main.go 17 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454
  1. package main
  2. import (
  3. "bytes"
  4. "encoding/json"
  5. "fmt"
  6. es "github.com/olivere/elastic/v7"
  7. "github.com/robfig/cron/v3"
  8. "go.mongodb.org/mongo-driver/bson"
  9. "go.mongodb.org/mongo-driver/bson/primitive"
  10. util "jygit.jydev.jianyu360.cn/data_processing/common_utils"
  11. "jygit.jydev.jianyu360.cn/data_processing/common_utils/elastic"
  12. "jygit.jydev.jianyu360.cn/data_processing/common_utils/mongodb"
  13. "log"
  14. "net/http"
  15. "strings"
  16. "time"
  17. )
  18. //定时任务,去统计bidding索引,排查问题、预警
  19. type T struct {
  20. Cron string
  21. Name string
  22. Min int
  23. Max int
  24. Type string
  25. Tjscope string
  26. Mgo string
  27. }
  28. // es数据最报警
  29. var (
  30. config map[string]interface{}
  31. to string
  32. api string
  33. esAddr, esAddr2, esAddr3 string
  34. esIndex, esIndex2, esIndex3 string
  35. username, password string
  36. username3, password3 string
  37. Ts = []*T{}
  38. esQ = `{"query": {"range": {"id": {"gte": "%s","lt": "%s"}}}}`
  39. esQ1 = `{"query": {"bool": {"must": [{"range": {"id": {"gte": "%s","lt": "%s"}}},{"terms": {"bidding.site": ["元博网(采购与招标网)","中国招标与采购网"]}}]}}}`
  40. )
  41. func Init() {
  42. util.ReadConfig(&config)
  43. jkmail := config["jkmail"].(map[string]interface{})
  44. to, _ = jkmail["to"].(string)
  45. api, _ = jkmail["api"].(string)
  46. //esAddr, _ = config["esAddr"].(string)
  47. esIndex, _ = config["esIndex"].(string)
  48. esAddr2, _ = config["esAddr2"].(string)
  49. esIndex2, _ = config["esIndex2"].(string) //
  50. if _, ok := config["esAddr3"]; ok {
  51. esAddr3, _ = config["esAddr3"].(string)
  52. esIndex3, _ = config["esIndex3"].(string) //
  53. username3, _ = config["username3"].(string)
  54. password3, _ = config["password3"].(string)
  55. }
  56. username, _ = config["username"].(string)
  57. password, _ = config["password"].(string)
  58. tasks, _ := config["task"].([]interface{})
  59. for _, t := range tasks {
  60. bs, _ := json.Marshal(t)
  61. var v *T
  62. json.Unmarshal(bs, &v)
  63. if v != nil {
  64. Ts = append(Ts, v)
  65. }
  66. }
  67. }
  68. func main() {
  69. Init()
  70. log.Println("start..")
  71. if len(Ts) > 0 {
  72. local, _ := time.LoadLocation("Asia/Shanghai")
  73. c := cron.New(cron.WithLocation(local), cron.WithSeconds())
  74. for _, v := range Ts {
  75. c.AddFunc(v.Cron, v.task)
  76. }
  77. c.Start()
  78. defer c.Stop()
  79. select {}
  80. }
  81. log.Println("end..")
  82. }
  83. func (t *T) task() {
  84. //初始化语句
  85. qt := strings.Split(t.Tjscope, ",")
  86. if len(qt) != 2 {
  87. return
  88. }
  89. st, et := int64(0), int64(0)
  90. now := time.Now()
  91. switch qt[1] {
  92. case "h":
  93. et = now.Unix()
  94. st = et + util.Int64All(qt[0])*3600
  95. case "d":
  96. st = util.GetDayStartSecond(util.IntAll(qt[0]))
  97. et = util.GetDayStartSecond(0)
  98. }
  99. st1 := fmt.Sprintf("%x0000000000000000", st)
  100. et1 := fmt.Sprintf("%x0000000000000000", et)
  101. eq := fmt.Sprintf(esQ, st1, et1)
  102. fmt.Println("eq", eq)
  103. eq1 := fmt.Sprintf(esQ1, st1, et1)
  104. fmt.Println("eq1", eq1)
  105. // 将时间戳转换为 time.Time 类型
  106. stt := time.Unix(st, 0)
  107. // 格式化时间为 "YYYY-MM-DD HH:MM:SS" 形式
  108. formattedTime := stt.Format("2006-01-02")
  109. rangeQuery := es.NewRangeQuery("id").Gte(st1).Lt(et1)
  110. //竞品网站统计
  111. termsQuery := es.NewTermsQuery("site", "元博网(采购与招标网)", "中国招标与采购网", "北京隆道网络科技有限公司", "友云采", "标800")
  112. countQuery := es.NewBoolQuery().
