kjh.api.55128.cn_old/SCC.Crawler/DT/HeBei20X5Job.cs

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
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>
    ///     河北20选5
    /// </summary>
    [DisallowConcurrentExecution]
    [PersistJobDataAfterExecution]
    public class HeBei20X5Job : IJob
    {
        /// <summary>
        ///     构造函数
        /// </summary>
        public HeBei20X5Job()
        {
            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 Fchebei20x5LongInfo;
            try
            {
                //服务启动时配置初始数据
                if (LatestItem == null)
                {
                    LatestItem = new Fchebei20x5LongInfo
                    {
                        qi = CommonHelper.GenerateQiHaoYYYYQQQ(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 Fchebei20x5LongInfo
                    {
                        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 > 12)
                {
                    DoMainUrl();
                    //DoBackUrl();
                }
            }
            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;
        }

        #region 通过主站爬取数据
        /// <summary>
        /// 执行主站的数据插入
        /// </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 = Fchebei20x5Data.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
                Fchebei20x5LongInfo matchItem = null;
                for (var i = startQiNum; i <= newestQiHao; i++)
                {
                    matchItem = openList.FirstOrDefault(r => r.qi.ToString() == i.ToString());

                    if (matchItem != null)
                    {
                        //add db
                        matchItem.addtime = DateTime.Now;
                        Fchebei20x5Data.Add(matchItem);
                        //Do Success Log
                        log.Info(GetType(), CommonHelper.GetJobMainLogInfo(Config, i.ToString()));
                        LatestItem = matchItem;
                        isGetData = true;
                    }
                }
            }
        }
        /// <summary>
        /// 获取主站的数据列表
        /// </summary>
        /// <param name="mainUrl"></param>
        /// <returns></returns>
        private List<Fchebei20x5LongInfo> GetOpenListFromMainUrl(string mainUrl)
        {
            var result = new List<Fchebei20x5LongInfo>();
            try
            {
                var url = new Uri(mainUrl);
                var htmlResource = NetHelper.GetUrlResponse(mainUrl, 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();
                Fchebei20x5LongInfo model = null;
#pragma warning disable CS0219 // 变量“nodeA”已被赋值，但从未使用过它的值
                HtmlNode nodeA = null;
#pragma warning restore CS0219 // 变量“nodeA”已被赋值，但从未使用过它的值
                var optimizeUrl = string.Empty;
                for (var i = 2; i < trs.Count; i++) //第一二行为表头
                {
                    var tds = trs[i].ChildNodes.Where(node => node.Name == "td").ToList();
                    if (tds.Count < 4) continue;
                    model = new Fchebei20x5LongInfo();
                    //nodeA = tds[6].ChildNodes.Where(n => n.Name == "a").FirstOrDefault();
                    //if (nodeA == null) continue;
                    model.qi = Convert.ToInt32(tds[0].InnerText.Trim());
                    //optimizeUrl = nodeA.Attributes["href"].Value;
                    //model.DetailUrl = new Uri(url, optimizeUrl).AbsoluteUri;
                    model.date = Convert.ToDateTime(tds[1].InnerText.Remove(10));
                    var opencodeNode = tds[2].ChildNodes.Where(n => n.Name.ToLower() == "span").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());
                    GetKaijiangDetails(ref model, model.qi.ToString());
                    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>
        ///     获取主站下开奖详情
        /// </summary>
        /// <param name="nodes"></param>
        /// <returns></returns>
        private void GetKaijiangDetails(ref Fchebei20x5LongInfo model, string qi)
        {
            string url = $"http://kaijiang.500.com/shtml/hbesxw/{qi}.shtml";
            var htmlResource = NetHelper.GetUrlResponse(url, Encoding.GetEncoding("gb2312"));
            if (htmlResource == null) return;
            var list = new List<Winbonus>();
            var doc = new HtmlDocument();
            doc.LoadHtml(htmlResource);
            var tables = doc.DocumentNode.SelectNodes("//table");
            if (tables == null) return;
            foreach (var table in tables)
            {
                var trs = table.ChildNodes.Where(node => node.Name == "tr").ToList();
                for (var i = 1; i < trs.Count; i++) //第一二行为表头
                {
                    var tds = trs[i].ChildNodes.Where(node => node.Name == "td").ToList();
                    if (tds.Count != 3) continue;
                    if (tds[0].InnerText.Trim() == "一等奖")
                    {
                        model.zj1 = tds[1].InnerText.Trim();
                        model.jo1 = tds[2].InnerText.Trim();
                        list.Add(new Winbonus()
                        {
                            item = "一等奖",
                            wincount = model.zj1,
                            winmoney = model.jo1
                        });
                    }
                    if (tds[0].InnerText.Trim() == "二等奖")
                    {
                        model.zj2 = tds[1].InnerText.Trim();
                        model.jo2 = tds[2].InnerText.Trim();
                        list.Add(new Winbonus()
                        {
                            item = "二等奖",
                            wincount = model.zj2,
                            winmoney = model.jo2
                        });
                    }
                    if (tds[0].InnerText.Trim() == "三等奖")
                    {
                        model.zj3 = tds[1].InnerText.Trim();
                        model.jo3 = tds[2].InnerText.Trim();
                        list.Add(new Winbonus()
                        {
                            item = "三等奖",
                            wincount = model.zj3,
                            winmoney = model.jo3
                        });
                    }
                }

