SSqSkillJob.cs 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355
  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.LotterySkill
  11. {
  12. public class SSqSkillJob : IJob
  13. {
  14. /// <summary>
  15. /// 构造函数
  16. /// </summary>
  17. public SSqSkillJob()
  18. {
  19. log = new LogHelper();
  20. services = IOC.Resolve<IDTOpenCode>();
  21. email = IOC.Resolve<IEmail>();
  22. }
  23. /// <summary>
  24. /// 执行入口
  25. /// </summary>
  26. /// <param name="context"></param>
  27. public void Execute(IJobExecutionContext context)
  28. {
  29. Config = CommonHelper.GetConfigFromDataMap(context.JobDetail.JobDataMap);
  30. DoBackUrl();
  31. DoMainUrl();
  32. }
  33. /// <summary>
  34. /// 执行主站技巧
  35. /// </summary>
  36. private void DoMainUrl()
  37. {
  38. List<string> urls = GetMainUrl(Config);
  39. #pragma warning disable CS0219 // 变量“lotterySkill”已被赋值,但从未使用过它的值
  40. LotterySkillModel lotterySkill = null;
  41. #pragma warning restore CS0219 // 变量“lotterySkill”已被赋值,但从未使用过它的值
  42. foreach (string url in urls)
  43. {
  44. List<LotterySkillModel> res = GetOpenListFromMainUrl(url);
  45. foreach (var lotterySkillModel in res)
  46. {
  47. if (services.LotterySkillModel(currentLottery, lotterySkillModel))
  48. {
  49. //Do Success Log
  50. log.Info(GetType(), CommonHelper.GetJobMainLogInfo(Config, lotterySkillModel.Title));
  51. isGetData = true;
  52. }
  53. }
  54. }
  55. }
  56. /// <summary>
  57. /// 执行副站技巧
  58. /// </summary>
  59. private void DoBackUrl()
  60. {
  61. List<string> urls = GetBackUrl(Config);
  62. #pragma warning disable CS0219 // 变量“lotterySkill”已被赋值,但从未使用过它的值
  63. LotterySkillModel lotterySkill = null;
  64. #pragma warning restore CS0219 // 变量“lotterySkill”已被赋值,但从未使用过它的值
  65. foreach (string url in urls)
  66. {
  67. List<LotterySkillModel> res = GetOpenListFromBackUrl(url);
  68. foreach (var lotterySkillModel in res)
  69. {
  70. if (services.LotterySkillModel(currentLottery, lotterySkillModel))
  71. {
  72. //Do Success Log
  73. log.Info(GetType(), CommonHelper.GetJobMainLogInfo(Config, lotterySkillModel.Title));
  74. isGetData = true;
  75. }
  76. }
  77. }
  78. }
  79. /// <summary>
  80. /// 组装主站爬取地址
  81. /// </summary>
  82. /// <param name="config"></param>
  83. /// <returns></returns>
  84. private List<string> GetMainUrl(SCCConfig config)
  85. {
  86. List<string> urlList = new List<string>();
  87. string url = config.MainUrl;
  88. int pages = config.MainUrlPages > 0 ? config.MainUrlPages : 1;
  89. for (int i = 1; i <= pages; i++)
  90. {
  91. string res = string.Format(url, i);
  92. if (!urlList.Contains(res))
  93. {
  94. urlList.Add(res);
  95. }
  96. }
  97. return urlList;
  98. }
  99. /// <summary>
  100. /// 组装副站爬取地址
  101. /// </summary>
  102. /// <param name="config"></param>
  103. /// <returns></returns>
  104. private List<string> GetBackUrl(SCCConfig config)
  105. {
  106. List<string> urlList = new List<string>();
  107. string url = config.BackUrl;
  108. int pages = config.MainUrlPages > 0 ? config.BackUrlPages : 1;
  109. for (int i = 1; i <= pages; i++)
  110. {
  111. string res = string.Format(url, i);
  112. if (!urlList.Contains(res))
  113. {
  114. urlList.Add(res);
  115. }
  116. }
  117. return urlList;
  118. }
  119. /// <summary>
  120. /// 爬取主站技巧列表
  121. /// </summary>
  122. /// <param name="mainUrl"></param>
  123. /// <returns></returns>
  124. private List<LotterySkillModel> GetOpenListFromMainUrl(string mainUrl)
