duckdb 0.7.1 → 0.7.2-dev1034.0

package/src/duckdb/src/optimizer/rule/ordered_aggregate_optimizer.cpp

@@ -0,0 +1,30 @@
+#include "duckdb/optimizer/rule/ordered_aggregate_optimizer.hpp"
+
+#include "duckdb/optimizer/matcher/expression_matcher.hpp"
+#include "duckdb/planner/expression/bound_aggregate_expression.hpp"
+
+namespace duckdb {
+
+OrderedAggregateOptimizer::OrderedAggregateOptimizer(ExpressionRewriter &rewriter) : Rule(rewriter) {
+	// we match on an OR expression within a LogicalFilter node
+	root = make_unique<ExpressionMatcher>();
+	root->expr_class = ExpressionClass::BOUND_AGGREGATE;
+}
+
+unique_ptr<Expression> OrderedAggregateOptimizer::Apply(LogicalOperator &op, vector<Expression *> &bindings,
+                                                        bool &changes_made, bool is_root) {
+	auto aggr = (BoundAggregateExpression *)bindings[0];
+	if (!aggr->order_bys) {
+		// no ORDER BYs defined
+		return nullptr;
+	}
+	if (aggr->function.order_dependent == AggregateOrderDependent::NOT_ORDER_DEPENDENT) {
+		// not an order dependent aggregate but we have an ORDER BY clause - remove it
+		aggr->order_bys.reset();
+		changes_made = true;
+		return nullptr;
+	}
+	return nullptr;
+}
+
+} // namespace duckdb

package/src/duckdb/src/optimizer/rule/regex_optimizations.cpp

@@ -35,7 +35,7 @@ unique_ptr<Expression> RegexOptimizationRule::Apply(LogicalOperator &op, vector<
 
 	auto constant_value = ExpressionExecutor::EvaluateScalar(GetContext(), *constant_expr);
 	D_ASSERT(constant_value.type() == constant_expr->return_type);
-	auto &patt_str = StringValue::Get(constant_value);
+	auto patt_str = StringValue::Get(constant_value);
 
 	duckdb_re2::RE2 pattern(patt_str);
 	if (!pattern.ok()) {
@@ -47,7 +47,14 @@ unique_ptr<Expression> RegexOptimizationRule::Apply(LogicalOperator &op, vector<
 		auto contains = make_unique<BoundFunctionExpression>(root->return_type, ContainsFun::GetFunction(),
 		                                                     std::move(root->children), nullptr);
 
-		contains->children[1] = make_unique<BoundConstantExpression>(Value(patt_str));
+		string min;
+		string max;
+		pattern.PossibleMatchRange(&min, &max, patt_str.size());
+		if (min == max) {
+			contains->children[1] = make_unique<BoundConstantExpression>(Value(std::move(min)));
+		} else {
+			contains->children[1] = make_unique<BoundConstantExpression>(Value(std::move(patt_str)));
+		}
 		return std::move(contains);
 	}
 	return nullptr;

package/src/duckdb/src/optimizer/statistics/expression/propagate_aggregate.cpp

@@ -5,15 +5,21 @@ namespace duckdb {
 
 unique_ptr<BaseStatistics> StatisticsPropagator::PropagateExpression(BoundAggregateExpression &aggr,
                                                                      unique_ptr<Expression> *expr_ptr) {
-	vector<unique_ptr<BaseStatistics>> stats;
+	vector<BaseStatistics> stats;
 	stats.reserve(aggr.children.size());
 	for (auto &child : aggr.children) {
-		stats.push_back(PropagateExpression(child));
+		auto stat = PropagateExpression(child);
+		if (!stat) {
+			stats.push_back(BaseStatistics::CreateUnknown(child->return_type));
+		} else {
+			stats.push_back(stat->Copy());
+		}
 	}
 	if (!aggr.function.statistics) {
 		return nullptr;
 	}
-	return aggr.function.statistics(context, aggr, aggr.bind_info.get(), stats, node_stats.get());
+	AggregateStatisticsInput input(aggr.bind_info.get(), stats, node_stats.get());
+	return aggr.function.statistics(context, aggr, input);
 }
 
 } // namespace duckdb

package/src/duckdb/src/optimizer/statistics/expression/propagate_and_compress.cpp

