I/O in node is asynchronous, so interacting with the disk and network involves passing callbacks to functions. You might be tempted to write code that serves up a file from disk like this:

var http = require('http');
var fs = require('fs');

var server = http.createServer(function (req, res) {
    fs.readFile(__dirname + '/data.txt', function (err, data) {
        res.end(data);
    });
});
server.listen(8000);

This code works but it's bulky and buffers up the entire data.txt file into memory for every request before writing the result back to clients. If data.txt is very large, your program could start eating a lot of memory as it serves lots of users concurrently, particularly for users on slow connections.

The user experience is poor too because users will need to wait for the whole file to be buffered into memory on your server before they can start receiving any contents.

Luckily both of the (req, res) arguments are streams, which means we can write this in a much better way using fs.createReadStream() instead of fs.readFile():

var http = require('http');
var fs = require('fs');

var server = http.createServer(function (req, res) {
    var stream = fs.createReadStream(__dirname + '/data.txt');
    stream.pipe(res);
});
server.listen(8000);

Here .pipe() takes care of listening for 'data' and 'end' events from the fs.createReadStream(). This code is not only cleaner, but now the data.txt file will be written to clients one chunk at a time immediately as they are received from the disk.

Readable streams can be "piped," or connected, to writable streams. This makes data flow from the source stream to the destination stream without much effort.

var fs = require('fs')

var readable = fs.createReadStream('file1.txt')
var writable = fs.createWriteStream('file2.txt')

readable.pipe(writable) // returns writable

When writable streams are also readable streams, i.e. when they're duplex streams, you can continue piping it to other writable streams.

var zlib = require('zlib')

fs.createReadStream('style.css')
  .pipe(zlib.createGzip()) // The returned object, zlib.Gzip, is a duplex stream.
  .pipe(fs.createWriteStream('style.css.gz')

Readable streams can also be piped into multiple streams.

var readable = fs.createReadStream('source.css')
readable.pipe(zlib.createGzip()).pipe(fs.createWriteStream('output.css.gz'))
readable.pipe(fs.createWriteStream('output.css')

Note that you must pipe to the output streams synchronously (at the same time) before any data 'flows'. Failure to do so might lead to incomplete data being streamed.

Also note that stream objects can emit error events; be sure to responsibly handle these events on every stream, as needed:

var readable = fs.createReadStream('file3.txt')
var writable = fs.createWriteStream('file4.txt')
readable.pipe(writable)
readable.on('error', console.error)
writable.on('error', console.error)

We will see stream objects being returned by modules like fs etc but what if we want to create our own streamable object.

To create Stream object we need to use the stream module provided by NodeJs

   var fs = require("fs");
    var stream = require("stream").Writable;
    
    /* 
     *  Implementing the write function in writable stream class.
     *  This is the function which will be used when other stream is piped into this 
     *  writable stream.
     */
    stream.prototype._write = function(chunk, data){
        console.log(data);
    }
    
    var customStream = new stream();
    
    fs.createReadStream("am1.js").pipe(customStream);

This will give us our own custom writable stream. we can implement anything within the _write function. Above method works in NodeJs 4.x.x version but in NodeJs 6.x ES6 introduced classes therefore syntax have changed. Below is the code for 6.x version of NodeJs

   const Writable = require('stream').Writable;
    
    class MyWritable extends Writable {
      constructor(options) {
        super(options);
      }
    
      _write(chunk, encoding, callback) {
        console.log(chunk);
      }
    }

Lets examine the following two examples for reading a file's contents:

The first one, which uses an async method for reading a file, and providing a callback function which is called once the file is fully read into the memory:

fs.readFile(`${__dirname}/utils.js`, (err, data) => {
  if (err) {
    handleError(err);
  } else {
    console.log(data.toString());
  }
})

And the second, which uses streams in order to read the file's content, piece by piece:

var fileStream = fs.createReadStream(`${__dirname}/file`);
var fileContent = '';
fileStream.on('data', data => {
  fileContent += data.toString();
})

fileStream.on('end', () => {
  console.log(fileContent);
})

fileStream.on('error', err => {
  handleError(err)
})

It's worth mentioning that both examples do the exact same thing. What's the difference then?

• The first one is shorter and looks more elegant
• The second lets you do some processing on the file while it is being read (!)

When the files you deal with are small then there is no real effect when using streams, but what happens when the file is big? (so big that it takes 10 seconds to read it into memory)

Without streams you'll be waiting, doing absolutely nothing (unless your process does other stuff), until the 10 seconds pass and the file is fully read, and only then you can start processing the file.

With streams, you get the file's contents piece by piece, right when they're available - and that lets you process the file while it is being read.

The above example does not illustrate how streams can be utilized for work that cannot be done when going the callback fashion, so lets look at another example:

I would like to download a gzip file, unzip it and save its content to the disk. Given the file's url this is what's need to be done:

• Download the file
• Unzip the file
• Save it to disk

Here's a [small file][1], which is stored in my S3 storage. The following code does the above in the callback fashion.

var startTime = Date.now()
s3.getObject({Bucket: 'some-bucket', Key: 'tweets.gz'}, (err, data) => {
  // here, the whole file was downloaded

  zlib.gunzip(data.Body, (err, data) => {
    // here, the whole file was unzipped

    fs.writeFile(`${__dirname}/tweets.json`, data, err => {
      if (err) console.error(err)

      // here, the whole file was written to disk
      var endTime = Date.now()
      console.log(`${endTime - startTime} milliseconds`) // 1339 milliseconds
    })
  })
})

// 1339 milliseconds

This is how it looks using streams:

s3.getObject({Bucket: 'some-bucket', Key: 'tweets.gz'}).createReadStream()
  .pipe(zlib.createGunzip())
  .pipe(fs.createWriteStream(`${__dirname}/tweets.json`));

// 1204 milliseconds

Yep, it's not faster when dealing with small files - the tested file weights 80KB. Testing this on a bigger file, 71MB gzipped (382MB unzipped), shows that the streams version is much faster

• It took 20925 milliseconds to download 71MB, unzip it and then write 382MB to disk - using the callback fashion.
• In comparison, it took 13434 milliseconds to do the same when using the streams version (35% faster, for a not-so-big file)

Parameters