duckdb 0.7.2-dev654.0 → 0.7.2-dev832.0

package/src/duckdb/src/execution/aggregate_hashtable.cpp

@@ -155,25 +155,30 @@ void GroupedAggregateHashTable::VerifyInternal() {
 	D_ASSERT(count == entries);
 }
 
-idx_t GroupedAggregateHashTable::MaxCapacity() {
-	idx_t max_pages = 0;
-	idx_t max_tuples = 0;
+idx_t GroupedAggregateHashTable::GetMaxCapacity(HtEntryType entry_type, idx_t tuple_size) {
+	idx_t max_pages;
+	idx_t max_tuples;
 
 	switch (entry_type) {
 	case HtEntryType::HT_WIDTH_32:
 		max_pages = NumericLimits<uint8_t>::Maximum();
 		max_tuples = NumericLimits<uint16_t>::Maximum();
 		break;
-	default:
-		D_ASSERT(entry_type == HtEntryType::HT_WIDTH_64);
+	case HtEntryType::HT_WIDTH_64:
 		max_pages = NumericLimits<uint32_t>::Maximum();
 		max_tuples = NumericLimits<uint16_t>::Maximum();
 		break;
+	default:
+		throw InternalException("Unsupported hash table width");
 	}
 
 	return max_pages * MinValue(max_tuples, (idx_t)Storage::BLOCK_SIZE / tuple_size);
 }
 
+idx_t GroupedAggregateHashTable::MaxCapacity() {
+	return GetMaxCapacity(entry_type, tuple_size);
+}
+
 void GroupedAggregateHashTable::Verify() {
 #ifdef DEBUG
 	switch (entry_type) {

package/src/duckdb/src/execution/expression_executor/execute_cast.cpp

@@ -11,7 +11,8 @@ unique_ptr<ExpressionState> ExpressionExecutor::InitializeState(const BoundCastE
 	result->AddChild(expr.child.get());
 	result->Finalize();
 	if (expr.bound_cast.init_local_state) {
-		result->local_state = expr.bound_cast.init_local_state(root.executor->GetContext());
+		CastLocalStateParameters parameters(root.executor->GetContext(), expr.bound_cast.cast_data);
+		result->local_state = expr.bound_cast.init_local_state(parameters);
 	}
 	return std::move(result);
 }

package/src/duckdb/src/execution/index/art/art.cpp

@@ -330,8 +330,7 @@ bool ART::ConstructFromSorted(idx_t count, vector<Key> &keys, Vector &row_identi
 //===--------------------------------------------------------------------===//
 // Insert / Verification / Constraint Checking
 //===--------------------------------------------------------------------===//
-
-bool ART::Insert(IndexLock &lock, DataChunk &input, Vector &row_ids) {
+PreservedError ART::Insert(IndexLock &lock, DataChunk &input, Vector &row_ids) {
 
 	D_ASSERT(row_ids.GetType().InternalType() == ROW_TYPE);
 	D_ASSERT(logical_types[0] == input.data[0].GetType());
@@ -375,12 +374,13 @@ bool ART::Insert(IndexLock &lock, DataChunk &input, Vector &row_ids) {
 
 	IncreaseAndVerifyMemorySize(old_memory_size);
 	if (failed_index != DConstants::INVALID_INDEX) {
-		return false;
+		return PreservedError(ConstraintException("PRIMARY KEY or UNIQUE constraint violated: duplicate key \"%s\"",
+		                                          AppendRowError(input, failed_index)));
 	}
-	return true;
+	return PreservedError();
 }
 
