Github user fhueske commented on a diff in the pull request:
https://github.com/apache/flink/pull/2792#discussion_r89101873
--- Diff:
flink-libraries/flink-table/src/main/scala/org/apache/flink/api/table/runtime/aggregate/AggregateUtil.scala
---
@@ -115,14 +124,210 @@ object AggregateUtil {
intermediateRowArity,
outputType.getFieldCount)
}
+ groupReduceFunction
+ }
+
+ /**
+ * Create IncrementalAggregateReduceFunction for Incremental
aggregates. It implement
+ * [[org.apache.flink.api.common.functions.ReduceFunction]]
+ *
+ */
+ private[flink] def createIncrementalAggregateReduceFunction(
+ aggregates: Array[Aggregate[_ <: Any]],
+ namedAggregates: Seq[CalcitePair[AggregateCall, String]],
+ inputType: RelDataType,
+ outputType: RelDataType,
+ groupings: Array[Int]): IncrementalAggregateReduceFunction = {
+ val groupingOffsetMapping =
+ getGroupingOffsetAndaggOffsetMapping(
+ namedAggregates,
+ inputType,
+ outputType,
+ groupings)._1
+ val intermediateRowArity = groupings.length +
aggregates.map(_.intermediateDataType.length).sum
+ val reduceFunction = new IncrementalAggregateReduceFunction(
+ aggregates,
+ groupingOffsetMapping,
+ intermediateRowArity)
+ reduceFunction
+ }
+
+ /**
+ * @return groupingOffsetMapping (mapping relation between field index
of intermediate
+ * aggregate Row and output Row.)
+ * and aggOffsetMapping (the mapping relation between aggregate
function index in list
+ * and its corresponding field index in output Row.)
+ */
+ def getGroupingOffsetAndaggOffsetMapping(
+ namedAggregates: Seq[CalcitePair[AggregateCall, String]],
+ inputType: RelDataType,
+ outputType: RelDataType,
+ groupings: Array[Int]): (Array[(Int, Int)], Array[(Int, Int)]) = {
+
+ // the mapping relation between field index of intermediate aggregate
Row and output Row.
+ val groupingOffsetMapping = getGroupKeysMapping(inputType, outputType,
groupings)
+
+ // the mapping relation between aggregate function index in list and
its corresponding
+ // field index in output Row.
+ val aggOffsetMapping = getAggregateMapping(namedAggregates, outputType)
- (mapFunction, reduceGroupFunction)
+ if (groupingOffsetMapping.length != groupings.length ||
+ aggOffsetMapping.length != namedAggregates.length) {
+ throw new TableException(
+ "Could not find output field in input data type " +
+ "or aggregate functions.")
+ }
+ (groupingOffsetMapping, aggOffsetMapping)
+ }
+
+
+ private[flink] def createAllWindowAggregationFunction(
+ window: LogicalWindow,
+ properties: Seq[NamedWindowProperty],
+ aggFunction: RichGroupReduceFunction[Row, Row])
+ : AllWindowFunction[Row, Row, DataStreamWindow] = {
+
+ if (isTimeWindow(window)) {
+ val (startPos, endPos) = computeWindowStartEndPropertyPos(properties)
+ new AggregateAllTimeWindowFunction(aggFunction, startPos, endPos)
+ .asInstanceOf[AllWindowFunction[Row, Row, DataStreamWindow]]
+ } else {
+ new AggregateAllWindowFunction(aggFunction)
+ }
+
+ }
+
+
+ private[flink] def createWindowAggregationFunction(
+ window: LogicalWindow,
+ properties: Seq[NamedWindowProperty],
+ aggFunction: RichGroupReduceFunction[Row, Row])
--- End diff --
Same as above.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
