Skip to content
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

Jump to bottom

fix: Comet should not translate try_sum to native sum expression #277

Merged
merged 4 commits into from
Apr 17, 2024
Merged

fix: Comet should not translate try_sum to native sum expression #277

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
ee349ad
fix: Comet should not translate try_sum to native sum expression
viirya Apr 16, 2024
a447471
For Spark 3.2 and 3.3
viirya Apr 16, 2024
8328391
Update spark/src/main/scala/org/apache/comet/shims/ShimQueryPlanSerde…
viirya Apr 17, 2024
8198590
Fix format
viirya Apr 17, 2024
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
4 changes: 2 additions & 2 deletions spark/src/main/scala/org/apache/comet/serde/QueryPlanSerde.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -202,7 +202,7 @@ object QueryPlanSerde extends Logging with ShimQueryPlanSerde {
inputs: Seq[Attribute],
binding: Boolean): Option[AggExpr] = {
aggExpr.aggregateFunction match {
case s @ Sum(child, _) if sumDataTypeSupported(s.dataType) =>
case s @ Sum(child, _) if sumDataTypeSupported(s.dataType) && isLegacyMode(s) =>
Copy link
Contributor

Choose a reason for hiding this comment

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

Not part of this PR.
But maybe we should add try_sum and try_xxx aggregation function support in the native side too?

Copy link
Member Author

Choose a reason for hiding this comment

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

Yea, we should do it. We probably need to implement native try_sum expression because DataFusion might not have it.

Copy link
Contributor

Choose a reason for hiding this comment

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

yeah, so I just went ahead and create an issue to track this one: #280.

val childExpr = exprToProto(child, inputs, binding)
val dataType = serializeDataType(s.dataType)

Expand All @@ -220,7 +220,7 @@ object QueryPlanSerde extends Logging with ShimQueryPlanSerde {
} else {
None
}
case s @ Average(child, _) if avgDataTypeSupported(s.dataType) =>
case s @ Average(child, _) if avgDataTypeSupported(s.dataType) && isLegacyMode(s) =>
val childExpr = exprToProto(child, inputs, binding)
val dataType = serializeDataType(s.dataType)

Expand Down
20 changes: 20 additions & 0 deletions spark/src/main/scala/org/apache/comet/shims/ShimQueryPlanSerde.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -45,6 +45,26 @@ trait ShimQueryPlanSerde {
}
}

// TODO: delete after drop Spark 3.2/3.3 support
// This method is used to check if the aggregate function is in legacy mode.
// EvalMode is an enum object in Spark 3.4.
def isLegacyMode(aggregate: DeclarativeAggregate): Boolean = {
val evalMode = aggregate.getClass.getDeclaredMethods
.flatMap(m =>
m.getName match {
case "evalMode" => Some(m.invoke(aggregate))
case _ => None
})

if (evalMode.isEmpty) {
true
} else {
// scalastyle:off caselocale
evalMode.head.toString.toLowerCase == "legacy"
// scalastyle:on caselocale
viirya marked this conversation as resolved.
Show resolved Hide resolved
}
}

// TODO: delete after drop Spark 3.2 support
def isBloomFilterMightContain(binary: BinaryExpression): Boolean = {
binary.getClass.getName == "org.apache.spark.sql.catalyst.expressions.BloomFilterMightContain"
Expand Down
20 changes: 18 additions & 2 deletions spark/src/test/scala/org/apache/comet/exec/CometExecSuite.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,6 +19,8 @@

package org.apache.comet.exec

import java.time.{Duration, Period}

import scala.collection.JavaConverters._
import scala.collection.mutable
import scala.util.Random
Expand All @@ -38,13 +40,13 @@ import org.apache.spark.sql.execution.{CollectLimitExec, ProjectExec, SQLExecuti
import org.apache.spark.sql.execution.exchange.BroadcastExchangeExec
import org.apache.spark.sql.execution.joins.{BroadcastNestedLoopJoinExec, CartesianProductExec, SortMergeJoinExec}
import org.apache.spark.sql.execution.window.WindowExec
import org.apache.spark.sql.functions.{date_add, expr, sum}
import org.apache.spark.sql.functions.{col, date_add, expr, sum}
import org.apache.spark.sql.internal.SQLConf
import org.apache.spark.sql.internal.SQLConf.SESSION_LOCAL_TIMEZONE
import org.apache.spark.unsafe.types.UTF8String

import org.apache.comet.CometConf
import org.apache.comet.CometSparkSessionExtensions.isSpark34Plus
import org.apache.comet.CometSparkSessionExtensions.{isSpark33Plus, isSpark34Plus}

class CometExecSuite extends CometTestBase {
import testImplicits._
Expand All @@ -58,6 +60,20 @@ class CometExecSuite extends CometTestBase {
}
}

test("try_sum should return null if overflow happens before merging") {
assume(isSpark33Plus, "try_sum is available in Spark 3.3+")
val longDf = Seq(Long.MaxValue, Long.MaxValue, 2).toDF("v")
val yearMonthDf = Seq(Int.MaxValue, Int.MaxValue, 2)
.map(Period.ofMonths)
.toDF("v")
val dayTimeDf = Seq(106751991L, 106751991L, 2L)
.map(Duration.ofDays)
.toDF("v")
Seq(longDf, yearMonthDf, dayTimeDf).foreach { df =>
checkSparkAnswer(df.repartitionByRange(2, col("v")).selectExpr("try_sum(v)"))
}
}

test("Fix corrupted AggregateMode when transforming plan parameters") {
withParquetTable((0 until 5).map(i => (i, i + 1)), "table") {
val df = sql("SELECT * FROM table").groupBy($"_1").agg(sum("_2"))
Expand Down
Loading