使用CInternetSession和CHttpFile提取网页源码

程序界面：

为编辑框添加一个对应的变量：m_sUrl，再为Button1添加点击事件，主要代码如下：

?

void CGetHtmlCodeDlg::OnButton1()

{

    CString sHtml;

 

    UpdateData();

    GetHtmlCode(m_sUrl, sHtml);

    WritetoFile(sHtml);

    MessageBox("提取网页源码完成");

}

 

void CGetHtmlCodeDlg::GetHtmlCode(const CString url, CString &html)

{

    CInternetSession session;

    CHttpFile * file = NULL;

     

    try

    {

        file = (CHttpFile *)session.OpenURL(url);

        if(file)

        {

            CString line;

            while(file->ReadString(line) != NULL)

            {

                html += line + "\n";

            }

            file->Close();

        }

    }

    catch (CInternetException* e)

    {

        e->Delete();

        html = "CInternet Exception";

    }

    file = NULL;

    session.Close();

}

 

BOOL CGetHtmlCodeDlg::WritetoFile(const CString& sValue)

{

    CString sFile = GetExePath() + "\\1.html";

     

    CStdioFile file;

    if(file.Open(sFile, CFile::modeCreate | CFile::modeWrite | CFile::modeNoTruncate))

    {

        file.SeekToEnd();               // 移动文件指针到末尾

        file.WriteString(sValue);

        file.Close();

    }

    return FALSE;

}

 

CString CGetHtmlCodeDlg::GetExePath()

{

    char sFileName[256] = {0};

    CString sPath = _T("");

     

    GetModuleFileName(AfxGetInstanceHandle(), sFileName, 255);

    sPath.Format("%s", sFileName);

    int pos = sPath.ReverseFind('\\');

    if(pos != -1)

        sPath = sPath.Left(pos);

    else

        sPath = _T("");

     

    return sPath;

}