-bool ART::Append(IndexLock &lock, DataChunk &appended_data, Vector &row_identifiers) {
+PreservedError ART::Append(IndexLock &lock, DataChunk &appended_data, Vector &row_identifiers) {
 	DataChunk expression_result;
 	expression_result.Initialize(Allocator::DefaultAllocator(), logical_types);
 

package/src/duckdb/src/execution/operator/persistent/base_csv_reader.cpp

@@ -121,6 +121,9 @@ bool TryCastFloatingValueCommaSeparated(const string_t &value_str, const Logical
 }
 
 bool BaseCSVReader::TryCastValue(const Value &value, const LogicalType &sql_type) {
+	if (value.IsNull()) {
+		return true;
+	}
 	if (options.has_format[LogicalTypeId::DATE] && sql_type.id() == LogicalTypeId::DATE) {
 		date_t result;
 		string error_message;

package/src/duckdb/src/execution/partitionable_hashtable.cpp

@@ -62,6 +62,18 @@ PartitionableHashTable::PartitionableHashTable(ClientContext &context, Allocator
 	for (hash_t r = 0; r < partition_info.n_partitions; r++) {
 		sel_vectors[r].Initialize();
 	}
+
+	RowLayout layout;
+	layout.Initialize(group_types, AggregateObject::CreateAggregateObjects(bindings));
+	tuple_size = layout.GetRowWidth();
+}
+
+HtEntryType PartitionableHashTable::GetHTEntrySize() {
+	// we need at least STANDARD_VECTOR_SIZE entries to fit in the hash table
+	if (GroupedAggregateHashTable::GetMaxCapacity(HtEntryType::HT_WIDTH_32, tuple_size) < STANDARD_VECTOR_SIZE) {
+		return HtEntryType::HT_WIDTH_64;
+	}
+	return HtEntryType::HT_WIDTH_32;
 }
 
 idx_t PartitionableHashTable::ListAddChunk(HashTableList &list, DataChunk &groups, Vector &group_hashes,
@@ -74,7 +86,7 @@ idx_t PartitionableHashTable::ListAddChunk(HashTableList &list, DataChunk &group
 			list.back()->Finalize();
 		}
 		list.push_back(make_unique<GroupedAggregateHashTable>(context, allocator, group_types, payload_types, bindings,
-		                                                      HtEntryType::HT_WIDTH_32));
+		                                                      GetHTEntrySize()));
 	}
 	return list.back()->AddChunk(groups, group_hashes, payload, filter);
 }
@@ -141,7 +153,7 @@ void PartitionableHashTable::Partition() {
 	for (auto &unpartitioned_ht : unpartitioned_hts) {
 		for (idx_t r = 0; r < partition_info.n_partitions; r++) {
 			radix_partitioned_hts[r].push_back(make_unique<GroupedAggregateHashTable>(
-			    context, allocator, group_types, payload_types, bindings, HtEntryType::HT_WIDTH_32));
+			    context, allocator, group_types, payload_types, bindings, GetHTEntrySize()));
 			partition_hts[r] = radix_partitioned_hts[r].back().get();
 		}
 		unpartitioned_ht->Partition(partition_hts, partition_info.radix_mask, partition_info.RADIX_SHIFT);

package/src/duckdb/src/function/cast/cast_function_set.cpp

@@ -6,7 +6,7 @@
 
 namespace duckdb {
 
-BindCastInput::BindCastInput(CastFunctionSet &function_set, BindCastInfo *info, ClientContext *context)
+BindCastInput::BindCastInput(CastFunctionSet &function_set, BindCastInfo *info, optional_ptr<ClientContext> context)
     : function_set(function_set), info(info), context(context) {
 }
 

package/src/duckdb/src/function/cast/enum_casts.cpp

@@ -109,17 +109,39 @@ unique_ptr<BoundCastData> BindEnumCast(BindCastInput &input, const LogicalType &
 	return make_unique<EnumBoundCastData>(std::move(to_varchar_cast), std::move(from_varchar_cast));
 }
 