            }

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

        #endregion

        #region 通过副站爬取数据

        /// <summary>
        ///     副站数据爬取 地址：https://fx.cp2y.com/draw/history_10065_Y/
        /// </summary>
        /// <returns></returns>
        private List<Fchebei20x5LongInfo> GetOpenListFromBackUrl()
        {
            var result = new List<Fchebei20x5LongInfo>();
            try
            {
                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.SelectNodes("//tbody[@id='pagedata']");
                if (table == null) return result;
                var trs = table[0].ChildNodes.Where(S => S.Name.ToLower() == "tr").ToList();
                List<HtmlNode> tds = null;
                string term = string.Empty, openCodeString = string.Empty, optimizeUrl = string.Empty;
                Fchebei20x5LongInfo model = null;
                for (var i = 0; i < trs.Count; i++)
                {
                    tds = trs[i].ChildNodes.Where(S => S.Name.ToLower() == "td").ToList();
                    if (tds.Count < 4) continue;

                    model = new Fchebei20x5LongInfo();
                    term = tds[1].InnerText.Trim();
                    if (term.StartsWith((CommonHelper.SCCSysDateTime.Year - 1).ToString())) break;

                    model.qi = Convert.ToInt32(term);
                    openCodeString = tds[2].InnerText.Replace("&nbsp;", "").Trim();
                    model.n1 = Convert.ToInt32(openCodeString.Substring(0, 2));
                    model.n2 = Convert.ToInt32(openCodeString.Substring(3, 2));
                    model.n3 = Convert.ToInt32(openCodeString.Substring(6, 2));
                    model.n4 = Convert.ToInt32(openCodeString.Substring(9, 2));
                    model.n5 = Convert.ToInt32(openCodeString.Substring(12, 2));
                    //组装开奖详情

                    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>
        /// 执行备用站的数据插入
        /// </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 = Fchebei20x5Data.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
                Fchebei20x5LongInfo 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;
                        Fchebei20x5Data.Add(matchItem);
                        //Do Success Log
                        log.Info(GetType(), CommonHelper.GetJobBackLogInfo(Config, i.ToString()));
                        LatestItem = matchItem;
                        isGetData = true;
                    }
                }
            }
        }

        #endregion

        #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 Fchebei20x5LongInfo LatestItem;

#pragma warning disable CS0414 // 字段“HeBei20X5Job.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“HeBei20X5Job.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.HeBei20X5;

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

#pragma warning disable CS0414 // 字段“HeBei20X5Job.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“HeBei20X5Job.isGetData”已被赋值，但从未使用过它的值

        #endregion
    }
}