init.go 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398
  1. package main
  2. import (
  3. "log"
  4. "math"
  5. mu "mfw/util"
  6. "qfw/util"
  7. "regexp"
  8. "sort"
  9. "strings"
  10. "sync"
  11. "go.mongodb.org/mongo-driver/bson/primitive"
  12. )
  13. const (
  14. ProjectCache = "info" //存放每条项目信息,key为项目ID
  15. )
  16. var (
  17. Sysconfig map[string]interface{} //读取配置文件
  18. MongoTool *MongodbSim //mongodb连接
  19. ExtractColl, ProjectColl, BackupColl, SiteColl string //抽取表、项目表、项目快照表、站点表
  20. //NextNode []interface{}
  21. )
  22. var (
  23. //判断是日期
  24. _datereg = regexp.MustCompile("20[0-2][0-9][年-][0-9]{1,2}[月-][0-9]{1,2}[日-]([0-9]{1,2}时[0-9]{0,2})?")
  25. _numreg1 = regexp.MustCompile("^[0-9-]{1,8}$")
  26. _zimureg1 = regexp.MustCompile("^[a-zA-Z-]{1,7}$")
  27. _nzreg = regexp.MustCompile("^[0-9a-zA-Z-]+$")
  28. _hanreg = regexp.MustCompile(`^[\p{Han}::【】\\[\\]()()--、]+$`)
  29. replaceStr = regexp.MustCompile("(工程|采购|项目|[?!、【】()—()--]|栏标价|中标候选人|招标代理)")
  30. //判断带有分包、等特定词的
  31. pStr = regexp.MustCompile("(勘察|监理|施工|设计|验收|标段|分包|子包|[0-9A-Z]包|[一二三四五六七八九十0-9]批)")
  32. //判断包含数值
  33. nreg1 = regexp.MustCompile("[0-9]{2,}")
  34. //判断包含字母
  35. zreg1 = regexp.MustCompile("[a-zA-Z]{1,}")
  36. //判断包含汉字
  37. hreg1 = regexp.MustCompile(`[\p{Han}]+`)
  38. //判断项目编号是在10以内的纯数字结构
  39. numCheckPc = regexp.MustCompile("^[0-9-]{1,10}$")
  40. //仅初始化使用
  41. compareNoPass = map[string]bool{}
  42. compareAB = map[string]bool{}
  43. compareAB2D = map[string]bool{}
  44. compareABD = map[string]bool{}
  45. compareAB2CD = map[string]bool{}
  46. compareABCD = map[string]bool{}
  47. )
  48. func init() {
  49. util.ReadConfig(&Sysconfig)
  50. MongoTool = &MongodbSim{
  51. MongodbAddr: Sysconfig["mongodbServers"].(string),
  52. Size: util.IntAll(Sysconfig["mongodbPoolSize"]),
  53. DbName: Sysconfig["mongodbName"].(string),
  54. }
  55. MongoTool.InitPool()
  56. ExtractColl = Sysconfig["extractColl"].(string)
  57. ProjectColl = Sysconfig["projectColl"].(string)
  58. BackupColl = Sysconfig["backupColl"].(string)
  59. SiteColl = Sysconfig["siteColl"].(string)
  60. //NextNode = Sysconfig["nextNode"].([]interface{})
  61. udpport, _ := Sysconfig["udpport"].(string)
  62. udpclient = mu.UdpClient{Local: udpport, BufSize: 1024}
  63. udpclient.Listen(processUdpMsg)
  64. log.Println("Udp服务监听", udpport)
  65. //加载项目数据
  66. //---不能通过
  67. vm := []string{"C", "D"}
  68. for i := 0; i < 2; i++ {
  69. for j := 0; j < 2; j++ {
  70. for k := 0; k < 2; k++ {
  71. key := vm[i] + vm[j] + vm[k]
  72. compareNoPass[key] = true
  73. //fmt.Println(key)
  74. }
  75. }
  76. }
  77. //fmt.Println("-------------------")
  78. //三个元素一致 [AB][AB][AB],分值最高
  79. vm = []string{"A", "B"}
  80. for i := 0; i < 2; i++ {
  81. for j := 0; j < 2; j++ {
  82. for k := 0; k < 2; k++ {
  83. key := vm[i] + vm[j] + vm[k]
  84. compareAB[key] = true
  85. //fmt.Println(key)
  86. }
  87. }
  88. }
  89. //fmt.Println("-------------------", len(compareAB))
