using System; using System.Collections.Generic; using System.Linq; using System.Text; using System.Text.RegularExpressions; using Quartz; using SCC.Common; using SCC.Models; using SCC.Interface; using Newtonsoft.Json; using HtmlAgilityPack; namespace SCC.Crawler.DT { /// /// 数据爬取类 /// 湖北30选5 /// [DisallowConcurrentExecution] [PersistJobDataAfterExecution] public class HuBei30X5Job : IJob { /// /// 构造函数 /// public HuBei30X5Job() { log = new LogHelper(); services = IOC.Resolve(); email = IOC.Resolve(); } /// /// 作业执行入口 /// /// 作业执行上下文 public void Execute(IJobExecutionContext context) { Config = CommonHelper.GetConfigFromDataMap(context.JobDetail.JobDataMap); //预设节假日不开奖 if (Config.SkipDate.Contains(CommonHelper.SCCSysDateTime.ToString("yyyyMMdd"))) return; LatestItem = context.JobDetail.JobDataMap["LatestItem"] as OpenCode5DTModel; try { //服务启动时配置初始数据 if (LatestItem == null) { LatestItem = services.GetOpenCode5DTLastItem(currentLottery); if (LatestItem == null) { //服务第一次启动,数据库一条数据都没有时 LatestItem = new OpenCode5DTModel() { Term = CommonHelper.GenerateQiHaoYYQQQ(0), OpenTime = new DateTime(CommonHelper.SCCSysDateTime.Year, 1, 1) }; } } //程序时间第二天,程序根据配置检查是否昨天有开奖 isGetData = false; if (CommonHelper.CheckDTIsNeedGetData(Config)) { CheckingOpenDayTheLotteryData(); } if (!LatestItem.Term.ToString().StartsWith(CommonHelper.SCCSysDateTime.ToString("yy"))) { //翻年时 LatestItem = new OpenCode5DTModel() { Term = CommonHelper.GenerateQiHaoYYQQQ(0), OpenTime = new DateTime(CommonHelper.SCCSysDateTime.Year, 1, 1) }; } //当今日开奖并且当前时间是晚上8点过后开始抓取 if (CommonHelper.CheckTodayIsOpenDay(Config) && CommonHelper.SCCSysDateTime.Hour > 12) { DoTodayJobByMainUrl(); DoTodayJobByBackUrl(); } if (isGetData) TrendChartHelper.GenerateHuBei30X5TrendChart(log); } catch (Exception ex) { log.Error(typeof(HuBei30X5Job), string.Format("【{0}】抓取时发生错误,错误信息【{1}】", Config.Area + Config.LotteryName, ex.Message)); } //保存最新期号 context.JobDetail.JobDataMap["LatestItem"] = LatestItem; } /// /// 自检爬取未爬取到的开奖数据,并对昨日开奖但未爬取到开奖数据的彩种添加邮件提醒 /// private void CheckingOpenDayTheLotteryData() { //从数据库中获取昨天数据抓取失败列表 FailedQiHaoList = services.GetFailedYYQQQList(currentLottery); if (FailedQiHaoList.Count > 0) { DoYesterdayFailedListByMainUrl(); DoYesterdayFailedListByBackUrl(); foreach (var fQiHao in FailedQiHaoList) { //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 if (email.AddEmail(Config.Area + Config.LotteryName, fQiHao, CommonHelper.GenerateDTOpenTime(Config))) log.Error(typeof(HuBei30X5Job), CommonHelper.GetJobLogError(Config, fQiHao)); } } if (LatestItem.OpenTime.ToString("yyyyMMdd") != CommonHelper.SCCSysDateTime.AddDays(-1).ToString("yyyyMMdd")) { //开奖时间(昨天)未抓取到最新开奖数据,则再抓取一次,若还不成功则写入待发送邮件列表 DoTodayJobByMainUrl(); DoTodayJobByBackUrl(); if (LatestItem.OpenTime.ToString("yyyyMMdd") != CommonHelper.SCCSysDateTime.AddDays(-1).ToString("yyyyMMdd")) { var openQiHao = (LatestItem.Term + 1).ToString(); if (email.AddEmail(Config.Area + Config.LotteryName, openQiHao, CommonHelper.GenerateDTOpenTime(Config))) log.Error(typeof(HuBei30X5Job), CommonHelper.GetJobLogError(Config, openQiHao)); } } } /// /// 通过主站点爬取开奖数据 /// (湖北福彩网) /// private void DoTodayJobByMainUrl() { if (!string.IsNullOrEmpty(Config.MainUrl)) { var OpenList = GetOpenListFromMainUrl(); if (OpenList.Count == 0) return;//无抓取数据 var newestQiHao = OpenList.First().Term.ToString(); var startQiNum = Convert.ToInt32(LatestItem.Term.ToString().Substring(2)) + 1; var newestQiNum = Convert.ToInt32(newestQiHao.Substring(2)); if (startQiNum > newestQiNum) return;//无最新数据 //处理最新开奖数据 string getQiHao = string.Empty; OpenCode5DTModel