| 1 | #include "duckdb/storage/statistics/numeric_stats.hpp" |
| 2 | #include "duckdb/storage/statistics/base_statistics.hpp" |
| 3 | #include "duckdb/common/field_writer.hpp" |
| 4 | #include "duckdb/common/types/vector.hpp" |
| 5 | #include "duckdb/common/operator/comparison_operators.hpp" |
| 6 | |
| 7 | namespace duckdb { |
| 8 | |
| 9 | template <> |
| 10 | void NumericStats::Update<interval_t>(BaseStatistics &stats, interval_t new_value) { |
| 11 | } |
| 12 | |
| 13 | template <> |
| 14 | void NumericStats::Update<list_entry_t>(BaseStatistics &stats, list_entry_t new_value) { |
| 15 | } |
| 16 | |
| 17 | //===--------------------------------------------------------------------===// |
| 18 | // NumericStats |
| 19 | //===--------------------------------------------------------------------===// |
| 20 | BaseStatistics NumericStats::CreateUnknown(LogicalType type) { |
| 21 | BaseStatistics result(std::move(type)); |
| 22 | result.InitializeUnknown(); |
| 23 | SetMin(stats&: result, val: Value(result.GetType())); |
| 24 | SetMax(stats&: result, val: Value(result.GetType())); |
| 25 | return result; |
| 26 | } |
| 27 | |
| 28 | BaseStatistics NumericStats::CreateEmpty(LogicalType type) { |
| 29 | BaseStatistics result(std::move(type)); |
| 30 | result.InitializeEmpty(); |
| 31 | SetMin(stats&: result, val: Value::MaximumValue(type: result.GetType())); |
| 32 | SetMax(stats&: result, val: Value::MinimumValue(type: result.GetType())); |
| 33 | return result; |
| 34 | } |
| 35 | |
| 36 | NumericStatsData &NumericStats::GetDataUnsafe(BaseStatistics &stats) { |
| 37 | D_ASSERT(stats.GetStatsType() == StatisticsType::NUMERIC_STATS); |
| 38 | return stats.stats_union.numeric_data; |
| 39 | } |
| 40 | |
| 41 | const NumericStatsData &NumericStats::GetDataUnsafe(const BaseStatistics &stats) { |
| 42 | D_ASSERT(stats.GetStatsType() == StatisticsType::NUMERIC_STATS); |
| 43 | return stats.stats_union.numeric_data; |
| 44 | } |
| 45 | |
| 46 | void NumericStats::Merge(BaseStatistics &stats, const BaseStatistics &other) { |
| 47 | if (other.GetType().id() == LogicalTypeId::VALIDITY) { |
| 48 | return; |
| 49 | } |
| 50 | D_ASSERT(stats.GetType() == other.GetType()); |
| 51 | if (NumericStats::HasMin(stats: other) && NumericStats::HasMin(stats)) { |
| 52 | auto other_min = NumericStats::Min(stats: other); |
| 53 | if (other_min < NumericStats::Min(stats)) { |
| 54 | NumericStats::SetMin(stats, val: other_min); |
| 55 | } |
| 56 | } else { |
| 57 | NumericStats::SetMin(stats, val: Value()); |
| 58 | } |
| 59 | if (NumericStats::HasMax(stats: other) && NumericStats::HasMax(stats)) { |
| 60 | auto other_max = NumericStats::Max(stats: other); |
| 61 | if (other_max > NumericStats::Max(stats)) { |
| 62 | NumericStats::SetMax(stats, val: other_max); |
| 63 | } |
| 64 | } else { |
| 65 | NumericStats::SetMax(stats, val: Value()); |
| 66 | } |
| 67 | } |
| 68 | |
| 69 | struct GetNumericValueUnion { |
| 70 | template <class T> |
| 71 | static T Operation(const NumericValueUnion &v); |
| 72 | }; |
| 73 | |
| 74 | template <> |
| 75 | int8_t GetNumericValueUnion::Operation(const NumericValueUnion &v) { |
| 76 | return v.value_.tinyint; |
| 77 | } |
| 78 | |
| 79 | template <> |
