SHSSLJob.cs 18 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407
  1. using System;
  2. using System.Collections.Generic;
  3. using System.Linq;
  4. using System.Text;
  5. using System.Xml;
  6. using Quartz;
  7. using SCC.Common;
  8. using SCC.Models;
  9. using SCC.Interface;
  10. using HtmlAgilityPack;
  11. namespace SCC.Crawler.GP
  12. {
  13. /// <summary>
  14. /// 数据爬取类
  15. /// 上海时时乐
  16. /// </summary>
  17. [DisallowConcurrentExecution]
  18. [PersistJobDataAfterExecution]
  19. public class SHSSLJob : IJob
  20. {
  21. /// <summary>
  22. /// 构造函数
  23. /// </summary>
  24. public SHSSLJob()
  25. {
  26. log = new LogHelper();
  27. services = IOC.Resolve<IOpen3Code>();
  28. email = IOC.Resolve<IEmail>();
  29. }
  30. /// <summary>
  31. /// 作业执行入口
  32. /// </summary>
  33. /// <param name="context">作业执行上下文</param>
  34. public void Execute(IJobExecutionContext context)
  35. {
  36. Config = CommonHelper.GetConfigFromDataMap(context.JobDetail.JobDataMap);
  37. //预设节假日不开奖
  38. if (Config.SkipDate.Contains(CommonHelper.SCCSysDateTime.ToString("yyyyMMdd"))) return;
  39. LatestQiHao = context.JobDetail.JobDataMap.GetString("LatestQiHao");
  40. try
  41. {
  42. //服务启动时配置初始数据
  43. if (string.IsNullOrEmpty(LatestQiHao))
  44. {
  45. var lastItem = services.GetLastItem(currentLottery);
  46. if (lastItem != null)
  47. {
  48. LatestQiHao = lastItem.Term.ToString();
  49. }
  50. }
  51. //第一次启动服务或最新期号为昨天的开奖期号,则自检昨天开奖数据是否抓取完毕(否则插入邮件数据),并重置当天期号和失败列表
  52. if (string.IsNullOrEmpty(LatestQiHao) || !LatestQiHao.StartsWith(CommonHelper.SCCSysDateTime.ToString("yyMMdd")))
  53. {
  54. CheckingYesterdayTheLotteryData();
  55. LatestQiHao = CommonHelper.GenerateTodayQiHaoYYMMDDQQ(0);
  56. }
  57. //当最新期号不符合当天总期数,执行当天作业
  58. if (Convert.ToInt32(LatestQiHao.Substring(6)) != Config.TimesPerDay)
  59. {
  60. DoTodayJobByMainUrl();
  61. DoTodayJobByBackUrl();
  62. }
  63. }
  64. catch (Exception ex)
  65. {
  66. log.Error(typeof(SHSSLJob), string.Format("【{0}】抓取时发生错误,错误信息【{1}】", Config.Area + Config.LotteryName, ex.Message));
  67. }
  68. //保存最新期号和失败期号列表
  69. context.JobDetail.JobDataMap["LatestQiHao"] = LatestQiHao;
  70. }
  71. /// <summary>
  72. /// 自检昨天开奖数据
  73. /// </summary>
  74. private void CheckingYesterdayTheLotteryData()
  75. {
  76. if (Config.SkipDate.Contains(CommonHelper.SCCSysDateTime.AddDays(-1).ToString("yyyyMMdd"))) return;//如果昨日设定不开奖则不自检昨日开奖数据
  77. //从数据库中获取昨天数据抓取失败列表
  78. FailedQiHaoList = services.GetYesterdayFailQQList(currentLottery, Config.TimesPerDay);
  79. if (FailedQiHaoList.Count > 0)
  80. {
  81. DoYesterdayFailedListByMainUrl();
  82. DoYesterdayFailedListByBackUrl();
  83. foreach (var fQiHao in FailedQiHaoList)
  84. {
  85. //将抓取失败数据推送至邮件列表,待邮件服务发送至配置管理员的邮箱中
  86. if (email.AddEmail(Config.Area + Config.LotteryName, fQiHao, CommonHelper.GenerateYesterdayOpenTime(Config, fQiHao)))
