Tiny Types

TinyTypes is an npm module that makes it easy for TypeScript and JavaScript projects to give domain meaning to primitive types. It also helps to avoid all sorts of bugs and makes your code easier to refactor. Learn more.

To install the module from npm:

npm install --save tiny-types

API documentation is available at jan-molak.github.io/tiny-types/.

TinyTypes are available as part of the Tidelift Subscription. The maintainers of TinyTypes and thousands of other packages are working with Tidelift to deliver one enterprise subscription that covers all of the open source you use. If you want the flexibility of open source and the confidence of commercial-grade software, this is for you. Learn more.

An int on its own is just a scalar with no meaning. With an object, even a small one, you are giving both the compiler and the programmer additional information about what the value is and why it is being used.

‐ Jeff Bay, Object Calisthenics

To define a single-value TinyType - extend from TinyTypeOf<T>():

import { TinyTypeOf } from 'tiny-types';

class FirstName extends TinyTypeOf<string>() {}
class LastName  extends TinyTypeOf<string>() {}
class Age       extends TinyTypeOf<number>() {}

Every tiny type defined this way has a readonly property value of type T, which you can use to access the wrapped primitive value. For example:

const firstName = new FirstName('Jan');

firstName.value === 'Jan';

Each tiny type object has an equals method, which you can use to compare it by value:

const 
    name1 = new FirstName('Jan'),
    name2 = new FirstName('Jan');

name1.equals(name2) === true; 

An additional feature of tiny types is a built-in toString() method:

const name = new FirstName('Jan');

name.toString() === 'FirstName(value=Jan)';

Which you can override if you want to:

class Timestamp extends TinyTypeOf<Date>() {
    toString() {
        return `Timestamp(value=${this.value.toISOString()})`;
    }
}

const timestamp = new Timestamp(new Date());

timestampt.toString() === 'Timestamp(value=2018-03-12T00:30:00.000Z))'

If the tiny type you want to model has more than one value, or you want to perform additional operations in the constructor, extend from TinyType directly:

import { TinyType } from 'tiny-types';

class Person extends TinyType {
    constructor(public readonly firstName: FirstName,
                public readonly lastName: LastName,
    ) {
        super();
    }
}

You can also mix and match both of the above definition styles:

import { TinyType, TinyTypeOf } from 'tiny-types';

class UserName extends TinyTypeOf<string>() {}

class Timestamp extends TinyTypeOf<Date>() {
    toString() {
        return `Timestamp(value=${this.value.toISOString()})`;
    }
}

abstract class DomainEvent extends TinyTypeOf<Timestamp>() {}

class AccountCreated extends DomainEvent {
    constructor(public readonly username: UserName, timestamp: Timestamp) {
        super(timestamp);
    }
}

const event = new AccountCreated(new UserName('jan-molak'), new Timestamp(new Date()));

Even such complex types still have both the equals and toString methods:

const 
    now = new Date(2018, 2, 12, 0, 30),
    event1 = new AccountCreated(new UserName('jan-molak'), new Timestamp(now)),
    event2 = new AccountCreated(new UserName('jan-molak'), new Timestamp(now));
    
event1.equals(event2) === true;

event1.toString() === 'AccountCreated(username=UserName(value=jan-molak), value=Timestamp(value=2018-03-12T00:30:00.000Z))'

The best way to guarantee runtime correctness of your domain models is to ensure that no tiny type can ever hold invalid data at runtime. This way, when a function receives an instance of a tiny type, it does not need to perform any checks on it and can simply trust that its value is correct. OK, but how do you guarantee that?

Let me show you an example.

Imagine that upon registering a customer on your website you need to ask them their age. How would you model the concept of "age" in your system?

You might consider using a number for this purpose:

const age = 35;

However, this is far from ideal as "age" is not just any number: it can't be negative, it has to be an integer, and it's highly unlikely that your customers would ever be 2⁵³-1 years old.

All that means that there are certain rules that an object representing "age" needs to obey, certain constraints that its value has to meet in order to be considered valid.

You might have already guessed that my recommendation to you would be to define a tiny type representing Age, but not just that. You should also take it a step further and use the ensure function together with other predicates to describe the constraints the underlying value has to meet:

import { TinyType, ensure, isDefined, isInteger, isInRange } from 'tiny-types'

class Age extends TinyType {
  constructor(public readonly value: number) {
    ensure('Age', value, isDefined(), isInteger(), isInRange(0, 125));
  }
} 

With a tiny type defined as per the above code sample you can eliminate entire classes of errors. You also have one place in your system where you define what "age" means.

Every TinyType defines a toJSON() method, which returns a JSON representation of the object. This means that you can use TinyTypes as Data Transfer Objects.

Single-value TinyTypes are serialised to the value itself:

import { TinyTypeOf } from 'tiny-types';

class FirstName extends TinyTypeOf<string>() {}

const firstName = new FirstName('Jan');

firstName.toJSON() === 'Jan'

Complex TinyTypes are serialised recursively:

import { TinyType, TinyTypeOf } from 'tiny-types';

class FirstName extends TinyTypeOf<string>() {}
class LastName extends TinyTypeOf<string>() {}
class Age extends TinyTypeOf<number>() {}
class Person extends TinyType {
    constructor(
        public readonly firstName: FirstName,
        public readonly lastName: LastName,
        public readonly age: Age,
    ) {
        super();
    }
}

const person = new Person(new FirstName('Bruce'), new LastName('Smith'), new Age(55));

person.toJSON() === { firstName: 'Bruce', lastName: 'Smith', age: 55 }

Although you could define standalone de-serialisers, I like to define them as static factory methods on the TinyTypes themselves:

import { TinyTypeOf } from 'tiny-types';

class FirstName extends TinyTypeOf<string>() {
    static fromJSON = (v: string) => new FirstName(v);
}

const firstName = new FirstName('Jan'),

FirstName.fromJSON(firstName.toJSON()).equals(firstName) === true

When working with complex TinyTypes, you can use the (experimental) Serialised interface to reduce the likelihood of your custom fromJSON method being incompatible with toJSON:

import { TinyTypeOf, TinyType, Serialised } from 'tiny-types';

class EmployeeId extends TinyTypeOf<number>() {
    static fromJSON = (id: number) => new EmployeeId(id);
}

class DepartmentId extends TinyTypeOf<string>() {
    static fromJSON = (id: string) => new DepartmentId(id);
}

class Allocation extends TinyType {
    static fromJSON = (o: Serialised<Allocation>) => new Allocation(
        EmployeeId.fromJSON(o.employeeId as number),
        DepartmentId.fromJSON(o.departmentId as string),
    )

    constructor(public readonly employeeId: EmployeeId, public readonly departmentId: DepartmentId) {
        super();
    }
}

This way de-serialising a complex type becomes trivial:

const allocation = new Allocation(new EmployeeId(1), new DepartmentId('engineering'));

const deserialised = Allocation.fromJSON({ departmentId: 'engineering', employeeId: 1 });

allocation.equals(deserialised) === true

Although Serialised is by no means 100% foolproof as it's only limited to checking whether your input JSON has the same fields as the object you're trying to de-serialise, it can at least help you to avoid errors caused by typos.

Do you find TinyTypes useful? Give it a star! ★

Found a bug? Need a feature? Raise an issue or submit a pull request.

Have feedback? Let me know on twitter: @JanMolak

☕ If TinyTypes have made your life a little bit easier and saved at least $5 worth of your time, please consider repaying the favour and buying me a coffee via Github Sponsors. Thanks! 🙏

TinyTypes library is licensed under the Apache-2.0 license.

- Copyright © 2018- Jan Molak