xmlparse.cpp

一个完整强大的xml文件解析的程序

		{
			bool bCodeConverted = false;

			// Is it a numeric character reference?
			if ( pSource[nChar+1] == _T('#') )
			{
				// Is it a hex number?
				int nBase = 10;
				int nNumericChar = nChar + 2;
				_TCHAR cChar = pSource[nNumericChar];
				if ( cChar == _T('x') )
				{
					++nNumericChar;
					cChar = pSource[nNumericChar];
					nBase = 16;
				}

				// Look for terminating semi-colon within 7 characters
				int nCodeLen = 0;
				while ( nCodeLen < 7 && cChar && cChar != _T(';') )
				{
					// only ASCII digits 0-9, A-F, a-f expected
					nCodeLen += (int)_tclen( &pSource[nNumericChar+nCodeLen] );
					cChar = pSource[nNumericChar + nCodeLen];
				}

				// Process unicode
				if ( cChar == _T(';') )
				{
					int nUnicode = _tcstol( &pSource[nNumericChar], NULL, nBase );
#if defined(_UNICODE)
					pDest[nLen++] = (_TCHAR)nUnicode;
#elif defined(_MBCS)
					int nMBLen = wctomb( &pDest[nLen], (wchar_t)nUnicode );
					if ( nMBLen > 0 )
						nLen += nMBLen;
					else
						nUnicode = 0;
#else
					if ( nUnicode < 0x80 )
						pDest[nLen++] = (_TCHAR)nUnicode;
					else if ( nUnicode < 0x800 )
					{
						// Convert to 2-byte UTF-8
						pDest[nLen++] = (_TCHAR)(((nUnicode&0x7c0)>>6) | 0xc0);
						pDest[nLen++] = (_TCHAR)((nUnicode&0x3f) | 0x80);
					}
					else
					{
						// Convert to 3-byte UTF-8
						pDest[nLen++] = (_TCHAR)(((nUnicode&0xf000)>>12) | 0xe0);
						pDest[nLen++] = (_TCHAR)(((nUnicode&0xfc0)>>6) | 0x80);
						pDest[nLen++] = (_TCHAR)((nUnicode&0x3f) | 0x80);
					}
#endif
					if ( nUnicode )
					{
						// Increment index past ampersand semi-colon
						nChar = nNumericChar + nCodeLen + 1;
						bCodeConverted = true;
					}
				}
			}
			else // does not start with #
			{
				// Look for matching &code;
				for ( int nMatch = 0; nMatch < 5; ++nMatch )
				{
					if ( nChar < nTextLength - anCodeLen[nMatch]
						&& _tcsncmp(szaCode[nMatch],&pSource[nChar+1],anCodeLen[nMatch]) == 0 )
					{
						// Insert symbol and increment index past ampersand semi-colon
						pDest[nLen++] = szSymbol[nMatch];
						nChar += anCodeLen[nMatch] + 1;
						bCodeConverted = true;
						break;
					}
				}
			}

			// If the code is not converted, leave it as is
			if ( ! bCodeConverted )
			{
				pDest[nLen++] = _T('&');
				++nChar;
			}
		}
		else // not &
		{
			nCharLen = (int)_tclen(&pSource[nChar]);
			_tccpy( &pDest[nLen], &pSource[nChar] );
			nLen += nCharLen;
			nChar += nCharLen;
		}
	}
	strText.ReleaseBuffer(nLen);
	return string(strText);
}


int CXMLParse::FindNode( int nType )
{
	// Change current node position only if a node is found
	// If nType is 0 find any node, otherwise find node of type nType
	// Return type of node or 0 if not found
	// If found node is an element, change m_iPos

	// Determine where in document to start scanning for node
	int nTypeFound = 0;
	int nNodeOffset = m_nNodeOffset;
	if ( m_nNodeType > 1 )
	{
		// By-pass current node
		nNodeOffset += m_nNodeLength;
	}
	else
	{
		// Set position to begin looking for node
		nNodeOffset = 0; // default to start of document
		if ( m_iPos )
		{
			// After element
			nNodeOffset = m_aPos[m_iPos].StartAfter();
		}
		else if ( m_iPosParent )
		{
			// Immediately after start tag of parent
			if ( m_aPos[m_iPosParent].IsEmptyElement() )
				return 0;
			else
				nNodeOffset = m_aPos[m_iPosParent].StartContent();
		}
	}

