Java Code Examples for org.apache.flink.runtime.operators.testutils.TestData#TupleGeneratorIterator

The following examples show how to use org.apache.flink.runtime.operators.testutils.TestData#TupleGeneratorIterator . These examples are extracted from open source projects. You can vote up the ones you like or vote down the ones you don't like, and go to the original project or source file by following the links above each example. You may check out the related API usage on the sidebar.
Example 1
Source Project: flink   File: HashVsSortMiniBenchmark.java    License: Apache License 2.0 5 votes vote down vote up
@Test
public void testBuildFirst() {
	try {
		TestData.TupleGenerator generator1 = new TestData.TupleGenerator(SEED1, INPUT_1_SIZE / 10, 100, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
		TestData.TupleGenerator generator2 = new TestData.TupleGenerator(SEED2, INPUT_2_SIZE, 100, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
		
		final TestData.TupleGeneratorIterator input1 = new TestData.TupleGeneratorIterator(generator1, INPUT_1_SIZE);
		final TestData.TupleGeneratorIterator input2 = new TestData.TupleGeneratorIterator(generator2, INPUT_2_SIZE);
		
		final FlatJoinFunction matcher = new NoOpMatcher();
		
		final Collector<Tuple2<Integer, String>> collector = new DiscardingOutputCollector<>();
		
		long start = System.nanoTime();
		
		// compare with iterator values
		final ReusingBuildFirstHashJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> iterator =
				new ReusingBuildFirstHashJoinIterator<>(
					input1, input2, this.serializer1.getSerializer(), this.comparator1, 
						this.serializer2.getSerializer(), this.comparator2, this.pairComparator11,
						this.memoryManager, this.ioManager, this.parentTask, 1, false, false, true);
		
		iterator.open();
		
		while (iterator.callWithNextKey(matcher, collector));
		
		iterator.close();
		
		long elapsed = System.nanoTime() - start;
		double msecs = elapsed / (1000 * 1000);
		
		System.out.println("Hash Build First Took " + msecs + " msecs.");
	}
	catch (Exception e) {
		e.printStackTrace();
		Assert.fail("An exception occurred during the test: " + e.getMessage());
	}
}
 
Example 2
@Test
public void test() throws Exception {
	final TupleGenerator generator1 =
			new TupleGenerator(SEED1, 500, 4096, KeyMode.SORTED, ValueMode.RANDOM_LENGTH);
	final TupleGenerator generator2 =
			new TupleGenerator(SEED2, 500, 2048, KeyMode.SORTED, ValueMode.RANDOM_LENGTH);

	final TestData.TupleGeneratorIterator input1 = new TestData.TupleGeneratorIterator(generator1, INPUT_FIRST_SIZE);
	final TestData.TupleGeneratorIterator input2 = new TestData.TupleGeneratorIterator(generator2, INPUT_SECOND_SIZE);

	// collect expected data
	final Map<Integer, Collection<Match>> expectedMatchesMap = matchValues(
			collectData(input1), collectData(input2));

	// reset the generators
	generator1.reset();
	generator2.reset();
	input1.reset();
	input2.reset();

	StreamOperator operator = getOperator();

	match(expectedMatchesMap, transformToBinary(join(operator, input1, input2)));

	// assert that each expected match was seen
	for (Map.Entry<Integer, Collection<Match>> entry : expectedMatchesMap.entrySet()) {
		Assert.assertTrue("Collection for key " + entry.getKey() + " is not empty", entry.getValue().isEmpty());
	}
}
 
Example 3
@Test
public void testSortBothMerge() {
	try {
		
		TestData.TupleGenerator generator1 = new TestData.TupleGenerator(SEED1, INPUT_1_SIZE / 10, 100, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
		TestData.TupleGenerator generator2 = new TestData.TupleGenerator(SEED2, INPUT_2_SIZE, 100, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);

		final TestData.TupleGeneratorIterator input1 = new TestData.TupleGeneratorIterator(generator1, INPUT_1_SIZE);
		final TestData.TupleGeneratorIterator input2 = new TestData.TupleGeneratorIterator(generator2, INPUT_2_SIZE);
		
		final FlatJoinFunction matcher = new NoOpMatcher();
		final Collector<Tuple2<Integer, String>> collector = new DiscardingOutputCollector<>();
		
		long start = System.nanoTime();
		
		final UnilateralSortMerger<Tuple2<Integer, String>> sorter1 = new UnilateralSortMerger<>(
				this.memoryManager, this.ioManager, input1, this.parentTask, this.serializer1, 
				this.comparator1.duplicate(), (double)MEMORY_FOR_SORTER/MEMORY_SIZE, 128, 0.8f,
				true /*use large record handler*/, true);
		
		final UnilateralSortMerger<Tuple2<Integer, String>> sorter2 = new UnilateralSortMerger<>(
				this.memoryManager, this.ioManager, input2, this.parentTask, this.serializer2, 
				this.comparator2.duplicate(), (double)MEMORY_FOR_SORTER/MEMORY_SIZE, 128, 0.8f,
				true /*use large record handler*/, true);
		
		final MutableObjectIterator<Tuple2<Integer, String>> sortedInput1 = sorter1.getIterator();
		final MutableObjectIterator<Tuple2<Integer, String>> sortedInput2 = sorter2.getIterator();
		
		// compare with iterator values
		ReusingMergeInnerJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> iterator =
			new ReusingMergeInnerJoinIterator<>(sortedInput1, sortedInput2,
					this.serializer1.getSerializer(), this.comparator1, this.serializer2.getSerializer(), this.comparator2, this.pairComparator11,
					this.memoryManager, this.ioManager, MEMORY_PAGES_FOR_MERGE, this.parentTask);
		
		iterator.open();
		
		while (iterator.callWithNextKey(matcher, collector));
		
		iterator.close();
		sorter1.close();
		sorter2.close();
		
		long elapsed = System.nanoTime() - start;
		double msecs = elapsed / (1000 * 1000);
		
		System.out.println("Sort-Merge Took " + msecs + " msecs.");
	}
	catch (Exception e) {
		e.printStackTrace();
		Assert.fail("An exception occurred during the test: " + e.getMessage());
	}
}
 
