Re: [xml] setting URL for xmlRelaxNGParserCtxt?

[Martijn Faassen <faassen infrae com>]

Daniel Veillard wrote:
> On Tue, Jan 25, 2005 at 08:14:22PM +0100, Martijn Faassen wrote:
>
> > Hi there,
> >
> > In my lxml Python binding for libxml2, I was using the following 
> > approach to construct a Relax NG schemas:
> >
> > xmlDocPtr* doc = ...
> > parser_ctxt = xmlRelaxNGNewDocParserCtxt(doc)
> >
> > unfortunately, this got me into trouble with Relax NG schemas that use 
> > the 'include' statement. They try to load the files to be included, but 
> > have no idea where to look, as this information is lost.
> >
> > Glancing through the source, it seems like the only way to make the 
> > 'include' statement work is to use xmlRelaxNGNewParserCtxt, which takes 
> > a URL.
>
>    Wrong. How did you built "doc" ?  If it's from a memory string you can
> still define the base for that document by providing the URL argument
> of xmlReadMemory().
>
> > Would it be useful to have a way to supply this URL to an existing 
> > xmlRelaxNGParserCtxtPtr? Something like:
>
>
>   no the base has to be set in the document. Since that part is just
> described by '...' in your mail there is little more indications I can
> provide.

Interesting. This is how a doc typically get parsed now:

pctxt = xmlNewParserCtxt()
.. (share dictionaries)
result = xmlCtxtReadFile(pctxt, filename,
                                   NULL, _getParseOptions())

the other approach is to create a doc with xmlNewDoc().

Thinking it through, it's quite possible that I often have a node being 
passed from a doc that's parsed into a doc that's created with 
xmlNewDoc(). I will investigate whether that happens here, and whether I 
can pass along the URL somehow when it happens.

[snip]
>   The base is a property of the document, not of the relax-NG transformation
> compilation, this is really the wrong approach.

> > I imagine the same issues come up when using XSLT that tries to include 
> > or import external stylesheets.
>  
>   yes the base is *in* the document.
>   xmlNodeSetBase(xmlNodePtr cur, const xmlChar* uri)
> to be called on your document. But it should really be there as the result
> of parsing.

Good, good. I have an idea of what to do to fix the issue. Thanks for 
the hints and my apologies for the misunderstanding.

Regards,

Martijn