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

The following examples show how to use org.apache.flink.runtime.operators.testutils.TestData#TupleGenerator . 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 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 3
@Test
public void testBuildSecondAndBuildSideOuterJoin() {
	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 = 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
		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, 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 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
Source Project: Flink-CEPplus   File: ChannelViewsTest.java    License: Apache License 2.0 4 votes vote down vote up
@Test
public void testWriteReadNotAll() throws Exception
{
	final TestData.TupleGenerator generator = new TestData.TupleGenerator(SEED, KEY_MAX, VALUE_SHORT_LENGTH, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
	final FileIOChannel.ID channel = this.ioManager.createChannel();
	final TypeSerializer<Tuple2<Integer, String>> serializer = TestData.getIntStringTupleSerializer();
	
	// create the writer output view
	List<MemorySegment> memory = this.memoryManager.allocatePages(this.parentTask, NUM_MEMORY_SEGMENTS);
	final BlockChannelWriter<MemorySegment> writer = this.ioManager.createBlockChannelWriter(channel);
	final ChannelWriterOutputView outView = new ChannelWriterOutputView(writer, memory, MEMORY_PAGE_SIZE);
	
	// write a number of pairs
	final Tuple2<Integer, String> rec = new Tuple2<>();
	for (int i = 0; i < NUM_PAIRS_SHORT; i++) {
		generator.next(rec);
		serializer.serialize(rec, outView);
	}
	this.memoryManager.release(outView.close());
	
	// create the reader input view
	memory = this.memoryManager.allocatePages(this.parentTask, NUM_MEMORY_SEGMENTS);
	final BlockChannelReader<MemorySegment> reader = this.ioManager.createBlockChannelReader(channel);
	final ChannelReaderInputView inView = new ChannelReaderInputView(reader, memory, outView.getBlockCount(), true);
	generator.reset();
	
	// read and re-generate all records and compare them
	final Tuple2<Integer, String> readRec = new Tuple2<>();
	for (int i = 0; i < NUM_PAIRS_SHORT / 2; i++) {
		generator.next(rec);
		serializer.deserialize(readRec, inView);
		
		int k1 = rec.f0;
		String v1 = rec.f1;
		
		int k2 = readRec.f0;
		String v2 = readRec.f1;
		
		Assert.assertTrue("The re-generated and the read record do not match.", k1 == k2 && v1.equals(v2));
	}
	
	this.memoryManager.release(inView.close());
	reader.deleteChannel();
}
 
Example 6
@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 7
@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 8
Source Project: flink   File: NormalizedKeySorterTest.java    License: Apache License 2.0 4 votes vote down vote up
@Test
public void testSortLongStringKeys() throws Exception {
	final int numSegments = MEMORY_SIZE / MEMORY_PAGE_SIZE;
	final List<MemorySegment> memory = this.memoryManager.allocatePages(new DummyInvokable(), numSegments);
	
	@SuppressWarnings("unchecked")
	TypeComparator<Tuple2<Integer, String>> accessors = TestData.getIntStringTupleTypeInfo().createComparator(new int[]{1}, new boolean[]{true}, 0, null);
	NormalizedKeySorter<Tuple2<Integer, String>> sorter = new NormalizedKeySorter<>(TestData.getIntStringTupleSerializer(), accessors, memory);
	
	TestData.TupleGenerator generator = new TestData.TupleGenerator(SEED, KEY_MAX, VALUE_LENGTH, KeyMode.RANDOM,
		ValueMode.FIX_LENGTH);
	
	// write the records
	Tuple2<Integer, String> record = new Tuple2<>();
	do {
		generator.next(record);
	}
	while (sorter.write(record));
	
	QuickSort qs = new QuickSort();
	qs.sort(sorter);
	
