Re: strange transformCtxt free-ing problem

Daniel Veillard <veillardredhat.com> writes:

> On Sun, Jan 14, 2007 at 02:40:49PM +0000, Nic James
Ferrier wrote:
>> I've got a document loader implemented in python
that handles xml or
>> html documents thus:
>> 
>> 
>>   if re.match("^[a-z]+://.*", url):
>>       cached_file, header =
hapi_lib.httputils.url_retrieve(url)
>>       if header["content-type"] ==
"text/html":
>>            doc =
parserContext.htmlCtxtReadFile(cached_file,
"UTF8", 0)
>>       else:
>>           # Guess that it's XML
>>           doc =
parserContext.ctxtReadFile(cached_file, "UTF8",
0)
>> 
>> The HTML works fine until I come to free the
transformContext that
>> this is all done within.
>> 
>> When I do I get a glibc error and my process
segfaults.
>
>   How do you free it ? And what how is that piece of
code related to a
> transformContext, I don't see any ?!? A transformation
context made from
> a document consumes the XSLT document, if you free it
it will die when
> python reclain the transformation context. But I see no
relationship to
> HTML or XML parsing of *transformed* document.

The code shown is in the loader.

The loader is set... and then:

- a stylesheet is created, 
- a transformContext is created
- the stylesheet is applied using the transformContext
(causing the
  document loader to be called with XML and HTML resources
requested) 
- the result is obtained
- the result is serialized to a string
- the result is printed
- the transformContext is freed

and then it segfaults.

It's only the calls to the HTML parsing code that cause the
free-ing
of the transformContext to fail.
-- 
Nic Ferrier
http://www.tapsellfer
rier.co.uk   for all your tapsell ferrier needs
_______________________________________________
xml mailing list, project page  http://xmlsoft.org/
xmlgnome.org
http://mai
l.gnome.org/mailman/listinfo/xml