The REST adapter allows your store to communicate with an HTTP server by transmitting JSON via XHR. Most Ember.js apps that consume a JSON API should use the REST adapter.

This adapter is designed around the idea that the JSON exchanged with the server should be conventional.

Success and failure

The REST adapter will consider a success any response with a status code of the 2xx family ("Success"), as well as 304 ("Not Modified"). Any other status code will be considered a failure.

On success, the request promise will be resolved with the full response payload.

Failed responses with status code 422 ("Unprocessable Entity") will be considered "invalid". The response will be discarded, except for the errors key. The request promise will be rejected with a DS.InvalidError. This error object will encapsulate the saved errors value.

Any other status codes will be treated as an "adapter error". The request promise will be rejected, similarly to the "invalid" case, but with an instance of DS.AdapterError instead.

JSON Structure

The REST adapter expects the JSON returned from your server to follow these conventions.

Object Root

The JSON payload should be an object that contains the record inside a root property. For example, in response to a GET request for /posts/1, the JSON should look like this:

{
  "posts": {
    "id": 1,
    "title": "I'm Running to Reform the W3C's Tag",
    "author": "Yehuda Katz"
  }
}

Similarly, in response to a GET request for /posts, the JSON should look like this:

{
  "posts": [
    {
      "id": 1,
      "title": "I'm Running to Reform the W3C's Tag",
      "author": "Yehuda Katz"
    },
    {
      "id": 2,
      "title": "Rails is omakase",
      "author": "D2H"
    }
  ]
}

Note that the object root can be pluralized for both a single-object response and an array response: the REST adapter is not strict on this. Further, if the HTTP server responds to a GET request to /posts/1 (e.g. the response to a findRecord query) with more than one object in the array, Ember Data will only display the object with the matching ID.

Conventional Names

Attribute names in your JSON payload should be the camelCased versions of the attributes in your Ember.js models.

For example, if you have a Person model:

import DS from 'ember-data';

export default DS.Model.extend({
  firstName: DS.attr('string'),
  lastName: DS.attr('string'),
  occupation: DS.attr('string')
});

The JSON returned should look like this:

{
  "people": {
    "id": 5,
    "firstName": "Barack",
    "lastName": "Obama",
    "occupation": "President"
  }
}

Relationships

Relationships are usually represented by ids to the record in the relationship. The related records can then be sideloaded in the response under a key for the type.

{
  "posts": {
    "id": 5,
    "title": "I'm Running to Reform the W3C's Tag",
    "author": "Yehuda Katz",
    "comments": [1, 2]
  },
  "comments": [{
    "id": 1,
    "author": "User 1",
    "message": "First!",
  }, {
    "id": 2,
    "author": "User 2",
    "message": "Good Luck!",
  }]
}

If the records in the relationship are not known when the response is serialized its also possible to represent the relationship as a url using the links key in the response. Ember Data will fetch this url to resolve the relationship when it is accessed for the first time.

{
  "posts": {
    "id": 5,
    "title": "I'm Running to Reform the W3C's Tag",
    "author": "Yehuda Katz",
    "links": {
      "comments": "/posts/5/comments"
    }
  }
}

Errors

If a response is considered a failure, the JSON payload is expected to include a top-level key errors, detailing any specific issues. For example:

{
  "errors": {
    "msg": "Something went wrong"
  }
}

This adapter does not make any assumptions as to the format of the errors object. It will simply be passed along as is, wrapped in an instance of DS.InvalidError or DS.AdapterError. The serializer can interpret it afterwards.

Customization

Endpoint path customization

Endpoint paths can be prefixed with a namespace by setting the namespace property on the adapter:

import DS from 'ember-data';

export default DS.RESTAdapter.extend({
  namespace: 'api/1'
});

Requests for the Person model would now target /api/1/people/1.

Host customization

An adapter can target other hosts by setting the host property.

import DS from 'ember-data';

export default DS.RESTAdapter.extend({
  host: 'https://api.example.com'
});

Headers customization

Some APIs require HTTP headers, e.g. to provide an API key. Arbitrary headers can be set as key/value pairs on the RESTAdapter's headers object and Ember Data will send them along with each ajax request.

import DS from 'ember-data';

export default DS.RESTAdapter.extend({
  headers: {
    "API_KEY": "secret key",
    "ANOTHER_HEADER": "Some header value"
  }
});

headers can also be used as a computed property to support dynamic headers. In the example below, the session object has been injected into an adapter by Ember's container.

import DS from 'ember-data';

export default DS.RESTAdapter.extend({
  headers: Ember.computed('session.authToken', function() {
    return {
      "API_KEY": this.get("session.authToken"),
      "ANOTHER_HEADER": "Some header value"
    };
  })
});

In some cases, your dynamic headers may require data from some object outside of Ember's observer system (for example document.cookie). You can use the volatile function to set the property into a non-cached mode causing the headers to be recomputed with every request.

import DS from 'ember-data';

export default DS.RESTAdapter.extend({
  headers: Ember.computed(function() {
    return {
      "API_KEY": Ember.get(document.cookie.match(/apiKey\=([^;]*)/), "1"),
      "ANOTHER_HEADER": "Some header value"
    };
  }).volatile()
});

Make a request using jQuery.ajax.