	MutableObjectIterator<Tuple2<Integer, String>> iter = sorter.getIterator();
	Tuple2<Integer, String> readTarget = new Tuple2<>();
	
	iter.next(readTarget);
	String last = readTarget.f1;
	
	while ((readTarget = iter.next(readTarget)) != null) {
		String current = readTarget.f1;
		
		final int cmp = last.compareTo(current);
		if (cmp > 0) {
			Assert.fail("Next value is not larger or equal to previous value.");
		}
		
		last = current;
	}
	
	// release the memory occupied by the buffers
	sorter.dispose();
	this.memoryManager.release(memory);
}
 
Example 9
Source Project: flink   File: NormalizedKeySorterTest.java    License: Apache License 2.0 4 votes vote down vote up
@Test
public void testSortShortStringKeys() throws Exception {
	final int numSegments = MEMORY_SIZE / MEMORY_PAGE_SIZE;
	final List<MemorySegment> memory = this.memoryManager.allocatePages(new DummyInvokable(), numSegments);
	
	@SuppressWarnings("unchecked")
	TypeComparator<Tuple2<Integer, String>> accessors = TestData.getIntStringTupleTypeInfo().createComparator(new int[]{1}, new boolean[]{true}, 0, null);
	NormalizedKeySorter<Tuple2<Integer, String>> sorter = new NormalizedKeySorter<>(TestData.getIntStringTupleSerializer(), accessors, memory);
	
	TestData.TupleGenerator generator = new TestData.TupleGenerator(SEED, KEY_MAX, 5, KeyMode.RANDOM,
		ValueMode.FIX_LENGTH);
	
	// write the records
	Tuple2<Integer, String> record = new Tuple2<>();
	do {
		generator.next(record);
	}
	while (sorter.write(record));
	
	QuickSort qs = new QuickSort();
	qs.sort(sorter);
	
	MutableObjectIterator<Tuple2<Integer, String>> iter = sorter.getIterator();
	Tuple2<Integer, String> readTarget = new Tuple2<>();

	iter.next(readTarget);
	String last = readTarget.f1;
	
	while ((readTarget = iter.next(readTarget)) != null) {
		String current = readTarget.f1;
		
		final int cmp = last.compareTo(current);
		if (cmp > 0) {
			Assert.fail("Next value is not larger or equal to previous value.");
		}
		
		last = current;
	}
	
	// release the memory occupied by the buffers
	sorter.dispose();
	this.memoryManager.release(memory);
}
 
Example 10
@Test
public void testSortAndValidate() throws Exception
{
	final Hashtable<Integer, Integer> countTable = new Hashtable<>(KEY_MAX);
	for (int i = 1; i <= KEY_MAX; i++) {
		countTable.put(i, 0);
	}

	// comparator
	final TypeComparator<Integer> keyComparator = new IntComparator(true);

	// reader
	TestData.MockTuple2Reader<Tuple2<Integer, String>> reader = TestData.getIntStringTupleReader();

	// merge iterator
	LOG.debug("initializing sortmerger");
	
	TestCountCombiner2 comb = new TestCountCombiner2();
	
	Sorter<Tuple2<Integer, String>> merger = new CombiningUnilateralSortMerger<>(comb,
			this.memoryManager, this.ioManager, reader, this.parentTask, this.serializerFactory1, this.comparator1,
			0.25, 2, 0.7f, true /* use large record handler */, false);

	// emit data
	LOG.debug("emitting data");
	TestData.TupleGenerator generator = new TestData.TupleGenerator(SEED, KEY_MAX, VALUE_LENGTH, KeyMode.RANDOM, ValueMode.FIX_LENGTH);
	Tuple2<Integer, String> rec = new Tuple2<>();
	
	for (int i = 0; i < NUM_PAIRS; i++) {
		Assert.assertTrue((rec = generator.next(rec)) != null);
		final Integer key = rec.f0;
		rec.setField("1", 1);
		reader.emit(rec);
		
		countTable.put(key, countTable.get(key) + 1);
	}
	reader.close();

	// check order
	MutableObjectIterator<Tuple2<Integer, String>> iterator = merger.getIterator();
	
	LOG.debug("checking results");
	