+struct EnumCastLocalState : public FunctionLocalState {
+public:
+	unique_ptr<FunctionLocalState> to_varchar_local;
+	unique_ptr<FunctionLocalState> from_varchar_local;
+};
+
+static unique_ptr<FunctionLocalState> InitEnumCastLocalState(CastLocalStateParameters &parameters) {
+	auto &cast_data = (EnumBoundCastData &)*parameters.cast_data;
+	auto result = make_unique<EnumCastLocalState>();
+
+	if (cast_data.from_varchar_cast.init_local_state) {
+		CastLocalStateParameters from_varchar_params(parameters, cast_data.from_varchar_cast.cast_data);
+		result->from_varchar_local = cast_data.from_varchar_cast.init_local_state(from_varchar_params);
+	}
+	if (cast_data.to_varchar_cast.init_local_state) {
+		CastLocalStateParameters from_varchar_params(parameters, cast_data.to_varchar_cast.cast_data);
+		result->from_varchar_local = cast_data.to_varchar_cast.init_local_state(from_varchar_params);
+	}
+	return std::move(result);
+}
+
 static bool EnumToAnyCast(Vector &source, Vector &result, idx_t count, CastParameters &parameters) {
 	auto &cast_data = (EnumBoundCastData &)*parameters.cast_data;
+	auto &lstate = (EnumCastLocalState &)*parameters.local_state;
 
 	Vector varchar_cast(LogicalType::VARCHAR, count);
 
 	// cast to varchar
-	CastParameters to_varchar_params(parameters, cast_data.to_varchar_cast.cast_data.get());
+	CastParameters to_varchar_params(parameters, cast_data.to_varchar_cast.cast_data, lstate.to_varchar_local);
 	cast_data.to_varchar_cast.function(source, varchar_cast, count, to_varchar_params);
 
 	// cast from varchar to the target
-	CastParameters from_varchar_params(parameters, cast_data.from_varchar_cast.cast_data.get());
+	CastParameters from_varchar_params(parameters, cast_data.from_varchar_cast.cast_data, lstate.from_varchar_local);
 	cast_data.from_varchar_cast.function(varchar_cast, result, count, from_varchar_params);
 	return true;
 }
@@ -152,7 +174,7 @@ BoundCastInfo DefaultCasts::EnumCastSwitch(BindCastInput &input, const LogicalTy
 			throw InternalException("ENUM can only have unsigned integers (except UINT64) as physical types");
 		}
 	default: {
-		return BoundCastInfo(EnumToAnyCast, BindEnumCast(input, source, target));
+		return BoundCastInfo(EnumToAnyCast, BindEnumCast(input, source, target), InitEnumCastLocalState);
 	}
 	}
 }

package/src/duckdb/src/function/cast/list_casts.cpp

@@ -1,5 +1,6 @@
 #include "duckdb/function/cast/default_casts.hpp"
 #include "duckdb/function/cast/cast_function_set.hpp"
+#include "duckdb/function/cast/bound_cast_data.hpp"
 
 namespace duckdb {
 
@@ -12,6 +13,15 @@ unique_ptr<BoundCastData> ListBoundCastData::BindListToListCast(BindCastInput &i
 	return make_unique<ListBoundCastData>(std::move(child_cast));
 }
 
+unique_ptr<FunctionLocalState> ListBoundCastData::InitListLocalState(CastLocalStateParameters &parameters) {
+	auto &cast_data = (ListBoundCastData &)*parameters.cast_data;
+	if (!cast_data.child_cast_info.init_local_state) {
+		return nullptr;
+	}
+	CastLocalStateParameters child_parameters(parameters, cast_data.child_cast_info.cast_data);
+	return cast_data.child_cast_info.init_local_state(child_parameters);
+}
+
 bool ListCast::ListToListCast(Vector &source, Vector &result, idx_t count, CastParameters &parameters) {
 	auto &cast_data = (ListBoundCastData &)*parameters.cast_data;
 
@@ -40,7 +50,7 @@ bool ListCast::ListToListCast(Vector &source, Vector &result, idx_t count, CastP
 	ListVector::Reserve(result, source_size);
 	auto &append_vector = ListVector::GetEntry(result);
 
