抓去网站上的信息插入到数据库代码,CSharp代码库,德仔网

代码语言
代码分类
【CSharp】抓去网站上的信息插入到数据库
作者:酷&酷 / 发布于2011/1/3/ 508
<div>private void button1_Click(object sender, EventArgs e)
 {
 dt.Rows.Clear();
 //要抓取的URL地址
 string Url = "<a href="http://list.mp3.baidu.com/topso/mp3topsong.html?id=1#top2">http://list.mp3.baidu.com/topso/mp3topsong.html?id=1#top2</a>";</div>
<div> //得到指定Url的源码
 string strWebContent = GetWebContent(Url);</div>
<div> // richTextBox1.Text = strWebContent;
 //取出和数据有关的那段源码
 int iBodyStart = strWebContent.IndexOf("<body", 0);
 int iStart = strWebContent.IndexOf("歌曲TOP500", iBodyStart);
 int iTableStart = strWebContent.IndexOf("<table", iStart);
 int iTableEnd = strWebContent.IndexOf("</table>", iTableStart);
 string strWeb = strWebContent.Substring(iTableStart, iTableEnd - iTableStart + 8);</div>
<div> //生成HtmlDocument
 WebBrowser webb = new WebBrowser();
 webb.Navigate("about:blank");
 HtmlDocument htmldoc = webb.Document.OpenNew(true);
 htmldoc.Write(strWeb);
 HtmlElementCollection htmlTR = htmldoc.GetElementsByTagName("TR");</div>
<div> // DataRow row = null;
 foreach (HtmlElement tr in htmlTR)
 {
 string strID = tr.GetElementsByTagName("TD")[0].InnerText;
 string[] info = tr.GetElementsByTagName("TD")[1].InnerText.Split('(');
 string strName = "";
 string strSinger = "";
 if (info.Length != 2)
 {
 strName = info[0];
 strSinger = "未知";
 }
 else
 {
 strName = info[0];
 strSinger = info[1].Replace(")", "");
 }
 strID = strID.Replace(".", "");
 DataRow row = dt.NewRow();
 row["ID"] = strID;
 row["voide"] = strName;
 row["name"] = strSinger;
 dt.Rows.Add(row);</div>
<div>
 string strID1 = tr.GetElementsByTagName("TD")[2].InnerText;
 string[] info1 = tr.GetElementsByTagName("TD")[3].InnerText.Split('(');
 string strName1 = "";
 string strSinger1 = "";
 if (info1.Length != 2)
 {
 strName1 = info1[0];
 strSinger1 = "未知";
 }
 else
 {
 strName1 = info1[0];
 strSinger1 = info1[1].Replace(")", "");
 }
 strID1 = strID1.Replace(".", "");
 DataRow row1 = dt.NewRow();
 row1["ID"] = strID1;
 row1["voide"] = strName1;
 row1["name"] = strSinger1;
 dt.Rows.Add(row1);</div>
<div></div>
<div> //string strID2 = tr.GetElementsByTagName("TD")[3].InnerText;</div>
<div> //string[] info2 = tr.GetElementsByTagName("TD")[4].InnerText.Split('(');
 //string strName2 = "";
 //string strSinger2 = "";
 //if (info1.Length != 2)
 //{
 // strName2 = info1[0];
 // strSinger2 = "未知";
 //}
 //else
 //{
 // strName2 = info1[0];
 // strSinger2 = info1[1].Replace(")", "");
 //}
 //strID1 = strID2.Replace(".", "");
 //DataRow row2 = dt.NewRow();
 //row2["ID"] = strID2;
 //row2["voide"] = strName2;
 //row2["name"] = strSinger2;
 //dt.Rows.Add(row2);</div>
<div> }
 ////插入数据库
 //// InsertData(dt);</div>
<div> dataGridView1.DataSource = dt.DefaultView;</div>
<div></div>
<div> }
private string GetWebContent(string Url)
 {
 string strResult = "";
 try
 {</div>
<div> //创建访问目标
 HttpWebRequest request = (HttpWebRequest)WebRequest.Create(Url);
 //声明一个HttpWebRequest请求
 request.Timeout = 30000;
 //设置连接超时时间
 request.Headers.Set("Pragma", "no-cache");</div>
<div> //得到回应
 HttpWebResponse response = (HttpWebResponse)request.GetResponse();</div>
<div> //得到数据流
 Stream streamReceive = response.GetResponseStream();</div>
<div> //对获取到的数据流进行编码解析，让我们可以进行正常读取
 Encoding encoding = Encoding.GetEncoding("GB2312");
 StreamReader streamReader = new StreamReader(streamReceive, encoding);</div>
<div> //读取出数据流中的信息
 strResult = streamReader.ReadToEnd();</div>
<div> //关闭流</div>
<div> streamReader.Close();</div>
<div> //关闭网络响应流</div>
<div> response.Close();
 }
 catch
 {
 MessageBox.Show("出错");
 }
 return strResult;
 }
 static DataTable dt = new DataTable();</div>
<div> private void Form1_Load(object sender, EventArgs e)
 {
 DataColumn ID = new DataColumn();
 DataColumn voideName = new DataColumn();
 DataColumn name = new DataColumn();
 ID.ColumnName = "ID";
 voideName.ColumnName = "voide";
 name.ColumnName = "name";
 dt.Columns.Add(ID);
 dt.Columns.Add(voideName);
 dt.Columns.Add(name);
 }
<div></div>
试试其它关键字
　抓取　
同语言下
可能有用的
酷&酷贡献的其它代码(3)