init.go 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386
  1. package main
  2. import (
  3. "log"
  4. "math"
  5. mu "mfw/util"
  6. "qfw/util"
  7. //"qfw/util/mongodb"
  8. "regexp"
  9. "sort"
  10. "strings"
  11. "sync"
  12. //"gopkg.in/mgo.v2/bson"
  13. //"go.mongodb.org/mongo-driver/bson"
  14. "go.mongodb.org/mongo-driver/bson/primitive"
  15. )
  16. const (
  17. ProjectCache = "info" //存放每条项目信息,key为项目ID
  18. )
  19. var (
  20. Sysconfig map[string]interface{} //读取配置文件
  21. MongoTool *MongodbSim //mongodb连接
  22. ExtractColl, ProjectColl, BackupColl, SiteColl string //抽取表、项目表、项目快照表、站点表
  23. //NextNode []interface{}
  24. )
  25. var (
  26. //判断是日期
  27. _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})?")
  28. _numreg1 = regexp.MustCompile("^[0-9-]{1,8}$")
  29. _zimureg1 = regexp.MustCompile("^[a-zA-Z-]{1,7}$")
  30. _nzreg = regexp.MustCompile("^[0-9a-zA-Z-]+$")
  31. _hanreg = regexp.MustCompile(`^[\p{Han}::【】\\[\\]()()--、]+$`)
  32. replaceStr = regexp.MustCompile("(工程|采购|项目|[?!、【】()—()--]|栏标价|中标候选人|招标代理)")
  33. //判断带有分包、等特定词的
  34. 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)")
  35. //判断包含数值
  36. nreg1 = regexp.MustCompile("[0-9]{2,}")
  37. //判断包含字母
  38. zreg1 = regexp.MustCompile("[a-zA-Z]{1,}")
  39. //判断包含汉字
  40. hreg1 = regexp.MustCompile(`[\p{Han}]+`)
  41. //判断项目编号是在10以内的纯数字结构
  42. numCheckPc = regexp.MustCompile("^[0-9-]{1,10}$")
  43. //仅初始化使用
  44. compareNoPass = map[string]bool{}
  45. compareAB = map[string]bool{}
  46. compareAB2D = map[string]bool{}
  47. compareABD = map[string]bool{}
  48. compareAB2CD = map[string]bool{}
  49. compareABCD = map[string]bool{}
  50. )
  51. func init() {
  52. util.ReadConfig(&Sysconfig)
  53. MongoTool = &MongodbSim{
  54. MongodbAddr: Sysconfig["mongodbServers"].(string),
  55. Size: util.IntAll(Sysconfig["mongodbPoolSize"]),
  56. DbName: Sysconfig["mongodbName"].(string),
  57. }
  58. MongoTool.InitPool()
  59. ExtractColl = Sysconfig["extractColl"].(string)
  60. ProjectColl = Sysconfig["projectColl"].(string)
  61. BackupColl = Sysconfig["backupColl"].(string)
  62. SiteColl = Sysconfig["siteColl"].(string)
  63. //NextNode = Sysconfig["nextNode"].([]interface{})
  64. udpport, _ := Sysconfig["udpport"].(string)
  65. udpclient = mu.UdpClient{Local: udpport, BufSize: 1024}
  66. udpclient.Listen(processUdpMsg)
  67. log.Println("Udp服务监听", udpport)
  68. //加载项目数据
  69. //---不能通过
  70. vm := []string{"C", "D"}
  71. for i := 0; i < 2; i++ {
  72. for j := 0; j < 2; j++ {
  73. for k := 0; k < 2; k++ {
  74. key := vm[i] + vm[j] + vm[k]
  75. compareNoPass[key] = true
  76. //fmt.Println(key)
  77. }
  78. }
  79. }
  80. //fmt.Println("-------------------")
  81. //三个元素一致 [AB][AB][AB],分值最高
  82. vm = []string{"A", "B"}
  83. for i := 0; i < 2; i++ {
  84. for j := 0; j < 2; j++ {
  85. for k := 0; k < 2; k++ {
  86. key := vm[i] + vm[j] + vm[k]
  87. compareAB[key] = true
  88. //fmt.Println(key)
  89. }
  90. }