matchItem = null; for (var i = startQiNum; i <= newestQiNum; i++) { getQiHao = LatestItem.Term.ToString().Substring(0, 2) + i.ToString().PadLeft(3, '0'); matchItem = OpenList.Where(R => R.Term.ToString() == getQiHao).FirstOrDefault(); if (matchItem != null && OptimizeMainModel(ref matchItem) && services.AddDTOpen5Code(currentLottery, matchItem)) { //Do Success Log log.Info(typeof(HuBei30X5Job), CommonHelper.GetJobMainLogInfo(Config, getQiHao)); LatestItem = matchItem; isGetData = true; } } } } /// /// 通过主站爬取错误期号列表中每一个期号 /// (湖北福彩网) /// private void DoYesterdayFailedListByMainUrl() { if (!string.IsNullOrEmpty(Config.MainUrl) && FailedQiHaoList.Count > 0) { var OpenList = GetOpenListFromMainUrl(); if (OpenList.Count == 0) return;//无抓取数据 OpenCode5DTModel matchItem = null; var SuccessList = new List(); foreach (string failedQiHao in FailedQiHaoList) { matchItem = OpenList.Where(R => R.Term.ToString() == failedQiHao).FirstOrDefault(); if (matchItem != null && OptimizeMainModel(ref matchItem) && services.AddDTOpen5Code(currentLottery, matchItem)) { //Do Success Log log.Info(typeof(HuBei30X5Job), CommonHelper.GetJobMainLogInfo(Config, failedQiHao)); if (matchItem.Term > LatestItem.Term) { LatestItem = matchItem; } isGetData = true; SuccessList.Add(failedQiHao); } } foreach (var successQiHao in SuccessList) { FailedQiHaoList.Remove(successQiHao); } } } /// /// 获取主站开奖列表数据 /// /// private List GetOpenListFromMainUrl() { List result = new List(); try { var htmlResource = NetHelper.GetUrlResponse(Config.MainUrl); if (!string.IsNullOrWhiteSpace(htmlResource)) { HtmlDocument doc = new HtmlDocument(); doc.LoadHtml(htmlResource); var select = doc.GetElementbyId("T30Xuan5"); if (select == null) return result; var options = select.ChildNodes.Where(N => N.Name.ToLower() == "option").ToList(); string term = string.Empty; foreach (var option in options) { term = option.NextSibling.InnerText.Trim(); if (!term.StartsWith(CommonHelper.SCCSysDateTime.ToString("yy"))) break; result.Add(new OpenCode5DTModel() { Term = Convert.ToInt64(term) }); } } } catch (Exception ex) { log.Error(typeof(HuBei30X5Job), string.Format("【{0}】通过主站点抓取开奖列表时发生错误,错误信息【{1}】", Config.Area + Config.LotteryName, ex.Message)); } return result; } /// /// 完善主站江苏体彩7位数开奖详情信息 /// /// private bool OptimizeMainModel(ref OpenCode5DTModel model) { try { var requestUrl = string.Format("http://stats.hbfcw.gov.cn/index.php/Index/show_result/type/5/qi/{0}", model.Term); var htmlResource = NetHelper.GetUrlResponse(requestUrl); if (!string.IsNullOrEmpty(htmlResource)) { HtmlDocument doc = new HtmlDocument(); doc.LoadHtml(htmlResource); var openCodes = doc.GetElementbyId("HaoMa").InnerText.Replace("开奖号码:", "").Trim().Split(new char[] { ',' }); model.OpenCode1 = Convert.ToInt32(openCodes[0]); model.OpenCode2 = Convert.ToInt32(openCodes[1]); model.OpenCode3 = Convert.ToInt32(openCodes[2]); model.OpenCode4 = Convert.ToInt32(openCodes[3]); model.OpenCode5 = Convert.ToInt32(openCodes[4]); model.OpenTime = Convert.ToDateTime(doc.GetElementbyId("KaiJiangRi").InnerText.Trim()); var trs = doc.GetElementbyId("GridView1").ChildNodes.Where(N => N.Name.ToLower() == "tr").ToList(); List tds = null; int Level1Num = 0, Level2Num = 0, Level3Num = 0; decimal Level1Money = 0, Level2Money = 0, Level3Money = 0, Sales = 0, Jackpot = 0; for (var i = 1; i < trs.Count; i++)//第一行为表头 { tds = trs[i].ChildNodes.Where(N => N.Name.ToLower() == "td").ToList(); if (tds[0].InnerText.Trim() == "一等奖") { Level1Num = Convert.ToInt32(tds[1].InnerText.Trim()); Level1Money = Convert.ToDecimal(tds[2].InnerText.Trim()); } if (tds[0].InnerText.Trim() == "二等奖") { Level2Num = Convert.ToInt32(tds[1].InnerText.Trim()); Level2Money = Convert.ToDecimal(tds[2].InnerText.Trim()); } if (tds[0].InnerText.Trim() == "三等奖") { Level3Num = Convert.ToInt32(tds[1].InnerText.Trim()); Level3Money = Convert.ToDecimal(tds[2].InnerText.Trim()); } } var reg = new Regex(@"湖北投注金额为:([\d.,]*?)