markupstl.cpp

解析xml文件的强大包

	{
		// Starting at token.nNext, bypass whitespace and find the next token
		nChar = token.nNext;
		nPreSpaceStart = nChar;
		if ( ! x_FindAny(szDoc,nChar) )
			break;
		nPreSpaceLength = nChar - nPreSpaceStart;

		// Is it an opening quote?
		cFirstChar = szDoc[nChar];
		if ( cFirstChar == '\"' || cFirstChar == '\'' )
		{
			token.nTokenFlags |= MNF_QUOTED;

			// Move past opening quote
			++nChar;
			token.nL = nChar;

			// Look for closing quote
			while ( szDoc[nChar] && szDoc[nChar] != cFirstChar )
				++nChar;

			// Set right to before closing quote
			token.nR = nChar - 1;

			// Set nChar past closing quote unless at end of document
			if ( szDoc[nChar] )
				++nChar;
		}
		else
		{
			token.nTokenFlags &= ~MNF_QUOTED;

			// Go until special char or whitespace
			token.nL = nChar;
			if ( bAfterEqual )
			{
				while ( szDoc[nChar] && ! strchr(" \t\n\r>",szDoc[nChar]) )
					++nChar;
			}
			else
			{
				while ( szDoc[nChar] && ! strchr("= \t\n\r>/?",szDoc[nChar]) )
					++nChar;
			}

			// Adjust end position if it is one special char
			if ( nChar == token.nL )
				++nChar; // it is a special char
			token.nR = nChar - 1;
		}

		// nNext points to one past last char of token
		token.nNext = nChar;

		if ( ! bAfterEqual && ! (token.nTokenFlags&MNF_QUOTED) )
		{
			// Is it an equal sign?
			char cChar = szDoc[token.nL];
			if ( cChar == '=' )
			{
				bAfterEqual = true;
				continue;
			}

			// Is it the right angle bracket?
			if ( cChar == '>' || cChar == '/' || cChar == '?' )
			{
				token.nNext = nPreSpaceStart;
				break; // attrib not found
			}

			if ( nFoundAttribNameR )
				break;

			// Attribute name
			if ( nAttrib != -1 )
			{
				if ( ! szAttrib )
				{
					if ( nAttrib == n )
						return true; // found by number
				}
				else if ( token.Match(szAttrib) )
				{
					// Matched attrib name, go forward to value
					nFoundAttribNameR = token.nR;
					token.nPreSpaceStart = nPreSpaceStart;
					token.nPreSpaceLength = nPreSpaceLength;
				}
			}
			++nAttrib;
		}
		else if ( nFoundAttribNameR )
			break;
		bAfterEqual = false;
	}

	if ( nFoundAttribNameR )
	{
		if ( ! bAfterEqual )
		{
			// when attribute has no value the value is the attribute name
			token.nL = token.nPreSpaceStart + token.nPreSpaceLength;
			token.nR = nFoundAttribNameR;
			token.nNext = nFoundAttribNameR + 1;
		}
		return true; // found by name
	}
	return false; // not found
}

string CMarkupSTL::x_GetAttrib( int iPos, const char* szAttrib ) const
{
	// Return the value of the attrib
	TokenPos token( m_strDoc, m_nFlags );
	if ( iPos && m_nNodeType == MNT_ELEMENT )
		token.nNext = m_aPos[iPos].nStart + 1;
	else if ( iPos == m_iPos && m_nNodeLength && m_nNodeType == MNT_PROCESSING_INSTRUCTION )
		token.nNext = m_nNodeOffset + 2;
	else
		return "";

	if ( szAttrib && x_FindAttrib( token, szAttrib ) )
		return UnescapeText( &token.szDoc[token.nL], token.Length() );
	return "";
}

bool CMarkupSTL::x_SetAttrib( int iPos, const char* szAttrib, int nValue )
{
	// Convert integer to string and call SetChildAttrib
	char szVal[25];
	sprintf( szVal, "%d", nValue );
	return x_SetAttrib( iPos, szAttrib, szVal );
}

bool CMarkupSTL::x_SetAttrib( int iPos, const char* szAttrib, const char* szValue )
{
	// Set attribute in iPos element
	TokenPos token( m_strDoc, m_nFlags );
	if ( iPos && m_nNodeType == MNT_ELEMENT )
		token.nNext = m_aPos[iPos].nStart + 1;
	else if ( iPos == m_iPos && m_nNodeLength && m_nNodeType == MNT_PROCESSING_INSTRUCTION )
		token.nNext = m_nNodeOffset + 2;
	else
		return false;

