Java TypePairComparatorFactory类代码示例

import org.apache.flink.api.common.typeutils.TypePairComparatorFactory; //导入依赖的package包/类
public <T1, T2> TypePairComparatorFactory<T1, T2> getPairComparatorFactory(ClassLoader cl) {
	final String className = this.config.getString(DRIVER_PAIR_COMPARATOR_FACTORY, null);
	if (className == null) {
		return null;
	}
	
	@SuppressWarnings("unchecked")
	final Class<TypePairComparatorFactory<T1, T2>> superClass = (Class<TypePairComparatorFactory<T1, T2>>) (Class<?>) TypePairComparatorFactory.class;
	try {
		final Class<? extends TypePairComparatorFactory<T1, T2>> clazz = Class.forName(className, true, cl).asSubclass(superClass);
		return InstantiationUtil.instantiate(clazz, superClass);
	}
	catch (ClassNotFoundException cnfex) {
		throw new RuntimeException("The class '" + className + "', noted in the configuration as " +
			"pair comparator factory, could not be found. It is not part of the user code's class loader resources.");
	}
	catch (ClassCastException ccex) {
		throw new CorruptConfigurationException("The class noted in the configuration as the pair comparator factory " +
			"is no subclass of TypePairComparatorFactory.");
	}
}
 

import org.apache.flink.api.common.typeutils.TypePairComparatorFactory; //导入依赖的package包/类
protected abstract JoinTaskIterator<IT1, IT2, OT> getReusingOuterJoinIterator(
		DriverStrategy driverStrategy,
		MutableObjectIterator<IT1> in1,
		MutableObjectIterator<IT2> in2,
		TypeSerializer<IT1> serializer1,
		TypeComparator<IT1> comparator1,
		TypeSerializer<IT2> serializer2,
		TypeComparator<IT2> comparator2,
		TypePairComparatorFactory<IT1, IT2> pairComparatorFactory,
		MemoryManager memoryManager,
		IOManager ioManager,
		double driverMemFraction
) throws Exception;
 

import org.apache.flink.api.common.typeutils.TypePairComparatorFactory; //导入依赖的package包/类
protected abstract JoinTaskIterator<IT1, IT2, OT> getNonReusingOuterJoinIterator(
		DriverStrategy driverStrategy,
		MutableObjectIterator<IT1> in1,
		MutableObjectIterator<IT2> in2,
		TypeSerializer<IT1> serializer1,
		TypeComparator<IT1> comparator1,
		TypeSerializer<IT2> serializer2,
		TypeComparator<IT2> comparator2,
		TypePairComparatorFactory<IT1, IT2> pairComparatorFactory,
		MemoryManager memoryManager,
		IOManager ioManager,
		double driverMemFraction
) throws Exception;
 

import org.apache.flink.api.common.typeutils.TypePairComparatorFactory; //导入依赖的package包/类
private static <T1 extends Tuple, T2 extends Tuple> TypePairComparatorFactory<T1,T2> createPairComparator(TypeInformation<?> typeInfo1, TypeInformation<?> typeInfo2) {
//		@SuppressWarnings("unchecked")
//		TupleTypeInfo<T1> info1 = (TupleTypeInfo<T1>) typeInfo1;
//		@SuppressWarnings("unchecked")
//		TupleTypeInfo<T2> info2 = (TupleTypeInfo<T2>) typeInfo2;
		
		return new RuntimePairComparatorFactory<T1,T2>();
	}
 

import org.apache.flink.api.common.typeutils.TypePairComparatorFactory; //导入依赖的package包/类
private static <T1 extends Tuple, T2 extends Tuple> TypePairComparatorFactory<T1,T2> createPairComparator(TypeInformation<?> typeInfo1, TypeInformation<?> typeInfo2) {
		if (!(typeInfo1.isTupleType() || typeInfo1 instanceof PojoTypeInfo) && (typeInfo2.isTupleType() || typeInfo2 instanceof PojoTypeInfo)) {
			throw new RuntimeException("The runtime currently supports only keyed binary operations (such as joins) on tuples and POJO types.");
		}
		
