MaxGekk commented on code in PR #47154:
URL: https://github.com/apache/spark/pull/47154#discussion_r1738971754
##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/Mode.scala:
##########
@@ -50,17 +50,19 @@ case class Mode(
override def inputTypes: Seq[AbstractDataType] = Seq(AnyDataType)
override def checkInputDataTypes(): TypeCheckResult = {
- if (UnsafeRowUtils.isBinaryStable(child.dataType) ||
child.dataType.isInstanceOf[StringType]) {
+ // TODO: SPARK-49358: Mode expression for map type with collated fields
+ if (UnsafeRowUtils.isBinaryStable(child.dataType) ||
+ !child.dataType.existsRecursively(f => f.isInstanceOf[MapType] &&
+ !UnsafeRowUtils.isBinaryStable(f))) {
/*
* The Mode class uses collation awareness logic to handle string data.
- * Complex types with collated fields are not yet supported.
+ * All complex types except MapType with collated fields are supported.
*/
- // TODO: SPARK-48700: Mode expression for complex types (all collations)
super.checkInputDataTypes()
} else {
- TypeCheckResult.TypeCheckFailure("The input to the function 'mode' was" +
- " a type of binary-unstable type that is " +
- s"not currently supported by ${prettyName}.")
+ TypeCheckResult.TypeCheckFailure("The input to the function 'mode'
includes" +
Review Comment:
It would be better to avoid using the legacy `TypeCheckFailure` and switch
to `DataTypeMismatch`.
Please, add a sub-class to `DATATYPE_MISMATCH`, and return
`TypeCheckResult.DataTypeMismatch` here.
##########
sql/core/src/test/scala/org/apache/spark/sql/CollationSQLExpressionsSuite.scala:
##########
@@ -1744,6 +1778,7 @@ class CollationSQLExpressionsSuite
ModeTestCase("unicode", Map("a" -> 3L, "b" -> 2L, "B" -> 2L), "a"),
ModeTestCase("unicode_ci", Map("a" -> 3L, "b" -> 2L, "B" -> 2L), "b")
)
+
Review Comment:
An unnecessary change
##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/aggregate/Mode.scala:
##########
@@ -86,6 +88,49 @@ case class Mode(
buffer
}
+ private def getCollationAwareBuffer(
+ childDataType: DataType,
+ buffer: OpenHashMap[AnyRef, Long]): Iterable[(AnyRef, Long)] = {
+ def groupAndReduceBuffer(groupingFunction: AnyRef => _): Iterable[(AnyRef,
Long)] = {
+ buffer.groupMapReduce(t =>
+ groupingFunction(t._1))(x => x)((x, y) => (x._1, x._2 + y._2)).values
+ }
+ def determineBufferingFunction(
+ childDataType: DataType): Option[AnyRef => _] = {
+ childDataType match {
+ case _ if UnsafeRowUtils.isBinaryStable(child.dataType) => None
+ case _ => Some(collationAwareTransform(_, childDataType))
+ }
+ }
+
determineBufferingFunction(childDataType).map(groupAndReduceBuffer).getOrElse(buffer)
+ }
+
+ private def collationAwareTransform(data: AnyRef, dataType: DataType):
AnyRef = {
+ dataType match {
+ case _ if UnsafeRowUtils.isBinaryStable(dataType) => data
+ case st: StructType =>
+
processStructTypeWithBuffer(data.asInstanceOf[InternalRow].toSeq(st).zip(st.fields))
+ case at: ArrayType => processArrayTypeWithBuffer(at,
data.asInstanceOf[ArrayData])
+ case st: StringType =>
+ CollationFactory.getCollationKey(data.asInstanceOf[UTF8String],
st.collationId)
+ case _ =>
+ throw new UnsupportedOperationException(
Review Comment:
Could you use `SparkUnsupportedOperationException` instead of it.
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