	Tuple2<Integer, String> rec1 = new Tuple2<>();
	Tuple2<Integer, String> rec2 = new Tuple2<>();
	
	Assert.assertTrue((rec1 = iterator.next(rec1)) != null);
	countTable.put(rec1.f0, countTable.get(rec1.f0) - (Integer.parseInt(rec1.f1)));

	while ((rec2 = iterator.next(rec2)) != null) {
		int k1 = rec1.f0;
		int k2 = rec2.f0;
		
		Assert.assertTrue(keyComparator.compare(k1, k2) <= 0); 
		countTable.put(k2, countTable.get(k2) - (Integer.parseInt(rec2.f1)));
		
		rec1 = rec2;
	}

	for (Integer cnt : countTable.values()) {
		Assert.assertTrue(cnt == 0);
	}
	
	merger.close();
	
	// if the combiner was opened, it must have been closed
	Assert.assertTrue(comb.opened == comb.closed);
}
 
Example 11
Source Project: flink   File: NormalizedKeySorterTest.java    License: Apache License 2.0 4 votes vote down vote up
@Test
public void testReset() throws Exception {
	final int numSegments = MEMORY_SIZE / MEMORY_PAGE_SIZE;
	final List<MemorySegment> memory = this.memoryManager.allocatePages(new DummyInvokable(), numSegments);
	
	NormalizedKeySorter<Tuple2<Integer, String>> sorter = newSortBuffer(memory);
	TestData.TupleGenerator generator = new TestData.TupleGenerator(SEED, KEY_MAX, VALUE_LENGTH, KeyMode.RANDOM, ValueMode.FIX_LENGTH);
	
	// write the buffer full with the first set of records
	Tuple2<Integer, String> record = new Tuple2<>();
	int num = -1;
	do {
		generator.next(record);
		num++;
	}
	while (sorter.write(record));
	
	sorter.reset();
	
	// write a second sequence of records. since the values are of fixed length, we must be able to write an equal number
	generator = new TestData.TupleGenerator(SEED2, KEY_MAX, VALUE_LENGTH, KeyMode.RANDOM, ValueMode.FIX_LENGTH);
	
	// write the buffer full with the first set of records
	int num2 = -1;
	do {
		generator.next(record);
		num2++;
	}
	while (sorter.write(record));
	
	Assert.assertEquals("The number of records written after the reset was not the same as before.", num, num2);
	
	// re-read the records
	generator.reset();
	Tuple2<Integer, String> readTarget = new Tuple2<>();
	
	int i = 0;
	while (i < num) {
		generator.next(record);
		readTarget = sorter.getRecord(readTarget, i++);
		
		int rk = readTarget.f0;
		int gk = record.f0;
		
		String rv = readTarget.f1;
		String gv = record.f1;
		
		Assert.assertEquals("The re-read key is wrong", gk, rk);
		Assert.assertEquals("The re-read value is wrong", gv, rv);
	}
	
	// release the memory occupied by the buffers
	sorter.dispose();
	this.memoryManager.release(memory);
}
 
Example 12
Source Project: flink   File: ChannelViewsTest.java    License: Apache License 2.0 4 votes vote down vote up
@Test
public void testReadWithoutKnownBlockCount() throws Exception
{
	final TestData.TupleGenerator generator = new TestData.TupleGenerator(SEED, KEY_MAX, VALUE_SHORT_LENGTH, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
	final FileIOChannel.ID channel = this.ioManager.createChannel();
	final TypeSerializer<Tuple2<Integer, String>> serializer = TestData.getIntStringTupleSerializer();
	
	// create the writer output view
	List<MemorySegment> memory = this.memoryManager.allocatePages(this.parentTask, NUM_MEMORY_SEGMENTS);
	final BlockChannelWriter<MemorySegment> writer = this.ioManager.createBlockChannelWriter(channel);
	final ChannelWriterOutputView outView = new ChannelWriterOutputView(writer, memory, MEMORY_PAGE_SIZE);
	
