jy-data-extract/fullproject/src_v1/task.go

package main

import (
	"encoding/json"
	"fmt"
	"log"
	mu "mfw/util"

	"qfw/util"
	"regexp"
	"strings"
	"sync"
	"time"

	"github.com/robfig/cron"
	"go.mongodb.org/mongo-driver/bson/primitive"
)

/**
任务入口
全量、增量合并
更新、插入，内存清理
转换成info对象
**/

//项目合并对象
type ProjectTask struct {
	InitMinTime int64 //最小时间，小于0的处理一次
	name        string
	thread      int //线程数
	//查找锁
	findLock sync.Mutex
	wg       sync.WaitGroup
	//map锁
	AllIdsMapLock sync.Mutex
	//对应的id
	AllIdsMap map[string]*ID
	//采购单位、项目名称、项目编号
	mapPb, mapPn, mapPc map[string]*Key
	//流程数据  字段相同，直接合并
	mapHref     map[string]string
	mapHrefLock sync.Mutex
	//站点
	mapSite     map[string]*Site
	mapSiteLock sync.Mutex
	//bidtype、bidstatus 锁
	mapBidLock sync.Mutex
	//更新或新增通道
	updatePool chan []map[string]interface{}
	//savePool   chan map[string]interface{}
	//saveSign, updateSign chan bool
	//表名
	coll string
	//当前状态是全量还是增量
	currentType string //当前是跑全量还是跑增量
	//
	clearContimes int
	//当前时间
	currentTime int64
	//保存长度
	saveSize  int
	pici      int64
	validTime int64
	statusTime int64
	//	LockPool     chan *sync.Mutex
	//	LockPoolLock sync.Mutex
	//	m1, m23, m4  map[int]int
	//	l1, l23, l4  map[int]*sync.Mutex
	Brun bool
}

func NewPT() *ProjectTask {
	p := &ProjectTask{
		InitMinTime: int64(1325347200),
		name:        "全/增量对象",
		thread:      4,
		updatePool:  make(chan []map[string]interface{}, 5000),
		//savePool:    make(chan map[string]interface{}, 2000),
		wg:        sync.WaitGroup{},
		AllIdsMap: make(map[string]*ID, 5000000),
		mapPb:     make(map[string]*Key, 1500000),
		mapPn:     make(map[string]*Key, 5000000),
		mapPc:     make(map[string]*Key, 5000000),
		mapHref:   make(map[string]string, 1500000),
		mapSite:   make(map[string]*Site, 1000000),
		saveSize:  400,

		//saveSign:   make(chan bool, 1),
		//updateSign: make(chan bool, 1),
		coll:      ProjectColl,
		validTime: int64(util.IntAllDef(Sysconfig["validdays"], 150) * 86400),
		statusTime: int64(util.IntAllDef(Sysconfig["statusdays"], 7) * 86400),
	}
	return p
}

var P_QL *ProjectTask
var sp = make(chan bool, 5)

//初始化全量合并对象
func init() {
	P_QL = NewPT()
	log.Println(len(P_QL.updatePool))
	go P_QL.updateAllQueue()
	go P_QL.clearMem()
}

func (p *ProjectTask) updateAllQueue() {
	arru := make([][]map[string]interface{}, p.saveSize)
	indexu := 0
	for {
		select {
		case v := <-p.updatePool:
			arru[indexu] = v
			indexu++
			if indexu == p.saveSize {
				sp <- true
				go func(arru [][]map[string]interface{}) {
					defer func() {
						<-sp
					}()
					MongoTool.UpSertBulk(p.coll, arru...)
				}(arru)
				arru = make([][]map[string]interface{}, p.saveSize)
				indexu = 0
			}
		case <-time.After(1000 * time.Millisecond):
			if indexu > 0 {
				sp <- true
				go func(arru [][]map[string]interface{}) {
					defer func() {
						<-sp
					}()
					MongoTool.UpSertBulk(p.coll, arru...)
				}(arru[:indexu])
				arru = make([][]map[string]interface{}, p.saveSize)
				indexu = 0
			}
		}
	}
}

