duckdb 0.7.2-dev12.0 → 0.7.2-dev1244.0

package/src/duckdb/src/include/duckdb/function/cast/cast_function_set.hpp

@@ -19,12 +19,12 @@ typedef BoundCastInfo (*bind_cast_function_t)(BindCastInput &input, const Logica
 typedef int64_t (*implicit_cast_cost_t)(const LogicalType &from, const LogicalType &to);
 
 struct GetCastFunctionInput {
-	GetCastFunctionInput(ClientContext *context = nullptr) : context(context) {
+	GetCastFunctionInput(optional_ptr<ClientContext> context = nullptr) : context(context) {
 	}
 	GetCastFunctionInput(ClientContext &context) : context(&context) {
 	}
 
-	ClientContext *context;
+	optional_ptr<ClientContext> context;
 };
 
 struct BindCastFunction {

package/src/duckdb/src/include/duckdb/function/cast/default_casts.hpp

@@ -10,6 +10,8 @@
 
 #include "duckdb/common/types.hpp"
 #include "duckdb/common/types/vector.hpp"
+#include "duckdb/common/optional_ptr.hpp"
+#include "duckdb/function/scalar_function.hpp"
 
 namespace duckdb {
 
@@ -31,25 +33,43 @@ struct BoundCastData {
 struct CastParameters {
 	CastParameters() {
 	}
-	CastParameters(BoundCastData *cast_data, bool strict, string *error_message, FunctionLocalState *local_state)
+	CastParameters(BoundCastData *cast_data, bool strict, string *error_message,
+	               optional_ptr<FunctionLocalState> local_state)
 	    : cast_data(cast_data), strict(strict), error_message(error_message), local_state(local_state) {
 	}
-	CastParameters(CastParameters &parent, BoundCastData *cast_data = nullptr)
-	    : cast_data(cast_data), strict(parent.strict), error_message(parent.error_message) {
+	CastParameters(CastParameters &parent, optional_ptr<BoundCastData> cast_data,
+	               optional_ptr<FunctionLocalState> local_state)
+	    : cast_data(cast_data), strict(parent.strict), error_message(parent.error_message), local_state(local_state) {
 	}
 
 	//! The bound cast data (if any)
-	BoundCastData *cast_data = nullptr;
+	optional_ptr<BoundCastData> cast_data;
 	//! whether or not to enable strict casting
 	bool strict = false;
 	// out: error message in case cast has failed
 	string *error_message = nullptr;
 	//! Local state
-	FunctionLocalState *local_state = nullptr;
+	optional_ptr<FunctionLocalState> local_state;
+};
+
+struct CastLocalStateParameters {
+	CastLocalStateParameters(optional_ptr<ClientContext> context_p, optional_ptr<BoundCastData> cast_data_p)
+	    : context(context_p), cast_data(cast_data_p) {
+	}
+	CastLocalStateParameters(ClientContext &context_p, optional_ptr<BoundCastData> cast_data_p)
+	    : context(&context_p), cast_data(cast_data_p) {
+	}
+	CastLocalStateParameters(CastLocalStateParameters &parent, optional_ptr<BoundCastData> cast_data_p)
+	    : context(parent.context), cast_data(cast_data_p) {
+	}
+
+	optional_ptr<ClientContext> context;
+	//! The bound cast data (if any)
+	optional_ptr<BoundCastData> cast_data;
 };
 
 typedef bool (*cast_function_t)(Vector &source, Vector &result, idx_t count, CastParameters &parameters);
-typedef unique_ptr<FunctionLocalState> (*init_cast_local_state_t)(ClientContext &context);
+typedef unique_ptr<FunctionLocalState> (*init_cast_local_state_t)(CastLocalStateParameters &parameters);
 
 struct BoundCastInfo {
 	DUCKDB_API
@@ -65,72 +85,16 @@ public:
 };
 
 struct BindCastInput {
-	DUCKDB_API BindCastInput(CastFunctionSet &function_set, BindCastInfo *info, ClientContext *context);
+	DUCKDB_API BindCastInput(CastFunctionSet &function_set, BindCastInfo *info, optional_ptr<ClientContext> context);
 
