limiter

Provides a generic rate limiter for the web and node.js. Useful for API clients, web crawling, or other tasks that need to be throttled. Two classes are exposed, RateLimiter and TokenBucket. TokenBucket provides a lower level interface to rate limiting with a configurable burst rate and drip rate. RateLimiter sits on top of the token bucket and adds a restriction on the maximum number of tokens that can be removed each interval to comply with common API restrictions such as "150 requests per hour maximum".

Installation

yarn add limiter

Usage

A simple example allowing 150 requests per hour:

1import { RateLimiter } from "limiter";
2
3// Allow 150 requests per hour (the Twitter search limit). Also understands
4// 'second', 'minute', 'day', or a number of milliseconds
5const limiter = new RateLimiter({ tokensPerInterval: 150, interval: "hour" });
6
7async function sendRequest() {
8  // This call will throw if we request more than the maximum number of requests
9  // that were set in the constructor
10  // remainingRequests tells us how many additional requests could be sent
11  // right this moment
12  const remainingRequests = await limiter.removeTokens(1);
13  callMyRequestSendingFunction(...);
14}

Another example allowing one message to be sent every 250ms:

1import { RateLimiter } from "limiter";
2
3const limiter = new RateLimiter({ tokensPerInterval: 1, interval: 250 });
4
5async function sendMessage() {
6  const remainingMessages = await limiter.removeTokens(1);
7  callMyMessageSendingFunction(...);
8}

The default behaviour is to wait for the duration of the rate limiting that's currently in effect before the promise is resolved, but if you pass in "fireImmediately": true, the promise will be resolved immediately with remainingRequests set to -1:

1import { RateLimiter } from "limiter";
2
3const limiter = new RateLimiter({
4  tokensPerInterval: 150,
5  interval: "hour",
6  fireImmediately: true
7});
8
9async function requestHandler(request, response) {
10  // Immediately send 429 header to client when rate limiting is in effect
11  const remainingRequests = await limiter.removeTokens(1);
12  if (remainingRequests < 0) {
13    response.writeHead(429, {'Content-Type': 'text/plain;charset=UTF-8'});
14    response.end('429 Too Many Requests - your IP is being rate limited');
15  } else {
16    callMyMessageSendingFunction(...);
17  }
18}

A synchronous method, tryRemoveTokens(), is available in both RateLimiter and TokenBucket. This will return immediately with a boolean value indicating if the token removal was successful.

1import { RateLimiter } from "limiter";
2
3const limiter = new RateLimiter({ tokensPerInterval: 10, interval: "second" });
4
5if (limiter.tryRemoveTokens(5))
6  console.log('Tokens removed');
7else
8  console.log('No tokens removed');

To get the number of remaining tokens outside the removeTokens promise, simply use the getTokensRemaining method.

1import { RateLimiter } from "limiter";
2
3const limiter = new RateLimiter({ tokensPerInterval: 1, interval: 250 });
4
5// Prints 1 since we did not remove a token and our number of tokens per
6// interval is 1
7console.log(limiter.getTokensRemaining());

Using the token bucket directly to throttle at the byte level:

1import { TokenBucket } from "limiter";
2
3const BURST_RATE = 1024 * 1024 * 150; // 150KB/sec burst rate
4const FILL_RATE = 1024 * 1024 * 50; // 50KB/sec sustained rate
5
6// We could also pass a parent token bucket in to create a hierarchical token
7// bucket
8// bucketSize, tokensPerInterval, interval
9const bucket = new TokenBucket({
10  bucketSize: BURST_RATE,
11  tokensPerInterval: FILL_RATE,
12  interval: "second"
13});
14
15async function handleData(myData) {
16  await bucket.removeTokens(myData.byteLength);
17  sendMyData(myData);
18}

Additional Notes

Both the token bucket and rate limiter should be used with a message queue or some way of preventing multiple simultaneous calls to removeTokens(). Otherwise, earlier messages may get held up for long periods of time if more recent messages are continually draining the token bucket. This can lead to out of order messages or the appearance of "lost" messages under heavy load.

License

MIT License