Validation

There are a bunch of different ways that an API client can send bad data: they might send malformed JSON... or send a blank title field... maybe because a user forgot to fill in a field on the frontend. The job of our API is to respond to all of these situations in an informative and consistent way so that errors can be easily understood, parsed and communicated back to humans.

Invalid JSON Handling

This is one of the areas where API Platform really excels. Let's do some experimenting: what happens if we accidentally send some invalid JSON? Remove the last curly brace.

Try it! Woh! Nice! This comes back with a new "type" of resource: a hydra:error. If an API client understands Hydra, they'll instantly know that this response contains error details. And even if someone has never heard of Hydra before, this is a super clear response. And, most importantly, every error has the same structure.

The status code is also 400 - which means the client made an error in the request - and hydra:description says "Syntax error". Without doing anything, API Platform is already handling this case. Oh, and the trace, while maybe useful right now during development, will not show up in the production environment.

Field Validation

What happens if we just delete everything and send an empty request? Oh... that's still technically invalid JSON. Try just {}.

Ah... this time we get a 500 error: the database is exploding because some of the columns cannot be null. Oh, and like I mentioned earlier, if you're using Symfony 4.3, you might already see a validation error instead of a database error because of a new feature where validation rules are automatically added by reading the Doctrine database rules.

But, whether you're seeing a 500 error, or Symfony is at least adding some basic validation for you, the input data that's allowed is something we want to control: I want to decide the exact rules for each field.

Adding validation rules is... oh, so nice. And, unless you're new to Symfony, this will look delightfully boring. Above title, to make it required, add @Assert\NotBlank(). Let's also add @Assert\Length() here with, how about, min=2 and max=50. Heck, let's even set the maxMessage to

Describe your cheese in 50 chars or less

What else? Above description, add @Assert\NotBlank. And for price, @Assert\NotBlank(). You could also add a GreaterThan constraint to make sure this is above zero.

Ok, switch back over and try sending no data again. Woh! It's awesome! The @type is ConstraintViolationList! That's one of the types that was described by our JSON-LD documentation!

Go to /api/docs.jsonld. Down under supportedClasses, there's EntryPoint and here is ConstraintViolation and ConstraintViolationList, which describes what each of these types look like.

And the data on the response is really useful: a violations array where each error has a propertyPath - so we know what field that error is coming from - and message. So... it all just... works!

And if you try passing a title that's longer than 50 characters... and execute, there's our custom message.

Validation for Passing Invalid Types

Perfect! We're done! But wait... aren't we missing a bit of validation on the price field? We have @NotBlank... but what's preventing us from sending text for this field? Anything?

Let's try it! Set the price to apple, and execute.

Ha! It fails with a 400 status code! That's awesome! It says:

The type of the price attribute must be int, string given

If you look closely, it's failing during the deserialization process. It's not technically a validation error - it's a serialization error. But to the API client, it looks just about the same, except that this returns an Error type instead of a ConstraintViolationList... which probably makes sense: if some JavaScript is making this request, that JavaScript should probably have some built-in validation rules to prevent the user from ever adding text to the price field.

The point is: API Platform, well, really, the serializer, knows the types of your fields and will make sure that nothing insane gets passed. It knows that price is an integer from two sources actually: the Doctrine @ORM\Column metadata on the field and the argument type-hint on setPrice().

The only thing we really need to worry about is adding "business rules" validation: adding the @Assert validation constraints to say that this field is required, that field has a min length, etc. Basically, validation in API Platform works exactly like validation in every Symfony app. And API Platform takes care of the boring work of mapping serialization and validation failures to a 400 status code and descriptive, consistent error responses.

Next, let's create a second API Resource! A User! Because things will get really interesting when we start creating relations between resources.