	// Get nodes until we find what we're looking for
	int iPosNew = m_iPos;
	TokenPos token( m_strDoc, m_nFlags );
	NodePos node;
	token.nNext = nNodeOffset;
	do
	{
		nNodeOffset = token.nNext;
		nTypeFound = x_ParseNode( token, node );
		if ( nTypeFound == 0 )
		{
			// Check if we have reached the end of the parent element
			// Otherwise it is a lone end tag
			if ( m_iPosParent && nNodeOffset == m_aPos[m_iPosParent].StartContent()
					+ m_aPos[m_iPosParent].ContentLen() )
				return 0;
			nTypeFound = MNT_LONE_END_TAG;
		}
		else if ( nTypeFound < 0 )
		{
			if ( nTypeFound == -2 )
				return 0;
			// -1 is node error
			nTypeFound = MNT_NODE_ERROR;
		}
		else if ( nTypeFound == MNT_ELEMENT )
		{
			if ( iPosNew )
				iPosNew = m_aPos[iPosNew].iElemNext;
			else
				iPosNew = m_aPos[m_iPosParent].iElemChild;
			if ( ! iPosNew )
				return 0;
			if ( ! nType || (nType & nTypeFound) )
			{
				// Found element node, move position to this element
				x_SetPos( m_iPosParent, iPosNew, 0 );
				return m_nNodeType;
			}
			token.nNext = m_aPos[iPosNew].StartAfter();
		}
	}
	while ( nType && ! (nType & nTypeFound) );

	m_iPos = iPosNew;
	m_iPosChild = 0;
	m_nNodeOffset = nNodeOffset;
	m_nNodeLength = token.nNext - nNodeOffset;
	m_nNodeType = nTypeFound;
	return m_nNodeType;
}

bool CXMLParse::RemoveNode()
{
	if ( m_iPos || m_nNodeLength )
	{
		x_RemoveNode( m_iPosParent, m_iPos, m_nNodeType, m_nNodeOffset, m_nNodeLength );
		m_iPosChild = 0;
		return true;
	}
	return false;
}

string CXMLParse::GetTagName() const
{
	// Return the tag name at the current main position
	CString strTagName;

	// This method is primarily for elements, however
	// it does return something for certain other nodes
	if ( m_nNodeLength )
	{
		switch ( m_nNodeType )
		{
		case MNT_PROCESSING_INSTRUCTION:
		case MNT_LONE_END_TAG:
			{
				// <?target or </tagname
				TokenPos token( m_strDoc, m_nFlags );
				token.nNext = m_nNodeOffset + 2;
				if ( x_FindName(token) )
					strTagName = x_GetToken( token ).c_str();
			}
			break;
		case MNT_COMMENT:
			strTagName = _T("#comment");
			break;
		case MNT_CDATA_SECTION:
			strTagName = _T("#cdata-section");
			break;
		case MNT_DOCUMENT_TYPE:
			{
				// <!DOCTYPE name
				TokenPos token( m_strDoc, m_nFlags );
				token.nNext = m_nNodeOffset + 2;
				if ( x_FindName(token) && x_FindName(token) )
					strTagName = x_GetToken( token ).c_str();
			}
			break;
		case MNT_TEXT:
		case MNT_WHITESPACE:
			strTagName = _T("#text");
			break;
		}
		return string(strTagName);
	}

	if ( m_iPos )
		strTagName = x_GetTagName( m_iPos ).c_str();
	return string(strTagName);
}

bool CXMLParse::IntoElem()
{
	// If there is no child position and IntoElem is called it will succeed in release 6.3
	// (A subsequent call to FindElem will find the first element)
	// The following short-hand behavior was never part of EDOM and was misleading
	// It would find a child element if there was no current child element position and go into it
	// It is removed in release 6.3, this change is NOT backwards compatible!
	// if ( ! m_iPosChild )
	//	FindChildElem();

