/// <summary>
/// 根据相传入的数据，得到相应页面数据
/// </summary>
/// <param name="item">参数类对象</param>
/// <returns>返回HttpResult类型</returns>
public HttpResult GetHtml(HttpItem item)
{
//返回参数
HttpResult result = new HttpResult();
try
{
//准备参数
SetRequest(item);
}
catch (Exception ex)
{
//配置参数时出错
return new HttpResult() { Cookie = string.Empty, Header = null, Html = ex.Message, StatusDescription = "配置参数时出错：" + ex.Message };
}
try
{
//请求数据
using (response = (HttpWebResponse)request.GetResponse())
{
GetData(item, result);
}
}
catch (WebException ex)
{
using (response = (HttpWebResponse)ex.Response)
{
GetData(item, result);
}
}
catch (Exception ex)
{
result.Html = ex.Message;
}
if (item.IsToLower) result.Html = result.Html.ToLower();
return result;
}
/// <summary>
/// 设置代理
/// </summary>
/// <param name="item">参数对象</param>
private void SetProxy(HttpItem item)
{
bool isIeProxy = false;
if (!string.IsNullOrEmpty(item.ProxyIp))
{
isIeProxy = item.ProxyIp.ToLower().Contains("ieproxy");
}
if (!string.IsNullOrEmpty(item.ProxyIp) && !isIeProxy)
{
//设置代理服务器
if (item.ProxyIp.Contains(":"))
{
string[] plist = item.ProxyIp.Split(':');
WebProxy myProxy = new WebProxy(plist[0].Trim(), Convert.ToInt32(plist[1].Trim()));
//建议连接
myProxy.Credentials = new NetworkCredential(item.ProxyUserName, item.ProxyPwd);
//给当前请求对象
request.Proxy = myProxy;
}
else
{
WebProxy myProxy = new WebProxy(item.ProxyIp, false);
//建议连接
myProxy.Credentials = new NetworkCredential(item.ProxyUserName, item.ProxyPwd);
//给当前请求对象
request.Proxy = myProxy;
}
}
else if (isIeProxy)
{
//设置为IE代理
}
else
{
request.Proxy = item.WebProxy;
}
}
/// <summary>
/// 为请求准备参数
/// </summary>
///<param name="item">参数列表</param>
private void SetRequest(HttpItem item)
{
// 验证证书
SetCer(item);
//设置Header参数
if (item.Header != null && item.Header.Count > 0) foreach (string key in item.Header.AllKeys)
{
request.Headers.Add(key, item.Header[key]);
}
// 设置代理
SetProxy(item);
if (item.ProtocolVersion != null) request.ProtocolVersion = item.ProtocolVersion;
request.ServicePoint.Expect100Continue = item.Expect100Continue;
//请求方式Get或者Post
request.Method = item.Method;
request.Timeout = item.Timeout;
request.KeepAlive = item.KeepAlive;
request.ReadWriteTimeout = item.ReadWriteTimeout;
if (item.IfModifiedSince != null) request.IfModifiedSince = Convert.ToDateTime(item.IfModifiedSince);
//Accept
request.Accept = item.Accept;
//ContentType返回类型
request.ContentType = item.ContentType;
//UserAgent客户端的访问类型，包括浏览器版本和操作系统信息
request.UserAgent = item.UserAgent;
// 编码
encoding = item.Encoding;
//设置安全凭证
request.Credentials = item.ICredentials;
//设置Cookie
SetCookie(item);
//来源地址
request.Referer = item.Referer;
//是否执行跳转功能
request.AllowAutoRedirect = item.Allowautoredirect;
if (item.MaximumAutomaticRedirections > 0)
{
request.MaximumAutomaticRedirections = item.MaximumAutomaticRedirections;
}
//设置Post数据
SetPostData(item);
//设置最大连接
if (item.Connectionlimit > 0) request.ServicePoint.ConnectionLimit = item.Connectionlimit;
}
/// <summary>
/// 设置编码
/// </summary>
/// <param name="item">HttpItem</param>
/// <param name="result">HttpResult</param>
/// <param name="ResponseByte">byte[]</param>
private void SetEncoding(HttpItem item, HttpResult result, byte[] ResponseByte)
{
//是否返回Byte类型数据
if (item.ResultType == ResultType.Byte) result.ResultByte = ResponseByte;
//从这里开始我们要无视编码了
if (encoding == null)
{
Match meta = Regex.Match(Encoding.Default.GetString(ResponseByte), "<meta[^<]*charset=([^<]*)[\"']", RegexOptions.IgnoreCase);
