es.go 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490
  1. package main
  2. import (
  3. "bytes"
  4. "context"
  5. "encoding/base64"
  6. "errors"
  7. "fmt"
  8. "github.com/olivere/elastic/v7"
  9. "go.uber.org/zap"
  10. "net/http"
  11. "strconv"
  12. "jygit.jydev.jianyu360.cn/data_processing/common_utils/log"
  13. "time"
  14. )
  15. var setting = fmt.Sprintf(` "settings": {
  16. "index": {
  17. "analysis": {
  18. "analyzer": {
  19. "my_ngram_title": {
  20. "filter": [
  21. "lowercase"
  22. ],
  23. "tokenizer": "my_ngram_title"
  24. },
  25. "douhao": {
  26. "type": "pattern",
  27. "pattern": ","
  28. },
  29. "my_ngram": {
  30. "filter": [
  31. "lowercase"
  32. ],
  33. "tokenizer": "my_ngram"
  34. }
  35. },
  36. "tokenizer": {
  37. "my_ngram_title": {
  38. "token_chars": [
  39. "letter",
  40. "digit",
  41. "punctuation",
  42. "symbol"
  43. ],
  44. "min_gram": "1",
  45. "type": "nGram",
  46. "max_gram": "1"
  47. },
  48. "my_ngram": {
  49. "token_chars": [
  50. "letter",
  51. "digit",
  52. "punctuation",
  53. "symbol"
  54. ],
  55. "min_gram": "2",
  56. "type": "nGram",
  57. "max_gram": "2"
  58. }
  59. }
  60. },
  61. "number_of_shards": "%s",
  62. "number_of_replicas": "0",
  63. "max_result_window": "20000"
  64. }
  65. }`, "2")
  66. var mapping = ` "mappings": {
  67. "dynamic": false,
  68. "properties": {
  69. "dataweight": {
  70. "type": "long"
  71. },
  72. "projectcode": {
  73. "type": "keyword"
  74. },
  75. "object_type": {
  76. "type": "keyword"
  77. },
  78. "bidopentime": {
  79. "type": "long"
  80. },
  81. "bidamount": {
  82. "type": "double"
  83. },
  84. "winner": {
  85. "type": "keyword"
  86. },
  87. "buyer": {
  88. "type": "keyword",
  89. "fields": {
  90. "mbuyer": {
  91. "analyzer": "my_ngram_title",
  92. "type": "text"
  93. }
  94. }
  95. },
  96. "budget": {
  97. "type": "double"
  98. },
  99. "projectname": {
  100. "type": "keyword",
  101. "fields": {
  102. "pname": {
  103. "analyzer": "my_ngram_title",
  104. "type": "text"
  105. }
  106. }
  107. },
  108. "area": {
  109. "type": "keyword"
  110. },
  111. "city": {
  112. "type": "keyword"
  113. },
  114. "district": {
  115. "type": "keyword"
  116. },
  117. "s_winner": {
  118. "analyzer": "douhao",
  119. "type": "text",
  120. "fields": {
  121. "mwinner": {
  122. "analyzer": "my_ngram_title",
  123. "type": "text"
  124. }
  125. }
  126. },
  127. "pici": {
  128. "type": "long"
  129. },
  130. "id": {
  131. "type": "keyword"
  132. },
  133. "title": {
  134. "analyzer": "my_ngram_title",
  135. "type": "text",
  136. "fields": {
  137. "mtitle": {
  138. "type": "keyword"
  139. }
  140. }
  141. },
  142. "detail": {
  143. "analyzer": "my_ngram",
  144. "type": "text"
  145. },
  146. "site": {
  147. "type": "keyword"
  148. },
  149. "comeintime": {
  150. "type": "long"
  151. },
  152. "href": {
  153. "type": "keyword"
  154. },
  155. "infoformat": {
  156. "type": "long"
  157. },
  158. "publishtime": {
  159. "type": "long"
  160. },
  161. "toptype": {
  162. "type": "keyword"
  163. },
  164. "subtype": {
  165. "type": "keyword"
  166. },
