main.go 23 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656
  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. c.AddFunc(v.Cron, v.task)
  78. }
  79. c.Start()
  80. defer c.Stop()
  81. select {}
  82. }
  83. log.Println("end..")
  84. }
  85. func (t *T) task() {
  86. //初始化语句
  87. qt := strings.Split(t.Tjscope, ",")
  88. if len(qt) != 2 {
  89. return
  90. }
  91. st, et := int64(0), int64(0)
  92. now := time.Now()
  93. switch qt[1] {
  94. case "h":
  95. et = now.Unix()
  96. st = et + util.Int64All(qt[0])*3600
  97. case "d":
  98. st = util.GetDayStartSecond(util.IntAll(qt[0]))
  99. et = util.GetDayStartSecond(0)
  100. }
  101. st1 := fmt.Sprintf("%x0000000000000000", st)
  102. et1 := fmt.Sprintf("%x0000000000000000", et)
  103. eq := fmt.Sprintf(esQ, st1, et1)
  104. fmt.Println("eq", eq)
  105. eq1 := fmt.Sprintf(esQ1, st1, et1)
  106. fmt.Println("eq1", eq1)
  107. // 将时间戳转换为 time.Time 类型
  108. stt := time.Unix(st, 0)
  109. // 格式化时间为 "YYYY-MM-DD HH:MM:SS" 形式
  110. formattedTime := stt.Format("2006-01-02")
  111. rangeQuery := es.NewRangeQuery("id").Gte(st1).Lt(et1)
  112. //竞品网站统计
  113. 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")
  114. countQuery := es.NewBoolQuery().
  115. Must(rangeQuery)
  116. //竞品网站总量
  117. count1Query := es.NewBoolQuery().Must(rangeQuery).Filter(termsQuery)
  118. //元博网(采购与招标网),竞品网站数量统计
  119. //countComptetQuery1 := es.NewBoolQuery().Must(rangeQuery).Filter(es.NewTermQuery("site", "元博网(采购与招标网)"))
  120. //countComptetQuery2 := es.NewBoolQuery().Must(rangeQuery).Filter(es.NewTermQuery("site", "中国招标与采购网"))
  121. //countComptetQuery3 := es.NewBoolQuery().Must(rangeQuery).Filter(es.NewTermQuery("site", "北京隆道网络科技有限公司"))
  122. //countComptetQuery4 := es.NewBoolQuery().Must(rangeQuery).Filter(es.NewTermQuery("site", "友云采"))
  123. //countComptetQuery5 := es.NewBoolQuery().Must(rangeQuery).Filter(es.NewTermQuery("site", "标800"))
  124. es2 := elastic.Elastic{S_esurl: esAddr2, I_size: 2, Username: username, Password: password}
  125. es2.InitElasticSize()
  126. es3 := elastic.Elastic{S_esurl: esAddr3, I_size: 5, Username: username3, Password: password3}
  127. es3.InitElasticSize()
  128. count := int(es2.Count(esIndex2, countQuery)) //公司 es集群 数量统计
  129. count1 := int(es2.Count(esIndex, count1Query)) //竞品网站 数量统计
  130. ////元博网(采购与招标网)
  131. //countComptet1 := int(es2.Count(esIndex, countComptetQuery1))
  132. ////中国招标与采购网
  133. //countComptet2 := int(es2.Count(esIndex, countComptetQuery2))
  134. ////北京隆道网络科技有限公司
  135. //countComptet3 := int(es2.Count(esIndex, countComptetQuery3))
  136. ////友云采
  137. //countComptet4 := int(es2.Count(esIndex, countComptetQuery4))
  138. ////标800
  139. //countComptet5 := int(es2.Count(esIndex, countComptetQuery5))
  140. countNew := int(es3.Count(esIndex3, countQuery)) //华为云 新集群
  141. countNewTmp := int(es3.Count("bidding_temp", countQuery)) //华为云 新集群
  142. log.Println("count", count)
  143. log.Println("count1", count1)
  144. log.Println("countNew", countNew)
  145. report := ""
  146. switch t.Type {
  147. case "alert":
  148. if count < t.Min || count > t.Max || countNew < t.Min || countNew > t.Max {
  149. report = fmt.Sprintf("告警%s,最小%d,最大%d,统计结果:%d", t.Name, t.Min, t.Max, count)
  150. t.SendMail(report)
