join_operator.go 7.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307
  1. package plans
  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 JoinPlan 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 *JoinPlan) Apply(ctx api.StreamContext, data interface{}) interface{} {
  15. log := ctx.GetLogger()
  16. var input xsql.WindowTuplesSet
  17. switch v := data.(type) {
  18. case error:
  19. return input
  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)
  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)
  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 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. return nil, fmt.Errorf("Not correct join expression, it requires exactly 2 sources at ON expression.")
  62. }
  63. return srcs, nil
  64. }
  65. func (jp *JoinPlan) evalSet(input xsql.WindowTuplesSet, join xsql.Join) (xsql.JoinTupleSets, error) {
  66. var leftStream, rightStream string
  67. if join.JoinType != xsql.CROSS_JOIN {
  68. streams, err := getStreamNames(&join)
  69. if err != nil {
  70. return nil, err
  71. }
  72. leftStream = streams[0]
  73. rightStream = streams[1]
  74. } else {
  75. if jp.From.Alias == "" {
  76. leftStream = jp.From.Name
  77. } else {
  78. leftStream = jp.From.Alias
  79. }
  80. if join.Alias == "" {
  81. rightStream = join.Name
  82. } else {
  83. rightStream = join.Alias
  84. }
  85. }
  86. var lefts, rights []xsql.Tuple
  87. lefts = input.GetBySrc(leftStream)
  88. rights = input.GetBySrc(rightStream)
  89. sets := xsql.JoinTupleSets{}
  90. if join.JoinType == xsql.RIGHT_JOIN {
  91. return jp.evalSetWithRightJoin(input, join, false)
  92. }
  93. for _, left := range lefts {
  94. merged := &xsql.JoinTuple{}
  95. if join.JoinType == xsql.LEFT_JOIN || join.JoinType == xsql.FULL_JOIN || join.JoinType == xsql.CROSS_JOIN {
  96. merged.AddTuple(left)
  97. }
  98. for _, right := range rights {
  99. if join.JoinType == xsql.CROSS_JOIN {
  100. merged.AddTuple(right)
  101. } else {
  102. temp := &xsql.JoinTuple{}
  103. temp.AddTuple(left)
  104. temp.AddTuple(right)
  105. ve := &xsql.ValuerEval{Valuer: xsql.MultiValuer(temp, &xsql.FunctionValuer{})}
  106. result := ve.Eval(join.Expr)
  107. switch val := result.(type) {
  108. case error:
  109. return nil, val
  110. case bool:
  111. if val {
  112. if join.JoinType == xsql.INNER_JOIN {
  113. merged.AddTuple(left)
  114. merged.AddTuple(right)
  115. sets = append(sets, *merged)
  116. merged = &xsql.JoinTuple{}
  117. } else {
  118. merged.AddTuple(right)
  119. }
  120. }
  121. default:
  122. return nil, fmt.Errorf("invalid join condition that returns non-bool value %[1]T(%[1]v)", val)
  123. }
  124. }
  125. }
  126. if len(merged.Tuples) > 0 {
  127. sets = append(sets, *merged)
  128. }
  129. }
  130. if join.JoinType == xsql.FULL_JOIN {
  131. if rightJoinSet, err := jp.evalSetWithRightJoin(input, join, true); err == nil {
  132. if len(rightJoinSet) > 0 {
  133. for _, jt := range rightJoinSet {
  134. sets = append(sets, jt)
  135. }
  136. }
  137. } else {
  138. return nil, err
  139. }
  140. }
  141. return sets, nil
  142. }
  143. func (jp *JoinPlan) evalSetWithRightJoin(input xsql.WindowTuplesSet, join xsql.Join, excludeJoint bool) (xsql.JoinTupleSets, error) {
  144. streams, err := getStreamNames(&join)
  145. if err != nil {
  146. return nil, err
  147. }
  148. leftStream := streams[0]
  149. rightStream := streams[1]
  150. var lefts, rights []xsql.Tuple
  151. lefts = input.GetBySrc(leftStream)
  152. rights = input.GetBySrc(rightStream)
  153. sets := xsql.JoinTupleSets{}
  154. for _, right := range rights {
  155. merged := &xsql.JoinTuple{}
  156. merged.AddTuple(right)
  157. isJoint := false
  158. for _, left := range lefts {
  159. temp := &xsql.JoinTuple{}
  160. temp.AddTuple(right)
