Rate limiting for asyncio
Project description
aiosteady is an MIT licensed library, written in Python, for rate limiting in asyncio applications using Redis and the aioredis library.
aiosteady currently implements the leaky bucket algorithm in a very efficient way.
max_capacity = 10 # The bucket can contain up to 10 drops, starts with 0
drop_recharge = 5.0 # One drop recharges every 5 seconds.
throttler = Throttler(aioredis, max_capacity, drop_recharge)
# consume() returns information about success, the current bucket level,
# how long until the next drop recharges, etc.
res = await throttler.consume(f'user:{user_id}')Installation
To install aiosteady, simply:
$ pip install aiosteadyUsage
The leaky bucket algorithm follows a simple model.
A single bucket contains a number of drops, called the bucket level. Buckets start with zero drops.
Buckets have a maximum capacity of drops.
Each use of the bucket (consumption) inserts one or more drops into the bucket, up until the maximum capacity. If the bucket would overflow, the consumption fails.
One drop leaks out every drop_recharge seconds, freeing space in the bucket for a new drop to be put into it.
In addition to making the consumption fail, full buckets can optionally be configured to block further attempts to consume for a period.
Create an instance of aiosteady.leakybucket.Throttler, giving it an instance of an aioredis client and rate limiting parameters (the maximum bucket capacity, the number of seconds it takes for a drop to leak out, and an optional blocking duration).
A Throttler supports two operations: consuming and peeking.
await Throttler.consume("a_key") (consume because it consumes bucket resources) attempts to put the given number of drops (default 1) into the bucket at the given key. It returns an instance of aiosteady.leakybucket.ThrottleResult, with fields for:
success: a boolean, describing whether the consumption was successful
level: an integer, describing the new level of the bucket
until_next_drop: a float, describing the number of seconds left after the next drop regenerates
blocked_for: an optional float, if blocking is being used and the bucket is blocked, the number of seconds until the block expires
await Throttler.peek("a_key") returns the same ThrottleResult but without attempting to consume any drops.
Both operations are implemented using a single Redis call, using Lua scripting.
Changelog
0.2.0 (2021-04-08)
Use the Redis evalsha instead of eval, for efficiency.
0.1.0 (2021-03-07)
Initial release.
Credits
The Lua Redis script for atomic leaky bucket has been taken and heavily adapted from the Prorate project.
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for aiosteady-0.2.0-py3-none-any.whl
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 | 24187c820bbee9dc8199db35f1f419265582721ff0105eb86d49e41be427bc20 |
|
| MD5 | 87d49bebf10691eef5d0853db91fa55d |
|
| BLAKE2b-256 | 6ae4a6e18bef8422124c5dde87e27e4c2f1a612876ebe175ff50560d67f32d01 |