//		@SuppressWarnings("unchecked")
//		TupleTypeInfo<T1> info1 = (TupleTypeInfo<T1>) typeInfo1;
//		@SuppressWarnings("unchecked")
//		TupleTypeInfo<T2> info2 = (TupleTypeInfo<T2>) typeInfo2;
		
		return new RuntimePairComparatorFactory<T1,T2>();
	}
 

import org.apache.flink.api.common.typeutils.TypePairComparatorFactory; //导入依赖的package包/类
@Override
public void prepare() throws Exception
{
	final TaskConfig config = this.taskContext.getTaskConfig();
	if (config.getDriverStrategy() != DriverStrategy.CO_GROUP) {
		throw new Exception("Unrecognized driver strategy for CoGoup driver: " + config.getDriverStrategy().name());
	}
	
	final MutableObjectIterator<IT1> in1 = this.taskContext.getInput(0);
	final MutableObjectIterator<IT2> in2 = this.taskContext.getInput(1);
	
	// get the key positions and types
	final TypeSerializer<IT1> serializer1 = this.taskContext.<IT1>getInputSerializer(0).getSerializer();
	final TypeSerializer<IT2> serializer2 = this.taskContext.<IT2>getInputSerializer(1).getSerializer();
	final TypeComparator<IT1> groupComparator1 = this.taskContext.getDriverComparator(0);
	final TypeComparator<IT2> groupComparator2 = this.taskContext.getDriverComparator(1);
	
	final TypePairComparatorFactory<IT1, IT2> pairComparatorFactory = config.getPairComparatorFactory(
				this.taskContext.getUserCodeClassLoader());
	if (pairComparatorFactory == null) {
		throw new Exception("Missing pair comparator factory for CoGroup driver");
	}

	// create CoGropuTaskIterator according to provided local strategy.
	this.coGroupIterator = new SortMergeCoGroupIterator<IT1, IT2>(in1, in2,
			serializer1, groupComparator1,  serializer2, groupComparator2,
			pairComparatorFactory.createComparator12(groupComparator1, groupComparator2));
	
	// open CoGroupTaskIterator - this triggers the sorting and blocks until the iterator is ready
	this.coGroupIterator.open();
	
	if (LOG.isDebugEnabled()) {
		LOG.debug(this.taskContext.formatLogString("CoGroup task iterator ready."));
	}
}
 

import org.apache.flink.api.common.typeutils.TypePairComparatorFactory; //导入依赖的package包/类
@Override
@SuppressWarnings("unchecked")
public void initialize() throws Exception {
	
	final TypeComparator<IT2> solutionSetComparator;
	
	// grab a handle to the hash table from the iteration broker
	if (taskContext instanceof AbstractIterativeTask) {
		AbstractIterativeTask<?, ?> iterativeTaskContext = (AbstractIterativeTask<?, ?>) taskContext;
		String identifier = iterativeTaskContext.brokerKey();
		Object table = SolutionSetBroker.instance().get(identifier);
		
		if (table instanceof CompactingHashTable) {
			this.hashTable = (CompactingHashTable<IT2>) table;
			solutionSetSerializer = this.hashTable.getBuildSideSerializer();
			solutionSetComparator = this.hashTable.getBuildSideComparator().duplicate();
		}
		else if (table instanceof JoinHashMap) {
			this.objectMap = (JoinHashMap<IT2>) table;
			solutionSetSerializer = this.objectMap.getBuildSerializer();
			solutionSetComparator = this.objectMap.getBuildComparator().duplicate();
		}
		else {
			throw new RuntimeException("Unrecognized solution set index: " + table);
		}
	}
	else {
		throw new Exception("The task context of this driver is no iterative task context.");
	}
	
	TaskConfig config = taskContext.getTaskConfig();
	ClassLoader classLoader = taskContext.getUserCodeClassLoader();
	
	TypeComparatorFactory<IT1> probeSideComparatorFactory = config.getDriverComparator(0, classLoader); 
	
	this.probeSideSerializer = taskContext.<IT1>getInputSerializer(0).getSerializer();
	this.probeSideComparator = probeSideComparatorFactory.createComparator();

	ExecutionConfig executionConfig = taskContext.getExecutionConfig();
	objectReuseEnabled = executionConfig.isObjectReuseEnabled();

	if (objectReuseEnabled) {
		solutionSideRecord = solutionSetSerializer.createInstance();
	};
	
