Search in sources :

Example 11 with DimensionsSpec

use of org.apache.druid.data.input.impl.DimensionsSpec in project druid by druid-io.

the class IndexMergerTestBase method testMultivalDim_mergeAcrossSegments_rollupWorks.

@Test
public void testMultivalDim_mergeAcrossSegments_rollupWorks() throws Exception {
    List<String> dims = Arrays.asList("dimA", "dimMultiVal");
    IncrementalIndexSchema indexSchema = new IncrementalIndexSchema.Builder().withDimensionsSpec(new DimensionsSpec(ImmutableList.of(new StringDimensionSchema("dimA", MultiValueHandling.SORTED_ARRAY, true), new StringDimensionSchema("dimMultiVal", MultiValueHandling.SORTED_ARRAY, true)))).withMetrics(new LongSumAggregatorFactory("sumCount", "sumCount")).withRollup(true).build();
    IncrementalIndex toPersistA = new OnheapIncrementalIndex.Builder().setIndexSchema(indexSchema).setMaxRowCount(1000).build();
    Map<String, Object> event1 = new HashMap<>();
    event1.put("dimA", "leek");
    event1.put("dimMultiVal", ImmutableList.of("1", "2", "4"));
    event1.put("sumCount", 1L);
    Map<String, Object> event2 = new HashMap<>();
    event2.put("dimA", "leek");
    event2.put("dimMultiVal", ImmutableList.of("1", "2", "3", "5"));
    event2.put("sumCount", 1L);
    toPersistA.add(new MapBasedInputRow(1, dims, event1));
    toPersistA.add(new MapBasedInputRow(1, dims, event2));
    IncrementalIndex toPersistB = new OnheapIncrementalIndex.Builder().setIndexSchema(indexSchema).setMaxRowCount(1000).build();
    Map<String, Object> event3 = new HashMap<>();
    event3.put("dimA", "leek");
    event3.put("dimMultiVal", ImmutableList.of("1", "2", "4"));
    event3.put("sumCount", 1L);
    Map<String, Object> event4 = new HashMap<>();
    event4.put("dimA", "potato");
    event4.put("dimMultiVal", ImmutableList.of("0", "1", "4"));
    event4.put("sumCount", 1L);
    toPersistB.add(new MapBasedInputRow(1, dims, event3));
    toPersistB.add(new MapBasedInputRow(1, dims, event4));
    final File tmpDirA = temporaryFolder.newFolder();
    final File tmpDirB = temporaryFolder.newFolder();
    final File tmpDirMerged = temporaryFolder.newFolder();
    QueryableIndex indexA = closer.closeLater(indexIO.loadIndex(indexMerger.persist(toPersistA, tmpDirA, indexSpec, null)));
    QueryableIndex indexB = closer.closeLater(indexIO.loadIndex(indexMerger.persist(toPersistB, tmpDirB, indexSpec, null)));
    final QueryableIndex merged = closer.closeLater(indexIO.loadIndex(indexMerger.mergeQueryableIndex(Arrays.asList(indexA, indexB), true, new AggregatorFactory[] { new LongSumAggregatorFactory("sumCount", "sumCount") }, tmpDirMerged, indexSpec, null, -1)));
    final QueryableIndexIndexableAdapter adapter = new QueryableIndexIndexableAdapter(merged);
    final List<DebugRow> rowList = RowIteratorHelper.toList(adapter.getRows());
    Assert.assertEquals(ImmutableList.of("dimA", "dimMultiVal"), ImmutableList.copyOf(adapter.getDimensionNames()));
    Assert.assertEquals(3, rowList.size());
    Assert.assertEquals(Arrays.asList("leek", Arrays.asList("1", "2", "3", "5")), rowList.get(0).dimensionValues());
    Assert.assertEquals(1L, rowList.get(0).metricValues().get(0));
    Assert.assertEquals(Arrays.asList("leek", Arrays.asList("1", "2", "4")), rowList.get(1).dimensionValues());
    Assert.assertEquals(2L, rowList.get(1).metricValues().get(0));
    Assert.assertEquals(Arrays.asList("potato", Arrays.asList("0", "1", "4")), rowList.get(2).dimensionValues());
    Assert.assertEquals(1L, rowList.get(2).metricValues().get(0));
    checkBitmapIndex(Arrays.asList(0, 1), adapter.getBitmapIndex("dimA", "leek"));
    checkBitmapIndex(Collections.singletonList(2), adapter.getBitmapIndex("dimA", "potato"));
    checkBitmapIndex(Collections.singletonList(2), adapter.getBitmapIndex("dimMultiVal", "0"));
    checkBitmapIndex(Arrays.asList(0, 1, 2), adapter.getBitmapIndex("dimMultiVal", "1"));
    checkBitmapIndex(Arrays.asList(0, 1), adapter.getBitmapIndex("dimMultiVal", "2"));
    checkBitmapIndex(Collections.singletonList(0), adapter.getBitmapIndex("dimMultiVal", "3"));
    checkBitmapIndex(Arrays.asList(1, 2), adapter.getBitmapIndex("dimMultiVal", "4"));
    checkBitmapIndex(Collections.singletonList(0), adapter.getBitmapIndex("dimMultiVal", "5"));
}
Also used : IncrementalIndex(org.apache.druid.segment.incremental.IncrementalIndex) OnheapIncrementalIndex(org.apache.druid.segment.incremental.OnheapIncrementalIndex) HashMap(java.util.HashMap) LongSumAggregatorFactory(org.apache.druid.query.aggregation.LongSumAggregatorFactory) OnheapIncrementalIndex(org.apache.druid.segment.incremental.OnheapIncrementalIndex) StringDimensionSchema(org.apache.druid.data.input.impl.StringDimensionSchema) DimensionsSpec(org.apache.druid.data.input.impl.DimensionsSpec) MapBasedInputRow(org.apache.druid.data.input.MapBasedInputRow) File(java.io.File) IncrementalIndexSchema(org.apache.druid.segment.incremental.IncrementalIndexSchema) IncrementalIndexTest(org.apache.druid.segment.data.IncrementalIndexTest) InitializedNullHandlingTest(org.apache.druid.testing.InitializedNullHandlingTest) Test(org.junit.Test)