  113. Must(rangeQuery)
  114. //竞品网站总量
  115. count1Query := es.NewBoolQuery().Must(rangeQuery).Filter(termsQuery)
  116. //元博网(采购与招标网),竞品网站数量统计
  117. countComptetQuery1 := es.NewBoolQuery().Must(rangeQuery).Filter(es.NewTermQuery("site", "元博网(采购与招标网)"))
  118. countComptetQuery2 := es.NewBoolQuery().Must(rangeQuery).Filter(es.NewTermQuery("site", "中国招标与采购网"))
  119. countComptetQuery3 := es.NewBoolQuery().Must(rangeQuery).Filter(es.NewTermQuery("site", "北京隆道网络科技有限公司"))
  120. countComptetQuery4 := es.NewBoolQuery().Must(rangeQuery).Filter(es.NewTermQuery("site", "友云采"))
  121. countComptetQuery5 := es.NewBoolQuery().Must(rangeQuery).Filter(es.NewTermQuery("site", "标800"))
  122. es2 := elastic.Elastic{S_esurl: esAddr2, I_size: 2, Username: username, Password: password}
  123. es2.InitElasticSize()
  124. es3 := elastic.Elastic{S_esurl: esAddr3, I_size: 5, Username: username3, Password: password3}
  125. es3.InitElasticSize()
  126. count := int(es2.Count(esIndex2, countQuery)) //公司 es集群 数量统计
  127. count1 := int(es2.Count(esIndex, count1Query)) //竞品网站 数量统计
  128. //元博网(采购与招标网)
  129. countComptet1 := int(es2.Count(esIndex, countComptetQuery1))
  130. //中国招标与采购网
  131. countComptet2 := int(es2.Count(esIndex, countComptetQuery2))
  132. //北京隆道网络科技有限公司
  133. countComptet3 := int(es2.Count(esIndex, countComptetQuery3))
  134. //友云采
  135. countComptet4 := int(es2.Count(esIndex, countComptetQuery4))
  136. //标800
  137. countComptet5 := int(es2.Count(esIndex, countComptetQuery5))
  138. countNew := int(es3.Count(esIndex3, countQuery)) //华为云 新集群
  139. log.Println("count", count)
  140. log.Println("count1", count1)
  141. log.Println("countNew", countNew)
  142. report := ""
  143. switch t.Type {
  144. case "alert":
  145. if count < t.Min || count > t.Max || countNew < t.Min || countNew > t.Max {
  146. report = fmt.Sprintf("告警%s,最小%d,最大%d,统计结果:%d", t.Name, t.Min, t.Max, count)
  147. t.SendMail(report)
  148. wxurl := util.ObjToString(config["wxurl"])
  149. if wxurl != "" {
  150. SendBot(wxurl, report)
  151. }
  152. }
  153. case "report":
  154. //report = fmt.Sprintf("统计报告%s,【统计结果】,es库数量:%d", t.Name, count)
  155. if len(t.Mgo) > 5 {
  156. fs := strings.Split(t.Mgo, "|")
  157. fmgo := mongodb.NewMgoWithUser(fs[0], fs[3], fs[1], fs[2], 1)
  158. id1 := mongodb.StringTOBsonId(st1)
  159. id2 := mongodb.StringTOBsonId(et1)
  160. mq := bson.M{"_id": bson.M{"$gte": id1, "$lt": id2}} //一天时间内的id段
  161. fd := bson.M{"extracttype": 1, "sensitive": 1, "dataging": 1, "site": 1, "infoformat": 1, "comeintime": 1, "pici": 1, "publishtime": 1, "competehref": 1, "attach_text": 1}
  162. sess := fmgo.GetMgoConn()
  163. defer fmgo.DestoryMongoConn(sess)
  164. /**
  165. count 一天内,es 中 数据总量
  166. count1 一天内,es 中 竞品总量
  167. count2 一天内,mgo 总入库量
  168. count3 一天内,mgo 有效数据 总数
  169. count4 一天内,mgo 中 竞品数据总量
  170. count5 一天内,mgo 有效数据中,竞品的数量
  171. countNew 一天内,es3新集群 中 数据总量
  172. timeCount pici-comeintime 时差在12小时内的数据
  173. */
  174. count2, count3 := 0, 0 //
  175. count4, count5 := 0, 0 //竟品
  176. es_comeintime_totaltime := int64(0) //comeintime 和 生索引 pici 时间 差值的总和
  177. es_publishtime_totaltime := int64(0) //publishtime 和 生索引 pici 时间 差值的总和
  178. es_comeintime_avgtime := int64(0) //comeintime 和 生索引 pici 时间 差值的平均值
