What’s the best way to define a relationship in your Laravel factories? I’ll share some tips that will help you avoid some common pitfalls, as well as give you a deeper insight into how factories work.
Factories were introduced back in Laravel 5.1. From the very beginning, the docs showed this sample code for defining a factory that had a relationship:
<?php
$factory->define(App\Post::class, function ($faker) {
return [
'title' => $faker->title,
'content' => $faker->paragraph,
'user_id' => function () {
return factory(App\User::class)->create()->id;
}
];
});
But I’ve seen projects that define that factory like this instead:
<?php
$factory->define(App\Post::class, function ($faker) {
return [
'title' => $faker->title,
'content' => $faker->paragraph,
'user_id' => factory(App\User::class)->create()->id,
];
});
Did you spot the difference? The docs recommend wrapping relationship factories in a closure, but this second example just has the relationship factory as the direct value. Why does this seemingly small difference matter?
Consider this chunk of test setup logic:
<?php
$user = factory(User::class)->create();
$post = factory(Post::class)->create([
'user_id' => $user->id,
])
This code works differently depending on whether or not your PostFactory wrapped that UserFactory in a closure or not. If it’s wrapped in a closure, as the docs recommend, you’ll end up with one User and one Post as expected. But if you don’t wrap it in the closure, you’ll end up with two Users and one Post.
But wait! I’m specifying the user_id when calling the factory, so why is the PostFactory creating a second user anyways? What’s going on?
If you don’t wrap the relationship’s factory in a closure, it will be evaluated when the factory returns its default values. At the moment the factory is invoked, and the default values are returned, a new user is created. When your attributes are merged in with the defaults, it’s already too late. That created user is now in your database.
If it’s wrapped in a closure, evaluation is delayed until after any attributes are merged in. This way, if you specify your own user_id, that closure will never be executed, and the UserFactory will not be invoked. No mysterious second user will be created!
If you think that closure syntax is too verbose, you could chain the lazy method to the UserFactory:
<?php
$factory->define(Post::class, function (Faker $faker) {
return [
'title' => $faker->sentence,
'user_id' => factory(\App\User::class)->lazy(),
];
});
If you look at the lazy helper, it’s literally just wrapping your FactoryBuilder in a closure for you:
<?php
public function lazy(array $attributes = [])
{
return function () use ($attributes) {
return $this->create($attributes);
};
}
Or better yet, just pass the FactoryBuilder itself as your value. Don’t reference the id property on your User:
<?php
$factory->define(Post::class, function (Faker $faker) {
return [
'title' => $faker->sentence,
'user_id' => factory(\App\User::class),
];
});
In fact, this is the recommendation in the docs for Laravel versions 6 and 7.
But what about the cool new class-based factories in Laravel 8? More on that in a future blog post.