1.1.7 • Public • Published


Fork Node streams into different destinations based on the input data

Sometimes you want to create complex data flows, that behave differently based on the input data. For example, if you want to split a log file into several files, one for errors and one for everything else. Using simple stream pipes makes it difficult. This minimal library provides the interface for constructing these branched data flows.


var forkable = require( "forkable" );

// split a log file into two destinations: errors and non errors
forkable( fs.createReadStream( "somefile" ) )
    .fork(function ( chunk ) {
        // called for every input data to determine the 
        // destination of that chunk
        var lines = chunk.toString().split( "\n" );
        var errors = lines.filter( function ( l ) {
            return l.indexOf( "ERROR:" ) == 0;
        }).join( "\n" );

        var logs = lines.filter( function ( l ) {
            return l.indexOf( "ERROR:" ) != 0;
        }).join( "\n" );

        return {
            "errors": errors,
            "logs": logs
    .pipe( function ( dest ) {
        return fs.createWriteStream( dest )


forkable([ readable ])

  • readable a Node readable stream, or a standard options object that is passed to Node's PassThrough stream. Optional
  • returns the readable stream itself, augmented with the .fork method

Adds the .fork() method to the provided readable stream, and returns it. This is the simplest way to attach the forkable functionality to any existing stream.

If you omit the readable argument, a PassThrough stream will be created and returned by default. You can also pass in an options object to be passed into the new PassThrough, for example:

fs.createReadStream( "file.log" )
    .pipe( forkable({ highWaterMark: 100 }) )
    .fork( fn );

fork( fn [, options ] )

  • fn a mapping function:
    • data the input chunk from the stream
    • returns a destination name string, or an object of destination names to data chunks
  • options a standard options object that is passed to stream.Transform
  • returns a ForkStream object, which is a subclass of stream.Transform

Creates a fork stream with the provided mapping function, which will be invoked for every input data piped to it from the readable stream. The purpose of this function is to map a destination name for the input data in order to determine to which destination stream it should be piped into.

forkable( stream )
    .fork( function ( data ) {
        return "destination1"; // writes all of the data to "destination1"

The example above will write all of the input data to a single destination. You can also break down the data into separate sub-chunks and write it to multiple destinations:

forkable( stream )
    .fork( function ( data ) {
        return {
            destination1: data[ 0 ],
            destination2: data[ 2 ]

Any data that is not included in the returned destination map is discarded

pipe( fn )

  • fn a dynamic pipe stream function
    • name the destination name returned from .fork()
    • returns the corresponding stream.Writable instance for that name
  • returns the ForkStream instance itself

Unlike to the normal .pipe() method, this augmented method receives a function which will dynamically pipe the data based on the input map from the .fork() function. It will be invoked exactly once for each unique destination name, allowing you to define where the data should be piped into for the given name:

forkable( stream )
    .fork( function ( data ) {
        return data > 100 ? "big" : "small";
    .pipe( function ( name ) { 
        // name is either "big" or "small"
        return fs.createWriteStream( name )

The example above will create two files: "big", with the data that is greater than 100, and "small" for everything else.



Package Sidebar


npm i forkable

Weekly Downloads






Last publish


  • avinoamr