Validation

This chapter will demonstrate how you can apply different validation strategies for the @marblejs/middleware-io validator using io-ts schemas.

io-ts is a nifty library that validates and checks at runtime that incoming data has the shape that you expect. This powerful library can extract a static type from the validator that is guaranteed to match any values that pass validation.

String literals

You can create string literals using combination of t.union and t.literal codecs. Lets say that we would like to validate users that can contain different roles from the given set: 'ADMIN' | 'GUEST'.

const userSchema = t.type({
  id: t.string,
  name: t.string,
  roles: t.array(t.union([
    t.literal('ADMIN'),
    t.literal('GUEST'),
  ])),
});

type User = t.TypeOf<typeof userSchema>;

👇

type User = {
  id: string;
  name: string;
  roles: ('ADMIN' | 'GUEST')[];
};

Branded types

io-ts allows you to create a custom codec validators that must match to given predicate. There are ton of use cases where you can use this mechanism, eg. you would like to validate users which are adult (age is bigger or equal 18).

interface AgeAdultBrand {
  readonly AgeAdult: unique symbol;
}

const AdultAge = t.brand(
  t.number,
  (age): age is t.Branded<number, AgeAdultBrand> => age >= 18,
  'AgeAdult'
);

const userSchema = t.type({
  id: t.string,
  name: t.string,
  age: AdultAge,
}),

type User = t.TypeOf<typeof userSchema>;

👇

type User = {
  id: string;
  name: string;
  age: t.Branded<number, NumberAdultBrand>;
};

As you can see, io-ts requires some boilerplate in order to have it properly typed, but the benefits are invaluable.

Optional properties

According to io-ts documentation you can define a validator with optional values as an intersection of optional and required properties.

const StorySchema = t.intersection([
  t.type({
    type: t.literal('story'),
    commentsTotal: t.number,
  }),
  t.partial({
    description: t.string,
    url: t.string,
  })
]);

type Story = t.TypeOf<typeof StorySchema>;

👇

type Story = {
  type: 'story';
  commentsTotal: number;
} & {
  description?: string | undefined;
  url?: string | undefined;
}

As you can see in the example above, the generated Story type is not as clean as it might be. Jasse Hallett in his article proposed a slightly different approach to defining optional values in validator schemas. Lets define a handy optional combinator!

export const optional = <T extends t.Any>(
  type: T,
  name = `${type.name} | undefined`
): t.UnionType<
  [T, t.UndefinedType],
  t.TypeOf<T> | undefined,
  t.OutputOf<T> | undefined,
  t.InputOf<T> | undefined
> =>
  t.union<[T, t.UndefinedType]>([type, t.undefined], name);

Technically this implies that we expect the given property to be present in every case, but that the value might be undefined. In practice, that distinction often does not matter, and io-ts will validate an object that is missing a required property if the type of that property is allowed to be undefined.

Using the introduced combinator our StorySchema definition can be much more cleaner and readable.

const StorySchema = t.type({
  type: t.literal('story'),
  commentsTotal: t.number,
  description: optional(t.string),
  url: optional(t.string),
});

type Story = t.TypeOf<typeof StorySchema>;

👇

type Story = {
  type: 'story';
  commentsTotal: number;
  description: string | undefined;
  url: string | undefined;
}

Last updated