odbcimporter.go 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429
  1. package importer
  2. import (
  3. "encoding/json"
  4. "fmt"
  5. "regexp"
  6. "strings"
  7. "sync"
  8. "sync/atomic"
  9. "time"
  10. "git.wecise.com/wecise/cgimport/graph"
  11. "git.wecise.com/wecise/cgimport/odbc"
  12. "git.wecise.com/wecise/cgimport/schema"
  13. "git.wecise.com/wecise/odb-go/odb"
  14. "git.wecise.com/wecise/util/cast"
  15. "git.wecise.com/wecise/util/cmap"
  16. "git.wecise.com/wecise/util/merrs"
  17. "github.com/dgryski/go-farm"
  18. "github.com/scylladb/go-set/strset"
  19. )
  20. type classdatainfo struct {
  21. *schema.ClassInfo
  22. insertcount int64
  23. lastlogtime time.Time
  24. lastlogicount int64
  25. mutex sync.Mutex
  26. }
  27. var classdatainfos = cmap.NewSingle[string, *classdatainfo]()
  28. type ODBCImporter struct {
  29. client odb.Client
  30. }
  31. func NewODBCImporter() *ODBCImporter {
  32. odbci := &ODBCImporter{}
  33. if odbc.DevPhase&(odbc.DP_CREATECLASS|odbc.DP_INSERTDATA) != 0 {
  34. odbci.client = odbc.ODBC()
  35. }
  36. return odbci
  37. }
  38. // 根据数据修正类定义
  39. func (odbci *ODBCImporter) ReviseClassStruct() (err error) {
  40. for _, classname := range schema.ClassNames {
  41. ci := schema.ClassInfos.GetIFPresent(classname)
  42. if ci == nil {
  43. return merrs.NewError("classinfo not found " + classname)
  44. }
  45. cdi, e := classdatainfos.GetWithNew(ci.Classaliasname, func() (cdi *classdatainfo, err error) {
  46. if odbci.client != nil {
  47. _, e := odbci.client.Query("select class,id from " + ci.Classfullname + " limit 1").Do()
  48. if e != nil {
  49. if !strings.Contains(e.Error(), "not find") {
  50. return nil, e
  51. }
  52. logger.Info("create class " + ci.Classfullname)
  53. _, e = odbci.client.Query(ci.Createmql).Do()
  54. if e != nil {
  55. return nil, e
  56. }
  57. }
  58. }
  59. cdi = &classdatainfo{ClassInfo: ci}
  60. return
  61. })
  62. if e != nil {
  63. return e
  64. }
  65. classdatainfos.Set(ci.Classfullname, cdi)
  66. }
  67. if odbci.client != nil {
  68. for _, createedgemql := range schema.CreateEdgeMqls {
  69. _, e := odbci.client.Query(createedgemql).Do()
  70. if e != nil && !strings.Contains(e.Error(), "already exist") {
  71. return e
  72. }
  73. }
  74. }
  75. return
  76. }
  77. // func (odbci *ODBCImporter) InsertEdge(data map[string]any) (err error) {
  78. // ei, e := graph.ParseEdgeInfo(data)
  79. // if e != nil {
  80. // return e
  81. // }
  82. // if odbci.client != nil {
  83. // // foid := get_object_id_from_cache("level1", fromuid)
  84. // // toid := to_object_id("level1", touid)
  85. // // eabs, _ := json.Marshal(extraattr)
  86. // // quadmql := `quad "` + foid + `" ` + edgetype + ` + "` + toid + `" ` + string(eabs)
  87. // // _, err = odbci.client.Query(quadmql).Do()
  88. // // if err != nil {
  89. // // err = merrs.NewError(err, merrs.SSMaps{{"mql": quadmql}})
  90. // // logger.Error(err)
  91. // // return
  92. // // }
  93. // updatemql := "update " + "level1" + " set " + " contain=contain+?" + " where id='" + foid + "'"
  94. // _, err = odbci.client.Query(updatemql, map[string][]string{
  95. // "_all": {toid},
  96. // toid: {string(eabs)},
  97. // }).Do()
  98. // if err != nil {
  99. // err = merrs.NewError(err, merrs.SSMaps{{"mql": updatemql}})
