parserengine.cpp

C++ class libraries for network-centric, portable applications, integrated perfectly with the C++ St

	}

	XML_SetUserData(_parser, this);
	XML_SetElementHandler(_parser, handleStartElement, handleEndElement);
	XML_SetCharacterDataHandler(_parser, handleCharacterData);
	XML_SetProcessingInstructionHandler(_parser, handleProcessingInstruction);
	if (_expandInternalEntities)
		XML_SetDefaultHandlerExpand(_parser, handleDefault);
	else
		XML_SetDefaultHandler(_parser, handleDefault);
	XML_SetUnparsedEntityDeclHandler(_parser, handleUnparsedEntityDecl);
	XML_SetNotationDeclHandler(_parser, handleNotationDecl);
	XML_SetExternalEntityRefHandler(_parser, handleExternalEntityRef);
	XML_SetCommentHandler(_parser, handleComment);
	XML_SetCdataSectionHandler(_parser, handleStartCdataSection, handleEndCdataSection);
	XML_SetDoctypeDeclHandler(_parser, handleStartDoctypeDecl, handleEndDoctypeDecl);
	XML_SetEntityDeclHandler(_parser, handleEntityDecl);
	XML_SetSkippedEntityHandler(_parser, handleSkippedEntity);
	XML_SetParamEntityParsing(_parser, _externalParameterEntities ? XML_PARAM_ENTITY_PARSING_ALWAYS : XML_PARAM_ENTITY_PARSING_NEVER);
	XML_SetUnknownEncodingHandler(_parser, handleUnknownEncoding, this);
}


void ParserEngine::handleError(int errorNo)
{
	try
	{
		switch (errorNo)
		{
		case XML_ERROR_NO_MEMORY:
			throw XMLException("No memory");
		case XML_ERROR_SYNTAX:
			throw SAXParseException("Syntax error", locator());
		case XML_ERROR_NO_ELEMENTS:
			throw SAXParseException("No element found", locator());
		case XML_ERROR_INVALID_TOKEN:
			throw SAXParseException("Invalid token", locator());
		case XML_ERROR_UNCLOSED_TOKEN:
			throw SAXParseException("Unclosed token", locator());
		case XML_ERROR_PARTIAL_CHAR:
			throw SAXParseException("Partial character", locator());
		case XML_ERROR_TAG_MISMATCH:
			throw SAXParseException("Tag mismatch", locator());
		case XML_ERROR_DUPLICATE_ATTRIBUTE:
			throw SAXParseException("Duplicate attribute", locator());
		case XML_ERROR_JUNK_AFTER_DOC_ELEMENT:
			throw SAXParseException("Junk after document element", locator());
		case XML_ERROR_PARAM_ENTITY_REF:
			throw SAXParseException("Illegal parameter entity reference", locator());
		case XML_ERROR_UNDEFINED_ENTITY:
			throw SAXParseException("Undefined entity", locator());
		case XML_ERROR_RECURSIVE_ENTITY_REF:
			throw SAXParseException("Recursive entity reference", locator());
		case XML_ERROR_ASYNC_ENTITY:
			throw SAXParseException("Asynchronous entity", locator());
		case XML_ERROR_BAD_CHAR_REF:
			throw SAXParseException("Reference to invalid character number", locator());
		case XML_ERROR_BINARY_ENTITY_REF:
			throw SAXParseException("Reference to binary entity", locator());
		case XML_ERROR_ATTRIBUTE_EXTERNAL_ENTITY_REF:
			throw SAXParseException("Reference to external entity in attribute", locator());
		case XML_ERROR_MISPLACED_XML_PI:
			throw SAXParseException("XML processing instruction not at start of external entity", locator());
		case XML_ERROR_UNKNOWN_ENCODING:
			throw SAXParseException("Unknown encoding", locator());
		case XML_ERROR_INCORRECT_ENCODING:
			throw SAXParseException("Encoding specified in XML declaration is incorrect", locator());
		case XML_ERROR_UNCLOSED_CDATA_SECTION:
			throw SAXParseException("Unclosed CDATA section", locator());
		case XML_ERROR_EXTERNAL_ENTITY_HANDLING:
			throw SAXParseException("Error in processing external entity reference", locator());
		case XML_ERROR_NOT_STANDALONE:
			throw SAXParseException("Document is not standalone", locator());
		case XML_ERROR_UNEXPECTED_STATE:
			throw SAXParseException("Unexpected parser state - please send a bug report", locator());		
		case XML_ERROR_ENTITY_DECLARED_IN_PE:
			throw SAXParseException("Entity declared in parameter entity", locator());
		case XML_ERROR_FEATURE_REQUIRES_XML_DTD:
			throw SAXParseException("Requested feature requires XML_DTD support in Expat", locator());
		case XML_ERROR_CANT_CHANGE_FEATURE_ONCE_PARSING:
			throw SAXParseException("Cannot change setting once parsing has begun", locator());
		case XML_ERROR_UNBOUND_PREFIX:
			throw SAXParseException("Unbound prefix", locator());
		case XML_ERROR_UNDECLARING_PREFIX:
			throw SAXParseException("Must not undeclare prefix", locator());
		case XML_ERROR_INCOMPLETE_PE:
			throw SAXParseException("Incomplete markup in parameter entity", locator());
		case XML_ERROR_XML_DECL:
			throw SAXParseException("XML declaration not well-formed", locator());
		case XML_ERROR_TEXT_DECL:
			throw SAXParseException("Text declaration not well-formed", locator());
		case XML_ERROR_PUBLICID:
			throw SAXParseException("Illegal character(s) in public identifier", locator());
		case XML_ERROR_SUSPENDED:
			throw SAXParseException("Parser suspended", locator());
		case XML_ERROR_NOT_SUSPENDED:
			throw SAXParseException("Parser not suspended", locator());
		case XML_ERROR_ABORTED:
			throw SAXParseException("Parsing aborted", locator());
		case XML_ERROR_FINISHED:
			throw SAXParseException("Parsing finished", locator());
		case XML_ERROR_SUSPEND_PE:
			throw SAXParseException("Cannot suspend in external parameter entity", locator());
		}
		throw XMLException("Unknown Expat error code");
	}
	catch (SAXException& exc)
	{
		if (_pErrorHandler) _pErrorHandler->error(exc);
		throw;
	}
	catch (Poco::Exception& exc)
	{
		if (_pErrorHandler) _pErrorHandler->fatalError(SAXParseException("Fatal error", locator(), exc));
		throw;
	}
}	