  167. "createtime": {
  168. "type": "long"
  169. }
  170. }
  171. }`
  172. //createIndex 创建索引
  173. func createIndex(clients map[string]*elastic.Client, PreBiddingIndex string) error {
  174. createJson := fmt.Sprintf(`{%s,%s}`, setting, mapping)
  175. for k, client := range clients {
  176. url := GF.ES[k].URL
  177. username := GF.ES[k].Username
  178. password := GF.ES[k].Password
  179. exist, err := client.IndexExists(PreBiddingIndex).Do(context.Background())
  180. if exist {
  181. log.Info("CreateIndex", zap.String(PreBiddingIndex, "已经存在了"))
  182. }
  183. //1. 开启节点平衡
  184. // 设置临时的节点平衡设置
  185. balanceSettings := `{
  186. "transient" : {
  187. "cluster.routing.allocation.enable" : "all"
  188. }
  189. }`
  190. requestURL := fmt.Sprintf("%s/_cluster/settings", url)
  191. req, err := http.NewRequest("PUT", requestURL, bytes.NewBuffer([]byte(balanceSettings)))
  192. if err != nil {
  193. log.Error("开启节点平衡", zap.Error(err))
  194. }
  195. req.Header.Set("Content-Type", "application/json")
  196. // 添加身份验证头部
  197. auth := username + ":" + password
  198. basicAuth := "Basic " + base64.StdEncoding.EncodeToString([]byte(auth))
  199. req.Header.Set("Authorization", basicAuth)
  200. clientQ := &http.Client{}
  201. resp, err := clientQ.Do(req)
  202. if err != nil {
  203. log.Error("开启节点平衡", zap.Error(err))
  204. return err
  205. }
  206. defer resp.Body.Close()
  207. if resp.StatusCode != http.StatusOK {
  208. log.Fatal("设置节点平衡失败")
  209. return err
  210. }
  211. fmt.Println(url, "节点平衡已开启")
  212. //2. 创建索引
  213. createIndexR, err := client.CreateIndex(PreBiddingIndex).BodyString(createJson).Do(context.Background())
  214. if err != nil {
  215. log.Error(PreBiddingIndex, zap.Error(err))
  216. return err
  217. }
  218. if !createIndexR.Acknowledged {
  219. log.Error("CreateIndex", zap.String(PreBiddingIndex, "创建索引失败"))
  220. return err
  221. }
  222. //3. 关闭节点平衡
  223. //设置临时的节点平衡设置
  224. disableSettings := `{
  225. "transient" : {
  226. "cluster.routing.allocation.enable" : "none"
  227. }
  228. }`
  229. req2, err2 := http.NewRequest("PUT", requestURL, bytes.NewBuffer([]byte(disableSettings)))
  230. if err2 != nil {
  231. log.Error("开启节点平衡", zap.Error(err))
  232. }
  233. req2.Header.Set("Content-Type", "application/json")
  234. // 添加身份验证头部
  235. req2.Header.Set("Authorization", basicAuth)
  236. //clientQ := &http.Client{}
  237. resp2, err2 := clientQ.Do(req)
  238. if err2 != nil {
  239. log.Error("关闭节点平衡", zap.Error(err))
  240. }
  241. defer resp2.Body.Close()
  242. fmt.Println(url, "节点平衡已关闭")
  243. //4. 新索引 添加别名
  244. for _, alias := range GF.Env.Alias {
  245. _, err = client.Alias().Add(PreBiddingIndex, alias).Do(context.Background())
  246. if err != nil {
  247. log.Error("添加别名失败:", zap.Error(err))
  248. }
  249. }
  250. log.Info("CreateIndex", zap.String(url, "索引别名 添加完毕"))
  251. }
  252. return nil
  253. }
  254. //deleteIndex 删除索引
  255. func deleteIndex(clients map[string]*elastic.Client, index string) error {
  256. if len(clients) == 0 {
  257. return errors.New("没有 配置 ES集群信息")
  258. }
  259. for k, client := range clients {