Example 4
@Test
public void testBuildFirstWithHighNumberOfCommonKeys()
{
	// the size of the left and right inputs
	final int INPUT_1_SIZE = 200;
	final int INPUT_2_SIZE = 100;
	
	final int INPUT_1_DUPLICATES = 10;
	final int INPUT_2_DUPLICATES = 2000;
	final int DUPLICATE_KEY = 13;
	
	try {
		TestData.TupleGenerator generator1 = new TestData.TupleGenerator(SEED1, 500, 4096, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
		TestData.TupleGenerator generator2 = new TestData.TupleGenerator(SEED2, 500, 2048, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
		
		final TestData.TupleGeneratorIterator gen1Iter = new TestData.TupleGeneratorIterator(generator1, INPUT_1_SIZE);
		final TestData.TupleGeneratorIterator gen2Iter = new TestData.TupleGeneratorIterator(generator2, INPUT_2_SIZE);
		
		final TestData.TupleConstantValueIterator const1Iter = new TestData.TupleConstantValueIterator(DUPLICATE_KEY, "LEFT String for Duplicate Keys", INPUT_1_DUPLICATES);
		final TestData.TupleConstantValueIterator const2Iter = new TestData.TupleConstantValueIterator(DUPLICATE_KEY, "RIGHT String for Duplicate Keys", INPUT_2_DUPLICATES);
		
		final List<MutableObjectIterator<Tuple2<Integer, String>>> inList1 = new ArrayList<>();
		inList1.add(gen1Iter);
		inList1.add(const1Iter);
		
		final List<MutableObjectIterator<Tuple2<Integer, String>>> inList2 = new ArrayList<>();
		inList2.add(gen2Iter);
		inList2.add(const2Iter);
		
		MutableObjectIterator<Tuple2<Integer, String>> input1 = new UnionIterator<>(inList1);
		MutableObjectIterator<Tuple2<Integer, String>> input2 = new UnionIterator<>(inList2);
		
		
		// collect expected data
		final Map<Integer, Collection<TupleMatch>> expectedMatchesMap = joinTuples(
				collectTupleData(input1),
				collectTupleData(input2));
		
		// re-create the whole thing for actual processing
		
		// reset the generators and iterators
		generator1.reset();
		generator2.reset();
		const1Iter.reset();
		const2Iter.reset();
		gen1Iter.reset();
		gen2Iter.reset();
		
		inList1.clear();
		inList1.add(gen1Iter);
		inList1.add(const1Iter);
		
		inList2.clear();
		inList2.add(gen2Iter);
		inList2.add(const2Iter);

		input1 = new UnionIterator<>(inList1);
		input2 = new UnionIterator<>(inList2);
		
		final FlatJoinFunction matcher = new TupleMatchRemovingJoin(expectedMatchesMap);
		final Collector<Tuple2<Integer, String>> collector = new DiscardingOutputCollector<>();

		ReusingBuildFirstHashJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> iterator =
				new ReusingBuildFirstHashJoinIterator<>(
					input1, input2, this.recordSerializer, this.record1Comparator, 
					this.recordSerializer, this.record2Comparator, this.recordPairComparator,
					this.memoryManager, ioManager, this.parentTask, 1.0, false, false, true);

		iterator.open();
		
		while (iterator.callWithNextKey(matcher, collector));
		
		iterator.close();

		// assert that each expected match was seen
		for (Entry<Integer, Collection<TupleMatch>> entry : expectedMatchesMap.entrySet()) {
			if (!entry.getValue().isEmpty()) {
				Assert.fail("Collection for key " + entry.getKey() + " is not empty");
			}
		}
	}
	catch (Exception e) {
		e.printStackTrace();
		Assert.fail("An exception occurred during the test: " + e.getMessage());
	}
}
 
Example 5
@Test
public void testInMemorySort() {
	try {
		// comparator
		final TypeComparator<Integer> keyComparator = new IntComparator(true);
		
		final TestData.TupleGenerator generator = new TestData.TupleGenerator(SEED, KEY_MAX, VALUE_LENGTH, KeyMode.RANDOM, ValueMode.CONSTANT, VAL);
		final MutableObjectIterator<Tuple2<Integer, String>> source = new TestData.TupleGeneratorIterator(generator, NUM_PAIRS);

		// merge iterator
		LOG.debug("Initializing sortmerger...");
		
		Sorter<Tuple2<Integer, String>> merger = new UnilateralSortMerger<>(this.memoryManager, this.ioManager,
			source, this.parentTask, this.pactRecordSerializer, this.pactRecordComparator,
				(double)64/78, 2, 0.9f, true /*use large record handler*/, true);

		// emit data
		LOG.debug("Reading and sorting data...");

		// check order
		MutableObjectIterator<Tuple2<Integer, String>> iterator = merger.getIterator();
		
		LOG.debug("Checking results...");
		int pairsEmitted = 1;

		Tuple2<Integer, String> rec1 = new Tuple2<>();
		Tuple2<Integer, String> rec2 = new Tuple2<>();
		
		Assert.assertTrue((rec1 = iterator.next(rec1)) != null);
		while ((rec2 = iterator.next(rec2)) != null) {
			pairsEmitted++;
			
			Assert.assertTrue(keyComparator.compare(rec1.f0, rec2.f0) <= 0);

			Tuple2<Integer, String> tmp = rec1;
			rec1 = rec2;
			rec2 = tmp;
		}
		Assert.assertTrue(NUM_PAIRS == pairsEmitted);
		
		merger.close();
		testSuccess = true;
	}
	catch (Exception e) {
		e.printStackTrace();
		Assert.fail(e.getMessage());
	}
}
 
Example 6
@Test
public void testBuildFirstAndProbeSideOuterJoin() {
	try {
		TestData.TupleGenerator generator1 = new TestData.TupleGenerator(SEED1, 500, 4096, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
		TestData.TupleGenerator generator2 = new TestData.TupleGenerator(SEED2, 1000, 2048, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);

		final TestData.TupleGeneratorIterator input1 = new TestData.TupleGeneratorIterator(generator1, INPUT_1_SIZE);
		final TestData.TupleGeneratorIterator input2 = new TestData.TupleGeneratorIterator(generator2, INPUT_2_SIZE);

		// collect expected data
		final Map<Integer, Collection<TupleMatch>> expectedMatchesMap = rightOuterJoinTuples(
				collectTupleData(input1),
				collectTupleData(input2));

		final FlatJoinFunction matcher = new TupleMatchRemovingJoin(expectedMatchesMap);
		final Collector<Tuple2<Integer, String>> collector = new DiscardingOutputCollector<>();

		// reset the generators
		generator1.reset();
		generator2.reset();
		input1.reset();
		input2.reset();

		// compare with iterator values
		ReusingBuildFirstHashJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> iterator =
				new ReusingBuildFirstHashJoinIterator<>(
						input1, input2, this.recordSerializer, this.record1Comparator,
						this.recordSerializer, this.record2Comparator, this.recordPairComparator,
						this.memoryManager, ioManager, this.parentTask, 1.0, true, false, false);

		iterator.open();

		while (iterator.callWithNextKey(matcher, collector));

		iterator.close();
	
