| 1 | #include <Functions/IFunctionImpl.h> |
| 2 | #include <Functions/FunctionFactory.h> |
| 3 | #include <Functions/FunctionHelpers.h> |
| 4 | #include <DataTypes/DataTypeArray.h> |
| 5 | #include <DataTypes/DataTypeNullable.h> |
| 6 | #include <DataTypes/DataTypeTuple.h> |
| 7 | #include <Core/ColumnNumbers.h> |
| 8 | #include <Columns/ColumnArray.h> |
| 9 | #include <Columns/ColumnNullable.h> |
| 10 | #include <Columns/ColumnsNumber.h> |
| 11 | #include <Columns/ColumnString.h> |
| 12 | #include <Columns/ColumnTuple.h> |
| 13 | #include <Common/typeid_cast.h> |
| 14 | #include <Common/assert_cast.h> |
| 15 | |
| 16 | |
| 17 | namespace DB |
| 18 | { |
| 19 | |
| 20 | namespace ErrorCodes |
| 21 | { |
| 22 | extern const int LOGICAL_ERROR; |
| 23 | extern const int ILLEGAL_COLUMN; |
| 24 | extern const int ILLEGAL_TYPE_OF_ARGUMENT; |
| 25 | extern const int ZERO_ARRAY_OR_TUPLE_INDEX; |
| 26 | } |
| 27 | |
| 28 | namespace ArrayImpl |
| 29 | { |
| 30 | class NullMapBuilder; |
| 31 | } |
| 32 | |
| 33 | /** arrayElement(arr, i) - get the array element by index. If index is not constant and out of range - return default value of data type. |
| 34 | * The index begins with 1. Also, the index can be negative - then it is counted from the end of the array. |
| 35 | */ |
| 36 | class FunctionArrayElement : public IFunction |
| 37 | { |
| 38 | public: |
| 39 | static constexpr auto name = "arrayElement" ; |
| 40 | static FunctionPtr create(const Context & context); |
| 41 | |
| 42 | String getName() const override; |
| 43 | |
| 44 | bool useDefaultImplementationForConstants() const override { return true; } |
| 45 | size_t getNumberOfArguments() const override { return 2; } |
| 46 | |
| 47 | DataTypePtr getReturnTypeImpl(const DataTypes & arguments) const override; |
| 48 | |
| 49 | void executeImpl(Block & block, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) override; |
| 50 | |
| 51 | private: |
| 52 | void perform(Block & block, const ColumnNumbers & arguments, size_t result, |
| 53 | ArrayImpl::NullMapBuilder & builder, size_t input_rows_count); |
| 54 | |
| 55 | template <typename DataType> |
| 56 | bool executeNumberConst(Block & block, const ColumnNumbers & arguments, size_t result, const Field & index, |
| 57 | ArrayImpl::NullMapBuilder & builder); |
| 58 | |
| 59 | template <typename IndexType, typename DataType> |
| 60 | bool executeNumber(Block & block, const ColumnNumbers & arguments, size_t result, const PaddedPODArray<IndexType> & indices, |
| 61 | ArrayImpl::NullMapBuilder & builder); |
| 62 | |
| 63 | bool executeStringConst(Block & block, const ColumnNumbers & arguments, size_t result, const Field & index, |
| 64 | ArrayImpl::NullMapBuilder & builder); |
| 65 | |
| 66 | template <typename IndexType> |
| 67 | bool executeString(Block & block, const ColumnNumbers & arguments, size_t result, const PaddedPODArray<IndexType> & indices, |
| 68 | ArrayImpl::NullMapBuilder & builder); |
| 69 | |
| 70 | bool executeGenericConst(Block & block, const ColumnNumbers & arguments, size_t result, const Field & index, |
| 71 | ArrayImpl::NullMapBuilder & builder); |
| 72 | |
| 73 | template <typename IndexType> |
| 74 | bool executeGeneric(Block & block, const ColumnNumbers & arguments, size_t result, const PaddedPODArray<IndexType> & indices, |
| 75 | ArrayImpl::NullMapBuilder & builder); |
| 76 | |
| 77 | template <typename IndexType> |
| 78 | bool executeConst(Block & block, const ColumnNumbers & arguments, size_t result, |
| 79 | const PaddedPODArray <IndexType> & indices, ArrayImpl::NullMapBuilder & builder, |
| 80 | size_t input_rows_count); |
| 81 | |
| 82 | template <typename IndexType> |
| 83 | bool executeArgument(Block & block, const ColumnNumbers & arguments, size_t result, |
| 84 | ArrayImpl::NullMapBuilder & builder, size_t input_rows_count); |
| 85 | |
| 86 | /** For a tuple array, the function is evaluated component-wise for each element of the tuple. |
| 87 | */ |
| 88 | bool executeTuple(Block & block, const ColumnNumbers & arguments, size_t result, size_t input_rows_count); |
| 89 | }; |
| 90 | |
| 91 | |
| 92 | namespace ArrayImpl |
| 93 | { |
| 94 | |
| 95 | class NullMapBuilder |
| 96 | { |
| 97 | public: |
| 98 | operator bool() const { return src_null_map; } |
| 99 | bool operator!() const { return !src_null_map; } |
| 100 | |
| 101 | void initSource(const UInt8 * src_null_map_) |
| 102 | { |
| 103 | src_null_map = src_null_map_; |
| 104 | } |
| 105 | |
