ASP.NET實現(xiàn)數(shù)據(jù)采集

2019-11-17 04:14:42

字體：大中小

供稿：網(wǎng)友

第一次寫博客，寫的不好請大家不要見怪，這是我自己寫的一個數(shù)據(jù)采集程序，是采集一個人才網(wǎng)里面人才的信息！

//先按照下面的字段創(chuàng)建一個數(shù)據(jù)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(xué)歷
        public static string XL = "";
        //婚姻狀況
        public static string HK = "";
        //所學(xué)專業(yè)
        public static string ZY = "";
        //工作經(jīng)驗
        public static string GZJY = "";
        //在職單位
        public static string ZZDW = "";
        //在職職位
        public static string ZZZW = "";
        //工作經(jīng)歷
        public static string GZJL = "";
        //要求月薪
        public static string YX = "";
        //工作性質(zhì)
        public static string GZXZ = "";
        //求職意向
        public static string QZYX = "";
        //具體職務(wù)
        public static string JTZW = "";
        //期望工作地
        public static string QWGZD = "";
        //教育情況，語言水平，技術(shù)專長
        public static string QT = "";

        PRivate void button1_Click(object sender, EventArgs e)
        {
            label1.Text = "正在采集數(shù)據(jù)……";

     //遍歷數(shù)據(jù)的頁數(shù)
            for (int i = 1; i <=50; i++)
            {
                CJ("http://www.xcjob.cn/renli.asp?pageno=" + i);
            }

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

        //采集數(shù)據(jù)
        private void CJ(string Url)
        {
     //獲得頁面源文件(Html)
            string strWebContent = YM(Url);

            //按照Ｈtml里面的標(biāo)簽取出和數(shù)據(jù)有關(guān)的那段源碼
            int iBodyStart = strWebContent.IndexOf("<body", 0);
            int aaa = strWebContent.IndexOf("關(guān)鍵字:", iBodyStart);
            int iTableStart = strWebContent.IndexOf("<table", aaa);
            int iTableEnd = strWebContent.IndexOf("</table>", iTableStart);
            string strWeb = strWebContent.Substring(iTableStart, iTableEnd - iTableStart);

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

            foreach (HtmlElement tr in htmlTR)
            {
                try
                {
      //姓名
                    XM = tr.GetElementsByTagName("a")[0].InnerText;
      //獲得詳細信息頁面的網(wǎng)址
                    string a = tr.GetElementsByTagName("a")[0].GetAttribute("href").ToString();
                    a = "

                    Content(a);
                }
                catch { }
            }
        }

        //采集詳細數(shù)據(jù)
        private void Content(string URL)
        {
            try
            {
                string strWebContent = YM(URL);

                //按照Ｈtml里面的標(biāo)簽取出和數(shù)據(jù)有關(guān)的那段源碼
                int iBodyStart = strWebContent.IndexOf("<body", 0);
                int iTableStart = strWebContent.IndexOf("瀏覽次數(shù)", 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");

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

      //將數(shù)據(jù)插入數(shù)據(jù)庫
      private void insert()
        {
            try
            {
                string str = "Provider=Microsoft.Jet.OleDb.4.0;Data Source=Data.mdb";
                string sql = "insert into 人才信息 (姓名,年齡,性別,身高,政治面貌,民族,學(xué)歷,婚煙狀況,所學(xué)專業(yè),";
                sql += "工作經(jīng)驗,在職單位,在職職位,工作經(jīng)歷,要求月薪,工作性質(zhì),求職意向,具體職務(wù),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,然后進行查詢內(nèi)容
        private HtmlElementCollection HtmlTR_Content(string strWeb, string tj)
        {
            try
            {
                //生成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(chǔ)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(wǎng)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;
        }
    }

//這個程序?qū)懙牟皇翘茫际怯胒or循環(huán)遍歷出來的，效率不是太高，那位高手可以使用多線程指點一下

上一篇：ASP.NET母版頁TreeView導(dǎo)航鏈接問題

下一篇：c#實現(xiàn)簡單的發(fā)送郵件