  161. temp.AddTuple(left)
  162. ve := &xsql.ValuerEval{Valuer: xsql.MultiValuer(temp, &xsql.FunctionValuer{})}
  163. result := ve.Eval(join.Expr)
  164. switch val := result.(type) {
  165. case error:
  166. return nil, val
  167. case bool:
  168. if val {
  169. merged.AddTuple(left)
  170. isJoint = true
  171. }
  172. default:
  173. return nil, fmt.Errorf("invalid join condition that returns non-bool value %[1]T(%[1]v)", val)
  174. }
  175. }
  176. if excludeJoint {
  177. if len(merged.Tuples) > 0 && (!isJoint) {
  178. sets = append(sets, *merged)
  179. }
  180. } else {
  181. if len(merged.Tuples) > 0 {
  182. sets = append(sets, *merged)
  183. }
  184. }
  185. }
  186. return sets, nil
  187. }
  188. func (jp *JoinPlan) evalJoinSets(set *xsql.JoinTupleSets, input xsql.WindowTuplesSet, join xsql.Join) (interface{}, error) {
  189. var rightStream string
  190. if join.Alias == "" {
  191. rightStream = join.Name
  192. } else {
  193. rightStream = join.Alias
  194. }
  195. rights := input.GetBySrc(rightStream)
  196. newSets := xsql.JoinTupleSets{}
  197. if join.JoinType == xsql.RIGHT_JOIN {
  198. return jp.evalRightJoinSets(set, input, join, false)
  199. }
  200. for _, left := range *set {
  201. merged := &xsql.JoinTuple{}
  202. if join.JoinType == xsql.LEFT_JOIN || join.JoinType == xsql.FULL_JOIN || join.JoinType == xsql.CROSS_JOIN {
  203. merged.AddTuples(left.Tuples)
  204. }
  205. innerAppend := false
  206. for _, right := range rights {
  207. if join.JoinType == xsql.CROSS_JOIN {
  208. merged.AddTuple(right)
  209. } else {
  210. ve := &xsql.ValuerEval{Valuer: xsql.MultiValuer(&left, &right, &xsql.FunctionValuer{})}
  211. result := ve.Eval(join.Expr)
  212. switch val := result.(type) {
  213. case error:
  214. return nil, val
  215. case bool:
  216. if val {
  217. if join.JoinType == xsql.INNER_JOIN && !innerAppend {
  218. merged.AddTuples(left.Tuples)
  219. innerAppend = true
  220. }
  221. merged.AddTuple(right)
  222. }
  223. default:
  224. return nil, fmt.Errorf("invalid join condition that returns non-bool value %[1]T(%[1]v)", val)
  225. }
  226. }
  227. }
  228. if len(merged.Tuples) > 0 {
  229. newSets = append(newSets, *merged)
  230. }
  231. }
  232. if join.JoinType == xsql.FULL_JOIN {
  233. if rightJoinSet, err := jp.evalRightJoinSets(set, input, join, true); err == nil && len(rightJoinSet) > 0 {
  234. for _, jt := range rightJoinSet {
  235. newSets = append(newSets, jt)
  236. }
  237. }
  238. }
  239. return newSets, nil
  240. }
  241. func (jp *JoinPlan) evalRightJoinSets(set *xsql.JoinTupleSets, input xsql.WindowTuplesSet, join xsql.Join, excludeJoint bool) (xsql.JoinTupleSets, error) {
  242. var rightStream string
  243. if join.Alias == "" {
  244. rightStream = join.Name
  245. } else {
  246. rightStream = join.Alias
  247. }
  248. rights := input.GetBySrc(rightStream)
  249. newSets := xsql.JoinTupleSets{}
  250. for _, right := range rights {
  251. merged := &xsql.JoinTuple{}
  252. merged.AddTuple(right)
  253. isJoint := false
  254. for _, left := range *set {
  255. ve := &xsql.ValuerEval{Valuer: xsql.MultiValuer(&right, &left, &xsql.FunctionValuer{})}
  256. result := ve.Eval(join.Expr)
  257. switch val := result.(type) {
  258. case error:
  259. return nil, val
  260. case bool:
  261. if val {
  262. isJoint = true
  263. merged.AddTuples(left.Tuples)
  264. }
  265. default:
  266. return nil, fmt.Errorf("invalid join condition that returns non-bool value %[1]T(%[1]v)", val)
  267. }
  268. }
  269. if excludeJoint {
  270. if len(merged.Tuples) > 0 && (!isJoint) {
  271. newSets = append(newSets, *merged)
  272. }
  273. } else {
  274. if len(merged.Tuples) > 0 {
  275. newSets = append(newSets, *merged)
  276. }
  277. }
  278. }
  279. return newSets, nil
  280. }