| 1 | #pragma once |
| 2 | |
| 3 | #include <Core/Names.h> |
| 4 | #include <Core/NamesAndTypes.h> |
| 5 | #include <Parsers/ASTTablesInSelectQuery.h> |
| 6 | #include <Interpreters/IJoin.h> |
| 7 | #include <Interpreters/asof.h> |
| 8 | #include <DataStreams/IBlockStream_fwd.h> |
| 9 | #include <DataStreams/SizeLimits.h> |
| 10 | |
| 11 | #include <utility> |
| 12 | #include <memory> |
| 13 | |
| 14 | namespace DB |
| 15 | { |
| 16 | |
| 17 | class Context; |
| 18 | class ASTSelectQuery; |
| 19 | struct DatabaseAndTableWithAlias; |
| 20 | class Block; |
| 21 | |
| 22 | struct Settings; |
| 23 | |
| 24 | class AnalyzedJoin |
| 25 | { |
| 26 | /** Query of the form `SELECT expr(x) AS k FROM t1 ANY LEFT JOIN (SELECT expr(x) AS k FROM t2) USING k` |
| 27 | * The join is made by column k. |
| 28 | * During the JOIN, |
| 29 | * - in the "right" table, it will be available by alias `k`, since `Project` action for the subquery was executed. |
| 30 | * - in the "left" table, it will be accessible by the name `expr(x)`, since `Project` action has not been executed yet. |
| 31 | * You must remember both of these options. |
| 32 | * |
| 33 | * Query of the form `SELECT ... from t1 ANY LEFT JOIN (SELECT ... from t2) ON expr(t1 columns) = expr(t2 columns)` |
| 34 | * to the subquery will be added expression `expr(t2 columns)`. |
| 35 | * It's possible to use name `expr(t2 columns)`. |
| 36 | */ |
| 37 | |
| 38 | friend class SyntaxAnalyzer; |
| 39 | |
| 40 | const SizeLimits size_limits; |
| 41 | const size_t default_max_bytes; |
| 42 | const bool join_use_nulls; |
| 43 | const bool partial_merge_join = false; |
| 44 | const bool partial_merge_join_optimizations = false; |
| 45 | const size_t partial_merge_join_rows_in_right_blocks = 0; |
| 46 | |
| 47 | Names key_names_left; |
| 48 | Names key_names_right; /// Duplicating names are qualified. |
| 49 | ASTs key_asts_left; |
| 50 | ASTs key_asts_right; |
| 51 | ASTTableJoin table_join; |
| 52 | ASOF::Inequality asof_inequality = ASOF::Inequality::GreaterOrEquals; |
| 53 | |
| 54 | /// All columns which can be read from joined table. Duplicating names are qualified. |
| 55 | NamesAndTypesList columns_from_joined_table; |
| 56 | /// Columns will be added to block by JOIN. It's a subset of columns_from_joined_table with corrected Nullability |
| 57 | NamesAndTypesList columns_added_by_join; |
| 58 | |
| 59 | /// Name -> original name. Names are the same as in columns_from_joined_table list. |
| 60 | std::unordered_map<String, String> original_names; |
| 61 | /// Original name -> name. Only ranamed columns. |
| 62 | std::unordered_map<String, String> renames; |
| 63 | |
| 64 | String tmp_path; |
| 65 | |
| 66 | public: |
| 67 | AnalyzedJoin(const Settings &, const String & tmp_path); |
| 68 | |
| 69 | /// for StorageJoin |
| 70 | AnalyzedJoin(SizeLimits limits, bool use_nulls, ASTTableJoin::Kind kind, ASTTableJoin::Strictness strictness, |
| 71 | const Names & key_names_right_) |
| 72 | : size_limits(limits) |
| 73 | , default_max_bytes(0) |
| 74 | , join_use_nulls(use_nulls) |
| 75 | , key_names_right(key_names_right_) |
| 76 | { |
| 77 | table_join.kind = kind; |
| 78 | table_join.strictness = strictness; |
| 79 | } |
| 80 | |
| 81 | ASTTableJoin::Kind kind() const { return table_join.kind; } |
| 82 | ASTTableJoin::Strictness strictness() const { return table_join.strictness; } |
| 83 | const SizeLimits & sizeLimits() const { return size_limits; } |
| 84 | const String & getTemporaryPath() const { return tmp_path; } |
| 85 | |
| 86 | bool forceNullableRight() const { return join_use_nulls && isLeftOrFull(table_join.kind); } |
| 87 | bool forceNullableLeft() const { return join_use_nulls && isRightOrFull(table_join.kind); } |
| 88 | size_t defaultMaxBytes() const { return default_max_bytes; } |
| 89 | size_t maxRowsInRightBlock() const { return partial_merge_join_rows_in_right_blocks; } |
| 90 | bool enablePartialMergeJoinOptimizations() const { return partial_merge_join_optimizations; } |
| 91 | |
| 92 | void addUsingKey(const ASTPtr & ast); |
| 93 | void addOnKeys(ASTPtr & left_table_ast, ASTPtr & right_table_ast); |
| 94 | |
| 95 | bool hasUsing() const { return table_join.using_expression_list != nullptr; } |
| 96 | bool hasOn() const { return table_join.on_expression != nullptr; } |
| 97 | |
| 98 | NameSet getQualifiedColumnsSet() const; |
| 99 | NamesWithAliases getNamesWithAliases(const NameSet & required_columns) const; |
| 100 | NamesWithAliases getRequiredColumns(const Block & sample, const Names & action_columns) const; |
| 101 | |
| 102 | void deduplicateAndQualifyColumnNames(const NameSet & left_table_columns, const String & right_table_prefix); |
| 103 | size_t rightKeyInclusion(const String & name) const; |
| 104 | NameSet requiredRightKeys() const; |
| 105 | |
| 106 | void addJoinedColumn(const NameAndTypePair & joined_column); |
| 107 | void addJoinedColumnsAndCorrectNullability(Block & sample_block) const; |
| 108 | |
| 109 | void setAsofInequality(ASOF::Inequality inequality) { asof_inequality = inequality; } |
| 110 | ASOF::Inequality getAsofInequality() { return asof_inequality; } |
| 111 | |
| 112 | ASTPtr leftKeysList() const; |
| 113 | ASTPtr rightKeysList() const; /// For ON syntax only |
| 114 | |
| 115 | Names requiredJoinedNames() const; |
| 116 | const Names & keyNamesLeft() const { return key_names_left; } |
| 117 | const Names & keyNamesRight() const { return key_names_right; } |
| 118 | const NamesAndTypesList & columnsFromJoinedTable() const { return columns_from_joined_table; } |
| 119 | const NamesAndTypesList & columnsAddedByJoin() const { return columns_added_by_join; } |
| 120 | |
| 121 | /// StorageJoin overrides key names (cause of different names qualification) |
| 122 | void setRightKeys(const Names & keys) { key_names_right = keys; } |
| 123 | |
| 124 | static bool sameJoin(const AnalyzedJoin * x, const AnalyzedJoin * y); |
| 125 | friend JoinPtr makeJoin(std::shared_ptr<AnalyzedJoin> table_join, const Block & right_sample_block); |
| 126 | }; |
| 127 | |
| 128 | bool isMergeJoin(const JoinPtr &); |
| 129 | |
| 130 | } |
| 131 | |