Question

This snippet of code generate huge memory leaks. Could you help me to find out where it happens?

This code does the following thing:

1) It gets IHTMLDocuments2 interface
2) Requests for all tags collection
3) Iterates over whole collection
4) and adds some of the tags’ data to list

IDispatch* pDisp;
pDisp = this->GetHtmlDocument();

if (pDisp != NULL ) 
{
    IHTMLDocument2* pHTMLDocument2;
    HRESULT hr;
    hr = pDisp->QueryInterface( IID_IHTMLDocument2,(void**)&pHTMLDocument2 );
    if (hr == S_OK)
    {
                    // I know that I could use IHTMLDocument3 interface to get collection by ID
                    // but it didn't worked and returned NULL on each call.
        IHTMLElementCollection* pColl = NULL;
                    // get all tags
        hr = pHTMLDocument2->get_all( &pColl );
        if (hr == S_OK && pColl != NULL)
        {
            LONG celem;
            hr = pColl->get_length( &celem );
            if ( hr == S_OK )
            {
                                    //iterate through all tags
                                    // if I iterate this block of code in cycle, it 
                                    // uses memory available upto 2GBs and then
                                    // app crashes
                for ( int i=0; i< celem; i++ )
                {                       
                    VARIANT varIndex;
                    varIndex.vt = VT_UINT;
                    varIndex.lVal = i;
                    VARIANT var2;
                    VariantInit( &var2 );
                    IDispatch* pElemDisp = NULL;
                    hr = pColl->item( varIndex, var2, &pElemDisp );
                    if ( hr == S_OK && pElemDisp != NULL)
                    {
                        IHTMLElement* pElem;
                        hr = pElemDisp->QueryInterface(IID_IHTMLElement,(void **)&pElem);
                        if ( hr == S_OK)
                        {                   
                                                            // check INPUT tags only
                            BSTR tagNameStr = L"";
                            pElem->get_tagName(&tagNameStr);
                            CString tagname(tagNameStr);
                            SysFreeString(tagNameStr);
                            tagname.MakeLower();
                            if (tagname != "input")
                            {
                                continue;
                            }
                                                            //get ID attribute
                            BSTR bstr = L"";
                            pElem->get_id(&bstr);
                            CString idStr(bstr);
                            SysFreeString(bstr);

                            if (RequiredTag(pElem)) 
                            {       
                                AddTagToList(pElem);
                            }
                                                            //release all objects
                            pElem->Release();
                        }
                        pElemDisp->Release();
                    }
                }
            }
                            // I looked over this code snippet many times and couldn't find what I'm missing here...
            pColl->Release();
        }
        pHTMLDocument2->Release();
    }
    pDisp->Release();       
}

Answer 1

Inside your loop, for each retreived element that does not have a tagname of "input" (which will be most elements), you are not calling pElem->Release() when calling continue, so you are leaking them:

if (tagname != "input") 
{ 
    pElem->Release(); // <-- add this
    continue; 
} 

With that said, you should re-write your code to use ATL’s smart pointer classes (CComPtr, CComQIPtr, CComBSTR, etc) to manage the memory for you so you do not have to manually release everything yourself anymore, eg:

CComPtr<IDispatch> pDisp;
pDisp.Attach(this->GetHtmlDocument());
if (pDisp.p != NULL)    
{   
    CComQIPtr<IHTMLDocument2> pHTMLDocument2(pDisp);   
    if (pHTMLDocument2.p != NULL)   
    {   
        CComPtr<IHTMLElementCollection> pColl;   
        pHTMLDocument2->get_all(&pColl);
        if (pColl.p != NULL)
        {   
            LONG celem;   
            if (SUCCEEDED(pColl->get_length(&celem)))   
            {   
                for (LONG i = 0; i < celem; ++i)   
                {                          
                    VARIANT varIndex;   
                    varIndex.vt = VT_UINT;   
                    varIndex.lVal = i;   
                    VARIANT var2;   
                    VariantInit( &var2 );   
                    CComPtr<IDispatch> pElemDisp;   
                    pColl->item( varIndex, var2, &pElemDisp );   
                    if (pElemDisp.p != NULL)   
                    {   
                        CComQIPtr<IHTMLElement> pElem(pElemDisp);   
                        if (pElem.p != NULL)   
                        {                      
                            CComBSTR tagNameStr;   
                            pElem->get_tagName(&tagNameStr);   

                            if (lstrcmpiW(tagNameStr.m_str, L"input") != 0)
                                continue;   

                            CComBSTR idStr;   
                            pElem->get_id(&idStr);   

                            if (RequiredTag(pElem))    
                                AddTagToList(pElem);   
                        }   
                    }   
                }   
            }   
        }   
    }   
}