| 106 | void initSink(size_t size) |
| 107 | { |
| 108 | auto sink = ColumnUInt8::create(size); |
| 109 | sink_null_map = sink->getData().data(); |
| 110 | sink_null_map_holder = std::move(sink); |
| 111 | } |
| 112 | |
| 113 | void update(size_t from) |
| 114 | { |
| 115 | sink_null_map[index] = bool(src_null_map && src_null_map[from]); |
| 116 | ++index; |
| 117 | } |
| 118 | |
| 119 | void update() |
| 120 | { |
| 121 | sink_null_map[index] = bool(src_null_map); |
| 122 | ++index; |
| 123 | } |
| 124 | |
| 125 | ColumnPtr getNullMapColumnPtr() && { return std::move(sink_null_map_holder); } |
| 126 | |
| 127 | private: |
| 128 | const UInt8 * src_null_map = nullptr; |
| 129 | UInt8 * sink_null_map = nullptr; |
| 130 | MutableColumnPtr sink_null_map_holder; |
| 131 | size_t index = 0; |
| 132 | }; |
| 133 | |
| 134 | } |
| 135 | |
| 136 | namespace |
| 137 | { |
| 138 | |
| 139 | template <typename T> |
| 140 | struct ArrayElementNumImpl |
| 141 | { |
| 142 | /** Implementation for constant index. |
| 143 | * If negative = false - index is from beginning of array, started from 0. |
| 144 | * If negative = true - index is from end of array, started from 0. |
| 145 | */ |
| 146 | template <bool negative> |
| 147 | static void vectorConst( |
| 148 | const PaddedPODArray<T> & data, const ColumnArray::Offsets & offsets, |
| 149 | const ColumnArray::Offset index, |
| 150 | PaddedPODArray<T> & result, ArrayImpl::NullMapBuilder & builder) |
| 151 | { |
| 152 | size_t size = offsets.size(); |
| 153 | result.resize(size); |
| 154 | |
| 155 | ColumnArray::Offset current_offset = 0; |
| 156 | for (size_t i = 0; i < size; ++i) |
| 157 | { |
| 158 | size_t array_size = offsets[i] - current_offset; |
| 159 | |
| 160 | if (index < array_size) |
| 161 | { |
| 162 | size_t j = !negative ? (current_offset + index) : (offsets[i] - index - 1); |
| 163 | result[i] = data[j]; |
| 164 | if (builder) |
| 165 | builder.update(j); |
| 166 | } |
| 167 | else |
| 168 | { |
| 169 | result[i] = T(); |
| 170 | |
| 171 | if (builder) |
| 172 | builder.update(); |
| 173 | } |
| 174 | |
| 175 | current_offset = offsets[i]; |
| 176 | } |
| 177 | } |
| 178 | |
| 179 | /** Implementation for non-constant index. |
| 180 | */ |
| 181 | template <typename TIndex> |
| 182 | static void vector( |
| 183 | const PaddedPODArray<T> & data, const ColumnArray::Offsets & offsets, |
| 184 | const PaddedPODArray<TIndex> & indices, |
| 185 | PaddedPODArray<T> & result, ArrayImpl::NullMapBuilder & builder) |
| 186 | { |
| 187 | size_t size = offsets.size(); |
| 188 | result.resize(size); |
| 189 | |
| 190 | ColumnArray::Offset current_offset = 0; |
| 191 | for (size_t i = 0; i < size; ++i) |
| 192 | { |
| 193 | size_t array_size = offsets[i] - current_offset; |
| 194 | |
| 195 | TIndex index = indices[i]; |
| 196 | if (index > 0 && static_cast<size_t>(index) <= array_size) |
| 197 | { |
| 198 | size_t j = current_offset + index - 1; |
| 199 | result[i] = data[j]; |
| 200 | |
| 201 | if (builder) |
| 202 | builder.update(j); |
| 203 | } |
| 204 | else if (index < 0 && static_cast<size_t>(-index) <= array_size) |
| 205 | { |
| 206 | size_t j = offsets[i] + index; |
| 207 | result[i] = data[j]; |
| 208 | |
| 209 | if (builder) |
| 210 | builder.update(j); |
| 211 | } |
| 212 | else |
| 213 | { |
| 214 | result[i] = T(); |
| 215 | |
| 216 | if (builder) |
| 217 | builder.update(); |
| 218 | } |
| 219 | |
| 220 | current_offset = offsets[i]; |
| 221 | } |
| 222 | } |
| 223 | }; |
| 224 | |
| 225 | struct ArrayElementStringImpl |
| 226 | { |
| 227 | template <bool negative> |
| 228 | static void vectorConst( |
| 229 | const ColumnString::Chars & data, const ColumnArray::Offsets & offsets, const ColumnString::Offsets & string_offsets, |
| 230 | const ColumnArray::Offset index, |
| 231 | ColumnString::Chars & result_data, ColumnArray::Offsets & result_offsets, |
| 232 | ArrayImpl::NullMapBuilder & builder) |
| 233 | { |
| 234 | size_t size = offsets.size(); |
| 235 | result_offsets.resize(size); |
| 236 | result_data.reserve(data.size()); |
| 237 | |
| 238 | ColumnArray::Offset current_offset = 0; |
| 239 | ColumnArray::Offset current_result_offset = 0; |
| 240 | for (size_t i = 0; i < size; ++i) |
| 241 | { |
| 242 | size_t array_size = offsets[i] - current_offset; |
| 243 | |
| 244 | if (index < array_size) |
| 245 | { |
| 246 | size_t adjusted_index = !negative ? index : (array_size - index - 1); |
| 247 | |
