DS.RESTSerializer

Normally, applications will use the RESTSerializer by implementing the normalize method.

This allows you to do whatever kind of munging you need, and is especially useful if your server is inconsistent and you need to do munging differently for many different kinds of responses.

See the normalize documentation for more information.

Across the Board Normalization

There are also a number of hooks that you might find useful to define across-the-board rules for your payload. These rules will be useful if your server is consistent, or if you're building an adapter for an infrastructure service, like Firebase, and want to encode service conventions.

For example, if all of your keys are underscored and all-caps, but otherwise consistent with the names you use in your models, you can implement across-the-board rules for how to convert an attribute name in your model to a key in your JSON.

import DS from 'ember-data';

export default DS.RESTSerializer.extend({
  keyForAttribute(attr, method) {
    return Ember.String.underscore(attr).toUpperCase();
  }
});

You can also implement keyForRelationship, which takes the name of the relationship as the first parameter, the kind of relationship (hasMany or belongsTo) as the second parameter, and the method (serialize or deserialize) as the third parameter.

Normalizes an array of resource payloads and returns a JSON-API Document with primary data and, if any, included data as { data, included }.

You can use this method to customize how a polymorphic relationship should be extracted.

keyForPolymorphicType can be used to define a custom key when serializing and deserializing a polymorphic type. By default, the returned key is ${key}Type.

Example

 import DS from 'ember-data';

 export default DS.RESTSerializer.extend({
   keyForPolymorphicType(key, relationship) {
     var relationshipKey = this.keyForRelationship(key);

     return 'type-' + relationshipKey;
   }
 });

This method is used to convert each JSON root key in the payload into a modelName that it can use to look up the appropriate model for that part of the payload.

For example, your server may send a model name that does not correspond with the name of the model in your app. Let's take a look at an example model, and an example payload:

import DS from 'ember-data';

export default DS.Model.extend({
});

  {
    "blog/post": {
      "id": "1
    }
  }

Ember Data is going to normalize the payload's root key for the modelName. As a result, it will try to look up the "blog/post" model. Since we don't have a model called "blog/post" (or a file called app/models/blog/post.js in ember-cli), Ember Data will throw an error because it cannot find the "blog/post" model.

Since we want to remove this namespace, we can define a serializer for the application that will remove "blog/" from the payload key whenver it's encountered by Ember Data:

import DS from 'ember-data';

export default DS.RESTSerializer.extend({
  modelNameFromPayloadKey(payloadKey) {
    if (payloadKey === 'blog/post') {
      return this._super(payloadKey.replace('blog/', ''));
    } else {
     return this._super(payloadKey);
    }
  }
});

After refreshing, Ember Data will appropriately look up the "post" model.

By default the modelName for a model is its name in dasherized form. This means that a payload key like "blogPost" would be normalized to "blog-post" when Ember Data looks up the model. Usually, Ember Data can use the correct inflection to do this for you. Most of the time, you won't need to override modelNameFromPayloadKey for this purpose.

modelNameFromPayloadType can be used to change the mapping for a DS model name, taken from the value in the payload.

Say your API namespaces the type of a model and returns the following payload for the post model, which has a polymorphic user relationship:

// GET /api/posts/1
{
  "post": {
    "id": 1,
    "user": 1,
    "userType: "api::v1::administrator"
  }
}

By overwriting modelNameFromPayloadType you can specify that the administrator model should be used:

import DS from 'ember-data';

export default DS.RESTSerializer.extend({
  modelNameFromPayloadType(payloadType) {
    return payloadType.replace('api::v1::', '');
  }
});

By default the modelName for a model is its name in dasherized form. Usually, Ember Data can use the correct inflection to do this for you. Most of the time, you won't need to override modelNameFromPayloadType for this purpose.

Also take a look at payloadTypeFromModelName to customize how the type of a record should be serialized.

Normalizes a part of the JSON payload returned by the server. You should override this method, munge the hash and call super if you have generic normalization to do.

It takes the type of the record that is being normalized (as a DS.Model class), the property where the hash was originally found, and the hash to normalize.

For example, if you have a payload that looks like this:

{
  "post": {
    "id": 1,
    "title": "Rails is omakase",
    "comments": [ 1, 2 ]
  },
  "comments": [{
    "id": 1,
    "body": "FIRST"
  }, {
    "id": 2,
    "body": "Rails is unagi"
  }]
}

The normalize method will be called three times:

• With App.Post, "posts" and { id: 1, title: "Rails is omakase", ... }
• With App.Comment, "comments" and { id: 1, body: "FIRST" }
• With App.Comment, "comments" and { id: 2, body: "Rails is unagi" }

You can use this method, for example, to normalize underscored keys to camelized or other general-purpose normalizations. You will only need to implement normalize and manipulate the payload as desired.

For example, if the IDs under "comments" are provided as _id instead of id, you can specify how to normalize just the comments:

import DS from 'ember-data';

export default DS.RESTSerializer.extend({
  normalize(model, hash, prop) {
    if (prop === 'comments') {
      hash.id = hash._id;
      delete hash._id;
    }

    return this._super(...arguments);
  }
});

On each call to the normalize method, the third parameter (prop) is always one of the keys that were in the original payload or in the result of another normalization as normalizeResponse.

You can use payloadKeyFromModelName to override the root key for an outgoing request. By default, the RESTSerializer returns a camelized version of the model's name.

