duckdb 0.6.2-dev735.0 → 0.6.2-dev758.0

package/package.json

@@ -2,7 +2,7 @@
   "name": "duckdb",
   "main": "./lib/duckdb.js",
   "types": "./lib/duckdb.d.ts",
-  "version": "0.6.2-dev735.0",
+  "version": "0.6.2-dev758.0",
   "description": "DuckDB node.js API",
   "gypfile": true,
   "dependencies": {

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

@@ -160,7 +160,6 @@ void BaseCSVReader::AddValue(string_t str_val, idx_t &column, vector<idx_t> &esc
 	} else {
 		row_empty = false;
 	}
-
 	if (!sql_types.empty() && column == sql_types.size() && length == 0) {
 		// skip a single trailing delimiter in last column
 		return;
@@ -249,7 +248,7 @@ bool BaseCSVReader::AddRow(DataChunk &insert_chunk, idx_t &column, string &error
 				return false;
 			} else {
 				throw InvalidInputException(
-				    "Error in file \"%s\" on line %s: expected %lld values per row, but got %d. (%s)",
+				    "Error in file \"%s\" on line %s: expected %lld values per row, but got %d.\nParser options:\n%s",
 				    options.file_path, GetLineNumberStr(linenr, linenr_estimated).c_str(), sql_types.size(), column,
 				    options.ToString());
 			}
@@ -309,7 +308,7 @@ void BaseCSVReader::VerifyUTF8(idx_t col_idx, idx_t row_idx, DataChunk &chunk, i
 		int64_t error_line = linenr - (chunk.size() - row_idx) + 1 + offset;
 		D_ASSERT(error_line >= 0);
 		throw InvalidInputException("Error in file \"%s\" at line %llu in column \"%s\": "
-		                            "%s. Parser options: %s",
+		                            "%s. Parser options:\n%s",
 		                            options.file_path, error_line, col_name,
 		                            ErrorManager::InvalidUnicodeError(s.GetString(), "CSV file"), options.ToString());
 	}
@@ -332,29 +331,27 @@ bool BaseCSVReader::Flush(DataChunk &insert_chunk, bool try_add_line) {
 	// convert the columns in the parsed chunk to the types of the table
 	insert_chunk.SetCardinality(parse_chunk);
 	for (idx_t col_idx = 0; col_idx < sql_types.size(); col_idx++) {
-		if (sql_types[col_idx].id() == LogicalTypeId::VARCHAR) {
+		auto insert_idx = insert_cols_idx[col_idx];
+		auto &type = sql_types[col_idx];
+		if (type.id() == LogicalTypeId::VARCHAR) {
 			// target type is varchar: no need to convert
 			// just test that all strings are valid utf-8 strings
 			VerifyUTF8(col_idx);
-			insert_chunk.data[insert_cols_idx[col_idx]].Reference(parse_chunk.data[col_idx]);
+			insert_chunk.data[insert_idx].Reference(parse_chunk.data[col_idx]);
 		} else {
 			string error_message;
 			bool success;
-			if (options.has_format[LogicalTypeId::DATE] && sql_types[col_idx].id() == LogicalTypeId::DATE) {
+			if (options.has_format[LogicalTypeId::DATE] && type.id() == LogicalTypeId::DATE) {
 				// use the date format to cast the chunk
-				success =
-				    TryCastDateVector(options, parse_chunk.data[col_idx], insert_chunk.data[insert_cols_idx[col_idx]],
-				                      parse_chunk.size(), error_message);
-			} else if (options.has_format[LogicalTypeId::TIMESTAMP] &&
-			           sql_types[col_idx].id() == LogicalTypeId::TIMESTAMP) {
+				success = TryCastDateVector(options, parse_chunk.data[col_idx], insert_chunk.data[insert_idx],
+				                            parse_chunk.size(), error_message);
+			} else if (options.has_format[LogicalTypeId::TIMESTAMP] && type.id() == LogicalTypeId::TIMESTAMP) {
 				// use the date format to cast the chunk
-				success = TryCastTimestampVector(options, parse_chunk.data[col_idx],
-				                                 insert_chunk.data[insert_cols_idx[col_idx]], parse_chunk.size(),
-				                                 error_message);
+				success = TryCastTimestampVector(options, parse_chunk.data[col_idx], insert_chunk.data[insert_idx],
+				                                 parse_chunk.size(), error_message);
 			} else {
 				// target type is not varchar: perform a cast
-				success = VectorOperations::DefaultTryCast(parse_chunk.data[col_idx],
-				                                           insert_chunk.data[insert_cols_idx[col_idx]],
+				success = VectorOperations::DefaultTryCast(parse_chunk.data[col_idx], insert_chunk.data[insert_idx],
 				                                           parse_chunk.size(), &error_message);
 			}
 			if (success) {
@@ -385,13 +382,13 @@ bool BaseCSVReader::Flush(DataChunk &insert_chunk, bool try_add_line) {
 			auto error_line = linenr - (parse_chunk.size() - row_idx) + 1;
 