 	CastFunctionSet &function_set;
 	BindCastInfo *info;
-	ClientContext *context;
+	optional_ptr<ClientContext> context;
 
 public:
 	DUCKDB_API BoundCastInfo GetCastFunction(const LogicalType &source, const LogicalType &target);
 };
 
-struct ListBoundCastData : public BoundCastData {
-	explicit ListBoundCastData(BoundCastInfo child_cast) : child_cast_info(std::move(child_cast)) {
-	}
-
-	BoundCastInfo child_cast_info;
-	static unique_ptr<BoundCastData> BindListToListCast(BindCastInput &input, const LogicalType &source,
-	                                                    const LogicalType &target);
-
-public:
-	unique_ptr<BoundCastData> Copy() const override {
-		return make_unique<ListBoundCastData>(child_cast_info.Copy());
-	}
-};
-
-struct ListCast {
-	static bool ListToListCast(Vector &source, Vector &result, idx_t count, CastParameters &parameters);
-};
-
-struct StructBoundCastData : public BoundCastData {
-	StructBoundCastData(vector<BoundCastInfo> child_casts, LogicalType target_p)
-	    : child_cast_info(std::move(child_casts)), target(std::move(target_p)) {
-	}
-
-	vector<BoundCastInfo> child_cast_info;
-	LogicalType target;
-
-	static unique_ptr<BoundCastData> BindStructToStructCast(BindCastInput &input, const LogicalType &source,
-	                                                        const LogicalType &target);
-
-public:
-	unique_ptr<BoundCastData> Copy() const override {
-		vector<BoundCastInfo> copy_info;
-		for (auto &info : child_cast_info) {
-			copy_info.push_back(info.Copy());
-		}
-		return make_unique<StructBoundCastData>(std::move(copy_info), target);
-	}
-};
-
-struct MapBoundCastData : public BoundCastData {
-	MapBoundCastData(BoundCastInfo key_cast, BoundCastInfo value_cast)
-	    : key_cast(std::move(key_cast)), value_cast(std::move(value_cast)) {
-	}
-
-	BoundCastInfo key_cast;
-	BoundCastInfo value_cast;
-
-	static unique_ptr<BoundCastData> BindMapToMapCast(BindCastInput &input, const LogicalType &source,
-	                                                  const LogicalType &target);
-
-public:
-	unique_ptr<BoundCastData> Copy() const override {
-		return make_unique<MapBoundCastData>(key_cast.Copy(), value_cast.Copy());
-	}
-};
-
 struct DefaultCasts {
 	DUCKDB_API static BoundCastInfo GetDefaultCastFunction(BindCastInput &input, const LogicalType &source,
 	                                                       const LogicalType &target);

package/src/duckdb/src/include/duckdb/function/function_binder.hpp

@@ -59,13 +59,10 @@ public:
 	DUCKDB_API unique_ptr<BoundAggregateExpression>
 	BindAggregateFunction(AggregateFunction bound_function, vector<unique_ptr<Expression>> children,
 	                      unique_ptr<Expression> filter = nullptr,
-	                      AggregateType aggr_type = AggregateType::NON_DISTINCT,
-	                      unique_ptr<BoundOrderModifier> order_bys = nullptr);
+	                      AggregateType aggr_type = AggregateType::NON_DISTINCT);
 
-	DUCKDB_API unique_ptr<FunctionData> BindSortedAggregate(AggregateFunction &bound_function,
-	                                                        vector<unique_ptr<Expression>> &children,
-	                                                        unique_ptr<FunctionData> bind_info,
-	                                                        unique_ptr<BoundOrderModifier> order_bys);
+	DUCKDB_API static void BindSortedAggregate(ClientContext &context, BoundAggregateExpression &expr,
+	                                           const vector<unique_ptr<Expression>> &groups);
 
 private:
 	//! Cast a set of expressions to the arguments of this function

package/src/duckdb/src/include/duckdb/function/scalar/bit_functions.hpp

@@ -25,4 +25,8 @@ struct BitPositionFun {
 	static void RegisterFunction(BuiltinFunctions &set);
 };
 
+struct BitStringFun {
+	static void RegisterFunction(BuiltinFunctions &set);
+};
+
 } // namespace duckdb

package/src/duckdb/src/include/duckdb/function/scalar/list/contains_or_position.hpp

