data_ai/clean/c_all.go

package clean

import (
	qu "jygit.jydev.jianyu360.cn/data_processing/common_utils"
	"regexp"
)

// 通用清洗方法
var (
	fieldReg1 = regexp.MustCompile("([\\s ]+)")
	pcodeReg1 = regexp.MustCompile("([(（]变更[)）])$")
	pcodeReg2 = regexp.MustCompile("([\\*]+)")
)

func CleanFieldInfo(zhipu map[string]interface{}, fns []string, isTable bool) map[string]interface{} {
	data := map[string]interface{}{}
	//重点字段
	if s_area, s_city := CleanRegion(qu.ObjToString(zhipu["省份"]), qu.ObjToString(zhipu["城市"])); s_area != "" || s_city != "" {
		data["s_area"] = s_area
		data["s_city"] = s_city
	}
	if s_buyer := CleanBuyer(qu.ObjToString(zhipu["采购单位"])); s_buyer != "" {
		data["s_buyer"] = s_buyer
	}
	if s_pname := CleanPname(qu.ObjToString(zhipu["项目名称"])); s_pname != "" {
		data["s_projectname"] = s_pname
	}
	if s_pcode := CleanPcode(qu.ObjToString(zhipu["项目编号"]), fns); s_pcode != "" {
		data["s_projectcode"] = s_pcode
	}
	if s_biddingcode := CleanOtherCode(qu.ObjToString(zhipu["招标编号"])); s_biddingcode != "" {
		data["s_biddingcode"] = s_biddingcode
	}
	if s_packagecode := CleanOtherCode(qu.ObjToString(zhipu["标段编号"])); s_packagecode != "" {
		data["s_packagecode"] = s_packagecode
	}
	if s_contractcode := CleanOtherCode(qu.ObjToString(zhipu["合同编号"])); s_contractcode != "" {
		data["s_contractcode"] = s_contractcode
	}
	if s_budget, s_budget_unit := CleanMoney([]interface{}{zhipu["预算金额"], ""}); s_budget > 0.0 && s_budget < 1000000000.0 {
		if !s_budget_unit {
			if n_s_budget := ConvertMoney(s_budget, qu.ObjToString(zhipu["预算金额单位"])); n_s_budget > 0.0 && n_s_budget < 1000000000.0 {
				data["s_budget"] = n_s_budget
			} else {
				data["s_budget"] = s_budget
			}
		} else {
			data["s_budget"] = s_budget
		}
	}
	if s_bidamount, s_bidamount_unit := CleanMoney([]interface{}{zhipu["中标金额"], ""}); s_bidamount > 0.0 && s_bidamount < 1000000000.0 {
		if !s_bidamount_unit {
			if n_s_bidamount := ConvertMoney(s_bidamount, qu.ObjToString(zhipu["中标金额单位"])); n_s_bidamount > 0.0 && n_s_bidamount < 1000000000.0 {
				data["s_bidamount"] = n_s_bidamount
			} else {
				data["s_bidamount"] = s_bidamount
			}
		} else {
			data["s_bidamount"] = s_bidamount
		}
	}
	if s_agency := CleanAgency(qu.ObjToString(zhipu["代理机构"])); s_agency != "" {
		data["s_agency"] = s_agency
	}
	if s_winner := CleanWinner(qu.ObjToString(zhipu["中标单位"])); s_winner != "" {
		data["s_winner"] = s_winner
	}

	//其他字段
	if s_bidopenaddress := CleanOtherName(qu.ObjToString(zhipu["开标地点"])); s_bidopenaddress != "" {
		data["s_bidopenaddress"] = s_bidopenaddress
	}
	if s_biddiscount := CleanDiscount(qu.ObjToString(zhipu["中标金额折扣率"])); s_biddiscount > 0.0 {
		data["s_biddiscount"] = s_biddiscount
	}

	//时间相关
	if s_bidopentime := CleanTime(qu.ObjToString(zhipu["开标日期"])); s_bidopentime > 0 {
		data["s_bidopentime"] = s_bidopentime
	}
	if s_bidendtime := CleanTime(qu.ObjToString(zhipu["投标截止时间"])); s_bidendtime > 0 {
		data["s_bidendtime"] = s_bidendtime
	}
	if s_docstarttime := CleanTime(qu.ObjToString(zhipu["招标文件获取开始时间"])); s_docstarttime > 0 {
		data["s_docstarttime"] = s_docstarttime
	}
	if s_docendtime := CleanTime(qu.ObjToString(zhipu["招标文件获取结束时间"])); s_docendtime > 0 {
		data["s_docendtime"] = s_docendtime
	}

	//联系方式方式
	if s_buyerperson := CleanContactPerson(qu.ObjToString(zhipu["采购单位联系人"])); s_buyerperson != "" {
		data["s_buyerperson"] = s_buyerperson
	}
	if s_buyertel := CleanContactTel(qu.ObjToString(zhipu["采购单位联系方式"])); s_buyertel != "" {
		data["s_buyertel"] = s_buyertel
	}
	if s_agencyperson := CleanContactPerson(qu.ObjToString(zhipu["代理机构联系人"])); s_agencyperson != "" {
		data["s_agencyperson"] = s_agencyperson
	}
	if s_agencytel := CleanContactTel(qu.ObjToString(zhipu["代理机构联系方式"])); s_agencytel != "" {
		data["s_agencytel"] = s_agencytel
	}
	if s_winnerperson := CleanContactPerson(qu.ObjToString(zhipu["中标单位联系人"])); s_winnerperson != "" {
		data["s_winnerperson"] = s_winnerperson
	}
	if s_winnertel := CleanContactTel(qu.ObjToString(zhipu["中标单位联系方式"])); s_winnertel != "" {
		data["s_winnertel"] = s_winnertel
	}
	//标的物字段
	if zhipu["s_purchasinglist"] != nil {
		data["s_purchasinglist"] = zhipu["s_purchasinglist"]
	}

	//分包字段
	if zhipu["s_pkg"] != nil {
		data["s_pkg"] = zhipu["s_pkg"]
	}
	if zhipu["ispkg"] != nil {
		data["ispkg"] = zhipu["ispkg"]
	}
	//分类字段
	s_toptype := qu.ObjToString(zhipu["s_toptype"])
	s_subtype := qu.ObjToString(zhipu["s_subtype"])
	data["s_toptype"] = s_toptype
	data["s_subtype"] = s_subtype

	if s_subtype == "合同" || s_subtype == "中标" || s_subtype == "成交" {
		if qu.Float64All(data["s_budget"]) == qu.Float64All(data["s_bidamount"]) && qu.Float64All(data["s_budget"]) > 0.0 {
			delete(data, "s_budget")
		}
	}

	//表格类数据-分包金额以及外围字段不准确
	if isTable {
		delete(data, "s_budget")
		delete(data, "s_bidamount")
	}

	return data
}