duckdb 0.7.2-dev16.0 → 0.7.2-dev314.0

package/src/duckdb/src/storage/statistics/list_stats.cpp

@@ -0,0 +1,117 @@
+#include "duckdb/storage/statistics/list_stats.hpp"
+#include "duckdb/storage/statistics/base_statistics.hpp"
+#include "duckdb/common/field_writer.hpp"
+#include "duckdb/common/string_util.hpp"
+#include "duckdb/common/types/vector.hpp"
+
+namespace duckdb {
+
+void ListStats::Construct(BaseStatistics &stats) {
+	stats.child_stats = unique_ptr<BaseStatistics[]>(new BaseStatistics[1]);
+	BaseStatistics::Construct(stats.child_stats[0], ListType::GetChildType(stats.GetType()));
+}
+
+BaseStatistics ListStats::CreateUnknown(LogicalType type) {
+	auto &child_type = ListType::GetChildType(type);
+	BaseStatistics result(std::move(type));
+	result.InitializeUnknown();
+	result.child_stats[0].Copy(BaseStatistics::CreateUnknown(child_type));
+	return result;
+}
+
+BaseStatistics ListStats::CreateEmpty(LogicalType type) {
+	auto &child_type = ListType::GetChildType(type);
+	BaseStatistics result(std::move(type));
+	result.InitializeEmpty();
+	result.child_stats[0].Copy(BaseStatistics::CreateEmpty(child_type));
+	return result;
+}
+
+void ListStats::Copy(BaseStatistics &stats, const BaseStatistics &other) {
+	D_ASSERT(stats.child_stats);
+	D_ASSERT(other.child_stats);
+	stats.child_stats[0].Copy(other.child_stats[0]);
+}
+
+const BaseStatistics &ListStats::GetChildStats(const BaseStatistics &stats) {
+	D_ASSERT(stats.GetStatsType() == StatisticsType::LIST_STATS);
+	D_ASSERT(stats.child_stats);
+	return stats.child_stats[0];
+}
+BaseStatistics &ListStats::GetChildStats(BaseStatistics &stats) {
+	D_ASSERT(stats.GetStatsType() == StatisticsType::LIST_STATS);
+	D_ASSERT(stats.child_stats);
+	return stats.child_stats[0];
+}
+
+void ListStats::SetChildStats(BaseStatistics &stats, unique_ptr<BaseStatistics> new_stats) {
+	if (!new_stats) {
+		stats.child_stats[0].Copy(BaseStatistics::CreateUnknown(ListType::GetChildType(stats.GetType())));
+	} else {
+		stats.child_stats[0].Copy(*new_stats);
+	}
+}
+
+void ListStats::Merge(BaseStatistics &stats, const BaseStatistics &other) {
+	if (other.GetType().id() == LogicalTypeId::VALIDITY) {
+		return;
+	}
+
+	auto &child_stats = ListStats::GetChildStats(stats);
+	auto &other_child_stats = ListStats::GetChildStats(other);
+	child_stats.Merge(other_child_stats);
+}
+
+void ListStats::Serialize(const BaseStatistics &stats, FieldWriter &writer) {
+	auto &child_stats = ListStats::GetChildStats(stats);
+	writer.WriteSerializable(child_stats);
+}
+
+BaseStatistics ListStats::Deserialize(FieldReader &reader, LogicalType type) {
+	D_ASSERT(type.InternalType() == PhysicalType::LIST);
+	auto &child_type = ListType::GetChildType(type);
+	BaseStatistics result(std::move(type));
+	result.child_stats[0].Copy(reader.ReadRequiredSerializable<BaseStatistics, BaseStatistics>(child_type));
+	return result;
+}
+
+string ListStats::ToString(const BaseStatistics &stats) {
+	auto &child_stats = ListStats::GetChildStats(stats);
+	return StringUtil::Format("[%s]", child_stats.ToString());
+}
+
+void ListStats::Verify(const BaseStatistics &stats, Vector &vector, const SelectionVector &sel, idx_t count) {
+	auto &child_stats = ListStats::GetChildStats(stats);
