Examples with StreamingAggregate org.apache.drill.exec.physical.config.StreamingAggregate used on opensource projects

Example 16 with StreamingAggregate

use of org.apache.drill.exec.physical.config.StreamingAggregate in project drill by axbaretto.

the class BasicPhysicalOpUnitTest method testSimpleStreamAgg.

@Test
public void testSimpleStreamAgg() {
    StreamingAggregate aggConf = new StreamingAggregate(null, parseExprs("a", "a"), parseExprs("sum(b)", "b_sum"), 1.0f);
    List<String> inputJsonBatches = Lists.newArrayList("[{\"a\": 5, \"b\" : 1 }]", "[{\"a\": 5, \"b\" : 5},{\"a\": 3, \"b\" : 8}]");
    opTestBuilder().physicalOperator(aggConf).inputDataStreamJson(inputJsonBatches).baselineColumns("b_sum", "a").baselineValues(6l, 5l).baselineValues(8l, 3l).go();
}

Example 17 with StreamingAggregate

use of org.apache.drill.exec.physical.config.StreamingAggregate in project drill by apache.

the class TestStreamingAggEmitOutcome method t12_testStreamingAggrNonEmptyBatchEmitOutcome.

/**
 *   Repeats t2_testStreamingAggrNonEmptyBatchEmitOutcome with no group by
 */
@Test
public void t12_testStreamingAggrNonEmptyBatchEmitOutcome() {
    final RowSet.SingleRowSet nonEmptyInputRowSet2 = operatorFixture.rowSetBuilder(inputSchema).addRow(13, 130, "item13").addRow(13, 130, "item13").addRow(2, 20, "item2").addRow(2, 20, "item2").addRow(4, 40, "item4").build();
    final RowSet.SingleRowSet expectedRowSet = operatorFixture.rowSetBuilder(resultSchemaNoGroupBy).addRow((long) 385).build();
    inputContainer.add(emptyInputRowSet.container());
    inputContainer.add(nonEmptyInputRowSet.container());
    inputContainer.add(nonEmptyInputRowSet2.container());
    inputOutcomes.add(RecordBatch.IterOutcome.OK_NEW_SCHEMA);
    inputOutcomes.add(RecordBatch.IterOutcome.OK_NEW_SCHEMA);
    inputOutcomes.add(RecordBatch.IterOutcome.EMIT);
    final MockRecordBatch mockInputBatch = new MockRecordBatch(operatorFixture.getFragmentContext(), opContext, inputContainer, inputOutcomes, emptyInputRowSet.container().getSchema());
    final StreamingAggregate streamAggrConfig = new StreamingAggregate(null, new ArrayList<NamedExpression>(), parseExprs("sum(id_left+cost_left)", "total_sum"));
    final StreamingAggBatch strAggBatch = new StreamingAggBatch(streamAggrConfig, mockInputBatch, operatorFixture.getFragmentContext());
    assertTrue(strAggBatch.next() == RecordBatch.IterOutcome.OK_NEW_SCHEMA);
    // Data before EMIT is returned with an OK_NEW_SCHEMA.
    assertTrue(strAggBatch.next() == RecordBatch.IterOutcome.OK_NEW_SCHEMA);
    assertEquals(1, strAggBatch.getRecordCount());
    RowSet actualRowSet = DirectRowSet.fromContainer(strAggBatch.getContainer());
    new RowSetComparison(expectedRowSet).verify(actualRowSet);
    // EMIT comes with an empty batch
    assertTrue(strAggBatch.next() == RecordBatch.IterOutcome.EMIT);
    assertTrue(strAggBatch.next() == RecordBatch.IterOutcome.NONE);
    // Release memory for row sets
    nonEmptyInputRowSet2.clear();
    expectedRowSet.clear();
}

Example 18 with StreamingAggregate

use of org.apache.drill.exec.physical.config.StreamingAggregate in project drill by apache.

the class TestStreamingAggEmitOutcome method t7_testStreamingAggrMultipleEMITOutcome.