| 80 | int16_t GetNumericValueUnion::Operation(const NumericValueUnion &v) { |
| 81 | return v.value_.smallint; |
| 82 | } |
| 83 | |
| 84 | template <> |
| 85 | int32_t GetNumericValueUnion::Operation(const NumericValueUnion &v) { |
| 86 | return v.value_.integer; |
| 87 | } |
| 88 | |
| 89 | template <> |
| 90 | int64_t GetNumericValueUnion::Operation(const NumericValueUnion &v) { |
| 91 | return v.value_.bigint; |
| 92 | } |
| 93 | |
| 94 | template <> |
| 95 | hugeint_t GetNumericValueUnion::Operation(const NumericValueUnion &v) { |
| 96 | return v.value_.hugeint; |
| 97 | } |
| 98 | |
| 99 | template <> |
| 100 | uint8_t GetNumericValueUnion::Operation(const NumericValueUnion &v) { |
| 101 | return v.value_.utinyint; |
| 102 | } |
| 103 | |
| 104 | template <> |
| 105 | uint16_t GetNumericValueUnion::Operation(const NumericValueUnion &v) { |
| 106 | return v.value_.usmallint; |
| 107 | } |
| 108 | |
| 109 | template <> |
| 110 | uint32_t GetNumericValueUnion::Operation(const NumericValueUnion &v) { |
| 111 | return v.value_.uinteger; |
| 112 | } |
| 113 | |
| 114 | template <> |
| 115 | uint64_t GetNumericValueUnion::Operation(const NumericValueUnion &v) { |
| 116 | return v.value_.ubigint; |
| 117 | } |
| 118 | |
| 119 | template <> |
| 120 | float GetNumericValueUnion::Operation(const NumericValueUnion &v) { |
| 121 | return v.value_.float_; |
| 122 | } |
| 123 | |
| 124 | template <> |
| 125 | double GetNumericValueUnion::Operation(const NumericValueUnion &v) { |
| 126 | return v.value_.double_; |
| 127 | } |
| 128 | |
| 129 | template <class T> |
| 130 | T NumericStats::GetMinUnsafe(const BaseStatistics &stats) { |
| 131 | return GetNumericValueUnion::Operation<T>(NumericStats::GetDataUnsafe(stats).min); |
| 132 | } |
| 133 | |
| 134 | template <class T> |
| 135 | T NumericStats::GetMaxUnsafe(const BaseStatistics &stats) { |
| 136 | return GetNumericValueUnion::Operation<T>(NumericStats::GetDataUnsafe(stats).max); |
| 137 | } |
| 138 | |
| 139 | template <class T> |
| 140 | bool ConstantExactRange(T min, T max, T constant) { |
| 141 | return Equals::Operation(constant, min) && Equals::Operation(constant, max); |
| 142 | } |
| 143 | |
| 144 | template <class T> |
| 145 | bool ConstantValueInRange(T min, T max, T constant) { |
| 146 | return !(LessThan::Operation(constant, min) || GreaterThan::Operation(constant, max)); |
| 147 | } |
| 148 | |
| 149 | template <class T> |
| 150 | FilterPropagateResult CheckZonemapTemplated(const BaseStatistics &stats, ExpressionType comparison_type, |
| 151 | const Value &constant_value) { |
| 152 | T min_value = NumericStats::GetMinUnsafe<T>(stats); |
| 153 | T max_value = NumericStats::GetMaxUnsafe<T>(stats); |
| 154 | T constant = constant_value.GetValueUnsafe<T>(); |
| 155 | switch (comparison_type) { |
| 156 | case ExpressionType::COMPARE_EQUAL: |
| 157 | if (ConstantExactRange(min_value, max_value, constant)) { |
| 158 | return FilterPropagateResult::FILTER_ALWAYS_TRUE; |
| 159 | } |
| 160 | if (ConstantValueInRange(min_value, max_value, constant)) { |
| 161 | return FilterPropagateResult::NO_PRUNING_POSSIBLE; |
| 162 | } |
| 163 | return FilterPropagateResult::FILTER_ALWAYS_FALSE; |
| 164 | case ExpressionType::COMPARE_NOTEQUAL: |