Get an object which contains all properties for a request which should be made.

Convert a request object into a hash which can be passed to jQuery.ajax.

Takes a URL, an HTTP method and a hash of data, and makes an HTTP request.

When the server responds with a payload, Ember Data will call into extractSingle or extractArray (depending on whether the original query was for one record or many records).

By default, ajax method has the following behavior:

• It sets the response dataType to "json"
• If the HTTP method is not "GET", it sets the Content-Type to be application/json; charset=utf-8
• If the HTTP method is not "GET", it stringifies the data passed in. The data is the serialized record in the case of a save.
• Registers success and failure handlers.

Called by the store when a newly created record is saved via the save method on a model record instance.

The createRecord method serializes the record and makes an Ajax (HTTP POST) request to a URL computed by buildURL.

See serialize for information on how to customize the serialized form of a record.

Get the data (body or query params) for a request.

Called by the store when a record is deleted.

The deleteRecord method makes an Ajax (HTTP DELETE) request to a URL computed by buildURL.

Called by the store in order to fetch a JSON array for all of the records for a given type.

The findAll method makes an Ajax (HTTP GET) request to a URL computed by buildURL, and returns a promise for the resulting payload.

Called by the store in order to fetch the JSON for the unloaded record in a belongs-to relationship that was originally specified as a URL (inside of links).

For example, if your original payload looks like this:

{
  "person": {
    "id": 1,
    "name": "Tom Dale",
    "links": { "group": "/people/1/group" }
  }
}

This method will be called with the parent record and /people/1/group.

The findBelongsTo method will make an Ajax (HTTP GET) request to the originally specified URL.

The format of your links value will influence the final request URL via the urlPrefix method:

• Links beginning with //, http://, https://, will be used as is, with no further manipulation.

• Links beginning with a single / will have the current adapter's host value prepended to it.

• Links with no beginning / will have a parentURL prepended to it, via the current adapter's buildURL.

Called by the store in order to fetch a JSON array for the unloaded records in a has-many relationship that were originally specified as a URL (inside of links).

For example, if your original payload looks like this:

{
  "post": {
    "id": 1,
    "title": "Rails is omakase",
    "links": { "comments": "/posts/1/comments" }
  }
}

This method will be called with the parent record and /posts/1/comments.

The findHasMany method will make an Ajax (HTTP GET) request to the originally specified URL.

The format of your links value will influence the final request URL via the urlPrefix method:

• Links beginning with //, http://, https://, will be used as is, with no further manipulation.

• Links beginning with a single / will have the current adapter's host value prepended to it.

• Links with no beginning / will have a parentURL prepended to it, via the current adapter's buildURL.

Called by the store in order to fetch several records together if coalesceFindRequests is true

For example, if the original payload looks like:

{
  "id": 1,
  "title": "Rails is omakase",
  "comments": [ 1, 2, 3 ]
}

The IDs will be passed as a URL-encoded Array of IDs, in this form:

ids[]=1&ids[]=2&ids[]=3

Many servers, such as Rails and PHP, will automatically convert this URL-encoded array into an Array for you on the server-side. If you want to encode the IDs, differently, just override this (one-line) method.

The findMany method makes an Ajax (HTTP GET) request to a URL computed by buildURL, and returns a promise for the resulting payload.

Called by the store in order to fetch the JSON for a given type and ID.

The findRecord method makes an Ajax request to a URL computed by buildURL, and returns a promise for the resulting payload.

This method performs an HTTP GET request with the id provided as part of the query string.

Generates a detailed ("friendly") error message, with plenty of information for debugging (good luck!)

Organize records into groups, each of which is to be passed to separate calls to findMany.

This implementation groups together records that have the same base URL but differing ids. For example /comments/1 and /comments/2 will be grouped together because we know findMany can coalesce them together as /comments?ids[]=1&ids[]=2

It also supports urls where ids are passed as a query param, such as /comments?id=1 but not those where there is more than 1 query param such as /comments?id=2&name=David Currently only the query param of id is supported. If you need to support others, please override this or the _stripIDFromURL method.