@@ -0,0 +1,138 @@
+#pragma once
+
+namespace duckdb {
+
+template <class CHILD_TYPE, class RETURN_TYPE, class OP, class LIST_ACCESSOR>
+static void TemplatedContainsOrPosition(DataChunk &args, Vector &result, bool is_nested = false) {
+	D_ASSERT(args.ColumnCount() == 2);
+	auto count = args.size();
+	Vector &list = LIST_ACCESSOR::GetList(args.data[0]);
+	Vector &value_vector = args.data[1];
+
+	// Create a result vector of type RETURN_TYPE
+	result.SetVectorType(VectorType::FLAT_VECTOR);
+	auto result_entries = FlatVector::GetData<RETURN_TYPE>(result);
+	auto &result_validity = FlatVector::Validity(result);
+
+	if (list.GetType().id() == LogicalTypeId::SQLNULL) {
+		result_validity.SetInvalid(0);
+		return;
+	}
+
+	auto list_size = LIST_ACCESSOR::GetListSize(list);
+	auto &child_vector = LIST_ACCESSOR::GetEntry(list);
+
+	UnifiedVectorFormat child_data;
+	child_vector.ToUnifiedFormat(list_size, child_data);
+
+	UnifiedVectorFormat list_data;
+	list.ToUnifiedFormat(count, list_data);
+	auto list_entries = (list_entry_t *)list_data.data;
+
+	UnifiedVectorFormat value_data;
+	value_vector.ToUnifiedFormat(count, value_data);
+
+	// not required for a comparison of nested types
+	auto child_value = (CHILD_TYPE *)child_data.data;
+	auto values = (CHILD_TYPE *)value_data.data;
+
+	for (idx_t i = 0; i < count; i++) {
+		auto list_index = list_data.sel->get_index(i);
+		auto value_index = value_data.sel->get_index(i);
+
+		if (!list_data.validity.RowIsValid(list_index) || !value_data.validity.RowIsValid(value_index)) {
+			result_validity.SetInvalid(i);
+			continue;
+		}
+
+		const auto &list_entry = list_entries[list_index];
+
+		result_entries[i] = OP::Initialize();
+		for (idx_t child_idx = 0; child_idx < list_entry.length; child_idx++) {
+
+			auto child_value_idx = child_data.sel->get_index(list_entry.offset + child_idx);
+			if (!child_data.validity.RowIsValid(child_value_idx)) {
+				continue;
+			}
+
+			if (!is_nested) {
+				if (Equals::Operation(child_value[child_value_idx], values[value_index])) {
+					result_entries[i] = OP::UpdateResultEntries(child_idx);
+					break; // Found value in list, no need to look further
+				}
+			} else {
+				// FIXME: using Value is less efficient than modifying the vector comparison code
+				// to more efficiently compare nested types
+
+				// Note: When using GetValue we don't first apply the selection vector
+				// because it is already done inside GetValue
+				auto lvalue = child_vector.GetValue(list_entry.offset + child_idx);
+				auto rvalue = value_vector.GetValue(i);
+				if (Value::NotDistinctFrom(lvalue, rvalue)) {
+					result_entries[i] = OP::UpdateResultEntries(child_idx);
+					break; // Found value in list, no need to look further
+				}
+			}
+		}
+	}
+
+	if (args.AllConstant()) {
+		result.SetVectorType(VectorType::CONSTANT_VECTOR);
+	}
+}
+
+template <class T, class OP, class LIST_ACCESSOR>
+void ListContainsOrPosition(DataChunk &args, Vector &result) {
+	const auto physical_type = args.data[1].GetType().InternalType();
+	switch (physical_type) {
+	case PhysicalType::BOOL:
+	case PhysicalType::INT8:
+		TemplatedContainsOrPosition<int8_t, T, OP, LIST_ACCESSOR>(args, result);
+		break;
+	case PhysicalType::INT16:
+		TemplatedContainsOrPosition<int16_t, T, OP, LIST_ACCESSOR>(args, result);
+		break;
+	case PhysicalType::INT32:
+		TemplatedContainsOrPosition<int32_t, T, OP, LIST_ACCESSOR>(args, result);
+		break;
+	case PhysicalType::INT64:
+		TemplatedContainsOrPosition<int64_t, T, OP, LIST_ACCESSOR>(args, result);
+		break;
+	case PhysicalType::INT128:
+		TemplatedContainsOrPosition<hugeint_t, T, OP, LIST_ACCESSOR>(args, result);
+		break;
+	case PhysicalType::UINT8:
+		TemplatedContainsOrPosition<uint8_t, T, OP, LIST_ACCESSOR>(args, result);
+		break;
+	case PhysicalType::UINT16:
+		TemplatedContainsOrPosition<uint16_t, T, OP, LIST_ACCESSOR>(args, result);
+		break;
+	case PhysicalType::UINT32:
+		TemplatedContainsOrPosition<uint32_t, T, OP, LIST_ACCESSOR>(args, result);
+		break;
+	case PhysicalType::UINT64:
+		TemplatedContainsOrPosition<uint64_t, T, OP, LIST_ACCESSOR>(args, result);
+		break;
+	case PhysicalType::FLOAT:
+		TemplatedContainsOrPosition<float, T, OP, LIST_ACCESSOR>(args, result);
+		break;
+	case PhysicalType::DOUBLE:
+		TemplatedContainsOrPosition<double, T, OP, LIST_ACCESSOR>(args, result);
+		break;
+	case PhysicalType::VARCHAR:
+		TemplatedContainsOrPosition<string_t, T, OP, LIST_ACCESSOR>(args, result);
+		break;
+	case PhysicalType::INTERVAL:
+		TemplatedContainsOrPosition<interval_t, T, OP, LIST_ACCESSOR>(args, result);
+		break;
+	case PhysicalType::STRUCT:
+	case PhysicalType::LIST:
+		TemplatedContainsOrPosition<int8_t, T, OP, LIST_ACCESSOR>(args, result, true);
+		break;
+	default:
+		throw NotImplementedException("This function has not been implemented for physical type %s",
+		                              TypeIdToString(physical_type));
+	}
+}
+
+} // namespace duckdb

