jy-data-extract/udpfilterdup/src1/dataMethodMerge.go

package src1

import "qfw/util"

//合并字段-并更新merge字段的值
func mergeDataFields(source *Info, info *Info) (*Info, []int64, bool) {

	merge_recordMap := make(map[string]interface{}, 0)
	mergeArr := make([]int64, 0)
	//是否替换数据了-记录原始的数据
	is_replace := false
	//1、城市
	if source.area == "" || source.area == "全国" {
		//为空
		if info.area != "全国" && info.area != "" {
			merge_recordMap["area"] = source.area
			merge_recordMap["city"] = source.city
			source.area = info.area
			source.city = info.city
			mergeArr = append(mergeArr, 1)
			is_replace = true
		}
	} else {
		//不为空-查看站点相关-有值必替换
		if source.is_site {
			//是站点替换的城市
			merge_recordMap["site_area"] = source.area
			merge_recordMap["site_city"] = source.city
			mergeArr = append(mergeArr, 0)
			is_replace = true
			source.is_site = false

		}
	}
	//2、项目名称
	if source.projectname == "" && info.projectname != "" {
		merge_recordMap["projectname"] = source.projectname
		source.projectname = info.projectname
		mergeArr = append(mergeArr, 2)
		is_replace = true
	}
	//3、项目编号
	if source.projectcode == "" && info.projectcode != "" {
		merge_recordMap["projectcode"] = source.projectcode
		source.projectcode = info.projectcode
		mergeArr = append(mergeArr, 3)
		is_replace = true
	}
	//4、采购单位
	if source.buyer == "" && info.buyer != "" {
		merge_recordMap["buyer"] = source.buyer
		source.buyer = info.buyer
		mergeArr = append(mergeArr, 4)
		is_replace = true
	}
	//5、预算
	if source.budget == 0 && info.budget != 0 {
		merge_recordMap["budget"] = source.budget
		source.budget = info.budget
		mergeArr = append(mergeArr, 5)
		is_replace = true
	}
	//6、中标单位
	if source.winner == "" && info.winner != "" {
		merge_recordMap["winner"] = source.winner
		source.winner = info.winner
		mergeArr = append(mergeArr, 6)
		is_replace = true
	}
	//7、中标金额
	if source.bidamount == 0 && info.bidamount != 0 {
		merge_recordMap["bidamount"] = source.bidamount
		source.bidamount = info.bidamount
		mergeArr = append(mergeArr, 7)
		is_replace = true
	}
	//8、开标时间-地点
	if source.bidopentime == 0 && info.bidopentime != 0 {
		merge_recordMap["bidopentime"] = source.bidopentime
		source.bidopentime = info.bidopentime
		mergeArr = append(mergeArr, 8)
		is_replace = true
	}

	//9、合同编号
	if source.contractnumber == "" && info.contractnumber != "" {
		merge_recordMap["contractnumber"] = source.contractnumber
		source.contractnumber = info.contractnumber
		mergeArr = append(mergeArr, 9)
		is_replace = true
	}

	//10、发布时间
	if source.publishtime == 0 && info.publishtime != 0 {
		merge_recordMap["publishtime"] = source.publishtime
		source.publishtime = info.publishtime
		mergeArr = append(mergeArr, 10)
		is_replace = true
	}
	//11、代理机构
	if source.agency == "" && info.agency != "" {
		merge_recordMap["agency"] = source.agency
		source.agency = info.agency
		mergeArr = append(mergeArr, 11)
		is_replace = true
	}

	if is_replace { //有过替换更新
		//总次数+1
		source.mergemap["total_num"] = util.Int64All(source.mergemap["total_num"]) + 1
		merge_recordMap["num"] = util.Int64All(source.mergemap["total_num"])
		//和哪一个数据id进行非空替换的-记录
		key := info.id
		source.mergemap[key] = merge_recordMap
	}

