RE: [xml] xmlNodeGetContent() for XML_ENTITY_REF_NODE

["Henke, Markus" <Markus_Henke ordat com>]

> -----Original Message-----
> From: Daniel Veillard [mailto:veillard redhat com]
> Sent: Thursday, February 21, 2002 6:19 PM
> To: Henke, Markus
> Cc: 'xml gnome org'
> Subject: Re: [xml] xmlNodeGetContent() for XML_ENTITY_REF_NODE
>
>
> On Thu, Feb 21, 2002 at 06:02:54PM +0100, Henke, Markus wrote:
> > Hello,
> >
> > with regard to 
> > http://mail.gnome.org/archives/xml/2001-October/msg00160.html
> > i'm trying to collect the content of a XML_ELEMENT_NODE without
> > consider the child-XML_ELEMENT_NODEs.
> > I'm wondering if it's intended that 'xmlNodeGetContent(myNode)'
> > generally results 'NULL' if 'myNode' is a XML_ENTITY_REF_NODE?
> > And if so, what's the appropriate method to get the (expanded)
> > content of a  XML_ENTITY_REF_NODE?
>
>    Get the ENTITY_DEF by doing an entity lookup for the entity
> with that name, concatenate the xmlNodeGetContent() of its 
> children.

Yep, works (at least for non-nested entity refs)!


> I'm not sure you can get this recursive.


Hum, you mean nested entity references? Weird, i've never
thought about this. My app instantly crashed as i've tested
it. Smell's like a new debugging session... 8]
I'd hoped that i could pilfer at 'xmlGetProp()', but it won't
do a recursiv entity substitution!?
However, i've to do this and i should manage it since cyclic
entity refs are illegal.
BTW, the libxml parser detects cyclic entity
references but it seems that it loops about 40 times? 


> Another way would be to change xmlNodeGetContent()
> behaviour on entities nodes.

I'll try to code a recursiv entity substitution for
XML_ENTITY_REF_NODEs. If it works for me i'll send
it to the list so you can decide if it makes sense to
include it in xmlNodeGetContent() (and possibly
xmlGetProp()...).


> > Another point: I've debugged my code to see what happens
> > and found that a node with type XML_ENTITY_REF_NODE
> > holds the expanded entity in it's 'content' component,
>
>   Actually it's a pointer to the associated ENTITY_DEF
> node content's, the subtree are shared.

I see. Debugging don't show that (at least not obvious).
 

> > although i've set 'xmlSubstituteEntitiesDefault(0)'.
> > Is this correct?
>
>   yes, with xmlSubstituteEntitiesDefault(1) you would get the content
> directly instead of the reference node.
>   xmlSubstituteEntitiesDefault(0) means the tree keeps track of the
> entities in the document (including in attribute values !) so it can
> save them back. xmlSubstituteEntitiesDefault(1) wipe them out and
> substitute them with their content.

Then xmlSubstituteEntitiesDefault(0) is what i need...8)


> Daniel

Thanx & Ciao, Markus