Example 12 with DimensionsSpec

use of org.apache.druid.data.input.impl.DimensionsSpec in project druid by druid-io.

the class IndexMergerTestBase method persistAndLoad.

private QueryableIndex persistAndLoad(List<DimensionSchema> schema, InputRow... rows) throws IOException {
    IncrementalIndex toPersist = IncrementalIndexTest.createIndex(null, new DimensionsSpec(schema));
    for (InputRow row : rows) {
        toPersist.add(row);
    }
    final File tempDir = temporaryFolder.newFolder();
    return closer.closeLater(indexIO.loadIndex(indexMerger.persist(toPersist, tempDir, indexSpec, null)));
}
Also used : IncrementalIndex(org.apache.druid.segment.incremental.IncrementalIndex) OnheapIncrementalIndex(org.apache.druid.segment.incremental.OnheapIncrementalIndex) InputRow(org.apache.druid.data.input.InputRow) MapBasedInputRow(org.apache.druid.data.input.MapBasedInputRow) DimensionsSpec(org.apache.druid.data.input.impl.DimensionsSpec) File(java.io.File)

Example 13 with DimensionsSpec

use of org.apache.druid.data.input.impl.DimensionsSpec in project druid by druid-io.

the class IndexMergerTestBase method testMergeWithDimensionsList.

