1 | #pragma once |
2 | |
3 | #include <DataStreams/IBlockInputStream.h> |
4 | #include <Interpreters/SetVariants.h> |
5 | #include <Core/SortDescription.h> |
6 | |
7 | |
8 | namespace DB |
9 | { |
10 | |
11 | /** This class is intended for implementation of SELECT DISTINCT clause and |
12 | * leaves only unique rows in the stream. |
13 | * |
14 | * Implementation for case, when input stream has rows for same DISTINCT key or at least its prefix, |
15 | * grouped together (going consecutively). |
16 | * |
17 | * To optimize the SELECT DISTINCT ... LIMIT clause we can |
18 | * set limit_hint to non zero value. So we stop emitting new rows after |
19 | * count of already emitted rows will reach the limit_hint. |
20 | */ |
21 | class DistinctSortedBlockInputStream : public IBlockInputStream |
22 | { |
23 | public: |
24 | /// Empty columns_ means all collumns. |
25 | DistinctSortedBlockInputStream(const BlockInputStreamPtr & input, const SizeLimits & set_size_limits_, UInt64 limit_hint_, const Names & columns); |
26 | |
27 | String getName() const override { return "DistinctSorted" ; } |
28 | |
29 | Block () const override { return children.at(0)->getHeader(); } |
30 | |
31 | protected: |
32 | Block readImpl() override; |
33 | |
34 | private: |
35 | ColumnRawPtrs getKeyColumns(const Block & block) const; |
36 | /// When clearing_columns changed, we can clean HashSet to memory optimization |
37 | /// clearing_columns is a left-prefix of SortDescription exists in key_columns |
38 | ColumnRawPtrs getClearingColumns(const Block & block, const ColumnRawPtrs & key_columns) const; |
39 | static bool rowsEqual(const ColumnRawPtrs & lhs, size_t n, const ColumnRawPtrs & rhs, size_t m); |
40 | |
41 | /// return true if has new data |
42 | template <typename Method> |
43 | bool buildFilter( |
44 | Method & method, |
45 | const ColumnRawPtrs & key_columns, |
46 | const ColumnRawPtrs & clearing_hint_columns, |
47 | IColumn::Filter & filter, |
48 | size_t rows, |
49 | ClearableSetVariants & variants) const; |
50 | |
51 | const SortDescription & description; |
52 | |
53 | struct PreviousBlock |
54 | { |
55 | Block block; |
56 | ColumnRawPtrs clearing_hint_columns; |
57 | }; |
58 | PreviousBlock prev_block; |
59 | |
60 | Names columns_names; |
61 | ClearableSetVariants data; |
62 | Sizes key_sizes; |
63 | UInt64 limit_hint; |
64 | |
65 | /// Restrictions on the maximum size of the output data. |
66 | SizeLimits set_size_limits; |
67 | }; |
68 | |
69 | } |
70 | |