Github user nongli commented on a diff in the pull request:
https://github.com/apache/spark/pull/11328#discussion_r53857174
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/joins/BroadcastNestedLoopJoin.scala
---
@@ -58,56 +58,173 @@ case class BroadcastNestedLoopJoin(
override def output: Seq[Attribute] = {
joinType match {
+ case Inner =>
+ left.output ++ right.output
case LeftOuter =>
left.output ++ right.output.map(_.withNullability(true))
case RightOuter =>
left.output.map(_.withNullability(true)) ++ right.output
case FullOuter =>
left.output.map(_.withNullability(true)) ++
right.output.map(_.withNullability(true))
- case Inner =>
- // TODO we can avoid breaking the lineage, since we union an empty
RDD for Inner Join case
- left.output ++ right.output
- case x => // TODO support the Left Semi Join
+ case LeftSemi =>
+ left.output
+ case x =>
throw new IllegalArgumentException(
s"BroadcastNestedLoopJoin should not take $x as the JoinType")
}
}
- @transient private lazy val boundCondition =
- newPredicate(condition.getOrElse(Literal(true)), left.output ++
right.output)
+ @transient private lazy val boundCondition = {
+ if (condition.isDefined) {
+ newPredicate(condition.get, left.output ++ right.output)
+ } else {
+ (r: InternalRow) => true
+ }
+ }
- protected override def doExecute(): RDD[InternalRow] = {
- val numOutputRows = longMetric("numOutputRows")
+ private def innerJoin(relation: Broadcast[Array[InternalRow]]):
RDD[InternalRow] = {
+ streamed.execute().mapPartitionsInternal { streamedIter =>
+ val buildRows = relation.value
+ val joinedRow = new JoinedRow
- val broadcastedRelation =
broadcast.executeBroadcast[Array[InternalRow]]()
+ streamedIter.flatMap { streamedRow =>
+ val joinedRows = buildSide match {
+ case BuildRight =>
+ buildRows.iterator.map(r => joinedRow(streamedRow, r))
+ case BuildLeft =>
+ buildRows.iterator.map(r => joinedRow(r, streamedRow))
+ }
+ if (condition.isDefined) {
+ joinedRows.filter(boundCondition)
+ } else {
+ joinedRows
+ }
+ }
+ }
+ }
+
+ private def leftOuterJoin(relation: Broadcast[Array[InternalRow]]):
RDD[InternalRow] = {
+ assert(buildSide == BuildRight)
+ streamed.execute().mapPartitionsInternal { streamedIter =>
+ val buildRows = relation.value
+ val joinedRow = new JoinedRow
+
+ streamedIter.flatMap { streamedRow =>
+ val joinedRows = buildRows.iterator.map(r =>
joinedRow(streamedRow, r))
+ if (condition.isDefined) {
+ joinedRows.filter(boundCondition)
--- End diff --
Why is this right? If it fails on the join condition, it should put null on
the other side.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
