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

package extract

import (
	"encoding/json"
	"fmt"
	"jy/clear"
	db "jy/mongodbutil"
	"jy/pretreated"
	ju "jy/util"
	"log"
	qu "qfw/util"
	redis "qfw/util/redis"
	"reflect"
	"regexp"
	"strconv"
	"sync"
	"time"

	"gopkg.in/mgo.v2/bson"
)

var (
	lock          sync.RWMutex
	cut           = ju.NewCut()                          //获取正文并清理
	ExtLogs       map[*TaskInfo][]map[string]interface{} //抽取日志
	TaskList      map[string]*ExtractTask                //任务列表
	ClearTaskList map[string]*ClearTask                  //清理任务列表
	saveLimit     = 200                                  //抽取日志批量保存
	PageSize      = 5000                                 //查询分页
	Fields        = `{"title":1,"detail":1,"contenthtml":1,"site":1,"spidercode":1,"toptype":1,"subtype":1,"area":1,"city":1,"comeintime":1,"publishtime":1,"sensitive":1}`
	Fields2       = `{"budget":1,"bidamount":1,"title":1,"projectname":1,"winner":1}`
)

//启动测试抽取
func StartExtractTestTask(taskId, startId, num, resultcoll, trackcoll string) bool {
	defer qu.Catch()
	ext := &ExtractTask{}
	ext.Id = taskId
	ext.IsRun = true
	ext.InitTestTaskInfo(resultcoll, trackcoll)
	ext.TaskInfo.FDB = db.MgoFactory(1, 3, 120, ext.TaskInfo.FromDbAddr, ext.TaskInfo.FromDB)
	ext.InitRulePres()
	ext.InitRuleBacks()
	ext.InitRuleCore()
	ext.InitPkgCore()
	ext.InitTag()
	ext.InitClearFn()
	if ext.IsExtractCity { //版本上控制是否开始城市抽取
		//初始化城市DFA信息
		ext.InitDFA()
	}
	//质量审核
	ext.InitAuditFields()
	ext.InitAuditRule()
	ext.InitAuditClass()
	ext.InitAuditRecogField()

	//品牌抽取是否开启
	ju.IsBrandGoods = ju.Config["brandgoods"].(bool)
	return RunExtractTestTask(ext, startId, num)
}

func IdTrans(startId string) bson.ObjectId {
	defer qu.Catch()
	return bson.ObjectIdHex(startId)
}

//开始测试任务抽取
func RunExtractTestTask(ext *ExtractTask, startId, num string) bool {
	n, _ := strconv.Atoi(num)
	id := IdTrans(startId)
	if id.Valid() {
		query := bson.M{"_id": bson.M{"$gte": bson.ObjectIdHex(startId)}}
		list, _ := ext.TaskInfo.FDB.Find(ext.TaskInfo.FromColl, query, nil, Fields, false, 0, n)
		for _, v := range *list {
			if qu.ObjToString(v["sensitive"]) != "" { //去除含敏感词数据
				continue
			}
			//log.Println(v["_id"])
			j := PreInfo(v)
			ext.TaskInfo.ProcessPool <- true
			go ext.ExtractProcess(j)
		}
		return true
	} else {
		return false
	}
}

//启动抽取
func StartExtractTaskId(taskId string) bool {
	defer qu.Catch()
	isgo := false
	ext := TaskList[taskId]
	if ext == nil {
		ext = &ExtractTask{}
		ext.Id = taskId
		ext.InitTaskInfo()
		isgo = true
	} else {
		ext.Id = taskId
		ext.InitTaskInfo()
	}
	ext.TaskInfo.FDB = db.MgoFactory(2, 3, 120, ext.TaskInfo.FromDbAddr, ext.TaskInfo.FromDB)
	ext.TaskInfo.TDB = db.MgoFactory(1, 3, 120, ext.TaskInfo.ToDbAddr, ext.TaskInfo.ToDB)
	ext.InitRulePres()
	ext.InitRuleBacks()
	ext.InitRuleCore()
	ext.InitPkgCore()
	ext.InitTag()
	ext.InitClearFn()
	if ext.IsExtractCity { //版本上控制是否开始城市抽取
		//初始化城市DFA信息
		ext.InitDFA()
	}
	//质量审核
	ext.InitAuditFields()
	ext.InitAuditRule()
	ext.InitAuditClass()
	ext.InitAuditRecogField()

	//品牌抽取是否开启
	ju.IsBrandGoods = ju.Config["brandgoods"].(bool)

	ext.IsRun = true
	go ext.ResultSave()
	go ext.BidSave()
	if isgo {
		go RunExtractTask(taskId)
	}
	TaskList[taskId] = ext
	return true
}

