Package selection guidance compiled from public package information, community data, and project usage.
Fast CSV is a high-performance CSV parsing library for Dart, designed for efficient reading and writing of CSV files across multiple platforms. It supports standard CSV parsing with good speed and includes experimental event-based streaming for handling large files without excessive memory usage. Ideal for applications needing reliable and fast CSV processing, such as data import/export, analytics, or configuration loading. The package is lightweight and integrates seamlessly into Flutter projects.
Use cases
Data import from CSV files
Exporting data to CSV format
Processing large CSV datasets
Reading configuration files
Parsing spreadsheet-like data
Integrating with backend data sources
Key features
High-speed parsing
Experimental streaming parser
Cross-platform support
Simple API for basic use
Lightweight and efficient
Supports UTF-8 encoding
Suitable for
Flutter apps needing CSV handling
Backend services processing CSV data
Desktop and web applications
Large file processing with low memory
Applications requiring fast data ingestion
Developers prioritizing performance
Considerations
Streaming feature is experimental
Limited advanced formatting options
No built-in CSV generation (only parsing)
Not optimized for very complex CSV structures
Requires manual handling of malformed input
No GUI integration out of the box
Original README
Original README snapshot from the linked source repository. Visit the repository for the latest content.
fast_csv
Classic CSV parsers suitable for most use cases. Pretty fast parsing. With experimental event-based streaming parsing.
Version: 0.2.11
About testing generated parsers
Both parsers have been tested in both operating modes, synchronous and asynchronous modes.
For testing, files from the following source are used: CSV is a data directory which contains examples of CSV files, a flat file format describing values in a table.
https://people.sc.fsu.edu/~jburkardt/data/csv/csv.html
All files are parsed except for the malformed file mlb_players.csv.
It has unclosed quote on row 1036.
Example of the parser usage
An example of a simple way to parse 1000K rows (53M UTF-16 code units) in chunks (by ~0.53M UTF-16 code units) using events without consuming a lot of memory (during parsing) and simultaneously saving this parsed data into a virtual database (by 10K rows per transaction).
import'dart:async';
import'package:fast_csv/csv_converter.dart';
Future<void> main(List<String> args) async {
_exampleParseString();
await_exampleParseStreamWithEvents();
}
const _data = '''
Year,Make,Model,Description,Price
1997,Ford,E350,"ac, abs, moon",3000.00
1999,Chevy,"Venture ""Extended Edition""","",4900.00
1999,Chevy,"Venture ""Extended Edition, Very Large""","",5000.00
1996,Jeep,Grand Cherokee,"MUST SELL!
air, moon roof, loaded",4799.00
''';
Stream<String> _createStream() {
// Create the stream with 1000000 rowsconst count = 1000 * 1000;
final controller = StreamController<String>();
final sink = controller.sink;
const row = '1999,Chevy,"Venture В«Extended EditionВ»","",4900.00';
const rowsInChunk = count ~/ 100;
final chunk = List.generate(rowsInChunk, (i) => row).join('\n');
print('Total data amount ${row.length * count} UTF-16 code units.');
print('The data will arrive in ${chunk.length} UTF-16 code unit chunks.');
var i = 0;
Timer.periodic(Duration.zero, (timer) {
sink.add(chunk);
i += rowsInChunk;
if (i < count) {
sink.add('\n');
}
if (i >= count) {
controller.close();
timer.cancel();
}
});
return controller.stream;
}
Future<void> _exampleParseStreamWithEvents() async {
print('=========================');
print('Start streaming parsing with events');
// Get external datafinal stream = _createStream();
final sw = Stopwatch();
sw.start();
final parser = _MyParser(() async {
print('Saving to virtual database complete in ${sw.elapsed}');
sw.stop();
});
await stream.transform(CsvConverter(parser: parser)).first;
}
void_exampleParseString() {
