花嫁喜铺数据抓取

 private void saveImage()
  {
   string baseUri = "
http://www.hjxsy.com/hjxp/showimg.asp?id=";
   
   for(int _number = 1; _number < 300; _number ++)
   {
    try
    {
     WebClient client = new WebClient();
     client.DownloadFile(baseUri+_number.ToString(),@"C:\Inetpub\wwwroot\TestWebServices\hj\"+_number.ToString()+".jpg");
    }
    catch
    {}
   }
  }

  private void saveInfo()
  {
   SqlConnection con = new SqlConnection(System.Configuration.ConfigurationSettings.AppSettings["SqlConnectStringPortal"]);
   SqlCommand cmd = new SqlCommand("insert into marriage values(@id,@name,@priceold,@pricenew,@unit,@intro)",con);
   cmd.Parameters.Add("@id",SqlDbType.Int);
   cmd.Parameters.Add("@name",SqlDbType.NVarChar);
   cmd.Parameters.Add("@priceold",SqlDbType.Money);
   cmd.Parameters.Add("@pricenew",SqlDbType.Money);
   cmd.Parameters.Add("@unit",SqlDbType.NVarChar);
   cmd.Parameters.Add("@intro",SqlDbType.NVarChar);
   con.Open();
   for(int i=1;i<300;i++)
   {
    try
    {
     WebClient client = new WebClient();
     Stream stm = client.OpenRead(txtUri.Text+i.ToString());
     StreamReader sdr = new StreamReader(stm,Encoding.Default);
     string line = sdr.ReadToEnd();

     int pm1 = line.IndexOf(@"品名:")+3;
     string pm2 = line.Substring(pm1);
     int pm3 = pm2.IndexOf(@"</td>");
     string _pm = pm2.Substring(0,pm3).Trim();

     int firstYuan = pm2.IndexOf(@"¥")+1;
     string priceOne = pm2.Substring(firstYuan);
     int _firstYuan = priceOne.IndexOf(@" ");
     string _priceOne = priceOne.Substring(0,_firstYuan).Trim();

     string unit = priceOne.Substring(_firstYuan) + 1;
     int _unit = unit.IndexOf(@"<br>");
     string _unitOne = unit.Substring(0,_unit).Trim();

     int secondYuan = unit.IndexOf(@"¥")+1;
     string priceTwo = unit.Substring(secondYuan);
     int _secondYuan = priceTwo.IndexOf(@" ");
     string _priceTwo = priceTwo.Substring(0,_secondYuan).Trim();

     int intro1 = priceTwo.IndexOf(@"85%")+9;
     string intro2 = priceTwo.Substring(intro1);
     int intro3 = intro2.IndexOf(@"</td>");
     string _intro = intro2.Substring(0,intro3).Replace(@"&nbsp;"," ").Trim();

     cmd.Parameters["@id"].Value = i;
     cmd.Parameters["@name"].Value = _pm;
     cmd.Parameters["@priceold"].Value = Convert.ToDecimal(_priceOne);
     cmd.Parameters["@pricenew"].Value = Convert.ToDecimal(_priceTwo);
     cmd.Parameters["@unit"].Value = _unitOne;
     cmd.Parameters["@intro"].Value = _intro;
     cmd.ExecuteNonQuery();
     stm.Close();
    }
    catch
    {}
   }
   con.Close();
  }


上海热线 -- wojiehun.com

  private void getInfo2()
  {
   StreamReader sr = new StreamReader(@"C:\Inetpub\wwwroot\TestWebServices\meirong.txt",Encoding.Default);
   string line = sr.ReadToEnd();
   int flag1;
   int flag2;
   int flag3;
   int flag4;
   int flag5;
   int flag6;
   string str1;
   string str2;
   string str3;
   string str4;
   string str5;
   string str6;
   
   do
   {
    flag1 = line.IndexOf(@"http");
    str1 = line.Substring(flag1);
    flag2 = str1.IndexOf(@"htm")+3;
    str2 = str1.Substring(0,flag2);

    flag3 = str1.IndexOf(@"htm>");
    str3 = str1.Substring(flag3);
    flag4 = str3.IndexOf(@"</a>");
    str4 = str3.Substring(4,flag4);

    WebClient client = new WebClient();
    Stream stm = client.OpenRead(str2);
    StreamReader sdr = new StreamReader(stm,Encoding.Default);
    string _line = sdr.ReadToEnd();
    
    try
    {
     flag5 = _line.IndexOf(@"<!--enpcontent-->") +17;
     str5 = _line.Substring(flag5);
     flag6 = str5.IndexOf(@"<script") -1;
     str6 = str5.Substring(0,flag6);
     str6 = str6.Remove(str6.IndexOf(@"<TABLE"),102);

     string title = System.Web.HttpUtility.UrlEncode(str4,Encoding.GetEncoding("GB2312"));
     string contenttype = System.Web.HttpUtility.UrlEncode("html", Encoding.GetEncoding("GB2312"));
     string typeid = System.Web.HttpUtility.UrlEncode("5", Encoding.GetEncoding("GB2312"));
     string nfrom = System.Web.HttpUtility.UrlEncode("", Encoding.GetEncoding("GB2312"));
     string zznews = System.Web.HttpUtility.UrlEncode("", Encoding.GetEncoding("GB2312"));
     string about = System.Web.HttpUtility.UrlEncode("", Encoding.GetEncoding("GB2312"));
     string tjnews = System.Web.HttpUtility.UrlEncode("", Encoding.GetEncoding("GB2312"));
     string shenghe = System.Web.HttpUtility.UrlEncode("checkbox", Encoding.GetEncoding("GB2312"));
     string content = System.Web.HttpUtility.UrlEncode(str6, Encoding.GetEncoding("GB2312"));

     String postStr = @"?";
     postStr = postStr + "&title=" + title;
     postStr = postStr + "&contenttype=" + contenttype;
     postStr = postStr + "&typeid=" + typeid;
     postStr = postStr + "&nfrom=" + nfrom;
     postStr = postStr + "&zznews=" + zznews;
     postStr = postStr + "&about=" + about;
     postStr = postStr + "&tjnews=" + tjnews;
     postStr = postStr + "&shenghe=" + shenghe;
     postStr = postStr + "&content=" + content;

     WebClient command = new WebClient();
     Stream st = command.OpenRead(postStr);
     st.Close();
    }
    catch
    {}
    stm.Close();
    line = str3;
   }
   while(line.Length >93);
  }

posted @ 2004-06-30 13:41  thanks  阅读(1756)  评论(0编辑  收藏  举报