use of org.apache.druid.query.groupby.ResultRow in project druid by druid-io.
the class SketchAggregationTest method testSketchDataIngestAndGpByQuery.
@Test
public void testSketchDataIngestAndGpByQuery() throws Exception {
final GroupByQuery groupByQuery = readQueryFromClasspath("sketch_test_data_group_by_query.json", helper.getObjectMapper(), vectorize);
final Sequence<ResultRow> seq = helper.createIndexAndRunQueryOnSegment(new File(SketchAggregationTest.class.getClassLoader().getResource("sketch_test_data.tsv").getFile()), readFileFromClasspathAsString("sketch_test_data_record_parser.json"), readFileFromClasspathAsString("sketch_test_data_aggregators.json"), 0, Granularities.NONE, 1000, groupByQuery);
final String expectedSummary = "\n### HeapCompactSketch SUMMARY: \n" + " Estimate : 50.0\n" + " Upper Bound, 95% conf : 50.0\n" + " Lower Bound, 95% conf : 50.0\n" + " Theta (double) : 1.0\n" + " Theta (long) : 9223372036854775807\n" + " Theta (long) hex : 7fffffffffffffff\n" + " EstMode? : false\n" + " Empty? : false\n" + " Ordered? : true\n" + " Retained Entries : 50\n" + " Seed Hash : 93cc | 37836\n" + "### END SKETCH SUMMARY\n";
List<ResultRow> results = seq.toList();
Assert.assertEquals(1, results.size());
Assert.assertEquals(ResultRow.fromLegacyRow(new MapBasedRow(DateTimes.of("2014-10-19T00:00:00.000Z"), ImmutableMap.<String, Object>builder().put("sids_sketch_count", 50.0).put("sids_sketch_count_with_err", new SketchEstimateWithErrorBounds(50.0, 50.0, 50.0, 2)).put("sketchEstimatePostAgg", 50.0).put("sketchEstimatePostAggWithErrorBounds", new SketchEstimateWithErrorBounds(50.0, 50.0, 50.0, 2)).put("sketchUnionPostAggEstimate", 50.0).put("sketchSummary", expectedSummary).put("sketchIntersectionPostAggEstimate", 50.0).put("sketchAnotBPostAggEstimate", 0.0).put("non_existing_col_validation", 0.0).build()), groupByQuery), results.get(0));
}
use of org.apache.druid.query.groupby.ResultRow in project druid by druid-io.
the class SketchAggregationTest method testThetaCardinalityOnSimpleColumn.
@Test
public void testThetaCardinalityOnSimpleColumn() throws Exception {
final GroupByQuery groupByQuery = readQueryFromClasspath("simple_test_data_group_by_query.json", helper.getObjectMapper(), vectorize);
final Sequence<ResultRow> seq = helper.createIndexAndRunQueryOnSegment(new File(SketchAggregationTest.class.getClassLoader().getResource("simple_test_data.tsv").getFile()), readFileFromClasspathAsString("simple_test_data_record_parser2.json"), "[" + " {" + " \"type\": \"count\"," + " \"name\": \"count\"" + " }" + "]", 0, Granularities.NONE, 1000, groupByQuery);
List<ResultRow> results = seq.toList();
Assert.assertEquals(5, results.size());
Assert.assertEquals(ImmutableList.of(new MapBasedRow(DateTimes.of("2014-10-19T00:00:00.000Z"), ImmutableMap.<String, Object>builder().put("product", "product_3").put("sketch_count", 38.0).put("sketchEstimatePostAgg", 38.0).put("sketchUnionPostAggEstimate", 38.0).put("sketchIntersectionPostAggEstimate", 38.0).put("sketchAnotBPostAggEstimate", 0.0).put("non_existing_col_validation", 0.0).build()), new MapBasedRow(DateTimes.of("2014-10-19T00:00:00.000Z"), ImmutableMap.<String, Object>builder().put("product", "product_1").put("sketch_count", 42.0).put("sketchEstimatePostAgg", 42.0).put("sketchUnionPostAggEstimate", 42.0).put("sketchIntersectionPostAggEstimate", 42.0).put("sketchAnotBPostAggEstimate", 0.0).put("non_existing_col_validation", 0.0).build()), new MapBasedRow(DateTimes.of("2014-10-19T00:00:00.000Z"), ImmutableMap.