Re: [xml] how to advance the xmlParserInput.cur?

[Wei Lu <welu cs indiana edu>]

On Sun, 12 Mar 2006, Daniel Veillard wrote:

> On Sat, Mar 11, 2006 at 11:00:01PM -0500, Wei Lu wrote:
> > Hi there
> >     I have a strang requirement for libxml2. During parsing a XML
> > document in the memory at some point I want to the parser to skip parsing
> > a chunk of memory, which is somehow guarnteed to just include a
> > well-balanced sub-tree. I modified the xmlParseContent function to skip a
> > chunk of memory when it detected a open tag. Any body can tell me the
> > correct and safe method to do that.
>
>   By definition it's not correct, it's not an XML parser anymore, "somehow
> it's XML" is not something I feel okay to provide help for. What do you intend
> to do with this bastardized code ?

The ability of skipping of a sub tree is useful for some perfomrance
optimization. For example lazy evaluation can skip a sub-tree once it
found user are not interested in the sub-tree. For tree
   A
  / \
 B   C, if B is not used now, then the generated tree could be just

   A
    \
     C

Thanks
Wei


> > By guess I tried the
> > xmlParserInputGrow, which seems doesn't change the xmpParserInput.cur at
> > all; Now my work around is just increaing the xmlParserInput.cur directly,
> > seems works well, but when file size is large, I got segment fault.
> >
> > Any clue is appreciated.
>
>   You will need to refill the buffers if you get outside of the chunk
> already availble in xmpParserInput. It is *not* trivial, encoding will
> play a role too.
>
> Daniel
>
> --
> Daniel Veillard      | Red Hat http://redhat.com/
> veillard redhat com  | libxml GNOME XML XSLT toolkit  http://xmlsoft.org/
> http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/