Java Code Examples for org.apache.hadoop.hive.ql.udf.generic.GenericUDAFEvaluator#terminate()

@Test
public void testComplete() throws HiveException {
    Funnel udaf = new Funnel();

    ObjectInspector[] inputObjectInspectorList = new ObjectInspector[]{
        PrimitiveObjectInspectorFactory.javaStringObjectInspector, // action_column
        PrimitiveObjectInspectorFactory.javaLongObjectInspector,   // timestamp_column
        ObjectInspectorFactory.getStandardListObjectInspector(PrimitiveObjectInspectorFactory.javaStringObjectInspector), // funnel_step_1
        ObjectInspectorFactory.getStandardListObjectInspector(PrimitiveObjectInspectorFactory.javaStringObjectInspector) // funnel_step_1
    };

    GenericUDAFParameterInfo paramInfo = new SimpleGenericUDAFParameterInfo(inputObjectInspectorList, false, false);
    GenericUDAFEvaluator udafEvaluator = udaf.getEvaluator(paramInfo);

    ObjectInspector outputObjectInspector = udafEvaluator.init(Mode.COMPLETE, inputObjectInspectorList);

    // Order will be "alpha, beta, gamma, delta" when ordered on timestamp_column
    // Funnel is "beta" -> "gamma" -> "epsilon"
    // Should return [1, 1, 0] as we don't have an epsilon
    Object[] parameters1 = new Object[]{ "beta", 200L, new ArrayList<Object>(), Arrays.asList("beta", "BAD"), null, "gamma", Arrays.asList("epsilon")}; // Test empty list funnel step, and null in funnel step
    Object[] parameters2 = new Object[]{"alpha", 100L,   Arrays.asList("beta", "BAD"), "gamma", Arrays.asList("epsilon")};
    Object[] parameters3 = new Object[]{"delta", 400L,   Arrays.asList("beta", "BAD"), "gamma", Arrays.asList("epsilon")};
    Object[] parameters4 = new Object[]{"gamma", 200L,   Arrays.asList("beta", "BAD"), "gamma", Arrays.asList("epsilon")}; // gamma and beta happen at the same time, beta should come first (sorted on action after timestamp)
    Object[] parameters5 = new Object[]{   null, 800L,   Arrays.asList("beta", "BAD"), "gamma", Arrays.asList("epsilon")}; // Check null action_column
    Object[] parameters6 = new Object[]{"omega", null,   Arrays.asList("beta", "BAD"), "gamma", Arrays.asList("epsilon")}; // Check null timestamp

    // Process the data
    AggregationBuffer agg = udafEvaluator.getNewAggregationBuffer();
    udafEvaluator.reset(agg);
    udafEvaluator.iterate(agg, parameters1);
    udafEvaluator.iterate(agg, parameters2);
    udafEvaluator.iterate(agg, parameters3);
    udafEvaluator.iterate(agg, parameters4);
    udafEvaluator.iterate(agg, parameters5);
    udafEvaluator.iterate(agg, parameters6);
    Object result = udafEvaluator.terminate(agg);

    // Expected
    List<Long> expected = new ArrayList<>();
    expected.add(1L);
    expected.add(1L);
    expected.add(0L);

    Assert.assertEquals(expected, result);
}
 

@Test
public void testFinal() throws HiveException {
    Funnel udaf = new Funnel();

    // Construct the object inspector for udaf evaluator
    ObjectInspector[] inputObjectInspectorList = new ObjectInspector[]{
        PrimitiveObjectInspectorFactory.javaStringObjectInspector, // action_column
        PrimitiveObjectInspectorFactory.javaLongObjectInspector,   // timestamp_column
        ObjectInspectorFactory.getStandardListObjectInspector(PrimitiveObjectInspectorFactory.javaStringObjectInspector), // funnel_step_1
        ObjectInspectorFactory.getStandardListObjectInspector(PrimitiveObjectInspectorFactory.javaStringObjectInspector) // funnel_step_1
    };
    GenericUDAFParameterInfo paramInfo = new SimpleGenericUDAFParameterInfo(inputObjectInspectorList, false, false);
    GenericUDAFEvaluator udafEvaluator = udaf.getEvaluator(paramInfo);

