HTML parser segfaults

Hi,

I've been getting segfaults when trying to write a SAX parser for HTML. It
looks like libxml2.htmlCreatePushParser works correctly but the first time
you call libxml2.htmlParseChunk it will segv because, in C land,
ctxt->input (and possibly ctxt) has been trashed somewhere. I got a little
lost trying to debug it as python seems to be doing some wierd threading
stuff (and I'm very tired now because of it :-/), so I thought I'd post it
here and see if anyone else can find it whilst I sleep ;)

The attached tarball contains some stuff to reproduce it, although it 
happened on every html or xhtml file I could find and with both python 1.5 
and python 2.2.

Try "python xml.py" (should work), and "python html.py" (should sig11). 
I'm using Red Hat Linux 7.3 with Daniel's latest libxml2 rpms from 
gnome.org (rpm -q says python-1.5.2-38, python2-2.2-16, libxml2-2.4.21-1
and libxml2-python-2.4.21-1).

Cheers,
Gary

[ [email protected] ][ GnuPG 85A8F78B ][ http://inauspicious.org/ ]