tinyxmlparser.cpp

xml 简单解析器

	if ( !p || !*p )
	{
		assert( 0 );
		return false;
	}

	const char* q = p;

	if ( ignoreCase )
	{
		while ( *q && *tag && ToLower( *q, encoding ) == ToLower( *tag, encoding ) )
		{
			++q;
			++tag;
		}

		if ( *tag == 0 )
			return true;
	}
	else
	{
		while ( *q && *tag && *q == *tag )
		{
			++q;
			++tag;
		}

		if ( *tag == 0 )		// Have we found the end of the tag, and everything equal?
			return true;
	}
	return false;
}

const char* TiXmlBase::ReadText(	const char* p, 
									TIXML_STRING * text, 
									bool trimWhiteSpace, 
									const char* endTag, 
									bool caseInsensitive,
									TiXmlEncoding encoding )
{
    *text = "";
	if (    !trimWhiteSpace			// certain tags always keep whitespace
		 || !condenseWhiteSpace )	// if true, whitespace is always kept
	{
		// Keep all the white space.
		while (	   p && *p
				&& !StringEqual( p, endTag, caseInsensitive, encoding )
			  )
		{
			int len;
			char cArr[4] = { 0, 0, 0, 0 };
			p = GetChar( p, cArr, &len, encoding );
			text->append( cArr, len );
		}
	}
	else
	{
		bool whitespace = false;

		// Remove leading white space:
		p = SkipWhiteSpace( p, encoding );
		while (	   p && *p
				&& !StringEqual( p, endTag, caseInsensitive, encoding ) )
		{
			if ( *p == '\r' || *p == '\n' )
			{
				whitespace = true;
				++p;
			}
			else if ( IsWhiteSpace( *p ) )
			{
				whitespace = true;
				++p;
			}
			else
			{
				// If we've found whitespace, add it before the
				// new character. Any whitespace just becomes a space.
				if ( whitespace )
				{
					(*text) += ' ';
					whitespace = false;
				}
				int len;
				char cArr[4] = { 0, 0, 0, 0 };
				p = GetChar( p, cArr, &len, encoding );
				if ( len == 1 )
					(*text) += cArr[0];	// more efficient
				else
					text->append( cArr, len );
			}
		}
	}
	return p + strlen( endTag );
}

#ifdef TIXML_USE_STL

void TiXmlDocument::StreamIn( TIXML_ISTREAM * in, TIXML_STRING * tag )
{
	// The basic issue with a document is that we don't know what we're
	// streaming. Read something presumed to be a tag (and hope), then
	// identify it, and call the appropriate stream method on the tag.
	//
	// This "pre-streaming" will never read the closing ">" so the
	// sub-tag can orient itself.

	if ( !StreamTo( in, '<', tag ) ) 
	{
		SetError( TIXML_ERROR_PARSING_EMPTY, 0, 0, TIXML_ENCODING_UNKNOWN );
		return;
	}

	while ( in->good() )
	{
		int tagIndex = (int) tag->length();
		while ( in->good() && in->peek() != '>' )
		{
			int c = in->get();
			if ( c <= 0 )
			{
				SetError( TIXML_ERROR_EMBEDDED_NULL, 0, 0, TIXML_ENCODING_UNKNOWN );
				break;
			}
			(*tag) += (char) c;
		}

		if ( in->good() )
		{
			// We now have something we presume to be a node of 
			// some sort. Identify it, and call the node to
			// continue streaming.
			TiXmlNode* node = Identify( tag->c_str() + tagIndex, TIXML_DEFAULT_ENCODING );

			if ( node )
			{
				node->StreamIn( in, tag );
				bool isElement = node->ToElement() != 0;
				delete node;
				node = 0;

				// If this is the root element, we're done. Parsing will be
				// done by the >> operator.
				if ( isElement )
				{
					return;
				}
			}
			else
			{
				SetError( TIXML_ERROR, 0, 0, TIXML_ENCODING_UNKNOWN );
				return;
			}
		}
	}
	// We should have returned sooner.
	SetError( TIXML_ERROR, 0, 0, TIXML_ENCODING_UNKNOWN );
}

#endif

const char* TiXmlDocument::Parse( const char* p, TiXmlParsingData* prevData, TiXmlEncoding encoding )
{
	ClearError();

	// Parse away, at the document level. Since a document
	// contains nothing but other tags, most of what happens
	// here is skipping white space.
	if ( !p || !*p )
	{
		SetError( TIXML_ERROR_DOCUMENT_EMPTY, 0, 0, TIXML_ENCODING_UNKNOWN );
		return 0;
	}