| 165 | if (!ConstantValueInRange(min_value, max_value, constant)) { |
| 166 | return FilterPropagateResult::FILTER_ALWAYS_TRUE; |
| 167 | } else if (ConstantExactRange(min_value, max_value, constant)) { |
| 168 | // corner case of a cluster with one numeric equal to the target constant |
| 169 | return FilterPropagateResult::FILTER_ALWAYS_FALSE; |
| 170 | } |
| 171 | return FilterPropagateResult::NO_PRUNING_POSSIBLE; |
| 172 | case ExpressionType::COMPARE_GREATERTHANOREQUALTO: |
| 173 | // GreaterThanEquals::Operation(X, C) |
| 174 | // this can be true only if max(X) >= C |
| 175 | // if min(X) >= C, then this is always true |
| 176 | if (GreaterThanEquals::Operation(min_value, constant)) { |
| 177 | return FilterPropagateResult::FILTER_ALWAYS_TRUE; |
| 178 | } else if (GreaterThanEquals::Operation(max_value, constant)) { |
| 179 | return FilterPropagateResult::NO_PRUNING_POSSIBLE; |
| 180 | } else { |
| 181 | return FilterPropagateResult::FILTER_ALWAYS_FALSE; |
| 182 | } |
| 183 | case ExpressionType::COMPARE_GREATERTHAN: |
| 184 | // GreaterThan::Operation(X, C) |
| 185 | // this can be true only if max(X) > C |
| 186 | // if min(X) > C, then this is always true |
| 187 | if (GreaterThan::Operation(min_value, constant)) { |
| 188 | return FilterPropagateResult::FILTER_ALWAYS_TRUE; |
| 189 | } else if (GreaterThan::Operation(max_value, constant)) { |
| 190 | return FilterPropagateResult::NO_PRUNING_POSSIBLE; |
| 191 | } else { |
| 192 | return FilterPropagateResult::FILTER_ALWAYS_FALSE; |
| 193 | } |
| 194 | case ExpressionType::COMPARE_LESSTHANOREQUALTO: |
| 195 | // LessThanEquals::Operation(X, C) |
| 196 | // this can be true only if min(X) <= C |
| 197 | // if max(X) <= C, then this is always true |
| 198 | if (LessThanEquals::Operation(max_value, constant)) { |
| 199 | return FilterPropagateResult::FILTER_ALWAYS_TRUE; |
| 200 | } else if (LessThanEquals::Operation(min_value, constant)) { |
| 201 | return FilterPropagateResult::NO_PRUNING_POSSIBLE; |
| 202 | } else { |
| 203 | return FilterPropagateResult::FILTER_ALWAYS_FALSE; |
| 204 | } |
| 205 | case ExpressionType::COMPARE_LESSTHAN: |
| 206 | // LessThan::Operation(X, C) |
| 207 | // this can be true only if min(X) < C |
| 208 | // if max(X) < C, then this is always true |
| 209 | if (LessThan::Operation(max_value, constant)) { |
| 210 | return FilterPropagateResult::FILTER_ALWAYS_TRUE; |
| 211 | } else if (LessThan::Operation(min_value, constant)) { |
| 212 | return FilterPropagateResult::NO_PRUNING_POSSIBLE; |
| 213 | } else { |
| 214 | return FilterPropagateResult::FILTER_ALWAYS_FALSE; |
| 215 | } |
| 216 | default: |
| 217 | throw InternalException("Expression type in zonemap check not implemented" ); |
| 218 | } |
| 219 | } |
| 220 | |
| 221 | FilterPropagateResult NumericStats::CheckZonemap(const BaseStatistics &stats, ExpressionType comparison_type, |
| 222 | const Value &constant) { |
| 223 | D_ASSERT(constant.type() == stats.GetType()); |
| 224 | if (constant.IsNull()) { |
| 225 | return FilterPropagateResult::FILTER_ALWAYS_FALSE; |
| 226 | } |
| 227 | if (!NumericStats::HasMinMax(stats)) { |
| 228 | return FilterPropagateResult::NO_PRUNING_POSSIBLE; |
| 229 | } |
