| 1 | //===----------------------------------------------------------------------===// |
| 2 | // DuckDB |
| 3 | // |
| 4 | // duckdb/execution/radix_partitioned_hashtable.hpp |
| 5 | // |
| 6 | // |
| 7 | //===----------------------------------------------------------------------===// |
| 8 | |
| 9 | #pragma once |
| 10 | |
| 11 | #include "duckdb/execution/partitionable_hashtable.hpp" |
| 12 | #include "duckdb/parser/group_by_node.hpp" |
| 13 | #include "duckdb/execution/physical_operator.hpp" |
| 14 | #include "duckdb/execution/operator/aggregate/grouped_aggregate_data.hpp" |
| 15 | |
| 16 | namespace duckdb { |
| 17 | class BufferManager; |
| 18 | class Executor; |
| 19 | class PhysicalHashAggregate; |
| 20 | class Pipeline; |
| 21 | class Task; |
| 22 | |
| 23 | class RadixPartitionedHashTable { |
| 24 | public: |
| 25 | RadixPartitionedHashTable(GroupingSet &grouping_set, const GroupedAggregateData &op); |
| 26 | |
| 27 | GroupingSet &grouping_set; |
| 28 | //! The indices specified in the groups_count that do not appear in the grouping_set |
| 29 | unsafe_vector<idx_t> null_groups; |
| 30 | const GroupedAggregateData &op; |
| 31 | |
| 32 | vector<LogicalType> group_types; |
| 33 | //! how many groups can we have in the operator before we switch to radix partitioning |
| 34 | idx_t radix_limit; |
| 35 | |
| 36 | //! The GROUPING values that belong to this hash table |
| 37 | vector<Value> grouping_values; |
| 38 | |
| 39 | public: |
| 40 | //! Sink Interface |
| 41 | unique_ptr<GlobalSinkState> GetGlobalSinkState(ClientContext &context) const; |
| 42 | unique_ptr<LocalSinkState> GetLocalSinkState(ExecutionContext &context) const; |
| 43 | |
| 44 | void Sink(ExecutionContext &context, DataChunk &chunk, OperatorSinkInput &input, DataChunk &aggregate_input_chunk, |
| 45 | const unsafe_vector<idx_t> &filter) const; |
| 46 | void Combine(ExecutionContext &context, GlobalSinkState &state, LocalSinkState &lstate) const; |
| 47 | bool Finalize(ClientContext &context, GlobalSinkState &gstate_p) const; |
| 48 | |
| 49 | void ScheduleTasks(Executor &executor, const shared_ptr<Event> &event, GlobalSinkState &state, |
| 50 | vector<shared_ptr<Task>> &tasks) const; |
| 51 | |
| 52 | //! Source interface |
| 53 | idx_t Size(GlobalSinkState &sink_state) const; |
| 54 | unique_ptr<GlobalSourceState> GetGlobalSourceState(ClientContext &context) const; |
| 55 | unique_ptr<LocalSourceState> GetLocalSourceState(ExecutionContext &context) const; |
| 56 | SourceResultType GetData(ExecutionContext &context, DataChunk &chunk, GlobalSinkState &sink_state, |
| 57 | OperatorSourceInput &input) const; |
| 58 | |
| 59 | static void SetMultiScan(GlobalSinkState &state); |
| 60 | bool ForceSingleHT(GlobalSinkState &state) const; |
| 61 | |
| 62 | private: |
| 63 | void SetGroupingValues(); |
| 64 | void PopulateGroupChunk(DataChunk &group_chunk, DataChunk &input_chunk) const; |
| 65 | }; |
| 66 | |
| 67 | } // namespace duckdb |
| 68 | |