duckdb 0.7.2-dev3353.0 → 0.7.2-dev3402.0

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

@@ -6,8 +6,8 @@
 #include "duckdb/function/replacement_scan.hpp"
 #include "duckdb/parser/expression/constant_expression.hpp"
 #include "duckdb/parser/expression/function_expression.hpp"
-#include "duckdb/parser/tableref/table_function_ref.hpp"
 #include "duckdb/parser/parsed_data/create_pragma_function_info.hpp"
+#include "duckdb/parser/tableref/table_function_ref.hpp"
 
 namespace duckdb {
 
@@ -115,6 +115,14 @@ unique_ptr<FunctionLocalState> JSONFunctionLocalState::Init(ExpressionState &sta
 	return make_uniq<JSONFunctionLocalState>(state.GetContext());
 }
 
+unique_ptr<FunctionLocalState> JSONFunctionLocalState::InitCastLocalState(CastLocalStateParameters &parameters) {
+	if (parameters.context) {
+		return make_uniq<JSONFunctionLocalState>(*parameters.context);
+	} else {
+		return make_uniq<JSONFunctionLocalState>(Allocator::DefaultAllocator());
+	}
+}
+
 JSONFunctionLocalState &JSONFunctionLocalState::ResetAndGet(ExpressionState &state) {
 	auto &lstate = ExecuteFunctionState::GetFunctionState(state)->Cast<JSONFunctionLocalState>();
 	lstate.json_allocator.Reset();
@@ -197,14 +205,6 @@ unique_ptr<TableRef> JSONFunctions::ReadJSONReplacement(ClientContext &context,
 	return std::move(table_function);
 }
 
-static duckdb::unique_ptr<FunctionLocalState> InitJSONCastLocalState(CastLocalStateParameters &parameters) {
-	if (parameters.context) {
-		return make_uniq<JSONFunctionLocalState>(*parameters.context);
-	} else {
-		return make_uniq<JSONFunctionLocalState>(Allocator::DefaultAllocator());
-	}
-}
-
 static bool CastVarcharToJSON(Vector &source, Vector &result, idx_t count, CastParameters &parameters) {
 	auto &lstate = parameters.local_state->Cast<JSONFunctionLocalState>();
 	lstate.json_allocator.Reset();
@@ -215,15 +215,17 @@ static bool CastVarcharToJSON(Vector &source, Vector &result, idx_t count, CastP
 	    source, result, count, [&](string_t input, ValidityMask &mask, idx_t idx) {
 		    auto data = (char *)(input.GetData());
 		    auto length = input.GetSize();
-		    yyjson_read_err error;
 
+		    yyjson_read_err error;
 		    auto doc = JSONCommon::ReadDocumentUnsafe(data, length, JSONCommon::READ_FLAG, alc, &error);
 
 		    if (!doc) {
-			    HandleCastError::AssignError(JSONCommon::FormatParseError(data, length, error),
-			                                 parameters.error_message);
 			    mask.SetInvalid(idx);
-			    success = false;
+			    if (success) {
+				    HandleCastError::AssignError(JSONCommon::FormatParseError(data, length, error),
+				                                 parameters.error_message);
+				    success = false;
+			    }
 		    }
 		    return input;
 	    });
@@ -231,13 +233,13 @@ static bool CastVarcharToJSON(Vector &source, Vector &result, idx_t count, CastP
 	return success;
 }
 
-void JSONFunctions::RegisterCastFunctions(CastFunctionSet &casts) {
+void JSONFunctions::RegisterSimpleCastFunctions(CastFunctionSet &casts) {
 	// JSON to VARCHAR is basically free
 	casts.RegisterCastFunction(JSONCommon::JSONType(), LogicalType::VARCHAR, DefaultCasts::ReinterpretCast, 1);
 
 	// VARCHAR to JSON requires a parse so it's not free. Let's make it 1 more than a cast to STRUCT
 	auto varchar_to_json_cost = casts.ImplicitCastCost(LogicalType::SQLNULL, LogicalTypeId::STRUCT) + 1;
-	BoundCastInfo info(CastVarcharToJSON, nullptr, InitJSONCastLocalState);
+	BoundCastInfo info(CastVarcharToJSON, nullptr, JSONFunctionLocalState::InitCastLocalState);
 	casts.RegisterCastFunction(LogicalType::VARCHAR, JSONCommon::JSONType(), std::move(info), varchar_to_json_cost);
 
