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) map[string]interface{} {
	data := map[string]interface{}{}
	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_budget := CleanMoney([]interface{}{zhipu["预算金额"], ""}); s_budget > 0.0 && s_budget < 1000000000.0 {
		data["s_budget"] = s_budget
	}
	if s_bidamount := CleanMoney([]interface{}{zhipu["中标金额"], ""}); s_bidamount > 0.0 && s_bidamount < 1000000000.0 {
		data["s_bidamount"] = s_bidamount
	}
	if s_winner := CleanWinner(qu.ObjToString(zhipu["中标单位"])); s_winner != "" {
		data["s_winner"] = s_winner
	}
	//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")
		}
	}

	//对于采购单位与中标单位一致的数据进行逻辑清洗

	return data
}