join_operator.go 8.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330
  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. merged := &xsql.JoinTuple{}
  104. if join.JoinType == xsql.LEFT_JOIN || join.JoinType == xsql.FULL_JOIN || join.JoinType == xsql.CROSS_JOIN {
  105. merged.AddTuple(left)
  106. }
  107. for _, right := range rights {
  108. if join.JoinType == xsql.CROSS_JOIN {
  109. merged.AddTuple(right)
  110. } else {
  111. temp := &xsql.JoinTuple{}
  112. temp.AddTuple(left)
  113. temp.AddTuple(right)
  114. ve := &xsql.ValuerEval{Valuer: xsql.MultiValuer(temp, fv)}
  115. result := evalOn(join, ve, &left, &right)
  116. switch val := result.(type) {
  117. case error:
  118. return nil, val
  119. case bool:
  120. if val {
  121. if join.JoinType == xsql.INNER_JOIN {
  122. merged.AddTuple(left)
  123. merged.AddTuple(right)
  124. sets = append(sets, *merged)
  125. merged = &xsql.JoinTuple{}
  126. } else {
  127. merged.AddTuple(right)
  128. }
  129. }
  130. default:
  131. return nil, fmt.Errorf("invalid join condition that returns non-bool value %[1]T(%[1]v)", val)
  132. }
  133. }
  134. }
  135. if len(merged.Tuples) > 0 {
  136. sets = append(sets, *merged)
  137. }
  138. }
  139. if join.JoinType == xsql.FULL_JOIN {
  140. if rightJoinSet, err := jp.evalSetWithRightJoin(input, join, true, fv); err == nil {
  141. if len(rightJoinSet) > 0 {
  142. for _, jt := range rightJoinSet {
  143. sets = append(sets, jt)
  144. }
  145. }
  146. } else {
  147. return nil, err
  148. }
  149. }
  150. return sets, nil
  151. }
  152. func evalOn(join xsql.Join, ve *xsql.ValuerEval, left interface{}, right *xsql.Tuple) interface{} {
  153. var result interface{}
  154. if join.Expr != nil {
  155. result = ve.Eval(join.Expr)
  156. } else if join.JoinType == xsql.INNER_JOIN { // if no on expression
  157. result = left != nil && right != nil
  158. } else {
  159. result = true
  160. }
  161. return result
  162. }
  163. func (jp *JoinOp) evalSetWithRightJoin(input xsql.WindowTuplesSet, join xsql.Join, excludeJoint bool, fv *xsql.FunctionValuer) (xsql.JoinTupleSets, error) {
  164. streams, err := jp.getStreamNames(&join)
  165. if err != nil {
  166. return nil, err
  167. }
  168. leftStream := streams[0]
  169. rightStream := streams[1]
  170. var lefts, rights []xsql.Tuple
  171. lefts = input.GetBySrc(leftStream)
  172. rights = input.GetBySrc(rightStream)
  173. sets := xsql.JoinTupleSets{}
  174. for _, right := range rights {
  175. merged := &xsql.JoinTuple{}
  176. merged.AddTuple(right)
  177. isJoint := false
  178. for _, left := range lefts {
  179. temp := &xsql.JoinTuple{}
  180. temp.AddTuple(right)
  181. temp.AddTuple(left)
  182. ve := &xsql.ValuerEval{Valuer: xsql.MultiValuer(temp, fv)}
  183. result := evalOn(join, ve, &left, &right)
  184. switch val := result.(type) {
  185. case error:
  186. return nil, val
  187. case bool:
  188. if val {
  189. merged.AddTuple(left)
  190. isJoint = true
  191. }
  192. default:
  193. return nil, fmt.Errorf("invalid join condition that returns non-bool value %[1]T(%[1]v)", val)
  194. }
  195. }
  196. if excludeJoint {
  197. if len(merged.Tuples) > 0 && (!isJoint) {
  198. sets = append(sets, *merged)
  199. }
  200. } else {