 	// Register NULL to JSON with a different cost than NULL to VARCHAR so the binder can disambiguate functions

package/src/duckdb/extension/json/json_scan.cpp

@@ -1,10 +1,10 @@
 #include "json_scan.hpp"
 
+#include "duckdb/common/multi_file_reader.hpp"
 #include "duckdb/main/database.hpp"
 #include "duckdb/main/extension_helper.hpp"
 #include "duckdb/parallel/task_scheduler.hpp"
 #include "duckdb/storage/buffer_manager.hpp"
-#include "duckdb/common/multi_file_reader.hpp"
 
 namespace duckdb {
 
@@ -59,11 +59,15 @@ unique_ptr<FunctionData> JSONScanData::Bind(ClientContext &context, TableFunctio
 }
 
 void JSONScanData::InitializeFormats() {
+	InitializeFormats(auto_detect);
+}
+
+void JSONScanData::InitializeFormats(bool auto_detect_p) {
 	// Set defaults for date/timestamp formats if we need to
-	if (!auto_detect && date_format.empty()) {
+	if (!auto_detect_p && date_format.empty()) {
 		date_format = "%Y-%m-%d";
 	}
-	if (!auto_detect && timestamp_format.empty()) {
+	if (!auto_detect_p && timestamp_format.empty()) {
 		timestamp_format = "%Y-%m-%dT%H:%M:%S.%fZ";
 	}
 
@@ -75,7 +79,7 @@ void JSONScanData::InitializeFormats() {
 		date_format_map.AddFormat(LogicalTypeId::TIMESTAMP, timestamp_format);
 	}
 
-	if (auto_detect) {
+	if (auto_detect_p) {
 		static const unordered_map<LogicalTypeId, vector<const char *>, LogicalTypeIdHash> FORMAT_TEMPLATES = {
 		    {LogicalTypeId::DATE, {"%m-%d-%Y", "%m-%d-%y", "%d-%m-%Y", "%d-%m-%y", "%Y-%m-%d", "%y-%m-%d"}},
 		    {LogicalTypeId::TIMESTAMP,

package/src/duckdb/extension/parquet/column_reader.cpp

@@ -589,8 +589,8 @@ StringColumnReader::StringColumnReader(ParquetReader &reader, LogicalType type_p
 	}
 }
 
-uint32_t StringColumnReader::VerifyString(const char *str_data, uint32_t str_len) {
-	if (Type() != LogicalTypeId::VARCHAR) {
+uint32_t StringColumnReader::VerifyString(const char *str_data, uint32_t str_len, const bool is_varchar) {
+	if (!is_varchar) {
 		return str_len;
 	}
 	// verify if a string is actually UTF8, and if there are no null bytes in the middle of the string
@@ -605,6 +605,10 @@ uint32_t StringColumnReader::VerifyString(const char *str_data, uint32_t str_len
 	return str_len;
 }
 
+uint32_t StringColumnReader::VerifyString(const char *str_data, uint32_t str_len) {
+	return VerifyString(str_data, str_len, Type() == LogicalTypeId::VARCHAR);
+}
+
 void StringColumnReader::Dictionary(shared_ptr<ResizeableBuffer> data, idx_t num_entries) {
 	dict = std::move(data);
 	dict_strings = duckdb::unique_ptr<string_t[]>(new string_t[num_entries]);

package/src/duckdb/extension/parquet/include/parquet_reader.hpp

@@ -80,15 +80,14 @@ public:
 
 class ParquetReader {
 public:
-	ParquetReader(Allocator &allocator, unique_ptr<FileHandle> file_handle_p);
 	ParquetReader(ClientContext &context, string file_name, ParquetOptions parquet_options);
 	ParquetReader(ClientContext &context, ParquetOptions parquet_options,
 	              shared_ptr<ParquetFileMetadataCache> metadata);
 	~ParquetReader();
 
