diff options
author | Takuya UESHIN <ueshin@happy-camper.st> | 2014-11-17 16:28:07 -0800 |
---|---|---|
committer | Michael Armbrust <michael@databricks.com> | 2014-11-17 16:28:07 -0800 |
commit | 566c791931645bfaaaf57ee5a15b9ffad534f81e (patch) | |
tree | befa3c51205752bdeb7d80d5888c0fbe2827b117 /sql | |
parent | 3a81a1c9e0963173534d96850f3c0b7a16350838 (diff) | |
download | spark-566c791931645bfaaaf57ee5a15b9ffad534f81e.tar.gz spark-566c791931645bfaaaf57ee5a15b9ffad534f81e.tar.bz2 spark-566c791931645bfaaaf57ee5a15b9ffad534f81e.zip |
[SPARK-4425][SQL] Handle NaN or Infinity cast to Timestamp correctly.
`Cast` from `NaN` or `Infinity` of `Double` or `Float` to `TimestampType` throws `NumberFormatException`.
Author: Takuya UESHIN <ueshin@happy-camper.st>
Closes #3283 from ueshin/issues/SPARK-4425 and squashes the following commits:
14def0c [Takuya UESHIN] Fix Cast to be able to handle NaN or Infinity to TimestampType.
Diffstat (limited to 'sql')
2 files changed, 17 insertions, 2 deletions
diff --git a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/Cast.scala b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/Cast.scala index b401096ce1..b47865f87a 100644 --- a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/Cast.scala +++ b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/Cast.scala @@ -32,6 +32,8 @@ case class Cast(child: Expression, dataType: DataType) extends UnaryExpression w override def nullable = (child.dataType, dataType) match { case (StringType, _: NumericType) => true case (StringType, TimestampType) => true + case (DoubleType, TimestampType) => true + case (FloatType, TimestampType) => true case (StringType, DateType) => true case (_: NumericType, DateType) => true case (BooleanType, DateType) => true @@ -117,10 +119,18 @@ case class Cast(child: Expression, dataType: DataType) extends UnaryExpression w buildCast[Decimal](_, d => decimalToTimestamp(d)) // TimestampWritable.doubleToTimestamp case DoubleType => - buildCast[Double](_, d => decimalToTimestamp(Decimal(d))) + buildCast[Double](_, d => try { + decimalToTimestamp(Decimal(d)) + } catch { + case _: NumberFormatException => null + }) // TimestampWritable.floatToTimestamp case FloatType => - buildCast[Float](_, f => decimalToTimestamp(Decimal(f))) + buildCast[Float](_, f => try { + decimalToTimestamp(Decimal(f)) + } catch { + case _: NumberFormatException => null + }) } private[this] def decimalToTimestamp(d: Decimal) = { diff --git a/sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/expressions/ExpressionEvaluationSuite.scala b/sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/expressions/ExpressionEvaluationSuite.scala index 3a6a0203af..3f5b9f698f 100644 --- a/sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/expressions/ExpressionEvaluationSuite.scala +++ b/sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/expressions/ExpressionEvaluationSuite.scala @@ -450,6 +450,11 @@ class ExpressionEvaluationSuite extends FunSuite { // A test for higher precision than millis checkEvaluation(Cast(Cast(0.00000001, TimestampType), DoubleType), 0.00000001) + + checkEvaluation(Cast(Literal(Double.NaN), TimestampType), null) + checkEvaluation(Cast(Literal(1.0 / 0.0), TimestampType), null) + checkEvaluation(Cast(Literal(Float.NaN), TimestampType), null) + checkEvaluation(Cast(Literal(1.0f / 0.0f), TimestampType), null) } test("null checking") { |