/**
 * Normal case
 */
@Test
public void t7_testStreamingAggrMultipleEMITOutcome() {
    final RowSet.SingleRowSet nonEmptyInputRowSet2 = operatorFixture.rowSetBuilder(inputSchema).addRow(2, 20, "item2").addRow(3, 30, "item3").build();
    inputContainer.add(emptyInputRowSet.container());
    inputContainer.add(nonEmptyInputRowSet.container());
    inputContainer.add(emptyInputRowSet.container());
    inputContainer.add(nonEmptyInputRowSet2.container());
    inputContainer.add(emptyInputRowSet.container());
    inputOutcomes.add(RecordBatch.IterOutcome.OK_NEW_SCHEMA);
    inputOutcomes.add(RecordBatch.IterOutcome.OK_NEW_SCHEMA);
    inputOutcomes.add(RecordBatch.IterOutcome.EMIT);
    inputOutcomes.add(RecordBatch.IterOutcome.EMIT);
    inputOutcomes.add(RecordBatch.IterOutcome.EMIT);
    final MockRecordBatch mockInputBatch = new MockRecordBatch(operatorFixture.getFragmentContext(), opContext, inputContainer, inputOutcomes, emptyInputRowSet.container().getSchema());
    final StreamingAggregate streamAggrConfig = new StreamingAggregate(null, parseExprs("name_left", "name"), parseExprs("sum(id_left+cost_left)", "total_sum"));
    final StreamingAggBatch strAggBatch = new StreamingAggBatch(streamAggrConfig, mockInputBatch, operatorFixture.getFragmentContext());
    assertTrue(strAggBatch.next() == RecordBatch.IterOutcome.OK_NEW_SCHEMA);
    assertTrue(strAggBatch.next() == RecordBatch.IterOutcome.OK_NEW_SCHEMA);
    assertEquals(1, strAggBatch.getRecordCount());
    assertTrue(strAggBatch.next() == RecordBatch.IterOutcome.EMIT);
    assertEquals(0, strAggBatch.getRecordCount());
    assertTrue(strAggBatch.next() == RecordBatch.IterOutcome.EMIT);
    assertEquals(2, strAggBatch.getRecordCount());
    assertTrue(strAggBatch.next() == RecordBatch.IterOutcome.EMIT);
    assertEquals(0, strAggBatch.getRecordCount());
    assertTrue(strAggBatch.next() == RecordBatch.IterOutcome.NONE);
    nonEmptyInputRowSet2.clear();
}

Example 19 with StreamingAggregate

use of org.apache.drill.exec.physical.config.StreamingAggregate in project drill by apache.

the class TestStreamingAggEmitOutcome method t19_testStreamingAgr_WithEmptyNonEmptyBatchesAndOKOutcome.

/**
 *   Repeats t9_testStreamingAgr_WithEmptyNonEmptyBatchesAndOKOutcome with no group by
 */
@Test
public void t19_testStreamingAgr_WithEmptyNonEmptyBatchesAndOKOutcome() {
    final RowSet.SingleRowSet nonEmptyInputRowSet2 = operatorFixture.rowSetBuilder(inputSchema).addRow(2, 20, "item1").addRow(13, 130, "item13").addRow(13, 130, "item13").addRow(13, 130, "item13").addRow(130, 1300, "item130").addRow(0, 0, "item130").build();
    final RowSet.SingleRowSet nonEmptyInputRowSet3 = operatorFixture.rowSetBuilder(inputSchema).addRow(23, 230, "item23").addRow(3, 33, "item3").addRow(7, 70, "item7").addRow(17, 170, "item7").build();
    final RowSet.SingleRowSet expectedRowSet = operatorFixture.rowSetBuilder(resultSchemaNoGroupBy).addRow((long) 2445).build();
    inputContainer.add(emptyInputRowSet.container());
    inputContainer.add(nonEmptyInputRowSet.container());
    inputContainer.add(emptyInputRowSet.container());
    inputContainer.add(nonEmptyInputRowSet2.container());
    inputContainer.add(emptyInputRowSet.container());
    inputContainer.add(nonEmptyInputRowSet3.container());
    inputContainer.add(emptyInputRowSet.container());
    inputOutcomes.add(RecordBatch.IterOutcome.OK_NEW_SCHEMA);
    inputOutcomes.add(RecordBatch.IterOutcome.OK_NEW_SCHEMA);
    inputOutcomes.add(RecordBatch.IterOutcome.OK);
    inputOutcomes.add(RecordBatch.IterOutcome.OK);
    inputOutcomes.add(RecordBatch.IterOutcome.OK);
    inputOutcomes.add(RecordBatch.IterOutcome.OK);
    inputOutcomes.add(RecordBatch.IterOutcome.OK);
    final MockRecordBatch mockInputBatch = new MockRecordBatch(operatorFixture.getFragmentContext(), opContext, inputContainer, inputOutcomes, emptyInputRowSet.container().getSchema());
    final StreamingAggregate streamAggrConfig = new StreamingAggregate(null, new ArrayList<>(), parseExprs("sum(id_left+cost_left)", "total_sum"));
    final StreamingAggBatch strAggBatch = new StreamingAggBatch(streamAggrConfig, mockInputBatch, operatorFixture.getFragmentContext());
    assertTrue(strAggBatch.next() == RecordBatch.IterOutcome.OK_NEW_SCHEMA);
    assertTrue(strAggBatch.next() == RecordBatch.IterOutcome.OK_NEW_SCHEMA);
    assertEquals(1, strAggBatch.getRecordCount());
    RowSet actualRowSet = DirectRowSet.fromContainer(strAggBatch.getContainer());
    new RowSetComparison(expectedRowSet).verify(actualRowSet);
    assertTrue(strAggBatch.next() == RecordBatch.IterOutcome.NONE);
    nonEmptyInputRowSet2.clear();
    nonEmptyInputRowSet3.clear();
    expectedRowSet.clear();
}

