kjh.api.55128.cn_old/SCC.Crawler/DT/JSTC7WSJob.cs

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>
    ///     江苏体彩7位数
    /// </summary>
    [DisallowConcurrentExecution]
    [PersistJobDataAfterExecution]
    public class JSTC7WSJob : IJob
    {
        /// <summary>
        /// 初始化函数
        /// </summary>
        public JSTC7WSJob()
        {
            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 Tcjs7wsLongInfo;
            try
            {
                //服务启动时配置初始数据
                if (LatestItem == null)
                {
                    LatestItem = new Tcjs7wsLongInfo
                    {
                        qi = CommonHelper.GenerateQiHaoYYQQQ(0),
                        date = 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))
                {
                    DoMainUrl();
                    DoBackUrl();
                }

                if (!LatestItem.qi.ToString().StartsWith(CommonHelper.SCCSysDateTime.ToString("yy")))
                    LatestItem = new Tcjs7wsLongInfo
                    {
                        qi = CommonHelper.GenerateQiHaoYYQQQ(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)
                {
                    DoMainUrl();
                    DoBackUrl();
                }
            }
            catch (Exception ex)
            {
                log.Error(typeof(JSTC7WSJob),
                    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
        ///     (江苏体彩网)
        /// </summary>
        private void DoMainUrl()
        {
            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 = Tcjs7wsData.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
                Tcjs7wsLongInfo 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;
                        Tcjs7wsData.Add(matchItem);
                        //Do Success Log
                        log.Info(typeof(JSTC7WSJob), 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<Tcjs7wsLongInfo> GetOpenListFromMainUrl(string mainUrl)
        {
            var result = new List<Tcjs7wsLongInfo>();
            try
            {
                var pageIndex = 1;
                var htmlResource = string.Empty;
                var resourceUrl = new Uri(mainUrl);
                var isLoop = true;
                var lastYear = (DateTime.Now.Year - 1).ToString().Substring(2);
                var postData = "current_page={0}&all_count=0&num=";
                var OpenTime = string.Empty;
                while (isLoop)
                {
                    htmlResource = NetHelper.GetUrlResponse(resourceUrl.AbsoluteUri, "POST",
                        string.Format(postData, pageIndex), Encoding.UTF8);
                    var jsonData = htmlResource.JsonToEntity<dynamic>();
                    var dataList = jsonData["items"];
                    foreach (var data in dataList)
                    {
                        if (data["num"].Value.StartsWith(lastYear))
                        {
                            isLoop = false;
                            break;
                        }
                        OpenTime = data["date_publish"].Value.Insert(6, "-").Insert(4, "-");

                        string detailUrl = string.Format(
                                "http://www.js-lottery.com/Article/news/group_id/3/article_id/{0}.html",
                                data["article_id"].Value);
                        string qihao = data["num"].Value;
                        qihao = qihao.Length < 7 ? $"20{qihao}" : qihao;
                        var model = new Tcjs7wsLongInfo
                        {
                            qi = Convert.ToInt32(qihao),
                            n1 = Convert.ToInt32(data["one"].Value),
                            n2 = Convert.ToInt32(data["two"].Value),
                            n3 = Convert.ToInt32(data["three"].Value),
                            n4 = Convert.ToInt32(data["four"].Value),
                            n5 = Convert.ToInt32(data["five"].Value),
                            n6 = Convert.ToInt32(data["six"].Value),
                            n7 = Convert.ToInt32(data["seven"].Value),
                            date = Convert.ToDateTime(OpenTime),
                        };
                        OptimizeMainModel(ref model, detailUrl);
                        result.Add(model);
                    }

                    pageIndex++;
                }
                //var checkDataHelper = new CheckDataHelper();
                //var dbdata = services.GetListS<OpenCode7DTModel>(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(typeof(JSTC7WSJob),
                    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 void OptimizeBackModel(ref Tcjs7wsLongInfo model, HtmlNode tr)
        {
            try
            {
                var tds = tr.ChildNodes.Where(w => w.Name == "td").ToList();
                var xiaoshoue = tds[2].InnerText.Trim().Replace(",", "").Replace("元", "");
                var jiangchi = "";
                //特等奖
                model.zj = tds[3].InnerText.Trim();
                model.jo = tds[4].InnerText;
                //一等奖
                model.zj1 = tds[5].InnerText.Trim();
                model.jo1 = tds[6].InnerText;
                //二等奖
                model.zj2 = tds[7].InnerText.Trim();
                model.jo2 = tds[8].InnerText;
                model.nextmoney = jiangchi;
                model.tzmoney = xiaoshoue;

                var list = new List<Winbonus>();
                list.Add(new Winbonus()
                {
                    item = "特等奖",
                    wincount = model.zj,
                    winmoney = model.jo
                });
                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
                });
                model.winbonus = JsonConvert.SerializeObject(list);
            }
            catch (Exception ex)
            {
                log.Error(typeof(JSTC7WSJob),
                    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 + currentLottery, ex.Message));
            }
        }