-	CastParameters child_parameters(parameters, cast_data.child_cast_info.cast_data.get());
+	CastParameters child_parameters(parameters, cast_data.child_cast_info.cast_data, parameters.local_state);
 	if (!cast_data.child_cast_info.function(source_cc, append_vector, source_size, child_parameters)) {
 		return false;
 	}
@@ -116,10 +126,13 @@ static bool ListToVarcharCast(Vector &source, Vector &result, idx_t count, CastP
 BoundCastInfo DefaultCasts::ListCastSwitch(BindCastInput &input, const LogicalType &source, const LogicalType &target) {
 	switch (target.id()) {
 	case LogicalTypeId::LIST:
-		return BoundCastInfo(ListCast::ListToListCast, ListBoundCastData::BindListToListCast(input, source, target));
+		return BoundCastInfo(ListCast::ListToListCast, ListBoundCastData::BindListToListCast(input, source, target),
+		                     ListBoundCastData::InitListLocalState);
 	case LogicalTypeId::VARCHAR:
-		return BoundCastInfo(ListToVarcharCast, ListBoundCastData::BindListToListCast(
-		                                            input, source, LogicalType::LIST(LogicalType::VARCHAR)));
+		return BoundCastInfo(
+		    ListToVarcharCast,
+		    ListBoundCastData::BindListToListCast(input, source, LogicalType::LIST(LogicalType::VARCHAR)),
+		    ListBoundCastData::InitListLocalState);
 	default:
 		return DefaultCasts::TryVectorNullCast;
 	}

package/src/duckdb/src/function/cast/map_cast.cpp

@@ -1,5 +1,6 @@
 #include "duckdb/function/cast/default_casts.hpp"
 #include "duckdb/function/cast/cast_function_set.hpp"
+#include "duckdb/function/cast/bound_cast_data.hpp"
 
 namespace duckdb {
 
@@ -78,10 +79,12 @@ static bool MapToVarcharCast(Vector &source, Vector &result, idx_t count, CastPa
 BoundCastInfo DefaultCasts::MapCastSwitch(BindCastInput &input, const LogicalType &source, const LogicalType &target) {
 	switch (target.id()) {
 	case LogicalTypeId::MAP:
-		return BoundCastInfo(ListCast::ListToListCast, ListBoundCastData::BindListToListCast(input, source, target));
+		return BoundCastInfo(ListCast::ListToListCast, ListBoundCastData::BindListToListCast(input, source, target),
+		                     ListBoundCastData::InitListLocalState);
 	case LogicalTypeId::VARCHAR: {
 		auto varchar_type = LogicalType::MAP(LogicalType::VARCHAR, LogicalType::VARCHAR);
-		return BoundCastInfo(MapToVarcharCast, ListBoundCastData::BindListToListCast(input, source, varchar_type));
+		return BoundCastInfo(MapToVarcharCast, ListBoundCastData::BindListToListCast(input, source, varchar_type),
+		                     ListBoundCastData::InitListLocalState);
 	}
 	default:
 		return TryVectorNullCast;

package/src/duckdb/src/function/cast/string_cast.cpp

@@ -3,6 +3,7 @@
 #include "duckdb/common/pair.hpp"
 #include "duckdb/common/vector.hpp"
 #include "duckdb/function/scalar/nested_functions.hpp"
+#include "duckdb/function/cast/bound_cast_data.hpp"
 
 namespace duckdb {
 
@@ -115,7 +116,9 @@ static BoundCastInfo VectorStringCastNumericSwitch(BindCastInput &input, const L
 	}
 }
 