 			if (options.auto_detect) {
-				throw InvalidInputException("%s in column %s, at line %llu. Parser "
-				                            "options: %s. Consider either increasing the sample size "
+				throw InvalidInputException("%s in column %s, at line %llu.\n\nParser "
+				                            "options:\n%s.\n\nConsider either increasing the sample size "
 				                            "(SAMPLE_SIZE=X [X rows] or SAMPLE_SIZE=-1 [all rows]), "
 				                            "or skipping column conversion (ALL_VARCHAR=1)",
 				                            error_message, col_name, error_line, options.ToString());
 			} else {
-				throw InvalidInputException("%s at line %llu in column %s. Parser options: %s ", error_message,
+				throw InvalidInputException("%s at line %llu in column %s. Parser options:\n%s ", error_message,
 				                            error_line, col_name, options.ToString());
 			}
 		}

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

@@ -851,10 +851,13 @@ vector<LogicalType> BufferedCSVReader::SniffCSV(const vector<LogicalType> &reque
 	// #######
 	// type candidates, ordered by descending specificity (~ from high to low)
 	vector<LogicalType> type_candidates = {
-	    LogicalType::VARCHAR, LogicalType::TIMESTAMP,
-	    LogicalType::DATE,    LogicalType::TIME,
-	    LogicalType::DOUBLE,  /* LogicalType::FLOAT,*/ LogicalType::BIGINT,
-	    LogicalType::INTEGER, /*LogicalType::SMALLINT, LogicalType::TINYINT,*/ LogicalType::BOOLEAN,
+	    LogicalType::VARCHAR,
+	    LogicalType::TIMESTAMP,
+	    LogicalType::DATE,
+	    LogicalType::TIME,
+	    LogicalType::DOUBLE,
+	    /* LogicalType::FLOAT,*/ LogicalType::BIGINT,
+	    /*LogicalType::INTEGER,*/ /*LogicalType::SMALLINT, LogicalType::TINYINT,*/ LogicalType::BOOLEAN,
 	    LogicalType::SQLNULL};
 	// format template candidates, ordered by descending specificity (~ from high to low)
 	std::map<LogicalTypeId, vector<const char *>> format_template_candidates = {

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

@@ -251,13 +251,14 @@ bool BufferedCSVReaderOptions::SetBaseOption(const string &loption, const Value
 }
 
 std::string BufferedCSVReaderOptions::ToString() const {
-	return "DELIMITER='" + delimiter + (has_delimiter ? "'" : (auto_detect ? "' (auto detected)" : "' (default)")) +
-	       ", QUOTE='" + quote + (has_quote ? "'" : (auto_detect ? "' (auto detected)" : "' (default)")) +
-	       ", ESCAPE='" + escape + (has_escape ? "'" : (auto_detect ? "' (auto detected)" : "' (default)")) +
-	       ", HEADER=" + std::to_string(header) +
+	return "  file=" + file_path + "\n  delimiter='" + delimiter +
+	       (has_delimiter ? "'" : (auto_detect ? "' (auto detected)" : "' (default)")) + "\n  quote='" + quote +
+	       (has_quote ? "'" : (auto_detect ? "' (auto detected)" : "' (default)")) + "\n  escape='" + escape +
+	       (has_escape ? "'" : (auto_detect ? "' (auto detected)" : "' (default)")) +
+	       "\n  header=" + std::to_string(header) +
 	       (has_header ? "" : (auto_detect ? " (auto detected)" : "' (default)")) +
-	       ", SAMPLE_SIZE=" + std::to_string(sample_chunk_size * sample_chunks) +
-	       ", IGNORE_ERRORS=" + std::to_string(ignore_errors) + ", ALL_VARCHAR=" + std::to_string(all_varchar);
+	       "\n  sample_size=" + std::to_string(sample_chunk_size * sample_chunks) +
+	       "\n  ignore_erros=" + std::to_string(ignore_errors) + "\n  all_varchar=" + std::to_string(all_varchar);
 }
 
