Skip to content

Latest commit

 

History

History
292 lines (246 loc) · 7.6 KB

discriminated-unions.md

File metadata and controls

292 lines (246 loc) · 7.6 KB

Discriminated Union

If you have a class with a literal member then you can use that property to discriminate between union members.

As an example consider the union of a Square and Rectangle, here we have a member kind that exists on both union members and is of a particular literal type:

interface Square {
    kind: "square";
    size: number;
}

interface Rectangle {
    kind: "rectangle";
    width: number;
    height: number;
}
type Shape = Square | Rectangle;

If you use a type guard style check (==, ===, !=, !==) or switch on the discriminant property (here kind) TypeScript will realize that the object must be of the type that has that specific literal and do a type narrowing for you :)

function area(s: Shape) {
    if (s.kind === "square") {
        // Now TypeScript *knows* that `s` must be a square ;)
        // So you can use its members safely :)
        return s.size * s.size;
    }
    else {
        // Wasn't a square? So TypeScript will figure out that it must be a Rectangle ;)
        // So you can use its members safely :)
        return s.width * s.height;
    }
}

Exhaustive Checks

Quite commonly you want to make sure that all members of a union have some code(action) against them.

interface Square {
    kind: "square";
    size: number;
}

interface Rectangle {
    kind: "rectangle";
    width: number;
    height: number;
}

// Someone just added this new `Circle` Type
// We would like to let TypeScript give an error at any place that *needs* to cater for this
interface Circle {
    kind: "circle";
    radius: number;
}

type Shape = Square | Rectangle | Circle;

As an example of where stuff goes bad:

function area(s: Shape) {
    if (s.kind === "square") {
        return s.size * s.size;
    }
    else if (s.kind === "rectangle") {
        return s.width * s.height;
    }
    // Would it be great if you could get TypeScript to give you an error?
}

You can do that by simply adding a fall through and making sure that the inferred type in that block is compatible with the never type. For example if you add the exhaustive check you get a nice error:

function area(s: Shape) {
    if (s.kind === "square") {
        return s.size * s.size;
    }
    else if (s.kind === "rectangle") {
        return s.width * s.height;
    }
    else {
        // ERROR : `Circle` is not assignable to `never`
        const _exhaustiveCheck: never = s;
    }
}

That forces you to handle this new case :

function area(s: Shape) {
    if (s.kind === "square") {
        return s.size * s.size;
    }
    else if (s.kind === "rectangle") {
        return s.width * s.height;
    }
    else if (s.kind === "circle") {
        return Math.PI * (s.radius **2);
    }
    else {
        // Okay once more
        const _exhaustiveCheck: never = s;
    }
}

Switch

TIP: of course you can also do it in a switch statement:

function area(s: Shape) {
    switch (s.kind) {
        case "square": return s.size * s.size;
        case "rectangle": return s.width * s.height;
        case "circle": return Math.PI * s.radius * s.radius;
        default: const _exhaustiveCheck: never = s;
    }
}

strictNullChecks

If using strictNullChecks and doing exhaustive checks, TypeScript might complain "not all code paths return a value". You can silence that by simply returning the _exhaustiveCheck variable (of type never). So:

function area(s: Shape) {
    switch (s.kind) {
        case "square": return s.size * s.size;
        case "rectangle": return s.width * s.height;
        case "circle": return Math.PI * s.radius * s.radius;
        default:
          const _exhaustiveCheck: never = s;
          return _exhaustiveCheck;
    }
}

Throw in exhaustive checks

You can write a function that takes a never (and therefore can only be called with a variable that is inferred as never) and then throws an error if its body ever executes:

function assertNever(x:never): never {
    throw new Error('Unexpected value. Should have been never.');
}

Example use with the area function:

interface Square {
    kind: "square";
    size: number;
}
interface Rectangle {
    kind: "rectangle";
    width: number;
    height: number;
}
type Shape = Square | Rectangle;

function area(s: Shape) {
    switch (s.kind) {
        case "square": return s.size * s.size;
        case "rectangle": return s.width * s.height;
		// If a new case is added at compile time you will get a compile error
		// If a new value appears at runtime you will get a runtime error
        default: return assertNever(s);
    }
}

Retrospective Versioning

Say you have a data structure of the form:

type DTO = {
  name: string
}

And after you have a bunch of DTOs you realize that name was a poor choice. You can add versioning retrospectively by creating a new union with literal number (or string if you want) of DTO. Mark the version 0 as undefined and if you have strictNullChecks enabled it will just work out:

type DTO = 
| { 
   version: undefined, // version 0
   name: string,
 }
| {
   version: 1,
   firstName: string,
   lastName: string, 
}
// Even later 
| {
    version: 2,
    firstName: string,
    middleName: string,
    lastName: string, 
} 
// So on

Example usage of such a DTO:

function printDTO(dto:DTO) {
  if (dto.version == null) {
      console.log(dto.name);
  } else if (dto.version == 1) {
      console.log(dto.firstName,dto.lastName);
  } else if (dto.version == 2) {
      console.log(dto.firstName, dto.middleName, dto.lastName);
  } else {
      const _exhaustiveCheck: never = dto;
  }
}

Redux

A popular library that makes use of this is redux.

Here is the gist of redux with TypeScript type annotations added:

import { createStore } from 'redux'

type Action
  = {
    type: 'INCREMENT'
  }
  | {
    type: 'DECREMENT'
  }

/**
 * This is a reducer, a pure function with (state, action) => state signature.
 * It describes how an action transforms the state into the next state.
 *
 * The shape of the state is up to you: it can be a primitive, an array, an object,
 * or even an Immutable.js data structure. The only important part is that you should
 * not mutate the state object, but return a new object if the state changes.
 *
 * In this example, we use a `switch` statement and strings, but you can use a helper that
 * follows a different convention (such as function maps) if it makes sense for your
 * project.
 */
function counter(state = 0, action: Action) {
  switch (action.type) {
  case 'INCREMENT':
    return state + 1
  case 'DECREMENT':
    return state - 1
  default:
    return state
  }
}

// Create a Redux store holding the state of your app.
// Its API is { subscribe, dispatch, getState }.
let store = createStore(counter)

// You can use subscribe() to update the UI in response to state changes.
// Normally you'd use a view binding library (e.g. React Redux) rather than subscribe() directly.
// However, it can also be handy to persist the current state in the localStorage.

store.subscribe(() =>
  console.log(store.getState())
)

// The only way to mutate the internal state is to dispatch an action.
// The actions can be serialized, logged or stored and later replayed.
store.dispatch({ type: 'INCREMENT' })
// 1
store.dispatch({ type: 'INCREMENT' })
// 2
store.dispatch({ type: 'DECREMENT' })
// 1

Using it with TypeScript gives you safety against typo errors, increased refactor-ability and self documenting code.