string c = string.Empty;
if (meta != null && meta.Groups.Count > 0)
{
c = meta.Groups[1].Value.ToLower().Trim();
}
if (c.Length > 2)
{
try
{
encoding = Encoding.GetEncoding(c.Replace("\"", string.Empty).Replace("'", "").Replace(";", "").Replace("iso-8859-1", "gbk").Trim());
}
catch
{
if (string.IsNullOrEmpty(response.CharacterSet))
{
encoding = Encoding.UTF8;
}
else
{
encoding = Encoding.GetEncoding(response.CharacterSet);
}
}
}
else
{
if (string.IsNullOrEmpty(response.CharacterSet))
{
encoding = Encoding.UTF8;
}
else
{
encoding = Encoding.GetEncoding(response.CharacterSet);
}
}
}
}
/// <summary>
/// 设置Post数据
/// </summary>
/// <param name="item">Http参数</param>
private void SetPostData(HttpItem item)
{
//验证在得到结果时是否有传入数据
if (request.Method.Trim().ToLower().Contains("post"))
{
if (item.PostEncoding != null)
{
postencoding = item.PostEncoding;
}
byte[] buffer = null;
//写入Byte类型
if (item.PostDataType == PostDataType.Byte && item.PostdataByte != null && item.PostdataByte.Length > 0)
{
//验证在得到结果时是否有传入数据
buffer = item.PostdataByte;
}//写入文件
else if (item.PostDataType == PostDataType.FilePath && !string.IsNullOrEmpty(item.Postdata))
{
StreamReader r = new StreamReader(item.Postdata, postencoding);
buffer = postencoding.GetBytes(r.ReadToEnd());
r.Close();
} //写入字符串
else if (!string.IsNullOrEmpty(item.Postdata))
{
buffer = postencoding.GetBytes(item.Postdata);
}
if (buffer != null)
{
request.ContentLength = buffer.Length;
request.GetRequestStream().Write(buffer, 0, buffer.Length);
}
}
}
/// <summary>
/// 设置多个证书
/// </summary>
/// <param name="item"></param>
private void SetCerList(HttpItem item)
{
if (item.ClentCertificates != null && item.ClentCertificates.Count > 0)
{
foreach (X509Certificate c in item.ClentCertificates)
{
request.ClientCertificates.Add(c);
}
}
}
/// <summary>
/// 设置Cookie
/// </summary>
/// <param name="item">Http参数</param>
private void SetCookie(HttpItem item)
{
if (!string.IsNullOrEmpty(item.Cookie)) request.Headers[HttpRequestHeader.Cookie] = item.Cookie;
//设置CookieCollection
if (item.ResultCookieType == ResultCookieType.CookieCollection)
{
request.CookieContainer = new CookieContainer();
if (item.CookieCollection != null && item.CookieCollection.Count > 0)
request.CookieContainer.Add(item.CookieCollection);
}
}
/// <summary>
/// 设置证书
/// </summary>
/// <param name="item"></param>
private void SetCer(HttpItem item)
{
if (!string.IsNullOrEmpty(item.CerPath))
{
//这一句一定要写在创建连接的前面。使用回调的方法进行证书验证。
ServicePointManager.ServerCertificateValidationCallback = new System.Net.Security.RemoteCertificateValidationCallback(CheckValidationResult);
//初始化对像，并设置请求的URL地址
request = (HttpWebRequest)WebRequest.Create(item.URL);
SetCerList(item);
//将证书添加到请求里
request.ClientCertificates.Add(new X509Certificate(item.CerPath));
}
else
{
//初始化对像，并设置请求的URL地址
request = (HttpWebRequest)WebRequest.Create(item.URL);
SetCerList(item);
}
}
/// <summary>
/// 获取数据的并解析的方法
/// </summary>
/// <param name="item"></param>
/// <param name="result"></param>
private void GetData(HttpItem item, HttpResult result)
{
#region base
//获取StatusCode
result.StatusCode = response.StatusCode;
//获取StatusDescription
result.StatusDescription = response.StatusDescription;
//获取Headers
result.Header = response.Headers;
//获取CookieCollection
if (response.Cookies != null) result.CookieCollection = response.Cookies;
//获取set-cookie
if (response.Headers["set-cookie"] != null) result.Cookie = response.Headers["set-cookie"];
#endregion
#region byte