@Test
public void testMergeWithDimensionsList() throws Exception {
    IncrementalIndexSchema schema = new IncrementalIndexSchema.Builder().withDimensionsSpec(new DimensionsSpec(makeDimensionSchemas(Arrays.asList("dimA", "dimB", "dimC")))).withMetrics(new CountAggregatorFactory("count")).build();
    IncrementalIndex toPersist1 = new OnheapIncrementalIndex.Builder().setIndexSchema(schema).setMaxRowCount(1000).build();
    IncrementalIndex toPersist2 = new OnheapIncrementalIndex.Builder().setIndexSchema(schema).setMaxRowCount(1000).build();
    IncrementalIndex toPersist3 = new OnheapIncrementalIndex.Builder().setIndexSchema(schema).setMaxRowCount(1000).build();
    addDimValuesToIndex(toPersist1, "dimA", Arrays.asList("1", "2"));
    addDimValuesToIndex(toPersist2, "dimA", Arrays.asList("1", "2"));
    addDimValuesToIndex(toPersist3, "dimC", Arrays.asList("1", "2"));
    final File tmpDir = temporaryFolder.newFolder();
    final File tmpDir2 = temporaryFolder.newFolder();
    final File tmpDir3 = temporaryFolder.newFolder();
    final File tmpDirMerged = temporaryFolder.newFolder();
    QueryableIndex index1 = closer.closeLater(indexIO.loadIndex(indexMerger.persist(toPersist1, tmpDir, indexSpec, null)));
    QueryableIndex index2 = closer.closeLater(indexIO.loadIndex(indexMerger.persist(toPersist2, tmpDir2, indexSpec, null)));
    QueryableIndex index3 = closer.closeLater(indexIO.loadIndex(indexMerger.persist(toPersist3, tmpDir3, indexSpec, null)));
    final QueryableIndex merged = closer.closeLater(indexIO.loadIndex(indexMerger.mergeQueryableIndex(Arrays.asList(index1, index2, index3), true, new AggregatorFactory[] { new CountAggregatorFactory("count") }, tmpDirMerged, indexSpec, null, -1)));
    final QueryableIndexIndexableAdapter adapter = new QueryableIndexIndexableAdapter(merged);
    final List<DebugRow> rowList = RowIteratorHelper.toList(adapter.getRows());
    Assert.assertEquals(ImmutableList.of("dimA", "dimC"), ImmutableList.copyOf(adapter.getDimensionNames()));
    Assert.assertEquals(4, rowList.size());
    Assert.assertEquals(Arrays.asList(null, "1"), rowList.get(0).dimensionValues());
    Assert.assertEquals(Collections.singletonList(1L), rowList.get(0).metricValues());
    Assert.assertEquals(Arrays.asList(null, "2"), rowList.get(1).dimensionValues());
    Assert.assertEquals(Collections.singletonList(1L), rowList.get(1).metricValues());
    Assert.assertEquals(Arrays.asList("1", null), rowList.get(2).dimensionValues());
    Assert.assertEquals(Collections.singletonList(2L), rowList.get(2).metricValues());
    Assert.assertEquals(Arrays.asList("2", null), rowList.get(3).dimensionValues());
    Assert.assertEquals(Collections.singletonList(2L), rowList.get(3).metricValues());
    Assert.assertEquals(useBitmapIndexes, adapter.getCapabilities("dimA").hasBitmapIndexes());
    Assert.assertEquals(useBitmapIndexes, adapter.getCapabilities("dimC").hasBitmapIndexes());
    if (useBitmapIndexes) {
        checkBitmapIndex(Arrays.asList(0, 1), adapter.getBitmapIndex("dimA", null));
        checkBitmapIndex(Collections.singletonList(2), adapter.getBitmapIndex("dimA", "1"));
        checkBitmapIndex(Collections.singletonList(3), adapter.getBitmapIndex("dimA", "2"));
        checkBitmapIndex(Collections.emptyList(), adapter.getBitmapIndex("dimB", null));
        checkBitmapIndex(Arrays.asList(2, 3), adapter.getBitmapIndex("dimC", null));
        checkBitmapIndex(Collections.singletonList(0), adapter.getBitmapIndex("dimC", "1"));
        checkBitmapIndex(Collections.singletonList(1), adapter.getBitmapIndex("dimC", "2"));
    }
    checkBitmapIndex(Collections.emptyList(), adapter.getBitmapIndex("dimB", ""));
}
Also used : CountAggregatorFactory(org.apache.druid.query.aggregation.CountAggregatorFactory) IncrementalIndex(org.apache.druid.segment.incremental.IncrementalIndex) OnheapIncrementalIndex(org.apache.druid.segment.incremental.OnheapIncrementalIndex) OnheapIncrementalIndex(org.apache.druid.segment.incremental.OnheapIncrementalIndex) DimensionsSpec(org.apache.druid.data.input.impl.DimensionsSpec) File(java.io.File) IncrementalIndexSchema(org.apache.druid.segment.incremental.IncrementalIndexSchema) IncrementalIndexTest(org.apache.druid.segment.data.IncrementalIndexTest) InitializedNullHandlingTest(org.apache.druid.testing.InitializedNullHandlingTest) Test(org.junit.Test)

Example 14 with DimensionsSpec

use of org.apache.druid.data.input.impl.DimensionsSpec in project druid by druid-io.

the class QueryableIndexColumnCapabilitiesTest method setup.

