瀏覽代碼

sensitive

unknown 6 年之前
父節點
當前提交
9c09986737
共有 3 個文件被更改,包括 23 次插入12 次删除
  1. 2 4
      src/jy/extract/exportask.go
  2. 9 4
      src/jy/extract/extract.go
  3. 12 4
      src/jy/extract/extractudp.go

+ 2 - 4
src/jy/extract/exportask.go

@@ -59,10 +59,8 @@ func extractAndExport(v string, t map[string]interface{}) {
 	limit := qu.IntAll(t["limit"])
 	list, _ := e.TaskInfo.FDB.Find(e.TaskInfo.FromColl, query, nil, Fields, false, 0, limit)
 	for _, v := range *list {
-		if v["sensitive"] != nil { //去除含敏感词数据
-			if v["sensitive"].(string) != "" {
-				continue
-			}
+		if qu.ObjToString(v["sensitive"]) != "" { //去除含敏感词数据
+			continue
 		}
 		j := PreInfo(v)
 		e.TaskInfo.ProcessPool <- true

+ 9 - 4
src/jy/extract/extract.go

@@ -148,10 +148,8 @@ func RunExtractTask(taskId string) {
 		log.Printf("page=%d,query=%v", i+1, query)
 		list, _ := ext.TaskInfo.FDB.Find(ext.TaskInfo.FromColl, query, nil, Fields, false, 0, limit)
 		for _, v := range *list {
-			if v["sensitive"] != nil { //去除含敏感词数据
-				if v["sensitive"].(string) != "" {
-					continue
-				}
+			if qu.ObjToString(v["sensitive"]) != "" { //去除含敏感词数据
+				continue
 			}
 			//log.Println(v["_id"])
 			if !ext.IsRun {
@@ -240,6 +238,13 @@ func (e *ExtractTask) ExtractProcess(j *ju.Job) {
 			}
 			//log.Println("抽取-规则", tmp)
 
+			//项目名称未能抽取到,标题来凑
+			if vc.Field == "projectname" {
+				if len(j.Result[vc.Field]) < 1 {
+					j.Result[vc.Field] = append(j.Result[vc.Field], &ju.ExtField{vc.Field, "title", "title", "regexp", "title", vc.ExtFrom, j.Title, 0})
+				}
+			}
+
 			//抽取-后置规则
 			for _, v := range vc.RuleBacks {
 				ExtRegBack(j, v, e.TaskInfo)

+ 12 - 4
src/jy/extract/extractudp.go

@@ -136,10 +136,15 @@ func ExtractByUdp(sid, eid string, instanceId ...string) {
 			if ext.TaskInfo.FDB.Count(ext.TaskInfo.FromColl, query) > 0 {
 				list, _ := ext.TaskInfo.FDB.Find(ext.TaskInfo.FromColl, query, nil, Fields, false, 0, limit)
 				for _, v := range *list {
+<<<<<<< HEAD
 					if v["sensitive"] != nil { //去除含敏感词数据
 						if v["sensitive"].(string) != "" {
 							continue
 						}
+=======
+					if qu.ObjToString(v["check_sensitive2"]) != "" { //去除含敏感词数据
+						continue
+>>>>>>> 60d47d50992bea38b0f4c3b40204314bac6a4985
 					}
 					//log.Println(v["_id"])
 					j := PreInfo(v)
@@ -157,10 +162,15 @@ func ExtractByUdp(sid, eid string, instanceId ...string) {
 			if ext.TaskInfo.FDB.Count(ext.TaskInfo.FromColl+"_back", queryback) > 0 {
 				list2, _ := ext.TaskInfo.FDB.Find(ext.TaskInfo.FromColl+"_back", queryback, nil, Fields, false, 0, limit)
 				for _, v := range *list2 {
+<<<<<<< HEAD
 					if v["sensitive"] != nil { //去除含敏感词数据
 						if v["sensitive"].(string) != "" {
 							continue
 						}
+=======
+					if qu.ObjToString(v["check_sensitive2"]) != "" { //去除含敏感词数据
+						continue
+>>>>>>> 60d47d50992bea38b0f4c3b40204314bac6a4985
 					}
 					//log.Println(v["_id"])
 					j := PreInfo(v)
@@ -192,10 +202,8 @@ func ExtractByUdp(sid, eid string, instanceId ...string) {
 			log.Printf("page=%d,query=%v", i+1, query)
 			list, _ := ext.TaskInfo.FDB.Find(ext.TaskInfo.FromColl, query, nil, Fields, false, 0, limit)
 			for _, v := range *list {
-				if v["sensitive"] != nil { //去除含敏感词数据
-					if v["sensitive"].(string) != "" {
-						continue
-					}
+				if qu.ObjToString(v["sensitive"]) != "" { //去除含敏感词数据
+					continue
 				}
 				//log.Println(v["_id"])
 				j := PreInfo(v)