Highlights
- 3 rate-limiting algorithms: fixed window, sliding window log, token bucket
- 2 backends: in-memory (asyncio.Lock) and Redis (atomic Lua scripts)
- ASGI middleware: drop-in for FastAPI/Starlette with X-RateLimit-* headers, 429 + Retry-After, configurable fail-open/closed
- Benchmarked: ~70k ops/s in-memory, ~27k ops/s distributed Redis
- 55 tests: behavioral, integration (live Redis), concurrency, middleware, edge cases
- CI: ruff + black + mypy + pytest with Redis service container
Install
pip install async-ratelimit
Full usage in README.md. Benchmark methodology in BENCHMARKS.md.