int CMarkup::x_ParseError( char* szError, char* szName )
{
if ( szName )
szError = szName;
else
m_csError = szError;
x_ReleasePos();
return -1;
}
int CMarkupSTL::x_ParseError( const char * szError, const char * szName )
{
if ( szName )
m_csError.Format( szError, szName );
else
m_csError = szError;
x_ReleasePos();
return -1;
}
int CMarkup::x_ParseError( LPCTSTR szError, LPCTSTR szName )
{
if ( szName )
m_csError.Format( szError, szName );
else
m_csError = szError;
x_ReleasePos();
return -1;
}
int CMarkupSTL::x_ParseElem( int iPosParent )
{
// This is either called by SetDoc, x_AddSubDoc, or itself recursively
// m_aPos[iPosParent].nEndL is where to start parsing for the child element
// This returns the new position if a tag is found, otherwise zero
// In all cases we need to get a new ElemPos, but release it if unused
//
int iPos = x_GetFreePos();
m_aPos[iPos].nStartL = m_aPos[iPosParent].nEndL;
m_aPos[iPos].iElemParent = iPosParent;
m_aPos[iPos].iElemChild = 0;
m_aPos[iPos].iElemNext = 0;
// Start Tag
// A loop is used to ignore all remarks tags and special tags
// i.e. <?xml version="1.0"?>, and <!-- comment here -->
// So any tag beginning with ? or ! is ignored
// Loop past ignored tags
TokenPos token( m_csDoc );
token.nNext = m_aPos[iPosParent].nEndL;
CStdString csName;
while ( csName.IsEmpty() )
{
// Look for left angle bracket of start tag
m_aPos[iPos].nStartL = token.nNext;
if ( ! x_FindChar( token.szDoc, m_aPos[iPos].nStartL, _T('<') ) )
return x_ParseError( _T("Element tag not found") );
// Set parent's End tag to start looking from here (or later)
m_aPos[iPosParent].nEndL = m_aPos[iPos].nStartL;
// Determine whether this is an element, or bypass other type of node
token.nNext = m_aPos[iPos].nStartL + 1;
if ( x_FindToken( token ) )
{
if ( token.bIsString )
return x_ParseError( _T("Tag starts with quote") );
_TCHAR cFirstChar = m_csDoc[token.nL];
if ( cFirstChar == _T('?') || cFirstChar == _T('!') )
{
token.nNext = m_aPos[iPos].nStartL;
if ( ! x_ParseNode(token) )
return x_ParseError( _T("Invalid node") );
}
else if ( cFirstChar != _T('/') )
{
csName = x_GetToken( token );
// Look for end of tag
if ( ! x_FindChar(token.szDoc, token.nNext, _T('>')) )
return x_ParseError( _T("End of tag not found") );
}
else
return x_ReleasePos(); // probably end tag of parent
}
else
return x_ParseError( _T("Abrupt end within tag") );
}
m_aPos[iPos].nStartR = token.nNext;
// Is ending mark within start tag, i.e. empty element?
if ( m_csDoc[m_aPos[iPos].nStartR-1] == _T('/') )
{
// Empty element
// Close tag left is set to ending mark, and right to open tag right
m_aPos[iPos].nEndL = m_aPos[iPos].nStartR-1;
m_aPos[iPos].nEndR = m_aPos[iPos].nStartR;
}
else // look for end tag
{
// Element probably has contents
// Determine where to start looking for left angle bracket of end tag
// This is done by recursively parsing the contents of this element
int iInner, iInnerPrev = 0;
m_aPos[iPos].nEndL = m_aPos[iPos].nStartR + 1;
while ( (iInner = x_ParseElem( iPos )) > 0 )
{
// Set links to iInner
if ( iInnerPrev )
m_aPos[iInnerPrev].iElemNext = iInner;
else
m_aPos[iPos].iElemChild = iInner;
iInnerPrev = iInner;
// Set offset to reflect child
m_aPos[iPos].nEndL = m_aPos[iInner].nEndR + 1;
}
if ( iInner == -1 )
return -1;
// Look for left angle bracket of end tag
if ( ! x_FindChar( token.szDoc, m_aPos[iPos].nEndL, _T('<') ) )
return x_ParseError( _T("End tag of %s element not found"), csName );
// Look through tokens of end tag
token.nNext = m_aPos[iPos].nEndL + 1;
int nTokenCount = 0;
while ( x_FindToken( token ) )
{
++nTokenCount;
if ( ! token.bIsString )
{
// Is first token not an end slash mark?
if ( nTokenCount == 1 && m_csDoc[token.nL] != _T('/') )
return x_ParseError( _T("Expecting end tag of element %s"), csName );
else if ( nTokenCount == 2 && ! token.Match(csName) )
return x_ParseError( _T("End tag does not correspond to %s"), csName );
// Else is it a right angle bracket?
else if ( m_csDoc[token.nL] == _T('>') )
break;
}
}
// Was a right angle bracket not found?
if ( ! token.szDoc[token.nL] || nTokenCount < 2 )
return x_ParseError( _T("End tag not completed for element %s"), csName );
m_aPos[iPos].nEndR = token.nL;
}
// Successfully parsed element (and contained elements)
return iPos;
}
