Update vendored DuckDB sources to c474827

src/duckdb/extension/json/buffered_json_reader.cpp

@@ -35,7 +35,7 @@ void JSONFileHandle::Reset() {
 	requested_reads = 0;
 	actual_reads = 0;
 	last_read_requested = false;
-	if (IsOpen() && CanSeek()) {
+	if (IsOpen() && !file_handle->IsPipe()) {
 		file_handle->Reset();
 	}
 }

src/duckdb/extension/json/json_functions/read_json_objects.cpp

@@ -64,8 +64,8 @@ TableFunction GetReadJSONObjectsTableFunction(bool list_parameter, shared_ptr<JS
 
 TableFunctionSet JSONFunctions::GetReadJSONObjectsFunction() {
 	TableFunctionSet function_set("read_json_objects");
-	auto function_info =
-	    make_shared_ptr<JSONScanInfo>(JSONScanType::READ_JSON_OBJECTS, JSONFormat::ARRAY, JSONRecordType::RECORDS);
+	auto function_info = make_shared_ptr<JSONScanInfo>(JSONScanType::READ_JSON_OBJECTS, JSONFormat::AUTO_DETECT,
+	                                                   JSONRecordType::RECORDS);
 	function_set.AddFunction(GetReadJSONObjectsTableFunction(false, function_info));
 	function_set.AddFunction(GetReadJSONObjectsTableFunction(true, function_info));
 	return function_set;

src/duckdb/src/common/enum_util.cpp

@@ -2399,19 +2399,20 @@ const StringUtil::EnumStringLiteral *GetMetricsTypeValues() {
 		{ static_cast<uint32_t>(MetricsType::OPTIMIZER_JOIN_FILTER_PUSHDOWN), "OPTIMIZER_JOIN_FILTER_PUSHDOWN" },
 		{ static_cast<uint32_t>(MetricsType::OPTIMIZER_EXTENSION), "OPTIMIZER_EXTENSION" },
 		{ static_cast<uint32_t>(MetricsType::OPTIMIZER_MATERIALIZED_CTE), "OPTIMIZER_MATERIALIZED_CTE" },
-		{ static_cast<uint32_t>(MetricsType::OPTIMIZER_SUM_REWRITER), "OPTIMIZER_SUM_REWRITER" }
+		{ static_cast<uint32_t>(MetricsType::OPTIMIZER_SUM_REWRITER), "OPTIMIZER_SUM_REWRITER" },
+		{ static_cast<uint32_t>(MetricsType::OPTIMIZER_LATE_MATERIALIZATION), "OPTIMIZER_LATE_MATERIALIZATION" }
 	};
 	return values;
 }
 
 template<>
 const char* EnumUtil::ToChars<MetricsType>(MetricsType value) {
-	return StringUtil::EnumToString(GetMetricsTypeValues(), 48, "MetricsType", static_cast<uint32_t>(value));
+	return StringUtil::EnumToString(GetMetricsTypeValues(), 49, "MetricsType", static_cast<uint32_t>(value));
 }
 
 template<>
 MetricsType EnumUtil::FromString<MetricsType>(const char *value) {
-	return static_cast<MetricsType>(StringUtil::StringToEnum(GetMetricsTypeValues(), 48, "MetricsType", value));
+	return static_cast<MetricsType>(StringUtil::StringToEnum(GetMetricsTypeValues(), 49, "MetricsType", value));
 }
 
 const StringUtil::EnumStringLiteral *GetNTypeValues() {
@@ -2584,19 +2585,20 @@ const StringUtil::EnumStringLiteral *GetOptimizerTypeValues() {
 		{ static_cast<uint32_t>(OptimizerType::JOIN_FILTER_PUSHDOWN), "JOIN_FILTER_PUSHDOWN" },
 		{ static_cast<uint32_t>(OptimizerType::EXTENSION), "EXTENSION" },
 		{ static_cast<uint32_t>(OptimizerType::MATERIALIZED_CTE), "MATERIALIZED_CTE" },
-		{ static_cast<uint32_t>(OptimizerType::SUM_REWRITER), "SUM_REWRITER" }
+		{ static_cast<uint32_t>(OptimizerType::SUM_REWRITER), "SUM_REWRITER" },
+		{ static_cast<uint32_t>(OptimizerType::LATE_MATERIALIZATION), "LATE_MATERIALIZATION" }
 	};
 	return values;
 }
 
