documentparser.cpp

OSB-PIK-OpenVXI-3.0.0源代码 “中国XML论坛 - 专业的XML技术讨论区--XML在语音技术中的应用”

 
   if (bufSize < 2047)
     bufSize = 2047;
 
   ++bufSize;
   VXIbyte * buffer = new VXIbyte[bufSize];
   if(buffer == NULL) {
     log.LogError(202);
     return -1;
   }
 
   bool reachedEnd = false; 
   read = 0;
 
   while (!reachedEnd) {
     VXIulong bytesRead = 0;
     switch (inet->Read(inet, buffer+read, bufSize-read, &bytesRead, stream)) {
     case VXIinet_RESULT_SUCCESS:
       read += bytesRead;
       break;
     case VXIinet_RESULT_END_OF_STREAM:
       read += bytesRead;
       reachedEnd = true;  // exit while
       break;
     case VXIinet_RESULT_WOULD_BLOCK:
       VXItrdThreadYield();
       break;
     default:
       inet->Close(inet, &stream);
       delete[] buffer;
 
       log.LogDiagnostic(0, L"DocumentParser::FetchBuffer - "
                         L"could not read from URL.");
       return 3;
     }
 
     if (read == static_cast<VXIulong>(bufSize)) {
       // The number of bytes read exceeds the number expected.  Double the
       // size and keep reading.
       VXIbyte * temp = new VXIbyte[2*bufSize];
       if(temp == NULL) {
         log.LogError(202);
         delete [] buffer;
         return -1;
       }
       memcpy(static_cast<void *>(temp), static_cast<void *>(buffer),
              bufSize * sizeof(VXIbyte));
       delete[] buffer;
       buffer = temp;
       bufSize *= 2;
     }
   }
 
   inet->Close(inet, &stream);
   result = buffer;
 
   log.LogDiagnostic(2, L"DocumentParser::FetchBuffer - success");
 
   return 0;
 }
 
 
 void DocumentParser::ReleaseBuffer(const VXIbyte * & buffer)
 {
   if (buffer != VALIDATOR_DATA + VXML_DEFAULTS)
     delete[] const_cast<VXIbyte *>(buffer);
 
   buffer = NULL;
 }
 
 
 //#############################################################################
 // FetchDocument
 //#############################################################################
 
 // -2: Internal error
 int DocumentParser::FetchDocument(const VXIchar * url,
                                   const VXIMapHolder & properties,
                                   VXIinetInterface * inet,
                                   VXIcacheInterface * cache,
                                   SimpleLogger & log,
                                   VXMLDocument & document,
                                   VXIMapHolder & docProperties,
                                   bool isDefaults,
                                   VXIbyte **content,
                                   VXIulong *size)
 {
   int result;
 
   if (log.IsLogging(2)) {
     log.StartDiagnostic(2) << L"DocumentParser::FetchDocument(" << url << L")";
     log.EndDiagnostic();
   }
 
   // (1) Load the VXML DTD for validation.  This will override an externally
   // specified DTD if the user provides a link.
 
   try {
     if (isDefaults) {
       parser->parse(MemBufInputSource(VALIDATOR_DATA + DUMMY_VXML_DEFAULTS_DOC,
                                       DUMMY_VXML_DEFAULTS_DOC_SIZE,
                                       "vxml 1.0 defaults"));
       converter->ResetDocument(); // Throw this document away.
     }
 
     if (!isDefaults && !loadedVXML20) {
       // Preload the VXML 2.0 schema.
       VXIcharToXMLCh name(L"http://www.w3.org/TR/voicexml20/vxml.xsd");
       LockLoadGrammar();
       parser->loadGrammar(name.c_str(), Grammar::SchemaGrammarType, true);
 
