Package org.apache.flink.table.runtime.hashtable

Class LongHybridHashTable

java.lang.Object

org.apache.flink.table.runtime.hashtable.BaseHybridHashTable

org.apache.flink.table.runtime.hashtable.LongHybridHashTable

All Implemented Interfaces:

org.apache.flink.core.memory.MemorySegmentSource, MemorySegmentPool

public abstract class LongHybridHashTable
extends BaseHybridHashTable

Special optimized hashTable with key long.

See LongHashPartition. TODO add min max long filter and bloomFilter to spilled partition.

Field Summary

Fields inherited from class org.apache.flink.table.runtime.hashtable.BaseHybridHashTable

buildRowCount, buildSpillRetBufferNumbers, buildSpillReturnBuffers, closed, compressionBlockSize, compressionCodecFactory, compressionEnabled, currentEnumerator, currentRecursionDepth, currentSpilledBuildSide, currentSpilledProbeSide, initPartitionFanOut, internalPool, ioManager, LOG, MAX_NUM_PARTITIONS, MAX_RECURSION_DEPTH, numSpillFiles, segmentSize, segmentSizeBits, segmentSizeMask, spillInBytes, totalNumBuffers, tryDistinctBuildRow

Constructor Summary

Constructors

Constructor	Description
LongHybridHashTable(Object owner, boolean compressionEnabled, int compressionBlockSize, BinaryRowDataSerializer buildSideSerializer, BinaryRowDataSerializer probeSideSerializer, org.apache.flink.runtime.memory.MemoryManager memManager, long reservedMemorySize, org.apache.flink.runtime.io.disk.iomanager.IOManager ioManager, int avgRecordLen, long buildRowCount)

Method Summary

Modifier and Type	Method	Description
protected void	clearPartitions()
void	close()	Closes the hash table.
int	compressionBlockSize()
org.apache.flink.runtime.io.compression.BlockCompressionFactory	compressionCodecFactory()
boolean	compressionEnabled()
void	endBuild()
void	free()
RowIterator<org.apache.flink.table.data.binary.BinaryRowData>	get(long probeKey)	This method is only used for operator fusion codegen to get build row from hash table.
abstract long	getBuildLongKey(org.apache.flink.table.data.RowData row)	For code gen get build side long key.
LongHashPartition.MatchIterator	getBuildSideIterator()
org.apache.flink.table.data.RowData	getCurrentProbeRow()
List<LongHashPartition>	getPartitionsPendingForSMJ()
abstract long	getProbeLongKey(org.apache.flink.table.data.RowData row)	For code gen get probe side long key.
RowIterator	getSpilledPartitionBuildSideIter(LongHashPartition p)
ProbeIterator	getSpilledPartitionProbeSideIter(LongHashPartition p)
void	insertIntoProbeBuffer(org.apache.flink.table.data.RowData probeRecord)	If the probe row corresponding partition has been spilled to disk, just call this method spill probe row to disk.
boolean	nextMatching()
abstract org.apache.flink.table.data.binary.BinaryRowData	probeToBinary(org.apache.flink.table.data.RowData row)	For code gen probe side to BinaryRowData.
void	putBuildRow(org.apache.flink.table.data.binary.BinaryRowData row)
int	spillPartition()
boolean	tryProbe(org.apache.flink.table.data.RowData record)

Methods inherited from class org.apache.flink.table.runtime.hashtable.BaseHybridHashTable

createInputView, ensureNumBuffersReturned, freeCurrent, freePages, getNextBuffer, getNextBuffers, getNotNullNextBuffer, getNumSpillFiles, getSpillInBytes, getUsedMemoryInBytes, hash, maxInitBufferOfBucketArea, maxNumPartition, nextSegment, pageSize, readAllBuffers, releaseMemoryCacheForSMJ, remainBuffers, returnAll, returnPage

Methods inherited from class java.lang.Object

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

Constructor Detail

LongHybridHashTable

public LongHybridHashTable(Object owner,
                           boolean compressionEnabled,
                           int compressionBlockSize,
                           BinaryRowDataSerializer buildSideSerializer,
                           BinaryRowDataSerializer probeSideSerializer,
                           org.apache.flink.runtime.memory.MemoryManager memManager,
                           long reservedMemorySize,
                           org.apache.flink.runtime.io.disk.iomanager.IOManager ioManager,
                           int avgRecordLen,
                           long buildRowCount)

Method Detail

putBuildRow

public void putBuildRow(org.apache.flink.table.data.binary.BinaryRowData row)
                 throws IOException

Throws:

IOException

endBuild

public void endBuild()
              throws IOException

Throws:

IOException

get

@Nullable
public final RowIterator<org.apache.flink.table.data.binary.BinaryRowData> get(long probeKey)
                                                                        throws IOException

This method is only used for operator fusion codegen to get build row from hash table. If the build partition has spilled to disk, return null directly which requires the join operator also spill probe row to disk.

Throws:

IOException

insertIntoProbeBuffer

public final void insertIntoProbeBuffer(org.apache.flink.table.data.RowData probeRecord)
                                 throws IOException

If the probe row corresponding partition has been spilled to disk, just call this method spill probe row to disk.

Note: This must be called only after get(long) method.

Throws:

IOException

tryProbe

public boolean tryProbe(org.apache.flink.table.data.RowData record)
                 throws IOException

Throws:

IOException

nextMatching

public boolean nextMatching()
                     throws IOException

Throws:

IOException

getCurrentProbeRow

public org.apache.flink.table.data.RowData getCurrentProbeRow()

getBuildSideIterator

public LongHashPartition.MatchIterator getBuildSideIterator()

close

public void close()

Description copied from class: BaseHybridHashTable

Closes the hash table. This effectively releases all internal structures and closes all open files and removes them. The call to this method is valid both as a cleanup after the complete inputs were properly processed, and as an cancellation call, which cleans up all resources that are currently held by the hash join.

Overrides:

close in class BaseHybridHashTable

free

public void free()

Overrides:

free in class BaseHybridHashTable

getBuildLongKey

public abstract long getBuildLongKey(org.apache.flink.table.data.RowData row)

For code gen get build side long key.

getProbeLongKey

public abstract long getProbeLongKey(org.apache.flink.table.data.RowData row)

For code gen get probe side long key.

probeToBinary

public abstract org.apache.flink.table.data.binary.BinaryRowData probeToBinary(org.apache.flink.table.data.RowData row)

For code gen probe side to BinaryRowData.

spillPartition

public int spillPartition()
                   throws IOException

Specified by:

spillPartition in class BaseHybridHashTable

Throws:

IOException

getPartitionsPendingForSMJ

public List<LongHashPartition> getPartitionsPendingForSMJ()

getSpilledPartitionBuildSideIter

public RowIterator getSpilledPartitionBuildSideIter(LongHashPartition p)
                                             throws IOException

Throws:

IOException

getSpilledPartitionProbeSideIter

public ProbeIterator getSpilledPartitionProbeSideIter(LongHashPartition p)
                                               throws IOException

Throws:

IOException

clearPartitions

protected void clearPartitions()

Specified by:

clearPartitions in class BaseHybridHashTable

compressionEnabled

public boolean compressionEnabled()

compressionCodecFactory

public org.apache.flink.runtime.io.compression.BlockCompressionFactory compressionCodecFactory()

compressionBlockSize

public int compressionBlockSize()