jy-data-extract/udpfilterdup/src/datamap.go

package main

import (
	"fmt"
	"log"
	qutil "qfw/util"
	"qfw/util/mongodb"
	"regexp"
	"strings"
	"sync"
	"time"
)

type Info struct {
	id    string //id
	title string //标题

	area           string  //省份
	city           string  //城市
	subtype        string  //信息类型
	buyer          string  //采购单位
	agency         string  //代理机构
	winner         string  //中标单位
	budget         float64 //预算金额
	bidamount      float64 //中标金额
	projectname    string  //项目名称
	projectcode    string  //项目编号
	contractnumber string  //合同编号
	publishtime    int64   //发布时间
	comeintime     int64   //入库时间
	bidopentime    int64   //开标时间
	bidopenaddress string  //开标地点

	site string //站点
	href string //正文的url

	repeatid         string                 //重复id
	titleSpecialWord bool                   //标题特殊词
	specialWord      bool                   //再次判断的特殊词
	mergemap         map[string]interface{} //合并记录
	is_site          bool                   //是否站点城市

}

var datelimit = float64(432000) //五天
var sitelock sync.Mutex         //锁

//一般数据判重
type datamap struct {
	lock   sync.Mutex //锁
	days   int        //保留几天数据
	data   map[string][]*Info
	keymap []string
	areakeys []string
	keys   map[string]bool
}

//历史更新数据
type historymap struct {
	lock   sync.Mutex //锁
	days   int        //保留几天数据
	data   map[string][]*Info
	keymap []string
	areakeys []string
	keys   map[string]bool
}

func TimedTaskDatamap(days int,lasttime int64) *datamap {
	log.Println("数据池开始重新构建")
	datelimit = qutil.Float64All(days * 86400)
	dm := &datamap{sync.Mutex{}, days, map[string][]*Info{}, []string{}, []string{},map[string]bool{}}
	if lasttime <0 {
		log.Println("数据池空数据")
		return dm
	}

	start := int(time.Now().Unix())
	sess := mgo.GetMgoConn()
	defer mgo.DestoryMongoConn(sess)
	query := map[string]interface{}{"publishtime": map[string]interface{}{
		"$lt": lasttime,
	}}
	log.Println("query", query)
	it := sess.DB(mgo.DbName).C(extract_back).Find(query).Sort("-publishtime").Iter()
	n, continuSum := 0, 0
	for tmp := make(map[string]interface{}); it.Next(&tmp); n++ {
		//qutil.IntAll(tmp["dataging"]) == 1
		if qutil.IntAll(tmp["repeat"]) == 1 || qutil.IntAll(tmp["repeat"]) == -1||qutil.IntAll(tmp["dataging"]) == 1  {

		} else {
			pt := tmp["publishtime"]
			pt_time := qutil.Int64All(pt)
			if qutil.Float64All(lasttime-pt_time) < datelimit {
				continuSum++
				info := NewInfo(tmp)
				dkey := qutil.FormatDateWithObj(&pt, qutil.Date_yyyyMMdd)
				k := fmt.Sprintf("%s_%s_%s", dkey, info.subtype, info.area)
				data := dm.data[k]
				if data == nil {
					data = []*Info{}
				}
				data = append(data, info)
				dm.data[k] = data
				dm.keys[dkey] = true
				//添加省
				isAreaExist :=false
				for _,v:= range dm.areakeys {
					if v==info.area {
						isAreaExist = true
					}
				}
				if !isAreaExist {
					areaArr := dm.areakeys
					areaArr = append(areaArr,info.area)
					dm.areakeys = areaArr
				}
			} else {
				break
			}
		}
		if n%50000 == 0 {
			log.Println("current 数据池:", n, continuSum)
		}
		tmp = make(map[string]interface{})
	}


	log.Printf("数据池构建完成：：%d秒,%d个\n", int(time.Now().Unix())-start, n)