	// write a number of pairs
	final Tuple2<Integer, String> rec = new Tuple2<>();
	for (int i = 0; i < NUM_PAIRS_SHORT; i++) {
		generator.next(rec);
		serializer.serialize(rec, outView);
	}
	this.memoryManager.release(outView.close());
	
	// create the reader input view
	memory = this.memoryManager.allocatePages(this.parentTask, NUM_MEMORY_SEGMENTS);
	final BlockChannelReader<MemorySegment> reader = this.ioManager.createBlockChannelReader(channel);
	final ChannelReaderInputView inView = new ChannelReaderInputView(reader, memory, true);
	generator.reset();
	
	// read and re-generate all records and compare them
	final Tuple2<Integer, String> readRec = new Tuple2<>();
	for (int i = 0; i < NUM_PAIRS_SHORT; i++) {
		generator.next(rec);
		serializer.deserialize(readRec, inView);
		
		int k1 = rec.f0;
		String v1 = rec.f1;
		
		int k2 = readRec.f0;
		String v2 = readRec.f1;
		
		Assert.assertTrue("The re-generated and the read record do not match.", k1 == k2 && v1.equals(v2));
	}
	
	this.memoryManager.release(inView.close());
	reader.deleteChannel();
}
 
Example 13
Source Project: flink   File: NormalizedKeySorterTest.java    License: Apache License 2.0 4 votes vote down vote up
@Test
public void testSortShortStringKeys() throws Exception {
	final int numSegments = MEMORY_SIZE / MEMORY_PAGE_SIZE;
	final List<MemorySegment> memory = this.memoryManager.allocatePages(new DummyInvokable(), numSegments);
	
	@SuppressWarnings("unchecked")
	TypeComparator<Tuple2<Integer, String>> accessors = TestData.getIntStringTupleTypeInfo().createComparator(new int[]{1}, new boolean[]{true}, 0, null);
	NormalizedKeySorter<Tuple2<Integer, String>> sorter = new NormalizedKeySorter<>(TestData.getIntStringTupleSerializer(), accessors, memory);
	
	TestData.TupleGenerator generator = new TestData.TupleGenerator(SEED, KEY_MAX, 5, KeyMode.RANDOM,
		ValueMode.FIX_LENGTH);
	
	// write the records
	Tuple2<Integer, String> record = new Tuple2<>();
	do {
		generator.next(record);
	}
	while (sorter.write(record));
	
	QuickSort qs = new QuickSort();
	qs.sort(sorter);
	
	MutableObjectIterator<Tuple2<Integer, String>> iter = sorter.getIterator();
	Tuple2<Integer, String> readTarget = new Tuple2<>();

	iter.next(readTarget);
	String last = readTarget.f1;
	
	while ((readTarget = iter.next(readTarget)) != null) {
		String current = readTarget.f1;
		
		final int cmp = last.compareTo(current);
		if (cmp > 0) {
			Assert.fail("Next value is not larger or equal to previous value.");
		}
		
		last = current;
	}
	
	// release the memory occupied by the buffers
	sorter.dispose();
	this.memoryManager.release(memory);
}
 
Example 14
Source Project: flink   File: ChannelViewsTest.java    License: Apache License 2.0 4 votes vote down vote up
@Test
public void testWriteReadOneBufferOnly() throws Exception
{
	final TestData.TupleGenerator generator = new TestData.TupleGenerator(SEED, KEY_MAX, VALUE_SHORT_LENGTH, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
	final FileIOChannel.ID channel = this.ioManager.createChannel();
	final TypeSerializer<Tuple2<Integer, String>> serializer = TestData.getIntStringTupleSerializer();
	
	// create the writer output view
	List<MemorySegment> memory = this.memoryManager.allocatePages(this.parentTask, 1);
	final BlockChannelWriter<MemorySegment> writer = this.ioManager.createBlockChannelWriter(channel);
	final ChannelWriterOutputView outView = new ChannelWriterOutputView(writer, memory, MEMORY_PAGE_SIZE);
	