+	auto &child_entry = ListVector::GetEntry(vector);
+	UnifiedVectorFormat vdata;
+	vector.ToUnifiedFormat(count, vdata);
+
+	auto list_data = (list_entry_t *)vdata.data;
+	idx_t total_list_count = 0;
+	for (idx_t i = 0; i < count; i++) {
+		auto idx = sel.get_index(i);
+		auto index = vdata.sel->get_index(idx);
+		auto list = list_data[index];
+		if (vdata.validity.RowIsValid(index)) {
+			for (idx_t list_idx = 0; list_idx < list.length; list_idx++) {
+				total_list_count++;
+			}
+		}
+	}
+	SelectionVector list_sel(total_list_count);
+	idx_t list_count = 0;
+	for (idx_t i = 0; i < count; i++) {
+		auto idx = sel.get_index(i);
+		auto index = vdata.sel->get_index(idx);
+		auto list = list_data[index];
+		if (vdata.validity.RowIsValid(index)) {
+			for (idx_t list_idx = 0; list_idx < list.length; list_idx++) {
+				list_sel.set_index(list_count++, list.offset + list_idx);
+			}
+		}
+	}
+
+	child_stats.Verify(child_entry, list_sel, list_count);
+}
+
+} // namespace duckdb

package/src/duckdb/src/storage/statistics/numeric_stats.cpp

@@ -0,0 +1,529 @@
+#include "duckdb/storage/statistics/numeric_stats.hpp"
+#include "duckdb/storage/statistics/base_statistics.hpp"
+#include "duckdb/common/field_writer.hpp"
+#include "duckdb/common/types/vector.hpp"
+
+namespace duckdb {
+
+template <>
+void NumericStats::Update<interval_t>(BaseStatistics &stats, interval_t new_value) {
+}
+
+template <>
+void NumericStats::Update<list_entry_t>(BaseStatistics &stats, list_entry_t new_value) {
+}
+
+BaseStatistics NumericStats::CreateUnknown(LogicalType type) {
+	BaseStatistics result(std::move(type));
+	result.InitializeUnknown();
+	SetMin(result, Value(result.GetType()));
+	SetMax(result, Value(result.GetType()));
+	return result;
+}
+
+BaseStatistics NumericStats::CreateEmpty(LogicalType type) {
+	BaseStatistics result(std::move(type));
+	result.InitializeEmpty();
+	SetMin(result, Value::MaximumValue(result.GetType()));
+	SetMax(result, Value::MinimumValue(result.GetType()));
+	return result;
+}
+
+NumericStatsData &NumericStats::GetDataUnsafe(BaseStatistics &stats) {
+	D_ASSERT(stats.GetStatsType() == StatisticsType::NUMERIC_STATS);
+	return stats.stats_union.numeric_data;
+}
+
+const NumericStatsData &NumericStats::GetDataUnsafe(const BaseStatistics &stats) {
+	D_ASSERT(stats.GetStatsType() == StatisticsType::NUMERIC_STATS);
+	return stats.stats_union.numeric_data;
+}
+
+void NumericStats::Merge(BaseStatistics &stats, const BaseStatistics &other) {
+	if (other.GetType().id() == LogicalTypeId::VALIDITY) {
+		return;
+	}
+	D_ASSERT(stats.GetType() == other.GetType());
+	if (NumericStats::HasMin(other) && NumericStats::HasMin(stats)) {
+		auto other_min = NumericStats::Min(other);
+		if (other_min < NumericStats::Min(stats)) {
+			NumericStats::SetMin(stats, other_min);
+		}
+	} else {
+		NumericStats::SetMin(stats, Value());
+	}
+	if (NumericStats::HasMax(other) && NumericStats::HasMax(stats)) {
+		auto other_max = NumericStats::Max(other);
+		if (other_max > NumericStats::Max(stats)) {
+			NumericStats::SetMax(stats, other_max);
+		}
+	} else {
+		NumericStats::SetMax(stats, Value());
+	}
+}
+
+FilterPropagateResult NumericStats::CheckZonemap(const BaseStatistics &stats, ExpressionType comparison_type,
