RE: [xml] Need example / help for using SAX parsing on IO stream (socket)

["Chris Hesterman" <hester nortel com>]

Daniel...

Thanks for your quick response. I had looked at the section of code that
you recommend, and will use that going forward. I am a little confused
as to why the method xmlCreateIOParserCtxt() isn't more appropriate than
xmlCreatePushParserCtxt() when dealing with a socket connection...

Is it mainly the issue of an xml message spanning across multiple reads?

Cheers,

Chris Hesterman 

-----Original Message-----
From: Daniel Veillard [mailto:veillard redhat com] 
Sent: Wednesday, July 13, 2005 12:29 PM
To: Hesterman, Chris 
Cc: xml gnome org
Subject: Re: [xml] Need example / help for using SAX parsing on IO
stream (socket)


On Wed, Jul 13, 2005 at 11:43:12AM -0400, Chris Hesterman wrote:
> Hello....
>
> I've been searching for documentation and / or code examples for using

> the SAX parser in libxml2 to parse an input stream (XML messages being

> received over a socket). I have played around a lot, but need some 
> guidance on how to use the IO methods and SAX....
>
> Here are the goals I'm trying to acheive with this code:
>
> 1. parse an input stream of messages (from a socket connection)

  the parser can't split the messages for you, you will have to take
control over the socket then. As I said repeatedly stacking multiple XML
document within a single stream is a design mistake error.

> 2. parse "on the fly" (i.e. don't require that the entire message has 
> been collected before parsing). I assume this implies using SAX and 
> I/O stream methods.

  No. This implies using a Push mode.

> 3. keep memory requirements low by not creating a tree (i.e. use SAX).

  or the XML reader but this may not be possible in push mode.

> 4. account for the retrieval of a message requiring multiple reads 
> from the socket (i.e. the message length is larger than the socket's 
> buffer).

  Since you're packing multiple message on a single socket you will need
to do the reads. Then push the data on the parser context


> My guess as to how to go about this is to use the following methods, 
> but I need some verification that I'm using the right ones, and an 
> example of how to piece it all together: For now I just want to get 
> the basics working, and have a print out when the start of a message 
> ("<rpc
> message-id="123">") or the end of a message ("</rpc>") is detected,
and
> I'll build from there...

  I don't have time to read and review other people code, so read mine:
    see xmllint.c around line 2107
    You need to use

      xmlCreatePushParserCtxt() passing your SAX block and user data
      xmlParseChunk() until all document input have been consumed.

Daniel


-- 
Daniel Veillard      | Red Hat Desktop team http://redhat.com/
veillard redhat com  | libxml GNOME XML XSLT toolkit
http://xmlsoft.org/ http://veillard.com/ | Rpmfind RPM search engine
http://rpmfind.net/