//停止抽取
func StopExtractTaskId(taskId string) bool {
	defer qu.Catch()
	ext := TaskList[taskId]
	if ext != nil {
		ext.IsRun = false
		TaskList[taskId] = ext
	}
	//更新task.s_extlastid
	db.Mgo.UpdateById("task", taskId, `{"$set":{"s_extlastid":"`+ext.TaskInfo.LastExtId+`"}}`)
	return true
}

//开始抽取
func RunExtractTask(taskId string) {
	defer qu.Catch()
	ext := TaskList[taskId]
	query := bson.M{"_id": bson.M{"$gte": bson.ObjectIdHex(ext.TaskInfo.LastExtId)}}
	count := ext.TaskInfo.FDB.Count(ext.TaskInfo.FromColl, query)
	pageNum := (count + PageSize - 1) / PageSize
	limit := PageSize
	if count < PageSize {
		limit = count
	}
	log.Printf("count=%d,pageNum=%d,query=%v", count, pageNum, query)
	for i := 0; i < pageNum; i++ {
		query = bson.M{"_id": bson.M{"$gte": bson.ObjectIdHex(ext.TaskInfo.LastExtId)}}
		log.Printf("page=%d,query=%v", i+1, query)
		list, _ := ext.TaskInfo.FDB.Find(ext.TaskInfo.FromColl, query, nil, Fields, false, 0, limit)
		for _, v := range *list {
			if qu.ObjToString(v["sensitive"]) != "" { //去除含敏感词数据
				continue
			}
			//log.Println(v["_id"])
			if !ext.IsRun {
				break
			}
			j := PreInfo(v)
			ext.TaskInfo.ProcessPool <- true
			go ext.ExtractProcess(j)
			ext.TaskInfo.LastExtId = qu.BsonIdToSId(v["_id"])
		}
		db.Mgo.UpdateById("task", ext.Id, `{"$set":{"s_extlastid":"`+ext.TaskInfo.LastExtId+`"}}`)
		if !ext.IsRun {
			break
		}
	}
	//更新task.s_extlastid
	time.AfterFunc(1*time.Minute, func() { RunExtractTask(taskId) })
}

//信息预处理
func PreInfo(doc map[string]interface{}) *ju.Job {
	defer qu.Catch()
	detail := ""
	d1, _ := doc["detail"].(string)
	d2, _ := doc["contenthtml"].(string)
	if len(d1) >= len(d2) || d2 == "" {
		detail = d1
	} else {
		detail = d2
	}
	detail = ju.CutLableStr(detail)
	detail = cut.ClearHtml(detail)
	doc["detail"] = detail
	toptype := qu.ObjToString(doc["toptype"])
	if qu.ObjToString(doc["type"]) == "bid" {
		toptype = "结果"
	}
	if toptype == "" {
		toptype = "*"
	}
	j := &ju.Job{
		SourceMid:  qu.BsonIdToSId(doc["_id"]),
		Category:   toptype,
		Content:    qu.ObjToString(doc["detail"]),
		SpiderCode: qu.ObjToString(doc["spidercode"]),
		//Domain:     qu.ObjToString(doc["domain"]),
		//Href:       qu.ObjToString(doc["href"]),
		Title:     qu.ObjToString(doc["title"]),
		Data:      &doc,
		City:      qu.ObjToString(doc["city"]),
		Province:  qu.ObjToString(doc["area"]),
		Result:    map[string][]*ju.ExtField{},
		BuyerAddr: qu.ObjToString(doc["buyeraddr"]),
	}
	qu.Try(func() {
		pretreated.AnalyStart(j)
	}, func(err interface{}) {
		log.Println("pretreated.AnalyStart", err)
	})
	return j
}

//抽取
func (e *ExtractTask) ExtractProcess(j *ju.Job) {
	qu.Try(func() {
		doc := *j.Data
		//全局前置规则,结果覆盖doc属性
		for _, v := range e.RulePres {
			doc = ExtRegPre(doc, j, v, e.TaskInfo)
		}
		//抽取规则
		for _, vc := range e.RuleCores {
			tmp := ju.DeepCopy(doc).(map[string]interface{})
			//是否进入逻辑
			if !ju.Logic(vc.LuaLogic, tmp) {
				continue
			}
			//抽取-前置规则
			for _, v := range vc.RulePres {
				tmp = ExtRegPre(tmp, j, v, e.TaskInfo)
			}
			//log.Println("抽取-前置规则", tmp)