@@ -5,7 +5,6 @@
 #include "duckdb/planner/expression/bound_constant_expression.hpp"
 #include "duckdb/planner/expression/bound_function_expression.hpp"
 #include "duckdb/storage/statistics/base_statistics.hpp"
-#include "duckdb/storage/statistics/numeric_statistics.hpp"
 #include "duckdb/common/operator/subtract.hpp"
 
 namespace duckdb {
@@ -44,14 +43,14 @@ bool GetCastType(hugeint_t range, LogicalType &cast_type) {
 }
 
 template <class T>
-unique_ptr<Expression> TemplatedCastToSmallestType(unique_ptr<Expression> expr, NumericStatistics &num_stats) {
+unique_ptr<Expression> TemplatedCastToSmallestType(unique_ptr<Expression> expr, BaseStatistics &stats) {
 	// Compute range
-	if (num_stats.min.IsNull() || num_stats.max.IsNull()) {
+	if (!NumericStats::HasMinMax(stats)) {
 		return expr;
 	}
 
-	auto signed_min_val = num_stats.min.GetValue<T>();
-	auto signed_max_val = num_stats.max.GetValue<T>();
+	auto signed_min_val = NumericStats::Min(stats).GetValue<T>();
+	auto signed_max_val = NumericStats::Max(stats).GetValue<T>();
 	if (signed_max_val < signed_min_val) {
 		return expr;
 	}
@@ -82,7 +81,7 @@ unique_ptr<Expression> TemplatedCastToSmallestType(unique_ptr<Expression> expr,
 	return BoundCastExpression::AddDefaultCastToType(std::move(minus_expr), cast_type);
 }
 
-unique_ptr<Expression> CastToSmallestType(unique_ptr<Expression> expr, NumericStatistics &num_stats) {
+unique_ptr<Expression> CastToSmallestType(unique_ptr<Expression> expr, BaseStatistics &num_stats) {
 	auto physical_type = expr->return_type.InternalType();
 	switch (physical_type) {
 	case PhysicalType::UINT8:
@@ -111,7 +110,7 @@ void StatisticsPropagator::PropagateAndCompress(unique_ptr<Expression> &expr, un
 	stats = PropagateExpression(expr);
 	if (stats) {
 		if (expr->return_type.IsIntegral()) {
-			expr = CastToSmallestType(std::move(expr), (NumericStatistics &)*stats);
+			expr = CastToSmallestType(std::move(expr), *stats);
 		}
 	}
 }

package/src/duckdb/src/optimizer/statistics/expression/propagate_cast.cpp

@@ -1,24 +1,27 @@
 #include "duckdb/optimizer/statistics_propagator.hpp"
 #include "duckdb/planner/expression/bound_cast_expression.hpp"
-#include "duckdb/storage/statistics/numeric_statistics.hpp"
 
 namespace duckdb {
 
-static unique_ptr<BaseStatistics> StatisticsOperationsNumericNumericCast(const BaseStatistics *input_p,
+static unique_ptr<BaseStatistics> StatisticsOperationsNumericNumericCast(const BaseStatistics &input,
                                                                          const LogicalType &target) {
-	auto &input = (NumericStatistics &)*input_p;
-
-	Value min = input.min, max = input.max;
+	if (!NumericStats::HasMinMax(input)) {
+		return nullptr;
+	}
+	Value min = NumericStats::Min(input);
+	Value max = NumericStats::Max(input);
 	if (!min.DefaultTryCastAs(target) || !max.DefaultTryCastAs(target)) {
 		// overflow in cast: bailout
 		return nullptr;
 	}
-	auto stats = make_unique<NumericStatistics>(target, std::move(min), std::move(max), input.stats_type);
-	stats->CopyBase(*input_p);
-	return std::move(stats);
+	auto result = NumericStats::CreateEmpty(target);
+	result.CopyBase(input);
+	NumericStats::SetMin(result, min);
+	NumericStats::SetMax(result, max);
+	return result.ToUnique();
 }
 
