main.go 23 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662
  1. package main
  2. import (
  3. "bytes"
  4. "context"
  5. "encoding/json"
  6. "fmt"
  7. es "github.com/olivere/elastic/v7"
  8. "github.com/robfig/cron/v3"
  9. "go.mongodb.org/mongo-driver/bson"
  10. "go.mongodb.org/mongo-driver/bson/primitive"
  11. "io"
  12. util "jygit.jydev.jianyu360.cn/data_processing/common_utils"
  13. "jygit.jydev.jianyu360.cn/data_processing/common_utils/elastic"
  14. "jygit.jydev.jianyu360.cn/data_processing/common_utils/mongodb"
  15. "log"
  16. "net/http"
  17. "strings"
  18. "time"
  19. )
  20. //定时任务,去统计bidding索引,排查问题、预警
  21. type T struct {
  22. Cron string
  23. Name string
  24. Min int
  25. Max int
  26. Type string
  27. Tjscope string
  28. Mgo string
  29. }
  30. // es数据最报警
  31. var (
  32. config map[string]interface{}
  33. to string
  34. api string
  35. esAddr, esAddr2, esAddr3 string
  36. esIndex, esIndex2, esIndex3 string
  37. username, password string
  38. username3, password3 string
  39. Ts = []*T{}
  40. esQ = `{"query": {"range": {"id": {"gte": "%s","lt": "%s"}}}}`
  41. 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"]}}]}}}`
  42. )
  43. func Init() {
  44. util.ReadConfig(&config)
  45. jkmail := config["jkmail"].(map[string]interface{})
  46. to, _ = jkmail["to"].(string)
  47. api, _ = jkmail["api"].(string)
  48. //esAddr, _ = config["esAddr"].(string)
  49. esIndex, _ = config["esIndex"].(string)
  50. esAddr2, _ = config["esAddr2"].(string)
  51. esIndex2, _ = config["esIndex2"].(string) //
  52. if _, ok := config["esAddr3"]; ok {
  53. esAddr3, _ = config["esAddr3"].(string)
  54. esIndex3, _ = config["esIndex3"].(string) //
  55. username3, _ = config["username3"].(string)
  56. password3, _ = config["password3"].(string)
  57. }
  58. username, _ = config["username"].(string)
  59. password, _ = config["password"].(string)
  60. tasks, _ := config["task"].([]interface{})
  61. for _, t := range tasks {
  62. bs, _ := json.Marshal(t)
  63. var v *T
  64. json.Unmarshal(bs, &v)
  65. if v != nil {
  66. Ts = append(Ts, v)
  67. }
  68. }
  69. }
  70. func main() {
  71. Init()
  72. log.Println("start..")
  73. if len(Ts) > 0 {
  74. local, _ := time.LoadLocation("Asia/Shanghai")
  75. c := cron.New(cron.WithLocation(local), cron.WithSeconds())
  76. for _, v := range Ts {
  77. _, err := c.AddFunc(v.Cron, v.task)
  78. if err != nil {
  79. panic(err)
  80. }
  81. }
  82. c.Start()
  83. defer c.Stop()
  84. select {}
  85. }
  86. log.Println("end..")
  87. }
  88. func (t *T) task() {
  89. //初始化语句
  90. qt := strings.Split(t.Tjscope, ",")
  91. if len(qt) != 2 {
  92. return
  93. }
  94. st, et := int64(0), int64(0)
  95. now := time.Now()
  96. switch qt[1] {
  97. case "m":
  98. et = now.Unix()
  99. st = et + util.Int64All(qt[0])*60
  100. case "h":
  101. et = now.Unix()
  102. st = et + util.Int64All(qt[0])*3600
  103. case "d":
  104. st = util.GetDayStartSecond(util.IntAll(qt[0]))
  105. et = util.GetDayStartSecond(0)
  106. }
  107. st1 := fmt.Sprintf("%x0000000000000000", st)
  108. et1 := fmt.Sprintf("%x0000000000000000", et)
  109. eq := fmt.Sprintf(esQ, st1, et1)
  110. fmt.Println("eq", eq)
  111. eq1 := fmt.Sprintf(esQ1, st1, et1)
  112. fmt.Println("eq1", eq1)
  113. // 将时间戳转换为 time.Time 类型
  114. stt := time.Unix(st, 0)
  115. // 格式化时间为 "YYYY-MM-DD HH:MM:SS" 形式
  116. formattedTime := stt.Format("2006-01-02")
  117. rangeQuery := es.NewRangeQuery("id").Gte(st1).Lt(et1)
  118. //竞品网站统计
  119. 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")
  120. countQuery := es.NewBoolQuery().
