Procházet zdrojové kódy

Merge branch 'dev3.2' of http://192.168.3.207:10080/qmx/jy-data-extract into dev3.2

fengweiqiang před 6 roky
rodič
revize
92a02b8288

+ 0 - 1
src/jy/pretreated/analystep.go

@@ -70,7 +70,6 @@ func AnalyStart(job *util.Job) {
 		bl.Text = HtmlToText(con)
 		//调用kv解析
 		bl.ColonKV = GetKVAll(bl.Text, "", nil, 1)
-		PrintKvTags(bl.ColonKV.KvTags)
 		bl.SpaceKV = SspacekvEntity.Entrance(bl.Text, "", nil)
 		//新加 未分块table中未能解析到中标候选人,从正文中解析
 		if job.Winnerorder == nil || len(job.Winnerorder) == 0 {

+ 6 - 1
src/jy/pretreated/analytable.go

@@ -112,6 +112,7 @@ var (
 	projectcodeReg3             = regexp.MustCompile("(^询价单编号[A-Za-z0-9/-]*|公告编号[A-Za-z0-9/-]*)")
 	jsonReg                     = regexp.MustCompile(`\{.+:[^}]*\} `) //  \{".*\":\".+\"}
 	regHz                       = regexp.MustCompile("[\u4e00-\u9fa5]")
+	winnerOrerReg               = regexp.MustCompile("(中标)?候选人")
 )
 
 //在解析时,判断表格元素是否隐藏
@@ -160,7 +161,11 @@ func CommonDataAnaly(k, tabletag, tabledesc string, v interface{}) (kvTags map[s
 			if MhSpilt.MatchString(v1) {
 				t1.Weight -= 50
 			}
-			kvTags[t1.Value] = append(kvTags[t1.Value], &u.Tag{Key: k1, Value: v1, Weight: t1.Weight})
+			if winnerOrerReg.MatchString(tabletag) && t1.Value == "采购单位联系人" { //处理table中项目负责人
+				kvTags[k] = append(kvTags[k], &u.Tag{Key: k, Value: v1, IsInvalid: true})
+			} else {
+				kvTags[t1.Value] = append(kvTags[t1.Value], &u.Tag{Key: k1, Value: v1, Weight: t1.Weight})
+			}
 		}
 		//k1 = res[0].Value
 	} else {

+ 1 - 1
src/res/fieldscore.json

@@ -206,7 +206,7 @@
             {
                 "describe": "非结尾",
                 "regstr": ".*[^集团|公司|学校|中心|家具城|门诊|\\[大中小\\]学|部|院|局|厂|店|所|队|社|室|厅|段|会|场|行]$",
-                "score": -10
+                "score": -5
             }
         ]
     },