	return dm
}


func NewDatamap(days int, lastid string) *datamap {
	datelimit = qutil.Float64All(days * 86400)
	dm := &datamap{sync.Mutex{}, days, map[string][]*Info{}, []string{},[]string{}, map[string]bool{}}
	if lastid == "" {
		return dm
	}
	//初始化加载数据
	sess := mgo.GetMgoConn()
	defer mgo.DestoryMongoConn(sess)
	query := map[string]interface{}{"_id": map[string]interface{}{
		"$lte": StringTOBsonId(lastid),
	}}
	log.Println("query", query)
	it := sess.DB(mgo.DbName).C(extract).Find(query).Sort("-_id").Iter()
	now1 := int64(0)
	n, continuSum := 0, 0
	for tmp := make(map[string]interface{}); it.Next(&tmp); n++ {
		if qutil.IntAll(tmp["repeat"]) == 1 || qutil.IntAll(tmp["repeat"]) == -1 {
			continuSum++
		} else {
			pt := tmp["comeintime"]
			if Is_Sort {
				pt = tmp["publishtime"]
			}
			pt_time := qutil.Int64All(pt)
			if pt_time <= 0 {
				continue
			}
			if now1 == 0 {
				now1 = pt_time
			}
			if qutil.Float64All(now1-pt_time) < datelimit {
				info := NewInfo(tmp)
				dkey := qutil.FormatDateWithObj(&pt, qutil.Date_yyyyMMdd)
				k := fmt.Sprintf("%s_%s_%s", dkey, info.subtype, info.area)
				data := dm.data[k]
				if data == nil {
					data = []*Info{}
				}
				data = append(data, info)
				dm.data[k] = data
				dm.keys[dkey] = true
				//添加省
				isAreaExist :=false
				for _,v:= range dm.areakeys {
					if v==info.area {
						isAreaExist = true
					}
				}
				if !isAreaExist {
					areaArr := dm.areakeys
					areaArr = append(areaArr,info.area)
					dm.areakeys = areaArr
				}
			} else {
				break
			}
		}
		if n%5000 == 0 {
			log.Println("current n:", n, continuSum)
		}
		tmp = make(map[string]interface{})
	}
	log.Println("load data:", n)
	return dm
}

//构建新历史数据池
func NewHistorymap(startid string, lastid string, startTime int64, lastTime int64) *historymap {
	datelimit = qutil.Float64All(5 * 86400)
	hm := &historymap{sync.Mutex{}, 5, map[string][]*Info{}, []string{},[]string{}, map[string]bool{}}
	if lastid == "" || startid == "" {
		return hm
	}
	//取startid之前5天
	sess_start := mgo.GetMgoConn()
	defer mgo.DestoryMongoConn(sess_start) //lte  gte
	it_start := sess_start.DB(mgo.DbName).C(extract).Find(mongodb.ObjToMQ(`{"_id":{"$lte":"`+startid+`"}}`,
		true)).Sort("-_id").Iter()
	m, n := 0, 0
	for tmp_start := make(map[string]interface{}); it_start.Next(&tmp_start); {

		if qutil.IntAll(tmp_start["repeat"]) == 1||qutil.IntAll(tmp_start["repeat"]) == -1 {
			continue
		}

		pt_s := tmp_start["comeintime"]
		if Is_Sort {
			pt_s = tmp_start["publishtime"]
		}
		pt_time := qutil.Int64All(pt_s)
		if pt_time <= 0 {
			continue
		}
		if qutil.Float64All(startTime-pt_time) <= datelimit {
			n++
			info := NewInfo(tmp_start)
			dkey := qutil.FormatDateWithObj(&pt_s, qutil.Date_yyyyMMdd)
			k := fmt.Sprintf("%s_%s_%s", dkey, info.subtype, info.area)
			data := hm.data[k]
			if data == nil {
				data = []*Info{}
			}
			data = append(data, info)
			hm.data[k] = data
			hm.keys[dkey] = true

