With Sequelize you can also specify associations between multiple classes. Doing so will help you to easily access and set those associated objects. The library therefore provides for each defined class different methods, which are explained in the following chapters.

Note: Associations with models that use custom primaryKeys (so not the field 'id') are currently unsupported.

One-To-One associations

One-To-One associations are connecting one source with exactly one target. In order to define a proper database schema, Sequelize utilizes the methodsbelongsToandhasOne. You can use them as follows:

var User = sequelize.define('User', {/* ... */})
var Project = sequelize.define('Project', {/* ... */})
// One-way associations
  In this example hasOne will add an attribute ProjectId to the User model!
  Furthermore, Project.prototype will gain the methods getUser and setUser according
  to the first parameter passed to define. If you have underscore style
  enabled, the added attribute will be project_id instead of ProjectId.
  You can also define the foreign key, e.g. if you already have an existing
  database and want to work on it:
Project.hasOne(User, { foreignKey: 'initiator_id' })
  Because Sequelize will use the model's name (first parameter of define) for
  the accessor methods, it is also possible to pass a special option to hasOne:
Project.hasOne(User, { as: 'Initiator' })
// Now you will get Project#getInitiator and Project#setInitiator
// Or let's define some self references
var Person = sequelize.define('Person', { /* ... */})
Person.hasOne(Person, {as: 'Father'})
// this will add the attribute FatherId to Person
// also possible:
Person.hasOne(Person, {as: 'Father', foreignKey: 'DadId'})
// this will add the attribute DadId to Person
// In both cases you will be able to do:
// If you need to join a table twice you can double join the same table
  .hasOne(Game, {foreignKey : 'homeTeamId'});
  .hasOne(Game, {foreignKey : 'awayTeamId'});
// Since v1.3.0 you can also chain associations:

To get the association working the other way around (so fromUsertoProject), it's necessary to do this:

var User = sequelize.define('User', {/* ... */})
var Project = sequelize.define('Project', {/* ... */})
// One-way back associations
  In this example belongsTo will add an attribute UserId to the Project model!
  That's the only difference to hasMany. Self references are working the very same way!

One-To-Many associations

One-To-Many associations are connecting one source with multiple targets. The targets however are again connected to exactly one specific source.

var User = sequelize.define('User', {/* ... */})
var Project = sequelize.define('Project', {/* ... */})
// OK. Now things get more complicated (not really visible to the user :)).
// First let's define a hasMany association
Project.hasMany(User, {as: 'Workers'})
  This will add the attribute ProjectId or project_id to User.
  Instances of Project will get the accessors getWorkers and setWorkers.
  We could just leave it the way it is and let it be a one-way association.
  But we want more! Let's define the other way around:

Many-To-Many associations

Many-To-Many associations are used to connect sources with multiple targets. Furthermore the targets can also have connections to multiple sources.

// again the Project association to User
Project.hasMany(User, { as: 'Workers' })
// now comes the association between User and Project
  This will remove the attribute ProjectId (or project_id) from User and create
  a new model called ProjectsUsers with the equivalent foreign keys ProjectId
  (or project_id) and UserId (or user_id). If the attributes are camelcase or
  not depends on the Model it represents.
  Now you can use Project#getWorkers, Project#setWorkers, User#getTasks and
// Of course you can also define self references with hasMany:
Person.hasMany(Person, { as: 'Children' })
// This will create the table ChildrenPersons which stores the ids of the objects.
// Since v1.5.0 you can also reference the same Model without creating a junction
// table (but only if each object will have just one 'parent'). If you need that,
// use the option foreignKey and set through to null
Person.hasMany(Person, { as: 'Children', foreignKey: 'ParentId', through: null })
// You can also use a predefined junction table using the option through:
Project.hasMany(User, {through: 'project_has_users'})
User.hasMany(Project, {through: 'project_has_users'})

If you want additional attributes in your join table, you can define a model for the join table in sequelize, before you define the association, and then tell sequelize that it should use that model for joining, instead of creating a new one:

User = sequelize.define('User', {})
Project = sequelize.define('Project', {})
UserProjects = sequelize.define('UserProjects', {
    status: DataTypes.STRING
User.hasMany(Project, { through: UserProjects })
Project.hasMany(User, { through: UserProjects })

The code above will add ProjectId and UserId to the UserProjects table, andremove any previsouly defined primary key attribute, - the table will be uniquely identified by the combination of the keys of the two tables, and there is no reason to have other PK columns.

