| 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537 |
- using System;
- using System.Collections.Generic;
- using System.Linq;
- using System.Text;
- using System.Text.RegularExpressions;
- using CP.Model;
- using HtmlAgilityPack;
- using Newtonsoft.Json;
- using Quartz;
- using SCC.Common;
- using SCC.Crawler.Tools;
- using SCC.Interface;
- using SCC.Models;
- namespace SCC.Crawler.DT
- {
- /// <summary>
- /// 河南22选5
- /// </summary>
- [DisallowConcurrentExecution]
- [PersistJobDataAfterExecution]
- public class HeNan22X5Job : IJob
- {
- /// <summary>
- /// 初始化函数
- /// </summary>
- public HeNan22X5Job()
- {
- log = new LogHelper();
- email = IOC.Resolve<IEmail>();
- }
- /// <summary>
- /// 作业执行入口
- /// </summary>
- /// <param name="context">作业执行上下文</param>
- 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 Fchn22x5LongInfo;
- try
- {
- //服务启动时配置初始数据
- if (LatestItem == null)
- {
- LatestItem = new Fchn22x5LongInfo
- {
- qi = CommonHelper.GenerateQiHaoYYYYQQQ(0),
- date = 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 > 0)
- {
- DoTodayJobByMainUrl();
- DoTodayJobByBackUrl();
- }
- }
- catch (Exception ex)
- {
- log.Error(GetType(), string.Format("【{0}】抓取时发生错误,错误信息【{1}】", Config.Area + currentLottery, ex.Message));
- }
- //保存最新期号
- context.JobDetail.JobDataMap["LatestItem"] = LatestItem;
- }
- /// <summary>
- /// 通过主站点爬取开奖数据
- /// (河南福彩网)
- /// </summary>
- private void DoTodayJobByMainUrl()
- {
- if (!string.IsNullOrEmpty(Config.MainUrl))
- {
- var openList = GetOpenListFromMainUrl(Config.MainUrl);
- if (openList == null || openList.Count == 0) return; //无抓取数据
- //抓取到的最新期数
- var newestQiHao = Convert.ToInt32(openList.OrderByDescending(m => m.qi).First().qi.ToString());
- //数据库里面最新期数
- LatestItem = Fchn22x5Data.GetLastOne();
- var startQiNum = Convert.ToInt32(LatestItem.qi.ToString());
- if (startQiNum > newestQiHao) return; //无最新数据
- //处理最新开奖数据
- Fchn22x5LongInfo matchItem = null;
- for (var i = startQiNum; i <= newestQiHao; i++)
- {
- matchItem = openList.Where(R => R.qi.ToString() == i.ToString()).FirstOrDefault();
- if (matchItem != null)
- {
- //add db
- matchItem.addtime = DateTime.Now;
- Fchn22x5Data.Add(matchItem);
- //Do Success Log
- log.Info(GetType(), CommonHelper.GetJobMainLogInfo(Config, i.ToString()));
- LatestItem = matchItem;
- isGetData = true;
- }
- }
- }
- }
- /// <summary>
- /// 获取主站开奖列表数据
- /// </summary>
- /// <param name="mainUrl">主站地址</param>
- /// <returns></returns>
- private List<Fchn22x5LongInfo> GetOpenListFromMainUrl(string mainUrl)
- {
- var result = new List<Fchn22x5LongInfo>();
- try
- {
- var url = new Uri(mainUrl);
- var htmlResource = NetHelper.GetUrlResponse(url.AbsoluteUri);
- if (htmlResource == null) return result;
- var doc = new HtmlDocument();
- doc.LoadHtml(htmlResource);
- var table = doc.DocumentNode.SelectSingleNode("//table");
- if (table == null) return result;
- var trs = table.ChildNodes.Where(S => S.Name.ToLower() == "tr").ToList();
- List<HtmlNode> tds = null;
- string term = string.Empty, openCodeString = string.Empty, optimizeUrl = string.Empty;
- decimal sales = 0, pool = 0;
- Fchn22x5LongInfo model = null;
- HtmlNode nodeA = null;
- for (var i = 1; i < trs.Count; i++)
- {
- tds = trs[i].ChildNodes.Where(S => S.Name.ToLower() == "td").ToList();
- if (tds.Count < 8) continue;
- model = new Fchn22x5LongInfo();
