Facebook's PropTypes-alike interface implementation for client and server, reusable and extensible. It produce error reports instead of throwing or printing into console. It works without React.

Gitter

Table of Contents

• Installation
• Usage
• Examples
• Standard checks
• Non-standard checks
• Decorators
• Checks and groups
• Extension
• API
• License

Installation

Via npm:

npm i typed-props

Or via unpkg.com:

<script src="https://unpkg.com/typed-props@1/dist/typed-props.js"></script>
<script src="https://unpkg.com/typed-props@1/dist/typed-props.min.js"></script>

Complete usage guide.

Usage

Custom types check:

import {Type as T, StrictType as ST, check} from 'typed-props'

const type = T.shape({
  id: T.number.isRequired,
  name: T.string.isRequired,
  email: T.string,
}).isRequired

// Or

const type = ST.shape({
  id: ST.number,
  name: ST.string,
  email: ST.string.optional,
})

const data = {
  id: '1',
  name: null
}

const issues = check(data, type)

Output

Result of validation is an Array of Issues. Issue is an object which describes validator rules violation.

Issue typing:

export type Issue = {
  rule: string
  path: Array<string|number>
  details: {
    reason: string
    [key:string]: any
  }
}

Example:

[
  {
    // Violated rule.
    rule: 'type',
    // Value location. It helps to receive value from nested object.
    path: ['user', 'messages', 0],
    // Details explain what exactly goes wrong.
    details: {
      // Reason helps to identify kind of problem within one validator.
      // Usual values are mismatch, no_matches, and redundant.
      reason: 'mismatch',
      // The next values are validator dependent.
      type: 'string',
      expect: true,
      is: false,
    }
  }
]

Examples

• Create UniqItems check.
• Describe API with TypedProps (with circular types resolution).

Standard checks

Standard checks provided by Facebook's PropTypes:

import {Type, TypeStore} from 'typed-props'

// Object properties should pass all checks.
const shape = Type.shape({
  // Any value except of undefined
  anything: Type.isRequired,
  // Which is equivalent of
  anythingElse: Type.any.isRequired,
  // Number property
  number: Type.number,
  // String property
  string: Type.string,
  // Boolean property
  bool: Type.bool,
  // Object property
  object: Type.object,
  // Array property
  array: Type.array,
  // Array property
  func: Type.func,
  // Symbol property
  symbol: Type.symbol,
  // Property which value is instance of Date
  instanceOf: Type.instanceOf(Date),

  // Complex rules

  // One of check if value is in list of passed primitives
  // It works like an enum
  oneOf: ['one', 'two'],
  // Check if all array values match the passed TypedProps
  arrayOf: Type.number,
  // Check if value is matched any of passed TypedProps.
  oneOfType: Type.oneOfType([
    Type.number,
    Type.string,
  ]),
  // Check if all object properties passes the TypedProps.
  objectOf: Type.objectOf(Type.number),
  // Check shape has described properties and no other props.
  exactShape: Type.exact({
    id: Type.number,
    name: Type.string,
  }),
  // Exact shape with custom props
  exactShape: Type.exactFuzzy({
    id: Type.number,
    name: Type.string,
  }, [/Id$/, Type.string]),
})

const issues = check({}, shape) // => [{path:['anything'], rule: 'isRequired', details: {is: false}}]

Result of check call is array of issues. If there is no issues, this array will be empty.

⚠️ If shape/exact rule property presented by function it should return type to check.

⚠️ .object will fail for arrays and vice versa.

Named and circular types

Using named types it's possible to created nested structures with cross references. It could be useful for validating complex data types. Also names help to produce more meaningful error reports.

const store = new TypeStore()

const userType = Type.shape({
  id: Type.number.isRequired,
  name: Type.string.isRequired,
  friends: Type.arrayOf(
    store.ref('User')
  ),
})

store.add('User', userType)

Non-standard checks

Type.shape({
  // Make type optional
  optionalValue: Type.optional,
  // Determine type in runtime
  propertyDependentType: Type.select(
    [
      ({type}) => type === 'ADD_TODO',
      Type.shape({/* ADD_TODO action payload shape */}),
    ],
    [
      ({type}) => type === 'SET_COMPLETED',
      Type.shape({/* SET_COMPLETED action payload shape */}),
    ],
    [
      () => true,
      Type.any
    ],
  ),
  // Create custom checks when TypedProps is not enough.
  customValue: Type.custom((value) => value === Math.PI),
})

TypedProps have it's own custom checks which make it more handful.

Checkable.optional

This is pseudo check. This rule allows to make some property optional. It can switch off previously defined .isRequired check. It's better to use with StrictType.

StrictType.number.optional

Checkable.is()

(value:*) -> Checkable

This check validates the checkable value to strict equal value argument.

Type.is('user')

Checkable.select()

(...Function|Checkable) -> Checkable

This checker allow to dynamically switch between types depending on input value. It selects the first arguments that is not a function or if it's function and it returns something truly. Then use it as type to check.