+                                                 const Value &constant) {
+	if (constant.IsNull()) {
+		return FilterPropagateResult::FILTER_ALWAYS_FALSE;
+	}
+	if (!NumericStats::HasMinMax(stats)) {
+		return FilterPropagateResult::NO_PRUNING_POSSIBLE;
+	}
+	auto min_value = NumericStats::Min(stats);
+	auto max_value = NumericStats::Max(stats);
+	switch (comparison_type) {
+	case ExpressionType::COMPARE_EQUAL:
+		if (constant == min_value && constant == max_value) {
+			return FilterPropagateResult::FILTER_ALWAYS_TRUE;
+		} else if (constant >= min_value && constant <= max_value) {
+			return FilterPropagateResult::NO_PRUNING_POSSIBLE;
+		} else {
+			return FilterPropagateResult::FILTER_ALWAYS_FALSE;
+		}
+	case ExpressionType::COMPARE_NOTEQUAL:
+		if (constant < min_value || constant > max_value) {
+			return FilterPropagateResult::FILTER_ALWAYS_TRUE;
+		} else if (min_value == max_value && min_value == constant) {
+			// corner case of a cluster with one numeric equal to the target constant
+			return FilterPropagateResult::FILTER_ALWAYS_FALSE;
+		}
+		return FilterPropagateResult::NO_PRUNING_POSSIBLE;
+	case ExpressionType::COMPARE_GREATERTHANOREQUALTO:
+		// X >= C
+		// this can be true only if max(X) >= C
+		// if min(X) >= C, then this is always true
+		if (min_value >= constant) {
+			return FilterPropagateResult::FILTER_ALWAYS_TRUE;
+		} else if (max_value >= constant) {
+			return FilterPropagateResult::NO_PRUNING_POSSIBLE;
+		} else {
+			return FilterPropagateResult::FILTER_ALWAYS_FALSE;
+		}
+	case ExpressionType::COMPARE_GREATERTHAN:
+		// X > C
+		// this can be true only if max(X) > C
+		// if min(X) > C, then this is always true
+		if (min_value > constant) {
+			return FilterPropagateResult::FILTER_ALWAYS_TRUE;
+		} else if (max_value > constant) {
+			return FilterPropagateResult::NO_PRUNING_POSSIBLE;
+		} else {
+			return FilterPropagateResult::FILTER_ALWAYS_FALSE;
+		}
+	case ExpressionType::COMPARE_LESSTHANOREQUALTO:
+		// X <= C
+		// this can be true only if min(X) <= C
+		// if max(X) <= C, then this is always true
+		if (max_value <= constant) {
+			return FilterPropagateResult::FILTER_ALWAYS_TRUE;
+		} else if (min_value <= constant) {
+			return FilterPropagateResult::NO_PRUNING_POSSIBLE;
+		} else {
+			return FilterPropagateResult::FILTER_ALWAYS_FALSE;
+		}
+	case ExpressionType::COMPARE_LESSTHAN:
+		// X < C
+		// this can be true only if min(X) < C
+		// if max(X) < C, then this is always true
+		if (max_value < constant) {
+			return FilterPropagateResult::FILTER_ALWAYS_TRUE;
+		} else if (min_value < constant) {
+			return FilterPropagateResult::NO_PRUNING_POSSIBLE;
+		} else {
+			return FilterPropagateResult::FILTER_ALWAYS_FALSE;
+		}
+	default:
+		throw InternalException("Expression type in zonemap check not implemented");
+	}
+}
+
+bool NumericStats::IsConstant(const BaseStatistics &stats) {
+	return NumericStats::Max(stats) <= NumericStats::Min(stats);
+}
+
+void SetNumericValueInternal(const Value &input, const LogicalType &type, NumericValueUnion &val, bool &has_val) {
+	if (input.IsNull()) {
+		has_val = false;
+		return;
+	}
+	if (input.type().InternalType() != type.InternalType()) {