+	FileSystem &fs;
 	Allocator &allocator;
 	string file_name;
-	FileOpener *file_opener;
 	vector<LogicalType> return_types;
 	vector<string> names;
 	shared_ptr<ParquetFileMetadataCache> metadata;

package/src/duckdb/extension/parquet/include/parquet_writer.hpp

@@ -32,8 +32,8 @@ struct PreparedRowGroup {
 
 class ParquetWriter {
 public:
-	ParquetWriter(FileSystem &fs, string file_name, FileOpener *file_opener, vector<LogicalType> types,
-	              vector<string> names, duckdb_parquet::format::CompressionCodec::type codec);
+	ParquetWriter(FileSystem &fs, string file_name, vector<LogicalType> types, vector<string> names,
+	              duckdb_parquet::format::CompressionCodec::type codec);
 
 public:
 	void PrepareRowGroup(ColumnDataCollection &buffer, PreparedRowGroup &result);

package/src/duckdb/extension/parquet/include/string_column_reader.hpp

@@ -39,6 +39,7 @@ public:
 	void PrepareDeltaByteArray(ResizeableBuffer &buffer) override;
 	void DeltaByteArray(uint8_t *defines, idx_t num_values, parquet_filter_t &filter, idx_t result_offset,
 	                    Vector &result) override;
+	static uint32_t VerifyString(const char *str_data, uint32_t str_len, const bool isVarchar);
 	uint32_t VerifyString(const char *str_data, uint32_t str_len);
 
 protected:

package/src/duckdb/extension/parquet/include/thrift_tools.hpp

@@ -51,8 +51,7 @@ struct ReadHeadComparator {
 // 1: register all ranges that will be read, merging ranges that are consecutive
 // 2: prefetch all registered ranges
 struct ReadAheadBuffer {
-	ReadAheadBuffer(Allocator &allocator, FileHandle &handle, FileOpener &opener)
-	    : allocator(allocator), handle(handle), file_opener(opener) {
+	ReadAheadBuffer(Allocator &allocator, FileHandle &handle) : allocator(allocator), handle(handle) {
 	}
 
 	// The list of read heads
@@ -62,7 +61,6 @@ struct ReadAheadBuffer {
 
 	Allocator &allocator;
 	FileHandle &handle;
-	FileOpener &file_opener;
 
 	idx_t total_size = 0;
 
@@ -124,8 +122,8 @@ class ThriftFileTransport : public duckdb_apache::thrift::transport::TVirtualTra
 public:
 	static constexpr uint64_t PREFETCH_FALLBACK_BUFFERSIZE = 1000000;
 
-	ThriftFileTransport(Allocator &allocator, FileHandle &handle_p, FileOpener &opener, bool prefetch_mode_p)
-	    : handle(handle_p), location(0), allocator(allocator), ra_buffer(ReadAheadBuffer(allocator, handle_p, opener)),
+	ThriftFileTransport(Allocator &allocator, FileHandle &handle_p, bool prefetch_mode_p)
+	    : handle(handle_p), location(0), allocator(allocator), ra_buffer(ReadAheadBuffer(allocator, handle_p)),
 	      prefetch_mode(prefetch_mode_p) {
 	}
 

package/src/duckdb/extension/parquet/parquet-extension.cpp

@@ -239,8 +239,7 @@ public:
 					// missing metadata entry in cache, no usable stats
 					return nullptr;
 				}
-				auto handle = fs.OpenFile(file_name, FileFlags::FILE_FLAGS_READ, FileSystem::DEFAULT_LOCK,
-				                          FileSystem::DEFAULT_COMPRESSION, FileSystem::GetFileOpener(context));
+				auto handle = fs.OpenFile(file_name, FileFlags::FILE_FLAGS_READ);
 				// we need to check if the metadata cache entries are current
 				if (fs.GetLastModifiedTime(*handle) >= metadata->read_time) {
 					// missing or invalid metadata entry in cache, no usable stats overall
@@ -627,8 +626,7 @@ unique_ptr<GlobalFunctionData> ParquetWriteInitializeGlobal(ClientContext &conte
 