  100. // return
  101. // }
  102. // logger.Info("relation immport " + foid + "->" + toid)
  103. // }
  104. // return
  105. // }
  106. // var cm_object_id_cache = cmap.New[string, chan string]()
  107. // func object_id_cache(classaliasname, suid string) chan string {
  108. // choid, _ := cm_object_id_cache.GetWithNew(classaliasname+":"+suid,
  109. // func() (chan string, error) {
  110. // ch := make(chan string, 2)
  111. // return ch, nil
  112. // })
  113. // return choid
  114. // }
  115. // func get_object_id_from_cache(classaliasname, suid string) string {
  116. // choid := object_id_cache(classaliasname, suid)
  117. // oid := <-choid
  118. // push_object_id_into_cache(choid, oid)
  119. // return oid
  120. // }
  121. // func push_object_id_into_cache(choid chan string, oid string) {
  122. // choid <- oid
  123. // if len(choid) == 2 {
  124. // // 最多保留 1 个
  125. // // chan cap = 2,第三个元素进不来
  126. // // 进第二个元素的协程,清除第一个元素,允许其它协程后续进入新元素
  127. // <-choid
  128. // }
  129. // }
  130. var masterlevel1mutex = make([]sync.Mutex, 256)
  131. var masterdatas = cmap.New[string, map[string]any]()
  132. var level1datas = cmap.New[string, map[string]any]()
  133. func (odbci *ODBCImporter) masterlevel1data(classaliasname string, suid string, data map[string]any) error {
  134. hidx := int(farm.Hash32([]byte(suid)) % uint32(len(masterlevel1mutex)))
  135. masterlevel1mutex[hidx].Lock()
  136. defer masterlevel1mutex[hidx].Unlock()
  137. switch classaliasname {
  138. case "master":
  139. level1data := level1datas.GetIFPresent(suid)
  140. if level1data == nil {
  141. // 先插入 master
  142. masterdatas.Set(suid, data)
  143. } else {
  144. // 后插入 master
  145. level1datas.Remove(suid)
  146. // 用 master 补齐 level1 数据
  147. // data 数据不能变,需要后续插入 master
  148. for k, v := range data {
  149. if _, has := level1data[k]; !has {
  150. level1data[k] = v
  151. }
  152. }
  153. // 重新插入完整的 level1
  154. e := odbci.insertData("level1", level1data)
  155. if e != nil {
  156. return e
  157. }
  158. }
  159. case "level1":
  160. masterdata := masterdatas.GetIFPresent(suid)
  161. if masterdata == nil {
  162. // 先插入 level 1
  163. level1datas.Set(suid, data)
  164. } else {
  165. // 后插入 level1
  166. masterdatas.Remove(suid)
  167. // 用 level1 补齐 master 数据
  168. for k, v := range data {
  169. masterdata[k] = v
  170. }
  171. // 完整 level1 数据
  172. data = masterdata
  173. }
  174. // 插入 level1 数据
  175. e := odbci.insertData("level1", data)
  176. if e != nil {
  177. return e
  178. }
  179. }
  180. return nil
  181. }
  182. // 插入数据
  183. func (odbci *ODBCImporter) InsertData(classaliasname string, data map[string]any) (err error) {
  184. oid, suid, e := graph.GetNodeId(classaliasname, data)
  185. if e != nil {
  186. return e
  187. }
  188. data["id"] = oid
  189. if classaliasname == "master" {
  190. e := odbci.masterlevel1data(classaliasname, suid, data)
  191. if e != nil {
  192. return e
  193. }
  194. } else if classaliasname == "level1" {
  195. ei := graph.GetEdgeInfo(oid)
  196. if ei != nil {
  197. data["contain"] = ei["contain"]
  198. data["depend"] = ei["depend"]
  199. data["topology"] = ei["topology"]
  200. }
  201. e := odbci.masterlevel1data(classaliasname, suid, data)
  202. if e != nil {
