GoalSingleDbListJob.cs 8.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213
  1. using System;
  2. using System.Collections.Generic;
  3. using FCS.Common;
  4. using FCS.Interface;
  5. using FCS.Models;
  6. using HtmlAgilityPack;
  7. using Quartz;
  8. using System.Data;
  9. using System.Linq;
  10. using System.Threading.Tasks;
  11. using System.Threading;
  12. using System.Diagnostics;
  13. namespace FCS.Crawler.ZCMatchRankingList
  14. {
  15. /// <summary>
  16. /// 进球数单双统计
  17. /// </summary>
  18. public class GoalSingleDbListJob : CommonJob, IJob
  19. {
  20. private static List<F_SingleDoubleList> SingleDoubleLists = new List<F_SingleDoubleList>();
  21. private static List<F_Team> TeamList = new List<F_Team>();
  22. public GoalSingleDbListJob()
  23. {
  24. log = new LogHelper();
  25. services = IOC.Resolve<IDTOpenCode>();
  26. }
  27. public void Execute(IJobExecutionContext context)
  28. {
  29. Config = CommonHelper.GetConfigFromDataMap(context.JobDetail.JobDataMap);
  30. GetAll();
  31. }
  32. #region 更新数据
  33. /// <summary>
  34. /// 获取所有数据
  35. /// </summary>
  36. public void GetAll()
  37. {
  38. GetSqlString(GetAllEventUrl,"");
  39. }
  40. /// <summary>
  41. /// 更新数据,默认为今年数据
  42. /// </summary>
  43. /// <param name="Year"></param>
  44. public void Click(string Year = "")
  45. {
  46. if (Year == "")
  47. {
  48. Year = DateTime.Now.Year.ToString();
  49. }
  50. var sql = string.Format(GetNewEventUrl, Year);
  51. GetSqlString(sql, Year);
  52. }
  53. #endregion
  54. /// <summary>
  55. /// 解析数据
  56. /// </summary>
  57. /// <param name="sqlstring"></param>
  58. private void GetSqlString(string sqlstring,string Year)
  59. {
  60. ThreadPool.SetMinThreads(10, 10);
  61. ThreadPool.SetMaxThreads(500, 500);
  62. TeamList = services.GetTeamList();
  63. var ds = SqlHelper.ExecuteDataset(CommandType.Text, sqlstring);
  64. if (ds != null && ds.Tables.Count > 0)
  65. {
  66. for (int i = 0; i < ds.Tables[0].Rows.Count; i++)
  67. {
  68. //拿去赛事组的链接
  69. var url = ds.Tables[0].Rows[i]["Remark"].ToString();
  70. var eventId = ds.Tables[0].Rows[i]["EventId"].ToString();
  71. var season = ds.Tables[0].Rows[i]["Season"].ToString();
  72. ////获取到总进球单双的链接
  73. Task.Run(() =>
  74. {
  75. HtmlDocument doc = CommonHelper.GetHtml(url, new Dictionary<string, string>(), "", "", 10000, 100);
  76. var TJList = doc.DocumentNode.SelectNodes("//*[@class='tongji_list']");
  77. if (TJList != null && TJList.Count() != 0 && TJList[0].InnerText.Contains("赛事统计"))
  78. {
  79. HtmlDocument doc1 = new HtmlDocument();
  80. doc1.LoadHtml(TJList[0].InnerHtml);
  81. var liList = doc1.DocumentNode.SelectNodes("//ul/li/a").ToList();
  82. var Glod = liList.Where(o => o.InnerText.Contains("进球单双")).First();
  83. if (Glod != null)
  84. {
  85. var jqdsurl = Glod.Attributes.SingleOrDefault(a => a.Name.Equals("href")).Value.ToString();
  86. Task.Run(() =>
  87. {
  88. GetJQDSHtml(jqdsurl, eventId, season);
  89. });
  90. }
  91. }
  92. });
  93. }
  94. int maxWorkerThreads, workerThreads;
  95. int maxportThreads, portThreads;
  96. while (true)
  97. {
  98. /*
  99. GetAvailableThreads():检索由 GetMaxThreads 返回的线程池线程的最大数目和当前活动数目之间的差值。
  100. 而GetMaxThreads 检索可以同时处于活动状态的线程池请求的数目。
  101. 通过最大数目减可用数目就可以得到当前活动线程的数目,如果为零,那就说明没有活动线程,说明所有线程运行完毕。
  102. */
  103. ThreadPool.GetMaxThreads(out maxWorkerThreads, out maxportThreads);
  104. ThreadPool.GetAvailableThreads(out workerThreads, out portThreads);
  105. Thread.Sleep(1000);
