浏览代码

清洗时间,单位等方法

zhengkun 9 月之前
父节点
当前提交
9f434089ea
共有 7 个文件被更改,包括 62 次插入47 次删除
  1. 0 22
      clean/c_buyer.go
  2. 0 10
      clean/c_discount.go
  3. 14 0
      clean/c_time.go
  4. 46 0
      clean/c_unit.go
  5. 0 15
      clean/c_winner.go
  6. 1 0
      extract/extension.go
  7. 1 0
      ul/attr.go

+ 0 - 22
clean/c_buyer.go

@@ -1,22 +0,0 @@
-package clean
-
-import (
-	"strings"
-	"unicode/utf8"
-)
-
-// 清洗采购单位
-func CleanBuyer(buyer string) string {
-	if buyer == "无" {
-		return ""
-	}
-	buyer = fieldReg1.ReplaceAllString(buyer, "")
-	//中文括弧
-	buyer = strings.ReplaceAll(buyer, "(", "(")
-	buyer = strings.ReplaceAll(buyer, ")", ")")
-
-	if utf8.RuneCountInString(buyer) < 4 {
-		buyer = ""
-	}
-	return buyer
-}

+ 0 - 10
clean/c_discount.go

@@ -1,10 +0,0 @@
-package clean
-
-// 投标折扣系数
-func CleanBiddiscount(biddiscount string) string {
-	if biddiscount == "无" {
-		return ""
-	}
-	biddiscount = fieldReg1.ReplaceAllString(biddiscount, "")
-	return biddiscount
-}

+ 14 - 0
clean/c_time.go

@@ -0,0 +1,14 @@
+package clean
+
+import (
+	"data_ai/ul"
+	"time"
+)
+
+func CleanTime(st string) int64 {
+	if st == "" {
+		return 0
+	}
+	t, _ := time.ParseInLocation(ul.TimeLayout, st, time.Local)
+	return t.Unix()
+}

+ 46 - 0
clean/c_unit.go

@@ -0,0 +1,46 @@
+package clean
+
+import (
+	"strings"
+	"unicode/utf8"
+)
+
+// 清洗采购单位
+func CleanBuyer(buyer string) string {
+	if buyer == "无" {
+		return ""
+	}
+	buyer = fieldReg1.ReplaceAllString(buyer, "")
+	//中文括弧
+	buyer = strings.ReplaceAll(buyer, "(", "(")
+	buyer = strings.ReplaceAll(buyer, ")", ")")
+
+	if utf8.RuneCountInString(buyer) < 4 {
+		buyer = ""
+	}
+	return buyer
+}
+
+// 清洗中标单位
+func CleanWinner(s_winner string) string {
+	if s_winner == "无" {
+		return ""
+	}
+	s_winner = fieldReg1.ReplaceAllString(s_winner, "")
+	if utf8.RuneCountInString(s_winner) < 3 {
+		s_winner = ""
+	}
+	return s_winner
+}
+
+// 清洗代理机构
+func CleanAgency(agency string) string {
+	if agency == "无" {
+		return ""
+	}
+	agency = fieldReg1.ReplaceAllString(agency, "")
+	if utf8.RuneCountInString(agency) < 4 {
+		agency = ""
+	}
+	return agency
+}

+ 0 - 15
clean/c_winner.go

@@ -1,15 +0,0 @@
-package clean
-
-import "unicode/utf8"
-
-// 清洗中标单位
-func CleanWinner(s_winner string) string {
-	if s_winner == "无" {
-		return ""
-	}
-	s_winner = fieldReg1.ReplaceAllString(s_winner, "")
-	if utf8.RuneCountInString(s_winner) < 3 {
-		s_winner = ""
-	}
-	return s_winner
-}

+ 1 - 0
extract/extension.go

@@ -160,4 +160,5 @@ func ForcedLogicDecideInfo(f_data map[string]interface{}) {
 			f_data["s_agency"] = ""
 		}
 	}
+
 }

+ 1 - 0
ul/attr.go

@@ -8,6 +8,7 @@ var (
 	SysConfig               map[string]interface{}
 	ToolConfig              map[string]interface{}
 	Bid_Name, Ext_Name      string
+	TimeLayout              = "2006-01-02 15:04:05"
 	Url                     = "https://www.jianyu360.cn/article/content/%s.html"
 	CleanResultReg          = regexp.MustCompile("((\\s|\n| |\\[|\\]|\\`|json)+)")
 	SaveResultReg           = regexp.MustCompile("([{].*[}])")