apple 5 years ago
parent
commit
547fd4bd0a
4 changed files with 50 additions and 45 deletions
  1. 2 2
      udpfilterdup/src/config.json
  2. 1 1
      udpfilterdup/src/datamap.go
  3. 41 40
      udpfilterdup/src/main.go
  4. 6 2
      udps/main.go

+ 2 - 2
udpfilterdup/src/config.json

@@ -5,8 +5,8 @@
         "addr": "192.168.3.207:27082",
         "pool": 15,
         "db": "extract_kf",
-        "extract": "bidding_20190910_01",
-        "extract_copy": "bidding_20190910_01_copy",
+        "extract": "a_testbidding",
+        "extract_copy": "a_testbidding_copy",
         "bidding": "bidding_126"
     },
     "jkmail": {

+ 1 - 1
udpfilterdup/src/datamap.go

@@ -201,7 +201,7 @@ L:
 								//同城
 								b = true
 								id = v.id
-								log.Print("同城满足的",info.id)
+								//log.Print("同城满足的",info.id)
 								break L
 							}
 						}

+ 41 - 40
udpfilterdup/src/main.go

@@ -32,9 +32,10 @@ var (
 	dupdays      = 5                      //初始化判重范围
 	DM           *datamap                 //判重数据
 	FilterRegexp = regexp.MustCompile("^_$")
-	lastid       = "5d767728a5cb26b9b7748868"
-
+	lastid       = "5da3f2c5a5cb26b9b79847fc"
 
+	//5d767728a5cb26b9b7748868  //9万
+	//5da3f2c5a5cb26b9b79847fc  //76万
 	FilterRegTitle = regexp.MustCompile("[0-9一二三四五六七八九十零123456789](次|包|标段)")
 	inV_n int   //无效数据数量
 	siteArr     []map[string]interface{} //站点
@@ -64,13 +65,13 @@ func init() {
 
 
 	//测试临时注释
-	//dupdays = util.IntAllDef(Sysconfig["dupdays"], 3)
-	////加载数据
-	//DM = NewDatamap(dupdays, lastid)
-	//sw := util.ObjToString(Sysconfig["specialwords"])
-	//if sw != "" {
-	//	FilterRegexp = regexp.MustCompile(sw)
-	//}
+	dupdays = util.IntAllDef(Sysconfig["dupdays"], 3)
+	//加载数据
+	DM = NewDatamap(dupdays, lastid)
+	sw := util.ObjToString(Sysconfig["specialwords"])
+	if sw != "" {
+		FilterRegexp = regexp.MustCompile(sw)
+	}
 
 
 	//数据库
@@ -87,38 +88,38 @@ func init() {
 }
 
 //新增一个方法 判断
-func main222()  {
-	log.Println("1")
-
-	sessTest :=mgoTest.GetMgoConn()
-	defer sessTest.Close()
-
-	sess := mgo.GetMgoConn()
-	defer sess.Close()
-
-	//var arr []map[string]interface{}
+func main22()  {
+	//log.Println("1")
 
-	res_test := sessTest.DB("qfw").C("bidding").Find(mongodb.ObjToMQ(`{"comeintime":{"$gte": 1571025600, "$lte": 1571976000}}`, true)).Iter()
-	res :=sess.DB("extract_kf").C("a_testbidding")
-
-
-
-
-
-	i:=0
-	for dict := make(map[string]interface{}); res_test.Next(&dict); i++{
-
-		//插入
-		if i%2000==0 {
-			log.Println("当前:",i)
-		}
-		res.Insert(dict)
-		//if len(arr)>=500 {
-		//	arr = make([]map[string]interface{},0)
-		//}else {
-		//	arr = append(arr,dict)
-		//}
-	}
+	//sessTest :=mgoTest.GetMgoConn()
+	//defer sessTest.Close()
+	//
+	//sess := mgo.GetMgoConn()
+	//defer sess.Close()
+	//
+	////var arr []map[string]interface{}
+	//
+	//res_test := sessTest.DB("qfw").C("bidding").Find(mongodb.ObjToMQ(`{"comeintime":{"$gte": 1571025600, "$lte": 1571976000}}`, true)).Iter()
+	//res :=sess.DB("extract_kf").C("a_testbidding")
+	//
+	//
+	//
+	//
+	//
+	//i:=0
+	//for dict := make(map[string]interface{}); res_test.Next(&dict); i++{
+	//
+	//	//插入
+	//	if i%2000==0 {
+	//		log.Println("当前:",i)
+	//	}
+	//	res.Insert(dict)
+	//	//if len(arr)>=500 {
+	//	//	arr = make([]map[string]interface{},0)
+	//	//}else {
+	//	//	arr = append(arr,dict)
+	//	//}
+	//}
 
 
 

+ 6 - 2
udps/main.go

@@ -29,9 +29,13 @@ func main() {
 
 	56404035af5374672e00059c
 	5d4da9c8a5cb26b9b7b6bbcd  100万
+
+
+	5da3f2c5a5cb26b9b79847fc
+	5db2735ba5cb26b9b7c99c6f   76万
 	*/
-	flag.StringVar(&sid, "sid", "5d767728a5cb26b9b7748868", "开始id")
-	flag.StringVar(&eid, "eid", "5d77c881a5cb26b9b7de209d", "结束id")
+	flag.StringVar(&sid, "sid", "5da3f2c5a5cb26b9b79847fc", "开始id")
+	flag.StringVar(&eid, "eid", "5db2735ba5cb26b9b7c99c6f", "结束id")
 	flag.StringVar(&startDate, "start", "", "开始日期2006-01-02")
 	flag.StringVar(&endDate, "end", "", "结束日期2006-01-02")
 	flag.StringVar(&ip, "ip", "127.0.0.1", "ip")