 template<>
 const char* EnumUtil::ToChars<OptimizerType>(OptimizerType value) {
-	return StringUtil::EnumToString(GetOptimizerTypeValues(), 27, "OptimizerType", static_cast<uint32_t>(value));
+	return StringUtil::EnumToString(GetOptimizerTypeValues(), 28, "OptimizerType", static_cast<uint32_t>(value));
 }
 
 template<>
 OptimizerType EnumUtil::FromString<OptimizerType>(const char *value) {
-	return static_cast<OptimizerType>(StringUtil::StringToEnum(GetOptimizerTypeValues(), 27, "OptimizerType", value));
+	return static_cast<OptimizerType>(StringUtil::StringToEnum(GetOptimizerTypeValues(), 28, "OptimizerType", value));
 }
 
 const StringUtil::EnumStringLiteral *GetOrderByNullTypeValues() {

src/duckdb/src/common/enums/metric_type.cpp

@@ -39,6 +39,7 @@ profiler_settings_t MetricsUtils::GetOptimizerMetrics() {
         MetricsType::OPTIMIZER_EXTENSION,
         MetricsType::OPTIMIZER_MATERIALIZED_CTE,
         MetricsType::OPTIMIZER_SUM_REWRITER,
+        MetricsType::OPTIMIZER_LATE_MATERIALIZATION,
     };
 }
 
@@ -109,6 +110,8 @@ MetricsType MetricsUtils::GetOptimizerMetricByType(OptimizerType type) {
             return MetricsType::OPTIMIZER_MATERIALIZED_CTE;
         case OptimizerType::SUM_REWRITER:
             return MetricsType::OPTIMIZER_SUM_REWRITER;
+        case OptimizerType::LATE_MATERIALIZATION:
+            return MetricsType::OPTIMIZER_LATE_MATERIALIZATION;
        default:
             throw InternalException("OptimizerType %s cannot be converted to a MetricsType", EnumUtil::ToString(type));
     };
@@ -168,6 +171,8 @@ OptimizerType MetricsUtils::GetOptimizerTypeByMetric(MetricsType type) {
             return OptimizerType::MATERIALIZED_CTE;
         case MetricsType::OPTIMIZER_SUM_REWRITER:
             return OptimizerType::SUM_REWRITER;
+        case MetricsType::OPTIMIZER_LATE_MATERIALIZATION:
+            return OptimizerType::LATE_MATERIALIZATION;
     default:
             return OptimizerType::INVALID;
     };
@@ -201,6 +206,7 @@ bool MetricsUtils::IsOptimizerMetric(MetricsType type) {
         case MetricsType::OPTIMIZER_EXTENSION:
         case MetricsType::OPTIMIZER_MATERIALIZED_CTE:
         case MetricsType::OPTIMIZER_SUM_REWRITER:
+        case MetricsType::OPTIMIZER_LATE_MATERIALIZATION:
             return true;
         default:
             return false;

src/duckdb/src/common/enums/optimizer_type.cpp

@@ -38,6 +38,7 @@ static const DefaultOptimizerType internal_optimizer_types[] = {
     {"extension", OptimizerType::EXTENSION},
     {"materialized_cte", OptimizerType::MATERIALIZED_CTE},
     {"sum_rewriter", OptimizerType::SUM_REWRITER},
+    {"late_materialization", OptimizerType::LATE_MATERIALIZATION},
     {nullptr, OptimizerType::INVALID}};
 
 string OptimizerTypeToString(OptimizerType type) {

src/duckdb/src/common/local_file_system.cpp

@@ -853,16 +853,27 @@ unique_ptr<FileHandle> LocalFileSystem::OpenFile(const string &path_p, FileOpenF
 	bool open_write = flags.OpenForWriting();
 	if (open_read && open_write) {
 		desired_access = GENERIC_READ | GENERIC_WRITE;
-		share_mode = 0;
 	} else if (open_read) {
 		desired_access = GENERIC_READ;
-		share_mode = FILE_SHARE_READ;
 	} else if (open_write) {
 		desired_access = GENERIC_WRITE;
-		share_mode = 0;
 	} else {
 		throw InternalException("READ, WRITE or both should be specified when opening a file");
 	}
+	switch (flags.Lock()) {
+	case FileLockType::NO_LOCK:
+		share_mode = FILE_SHARE_READ | FILE_SHARE_WRITE;
+		break;
+	case FileLockType::READ_LOCK:
+		share_mode = FILE_SHARE_READ;
+		break;
+	case FileLockType::WRITE_LOCK:
+		share_mode = 0;
+		break;
+	default:
+		throw InternalException("Unknown FileLockType");
+	}
+
 	if (open_write) {
 		if (flags.CreateFileIfNotExists()) {
 			creation_disposition = OPEN_ALWAYS;