- term = tds[0].InnerText.Trim();
- if (term.StartsWith((CommonHelper.SCCSysDateTime.Year - 1).ToString())) break;
- model.qi = Convert.ToInt32(tds[0].InnerText.Trim());
- openCodeString = tds[1].InnerText.Trim();
- model.n1 = Convert.ToInt32(openCodeString.Substring(0, 2));
- model.n2 = Convert.ToInt32(openCodeString.Substring(2, 2));
- model.n3 = Convert.ToInt32(openCodeString.Substring(4, 2));
- model.n4 = Convert.ToInt32(openCodeString.Substring(6, 2));
- model.n5 = Convert.ToInt32(openCodeString.Substring(8, 2));
- model.date = Convert.ToDateTime(tds[6].InnerText.Trim());
- sales = Convert.ToDecimal(tds[2].InnerText.Trim());
- pool = Convert.ToDecimal(tds[5].InnerText.Trim());
- //model.Spare = sales + "|" + pool;
- model.tzmoney = sales.ToString();
- model.nextmoney = pool.ToString();
- nodeA = tds[7].ChildNodes.Where(N => N.Name.ToLower() == "a").FirstOrDefault();
- if (nodeA == null) continue;
- optimizeUrl = nodeA.Attributes["href"].Value;
- string detailUrl = new Uri(url, optimizeUrl).AbsoluteUri;
- OptimizeMainModel(ref model, detailUrl);
- result.Add(model);
- }
- //var checkDataHelper = new CheckDataHelper();
- //var dbdata = services.GetListS<OpenCode5DTModel>(currentLottery)
- // .ToDictionary(w => w.Term.ToString(), w => w.GetCodeStr());
- //checkDataHelper.CheckData(dbdata, result.ToDictionary(w => w.Term.ToString(), w => w.GetCodeStr()),
- // Config.Area, currentLottery);
- }
- catch (Exception ex)
- {
- log.Error(GetType(),
- 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 + currentLottery, ex.Message));
- }
- return result;
- }
- /// <summary>
- /// 完善主站江苏体彩7位数开奖详情信息
- /// </summary>
- /// <param name="model"></param>
- private bool OptimizeMainModel(ref Fchn22x5LongInfo model,string detailUrl)
- {
- try
- {
- var htmlResource = NetHelper.GetUrlResponse(detailUrl);
- if (string.IsNullOrWhiteSpace(htmlResource)) return false;
- var doc = new HtmlDocument();
- doc.LoadHtml(htmlResource);
- var table = doc.DocumentNode.SelectSingleNode("//table");
- if (table == null) return false;
- var trs = table.ChildNodes.Where(C => C.Name == "tr").ToList();
- List<HtmlNode> tds = null;
- foreach (var tr in trs)
- {
- tds = tr.ChildNodes.Where(S => S.Name.ToLower() == "td").ToList();
- if (tds.Count < 3) continue;
- if (tds[0].InnerText == "一等奖")
- {
- //一等奖
- model.zj1 = tds[1].InnerText.Replace("注", "");
- model.jo1 = tds[2].InnerText.Replace("元", "");
- }
- else if (tds[0].InnerText == "二等奖")
- {
- //二等奖
- model.zj2 = tds[1].InnerText.Replace("注", "");
- model.jo2 = tds[2].InnerText.Replace("元", "");
- }
- else if (tds[0].InnerText == "三等奖")
- {
- //三等奖
- model.zj3 = tds[1].InnerText.Replace("注", "");
- model.jo3 = tds[2].InnerText.Replace("元", "");
- }
- else if (tds[0].InnerText == "好运二")
- {
- //好运二
- model.hj2 = tds[1].InnerText.Replace("注", "");
- model.ho2 = tds[2].InnerText.Replace("元", "");
- }
- else if (tds[0].InnerText == "好运三")
- {
- //好运三
- model.hj3 = tds[1].InnerText.Replace("注", "");
- model.ho3 = tds[2].InnerText.Replace("元", "");
- }
- else if (tds[0].InnerText == "好运四")
- {
- //好运四
- model.hj4 = tds[1].InnerText.Replace("注", "");
- model.ho4 = tds[2].InnerText.Replace("元", "");
- }
- var list = new List<Winbonus>();
- list.Add(new Winbonus()
- {
- item = "一等奖",
- wincount = model.zj1,
- winmoney = model.jo1
- });
- list.Add(new Winbonus()
- {
- item = "二等奖",
