Ver Fonte

dataging

fengweiqiang há 5 anos atrás
pai
commit
037aa132cc

+ 4 - 1
src/jy/extract/extract.go

@@ -33,7 +33,7 @@ var (
 	ClearTaskList map[string]*ClearTask   //清理任务列表
 	saveLimit     = 100                   //抽取日志批量保存
 	PageSize      = 5000                  //查询分页
-	Fields        = `{"title":1,"summary":1,"detail":1,"contenthtml":1,"site":1,"spidercode":1,"toptype":1,"subtype":1,"bidstatus":1,"area":1,"city":1,"comeintime":1,"publishtime":1,"sensitive":1,"projectinfo":1,"jsondata":1,"href":1,"infoformat":1,"attach_text":1}`
+	Fields        = `{"title":1,"summary":1,"detail":1,"contenthtml":1,"site":1,"spidercode":1,"toptype":1,"subtype":1,"bidstatus":1,"area":1,"city":1,"comeintime":1,"publishtime":1,"sensitive":1,"projectinfo":1,"jsondata":1,"href":1,"infoformat":1,"attach_text":1,"dataging:":1}`
 	Fields2       = `{"budget":1,"bidamount":1,"title":1,"projectname":1,"winner":1}`
 )
 
@@ -331,6 +331,7 @@ func (e *ExtractTask) PreInfo(doc map[string]interface{}) (j, jf *ju.Job, isSite
 		Result:    map[string][]*ju.ExtField{},
 		BuyerAddr: qu.ObjToString(doc["buyeraddr"]),
 		RuleBlock: e.RuleBlock,
+		Dataging: qu.IntAll(doc["dataging"]),
 	}
 	if (j.Jsondata != nil || (*j.Jsondata) != nil) && (*j.Jsondata)["jsoncontent"] != nil {
 		delete((*j.Jsondata), "jsoncontent")
@@ -352,6 +353,7 @@ func (e *ExtractTask) PreInfo(doc map[string]interface{}) (j, jf *ju.Job, isSite
 			BuyerAddr:      qu.ObjToString(doc["buyeraddr"]),
 			RuleBlock:      e.RuleBlock,
 			IsFile:         isextFile,
+			Dataging: qu.IntAll(doc["dataging"]),
 		}
 		if (jf.Jsondata != nil || (*jf.Jsondata) != nil) && (*jf.Jsondata)["jsoncontent"] != nil {
 			delete((*jf.Jsondata), "jsoncontent")
@@ -1927,6 +1929,7 @@ func AnalysisSaveResult(j, jf *ju.Job, e *ExtractTask) {
 		//			qu.Debug(k, "---", v)
 		//		}
 		//tmp["extract_content"] = j.Content
+		tmp["dataging"] = j.Dataging
 		if e.TaskInfo.TestColl == "" {
 			if len(tmp) > 0 { //保存抽取结果
 				/*	if len(e.SiteFields) <= 0 {

+ 1 - 0
src/jy/extract/extractudp.go

@@ -130,6 +130,7 @@ func ExtractByUdp(sid, eid string, ra *net.UDPAddr, instanceId ...string) {
 		ext.ResultSave(true)
 		ext.BidSave(true)
 		ext.IsRun = true
+		ext.InitFile()
 	} else {
 		ext.BidTotal = 0
 	}

+ 1 - 0
src/jy/util/article.go

@@ -40,6 +40,7 @@ type Job struct {
 	SimAreaScore      map[string]float64                //简称province得分
 	SimCityScore      map[string]float64                //简称city得分
 	SimDistrictScore  map[string]float64                //简称district得分
+	Dataging int
 }
 
 type ExtField struct {

+ 1 - 0
src/main_blocktest.go

@@ -119,6 +119,7 @@ func com(doc map[string]interface{}) {
 		Result:    map[string][]*ju.ExtField{},
 		BuyerAddr: qu.ObjToString(doc["buyeraddr"]),
 		RuleBlock: e.RuleBlock,
+		Dataging: qu.IntAll(doc["dataging"]),
 	}
 	e.TaskInfo.ProcessPool <- true
 	pretreated.AnalyStart(j, false, "")