data_ai/clean/c_unit.go

package clean

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

var c_unit_1 = regexp.MustCompile("([(（]本级[)）])$")
var c_unit_2 = regexp.MustCompile("(局)$")

// 清洗采购单位
func CleanBuyer(buyer string) string {
	if buyer == "无" || buyer == "有限公司" {
		return ""
	}
	buyer = fieldReg1.ReplaceAllString(buyer, "")
	//中文括弧
	buyer = strings.ReplaceAll(buyer, "(", "（")
	buyer = strings.ReplaceAll(buyer, ")", "）")
	//清除后缀-本级
	buyer = c_unit_1.ReplaceAllString(buyer, "")

	if utf8.RuneCountInString(buyer) == 3 {
		if c_unit_2.MatchString(buyer) {
			return 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" || s_winner == "有限公司" {
		return ""
	}
	s_winner = fieldReg1.ReplaceAllString(s_winner, "")
	if utf8.RuneCountInString(s_winner) < 3 {
		s_winner = ""
	}
	s_winner = strings.ReplaceAll(s_winner, "(", "（")
	s_winner = strings.ReplaceAll(s_winner, ")", "）")

	return s_winner
}

// 清洗代理机构
func CleanAgency(agency string) string {
	if agency == "无" || agency == "有限公司" {
		return ""
	}
	agency = fieldReg1.ReplaceAllString(agency, "")
	if utf8.RuneCountInString(agency) < 4 {
		agency = ""
	}
	return agency
}

// 联系人
func CleanContactPerson(person string) string {
	if person == "无" {
		return ""
	}
	person = fieldReg1.ReplaceAllString(person, "")
	if utf8.RuneCountInString(person) < 2 {
		person = ""
	}
	return person
}

// 联系方式
func CleanContactTel(tel string) string {
	if tel == "无" {
		return ""
	}
	tel = fieldReg1.ReplaceAllString(tel, "")
	if utf8.RuneCountInString(tel) < 6 {
		tel = ""
	}
	return tel
}