package/src/duckdb/src/include/duckdb/function/scalar/math_functions.hpp

@@ -122,4 +122,12 @@ struct IsFiniteFun {
 	static void RegisterFunction(BuiltinFunctions &set);
 };
 
+struct GreatestCommonDivisorFun {
+	static void RegisterFunction(BuiltinFunctions &set);
+};
+
+struct LeastCommonMultipleFun {
+	static void RegisterFunction(BuiltinFunctions &set);
+};
+
 } // namespace duckdb

package/src/duckdb/src/include/duckdb/function/scalar/nested_functions.hpp

@@ -14,12 +14,59 @@
 #include "duckdb/common/unordered_map.hpp"
 #include "duckdb/common/field_writer.hpp"
 #include "duckdb/function/built_in_functions.hpp"
+#include "duckdb/function/scalar/list/contains_or_position.hpp"
 
 namespace duckdb {
 
 enum class MapInvalidReason : uint8_t { VALID, NULL_KEY_LIST, NULL_KEY, DUPLICATE_KEY };
 enum class UnionInvalidReason : uint8_t { VALID, TAG_OUT_OF_RANGE, NO_MEMBERS, VALIDITY_OVERLAP };
 
+struct ListArgFunctor {
+	static Vector &GetList(Vector &list) {
+		return list;
+	}
+	static idx_t GetListSize(Vector &list) {
+		return ListVector::GetListSize(list);
+	}
+	static Vector &GetEntry(Vector &list) {
+		return ListVector::GetEntry(list);
+	}
+};
+
+struct MapKeyArgFunctor {
+	// MAP is a LIST(STRUCT(K,V))
+	// meaning the MAP itself is a List, but the child vector that we're interested in (the keys)
+	// are a level deeper than the initial child vector
+
+	static Vector &GetList(Vector &map) {
+		return map;
+	}
+	static idx_t GetListSize(Vector &map) {
+		return ListVector::GetListSize(map);
+	}
+	static Vector &GetEntry(Vector &map) {
+		return MapVector::GetKeys(map);
+	}
+};
+
+struct ContainsFunctor {
+	static inline bool Initialize() {
+		return false;
+	}
+	static inline bool UpdateResultEntries(idx_t child_idx) {
+		return true;
+	}
+};
+
+struct PositionFunctor {
+	static inline int32_t Initialize() {
+		return 0;
+	}
+	static inline int32_t UpdateResultEntries(idx_t child_idx) {
+		return child_idx + 1;
+	}
+};
+
 struct VariableReturnBindData : public FunctionData {
 	LogicalType stype;
 