+//===--------------------------------------------------------------------===//
 // string -> list casting
+//===--------------------------------------------------------------------===//
 bool VectorStringToList::StringToNestedTypeCastLoop(string_t *source_data, ValidityMask &source_mask, Vector &result,
                                                     ValidityMask &result_mask, idx_t count, CastParameters &parameters,
                                                     const SelectionVector *sel) {
@@ -163,7 +166,7 @@ bool VectorStringToList::StringToNestedTypeCastLoop(string_t *source_data, Valid
 
 	auto &result_child = ListVector::GetEntry(result);
 	auto &cast_data = (ListBoundCastData &)*parameters.cast_data;
-	CastParameters child_parameters(parameters, cast_data.child_cast_info.cast_data.get());
+	CastParameters child_parameters(parameters, cast_data.child_cast_info.cast_data, parameters.local_state);
 	return cast_data.child_cast_info.function(varchar_vector, result_child, total_list_size, child_parameters) &&
 	       all_converted;
 }
@@ -177,11 +180,12 @@ static LogicalType InitVarcharStructType(const LogicalType &target) {
 	return LogicalType::STRUCT(child_types);
 }
 
+//===--------------------------------------------------------------------===//
 // string -> struct casting
+//===--------------------------------------------------------------------===//
 bool VectorStringToStruct::StringToNestedTypeCastLoop(string_t *source_data, ValidityMask &source_mask, Vector &result,
                                                       ValidityMask &result_mask, idx_t count,
                                                       CastParameters &parameters, const SelectionVector *sel) {
-
 	auto varchar_struct_type = InitVarcharStructType(result.GetType());
 	Vector varchar_vector(varchar_struct_type, count);
 	auto &child_vectors = StructVector::GetEntries(varchar_vector);
@@ -216,21 +220,39 @@ bool VectorStringToStruct::StringToNestedTypeCastLoop(string_t *source_data, Val
 	}
 
 	auto &cast_data = (StructBoundCastData &)*parameters.cast_data;
+	auto &lstate = (StructCastLocalState &)*parameters.local_state;
 	D_ASSERT(cast_data.child_cast_info.size() == result_children.size());
 
 	for (idx_t child_idx = 0; child_idx < result_children.size(); child_idx++) {
-		auto &varchar_vector = *child_vectors[child_idx];
+		auto &child_varchar_vector = *child_vectors[child_idx];
 		auto &result_child_vector = *result_children[child_idx];
 		auto &child_cast_info = cast_data.child_cast_info[child_idx];
-		CastParameters child_parameters(parameters, child_cast_info.cast_data.get());
-		if (!child_cast_info.function(varchar_vector, result_child_vector, count, child_parameters)) {
+		CastParameters child_parameters(parameters, child_cast_info.cast_data, lstate.local_states[child_idx]);
+		if (!child_cast_info.function(child_varchar_vector, result_child_vector, count, child_parameters)) {
 			all_converted = false;
 		}
 	}
 	return all_converted;
 }
 
+//===--------------------------------------------------------------------===//
 // string -> map casting
+//===--------------------------------------------------------------------===//
+unique_ptr<FunctionLocalState> InitMapCastLocalState(CastLocalStateParameters &parameters) {
+	auto &cast_data = (MapBoundCastData &)*parameters.cast_data;
+	auto result = make_unique<MapCastLocalState>();
+
+	if (cast_data.key_cast.init_local_state) {
+		CastLocalStateParameters child_params(parameters, cast_data.key_cast.cast_data);
+		result->key_state = cast_data.key_cast.init_local_state(child_params);
+	}
+	if (cast_data.value_cast.init_local_state) {
+		CastLocalStateParameters child_params(parameters, cast_data.value_cast.cast_data);
+		result->value_state = cast_data.value_cast.init_local_state(child_params);
+	}
+	return std::move(result);
+}
+
 bool VectorStringToMap::StringToNestedTypeCastLoop(string_t *source_data, ValidityMask &source_mask, Vector &result,
                                                    ValidityMask &result_mask, idx_t count, CastParameters &parameters,
                                                    const SelectionVector *sel) {
@@ -282,12 +304,13 @@ bool VectorStringToMap::StringToNestedTypeCastLoop(string_t *source_data, Validi
 	auto &result_key_child = MapVector::GetKeys(result);
 	auto &result_val_child = MapVector::GetValues(result);
 	auto &cast_data = (MapBoundCastData &)*parameters.cast_data;
+	auto &lstate = (MapCastLocalState &)*parameters.local_state;
 