<String, Object>builder().put("product", "product_2").put("sketch_count", 42.0).put("sketchEstimatePostAgg", 42.0).put("sketchUnionPostAggEstimate", 42.0).put("sketchIntersectionPostAggEstimate", 42.0).put("sketchAnotBPostAggEstimate", 0.0).put("non_existing_col_validation", 0.0).build()), new MapBasedRow(DateTimes.of("2014-10-19T00:00:00.000Z"), ImmutableMap.<String, Object>builder().put("product", "product_4").put("sketch_count", 42.0).put("sketchEstimatePostAgg", 42.0).put("sketchUnionPostAggEstimate", 42.0).put("sketchIntersectionPostAggEstimate", 42.0).put("sketchAnotBPostAggEstimate", 0.0).put("non_existing_col_validation", 0.0).build()), new MapBasedRow(DateTimes.of("2014-10-19T00:00:00.000Z"), ImmutableMap.<String, Object>builder().put("product", "product_5").put("sketch_count", 42.0).put("sketchEstimatePostAgg", 42.0).put("sketchUnionPostAggEstimate", 42.0).put("sketchIntersectionPostAggEstimate", 42.0).put("sketchAnotBPostAggEstimate", 0.0).put("non_existing_col_validation", 0.0).build())).stream().map(row -> ResultRow.fromLegacyRow(row, groupByQuery)).collect(Collectors.toList()), results);
}
use of org.apache.druid.query.groupby.ResultRow in project druid by druid-io.
the class ArrayOfDoublesSketchAggregationTest method buildingSketchesAtQueryTimeUseNumerical.
@Test
public void buildingSketchesAtQueryTimeUseNumerical() throws Exception {
Sequence<ResultRow> seq = helper.createIndexAndRunQueryOnSegment(new File(this.getClass().getClassLoader().getResource("tuple/array_of_doubles_build_data.tsv").getFile()), String.join("\n", "{", " \"type\": \"string\",", " \"parseSpec\": {", " \"format\": \"tsv\",", " \"timestampSpec\": {\"column\": \"timestamp\", \"format\": \"yyyyMMddHH\"},", " \"dimensionsSpec\": {", " \"dimensions\": [\"product\", \"key\", {\"type\": \"long\", \"name\": \"key_num\"}],", " \"dimensionExclusions\": [],", " \"spatialDimensions\": []", " },", " \"columns\": [\"timestamp\", \"product\", \"key\", \"key_num\", \"value\"]", " }", "}"), String.join("\n", "[", " {\"type\": \"doubleSum\", \"name\": \"value\", \"fieldName\": \"value\"}", "]"), // minTimestamp
0, Granularities.NONE, // maxRowCount
40, String.join("\n", "{", " \"queryType\": \"groupBy\",", " \"dataSource\": \"test_datasource\",", " \"granularity\": \"ALL\",", " \"dimensions\": [],", " \"aggregations\": [", " {\"type\": \"arrayOfDoublesSketch\", \"name\": \"sketch\", \"fieldName\": \"key_num\", \"metricColumns\": [\"value\"], \"nominalEntries\": 1024},", " {\"type\": \"count\", \"name\":\"cnt\"}", " ],", " \"postAggregations\": [", " {\"type\": \"arrayOfDoublesSketchToEstimate\", \"name\": \"estimate\", \"field\": {\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}},", " {\"type\": \"arrayOfDoublesSketchToQuantilesSketch\", \"name\": \"quantiles-sketch\", \"field\": {\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}},", " {\"type\": \"arrayOfDoublesSketchToEstimate\", \"name\": \"union\", \"field\": {", " \"type\": \"arrayOfDoublesSketchSetOp\",", " \"name\": \"union\",", " \"operation\": \"UNION\",", " \"nominalEntries\": 