@@ -80,6 +127,18 @@ struct MapFromEntriesFun {
 	static void RegisterFunction(BuiltinFunctions &set);
 };
 
+struct MapEntriesFun {
+	static void RegisterFunction(BuiltinFunctions &set);
+};
+
+struct MapValuesFun {
+	static void RegisterFunction(BuiltinFunctions &set);
+};
+
+struct MapKeysFun {
+	static void RegisterFunction(BuiltinFunctions &set);
+};
+
 struct MapExtractFun {
 	static void RegisterFunction(BuiltinFunctions &set);
 };

package/src/duckdb/src/include/duckdb/function/scalar/regexp.hpp

@@ -11,9 +11,36 @@
 #include "duckdb/function/function_set.hpp"
 #include "re2/re2.h"
 #include "duckdb/function/built_in_functions.hpp"
+#include "re2/stringpiece.h"
 
 namespace duckdb {
 
+namespace regexp_util {
+
+bool TryParseConstantPattern(ClientContext &context, Expression &expr, string &constant_string);
+void ParseRegexOptions(const string &options, duckdb_re2::RE2::Options &result, bool *global_replace = nullptr);
+void ParseRegexOptions(ClientContext &context, Expression &expr, RE2::Options &target, bool *global_replace = nullptr);
+
+inline duckdb_re2::StringPiece CreateStringPiece(const string_t &input) {
+	return duckdb_re2::StringPiece(input.GetDataUnsafe(), input.GetSize());
+}
+
+inline string_t Extract(const string_t &input, Vector &result, const RE2 &re, const duckdb_re2::StringPiece &rewrite) {
+	string extracted;
+	RE2::Extract(input.GetString(), re, rewrite, &extracted);
+	return StringVector::AddString(result, extracted.c_str(), extracted.size());
+}
+
+} // namespace regexp_util
+
+struct RegexpExtractAll {
+	static void Execute(DataChunk &args, ExpressionState &state, Vector &result);
+	static unique_ptr<FunctionData> Bind(ClientContext &context, ScalarFunction &bound_function,
+	                                     vector<unique_ptr<Expression>> &arguments);
+	static unique_ptr<FunctionLocalState> InitLocalState(ExpressionState &state, const BoundFunctionExpression &expr,
+	                                                     FunctionData *bind_data);
+};
+
 struct RegexpBaseBindData : public FunctionData {
 	RegexpBaseBindData();
 	RegexpBaseBindData(duckdb_re2::RE2::Options options, string constant_string, bool constant_pattern = true);
@@ -61,14 +88,67 @@ struct RegexpExtractBindData : public RegexpBaseBindData {
 	bool Equals(const FunctionData &other_p) const override;
 };
 
+struct RegexStringPieceArgs {
+	RegexStringPieceArgs() : size(0), capacity(0), group_buffer(nullptr) {
+	}
+	void Init(idx_t size) {
+		this->size = size;
+		// Allocate for one extra, for the all-encompassing match group
+		this->capacity = size + 1;
+		group_buffer = AllocateArray<duckdb_re2::StringPiece>(capacity);
+	}
+	void SetSize(idx_t size) {
+		this->size = size;
+		if (size + 1 > capacity) {
+			Clear();
+			Init(size);
+		}
+	}
+
+	RegexStringPieceArgs &operator=(RegexStringPieceArgs &&other) {
+		std::swap(this->size, other.size);
+		std::swap(this->capacity, other.capacity);
+		std::swap(this->group_buffer, other.group_buffer);
+		return *this;
+	}
+
+	~RegexStringPieceArgs() {
+		Clear();
+	}
+
+private:
+	void Clear() {
+		DeleteArray<duckdb_re2::StringPiece>(group_buffer, capacity);
+		group_buffer = nullptr;
+
+		size = 0;
+		capacity = 0;
+	}
+
+public:
+	idx_t size;
+	//! The currently allocated capacity for the groups
+	idx_t capacity;
+	//! Used by ExtractAll to pre-allocate the storage for the groups
+	duckdb_re2::StringPiece *group_buffer;
+};
+
 struct RegexLocalState : public FunctionLocalState {
-	explicit RegexLocalState(RegexpBaseBindData &info)
+	explicit RegexLocalState(RegexpBaseBindData &info, bool extract_all = false)
 	    : constant_pattern(duckdb_re2::StringPiece(info.constant_string.c_str(), info.constant_string.size()),
 	                       info.options) {
+		if (extract_all) {
+			auto group_count_p = constant_pattern.NumberOfCapturingGroups();
+			if (group_count_p != -1) {
+				group_buffer.Init(group_count_p);
+			}
+		}
 		D_ASSERT(info.constant_pattern);
 	}
 