  203. return e
  204. }
  205. // 数据已经在 masterlevel1data 中插入完成
  206. return
  207. } else {
  208. data["depend"] = referencedata(classaliasname, data)
  209. }
  210. return odbci.insertData(classaliasname, data)
  211. }
  212. type InnerData struct {
  213. oid string
  214. suid string
  215. contain map[string][]string
  216. depend map[string][]string
  217. topology map[string][]string
  218. }
  219. func referencedata(classaliasname string, data map[string]any) (depend map[string][]string) {
  220. refer := data["_references"]
  221. switch vv := refer.(type) {
  222. case []interface{}:
  223. for _, v := range vv {
  224. switch vv := v.(type) {
  225. case map[string]interface{}:
  226. for k, v := range vv {
  227. switch k {
  228. case "_edgeType":
  229. case "_toUniqueId":
  230. suid := cast.ToString(v)
  231. toclassname := "master"
  232. switch classaliasname {
  233. case "level1":
  234. toclassname = "level1"
  235. case "level2":
  236. toclassname = "level1"
  237. case "level3":
  238. toclassname = "level2"
  239. case "level4":
  240. toclassname = "level3"
  241. case "level5":
  242. toclassname = "level4"
  243. case "level6":
  244. toclassname = "level5"
  245. case "level7":
  246. toclassname = "level6"
  247. case "level8":
  248. toclassname = "level7"
  249. }
  250. toid := graph.ToNodeId(toclassname, suid)
  251. m := map[string]string{"_direction": "out"}
  252. mbs, _ := json.Marshal(m)
  253. depend = map[string][]string{
  254. "_all": {toid},
  255. toid: {string(mbs)},
  256. }
  257. }
  258. }
  259. }
  260. }
  261. }
  262. return
  263. }
  264. func (odbci *ODBCImporter) insertData(classaliasname string, data map[string]any) (err error) {
  265. cdi := classdatainfos.GetIFPresent(classaliasname)
  266. if cdi == nil {
  267. return merrs.NewError("class not defined " + classaliasname)
  268. }
  269. if cdi.Insertmql == "" {
  270. return merrs.NewError("class no fields to insert " + classaliasname)
  271. }
  272. values := []any{}
  273. for _, fn := range cdi.Fieldslist {
  274. fi := cdi.Fieldinfos[fn]
  275. if fi == nil {
  276. values = append(values, data[fn])
  277. continue
  278. }
  279. // 合并扩展字段
  280. if strset.New(fi.Datakey...).Has("*") {
  281. if fi.Fieldtype != "map<varchar,varchar>" {
  282. return merrs.NewError("fi.Fieldtype=" + fi.Fieldtype + " != map<varchar,varchar>")
  283. }
  284. td := map[string]any{}
  285. for k, v := range data {
  286. if cdi.DatakeyFieldinfos[k] == nil {
  287. td[k] = cast.ToString(v)
  288. }
  289. }
  290. values = append(values, td)
  291. continue
  292. }
  293. // 字段类型修正
  294. var v any
  295. for _, dk := range fi.Datakey {
  296. v = data[dk]
  297. if v != nil {
  298. switch fi.Fieldtype {
  299. case "set<varchar>":
  300. v = cast.ToStringSlice(v)
  301. case "timestamp":
  302. tv, e := cast.ToDateTimeE(v, "2006-01-02-15.04.05.000000")
  303. if e != nil {
  304. return merrs.NewError(fmt.Sprint("can't parse datetime value '", v, "'"))
  305. }
  306. v = tv.Format("2006-01-02 15:04:05.000000")
  307. }
  308. break
  309. }
  310. }
  311. if fn == "tags" {
  312. v = append(cast.ToStringSlice(v), classaliasname)
  313. }
  314. values = append(values, v)
  315. }
  316. if odbci.client != nil {
  317. // logger.Info(values...)
