Examples with LookupExtractionFn io.druid.query.lookup.LookupExtractionFn used on opensource projects

Example 6 with LookupExtractionFn

use of io.druid.query.lookup.LookupExtractionFn in project druid by druid-io.

the class GroupByQueryRunnerTest method testGroupByWithSimpleRename.

@Test
public void testGroupByWithSimpleRename() {
    Map<String, String> map = new HashMap<>();
    map.put("automotive", "automotive0");
    map.put("business", "business0");
    map.put("entertainment", "entertainment0");
    map.put("health", "health0");
    map.put("mezzanine", "mezzanine0");
    map.put("news", "news0");
    map.put("premium", "premium0");
    map.put("technology", "technology0");
    map.put("travel", "travel0");
    GroupByQuery query = GroupByQuery.builder().setDataSource(QueryRunnerTestHelper.dataSource).setQuerySegmentSpec(QueryRunnerTestHelper.firstToThird).setDimensions(Lists.<DimensionSpec>newArrayList(new ExtractionDimensionSpec("quality", "alias", new LookupExtractionFn(new MapLookupExtractor(map, false), false, null, true, false)))).setAggregatorSpecs(Arrays.asList(QueryRunnerTestHelper.rowsCount, new LongSumAggregatorFactory("idx", "index"))).setGranularity(QueryRunnerTestHelper.dayGran).build();
    List<Row> expectedResults = Arrays.asList(GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "automotive0", "rows", 1L, "idx", 135L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "business0", "rows", 1L, "idx", 118L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "entertainment0", "rows", 1L, "idx", 158L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "health0", "rows", 1L, "idx", 120L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "mezzanine0", "rows", 3L, "idx", 2870L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "news0", "rows", 1L, "idx", 121L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "premium0", "rows", 3L, "idx", 2900L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "technology0", "rows", 1L, "idx", 78L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "travel0", "rows", 1L, "idx", 119L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "automotive0", "rows", 1L, "idx", 147L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "business0", "rows", 1L, "idx", 112L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "entertainment0", "rows", 1L, "idx", 166L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "health0", "rows", 1L, "idx", 113L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "mezzanine0", "rows", 3L, "idx", 2447L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "news0", "rows", 1L, "idx", 114L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "premium0", "rows", 3L, "idx", 2505L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "technology0", "rows", 1L, "idx", 97L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "travel0", "rows", 1L, "idx", 126L));
    Iterable<Row> results = GroupByQueryRunnerTestHelper.runQuery(factory, runner, query);
    TestHelper.assertExpectedObjects(expectedResults, results, "");
}

Example 7 with LookupExtractionFn

use of io.druid.query.lookup.LookupExtractionFn in project druid by druid-io.

the class GroupByQueryRunnerTest method testGroupByWithAggregatorFilterAndExtractionFunction.

@Test
public void testGroupByWithAggregatorFilterAndExtractionFunction() {
    Map<String, String> extractionMap = new HashMap<>();
    extractionMap.put("automotive", "automotive0");
    extractionMap.put("business", "business0");
    extractionMap.put("entertainment", "entertainment0");
    extractionMap.put("health", "health0");
    extractionMap.put("mezzanine", "mezzanineANDnews");
    extractionMap.put("news", "mezzanineANDnews");
    extractionMap.put("premium", "premium0");
    extractionMap.put("technology", "technology0");
    extractionMap.put("travel", "travel0");
    MapLookupExtractor mapLookupExtractor = new MapLookupExtractor(extractionMap, false);
    LookupExtractionFn lookupExtractionFn = new LookupExtractionFn(mapLookupExtractor, false, "missing", true, false);
    DimFilter filter = new ExtractionDimFilter("quality", "mezzanineANDnews", lookupExtractionFn, null);
    GroupByQuery query = GroupByQuery.builder().setDataSource(QueryRunnerTestHelper.dataSource).setQuerySegmentSpec(QueryRunnerTestHelper.firstToThird).setDimensions(Lists.<DimensionSpec>newArrayList(new DefaultDimensionSpec("quality", "alias"))).setAggregatorSpecs(Arrays.asList(new FilteredAggregatorFactory(QueryRunnerTestHelper.rowsCount, filter), (AggregatorFactory) new FilteredAggregatorFactory(new LongSumAggregatorFactory("idx", "index"), filter))).setGranularity(QueryRunnerTestHelper.dayGran).build();
    List<Row> expectedResults = Arrays.asList(GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "automotive", "rows", 0L, "idx", 0L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "business", "rows", 0L, "idx", 0L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "entertainment", "rows", 0L, "idx", 0L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "health", "rows", 0L, "idx", 0L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "mezzanine", "rows", 3L, "idx", 2870L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "news", "rows", 1L, "idx", 121L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "premium", "rows", 0L, "idx", 0L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "technology", "rows", 0L, "idx", 0L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "travel", "rows", 0L, "idx", 0L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "automotive", "rows", 0L, "idx", 0L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "business", "rows", 0L, "idx", 0L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "entertainment", "rows", 0L, "idx", 0L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "health", "rows", 0L, "idx", 0L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "mezzanine", "rows", 3L, "idx", 2447L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "news", "rows", 1L, "idx", 114L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "premium", "rows", 0L, "idx", 0L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "technology", "rows", 0L, "idx", 0L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "travel", "rows", 0L, "idx", 0L));
    Iterable<Row> results = GroupByQueryRunnerTestHelper.runQuery(factory, runner, query);
    TestHelper.assertExpectedObjects(expectedResults, results, "");
}