//处理网页Byte
byte[] ResponseByte = GetByte();
#endregion
#region Html
if (ResponseByte != null & ResponseByte.Length > 0)
{
//设置编码
SetEncoding(item, result, ResponseByte);
//得到返回的HTML
result.Html = encoding.GetString(ResponseByte);
}
else
{
//没有返回任何Html代码
result.Html = string.Empty;
}
#endregion
}
public void LoadAllUrl()
{
Dictionary<string, string> combine = new Dictionary<string, string>();
//学段
string[] ids = new string[] { "3", "1063", "2075" };
Queue<string> idqueue = new Queue<string>();
for (int i = 0; i < ids.Length; i++)
{
idqueue.Enqueue(ids[i]);
combine.Add(ids[i], ids[i]);
}
//end for
while (idqueue.Count != 0)
{
string nowid = idqueue.Dequeue();
string url = string.Format("http://www.91taoke.com/Taocan/taocan_list/id/{0}", nowid);
List<string> list = GetCousreList(url);
HttpItem item = new HttpItem()
{
URL = url,
Method = "GET",
IsToLower = false,
Timeout = 100000,
ReadWriteTimeout = 60000,
UserAgent = "Mozilla/5.0 (compatible; MSIE 9.0; Windows NT 6.1; Trident/5.0)",
ContentType = "text/html",
Allowautoredirect = false,
ResultType = ResultType.String
};
HttpResult result = http.GetHtml(item);
Document doc = NSoupClient.Parse(result.Html);
Elements ems = doc.Body.GetElementsByAttributeValue("class", "menuList cf");
Elements xueduanlinks = ems[0].GetElementsByTag("a");
List<string> xueduan = new List<string>();
foreach (Element e in xueduanlinks)
{
string href = e.Attr("href").ToString();
xueduan.Add(href.Substring(href.IndexOf(',') + 1, href.Length - href.IndexOf(',') - 2));
}
Elements nianjilinks = ems[1].GetElementsByTag("a");
List<string> nianji = new List<string>();
foreach (Element e in nianjilinks)
{
string href = e.Attr("href").ToString();
nianji.Add(href.Substring(href.IndexOf(',') + 1, href.Length - href.IndexOf(',') - 2));
}
Elements xuekelinks = ems[2].GetElementsByTag("a");
List<string> xueke = new List<string>();
foreach (Element e in xuekelinks)
{
string href = e.Attr("href").ToString();
xueke.Add(href.Substring(href.IndexOf(',') + 1, href.Length - href.IndexOf(',') - 2));
}
Elements versionlinks = ems[3].GetElementsByTag("a");
List<string> version = new List<string>();
foreach (Element e in versionlinks)
{
string href = e.Attr("href").ToString();
version.Add(href.Substring(href.IndexOf(',') + 1, href.Length - href.IndexOf(',') - 2));
}
if (nowid.Split(',').Length == 1)
{
for (int j = 0; j < nianji.Count; j++)
{
if (!combine.ContainsKey(nowid + "," + nianji[j]))
{
combine.Add(nowid + "," + nianji[j], nowid + "," + nianji[j]);
idqueue.Enqueue(nowid + "," + nianji[j]);
}
for (int k = 0; k < xueke.Count; k++)
{
if (!combine.ContainsKey(nowid + "," + nianji[j] + "," + xueke[k]))
{
combine.Add(nowid + "," + nianji[j] + "," + xueke[k], nowid + "," + nianji[j] + "," + xueke[k]);
idqueue.Enqueue(nowid + "," + nianji[j] + "," + xueke[k]);
}
for (int l = 0; l < version.Count; l++)
{
if (!combine.ContainsKey(nowid + "," + nianji[j] + "," + xueke[k] + "," + version[l]))
{
combine.Add(nowid + "," + nianji[j] + "," + xueke[k] + "," + version[l], nowid + "," + nianji[j] + "," + xueke[k] + "," + version[l]);
idqueue.Enqueue(nowid + "," + nianji[j] + "," + xueke[k] + "," + version[l]);
}
}
}
}
}
if (nowid.Split(',').Length == 2)
{
for (int k = 0; k < xueke.Count; k++)
{
if (!combine.ContainsKey(nowid + "," + xueke[k]))
{
combine.Add(nowid + "," + xueke[k], nowid + "," + "," + xueke[k]);
idqueue.Enqueue(nowid + "," + xueke[k]);
}
for (int l = 0; l < version.Count; l++)
{