@BeforeClass
public static void setup() throws IOException {
    MapInputRowParser parser = new MapInputRowParser(new TimeAndDimsParseSpec(new TimestampSpec("time", "auto", null), new DimensionsSpec(ImmutableList.<DimensionSchema>builder().addAll(DimensionsSpec.getDefaultSchemas(ImmutableList.of("d1", "d2"))).add(new DoubleDimensionSchema("d3")).add(new FloatDimensionSchema("d4")).add(new LongDimensionSchema("d5")).build())));
    AggregatorFactory[] metricsSpecs = new AggregatorFactory[] { new CountAggregatorFactory("cnt"), new DoubleSumAggregatorFactory("m1", "d3"), new FloatSumAggregatorFactory("m2", "d4"), new LongSumAggregatorFactory("m3", "d5"), new HyperUniquesAggregatorFactory("m4", "d1") };
    List<InputRow> rows = new ArrayList<>();
    Map<String, Object> event = ImmutableMap.<String, Object>builder().put("time", DateTimes.nowUtc().getMillis()).put("d1", "some string").put("d2", ImmutableList.of("some", "list")).put("d3", 1.234).put("d4", 1.234f).put("d5", 10L).build();
    rows.add(Iterables.getOnlyElement(parser.parseBatch(event)));
    IndexBuilder builder = IndexBuilder.create().rows(rows).schema(new IncrementalIndexSchema.Builder().withMetrics(metricsSpecs).withDimensionsSpec(parser).withRollup(false).build()).tmpDir(temporaryFolder.newFolder());
    INC_INDEX = builder.buildIncrementalIndex();
    MMAP_INDEX = builder.buildMMappedIndex();
    List<InputRow> rowsWithNulls = new ArrayList<>();
    rowsWithNulls.add(Iterables.getOnlyElement(parser.parseBatch(event)));
    Map<String, Object> eventWithNulls = new HashMap<>();
    eventWithNulls.put("time", DateTimes.nowUtc().getMillis());
    eventWithNulls.put("d1", null);
    eventWithNulls.put("d2", ImmutableList.of());
    eventWithNulls.put("d3", null);
    eventWithNulls.put("d4", null);
    eventWithNulls.put("d5", null);
    rowsWithNulls.add(Iterables.getOnlyElement(parser.parseBatch(eventWithNulls)));
    IndexBuilder builderWithNulls = IndexBuilder.create().rows(rowsWithNulls).schema(new IncrementalIndexSchema.Builder().withMetrics(metricsSpecs).withDimensionsSpec(parser).withRollup(false).build()).tmpDir(temporaryFolder.newFolder());
    INC_INDEX_WITH_NULLS = builderWithNulls.buildIncrementalIndex();
    MMAP_INDEX_WITH_NULLS = builderWithNulls.buildMMappedIndex();
}
Also used : MapInputRowParser(org.apache.druid.data.input.impl.MapInputRowParser) DoubleSumAggregatorFactory(org.apache.druid.query.aggregation.DoubleSumAggregatorFactory) HashMap(java.util.HashMap) LongDimensionSchema(org.apache.druid.data.input.impl.LongDimensionSchema) LongSumAggregatorFactory(org.apache.druid.query.aggregation.LongSumAggregatorFactory) ArrayList(java.util.ArrayList) FloatDimensionSchema(org.apache.druid.data.input.impl.FloatDimensionSchema) FloatSumAggregatorFactory(org.apache.druid.query.aggregation.FloatSumAggregatorFactory) DoubleSumAggregatorFactory(org.apache.druid.query.aggregation.DoubleSumAggregatorFactory) LongSumAggregatorFactory(org.apache.druid.query.aggregation.LongSumAggregatorFactory) CountAggregatorFactory(org.apache.druid.query.aggregation.CountAggregatorFactory) HyperUniquesAggregatorFactory(org.apache.druid.query.aggregation.hyperloglog.HyperUniquesAggregatorFactory) AggregatorFactory(org.apache.druid.query.aggregation.AggregatorFactory) FloatSumAggregatorFactory(org.apache.druid.query.aggregation.FloatSumAggregatorFactory) DoubleDimensionSchema(org.apache.druid.data.input.impl.DoubleDimensionSchema) LongDimensionSchema(org.apache.druid.data.input.impl.LongDimensionSchema) FloatDimensionSchema(org.apache.druid.data.input.impl.FloatDimensionSchema) DimensionSchema(org.apache.druid.data.input.impl.DimensionSchema) TimeAndDimsParseSpec(org.apache.druid.data.input.impl.TimeAndDimsParseSpec) DoubleDimensionSchema(org.apache.druid.data.input.impl.DoubleDimensionSchema) CountAggregatorFactory(org.apache.druid.query.aggregation.CountAggregatorFactory) TimestampSpec(org.apache.druid.data.input.impl.TimestampSpec) HyperUniquesAggregatorFactory(org.apache.druid.query.aggregation.hyperloglog.HyperUniquesAggregatorFactory) InputRow(org.apache.druid.data.input.InputRow) DimensionsSpec(org.apache.druid.data.input.impl.DimensionsSpec) BeforeClass(org.junit.BeforeClass)

Example 15 with DimensionsSpec

use of org.apache.druid.data.input.impl.DimensionsSpec in project druid by druid-io.

the class IndexMergerTestBase method testMultivalDim_persistAndMerge_dimensionValueOrderingRules.