src/duckdb/src/execution/operator/csv_scanner/sniffer/header_detection.cpp

@@ -335,6 +335,13 @@ void CSVSniffer::DetectHeader() {
 	auto &sniffer_state_machine = best_candidate->GetStateMachine();
 	names = DetectHeaderInternal(buffer_manager->context, best_header_row, sniffer_state_machine, set_columns,
 	                             best_sql_types_candidates_per_column_idx, options, *error_handler);
+	if (single_row_file && sniffer_state_machine.dialect_options.header.GetValue()) {
+		// This file only contains a header, lets default to the lowest type of all.
+		detected_types.clear();
+		for (idx_t i = 0; i < names.size(); i++) {
+			detected_types.push_back(LogicalType::BOOLEAN);
+		}
+	}
 	for (idx_t i = max_columns_found; i < names.size(); i++) {
 		detected_types.push_back(LogicalType::VARCHAR);
 	}

src/duckdb/src/execution/operator/csv_scanner/sniffer/type_detection.cpp

@@ -467,6 +467,7 @@ void CSVSniffer::DetectTypes() {
 				best_format_candidates[format_candidate.first] = format_candidate.second.format;
 			}
 			if (chunk_size > 0) {
+				single_row_file = chunk_size == 1;
 				for (idx_t col_idx = 0; col_idx < data_chunk.ColumnCount(); col_idx++) {
 					auto &cur_vector = data_chunk.data[col_idx];
 					auto vector_data = FlatVector::GetData<string_t>(cur_vector);

src/duckdb/src/function/scalar/strftime_format.cpp

@@ -1412,6 +1412,16 @@ StrpTimeFormat::ParseResult StrpTimeFormat::Parse(const string &format_string, c
 	return result;
 }
 
+bool StrpTimeFormat::TryParse(const string &format_string, const string &text, ParseResult &result) {
+	StrpTimeFormat format;
+	format.format_specifier = format_string;
+	string error = StrTimeFormat::ParseFormatSpecifier(format_string, format);
+	if (!error.empty()) {
+		throw InvalidInputException("Failed to parse format specifier %s: %s", format_string, error);
+	}
+	return format.Parse(text, result);
+}
+
 bool StrTimeFormat::Empty() const {
 	return format_specifier.empty();
 }

src/duckdb/src/function/table/table_scan.cpp

@@ -105,7 +105,7 @@ class DuckIndexScanState : public TableScanGlobalState {
 
 	//! The batch index of the next Sink.
 	//! Also determines the offset of the next chunk. I.e., offset = next_batch_index * STANDARD_VECTOR_SIZE.
-	idx_t next_batch_index;
+	atomic<idx_t> next_batch_index;
 	//! The total scanned row IDs.
 	unsafe_vector<row_t> row_ids;
 	//! The column IDs of the to-be-scanned columns.

src/duckdb/src/function/table/version/pragma_version.cpp

@@ -1,5 +1,5 @@
 #ifndef DUCKDB_PATCH_VERSION
-#define DUCKDB_PATCH_VERSION "4-dev4516"
+#define DUCKDB_PATCH_VERSION "4-dev4570"
 #endif
 #ifndef DUCKDB_MINOR_VERSION
 #define DUCKDB_MINOR_VERSION 1
@@ -8,10 +8,10 @@
 #define DUCKDB_MAJOR_VERSION 1
 #endif
 #ifndef DUCKDB_VERSION
-#define DUCKDB_VERSION "v1.1.4-dev4516"
+#define DUCKDB_VERSION "v1.1.4-dev4570"
 #endif
 #ifndef DUCKDB_SOURCE_ID
-#define DUCKDB_SOURCE_ID "2e533ec9df"
+#define DUCKDB_SOURCE_ID "d1740d6cf7"
 #endif
 #include "duckdb/function/table/system_functions.hpp"
 #include "duckdb/main/database.hpp"

src/duckdb/src/include/duckdb/common/enums/metric_type.hpp

@@ -68,6 +68,7 @@ enum class MetricsType : uint8_t {
     OPTIMIZER_EXTENSION,
     OPTIMIZER_MATERIALIZED_CTE,
     OPTIMIZER_SUM_REWRITER,
+    OPTIMIZER_LATE_MATERIALIZATION,
 };
 
 struct MetricsTypeHashFunction {

src/duckdb/src/include/duckdb/common/enums/optimizer_type.hpp

@@ -40,7 +40,8 @@ enum class OptimizerType : uint32_t {
 	JOIN_FILTER_PUSHDOWN,
 	EXTENSION,
 	MATERIALIZED_CTE,
-	SUM_REWRITER
+	SUM_REWRITER,
+	LATE_MATERIALIZATION
 };
 
 string OptimizerTypeToString(OptimizerType type);

src/duckdb/src/include/duckdb/common/helper.hpp

@@ -195,6 +195,12 @@ static inline T AlignValue(T n) {
 	return ((n + (val - 1)) / val) * val;
 }
 
