documentconverter.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”

       ParseException(L"exactly one of 'aai' or 'aaiexpr' may be specified");
     break;
   }
 
   // (6) Verify the node.
   ProcessNodeFinal(elemType);
 }
 
 
 void DocumentConverter::endElement(const XMLCh* const uri,
                                    const XMLCh* const localname,
                                    const XMLCh* const qname)
 {
   // (0) Just ignore this?
   if (ignoreDepth > 0) {
     --ignoreDepth;
     return;
   }
 
   // (1) Copy this element or look it up?
 
   // (1.1) Inside a <grammar>, we can blindly copy this data.
   if (inGrammar && copyDepth == 0) inGrammar = false;
   if (inGrammar) {
     CopyElementClose(localname);
     return;
   }
 
   // (1.2) Convert name string to integer value.
   XMLChToVXIchar elementName(localname);
 
   int elemType;
   bool conversionFailed = !ConvertElement(elementName.c_str(), elemType);
 
   // (1.3) Copy almost anything inside a <prompt>.
   if (explicitPrompt) {
     if (conversionFailed) {
       CopyElementClose(localname);
       return;
     }
     switch (elemType) {
     case NODE_AUDIO:
     case NODE_VALUE:
     case NODE_ENUMERATE:
       break;
     case NODE_PROMPT:
       doc->AddContent(L"</speak>", 8);
       explicitPrompt = false;
       break;
     default:
       CopyElementClose(localname);
       return;
     }
   }
 
   // (1.4) Print errors for all other conversion failures.
   if (conversionFailed) {
     vxistring temp(L"unrecognized element - ");
     temp += elementName.c_str();
     ParseException(temp.c_str());
   }
 
   // (2) Handle implicit prompts.
 
   if (implicitPrompt) {
     switch (elemType) {
     case NODE_AUDIO:
     case NODE_VALUE:
     case NODE_ENUMERATE:
       break;
     case PRIV_ELEM_BREAK:   // These implicitly imply a prompt.
     case PRIV_ELEM_EMPHASIS:
     case PRIV_ELEM_MARK:
     case PRIV_ELEM_PARAGRAPH:
     case PRIV_ELEM_PHONEME:
     case PRIV_ELEM_PROSODY:
     case PRIV_ELEM_SAYAS:
     case PRIV_ELEM_SENTENCE:
     case PRIV_ELEM_VOICE:
       CopyElementClose(localname);
       return;
     default:
       EndImplicitPrompt();
       break;
     }
   }
 
   // (3) Reverse special handling from startElement()
 
   switch (elemType) {
   case NODE_CHOICE:          // These may have PCDATA which is not a prompt
   case NODE_OPTION:
     doc->PruneWhitespace();
     pcdataImpliesPrompt = true;
   case NODE_LOG:
     pcdataImpliesPrompt = true;
     break;
   case NODE_SCRIPT:
     pcdataImpliesPrompt = true;
     // Intentional fall through!
   case NODE_GRAMMAR:
     if (contentForbidden ^ hasContent)
       contentForbidden = false;
     else
       ParseException(L"either the 'src' attribute or inlined content "
                      L"may be provided, not both");
     break;
   default:
     break;
   }
 
   // (4) This element is complete.
 
   try {
     doc->EndElement();
   }
   catch (const VXMLDocumentModel::InternalError &) {
     ParseException(L"corrupted document tree; unable to complete element");
   }
 }
 
 
 void DocumentConverter::characters(const XMLCh* const chars,
                                    const unsigned int length)
 {
   // (0) Just ignore this?
   if (chars == NULL || length == 0 || ignoreDepth > 0) return;
 
   // (1) Ignore non-empty prompts consisting of pure whitespace
   unsigned int l;
   for (l = 0; l < length; ++l) {
     XMLCh c = chars[l];
     if (c == ' ' || c == '\n' || c == '\r' || c == '\t') continue;
     break;
   }
   if (l == length) return;
 
   // (2) Is this an implicit prompt?
   if (pcdataImpliesPrompt && !explicitPrompt && !implicitPrompt && !inGrammar)
     StartImplicitPrompt();
 
   // (3) Add content to the document model.
   try {
     // This conversion should be safe.  CDATA may only contain valid XML
     // characters.  These are defined in the XML 1.0 specification as the set
     // 0x9, 0xA, 0xD, 0x20-0xD7FF, 0xE000-0xFFFD, 0x10000-0x10FFFF.
     XMLChToVXIchar data(chars);
     hasContent = true;
     
     unsigned int start = 0;
     unsigned int pos = 0;
     VXIchar c;
     // We need to escape two characters: (<,&) -> (&lt;, &amp;)
     if((inGrammar && (GetGrammarType().empty() || GetGrammarType() == L"application/srgs+xml")) ||
         (explicitPrompt || implicitPrompt)
         ) 
     {
       do {
         c = *(data.c_str() + pos);
         if (c == L'<') {
           if (start != pos)
              doc->AddContent(data.c_str() + start, pos - start);
           start = pos + 1;
           doc->AddContent(L"&lt;", 4);
         }
         else if (c == L'&') {
           if (start != pos)
              doc->AddContent(data.c_str() + start, pos - start);
           start = pos + 1;
           doc->AddContent(L"&amp;", 5);
         }
         