void ParserEngine::pushContext(XML_Parser parser, InputSource* pInputSource)
{
	ContextLocator* pLocator = new ContextLocator(parser, pInputSource->getPublicId(), pInputSource->getSystemId());
	_context.push_back(pLocator);
}


void ParserEngine::popContext()
{
	poco_assert (!_context.empty());
	delete _context.back();
	_context.pop_back();
}


void ParserEngine::resetContext()
{
	for (ContextStack::iterator it = _context.begin(); it != _context.end(); ++it)
	{
		delete *it;
	}
	_context.clear();
}


void ParserEngine::handleStartElement(void* userData, const XML_Char* name, const XML_Char** atts)
{
	ParserEngine* pThis = reinterpret_cast<ParserEngine*>(userData);
	
	if (pThis->_pContentHandler)
	{
		try
		{
			pThis->_pNamespaceStrategy->startElement(name, atts, XML_GetSpecifiedAttributeCount(pThis->_parser)/2, pThis->_pContentHandler);	
		}
		catch (XMLException& exc)
		{
			throw SAXParseException(exc.message(), pThis->locator());
		}
	}
}


void ParserEngine::handleEndElement(void* userData, const XML_Char* name)
{
	ParserEngine* pThis = reinterpret_cast<ParserEngine*>(userData);
	
	if (pThis->_pContentHandler)
	{
		try
		{
			pThis->_pNamespaceStrategy->endElement(name, pThis->_pContentHandler);	
		}
		catch (XMLException& exc)
		{
			throw SAXParseException(exc.message(), pThis->locator());
		}
	}
}


void ParserEngine::handleCharacterData(void* userData, const XML_Char* s, int len)
{
	ParserEngine* pThis = reinterpret_cast<ParserEngine*>(userData);
	
	if (pThis->_pContentHandler)
		pThis->_pContentHandler->characters(s, 0, len);
}


void ParserEngine::handleProcessingInstruction(void* userData, const XML_Char* target, const XML_Char* data)
{
	ParserEngine* pThis = reinterpret_cast<ParserEngine*>(userData);
	
	if (pThis->_pContentHandler)
		pThis->_pContentHandler->processingInstruction(target, data);
}


void ParserEngine::handleDefault(void* userData, const XML_Char* s, int len)
{
}


void ParserEngine::handleUnparsedEntityDecl(void* userData, const XML_Char* entityName, const XML_Char* base, const XML_Char* systemId, const XML_Char* publicId, const XML_Char* notationName)
{
	ParserEngine* pThis = reinterpret_cast<ParserEngine*>(userData);
	