	// write a number of pairs
	final Tuple2<Integer, String> rec = new Tuple2<>();
	for (int i = 0; i < NUM_PAIRS_SHORT; i++) {
		generator.next(rec);
		serializer.serialize(rec, outView);
	}
	this.memoryManager.release(outView.close());
	
	// create the reader input view
	memory = this.memoryManager.allocatePages(this.parentTask, 1);
	final BlockChannelReader<MemorySegment> reader = this.ioManager.createBlockChannelReader(channel);
	final ChannelReaderInputView inView = new ChannelReaderInputView(reader, memory, outView.getBlockCount(), true);
	generator.reset();
	
	// read and re-generate all records and compare them
	final Tuple2<Integer, String> readRec = new Tuple2<>();
	for (int i = 0; i < NUM_PAIRS_SHORT; i++) {
		generator.next(rec);
		serializer.deserialize(readRec, inView);
		
		int k1 = rec.f0;
		String v1 = rec.f1;
		
		int k2 = readRec.f0;
		String v2 = readRec.f1;
		
		Assert.assertTrue("The re-generated and the read record do not match.", k1 == k2 && v1.equals(v2));
	}
	
	this.memoryManager.release(inView.close());
	reader.deleteChannel();
}
 
Example 15
Source Project: Flink-CEPplus   File: ChannelViewsTest.java    License: Apache License 2.0 4 votes vote down vote up
@Test
public void testReadTooMany() throws Exception
{
	final TestData.TupleGenerator generator = new TestData.TupleGenerator(SEED, KEY_MAX, VALUE_SHORT_LENGTH, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
	final FileIOChannel.ID channel = this.ioManager.createChannel();
	final TypeSerializer<Tuple2<Integer, String>> serializer = TestData.getIntStringTupleSerializer();
	
	// create the writer output view
	List<MemorySegment> memory = this.memoryManager.allocatePages(this.parentTask, NUM_MEMORY_SEGMENTS);
	final BlockChannelWriter<MemorySegment> writer = this.ioManager.createBlockChannelWriter(channel);
	final ChannelWriterOutputView outView = new ChannelWriterOutputView(writer, memory, MEMORY_PAGE_SIZE);

	// write a number of pairs
	final Tuple2<Integer, String> rec = new Tuple2<>();
	for (int i = 0; i < NUM_PAIRS_SHORT; i++) {
		generator.next(rec);
		serializer.serialize(rec, outView);
	}
	this.memoryManager.release(outView.close());

	// create the reader input view
	memory = this.memoryManager.allocatePages(this.parentTask, NUM_MEMORY_SEGMENTS);
	final BlockChannelReader<MemorySegment> reader = this.ioManager.createBlockChannelReader(channel);
	final ChannelReaderInputView inView = new ChannelReaderInputView(reader, memory, outView.getBlockCount(), true);
	generator.reset();

	// read and re-generate all records and compare them
	try {
		final Tuple2<Integer, String> readRec = new Tuple2<>();
		for (int i = 0; i < NUM_PAIRS_SHORT + 1; i++) {
			generator.next(rec);
			serializer.deserialize(readRec, inView);
			final int k1 = rec.f0;
			final String v1 = rec.f1;
			final int k2 = readRec.f0;
			final String v2 = readRec.f1;
			Assert.assertTrue("The re-generated and the read record do not match.", k1 == k2 && v1.equals(v2));
		}
		Assert.fail("Expected an EOFException which did not occur.");
	}
	catch (EOFException eofex) {
		// expected
	}
	catch (Throwable t) {
		// unexpected
		Assert.fail("Unexpected Exception: " + t.getMessage());
	}
	
	this.memoryManager.release(inView.close());
	reader.deleteChannel();
}
 
Example 16
@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 17
@Test
public void testWriteAndRead() throws Exception {
	final int numSegments = MEMORY_SIZE / MEMORY_PAGE_SIZE;
	final List<MemorySegment> memory = this.memoryManager.allocatePages(new DummyInvokable(), numSegments);
	