        /// <summary>
        ///     完善主站江苏体彩7位数开奖详情信息
        /// </summary>
        /// <param name="model"></param>
        private void OptimizeMainModel(ref Tcjs7wsLongInfo model, string detailUrl)
        {
            try
            {
                var htmlResource = NetHelper.GetUrlResponse(detailUrl);
                if (htmlResource == null) return;
                if (!string.IsNullOrEmpty(htmlResource))
                {
                    var doc = new HtmlDocument();
                    doc.LoadHtml(htmlResource);
                    var table = doc.DocumentNode.SelectNodes("//table");
                    if (table != null && table.Count > 1)
                    {
                        var trs = table[1].ChildNodes.Where(N => N.Name.ToLower() == "tbody").First().ChildNodes
                            .Where(N => N.Name.ToLower() == "tr").ToList();
                        for (var i = 0; i < trs.Count; i++)
                        {
                            var tds = trs[i].ChildNodes.Where(N => N.Name.ToLower() == "td").ToList();
                            if (tds[0].InnerText == "特等奖")
                            {
                                model.zj = tds[1].InnerText.Replace(",", string.Empty).Replace("注", string.Empty);
                                model.jo = tds[2].InnerText.Replace("元", string.Empty).Replace("--", "0").Replace(",", "").Trim();
                            }
                            else if (tds[0].InnerText == "一等奖")
                            {
                                model.zj1 = tds[1].InnerText.Replace(",", string.Empty).Replace("注", string.Empty);
                                model.jo1 = tds[2].InnerText.Replace("元", string.Empty).Replace("--", "0").Replace(",", "").Trim();
                            }
                            else if (tds[0].InnerText == "二等奖")
                            {
                                model.zj2 = tds[1].InnerText.Replace(",", string.Empty).Replace("注", string.Empty);
                                model.jo2 = tds[2].InnerText.Replace("元", string.Empty).Replace("--", "0").Replace(",", "").Trim();
                            }
                            else if (tds[0].InnerText == "三等奖")
                            {
                                model.zj3 = tds[1].InnerText.Replace(",", string.Empty).Replace("注", string.Empty);
                                model.jo3 = tds[2].InnerText.Replace("元", string.Empty).Replace("--", "0").Replace(",", "").Trim();
                            }
                            else if (tds[0].InnerText == "四等奖")
                            {
                                model.zj4 = tds[1].InnerText.Replace(",", string.Empty).Replace("注", string.Empty);
                                model.jo4 = tds[2].InnerText.Replace("元", string.Empty).Replace("--", "0").Replace(",", "").Trim();
                            }
                            else if (tds[0].InnerText == "五等奖")
                            {
                                model.zj5 = tds[1].InnerText.Replace(",", string.Empty).Replace("注", string.Empty);
                                model.jo5 = tds[2].InnerText.Replace("元", string.Empty).Replace("--", "0").Replace(",", "").Trim();
                            };

                            var list = new List<Winbonus>();
                            list.Add(new Winbonus()
                            {
                                item = "特等奖",
                                wincount = model.zj,
                                winmoney = model.jo
                            });
                            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 = "四等奖",
                                wincount = model.zj4,
                                winmoney = model.jo4
                            });
                            list.Add(new Winbonus()
                            {
                                item = "五等奖",
                                wincount = model.zj5,
                                winmoney = model.jo5
                            });