 	RE2 constant_pattern;
+	//! Used by regexp_extract_all to pre-allocate the args
+	RegexStringPieceArgs group_buffer;
 };
 
 unique_ptr<FunctionLocalState> RegexInitLocalState(ExpressionState &state, const BoundFunctionExpression &expr,

package/src/duckdb/src/include/duckdb/function/scalar/string_functions.hpp

@@ -213,4 +213,8 @@ struct JaroWinklerFun {
 	static void RegisterFunction(BuiltinFunctions &set);
 };
 
+struct HexFun {
+	static void RegisterFunction(BuiltinFunctions &set);
+};
+
 } // namespace duckdb

package/src/duckdb/src/include/duckdb/function/scalar_function.hpp

@@ -29,13 +29,13 @@ class ScalarFunctionCatalogEntry;
 
 struct FunctionStatisticsInput {
 	FunctionStatisticsInput(BoundFunctionExpression &expr_p, FunctionData *bind_data_p,
-	                        vector<unique_ptr<BaseStatistics>> &child_stats_p, unique_ptr<Expression> *expr_ptr_p)
+	                        vector<BaseStatistics> &child_stats_p, unique_ptr<Expression> *expr_ptr_p)
 	    : expr(expr_p), bind_data(bind_data_p), child_stats(child_stats_p), expr_ptr(expr_ptr_p) {
 	}
 
 	BoundFunctionExpression &expr;
 	FunctionData *bind_data;
-	vector<unique_ptr<BaseStatistics>> &child_stats;
+	vector<BaseStatistics> &child_stats;
 	unique_ptr<Expression> *expr_ptr;
 };
 

package/src/duckdb/src/include/duckdb/function/table/arrow.hpp

@@ -32,7 +32,18 @@ enum class ArrowDateTimeType : uint8_t {
 	NANOSECONDS = 2,
 	SECONDS = 3,
 	DAYS = 4,
-	MONTHS = 5
+	MONTHS = 5,
+	MONTH_DAY_NANO = 6
+};
+
+struct ArrowInterval {
+	int32_t months;
+	int32_t days;
+	int64_t nanoseconds;
+
+	inline bool operator==(const ArrowInterval &rhs) const {
+		return this->days == rhs.days && this->months == rhs.months && this->nanoseconds == rhs.nanoseconds;
+	}
 };
 
 struct ArrowConvertData {

package/src/duckdb/src/include/duckdb/function/table_function.hpp

@@ -11,6 +11,8 @@
 #include "duckdb/common/enums/operator_result_type.hpp"
 #include "duckdb/execution/execution_context.hpp"
 #include "duckdb/function/function.hpp"
+#include "duckdb/planner/bind_context.hpp"
+#include "duckdb/planner/logical_operator.hpp"
 #include "duckdb/storage/statistics/node_statistics.hpp"
 