  179. es_publishtime_avgtime := int64(0) //publishtime 和 生索引 pici 时间 差值的平均值
  180. file_totaltime := int64(0)
  181. no_file_totaltime := int64(0)
  182. file_avgltime := int64(0)
  183. no_file_avgltime := int64(0)
  184. timeCount := 0 //12小时 统计时间差的数据量
  185. fileCount := 0
  186. noFileCount := 0
  187. //统计pici -comeintime 时间差,1、3、5、10、15、30、30+ 分钟
  188. var pc_diff1 int64
  189. var pc_diff3 int64
  190. var pc_diff5 int64
  191. var pc_diff10 int64
  192. var pc_diff15 int64
  193. var pc_diff30 int64
  194. var pc_diff301 int64
  195. var competCount1 int
  196. var competCount2 int
  197. var competCount3 int
  198. var competCount4 int
  199. var competCount5 int
  200. competeReal1 := 0
  201. competeReal2 := 0
  202. competeReal3 := 0
  203. competeReal4 := 0
  204. competeReal5 := 0
  205. query := sess.DB(fs[3]).C(fs[4]).Find(mq).Select(fd).Iter()
  206. for tmp := make(map[string]interface{}); query.Next(tmp); count2++ {
  207. if util.ObjToString(tmp["site"]) == "元博网(采购与招标网)" || util.ObjToString(tmp["site"]) == "中国招标与采购网" || util.ObjToString(tmp["site"]) == "北京隆道网络科技有限公司" || util.ObjToString(tmp["site"]) == "友云采" || util.ObjToString(tmp["site"]) == "标800" {
  208. count4++
  209. if util.ObjToString(tmp["site"]) == "元博网(采购与招标网)" {
  210. competCount1++
  211. }
  212. if util.ObjToString(tmp["site"]) == "中国招标与采购网" {
  213. competCount2++
  214. }
  215. if util.ObjToString(tmp["site"]) == "北京隆道网络科技有限公司" {
  216. competCount3++
  217. }
  218. if util.ObjToString(tmp["site"]) == "友云采" {
  219. competCount4++
  220. }
  221. if util.ObjToString(tmp["site"]) == "标800" {
  222. competCount5++
  223. }
  224. }
  225. if util.IntAll(tmp["extracttype"]) != -1 && util.ObjToString(tmp["sensitive"]) != "测试" && util.IntAll(tmp["dataging"]) != 1 && util.Float64All(tmp["infoformat"]) != 3 {
  226. count3++
  227. comeintime := util.Int64All(tmp["comeintime"])
  228. publishtime := util.Int64All(tmp["publishtime"])
  229. pici := util.Int64All(tmp["pici"])
  230. tp := time.Unix(publishtime, 0)
  231. isZero := false
  232. if (tp.Hour() == 0 && tp.Minute() == 0) && tp.Second() == 0 {
  233. isZero = true
  234. }
  235. if (comeintime-publishtime) < 12*60*60 && !isZero {
  236. if pici > 0 {
  237. diff1 := pici - comeintime
  238. diff2 := pici - publishtime
  239. if diff1 < 0 {
  240. fmt.Println("diff1", diff1, tmp["_id"])
  241. } else if diff1 <= 60 {
  242. pc_diff1++
  243. } else if diff1 <= 3*60 {
  244. pc_diff3++
  245. } else if diff1 <= 5*60 {
  246. pc_diff5++
  247. } else if diff1 <= 10*60 {
  248. pc_diff10++
  249. } else if diff1 <= 15*60 {
  250. pc_diff15++
  251. } else if diff1 <= 30*60 {
  252. pc_diff30++
  253. } else {
  254. pc_diff301++
  255. }
  256. if diff2 < 0 {
  257. fmt.Println("diff2", diff2, tmp["_id"])
  258. }
  259. es_comeintime_totaltime += diff1
  260. es_publishtime_totaltime += diff2
  261. timeCount++
  262. if _, ok := tmp["attach_text"]; ok {
  263. curtime := tmp["_id"].(primitive.ObjectID).Timestamp().Unix()
  264. diff3 := curtime - comeintime
  265. if diff3 >= 0 {
  266. file_totaltime += diff3
  267. }
  268. fileCount++
  269. } else {
  270. curtime := tmp["_id"].(primitive.ObjectID).Timestamp().Unix()