			//添加省
			isAreaExist :=false
			for _,v:= range hm.areakeys {
				if v==info.area {
					isAreaExist = true
				}
			}
			if !isAreaExist {
				areaArr := hm.areakeys
				areaArr = append(areaArr,info.area)
				hm.areakeys = areaArr
			}
		} else {
			break
		}
		tmp_start = make(map[string]interface{})
	}

	log.Println("load history 前:", n)
	//取lastid之后5天
	sess_last := mgo.GetMgoConn()
	defer mgo.DestoryMongoConn(sess_last) //lte  gte
	it_last := sess_last.DB(mgo.DbName).C(extract).Find(mongodb.ObjToMQ(`{"_id":{"$gte":"`+lastid+`"}}`,
		true)).Sort("_id").Iter()

	for tmp_last := make(map[string]interface{}); it_last.Next(&tmp_last); {

		if qutil.IntAll(tmp_last["repeat"]) == 1||qutil.IntAll(tmp_last["repeat"]) == -1 {
			continue
		}

		pt_l := tmp_last["comeintime"]
		if Is_Sort {
			pt_l = tmp_last["publishtime"]
		}
		pt_time := qutil.Int64All(pt_l)
		if pt_time <= 0 {
			continue
		}
		if qutil.Float64All(pt_time-lastTime) <= datelimit {
			m++
			info := NewInfo(tmp_last)
			dkey := qutil.FormatDateWithObj(&pt_l, qutil.Date_yyyyMMdd)
			k := fmt.Sprintf("%s_%s_%s", dkey, info.subtype, info.area)
			data := hm.data[k]
			if data == nil {
				data = []*Info{}
			}
			data = append(data, info)
			hm.data[k] = data
			hm.keys[dkey] = true

			//添加省
			isAreaExist :=false
			for _,v:= range hm.areakeys {
				if v==info.area {
					isAreaExist = true
				}
			}
			if !isAreaExist {
				areaArr := hm.areakeys
				areaArr = append(areaArr,info.area)
				hm.areakeys = areaArr
			}
		} else {
			break
		}
		tmp_last = make(map[string]interface{})
	}

	log.Println("load history 后:", m)
	return hm
}

func NewInfo(tmp map[string]interface{}) *Info {
	subtype := qutil.ObjToString(tmp["subtype"])
	area := qutil.ObjToString(tmp["area"])
	if area == "A" {
		area = "全国"
	}
	info := &Info{}
	info.id = BsonTOStringId(tmp["_id"])
	info.title = qutil.ObjToString(tmp["title"])
	info.area = area
	info.subtype = subtype
	info.buyer = qutil.ObjToString(tmp["buyer"])
	info.projectname = qutil.ObjToString(tmp["projectname"])
	info.contractnumber = qutil.ObjToString(tmp["contractnumber"])
	info.projectcode = qutil.ObjToString(tmp["projectcode"])
	info.city = qutil.ObjToString(tmp["city"])
	info.agency = qutil.ObjToString(tmp["agency"])
	info.winner = qutil.ObjToString(tmp["winner"])
	info.budget = qutil.Float64All(tmp["budget"])
	info.bidamount = qutil.Float64All(tmp["bidamount"])
	info.publishtime = qutil.Int64All(tmp["publishtime"])
	info.comeintime = qutil.Int64All(tmp["comeintime"])
	info.bidopentime = qutil.Int64All(tmp["bidopentime"])
	info.bidopenaddress = qutil.ObjToString(tmp["bidopenaddress"])
	info.site = qutil.ObjToString(tmp["site"])
	info.href = qutil.ObjToString(tmp["href"])
	info.repeatid = qutil.ObjToString(tmp["repeatid"])

	info.specialWord = FilterRegTitle.MatchString(info.title)
	info.titleSpecialWord = FilterRegTitle_0.MatchString(info.title) ||FilterRegTitle_1.MatchString(info.title) || FilterRegTitle_2.MatchString(info.title)
	info.mergemap = *qutil.ObjToMap(tmp["merge_map"])
	if info.mergemap == nil {
		info.mergemap = make(map[string]interface{}, 0)
	}