元"); var m = reg.Match(htmlResource); if (m.Success) Sales = Convert.ToDecimal(m.Result("$1")); reg = new Regex(@"奖池累积金额为:([\d.,]*?)元"); m = reg.Match(htmlResource); if (m.Success) Jackpot = Convert.ToDecimal(m.Result("$1")); model.Detail = string.Format("{0},{1}^一等奖|{2}|{3},二等奖|{4}|{5},三等奖|{6}|{7}", Sales, Jackpot, Level1Num, Level1Money, Level2Num, Level2Money, Level3Num, Level3Money); return true; } } catch (Exception ex) { log.Error(typeof(HuBei30X5Job), string.Format("【{0}】通过主站点优化开奖列表时发生错误,错误信息【{1}】", Config.Area + Config.LotteryName, ex.Message)); } return false; } /// /// 通过备用站点抓取开奖数据 /// (彩票两元网) /// private void DoTodayJobByBackUrl() { if (!string.IsNullOrEmpty(Config.BackUrl)) { var OpenList = GetOpenListFromBackUrl(); if (OpenList.Count == 0) return;//无抓取数据 var newestQiHao = OpenList.First().Term.ToString(); var startQiNum = Convert.ToInt32(LatestItem.Term.ToString().Substring(2)) + 1; var newestQiNum = Convert.ToInt32(newestQiHao.Substring(2)); if (startQiNum > newestQiNum) return;//无最新数据 //处理最新开奖数据 var getQiHao = string.Empty; OpenCode5DTModel matchItem = null; for (var i = startQiNum; i <= newestQiNum; i++) { //Modify By 大师兄 2018年以后,期数上加了20 getQiHao = LatestItem.Term.ToString().Substring(0, 2) + i.ToString().PadLeft(3, '0'); matchItem = OpenList.Where(R => R.Term.ToString() == getQiHao).FirstOrDefault(); if (matchItem != null && OptimizeBackModel(ref matchItem) && services.AddDTOpen5Code(currentLottery, matchItem)) { //Do Success Log log.Info(typeof(HuBei30X5Job), CommonHelper.GetJobBackLogInfo(Config, getQiHao)); LatestItem = matchItem; isGetData = true; } } } } /// /// 通过备用地址抓取错误期号列表中每一个期号 /// (彩票两元网) /// private void DoYesterdayFailedListByBackUrl() { if (!string.IsNullOrEmpty(Config.BackUrl) && FailedQiHaoList.Count > 0) { var OpenList = GetOpenListFromBackUrl(); if (OpenList.Count == 0) return;//无抓取数据 OpenCode5DTModel matchItem = null; var SuccessList = new List(); foreach (string failedQiHao in FailedQiHaoList) { matchItem = OpenList.Where(R => R.Term.ToString() == failedQiHao).FirstOrDefault(); if (matchItem != null && OptimizeBackModel(ref matchItem) && services.AddDTOpen5Code(currentLottery, matchItem)) { //Do Success Log log.Info(typeof(HuBei30X5Job), CommonHelper.GetJobBackLogInfo(Config, failedQiHao)); if (matchItem.Term > LatestItem.Term) { LatestItem = matchItem; } SuccessList.Add(failedQiHao); isGetData = true; } } foreach (var successQiHao in SuccessList) { FailedQiHaoList.Remove(successQiHao); } } } /// /// 获取备用站点开奖列表数据 /// /// private List GetOpenListFromBackUrl() { List result = new List(); try { Uri url = new Uri(Config.BackUrl); var htmlResource = NetHelper.GetUrlResponse(Config.BackUrl, Encoding.GetEncoding("gb2312")); if (string.IsNullOrWhiteSpace(htmlResource)) return result; HtmlDocument doc = new HtmlDocument(); doc.LoadHtml(htmlResource); var table = doc.DocumentNode.SelectSingleNode("//table"); if (table == null) return result; var trs = table.ChildNodes.Where(node => node.Name == "tr").ToList(); OpenCode5DTModel model = null; HtmlNode nodeA = null; string optimizeUrl = string.Empty; for (var i = 2; i < trs.Count; i++)//第一二行为表头 { var trstyle = trs[i].Attributes["style"]; if (trstyle != null && trstyle.Value == "display:none") { continue; } var