 	auto &fs = FileSystem::GetFileSystem(context);
 	global_state->writer =
-	    make_uniq<ParquetWriter>(fs, file_path, FileSystem::GetFileOpener(context), parquet_bind.sql_types,
-	                             parquet_bind.column_names, parquet_bind.codec);
+	    make_uniq<ParquetWriter>(fs, file_path, parquet_bind.sql_types, parquet_bind.column_names, parquet_bind.codec);
 	return std::move(global_state);
 }
 

package/src/duckdb/extension/parquet/parquet_reader.cpp

@@ -49,16 +49,15 @@ using duckdb_parquet::format::Statistics;
 using duckdb_parquet::format::Type;
 
 static duckdb::unique_ptr<duckdb_apache::thrift::protocol::TProtocol>
-CreateThriftProtocol(Allocator &allocator, FileHandle &file_handle, FileOpener &opener, bool prefetch_mode) {
-	auto transport = make_shared<ThriftFileTransport>(allocator, file_handle, opener, prefetch_mode);
+CreateThriftProtocol(Allocator &allocator, FileHandle &file_handle, bool prefetch_mode) {
+	auto transport = make_shared<ThriftFileTransport>(allocator, file_handle, prefetch_mode);
 	return make_uniq<duckdb_apache::thrift::protocol::TCompactProtocolT<ThriftFileTransport>>(std::move(transport));
 }
 
-static shared_ptr<ParquetFileMetadataCache> LoadMetadata(Allocator &allocator, FileHandle &file_handle,
-                                                         FileOpener &opener) {
+static shared_ptr<ParquetFileMetadataCache> LoadMetadata(Allocator &allocator, FileHandle &file_handle) {
 	auto current_time = std::chrono::system_clock::to_time_t(std::chrono::system_clock::now());
 
-	auto proto = CreateThriftProtocol(allocator, file_handle, opener, false);
+	auto proto = CreateThriftProtocol(allocator, file_handle, false);
 	auto &transport = ((ThriftFileTransport &)*proto->getTransport());
 	auto file_size = transport.GetSize();
 	if (file_size < 12) {
@@ -428,20 +427,11 @@ ParquetOptions::ParquetOptions(ClientContext &context) {
 	}
 }
 
-ParquetReader::ParquetReader(Allocator &allocator_p, unique_ptr<FileHandle> file_handle_p) : allocator(allocator_p) {
-	file_name = file_handle_p->path;
-	file_handle = std::move(file_handle_p);
-	metadata = LoadMetadata(allocator, *file_handle, *file_opener);
-	InitializeSchema();
-}
-
 ParquetReader::ParquetReader(ClientContext &context_p, string file_name_p, ParquetOptions parquet_options_p)
-    : allocator(BufferAllocator::Get(context_p)), file_opener(FileSystem::GetFileOpener(context_p)),
+    : fs(FileSystem::GetFileSystem(context_p)), allocator(BufferAllocator::Get(context_p)),
       parquet_options(parquet_options_p) {
-	auto &fs = FileSystem::GetFileSystem(context_p);
 	file_name = std::move(file_name_p);
-	file_handle = fs.OpenFile(file_name, FileFlags::FILE_FLAGS_READ, FileSystem::DEFAULT_LOCK,
-	                          FileSystem::DEFAULT_COMPRESSION, file_opener);
+	file_handle = fs.OpenFile(file_name, FileFlags::FILE_FLAGS_READ);
 	if (!file_handle->CanSeek()) {
 		throw NotImplementedException(
 		    "Reading parquet files from a FIFO stream is not supported and cannot be efficiently supported since "
@@ -451,12 +441,12 @@ ParquetReader::ParquetReader(ClientContext &context_p, string file_name_p, Parqu
 	// or if this file has cached metadata
 	// or if the cached version already expired
 	if (!ObjectCache::ObjectCacheEnabled(context_p)) {
-		metadata = LoadMetadata(allocator, *file_handle, *file_opener);
+		metadata = LoadMetadata(allocator, *file_handle);
 	} else {
 		auto last_modify_time = fs.GetLastModifiedTime(*file_handle);
 		metadata = ObjectCache::GetObjectCache(context_p).Get<ParquetFileMetadataCache>(file_name);
 		if (!metadata || (last_modify_time + 10 >= metadata->read_time)) {
-			metadata = LoadMetadata(allocator, *file_handle, *file_opener);
+			metadata = LoadMetadata(allocator, *file_handle);
 			ObjectCache::GetObjectCache(context_p).Put(file_name, metadata);
 		}
 	}
@@ -466,7 +456,7 @@ ParquetReader::ParquetReader(ClientContext &context_p, string file_name_p, Parqu
 