For a model called TacoParty, its modelName would be the string taco-party. The RESTSerializer will send it to the server with tacoParty as the root key in the JSON payload:

{
  "tacoParty": {
    "id": "1",
    "location": "Matthew Beale's House"
  }
}

For example, your server may expect dasherized root objects:

import DS from 'ember-data';

export default DS.RESTSerializer.extend({
  payloadKeyFromModelName(modelName) {
    return Ember.String.dasherize(modelName);
  }
});

Given a TacoParty model, calling save on it would produce an outgoing request like:

{
  "taco-party": {
    "id": "1",
    "location": "Matthew Beale's House"
  }
}

payloadTypeFromModelName can be used to change the mapping for the type in the payload, taken from the model name.

Say your API namespaces the type of a model and expects the following payload when you update the post model, which has a polymorphic user relationship:

// POST /api/posts/1
{
  "post": {
    "id": 1,
    "user": 1,
    "userType": "api::v1::administrator"
  }
}

By overwriting payloadTypeFromModelName you can specify that the namespaces model name for the administrator should be used:

import DS from 'ember-data';

export default DS.RESTSerializer.extend({
  payloadTypeFromModelName(modelName) {
    return 'api::v1::' + modelName;
  }
});

By default the payload type is the camelized model name. Usually, Ember Data can use the correct inflection to do this for you. Most of the time, you won't need to override payloadTypeFromModelName for this purpose.

Also take a look at modelNameFromPayloadType to customize how the model name from should be mapped from the payload.

This method allows you to push a payload containing top-level collections of records organized per type.

{
  "posts": [{
    "id": "1",
    "title": "Rails is omakase",
    "author", "1",
    "comments": [ "1" ]
  }],
  "comments": [{
    "id": "1",
    "body": "FIRST"
  }],
  "users": [{
    "id": "1",
    "name": "@d2h"
  }]
}

It will first normalize the payload, so you can use this to push in data streaming in from your server structured the same way that fetches and saves are structured.

Called when a record is saved in order to convert the record into JSON.

By default, it creates a JSON object with a key for each attribute and belongsTo relationship.

For example, consider this model:

import DS from 'ember-data';

export default DS.Model.extend({
  title: DS.attr(),
  body: DS.attr(),

  author: DS.belongsTo('user')
});

The default serialization would create a JSON object like:

{
  "title": "Rails is unagi",
  "body": "Rails? Omakase? O_O",
  "author": 12
}

By default, attributes are passed through as-is, unless you specified an attribute type (DS.attr('date')). If you specify a transform, the JavaScript value will be serialized when inserted into the JSON hash.

By default, belongs-to relationships are converted into IDs when inserted into the JSON hash.

IDs

serialize takes an options hash with a single option: includeId. If this option is true, serialize will, by default include the ID in the JSON object it builds.

The adapter passes in includeId: true when serializing a record for createRecord, but not for updateRecord.

Customization

Your server may expect a different JSON format than the built-in serialization format.

In that case, you can implement serialize yourself and return a JSON hash of your choosing.

import DS from 'ember-data';

export default DS.RESTSerializer.extend({
  serialize(snapshot, options) {
    var json = {
      POST_TTL: snapshot.attr('title'),
      POST_BDY: snapshot.attr('body'),
      POST_CMS: snapshot.hasMany('comments', { ids: true })
    };

    if (options.includeId) {
      json.POST_ID_ = snapshot.id;
    }

    return json;
  }
});

Customizing an App-Wide Serializer

If you want to define a serializer for your entire application, you'll probably want to use eachAttribute and eachRelationship on the record.

import DS from 'ember-data';

export default DS.RESTSerializer.extend({
  serialize(snapshot, options) {
    var json = {};

    snapshot.eachAttribute(function(name) {
      json[serverAttributeName(name)] = snapshot.attr(name);
    });

    snapshot.eachRelationship(function(name, relationship) {
      if (relationship.kind === 'hasMany') {
        json[serverHasManyName(name)] = snapshot.hasMany(name, { ids: true });
      }
    });

    if (options.includeId) {
      json.ID_ = snapshot.id;
    }

    return json;
  }
});

function serverAttributeName(attribute) {
  return attribute.underscore().toUpperCase();
}

function serverHasManyName(name) {
  return serverAttributeName(name.singularize()) + "_IDS";
}

This serializer will generate JSON that looks like this:

{
  "TITLE": "Rails is omakase",
  "BODY": "Yep. Omakase.",
  "COMMENT_IDS": [ 1, 2, 3 ]
}

Tweaking the Default JSON

If you just want to do some small tweaks on the default JSON, you can call super first and make the tweaks on the returned JSON.

import DS from 'ember-data';

export default DS.RESTSerializer.extend({
  serialize(snapshot, options) {
    var json = this._super(snapshot, options);

    json.subject = json.title;
    delete json.title;

    return json;
  }
});

You can use this method to customize the root keys serialized into the JSON. The hash property should be modified by reference (possibly using something like _.extend) By default the REST Serializer sends the modelName of a model, which is a camelized version of the name.

For example, your server may expect underscored root objects.

import DS from 'ember-data';

export default DS.RESTSerializer.extend({
  serializeIntoHash(data, type, record, options) {
    var root = Ember.String.decamelize(type.modelName);
    data[root] = this.serialize(record, options);
  }
});

You can use this method to customize how polymorphic objects are serialized. By default the REST Serializer creates the key by appending Type to the attribute and value from the model's camelcased model name.