@Test
public void testMultivalDim_persistAndMerge_dimensionValueOrderingRules() throws Exception {
    List<String> dims = Arrays.asList("dimA", "dimMultiVal");
    IncrementalIndexSchema indexSchema = new IncrementalIndexSchema.Builder().withDimensionsSpec(new DimensionsSpec(ImmutableList.of(new StringDimensionSchema("dimA", MultiValueHandling.SORTED_ARRAY, true), new StringDimensionSchema("dimMultiVal", MultiValueHandling.SORTED_ARRAY, true)))).withMetrics(new LongSumAggregatorFactory("sumCount", "sumCount")).withRollup(true).build();
    Map<String, Object> nullEvent = new HashMap<>();
    nullEvent.put("dimA", "leek");
    nullEvent.put("sumCount", 1L);
    Map<String, Object> nullEvent2 = new HashMap<>();
    nullEvent2.put("dimA", "leek");
    nullEvent2.put("dimMultiVal", null);
    nullEvent2.put("sumCount", 1L);
    Map<String, Object> emptyListEvent = new HashMap<>();
    emptyListEvent.put("dimA", "leek");
    emptyListEvent.put("dimMultiVal", ImmutableList.of());
    emptyListEvent.put("sumCount", 1L);
    List<String> listWithNull = new ArrayList<>();
    listWithNull.add(null);
    Map<String, Object> listWithNullEvent = new HashMap<>();
    listWithNullEvent.put("dimA", "leek");
    listWithNullEvent.put("dimMultiVal", listWithNull);
    listWithNullEvent.put("sumCount", 1L);
    Map<String, Object> emptyStringEvent = new HashMap<>();
    emptyStringEvent.put("dimA", "leek");
    emptyStringEvent.put("dimMultiVal", "");
    emptyStringEvent.put("sumCount", 1L);
    Map<String, Object> listWithEmptyStringEvent = new HashMap<>();
    listWithEmptyStringEvent.put("dimA", "leek");
    listWithEmptyStringEvent.put("dimMultiVal", ImmutableList.of(""));
    listWithEmptyStringEvent.put("sumCount", 1L);
    Map<String, Object> singleValEvent = new HashMap<>();
    singleValEvent.put("dimA", "leek");
    singleValEvent.put("dimMultiVal", "1");
    singleValEvent.put("sumCount", 1L);
    Map<String, Object> singleValEvent2 = new HashMap<>();
    singleValEvent2.put("dimA", "leek");
    singleValEvent2.put("dimMultiVal", "2");
    singleValEvent2.put("sumCount", 1L);
    Map<String, Object> singleValEvent3 = new HashMap<>();
    singleValEvent3.put("dimA", "potato");
    singleValEvent3.put("dimMultiVal", "2");
    singleValEvent3.put("sumCount", 1L);
    Map<String, Object> listWithSingleValEvent = new HashMap<>();
    listWithSingleValEvent.put("dimA", "leek");
    listWithSingleValEvent.put("dimMultiVal", ImmutableList.of("1"));
    listWithSingleValEvent.put("sumCount", 1L);
    Map<String, Object> listWithSingleValEvent2 = new HashMap<>();
    listWithSingleValEvent2.put("dimA", "leek");
    listWithSingleValEvent2.put("dimMultiVal", ImmutableList.of("2"));
    listWithSingleValEvent2.put("sumCount", 1L);
    Map<String, Object> listWithSingleValEvent3 = new HashMap<>();
    listWithSingleValEvent3.put("dimA", "potato");
    listWithSingleValEvent3.put("dimMultiVal", ImmutableList.of("2"));
    listWithSingleValEvent3.put("sumCount", 1L);
    Map<String, Object> multivalEvent = new HashMap<>();
    multivalEvent.put("dimA", "leek");
    multivalEvent.put("dimMultiVal", ImmutableList.of("1", "3"));
    multivalEvent.put("sumCount", 1L);
    Map<String, Object> multivalEvent2 = new HashMap<>();
    multivalEvent2.put("dimA", "leek");
    multivalEvent2.put("dimMultiVal", ImmutableList.of("1", "4"));
    multivalEvent2.put("sumCount", 1L);
    Map<String, Object> multivalEvent3 = new HashMap<>();
    multivalEvent3.put("dimA", "leek");
    multivalEvent3.put("dimMultiVal", ImmutableList.of("1", "3", "5"));
    multivalEvent3.put("sumCount", 1L);
    Map<String, Object> multivalEvent4 = new HashMap<>();
    multivalEvent4.put("dimA", "leek");
    multivalEvent4.put("dimMultiVal", ImmutableList.of("1", "2", "3"));
    multivalEvent4.put("sumCount", 1L);
    List<String> multivalEvent5List = Arrays.asList("1", "2", "3", null);
    Map<String, Object> multivalEvent5 = new HashMap<>();
    multivalEvent5.put("dimA", "leek");
    multivalEvent5.put("dimMultiVal", multivalEvent5List);
    multivalEvent5.put("sumCount", 1L);
    List<String> multivalEvent6List = Arrays.asList(null, "3");
    Map<String, Object> multivalEvent6 = new HashMap<>();
    multivalEvent6.put("dimA", "leek");
    multivalEvent6.put("dimMultiVal", multivalEvent6List);
    multivalEvent6.put("sumCount", 1L);
    Map<String, Object> multivalEvent7 = new HashMap<>();
    multivalEvent7.put("dimA", "leek");
    multivalEvent7.put("dimMultiVal", ImmutableList.of("1", "2", "3", ""));
    multivalEvent7.put("sumCount", 1L);
    Map<String, Object> multivalEvent8 = new HashMap<>();
    multivalEvent8.put("dimA", "leek");
    multivalEvent8.put("dimMultiVal", ImmutableList.of("", "3"));