It does not group records that have differing base urls, such as for example: /posts/1/comments/2 and /posts/2/comments/3

Takes an ajax response, and returns the json payload or an error.

By default this hook just returns the json payload passed to it. You might want to override it in two cases:

1. Your API might return useful results in the response headers. Response headers are passed in as the second argument.

2. Your API might return errors as successful responses with status code 200 and an Errors text or object. You can return a DS.InvalidError or a DS.AdapterError (or a sub class) from this hook and it will automatically reject the promise and put your record into the invalid or error state.

Returning a DS.InvalidError from this method will cause the record to transition into the invalid state and make the errors object available on the record. When returning an DS.InvalidError the store will attempt to normalize the error data returned from the server using the serializer's extractErrors method.

Get the headers for a request.

By default the value of the headers property of the adapter is returned.

Default handleResponse implementation uses this hook to decide if the response is an invalid error.

Default handleResponse implementation uses this hook to decide if the response is a success.

Get the HTTP method for a request.

Called by the store in order to fetch a JSON array for the records that match a particular query.

The query method makes an Ajax (HTTP GET) request to a URL computed by buildURL, and returns a promise for the resulting payload.

The query argument is a simple JavaScript object that will be passed directly to the server as parameters.

Called by the store in order to fetch a JSON object for the record that matches a particular query.

The queryRecord method makes an Ajax (HTTP GET) request to a URL computed by buildURL, and returns a promise for the resulting payload.

The query argument is a simple JavaScript object that will be passed directly to the server as parameters.

By default, the RESTAdapter will send the query params sorted alphabetically to the server.

For example:

  store.query('posts', { sort: 'price', category: 'pets' });

will generate a requests like this /posts?category=pets&sort=price, even if the parameters were specified in a different order.

That way the generated URL will be deterministic and that simplifies caching mechanisms in the backend.

Setting sortQueryParams to a falsey value will respect the original order.

In case you want to sort the query parameters with a different criteria, set sortQueryParams to your custom sort function.

import DS from 'ember-data';

export default DS.RESTAdapter.extend({
  sortQueryParams: function(params) {
    var sortedKeys = Object.keys(params).sort().reverse();
    var len = sortedKeys.length, newParams = {};

    for (var i = 0; i < len; i++) {
      newParams[sortedKeys[i]] = params[sortedKeys[i]];
    }
    return newParams;
  }
});

Called by the store when an existing record is saved via the save method on a model record instance.

The updateRecord method serializes the record and makes an Ajax (HTTP PUT) request to a URL computed by buildURL.

See serialize for information on how to customize the serialized form of a record.

Get the URL for a request.

By default the RESTAdapter will send each find request coming from a store.find or from accessing a relationship separately to the server. If your server supports passing ids as a query string, you can set coalesceFindRequests to true to coalesce all find requests within a single runloop.

For example, if you have an initial payload of:

{
  post: {
    id: 1,
    comments: [1, 2]
  }
}

By default calling post.get('comments') will trigger the following requests(assuming the comments haven't been loaded before):

GET /comments/1
GET /comments/2

If you set coalesceFindRequests to true it will instead trigger the following request:

GET /comments?ids[]=1&ids[]=2

Setting coalesceFindRequests to true also works for store.find requests and belongsTo relationships accessed within the same runloop. If you set coalesceFindRequests: true

store.findRecord('comment', 1);
store.findRecord('comment', 2);

will also send a request to: GET /comments?ids[]=1&ids[]=2

Note: Requests coalescing rely on URL building strategy. So if you override buildURL in your app groupRecordsForFindMany more likely should be overridden as well in order for coalescing to work.

Some APIs require HTTP headers, e.g. to provide an API key. Arbitrary headers can be set as key/value pairs on the RESTAdapter's headers object and Ember Data will send them along with each ajax request. For dynamic headers see headers customization.

import DS from 'ember-data';

export default DS.RESTAdapter.extend({
  headers: {
    "API_KEY": "secret key",
    "ANOTHER_HEADER": "Some header value"
  }
});

An adapter can target other hosts by setting the host property.

import DS from 'ember-data';

export default DS.RESTAdapter.extend({
  host: 'https://api.example.com'
});

Requests for the Post model would now target https://api.example.com/post/.

Endpoint paths can be prefixed with a namespace by setting the namespace property on the adapter:

import DS from 'ember-data';

export default DS.RESTAdapter.extend({
  namespace: 'api/1'
});

Requests for the Post model would now target /api/1/post/.