			//抽取-规则
			for _, v := range vc.RuleCores {
				ExtRegCore(vc.ExtFrom, tmp, j, v, e)
			}
			//log.Println("抽取-规则", tmp)

			//项目名称未能抽取到，标题来凑
			if vc.Field == "projectname" {
				if len(j.Result[vc.Field]) < 1 {
					j.Result[vc.Field] = append(j.Result[vc.Field], &ju.ExtField{vc.Field, "title", "title", "regexp", "title", vc.ExtFrom, j.Title, 0})
				}
			}

			//抽取-后置规则
			for _, v := range vc.RuleBacks {
				ExtRegBack(j, v, e.TaskInfo)
			}
			//log.Println("抽取-后置规则", tmp)
		}
		//全局后置规则
		for _, v := range e.RuleBacks {
			ExtRegBack(j, v, e.TaskInfo)
		}
		//候选人加入
		if len(j.Winnerorder) > 0 {
			winner := &ju.ExtField{
				Field:     "winner",
				Code:      "",
				RuleText:  "",
				Type:      "winnerorder",
				MatchType: "winnerorder",
				ExtFrom:   "",
				Value:     j.Winnerorder[0]["entname"],
				Score:     0,
			}
			if len([]rune(qu.ObjToString(j.Winnerorder[0]["entname"]))) < 4 {
				winner.Score = -5
			}
			winners := j.Result["winner"]
			if winners != nil {
				winners = append(winners, winner)
			} else {
				winners = []*ju.ExtField{}
				winners = append(winners, winner)
			}
			j.Result["winner"] = winners
		}
		//函数清理
		for key, val := range j.Result {
			for _, v := range val {
				lock.Lock()
				cfn := e.ClearFn[key]
				lock.Unlock()
				data := clear.DoClearFn(cfn, []interface{}{v.Value, j.Content})
				v.Value = data[0]
				//清理特殊符号
				if clear.AsyField[key] != nil || clear.SymField[key] != nil ||
					clear.MesField[key] != nil {
					text := qu.ObjToString(v.Value)
					text = clear.OtherClean(key, text)
					v.Value = text
				}
			}
		}
		PackageDetail(j, e) //处理分包信息
		//		bs, _ := json.Marshal(j.Result)
		//		log.Println("抽取结果", j.Title, j.SourceMid, string(bs))
		//分析抽取结果并保存 todo
		AnalysisSaveResult(j, e)
		<-e.TaskInfo.ProcessPool
	}, func(err interface{}) {
		log.Println("ExtractProcess err", err, (*j.Data)["_id"])
		<-e.TaskInfo.ProcessPool
	})
}

//前置过滤
func ExtRegPre(doc map[string]interface{}, j *ju.Job, in *RegLuaInfo, t *TaskInfo) map[string]interface{} {
	defer qu.Catch()
	before := ju.DeepCopy(doc).(map[string]interface{})
	extinfo := map[string]interface{}{}
	if in.IsLua {
		lua := ju.LuaScript{Code: in.Code, Name: in.Name, Doc: doc, Script: in.RuleText}
		if j != nil {
			lua.Block = j.Block
		}
		extinfo = lua.RunScript("pre")
		for k, v := range extinfo { //结果覆盖原doc
			doc[k] = v
		}
		AddExtLog("prereplace", j.SourceMid, before, extinfo, in, t) //抽取日志
	} else {
		key := qu.If(in.Field == "", "detail", in.Field).(string)
		text := qu.ObjToString(doc[key])
		extinfo[key] = in.RegPreBac.Reg.ReplaceAllString(text, "")
		doc[key] = extinfo[key]                                      //结果覆盖原doc
		AddExtLog("prereplace", j.SourceMid, before, extinfo, in, t) //抽取日志
	}
	return doc
}

