markupstl.cpp

解析xml文件的强大包

		}
		bPosFirst = false;

		// Shift indexes at iPos
		if ( iPos != iPosTop )
			m_aPos[iPos].nStart += nShift;
		else
			m_aPos[iPos].nLength += nShift;
	}
}

int CMarkupSTL::x_InsertNew( int iPosParent, int& iPosRel, CMarkupSTL::NodePos& node )
{
	// Parent empty tag or tags with no content?
	bool bEmptyParentTag = iPosParent && m_aPos[iPosParent].IsEmptyElement();
	bool bNoContentParentTags = iPosParent && ! m_aPos[iPosParent].ContentLen();
	if ( node.nLength )
	{
		// Located at a non-element node
		if ( ! (node.nFlags & MNF_INSERT) )
			node.nStart += node.nLength;
	}
	else if ( iPosRel )
	{
		// Located at an element
		node.nStart = m_aPos[iPosRel].nStart;
		if ( ! (node.nFlags & MNF_INSERT) ) // follow iPosRel
			node.nStart += m_aPos[iPosRel].nLength;
	}
	else if ( bEmptyParentTag )
	{
		// Parent has no separate end tag, so split empty element
		if ( m_aPos[iPosParent].nFlags & MNF_NONENDED )
			node.nStart = m_aPos[iPosParent].StartContent();
		else
			node.nStart = m_aPos[iPosParent].StartContent() - 1;
	}
	else
	{
		if ( node.nFlags & (MNF_INSERT|MNF_REPLACE) )
			node.nStart = m_aPos[iPosParent].StartContent();
		else // before end tag
			node.nStart = m_aPos[iPosParent].StartAfter() - m_aPos[iPosParent].EndTagLen();
	}

	// Go up to start of next node, unless its splitting an empty element
	if ( ! (node.nFlags&(MNF_WITHNOLINES|MNF_REPLACE)) && ! bEmptyParentTag )
	{
		const char* szDoc = m_strDoc.c_str();
		int nChar = node.nStart;
		if ( ! x_FindAny(szDoc,nChar) || szDoc[nChar] == '<' )
			node.nStart = nChar;
	}

	// Is insert relative to element position? (i.e. not other kind of node)
	if ( ! node.nLength )
	{
		// Modify iPosRel to reflect position before
		if ( iPosRel )
		{
			if ( node.nFlags & MNF_INSERT )
			{
				if ( ! (m_aPos[iPosRel].nFlags & MNF_FIRST) )
					iPosRel = m_aPos[iPosRel].iElemPrev;
				else
					iPosRel = 0;
			}
		}
		else if ( ! (node.nFlags & MNF_INSERT) )
		{
			// If parent has a child, add after last child
			if ( m_aPos[iPosParent].iElemChild )
				iPosRel = m_aPos[m_aPos[iPosParent].iElemChild].iElemPrev;
		}
	}

	// Get node length (used only by x_AddNode)
	node.nLength = (int)node.strMeta.size();

	// Prepare end of lines
	if ( (! (node.nFlags & MNF_WITHNOLINES)) && (bEmptyParentTag || bNoContentParentTags) )
		node.nStart += x_EOLLEN;
	if ( ! (node.nFlags & MNF_WITHNOLINES) )
		node.strMeta += x_EOL;

	// Calculate insert offset and replace length
	int nReplace = 0;
	int nInsertAt = node.nStart;
	if ( bEmptyParentTag )
	{
		string strTagName = x_GetTagName( iPosParent );
		string strFormat;
		if ( node.nFlags & MNF_WITHNOLINES )
			strFormat = ">";
		else
			strFormat = ">" x_EOL;
		strFormat += node.strMeta;
		strFormat += "</";
		strFormat += strTagName;
		node.strMeta = strFormat;
		if ( m_aPos[iPosParent].nFlags & MNF_NONENDED )
		{
			nInsertAt = m_aPos[iPosParent].StartAfter() - 1;
			nReplace = 0;
			m_aPos[iPosParent].nFlags ^= MNF_NONENDED;
		}
		else
		{
			nInsertAt = m_aPos[iPosParent].StartAfter() - 2;
			nReplace = 1;
			m_aPos[iPosParent].AdjustStartTagLen( -1 );
		}
		m_aPos[iPosParent].SetEndTagLen( 3 + (int)strTagName.size() );
	}
	else
	{
		if ( node.nFlags & MNF_REPLACE )
		{
			nInsertAt = m_aPos[iPosParent].StartContent();
			nReplace = m_aPos[iPosParent].ContentLen();
		}
		else if ( bNoContentParentTags )
		{
			node.strMeta = x_EOL + node.strMeta;
			nInsertAt = m_aPos[iPosParent].StartContent();
		}
	}
	x_DocChange( nInsertAt, nReplace, node.strMeta );
	return nReplace;
}