  87. log.Error(typeof(SHSSLJob), CommonHelper.GetJobLogError(Config, fQiHao));
  88. }
  89. }
  90. }
  91. /// <summary>
  92. /// 通过主站点抓取开奖数据
  93. /// (上海福彩)
  94. /// </summary>
  95. private void DoTodayJobByMainUrl()
  96. {
  97. if (!string.IsNullOrEmpty(Config.MainUrl))
  98. {
  99. var OpenList = GetTodayOpenListFromMainUrl();
  100. if (OpenList.Count == 0) return;//无抓取数据
  101. var newestQiHao = OpenList.First().Key;
  102. var startQiNum = Convert.ToInt32(LatestQiHao.Substring(6)) + 1;
  103. var newestQiNum = Convert.ToInt32(newestQiHao.Substring(6));
  104. if (startQiNum > newestQiNum) return;//无最新数据
  105. //处理最新开奖数据
  106. string getQiHao = string.Empty;
  107. for (var i = startQiNum; i <= newestQiNum; i++)
  108. {
  109. getQiHao = CommonHelper.GenerateTodayQiHaoYYMMDDQQ(i);
  110. var matchItem = OpenList.Where(R => R.Key == getQiHao).FirstOrDefault();
  111. if (matchItem.Key != null && SaveRecord(getQiHao, matchItem.Value, false))
  112. {
  113. //处理成功写入日志
  114. log.Info(typeof(SHSSLJob), CommonHelper.GetJobMainLogInfo(Config, getQiHao));
  115. LatestQiHao = getQiHao;
  116. }
  117. }
  118. }
  119. }
  120. /// <summary>
  121. /// 通过主站抓取错误期号列表中每一个期号
  122. /// (上海福彩)
  123. /// </summary>
  124. private void DoYesterdayFailedListByMainUrl()
  125. {
  126. if (!string.IsNullOrEmpty(Config.MainUrl) && FailedQiHaoList.Count > 0)
  127. {
  128. var OpenList = GetYesterdayOpenListFromMainUrl();
  129. if (OpenList.Count == 0) return;//无抓取数据
  130. var SuccessList = new List<string>();
  131. foreach (string failedQiHao in FailedQiHaoList)
  132. {
  133. var matchItem = OpenList.Where(R => R.Key == failedQiHao).FirstOrDefault();
  134. if (matchItem.Key != null && SaveRecord(failedQiHao, matchItem.Value, true))
  135. {
  136. //处理成功写入日志
  137. log.Info(typeof(SHSSLJob), CommonHelper.GetJobMainLogInfo(Config, failedQiHao));
  138. SuccessList.Add(failedQiHao);
  139. continue;
  140. }
  141. }
  142. foreach (var successQiHao in SuccessList)
  143. {
  144. FailedQiHaoList.Remove(successQiHao);
  145. }
  146. }
  147. }
  148. /// <summary>
  149. /// 通过主站点抓取今日最新开奖数据
  150. /// </summary>
  151. /// <returns></returns>
  152. private Dictionary<string, string> GetTodayOpenListFromMainUrl()
  153. {
  154. Dictionary<string, string> result = new Dictionary<string, string>();
  155. try
  156. {
  157. var HtmlResource = NetHelper.GetUrlResponse(Config.MainUrl);
  158. if (!string.IsNullOrWhiteSpace(HtmlResource))
  159. {
  160. HtmlDocument doc = new HtmlDocument();
  161. doc.LoadHtml(HtmlResource);
  162. var divs = doc.DocumentNode.SelectNodes("//div").ToList();
  163. var targetDiv = divs.Where(R => R.Attributes["class"] != null && R.Attributes["class"].Value == "lm2block1").FirstOrDefault();
  164. if (targetDiv == null) return result;
