es.go 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415
  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(client *elastic.Client, PreBiddingIndex string) error {
  174. createJson := fmt.Sprintf(`{%s,%s}`, setting, mapping)
  175. //fmt.Println(createJson)
  176. //month := int(time.Now().Month())
  177. //monthStr := strconv.Itoa(month)
  178. //year := time.Now().Year()
  179. //yearStr := strconv.Itoa(year)
  180. ////预处理索引名称
  181. //PreBiddingIndex := "bidding_" + yearStr + monthStr
  182. url := GF.ES.URL
  183. username := GF.ES.Username
  184. password := GF.ES.Password
  185. // 创建 Elasticsearch 客户端
  186. //client, err := elastic.NewClient(
  187. // elastic.SetURL(url),
  188. // elastic.SetBasicAuth(username, password),
  189. // elastic.SetSniff(false),
  190. //)
  191. //if err != nil {
  192. // log.Error("创建 Elasticsearch 客户端失败:", zap.Error(err))
  193. //}
  194. exist, err := client.IndexExists(PreBiddingIndex).Do(context.Background())
  195. if exist {
  196. log.Info("CreateIndex", zap.String(PreBiddingIndex, "已经存在了"))
  197. return err
  198. }
  199. //1. 开启节点平衡
  200. // 设置临时的节点平衡设置
  201. balanceSettings := `{
  202. "transient" : {
  203. "cluster.routing.allocation.enable" : "all"
  204. }
  205. }`
  206. requestURL := fmt.Sprintf("%s/_cluster/settings", url)
  207. req, err := http.NewRequest("PUT", requestURL, bytes.NewBuffer([]byte(balanceSettings)))
  208. if err != nil {
  209. log.Error("开启节点平衡", zap.Error(err))
  210. }
  211. req.Header.Set("Content-Type", "application/json")
  212. // 添加身份验证头部
  213. auth := username + ":" + password
  214. basicAuth := "Basic " + base64.StdEncoding.EncodeToString([]byte(auth))
  215. req.Header.Set("Authorization", basicAuth)
  216. clientQ := &http.Client{}
  217. resp, err := clientQ.Do(req)
  218. if err != nil {
  219. log.Error("开启节点平衡", zap.Error(err))
  220. return err
  221. }
  222. defer resp.Body.Close()
  223. if resp.StatusCode != http.StatusOK {
  224. log.Fatal("设置节点平衡失败")
  225. return err
  226. }
  227. fmt.Println("节点平衡已开启")
  228. createIndexR, err := client.CreateIndex(PreBiddingIndex).BodyString(createJson).Do(context.Background())
  229. if err != nil {
  230. log.Error(PreBiddingIndex, zap.Error(err))
  231. return err
  232. }
  233. if !createIndexR.Acknowledged {
  234. log.Error("CreateIndex", zap.String(PreBiddingIndex, "创建索引失败"))
  235. return err
  236. }
  237. defer func() {
  238. //3. 关闭节点平衡
  239. //设置临时的节点平衡设置
  240. disableSettings := `{
  241. "transient" : {
  242. "cluster.routing.allocation.enable" : "none"
  243. }
  244. }`
  245. req2, err2 := http.NewRequest("PUT", requestURL, bytes.NewBuffer([]byte(disableSettings)))
  246. if err2 != nil {
  247. log.Error("开启节点平衡", zap.Error(err))
  248. }
  249. req2.Header.Set("Content-Type", "application/json")
  250. // 添加身份验证头部
  251. req2.Header.Set("Authorization", basicAuth)
  252. //clientQ := &http.Client{}
  253. resp2, err2 := clientQ.Do(req)
  254. if err2 != nil {
  255. log.Error("关闭节点平衡", zap.Error(err))
  256. }
  257. defer resp2.Body.Close()
  258. fmt.Println("节点平衡已关闭")
  259. }()
  260. return nil
  261. }
  262. //deleteIndex 删除索引
  263. func deleteIndex(client *elastic.Client, index string) error {
  264. exist, err := client.IndexExists(index).Do(context.Background())
  265. if !exist {
  266. return errors.New("索引" + index + "不存在")
  267. }
  268. if err != nil {
  269. return err
  270. }
  271. _, err = client.DeleteIndex(index).Do(context.Background())
  272. if err != nil {
  273. return err
  274. }
  275. return nil
  276. }
  277. //dealIndexByDay 处理预处理索引,根据天;
  278. func dealIndexByDay() {
