HomeArticles

TypeScript: Built-in generic types

Stefan Baumgartner

Stefan on Mastodon

More on TypeScript

TypeScript comes with a ton of built in generic types that ease your development workflow. Here’s a list of all built-in generic types, with examples!

Note: This list is probably (most definitely) not complete. If you miss something and want to have it added, please reach out to me via Twitter

See more on:

Readonly #

const in JavaScript is tricky, because it only means you can’t reassign any other values to this name. It allows however for changing properties of an object. The Readonly built-in type helps:

type Point = {
x: number,
y: number
};

const p: Readonly<Point> = {
x: 4,
y: 2
};

p.x = 5; // ⚡️ compile error!

ReadonlyArray #

The generic, built-in type ReadonlyArray allows us to throw errors once we use an array function that mutates the original array. See for yourself:

const values: ReadonlyArray<number> = [1, 2, 3, 4, 5];

values.push(6); // ⚡️ compile error! This mutates the array
values.filter(x => x > 4); // ✔︎ compiles! filter returns a new array

Handy if you want to keep an immutable array!

ReturnType #

This built-in type gets you the return type of any function.

type Point = {
x: number,
y: number
}

function generateSquare(bottomLeft: Point, topRight: Point) {
return {
bottomLeft,
bottomRight: {
x: topRight.x,
y: bottomLeft.y,
},
topLeft: {
x: bottomLeft.x,
y: topRight.y
},
topRight
}
}

type Square = ReturnType<typeof generateSquare>;
// here's the magic! I can use this return type now everywhere
// in my code

function areaOfSquare(square: Square): number {
//do something
return result;
}

You can also access functions inside classes:

class Square {
generate(bottomLeft, topRight) {
return {
bottomLeft,
bottomRight: {
x: topRight.x,
y: bottomLeft.y,
},
topLeft: {
x: bottomLeft.x,
y: topRight.y
},
topRight
}
}
}

type TSquare = ReturnType<Square['generate']>;
declare let result: TSquare;

Partial #

Partial<T> is a beauty. It takes all properties from one type, and makes them optional. What is it good for? Think about having a set of default options, and you want to override just parts and pieces from them. Partial<T> helps you getting autocomplete and type-checking for that case:

const defaultOptions = {
directory: '.',
incremental: true,
filePattern: '**/*',
}

function start(options: Partial<typeof defaultOptions>) {
const allOptions = Object.assign({}, defaultOptions, options);
console.log(allOptions);
}

start({
incremental: false, // Awesome! Typechecks!
});

start({
flatten: true // ⚡️ Error! This property has nothing to do with our options
});

Required #

Required<T> is the opposite to Partial<T>. Where Partial<T> makes every property optional, required makes ever property necessary.

type Container = {
width: number,
height: number,
children?: Container[]
}

function getChildrenArea(parent: Required<Container>) {
let sum = 0;
for (let child of parent.children) {
sum = sum + (child.width * child.height)
}
return sum;
}

const mainContainer: Container = {
width: 200,
height: 100
}

getChildrenArea(mainContainer); // ⚡️ Error: Needs children

NonNullable #

NonNullable<T> helps you to ensure you don’t pass null or undefined to your functions. This complements the strictNullChecks compiler flag, so make sure you activate it.

function print<T>(x: NonNullable<T>) {
console.log(x.toString());
}

print('Hello');
print(2);
print(null); // ⚡️ Error
print(undefined); // ⚡️ Error

Pick #

With Pick<T, K extends keyof T> you can create a new type from an existing object, by only using a selected list of properties. Lodash’s eponymous pick function is a good example of its usage:

/**
* The pick function is generic as well. It has two generic types:
* - T ... the type of the object we want to pick props from
* - K ... a subset of all keys in T
*
* Our method signature takes an object of type T, the other parameters
* are collected in an array of type K.
*
* The return type is a subset of keys of T.
*/

declare function pick<T, K extends keyof T>(obj: T, ...propsToPick: K[]): Pick<T, K>;

const point3D = {
x: 2,
y: 0,
z: 4
}

const point2D = pick(point3D, 'x', 'y'); // returns a type { x: number, y: number }

This one is especially useful when used with other generic types, e.g. Exclude.