  121. Must(rangeQuery)
  122. //竞品网站总量
  123. count1Query := es.NewBoolQuery().Must(rangeQuery).Filter(termsQuery)
  124. //元博网(采购与招标网),竞品网站数量统计
  125. //countComptetQuery1 := es.NewBoolQuery().Must(rangeQuery).Filter(es.NewTermQuery("site", "元博网(采购与招标网)"))
  126. //countComptetQuery2 := es.NewBoolQuery().Must(rangeQuery).Filter(es.NewTermQuery("site", "中国招标与采购网"))
  127. //countComptetQuery3 := es.NewBoolQuery().Must(rangeQuery).Filter(es.NewTermQuery("site", "北京隆道网络科技有限公司"))
  128. //countComptetQuery4 := es.NewBoolQuery().Must(rangeQuery).Filter(es.NewTermQuery("site", "友云采"))
  129. //countComptetQuery5 := es.NewBoolQuery().Must(rangeQuery).Filter(es.NewTermQuery("site", "标800"))
  130. es2 := elastic.Elastic{S_esurl: esAddr2, I_size: 2, Username: username, Password: password}
  131. es2.InitElasticSize()
  132. es3 := elastic.Elastic{S_esurl: esAddr3, I_size: 5, Username: username3, Password: password3}
  133. es3.InitElasticSize()
  134. count := int(es2.Count(esIndex2, countQuery)) //公司 es集群 数量统计
  135. count1 := int(es2.Count(esIndex, count1Query)) //竞品网站 数量统计
  136. ////元博网(采购与招标网)
  137. //countComptet1 := int(es2.Count(esIndex, countComptetQuery1))
  138. ////中国招标与采购网
  139. //countComptet2 := int(es2.Count(esIndex, countComptetQuery2))
  140. ////北京隆道网络科技有限公司
  141. //countComptet3 := int(es2.Count(esIndex, countComptetQuery3))
  142. ////友云采
  143. //countComptet4 := int(es2.Count(esIndex, countComptetQuery4))
  144. ////标800
  145. //countComptet5 := int(es2.Count(esIndex, countComptetQuery5))
  146. countNew := int(es3.Count(esIndex3, countQuery)) //华为云 新集群
  147. countNewTmp := int(es3.Count("bidding_temp", countQuery)) //华为云 新集群
  148. log.Println("count", count)
  149. log.Println("count1", count1)
  150. log.Println("countNew", countNew)
  151. report := ""
  152. switch t.Type {
  153. case "alert":
  154. if count < t.Min || count > t.Max || countNew < t.Min || countNew > t.Max {
  155. report = fmt.Sprintf("告警%s,最小%d,最大%d,统计结果:%d", t.Name, t.Min, t.Max, count)
  156. t.SendMail(report)
  157. wxurl := util.ObjToString(config["wxurl"])
  158. if wxurl != "" {
  159. SendBot(wxurl, report)
  160. }
  161. }
  162. case "report":
  163. //report = fmt.Sprintf("统计报告%s,【统计结果】,es库数量:%d", t.Name, count)
  164. if len(t.Mgo) > 5 {
  165. fs := strings.Split(t.Mgo, "|")
  166. fmgo := mongodb.NewMgoWithUser(fs[0], fs[3], fs[1], fs[2], 1)
  167. id1 := mongodb.StringTOBsonId(st1)
  168. id2 := mongodb.StringTOBsonId(et1)
  169. mq := bson.M{"_id": bson.M{"$gte": id1, "$lt": id2}} //一天时间内的id段
  170. fd := bson.M{"extracttype": 1, "sensitive": 1, "dataging": 1, "site": 1, "infoformat": 1, "comeintime": 1, "pici": 1, "publishtime": 1, "competehref": 1, "attach_text": 1}
  171. sess := fmgo.GetMgoConn()
  172. defer fmgo.DestoryMongoConn(sess)
  173. /**
  174. count 一天内,es 中 数据总量
  175. count1 一天内,es 中 竞品总量
  176. count2 一天内,mgo 总入库量
  177. count3 一天内,mgo 有效数据 总数
  178. count4 一天内,mgo 中 竞品数据总量
  179. count5 一天内,mgo 有效数据中,竞品的数量
  180. countNew 一天内,es3新集群 中 数据总量
  181. timeCount pici-comeintime 时差在12小时内的数据
  182. */