	info.is_site = false

	return info
}

//判重方法
func (d *datamap) check(info *Info) (b bool, source *Info, reasons string) {

	reason := ""
	keys := []string{}
	d.lock.Lock()
	for k, _ := range d.keys { //不同时间段
		if info.area=="全国" {
			//匹配所有省
			for _,v := range d.areakeys{
				keys = append(keys, fmt.Sprintf("%s_%s_%s", k, info.subtype, v))
			}
		}else {
			//匹配指定省
			keys = append(keys, fmt.Sprintf("%s_%s_%s", k, info.subtype, info.area))
		}
		keys = append(keys, fmt.Sprintf("%s_%s_%s", k, info.subtype, "全国"))

	}
	d.lock.Unlock()
L:
	for _, k := range keys {
		d.lock.Lock()
		data := d.data[k]
		d.lock.Unlock()
		if len(data) > 0 { //对比v   找到同类型，同省或全国的数据作对比
			//log.Println(info.area,info.subtype,k)
			for _, v := range data {
				reason = ""
				if v.id == info.id { //正常重复
					//log.Println("相同id",info.id)
					return false, v, ""
				}

				//if v.id == "5c761a4fa5cb26b9b73d9512" &&info.id=="5c767bd1a5cb26b9b7a61597" {
				//	log.Println("测试数据")
				//}

				if info.subtype == v.subtype {
					if info.site != "" {
						sitelock.Lock()
						dict := SiteMap[info.site]
						sitelock.Unlock()
						if dict != nil {
							if info.area == "全国" && dict["area"] != "" {
								info.is_site = true
								info.area = qutil.ObjToString(dict["area"])
								info.city = qutil.ObjToString(dict["city"])
							} else {
								if info.city == "" && dict["city"] != "" {
									info.is_site = true
									info.area = qutil.ObjToString(dict["area"])
									info.city = qutil.ObjToString(dict["city"])
								}
							}
						}
					}
					//前置条件1 - 站点相关
					if info.site != "" && info.site == v.site {
						if info.href != "" && info.href == v.href {
							reason = "href相同"
							b = true
							source = v
							reasons = reason
							break L
						}
						if info.href != "" && info.href != v.href {
							reason = "href不同-"
						}
					}

					//前置条件2 - 标题相关,有且一个关键词
					if ((info.titleSpecialWord && !v.titleSpecialWord) || (info.specialWord && !v.specialWord)) &&
						info.title != v.title && v.title != "" && info.title != "" {
						continue
					}

					//前置条件3 - 标题相关,均含有关键词
					if ((info.titleSpecialWord && v.titleSpecialWord) || (info.specialWord && v.specialWord)) &&
						len([]rune(v.title)) > 10 && len([]rune(info.title)) > 10 && v.title != "" && info.title != "" {
						if !(strings.Contains(v.title, info.title) || strings.Contains(info.title, v.title)) {
							continue //无包含关系
						}
						if strings.Contains(v.title, info.title) || strings.Contains(info.title, v.title) {
							reason = reason + "标题关键词且包含关系"
							//继续二级金额判断
							if !againRepeat(v, info) {
								b = true
								source = v
								reasons = reason
								break
							}
						}
					}


					//新增快速数据过少判重
					if LowHeavy {
						repeat := false
						if repeat, reason = fastLowQualityHeavy(v, info, reason); repeat {
							b = true
							source = v
							reasons = reason
							break
						}
					}


					//代理机构相同-非空相等
					if v.agency != "" && info.agency != "" && v.agency == info.agency {
						reason = reason + "同机构-"
						repeat := false
						if repeat, reason = quickHeavyMethodTwo(v, info, reason); repeat {
							b = true
							source = v
							reasons = reason
							break
						}
					} else {
						reason = reason + "非同机构-"
						if info.city != "" && info.city == v.city {
							reason = reason + "同城-"
							repeat := false
							if repeat, reason = quickHeavyMethodTwo(v, info, reason); repeat {
								b = true
								source = v
								reasons = reason
								break
							}
						} else {
							reason = reason + "不同城-"
							repeat := false
							if repeat, reason = quickHeavyMethodOne(v, info, reason); repeat {
								b = true
								source = v
								reasons = reason
								break
							}
						}
					}
				}
			}