	TypePairComparatorFactory<IT1, IT2> factory = taskContext.getTaskConfig().getPairComparatorFactory(taskContext.getUserCodeClassLoader());
	pairComparator = factory.createComparator12(this.probeSideComparator, solutionSetComparator);
}
 

import org.apache.flink.api.common.typeutils.TypePairComparatorFactory; //导入依赖的package包/类
@Override
@SuppressWarnings("unchecked")
public void initialize() throws Exception {
	
	final TypeSerializer<IT2> solutionSetSerializer;
	final TypeComparator<IT2> solutionSetComparator;
	
	// grab a handle to the hash table from the iteration broker
	if (taskContext instanceof AbstractIterativeTask) {
		AbstractIterativeTask<?, ?> iterativeTaskContext = (AbstractIterativeTask<?, ?>) taskContext;
		String identifier = iterativeTaskContext.brokerKey();
		Object table = SolutionSetBroker.instance().get(identifier);
		
		if (table instanceof CompactingHashTable) {
			this.hashTable = (CompactingHashTable<IT2>) table;
			solutionSetSerializer = this.hashTable.getBuildSideSerializer();
			solutionSetComparator = this.hashTable.getBuildSideComparator().duplicate();
		}
		else if (table instanceof JoinHashMap) {
			this.objectMap = (JoinHashMap<IT2>) table;
			solutionSetSerializer = this.objectMap.getBuildSerializer();
			solutionSetComparator = this.objectMap.getBuildComparator().duplicate();
		}
		else {
			throw new RuntimeException("Unrecognized solution set index: " + table);
		}
	}
	else {
		throw new Exception("The task context of this driver is no iterative task context.");
	}
	
	TaskConfig config = taskContext.getTaskConfig();
	ClassLoader classLoader = taskContext.getUserCodeClassLoader();
	
	TypeSerializer<IT1> probeSideSerializer = taskContext.<IT1>getInputSerializer(0).getSerializer();
	
	TypeComparatorFactory<IT1> probeSideComparatorFactory = config.getDriverComparator(0, classLoader); 
	
	this.probeSideComparator = probeSideComparatorFactory.createComparator();
	
	ExecutionConfig executionConfig = taskContext.getExecutionConfig();
	objectReuseEnabled = executionConfig.isObjectReuseEnabled();

	if (objectReuseEnabled) {
		solutionSideRecord = solutionSetSerializer.createInstance();
		probeSideRecord = probeSideSerializer.createInstance();
	}
	
	TypePairComparatorFactory<IT1, IT2> factory = taskContext.getTaskConfig().getPairComparatorFactory(taskContext.getUserCodeClassLoader());
	pairComparator = factory.createComparator12(this.probeSideComparator, solutionSetComparator);
}
 

import org.apache.flink.api.common.typeutils.TypePairComparatorFactory; //导入依赖的package包/类
@Override
protected JoinTaskIterator<IT1, IT2, OT> getReusingOuterJoinIterator(
		DriverStrategy driverStrategy,
		MutableObjectIterator<IT1> in1,
		MutableObjectIterator<IT2> in2,
		TypeSerializer<IT1> serializer1,
		TypeComparator<IT1> comparator1,
		TypeSerializer<IT2> serializer2,
		TypeComparator<IT2> comparator2,
		TypePairComparatorFactory<IT1, IT2> pairComparatorFactory,
		MemoryManager memoryManager,
		IOManager ioManager,
		double driverMemFraction
) throws Exception {
	switch (driverStrategy) {
		case LEFT_OUTER_MERGE:
			int numPages = memoryManager.computeNumberOfPages(driverMemFraction);
			return new ReusingMergeOuterJoinIterator<>(
					OuterJoinType.LEFT,
					in1,
					in2,
					serializer1,
					comparator1,
					serializer2,
					comparator2,
					pairComparatorFactory.createComparator12(comparator1, comparator2),
					memoryManager,
					ioManager,
					numPages,
					super.taskContext.getContainingTask()
			);
		case LEFT_HYBRIDHASH_BUILD_FIRST:
			return new ReusingBuildFirstHashJoinIterator<>(in1, in2,
					serializer1, comparator1,
					serializer2, comparator2,
					pairComparatorFactory.createComparator21(comparator1, comparator2),
					memoryManager, ioManager,
					this.taskContext.getContainingTask(),
					driverMemFraction,
					false,
					true,
					false);
		case LEFT_HYBRIDHASH_BUILD_SECOND:
			return new ReusingBuildSecondHashJoinIterator<>(in1, in2,
					serializer1, comparator1,
					serializer2, comparator2,
					pairComparatorFactory.createComparator12(comparator1, comparator2),
					memoryManager, ioManager,
					this.taskContext.getContainingTask(),
					driverMemFraction,
					true,
					false,
					false);
		default:
			throw new Exception("Unsupported driver strategy for left outer join driver: " + driverStrategy.name());
	}
}
 