| 248 | size_t j = current_offset + adjusted_index; |
| 249 | if (builder) |
| 250 | builder.update(j); |
| 251 | |
| 252 | ColumnArray::Offset string_pos = current_offset == 0 && adjusted_index == 0 |
| 253 | ? 0 |
| 254 | : string_offsets[current_offset + adjusted_index - 1]; |
| 255 | |
| 256 | ColumnArray::Offset string_size = string_offsets[current_offset + adjusted_index] - string_pos; |
| 257 | |
| 258 | result_data.resize(current_result_offset + string_size); |
| 259 | memcpySmallAllowReadWriteOverflow15(&result_data[current_result_offset], &data[string_pos], string_size); |
| 260 | current_result_offset += string_size; |
| 261 | result_offsets[i] = current_result_offset; |
| 262 | } |
| 263 | else |
| 264 | { |
| 265 | /// Insert an empty row. |
| 266 | result_data.resize(current_result_offset + 1); |
| 267 | result_data[current_result_offset] = 0; |
| 268 | current_result_offset += 1; |
| 269 | result_offsets[i] = current_result_offset; |
| 270 | |
| 271 | if (builder) |
| 272 | builder.update(); |
| 273 | } |
| 274 | |
| 275 | current_offset = offsets[i]; |
| 276 | } |
| 277 | } |
| 278 | |
| 279 | /** Implementation for non-constant index. |
| 280 | */ |
| 281 | template <typename TIndex> |
| 282 | static void vector( |
| 283 | const ColumnString::Chars & data, const ColumnArray::Offsets & offsets, const ColumnString::Offsets & string_offsets, |
| 284 | const PaddedPODArray<TIndex> & indices, |
| 285 | ColumnString::Chars & result_data, ColumnArray::Offsets & result_offsets, |
| 286 | ArrayImpl::NullMapBuilder & builder) |
| 287 | { |
| 288 | size_t size = offsets.size(); |
| 289 | result_offsets.resize(size); |
| 290 | result_data.reserve(data.size()); |
| 291 | |
| 292 | ColumnArray::Offset current_offset = 0; |
| 293 | ColumnArray::Offset current_result_offset = 0; |
| 294 | for (size_t i = 0; i < size; ++i) |
| 295 | { |
| 296 | size_t array_size = offsets[i] - current_offset; |
| 297 | size_t adjusted_index; /// index in array from zero |
| 298 | |
| 299 | TIndex index = indices[i]; |
| 300 | if (index > 0 && static_cast<size_t>(index) <= array_size) |
| 301 | adjusted_index = index - 1; |
| 302 | else if (index < 0 && static_cast<size_t>(-index) <= array_size) |
| 303 | adjusted_index = array_size + index; |
| 304 | else |
| 305 | adjusted_index = array_size; /// means no element should be taken |
| 306 | |
| 307 | if (adjusted_index < array_size) |
| 308 | { |
| 309 | size_t j = current_offset + adjusted_index; |
| 310 | if (builder) |
| 311 | builder.update(j); |
| 312 | |
| 313 | ColumnArray::Offset string_pos = current_offset == 0 && adjusted_index == 0 |
| 314 | ? 0 |
| 315 | : string_offsets[current_offset + adjusted_index - 1]; |
| 316 | |
| 317 | ColumnArray::Offset string_size = string_offsets[current_offset + adjusted_index] - string_pos; |
| 318 | |
| 319 | result_data.resize(current_result_offset + string_size); |
| 320 | memcpySmallAllowReadWriteOverflow15(&result_data[current_result_offset], &data[string_pos], string_size); |
| 321 | current_result_offset += string_size; |
| 322 | result_offsets[i] = current_result_offset; |
| 323 | } |
| 324 | else |
| 325 | { |
| 326 | /// Insert empty string |
| 327 | result_data.resize(current_result_offset + 1); |
| 328 | result_data[current_result_offset] = 0; |
| 329 | current_result_offset += 1; |
| 330 | result_offsets[i] = current_result_offset; |
| 331 | |
| 332 | if (builder) |
| 333 | builder.update(); |
| 334 | } |
| 335 | |
| 336 | current_offset = offsets[i]; |
| 337 | } |
| 338 | } |
| 339 | }; |
| 340 | |
| 341 | /// Generic implementation for other nested types. |
| 342 | struct ArrayElementGenericImpl |
| 343 | { |
| 344 | template <bool negative> |
| 345 | static void vectorConst( |
| 346 | const IColumn & data, const ColumnArray::Offsets & offsets, |
| 347 | const ColumnArray::Offset index, |
| 348 | IColumn & result, ArrayImpl::NullMapBuilder & builder) |
| 349 | { |
| 350 | size_t size = offsets.size(); |
| 351 | result.reserve(size); |
| 352 | |
| 353 | ColumnArray::Offset current_offset = 0; |
| 354 | for (size_t i = 0; i < size; ++i) |
| 355 | { |
| 356 | size_t array_size = offsets[i] - current_offset; |
| 357 | |
| 358 | if (index < array_size) |
| 359 | { |
| 360 | size_t j = !negative ? current_offset + index : offsets[i] - index - 1; |
| 361 | result.insertFrom(data, j); |
| 362 | if (builder) |
| 363 | builder.update(j); |
| 364 | } |
| 365 | else |
| 366 | { |
| 367 | result.insertDefault(); |
| 368 | if (builder) |
