You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@spark.apache.org by we...@apache.org on 2025/05/16 09:20:27 UTC
(spark) branch master updated: [SPARK-52157][SQL] Use earlier computed name for OuterReference in single-pass resolver
This is an automated email from the ASF dual-hosted git repository.
wenchen pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/spark.git
The following commit(s) were added to refs/heads/master by this push:
new 71a046b73abe [SPARK-52157][SQL] Use earlier computed name for OuterReference in single-pass resolver
71a046b73abe is described below
commit 71a046b73abe7ac244afa4cd09d80dffb133509d
Author: mihailoale-db <mi...@databricks.com>
AuthorDate: Fri May 16 11:20:14 2025 +0200
[SPARK-52157][SQL] Use earlier computed name for OuterReference in single-pass resolver
### What changes were proposed in this pull request?
In the following query we would have min(outer(t2.t2a)) as a name for min(t2a) expression.
```
SELECT t1a
FROM t1
WHERE t1a IN (SELECT t2a
FROM t2
WHERE EXISTS (SELECT min(t2a)
FROM t3))
```
This is a problem in compatibility between single-pass resolver and fixed-point analyzer because names in single-pass are generated after we finish resolution of aggregate expression min(t2a) (bottom-up manner) and at that point we have OuterReference wrapped around aggregate expression (name looks like outer(min(t2a))).
In this PR I propose that we store the name earlier (in a form of a tag) and use it when needed
### Why are the changes needed?
To make single-pass and fixed-point compatible.
### Does this PR introduce _any_ user-facing change?
No.
### How was this patch tested?
Existing tests.
### Was this patch authored or co-authored using generative AI tooling?
No.
Closes #50903 from mihailoale-db/outerreferencetag.
Authored-by: mihailoale-db <mi...@databricks.com>
Signed-off-by: Wenchen Fan <we...@databricks.com>
---
.../resolver/AggregateExpressionResolver.scala | 42 ++++++++++++++--------
.../catalyst/expressions/namedExpressions.scala | 26 ++++++++++++--
2 files changed, 52 insertions(+), 16 deletions(-)
diff --git a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/resolver/AggregateExpressionResolver.scala b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/resolver/AggregateExpressionResolver.scala
index 63642e3ab1f1..c13a01115429 100644
--- a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/resolver/AggregateExpressionResolver.scala
+++ b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/resolver/AggregateExpressionResolver.scala
@@ -29,6 +29,7 @@ import org.apache.spark.sql.catalyst.analysis.{
import org.apache.spark.sql.catalyst.expressions.{Alias, Expression, OuterReference, SubExprUtils}
import org.apache.spark.sql.catalyst.expressions.aggregate.AggregateExpression
import org.apache.spark.sql.catalyst.plans.logical.{Aggregate, Sort}
+import org.apache.spark.sql.catalyst.util.toPrettySQL
import org.apache.spark.sql.errors.QueryCompilationErrors
/**
@@ -139,7 +140,9 @@ class AggregateExpressionResolver(
* - Alias this subtree and put it inside the current [[SubqueryScope]];
* - If outer aggregates are allowed, replace the [[AggregateExpression]] with an
* [[OuterReference]] to the auto-generated [[Alias]] that we created. This alias will later
- * be injected into the outer [[Aggregate]];
+ * be injected into the outer [[Aggregate]]; We store the name that needs to be used for the
+ * [[OuterReference]] in [[OuterReference.SINGLE_PASS_SQL_STRING_OVERRIDE]] computed based on
+ * the [[AggregateExpression]] without [[OuterReference]] pulled out.
* - In case we have an [[AggregateExpression]] inside a [[Sort]] operator, we need to handle it
* in a special way (see [[handleAggregateExpressionInSort]] for more details).
