#rate-limiting #warp #middleware #web-framework #web #async

warp-rate-limit

Rate limiting middleware for the Warp web framework

1 unstable release

0.1.0 Nov 23, 2024

#1744 in Web programming

Download history 123/week @ 2024-11-18 23/week @ 2024-11-25 14/week @ 2024-12-02 5/week @ 2024-12-09

165 downloads per month

MIT/Apache

15KB
141 lines

Rate-limiting Middleware for Warp

Crates.io Documentation

A rate-limiting middleware for the Warp web framework.

Features:

  • In memory, IP-based rate limiting
  • Configurable time windows and request limits
  • Thread-safe request tracking
  • Idiomatic integration with existing Warp routes

Usage

Add this to your Cargo.toml:

[dependencies]
warp-rate-limit = "0.1"

Then, you'll need to:

  1. Create a rate limiter, and
  2. Add the rate limiter to a route.

Here's an example:

use std::time::Duration;
use warp::{Filter, Reply};
use warp_rate_limit::RateLimit;

#[tokio::main]
async fn main() {
    // STEP 1. Create a rate limiter that allows, at most,
    // 5 requests per every 30 seconds:
    let rate_limit = RateLimit::new()
        .with_window(Duration::from_secs(30))
        .with_max_requests(5);

    // STEP 2. Add the rate limiter to a route:
    let route = warp::path("hello")
        .and(rate_limit)
        .map(|_| "Hello, World!");

    warp::serve(route).run(([127, 0, 0, 1], 3030)).await;
}

Configuration Options

See the documentation.

Limitations

For most basic web applications these limitations are acceptable, but if you need more advanced features, consider using a dedicated rate-limiting solution with persistent storage.

  • IP-based only: Rate limiting is performed based on IP address. This may not be suitable if your application is behind a proxy or if you need to rate limit by other identifiers (e.g., API keys, user IDs).
  • In-memory storage: Rate limit data is stored in memory using a HashMap. This means:
    • Rate limit data is not persisted across server restarts
    • May not be suitable for high-scale deployments with multiple instances
    • Memory usage grows with the number of unique IPs making requests
  • Single window: Only supports a single fixed-time window. Does not support more complex rate limiting strategies like sliding windows or token bucket algorithms.
  • No clustering support: When running multiple instances of your application, each instance maintains its own rate limit counters. This could allow more requests than intended in a distributed setup.
  • No advanced configuration: This is intended to be a simple, small solution for basic rate limiting, and as such, there is no support for per-route rate limiting, burst allowances, customr response headers, or allow/deny list support.

If any of these limitations are a blocker for you, consider augmenting the crate and submitting a PR.

License

MIT license.

Contributing

Contributions are welcome and appreciated. Submit a PR when you're ready.

Dependencies

~9–18MB
~256K SLoC