//项目合并内存更新
func (p *ProjectTask) clearMem() {
	c := cron.New()
	//在内存中保留最近6个月的信息
	//跑全量时每4分钟跑一次，跑增量时400分钟跑一次
	_ = c.AddFunc("50 0/15 * * * *", func() {
		if p.currentType == "ql" || p.clearContimes >= 60 {
			//跳过的次数清零
			p.clearContimes = 0
			//信息进入查找对比全局锁
			p.findLock.Lock()
			//defer p.findLock.Unlock()
			//合并进行的任务都完成
			p.wg.Wait()
			//遍历id
			//所有内存中的项目信息
			p.AllIdsMapLock.Lock()
			p.mapHrefLock.Lock()

			//清除计数
			clearNum := 0
			for k, v := range p.AllIdsMap {
				if p.currentTime-v.P.LastTime > p.validTime {
					clearNum++
					//删除id的map
					delete(p.AllIdsMap, k)
					//删除pb
					if v.P.Buyer != "" {
						ids := p.mapPb[v.P.Buyer]
						if ids != nil {
							ids.Lock.Lock()
							ids.Arr = deleteSlice(ids.Arr, k)
							if len(ids.Arr) == 0 {
								delete(p.mapPb, v.P.Buyer)
							}
							ids.Lock.Unlock()
						}
					}
					//删除mapPn
					for _, vn := range append([]string{v.P.ProjectName}, v.P.MPN...) {
						if vn != "" {
							ids := p.mapPn[vn]
							if ids != nil {
								ids.Lock.Lock()
								ids.Arr = deleteSlice(ids.Arr, k)
								if len(ids.Arr) == 0 {
									delete(p.mapPn, vn)
								}
								ids.Lock.Unlock()
							}
						}
					}
					//删除mapPc
					for _, vn := range append([]string{v.P.ProjectCode}, v.P.MPC...) {
						if vn != "" {
							ids := p.mapPc[vn]
							if ids != nil {
								ids.Lock.Lock()
								ids.Arr = deleteSlice(ids.Arr, k)
								if len(ids.Arr) == 0 {
									delete(p.mapPc, vn)
								}
								ids.Lock.Unlock()
							}
						}
					}
					for kHref, pid := range p.mapHref {
						if pid == k {
							delete(p.mapHref, kHref)
						}
					}
					v = nil
				}
			}
			p.mapHrefLock.Unlock()
			p.AllIdsMapLock.Unlock()
			p.findLock.Unlock()
			log.Println("清除完成:", clearNum, len(p.AllIdsMap), len(p.mapPn), len(p.mapPc), len(p.mapPb), len(p.mapHref))
		} else {
			p.clearContimes++
		}
	})
	c.Start()
	select {}
}

//全量合并
func (p *ProjectTask) taskQl(udpInfo map[string]interface{}) {
	defer util.Catch()
	//1、检查pubilshtime索引
	db, _ := udpInfo["db"].(string)
	if db == "" {
		db = MongoTool.DbName
	}
	coll, _ := udpInfo["coll"].(string)
	if coll == "" {
		coll = ExtractColl
	}
	thread := util.IntAllDef(udpInfo["thread"], 4)
	if thread > 0 {
		p.thread = thread
	}
	q, _ := udpInfo["query"].(map[string]interface{})
	if q == nil {
		q = map[string]interface{}{}
		lteid, _ := udpInfo["lteid"].(string)
		var idmap map[string]interface{}
		if len(lteid) > 15 {
			idmap = map[string]interface{}{
				"$lte": StringTOBsonId(lteid),
			}
		}
		gtid, _ := udpInfo["gtid"].(string)
		if len(gtid) > 15 {
			if idmap == nil {
				idmap = map[string]interface{}{}
			}
			idmap["$gt"] = StringTOBsonId(gtid)
		}
		if idmap != nil {
			q["_id"] = idmap
		}
	}
	//生成查询语句执行
	log.Println("查询语句：", q)
	p.enter(db, coll, q)

