采集58信息的一些总结

  1  /// <summary>
  2     /// 封装Http类
  3     /// </summary>
  4     class HttpWebHelper
  5     {
  6         /// <summary>
  7         /// 显示验证码页面容器
  8         /// </summary>
  9         public static WebBrowser webBrowser { get; set; }
 10 
 11         /// <summary>
 12         /// 验证码需要的唯一id
 13         /// </summary>
 14         public static string uuid { get; set; }
 15 
 16         /// <summary>
 17         /// 验证码是否通过
 18         /// </summary>
 19         public static bool isPass { get; set; }
 20 
 21         /// <summary>
 22         /// 首页关键字对应的url
 23         /// </summary>
 24         public static Dictionary<string, string> list;
 25 
 26         /// <summary>
 27         /// 抓取页面前缀
 28         /// </summary>
 29         public static string prefix;
 30         /// <summary>
 31         /// 显示验证码页面
 32         /// </summary>
 33         public string codeUrl { get; set; }
 34         /// <summary>
 35         /// 抓取首页
 36         /// </summary>
 37         public string dataUrl { get; set; }
 38         /// <summary>
 39         /// 验证码提交页面
 40         /// </summary>
 41         public static string verCode { get; set; }
 42 
 43         /// <summary>
 44         /// 页面请求方式
 45         /// </summary>
 46         public string Method { get; set; }
 47         /// <summary>
 48         /// RefererHTTP 表头值
 49         /// </summary>
 50         public string Referer { get; set; }
 51         /// <summary>
 52         /// 主机
 53         /// </summary>
 54         public string Host { get; set; }
 55         /// <summary>
 56         /// cookie
 57         /// </summary>
 58         public CookieContainer cookie { get; set; }
 59         /// <summary>
 60         /// 
 61         /// </summary>
 62         public string Accept { get; set; }
 63         public string UserAgent { get; set; }
 64         public string ContentType { get; set; }
 65         public string Accept_Language { get; set; }
 66         public Encoding encoding { get; set; }
 67 
 68         public Image PictureBox { get; set; }
 69 
 70 
 71 
 72 
 73         public HttpWebHelper()
 74         {
 75             this.codeUrl = "http://support.58.com/firewall/valid/3071088800.do";
 76             //this.verCode = "http://support.58.com/firewall/valid/3071088800.do";
 77 
 78             Method = "post";
 79             Referer = "http://support.58.com/firewall/valid/3071088800.do";
 80             Host = "support.58.com";
 81             Accept_Language = "zh-CN,zh;q=0.8,en-US;q=0.5,en;q=0.3";
 82             Accept = "*/*";
 83             ContentType = "application/x-www-form-urlencoded; charset=UTF-8";
 84             UserAgent = "Mozilla/5.0 (Windows NT 6.1; WOW64; rv:37.0) Gecko/20100101 Firefox/37.0";
 85             encoding = Encoding.UTF8;
 86 
 87 
 88         }
 89 
 90         public HttpWebHelper(WebBrowser webBrowser, string uuid, string dataUrl)
 91         {
 92             //this.codeUrl = "http://support.58.com/firewall/valid/3071088800.do";
 93             //HttpWebHelper.webBrowser = webBrowser;
 94             //this.uuid = uuid;
 95             this.dataUrl = dataUrl;
 96         }
 97 
 98         /// <summary>
 99         /// 验证 验证码，验证码和页面生成的一个id值同时post到服务器
100         /// </summary>
101         /// <param name="code">验证码</param>
102         public void postVerCode(string code, string uuid)
103         {
104             try
105             {
106                 //HtmlElement d = webBrowser.Document.GetElementById("uuid");
107 
108                 //获取页面uid。
109                 /*
110                  * 验证方式：验证码和页面生成的一个id值
111                  */
112                 //string y = webBrowser.Document.GetElementById("uuid").GetAttribute("value");
113 
114                 // string postUrl = "http://support.58.com/firewall/valid/3071088800.do";
115                 HttpWebHelper h = new HttpWebHelper();
116 
117                 HttpWebRequest request = (HttpWebRequest)WebRequest.Create(verCode);
118                 request.Method = Method;
119                 request.Referer = Referer;
120                 request.Headers.Add("X-Requested-With", "XMLHttpRequest");
121                 request.Host = Host;
122                 CookieContainer cookie = new CookieContainer();
123                 request.CookieContainer = cookie;
124                 request.Accept = Accept;
125                 request.ContentType = ContentType;
126                 request.Headers.Add("Accept-Language", Accept_Language);
127                 request.UserAgent = UserAgent;
128                 string parameter = string.Format("inputcode={0}&namespace=infodetailweb&uuid={1}", HttpUtility.UrlEncode(code), uuid);
129 
130                 byte[] buffer = Encoding.Default.GetBytes(parameter);
131 
132                 string result = string.Empty;
133                 Stream reqStr = request.GetRequestStream();
134                 reqStr.Write(buffer, 0, buffer.Length);
135                 using (HttpWebResponse response1 = (HttpWebResponse)request.GetResponse())
136                 {
137 
138                     using (StreamReader reader = new StreamReader(response1.GetResponseStream(), encoding))
139                     {
140                         result = reader.ReadToEnd().Trim();
141                     }
142                 }
143                 HttpWebHelper.isPass = (result == "1" ? true : false);
144             }
145             catch (Exception ex)
146             {
147                 MessageBox.Show(ex.StackTrace);
148             }
149         }
150 
151         /// <summary>
152         /// WebClient简单下载页面
153         /// </summary>
154         /// <param name="url">下载html的页面</param>
155         /// <returns></returns>
156         public string webClient(string url)
157         {
158             string html = string.Empty;
159             try
160             {
161                 //WebClient client = new WebClient();
162                 //client.Encoding = encoding;
163                 //string html = client.DownloadString(url);
164                 HttpWebRequest request = (HttpWebRequest)WebRequest.Create(url);
165 
166                 request.Method = "get";
167                 //request.Timeout = 300;
168                 using (HttpWebResponse response1 = (HttpWebResponse)request.GetResponse())
169                 {
170                     using (StreamReader reader = new StreamReader(response1.GetResponseStream(), encoding))
171                     {
172                         html = reader.ReadToEnd().Trim();
173                     }
174                 }
175             }
176             catch (Exception ex)
177             {
178                 MessageBox.Show(ex.StackTrace);
179             }
180             return html;
181         }
182     }

