Node CSV

This project provides CSV parsing and has been tested and used on large input files.

  • Follow the Node.js streaming API
  • Async and event based
  • Support delimiters, quotes, escape characters and comments
  • Line breaks discovery: detected in source and reported to destination
  • Data transformation
  • Support for large datasets
  • Complete test coverage as sample and inspiration
  • no external dependencies

Important: this documentation covers the current version (0.2.x) of node-csv-parser. The documentation for the previous version (0.1.0) is available here.


npm install csv

Quick example

This take a string with a comment and convert it to an array:

// node samples/string.js
  {comment: '#'} )
.to.array( function(data){
} );
// [ [ '1', '2', '3', '4' ], [ 'a', 'b', 'c', 'd' ] ]

Advanced example

The following example illustrates 4 usages of the library:

  • Plug a readable stream from a file
  • Direct output to a file path
  • Transform each row
  • Listen to events

// node samples/sample.js
var csv = require('csv');
var fs = require('fs');
.transform( function(row){
  return row;
.on('record', function(row,index){
  console.log('#'+index+' '+JSON.stringify(row));
.on('end', function(count){
  console.log('Number of lines: '+count);
.on('error', function(error){
// #0 ["2000-01-01","20322051544","1979.0","8.8017226E7","ABC","45"]
// #1 ["2050-11-27","28392898392","1974.0","8.8392926E7","DEF","23"]
// Number of lines: 2

Pipe example

The module follow a Stream architecture. At its core, the parser and the stringifier utilities provide a Stream Writer and a Stream Reader implementation available in the CSV API.

+--------+      +----------+----------+       +--------+
|        |      |          |          |       |        |
|        |      |         CSV         |       |        |
|        |      |          |          |       |        |
| Stream |      |  Writer  |  Reader  |       | Stream |
| Reader |.pipe(|   API    |   API    |).pipe(| Writer |)
|        |      |          |          |       |        |
|        |      |          |          |       |        |
+--------+      +----------+----------+       +--------+

Here's a quick example:

in = fs.createReadStream('./in')
out = fs.createWriteStream('./out')


Via npm:

npm install csv

Via git (or downloaded tarball):

git clone


The library extends Node EventEmitter class and emit all the events of the Writable and Readable Stream API. Additionally, the useful "records" event is emitted.

  • record
    Emitted by the stringifier when a new row is parsed and transformed. The data is the value returned by the user transform callback if any. Note however that the event won't be called if transform returns null since the record is skipped. The callback provides two arguments. row is the CSV line being processed (an array or an object) and index is the index number of the line starting at zero
  • data
    Emitted by the stringifier on each line once the data has been transformed and stringified.
  • drain
  • end
    Emitted when the CSV content has been parsed.
  • finish
    Emitted when all data has been flushed to the underlying system. For example, when writting to a file with csv().to.path(), the event will be called once the writing process is complete and the file closed.
  • error
    Thrown whenever an error occured.


The code is organized mainly around 5 main components. The "from" properties provide convenient functions to write CSV text or to plug a Stream Reader. The "parser" takes this CSV content and transform it into an array or an object for each line. The "transformer" provides the ability to work on each line in a synchronous or asynchronous mode. The "stringifier" takes an array or an object and serializes it into CSV text. Finally, the "to" properties provide convenient functions to retrieve the text or to write to plug a Stream Writer.

|       |        |              |              |     |
| from -> parser -> transformer -> stringifier -> to |
|       |        |              |              |     |

Note, even though the "parser", "transformer" and "singifier" are available as properties, you won't have to interact with those.

Stream 1 API

  • pause()
    Implementation of the Readable Stream API, requesting that no further data be sent until resume() is called.

  • resume()
    Implementation of the Readable Stream API, resuming the incoming 'data' events after a pause().

  • write(data, [preserve])
    Implementation of the Writable Stream API with a larger signature. Data may be a string, a buffer, an array or an object.
    If data is a string or a buffer, it could span multiple lines. If data is an object or an array, it must represent a single line.
    Preserve is for line which are not considered as CSV data.

  • end()
    Terminate the parsing. Call this method when no more csv data is to be parsed. It implement the StreamWriter API by setting the writable property to "false" and emitting the end event.

  • transform(callback, [options])
    Register the transformer callback. The callback is a user provided function call on each line to filter, enrich or modify the dataset. More information in the "transforming data" section.

  • error(error)
    Unified mechanism to handle error, emit the error and mark the stream as non readable and non writable.

Fork me on GitHub