Re: [xml] context reuse for push parser

[Daniel Veillard <veillard redhat com>]

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.

> static void xmlCtxtInitPushParser(xmlParserCtxtPtr ctxt, const char
> *chunk, int size, const char *filename)
> {
>   xmlParserInputPtr inputStream;
>   xmlParserInputBufferPtr buf;
>   xmlCharEncoding enc = XML_CHAR_ENCODING_NONE;
>
>   if ((chunk != NULL) && (size >= 4))
>     enc = xmlDetectCharEncoding((const xmlChar *) chunk, size);
>
>   buf = xmlAllocParserInputBuffer(enc);
>   if (buf == NULL) return;
>
>   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).

>     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,

>   if (ctxt->pushTab == NULL) {
>     xmlErrMemory(ctxt, NULL);
>     xmlFreeParserInputBuffer(buf);
>     return;
>   }
>
>   if (filename == NULL) {
>     ctxt->directory = NULL;
>   } else {
>     ctxt->directory = xmlParserGetDirectory(filename);
>   }
>
>   inputStream = xmlNewInputStream(ctxt);
>   if (inputStream == NULL) {
>     xmlFreeParserInputBuffer(buf);
>     return;
>   }
>
>   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);
>   }

  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.

> }

  it's a good idea, that could be reused within the xmlreader code.

Daniel

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