| 1 | #include <DataStreams/DistinctSortedBlockInputStream.h> |
| 2 | |
| 3 | namespace DB |
| 4 | { |
| 5 | |
| 6 | namespace ErrorCodes |
| 7 | { |
| 8 | extern const int SET_SIZE_LIMIT_EXCEEDED; |
| 9 | } |
| 10 | |
| 11 | DistinctSortedBlockInputStream::DistinctSortedBlockInputStream( |
| 12 | const BlockInputStreamPtr & input, const SizeLimits & set_size_limits_, UInt64 limit_hint_, const Names & columns) |
| 13 | : description(input->getSortDescription()) |
| 14 | , columns_names(columns) |
| 15 | , limit_hint(limit_hint_) |
| 16 | , set_size_limits(set_size_limits_) |
| 17 | { |
| 18 | children.push_back(input); |
| 19 | } |
| 20 | |
| 21 | Block DistinctSortedBlockInputStream::readImpl() |
| 22 | { |
| 23 | /// Execute until end of stream or until |
| 24 | /// a block with some new records will be gotten. |
| 25 | for (;;) |
| 26 | { |
| 27 | /// Stop reading if we already reached the limit. |
| 28 | if (limit_hint && data.getTotalRowCount() >= limit_hint) |
| 29 | return Block(); |
| 30 | |
| 31 | Block block = children.back()->read(); |
| 32 | if (!block) |
| 33 | return Block(); |
| 34 | |
| 35 | const ColumnRawPtrs column_ptrs(getKeyColumns(block)); |
| 36 | if (column_ptrs.empty()) |
| 37 | return block; |
| 38 | |
| 39 | const ColumnRawPtrs clearing_hint_columns(getClearingColumns(block, column_ptrs)); |
| 40 | |
| 41 | if (data.type == ClearableSetVariants::Type::EMPTY) |
| 42 | data.init(ClearableSetVariants::chooseMethod(column_ptrs, key_sizes)); |
| 43 | |
| 44 | const size_t rows = block.rows(); |
| 45 | IColumn::Filter filter(rows); |
| 46 | |
| 47 | bool has_new_data = false; |
| 48 | switch (data.type) |
| 49 | { |
| 50 | case ClearableSetVariants::Type::EMPTY: |
| 51 | break; |
| 52 | #define M(NAME) \ |
| 53 | case ClearableSetVariants::Type::NAME: \ |
| 54 | has_new_data = buildFilter(*data.NAME, column_ptrs, clearing_hint_columns, filter, rows, data); \ |
| 55 | break; |
| 56 | APPLY_FOR_SET_VARIANTS(M) |
| 57 | #undef M |
| 58 | } |
| 59 | |
| 60 | /// Just go to the next block if there isn't any new record in the current one. |
| 61 | if (!has_new_data) |
| 62 | continue; |
| 63 | |
| 64 | if (!set_size_limits.check(data.getTotalRowCount(), data.getTotalByteCount(), "DISTINCT" , ErrorCodes::SET_SIZE_LIMIT_EXCEEDED)) |
| 65 | return {}; |
| 66 | |
| 67 | prev_block.block = block; |
| 68 | prev_block.clearing_hint_columns = std::move(clearing_hint_columns); |
| 69 | |
| 70 | size_t all_columns = block.columns(); |
| 71 | for (size_t i = 0; i < all_columns; ++i) |
| 72 | block.safeGetByPosition(i).column = block.safeGetByPosition(i).column->filter(filter, -1); |
| 73 | |
| 74 | return block; |
| 75 | } |
| 76 | } |
| 77 | |
| 78 | |
| 79 | template <typename Method> |
| 80 | bool DistinctSortedBlockInputStream::buildFilter( |
| 81 | Method & method, |
| 82 | const ColumnRawPtrs & columns, |
| 83 | const ColumnRawPtrs & clearing_hint_columns, |
| 84 | IColumn::Filter & filter, |
| 85 | size_t rows, |
| 86 | ClearableSetVariants & variants) const |
| 87 | { |
| 88 | typename Method::State state(columns, key_sizes, nullptr); |
| 89 | |
| 90 | /// Compare last row of previous block and first row of current block, |
| 91 | /// If rows not equal, we can clear HashSet, |
| 92 | /// If clearing_hint_columns is empty, we CAN'T clear HashSet. |
