Nest Starter

A boilerplate for NestJS, using Fastify.

Features

• Config
• Database
• Swagger (API docs)
• Query Parsing
• File Upload
• Logging
• Request Context
• Auth guard
• Rate limiting
• Request body validation
• Exception Handling
• Health Check
• OpenTelemetry
• Docker
• Testing
• Continuous Integration

Installation

$ npm install

Running

# development
$ npm run start
# watch mode
$ npm run start:dev
# production mode
$ npm run start:prod

Config

The node-config package to manage configs.

Default config values are found in default.json.
These values can be overridden by:

• Creating config files as described in node-config docs
• Creating a local.json file in config/
• Creating a .env file in the project directory. (supported via dotenv)
• Setting environment variables. See the environment variable mappings
  in custom-environment-variables.json.

CORS

CORS is configurable via the config.

The following values are acceptable for config.cors.origins:

• * - Will accept any origin. The access-control-allow-origin header will always respond with the exact origin of the request, not a *.
• https://example.com,https://example2.com - Will accept all domains in the comma separated list.

Database

Objection (which uses Knex.js)
is used for database operations.

It uses PostgreSQL by default, but that can be changed by changing the client
in knexfile.ts.
See Knex documentation for supported databases.

Conversion of column names from camel case to snake case is automatically done in most cases.
It however does not work in all cases. See excerpt below from Objection docs
(Snake case to camel case conversion):

When the conversion is done on objection level only database columns of the returned
rows (model instances) are convered to camel case. You still need to use snake case in
relationMappings and queries. Note that insert, patch, update and their variants still
take objects in camel case. The reasoning is that objects passed to those methods
usually come from the client that also uses camel case.

Migrations

Schema changes require migrations.

Migrations can be created using:

npm run migration:make

When the server starts, migrations will run automatically, or, you can run the migrations
using npm run migration:latest

Swagger

Swagger documentation is automatically generated from the routes.

By default it is available at http://127.0.0.1:3000/docs

Query Parsing

URL query to DB query parsing is available. See query-parser.ts.

Note: There is currently no limitation put on the complexity of the query, so this should be exposed with caution.

Example:

limit=10&page=1&orderBy=(averageRating,DESC)&filter=(authors,=,Bill Bryson):(yearPublished,>,2000)

SELECT "*"
FROM   "..."
WHERE  "authors" = 'Bill Bryson'
  AND "year_published" > '2000'
ORDER  BY "average_rating" DESC
  LIMIT  10

Paging

Paging can be done using limit and page.

Filters

Filters take the format of (column,operand,value) where the operand can be =,>,>=,<,<=.
Column names are automatically converted to snake case.
Multiple filters can be specified using a colon : as the delimiter.

Ordering

Ordering takes the format of (column,direction) where direction can be desc,DESC,asc,ASC.
Multiple orderings can be specified using a colon : as the delimiter.

File Upload

File uploads are configured.

See FileUploadDto for an example dto.

Config

maxSize: number - Max size in bytes. Default 5MB.
uploadDir: string - Upload directory. If blank, it will default to the OS’s temp directory.
removeAfterUpload: boolean - Whether to delete files after the request completes.

Logging

Winston is used as a logging library.

Create a logger instance using new Logger().
A parameter can be passed into the constructor and to be used as a tag (defaults to “Application”).

For example,

import { Logger } from "@/logging/Logger";
const logger = new Logger("AppService");

this.logger.debug("Hello!");
// 2019-05-10 19:47:21.570 | debug: [AppService] Hello!

A custom tag can be passed into the log functions.

this.logger.debug("Hello!", { tag: "AppService.getHello" });
// 2019-05-10 19:54:43.062 | debug: [AppService.getHello] Hello!

Extra data can be passed into the log functions. To enable printing it to the console, set the
config logging.logDataConsole to true.

this.logger.debug(`Hello ${ctx.traceId}`, { data: { traceId: ctx.traceId } });
// 2025-01-15 15:11:57.090 | debug: [AppService.getHello] Hello 47e4a86ea7c0676916b45bed6c80d1bb
// {
//   "traceId": "47e4a86ea7c0676916b45bed6c80d1bb"
// }

Custom Transports

See SampleTransport for an example.