 /// <summary>
        /// 
        /// </summary>
        /// <param name="url">首页抓取</param>
        /// <param name="keyword">首页关键字</param>
        private void ProcessDownload(string url, string keyword)
        {
            this.Invoke(
                         new Action(() => { richTextBoxInfo.AppendText(url + "开始下载中．．．．．．\n"); })
                        );

            //抓取关键字对应的url
            WebClient client = new WebClient();
            string html = client.DownloadString(url);
            IHtmlDocument document = new JumonyParser().Parse(html);
            IEnumerable<IHtmlElement> result = document.Find("a").Where(t => t.InnerText().Contains(keyword));

            Dictionary<string, string> dir = new Dictionary<string, string>();
            foreach (var item in result)
            {
                var href = item.Attribute("href").Value();
                var text = item.InnerText();
                if (!dir.ContainsKey(href)) dir.Add(text, href);
            }

            //左边菜单栏赋值
            this.Invoke(new Action(() =>
            {
                foreach (var item in dir)
                {
                    listBoxMenu.Items.Add(item.Key);
                }
            }));

            //共享数据
            HttpWebHelper.list = dir;
            HttpWebHelper.prefix = url;

            //开启多线程下载。

            //foreach (var item in dir)
            //{
            //    Thread thread = new Thread(() => { DownloadHtml(item.Key); });
            //    thread.Name = item.Key; //线程取名字
            //}

            try
            {
                foreach (var item in dir)
                {
                    //ThreadPool.QueueUserWorkItem(new WaitCallback(DownloadHtml), item.Key);

                    Thread thread = new Thread(ThreadDownload);
                    thread.Name = item.Key;
                    thread.Start(item.Key + "," + item.Value);