+		throw InternalException("SetMin or SetMax called with Value that does not match statistics' column value");
+	}
+	has_val = true;
+	switch (type.InternalType()) {
+	case PhysicalType::BOOL:
+		val.value_.boolean = BooleanValue::Get(input);
+		break;
+	case PhysicalType::INT8:
+		val.value_.tinyint = TinyIntValue::Get(input);
+		break;
+	case PhysicalType::INT16:
+		val.value_.smallint = SmallIntValue::Get(input);
+		break;
+	case PhysicalType::INT32:
+		val.value_.integer = IntegerValue::Get(input);
+		break;
+	case PhysicalType::INT64:
+		val.value_.bigint = BigIntValue::Get(input);
+		break;
+	case PhysicalType::UINT8:
+		val.value_.utinyint = UTinyIntValue::Get(input);
+		break;
+	case PhysicalType::UINT16:
+		val.value_.usmallint = USmallIntValue::Get(input);
+		break;
+	case PhysicalType::UINT32:
+		val.value_.uinteger = UIntegerValue::Get(input);
+		break;
+	case PhysicalType::UINT64:
+		val.value_.ubigint = UBigIntValue::Get(input);
+		break;
+	case PhysicalType::INT128:
+		val.value_.hugeint = HugeIntValue::Get(input);
+		break;
+	case PhysicalType::FLOAT:
+		val.value_.float_ = FloatValue::Get(input);
+		break;
+	case PhysicalType::DOUBLE:
+		val.value_.double_ = DoubleValue::Get(input);
+		break;
+	default:
+		throw InternalException("Unsupported type for NumericStatistics::SetValueInternal");
+	}
+}
+
+void NumericStats::SetMin(BaseStatistics &stats, const Value &new_min) {
+	auto &data = NumericStats::GetDataUnsafe(stats);
+	SetNumericValueInternal(new_min, stats.GetType(), data.min, data.has_min);
+}
+
+void NumericStats::SetMax(BaseStatistics &stats, const Value &new_max) {
+	auto &data = NumericStats::GetDataUnsafe(stats);
+	SetNumericValueInternal(new_max, stats.GetType(), data.max, data.has_max);
+}
+
+Value NumericValueUnionToValueInternal(const LogicalType &type, const NumericValueUnion &val) {
+	switch (type.InternalType()) {
+	case PhysicalType::BOOL:
+		return Value::BOOLEAN(val.value_.boolean);
+	case PhysicalType::INT8:
+		return Value::TINYINT(val.value_.tinyint);
+	case PhysicalType::INT16:
+		return Value::SMALLINT(val.value_.smallint);
+	case PhysicalType::INT32:
+		return Value::INTEGER(val.value_.integer);
+	case PhysicalType::INT64:
+		return Value::BIGINT(val.value_.bigint);
+	case PhysicalType::UINT8:
+		return Value::UTINYINT(val.value_.utinyint);
+	case PhysicalType::UINT16:
+		return Value::USMALLINT(val.value_.usmallint);
+	case PhysicalType::UINT32:
+		return Value::UINTEGER(val.value_.uinteger);
+	case PhysicalType::UINT64:
+		return Value::UBIGINT(val.value_.ubigint);
+	case PhysicalType::INT128:
+		return Value::HUGEINT(val.value_.hugeint);
+	case PhysicalType::FLOAT:
+		return Value::FLOAT(val.value_.float_);
+	case PhysicalType::DOUBLE:
+		return Value::DOUBLE(val.value_.double_);
+	default:
+		throw InternalException("Unsupported type for NumericValueUnionToValue");
+	}
+}
+
+Value NumericValueUnionToValue(const LogicalType &type, const NumericValueUnion &val) {
+	Value result = NumericValueUnionToValueInternal(type, val);
+	result.GetTypeMutable() = type;
+	return result;
+}
+
+bool NumericStats::HasMinMax(const BaseStatistics &stats) {
+	return NumericStats::HasMin(stats) && NumericStats::HasMax(stats);
