httpwebrequest 模拟登录 获取cookies 以前的代码,记录备忘!

2个类,一个基类,一个构建头信息调用类

关于如何获取到post中的内容,你之需要用http抓包工具把你与目标网站的请求信息抓下来后,打开分析下按照抓下来的包中的数

据进行构建就行了

[Csharp] view plaincopy
  1. using System;  
  2. using System.Collections.Generic;  
  3. using System.Text;  
  4. using System.Net;  
  5. using System.IO;  
  6. namespace bot  
  7. {  
  8.    public class Html  
  9.     {  
  10.         /// <summary>  
  11.         /// httpwebrequest类中的一些属性的集合  
  12.         /// </summary>  
  13.         public struct RequestPPT  
  14.         {  
  15.             private string strAccept;  
  16.             /// <summary>  
  17.             /// 获取或设置request类中的Accept属性  
  18.             /// 用以设置接受的文件类型  
  19.             /// </summary>              
  20.             public string Accept  
  21.             {  
  22.                 get  
  23.                 {  
  24.                     return strAccept;  
  25.                 }  
  26.                 set  
  27.                 {  
  28.                     strAccept = value;  
  29.                 }  
  30.             }  
  31.             private string strContentType;  
  32.             /// <summary>  
  33.             /// 获取或设置request类中的ContentType属性  
  34.             /// 用以设置请求的媒体类型  
  35.             /// </summary>            
  36.             public string ContentType  
  37.             {  
  38.                 get  
  39.                 {  
  40.                     return strContentType;  
  41.                 }  
  42.                 set  
  43.                 {  
  44.                     strContentType = value;  
  45.                 }  
  46.             }  
  47.             /// <summary>  
  48.             /// 获取或设置request类中的UserAgent属性  
  49.             /// 用以设置请求的客户端信息  
  50.             /// </summary>  
  51.             private string strUserAgent;  
  52.             public string UserAgent  
  53.             {  
  54.                 get  
  55.                 {  
  56.                     return strUserAgent;  
  57.                 }  
  58.                 set  
  59.                 {  
  60.                     strUserAgent = value;  
  61.                 }  
  62.             }  
  63.             private string strMethod;  
  64.             /// <summary>  
  65.             /// 获取或设置request类中的Method属性  
  66.             /// 可以将 Method 属性设置为任何 HTTP 1.1 协议谓词：GET、HEAD、POST、PUT、DELETE、TRACE 或 OPTIONS。  
  67.             /// 如果 ContentLength 属性被设置为 -1 以外的任何值，则必须将 Method 属性设置为上载数据的协议属性。  
  68.             /// </summary>              
  69.             public string Method  
  70.             {  
  71.                 get  
  72.                 {  
  73.                     return strMethod;  
  74.                 }  
  75.                 set  
  76.                 {  
  77.                     strMethod = value;  
  78.                 }  
  79.             }  
  80.         }  
  81.         /// <summary>  
  82.         /// 构建一个httt请求以获取目标链接的cookies,需要传入目标的登录地址和相关的post信息,返回完成登录的cookies,以及返回的html内容  
  83.         /// </summary>  
  84.         /// <param name="url">登录页面的地址  
  85.         /// <param name="post">post信息  
  86.         /// <param name="strHtml">输出的html代码  
  87.         /// <param name="rppt">请求的标头所需要的相关属性设置  
  88.         /// <returns>请求完成后的cookies</returns>  
  89.         public CookieCollection funGetCookie(string url, byte[] post, out string strHtml, RequestPPT rppt,string server)  
  90.         {  
  91.               