//抽取-规则
func ExtRegCore(extfrom string, doc map[string]interface{}, j *ju.Job, in *RegLuaInfo, et *ExtractTask) {
	defer qu.Catch()
	//废标、流标、ppp等跳过
	b := IsExtract(in.Field, j.Title, j.Content)
	if !b {
		return
	}
	if in.IsLua {
		lua := ju.LuaScript{Code: in.Code, Name: in.Name, Doc: doc, Script: in.RuleText}
		lua.KvMap = getKvByLuaFields(extfrom, j, in, et.Tag)
		lua.Block = j.Block
		extinfo := lua.RunScript("core")
		for k, v := range extinfo {
			if k == in.Field {
				if j.Result[k] == nil {
					j.Result[k] = [](*ju.ExtField){}
				}
				if tmps, ok := v.([]map[string]interface{}); ok {
					for _, tmp := range tmps {
						j.Result[k] = append(j.Result[k],
							&ju.ExtField{k, qu.ObjToString(tmp["code"]), qu.ObjToString(tmp["ruletext"]), qu.ObjToString(tmp["type"]), qu.ObjToString(tmp["matchtype"]), extfrom, tmp["value"], 0})
					}
				}
			}
		}
		if len(extinfo) > 0 {
			AddExtLog("extract", j.SourceMid, nil, extinfo, in, et.TaskInfo) //抽取日志
		}
	} else {
		//全文正则
		text := qu.ObjToString(doc[extfrom])
		if in.Field != "" {
			extinfo := extRegCoreToResult(extfrom, text, j, in)
			if len(extinfo) > 0 {
				AddExtLog("extract", j.SourceMid, nil, extinfo, in, et.TaskInfo) //抽取日志
			}
		}
	}
}

//lua脚本根据属性设置提取kv值
func getKvByLuaFields(extfrom string, j *ju.Job, in *RegLuaInfo, t map[string][]*Tag) map[string][]map[string]interface{} {
	defer qu.Catch()
	kvmap := map[string][]map[string]interface{}{}
	for fieldname, field := range in.LFields {
		lock.Lock()
		tags := t[field] //获取对应标签库
		lock.Unlock()
		for _, bl := range j.Block {
			//冒号kv
			if bl.ColonKV != nil {
				kvs := bl.ColonKV.Kvs
				kvs2 := bl.ColonKV.Kvs_2
				//log.Println("ColonKV1", kvs)
				//log.Println("ColonKV2", kvs2)
				for _, tag := range tags {
					for _, kv := range kvs {
						if tag.Type == "string" {
							if kv.Key == tag.Key {
								text := ju.TrimLRSpace(kv.Value, "")
								if text != "" {
									kvmap[field] = append(kvmap[field], map[string]interface{}{
										"field":     field,
										"code":      in.Code,
										"ruletext":  tag.Key,
										"extfrom":   extfrom,
										"value":     text,
										"type":      "colon1",
										"matchtype": "tag_string",
									})
								}
								break
							}
						} else if tag.Type == "regexp" {
							if tag.Reg.MatchString(kv.Key) {
								text := ju.TrimLRSpace(kv.Value, "")
								if text != "" {
									kvmap[field] = append(kvmap[field], map[string]interface{}{
										"field":     field,
										"code":      in.Code,
										"ruletext":  tag.Key,
										"extfrom":   extfrom,
										"value":     text,
										"type":      "colon1",
										"matchtype": "tag_regexp",
									})
								}
								break
							}
						}
					}
					for _, kv := range kvs2 {
						if tag.Type == "string" {
							if kv.Key == tag.Key {
								text := ju.TrimLRSpace(kv.Value, "")
								if text != "" {
									kvmap[field] = append(kvmap[field], map[string]interface{}{
										"field":     field,
										"code":      in.Code,
										"ruletext":  tag.Key,
										"extfrom":   extfrom,
										"value":     text,
										"type":      "colon2",
										"matchtype": "tag_string",
									})
								}
								break
							}
						} else if tag.Type == "regexp" {
							if tag.Reg.MatchString(kv.Key) {
								text := ju.TrimLRSpace(kv.Value, "")
								if text != "" {
									kvmap[field] = append(kvmap[field], map[string]interface{}{
										"field":     field,
										"code":      in.Code,
										"ruletext":  tag.Key,
										"extfrom":   extfrom,
										"value":     text,
										"type":      "colon2",
										"matchtype": "tag_regexp",
									})
								}
								break
							}
						}
					}
				}
			}
			//空格kv
			if bl.SpaceKV != nil {
				kvs := bl.SpaceKV.Kvs
				//log.Println("SpaceKV", kvs)
				for _, tag := range tags {
					for _, kv := range kvs {
						if tag.Type == "string" {
							if kv.Key == tag.Key {
								text := ju.TrimLRSpace(kv.Value, "")
								if text != "" {
									kvmap[field] = append(kvmap[field], map[string]interface{}{
										"field":     field,
										"code":      in.Code,
										"ruletext":  tag.Key,
										"extfrom":   extfrom,
										"value":     text,
										"type":      "space",
										"matchtype": "tag_string",
									})
								}
								break
							}
						} else if tag.Type == "regexp" {
							if tag.Reg.MatchString(kv.Key) {
								text := ju.TrimLRSpace(kv.Value, "")
								if text != "" {
									kvmap[field] = append(kvmap[field], map[string]interface{}{
										"field":     field,
										"code":      in.Code,
										"ruletext":  tag.Key,
										"extfrom":   extfrom,
										"value":     text,
										"type":      "space",
										"matchtype": "tag_regexp",
									})
								}
								break
							}
						}
					}
				}
			}
			//表格kv
			if bl.TableKV != nil {
				tkv := bl.TableKV
				//log.Println("tkv", tkv)
				for k, v := range tkv.Kv {
					if k == fieldname {
						if len(tags) > -tkv.KvIndex[fieldname] {
							ruletext := ""
							if fieldname == "项目名称" && -tkv.KvIndex[fieldname] == -100 {
								ruletext = "项目名称"
							} else {
								ruletext = tags[-tkv.KvIndex[fieldname]].Key
							}
							kvmap[field] = append(kvmap[field], map[string]interface{}{
								"field":     field,
								"code":      in.Code,
								"ruletext":  ruletext,
								"extfrom":   "table",
								"value":     v,
								"type":      "table",
								"matchtype": "tag_string",
							})
						} else { //涉及其他待处理
							//log.Println(tags)
						}
					}
				}
			}
		}
	}
	return kvmap
}

