type-plus
Provides additional types and type adjusted utilities for TypeScript.
Feature Highlights
Installation
npm install type-plus
// or
yarn add type-plus
Runtime type checker
Bringing the power of TypeScript to JavaScript runtime.
const eslintConfig = T.object.create({
env: O.object.create({
es6: O.boolean
}),
parseOptions: O.object.create({
ecmaVersion: O.number.list(3, 5, 6, 7, 8, 9, 10, 11, 12),
sourceType: O.string.list('script', 'module'),
ecmaFeatures: O.object.create({
globalReturn: O.boolean,
impliedStrict: O.boolean,
jsx: O.boolean
})
}),
...
})
const config: unknown = require('.eslintrc.json')
if (T.satisfy(eslintConfig, config)) {
config.parseOptions?.ecmaVersion
}
else {
console.error(T.satify.getReport())
}
All type checker functionalities are exposed as types
(alias to T
).
In addition, O
and R
are exposed to make it easier to access optional types and required types respectively.
Types supported: any
, array
, boolean
, null
, number
, object
, record
, string
, symbol
, tuple
, undefined
, union
, unknown
.
i.e., most of the basic types are supported except bigint
.
It is left out for backward compatibility reasons.
You can use one of the three functions to perform type check:
satisfy(type, subject)
:
A loose type check that permits extra elements in Tuple
and properties in Object
.
conform(type, subject)
:
A strick type check that does not allow extra elements in Tuple
and properties in Object
.
check(options, type, subject)
:
A general form of satisfy()
and conform()
.
Type Assertion
Besides the runtime type checker,
type-plus
also provides a few other ways to do type assertions.
There are actually at least 5 kinds of type assertions:
runtime
: validates during runtime.immediate
: validates at compile time.type guard
: User-defined type guard functions (if (isBool(s))
) introduced in TypeScript 1.6.assertion function
: assertion functions (assertIsBool(a)
) introduced in TypeScript 3.7.logical
: functions or generic types that returns true
or false
type to be used in type level programming.filter
: generic types that returns never
if the test fails.
Here are the type assertions provided in type-plus
.
Use the one that fits your specific needs.
assertType<T>(subject)
:
✔️ immediate
It ensures subject
satisfies T
.
It is similar to const x: T = subject
without introducing unused variable.
You need to specify T
for it to work.
assertType<T>(subject, validator)
:
assertType<T>(subject, Class)
:
✔️ assertion function
, runtime
These overloads of assertType
allows you to specify a validator
.
With these overloads, subject
can be unknown
or any
.
If subject
fails the assertion,
a standard TypeError
will be thrown and provide better error info.
For example:
const s: any = 1
assertType<boolean>(s, s => typeof s === 'boolean')
The message beautification is provided by tersify
.
assertType.isUndefiend(subject)
:
assertType.isNull(subject)
:
assertType.isNumber(subject)
:
assertType.isBoolean(subject)
:
assertType.isTrue(subject)
:
assertType.isFalse(subject)
:
assertType.isString(subject)
:
assertType.isFunction(subject)
:
assertType.isConstructor(subject)
:
assertType.isError(subject)
:
✔️ immediate
, assertion function
, runtime
Compiler and runtime assertion with type narrowing from any
.
They assert the type of subject
is that specific type.
i.e. union type will fail at type level:
const s: number | undefined = undefined
assertType.isUndefined(s)
They accepts any
and will be narrowed to the specific type.
const s: any = undefined
assertType.isUndefined(s)
s
assertType.isNever(subject)
:
✔️ immediate
Check if the subject type is never
.
This function is not very useful in actual code as TypeScript will indicate the error.
But it can be useful when writing tests for type.
This is useful for variable. For type level only check, do the following:
assertType.isTrue(true as Equal<YourType, never>)
assertType.noUndefiend(subject)
:
assertType.noNull(subject)
:
assertType.noNumber(subject)
:
assertType.noBoolean(subject)
:
assertType.noTrue(subject)
:
assertType.noFalse(subject)
:
assertType.noString(subject)
:
assertType.noFunction(subject)
:
assertType.noError(subject)
:
✔️ immediate
, runtime
Compiler and runtime assertion.
