| 1 | #include <Columns/ColumnTuple.h> |
| 2 | #include <DataStreams/ColumnGathererStream.h> |
| 3 | #include <IO/WriteBufferFromString.h> |
| 4 | #include <IO/Operators.h> |
| 5 | #include <ext/map.h> |
| 6 | #include <ext/range.h> |
| 7 | #include <Common/typeid_cast.h> |
| 8 | #include <Common/assert_cast.h> |
| 9 | #include <Core/Field.h> |
| 10 | |
| 11 | |
| 12 | namespace DB |
| 13 | { |
| 14 | |
| 15 | namespace ErrorCodes |
| 16 | { |
| 17 | extern const int ILLEGAL_COLUMN; |
| 18 | extern const int NOT_IMPLEMENTED; |
| 19 | extern const int CANNOT_INSERT_VALUE_OF_DIFFERENT_SIZE_INTO_TUPLE; |
| 20 | } |
| 21 | |
| 22 | |
| 23 | std::string ColumnTuple::getName() const |
| 24 | { |
| 25 | WriteBufferFromOwnString res; |
| 26 | res << "Tuple(" ; |
| 27 | bool is_first = true; |
| 28 | for (const auto & column : columns) |
| 29 | { |
| 30 | if (!is_first) |
| 31 | res << ", " ; |
| 32 | is_first = false; |
| 33 | res << column->getName(); |
| 34 | } |
| 35 | res << ")" ; |
| 36 | return res.str(); |
| 37 | } |
| 38 | |
| 39 | ColumnTuple::ColumnTuple(MutableColumns && mutable_columns) |
| 40 | { |
| 41 | columns.reserve(mutable_columns.size()); |
| 42 | for (auto & column : mutable_columns) |
| 43 | { |
| 44 | if (isColumnConst(*column)) |
| 45 | throw Exception{"ColumnTuple cannot have ColumnConst as its element" , ErrorCodes::ILLEGAL_COLUMN}; |
| 46 | |
| 47 | columns.push_back(std::move(column)); |
| 48 | } |
| 49 | } |
| 50 | |
| 51 | ColumnTuple::Ptr ColumnTuple::create(const Columns & columns) |
| 52 | { |
| 53 | for (const auto & column : columns) |
| 54 | if (isColumnConst(*column)) |
| 55 | throw Exception{"ColumnTuple cannot have ColumnConst as its element" , ErrorCodes::ILLEGAL_COLUMN}; |
| 56 | |
| 57 | auto column_tuple = ColumnTuple::create(MutableColumns()); |
| 58 | column_tuple->columns.assign(columns.begin(), columns.end()); |
| 59 | |
| 60 | return column_tuple; |
| 61 | } |
| 62 | |
| 63 | ColumnTuple::Ptr ColumnTuple::create(const TupleColumns & columns) |
| 64 | { |
| 65 | for (const auto & column : columns) |
| 66 | if (isColumnConst(*column)) |
| 67 | throw Exception{"ColumnTuple cannot have ColumnConst as its element" , ErrorCodes::ILLEGAL_COLUMN}; |
| 68 | |
| 69 | auto column_tuple = ColumnTuple::create(MutableColumns()); |
| 70 | column_tuple->columns = columns; |
| 71 | |
| 72 | return column_tuple; |
| 73 | } |
| 74 | |
| 75 | MutableColumnPtr ColumnTuple::cloneEmpty() const |
| 76 | { |
| 77 | const size_t tuple_size = columns.size(); |
| 78 | MutableColumns new_columns(tuple_size); |
| 79 | for (size_t i = 0; i < tuple_size; ++i) |
| 80 | new_columns[i] = columns[i]->cloneEmpty(); |
| 81 | |
| 82 | return ColumnTuple::create(std::move(new_columns)); |
| 83 | } |
| 84 | |
| 85 | MutableColumnPtr ColumnTuple::cloneResized(size_t new_size) const |
| 86 | { |
| 87 | const size_t tuple_size = columns.size(); |
| 88 | MutableColumns new_columns(tuple_size); |
| 89 | for (size_t i = 0; i < tuple_size; ++i) |
| 90 | new_columns[i] = columns[i]->cloneResized(new_size); |
| 91 | |
| 92 | return ColumnTuple::create(std::move(new_columns)); |
| 93 | } |
| 94 | |
| 95 | Field ColumnTuple::operator[](size_t n) const |
| 96 | { |
| 97 | return ext::map<Tuple>(columns, [n] (const auto & column) { return (*column)[n]; }); |
| 98 | } |
| 99 | |
| 100 | void ColumnTuple::get(size_t n, Field & res) const |