  165. var childDivs = targetDiv.ChildNodes.Where(R => R.Name == "div").ToList();
  166. if (childDivs.Count >= 4 && (childDivs.Count - 1) % 3 == 0)//第一个div为表头,后面每3个div一期开奖记录
  167. {
  168. string matchQiHao = string.Empty;
  169. string matchKJHaoMa = string.Empty;
  170. for (var i = 0; i < childDivs.Count / 3; i++)
  171. {
  172. matchQiHao = childDivs[i * 3 + 1].InnerText.Replace("-", string.Empty).Substring(2);
  173. matchKJHaoMa = childDivs[i * 3 + 3].InnerText.Replace(" ", string.Empty).Insert(2, ",").Insert(1, ",");
  174. if (!result.ContainsKey(matchQiHao))
  175. result.Add(matchQiHao, matchKJHaoMa);
  176. }
  177. }
  178. }
  179. }
  180. catch (Exception ex)
  181. {
  182. log.Error(typeof(SHSSLJob), string.Format("【{0}】通过主站点抓取今日最新开奖列表时发生错误,错误信息【{1}】", Config.Area + Config.LotteryName, ex.Message));
  183. }
  184. return result;
  185. }
  186. /// <summary>
  187. /// 通过主站点抓取昨日开奖数据
  188. /// </summary>
  189. /// <returns></returns>
  190. private Dictionary<string, string> GetYesterdayOpenListFromMainUrl()
  191. {
  192. Dictionary<string, string> result = new Dictionary<string, string>();
  193. try
  194. {
  195. var HtmlResource = NetHelper.GetUrlResponse(Config.MainUrl, "POST", string.Format("dateOptions={0}", CommonHelper.SCCSysDateTime.AddDays(-1).ToString("yyyy-MM-dd")));
  196. if (!string.IsNullOrWhiteSpace(HtmlResource))
  197. {
  198. HtmlDocument doc = new HtmlDocument();
  199. doc.LoadHtml(HtmlResource);
  200. var divs = doc.DocumentNode.SelectNodes("//div").ToList();
  201. var targetDiv = divs.Where(R => R.Attributes["class"] != null && R.Attributes["class"].Value == "lm2block1").FirstOrDefault();
  202. if (targetDiv == null) return result;
  203. var childDivs = targetDiv.ChildNodes.Where(R => R.Name == "div").ToList();
  204. if (childDivs.Count >= 4 && (childDivs.Count - 1) % 3 == 0)//第一个div为表头,后面每3个div一期开奖记录
  205. {
  206. string matchQiHao = string.Empty;
  207. string matchKJHaoMa = string.Empty;
  208. for (var i = 0; i < childDivs.Count / 3; i++)
  209. {
  210. matchQiHao = childDivs[i * 3 + 1].InnerText.Replace("-", string.Empty).Substring(2);
  211. matchKJHaoMa = childDivs[i * 3 + 3].InnerText.Replace(" ", string.Empty).Insert(2, ",").Insert(1, ",");
  212. if (!result.ContainsKey(matchQiHao))
  213. result.Add(matchQiHao, matchKJHaoMa);
  214. }
  215. }
  216. }
  217. }
  218. catch (Exception ex)
  219. {
  220. log.Error(typeof(SHSSLJob), string.Format("【{0}】通过主站点抓取昨日开奖列表时发生错误,错误信息【{1}】", Config.Area + Config.LotteryName, ex.Message));
  221. }
  222. return result;
  223. }
  224. /// <summary>
  225. /// 通过备用站点抓取开奖数据
  226. /// (500彩票网)
  227. /// </summary>
  228. private void DoTodayJobByBackUrl()
  229. {
  230. if (!string.IsNullOrEmpty(Config.BackUrl))
  231. {
  232. var OpenList = GetTodayOpenListFromBackUrl();
  233. if (OpenList.Count == 0) return;//无抓取数据
  234. var newestQiHao = OpenList.First().Key;
