c21 commented on a change in pull request #29277:
URL: https://github.com/apache/spark/pull/29277#discussion_r462830406
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/joins/ShuffledHashJoinExec.scala
##########
@@ -70,4 +74,54 @@ case class ShuffledHashJoinExec(
join(streamIter, hashed, numOutputRows)
}
}
+
+ override def inputRDDs(): Seq[RDD[InternalRow]] = {
+ streamedPlan.execute() :: buildPlan.execute() :: Nil
+ }
+
+ override def needCopyResult: Boolean = true
+
+ override protected def doProduce(ctx: CodegenContext): String = {
+ // inline mutable state since not many join operations in a task
+ val streamedInput = ctx.addMutableState(
+ "scala.collection.Iterator", "streamedInput", v => s"$v = inputs[0];",
forceInline = true)
+ val streamedRow = ctx.addMutableState(
+ "InternalRow", "streamedRow", forceInline = true)
+ val (streamInputVar, streamInputVarDecl) = createVars(ctx, streamedRow,
streamedPlan.output)
+
+ val join = joinType match {
+ case _: InnerLike => codegenInner(ctx, streamInputVar)
+ case LeftOuter | RightOuter => codegenOuter(ctx, streamInputVar)
+ case LeftSemi => codegenSemi(ctx, streamInputVar)
+ case LeftAnti => codegenAnti(ctx, streamInputVar)
+ case _: ExistenceJoin => codegenExistence(ctx, streamInputVar)
+ case x =>
+ throw new IllegalArgumentException(
+ s"ShuffledHashJoin should not take $x as the JoinType")
+ }
+
+ s"""
+ |while ($streamedInput.hasNext()) {
Review comment:
@cloud-fan - sorry I got your point now, updated to do join part in
`doConsume()` now. Moved BHJ's `doProduce()` and `doConsume()` into `HashJoin`,
to be shared between BHJ and SHJ. Also updated PR with latest codegen code for
example query. Thanks for suggestion.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