		// assert that each expected match was seen
		for (Entry<Integer, Collection<TupleMatch>> entry : expectedMatchesMap.entrySet()) {
			if (!entry.getValue().isEmpty()) {
				Assert.fail("Collection for key " + entry.getKey() + " is not empty");
			}
		}
	}
	catch (Exception e) {
		e.printStackTrace();
		Assert.fail("An exception occurred during the test: " + e.getMessage());
	}
}
 
Example 7
Source Project: flink   File: ExternalSortITCase.java    License: Apache License 2.0 4 votes vote down vote up
@Test
public void testSpillingSort() {
	try {
		// comparator
		final TypeComparator<Integer> keyComparator = new IntComparator(true);
		
		final TestData.TupleGenerator generator = new TestData.TupleGenerator(SEED, KEY_MAX, VALUE_LENGTH, KeyMode.RANDOM, ValueMode.CONSTANT, VAL);
		final MutableObjectIterator<Tuple2<Integer, String>> source = new TestData.TupleGeneratorIterator(generator, NUM_PAIRS);

		// merge iterator
		LOG.debug("Initializing sortmerger...");
		
		Sorter<Tuple2<Integer, String>> merger = new UnilateralSortMerger<>(this.memoryManager, this.ioManager,
				source, this.parentTask, this.pactRecordSerializer, this.pactRecordComparator,
				(double)16/78, 64, 0.7f, true /*use large record handler*/, true);

		// emit data
		LOG.debug("Reading and sorting data...");

		// check order
		MutableObjectIterator<Tuple2<Integer, String>> iterator = merger.getIterator();
		
		LOG.debug("Checking results...");
		int pairsEmitted = 1;

		Tuple2<Integer, String> rec1 = new Tuple2<>();
		Tuple2<Integer, String> rec2 = new Tuple2<>();
		
		Assert.assertTrue((rec1 = iterator.next(rec1)) != null);
		while ((rec2 = iterator.next(rec2)) != null) {
			pairsEmitted++;
			
			Assert.assertTrue(keyComparator.compare(rec1.f0, rec2.f0) <= 0);

			Tuple2<Integer, String> tmp = rec1;
			rec1 = rec2;
			rec2 = tmp;
		}
		Assert.assertTrue(NUM_PAIRS == pairsEmitted);
		
		merger.close();
		testSuccess = true;
	}
	catch (Exception e) {
		e.printStackTrace();
		Assert.fail(e.getMessage());
	}
}
 
Example 8
@Test
public void testMerge() {
	try {

		final TupleGenerator generator1 = new TupleGenerator(SEED1, 500, 4096, KeyMode.SORTED, ValueMode.RANDOM_LENGTH);
		final TupleGenerator generator2 = new TupleGenerator(SEED2, 500, 2048, KeyMode.SORTED, ValueMode.RANDOM_LENGTH);

		final TestData.TupleGeneratorIterator input1 = new TestData.TupleGeneratorIterator(generator1, INPUT_1_SIZE);
		final TestData.TupleGeneratorIterator input2 = new TestData.TupleGeneratorIterator(generator2, INPUT_2_SIZE);

		// collect expected data
		final Map<Integer, Collection<Match>> expectedMatchesMap = matchValues(
			collectData(input1),
			collectData(input2));

		final FlatJoinFunction<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> joinFunction =
				new MatchRemovingJoiner(expectedMatchesMap);

		final Collector<Tuple2<Integer, String>> collector = new DiscardingOutputCollector<Tuple2<Integer, String>>();

		// reset the generators
		generator1.reset();
		generator2.reset();
		input1.reset();
		input2.reset();

		// compare with iterator values
		ReusingMergeInnerJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> iterator =
			new ReusingMergeInnerJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>>(
				input1, input2, this.serializer1, this.comparator1, this.serializer2, this.comparator2,
				this.pairComparator, this.memoryManager, this.ioManager, PAGES_FOR_BNLJN, this.parentTask);

		iterator.open();

		while (iterator.callWithNextKey(joinFunction, collector));

		iterator.close();

		// assert that each expected match was seen
		for (Entry<Integer, Collection<Match>> entry : expectedMatchesMap.entrySet()) {
			Assert.assertTrue("Collection for key " + entry.getKey() + " is not empty", entry.getValue().isEmpty());
		}
	}
	catch (Exception e) {
		e.printStackTrace();
		Assert.fail("An exception occurred during the test: " + e.getMessage());
	}
}
 
Example 9
@Test
public void testBuildFirstAndFullOuterJoin() {
	try {
		TestData.TupleGenerator generator1 = new TestData.TupleGenerator(SEED1, 500, 4096, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
		TestData.TupleGenerator generator2 = new TestData.TupleGenerator(SEED2, 1000, 2048, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);

		final TestData.TupleGeneratorIterator input1 = new TestData.TupleGeneratorIterator(generator1, INPUT_1_SIZE);
		final TestData.TupleGeneratorIterator input2 = new TestData.TupleGeneratorIterator(generator2, INPUT_2_SIZE);

		// collect expected data
		final Map<Integer, Collection<TupleMatch>> expectedMatchesMap = fullOuterJoinTuples(
			collectTupleData(input1),
			collectTupleData(input2));

		final FlatJoinFunction matcher = new TupleMatchRemovingJoin(expectedMatchesMap);
		final Collector<Tuple2<Integer, String>> collector = new DiscardingOutputCollector<>();

		// reset the generators
		generator1.reset();
		generator2.reset();
		input1.reset();
		input2.reset();

		// compare with iterator values
		ReusingBuildFirstHashJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> iterator =
			new ReusingBuildFirstHashJoinIterator<>(
				input1, input2, this.recordSerializer, this.record1Comparator,
				this.recordSerializer, this.record2Comparator, this.recordPairComparator,
				this.memoryManager, ioManager, this.parentTask, 1.0, true, true, false);

		iterator.open();

		while (iterator.callWithNextKey(matcher, collector));

		iterator.close();

		// assert that each expected match was seen
		for (Entry<Integer, Collection<TupleMatch>> entry : expectedMatchesMap.entrySet()) {
			if (!entry.getValue().isEmpty()) {
				Assert.fail("Collection for key " + entry.getKey() + " is not empty");
			}
		}
	}
	catch (Exception e) {
		e.printStackTrace();
		Assert.fail("An exception occurred during the test: " + e.getMessage());
	}
}
 
