data_ai/clean/c_time.go

package clean

import (
	"data_ai/ul"
	"github.com/shopspring/decimal"
	qu "jygit.jydev.jianyu360.cn/data_processing/common_utils"
	"regexp"
	"strings"
	"time"
)

var numReg = regexp.MustCompile("[0-9.]+")
var symbolReg = regexp.MustCompile("[%％﹪!！]")

var YMD_Reg1 = regexp.MustCompile("(\\d{4}年\\d{1,2}月\\d{1,2}日)")
var YMD_Reg2 = regexp.MustCompile("((\\d{4})[-/](\\d{1,2})[-/](\\d{1,2}))")

var HMS_Reg1 = regexp.MustCompile("(\\d{1,2}时\\d{1,2}分\\d{1,2}秒)")
var HMS_Reg2 = regexp.MustCompile("(\\d{1,2}[:]\\d{1,2}[:]\\d{1,2})")
var HMS_Reg3 = regexp.MustCompile("((\\d{1,2})[时点:]((\\d{1,2})[分:])?(\\d{1,2})?)")

func fillPrefix(s string) string {
	if len(s) == 2 {
		return s
	} else if len(s) == 1 {
		return "0" + s
	} else {
		return "00"
	}
}

// 获取年月日
func convertYMD(ymd string) string {
	if ymd1 := YMD_Reg1.FindString(ymd); ymd1 != "" {
		ymd1 = strings.ReplaceAll(ymd1, "年", "-")
		ymd1 = strings.ReplaceAll(ymd1, "月", "-")
		ymd1 = strings.ReplaceAll(ymd1, "日", "")
		if arr := strings.Split(ymd1, "-"); len(arr) == 3 {
			y, m, d := arr[0], arr[1], arr[2]
			return y + "-" + fillPrefix(m) + "-" + fillPrefix(d)
		} else {
			return ""
		}
	}
	if ymd2 := YMD_Reg2.FindString(ymd); ymd2 != "" {
		y := YMD_Reg2.ReplaceAllString(ymd2, "${2}")
		m := YMD_Reg2.ReplaceAllString(ymd2, "${3}")
		d := YMD_Reg2.ReplaceAllString(ymd2, "${4}")
		return y + "-" + fillPrefix(m) + "-" + fillPrefix(d)
	}
	return ""
}

// 获取时分秒
func convertHMS(hms string) string {
	if hms1 := HMS_Reg1.FindString(hms); hms1 != "" {
		hms1 = strings.ReplaceAll(hms1, "时", ":")
		hms1 = strings.ReplaceAll(hms1, "分", ":")
		hms1 = strings.ReplaceAll(hms1, "秒", "")
		if arr := strings.Split(hms1, ":"); len(arr) == 3 {
			h, m, s := arr[0], arr[1], arr[2]
			return fillPrefix(h) + ":" + fillPrefix(m) + ":" + fillPrefix(s)
		} else {
			return "00:00:00"
		}
	}
	if hms2 := HMS_Reg2.FindString(hms); hms2 != "" {
		arr := strings.Split(hms2, ":")
		h, m, s := "", "", ""
		if len(arr) == 3 {
			h, m, s = arr[0], arr[1], arr[2]
		} else if len(arr) == 2 {
			h, m, s = arr[0], arr[1], "00"
		} else {
			return "00:00:00"
		}
		return fillPrefix(h) + ":" + fillPrefix(m) + ":" + fillPrefix(s)
	}
	if hms3 := HMS_Reg3.FindString(hms); hms3 != "" {
		h := HMS_Reg3.ReplaceAllString(hms3, "${2}")
		m := HMS_Reg3.ReplaceAllString(hms3, "${4}")
		s := HMS_Reg3.ReplaceAllString(hms3, "${5}")
		return fillPrefix(h) + ":" + fillPrefix(m) + ":" + fillPrefix(s)
	}
	return "00:00:00"
}

// 清洗时间
func CleanTime(st string) int64 {
	if st == "" || st == "无" {
		return 0
	}
	st = strings.ReplaceAll(st, "：", ":")
	ymd, hms := convertYMD(st), convertHMS(st)
	if ymd == "" {
		return 0
	}
	st = ymd + " " + hms
	t, _ := time.ParseInLocation(ul.TimeLayout, st, time.Local)
	return t.Unix()
}

// 清洗折扣率
func CleanDiscount(str string) float64 {
	str = fieldReg1.ReplaceAllString(str, "")
	if str == "" || str == "无" {
		return 0.0
	}
	if biddiscount := RateToFloat(str); biddiscount > 0.0 {
		baseCount := 1.0
		num1 := decimal.NewFromFloat(baseCount)
		num2 := decimal.NewFromFloat(biddiscount)
		if strings.Contains(str, "上浮") {
			decimalValue := num1.Add(num2)
			res, _ := decimalValue.Float64()
			return res
		} else if strings.Contains(str, "下浮") {
			decimalValue := num1.Sub(num2)
			res, _ := decimalValue.Float64()
			return res
		} else {
			return biddiscount
		}
	}
	return 0.0
}

// 转换系数
func RateToFloat(str string) float64 {
	if num0 := qu.Float64All(numReg.FindString(str)); num0 > 0.0 {
		num1 := decimal.NewFromFloat(100.0)
		num2 := decimal.NewFromFloat(num0)
		if symbolReg.MatchString(str) {
			decimalValue := num2.Div(num1)
			res, _ := decimalValue.Float64()
			if res < 1.0 {
				return res
			} else {
				return 0.0
			}
		} else {
			if num0 < 1.0 {
				return num0
			} else if num0 == 1 {
				return 0.0
			} else {
				decimalValue := num2.Div(num1)
				res, _ := decimalValue.Float64()
				if res < 1.0 {
					return res
				} else {
					return 0.0
				}
			}
		}
	}
	return 0.0
}