@kofile/config-factory

Config client with optional validation

Stats

StarsIssuesVersionUpdatedCreatedSize
@kofile/config-factory
2.1.04 years ago4 years agoMinified + gzip package size for @kofile/config-factory in KB

Readme

ConfigFactory

Build Status Coverage Status Commitizen friendly semantic-release

ConfigFactory is a small helper function that helps you to safely and easily create configuration objects from an externally injected source. You can also pass-in default values and Joi validation schemas, and you'll get easy-to-handle errors if your config object doesn't pass validation.

Usage

The default export is a curried function that expects a map object that translates the from the data you have to the shape you want. Here's a simple example:

// assuming `process.env.SOME_EXISTING_VAR === 'hello world'`

const map = { myCoolNewKey: 'SOME_EXISTING_VAR' }
const configFactory = makeConfigFactory(map)
const config = configFactory(process.env)

config.get(['myCoolNewKey']) === 'hello world'

In your tests, you might not want to muck around with process.env. Define an object to use instead:

const map = { myCoolNewKey: 'SOME_EXISTING_VAR' }
const configFactory = makeConfigFactory(map)
const config = configFactory({ SOME_EXISTING_VAR: 'hello world' })

config.get(['myCoolNewKey']) === 'hello world'

You can provide a default value as the third value in a tuple:

const map = { myCoolNewKey: ['SOME_EXISTING_VAR', null, 'banans'] }
const configFactory = makeConfigFactory(map)
const config = configFactory({})

config.get(['myCoolNewKey']) === 'bananas'

Also, you can pass in Joi schemas per-key as the second value in a tuple:

const map = { myCoolNewKey: ['SOME_EXISTING_VAR', joi.string().required()] }
const configFactory = makeConfigFactory(map)
const config = configFactory({})

try {
  config.get(['myCoolNewKey'])
} catch (error) {
  error.message === 'Invalid config for myCoolNewKey!'
}

You can validate your entire config by running the validate() method:

const map = { myCoolNewKey: ['SOME_EXISTING_VAR', joi.string().required()] }
const configFactory = makeConfigFactory(map)
const config = configFactory({})

config.validate()

This will appear to do nothing, and that's because when called this way, you need to subscribe to the invalid event handler like so:

config.on('invalid', error => {
  console.error(error)
  process.exit(1)
})

config.validate()

The event listener will receive error from the validation result. Get the message of the failure with error.message.

Subscribing an event handler and invoking config.validate() is the preferred way of handling configuration validation. The previous method (throwing an error on a property-by-property basis) is only there as a fail-safe against forgetting to invoke validate().

Additionally, calling validate() will skip future checks when calling get on a keypath.

NOTE: Incomplete or invalid configurations can lead a service to not start at all or start in a thoroughly broken state; therefor it's advisable to bail hard, fast, and noisily with process.exit(1) if validate() fails.

Lastly, you can also use computed properties: the function for a computed property will receive the all of the non-computed config as an argument.

NOTE: Order of computed property invocations is not guaranteed, so please don't make computed properties that depend on other computed properties. That just sounds like a headache anyways.

const env = { A: 'aaaa' }
const map = { scarletLetter: 'A', phrase: config => `${config.scarletLetter} is not B`) }
const configFactory = makeConfigFactory(map)
const config = configFactory(process.env)

config.get(['phrase']) === 'aaaa is not B'

You can use computed properties to return hard-coded values:

const map = { theAnswer: () => 42 }
const configFactory = makeConfigFactory(map)
const config = configFactory(process.env)

config.get(['theAnswer']) === 42

Computed properties can also be validated and use default values.

Why?

  • Separation of concerns We've described the final shape we want, complete with validations and default values, without hard-coding where the values come from.
  • Ease of testing We've full control over the values of env without having to fuss with globals. This is a nice side effect of the point above.
  • State isolation By treating our configuration as the product of inputs into a function, we make it much harder to pass configuration state around via require/import statements. This is a good thing!

Supported Platforms

Node 7+

Testing

Run tests with yarn test

Authors

If you find any bugs or have a feature request, please open an issue on github!

The npm package download data comes from npm's download counts api and package details come from npms.io.