//正则提取结果
func extRegCoreToResult(extfrom, text string, j *ju.Job, v *RegLuaInfo) map[string][]map[string]interface{} {
	defer qu.Catch()
	extinfo := map[string][]map[string]interface{}{}
	if v.RegCore.Bextract { //正则是两部分的，可以直接抽取的（含下划线）
		apos := v.RegCore.Reg.FindAllStringSubmatchIndex(text, -1)
		if len(apos) > 0 {
			pos := apos[0]
			for k, p := range v.RegCore.ExtractPos {
				if len(pos) > p {
					if pos[p] == -1 || pos[p+1] == -1 {
						continue
					}
					val := text[pos[p]:pos[p+1]]
					tmps := []map[string]interface{}{}
					tmp := map[string]interface{}{
						"field":     v.Field,
						"code":      v.Code,
						"ruletext":  v.RuleText,
						"extfrom":   extfrom,
						"value":     val,
						"type":      "regexp",
						"matchtype": "regcontent",
					}
					tmps = append(tmps, tmp)
					extinfo[k] = tmps
					if val != "" {
						if j.Result[v.Field] == nil {
							j.Result[k] = [](*ju.ExtField){}
						}
						j.Result[k] = append(j.Result[k], &ju.ExtField{k, v.Code, v.RuleText, "regexp", "regcontent", extfrom, val, 0})
					}
				}
			}
		}
	} else {
		pos := v.RegCore.Reg.FindStringIndex(text)
		val := ""
		if len(pos) == 2 {
			text = text[pos[1]:]
			rs := regexp.MustCompile("[^\r\n\t]+")
			tmp := rs.FindAllString(text, -1)
			if len(tmp) > 0 {
				val = tmp[0]
			}
		}
		if val != "" {
			tmps := []map[string]interface{}{}
			tmp := map[string]interface{}{
				"field":     v.Field,
				"code":      v.Code,
				"ruletext":  v.RuleText,
				"extfrom":   extfrom,
				"value":     val,
				"type":      "regexp",
				"matchtype": "regcontent",
			}
			tmps = append(tmps, tmp)
			extinfo[v.Field] = tmps
			if j.Result[v.Field] == nil {
				j.Result[v.Field] = [](*ju.ExtField){}
			}
			j.Result[v.Field] = append(j.Result[v.Field], &ju.ExtField{v.Field, v.Code, v.RuleText, "regexp", "regcontent", extfrom, val, 0})
		}
	}
	return extinfo
}

