| 1 | #pragma once |
| 2 | |
| 3 | #include <Core/QueryProcessingStage.h> |
| 4 | #include <Storages/SelectQueryInfo.h> |
| 5 | #include <Storages/MergeTree/MergeTreeData.h> |
| 6 | #include <Storages/MergeTree/RangesInDataPart.h> |
| 7 | |
| 8 | |
| 9 | namespace DB |
| 10 | { |
| 11 | |
| 12 | class KeyCondition; |
| 13 | |
| 14 | |
| 15 | /** Executes SELECT queries on data from the merge tree. |
| 16 | */ |
| 17 | class MergeTreeDataSelectExecutor |
| 18 | { |
| 19 | public: |
| 20 | explicit MergeTreeDataSelectExecutor(const MergeTreeData & data_); |
| 21 | |
| 22 | /** When reading, selects a set of parts that covers the desired range of the index. |
| 23 | * max_blocks_number_to_read - if not nullptr, do not read all the parts whose right border is greater than max_block in partition. |
| 24 | */ |
| 25 | using PartitionIdToMaxBlock = std::unordered_map<String, Int64>; |
| 26 | |
| 27 | Pipes read( |
| 28 | const Names & column_names, |
| 29 | const SelectQueryInfo & query_info, |
| 30 | const Context & context, |
| 31 | UInt64 max_block_size, |
| 32 | unsigned num_streams, |
| 33 | const PartitionIdToMaxBlock * max_block_numbers_to_read = nullptr) const; |
| 34 | |
| 35 | Pipes readFromParts( |
| 36 | MergeTreeData::DataPartsVector parts, |
| 37 | const Names & column_names, |
| 38 | const SelectQueryInfo & query_info, |
| 39 | const Context & context, |
| 40 | UInt64 max_block_size, |
| 41 | unsigned num_streams, |
| 42 | const PartitionIdToMaxBlock * max_block_numbers_to_read = nullptr) const; |
| 43 | |
| 44 | private: |
| 45 | const MergeTreeData & data; |
| 46 | |
| 47 | Logger * log; |
| 48 | |
| 49 | Pipes spreadMarkRangesAmongStreams( |
| 50 | RangesInDataParts && parts, |
| 51 | size_t num_streams, |
| 52 | const Names & column_names, |
| 53 | UInt64 max_block_size, |
| 54 | bool use_uncompressed_cache, |
| 55 | const SelectQueryInfo & query_info, |
| 56 | const Names & virt_columns, |
| 57 | const Settings & settings) const; |
| 58 | |
| 59 | Pipes spreadMarkRangesAmongStreamsWithOrder( |
| 60 | RangesInDataParts && parts, |
| 61 | size_t num_streams, |
| 62 | const Names & column_names, |
| 63 | UInt64 max_block_size, |
| 64 | bool use_uncompressed_cache, |
| 65 | const SelectQueryInfo & query_info, |
| 66 | const ExpressionActionsPtr & sorting_key_prefix_expr, |
| 67 | const Names & virt_columns, |
| 68 | const Settings & settings) const; |
| 69 | |
| 70 | Pipes spreadMarkRangesAmongStreamsFinal( |
| 71 | RangesInDataParts && parts, |
| 72 | const Names & column_names, |
| 73 | UInt64 max_block_size, |
| 74 | bool use_uncompressed_cache, |
| 75 | const SelectQueryInfo & query_info, |
| 76 | const Names & virt_columns, |
| 77 | const Settings & settings) const; |
| 78 | |
| 79 | /// Get the approximate value (bottom estimate - only by full marks) of the number of rows falling under the index. |
| 80 | size_t getApproximateTotalRowsToRead( |
| 81 | const MergeTreeData::DataPartsVector & parts, |
| 82 | const KeyCondition & key_condition, |
| 83 | const Settings & settings) const; |
| 84 | |
| 85 | /// Create the expression "Sign == 1". |
| 86 | void createPositiveSignCondition( |
| 87 | ExpressionActionsPtr & out_expression, |
| 88 | String & out_column, |
| 89 | const Context & context) const; |
| 90 | |
| 91 | MarkRanges markRangesFromPKRange( |
| 92 | const MergeTreeData::DataPartPtr & part, |
| 93 | const KeyCondition & key_condition, |
| 94 | const Settings & settings) const; |
| 95 | |
| 96 | MarkRanges filterMarksUsingIndex( |
| 97 | MergeTreeIndexPtr index, |
| 98 | MergeTreeIndexConditionPtr condition, |
| 99 | MergeTreeData::DataPartPtr part, |
| 100 | const MarkRanges & ranges, |
| 101 | const Settings & settings) const; |
| 102 | }; |
| 103 | |
| 104 | } |
| 105 | |