Skip to main content
Version: Next

Defining Entities via EntitySchema

With EntitySchema helper you define the schema programmatically.

export interface Book extends BaseEntity {
title: string;
author: Author;
publisher: Publisher;
tags: Collection<BookTag>;
export const schema = new EntitySchema<Book, BaseEntity>({
name: 'Book',
extends: 'BaseEntity',
properties: {
title: { type: 'string' },
author: { reference: 'm:1', entity: 'Author', inversedBy: 'books' },
publisher: { reference: 'm:1', entity: 'Publisher', inversedBy: 'books' },
tags: { reference: 'm:n', entity: 'BookTag', inversedBy: 'books', fixedOrder: true },

When creating new entity instances, you will need to use em.create() method that will create instance of internally created class.

const repo = em.getRepository<Author>('Author');
const author = repo.create('Author', { name: 'name', email: 'email' }); // instance of internal Author class
await repo.persistAndFlush(author);

Using this approach, metadata caching is automatically disabled as it is not needed.

Using DTO class#

It is very common to define a DTO (Data Transfer Object) to validate incoming request bodies and pass the request body data on to the other parts of your application. If you pass the DTO directly to the create method it could lead to unexpected results. The data for the create method should be provided as a POJO (Plain Old JS Object {}).

You can achieve this by letting your DTO class extend the PlainObject class. This way MikroORM knows it should be treated as such.

Using custom entity classes#

You can optionally use custom class for entity instances.

export class Author extends BaseEntity {
name: string;
email: string;
age?: number;
termsAccepted?: boolean;
identities?: string[];
born?: Date;
books = new Collection<Book>(this);
favouriteBook?: Book;
version?: number;
constructor(name: string, email: string) { = name; = email;
export const schema = new EntitySchema<Author, BaseEntity>({
class: Author,
properties: {
name: { type: 'string' },
email: { type: 'string', unique: true },
age: { type: 'number', nullable: true },
termsAccepted: { type: 'boolean', default: 0, onCreate: () => false },
identities: { type: 'string[]', nullable: true },
born: { type: DateType, nullable: true, length: 3 },
books: { reference: '1:m', entity: () => 'Book', mappedBy: book => },
favouriteBook: { reference: 'm:1', type: 'Book' },
version: { type: 'number', persist: false },

Then you can use the entity class as usual:

const repo = em.getRepository(Author);
const author = new Author('name', 'email');
await repo.persistAndFlush(author);

Using BaseEntity#

Do not forget that base entities needs to be discovered just like normal entities.

export interface BaseEntity {
id: number;
createdAt: Date;
updatedAt: Date;
export const schema = new EntitySchema<BaseEntity>({
name: 'BaseEntity',
abstract: true,
properties: {
id: { type: 'number', primary: true },
createdAt: { type: 'Date', onCreate: () => new Date(), nullable: true },
updatedAt: { type: 'Date', onCreate: () => new Date(), onUpdate: () => new Date(), nullable: true },

Configuration Reference#

The parameter of EntitySchema requires to provide either name or class parameters. When using class, extends will be automatically inferred. You can optionally pass these additional parameters:

name: string;
class: Constructor<T>;
extends: string;
tableName: string; // alias for `collection: string`
properties: { [K in keyof T & string]: EntityProperty<T[K]> };
indexes: { properties: string | string[]; name?: string; type?: string }[];
uniques: { properties: string | string[]; name?: string }[];
customRepository: () => Constructor<EntityRepository<T>>;
hooks: Partial<Record<HookType, (string & keyof T)[]>>;
abstract: boolean;

Every property then needs to contain a type specification - one of type/customType/entity. Here are some examples of various property types:

export enum MyEnum {
LOCAL = 'local',
GLOBAL = 'global',
export const schema = new EntitySchema<FooBar>({
name: 'FooBar',
tableName: 'tbl_foo_bar',
indexes: [{ name: 'idx1', properties: 'name' }],
uniques: [{ name: 'unq1', properties: ['name', 'email'] }],
customRepository: () => FooBarRepository,
properties: {
id: { type: 'number', primary: true },
name: { type: 'string' },
baz: { reference: '1:1', entity: 'FooBaz', orphanRemoval: true, nullable: true },
fooBar: { reference: '1:1', entity: 'FooBar', nullable: true },
publisher: { reference: 'm:1', entity: 'Publisher', inversedBy: 'books' },
books: { reference: '1:m', entity: () => 'Book', mappedBy: book => },
tags: { reference: 'm:n', entity: 'BookTag', inversedBy: 'books', fixedOrder: true },
version: { type: 'Date', version: true, length: 0 },
type: { enum: true, items: () => MyEnum, default: MyEnum.LOCAL },

As a value for type you can also use one of String/Number/Boolean/Date.

MongoDB example#

export class BookTag {
_id!: ObjectId;
id!: string;
name: string;
books = new Collection<Book>(this);
constructor(name: string) { = name;
export const schema = new EntitySchema<BookTag>({
class: BookTag,
properties: {
_id: { type: 'ObjectId', primary: true },
id: { type: 'string', serializedPrimaryKey: true },
name: { type: 'string' },
books: { reference: 'm:n', entity: () => Book, mappedBy: book => book.tags },
Last updated on by Renovate Bot