| 230 | switch (stats.GetType().InternalType()) { |
| 231 | case PhysicalType::INT8: |
| 232 | return CheckZonemapTemplated<int8_t>(stats, comparison_type, constant_value: constant); |
| 233 | case PhysicalType::INT16: |
| 234 | return CheckZonemapTemplated<int16_t>(stats, comparison_type, constant_value: constant); |
| 235 | case PhysicalType::INT32: |
| 236 | return CheckZonemapTemplated<int32_t>(stats, comparison_type, constant_value: constant); |
| 237 | case PhysicalType::INT64: |
| 238 | return CheckZonemapTemplated<int64_t>(stats, comparison_type, constant_value: constant); |
| 239 | case PhysicalType::UINT8: |
| 240 | return CheckZonemapTemplated<uint8_t>(stats, comparison_type, constant_value: constant); |
| 241 | case PhysicalType::UINT16: |
| 242 | return CheckZonemapTemplated<uint16_t>(stats, comparison_type, constant_value: constant); |
| 243 | case PhysicalType::UINT32: |
| 244 | return CheckZonemapTemplated<uint32_t>(stats, comparison_type, constant_value: constant); |
| 245 | case PhysicalType::UINT64: |
| 246 | return CheckZonemapTemplated<uint64_t>(stats, comparison_type, constant_value: constant); |
| 247 | case PhysicalType::INT128: |
| 248 | return CheckZonemapTemplated<hugeint_t>(stats, comparison_type, constant_value: constant); |
| 249 | case PhysicalType::FLOAT: |
| 250 | return CheckZonemapTemplated<float>(stats, comparison_type, constant_value: constant); |
| 251 | case PhysicalType::DOUBLE: |
| 252 | return CheckZonemapTemplated<double>(stats, comparison_type, constant_value: constant); |
| 253 | default: |
| 254 | throw InternalException("Unsupported type for NumericStats::CheckZonemap" ); |
| 255 | } |
| 256 | } |
| 257 | |
| 258 | bool NumericStats::IsConstant(const BaseStatistics &stats) { |
| 259 | return NumericStats::Max(stats) <= NumericStats::Min(stats); |
| 260 | } |
| 261 | |
| 262 | void SetNumericValueInternal(const Value &input, const LogicalType &type, NumericValueUnion &val, bool &has_val) { |
| 263 | if (input.IsNull()) { |
| 264 | has_val = false; |
| 265 | return; |
| 266 | } |
| 267 | if (input.type().InternalType() != type.InternalType()) { |
| 268 | throw InternalException("SetMin or SetMax called with Value that does not match statistics' column value" ); |
| 269 | } |
| 270 | has_val = true; |
| 271 | switch (type.InternalType()) { |
| 272 | case PhysicalType::BOOL: |
| 273 | val.value_.boolean = BooleanValue::Get(value: input); |
| 274 | break; |
| 275 | case PhysicalType::INT8: |
| 276 | val.value_.tinyint = TinyIntValue::Get(value: input); |
| 277 | break; |
| 278 | case PhysicalType::INT16: |
| 279 | val.value_.smallint = SmallIntValue::Get(value: input); |
| 280 | break; |
| 281 | case PhysicalType::INT32: |
| 282 | val.value_.integer = IntegerValue::Get(value: input); |
| 283 | break; |
| 284 | case PhysicalType::INT64: |
| 285 | val.value_.bigint = BigIntValue::Get(value: input); |
| 286 | break; |
| 287 | case PhysicalType::UINT8: |
| 288 | val.value_.utinyint = UTinyIntValue::Get(value: input); |
| 289 | break; |
| 290 | case PhysicalType::UINT16: |
| 291 | val.value_.usmallint = USmallIntValue::Get(value: input); |
| 292 | break; |
| 293 | case PhysicalType::UINT32: |