| 369 | builder.update(); |
| 370 | } |
| 371 | |
| 372 | current_offset = offsets[i]; |
| 373 | } |
| 374 | } |
| 375 | |
| 376 | /** Implementation for non-constant index. |
| 377 | */ |
| 378 | template <typename TIndex> |
| 379 | static void vector( |
| 380 | const IColumn & data, const ColumnArray::Offsets & offsets, |
| 381 | const PaddedPODArray<TIndex> & indices, |
| 382 | IColumn & result, ArrayImpl::NullMapBuilder & builder) |
| 383 | { |
| 384 | size_t size = offsets.size(); |
| 385 | result.reserve(size); |
| 386 | |
| 387 | ColumnArray::Offset current_offset = 0; |
| 388 | for (size_t i = 0; i < size; ++i) |
| 389 | { |
| 390 | size_t array_size = offsets[i] - current_offset; |
| 391 | |
| 392 | TIndex index = indices[i]; |
| 393 | if (index > 0 && static_cast<size_t>(index) <= array_size) |
| 394 | { |
| 395 | size_t j = current_offset + index - 1; |
| 396 | result.insertFrom(data, j); |
| 397 | if (builder) |
| 398 | builder.update(j); |
| 399 | } |
| 400 | else if (index < 0 && static_cast<size_t>(-index) <= array_size) |
| 401 | { |
| 402 | size_t j = offsets[i] + index; |
| 403 | result.insertFrom(data, j); |
| 404 | if (builder) |
| 405 | builder.update(j); |
| 406 | } |
| 407 | else |
| 408 | { |
| 409 | result.insertDefault(); |
| 410 | if (builder) |
| 411 | builder.update(); |
| 412 | } |
| 413 | |
| 414 | current_offset = offsets[i]; |
| 415 | } |
| 416 | } |
| 417 | }; |
| 418 | |
| 419 | } |
| 420 | |
| 421 | |
| 422 | FunctionPtr FunctionArrayElement::create(const Context &) |
| 423 | { |
| 424 | return std::make_shared<FunctionArrayElement>(); |
| 425 | } |
| 426 | |
| 427 | |
| 428 | template <typename DataType> |
| 429 | bool FunctionArrayElement::executeNumberConst(Block & block, const ColumnNumbers & arguments, size_t result, const Field & index, |
| 430 | ArrayImpl::NullMapBuilder & builder) |
| 431 | { |
| 432 | const ColumnArray * col_array = checkAndGetColumn<ColumnArray>(block.getByPosition(arguments[0]).column.get()); |
| 433 | |
| 434 | if (!col_array) |
| 435 | return false; |
| 436 | |
| 437 | const ColumnVector<DataType> * col_nested = checkAndGetColumn<ColumnVector<DataType>>(&col_array->getData()); |
| 438 | |
| 439 | if (!col_nested) |
| 440 | return false; |
| 441 | |
| 442 | auto col_res = ColumnVector<DataType>::create(); |
| 443 | |
| 444 | if (index.getType() == Field::Types::UInt64) |
| 445 | ArrayElementNumImpl<DataType>::template vectorConst<false>( |
| 446 | col_nested->getData(), col_array->getOffsets(), safeGet<UInt64>(index) - 1, col_res->getData(), builder); |
| 447 | else if (index.getType() == Field::Types::Int64) |
| 448 | ArrayElementNumImpl<DataType>::template vectorConst<true>( |
| 449 | col_nested->getData(), col_array->getOffsets(), -safeGet<Int64>(index) - 1, col_res->getData(), builder); |
| 450 | else |
| 451 | throw Exception("Illegal type of array index" , ErrorCodes::LOGICAL_ERROR); |
| 452 | |
| 453 | block.getByPosition(result).column = std::move(col_res); |
| 454 | return true; |
| 455 | } |
| 456 | |
| 457 | template <typename IndexType, typename DataType> |
| 458 | bool FunctionArrayElement::executeNumber(Block & block, const ColumnNumbers & arguments, size_t result, const PaddedPODArray<IndexType> & indices, |
| 459 | ArrayImpl::NullMapBuilder & builder) |
| 460 | { |
| 461 | const ColumnArray * col_array = checkAndGetColumn<ColumnArray>(block.getByPosition(arguments[0]).column.get()); |
| 462 | |
| 463 | if (!col_array) |
| 464 | return false; |
| 465 | |
| 466 | const ColumnVector<DataType> * col_nested = checkAndGetColumn<ColumnVector<DataType>>(&col_array->getData()); |
| 467 | |
| 468 | if (!col_nested) |
| 469 | return false; |
| 470 | |
| 471 | auto col_res = ColumnVector<DataType>::create(); |
| 472 | |
| 473 | ArrayElementNumImpl<DataType>::template vector<IndexType>( |
| 474 | col_nested->getData(), col_array->getOffsets(), indices, col_res->getData(), builder); |
| 475 | |
| 476 | block.getByPosition(result).column = std::move(col_res); |
| 477 | return true; |
| 478 | } |
| 479 | |
| 480 | bool FunctionArrayElement::executeStringConst(Block & block, const ColumnNumbers & arguments, size_t result, const Field & index, |
| 481 | ArrayImpl::NullMapBuilder & builder) |
| 482 | { |
| 483 | const ColumnArray * col_array = checkAndGetColumn<ColumnArray>(block.getByPosition(arguments[0]).column.get()); |
| 484 | |
| 485 | if (!col_array) |
| 486 | return false; |
| 487 | |