+}
+
+bool NumericStats::HasMin(const BaseStatistics &stats) {
+	if (stats.GetType().id() == LogicalTypeId::SQLNULL) {
+		return false;
+	}
+	return NumericStats::GetDataUnsafe(stats).has_min;
+}
+
+bool NumericStats::HasMax(const BaseStatistics &stats) {
+	if (stats.GetType().id() == LogicalTypeId::SQLNULL) {
+		return false;
+	}
+	return NumericStats::GetDataUnsafe(stats).has_max;
+}
+
+Value NumericStats::Min(const BaseStatistics &stats) {
+	if (!NumericStats::HasMin(stats)) {
+		throw InternalException("Min() called on statistics that does not have min");
+	}
+	return NumericValueUnionToValue(stats.GetType(), NumericStats::GetDataUnsafe(stats).min);
+}
+
+Value NumericStats::Max(const BaseStatistics &stats) {
+	if (!NumericStats::HasMax(stats)) {
+		throw InternalException("Max() called on statistics that does not have max");
+	}
+	return NumericValueUnionToValue(stats.GetType(), NumericStats::GetDataUnsafe(stats).max);
+}
+
+Value NumericStats::MinOrNull(const BaseStatistics &stats) {
+	if (!NumericStats::HasMin(stats)) {
+		return Value(stats.GetType());
+	}
+	return NumericStats::Min(stats);
+}
+
+Value NumericStats::MaxOrNull(const BaseStatistics &stats) {
+	if (!NumericStats::HasMax(stats)) {
+		return Value(stats.GetType());
+	}
+	return NumericStats::Max(stats);
+}
+
+void SerializeNumericStatsValue(const LogicalType &type, NumericValueUnion val, bool has_value, FieldWriter &writer) {
+	writer.WriteField<bool>(!has_value);
+	if (!has_value) {
+		return;
+	}
+	switch (type.InternalType()) {
+	case PhysicalType::BOOL:
+		writer.WriteField<bool>(val.value_.boolean);
+		break;
+	case PhysicalType::INT8:
+		writer.WriteField<int8_t>(val.value_.tinyint);
+		break;
+	case PhysicalType::INT16:
+		writer.WriteField<int16_t>(val.value_.smallint);
+		break;
+	case PhysicalType::INT32:
+		writer.WriteField<int32_t>(val.value_.integer);
+		break;
+	case PhysicalType::INT64:
+		writer.WriteField<int64_t>(val.value_.bigint);
+		break;
+	case PhysicalType::UINT8:
+		writer.WriteField<int8_t>(val.value_.utinyint);
+		break;
+	case PhysicalType::UINT16:
+		writer.WriteField<int16_t>(val.value_.usmallint);
+		break;
+	case PhysicalType::UINT32:
+		writer.WriteField<int32_t>(val.value_.uinteger);
+		break;
+	case PhysicalType::UINT64:
+		writer.WriteField<int64_t>(val.value_.ubigint);
+		break;
+	case PhysicalType::INT128:
+		writer.WriteField<hugeint_t>(val.value_.hugeint);
+		break;
+	case PhysicalType::FLOAT:
+		writer.WriteField<float>(val.value_.float_);
+		break;
+	case PhysicalType::DOUBLE:
+		writer.WriteField<double>(val.value_.double_);
+		break;
+	default:
+		throw InternalException("Unsupported type for serializing numeric statistics");
+	}
+}
+
+void NumericStats::Serialize(const BaseStatistics &stats, FieldWriter &writer) {
+	auto &numeric_stats = NumericStats::GetDataUnsafe(stats);
+	SerializeNumericStatsValue(stats.GetType(), numeric_stats.min, numeric_stats.has_min, writer);
+	SerializeNumericStatsValue(stats.GetType(), numeric_stats.max, numeric_stats.has_max, writer);
+}
+
+void DeserializeNumericStatsValue(const LogicalType &type, FieldReader &reader, NumericValueUnion &result,
+                                  bool &has_stats) {