  260. exist, err := client.IndexExists(index).Do(context.Background())
  261. if !exist {
  262. log.Info("deleteIndex", zap.String(k, index+" 索引文件不存在"))
  263. }
  264. if err != nil {
  265. return err
  266. }
  267. num, err := client.Count(index).Do(context.Background())
  268. if err != nil {
  269. return err
  270. }
  271. if num > 0 {
  272. log.Info("deleteIndex", zap.String(k, "索引"+index+"还存在有效数据"))
  273. }
  274. _, err = client.DeleteIndex(index).Do(context.Background())
  275. if err != nil {
  276. return err
  277. }
  278. }
  279. return nil
  280. }
  281. //dealIndexByDay 处理预处理索引,根据天;
  282. func dealIndexByDay() {
  283. now := time.Now()
  284. PreBiddingIndex := ""
  285. //hour := now.Hour()
  286. // 判断当前时间是否时最后一个小时
  287. //if hour == 23 {
  288. //当天最后一小时
  289. var clients = make(map[string]*elastic.Client, 0)
  290. for k, v := range GF.ES {
  291. url := v.URL
  292. username := v.Username
  293. password := v.Password
  294. // 创建 Elasticsearch 客户端
  295. client, err := elastic.NewClient(
  296. elastic.SetURL(url),
  297. elastic.SetBasicAuth(username, password),
  298. elastic.SetSniff(false),
  299. )
  300. if err != nil {
  301. log.Error("创建 Elasticsearch 客户端失败:", zap.Error(err))
  302. }
  303. clients[k] = client
  304. }
  305. next := now.AddDate(0, 0, 1)
  306. month := int(next.Month())
  307. monthStr := strconv.Itoa(month)
  308. year := next.Year()
  309. yearStr := strconv.Itoa(year)
  310. dayStr := strconv.Itoa(next.Day())
  311. //下一天的索引名称
  312. PreBiddingIndex = "bidding_" + yearStr + monthStr + dayStr
  313. err := createIndex(clients, PreBiddingIndex)
  314. if err != nil {
  315. log.Info("dealIndexByDay", zap.Error(err))
  316. SendMail("预处理索引", "预处理索引创建失败,请检查")
  317. }
  318. log.Info("dealIndexByDay", zap.String(PreBiddingIndex, "创建成功"))
  319. //3. 删除昨天的索引
  320. last := now.AddDate(0, 0, -1)
  321. month2 := int(last.Month())
  322. monthStr2 := strconv.Itoa(month2)
  323. year2 := last.Year()
  324. yearStr2 := strconv.Itoa(year2)
  325. dayStr2 := strconv.Itoa(last.Day())
  326. //索引名称
  327. lastIndex := "bidding_" + yearStr2 + monthStr2 + dayStr2
  328. err = deleteIndex(clients, lastIndex)
  329. if err != nil {
  330. log.Info("dealIndexByDay", zap.Error(err))
  331. }
  332. //}
  333. }
  334. //dealIndexByMonth 处理预处理索引,根据月份;提前一天创建好 下个月的索引
  335. func dealIndexByMonth() {
  336. now := time.Now()
  337. PreBiddingIndex := ""
  338. // 获取当前月份的最后一天
  339. lastDayOfMonth := time.Date(now.Year(), now.Month()+1, 0, 0, 0, 0, 0, time.UTC)
  340. // 判断当前时间是否为当前月份的最后一天
  341. if now.Day() == lastDayOfMonth.Day() {
  342. //当月最后一天,需要提前创建好索引
  343. fmt.Println("当前时间是当前月份的最后一天")
  344. var clients = make(map[string]*elastic.Client, 0)
  345. for k, v := range GF.ES {
  346. url := v.URL
  347. username := v.Username
  348. password := v.Password
  349. // 创建 Elasticsearch 客户端
  350. client, err := elastic.NewClient(
  351. elastic.SetURL(url),
  352. elastic.SetBasicAuth(username, password),
  353. elastic.SetSniff(false),
  354. )
  355. if err != nil {
  356. log.Error("创建 Elasticsearch 客户端失败:", zap.Error(err))
  357. }
  358. clients[k] = client
  359. }
  360. next := now.AddDate(0, 0, 1)