| 488 | const ColumnString * col_nested = checkAndGetColumn<ColumnString>(&col_array->getData()); |
| 489 | |
| 490 | if (!col_nested) |
| 491 | return false; |
| 492 | |
| 493 | auto col_res = ColumnString::create(); |
| 494 | |
| 495 | if (index.getType() == Field::Types::UInt64) |
| 496 | ArrayElementStringImpl::vectorConst<false>( |
| 497 | col_nested->getChars(), |
| 498 | col_array->getOffsets(), |
| 499 | col_nested->getOffsets(), |
| 500 | safeGet<UInt64>(index) - 1, |
| 501 | col_res->getChars(), |
| 502 | col_res->getOffsets(), |
| 503 | builder); |
| 504 | else if (index.getType() == Field::Types::Int64) |
| 505 | ArrayElementStringImpl::vectorConst<true>( |
| 506 | col_nested->getChars(), |
| 507 | col_array->getOffsets(), |
| 508 | col_nested->getOffsets(), |
| 509 | -safeGet<Int64>(index) - 1, |
| 510 | col_res->getChars(), |
| 511 | col_res->getOffsets(), |
| 512 | builder); |
| 513 | else |
| 514 | throw Exception("Illegal type of array index" , ErrorCodes::LOGICAL_ERROR); |
| 515 | |
| 516 | block.getByPosition(result).column = std::move(col_res); |
| 517 | return true; |
| 518 | } |
| 519 | |
| 520 | template <typename IndexType> |
| 521 | bool FunctionArrayElement::executeString(Block & block, const ColumnNumbers & arguments, size_t result, const PaddedPODArray<IndexType> & indices, |
| 522 | ArrayImpl::NullMapBuilder & builder) |
| 523 | { |
| 524 | const ColumnArray * col_array = checkAndGetColumn<ColumnArray>(block.getByPosition(arguments[0]).column.get()); |
| 525 | |
| 526 | if (!col_array) |
| 527 | return false; |
| 528 | |
| 529 | const ColumnString * col_nested = checkAndGetColumn<ColumnString>(&col_array->getData()); |
| 530 | |
| 531 | if (!col_nested) |
| 532 | return false; |
| 533 | |
| 534 | auto col_res = ColumnString::create(); |
| 535 | |
| 536 | ArrayElementStringImpl::vector<IndexType>( |
| 537 | col_nested->getChars(), |
| 538 | col_array->getOffsets(), |
| 539 | col_nested->getOffsets(), |
| 540 | indices, |
| 541 | col_res->getChars(), |
| 542 | col_res->getOffsets(), |
| 543 | builder); |
| 544 | |
| 545 | block.getByPosition(result).column = std::move(col_res); |
| 546 | return true; |
| 547 | } |
| 548 | |
| 549 | bool FunctionArrayElement::executeGenericConst(Block & block, const ColumnNumbers & arguments, size_t result, const Field & index, |
| 550 | ArrayImpl::NullMapBuilder & builder) |
| 551 | { |
| 552 | const ColumnArray * col_array = checkAndGetColumn<ColumnArray>(block.getByPosition(arguments[0]).column.get()); |
| 553 | |
| 554 | if (!col_array) |
| 555 | return false; |
| 556 | |
| 557 | const auto & col_nested = col_array->getData(); |
| 558 | auto col_res = col_nested.cloneEmpty(); |
| 559 | |
| 560 | if (index.getType() == Field::Types::UInt64) |
| 561 | ArrayElementGenericImpl::vectorConst<false>( |
| 562 | col_nested, col_array->getOffsets(), safeGet<UInt64>(index) - 1, *col_res, builder); |
| 563 | else if (index.getType() == Field::Types::Int64) |
| 564 | ArrayElementGenericImpl::vectorConst<true>( |
| 565 | col_nested, col_array->getOffsets(), -safeGet<Int64>(index) - 1, *col_res, builder); |
| 566 | else |
| 567 | throw Exception("Illegal type of array index" , ErrorCodes::LOGICAL_ERROR); |
| 568 | |
| 569 | block.getByPosition(result).column = std::move(col_res); |
| 570 | return true; |
| 571 | } |
| 572 | |
| 573 | template <typename IndexType> |
| 574 | bool FunctionArrayElement::executeGeneric(Block & block, const ColumnNumbers & arguments, size_t result, const PaddedPODArray<IndexType> & indices, |
| 575 | ArrayImpl::NullMapBuilder & builder) |
| 576 | { |
| 577 | const ColumnArray * col_array = checkAndGetColumn<ColumnArray>(block.getByPosition(arguments[0]).column.get()); |
| 578 | |
| 579 | if (!col_array) |
| 580 | return false; |
| 581 | |
| 582 | const auto & col_nested = col_array->getData(); |
| 583 | auto col_res = col_nested.cloneEmpty(); |
| 584 | |
| 585 | ArrayElementGenericImpl::vector<IndexType>( |
| 586 | col_nested, col_array->getOffsets(), indices, *col_res, builder); |
| 587 | |
| 588 | block.getByPosition(result).column = std::move(col_res); |
| 589 | return true; |
| 590 | } |
| 591 | |
| 592 | template <typename IndexType> |
| 593 | bool FunctionArrayElement::executeConst(Block & block, const ColumnNumbers & arguments, size_t result, |
| 594 | const PaddedPODArray <IndexType> & indices, ArrayImpl::NullMapBuilder & builder, |
| 595 | size_t input_rows_count) |
| 596 | { |