         ++pos;
       } while (c != L'\0');
   
       if (pos > start)
         doc->AddContent(data.c_str() + start, pos - start - 1); // -1 to kill off final \0
   
     } else {
       doc->AddContent(data.c_str(), wcslen(data.c_str()));
     }
   }
   catch (const VXMLDocumentModel::OutOfMemory &) {
     ParseException(L"unable to allocate memory for content");
   }
   catch (const VXMLDocumentModel::InternalError &) {
     ParseException(L"corrupted document tree; unable to add content");
   }
 }
 
 
 void DocumentConverter::ignorableWhitespace(const XMLCh* const chars,
                                             const unsigned int l)
 { }
 
 
 void DocumentConverter::processingInstruction(const XMLCh* const target,
                                               const XMLCh* const data)
 { }
 
 //#############################################################################
 
 void DocumentConverter::ProcessNodeAttribute(VXMLElementType elemType,
                                              int attrType,
                                              const VXIchar* const value)
 {
   // This shouldn't ever happen, if it does, we'll just ignore it.
   if (value == NULL) return;
 
   switch (elemType) {
   case NODE_BLOCK:
   case NODE_FIELD:
   case NODE_INITIAL:
   case NODE_OBJECT:
   case NODE_RECORD:
   case NODE_TRANSFER:
     // We associate a hidden interal name with this element.
     if (attrType == ATTRIBUTE_NAME) attrType = ATTRIBUTE__ITEMNAME;
     break;
   case NODE_CLEAR:
     if (attrType == ATTRIBUTE_NAMELIST && !value[0])
       ParseException(L"the namelist attribute on clear cannot be empty");
     break;
   case NODE_FORM:
   case NODE_MENU:
     // We associate a hidden interal name with this element.
     if (attrType == ATTRIBUTE_ID) attrType = ATTRIBUTE__ITEMNAME;
     break;
   case NODE_FILLED:
     if (doc->GetParentType() != NODE_FORM)
       ParseException(L"attributes valid on filled only at form level");
     if (attrType == ATTRIBUTE_NAMELIST && !value[0])
       ParseException(L"the namelist attribute on filled cannot be empty");
     break;
   case NODE_GRAMMAR:
     if (doc->GetParentType() != NODE_FORM && attrType == ATTRIBUTE_SCOPE)
       ParseException(L"the scope attribute is valid only on grammars at form "
                      L"level");
     break;
   case NODE_SUBDIALOG:
     // We associate a hidden interal name with this element.
     if (attrType == ATTRIBUTE_NAME) attrType = ATTRIBUTE__ITEMNAME;
     // modal was dropped in 2.0.
     if (attrType == ATTRIBUTE_MODAL && version != 1.0f)
       ParseException(L"the modal attribute on subdialog was dropped after "
                      L"1.0");
     break;
   case NODE_VALUE:
     if (attrType == ATTRIBUTE_EXPR) break;
 
     if (doc->GetParentType() == NODE_LOG)
       ParseException(L"only the expr attribute is valid on value elements "
                      L"within a log element");
 
 #pragma message("JC: Obsolete???")
     /*
     if (doc->GetParentType() == NODE_SAYAS)
       ParseException(L"only the expr attribute is valid on value elements "
                      L"within a say-as element");
     */
     break;
   case NODE_VXML:
     // version is processed elsewhere and may be ignored.
     if (attrType == ATTRIBUTE_VERSION) return;
     if (attrType == PRIV_ATTRIB_SCHEMALOC) return;
     break;
   default:
     break;
   }
 
   // (4.3) Add the attribute to the element.
 
   if (attrType > PRIV_ATTRIB_RangeStart)
     ParseException(L"internal error during attribute processing");
 
   doc->AddAttribute(VXMLAttributeType(attrType), value);
 }
 
 
 void DocumentConverter::ProcessNodeFinal(VXMLElementType elemType)
 {
   // Convert attributes.
   vxistring attr;
 
   switch (elemType) {
   case NODE_BLOCK:
   case NODE_FIELD:
   case NODE_FORM:
   case NODE_INITIAL:
   case NODE_MENU:
   case NODE_OBJECT:
   case NODE_RECORD:
   case NODE_SUBDIALOG:
   case NODE_TRANSFER:
     // Name the 'unnamed' elements as neccessary.
     if (!doc->GetAttribute(ATTRIBUTE__ITEMNAME, attr)) {
       vxistring variable;
       VXMLDocumentModel::CreateHiddenVariable(variable);
       doc->AddAttribute(ATTRIBUTE__ITEMNAME, variable);
     }
     break;
   case NODE_FILLED:
     if (!doc->GetAttribute(ATTRIBUTE_MODE, attr))
       doc->AddAttribute(ATTRIBUTE_MODE, L"all");
     break;
   case NODE_VXML:
   case DEFAULTS_ROOT:
     VXMLDocumentModel::CreateHiddenVariable(attr);
     doc->AddAttribute(ATTRIBUTE__ITEMNAME, attr);
     break;
   default:
     break;
   }
 