	// Create insertion text depending on whether attribute already exists
	// Decision: for empty value leaving attrib="" instead of removing attrib
	int nReplace = 0;
	int nInsertAt;
	string strInsert;
	strInsert += x_ATTRIBQUOTE;
	strInsert += EscapeText( szValue, MNF_ESCAPEQUOTES );
	strInsert += x_ATTRIBQUOTE;
	if ( x_FindAttrib( token, szAttrib ) )
	{
		// Replace value
		nInsertAt = token.nL - ((token.nTokenFlags&MNF_QUOTED)?1:0);
		nReplace = token.Length() + ((token.nTokenFlags&MNF_QUOTED)?2:0);
	}
	else
	{
		// Insert string name value pair
		string strFormat;
		strFormat = " ";
		strFormat += szAttrib;
		strFormat += "=";
		strFormat += strInsert;
		strInsert = strFormat;
		nInsertAt = token.nNext;
	}

	x_DocChange( nInsertAt, nReplace, strInsert );
	int nAdjust = (int)strInsert.size() - nReplace;
	if ( m_nNodeType == MNT_PROCESSING_INSTRUCTION )
	{
		x_AdjustForNode( m_iPosParent, m_iPos, nAdjust );
		m_nNodeLength += nAdjust;
		MARKUP_SETDEBUGSTATE;
		return true;
	}
	m_aPos[iPos].AdjustStartTagLen( nAdjust );
	m_aPos[iPos].nLength += nAdjust;
	x_Adjust( iPos, nAdjust );
	MARKUP_SETDEBUGSTATE;
	return true;
}


bool CMarkupSTL::x_CreateNode( string& strNode, int nNodeType, const char* szText )
{
	// Set strNode based on nNodeType and szData
	// Return false if szData would jeopardize well-formed document
	//
	switch ( nNodeType )
	{
	case MNT_PROCESSING_INSTRUCTION:
		strNode = "<?";
		strNode += szText;
		strNode += "?>";
		break;
	case MNT_COMMENT:
		strNode = "<!--";
		strNode += szText;
		strNode += "-->";
		break;
	case MNT_ELEMENT:
		strNode = "<";
		strNode += szText;
		strNode += "/>";
		break;
	case MNT_TEXT:
	case MNT_WHITESPACE:
		strNode = EscapeText( szText );
		break;
	case MNT_DOCUMENT_TYPE:
		strNode = szText;
		break;
	case MNT_LONE_END_TAG:
		return false;
	case MNT_CDATA_SECTION:
		if ( strstr(szText,"]]>") != NULL )
			return false;
		strNode = "<![CDATA[";
		strNode += szText;
		strNode += "]]>";
		break;
	}
	return true;
}

string CMarkupSTL::x_EncodeCDATASection( const char* szData )
{
	// Split CDATA Sections if there are any end delimiters
	string strData = "<![CDATA[";
	const char* pszNextStart = szData;
	const char* pszEnd = strstr( szData, "]]>" );
	while ( pszEnd )
	{
		strData += string( pszNextStart, (int)(pszEnd - pszNextStart) );
		strData += "]]]]><![CDATA[>";
		pszNextStart = pszEnd + 3;
		pszEnd = strstr( pszNextStart, "]]>" );
	}
	strData += pszNextStart;
	strData += "]]>";
	return strData;
}

bool CMarkupSTL::x_SetData( int iPos, int nValue )
{
	// Convert integer to string
	char szVal[25];
	sprintf( szVal, "%d", nValue );
	return x_SetData( iPos, szVal, 0 );
}

bool CMarkupSTL::x_SetData( int iPos, const char* szData, int nFlags )
{
	// Set data at specified position
	// if nFlags==1, set content of element to a CDATA Section
	string strInsert;

	if ( iPos == m_iPos && m_nNodeLength )
	{
		// Not an element
		if ( ! x_CreateNode(strInsert, m_nNodeType, szData) )
			return false;
		x_DocChange( m_nNodeOffset, m_nNodeLength, strInsert );
		x_AdjustForNode( m_iPosParent, iPos, (int)strInsert.size() - m_nNodeLength );
		m_nNodeLength = (int)strInsert.size();
		MARKUP_SETDEBUGSTATE;
		return true;
	}