  151. wxurl := util.ObjToString(config["wxurl"])
  152. if wxurl != "" {
  153. SendBot(wxurl, report)
  154. }
  155. }
  156. case "report":
  157. //report = fmt.Sprintf("统计报告%s,【统计结果】,es库数量:%d", t.Name, count)
  158. if len(t.Mgo) > 5 {
  159. fs := strings.Split(t.Mgo, "|")
  160. fmgo := mongodb.NewMgoWithUser(fs[0], fs[3], fs[1], fs[2], 1)
  161. id1 := mongodb.StringTOBsonId(st1)
  162. id2 := mongodb.StringTOBsonId(et1)
  163. mq := bson.M{"_id": bson.M{"$gte": id1, "$lt": id2}} //一天时间内的id段
  164. fd := bson.M{"extracttype": 1, "sensitive": 1, "dataging": 1, "site": 1, "infoformat": 1, "comeintime": 1, "pici": 1, "publishtime": 1, "competehref": 1, "attach_text": 1}
  165. sess := fmgo.GetMgoConn()
  166. defer fmgo.DestoryMongoConn(sess)
  167. /**
  168. count 一天内,es 中 数据总量
  169. count1 一天内,es 中 竞品总量
  170. count2 一天内,mgo 总入库量
  171. count3 一天内,mgo 有效数据 总数
  172. count4 一天内,mgo 中 竞品数据总量
  173. count5 一天内,mgo 有效数据中,竞品的数量
  174. countNew 一天内,es3新集群 中 数据总量
  175. timeCount pici-comeintime 时差在12小时内的数据
  176. */
  177. count2, count3 := 0, 0 //
  178. count4, count5 := 0, 0 //竟品
  179. es_comeintime_totaltime := int64(0) //comeintime 和 生索引 pici 时间 差值的总和
  180. es_publishtime_totaltime := int64(0) //publishtime 和 生索引 pici 时间 差值的总和
  181. es_comeintime_avgtime := int64(0) //comeintime 和 生索引 pici 时间 差值的平均值
  182. es_publishtime_avgtime := int64(0) //publishtime 和 生索引 pici 时间 差值的平均值
  183. file_totaltime := int64(0)
  184. no_file_totaltime := int64(0)
  185. file_avgltime := int64(0)
  186. no_file_avgltime := int64(0)
  187. timeCount := 0 //12小时 统计时间差的数据量
  188. fileCount := 0
  189. noFileCount := 0
  190. //统计pici -comeintime 时间差,1、3、5、10、15、30、30+ 分钟
  191. var pc_diff1 int64
  192. var pc_diff3 int64
  193. var pc_diff5 int64
  194. var pc_diff10 int64
  195. var pc_diff15 int64
  196. var pc_diff30 int64
  197. var pc_diff301 int64
  198. var competCount1 int
  199. var competCount2 int
  200. var competCount3 int
  201. var competCount4 int
  202. var competCount5 int
  203. competeReal1 := 0
  204. competeReal2 := 0
  205. competeReal3 := 0
  206. competeReal4 := 0
  207. competeReal5 := 0
  208. query := sess.DB(fs[3]).C(fs[4]).Find(mq).Select(fd).Iter()
  209. for tmp := make(map[string]interface{}); query.Next(tmp); count2++ {
  210. 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" {
  211. count4++
  212. if util.ObjToString(tmp["site"]) == "元博网(采购与招标网)" {
  213. competCount1++
  214. }
  215. if util.ObjToString(tmp["site"]) == "中国招标与采购网" {
  216. competCount2++
  217. }
  218. if util.ObjToString(tmp["site"]) == "北京隆道网络科技有限公司" {
  219. competCount3++
  220. }
  221. if util.ObjToString(tmp["site"]) == "友云采" {
  222. competCount4++
  223. }
  224. if util.ObjToString(tmp["site"]) == "标800" {
  225. competCount5++
  226. }
  227. }
  228. if util.IntAll(tmp["extracttype"]) != -1 && util.ObjToString(tmp["sensitive"]) != "测试" && util.IntAll(tmp["dataging"]) != 1 && util.Float64All(tmp["infoformat"]) != 3 {
  229. count3++
  230. comeintime := util.Int64All(tmp["comeintime"])
  231. publishtime := util.Int64All(tmp["publishtime"])
  232. pici := util.Int64All(tmp["pici"])
  233. tp := time.Unix(publishtime, 0)
  234. isZero := false
  235. if (tp.Hour() == 0 && tp.Minute() == 0) && tp.Second() == 0 {
  236. isZero = true