  271. diff4 := curtime - comeintime
  272. if diff4 >= 0 {
  273. no_file_totaltime += diff4
  274. }
  275. noFileCount++
  276. }
  277. }
  278. }
  279. if util.ObjToString(tmp["site"]) == "元博网(采购与招标网)" || util.ObjToString(tmp["site"]) == "中国招标与采购网" || util.ObjToString(tmp["site"]) == "北京隆道网络科技有限公司" || util.ObjToString(tmp["site"]) == "友云采" || util.ObjToString(tmp["site"]) == "标800" {
  280. count5++
  281. if util.ObjToString(tmp["site"]) == "元博网(采购与招标网)" {
  282. competeReal1++
  283. }
  284. if util.ObjToString(tmp["site"]) == "中国招标与采购网" {
  285. competeReal2++
  286. }
  287. if util.ObjToString(tmp["site"]) == "北京隆道网络科技有限公司" {
  288. competeReal3++
  289. }
  290. if util.ObjToString(tmp["site"]) == "友云采" {
  291. competeReal4++
  292. }
  293. if util.ObjToString(tmp["site"]) == "标800" {
  294. competeReal5++
  295. }
  296. }
  297. }
  298. }
  299. if timeCount > 0 {
  300. es_comeintime_avgtime = es_comeintime_totaltime / int64(timeCount)
  301. es_publishtime_avgtime = es_publishtime_totaltime / int64(timeCount)
  302. }
  303. if fileCount > 0 {
  304. file_avgltime = file_totaltime / int64(fileCount)
  305. }
  306. if noFileCount > 0 {
  307. no_file_avgltime = no_file_totaltime / int64(noFileCount)
  308. }
  309. //report += `,mgo统计:` + fmt.Sprint(count3) + `,差值:` + fmt.Sprint(count3-count) + `,mgo总入库量:` + fmt.Sprint(count2) +
  310. // `<br>竟品统计结果_es统计:` + strconv.Itoa(count1) + `,mgo统计:` + fmt.Sprint(count5) + `,差值:` + fmt.Sprint(count5-count1) + `,mgo总入库量` + fmt.Sprint(count4) +
  311. // `<br>元博网_es统计:` + strconv.Itoa(countComptet1) + `,mgo统计:` + fmt.Sprint(competeReal1) + `,差值:` + fmt.Sprint(competeReal1-countComptet1) + `,mgo总入库量` + fmt.Sprint(competCount1) +
  312. // `<br>中国招标与采购网_es统计` + strconv.Itoa(countComptet2) + `,mgo统计:` + fmt.Sprint(competeReal2) + `,差值:` + fmt.Sprint(competeReal2-countComptet2) + `,mgo总入库量` + fmt.Sprint(competCount2) +
  313. // `<br>北京隆道网络科技有限公司_es统计` + strconv.Itoa(countComptet3) + `,mgo统计:` + fmt.Sprint(competeReal3) + `,差值:` + fmt.Sprint(competeReal3-countComptet3) + `,mgo总入库量` + fmt.Sprint(competCount3) +
  314. // `<br>友云采_es统计` + strconv.Itoa(countComptet4) + `,mgo统计:` + fmt.Sprint(competeReal4) + `,差值:` + fmt.Sprint(competeReal4-countComptet4) + `,mgo总入库量` + fmt.Sprint(competCount4) +
  315. // `<br>新集群统计结果_es数量:` + strconv.Itoa(countNew) + `,mgo统计:` + fmt.Sprint(count3) + `,差值:` + fmt.Sprint(count3-countNew)
  316. //
  317. var reportBuilder strings.Builder
  318. // 开始表格
  319. reportBuilder.WriteString("<table>")
  320. reportBuilder.WriteString(formattedTime)
  321. reportBuilder.WriteString("<tr><th>统计项</th><th>es统计</th><th>mgo统计</th><th>差值</th><th>mgo总入库量</th></tr>")
  322. addTableRow(&reportBuilder, "阿里云es集群", count, count3, count3-count, count2)
  323. addTableRow(&reportBuilder, "竟品统计结果", count1, count5, count5-count1, count4)
  324. addTableRow(&reportBuilder, "元博网", countComptet1, competeReal1, competeReal1-countComptet1, competCount1)
  325. addTableRow(&reportBuilder, "中国招标与采购网", countComptet2, competeReal2, competeReal2-countComptet2, competCount2)
  326. addTableRow(&reportBuilder, "北京隆道网络科技有限公司", countComptet3, competeReal3, competeReal3-countComptet3, competCount3)