* - Return the original [[AggregateExpression]] otherwise. This is done to stay compatible
@@ -155,21 +158,32 @@ class AggregateExpressionResolver(
)
}
- if (subqueryRegistry.currentScope.isOuterAggregateAllowed) {
- val aggregateExpressionWithStrippedOuterReferences =
- SubExprUtils.stripOuterReference(aggregateExpression)
+ val resolvedOuterAggregateExpression =
+ if (subqueryRegistry.currentScope.isOuterAggregateAllowed) {
+ val aggregateExpressionWithStrippedOuterReferences =
+ SubExprUtils.stripOuterReference(aggregateExpression)
- val outerAggregateExpressionAlias = autoGeneratedAliasProvider.newOuterAlias(
- child = aggregateExpressionWithStrippedOuterReferences
- )
- subqueryRegistry.currentScope.addOuterAggregateExpression(
- outerAggregateExpressionAlias,
- aggregateExpressionWithStrippedOuterReferences
- )
+ val outerAggregateExpressionAlias = autoGeneratedAliasProvider.newOuterAlias(
+ child = aggregateExpressionWithStrippedOuterReferences
+ )
+ subqueryRegistry.currentScope.addOuterAggregateExpression(
+ outerAggregateExpressionAlias,
+ aggregateExpressionWithStrippedOuterReferences
+ )
- OuterReference(outerAggregateExpressionAlias.toAttribute)
- } else {
- aggregateExpression
+ OuterReference(outerAggregateExpressionAlias.toAttribute)
+ } else {
+ aggregateExpression
+ }
+
+ resolvedOuterAggregateExpression match {
+ case outerReference: OuterReference =>
+ outerReference.setTagValue(
+ OuterReference.SINGLE_PASS_SQL_STRING_OVERRIDE,
+ toPrettySQL(aggregateExpression)
+ )
+ outerReference
+ case other => other
}
}
diff --git a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/namedExpressions.scala b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/namedExpressions.scala
index 1d0082582114..0bfba04d6372 100644
--- a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/namedExpressions.scala
+++ b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/namedExpressions.scala
@@ -24,7 +24,7 @@ import org.apache.spark.sql.catalyst.InternalRow
import org.apache.spark.sql.catalyst.analysis.UnresolvedAttribute
import org.apache.spark.sql.catalyst.expressions.codegen._
import org.apache.spark.sql.catalyst.plans.logical.EventTimeWatermark
-import org.apache.spark.sql.catalyst.trees.TreePattern
+import org.apache.spark.sql.catalyst.trees.{TreeNodeTag, TreePattern}
import org.apache.spark.sql.catalyst.trees.TreePattern._
import org.apache.spark.sql.catalyst.types._
import org.apache.spark.sql.catalyst.util.{quoteIfNeeded, METADATA_COL_ATTR_KEY}
@@ -443,7 +443,13 @@ case class OuterReference(e: NamedExpression)
override def nullable: Boolean = e.nullable
override def prettyName: String = "outer"
- override def sql: String = s"$prettyName(${e.sql})"
+ override def sql: String =
+ getTagValue(OuterReference.SINGLE_PASS_SQL_STRING_OVERRIDE) match {
+ case Some(name) =>
+ name
+ case None =>
+ s"$prettyName(${e.sql})"
+ }
override def name: String = e.name
override def qualifier: Seq[String] = e.qualifier
override def exprId: ExprId = e.exprId
@@ -510,6 +516,22 @@ case class LateralColumnAliasReference(ne: NamedExpression, nameParts: Seq[Strin
final override val nodePatterns: Seq[TreePattern] = Seq(LATERAL_COLUMN_ALIAS_REFERENCE)
}
+object OuterReference {
+
+ /**
+ * In fixed-point [[OuterReference]] is extracted in [[UpdateOuterReferences]] which is invoked
+ * after the alias assignment ([[ResolveAliases]]) which is opposite of how it is done in the
+ * single-pass implementation (first we extract the [[OuterReference]] and then assign the name -
+ * in bottom-up manner).
+ *
+ * In order to make single-pass and fixed-point implementations compatible use earlier computed
+ * name (if defined) for [[OuterReference]] (defined in
+ * `AggregateExpressionResolver.handleOuterAggregateExpression`).
+ */
+ val SINGLE_PASS_SQL_STRING_OVERRIDE =
+ TreeNodeTag[String]("single_pass_sql_string_override")
+}
+
object VirtualColumn {
// The attribute name used by Hive, which has different result than Spark, deprecated.
val hiveGroupingIdName: String = "grouping__id"
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@spark.apache.org
For additional commands, e-mail: commits-help@spark.apache.org