-static unique_ptr<BaseStatistics> StatisticsNumericCastSwitch(const BaseStatistics *input, const LogicalType &target) {
+static unique_ptr<BaseStatistics> StatisticsNumericCastSwitch(const BaseStatistics &input, const LogicalType &target) {
 	switch (target.InternalType()) {
 	case PhysicalType::INT8:
 	case PhysicalType::INT16:
@@ -48,13 +51,13 @@ unique_ptr<BaseStatistics> StatisticsPropagator::PropagateExpression(BoundCastEx
 	case PhysicalType::INT128:
 	case PhysicalType::FLOAT:
 	case PhysicalType::DOUBLE:
-		result_stats = StatisticsNumericCastSwitch(child_stats.get(), cast.return_type);
+		result_stats = StatisticsNumericCastSwitch(*child_stats, cast.return_type);
 		break;
 	default:
 		return nullptr;
 	}
 	if (cast.try_cast && result_stats) {
-		result_stats->validity_stats = make_unique<ValidityStatistics>(true, true);
+		result_stats->Set(StatsInfo::CAN_HAVE_NULL_VALUES);
 	}
 	return result_stats;
 }

package/src/duckdb/src/optimizer/statistics/expression/propagate_columnref.cpp

@@ -9,7 +9,7 @@ unique_ptr<BaseStatistics> StatisticsPropagator::PropagateExpression(BoundColumn
 	if (stats == statistics_map.end()) {
 		return nullptr;
 	}
-	return stats->second->Copy();
+	return stats->second->ToUnique();
 }
 
 } // namespace duckdb

package/src/duckdb/src/optimizer/statistics/expression/propagate_comparison.cpp

@@ -1,15 +1,14 @@
 #include "duckdb/optimizer/statistics_propagator.hpp"
 #include "duckdb/planner/expression/bound_comparison_expression.hpp"
 #include "duckdb/planner/expression/bound_constant_expression.hpp"
-#include "duckdb/storage/statistics/numeric_statistics.hpp"
 #include "duckdb/optimizer/expression_rewriter.hpp"
 
 namespace duckdb {
 
-FilterPropagateResult StatisticsPropagator::PropagateComparison(BaseStatistics &left, BaseStatistics &right,
+FilterPropagateResult StatisticsPropagator::PropagateComparison(BaseStatistics &lstats, BaseStatistics &rstats,
                                                                 ExpressionType comparison) {
 	// only handle numerics for now
-	switch (left.type.InternalType()) {
+	switch (lstats.GetType().InternalType()) {
 	case PhysicalType::BOOL:
 	case PhysicalType::UINT8:
 	case PhysicalType::UINT16:
@@ -26,9 +25,7 @@ FilterPropagateResult StatisticsPropagator::PropagateComparison(BaseStatistics &
 	default:
 		return FilterPropagateResult::NO_PRUNING_POSSIBLE;
 	}
-	auto &lstats = (NumericStatistics &)left;
-	auto &rstats = (NumericStatistics &)right;
-	if (lstats.min.IsNull() || lstats.max.IsNull() || rstats.min.IsNull() || rstats.max.IsNull()) {
+	if (!NumericStats::HasMinMax(lstats) || !NumericStats::HasMinMax(rstats)) {
 		// no stats available: nothing to prune
 		return FilterPropagateResult::NO_PRUNING_POSSIBLE;
 	}
@@ -38,52 +35,53 @@ FilterPropagateResult StatisticsPropagator::PropagateComparison(BaseStatistics &
 	switch (comparison) {
 	case ExpressionType::COMPARE_EQUAL:
 		// l = r, if l.min > r.max or r.min > l.max equality is not possible
-		if (lstats.min > rstats.max || rstats.min > lstats.max) {
+		if (NumericStats::Min(lstats) > NumericStats::Max(rstats) ||
+		    NumericStats::Min(rstats) > NumericStats::Max(lstats)) {
 			return has_null ? FilterPropagateResult::FILTER_FALSE_OR_NULL : FilterPropagateResult::FILTER_ALWAYS_FALSE;
 		} else {
 			return FilterPropagateResult::NO_PRUNING_POSSIBLE;
 		}
 	case ExpressionType::COMPARE_GREATERTHAN:
 		// l > r
-		if (lstats.min > rstats.max) {
+		if (NumericStats::Min(lstats) > NumericStats::Max(rstats)) {
 			// if l.min > r.max, it is always true ONLY if neither side contains nulls
 			return has_null ? FilterPropagateResult::FILTER_TRUE_OR_NULL : FilterPropagateResult::FILTER_ALWAYS_TRUE;
 		}
 		// if r.min is bigger or equal to l.max, the filter is always false
-		if (rstats.min >= lstats.max) {
+		if (NumericStats::Min(rstats) >= NumericStats::Max(lstats)) {
 			return has_null ? FilterPropagateResult::FILTER_FALSE_OR_NULL : FilterPropagateResult::FILTER_ALWAYS_FALSE;
 		}
 		return FilterPropagateResult::NO_PRUNING_POSSIBLE;
 	case ExpressionType::COMPARE_GREATERTHANOREQUALTO:
 		// l >= r
-		if (lstats.min >= rstats.max) {
+		if (NumericStats::Min(lstats) >= NumericStats::Max(rstats)) {
 			// if l.min >= r.max, it is always true ONLY if neither side contains nulls
 			return has_null ? FilterPropagateResult::FILTER_TRUE_OR_NULL : FilterPropagateResult::FILTER_ALWAYS_TRUE;
 		}
 		// if r.min > l.max, the filter is always false
-		if (rstats.min > lstats.max) {
+		if (NumericStats::Min(rstats) > NumericStats::Max(lstats)) {
 			return has_null ? FilterPropagateResult::FILTER_FALSE_OR_NULL : FilterPropagateResult::FILTER_ALWAYS_FALSE;
 		}
 		return FilterPropagateResult::NO_PRUNING_POSSIBLE;
 	case ExpressionType::COMPARE_LESSTHAN:
 		// l < r
-		if (lstats.max < rstats.min) {
+		if (NumericStats::Max(lstats) < NumericStats::Min(rstats)) {
 			// if l.max < r.min, it is always true ONLY if neither side contains nulls
 			return has_null ? FilterPropagateResult::FILTER_TRUE_OR_NULL : FilterPropagateResult::FILTER_ALWAYS_TRUE;
 		}
 		// if l.min >= rstats.max, the filter is always false
-		if (lstats.min >= rstats.max) {
+		if (NumericStats::Min(lstats) >= NumericStats::Max(rstats)) {
 			return has_null ? FilterPropagateResult::FILTER_FALSE_OR_NULL : FilterPropagateResult::FILTER_ALWAYS_FALSE;
 		}
 		return FilterPropagateResult::NO_PRUNING_POSSIBLE;
 	case ExpressionType::COMPARE_LESSTHANOREQUALTO:
 		// l <= r
-		if (lstats.max <= rstats.min) {
+		if (NumericStats::Max(lstats) <= NumericStats::Min(rstats)) {
 			// if l.max <= r.min, it is always true ONLY if neither side contains nulls
 			return has_null ? FilterPropagateResult::FILTER_TRUE_OR_NULL : FilterPropagateResult::FILTER_ALWAYS_TRUE;
 		}
 		// if l.min > rstats.max, the filter is always false
-		if (lstats.min > rstats.max) {
+		if (NumericStats::Min(lstats) > NumericStats::Max(rstats)) {
 			return has_null ? FilterPropagateResult::FILTER_FALSE_OR_NULL : FilterPropagateResult::FILTER_ALWAYS_FALSE;
 		}
 		return FilterPropagateResult::NO_PRUNING_POSSIBLE;

