TypeScript icon, indicating that this package has built-in type declarations

4.36.1 • Public • Published

Scramjet Logo

Version 4

Master Build Status Develop Build Status FOSSA Status Known Vulnerabilities DeepScan grade Donate

What does it do?

Scramjet is a fast, simple, functional reactive stream programming framework written on top of node.js object streams. The code is written by chaining functions that transform the streamed data, including well known map, filter and reduce and fully compatible with ES7 async/await. Thanks to it some built in optimizations scramjet is much faster and much much simpler than similar frameworks when using asynchronous operations.

The main advantage of scramjet is running asynchronous operations on your data streams. First of all it allows you to perform the transformations both synchronously and asynchronously by using the same API - so now you can "map" your stream from whatever source and call any number of API's consecutively. And if you're after some heavy maths there's an option of running your stream as multi-threaded!

The benchmarks are published in the scramjet-benchmark repo.


How about a full API to API migration, reading a long list of items from one API and checking them one after another, pushing them to another API? With simultaneous request control? And outputting the log of the conversion? Easy!

const fetch = require("node-fetch");
const get = async (url, options = {}) => (await fetch(url, options)).json;
const { StringStream } = require("scramjet");

StringStream.from(                                 // fetch your API to a scramjet stream
    () => get("https://api.example.org/v1/shows/list")
    .setOptions({maxParallel: 4})                  // set your options
    .lines()                                       // split the stream by line
    .parse(line => {                               // parse strings to data
        const [id, title, url] = line.split(",");
        return { id, title, url };
    .map(async myShow => get({                      // use asynchronous mapping (for example send requests)
        uri: `http://api.local/set/${myShow.id}`,
        body: JSON.stringify(myShow)
    .stringify(resp => `+ Updated "${resp}"`)
    .catch(err => `! Error occured ${err.uri}`)    // handle errors
    .pipe(process.stdout)                          // use any stream

Here you can find a most basic guide on how to execute the above example starting from just having access to some command line: Scramjet from Scratch

Execution and deployment

You can now run stream processing programs with our Scramjet Transform Hub. It will allow you to deploy and execute programs on local and remote environments of your choice and it's as easy as:

npm i -g @scramjet/sth @scramjet/cli
scramjet-transform-hub &
si run <path-to-your-program-dir>

See more info:


Scramjet uses functional programming to run transformations on your data streams in a fashion very similar to the well known event-stream node module. First create a stream from a source:

Use DataStream.from(someThing) to create a new stream from an Array, Generator, AsyncGenerator, Iterator or Readable stream. See the DataStream.from docs for more information, here's a sample.

/* global StringStream, fs */
    .from(fs.createReadStream("./log.txt"))     // get from any readable stream
    .lines()                                 // split the stream by line
    .use("./your-file")                      // use some transforms from another file

Use DataStream.pipeline(readable, transforms) to create a pipeline of transform streams and/or stream modules. Any number of consecutive arguments will get piped one into another.

/* global StringStream, fs, gzip */
    .pipeline(                              // process a number of streams
        gzip.unzip()                        // all errors here will get forwarded
    .lines()                                // split the stream by line
    .use("./your-file")                     // use some transforms from another file

Some methods like from, use, flatMap allow using ES6 generators and ES7 async generators:

const fetch = require("node-fetch");
const { StringStream } = require("scramjet");

        async function* () {                       // construct a stream from an async generator
            yield "houses\n";                      // yield - push a stream chunk
                                                   // yield - push a whole stream
            yield* (await fetch("https://example.org/categories")).body;
        {maxParallel: 4}                           // set your options
    .lines()                                       // split the stream by line
    .flatMap(async function* (category) {
        const req = await fetch(`https://example.org/posts/${category}/`);
        yield* await req.json();                   // yield - push a whole array
    .catch(err => `! Error occured ${err.uri}`)
    .pipe(process.stdout)   // pipe to any output

Most transformations are done by passing a transform function. You can write your function in three ways:

  1. Synchronous

Example: a simple stream transform that outputs a stream of objects of the same id property and the length of the value string.

       (item) => ({id: item.id, length: item.value.length})
  1. Asynchronous using ES2015 async await

Example: A simple stream that uses Fetch API to get all the contents of all entries in the stream

        async (url) => fetch(url).then(res => res.json())
  1. Asynchronous using Promises

Example: A simple stream that fetches an url mentioned in the incoming object

       (item) => new Promise((resolve, reject) => {
           request(item.url, (err, res, data) => {
               if (err)
                   reject(err); // will emit an "error" event on the stream

The actual logic of this transform function is as if you passed your function to the then method of a Promise resolved with the data from the input stream.