bool CMarkupSTL::x_AddElem( const char* szName, int nValue, int nFlags )
{
	// Convert integer to string
	char szVal[25];
	sprintf( szVal, "%d", nValue );
	return x_AddElem( szName, szVal, nFlags );
}

bool CMarkupSTL::x_AddElem( const char* szName, const char* szValue, int nFlags )
{
	if ( nFlags & MNF_CHILD )
	{
		// Adding a child element under main position
		if ( ! m_iPos )
			return false;
	}

	// Locate where to add element relative to current node
	NodePos node( nFlags );
	int iPosParent, iPosBefore;
	if ( nFlags & MNF_CHILD )
	{
		iPosParent = m_iPos;
		iPosBefore = m_iPosChild;
	}
	else
	{
		iPosParent = m_iPosParent;
		iPosBefore = m_iPos;
		node.nStart = m_nNodeOffset;
		node.nLength = m_nNodeLength;
	}

	// Cannot have data in non-ended element
	if ( (nFlags&MNF_WITHNOEND) && szValue && szValue[0] )
		return false;

	// Allocate ElemPos structure for this element
	int iPos = x_GetFreePos();

	// Create string for insert
	// If no szValue is specified, an empty element is created
	// i.e. either <NAME>value</NAME> or <NAME/>
	//
	ElemPos* pElem = &m_aPos[iPos];
	int nLenName = (int)strlen(szName);
	if ( ! szValue || ! szValue[0] )
	{
		// <NAME/> empty element
		node.strMeta = "<";
		node.strMeta += szName;
		if ( nFlags & MNF_WITHNOEND )
		{
			node.strMeta += ">";
			pElem->SetStartTagLen( nLenName + 2 );
			pElem->nLength = nLenName + 2;
		}
		else
		{
			if ( nFlags & MNF_WITHXHTMLSPACE )
			{
				node.strMeta += " />";
				pElem->SetStartTagLen( nLenName + 4 );
				pElem->nLength = nLenName + 4;
			}
			else
			{
				node.strMeta += "/>";
				pElem->SetStartTagLen( nLenName + 3 );
				pElem->nLength = nLenName + 3;
			}
		}
		pElem->SetEndTagLen( 0 );
	}
	else
	{
		// <NAME>value</NAME>
		string strValue;
		if ( nFlags & MNF_WITHCDATA )
			strValue = x_EncodeCDATASection( szValue );
		else
			strValue = EscapeText( szValue, nFlags );
		int nLenValue = (int)strValue.size();
		node.strMeta = "<";
		node.strMeta += szName;
		node.strMeta += ">";
		node.strMeta += strValue;
		node.strMeta += "</";
		node.strMeta += szName;
		node.strMeta += ">";
		pElem->SetEndTagLen( nLenName + 3 );
		pElem->nLength = nLenName * 2 + nLenValue + 5;
		pElem->SetStartTagLen( nLenName + 2 );
	}