  183. count2, count3 := 0, 0 //
  184. count4, count5 := 0, 0 //竟品
  185. es_comeintime_totaltime := int64(0) //comeintime 和 生索引 pici 时间 差值的总和
  186. es_publishtime_totaltime := int64(0) //publishtime 和 生索引 pici 时间 差值的总和
  187. es_comeintime_avgtime := int64(0) //comeintime 和 生索引 pici 时间 差值的平均值
  188. es_publishtime_avgtime := int64(0) //publishtime 和 生索引 pici 时间 差值的平均值
  189. file_totaltime := int64(0)
  190. no_file_totaltime := int64(0)
  191. file_avgltime := int64(0)
  192. no_file_avgltime := int64(0)
  193. timeCount := 0 //12小时 统计时间差的数据量
  194. fileCount := 0
  195. noFileCount := 0
  196. //统计pici -comeintime 时间差,1、3、5、10、15、30、30+ 分钟
  197. var pc_diff1 int64
  198. var pc_diff3 int64
  199. var pc_diff5 int64
  200. var pc_diff10 int64
  201. var pc_diff15 int64
  202. var pc_diff30 int64
  203. var pc_diff301 int64
  204. var competCount1 int
  205. var competCount2 int
  206. var competCount3 int
  207. var competCount4 int
  208. var competCount5 int
  209. competeReal1 := 0
  210. competeReal2 := 0
  211. competeReal3 := 0
  212. competeReal4 := 0
  213. competeReal5 := 0
  214. query := sess.DB(fs[3]).C(fs[4]).Find(mq).Select(fd).Iter()
  215. for tmp := make(map[string]interface{}); query.Next(tmp); count2++ {
  216. 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" {
  217. count4++
  218. if util.ObjToString(tmp["site"]) == "元博网(采购与招标网)" {
  219. competCount1++
  220. }
  221. if util.ObjToString(tmp["site"]) == "中国招标与采购网" {
  222. competCount2++
  223. }
  224. if util.ObjToString(tmp["site"]) == "北京隆道网络科技有限公司" {
  225. competCount3++
  226. }
  227. if util.ObjToString(tmp["site"]) == "友云采" {
  228. competCount4++
  229. }
  230. if util.ObjToString(tmp["site"]) == "标800" {
  231. competCount5++
  232. }
  233. }
  234. if util.IntAll(tmp["extracttype"]) != -1 && util.ObjToString(tmp["sensitive"]) != "测试" && util.IntAll(tmp["dataging"]) != 1 && util.Float64All(tmp["infoformat"]) != 3 {
  235. count3++
  236. comeintime := util.Int64All(tmp["comeintime"])
  237. publishtime := util.Int64All(tmp["publishtime"])
  238. pici := util.Int64All(tmp["pici"])
  239. tp := time.Unix(publishtime, 0)
  240. isZero := false
  241. if (tp.Hour() == 0 && tp.Minute() == 0) && tp.Second() == 0 {
  242. isZero = true
  243. }
  244. if (comeintime-publishtime) < 12*60*60 && !isZero {
  245. if pici > 0 {
  246. diff1 := pici - comeintime
  247. diff2 := pici - publishtime
  248. if diff1 < 0 {
  249. fmt.Println("diff1", diff1, tmp["_id"])
  250. } else if diff1 <= 60 {
  251. pc_diff1++
  252. } else if diff1 <= 3*60 {
  253. pc_diff3++
  254. } else if diff1 <= 5*60 {
  255. pc_diff5++
  256. } else if diff1 <= 10*60 {
  257. pc_diff10++
  258. } else if diff1 <= 15*60 {
  259. pc_diff15++
  260. } else if diff1 <= 30*60 {
  261. pc_diff30++
  262. } else {
  263. pc_diff301++
  264. }
  265. if diff2 < 0 {
  266. fmt.Println("diff2", diff2, tmp["_id"])
  267. }
  268. es_comeintime_totaltime += diff1
  269. es_publishtime_totaltime += diff2
  270. timeCount++
  271. if _, ok := tmp["attach_text"]; ok {