	NormalizedKeySorter<Tuple2<Integer, String>> sorter = newSortBuffer(memory);
	TestData.TupleGenerator generator = new TestData.TupleGenerator(SEED, KEY_MAX, VALUE_LENGTH, KeyMode.RANDOM,
		ValueMode.RANDOM_LENGTH);
	
	// write the records
	Tuple2<Integer, String> record = new Tuple2<>();
	int num = -1;
	do {
		generator.next(record);
		num++;
	}
	while (sorter.write(record));
	
	// re-read the records
	generator.reset();
	Tuple2<Integer, String> readTarget = new Tuple2<>();
	
	int i = 0;
	while (i < num) {
		generator.next(record);
		readTarget = sorter.getRecord(readTarget, i++);
		
		int rk = readTarget.f0;
		int gk = record.f0;
		
		String rv = readTarget.f1;
		String gv = record.f1;
		
		Assert.assertEquals("The re-read key is wrong", gk, rk);
		Assert.assertEquals("The re-read value is wrong", gv, rv);
	}
	
	// release the memory occupied by the buffers
	sorter.dispose();
	this.memoryManager.release(memory);
}
 
Example 18
@Test
public void testSortAndValidate() throws Exception
{
	final Hashtable<Integer, Integer> countTable = new Hashtable<>(KEY_MAX);
	for (int i = 1; i <= KEY_MAX; i++) {
		countTable.put(i, 0);
	}

	// comparator
	final TypeComparator<Integer> keyComparator = new IntComparator(true);

	// reader
	TestData.MockTuple2Reader<Tuple2<Integer, String>> reader = TestData.getIntStringTupleReader();

	// merge iterator
	LOG.debug("initializing sortmerger");
	
	TestCountCombiner2 comb = new TestCountCombiner2();
	
	Sorter<Tuple2<Integer, String>> merger = new CombiningUnilateralSortMerger<>(comb,
			this.memoryManager, this.ioManager, reader, this.parentTask, this.serializerFactory1, this.comparator1,
			0.25, 2, 0.7f, true /* use large record handler */, false);

	// emit data
	LOG.debug("emitting data");
	TestData.TupleGenerator generator = new TestData.TupleGenerator(SEED, KEY_MAX, VALUE_LENGTH, KeyMode.RANDOM, ValueMode.FIX_LENGTH);
	Tuple2<Integer, String> rec = new Tuple2<>();
	
	for (int i = 0; i < NUM_PAIRS; i++) {
		Assert.assertTrue((rec = generator.next(rec)) != null);
		final Integer key = rec.f0;
		rec.setField("1", 1);
		reader.emit(rec);
		
		countTable.put(key, countTable.get(key) + 1);
	}
	reader.close();

	// check order
	MutableObjectIterator<Tuple2<Integer, String>> iterator = merger.getIterator();
	
	LOG.debug("checking results");
	
	Tuple2<Integer, String> rec1 = new Tuple2<>();
	Tuple2<Integer, String> rec2 = new Tuple2<>();
	
	Assert.assertTrue((rec1 = iterator.next(rec1)) != null);
	countTable.put(rec1.f0, countTable.get(rec1.f0) - (Integer.parseInt(rec1.f1)));

	while ((rec2 = iterator.next(rec2)) != null) {
		int k1 = rec1.f0;
		int k2 = rec2.f0;
		
		Assert.assertTrue(keyComparator.compare(k1, k2) <= 0); 
		countTable.put(k2, countTable.get(k2) - (Integer.parseInt(rec2.f1)));
		
		rec1 = rec2;
	}

	for (Integer cnt : countTable.values()) {
		Assert.assertTrue(cnt == 0);
	}
	
	merger.close();
	
	// if the combiner was opened, it must have been closed
	Assert.assertTrue(comb.opened == comb.closed);
}
 