  237. }
  238. if (comeintime-publishtime) < 12*60*60 && !isZero {
  239. if pici > 0 {
  240. diff1 := pici - comeintime
  241. diff2 := pici - publishtime
  242. if diff1 < 0 {
  243. fmt.Println("diff1", diff1, tmp["_id"])
  244. } else if diff1 <= 60 {
  245. pc_diff1++
  246. } else if diff1 <= 3*60 {
  247. pc_diff3++
  248. } else if diff1 <= 5*60 {
  249. pc_diff5++
  250. } else if diff1 <= 10*60 {
  251. pc_diff10++
  252. } else if diff1 <= 15*60 {
  253. pc_diff15++
  254. } else if diff1 <= 30*60 {
  255. pc_diff30++
  256. } else {
  257. pc_diff301++
  258. }
  259. if diff2 < 0 {
  260. fmt.Println("diff2", diff2, tmp["_id"])
  261. }
  262. es_comeintime_totaltime += diff1
  263. es_publishtime_totaltime += diff2
  264. timeCount++
  265. if _, ok := tmp["attach_text"]; ok {
  266. curtime := tmp["_id"].(primitive.ObjectID).Timestamp().Unix()
  267. diff3 := curtime - comeintime
  268. if diff3 >= 0 {
  269. file_totaltime += diff3
  270. }
  271. fileCount++
  272. } else {
  273. curtime := tmp["_id"].(primitive.ObjectID).Timestamp().Unix()
  274. diff4 := curtime - comeintime
  275. if diff4 >= 0 {
  276. no_file_totaltime += diff4
  277. }
  278. noFileCount++
  279. }
  280. }
  281. }
  282. 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" {
  283. count5++
  284. if util.ObjToString(tmp["site"]) == "元博网(采购与招标网)" {
  285. competeReal1++
  286. }
  287. if util.ObjToString(tmp["site"]) == "中国招标与采购网" {
  288. competeReal2++
  289. }
  290. if util.ObjToString(tmp["site"]) == "北京隆道网络科技有限公司" {
  291. competeReal3++
  292. }
  293. if util.ObjToString(tmp["site"]) == "友云采" {
  294. competeReal4++
  295. }
  296. if util.ObjToString(tmp["site"]) == "标800" {
  297. competeReal5++
  298. }
  299. }
  300. }
  301. }
  302. if timeCount > 0 {
  303. es_comeintime_avgtime = es_comeintime_totaltime / int64(timeCount)
  304. es_publishtime_avgtime = es_publishtime_totaltime / int64(timeCount)
  305. }
  306. if fileCount > 0 {
  307. file_avgltime = file_totaltime / int64(fileCount)
  308. }
  309. if noFileCount > 0 {
  310. no_file_avgltime = no_file_totaltime / int64(noFileCount)
  311. }
  312. //report += `,mgo统计:` + fmt.Sprint(count3) + `,差值:` + fmt.Sprint(count3-count) + `,mgo总入库量:` + fmt.Sprint(count2) +
  313. // `<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) +
  314. // `<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) +
  315. // `<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) +
  316. // `<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) +
  317. // `<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) +
  318. // `<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)
  319. //html 格式
  320. //var reportBuilder strings.Builder
  321. //// 开始表格
  322. //reportBuilder.WriteString("<table>")
  323. //reportBuilder.WriteString(formattedTime)
  324. //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>")
  325. //
  326. //addTableRow(&reportBuilder, "阿里云es集群", count, count3, count3-count, count2)
  327. //addTableRow(&reportBuilder, "竟品统计结果", count1, count5, count5-count1, count4)
  328. //addTableRow(&reportBuilder, "元博网", countComptet1, competeReal1, competeReal1-countComptet1, competCount1)
  329. //addTableRow(&reportBuilder, "中国招标与采购网", countComptet2, competeReal2, competeReal2-countComptet2, competCount2)
