use of org.apache.druid.query.FinalizeResultsQueryRunner in project druid by druid-io.
the class CachingClusteredClientTest method testGroupByCaching.
@Test
public void testGroupByCaching() {
List<AggregatorFactory> aggsWithUniques = ImmutableList.<AggregatorFactory>builder().addAll(AGGS).add(new HyperUniquesAggregatorFactory("uniques", "uniques")).build();
final HashFunction hashFn = Hashing.murmur3_128();
GroupByQuery.Builder builder = new GroupByQuery.Builder().setDataSource(DATA_SOURCE).setQuerySegmentSpec(SEG_SPEC).setDimFilter(DIM_FILTER).setGranularity(GRANULARITY).setDimensions(new DefaultDimensionSpec("a", "a")).setAggregatorSpecs(aggsWithUniques).setPostAggregatorSpecs(POST_AGGS).setContext(CONTEXT);
final HyperLogLogCollector collector = HyperLogLogCollector.makeLatestCollector();
collector.add(hashFn.hashString("abc123", StandardCharsets.UTF_8).asBytes());
collector.add(hashFn.hashString("123abc", StandardCharsets.UTF_8).asBytes());
final GroupByQuery query = builder.randomQueryId().build();
testQueryCaching(getDefaultQueryRunner(), query, Intervals.of("2011-01-01/2011-01-02"), makeGroupByResults(query, DateTimes.of("2011-01-01"), ImmutableMap.of("a", "a", "rows", 1, "imps", 1, "impers", 1, "uniques", collector)), Intervals.of("2011-01-02/2011-01-03"), makeGroupByResults(query, DateTimes.of("2011-01-02"), ImmutableMap.of("a", "b", "rows", 2, "imps", 2, "impers", 2, "uniques", collector)), Intervals.of("2011-01-05/2011-01-10"), makeGroupByResults(query, DateTimes.of("2011-01-05"), ImmutableMap.of("a", "c", "rows", 3, "imps", 3, "impers", 3, "uniques", collector), DateTimes.of("2011-01-06"), ImmutableMap.of("a", "d", "rows", 4, "imps", 4, "impers", 4, "uniques", collector), DateTimes.of("2011-01-07"), ImmutableMap.of("a", "e", "rows", 5, "imps", 5, "impers", 5, "uniques", collector), DateTimes.of("2011-01-08"), ImmutableMap.of("a", "f", "rows", 6, "imps", 6, "impers", 6, "uniques", collector), DateTimes.of("2011-01-09"), ImmutableMap.of("a", "g", "rows", 7, "imps", 7, "impers", 7, "uniques", collector)), Intervals.of("2011-01-05/2011-01-10"), makeGroupByResults(query, DateTimes.of("2011-01-05T01"), ImmutableMap.of("a", "c", "rows", 3, "imps", 3, "impers", 3, "uniques", collector), DateTimes.of("2011-01-06T01"), ImmutableMap.of("a", "d", "rows", 4, "imps", 4, "impers", 4, "uniques", collector), DateTimes.of("2011-01-07T01"), ImmutableMap.of("a", "e", "rows", 5, "imps", 5, "impers", 5, "uniques", collector), DateTimes.of("2011-01-08T01"), ImmutableMap.of("a", "f", "rows", 6, "imps", 6, "impers", 6, "uniques", collector), DateTimes.of("2011-01-09T01"), ImmutableMap.of("a", "g", "rows", 7, "imps", 7, "impers", 7, "uniques", collector)));
QueryRunner runner = new FinalizeResultsQueryRunner(getDefaultQueryRunner(), WAREHOUSE.getToolChest(query));
TestHelper.assertExpectedObjects(makeGroupByResults(query, DateTimes.of("2011-01-05T"), ImmutableMap.of("a", "c", "rows", 3, "imps", 3, "impers", 3, "uniques", collector), DateTimes.of("2011-01-05T01"), ImmutableMap.of("a", "c", "rows", 3, "imps", 3, "impers", 3, "uniques", collector), DateTimes.of("2011-01-06T"), ImmutableMap.of("a", "d", "rows", 4, "imps", 4, "impers", 4, "uniques", collector), DateTimes.of("2011-01-06T01"), ImmutableMap.of("a", "d", "rows", 4, "imps", 4, "impers", 4, "uniques", collector), DateTimes.of("2011-01-07T"), ImmutableMap.of("a", "e", "rows", 5, "imps", 5, "impers", 5, "uniques", collector), DateTimes.of("2011-01-07T01"), ImmutableMap.of("a", "e", "rows", 