diff options
author | Reynold Xin <rxin@apache.org> | 2014-06-19 23:58:23 -0700 |
---|---|---|
committer | Reynold Xin <rxin@apache.org> | 2014-06-19 23:58:23 -0700 |
commit | 61756409736a64bd42577782cb7468557fa0b642 (patch) | |
tree | bc3a2f945e76037cf67cef450a77c3cdd7dcc3d0 | |
parent | f479cf3743e416ee08e62806e1b34aff5998ac22 (diff) | |
download | spark-61756409736a64bd42577782cb7468557fa0b642.tar.gz spark-61756409736a64bd42577782cb7468557fa0b642.tar.bz2 spark-61756409736a64bd42577782cb7468557fa0b642.zip |
[SPARK-2210] cast to boolean on boolean value gets turned into NOT((boolean_condition) = 0)
```
explain select cast(cast(key=0 as boolean) as boolean) aaa from src
```
should be
```
[Physical execution plan:]
[Project [(key#10:0 = 0) AS aaa#7]]
[ HiveTableScan [key#10], (MetastoreRelation default, src, None), None]
```
However, it is currently
```
[Physical execution plan:]
[Project [NOT((key#10=0) = 0) AS aaa#7]]
[ HiveTableScan [key#10], (MetastoreRelation default, src, None), None]
```
Author: Reynold Xin <rxin@apache.org>
Closes #1144 from rxin/booleancast and squashes the following commits:
c4e543d [Reynold Xin] [SPARK-2210] boolean cast on boolean value should be removed.
2 files changed, 27 insertions, 2 deletions
diff --git a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/HiveTypeCoercion.scala b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/HiveTypeCoercion.scala index 6d331fb501..c0714bcdd0 100644 --- a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/HiveTypeCoercion.scala +++ b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/HiveTypeCoercion.scala @@ -251,7 +251,9 @@ trait HiveTypeCoercion { def apply(plan: LogicalPlan): LogicalPlan = plan transformAllExpressions { // Skip nodes who's children have not been resolved yet. case e if !e.childrenResolved => e - + // Skip if the type is boolean type already. Note that this extra cast should be removed + // by optimizer.SimplifyCasts. + case Cast(e, BooleanType) if e.dataType == BooleanType => e case Cast(e, BooleanType) => Not(Equals(e, Literal(0))) case Cast(e, dataType) if e.dataType == BooleanType => Cast(If(e, Literal(1), Literal(0)), dataType) diff --git a/sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveTypeCoercionSuite.scala b/sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveTypeCoercionSuite.scala index e030c8ee3d..cc8744c966 100644 --- a/sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveTypeCoercionSuite.scala +++ b/sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveTypeCoercionSuite.scala @@ -17,8 +17,12 @@ package org.apache.spark.sql.hive.execution +import org.apache.spark.sql.catalyst.expressions.{Cast, Equals} +import org.apache.spark.sql.execution.Project +import org.apache.spark.sql.hive.test.TestHive + /** - * A set of tests that validate type promotion rules. + * A set of tests that validate type promotion and coercion rules. */ class HiveTypeCoercionSuite extends HiveComparisonTest { val baseTypes = Seq("1", "1.0", "1L", "1S", "1Y", "'1'") @@ -28,4 +32,23 @@ class HiveTypeCoercionSuite extends HiveComparisonTest { createQueryTest(s"$i + $j", s"SELECT $i + $j FROM src LIMIT 1") } } + + test("[SPARK-2210] boolean cast on boolean value should be removed") { + val q = "select cast(cast(key=0 as boolean) as boolean) from src" + val project = TestHive.hql(q).queryExecution.executedPlan.collect { case e: Project => e }.head + + // No cast expression introduced + project.transformAllExpressions { case c: Cast => + assert(false, "unexpected cast " + c) + c + } + + // Only one Equals + var numEquals = 0 + project.transformAllExpressions { case e: Equals => + numEquals += 1 + e + } + assert(numEquals === 1) + } } |