  125. {
  126. var result = new List<LotterySkillModel>();
  127. try
  128. {
  129. var url = new Uri(mainUrl);
  130. var htmlResource = NetHelper.GetUrlResponse(mainUrl, Encoding.GetEncoding("utf-8"));
  131. if (htmlResource == null) return result;
  132. HtmlDocument doc = new HtmlDocument();
  133. doc.LoadHtml(htmlResource);
  134. //获取li下面所有a标签
  135. HtmlNodeCollection nodeList = doc.DocumentNode.SelectNodes("//*[@class='art-list']/ul/li/a");
  136. if (nodeList == null) return result;
  137. List<string> urls = new List<string>();
  138. //遍历a标签
  139. foreach (HtmlNode node in nodeList)
  140. {
  141. HtmlAttribute attr = node.Attributes.SingleOrDefault(a => a.Name.Equals("href"));
  142. if (attr != null)
  143. {
  144. string href = Host + attr.Value;
  145. //去重
  146. if (!urls.Contains(href))
  147. {
  148. urls.Add(href);
  149. }
  150. }
  151. }
  152. foreach (var url1 in urls)
  153. {
  154. var LotterySkill = GetSkillModel(url1);
  155. result.Add(LotterySkill);
  156. }
  157. }
  158. catch (Exception ex)
  159. {
  160. log.Error(GetType(),
  161. 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));
  162. }
  163. return result;
  164. }
  165. /// <summary>
  166. /// 爬取副站技巧列表
  167. /// </summary>
  168. /// <param name="backUrl"></param>
  169. /// <returns></returns>
  170. private List<LotterySkillModel> GetOpenListFromBackUrl(string backUrl)
  171. {
  172. var result = new List<LotterySkillModel>();
  173. try
  174. {
  175. var url = new Uri(backUrl);
  176. var htmlResource = NetHelper.GetUrlResponse(backUrl, Encoding.GetEncoding("utf-8"));
  177. if (htmlResource == null) return result;
  178. HtmlDocument doc = new HtmlDocument();
  179. doc.LoadHtml(htmlResource);
  180. //获取li下面所有a标签
  181. HtmlNodeCollection nodeList = doc.DocumentNode.SelectNodes("//*[@class='listContainer']/ul/li/a");
  182. if (nodeList == null) return result;
  183. List<string> urls = new List<string>();
  184. //遍历a标签
  185. foreach (HtmlNode node in nodeList)
  186. {
  187. HtmlAttribute attr = node.Attributes.SingleOrDefault(a => a.Name.Equals("href"));
  188. if (attr != null)
  189. {
  190. string href = HostBackUrl + attr.Value;
  191. //去重
  192. if (!urls.Contains(href))
  193. {
  194. urls.Add(href);
  195. }
  196. }
  197. }
  198. foreach (var url1 in urls)
  199. {
  200. var LotterySkill = GetSkillModelBackUrl(url1);
  201. result.Add(LotterySkill);
  202. }
  203. }
  204. catch (Exception ex)
  205. {
  206. log.Error(GetType(),
  207. 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));
  208. }
  209. return result;
  210. }
  211. /// <summary>
  212. /// 根据副站url获取技巧详情
  213. /// </summary>
  214. /// <param name="url"></param>
  215. /// <returns></returns>
  216. private LotterySkillModel GetSkillModelBackUrl(string url)
  217. {
  218. LotterySkillModel lotterySkill = new LotterySkillModel();
  219. try
  220. {
  221. var htmlResource = NetHelper.GetUrlResponse(url, Encoding.GetEncoding("utf-8"));
  222. if (htmlResource == null) return lotterySkill;
  223. HtmlDocument doc = new HtmlDocument();
  224. doc.LoadHtml(htmlResource);
  225. //获取li下面所有a标签
  226. var div = doc.DocumentNode.SelectSingleNode("//*[@class='article mb-10']");
  227. var Title = div.ChildNodes.Where(node => node.Name == "h5").ToList();
  228. var div1 = div.ChildNodes.Where(node => node.Name == "div").ToList();
  229. lotterySkill.Title = Title[0].InnerText.Trim();
  230. lotterySkill.Author = "cn55128";
