|
@@ -80,6 +80,7 @@ type Spider struct {
|
|
IsHistoricalMend bool //是否是历史补漏爬虫
|
|
IsHistoricalMend bool //是否是历史补漏爬虫
|
|
IsMustDownload bool //是否强制下载
|
|
IsMustDownload bool //是否强制下载
|
|
IsCompete bool //区分新老爬虫
|
|
IsCompete bool //区分新老爬虫
|
|
|
|
+ Infoformat int //区分爬虫类型 1:招标;2:拟建/审批;3:产权
|
|
IsMainThread bool //是否为主线程(多线程采集时区分是否为主线程)
|
|
IsMainThread bool //是否为主线程(多线程采集时区分是否为主线程)
|
|
}
|
|
}
|
|
|
|
|
|
@@ -658,7 +659,8 @@ func (s *Spider) DownloadDetailItem(p interface{}, num *int) {
|
|
//atomic.AddInt32(&s.TodayDowncount, 1)
|
|
//atomic.AddInt32(&s.TodayDowncount, 1)
|
|
//atomic.AddInt32(&s.TotalDowncount, 1)
|
|
//atomic.AddInt32(&s.TotalDowncount, 1)
|
|
data["spidercode"] = s.Code
|
|
data["spidercode"] = s.Code
|
|
- data["iscompete"] = s.IsCompete //2021-11-01以后新增的爬虫不在展示原文链接(保存服务判断)
|
|
|
|
|
|
+ data["iscompete"] = s.IsCompete //2021-11-01以后新增的爬虫不在展示原文链接(保存服务判断)
|
|
|
|
+ data["infoformat"] = s.Infoformat //爬虫类型
|
|
Store(s.StoreMode, s.StoreToMsgEvent, s.Collection, s.CoverAttr, data, true)
|
|
Store(s.StoreMode, s.StoreToMsgEvent, s.Collection, s.CoverAttr, data, true)
|
|
}
|
|
}
|
|
|
|
|
|
@@ -995,7 +997,6 @@ func (s *Spider) DownloadDetail(reload bool, isHistory bool) {
|
|
}
|
|
}
|
|
delete(data, "exit")
|
|
delete(data, "exit")
|
|
delete(data, "checkpublishtime")
|
|
delete(data, "checkpublishtime")
|
|
- data["comeintime"] = time.Now().Unix()
|
|
|
|
//计数
|
|
//计数
|
|
//tmpsp1, b := Allspiders.Load(s.Code)
|
|
//tmpsp1, b := Allspiders.Load(s.Code)
|
|
//if b {
|
|
//if b {
|
|
@@ -1006,9 +1007,11 @@ func (s *Spider) DownloadDetail(reload bool, isHistory bool) {
|
|
// atomic.AddInt32(&sp1.TotalDowncount, 1)
|
|
// atomic.AddInt32(&sp1.TotalDowncount, 1)
|
|
// }
|
|
// }
|
|
//}
|
|
//}
|
|
|
|
+ data["comeintime"] = time.Now().Unix()
|
|
data["spidercode"] = s.Code
|
|
data["spidercode"] = s.Code
|
|
data["dataging"] = 0
|
|
data["dataging"] = 0
|
|
- data["iscompete"] = s.IsCompete //2021-11-01以后新增的爬虫不在展示原文链接(保存服务判断)
|
|
|
|
|
|
+ data["iscompete"] = s.IsCompete //2021-11-01以后新增的爬虫不在展示原文链接(保存服务判断)
|
|
|
|
+ data["infoformat"] = s.Infoformat //爬虫类型
|
|
Store(s.StoreMode, s.StoreToMsgEvent, s.Collection, s.CoverAttr, data, true)
|
|
Store(s.StoreMode, s.StoreToMsgEvent, s.Collection, s.CoverAttr, data, true)
|
|
set := map[string]interface{}{"$set": map[string]interface{}{"state": 1, "updatetime": time.Now().Unix()}} //下载成功state置为1
|
|
set := map[string]interface{}{"$set": map[string]interface{}{"state": 1, "updatetime": time.Now().Unix()}} //下载成功state置为1
|
|
update = append(update, query)
|
|
update = append(update, query)
|