//后置过滤
func ExtRegBack(j *ju.Job, in *RegLuaInfo, t *TaskInfo) {
	defer qu.Catch()
	if in.IsLua {
		result := GetResultMapForLua(j)
		lua := ju.LuaScript{Code: in.Code, Name: in.Name, Result: result, Script: in.RuleText}
		if j != nil {
			lua.Block = j.Block
		}
		extinfo := lua.RunScript("back")
		for k, v := range extinfo {
			if tmps, ok := v.([]map[string]interface{}); ok {
				j.Result[k] = [](*ju.ExtField){}
				for _, tmp := range tmps {
					j.Result[k] = append(j.Result[k], &ju.ExtField{k, qu.ObjToString(tmp["code"]), qu.ObjToString(tmp["ruletext"]), qu.ObjToString(tmp["type"]), qu.ObjToString(tmp["matchtype"]), qu.ObjToString(tmp["extfrom"]), tmp["value"], 0})
				}
			}
		}
		if len(extinfo) > 0 {
			AddExtLog("clear", j.SourceMid, result, extinfo, in, t) //抽取日志
		}
	} else {
		extinfo := map[string]interface{}{}
		if in.Field != "" {
			if j.Result[in.Field] != nil {
				tmp := j.Result[in.Field]
				exts := []interface{}{}
				for k, v := range tmp {
					if v.Type == "table" { //table抽取到的数据不清理
						continue
					}
					text := qu.ObjToString(v.Value)
					if text != "" {
						text = in.RegPreBac.Reg.ReplaceAllString(text, in.RegPreBac.Replace)
					}
					j.Result[in.Field][k].Value = text
					exts = append(exts, map[string]interface{}{
						"field":     v.Field,
						"code":      v.Code,
						"ruletext":  v.RuleText,
						"type":      v.Type,
						"matchtype": v.MatchType,
						"extfrom":   v.ExtFrom,
						"value":     text,
					})
				}
				extinfo[in.Field] = exts
				if len(extinfo) > 0 {
					AddExtLog("clear", j.SourceMid, tmp, extinfo, in, t) //抽取日志
				}
			}
		} else {
			for key, tmp := range j.Result {
				exts := []interface{}{}
				for k, v := range tmp {
					if v.Type == "table" { //table抽取到的数据不清理
						continue
					}
					text := qu.ObjToString(v.Value)
					if text != "" {
						text = in.RegPreBac.Reg.ReplaceAllString(text, in.RegPreBac.Replace)
					}
					j.Result[key][k].Value = text
					exts = append(exts, map[string]interface{}{
						"field":     v.Field,
						"code":      v.Code,
						"ruletext":  v.RuleText,
						"type":      v.Type,
						"matchtype": v.MatchType,
						"extfrom":   v.ExtFrom,
						"value":     text,
					})
				}
				extinfo[key] = exts
			}
			if len(extinfo) > 0 {
				AddExtLog("clear", j.SourceMid, j.Result, extinfo, in, t) //抽取日志
			}
		}
	}
}

//获取抽取结果map[string][]interface{},lua脚本使用
func GetResultMapForLua(j *ju.Job) map[string][]map[string]interface{} {
	defer qu.Catch()
	result := map[string][]map[string]interface{}{}
	for key, val := range j.Result {
		if result[key] == nil {
			result[key] = []map[string]interface{}{}
		}
		for _, v := range val {
			tmp := map[string]interface{}{
				"field":     v.Field,
				"code":      v.Code,
				"ruletext":  v.RuleText,
				"value":     v.Value,
				"type":      v.Type,
				"matchtype": v.MatchType,
				"extfrom":   v.ExtFrom,
			}
			result[key] = append(result[key], tmp)
		}
	}
	return result
}

//抽取日志
func AddExtLog(ftype, sid string, before interface{}, extinfo interface{}, v *RegLuaInfo, t *TaskInfo) {
	defer qu.Catch()
	if !t.IsEtxLog {
		return
	}
	logdata := map[string]interface{}{
		"code":       v.Code,
		"name":       v.Name,
		"type":       ftype,
		"ruletext":   v.RuleText,
		"islua":      v.IsLua,
		"field":      v.Field,
		"version":    t.Version,
		"taskname":   t.Name,
		"before":     before,
		"extinfo":    extinfo,
		"sid":        sid,
		"comeintime": time.Now().Unix(),
	}
	lock.Lock()
	ExtLogs[t] = append(ExtLogs[t], logdata)
	lock.Unlock()
}

//保存抽取日志
func SaveExtLog() {
	defer qu.Catch()
	tmpLogs := map[*TaskInfo][]map[string]interface{}{}
	lock.Lock()
	tmpLogs = ExtLogs
	ExtLogs = map[*TaskInfo][]map[string]interface{}{}
	lock.Unlock()
	for k, v := range tmpLogs {
		if len(v) < saveLimit {
			db.Mgo.SaveBulk(k.TrackColl, v...)
		} else {
			for {
				if len(v) > saveLimit {
					tmp := v[:saveLimit]
					db.Mgo.SaveBulk(k.TrackColl, tmp...)
					v = v[saveLimit:]
				} else {
					db.Mgo.SaveBulk(k.TrackColl, v...)
					break
				}
			}
		}
	}
	time.AfterFunc(10*time.Second, SaveExtLog)
}