		}
	}

	//往预存数据 d 添加
	if !b {
		ct := info.comeintime
		if Is_Sort ||TimingTask{
			ct = info.publishtime
		}
		dkey := qutil.FormatDateByInt64(&ct, qutil.Date_yyyyMMdd)
		k := fmt.Sprintf("%s_%s_%s", dkey, info.subtype, info.area)
		d.lock.Lock()
		data := d.data[k]
		if data == nil {
			data = []*Info{info}
			d.data[k] = data
			if !d.keys[dkey] {
				d.keys[dkey] = true
				d.update(ct)
			}
		} else {
			data = append(data, info)
			d.data[k] = data
		}

		//添加省
		isAreaExist :=false
		for _,v:= range d.areakeys {
			if v==info.area {
				isAreaExist = true
			}
		}
		if !isAreaExist {
			areaArr := d.areakeys
			areaArr = append(areaArr,info.area)
			d.areakeys = areaArr
		}

		d.lock.Unlock()
	}

	return
}

func (h *historymap) checkHistory(info *Info) (b bool, source *Info, reasons string) {
	reason := ""
	keys := []string{}
	h.lock.Lock()
	for k, _ := range h.keys { //不同时间段
		if info.area=="全国" {
			//匹配所有省
			for _,v := range h.areakeys{
				keys = append(keys, fmt.Sprintf("%s_%s_%s", k, info.subtype, v))
			}
		}else {
			//匹配指定省
			keys = append(keys, fmt.Sprintf("%s_%s_%s", k, info.subtype, info.area))
		}
		keys = append(keys, fmt.Sprintf("%s_%s_%s", k, info.subtype, "全国"))

	}
	h.lock.Unlock()

L:
	for _, k := range keys {
		h.lock.Lock()
		data := h.data[k]
		h.lock.Unlock()
		if len(data) > 0 { //对比v   找到同类型，同省或全国的数据作对比
			for _, v := range data {
				reason = ""
				if v.id == info.id { //正常重复
					return false, v, ""
				}
				if info.subtype == v.subtype {
					if info.site != "" {
						sitelock.Lock()
						dict := SiteMap[info.site]
						sitelock.Unlock()
						if dict != nil {
							if info.area == "全国" && dict["area"] != "" {
								info.area = qutil.ObjToString(dict["area"])
								info.city = qutil.ObjToString(dict["city"])
							} else {
								if info.city == "" && dict["city"] != "" {
									info.area = qutil.ObjToString(dict["area"])
									info.city = qutil.ObjToString(dict["city"])
								}
							}
						}
					}
					//前置条件1 - 站点相关
					if info.site != "" && info.site == v.site {
						if info.href != "" && info.href == v.href {
							reason = "href相同"
							b = true
							source = v
							reasons = reason
							break L
						}
						if info.href != "" && info.href != v.href {
							reason = "href不同-"
						}
					}

					//前置条件2 - 标题相关,有且一个关键词
					if ((info.titleSpecialWord && !v.titleSpecialWord) || (info.specialWord && !v.specialWord)) &&
						info.title != v.title && v.title != "" && info.title != "" {
						continue
					}

					//前置条件3 - 标题相关,均含有关键词
					if ((info.titleSpecialWord && v.titleSpecialWord) || (info.specialWord && v.specialWord)) &&
						len([]rune(v.title)) > 10 && len([]rune(info.title)) > 10 && v.title != "" && info.title != "" {
						if !(strings.Contains(v.title, info.title) || strings.Contains(info.title, v.title)) {
							continue //无包含关系
						}
						if strings.Contains(v.title, info.title) || strings.Contains(info.title, v.title) {
							reason = reason + "标题关键词且包含关系"
							//继续二级金额判断
							if !againRepeat(v, info) {
								b = true
								source = v
								reasons = reason
								break
							}