}

//增量合并
func (p *ProjectTask) taskZl(udpInfo map[string]interface{}) {
	defer util.Catch()
	//1、检查pubilshtime索引
	db, _ := udpInfo["db"].(string)
	if db == "" {
		db = MongoTool.DbName
	}
	coll, _ := udpInfo["coll"].(string)
	if coll == "" {
		coll = ExtractColl
	}
	thread := util.IntAllDef(udpInfo["thread"], 4)
	if thread > 0 {
		p.thread = thread
	}
	//开始id和结束id
	q, _ := udpInfo["query"].(map[string]interface{})
	gtid := udpInfo["gtid"].(string)
	lteid := udpInfo["lteid"].(string)
	if q == nil {
		q = map[string]interface{}{
			"_id": map[string]interface{}{
				"$gt":  StringTOBsonId(gtid),
				"$lte": StringTOBsonId(lteid),
			},
		}
	}
	if q != nil {
		//生成查询语句执行
		p.enter(db, coll, q)
	}
	if udpInfo["stop"] == nil {
		for i := 0; i < 5; i++ {
			sp <- true
		}
		for i := 0; i < 5; i++ {
			<-sp
		}
		log.Println("保存完成，生索引", p.pici)
		time.Sleep(5 * time.Second)
		nextNode(udpInfo, p.pici)
	}
}

//招标字段更新
func (p *ProjectTask) taskUpdateInfo(udpInfo map[string]interface{}) {
	defer util.Catch()
	db, _ := udpInfo["db"].(string)
	if db == "" {
		db = MongoTool.DbName
	}
	coll, _ := udpInfo["coll"].(string)
	if coll == "" {
		coll = ExtractColl
	}
	thread := util.IntAllDef(udpInfo["thread"], 4)
	if thread > 0 {
		p.thread = thread
	}
	q, _ := udpInfo["query"].(map[string]interface{})
	gtid := udpInfo["gtid"].(string)
	lteid := udpInfo["lteid"].(string)
	if q == nil {
		q = map[string]interface{}{
			"_id": map[string]interface{}{
				"$gt":  StringTOBsonId(gtid),
				"$lte": StringTOBsonId(lteid),
			},
			"is_m": 1,
		}
	}
	log.Println("查询语句：", q)
	p.enter(db, coll, q)
}

func StringTOBsonId(id string) primitive.ObjectID {
	objectId, _ := primitive.ObjectIDFromHex(id)
	return objectId
}

//通知下个节点nextNode
func nextNode(mapInfo map[string]interface{}, pici int64) {
	mapInfo["stype"] = "project"
	mapInfo["query"] = map[string]interface{}{
		"pici": pici,
	}
	for n, to := range toaddr {
		key := fmt.Sprintf("%d-%s-%d", pici, "project", n)
		mapInfo["key"] = key
		datas, _ := json.Marshal(mapInfo)
		node := &udpNode{datas, to, time.Now().Unix(), 0}
		udptaskmap.Store(key, node)
		_ = udpclient.WriteUdp(datas, mu.OP_TYPE_DATA, to)
	}
}

func (p *ProjectTask) enter(db, coll string, q map[string]interface{}) {
	defer util.Catch()
	defer func() {
		p.Brun = false
	}()
	p.Brun = true
	count, taskcount := 0, 0

	pool := make(chan bool, p.thread)
	log.Println("start project", q)
	sess := MongoTool.GetMgoConn()
	defer MongoTool.DestoryMongoConn(sess)

	infoPool := make(chan map[string]interface{}, 2000)
	over := make(chan bool)
	go func() {
	L:
		for {
			select {
			case tmp := <-infoPool:
				pool <- true
				taskcount++
				go func(tmp map[string]interface{}) {
					defer func() {
						<-pool
					}()
					if util.IntAll(tmp["repeat"]) == 0 {
						p.fillInPlace(tmp)
						info := ParseInfo(tmp)
						p.currentTime = info.Publishtime
						if p.currentType == "updateInfo" {
							//招标信息更改合并
							p.updateJudge(tmp, info)
						} else {
							//普通合并
							p.CommonMerge(tmp, info)
						}
					} else {
						//信息错误,进行更新
					}
				}(tmp)
			case <-over:
				break L
			}
		}