 } // namespace duckdb

package/src/duckdb/src/function/table/read_csv.cpp

@@ -271,6 +271,24 @@ public:
 	atomic<idx_t> bytes_read;
 	//! Size of current file
 	idx_t file_size;
+	//! The index of the next file to read (i.e. current file + 1)
+	idx_t file_index = 1;
+
+	double GetProgress(ReadCSVData &bind_data) const {
+		idx_t total_files = bind_data.files.size();
+
+		// get the progress WITHIN the current file
+		double progress;
+		if (file_size == 0) {
+			progress = 1.0;
+		} else {
+			progress = double(bytes_read) / double(file_size);
+		}
+		// now get the total percentage of files read
+		double percentage = double(file_index) / total_files;
+		percentage += (double(1) / double(total_files)) * progress;
+		return percentage * 100;
+	}
 
 private:
 	//! File Handle for current file
@@ -278,8 +296,6 @@ private:
 
 	shared_ptr<CSVBuffer> current_buffer;
 	shared_ptr<CSVBuffer> next_buffer;
-	//! The index of the next file to read (i.e. current file + 1)
-	idx_t file_index = 1;
 
 	//! Mutex to lock when getting next batch of bytes (Parallel Only)
 	mutex main_mutex;
@@ -348,6 +364,7 @@ unique_ptr<CSVBufferRead> ParallelCSVGlobalState::Next(ClientContext &context, R
 	}
 	return result;
 }
+
 static unique_ptr<GlobalTableFunctionState> ParallelCSVInitGlobal(ClientContext &context,
                                                                   TableFunctionInitInput &input) {
 	auto &bind_data = (ReadCSVData &)*input.bind_data;
@@ -359,7 +376,6 @@ static unique_ptr<GlobalTableFunctionState> ParallelCSVInitGlobal(ClientContext
 
 	bind_data.options.file_path = bind_data.files[0];
 	file_handle = ReadCSV::OpenCSV(bind_data.options, context);
-
 	idx_t rows_to_skip = bind_data.options.skip_rows + (bind_data.options.has_header ? 1 : 0);
 	return make_unique<ParallelCSVGlobalState>(context, move(file_handle), bind_data.files,
 	                                           context.db->NumberOfThreads(), bind_data.options.buffer_size,
@@ -379,12 +395,9 @@ public:
 	CSVBufferRead previous_buffer;
 };
 
-unique_ptr<LocalTableFunctionState> ReadCSVInitLocal(ExecutionContext &context, TableFunctionInitInput &input,
-                                                     GlobalTableFunctionState *global_state_p) {
+unique_ptr<LocalTableFunctionState> ParallelReadCSVInitLocal(ExecutionContext &context, TableFunctionInitInput &input,
+                                                             GlobalTableFunctionState *global_state_p) {
 	auto &csv_data = (ReadCSVData &)*input.bind_data;
-	if (csv_data.single_threaded) {
-		return nullptr;
-	}
 	auto &global_state = (ParallelCSVGlobalState &)*global_state_p;
 	auto next_local_buffer = global_state.Next(context.client, csv_data);
 	unique_ptr<ParallelCSVReader> csv_reader;
@@ -416,7 +429,6 @@ static void ParallelReadCSVFunction(ClientContext &context, TableFunctionInput &
 			if (!next_chunk) {
 				break;
 			}
-			//			csv_local_state.previous_buffer = csv_local_state.csv_reader->buffer;
 			csv_local_state.csv_reader->SetBufferRead(move(next_chunk));
 		}
 		csv_local_state.csv_reader->ParseCSV(output);
@@ -434,91 +446,172 @@ static void ParallelReadCSVFunction(ClientContext &context, TableFunctionInput &
 	}
 }
 
-static idx_t CSVReaderGetBatchIndex(ClientContext &context, const FunctionData *bind_data_p,
-                                    LocalTableFunctionState *local_state, GlobalTableFunctionState *global_state) {
-	auto &bind_data = (ReadCSVData &)*bind_data_p;
-	if (bind_data.single_threaded) {
-		return 0;
-	}
-	auto &data = (ParallelCSVLocalState &)*local_state;
-	return data.csv_reader->buffer->batch_index;
-}
-
 //===--------------------------------------------------------------------===//
 // Single-Threaded CSV Reader
 //===--------------------------------------------------------------------===//
 struct SingleThreadedCSVState : public GlobalTableFunctionState {
-	//! The CSV reader
-	unique_ptr<BufferedCSVReader> csv_reader;
+	explicit SingleThreadedCSVState(idx_t total_files) : total_files(total_files), next_file(0), progress_in_files(0) {
+	}
+
+	mutex csv_lock;
+	unique_ptr<BufferedCSVReader> initial_reader;
+	//! The total number of files to read from
+	idx_t total_files;
 	//! The index of the next file to read (i.e. current file + 1)
-	idx_t file_index;
-	//! Total File Size
-	idx_t file_size;
-	//! How many bytes were read up to this point
-	atomic<idx_t> bytes_read;
+	atomic<idx_t> next_file;
+	//! How far along we are in reading the current set of open files
+	//! This goes from [0...next_file] * 100
+	atomic<idx_t> progress_in_files;
+	//! The set of SQL types
+	vector<LogicalType> sql_types;
 