						}
					}

					//新增快速数据过少判重
					if LowHeavy {
						repeat := false
						if repeat, reason = fastLowQualityHeavy(v, info, reason); repeat {
							b = true
							source = v
							reasons = reason
							break
						}
					}

					//代理机构相同-非空相等
					if v.agency != "" && info.agency != "" && v.agency == info.agency {
						reason = reason + "同机构-"
						repeat := false
						if repeat, reason = quickHeavyMethodTwo(v, info, reason); repeat {
							b = true
							source = v
							reasons = reason
							break
						}
					} else {
						reason = reason + "非同机构-"
						if info.city != "" && info.city == v.city {
							reason = reason + "同城-"
							repeat := false
							if repeat, reason = quickHeavyMethodTwo(v, info, reason); repeat {
								b = true
								source = v
								reasons = reason
								break
							}
						} else {
							reason = reason + "不同城-"
							repeat := false
							if repeat, reason = quickHeavyMethodOne(v, info, reason); repeat {
								b = true
								source = v
								reasons = reason
								break
							}
						}
					}
				}
			}
		}
	}

	//
	if b {
		if info.repeatid == source.id {
			b = false //重复-无变化-不处理
		}
	} else {
		if source != nil {
			if source.repeatid != "" { //未判重-有变化--记录
				b = true
				reason = "未判重记录"
				reasons = reason
			}
		}
	}
	//往预存数据 d 添加
	if !b {
		ct := info.comeintime
		if Is_Sort {
			ct = info.publishtime
		}
		dkey := qutil.FormatDateByInt64(&ct, qutil.Date_yyyyMMdd)
		k := fmt.Sprintf("%s_%s_%s", dkey, info.subtype, info.area)
		data := h.data[k]
		if data == nil {
			data = []*Info{info}
			h.data[k] = data
			if !h.keys[dkey] {
				h.keys[dkey] = true
				//h.update(ct)
			}
		} else {
			data = append(data, info)
			h.data[k] = data
		}

		//添加省
		isAreaExist :=false
		for _,v:= range h.areakeys {
			if v==info.area {
				isAreaExist = true
			}
		}
		if !isAreaExist {
			areaArr := h.areakeys
			areaArr = append(areaArr,info.area)
			h.areakeys = areaArr
		}
	}
	return
}

//替换原始数据池
func (d *datamap) replaceSourceData(replaceData *Info, replaceId string) {
	ct := replaceData.comeintime
	if Is_Sort||TimingTask {
		ct = replaceData.publishtime
	}
	dkey := qutil.FormatDateByInt64(&ct, qutil.Date_yyyyMMdd)
	k := fmt.Sprintf("%s_%s_%s", dkey, replaceData.subtype, replaceData.area)
	d.lock.Lock()
	data := d.data[k]
	if data == nil {
		data = []*Info{replaceData}
		d.data[k] = data
		if !d.keys[dkey] {
			d.keys[dkey] = true
		}
	} else {
		//遍历替换
		for k, v := range data {
			if v.id == replaceId {
				data[k] = replaceData
				break
			}
		}
		d.data[k] = data
	}
	d.lock.Unlock()
}

func (h *historymap) replaceSourceData(replaceData *Info, replaceId string) {
	ct := replaceData.comeintime
	if Is_Sort {
		ct = replaceData.publishtime
	}
	dkey := qutil.FormatDateByInt64(&ct, qutil.Date_yyyyMMdd)
	k := fmt.Sprintf("%s_%s_%s", dkey, replaceData.subtype, replaceData.area)
	h.lock.Lock()
	data := h.data[k]
	if data == nil {
		data = []*Info{replaceData}
		h.data[k] = data
		if !h.keys[dkey] {
			h.keys[dkey] = true
		}
	} else {
		//遍历替换
		for k, v := range data {
			if v.id == replaceId {
				data[k] = replaceData
				break
			}
		}
		h.data[k] = data
	}
	h.lock.Unlock()
}

