"Re: [xml] xmlParseChunk with UTF-16LE"

[Kasimier Buchcik <kbuchcik 4commerce de>]

Hi,

On 12/1/2003 1:20 PM Kasimier Buchcik wrote:
> On 12/1/2003 1:01 PM Daniel Veillard wrote:
> > On Mon, Dec 01, 2003 at 12:34:24PM +0100, Kasimier Buchcik wrote:

> > > I seem to be confused here. How can I pass the real encoding (UTF-16LE) 
> > > at parser creation time?

> >  With xmlCtxtResetPush() for example.

> I forgot to write that we are using libxml2 ver. 2.5.10 at this time. We 
> cannot switch to the current version yet. "xmlCtxtResetPush" seems not 
> to be existent in that version.
>
> > > At last: I don't have to use the push parser necessarily, for I just 
> > > used it to fake easily the first 4 bytes of an UTF-16 encoded entity, 
> > > which are autodetected by libxml2. If I could set the "real" encoding, I 
> > > would prefer to use "xmlCreateMemoryParserCtxt" and "xmlParseDocument".

> >  Then there is a *lot* of parser creation interface taking an
> > encoding declaration, why don't you use one of them ?!? ALL the
> > xmlReadxxx APIs take a const char *encoding argument.
>
> Well, we did not want to use the xmlReadxxx API since it is slower 
> (isn't it still?). Could you think of a way to accompish this with 
> version 2.5.10?

Sorry for asking once again, but since there was no resonance to my last 
mail, I assume that there is no way of correctly parsing a XML document 
with an actual encoding diverging from the declared encoding with the 
version 2.5.10.

I need this to be cofirmed. Note that this is not a big problem, but I 
just need to know if I have to freeze this issue on my side until we can 
change to a more up-to-date version of libxml2.

Thanks,

Kasimier