xmlparse.cpp

一个完整强大的xml文件解析的程序

		}
		else if ( nParseFlags )
		{
			if ( nParseFlags & PD_TEXTORWS )
			{
				if ( cD == _T('<') )
				{
					nR = (int)(pDoc - token.szDoc) - 1;
					nNodeType = MNT_WHITESPACE;
					break;
				}
				else if ( ! _tcschr(_T(" \t\n\r"),(_TCHAR)cD) )
				{
					nParseFlags ^= PD_TEXTORWS;
					FINDNODETYPE( _T("<"), MNT_TEXT, 0 )
				}
			}
			else if ( nParseFlags & PD_OPENTAG )
			{
				nParseFlags ^= PD_OPENTAG;
				if ( cD > 0x60 || ( cD > 0x40 && cD < 0x5b ) || cD == 0x5f || cD == 0x3a )
					FINDNODETYPE( _T(">"), MNT_ELEMENT, 1 )
				else if ( cD == _T('/') )
					FINDNODETYPE( _T(">"), 0, 2 )
				else if ( cD == _T('!') )
					nParseFlags |= PD_BANG;
				else if ( cD == _T('?') )
					FINDNODETYPE( _T("?>"), MNT_PROCESSING_INSTRUCTION, 2 )
				else
					FINDNODEBAD( _T("tag name character") )
			}
			else if ( nParseFlags & PD_BANG )
			{
				nParseFlags ^= PD_BANG;
				if ( cD == _T('-') )
					nParseFlags |= PD_DASH;
				else if ( cD == _T('[') && !(nParseFlags & PD_DOCTYPE) )
					nParseFlags |= PD_BRACKET;
				else if ( cD == _T('D') && !(nParseFlags & PD_DOCTYPE) )
					nParseFlags |= PD_DOCTYPE;
				else if ( _tcschr(_T("EAN"),(_TCHAR)cD) ) // <!ELEMENT ATTLIST ENTITY NOTATION
					FINDNODETYPE( _T(">"), MNT_DOCUMENT_TYPE, 0 )
				else
					FINDNODEBAD( _T("! tag") )
			}
			else if ( nParseFlags & PD_DASH )
			{
				nParseFlags ^= PD_DASH;
				if ( cD == _T('-') )
					FINDNODETYPE( _T("-->"), MNT_COMMENT, 0 )
				else
					FINDNODEBAD( _T("comment tag") )
			}
			else if ( nParseFlags & PD_BRACKET )
			{
				nParseFlags ^= PD_BRACKET;
				if ( cD == _T('C') )
					FINDNODETYPE( _T("]]>"), MNT_CDATA_SECTION, 0 )
				else
					FINDNODEBAD( _T("tag") )
			}
			else if ( nParseFlags & PD_DOCTYPE )
			{
				if ( cD == _T('<') )
					nParseFlags |= PD_OPENTAG;
				else if ( cD == _T('>') )
				{
					nR = (int)(pDoc - token.szDoc);
					nNodeType = MNT_DOCUMENT_TYPE;
					break;
				}
			}
		}
		else if ( cD == _T('<') )
		{
			nParseFlags |= PD_OPENTAG;
		}
		else
		{
			nNodeType = MNT_WHITESPACE;
			if ( _tcschr(_T(" \t\n\r"),(_TCHAR)cD) )
				nParseFlags |= PD_TEXTORWS;
			else
				FINDNODETYPE( _T("<"), MNT_TEXT, 0 )
		}
		pDoc += _tclen( pDoc );
	}
	token.nNext = nR + 1;
	node.nLength = token.nNext - node.nStart;
	node.nNodeType = nNodeType;
	return nNodeType;
}

string CXMLParse::x_GetPath( int iPos ) const
{
	CString strPath;
	while ( iPos )
	{
		CString strTagName = x_GetTagName( iPos ).c_str();
		int iPosParent = m_aPos[iPos].iElemParent;
		int iPosSib = 0;
		int nCount = 0;
		while ( iPosSib != iPos )
		{
			iPosSib = x_FindElem( iPosParent, iPosSib, strTagName );
			++nCount;
		}
		if ( nCount > 1 )
		{
			_TCHAR szPred[25];
			_stprintf( szPred, _T("[%d]"), nCount );
			strPath = _T("/") + strTagName + szPred + strPath;
		}
		else
			strPath = _T("/") + strTagName + strPath;
		iPos = iPosParent;
	}
	return string(strPath);
}

string CXMLParse::x_GetTagName( int iPos ) const
{
	// Return the tag name at specified element
	TokenPos token( m_strDoc, m_nFlags );
	token.nNext = m_aPos[iPos].nStart + 1;
	if ( ! iPos || ! x_FindName( token ) )
		return _T("");

