maxiaoshan 3 gadi atpakaļ
vecāks
revīzija
d6287ef1ce
1 mainītis faili ar 9 papildinājumiem un 9 dzēšanām
  1. 9 9
      src/spider/spider.go

+ 9 - 9
src/spider/spider.go

@@ -549,9 +549,9 @@ func (s *Spider) DownloadDetailItem(p interface{}, num *int) {
 		isEsRepeat := false
 		if delayDay := util.Config.DelaySites[s.Name]; delayDay > 0 { //类竞品站点爬虫title做es7天内判重检验(顺序采集无法延迟,只能判重)
 			title := qu.ObjToString(paramdata["title"])
-			eTime := fmt.Sprint(GetTime(0))
-			sTime := fmt.Sprint(GetTime(-7))
-			esQuery := `{"query": {"filtered": {"filter": {"bool": {"must": [{"range": {"comeintime": {"gte": "` + sTime + `","lte": "` + eTime + `"}}}]}},"query": {"bool": {"must": [{"multi_match": {"query": "` + title + `","type": "phrase","fields": ["title"]}}]}}}}}`
+			eTime := time.Now().Unix()
+			sTime := eTime - int64(7*86400)
+			esQuery := `{"query": {"filtered": {"filter": {"bool": {"must": [{"range": {"comeintime": {"gte": "` + fmt.Sprint(sTime) + `","lte": "` + fmt.Sprint(eTime) + `"}}}]}},"query": {"bool": {"must": [{"multi_match": {"query": "` + title + `","type": "phrase","fields": ["title"]}}]}}}}}`
 			if Es.Count(EsIndex, EsType, esQuery) > 0 { //es中含本title数据,不再采集,更新list表数据状态
 				isEsRepeat = true
 			}
@@ -774,9 +774,9 @@ func (s *Spider) DownloadHighDetail() {
 					href := qu.ObjToString(tmp["href"])
 					if isEsRepeat { //es数据title判重
 						title := qu.ObjToString(tmp["title"])
-						eTime := fmt.Sprint(GetTime(0))
-						sTime := fmt.Sprint(GetTime(-7))
-						esQuery := `{"query": {"filtered": {"filter": {"bool": {"must": [{"range": {"comeintime": {"gte": "` + sTime + `","lte": "` + eTime + `"}}}]}},"query": {"bool": {"must": [{"multi_match": {"query": "` + title + `","type": "phrase","fields": ["title"]}}]}}}}}`
+						eTime := time.Now().Unix()
+						sTime := eTime - int64(7*86400)
+						esQuery := `{"query": {"filtered": {"filter": {"bool": {"must": [{"range": {"comeintime": {"gte": "` + fmt.Sprint(sTime) + `","lte": "` + fmt.Sprint(eTime) + `"}}}]}},"query": {"bool": {"must": [{"multi_match": {"query": "` + title + `","type": "phrase","fields": ["title"]}}]}}}}}`
 						count := Es.Count(EsIndex, EsType, esQuery)
 						if count > 0 { //es中含本title数据,不再采集,更新list表数据状态
 							set := map[string]interface{}{"$set": map[string]interface{}{"state": 1, "exist": true}} //已存在state置为1
@@ -910,9 +910,9 @@ func (s *Spider) DownloadListDetail() {
 			href := qu.ObjToString(tmp["href"])
 			if isEsRepeat { //es数据title判重
 				title := qu.ObjToString(tmp["title"])
-				eTime := fmt.Sprint(GetTime(0))
-				sTime := fmt.Sprint(GetTime(-7))
-				esQuery := `{"query": {"filtered": {"filter": {"bool": {"must": [{"range": {"comeintime": {"gte": "` + sTime + `","lte": "` + eTime + `"}}}]}},"query": {"bool": {"must": [{"multi_match": {"query": "` + title + `","type": "phrase","fields": ["title"]}}]}}}}}`
+				eTime := time.Now().Unix()
+				sTime := eTime - int64(7*86400)
+				esQuery := `{"query": {"filtered": {"filter": {"bool": {"must": [{"range": {"comeintime": {"gte": "` + fmt.Sprint(sTime) + `","lte": "` + fmt.Sprint(eTime) + `"}}}]}},"query": {"bool": {"must": [{"multi_match": {"query": "` + title + `","type": "phrase","fields": ["title"]}}]}}}}}`
 				if Es.Count(EsIndex, EsType, esQuery) > 0 { //es中含本title数据,不再采集,更新list表数据状态
 					set := map[string]interface{}{"$set": map[string]interface{}{"state": 1, "exist": true}} //已存在state置为1
 					Mgo.Update("spider_highlistdata", query, set, false, false)