 ParquetReader::ParquetReader(ClientContext &context_p, ParquetOptions parquet_options_p,
                              shared_ptr<ParquetFileMetadataCache> metadata_p)
-    : allocator(BufferAllocator::Get(context_p)), file_opener(FileSystem::GetFileOpener(context_p)),
+    : fs(FileSystem::GetFileSystem(context_p)), allocator(BufferAllocator::Get(context_p)),
       metadata(std::move(metadata_p)), parquet_options(parquet_options_p) {
 	InitializeSchema();
 }
@@ -634,11 +624,10 @@ void ParquetReader::InitializeScan(ParquetReaderScanState &state, vector<idx_t>
 			state.prefetch_mode = false;
 		}
 
-		state.file_handle = file_handle->file_system.OpenFile(file_handle->path, flags, FileSystem::DEFAULT_LOCK,
-		                                                      FileSystem::DEFAULT_COMPRESSION, file_opener);
+		state.file_handle = fs.OpenFile(file_handle->path, flags);
 	}
 
-	state.thrift_file_proto = CreateThriftProtocol(allocator, *state.file_handle, *file_opener, state.prefetch_mode);
+	state.thrift_file_proto = CreateThriftProtocol(allocator, *state.file_handle, state.prefetch_mode);
 	state.root_reader = CreateReader();
 	state.define_buf.resize(allocator, STANDARD_VECTOR_SIZE);
 	state.repeat_buf.resize(allocator, STANDARD_VECTOR_SIZE);

package/src/duckdb/extension/parquet/parquet_statistics.cpp

@@ -1,6 +1,7 @@
 #include "parquet_statistics.hpp"
 #include "parquet_decimal_utils.hpp"
 #include "parquet_timestamp.hpp"
+#include "string_column_reader.hpp"
 #include "duckdb.hpp"
 #ifndef DUCKDB_AMALGAMATION
 #include "duckdb/common/types/blob.hpp"
@@ -253,15 +254,19 @@ unique_ptr<BaseStatistics> ParquetStatisticsUtils::TransformColumnStatistics(con
 	case LogicalTypeId::VARCHAR: {
 		auto string_stats = StringStats::CreateEmpty(type);
 		if (parquet_stats.__isset.min) {
+			StringColumnReader::VerifyString(parquet_stats.min.c_str(), parquet_stats.min.size(), true);
 			StringStats::Update(string_stats, parquet_stats.min);
 		} else if (parquet_stats.__isset.min_value) {
+			StringColumnReader::VerifyString(parquet_stats.min_value.c_str(), parquet_stats.min_value.size(), true);
 			StringStats::Update(string_stats, parquet_stats.min_value);
 		} else {
 			return nullptr;
 		}
 		if (parquet_stats.__isset.max) {
+			StringColumnReader::VerifyString(parquet_stats.max.c_str(), parquet_stats.max.size(), true);
 			StringStats::Update(string_stats, parquet_stats.max);
 		} else if (parquet_stats.__isset.max_value) {
+			StringColumnReader::VerifyString(parquet_stats.max_value.c_str(), parquet_stats.max_value.size(), true);
 			StringStats::Update(string_stats, parquet_stats.max_value);
 		} else {
 			return nullptr;

package/src/duckdb/extension/parquet/parquet_writer.cpp

@@ -225,12 +225,12 @@ void VerifyUniqueNames(const vector<string> &names) {
 #endif
 }
 