  235. var startQiNum = Convert.ToInt32(LatestQiHao.Substring(6)) + 1;
  236. var newestQiNum = Convert.ToInt32(newestQiHao.Substring(6));
  237. if (startQiNum > newestQiNum) return;//无最新数据
  238. //处理最新开奖数据
  239. var getQiHao = string.Empty;
  240. for (var i = startQiNum; i <= newestQiNum; i++)
  241. {
  242. getQiHao = CommonHelper.GenerateTodayQiHaoYYMMDDQQ(i);
  243. var matchItem = OpenList.Where(R => R.Key == getQiHao).FirstOrDefault();
  244. if (matchItem.Key != null && SaveRecord(getQiHao, matchItem.Value, false))
  245. {
  246. //处理成功写入日志
  247. log.Info(typeof(SHSSLJob), CommonHelper.GetJobBackLogInfo(Config, getQiHao));
  248. LatestQiHao = getQiHao;
  249. }
  250. }
  251. }
  252. }
  253. /// <summary>
  254. /// 通过备用地址抓取错误期号列表中每一个期号
  255. /// (500彩票网)
  256. /// </summary>
  257. private void DoYesterdayFailedListByBackUrl()
  258. {
  259. if (!string.IsNullOrEmpty(Config.BackUrl) && FailedQiHaoList.Count > 0)
  260. {
  261. var OpenList = GetYesterdayOpenListFromBackUrl();
  262. if (OpenList.Count == 0) return;//无抓取数据
  263. var SuccessList = new List<string>();
  264. foreach (var failedQiHao in FailedQiHaoList)
  265. {
  266. var matchItem = OpenList.Where(R => R.Key == failedQiHao).FirstOrDefault();
  267. if (matchItem.Key != null && SaveRecord(failedQiHao, matchItem.Value, true))
  268. {
  269. //处理成功写入日志
  270. log.Info(typeof(SHSSLJob), CommonHelper.GetJobBackLogInfo(Config, failedQiHao));
  271. SuccessList.Add(failedQiHao);
  272. continue;
  273. }
  274. }
  275. foreach (var successQiHao in SuccessList)
  276. {
  277. FailedQiHaoList.Remove(successQiHao);
  278. }
  279. }
  280. }
  281. /// <summary>
  282. /// 通过备用站点抓取今日最新开奖列表
  283. /// </summary>
  284. /// <returns></returns>
  285. private Dictionary<string, string> GetTodayOpenListFromBackUrl()
  286. {
  287. var requestUrl = string.Format(Config.BackUrl, CommonHelper.SCCSysDateTime.ToString("yyyyMMdd"));
  288. return GetOpenListFromBackUrl(requestUrl);
  289. }
  290. /// <summary>
  291. /// 通过备用站点抓取昨日开奖列表
  292. /// </summary>
  293. /// <returns></returns>
  294. private Dictionary<string, string> GetYesterdayOpenListFromBackUrl()
  295. {
  296. var requestUrl = string.Format(Config.BackUrl, CommonHelper.SCCSysDateTime.AddDays(-1).ToString("yyyyMMdd"));
  297. return GetOpenListFromBackUrl(requestUrl);
  298. }
  299. /// <summary>
  300. /// 抓取备用站点开奖数据
  301. /// </summary>
  302. /// <param name="url">备用站点</param>
  303. /// <returns></returns>
  304. private Dictionary<string, string> GetOpenListFromBackUrl(string url)
  305. {
  306. Dictionary<string, string> result = new Dictionary<string, string>();
  307. try
  308. {
  309. var HtmlResource = NetHelper.GetUrlResponse(url, Encoding.GetEncoding("gb2312"));
  310. if (!string.IsNullOrWhiteSpace(HtmlResource))
  311. {
  312. XmlDocument doc = new XmlDocument();