  92.             CookieCollection ckclReturn = new CookieCollection();  
  93.             CookieContainer cc = new CookieContainer();  
  94.             HttpWebRequest hwRequest;  
  95.             HttpWebResponse hwResponse;  
  96.             //请求cookies的格式  
  97.             //hwRequest = (HttpWebRequest)HttpWebRequest.Create(new Uri(url));  
  98.             //hwResponse = (HttpWebResponse)hwRequest.GetResponse();  
  99.             //string cookie = hwResponse.Headers.Get("Set-Cookie");  
  100.             //cookie = cookie.Split(';')[0];  
  101.             //hwRequest = null;  
  102.             //hwResponse = null;  
  103.             //构建即将发送的包头  
  104.             //cc.SetCookies(new Uri(server), cookie);             
  105.             hwRequest = (HttpWebRequest)HttpWebRequest.Create(new Uri(url));  
  106.             hwRequest.CookieContainer = cc;  
  107.             hwRequest.Accept = rppt.Accept;  
  108.             hwRequest.ContentType = rppt.ContentType;  
  109.             hwRequest.UserAgent = rppt.UserAgent;  
  110.             hwRequest.Method = rppt.Method;  
  111.             hwRequest.ContentLength = post.Length;  
  112.             //写入标头  
  113.             Stream stream;  
  114.             stream = hwRequest.GetRequestStream();  
  115.             stream.Write(post, 0, post.Length);  
  116.             stream.Close();  
  117.             //发送请求获取响应内容  
  118.             try  
  119.             {  
  120.                 hwResponse = (HttpWebResponse)hwRequest.GetResponse();  
  121.             }  
  122.             catch  
  123.             {  
  124.                 strHtml = "";  
  125.                 return ckclReturn;  
  126.             }  
  127.             stream = hwResponse.GetResponseStream();  
  128.             StreamReader sReader = new StreamReader(stream, Encoding.Default);  
  129.             strHtml = sReader.ReadToEnd();  
  130.             sReader.Close();  
  131.             stream.Close();  
  132.             //获取缓存内容  
  133.             ckclReturn = hwResponse.Cookies;  
  134.             return ckclReturn;  
  135.         }  
  136.        /// <summary>  
  137.        /// 根据已经获取的有效cookies来获取目标链接的内容  
  138.        /// </summary>  
  139.        /// <param name="strUri">目标链接的url  
  140.        /// <param name="ccl">已经获取到的有效cookies  
  141.        /// <param name="rppt">头属性的相关设置  
  142.        /// <returns>目标连接的纯文本:"txt/html"</returns>  
  143.        public string funGetHtmlByCookies(string strUri, CookieCollection ccl, RequestPPT rppt)  
  144.        {  
  145.            CookieContainer cc = new CookieContainer();  
  146.            HttpWebRequest hwRequest;  
  147.            HttpWebResponse hwResponse;        
  148.      
  149.            //构建即将发送的包头         
  150.            hwRequest = (HttpWebRequest)HttpWebRequest.Create(new Uri(strUri));  
  151.            cc.Add(ccl);  
  152.            hwRequest.CookieContainer = cc;  
  153.            hwRequest.Accept = rppt.Accept;  
  154.            hwRequest.ContentType = rppt.ContentType;  
  155.            hwRequest.UserAgent = rppt.UserAgent;  
  156.            hwRequest.Method = rppt.Method;  
  157.            hwRequest.ContentLength = 0;      
  158.       
  159.            //发送请求获取响应内容  
  160.            try  
  161.            {  
  162.                hwResponse = (HttpWebResponse)hwRequest.GetResponse();  
  163.            }  
  164.            catch  
  165.            {  
  166.                return "";  
  167.            }  
  168.   
  169.            Stream stream;  
  170.            stream = hwResponse.GetResponseStream();  
  171.            StreamReader sReader = new StreamReader(stream, Encoding.Default);  
  172.            string strHtml = sReader.ReadToEnd();  
  173.            sReader.Close();  
  174.            stream.Close();  
  175.   
  176.            //返回值            
  177.            return strHtml;  
  178.        }  
  179.        /// <summary>  
  180.        /// 根据已经获取的有效cookies来获取目标链接的内容  
  181.        /// </summary>  
  182.        /// <param name="strUri">目标链接的url  
  183.        ///<param name="post">post的byte信息  
  184.        /// <param name="ccl">已经获取到的有效cookies  
  185.        /// <param name="rppt">头属性的相关设置  
  186.        /// <returns>目标连接的纯文本:"txt/html"</returns>  
  187.        public string funGetHtmlByCookies(string strUri,byte[] post, CookieCollection ccl, RequestPPT rppt)  
  188.        {  
  189.            CookieContainer cc = new CookieContainer();  
  190.            HttpWebRequest hwRequest;  
  191.            HttpWebResponse hwResponse;  
  192.   
  193.            //构建即将发送的包头         
  194.            hwRequest = (HttpWebRequest)HttpWebRequest.Create(new Uri(strUri));  
  195.            cc.Add(ccl);  
  196.            hwRequest.CookieContainer = cc;  
  197.            hwRequest.Accept = rppt.Accept;  
  198.            hwRequest.ContentType = rppt.ContentType;  
  199.            hwRequest.UserAgent = rppt.UserAgent;  
  200.            hwRequest.Method = rppt.Method;  
  201.            hwRequest.ContentLength = post.Length;  
  202.            //写入post信息  
  203.            Stream stream;  
  204.            stream = hwRequest.GetRequestStream();  
  205.            stream.Write(post, 0, post.Length);  
  206.            stream.Close();  
  207.            //发送请求获取响应内容  
  208.            try  
  209.            {  
  210.                hwResponse = (HttpWebResponse)hwRequest.GetResponse();  
  211.            }  
  212.            catch  
  213.            {  
  214.                return"" ;  
  215.            }  
  216.              