+template<uintptr_t alignment>
+inline data_ptr_t AlignValue(data_ptr_t addr) {
+	static_assert((alignment & (alignment - 1)) == 0, "'alignment' has to be a power of 2");
+	return reinterpret_cast<data_ptr_t>((reinterpret_cast<uintptr_t>(addr) + alignment - 1) & ~(alignment - 1));
+}
+
 template<class T, T val=8>
 constexpr inline T AlignValueFloor(T n) {
 	return (n / val) * val;

src/duckdb/src/include/duckdb/execution/operator/csv_scanner/sniffer/csv_sniffer.hpp

@@ -236,6 +236,8 @@ class CSVSniffer {
 	                     unordered_map<idx_t, vector<LogicalType>> &best_sql_types_candidates_per_column_idx,
 	                     CSVReaderOptions &options, CSVErrorHandler &error_handler);
 	vector<string> names;
+	//! If the file only has a header
+	bool single_row_file = false;
 
 	//! ------------------------------------------------------//
 	//! ------------------ Type Replacement ----------------- //

src/duckdb/src/include/duckdb/function/scalar/strftime_format.hpp

@@ -169,6 +169,7 @@ struct StrpTimeFormat : public StrTimeFormat { // NOLINT: work-around bug in cla
 		return format_specifier != other.format_specifier;
 	}
 	DUCKDB_API static ParseResult Parse(const string &format, const string &text);
+	DUCKDB_API static bool TryParse(const string &format, const string &text, ParseResult &result);
 
 	DUCKDB_API bool Parse(string_t str, ParseResult &result, bool strict = false) const;
 

src/duckdb/src/include/duckdb/function/table/table_scan.hpp

@@ -34,6 +34,13 @@ struct TableScanBindData : public TableFunctionData {
 		auto &other = other_p.Cast<TableScanBindData>();
 		return &other.table == &table;
 	}
+	unique_ptr<FunctionData> Copy() const override {
+		auto bind_data = make_uniq<TableScanBindData>(table);
+		bind_data->is_index_scan = is_index_scan;
+		bind_data->is_create_index = is_create_index;
+		bind_data->column_ids = column_ids;
+		return std::move(bind_data);
+	}
 };
 
 //! The table scan function represents a sequential or index scan over one of DuckDB's base tables.

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

@@ -120,6 +120,9 @@ struct ClientConfig {
 	//! The maximum amount of OR filters we generate dynamically from a hash join
 	idx_t dynamic_or_filter_threshold = 50;
 
+	//! The maximum amount of rows in the LIMIT/SAMPLE for which we trigger late materialization
+	idx_t late_materialization_max_rows = 50;
+
 	//! Whether the "/" division operator defaults to integer division or floating point division
 	bool integer_division = false;
 	//! When a scalar subquery returns multiple rows - return a random row instead of returning an error

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

@@ -56,6 +56,7 @@ class CompressionInfo;
 class EncryptionUtil;
 
 struct CompressionFunctionSet;
+struct DatabaseCacheEntry;
 struct DBConfig;
 
 enum class CheckpointAbort : uint8_t {
@@ -331,6 +332,8 @@ struct DBConfig {
 	vector<unique_ptr<ExtensionCallback>> extension_callbacks;
 	//! Encryption Util for OpenSSL
 	shared_ptr<EncryptionUtil> encryption_util;
+	//! Reference to the database cache entry (if any)
+	shared_ptr<DatabaseCacheEntry> db_cache_entry;
 
 public:
 	DUCKDB_API static DBConfig &GetConfig(ClientContext &context);

src/duckdb/src/include/duckdb/main/database.hpp

@@ -77,7 +77,6 @@ class DatabaseInstance : public enable_shared_from_this<DatabaseInstance> {
 	                                                    const AttachOptions &options);
 
 	void AddExtensionInfo(const string &name, const ExtensionLoadedInfo &info);
-	void SetDatabaseCacheEntry(shared_ptr<DatabaseCacheEntry> entry);
 