  106. Trace.WriteLine("正在执行任务的线程数" + (maxWorkerThreads - workerThreads));
  107. if (maxWorkerThreads - workerThreads == 0)
  108. {
  109. Console.WriteLine("Thread Finished!");
  110. break;
  111. }
  112. }
  113. try
  114. {
  115. if (Year!="")
  116. {
  117. services.Delete<F_SingleDoubleList>(" and Season like '%"+Year+"%'");
  118. }
  119. services.SqlBulkCopyAdd<F_SingleDoubleList>(SingleDoubleLists);
  120. Trace.WriteLine("F_SingleDoubleList更新完毕");
  121. }
  122. catch (Exception ex)
  123. {
  124. throw;
  125. }
  126. }
  127. }
  128. /// <summary>
  129. /// 解析进球单双页面
  130. /// </summary>
  131. /// <param name="url"></param>
  132. /// <param name="eventId"></param>
  133. /// <param name="season"></param>
  134. public void GetJQDSHtml(string url, string eventId, string season)
  135. {
  136. if (url.Contains("wwaattssuunn"))
  137. {
  138. return;
  139. }
  140. HtmlDocument doc = CommonHelper.GetHtml(url,new Dictionary<string,string>(),"","",1000,100);
  141. var trData = doc.DocumentNode.SelectNodes("//*[@class='table_out']/table/tbody/tr");
  142. if (trData == null || trData.Count() == 0) return;
  143. //读取进球单双信息并加入到数据库
  144. foreach (var tritem in trData)
  145. {
  146. try
  147. {
  148. HtmlDocument doc11 = new HtmlDocument();
  149. doc11.LoadHtml(tritem.InnerHtml);
  150. var tdList = doc11.DocumentNode.SelectNodes("//td");
  151. if (tdList == null || tdList.Count == 0) continue;
  152. F_SingleDoubleList f_SingleDoubleList = new F_SingleDoubleList();
  153. f_SingleDoubleList.Id = Guid.NewGuid().ToString();
  154. f_SingleDoubleList.EventId = eventId;
  155. f_SingleDoubleList.Season = season;
  156. f_SingleDoubleList.Sort = tdList[0].InnerText.Trim() == null ? 0 : int.Parse(tdList[0].InnerText.Trim().ToString());
  157. var TeamName = tdList[1].InnerText.Trim().ToString();
  158. var Team = TeamList.Where(o => o.Name == TeamName);
  159. f_SingleDoubleList.TeamId = Team.Count() > 0 ? Team.FirstOrDefault().Id : "";
  160. if (Team.Count() == 0)
  161. {
  162. f_SingleDoubleList.Remark = TeamName;
  163. }
  164. f_SingleDoubleList.OneNum = tdList.Count()>3? int.Parse(tdList[2].InnerText.Trim().ToString()):0;
  165. f_SingleDoubleList.ZeroNum = tdList.Count() > 4 ? int.Parse(tdList[3].InnerText.Trim().ToString()) : 0;
  166. f_SingleDoubleList.TwoNum = tdList.Count() > 5 ? int.Parse(tdList[4].InnerText.Trim().ToString()) : 0;
  167. f_SingleDoubleList.ThreeNum = tdList.Count() > 6 ? int.Parse(tdList[5].InnerText.Trim().ToString()) : 0;
  168. f_SingleDoubleList.FourNum = tdList.Count() > 7 ? int.Parse(tdList[6].InnerText.Trim().ToString()) : 0;
  169. f_SingleDoubleList.OddNum = tdList.Count() > 8 ? int.Parse(tdList[7].InnerText.Trim().ToString()) : 0;
  170. f_SingleDoubleList.EvenNum = tdList.Count() > 9 ? int.Parse(tdList[8].InnerText.Trim().ToString()) : 0;
  171. SingleDoubleLists.Add(f_SingleDoubleList);
  172. }
  173. catch (Exception ex)
  174. {
  175. throw;
  176. }
  177. }
  178. }
  179. #region SQL语句
  180. /// <summary>
  181. /// 获取所有的赛事
  182. /// </summary>
  183. private static string GetAllEventUrl = @"select distinct EventId,Season,Remark from F_Grouping where Remark is not null";
  184. /// <summary>
  185. /// 获取最新两年的数据赛事
  186. /// </summary>
  187. private static string GetNewEventUrl = @"select distinct EventId,Season,Remark from F_Grouping where Remark is not null and Season like '%{0}%'";
  188. private static string DeleteNewEventUrl = @"delete F_SingleDoubleList where Season like '{0}%'";
  189. #endregion
  190. }
  191. }