 #include <functional>
@@ -136,6 +138,7 @@ public:
 
 typedef unique_ptr<FunctionData> (*table_function_bind_t)(ClientContext &context, TableFunctionBindInput &input,
                                                           vector<LogicalType> &return_types, vector<string> &names);
+typedef unique_ptr<TableRef> (*table_function_bind_replace_t)(ClientContext &context, TableFunctionBindInput &input);
 typedef unique_ptr<GlobalTableFunctionState> (*table_function_init_global_t)(ClientContext &context,
                                                                              TableFunctionInitInput &input);
 typedef unique_ptr<LocalTableFunctionState> (*table_function_init_local_t)(ExecutionContext &context,
@@ -185,6 +188,11 @@ public:
 	//! This function is used for determining the return type of a table producing function and returning bind data
 	//! The returned FunctionData object should be constant and should not be changed during execution.
 	table_function_bind_t bind;
+	//! (Optional) Bind replace function
+	//! This function is called before the regular bind function. It allows returning a TableRef will be used to
+	//! to generate a logical plan that replaces the LogicalGet of a regularly bound TableFunction. The BindReplace can
+	//! also return a nullptr to indicate a regular bind needs to be performed instead.
+	table_function_bind_replace_t bind_replace;
 	//! (Optional) global init function
 	//! Initialize the global operator state of the function.
 	//! The global operator state is used to keep track of the progress in the table function and is shared between
@@ -235,6 +243,8 @@ public:
 	bool filter_prune;
 	//! Additional function info, passed to the bind
 	shared_ptr<TableFunctionInfo> function_info;
+
+	DUCKDB_API bool Equal(const TableFunction &rhs) const;
 };
 
 } // namespace duckdb

package/src/duckdb/src/include/duckdb/main/capi/capi_internal.hpp

@@ -40,6 +40,7 @@ struct ExtractStatementsWrapper {
 
 struct PendingStatementWrapper {
 	unique_ptr<PendingQueryResult> statement;
+	bool allow_streaming;
 };
 
 struct ArrowResultWrapper {
@@ -56,6 +57,7 @@ struct AppenderWrapper {
 enum class CAPIResultSetType : uint8_t {
 	CAPI_RESULT_TYPE_NONE = 0,
 	CAPI_RESULT_TYPE_MATERIALIZED,
+	CAPI_RESULT_TYPE_STREAMING,
 	CAPI_RESULT_TYPE_DEPRECATED
 };
 

package/src/duckdb/src/include/duckdb/main/client_config.hpp

@@ -40,6 +40,8 @@ struct ClientConfig {
 	//! to output anything
 	bool emit_profiler_output = true;
 
+	//! system-wide progress bar disable.
+	const char *system_progress_bar_disable_reason = nullptr;
 	//! If the progress bar is enabled or not.
 	bool enable_progress_bar = false;
 	//! If the print of the progress bar is enabled

package/src/duckdb/src/include/duckdb/main/client_data.hpp

@@ -20,7 +20,7 @@ class BufferedFileWriter;
 class ClientContext;
 class CatalogSearchPath;
 class FileOpener;
-class HTTPStats;
+class HTTPState;
 class QueryProfiler;
 class QueryProfilerHistory;
 class PreparedStatementData;
@@ -52,8 +52,8 @@ struct ClientData {
 	//! The file opener of the client context
 	unique_ptr<FileOpener> file_opener;
 
-	//! Statistics on HTTP traffic
-	unique_ptr<HTTPStats> http_stats;
+	//! HTTP State in this query
+	unique_ptr<HTTPState> http_state;
 
 	//! The file search path
 	string file_search_path;

package/src/duckdb/src/include/duckdb/main/config.hpp

@@ -30,6 +30,7 @@
 #include "duckdb/planner/operator_extension.hpp"
 
 namespace duckdb {
+class BufferPool;
 class CastFunctionSet;
 class ClientContext;
 class ErrorManager;
@@ -186,6 +187,8 @@ public:
 	vector<std::unique_ptr<OperatorExtension>> operator_extensions;
 	//! Extensions made to storage
 	case_insensitive_map_t<std::unique_ptr<StorageExtension>> storage_extensions;
+	//! A buffer pool can be shared across multiple databases (if desired).
+	shared_ptr<BufferPool> buffer_pool;
 
 public:
 	DUCKDB_API static DBConfig &GetConfig(ClientContext &context);

package/src/duckdb/src/include/duckdb/main/connection_manager.hpp

@@ -47,6 +47,8 @@ public:
 		return result;
 	}
 
+	ClientContext *GetConnection(DatabaseInstance *db);
+
 	static ConnectionManager &Get(DatabaseInstance &db);
 	static ConnectionManager &Get(ClientContext &context);