ASP.NET实现数据采集

        private void button1_Click(object sender, EventArgs e)
        {
            label1.Text = "正在采集数据……";

     //遍历数据的页数 
            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！");
        }

        //采集数据
        private void CJ(string Url)
        {
     //获得页面源文件(Html)
            string strWebContent = YM(Url);

            //按照Ｈtml里面的标签  取出和数据有关的那段源码
            int iBodyStart = strWebContent.IndexOf("<body", 0);
            int aaa = strWebContent.IndexOf("关键字:", 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;
      //获得详细信息页面的网址
                    string a = tr.GetElementsByTagName("a")[0].GetAttribute("href").ToString();
                    a = "http://www.xcjob.cn" + a.Substring(11);

                    Content(a);
                }
                catch { }
            }
        }

        //采集详细数据
        private void Content(string URL)
        {
            try
            {
                string strWebContent = YM(URL);

                //按照Ｈtml里面的标签 取出和数据有关的那段源码
                int iBodyStart = strWebContent.IndexOf("<body", 0);
                int iTableStart = strWebContent.IndexOf("浏览次数", 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
                    {
                        //年龄
                        nl = tr.GetElementsByTagName("tr")[1].GetElementsByTagName("td")[1].InnerText;
                        //性别
                        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;
                        //学历
                        XL = tr.GetElementsByTagName("tr")[3].GetElementsByTagName("td")[1].InnerText;
                        //婚烟状况
                        HK = tr.GetElementsByTagName("tr")[3].GetElementsByTagName("td")[3].InnerText;
                        //所学专业
                        ZY = tr.GetElementsByTagName("tr")[5].GetElementsByTagName("td")[1].InnerText;
                        //工作经验
                        GZJY = tr.GetElementsByTagName("tr")[5].GetElementsByTagName("td")[3].InnerText;
                        //在职单位
                        ZZDW = tr.GetElementsByTagName("tr")[6].GetElementsByTagName("td")[1].InnerText;
                        //在职职位
                        ZZZW = tr.GetElementsByTagName("tr")[6].GetElementsByTagName("td")[3].InnerText;
                        //工作经历
                        GZJY = tr.GetElementsByTagName("tr")[7].GetElementsByTagName("td")[1].InnerText;
                        //要求月薪
                        YX = tr.GetElementsByTagName("tr")[9].GetElementsByTagName("td")[1].InnerText;
                        //工作性质
                        GZXZ = tr.GetElementsByTagName("tr")[9].GetElementsByTagName("td")[3].InnerText;
                        //求职意向
                        QZYX = tr.GetElementsByTagName("tr")[10].GetElementsByTagName("td")[1].InnerText;
                        //具体职务
                        JTZW = tr.GetElementsByTagName("tr")[10].GetElementsByTagName("td")[3].InnerText;
                        //期望工作地
                        QWGZD = tr.GetElementsByTagName("tr")[11].GetElementsByTagName("td")[1].InnerText;
                        //教育情况，语言水平，技术专长
                        QT = tr.GetElementsByTagName("tr")[13].GetElementsByTagName("td")[1].InnerText;

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

      //将数据插入数据库 
      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,年龄,性别,身高,政治面貌,民族,学历,婚烟状况,所学专业,";
                sql += "工作经验,在职单位,在职职位,工作经历,要求月薪,工作性质,求职意向,具体职务,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 { }
        }

        //返回一个HtmlElementCollection,然后进行查询内容
        private HtmlElementCollection HtmlTR_Content(string strWeb, string tj)
        {
            try
            {
                //生成HtmlDocument
                WebBrowser webb = new WebBrowser();
                webb.Navigate("about:blank");
                //window.document返回一个htmldocument对象，表示对一个html文档的操作
                //htmldocument对象是在xmldocument基础上建立的，具有xmldocument的一切方法属性
                HtmlDocument htmldoc = webb.Document.OpenNew(true);
                htmldoc.Write(strWeb);
                HtmlElementCollection htmlTR = htmldoc.GetElementsByTagName(tj);

                return htmlTR;
            }
            catch { return null; }
        }

标签：

版权申明：本站文章部分自网络，如有侵权，请联系：west999com@outlook.com
特别注意：本站所有转载文章言论不代表本站观点，本站所提供的摄影照片，插画，设计作品，如需使用，请与原作者联系，版权归原作者所有

热门词条

最新资讯

• Photoshop绘制立体风格的微笑表情
• PS文字特效教程：制作公路上个性的涂鸦
• Photoshop设计卷边效果的变形金刚电影
• PS色彩修复教程：利用色阶工具快速给红
• PS半透明物体抠图：利用通道选区工具抠
• PS海报设计技巧教程：学习制作个性的扭
• PS图片特效制作教程：学习给铁塔图片制
• 学习用photoshop把浑浊的海水照片后期
• PS古风照片教程：给古风美女打造出女侠
• PS个性人物海报制作：设计创意时尚的玻

热门关注

• 最近在折腾在线编辑，研究了下Wopi，下
• Vue+Element UI 实现视频上传
• ABP 框架代码批量生成器
• VB中使用DirectX库的简明教程
• 【半小时大话.net依赖注入】（一）理论
• 用VB6.0自制压缩与解压缩程序(一)
• 竖向照片上传后变横向
• 尝试加载 Oracle 客户端库时引发 BadIm
• 企业微信平台开发--解决access_token过
• 从零开始学习jQuery (二) 万能的选择器

热门标签