main.go 5.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256
  1. package main
  2. import (
  3. "data_ai/clean"
  4. "data_ai/extract"
  5. "data_ai/udp"
  6. "data_ai/ul"
  7. log "github.com/donnie4w/go-logger/logger"
  8. qu "jygit.jydev.jianyu360.cn/data_processing/common_utils"
  9. "sync"
  10. "unicode/utf8"
  11. )
  12. func init() {
  13. ul.IsTool = false //是否工具
  14. ul.IsLocal = true //是否本地
  15. ul.IsFull = false //是否全量
  16. if ul.IsTool {
  17. ul.InitToolVar()
  18. } else {
  19. ul.InitGlobalVar()
  20. if !ul.IsFull {
  21. udp.InitProcessVar()
  22. }
  23. }
  24. ul.InitOss(ul.IsLocal)
  25. }
  26. func main() {
  27. if ul.IsTool {
  28. //tool.StartToolUpdateInfo()
  29. return
  30. }
  31. //extract.TestSingleFieldInfo("bidding", "6722de29b25c3e1debe624c9")
  32. test1()
  33. lock := make(chan bool)
  34. <-lock
  35. }
  36. func test() {
  37. log.Debug("开始大模型验证内存数据···")
  38. q := map[string]interface{}{}
  39. pool_mgo := make(chan bool, 500)
  40. wg_mgo := &sync.WaitGroup{}
  41. sess := ul.SourceMgo.GetMgoConn()
  42. defer ul.SourceMgo.DestoryMongoConn(sess)
  43. total, isok := 0, 0
  44. it := sess.DB(ul.SourceMgo.DbName).C(ul.Bid_Name).Find(&q).Sort("-_id").Iter()
  45. for tmp := make(map[string]interface{}); it.Next(&tmp); total++ {
  46. if total%1000 == 0 {
  47. log.Debug("cur index ", total)
  48. }
  49. isok++
  50. pool_mgo <- true
  51. wg_mgo.Add(1)
  52. go func(tmp map[string]interface{}) {
  53. defer func() {
  54. <-pool_mgo
  55. wg_mgo.Done()
  56. }()
  57. detail := qu.ObjToString(tmp["detail"])
  58. if utf8.RuneCountInString(detail) < 100 {
  59. data := extract.ResolveInfo(tmp)
  60. if len(data) > 0 {
  61. tmp["ai_zhipu"] = data
  62. ul.SourceMgo.Save("zktest_1031", tmp)
  63. }
  64. }
  65. }(tmp)
  66. tmp = make(map[string]interface{})
  67. }
  68. wg_mgo.Wait()
  69. log.Debug("ai is over ...")
  70. }
  71. func test1() {
  72. arr := []string{
  73. "2024年3月4日 08:00:00",
  74. "2024年03月01日08时00分00秒",
  75. "2024年3月8日 17:00:00",
  76. "2024年03月08日09时30分00秒",
  77. "2024-01-24 09:15:00",
  78. "2024-02-29 00:00:00",
  79. "2024-02-23 00:00:00",
  80. "2024-03-05 09:30:00",
  81. "2024-02-21 09:30:00",
  82. "2024-02-28",
  83. "2024-02-28 09:00:00",
  84. "2024-03-01 00:00:00",
  85. "2024-02-06 17:30:00",
  86. "2024-03-01 00:00:00",
  87. "2024-02-06 17:30:00",
  88. "2024-02-29 00:00:00",
  89. "2024-02-29 00:00:00",
  90. "2024-02-29 00:00:00",
  91. "2024-02-28 08:30",
  92. "2024-02-29 19:56:29",
  93. "2024-03-29 10:30:00",
  94. "2024-03-29 10:30:00",
  95. "2024-01-12 00:00:00",
  96. "2024-03-23 09:00:00",
  97. "2024-02-28",
  98. "2024-02-28 09:00:00",
  99. "2024-02-02 10:00:00",
  100. "2024-02-28 00:00:00",
  101. "2024-02-27 00:00:00",
  102. "2024-02-28",
  103. "2024-02-29 15:30:00",
  104. "2024-02-29 12:00:00",
  105. "2024-02-28 09:00:00",
  106. "2024-02-27 09:00:00",
  107. "2024-02-27 09:00:00",
  108. "2024-02-29 00:00:00",
  109. "2024-02-29 00:00:00",
  110. "2024-01-30 09:30:00",
  111. "2024-02-27 09:00:00",
  112. "2024-02-27 09:30:00",
  113. "2024-03-01 00:00:00",
  114. "2024-02-28 09:00:00",
  115. "2024-02-27 09:00:00",
  116. "2024-02-28 09:00",
  117. "2024-02-26 08:30",
  118. "2024-02-27 09:00:00",
  119. "2024-02-27 09:30:00",
  120. "2024-02-28 09:00:00",
  121. "2024-02-27 09:00:00",