Example 19
Source Project: flink   File: ChannelViewsTest.java    License: Apache License 2.0 4 votes vote down vote up
@Test
public void testWriteReadNotAll() throws Exception
{
	final TestData.TupleGenerator generator = new TestData.TupleGenerator(SEED, KEY_MAX, VALUE_SHORT_LENGTH, KeyMode.RANDOM, ValueMode.RANDOM_LENGTH);
	final FileIOChannel.ID channel = this.ioManager.createChannel();
	final TypeSerializer<Tuple2<Integer, String>> serializer = TestData.getIntStringTupleSerializer();
	
	// create the writer output view
	List<MemorySegment> memory = this.memoryManager.allocatePages(this.parentTask, NUM_MEMORY_SEGMENTS);
	final BlockChannelWriter<MemorySegment> writer = this.ioManager.createBlockChannelWriter(channel);
	final ChannelWriterOutputView outView = new ChannelWriterOutputView(writer, memory, MEMORY_PAGE_SIZE);
	
	// write a number of pairs
	final Tuple2<Integer, String> rec = new Tuple2<>();
	for (int i = 0; i < NUM_PAIRS_SHORT; i++) {
		generator.next(rec);
		serializer.serialize(rec, outView);
	}
	this.memoryManager.release(outView.close());
	
	// create the reader input view
	memory = this.memoryManager.allocatePages(this.parentTask, NUM_MEMORY_SEGMENTS);
	final BlockChannelReader<MemorySegment> reader = this.ioManager.createBlockChannelReader(channel);
	final ChannelReaderInputView inView = new ChannelReaderInputView(reader, memory, outView.getBlockCount(), true);
	generator.reset();
	
	// read and re-generate all records and compare them
	final Tuple2<Integer, String> readRec = new Tuple2<>();
	for (int i = 0; i < NUM_PAIRS_SHORT / 2; i++) {
		generator.next(rec);
		serializer.deserialize(readRec, inView);
		
		int k1 = rec.f0;
		String v1 = rec.f1;
		
		int k2 = readRec.f0;
		String v2 = readRec.f1;
		
		Assert.assertTrue("The re-generated and the read record do not match.", k1 == k2 && v1.equals(v2));
	}
	
	this.memoryManager.release(inView.close());
	reader.deleteChannel();
}
 
Example 20
Source Project: flink   File: NormalizedKeySorterTest.java    License: Apache License 2.0 4 votes vote down vote up
@Test
public void testSortLongStringKeys() throws Exception {
	final int numSegments = MEMORY_SIZE / MEMORY_PAGE_SIZE;
	final List<MemorySegment> memory = this.memoryManager.allocatePages(new DummyInvokable(), numSegments);
	
	@SuppressWarnings("unchecked")
	TypeComparator<Tuple2<Integer, String>> accessors = TestData.getIntStringTupleTypeInfo().createComparator(new int[]{1}, new boolean[]{true}, 0, null);
	NormalizedKeySorter<Tuple2<Integer, String>> sorter = new NormalizedKeySorter<>(TestData.getIntStringTupleSerializer(), accessors, memory);
	
	TestData.TupleGenerator generator = new TestData.TupleGenerator(SEED, KEY_MAX, VALUE_LENGTH, KeyMode.RANDOM,
		ValueMode.FIX_LENGTH);
	
	// write the records
	Tuple2<Integer, String> record = new Tuple2<>();
	do {
		generator.next(record);
	}
	while (sorter.write(record));
	
	QuickSort qs = new QuickSort();
	qs.sort(sorter);
	
	MutableObjectIterator<Tuple2<Integer, String>> iter = sorter.getIterator();
	Tuple2<Integer, String> readTarget = new Tuple2<>();
	
	iter.next(readTarget);
	String last = readTarget.f1;
	
	while ((readTarget = iter.next(readTarget)) != null) {
		String current = readTarget.f1;
		
		final int cmp = last.compareTo(current);
		if (cmp > 0) {
			Assert.fail("Next value is not larger or equal to previous value.");
		}
		
		last = current;
	}
	
	// release the memory occupied by the buffers
	sorter.dispose();
	this.memoryManager.release(memory);
}