  361. month := int(next.Month())
  362. monthStr := strconv.Itoa(month)
  363. year := next.Year()
  364. yearStr := strconv.Itoa(year)
  365. //下一个月的索引名称
  366. PreBiddingIndex = "bidding_" + yearStr + monthStr
  367. //2 创建下个月索引结构
  368. err := createIndex(clients, PreBiddingIndex)
  369. if err != nil {
  370. log.Info("dealIndexByMonth", zap.Error(err))
  371. SendMail("预处理索引", "预处理索引创建失败,请检查")
  372. }
  373. log.Info("dealIndexByMonth", zap.String(PreBiddingIndex, "创建成功"))
  374. //3. 删除上个月的索引
  375. last := now.AddDate(0, -1, 1)
  376. month2 := int(last.Month())
  377. monthStr2 := strconv.Itoa(month2)
  378. year2 := last.Year()
  379. yearStr2 := strconv.Itoa(year2)
  380. //上个月的索引名称
  381. lastIndex := "bidding_" + yearStr2 + monthStr2
  382. err = deleteIndex(clients, lastIndex)
  383. if err != nil {
  384. log.Info("dealIndexByMonth", zap.Error(err))
  385. }
  386. }
  387. }
  388. ////SwitchAlias 切换别名
  389. //func SwitchAlias() {
  390. // now := time.Now()
  391. // // 判断当前时间是否为当前月份的最后一天
  392. // //按日创建索引
  393. // if GF.Env.SpecType == "day" {
  394. // dealAlias()
  395. // } else if GF.Env.SpecType == "month" {
  396. // // 获取当前月份的最后一天
  397. // lastDayOfMonth := time.Date(now.Year(), now.Month()+1, 0, 0, 0, 0, 0, time.UTC)
  398. // //按月创建索引
  399. // if now.Day() == lastDayOfMonth.Day() {
  400. // dealAlias()
  401. // }
  402. // }
  403. //}
  404. //
  405. ////dealAlias 处理 索引别名
  406. //func dealAlias() {
  407. // now := time.Now()
  408. //
  409. // for _, v := range GF.ES {
  410. // url := v.URL
  411. // username := v.Username
  412. // password := v.Password
  413. // // 创建 Elasticsearch 客户端
  414. // client, err := elastic.NewClient(
  415. // elastic.SetURL(url),
  416. // elastic.SetBasicAuth(username, password),
  417. // elastic.SetSniff(false),
  418. // )
  419. // if err != nil {
  420. // log.Error("创建 Elasticsearch 客户端失败:", zap.Error(err))
  421. // }
  422. //
  423. // preBiddingIndex := "" //下一个索引名称
  424. // next := now.AddDate(0, 0, 1)
  425. // month := int(next.Month())
  426. // monthStr := strconv.Itoa(month)
  427. // year := next.Year()
  428. // yearStr := strconv.Itoa(year)
  429. // dayStr := strconv.Itoa(next.Day())
  430. // if GF.Env.SpecType == "month" {
  431. // //下一个月的索引名称
  432. // preBiddingIndex = "bidding_" + yearStr + monthStr
  433. // } else if GF.Env.SpecType == "day" {
  434. // //下一天的索引名称
  435. // preBiddingIndex = "bidding_" + yearStr + monthStr + dayStr
  436. // }
  437. //
  438. // for _, alias := range GF.Env.Alias {
  439. // _, err = client.Alias().Add(preBiddingIndex, alias).Do(context.Background())
  440. // if err != nil {
  441. // log.Error("添加别名失败:", zap.Error(err))
  442. // }
  443. // }
  444. // log.Info("dealAlias", zap.String(url, "索引别名处理完毕"))
  445. // }
  446. //}
  447. //SendMail 发送邮件
  448. func SendMail(title, content string) {
  449. url := fmt.Sprintf("%s?to=%s&title=%s&body=%s", GF.Email.Api, GF.Email.To, title, content)
  450. fmt.Println("url=>", url)
  451. res, err := http.Get(url)
  452. if err != nil {
  453. log.Info("SendMail", zap.Any("err", err))
  454. } else {
  455. log.Info("SendMail", zap.Any("res", res))
  456. }
  457. }