  122. "2024-02-28 09:00",
  123. "2024-02-26 08:30",
  124. "2024-03-05 09:30:00",
  125. "2024-02-27 17:54:28",
  126. "2024-03-01 10:00:00",
  127. "2024-03-01 10:00:00",
  128. "2024-02-08 00:00:00",
  129. "2024-03-01 09:00:00",
  130. "2024-03-01 09:00:00",
  131. "2024-03-08 19:30:00",
  132. "2024-03-08 19:30:00",
  133. "2024年3月4日 08:00:00",
  134. "2024年03月01日08时00分00秒",
  135. "2024年3月8日 17:00:00",
  136. "2024年03月08日09时30分00秒",
  137. "2024-03-07 17:00:00",
  138. "2024-03-07 17:00:00",
  139. "2024-02-26 08:30",
  140. "2024-02-28 09:00",
  141. "2024-02-27 09:00:00",
  142. "2024-02-28 09:00:00",
  143. "2024-02-27 09:30:00",
  144. "2024-02-27 09:00:00",
  145. "2024-01-30 09:30:00",
  146. "2024-02-29 00:00:00",
  147. "2024-02-29 00:00:00",
  148. "2024-02-27 09:00:00",
  149. "2024-02-27 09:00:00",
  150. "2024-02-28 09:00:00",
  151. "2024-02-29 12:00:00",
  152. "2024-02-29 16:00:00",
  153. "2024-02-29 15:30:00",
  154. "2024-02-28 00:00:00",
  155. "2024-02-27 00:00:00",
  156. "2024-02-28 00:00:00",
  157. "2024-02-02 10:00:00",
  158. "2024-02-28 09:00:00",
  159. "2024-02-28",
  160. "2024-03-23 09:00:00",
  161. "2024-01-12 00:00:00",
  162. "2024-03-29 10:30:00",
  163. "2024-03-29 10:30:00",
  164. "2024-02-29 19:56:29",
  165. "2024-02-28 08:30",
  166. "2024-02-29 00:00:00",
  167. "2024-02-29 00:00:00",
  168. "2024-02-29 00:00:00",
  169. "2024-02-06 17:30:00",
  170. "2024-03-01 00:00:00",
  171. "2024-02-06 17:30:00",
  172. "2024-03-01 00:00:00",
  173. "2024-02-28 09:00:00",
  174. "2024-02-28",
  175. "2024-02-21 09:30:00",
  176. "2024-03-05 09:30:00",
  177. "2024-02-23 00:00:00",
  178. "2024-02-29 00:00:00",
  179. "2024-01-24 09:15:00",
  180. "2024-03-06 14:00:00",
  181. "2024-03-06 00:00:00",
  182. "2024-03-07 23:59:00",
  183. }
  184. for _, v := range arr {
  185. nv := clean.CleanTime(v)
  186. log.Debug(v, "~~~", nv)
  187. }
  188. }
  189. func test2() {
  190. arr := []string{
  191. "ZFGG包件",
  192. "GC530100202400091001001",
  193. "GC532300202400053001001",
  194. "广州市白云区金沙街沙凤经济联合社留用地项目S2地块(金沙洲AB3708005地块)新建商业综合体全过程造价咨询服务",
  195. "一标段",
  196. "一标段",
  197. "E6111013556240003001001",
  198. "E6111013556240003001002",
  199. "一标段",
  200. "二标段",
  201. "一标段",
  202. "三标段",
  203. "二标段",
  204. "E61042235168rf556pk7001",
  205. "第三标段",
  206. "第二标段",
  207. "第三标段",
  208. "GC530500202400037001001",
  209. "第17标段",
  210. "第18标段",
  211. "1",
  212. "第23标段",
  213. "标段二",
  214. "标段二",
  215. "一标段",
  216. "一标段",
  217. "一标段",
  218. "e3100000151040117001001",
  219. "e3100000151027733043001",
  220. "e3100000151032001003001",
  221. "2024ANNGZ00007",
  222. "SYS20240206NTGC05002004",
  223. "SYS20240206NTGC05002010",
  224. "SYS20240206NTGC05002003",
  225. "E341522001002634001",
  226. "第1标段",
  227. "第三标段",
  228. "KJ06",
  229. "四川省遂宁市第一中学西校区第三期建设项目及附属工程(遂宁一中新校区改、扩建工程)主体工程增加建设内容-施工-1",
  230. "四川省遂宁市第一中学西校区第三期建设项目及附属工程(遂宁一中新校区改、扩建工程)增加建设内容-监理-1",
  231. "射洪市绿色农业产业带建设项目(一期)-监理-1",
  232. "HJFFJSZ2024020002001",
  233. "一标段",
  234. "E3702002313017682001001",
  235. "E3702002313017722001001",
  236. "1",
  237. "1",
  238. "标包一",
  239. "C01",
  240. "KJ06",
  241. "五标段",
  242. "二标段",
  243. "1标段",
  244. }
  245. for _, v := range arr {
  246. nv := clean.CleanOtherCode(v)
  247. log.Debug(nv, "~~~", v)
  248. }
  249. }