[GitHub] [spark] huaxingao commented on a change in pull request #35691: [SPARK-38357][SQL][3.2] Fix StackOverflowError with OR(data filter, partition filter)

[GitBox <gi...@apache.org>]

huaxingao commented on a change in pull request #35691:
URL: https://github.com/apache/spark/pull/35691#discussion_r817422198



##########
File path: sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/PruneFileSourcePartitions.scala
##########
@@ -46,21 +46,27 @@ private[sql] object PruneFileSourcePartitions
   private def getPartitionKeyFiltersAndDataFilters(
       sparkSession: SparkSession,
       relation: LeafNode,
+      dataSchema: StructType,
       partitionSchema: StructType,
       filters: Seq[Expression],
       output: Seq[AttributeReference]): (ExpressionSet, Seq[Expression]) = {
     val normalizedFilters = DataSourceStrategy.normalizeExprs(
       filters.filter(f => f.deterministic && !SubqueryExpression.hasSubquery(f)), output)
     val partitionColumns =
       relation.resolve(partitionSchema, sparkSession.sessionState.analyzer.resolver)
+    val dataColumns =
+      relation.resolve(dataSchema, sparkSession.sessionState.analyzer.resolver)
     val partitionSet = AttributeSet(partitionColumns)
-    val (partitionFilters, dataFilters) = normalizedFilters.partition(f =>
-      f.references.subsetOf(partitionSet)
-    )
-    val extraPartitionFilter =
-      dataFilters.flatMap(extractPredicatesWithinOutputSet(_, partitionSet))

Review comment:
       Yes. The `datafilters` is 
   ```
   (p = 0 AND id > 0) OR (p = 1 AND id = 2)
   ```
    and the `extraPartitionFilter` is 
   ```
   (p = 0) OR (p = 1)
   ```
   Then later on, the condition 
   ```
   if (partitionKeyFilters.nonEmpty || (dataFilters.nonEmpty && scan.dataFilters.isEmpty))
   ```
   is always true. It keeps on going into that block until StackOverFlow.




-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org