- wincount = model.zj2,
- winmoney = model.jo2
- });
- list.Add(new Winbonus()
- {
- item = "三等奖",
- wincount = model.zj3,
- winmoney = model.jo3
- });
- list.Add(new Winbonus()
- {
- item = "好运二",
- wincount = model.hj2,
- winmoney = model.ho2
- });
- list.Add(new Winbonus()
- {
- item = "好运三",
- wincount = model.hj3,
- winmoney = model.ho3
- });
- list.Add(new Winbonus()
- {
- item = "好运四",
- wincount = model.hj4,
- winmoney = model.ho4
- });
- model.winbonus = JsonConvert.SerializeObject(list);
- }
- return true;
- }
- catch (Exception ex)
- {
- log.Error(GetType(),
- string.Format("【{0}】通过主站点完善抓取开奖列表时发生错误,错误信息【{1}】", Config.Area + currentLottery, ex.Message));
- }
- return false;
- }
- /// <summary>
- /// 通过备用站点抓取开奖数据
- /// (彩票两元网)
- /// </summary>
- private void DoTodayJobByBackUrl()
- {
- if (!string.IsNullOrEmpty(Config.BackUrl))
- {
- var openList = GetOpenListFromBackUrl();
- if (openList == null || openList.Count == 0) return; //无抓取数据
- //抓取到的最新期数
- var newestQiHao = Convert.ToInt32(openList.OrderByDescending(m => m.qi).First().qi.ToString());
- //数据库里面最新期数
- //LatestItem = Fchn22x5Data.GetLastOne();
- var startQiNum = Convert.ToInt32(LatestItem.qi.ToString());
- if (startQiNum > newestQiHao) return; //无最新数据
- //处理最新开奖数据
- Fchn22x5LongInfo matchItem = null;
- for (var i = startQiNum; i <= newestQiHao; i++)
- {
- matchItem = openList.Where(R => R.qi.ToString() == i.ToString()).FirstOrDefault();
- if (matchItem != null)
- {
- //add db
- matchItem.addtime = DateTime.Now;
- Fchn22x5Data.Add(matchItem);
- //Do Success Log
- log.Info(GetType(), CommonHelper.GetJobBackLogInfo(Config, i.ToString()));
- LatestItem = matchItem;
- isGetData = true;
- }
- }
- }
- }
- /// <summary>
- /// 获取备用站点开奖列表数据
- /// </summary>
- /// <returns></returns>
- private List<Fchn22x5LongInfo> GetOpenListFromBackUrl()
- {
- var result = new List<Fchn22x5LongInfo>();
- try
- {
- var url = new Uri(Config.BackUrl);
- var htmlResource = NetHelper.GetUrlResponse(Config.BackUrl, Encoding.GetEncoding("gb2312"));
- if (htmlResource == null) return result;
- var 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();
- Fchn22x5LongInfo model = null;
- HtmlNode nodeA = null;
- var 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 Fchn22x5LongInfo();
- nodeA = tds[0].ChildNodes.Where(n => n.Name == "a").FirstOrDefault();
- if (nodeA == null) continue;
- model.qi = Convert.ToInt32(nodeA.InnerText.Trim());
- optimizeUrl = nodeA.Attributes["href"].Value;
- string detailUrl = new Uri(url, optimizeUrl).AbsoluteUri;
- model.date = 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.n1 = Convert.ToInt32(opencodeNode[0].InnerText.Trim());
- model.n2 = Convert.ToInt32(opencodeNode[1].InnerText.Trim());
- model.n3 = Convert.ToInt32(opencodeNode[2].InnerText.Trim());
- model.n4 = Convert.ToInt32(opencodeNode[3].InnerText.Trim());
- model.n5 = Convert.ToInt32(opencodeNode[4].InnerText.Trim());
- OptimizeBackModel(ref model, detailUrl);
- result.Add(model);
- }
- //var checkDataHelper = new CheckDataHelper();
- //var dbdata = services.GetListS<OpenCode5DTModel>(currentLottery)
- // .ToDictionary(w => w.Term.ToString(), w => w.GetCodeStr());
- //checkDataHelper.CheckData(dbdata, result.ToDictionary(w => w.Term.ToString(), w => w.GetCodeStr()),
- // Config.Area, currentLottery);
- //result = result.OrderByDescending(S => S.Term).ToList();