-	CastParameters key_params(parameters, cast_data.key_cast.cast_data.get());
+	CastParameters key_params(parameters, cast_data.key_cast.cast_data, lstate.key_state);
 	if (!cast_data.key_cast.function(varchar_key_vector, result_key_child, total_elements, key_params)) {
 		all_converted = false;
 	}
-	CastParameters val_params(parameters, cast_data.value_cast.cast_data.get());
+	CastParameters val_params(parameters, cast_data.value_cast.cast_data, lstate.value_state);
 	if (!cast_data.value_cast.function(varchar_val_vector, result_val_child, total_elements, val_params)) {
 		all_converted = false;
 	}
@@ -373,14 +396,17 @@ BoundCastInfo DefaultCasts::StringCastSwitch(BindCastInput &input, const Logical
 		// the second argument allows for a secondary casting function to be passed in the CastParameters
 		return BoundCastInfo(
 		    &StringToNestedTypeCast<VectorStringToList>,
-		    ListBoundCastData::BindListToListCast(input, LogicalType::LIST(LogicalType::VARCHAR), target));
+		    ListBoundCastData::BindListToListCast(input, LogicalType::LIST(LogicalType::VARCHAR), target),
+		    ListBoundCastData::InitListLocalState);
 	case LogicalTypeId::STRUCT:
 		return BoundCastInfo(&StringToNestedTypeCast<VectorStringToStruct>,
-		                     StructBoundCastData::BindStructToStructCast(input, InitVarcharStructType(target), target));
+		                     StructBoundCastData::BindStructToStructCast(input, InitVarcharStructType(target), target),
+		                     StructBoundCastData::InitStructCastLocalState);
 	case LogicalTypeId::MAP:
 		return BoundCastInfo(&StringToNestedTypeCast<VectorStringToMap>,
 		                     MapBoundCastData::BindMapToMapCast(
-		                         input, LogicalType::MAP(LogicalType::VARCHAR, LogicalType::VARCHAR), target));
+		                         input, LogicalType::MAP(LogicalType::VARCHAR, LogicalType::VARCHAR), target),
+		                     InitMapCastLocalState);
 	default:
 		return VectorStringCastNumericSwitch(input, source, target);
 	}

package/src/duckdb/src/function/cast/struct_cast.cpp

@@ -1,5 +1,6 @@
 #include "duckdb/function/cast/default_casts.hpp"
 #include "duckdb/function/cast/cast_function_set.hpp"
+#include "duckdb/function/cast/bound_cast_data.hpp"
 
 namespace duckdb {
 
@@ -18,8 +19,24 @@ unique_ptr<BoundCastData> StructBoundCastData::BindStructToStructCast(BindCastIn
 	return make_unique<StructBoundCastData>(std::move(child_cast_info), target);
 }
 
