writing a code generator

[Srinivas Nedunuri <nedunuri cs utexas edu>]

hello I am interested in generating an XML representation from a UML 
model drawn as a Dia diagram in a custom format. I took a look at the 
Dia2Code source for inspiration. However I would like to write the 
generator either in Java (if possible) or XSL (I am not familiar with 
C). I wondered if you could give me some pointers about how to get 
started - do you recommend XSL or an actual scripting languge, is there 
a DTD available, is there a pure export of a Dia UML digram minus the 
layout information, what are the gotchas I need to worry about, etc

Any help you could give is much appreciated

thanks