Example 10
@Test
public void testMergeWithHighNumberOfCommonKeys() {
	// the size of the left and right inputs
	final int input1Size = 200;
	final int input2Size = 100;

	final int input1Duplicates = 10;
	final int input2Duplicates = 4000;
	final int duplicateKey = 13;

	try {
		final TupleGenerator generator1 = new TupleGenerator(SEED1, 500, 4096, KeyMode.SORTED, ValueMode.RANDOM_LENGTH);
		final TupleGenerator generator2 = new TupleGenerator(SEED2, 500, 2048, KeyMode.SORTED, ValueMode.RANDOM_LENGTH);

		final TestData.TupleGeneratorIterator gen1Iter = new TestData.TupleGeneratorIterator(generator1, input1Size);
		final TestData.TupleGeneratorIterator gen2Iter = new TestData.TupleGeneratorIterator(generator2, input2Size);

		final TestData.TupleConstantValueIterator const1Iter = new TestData.TupleConstantValueIterator(duplicateKey, "LEFT String for Duplicate Keys", input1Duplicates);
		final TestData.TupleConstantValueIterator const2Iter = new TestData.TupleConstantValueIterator(duplicateKey, "RIGHT String for Duplicate Keys", input2Duplicates);

		final List<MutableObjectIterator<Tuple2<Integer, String>>> inList1 = new ArrayList<>();
		inList1.add(gen1Iter);
		inList1.add(const1Iter);

		final List<MutableObjectIterator<Tuple2<Integer, String>>> inList2 = new ArrayList<>();
		inList2.add(gen2Iter);
		inList2.add(const2Iter);

		MutableObjectIterator<Tuple2<Integer, String>> input1 = new MergeIterator<>(inList1, comparator1.duplicate());
		MutableObjectIterator<Tuple2<Integer, String>> input2 = new MergeIterator<>(inList2, comparator2.duplicate());

		// collect expected data
		final Map<Integer, Collection<Match>> expectedMatchesMap = matchValues(
				collectData(input1),
				collectData(input2));

		// re-create the whole thing for actual processing

		// reset the generators and iterators
		generator1.reset();
		generator2.reset();
		const1Iter.reset();
		const2Iter.reset();
		gen1Iter.reset();
		gen2Iter.reset();

		inList1.clear();
		inList1.add(gen1Iter);
		inList1.add(const1Iter);

		inList2.clear();
		inList2.add(gen2Iter);
		inList2.add(const2Iter);

		input1 = new MergeIterator<>(inList1, comparator1.duplicate());
		input2 = new MergeIterator<>(inList2, comparator2.duplicate());

		StreamOperator operator = getOperator();

		match(expectedMatchesMap, transformToBinary(join(operator, input1, input2)));

		// assert that each expected match was seen
		for (Map.Entry<Integer, Collection<Match>> entry : expectedMatchesMap.entrySet()) {
			if (!entry.getValue().isEmpty()) {
				Assert.fail("Collection for key " + entry.getKey() + " is not empty");
			}
		}
	}
	catch (Exception e) {
		e.printStackTrace();
		Assert.fail("An exception occurred during the test: " + e.getMessage());
	}
}
 
Example 11
@Test
public void testMergeWithHighNumberOfCommonKeys()
{
	// the size of the left and right inputs
	final int INPUT_1_SIZE = 200;
	final int INPUT_2_SIZE = 100;
	
	final int INPUT_1_DUPLICATES = 10;
	final int INPUT_2_DUPLICATES = 4000;
	final int DUPLICATE_KEY = 13;
	
	try {
		final TupleGenerator generator1 = new TupleGenerator(SEED1, 500, 4096, KeyMode.SORTED, ValueMode.RANDOM_LENGTH);
		final TupleGenerator generator2 = new TupleGenerator(SEED2, 500, 2048, KeyMode.SORTED, ValueMode.RANDOM_LENGTH);

		final TestData.TupleGeneratorIterator gen1Iter = new TestData.TupleGeneratorIterator(generator1, INPUT_1_SIZE);
		final TestData.TupleGeneratorIterator gen2Iter = new TestData.TupleGeneratorIterator(generator2, INPUT_2_SIZE);

		final TestData.TupleConstantValueIterator const1Iter = new TestData.TupleConstantValueIterator(DUPLICATE_KEY, "LEFT String for Duplicate Keys", INPUT_1_DUPLICATES);
		final TestData.TupleConstantValueIterator const2Iter = new TestData.TupleConstantValueIterator(DUPLICATE_KEY, "RIGHT String for Duplicate Keys", INPUT_2_DUPLICATES);

		final List<MutableObjectIterator<Tuple2<Integer, String>>> inList1 = new ArrayList<MutableObjectIterator<Tuple2<Integer, String>>>();
		inList1.add(gen1Iter);
		inList1.add(const1Iter);

		final List<MutableObjectIterator<Tuple2<Integer, String>>> inList2 = new ArrayList<MutableObjectIterator<Tuple2<Integer, String>>>();
		inList2.add(gen2Iter);
		inList2.add(const2Iter);

		MutableObjectIterator<Tuple2<Integer, String>> input1 = new MergeIterator<Tuple2<Integer, String>>(inList1, comparator1.duplicate());
		MutableObjectIterator<Tuple2<Integer, String>> input2 = new MergeIterator<Tuple2<Integer, String>>(inList2, comparator2.duplicate());
		
		// collect expected data
		final Map<Integer, Collection<Match>> expectedMatchesMap = matchValues(
			collectData(input1),
			collectData(input2));
		
		// re-create the whole thing for actual processing
		
		// reset the generators and iterators
		generator1.reset();
		generator2.reset();
		const1Iter.reset();
		const2Iter.reset();
		gen1Iter.reset();
		gen2Iter.reset();
		
		inList1.clear();
		inList1.add(gen1Iter);
		inList1.add(const1Iter);
		
		inList2.clear();
		inList2.add(gen2Iter);
		inList2.add(const2Iter);

		input1 = new MergeIterator<Tuple2<Integer, String>>(inList1, comparator1.duplicate());
		input2 = new MergeIterator<Tuple2<Integer, String>>(inList2, comparator2.duplicate());
		
		final FlatJoinFunction<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> joinFunction = new MatchRemovingJoiner(expectedMatchesMap);
		
		final Collector<Tuple2<Integer, String>> collector = new DiscardingOutputCollector<Tuple2<Integer, String>>();

		
		// we create this sort-merge iterator with little memory for the block-nested-loops fall-back to make sure it
		// needs to spill for the duplicate keys
		NonReusingMergeInnerJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> iterator =
			new NonReusingMergeInnerJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>>(
				input1, input2, this.serializer1, this.comparator1, this.serializer2, this.comparator2,
				this.pairComparator, this.memoryManager, this.ioManager, PAGES_FOR_BNLJN, this.parentTask);

		iterator.open();
		
		while (iterator.callWithNextKey(joinFunction, collector));
		
		iterator.close();

		// assert that each expected match was seen
		for (Entry<Integer, Collection<Match>> entry : expectedMatchesMap.entrySet()) {
			if (!entry.getValue().isEmpty()) {
				Assert.fail("Collection for key " + entry.getKey() + " is not empty");
			}
		}
	}
	catch (Exception e) {
		e.printStackTrace();
		Assert.fail("An exception occurred during the test: " + e.getMessage());
	}
}
 