 	idx_t MaxThreads() const override {
-		return 1;
+		return total_files;
+	}
+
+	double GetProgress(ReadCSVData &bind_data) const {
+		D_ASSERT(total_files == bind_data.files.size());
+		D_ASSERT(progress_in_files <= total_files * 100);
+		return (double(progress_in_files) / double(total_files));
+	}
+
+	unique_ptr<BufferedCSVReader> GetCSVReader(ClientContext &context, ReadCSVData &bind_data, idx_t &file_index,
+	                                           idx_t &total_size) {
+		BufferedCSVReaderOptions options;
+		{
+			lock_guard<mutex> l(csv_lock);
+			if (initial_reader) {
+				return move(initial_reader);
+			}
+			if (next_file >= total_files) {
+				return nullptr;
+			}
+			options = bind_data.options;
+			file_index = next_file;
+			next_file++;
+		}
+		// reuse csv_readers was created during binding
+		unique_ptr<BufferedCSVReader> result;
+		if (options.union_by_name) {
+			result = move(bind_data.union_readers[file_index]);
+		} else {
+			options.file_path = bind_data.files[file_index];
+			result = make_unique<BufferedCSVReader>(context, move(options), sql_types);
+		}
+		total_size = result->file_handle->FileSize();
+		return result;
+	}
+};
+
+struct SingleThreadedCSVLocalState : public LocalTableFunctionState {
+public:
+	explicit SingleThreadedCSVLocalState() : bytes_read(0), total_size(0), current_progress(0), file_index(0) {
 	}
+
+	//! The CSV reader
+	unique_ptr<BufferedCSVReader> csv_reader;
+	//! The current amount of bytes read by this reader
+	idx_t bytes_read;
+	//! The total amount of bytes in the file
+	idx_t total_size;
+	//! The current progress from 0..100
+	idx_t current_progress;
+	//! The file index of this reader
+	idx_t file_index;
 };
 
 static unique_ptr<GlobalTableFunctionState> SingleThreadedCSVInit(ClientContext &context,
                                                                   TableFunctionInitInput &input) {
 	auto &bind_data = (ReadCSVData &)*input.bind_data;
-	auto result = make_unique<SingleThreadedCSVState>();
+	auto result = make_unique<SingleThreadedCSVState>(bind_data.files.size());
 	if (bind_data.initial_reader) {
-		result->csv_reader = move(bind_data.initial_reader);
+		result->initial_reader = move(bind_data.initial_reader);
 	} else if (bind_data.files.empty()) {
 		// This can happen when a filename based filter pushdown has eliminated all possible files for this scan.
 		return move(result);
 	} else {
 		bind_data.options.file_path = bind_data.files[0];
-		result->csv_reader = make_unique<BufferedCSVReader>(context, bind_data.options, bind_data.sql_types);
+		result->initial_reader = make_unique<BufferedCSVReader>(context, bind_data.options, bind_data.sql_types);
+		if (bind_data.options.auto_detect) {
+			bind_data.options = result->initial_reader->options;
+		}
+	}
+	if (!bind_data.options.union_by_name) {
+		// if we are reading multiple files - run auto-detect only on the first file
+		// UNLESS union by name is turned on - in that case we assume that different files have different schemas
+		// as such, we need to re-run the auto detection on each file
+		bind_data.options.auto_detect = false;
+	}
+	result->next_file = 1;
+	if (result->initial_reader) {
+		result->sql_types = result->initial_reader->sql_types;
 	}
-	result->file_size = result->csv_reader->file_handle->FileSize();
-	result->file_index = 1;
+	return move(result);
+}
+
+unique_ptr<LocalTableFunctionState> SingleThreadedReadCSVInitLocal(ExecutionContext &context,
+                                                                   TableFunctionInitInput &input,
+                                                                   GlobalTableFunctionState *global_state_p) {
+	auto &bind_data = (ReadCSVData &)*input.bind_data;
+	auto &data = (SingleThreadedCSVState &)*global_state_p;
+	auto result = make_unique<SingleThreadedCSVLocalState>();
+	result->csv_reader = data.GetCSVReader(context.client, bind_data, result->file_index, result->total_size);
 	return move(result);
 }
 