	}()
	ms := sess.DB(db).C(coll).Find(q).Sort("publishtime")
	if Sysconfig["hints"] != nil {
		ms.Hint(Sysconfig["hints"])
	}
	query := ms.Iter()
	var lastid interface{}
L:
	for {
		select {
		case <-queryClose:
			log.Println("receive interrupt sign")
			log.Println("close iter..", lastid, query.Cursor.Close(nil))
			queryCloseOver <- true
			break L
		default:
			tmp := make(map[string]interface{})
			if query.Next(&tmp) {
				lastid = tmp["_id"]
				if count%10000 == 0 {
					log.Println("current", count, lastid)
				}
				infoPool <- tmp
				count++
			} else {
				break L
			}
		}
	}
	time.Sleep(5 * time.Second)
	over <- true
	//阻塞
	for n := 0; n < p.thread; n++ {
		pool <- true
	}
	log.Println("所有线程执行完成...", count, taskcount)

}

var (
	//从标题获取项目编号
	titleGetPc  = regexp.MustCompile("^([-0-9a-zA-Z第号采招政询电审竞#]{8,}[-0-9a-zA-Z#]+)")
	titleGetPc1 = regexp.MustCompile("[\\[【(（](.{0,6}(编号|编码|项号|包号|代码|标段?号)[:：为])?([-0-9a-zA-Z第号采招政询电审竞#]{5,}([\\[\\]()（）][-0-9a-zA-Z第号采招审竞#]+[\\[\\]()（）][-0-9a-zA-Z第号采招审竞#]+)?)[\\]】)）]")
	titleGetPc2 = regexp.MustCompile("([-0-9a-zA-Z第号采政招询电审竞#]{8,}[-0-9a-zA-Z#]+)(.{0,5}公告)?$")
	//项目编号过滤
	pcReplace = regexp.MustCompile("([\\[【(（〖〔《｛﹝{](重|第?[二三四再]次.{0,4})[\\]】)）〗〕》｝﹞}])$|[\\[\\]【】()（）〖〗〔〕《》｛｝﹝﹞－；{}– 　]+|(号|重|第?[二三四五再]次(招标)?)$|[ _＿]+|((采购)?项目|采购(项目)?)$")
	//项目编号只是数字或只是字母4个以下
	StrOrNum = regexp.MustCompile("^[0-9_-]{1,4}$|^[a-zA-Z_-]{1,4}$")
	//纯数字或纯字母
	StrOrNum2 = regexp.MustCompile("^[0-9_-]+$|^[a-zA-Z_-]+$")
	//含分包词，招标未识别分包  合并到一个项目
	KeyPackage = regexp.MustCompile("[0-9a-zA-Z一二三四五六七八九十ⅠⅡⅢⅣⅤⅥⅦⅧⅨⅩ]+.{0,2}(包|段)|(包|段)[0-9a-zA-Z一二三四五六七八九十ⅠⅡⅢⅣⅤⅥⅦⅧⅨⅩ]+.{0,2}")
)

func (p *ProjectTask) CommonMerge(tmp map[string]interface{}, info *Info) {
	if info != nil && !((info.pnbval == 1 && info.Buyer != "") || info.pnbval == 0) {
		if jsonData, ok := tmp["jsondata"].(map[string]interface{}); ok {
			proHref := util.ObjToString(jsonData["projecthref"])
			if jsonData != nil && proHref != "" {
				//projectHref字段合并
				tmp["projecthref"] = proHref
				p.mapHrefLock.Lock()
				pid := p.mapHref[proHref]
				p.mapHrefLock.Unlock()
				if pid != "" {
					p.AllIdsMapLock.Lock()
					comparePro := p.AllIdsMap[pid].P
					p.AllIdsMapLock.Unlock()
					_, ex := p.CompareStatus(comparePro, info)
					p.UpdateProject(tmp, info, comparePro, -1, "AAAAAAAAAA", ex)
				} else {
					id, p1 := p.NewProject(tmp, info)
					p.mapHrefLock.Lock()
					p.mapHref[proHref] = id
					p.mapHrefLock.Unlock()
					p.AllIdsMapLock.Lock()
					p.AllIdsMap[id] = &ID{Id: id, P: p1}
					p.AllIdsMapLock.Unlock()
				}
			} else {
				//项目合并
				p.startProjectMerge(info, tmp)
			}
		} else {
			//项目合并
			p.startProjectMerge(info, tmp)
		}
	}
}