  279. now := time.Now()
  280. PreBiddingIndex := ""
  281. hour := now.Hour()
  282. // 判断当前时间是否时最后一个小时
  283. if hour == 23 {
  284. //当天最后一小时
  285. url := GF.ES.URL
  286. username := GF.ES.Username
  287. password := GF.ES.Password
  288. // 创建 Elasticsearch 客户端
  289. client, err := elastic.NewClient(
  290. elastic.SetURL(url),
  291. elastic.SetBasicAuth(username, password),
  292. elastic.SetSniff(false),
  293. )
  294. if err != nil {
  295. log.Error("创建 Elasticsearch 客户端失败:", zap.Error(err))
  296. }
  297. next := now.AddDate(0, 0, 1)
  298. month := int(next.Month())
  299. monthStr := strconv.Itoa(month)
  300. year := next.Year()
  301. yearStr := strconv.Itoa(year)
  302. dayStr := strconv.Itoa(next.Day())
  303. //下一天的索引名称
  304. PreBiddingIndex = "bidding_" + yearStr + monthStr + dayStr
  305. err = createIndex(client, PreBiddingIndex)
  306. if err != nil {
  307. log.Info("dealIndexByDay", zap.Error(err))
  308. SendMail("预处理索引", "预处理索引创建失败,请检查")
  309. }
  310. log.Info("dealIndexByDay", zap.String(PreBiddingIndex, "创建成功"))
  311. //3. 删除昨天的索引
  312. last := now.AddDate(0, 0, -1)
  313. month2 := int(last.Month())
  314. monthStr2 := strconv.Itoa(month2)
  315. year2 := last.Year()
  316. yearStr2 := strconv.Itoa(year2)
  317. dayStr2 := strconv.Itoa(last.Day())
  318. //索引名称
  319. lastIndex := "bidding_" + yearStr2 + monthStr2 + dayStr2
  320. err = deleteIndex(client, lastIndex)
  321. if err != nil {
  322. log.Info("dealIndexByDay", zap.Error(err))
  323. }
  324. }
  325. }
  326. //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
  327. func dealIndexByMonth() {
  328. now := time.Now()
  329. PreBiddingIndex := ""
  330. // 获取当前月份的最后一天
  331. lastDayOfMonth := time.Date(now.Year(), now.Month()+1, 0, 0, 0, 0, 0, time.UTC)
  332. // 判断当前时间是否为当前月份的最后一天
  333. if now.Day() == lastDayOfMonth.Day() {
  334. //当月最后一天,需要提前创建好索引
  335. fmt.Println("当前时间是当前月份的最后一天")
  336. url := GF.ES.URL
  337. username := GF.ES.Username
  338. password := GF.ES.Password
  339. // 创建 Elasticsearch 客户端
  340. client, err := elastic.NewClient(
  341. elastic.SetURL(url),
  342. elastic.SetBasicAuth(username, password),
  343. elastic.SetSniff(false),
  344. )
  345. if err != nil {
  346. log.Error("创建 Elasticsearch 客户端失败:", zap.Error(err))
  347. }
  348. next := now.AddDate(0, 0, 1)
  349. month := int(next.Month())
  350. monthStr := strconv.Itoa(month)
  351. year := next.Year()
  352. yearStr := strconv.Itoa(year)
  353. //下一个月的索引名称
  354. PreBiddingIndex = "bidding_" + yearStr + monthStr
  355. //2 创建下个月索引结构
  356. err = createIndex(client, PreBiddingIndex)
  357. if err != nil {
  358. log.Info("dealIndexByMonth", zap.Error(err))
  359. SendMail("预处理索引", "预处理索引创建失败,请检查")
  360. }
  361. log.Info("dealIndexByMonth", zap.String(PreBiddingIndex, "创建成功"))
  362. //3. 删除上个月的索引
  363. last := now.AddDate(0, -1, 1)
  364. month2 := int(last.Month())
  365. monthStr2 := strconv.Itoa(month2)
  366. year2 := last.Year()
  367. yearStr2 := strconv.Itoa(year2)
  368. //上个月的索引名称
  369. lastIndex := "bidding_" + yearStr2 + monthStr2
  370. err = deleteIndex(client, lastIndex)
  371. if err != nil {
  372. log.Info("dealIndexByMonth", zap.Error(err))
  373. }
  374. }
  375. }
  376. //SendMail 发送邮件
  377. func SendMail(title, content string) {
  378. url := fmt.Sprintf("%s?to=%s&title=%s&body=%s", GF.Email.Api, GF.Email.To, title, content)
  379. fmt.Println("url=>", url)
  380. res, err := http.Get(url)
  381. if err != nil {
  382. log.Info("SendMail", zap.Any("err", err))
  383. } else {
  384. log.Info("SendMail", zap.Any("res", res))
  385. }
  386. }