                }
            }
            catch (Exception ex)
            {

                MessageBox.Show(ex.StackTrace);
            }
        }

try
            {
                foreach (var item in dir)
                {
                    //ThreadPool.QueueUserWorkItem(new WaitCallback(DownloadHtml), item.Key);

                    Thread thread = new Thread(ThreadDownload);
                    thread.Name = item.Key;
                    thread.Start(item.Key + "," + item.Value);

                }
            }
            catch (Exception ex)
            {

                MessageBox.Show(ex.StackTrace);
            }

  1 /// <summary>
  2         /// 
  3         /// </summary>
  4         /// <param name="title">当前抓取的关键字</param>
  5         private void ThreadDownload(object obj)
  6         {
  7             //因为58有采集频率限制。所以改成同步
  8             Monitor.Enter(this);
  9 
 10             string[] ob = obj.ToString().Split(‘,‘);
 11             this.Invoke(
 12                          new Action(() => { richTextBoxInfo.AppendText(string.Format("正在抓取：{0}\n", ob[0])); })
 13                      );
 14             Dictionary<string, string> list = HttpWebHelper.list;
 15             string prefix = HttpWebHelper.prefix;
 16 
 17 
 18             HttpWebHelper client = new HttpWebHelper();
 19             client.encoding = Encoding.UTF8;
 20             //client.webClient(prefix);
 21 
 22 
 23             DataTable dt = new DataTable();
 24             dt.Columns.Add("公司名字", typeof(string));
 25             dt.Columns.Add("联系人", typeof(string));
 26             dt.Columns.Add("联系电话", typeof(string));
 27 
 28             //遍历每个信息对象的url 如：家庭维修==》 www.baidu.com
 29             //foreach (var item in list)
 30             //{
 31             //获取列表
 32             string fullurl = string.Format("{0}{1}", prefix, ob[1]);
 33             string html = client.webClient(fullurl);
 34 
 35             IHtmlDocument document = new JumonyParser().Parse(html);
 36             IEnumerable<IHtmlElement> result = document.Find("table[id=jingzhun]");
 37 
 38             var items = result.Find("tr");
 39 
 40             foreach (var o in items)
 41             {
 42                 if (o.Find("a").Count() > 0)
 43                 {
 44                     /*
 45                      * 执行该url的时候。服务器判断了请求的频繁度，需要输入验证码。
 46                      * 输入验证码成功后。会执行该url  即下面的referer
 47                      */
 48                     //列表中找到a标签转到详细页面
 49                     string referer = o.FindFirst("a").Attribute("href").Value();
 50 
 51 
 52                     //http://support.58.com/firewall/valid/1032910901.do?namespace=infodetailweb&url=http://sz.58.com/qichejx/19720429696131x.shtml
 53 
 54                     //等待5秒，防止抓取频率过高 时间根据当前的环境来定
 55                     Thread.Sleep(5000);
 56 
 57 
 58 
 59                     string n = Thread.CurrentThread.Name;
 60                     string i = Thread.CurrentThread.ManagedThreadId.ToString();
 61 
 62                     //抓取详细页面。这里如果过于频繁，会跳到输入验证码页面
 63                     string sonHtml = client.webClient(referer);
 64 
 65                     //Monitor.Enter(this);
 66 
 67                     if (sonHtml.Contains("验证码"))
 68                     {
 69 
 70                         HttpWebRequest request = (HttpWebRequest)WebRequest.Create(referer);
 71                         request.Method = "get";
 72                         string responseUrl = string.Empty;
 73                         string rediect = string.Empty;
 74                         using (HttpWebResponse response1 = (HttpWebResponse)request.GetResponse())
 75                         {
 76                             //"http://support.58.com/firewall/valid/1903444021.do?namespace=infodetailweb&url=http://sz.58.com/shoujiweixiu/21147587557513x.shtml"
 77                             responseUrl = response1.ResponseUri.ToString();
 78 
 79                             //获取绝对路径 "/firewall/valid/1032910901.do"
 80                             string absolutePath = response1.ResponseUri.AbsolutePath;
 81 
 82                             //ResponseUri.Authority  "support.58.com"
 83                             HttpWebHelper.verCode = "http://" + response1.ResponseUri.Authority + absolutePath;
 84 
 85                             //获取?后面的字符串
 86                             string query = response1.ResponseUri.Query;
 87 
 88                             //验证码成功后，重定向的url
 89                             rediect = query.Substring(query.LastIndexOf("=") + 1);
 90                         }