	// Set data in iPos element
	if ( ! iPos || m_aPos[iPos].iElemChild )
		return false;

	// Build strInsert from szData based on nFlags
	if ( nFlags & MNF_WITHCDATA )
		strInsert = x_EncodeCDATASection( szData );
	else
		strInsert = EscapeText( szData, nFlags );

	// Insert
	NodePos node( MNF_WITHNOLINES|MNF_REPLACE );
	node.strMeta = strInsert;
	int iPosBefore = 0;
	int nReplace = x_InsertNew( iPos, iPosBefore, node );
	int nAdjust = (int)node.strMeta.size() - nReplace;
	x_Adjust( iPos, nAdjust );
	m_aPos[iPos].nLength += nAdjust;
	if ( m_aPos[iPos].nFlags & MNF_ILLDATA )
		m_aPos[iPos].nFlags &= ~MNF_ILLDATA;
	MARKUP_SETDEBUGSTATE;
	return true;
}

string CMarkupSTL::x_GetData( int iPos ) const
{
	if ( iPos == m_iPos && m_nNodeLength )
	{
		if ( m_nNodeType == MNT_COMMENT )
			return m_strDoc.substr( m_nNodeOffset+4, m_nNodeLength-7 );
		else if ( m_nNodeType == MNT_PROCESSING_INSTRUCTION )
			return m_strDoc.substr( m_nNodeOffset+2, m_nNodeLength-4 );
		else if ( m_nNodeType == MNT_CDATA_SECTION )
			return m_strDoc.substr( m_nNodeOffset+9, m_nNodeLength-12 );
		else if ( m_nNodeType == MNT_TEXT )
			return UnescapeText( &(m_strDoc.c_str())[m_nNodeOffset], m_nNodeLength );
		else if ( m_nNodeType == MNT_LONE_END_TAG )
			return m_strDoc.substr( m_nNodeOffset+2, m_nNodeLength-3 );
		else
			return m_strDoc.substr( m_nNodeOffset, m_nNodeLength );
	}

	// Return a string representing data between start and end tag
	// Return empty string if there are any children elements
	string strData;
	if ( ! m_aPos[iPos].iElemChild && ! m_aPos[iPos].IsEmptyElement() )
	{
		// Quick scan for any tags inside content
		int nContentLen = m_aPos[iPos].ContentLen();
		int nStartContent = m_aPos[iPos].StartContent();
		const char* pszContent = &(m_strDoc.c_str())[nStartContent];
		const char* pszTag = strchr( pszContent, '<' );
		if ( pszTag && ((int)(pszTag-pszContent) < nContentLen) )
		{
			// Concatenate all CDATA Sections and text nodes, ignore other nodes
			TokenPos token( m_strDoc, m_nFlags );
			token.nNext = nStartContent;
			NodePos node;
			while ( token.nNext < nStartContent + nContentLen )
			{
				x_ParseNode( token, node );
				if ( node.nNodeType == MNT_TEXT )
					strData += UnescapeText( &token.szDoc[node.nStart], node.nLength );
				else if ( node.nNodeType == MNT_CDATA_SECTION )
					strData += m_strDoc.substr( node.nStart+9, node.nLength-12 );
			}
		}
		else // no tags
			strData = UnescapeText( &(m_strDoc.c_str())[nStartContent], nContentLen );
	}
	return strData;
}

string CMarkupSTL::x_GetElemContent( int iPos ) const
{
	if ( iPos && m_aPos[iPos].ContentLen() )
		return m_strDoc.substr( m_aPos[iPos].StartContent(), m_aPos[iPos].ContentLen() );
	return "";
}

bool CMarkupSTL::x_SetElemContent( const char* szContent )
{
	// Set data in iPos element only
	if ( ! m_iPos )
		return false;

	if ( m_nNodeLength )
		return false; // not an element

	// Unlink all children
	int iPos = m_iPos;
	int iPosChild = m_aPos[iPos].iElemChild;
	bool bHadChild = (iPosChild != 0);
	while ( iPosChild )
		iPosChild = x_ReleaseSubDoc( iPosChild );
	if ( bHadChild )
		x_CheckSavedPos();

