c21 commented on a change in pull request #34444:
URL: https://github.com/apache/spark/pull/34444#discussion_r741525180
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/joins/ShuffledHashJoinExec.scala
##########
@@ -332,6 +327,266 @@ case class ShuffledHashJoinExec(
HashedRelationInfo(relationTerm, keyIsUnique = false, isEmpty = false)
}
+ override def doProduce(ctx: CodegenContext): String = {
+ // Specialize `doProduce` code for full outer join, because full outer
join needs to
+ // iterate streamed and build side separately.
+ if (joinType != FullOuter) {
+ return super.doProduce(ctx)
+ }
+
+ val HashedRelationInfo(relationTerm, _, _) = prepareRelation(ctx)
+
+ // Inline mutable state since not many join operations in a task
+ val keyIsUnique = ctx.addMutableState("boolean", "keyIsUnique",
+ v => s"$v = $relationTerm.keyIsUnique();", forceInline = true)
+ val streamedInput = ctx.addMutableState("scala.collection.Iterator",
"streamedInput",
+ v => s"$v = inputs[0];", forceInline = true)
+ val buildInput = ctx.addMutableState("scala.collection.Iterator",
"buildInput",
+ v => s"$v = $relationTerm.valuesWithKeyIndex();", forceInline = true)
+ val streamedRow = ctx.addMutableState("InternalRow", "streamedRow",
forceInline = true)
+ val buildRow = ctx.addMutableState("InternalRow", "buildRow", forceInline
= true)
+
+ // Generate variables and related code from streamed side
+ val streamedVars = genOneSideJoinVars(ctx, streamedRow, streamedPlan,
setDefaultValue = false)
+ val streamedKeyVariables = evaluateRequiredVariables(streamedOutput,
streamedVars,
+
AttributeSet.fromAttributeSets(HashJoin.rewriteKeyExpr(streamedKeys).map(_.references)))
+ ctx.currentVars = streamedVars
+ val streamedKeyExprCode = GenerateUnsafeProjection.createCode(ctx,
streamedBoundKeys)
+ val streamedKeyEv =
+ s"""
+ |$streamedKeyVariables
+ |${streamedKeyExprCode.code}
+ """.stripMargin
+ val streamedKeyAnyNull = s"${streamedKeyExprCode.value}.anyNull()"
Review comment:
@cloud-fan - sorry if I am missing anything, `anyNull` returns whether
any null of the current input row (`UnsafeRow.anyNull()`). The result is varied
for each input row in while loop. Just to make sure we are in the same page, I
am referring to `anyNull` call in [line 094 & 157 of example
code-gen](https://gist.github.com/c21/828b782ee81827f4148939cb50314a7b).
--
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.
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
