FC3DGlossaryJob.cs 6.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204
  1. using System;
  2. using System.Collections.Generic;
  3. using System.Linq;
  4. using System.Text;
  5. using HtmlAgilityPack;
  6. using Quartz;
  7. using SCC.Common;
  8. using SCC.Interface;
  9. using SCC.Models;
  10. namespace SCC.Crawler.LotteryGlossary
  11. {
  12. public class FC3DGlossaryJob : IJob
  13. { /// <summary>
  14. /// 构造函数
  15. /// </summary>
  16. public FC3DGlossaryJob()
  17. {
  18. log = new LogHelper();
  19. services = IOC.Resolve<IDTOpenCode>();
  20. email = IOC.Resolve<IEmail>();
  21. }
  22. /// <summary>
  23. /// 执行入口
  24. /// </summary>
  25. /// <param name="context"></param>
  26. public void Execute(IJobExecutionContext context)
  27. {
  28. Config = CommonHelper.GetConfigFromDataMap(context.JobDetail.JobDataMap);
  29. DoMainUrl();
  30. }
  31. /// <summary>
  32. /// 执行主站技巧
  33. /// </summary>
  34. private void DoMainUrl()
  35. {
  36. List<LotteryGlossaryModel> res = GetOpenListFromMainUrl(Config.MainUrl);
  37. foreach (var LotteryGlossaryModel in res)
  38. {
  39. if (services.LotteryGlossaryModel(currentLottery, LotteryGlossaryModel))
  40. {
  41. //Do Success Log
  42. log.Info(GetType(), CommonHelper.GetJobMainLogInfo(Config, LotteryGlossaryModel.Title));
  43. isGetData = true;
  44. }
  45. }
  46. }
  47. /// <summary>
  48. /// 爬取主站技巧列表
  49. /// </summary>
  50. /// <param name="mainUrl"></param>
  51. /// <returns></returns>
  52. private List<LotteryGlossaryModel> GetOpenListFromMainUrl(string mainUrl)
  53. {
  54. var result = new List<LotteryGlossaryModel>();
  55. try
  56. {
  57. var url = new Uri(mainUrl);
  58. var htmlResource = NetHelper.GetUrlResponse(mainUrl, Encoding.GetEncoding("utf-8"));
  59. if (htmlResource == null) return result;
  60. HtmlDocument doc = new HtmlDocument();
  61. doc.LoadHtml(htmlResource);
  62. //获取li下面所有a标签
  63. HtmlNodeCollection nodeList = doc.DocumentNode.SelectNodes("//*[@id='3d']/div[3]/ul/li/a");//*[@id="ssq"]/div[3]
  64. if (nodeList == null) return result;
  65. List<string> urls = new List<string>();
  66. //遍历a标签
  67. foreach (HtmlNode node in nodeList)
  68. {
  69. HtmlAttribute attr = node.Attributes.SingleOrDefault(a => a.Name.Equals("href"));
  70. if (attr != null)
  71. {
  72. string href = Host + attr.Value;
  73. //去重
  74. if (!urls.Contains(href))
  75. {
  76. urls.Add(href);
  77. }
  78. }
  79. }
  80. foreach (var url1 in urls)
  81. {
  82. var LotterySkill = GetSkillModel(url1);
  83. result.Add(LotterySkill);
  84. }
  85. }
  86. catch (Exception ex)
  87. {
  88. log.Error(GetType(),
  89. string.Format("【{0}】通过主抓取开奖列表时发生错误,错误信息【{1}】", Config.Area + currentLottery, ex.Message));
  90. }
  91. return result;
  92. }
  93. /// <summary>
  94. /// 根据主站url获取技巧详情
  95. /// </summary>
  96. /// <param name="url"></param>
  97. /// <returns></returns>
  98. private LotteryGlossaryModel GetSkillModel(string url)
  99. {
  100. LotteryGlossaryModel lotterySkill = new LotteryGlossaryModel();
  101. try
  102. {
  103. var htmlResource = NetHelper.GetUrlResponse(url, Encoding.GetEncoding("utf-8"));
  104. if (htmlResource == null) return lotterySkill;
  105. HtmlDocument doc = new HtmlDocument();
  106. doc.LoadHtml(htmlResource);
  107. //获取li下面所有a标签
  108. var div = doc.DocumentNode.SelectSingleNode("//*[@class='artile']");
  109. var Title = div.ChildNodes.Where(node => node.Name == "h1").ToList();
  110. var div1 = div.ChildNodes.Where(node => node.Name == "div").ToList();
  111. lotterySkill.Title = Title[0].InnerText.Trim();
  112. lotterySkill.Author = "cn55128";
  113. lotterySkill.Content = div1[1].InnerHtml.Trim();
  114. lotterySkill.IsDelete = false;
  115. lotterySkill.SourceUrl = url.ToString();
  116. lotterySkill.TypeID = lotterySkillType;
  117. lotterySkill.TypeName = lotterySkillType.GetEnumDescription();
  118. }
  119. catch (Exception ex)
  120. {
  121. log.Error(GetType(),
  122. string.Format("【{0}】通过主抓取开奖列表时发生错误,错误信息【{1}】", Config.Area + currentLottery, ex.Message));
  123. }
  124. return lotterySkill;
  125. }
  126. #region Attribute
  127. /// <summary>
  128. /// 主机地址
  129. /// </summary>
  130. public string Host = "http://www.55125.cn/";
  131. /// <summary>
  132. /// 配置信息
  133. /// </summary>
  134. private SCCConfig Config;
  135. /// <summary>
  136. /// 当天抓取的最新一期开奖记录
  137. /// </summary>
  138. private LotteryGlossaryModel LatestItem = null;
  139. /// <summary>
  140. /// 当天抓取失败列表
  141. /// </summary>
  142. private List<string> FailedQiHaoList = null;
  143. /// <summary>
  144. /// 日志对象
  145. /// </summary>
  146. private readonly LogHelper log;
  147. /// <summary>
  148. /// 数据服务
  149. /// </summary>
  150. private readonly IDTOpenCode services;
  151. /// <summary>
  152. /// 当前彩种
  153. /// </summary>
  154. private SCCLottery currentLottery => SCCLottery.LotteryGlossary;
  155. /// <summary>
  156. /// 福彩3D技巧
  157. /// </summary>
  158. private LotteryGlossaryType lotterySkillType = LotteryGlossaryType.FC3D;
  159. /// <summary>
  160. /// 邮件接口
  161. /// </summary>
  162. private IEmail email;
  163. /// <summary>
  164. /// 是否本次运行抓取到开奖数据
  165. /// </summary>
  166. private bool isGetData = false;
  167. #endregion
  168. }
  169. }