You are viewing a plain text version of this content. The canonical link for it is here.
Posted to reviews@spark.apache.org by GitBox <gi...@apache.org> on 2020/09/22 05:06:35 UTC
[GitHub] [spark] liancheng commented on a change in pull request #29831: [SPARK-32351][SQL] pushdown partition filter in FileSourceScanExec and HiveTableScanExec
liancheng commented on a change in pull request #29831:
URL: https://github.com/apache/spark/pull/29831#discussion_r492476018
##########
File path: sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/PrunePartitionSuiteBase.scala
##########
@@ -46,30 +46,42 @@ abstract class PrunePartitionSuiteBase extends QueryTest with SQLTestUtils with
}
assertPrunedPartitions(
- "SELECT * FROM t WHERE p = '1' OR (p = '2' AND i = 1)", 2)
+ "SELECT * FROM t WHERE p = '1' OR (p = '2' AND i = 1)", 2, 1)
assertPrunedPartitions(
- "SELECT * FROM t WHERE (p = '1' AND i = 2) OR (i = 1 OR p = '2')", 4)
+ "SELECT * FROM t WHERE (p = '1' AND i = 2) OR (i = 1 OR p = '2')", 4, 0)
assertPrunedPartitions(
- "SELECT * FROM t WHERE (p = '1' AND i = 2) OR (p = '3' AND i = 3 )", 2)
+ "SELECT * FROM t WHERE (p = '1' AND i = 2) OR (p = '3' AND i = 3 )", 2, 1)
assertPrunedPartitions(
- "SELECT * FROM t WHERE (p = '1' AND i = 2) OR (p = '2' OR p = '3')", 3)
+ "SELECT * FROM t WHERE (p = '1' AND i = 2) OR (p = '2' OR p = '3')", 3, 1)
assertPrunedPartitions(
- "SELECT * FROM t", 4)
+ "SELECT * FROM t", 4, expectedPushedDownFilterCount = 0)
assertPrunedPartitions(
- "SELECT * FROM t WHERE p = '1' AND i = 2", 1)
+ "SELECT * FROM t WHERE p = '1' AND i = 2", 1, 2)
assertPrunedPartitions(
"""
|SELECT i, COUNT(1) FROM (
|SELECT * FROM t WHERE p = '1' OR (p = '2' AND i = 1)
|) tmp GROUP BY i
- """.stripMargin, 2)
+ """.stripMargin, 2, 1)
}
}
}
- protected def assertPrunedPartitions(query: String, expected: Long): Unit = {
- val plan = sql(query).queryExecution.sparkPlan
- assert(getScanExecPartitionSize(plan) == expected)
+ protected def assertPrunedPartitions(
+ query: String,
+ expectedPartitionCount: Long,
+ expectedPushedDownFilterCount: Int): Unit = {
+ val qe = sql(query).queryExecution
+ val plan = qe.sparkPlan
+ assert(getScanExecPartitionSize(plan) == expectedPartitionCount)
+ val pushedDownPartitionFilters = qe.executedPlan.collectFirst {
+ case FileSourceScanExec(_, _, _, partitionFilters, _, _, _, _) =>
+ partitionFilters
+ case HiveTableScanExec(_, _, partitionFilters) =>
+ partitionFilters
Review comment:
```suggestion
case scan: FileSourceScanExec => scan.partitionFilters
case scan: HiveTableScanExec => scan.partitionFilters
```
I'd prefer the above style for the following reasons:
1. It's more resilient to future physical plan changes, e.g., adding or removing fields.
2. Shorter.
----------------------------------------------------------------
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.
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