-ParquetWriter::ParquetWriter(FileSystem &fs, string file_name_p, FileOpener *file_opener_p, vector<LogicalType> types_p,
-                             vector<string> names_p, CompressionCodec::type codec)
+ParquetWriter::ParquetWriter(FileSystem &fs, string file_name_p, vector<LogicalType> types_p, vector<string> names_p,
+                             CompressionCodec::type codec)
     : file_name(std::move(file_name_p)), sql_types(std::move(types_p)), column_names(std::move(names_p)), codec(codec) {
 	// initialize the file writer
-	writer = make_uniq<BufferedFileWriter>(
-	    fs, file_name.c_str(), FileFlags::FILE_FLAGS_WRITE | FileFlags::FILE_FLAGS_FILE_CREATE_NEW, file_opener_p);
+	writer = make_uniq<BufferedFileWriter>(fs, file_name.c_str(),
+	                                       FileFlags::FILE_FLAGS_WRITE | FileFlags::FILE_FLAGS_FILE_CREATE_NEW);
 	// parquet files start with the string "PAR1"
 	writer->WriteData((const_data_ptr_t) "PAR1", 4);
 	TCompactProtocolFactoryT<MyTransport> tproto_factory;

package/src/duckdb/src/common/file_system.cpp

@@ -40,11 +40,8 @@ FileSystem::~FileSystem() {
 }
 
 FileSystem &FileSystem::GetFileSystem(ClientContext &context) {
-	return FileSystem::GetFileSystem(*context.db);
-}
-
-FileOpener *FileSystem::GetFileOpener(ClientContext &context) {
-	return ClientData::Get(context).file_opener.get();
+	auto &client_data = ClientData::Get(context);
+	return *client_data.client_file_system;
 }
 
 bool PathMatched(const string &path, const string &sub_path) {
@@ -193,7 +190,7 @@ string FileSystem::ExtractBaseName(const string &path) {
 	return vec[0];
 }
 
-string FileSystem::GetHomeDirectory(FileOpener *opener) {
+string FileSystem::GetHomeDirectory(optional_ptr<FileOpener> opener) {
 	// read the home_directory setting first, if it is set
 	if (opener) {
 		Value result;
@@ -215,7 +212,11 @@ string FileSystem::GetHomeDirectory(FileOpener *opener) {
 	return string();
 }
 
-string FileSystem::ExpandPath(const string &path, FileOpener *opener) {
+string FileSystem::GetHomeDirectory() {
+	return GetHomeDirectory(nullptr);
+}
+
+string FileSystem::ExpandPath(const string &path, optional_ptr<FileOpener> opener) {
 	if (path.empty()) {
 		return path;
 	}
@@ -225,6 +226,10 @@ string FileSystem::ExpandPath(const string &path, FileOpener *opener) {
 	return path;
 }
 
+string FileSystem::ExpandPath(const string &path) {
+	return FileSystem::ExpandPath(path, nullptr);
+}
+
 // LCOV_EXCL_START
 unique_ptr<FileHandle> FileSystem::OpenFile(const string &path, uint8_t flags, FileLockType lock,
                                             FileCompressionType compression, FileOpener *opener) {
@@ -247,14 +252,6 @@ int64_t FileSystem::Write(FileHandle &handle, void *buffer, int64_t nr_bytes) {
 	throw NotImplementedException("%s: Write is not implemented!", GetName());
 }
 
-string FileSystem::GetFileExtension(FileHandle &handle) {
-	auto dot_location = handle.path.rfind('.');
-	if (dot_location != std::string::npos) {
-		return handle.path.substr(dot_location + 1, std::string::npos);
-	}
-	return string();
-}
-
 int64_t FileSystem::GetFileSize(FileHandle &handle) {
 	throw NotImplementedException("%s: GetFileSize is not implemented!", GetName());
 }
@@ -312,10 +309,6 @@ vector<string> FileSystem::Glob(const string &path, FileOpener *opener) {
 	throw NotImplementedException("%s: Glob is not implemented!", GetName());
 }
 
