On Mon, Oct 05, 2009 at 01:51:02PM +0200, Andreas Haufler (scireum) <[email protected]> wrote: > I was debugging why sometimes an error like ? Illegal entry in bfchar block > in ToUnicode CMap? occurs when reading valid PDFs. I think I spotted the > problem in CharCodeToUnicode.cc in line 245. There is a check that the first > token is exactly 2 digits long, however: 0022 is also valid?
Ah, thanks for the hint. I'll add a fix for this to swftools' xpdf changeset. > While debugging, I learned that there is a fork of xpdf called libpoppeler > which is intended for programs using xpdf as library. Since libpoppeler is > maintained and updated as xpdf is, pdf2swf would benefit from using this > library instead of managing an own ?fork? of xpdf. So I wonder if there are > deep modifications within the used xpdf sources or if they could possibly be > replaced by libpoppeler (?) It's not straightforward to switch from xpdf to poppler. In particular, all the changes that swftools does to xpdf would need to be ported to poppler. Also, Derek Noonburg (the xpdf author) is working on a new version of xpdf- so xpdf is not unmaintained, it just has a different release granularity. > Otherwise it would be kind to integrate the current version of > CharCodeToUnicode.cc from xpdf or libpoppeler to pdf2swf since it seems to > handle Unicode-stuff better when it comes to multi-char mappings like > ligatures (fl, fi, etc.). If you're using pdf2swf commercially, you might want to consider buying an upgraded version of xpdf from the xpdf author. Unlike poppler, it's pretty straightforward to make pdf2swf use a new xpdf version. Matthias