| 101 | { |
| 102 | const size_t tuple_size = columns.size(); |
| 103 | Tuple tuple(tuple_size); |
| 104 | for (const auto i : ext::range(0, tuple_size)) |
| 105 | columns[i]->get(n, tuple[i]); |
| 106 | |
| 107 | res = tuple; |
| 108 | } |
| 109 | |
| 110 | StringRef ColumnTuple::getDataAt(size_t) const |
| 111 | { |
| 112 | throw Exception("Method getDataAt is not supported for " + getName(), ErrorCodes::NOT_IMPLEMENTED); |
| 113 | } |
| 114 | |
| 115 | void ColumnTuple::insertData(const char *, size_t) |
| 116 | { |
| 117 | throw Exception("Method insertData is not supported for " + getName(), ErrorCodes::NOT_IMPLEMENTED); |
| 118 | } |
| 119 | |
| 120 | void ColumnTuple::insert(const Field & x) |
| 121 | { |
| 122 | auto & tuple = DB::get<const Tuple &>(x); |
| 123 | |
| 124 | const size_t tuple_size = columns.size(); |
| 125 | if (tuple.size() != tuple_size) |
| 126 | throw Exception("Cannot insert value of different size into tuple" , ErrorCodes::CANNOT_INSERT_VALUE_OF_DIFFERENT_SIZE_INTO_TUPLE); |
| 127 | |
| 128 | for (size_t i = 0; i < tuple_size; ++i) |
| 129 | columns[i]->insert(tuple[i]); |
| 130 | } |
| 131 | |
| 132 | void ColumnTuple::insertFrom(const IColumn & src_, size_t n) |
| 133 | { |
| 134 | const ColumnTuple & src = assert_cast<const ColumnTuple &>(src_); |
| 135 | |
| 136 | const size_t tuple_size = columns.size(); |
| 137 | if (src.columns.size() != tuple_size) |
| 138 | throw Exception("Cannot insert value of different size into tuple" , ErrorCodes::CANNOT_INSERT_VALUE_OF_DIFFERENT_SIZE_INTO_TUPLE); |
| 139 | |
| 140 | for (size_t i = 0; i < tuple_size; ++i) |
| 141 | columns[i]->insertFrom(*src.columns[i], n); |
| 142 | } |
| 143 | |
| 144 | void ColumnTuple::insertDefault() |
| 145 | { |
| 146 | for (auto & column : columns) |
| 147 | column->insertDefault(); |
| 148 | } |
| 149 | |
| 150 | void ColumnTuple::popBack(size_t n) |
| 151 | { |
| 152 | for (auto & column : columns) |
| 153 | column->popBack(n); |
| 154 | } |
| 155 | |
| 156 | StringRef ColumnTuple::serializeValueIntoArena(size_t n, Arena & arena, char const *& begin) const |
| 157 | { |
| 158 | StringRef res(begin, 0); |
| 159 | for (auto & column : columns) |
| 160 | { |
| 161 | auto value_ref = column->serializeValueIntoArena(n, arena, begin); |
| 162 | res.data = value_ref.data - res.size; |
| 163 | res.size += value_ref.size; |
| 164 | } |
| 165 | |
| 166 | return res; |
| 167 | } |
| 168 | |
| 169 | const char * ColumnTuple::deserializeAndInsertFromArena(const char * pos) |
| 170 | { |
| 171 | for (auto & column : columns) |
| 172 | pos = column->deserializeAndInsertFromArena(pos); |
| 173 | |
| 174 | return pos; |
| 175 | } |
| 176 | |
| 177 | void ColumnTuple::updateHashWithValue(size_t n, SipHash & hash) const |
| 178 | { |
| 179 | for (auto & column : columns) |
| 180 | column->updateHashWithValue(n, hash); |
| 181 | } |
| 182 | |
| 183 | void ColumnTuple::insertRangeFrom(const IColumn & src, size_t start, size_t length) |
| 184 | { |
| 185 | const size_t tuple_size = columns.size(); |
| 186 | for (size_t i = 0; i < tuple_size; ++i) |
| 187 | columns[i]->insertRangeFrom( |
| 188 | *assert_cast<const ColumnTuple &>(src).columns[i], |
| 189 | start, length); |
| 190 | } |
| 191 | |
| 192 | ColumnPtr ColumnTuple::filter(const Filter & filt, ssize_t result_size_hint) const |
