io-ts

The idea

A value of type Type<S, A> (called "runtime type") is the runtime representation of the static type A:

class Type<S, A> {
  readonly _A: A
  readonly _S: S
  constructor(
    readonly name: string,
    readonly is: Is<A>,
    readonly validate: Validate<S, A>,
    readonly serialize: Serialize<S, A>
  ) {}
}

where Validate<A> is a specific validation function for the type A

export interface ContextEntry {
  readonly key: string
  readonly type: Any | NeverType
}
export type Context = Array<ContextEntry>
export interface ValidationError {
  readonly value: any
  readonly context: Context
}
export type Errors = Array<ValidationError>
export type Validation<A> = Either<Errors, A>
export type Is<A> = (v: any) => v is A
export type Validate<S, A> = (s: S, context: Context) => Validation<A>
export type Serialize<S, A> = (a: A) => S

Note. The Either type is defined in fp-ts, a library containing implementations of common algebraic types in TypeScript.

Example

A runtime type representing string can be defined as

import * as t from 'io-ts'

export class StringType extends Type<any, string> {
  constructor() {
    super(
      'string',
      (v): v is string => typeof v === 'string',
      (s, c) => (this.is(s) ? success(s) : failure(s, c)),
      a => a
    )
  }
}

A runtime type can be used to validate an object in memory (for example an API payload)

const Person = t.interface({
  name: t.string,
  age: t.number
})

// ok
t.validate(JSON.parse('{"name":"Giulio","age":43}'), Person) // => Right({name: "Giulio", age: 43})

// ko
t.validate(JSON.parse('{"name":"Giulio"}'), Person) // => Left([...])

Error reporters

A reporter implements the following interface

interface Reporter<A> {
  report: (validation: Validation<any>) => A;
}

This package exports two default reporters

• PathReporter: Reporter<Array<string>>
• ThrowReporter: Reporter<void>

Example

import { PathReporter } from 'io-ts/lib/PathReporter'
import { ThrowReporter } from 'io-ts/lib/ThrowReporter'

const validation = t.validate({ name: 'Giulio' }, Person)

console.log(PathReporter.report(validation))
// => ['Invalid value undefined supplied to : { name: string, age: number }/age: number']

ThrowReporter.report(validation)
// => throws 'Invalid value undefined supplied to : { name: string, age: number }/age: number'

Community

• io-ts-reporters - Error reporters for io-ts
• geojson-iots - Runtime types for GeoJSON as defined in rfc7946 made with io-ts

TypeScript integration

Runtime types can be inspected

This library uses TypeScript extensively. Its API is defined in a way which automatically infers types for produced values

Note that the type annotation isn't needed, TypeScript infers the type automatically based on a schema.

Static types can be extracted from runtime types with the TypeOf operator

type IPerson = t.TypeOf<typeof Person>

// same as
type IPerson = {
  name: string,
  age: number
}

Recursive types

Recursive types can't be inferred by TypeScript so you must provide the static type as a hint

// helper type
type ICategory = {
  name: string,
  categories: Array<ICategory>
}

const Category =
  t.recursion <
  ICategory >
  ('Category',
  self =>
    t.interface({
      name: t.string,
      categories: t.array(self)
    }))

Implemented types / combinators

import * as t from 'io-ts'

