Package io.trino.operator

Class GroupedTopNRankAccumulator

java.lang.Object

io.trino.operator.GroupedTopNRankAccumulator

public class GroupedTopNRankAccumulator
extends Object

Memory Layout:

                    +----------------------+
                    |PeerGroupLookup       |
                    +----------------------+
                    |Map:                  |
 (Group1, RowID4)+->+(Group,RowID):HeapNode+-+
                    +----------------------+ |
                                             |
                    +--------------------+   |  +---------------+    +---------------+
                    |GroupIdToHeapBuffer |   |  |HeapNodeBuffer |    |PeerGroupBuffer|
                    +--------------------+   |  +---------------+    +---------------+
           Group1+->+RootNodeIndex1+----------->+PeerGroupIndex1+--->+RowID1         |
                    |RootNodeIndex2      |   |  |PeerGroupCount1|    |NextPeerIndex1+--+
                    |...                 |   |  |LeftChild1   +----+ |RowID2         | |
                    +--------------------+   |  |RightChild1    |  | |NextPeerIndex2 | |
                    |ValueCount1         |   +->+PeerGroupIndex2+invalid input: '<'-+ |RowID3   invalid input: '<'-------+
                    |HeapSize1           |      |PeerGroupCount2|    |NextPeerIndex3 |
                    |ValueCount2         |      |LeftChild2     |    |...            |
                    |HeapSize2           |      |RightChild2    |    +---------------+
                    |...                 |      |...            |
                    +--------------------+      +---------------+
 

Constructor Summary

Constructors

Constructor	Description
GroupedTopNRankAccumulator(RowIdComparisonHashStrategy strategy, int topN, LongConsumer rowIdEvictionListener)

Method Summary

Modifier and Type	Method	Description
boolean	add(int groupId, RowReference rowReference)	Add the specified row to this accumulator.
long	drainTo(int groupId, LongBigArray rowIdOutput)	Drain the contents of this accumulator to the provided output row ID.
long	drainTo(int groupId, LongBigArray rowIdOutput, LongBigArray rankingOutput)	Drain the contents of this accumulator to the provided output row ID and ranking buffer.
int	findFirstPositionToAdd(Page newPage, int groupCount, int[] groupIds, PageWithPositionComparator comparator, RowReferencePageManager pageManager)
long	sizeOf()

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Details

GroupedTopNRankAccumulator

public GroupedTopNRankAccumulator(RowIdComparisonHashStrategy strategy,
 int topN,
 LongConsumer rowIdEvictionListener)

Method Details

sizeOf

public long sizeOf()

findFirstPositionToAdd

public int findFirstPositionToAdd(Page newPage,
 int groupCount,
 int[] groupIds,
 PageWithPositionComparator comparator,
 RowReferencePageManager pageManager)

add

public boolean add(int groupId,
 RowReference rowReference)

Add the specified row to this accumulator.

This may trigger row eviction callbacks if other rows have to be evicted to make space.

Returns:

true if this row was incorporated, false otherwise

drainTo

public long drainTo(int groupId,
 LongBigArray rowIdOutput,
 LongBigArray rankingOutput)

Drain the contents of this accumulator to the provided output row ID and ranking buffer.

Rows will be presented in increasing rank order. Draining will not trigger any row eviction callbacks. After this method completion, the Accumulator will contain zero rows for the specified groupId.

Returns:

number of rows deposited to the output buffers

drainTo

public long drainTo(int groupId,
 LongBigArray rowIdOutput)

Drain the contents of this accumulator to the provided output row ID.

Rows will be presented in increasing rank order. Draining will not trigger any row eviction callbacks. After this method completion, the Accumulator will contain zero rows for the specified groupId.

Returns:

number of rows deposited to the output buffer