From 294494e5559dcc92670dd9800caf976853aff876 Mon Sep 17 00:00:00 2001 From: username@email.com <yzy2002yzy@163.com> Date: 星期三, 24 七月 2024 14:23:02 +0800 Subject: [PATCH] 提交 --- zhengcaioa/Crawler/sichuan/CcgpSichuanoperation.cs | 716 +++++++++++++++++++++++++++++++++++++++++++--------------- 1 files changed, 525 insertions(+), 191 deletions(-) diff --git a/zhengcaioa/Crawler/sichuan/CcgpSichuanoperation.cs b/zhengcaioa/Crawler/sichuan/CcgpSichuanoperation.cs index 53b2a2c..37c8ea7 100644 --- a/zhengcaioa/Crawler/sichuan/CcgpSichuanoperation.cs +++ b/zhengcaioa/Crawler/sichuan/CcgpSichuanoperation.cs @@ -3,10 +3,14 @@ using System; using System.Collections.Generic; using System.Globalization; +using System.IO; +using System.IO.Compression; using System.Linq; +using System.Net; using System.Net.Http; using System.Net.Http.Headers; using System.Text; +using System.Text.RegularExpressions; using System.Threading; using zhengcaioa.Models; namespace Crawler.sichuan @@ -19,142 +23,139 @@ { try { + + //鑾峰彇鏇存柊鏃堕棿 string operationStartTime = ""; string operationEndTime = ""; - Updatetime updatetime = _ccontext.Updatetimes.Where(x => x.Sheng == "涓浗鏀块噰缃戝洓宸濈渷").FirstOrDefault(); - if (updatetime != null) + + logg.WriteLog("寮�濮嬪垽鏂椂闂�", "涓浗鏀块噰缃戝洓宸濈渷"); + + if (DateTime.Now.Hour == 0 ) { - operationStartTime = updatetime.Updatetime1.ToString("yyyy-MM-dd"); - operationEndTime = DateTime.Now.ToString("yyyy-MM-dd"); - //updatetime.Updatetime1 = DateTime.Now.AddDays(-1); - //_ccontext.SaveChanges(); + DateTime datenow = DateTime.Now.Date; + Updatetime updatetime = _ccontext.Updatetimes.Where(x => x.Sheng == "涓浗鏀块噰缃戝洓宸濈渷").FirstOrDefault(); + logg.WriteLog("datenow="+datenow.ToString("yyyy-MM-dd HH:mm:ss"), "涓浗鏀块噰缃戝洓宸濈渷"); + operationStartTime = datenow.AddDays(-1).ToString("yyyy-MM-dd"); + operationEndTime = datenow.AddDays(-1).ToString("yyyy-MM-dd"); + logg.WriteLog("operationStartTime=" + operationStartTime, "涓浗鏀块噰缃戝洓宸濈渷"); + logg.WriteLog("operationEndTime=" + operationEndTime, "涓浗鏀块噰缃戝洓宸濈渷"); + var currPage = 1; + string sichuanpageurll = ""; + //Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 + #region 鎰忓悜鍏紑 + currPage = 1; + sichuanpageurll = "https://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=9&dbselect=bidx&kw=%E6%84%8F%E5%90%91&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷鎰忓悜鍏紑 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); + yixianggonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage); + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷鎰忓悜鍏紑 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); + #endregion + + //Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 + + #region 鍏紑鎷涙爣 + currPage = 1; + sichuanpageurll = "https://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=1&dbselect=bidx&kw=&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 鍏紑鎷涙爣 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); + caigougonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage, "1", "鍏紑鎷涙爣"); + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 鍏紑鎷涙爣 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); + #endregion + //Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 + + #region 璇环 + currPage = 1; + sichuanpageurll = "https://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=2&dbselect=bidx&kw=&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 璇环 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); + caigougonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage, "4", "璇环"); + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 璇环 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); + #endregion + //Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 + #region 绔炰簤鎬ц皥鍒� + currPage = 1; + sichuanpageurll = "https://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=3&dbselect=bidx&kw=&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 绔炰簤鎬ц皥鍒� 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); + caigougonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage, "5", "绔炰簤鎬ц皥鍒�"); + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 绔炰簤鎬ц皥鍒� 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); + #endregion + //Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 + #region 鍗曚竴鏉ユ簮 + currPage = 1; + sichuanpageurll = "https://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=4&dbselect=bidx&kw=&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 鍗曚竴鏉ユ簮 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); + caigougonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage, "7", "鍗曚竴鏉ユ簮"); + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 鍗曚竴鏉ユ簮 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); + #endregion + //Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 + #region 閭�璇锋嫑鏍� + currPage = 1; + sichuanpageurll = "https://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=6&dbselect=bidx&kw=&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 閭�璇锋嫑鏍� 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); + caigougonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage, "2", "閭�璇锋嫑鏍�"); + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 閭�璇锋嫑鏍� 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); + #endregion + //Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 + #region 绔炰簤鎬х鍟� + currPage = 1; + sichuanpageurll = "https://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=10&dbselect=bidx&kw=&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 绔炰簤鎬х鍟� 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); + caigougonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage, "3", "绔炰簤鎬х鍟�"); + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 绔炰簤鎬х鍟� 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); + #endregion + //Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 + #region 涓爣鍏憡 + currPage = 1; + sichuanpageurll = "https://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=7&dbselect=bidx&kw=&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷涓爣鍏憡 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); + zhongbiaogonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage); + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷涓爣鍏憡 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); + #endregion + //Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 + #region 鎴愪氦鍏憡 + currPage = 1; + sichuanpageurll = "https://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=11&dbselect=bidx&kw=&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷鎴愪氦鍏憡 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); + zhongbiaogonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage); + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷鎴愪氦鍏憡 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); + #endregion + //Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 + #region 鏇存鍏憡 + currPage = 1; + sichuanpageurll = "https://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=8&dbselect=bidx&kw=&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷鏇存鍏憡 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); + gengzhenggonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage); + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷鏇存鍏憡 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); + #endregion + //Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 + #region 搴熸爣鍏憡 + currPage = 1; + sichuanpageurll = "https://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=12&dbselect=bidx&kw=&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷搴熸爣鍏憡 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); + feibiaogonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage); + logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷搴熸爣鍏憡 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); + #endregion + + + + //淇濆瓨鏈�鍚庢洿鏂版椂闂� + if (updatetime != null) + { + updatetime.Updatetime1 = datenow; + _ccontext.SaveChanges(); + } + else + { + updatetime = new Updatetime(); + updatetime.Id = Guid.NewGuid().ToString(); + updatetime.Sheng = "涓浗鏀块噰缃戝洓宸濈渷"; + updatetime.Updatetime1 = datenow; + _ccontext.Updatetimes.Add(updatetime); + _ccontext.SaveChanges(); + } + + Thread.CurrentThread.Join(1000 * 60 * 60);//闃绘璁惧畾鏃堕棿 } - else - { - operationStartTime = "2021-12-01"; //DateTime.Now.AddYears(-1).ToString("yyyy-MM-dd"); - operationEndTime = "2022-12-01";//DateTime.Now.ToString("yyyy-MM-dd"); - } - var currPage = 1; - string sichuanpageurll = ""; - - #region 鎰忓悜鍏紑 - currPage = 1; - sichuanpageurll = "http://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=9&dbselect=bidx&kw=%E6%84%8F%E5%90%91&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷鎰忓悜鍏紑 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); - yixianggonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage); - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷鎰忓悜鍏紑 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); - #endregion - Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 - - - #region 鍏紑鎷涙爣 - currPage = 1; - sichuanpageurll = "http://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=1&dbselect=bidx&kw=&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 鍏紑鎷涙爣 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); - caigougonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage, "1", "鍏紑鎷涙爣"); - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 鍏紑鎷涙爣 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); - #endregion - Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 - - #region 璇环 - currPage = 1; - sichuanpageurll = "http://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=2&dbselect=bidx&kw=&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 璇环 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); - caigougonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage, "4", "璇环"); - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 璇环 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); - #endregion - Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 - - #region 绔炰簤鎬ц皥鍒� - currPage = 1; - sichuanpageurll = "http://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=3&dbselect=bidx&kw=&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 绔炰簤鎬ц皥鍒� 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); - caigougonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage, "5", "绔炰簤鎬ц皥鍒�"); - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 绔炰簤鎬ц皥鍒� 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); - #endregion - Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 - - #region 鍗曚竴鏉ユ簮 - currPage = 1; - sichuanpageurll = "http://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=4&dbselect=bidx&kw=&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 鍗曚竴鏉ユ簮 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); - caigougonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage, "7", "鍗曚竴鏉ユ簮"); - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 鍗曚竴鏉ユ簮 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); - #endregion - Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 - - #region 閭�璇锋嫑鏍� - currPage = 1; - sichuanpageurll = "http://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=6&dbselect=bidx&kw=&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 閭�璇锋嫑鏍� 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); - caigougonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage, "2", "閭�璇锋嫑鏍�"); - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 閭�璇锋嫑鏍� 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); - #endregion - Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 - - #region 绔炰簤鎬х鍟� - currPage = 1; - sichuanpageurll = "http://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=10&dbselect=bidx&kw=&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 绔炰簤鎬х鍟� 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); - caigougonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage, "3", "绔炰簤鎬х鍟�"); - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷閲囪喘鍏憡 绔炰簤鎬х鍟� 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); - #endregion - Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 - - - #region 涓爣鍏憡 - currPage = 1; - sichuanpageurll = "http://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=7&dbselect=bidx&kw=&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷涓爣鍏憡 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); - zhongbiaogonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage); - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷涓爣鍏憡 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); - #endregion - Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 - - #region 鎴愪氦鍏憡 - currPage = 1; - sichuanpageurll = "http://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=11&dbselect=bidx&kw=&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷鎴愪氦鍏憡 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); - zhongbiaogonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage); - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷鎴愪氦鍏憡 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); - #endregion - Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 - - #region 鏇存鍏憡 - currPage = 1; - sichuanpageurll = "http://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=8&dbselect=bidx&kw=&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷鏇存鍏憡 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); - gengzhenggonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime, currPage); - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷鏇存鍏憡 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); - #endregion - Thread.CurrentThread.Join(1000 * 60 * 30);//闃绘璁惧畾鏃堕棿 - - #region 搴熸爣鍏憡 - currPage = 1; - sichuanpageurll = "http://search.ccgp.gov.cn/bxsearch?searchtype=1&page_index=1&bidSort=0&buyerName=&projectId=&pinMu=0&bidType=12&dbselect=bidx&kw=&start_time=&end_time=&timeType=6&displayZone=%E5%9B%9B%E5%B7%9D&zoneId=51&pppStatus=&agentName="; - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷搴熸爣鍏憡 寮�濮嬭幏鍙�", "涓浗鏀块噰缃戝洓宸濈渷"); - feibiaogonggao(_ccontext, sichuanpageurll, operationStartTime, operationEndTime , currPage); - logg.WriteLog("涓浗鏀块噰缃戝洓宸濈渷搴熸爣鍏憡 寮�濮嬭幏鍙栫粨鏉�", "涓浗鏀块噰缃戝洓宸濈渷"); - #endregion - - - - //淇濆瓨鏈�鍚庢洿鏂版椂闂� - if (updatetime != null) - { - updatetime.Updatetime1 = DateTime.Now.AddDays(-1); - _ccontext.SaveChanges(); - } - else - { - updatetime = new Updatetime(); - updatetime.Id = Guid.NewGuid().ToString(); - updatetime.Sheng = "涓浗鏀块噰缃戝洓宸濈渷"; - updatetime.Updatetime1 = DateTime.Now.AddDays(-1); - _ccontext.Updatetimes.Add(updatetime); - _ccontext.SaveChanges(); - } + //datenow = DateTime.Now.Date.AddDays(1); + } catch (Exception ex) { @@ -174,7 +175,7 @@ //閲囪喘鍏憡 - public static void caigougonggao(WebCrawlerContext _ccontext, string sichuanpageurll, string startTime, string endTime , int page, string cgfs, string cgfsName) + public static async void caigougonggao(WebCrawlerContext _ccontext, string sichuanpageurll, string startTime, string endTime , int page, string cgfs, string cgfsName) { sichuanpageurll = sichuanpageurll.Replace("start_time=", "start_time=" + startTime.Replace("-", "%3A")); @@ -183,17 +184,41 @@ var list11 = new List<AngleSharp.Dom.IElement>(); while (true) { - Thread.CurrentThread.Join(1000 * 5);//闃绘璁惧畾鏃堕棿 + Thread.CurrentThread.Join(1000 * 10);//闃绘璁惧畾鏃堕棿 logg.WriteLog(page.ToString(), "涓浗鏀块噰缃戝洓宸濈渷"); string sichuanpageurl2 = sichuanpageurll.Replace("page_index=1", "page_index=" + page); try { - using (HttpClient client = new HttpClient()) + HttpClientHandler handler = new HttpClientHandler(); + handler.CookieContainer = new CookieContainer(); + using (HttpClient client = new HttpClient(handler)) { client.Timeout = TimeSpan.FromSeconds(60); + client.DefaultRequestHeaders.Add("Accept", "*/*"); + client.DefaultRequestHeaders.Add("Accept-Encoding", "gzip, deflate, br"); + client.DefaultRequestHeaders.Add("Host", "search.ccgp.gov.cn"); //List<sichuanjieshoudtl> data = new List<sichuanjieshoudtl>(); HttpResponseMessage response = client.GetAsync(sichuanpageurl2).Result; - var res = response.Content.ReadAsStringAsync().Result; + string res = ""; + if (response.IsSuccessStatusCode) + { + using (var responseStream = await response.Content.ReadAsStreamAsync()) + { + using (var decompressedStream = new GZipStream(responseStream, CompressionMode.Decompress)) + { + using (var reader = new StreamReader(decompressedStream)) + { + res = await reader.ReadToEndAsync(); + // 澶勭悊瑙e帇缂╁悗鐨勫搷搴斿唴瀹� + } + } + } + } + else + { + // 澶勭悊璇锋眰澶辫触鐨勬儏鍐� + } + //var res = response.Content.ReadAsStringAsync().Result; var document = parser.ParseDocument(res); var sssdfsdfsd = document.All.Where(m => m.ClassName == "vT-srch-result-list").FirstOrDefault(); var contentList = sssdfsdfsd.QuerySelector("ul"); @@ -232,15 +257,50 @@ list11 = lists.ToList(); } + string[] bbb = null; + var scriptElements = document.QuerySelectorAll("script"); + foreach (var scriptElement in scriptElements) + { + var scriptText = scriptElement.TextContent; + if (scriptText.Contains("var ohtmlurls")) // 妫�鏌ユ枃鏈腑鏄惁鍖呭惈鐗瑰畾鍙橀噺 + { + + int startIndex = scriptText.IndexOf('"') + 1; + int endIndex = scriptText.IndexOf('"', startIndex); + if (startIndex > 0 && endIndex > startIndex) + { + var aaaa = scriptText.Substring(startIndex, endIndex - startIndex); + if (!string.IsNullOrEmpty(aaaa)) + { + bbb = aaaa.Split(','); + } + } + } + } + + + + + int ccc = 0; foreach (var sichuanjieshoudtl1 in lists) { - Thread.CurrentThread.Join(1000 * 5);//闃绘璁惧畾鏃堕棿 + Thread.CurrentThread.Join(1000 * 10);//闃绘璁惧畾鏃堕棿 try { sichuanjieshoudtl aaaaaaaa = new sichuanjieshoudtl(); - aaaaaaaa.pageurl = sichuanjieshoudtl1.QuerySelector("a").GetAttribute("href"); + if(bbb!=null && bbb.Length>= ccc && !string.IsNullOrEmpty(bbb[ccc])) + { + aaaaaaaa.pageurl = bbb[ccc]; + ccc = ccc + 1; + } + else + { + ccc = ccc + 1; + continue; + } + // aaaaaaaa.pageurl = sichuanjieshoudtl1.QuerySelector("a").GetAttribute("href"); logg.WriteLog(aaaaaaaa.pageurl, "涓浗鏀块噰缃戝洓宸濈渷"); aaaaaaaa.title = sichuanjieshoudtl1.QuerySelector("a").TextContent.Replace("\n", "").Trim(); @@ -265,7 +325,10 @@ } using (HttpClient clientdtl = new HttpClient()) { - var zhengfuProjectcount = _ccontext.ZhengfuProjects.Count(x => (x.RecStatus == "A" && x.Pageurl == aaaaaaaa.pageurl) || (x.RecStatus == "A" && x.Title == aaaaaaaa.title && x.Gglx == "02" && x.OpenTenderTime == DateTime.Parse(NoticeTime))); + var notime = DateTime.Parse(NoticeTime).ToString("yyyy-MM-dd"); + var notimestart = DateTime.Parse(notime); + var notimeend = notimestart.AddDays(1); + var zhengfuProjectcount = _ccontext.ZhengfuProjects.Count(x => (x.RecStatus == "A" && x.Pageurl == aaaaaaaa.pageurl) || (x.RecStatus == "A" && x.Title == aaaaaaaa.title && x.Gglx == "02" && x.NoticeTime >= notimestart && x.NoticeTime < notimeend) ); if (zhengfuProjectcount <= 0) @@ -289,7 +352,7 @@ { var ssss = fujianya.Id; var sssss = fujianya.InnerHtml; - fujianhtml += "<tr><td class=\"bid_attachtab_content\">闄勪欢涓嬭浇锛�<a class=\"bizDownload\" target=\"_blank\" href =\"http://download.ccgp.gov.cn/oss/download?uuid=" + ssss + "\" id=\"0E1723104D34335C527765FF6CD28A\" title=\"鐐瑰嚮涓嬭浇\">" + sssss + "</a><br></td></tr>"; + fujianhtml += "<tr><td class=\"bid_attachtab_content\">闄勪欢涓嬭浇锛�<a class=\"bizDownload\" target=\"_blank\" href =\"https://download.ccgp.gov.cn/oss/download?uuid=" + ssss + "\" id=\"0E1723104D34335C527765FF6CD28A\" title=\"鐐瑰嚮涓嬭浇\">" + sssss + "</a><br></td></tr>"; } fujianhtml += "</tbody></table></div>"; @@ -498,7 +561,7 @@ _ccontext.SaveChanges(); - Thread.CurrentThread.Join(1000 * 5);//闃绘璁惧畾鏃堕棿 + Thread.CurrentThread.Join(1000 * 10);//闃绘璁惧畾鏃堕棿 } } } @@ -515,6 +578,8 @@ } } } + + page += 1; } catch (Exception ex) { @@ -523,9 +588,10 @@ logg.WriteLog(ex.StackTrace, "涓浗鏀块噰缃戝洓宸濈渷"); logg.WriteLog(ex.ToString(), "涓浗鏀块噰缃戝洓宸濈渷"); + Thread.CurrentThread.Join(1000 * 60 * 5);//闃绘璁惧畾鏃堕棿 } - page += 1; + } @@ -539,7 +605,7 @@ //鎰忓悜鍏紑 - public static void yixianggonggao(WebCrawlerContext _ccontext, string sichuanpageurll, string startTime,string endTime, int page) + public static async void yixianggonggao(WebCrawlerContext _ccontext, string sichuanpageurll, string startTime,string endTime, int page) { sichuanpageurll = sichuanpageurll.Replace("start_time=", "start_time=" + startTime.Replace("-", "%3A")); @@ -548,17 +614,41 @@ var list11 = new List<AngleSharp.Dom.IElement>(); while (true) { - Thread.CurrentThread.Join(1000 * 5);//闃绘璁惧畾鏃堕棿 + Thread.CurrentThread.Join(1000 * 10);//闃绘璁惧畾鏃堕棿 logg.WriteLog(page.ToString(), "涓浗鏀块噰缃戝洓宸濈渷"); string sichuanpageurl2 = sichuanpageurll.Replace("page_index=1", "page_index=" + page); try { - using (HttpClient client = new HttpClient()) + HttpClientHandler handler = new HttpClientHandler(); + handler.CookieContainer = new CookieContainer(); + using (HttpClient client = new HttpClient(handler)) { client.Timeout = TimeSpan.FromSeconds(60); + client.DefaultRequestHeaders.Add("Accept", "*/*"); + client.DefaultRequestHeaders.Add("Accept-Encoding", "gzip, deflate, br"); + client.DefaultRequestHeaders.Add("Host", "search.ccgp.gov.cn"); //List<sichuanjieshoudtl> data = new List<sichuanjieshoudtl>(); HttpResponseMessage response = client.GetAsync(sichuanpageurl2).Result; - var res = response.Content.ReadAsStringAsync().Result; + //var res = response.Content.ReadAsStringAsync().Result; + string res = ""; + if (response.IsSuccessStatusCode) + { + using (var responseStream = await response.Content.ReadAsStreamAsync()) + { + using (var decompressedStream = new GZipStream(responseStream, CompressionMode.Decompress)) + { + using (var reader = new StreamReader(decompressedStream)) + { + res = await reader.ReadToEndAsync(); + // 澶勭悊瑙e帇缂╁悗鐨勫搷搴斿唴瀹� + } + } + } + } + else + { + // 澶勭悊璇锋眰澶辫触鐨勬儏鍐� + } var document = parser.ParseDocument(res); var sssdfsdfsd = document.All.Where(m => m.ClassName == "vT-srch-result-list").FirstOrDefault(); var contentList = sssdfsdfsd.QuerySelector("ul"); @@ -596,13 +686,51 @@ { list11 = lists.ToList(); } + + string[] bbb = null; + var scriptElements = document.QuerySelectorAll("script"); + foreach (var scriptElement in scriptElements) + { + var scriptText = scriptElement.TextContent; + + if (scriptText.Contains("var ohtmlurls")) // 妫�鏌ユ枃鏈腑鏄惁鍖呭惈鐗瑰畾鍙橀噺 + { + + int startIndex = scriptText.IndexOf('"') + 1; + int endIndex = scriptText.IndexOf('"', startIndex); + if (startIndex > 0 && endIndex > startIndex) + { + var aaaa = scriptText.Substring(startIndex, endIndex - startIndex); + if (!string.IsNullOrEmpty(aaaa)) + { + bbb = aaaa.Split(','); + } + } + } + } + + + + + int ccc = 0; + foreach (var sichuanjieshoudtl1 in lists) { - Thread.CurrentThread.Join(1000 * 5);//闃绘璁惧畾鏃堕棿 + Thread.CurrentThread.Join(1000 * 10);//闃绘璁惧畾鏃堕棿 try { sichuanjieshoudtl aaaaaaaa = new sichuanjieshoudtl(); - aaaaaaaa.pageurl = sichuanjieshoudtl1.QuerySelector("a").GetAttribute("href"); + if (bbb != null && bbb.Length >= ccc && !string.IsNullOrEmpty(bbb[ccc])) + { + aaaaaaaa.pageurl = bbb[ccc]; + ccc = ccc + 1; + } + else + { + ccc = ccc + 1; + continue; + } + // aaaaaaaa.pageurl = sichuanjieshoudtl1.QuerySelector("a").GetAttribute("href"); logg.WriteLog(aaaaaaaa.pageurl, "涓浗鏀块噰缃戝洓宸濈渷"); aaaaaaaa.title = sichuanjieshoudtl1.QuerySelector("a").TextContent.Replace("\n", "").Trim(); @@ -627,7 +755,10 @@ } using (HttpClient clientdtl = new HttpClient()) { - var zhengfuProjectCount = _ccontext.ZhengfuProjects.Count(x => (x.RecStatus == "A" && x.Pageurl == aaaaaaaa.pageurl) || (x.RecStatus == "A" && x.Title == aaaaaaaa.title && x.Gglx == "01" && x.OpenTenderTime == DateTime.Parse(NoticeTime))); + var notime = DateTime.Parse(NoticeTime).ToString("yyyy-MM-dd"); + var notimestart = DateTime.Parse(notime); + var notimeend = notimestart.AddDays(1); + var zhengfuProjectCount = _ccontext.ZhengfuProjects.Count(x => (x.RecStatus == "A" && x.Pageurl == aaaaaaaa.pageurl) || (x.RecStatus == "A" && x.Title == aaaaaaaa.title && x.Gglx == "01" && x.NoticeTime >= notimestart && x.NoticeTime < notimeend)); if (zhengfuProjectCount <= 0) { clientdtl.Timeout = TimeSpan.FromSeconds(60); @@ -649,7 +780,7 @@ { var ssss = fujianya.Id; var sssss = fujianya.InnerHtml; - fujianhtml += "<tr><td class=\"bid_attachtab_content\">闄勪欢涓嬭浇锛�<a class=\"bizDownload\" target=\"_blank\" href =\"http://download.ccgp.gov.cn/oss/download?uuid=" + ssss + "\" id=\"0E1723104D34335C527765FF6CD28A\" title=\"鐐瑰嚮涓嬭浇\">" + sssss + "</a><br></td></tr>"; + fujianhtml += "<tr><td class=\"bid_attachtab_content\">闄勪欢涓嬭浇锛�<a class=\"bizDownload\" target=\"_blank\" href =\"https://download.ccgp.gov.cn/oss/download?uuid=" + ssss + "\" id=\"0E1723104D34335C527765FF6CD28A\" title=\"鐐瑰嚮涓嬭浇\">" + sssss + "</a><br></td></tr>"; } fujianhtml += "</tbody></table></div>"; @@ -672,8 +803,8 @@ { if (td[i].TextContent.IndexOf("椤圭洰缂栧彿锛�") >= 0) { - - OpenTenderCode = td[i].TextContent.Replace("椤圭洰缂栧彿锛�", "").Replace("/", "").Trim(); + if (string.IsNullOrEmpty(OpenTenderCode)) + OpenTenderCode = td[i].TextContent.Replace("椤圭洰缂栧彿锛�", "").Replace("/", "").Trim(); } if (td[i].TextContent.IndexOf("棰勭畻閲戦锛�") >= 0) @@ -806,7 +937,7 @@ _ccontext.SaveChanges(); - Thread.CurrentThread.Join(1000 * 5);//闃绘璁惧畾鏃堕棿 + Thread.CurrentThread.Join(1000 * 10);//闃绘璁惧畾鏃堕棿 } } } @@ -823,6 +954,8 @@ } } } + + page += 1; } catch (Exception ex) { @@ -831,10 +964,11 @@ logg.WriteLog(ex.StackTrace, "涓浗鏀块噰缃戝洓宸濈渷"); logg.WriteLog(ex.ToString(), "涓浗鏀块噰缃戝洓宸濈渷"); + Thread.CurrentThread.Join(1000 * 60 * 5);//闃绘璁惧畾鏃堕棿 } - page += 1; + } @@ -848,7 +982,7 @@ //缁撴灉鍏憡 - public static void zhongbiaogonggao(WebCrawlerContext _ccontext, string sichuanpageurll, string startTime, string endTime , int page) + public static async void zhongbiaogonggao(WebCrawlerContext _ccontext, string sichuanpageurll, string startTime, string endTime , int page) { sichuanpageurll = sichuanpageurll.Replace("start_time=", "start_time=" + startTime.Replace("-", "%3A")); @@ -857,17 +991,41 @@ var list11 = new List<AngleSharp.Dom.IElement>(); while (true) { - Thread.CurrentThread.Join(1000 * 5);//闃绘璁惧畾鏃堕棿 + Thread.CurrentThread.Join(1000 * 10);//闃绘璁惧畾鏃堕棿 logg.WriteLog(page.ToString(), "涓浗鏀块噰缃戝洓宸濈渷"); string sichuanpageurl2 = sichuanpageurll.Replace("page_index=1", "page_index=" + page); try { - using (HttpClient client = new HttpClient()) + HttpClientHandler handler = new HttpClientHandler(); + handler.CookieContainer = new CookieContainer(); + using (HttpClient client = new HttpClient(handler)) { client.Timeout = TimeSpan.FromSeconds(60); + client.DefaultRequestHeaders.Add("Accept", "*/*"); + client.DefaultRequestHeaders.Add("Accept-Encoding", "gzip, deflate, br"); + client.DefaultRequestHeaders.Add("Host", "search.ccgp.gov.cn"); //List<sichuanjieshoudtl> data = new List<sichuanjieshoudtl>(); HttpResponseMessage response = client.GetAsync(sichuanpageurl2).Result; - var res = response.Content.ReadAsStringAsync().Result; + //var res = response.Content.ReadAsStringAsync().Result; + string res = ""; + if (response.IsSuccessStatusCode) + { + using (var responseStream = await response.Content.ReadAsStreamAsync()) + { + using (var decompressedStream = new GZipStream(responseStream, CompressionMode.Decompress)) + { + using (var reader = new StreamReader(decompressedStream)) + { + res = await reader.ReadToEndAsync(); + // 澶勭悊瑙e帇缂╁悗鐨勫搷搴斿唴瀹� + } + } + } + } + else + { + // 澶勭悊璇锋眰澶辫触鐨勬儏鍐� + } var document = parser.ParseDocument(res); var sssdfsdfsd = document.All.Where(m => m.ClassName == "vT-srch-result-list").FirstOrDefault(); var contentList = sssdfsdfsd.QuerySelector("ul"); @@ -906,13 +1064,50 @@ list11 = lists.ToList(); } + string[] bbb = null; + var scriptElements = document.QuerySelectorAll("script"); + foreach (var scriptElement in scriptElements) + { + var scriptText = scriptElement.TextContent; + + if (scriptText.Contains("var ohtmlurls")) // 妫�鏌ユ枃鏈腑鏄惁鍖呭惈鐗瑰畾鍙橀噺 + { + + int startIndex = scriptText.IndexOf('"') + 1; + int endIndex = scriptText.IndexOf('"', startIndex); + if (startIndex > 0 && endIndex > startIndex) + { + var aaaa = scriptText.Substring(startIndex, endIndex - startIndex); + if (!string.IsNullOrEmpty(aaaa)) + { + bbb = aaaa.Split(','); + } + } + } + } + + + + + int ccc = 0; + foreach (var sichuanjieshoudtl1 in lists) { - Thread.CurrentThread.Join(1000 * 5);//闃绘璁惧畾鏃堕棿 + Thread.CurrentThread.Join(1000 * 10);//闃绘璁惧畾鏃堕棿 try { sichuanjieshoudtl aaaaaaaa = new sichuanjieshoudtl(); - aaaaaaaa.pageurl = sichuanjieshoudtl1.QuerySelector("a").GetAttribute("href"); + if (bbb != null && bbb.Length >= ccc && !string.IsNullOrEmpty(bbb[ccc])) + { + aaaaaaaa.pageurl = bbb[ccc]; + ccc = ccc + 1; + } + else + { + ccc = ccc + 1; + continue; + } + // aaaaaaaa.pageurl = sichuanjieshoudtl1.QuerySelector("a").GetAttribute("href"); logg.WriteLog(aaaaaaaa.pageurl, "涓浗鏀块噰缃戝洓宸濈渷"); aaaaaaaa.title = sichuanjieshoudtl1.QuerySelector("a").TextContent.Replace("\n", "").Trim(); @@ -937,7 +1132,10 @@ } using (HttpClient clientdtl = new HttpClient()) { - var zhengfuProjectCount = _ccontext.ZhengfuProjects.Count(x => (x.RecStatus == "A" && x.Pageurl == aaaaaaaa.pageurl) || (x.RecStatus == "A" && x.Title == aaaaaaaa.title && x.Gglx == "03" && x.OpenTenderTime == DateTime.Parse(NoticeTime))); + var notime = DateTime.Parse(NoticeTime).ToString("yyyy-MM-dd"); + var notimestart = DateTime.Parse(notime); + var notimeend = notimestart.AddDays(1); + var zhengfuProjectCount = _ccontext.ZhengfuProjects.Count(x => (x.RecStatus == "A" && x.Pageurl == aaaaaaaa.pageurl) || (x.RecStatus == "A" && x.Title == aaaaaaaa.title && x.Gglx == "03" && x.NoticeTime >= notimestart && x.NoticeTime < notimeend)); if (zhengfuProjectCount <= 0) { clientdtl.Timeout = TimeSpan.FromSeconds(60); @@ -959,7 +1157,7 @@ { var ssss = fujianya.Id; var sssss = fujianya.InnerHtml; - fujianhtml += "<tr><td class=\"bid_attachtab_content\">闄勪欢涓嬭浇锛�<a class=\"bizDownload\" target=\"_blank\" href =\"http://download.ccgp.gov.cn/oss/download?uuid=" + ssss + "\" id=\"0E1723104D34335C527765FF6CD28A\" title=\"鐐瑰嚮涓嬭浇\">" + sssss + "</a><br></td></tr>"; + fujianhtml += "<tr><td class=\"bid_attachtab_content\">闄勪欢涓嬭浇锛�<a class=\"bizDownload\" target=\"_blank\" href =\"https://download.ccgp.gov.cn/oss/download?uuid=" + ssss + "\" id=\"0E1723104D34335C527765FF6CD28A\" title=\"鐐瑰嚮涓嬭浇\">" + sssss + "</a><br></td></tr>"; } fujianhtml += "</tbody></table></div>"; @@ -982,8 +1180,8 @@ { if (td[i].TextContent.IndexOf("椤圭洰缂栧彿锛�") >= 0) { - - OpenTenderCode = td[i].TextContent.Replace("椤圭洰缂栧彿锛�", "").Replace("涓�銆�", "").Trim(); + if (string.IsNullOrEmpty(OpenTenderCode)) + OpenTenderCode = td[i].TextContent.Replace("椤圭洰缂栧彿锛�", "").Replace("涓�銆�", "").Trim(); if (OpenTenderCode.IndexOf("锛堟嫑鏍囨枃浠剁紪鍙�") >= 0) { @@ -1030,8 +1228,8 @@ { if (td[i].TextContent.IndexOf("椤圭洰缂栧彿锛�") >= 0) { - - OpenTenderCode = td[i].TextContent.Replace("椤圭洰缂栧彿锛�", "").Replace("涓�銆�", "").Trim(); + if (string.IsNullOrEmpty(OpenTenderCode)) + OpenTenderCode = td[i].TextContent.Replace("椤圭洰缂栧彿锛�", "").Replace("涓�銆�", "").Trim(); if (OpenTenderCode.IndexOf("锛堟嫑鏍囨枃浠剁紪鍙�") >= 0) { @@ -1169,7 +1367,7 @@ _ccontext.SaveChanges(); - Thread.CurrentThread.Join(1000 * 5);//闃绘璁惧畾鏃堕棿 + Thread.CurrentThread.Join(1000 * 10);//闃绘璁惧畾鏃堕棿 } } } @@ -1186,6 +1384,7 @@ } } } + page += 1; } catch (Exception ex) { @@ -1194,10 +1393,11 @@ logg.WriteLog(ex.StackTrace, "涓浗鏀块噰缃戝洓宸濈渷"); logg.WriteLog(ex.ToString(), "涓浗鏀块噰缃戝洓宸濈渷"); + Thread.CurrentThread.Join(1000 * 60 * 5);//闃绘璁惧畾鏃堕棿 } - page += 1; + } @@ -1210,7 +1410,7 @@ } //鏇存鍏憡 - public static void gengzhenggonggao(WebCrawlerContext _ccontext, string sichuanpageurll, string startTime, string endTime , int page) + public static async void gengzhenggonggao(WebCrawlerContext _ccontext, string sichuanpageurll, string startTime, string endTime , int page) { sichuanpageurll = sichuanpageurll.Replace("start_time=", "start_time=" + startTime.Replace("-", "%3A")); @@ -1219,17 +1419,41 @@ var list11 = new List<AngleSharp.Dom.IElement>(); while (true) { - Thread.CurrentThread.Join(1000 * 5);//闃绘璁惧畾鏃堕棿 + Thread.CurrentThread.Join(1000 * 10);//闃绘璁惧畾鏃堕棿 logg.WriteLog(page.ToString(), "涓浗鏀块噰缃戝洓宸濈渷"); string sichuanpageurl2 = sichuanpageurll.Replace("page_index=1", "page_index=" + page); try { - using (HttpClient client = new HttpClient()) + HttpClientHandler handler = new HttpClientHandler(); + handler.CookieContainer = new CookieContainer(); + using (HttpClient client = new HttpClient(handler)) { client.Timeout = TimeSpan.FromSeconds(60); + client.DefaultRequestHeaders.Add("Accept", "*/*"); + client.DefaultRequestHeaders.Add("Accept-Encoding", "gzip, deflate, br"); + client.DefaultRequestHeaders.Add("Host", "search.ccgp.gov.cn"); //List<sichuanjieshoudtl> data = new List<sichuanjieshoudtl>(); HttpResponseMessage response = client.GetAsync(sichuanpageurl2).Result; - var res = response.Content.ReadAsStringAsync().Result; + //var res = response.Content.ReadAsStringAsync().Result; + string res = ""; + if (response.IsSuccessStatusCode) + { + using (var responseStream = await response.Content.ReadAsStreamAsync()) + { + using (var decompressedStream = new GZipStream(responseStream, CompressionMode.Decompress)) + { + using (var reader = new StreamReader(decompressedStream)) + { + res = await reader.ReadToEndAsync(); + // 澶勭悊瑙e帇缂╁悗鐨勫搷搴斿唴瀹� + } + } + } + } + else + { + // 澶勭悊璇锋眰澶辫触鐨勬儏鍐� + } var document = parser.ParseDocument(res); var sssdfsdfsd = document.All.Where(m => m.ClassName == "vT-srch-result-list").FirstOrDefault(); var contentList = sssdfsdfsd.QuerySelector("ul"); @@ -1268,13 +1492,50 @@ list11 = lists.ToList(); } + string[] bbb = null; + var scriptElements = document.QuerySelectorAll("script"); + foreach (var scriptElement in scriptElements) + { + var scriptText = scriptElement.TextContent; + + if (scriptText.Contains("var ohtmlurls")) // 妫�鏌ユ枃鏈腑鏄惁鍖呭惈鐗瑰畾鍙橀噺 + { + + int startIndex = scriptText.IndexOf('"') + 1; + int endIndex = scriptText.IndexOf('"', startIndex); + if (startIndex > 0 && endIndex > startIndex) + { + var aaaa = scriptText.Substring(startIndex, endIndex - startIndex); + if (!string.IsNullOrEmpty(aaaa)) + { + bbb = aaaa.Split(','); + } + } + } + } + + + + + int ccc = 0; + foreach (var sichuanjieshoudtl1 in lists) { - Thread.CurrentThread.Join(1000 * 5);//闃绘璁惧畾鏃堕棿 + Thread.CurrentThread.Join(1000 * 10);//闃绘璁惧畾鏃堕棿 try { sichuanjieshoudtl aaaaaaaa = new sichuanjieshoudtl(); - aaaaaaaa.pageurl = sichuanjieshoudtl1.QuerySelector("a").GetAttribute("href"); + if (bbb != null && bbb.Length >= ccc && !string.IsNullOrEmpty(bbb[ccc])) + { + aaaaaaaa.pageurl = bbb[ccc]; + ccc = ccc + 1; + } + else + { + ccc = ccc + 1; + continue; + } + // aaaaaaaa.pageurl = sichuanjieshoudtl1.QuerySelector("a").GetAttribute("href"); logg.WriteLog(aaaaaaaa.pageurl, "涓浗鏀块噰缃戝洓宸濈渷"); aaaaaaaa.title = sichuanjieshoudtl1.QuerySelector("a").TextContent.Replace("\n", "").Trim(); @@ -1299,7 +1560,10 @@ } using (HttpClient clientdtl = new HttpClient()) { - var zhengfuProjectCount = _ccontext.ZhengfuProjects.Count(x => (x.RecStatus == "A" && x.Pageurl == aaaaaaaa.pageurl) || (x.RecStatus == "A" && x.Title == aaaaaaaa.title && x.Gglx == "04" && x.OpenTenderTime == DateTime.Parse(NoticeTime))); + var notime = DateTime.Parse(NoticeTime).ToString("yyyy-MM-dd"); + var notimestart = DateTime.Parse(notime); + var notimeend = notimestart.AddDays(1); + var zhengfuProjectCount = _ccontext.ZhengfuProjects.Count(x => (x.RecStatus == "A" && x.Pageurl == aaaaaaaa.pageurl) || (x.RecStatus == "A" && x.Title == aaaaaaaa.title && x.Gglx == "04" && x.NoticeTime >= notimestart && x.NoticeTime < notimeend)); if (zhengfuProjectCount <= 0) { clientdtl.Timeout = TimeSpan.FromSeconds(60); @@ -1321,7 +1585,7 @@ { var ssss = fujianya.Id; var sssss = fujianya.InnerHtml; - fujianhtml += "<tr><td class=\"bid_attachtab_content\">闄勪欢涓嬭浇锛�<a class=\"bizDownload\" target=\"_blank\" href =\"http://download.ccgp.gov.cn/oss/download?uuid=" + ssss + "\" id=\"0E1723104D34335C527765FF6CD28A\" title=\"鐐瑰嚮涓嬭浇\">" + sssss + "</a><br></td></tr>"; + fujianhtml += "<tr><td class=\"bid_attachtab_content\">闄勪欢涓嬭浇锛�<a class=\"bizDownload\" target=\"_blank\" href =\"https://download.ccgp.gov.cn/oss/download?uuid=" + ssss + "\" id=\"0E1723104D34335C527765FF6CD28A\" title=\"鐐瑰嚮涓嬭浇\">" + sssss + "</a><br></td></tr>"; } fujianhtml += "</tbody></table></div>"; @@ -1344,8 +1608,8 @@ { if (td[i].TextContent.IndexOf("鍘熷叕鍛婄殑閲囪喘椤圭洰缂栧彿锛�") >= 0) { - - OpenTenderCode = td[i].TextContent.Replace("鍘熷叕鍛婄殑閲囪喘椤圭洰缂栧彿锛�", "").Trim(); + if (string.IsNullOrEmpty(OpenTenderCode)) + OpenTenderCode = td[i].TextContent.Replace("鍘熷叕鍛婄殑閲囪喘椤圭洰缂栧彿锛�", "").Trim(); } if (td[i].TextContent.IndexOf("棰勭畻閲戦锛�") >= 0) @@ -1477,7 +1741,7 @@ _ccontext.SaveChanges(); - Thread.CurrentThread.Join(1000 * 5);//闃绘璁惧畾鏃堕棿 + Thread.CurrentThread.Join(1000 * 10);//闃绘璁惧畾鏃堕棿 } } } @@ -1494,6 +1758,8 @@ } } } + + page += 1; } catch (Exception ex) { @@ -1502,10 +1768,11 @@ logg.WriteLog(ex.StackTrace, "涓浗鏀块噰缃戝洓宸濈渷"); logg.WriteLog(ex.ToString(), "涓浗鏀块噰缃戝洓宸濈渷"); + Thread.CurrentThread.Join(1000 * 60 * 5);//闃绘璁惧畾鏃堕棿 } - page += 1; + } @@ -1519,7 +1786,7 @@ //搴熸爣鍏憡 - public static void feibiaogonggao(WebCrawlerContext _ccontext, string sichuanpageurll, string startTime, string endTime , int page) + public static async void feibiaogonggao(WebCrawlerContext _ccontext, string