	// Insert
	int nReplace = x_InsertNew( iPosParent, iPosBefore, node );

	pElem->nStart = node.nStart;
	pElem->iElemChild = 0;
	if ( nFlags & MNF_WITHNOEND )
		pElem->nFlags = MNF_NONENDED;
	else
		pElem->nFlags = 0;
	x_LinkElem( iPosParent, iPosBefore, iPos );

	x_Adjust( iPos, (int)node.strMeta.size() - nReplace );

	if ( nFlags & MNF_CHILD )
		x_SetPos( m_iPosParent, iPosParent, iPos );
	else
		x_SetPos( iPosParent, iPos, 0 );
	return true;
}

string CMarkupSTL::x_GetSubDoc( int iPos ) const
{
	if ( iPos )
	{
		int nStart = m_aPos[iPos].nStart;
		int nNext = nStart + m_aPos[iPos].nLength;
		const char* szDoc = m_strDoc.c_str();
		int nChar = nNext;
		if ( ! x_FindAny(szDoc,nChar) || szDoc[nChar] == '<' )
			nNext = nChar;
		return m_strDoc.substr( nStart, nNext - nStart );
	}
	return "";
}

bool CMarkupSTL::x_AddSubDoc( const char* szSubDoc, int nFlags )
{
	// Add subdocument, parse, and modify positions of affected elements
	//
	NodePos node( nFlags );
	int iPosParent, iPosBefore;
	if ( nFlags & MNF_CHILD )
	{
		// Add a subdocument under main position, before or after child
		if ( ! m_iPos )
			return false;
		iPosParent = m_iPos;
		iPosBefore = m_iPosChild;
	}
	else
	{
		// Add a subdocument under parent position, before or after main
		iPosParent = m_iPosParent;
		iPosBefore = m_iPos;
		node.nStart = m_nNodeOffset;
		node.nLength = m_nNodeLength;
	}

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

	// Extract subdocument without leading/trailing nodes
	int nExtractStart = 0;
	int iPosLast = m_aPos[iPos].iElemPrev;
	if ( bWellFormed )
	{
		nExtractStart = m_aPos[iPos].nStart;
		int nExtractLength = m_aPos[iPos].nLength;
		if ( iPos != iPosLast )
		{
			nExtractLength = m_aPos[iPosLast].nStart - nExtractStart + m_aPos[iPosLast].nLength;
			bWellFormed = false; // treat as subdoc here, but return not well-formed
		}
		node.strMeta.assign( &szSubDoc[nExtractStart], nExtractLength );
	}
	else
	{
		node.strMeta = szSubDoc;
		node.nFlags |= MNF_WITHNOLINES;
	}

	// Insert
	int nReplace = x_InsertNew( iPosParent, iPosBefore, node );

	// Adjust and link in the inserted elements
	// iPosVirtual will stop it from affecting rest of document
	int nAdjust = node.nStart - nExtractStart;
	if ( iPos && nAdjust )
	{
		x_Adjust( iPos, nAdjust );
		m_aPos[iPos].nStart += nAdjust;
	}
	int iPosChild = iPos;
	while ( iPosChild )
	{
		int iPosNext = m_aPos[iPosChild].iElemNext;
		x_LinkElem( iPosParent, iPosBefore, iPosChild );
		iPosBefore = iPosChild;
		iPosChild = iPosNext;
	}
	x_ReleasePos( iPosVirtual );

	// Now adjust remainder of document
	x_Adjust( iPosLast, (int)node.strMeta.size() - nReplace, true );

	// Set position to top element of subdocument
	if ( nFlags & MNF_CHILD )
		x_SetPos( m_iPosParent, iPosParent, iPos );
	else // Main
		x_SetPos( m_iPosParent, iPos, 0 );
	return bWellFormed;
}

int CMarkupSTL::x_RemoveElem( int iPos )
{
	// Remove element and all contained elements
	// Return new position
	//
	if ( ! iPos )
		return 0;

	// Determine whether any whitespace up to next tag
	int nAfterEnd = m_aPos[iPos].StartAfter();
	const char* szDoc = m_strDoc.c_str();
	int nChar = nAfterEnd;
	if ( ! x_FindAny(szDoc,nChar) || szDoc[nChar] == '<' )
		nAfterEnd = nChar;

