Awesome

Summary

The work that gcanti has done with io-ts is really useful, but it relies on a lot of language semantics that are exclusive to Typescript and the work on flow-io is now deprecated and no longer maintained. As a result, I have forked his work and refactored a lot of the existing combinators/classes so they more easily align with the semantics of Flow.

Table of Contents

• The idea
• Error handling
  • Error reporters
• Implemented types / combinators
• Flow compatibility
• Flow integration
• Mixing required and optional props
• Union Maps
• Refinements
• Custom types
  • Custom Error Messages
• Tips and Tricks
  • Is there a way to turn the checks off in production code?
  • Union of string literals

The idea

Blog post: "Typescript and validations at runtime boundaries" by @lorefnon

Similiar to io-ts, a value of type Type<A, O, I> (called "runtime type") is the runtime representation of the static type A.

A runtime type can

• decode inputs of type I (through decode)
• encode outputs of type O (through encode)
• be used as a custom type guard (through is)

type Validation<A> = Either<Errors, A>;

class Type<A, +O = A, I = mixed> {
  +_A: A
  +_O: O
  +_I: I
  constructor(
    /** a unique name for this runtime type */
    name: string,
    /** a custom type guard */
    is: (v: mixed) => boolean,
    /** succeeds if a value of type I can be decoded to a value of type A */
    validate: (input: I, context: Context) => Validation<A>,
    /** converts a value of type A to a value of type O */
    encode: (a: A) => O
  ) {}
  /** a version of `validate` with a default context */
  decode(i: I): Either<Errors, A>
  /** a version of `validate` which will throw if invalid */
  assert(i: I): A
}

Example

A runtime type representing string can be defined as

import * as t from 'io-flow-types'

// runtime type definition
export class StringType extends t.Type<string> {
  // equivalent to Type<string, string, mixed> as per type parameter defaults
  constructor() {
    super(
      'string',
      (m): m is string => typeof m === 'string',
      (m, c) => (this.is(m) ? t.success(m) : t.failure(m, c)),
      t.identity
    )
  }
}

// runtime type instance: use this when building other runtime types instances
export const string = new StringType()

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

const Person = t.inexactAll({name: t.String, age: t.Number})

// validation succeeded
Person.decode(JSON.parse('{"name":"John","age":43}')) // => right({name: "John", age: 43})

// validation failed
Person.decode(JSON.parse('{"name":"John"}')) // => left([...])

//assertion succeeded
Person.assert(JSON.parse('{"name":"John","age":43}')) // => {name: "John", age: 43}

//async assertion succeeded
Promise.resolve('{"name":"John","age":43}')
  .then(JSON.parse)
  .then(Person.getAssert())
  .then(val => {
    return val // => {name: "John", age: 43})
  });

Error handling

An error that is uncovered during decoding will be packed into an instance of the ValidationError class.

class ValidationError extends Error {
  +value: mixed;
  +context: Context;
  +message: string;
  constructor(value: mixed, context: Context, message?: string)
}

Besides having a message property, as is standard for Error classes in JavaScript, it also references the value which failed validation along with the context that was used in decoding. By default, if a message isn't supplied, a default one will be constructed based on the context reference

All errors that are uncovered during decoding will be packed into an instance of the AggregateErrors class, which is a subclass of Array<ValidationError>.

class AggregateError extends Array<ValidationError> {
  constructor(...args: ValidationError[])
  messages(): Array<string>
}

Errors can be still be extracted individually as elements of the wrapped array, and the messages can be extracted all at once via the introduction of the messages() method on this class.

An example of Error inspection is shown below:

const Person = t.inexactAll({name: t.String, age: t.Number})

// validation failed with decode
const leftErr = Person.decode(JSON.parse('{}')) // => left([...])

if (leftErr.tag === 'Left') {
  const errs = leftErr.value;
  console.log(errs[0].message)
  // => Invalid value undefined supplied to : { name: string, age: number }/name: string
  console.log(errs[1].message)
  // => Invalid value undefined supplied to : { name: string, age: number }/age: number
}

try {
 // validation throws with assert
 Person.assert(JSON.parse('{}')) // => left([...])
} catch (errs) {
  console.log(errs[0].message)
  // => Invalid value undefined supplied to : { name: string, age: number }/name: string
  console.log(errs[1].message)
  // => Invalid value undefined supplied to : { name: string, age: number }/age: number
}

////async validation
Promise.resolve('{}')
  .then(JSON.parse)
  .then(Person.getAssert())
  .catch(errs => {
    console.log(errs[0].message)
    // => Invalid value undefined supplied to : { name: string, age: number }/name: string
    console.log(errs[1].message)
    // => Invalid value undefined supplied to : { name: string, age: number }/age: number
  });

Error reporters

To implement custom error reporters, it's recommended to design an object which can implement the following interface.

