jy-data-extract/udpfilterdup/src4/dataMethodHeavy.go

package main

import "strings"

//判重方法1
func quickHeavyMethodOne(v *Info, info *Info, reason string) (bool, string) {

	isMeet := false
	if info.subtype == "招标" || info.subtype == "邀标" || info.subtype == "询价" ||
		info.subtype == "竞谈" || info.subtype == "单一" || info.subtype == "竞价" ||
		info.subtype == "变更" || info.subtype == "其他" {
		//招标结果
		if isMeet, reason = tenderRepeat_A(v, info, reason); isMeet {
			if tenderRepeat_C(v, info) {
				return false, reason
			} else {
				reason = reason + "---招标类"
				return true, reason
			}
		} else {
			return false, reason
		}

	} else if info.subtype == "中标" || info.subtype == "成交" || info.subtype == "废标" || info.subtype == "流标" {
		//中标结果
		if isMeet, reason = winningRepeat_A(v, info, reason); isMeet {
			if winningRepeat_C(v, info) {
				return false, reason
			} else {
				reason = reason + "---中标类"
				return true, reason
			}
		} else {
			return false, reason
		}

	} else if info.subtype == "合同" || info.subtype == "验收" || info.subtype == "违规" {
		//合同
		if isMeet, reason = contractRepeat_A(v, info, reason); isMeet {
			if contractRepeat_C(v, info) {
				return false, reason
			} else {
				reason = reason + "---合同类"
				return true, reason
			}
		} else {
			return false, reason
		}
	} else {
		//招标结果
		if isMeet, reason = tenderRepeat_A(v, info, reason); isMeet {
			if tenderRepeat_C(v, info) {
				return false, reason
			} else {
				reason = reason + "---类别空-招标类"
				return true, reason
			}
		} else {
			return false, reason
		}
	}

	return false, reason
}

//判重方法2
func quickHeavyMethodTwo(v *Info, info *Info, reason string) (bool, string) {
	isMeet := false
	isAgency :=false
	//招标类-代理机构不同-广泛前后缀比较
	if v.agency != info.agency && v.agency != "" && info.agency != "" {
		//新增一层判断
		if strings.Contains(v.agency, info.agency) || strings.Contains(info.agency, v.agency) {
			isAgency = true
		}else {
			return false, reason
		}
	}

	if (v.agency == info.agency && v.agency != "" && info.agency != "")|| isAgency {
		if info.subtype == "招标" || info.subtype == "邀标" || info.subtype == "询价" ||
			info.subtype == "竞谈" || info.subtype == "单一" || info.subtype == "竞价" ||
			info.subtype == "变更" || info.subtype == "其他" {
			//招标结果
			if isMeet, reason = tenderRepeat_B(v, info, reason); isMeet {
				if tenderRepeat_C(v, info) { //有不同
					return false, reason
				} else {
					reason = reason + "---招标类"
					return true, reason
				}
			} else {
				return false, reason
			}

		} else if info.subtype == "中标" || info.subtype == "成交" || info.subtype == "废标" || info.subtype == "流标" {
			//中标结果
			if isMeet, reason = winningRepeat_B(v, info, reason); isMeet {
				if winningRepeat_C(v, info) { //有不同
					return false, reason
				} else {
					reason = reason + "---中标类"
					return true, reason
				}
			} else {
				return false, reason
			}

		} else if info.subtype == "合同" || info.subtype == "验收" || info.subtype == "违规" {
			//合同
			if isMeet, reason = contractRepeat_B(v, info, reason); isMeet {
				if contractRepeat_C(v, info) { //有不同
					return false, reason
				} else {
					reason = reason + "---合同类"
					return true, reason
				}
			} else {
				return false, reason
			}
		} else {
			//招标结果
			if isMeet, reason = tenderRepeat_B(v, info, reason); isMeet {
				if tenderRepeat_C(v, info) { //有不同
					return false, reason
				} else {
					reason = reason + "---类别空-招标类"
					return true, reason
				}
			} else {
				return false, reason
			}
		}
	}