| 597 | const ColumnArray * col_array = checkAndGetColumnConstData<ColumnArray>(block.getByPosition(arguments[0]).column.get()); |
| 598 | |
| 599 | if (!col_array) |
| 600 | return false; |
| 601 | |
| 602 | auto res = block.getByPosition(result).type->createColumn(); |
| 603 | |
| 604 | size_t rows = input_rows_count; |
| 605 | const IColumn & array_elements = col_array->getData(); |
| 606 | size_t array_size = array_elements.size(); |
| 607 | |
| 608 | for (size_t i = 0; i < rows; ++i) |
| 609 | { |
| 610 | IndexType index = indices[i]; |
| 611 | if (index > 0 && static_cast<size_t>(index) <= array_size) |
| 612 | { |
| 613 | size_t j = index - 1; |
| 614 | res->insertFrom(array_elements, j); |
| 615 | if (builder) |
| 616 | builder.update(j); |
| 617 | } |
| 618 | else if (index < 0 && static_cast<size_t>(-index) <= array_size) |
| 619 | { |
| 620 | size_t j = array_size + index; |
| 621 | res->insertFrom(array_elements, j); |
| 622 | if (builder) |
| 623 | builder.update(j); |
| 624 | } |
| 625 | else |
| 626 | { |
| 627 | res->insertDefault(); |
| 628 | if (builder) |
| 629 | builder.update(); |
| 630 | } |
| 631 | } |
| 632 | |
| 633 | block.getByPosition(result).column = std::move(res); |
| 634 | return true; |
| 635 | } |
| 636 | |
| 637 | template <typename IndexType> |
| 638 | bool FunctionArrayElement::executeArgument(Block & block, const ColumnNumbers & arguments, size_t result, |
| 639 | ArrayImpl::NullMapBuilder & builder, size_t input_rows_count) |
| 640 | { |
| 641 | auto index = checkAndGetColumn<ColumnVector<IndexType>>(block.getByPosition(arguments[1]).column.get()); |
| 642 | |
| 643 | if (!index) |
| 644 | return false; |
| 645 | |
| 646 | const auto & index_data = index->getData(); |
| 647 | |
| 648 | if (builder) |
| 649 | builder.initSink(index_data.size()); |
| 650 | |
| 651 | if (!(executeNumber<IndexType, UInt8>(block, arguments, result, index_data, builder) |
| 652 | || executeNumber<IndexType, UInt16>(block, arguments, result, index_data, builder) |
| 653 | || executeNumber<IndexType, UInt32>(block, arguments, result, index_data, builder) |
| 654 | || executeNumber<IndexType, UInt64>(block, arguments, result, index_data, builder) |
| 655 | || executeNumber<IndexType, Int8>(block, arguments, result, index_data, builder) |
| 656 | || executeNumber<IndexType, Int16>(block, arguments, result, index_data, builder) |
| 657 | || executeNumber<IndexType, Int32>(block, arguments, result, index_data, builder) |
| 658 | || executeNumber<IndexType, Int64>(block, arguments, result, index_data, builder) |
| 659 | || executeNumber<IndexType, Float32>(block, arguments, result, index_data, builder) |
| 660 | || executeNumber<IndexType, Float64>(block, arguments, result, index_data, builder) |
| 661 | || executeConst<IndexType>(block, arguments, result, index_data, builder, input_rows_count) |
| 662 | || executeString<IndexType>(block, arguments, result, index_data, builder) |
| 663 | || executeGeneric<IndexType>(block, arguments, result, index_data, builder))) |
| 664 | throw Exception("Illegal column " + block.getByPosition(arguments[0]).column->getName() |
| 665 | + " of first argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); |
| 666 | |
| 667 | return true; |
| 668 | } |
| 669 | |
| 670 | bool FunctionArrayElement::executeTuple(Block & block, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) |
| 671 | { |
| 672 | const ColumnArray * col_array = typeid_cast<const ColumnArray *>(block.getByPosition(arguments[0]).column.get()); |
| 673 | |
| 674 | if (!col_array) |
| 675 | return false; |
| 676 | |
| 677 | const ColumnTuple * col_nested = typeid_cast<const ColumnTuple *>(&col_array->getData()); |
| 678 | |
| 679 | if (!col_nested) |
| 680 | return false; |
| 681 | |
| 682 | const auto & tuple_columns = col_nested->getColumns(); |
| 683 | size_t tuple_size = tuple_columns.size(); |
| 684 | |
| 685 | const DataTypes & tuple_types = typeid_cast<const DataTypeTuple &>( |
| 686 | *typeid_cast<const DataTypeArray &>(*block.getByPosition(arguments[0]).type).getNestedType()).getElements(); |
| 687 | |
| 688 | /** We will calculate the function for the tuple of the internals of the array. |
| 689 | * To do this, create a temporary block. |
| 690 | * It will consist of the following columns |
| 691 | * - the index of the array to be taken; |
| 692 | * - an array of the first elements of the tuples; |
| 693 | * - the result of taking the elements by the index for an array of the first elements of the tuples; |