Example 8 with LookupExtractionFn

use of io.druid.query.lookup.LookupExtractionFn in project druid by druid-io.

the class GroupByQueryRunnerTest method testGroupByWithExtractionDimFilterCaseMappingValueIsNullOrEmpty.

@Test
public void testGroupByWithExtractionDimFilterCaseMappingValueIsNullOrEmpty() {
    Map<String, String> extractionMap = new HashMap<>();
    extractionMap.put("automotive", "automotive0");
    extractionMap.put("business", "business0");
    extractionMap.put("entertainment", "entertainment0");
    extractionMap.put("health", "health0");
    extractionMap.put("mezzanine", null);
    extractionMap.put("news", "");
    extractionMap.put("premium", "premium0");
    extractionMap.put("technology", "technology0");
    extractionMap.put("travel", "travel0");
    MapLookupExtractor mapLookupExtractor = new MapLookupExtractor(extractionMap, false);
    LookupExtractionFn lookupExtractionFn = new LookupExtractionFn(mapLookupExtractor, false, null, true, false);
    GroupByQuery query = GroupByQuery.builder().setDataSource(QueryRunnerTestHelper.dataSource).setQuerySegmentSpec(QueryRunnerTestHelper.firstToThird).setDimensions(Lists.<DimensionSpec>newArrayList(new DefaultDimensionSpec("quality", "alias"))).setAggregatorSpecs(Arrays.asList(QueryRunnerTestHelper.rowsCount, new LongSumAggregatorFactory("idx", "index"))).setGranularity(QueryRunnerTestHelper.dayGran).setDimFilter(new ExtractionDimFilter("quality", "", lookupExtractionFn, null)).build();
    List<Row> expectedResults = Arrays.asList(GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "mezzanine", "rows", 3L, "idx", 2870L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "news", "rows", 1L, "idx", 121L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "mezzanine", "rows", 3L, "idx", 2447L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "news", "rows", 1L, "idx", 114L));
    Iterable<Row> results = GroupByQueryRunnerTestHelper.runQuery(factory, runner, query);
    TestHelper.assertExpectedObjects(expectedResults, results, "");
}

Example 9 with LookupExtractionFn

use of io.druid.query.lookup.LookupExtractionFn in project druid by druid-io.

the class SelectQueryRunnerTest method testFullOnSelectWithDimensionSpec.

@Test
public void testFullOnSelectWithDimensionSpec() {
    Map<String, String> map = new HashMap<>();
    map.put("automotive", "automotive0");
    map.put("business", "business0");
    map.put("entertainment", "entertainment0");
    map.put("health", "health0");
    map.put("mezzanine", "mezzanine0");
    map.put("news", "news0");
    map.put("premium", "premium0");
    map.put("technology", "technology0");
    map.put("travel", "travel0");
    SelectQuery query = newTestQuery().dimensionSpecs(Arrays.<DimensionSpec>asList(new DefaultDimensionSpec(QueryRunnerTestHelper.marketDimension, "mar"), new ExtractionDimensionSpec(QueryRunnerTestHelper.qualityDimension, "qual", new LookupExtractionFn(new MapLookupExtractor(map, true), false, null, true, false)), new DefaultDimensionSpec(QueryRunnerTestHelper.placementDimension, "place"))).build();
    HashMap<String, Object> context = new HashMap<String, Object>();
    Iterable<Result<SelectResultValue>> results = Sequences.toList(runner.run(query, context), Lists.<Result<SelectResultValue>>newArrayList());
    List<Result<SelectResultValue>> expectedResultsAsc = Arrays.asList(new Result<SelectResultValue>(new DateTime("2011-01-12T00:00:00.000Z"), new SelectResultValue(ImmutableMap.of(QueryRunnerTestHelper.segmentId, 2), Sets.newHashSet("mar", "qual", "place"), Sets.newHashSet("index", "quality_uniques", "indexMin", "indexMaxPlusTen"), Arrays.asList(new EventHolder(QueryRunnerTestHelper.segmentId, 0, new ImmutableMap.Builder<String, Object>().put(EventHolder.timestampKey, new DateTime("2011-01-12T00:00:00.000Z")).put("mar", "spot").put("qual", "automotive0").put("place", "preferred").put(QueryRunnerTestHelper.indexMetric, 100.000000F).build()), new EventHolder(QueryRunnerTestHelper.segmentId, 1, new ImmutableMap.Builder<String, Object>().put(EventHolder.timestampKey, new DateTime("2011-01-12T00:00:00.000Z")).put("mar", "spot").put("qual", "business0").put("place", "preferred").put(QueryRunnerTestHelper.indexMetric, 100.000000F).build()), new EventHolder(QueryRunnerTestHelper.segmentId, 2, new ImmutableMap.Builder<String, Object>().put(EventHolder.timestampKey, new DateTime("2011-01-12T00:00:00.000Z")).put("mar", "spot").put("qual", "entertainment0").put("place", "preferred").put(QueryRunnerTestHelper.indexMetric, 100.000000F).build())))));
    List<Result<SelectResultValue>> expectedResultsDsc = Arrays.asList(new Result<SelectResultValue>(new DateTime("2011-01-12T00:00:00.000Z"), new SelectResultValue(ImmutableMap.of(QueryRunnerTestHelper.segmentId, -3), Sets.newHashSet("mar", "qual", "place"), Sets.newHashSet("index", "quality_uniques", "indexMin", "indexMaxPlusTen"), Arrays.asList(new EventHolder(QueryRunnerTestHelper.segmentId, -1, new ImmutableMap.Builder<String, Object>().put(EventHolder.timestampKey, new DateTime("2011-04-15T00:00:00.000Z")).put("mar", "upfront").put("qual", "premium0").put("place", "preferred").put(QueryRunnerTestHelper.indexMetric, 780.27197265625F).build()), new EventHolder(QueryRunnerTestHelper.segmentId, -2, new ImmutableMap.Builder<String, Object>().put(EventHolder.timestampKey, new DateTime("2011-04-15T00:00:00.000Z")).put("mar", "upfront").put("qual", "mezzanine0").put("place", "preferred").put(QueryRunnerTestHelper.indexMetric, 962.731201171875F).build()), new EventHolder(QueryRunnerTestHelper.segmentId, -3, new ImmutableMap.Builder<String, Object>().put(EventHolder.timestampKey, new DateTime("2011-04-15T00:00:00.000Z")).put("mar", "total_market").put("qual", "premium0").put("place", "preferred").put(QueryRunnerTestHelper.indexMetric, 1029.0570068359375F).build())))));
    verify(descending ? expectedResultsDsc : expectedResultsAsc, results);
}