 static void SingleThreadedCSVFunction(ClientContext &context, TableFunctionInput &data_p, DataChunk &output) {
 	auto &bind_data = (ReadCSVData &)*data_p.bind_data;
 	auto &data = (SingleThreadedCSVState &)*data_p.global_state;
-
-	if (!data.csv_reader) {
+	auto &lstate = (SingleThreadedCSVLocalState &)*data_p.local_state;
+	if (!lstate.csv_reader) {
 		// no csv_reader was set, this can happen when a filename-based filter has filtered out all possible files
 		return;
 	}
 
 	do {
-		data.csv_reader->ParseCSV(output);
-		data.bytes_read = data.csv_reader->bytes_in_chunk;
-		if (output.size() == 0 && data.file_index < bind_data.files.size()) {
-			// exhausted this file, but we have more files we can read
-			// open the next file and increment the counter
-			bind_data.options.file_path = bind_data.files[data.file_index];
-			// reuse csv_readers was created during binding
-			if (bind_data.options.union_by_name) {
-				data.csv_reader = move(bind_data.union_readers[data.file_index]);
-			} else {
-				data.csv_reader =
-				    make_unique<BufferedCSVReader>(context, bind_data.options, data.csv_reader->sql_types);
+		lstate.csv_reader->ParseCSV(output);
+		// update the number of bytes read
+		D_ASSERT(lstate.bytes_read <= lstate.csv_reader->bytes_in_chunk);
+		auto bytes_read = MinValue<idx_t>(lstate.total_size, lstate.csv_reader->bytes_in_chunk);
+		auto current_progress = lstate.total_size == 0 ? 100 : 100 * bytes_read / lstate.total_size;
+		if (current_progress > lstate.current_progress) {
+			if (current_progress > 100) {
+				throw InternalException("Progress should never exceed 100");
+			}
+			data.progress_in_files += current_progress - lstate.current_progress;
+			lstate.current_progress = current_progress;
+		}
+		if (output.size() == 0) {
+			// exhausted this file, but we might have more files we can read
+			auto csv_reader = data.GetCSVReader(context, bind_data, lstate.file_index, lstate.total_size);
+			// add any left-over progress for this file to the progress bar
+			if (lstate.current_progress < 100) {
+				data.progress_in_files += 100 - lstate.current_progress;
+			}
+			// reset the current progress
+			lstate.current_progress = 0;
+			lstate.bytes_read = 0;
+			lstate.csv_reader = move(csv_reader);
+			if (!lstate.csv_reader) {
+				// no more files - we are done
+				return;
 			}
-			data.file_index++;
+			lstate.bytes_read = 0;
 		} else {
 			break;
 		}
 	} while (true);
 
 	if (bind_data.options.union_by_name) {
-		data.csv_reader->SetNullUnionCols(output);
+		lstate.csv_reader->SetNullUnionCols(output);
 	}
 	if (bind_data.options.include_file_name) {
 		auto &col = output.data[bind_data.filename_col_idx];
-		col.SetValue(0, Value(data.csv_reader->options.file_path));
+		col.SetValue(0, Value(lstate.csv_reader->options.file_path));
 		col.SetVectorType(VectorType::CONSTANT_VECTOR);
 	}
 	if (bind_data.options.include_parsed_hive_partitions) {
-		auto partitions = HivePartitioning::Parse(data.csv_reader->options.file_path);
+		auto partitions = HivePartitioning::Parse(lstate.csv_reader->options.file_path);
 
 		idx_t i = bind_data.hive_partition_col_idx;
 
@@ -531,7 +624,7 @@ static void SingleThreadedCSVFunction(ClientContext &context, TableFunctionInput
 		for (auto &part : partitions) {
 			if (bind_data.options.names[i] != part.first) {
 				throw IOException("Hive partition names mismatch, expected '" + bind_data.options.names[i] +
-				                  "' but found '" + part.first + "' for file '" + data.csv_reader->options.file_path +
+				                  "' but found '" + part.first + "' for file '" + lstate.csv_reader->options.file_path +
 				                  "'");
 			}
 			auto &col = output.data[i++];
@@ -553,6 +646,16 @@ static unique_ptr<GlobalTableFunctionState> ReadCSVInitGlobal(ClientContext &con
 	}
 }
 