1024,", " \"fields\": [{\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}, {\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}]", " }},", " {\"type\": \"arrayOfDoublesSketchToEstimate\", \"name\": \"intersection\", \"field\": {", " \"type\": \"arrayOfDoublesSketchSetOp\",", " \"name\": \"intersection\",", " \"operation\": \"INTERSECT\",", " \"nominalEntries\": 1024,", " \"fields\": [{\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}, {\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}]", " }},", " {\"type\": \"arrayOfDoublesSketchToEstimate\", \"name\": \"anotb\", \"field\": {", " \"type\": \"arrayOfDoublesSketchSetOp\",", " \"name\": \"anotb\",", " \"operation\": \"NOT\",", " \"nominalEntries\": 1024,", " \"fields\": [{\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}, {\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}]", " }}", " ],", " \"intervals\": [\"2015-01-01T00:00:00.000Z/2015-01-31T00:00:00.000Z\"]", "}"));
List<ResultRow> results = seq.toList();
Assert.assertEquals(1, results.size());
ResultRow row = results.get(0);
Assert.assertEquals("cnt", 40.0, new Double(row.get(1).toString()), 0);
Assert.assertEquals("sketch", 40.0, (double) row.get(0), 0);
Assert.assertEquals("estimate", 40.0, new Double(row.get(2).toString()), 0);
Assert.assertEquals("union", 40.0, new Double(row.get(4).toString()), 0);
Assert.assertEquals("intersection", 40.0, new Double(row.get(5).toString()), 0);
Assert.assertEquals("anotb", 0, new Double(row.get(6).toString()), 0);
// quantiles-sketch
Object obj = row.get(3);
Assert.assertTrue(obj instanceof DoublesSketch);
DoublesSketch ds = (DoublesSketch) obj;
Assert.assertEquals(40, ds.getN());
Assert.assertEquals(1.0, ds.getMinValue(), 0);
Assert.assertEquals(1.0, ds.getMaxValue(), 0);
}
use of org.apache.druid.query.groupby.ResultRow in project druid by druid-io.
the class ArrayOfDoublesSketchAggregationTest method ingestingSketches.
@Test
public void ingestingSketches() throws Exception {
Sequence<ResultRow> seq = helper.createIndexAndRunQueryOnSegment(new File(this.getClass().getClassLoader().getResource("tuple/array_of_doubles_sketch_data.tsv").getFile()), String.join("\n", "{", " \"type\": \"string\",", " \"parseSpec\": {", " \"format\": \"tsv\",", " \"timestampSpec\": {\"column\": \"timestamp\", \"format\": \"yyyyMMddHH\"},", " \"dimensionsSpec\": {", " \"dimensions\": [\"product\"],", " \"dimensionExclusions\": [],", " \"spatialDimensions\": []", " },", " \"columns\": [\"timestamp\", \"product\", \"sketch\"]", " }", "}"), String.join("\n", "[", " {\"type\": \"arrayOfDoublesSketch\", \"name\": \"sketch\", \"fieldName\": \"sketch\", \"nominalEntries\": 1024},", " {\"type\": \"arrayOfDoublesSketch\", \"name\": \"non_existing_sketch\", \"fieldName\": \"non_existing_sketch\"}", "]"), // minTimestamp
0, Granularities.NONE, // maxRowCount
10, String.join("\n", "{", " \"queryType\": \"groupBy\",", " \"dataSource\": \"test_datasource\",", " \"granularity\": \"ALL\",", " \"dimensions\": [],", " \"aggregations\": [", " {\"type\": \"arrayOfDoublesSketch\", \"name\": \"sketch\", \"fieldName\": \"sketch\", \"nominalEntries\": 1024},", " {\"type\": \"arrayOfDoublesSketch\", \"name\": \"non_existing_sketch\", \"fieldName\": \"non_existing_sketch\"}", " ],", " \"postAggregations\": [", " {\"type\": \"arrayOfDoublesSketchToEstimate\", \"name\": \"estimate\", \"field\": {\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}},", " {\"type\": \"arrayOfDoublesSketchToEstimateAndBounds\", \"name\": \"estimateAndBounds\", \"field\": {\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}, \"numStdDevs\": 