Example 12
Source Project: flink   File: HashVsSortMiniBenchmark.java    License: Apache License 2.0 4 votes vote down vote up
@Test
public void testSortBothMerge() {
	try {
		
		TestData.TupleGenerator generator1 = new TestData.TupleGenerator(SEED1, INPUT_1_SIZE / 10, 100, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
		TestData.TupleGenerator generator2 = new TestData.TupleGenerator(SEED2, INPUT_2_SIZE, 100, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);

		final TestData.TupleGeneratorIterator input1 = new TestData.TupleGeneratorIterator(generator1, INPUT_1_SIZE);
		final TestData.TupleGeneratorIterator input2 = new TestData.TupleGeneratorIterator(generator2, INPUT_2_SIZE);
		
		final FlatJoinFunction matcher = new NoOpMatcher();
		final Collector<Tuple2<Integer, String>> collector = new DiscardingOutputCollector<>();
		
		long start = System.nanoTime();
		
		final UnilateralSortMerger<Tuple2<Integer, String>> sorter1 = new UnilateralSortMerger<>(
				this.memoryManager, this.ioManager, input1, this.parentTask, this.serializer1, 
				this.comparator1.duplicate(), (double)MEMORY_FOR_SORTER/MEMORY_SIZE, 128, 0.8f,
				true /*use large record handler*/, true);
		
		final UnilateralSortMerger<Tuple2<Integer, String>> sorter2 = new UnilateralSortMerger<>(
				this.memoryManager, this.ioManager, input2, this.parentTask, this.serializer2, 
				this.comparator2.duplicate(), (double)MEMORY_FOR_SORTER/MEMORY_SIZE, 128, 0.8f,
				true /*use large record handler*/, true);
		
		final MutableObjectIterator<Tuple2<Integer, String>> sortedInput1 = sorter1.getIterator();
		final MutableObjectIterator<Tuple2<Integer, String>> sortedInput2 = sorter2.getIterator();
		
		// compare with iterator values
		ReusingMergeInnerJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> iterator =
			new ReusingMergeInnerJoinIterator<>(sortedInput1, sortedInput2,
					this.serializer1.getSerializer(), this.comparator1, this.serializer2.getSerializer(), this.comparator2, this.pairComparator11,
					this.memoryManager, this.ioManager, MEMORY_PAGES_FOR_MERGE, this.parentTask);
		
		iterator.open();
		
		while (iterator.callWithNextKey(matcher, collector));
		
		iterator.close();
		sorter1.close();
		sorter2.close();
		
		long elapsed = System.nanoTime() - start;
		double msecs = elapsed / (1000 * 1000);
		
		System.out.println("Sort-Merge Took " + msecs + " msecs.");
	}
	catch (Exception e) {
		e.printStackTrace();
		Assert.fail("An exception occurred during the test: " + e.getMessage());
	}
}
 
Example 13
@Test
public void testBuildFirst() {
	try {
		TupleGenerator generator1 = new TupleGenerator(SEED1, 500, 4096, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
		TupleGenerator generator2 = new TupleGenerator(SEED2, 500, 2048, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
		
		final TestData.TupleGeneratorIterator input1 = new TestData.TupleGeneratorIterator(generator1, INPUT_1_SIZE);
		final TestData.TupleGeneratorIterator input2 = new TestData.TupleGeneratorIterator(generator2, INPUT_2_SIZE);
		
		// collect expected data
		final Map<Integer, Collection<TupleMatch>> expectedMatchesMap = joinTuples(
				collectTupleData(input1),
				collectTupleData(input2));
		
		final TupleMatchRemovingJoin matcher = new TupleMatchRemovingJoin(expectedMatchesMap);
		final Collector<Tuple2<Integer, String>> collector = new DiscardingOutputCollector<Tuple2<Integer, String>>();

		// reset the generators
		generator1.reset();
		generator2.reset();
		input1.reset();
		input2.reset();

		// compare with iterator values
		NonReusingBuildFirstHashJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> iterator =
				new NonReusingBuildFirstHashJoinIterator<>(
					input1, input2, this.recordSerializer, this.record1Comparator, 
					this.recordSerializer, this.record2Comparator, this.recordPairComparator,
					this.memoryManager, ioManager, this.parentTask, 1.0, false, false, true);
		
		iterator.open();

		//noinspection StatementWithEmptyBody
		while (iterator.callWithNextKey(matcher, collector));
		
		iterator.close();

		// assert that each expected match was seen
		for (Entry<Integer, Collection<TupleMatch>> entry : expectedMatchesMap.entrySet()) {
			if (!entry.getValue().isEmpty()) {
				Assert.fail("Collection for key " + entry.getKey() + " is not empty");
			}
		}
	}
	catch (Exception e) {
		e.printStackTrace();
		Assert.fail("An exception occurred during the test: " + e.getMessage());
	}
}
 
Example 14
@Test
public void testBuildFirstWithHighNumberOfCommonKeys()
{
	// the size of the left and right inputs
	final int INPUT_1_SIZE = 200;
	final int INPUT_2_SIZE = 100;
	
	final int INPUT_1_DUPLICATES = 10;
	final int INPUT_2_DUPLICATES = 2000;
	final int DUPLICATE_KEY = 13;
	
	try {
		TupleGenerator generator1 = new TupleGenerator(SEED1, 500, 4096, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
		TupleGenerator generator2 = new TupleGenerator(SEED2, 500, 2048, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
		
		final TestData.TupleGeneratorIterator gen1Iter = new TestData.TupleGeneratorIterator(generator1, INPUT_1_SIZE);
		final TestData.TupleGeneratorIterator gen2Iter = new TestData.TupleGeneratorIterator(generator2, INPUT_2_SIZE);
		
		final TestData.TupleConstantValueIterator const1Iter = new TestData.TupleConstantValueIterator(DUPLICATE_KEY, "LEFT String for Duplicate Keys", INPUT_1_DUPLICATES);
		final TestData.TupleConstantValueIterator const2Iter = new TestData.TupleConstantValueIterator(DUPLICATE_KEY, "RIGHT String for Duplicate Keys", INPUT_2_DUPLICATES);
		
		final List<MutableObjectIterator<Tuple2<Integer, String>>> inList1 = new ArrayList<>();
		inList1.add(gen1Iter);
		inList1.add(const1Iter);
		
		final List<MutableObjectIterator<Tuple2<Integer, String>>> inList2 = new ArrayList<>();
		inList2.add(gen2Iter);
		inList2.add(const2Iter);
		
