Re: [xml] why is htmlNewParserCtxt static?

[Michael Day <mikeday yeslogic com>]

Hi Daniel,

>   But shouldn't you use one of the wrapper functions where you specify
> what input will be used ? Like htmlCreateMemoryParserCtxt() or one of the
> htmlRead... function ? What is your use case ?

The use case is replacing xmlDefaultExternalEntityLoader() with a 
customised entity loader function.

The entity loader is passed an xmlParserCtxtPtr, which allows 
document-specific options to be passed in using the _private field and 
used by the entity loader (as long as you don't use xmlReader :) So when 
I parse a document I want to pass in some stuff in the _private field of 
the context.

Normally I would use xmlReadFile(), but that creates its own context and 
destroys it afterwards and doesn't give me a chance to put anything in 
_private. However, xmlCtxtReadFile() looks like just what I want, as it 
takes a context that I've already created with the appropriate stuff in 
_private.

On the HTMLparser side of things I can't do this, as htmlNewParserCtxt() 
is static and I can't call it. I want to call htmlCtxtReadFile(), but 
first I need to create a context, set the _private field and pass it in.

The htmlCreateFileParserCtxt() is the only way I can create a context, 
but I can't use it, because it calls xmlLoadExternalEntity() itself, 
before I have had a chance to set _private. Then when I pass the context 
into htmlCtxtReadFile() it will call xmlLoadExternalEntity() again!

So I need a way to create a HTML parsing context that will let me set 
_private before trying to load any external entities. Does this seem 
reasonable? I can do it with the XML parsing API, so I just assume that 
the HTML parsing API should work the same way.

Best regards,

Michael

--
Print XML with Prince!
http://www.princexml.com