linaGirl/request-rate-limiter

Call HTTP APIs that have rate limits and allow request bursts

request-rate-limiter

A simple leaky-bucket based request rate limiter. Mostly used for clients that work against an API that makes use of a leaky bucket for rate limiting incoming requests. Backs off when an API returns the HTTP 429 HTTP «Too Many Requests» status is encountered.

Provides a request module based on the request module by mikael. You may also implement your own request handler module in a very simple way.

ATTENTION: The API of Version 2+ is not backwards compatible with the API of version 1!

ATTENTION: This module makes use of es modules and thus needs to be started using the --experimental-modules flag in node 10 and 12.

API

constructor

The constructor accepts 4 additional options, which can be used to configure the behaviour of the limiter:

• backoffTime: how many seconds to back off when the remote end indicates to back off
• requestRate: how many requests can be sent within the interval
• interval: the interval within which all requests of the requestRate should be executed
• timeout: no request will stay in the queue any longer than the timeout. if the queue is full, the requst will be rejected

import RequestRateLimiter from 'request-rate-limiter';

const limiter = new RequestRateLimiter({
    backoffTime: 10,
    requestRate: 60,
    interval: 60,
    timeout: 600,
});

setRequestHandler

Used to pass a request handler to the limiter

Make use of the request handler provided by this module

import RequestRateLimiter, { RequestsRequestHandler } from 'request-rate-limiter';

const limiter = new RequestRateLimiter();

limiter.setRequestHandler(new RequestsRequestHandler({
    backoffHTTPCode: 429,
}));

Implement your own request implementation

import RequestRateLimiter, { BackoffError } from 'request-rate-limiter';

const limiter = new RequestRateLimiter();



class MyRequestHandler {

    // this method is the only required interface to implement
    // it gets passed the request config that is passed by the 
    // user to the request method of the limiter. The method must
    // return an instance of the BackoffError when the limiter 
    // needs to back off
    async request(requestConfig) {
        const response = sendRequestUsingSomeLibrary(requestConfig);

        if (response.statusCode === 429) throw new BackoffError(`Need to nack off guys!`);
        else return response;
    }
}

limiter.setRequestHandler(new MyRequestHandler());

request

The request method is used to send rate limited requests. You need to pass the configuration of the request to it, which later will be passed to the request handler (see above).

import RequestRateLimiter, { RequestsRequestHandler } from 'request-rate-limiter';

const limiter = new RequestRateLimiter();
limiter.setRequestHandler(new RequestsRequestHandler());


// just send one request
const response = await limiter.request('https://joinbox.com/');


// send requests one after another, waiting for each one to finish 
// before the next one is sent
for (const requestConfig of requests) {
    const response = await limiter.request(requestConfig);
}

// send a buch of requests in parallel
await Promise.all(requests.map(async(requestConfig) => {
    const response = await limiter.request(requestConfig);
}));

idle

The idle method returns a promise which is called when the limiter becomes idle. It's like a once event listener which means that once the promise is resolved one must call the idle method again to wait on the next bunch of requests to complete and the limiter to become idle

import RequestRateLimiter, { RequestsRequestHandler } from 'request-rate-limiter';

const limiter = new RequestRateLimiter();
limiter.setRequestHandler(new RequestsRequestHandler());



// send a buch of requests in parallel
Promise.all(requests.map(async(requestConfig) => {
    const response = await limiter.request(requestConfig);
})).catch((err) => {
    // this happens if the bucket is overflowing
});

// wait until th elimiter becomes idle
await limiter.idle();