		MutableObjectIterator<Tuple2<Integer, String>> input1 = new UnionIterator<>(inList1);
		MutableObjectIterator<Tuple2<Integer, String>> input2 = new UnionIterator<>(inList2);
		
		
		// collect expected data
		final Map<Integer, Collection<TupleMatch>> expectedMatchesMap = joinTuples(
				collectTupleData(input1),
				collectTupleData(input2));
		
		// re-create the whole thing for actual processing
		
		// reset the generators and iterators
		generator1.reset();
		generator2.reset();
		const1Iter.reset();
		const2Iter.reset();
		gen1Iter.reset();
		gen2Iter.reset();
		
		inList1.clear();
		inList1.add(gen1Iter);
		inList1.add(const1Iter);
		
		inList2.clear();
		inList2.add(gen2Iter);
		inList2.add(const2Iter);

		input1 = new UnionIterator<>(inList1);
		input2 = new UnionIterator<>(inList2);
		
		final TupleMatchRemovingJoin matcher = new TupleMatchRemovingJoin(expectedMatchesMap);
		final Collector<Tuple2<Integer, String>> collector = new DiscardingOutputCollector<>();

		NonReusingBuildFirstHashJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> iterator =
				new NonReusingBuildFirstHashJoinIterator<>(
					input1, input2, this.recordSerializer, this.record1Comparator, 
					this.recordSerializer, this.record2Comparator, this.recordPairComparator,
					this.memoryManager, ioManager, this.parentTask, 1.0, false, false, true);

		iterator.open();
		
		while (iterator.callWithNextKey(matcher, collector));
		
		iterator.close();

		// assert that each expected match was seen
		for (Entry<Integer, Collection<TupleMatch>> entry : expectedMatchesMap.entrySet()) {
			if (!entry.getValue().isEmpty()) {
				Assert.fail("Collection for key " + entry.getKey() + " is not empty");
			}
		}
	}
	catch (Exception e) {
		e.printStackTrace();
		Assert.fail("An exception occurred during the test: " + e.getMessage());
	}
}
 
Example 15
@Test
public void testBuildSecondAndProbeSideOuterJoin() {
	try {
		TestData.TupleGenerator generator1 = new TestData.TupleGenerator(SEED1, 1000, 4096, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
		TestData.TupleGenerator generator2 = new TestData.TupleGenerator(SEED2, 500, 2048, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);

		final TestData.TupleGeneratorIterator input1 = new TestData.TupleGeneratorIterator(generator1, INPUT_1_SIZE);
		final TestData.TupleGeneratorIterator input2 = new TestData.TupleGeneratorIterator(generator2, INPUT_2_SIZE);

		// collect expected data
		final Map<Integer, Collection<TupleMatch>> expectedMatchesMap = leftOuterJoinTuples(
				collectTupleData(input1),
				collectTupleData(input2));

		final FlatJoinFunction matcher = new TupleMatchRemovingJoin(expectedMatchesMap);
		final Collector<Tuple2<Integer, String>> collector = new DiscardingOutputCollector<>();

		// reset the generators
		generator1.reset();
		generator2.reset();
		input1.reset();
		input2.reset();

		// compare with iterator values
		ReusingBuildSecondHashJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> iterator =
				new ReusingBuildSecondHashJoinIterator<>(
						input1, input2, this.recordSerializer, this.record1Comparator,
						this.recordSerializer, this.record2Comparator, this.recordPairComparator,
						this.memoryManager, ioManager, this.parentTask, 1.0, true, false, false);

		iterator.open();

		while (iterator.callWithNextKey(matcher, collector));

		iterator.close();

		// assert that each expected match was seen
		for (Entry<Integer, Collection<TupleMatch>> entry : expectedMatchesMap.entrySet()) {
			if (!entry.getValue().isEmpty()) {
				Assert.fail("Collection for key " + entry.getKey() + " is not empty");
			}
		}
	}
	catch (Exception e) {
		e.printStackTrace();
		Assert.fail("An exception occurred during the test: " + e.getMessage());
	}
}
 
Example 16
@Test
public void testBuildFirst() {
	try {
		TestData.TupleGenerator generator1 = new TestData.TupleGenerator(SEED1, 500, 4096, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
		TestData.TupleGenerator generator2 = new TestData.TupleGenerator(SEED2, 500, 2048, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
		
		final TestData.TupleGeneratorIterator input1 = new TestData.TupleGeneratorIterator(generator1, INPUT_1_SIZE);
		final TestData.TupleGeneratorIterator input2 = new TestData.TupleGeneratorIterator(generator2, INPUT_2_SIZE);
		
		// collect expected data
		final Map<Integer, Collection<TupleMatch>> expectedMatchesMap = joinTuples(
				collectTupleData(input1),
				collectTupleData(input2));
		
		final FlatJoinFunction matcher = new TupleMatchRemovingJoin(expectedMatchesMap);
		final Collector<Tuple2<Integer, String>> collector = new DiscardingOutputCollector<>();

		// reset the generators
		generator1.reset();
		generator2.reset();
		input1.reset();
		input2.reset();

		// compare with iterator values
		ReusingBuildFirstHashJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> iterator =
				new ReusingBuildFirstHashJoinIterator<>(
					input1, input2, this.recordSerializer, this.record1Comparator, 
					this.recordSerializer, this.record2Comparator, this.recordPairComparator,
					this.memoryManager, ioManager, this.parentTask, 1.0, false, false, true);
		
		iterator.open();
		
		while (iterator.callWithNextKey(matcher, collector));
		
		iterator.close();

		// assert that each expected match was seen
		for (Entry<Integer, Collection<TupleMatch>> entry : expectedMatchesMap.entrySet()) {
			if (!entry.getValue().isEmpty()) {
				Assert.fail("Collection for key " + entry.getKey() + " is not empty");
			}
		}
	}
	catch (Exception e) {
		e.printStackTrace();
		Assert.fail("An exception occurred during the test: " + e.getMessage());
	}
}
 
Example 17
@Test
public void testBuildSecondAndFullOuterJoin() {
	try {
		TestData.TupleGenerator generator1 = new TestData.TupleGenerator(SEED1, 1000, 4096, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
		TestData.TupleGenerator generator2 = new TestData.TupleGenerator(SEED2, 500, 2048, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);

		final TestData.TupleGeneratorIterator input1 = new TestData.TupleGeneratorIterator(generator1, INPUT_1_SIZE);
		final TestData.TupleGeneratorIterator input2 = new TestData.TupleGeneratorIterator(generator2, INPUT_2_SIZE);

		// collect expected data
		final Map<Integer, Collection<TupleMatch>> expectedMatchesMap = fullOuterJoinTuples(
			collectTupleData(input1),
			collectTupleData(input2));

		final FlatJoinFunction matcher = new TupleMatchRemovingJoin(expectedMatchesMap);
		final Collector<Tuple2<Integer, String>> collector = new DiscardingOutputCollector<>();