+	auto is_null = reader.ReadRequired<bool>();
+	if (is_null) {
+		has_stats = false;
+		return;
+	}
+	has_stats = true;
+	switch (type.InternalType()) {
+	case PhysicalType::BOOL:
+		result.value_.boolean = reader.ReadRequired<bool>();
+		break;
+	case PhysicalType::INT8:
+		result.value_.tinyint = reader.ReadRequired<int8_t>();
+		break;
+	case PhysicalType::INT16:
+		result.value_.smallint = reader.ReadRequired<int16_t>();
+		break;
+	case PhysicalType::INT32:
+		result.value_.integer = reader.ReadRequired<int32_t>();
+		break;
+	case PhysicalType::INT64:
+		result.value_.bigint = reader.ReadRequired<int64_t>();
+		break;
+	case PhysicalType::UINT8:
+		result.value_.utinyint = reader.ReadRequired<uint8_t>();
+		break;
+	case PhysicalType::UINT16:
+		result.value_.usmallint = reader.ReadRequired<uint16_t>();
+		break;
+	case PhysicalType::UINT32:
+		result.value_.uinteger = reader.ReadRequired<uint32_t>();
+		break;
+	case PhysicalType::UINT64:
+		result.value_.ubigint = reader.ReadRequired<uint64_t>();
+		break;
+	case PhysicalType::INT128:
+		result.value_.hugeint = reader.ReadRequired<hugeint_t>();
+		break;
+	case PhysicalType::FLOAT:
+		result.value_.float_ = reader.ReadRequired<float>();
+		break;
+	case PhysicalType::DOUBLE:
+		result.value_.double_ = reader.ReadRequired<double>();
+		break;
+	default:
+		throw InternalException("Unsupported type for deserializing numeric statistics");
+	}
+}
+
+BaseStatistics NumericStats::Deserialize(FieldReader &reader, LogicalType type) {
+	BaseStatistics result(std::move(type));
+	auto &numeric_stats = NumericStats::GetDataUnsafe(result);
+	DeserializeNumericStatsValue(result.GetType(), reader, numeric_stats.min, numeric_stats.has_min);
+	DeserializeNumericStatsValue(result.GetType(), reader, numeric_stats.max, numeric_stats.has_max);
+	return result;
+}
+
+string NumericStats::ToString(const BaseStatistics &stats) {
+	return StringUtil::Format("[Min: %s, Max: %s]", NumericStats::MinOrNull(stats).ToString(),
+	                          NumericStats::MaxOrNull(stats).ToString());
+}
+
+template <class T>
+void NumericStats::TemplatedVerify(const BaseStatistics &stats, Vector &vector, const SelectionVector &sel,
+                                   idx_t count) {
+	UnifiedVectorFormat vdata;
+	vector.ToUnifiedFormat(count, vdata);
+
+	auto data = (T *)vdata.data;
+	auto min_value = NumericStats::MinOrNull(stats);
+	auto max_value = NumericStats::MaxOrNull(stats);
+	for (idx_t i = 0; i < count; i++) {
+		auto idx = sel.get_index(i);
+		auto index = vdata.sel->get_index(idx);
+		if (!vdata.validity.RowIsValid(index)) {
+			continue;
+		}
+		if (!min_value.IsNull() && LessThan::Operation(data[index], min_value.GetValueUnsafe<T>())) { // LCOV_EXCL_START
+			throw InternalException("Statistics mismatch: value is smaller than min.\nStatistics: %s\nVector: %s",
+			                        stats.ToString(), vector.ToString(count));
+		} // LCOV_EXCL_STOP
+		if (!max_value.IsNull() && GreaterThan::Operation(data[index], max_value.GetValueUnsafe<T>())) {
+			throw InternalException("Statistics mismatch: value is bigger than max.\nStatistics: %s\nVector: %s",
+			                        stats.ToString(), vector.ToString(count));
