| 1 | #include <Processors/Transforms/DistinctTransform.h> |
| 2 | |
| 3 | namespace DB |
| 4 | { |
| 5 | |
| 6 | namespace ErrorCodes |
| 7 | { |
| 8 | extern const int SET_SIZE_LIMIT_EXCEEDED; |
| 9 | } |
| 10 | |
| 11 | DistinctTransform::DistinctTransform( |
| 12 | const Block & , |
| 13 | const SizeLimits & set_size_limits_, |
| 14 | UInt64 limit_hint_, |
| 15 | const Names & columns_) |
| 16 | : ISimpleTransform(header_, header_, true) |
| 17 | , limit_hint(limit_hint_) |
| 18 | , set_size_limits(set_size_limits_) |
| 19 | { |
| 20 | size_t num_columns = columns_.empty() ? header_.columns() : columns_.size(); |
| 21 | |
| 22 | key_columns_pos.reserve(columns_.size()); |
| 23 | for (size_t i = 0; i < num_columns; ++i) |
| 24 | { |
| 25 | auto pos = columns_.empty() ? i |
| 26 | : header_.getPositionByName(columns_[i]); |
| 27 | |
| 28 | auto & col = header_.getByPosition(pos).column; |
| 29 | |
| 30 | if (!(col && isColumnConst(*col))) |
| 31 | key_columns_pos.emplace_back(pos); |
| 32 | } |
| 33 | } |
| 34 | |
| 35 | template <typename Method> |
| 36 | void DistinctTransform::buildFilter( |
| 37 | Method & method, |
| 38 | const ColumnRawPtrs & columns, |
| 39 | IColumn::Filter & filter, |
| 40 | size_t rows, |
| 41 | SetVariants & variants) const |
| 42 | { |
| 43 | typename Method::State state(columns, key_sizes, nullptr); |
| 44 | |
| 45 | for (size_t i = 0; i < rows; ++i) |
| 46 | { |
| 47 | auto emplace_result = state.emplaceKey(method.data, i, variants.string_pool); |
| 48 | |
| 49 | /// Emit the record if there is no such key in the current set yet. |
| 50 | /// Skip it otherwise. |
| 51 | filter[i] = emplace_result.isInserted(); |
| 52 | } |
| 53 | } |
| 54 | |
| 55 | void DistinctTransform::transform(Chunk & chunk) |
| 56 | { |
| 57 | auto num_rows = chunk.getNumRows(); |
| 58 | auto columns = chunk.detachColumns(); |
| 59 | |
| 60 | /// Stop reading if we already reach the limit. |
| 61 | if (no_more_rows || (limit_hint && data.getTotalRowCount() >= limit_hint)) |
| 62 | { |
| 63 | stopReading(); |
| 64 | return; |
| 65 | } |
| 66 | |
| 67 | ColumnRawPtrs column_ptrs; |
| 68 | column_ptrs.reserve(key_columns_pos.size()); |
| 69 | for (auto pos : key_columns_pos) |
| 70 | column_ptrs.emplace_back(columns[pos].get()); |
| 71 | |
| 72 | if (column_ptrs.empty()) |
| 73 | { |
| 74 | /// Only constants. We need to return single row. |
| 75 | no_more_rows = true; |
| 76 | for (auto & column : columns) |
| 77 | column = column->cut(0, 1); |
| 78 | |
| 79 | chunk.setColumns(std::move(columns), 1); |
| 80 | return; |
| 81 | } |
| 82 | |
| 83 | if (data.empty()) |
| 84 | data.init(SetVariants::chooseMethod(column_ptrs, key_sizes)); |
| 85 | |
| 86 | const auto old_set_size = data.getTotalRowCount(); |
| 87 | IColumn::Filter filter(num_rows); |
| 88 | |
| 89 | switch (data.type) |
| 90 | { |
| 91 | case SetVariants::Type::EMPTY: |
| 92 | break; |
| 93 | #define M(NAME) \ |
| 94 | case SetVariants::Type::NAME: \ |
| 95 | buildFilter(*data.NAME, column_ptrs, filter, num_rows, data); \ |
| 96 | break; |
| 97 | APPLY_FOR_SET_VARIANTS(M) |
| 98 | #undef M |
| 99 | } |
| 100 | |
| 101 | /// Just go to the next chunk if there isn't any new record in the current one. |
| 102 | if (data.getTotalRowCount() == old_set_size) |
| 103 | return; |
| 104 | |
| 105 | if (!set_size_limits.check(data.getTotalRowCount(), data.getTotalByteCount(), "DISTINCT" , ErrorCodes::SET_SIZE_LIMIT_EXCEEDED)) |
| 106 | return; |
| 107 | |
| 108 | for (auto & column : columns) |
| 109 | column = column->filter(filter, -1); |
| 110 | |
| 111 | chunk.setColumns(std::move(columns), data.getTotalRowCount() - old_set_size); |
| 112 | } |
| 113 | |
| 114 | } |
| 115 | |