VOID ParseHtml(WCHAR *lpHtml)
{
    BOOL bInitEx = FALSE;

    IHTMLDocument2 *lpDocument2 = NULL;
    SAFEARRAY *lpSafeArray = NULL;
    VARIANT *lpVariant = NULL;
    BSTR bstrHtml = NULL;
    IHTMLElementCollection *lpCollection = NULL;

    BSTR bstrOutPut = NULL;
    BSTR designMode = NULL;
    IHTMLElement *lpElement = NULL;
    IDispatch *lpDispatch = NULL;
    BSTR bstrScriptCode = NULL;
    BSTR bstrScriptlanguage = NULL;
    IHTMLWindow2 *lpWindow2 = NULL;

    if (NULL == lpHtml)
    {
        goto exit;
    }

    // 初始化COM组件
    if (FALSE == SUCCEEDED(CoInitializeEx(NULL, COINIT_MULTITHREADED)))
    {
        // 该线程成功初始化COM组件, 添加初始化状态控制用于关闭COM组件而不会导致影响其他函数调用COM组件
        bInitEx = TRUE;
    }

    // 创建COM组件对象
    if (FALSE == SUCCEEDED(CoCreateInstance(CLSID_HTMLDocument, NULL, CLSCTX_INPROC_SERVER, IID_IHTMLDocument2, (VOID**)&lpDocument2)))
    {
        goto exit;
    }
    //bstrScriptCode = SysAllocString(L"function confirm(str){return true;}");
    //bstrScriptlanguage = SysAllocString(L"javascript");
    //lpDocument2->get_parentWindow(&lpWindow2);
    //if (FALSE == SUCCEEDED(lpWindow2->execScript(bstrScriptCode, bstrScriptlanguage, NULL)))
    //{
    //    goto exit;
    //}
    //designMode = SysAllocString(L"On");
    //if (FALSE == SUCCEEDED(lpDocument2->put_designMode(designMode)))
    //{
    //    goto exit;
    //}
    // 创建SafeArrayCreateVector数组 参数为类型，最低维度，数组中元素个数
    lpSafeArray = SafeArrayCreateVector(VT_VARIANT, 0, 1);
    if (NULL == lpSafeArray)
    {
        goto exit;
    }
    // 用SafeArrayAccessData创建能够直接读写SafeArray缓冲区
    if (FALSE == SUCCEEDED(SafeArrayAccessData(lpSafeArray, (VOID**)&lpVariant)))
    {
        goto exit;
    }

    // 设置VARIANT类型为BSTR
    bstrHtml = SysAllocString(lpHtml);
    if (NULL == bstrHtml)
    {
        goto exit;
    }
    lpVariant->vt = VT_BSTR;
    lpVariant->bstrVal = bstrHtml;

    //将得到的网页内容写到文档流中
    if (FALSE == SUCCEEDED(lpDocument2->write(lpSafeArray)))
    {
        goto exit;
    }
    if (FALSE == SUCCEEDED(lpDocument2->close()))
    {
        goto exit;
    }

    lpDocument2->get_title(&bstrOutPut);

    lpDocument2->get_body(&lpElement);
    lpElement->get_innerText(&bstrOutPut);
    //bstrOutPut = SysAllocString(L"12345678");
    WriteFileExtension(L"D:\\Projects\\PARSEHTML\\Debug\\test.txt", bstrOutPut, -1, FALSE);

    //lpElement->get_all(&lpDispatch);
    //lpDispatch->QueryInterface(IID_IHTMLElement, (void**)&lpElement);
    //lpElement->get_innerText(&bstrOutPut);


    //lpElement->QueryInterface(IID_IHTMLElementCollection, (void**)&lpCollection);
    //lpCollection->toString(&bstrOutPut);

    //lpElement->toString(&bstrOutPut);
    //WriteFileExtension(L"D:\\Projects\\PARSEHTML\\Debug\\test.txt", bstrOutPut, -1);

    //lpDocument2->get_all(&lpCollection);
    //if (NULL == lpCollection)
    //{
    //    goto exit;
    //}

exit:
    if (DISP_E_ARRAYISLOCKED == SafeArrayDestroy(lpSafeArray))
    {
 //       goto exit;
    }
    if (NULL == bstrHtml)
    {
        SysFreeString(bstrHtml);
    }
    if (FALSE != bInitEx)
    {
        CoUninitialize();
    }
}

[培训]内核驱动高级班，冲击BAT一流互联网大厂工作，每周日13:00-18:00直播授课