func ParseInfo(tmp map[string]interface{}) (info *Info) {
	bys, _ := json.Marshal(tmp)
	var thisinfo *Info
	_ = json.Unmarshal(bys, &thisinfo)
	if thisinfo == nil {
		return nil
	}
	if len(thisinfo.Topscopeclass) == 0 {
		thisinfo.Topscopeclass = []string{}
	}
	if len(thisinfo.Subscopeclass) == 0 {
		thisinfo.Subscopeclass = []string{}
	}

	if thisinfo.Publishtime == 0 {
		thisinfo.Publishtime = thisinfo.Comeintime
		tmp["publishtime"] = thisinfo.Comeintime
	}

	//从标题中查找项目编号
	res := titleGetPc.FindStringSubmatch(thisinfo.Title)
	if len(res) > 1 && len(res[1]) > 6 && thisinfo.ProjectCode != res[1] && !numCheckPc.MatchString(res[1]) && !_zimureg1.MatchString(res[1]) {
		thisinfo.PTC = res[1]
	} else {
		res = titleGetPc1.FindStringSubmatch(thisinfo.Title)
		if len(res) > 3 && len(res[3]) > 6 && thisinfo.ProjectCode != res[3] && !numCheckPc.MatchString(res[3]) && !_zimureg1.MatchString(res[3]) {
			thisinfo.PTC = res[3]
		} else {
			res = titleGetPc2.FindStringSubmatch(thisinfo.Title)
			if len(res) > 1 && len(res[1]) > 6 && thisinfo.ProjectCode != res[1] && !numCheckPc.MatchString(res[1]) && !_zimureg1.MatchString(res[1]) {
				thisinfo.PTC = res[1]
			}
		}
	}

	if thisinfo.ProjectName != "" && len([]rune(thisinfo.ProjectName)) > 0 {
		thisinfo.ProjectName = pcReplace.ReplaceAllString(thisinfo.ProjectName, "")
		if thisinfo.ProjectName != "" {
			thisinfo.pnbval++
		}
	}

	if thisinfo.ProjectCode != "" || thisinfo.PTC != "" {
		if thisinfo.ProjectCode != "" {
			thisinfo.ProjectCode = pcReplace.ReplaceAllString(thisinfo.ProjectCode, "")
			if thisinfo.pnbval == 0 && len([]rune(thisinfo.ProjectCode)) < 5 {
				thisinfo.ProjectCode = StrOrNum.ReplaceAllString(thisinfo.ProjectCode, "")
			}
		} else {
			thisinfo.PTC = pcReplace.ReplaceAllString(thisinfo.PTC, "")
			if thisinfo.pnbval == 0 && len([]rune(thisinfo.PTC)) < 5 {
				thisinfo.PTC = StrOrNum.ReplaceAllString(thisinfo.PTC, "")
			}
		}
		if thisinfo.ProjectCode != "" || thisinfo.PTC != "" {
			thisinfo.pnbval++
		}
	}
	if thisinfo.ProjectCode == thisinfo.PTC || strings.Index(thisinfo.ProjectCode, thisinfo.PTC) > -1 {
		thisinfo.PTC = ""
	}

	if thisinfo.Buyer != "" && len([]rune(thisinfo.Buyer)) > 2 {
		thisinfo.pnbval++
	} else {
		thisinfo.Buyer = ""
	}
	//winners整理
	winner, _ := tmp["winner"].(string)
	m1 := map[string]bool{}
	winners := []string{}
	if winner != "" {
		m1[winner] = true
		winners = append(winners, winner)
	}
	packageM, _ := tmp["package"].(map[string]interface{})
	if packageM != nil {
		thisinfo.HasPackage = true
		for _, p := range packageM {
			pm, _ := p.(map[string]interface{})
			pw, _ := pm["winner"].(string)
			if pw != "" && !m1[pw] {
				m1[pw] = true
				winners = append(winners, pw)
			}
		}
	}
	thisinfo.Winners = winners