  272. curtime := tmp["_id"].(primitive.ObjectID).Timestamp().Unix()
  273. diff3 := curtime - comeintime
  274. if diff3 >= 0 {
  275. file_totaltime += diff3
  276. }
  277. fileCount++
  278. } else {
  279. curtime := tmp["_id"].(primitive.ObjectID).Timestamp().Unix()
  280. diff4 := curtime - comeintime
  281. if diff4 >= 0 {
  282. no_file_totaltime += diff4
  283. }
  284. noFileCount++
  285. }
  286. }
  287. }
  288. 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" {
  289. count5++
  290. if util.ObjToString(tmp["site"]) == "元博网(采购与招标网)" {
  291. competeReal1++
  292. }
  293. if util.ObjToString(tmp["site"]) == "中国招标与采购网" {
  294. competeReal2++
  295. }
  296. if util.ObjToString(tmp["site"]) == "北京隆道网络科技有限公司" {
  297. competeReal3++
  298. }
  299. if util.ObjToString(tmp["site"]) == "友云采" {
  300. competeReal4++
  301. }
  302. if util.ObjToString(tmp["site"]) == "标800" {
  303. competeReal5++
  304. }
  305. }
  306. }
  307. }
  308. if timeCount > 0 {
  309. es_comeintime_avgtime = es_comeintime_totaltime / int64(timeCount)
  310. es_publishtime_avgtime = es_publishtime_totaltime / int64(timeCount)
  311. }
  312. if fileCount > 0 {
  313. file_avgltime = file_totaltime / int64(fileCount)
  314. }
  315. if noFileCount > 0 {
  316. no_file_avgltime = no_file_totaltime / int64(noFileCount)
  317. }
  318. //report += `,mgo统计:` + fmt.Sprint(count3) + `,差值:` + fmt.Sprint(count3-count) + `,mgo总入库量:` + fmt.Sprint(count2) +
  319. // `<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) +
  320. // `<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) +
  321. // `<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) +
  322. // `<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) +
  323. // `<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) +
  324. // `<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)
  325. //html 格式
  326. //var reportBuilder strings.Builder
  327. //// 开始表格
  328. //reportBuilder.WriteString("<table>")
  329. //reportBuilder.WriteString(formattedTime)
  330. //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>")
  331. //
  332. //addTableRow(&reportBuilder, "阿里云es集群", count, count3, count3-count, count2)
  333. //addTableRow(&reportBuilder, "竟品统计结果", count1, count5, count5-count1, count4)
  334. //addTableRow(&reportBuilder, "元博网", countComptet1, competeReal1, competeReal1-countComptet1, competCount1)
  335. //addTableRow(&reportBuilder, "中国招标与采购网", countComptet2, competeReal2, competeReal2-countComptet2, competCount2)
  336. //addTableRow(&reportBuilder, "北京隆道网络科技有限公司", countComptet3, competeReal3, competeReal3-countComptet3, competCount3)
  337. //addTableRow(&reportBuilder, "友云采", countComptet4, competeReal4, competeReal4-countComptet4, competCount4)
  338. //addTableRow(&reportBuilder, "标800", countComptet5, competeReal5, competeReal5-countComptet5, competCount5)
  339. //addTableRow(&reportBuilder, "华为云es集群", countNew, count3, count3-countNew, count2)
  340. //// 结束表格
  341. //reportBuilder.WriteString("</table>")
  342. //// 获取最终的报告字符串
  343. //report = reportBuilder.String()
