Generating Seed Data

Hard-coding data works great for small datasets. But as your application data requirements grow you're going to find it hard to keep up. You can certainly use production data to help you come up with your seed data, but there are privacy concerns there etc.

Faker.js

A better way is to use a library like faker to

Generate massive amounts of fake (but realistic) data for testing and development.

Faker has a good guide to get you an introduction to how it works in the usage docs. But here's a quick snippet from that to give you an idea:

import { faker } from '@faker-js/faker'

function createRandomUser() {
	return {
		_id: faker.datatype.uuid(),
		avatar: faker.image.avatar(),
		birthday: faker.date.birthdate(),
		email: faker.internet.email(),
		firstName: faker.person.firstName(),
		lastName: faker.person.lastName(),
		sex: faker.person.sexType(),
		subscriptionTier: faker.helpers.arrayElement(['free', 'basic', 'business']),
	}
}

const user = createRandomUser()

Dynamic Data

Another thing we'll be doing in our workshop involves generating a random number of records (like a random number of images or notes). You can use faker to help with that as well:

import { faker } from '@faker-js/faker'

const numberOfThings = faker.number.int({ min: 1, max: 10 })

// then you can use that number to generate an array of things:
const things = Array.from({ length: numberOfThings }, () => {
	// create your thing...
})

Tip: if the thing you're doing is async, you can wrap it in Promise.all:

const things = await Promise.all(
	Array.from({ length: numberOfThings }, async () => {
		// create your async thing...
	}),
)

Unique Data

Another thing you'll want to be careful of when generating data is you want to avoid creating duplicate data for unique fields. For example, if email is a unique field, you'll need to make sure you don't generate the same email address twice. Faker's mechanism for doing this is deprecated in favor of enforce-unique:

const uniqueEmailEnforcer = new UniqueEnforcer()

const email = uniqueEmailEnforcer.enforce(() => faker.internet.email())

This will keep track of the generated emails and call the function again if it finds a duplicate.

Schema Validation

Finally, one more thing you'll want to consider when generating fake data is that the data you generate matches your schema requirements. The seed script isn't necessarily the place to test your application's resilience to bad data (you should have tests for that).

For example, let's say you have a limit of 50 characters for the title of a Review. You'll want to make sure you don't generate a title that's too long. For example:

const reviewTitle = faker.lorem.words(10).slice(0, 50)

Balancing realistic with practical is definitely in force here.