  313. doc.LoadXml(HtmlResource);
  314. string matchQiHao = string.Empty;
  315. string matchKJHaoMa = string.Empty;
  316. var records = doc.SelectNodes("//row");
  317. foreach (XmlNode xmlnode in records)
  318. {
  319. if (xmlnode.Attributes["expect"] == null || string.IsNullOrWhiteSpace(xmlnode.Attributes["expect"].Value) || xmlnode.Attributes["opencode"] == null || string.IsNullOrWhiteSpace(xmlnode.Attributes["opencode"].Value)) continue;
  320. matchQiHao = xmlnode.Attributes["expect"].Value.Replace("-", string.Empty).Substring(2);
  321. matchKJHaoMa = xmlnode.Attributes["opencode"].Value.Trim();
  322. if (!result.ContainsKey(matchQiHao))
  323. result.Add(matchQiHao, matchKJHaoMa);
  324. }
  325. }
  326. }
  327. catch (Exception ex)
  328. {
  329. log.Error(typeof(SHSSLJob), string.Format("【{0}】通过备用站点抓取开奖列表时发生错误,错误信息【{1}】", Config.Area + Config.LotteryName, ex.Message));
  330. }
  331. return result;
  332. }
  333. /// <summary>
  334. /// 将此彩种指定期号和开奖号码保存至数据库
  335. /// </summary>
  336. /// <param name="QiHao">期号</param>
  337. /// <param name="OpenCode">开奖号码(形如01,02,03,04,05)</param>
  338. /// <param name="IsYesterdayRecord">是否是保存昨天的记录</param>
  339. /// <returns></returns>
  340. private bool SaveRecord(string QiHao, string OpenCode, bool IsYesterdayRecord)
  341. {
  342. if (!string.IsNullOrWhiteSpace(QiHao) && !string.IsNullOrWhiteSpace(OpenCode))
  343. {
  344. OpenCode3Model model = new OpenCode3Model();
  345. model.Term = Convert.ToInt64(QiHao);
  346. var haoMaArray = OpenCode.Split(new char[] { ',' });
  347. model.OpenCode1 = Convert.ToInt32(haoMaArray[0]);
  348. model.OpenCode2 = Convert.ToInt32(haoMaArray[1]);
  349. model.OpenCode3 = Convert.ToInt32(haoMaArray[2]);
  350. if (IsYesterdayRecord)
  351. model.OpenTime = CommonHelper.GenerateYesterdayOpenTime(Config, QiHao);
  352. else
  353. model.OpenTime = CommonHelper.GenerateTodayOpenTime(Config, QiHao);
  354. return services.AddOpen3Code(currentLottery, model);
  355. }
  356. return false;
  357. }
  358. #region Attribute
  359. /// <summary>
  360. /// 配置信息
  361. /// </summary>
  362. private SCCConfig Config = null;
  363. /// <summary>
  364. /// 当天抓取的最新一期期号
  365. /// </summary>
  366. private string LatestQiHao = null;
  367. /// <summary>
  368. /// 当天抓取失败列表
  369. /// </summary>
  370. private List<string> FailedQiHaoList = null;
  371. /// <summary>
  372. /// 日志对象
  373. /// </summary>
  374. private LogHelper log = null;
  375. /// <summary>
  376. /// 数据服务
  377. /// </summary>
  378. private IOpen3Code services = null;
  379. /// <summary>
  380. /// 当前彩种
  381. /// </summary>
  382. private SCCLottery currentLottery
  383. {
  384. get
  385. {
  386. return SCCLottery.ShangHaiSSL;
  387. }
  388. }
  389. /// <summary>
  390. /// 邮件接口
  391. /// </summary>
  392. private IEmail email = null;
  393. #endregion
  394. }
  395. }