package/src/duckdb/src/optimizer/statistics/expression/propagate_conjunction.cpp

@@ -2,7 +2,6 @@
 #include "duckdb/optimizer/statistics_propagator.hpp"
 #include "duckdb/planner/expression/bound_conjunction_expression.hpp"
 #include "duckdb/planner/expression/bound_constant_expression.hpp"
-#include "duckdb/storage/statistics/numeric_statistics.hpp"
 #include "duckdb/optimizer/expression_rewriter.hpp"
 #include "duckdb/execution/expression_executor.hpp"
 

package/src/duckdb/src/optimizer/statistics/expression/propagate_constant.cpp

@@ -1,85 +1,13 @@
 #include "duckdb/optimizer/statistics_propagator.hpp"
 #include "duckdb/planner/expression/bound_constant_expression.hpp"
 #include "duckdb/storage/statistics/distinct_statistics.hpp"
-#include "duckdb/storage/statistics/list_statistics.hpp"
-#include "duckdb/storage/statistics/numeric_statistics.hpp"
-#include "duckdb/storage/statistics/string_statistics.hpp"
-#include "duckdb/storage/statistics/struct_statistics.hpp"
+#include "duckdb/storage/statistics/list_stats.hpp"
+#include "duckdb/storage/statistics/struct_stats.hpp"
 