-vector<string> FileSystem::Glob(const string &path, ClientContext &context) {
-	return Glob(path, GetFileOpener(context));
-}
-
 void FileSystem::RegisterSubSystem(unique_ptr<FileSystem> sub_fs) {
 	throw NotImplementedException("%s: Can't register a sub system on a non-virtual file system", GetName());
 }
@@ -337,7 +330,7 @@ bool FileSystem::CanHandleFile(const string &fpath) {
 }
 
 vector<string> FileSystem::GlobFiles(const string &pattern, ClientContext &context, FileGlobOptions options) {
-	auto result = Glob(pattern, context);
+	auto result = Glob(pattern);
 	if (result.empty()) {
 		string required_extension;
 		const string prefixes[] = {"http://", "https://", "s3://"};

package/src/duckdb/src/common/serializer/buffered_file_writer.cpp

@@ -8,9 +8,9 @@ namespace duckdb {
 // Remove this when we switch C++17: https://stackoverflow.com/a/53350948
 constexpr uint8_t BufferedFileWriter::DEFAULT_OPEN_FLAGS;
 
-BufferedFileWriter::BufferedFileWriter(FileSystem &fs, const string &path_p, uint8_t open_flags, FileOpener *opener)
+BufferedFileWriter::BufferedFileWriter(FileSystem &fs, const string &path_p, uint8_t open_flags)
     : fs(fs), path(path_p), data(unique_ptr<data_t[]>(new data_t[FILE_BUFFER_SIZE])), offset(0), total_written(0) {
-	handle = fs.OpenFile(path, open_flags, FileLockType::WRITE_LOCK, FileSystem::DEFAULT_COMPRESSION, opener);
+	handle = fs.OpenFile(path, open_flags, FileLockType::WRITE_LOCK);
 }
 
 int64_t BufferedFileWriter::GetFileSize() {

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

@@ -130,6 +130,9 @@ static void TemplatedGenerateKeys(ArenaAllocator &allocator, Vector &input, idx_
 		auto idx = idata.sel->get_index(i);
 		if (idata.validity.RowIsValid(idx)) {
 			ARTKey::CreateARTKey<T>(allocator, input.GetType(), keys[i], input_data[idx]);
+		} else {
+			// we need to possibly reset the former key value in the keys vector
+			keys[i] = ARTKey();
 		}
 	}
 }
@@ -680,7 +683,6 @@ Node ART::Lookup(Node node, const ARTKey &key, idx_t depth) {
 			}
 			return node;
 		}
-
 		auto &node_prefix = node.GetPrefix(*this);
 		if (node_prefix.count) {
 			for (idx_t pos = 0; pos < node_prefix.count; pos++) {

package/src/duckdb/src/execution/operator/join/physical_index_join.cpp

@@ -167,7 +167,6 @@ void PhysicalIndexJoin::GetRHSMatches(ExecutionContext &context, DataChunk &inpu
 
 	auto &state = state_p.Cast<IndexJoinOperatorState>();
 	auto &art = index.Cast<ART>();
-	;
 
 	// generate the keys for this chunk
 	state.arena_allocator.Reset();

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

@@ -35,7 +35,7 @@ string BaseCSVReader::GetLineNumberStr(idx_t line_error, bool is_line_estimated,
 BaseCSVReader::BaseCSVReader(ClientContext &context_p, BufferedCSVReaderOptions options_p,
                              const vector<LogicalType> &requested_types)
     : context(context_p), fs(FileSystem::GetFileSystem(context)), allocator(Allocator::Get(context)),
-      opener(FileSystem::GetFileOpener(context)), options(std::move(options_p)) {
+      options(std::move(options_p)) {
 }
 
 BaseCSVReader::~BaseCSVReader() {
@@ -43,7 +43,7 @@ BaseCSVReader::~BaseCSVReader() {
 
 unique_ptr<CSVFileHandle> BaseCSVReader::OpenCSV(const BufferedCSVReaderOptions &options_p) {
 	auto file_handle = fs.OpenFile(options_p.file_path.c_str(), FileFlags::FILE_FLAGS_READ, FileLockType::NO_LOCK,
-	                               options_p.compression, this->opener);
+	                               options_p.compression);
 	if (file_handle->CanSeek()) {
 		file_handle->Reset();
 	}

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

@@ -636,10 +636,10 @@ void ParallelCSVReader::ParseCSV(DataChunk &insert_chunk) {
 }
 
 idx_t ParallelCSVReader::GetLineError(idx_t line_error, idx_t buffer_idx) {
-
 	while (true) {
 		if (buffer->line_info->CanItGetLine(file_idx, buffer_idx)) {
 			auto cur_start = verification_positions.beginning_of_first_line + buffer->buffer->GetCSVGlobalStart();
+			// line errors are 1-indexed
 			return buffer->line_info->GetLine(buffer_idx, line_error, file_idx, cur_start, false);
 		}
 	}

