FootballPlayerJob.cs 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237
  1. using System;
  2. using System.Collections.Generic;
  3. using FCS.Common;
  4. using FCS.Crawler.Tools;
  5. using FCS.Interface;
  6. using FCS.Models;
  7. using HtmlAgilityPack;
  8. using Newtonsoft.Json;
  9. using Quartz;
  10. using System.Data;
  11. using System.Linq;
  12. using System.Text;
  13. using System.Text.RegularExpressions;
  14. using System.Threading;
  15. using System.Threading.Tasks;
  16. using System.Diagnostics;
  17. using FCS.Models.DTO;
  18. namespace FCS.Crawler.ZCLotteryTeam
  19. {
  20. /// <summary>
  21. /// 球員爬取
  22. /// </summary>
  23. public class FootballPlayerJob : CommonJob, IJob
  24. {
  25. public List<F_Relation> relations = new List<F_Relation>();//球员球队关系
  26. public List<F_Players> players = new List<F_Players>();//球员列表
  27. public List<F_Players> allPlayer = new List<F_Players>();
  28. F_Players g;
  29. public FootballPlayerJob()
  30. {
  31. log = new LogHelper();
  32. services = IOC.Resolve<IDTOpenCode>();
  33. g = new F_Players();
  34. }
  35. public void Execute(IJobExecutionContext context)
  36. {
  37. Config = CommonHelper.GetConfigFromDataMap(context.JobDetail.JobDataMap);
  38. GetAll();
  39. }
  40. public void GetAll()
  41. {
  42. var ds = services.Query<F_Team>("and Remark is not null").ToList();
  43. allPlayer = services.GetPlayerList();
  44. if (ds != null && ds.Count > 0)
  45. {
  46. //获取赛事存取的url ds.Tables[0].Rows.Count
  47. //int max = ds.Count;
  48. //int num = 0;
  49. ds.ForEach(p =>
  50. {
  51. //异步加载分组
  52. var url = p.Remark.ToString();
  53. var TeamId = p.Id.ToString();
  54. var TeamName = p.Name.ToString();
  55. Task.Run(() =>
  56. {
  57. GetPlayerListHtml(url, TeamId, TeamName);
  58. });
  59. //lock (g)
  60. //{
  61. // num++;
  62. // Monitor.Pulse(g); //完成,通知等待队列,告知已完,执行下一个。
  63. //}
  64. });
  65. //lock (g)
  66. //{
  67. // while (num < max)
  68. // {
  69. // Monitor.Wait(g);//等待
  70. // }
  71. //}
  72. //Trace.WriteLine("获取球员 完结");
  73. while (true)
  74. {
  75. if (CommonHelper.ThreadsFinsh())
  76. break;
  77. }
  78. }
  79. try
  80. {
  81. services.SqlBulkCopyAdd<F_Players>(players);
  82. services.SqlBulkCopyAdd<F_Relation>(relations);
  83. }
  84. catch (Exception ex)
  85. {
  86. throw;
  87. }
  88. }
  89. /// <summary>
  90. /// 读取球员列表
  91. /// </summary>
  92. /// <param name="url"></param>
  93. /// <param name="teamId"></param>
  94. /// <param name="teamName"></param>
  95. /// <returns></returns>
  96. public bool GetPlayerListHtml(string url, string teamId, string teamName)
  97. {
  98. if (url.Contains("wwaattssuunn"))
  99. {
  100. return false;
  101. }
  102. HtmlParameterDTO dtomodel = new HtmlParameterDTO();
  103. dtomodel.Url = url;
  104. dtomodel.Timeout = 10 * 1000;
  105. dtomodel.IsCheckEmpty = false;
  106. HtmlDocument doc = CommonHelper.GetHtmlHtmlDocument(dtomodel);// CommonHelper.GetHtml(url, new Dictionary<string, string>(), "", "", 90000, 10);
  107. if (doc.DocumentNode.InnerHtml == "Termination")
  108. {
  109. return false;
  110. }
  111. //获取球员的列表
  112. var TongJiListData = doc.DocumentNode.SelectNodes("//*[@class='mingd_right']/dl");
  113. if (TongJiListData == null || TongJiListData.Count == 0) return false;
  114. foreach (var list in TongJiListData)
  115. {
  116. HtmlDocument doc1 = new HtmlDocument();
  117. doc1.LoadHtml(list.InnerHtml);
  118. var playerLIst = doc1.DocumentNode.SelectNodes("//dd/a");
  119. var playeerType = doc1.DocumentNode.SelectNodes("//dt");
  120. var position = "";
  121. if (playeerType.Count > 0)
  122. {
  123. position = playeerType.FirstOrDefault().InnerText.Replace("/t", "").Trim();
  124. }
  125. if (playerLIst == null)
  126. {
  127. continue;
  128. }
  129. Task.Run(() =>
  130. {
  131. foreach (var playitem in playerLIst)
  132. {
  133. var playerurl = playitem.Attributes.SingleOrDefault(a => a.Name.Equals("href")).Value;
  134. var playerName = playitem.InnerText.Trim();
  135. var playerInfo = "";
  136. if (allPlayer.Where(o => o.ChineseName == playerName).Count() == 0)
  137. {
  138. playerInfo = GetPlayerInfo(playerurl, playerName);
  139. }
  140. var playid = ""; var num = "";
  141. if (playerInfo != "")//不为空表示需要新增
