Few years ago I have successfully used JODConverter with Open Office running in the headless mode to convert all sorts of files.
To be honest I cannot remember the exact limitations of the software, but there is a version of what I did still running as a webservice here:
http://oxgarage.oucs.ox.ac.uk:8080/ege-webclient/
and here: http://www.tei-c.org/oxgarage/.
No idea about what is the current version of the converter they are using and it is possible that it is a bit outdated and at the moment it works a lot better, but you can try to convert your files and see how well they come up before you go head first into coding.
I also found this article and it looks like it might be quite helpful for you, have a look at it:
http://angelozerr.wordpress.com/2012/12/06/how-to-convert-docxodt-to-pdfhtml-with-java/.
Oh and by the way, that webservice is opensource:
https://github.com/sebastianrahtz/oxgarage , so you can have a look at how it is implemented as well.