	// Note that, for a document, this needs to come
	// before the while space skip, so that parsing
	// starts from the pointer we are given.
	location.Clear();
	if ( prevData )
	{
		location.row = prevData->cursor.row;
		location.col = prevData->cursor.col;
	}
	else
	{
		location.row = 0;
		location.col = 0;
	}
	TiXmlParsingData data( p, TabSize(), location.row, location.col );
	location = data.Cursor();

	if ( encoding == TIXML_ENCODING_UNKNOWN )
	{
		// Check for the Microsoft UTF-8 lead bytes.
		const unsigned char* pU = (const unsigned char*)p;
		if (	*(pU+0) && *(pU+0) == TIXML_UTF_LEAD_0
			 && *(pU+1) && *(pU+1) == TIXML_UTF_LEAD_1
			 && *(pU+2) && *(pU+2) == TIXML_UTF_LEAD_2 )
		{
			encoding = TIXML_ENCODING_UTF8;
			useMicrosoftBOM = true;
		}
	}

    p = SkipWhiteSpace( p, encoding );
	if ( !p )
	{
		SetError( TIXML_ERROR_DOCUMENT_EMPTY, 0, 0, TIXML_ENCODING_UNKNOWN );
		return 0;
	}

	while ( p && *p )
	{
		TiXmlNode* node = Identify( p, encoding );
		if ( node )
		{
			p = node->Parse( p, &data, encoding );
			LinkEndChild( node );
		}
		else
		{
			break;
		}

		// Did we get encoding info?
		if (    encoding == TIXML_ENCODING_UNKNOWN
			 && node->ToDeclaration() )
		{
			TiXmlDeclaration* dec = node->ToDeclaration();
			const char* enc = dec->Encoding();
			assert( enc );

			if ( *enc == 0 )
				encoding = TIXML_ENCODING_UTF8;
			else if ( StringEqual( enc, "UTF-8", true, TIXML_ENCODING_UNKNOWN ) )
				encoding = TIXML_ENCODING_UTF8;
			else if ( StringEqual( enc, "UTF8", true, TIXML_ENCODING_UNKNOWN ) )
				encoding = TIXML_ENCODING_UTF8;	// incorrect, but be nice
			else 
				encoding = TIXML_ENCODING_LEGACY;
		}

		p = SkipWhiteSpace( p, encoding );
	}

	// Was this empty?
	if ( !firstChild ) {
		SetError( TIXML_ERROR_DOCUMENT_EMPTY, 0, 0, encoding );
		return 0;
	}

	// All is well.
	return p;
}

void TiXmlDocument::SetError( int err, const char* pError, TiXmlParsingData* data, TiXmlEncoding encoding )
{	
	// The first error in a chain is more accurate - don't set again!
	if ( error )
		return;

	assert( err > 0 && err < TIXML_ERROR_STRING_COUNT );
	error   = true;
	errorId = err;
	errorDesc = errorString[ errorId ];

	errorLocation.Clear();
	if ( pError && data )
	{
		data->Stamp( pError, encoding );
		errorLocation = data->Cursor();
	}
}


TiXmlNode* TiXmlNode::Identify( const char* p, TiXmlEncoding encoding )
{
	TiXmlNode* returnNode = 0;

	p = SkipWhiteSpace( p, encoding );
	if( !p || !*p || *p != '<' )
	{
		return 0;
	}

	TiXmlDocument* doc = GetDocument();
	p = SkipWhiteSpace( p, encoding );

	if ( !p || !*p )
	{
		return 0;
	}

	// What is this thing? 
	// - Elements start with a letter or underscore, but xml is reserved.
	// - Comments: <!--
	// - Decleration: <?xml
	// - Everthing else is unknown to tinyxml.
	//

	const char* xmlHeader = { "<?xml" };
	const char* commentHeader = { "<!--" };
	const char* dtdHeader = { "<!" };
	const char* cdataHeader = { "<![CDATA[" };

	if ( StringEqual( p, xmlHeader, true, encoding ) )
	{
		#ifdef DEBUG_PARSER
			TIXML_LOG( "XML parsing Declaration\n" );
		#endif
		returnNode = new TiXmlDeclaration();
	}
	else if ( StringEqual( p, commentHeader, false, encoding ) )
	{
		#ifdef DEBUG_PARSER
			TIXML_LOG( "XML parsing Comment\n" );
		#endif
		returnNode = new TiXmlComment();
	}
	else if ( StringEqual( p, cdataHeader, false, encoding ) )
	{
		#ifdef DEBUG_PARSER
			TIXML_LOG( "XML parsing CDATA\n" );
		#endif
		TiXmlText* text = new TiXmlText( "" );
		text->SetCDATA( true );
		returnNode = text;
	}
	else if ( StringEqual( p, dtdHeader, false, encoding ) )
	{
		#ifdef DEBUG_PARSER
			TIXML_LOG( "XML parsing Unknown(1)\n" );
		#endif
		returnNode = new TiXmlUnknown();
	}
	else if (    IsAlpha( *(p+1), encoding )
			  || *(p+1) == '_' )
	{
		#ifdef DEBUG_PARSER
			TIXML_LOG( "XML parsing Element\n" );
		#endif
		returnNode = new TiXmlElement( "" );
	}
	else
	{
		#ifdef DEBUG_PARSER
			TIXML_LOG( "XML parsing Unknown(2)\n" );
		#endif
		returnNode = new TiXmlUnknown();
	}

	if ( returnNode )
	{
		// Set the parent, so it can report errors
		returnNode->parent = this;
	}
	else
	{
		if ( doc )
			doc->SetError( TIXML_ERROR_OUT_OF_MEMORY, 0, 0, TIXML_ENCODING_UNKNOWN );
	}
	return returnNode;
}