import org.apache.flink.api.common.typeutils.TypePairComparatorFactory; //导入依赖的package包/类
@Override
protected JoinTaskIterator<IT1, IT2, OT> getNonReusingOuterJoinIterator(
		DriverStrategy driverStrategy,
		MutableObjectIterator<IT1> in1,
		MutableObjectIterator<IT2> in2,
		TypeSerializer<IT1> serializer1,
		TypeComparator<IT1> comparator1,
		TypeSerializer<IT2> serializer2,
		TypeComparator<IT2> comparator2,
		TypePairComparatorFactory<IT1, IT2> pairComparatorFactory,
		MemoryManager memoryManager,
		IOManager ioManager,
		double driverMemFraction
) throws Exception {
	switch (driverStrategy) {
		case LEFT_OUTER_MERGE:
			int numPages = memoryManager.computeNumberOfPages(driverMemFraction);
			return new NonReusingMergeOuterJoinIterator<>(
					OuterJoinType.LEFT,
					in1,
					in2,
					serializer1,
					comparator1,
					serializer2,
					comparator2,
					pairComparatorFactory.createComparator12(comparator1, comparator2),
					memoryManager,
					ioManager,
					numPages,
					super.taskContext.getContainingTask()
			);
		case LEFT_HYBRIDHASH_BUILD_FIRST:
			return new NonReusingBuildFirstHashJoinIterator<>(in1, in2,
					serializer1, comparator1,
					serializer2, comparator2,
					pairComparatorFactory.createComparator21(comparator1, comparator2),
					memoryManager, ioManager,
					this.taskContext.getContainingTask(),
					driverMemFraction,
					false,
					true,
					false);
		case LEFT_HYBRIDHASH_BUILD_SECOND:
			return new NonReusingBuildSecondHashJoinIterator<>(in1, in2,
					serializer1, comparator1,
					serializer2, comparator2,
					pairComparatorFactory.createComparator12(comparator1, comparator2),
					memoryManager, ioManager,
					this.taskContext.getContainingTask(),
					driverMemFraction,
					true,
					false,
					false);
		default:
			throw new Exception("Unsupported driver strategy for left outer join driver: " + driverStrategy.name());
	}
}
 

import org.apache.flink.api.common.typeutils.TypePairComparatorFactory; //导入依赖的package包/类
@Override
@SuppressWarnings("unchecked")
public void initialize() {
	
	final TypeComparator<IT1> solutionSetComparator;
	
	// grab a handle to the hash table from the iteration broker
	if (taskContext instanceof AbstractIterativeTask) {
		AbstractIterativeTask<?, ?> iterativeTaskContext = (AbstractIterativeTask<?, ?>) taskContext;
		String identifier = iterativeTaskContext.brokerKey();
		
		Object table = SolutionSetBroker.instance().get(identifier);
		if (table instanceof CompactingHashTable) {
			this.hashTable = (CompactingHashTable<IT1>) table;
			solutionSetSerializer = this.hashTable.getBuildSideSerializer();
			solutionSetComparator = this.hashTable.getBuildSideComparator().duplicate();
		}
		else if (table instanceof JoinHashMap) {
			this.objectMap = (JoinHashMap<IT1>) table;
			solutionSetSerializer = this.objectMap.getBuildSerializer();
			solutionSetComparator = this.objectMap.getBuildComparator().duplicate();
		}
		else {
			throw new RuntimeException("Unrecognized solution set index: " + table);
		}
	} else {
		throw new RuntimeException("The task context of this driver is no iterative task context.");
	}
	
	TaskConfig config = taskContext.getTaskConfig();
	ClassLoader classLoader = taskContext.getUserCodeClassLoader();
	