  330. //addTableRow(&reportBuilder, "北京隆道网络科技有限公司", countComptet3, competeReal3, competeReal3-countComptet3, competCount3)
  331. //addTableRow(&reportBuilder, "友云采", countComptet4, competeReal4, competeReal4-countComptet4, competCount4)
  332. //addTableRow(&reportBuilder, "标800", countComptet5, competeReal5, competeReal5-countComptet5, competCount5)
  333. //addTableRow(&reportBuilder, "华为云es集群", countNew, count3, count3-countNew, count2)
  334. //// 结束表格
  335. //reportBuilder.WriteString("</table>")
  336. //// 获取最终的报告字符串
  337. //report = reportBuilder.String()
  338. var reportBuilder strings.Builder
  339. // 添加时间
  340. reportBuilder.WriteString(formattedTime + "\n")
  341. // Markdown 表头
  342. reportBuilder.WriteString("| 统计项 | es统计 | mgo统计 | 差值 | mgo总入库量 |\n")
  343. reportBuilder.WriteString("|--------|--------|---------|------|--------------|\n")
  344. addMarkdownRow(&reportBuilder, "阿里云es集群", count, count3, count3-count, count2)
  345. //addMarkdownRow(&reportBuilder, "竟品统计结果", count1, count5, count5-count1, count4)
  346. //addMarkdownRow(&reportBuilder, "元博网", countComptet1, competeReal1, competeReal1-countComptet1, competCount1)
  347. //addMarkdownRow(&reportBuilder, "中国招标与采购网", countComptet2, competeReal2, competeReal2-countComptet2, competCount2)
  348. //addMarkdownRow(&reportBuilder, "北京隆道网络科技有限公司", countComptet3, competeReal3, competeReal3-countComptet3, competCount3)
  349. //addMarkdownRow(&reportBuilder, "友云采", countComptet4, competeReal4, competeReal4-countComptet4, competCount4)
  350. //addMarkdownRow(&reportBuilder, "标800", countComptet5, competeReal5, competeReal5-countComptet5, competCount5)
  351. addMarkdownRow(&reportBuilder, "华为云es集群", countNew, count3, count3-countNew, count2)
  352. addMarkdownRow(&reportBuilder, "华为云es集群 bidding_temp", countNewTmp, count3, count3-countNewTmp, count2)
  353. report = reportBuilder.String()
  354. //存入数据库
  355. yesterday := now.AddDate(0, 0, -1)
  356. insert := map[string]interface{}{
  357. "es_count": count, //
  358. "es3_count": countNew, //
  359. "mgo_count": count3,
  360. "es_mgo_diff": count3 - count,
  361. "mgo_total": count2,
  362. "competitor_es_count": count1, //竞品网站es 数量
  363. "competitor_mgo_count": count5,
  364. "competitor_diff": count5 - count1,
  365. "competitor_mgo_total": count4,
  366. "date": yesterday.Format("2006-01-02"),
  367. "es_comeintime_totaltime": es_comeintime_totaltime,
  368. "es_publishtime_totaltime": es_publishtime_totaltime,
  369. "es_comeintime_avgtime": es_comeintime_avgtime,
  370. "es_publishtime_avgtime": es_publishtime_avgtime,
  371. "file_avgltime": file_avgltime,
  372. "no_file_avgltime": no_file_avgltime,
  373. "file_totaltime": file_totaltime,
  374. "no_file_totaltime": no_file_totaltime,
  375. "file_count": fileCount,
  376. "no_file_count": noFileCount,
  377. "pc_diff1": pc_diff1,
  378. "pc_diff3": pc_diff3,
  379. "pc_diff5": pc_diff5,
  380. "pc_diff10": pc_diff10,
  381. "pc_diff15": pc_diff15,
  382. "pc_diff30": pc_diff30,
  383. "pc_diff301": pc_diff301,
  384. "timeCount": timeCount,
  385. }
  386. fmgo.Save("bidding_ribao", insert)
  387. }
  388. t.SendMail(report)
  389. //t.SendMail("【竟品统计结果】")
  390. //fmt.Println(report)