Associating objects

Because Sequelize is doing a lot of magic, you have to callSequelize#syncafter setting the associations! Doing so will allow you the following:

// save them... and then:
project.setTasks([task1, task2]).success(function() {
  // saved!
// ok now they are save... how do I get them later on?
project.getTasks().success(function(associatedTasks) {
  // associatedTasks is an array of tasks
// You can also pass filters to the getter method.
// They are equal to the options you can pass to a usual finder method.
project.getTasks({ where: 'id > 10' }).success(function(tasks) {
  // tasks with an id greater than 10 :)
// You can also only retrieve certain fields of a associated object.
// This example will retrieve the attibutes "title" and "id"
project.getTasks({attributes: ['title']}).success(function(tasks) {
  // tasks with an id greater than 10 :)

To remove created associations you can just call the set method without a specific id:

// remove the association with task1
project.setTasks([task2]).success(function(associatedTasks) {
  // you will get task2 only
// remove 'em all
project.setTasks([]).success(function(associatedTasks) {
  // you will get an empty array
// or remove 'em more directly
project.removeTask(task1).success(function() {
  // it's gone
// and add 'em again
project.addTask(task1).success(function() {
  // it's back again

You can of course also do it vice versa:

// project is associated with task1 and task2
task2.setProject(null).success(function() {
  // and it's gone

For hasOne/belongsTo its basically the same:

Task.hasOne(User, {as: "Author"})

Adding associations to a relation with a custom join table can be done in two ways (continuing with the associations defined in the previous chapter):

// Either by adding a property with the name of the join table model to the object, before creating the association
project.UserProjects = {
  status: 'active'
// Or by providing a second argument when adding the association, containing the data that should go in the join table
u.addProject(project, { status: 'active' })
// When associating multiple objects, you can combine the two options above. In this case the second argument
// will be treated as a defaults object, that will be used if no data is provided
project1.UserProjects = {
    status: 'inactive'
u.setProjects([project1, project2], { status: 'active' })
// The code above will record inactive for project one, and active for project two in the join table

When getting data on an association that has a custom join table, the data from the join table will be returned as a DAO instance:

u.getProjects().success(function(projects) {
  var project = projects[0]
  if (project.UserProjects.status === 'active') {
    // .. do magic
    // since this is a real DAO instance, you can save it directly after you are done doing magic

If you only need some of the attributes from the join table, you can provide an array with the attributes you want:

// This will select only name from the Projects table, and only status from the UserProjects table
user.getProjects({ attributes: ['name'], joinTableAttributes: ['status']})

Check associations

Sequelizev1.5.0introduced methods which allows you, to check if an object is already associated with another one (N:M only). Here is how you'd do it:

// check if an object is one of associated ones:
Project.create({ /* */ }).success(function(project) {
  User.create({ /* */ }).success(function(user) {
    project.hasUser(user).success(function(result) {
      // result would be false
      project.addUser(user).success(function() {
        project.hasUser(user).success(function(result) {
          // result would be true
// check if all associated objects are as expected:
// let's assume we have already a project and two users
project.setUsers([user1, user2]).success(function() {
  project.hasUsers([user1]).success(function(result) {
    // result would be false
    project.hasUsers([user1, user2]).success(function(result) {
      // result would be true

Foreign Keys

Foreign key constraints

// First setup our models
var Task = this.sequelize.define('Task', { title: Sequelize.STRING })
  , User = this.sequelize.define('User', { username: Sequelize.STRING })
// Then our associations with a constraint
User.hasMany(Task, {onDelete: 'restrict'})

The same works forhasOne()andbelongsTo(). Valid options are:

  • {onDelete: 'restrict|cascade'}
  • {onUpdate: 'restrict|cascade'}
  • {foreignKeyConstraint: true}- Set to true to gain aREFERENCESdeclaration without eitheronDeleteoronUpdate(that is,foreignKeyConstraintis implied ifonDeleteoronUpdateare set). This may be a helpful performance optimization in some cases.

