duckdb 0.7.2-dev225.0 → 0.7.2-dev314.0

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:

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 {
 
@@ -210,7 +209,7 @@ 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 +218,7 @@ 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 +264,7 @@ 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 +273,7 @@ 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);
 		}
 	}
 

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

@@ -47,18 +47,18 @@ unique_ptr<NodeStatistics> StatisticsPropagator::PropagateStatistics(LogicalSetO
 		switch (setop.type) {
 		case LogicalOperatorType::LOGICAL_UNION:
 			// union: merge the stats of the LHS and RHS together
-			new_stats = left_entry->second->Copy();
+			new_stats = left_entry->second->ToUnique();
 			new_stats->Merge(*right_entry->second);
 			break;
 		case LogicalOperatorType::LOGICAL_EXCEPT:
 			// except: use the stats of the LHS
-			new_stats = left_entry->second->Copy();
+			new_stats = left_entry->second->ToUnique();
 			break;
 		case LogicalOperatorType::LOGICAL_INTERSECT:
 			// intersect: intersect the two stats
 			// FIXME: for now we just use the stats of the LHS, as this is correct
 			// however, the stats can be further refined to the minimal subset of the LHS and RHS
-			new_stats = left_entry->second->Copy();
+			new_stats = left_entry->second->ToUnique();
 			break;
 		default:
 			throw InternalException("Unsupported setop type");

package/src/duckdb/src/optimizer/statistics_propagator.cpp

@@ -92,7 +92,7 @@ unique_ptr<BaseStatistics> StatisticsPropagator::PropagateExpression(Expression
 unique_ptr<BaseStatistics> StatisticsPropagator::PropagateExpression(unique_ptr<Expression> &expr) {
 	auto stats = PropagateExpression(*expr, &expr);
 	if (ClientConfig::GetConfig(context).query_verification_enabled && stats) {
-		expr->verification_stats = stats->Copy();
+		expr->verification_stats = stats->ToUnique();
 	}
 	return stats;
 }

package/src/duckdb/src/parser/transform/tableref/transform_join.cpp

@@ -28,6 +28,10 @@ unique_ptr<TableRef> Transformer::TransformJoin(duckdb_libpgquery::PGJoinExpr *r
 		result->type = JoinType::SEMI;
 		break;
 	}
+	case duckdb_libpgquery::PG_JOIN_ANTI: {
+		result->type = JoinType::ANTI;
+		break;
+	}
 	case duckdb_libpgquery::PG_JOIN_POSITION: {
 		result->ref_type = JoinRefType::POSITIONAL;
 		break;

package/src/duckdb/src/planner/bind_context.cpp

@@ -555,4 +555,20 @@ void BindContext::AddContext(BindContext other) {
 	}
 }
 
+void BindContext::RemoveContext(vector<std::pair<string, duckdb::Binding *>> &other_bindings_list) {
+	for (auto &other_binding : other_bindings_list) {
+		if (bindings.find(other_binding.first) != bindings.end()) {
+			bindings.erase(other_binding.first);
+		}
+	}
+
+	vector<idx_t> delete_list_indexes;
+	for (auto &other_binding : other_bindings_list) {
+		auto it =
+		    std::remove_if(bindings_list.begin(), bindings_list.end(),
+		                   [other_binding](std::pair<string, Binding *> &x) { return x.first == other_binding.first; });
+		bindings_list.erase(it, bindings_list.end());
+	}
+}
+
 } // namespace duckdb

package/src/duckdb/src/planner/binder/query_node/plan_select_node.cpp

@@ -104,7 +104,6 @@ unique_ptr<LogicalOperator> Binder::CreatePlan(BoundSelectNode &statement) {
 		PlanSubqueries(&expr, &root);
 	}
 
-	// create the projection
 	auto proj = make_unique<LogicalProjection>(statement.projection_index, std::move(statement.select_list));
 	auto &projection = *proj;
 	proj->AddChild(std::move(root));

package/src/duckdb/src/planner/binder/tableref/bind_joinref.cpp

@@ -7,9 +7,11 @@
 #include "duckdb/parser/expression/constant_expression.hpp"
 #include "duckdb/parser/expression/conjunction_expression.hpp"
 #include "duckdb/parser/expression/bound_expression.hpp"
+#include "duckdb/parser/expression/star_expression.hpp"
 #include "duckdb/common/string_util.hpp"
 #include "duckdb/common/case_insensitive_map.hpp"
 #include "duckdb/planner/expression_binder/lateral_binder.hpp"
+#include "duckdb/planner/query_node/bound_select_node.hpp"
 
 namespace duckdb {
 
@@ -253,6 +255,8 @@ unique_ptr<BoundTableRef> Binder::Bind(JoinRef &ref) {
 		}
 	}
 