  201. if len(merged.Tuples) > 0 {
  202. sets = append(sets, *merged)
  203. }
  204. }
  205. }
  206. return sets, nil
  207. }
  208. func (jp *JoinOp) evalJoinSets(set *xsql.JoinTupleSets, input xsql.WindowTuplesSet, join xsql.Join, fv *xsql.FunctionValuer) (interface{}, error) {
  209. var rightStream string
  210. if join.Alias == "" {
  211. rightStream = join.Name
  212. } else {
  213. rightStream = join.Alias
  214. }
  215. rights := input.GetBySrc(rightStream)
  216. newSets := xsql.JoinTupleSets{}
  217. if join.JoinType == xsql.RIGHT_JOIN {
  218. return jp.evalRightJoinSets(set, input, join, false, fv)
  219. }
  220. for _, left := range *set {
  221. merged := &xsql.JoinTuple{}
  222. if join.JoinType == xsql.LEFT_JOIN || join.JoinType == xsql.FULL_JOIN || join.JoinType == xsql.CROSS_JOIN {
  223. merged.AddTuples(left.Tuples)
  224. }
  225. innerAppend := false
  226. for _, right := range rights {
  227. if join.JoinType == xsql.CROSS_JOIN {
  228. merged.AddTuple(right)
  229. } else {
  230. ve := &xsql.ValuerEval{Valuer: xsql.MultiValuer(&left, &right, fv)}
  231. result := evalOn(join, ve, &left, &right)
  232. switch val := result.(type) {
  233. case error:
  234. return nil, val
  235. case bool:
  236. if val {
  237. if join.JoinType == xsql.INNER_JOIN && !innerAppend {
  238. merged.AddTuples(left.Tuples)
  239. innerAppend = true
  240. }
  241. merged.AddTuple(right)
  242. }
  243. default:
  244. return nil, fmt.Errorf("invalid join condition that returns non-bool value %[1]T(%[1]v)", val)
  245. }
  246. }
  247. }
  248. if len(merged.Tuples) > 0 {
  249. newSets = append(newSets, *merged)
  250. }
  251. }
  252. if join.JoinType == xsql.FULL_JOIN {
  253. if rightJoinSet, err := jp.evalRightJoinSets(set, input, join, true, fv); err == nil && len(rightJoinSet) > 0 {
  254. for _, jt := range rightJoinSet {
  255. newSets = append(newSets, jt)
  256. }
  257. }
  258. }
  259. return newSets, nil
  260. }
  261. func (jp *JoinOp) evalRightJoinSets(set *xsql.JoinTupleSets, input xsql.WindowTuplesSet, join xsql.Join, excludeJoint bool, fv *xsql.FunctionValuer) (xsql.JoinTupleSets, error) {
  262. var rightStream string
  263. if join.Alias == "" {
  264. rightStream = join.Name
  265. } else {
  266. rightStream = join.Alias
  267. }
  268. rights := input.GetBySrc(rightStream)
  269. newSets := xsql.JoinTupleSets{}
  270. for _, right := range rights {
  271. merged := &xsql.JoinTuple{}
  272. merged.AddTuple(right)
  273. isJoint := false
  274. for _, left := range *set {
  275. ve := &xsql.ValuerEval{Valuer: xsql.MultiValuer(&right, &left, fv)}
  276. result := evalOn(join, ve, &left, &right)
  277. switch val := result.(type) {
  278. case error:
  279. return nil, val
  280. case bool:
  281. if val {
  282. isJoint = true
  283. merged.AddTuples(left.Tuples)
  284. }
  285. default:
  286. return nil, fmt.Errorf("invalid join condition that returns non-bool value %[1]T(%[1]v)", val)
  287. }
  288. }
  289. if excludeJoint {
  290. if len(merged.Tuples) > 0 && (!isJoint) {
  291. newSets = append(newSets, *merged)
  292. }
  293. } else {
  294. if len(merged.Tuples) > 0 {
  295. newSets = append(newSets, *merged)
  296. }
  297. }
  298. }
  299. return newSets, nil
  300. }