       // Reuse cached grammars if available.
       parser->setFeature(XMLUni::fgXercesUseCachedGrammarInParse, true);
       
       loadedVXML20 = true;
       UnlockLoadGrammar();
     }
   }
   catch (const XMLException & exception) {
     log.StartDiagnostic(0) << L"DocumentParser::FetchDocument - XML parsing "
       L"error from DOM: " << XMLChToVXIchar(exception.getMessage());
     log.EndDiagnostic();
     log.LogError(999, SimpleLogger::MESSAGE, L"unable to load VXML DTD");
     return 4;
   }
   catch (const SAXParseException & exception) {
     log.StartDiagnostic(0) << L"DocumentParser::FetchDocument - Parse error "
                            << L"in file \"" 
 			   << XMLChToVXIchar(exception.getSystemId()) 
                            << L"\", line " << exception.getLineNumber()
                            << L", column " << exception.getColumnNumber()
                            << L" - " << XMLChToVXIchar(exception.getMessage());
     log.EndDiagnostic();
     log.LogError(999, SimpleLogger::MESSAGE, L"unable to load VXML DTD");
     return 4;
   }
 
   // (2) Load the url into memory.
 
   const VXIbyte * buffer;
   VXIulong bufSize;
   vxistring docURL;
   bool isDefaultDoc = false;
   
 
   if (isDefaults && wcslen(url) == 0) {
     buffer  = VALIDATOR_DATA + VXML_DEFAULTS;
     bufSize = VXML_DEFAULTS_SIZE;
     docURL = L"builtin defaults";
     result = 0;
     isDefaultDoc = true;
   }
   else {
     result = DocumentParser::FetchBuffer(url, properties, docProperties, inet, log,
                                          buffer, bufSize, docURL);
   }
 
   if (result != 0) {
     if (log.IsLogging(0)) {
       log.StartDiagnostic(0) << L"DocumentParser::FetchDocument - exiting "
         L"with error result " << result;
       log.EndDiagnostic();
     }
     return result; // may return { -1, 1, 2, 3 }
   }
 
   // store buffer for reference
   if (content) {
     VXIbyte *tempbuf = new VXIbyte[bufSize];
     if(tempbuf == NULL) {
       log.LogError(202);
       DocumentParser::ReleaseBuffer(buffer);
       return -1;
     }
     memcpy(tempbuf, buffer, bufSize);
     if (size != NULL) *size = bufSize;
     *content = tempbuf;
   }
   
   // (3) Pull the document from cache.
   
   // (3.1) Set the base uri for this document, but
   // ignore if it is the default doc
   if(!isDefaultDoc) 
     converter->SetBaseUri(docURL.c_str());
 
   VXMLDocument doc;
   if (!DocumentStorageSingleton::Instance()->Retrieve(doc, buffer, bufSize)) {
 
     // (4) Not in cache; parse buffer into our VXML document representation
     try {
       VXIcharToXMLCh membufURL(docURL.c_str());
       parser->parse(MemBufInputSource(buffer, bufSize, membufURL.c_str(), false));
     }
     catch (const XMLException & exception) {
       DocumentParser::ReleaseBuffer(buffer);
       if (log.IsLogging(0)) {
         log.StartDiagnostic(0) << L"DocumentParser::FetchDocument - XML "
           L"parsing error from DOM: " <<XMLChToVXIchar(exception.getMessage());
         log.EndDiagnostic();
       }
       return 4;
     }
     catch (const SAXParseException & exception) {
       DocumentParser::ReleaseBuffer(buffer);
       if (log.IsLogging(0)) {
         log.StartDiagnostic(0) << L"DocumentParser::FetchDocument - Parse "
                                << L"error in file \"" 
                                << XMLChToVXIchar(exception.getSystemId())
                                << L"\", line " << exception.getLineNumber()
                                << L", column " << exception.getColumnNumber()
                                << L" - " 
                                << XMLChToVXIchar(exception.getMessage());
         log.EndDiagnostic();
       }
 
       return 4;
     }
 
     // (5) Write the compiled form out to the cache.
 
     doc = converter->GetDocument();
     DocumentStorageSingleton::Instance()->Store(doc, buffer, bufSize);
   }
 
   DocumentParser::ReleaseBuffer(buffer);
 
   // (6) Parse was successful, process document.  We want only the top level
   // <vxml> node.
 
   const VXMLElement root = doc.GetRoot();
   VXMLElementType nodeName = root.GetName();
 
   // If we're looking for the defaults, we can exit early.
   if (isDefaults && nodeName == DEFAULTS_ROOT) {
     log.LogDiagnostic(2, L"DocumentParser::FetchDocument - success");
     document = doc;
     return 0;
   }
   else if (nodeName != NODE_VXML) {
     document = VXMLDocument();
     if (log.IsLogging(0)) {
       log.StartDiagnostic(0) << L"DocumentParser::FetchDocument - unable to "
         L"find " << NODE_VXML << L" in document.";
       log.EndDiagnostic();
     }
     return 4;
   }
 
   vxistring temp;
 
   // If the properties map is NULL, don't bother with the remaining settings
   if (docProperties.GetValue() != NULL) {
     // Retrieve the properties and put them into the map.
 