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

@@ -207,8 +207,7 @@ unique_ptr<GlobalSinkState> PhysicalCopyToFile::GetGlobalSinkState(ClientContext
 			fs.CreateDirectory(file_path);
 		} else if (!overwrite_or_ignore) {
 			idx_t n_files = 0;
-			fs.ListFiles(
-			    file_path, [&n_files](const string &path, bool) { n_files++; }, FileOpener::Get(context));
+			fs.ListFiles(file_path, [&n_files](const string &path, bool) { n_files++; });
 			if (n_files > 0) {
 				throw IOException("Directory %s is not empty! Enable OVERWRITE_OR_IGNORE option to force writing",
 				                  file_path);

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

@@ -27,10 +27,10 @@ static void WriteCatalogEntries(stringstream &ss, vector<reference<CatalogEntry>
 	ss << std::endl;
 }
 
-static void WriteStringStreamToFile(FileSystem &fs, FileOpener *opener, stringstream &ss, const string &path) {
+static void WriteStringStreamToFile(FileSystem &fs, stringstream &ss, const string &path) {
 	auto ss_string = ss.str();
 	auto handle = fs.OpenFile(path, FileFlags::FILE_FLAGS_WRITE | FileFlags::FILE_FLAGS_FILE_CREATE_NEW,
-	                          FileLockType::WRITE_LOCK, FileSystem::DEFAULT_COMPRESSION, opener);
+	                          FileLockType::WRITE_LOCK);
 	fs.Write(*handle, (void *)ss_string.c_str(), ss_string.size());
 	handle.reset();
 }
@@ -108,7 +108,6 @@ SourceResultType PhysicalExport::GetData(ExecutionContext &context, DataChunk &c
 
 	auto &ccontext = context.client;
 	auto &fs = FileSystem::GetFileSystem(ccontext);
-	auto *opener = FileSystem::GetFileOpener(ccontext);
 
 	// gather all catalog types to export
 	vector<reference<CatalogEntry>> schemas;
@@ -172,7 +171,7 @@ SourceResultType PhysicalExport::GetData(ExecutionContext &context, DataChunk &c
 	WriteCatalogEntries(ss, indexes);
 	WriteCatalogEntries(ss, macros);
 
-	WriteStringStreamToFile(fs, opener, ss, fs.JoinPath(info->file_path, "schema.sql"));
+	WriteStringStreamToFile(fs, ss, fs.JoinPath(info->file_path, "schema.sql"));
 
 	// write the load.sql file
 	// for every table, we write COPY INTO statement with the specified options
@@ -181,7 +180,7 @@ SourceResultType PhysicalExport::GetData(ExecutionContext &context, DataChunk &c
 		auto exported_table_info = exported_tables.data[i].table_data;
 		WriteCopyStatement(fs, load_ss, *info, exported_table_info, function);
 	}
-	WriteStringStreamToFile(fs, opener, load_ss, fs.JoinPath(info->file_path, "load.sql"));
+	WriteStringStreamToFile(fs, load_ss, fs.JoinPath(info->file_path, "load.sql"));
 	state.finished = true;
 
 	return SourceResultType::FINISHED;

package/src/duckdb/src/execution/physical_plan/plan_copy_to_file.cpp

@@ -11,7 +11,7 @@ unique_ptr<PhysicalOperator> PhysicalPlanGenerator::CreatePlan(LogicalCopyToFile
 	bool preserve_insertion_order = PhysicalPlanGenerator::PreserveInsertionOrder(context, *plan);
 	bool supports_batch_index = PhysicalPlanGenerator::UseBatchIndex(context, *plan);
 	auto &fs = FileSystem::GetFileSystem(context);
-	op.file_path = fs.ExpandPath(op.file_path, FileSystem::GetFileOpener(context));
+	op.file_path = fs.ExpandPath(op.file_path);
 	if (op.use_tmp_file) {
 		op.file_path += ".tmp";
 	}