用ASP.NET實現的資料擷取

//先按照下面的欄位建立一個資料表 //這個程式寫的不是太好，全都是用for迴圈遍曆出來的，效率不是太高，那位高手可以使用 多線程 指點一下

//先按照下面的欄位建立一個資料表
public partial class Form2 : Form
    {
        public Form2()
        {
            InitializeComponent();
        }
        //姓名
        public static string XM = "";
        //年齡
        public static string nl = "";
        //性別
        public static string XB = "";
        //身高
        public static string SG = "";
        //政治面貌
        public static string mm = "";
        //民族
        public static string MZ = "";
        //學曆
        public static string XL = "";
        //婚姻狀況
        public static string HK = "";
        //所學專業
        public static string ZY = "";
        //工作經驗
        public static string GZJY = "";
        //在職單位
        public static string ZZDW = "";
        //在職職位
        public static string ZZZW = "";
        //工作經曆
        public static string GZJL = "";
        //要求月薪
        public static string YX = "";
        //工作性質
        public static string GZXZ = "";
        //求職意向
        public static string QZYX = "";
        //具體職務
        public static string JTZW = "";
        //期望工作地
        public static string QWGZD = "";
        //教育情況，語言水平，技術專長
        public static string QT = "";
        private void button1_Click(object sender, EventArgs e)
        {
            label1.Text = "正在採集資料……";

     //遍曆資料的頁數
            for (int i = 1; i <=50; i++)
            {
                CJ("http://www.xcjob.cn/renli.asp?pageno=" + i);
            }

            label1.Text = "恭喜你採集完成！";
            MessageBox.Show("恭喜你採集完成！");
        }

        //採集資料
        private void CJ(string Url)
        {
     //獲得頁面源檔案(Html)
            string strWebContent = YM(Url);

            //按照Ｈtml裡面的標籤  取出和資料有關的那段源碼
            int iBodyStart = strWebContent.IndexOf("<body", 0);
            int aaa = strWebContent.IndexOf("關鍵字:", iBodyStart);
            int iTableStart = strWebContent.IndexOf("<table", aaa);
            int iTableEnd = strWebContent.IndexOf("</table>", iTableStart);
            string strWeb = strWebContent.Substring(iTableStart, iTableEnd - iTableStart);

            //產生HtmlDocument
            HtmlElementCollection htmlTR = HtmlTR_Content(strWeb, "tr");

            foreach (HtmlElement tr in htmlTR)
            {
                try
                {
      //姓名
                    XM = tr.GetElementsByTagName("a")[0].InnerText;
      //獲得詳細資料頁面的網址
                    string a = tr.GetElementsByTagName("a")[0].GetAttribute("href").ToString();
                    a = "http://www.xcjob.cn" + a.Substring(11);

                    Content(a);
                }
                catch { }
            }
        }

        //採集詳細資料
        private void Content(string URL)
        {
            try
            {
                string strWebContent = YM(URL);

                //按照Ｈtml裡面的標籤 取出和資料有關的那段源碼
                int iBodyStart = strWebContent.IndexOf("<body", 0);
                int iTableStart = strWebContent.IndexOf("瀏覽次數", iBodyStart);
                int iTableEnd = strWebContent.IndexOf("<table", iTableStart);
                int dd = strWebContent.IndexOf("</table>", iTableEnd);
                string strWeb = strWebContent.Substring(iTableEnd, dd - iTableEnd + 8);

                HtmlElementCollection htmlTR = HtmlTR_Content(strWeb, "table");