Example 10 with LookupExtractionFn

use of io.druid.query.lookup.LookupExtractionFn in project druid by druid-io.

the class SelectQueryRunnerTest method testSelectWithFilterLookupExtractionFn.

@Test
public void testSelectWithFilterLookupExtractionFn() {
    Map<String, String> extractionMap = new HashMap<>();
    extractionMap.put("total_market", "replaced");
    MapLookupExtractor mapLookupExtractor = new MapLookupExtractor(extractionMap, false);
    LookupExtractionFn lookupExtractionFn = new LookupExtractionFn(mapLookupExtractor, false, null, true, true);
    SelectQuery query = newTestQuery().intervals(I_0112_0114).filters(new SelectorDimFilter(QueryRunnerTestHelper.marketDimension, "replaced", lookupExtractionFn)).granularity(QueryRunnerTestHelper.dayGran).dimensionSpecs(DefaultDimensionSpec.toSpec(QueryRunnerTestHelper.qualityDimension)).metrics(Lists.<String>newArrayList(QueryRunnerTestHelper.indexMetric)).build();
    Iterable<Result<SelectResultValue>> results = Sequences.toList(runner.run(query, Maps.newHashMap()), Lists.<Result<SelectResultValue>>newArrayList());
    Iterable<Result<SelectResultValue>> resultsOptimize = Sequences.toList(toolChest.postMergeQueryDecoration(toolChest.mergeResults(toolChest.preMergeQueryDecoration(runner))).run(query, Maps.<String, Object>newHashMap()), Lists.<Result<SelectResultValue>>newArrayList());
    final List<List<Map<String, Object>>> events = toEvents(new String[] { EventHolder.timestampKey + ":TIME", null, QueryRunnerTestHelper.qualityDimension + ":STRING", null, null, QueryRunnerTestHelper.indexMetric + ":FLOAT" }, // filtered values with day granularity
    new String[] { "2011-01-12T00:00:00.000Z\ttotal_market\tmezzanine\tpreferred\tm
preferred\t1000.000000", "2011-01-12T00:00:00.000Z\ttotal_market\tpremium\tpreferred\tp
preferred\t1000.000000" }, new String[] { "2011-01-13T00:00:00.000Z\ttotal_market\tmezzanine\tpreferred\tm
preferred\t1040.945505", "2011-01-13T00:00:00.000Z\ttotal_market\tpremium\tpreferred\tp
preferred\t1689.012875" });
    PagingOffset offset = query.getPagingOffset(QueryRunnerTestHelper.segmentId);
    List<Result<SelectResultValue>> expectedResults = toExpected(events, Lists.newArrayList(QueryRunnerTestHelper.qualityDimension), Lists.<String>newArrayList(QueryRunnerTestHelper.indexMetric), offset.startOffset(), offset.threshold());
    verify(expectedResults, results);
    verify(expectedResults, resultsOptimize);
}