join_operator.go 9.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354
  1. package operators
  2. import (
  3. "fmt"
  4. "github.com/emqx/kuiper/xsql"
  5. "github.com/emqx/kuiper/xstream/api"
  6. )
  7. //TODO join expr should only be the equal op between 2 streams like tb1.id = tb2.id
  8. type JoinOp struct {
  9. From *xsql.Table
  10. Joins xsql.Joins
  11. }
  12. // input: xsql.WindowTuplesSet from windowOp, window is required for join
  13. // output: xsql.JoinTupleSets
  14. func (jp *JoinOp) Apply(ctx api.StreamContext, data interface{}, fv *xsql.FunctionValuer, _ *xsql.AggregateFunctionValuer) interface{} {
  15. log := ctx.GetLogger()
  16. var input xsql.WindowTuplesSet
  17. switch v := data.(type) {
  18. case error:
  19. return v
  20. case xsql.WindowTuplesSet:
  21. input = v
  22. log.Debugf("join plan receive %v", data)
  23. default:
  24. return fmt.Errorf("run Join error: join is only supported in window")
  25. }
  26. result := xsql.JoinTupleSets{}
  27. for i, join := range jp.Joins {
  28. if i == 0 {
  29. v, err := jp.evalSet(input, join, fv)
  30. if err != nil {
  31. return fmt.Errorf("run Join error: %s", err)
  32. }
  33. result = v
  34. } else {
  35. r1, err := jp.evalJoinSets(&result, input, join, fv)
  36. if err != nil {
  37. return fmt.Errorf("run Join error: %s", err)
  38. }
  39. if v1, ok := r1.(xsql.JoinTupleSets); ok {
  40. result = v1
  41. }
  42. }
  43. }
  44. if result.Len() <= 0 {
  45. log.Debugf("join plan yields nothing")
  46. return nil
  47. }
  48. return result
  49. }
  50. func (jp *JoinOp) getStreamNames(join *xsql.Join) ([]string, error) {
  51. var srcs []string
  52. xsql.WalkFunc(join, func(node xsql.Node) {
  53. if f, ok := node.(*xsql.FieldRef); ok {
  54. if string(f.StreamName) == "" {
  55. return
  56. }
  57. srcs = append(srcs, string(f.StreamName))
  58. }
  59. })
  60. if len(srcs) != 2 {
  61. if jp.From.Alias != "" {
  62. srcs = append(srcs, jp.From.Alias)
  63. } else {
  64. srcs = append(srcs, jp.From.Name)
  65. }
  66. if join.Alias != "" {
  67. srcs = append(srcs, join.Alias)
  68. } else {
  69. srcs = append(srcs, join.Name)
  70. }
  71. }
  72. return srcs, nil
  73. }
  74. func (jp *JoinOp) evalSet(input xsql.WindowTuplesSet, join xsql.Join, fv *xsql.FunctionValuer) (xsql.JoinTupleSets, error) {
  75. var leftStream, rightStream string
  76. if join.JoinType != xsql.CROSS_JOIN {
  77. streams, err := jp.getStreamNames(&join)
  78. if err != nil {
  79. return nil, err
  80. }
  81. leftStream = streams[0]
  82. rightStream = streams[1]
  83. } else {
  84. if jp.From.Alias == "" {
  85. leftStream = jp.From.Name
  86. } else {
  87. leftStream = jp.From.Alias
  88. }
  89. if join.Alias == "" {
  90. rightStream = join.Name
  91. } else {
  92. rightStream = join.Alias
  93. }
  94. }
  95. var lefts, rights []xsql.Tuple
  96. lefts = input.GetBySrc(leftStream)
  97. rights = input.GetBySrc(rightStream)
  98. sets := xsql.JoinTupleSets{}
  99. if join.JoinType == xsql.RIGHT_JOIN {
  100. return jp.evalSetWithRightJoin(input, join, false, fv)
  101. }
  102. for _, left := range lefts {
  103. leftJoined := false
  104. for index, right := range rights {
  105. tupleJoined := false
  106. merged := &xsql.JoinTuple{}
  107. if join.JoinType == xsql.LEFT_JOIN || join.JoinType == xsql.FULL_JOIN || join.JoinType == xsql.CROSS_JOIN {
  108. merged.AddTuple(left)
  109. }
  110. if join.JoinType == xsql.CROSS_JOIN {
  111. tupleJoined = true
  112. merged.AddTuple(right)