| 294 | val.value_.uinteger = UIntegerValue::Get(value: input); |
| 295 | break; |
| 296 | case PhysicalType::UINT64: |
| 297 | val.value_.ubigint = UBigIntValue::Get(value: input); |
| 298 | break; |
| 299 | case PhysicalType::INT128: |
| 300 | val.value_.hugeint = HugeIntValue::Get(value: input); |
| 301 | break; |
| 302 | case PhysicalType::FLOAT: |
| 303 | val.value_.float_ = FloatValue::Get(value: input); |
| 304 | break; |
| 305 | case PhysicalType::DOUBLE: |
| 306 | val.value_.double_ = DoubleValue::Get(value: input); |
| 307 | break; |
| 308 | default: |
| 309 | throw InternalException("Unsupported type for NumericStatistics::SetValueInternal" ); |
| 310 | } |
| 311 | } |
| 312 | |
| 313 | void NumericStats::SetMin(BaseStatistics &stats, const Value &new_min) { |
| 314 | auto &data = NumericStats::GetDataUnsafe(stats); |
| 315 | SetNumericValueInternal(input: new_min, type: stats.GetType(), val&: data.min, has_val&: data.has_min); |
| 316 | } |
| 317 | |
| 318 | void NumericStats::SetMax(BaseStatistics &stats, const Value &new_max) { |
| 319 | auto &data = NumericStats::GetDataUnsafe(stats); |
| 320 | SetNumericValueInternal(input: new_max, type: stats.GetType(), val&: data.max, has_val&: data.has_max); |
| 321 | } |
| 322 | |
| 323 | Value NumericValueUnionToValueInternal(const LogicalType &type, const NumericValueUnion &val) { |
| 324 | switch (type.InternalType()) { |
| 325 | case PhysicalType::BOOL: |
| 326 | return Value::BOOLEAN(value: val.value_.boolean); |
| 327 | case PhysicalType::INT8: |
| 328 | return Value::TINYINT(value: val.value_.tinyint); |
| 329 | case PhysicalType::INT16: |
| 330 | return Value::SMALLINT(value: val.value_.smallint); |
| 331 | case PhysicalType::INT32: |
| 332 | return Value::INTEGER(value: val.value_.integer); |
| 333 | case PhysicalType::INT64: |
| 334 | return Value::BIGINT(value: val.value_.bigint); |
| 335 | case PhysicalType::UINT8: |
| 336 | return Value::UTINYINT(value: val.value_.utinyint); |
| 337 | case PhysicalType::UINT16: |
| 338 | return Value::USMALLINT(value: val.value_.usmallint); |
| 339 | case PhysicalType::UINT32: |
| 340 | return Value::UINTEGER(value: val.value_.uinteger); |
| 341 | case PhysicalType::UINT64: |
| 342 | return Value::UBIGINT(value: val.value_.ubigint); |
| 343 | case PhysicalType::INT128: |
| 344 | return Value::HUGEINT(value: val.value_.hugeint); |
| 345 | case PhysicalType::FLOAT: |
| 346 | return Value::FLOAT(value: val.value_.float_); |
| 347 | case PhysicalType::DOUBLE: |
| 348 | return Value::DOUBLE(value: val.value_.double_); |
| 349 | default: |
| 350 | throw InternalException("Unsupported type for NumericValueUnionToValue" ); |
| 351 | } |
| 352 | } |
| 353 | |
| 354 | Value NumericValueUnionToValue(const LogicalType &type, const NumericValueUnion &val) { |
| 355 | Value result = NumericValueUnionToValueInternal(type, val); |
| 356 | result.GetTypeMutable() = type; |
| 357 | return result; |
| 358 | } |
| 359 | |
| 360 | bool NumericStats::HasMinMax(const BaseStatistics &stats) { |
| 361 | return NumericStats::HasMin(stats) && NumericStats::HasMax(stats); |
| 362 | } |
| 363 | |
| 364 | bool NumericStats::HasMin(const BaseStatistics &stats) { |
| 365 | if (stats.GetType().id() == LogicalTypeId::SQLNULL) { |
