123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973 |
- package main
- import (
- "fmt"
- "log"
- qutil "qfw/util"
- "qfw/util/mongodb"
- "strconv"
- "strings"
- "sync"
- "time"
- )
- type Info struct {
- id string
- 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
- publishtime int64 //发布时间
- comeintime int64 //采集时间
- bidopentime int64 //开标时间
- agencyaddr 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
- }
- var datelimit = float64(432000) //五天
- var sitelock sync.Mutex //锁
- //判重数据
- type datamap struct {
- lock sync.Mutex //锁
- days int //保留几天数据
- data map[string][]*Info
- keymap []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
- keys map[string]bool
- }
- func NewDatamap(days int, lastid string) *datamap {
- datelimit = qutil.Float64All(days * 86400)
- dm := &datamap{sync.Mutex{}, days, map[string][]*Info{}, []string{}, map[string]bool{}}
- if lastid == "" {
- return dm
- }
- //初始化加载数据
- sess := mgo.GetMgoConn()
- defer mgo.DestoryMongoConn(sess)
- it := sess.DB(mgo.DbName).C(extract).Find(mongodb.ObjToMQ(`{"_id":{"$lte":"`+lastid+`"}}`, true)).Sort("-_id").Iter()
- now1 := int64(0)
- n, continuSum := 0, 0
- for tmp := make(map[string]interface{}); it.Next(&tmp); n++ {
- if qutil.IntAll(tmp["repeat"]) == 1 || qutil.IntAll(tmp["repeat"]) == -1 {
- continuSum++
- } else {
- cm := tmp["comeintime"] //时间单位?
- comeintime := qutil.Int64All(cm)
- if comeintime == 0 {
- id := qutil.BsonIdToSId(tmp["_id"])[0:8]
- comeintime, _ = strconv.ParseInt(id, 16, 64)
- }
- if now1 == 0 {
- now1 = comeintime
- }
- if qutil.Float64All(now1-comeintime) < datelimit {
- info := NewInfo(tmp)
- //时间字符串
- dkey := qutil.FormatDateWithObj(&cm, qutil.Date_yyyyMMdd)
- //拼接的一个时间字符串 xxxx_类型_省份
- k := fmt.Sprintf("%s_%s_%s", dkey, info.subtype, info.area)
- data := dm.data[k]
- if data == nil {
- data = []*Info{}
- }
- data = append(data, info)
- dm.data[k] = data
- dm.keys[dkey] = true
- } else {
- break
- }
- }
- if n%5000 == 0 {
- log.Println("current n:", n, continuSum)
- }
- tmp = make(map[string]interface{})
- }
- log.Println("load data:", n)
- return dm
- }
- //构建新历史数据池
- func NewHistorymap(startid string, lastid string, startTime int64, lastTime int64) *historymap {
- datelimit = qutil.Float64All(5 * 86400)
- hm := &historymap{sync.Mutex{}, 5, map[string][]*Info{}, []string{}, map[string]bool{}}
- if lastid == "" || startid == "" {
- return hm
- }
- //取startid之前5天
- sess_start := mgo.GetMgoConn()
- defer mgo.DestoryMongoConn(sess_start) //lte gte
- it_start := sess_start.DB(mgo.DbName).C(extract).Find(mongodb.ObjToMQ(`{"_id":{"$lte":"`+startid+`"}}`,
- true)).Sort("-_id").Iter()
- m, n := 0, 0
- for tmp_start := make(map[string]interface{}); it_start.Next(&tmp_start); {
- cm := tmp_start["comeintime"]
- comeintime := qutil.Int64All(tmp_start["comeintime"])
- if comeintime == 0 {
- id := qutil.BsonIdToSId(tmp_start["_id"])[0:8]
- comeintime, _ = strconv.ParseInt(id, 16, 64)
- }
- if qutil.Float64All(startTime-comeintime) <= datelimit {
- n++
- info := NewInfo(tmp_start)
- dkey := qutil.FormatDateWithObj(&cm, 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
- } 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); {
- cm := tmp_last["comeintime"]
- comeintime := qutil.Int64All(tmp_last["comeintime"])
- if comeintime == 0 {
- id := qutil.BsonIdToSId(tmp_last["_id"])[0:8]
- comeintime, _ = strconv.ParseInt(id, 16, 64)
- }
- if qutil.Float64All(comeintime-lastTime) <= datelimit {
- m++
- info := NewInfo(tmp_last)
- dkey := qutil.FormatDateWithObj(&cm, 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
- } 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 = qutil.BsonIdToSId(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.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.bidopentime = qutil.Int64All(tmp["bidopentime"])
- info.agencyaddr = qutil.ObjToString(tmp["agencyaddr"])
- //info.detail = qutil.ObjToString(tmp["detail"])
- 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_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)
- }
- return info
- }
- //判重方法
- func (d *datamap) check(info *Info) (b bool, source *Info, reason string) {
- keys := []string{}
- d.lock.Lock()
- for k, _ := range d.keys { //不同时间段
- //...代码
- keys = append(keys, fmt.Sprintf("%s_%s_%s", k, info.subtype, info.area))
- if info.area != "全国" { //这个后续可以不要
- 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
- for _, v := range data {
- reason = ""
- if v.id == info.id { //正常重复
- return false, v, ""
- }
- //类型分组
- if info.subtype == v.subtype {
- //站点配置--
- if info.site != "" {
- sitelock.Lock()
- dict := SiteMap[info.site]
- sitelock.Unlock()
- if dict != nil {
- //临时改变--具体值
- 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
- 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 = "标题关键词且包含关系"
- b = true
- source = v
- break L
- }
- }
- //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
- 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
- break
- }
- } else {
- reason = reason + "不同城-"
- repeat := false
- if repeat, reason = quickHeavyMethodOne(v, info, reason); repeat {
- b = true
- source = v
- break
- }
- }
- }
- }
- }
- }
- }
- //往预存数据 d 添加
- if !b {
- 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
- }
- d.lock.Unlock()
- }
- return
- }
- func (h *historymap) checkHistory(info *Info) (b bool, source *Info, reasons string) {
- h.lock.Lock()
- defer h.lock.Unlock()
- keys := []string{}