 namespace duckdb {
 
-void UpdateDistinctStats(BaseStatistics &distinct_stats, const Value &input) {
-	Vector v(input);
-	auto &d_stats = (DistinctStatistics &)distinct_stats;
-	d_stats.Update(v, 1);
-}
-
 unique_ptr<BaseStatistics> StatisticsPropagator::StatisticsFromValue(const Value &input) {
-	switch (input.type().InternalType()) {
-	case PhysicalType::BOOL:
-	case PhysicalType::UINT8:
-	case PhysicalType::UINT16:
-	case PhysicalType::UINT32:
-	case PhysicalType::UINT64:
-	case PhysicalType::INT8:
-	case PhysicalType::INT16:
-	case PhysicalType::INT32:
-	case PhysicalType::INT64:
-	case PhysicalType::INT128:
-	case PhysicalType::FLOAT:
-	case PhysicalType::DOUBLE: {
-		auto result = make_unique<NumericStatistics>(input.type(), input, input, StatisticsType::GLOBAL_STATS);
-		result->validity_stats = make_unique<ValidityStatistics>(input.IsNull(), !input.IsNull());
-		UpdateDistinctStats(*result->distinct_stats, input);
-		return std::move(result);
-	}
-	case PhysicalType::VARCHAR: {
-		auto result = make_unique<StringStatistics>(input.type(), StatisticsType::GLOBAL_STATS);
-		result->validity_stats = make_unique<ValidityStatistics>(input.IsNull(), !input.IsNull());
-		UpdateDistinctStats(*result->distinct_stats, input);
-		if (!input.IsNull()) {
-			auto &string_value = StringValue::Get(input);
-			result->Update(string_t(string_value));
-		}
-		return std::move(result);
-	}
-	case PhysicalType::STRUCT: {
-		auto result = make_unique<StructStatistics>(input.type());
-		result->validity_stats = make_unique<ValidityStatistics>(input.IsNull(), !input.IsNull());
-		if (input.IsNull()) {
-			for (auto &child_stat : result->child_stats) {
-				child_stat.reset();
-			}
-		} else {
-			auto &struct_children = StructValue::GetChildren(input);
-			D_ASSERT(result->child_stats.size() == struct_children.size());
-			for (idx_t i = 0; i < result->child_stats.size(); i++) {
-				result->child_stats[i] = StatisticsFromValue(struct_children[i]);
-			}
-		}
-		return std::move(result);
-	}
-	case PhysicalType::LIST: {
-		auto result = make_unique<ListStatistics>(input.type());
-		result->validity_stats = make_unique<ValidityStatistics>(input.IsNull(), !input.IsNull());
-		if (input.IsNull()) {
-			result->child_stats.reset();
-		} else {
-			auto &list_children = ListValue::GetChildren(input);
-			for (auto &child_element : list_children) {
-				auto child_element_stats = StatisticsFromValue(child_element);
-				if (child_element_stats) {
-					result->child_stats->Merge(*child_element_stats);
-				} else {
-					result->child_stats.reset();
-				}
-			}
-		}
-		return std::move(result);
-	}
-	default:
-		return nullptr;
-	}
+	return BaseStatistics::FromConstant(input).ToUnique();
 }
 
 unique_ptr<BaseStatistics> StatisticsPropagator::PropagateExpression(BoundConstantExpression &constant,

package/src/duckdb/src/optimizer/statistics/expression/propagate_function.cpp

@@ -5,10 +5,15 @@ namespace duckdb {
 
 unique_ptr<BaseStatistics> StatisticsPropagator::PropagateExpression(BoundFunctionExpression &func,
                                                                      unique_ptr<Expression> *expr_ptr) {
-	vector<unique_ptr<BaseStatistics>> stats;
+	vector<BaseStatistics> stats;
 	stats.reserve(func.children.size());
 	for (idx_t i = 0; i < func.children.size(); i++) {
-		stats.push_back(PropagateExpression(func.children[i]));
+		auto stat = PropagateExpression(func.children[i]);
+		if (!stat) {
+			stats.push_back(BaseStatistics::CreateUnknown(func.children[i]->return_type));
+		} else {
+			stats.push_back(stat->Copy());
+		}
 	}
 	if (!func.function.statistics) {
 		return nullptr;