		// reset the generators
		generator1.reset();
		generator2.reset();
		input1.reset();
		input2.reset();

		// compare with iterator values
		ReusingBuildSecondHashJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> iterator =
			new ReusingBuildSecondHashJoinIterator<>(
				input1, input2, this.recordSerializer, this.record1Comparator,
				this.recordSerializer, this.record2Comparator, this.recordPairComparator,
				this.memoryManager, ioManager, this.parentTask, 1.0, true, true, false);

		iterator.open();

		while (iterator.callWithNextKey(matcher, collector));

		iterator.close();

		// assert that each expected match was seen
		for (Entry<Integer, Collection<TupleMatch>> entry : expectedMatchesMap.entrySet()) {
			if (!entry.getValue().isEmpty()) {
				Assert.fail("Collection for key " + entry.getKey() + " is not empty");
			}
		}
	}
	catch (Exception e) {
		e.printStackTrace();
		Assert.fail("An exception occurred during the test: " + e.getMessage());
	}
}
 
Example 18
@Test
public void testBuildSecondWithHighNumberOfCommonKeys()
{
	// the size of the left and right inputs
	final int INPUT_1_SIZE = 200;
	final int INPUT_2_SIZE = 100;
	
	final int INPUT_1_DUPLICATES = 10;
	final int INPUT_2_DUPLICATES = 2000;
	final int DUPLICATE_KEY = 13;
	
	try {
		TestData.TupleGenerator generator1 = new TestData.TupleGenerator(SEED1, 500, 4096, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
		TestData.TupleGenerator generator2 = new TestData.TupleGenerator(SEED2, 500, 2048, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
		
		final TestData.TupleGeneratorIterator gen1Iter = new TestData.TupleGeneratorIterator(generator1, INPUT_1_SIZE);
		final TestData.TupleGeneratorIterator gen2Iter = new TestData.TupleGeneratorIterator(generator2, INPUT_2_SIZE);
		
		final TestData.TupleConstantValueIterator const1Iter = new TestData.TupleConstantValueIterator(DUPLICATE_KEY, "LEFT String for Duplicate Keys", INPUT_1_DUPLICATES);
		final TestData.TupleConstantValueIterator const2Iter = new TestData.TupleConstantValueIterator(DUPLICATE_KEY, "RIGHT String for Duplicate Keys", INPUT_2_DUPLICATES);
		
		final List<MutableObjectIterator<Tuple2<Integer, String>>> inList1 = new ArrayList<>();
		inList1.add(gen1Iter);
		inList1.add(const1Iter);
		
		final List<MutableObjectIterator<Tuple2<Integer, String>>> inList2 = new ArrayList<>();
		inList2.add(gen2Iter);
		inList2.add(const2Iter);
		
		MutableObjectIterator<Tuple2<Integer, String>> input1 = new UnionIterator<>(inList1);
		MutableObjectIterator<Tuple2<Integer, String>> input2 = new UnionIterator<>(inList2);
		
		
		// collect expected data
		final Map<Integer, Collection<TupleMatch>> expectedMatchesMap = joinTuples(
				collectTupleData(input1),
				collectTupleData(input2));
		
		// re-create the whole thing for actual processing
		
		// reset the generators and iterators
		generator1.reset();
		generator2.reset();
		const1Iter.reset();
		const2Iter.reset();
		gen1Iter.reset();
		gen2Iter.reset();
		
		inList1.clear();
		inList1.add(gen1Iter);
		inList1.add(const1Iter);
		
		inList2.clear();
		inList2.add(gen2Iter);
		inList2.add(const2Iter);

		input1 = new UnionIterator<>(inList1);
		input2 = new UnionIterator<>(inList2);
		
		final FlatJoinFunction matcher = new TupleMatchRemovingJoin(expectedMatchesMap);
		final Collector<Tuple2<Integer, String>> collector = new DiscardingOutputCollector<>();

		ReusingBuildSecondHashJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> iterator =
			new ReusingBuildSecondHashJoinIterator<>(
				input1, input2, this.recordSerializer, this.record1Comparator, 
				this.recordSerializer, this.record2Comparator, this.recordPairComparator,
				this.memoryManager, ioManager, this.parentTask, 1.0, false, false, true);
		
		iterator.open();
		
		while (iterator.callWithNextKey(matcher, collector));
		
		iterator.close();

		// assert that each expected match was seen
		for (Entry<Integer, Collection<TupleMatch>> entry : expectedMatchesMap.entrySet()) {
			if (!entry.getValue().isEmpty()) {
				Assert.fail("Collection for key " + entry.getKey() + " is not empty");
			}
		}
	}
	catch (Exception e) {
		e.printStackTrace();
		Assert.fail("An exception occurred during the test: " + e.getMessage());
	}
}
 
Example 19
@Test
public void testMergeWithHighNumberOfCommonKeys()
{
	// the size of the left and right inputs
	final int INPUT_1_SIZE = 200;
	final int INPUT_2_SIZE = 100;
	
	final int INPUT_1_DUPLICATES = 10;
	final int INPUT_2_DUPLICATES = 4000;
	final int DUPLICATE_KEY = 13;
	
	try {
		final TupleGenerator generator1 = new TupleGenerator(SEED1, 500, 4096, KeyMode.SORTED, ValueMode.RANDOM_LENGTH);
		final TupleGenerator generator2 = new TupleGenerator(SEED2, 500, 2048, KeyMode.SORTED, ValueMode.RANDOM_LENGTH);

		final TestData.TupleGeneratorIterator gen1Iter = new TestData.TupleGeneratorIterator(generator1, INPUT_1_SIZE);
		final TestData.TupleGeneratorIterator gen2Iter = new TestData.TupleGeneratorIterator(generator2, INPUT_2_SIZE);

		final TestData.TupleConstantValueIterator const1Iter = new TestData.TupleConstantValueIterator(DUPLICATE_KEY, "LEFT String for Duplicate Keys", INPUT_1_DUPLICATES);
		final TestData.TupleConstantValueIterator const2Iter = new TestData.TupleConstantValueIterator(DUPLICATE_KEY, "RIGHT String for Duplicate Keys", INPUT_2_DUPLICATES);

		final List<MutableObjectIterator<Tuple2<Integer, String>>> inList1 = new ArrayList<MutableObjectIterator<Tuple2<Integer, String>>>();
		inList1.add(gen1Iter);
		inList1.add(const1Iter);

		final List<MutableObjectIterator<Tuple2<Integer, String>>> inList2 = new ArrayList<MutableObjectIterator<Tuple2<Integer, String>>>();
		inList2.add(gen2Iter);
		inList2.add(const2Iter);

