zhangjinkun 6 years ago
parent
commit
0380e4ec80
3 changed files with 17 additions and 7 deletions
  1. 7 1
      src/jy/extract/extract.go
  2. 8 4
      src/jy/extract/score.go
  3. 2 2
      src/res/fieldscore.json

+ 7 - 1
src/jy/extract/extract.go

@@ -660,7 +660,7 @@ func ExtRegCore(extfrom string, doc map[string]interface{}, j *ju.Job, in *RegLu
 						if tmp["blocktag"] != nil {
 							btag := make(map[string]string)
 							for k := range tmp["blocktag"].(map[string]bool) {
-								if TagConfigDesc[k] != ""{
+								if TagConfigDesc[k] != "" {
 									btag[k] = TagConfigDesc[k]
 								}
 							}
@@ -919,6 +919,9 @@ func ExtRegBack(j *ju.Job, in *RegLuaInfo, t *TaskInfo) {
 						text = in.RegPreBac.Reg.ReplaceAllString(text, in.RegPreBac.Replace)
 					}
 					j.Result[in.Field][k].Value = text
+					if text == qu.ObjToString(v.Value) { //值未发生改变,不存日志
+						continue
+					}
 					exts = append(exts, map[string]interface{}{
 						"field":     v.Field,
 						"code":      v.Code,
@@ -946,6 +949,9 @@ func ExtRegBack(j *ju.Job, in *RegLuaInfo, t *TaskInfo) {
 						text = in.RegPreBac.Reg.ReplaceAllString(text, in.RegPreBac.Replace)
 					}
 					j.Result[key][k].Value = text
+					if text == qu.ObjToString(v.Value) { //值未发生改变,不存日志
+						continue
+					}
 					exts = append(exts, map[string]interface{}{
 						"field":     v.Field,
 						"code":      v.Code,

+ 8 - 4
src/jy/extract/score.go

@@ -126,10 +126,6 @@ func ScoreFields(j *ju.Job) map[string][]*ju.ExtField {
 	for field, tmps := range result {
 		for tmpsindex, tmpsvalue := range tmps {
 			describe := qu.ObjToString(SoreConfig["extractype"]["describe"])
-			if tmpsvalue.ExtFrom == "title" { //标题打分初始化
-				tmps[tmpsindex].Score += CommonScore["title"]
-				tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "title初始化", Code: "fieldscore.title", RuleText: describe, ScoreFrom: "fieldscore.json", Value: tmpsvalue.Value, Score: CommonScore["title"]})
-			}
 			//是否有段标签
 			if len(tmpsvalue.BlockTag) > 0 {
 				//有标签段
@@ -151,6 +147,10 @@ func ScoreFields(j *ju.Job) map[string][]*ju.ExtField {
 			//抽取类型打分
 			if FieldsScore[field] != nil { //指定抽取属性打分配置
 				fieldscore := FieldsScore[field]
+				if tmpsvalue.ExtFrom == "title" { //标题打分初始化
+					tmps[tmpsindex].Score += fieldscore["title"]
+					tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "title初始化", Code: "fieldscore.title", RuleText: describe, ScoreFrom: "fieldscore.json", Value: tmpsvalue.Value, Score: fieldscore["title"]})
+				}
 				if strings.Contains(tmpsvalue.Type, "colon") {
 					tmps[tmpsindex].Score += fieldscore["colon"]
 					tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "colonkv", Code: "fieldscore.colon", RuleText: describe, ScoreFrom: "fieldscore.json", Value: tmpsvalue.Value, Score: fieldscore["colon"]})
@@ -165,6 +165,10 @@ func ScoreFields(j *ju.Job) map[string][]*ju.ExtField {
 					tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "regexp", Code: "fieldscore.regexp", RuleText: describe, ScoreFrom: "fieldscore.json", Value: tmpsvalue.Value, Score: fieldscore["regexp"]})
 				}
 			} else { //通用抽取属性打分配置
+				if tmpsvalue.ExtFrom == "title" { //标题打分初始化
+					tmps[tmpsindex].Score += CommonScore["title"]
+					tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "title初始化", Code: "fieldscore.title", RuleText: describe, ScoreFrom: "fieldscore.json", Value: tmpsvalue.Value, Score: CommonScore["title"]})
+				}
 				if strings.Contains(tmpsvalue.Type, "colon") {
 					tmps[tmpsindex].Score += CommonScore["colon"]
 					tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "colonkv", Code: "fieldscore.colon", RuleText: describe, ScoreFrom: "fieldscore.json", Value: tmpsvalue.Value, Score: CommonScore["colon"]})

+ 2 - 2
src/res/fieldscore.json

@@ -14,7 +14,7 @@
                 "table": 3,
                 "colon": 3,
                 "space": 3,
-                "regexp": 2
+                "regexp": 1
             },
             "winner": {
                 "table": 3,
@@ -41,7 +41,7 @@
             {
                 "describe": "以*结尾",
                 "regstr": ".{2,100}(项目|工程|施工|服务|设备|采购|设计|系统)$",
-                "score": 3
+                "score": 2
             }
         ],
         "negativewords": [