A progressive Node.js framework for building efficient and scalable server-side applications.

Description

A Rate-Limiter for NestJS, regardless of the context.

For an overview of the community storage providers, see Community Storage Providers.

This package comes with a couple of goodies that should be mentioned, first is the ThrottlerModule.

Installation

1$ npm i --save @nestjs/throttler

Versions

@nestjs/throttler@^1 is compatible with Nest v7 while @nestjs/throttler@^2 is compatible with Nest v7 and Nest v8, but it is suggested to be used with only v8 in case of breaking changes against v7 that are unseen.

For NestJS v10, please use version 4.1.0 or above

Table of Contents

• Description
• Versions
• Table of Contents
• Usage
  • ThrottlerModule
  • Customization
  • ThrottlerStorage
  • Proxies
  • Working with Websockets
  • Working with GraphQL
• Community Storage Providers

Usage

ThrottlerModule

Once the installation is complete, the ThrottlerModule can be configured as any other Nest package with forRoot or forRootAsync methods.

1@@filename(app.module)
2@Module({
3  imports: [
4    ThrottlerModule.forRoot([{
5      ttl: 60000,
6      limit: 10,
7    }]),
8  ],
9})
10export class AppModule {}

The above will set the global options for the ttl, the time to live in milliseconds, and the limit, the maximum number of requests within the ttl, for the routes of your application that are guarded.

Once the module has been imported, you can then choose how you would like to bind the ThrottlerGuard. Any kind of binding as mentioned in the guards section is fine. If you wanted to bind the guard globally, for example, you could do so by adding this provider to any module:

1{
2  provide: APP_GUARD,
3  useClass: ThrottlerGuard
4}

Multiple Throttler Definitions

There may come upon times where you want to set up multiple throttling definitions, like no more than 3 calls in a second, 20 calls in 10 seconds, and 100 calls in a minute. To do so, you can set up your definitions in the array with named options, that can later be referenced in the @SkipThrottle() and @Throttle() decorators to change the options again.

1@@filename(app.module)
2@Module({
3  imports: [
4    ThrottlerModule.forRoot([
5      {
6        name: 'short',
7        ttl: 1000,
8        limit: 3,
9      },
10      {
11        name: 'medium',
12        ttl: 10000,
13        limit: 20
14      },
15      {
16        name: 'long',
17        ttl: 60000,
18        limit: 100
19      }
20    ]),
21  ],
22})
23export class AppModule {}

Customization

There may be a time where you want to bind the guard to a controller or globally, but want to disable rate limiting for one or more of your endpoints. For that, you can use the @SkipThrottle() decorator, to negate the throttler for an entire class or a single route. The @SkipThrottle() decorator can also take in an object of string keys with boolean values for if there is a case where you want to exclude most of a controller, but not every route, and configure it per throttler set if you have more than one. If you do not pass an object, the default is to use {{ '{' }} default: true {{ '}' }}

1@SkipThrottle()
2@Controller('users')
3export class UsersController {}

This @SkipThrottle() decorator can be used to skip a route or a class or to negate the skipping of a route in a class that is skipped.

1@SkipThrottle()
2@Controller('users')
3export class UsersController {
4  // Rate limiting is applied to this route.
5  @SkipThrottle({ default: false })
6  dontSkip() {
7    return 'List users work with Rate limiting.';
8  }
9  // This route will skip rate limiting.
10  doSkip() {
11    return 'List users work without Rate limiting.';
12  }
13}

There is also the @Throttle() decorator which can be used to override the limit and ttl set in the global module, to give tighter or looser security options. This decorator can be used on a class or a function as well. With version 5 and onwards, the decorator takes in an object with the string relating to the name of the throttler set, and an object with the limit and ttl keys and integer values, similar to the options passed to the root module. If you do not have a name set in your original options, use the string default You have to configure it like this:

1// Override default configuration for Rate limiting and duration.
2@Throttle({ default: { limit: 3, ttl: 60000 } })
3@Get()
4findAll() {
5  return "List users works with custom rate limiting.";
6}

