Since I've already implemented XML to EPUB2/EPUB3/LaTeX/XSL-FO converters, those transformations aren't rocket science to me. However, do you know of some special features of writer2epub and/or writer2latex which you think I could overtake from those tools? Do you want to produce PDF via LaTeX while writer2latex isn't working yet? Do you need any bug fixing for those tools or do they just work fine? As for writer2epub, the EPUB output could also be used as input for an automated processing workflow, so in general it could make sense to improve and to support writer2epub. The only disadvantage of writer2latex is that LaTeX isn't XML, so I would consider it as a target output format, not as an intermediate format for automated processing workflows. Besides of that, I still appreciate such a tool - but I guess OpenOffice/LibreOffice PDF output is already fairly nice, isn't it?

I'm also confident that the code of those tools handle much more than my simple basic first version of html2epub, because those tools are quite around for some time and I just developed in less than a month in spare time, so probably there's lots of things which I could look for solutions in the code of those tools. However, I just implemented what I needed for my own purposes, those tools probably aim for most complete support of ODT features. I'm not sure if I should have the same goal right from the start, as there is much of other things which need work, too. But over time, more features may be added, maybe based upon writer2epub and writer2latex solutions.

Edit: Actually, I guess writer2latex is strongly influenced by the direct formatting issue, so writer2latex has to implement a LaTeX replacement for whatever may occur in ODT, and might still not be able to produce an identical result (but maybe is able). Semantic markup would simplify the case, so that OpenOffice/LibreOffice would be just the tool to write in and to apply style templates, and print typesetting by LaTeX would be an entirely separated step from it, resulting in a different visual appearance as the OpenOffice/LibreOffice WYSIWYG.