	//机构最2少一个为空
	if v.agency == "" || info.agency == "" {
		var repeat = false
		if repeat, reason = quickHeavyMethodOne(v, info, reason); repeat {
			reason = reason + "---机构最少一个空"
			return true, reason
		} else {
			return false, reason
		}
	}

	return false, reason
}

//招标_A
func tenderRepeat_A(v *Info, info *Info, reason string) (bool, string) {

	var ss string
	p1, p2, p3, p4, p9, p10, p11 := false, false, false, false, false, false, false
	if v.projectname != "" && v.projectname == info.projectname {
		ss = ss + "p1-名称-"
		p1 = true
	}
	if v.buyer != "" && v.buyer == info.buyer {
		ss = ss + "p2-单位-"
		p2 = true
	}
	if (v.projectcode != "" && v.projectcode == info.projectcode && len(v.projectcode) >= 5) ||
		(v.contractnumber != "" && v.contractnumber == info.contractnumber && len(v.contractnumber) >= 5) {
		ss = ss + "p3-编号组-"
		p3 = true
	}
	if v.budget != 0 && v.budget == info.budget {
		ss = ss + "p4-预算-"
		p4 = true
	}
	if v.bidopentime != 0 && v.bidopentime == info.bidopentime {
		ss = ss + "p9-开标时间相同-"
		p9 = true
	}
	if v.bidopenaddress != "" && v.bidopenaddress == info.bidopenaddress {
		ss = ss + "p10-开标地点-"
		p10 = true
	}
	if len([]rune(v.title)) > 10 && len([]rune(info.title)) > 10 &&
		(strings.Contains(v.title, info.title) || strings.Contains(info.title, v.title)) {
		ss = ss + "p11-标题-"
		p11 = true
	}

	if info.subtype !=""&&(p1 && p3 && p11)  {
		reason = reason + "满足招标A，3要素组合-" + ss + ","
		return true, reason
	}

	if  (p1 && p2 && p3) || (p1 && p2 && p4) || (p1 && p2 && p9) ||
		(p1 && p2 && p10) || (p1 && p2 && p11) || (p1 && p3 && p9) || (p1 && p3 && p10) ||
		(p1 && p4 && p9) || (p1 && p4 && p10) || (p2 && p3 && p4) ||
		(p2 && p3 && p9) || (p2 && p3 && p10) || (p2 && p3 && p11) ||
		(p2 && p4 && p9) || (p2 && p4 && p10) || (p2 && p4 && p11) ||
		(p3 && p4 && p9) || (p3 && p4 && p10) || (p3 && p4 && p11) ||
		(p4 && p9 && p10) || (p4 && p9 && p11) || (p9 && p10 && p11) {
		reason = reason + "满足招标A，3要素组合-" + ss + ","
		return true, reason
	}
	return false, reason
}

//招标_B
func tenderRepeat_B(v *Info, info *Info, reason string) (bool, string) {

	m, n := 0, 0
	if v.projectname != "" && v.projectname == info.projectname {
		m++
		n++
	}
	if v.buyer != "" && v.buyer == info.buyer {
		m++
	}
	if (v.projectcode != "" && v.projectcode == info.projectcode && len(v.projectcode) >= 5) ||
		(v.contractnumber != "" && v.contractnumber == info.contractnumber && len(v.contractnumber) >= 5) {
		m++
	}
	if v.budget != 0 && v.budget == info.budget {
		m++
	}
	if v.bidopentime != 0 && v.bidopentime == info.bidopentime {
		m++
	}
	//if v.bidopenaddress != "" && v.bidopenaddress == info.bidopenaddress {
	//	m++
	//}
	if len([]rune(v.title)) > 10 && len([]rune(info.title)) > 10 &&
		(strings.Contains(v.title, info.title) || strings.Contains(info.title, v.title)) {
		m++
		n++
	}
	if m >= 2 {
		if n == 2 && m == 2 {
			return false, reason
		} else {
			reason = reason + "满足招标B,六选二,"
			return true, reason
		}
	}
	return false, reason
}