	if ( m_iPos && m_nNodeType == MNT_ELEMENT )
	{
		x_SetPos( m_iPos, m_iPosChild, 0 );
		return true;
	}
	return false;
}

bool CXMLParse::OutOfElem()
{
	// Go to parent element
	if ( m_iPosParent )
	{
		x_SetPos( m_aPos[m_iPosParent].iElemParent, m_iPosParent, m_iPos );
		return true;
	}
	return false;
}

string CXMLParse::GetAttribName( int n ) const
{
	// Return nth attribute name of main position
	TokenPos token( m_strDoc, m_nFlags );
	if ( m_iPos && m_nNodeType == MNT_ELEMENT )
		token.nNext = m_aPos[m_iPos].nStart + 1;
	else if ( m_nNodeLength && m_nNodeType == MNT_PROCESSING_INSTRUCTION )
		token.nNext = m_nNodeOffset + 2;
	else
		return _T("");
	if ( x_FindAttrib(token,NULL,n) )
		return x_GetToken( token );
	return _T("");
}

bool CXMLParse::SavePos( const char * szPosName )
{
	// Save current element position in saved position map
	if ( szPosName )
	{
		SavedPos savedpos;
		if ( szPosName )
			savedpos.strName = szPosName;
		if ( m_iPosChild )
		{
			savedpos.iPos = m_iPosChild;
			savedpos.nSavedPosFlags |= SavedPosMap::SPM_CHILD;
		}
		else if ( m_iPos )
		{
			savedpos.iPos = m_iPos;
			savedpos.nSavedPosFlags |= SavedPosMap::SPM_MAIN;
		}
		else
		{
			savedpos.iPos = m_iPosParent;
		}
		savedpos.nSavedPosFlags |= SavedPosMap::SPM_USED;

		if ( ! m_mapSavedPos.pTable )
			m_mapSavedPos.AllocMapTable();
		int nSlot = m_mapSavedPos.Hash( szPosName );
		SavedPos* pSavedPos = m_mapSavedPos.pTable[nSlot];
		int nOffset = 0;
		if ( ! pSavedPos )
		{
			pSavedPos = new SavedPos[2];
			pSavedPos[1].nSavedPosFlags = SavedPosMap::SPM_LAST;
			m_mapSavedPos.pTable[nSlot] = pSavedPos;
		}
		else
		{
			while ( pSavedPos[nOffset].nSavedPosFlags & SavedPosMap::SPM_USED )
			{
				if ( pSavedPos[nOffset].strName == szPosName )
					break;
				if ( pSavedPos[nOffset].nSavedPosFlags & SavedPosMap::SPM_LAST )
				{
					int nNewSize = (nOffset + 6) * 2;
					SavedPos* pNewSavedPos = new SavedPos[nNewSize];
					for ( int nCopy=0; nCopy<=nOffset; ++nCopy )
						pNewSavedPos[nCopy] = pSavedPos[nCopy];
					pNewSavedPos[nOffset].nSavedPosFlags ^= SavedPosMap::SPM_LAST;
					pNewSavedPos[nNewSize-1].nSavedPosFlags = SavedPosMap::SPM_LAST;
					delete [] pSavedPos;
					pSavedPos = pNewSavedPos;
					m_mapSavedPos.pTable[nSlot] = pSavedPos;
					++nOffset;
					break;
				}
				++nOffset;
			}
		}
		if ( pSavedPos[nOffset].nSavedPosFlags & SavedPosMap::SPM_LAST )
			savedpos.nSavedPosFlags |= SavedPosMap::SPM_LAST;
		pSavedPos[nOffset] = savedpos;

