Пример #1
0
UT_Error UT_XML::parse (const char * szFilename)
{
	UT_ASSERT (m_pListener || m_pExpertListener);
	UT_ASSERT (szFilename);
	
	if ((szFilename == 0) || ((m_pListener == 0) && (m_pExpertListener == 0))) return UT_ERROR;
	if (!reset_all ()) return UT_OUTOFMEM;
	
	UT_Error ret = UT_OK;
	
	DefaultReader defaultReader;
	Reader * reader = &defaultReader;
	if (m_pReader)
		reader = m_pReader;
	
	if (!reader->openFile (szFilename))
    {
		UT_DEBUGMSG (("Could not open file %s\n", szFilename));
		return UT_errnoToUTError ();
    }
	
	char buffer[2048];
	
	m_bStopped = false;
	
	xmlSAXHandler hdl;
	xmlParserCtxtPtr ctxt = 0;
	
	memset(&hdl, 0, sizeof(hdl));
	
	hdl.getEntity    = _getEntity;
	hdl.startElement = _startElement;
	hdl.endElement   = _endElement;
	hdl.characters   = _charData;
	hdl.error        = _errorSAXFunc;
	hdl.fatalError   = _fatalErrorSAXFunc;
	hdl.processingInstruction = _processingInstruction;
	hdl.comment      = _comment;
	hdl.cdataBlock   = _cdata;

	size_t length = reader->readBytes (buffer, sizeof (buffer));
	int done = (length < sizeof (buffer));
	
	if (length != 0)
    {
		ctxt = xmlCreatePushParserCtxt (&hdl, static_cast<void *>(this), buffer, static_cast<int>(length), szFilename);
		if (ctxt == NULL)
		{
			UT_DEBUGMSG (("Unable to create libxml2 push-parser context!\n"));
			reader->closeFile ();
			return UT_ERROR;
		}
		xmlSubstituteEntitiesDefault (1);
		UT_sint32 chucks = -1;
		while (!done && !m_bStopped)
		{
			chucks++;
			length = reader->readBytes (buffer, sizeof (buffer));
			UT_DEBUGMSG(("Done chunk %d length %zd \n",chucks,length));
			done = (length < sizeof (buffer));
			
			if (xmlParseChunk (ctxt, buffer, static_cast<int>(length), 0))
			{
			  if(getNumMinorErrors() > getNumRecoveredErrors())
			    {
				UT_DEBUGMSG (("Error - 1 parsing '%s' (Line: %d, Column: %d)\n", szFilename, xmlSAX2GetLineNumber(ctxt), xmlSAX2GetColumnNumber(ctxt)));
				ret = UT_IE_IMPORTERROR;
				break;
			    }
			}
		}
		if (ret == UT_OK)
		  if (!m_bStopped && (getNumMinorErrors() == 0))
			{
				if (xmlParseChunk (ctxt, "", 0, 1))
				{
					UT_DEBUGMSG (("Error -2 parsing '%s' (Line: %d, Column: %d)\n", szFilename, xmlSAX2GetLineNumber(ctxt), xmlSAX2GetColumnNumber(ctxt)));
					ret = UT_IE_IMPORTERROR;
				}
			}
		if (ret == UT_OK && (getNumMinorErrors() == 0))
			if (!ctxt->wellFormed && !m_bStopped) ret = UT_IE_IMPORTERROR; // How does stopping mid-file affect wellFormed?

		xmlDocPtr myXmlDoc = ctxt->myDoc;
		xmlFreeParserCtxt (ctxt);
		xmlFreeDoc(myXmlDoc);
    }
	else
    {
		UT_DEBUGMSG(("Empty file to parse - not sure how to proceed\n"));
    }
	
	reader->closeFile ();
	
	return ret;
}