tds = trs[i].ChildNodes.Where(node => node.Name == "td").ToList(); if (tds.Count < 10) continue; model = new OpenCode5DTModel(); nodeA = tds[0].ChildNodes.Where(n => n.Name == "a").FirstOrDefault(); if (nodeA == null) continue; model.Term = Convert.ToInt64(nodeA.InnerText.Trim().Remove(0,1)); optimizeUrl = nodeA.Attributes["href"].Value; model.DetailUrl = new Uri(url, optimizeUrl).AbsoluteUri; model.OpenTime = Convert.ToDateTime(tds[9].InnerText); if (tds[1].ChildNodes.Count == 0) continue; var opencodeNode = tds[1].ChildNodes.Where(n => n.Name.ToLower() == "i").ToList(); if (opencodeNode.Count < 5) continue; model.OpenCode1 = Convert.ToInt32(opencodeNode[0].InnerText.Trim()); model.OpenCode2 = Convert.ToInt32(opencodeNode[1].InnerText.Trim()); model.OpenCode3 = Convert.ToInt32(opencodeNode[2].InnerText.Trim()); model.OpenCode4 = Convert.ToInt32(opencodeNode[3].InnerText.Trim()); model.OpenCode5 = Convert.ToInt32(opencodeNode[4].InnerText.Trim()); result.Add(model); } result = result.OrderByDescending(S => S.Term).ToList(); } catch (Exception ex) { log.Error(typeof(HuBei30X5Job), 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)); } return result; } /// /// 完善备用站点江苏体彩7位数开奖实体信息 /// /// private bool OptimizeBackModel(ref OpenCode5DTModel model) { try { var htmlResource = NetHelper.GetUrlResponse(model.DetailUrl, Encoding.GetEncoding("gb2312")); if (string.IsNullOrWhiteSpace(htmlResource)) return false; HtmlDocument doc = new HtmlDocument(); doc.LoadHtml(htmlResource); var table = doc.DocumentNode.SelectSingleNode("//table"); if (table == null) return false; var trs = table.ChildNodes.Where(N => N.Name.ToLower() == "tr").ToList(); List tds = null; int Level1Num = 0, Level2Num = 0, Level3Num = 0; decimal Level1Money = 0, Level2Money = 0, Level3Money = 0, Sales = 0, Jackpot = 0; for (var i = 1; i < trs.Count; i++)//第一行为表头 { tds = trs[i].ChildNodes.Where(N => N.Name.ToLower() == "td").ToList(); if (tds.Count < 5) continue; if (tds[1].InnerText == "一等奖") { Level1Num = Convert.ToInt32(tds[2].InnerText); Level1Money = Convert.ToDecimal(tds[3].InnerText); } if (tds[1].InnerText == "二等奖") { Level2Num = Convert.ToInt32(tds[2].InnerText); Level2Money = Convert.ToDecimal(tds[3].InnerText); } if (tds[1].InnerText == "三等奖") { Level3Num = Convert.ToInt32(tds[2].InnerText); Level3Money = Convert.ToDecimal(tds[3].InnerText); } } Regex reg = new Regex(@"本期投注总额:([\d.,]*?)元"); var match = reg.Match(htmlResource); if (match.Success) { Sales = Convert.ToDecimal(match.Result("$1").Replace(",", string.Empty)); } reg = new Regex(@"奖池资金累计金额:([\d.,]*?)元"); match = reg.Match(htmlResource); if (match.Success) { Jackpot = Convert.ToDecimal(match.Result("$1").Replace(",", string.Empty)); } model.Detail = string.Format("{0},{1}^一等奖|{2}|{3},二等奖|{4}|{5},三等奖|{6}|{7}", Sales, Jackpot, Level1Num, Level1Money, Level2Num, Level2Money, Level3Num, Level3Money); return true; } catch (Exception ex) { log.Error(typeof(HuBei30X5Job), string.Format("【{0}】通过备用站点优化开奖列表时发生错误,错误信息【{1}】", Config.Area + Config.LotteryName, ex.Message)); } return false; } #region Attribute /// /// 配置信息 /// private SCCConfig Config = null; /// /// 当天抓取的最新一期开奖记录 /// private OpenCode5DTModel LatestItem = null; /// /// 当天抓取失败列表 /// private List FailedQiHaoList = null; /// /// 日志对象 /// private LogHelper log = null; /// /// 数据服务 /// private IDTOpenCode services = null; /// /// 当前彩种 /// private SCCLottery currentLottery { get { return SCCLottery.HuBei30X5; } } /// /// 邮件接口 /// private IEmail email = null; /// /// 是否本次运行抓取到开奖数据 /// private bool isGetData = false; #endregion } }