     VXIString * str = NULL;
     temp = docURL;
     str = VXIStringCreate(temp.c_str());
     if (str == NULL) 
       throw VXIException::OutOfMemory();
 
     VXIMapSetProperty(docProperties.GetValue(), PropertyList::AbsoluteURI,
                       reinterpret_cast<VXIValue *>(str));
 
     str = VXIStringCreate(converter->GetBaseUri().empty() ? L"" : 
                           converter->GetBaseUri().c_str());
     
     if (str == NULL)
       throw VXIException::OutOfMemory();
 
     VXIMapSetProperty(docProperties.GetValue(), PropertyList::BaseURI,
                       reinterpret_cast<VXIValue *>(str));
 
     if (root.GetAttribute(ATTRIBUTE_XMLLANG, temp)) {
       str = VXIStringCreate(temp.c_str());
       if (str == NULL)
         throw VXIException::OutOfMemory();
 
       VXIMapSetProperty(docProperties.GetValue(), PropertyList::Language,
                         reinterpret_cast<VXIValue *>(str));
     }
   }
 
   log.LogDiagnostic(2, L"DocumentParser::FetchDocument - success");
 
   document = doc;
   return 0;
 }
 
 
 int DocumentParser::FetchContent(const VXIchar * uri,
                                  const VXIMapHolder & properties,
                                  VXIMapHolder & fetchInfo,
                                  VXIinetInterface * inet,
                                  SimpleLogger & log,
                                  const vxistring & encoding,
                                  vxistring & content)
 {
   const VXIbyte * buffer;
   VXIulong bufSize;
   vxistring docURL;
 
   // (1) Retrieve the URI.
   switch (DocumentParser::FetchBuffer(uri, properties, fetchInfo, inet, log,
                                       buffer, bufSize, docURL)) {
   case -1: // Out of memory?
     return -1;
   case  0: // Success
     break;
   case  2: // Unable to open URI
     return 2;
   case  3: // Unable to read from URI
     return 3;
   case  1: // Invalid parameter
   default:
     return 1;
   }
 
   // (2) Create a transcoder for the requested type.
 
   VXIcharToXMLCh encName(encoding.c_str());
   XMLTransService::Codes failReason;
   XMLTranscoder* transcoder = 
       XMLPlatformUtils::fgTransService->makeNewTranscoderFor(encName.c_str(),
                                                              failReason,
                                                              8*1064
 #if _XERCES_VERSION <= 20200                                                             
                                                               );
 #else
                                                              ,XMLPlatformUtils::fgMemoryManager);
 #endif
 
   if (transcoder == NULL) return 4;
 
   // (3) Allocate memory for the conversion.
 
   XMLCh * convertedString = new XMLCh[bufSize+1];
   unsigned char* charSizes = new unsigned char[bufSize];
 
   if (convertedString == NULL || charSizes == NULL) {
     delete[] convertedString;
     delete[] charSizes;
     return -1;
   }
 
   // (4) Transcode the values into our string.
 
   unsigned int bytesEaten;
   unsigned int charsDone = transcoder->transcodeFrom(buffer, bufSize,
                                                      convertedString, bufSize,
                                                      bytesEaten, charSizes);
 
   // (5) Finally convert from XMLCh to VXIchar.
   convertedString[charsDone] = '\0';  // Add terminator. 
   XMLChToVXIchar result(convertedString);
 
   // (6) Done.  Release memory.
 
   content = result.c_str();
   delete[] convertedString;
   delete[] charSizes;
   ReleaseBuffer(buffer);
 
   return 0;
 }