| 93 | if (!clearing_hint_columns.empty() && !prev_block.clearing_hint_columns.empty() |
| 94 | && !rowsEqual(clearing_hint_columns, 0, prev_block.clearing_hint_columns, prev_block.block.rows() - 1)) |
| 95 | { |
| 96 | method.data.clear(); |
| 97 | } |
| 98 | |
| 99 | bool has_new_data = false; |
| 100 | for (size_t i = 0; i < rows; ++i) |
| 101 | { |
| 102 | /// Compare i-th row and i-1-th row, |
| 103 | /// If rows are not equal, we can clear HashSet, |
| 104 | /// If clearing_hint_columns is empty, we CAN'T clear HashSet. |
| 105 | if (i > 0 && !clearing_hint_columns.empty() && !rowsEqual(clearing_hint_columns, i, clearing_hint_columns, i - 1)) |
| 106 | method.data.clear(); |
| 107 | |
| 108 | auto emplace_result = state.emplaceKey(method.data, i, variants.string_pool); |
| 109 | |
| 110 | if (emplace_result.isInserted()) |
| 111 | has_new_data = true; |
| 112 | |
| 113 | /// Emit the record if there is no such key in the current set yet. |
| 114 | /// Skip it otherwise. |
| 115 | filter[i] = emplace_result.isInserted(); |
| 116 | } |
| 117 | return has_new_data; |
| 118 | } |
| 119 | |
| 120 | ColumnRawPtrs DistinctSortedBlockInputStream::getKeyColumns(const Block & block) const |
| 121 | { |
| 122 | size_t columns = columns_names.empty() ? block.columns() : columns_names.size(); |
| 123 | |
| 124 | ColumnRawPtrs column_ptrs; |
| 125 | column_ptrs.reserve(columns); |
| 126 | |
| 127 | for (size_t i = 0; i < columns; ++i) |
| 128 | { |
| 129 | auto & column = columns_names.empty() |
| 130 | ? block.safeGetByPosition(i).column |
| 131 | : block.getByName(columns_names[i]).column; |
| 132 | |
| 133 | /// Ignore all constant columns. |
| 134 | if (!isColumnConst(*column)) |
| 135 | column_ptrs.emplace_back(column.get()); |
| 136 | } |
| 137 | |
| 138 | return column_ptrs; |
| 139 | } |
| 140 | |
| 141 | ColumnRawPtrs DistinctSortedBlockInputStream::getClearingColumns(const Block & block, const ColumnRawPtrs & key_columns) const |
| 142 | { |
| 143 | ColumnRawPtrs clearing_hint_columns; |
| 144 | clearing_hint_columns.reserve(description.size()); |
| 145 | for (const auto & sort_column_description : description) |
| 146 | { |
| 147 | const auto sort_column_ptr = block.safeGetByPosition(sort_column_description.column_number).column.get(); |
| 148 | const auto it = std::find(key_columns.cbegin(), key_columns.cend(), sort_column_ptr); |
| 149 | if (it != key_columns.cend()) /// if found in key_columns |
| 150 | clearing_hint_columns.emplace_back(sort_column_ptr); |
| 151 | else |
| 152 | return clearing_hint_columns; /// We will use common prefix of sort description and requested DISTINCT key. |
| 153 | } |
| 154 | return clearing_hint_columns; |
| 155 | } |
| 156 | |
| 157 | bool DistinctSortedBlockInputStream::rowsEqual(const ColumnRawPtrs & lhs, size_t n, const ColumnRawPtrs & rhs, size_t m) |
| 158 | { |
| 159 | for (size_t column_index = 0, num_columns = lhs.size(); column_index < num_columns; ++column_index) |
| 160 | { |
| 161 | const auto & lhs_column = *lhs[column_index]; |
| 162 | const auto & rhs_column = *rhs[column_index]; |
| 163 | if (lhs_column.compareAt(n, m, rhs_column, 0) != 0) /// not equal |
| 164 | return false; |
| 165 | } |
| 166 | return true; |
| 167 | } |
| 168 | |
| 169 | } |
| 170 | |