                            model.winbonus = JsonConvert.SerializeObject(list);
                        }
                    }

                    var reg1 = new Regex(@"本省（区、市）销售额：([\s\S]*?)元");
                    var match1 = reg1.Match(htmlResource);
                    if (match1.Success)
                    {
                        //2016年182期及以前期数
                        //Sales = Convert.ToDecimal(match1.Result("$1"));
                        model.tzmoney = match1.Result("$1");
                    }
                    else
                    {
                        //2016年183期及以后期数
                        reg1 = new Regex(@"本期销售金额：([\s\S]*?)元");
                        match1 = reg1.Match(htmlResource);
                        if (match1.Success) model.tzmoney = match1.Result("$1");
                    }

                    var ps = table[1].ParentNode.ChildNodes.Where(N => N.Name.ToLower() == "p").ToList();
                    var potString = ps.Last().InnerHtml;
                    reg1 = new Regex(@"<br>([\s\S]*?)元");
                    match1 = reg1.Match(potString);
                    if (match1.Success)
                    {
                        var potValue = match1.Result("$1").Replace("&nbsp;", string.Empty);
                        if (potValue.Contains("<br>"))
                        {
                            while (potValue.IndexOf("<br>") > 0)
                            {
                                potValue = potValue.Substring(potValue.IndexOf("<br>") + 4);
                            }
                            model.nextmoney = potValue;
                        }// Jackpot = Convert.ToDecimal(potValue.Substring(potValue.IndexOf("<br>") + 4));
                        else
                            // Jackpot = Convert.ToDecimal(potValue);
                            model.tzmoney = potValue;
                    }
                }
            }
            catch (Exception ex)
            {
                log.Error(typeof(JSTC7WSJob),
                    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 + currentLottery, ex.Message));
            }
        }

        /// <summary>
        ///     通过备用站点抓取开奖数据
        ///     (百度乐彩)
        /// </summary>
        private void DoBackUrl()
        {
            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 = Tcjs7wsData.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
                Tcjs7wsLongInfo 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;
                        Tcjs7wsData.Add(matchItem);
                        //Do Success Log
                        log.Info(typeof(JSTC7WSJob), CommonHelper.GetJobBackLogInfo(Config, i.ToString()));
                        LatestItem = matchItem;
                        isGetData = true;
                    }
                }
            }
        }

        private List<Tcjs7wsLongInfo> GetOpenListFromBackUrl()
        {
            var result = new List<Tcjs7wsLongInfo>();
            try
            {
                var url = Config.BackUrl;

                try
                {
                    var htmlResource = NetHelper.GetUrlResponse(url, Encoding.GetEncoding("utf-8"));
                    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();
                    Tcjs7wsLongInfo 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 < 14) continue;
                        model = new Tcjs7wsLongInfo();
                        nodeA = tds[13].ChildNodes.Where(n => n.Name == "a").FirstOrDefault();
                        if (nodeA == null) continue;
                        model.qi = Convert.ToInt32(tds[0].InnerText.Trim());
                        //model.DetailUrl = new Uri(url, optimizeUrl).AbsoluteUri;
                        model.date = Convert.ToDateTime(tds[1].InnerText.Substring(0, 10));
                        if (tds[2].ChildNodes.Count == 0) continue;
                        var opencodeNode = tds[2].ChildNodes.Where(n => n.Name.ToLower() == "em").ToList();
                        if (opencodeNode.Count < 6) 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());
                        model.n6 = Convert.ToInt32(opencodeNode[5].InnerText.Trim());
                        model.n7 = Convert.ToInt32(opencodeNode[6].InnerText.Trim());
                        
                        OptimizeBackModel(ref model, trs[i]);
                        result.Add(model);
                    }
                    //var checkDataHelper = new CheckDataHelper();
                    //var dbdata = services.GetListS<OpenCode8DTModel>(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;
            }
            catch (Exception ex)
            {
                log.Error(typeof(JSTC7WSJob),
                    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;
        }


        #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 Tcjs7wsLongInfo LatestItem;

#pragma warning disable CS0414 // 字段“JSTC7WSJob.FailedQiHaoList”已被赋值，但从未使用过它的值
        /// <summary>
        ///     当天抓取失败列表
        /// </summary>
        private List<string> FailedQiHaoList = null;
#pragma warning restore CS0414 // 字段“JSTC7WSJob.FailedQiHaoList”已被赋值，但从未使用过它的值

        /// <summary>
        ///     日志对象
        /// </summary>
        private readonly LogHelper log;

        /// <summary>
        ///     当前彩种
        /// </summary>
        private SCCLottery currentLottery => SCCLottery.JiangSuTC7WS;

        /// <summary>
        ///     邮件接口
        /// </summary>
        private IEmail email;

#pragma warning disable CS0414 // 字段“JSTC7WSJob.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“JSTC7WSJob.isGetData”已被赋值，但从未使用过它的值

        #endregion
    }
}