markupstl.cpp

解析xml文件的强大包

		{
			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;
	MARKUP_SETDEBUGSTATE;
	return m_nNodeType;
}

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

string CMarkupSTL::GetTagName() const
{
	// Return the tag name at the current main position
	string 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 );
			}
			break;
		case MNT_COMMENT:
			strTagName = "#comment";
			break;
		case MNT_CDATA_SECTION:
			strTagName = "#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 );
			}
			break;
		case MNT_TEXT:
		case MNT_WHITESPACE:
			strTagName = "#text";
			break;
		}
		return strTagName;
	}

	if ( m_iPos )
		strTagName = x_GetTagName( m_iPos );
	return strTagName;
}

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

string CMarkupSTL::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 "";
	if ( x_FindAttrib(token,NULL,n) )
		return x_GetToken( token );
	return "";
}

bool CMarkupSTL::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
		string strBalance;
		char szSlot[20];
		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;
			}
			sprintf( szSlot, "%d ", nCount );
			strBalance += szSlot;
		}
		*/

		return true;
	}
	return false;
}

bool CMarkupSTL::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 CMarkupSTL::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 CMarkupSTL::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
//////////////////////////////////////////////////////////////////////

string CMarkupSTL::x_GetLastError()
{
	string strError = strerror(errno);
	for ( int nChar=0; nChar<(int)strError.size(); ++nChar )
		if ( strError[nChar] == '\r' || strError[nChar] == '\n' )
		{
			strError.resize( nChar ); // no trailing newline
			break;
		}
	return strError;
}

bool CMarkupSTL::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
		int nFullSegSize = 1 << m_aPos.PA_SEGBITS;
		if ( nSeg < nNewSeg && nSegSize < nFullSegSize )
		{
			char* pNewFirstSeg = new char[ nFullSegSize * sizeof(ElemPos) ];
			if ( nSegSize )
			{
				// Reallocate
				memcpy( pNewFirstSeg, m_aPos.pSegs[nSeg], nSegSize * sizeof(ElemPos) );
				delete[] (char*)m_aPos.pSegs[nSeg];
			}
			m_aPos.pSegs[nSeg] = (ElemPos*)pNewFirstSeg;
		}

		// New segment
		char* pNewSeg = new char[ nNewSegSize * sizeof(ElemPos) ];
		if ( nNewSeg == nSeg && nSegSize )
		{
			// Reallocate
			memcpy( pNewSeg, m_aPos.pSegs[nSeg], nSegSize * sizeof(ElemPos) );
			delete[] (char*)m_aPos.pSegs[nSeg];
		}
		m_aPos.pSegs[nNewSeg] = (ElemPos*)pNewSeg;
		m_aPos.nSize = nNewSize;
	}
	return true;
}

bool CMarkupSTL::x_ParseDoc()
{
	// Preserve pre-parse result
	string strResult = m_strError;

	// Reset indexes
	ResetPos();
	m_mapSavedPos.RemoveAll();

	// Starting size of position array: 1 element per 64 bytes of document
	// Tight fit when parsing small doc, only 0 to 2 reallocs when parsing large doc
	// Start at 8 when creating new document
	m_iPosFree = 1;
	x_AllocPosArray( (int)m_strDoc.size() / 64 + 8 );
	m_iPosDeleted = 0;

	// Parse document
	m_aPos[0].ClearVirtualParent();
	if ( m_strDoc.size() )
	{
		TokenPos token( m_strDoc, m_nFlags );
		int iPos = x_ParseElem( 0, token );
		m_aPos[0].nLength = (int)m_strDoc.size();
		if ( iPos > 0 )
		{
			m_aPos[0].iElemChild = iPos;
			if ( m_aPos[iPos].iElemNext )
				m_strError = "Root element has sibling";
		}
		else
			m_strError = "No root element";
	}
	else
		m_strError = "Empty document";

	ResetPos();

	// Combine preserved result with parse error
	if ( ! strResult.empty() )
	{
		if ( m_strError.empty() )
			m_strError = strResult;
		else
			m_strError = strResult + ", " + m_strError;
	}

	return IsWellFormed();
};

int CMarkupSTL::x_ParseElem( int iPosParent, TokenPos& token )
{
	// This is either called by x_ParseDoc or x_AddSubDoc or x_SetElemContent
	// Returns index of the first element encountered or zero if no elements
	//
	int iElemRoot = 0;
	int iPos = iPosParent;
	int iVirtualParent = iPosParent;
	int nRootDepth = m_aPos[iPos].Level();
	token.nNext = 0;
	m_strError.erase();

	// Loop through the nodes of the document
	NodeStack aNodes;
	aNodes.Add();
	int nDepth = 0;
	int nMatchDepth;
	int iPosChild;
	int iPosMatch;
	int nTypeFound = 0;
	ElemPos* pElem;
	int iElemFirst, iElemLast;
	while ( 1 )
	{
		nTypeFound = x_ParseNode( token, aNodes.Top() );
		nMatchDepth = 0;
		if ( nTypeFound == MNT_ELEMENT ) // start tag
		{
			iPos = x_GetFreePos();
			if ( ! iElemRoot )
				iElemRoot = iPos;
			pElem = &m_aPos[iPos];
			pElem->iElemParent = iPosParent;
			pElem->iElemNext = 0;
			if ( m_aPos[iPosParent].iElemChild )
			{
				iElemFirst = m_aPos[iPosParent].iElemChild;
				iElemLast = m_aPos[iElemFirst].iElemPrev;
				m_aPos[iElemLast].iElemNext = iPos;
				pElem->iElemPrev = iElemLast;
				m_aPos[iElemFirst].iElemPrev = iPos;
				pElem->nFlags = 0;
			}
			else
			{
				m_aPos[iPosParent].iElemChild = iPos;
				pElem->iElemPrev = iPos;
				pElem->nFlags = MNF_FIRST;
			}
			pElem->SetLevel( nRootDepth + nDepth );
			pElem->iElemChild = 0;
			pElem->nStart = aNodes.Top().nStart;
			pElem->SetStartTagLen( aNodes.Top().nLength );
			if ( aNodes.Top().nFlags & MNF_EMPTY )
			{
				iPos = iPosParent;
				pElem->SetEndTagLen( 0 );
				pElem->nLength = aNodes.Top().nLength;
			}
			else
			{
				iPosParent = iPos;
				++nDepth;
				aNodes.Add();
			}
		}
		else if ( nTypeFound == 0 ) // end tag
		{
			nMatchDepth = nDepth;
			iPosMatch = iPos;
			while ( nMatchDepth && ! token.Match(aNodes.At(nMatchDepth-1).strMeta) )
			{
				/*
				// Auto-switch case sensitivity
				if ( ! (token.nTokenFlags & MDF_IGNORECASE ) )
				{
					token.nTokenFlags |= MDF_IGNORECASE;
					if ( token.Match(aNodes.At(nMatchDepth-1).strMeta) )