Skip to content

[SPARK-39964][SQL] DS V2 pushdown should unify the translate path #37391

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
beliefer wants to merge 4 commits into from
Closed

[SPARK-39964][SQL] DS V2 pushdown should unify the translate path #37391

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
f10fcdf
[SPARK-39964][SQL] DS V2 pushdown should unify the translate API
beliefer Aug 3, 2022
51399f0
Update code
beliefer Aug 4, 2022
84365ec
Update sql/core/src/main/scala/org/apache/spark/sql/catalyst/util/V2E…
beliefer Aug 4, 2022
328539d
Update code
beliefer Aug 4, 2022
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
49 changes: 49 additions & 0 deletions sql/core/src/main/scala/org/apache/spark/sql/catalyst/util/V2ExpressionBuilder.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,8 +18,11 @@
package org.apache.spark.sql.catalyst.util

import org.apache.spark.sql.catalyst.expressions._
import org.apache.spark.sql.catalyst.expressions.aggregate.AggregateExpression
import org.apache.spark.sql.connector.expressions.{Cast => V2Cast, Expression => V2Expression, Extract => V2Extract, FieldReference, GeneralScalarExpression, LiteralValue, UserDefinedScalarFunc}
import org.apache.spark.sql.connector.expressions.aggregate.{Avg, Count, CountStar, GeneralAggregateFunc, Max, Min, Sum, UserDefinedAggregateFunc}
import org.apache.spark.sql.connector.expressions.filter.{AlwaysFalse, AlwaysTrue, And => V2And, Not => V2Not, Or => V2Or, Predicate => V2Predicate}
import org.apache.spark.sql.execution.datasources.PushableExpression
import org.apache.spark.sql.types.{BooleanType, IntegerType}

/**
Expand Down Expand Up @@ -90,6 +93,52 @@ class V2ExpressionBuilder(e: Expression, isPredicate: Boolean = false) {
}
case Cast(child, dataType, _, true) =>
generateExpression(child).map(v => new V2Cast(v, dataType))
case AggregateExpression(aggregateFunction, _, isDistinct, None, _) =>
aggregateFunction match {
Copy link
Contributor

@cloud-fan cloud-fan Aug 3, 2022

Choose a reason for hiding this comment

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

can we put the new code in a private method? The current method is already super long.

Copy link
Contributor

@huaxingao huaxingao Aug 3, 2022

Choose a reason for hiding this comment

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

+1

case aggregate.Min(PushableExpression(expr)) => Some(new Min(expr))
case aggregate.Max(PushableExpression(expr)) => Some(new Max(expr))
case count: aggregate.Count if count.children.length == 1 =>
count.children.head match {
// COUNT(any literal) is the same as COUNT(*)
case Literal(_, _) => Some(new CountStar())
case PushableExpression(expr) => Some(new Count(expr, isDistinct))
case _ => None
}
case aggregate.Sum(PushableExpression(expr), _) => Some(new Sum(expr, isDistinct))
case aggregate.Average(PushableExpression(expr), _) => Some(new Avg(expr, isDistinct))
case aggregate.VariancePop(PushableExpression(expr), _) =>
Some(new GeneralAggregateFunc("VAR_POP", isDistinct, Array(expr)))
case aggregate.VarianceSamp(PushableExpression(expr), _) =>
Some(new GeneralAggregateFunc("VAR_SAMP", isDistinct, Array(expr)))
case aggregate.StddevPop(PushableExpression(expr), _) =>
Some(new GeneralAggregateFunc("STDDEV_POP", isDistinct, Array(expr)))
case aggregate.StddevSamp(PushableExpression(expr), _) =>
Some(new GeneralAggregateFunc("STDDEV_SAMP", isDistinct, Array(expr)))
case aggregate.CovPopulation(PushableExpression(left), PushableExpression(right), _) =>
Some(new GeneralAggregateFunc("COVAR_POP", isDistinct, Array(left, right)))
case aggregate.CovSample(PushableExpression(left), PushableExpression(right), _) =>
Some(new GeneralAggregateFunc("COVAR_SAMP", isDistinct, Array(left, right)))
case aggregate.Corr(PushableExpression(left), PushableExpression(right), _) =>
Some(new GeneralAggregateFunc("CORR", isDistinct, Array(left, right)))
case aggregate.RegrIntercept(PushableExpression(left), PushableExpression(right)) =>
Some(new GeneralAggregateFunc("REGR_INTERCEPT", isDistinct, Array(left, right)))
case aggregate.RegrR2(PushableExpression(left), PushableExpression(right)) =>
Some(new GeneralAggregateFunc("REGR_R2", isDistinct, Array(left, right)))
case aggregate.RegrSlope(PushableExpression(left), PushableExpression(right)) =>
Some(new GeneralAggregateFunc("REGR_SLOPE", isDistinct, Array(left, right)))
case aggregate.RegrSXY(PushableExpression(left), PushableExpression(right)) =>
Some(new GeneralAggregateFunc("REGR_SXY", isDistinct, Array(left, right)))
// TODO supports other aggregate functions
case aggregate.V2Aggregator(aggrFunc, children, _, _) =>
val translatedExprs = children.flatMap(PushableExpression.unapply(_))
if (translatedExprs.length == children.length) {
Some(new UserDefinedAggregateFunc(aggrFunc.name(),
aggrFunc.canonicalName(), isDistinct, translatedExprs.toArray[V2Expression]))
} else {
None
}
case _ => None
}
case Abs(child, true) => generateExpressionWithName("ABS", Seq(child))
case Coalesce(children) => generateExpressionWithName("COALESCE", children)
case Greatest(children) => generateExpressionWithName("GREATEST", children)
Expand Down
59 changes: 4 additions & 55 deletions sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/DataSourceStrategy.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -42,7 +42,7 @@ import org.apache.spark.sql.catalyst.util.{ResolveDefaultColumns, V2ExpressionBu
import org.apache.spark.sql.connector.catalog.SupportsRead
import org.apache.spark.sql.connector.catalog.TableCapability._
import org.apache.spark.sql.connector.expressions.{Expression => V2Expression, NullOrdering, SortDirection, SortOrder => V2SortOrder, SortValue}
import org.apache.spark.sql.connector.expressions.aggregate.{AggregateFunc, Aggregation, Avg, Count, CountStar, GeneralAggregateFunc, Max, Min, Sum, UserDefinedAggregateFunc}
import org.apache.spark.sql.connector.expressions.aggregate.{AggregateFunc, Aggregation}
import org.apache.spark.sql.errors.QueryCompilationErrors
import org.apache.spark.sql.execution.{RowDataSourceScanExec, SparkPlan}
import org.apache.spark.sql.execution.command._
Expand Down Expand Up @@ -694,57 +694,6 @@ object DataSourceStrategy
(nonconvertiblePredicates ++ unhandledPredicates, pushedFilters, handledFilters)
}