func (d *datamap) update(t int64) {
	//每天0点清除历史数据
	d.keymap = d.GetLatelyFiveDay(t)
	m := map[string]bool{}
	for _, v := range d.keymap {
		m[v] = true
	}
	all, all1 := 0, 0
	for k, v := range d.data {
		all += len(v)
		if !m[k[:8]] {
			delete(d.data, k)
		}
	}
	for k, _ := range d.keys {
		if !m[k] {
			delete(d.keys, k)
		}
	}
	for _, v := range d.data {
		all1 += len(v)
	}
	//log.Println("更新前后数据:", all, all1)
}

func (d *datamap) GetLatelyFiveDay(t int64) []string  {
	array := make([]string, d.days)
	now := time.Unix(t, 0)
	for i := 0; i < d.days; i++ {
		array[i] = now.Format(qutil.Date_yyyyMMdd)
		now = now.AddDate(0, 0, -1)
	}
	return array
}

/*
**************************
******** 以下为判重 ********
**************************
 */

 //快速低质量数据判重
func fastLowQualityHeavy(v *Info, info *Info, reason string) (bool, string) {
	//首先判定是否为低质量数据    info目标数据
	if info.agency==v.agency&&info.title!=""&&
		info.title==v.title &&
		info.projectname==""&&info.projectcode==""&&info.contractnumber==""&&info.buyer=="" {
		isValue:=0//五要素判断
		if info.budget != 0 {//预算
			isValue++
		}
		if info.bidopentime != 0{//开标时间
			isValue++
		}
		if info.bidopenaddress!=""{//开标地点
			isValue++
		}
		if info.winner != ""{//中标单位
			isValue++
		}
		if info.bidamount != 0 {//中标金额
			isValue++
		}
		if isValue==0 {
			//if info.site!=v.site {
			//	log.Println("符合低质量条件条件0",info.id,"--",v.id)
			//}
			//log.Println("符合低质量条件条件0",info.id,"--",v.id)
			reason = reason + "---要素均为空，标题包含关系"
			return true, reason
		}else if isValue==1 {
			isMeet := false
			if isMeet, reason = judgeLowQualityData(v, info, reason); isMeet {
				//log.Println("符合低质量条件条件1",info.id,"--",v.id)
				reason = reason + "---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.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 info.budget != 0 && info.budget == v.budget{//预算
			reason = reason + "---招标类:预算"
			return true,reason
		}
		if info.bidopentime != 0 && info.bidopentime == v.bidopentime{//开标时间
			reason = reason + "---招标类:开标时间"
			return true,reason
		}
		if info.bidopenaddress!="" && info.bidopenaddress == v.bidopenaddress{//开标地点
			reason = reason + "---招标类:开标地点"
			return true,reason
		}
	} else if info.subtype == "中标" || info.subtype == "成交" || info.subtype == "废标" || info.subtype == "流标" {
		//中标结果
		if v.winner != "" && info.winner == v.winner{//中标单位
			reason = reason + "---中标类:中标单位"
			return true,reason
		}
		if v.bidamount != 0 && info.bidamount == v.bidamount{//中标金额
			reason = reason + "---中标类:中标金额"
			return true,reason
		}
	} else if info.subtype == "合同" || info.subtype == "验收" || info.subtype == "违规" {
		//合同
		if info.budget != 0 && info.budget == v.budget{//预算
			reason = reason + "---合同类:预算"
			return true,reason
		}
		if info.bidopentime != 0 && info.bidopentime == v.bidopentime{//开标时间
			reason = reason + "---合同类:开标时间"
			return true,reason
		}
		if info.bidopenaddress!="" && info.bidopenaddress == v.bidopenaddress{//开标地点
			reason = reason + "---合同类:开标地点"
			return true,reason
		}
		if v.winner != "" && info.winner == v.winner{//中标单位
			reason = reason + "---合同类:中标单位"
			return true,reason
		}
		if v.bidamount != 0 && info.bidamount == v.bidamount{//中标金额
			reason = reason + "---合同类:中标金额"
			return true,reason
		}
	} else {
		//招标结果
		if info.budget != 0 && info.budget == v.budget{//预算
			reason = reason + "---类别空-招标类:预算"
			return true,reason
		}
		if info.bidopentime != 0 && info.bidopentime == v.bidopentime{//开标时间
			reason = reason + "---类别空-招标类:开标时间"
			return true,reason
		}
		if info.bidopenaddress!="" && info.bidopenaddress == v.bidopenaddress{//开标地点
			reason = reason + "---类别空-招标类:开标地点"
			return true,reason
		}
	}
	return false,reason
}

