duckdb 0.5.2-dev486.0 → 0.5.2-dev492.0

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "duckdb",
   "main": "./lib/duckdb.js",
-  "version": "0.5.2-dev486.0",
+  "version": "0.5.2-dev492.0",
   "description": "DuckDB node.js API",
   "gypfile": true,
   "dependencies": {

package/src/duckdb.cpp

@@ -76745,6 +76745,7 @@ bool BufferedCSVReader::TryParseComplexCSV(DataChunk &insert_chunk, string &erro
 	bool finished_chunk = false;
 	idx_t column = 0;
 	vector<idx_t> escape_positions;
+	bool has_quotes = false;
 	uint8_t delimiter_pos = 0, escape_pos = 0, quote_pos = 0;
 	idx_t offset = 0;
 
@@ -76807,9 +76808,10 @@ normal:
 	} while (ReadBuffer(start));
 	goto final_state;
 add_value:
-	AddValue(buffer.get() + start, position - start - offset, column, escape_positions);
+	AddValue(buffer.get() + start, position - start - offset, column, escape_positions, has_quotes);
 	// increase position by 1 and move start to the new position
 	offset = 0;
+	has_quotes = false;
 	start = ++position;
 	if (position >= buffer_size && !ReadBuffer(start)) {
 		// file ends right after delimiter, go to final state
@@ -76819,10 +76821,11 @@ add_value:
 add_row : {
 	// check type of newline (\r or \n)
 	bool carriage_return = buffer[position] == '\r';
-	AddValue(buffer.get() + start, position - start - offset, column, escape_positions);
+	AddValue(buffer.get() + start, position - start - offset, column, escape_positions, has_quotes);
 	finished_chunk = AddRow(insert_chunk, column);
 	// increase position by 1 and move start to the new position
 	offset = 0;
+	has_quotes = false;
 	start = ++position;
 	if (position >= buffer_size && !ReadBuffer(start)) {
 		// file ends right after newline, go to final state
@@ -76844,6 +76847,7 @@ in_quotes:
 	// this state parses the remainder of a quoted value
 	quote_pos = 0;
 	escape_pos = 0;
+	has_quotes = true;
 	position++;
 	do {
 		for (; position < buffer_size; position++) {
@@ -76955,7 +76959,7 @@ final_state:
 	}
 	if (column > 0 || position > start) {
 		// remaining values to be added to the chunk
-		AddValue(buffer.get() + start, position - start - offset, column, escape_positions);
+		AddValue(buffer.get() + start, position - start - offset, column, escape_positions, has_quotes);
 		finished_chunk = AddRow(insert_chunk, column);
 	}
 	// final stage, only reached after parsing the file is finished
@@ -76973,6 +76977,7 @@ bool BufferedCSVReader::TryParseSimpleCSV(DataChunk &insert_chunk, string &error
 	bool finished_chunk = false;
 	idx_t column = 0;
 	idx_t offset = 0;
+	bool has_quotes = false;
 	vector<idx_t> escape_positions;
 
 	// read values into the buffer (if any)
@@ -77014,9 +77019,10 @@ normal:
 	// file ends during normal scan: go to end state
 	goto final_state;
 add_value:
-	AddValue(buffer.get() + start, position - start - offset, column, escape_positions);
+	AddValue(buffer.get() + start, position - start - offset, column, escape_positions, has_quotes);
 	// increase position by 1 and move start to the new position
 	offset = 0;
+	has_quotes = false;
 	start = ++position;
 	if (position >= buffer_size && !ReadBuffer(start)) {
 		// file ends right after delimiter, go to final state
@@ -77026,10 +77032,11 @@ add_value:
 add_row : {
 	// check type of newline (\r or \n)
 	bool carriage_return = buffer[position] == '\r';
-	AddValue(buffer.get() + start, position - start - offset, column, escape_positions);
+	AddValue(buffer.get() + start, position - start - offset, column, escape_positions, has_quotes);
 	finished_chunk = AddRow(insert_chunk, column);
 	// increase position by 1 and move start to the new position
 	offset = 0;
+	has_quotes = false;
 	start = ++position;
 	if (position >= buffer_size && !ReadBuffer(start)) {
 		// file ends right after delimiter, go to final state
@@ -77049,6 +77056,7 @@ add_row : {
 in_quotes:
 	/* state: in_quotes */
 	// this state parses the remainder of a quoted value
+	has_quotes = true;
 	position++;
 	do {
 		for (; position < buffer_size; position++) {
@@ -77135,7 +77143,7 @@ final_state:
 
 	if (column > 0 || position > start) {
 		// remaining values to be added to the chunk
-		AddValue(buffer.get() + start, position - start - offset, column, escape_positions);
+		AddValue(buffer.get() + start, position - start - offset, column, escape_positions, has_quotes);
 		finished_chunk = AddRow(insert_chunk, column);
 	}
 
@@ -77235,7 +77243,8 @@ bool BufferedCSVReader::TryParseCSV(ParserMode parser_mode, DataChunk &insert_ch
 	}
 }
 
-void BufferedCSVReader::AddValue(char *str_val, idx_t length, idx_t &column, vector<idx_t> &escape_positions) {
+void BufferedCSVReader::AddValue(char *str_val, idx_t length, idx_t &column, vector<idx_t> &escape_positions,
+                                 bool has_quotes) {
 	if (length == 0 && column == 0) {
 		row_empty = true;
 	} else {
@@ -77266,8 +77275,9 @@ void BufferedCSVReader::AddValue(char *str_val, idx_t length, idx_t &column, vec
 
 	str_val[length] = '\0';
 
-	// test against null string
-	if (!options.force_not_null[column] && strcmp(options.null_str.c_str(), str_val) == 0) {
+	// test against null string, but only if the value was not quoted
+	if ((!has_quotes || sql_types[column].id() != LogicalTypeId::VARCHAR) && !options.force_not_null[column] &&
+	    strcmp(options.null_str.c_str(), str_val) == 0) {
 		FlatVector::SetNull(parse_chunk.data[column], row_entry, true);
 	} else {
 		auto &v = parse_chunk.data[column];

package/src/duckdb.hpp

@@ -11,8 +11,8 @@ THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLI
 #pragma once
 #define DUCKDB_AMALGAMATION 1
 #define DUCKDB_AMALGAMATION_EXTENDED 1
-#define DUCKDB_SOURCE_ID "33b4ebcc3"
-#define DUCKDB_VERSION "v0.5.2-dev486"
+#define DUCKDB_SOURCE_ID "27a7b4d19"
+#define DUCKDB_VERSION "v0.5.2-dev492"
 //===----------------------------------------------------------------------===//
 //                         DuckDB
 //
@@ -27270,7 +27270,7 @@ private:
 	bool TryParseComplexCSV(DataChunk &insert_chunk, string &error_message);
 
 	//! Adds a value to the current row
-	void AddValue(char *str_val, idx_t length, idx_t &column, vector<idx_t> &escape_positions);
+	void AddValue(char *str_val, idx_t length, idx_t &column, vector<idx_t> &escape_positions, bool has_quotes);
 	//! Adds a row to the insert_chunk, returns true if the chunk is filled as a result of this row being added
 	bool AddRow(DataChunk &insert_chunk, idx_t &column);
 	//! Finalizes a chunk, parsing all values that have been added so far and adding them to the insert_chunk