limiter

Rate limiter for Crystal. Memory and Redis based. Redis limiter is shared (unlike Memory limiter which is local for process), so it can be used across multiple processes.

Installation

Add this to your application's shard.yml:

dependencies:
  limiter:
    github: kostya/limiter

Basic Limiter Usage

require "limiter"
limiter = Limiter::Memory.new
limiter.add_limit(2.seconds, 10) # allow 10 requests per 2.seconds
limiter.add_limit(1.hour, 1000)  # allow 1000 requests per 1.hour

res = limiter.request? { some_high_cost_action } # => return value of block or nil
res = limiter.request! { some_high_cost_action } # => return value of block or raise Limiter::Error

Example Memory Limiter

require "limiter"

limiter = Limiter::Memory.new
limiter.add_limit(2.seconds, 10) # allow 10 requests per 2.seconds
limiter.add_limit(1.hour, 1000)  # allow 1000 requests per 1.hour

record Result, val : Float64

def some_high_cost_action : Result
  # ...
  sleep 0.1
  # ...
  return Result.new(rand)
end

res = [] of Result
limited_count = 0

1000.times do
  if val = limiter.request? { some_high_cost_action }
    res << val
  else
    limited_count += 1
  end
end

p res.size
p limited_count

Example Redis Limiter

require "redis"
require "limiter"

redis_client = Redis::PooledClient.new

limiter = Limiter::Redis(Redis::PooledClient).new(redis_client, "my_limiter1")
limiter.add_limit(2.seconds, 10) # allow 10 requests per 2.seconds
limiter.add_limit(1.hour, 1000)  # allow 1000 requests per 1.hour

record Result, val : Float64

def some_high_cost_action : Result
  # ...
  sleep 0.1
  # ...
  return Result.new(rand)
end

res = [] of Result

50.times do
  if val = limiter.request? { some_high_cost_action }
    res << val
  else
    x = limiter.next_usage_after
    puts "processed: #{res.size}, next usage after #{x} seconds"
    sleep(x)
  end
end

puts "processed #{res.size}"