| 366 | return false; |
| 367 | } |
| 368 | return NumericStats::GetDataUnsafe(stats).has_min; |
| 369 | } |
| 370 | |
| 371 | bool NumericStats::HasMax(const BaseStatistics &stats) { |
| 372 | if (stats.GetType().id() == LogicalTypeId::SQLNULL) { |
| 373 | return false; |
| 374 | } |
| 375 | return NumericStats::GetDataUnsafe(stats).has_max; |
| 376 | } |
| 377 | |
| 378 | Value NumericStats::Min(const BaseStatistics &stats) { |
| 379 | if (!NumericStats::HasMin(stats)) { |
| 380 | throw InternalException("Min() called on statistics that does not have min" ); |
| 381 | } |
| 382 | return NumericValueUnionToValue(type: stats.GetType(), val: NumericStats::GetDataUnsafe(stats).min); |
| 383 | } |
| 384 | |
| 385 | Value NumericStats::Max(const BaseStatistics &stats) { |
| 386 | if (!NumericStats::HasMax(stats)) { |
| 387 | throw InternalException("Max() called on statistics that does not have max" ); |
| 388 | } |
| 389 | return NumericValueUnionToValue(type: stats.GetType(), val: NumericStats::GetDataUnsafe(stats).max); |
| 390 | } |
| 391 | |
| 392 | Value NumericStats::MinOrNull(const BaseStatistics &stats) { |
| 393 | if (!NumericStats::HasMin(stats)) { |
| 394 | return Value(stats.GetType()); |
| 395 | } |
| 396 | return NumericStats::Min(stats); |
| 397 | } |
| 398 | |
| 399 | Value NumericStats::MaxOrNull(const BaseStatistics &stats) { |
| 400 | if (!NumericStats::HasMax(stats)) { |
| 401 | return Value(stats.GetType()); |
| 402 | } |
| 403 | return NumericStats::Max(stats); |
| 404 | } |
| 405 | |
| 406 | void SerializeNumericStatsValue(const LogicalType &type, NumericValueUnion val, bool has_value, FieldWriter &writer) { |
| 407 | writer.WriteField<bool>(element: !has_value); |
| 408 | if (!has_value) { |
| 409 | return; |
| 410 | } |
| 411 | switch (type.InternalType()) { |
| 412 | case PhysicalType::BOOL: |
| 413 | writer.WriteField<bool>(element: val.value_.boolean); |
| 414 | break; |
| 415 | case PhysicalType::INT8: |
| 416 | writer.WriteField<int8_t>(element: val.value_.tinyint); |
| 417 | break; |
| 418 | case PhysicalType::INT16: |
| 419 | writer.WriteField<int16_t>(element: val.value_.smallint); |
| 420 | break; |
| 421 | case PhysicalType::INT32: |
| 422 | writer.WriteField<int32_t>(element: val.value_.integer); |
| 423 | break; |
| 424 | case PhysicalType::INT64: |
| 425 | writer.WriteField<int64_t>(element: val.value_.bigint); |
| 426 | break; |
| 427 | case PhysicalType::UINT8: |
| 428 | writer.WriteField<int8_t>(element: val.value_.utinyint); |
| 429 | break; |
| 430 | case PhysicalType::UINT16: |
| 431 | writer.WriteField<int16_t>(element: val.value_.usmallint); |
| 432 | break; |
| 433 | case PhysicalType::UINT32: |
| 434 | writer.WriteField<int32_t>(element: val.value_.uinteger); |
| 435 | break; |
| 436 | case PhysicalType::UINT64: |
| 437 | writer.WriteField<int64_t>(element: val.value_.ubigint); |
| 438 | break; |
| 439 | case PhysicalType::INT128: |
| 440 | writer.WriteField<hugeint_t>(element: val.value_.hugeint); |
| 441 | break; |
| 442 | case PhysicalType::FLOAT: |
| 443 | writer.WriteField<float>(element: val.value_.float_); |
| 444 | break; |
| 445 | case PhysicalType::DOUBLE: |
| 446 | writer.WriteField<double>(element: val.value_.double_); |
