[SPARK-44340][SPARK-44341][SQL][PYTHON][FOLLOWUP] Set partition index…

… correctly for WindowGroupLimitExec,WindowExec and WindowInPandasExec ### What changes were proposed in this pull request? This is a followup of apache#41899 and apache#41939, to set the partition index correctly even if it's not used for now. It also contains a few code cleanup. ### Why are the changes needed? future-proof ### Does this PR introduce _any_ user-facing change? 'No'. ### How was this patch tested? existing tests Closes apache#42208 from beliefer/SPARK-44340_followup. Authored-by: Jiaan Geng <beliefer@163.com> Signed-off-by: Wenchen Fan <wenchen@databricks.com>
nart4hire · Jul 31, 2023 · ea3061b · ea3061b
1 parent 32498b3
commit ea3061b
Show file tree

Hide file tree

Showing 3 changed files with 9 additions and 15 deletions.
diff --git a/sql/core/src/main/scala/org/apache/spark/sql/execution/python/WindowInPandasExec.scala b/sql/core/src/main/scala/org/apache/spark/sql/execution/python/WindowInPandasExec.scala
@@ -80,24 +80,22 @@ case class WindowInPandasExec(
   )
 
   protected override def doExecute(): RDD[InternalRow] = {
-    val spillSize = longMetric("spillSize")
-
     val evaluatorFactory =
       new WindowInPandasEvaluatorFactory(
         windowExpression,
         partitionSpec,
         orderSpec,
         child.output,
-        spillSize,
+        longMetric("spillSize"),
         pythonMetrics)
 
     // Start processing.
     if (conf.usePartitionEvaluator) {
       child.execute().mapPartitionsWithEvaluator(evaluatorFactory)
     } else {
-      child.execute().mapPartitions { iter =>
+      child.execute().mapPartitionsWithIndex { (index, rowIterator) =>
         val evaluator = evaluatorFactory.createEvaluator()
-        evaluator.eval(0, iter)
+        evaluator.eval(index, rowIterator)
       }
     }
   }

diff --git a/sql/core/src/main/scala/org/apache/spark/sql/execution/window/WindowExec.scala b/sql/core/src/main/scala/org/apache/spark/sql/execution/window/WindowExec.scala
@@ -95,23 +95,21 @@ case class WindowExec(
   )
 
   protected override def doExecute(): RDD[InternalRow] = {
-    val spillSize = longMetric("spillSize")
-
     val evaluatorFactory =
       new WindowEvaluatorFactory(
         windowExpression,
         partitionSpec,
         orderSpec,
         child.output,
-        spillSize)
+        longMetric("spillSize"))
 
     // Start processing.
     if (conf.usePartitionEvaluator) {
       child.execute().mapPartitionsWithEvaluator(evaluatorFactory)
     } else {
-      child.execute().mapPartitions { iter =>
+      child.execute().mapPartitionsWithIndex { (index, rowIterator) =>
         val evaluator = evaluatorFactory.createEvaluator()
-        evaluator.eval(0, iter)
+        evaluator.eval(index, rowIterator)
       }
     }
   }

diff --git a/sql/core/src/main/scala/org/apache/spark/sql/execution/window/WindowGroupLimitExec.scala b/sql/core/src/main/scala/org/apache/spark/sql/execution/window/WindowGroupLimitExec.scala
@@ -72,23 +72,21 @@ case class WindowGroupLimitExec(
     "numOutputRows" -> SQLMetrics.createMetric(sparkContext, "number of output rows"))
 
   protected override def doExecute(): RDD[InternalRow] = {
-    val numOutputRows = longMetric("numOutputRows")
-
     val evaluatorFactory =
       new WindowGroupLimitEvaluatorFactory(
         partitionSpec,
         orderSpec,
         rankLikeFunction,
         limit,
         child.output,
-        numOutputRows)
+        longMetric("numOutputRows"))
 
     if (conf.usePartitionEvaluator) {
       child.execute().mapPartitionsWithEvaluator(evaluatorFactory)
     } else {
-      child.execute().mapPartitionsInternal { iter =>
+      child.execute().mapPartitionsWithIndexInternal { (index, rowIterator) =>
         val evaluator = evaluatorFactory.createEvaluator()
-        evaluator.eval(0, iter)
+        evaluator.eval(index, rowIterator)
       }
     }
   }