Examples with SampleElementsForSplitPoints uk.gov.gchq.gaffer.operation.impl.SampleElementsForSplitPoints used on opensource projects

Example 6 with SampleElementsForSplitPoints

use of uk.gov.gchq.gaffer.operation.impl.SampleElementsForSplitPoints in project Gaffer by gchq.

the class AbstractSampleElementsForSplitPointsHandlerTest method shouldSampleApproximatelyHalfOfElements.

@Test
public void shouldSampleApproximatelyHalfOfElements() throws OperationException {
    // Given
    final int numSplits = 3;
    final List<Element> elements = IntStream.range(0, 1000 * numSplits).mapToObj(i -> new Entity(TestGroups.ENTITY, "vertex_" + i)).collect(Collectors.toList());
    final AbstractSampleElementsForSplitPointsHandler<?, S> handler = createHandler();
    final SampleElementsForSplitPoints operation = new SampleElementsForSplitPoints.Builder<>().input(elements).numSplits(numSplits).proportionToSample(0.5f).build();
    final S store = createStore();
    // When
    handler.doOperation(operation, new Context(), store);
    // Then
    final ArgumentCaptor<GenerateSplitPointsFromSample> generateSplitPointsFromSampleCaptor = ArgumentCaptor.forClass(GenerateSplitPointsFromSample.class);
    verify(store).execute(generateSplitPointsFromSampleCaptor.capture(), any(Context.class));
    final int maximumExpectedSampleSize = (int) ((elements.size() / 2) * 1.1);
    assertExpectedNumberOfSplitPointsAndSampleSizeOfNoMoreThan(generateSplitPointsFromSampleCaptor, numSplits, maximumExpectedSampleSize);
}

Example 7 with SampleElementsForSplitPoints

use of uk.gov.gchq.gaffer.operation.impl.SampleElementsForSplitPoints in project Gaffer by gchq.

the class AbstractSampleElementsForSplitPointsHandlerTest method shouldThrowExceptionIfNumberOfSampledElementsIsMoreThanMaxAllowed.

@Test
public void shouldThrowExceptionIfNumberOfSampledElementsIsMoreThanMaxAllowed() throws OperationException {
    // Given
    int maxSampledElements = 5;
    final AbstractSampleElementsForSplitPointsHandler<?, ?> handler = createHandler();
    handler.setMaxSampledElements(maxSampledElements);
    final List<Element> elements = IntStream.range(0, 6).mapToObj(i -> new Entity(TestGroups.ENTITY, "vertex_" + i)).collect(Collectors.toList());
    final SampleElementsForSplitPoints operation = new SampleElementsForSplitPoints.Builder<>().input(elements).numSplits(3).build();
    // When / Then
    try {
        handler.doOperation(operation, new Context(), createStore());
        fail("Exception expected");
    } catch (final LimitExceededException e) {
        assertTrue(e.getMessage().equals("Limit of " + maxSampledElements + " exceeded."), e.getMessage());
    }
}

Example 8 with SampleElementsForSplitPoints

use of uk.gov.gchq.gaffer.operation.impl.SampleElementsForSplitPoints in project Gaffer by gchq.

the class SampleElementsForSplitPointsHandlerTest method shouldUseTheNumberOfTabletServersToCalculateNumSplits.

@Test
public void shouldUseTheNumberOfTabletServersToCalculateNumSplits() throws OperationException {
    // Given
    final Integer numSplits = null;
    final List<Element> elements = IntStream.range(0, 30).mapToObj(i -> new Edge(TestGroups.EDGE, "source_" + i, "dest_" + i, true)).collect(Collectors.toList());
    final AbstractSampleElementsForSplitPointsHandler<String, AccumuloStore> handler = createHandler();
    final SampleElementsForSplitPoints<String> operation = new SampleElementsForSplitPoints.Builder<String>().input(elements).numSplits(numSplits).build();
    // When
    createHandler().doOperation(operation, new Context(), store);
    // Then
    final ArgumentCaptor<GenerateSplitPointsFromSample> generateSplitPointsFromSampleCaptor = ArgumentCaptor.forClass(GenerateSplitPointsFromSample.class);
    verify(store).execute(generateSplitPointsFromSampleCaptor.capture(), any(Context.class));
    final int expectedNumOfSplits = NUM_TABLET_SERVERS - 1;
    final int expectedElementCount = elements.size() * 2;
    assertExpectedNumberOfSplitPointsAndSampleSize(generateSplitPointsFromSampleCaptor, expectedNumOfSplits, expectedElementCount);
}

Example 9 with SampleElementsForSplitPoints

use of uk.gov.gchq.gaffer.operation.impl.SampleElementsForSplitPoints in project Gaffer by gchq.

the class SampleElementsForSplitPointsHandlerTest method shouldUseTheNumberOfRegionsToCalculateNumSplits.

@Test
public void shouldUseTheNumberOfRegionsToCalculateNumSplits() throws OperationException {
    // Given
    final Integer numSplits = null;
    final List<Element> elements = IntStream.range(0, 30).mapToObj(i -> new Edge(TestGroups.EDGE, "source_" + i, "dest_" + i, true)).collect(Collectors.toList());
    final AbstractSampleElementsForSplitPointsHandler<String, HBaseStore> handler = createHandler();
    final SampleElementsForSplitPoints<String> operation = new SampleElementsForSplitPoints.Builder<String>().input(elements).numSplits(numSplits).build();
    // When
    createHandler().doOperation(operation, new Context(), store);
    // Then
    final ArgumentCaptor<GenerateSplitPointsFromSample> generateSplitPointsFromSampleCaptor = ArgumentCaptor.forClass(GenerateSplitPointsFromSample.class);
    verify(store).execute(generateSplitPointsFromSampleCaptor.capture(), any(Context.class));
    final int expectedNumOfSplits = NUM_TABLE_REGIONS - 1;
    final int expectedElementCount = elements.size() * 2;
    assertExpectedNumberOfSplitPointsAndSampleSize(generateSplitPointsFromSampleCaptor, expectedNumOfSplits, expectedElementCount);
}

Example 10 with SampleElementsForSplitPoints

use of uk.gov.gchq.gaffer.operation.impl.SampleElementsForSplitPoints in project Gaffer by gchq.

the class SampleElementsForSplitPointsHandlerTest method shouldCalculateRequiredNumberOfSplitsFromEdges.

@Test
public void shouldCalculateRequiredNumberOfSplitsFromEdges() throws OperationException {
    // Given
    final int numSplits = 3;
    final List<Element> elements = IntStream.range(0, numSplits * 10).mapToObj(i -> new Edge(TestGroups.EDGE, "source_" + i, "dest_" + i, true)).collect(Collectors.toList());
    final AbstractSampleElementsForSplitPointsHandler<String, HBaseStore> handler = createHandler();
    final SampleElementsForSplitPoints<String> operation = new SampleElementsForSplitPoints.Builder<String>().input(elements).numSplits(numSplits).build();
    // When
    createHandler().doOperation(operation, new Context(), createStore());
    // Then
    final ArgumentCaptor<GenerateSplitPointsFromSample> generateSplitPointsFromSampleCaptor = ArgumentCaptor.forClass(GenerateSplitPointsFromSample.class);
    verify(store).execute(generateSplitPointsFromSampleCaptor.capture(), any(Context.class));
    final int expectedElementCount = elements.size() * 2;
    assertExpectedNumberOfSplitPointsAndSampleSize(generateSplitPointsFromSampleCaptor, numSplits, expectedElementCount);
}