+		}
+	}
+}
+
+void NumericStats::Verify(const BaseStatistics &stats, Vector &vector, const SelectionVector &sel, idx_t count) {
+	auto &type = stats.GetType();
+	switch (type.InternalType()) {
+	case PhysicalType::BOOL:
+		break;
+	case PhysicalType::INT8:
+		TemplatedVerify<int8_t>(stats, vector, sel, count);
+		break;
+	case PhysicalType::INT16:
+		TemplatedVerify<int16_t>(stats, vector, sel, count);
+		break;
+	case PhysicalType::INT32:
+		TemplatedVerify<int32_t>(stats, vector, sel, count);
+		break;
+	case PhysicalType::INT64:
+		TemplatedVerify<int64_t>(stats, vector, sel, count);
+		break;
+	case PhysicalType::UINT8:
+		TemplatedVerify<uint8_t>(stats, vector, sel, count);
+		break;
+	case PhysicalType::UINT16:
+		TemplatedVerify<uint16_t>(stats, vector, sel, count);
+		break;
+	case PhysicalType::UINT32:
+		TemplatedVerify<uint32_t>(stats, vector, sel, count);
+		break;
+	case PhysicalType::UINT64:
+		TemplatedVerify<uint64_t>(stats, vector, sel, count);
+		break;
+	case PhysicalType::INT128:
+		TemplatedVerify<hugeint_t>(stats, vector, sel, count);
+		break;
+	case PhysicalType::FLOAT:
+		TemplatedVerify<float>(stats, vector, sel, count);
+		break;
+	case PhysicalType::DOUBLE:
+		TemplatedVerify<double>(stats, vector, sel, count);
+		break;
+	default:
+		throw InternalException("Unsupported type %s for numeric statistics verify", type.ToString());
+	}
+}
+
+template <>
+int8_t &NumericValueUnion::GetReferenceUnsafe() {
+	return value_.tinyint;
+}
+
+template <>
+int16_t &NumericValueUnion::GetReferenceUnsafe() {
+	return value_.smallint;
+}
+
+template <>
+int32_t &NumericValueUnion::GetReferenceUnsafe() {
+	return value_.integer;
+}
+
+template <>
+int64_t &NumericValueUnion::GetReferenceUnsafe() {
+	return value_.bigint;
+}
+
+template <>
+hugeint_t &NumericValueUnion::GetReferenceUnsafe() {
+	return value_.hugeint;
+}
+
+template <>
+uint8_t &NumericValueUnion::GetReferenceUnsafe() {
+	return value_.utinyint;
+}
+
+template <>
+uint16_t &NumericValueUnion::GetReferenceUnsafe() {
+	return value_.usmallint;
+}
+
+template <>
+uint32_t &NumericValueUnion::GetReferenceUnsafe() {
+	return value_.uinteger;
+}
+
+template <>
+uint64_t &NumericValueUnion::GetReferenceUnsafe() {
+	return value_.ubigint;
+}
+
+template <>
+float &NumericValueUnion::GetReferenceUnsafe() {
+	return value_.float_;
+}
+
+template <>
+double &NumericValueUnion::GetReferenceUnsafe() {
+	return value_.double_;
+}
+
+} // namespace duckdb

package/src/duckdb/src/storage/statistics/segment_statistics.cpp

@@ -4,19 +4,10 @@
 
 namespace duckdb {
 
-SegmentStatistics::SegmentStatistics(LogicalType type) : type(std::move(type)) {
-	Reset();
+SegmentStatistics::SegmentStatistics(LogicalType type) : statistics(BaseStatistics::CreateEmpty(std::move(type))) {
 }
 
-SegmentStatistics::SegmentStatistics(LogicalType type, unique_ptr<BaseStatistics> stats)
-    : type(std::move(type)), statistics(std::move(stats)) {
-	if (!statistics) {
-		Reset();
-	}
-}
-
-void SegmentStatistics::Reset() {
-	statistics = BaseStatistics::CreateEmpty(type, StatisticsType::LOCAL_STATS);
+SegmentStatistics::SegmentStatistics(BaseStatistics stats) : statistics(std::move(stats)) {
 }
 
 } // namespace duckdb