	// Remove from document, adjust affected indexes, and unlink
	int nLen = nAfterEnd - m_aPos[iPos].nStart;
	x_DocChange( m_aPos[iPos].nStart, nLen, string() );
	x_Adjust( iPos, - nLen, true );
	int iPosPrev = x_UnlinkElem( iPos );
	x_CheckSavedPos();
	return iPosPrev;
}

void CMarkupSTL::x_LinkElem( int iPosParent, int iPosBefore, int iPos )
{
	// Link in element, and initialize nFlags, and iElem indexes
	ElemPos* pElem = &m_aPos[iPos];
	pElem->iElemParent = iPosParent;
	if ( iPosBefore )
	{
		// Link in after iPosBefore
		pElem->nFlags &= ~MNF_FIRST;
		pElem->iElemNext = m_aPos[iPosBefore].iElemNext;
		if ( pElem->iElemNext )
			m_aPos[pElem->iElemNext].iElemPrev = iPos;
		else
			m_aPos[m_aPos[iPosParent].iElemChild].iElemPrev = iPos;
		m_aPos[iPosBefore].iElemNext = iPos;
		pElem->iElemPrev = iPosBefore;
	}
	else
	{
		// Link in as first child
		pElem->nFlags |= MNF_FIRST;
		if ( m_aPos[iPosParent].iElemChild )
		{
			pElem->iElemNext = m_aPos[iPosParent].iElemChild;
			pElem->iElemPrev = m_aPos[pElem->iElemNext].iElemPrev;
			m_aPos[pElem->iElemNext].iElemPrev = iPos;
			m_aPos[pElem->iElemNext].nFlags ^= MNF_FIRST;
		}
		else
		{
			pElem->iElemNext = 0;
			pElem->iElemPrev = iPos;
		}
		m_aPos[iPosParent].iElemChild = iPos;
	}
	if ( iPosParent )
		pElem->SetLevel( m_aPos[iPosParent].Level() + 1 );
}

int CMarkupSTL::x_UnlinkElem( int iPos )
{
	// Fix links to remove element and mark as deleted
	// return previous position or zero if none
	ElemPos* pElem = &m_aPos[iPos];

	// Find previous sibling and bypass removed element
	int iPosPrev = 0;
	if ( pElem->nFlags & MNF_FIRST )
	{
		if ( pElem->iElemNext ) // set next as first child
		{
			m_aPos[pElem->iElemParent].iElemChild = pElem->iElemNext;
			m_aPos[pElem->iElemNext].iElemPrev = pElem->iElemPrev;
			m_aPos[pElem->iElemNext].nFlags |= MNF_FIRST;
		}
		else // no children remaining
			m_aPos[pElem->iElemParent].iElemChild = 0;
	}
	else
	{
		iPosPrev = pElem->iElemPrev;
		m_aPos[iPosPrev].iElemNext = pElem->iElemNext;
		if ( pElem->iElemNext )
			m_aPos[pElem->iElemNext].iElemPrev = iPosPrev;
		else
			m_aPos[m_aPos[pElem->iElemParent].iElemChild].iElemPrev = iPosPrev;
	}
	x_ReleaseSubDoc( iPos );
	return iPosPrev;
}

int CMarkupSTL::x_ReleasePos( int iPos )
{
	int iPosNext = m_aPos[iPos].iElemNext;
	m_aPos[iPos].iElemNext = m_iPosDeleted;
	m_aPos[iPos].nFlags = MNF_DELETED;
	m_iPosDeleted = iPos;
	return iPosNext;
}

int CMarkupSTL::x_ReleaseSubDoc( int iPos )
{
	// Mark position structures as deleted by depth first traversal
	// Tricky because iElemNext used in traversal is overwritten for linked list of deleted
	// Return value is what iElemNext was before being overwritten
	//
	int iPosNext = 0, iPosTop = iPos;
	while ( 1 )
	{
		if ( m_aPos[iPos].iElemChild )
			iPos = m_aPos[iPos].iElemChild;
		else
		{
			while ( 1 )
			{
				iPosNext = x_ReleasePos( iPos );
				if ( iPos == iPosTop )
					return iPosNext;
				if ( iPosNext )
					break;
				iPos = m_aPos[iPos].iElemParent;
			}
			iPos = iPosNext;
		}
	}
	return iPosNext;
}

void CMarkupSTL::x_CheckSavedPos()
{
	// Remove any saved