protected[sql] def translateAggregate(agg: AggregateExpression): Option[AggregateFunc] = {
if (agg.filter.isEmpty) {
agg.aggregateFunction match {
case aggregate.Min(PushableExpression(expr)) => Some(new Min(expr))
case aggregate.Max(PushableExpression(expr)) => Some(new Max(expr))
case count: aggregate.Count if count.children.length == 1 =>
count.children.head match {
// COUNT(any literal) is the same as COUNT(*)
case Literal(_, _) => Some(new CountStar())
case PushableExpression(expr) => Some(new Count(expr, agg.isDistinct))
case _ => None
}
case aggregate.Sum(PushableExpression(expr), _) => Some(new Sum(expr, agg.isDistinct))
case aggregate.Average(PushableExpression(expr), _) => Some(new Avg(expr, agg.isDistinct))
case aggregate.VariancePop(PushableExpression(expr), _) =>
Some(new GeneralAggregateFunc("VAR_POP", agg.isDistinct, Array(expr)))
case aggregate.VarianceSamp(PushableExpression(expr), _) =>
Some(new GeneralAggregateFunc("VAR_SAMP", agg.isDistinct, Array(expr)))
case aggregate.StddevPop(PushableExpression(expr), _) =>
Some(new GeneralAggregateFunc("STDDEV_POP", agg.isDistinct, Array(expr)))
case aggregate.StddevSamp(PushableExpression(expr), _) =>
Some(new GeneralAggregateFunc("STDDEV_SAMP", agg.isDistinct, Array(expr)))
case aggregate.CovPopulation(PushableExpression(left), PushableExpression(right), _) =>
Some(new GeneralAggregateFunc("COVAR_POP", agg.isDistinct, Array(left, right)))
case aggregate.CovSample(PushableExpression(left), PushableExpression(right), _) =>
Some(new GeneralAggregateFunc("COVAR_SAMP", agg.isDistinct, Array(left, right)))
case aggregate.Corr(PushableExpression(left), PushableExpression(right), _) =>
Some(new GeneralAggregateFunc("CORR", agg.isDistinct, Array(left, right)))
case aggregate.RegrIntercept(PushableExpression(left), PushableExpression(right)) =>
Some(new GeneralAggregateFunc("REGR_INTERCEPT", agg.isDistinct, Array(left, right)))
case aggregate.RegrR2(PushableExpression(left), PushableExpression(right)) =>
Some(new GeneralAggregateFunc("REGR_R2", agg.isDistinct, Array(left, right)))
case aggregate.RegrSlope(PushableExpression(left), PushableExpression(right)) =>
Some(new GeneralAggregateFunc("REGR_SLOPE", agg.isDistinct, Array(left, right)))
case aggregate.RegrSXY(PushableExpression(left), PushableExpression(right)) =>
Some(new GeneralAggregateFunc("REGR_SXY", agg.isDistinct, Array(left, right)))
case aggregate.V2Aggregator(aggrFunc, children, _, _) =>
val translatedExprs = children.flatMap(PushableExpression.unapply(_))
if (translatedExprs.length == children.length) {
Some(new UserDefinedAggregateFunc(aggrFunc.name(),
aggrFunc.canonicalName(), agg.isDistinct, translatedExprs.toArray[V2Expression]))
} else {
None
}
case _ => None
}
} else {
None
}
}

/**
* Translate aggregate expressions and group by expressions.
*
Expand All @@ -753,13 +702,13 @@ object DataSourceStrategy
protected[sql] def translateAggregation(
aggregates: Seq[AggregateExpression], groupBy: Seq[Expression]): Option[Aggregation] = {

def translateGroupBy(e: Expression): Option[V2Expression] = e match {
def translate(e: Expression): Option[V2Expression] = e match {
case PushableExpression(expr) => Some(expr)
case _ => None
}

val translatedAggregates = aggregates.flatMap(translateAggregate)
val translatedGroupBys = groupBy.flatMap(translateGroupBy)
val translatedAggregates = aggregates.flatMap(translate).asInstanceOf[Seq[AggregateFunc]]
val translatedGroupBys = groupBy.flatMap(translate)

if (translatedAggregates.length != aggregates.length ||
translatedGroupBys.length != groupBy.length) {
Expand Down