BB said: "unless you perform the conversions using the .txt version... but that means you need to beef up the strength of those... _and_ ensure that they have the consistency that's needed... and thus we end up at what i have been saying all along..." I am delighted to say that I _agree_ with BB! You _must_ have a canonical text version from which all other versions derive. Even better would be a marked-up text version and a stripping script to remove all markup. The option is then there to either remove the markup or use it to generate XML, etc. Easily said, but this would require an enormous house-cleaning effort. Marcello said: "Tcl is *not* our language of choice and integrating your tcl script into our python framework will probably take longer than rewriting the interesting bits from scratch. The only thing that really interest me is your experience with PDF toolkits. Which one are you using? What can you say pro and contra it?" Tcl or Python or...? This is just detail and any porting would be simple, as the whole system pretty much floats on regular expressions. PDF toolkits: The c2xsl script uses one or more CSS-style stylesheets to generate an XSLT stylesheet. This is then used to generate LaTeX text from the XML file. This is then processed with pdflatex. John Redmond