Example 20 with StreamingAggregate

use of org.apache.drill.exec.physical.config.StreamingAggregate in project drill by apache.

the class TestStreamingAggEmitOutcome method t6_testStreamingAggrOkFollowedByNone.

/**
 * Verifies that if StreamingAggr receives multiple non-empty record batch with EMIT outcome in between then it produces
 * output for those input batch correctly. In this case it receives first non-empty batch with OK_NEW_SCHEMA in
 * buildSchema phase followed by an empty batch with EMIT outcome. For this combination it produces output for the
 * record received so far along with EMIT outcome. Then it receives second non-empty batch with OK outcome and
 * produces output for it differently. The test validates that for each output received the order of the records are
 * correct.
 * @throws Exception
 */
@Test
public void t6_testStreamingAggrOkFollowedByNone() {
    final RowSet.SingleRowSet nonEmptyInputRowSet2 = operatorFixture.rowSetBuilder(inputSchema).addRow(2, 20, "item2").addRow(3, 30, "item3").addRow(4, 40, "item4").addRow(4, 40, "item4").addRow(5, 50, "item5").addRow(5, 50, "item5").build();
    final RowSet.SingleRowSet expectedRowSet1 = operatorFixture.rowSetBuilder(resultSchema).addRow("item1", (long) 11).build();
    final RowSet.SingleRowSet expectedRowSet2 = operatorFixture.rowSetBuilder(resultSchema).addRow("item2", (long) 22).addRow("item3", (long) 33).addRow("item4", (long) 88).addRow("item5", (long) 110).build();
    inputContainer.add(emptyInputRowSet.container());
    inputContainer.add(nonEmptyInputRowSet.container());
    inputContainer.add(emptyInputRowSet.container());
    inputContainer.add(nonEmptyInputRowSet2.container());
    inputOutcomes.add(RecordBatch.IterOutcome.OK_NEW_SCHEMA);
    inputOutcomes.add(RecordBatch.IterOutcome.OK_NEW_SCHEMA);
    inputOutcomes.add(RecordBatch.IterOutcome.EMIT);
    inputOutcomes.add(RecordBatch.IterOutcome.OK);
    final MockRecordBatch mockInputBatch = new MockRecordBatch(operatorFixture.getFragmentContext(), opContext, inputContainer, inputOutcomes, emptyInputRowSet.container().getSchema());
    final StreamingAggregate streamAggrConfig = new StreamingAggregate(null, parseExprs("name_left", "name"), parseExprs("sum(id_left+cost_left)", "total_sum"));
    final StreamingAggBatch strAggBatch = new StreamingAggBatch(streamAggrConfig, mockInputBatch, operatorFixture.getFragmentContext());
    assertTrue(strAggBatch.next() == RecordBatch.IterOutcome.OK_NEW_SCHEMA);
    assertTrue(strAggBatch.next() == RecordBatch.IterOutcome.OK_NEW_SCHEMA);
    assertEquals(1, strAggBatch.getRecordCount());
    RowSet actualRowSet = DirectRowSet.fromContainer(strAggBatch.getContainer());
    new RowSetComparison(expectedRowSet1).verify(actualRowSet);
    assertTrue(strAggBatch.next() == RecordBatch.IterOutcome.EMIT);
    assertEquals(0, strAggBatch.getRecordCount());
    assertTrue(strAggBatch.next() == RecordBatch.IterOutcome.OK);
    assertEquals(4, strAggBatch.getRecordCount());
    actualRowSet = DirectRowSet.fromContainer(strAggBatch.getContainer());
    new RowSetComparison(expectedRowSet2).verify(actualRowSet);
    assertTrue(strAggBatch.next() == RecordBatch.IterOutcome.NONE);
    // Release memory for row sets
    nonEmptyInputRowSet2.clear();
    expectedRowSet2.clear();
    expectedRowSet1.clear();
}