  344. var reportBuilder strings.Builder
  345. // 添加时间
  346. reportBuilder.WriteString(formattedTime + "\n")
  347. // Markdown 表头
  348. reportBuilder.WriteString("| 统计项 | es统计 | mgo统计 | 差值 | mgo总入库量 |\n")
  349. reportBuilder.WriteString("|--------|--------|---------|------|--------------|\n")
  350. addMarkdownRow(&reportBuilder, "阿里云es集群", count, count3, count3-count, count2)
  351. //addMarkdownRow(&reportBuilder, "竟品统计结果", count1, count5, count5-count1, count4)
  352. //addMarkdownRow(&reportBuilder, "元博网", countComptet1, competeReal1, competeReal1-countComptet1, competCount1)
  353. //addMarkdownRow(&reportBuilder, "中国招标与采购网", countComptet2, competeReal2, competeReal2-countComptet2, competCount2)
  354. //addMarkdownRow(&reportBuilder, "北京隆道网络科技有限公司", countComptet3, competeReal3, competeReal3-countComptet3, competCount3)
  355. //addMarkdownRow(&reportBuilder, "友云采", countComptet4, competeReal4, competeReal4-countComptet4, competCount4)
  356. //addMarkdownRow(&reportBuilder, "标800", countComptet5, competeReal5, competeReal5-countComptet5, competCount5)
  357. addMarkdownRow(&reportBuilder, "华为云es集群", countNew, count3, count3-countNew, count2)
  358. addMarkdownRow(&reportBuilder, "华为云es集群 bidding_temp", countNewTmp, count3, count3-countNewTmp, count2)
  359. report = reportBuilder.String()
  360. //存入数据库
  361. yesterday := now.AddDate(0, 0, -1)
  362. insert := map[string]interface{}{
  363. "es_count": count, //
  364. "es3_count": countNew, //
  365. "mgo_count": count3,
  366. "es_mgo_diff": count3 - count,
  367. "mgo_total": count2,
  368. "competitor_es_count": count1, //竞品网站es 数量
  369. "competitor_mgo_count": count5,
  370. "competitor_diff": count5 - count1,
  371. "competitor_mgo_total": count4,
  372. "date": yesterday.Format("2006-01-02"),
  373. "es_comeintime_totaltime": es_comeintime_totaltime,
  374. "es_publishtime_totaltime": es_publishtime_totaltime,
  375. "es_comeintime_avgtime": es_comeintime_avgtime,
  376. "es_publishtime_avgtime": es_publishtime_avgtime,
  377. "file_avgltime": file_avgltime,
  378. "no_file_avgltime": no_file_avgltime,
  379. "file_totaltime": file_totaltime,
  380. "no_file_totaltime": no_file_totaltime,
  381. "file_count": fileCount,
  382. "no_file_count": noFileCount,
  383. "pc_diff1": pc_diff1,
  384. "pc_diff3": pc_diff3,
  385. "pc_diff5": pc_diff5,
  386. "pc_diff10": pc_diff10,
  387. "pc_diff15": pc_diff15,
  388. "pc_diff30": pc_diff30,
  389. "pc_diff301": pc_diff301,
  390. "timeCount": timeCount,
  391. }
  392. fmgo.Save("bidding_ribao", insert)
  393. }
  394. t.SendMail(report)
  395. //t.SendMail("【竟品统计结果】")
  396. //fmt.Println(report)
  397. // 发送企业微信
  398. wxurl := util.ObjToString(config["wxurl"])
  399. if wxurl != "" {
  400. SendBot(wxurl, report)
  401. }
  402. //9008 bidding 数据和 9005 bidding 数据量不一致
  403. if countNew != count {
  404. client1, _ := es.NewClient(
  405. es.SetURL(esAddr2),
  406. es.SetBasicAuth(username, password),
  407. es.SetSniff(false),
  408. )
  409. client2, _ := es.NewClient(
  410. es.SetURL(esAddr3),
