On Fri, Apr 07, 2006 at 11:56:18PM +0100, Nic wrote: > I'm in a situation where I'm doing some processing on XSLT documents > when they are loaded and I'm having a lot of trouble doing this > efficiently. > > What I have right now is a python entity loader which: > > - reads in the associated document > - parses it in memory > - processes the XSLT > - caches the dom for later > > > Because I'm loading the document and parsing it myself I tried (for a > while) to have the entity loader pass back the document itself. > > But that didn't work. > > So now I have the entity loader return an open python file handle. > > Clearly that is going to mean that libxml2 duplicates my effort by > parsing the document again.
By definition an entity is a sequence of bytes. Doing XSLT processing at that level forces a serialization. Remember that the parsing of an entity can be dependant of the context (e.g. if you parse part of the external subset, this will be directly dependant on existing defined entities), so there is no way to avoid that reparsing in general. Daniel -- Daniel Veillard | Red Hat http://redhat.com/ [EMAIL PROTECTED] | libxml GNOME XML XSLT toolkit http://xmlsoft.org/ http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/ _______________________________________________ xml mailing list, project page http://xmlsoft.org/ [email protected] http://mail.gnome.org/mailman/listinfo/xml