| 447 | break; |
| 448 | default: |
| 449 | throw InternalException("Unsupported type for serializing numeric statistics" ); |
| 450 | } |
| 451 | } |
| 452 | |
| 453 | void NumericStats::Serialize(const BaseStatistics &stats, FieldWriter &writer) { |
| 454 | auto &numeric_stats = NumericStats::GetDataUnsafe(stats); |
| 455 | SerializeNumericStatsValue(type: stats.GetType(), val: numeric_stats.min, has_value: numeric_stats.has_min, writer); |
| 456 | SerializeNumericStatsValue(type: stats.GetType(), val: numeric_stats.max, has_value: numeric_stats.has_max, writer); |
| 457 | } |
| 458 | |
| 459 | void DeserializeNumericStatsValue(const LogicalType &type, FieldReader &reader, NumericValueUnion &result, |
| 460 | bool &has_stats) { |
| 461 | auto is_null = reader.ReadRequired<bool>(); |
| 462 | if (is_null) { |
| 463 | has_stats = false; |
| 464 | return; |
| 465 | } |
| 466 | has_stats = true; |
| 467 | switch (type.InternalType()) { |
| 468 | case PhysicalType::BOOL: |
| 469 | result.value_.boolean = reader.ReadRequired<bool>(); |
| 470 | break; |
| 471 | case PhysicalType::INT8: |
| 472 | result.value_.tinyint = reader.ReadRequired<int8_t>(); |
| 473 | break; |
| 474 | case PhysicalType::INT16: |
| 475 | result.value_.smallint = reader.ReadRequired<int16_t>(); |
| 476 | break; |
| 477 | case PhysicalType::INT32: |
| 478 | result.value_.integer = reader.ReadRequired<int32_t>(); |
| 479 | break; |
| 480 | case PhysicalType::INT64: |
| 481 | result.value_.bigint = reader.ReadRequired<int64_t>(); |
| 482 | break; |
| 483 | case PhysicalType::UINT8: |
| 484 | result.value_.utinyint = reader.ReadRequired<uint8_t>(); |
| 485 | break; |
| 486 | case PhysicalType::UINT16: |
| 487 | result.value_.usmallint = reader.ReadRequired<uint16_t>(); |
| 488 | break; |
| 489 | case PhysicalType::UINT32: |
| 490 | result.value_.uinteger = reader.ReadRequired<uint32_t>(); |
| 491 | break; |
| 492 | case PhysicalType::UINT64: |
| 493 | result.value_.ubigint = reader.ReadRequired<uint64_t>(); |
| 494 | break; |
| 495 | case PhysicalType::INT128: |
| 496 | result.value_.hugeint = reader.ReadRequired<hugeint_t>(); |
| 497 | break; |
| 498 | case PhysicalType::FLOAT: |
| 499 | result.value_.float_ = reader.ReadRequired<float>(); |
| 500 | break; |
| 501 | case PhysicalType::DOUBLE: |
| 502 | result.value_.double_ = reader.ReadRequired<double>(); |
| 503 | break; |
| 504 | default: |
| 505 | throw InternalException("Unsupported type for deserializing numeric statistics" ); |
| 506 | } |
| 507 | } |
| 508 | |
| 509 | BaseStatistics NumericStats::Deserialize(FieldReader &reader, LogicalType type) { |
| 510 | BaseStatistics result(std::move(type)); |
| 511 | auto &numeric_stats = NumericStats::GetDataUnsafe(stats&: result); |
| 512 | DeserializeNumericStatsValue(type: result.GetType(), reader, result&: numeric_stats.min, has_stats&: numeric_stats.has_min); |
| 513 | DeserializeNumericStatsValue(type: result.GetType(), reader, result&: numeric_stats.max, has_stats&: numeric_stats.has_max); |
| 514 | return result; |
| 515 | } |
| 516 | |
| 517 | string NumericStats::ToString(const BaseStatistics &stats) { |