    multivalEvent8.put("sumCount", 1L);
    Map<String, Object> multivalEvent9 = new HashMap<>();
    multivalEvent9.put("dimA", "potato");
    multivalEvent9.put("dimMultiVal", ImmutableList.of("1", "3"));
    multivalEvent9.put("sumCount", 1L);
    List<Map<String, Object>> events = ImmutableList.of(nullEvent, nullEvent2, emptyListEvent, listWithNullEvent, emptyStringEvent, listWithEmptyStringEvent, singleValEvent, singleValEvent2, singleValEvent3, listWithSingleValEvent, listWithSingleValEvent2, listWithSingleValEvent3, multivalEvent, multivalEvent2, multivalEvent3, multivalEvent4, multivalEvent5, multivalEvent6, multivalEvent7, multivalEvent8, multivalEvent9);
    IncrementalIndex toPersistA = new OnheapIncrementalIndex.Builder().setIndexSchema(indexSchema).setMaxRowCount(1000).build();
    for (Map<String, Object> event : events) {
        toPersistA.add(new MapBasedInputRow(1, dims, event));
    }
    final File tmpDirA = temporaryFolder.newFolder();
    QueryableIndex indexA = closer.closeLater(indexIO.loadIndex(indexMerger.persist(toPersistA, tmpDirA, indexSpec, null)));
    List<QueryableIndex> singleEventIndexes = new ArrayList<>();
    for (Map<String, Object> event : events) {
        IncrementalIndex toPersist = new OnheapIncrementalIndex.Builder().setIndexSchema(indexSchema).setMaxRowCount(1000).build();
        toPersist.add(new MapBasedInputRow(1, dims, event));
        final File tmpDir = temporaryFolder.newFolder();
        QueryableIndex queryableIndex = closer.closeLater(indexIO.loadIndex(indexMerger.persist(toPersist, tmpDir, indexSpec, null)));
        singleEventIndexes.add(queryableIndex);
    }
    singleEventIndexes.add(indexA);
    final File tmpDirMerged = temporaryFolder.newFolder();
    final QueryableIndex merged = closer.closeLater(indexIO.loadIndex(indexMerger.mergeQueryableIndex(singleEventIndexes, true, new AggregatorFactory[] { new LongSumAggregatorFactory("sumCount", "sumCount") }, tmpDirMerged, indexSpec, null, -1)));
    final QueryableIndexIndexableAdapter adapter = new QueryableIndexIndexableAdapter(merged);
    final List<DebugRow> rowList = RowIteratorHelper.toList(adapter.getRows());
    Assert.assertEquals(ImmutableList.of("dimA", "dimMultiVal"), ImmutableList.copyOf(adapter.getDimensionNames()));
    if (NullHandling.replaceWithDefault()) {
        Assert.assertEquals(11, rowList.size());
        Assert.assertEquals(Arrays.asList("leek", null), rowList.get(0).dimensionValues());
        Assert.assertEquals(12L, rowList.get(0).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("leek", Arrays.asList(null, "1", "2", "3")), rowList.get(1).dimensionValues());
        Assert.assertEquals(4L, rowList.get(1).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("leek", Arrays.asList(null, "3")), rowList.get(2).dimensionValues());
        Assert.assertEquals(4L, rowList.get(2).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("leek", "1"), rowList.get(3).dimensionValues());
        Assert.assertEquals(4L, rowList.get(3).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("leek", Arrays.asList("1", "2", "3")), rowList.get(4).dimensionValues());
        Assert.assertEquals(2L, rowList.get(4).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("leek", Arrays.asList("1", "3")), rowList.get(5).dimensionValues());
        Assert.assertEquals(2L, rowList.get(5).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("leek", Arrays.asList("1", "3", "5")), rowList.get(6).dimensionValues());
        Assert.assertEquals(2L, rowList.get(6).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("leek", Arrays.asList("1", "4")), rowList.get(7).dimensionValues());
        Assert.assertEquals(2L, rowList.get(7).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("leek", "2"), rowList.get(8).dimensionValues());
        Assert.assertEquals(4L, rowList.get(8).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("potato", Arrays.asList("1", "3")), rowList.get(9).dimensionValues());
        Assert.assertEquals(2L, rowList.get(9).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("potato", "2"), rowList.get(10).dimensionValues());
        Assert.assertEquals(4L, rowList.get(10).metricValues().get(0));
        checkBitmapIndex(Arrays.asList(0, 1, 2, 3, 4, 5, 6, 7, 8), adapter.getBitmapIndex("dimA", "leek"));
        checkBitmapIndex(Arrays.asList(9, 10), adapter.getBitmapIndex("dimA", "potato"));
        checkBitmapIndex(Arrays.asList(0, 1, 2), adapter.getBitmapIndex("dimMultiVal", null));
        checkBitmapIndex(ImmutableList.of(), adapter.getBitmapIndex("dimMultiVal", ""));
        checkBitmapIndex(Arrays.asList(1, 3, 4, 5, 6, 7, 9), adapter.getBitmapIndex("dimMultiVal", "1"));
