优化

src/jy/extract/extractudp.go

@@ -103,8 +103,8 @@ func processUdpMsg(act byte, data []byte, ra *net.UDPAddr) {
 }
 func sendMail(content string) {
 	res, err := http.Get(fmt.Sprintf("%s?to=%s&title=%s&body=%s", ju.Config["api"], ju.Config["tomail"], "jy-data-extract_3.2", "抽取完成："+content))
+	defer res.Body.Close()
 	if err == nil {
-		defer res.Body.Close()
 		read, err := ioutil.ReadAll(res.Body)
 		log.Debug("邮件发送：", string(read), err)
 	}
@@ -159,7 +159,7 @@ func ExtractByUdp(sid, eid string, ra *net.UDPAddr, instanceId ...string) {
 		if count < PageSize {
 			limit = count
 		}
-		fmt.Printf("count=%d,pageNum=%d,query=%v", count, pageNum, query)
+		fmt.Printf("count=%d,pageNum=%d,query=%v\n", count, pageNum, query)
 
 		startI := 0 //接着上次任务执行
 		sidback := sid
@@ -174,7 +174,7 @@ func ExtractByUdp(sid, eid string, ra *net.UDPAddr, instanceId ...string) {
 
 		for i := startI; i < pageNum; i++ {
 			query = bson.M{"_id": bson.M{"$gte": bson.ObjectIdHex(sid), "$lte": bson.ObjectIdHex(eid)}}
-			fmt.Printf("page=%d,query=%v", i+1, query)
+			fmt.Printf("page=%d,query=%v\n", i+1, query)
 			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 {
@@ -201,7 +201,7 @@ func ExtractByUdp(sid, eid string, ra *net.UDPAddr, instanceId ...string) {
 					}}, true, false)
 			}
 			queryback := bson.M{"_id": bson.M{"$gte": bson.ObjectIdHex(sidback)}}
-			fmt.Printf("page=%d,queryback=%v", i+1, queryback)
+			fmt.Printf("page=%d,queryback=%v\n", i+1, queryback)
 			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 {

src/jy/pretreated/analytable.go

@@ -509,11 +509,11 @@ func (table *Table) MergerToTableresult() {
 			package1 := table.TableResult.PackageMap.Map[v2]
 			if package1 == nil {
 				table.TableResult.PackageMap.AddKey(v2, table.BlockPackage.Map[v2])
-				/*if vvv, ok := table.BlockPackage.Map[v2].(*u.BlockPackage); ok {
-					if vvv.TableKV != nil && len(vvv.TableKV.KvTags) > 0 {
+				if vvv, ok := table.BlockPackage.Map[v2].(*u.BlockPackage); ok {
+					if vvv.TableKV != nil && len(vvv.TableKV.KvTags) == 0 {
 						MergeKvTags(table.TableResult.KvTags, vvv.TableKV.KvTags)
 					}
-				}*/
+				}
 			} else {
 				bp := package1.(*u.BlockPackage)
 				if bp.TableKV == nil {
@@ -579,7 +579,7 @@ func (table *Table) MergerToTableresult() {
 			}
 		}
 	}
-	/*if table.BlockPackage != nil && len(table.BlockPackage.Keys) > 0 {
+	if table.BlockPackage != nil && len(table.BlockPackage.Keys) == 0 {
 		for _, v := range table.BlockPackage.Keys {
 			if table.BlockPackage.Map[v] != nil {
 				if vvv, ok := table.BlockPackage.Map[v].((*u.BlockPackage)); ok {
@@ -597,7 +597,7 @@ func (table *Table) MergerToTableresult() {
 			}
 
 		}
-	}*/
+	}
 }
 
 /**

src/jy/pretreated/winnerorder.go

@@ -41,7 +41,7 @@ var (
 	findCompanyReg = regexp.MustCompile("[^:：]+公司")
 	colonSpaceReg  = regexp.MustCompile("[:：]\\s+")
 	findCandidate  = regexp.MustCompile("(^.{5,}(公司|集团|单位|机构|企业|厂|场|院|所|店|中心|市|局|站|城|处|行|部|队|联合(会|体))|工作室)")
-	findCandidate2 = regexp.MustCompile("(^.{5,}(公司|集团|单位|机构|企业|厂|场|院|所|店|中心|市|局|站|城|处|行|部|队|联合(会|体)|工作室)$)")
+	findCandidate2 = regexp.MustCompile("(^.{5,}(公司|集团|单位|机构|企业|厂|场|院|所|店|中心|局|站|城|处|行|部|队|联合(会|体)|工作室)$)")
 	clearSpace1    = regexp.MustCompile("([（(][\\d一二三四五六七八九十][)）][\\s\u3000\u2003\u00a0\\t]*|<[^>].+?>)")
 	offerReg       = regexp.MustCompile("(中标|磋商|投标|报|单|成交)总?(价|金额)")
 )