  90. //---至少两个一致,其他可能不存在
  91. //[AB][AB][ABD]
  92. //[AB][ABD][AB]
  93. vm = []string{"A", "B"}
  94. vm2 := []string{"A", "B", "D"}
  95. for i := 0; i < 2; i++ {
  96. for j := 0; j < 2; j++ {
  97. for k := 0; k < 3; k++ {
  98. key := vm[i] + vm[j] + vm2[k]
  99. if !compareAB[key] {
  100. compareAB2D[key] = true
  101. //fmt.Println(key)
  102. }
  103. }
  104. }
  105. }
  106. for i := 0; i < 2; i++ {
  107. for j := 0; j < 3; j++ {
  108. for k := 0; k < 2; k++ {
  109. key := vm[i] + vm2[j] + vm[k]
  110. if !compareAB[key] {
  111. compareAB2D[key] = true
  112. //fmt.Println(key)
  113. }
  114. }
  115. }
  116. }
  117. //fmt.Println("-------------------", len(compareAB2D))
  118. //---至少一个一致,其他可能不存在
  119. //[ABD][ABD][ABD] //已经删除DDD
  120. vm = []string{"A", "B", "D"}
  121. for i := 0; i < 3; i++ {
  122. for j := 0; j < 3; j++ {
  123. for k := 0; k < 3; k++ {
  124. key := vm[i] + vm[j] + vm[k]
  125. if !compareAB[key] && !compareAB2D[key] && !compareNoPass[key] {
  126. compareABD[key] = true
  127. //fmt.Println(key)
  128. }
  129. }
  130. }
  131. }
  132. //fmt.Println("-------------------", len(compareABD))
  133. //[AB][ABCD][AB]
  134. //[AB][AB][ABCD]
  135. vm = []string{"A", "B"}
  136. vm2 = []string{"A", "B", "C", "D"}
  137. for i := 0; i < 2; i++ {
  138. for j := 0; j < 4; j++ {
  139. for k := 0; k < 2; k++ {
  140. key := vm[i] + vm2[j] + vm[k]
  141. if !compareAB[key] && !compareAB2D[key] && !compareNoPass[key] && !compareABD[key] {
  142. compareAB2CD[key] = true
  143. //fmt.Println(key)
  144. }
  145. }
  146. }
  147. }
  148. for i := 0; i < 2; i++ {
  149. for j := 0; j < 2; j++ {
  150. for k := 0; k < 4; k++ {
  151. key := vm[i] + vm[j] + vm2[k]
  152. if !compareAB[key] && !compareAB2D[key] && !compareNoPass[key] && !compareABD[key] {
  153. compareAB2CD[key] = true
  154. //fmt.Println(key)
  155. }
  156. }
  157. }
  158. }
  159. //fmt.Println("-------------------", len(compareAB2CD))
  160. //[ABECD][ABECD][ABECD] //已经删除[CD][CD][CD] //这个要重点讨论
  161. vm = []string{"A", "B", "C", "D"}
  162. for i := 0; i < 4; i++ {
  163. for j := 0; j < 4; j++ {
  164. for k := 0; k < 4; k++ {
  165. key := vm[i] + vm[j] + vm[k]
  166. if !compareAB[key] && !compareAB2D[key] && !compareABD[key] && !compareNoPass[key] && !compareAB2CD[key] {
  167. compareABCD[key] = true
  168. //fmt.Println(key)
  169. }
  170. }
  171. }
  172. }
  173. }
  174. func CheckHanAndNum(str string) (b bool) {
  175. return nreg1.MatchString(str) && hreg1.MatchString(str)
  176. }
  177. func CheckZimuAndNum(str string) (b bool) {
  178. return zreg1.MatchString(str) && nreg1.MatchString(str)
  179. }
  180. type KeyMap struct {
  181. Lock sync.Mutex
  182. Map map[string]*Key
  183. }
  184. type ID struct {
  185. Id string
  186. Lock sync.Mutex
  187. P *ProjectInfo
  188. }
  189. type Key struct {
  190. Arr []string
  191. Lock sync.Mutex
  192. }
  193. type IdAndLock struct {
  194. Id string