Proxies

If your application runs behind a proxy server, check the specific HTTP adapter options (express and fastify) for the trust proxy option and enable it. Doing so will allow you to get the original IP address from the X-Forwarded-For header, and you can override the getTracker() method to pull the value from the header rather than from req.ip. The following example works with both express and fastify:

1// throttler-behind-proxy.guard.ts
2import { ThrottlerGuard } from '@nestjs/throttler';
3import { Injectable } from '@nestjs/common';
4
5@Injectable()
6export class ThrottlerBehindProxyGuard extends ThrottlerGuard {
7  protected getTracker(req: Record<string, any>): Promise<string> {
8  return new Promise<string>((resolve, reject) => {
9    const tracker = req.ips.length > 0 ? req.ips[0] : req.ip; // individualize IP extraction to meet your own needs
10    resolve(tracker);
11  });
12  }
13}
14
15// app.controller.ts
16import { ThrottlerBehindProxyGuard } from './throttler-behind-proxy.guard';
17
18@UseGuards(ThrottlerBehindProxyGuard)

info Hint You can find the API of the req Request object for express here and for fastify here.

Websockets

This module can work with websockets, but it requires some class extension. You can extend the ThrottlerGuard and override the handleRequest method like so:

1@Injectable()
2export class WsThrottlerGuard extends ThrottlerGuard {
3  async handleRequest(requestProps: ThrottlerRequest): Promise<boolean> {
4    const { context, limit, ttl, throttler, blockDuration, getTracker, generateKey } = requestProps;
5
6    const client = context.switchToWs().getClient();
7    const tracker = client._socket.remoteAddress;
8    const key = generateKey(context, tracker, throttler.name);
9    const { totalHits, timeToExpire, isBlocked, timeToBlockExpire } =
10      await this.storageService.increment(key, ttl, limit, blockDuration, throttler.name);
11
12    const getThrottlerSuffix = (name: string) => (name === 'default' ? '' : `-${name}`);
13
14    // Throw an error when the user reached their limit.
15    if (isBlocked) {
16      await this.throwThrottlingException(context, {
17        limit,
18        ttl,
19        key,
20        tracker,
21        totalHits,
22        timeToExpire,
23        isBlocked,
24        timeToBlockExpire,
25      });
26    }
27
28    return true;
29  }
30}

info Hint If you are using ws, it is necessary to replace the _socket with conn

There's a few things to keep in mind when working with WebSockets:

• Guard cannot be registered with the APP_GUARD or app.useGlobalGuards()
• When a limit is reached, Nest will emit an exception event, so make sure there is a listener ready for this

info Hint If you are using the @nestjs/platform-ws package you can use client._socket.remoteAddress instead.

GraphQL

The ThrottlerGuard can also be used to work with GraphQL requests. Again, the guard can be extended, but this time the getRequestResponse method will be overridden

1@Injectable()
2export class GqlThrottlerGuard extends ThrottlerGuard {
3  getRequestResponse(context: ExecutionContext) {
4    const gqlCtx = GqlExecutionContext.create(context);
5    const ctx = gqlCtx.getContext();
6    return { req: ctx.req, res: ctx.res };
7  }
8}

However, when using Apollo Express/Fastify or Mercurius, it's important to configure the context correctly in the GraphQLModule to avoid any problems.

Apollo Server (for Express):

For Apollo Server running on Express, you can set up the context in your GraphQLModule configuration as follows:

1GraphQLModule.forRoot({
2  // ... other GraphQL module options
3  context: ({ req, res }) => ({ req, res }),
4});

Apollo Server (for Fastify) & Mercurius:

When using Apollo Server with Fastify or Mercurius, you need to configure the context differently. You should use request and reply objects. Here's an example:

1GraphQLModule.forRoot({
2  // ... other GraphQL module options
3  context: (request, reply) => ({ request, reply }),
4});