Redact Private Keys

Private keys are automatically redacted in logged data.

The private keys are specified in redact.ts

{
  "username": "mark",
  "contact": {
    "email": "REDACTED"
  }
}

Request Context

nestjs-cls is implemented to maintain the request state.

The request ID is the trace ID if observability is enabled, otherwise it’s a random string.

It can be accessed using the AppClsService.

class Service {
  constructor(
    // This needs to be ClsService<AppClsStore>. AppClsService will not work.
    private readonly clsService: ClsService<AppClsStore>
  ) {}
  handler() {
    this.clsService.getId();
  }
}

Alternatively, the @ReqCtx() can be used.

class Controller {
  @Get()
  getHello(@ReqCtx() ctx: IReqCtx) {
    console.log(ctx.traceId) // 0d8df9931b05fbcd2262bc696a1410a6
  }
}

Authentication

A simple authentication scheme is implemented in auth.guard.ts

Rate Limiting

A rate limiter is configured
using @nestjs/throttler.
It defaults to 100 request per minute (configurable in default.json)
.

Request Body Validation

class-validator
and class-transformer are used to validate
request bodies.

See class-validator decorators
for available validation options.

An example of a response to an invalid body:

{
  "status": 400,
  "message": "Validation errors with properties [name,username,contact.mail,contact.email]",
  "code": "ValidationError",
  "meta": [
    {
      "property": "name",
      "constraints": [
        "property name should not exist"
      ]
    },
    {
      "property": "username",
      "constraints": [
        "username should not be empty"
      ]
    },
    {
      "property": "contact.mail",
      "constraints": [
        "property mail should not exist"
      ]
    },
    {
      "property": "contact.email",
      "constraints": [
        "email should not be empty"
      ]
    }
  ]
}

NB: Raising an error when unknown values are passed can be disabled by
setting validator.forbidUnknown to false in the config.

Response Mapping

Cleaning response objects using can be enabled using the @Serialize(ClassName) decorator.
It uses class-transformer.

Errors & Exception Handling

Exceptions should be thrown using the custom HttpException class.

throw new HttpException(404, `User ${1} was not found`, ErrorCodes.INVALID_USER, { id: 1 });

{
  "status": 404,
  "message": "User 1 was not found",
  "code": "InvalidUser",
  "traceId": "775523bae019485d",
  "meta": {
    "id": 1
  }
}

Regular errors and unhandled exceptions are also caught and returned as a 500 response.

{
  "status": 500,
  "message": "Uncaught exception",
  "code": "InternalError",
  "traceId": "d3cb1b2b3388e3b1"
}

Health Check

Health check endpoints are set up at /ready and /live.

OpenTelemetry

OpenTelemetry support in included with support for traces, metrics and logs.

@opentelemetry/auto-instrumentations-node"">@opentelemetry/auto-instrumentations-node is set up to automatically collect metrics and spans for various services

See the observability README for a compose file with various services for collecting and viewing signals.

Note: Global and per signal instrumentation needs to be enabled via config

Traces

Automatic instrumentation is enabled and will suite most needs.

Custom spans can be created as described in the OpenTelemetry docs.

Metrics

HTTP metrics are automatically collected by @opentelemetry/instrumentation-http

sum by(http_route) (rate(nb_http_server_duration_milliseconds_count[1m]))

See OpenTelemetry docs for how to create custom metrics.

const meter = opentelemetry.metrics.getMeter("UserService");
const getUserCounter = this.meter.createCounter("get_user")
getUserCounter.add(1, { user_id: id });

Logs

Logs are sent to the OpenTelemetry Collector using the OtelTransport.

See logging for how to log.

Docker

The application can be run using docker.

Build

docker build -t nest-boilerplate .
docker run -p 3000:3000 nest-boilerplate

Docker Compose can be used to start the application and a database.

docker compose up -d

Testing

Unit test

There exist unit tests for controllers and services.

Dependencies are mocked using jest.

npm run test

End-to-end test

A Docker container is created to run end to end tests.

See docker-compose-e2e.yml

# e2e tests (docker)
npm run test:e2e
# e2e test (locally)
npm run test:e2e:local

Load test

Load tests are written using Grafana k6.

See load-test/ directory.

CI

Github Actions config