ent_dispose/qyxy_std_new/task.go

package main

import (
	"context"
	"fmt"
	util "jygit.jydev.jianyu360.cn/data_processing/common_utils"
	//"github.com/cron"
	"github.com/wangbin/jiebago"
	"go.mongodb.org/mongo-driver/bson"
	"go.mongodb.org/mongo-driver/mongo/options"
	"log"
	"math/rand"
	"regexp"
	"strconv"
	"strings"
	"time"
)

var (
	AreaFiled = []string{"credit_no", "company_code"}
	WordsArr  = []string{"研发", "研制", "开发", "生产", "制造", "制作", "加工", "种植"}

	seg        jiebago.Segmenter
	regPre, _  = regexp.Compile(`^(.+[省|市|区|县|州])?(.+)`)
	seed       = 188
	startSeoId int64
)

// 企业基本信息
var company_base = []string{"company_name", "company_code", "credit_no", "org_code", "legal_person", "company_status",
	"authority", "establish_date", "issue_date", "operation_startdate", "operation_enddate", "capital", "company_type",
	"company_status", "company_address", "business_scope", "cancel_date", "cancel_reason", "revoke_date", "revoke_reason",
	"legal_person_type", "real_capital", "en_name", "list_code", "tax_code", "use_flag",
}

var province_map = map[string]string{
	"BJ": "北京", "TJ": "天津", "SH": "上海", "CQ": "重庆", "HB": "河北", "SX": "山西", "NMG": "内蒙古", "LN": "辽宁", "JL": "吉林",
	"HLJ": "黑龙江", "JS": "江苏", "ZJ": "浙江", "AH": "安徽", "FJ": "福建", "JX": "江西", "SD": "山东", "HEN": "河南", "HUB": "湖北",
	"HUN": "湖南", "GD": "广东", "GX": "广西", "HAIN": "海南", "SC": "四川", "GZ": "贵州", "YN": "云南", "XZ": "西藏", "SAX": "陕西",
	"GS": "甘肃", "QH": "青海", "NX": "宁夏", "XJ": "新疆",
}

// company_shortname 企业简称
func getStName(name string) string {
	regnames := regPre.FindStringSubmatch(name)
	lenth := len(regnames)
	if lenth < 1 {
		return ""
	}
	newstr := regnames[lenth-1]

	ch := seg.Cut(newstr, false)
	val := []string{}
	for word := range ch {
		val = append(val, word)
	}
	name2 := ""
	for _, v := range val {
		name2 = name2 + v
		if len([]rune(name2)) >= 4 {
			break
		}
	}
	return name2
}

func DealMemberNo(memberNo string) (no int) {
	memberNo = strings.Trim(memberNo, " ")
	if memberNo == "" {
		return no
	}
	if memberNo == "企业选择不公示" {
		return no
	}
	if strings.HasSuffix(memberNo, "人") {
		cou := strings.Split(memberNo, "人")[0]
		count, _ := strconv.Atoi(cou)
		return count
	}

	no, _ = strconv.Atoi(memberNo)
	return no
}

// updateStd 根据qyxy_std 表，更新 nseo_id
func updateStd() {
	id := int64(GF.Env.Autoid) //起始 autoid
	//生成seoid
	//31395235740
	startSeoId = int64(GF.Env.Seoid) //起始 nseo_id

	rand.Seed(time.Now().UnixNano())

	count := int64(0)
	//var wg sync.WaitGroup
	sess := MongoTool.GetMgoConn()
	defer MongoTool.DestoryMongoConn(sess)

	ctx, _ := context.WithTimeout(context.Background(), 99999*time.Hour)
	coll := sess.M.C.Database("mixdata").Collection("qyxy_std")
	find := options.Find().SetBatchSize(200).SetSort(bson.D{bson.E{"autoid", 1}}).SetProjection(bson.M{"_id": 1, "autoid": 1, "establish_date": 1, "create_time_msql": 1, "nseo_id": 1})
	cur, err := coll.Find(ctx, bson.M{"autoid": bson.M{"$gt": id}}, find)

	if err != nil {
		log.Println("mgo find err", err.Error())
		return
	}

	for tmp := make(map[string]interface{}); cur.Next(ctx); {
		count++
		if cur != nil {
			cur.Decode(&tmp)
			id := util.Int64All(tmp["autoid"])
			//存在就不处理
			if nseoid, ok := tmp["nseo_id"]; ok && nseoid != "" {
				continue
			}

			ed, _ := tmp["establish_date"].(string)
			pre := ""
			if len(ed) == 10 {
				pre = ed[2:4] + ed[5:7] + ed[8:10]
			} else {
				cd, _ := tmp["create_time_msql"].(string)
				if len(cd) > 9 {
					pre = cd[2:4] + cd[5:7] + cd[8:10]
				}
			}

			startSeoId += int64(rand.Intn(seed) + 2)
			nseo_id := fmt.Sprintf("%s%d", pre, startSeoId)
			update := make(map[string]interface{})
			update["$set"] = bson.M{
				"nseo_id": nseo_id,
			}
			where := map[string]interface{}{
				"_id": tmp["_id"],
			}

			MongoTool.Update("qyxy_std", where, update, true, false)

			if count%10000 == 0 {
				log.Println("current,autoid,startSeoId,nseo_id", count, id, startSeoId, nseo_id)
			}

			tmp = make(map[string]interface{})
		} else {
			cur.Close(ctx)
			break
		}

	}

	log.Println("over ----;", "seo_id:", startSeoId)

}