Skip to content

[SPARK-36647][SQL][TESTS] Push down Aggregate (Min/Max/Count) for Parquet if filter is on partition col #34248

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
wants to merge 5 commits into from
Closed

[SPARK-36647][SQL][TESTS] Push down Aggregate (Min/Max/Count) for Parquet if filter is on partition col #34248

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
cd22629
[SPARK-36647][SQL] Push down Aggregate (Min/Max/Count) for Parquet if…
huaxingao Oct 11, 2021
6cc6f7b
address comments
huaxingao Oct 13, 2021
d7ecec4
address comments
huaxingao Oct 13, 2021
1c96138
fix
huaxingao Oct 13, 2021
1293ae0
fix comment
huaxingao Oct 18, 2021
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
7 changes: 3 additions & 4 deletions ...main/scala/org/apache/spark/sql/execution/datasources/v2/parquet/ParquetScanBuilder.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -130,10 +130,9 @@ case class ParquetScanBuilder(
// are combined with filter or group by
// e.g. SELECT COUNT(col1) FROM t WHERE col2 = 8
// SELECT COUNT(col1) FROM t GROUP BY col2
// Todo: 1. add support if groupby column is partition col
// (https://issues.apache.org/jira/browse/SPARK-36646)
// 2. add support if filter col is partition col
// (https://issues.apache.org/jira/browse/SPARK-36647)
// However, if the filter is on partition column, max/min/count can still be pushed down
// Todo: add support if groupby column is partition col
// (https://issues.apache.org/jira/browse/SPARK-36646)
return false
}

Expand Down
40 changes: 30 additions & 10 deletions ...la/org/apache/spark/sql/execution/datasources/parquet/ParquetAggregatePushDownSuite.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -129,10 +129,9 @@ abstract class ParquetAggregatePushDownSuite
.write.partitionBy("p").parquet(dir.getCanonicalPath)
withTempView("tmp") {
spark.read.parquet(dir.getCanonicalPath).createOrReplaceTempView("tmp");
val enableVectorizedReader = Seq("false", "true")
for (testVectorizedReader <- enableVectorizedReader) {
Seq("false", "true").foreach { enableVectorizedReader =>
withSQLConf(SQLConf.PARQUET_AGGREGATE_PUSHDOWN_ENABLED.key -> "true",
vectorizedReaderEnabledKey -> testVectorizedReader) {
vectorizedReaderEnabledKey -> enableVectorizedReader) {
val count = sql("SELECT COUNT(p) FROM tmp")
count.queryExecution.optimizedPlan.collect {
case _: DataSourceV2ScanRelation =>
Expand Down Expand Up @@ -221,7 +220,7 @@ abstract class ParquetAggregatePushDownSuite
}
}

test("aggregate push down - query with filter not push down") {
test("aggregate push down - aggregate with data filter cannot be pushed down") {
val data = Seq((-2, "abc", 2), (3, "def", 4), (6, "ghi", 2), (0, null, 19),
(9, "mno", 7), (2, null, 7))
withParquetTable(data, "t") {
Expand All @@ -240,6 +239,29 @@ abstract class ParquetAggregatePushDownSuite
}
}

test("aggregate push down - aggregate with partition filter can be pushed down") {
withTempPath { dir =>
spark.range(10).selectExpr("id", "id % 3 as p")
.write.partitionBy("p").parquet(dir.getCanonicalPath)
withTempView("tmp") {
spark.read.parquet(dir.getCanonicalPath).createOrReplaceTempView("tmp");
Seq("false", "true").foreach { enableVectorizedReader =>
withSQLConf(SQLConf.PARQUET_AGGREGATE_PUSHDOWN_ENABLED.key -> "true",
vectorizedReaderEnabledKey -> enableVectorizedReader) {
val max = sql("SELECT max(id), min(id), count(id) FROM tmp WHERE p = 0")
max.queryExecution.optimizedPlan.collect {
case _: DataSourceV2ScanRelation =>
val expected_plan_fragment =
"PushedAggregation: [MAX(id), MIN(id), COUNT(id)]"
checkKeywordsExistsInExplain(max, expected_plan_fragment)
}
checkAnswer(max, Seq(Row(9, 0, 4)))
}
}
}
}
}

test("aggregate push down - push down only if all the aggregates can be pushed down") {
val data = Seq((-2, "abc", 2), (3, "def", 4), (6, "ghi", 2), (0, null, 19),
(9, "mno", 7), (2, null, 7))
Expand Down Expand Up @@ -356,10 +378,9 @@ abstract class ParquetAggregatePushDownSuite
spark.createDataFrame(rdd, schema).write.parquet(file.getCanonicalPath)
withTempView("test") {
spark.read.parquet(file.getCanonicalPath).createOrReplaceTempView("test")
val enableVectorizedReader = Seq("false", "true")
for (testVectorizedReader <- enableVectorizedReader) {
Seq("false", "true").foreach { enableVectorizedReader =>
withSQLConf(SQLConf.PARQUET_AGGREGATE_PUSHDOWN_ENABLED.key -> "true",
vectorizedReaderEnabledKey -> testVectorizedReader) {
vectorizedReaderEnabledKey -> enableVectorizedReader) {

val testMinWithTS = sql("SELECT min(StringCol), min(BooleanCol), min(ByteCol), " +
"min(BinaryCol), min(ShortCol), min(IntegerCol), min(LongCol), min(FloatCol), " +
Expand Down Expand Up @@ -477,10 +498,9 @@ abstract class ParquetAggregatePushDownSuite
}

test("aggregate push down - column name case sensitivity") {
val enableVectorizedReader = Seq("false", "true")
for (testVectorizedReader <- enableVectorizedReader) {
Seq("false", "true").foreach { enableVectorizedReader =>
withSQLConf(SQLConf.PARQUET_AGGREGATE_PUSHDOWN_ENABLED.key -> "true",
vectorizedReaderEnabledKey -> testVectorizedReader) {
vectorizedReaderEnabledKey -> enableVectorizedReader) {
withTempPath { dir =>
spark.range(10).selectExpr("id", "id % 3 as p")
.write.partitionBy("p").parquet(dir.getCanonicalPath)
Expand Down