I've had some limited experience with this, specifically html to PDF conversion. I ended up using a product called
htmldoc, which gave me reasonable results and speed. I recall using ghostscript in conjunction with another parser which produced postscript, but if I recall correctly, the ghostscript execution was rather slow (of course, I was converting >100 pages with tables in one shot). A potential solution for you would be to use
wvWare and friends in conjunction with ghostscript.