- //不同时间段
- for k, _ := range h.keys {
- //...代码
- keys = append(keys, fmt.Sprintf("%s_%s_%s", k, info.subtype, info.area))
- if info.area != "全国" { //这个后续可以不要
- keys = append(keys, fmt.Sprintf("%s_%s_%s", k, info.subtype, "全国"))
- }
- }
- L:
- for _, k := range keys {
- data := h.data[k]
- 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, ""
- }
- //类型分组
- if info.subtype == v.subtype {
- //站点配置--
- if info.site != "" {
- dict := SiteMap[info.site]
- if dict != nil {
- //临时改变--具体值
- 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 = "标题关键词且包含关系"
- b = true
- source = v
- reasons = reason
- break L
- }
- }
- //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
- 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
- break
- }
- } else {
- reason = reason + "不同城-"
- repeat := false
- if repeat, reason = quickHeavyMethodOne(v, info, reason); repeat {
- b = true
- source = v
- break
- }
- }
- }
- }
- }
- }
- }
- //
- if b {
- //判重
- if info.repeatid == source.id {
- //重复-无变化-不处理
- b = false
- } else {
- if source.id != "" {
- //重复-有变化-覆盖记录处理
- }
- }
- } else {
- if source != nil {
- if source.repeatid != "" {
- //未判重-有变化--记录
- b = true
- reasons = "未判重记录"
- }
- }
- }
- //往预存数据 d 添加
- if !b {
- ct, _ := strconv.ParseInt(info.id[:8], 16, 64)
- 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
- }
- }
- return
- }
- //替换原始数据池
- func (d *datamap) replaceSourceData(replaceData *Info, replaceId string) {
- ct, _ := strconv.ParseInt(replaceId[:8], 16, 64)
- 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
- d.update(ct)
- }
- } 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, _ := strconv.ParseInt(replaceId[:8], 16, 64)
- dkey := qutil.FormatDateByInt64(&ct, qutil.Date_yyyyMMdd)
- k := fmt.Sprintf("%s_%s_%s", dkey, replaceData.subtype, replaceData.area)
- data := h.data[k]
- if data == nil {
- data = []*Info{replaceData}
- h.data[k] = data
- if !h.keys[dkey] {
- h.keys[dkey] = true
- //h.update(ct)
- }
- } else {
- //遍历替换
- for k, v := range data {
- if v.id == replaceId {
- data[k] = replaceData
- break
- }
- }
- h.data[k] = data
- }
- }
- //以下为判重 - 一揽子的方法
- //判重方法1
- func quickHeavyMethodOne(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 tenderRepeat_A(v, info, reason) {
- 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 winningRepeat_A(v, info, reason) {
- 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 contractRepeat_A(v, info, reason) {
- 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 tenderRepeat_A(v, info, reason) {
- 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) {
- //相同
- 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 tenderRepeat_B(v, info, reason) {
- 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 winningRepeat_B(v, info, reason) {
- 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 contractRepeat_B(v, info, reason) {
- 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 tenderRepeat_B(v, info, reason) {
- 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 {
- 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 {
- ss = ss + "p3(编号)-"
- 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.agencyaddr != "" && v.agencyaddr == info.agencyaddr {
- 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
- }
- return false
- }
- //招标_B
- func tenderRepeat_B(v *Info, info *Info, reason string) bool {
- 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 {
- m++
- }
- if v.budget != 0 && v.budget == info.budget {
- m++
- }
- if v.bidopentime != 0 && v.bidopentime == info.bidopentime {
- m++
- }
- if v.agencyaddr != "" && v.agencyaddr == info.agencyaddr {
- 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
- } else {
- reason = reason + "满足招标B,七选二,"
- return true
- }
- }
- return false
- }
- //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.bidopentime != 0 && info.bidopentime != 0 && v.bidopentime != info.bidopentime {
- return true
- }
- if v.agencyaddr != "" && info.agencyaddr != "" && v.agencyaddr != info.agencyaddr {
- return true
- }
- return false
- }
- //中标_A
- func winningRepeat_A(v *Info, info *Info, reason string) bool {
- 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 {
- ss = ss + "p3(编号)-"
- 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 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
- }
- return false
- }
- //中标_B
- func winningRepeat_B(v *Info, info *Info, reason string) bool {
- 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 {
- m++
- }
- if v.bidamount != 0 && v.bidamount == info.bidamount {
- m++
- }
- if v.winner != "" && v.winner == 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
- } else {
- reason = reason + "满足中标B.六选二,"
- return true
- }
- }
- return false
- }
- //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.winner != "" && info.winner != "" && v.winner != info.winner {
- return true
- }
- //原始地址...
- return false
- }
- //合同_A
- func contractRepeat_A(v *Info, info *Info, reason string) bool {
- if tenderRepeat_A(v, info, reason) {
- return true
- }
- if winningRepeat_A(v, info, reason) {
- return true
- }
- return false
- }
- //合同_B
- func contractRepeat_B(v *Info, info *Info, reason string) bool {
- if tenderRepeat_B(v, info, reason) {
- return true
- }
- if winningRepeat_B(v, info, reason) {
- return true
- }
- return false
- }
- //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
- }
- 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
- }
|