  91. }
  92. //fmt.Println("-------------------", len(compareAB))
  93. //---至少两个一致,其他可能不存在
  94. //[AB][AB][ABD]
  95. //[AB][ABD][AB]
  96. vm = []string{"A", "B"}
  97. vm2 := []string{"A", "B", "D"}
  98. for i := 0; i < 2; i++ {
  99. for j := 0; j < 2; j++ {
  100. for k := 0; k < 3; k++ {
  101. key := vm[i] + vm[j] + vm2[k]
  102. if !compareAB[key] {
  103. compareAB2D[key] = true
  104. //fmt.Println(key)
  105. }
  106. }
  107. }
  108. }
  109. for i := 0; i < 2; i++ {
  110. for j := 0; j < 3; j++ {
  111. for k := 0; k < 2; k++ {
  112. key := vm[i] + vm2[j] + vm[k]
  113. if !compareAB[key] {
  114. compareAB2D[key] = true
  115. //fmt.Println(key)
  116. }
  117. }
  118. }
  119. }
  120. //fmt.Println("-------------------", len(compareAB2D))
  121. //---至少一个一致,其他可能不存在
  122. //[ABD][ABD][ABD] //已经删除DDD
  123. vm = []string{"A", "B", "D"}
  124. for i := 0; i < 3; i++ {
  125. for j := 0; j < 3; j++ {
  126. for k := 0; k < 3; k++ {
  127. key := vm[i] + vm[j] + vm[k]
  128. if !compareAB[key] && !compareAB2D[key] && !compareNoPass[key] {
  129. compareABD[key] = true
  130. //fmt.Println(key)
  131. }
  132. }
  133. }
  134. }
  135. //fmt.Println("-------------------", len(compareABD))
  136. //[AB][ABCD][AB]
  137. //[AB][AB][ABCD]
  138. vm = []string{"A", "B"}
  139. vm2 = []string{"A", "B", "C", "D"}
  140. for i := 0; i < 2; i++ {
  141. for j := 0; j < 4; j++ {
  142. for k := 0; k < 2; k++ {
  143. key := vm[i] + vm2[j] + vm[k]
  144. if !compareAB[key] && !compareAB2D[key] && !compareNoPass[key] && !compareABD[key] {
  145. compareAB2CD[key] = true
  146. //fmt.Println(key)
  147. }
  148. }
  149. }
  150. }
  151. for i := 0; i < 2; i++ {
  152. for j := 0; j < 2; j++ {
  153. for k := 0; k < 4; k++ {
  154. key := vm[i] + vm[j] + vm2[k]
  155. if !compareAB[key] && !compareAB2D[key] && !compareNoPass[key] && !compareABD[key] {
  156. compareAB2CD[key] = true
  157. //fmt.Println(key)
  158. }
  159. }
  160. }
  161. }
  162. //fmt.Println("-------------------", len(compareAB2CD))
  163. //[ABECD][ABECD][ABECD] //已经删除[CD][CD][CD] //这个要重点讨论
  164. vm = []string{"A", "B", "C", "D"}
  165. for i := 0; i < 4; i++ {
  166. for j := 0; j < 4; j++ {
  167. for k := 0; k < 4; k++ {
  168. key := vm[i] + vm[j] + vm[k]
  169. if !compareAB[key] && !compareAB2D[key] && !compareABD[key] && !compareNoPass[key] && !compareAB2CD[key] {
  170. compareABCD[key] = true
  171. //fmt.Println(key)
  172. }
  173. }
  174. }
  175. }
  176. }
  177. func CheckHanAndNum(str string) (b bool) {
  178. return nreg1.MatchString(str) && hreg1.MatchString(str)
  179. }
  180. func CheckZimuAndNum(str string) (b bool) {
  181. return zreg1.MatchString(str) && nreg1.MatchString(str)
  182. }
  183. type KeyMap struct {
  184. Lock sync.Mutex
  185. Map map[string]*Key
  186. }
  187. type ID struct {
  188. Id string
  189. Lock sync.Mutex
  190. P *ProjectInfo
  191. }
  192. type Key struct {
  193. Arr []string
