How to get the most current IHTMLDocument2 object from IE DOM - c++

Question

Currently, I use MSAA to get an IHTMLDocument2 object from a IE HWND. However, with some complicated web applications, this IHTMLDocument2 object may contain several IHTMLDocument2 objects, some of them are not belong to the current displaying page, but the previous page.
It seems to me, IE sometimes doesn't refesh its DOM object, but keep adding more IHTMLDocument2 object into its DOM. My question is how can I get the current displaying IHTMLDocument2 object from the DOM object.
Thanks in advance
Update
Hi Remy,
Thanks for your answer.
Yes, you are right, I do use frames to get to other IHTMLDocument2 objects. My understanding is that the IHTMLDocument2 object that I get from a HWND is the top object in its DOM. IE sometimes puts the prevous IHTMLDocument2 objects inside one of the frames as well.
Here is part of my code.
BOOL IESpy::GetHTMLText( CComPtr<IHTMLDocument2> spDoc, int tagNo, int schNo)
{
USES_CONVERSION;
HRESULT hr = NULL;
BOOL res = TRUE;
BOOL doneSearch = FALSE;
// Extract the source code of the document
if (spDoc) {
IHTMLFramesCollection2* pFrames = NULL;
if (hr = (spDoc->get_frames(&pFrames)) == S_OK){
LONG framesCount;
pFrames->get_length(&framesCount);
if (framesCount > 0) {
for( long i=0; i < framesCount; i++) {
VARIANT varIdx;
varIdx.vt=VT_I4;
VARIANT varResult;
varIdx.lVal=i;
VariantInit(&varResult);
hr = pFrames->item(&varIdx, &varResult);
if (SUCCEEDED(hr) && (varResult.vt == VT_DISPATCH)){
CComQIPtr<IHTMLWindow2> pFrameWnd;
CComQIPtr<IHTMLDocument2> pFrameDoc;
CComBSTR description=NULL;
pFrameWnd = varResult.pdispVal;
VariantClear(&varResult);
if (pFrameWnd == 0) {
continue;
}
hr = pFrameWnd->get_document(&pFrameDoc);
if (SUCCEEDED(hr) && pFrameDoc){
GetHTMLText( pFrameDoc, tagNo, schNo );
if ( m_foundText ) {
break;
}
} else if ( hr == E_ACCESSDENIED ) {
CComQIPtr<IWebBrowser2> spBrws = HtmlWindowToHtmlWebBrowser(pFrameWnd);
if ( spBrws != NULL) {
// Get the document object from the IWebBrowser2 object.
CComQIPtr<IDispatch> spDisp;
hr = spBrws->get_Document(&spDisp);
if ( hr == S_OK ) {
pFrameDoc = spDisp;
if ( pFrameDoc ) {
GetHTMLText( pFrameDoc, tagNo, schNo );
if ( m_foundText ) {
break;
}
}
}
}
}
}
}
}
pFrames->Release();
if ( !m_foundText ) {
res = ReadSearchText(spDoc, tagNo, schNo );
doneSearch = TRUE;
}
}
if ( !m_foundText && doneSearch == FALSE ) {
res = ReadSearchText(spDoc, tagNo, schNo );
}
}
return res;
}
BOOL IESpy::ReadSearchText(CComPtr<IHTMLDocument2> spDoc, int tagNo, int schNo )
{
USES_CONVERSION;
HRESULT hr = NULL;
BOOL found = FALSE;
IHTMLElementCollection *pAll;
hr = spDoc->get_all(&pAll);
if (FAILED(hr)) {
return FALSE;
}
long items;
IDispatch *ppvDisp;
IHTMLElement *ppvElement;
pAll->get_length(&items);
std::wstring foundText = L"";
for ( long j = 0; j < items; j++ ) {
VARIANT index;
index.vt = VT_I4;
index.lVal = j;
hr = pAll->item( index, index, &ppvDisp );
if (FAILED(hr)) {
return FALSE;
}
if ( ppvDisp ) {
ppvDisp->QueryInterface(IID_IHTMLElement, (void **)&ppvElement);
if ( ppvElement ) {
CComBSTR bstrTag;
ppvElement->get_tagName(&bstrTag);
wchar_t *wtemp = OLE2W(bstrTag);
if ( wtemp ) {
std::wstring text = ReadSearchText(ppvElement, wtemp, tagNo, schNo, found);
if ( !text.empty() ) {
if ( !foundText.empty() ) {
foundText += concat_string;
}
foundText += text;
}
ppvElement->Release();
if ( found ) {
BOOL stop = FALSE;
for ( size_t i = 0; i < m_tagName[tagNo]->size(); i++ ) {
if ( wcscmp(m_tagName[tagNo]->at(i).c_str(), L"HTML") == 0
|| wcscmp(m_tagName[tagNo]->at(i).c_str(), L"HEAD") == 0
|| wcscmp(m_tagName[tagNo]->at(i).c_str(), L"BODY") == 0 ) {
stop = TRUE;
break;
}
}
if ( stop ) {
break;
}
}
} else {
ppvElement->Release();
}
}
}
}
if ( !foundText.empty() ) {
if ( m_screenCompare ) {
// long timeStamp = GetHPTimeStamp(spDoc);
// m_temp_results[timeStamp] = foundText;
m_temp_results.push_back(foundText);
} else {
m_result += foundText;
m_result += L" ";
m_foundText = TRUE;
}
}
return TRUE;
}

Answer 1

An IHTMLDocument2 cannot contain other IHTMLDocument2 objects (unless they belong to frames on the page), and certainly not from previous pages. How are you determining that exactly? Can you show some code?