+unique_ptr<FunctionLocalState> StructBoundCastData::InitStructCastLocalState(CastLocalStateParameters &parameters) {
+	auto &cast_data = (StructBoundCastData &)*parameters.cast_data;
+	auto result = make_unique<StructCastLocalState>();
+
+	for (auto &entry : cast_data.child_cast_info) {
+		unique_ptr<FunctionLocalState> child_state;
+		if (entry.init_local_state) {
+			CastLocalStateParameters child_params(parameters, entry.cast_data);
+			child_state = entry.init_local_state(child_params);
+		}
+		result->local_states.push_back(std::move(child_state));
+	}
+	return std::move(result);
+}
+
 static bool StructToStructCast(Vector &source, Vector &result, idx_t count, CastParameters &parameters) {
 	auto &cast_data = (StructBoundCastData &)*parameters.cast_data;
+	auto &lstate = (StructCastLocalState &)*parameters.local_state;
 	auto &source_child_types = StructType::GetChildTypes(source.GetType());
 	auto &source_children = StructVector::GetEntries(source);
 	D_ASSERT(source_children.size() == StructType::GetChildTypes(result.GetType()).size());
@@ -29,7 +46,8 @@ static bool StructToStructCast(Vector &source, Vector &result, idx_t count, Cast
 	for (idx_t c_idx = 0; c_idx < source_child_types.size(); c_idx++) {
 		auto &result_child_vector = *result_children[c_idx];
 		auto &source_child_vector = *source_children[c_idx];
-		CastParameters child_parameters(parameters, cast_data.child_cast_info[c_idx].cast_data.get());
+		CastParameters child_parameters(parameters, cast_data.child_cast_info[c_idx].cast_data,
+		                                lstate.local_states[c_idx]);
 		if (!cast_data.child_cast_info[c_idx].function(source_child_vector, result_child_vector, count,
 		                                               child_parameters)) {
 			all_converted = false;
@@ -121,7 +139,8 @@ BoundCastInfo DefaultCasts::StructCastSwitch(BindCastInput &input, const Logical
                                              const LogicalType &target) {
 	switch (target.id()) {
 	case LogicalTypeId::STRUCT:
-		return BoundCastInfo(StructToStructCast, StructBoundCastData::BindStructToStructCast(input, source, target));
+		return BoundCastInfo(StructToStructCast, StructBoundCastData::BindStructToStructCast(input, source, target),
+		                     StructBoundCastData::InitStructCastLocalState);
 	case LogicalTypeId::VARCHAR: {
 		// bind a cast in which we convert all child entries to VARCHAR entries
 		auto &struct_children = StructType::GetChildTypes(source);
@@ -131,7 +150,8 @@ BoundCastInfo DefaultCasts::StructCastSwitch(BindCastInput &input, const Logical
 		}
 		auto varchar_type = LogicalType::STRUCT(varchar_children);
 		return BoundCastInfo(StructToVarcharCast,
-		                     StructBoundCastData::BindStructToStructCast(input, source, varchar_type));
+		                     StructBoundCastData::BindStructToStructCast(input, source, varchar_type),
+		                     StructBoundCastData::InitStructCastLocalState);
 	}
 	default:
 		return TryVectorNullCast;

package/src/duckdb/src/function/cast/union_casts.cpp

@@ -1,6 +1,6 @@
 #include "duckdb/function/cast/cast_function_set.hpp"
 #include "duckdb/function/cast/default_casts.hpp"
-#include "duckdb/function/cast/vector_cast_helpers.hpp"
+#include "duckdb/function/cast/bound_cast_data.hpp"
 
 #include <algorithm> // for std::sort
 
@@ -98,12 +98,21 @@ unique_ptr<BoundCastData> BindToUnionCast(BindCastInput &input, const LogicalTyp
 	return make_unique<ToUnionBoundCastData>(std::move(selected_cast));
 }
 
+unique_ptr<FunctionLocalState> InitToUnionLocalState(CastLocalStateParameters &parameters) {
+	auto &cast_data = (ToUnionBoundCastData &)*parameters.cast_data;
+	if (!cast_data.member_cast_info.init_local_state) {
+		return nullptr;
+	}
+	CastLocalStateParameters child_parameters(parameters, cast_data.member_cast_info.cast_data);
+	return cast_data.member_cast_info.init_local_state(child_parameters);
+}
+
 static bool ToUnionCast(Vector &source, Vector &result, idx_t count, CastParameters &parameters) {
 	D_ASSERT(result.GetType().id() == LogicalTypeId::UNION);
 	auto &cast_data = (ToUnionBoundCastData &)*parameters.cast_data;
 	auto &selected_member_vector = UnionVector::GetMember(result, cast_data.tag);
 
