Skip to content

[SPARK-16391][SQL] KeyValueGroupedDataset.reduceGroups should support partial aggregation #14222

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
viirya wants to merge 3 commits into from
Closed

[SPARK-16391][SQL] KeyValueGroupedDataset.reduceGroups should support partial aggregation #14222

Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
1135773
Support partial aggregation for reduceGroups.
viirya Jul 15, 2016
7e8d8c1
Add ReduceAggregator.
viirya Jul 18, 2016
6032325
For comment.
viirya Jul 20, 2016
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
30 changes: 27 additions & 3 deletions sql/core/src/main/scala/org/apache/spark/sql/KeyValueGroupedDataset.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,6 +25,7 @@ import org.apache.spark.sql.catalyst.encoders.{encoderFor, ExpressionEncoder, Ou
import org.apache.spark.sql.catalyst.expressions.{Alias, Attribute, CreateStruct}
import org.apache.spark.sql.catalyst.plans.logical._
import org.apache.spark.sql.execution.QueryExecution
import org.apache.spark.sql.expressions.Aggregator

/**
* :: Experimental ::
Expand Down Expand Up @@ -177,10 +178,33 @@ class KeyValueGroupedDataset[K, V] private[sql](
* @since 1.6.0
*/
def reduceGroups(f: (V, V) => V): Dataset[(K, V)] = {
val func = (key: K, it: Iterator[V]) => Iterator((key, it.reduce(f)))
val encoder = encoderFor[V]
val intEncoder: ExpressionEncoder[Int] = ExpressionEncoder()
val aggregator: TypedColumn[V, V] = new Aggregator[V, (Int, V), V] {
def bufferEncoder: Encoder[(Int, V)] = ExpressionEncoder.tuple(intEncoder, encoder)
def outputEncoder: Encoder[V] = encoder

implicit val resultEncoder = ExpressionEncoder.tuple(kExprEnc, vExprEnc)
flatMapGroups(func)
def zero: (Int, V) = (0, null.asInstanceOf[V])
Copy link
Member Author

@viirya viirya Jul 15, 2016

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

One problem with Aggregator here is the zero value. This PR uses an Int (can be Boolean too) to indicate if the buffer is initialized.

Copy link
Contributor

@rxin rxin Jul 15, 2016

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

pull this out to be a generic ReduceAggregator and add unit test for it?

Copy link
Member Author

@viirya viirya Jul 16, 2016

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

ok. I will update this.

def reduce(reducedValue: (Int, V), value: V): (Int, V) = {
if (reducedValue._1 == 0) {
(1, value)
} else {
(1, f(reducedValue._2, value))
}
}
def merge(buf1: (Int, V), buf2: (Int, V)): (Int, V) = {
if (buf1._1 == 0) {
buf2
} else if (buf2._2 == 0) {
buf1
} else {
(1, f(buf1._2, buf2._2))
}
}
def finish(result: (Int, V)): V = result._2
}.toColumn

agg(aggregator)
}

/**
Expand Down