  391. // 发送企业微信
  392. wxurl := util.ObjToString(config["wxurl"])
  393. if wxurl != "" {
  394. SendBot(wxurl, report)
  395. }
  396. //9008 bidding 数据和 9005 bidding 数据量不一致
  397. if countNew != count {
  398. client1, _ := es.NewClient(
  399. es.SetURL(esAddr2),
  400. es.SetBasicAuth(username, password),
  401. es.SetSniff(false),
  402. )
  403. client2, _ := es.NewClient(
  404. es.SetURL(esAddr3),
  405. es.SetBasicAuth(username3, password3),
  406. es.SetSniff(false),
  407. )
  408. // 9008 比 9005 的索引数据量多
  409. if count > countNew {
  410. ids1, err := fetchIDs(client1, "bidding", rangeQuery)
  411. if err != nil {
  412. log.Fatal(err)
  413. }
  414. fmt.Printf("集群 9008 bidding 总数: %d\n", len(ids1))
  415. ids2, err := fetchIDs(client2, "bidding", rangeQuery)
  416. if err != nil {
  417. log.Fatal(err)
  418. }
  419. fmt.Printf("集群9005 bidding 总数: %d\n", len(ids2))
  420. fmt.Println("差集 (集群9008 - 集群9005):")
  421. diffIDS := make([]string, 0)
  422. for id := range ids1 {
  423. if _, ok := ids2[id]; !ok {
  424. fmt.Println(id)
  425. diffIDS = append(diffIDS, id)
  426. }
  427. }
  428. text := fmt.Sprintf("集群9008 - 集群9005 bidding 的差集数据是:\n %v", diffIDS)
  429. SendBot(wxurl, text)
  430. } else {
  431. ids1, err := fetchIDs(client1, "bidding", rangeQuery)
  432. if err != nil {
  433. log.Fatal(err)
  434. }
  435. fmt.Printf("集群 9008 bidding 总数: %d\n", len(ids1))
  436. ids2, err := fetchIDs(client2, "bidding", rangeQuery)
  437. if err != nil {
  438. log.Fatal(err)
  439. }
  440. fmt.Printf("集群9005 bidding 总数: %d\n", len(ids2))
  441. fmt.Println("差集 (集群9005 - 集群9008):")
  442. diffIDS := make([]string, 0)
  443. for id := range ids2 {
  444. if _, ok := ids1[id]; !ok {
  445. fmt.Println(id)
  446. }
  447. }
  448. text := fmt.Sprintf("集群9005 - 集群9008 bidding 的差集数据是:\n %v", diffIDS)
  449. SendBot(wxurl, text)
  450. }
  451. }
  452. if countNewTmp != count {
  453. client1, _ := es.NewClient(
  454. es.SetURL(esAddr2),
  455. es.SetBasicAuth(username, password),
  456. es.SetSniff(false),
  457. )
  458. client2, _ := es.NewClient(
  459. es.SetURL(esAddr3),
  460. es.SetBasicAuth(username3, password3),
  461. es.SetSniff(false),
  462. )
  463. if count > countNewTmp {
  464. ids1, err := fetchIDs(client1, "bidding", rangeQuery)
  465. if err != nil {
  466. log.Fatal(err)
  467. }
  468. fmt.Printf("集群 9008 bidding 总数: %d\n", len(ids1))
  469. ids2, err := fetchIDs(client2, "bidding_temp", rangeQuery)
  470. if err != nil {
  471. log.Fatal(err)
  472. }
  473. fmt.Printf("集群9005 bidding_temp 总数: %d\n", len(ids2))
  474. fmt.Println("差集 (集群9008 bidding - 集群9005 bidding_temp ):")
  475. diffIDS := make([]string, 0)
  476. for id := range ids1 {
  477. if _, ok := ids2[id]; !ok {
  478. fmt.Println(id)
  479. diffIDS = append(diffIDS, id)
  480. }
  481. }
  482. text := fmt.Sprintf("集群9008 bidding - 集群9005 bidding_temp 的差集数据是:\n %v", diffIDS)
  483. SendBot(wxurl, text)
  484. } else {
  485. ids1, err := fetchIDs(client1, "bidding", rangeQuery)
  486. if err != nil {
  487. log.Fatal(err)
  488. }
  489. fmt.Printf("集群 9008 bidding 总数: %d\n", len(ids1))
  490. ids2, err := fetchIDs(client2, "bidding_temp", rangeQuery)
  491. if err != nil {
  492. log.Fatal(err)
  493. }