//招标_C
func tenderRepeat_C(v *Info, info *Info) bool {

	if v.budget != 0 && info.budget != 0 && v.budget != info.budget {
		return true
	}
	if v.bidopentime != 0 && info.bidopentime != 0 && isBidopentimeInterval(info.bidopentime,v.bidopentime) {
		return true
	}
	return false
}

//中标_A
func winningRepeat_A(v *Info, info *Info, reason string) (bool, string) {

	var ss string
	p1, p2, p3, p5, p6, p11 := false, false, false, false, false, false
	if v.projectname != "" && v.projectname == info.projectname {
		ss = ss + "p1-项目名称-"
		p1 = true
	}
	if v.buyer != "" && v.buyer == info.buyer {
		ss = ss + "p2-单位-"
		p2 = true
	}
	if (v.projectcode != "" && v.projectcode == info.projectcode && len(v.projectcode) >= 5) ||
		(v.contractnumber != "" && v.contractnumber == info.contractnumber && len(v.contractnumber) >= 5) {
		ss = ss + "p3-编号组--"
		p3 = true
	}
	if v.bidamount != 0 && !isBidWinningAmount(v.bidamount,info.bidamount) {
		ss = ss + "p5-中标金-"
		p5 = true
	}
	if v.winner != "" && deleteExtraSpace(v.winner) == deleteExtraSpace(info.winner) {
		ss = ss + "p6-中标人-"
		p6 = true
	}


	if len([]rune(v.title)) > 10 && len([]rune(info.title)) > 10 &&
		(strings.Contains(v.title, info.title) || strings.Contains(info.title, v.title)) {
		ss = ss + "p11-标题-"
		p11 = true
	}

	if 	(p1 && p2 && p3) || (p1 && p2 && p5) || (p1 && p2 && p6) ||
		(p1 && p2 && p11)|| (p1 && p3 && p11)||
		(p1 && p3 && p5) || (p1 && p3 && p6) || (p1 && p5 && p6) ||
		(p2 && p3 && p5) || (p2 && p3 && p6) || (p2 && p3 && p11) ||
		(p2 && p5 && p6) || (p2 && p5 && p11) || (p2 && p6 && p11) ||
		(p3 && p5 && p6) || (p3 && p5 && p11) || (p3 && p6 && p11) ||
		(p5 && p6 && p11) {
		reason = reason + "满足中标A,3要素组合-" + ss + ","
		return true, reason
	}

	return false, reason
}

//中标_B
func winningRepeat_B(v *Info, info *Info, reason string) (bool, string) {

	m, n := 0, 0
	if v.projectname != "" && v.projectname == info.projectname {
		m++
		n++
	}
	if v.buyer != "" && v.buyer == info.buyer {
		m++
	}
	if (v.projectcode != "" && v.projectcode == info.projectcode && len(v.projectcode) >= 5) ||
		(v.contractnumber != "" && v.contractnumber == info.contractnumber && len(v.contractnumber) >= 5) {
		m++
	}
	if v.bidamount != 0 && !isBidWinningAmount(v.bidamount,info.bidamount) {
		m++
	}
	if v.winner != "" && deleteExtraSpace(v.winner) == deleteExtraSpace(info.winner) {
		m++
	}
	if len([]rune(v.title)) > 10 && len([]rune(info.title)) > 10 &&
		(strings.Contains(v.title, info.title) || strings.Contains(info.title, v.title)) {
		m++
		n++
	}
	if m >= 2 {
		if n == 2 && m == 2 {
			return false, reason
		} else {
			reason = reason + "满足中标B.六选二,"
			return true, reason
		}
	}
	return false, reason
}

