AddHasFactoryToModelsRector doesn't check if factory exists

[hackel]

After enabling the new LARAVEL_FACTORIES set, I had to revert 120 changed models in a project where the HasFactory trait was added when no factory class actually exists. I would argue that in its current state, AddHasFactoryToModelsRector doesn't belong in this set as it's not reasonable to assume every app has a factory for every single model in their application.

Ideally, the rector would check if the factory exists first prior to adding the HasFactory trait. It would be nice if it added the appropriate @use docblock referencing the factory class as well. I don't know if this is currently feasible, though.

Just my two cents. Skipping that rule is an easy enough solution as well.

[GeniJaho]

You're right, we should check in the rule if the factory actually exists. I think that is possible using reflection.
There's this open PR that attempts to add the @use annotation to the trait, and it's doing some guesswork regarding the correct factory name. We could attempt to 'merge' both approaches, or just keep these as separate rules.

For now, I think it's better to fix this rule to check for factory existence, instead of removing it from the set.

However, it's not that easy since the rule is already configurable with a list of allowed models to add this trait to.

->withConfiguredRule(AddHasFactoryToModelsRector::class, [
    \App\Models\User::class,
    \App\Models\Photo::class,
])

Adding a new configuration option (for the factories namespace, which would default to \Database\Factories) would break the existing Rector configs.

We might have to make the rule check for factory existence in the Database\Factories namespace by default, which is not ideal, but at least it would not produce invalid HasFactory traits, and it would not break the existing configs.