type FieldValue struct {
	Value interface{}
	Count int
}

//分析抽取结果并保存
func AnalysisSaveResult(j *ju.Job, e *ExtractTask) {
	qu.Try(func() {
		doc := j.Data
		result := j.Result
		_id := qu.BsonIdToSId((*doc)["_id"])
		iscore, _ := ju.Config["fieldscore"].(bool)
		if iscore { //打分
			result = ScoreFields(j)
		}
		//结果排序
		values := map[string][]*ju.SortObject{}
		for key, val := range result {
			fieldValue := map[string][]interface{}{}
			if iscore { //走打分
				for _, v := range val {
					if len(fmt.Sprint(v.Value)) < 1 {
						continue //去除空串
					}
					fieldValue[fmt.Sprint(v.Value)+v.Type] = []interface{}{v.Score, v.Value}
				}
			} else { //不走打分，按出现频次
				for _, v := range val {
					if len(fmt.Sprint(v.Value)) < 1 {
						continue //去除空串
					}
					if fieldValue[fmt.Sprint(v.Value)] == nil {
						fieldValue[fmt.Sprint(v.Value)] = []interface{}{0, v.Value}
					} else {
						fieldValue[fmt.Sprint(v.Value)][0] = qu.IntAll(fieldValue[fmt.Sprint(v.Value)][0]) + 1
					}
				}
			}
			objects := []*ju.SortObject{}
			for k, v := range fieldValue {
				ValueStr := "" //第二排序
				if reflect.TypeOf(v[1]).String() == "string" {
					ValueStr = qu.ObjToString(v[1])
				}
				tmp := &ju.SortObject{
					Key:      k,
					Value:    qu.IntAll(v[0]),
					Object:   v[1],
					ValueStr: ValueStr,
				}
				objects = append(objects, tmp)
			}
			values[key] = ju.ExtSort(objects)
		}
		//从排序结果中取值
		tmp := map[string]interface{}{} //抽取值
		for key, val := range values {
			for _, v := range val { //取第一个非负数
				if v.Key != "" && v.Value > -1 {
					tmp[key] = v.Object
					break
				}
			}
		}
		if len(j.PackageInfo) > 0 { //分包信息
			tmp["package"] = j.PackageInfo
		}
		if len(j.Winnerorder) > 0 { //候选人信息
			tmp["winnerorder"] = j.Winnerorder
		}
		for k, v := range *doc {
			//去重冗余字段
			if k == "detail" || k == "contenthtml" || k == "site" || k == "spidercode" {
				continue
			}
			if tmp[k] == nil {
				tmp[k] = v
			}
		}

	//质量审核
	if ju.Config["qualityaudit"].(bool) {
		e.QualityAudit(tmp)
	}
	if e.IsExtractCity { //城市抽取
		b, p, c, d := e.TransmitData(tmp, _id) //抽取省份城市
		//log.Println("省份---", p, "城市---", c, "区---", d)
		tmp["district"] = d
		if b {
			tmp["city"] = c
			tmp["area"] = p
		}
	}
	//品牌抽取
	if ju.IsBrandGoods {
		tmp["checkhas"] = map[string]int{
			"hastable": j.HasTable,
			"hasgoods": j.HasGoods,
			"hasbrand": j.HasBrand,
			"haskey":   j.HasKey,
		}
		if len(j.BrandData) > 0 {
			tmp["brand"] = j.BrandData
		}
		//log.Println("============", j.HasBrand, j.HasGoods, j.HasKey, j.HasTable, j.BrandData)
	}
	if e.TaskInfo.TestColl == "" {
		if len(tmp) > 0 { //保存抽取结果
			tmparr := []map[string]interface{}{
				map[string]interface{}{
					"_id": qu.StringTOBsonId(_id),
				},
				map[string]interface{}{"$set": tmp},
			}
			e.BidArr = append(e.BidArr, tmparr)
		}
		if b, ok := ju.Config["saveresult"].(bool); ok && b {
			id := tmp["_id"]
			tmp["result"] = result
			delete(tmp, "_id")
			tmparr := []map[string]interface{}{
				map[string]interface{}{
					"_id": id,
				},
				map[string]interface{}{"$set": tmp},
			}
			e.ResultArr = append(e.ResultArr, tmparr)
		}
	} else { //测试结果
		delete(tmp, "_id")
		if len(j.BlockPackage) > 0 { //分包详情
			bs, _ := json.Marshal(j.BlockPackage)
			tmp["epackage"] = string(bs)
		}
		tmp["result"] = result
		b := db.Mgo.Update(e.TaskInfo.TestColl, `{"_id":"`+_id+`"}`, map[string]interface{}{"$set": tmp}, true, false)
		if !b {
			log.Println(e.TaskInfo.TestColl, _id)
		}
	}
	}, func(err interface{}) {
		log.Println("AnalysisSaveResult err", err)
	})
}