- }
- catch (Exception ex)
- {
- log.Error(GetType(),
- 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 + currentLottery, ex.Message));
- }
- return result;
- }
- /// <summary>
- /// 完善备用站点江苏体彩7位数开奖实体信息
- /// </summary>
- /// <param name="model"></param>
- private bool OptimizeBackModel(ref Fchn22x5LongInfo model,string detailUrl)
- {
- try
- {
- var htmlResource = NetHelper.GetUrlResponse(detailUrl, Encoding.GetEncoding("gb2312"));
- if (string.IsNullOrWhiteSpace(htmlResource)) return false;
- var 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<HtmlNode> tds = null;
- 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")
- {
- //一等奖
- model.zj1 = tds[2].InnerText;
- model.jo1 = tds[3].InnerText;
- }
- if (tds[1].InnerText == "二等奖")
- {
- //二等奖
- model.zj2 = tds[2].InnerText;
- model.jo2 = tds[3].InnerText;
- }
- if (tds[1].InnerText == "三等奖")
- {
- //三等奖
- model.zj3 = tds[2].InnerText;
- model.jo3 = tds[3].InnerText;
- }
- if (tds[1].InnerText == "好运二")
- {
- //好运二
- model.hj2 = tds[2].InnerText;
- model.ho2 = tds[3].InnerText;
- }
- if (tds[1].InnerText == "好运三")
- {
- //好运三
- model.hj3 = tds[2].InnerText;
- model.ho3 = tds[3].InnerText;
- }
- if (tds[1].InnerText == "好运四")
- {
- //好运四
- model.hj4 = tds[2].InnerText;
- model.ho4 = tds[3].InnerText;
- }
- var list = new List<Winbonus>();
- list.Add(new Winbonus()
- {
- item = "一等奖",
- wincount = model.zj1,
- winmoney = model.jo1
- });
- list.Add(new Winbonus()
- {
- item = "二等奖",
- wincount = model.zj2,
- winmoney = model.jo2
- });
- list.Add(new Winbonus()
- {
- item = "三等奖",
- wincount = model.zj3,
- winmoney = model.jo3
- });
- list.Add(new Winbonus()
- {
- item = "好运二",
- wincount = model.hj2,
- winmoney = model.ho2
- });
- list.Add(new Winbonus()
- {
- item = "好运三",
- wincount = model.hj3,
- winmoney = model.ho3
- });
- list.Add(new Winbonus()
- {
- item = "好运四",
- wincount = model.hj4,
- winmoney = model.ho4
- });
- model.winbonus = JsonConvert.SerializeObject(list);
- }
- var reg = new Regex(@"本期投注总额:([\d.,]*?)元");
- var match = reg.Match(htmlResource);
- if (match.Success) model.tzmoney = 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) model.nextmoney = match.Result("$1").Replace(",", string.Empty);
- return true;
- }
- catch (Exception ex)
- {
- log.Error(GetType(),
- 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 + currentLottery, ex.Message));
- }
- return false;
- }
- #region Attribute
- /// <summary>
- /// 配置信息
- /// </summary>
- private SCCConfig Config;
- /// <summary>
- /// 当天抓取的最新一期开奖记录
- /// </summary>
- private Fchn22x5LongInfo LatestItem;
- #pragma warning disable CS0169 // 从不使用字段“HeNan22X5Job.FailedQiHaoList”
- /// <summary>
- /// 当天抓取失败列表
- /// </summary>
- private List<string> FailedQiHaoList;
- #pragma warning restore CS0169 // 从不使用字段“HeNan22X5Job.FailedQiHaoList”
- /// <summary>
- /// 日志对象
- /// </summary>
- private readonly LogHelper log;
- /// <summary>
- /// 当前彩种
- /// </summary>
- private SCCLottery currentLottery => SCCLottery.HeNan22X5;
- /// <summary>
- /// 邮件接口
- /// </summary>
- private readonly IEmail email;
- #pragma warning disable CS0414 // 字段“HeNan22X5Job.isGetData”已被赋值,但从未使用过它的值
- /// <summary>
- /// 是否本次运行抓取到开奖数据
- /// </summary>
- private bool isGetData;
- #pragma warning restore CS0414 // 字段“HeNan22X5Job.isGetData”已被赋值,但从未使用过它的值
- #endregion
- }
- }
|