5, "imps", 5, "impers", 5, "uniques", collector), DateTimes.of("2011-01-08T"), ImmutableMap.of("a", "f", "rows", 6, "imps", 6, "impers", 6, "uniques", collector), DateTimes.of("2011-01-08T01"), ImmutableMap.of("a", "f", "rows", 6, "imps", 6, "impers", 6, "uniques", collector), DateTimes.of("2011-01-09T"), ImmutableMap.of("a", "g", "rows", 7, "imps", 7, "impers", 7, "uniques", collector), DateTimes.of("2011-01-09T01"), ImmutableMap.of("a", "g", "rows", 7, "imps", 7, "impers", 7, "uniques", collector)), runner.run(QueryPlus.wrap(builder.randomQueryId().setInterval("2011-01-05/2011-01-10").build())), "");
}
use of org.apache.druid.query.FinalizeResultsQueryRunner in project druid by druid-io.
the class CachingClusteredClientTest method testTopNOnPostAggMetricCaching.
@Test
public void testTopNOnPostAggMetricCaching() {
final TopNQueryBuilder builder = new TopNQueryBuilder().dataSource(DATA_SOURCE).dimension(TOP_DIM).metric("avg_imps_per_row_double").threshold(3).intervals(SEG_SPEC).filters(DIM_FILTER).granularity(GRANULARITY).aggregators(AGGS).postAggregators(POST_AGGS).context(CONTEXT);
QueryRunner runner = new FinalizeResultsQueryRunner(getDefaultQueryRunner(), new TopNQueryQueryToolChest(new TopNQueryConfig()));
testQueryCaching(runner, builder.randomQueryId().build(), Intervals.of("2011-01-01/2011-01-02"), makeTopNResultsWithoutRename(), Intervals.of("2011-01-02/2011-01-03"), makeTopNResultsWithoutRename(), Intervals.of("2011-01-05/2011-01-10"), makeTopNResultsWithoutRename(DateTimes.of("2011-01-05"), "a", 50, 4994, "b", 50, 4993, "c", 50, 4992, DateTimes.of("2011-01-06"), "a", 50, 4991, "b", 50, 4990, "c", 50, 4989, DateTimes.of("2011-01-07"), "a", 50, 4991, "b", 50, 4990, "c", 50, 4989, DateTimes.of("2011-01-08"), "a", 50, 4988, "b", 50, 4987, "c", 50, 4986, DateTimes.of("2011-01-09"), "c1", 50, 4985, "b", 50, 4984, "c", 50, 4983), Intervals.of("2011-01-05/2011-01-10"), makeTopNResultsWithoutRename(DateTimes.of("2011-01-05T01"), "a", 50, 4994, "b", 50, 4993, "c", 50, 4992, DateTimes.of("2011-01-06T01"), "a", 50, 4991, "b", 50, 4990, "c", 50, 4989, DateTimes.of("2011-01-07T01"), "a", 50, 4991, "b", 50, 4990, "c", 50, 4989, DateTimes.of("2011-01-08T01"), "a", 50, 4988, "b", 50, 4987, "c", 50, 4986, DateTimes.of("2011-01-09T01"), "c2", 50, 4985, "b", 50, 4984, "c", 50, 4983));
TopNQuery query = builder.intervals("2011-01-01/2011-01-10").metric("avg_imps_per_row_double").aggregators(AGGS).postAggregators(DIFF_ORDER_POST_AGGS).randomQueryId().build();
TestHelper.assertExpectedResults(makeTopNResultsWithoutRename(DateTimes.of("2011-01-05"), "a", 50, 4994, "b", 50, 4993, "c", 50, 4992, DateTimes.of("2011-01-05T01"), "a", 50, 4994, "b", 50, 4993, "c", 50, 4992, DateTimes.of("2011-01-06"), "a", 50, 4991, "b", 50, 4990, "c", 50, 4989, DateTimes.of("2011-01-06T01"), "a", 50, 4991, "b", 50, 4990, "c", 50, 4989, DateTimes.of("2011-01-07"), "a", 50, 4991, "b", 50, 4990, "c", 50, 4989, DateTimes.of("2011-01-07T01"), "a", 50, 4991, "b", 50, 4990, "c", 50, 4989, DateTimes.of("2011-01-08"), "a", 50, 4988, "b", 50, 4987, "c", 50, 4986, DateTimes.of("2011-01-08T01"), "a", 50, 4988, "b", 50, 4987, "c", 50, 4986, DateTimes.of("2011-01-09"), "c1", 50, 4985, "b", 50, 4984, "c", 50, 4983, DateTimes.of("2011-01-09T01"), "c2", 50, 4985, "b", 50, 4984, "c", 50, 4983), runner.run(QueryPlus.wrap(query)));
}
use of org.apache.druid.query.FinalizeResultsQueryRunner in project druid by druid-io.