| 518 | return StringUtil::Format(fmt_str: "[Min: %s, Max: %s]" , params: NumericStats::MinOrNull(stats).ToString(), |
| 519 | params: NumericStats::MaxOrNull(stats).ToString()); |
| 520 | } |
| 521 | |
| 522 | template <class T> |
| 523 | void NumericStats::TemplatedVerify(const BaseStatistics &stats, Vector &vector, const SelectionVector &sel, |
| 524 | idx_t count) { |
| 525 | UnifiedVectorFormat vdata; |
| 526 | vector.ToUnifiedFormat(count, data&: vdata); |
| 527 | |
| 528 | auto data = UnifiedVectorFormat::GetData<T>(vdata); |
| 529 | auto min_value = NumericStats::MinOrNull(stats); |
| 530 | auto max_value = NumericStats::MaxOrNull(stats); |
| 531 | for (idx_t i = 0; i < count; i++) { |
| 532 | auto idx = sel.get_index(idx: i); |
| 533 | auto index = vdata.sel->get_index(idx); |
| 534 | if (!vdata.validity.RowIsValid(row_idx: index)) { |
| 535 | continue; |
| 536 | } |
| 537 | if (!min_value.IsNull() && LessThan::Operation(data[index], min_value.GetValueUnsafe<T>())) { // LCOV_EXCL_START |
| 538 | throw InternalException("Statistics mismatch: value is smaller than min.\nStatistics: %s\nVector: %s" , |
| 539 | stats.ToString(), vector.ToString(count)); |
| 540 | } // LCOV_EXCL_STOP |
| 541 | if (!max_value.IsNull() && GreaterThan::Operation(data[index], max_value.GetValueUnsafe<T>())) { |
| 542 | throw InternalException("Statistics mismatch: value is bigger than max.\nStatistics: %s\nVector: %s" , |
| 543 | stats.ToString(), vector.ToString(count)); |
| 544 | } |
| 545 | } |
| 546 | } |
| 547 | |
| 548 | void NumericStats::Verify(const BaseStatistics &stats, Vector &vector, const SelectionVector &sel, idx_t count) { |
| 549 | auto &type = stats.GetType(); |
| 550 | switch (type.InternalType()) { |
| 551 | case PhysicalType::BOOL: |
| 552 | break; |
| 553 | case PhysicalType::INT8: |
| 554 | TemplatedVerify<int8_t>(stats, vector, sel, count); |
| 555 | break; |
| 556 | case PhysicalType::INT16: |
| 557 | TemplatedVerify<int16_t>(stats, vector, sel, count); |
| 558 | break; |
| 559 | case PhysicalType::INT32: |
| 560 | TemplatedVerify<int32_t>(stats, vector, sel, count); |
| 561 | break; |
| 562 | case PhysicalType::INT64: |
| 563 | TemplatedVerify<int64_t>(stats, vector, sel, count); |
| 564 | break; |
| 565 | case PhysicalType::UINT8: |
| 566 | TemplatedVerify<uint8_t>(stats, vector, sel, count); |
| 567 | break; |
| 568 | case PhysicalType::UINT16: |
| 569 | TemplatedVerify<uint16_t>(stats, vector, sel, count); |
| 570 | break; |
| 571 | case PhysicalType::UINT32: |
| 572 | TemplatedVerify<uint32_t>(stats, vector, sel, count); |
| 573 | break; |
| 574 | case PhysicalType::UINT64: |
| 575 | TemplatedVerify<uint64_t>(stats, vector, sel, count); |
| 576 | break; |
| 577 | case PhysicalType::INT128: |
| 578 | TemplatedVerify<hugeint_t>(stats, vector, sel, count); |
| 579 | break; |
| 580 | case PhysicalType::FLOAT: |
| 581 | TemplatedVerify<float>(stats, vector, sel, count); |
| 582 | break; |
| 583 | case PhysicalType::DOUBLE: |
| 584 | TemplatedVerify<double>(stats, vector, sel, count); |
| 585 | break; |
| 586 | default: |
| 587 | throw InternalException("Unsupported type %s for numeric statistics verify" , type.ToString()); |
| 588 | } |
| 589 | } |
| 590 | |
| 591 | } // namespace duckdb |
| 592 | |