Record #

Record<K, T> is funny. With it you can say that *every key K should be of type T. With it you can do things like

type Person = Record<'firstName' | 'lastName', string>

which is the same as { firstName: string, lastName: string }. Or, something like

type MetaInfo = {
title: string,
url: string
}

type Episodes = Record<string, MetaInfo>

Which allows an object with any key possible, but values of type MetaInfo. This is very much alike to { [k: string]: MetaInfo }.

So far, so good. But why have this generic Record type if we can achieve similar, if not the same results with other methods? Record helps when you deal with other generic types. Let’s look at that example: We can create a function that transforms all values of an object to a string representation:

// The implementation is somewhere else. It converts all values to strings.
declare function allToString<T>(obj: T): Record<keyof T, string>;

const person = {
firstName: 'Stefan',
lastName: 'Baumgartner',
age: Number.MAX_VALUE
}

// all properites in strPerson are now strings
const strPerson = allToString(person);

Check it out here;

Extract #

Extract<T, K> extracts all types from T that are assignable to K. Let’s say you have two different types of shape. Circles and rectangles. They look something like that:

const rect = {
width: 200,
height: 100,
area() {
return this.width * this.height;
},
perimeter() {
return this.width * 2 + this.height * 2;
}
}

const circle = {
r: 50,
area() {
return this.r * this.r * Math.PI;
},
perimeter() {
return 2 * this.r * Math.PI;
}
}

Their types are

type Rect = typeof rect;
type Circle = typeof circle;

They have something in common: They both have the same methods for area and perimeter. These objects might change over time. We still want to make sure that we only access methods that are available in both of them. With Extract, we can get the keys of those functions:

// extracts: 'area' | 'perimeter'
type ShapeFunctionProps = Extract<keyof Rect, keyof Circle>

To create a type that has access to all those functions again, we can use the Record type from earlier on:

type ShapeFunctions = Record<ShapeFunctionProps, () => number>

Now we have this extra type safety in functions that may apply to all of those objects:

declare function extensiveMath<T extends ShapeFunctions>(obj: T)

extensiveMath(rect); // ✅ compiles
extensiveMath(circle); // ✅ compiles

Exclude #

Exclude<T, K> excludes all types from T that are assignable to K. It’s like the opposite of Extract. Let’s take the example from earlier:

type Rect = typeof rect;
type Circle = typeof circle;

// only has 'width' | 'height';
type RectOnlyProperties = Exclude<keyof Rect, keyof Circle>;

// An object of type { width: number, height: number }
type RectOnly = Pick<Rect, RectOnlyProperties>;

declare function area<T extends RectOnly>(obj: T)

area({
width: 100,
height: 100
}) // ✅ compiles;

Omit #

Since version 3.5, TypeScript includes the Omit<T, K> helper type. Omit<T, K> is a shortcut for Pick<T, Exclude<keyof T, K>>, but since it was used so often, the team decided to add it:

type Person = {
name: string;
age: number;
location: string;
};

type Cosmopolitan = Omit<Person, 'location'>;

// equivalent to
type Cosmopolitan = {
name: string;
age: number;
};

Bottom line #

That’s a couple of generic built-in types that might be useful for different cases. I think the combination of all those generic types is super helpful if you want to strengthen your library or application without interfering too much. Especially when you generate type definitions for methods that allow different parameter types (as it happens often in JavaScript), they might be much more useful than defining every possible type outcome on its own.

Again, I’m sure I missed a ton. So if you have any generic built-in type that you want to have convered here, let me know.

Update: My buddy Martin Hochel has a couple of mapped types that build on top of those. They are pretty good, so check them out

More articles on TypeScript

Stay up to date!

3-4 updates per month, no tracking, spam-free, hand-crafted. Our newsletter gives you links, updates on oida.dev, conference talks, coding soundtracks, and much more.