Re: [xml] context reuse for push parser

[Graham Bennett <graham-libxml simulcra org>]

I have updated my function:

static int xmlCtxtInitPushParser(xmlParserCtxtPtr ctxt, const char
*chunk, int size, const char *filename, xmlCharEncoding enc)
{
  xmlParserInputPtr inputStream;
  xmlParserInputBufferPtr buf;
                                                                                
  if (enc == XML_CHAR_ENCODING_NONE && (chunk != NULL) && (size >= 4))
    enc = xmlDetectCharEncoding((const xmlChar *) chunk, size);
                                                                                
  buf = xmlAllocParserInputBuffer(enc);
  if (buf == NULL) return 1;
                                                                                
  if (ctxt == NULL) {
    xmlFreeParserInputBuffer(buf);
    return 1;
  }
                                                                                
  xmlCtxtReset(ctxt);
                                                                                
  if (ctxt->pushTab == NULL) {
    if ((ctxt->pushTab = (void **) xmlMalloc(ctxt->nameMax * 3 *
sizeof(xmlChar
*))) == NULL)
    {
      xmlFreeParserInputBuffer(buf);
      return 1;
    }
  }
                                                                                
  if (filename == NULL) {
    ctxt->directory = NULL;
  } else {
    ctxt->directory = xmlParserGetDirectory(filename);
  }
                                                                                
  inputStream = xmlNewInputStream(ctxt);
  if (inputStream == NULL) {
    xmlFreeParserInputBuffer(buf);
    return 1;
  }
                                                                                
  if (filename == NULL)
    inputStream->filename = NULL;
  else
    inputStream->filename = (char *)
      xmlCanonicPath((const xmlChar *) filename);
  inputStream->buf = buf;
  inputStream->base = inputStream->buf->buffer->content;
  inputStream->cur = inputStream->buf->buffer->content;
  inputStream->end =
    &inputStream->buf->buffer->content[inputStream->buf->buffer->use];
                                                                                
  inputPush(ctxt, inputStream);
                                                                                
  if ((size > 0) && (chunk != NULL) && (ctxt->input != NULL) &&
      (ctxt->input->buf != NULL))  {
    int base = ctxt->input->base - ctxt->input->buf->buffer->content;
    int cur = ctxt->input->cur - ctxt->input->base;
                                                                                
    xmlParserInputBufferPush(ctxt->input->buf, size, chunk);
                                                                                
    ctxt->input->base = ctxt->input->buf->buffer->content + base;
    ctxt->input->cur = ctxt->input->base + cur;
    ctxt->input->end =
      &ctxt->input->buf->buffer->content[ctxt->input->buf->buffer->use];
#ifdef DEBUG_PUSH
    xmlGenericError(xmlGenericErrorContext, "PP: pushed %d\n", size);
#endif
  }
                                                                                
  if (enc != XML_CHAR_ENCODING_NONE) {
    xmlSwitchEncoding(ctxt, enc);
  }
                                                                                
  return 0;
}


On Wed, Oct 22, 2003 at 08:06:35AM +0100, Graham Bennett wrote:
> On Tue, Oct 21, 2003 at 05:58:45PM -0400, Daniel Veillard wrote:
> > On Tue, Oct 21, 2003 at 10:19:43PM +0100, Graham Bennett wrote:
> > > Hi all,
> > >
> > > I added the following function to enable me to reuse the parser context
> > > using the push parser.  It corresponds to xmlCreatePushParserCtxt, but
> > > takes an existing context.  Not sure if it's generally useful but it
> > > seems to work for my code.
> >
> >   Okay might make a good addition to the new parser interfaces.
>
> [ snip ]
>
> > >   if (ctxt == NULL) {
> > >     xmlErrMemory(NULL, "No parser context\n");
> >
> >   Hum, this isn't really a memory error. It's an API use error, I would rather
> > make the function return 0 in case of success and -1 in case of error, and
> > return -1 in that case (and check on function entry).
>
> Yep sounds better.  I'm sure a lot of it needs to be cleaned up, I just
> did the bare minimum to get it working.
>  
> > >     xmlFreeParserInputBuffer(buf);
> > >     return;
> > >   }
> > >
> > >   xmlCtxtReset(ctxt);
> > >
> > >   ctxt->pushTab = (void **) xmlMalloc(ctxt->nameMax * 3 * sizeof(xmlChar
> > > *));
> >
> >   Hum, you're likely to memleak there. ctxt->pushTab is probably not NULL,
>
> Oops yes, that's wrong :)
>
> [ snip ]
>
> > >   if (enc != XML_CHAR_ENCODING_NONE) {
> > >     xmlSwitchEncoding(ctxt, enc);
> > >   }
> >
> >   maybe an extra encoding passed as a const char * is needed. C.f. 
> > section F.2. of the spec:
> >    http://www.w3.org/TR/REC-xml#sec-guessing-with-ext-info
> > in that case the context encoding overrides the one fine in the XML
> > declaration or autodetected by xmlDetectCharEncoding() which implement
> > the F.1. detection algorithm.
>  
> Sounds reasonable.  In my case I think I'd always want to autodetect it,
> but still it would be useful.
>
> >   it's a good idea, that could be reused within the xmlreader code.
>
> Ok cool.
>
> Thanks,
>
> Graham.
>
> -- 
> Graham Bennett
> _______________________________________________
> xml mailing list, project page  http://xmlsoft.org/
> xml gnome org
> http://mail.gnome.org/mailman/listinfo/xml

-- 
Graham Bennett