func (e *ExtractTask) QualityAudit(resulttmp map[string]interface{}) {
	defer qu.Catch()
	//获取审核字段
	for _, field := range e.AuditFields {
		//1.分包
		if resulttmp["package"] != nil {
			packagedata := resulttmp["package"].(map[string]map[string]interface{})
			for _, val := range packagedata {
				if val[field] != nil {
					fv := qu.ObjToString(val[field])
					if fv != "" {
						if field == "buyer" || field == "winner" { //field为buyer和winner时特殊处理，先从Redis中查，有直接通过，没有走匹配规则
							e.RedisMatch(field, fv, val) //redis匹配
						} else { //除了buyer和winner，其他字段走规则匹配
							e.RuleMatch(field, fv, val)
						}
					}
				}
			}
		}
		//2.外围
		if resulttmp[field] != nil {
			fv := qu.ObjToString(resulttmp[field])
			if fv != "" {
				if field == "buyer" || field == "winner" { //field为buyer和winner时特殊处理，先从Redis中查，有直接通过，没有走匹配规则
					e.RedisMatch(field, fv, resulttmp) //redis匹配
				} else { //除了buyer和winner，其他字段走规则匹配
					e.RuleMatch(field, fv, resulttmp)
				}
			}
		}
	}
}

//Redis匹配
func (e *ExtractTask) RedisMatch(field, fv string, val map[string]interface{}) {
	defer qu.Catch()
	i := redis.GetInt(field, field+"_"+fv) //查找redis
	if i == 0 {                            //reids未找到，执行规则匹配
		val[field+"_isredis"] = false
		e.RuleMatch(field, fv, val) //规则匹配
	} else { //redis找到，打标识存库
		val[field+"_isredis"] = true
	}
}

//规则匹配
func (e *ExtractTask) RuleMatch(field, fieldval string, tmpMap map[string]interface{}) {
	defer qu.Catch()
	if fieldval != "" {
		SMap := e.StartMatch(field, fieldval)
		//SMap.AddKey(field+"_isaudit", false)
		for _, k := range SMap.Keys {
			tmpMap[k] = SMap.Map[k]
		}
		tmpMap[field+"_isaudit"] = false //添加字段未审核信息
	}
}

//开始规则匹配
func (e *ExtractTask) StartMatch(field, text string) *pretreated.SortMap {
	defer qu.Catch()
	SMap := pretreated.NewSortMap()
	lock.Lock()
	f := e.RecogFieldMap[field]
	lock.Unlock()
	if len(f) > 0 {
		fid := qu.BsonIdToSId(f["_id"])
		recogFieldPreRule := qu.ObjToString(f["s_recogfield_prerule"])
		textAfterRecogFieldPrerule := ju.PreFilter(text, recogFieldPreRule) //识别字段的前置过滤
		if textAfterRecogFieldPrerule != "" {
			lock.Lock()
			classMap := e.FidClassMap[fid]
			lock.Unlock()
		L:
			for _, c := range classMap { //class
				classid := qu.BsonIdToSId(c["_id"])
				classPrerule := qu.ObjToString(c["s_class_prerule"])
				savefield := qu.ObjToString(c["s_savefield"])                                   //保存字段
				textAfterClassPrerule := ju.PreFilter(textAfterRecogFieldPrerule, classPrerule) //class的前置过滤
				if textAfterClassPrerule != "" {
					lock.Lock()
					ruleMap := e.CidRuleMap[classid]
					lock.Unlock()
					for _, r := range ruleMap { //rule
						rulePrerule := qu.ObjToString(r["s_rule_prerule"])
						s_name := qu.ObjToString(r["s_name"])
						rule := r["rule"].([]interface{})
						textAfterRulePrerule := ju.PreFilter(textAfterClassPrerule, rulePrerule) //class的前置过滤
						if textAfterRulePrerule != "" {
							b, _ := ju.RecogAnalyRules(textAfterRulePrerule, rule)
							if b { //匹配到一个分类下某个规则时，不再继续匹配
								if savefield != "" { //保存字段不为空，存储代码信息
									SMap.AddKey(field+"_"+savefield, s_name)
								}
								break L
							}
						}
					}
				}
			}
		}
	}
	return SMap
}