gengliangwang commented on a change in pull request #25300: [SPARK-28503][SQL]
Return null result on cast an out-of-range value to a integral type
URL: https://github.com/apache/spark/pull/25300#discussion_r309070794
##########
File path:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/Cast.scala
##########
@@ -456,9 +474,52 @@ case class Cast(child: Expression, dataType: DataType,
timeZoneId: Option[String
case DateType =>
buildCast[Int](_, d => null)
case TimestampType =>
- buildCast[Long](_, t => timestampToLong(t).toInt)
- case x: NumericType =>
- b => x.numeric.asInstanceOf[Numeric[Any]].toInt(b)
+ buildCast[Long](_, t => {
+ val longValue = timestampToLong(t)
+ if (longValue == longValue.toInt) {
+ longValue.toInt
+ } else {
+ null
+ }
+ })
+ case ByteType =>
+ b => b.asInstanceOf[Byte].toInt
+ case ShortType =>
+ b => b.asInstanceOf[Short].toInt
+ case LongType =>
+ buildCast[Long](_, l =>
+ if (l == l.toInt) {
+ l.toInt
+ } else {
+ null
+ }
+ )
+ case x: FloatType =>
+ buildCast[Float](_, f =>
+ if (f <= Int.MaxValue && f >= Int.MinValue) {
+ f.toInt
Review comment:
Oh I thought you were talking about the corner case.
As per the SQL standard in section 9.2
> it is implementation- defined whether the approximation is obtained by
rounding or by truncation
Spark always uses truncation. So I think we can simply follow the previous
behavior on this. It is not the scope of this PR.
----------------------------------------------------------------
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:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