Implementation notes:

  • Enabling constraints is opt-in: they are only created if one or more of the options above are used when defining an association.
  • MySQL (with InnoDB tables) and Postgres support foreign key references by default. SQLite does not unlessPRAGMA FOREIGN_KEYS=ONis issued. We do so when opening the database connection, unless explicitly disabled with a global option, to get parity across implementations.Note:enabling this doesn't make any difference unless there are actual constraints in place.
  • Only simple foreign keys are supported (because associations only support simple keys). This is the "80%" use case of course. Setting one of foreign key options in a situation where there is more than one primary key defined will cause the option to be ignored.
  • SQL allows two ways to define foreign keys: "inline" (someId INTEGER REFERENCES someTable(id) ON DELETE CASCADE) or "standalone" (FOREIGN KEY(someId) REFERENCES someTable(id) ON DELETE CASCADEplaced as a separate clause inside aCREATE TABLEstatement). Since associations in sequelize create foreign key attributes the "inline" syntax is used, andattributesToSQLis taught how to add the relevant suffix to any "raw" attribute with the relevant meta-data attached. This works for Postgres and SQLite (MySQL ignores this syntax) requiring the "standalone" approach. For MySQL, we move the declaration to the end with a bit of string manipulation. This is analogous to howPRIMARY KEYis handled and allows this to be done without major refactoring.
  • If we have foreign key constraints, the order in which tables are created matters: iffoohas a foreign key tobarwith a constraint, thenbarhas to exist beforefoocan be created. To make sure this happens, we use a topological sort of relationships (via theToposort-classmodule) to sequence calls toCREATE TABLEinsync(). This also necessitatessync()being serialized, but given it's an "on startup" operation that shouldn't be too much of an issue.
  • A similar concern happens withdropAllTables(), but here we don't have enough information to sort the list. Instead, we do one of two things: for SQLite and MySQL, we temporarily disable constraint checking. For Postgres, we useDROP TABLE ... CASCADEto drop relevant constraints when required. (MySQL and SQLite only support the former and Postgres only supports the latter). This is blunt, but OK given that the function is attempting to dropallthe tables.
  • For other calls todropTable()the caller is expected to sequence calls appropriately, or wrap the call indisableForeignKeyConstraints()andenableForeignKeyConstraints()(MySQL and SQLite; no-ops in Postgres) and/or pass {cascade: true} in options (Postgres; no-op in MySQL and SQLite).

Enforcing a foreign key reference

var Series, Trainer, Video
// Series has a foreign reference key after we call Trainer.hasMany(series)
Series = sequelize.define('Series', {
  title:        DataTypes.STRING,
  sub_title:    DataTypes.STRING,
  description:  DataTypes.TEXT,
  // Set FK relationship (hasMany) with `Trainer`
  trainer_id: {
    type: DataTypes.INTEGER,
    references: "Trainer",
    referencesKey: "id"
Trainer = sequelize.define('Trainer', {
  first_name: DataTypes.STRING,
  last_name:  DataTypes.STRING
// Video has a foreign reference key after we call Series.hasOne(Video)...
Video = sequelize.define('Video', {
  title:        DataTypes.STRING,
  sequence:     DataTypes.INTEGER,
  description:  DataTypes.TEXT,
  // set relationship (hasOne) with `Series`
  series_id: {
    type: DataTypes.INTEGER,
    references: "Series",
    referencesKey: "id"
© Sascha Depold, et al. 2006 - 2014