   // Generate DTMF sequences for <choice> elements in <menu> if necessary.
   if (elemType == NODE_MENU) {
     if (doc->GetAttribute(ATTRIBUTE_DTMF, attr) && attr == L"true")
       choiceNumber = 1;
     else
       choiceNumber = 0;
   }
 
   if (elemType == NODE_CHOICE && choiceNumber > 0 && choiceNumber < 10 &&
       !doc->GetAttribute(ATTRIBUTE_DTMF, attr))
   {
     std::basic_stringstream<VXIchar> countString;
     countString << choiceNumber;
     doc->AddAttribute(ATTRIBUTE_DTMF, countString.str());
     ++choiceNumber;
   }
 
   if (elemType == NODE_PROMPT) {
     vxistring base, lang;
     doc->GetAttribute(ATTRIBUTE_BASE, base);
     doc->GetAttribute(ATTRIBUTE_XMLLANG, lang);
 
     doc->AddContent(L"<?xml version='1.0'?><speak version='1.0' xmlns="
                     L"'http://www.w3.org/2001/10/synthesis' xml:base='", 96);
     doc->AddContent(base.c_str(), base.length());
     doc->AddContent(L"' xml:lang='", 12);
     doc->AddContent(lang.c_str(), lang.length());
     doc->AddContent(L"'>", 2);
   }
 }
 
 
 void DocumentConverter::CopyElementStart(const XMLCh* const localname,
                                          const Attributes &attrs)
 {
   ++copyDepth;
 
   doc->AddContent(L"<", 1);
 
   XMLChToVXIchar data(localname);
   doc->AddContent(data.c_str(), wcslen(data.c_str()));
 
   for (unsigned int index = 0; index < attrs.getLength(); ++index) {
     doc->AddContent(L" ", 1);
 
     // Determine the attribute name
     if (Compare(attrs.getQName(index), L"xml:lang"))
       doc->AddContent(L"xml:lang", 8); 
     else {
       XMLChToVXIchar name(attrs.getLocalName(index));
       doc->AddContent(name.c_str(), wcslen(name.c_str()));
     }
       
     doc->AddContent(L"=\"", 2);
 
     XMLChToVXIchar value(attrs.getValue(index));
 
     unsigned int start = 0;
     unsigned int pos = 0;
     VXIchar c;
     // We need to escape three characters: (<,>,&) -> (&lt;,&gt;,&amp;)
     do {
       c = *(value.c_str() + pos);
       if (c == L'<') {
         if (start != pos)
            doc->AddContent(value.c_str() + start, pos - start);
         start = pos + 1;
         doc->AddContent(L"&lt;", 4);
       }
       else if (c == L'>') {
         if (start != pos)
            doc->AddContent(value.c_str() + start, pos - start);
         start = pos + 1;
         doc->AddContent(L"&gt;", 4);
       }
       else if (c == L'&') {
         if (start != pos)
            doc->AddContent(value.c_str() + start, pos - start);
         start = pos + 1;
         doc->AddContent(L"&amp;", 5);
       }
       
       ++pos;
     } while (c != L'\0');
 
     if (pos > start)
       doc->AddContent(value.c_str() + start, pos - start - 1); // -1 to kill off final \0
 
     doc->AddContent(L"\"", 1);
   }
 
   doc->AddContent(L">", 1);
 }
 
 
 void DocumentConverter::CopyElementClose(const XMLCh* const localname)
 {
   doc->AddContent(L"</", 2);
 
   XMLChToVXIchar data(localname);
   doc->AddContent(data.c_str(), wcslen(data.c_str()));
 
   doc->AddContent(L">", 1);
 
   --copyDepth;
 }
 
 
 void DocumentConverter::StartImplicitPrompt()
 {
   implicitPrompt = true;
   doc->StartElement(NODE_PROMPT);
 
   doc->AddContent(L"<?xml version='1.0'?><speak version='1.0' xmlns="
                   L"'http://www.w3.org/2001/10/synthesis' xml:base='", 96);
   doc->AddContent(baseUri.c_str(), baseUri.length());
   doc->AddContent(L"' xml:lang='", 12);
   doc->AddContent(documentLang.c_str(), documentLang.length());
   doc->AddContent(L"'>", 2);
 
   copyDepth = 0;
 }
 
 
 void DocumentConverter::EndImplicitPrompt()
 {
   doc->AddContent(L"</speak>", 8);
   doc->EndElement();
   implicitPrompt = false;
 }