 private:
 	void Initialize(const char *path, DBConfig *config);
@@ -95,7 +94,6 @@ class DatabaseInstance : public enable_shared_from_this<DatabaseInstance> {
 	unordered_map<string, ExtensionInfo> loaded_extensions_info;
 	ValidChecker db_validity;
 	unique_ptr<DatabaseFileSystem> db_file_system;
-	shared_ptr<DatabaseCacheEntry> db_cache_entry;
 	shared_ptr<LogManager> log_manager;
 
 	duckdb_ext_api_v1 (*create_api_v1)();

src/duckdb/src/include/duckdb/main/db_instance_cache.hpp

@@ -18,6 +18,7 @@ namespace duckdb {
 class DBInstanceCache;
 
 struct DatabaseCacheEntry {
+	DatabaseCacheEntry();
 	explicit DatabaseCacheEntry(const shared_ptr<DuckDB> &database);
 	~DatabaseCacheEntry();
 

src/duckdb/src/include/duckdb/main/extension_entries.hpp

@@ -645,6 +645,7 @@ static constexpr ExtensionFunctionEntry EXTENSION_FUNCTIONS[] = {
     {"st_zmax", "spatial", CatalogType::SCALAR_FUNCTION_ENTRY},
     {"st_zmflag", "spatial", CatalogType::SCALAR_FUNCTION_ENTRY},
     {"st_zmin", "spatial", CatalogType::SCALAR_FUNCTION_ENTRY},
+    {"start_ui", "motherduck", CatalogType::TABLE_FUNCTION_ENTRY},
     {"starts_with", "core_functions", CatalogType::SCALAR_FUNCTION_ENTRY},
     {"stats", "core_functions", CatalogType::SCALAR_FUNCTION_ENTRY},
     {"stddev", "core_functions", CatalogType::AGGREGATE_FUNCTION_ENTRY},

src/duckdb/src/include/duckdb/main/settings.hpp

@@ -783,6 +783,17 @@ struct IntegerDivisionSetting {
 	static Value GetSetting(const ClientContext &context);
 };
 
+struct LateMaterializationMaxRowsSetting {
+	using RETURN_TYPE = idx_t;
+	static constexpr const char *Name = "late_materialization_max_rows";
+	static constexpr const char *Description =
+	    "The maximum amount of rows in the LIMIT/SAMPLE for which we trigger late materialization";
+	static constexpr const char *InputType = "UBIGINT";
+	static void SetLocal(ClientContext &context, const Value &parameter);
+	static void ResetLocal(ClientContext &context);
+	static Value GetSetting(const ClientContext &context);
+};
+
 struct LockConfigurationSetting {
 	using RETURN_TYPE = bool;
 	static constexpr const char *Name = "lock_configuration";

src/duckdb/src/include/duckdb/optimizer/late_materialization.hpp

@@ -0,0 +1,45 @@
+//===----------------------------------------------------------------------===//
+//                         DuckDB
+//
+// duckdb/optimizer/late_materialization.hpp
+//
+//
+//===----------------------------------------------------------------------===//
+
+#pragma once
+
+#include "duckdb/common/constants.hpp"
+#include "duckdb/optimizer/remove_unused_columns.hpp"
+
+namespace duckdb {
+class LogicalOperator;
+class LogicalGet;
+class Optimizer;
+
+//! Transform
+class LateMaterialization : public BaseColumnPruner {
+public:
+	explicit LateMaterialization(Optimizer &optimizer);
+
+	unique_ptr<LogicalOperator> Optimize(unique_ptr<LogicalOperator> op);
+
+private:
+	bool TryLateMaterialization(unique_ptr<LogicalOperator> &op);
+
+	unique_ptr<LogicalGet> ConstructLHS(LogicalGet &get);
+	ColumnBinding ConstructRHS(unique_ptr<LogicalOperator> &op);
+	idx_t GetOrInsertRowId(LogicalGet &get);
+
+	void ReplaceTopLevelTableIndex(LogicalOperator &op, idx_t new_index);
+	void ReplaceTableReferences(Expression &expr, idx_t new_table_index);
+	unique_ptr<Expression> GetExpression(LogicalOperator &op, idx_t column_index);
+	void ReplaceExpressionReferences(LogicalOperator &next_op, unique_ptr<Expression> &expr);
+	bool OptimizeLargeLimit(LogicalOperator &child);
+
+private:
+	Optimizer &optimizer;
+	//! The max row count for which we will consider late materialization
+	idx_t max_row_count;
+};
+
+} // namespace duckdb