Question

I am working with a webbrowser host on c++, I managed to sink event and I am running this void on DISPID_DOCUMENTCOMPLETE:

void DocumentComplete(LPDISPATCH pDisp, VARIANT *url)
{
    READYSTATE rState;

    iBrowser->get_ReadyState(&rState);

    if(rState == READYSTATE_COMPLETE)
    {
        HRESULT hr;
        IDispatch *pHtmlDoc = NULL;
        IHTMLDocument2 *pDocument = NULL;
        IHTMLElement *pBody = NULL;
        IHTMLElement *lpParentElm = NULL;
        BSTR bstrHTMLText;

        hr = iBrowser->get_Document(&pHtmlDoc);

        hr = pHtmlDoc->QueryInterface(IID_IHTMLDocument2, (void**)&pDocument);

        if( (FAILED(hr)) || !pDocument)
        {
            MessageBox(NULL, "QueryInterface failed", "WebBrowser", MB_OK);
        }

        hr = pDocument->get_body( &pBody );

        if( (!SUCCEEDED(hr)) || !pBody)
        {
            MessageBox(NULL, "get_body failed", "WebBrowser", MB_OK);
        }

        pBody->get_parentElement(&lpParentElm);

        lpParentElm->get_outerHTML(&bstrHTMLText);

        _bstr_t bstr_t(bstrHTMLText);

        std::string sTemp(bstr_t);

        MessageBox(NULL, sTemp.c_str(), "WebBrowser", MB_OK);
    }
}

I don’t much about c++, I built this code from watching other codes in google. Now I know I have to use ->Release, but do I have to use all these?:

    pHtmlDoc->Release();
    pDocument->Release();
    pBody->Release();
    lpParentElm->Release();
    iBrowser->Release();

Because on the examples I used to build my code it was using it only for the IHTMLElement(s).

Answer 1

Yes, you do have to call Release() on those pointers, otherwise objects will leak. The same goes for BSTRs.

You’ll be much better off if you use smart pointers for that – ATL::CComPtr/ATL::CComBSTR or _com_ptr_t/_bstr_t.