Skip to content

hachreak/limitless

Repository files navigation

limitless

Build Status

Limitless is a lightweight, generic library for Erlang to quickly integrate a rate-limiter inside your application.

Example

See examples directory to see how use the library with the mongodb backend.

Configuration

Add limitless to your rebar.config deps:

{deps,
 [
  {limitless, "",
   {git, "https://github.com/rpt/limitless.git",
    {tag, "v1.1.0"}}}
 ]}.

Remember to add also the libraries needed by the specific backend (in the mongodb example: uuid, mongopool).

Add a configuration for specify which backend are you using and which kind of limits are you imposing.

In this example we define that every request of type token will be controlled by two different limits:

  1. The first limit is 1000 req/day
  2. The other limit is 100 req/15min
[
  {limitless, [
    {backend, [
      % specify the backend to use
      {name, limitless_backend_mongopool},
      % and the backend configuration
      {config, [
        {table, limitless},
        {pool, mymongopool}
      ]}
    ]},
    {limits, [
      {token, [
        [
          % max 1000 req/day
          {type, <<"Token-Daily">>},
          {frequency, 86400}, % 1 day = 3600 * 24h
          {requests, 1000}
        ],
        [
          % max 100 req/15min
          {type, <<"Token-15min">>},
          {frequency, 900}, % 15 min = 60 * 15
          {requests, 100}
        ]
      ]}
    ]}
  ]}
]

Note: in the configuration there is also a term mongopool to complete the backend configuration.

On the application running, when a new object token we'll be created, also should be registered in the limits lists with:

{ok, Ctx} = limitless:init().
limitless:setup(RequestToken, token, ctx).

In this way, every time you receive a request from a token RequestToken, you be able to check if it reached any of its limits (in the example: 1000 req/day and 100 req/15min):

{Result, ConsumedTokens, InfoTokens} = limitless:is_reached(
                                                [RequestToken], LimitlessCtx)

If Result is true, it means that at least one limit is reached (see ConsumedTokens to know which one).

The InfoTokens are useful information to build a table about the current situation.

In the example, it's used to build the HTTP relative headers:

Req2 = limitless_cowboy_utils:set_rate_limiter_headers(InfoTokens, Req),

It'll add the following headers to the cowboy response:

X-RateLimit-Token-Daily-Limit: 1000
X-RateLimit-Token-Daily-Remaining: 1000
X-RateLimit-Token-Daily-Reset: 86250
X-RateLimit-Token-15min-Limit: 100
X-RateLimit-Token-15min-Remaining: 100
X-RateLimit-Token-15min-Reset: 750

Status

Stable release.