data_ai/clean/c_pcode.go

package clean

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

// 编号
var codeUnConReg *regexp.Regexp = regexp.MustCompile("(null|勘察|包件|测试|设计|设备|项目|标段|工程|监理|范围|分包|月|日|天)")
var codeUnLenReg *regexp.Regexp = regexp.MustCompile("([\u4e00-\u9fa5]{7,})")

// 清洗项目编号
func CleanPcode(pcode string, fns []string) string {
	if pcode == "无" {
		return ""
	}
	pcode = fieldReg1.ReplaceAllString(pcode, "")
	pcode = pcodeReg1.ReplaceAllString(pcode, "")
	pcode = pcodeReg2.ReplaceAllString(pcode, "")
	if utf8.RuneCountInString(pcode) < 5 {
		pcode = ""
	}
	if codeUnConReg.MatchString(pcode) || codeUnLenReg.MatchString(pcode) || !isAlphanumeric(pcode) || isRegTimeDateCode(pcode) {
		return ""
	}
	//校验与附件名字否是一致-舍弃
	for _, v := range fns {
		if utf8.RuneCountInString(v) >= utf8.RuneCountInString(pcode) {
			if strings.Contains(v, pcode) {
				return ""
			}
		}
	}
	return pcode
}

// 清洗其他编号
func CleanOtherCode(ocode string) string {
	if ocode == "无" || ocode == "" {
		return ""
	}
	ocode = fieldReg1.ReplaceAllString(ocode, "")
	ocode = pcodeReg1.ReplaceAllString(ocode, "")
	ocode = pcodeReg2.ReplaceAllString(ocode, "")
	if utf8.RuneCountInString(ocode) < 5 {
		return ""
	}
	if codeUnConReg.MatchString(ocode) || codeUnLenReg.MatchString(ocode) || !isAlphanumeric(ocode) || isRegTimeDateCode(ocode) {
		return ""
	}
	return ocode
}

// 是否含字母数字
func isAlphanumeric(str string) bool {
	var count int
	for _, v := range str {
		if unicode.IsNumber(v) || unicode.IsLetter(v) {
			count++
			break
		}
	}
	return count > 0
}

// 连续数字
func isRegTimeDateCode(str string) bool {
	reg := `\d{8}`
	regx, _ := regexp.Compile(reg)
	if regx.FindString(str) != "" {
		return false
	}
	if utf8.RuneCountInString(str) == 8 {
		return true
	}
	return false
}