Type	TypeScript	Flow	Runtime type / combinator
null	null	null	t.null or t.nullType
undefined	undefined	void	t.undefined
string	string	string	t.string
number	number	number	t.number
boolean	boolean	boolean	t.boolean
any	any	any	t.any
never	never	empty	t.never
object	object	✘	t.object
integer	✘	✘	t.Integer
array of any	Array<any>	Array<any>	t.Array
array of type	Array<A>	Array<A>	t.array(A)
dictionary of any	{ [key: string]: any }	{ [key: string]: any }	t.Dictionary
dictionary of type	{ [K in A]: B }	{ [key: A]: B }	t.dictionary(A, B)
function	Function	Function	t.Function
literal	's'	's'	t.literal('s')
partial	Partial<{ name: string }>	$Shape<{ name: string }>	t.partial({ name: t.string })
readonly	Readonly<T>	ReadOnly<T>	t.readonly(T)
readonly array	ReadonlyArray<number>	ReadOnlyArray<number>	t.readonlyArray(t.number)
interface	interface A { name: string }	interface A { name: string }	t.interface({ name: t.string }) or t.type({ name: t.string })
interface inheritance	interface B extends A {}	interface B extends A {}	t.intersection([ A, t.interface({}) ])
tuple	[ A, B ]	[ A, B ]	t.tuple([ A, B ])
union	A | B	A | B	t.union([ A, B ])
intersection	A & B	A & B	t.intersection([ A, B ])
keyof	keyof M	$Keys<M>	t.keyof(M)
recursive types	see Recursive types	see Recursive types	t.recursion(name, definition)
refinement	✘	✘	t.refinement(A, predicate)
strict/exact types	✘	$Exact<{{ name: t.string }}>	t.strict({ name: t.string })

Refinements

You can refine a type (any type) using the refinement combinator

const Positive = t.refinement(t.number, n => n >= 0, 'Positive')

const Adult = t.refinement(Person, person => person.age >= 18, 'Adult')

Strict/Exact interfaces

You can make an interface strict (which means that only the given properties are allowed) using the strict combinator

const Person = t.interface({
  name: t.string,
  age: t.number
})

const StrictPerson = t.strict(Person.props)

t.validate({ name: 'Giulio', age: 43, surname: 'Canti' }, Person) // ok
t.validate({ name: 'Giulio', age: 43, surname: 'Canti' }, StrictPerson) // fails

Mixing required and optional props

Note. You can mix required and optional props using an intersection

const A = t.interface({
  foo: t.string
})

const B = t.partial({
  bar: t.number
})

const C = t.intersection([A, B])

type CT = t.TypeOf<typeof C>

// same as
type CT = {
  foo: string
  bar?: number
}

Custom types

You can define your own types. Let's see an example

import * as t from 'io-ts'

// represents a Date from an ISO string
const DateFromString = new t.Type<any, Date>(
  'DateFromString',
  (v): v is Date => v instanceof Date,
  (v, c) =>
    t.string.validate(v, c).chain(s => {
      const d = new Date(s)
      return isNaN(d.getTime()) ? t.failure(s, c) : t.success(d)
    }),
  a => a.toISOString()
)

const s = new Date(1973, 10, 30).toISOString()

t.validate(s, DateFromString)
// right(new Date('1973-11-29T23:00:00.000Z'))

t.validate('foo', DateFromString)
// left(errors...)

Note that you can deserialize while validating.

Custom combinators

You can define your own combinators. Let's see some examples

The maybe combinator

An equivalent to T | null

export function maybe<RT extends t.Any>(type: RT, name?: string): t.UnionType<[RT, t.NullType], t.TypeOf<RT> | null> {
  return t.union<[RT, t.NullType]>([type, t.null], name)
}

Recipes

Is there a way to turn the checks off in production code?

No, however you can define your own logic for that (if you really trust the input)

import * as t from 'io-ts'
import { failure } from 'io-ts/lib/PathReporter'

const { NODE_ENV } = process.env

export function unsafeValidate<S, A>(value: any, type: t.Type<S, A>): A {
  if (NODE_ENV !== 'production') {
    return t.validate(value, type).fold(errors => {
      throw new Error(failure(errors).join('\n'))
    }, t.identity)
  }
  // unsafe cast
  return value as A
}

Known issues

Due to an upstream bug, VS Code might display weird types for nested interfaces

const NestedInterface = t.interface({
  foo: t.interface({
    bar: t.string
  })
})

type NestedInterfaceType = t.TypeOf<typeof NestedInterface>
/*
Hover on NestedInterfaceType will display

type NestedInterfaceType = {
  foo: t.InterfaceOf<{
    bar: t.StringType;
  }>;
}

instead of

type NestedInterfaceType = {
  foo: {
    bar: string;
  };
}
*/