| 193 | { |
| 194 | const size_t tuple_size = columns.size(); |
| 195 | Columns new_columns(tuple_size); |
| 196 | |
| 197 | for (size_t i = 0; i < tuple_size; ++i) |
| 198 | new_columns[i] = columns[i]->filter(filt, result_size_hint); |
| 199 | |
| 200 | return ColumnTuple::create(new_columns); |
| 201 | } |
| 202 | |
| 203 | ColumnPtr ColumnTuple::permute(const Permutation & perm, size_t limit) const |
| 204 | { |
| 205 | const size_t tuple_size = columns.size(); |
| 206 | Columns new_columns(tuple_size); |
| 207 | |
| 208 | for (size_t i = 0; i < tuple_size; ++i) |
| 209 | new_columns[i] = columns[i]->permute(perm, limit); |
| 210 | |
| 211 | return ColumnTuple::create(new_columns); |
| 212 | } |
| 213 | |
| 214 | ColumnPtr ColumnTuple::index(const IColumn & indexes, size_t limit) const |
| 215 | { |
| 216 | const size_t tuple_size = columns.size(); |
| 217 | Columns new_columns(tuple_size); |
| 218 | |
| 219 | for (size_t i = 0; i < tuple_size; ++i) |
| 220 | new_columns[i] = columns[i]->index(indexes, limit); |
| 221 | |
| 222 | return ColumnTuple::create(new_columns); |
| 223 | } |
| 224 | |
| 225 | ColumnPtr ColumnTuple::replicate(const Offsets & offsets) const |
| 226 | { |
| 227 | const size_t tuple_size = columns.size(); |
| 228 | Columns new_columns(tuple_size); |
| 229 | |
| 230 | for (size_t i = 0; i < tuple_size; ++i) |
| 231 | new_columns[i] = columns[i]->replicate(offsets); |
| 232 | |
| 233 | return ColumnTuple::create(new_columns); |
| 234 | } |
| 235 | |
| 236 | MutableColumns ColumnTuple::scatter(ColumnIndex num_columns, const Selector & selector) const |
| 237 | { |
| 238 | const size_t tuple_size = columns.size(); |
| 239 | std::vector<MutableColumns> scattered_tuple_elements(tuple_size); |
| 240 | |
| 241 | for (size_t tuple_element_idx = 0; tuple_element_idx < tuple_size; ++tuple_element_idx) |
| 242 | scattered_tuple_elements[tuple_element_idx] = columns[tuple_element_idx]->scatter(num_columns, selector); |
| 243 | |
| 244 | MutableColumns res(num_columns); |
| 245 | |
| 246 | for (size_t scattered_idx = 0; scattered_idx < num_columns; ++scattered_idx) |
| 247 | { |
| 248 | MutableColumns new_columns(tuple_size); |
| 249 | for (size_t tuple_element_idx = 0; tuple_element_idx < tuple_size; ++tuple_element_idx) |
| 250 | new_columns[tuple_element_idx] = std::move(scattered_tuple_elements[tuple_element_idx][scattered_idx]); |
| 251 | res[scattered_idx] = ColumnTuple::create(std::move(new_columns)); |
| 252 | } |
| 253 | |
| 254 | return res; |
| 255 | } |
| 256 | |
| 257 | int ColumnTuple::compareAt(size_t n, size_t m, const IColumn & rhs, int nan_direction_hint) const |
| 258 | { |
| 259 | const size_t tuple_size = columns.size(); |
| 260 | for (size_t i = 0; i < tuple_size; ++i) |
| 261 | if (int res = columns[i]->compareAt(n, m, *assert_cast<const ColumnTuple &>(rhs).columns[i], nan_direction_hint)) |
| 262 | return res; |
| 263 | |
| 264 | return 0; |
| 265 | } |
| 266 | |
| 267 | template <bool positive> |
| 268 | struct ColumnTuple::Less |
| 269 | { |
| 270 | TupleColumns columns; |
| 271 | int nan_direction_hint; |
| 272 | |
| 273 | Less(const TupleColumns & columns_, int nan_direction_hint_) |
| 274 | : columns(columns_), nan_direction_hint(nan_direction_hint_) |
| 275 | { |
| 276 | } |
| 277 | |
| 278 | bool operator() (size_t a, size_t b) const |
| 279 | { |