Assert subject
type does not contain the specific type.
Work againsts unions.
const s: number | undefined = 1
assertType.noUndefined(s)
They accepts subject
with type any
or unknown
,
assertion will happens in runtime to ensure subject
is the specific type.
isType<T>(subject: T)
:
✔️ immediate
It ensures subject
satisfies T
.
It is identical to assertType<T>(subject: T)
.
You need to specify T
.
isType<T>(subject, validator)
:
isType<T>(subject, Class)
:
✔️ type guard
, runtime
These overloads of isType
allows you to specify a validator
.
With these overloads, subject
can be unknown
or any
.
Equal<A, B>
:
IsEqual<A, B>
:
✔️ logical
Check if A
and B
are the same.
NotEqual<A, B>
:
IsNotEqual<A, B>
:
✔️ logical
Check if A
and B
are not the same.
IsExtend<A, B>
:
IsNotExtend<A, B>
:
✔️ logical
Check if A
extends or not extends B
.
Extendable<A, B>
:
NotExtendable<A, B>
:
✔️ filter
Check if A
extends or not extends B
.
IsAssign<A, B>
:
CanAssign<A, B>
:
✔️ logical
Check if A
can be assigned to B
.
A typical usage is using it with assertType
:
assertType.isFalse(false as CanAssign<boolean, { a: string }>)
assertType.isTrue(true as CanAssign<{ a:string, b:number }, { a: string }>)
canAssign<T>(): (subject) => true
:
✔️ immediate
, logical
Returns a compile time validating function to ensure subject
is assignable to T
.
const isConfig = canAssign<{ a: string }>()
assertType.isTrue(isConfig({ a: 'a' }))
canAssign<T>(false): (subject) => false
:
✔️ immediate
, logical
Returns a compile time validating function to ensure subject
is not assignable to T
.
const notA = canAssign<{ a: string }>(false)
assertType.isTrue(notA({ a: 1 }))
notA({ a: '' })
Nominal Type
TypeScript type system is structural.
In some cases, we want to express a type with nominal behavior.
type-plus
provides two kinds of nominal types: Brand
and Flavor
.
Brand<B, T>
:
brand(type, subject?)
:
Branded nominal type is the stronger nominal type of the two.
It disallows unbranded type assigned to it:
const a = brand('a', { a: 1 })
const b = { a: 1 }
a = b
subject
can be any type, from primitive to strings to objects.
brand(type)
:
If you do not provide subject
, brand(type)
will return a brand creator,
so that you can use it to create multiple branded values:
const nike = brand('nike')
const shirt = nike('shirt')
const socks = nike('socks')
Flavor<F, T>
:
flavor(type, subject?)
:
The key difference between Flavor
and Brand
is that
unflavored type can be assigned to Flavor
:
let f = flavor('orange', 'soda')
f = 'mist'
Also, Brand
of the same name can be assigned to Flavor
,
but Flavor
of the same name cannot be assigned to Brand
.
nominalMatch(a, b)
:
nominalMatch()
can be used to compare Brand
or Flavor
.
const b1 = brand('x', 1)
const b2 = brand('y', 1)
nominalMatch(b1, b2)
Functional Types
ChainFn<T>: T
: chain function that return the input type.
Type Utilities
type-plus
also provides additional type utilities.
These utilities includes utiltiy types and type adjusted functions.
Array function
CommonKeys<A>
: gets common keys inside the records in the array A
.Head<A>
: gets the first entry in the array.IsArray<T>
: logical
predicate for Array
.literalArray(...entries)
: return an array those items are restricted to the provided literals.MapToProp<A, K>
: Map A: Array<E>
to Array<E[K]>
.reduceWhile()
: reduce()
with predicate for early termination.
A simple version of the same function in the ramda
package.Tail<A>
: gets the remaining entries in the array except the first.
Constant Types
JSONTypes
: all JSON compatible types.KeyTypes
: type of all keys.PrimitiveTypes
: all primitive types, including Function
, symbol
, and bigint
.