	thisinfo.LenPC = len([]rune(thisinfo.ProjectCode))
	thisinfo.LenPTC = len([]rune(thisinfo.PTC))
	thisinfo.LenPN = len([]rune(thisinfo.ProjectName))
	return thisinfo
}

func (p *ProjectTask) updateJudge(tmp map[string]interface{}, info *Info) {
	index := -1
	pInfoId := ""
	p.AllIdsMapLock.Lock()
F:
	for k, ID := range p.AllIdsMap {
		for i, id := range ID.P.Ids {
			if info.Id == id {
				pInfoId = k
				index = i
				break F
			}
		}
	}
	p.AllIdsMapLock.Unlock()
	//未找到招标信息
	if index == -1 {
		if info != nil && !((info.pnbval == 1 && info.Buyer != "") || info.pnbval == 0) {
			p.currentTime = info.Publishtime
			p.startProjectMerge(info, tmp)
		}
	} else {
		tmpPro := MongoTool.FindById(ProjectColl, pInfoId)
		infoList := []interface{}(tmpPro["list"].(primitive.A))
		infoMap := infoList[index].(map[string]interface{})
		modifyMap, f := modifyEle(infoMap, tmp)
		//projecthref字段
		jsonData := tmp["jsondata"].(map[string]interface{})
		if jsonData != nil && jsonData["projecthref"] != nil {
			proHref := jsonData["projecthref"].(string)
			tmp["projecthref"] = proHref
			p.mapHrefLock.Lock()
			pid := p.mapHref[proHref]
			p.mapHrefLock.Unlock()
			if pid == pInfoId {
				p.modifyUpdate(pInfoId, index, info, tmp, tmpPro, modifyMap)
				return
			}
		}

		if f {
			//合并、修改
			log.Println("合并修改更新", "----------------------------")
			p.mergeAndModify(pInfoId, index, info, tmp, tmpPro, modifyMap)
		} else {
			//修改
			log.Println("修改更新", "----------------------------")
			p.modifyUpdate(pInfoId, index, info, tmp, tmpPro, modifyMap)
		}
	}
}

var Elements = []string{
	"projectname",
	"projectcode",
	"agency",
	"budget",
	"bidamount",
	"buyerperson",
	"area",
	"city",
	"publishtime",
}

/**
判断修改的字段是否是影响合并流程的要素字段
*/
func modifyEle(tmpPro map[string]interface{}, tmp map[string]interface{}) (map[string]interface{}, bool) {
	modifyMap := map[string]interface{}{}
	for k, _ := range tmpPro {
		for k1, _ := range tmp {
			if k == k1 && tmpPro[k] != tmp[k1] {
				modifyMap[k] = tmp[k1]
				break
			}
		}
	}
	for k, _ := range modifyMap {
		for _, str := range Elements {
			if k == str {
				return modifyMap, true
			}
		}
	}
	delete(modifyMap, "_id")
	return modifyMap, false
}

//补全位置信息
func (p *ProjectTask) fillInPlace(tmp map[string]interface{}) {
	area := util.ObjToString(tmp["area"])
	city := util.ObjToString(tmp["city"])
	district := util.ObjToString(tmp["district"])
	if area != "" && city != "" && district != "" {
		return
	}

	tmpSite := util.ObjToString(tmp["site"])
	if tmpSite == "" {
		return
	}
	p.mapSiteLock.Lock()
	defer p.mapSiteLock.Unlock()
	site := p.mapSite[tmpSite]
	if site != nil {
		if area != "" {
			if area == "全国" {
				tmp["area"] = site.Area
				tmp["city"] = site.City
				tmp["district"] = site.District
				return
			}
			if area != site.Area {
				return
			} else {
				if city == site.City {
					if district == "" {
						tmp["district"] = site.District
						return
					}
				} else if city == "" {
					tmp["city"] = site.City
					tmp["district"] = site.District
					return
				}
			}
		} else {
			tmp["area"] = site.Area
			tmp["city"] = site.City
			tmp["district"] = site.District
			return
		}
	}
}

//从数组中删除元素
func deleteSlice(arr []string, v string) []string {
	for k, v1 := range arr {
		if v1 == v {
			return append(arr[:k], arr[k+1:]...)
		}
	}
	return arr
}