 91                         //response1.ResponseUri.GetComponents(UriComponents.Query, UriFormat.UriEscaped);
 92                         //HttpWebHelper http = new HttpWebHelper();
 93                         //HttpWebHelper.webBrowser = new WebBrowser();
 94                         //HttpWebHelper.webBrowser.Url = new Uri(http.codeUrl);
 95 
 96                         //http.webBrowser.Navigate(http.codeUrl);
 97                         //HttpWebHelper.webBrowser.DocumentCompleted += new WebBrowserDocumentCompletedEventHandler(webBrowser_DocumentCompleted);
 98                         //HttpWebHelper.webBrowser.NewWindow += new CancelEventHandler(webBrowser_NewWindow);
 99                         //http://blog.csdn.net/jinjazz/article/details/1916883
100                         //while (waitHandle.WaitOne(10, false) == false) { Application.DoEvents(); }
101 
102                         //Thread thread = new Thread(() =>
103                         //{
104                         //    showCode code = new showCode();
105                         //    code.codeHandler = new HttpWebHelper().postVerCode;
106                         //    //code.p = h.PictureBox;
107                         //    if (code.ShowDialog() == DialogResult.OK)
108                         //    {
109                         //        code.Hide();
110                         //    }
111                         //});
112 
113                         this.Invoke(new Action(() =>
114                         {
115 
116                             showCode code = new showCode();
117                             code.codeHandler = new HttpWebHelper().postVerCode;
118                             code.showCodeUrl = responseUrl;
119                             //code.p = h.PictureBox;
120                             //this.dia
121                             if (code.ShowDialog() == DialogResult.OK)
122                             {
123                                 code.Hide();
124                                 if (HttpWebHelper.isPass)
125                                 {
126                                     sonHtml = client.webClient(rediect);
127 
128                                     getTable(sonHtml, ref dt);
129                                 }
130                             }
131                             //waitHandle.Set();
132 
133                             //waitHandle.WaitOne();
134                         }));
135                         //waitHandle.WaitOne();
136                     }
137                     else
138                         getTable(sonHtml, ref dt);
139 
140                     //获取当前线程
141                     Thread th = Thread.CurrentThread;
142                     string name = th.Name;
143 
144                     this.Invoke(new Action(() =>
145                     {
146                         //MessageBox.Show(name.ToString());
147 
148 
149                         //创建tab选项卡，如果不存在
150                         if (!tabControlWarp.TabPages.ContainsKey(name))
151                             tabControlWarp.TabPages.Add(name, name);
152 
153                         //动态创建选项卡中显示的数据，和一些属性设置
154                         DataGridView view = new DataGridView();
155                         view.AllowUserToAddRows = false;
156                         view.AllowUserToDeleteRows = false;
157                         view.AllowUserToResizeColumns = false;
158                         view.AllowUserToResizeRows = false;
159                         view.AutoSizeColumnsMode = DataGridViewAutoSizeColumnsMode.Fill;
160                         view.ColumnHeadersHeightSizeMode = DataGridViewColumnHeadersHeightSizeMode.AutoSize;
161                         view.MultiSelect = false;
162                         view.ReadOnly = true;
163                         view.RowHeadersVisible = false;
164                         view.BackgroundColor = Color.White;
165                         view.ScrollBars = ScrollBars.Vertical;
166                         view.SelectionMode = DataGridViewSelectionMode.FullRowSelect;
167                         view.Dock = DockStyle.Fill;
168                         view.DataSource = dt;
169                         //把DataGridView添加到当前选项卡
170                         tabControlWarp.TabPages[name].Controls.Add(view);
171 
172                         //刷新窗体，否则DataGridView数据没有变化
173                         this.Refresh();
174                     }));
175                 }
176             }
177             //当前线程执行完毕，把当前的数据导出为excel
178             ExcelRender.ExcelRender.RenderToExcel(dt, ob[0] + ".xls");
179             Monitor.Exit(this);
180         }