	TypeComparatorFactory<IT2> probeSideComparatorFactory = config.getDriverComparator(0, classLoader);
	
	this.probeSideSerializer = taskContext.<IT2>getInputSerializer(0).getSerializer();
	this.probeSideComparator = probeSideComparatorFactory.createComparator();
	
	ExecutionConfig executionConfig = taskContext.getExecutionConfig();
	objectReuseEnabled = executionConfig.isObjectReuseEnabled();

	if (objectReuseEnabled) {
		solutionSideRecord = solutionSetSerializer.createInstance();
	}
	
	TypePairComparatorFactory<IT1, IT2> factory = taskContext.getTaskConfig().getPairComparatorFactory(taskContext.getUserCodeClassLoader());
	pairComparator = factory.createComparator21(solutionSetComparator, this.probeSideComparator);
}
 

import org.apache.flink.api.common.typeutils.TypePairComparatorFactory; //导入依赖的package包/类
@Override
public void prepare() throws Exception
{
	final TaskConfig config = this.taskContext.getTaskConfig();
	if (config.getDriverStrategy() != DriverStrategy.CO_GROUP) {
		throw new Exception("Unrecognized driver strategy for CoGoup driver: " + config.getDriverStrategy().name());
	}

	final Counter numRecordsIn = this.taskContext.getMetricGroup().getIOMetricGroup().getNumRecordsInCounter();
	
	final MutableObjectIterator<IT1> in1 = new CountingMutableObjectIterator<>(this.taskContext.<IT1>getInput(0), numRecordsIn);
	final MutableObjectIterator<IT2> in2 = new CountingMutableObjectIterator<>(this.taskContext.<IT2>getInput(1), numRecordsIn);
	
	// get the key positions and types
	final TypeSerializer<IT1> serializer1 = this.taskContext.<IT1>getInputSerializer(0).getSerializer();
	final TypeSerializer<IT2> serializer2 = this.taskContext.<IT2>getInputSerializer(1).getSerializer();
	final TypeComparator<IT1> groupComparator1 = this.taskContext.getDriverComparator(0);
	final TypeComparator<IT2> groupComparator2 = this.taskContext.getDriverComparator(1);
	
	final TypePairComparatorFactory<IT1, IT2> pairComparatorFactory = config.getPairComparatorFactory(
				this.taskContext.getUserCodeClassLoader());
	if (pairComparatorFactory == null) {
		throw new Exception("Missing pair comparator factory for CoGroup driver");
	}

	ExecutionConfig executionConfig = taskContext.getExecutionConfig();
	this.objectReuseEnabled = executionConfig.isObjectReuseEnabled();

	if (LOG.isDebugEnabled()) {
		LOG.debug("CoGroupDriver object reuse: " + (this.objectReuseEnabled ? "ENABLED" : "DISABLED") + ".");
	}

	if (objectReuseEnabled) {
		// create CoGroupTaskIterator according to provided local strategy.
		this.coGroupIterator = new ReusingSortMergeCoGroupIterator<IT1, IT2>(
				in1, in2,
				serializer1, groupComparator1,
				serializer2, groupComparator2,
				pairComparatorFactory.createComparator12(groupComparator1, groupComparator2));
	} else {
		// create CoGroupTaskIterator according to provided local strategy.
		this.coGroupIterator = new NonReusingSortMergeCoGroupIterator<IT1, IT2>(
				in1, in2,
				serializer1, groupComparator1,
				serializer2, groupComparator2,
				pairComparatorFactory.createComparator12(groupComparator1, groupComparator2));
	}
	
	// open CoGroupTaskIterator - this triggers the sorting and blocks until the iterator is ready
	this.coGroupIterator.open();
	
	if (LOG.isDebugEnabled()) {
		LOG.debug(this.taskContext.formatLogString("CoGroup task iterator ready."));
	}
}
 