Configuration

The following options are valid for the object passed to the array of the ThrottlerModule's options:

name	the name for internal tracking of which throttler set is being used. Defaults to `default` if not passed
ttl	the number of milliseconds that each request will last in storage
limit	the maximum number of requests within the TTL limit
blockDuration	the number of milliseconds that request will be blocked for that time
ignoreUserAgents	an array of regular expressions of user-agents to ignore when it comes to throttling requests
skipIf	a function that takes in the ExecutionContext and returns a boolean to short circuit the throttler logic. Like @SkipThrottler(), but based on the request
getTracker	a function that takes in the Request and returns a string to override the default logic of the getTracker method
generateKey	a function that takes in the ExecutionContext, the tacker string and the throttler name as a string and returns a string to override the final key which will be used to store the rate limit value. This overrides the default logic of the generateKey method

If you need to set up storages instead, or want to use a some of the above options in a more global sense, applying to each throttler set, you can pass the options above via the throttlers option key and use the below table

storage	a custom storage service for where the throttling should be kept track. See here.
ignoreUserAgents	an array of regular expressions of user-agents to ignore when it comes to throttling requests
skipIf	a function that takes in the ExecutionContext and returns a boolean to short circuit the throttler logic. Like @SkipThrottler(), but based on the request
throttlers	an array of throttler sets, defined using the table above
errorMessage	a string OR a function that takes in the ExecutionContext and the ThrottlerLimitDetail and returns a string which overrides the default throttler error message
getTracker	a function that takes in the Request and returns a string to override the default logic of the getTracker method
generateKey	a function that takes in the ExecutionContext, the tacker string and the throttler name as a string and returns a string to override the final key which will be used to store the rate limit value. This overrides the default logic of the generateKey method

Async Configuration

You may want to get your rate-limiting configuration asynchronously instead of synchronously. You can use the forRootAsync() method, which allows for dependency injection and async methods.

One approach would be to use a factory function:

1@Module({
2  imports: [
3    ThrottlerModule.forRootAsync({
4      imports: [ConfigModule],
5      inject: [ConfigService],
6      useFactory: (config: ConfigService) => [
7        {
8          ttl: config.get('THROTTLE_TTL'),
9          limit: config.get('THROTTLE_LIMIT'),
10        },
11      ],
12    }),
13  ],
14})
15export class AppModule {}

You can also use the useClass syntax:

1@Module({
2  imports: [
3    ThrottlerModule.forRootAsync({
4      imports: [ConfigModule],
5      useClass: ThrottlerConfigService,
6    }),
7  ],
8})
9export class AppModule {}

This is doable, as long as ThrottlerConfigService implements the interface ThrottlerOptionsFactory.

Storages

The built in storage is an in memory cache that keeps track of the requests made until they have passed the TTL set by the global options. You can drop in your own storage option to the storage option of the ThrottlerModule so long as the class implements the ThrottlerStorage interface.

info Note ThrottlerStorage can be imported from @nestjs/throttler.

Time Helpers

There are a couple of helper methods to make the timings more readable if you prefer to use them over the direct definition. @nestjs/throttler exports five different helpers, seconds, minutes, hours, days, and weeks. To use them, simply call seconds(5) or any of the other helpers, and the correct number of milliseconds will be returned.

Migration Guide

For most people, wrapping your options in an array will be enough.

If you are using a custom storage, you should wrap you ttl and limit in an array and assign it to the throttlers property of the options object.

Any @ThrottleSkip() should now take in an object with string: boolean props. The strings are the names of the throttlers. If you do not have a name, pass the string 'default', as this is what will be used under the hood otherwise.

Any @Throttle() decorators should also now take in an object with string keys, relating to the names of the throttler contexts (again, 'default' if no name) and values of objects that have limit and ttl keys.

Warning Important The ttl is now in milliseconds. If you want to keep your ttl in seconds for readability, use the seconds helper from this package. It just multiplies the ttl by 1000 to make it in milliseconds.

For more info, see the Changelog

Community Storage Providers

• Redis
• Mongo

Feel free to submit a PR with your custom storage provider being added to this list.

License

Nest is MIT licensed.

🔼 Back to TOC