	XMLString pubId;
	if (publicId) pubId.assign(publicId);
	if (pThis->_pDTDHandler) 
		pThis->_pDTDHandler->unparsedEntityDecl(entityName, publicId ? &pubId : 0, systemId, notationName);
}


void ParserEngine::handleNotationDecl(void* userData, const XML_Char* notationName, const XML_Char* base, const XML_Char* systemId, const XML_Char* publicId)
{
	ParserEngine* pThis = reinterpret_cast<ParserEngine*>(userData);
	
	XMLString pubId;
	if (publicId) pubId.assign(publicId);
	XMLString sysId;
	if (systemId) sysId.assign(systemId);
	if (pThis->_pDTDHandler) 
		pThis->_pDTDHandler->notationDecl(notationName, publicId ? &pubId : 0, systemId ? &sysId : 0);
}


int ParserEngine::handleExternalEntityRef(XML_Parser parser, const XML_Char* context, const XML_Char* base, const XML_Char* systemId, const XML_Char* publicId)
{
	ParserEngine* pThis = reinterpret_cast<ParserEngine*>(XML_GetUserData(parser));

	if (!context && !pThis->_externalParameterEntities) return XML_STATUS_ERROR;
	if (context && !pThis->_externalGeneralEntities) return XML_STATUS_ERROR;

	InputSource* pInputSource = 0;
	EntityResolver* pEntityResolver = 0;
	EntityResolverImpl defaultResolver;

	XMLString sysId(systemId);
	XMLString pubId;
	if (publicId) pubId.assign(publicId);
	
	URI uri(fromXMLString(pThis->_context.back()->getSystemId()));
	uri.resolve(fromXMLString(sysId));

	if (pThis->_pEntityResolver)
	{
		pEntityResolver = pThis->_pEntityResolver;
		pInputSource = pEntityResolver->resolveEntity(publicId ? &pubId : 0, toXMLString(uri.toString()));
	}
	if (!pInputSource && pThis->_externalGeneralEntities)
	{
		pEntityResolver = &defaultResolver;
		pInputSource = pEntityResolver->resolveEntity(publicId ? &pubId : 0, toXMLString(uri.toString()));
	}

	if (pInputSource)
	{
		XML_Parser extParser = XML_ExternalEntityParserCreate(pThis->_parser, context, 0);
		try
		{
			pThis->parseExternal(extParser, pInputSource);
		}
		catch (XMLException&)
		{
			pEntityResolver->releaseInputSource(pInputSource);
			XML_ParserFree(extParser);
			throw;
		}
		pEntityResolver->releaseInputSource(pInputSource);
		XML_ParserFree(extParser);
		return XML_STATUS_OK;
	}
	else return XML_STATUS_ERROR;
}


int ParserEngine::handleUnknownEncoding(void* encodingHandlerData, const XML_Char* name, XML_Encoding* info)
{
	ParserEngine* pThis = reinterpret_cast<ParserEngine*>(encodingHandlerData);
	
	XMLString encoding(name);
	EncodingMap::const_iterator it = pThis->_encodings.find(encoding);
	if (it != pThis->_encodings.end())
	{
		const TextEncoding::CharacterMap& map = it->second->characterMap();
		for (int i = 0; i < 256; ++i)
			info->map[i] = map[i];
			
		info->data    = it->second;
		info->convert = &ParserEngine::convert;
		info->release = 0;
		return XML_STATUS_OK;
	}
	else return XML_STATUS_ERROR;
}


void ParserEngine::handleComment(void* userData, const XML_Char* data)
{
	ParserEngine* pThis = reinterpret_cast<ParserEngine*>(userData);