//中标_C
func winningRepeat_C(v *Info, info *Info) bool {

	if v.bidamount != 0 && info.bidamount != 0 && isBidWinningAmount(v.bidamount,info.bidamount) {
		//避免抽错金额-
		if ((v.projectcode!=""&&info.projectcode!=""&&v.projectcode==info.projectcode)||
			(v.contractnumber!=""&&info.contractnumber!=""&&v.contractnumber==info.contractnumber)) &&
			(v.winner!=""&&info.winner!=""&&v.winner==info.winner) {
			return false
		}
		return true
	}
	if v.winner != "" && info.winner != "" && deleteExtraSpace(v.winner) != deleteExtraSpace(info.winner) {
		return true
	}
	return false
}

//合同_A
func contractRepeat_A(v *Info, info *Info, reason string) (bool, string) {

	isMeet_1 := false
	if isMeet_1, reason = tenderRepeat_A(v, info, reason); isMeet_1 {
		return true, reason
	}

	isMeet_2 := false
	if isMeet_2, reason = winningRepeat_A(v, info, reason); isMeet_2 {
		return true, reason
	}
	return false, reason
}

//合同_B
func contractRepeat_B(v *Info, info *Info, reason string) (bool, string) {

	isMeet_1 := false
	if isMeet_1, reason = tenderRepeat_B(v, info, reason); isMeet_1 {
		return true, reason
	}
	isMeet_2 := false
	if isMeet_2, reason = winningRepeat_B(v, info, reason); isMeet_2 {
		return true, reason
	}
	return false, reason
}

//合同_C
func contractRepeat_C(v *Info, info *Info) bool {

	if tenderRepeat_C(v, info) {
		return true
	}
	if winningRepeat_C(v, info) {
		return true
	}

	//合同类 - 新增编号
	if v.contractnumber != "" && info.contractnumber != "" && v.contractnumber != info.contractnumber {
		return true
	}
	if v.projectcode != "" && info.projectcode != "" && v.projectcode != info.projectcode {
		return true
	}

	return false
}









//快速低质量数据判重
func fastLowQualityHeavy(v *Info, info *Info, reason string) (bool, string) {
	//if !isTheSameDay(v.publishtime,info.publishtime) {
	//	return false,reason
	//}

	//区间间隔24小时
	if !isTimeIntervalPeriod(v.publishtime,info.publishtime) {
		return false,reason
	}


	//首先判定是否为低质量数据    info目标数据
	if info.title!=""&&(info.agency==""||v.agency=="")&&
		info.title==v.title&&info.projectcode==""&&info.contractnumber==""&&info.buyer=="" {
		isValue:=0//五要素判断
		if info.projectname != "" {//项目名称
			isValue++
		}
		if info.budget != 0 {//预算
			isValue++
		}
		if info.winner != ""{//中标单位
			isValue++
		}
		if info.bidamount != 0 {//中标金额
			isValue++
		}
		if isValue==0 {
			reason = reason + "---低质量-要素均为空-标题满足"
			return true, reason
		}else if isValue==1 {
			isMeet := false
			if isMeet, reason = judgeLowQualityData(v, info, reason); isMeet {
				reason = reason + "---低质量-有且一个要素组合"
				return true, reason
			}
		}else {

		}
	}
	return false,reason
}


//类别细节原因记录
func judgeLowQualityData(v *Info, info *Info, reason string) (bool, string) {
	if info.projectname!="" && info.projectname == v.projectname{//项目名称
		reason = reason + "---项目名称"
		return true,reason
	}
	if info.budget != 0 && info.budget == v.budget{//预算
		reason = reason + "---预算"
		return true,reason
	}
	if v.winner != "" && info.winner == v.winner{//中标单位
		reason = reason + "---中标单位"
		return true,reason
	}
	if v.bidamount != 0 && info.bidamount == v.bidamount{//中标金额
		reason = reason + "---中标金额"
		return true,reason
	}
	return false,reason
}