| 280 | for (const auto & column : columns) |
| 281 | { |
| 282 | int res = column->compareAt(a, b, *column, nan_direction_hint); |
| 283 | if (res < 0) |
| 284 | return positive; |
| 285 | else if (res > 0) |
| 286 | return !positive; |
| 287 | } |
| 288 | return false; |
| 289 | } |
| 290 | }; |
| 291 | |
| 292 | void ColumnTuple::getPermutation(bool reverse, size_t limit, int nan_direction_hint, Permutation & res) const |
| 293 | { |
| 294 | size_t rows = size(); |
| 295 | res.resize(rows); |
| 296 | for (size_t i = 0; i < rows; ++i) |
| 297 | res[i] = i; |
| 298 | |
| 299 | if (limit >= rows) |
| 300 | limit = 0; |
| 301 | |
| 302 | if (limit) |
| 303 | { |
| 304 | if (reverse) |
| 305 | std::partial_sort(res.begin(), res.begin() + limit, res.end(), Less<false>(columns, nan_direction_hint)); |
| 306 | else |
| 307 | std::partial_sort(res.begin(), res.begin() + limit, res.end(), Less<true>(columns, nan_direction_hint)); |
| 308 | } |
| 309 | else |
| 310 | { |
| 311 | if (reverse) |
| 312 | std::sort(res.begin(), res.end(), Less<false>(columns, nan_direction_hint)); |
| 313 | else |
| 314 | std::sort(res.begin(), res.end(), Less<true>(columns, nan_direction_hint)); |
| 315 | } |
| 316 | } |
| 317 | |
| 318 | void ColumnTuple::gather(ColumnGathererStream & gatherer) |
| 319 | { |
| 320 | gatherer.gather(*this); |
| 321 | } |
| 322 | |
| 323 | void ColumnTuple::reserve(size_t n) |
| 324 | { |
| 325 | const size_t tuple_size = columns.size(); |
| 326 | for (size_t i = 0; i < tuple_size; ++i) |
| 327 | getColumn(i).reserve(n); |
| 328 | } |
| 329 | |
| 330 | size_t ColumnTuple::byteSize() const |
| 331 | { |
| 332 | size_t res = 0; |
| 333 | for (const auto & column : columns) |
| 334 | res += column->byteSize(); |
| 335 | return res; |
| 336 | } |
| 337 | |
| 338 | size_t ColumnTuple::allocatedBytes() const |
| 339 | { |
| 340 | size_t res = 0; |
| 341 | for (const auto & column : columns) |
| 342 | res += column->allocatedBytes(); |
| 343 | return res; |
| 344 | } |
| 345 | |
| 346 | void ColumnTuple::protect() |
| 347 | { |
| 348 | for (auto & column : columns) |
| 349 | column->protect(); |
| 350 | } |
| 351 | |
| 352 | void ColumnTuple::getExtremes(Field & min, Field & max) const |
| 353 | { |
| 354 | const size_t tuple_size = columns.size(); |
| 355 | |
| 356 | Tuple min_tuple(tuple_size); |
| 357 | Tuple max_tuple(tuple_size); |
| 358 | |
| 359 | for (const auto i : ext::range(0, tuple_size)) |
| 360 | columns[i]->getExtremes(min_tuple[i], max_tuple[i]); |
| 361 | |
| 362 | min = min_tuple; |
| 363 | max = max_tuple; |
| 364 | } |
| 365 | |
| 366 | void ColumnTuple::forEachSubcolumn(ColumnCallback callback) |
| 367 | { |
| 368 | for (auto & column : columns) |
| 369 | callback(column); |
| 370 | } |
| 371 | |
| 372 | bool ColumnTuple::structureEquals(const IColumn & rhs) const |
| 373 | { |
| 374 | if (auto rhs_tuple = typeid_cast<const ColumnTuple *>(&rhs)) |
| 375 | { |
| 376 | const size_t tuple_size = columns.size(); |
| 377 | if (tuple_size != rhs_tuple->columns.size()) |
| 378 | return false; |
| 379 | |
| 380 | for (const auto i : ext::range(0, tuple_size)) |
| 381 | if (!columns[i]->structureEquals(*rhs_tuple->columns[i])) |
| 382 | return false; |
| 383 | |
| 384 | return true; |
| 385 | } |
| 386 | else |
| 387 | return false; |
| 388 | } |
| 389 | |
| 390 | |
| 391 | } |
| 392 | |