	// Return substring of document
	return x_GetToken( token );
}

bool CXMLParse::x_FindAttrib( CXMLParse::TokenPos& token, const char * szAttrib, int n/*=0*/ )
{
	// Return true if found, otherwise false and token.nNext is new insertion point
	// If szAttrib is NULL find attrib n and leave token at attrib name
	// If szAttrib is given, find matching attrib and leave token at value
	// support non-well-formed attributes e.g. href=/advanced_search?hl=en, nowrap
	// token also holds start and length of preceeding whitespace to support remove
	//
	int nPreSpaceStart;
	int nPreSpaceLength;
	int nChar;
	_TCHAR cFirstChar;
	LPCTSTR szDoc = token.szDoc;
	int nAttrib = -1; // starts at tag name
	int nFoundAttribNameR = 0;
	bool bAfterEqual = false;
	while ( 1 )
	{
		// 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 == _T('\"') || cFirstChar == _T('\'') )
		{
			token.nTokenFlags |= MNF_QUOTED;

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

			// Look for closing quote
			while ( szDoc[nChar] && szDoc[nChar] != cFirstChar )
				nChar += (int)_tclen( &szDoc[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] && ! _tcschr(_T(" \t\n\r>"),szDoc[nChar]) )
					nChar += (int)_tclen( &szDoc[nChar] );
			}
			else
			{
				while ( szDoc[nChar] && ! _tcschr(_T("= \t\n\r>/?"),szDoc[nChar]) )
					nChar += (int)_tclen( &szDoc[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?
			_TCHAR cChar = szDoc[token.nL];
			if ( cChar == _T('=') )
			{
				bAfterEqual = true;
				continue;
			}

			// Is it the right angle bracket?
			if ( cChar == _T('>') || cChar == _T('/') || cChar == _T('?') )
			{
				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
}

std::string CXMLParse::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 _T("");

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

bool CXMLParse::x_SetAttrib( int iPos, const char * szAttrib, __int64 nValue)
{
	_TCHAR szVal[25];
	_stprintf( szVal, _T("%I64d"), nValue );
	return x_SetAttrib( iPos, szAttrib, szVal );
}
bool CXMLParse::x_SetAttrib( int iPos, const char * szAttrib, int nValue )
{
	// Convert integer to string
	_TCHAR szVal[25];
	_stprintf( szVal, _T("%d"), nValue );
	return x_SetAttrib( iPos, szAttrib, szVal );
}

bool CXMLParse::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;
	CString strInsert;
	strInsert += x_ATTRIBQUOTE;
	strInsert += EscapeText( szValue, MNF_ESCAPEQUOTES ).c_str();
	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
		CString strFormat;
		strFormat = _T(" ");
		strFormat += szAttrib;
		strFormat += _T("=");
		strFormat += strInsert;
		strInsert = strFormat;
		nInsertAt = token.nNext;
	}

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


bool CXMLParse::x_CreateNode( CString& 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 ).c_str();
		break;
	case MNT_DOCUMENT_TYPE:
		strNode = szText;
		break;
	case MNT_LONE_END_TAG:
		return false;
	case MNT_CDATA_SECTION:
		if ( _tcsstr(szText,_T("]]>")) != NULL )
			return false;
		strNode = "<![CDATA[";
		strNode += szText;
		strNode += "]]>";
		break;
	}
	return true;
}

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

bool CXMLParse::x_SetData( int iPos, int nValue )
{
	// Convert integer to string
	_TCHAR szVal[25];
	_stprintf( szVal, _T("%d"), nValue );
	return x_SetData( iPos, szVal, 0 );
}

bool CXMLParse::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
	CString 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, strInsert.GetLength() - m_nNodeLength );
		m_nNodeLength = strInsert.GetLength();
		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 ).c_str();

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

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

	// Return a string representing data between start and end tag
	// Return empty string if there are any children elements
	CString strData;
	if ( ! m_aPos[iPos].iElemChild && ! m_aPos[iPos].IsEmptyElement() )
	{
		// Quick scan for any tags inside content