Github user fhueske commented on a diff in the pull request:
https://github.com/apache/flink/pull/3266#discussion_r101016955
--- Diff:
flink-libraries/flink-table/src/main/scala/org/apache/flink/table/runtime/aggregate/AggregateUtil.scala
---
@@ -306,6 +307,85 @@ object AggregateUtil {
}
/**
+ * Create a
[[org.apache.flink.api.common.functions.MapPartitionFunction]] that aggregation
+ * for aggregates.
+ * The function returns aggregate values of all aggregate function
which are
+ * organized by the following format:
+ *
+ * {{{
+ * avg(x) aggOffsetInRow = 2 count(z) aggOffsetInRow = 5
+ * | |
windowEnd(max(rowtime)
+ * | | |
+ * v v v
+ * +--------+--------+--------+--------+-----------+---------+
+ * | sum1 | count1 | sum2 | count2 |windowStart|windowEnd|
+ * +--------+--------+--------+--------+-----------+---------+
+ * ^ ^
+ * | |
+ * sum(y) aggOffsetInRow = 4 windowStart(min(rowtime))
+ *
+ * }}}
+ *
+ */
+ def createDataSetWindowAggregationMapPartitionFunction(
+ window: LogicalWindow,
+ namedAggregates: Seq[CalcitePair[AggregateCall, String]],
+ inputType: RelDataType,
+ outputType: RelDataType = null,
+ properties: Seq[NamedWindowProperty] = null,
+ isPreMapPartition: Boolean = true,
+ isInputCombined: Boolean = false): MapPartitionFunction[Row, Row] = {
+
+ val aggregates = transformToAggregateFunctions(
+ namedAggregates.map(_.getKey),
+ inputType,
+ 0)._2
+
+ val intermediateRowArity =
aggregates.map(_.intermediateDataType.length).sum
+
+ window match {
+ case EventTimeSessionGroupWindow(_, _, gap) =>
+ if (isPreMapPartition) {
+ val preMapReturnType: RowTypeInfo =
+ createAggregateBufferDataType(
+ Array(),
+ aggregates,
+ inputType,
+ Option(Array(BasicTypeInfo.LONG_TYPE_INFO,
BasicTypeInfo.LONG_TYPE_INFO)))
+
+ new DataSetSessionWindowAggregatePreProcessor(
+ aggregates,
+ Array(),
+ // the addition two fields are used to store window-start and
window-end attributes
+ intermediateRowArity + 2,
+ asLong(gap),
+ preMapReturnType)
+
+ } else {
+ val (startPos, endPos) =
computeWindowStartEndPropertyPos(properties)
--- End diff --
We do not need this case if we compute the final aggregates with a
`GroupReduceFunction`.
---
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.
---