  194. Lock sync.Mutex
  195. }
  196. type IdAndLock struct {
  197. Id string
  198. Lock sync.Mutex
  199. }
  200. func NewKeyMap() *KeyMap {
  201. return &KeyMap{
  202. Map: map[string]*Key{},
  203. Lock: sync.Mutex{},
  204. }
  205. }
  206. //招标信息实体类
  207. type Info struct {
  208. Id string `json:"_id"`
  209. Href string `json:"href"` //源地址
  210. Publishtime int64 `json:"publishtime"`
  211. Title string `json:"title"`
  212. TopType string `json:"toptype"`
  213. SubType string `json:"subtype"`
  214. ProjectName string `json:"projectname"`
  215. ProjectCode string `json:"projectcode"`
  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. }
  281. type Site struct {
  282. Id string `json:"_id"`
  283. Site string `json:"site"` //站点名字
  284. Area string `json:"area"` //省
  285. City string `json:"city"` //市
  286. District string `json:"district"` //区、县
  287. Domain string `json:"domain"` //地址
  288. }
  289. //二分字符串查找
  290. func BinarySearch(s []string, k string) int {
  291. sort.Strings(s)
  292. lo, hi := 0, len(s)-1
  293. for lo <= hi {
  294. m := (lo + hi) >> 1
  295. if s[m] < k {
  296. lo = m + 1
  297. } else if s[m] > k {
  298. hi = m - 1
  299. } else {
  300. return m
  301. }
  302. }
  303. return -1
  304. }
  305. //计算文本相似度
  306. func CosineSimilar(srcWords1, dstWords1 string) float64 {
  307. srcWords, dstWords := strings.Split(srcWords1, ""), strings.Split(dstWords1, "")
  308. // get all words
  309. allWordsMap := make(map[string]int, 0)
  310. for _, word := range srcWords {
  311. if _, found := allWordsMap[word]; !found {
  312. allWordsMap[word] = 1
  313. } else {
  314. allWordsMap[word] += 1
  315. }
  316. }
  317. for _, word := range dstWords {
  318. if _, found := allWordsMap[word]; !found {
  319. allWordsMap[word] = 1
  320. } else {
  321. allWordsMap[word] += 1
  322. }
  323. }
  324. // stable the sort
  325. allWordsSlice := make([]string, 0)
  326. for word, _ := range allWordsMap {
  327. allWordsSlice = append(allWordsSlice, word)
  328. }
  329. // assemble vector
  330. srcVector := make([]int, len(allWordsSlice))
  331. dstVector := make([]int, len(allWordsSlice))
  332. for _, word := range srcWords {
  333. if index := BinarySearch(allWordsSlice, word); index != -1 {
  334. srcVector[index] += 1
  335. }
  336. }
  337. for _, word := range dstWords {
  338. if index := BinarySearch(allWordsSlice, word); index != -1 {
  339. dstVector[index] += 1
  340. }
  341. }
  342. // calc cos
  343. numerator := float64(0)
  344. srcSq := 0
  345. dstSq := 0
  346. for i, srcCount := range srcVector {
  347. dstCount := dstVector[i]
  348. numerator += float64(srcCount * dstCount)
  349. srcSq += srcCount * srcCount
  350. dstSq += dstCount * dstCount
  351. }
  352. denominator := math.Sqrt(float64(srcSq * dstSq))
  353. v1 := numerator / denominator
  354. // if v1 > 0.6 {
  355. // log.Println(v1, srcWords1, dstWords1)
  356. // }
  357. return v1
  358. }