package/src/duckdb/src/optimizer/statistics/expression/propagate_operator.cpp

@@ -62,6 +62,11 @@ unique_ptr<BaseStatistics> StatisticsPropagator::PropagateExpression(BoundOperat
 			*expr_ptr = make_unique<BoundConstantExpression>(Value::BOOLEAN(false));
 			return PropagateExpression(*expr_ptr);
 		}
+		if (!child_stats[0]->CanHaveNoNull()) {
+			// child has no valid values: x IS NULL will always be true
+			*expr_ptr = make_unique<BoundConstantExpression>(Value::BOOLEAN(true));
+			return PropagateExpression(*expr_ptr);
+		}
 		return nullptr;
 	case ExpressionType::OPERATOR_IS_NOT_NULL:
 		if (!child_stats[0]->CanHaveNull()) {
@@ -69,6 +74,11 @@ unique_ptr<BaseStatistics> StatisticsPropagator::PropagateExpression(BoundOperat
 			*expr_ptr = make_unique<BoundConstantExpression>(Value::BOOLEAN(true));
 			return PropagateExpression(*expr_ptr);
 		}
+		if (!child_stats[0]->CanHaveNoNull()) {
+			// child has no valid values: x IS NOT NULL will always be false
+			*expr_ptr = make_unique<BoundConstantExpression>(Value::BOOLEAN(false));
+			return PropagateExpression(*expr_ptr);
+		}
 		return nullptr;
 	default:
 		return nullptr;

package/src/duckdb/src/optimizer/statistics/operator/propagate_aggregate.cpp

@@ -1,6 +1,5 @@
 #include "duckdb/optimizer/statistics_propagator.hpp"
 #include "duckdb/planner/operator/logical_aggregate.hpp"
-#include "duckdb/storage/statistics/validity_statistics.hpp"
 
 namespace duckdb {
 
@@ -13,14 +12,14 @@ unique_ptr<NodeStatistics> StatisticsPropagator::PropagateStatistics(LogicalAggr
 	aggr.group_stats.resize(aggr.groups.size());
 	for (idx_t group_idx = 0; group_idx < aggr.groups.size(); group_idx++) {
 		auto stats = PropagateExpression(aggr.groups[group_idx]);
-		aggr.group_stats[group_idx] = stats ? stats->Copy() : nullptr;
+		aggr.group_stats[group_idx] = stats ? stats->ToUnique() : nullptr;
 		if (!stats) {
 			continue;
 		}
 		if (aggr.grouping_sets.size() > 1) {
 			// aggregates with multiple grouping sets can introduce NULL values to certain groups
 			// FIXME: actually figure out WHICH groups can have null values introduced
-			stats->validity_stats = make_unique<ValidityStatistics>(true, true);
+			stats->Set(StatsInfo::CAN_HAVE_NULL_VALUES);
 			continue;
 		}
 		ColumnBinding group_binding(aggr.group_index, group_idx);

package/src/duckdb/src/optimizer/statistics/operator/propagate_filter.cpp

@@ -5,7 +5,7 @@
 #include "duckdb/planner/expression/bound_comparison_expression.hpp"
 #include "duckdb/planner/expression/bound_constant_expression.hpp"
 #include "duckdb/planner/operator/logical_filter.hpp"
-#include "duckdb/storage/statistics/numeric_statistics.hpp"
+#include "duckdb/storage/statistics/base_statistics.hpp"
 
 namespace duckdb {
 
@@ -35,21 +35,20 @@ void StatisticsPropagator::SetStatisticsNotNull(ColumnBinding binding) {
 	if (entry == statistics_map.end()) {
 		return;
 	}
-	entry->second->validity_stats = make_unique<ValidityStatistics>(false);
+	entry->second->Set(StatsInfo::CANNOT_HAVE_NULL_VALUES);
 }
 