import org.apache.flink.api.common.typeutils.TypePairComparatorFactory; //导入依赖的package包/类
@Override
protected JoinTaskIterator<IT1, IT2, OT> getReusingOuterJoinIterator(
		DriverStrategy driverStrategy,
		MutableObjectIterator<IT1> in1,
		MutableObjectIterator<IT2> in2,
		TypeSerializer<IT1> serializer1,
		TypeComparator<IT1> comparator1,
		TypeSerializer<IT2> serializer2,
		TypeComparator<IT2> comparator2,
		TypePairComparatorFactory<IT1, IT2> pairComparatorFactory,
		MemoryManager memoryManager,
		IOManager ioManager,
		double driverMemFraction
) throws Exception {
	switch (driverStrategy) {
		case FULL_OUTER_MERGE:
			int numPages = memoryManager.computeNumberOfPages(driverMemFraction);
			return new ReusingMergeOuterJoinIterator<>(
					OuterJoinType.FULL,
					in1,
					in2,
					serializer1,
					comparator1,
					serializer2,
					comparator2,
					pairComparatorFactory.createComparator12(comparator1, comparator2),
					memoryManager,
					ioManager,
					numPages,
					super.taskContext.getContainingTask()
			);
	case FULL_OUTER_HYBRIDHASH_BUILD_FIRST:
		return new ReusingBuildFirstHashJoinIterator<>(in1, in2,
				serializer1, comparator1,
				serializer2, comparator2,
				pairComparatorFactory.createComparator21(comparator1, comparator2),
				memoryManager, ioManager,
				this.taskContext.getContainingTask(),
				driverMemFraction,
				true,
				true,
				false);
	case FULL_OUTER_HYBRIDHASH_BUILD_SECOND:
		return new ReusingBuildSecondHashJoinIterator<>(in1, in2,
				serializer1, comparator1,
				serializer2, comparator2,
				pairComparatorFactory.createComparator12(comparator1, comparator2),
				memoryManager, ioManager,
				this.taskContext.getContainingTask(),
				driverMemFraction,
				true,
				true,
				false);
		default:
			throw new Exception("Unsupported driver strategy for full outer join driver: " + driverStrategy.name());
	}
}
 

import org.apache.flink.api.common.typeutils.TypePairComparatorFactory; //导入依赖的package包/类
@Override
protected JoinTaskIterator<IT1, IT2, OT> getNonReusingOuterJoinIterator(
		DriverStrategy driverStrategy,
		MutableObjectIterator<IT1> in1,
		MutableObjectIterator<IT2> in2,
		TypeSerializer<IT1> serializer1,
		TypeComparator<IT1> comparator1,
		TypeSerializer<IT2> serializer2,
		TypeComparator<IT2> comparator2,
		TypePairComparatorFactory<IT1, IT2> pairComparatorFactory,
		MemoryManager memoryManager,
		IOManager ioManager,
		double driverMemFraction
) throws Exception {
	switch (driverStrategy) {
		case FULL_OUTER_MERGE:
			int numPages = memoryManager.computeNumberOfPages(driverMemFraction);
			return new NonReusingMergeOuterJoinIterator<>(
					OuterJoinType.FULL,
					in1,
					in2,
					serializer1,
					comparator1,
					serializer2,
					comparator2,
					pairComparatorFactory.createComparator12(comparator1, comparator2),
					memoryManager,
					ioManager,
					numPages,
					super.taskContext.getContainingTask()
			);
		case FULL_OUTER_HYBRIDHASH_BUILD_FIRST:
			return new NonReusingBuildFirstHashJoinIterator<>(in1, in2,
				serializer1, comparator1,
				serializer2, comparator2,
				pairComparatorFactory.createComparator21(comparator1, comparator2),
				memoryManager, ioManager,
				this.taskContext.getContainingTask(),
				driverMemFraction,
				true,
				true,
				false);
		case FULL_OUTER_HYBRIDHASH_BUILD_SECOND:
			return new NonReusingBuildSecondHashJoinIterator<>(in1, in2,
				serializer1, comparator1,
				serializer2, comparator2,
				pairComparatorFactory.createComparator12(comparator1, comparator2),
				memoryManager, ioManager,
				this.taskContext.getContainingTask(),
				driverMemFraction,
				true,
				true,
				false);
		default:
			throw new Exception("Unsupported driver strategy for full outer join driver: " + driverStrategy.name());
	}
}
 

import org.apache.flink.api.common.typeutils.TypePairComparatorFactory; //导入依赖的package包/类
public void setDriverPairComparator(TypePairComparatorFactory<?, ?> factory) {
	final Class<?> clazz = factory.getClass();
	InstantiationUtil.checkForInstantiation(clazz);
	this.config.setString(DRIVER_PAIR_COMPARATOR_FACTORY, clazz.getName());
}