        checkBitmapIndex(Arrays.asList(1, 4, 8, 10), adapter.getBitmapIndex("dimMultiVal", "2"));
        checkBitmapIndex(Arrays.asList(1, 2, 4, 5, 6, 9), adapter.getBitmapIndex("dimMultiVal", "3"));
        checkBitmapIndex(Collections.singletonList(7), adapter.getBitmapIndex("dimMultiVal", "4"));
        checkBitmapIndex(Collections.singletonList(6), adapter.getBitmapIndex("dimMultiVal", "5"));
    } else {
        Assert.assertEquals(14, rowList.size());
        Assert.assertEquals(Arrays.asList("leek", null), rowList.get(0).dimensionValues());
        Assert.assertEquals(8L, rowList.get(0).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("leek", Arrays.asList(null, "1", "2", "3")), rowList.get(1).dimensionValues());
        Assert.assertEquals(2L, rowList.get(1).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("leek", Arrays.asList(null, "3")), rowList.get(2).dimensionValues());
        Assert.assertEquals(2L, rowList.get(2).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("leek", ""), rowList.get(3).dimensionValues());
        Assert.assertEquals(4L, rowList.get(3).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("leek", Arrays.asList("", "1", "2", "3")), rowList.get(4).dimensionValues());
        Assert.assertEquals(2L, rowList.get(4).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("leek", Arrays.asList("", "3")), rowList.get(5).dimensionValues());
        Assert.assertEquals(2L, rowList.get(5).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("leek", "1"), rowList.get(6).dimensionValues());
        Assert.assertEquals(4L, rowList.get(6).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("leek", Arrays.asList("1", "2", "3")), rowList.get(7).dimensionValues());
        Assert.assertEquals(2L, rowList.get(7).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("leek", Arrays.asList("1", "3")), rowList.get(8).dimensionValues());
        Assert.assertEquals(2L, rowList.get(8).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("leek", Arrays.asList("1", "3", "5")), rowList.get(9).dimensionValues());
        Assert.assertEquals(2L, rowList.get(9).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("leek", Arrays.asList("1", "4")), rowList.get(10).dimensionValues());
        Assert.assertEquals(2L, rowList.get(10).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("leek", "2"), rowList.get(11).dimensionValues());
        Assert.assertEquals(4L, rowList.get(11).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("potato", Arrays.asList("1", "3")), rowList.get(12).dimensionValues());
        Assert.assertEquals(2L, rowList.get(12).metricValues().get(0));
        Assert.assertEquals(Arrays.asList("potato", "2"), rowList.get(13).dimensionValues());
        Assert.assertEquals(4L, rowList.get(13).metricValues().get(0));
        checkBitmapIndex(Arrays.asList(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11), adapter.getBitmapIndex("dimA", "leek"));
        checkBitmapIndex(Arrays.asList(12, 13), adapter.getBitmapIndex("dimA", "potato"));
        checkBitmapIndex(Arrays.asList(0, 1, 2), adapter.getBitmapIndex("dimMultiVal", null));
        checkBitmapIndex(ImmutableList.of(3, 4, 5), adapter.getBitmapIndex("dimMultiVal", ""));
        checkBitmapIndex(Arrays.asList(1, 4, 6, 7, 8, 9, 10, 12), adapter.getBitmapIndex("dimMultiVal", "1"));
        checkBitmapIndex(Arrays.asList(1, 4, 7, 11, 13), adapter.getBitmapIndex("dimMultiVal", "2"));
        checkBitmapIndex(Arrays.asList(1, 2, 4, 5, 7, 8, 9, 12), adapter.getBitmapIndex("dimMultiVal", "3"));
        checkBitmapIndex(Collections.singletonList(10), adapter.getBitmapIndex("dimMultiVal", "4"));
        checkBitmapIndex(Collections.singletonList(9), adapter.getBitmapIndex("dimMultiVal", "5"));
    }
}
Also used : HashMap(java.util.HashMap) IncrementalIndex(org.apache.druid.segment.incremental.IncrementalIndex) OnheapIncrementalIndex(org.apache.druid.segment.incremental.OnheapIncrementalIndex) LongSumAggregatorFactory(org.apache.druid.query.aggregation.LongSumAggregatorFactory) ArrayList(java.util.ArrayList) OnheapIncrementalIndex(org.apache.druid.segment.incremental.OnheapIncrementalIndex) StringDimensionSchema(org.apache.druid.data.input.impl.StringDimensionSchema) DimensionsSpec(org.apache.druid.data.input.impl.DimensionsSpec) MapBasedInputRow(org.apache.druid.data.input.MapBasedInputRow) Map(java.util.Map) ImmutableMap(com.google.common.collect.ImmutableMap) HashMap(java.util.HashMap) File(java.io.File) IncrementalIndexSchema(org.apache.druid.segment.incremental.IncrementalIndexSchema) IncrementalIndexTest(org.apache.druid.segment.data.IncrementalIndexTest) InitializedNullHandlingTest(org.apache.druid.testing.InitializedNullHandlingTest) Test(org.junit.Test)