	// Parse content
	bool bWellFormed = true;
	TokenPos token( szContent, m_nFlags );
	int iPosVirtual = x_GetFreePos();
	m_aPos[iPosVirtual].ClearVirtualParent();
	m_aPos[iPosVirtual].SetLevel( m_aPos[iPos].Level() + 1 );
	iPosChild = x_ParseElem( iPosVirtual, token );
	if ( m_aPos[iPosVirtual].nFlags & MNF_ILLFORMED )
		bWellFormed = false;
	m_aPos[iPos].nFlags = (m_aPos[iPos].nFlags & ~MNF_ILLDATA) | (m_aPos[iPosVirtual].nFlags & MNF_ILLDATA);

	// Prepare insert and adjust offsets
	NodePos node( MNF_WITHNOLINES|MNF_REPLACE );
	node.strMeta = szContent;
	int iPosBefore = 0;
	int nReplace = x_InsertNew( iPos, iPosBefore, node );
	
	// Adjust and link in the inserted elements
	x_Adjust( iPosChild, node.nStart );
	m_aPos[iPosChild].nStart += node.nStart;
	m_aPos[iPos].iElemChild = iPosChild;
	while ( iPosChild )
	{
		m_aPos[iPosChild].iElemParent = iPos;
		iPosChild = m_aPos[iPosChild].iElemNext;
	}
	x_ReleasePos( iPosVirtual );

	int nAdjust = (int)node.strMeta.size() - nReplace;
	x_Adjust( iPos, nAdjust, true );
	m_aPos[iPos].nLength += nAdjust;

	x_SetPos( m_iPosParent, m_iPos, 0 );
	return bWellFormed;
}

void CMarkupSTL::x_DocChange( int nLeft, int nReplace, const string& strInsert )
{
	// Insert strInsert int m_strDoc at nLeft replacing nReplace chars
	// Do this with only one buffer reallocation if it grows
	//
	int nDocLength = (int)m_strDoc.size();
	int nInsLength = (int)strInsert.size();
	int nNewLength = nInsLength + nDocLength - nReplace;

	// When creating a document, reduce reallocs by reserving string space
	// Allow for 1.5 times the current allocation
	int nBufferLen = nNewLength;
	int nAllocLen = (int)m_strDoc.capacity();
	if ( nNewLength > nAllocLen )
	{
		nBufferLen += nBufferLen/2 + 128;
		if ( nBufferLen < nNewLength )
			nBufferLen = nNewLength;
		m_strDoc.reserve( nBufferLen );
	}

	m_strDoc.replace( nLeft, nReplace, strInsert );

}

void CMarkupSTL::x_Adjust( int iPos, int nShift, bool bAfterPos /*=false*/ )
{
	// Loop through affected elements and adjust indexes
	// Algorithm:
	// 1. update children unless bAfterPos
	//    (if no children or bAfterPos is true, length of iPos not affected)
	// 2. update starts of next siblings and their children
	// 3. go up until there is a next sibling of a parent and update starts
	// 4. step 2
	int iPosTop = m_aPos[iPos].iElemParent;
	bool bPosFirst = bAfterPos; // mark as first to skip its children

	// Stop when we've reached the virtual parent (which has no tags)
	while ( m_aPos[iPos].StartTagLen() )
	{
		// Were we at containing parent of affected position?
		bool bPosTop = false;
		if ( iPos == iPosTop )
		{
			// Move iPosTop up one towards root
			iPosTop = m_aPos[iPos].iElemParent;
			bPosTop = true;
		}

		// Traverse to the next update position
		if ( ! bPosTop && ! bPosFirst && m_aPos[iPos].iElemChild )
		{
			// Depth first
			iPos = m_aPos[iPos].iElemChild;
		}
		else if ( m_aPos[iPos].iElemNext )
		{
			iPos = m_aPos[iPos].iElemNext;
		}
		else
		{
			// Look for next sibling of a parent of iPos
			// When going back up, parents have already been done except iPosTop
			while ( 1 )
			{
				iPos = m_aPos[iPos].iElemParent;
				if ( iPos == iPosTop )
					break;
				if ( m_aPos[iPos].iElemNext )
				{
					iPos = m_aPos[iPos].iElemNext;
					break;
				}
			}