| 1 | #include <Processors/Chunk.h> |
| 2 | #include <IO/WriteHelpers.h> |
| 3 | #include <IO/Operators.h> |
| 4 | |
| 5 | namespace DB |
| 6 | { |
| 7 | |
| 8 | namespace ErrorCodes |
| 9 | { |
| 10 | extern const int LOGICAL_ERROR; |
| 11 | extern const int POSITION_OUT_OF_BOUND; |
| 12 | } |
| 13 | |
| 14 | Chunk::Chunk(DB::Columns columns_, UInt64 num_rows_) : columns(std::move(columns_)), num_rows(num_rows_) |
| 15 | { |
| 16 | checkNumRowsIsConsistent(); |
| 17 | } |
| 18 | |
| 19 | Chunk::Chunk(Columns columns_, UInt64 num_rows_, ChunkInfoPtr chunk_info_) |
| 20 | : columns(std::move(columns_)), num_rows(num_rows_), chunk_info(std::move(chunk_info_)) |
| 21 | { |
| 22 | checkNumRowsIsConsistent(); |
| 23 | } |
| 24 | |
| 25 | static Columns unmuteColumns(MutableColumns && mut_columns) |
| 26 | { |
| 27 | Columns columns; |
| 28 | columns.reserve(mut_columns.size()); |
| 29 | for (auto & col : mut_columns) |
| 30 | columns.emplace_back(std::move(col)); |
| 31 | |
| 32 | return columns; |
| 33 | } |
| 34 | |
| 35 | Chunk::Chunk(MutableColumns columns_, UInt64 num_rows_) |
| 36 | : columns(unmuteColumns(std::move(columns_))), num_rows(num_rows_) |
| 37 | { |
| 38 | checkNumRowsIsConsistent(); |
| 39 | } |
| 40 | |
| 41 | Chunk::Chunk(MutableColumns columns_, UInt64 num_rows_, ChunkInfoPtr chunk_info_) |
| 42 | : columns(unmuteColumns(std::move(columns_))), num_rows(num_rows_), chunk_info(std::move(chunk_info_)) |
| 43 | { |
| 44 | checkNumRowsIsConsistent(); |
| 45 | } |
| 46 | |
| 47 | Chunk Chunk::clone() const |
| 48 | { |
| 49 | return Chunk(getColumns(), getNumRows()); |
| 50 | } |
| 51 | |
| 52 | void Chunk::setColumns(Columns columns_, UInt64 num_rows_) |
| 53 | { |
| 54 | columns = std::move(columns_); |
| 55 | num_rows = num_rows_; |
| 56 | checkNumRowsIsConsistent(); |
| 57 | } |
| 58 | |
| 59 | void Chunk::setColumns(MutableColumns columns_, UInt64 num_rows_) |
| 60 | { |
| 61 | columns = unmuteColumns(std::move(columns_)); |
| 62 | num_rows = num_rows_; |
| 63 | checkNumRowsIsConsistent(); |
| 64 | } |
| 65 | |
| 66 | void Chunk::checkNumRowsIsConsistent() |
| 67 | { |
| 68 | for (auto & column : columns) |
| 69 | if (column->size() != num_rows) |
| 70 | throw Exception("Invalid number of rows in Chunk column " + column->getName()+ ": expected " + |
| 71 | toString(num_rows) + ", got " + toString(column->size()), ErrorCodes::LOGICAL_ERROR); |
| 72 | } |
| 73 | |
| 74 | MutableColumns Chunk::mutateColumns() |
| 75 | { |
| 76 | size_t num_columns = columns.size(); |
| 77 | MutableColumns mut_columns(num_columns); |
| 78 | for (size_t i = 0; i < num_columns; ++i) |
| 79 | mut_columns[i] = (*std::move(columns[i])).mutate(); |
| 80 | |
| 81 | columns.clear(); |
| 82 | num_rows = 0; |
| 83 | |
| 84 | return mut_columns; |
| 85 | } |
| 86 | |
| 87 | MutableColumns Chunk::cloneEmptyColumns() const |
| 88 | { |
| 89 | size_t num_columns = columns.size(); |
| 90 | MutableColumns mut_columns(num_columns); |
| 91 | for (size_t i = 0; i < num_columns; ++i) |
| 92 | mut_columns[i] = columns[i]->cloneEmpty(); |
| 93 | return mut_columns; |
| 94 | } |
| 95 | |
| 96 | Columns Chunk::detachColumns() |
| 97 | { |
| 98 | num_rows = 0; |
| 99 | return std::move(columns); |
| 100 | } |
| 101 | |
| 102 | void Chunk::addColumn(ColumnPtr column) |
| 103 | { |
| 104 | if (column->size() != num_rows) |
| 105 | throw Exception("Invalid number of rows in Chunk column " + column->getName()+ ": expected " + |
| 106 | toString(num_rows) + ", got " + toString(column->size()), ErrorCodes::LOGICAL_ERROR); |
| 107 | |
| 108 | columns.emplace_back(std::move(column)); |
| 109 | } |
| 110 | |
| 111 | void Chunk::erase(size_t position) |
| 112 | { |
| 113 | if (columns.empty()) |
| 114 | throw Exception("Chunk is empty" , ErrorCodes::POSITION_OUT_OF_BOUND); |
| 115 | |
| 116 | if (position >= columns.size()) |
| 117 | throw Exception("Position " + toString(position) + " out of bound in Chunk::erase(), max position = " |
| 118 | + toString(columns.size() - 1), ErrorCodes::POSITION_OUT_OF_BOUND); |
| 119 | |
| 120 | columns.erase(columns.begin() + position); |
| 121 | } |
| 122 | |
| 123 | UInt64 Chunk::bytes() const |
| 124 | { |
| 125 | UInt64 res = 0; |
| 126 | for (const auto & column : columns) |
| 127 | res += column->byteSize(); |
| 128 | |
| 129 | return res; |
| 130 | } |
| 131 | |
| 132 | UInt64 Chunk::allocatedBytes() const |
| 133 | { |
| 134 | UInt64 res = 0; |
| 135 | for (const auto & column : columns) |
| 136 | res += column->allocatedBytes(); |
| 137 | |
| 138 | return res; |
| 139 | } |
| 140 | |
| 141 | std::string Chunk::dumpStructure() const |
| 142 | { |
| 143 | WriteBufferFromOwnString out; |
| 144 | for (auto & column : columns) |
| 145 | out << ' ' << column->dumpStructure(); |
| 146 | |
| 147 | return out.str(); |
| 148 | } |
| 149 | |
| 150 | |
| 151 | void ChunkMissingValues::setBit(size_t column_idx, size_t row_idx) |
| 152 | { |
| 153 | RowsBitMask & mask = rows_mask_by_column_id[column_idx]; |
| 154 | mask.resize(row_idx + 1); |
| 155 | mask[row_idx] = true; |
| 156 | } |
| 157 | |
| 158 | const ChunkMissingValues::RowsBitMask & ChunkMissingValues::getDefaultsBitmask(size_t column_idx) const |
| 159 | { |
| 160 | static RowsBitMask none; |
| 161 | auto it = rows_mask_by_column_id.find(column_idx); |
| 162 | if (it != rows_mask_by_column_id.end()) |
| 163 | return it->second; |
| 164 | return none; |
| 165 | } |
| 166 | |
| 167 | } |
| 168 | |