                foreach (HtmlElement tr in htmlTR)
                {
                    try
                    {
                        //年齡
                        nl = tr.GetElementsByTagName("tr")[1].GetElementsByTagName("td")[1].InnerText;
                        //性別
                        string XB_SG = tr.GetElementsByTagName("tr")[1].GetElementsByTagName("td")[3].InnerText;
                        XB = XB_SG.Substring(0, 1);
                        //身高
                        SG = XB_SG.Substring(11);
                        //政治面貌
                        mm = tr.GetElementsByTagName("tr")[2].GetElementsByTagName("td")[1].InnerText;
                        //民族
                        MZ = tr.GetElementsByTagName("tr")[2].GetElementsByTagName("td")[3].InnerText;
                        //學曆
                        XL = tr.GetElementsByTagName("tr")[3].GetElementsByTagName("td")[1].InnerText;
                        //婚煙狀況
                        HK = tr.GetElementsByTagName("tr")[3].GetElementsByTagName("td")[3].InnerText;
                        //所學專業
                        ZY = tr.GetElementsByTagName("tr")[5].GetElementsByTagName("td")[1].InnerText;
                        //工作經驗
                        GZJY = tr.GetElementsByTagName("tr")[5].GetElementsByTagName("td")[3].InnerText;
                        //在職單位
                        ZZDW = tr.GetElementsByTagName("tr")[6].GetElementsByTagName("td")[1].InnerText;
                        //在職職位
                        ZZZW = tr.GetElementsByTagName("tr")[6].GetElementsByTagName("td")[3].InnerText;
                        //工作經曆
                        GZJY = tr.GetElementsByTagName("tr")[7].GetElementsByTagName("td")[1].InnerText;
                        //要求月薪
                        YX = tr.GetElementsByTagName("tr")[9].GetElementsByTagName("td")[1].InnerText;
                        //工作性質
                        GZXZ = tr.GetElementsByTagName("tr")[9].GetElementsByTagName("td")[3].InnerText;
                        //求職意向
                        QZYX = tr.GetElementsByTagName("tr")[10].GetElementsByTagName("td")[1].InnerText;
                        //具體職務
                        JTZW = tr.GetElementsByTagName("tr")[10].GetElementsByTagName("td")[3].InnerText;
                        //期望工作地
                        QWGZD = tr.GetElementsByTagName("tr")[11].GetElementsByTagName("td")[1].InnerText;
                        //教育情況，語言水平，技術專長
                        QT = tr.GetElementsByTagName("tr")[13].GetElementsByTagName("td")[1].InnerText;

                        insert();
                    }
                    catch
                    { }
                }
            }
            catch { }
        }

      //將資料插入資料庫 
      private void insert()
        {
            try
            {
                string str = "Provider=Microsoft.Jet.OleDb.4.0;Data Source=Data.mdb";
                string sql = "insert into 人才資訊 (姓名,年齡,性別,身高,政治面貌,民族,學曆,婚煙狀況,所學專業,";
                sql += "工作經驗,在職單位,在職職位,工作經曆,要求月薪,工作性質,求職意向,具體職務,期望工作地,其他) values ";
                sql += "('" + XM + "'," + nl + ",'" + XB + "','" + SG + "','" + mm + "','" + MZ + "','" + XL + "','" + HK + "','" + ZY + "','" + GZJY + "','" + ZZDW + "','" + ZZZW + "',";
                sql += "'" + GZJY + "','" + YX + "','" + GZXZ + "','" + QZYX + "','" + JTZW + "','" + QWGZD + "','" + QT + "')";

                OleDbConnection con = new OleDbConnection(str);
                OleDbCommand com = new OleDbCommand(sql, con);
                con.Open();
                com.ExecuteNonQuery();
                con.Close();
            }
     catch { }
        }

        //返回一個HtmlElementCollection,然後進行查詢內容
        private HtmlElementCollection HtmlTR_Content(string strWeb, string tj)
        {
            try
            {
                //產生HtmlDocument
                WebBrowser webb = new WebBrowser();
                webb.Navigate("about:blank");
                //window.document返回一個htmldocument對象，表示對一個html文檔的操作
                //htmldocument對象是在xmldocument基礎上建立的，具有xmldocument的一切方法屬性
                HtmlDocument htmldoc = webb.Document.OpenNew(true);
                htmldoc.Write(strWeb);
                HtmlElementCollection htmlTR = htmldoc.GetElementsByTagName(tj);

                return htmlTR;
            }
            catch { return null; }
        }

        //獲得網址原代碼
        private string YM(string Url)
        {
            string strResult = "";

            try
            {
                HttpWebRequest request = (HttpWebRequest)WebRequest.Create(Url);
                request.Method = "GET";
                HttpWebResponse response = (HttpWebResponse)request.GetResponse();
                Stream streamReceive = response.GetResponseStream();
                Encoding encoding = Encoding.GetEncoding("GB2312");
                StreamReader streamReader = new StreamReader(streamReceive, encoding);
                strResult = streamReader.ReadToEnd();
            }
            catch { }

            return strResult;
        }
    }

    //這個程式寫的不是太好，全都是用for迴圈遍曆出來的，效率不是太高，那位高手可以使用 多線程 指點一下