  217.            stream = hwResponse.GetResponseStream();  
  218.            StreamReader sReader = new StreamReader(stream, Encoding.Default);  
  219.            string strHtml = sReader.ReadToEnd();  
  220.            sReader.Close();  
  221.            stream.Close();  
  222.   
  223.            //返回值            
  224.            return strHtml;  
  225.        }  
  226.     }  
  227. }  
[Csharp] view plaincopy
  1.    
[Csharp] view plaincopy
  1.    
[Csharp] view plaincopy
  1. 第二个  
[Csharp] view plaincopy
  1.    
[Csharp] view plaincopy
  1.    
[Csharp] view plaincopy
    1. <pre class="">using System;  
    2. using System.IO;  
    3. using System.Collections.Generic;  
    4. using System.Text;  
    5. using System.Net;  
    6. using System.Data;  
    7. using System.Xml;  
    8. using System.Text.RegularExpressions;  
    9. namespace bot  
    10. {  
    11.     public class SisHtml :Html  
    12.     {  
    13.         public SisHtml()  
    14.         {  
    15.    
    16.         }  
    17.   
    18.         /// <summary>  
    19.         /// 设置主机ip地址  
    20.         /// </summary>  
    21.         public string Host  
    22.         {  
    23.             get {  
    24.                 return strHost;  
    25.   
    26.             }  
    27.             set {  
    28.                 strHost = value;  
    29.             }  
    30.         }  
    31.   
    32.         private string strHost;  
    33.   
    34.         /// <summary>  
    35.         /// 获取目标登录链接的cookies  
    36.         /// </summary>  
    37.         /// <param name="url">目标的登录链接</param>  
    38.         /// <param name="dir">构造头的泛型键值对</param>  
    39.         /// <param name="strHtml">登录后返回的页面内容</param>  
    40.         /// <returns>登录后的cookies</returns>  
    41.         public CookieCollection funGetCookie(string url, Dictionary<string, string> dir, out string strHtml)  
    42.         {  
    43.             CookieCollection cc = new CookieCollection();  
    44.             RequestPPT rppt = new RequestPPT();  
    45.   
    46.             //构建post内容  
    47.             string strPost = funMakePost(dir);  
    48.             byte[] post = Encoding.Default.GetBytes(strPost);  
    49.   
    50.             //设置标头属性  
    51.             rppt.Accept = "image/gif, image/x-xbitmap, image/jpeg, image/pjpeg, application/x-shockwave-flash, application/vnd.ms-excel, application/vnd.ms-powerpoint, application/msword, */*";  
    52.             rppt.ContentType = "application/x-www-form-urlencoded";  
    53.             rppt.Method = "Post";  
    54.             rppt.UserAgent = "Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.2; SV1; .NET CLR 1.1.4322; InfoPath.1; .NET CLR 2.0.50727; .NET CLR 3.0.04506.648; .NET CLR 3.5.21022)";  
    55.             string server ="http://"+ new Uri(url).Host;  
    56.             return cc = base.funGetCookie(url, post, out strHtml, rppt, server);  
    57.         }  
    58.   
    59.         /// <summary>  
    60.         /// 根据已经获取到cookies来获取目标链接的内容  
    61.         /// </summary>  
    62.         /// <param name="strUri">目标的url</param>  
    63.         /// <param name="ccl">已经获取好的cookies</param>  
    64.         /// <returns>目标url的纯文本:"txt/html"</returns>  
    65.         public string funGetHtmlByCookies(string strUri,CookieCollection ccl )  
    66.         {  
    67.               
    68.             RequestPPT rppt = new RequestPPT();       
    69.             //设置头属性  
    70.             rppt.Accept = "txt/html";  
    71.             rppt.ContentType = "application/x-www-form-urlencoded";  
    72.             rppt.Method = "Post";  
    73.             rppt.UserAgent = "Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.2; SV1; .NET CLR 1.1.4322; InfoPath.1; .NET CLR 2.0.50727; .NET CLR 3.0.04506.648; .NET CLR 3.5.21022)";  
    74.             return base.funGetHtmlByCookies(strUri, ccl, rppt);  
    75.         }  
    76.   
    77.         