    // Construct the struct object inspector
    List<String> fieldNames = new ArrayList<>();
    fieldNames.add("action");
    fieldNames.add("timestamp");
    fieldNames.add("funnel");
    List<ObjectInspector> fieldInspectors = new ArrayList<>();
    fieldInspectors.add(ObjectInspectorFactory.getStandardListObjectInspector(PrimitiveObjectInspectorFactory.javaStringObjectInspector));
    fieldInspectors.add(ObjectInspectorFactory.getStandardListObjectInspector(PrimitiveObjectInspectorFactory.javaLongObjectInspector));
    fieldInspectors.add(ObjectInspectorFactory.getStandardListObjectInspector(PrimitiveObjectInspectorFactory.javaStringObjectInspector));
    ObjectInspector structObjectInspector = ObjectInspectorFactory.getStandardStructObjectInspector(fieldNames, fieldInspectors);
    ObjectInspector[] evaluatorInputObjectInspectorList = new ObjectInspector[]{structObjectInspector};

    ObjectInspector outputObjectInspector = udafEvaluator.init(Mode.FINAL, evaluatorInputObjectInspectorList);

    // Create the two structs to merge
    List<Object> parameter1 = new ArrayList<>();
    parameter1.add(Arrays.asList("beta"));
    parameter1.add(Arrays.asList(300L));
    parameter1.add(Arrays.asList("alpha", null, "beta", null, "gamma", null, "epsilon", null));

    List<Object> parameter2 = new ArrayList<>();
    parameter2.add(Arrays.asList("gamma", "alpha"));
    parameter2.add(Arrays.asList(400L, 200L));
    parameter2.add(Arrays.asList("alpha", null, "beta", null, "gamma", null, "epsilon", null));

    // Process the data
    AggregationBuffer agg = udafEvaluator.getNewAggregationBuffer();
    udafEvaluator.reset(agg);
    udafEvaluator.merge(agg, parameter1);
    udafEvaluator.merge(agg, parameter2);
    Object result = udafEvaluator.terminate(agg);

    // Expected
    List<Long> expected = Arrays.asList(1L, 1L, 1L, 0L);

    Assert.assertEquals(expected, result);
}
 

@Test
public void testComplete() throws HiveException {
    Merge udaf = new Merge();
    ObjectInspector[] inputObjectInspectorList = new ObjectInspector[]{
            ObjectInspectorFactory.getStandardListObjectInspector(PrimitiveObjectInspectorFactory.javaLongObjectInspector)
    };

    GenericUDAFParameterInfo paramInfo = new SimpleGenericUDAFParameterInfo(inputObjectInspectorList, false, false);
    GenericUDAFEvaluator udafEvaluator = udaf.getEvaluator(paramInfo);

    ObjectInspector outputObjectInspector = udafEvaluator.init(Mode.COMPLETE, inputObjectInspectorList);

    List<Long> funnel1 = new ArrayList<>();
    funnel1.add(1L);
    funnel1.add(1L);
    funnel1.add(0L);

    List<Long> funnel2 = new ArrayList<>();
    funnel2.add(1L);
    funnel2.add(0L);
    funnel2.add(0L);

    Object[] parameters1 = new Object[]{null};
    Object[] parameters2 = new Object[]{funnel1};
    Object[] parameters3 = new Object[]{funnel2};

    AggregationBuffer agg = udafEvaluator.getNewAggregationBuffer();
    udafEvaluator.reset(agg);
    udafEvaluator.iterate(agg, parameters1);
    udafEvaluator.iterate(agg, parameters2);
    udafEvaluator.iterate(agg, parameters3);
    Object result = udafEvaluator.terminate(agg);

    // Expected
    List<Long> expected = new ArrayList<>();
    expected.add(2L);
    expected.add(1L);
    expected.add(0L);

    Assert.assertEquals(expected, result);
}