the class CachingClusteredClientTest method testSearchCaching.
@Test
public void testSearchCaching() {
final Druids.SearchQueryBuilder builder = Druids.newSearchQueryBuilder().dataSource(DATA_SOURCE).filters(DIM_FILTER).granularity(GRANULARITY).limit(1000).intervals(SEG_SPEC).dimensions(Collections.singletonList(TOP_DIM)).query("how").context(CONTEXT);
testQueryCaching(getDefaultQueryRunner(), builder.randomQueryId().build(), Intervals.of("2011-01-01/2011-01-02"), makeSearchResults(TOP_DIM, DateTimes.of("2011-01-01"), "how", 1, "howdy", 2, "howwwwww", 3, "howwy", 4), Intervals.of("2011-01-02/2011-01-03"), makeSearchResults(TOP_DIM, DateTimes.of("2011-01-02"), "how1", 1, "howdy1", 2, "howwwwww1", 3, "howwy1", 4), Intervals.of("2011-01-05/2011-01-10"), makeSearchResults(TOP_DIM, DateTimes.of("2011-01-05"), "how2", 1, "howdy2", 2, "howwwwww2", 3, "howww2", 4, DateTimes.of("2011-01-06"), "how3", 1, "howdy3", 2, "howwwwww3", 3, "howww3", 4, DateTimes.of("2011-01-07"), "how4", 1, "howdy4", 2, "howwwwww4", 3, "howww4", 4, DateTimes.of("2011-01-08"), "how5", 1, "howdy5", 2, "howwwwww5", 3, "howww5", 4, DateTimes.of("2011-01-09"), "how6", 1, "howdy6", 2, "howwwwww6", 3, "howww6", 4), Intervals.of("2011-01-05/2011-01-10"), makeSearchResults(TOP_DIM, DateTimes.of("2011-01-05T01"), "how2", 1, "howdy2", 2, "howwwwww2", 3, "howww2", 4, DateTimes.of("2011-01-06T01"), "how3", 1, "howdy3", 2, "howwwwww3", 3, "howww3", 4, DateTimes.of("2011-01-07T01"), "how4", 1, "howdy4", 2, "howwwwww4", 3, "howww4", 4, DateTimes.of("2011-01-08T01"), "how5", 1, "howdy5", 2, "howwwwww5", 3, "howww5", 4, DateTimes.of("2011-01-09T01"), "how6", 1, "howdy6", 2, "howwwwww6", 3, "howww6", 4));
QueryRunner runner = new FinalizeResultsQueryRunner(getDefaultQueryRunner(), new SearchQueryQueryToolChest(new SearchQueryConfig()));
TestHelper.assertExpectedResults(makeSearchResults(TOP_DIM, DateTimes.of("2011-01-01"), "how", 1, "howdy", 2, "howwwwww", 3, "howwy", 4, DateTimes.of("2011-01-02"), "how1", 1, "howdy1", 2, "howwwwww1", 3, "howwy1", 4, DateTimes.of("2011-01-05"), "how2", 1, "howdy2", 2, "howwwwww2", 3, "howww2", 4, DateTimes.of("2011-01-05T01"), "how2", 1, "howdy2", 2, "howwwwww2", 3, "howww2", 4, DateTimes.of("2011-01-06"), "how3", 1, "howdy3", 2, "howwwwww3", 3, "howww3", 4, DateTimes.of("2011-01-06T01"), "how3", 1, "howdy3", 2, "howwwwww3", 3, "howww3", 4, DateTimes.of("2011-01-07"), "how4", 1, "howdy4", 2, "howwwwww4", 3, "howww4", 4, DateTimes.of("2011-01-07T01"), "how4", 1, "howdy4", 2, "howwwwww4", 3, "howww4", 4, DateTimes.of("2011-01-08"), "how5", 1, "howdy5", 2, "howwwwww5", 3, "howww5", 4, DateTimes.of("2011-01-08T01"), "how5", 1, "howdy5", 2, "howwwwww5", 3, "howww5", 4, DateTimes.of("2011-01-09"), "how6", 1, "howdy6", 2, "howwwwww6", 3, "howww6", 4, DateTimes.of("2011-01-09T01"), "how6", 1, "howdy6", 2, "howwwwww6", 3, "howww6", 4), runner.run(QueryPlus.wrap(builder.randomQueryId().intervals("2011-01-01/2011-01-10").build())));