Object Key functions
filterKey()
: type adjusted filter by key.findKey()
: type adjusted find by key.forEachKey()
: type adjusted for each by key.HasKey<T, K>
: predicate type checking T
has key K
.hasKey()
: function of HasKey
.IsRecord<T>
: logical
predicate for Record
.KeysWithDiffTypes<A, B>
: gets the keys common in A
and B
but with differnt value type.mapKey()
: type adjusted map by key.reduceKey()
: type adjusted reduce by key.someKey()
: type adjusted some by key.SpreadRecord<A, B>
: type for {...a, ...b}
when both a
and b
are Record
.
for array, just do [...A, ...B]
.
Promise function
isPromise<R>(subject: any)
: isPromise()
type guard.PromiseValue<P>
: Gets the type within the Promise.PromiseValueMerge<P1, P2, ...P9>
: Merge the values of multiple promises.mapSeries()
: Similar to bluebird.mapSeries()
but works with async
/await
.
Type manipulation
ANotB<A, B>
: get object with properties in A
and not in B
, including properties with differnt value type.BNotA<A, B>
: flip of ANotB
Except<T, K>
: Deprecated. Same as Omit<T, K>
.ExcludePropType<T, U>
: excludes type U
from properties in T
.KeyofOptional<T>
: keyof
that works with Record<any, any> | undefined
.KnownKeys<T>
: extract known (defined) keys from type T
.LeftJoin<A, B>
: left join A
with B
Omit<T, K>
: From T
, pick a set of properties whose keys are not in the union K
. This is the opposite of Pick<T, K>
.OptionalKeys<T>
: gets keys of optional properties in T
.PartialExcept<T, U>
: Deprecated. Same as PartialOmit<T, U>
.PartialOmit<T, U>
: makes the properties not specified in U
becomes optional.PartialPick<T, U>
: makes the properties specified in U
becomes optional.Pick<T, K>
: pick properties K
from T
. Works with union.RecursivePartial<T>
: make type T
optional recursively.RecursiveRequired<T>
: make type T
required recursively.ReplaceProperty<T, K, V>
: replace property K
in T
with V
.RequiredKeys<T>
: gets keys of required properties in T
.RequiredPick<T, U>
: makes the properties specified in U
becomes required.RequiredExcept<T, U>
: makes the properties not specified in U
becomes required.RecursiveIntersect<T, U>
: intersect type U
onto T
recursively.ValueOf<T>
: type of the value of the properties of T
.Widen<T>
: widen literal types.- PropType: ...no helper type for this. Just do
YourType['propName']
.
Type Predicates
Type predicates are type alias that returns true
or false
.
They can be used to compose complex types.
HasKey<T, K>
: predicate type checking T
has key K
.IsDisjoint<A, B>
: is A
and B
is a disjoint set.
Logical
And<A, B>
: logical AND
.Or<A, B>
: logical OR
.Xor<A, B>
: logical XOR
.Not<X>
: logical NOT
.
Note that these types work correctly with boolean
type.
e.g.:
And<boolean, true> -> boolean
Not<boolean> -> boolean
There is a problem with generic distribution: https://github.com/microsoft/TypeScript/issues/41053
So you may encounter some weird behavior if your logic is complex.
Utility Functions
facade(subject, ...props)
: create a facade of subject
.getField(subject, key, defaultValue)
: get a field from a subject. Works against nullable and optional subject.hasKey()
: function of HasKey
.hasProperty(value, prop)
: assert value
has property prop
. This will pick the correct union type.isConstructor(subject)
: type guard subject
is a constructor.pick(obj, ...props)
: pick properties from obj
.omit(obj, ...props)
: omit properties from obj
.required(...)
: merge options and removing Partial<T>
. From unpartial
requiredDeep(...)
: merge options deeply and removing Partial<T>
. From unpartial
typeOverrideIncompatible<T>()
: override only the incompatiable portion between two types.
type A = {
foo: boolean,
bar: string,
baz: string
}
const overrider = typeOverrideIncompatible<A>()
const source = {
foo: 1,
bar: 'bar',
baz: 'baz'
}
overrider(source, { foo: !!source.foo })
Attribution
Some of the code in this library are created by other people in the TypeScript community.
I merely adding them in and may be making some adjustments.
When ever possible, I add attribution to the person who created those code in the file.
Contribute
npm install
git checkout -b <branch>
npm run watch
git commit -m "<commit message>"
git push