 1                         HttpWebRequest request = (HttpWebRequest)WebRequest.Create(referer);
 2                         request.Method = "get";
 3                         string responseUrl = string.Empty;
 4                         string rediect = string.Empty;
 5                         using (HttpWebResponse response1 = (HttpWebResponse)request.GetResponse())
 6                         {
 7                             //"http://support.58.com/firewall/valid/1903444021.do?namespace=infodetailweb&url=http://sz.58.com/shoujiweixiu/21147587557513x.shtml"
 8                             responseUrl = response1.ResponseUri.ToString();
 9 
10                             //获取绝对路径 "/firewall/valid/1032910901.do"
11                             string absolutePath = response1.ResponseUri.AbsolutePath;
12 
13                             //ResponseUri.Authority  "support.58.com"
14                             HttpWebHelper.verCode = "http://" + response1.ResponseUri.Authority + absolutePath;
15 
16                             //获取?后面的字符串
17                             string query = response1.ResponseUri.Query;
18 
19                             //验证码成功后，重定向的url
20                             rediect = query.Substring(query.LastIndexOf("=") + 1);
21                         }

 1 /// <summary>
 2         /// 返回指定WebBrowser中图片<IMG></IMG>中的图内容
 3         /// </summary>
 4         /// <param name="WebCtl">WebBrowser控件</param>
 5         /// <param name="ImgeTag">IMG元素</param>
 6         /// <returns>IMG对象</returns>
 7         private Image GetWebImage(WebBrowser WebCtl, HtmlElement ImgeTag)
 8         {
 9 
10             /*
11              * 这种方法有时候会因为剪切板没有头像而报异常，
12              * 初步判断是页面(我这里是js对图片赋值)图片没有加载完成，而没获取到图片
13              * System.Threading.Thread.Sleep(8000);测试通过。但每次时间是不确定的。
14              */
15 
16             HTMLDocument doc = (HTMLDocument)WebCtl.Document.DomDocument;
17             HTMLBody body = (HTMLBody)doc.body;
18             IHTMLControlRange rang = (IHTMLControlRange)body.createControlRange();
19             IHTMLControlElement Img = (IHTMLControlElement)ImgeTag.DomElement; //图片地址
20             Image oldImage = Clipboard.GetImage();
21             rang.add(Img);
22             rang.execCommand("Copy", false, null);  //拷贝到内存
23             Image numImage = Clipboard.GetImage(); //如果为null则保存
24 
25             //判断剪切板是否有图片 
26             //https://msdn.microsoft.com/zh-cn/library/system.windows.forms.clipboard.getimage.aspx
27             if (Clipboard.ContainsImage())
28             { }
29 
30 
31             try
32             {
33                 Clipboard.SetImage(oldImage);
34             }
35             catch (Exception ex)
36             {
37                 MessageBox.Show(ex.Message);
38             }
39             return numImage;
40         }

1            //找到图片
2             HtmlElement ImgeTag = webBrowser1.Document.GetElementById("imgCode");
3             
4             Image numPic = GetWebImage(webBrowser1, ImgeTag); // 得到验证码图片
5             pictureBox1.Image = numPic; //图片赋值

 AutoResetEvent waitHandle = new AutoResetEvent(false);
 while (waitHandle.WaitOne(10, false) == false) { Application.DoEvents(); }

 WebBrowser we = new WebBrowser();
            we.Url = new Uri("http://support.58.com/firewall/valid/3071088800.do");
            we.DocumentCompleted += new WebBrowserDocumentCompletedEventHandler(we_DocumentCompleted);