-	CastParameters child_parameters(parameters, cast_data.member_cast_info.cast_data.get());
+	CastParameters child_parameters(parameters, cast_data.member_cast_info.cast_data, parameters.local_state);
 	if (!cast_data.member_cast_info.function(source, selected_member_vector, count, child_parameters)) {
 		return false;
 	}
@@ -118,7 +127,7 @@ static bool ToUnionCast(Vector &source, Vector &result, idx_t count, CastParamet
 
 BoundCastInfo DefaultCasts::ImplicitToUnionCast(BindCastInput &input, const LogicalType &source,
                                                 const LogicalType &target) {
-	return BoundCastInfo(&ToUnionCast, BindToUnionCast(input, source, target));
+	return BoundCastInfo(&ToUnionCast, BindToUnionCast(input, source, target), InitToUnionLocalState);
 }
 
 //--------------------------------------------------------------------------------------------------
@@ -197,8 +206,24 @@ unique_ptr<BoundCastData> BindUnionToUnionCast(BindCastInput &input, const Logic
 	return make_unique<UnionToUnionBoundCastData>(tag_map, std::move(member_casts), target);
 }
 
+unique_ptr<FunctionLocalState> InitUnionToUnionLocalState(CastLocalStateParameters &parameters) {
+	auto &cast_data = (UnionToUnionBoundCastData &)*parameters.cast_data;
+	auto result = make_unique<StructCastLocalState>();
+
+	for (auto &entry : cast_data.member_casts) {
+		unique_ptr<FunctionLocalState> child_state;
+		if (entry.init_local_state) {
+			CastLocalStateParameters child_params(parameters, entry.cast_data);
+			child_state = entry.init_local_state(child_params);
+		}
+		result->local_states.push_back(std::move(child_state));
+	}
+	return std::move(result);
+}
+
 static bool UnionToUnionCast(Vector &source, Vector &result, idx_t count, CastParameters &parameters) {
 	auto &cast_data = (UnionToUnionBoundCastData &)*parameters.cast_data;
+	auto &lstate = (StructCastLocalState &)*parameters.local_state;
 
 	auto source_member_count = UnionType::GetMemberCount(source.GetType());
 	auto target_member_count = UnionType::GetMemberCount(result.GetType());
@@ -213,7 +238,7 @@ static bool UnionToUnionCast(Vector &source, Vector &result, idx_t count, CastPa
 		auto &target_member_vector = UnionVector::GetMember(result, target_member_idx);
 		auto &member_cast = cast_data.member_casts[member_idx];
 
-		CastParameters child_parameters(parameters, member_cast.cast_data.get());
+		CastParameters child_parameters(parameters, member_cast.cast_data, lstate.local_states[member_idx]);
 		if (!member_cast.function(source_member_vector, target_member_vector, count, child_parameters)) {
 			return false;
 		}
@@ -339,10 +364,11 @@ BoundCastInfo DefaultCasts::UnionCastSwitch(BindCastInput &input, const LogicalT
 			varchar_members.push_back(make_pair(UnionType::GetMemberName(source, member_idx), LogicalType::VARCHAR));
 		}
 		auto varchar_type = LogicalType::UNION(std::move(varchar_members));
-		return BoundCastInfo(UnionToVarcharCast, BindUnionToUnionCast(input, source, varchar_type));
-	} break;
+		return BoundCastInfo(UnionToVarcharCast, BindUnionToUnionCast(input, source, varchar_type),
+		                     InitUnionToUnionLocalState);
+	}
 	case LogicalTypeId::UNION:
-		return BoundCastInfo(UnionToUnionCast, BindUnionToUnionCast(input, source, target));
+		return BoundCastInfo(UnionToUnionCast, BindUnionToUnionCast(input, source, target), InitUnionToUnionLocalState);
 	default:
 		return TryVectorNullCast;
 	}