		MutableObjectIterator<Tuple2<Integer, String>> input1 = new MergeIterator<Tuple2<Integer, String>>(inList1, comparator1.duplicate());
		MutableObjectIterator<Tuple2<Integer, String>> input2 = new MergeIterator<Tuple2<Integer, String>>(inList2, comparator2.duplicate());
		
		// collect expected data
		final Map<Integer, Collection<Match>> expectedMatchesMap = matchValues(
			collectData(input1),
			collectData(input2));
		
		// re-create the whole thing for actual processing
		
		// reset the generators and iterators
		generator1.reset();
		generator2.reset();
		const1Iter.reset();
		const2Iter.reset();
		gen1Iter.reset();
		gen2Iter.reset();
		
		inList1.clear();
		inList1.add(gen1Iter);
		inList1.add(const1Iter);
		
		inList2.clear();
		inList2.add(gen2Iter);
		inList2.add(const2Iter);

		input1 = new MergeIterator<Tuple2<Integer, String>>(inList1, comparator1.duplicate());
		input2 = new MergeIterator<Tuple2<Integer, String>>(inList2, comparator2.duplicate());
		
		final FlatJoinFunction<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> joinFunction = new MatchRemovingJoiner(expectedMatchesMap);
		
		final Collector<Tuple2<Integer, String>> collector = new DiscardingOutputCollector<Tuple2<Integer, String>>();

		
		// we create this sort-merge iterator with little memory for the block-nested-loops fall-back to make sure it
		// needs to spill for the duplicate keys
		NonReusingMergeInnerJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> iterator =
			new NonReusingMergeInnerJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>>(
				input1, input2, this.serializer1, this.comparator1, this.serializer2, this.comparator2,
				this.pairComparator, this.memoryManager, this.ioManager, PAGES_FOR_BNLJN, this.parentTask);

		iterator.open();
		
		while (iterator.callWithNextKey(joinFunction, collector));
		
		iterator.close();

		// assert that each expected match was seen
		for (Entry<Integer, Collection<Match>> entry : expectedMatchesMap.entrySet()) {
			if (!entry.getValue().isEmpty()) {
				Assert.fail("Collection for key " + entry.getKey() + " is not empty");
			}
		}
	}
	catch (Exception e) {
		e.printStackTrace();
		Assert.fail("An exception occurred during the test: " + e.getMessage());
	}
}
 
Example 20
@Test
public void testMergeWithHighNumberOfCommonKeys()
{
	// the size of the left and right inputs
	final int INPUT_1_SIZE = 200;
	final int INPUT_2_SIZE = 100;
	
	final int INPUT_1_DUPLICATES = 10;
	final int INPUT_2_DUPLICATES = 4000;
	final int DUPLICATE_KEY = 13;
	
	try {
		final TupleGenerator generator1 = new TupleGenerator(SEED1, 500, 4096, KeyMode.SORTED, ValueMode.RANDOM_LENGTH);
		final TupleGenerator generator2 = new TupleGenerator(SEED2, 500, 2048, KeyMode.SORTED, ValueMode.RANDOM_LENGTH);

		final TestData.TupleGeneratorIterator gen1Iter = new TestData.TupleGeneratorIterator(generator1, INPUT_1_SIZE);
		final TestData.TupleGeneratorIterator gen2Iter = new TestData.TupleGeneratorIterator(generator2, INPUT_2_SIZE);

		final TestData.TupleConstantValueIterator const1Iter = new TestData.TupleConstantValueIterator(DUPLICATE_KEY, "LEFT String for Duplicate Keys", INPUT_1_DUPLICATES);
		final TestData.TupleConstantValueIterator const2Iter = new TestData.TupleConstantValueIterator(DUPLICATE_KEY, "RIGHT String for Duplicate Keys", INPUT_2_DUPLICATES);

		final List<MutableObjectIterator<Tuple2<Integer, String>>> inList1 = new ArrayList<MutableObjectIterator<Tuple2<Integer, String>>>();
		inList1.add(gen1Iter);
		inList1.add(const1Iter);

		final List<MutableObjectIterator<Tuple2<Integer, String>>> inList2 = new ArrayList<MutableObjectIterator<Tuple2<Integer, String>>>();
		inList2.add(gen2Iter);
		inList2.add(const2Iter);

		MutableObjectIterator<Tuple2<Integer, String>> input1 = new MergeIterator<Tuple2<Integer, String>>(inList1, comparator1.duplicate());
		MutableObjectIterator<Tuple2<Integer, String>> input2 = new MergeIterator<Tuple2<Integer, String>>(inList2, comparator2.duplicate());
		
		// collect expected data
		final Map<Integer, Collection<Match>> expectedMatchesMap = matchValues(
			collectData(input1),
			collectData(input2));
		
		// re-create the whole thing for actual processing
		
		// reset the generators and iterators
		generator1.reset();
		generator2.reset();
		const1Iter.reset();
		const2Iter.reset();
		gen1Iter.reset();
		gen2Iter.reset();
		
		inList1.clear();
		inList1.add(gen1Iter);
		inList1.add(const1Iter);
		
		inList2.clear();
		inList2.add(gen2Iter);
		inList2.add(const2Iter);

		input1 = new MergeIterator<Tuple2<Integer, String>>(inList1, comparator1.duplicate());
		input2 = new MergeIterator<Tuple2<Integer, String>>(inList2, comparator2.duplicate());
		
		final FlatJoinFunction<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> joinFunction = new MatchRemovingJoiner(expectedMatchesMap);
		
		final Collector<Tuple2<Integer, String>> collector = new DiscardingOutputCollector<Tuple2<Integer, String>>();

		
		// we create this sort-merge iterator with little memory for the block-nested-loops fall-back to make sure it
		// needs to spill for the duplicate keys
		NonReusingMergeInnerJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>> iterator =
			new NonReusingMergeInnerJoinIterator<Tuple2<Integer, String>, Tuple2<Integer, String>, Tuple2<Integer, String>>(
				input1, input2, this.serializer1, this.comparator1, this.serializer2, this.comparator2,
				this.pairComparator, this.memoryManager, this.ioManager, PAGES_FOR_BNLJN, this.parentTask);

		iterator.open();
		
		while (iterator.callWithNextKey(joinFunction, collector));
		
		iterator.close();

		// assert that each expected match was seen
		for (Entry<Integer, Collection<Match>> entry : expectedMatchesMap.entrySet()) {
			if (!entry.getValue().isEmpty()) {
				Assert.fail("Collection for key " + entry.getKey() + " is not empty");
			}
		}
	}
	catch (Exception e) {
		e.printStackTrace();
		Assert.fail("An exception occurred during the test: " + e.getMessage());
	}
}