  318. _, err = odbci.client.Query(cdi.Insertmql, values...).Do()
  319. if err != nil {
  320. databs, _ := json.MarshalIndent(data, "", " ")
  321. err = merrs.NewError(err, merrs.SSMaps{{"mql": cdi.Insertmql}, {"values": fmt.Sprint(values)}, {"data": string(databs)}})
  322. logger.Error(err)
  323. return
  324. }
  325. }
  326. atomic.AddInt64(&cdi.insertcount, 1)
  327. cdi.mutex.Lock()
  328. if time.Since(cdi.lastlogtime) > 5*time.Second && cdi.lastlogicount != cdi.insertcount {
  329. cdi.lastlogtime = time.Now()
  330. cdi.lastlogicount = cdi.insertcount
  331. logger.Info("class", cdi.Classfullname, "import", cdi.insertcount, "records")
  332. }
  333. cdi.mutex.Unlock()
  334. return
  335. }
  336. func (odbci *ODBCImporter) done() {
  337. classdatainfos.Fetch(func(cn string, cdi *classdatainfo) bool {
  338. cdi.mutex.Lock()
  339. if cdi.lastlogicount != cdi.insertcount {
  340. cdi.lastlogtime = time.Now()
  341. cdi.lastlogicount = cdi.insertcount
  342. logger.Info("class", cdi.Classfullname, "import", cdi.insertcount, "records")
  343. }
  344. cdi.mutex.Unlock()
  345. return true
  346. })
  347. }
  348. func (odbci *ODBCImporter) alldone() {
  349. classdatainfos.Fetch(func(cn string, cdi *classdatainfo) bool {
  350. cdi.mutex.Lock()
  351. if cdi.insertcount != 0 {
  352. cdi.lastlogtime = time.Now()
  353. cdi.lastlogicount = cdi.insertcount
  354. logger.Info("class", cdi.Classfullname, "import", cdi.insertcount, "records")
  355. }
  356. cdi.mutex.Unlock()
  357. return true
  358. })
  359. }
  360. func (odbci *ODBCImporter) reload() error {
  361. if odbci.client != nil {
  362. for i := len(schema.ClassNames) - 1; i >= 0; i-- {
  363. classname := schema.ClassNames[i]
  364. ci := schema.ClassInfos.GetIFPresent(classname)
  365. if ci == nil {
  366. continue
  367. }
  368. e := odbci.dropclass(ci.Classfullname)
  369. if e != nil {
  370. return e
  371. }
  372. }
  373. }
  374. return nil
  375. }
  376. func (odbci *ODBCImporter) dropclass(classnames ...string) error {
  377. for _, classname := range classnames {
  378. for retry := 2; retry >= 0; retry-- {
  379. _, e := odbci.client.Query(`delete from "` + classname + `" with version`).Do()
  380. _ = e
  381. _, e = odbci.client.Query(`drop class if exists "` + classname + `"`).Do()
  382. if e != nil {
  383. matchstr := regexp.MustCompile(`refer by ([^,]+)`).FindStringSubmatch(e.Error())
  384. if len(matchstr) >= 2 {
  385. e = odbci.dropclass(matchstr[1])
  386. if e != nil {
  387. return e
  388. }
  389. } else {
  390. matchstr := regexp.MustCompile(`has children \[([^\]]+)\]`).FindStringSubmatch(e.Error())
  391. if len(matchstr) >= 2 {
  392. e = odbci.dropclass(strings.Split(matchstr[1], ",")...)
  393. if e != nil {
  394. return e
  395. }
  396. }
  397. }
  398. if retry > 0 {
  399. continue
  400. }
  401. return e
  402. }
  403. }
  404. logger.Info("drop class " + classname)
  405. }
  406. return nil
  407. }