  1. Streams with multi-threading

To distribute your code among the processor cores, just use the method distribute:

       16, // number of threads
       (stream) => {
           // multi-threaded code goes here.
           // it MUST return a valid stream back to the main thread.

Writing modules

Scramjet allows writing simple modules that are resolved in the same way as node's require. A module is a simple javascript file that exposes a function taking a stream and any number of following arguments as default export.

Here's an example:

module.exports = (stream, arg1) => {
    const mapper = (chunk) => mapper(chunk, arg1);
    return stream.map(mapper);

Then it can be used with DataStream.use function like this:

myStream.use("./path/to/my-module", "arg1");

If these modules are published you can also simply use myStream.use("published-module").

For more universal modules you can use helper methods createTransformModule and createReadModule that scramjet exports. See more in about this in this blog post Scramjet Modules.

Typescript support

Scramjet aims to be fully documented and expose TypeScript declarations. First version to include definitions in .d.ts folder is 4.15.0. More TypeScript support will be added with next versions, so feel free to report issues in GitHub.

Detailed docs

Here's the list of the exposed classes and methods, please review the specific documentation for details:

Note that:

  • Most of the methods take a Function argument that operates on the stream items.
  • The Function, unless it's stated otherwise, will receive an argument with the next chunk.
  • If you want to perform your operations asynchronously, return a Promise, otherwise just return the right value.


Check out the command line interface for simplified scramjet usage with scramjet-cli

$ sjr -i http://datasource.org/file.csv ./transform-module-1 ./transform-module-1 | gzip > logs.gz

Quick reference of some methods


DataStream is the primary stream type for Scramjet. When you parse your stream, just pipe it you can then perform calculations on the data objects streamed through your flow.

Use as:

const { DataStream } = require('scramjet');

await (DataStream.from(aStream) // create a DataStream
    .map(findInFiles)           // read some data asynchronously
    .map(sendToAPI)             // send the data somewhere
    .run());                    // wait until end

Detailed :DataStream docs here

Most popular methods:


A stream of string objects for further transformation on top of DataStream.


StringStream.from(async () => (await fetch('https://example.com/data/article.txt')).text())

Detailed :StringStream docs here

Most popular methods:


A facilitation stream created for easy splitting or parsing buffers.

Useful for working on built-in Node.js streams from files, parsing binary formats etc.

A simple use case would be:

     .pipe(new BufferStream)         // pipe a buffer stream into scramjet
     .breakup(4)                     // split into 4 byte fragments
     .parse(buffer => [
         buffer.readInt8(0),            // the output is a stream of R,G,B and Alpha
         buffer.readInt8(1),            // values from 0-255 in an array.

Detailed :BufferStream docs here

Most popular methods:


An object consisting of multiple streams than can be refined or muxed.

The idea behind a MultiStream is being able to mux and demux streams when needed.


new MultiStream([...streams])

new MultiStream(function*(){ yield* streams; })
 .map(stream => stream.filter(myFilter))

Detailed :MultiStream docs here

Most popular methods:


Simple scramjet stream that by default contains numbers or other containing with valueOf method. The streams provides simple methods like sum, average. It derives from DataStream so it's still fully supporting all map, reduce etc.

Detailed :NumberStream docs here

Most popular methods:


A stream for moving window calculation with some simple methods.

In essence it's a stream of Array's containing a list of items - a window. It's best used when created by the `DataStream..window`` method.

Detailed :WindowStream docs here

Most popular methods:


StreamWorker class - intended for internal use

This class provides control over the subprocesses, including:

  • spawning
  • communicating
  • delivering streams

Detailed :StreamWorker docs here

Most popular methods:

Scramjet core

Don't like dependencies? Scramjet packs just a couple of those, but if you are really really annoyed by second depth of deps, please try scramjet-core.

Only the most vital methods there, but the library is dependency free.

License and contributions

As of version 2.0 Scramjet is MIT Licensed.

FOSSA Status

Help wanted

The project need's your help! There's lots of work to do - transforming and muxing, joining and splitting, browserifying, modularizing, documenting and issuing those issues.

If you want to help and be part of the Scramjet team, please reach out to us, on slack or email us: opensource@scramjet.org.


Do you like this project? It helped you to reduce time spent on delivering your solution? You are welcome to buy us a coffee ;)

Support us with Github Sponsors


Package Sidebar


npm i @winton/scramjet

Weekly Downloads






Unpacked Size

327 kB

Total Files


Last publish


  • winton