  494. fmt.Printf("集群9005 bidding_temp 总数: %d\n", len(ids2))
  495. fmt.Println("差集 (集群9005 biding_temp - 集群9008 bidding ):")
  496. diffIDS := make([]string, 0)
  497. for id := range ids2 {
  498. if _, ok := ids1[id]; !ok {
  499. fmt.Println(id)
  500. }
  501. }
  502. text := fmt.Sprintf("集群9005 biding_temp - 集群9008 bidding的差集数据是:\n %v", diffIDS)
  503. SendBot(wxurl, text)
  504. }
  505. }
  506. }
  507. log.Println("task over:", t.Name, eq, count)
  508. }
  509. func (t *T) SendMail(report string) {
  510. url := fmt.Sprintf("%s?to=%s&title=%s&body=%s", api, to, t.Name, report)
  511. fmt.Println("url=>", url)
  512. res, err := http.Get(url)
  513. if err != nil {
  514. fmt.Println("SendMail err ", err)
  515. } else {
  516. fmt.Println("SendMail res ", res)
  517. }
  518. }
  519. func addTableRow(builder *strings.Builder, itemName string, esCount, mgoCount, diff, mgoTotal int) {
  520. 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))
  521. }
  522. func addMarkdownRow(builder *strings.Builder, itemName string, esCount, mgoCount, diff, mgoTotal int) {
  523. builder.WriteString(fmt.Sprintf("| %s | %d | %d | %d | %d |\n", itemName, esCount, mgoCount, diff, mgoTotal))
  524. }
  525. // SendBot 发送企业微信
  526. func SendBot(webhookURL, msg string) (b bool) {
  527. // 构造请求体
  528. payload := map[string]interface{}{
  529. "msgtype": "text",
  530. "text": map[string]string{
  531. "content": msg,
  532. },
  533. }
  534. // 转换为 JSON 字符串
  535. payloadBytes, err := json.Marshal(payload)
  536. if err != nil {
  537. log.Println("Error:", err)
  538. return
  539. }
  540. // 发送 POST 请求
  541. resp, err := http.Post(webhookURL, "application/json", bytes.NewReader(payloadBytes))
  542. if err != nil {
  543. log.Println("Error:", err)
  544. return
  545. }
  546. defer resp.Body.Close()
  547. b = true
  548. //打印响应结果
  549. log.Println("send bot Status:", resp.Status)
  550. return
  551. }
  552. func fetchIDs(client *es.Client, index string, query es.Query) (map[string]struct{}, error) {
  553. ctx := context.Background()
  554. ids := make(map[string]struct{})
  555. scrollID := ""
  556. scroll := "10m"
  557. searchSource := es.NewSearchSource().
  558. Query(query).
  559. Size(10000).
  560. Sort("_doc", true)
  561. searchService := client.Scroll(index).
  562. Size(10000).
  563. Scroll(scroll).
  564. SearchSource(searchSource)
  565. res, err := searchService.Do(ctx)
  566. if err != nil {
  567. if err == io.EOF {
  568. fmt.Println("没有数据")
  569. return ids, nil
  570. }
  571. return nil, fmt.Errorf("初始化滚动搜索失败: %w", err)
  572. }
  573. fmt.Println("命中总数:", res.TotalHits())
  574. total := 0
  575. for len(res.Hits.Hits) > 0 {
  576. for _, hit := range res.Hits.Hits {
  577. // 直接收集 _id
  578. ids[hit.Id] = struct{}{}
  579. }
  580. total += len(res.Hits.Hits)
  581. scrollID = res.ScrollId
  582. res, err = client.Scroll().
  583. ScrollId(scrollID).
  584. Scroll(scroll).
  585. Do(ctx)
  586. //log.Println("已获取数量:", total)
  587. if err != nil {
  588. if err == io.EOF {
  589. break
  590. }
  591. return ids, fmt.Errorf("滚动搜索失败: %w", err)
  592. }
  593. }
  594. _, err = client.ClearScroll().ScrollId(scrollID).Do(ctx)
  595. if err != nil {
  596. log.Printf("清理滚动搜索失败:%s", err)
  597. }
  598. return ids, nil
  599. }