  195. Lock sync.Mutex
  196. }
  197. func NewKeyMap() *KeyMap {
  198. return &KeyMap{
  199. Map: map[string]*Key{},
  200. Lock: sync.Mutex{},
  201. }
  202. }
  203. //招标信息实体类
  204. type Info struct {
  205. Id string `json:"_id"`
  206. Href string `json:"href"` //源地址
  207. Publishtime int64 `json:"publishtime"`
  208. Comeintime int64 `json:"comeintime"`
  209. Title string `json:"title"`
  210. TopType string `json:"toptype"`
  211. SubType string `json:"subtype"`
  212. ProjectName string `json:"projectname"`
  213. ProjectCode string `json:"projectcode"`
  214. ProjectScope string `json:"projectscope"`
  215. ContractCode string `json:"contractcode"`
  216. Buyer string `json:"buyer"`
  217. Buyerperson string `json:"buyerperson"`
  218. Buyertel string `json:"buyertel"`
  219. Agency string `json:"agency"`
  220. Area string `json:"area"`
  221. City string `json:"city"`
  222. District string `json:"district"`
  223. Infoformat int `json:"infoformat"`
  224. HasPackage bool // `json:"haspackage"`
  225. Package map[string]interface{} `json:"package"`
  226. //PNum string `json:"pnum"`
  227. Topscopeclass []string `json:"topscopeclass"`
  228. Subscopeclass []string `json:"subscopeclass"`
  229. Buyerclass string `json:"buyerclass"`
  230. Bidopentime int64 `json:"bidopentime"`
  231. Budget float64 `json:"budget"`
  232. Bidamount float64 `json:"bidamount"`
  233. Winners []string
  234. dealtype int
  235. Winnerorder []string
  236. PTC string //从标题中抽的项目编号
  237. pnbval int //项目名称、编号、采购单位存在的个数
  238. LenPC int //项目编号长度
  239. LenPN int //项目名称长度
  240. LenPTC int //标题抽的项目编号长度
  241. //以下三个元素做对比,计算包含时候使用
  242. PNBH int //0初始,+包含,-被包含
  243. PCBH int
  244. PTCBH int
  245. }
  246. //项目实体类
  247. type ProjectInfo struct {
  248. Id primitive.ObjectID `json:"_id"`
  249. FirstTime int64 `json:"firsttime,omitempty"` //项目的最早时间
  250. LastTime int64 `json:"lasttime,omitempty"` //项目的最后时间
  251. Ids []string `json:"ids,omitempty"`
  252. Topscopeclass []string `json:"topscopeclass,omitempty"`
  253. Subscopeclass []string `json:"subscopeclass,omitempty"` //子行业分类
  254. Winners []string `json:"winners,omitempty"` //中标人
  255. ProjectName string `json:"projectname,omitempty"` //项目名称
  256. ProjectCode string `json:"projectcode,omitempty"` //项目代码唯一(纯数字的权重低)
  257. Buyer string `json:"buyer,omitempty"` //采购单位唯一
  258. MPN []string `json:"mpn,omitempty"` //合并后多余的项目名称
  259. MPC []string `json:"mpc,omitempty"` //合并后多余的项目编号
  260. Buyerperson string `json:"buyerperson"` //采购联系人
  261. Buyertel string `json:"buyertel"` //采购联系人电话
  262. Agency string `json:"agency"` //代理机构
  263. Area string `json:"area"` //地区
  264. City string `json:"city"` //地市
  265. District string `json:"district"` //区县
  266. Bidstatus string `json:"bidstatus"` //
  267. Bidtype string `json:"bidtype"` //
  268. //HasPackage bool `json:"haspackage"` //是否有分包
  269. Package map[string]interface{} `json:"package,omitempty"` //分包的对比对象
  270. Buyerclass string `json:"buyerclass"` //采购单位分类