#if defined(XML_UNICODE_WCHAR_T)
	if (pThis->_pLexicalHandler)
		pThis->_pLexicalHandler->comment(data, 0, (int) wcslen(data));
#else
	if (pThis->_pLexicalHandler)
		pThis->_pLexicalHandler->comment(data, 0, (int) strlen(data));
#endif
}


void ParserEngine::handleStartCdataSection(void* userData)
{
	ParserEngine* pThis = reinterpret_cast<ParserEngine*>(userData);

	if (pThis->_pLexicalHandler)
		pThis->_pLexicalHandler->startCDATA();
}


void ParserEngine::handleEndCdataSection(void* userData)
{
	ParserEngine* pThis = reinterpret_cast<ParserEngine*>(userData);

	if (pThis->_pLexicalHandler)
		pThis->_pLexicalHandler->endCDATA();
}


void ParserEngine::handleStartNamespaceDecl(void* userData, const XML_Char* prefix, const XML_Char* uri)
{
	ParserEngine* pThis = reinterpret_cast<ParserEngine*>(userData);

	if (pThis->_pContentHandler)
		pThis->_pContentHandler->startPrefixMapping((prefix ? XMLString(prefix) : EMPTY_STRING), (uri ? XMLString(uri) : EMPTY_STRING));
}


void ParserEngine::handleEndNamespaceDecl(void* userData, const XML_Char* prefix)
{
	ParserEngine* pThis = reinterpret_cast<ParserEngine*>(userData);

	if (pThis->_pContentHandler)
		pThis->_pContentHandler->endPrefixMapping(prefix ? XMLString(prefix) : EMPTY_STRING);
}


void ParserEngine::handleStartDoctypeDecl(void* userData, const XML_Char* doctypeName, const XML_Char *systemId, const XML_Char* publicId, int hasInternalSubset)
{
	ParserEngine* pThis = reinterpret_cast<ParserEngine*>(userData);

	if (pThis->_pLexicalHandler)
	{
		XMLString sysId = systemId ? XMLString(systemId) : EMPTY_STRING;
		XMLString pubId = publicId ? XMLString(publicId) : EMPTY_STRING;
		pThis->_pLexicalHandler->startDTD(doctypeName, pubId, sysId);
	}
}


void ParserEngine::handleEndDoctypeDecl(void* userData)
{
	ParserEngine* pThis = reinterpret_cast<ParserEngine*>(userData);

	if (pThis->_pLexicalHandler)
		pThis->_pLexicalHandler->endDTD();
}


void ParserEngine::handleEntityDecl(void *userData, const XML_Char *entityName, int isParamEntity, const XML_Char *value, int valueLength, 
	                                const XML_Char *base, const XML_Char *systemId, const XML_Char *publicId, const XML_Char *notationName)
{
	if (value)
		handleInternalParsedEntityDecl(userData, entityName, value, valueLength);
	else
		handleExternalParsedEntityDecl(userData, entityName, base, systemId, publicId);
}


void ParserEngine::handleExternalParsedEntityDecl(void* userData, const XML_Char* entityName, const XML_Char* base, const XML_Char* systemId, const XML_Char* publicId)
{
	ParserEngine* pThis = reinterpret_cast<ParserEngine*>(userData);

	XMLString pubId;
	if (publicId) pubId.assign(publicId);
	if (pThis->_pDeclHandler)
		pThis->_pDeclHandler->externalEntityDecl(entityName, publicId ? &pubId : 0, systemId);
}


void ParserEngine::handleInternalParsedEntityDecl(void* userData, const XML_Char* entityName, const XML_Char* replacementText, int replacementTextLength)
{
	ParserEngine* pThis = reinterpret_cast<ParserEngine*>(userData);

	XMLString replText(replacementText, replacementTextLength);
	if (pThis->_pDeclHandler)
		pThis->_pDeclHandler->internalEntityDecl(entityName, replText);
}


void ParserEngine::handleSkippedEntity(void* userData, const XML_Char* entityName, int isParameterEntity)
{
	ParserEngine* pThis = reinterpret_cast<ParserEngine*>(userData);
	
	if (pThis->_pContentHandler)
		pThis->_pContentHandler->skippedEntity(entityName);
}


int ParserEngine::convert(void* data, const char* s)
{
	TextEncoding* pEncoding = reinterpret_cast<TextEncoding*>(data);
	return pEncoding->convert((const unsigned char*) s);
}


} } // namespace Poco::XML