if (!combine.ContainsKey(nowid + "," + xueke[k] + "," + version[l]))
{
combine.Add(nowid + "," + xueke[k] + "," + version[l], nowid + "," + xueke[k] + "," + version[l]);
idqueue.Enqueue(nowid + "," + xueke[k] + "," + version[l]);
}
}
}
}
if (nowid.Split(',').Length == 3)
{
for (int l = 0; l < version.Count; l++)
{
if (!combine.ContainsKey(nowid + "," + version[l]))
{
combine.Add(nowid + "," + version[l], nowid + "," + version[l]);
idqueue.Enqueue(nowid + "," + version[l]);
}
}
}
//for (int i = 0; i < xueduan.Count; i++)
//{
// if (!combine.ContainsKey(xueduan[i]))
// {
// combine.Add(xueduan[i], xueduan[i]);
// idqueue.Enqueue(xueduan[i]);
// }
// for (int j = 0; j < nianji.Count; j++)
// {
// if (!combine.ContainsKey(xueduan[i] + "," + nianji[j]))
// {
// combine.Add(xueduan[i] + "," + nianji[j], xueduan[i] + "," + nianji[j]);
// idqueue.Enqueue(xueduan[i] + "," + nianji[j]);
// }
// for (int k = 0; k < xueke.Count; k++)
// {
// if (!combine.ContainsKey(xueduan[i] + "," + nianji[j] + "," + xueke[k]))
// {
// combine.Add(xueduan[i] + "," + nianji[j] + "," + xueke[k], xueduan[i] + "," + nianji[j] + "," + xueke[k]);
// idqueue.Enqueue(xueduan[i] + "," + nianji[j] + "," + xueke[k]);
// }
// for (int l = 0; l < version.Count; l++)
// {
// if (!combine.ContainsKey(xueduan[i] + "," + nianji[j] + "," + xueke[k] + "," + version[l]))
// {
// combine.Add(xueduan[i] + "," + nianji[j] + "," + xueke[k] + "," + version[l], xueduan[i] + "," + nianji[j] + "," + xueke[k] + "," + version[l]);
// idqueue.Enqueue(xueduan[i] + "," + nianji[j] + "," + xueke[k] + "," + version[l]);
// }
// }
// }
// }
//}
WriteToLog(idqueue.Count + "----" + combine.Count + "----" + nowid + "----" + url);
}
foreach (var item in combine)
{
try
{
FileStream file = new FileStream("d://url.txt", FileMode.Append);
StreamWriter sw = new StreamWriter(file);
sw.WriteLine(string.Format("http://www.91taoke.com/Taocan/taocan_list/id/{0}", item.Value));
sw.Close();
file.Close();
}
catch (Exception)
{
}
}
}
public string GetPlayHtml(string id,string nid)
{
string url = string.Format("http://www.91taoke.com/index.php?m=Taocan&a=taocan_play&id={0}&nid={1}",id ,nid);
HttpItem item = new HttpItem()
{
URL = url,
Method = "GET",
IsToLower = false,
Timeout = 100000,
ReadWriteTimeout = 60000,
UserAgent = "Mozilla/5.0 (compatible; MSIE 9.0; Windows NT 6.1; Trident/5.0)",
ContentType = "text/html",
Allowautoredirect = false,
ResultType = ResultType.String
};
HttpResult result = http.GetHtml(item);
return result.Html;
}
//从url获取课程id
public List<string> GetCousreList(string url)
{
List<string> list = new List<string>();
HttpItem item = new HttpItem()
{
URL = url,
Method = "GET",
IsToLower = false,
Timeout = 100000,
ReadWriteTimeout = 60000,
UserAgent = "Mozilla/5.0 (compatible; MSIE 9.0; Windows NT 6.1; Trident/5.0)",
ContentType = "text/html",
Allowautoredirect = false,
ResultType = ResultType.String
};
HttpResult result = http.GetHtml(item);
string html = result.Html;
Document doc = NSoupClient.Parse(html);
try
{
Elements uls = doc.GetElementsByClass("xhr_wksp_left");
if (uls.Count == 0)
return list;
Element ul = uls[0];
Elements lis = ul.GetElementsByTag("li");
foreach (Element e in lis)
{
Elements divs = e.GetElementsByClass("xhr_wksp_r_h");
if (divs.Count == 0)
continue;
Element a = divs[0].GetElementsByTag("a")[0];
string[] array = a.Attr("href").Split('/');
string courseid = array[array.Length - 1];
list.Add(courseid);
}
}
catch (Exception)
{
return list;
}
return list;
}