  411. es.SetBasicAuth(username3, password3),
  412. es.SetSniff(false),
  413. )
  414. // 9008 比 9005 的索引数据量多
  415. if count > countNew {
  416. ids1, err := fetchIDs(client1, "bidding", rangeQuery)
  417. if err != nil {
  418. log.Fatal(err)
  419. }
  420. fmt.Printf("集群 9008 bidding 总数: %d\n", len(ids1))
  421. ids2, err := fetchIDs(client2, "bidding", rangeQuery)
  422. if err != nil {
  423. log.Fatal(err)
  424. }
  425. fmt.Printf("集群9005 bidding 总数: %d\n", len(ids2))
  426. fmt.Println("差集 (集群9008 - 集群9005):")
  427. diffIDS := make([]string, 0)
  428. for id := range ids1 {
  429. if _, ok := ids2[id]; !ok {
  430. fmt.Println(id)
  431. diffIDS = append(diffIDS, id)
  432. }
  433. }
  434. text := fmt.Sprintf("集群9008 - 集群9005 bidding 的差集数据是:\n %v", diffIDS)
  435. SendBot(wxurl, text)
  436. } else {
  437. ids1, err := fetchIDs(client1, "bidding", rangeQuery)
  438. if err != nil {
  439. log.Fatal(err)
  440. }
  441. fmt.Printf("集群 9008 bidding 总数: %d\n", len(ids1))
  442. ids2, err := fetchIDs(client2, "bidding", rangeQuery)
  443. if err != nil {
  444. log.Fatal(err)
  445. }
  446. fmt.Printf("集群9005 bidding 总数: %d\n", len(ids2))
  447. fmt.Println("差集 (集群9005 - 集群9008):")
  448. diffIDS := make([]string, 0)
  449. for id := range ids2 {
  450. if _, ok := ids1[id]; !ok {
  451. fmt.Println(id)
  452. }
  453. }
  454. text := fmt.Sprintf("集群9005 - 集群9008 bidding 的差集数据是:\n %v", diffIDS)
  455. SendBot(wxurl, text)
  456. }
  457. }
  458. if countNewTmp != count {
  459. client1, _ := es.NewClient(
  460. es.SetURL(esAddr2),
  461. es.SetBasicAuth(username, password),
  462. es.SetSniff(false),
  463. )
  464. client2, _ := es.NewClient(
  465. es.SetURL(esAddr3),
  466. es.SetBasicAuth(username3, password3),
  467. es.SetSniff(false),
  468. )
  469. if count > countNewTmp {
  470. ids1, err := fetchIDs(client1, "bidding", rangeQuery)
  471. if err != nil {
  472. log.Fatal(err)
  473. }
  474. fmt.Printf("集群 9008 bidding 总数: %d\n", len(ids1))
  475. ids2, err := fetchIDs(client2, "bidding_temp", rangeQuery)
  476. if err != nil {
  477. log.Fatal(err)
  478. }
  479. fmt.Printf("集群9005 bidding_temp 总数: %d\n", len(ids2))
  480. fmt.Println("差集 (集群9008 bidding - 集群9005 bidding_temp ):")
  481. diffIDS := make([]string, 0)
  482. for id := range ids1 {
  483. if _, ok := ids2[id]; !ok {
  484. fmt.Println(id)
  485. diffIDS = append(diffIDS, id)
  486. }
  487. }
  488. text := fmt.Sprintf("集群9008 bidding - 集群9005 bidding_temp 的差集数据是:\n %v", diffIDS)
  489. SendBot(wxurl, text)
  490. } else {
  491. ids1, err := fetchIDs(client1, "bidding", rangeQuery)
  492. if err != nil {
  493. log.Fatal(err)
  494. }
  495. fmt.Printf("集群 9008 bidding 总数: %d\n", len(ids1))
  496. ids2, err := fetchIDs(client2, "bidding_temp", rangeQuery)
  497. if err != nil {
  498. log.Fatal(err)
  499. }
  500. fmt.Printf("集群9005 bidding_temp 总数: %d\n", len(ids2))
  501. fmt.Println("差集 (集群9005 biding_temp - 集群9008 bidding ):")
  502. diffIDS := make([]string, 0)
  503. for id := range ids2 {
  504. if _, ok := ids1[id]; !ok {