	//待进一步优化
	return source, mergeArr, is_replace
}

//权重评估
func basicDataScore(v *Info, info *Info) bool {

	/*
	  权重评估
	  网站优先级判定规则：
	  1、国家>省级>市级>县区
	  2、政府采购>公共资源>官方网站|政府门户>社会公共招标平台|企业招标平台
	  3、同sitetype-分析weight
	  4、要素打分-分析
	*/
	v_score, info_score := -1, -1
	dict_v := SiteMap[v.site]
	dict_info := SiteMap[info.site]
	//先判断level
	if dict_v != nil {
		v_level := util.ObjToString(dict_v["level"])
		if v_level == "国家" {
			v_score = 4
		} else if v_level == "省级" {
			v_score = 3
		} else if v_level == "市级" {
			v_score = 2
		} else if v_level == "县区" {
			v_score = 1
		} else if v_level == "" {
		} else {
			v_score = 0
		}
	}

	if dict_info != nil {
		info_level := util.ObjToString(dict_info["level"])
		if info_level == "国家" {
			info_score = 4
		} else if info_level == "省级" {
			info_score = 3
		} else if info_level == "市级" {
			info_score = 2
		} else if info_level == "县区" {
			info_score = 1
		} else if info_level == "" {

		} else {
			v_score = 0
		}
	}

	if v_score > info_score {
		return true
	}
	if v_score < info_score {
		return false
	}

	//判断sitetype
	if dict_v != nil {
		v_sitetype := util.ObjToString(dict_v["sitetype"])
		if v_sitetype == "政府采购" {
			v_score = 4
		} else if v_sitetype == "公共资源" {
			v_score = 3
		} else if v_sitetype == "官方网站"|| v_sitetype == "政府门户" {
			v_score = 2
		} else if v_sitetype == "社会公共招标平台" || v_sitetype == "企业招标平台" {
			v_score = 1
		} else if v_sitetype == "" {
		} else {
			v_score = 0
		}
	}

	if dict_info != nil {
		info_sitetype := util.ObjToString(dict_info["sitetype"])
		if info_sitetype == "政府采购" {
			info_score = 4
		} else if info_sitetype == "公共资源" {
			info_score = 3
		} else if info_sitetype == "官方网站"|| info_sitetype == "政府门户" {
			info_score = 2
		} else if info_sitetype == "社会公共招标平台" || info_sitetype == "企业招标平台" {
			info_score = 1
		} else if info_sitetype == "" {
		} else {
			info_score = 0
		}
	}

	if v_score > info_score {
		return true
	}
	if v_score < info_score {
		return false
	}

	if v_score == info_score {//同sitetype 情况下   分析weight
		v_weight := util.IntAll(dict_v["weight"])
		info_weight := util.IntAll(dict_info["weight"])
		if v_weight>info_weight {
			return true
		}
		if info_weight>v_weight {
			return false
		}
	}

	//网站评估
	m, n := 0, 0
	if v.projectname != "" {
		m++
	}
	if v.buyer != "" {
		m++
	}
	if v.projectcode != "" || v.contractnumber != "" {
		m++
	}
	if v.budget != 0 {
		m++
	}
	if v.bidamount != 0 {
		m++
	}
	if v.winner != "" {
		m++
	}
	if v.bidopentime != 0 {
		m++
	}
	if v.bidopenaddress != "" {
		m++
	}
	if v.agency != "" {
		m = m + 2
	}
	if v.city != "" {
		m = m + 2
	}

	if info.projectname != "" {
		n++
	}
	if info.buyer != "" {
		n++
	}
	if info.projectcode != "" || info.contractnumber != "" {
		n++
	}
	if info.budget != 0 {
		n++
	}
	if info.bidamount != 0 {
		n++
	}
	if info.winner != "" {
		n++
	}
	if info.bidopentime != 0 {
		n++
	}
	if info.bidopenaddress != "" {
		n++
	}
	if info.agency != "" {
		n = n + 2
	}
	if info.city != "" {
		n = n + 2
	}

	if m > n {
		return true
	} else if m == n {
		if v.publishtime >= info.publishtime {
			return true
		} else {
			return false
		}
	} else {
		return false
	}
}