print('=========================');
print('Parsing string');
final result = CsvConverter().convert(_data);
print(result.join('\n'));
for (var i = 1; i < result.length; i++) {
final row = result[i];
final car = row[1];
final price = num.parse(row[4]);
print('$car$price');
}
}
class_MyParserextendsCsvParser{
finalFuture<void> Function()? onComplete;
int _count = 0;
int _totalCount = 0;
int _transactionCount = 0;
finalList<List<String>> _rows = [];
_MyParser([this.onComplete]);
@overridevoidbeginEvent(CsvParserEvent event) {
if (event == CsvParserEvent.startEvent) {
_count = 0;
_totalCount = 0;
_transactionCount = 0;
_rows.clear();
}
}
@override
R? endEvent<R>(CsvParserEvent event, R? result, bool ok) {
voidsaveRows(bool isLast) {
final rows = _rows.toList();
_rows.clear();
Timer.run(() async {
// Asynchronous saving to the database.await_saveToDatabase(rows, isLast);
});
}
if (ok) {
switch (event) {
caseCsvParserEvent.rowEvent:
final row = result asList<String>;
_rows.add(row);
if (_rows.length > 10000) {
saveRows(false);
}
// Free memory
result = const <String>[] as R;
break;
caseCsvParserEvent.startEvent:
saveRows(true);
if (onComplete != null) {
Timer.run(onComplete!);
}
// Completely freeing memory from the entire list
result = const <List<String>>[] as R;
default:
}
}
return result;
}
Future<void> _saveToDatabase(List<List<String>> rows, bool isLast) async {
_transactionCount++;
_count += rows.length;
_totalCount += rows.length;
if (_count > 100000 || isLast) {
print(
'Saved to virtual database $_totalCount row(s) in $_transactionCount transaction(s)');
_count = 0;
}
if (isLast) {
print(
'Totally saved to virtual database $_totalCount row(s) in $_transactionCount transaction(s)');
}
}
}
Output:
=========================
Parsing string
[Year, Make, Model, Description, Price]
[1997, Ford, E350, ac, abs, moon, 3000.00]
[1999, Chevy, Venture"Extended Edition", , 4900.00]
[1999, Chevy, Venture"Extended Edition, Very Large", , 5000.00]
[1996, Jeep, GrandCherokee, MUST SELL!
air, moon roof, loaded, 4799.00]
Ford3000.0Chevy4900.0Chevy5000.0Jeep4799.0
=========================
Start streaming parsing with events
Total data amount 52000000 UTF-16 code units.
The data will arrive in529999 UTF-16 code unit chunks.
Saved to virtual database 100010row(s) in10transaction(s)
Saved to virtual database 200020row(s) in20transaction(s)
Saved to virtual database 300030row(s) in30transaction(s)
Saved to virtual database 400040row(s) in40transaction(s)
Saved to virtual database 500050row(s) in50transaction(s)
Saved to virtual database 600060row(s) in60transaction(s)
Saved to virtual database 700070row(s) in70transaction(s)
Saved to virtual database 800080row(s) in80transaction(s)
Saved to virtual database 900090row(s) in90transaction(s)
Saved to virtual database 1000000row(s) in100transaction(s)
Totally saved to virtual database 1000000row(s) in100transaction(s)
Saving to virtual database complete in0:00:04.087358
An example of using a configurable parser
This parser is slightly slower than the non-configurable parser.
The difference between using a normal parser and using a configurable parser is that you can specify a field separator.
Any value (such as a space or semicolon).
import'package:fast_csv/csv_ex_converter.dart';
voidmain(List<String> args) {
final parser = CsvExParser(separator: ';');
final result = CsvExConverter(parser: parser).convert(_data);
print(result.join('\n'));
for (var i = 1; i < result.length; i++) {
final row = result[i];
final car = row[1];
final price = num.parse(row[4]);
print('$car$price');
}
}
const _data = '''
Year;Make;Model;Description;Price
1997;Ford;E350;"ac, abs, moon";3000.00
1999;Chevy;"Venture ""Extended Edition""";"";4900.00
1999;Chevy;"Venture ""Extended Edition, Very Large""";"";5000.00
1996;Jeep;Grand Cherokee;"MUST SELL!
air, moon roof, loaded";4799.00
''';
About the implementation of parsers
Parsers are generated from PEG grammars.
Software used to generate parsers
Below is the source code for one of the grammars.