| 694 | * - array of the second elements of the tuples; |
| 695 | * - result of taking elements by index for an array of second elements of tuples; |
| 696 | * ... |
| 697 | */ |
| 698 | Block block_of_temporary_results; |
| 699 | block_of_temporary_results.insert(block.getByPosition(arguments[1])); |
| 700 | |
| 701 | /// results of taking elements by index for arrays from each element of the tuples; |
| 702 | Columns result_tuple_columns; |
| 703 | |
| 704 | for (size_t i = 0; i < tuple_size; ++i) |
| 705 | { |
| 706 | ColumnWithTypeAndName array_of_tuple_section; |
| 707 | array_of_tuple_section.column = ColumnArray::create(tuple_columns[i], col_array->getOffsetsPtr()); |
| 708 | array_of_tuple_section.type = std::make_shared<DataTypeArray>(tuple_types[i]); |
| 709 | block_of_temporary_results.insert(array_of_tuple_section); |
| 710 | |
| 711 | ColumnWithTypeAndName array_elements_of_tuple_section; |
| 712 | array_elements_of_tuple_section.type = getReturnTypeImpl( |
| 713 | {block_of_temporary_results.getByPosition(i * 2 + 1).type, block_of_temporary_results.getByPosition(0).type}); |
| 714 | block_of_temporary_results.insert(array_elements_of_tuple_section); |
| 715 | |
| 716 | executeImpl(block_of_temporary_results, ColumnNumbers{i * 2 + 1, 0}, i * 2 + 2, input_rows_count); |
| 717 | |
| 718 | result_tuple_columns.emplace_back(std::move(block_of_temporary_results.getByPosition(i * 2 + 2).column)); |
| 719 | } |
| 720 | |
| 721 | block.getByPosition(result).column = ColumnTuple::create(result_tuple_columns); |
| 722 | |
| 723 | return true; |
| 724 | } |
| 725 | |
| 726 | String FunctionArrayElement::getName() const |
| 727 | { |
| 728 | return name; |
| 729 | } |
| 730 | |
| 731 | DataTypePtr FunctionArrayElement::getReturnTypeImpl(const DataTypes & arguments) const |
| 732 | { |
| 733 | const DataTypeArray * array_type = checkAndGetDataType<DataTypeArray>(arguments[0].get()); |
| 734 | if (!array_type) |
| 735 | throw Exception("First argument for function " + getName() + " must be array." , ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); |
| 736 | |
| 737 | if (!isInteger(arguments[1])) |
| 738 | throw Exception("Second argument for function " + getName() + " must be integer." , ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); |
| 739 | |
| 740 | return array_type->getNestedType(); |
| 741 | } |
| 742 | |
| 743 | void FunctionArrayElement::executeImpl(Block & block, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) |
| 744 | { |
| 745 | /// Check nullability. |
| 746 | bool is_array_of_nullable = false; |
| 747 | |
| 748 | const ColumnArray * col_array = nullptr; |
| 749 | const ColumnArray * col_const_array = nullptr; |
| 750 | |
| 751 | col_array = checkAndGetColumn<ColumnArray>(block.getByPosition(arguments[0]).column.get()); |
| 752 | if (col_array) |
| 753 | is_array_of_nullable = isColumnNullable(col_array->getData()); |
| 754 | else |
| 755 | { |
| 756 | col_const_array = checkAndGetColumnConstData<ColumnArray>(block.getByPosition(arguments[0]).column.get()); |
| 757 | if (col_const_array) |
| 758 | is_array_of_nullable = isColumnNullable(col_const_array->getData()); |
| 759 | else |
| 760 | throw Exception("Illegal column " + block.getByPosition(arguments[0]).column->getName() |
| 761 | + " of first argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); |
| 762 | } |
| 763 | |
| 764 | if (!is_array_of_nullable) |
| 765 | { |
| 766 | ArrayImpl::NullMapBuilder builder; |
| 767 | perform(block, arguments, result, builder, input_rows_count); |
| 768 | } |
| 769 | else |
| 770 | { |
| 771 | /// Perform initializations. |
| 772 | ArrayImpl::NullMapBuilder builder; |
| 773 | Block source_block; |
| 774 | |
| 775 | const auto & input_type = typeid_cast<const DataTypeNullable &>(*typeid_cast<const DataTypeArray &>(*block.getByPosition(arguments[0]).type).getNestedType()).getNestedType(); |
| 776 | const auto & tmp_ret_type = typeid_cast<const DataTypeNullable &>(*block.getByPosition(result).type).getNestedType(); |
| 777 | |
| 778 | if (col_array) |
| 779 | { |
| 780 | const auto & nullable_col = typeid_cast<const ColumnNullable &>(col_array->getData()); |
| 781 | const auto & nested_col = nullable_col.getNestedColumnPtr(); |
| 782 | |
| 783 | /// Put nested_col inside a ColumnArray. |
| 784 | source_block = |
| 785 | { |
| 786 | { |
| 787 | ColumnArray::create(nested_col, col_array->getOffsetsPtr()), |
| 788 | std::make_shared<DataTypeArray>(input_type), |