}
use of org.apache.druid.query.FinalizeResultsQueryRunner in project druid by druid-io.
the class TestClusterQuerySegmentWalker method getQueryRunnerForSegments.
@Override
public <T> QueryRunner<T> getQueryRunnerForSegments(final Query<T> query, final Iterable<SegmentDescriptor> specs) {
final QueryRunnerFactory<T, Query<T>> factory = conglomerate.findFactory(query);
if (factory == null) {
throw new ISE("Unknown query type[%s].", query.getClass());
}
final DataSourceAnalysis analysis = DataSourceAnalysis.forDataSource(query.getDataSource());
if (!analysis.isConcreteTableBased()) {
throw new ISE("Cannot handle datasource: %s", query.getDataSource());
}
final String dataSourceName = ((TableDataSource) analysis.getBaseDataSource()).getName();
final QueryToolChest<T, Query<T>> toolChest = factory.getToolchest();
// Make sure this query type can handle the subquery, if present.
if (analysis.isQuery() && !toolChest.canPerformSubquery(((QueryDataSource) analysis.getDataSource()).getQuery())) {
throw new ISE("Cannot handle subquery: %s", analysis.getDataSource());
}
final Function<SegmentReference, SegmentReference> segmentMapFn = joinableFactoryWrapper.createSegmentMapFn(analysis.getJoinBaseTableFilter().map(Filters::toFilter).orElse(null), analysis.getPreJoinableClauses(), new AtomicLong(), analysis.getBaseQuery().orElse(query));
final QueryRunner<T> baseRunner = new FinalizeResultsQueryRunner<>(toolChest.postMergeQueryDecoration(toolChest.mergeResults(toolChest.preMergeQueryDecoration(makeTableRunner(toolChest, factory, getSegmentsForTable(dataSourceName, specs), segmentMapFn)))), toolChest);
// to actually serve the queries
return (theQuery, responseContext) -> {
responseContext.initializeRemainingResponses();
responseContext.addRemainingResponse(theQuery.getQuery().getMostSpecificId(), 0);
if (scheduler != null) {
Set<SegmentServerSelector> segments = new HashSet<>();
specs.forEach(spec -> segments.add(new SegmentServerSelector(spec)));
return scheduler.run(scheduler.prioritizeAndLaneQuery(theQuery, segments), new LazySequence<>(() -> baseRunner.run(theQuery.withQuery(Queries.withSpecificSegments(theQuery.getQuery(), ImmutableList.copyOf(specs))), responseContext)));
} else {
return baseRunner.run(theQuery.withQuery(Queries.withSpecificSegments(theQuery.getQuery(), ImmutableList.copyOf(specs))), responseContext);
}
};
}
Aggregations