data_ai/clean/c_pname.go

package clean

import (
	"regexp"
	"strings"
	"unicode/utf8"
)

var pname_reg1 = regexp.MustCompile("([\\.。]+)$")
var pname_reg2 = regexp.MustCompile("(中标候选人公示)$")
var pname_reg3 = regexp.MustCompile("(采购项目采购项目)$")
var pname_reg4 = regexp.MustCompile("(XX项目)$")

// 清洗项目名称
func CleanPname(pname string) string {
	if pname == "无" {
		return ""
	}
	pname = fieldReg1.ReplaceAllString(pname, "")
	pname = pname_reg1.ReplaceAllString(pname, "")
	pname = pname_reg2.ReplaceAllString(pname, "")
	pname = pname_reg3.ReplaceAllString(pname, "采购项目")
	if pname_reg4.MatchString(pname) {
		return ""
	}
	pname = strings.ReplaceAll(pname, "（", "(")
	pname = strings.ReplaceAll(pname, "）", ")")

	if utf8.RuneCountInString(pname) < 5 {
		pname = ""
	}

	return pname
}