 void StatisticsPropagator::UpdateFilterStatistics(BaseStatistics &stats, ExpressionType comparison_type,
                                                   const Value &constant) {
 	// regular comparisons removes all null values
 	if (!IsCompareDistinct(comparison_type)) {
-		stats.validity_stats = make_unique<ValidityStatistics>(false);
+		stats.Set(StatsInfo::CANNOT_HAVE_NULL_VALUES);
 	}
-	if (!stats.type.IsNumeric()) {
+	if (!stats.GetType().IsNumeric()) {
 		// don't handle non-numeric columns here (yet)
 		return;
 	}
-	auto &numeric_stats = (NumericStatistics &)stats;
-	if (numeric_stats.min.IsNull() || numeric_stats.max.IsNull()) {
+	if (!NumericStats::HasMinMax(stats)) {
 		// no stats available: skip this
 		return;
 	}
@@ -58,19 +57,19 @@ void StatisticsPropagator::UpdateFilterStatistics(BaseStatistics &stats, Express
 	case ExpressionType::COMPARE_LESSTHANOREQUALTO:
 		// X < constant OR X <= constant
 		// max becomes the constant
-		numeric_stats.max = constant;
+		NumericStats::SetMax(stats, constant);
 		break;
 	case ExpressionType::COMPARE_GREATERTHAN:
 	case ExpressionType::COMPARE_GREATERTHANOREQUALTO:
 		// X > constant OR X >= constant
 		// min becomes the constant
-		numeric_stats.min = constant;
+		NumericStats::SetMin(stats, constant);
 		break;
 	case ExpressionType::COMPARE_EQUAL:
 		// X = constant
 		// both min and max become the constant
-		numeric_stats.min = constant;
-		numeric_stats.max = constant;
+		NumericStats::SetMin(stats, constant);
+		NumericStats::SetMax(stats, constant);
 		break;
 	default:
 		break;
@@ -81,17 +80,15 @@ void StatisticsPropagator::UpdateFilterStatistics(BaseStatistics &lstats, BaseSt
                                                   ExpressionType comparison_type) {
 	// regular comparisons removes all null values
 	if (!IsCompareDistinct(comparison_type)) {
-		lstats.validity_stats = make_unique<ValidityStatistics>(false);
-		rstats.validity_stats = make_unique<ValidityStatistics>(false);
+		lstats.Set(StatsInfo::CANNOT_HAVE_NULL_VALUES);
+		rstats.Set(StatsInfo::CANNOT_HAVE_NULL_VALUES);
 	}
-	D_ASSERT(lstats.type == rstats.type);
-	if (!lstats.type.IsNumeric()) {
+	D_ASSERT(lstats.GetType() == rstats.GetType());
+	if (!lstats.GetType().IsNumeric()) {
 		// don't handle non-numeric columns here (yet)
 		return;
 	}
-	auto &left_stats = (NumericStatistics &)lstats;
-	auto &right_stats = (NumericStatistics &)rstats;
-	if (left_stats.min.IsNull() || left_stats.max.IsNull() || right_stats.min.IsNull() || right_stats.max.IsNull()) {
+	if (!NumericStats::HasMinMax(lstats) || !NumericStats::HasMinMax(rstats)) {
 		// no stats available: skip this
 		return;
 	}
@@ -104,14 +101,14 @@ void StatisticsPropagator::UpdateFilterStatistics(BaseStatistics &lstats, BaseSt
 
 		// we know that left.max is AT MOST equal to right.max
 		// because any value in left that is BIGGER than right.max will not pass the filter
-		if (left_stats.max > right_stats.max) {
-			left_stats.max = right_stats.max;
+		if (NumericStats::Max(lstats) > NumericStats::Max(rstats)) {
+			NumericStats::SetMax(lstats, NumericStats::Max(rstats));
 		}
 
 		// we also know that right.min is AT MOST equal to left.min
 		// because any value in right that is SMALLER than left.min will not pass the filter
-		if (right_stats.min < left_stats.min) {
-			right_stats.min = left_stats.min;
+		if (NumericStats::Min(rstats) < NumericStats::Min(lstats)) {
+			NumericStats::SetMin(rstats, NumericStats::Min(lstats));
 		}
 		// so in our example, the bounds get updated as follows:
 		// left: [-50, 100], right: [-50, 100]
@@ -121,11 +118,11 @@ void StatisticsPropagator::UpdateFilterStatistics(BaseStatistics &lstats, BaseSt
 		// LEFT > RIGHT OR LEFT >= RIGHT
 		// we know that every value of left is bigger (or equal to) every value in right
 		// this is essentially the inverse of the less than (or equal to) scenario
-		if (right_stats.max > left_stats.max) {
-			right_stats.max = left_stats.max;
+		if (NumericStats::Max(rstats) > NumericStats::Max(lstats)) {
+			NumericStats::SetMax(rstats, NumericStats::Max(lstats));
 		}
-		if (left_stats.min < right_stats.min) {
-			left_stats.min = right_stats.min;
+		if (NumericStats::Min(lstats) < NumericStats::Min(rstats)) {
+			NumericStats::SetMin(lstats, NumericStats::Min(rstats));
 		}
 		break;
 	case ExpressionType::COMPARE_EQUAL:
@@ -135,16 +132,16 @@ void StatisticsPropagator::UpdateFilterStatistics(BaseStatistics &lstats, BaseSt
 		// so if we have e.g. left = [-50, 250] and right = [-100, 100]
 		// the tighest bounds are [-50, 100]
 		// select the highest min
-		if (left_stats.min > right_stats.min) {
-			right_stats.min = left_stats.min;
+		if (NumericStats::Min(lstats) > NumericStats::Min(rstats)) {
+			NumericStats::SetMin(rstats, NumericStats::Min(lstats));
 		} else {
-			left_stats.min = right_stats.min;
+			NumericStats::SetMin(lstats, NumericStats::Min(rstats));
 		}
 		// select the lowest max
-		if (left_stats.max < right_stats.max) {
-			right_stats.max = left_stats.max;
+		if (NumericStats::Max(lstats) < NumericStats::Max(rstats)) {
+			NumericStats::SetMax(rstats, NumericStats::Max(lstats));
 		} else {
-			left_stats.max = right_stats.max;
+			NumericStats::SetMax(lstats, NumericStats::Max(rstats));
 		}
 		break;
 	default:
@@ -168,7 +165,7 @@ void StatisticsPropagator::UpdateFilterStatistics(Expression &left, Expression &
 	if (left.type == ExpressionType::VALUE_CONSTANT && right.type == ExpressionType::BOUND_COLUMN_REF) {
 		constant = (BoundConstantExpression *)&left;
 		columnref = (BoundColumnRefExpression *)&right;
-		comparison_type = FlipComparisionExpression(comparison_type);
+		comparison_type = FlipComparisonExpression(comparison_type);
 	} else if (left.type == ExpressionType::BOUND_COLUMN_REF && right.type == ExpressionType::VALUE_CONSTANT) {
 		columnref = (BoundColumnRefExpression *)&left;
 		constant = (BoundConstantExpression *)&right;

