使用IHTMLDocument2分析Html

http://hi.baidu.com/ahstudy2010/blog/item/102d7df3468f3ca4a40f52be.html

利用CWebBrowser2获取IHTMLDocument2接口，IHTMLDocument2接口是mshtml中最重要的接口之一，取得这个接口，就可以开始分析html代码

下面的代码将获取特定的html 文本标签

// 从ctlBrowser当前html文档选中区域中获取名为strTagName的标签，存入*ppElement中
BOOL GetCrntHtmlElement(CWebBrowser2  &ctlBrowser, const CString &strTagName, IHTMLElement **ppElement)
{
    *ppElement = NULL; // initial as NULL
     CString strText;
     CComBSTR cbsText, cbsTagName;
     CComPtr<IHTMLTxtRange> spTxtRange = NULL;

    static CString SELECTTION_TYPE_TEXT = _T("text");
    static CString SELECTTION_TYPE_CONTROL = _T("control");
    static CString SELECTTION_TYPE_NONE = _T("none");

    try
     {
        // calculate the first control based on the user selection
         HRESULT hr = E_FAIL;
         CComPtr<IDispatch> spDispDoc = ctlBrowser.get_Document();
         CComPtr<IHTMLDocument2> spHtmlDoc = NULL;
         hr = spDispDoc->QueryInterface(IID_IHTMLDocument2, (void**)&spHtmlDoc);
        if (FAILED(hr) || NULL == spHtmlDoc)
         {
            return FALSE;
         }
         CComPtr<IHTMLSelectionObject> spSelection = NULL;
         hr = spHtmlDoc->get_selection(&spSelection);
        if(SUCCEEDED(hr) && spSelection)
         {
             hr = spSelection->get_type(&cbsText);
             strText = cbsText;
            if (0 == strText.CompareNoCase(SELECTTION_TYPE_TEXT)
                || 0 == strText.CompareNoCase(SELECTTION_TYPE_NONE) )
             {
                 CComPtr<IDispatch> spTextRangeDisp = NULL;
                 hr = spSelection->createRange(&spTextRangeDisp);
                if(SUCCEEDED(hr) && spTextRangeDisp)
                 {
                     hr = spTextRangeDisp->QueryInterface(IID_IHTMLTxtRange, (void**)&spTxtRange);
                    if(SUCCEEDED(hr) && spTxtRange) // get a text range here
                     {                
                         CComPtr<IHTMLElement> spElement=NULL;
                        // get self element first for tag: such as "h6"
                         hr = spTxtRange->QueryInterface(IID_IHTMLElement, (void**)&spElement);
                        if (SUCCEEDED(hr) && spElement)
                         {
                             hr = spElement->get_tagName(&cbsTagName);
                            if (SUCCEEDED(hr) && 0 == strTagName.CompareNoCase(CString(cbsTagName)) )
                             {    // found   tag now
                                return TRUE;
                             }
                         }
                        // loop search in parent for tag: such as "h6"
                         spElement=NULL;
                         CComPtr<IHTMLElement> spParent = NULL;
                         hr  =spTxtRange->parentElement (&spParent);
                        while (SUCCEEDED(hr) && spParent)
                         {
                             hr = spParent->get_tagName(&cbsTagName);
                            if (SUCCEEDED(hr) && 0 == strTagName.CompareNoCase(CString(cbsTagName)) )
                             {    // found tag now
                                return TRUE;
                             }
                             spElement = spParent;
                             spParent = NULL;
                             hr = spElement->get_parentElement(&spParent);
                         }
                     }
                 }
             }
         }
     }
    catch(CException* e)
     {
         e->ReportError();
         e->Delete();        
     }

    return FALSE;
}