Can I use FOP to convert XHTML or HTML files to PDF?
You will get better results if you transform your source XML directly into XSLFO. It is tempting to use the XML->XHTML->XSLFO approach, because this apparently means there’s only one transformation to maintain, but XHTML and publishing grade PDF are dissimilar enough that an “generic” XHTML->XSLFO(PDF) usually wont cut it, because the XHTML already lacks information from the original XML. It may be worthwhile to go from the source XML to an intermediate custom XML which on one hand is close enough to representation to allow simple transformation to either XHTML or XSLFO for most of the structure, but on the other hand retains enough semantic from the original XML that elements which must be handled differently in the two representations can be handled differently. Setting this up requires experience and most likely extensive prototyping.