  113. } else {
  114. temp := &xsql.JoinTuple{}
  115. temp.AddTuple(left)
  116. temp.AddTuple(right)
  117. ve := &xsql.ValuerEval{Valuer: xsql.MultiValuer(temp, fv)}
  118. result := evalOn(join, ve, &left, &right)
  119. switch val := result.(type) {
  120. case error:
  121. return nil, val
  122. case bool:
  123. if val {
  124. leftJoined = true
  125. tupleJoined = true
  126. if join.JoinType == xsql.INNER_JOIN {
  127. merged.AddTuple(left)
  128. merged.AddTuple(right)
  129. } else {
  130. merged.AddTuple(right)
  131. }
  132. }
  133. default:
  134. return nil, fmt.Errorf("invalid join condition that returns non-bool value %[1]T(%[1]v)", val)
  135. }
  136. }
  137. if tupleJoined || (!leftJoined && index == len(rights)-1 && len(merged.Tuples) > 0) {
  138. leftJoined = true
  139. sets = append(sets, *merged)
  140. }
  141. }
  142. // If no messages in the right
  143. if !leftJoined && join.JoinType != xsql.INNER_JOIN {
  144. merged := &xsql.JoinTuple{}
  145. merged.AddTuple(left)
  146. sets = append(sets, *merged)
  147. }
  148. }
  149. if join.JoinType == xsql.FULL_JOIN {
  150. if rightJoinSet, err := jp.evalSetWithRightJoin(input, join, true, fv); err == nil {
  151. if len(rightJoinSet) > 0 {
  152. for _, jt := range rightJoinSet {
  153. sets = append(sets, jt)
  154. }
  155. }
  156. } else {
  157. return nil, err
  158. }
  159. }
  160. return sets, nil
  161. }
  162. func evalOn(join xsql.Join, ve *xsql.ValuerEval, left interface{}, right *xsql.Tuple) interface{} {
  163. var result interface{}
  164. if join.Expr != nil {
  165. result = ve.Eval(join.Expr)
  166. } else if join.JoinType == xsql.INNER_JOIN { // if no on expression
  167. result = left != nil && right != nil
  168. } else {
  169. result = true
  170. }
  171. return result
  172. }
  173. func (jp *JoinOp) evalSetWithRightJoin(input xsql.WindowTuplesSet, join xsql.Join, excludeJoint bool, fv *xsql.FunctionValuer) (xsql.JoinTupleSets, error) {
  174. streams, err := jp.getStreamNames(&join)
  175. if err != nil {
  176. return nil, err
  177. }
  178. leftStream := streams[0]
  179. rightStream := streams[1]
  180. var lefts, rights []xsql.Tuple
  181. lefts = input.GetBySrc(leftStream)
  182. rights = input.GetBySrc(rightStream)
  183. sets := xsql.JoinTupleSets{}
  184. for _, right := range rights {
  185. isJoint := false
  186. for index, left := range lefts {
  187. tupleJoined := false
  188. merged := &xsql.JoinTuple{}
  189. merged.AddTuple(right)
  190. temp := &xsql.JoinTuple{}
  191. temp.AddTuple(right)
  192. temp.AddTuple(left)
  193. ve := &xsql.ValuerEval{Valuer: xsql.MultiValuer(temp, fv)}
  194. result := evalOn(join, ve, &left, &right)
  195. switch val := result.(type) {
  196. case error:
  197. return nil, val
  198. case bool:
  199. if val {
  200. merged.AddTuple(left)
  201. isJoint = true
  202. tupleJoined = true
  203. }
  204. default:
  205. return nil, fmt.Errorf("invalid join condition that returns non-bool value %[1]T(%[1]v)", val)
  206. }
  207. if !excludeJoint && (tupleJoined || (!isJoint && index == len(lefts)-1 && len(merged.Tuples) > 0)) {
  208. isJoint = true
  209. sets = append(sets, *merged)
  210. }
  211. }
  212. if !isJoint {
  213. merged := &xsql.JoinTuple{}
  214. merged.AddTuple(right)
  215. sets = append(sets, *merged)
  216. }
  217. }
  218. return sets, nil
  219. }
  220. func (jp *JoinOp) evalJoinSets(set *xsql.JoinTupleSets, input xsql.WindowTuplesSet, join xsql.Join, fv *xsql.FunctionValuer) (interface{}, error) {