2},", " {\"type\": \"arrayOfDoublesSketchToQuantilesSketch\", \"name\": \"quantiles-sketch\", \"field\": {\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}},", " {\"type\": \"arrayOfDoublesSketchToEstimate\", \"name\": \"union\", \"field\": {", " \"type\": \"arrayOfDoublesSketchSetOp\",", " \"name\": \"union\",", " \"operation\": \"UNION\",", " \"nominalEntries\": 1024,", " \"fields\": [{\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}, {\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}]", " }},", " {\"type\": \"arrayOfDoublesSketchToEstimate\", \"name\": \"intersection\", \"field\": {", " \"type\": \"arrayOfDoublesSketchSetOp\",", " \"name\": \"intersection\",", " \"operation\": \"INTERSECT\",", " \"nominalEntries\": 1024,", " \"fields\": [{\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}, {\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}]", " }},", " {\"type\": \"arrayOfDoublesSketchToEstimate\", \"name\": \"anotb\", \"field\": {", " \"type\": \"arrayOfDoublesSketchSetOp\",", " \"name\": \"anotb\",", " \"operation\": \"NOT\",", " \"nominalEntries\": 1024,", " \"fields\": [{\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}, {\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}]", " }},", " {\"type\": \"arrayOfDoublesSketchToString\", \"name\": \"summary\", \"field\": {\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}},", " {\"type\": \"arrayOfDoublesSketchToVariances\", \"name\": \"variances\", \"field\": {\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}}", " ],", " \"intervals\": [\"2015-01-01T00:00:00.000Z/2015-01-31T00:00:00.000Z\"]", "}"));
List<ResultRow> results = seq.toList();
Assert.assertEquals(1, results.size());
ResultRow row = results.get(0);
Assert.assertEquals("sketch", 40.0, (double) row.get(0), 0);
Assert.assertEquals("non_existing_sketch", 0, (double) row.get(1), 0);
Assert.assertEquals("estimate", 40.0, (double) row.get(2), 0);
Assert.assertArrayEquals("estimateAndBounds", new double[] { 40.0, 40.0, 40.0 }, (double[]) row.get(3), 0);
Assert.assertEquals("union", 40.0, (double) row.get(5), 0);
Assert.assertEquals("intersection", 40.0, (double) row.get(6), 0);
Assert.assertEquals("anotb", 0, (double) row.get(7), 0);
Assert.assertArrayEquals("variances", new double[] { 0.0 }, (double[]) row.get(9), 0);
// quantiles-sketch
Object obj = row.get(4);
Assert.assertTrue(obj instanceof DoublesSketch);
DoublesSketch ds = (DoublesSketch) obj;
Assert.assertEquals(40, ds.getN());
Assert.assertEquals(1.0, ds.getMinValue(), 0);
Assert.assertEquals(1.0, ds.getMaxValue(), 0);
final String expectedSummary = "### HeapArrayOfDoublesCompactSketch SUMMARY: \n" + " Estimate : 40.0\n" + " Upper Bound, 95% conf : 40.0\n" + " Lower Bound, 95% conf : 40.0\n" + " Theta (double) : 1.0\n" + " Theta (long) : 9223372036854775807\n" + " EstMode? : false\n" + " Empty? : false\n" + " Retained Entries : 40\n" + " Seed Hash : 93cc | 37836\n" + "### END SKETCH SUMMARY\n";
Assert.assertEquals("summary", expectedSummary, row.get(8));
}
use of org.apache.druid.query.groupby.ResultRow in project druid by druid-io.
the class ArrayOfDoublesSketchAggregationTest method buildingSketchesAtIngestionTime.