  505. fmt.Println(id)
  506. }
  507. }
  508. text := fmt.Sprintf("集群9005 biding_temp - 集群9008 bidding的差集数据是:\n %v", diffIDS)
  509. SendBot(wxurl, text)
  510. }
  511. }
  512. }
  513. log.Println("task over:", t.Name, eq, count)
  514. }
  515. func (t *T) SendMail(report string) {
  516. url := fmt.Sprintf("%s?to=%s&title=%s&body=%s", api, to, t.Name, report)
  517. fmt.Println("url=>", url)
  518. res, err := http.Get(url)
  519. if err != nil {
  520. fmt.Println("SendMail err ", err)
  521. } else {
  522. fmt.Println("SendMail res ", res)
  523. }
  524. }
  525. func addTableRow(builder *strings.Builder, itemName string, esCount, mgoCount, diff, mgoTotal int) {
  526. 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))
  527. }
  528. func addMarkdownRow(builder *strings.Builder, itemName string, esCount, mgoCount, diff, mgoTotal int) {
  529. builder.WriteString(fmt.Sprintf("| %s | %d | %d | %d | %d |\n", itemName, esCount, mgoCount, diff, mgoTotal))
  530. }
  531. // SendBot 发送企业微信
  532. func SendBot(webhookURL, msg string) (b bool) {
  533. // 构造请求体
  534. payload := map[string]interface{}{
  535. "msgtype": "text",
  536. "text": map[string]string{
  537. "content": msg,
  538. },
  539. }
  540. // 转换为 JSON 字符串
  541. payloadBytes, err := json.Marshal(payload)
  542. if err != nil {
  543. log.Println("Error:", err)
  544. return
  545. }
  546. // 发送 POST 请求
  547. resp, err := http.Post(webhookURL, "application/json", bytes.NewReader(payloadBytes))
  548. if err != nil {
  549. log.Println("Error:", err)
  550. return
  551. }
  552. defer resp.Body.Close()
  553. b = true
  554. //打印响应结果
  555. log.Println("send bot Status:", resp.Status)
  556. return
  557. }
  558. func fetchIDs(client *es.Client, index string, query es.Query) (map[string]struct{}, error) {
  559. ctx := context.Background()
  560. ids := make(map[string]struct{})
  561. scrollID := ""
  562. scroll := "10m"
  563. searchSource := es.NewSearchSource().
  564. Query(query).
  565. Size(10000).
  566. Sort("_doc", true)
  567. searchService := client.Scroll(index).
  568. Size(10000).
  569. Scroll(scroll).
  570. SearchSource(searchSource)
  571. res, err := searchService.Do(ctx)
  572. if err != nil {
  573. if err == io.EOF {
  574. fmt.Println("没有数据")
  575. return ids, nil
  576. }
  577. return nil, fmt.Errorf("初始化滚动搜索失败: %w", err)
  578. }
  579. fmt.Println("命中总数:", res.TotalHits())
  580. total := 0
  581. for len(res.Hits.Hits) > 0 {
  582. for _, hit := range res.Hits.Hits {
  583. // 直接收集 _id
  584. ids[hit.Id] = struct{}{}
  585. }
  586. total += len(res.Hits.Hits)
  587. scrollID = res.ScrollId
  588. res, err = client.Scroll().
  589. ScrollId(scrollID).
  590. Scroll(scroll).
  591. Do(ctx)
  592. //log.Println("已获取数量:", total)
  593. if err != nil {
  594. if err == io.EOF {
  595. break
  596. }
  597. return ids, fmt.Errorf("滚动搜索失败: %w", err)
  598. }
  599. }
  600. _, err = client.ClearScroll().ScrollId(scrollID).Do(ctx)
  601. if err != nil {
  602. log.Printf("清理滚动搜索失败:%s", err)
  603. }
  604. return ids, nil
  605. }