  221. var rightStream string
  222. if join.Alias == "" {
  223. rightStream = join.Name
  224. } else {
  225. rightStream = join.Alias
  226. }
  227. rights := input.GetBySrc(rightStream)
  228. newSets := xsql.JoinTupleSets{}
  229. if join.JoinType == xsql.RIGHT_JOIN {
  230. return jp.evalRightJoinSets(set, input, join, false, fv)
  231. }
  232. for _, left := range *set {
  233. leftJoined := false
  234. innerAppend := false
  235. for index, right := range rights {
  236. tupleJoined := false
  237. merged := &xsql.JoinTuple{}
  238. if join.JoinType == xsql.LEFT_JOIN || join.JoinType == xsql.FULL_JOIN || join.JoinType == xsql.CROSS_JOIN {
  239. merged.AddTuples(left.Tuples)
  240. }
  241. if join.JoinType == xsql.CROSS_JOIN {
  242. tupleJoined = true
  243. merged.AddTuple(right)
  244. } else {
  245. ve := &xsql.ValuerEval{Valuer: xsql.MultiValuer(&left, &right, fv)}
  246. result := evalOn(join, ve, &left, &right)
  247. switch val := result.(type) {
  248. case error:
  249. return nil, val
  250. case bool:
  251. if val {
  252. leftJoined = true
  253. tupleJoined = true
  254. if join.JoinType == xsql.INNER_JOIN && !innerAppend {
  255. merged.AddTuples(left.Tuples)
  256. innerAppend = true
  257. }
  258. merged.AddTuple(right)
  259. }
  260. default:
  261. return nil, fmt.Errorf("invalid join condition that returns non-bool value %[1]T(%[1]v)", val)
  262. }
  263. }
  264. if tupleJoined || (!leftJoined && index == len(rights)-1 && len(merged.Tuples) > 0) {
  265. leftJoined = true
  266. newSets = append(newSets, *merged)
  267. }
  268. }
  269. if !leftJoined && join.JoinType != xsql.INNER_JOIN {
  270. merged := &xsql.JoinTuple{}
  271. merged.AddTuples(left.Tuples)
  272. newSets = append(newSets, *merged)
  273. }
  274. }
  275. if join.JoinType == xsql.FULL_JOIN {
  276. if rightJoinSet, err := jp.evalRightJoinSets(set, input, join, true, fv); err == nil && len(rightJoinSet) > 0 {
  277. for _, jt := range rightJoinSet {
  278. newSets = append(newSets, jt)
  279. }
  280. }
  281. }
  282. return newSets, nil
  283. }
  284. func (jp *JoinOp) evalRightJoinSets(set *xsql.JoinTupleSets, input xsql.WindowTuplesSet, join xsql.Join, excludeJoint bool, fv *xsql.FunctionValuer) (xsql.JoinTupleSets, error) {
  285. var rightStream string
  286. if join.Alias == "" {
  287. rightStream = join.Name
  288. } else {
  289. rightStream = join.Alias
  290. }
  291. rights := input.GetBySrc(rightStream)
  292. newSets := xsql.JoinTupleSets{}
  293. for _, right := range rights {
  294. isJoint := false
  295. for index, left := range *set {
  296. tupleJoined := false
  297. merged := &xsql.JoinTuple{}
  298. merged.AddTuple(right)
  299. ve := &xsql.ValuerEval{Valuer: xsql.MultiValuer(&right, &left, fv)}
  300. result := evalOn(join, ve, &left, &right)
  301. switch val := result.(type) {
  302. case error:
  303. return nil, val
  304. case bool:
  305. if val {
  306. isJoint = true
  307. tupleJoined = true
  308. merged.AddTuples(left.Tuples)
  309. }
  310. default:
  311. return nil, fmt.Errorf("invalid join condition that returns non-bool value %[1]T(%[1]v)", val)
  312. }
  313. if !excludeJoint && (tupleJoined || (!isJoint && index == len(*set)-1 && len(merged.Tuples) > 0)) {
  314. isJoint = true
  315. newSets = append(newSets, *merged)
  316. }
  317. }
  318. if !isJoint {
  319. merged := &xsql.JoinTuple{}
  320. merged.AddTuple(right)
  321. newSets = append(newSets, *merged)
  322. }
  323. }
  324. return newSets, nil
  325. }