sichuanpageurll, string startTime, string endTime , int page) { sichuanpageurll = sichuanpageurll.Replace("start_time=", "start_time=" + startTime.Replace("-", "%3A")); @@ -1528,17 +1795,41 @@ var list11 = new List<AngleSharp.Dom.IElement>(); while (true) { - Thread.CurrentThread.Join(1000 * 5);//闃绘璁惧畾鏃堕棿 + Thread.CurrentThread.Join(1000 * 10);//闃绘璁惧畾鏃堕棿 logg.WriteLog(page.ToString(), "涓浗鏀块噰缃戝洓宸濈渷"); string sichuanpageurl2 = sichuanpageurll.Replace("page_index=1", "page_index=" + page); try { - using (HttpClient client = new HttpClient()) + HttpClientHandler handler = new HttpClientHandler(); + handler.CookieContainer = new CookieContainer(); + using (HttpClient client = new HttpClient(handler)) { //List<sichuanjieshoudtl> data = new List<sichuanjieshoudtl>(); client.Timeout = TimeSpan.FromSeconds(60); + client.DefaultRequestHeaders.Add("Accept", "*/*"); + client.DefaultRequestHeaders.Add("Accept-Encoding", "gzip, deflate, br"); + client.DefaultRequestHeaders.Add("Host", "search.ccgp.gov.cn"); HttpResponseMessage response = client.GetAsync(sichuanpageurl2).Result; - var res = response.Content.ReadAsStringAsync().Result; + //var res = response.Content.ReadAsStringAsync().Result; + string res = ""; + if (response.IsSuccessStatusCode) + { + using (var responseStream = await response.Content.ReadAsStreamAsync()) + { + using (var decompressedStream = new GZipStream(responseStream, CompressionMode.Decompress)) + { + using (var reader = new StreamReader(decompressedStream)) + { + res = await reader.ReadToEndAsync(); + // 澶勭悊瑙e帇缂╁悗鐨勫搷搴斿唴瀹� + } + } + } + } + else + { + // 澶勭悊璇锋眰澶辫触鐨勬儏鍐� + } var document = parser.ParseDocument(res); var sssdfsdfsd = document.All.Where(m => m.ClassName == "vT-srch-result-list").FirstOrDefault(); var contentList = sssdfsdfsd.QuerySelector("ul"); @@ -1577,13 +1868,51 @@ list11 = lists.ToList(); } + + string[] bbb = null; + var scriptElements = document.QuerySelectorAll("script"); + foreach (var scriptElement in scriptElements) + { + var scriptText = scriptElement.TextContent; + + if (scriptText.Contains("var ohtmlurls")) // 妫�鏌ユ枃鏈腑鏄惁鍖呭惈鐗瑰畾鍙橀噺 + { + + int startIndex = scriptText.IndexOf('"') + 1; + int endIndex = scriptText.IndexOf('"', startIndex); + if (startIndex > 0 && endIndex > startIndex) + { + var aaaa = scriptText.Substring(startIndex, endIndex - startIndex); + if (!string.IsNullOrEmpty(aaaa)) + { + bbb = aaaa.Split(','); + } + } + } + } + + + + + int ccc = 0; + foreach (var sichuanjieshoudtl1 in lists) { - Thread.CurrentThread.Join(1000 * 5);//闃绘璁惧畾鏃堕棿 + Thread.CurrentThread.Join(1000 * 10);//闃绘璁惧畾鏃堕棿 try { sichuanjieshoudtl aaaaaaaa = new sichuanjieshoudtl(); - aaaaaaaa.pageurl = sichuanjieshoudtl1.QuerySelector("a").GetAttribute("href"); + if (bbb != null && bbb.Length >= ccc && !string.IsNullOrEmpty(bbb[ccc])) + { + aaaaaaaa.pageurl = bbb[ccc]; + ccc = ccc + 1; + } + else + { + ccc = ccc + 1; + continue; + } + // aaaaaaaa.pageurl = sichuanjieshoudtl1.QuerySelector("a").GetAttribute("href"); logg.WriteLog(aaaaaaaa.pageurl, "涓浗鏀块噰缃戝洓宸濈渷"); aaaaaaaa.title = sichuanjieshoudtl1.QuerySelector("a").TextContent.Replace("\n", "").Trim(); @@ -1608,7 +1937,10 @@ } using (HttpClient clientdtl = new HttpClient()) { - var zhengfuProjectCount = _ccontext.ZhengfuProjects.Count(x => (x.RecStatus == "A" && x.Pageurl == aaaaaaaa.pageurl) || (x.RecStatus == "A" && x.Title == aaaaaaaa.title && x.Gglx == "11" && x.OpenTenderTime == DateTime.Parse(NoticeTime))); + var notime = DateTime.Parse(NoticeTime).ToString("yyyy-MM-dd"); + var notimestart = DateTime.Parse(notime); + var notimeend = notimestart.AddDays(1); + var zhengfuProjectCount = _ccontext.ZhengfuProjects.Count(x => (x.RecStatus == "A" && x.Pageurl == aaaaaaaa.pageurl) || (x.RecStatus == "A" && x.Title == aaaaaaaa.title && x.Gglx == "11" && x.NoticeTime >= notimestart && x.NoticeTime < notimeend)); if (zhengfuProjectCount <= 0) { clientdtl.Timeout = TimeSpan.FromSeconds(60); @@ -1630,7 +1962,7 @@ { var ssss = fujianya.Id; var sssss = fujianya.InnerHtml; - fujianhtml += "<tr><td class=\"bid_attachtab_content\">闄勪欢涓嬭浇锛�<a class=\"bizDownload\" target=\"_blank\" href =\"http://download.ccgp.gov.cn/oss/download?uuid=" + ssss + "\" id=\"0E1723104D34335C527765FF6CD28A\" title=\"鐐瑰嚮涓嬭浇\">" + sssss + "</a><br></td></tr>"; + fujianhtml += "<tr><td class=\"bid_attachtab_content\">闄勪欢涓嬭浇锛�<a class=\"bizDownload\" target=\"_blank\" href =\"https://download.ccgp.gov.cn/oss/download?uuid=" + ssss + "\" id=\"0E1723104D34335C527765FF6CD28A\" title=\"鐐瑰嚮涓嬭浇\">" + sssss + "</a><br></td></tr>"; } fujianhtml += "</tbody></table></div>"; @@ -1653,8 +1985,8 @@ { if (td[i].TextContent.IndexOf("閲囪喘椤圭洰缂栧彿锛�") >= 0) { - - OpenTenderCode = td[i].TextContent.Replace("閲囪喘椤圭洰缂栧彿锛�", "").Trim(); + if (string.IsNullOrEmpty(OpenTenderCode)) + OpenTenderCode = td[i].TextContent.Replace("閲囪喘椤圭洰缂栧彿锛�", "").Trim(); } if (td[i].TextContent.IndexOf("棰勭畻閲戦锛�") >= 0) @@ -1787,7 +2119,7 @@ _ccontext.SaveChanges(); - Thread.CurrentThread.Join(1000 * 5);//闃绘璁惧畾鏃堕棿 + Thread.CurrentThread.Join(1000 * 10);//闃绘璁惧畾鏃堕棿 } } } @@ -1804,6 +2136,7 @@ } } } + page += 1; } catch (Exception ex) { @@ -1812,10 +2145,11 @@ logg.WriteLog(ex.StackTrace, "涓浗鏀块噰缃戝洓宸濈渷"); logg.WriteLog(ex.ToString(), "涓浗鏀块噰缃戝洓宸濈渷"); + Thread.CurrentThread.Join(1000 * 60 * 5);//闃绘璁惧畾鏃堕棿 } - page += 1; + } -- Gitblit v1.9.1