  142. {
  143. playid = playerInfo.Split(',').Count() > 0 ? playerInfo.Split(',')[0] : "";
  144. num = playerInfo.Split(',').Count() > 1 ? playerInfo.Split(',')[1] : "0";
  145. F_Relation f_Relation = new F_Relation();
  146. f_Relation.Id = Guid.NewGuid().ToString();
  147. f_Relation.TeamId = teamId;
  148. f_Relation.PlayerId = playid;
  149. f_Relation.PoloShirt = num == "" ? 0 : int.Parse(num);
  150. f_Relation.Position = position;
  151. f_Relation.CreateDateTime = DateTime.Now;
  152. f_Relation.ContractTime = DateTime.Now;
  153. if (f_Relation.PlayerId != null && f_Relation.PlayerId != "")
  154. {
  155. relations.Add(f_Relation);
  156. }
  157. }
  158. }
  159. });
  160. }
  161. return true;
  162. }
  163. /// <summary>
  164. /// 解析球员信息,并返回id
  165. /// </summary>
  166. /// <param name="url"></param>
  167. /// <param name="name"></param>
  168. /// <returns></returns>
  169. public string GetPlayerInfo(string url, string name)
  170. {
  171. var mainUrl = "http://saishi.zgzcw.com/" + url;
  172. HtmlParameterDTO dtomodel = new HtmlParameterDTO();
  173. dtomodel.Url = url;
  174. dtomodel.Timeout = 10 * 1000;
  175. dtomodel.IsCheckEmpty = false;
  176. HtmlDocument doc = CommonHelper.GetHtmlHtmlDocument(dtomodel); //CommonHelper.GetHtml(mainUrl,new Dictionary<string, string>() ,"","",90000,10);
  177. if (doc.DocumentNode.InnerHtml == "Termination")
  178. {
  179. return "";
  180. }
  181. //球员图片
  182. var teamImg = doc.DocumentNode.SelectNodes("//*[@class='star_dl']/dt/img");
  183. //球员介绍
  184. var teamIntroData = doc.DocumentNode.SelectNodes("//*[@class='introduceDiv']") == null || doc.DocumentNode.SelectNodes("//*[@class='introduceDiv']").Count == 0 ? "" : doc.DocumentNode.SelectNodes("//*[@class='introduceDiv']").FirstOrDefault().InnerHtml.Trim();
  185. //球员的基本信息
  186. if (doc.DocumentNode.SelectNodes("//*[@class='star_dl']/dd") == null || doc.DocumentNode.SelectNodes("//*[@class='star_dl']/dd").Count == 0)
  187. {
  188. return "";
  189. }
  190. var playerData = doc.DocumentNode.SelectNodes("//*[@class='star_dl']/dd").ToList();
  191. F_Players f_Players = new F_Players();
  192. f_Players.Id = Guid.NewGuid().ToString();
  193. f_Players.ChineseName = name;
  194. f_Players.Describe = teamIntroData;
  195. f_Players.Remark = mainUrl;
  196. f_Players.LogoImage = teamImg.FirstOrDefault().Attributes.SingleOrDefault(a => a.Name.Equals("src")).Value;
  197. var contry = playerData.Where(o => o.InnerText.Contains("国家")).Count() > 0 ? playerData.Where(o => o.InnerText.Contains("国家")).FirstOrDefault().InnerText.Replace("\t", "").Split(':') : null;
  198. f_Players.Nationality = contry != null && contry.Count() > 1 ? contry[1].Trim() : "";
  199. var Weight = playerData.Where(o => o.InnerText.Contains("体重")).Count() > 0 ? playerData.Where(o => o.InnerText.Contains("体重")).FirstOrDefault().InnerText.Replace("\t", "").Split(':') : null;
  200. f_Players.Weight = Weight != null && Weight.Count() > 1 ? Weight[1].Trim() : "";
  201. var Stature = playerData.Where(o => o.InnerText.Contains("身高")).Count() > 0 ? playerData.Where(o => o.InnerText.Contains("身高")).FirstOrDefault().InnerText.Replace("\t", "").Split(':') : null;
  202. f_Players.Stature = Stature != null && Stature.Count() > 1 ? Stature[1].Trim() : "";
  203. var Birthday = playerData.Where(o => o.InnerText.Contains("生日")).Count() > 0 ? playerData.Where(o => o.InnerText.Contains("生日")).FirstOrDefault().InnerText.Replace("\t", "").Split(':') : null;
  204. f_Players.Birthday = Birthday != null && Birthday.Count() > 1 ? Birthday[1].Trim() : "";
  205. var num = playerData.Where(o => o.InnerText.Contains("号码")).Count() > 0 ? playerData.Where(o => o.InnerText.Contains("号码")).FirstOrDefault().InnerText.Replace("\t", "").Split(':') : null;
  206. var playerNum = num != null && num.Count() > 1 ? num[1].Trim() : "";
  207. //if (allPlayer.Where(o => o.ChineseName == f_Players.ChineseName).Count() == 0)
  208. //{
  209. players.Add(f_Players);
  210. return f_Players.Id + "," + playerNum;
  211. //}
  212. //services.AddPlayers(FCSLottery.F_Players, f_Players);
  213. //return "";
  214. }
  215. #region SQL语句
  216. private static string GetAllTeamUrl = @"select Id,Name, Remark from F_Team where Remark is not null";
  217. #endregion
  218. }
  219. }