org.apache.druid.query.groupby.epinephelinae

Class GroupByQueryEngineV2

java.lang.Object

org.apache.druid.query.groupby.epinephelinae.GroupByQueryEngineV2

public class GroupByQueryEngineV2
extends Object

Class that knows how to process a groupBy query on a single StorageAdapter. It returns a Sequence of ResultRow objects that are not guaranteed to be in any particular order, and may not even be fully grouped. It is expected that a downstream GroupByMergingQueryRunnerV2 will finish grouping these results. This code runs on data servers, like Historicals. Used by GroupByStrategyV2.process(GroupByQuery, StorageAdapter).

Method Summary

Modifier and Type	Method and Description
static boolean	canPushDownLimit(ColumnSelectorFactory columnSelectorFactory, String columnName) check if a column will operate correctly with LimitedBufferHashGrouper for query limit pushdown
static void	convertRowTypesToOutputTypes(List<DimensionSpec> dimensionSpecs, ResultRow resultRow, int resultRowDimensionStart)
static int	getCardinalityForArrayAggregation(GroupByQueryConfig querySpecificConfig, GroupByQuery query, StorageAdapter storageAdapter, ByteBuffer buffer) Returns the cardinality of array needed to do array-based aggregation, or -1 if array-based aggregation is impossible.
static boolean	isAllSingleValueDims(Function<String,ColumnCapabilities> capabilitiesFunction, List<DimensionSpec> dimensions, boolean missingMeansNonExistent) Checks whether all "dimensions" are either single-valued, or if allowed, nonexistent.
static Sequence<ResultRow>	process(GroupByQuery query, StorageAdapter storageAdapter, NonBlockingPool<ByteBuffer> intermediateResultsBufferPool, GroupByQueryConfig querySpecificConfig, QueryConfig queryConfig)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Detail

process

public static Sequence<ResultRow> process(GroupByQuery query,
                                          @Nullable
                                          StorageAdapter storageAdapter,
                                          NonBlockingPool<ByteBuffer> intermediateResultsBufferPool,
                                          GroupByQueryConfig querySpecificConfig,
                                          QueryConfig queryConfig)

getCardinalityForArrayAggregation

public static int getCardinalityForArrayAggregation(GroupByQueryConfig querySpecificConfig,
                                                    GroupByQuery query,
                                                    StorageAdapter storageAdapter,
                                                    ByteBuffer buffer)

Returns the cardinality of array needed to do array-based aggregation, or -1 if array-based aggregation is impossible.

isAllSingleValueDims

public static boolean isAllSingleValueDims(Function<String,ColumnCapabilities> capabilitiesFunction,
                                           List<DimensionSpec> dimensions,
                                           boolean missingMeansNonExistent)

Checks whether all "dimensions" are either single-valued, or if allowed, nonexistent. Since non-existent column selectors will show up as full of nulls they are effectively single valued, however they can also be null during broker merge, for example with an 'inline' datasource subquery. 'missingMeansNonExistent' is sort of a hack to let the vectorized engine, which only operates on actual segments, to still work in this case for non-existent columns.

convertRowTypesToOutputTypes

public static void convertRowTypesToOutputTypes(List<DimensionSpec> dimensionSpecs,
                                                ResultRow resultRow,
                                                int resultRowDimensionStart)

canPushDownLimit

public static boolean canPushDownLimit(ColumnSelectorFactory columnSelectorFactory,
                                       String columnName)

check if a column will operate correctly with LimitedBufferHashGrouper for query limit pushdown