    78.         /// <summary>  
    79.         /// 投票帖子用的方法  
    80.         /// </summary>  
    81.         /// <param name="strHtml">投票帖子的htmlcode</param>  
    82.         /// <param name="ccl">有效的cookies</param>  
    83.         /// <returns>投票完成以后的htmlcode</returns>  
    84.         public string  funVote(string strHtml,CookieCollection ccl)  
    85.         {  
    86.            //判断是不是选取投票  
    87.             try  
    88.             {  
    89.                 strHtml = strHtml.Substring(strHtml.IndexOf("<form"), strHtml.LastIndexOf("</form>") - strHtml.IndexOf("<form") + 7);  
    90.             }  
    91.             catch  
    92.             {  
    93.                 return "";  
    94.             }  
    95.             string strCheck = @"name=""pollanswers[]""";  
    96.             //如果代码中包含关键信息说明没有被投票过  
    97.             if(strHtml.IndexOf(strCheck)>0)  
    98.             {  
    99.                 //获取post头的需求信息  
    100.                 string strFormHash = "77b49df4";  
    101.                 string strPollanswers;  
    102.                 strPollanswers = strHtml.Substring(strHtml.IndexOf(strCheck)+strCheck.Length, 20).Split('"')[1];  
    103.                 string strPollansubmit = "提交";  
    104.                 Dictionary<string,string>dir = new Dictionary<string,string>();  
    105.                 dir.Add("formhash",strFormHash);  
    106.                 dir.Add("pollanswers[]",strPollanswers);  
    107.                 dir.Add("pollsubmit",strPollansubmit);  
    108.                 string strPost = funMakePost(dir);  
    109.                 byte[] post = Encoding.Default.GetBytes(strPost);  
    110.                 //获取请求的路径                 
    111.                 string strUrl= "http://"+Host+"/bbs/";  
    112.                 string strActionUrl =@"method=""post""";  
    113.                 strUrl+= strHtml.Substring(strHtml.IndexOf(strActionUrl)+strActionUrl.Length,100).Split('"')[1].Replace("amp;","");  
    114.                  //构建头  
    115.                 RequestPPT rppt = new RequestPPT();  
    116.                 rppt.Accept = "txt/html";  
    117.                 rppt.ContentType = "application/x-www-form-urlencoded";  
    118.                 rppt.Method = "Post";  
    119.                 rppt.UserAgent = "Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.2; SV1; .NET CLR 1.1.4322; InfoPath.1; .NET CLR 2.0.50727; .NET CLR 3.0.04506.648; .NET CLR 3.5.21022)";  
    120.                 strHtml = base.funGetHtmlByCookies(strUrl, post, ccl, rppt);  
    121.             }  
    122.             return strHtml;  
    123.         }  
    124.   
    125.         /// <summary>  
    126.         /// 根据泛型来构建字符串用于post  
    127.         /// </summary>  
    128.         /// <param name="dir">带有键值对的泛型</param>  
    129.         /// <returns>构建完毕的字符串</returns>  
    130.         private string funMakePost(Dictionary<string,string> dir)  
    131.         {  
    132.             string strPost="";  
    133.             foreach (KeyValuePair<string, string> kvp in dir)  
    134.             {  
    135.                 strPost += kvp.Key + "=";  
    136.                 if (kvp.Value == "")  
    137.                 {  
    138.                     strPost += "''";  
    139.                 }  
    140.                 else  
    141.                 {  
    142.                     strPost += kvp.Value;  
    143.                 }  
    144.                 strPost += "&";  
    145.             }  
    146.             strPost = strPost.Substring(0, strPost.Length - 1);  
    147.             return strPost;  
    148.         }  
    149.   
    150.         /// <summary>  
    151.         /// 获取下一个列表页面的路径  
    152.         /// </summary>  
    153.         /// <param name="strHtml">当前页面的htmlcode</param>  
    154.         /// <returns>下一个列表页面的路径</returns>  
    155.         public string funGetNextUrl(string strHtml)  
    156.         {  
    157.             string strUrl = "";  
    158.             //判断是否是列表型页面  
    159.             if (strHtml.IndexOf("<form") != -1)  
    160.             {  
    161.                 return strUrl;  
    162.             }  
    163.             string strKey =@"class=""next""";  
    164.             strUrl = "http://"+Host+"/bbs/"+strHtml.Substring(strHtml.IndexOf(strKey) - 100, 100).Split('"')[1].Replace("amp;", "");  
    165.             return strUrl;  
    166.         }  
    167.   
    168.         public DataTable funGetListTable(string strHtml)  
    169.         {  
    170.             DataTable dt = new DataTable();  
    171.             DataColumn dc = new DataColumn("Url");  
    172.             dt.Columns.Add(dc);  
    173.             DataRow dr ;  
    174.             string strReg = @"viewthread.php(/S)+highlight=";  
    175.             Regex rg = new Regex(strReg);  
    176.             MatchCollection mc = rg.Matches(strHtml);  
    177.   
    178.             foreach (Match ms in mc)  
    179.             {  
    180.                 dr = dt.NewRow();  
    181.                 dr[0] = "http://" + Host + "/bbs/" + ms.ToString().Replace("amp;", "");  
    182.                 dt.Rows.Add(dr);  
    183.             }  
    184.               
    185.             return dt;  
    186.         }  
    187.     }  
    188. }  
    189. </pre> 
posted @ 2014-03-27 18:13  程序员徐坤  阅读(347)  评论(2)  编辑  收藏  举报