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

The report method above should be prepared to take a Validation instance, which will be either left(AggregateErrors) or right(T) where T is the correctly decoded type. The reporter should check for and then use an instance of the AggregateErrors class and perform the necessary transformation, using the context and value properties on ValidationError instances packed inside AggregateErrors instance.

The example below implements a getPaths function which can be used as the report method on a Reporter-like object. It basically will take validations that fail and print the paths that failed.

import * as t from 'io-flow-types'

const getPathFromError = (error: t.ValidationError) => {
  return error.context.map(({ key }) => key).join('.');
}

const getPaths = <A>(v: t.Validation<A>): Array<string> => {
  return v.fold(errors => [...errors].map(getPathFromError), () => ['no errors'])
}

const Person = t.exactAll({name: t.String, age: t.Number })

console.log(getPaths(Person.decode({}))) // => [ '.name', '.age' ]

Implemented types / combinators

import * as t from 'io-flow-types'

Note: Assume A and B are instances of the t.Type class

Flow compatibility

The library is tested against a range of flow-bin versions, which is listed as the peerDependencies section of this NPM package.

Flow integration

Runtime types can be inspected

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

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

type IPerson = t.TypeOf<typeof Person>;

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

// also the same as
type IPerson = $PropertyType<typeof Person, '_A'>;

Mixing required and optional props

You can mix required and optional props using an intersection

const required = {foo: t.string};
const optional = { bar: t.number }
const C = t.exact<typeof required, typeof optional>({required, optional})
type CT = t.TypeOf<typeof C>;

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

You can call shape to an already defined runtime type if created with one of the exact or inexact functions

const PersonType = t.exactAll({
  name: t.string,
  age: t.number
})

const PartialPersonType = Person.shape();

type PartialPerson = t.TypeOf<typeof PartialPersonType>;

// same as
type PartialPerson = {
  name?: string
  age?: number
}

Union Maps

If you are encoding tagged unions, instead of the general purpose union combinator, you may want to use the unionMap combinator in order to get better performances

const A = t.exactAll({
  tag: t.literal('A'),
  foo: t.string
})

const B = t.exactAll({
  tag: t.literal('B'),
  bar: t.number
})

const U = t.unionMap({A, B}, 'tag')

Refinements

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

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

However, unless you utilize Flow's opaque types, this can't be enforced via a static check. For stricter safety, you should use the opaqueRefine function and supply the opaque type as a generic

opaque type Positive: number = number;
const positive = t.opaqueRefine<typeof t.Number, Positive>(t.Number, num => num > 0, 'Positive')

Custom types

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

import * as t from 'io-flow-types'

// represents a Date from an ISO string
const DateFromString = new t.Type<Date, string>(
  'DateFromString',
  (m): m is Date => m instanceof Date,
  (m, c) =>
    t.string.validate(m, 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()

DateFromString.decode(s)
// right(new Date('1973-11-29T23:00:00.000Z'))

DateFromString.decode('foo')
// left(errors...)

Note that you can deserialize while validating.

Custom Error Messages

You can set your own error message by providing a message argument to failure

Example

const NumberFromString = new t.Type<number, string, unknown>(
  'NumberFromString',
  t.number.is,
  (u, c) =>
    t.string.validate(u, c).chain(s => {
      const n = +s
      return isNaN(n) ? t.failure(u, c, 'cannot parse to a number') : t.success(n)
    }),
  String
)

console.log(PathReporter.report(NumberFromString.decode('a')))
// => ['cannot parse to a number']

Tips and Tricks

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-flow-types';

const { NODE_ENV } = process.env

export function unsafeDecode<A, O>(value: mixed, type: t.Type<A, O>): t.Either<t.Errors, A> {
  if (NODE_ENV !== 'production' || type.encode !== t.identity) {
    return type.decode(value)
  } else {
    // unsafe cast
    return t.Right((value: any): A)
  }
}

// or...

import { failure } from 'io-flow-types/lib/PathReporter'

export function unsafeGet<A, O>(value: mixed, type: t.Type<A, O>): A {
  if (NODE_ENV !== 'production' || type.encode !== t.identity) {
    return type.decode(value).getOrElseL(errors => {
      throw new Error(failure(errors).join('\n'))
    })
  } else {
    // unsafe cast
    return ((value: any): A)
  }
}

Union of string literals

Use keyof instead of union when defining a union of string literals

const Bad = t.union([
  t.literal<'foo'>('foo'),
  t.literal<'bar'>('bar'),
  t.literal<'baz'>('baz')
  // etc...
])

const Good = t.keyof({
  foo: null,
  bar: null,
  baz: null
  // etc...
})

Benefits

• unique check for free
• better performance
• quick info stays responsive

Known issues

• TODO