		/*
		// To review hash table balance, uncomment and watch strBalance
		CString strBalance, strSlot;
		for ( nSlot=0; nSlot < SavedPosMap::SPM_SIZE; ++nSlot )
		{
			pSavedPos = m_mapSavedPos.pTable[nSlot];
			int nCount = 0;
			while ( pSavedPos && pSavedPos->nSavedPosFlags & SavedPosMap::SPM_USED )
			{
				++nCount;
				if ( pSavedPos->nSavedPosFlags & SavedPosMap::SPM_LAST )
					break;
				++pSavedPos;
			}
			strSlot.Format( _T("%d "), nCount );
			strBalance += strSlot;
		}
		*/

		return true;
	}
	return false;
}

bool CXMLParse::RestorePos( const char * szPosName )
{
	// Restore element position if found in saved position map
	if ( szPosName && m_mapSavedPos.pTable )
	{
		int nSlot = m_mapSavedPos.Hash( szPosName );
		SavedPos* pSavedPos = m_mapSavedPos.pTable[nSlot];
		if ( pSavedPos )
		{
			int nOffset = 0;
			while ( pSavedPos[nOffset].nSavedPosFlags & SavedPosMap::SPM_USED )
			{
				if ( pSavedPos[nOffset].strName == szPosName )
				{
					int i = pSavedPos[nOffset].iPos;
					if ( pSavedPos[nOffset].nSavedPosFlags & SavedPosMap::SPM_CHILD )
						x_SetPos( m_aPos[m_aPos[i].iElemParent].iElemParent, m_aPos[i].iElemParent, i );
					else if ( pSavedPos[nOffset].nSavedPosFlags & SavedPosMap::SPM_MAIN )
						x_SetPos( m_aPos[i].iElemParent, i, 0 );
					else
						x_SetPos( i, 0, 0 );
					return true;
				}
				if ( pSavedPos[nOffset].nSavedPosFlags & SavedPosMap::SPM_LAST )
					break;
				++nOffset;
			}
		}
	}
	return false;
}

bool CXMLParse::RemoveElem()
{
	// Remove current main position element
	if ( m_iPos && m_nNodeType == MNT_ELEMENT )
	{
		int iPos = x_RemoveElem( m_iPos );
		x_SetPos( m_iPosParent, iPos, 0 );
		return true;
	}
	return false;
}

bool CXMLParse::RemoveChildElem()
{
	// Remove current child position element
	if ( m_iPosChild )
	{
		int iPosChild = x_RemoveElem( m_iPosChild );
		x_SetPos( m_iPosParent, m_iPos, iPosChild );
		return true;
	}
	return false;
}


//////////////////////////////////////////////////////////////////////
// Private Methods
//////////////////////////////////////////////////////////////////////

bool CXMLParse::x_AllocPosArray( int nNewSize /*=0*/ )
{
	// Resize m_aPos when the document is created or the array is filled
	// The PosArray class is implemented using segments to reduce contiguous memory requirements
	// It reduces reallocations (copying of memory) since this only occurs within one segment
	// The "Grow By" algorithm ensures there are no reallocations after 2 segments
	//
	if ( ! nNewSize )
		nNewSize = m_iPosFree + (m_iPosFree>>1); // Grow By: multiply size by 1.5
	if ( m_aPos.GetSize() < nNewSize )
	{
		// Grow By: new size can be at most one more complete segment
		int nSeg = (m_aPos.GetSize()?m_aPos.GetSize()-1:0) >> m_aPos.PA_SEGBITS;
		int nNewSeg = (nNewSize-1) >> m_aPos.PA_SEGBITS;
		if ( nNewSeg > nSeg + 1 )
		{
			nNewSeg = nSeg + 1;
			nNewSize = (nNewSeg+1) << m_aPos.PA_SEGBITS;
		}

		// Allocate array of segments
		if ( m_aPos.nSegs <= nNewSeg )
		{
			int nNewSegments = 4 + nNewSeg * 2;
			char* pNewSegments = new char[nNewSegments*sizeof(char*)];
			if ( m_aPos.SegsUsed() )
				memcpy( pNewSegments, m_aPos.pSegs, m_aPos.SegsUsed()*sizeof(char*) );
			if ( m_aPos.pSegs )
				delete[] (char*)m_aPos.pSegs;
			m_aPos.pSegs = (ElemPos**)pNewSegments;
			m_aPos.nSegs = nNewSegments;
		}

		// Calculate segment sizes
		int nSegSize = m_aPos.GetSize() - (nSeg << m_aPos.PA_SEGBITS);
		int nNewSegSize = nNewSize - (nNewSeg << m_aPos.PA_SEGBITS);

		// Complete first segment