@Test
public void buildingSketchesAtIngestionTime() throws Exception {
Sequence<ResultRow> seq = helper.createIndexAndRunQueryOnSegment(new File(this.getClass().getClassLoader().getResource("tuple/array_of_doubles_build_data.tsv").getFile()), String.join("\n", "{", " \"type\": \"string\",", " \"parseSpec\": {", " \"format\": \"tsv\",", " \"timestampSpec\": {\"column\": \"timestamp\", \"format\": \"yyyyMMddHH\"},", " \"dimensionsSpec\": {", " \"dimensions\": [\"product\"],", " \"dimensionExclusions\": [],", " \"spatialDimensions\": []", " },", " \"columns\": [\"timestamp\", \"product\", \"key\", \"key_num\", \"value\"]", " }", "}"), String.join("\n", "[", " {\"type\": \"arrayOfDoublesSketch\", \"name\": \"sketch\", \"fieldName\": \"key\", \"metricColumns\": [\"value\"], \"nominalEntries\": 1024}", "]"), // minTimestamp
0, Granularities.NONE, // maxRowCount
10, String.join("\n", "{", " \"queryType\": \"groupBy\",", " \"dataSource\": \"test_datasource\",", " \"granularity\": \"ALL\",", " \"dimensions\": [],", " \"aggregations\": [", " {\"type\": \"arrayOfDoublesSketch\", \"name\": \"sketch\", \"fieldName\": \"sketch\", \"size\": 1024}", " ],", " \"postAggregations\": [", " {\"type\": \"arrayOfDoublesSketchToEstimate\", \"name\": \"estimate\", \"field\": {\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}},", " {\"type\": \"arrayOfDoublesSketchToQuantilesSketch\", \"name\": \"quantiles-sketch\", \"field\": {\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}},", " {\"type\": \"arrayOfDoublesSketchToEstimate\", \"name\": \"union\", \"field\": {", " \"type\": \"arrayOfDoublesSketchSetOp\",", " \"name\": \"union\",", " \"operation\": \"UNION\",", " \"nominalEntries\": 1024,", " \"fields\": [{\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}, {\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}]", " }},", " {\"type\": \"arrayOfDoublesSketchToEstimate\", \"name\": \"intersection\", \"field\": {", " \"type\": \"arrayOfDoublesSketchSetOp\",", " \"name\": \"intersection\",", " \"operation\": \"INTERSECT\",", " \"nominalEntries\": 1024,", " \"fields\": [{\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}, {\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}]", " }},", " {\"type\": \"arrayOfDoublesSketchToEstimate\", \"name\": \"anotb\", \"field\": {", " \"type\": \"arrayOfDoublesSketchSetOp\",", " \"name\": \"anotb\",", " \"operation\": \"NOT\",", " \"nominalEntries\": 1024,", " \"fields\": [{\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}, {\"type\": \"fieldAccess\", \"fieldName\": \"sketch\"}]", " }}", " ],", " \"intervals\": [\"2015-01-01T00:00:00.000Z/2015-01-31T00:00:00.000Z\"]", "}"));
List<ResultRow> results = seq.toList();
Assert.assertEquals(1, results.size());
ResultRow row = results.get(0);
Assert.assertEquals("sketch", 40.0, (double) row.get(0), 0);
Assert.assertEquals("estimate", 40.0, (double) row.get(1), 0);
Assert.assertEquals("union", 40.0, (double) row.get(3), 0);
Assert.assertEquals("intersection", 40.0, (double) row.get(4), 0);
Assert.assertEquals("anotb", 0, (double) row.get(5), 0);
// quantiles-sketch
Object obj = row.get(2);
Assert.assertTrue(obj instanceof DoublesSketch);
DoublesSketch ds = (DoublesSketch) obj;
Assert.assertEquals(40, ds.getN());
Assert.assertEquals(1.0, ds.getMinValue(), 0);
Assert.assertEquals(1.0, ds.getMaxValue(), 0);
}
Aggregations