Type.select(
  [(value) => (typeof value === 'string'), Type.string],
  [(value) => (typeof value === 'number'), Type.number],
  [() => true, Type.any] // Otherwise accept anything
)

Checkable.exactFuzzy()

(shape:ShapeType, fuzzy:...[RegExp, Checkable]) -> Checkable

ShapeType = Object<key,Checkable|Shape>

Check exact shape with fuzzy properties which keys should match regexp from fuzzy params.

const fuzzy = Type.exactFuzzy({}, [/Id$/, Type.number])
check({userId: 1, postId: 2}, fuzzy)

Checkable.custom()

(check:(it:*) -> bool) -> Checkable

Custom check accepts check argument and use it to validate the value.

Decorators

Experimental Feature. Decorators can check function arguments and value it returns in runtime.

⚠️ Decorators throw CheckError with issues property.

import {StrictType as T, args, result} from 'typed-props'

class Arith {
  // Fixed arguments length example
  @args(T.number, T.number)
  @result(T.number)
  add(a, b) {
    return a + b
  }

  // Variadic argument's length example
  @args(T.number, [T.number])
  @result(T.number)
  addAll(a = 0, ...numbers) {
    return numbers.reduce((sum, b) => sum + b, a)
  }
}

Checks and groups

Currently there are several groups of checkers: existence, type, exact, and complex checks.

• Existance:
  • isRequired
  • optional: removes isRequired
• Type:
  • bool
  • number
  • string
  • func
  • object
  • array
  • any: removes any other type check
• Exact:
  • is
  • oneOf
• Complex:
  • oneOfType
  • arrayOf: overwirites type check with array
  • objectOf: overwirites type check with object
  • shape: replaces exact
  • exact: replaces shape
  • select
  • custom

Type and existence checks are switchable and can replace each other. It's made for flexibility.

Type.isRequired.number.string.func // -> final check is "required function".

In the same time the complex checks like shape or arrayOf require the input value to has certain type object and array respectively. They will overwrite type checks too. But overwiting of complex types currently isn't possible because of unexpected behaviour. It could leads to runtime errors. So it strongly recommended not to overwrite such checks.

It could be changed in the future.

Type.arrayOf(Type.number).isRequired.number // Will throw an error in runtime

Extension

TypedProps support extension throught inheritance. So you can create new type using extend on Type class or mixing in new rules with Object.defineProperty. More real life example is in repository examples directory.

class IsFinite extends SimpleRule {
  static ruleName = 'isFinite'

  static checkIt(it) {
    return isFinite(it)
  }
}

class MyTypes extends Type {
  static get isFinite() {
    return new this([
      IsFinite.create([], isFinite.config(true))
    ])
  }

  get isFinite() {
    return new this.constructor(
      isFinite.create(
        this.getChecks(),
        isFinite.config(true),
      )
    )
  }
}
// Mixin into TypedProps
Object.defineProperty(Type, 'isFinite', {
    get() {
        return new this(
            IsFinite.create([], IsFinite.config())
        )
    },
})

Object.define(Type.prototype, 'isFinite', {
  get() {
    const checks = IsFinite.create(
      this.getChecks(), IsFinite.config()
    )
    return new this.constructor(checks)
  },
})

Rule types

There is several Rules which are using internally.

UniqRule

This is probably the most useful type. This rule overwrites any other occurencies of the rule in type's rules array. It defines it's own create method which shouldn't be overriden by custom implementation. Mostly because it senseless.

SimpleRule

This rule extends UniqRule and define it's own check method which take all work by creating non-check logic (issue createsion, result comparision). It requires ckeckIt method wich should return boolean value.

class IsArray extends SimpleRule {
    static ruleName = 'isArray'

    static checkIt(it) {
        return Array.isArray(it)
    }
}

API

Checkable()

Checkable is a prototype of TypeProps. It contains only check logic and has no own rules.

Checkable#getChecks()

() -> Checks[]

Returns a list of all defined checks from the TypedProps instance.

check()

(value:*, type:Checkable) -> Array.<Issue>

Validate value to satisfy type requirements. Always produce an Array of Issues.

getRuleParams()

(type:TypedProps, name:string) -> null|*[]

• type - Target type.
• name - Rule name.
• = Returns check params object or null if rule isn't found.

Receive rule params by its' name.

listRules()

(type:TypedProps) -> string[]

• type - Target type.
• = Returns list of rule name in order they were defined.

Rule{}

{
    ruleName: string
    config: (...args[]) -> RuleParams
    create: (options:Array.<Check>, params:RuleParams) -> Array.<Check>
    check: (it:*, params:RuleParams, self:Checkable) -> Array.<Issue>
}

Rule is an object which contains several methods required for TypeProps to work:

Rule.config()

(...args:*) -> Object

Receives arguments from call and transform them to Rule internal representation. This representation could be different from rule to rule.

Rule.create()

(checks:Check[], params:object) -> Checks[]

Adds new check into array of already existing checks. It can overwrite any rule defined earlier when it is necessary. Also it can modify a rule defined previously. But in most cases it just remove previously defined rule and push new one to the end.

Rule.check()

CheckFunction

Receives value and checks that value is satisfying rule logic with params. If it's not, then method returns list of found issues.

Issue{}

{
  path: Array.<String|Number>
  rule: String
  details: Object
}

Check{}

{
  rule: String,
  params: Object
  check: CheckFunction
}

CheckFunction()

(value:*, params:Object) -> Issue[]

Object representing validation failure.

License

Copyright © 2018–2019, Rumkin. Released under MIT License.