+	auto right_bindings_list_copy = right_binder.bind_context.GetBindingsList();
+
 	bind_context.AddContext(std::move(left_binder.bind_context));
 	bind_context.AddContext(std::move(right_binder.bind_context));
 	MoveCorrelatedExpressions(left_binder);
@@ -269,6 +273,11 @@ unique_ptr<BoundTableRef> Binder::Bind(JoinRef &ref) {
 		WhereBinder binder(*this, context);
 		result->condition = binder.Bind(ref.condition);
 	}
+
+	if (result->type == JoinType::SEMI || result->type == JoinType::ANTI) {
+		bind_context.RemoveContext(right_bindings_list_copy);
+	}
+
 	return std::move(result);
 }
 

package/src/duckdb/src/planner/binder.cpp

@@ -3,8 +3,10 @@
 #include "duckdb/catalog/catalog_entry/table_catalog_entry.hpp"
 #include "duckdb/catalog/catalog_entry/view_catalog_entry.hpp"
 #include "duckdb/main/config.hpp"
+#include "duckdb/parser/parsed_expression_iterator.hpp"
 #include "duckdb/parser/query_node/select_node.hpp"
 #include "duckdb/parser/statement/list.hpp"
+#include "duckdb/parser/tableref/joinref.hpp"
 #include "duckdb/parser/tableref/table_function_ref.hpp"
 #include "duckdb/planner/bound_query_node.hpp"
 #include "duckdb/planner/bound_tableref.hpp"
@@ -13,7 +15,6 @@
 #include "duckdb/planner/expression_iterator.hpp"
 #include "duckdb/planner/operator/logical_projection.hpp"
 #include "duckdb/planner/operator/logical_sample.hpp"
-#include "duckdb/parser/parsed_expression_iterator.hpp"
 
 #include <algorithm>
 

package/src/duckdb/src/planner/bound_result_modifier.cpp

@@ -19,7 +19,7 @@ BoundOrderByNode::BoundOrderByNode(OrderType type, OrderByNullType null_order, u
 
 BoundOrderByNode BoundOrderByNode::Copy() const {
 	if (stats) {
-		return BoundOrderByNode(type, null_order, expression->Copy(), stats->Copy());
+		return BoundOrderByNode(type, null_order, expression->Copy(), stats->ToUnique());
 	} else {
 		return BoundOrderByNode(type, null_order, expression->Copy());
 	}

package/src/duckdb/src/planner/expression/bound_window_expression.cpp

@@ -102,7 +102,7 @@ unique_ptr<Expression> BoundWindowExpression::Copy() {
 	}
 	for (auto &ps : partitions_stats) {
 		if (ps) {
-			new_window->partitions_stats.push_back(ps->Copy());
+			new_window->partitions_stats.push_back(ps->ToUnique());
 		} else {
 			new_window->partitions_stats.push_back(nullptr);
 		}

package/src/duckdb/src/planner/filter/constant_filter.cpp

@@ -1,8 +1,6 @@
 #include "duckdb/planner/filter/constant_filter.hpp"
-
+#include "duckdb/storage/statistics/base_statistics.hpp"
 #include "duckdb/common/field_writer.hpp"
-#include "duckdb/storage/statistics/numeric_statistics.hpp"
-#include "duckdb/storage/statistics/string_statistics.hpp"
 
 namespace duckdb {
 
@@ -12,7 +10,7 @@ ConstantFilter::ConstantFilter(ExpressionType comparison_type_p, Value constant_
 }
 
 FilterPropagateResult ConstantFilter::CheckStatistics(BaseStatistics &stats) {
-	D_ASSERT(constant.type().id() == stats.type.id());
+	D_ASSERT(constant.type().id() == stats.GetType().id());
 	switch (constant.type().InternalType()) {
 	case PhysicalType::UINT8:
 	case PhysicalType::UINT16:
@@ -25,9 +23,9 @@ FilterPropagateResult ConstantFilter::CheckStatistics(BaseStatistics &stats) {
 	case PhysicalType::INT128:
 	case PhysicalType::FLOAT:
 	case PhysicalType::DOUBLE:
-		return ((NumericStatistics &)stats).CheckZonemap(comparison_type, constant);
+		return NumericStats::CheckZonemap(stats, comparison_type, constant);
 	case PhysicalType::VARCHAR:
-		return ((StringStatistics &)stats).CheckZonemap(comparison_type, StringValue::Get(constant));
+		return StringStats::CheckZonemap(stats, comparison_type, StringValue::Get(constant));
 	default:
 		return FilterPropagateResult::NO_PRUNING_POSSIBLE;
 	}

package/src/duckdb/src/storage/checkpoint/row_group_writer.cpp

@@ -32,7 +32,7 @@ void SingleFileRowGroupWriter::WriteColumnDataPointers(ColumnCheckpointState &co
 		meta_writer.Write<block_id_t>(data_pointer.block_pointer.block_id);
 		meta_writer.Write<uint32_t>(data_pointer.block_pointer.offset);
 		meta_writer.Write<CompressionType>(data_pointer.compression_type);
-		data_pointer.statistics->Serialize(meta_writer);
+		data_pointer.statistics.Serialize(meta_writer);
 	}
 }
 

package/src/duckdb/src/storage/checkpoint/table_data_reader.cpp

@@ -24,10 +24,7 @@ void TableDataReader::ReadTableData() {
 	D_ASSERT(!columns.empty());
 
 	// deserialize the total table statistics
-	info.data->column_stats.reserve(columns.PhysicalColumnCount());
-	for (auto &col : columns.Physical()) {
-		info.data->column_stats.push_back(BaseStatistics::Deserialize(reader, col.Type()));
-	}
+	info.data->table_stats.Deserialize(reader, columns);
 
 	// deserialize each of the individual row groups
 	auto row_group_count = reader.Read<uint64_t>();