  327. addTableRow(&reportBuilder, "友云采", countComptet4, competeReal4, competeReal4-countComptet4, competCount4)
  328. addTableRow(&reportBuilder, "标800", countComptet5, competeReal5, competeReal5-countComptet5, competCount5)
  329. addTableRow(&reportBuilder, "华为云es集群", countNew, count3, count3-countNew, count2)
  330. // 结束表格
  331. reportBuilder.WriteString("</table>")
  332. // 获取最终的报告字符串
  333. report = reportBuilder.String()
  334. //存入数据库
  335. yesterday := now.AddDate(0, 0, -1)
  336. insert := map[string]interface{}{
  337. "es_count": count, //
  338. "es3_count": countNew, //
  339. "mgo_count": count3,
  340. "es_mgo_diff": count3 - count,
  341. "mgo_total": count2,
  342. "competitor_es_count": count1, //竞品网站es 数量
  343. "competitor_mgo_count": count5,
  344. "competitor_diff": count5 - count1,
  345. "competitor_mgo_total": count4,
  346. "date": yesterday.Format("2006-01-02"),
  347. "es_comeintime_totaltime": es_comeintime_totaltime,
  348. "es_publishtime_totaltime": es_publishtime_totaltime,
  349. "es_comeintime_avgtime": es_comeintime_avgtime,
  350. "es_publishtime_avgtime": es_publishtime_avgtime,
  351. "file_avgltime": file_avgltime,
  352. "no_file_avgltime": no_file_avgltime,
  353. "file_totaltime": file_totaltime,
  354. "no_file_totaltime": no_file_totaltime,
  355. "file_count": fileCount,
  356. "no_file_count": noFileCount,
  357. "pc_diff1": pc_diff1,
  358. "pc_diff3": pc_diff3,
  359. "pc_diff5": pc_diff5,
  360. "pc_diff10": pc_diff10,
  361. "pc_diff15": pc_diff15,
  362. "pc_diff30": pc_diff30,
  363. "pc_diff301": pc_diff301,
  364. "timeCount": timeCount,
  365. }
  366. fmgo.Save("bidding_ribao", insert)
  367. }
  368. t.SendMail(report)
  369. //t.SendMail("【竟品统计结果】")
  370. //fmt.Println(report)
  371. // 发送企业微信
  372. wxurl := util.ObjToString(config["wxurl"])
  373. if wxurl != "" {
  374. SendBot(wxurl, report)
  375. }
  376. }
  377. log.Println("task over:", t.Name, eq, count)
  378. }
  379. func (t *T) SendMail(report string) {
  380. url := fmt.Sprintf("%s?to=%s&title=%s&body=%s", api, to, t.Name, report)
  381. fmt.Println("url=>", url)
  382. res, err := http.Get(url)
  383. if err != nil {
  384. fmt.Println("SendMail err ", err)
  385. } else {
  386. fmt.Println("SendMail res ", res)
  387. }
  388. }
  389. func addTableRow(builder *strings.Builder, itemName string, esCount, mgoCount, diff, mgoTotal int) {
  390. builder.WriteString(fmt.Sprintf(`<tr><th>%s</th><th>%d</th><th>%d</th><th>%d</th><th>%d</th></tr>`, itemName, esCount, mgoCount, diff, mgoTotal))
  391. }
  392. // SendBot 发送企业微信
  393. func SendBot(webhookURL, msg string) (b bool) {
  394. // 构造请求体
  395. payload := map[string]interface{}{
  396. "msgtype": "text",
  397. "text": map[string]string{
  398. "content": msg,
  399. },
  400. }
  401. // 转换为 JSON 字符串
  402. payloadBytes, err := json.Marshal(payload)
  403. if err != nil {
  404. log.Println("Error:", err)
  405. return
  406. }
  407. // 发送 POST 请求
  408. resp, err := http.Post(webhookURL, "application/json", bytes.NewReader(payloadBytes))
  409. if err != nil {
  410. log.Println("Error:", err)
  411. return
  412. }
  413. defer resp.Body.Close()
  414. b = true
  415. //打印响应结果
  416. log.Println("send bot Status:", resp.Status)
  417. return
  418. }