 1 void we_DocumentCompleted(object sender, WebBrowserDocumentCompletedEventArgs e)
 2         {
 3      
 4             //HtmlElement d = webBrowser1.Document.GetElementById("uuid");
 5 
 6             //string y = webBrowser1.Document.GetElementById("uuid").GetAttribute("value");
 7 
 8 
 9 
10             //var wb = new WebBrowser();
11 
12             HtmlElementCollection docs = we.Document.All;
13             foreach (HtmlElement item in docs)
14             {
15                 string ii = item.Id;
16 
17                 if (item.Id == "uuid")
18                 {
19                     string c = item.GetAttribute("value");
20                 }
21                 else if (item.Id == "imgCode")
22                 {
23                     string c = item.GetAttribute("src");
24                     item.Style = "position: absolute; z-index: 9999; top: 0px; left: 0px";
25 
26                     //抓图
27                     var b = new Bitmap(item.ClientRectangle.Width, item.ClientRectangle.Height);
28                     we.DrawToBitmap(b, new Rectangle(new Point(), item.ClientRectangle.Size));
29                     pictureBox1.Image = b;
30                     break;
31 
32                 }
33             }
34         }

 1  public partial class showCode : Form
 2     {
 3         public Image p { get; set; }
 4         public string showCodeUrl { get; set; } //显示验证码页面
 5         public delegate void delegateCode(string code, string uuid);
 6         public delegateCode codeHandler;
 7 
 8 
 9         public showCode()
10         {
11             InitializeComponent();
12             //InitializeEvents();
13         }
14         /// <summary>
15         /// 初始化
16         /// </summary>
17         //private void InitializeEvents()
18         //{
19         //    this.webBrowser.DocumentCompleted += new WebBrowserDocumentCompletedEventHandler(webBrowser_DocumentCompleted);
20         //}
21 
22         void webBrowser_DocumentCompleted(object sender, WebBrowserDocumentCompletedEventArgs e)
23         {
24             WebBrowser bro = (WebBrowser)sender;
25 
26             HtmlElement img = bro.Document.GetElementById("imgCode");
27 
28             bro.Document.GetElementById("uuid").GetAttribute("value");
29 
30             img.Style = "position: absolute; z-index: 9999; top: 0px; left: 0px"; //使其显示在最右上角
31             img.SetAttribute("onclick", "javascript:void(0)"); //取消单击图片刷新验证码操作
32         }
33         private void btnOk_Click(object sender, EventArgs e)
34         {
35             string code = textCode.Text;
36             if (string.IsNullOrEmpty(code))
37             {
38                 MessageBox.Show("请输入验证码", "验证码", MessageBoxButtons.OK, MessageBoxIcon.Information);
39                 textCode.Focus();
40                 return;
41             }
42             if (codeHandler != null)
43             {
44                 string uuid = webBrowser.Document.GetElementById("uuid").GetAttribute("value");
45 
46                 this.DialogResult = DialogResult.OK;
47                 codeHandler(code, uuid);
48             }
49         }
50 
51         private void showCode_Load(object sender, EventArgs e)
52         {
53             //pictureBoxCode.Image = p;
54             webBrowser.Url = new Uri(showCodeUrl);
55             this.webBrowser.DocumentCompleted += new WebBrowserDocumentCompletedEventHandler(webBrowser_DocumentCompleted);
56         }
57     }