//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
	if v.agency == info.agency && v.agency != "" && info.agency != "" {
		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
			}
		}
	}

	//不同
	if v.agency != info.agency && v.agency != "" && info.agency != "" {
		return false, reason
	}
	//机构最少一个为空
	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)-"
		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 (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
	}
	//原始地址...
	if v.buyer != "" && info.buyer != "" && v.buyer != info.buyer {
		return true
	}
	if v.bidopentime != 0 && info.bidopentime != 0 && v.bidopentime != info.bidopentime {
		return true
	}
	if v.bidopenaddress != "" && info.bidopenaddress != "" && v.bidopenaddress != info.bidopenaddress {
		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 && v.bidamount == info.bidamount {
	//	ss = ss + "p5(中标金)-"
	//	p5 = true
	//}
	//if v.winner != "" && v.winner == info.winner {
	//	ss = ss + "p6(中标人)-"
	//	p6 = 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 && 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 && v.bidamount != info.bidamount {
	//	return true
	//}
	if v.bidamount != 0 && info.bidamount != 0 && isBidWinningAmount(v.bidamount,info.bidamount) {
		return true
	}

	if v.winner != "" && info.winner != "" && deleteExtraSpace(v.winner) != deleteExtraSpace(info.winner) {
		return true
	}
	//原始地址...
	if v.buyer != "" && info.buyer != "" && v.buyer != info.buyer {
		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
	}
	return false
}

//再次金额判断
func againRepeat(v *Info, info *Info) bool {
	//相同采购单位下
	if info.buyer != "" && v.buyer == info.buyer {
		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 == "其他" || info.subtype == "变更" {
			//预算金额满足条件
			if v.budget != info.budget && v.budget != 0 && info.budget != 0 {
				return true
			}
		} else if info.subtype == "中标" || info.subtype == "成交" || info.subtype == "废标" ||
			info.subtype == "流标" || info.subtype == "合同" || info.subtype == "验收" ||
			info.subtype == "违规" {
			//中标金额单位满足条件
			if (isBidWinningAmount(v.bidamount,info.bidamount) && v.bidamount != 0 && info.bidamount != 0) ||
				(deleteExtraSpace(v.winner) != deleteExtraSpace(info.winner) && v.winner != "" && info.winner != "") {
				return true
			}
		} else {

		}
	}

	return false
}

//删除中标单位字符串中多余的空格(含tab)
func deleteExtraSpace(s string) string {
	//删除字符串中的多余空格，有多个空格时，仅保留一个空格
	s1 := strings.Replace(s, "  ", " ", -1)      //替换tab为空格
	regstr := "\\s{2,}"                          //两个及两个以上空格的正则表达式
	reg, _ := regexp.Compile(regstr)             //编译正则表达式
	s2 := make([]byte, len(s1))                  //定义字符数组切片
	copy(s2, s1)                                 //将字符串复制到切片
	spc_index := reg.FindStringIndex(string(s2)) //在字符串中搜索
	for len(spc_index) > 0 {                     //找到适配项
		s2 = append(s2[:spc_index[0]+1], s2[spc_index[1]:]...) //删除多余空格
		spc_index = reg.FindStringIndex(string(s2))            //继续在字符串中搜索
	}
	return string(s2)
}

//中标金额倍率：10000
func isBidWinningAmount(f1 float64 ,f2 float64) bool {

	if f1==f2||f1*10000==f2||f2*10000==f1 {
		return false
	}
	return true
}