#ifdef TIXML_USE_STL

void TiXmlElement::StreamIn (TIXML_ISTREAM * in, TIXML_STRING * tag)
{
	// We're called with some amount of pre-parsing. That is, some of "this"
	// element is in "tag". Go ahead and stream to the closing ">"
	while( in->good() )
	{
		int c = in->get();
		if ( c <= 0 )
		{
			TiXmlDocument* document = GetDocument();
			if ( document )
				document->SetError( TIXML_ERROR_EMBEDDED_NULL, 0, 0, TIXML_ENCODING_UNKNOWN );
			return;
		}
		(*tag) += (char) c ;
		
		if ( c == '>' )
			break;
	}

	if ( tag->length() < 3 ) return;

	// Okay...if we are a "/>" tag, then we're done. We've read a complete tag.
	// If not, identify and stream.

	if (    tag->at( tag->length() - 1 ) == '>' 
		 && tag->at( tag->length() - 2 ) == '/' )
	{
		// All good!
		return;
	}
	else if ( tag->at( tag->length() - 1 ) == '>' )
	{
		// There is more. Could be:
		//		text
		//		closing tag
		//		another node.
		for ( ;; )
		{
			StreamWhiteSpace( in, tag );

			// Do we have text?
			if ( in->good() && in->peek() != '<' ) 
			{
				// Yep, text.
				TiXmlText text( "" );
				text.StreamIn( in, tag );

				// What follows text is a closing tag or another node.
				// Go around again and figure it out.
				continue;
			}

			// We now have either a closing tag...or another node.
			// We should be at a "<", regardless.
			if ( !in->good() ) return;
			assert( in->peek() == '<' );
			int tagIndex = (int) tag->length();

			bool closingTag = false;
			bool firstCharFound = false;

			for( ;; )
			{
				if ( !in->good() )
					return;

				int c = in->peek();
				if ( c <= 0 )
				{
					TiXmlDocument* document = GetDocument();
					if ( document )
						document->SetError( TIXML_ERROR_EMBEDDED_NULL, 0, 0, TIXML_ENCODING_UNKNOWN );
					return;
				}
				
				if ( c == '>' )
					break;

				*tag += (char) c;
				in->get();

				if ( !firstCharFound && c != '<' && !IsWhiteSpace( c ) )
				{
					firstCharFound = true;
					if ( c == '/' )
						closingTag = true;
				}
			}
			// If it was a closing tag, then read in the closing '>' to clean up the input stream.
			// If it was not, the streaming will be done by the tag.
			if ( closingTag )
			{
				if ( !in->good() )
					return;

				int c = in->get();
				if ( c <= 0 )
				{
					TiXmlDocument* document = GetDocument();
					if ( document )
						document->SetError( TIXML_ERROR_EMBEDDED_NULL, 0, 0, TIXML_ENCODING_UNKNOWN );
					return;
				}
				assert( c == '>' );
				*tag += (char) c;

				// We are done, once we've found our closing tag.
				return;
			}
			else
			{
				// If not a closing tag, id it, and stream.
				const char* tagloc = tag->c_str() + tagIndex;
				TiXmlNode* node = Identify( tagloc, TIXML_DEFAULT_ENCODING );
				if ( !node )
					return;
				node->StreamIn( in, tag );
				delete node;
				node = 0;

				// No return: go around from the beginning: text, closing tag, or node.
			}
		}
	}
}
#endif

const char* TiXmlElement::Parse( const char* p, TiXmlParsingData* data, TiXmlEncoding encoding )
{
	p = SkipWhiteSpace( p, encoding );
	TiXmlDocument* document = GetDocument();

	if ( !p || !*p )
	{
		if ( document ) document->SetError( TIXML_ERROR_PARSING_ELEMENT, 0, 0, encoding );
		return 0;
	}

	if ( data )
	{
		data->Stamp( p, encoding );
		location = data->Cursor();
	}

	if ( *p != '<' )
	{
		if ( document ) document->SetError( TIXML_ERROR_PARSING_ELEMENT, p, data, encoding );
		return 0;
	}

	p = SkipWhiteSpace( p+1, encoding );

	// Read the name.
	const char* pErr = p;

    p = ReadName( p, &value, encoding );
	if ( !p || !*p )
	{
		if ( document )	document->SetError( TIXML_ERROR_FAILED_TO_READ_ELEMENT_NAME, pErr, data, encoding );
		return 0;