 1   private DataTable getTable(string document, ref DataTable dt)
 2         {
 3             try
 4             {
 5                 //if (IsDisposed) return null;
 6                 //this.Invoke(
 7                 //           new Action(() => { richTextBoxInfo.AppendText("正在下载\n"); })
 8                 //       );
 9 
10                 IHtmlDocument hd = new JumonyParser().Parse(document);
11                 //string company = hd.FindFirst("div[class=su_tit]").InnerText();
12 
13                 string company = "未知";
14                 string phone = "未知";
15                 string linkman = "未知";
16 
17                 //判断是个人还是企业
18                 var su = hd.Find("ul[class=suUl]");
19 
20                 //顶部html包含联系人。电话
21                 IHtmlDocument top = new JumonyParser().Parse(hd.FindFirst("ul[class=suUl]").InnerHtml());
22 
23                 if (su.Count() > 0)
24                 {
25                     if (top.Find("div[class=su_tit]").Count() > 0)
26                     {
27                         string txt = top.FindFirst("div[class=su_tit]").InnerText();
28                         if (txt.Contains("公司名称"))
29                         {
30                             if (top.Find("div[class=su_con]").Count() > 0)
31                                 //company = top.FindFirst("div[class=su_con]").FindFirst("a").InnerText();
32                                 company = top.FindFirst("div[class=su_con]").InnerText().Split(new char[] { ‘ ‘ }, StringSplitOptions.RemoveEmptyEntries)[0];
33                             if (top.Find("li:nth-child(1)").Count() > 0)
34                                 linkman = top.FindFirst("li:nth-child(2)").FindFirst("div[class=su_con]").FindFirst("a").InnerText();
35                             if (top.Find("span[class=l_phone]").Count() > 0)
36                                 phone = top.FindFirst("span[class=l_phone]").InnerText();
37                         }
38                         else if (txt.Contains("联系人"))
39                         {
40                             if (top.Find("li:nth-child(1)").Count() > 0)
41                                 linkman = top.FindFirst("li:nth-child(1)").FindFirst("div[class=su_con]").InnerText();
42                             if (top.Find("li:nth-child(2)").Count() > 0)
43                                 phone = top.FindFirst("li:nth-child(2)").FindFirst("span[id=t_phone]").InnerText();
44                         }
45                     }
46                 }
47 
48                 DataRow row = dt.NewRow();
49                 row["公司名字"] = company;
50                 row["联系电话"] = phone;
51                 row["联系人"] = linkman;
52 
53                 dt.Rows.Add(row);
54 
55 
56                 return dt;
57             }
58             catch (Exception)
59             {
60 
61                 return null;
62             }
63         }

 /// <summary>
        /// 开始抓取
        /// </summary>
        /// <param name="sender"></param>
        /// <param name="e"></param>
        /// 
        void btnStart_Click(object sender, EventArgs e)
        {
            btnStart.Enabled = false;

            //Dictionary<string, string> result = new Dictionary<string, string>();
            //string url = "http://sz.58.com/";
            //string keyword = "维修";

            string url = textBoxUrl.Text;
            string keyword = textBoxKeyword.Text;

            if (string.IsNullOrEmpty(url))
            {
                MessageBox.Show("请输入要抓取的网址", "网址", MessageBoxButtons.OK, MessageBoxIcon.Information);
                textBoxUrl.Focus();
                return;
            }
            else if (string.IsNullOrEmpty(keyword))
            {
                MessageBox.Show("请输入要抓取的关键字", "关键字", MessageBoxButtons.OK, MessageBoxIcon.Information);
                textBoxKeyword.Focus();
                return;
            }

            //string prefix = "http://sz.58.com";

            // 声明一个异步委托去处理下载操作
            Action downloadAction = new Action(() =>
            {
                ProcessDownload(url, keyword);
            });

            //Action<string, string> an = new Action<string, string>(ProcessDownload);

            //声明一个下载完成后的回调函数
            AsyncCallback callback = new AsyncCallback((asyncResult) =>
            {
                this.Invoke(
                         new Action(() => { richTextBoxInfo.AppendText("首页关键字匹配完成，显示在左侧列表中.....\n"); })
                     );
            });
            downloadAction.BeginInvoke(callback, null);
        }

 /// <summary>
        /// 窗体关闭提醒
        /// </summary>
        /// <param name="sender"></param>
        /// <param name="e"></param>
        void Main_FormClosing(object sender, FormClosingEventArgs e)
        {
            if (MessageBox.Show("是否退出当前程序", "关闭", MessageBoxButtons.YesNo, MessageBoxIcon.Question) == DialogResult.No) e.Cancel = true;
            else Environment.Exit(0); //强制退出所以线程
        }

        /// <summary>
        /// 单击左边菜单栏
        /// </summary>
        /// <param name="sender"></param>
        /// <param name="e"></param>
        void listBoxMenu_MouseClick(object sender, MouseEventArgs e)
        {
            string txt = listBoxMenu.Text;
            if (tabControlWarp.TabPages.ContainsKey(txt) && !string.IsNullOrEmpty(txt))
            {
                //tabControlWarp.TabPages.Add(txt, txt); //创建选项卡
                tabControlWarp.SelectedTab = tabControlWarp.TabPages[txt];//并且选中
            }
            //else tabControlWarp.SelectedTab = tabControlWarp.TabPages[txt];
        }