  231. lotterySkill.Content = div1[2].InnerHtml.Trim();
  232. lotterySkill.IsDelete = false;
  233. lotterySkill.SourceUrl = url.ToString();
  234. lotterySkill.TypeId = lotterySkillType;
  235. lotterySkill.TypeName = lotterySkillType.GetEnumDescription();
  236. }
  237. catch (Exception ex)
  238. {
  239. log.Error(GetType(),
  240. 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));
  241. }
  242. return lotterySkill;
  243. }
  244. /// <summary>
  245. /// 根据主站url获取技巧详情
  246. /// </summary>
  247. /// <param name="url"></param>
  248. /// <returns></returns>
  249. private LotterySkillModel GetSkillModel(string url)
  250. {
  251. LotterySkillModel lotterySkill = new LotterySkillModel();
  252. try
  253. {
  254. var htmlResource = NetHelper.GetUrlResponse(url, Encoding.GetEncoding("utf-8"));
  255. if (htmlResource == null) return lotterySkill;
  256. HtmlDocument doc = new HtmlDocument();
  257. doc.LoadHtml(htmlResource);
  258. //获取li下面所有a标签
  259. var div = doc.DocumentNode.SelectSingleNode("//*[@class='artile']");
  260. var Title = div.ChildNodes.Where(node => node.Name == "h1").ToList();
  261. var div1 = div.ChildNodes.Where(node => node.Name == "div").ToList();
  262. lotterySkill.Title = Title[0].InnerText.Trim();
  263. lotterySkill.Author = "cn55128";
  264. lotterySkill.Content = div1[1].InnerHtml.Trim();
  265. lotterySkill.IsDelete = false;
  266. lotterySkill.SourceUrl = url.ToString();
  267. lotterySkill.TypeId = lotterySkillType;
  268. lotterySkill.TypeName = lotterySkillType.GetEnumDescription();
  269. }
  270. catch (Exception ex)
  271. {
  272. log.Error(GetType(),
  273. 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));
  274. }
  275. return lotterySkill;
  276. }
  277. #region Attribute
  278. /// <summary>
  279. /// 主机地址
  280. /// </summary>
  281. public string Host = "http://www.55125.cn/";
  282. /// <summary>
  283. /// 副站地址
  284. /// </summary>
  285. public string HostBackUrl = "https://www.cz89.com/";
  286. /// <summary>
  287. /// 配置信息
  288. /// </summary>
  289. private SCCConfig Config;
  290. #pragma warning disable CS0414 // 字段“SSqSkillJob.LatestItem”已被赋值,但从未使用过它的值
  291. /// <summary>
  292. /// 当天抓取的最新一期开奖记录
  293. /// </summary>
  294. private LotterySkillModel LatestItem = null;
  295. #pragma warning restore CS0414 // 字段“SSqSkillJob.LatestItem”已被赋值,但从未使用过它的值
  296. #pragma warning disable CS0414 // 字段“SSqSkillJob.FailedQiHaoList”已被赋值,但从未使用过它的值
  297. /// <summary>
  298. /// 当天抓取失败列表
  299. /// </summary>
  300. private List<string> FailedQiHaoList = null;
  301. #pragma warning restore CS0414 // 字段“SSqSkillJob.FailedQiHaoList”已被赋值,但从未使用过它的值
  302. /// <summary>
  303. /// 日志对象
  304. /// </summary>
  305. private readonly LogHelper log;
  306. /// <summary>
  307. /// 数据服务
  308. /// </summary>
  309. private readonly IDTOpenCode services;
  310. /// <summary>
  311. /// 当前彩种
  312. /// </summary>
  313. private SCCLottery currentLottery => SCCLottery.LotterySkill;
  314. /// <summary>
  315. /// 福彩3D技巧
  316. /// </summary>
  317. private LotterySkillType lotterySkillType = LotterySkillType.SSQ;
  318. /// <summary>
  319. /// 邮件接口
  320. /// </summary>
  321. private IEmail email;
  322. #pragma warning disable CS0414 // 字段“SSqSkillJob.isGetData”已被赋值,但从未使用过它的值
  323. /// <summary>
  324. /// 是否本次运行抓取到开奖数据
  325. /// </summary>
  326. private bool isGetData = false;
  327. #pragma warning restore CS0414 // 字段“SSqSkillJob.isGetData”已被赋值,但从未使用过它的值
  328. #endregion
  329. }
  330. }