+unique_ptr<LocalTableFunctionState> ReadCSVInitLocal(ExecutionContext &context, TableFunctionInitInput &input,
+                                                     GlobalTableFunctionState *global_state_p) {
+	auto &csv_data = (ReadCSVData &)*input.bind_data;
+	if (csv_data.single_threaded) {
+		return SingleThreadedReadCSVInitLocal(context, input, global_state_p);
+	} else {
+		return ParallelReadCSVInitLocal(context, input, global_state_p);
+	}
+}
+
 static void ReadCSVFunction(ClientContext &context, TableFunctionInput &data_p, DataChunk &output) {
 	auto &bind_data = (ReadCSVData &)*data_p.bind_data;
 	if (bind_data.single_threaded) {
@@ -562,6 +665,17 @@ static void ReadCSVFunction(ClientContext &context, TableFunctionInput &data_p,
 	}
 }
 
+static idx_t CSVReaderGetBatchIndex(ClientContext &context, const FunctionData *bind_data_p,
+                                    LocalTableFunctionState *local_state, GlobalTableFunctionState *global_state) {
+	auto &bind_data = (ReadCSVData &)*bind_data_p;
+	if (bind_data.single_threaded) {
+		auto &data = (SingleThreadedCSVLocalState &)*local_state;
+		return data.file_index;
+	}
+	auto &data = (ParallelCSVLocalState &)*local_state;
+	return data.csv_reader->buffer->batch_index;
+}
+
 static void ReadCSVAddNamedParameters(TableFunction &table_function) {
 	table_function.named_parameters["sep"] = LogicalType::VARCHAR;
 	table_function.named_parameters["delim"] = LogicalType::VARCHAR;
@@ -592,21 +706,13 @@ static void ReadCSVAddNamedParameters(TableFunction &table_function) {
 double CSVReaderProgress(ClientContext &context, const FunctionData *bind_data_p,
                          const GlobalTableFunctionState *global_state) {
 	auto &bind_data = (ReadCSVData &)*bind_data_p;
-	idx_t file_size, bytes_read;
 	if (bind_data.single_threaded) {
-		auto &data = (const SingleThreadedCSVState &)*global_state;
-		file_size = data.file_size;
-		bytes_read = data.bytes_read;
+		auto &data = (SingleThreadedCSVState &)*global_state;
+		return data.GetProgress(bind_data);
 	} else {
 		auto &data = (const ParallelCSVGlobalState &)*global_state;
-		file_size = data.file_size;
-		bytes_read = data.bytes_read;
-	}
-	if (file_size == 0) {
-		return 100;
+		return data.GetProgress(bind_data);
 	}
-	auto percentage = (bytes_read * 100.0) / file_size;
-	return percentage;
 }
 
 void CSVComplexFilterPushdown(ClientContext &context, LogicalGet &get, FunctionData *bind_data_p,

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

@@ -1,8 +1,8 @@
 #ifndef DUCKDB_VERSION
-#define DUCKDB_VERSION "0.6.2-dev735"
+#define DUCKDB_VERSION "0.6.2-dev758"
 #endif
 #ifndef DUCKDB_SOURCE_ID
-#define DUCKDB_SOURCE_ID "b85fb31ebf"
+#define DUCKDB_SOURCE_ID "cd29769dcd"
 #endif
 #include "duckdb/function/table/system_functions.hpp"
 #include "duckdb/main/database.hpp"

package/src/duckdb/src/include/duckdb/common/http_stats.hpp

@@ -41,6 +41,11 @@ public:
 		}
 		return nullptr;
 	}
+
+	bool IsEmpty() {
+		return head_count == 0 && get_count == 0 && put_count == 0 && post_count == 0 && total_bytes_received == 0 &&
+		       total_bytes_sent == 0;
+	}
 };
 
 } // namespace duckdb

package/src/duckdb/src/main/query_profiler.cpp

@@ -377,7 +377,7 @@ void QueryProfiler::QueryTreeToStream(std::ostream &ss) const {
 		return;
 	}
 
-	if (context.client_data->http_stats) {
+	if (context.client_data->http_stats && !context.client_data->http_stats->IsEmpty()) {
 		string read =
 		    "in: " + StringUtil::BytesToHumanReadableString(context.client_data->http_stats->total_bytes_received);
 		string written =