| 789 | "" |
| 790 | }, |
| 791 | block.getByPosition(arguments[1]), |
| 792 | { |
| 793 | nullptr, |
| 794 | tmp_ret_type, |
| 795 | "" |
| 796 | } |
| 797 | }; |
| 798 | |
| 799 | builder.initSource(nullable_col.getNullMapData().data()); |
| 800 | } |
| 801 | else |
| 802 | { |
| 803 | /// ColumnConst(ColumnArray(ColumnNullable(...))) |
| 804 | const auto & nullable_col = assert_cast<const ColumnNullable &>(col_const_array->getData()); |
| 805 | const auto & nested_col = nullable_col.getNestedColumnPtr(); |
| 806 | |
| 807 | source_block = |
| 808 | { |
| 809 | { |
| 810 | ColumnConst::create(ColumnArray::create(nested_col, col_const_array->getOffsetsPtr()), input_rows_count), |
| 811 | std::make_shared<DataTypeArray>(input_type), |
| 812 | "" |
| 813 | }, |
| 814 | block.getByPosition(arguments[1]), |
| 815 | { |
| 816 | nullptr, |
| 817 | tmp_ret_type, |
| 818 | "" |
| 819 | } |
| 820 | }; |
| 821 | |
| 822 | builder.initSource(nullable_col.getNullMapData().data()); |
| 823 | } |
| 824 | |
| 825 | perform(source_block, {0, 1}, 2, builder, input_rows_count); |
| 826 | |
| 827 | /// Store the result. |
| 828 | const ColumnWithTypeAndName & source_col = source_block.getByPosition(2); |
| 829 | ColumnWithTypeAndName & dest_col = block.getByPosition(result); |
| 830 | dest_col.column = ColumnNullable::create(source_col.column, builder ? std::move(builder).getNullMapColumnPtr() : ColumnUInt8::create()); |
| 831 | } |
| 832 | } |
| 833 | |
| 834 | void FunctionArrayElement::perform(Block & block, const ColumnNumbers & arguments, size_t result, |
| 835 | ArrayImpl::NullMapBuilder & builder, size_t input_rows_count) |
| 836 | { |
| 837 | if (executeTuple(block, arguments, result, input_rows_count)) |
| 838 | { |
| 839 | } |
| 840 | else if (!isColumnConst(*block.getByPosition(arguments[1]).column)) |
| 841 | { |
| 842 | if (!(executeArgument<UInt8>(block, arguments, result, builder, input_rows_count) |
| 843 | || executeArgument<UInt16>(block, arguments, result, builder, input_rows_count) |
| 844 | || executeArgument<UInt32>(block, arguments, result, builder, input_rows_count) |
| 845 | || executeArgument<UInt64>(block, arguments, result, builder, input_rows_count) |
| 846 | || executeArgument<Int8>(block, arguments, result, builder, input_rows_count) |
| 847 | || executeArgument<Int16>(block, arguments, result, builder, input_rows_count) |
| 848 | || executeArgument<Int32>(block, arguments, result, builder, input_rows_count) |
| 849 | || executeArgument<Int64>(block, arguments, result, builder, input_rows_count))) |
| 850 | throw Exception("Second argument for function " + getName() + " must must have UInt or Int type." , |
| 851 | ErrorCodes::ILLEGAL_COLUMN); |
| 852 | } |
| 853 | else |
| 854 | { |
| 855 | Field index = (*block.getByPosition(arguments[1]).column)[0]; |
| 856 | |
| 857 | if (builder) |
| 858 | builder.initSink(input_rows_count); |
| 859 | |
| 860 | if (index == 0u) |
| 861 | throw Exception("Array indices are 1-based" , ErrorCodes::ZERO_ARRAY_OR_TUPLE_INDEX); |
| 862 | |
| 863 | if (!(executeNumberConst<UInt8>(block, arguments, result, index, builder) |
| 864 | || executeNumberConst<UInt16>(block, arguments, result, index, builder) |
| 865 | || executeNumberConst<UInt32>(block, arguments, result, index, builder) |
| 866 | || executeNumberConst<UInt64>(block, arguments, result, index, builder) |
| 867 | || executeNumberConst<Int8>(block, arguments, result, index, builder) |
| 868 | || executeNumberConst<Int16>(block, arguments, result, index, builder) |
| 869 | || executeNumberConst<Int32>(block, arguments, result, index, builder) |
| 870 | || executeNumberConst<Int64>(block, arguments, result, index, builder) |
| 871 | || executeNumberConst<Float32>(block, arguments, result, index, builder) |
| 872 | || executeNumberConst<Float64>(block, arguments, result, index, builder) |
| 873 | || executeStringConst (block, arguments, result, index, builder) |
| 874 | || executeGenericConst (block, arguments, result, index, builder))) |
| 875 | throw Exception("Illegal column " + block.getByPosition(arguments[0]).column->getName() |
| 876 | + " of first argument of function " + getName(), |
| 877 | ErrorCodes::ILLEGAL_COLUMN); |
| 878 | } |
| 879 | } |
| 880 | |
| 881 | |
| 882 | void registerFunctionArrayElement(FunctionFactory & factory) |
| 883 | { |
| 884 | factory.registerFunction<FunctionArrayElement>(); |
| 885 | } |
| 886 | |
| 887 | } |
| 888 | |