  271. Bidopentime int64 `json:"bidopentime,omitempty"` //开标时间
  272. // Zbtime int64 `json:"zbtime"` //招标时间
  273. // Jgtime int64 `json:"jgtime"` //结果中标时间
  274. Bidamount float64 `json:"bidamount,omitempty"` //中标金额
  275. Budget float64 `json:"budget,omitempty"` //预算
  276. //Winnerorder []string `json:"winnerorder"` //中标候选人
  277. score int
  278. comStr string
  279. resVal, pjVal int
  280. InfoFiled map[string]InfoFiled
  281. }
  282. //存储部分招标信息字段,业务逻辑处理需要
  283. type InfoFiled struct {
  284. budget float64
  285. bidamount float64
  286. contractCode string
  287. projectName string
  288. projectCode string
  289. bidstatus string
  290. }
  291. //站点信息
  292. type Site struct {
  293. Id string `json:"_id"`
  294. Site string `json:"site"` //站点名字
  295. Area string `json:"area"` //省
  296. City string `json:"city"` //市
  297. District string `json:"district"` //区、县
  298. Domain string `json:"domain"` //地址
  299. }
  300. //二分字符串查找
  301. func BinarySearch(s []string, k string) int {
  302. sort.Strings(s)
  303. lo, hi := 0, len(s)-1
  304. for lo <= hi {
  305. m := (lo + hi) >> 1
  306. if s[m] < k {
  307. lo = m + 1
  308. } else if s[m] > k {
  309. hi = m - 1
  310. } else {
  311. return m
  312. }
  313. }
  314. return -1
  315. }
  316. //计算文本相似度
  317. func CosineSimilar(srcWords1, dstWords1 string) float64 {
  318. srcWords, dstWords := strings.Split(srcWords1, ""), strings.Split(dstWords1, "")
  319. // get all words
  320. allWordsMap := make(map[string]int, 0)
  321. for _, word := range srcWords {
  322. if _, found := allWordsMap[word]; !found {
  323. allWordsMap[word] = 1
  324. } else {
  325. allWordsMap[word] += 1
  326. }
  327. }
  328. for _, word := range dstWords {
  329. if _, found := allWordsMap[word]; !found {
  330. allWordsMap[word] = 1
  331. } else {
  332. allWordsMap[word] += 1
  333. }
  334. }
  335. // stable the sort
  336. allWordsSlice := make([]string, 0)
  337. for word, _ := range allWordsMap {
  338. allWordsSlice = append(allWordsSlice, word)
  339. }
  340. // assemble vector
  341. srcVector := make([]int, len(allWordsSlice))
  342. dstVector := make([]int, len(allWordsSlice))
  343. for _, word := range srcWords {
  344. if index := BinarySearch(allWordsSlice, word); index != -1 {
  345. srcVector[index] += 1
  346. }
  347. }
  348. for _, word := range dstWords {
  349. if index := BinarySearch(allWordsSlice, word); index != -1 {
  350. dstVector[index] += 1
  351. }
  352. }
  353. // calc cos
  354. numerator := float64(0)
  355. srcSq := 0
  356. dstSq := 0
  357. for i, srcCount := range srcVector {
  358. dstCount := dstVector[i]
  359. numerator += float64(srcCount * dstCount)
  360. srcSq += srcCount * srcCount
  361. dstSq += dstCount * dstCount
  362. }
  363. denominator := math.Sqrt(float64(srcSq * dstSq))
  364. v1 := numerator / denominator
  365. // if v1 > 0.6 {
  366. // log.Println(v1, srcWords1, dstWords1)
  367. // }
  368. return v1
  369. }