package/src/duckdb/src/optimizer/statistics/operator/propagate_join.cpp

@@ -7,7 +7,6 @@
 #include "duckdb/planner/operator/logical_join.hpp"
 #include "duckdb/planner/operator/logical_limit.hpp"
 #include "duckdb/planner/operator/logical_positional_join.hpp"
-#include "duckdb/storage/statistics/validity_statistics.hpp"
 
 namespace duckdb {
 
@@ -196,6 +195,7 @@ unique_ptr<NodeStatistics> StatisticsPropagator::PropagateStatistics(LogicalJoin
 	switch (join.type) {
 	case LogicalOperatorType::LOGICAL_COMPARISON_JOIN:
 	case LogicalOperatorType::LOGICAL_DELIM_JOIN:
+	case LogicalOperatorType::LOGICAL_ASOF_JOIN:
 		PropagateStatistics((LogicalComparisonJoin &)join, node_ptr);
 		break;
 	case LogicalOperatorType::LOGICAL_ANY_JOIN:
@@ -210,7 +210,7 @@ unique_ptr<NodeStatistics> StatisticsPropagator::PropagateStatistics(LogicalJoin
 		for (auto &binding : right_bindings) {
 			auto stats = statistics_map.find(binding);
 			if (stats != statistics_map.end()) {
-				stats->second->validity_stats = make_unique<ValidityStatistics>(true);
+				stats->second->Set(StatsInfo::CAN_HAVE_NULL_VALUES);
 			}
 		}
 	}
@@ -219,7 +219,7 @@ unique_ptr<NodeStatistics> StatisticsPropagator::PropagateStatistics(LogicalJoin
 		for (auto &binding : left_bindings) {
 			auto stats = statistics_map.find(binding);
 			if (stats != statistics_map.end()) {
-				stats->second->validity_stats = make_unique<ValidityStatistics>(true);
+				stats->second->Set(StatsInfo::CAN_HAVE_NULL_VALUES);
 			}
 		}
 	}
@@ -265,7 +265,7 @@ unique_ptr<NodeStatistics> StatisticsPropagator::PropagateStatistics(LogicalPosi
 	for (auto &binding : left_bindings) {
 		auto stats = statistics_map.find(binding);
 		if (stats != statistics_map.end()) {
-			stats->second->validity_stats = make_unique<ValidityStatistics>(true);
+			stats->second->Set(StatsInfo::CAN_HAVE_NULL_VALUES);
 		}
 	}
 
@@ -274,7 +274,7 @@ unique_ptr<NodeStatistics> StatisticsPropagator::PropagateStatistics(LogicalPosi
 	for (auto &binding : right_bindings) {
 		auto stats = statistics_map.find(binding);
 		if (stats != statistics_map.end()) {
-			stats->second->validity_stats = make_unique<ValidityStatistics>(true);
+			stats->second->Set(StatsInfo::CAN_HAVE_NULL_VALUES);
 		}
 	}