Aggregations

DimensionsSpec (org.apache.druid.data.input.impl.DimensionsSpec)169 Test (org.junit.Test)129 TimestampSpec (org.apache.druid.data.input.impl.TimestampSpec)114 InputRow (org.apache.druid.data.input.InputRow)52 AggregatorFactory (org.apache.druid.query.aggregation.AggregatorFactory)47 LongSumAggregatorFactory (org.apache.druid.query.aggregation.LongSumAggregatorFactory)47 UniformGranularitySpec (org.apache.druid.segment.indexing.granularity.UniformGranularitySpec)42 DataSchema (org.apache.druid.segment.indexing.DataSchema)39 StringDimensionSchema (org.apache.druid.data.input.impl.StringDimensionSchema)37 CountAggregatorFactory (org.apache.druid.query.aggregation.CountAggregatorFactory)37 InputRowSchema (org.apache.druid.data.input.InputRowSchema)36 Map (java.util.Map)32 InitializedNullHandlingTest (org.apache.druid.testing.InitializedNullHandlingTest)32 InputEntityReader (org.apache.druid.data.input.InputEntityReader)31 ArrayList (java.util.ArrayList)29 CsvInputFormat (org.apache.druid.data.input.impl.CsvInputFormat)25 MapBasedInputRow (org.apache.druid.data.input.MapBasedInputRow)24 JSONPathSpec (org.apache.druid.java.util.common.parsers.JSONPathSpec)24 HashMap (java.util.HashMap)23 ImmutableMap (com.google.common.collect.ImmutableMap)21