Navigation Menu

Skip to content
This repository has been archived by the owner on Jun 7, 2019. It is now read-only.


Folders and files

Last commit message
Last commit date

Latest commit



11 Commits

Repository files navigation



Hi. This project has been archived, we have moved away from Firebase to Hasura and re no longer maintaining this project. Thanks for your interest.


ninjafire is an ORM for Firebase written in Typescript. It is heavily based on the API for Ember Data. It is compatible with data in Firebase created by the emberfire project.

It is intended to be used in node, and can be used in a Firebase Cloud Function, and it has the following features

  • Strongly Typed
  • Attribute types:
    • Boolean
    • Date
    • JSON
    • Number
    • String
  • Relationships:
    • belongsTo <-> belongsTo (1-1)
    • belongsTo <-> hasMany (1-many)
    • hasMany <-> hasMany (many-many)
  • Embedded Records. Both belongsTo and hasMany. Compatible with emberfire
  • Atomic saving of relationship changes and, optionally, all changes
  • An option to 'chroot' all the records under a base path
  • An option to group a collection of records under an alternative path
  • Optionally use Firebase push id's or UUID v1 or v4 for new records

Defining Models

Models are defined by providing the following items

  • modelName A string that identifies the name of the model. It should be unique to the model
  • modelPath A string that defines the name of a collection of these records. It is used to define the path to store the data in firebase. In Ember Data this is automatically 'pluralized' from modelName. Currently you need to explicitly specify it.
  • schema An object whose keys are the attributes of the model and whose values are the types of each attribute.
  • The types of the attributes. These are used so Typescript can type check the model. Attributes that return relationships (that are not embedded) should be ModelOrPromise<...> or ModelOrPromise<...>[].

For example (you can find this model at tests/models/blog.ts)

import { attr, belongsTo, hasMany, Model, ModelOrPromise, Schema, Serializers} from 'ninjafire';
import { Post, User } from '../models';

export class Blog extends Model {

    public static modelName: string = 'blog';
    public static modelPath: string = 'blogs';

    public schema: Schema = {

        name: attr(Serializers.String),
        description: attr(Serializers.String),
        createdDate: attr(Serializers.Date, { defaultValue: (): Date => new Date() }),
        updatedTime: attr(Serializers.Number, { setToServerTimestampOnSave: true }),
        published: attr(Serializers.Boolean),
        featured: attr(Serializers.Boolean),
        ranking: attr(Serializers.Number),
        config: attr(Serializers.JSON),

        owner: belongsTo(User, { inverse: 'blog' }),
        posts: hasMany(Post, { inverse: 'blog' }),

    public name: string;
    public description: string | null;
    public createdDate: Date;
    public updatedTime: number | null;
    public published: boolean;
    public featured: boolean;
    public ranking: number;
    public config: {} | null;

    public owner: ModelOrPromise<User> | null;
    public posts: ModelOrPromise<Post>[];

Schema Attributes

To define an attribute set the schema to attr(Serializers.[Type]) the following are included:

  • Serializers.Boolean
  • Serializers.Date
  • Serializers.JSON This will call JSON.stringify and JSON.parse on the value when serializing / deserializing
  • Serializers.Number You can configure this type to automatically set the value to the server timestamp on save by passing options including setToServerTimestampOnSave: true
  • Serializers.String

All serializers take a options object that can include a defaultValue function. It will be called when accessing the attribute if a value has not been specified and also .save()ing a record with no value set.

Schema Relationships

Relationships to other records are either belongsTo or hasMany. Specify the class of the model for the relationship eg

owner: belongsTo(User),
posts: hasMany(Post),

Inverse relationships are not automatically discovered (unlike Ember Data). You must explicitly state the key on the inverse model that is the inverse. You need to do this on both sides of the relationship.

owner: belongsTo(User, { inverse: 'blog' }),
posts: hasMany(Post, { inverse: 'blog' }),

belongsTo relationships return an object of ModelOrPromise<T>, where T is the model type. eg:

public owner: ModelOrPromise<User>;

hasMany relationships return an object of ModelOrPromise<T>[], which is to say an array containing ModelOrPromise<T> objects. eg:

public posts: ModelOrPromise<Post>[];

If the related record(s) are loaded then they will be returned immediately, otherwise it will return a promise resolving to the record(s).

Unset relationships will return null. If the relationship may be null then specify this in the type definition, eg:

public owner: ModelOrPromise<User> | null;

This will then ensure that you have appropriate null checks in place when using Typescript. Whether a record can be null or not should match your Firebase rule configuration.

When the ModelOrPromise<T> is currently a Promise it has id and isLoading properties in addition to the regular Promise properties.

If the related record is not found, i.e. the relationship is invalid, then an exception will be thrown.

Initializing a Store

A store is initialized like so (where admin is an initialized firebase-admin instance). The module should also work with the regular firebase client but that has not yet been tested.

const store: Store = new Store(admin.database());

You can optionally provide a basePath, eg /local that will 'chroot' all records under this location

const store: Store = new Store(admin.database(), { basePath });

This 'basePath' functionality is not supported natively in emberfire yet, but there is a PR to implement it

Additionally you can configure the store to use create records using a v1 or v4 UUID, rather than a Firebase push id. Like so

const store: Store = new Store(admin.database(), { useUUID: 4 });

It is important to note that the created store is not a singleton. Each store created will be independent. If you require a singleton then wrap a store in your own singleton.

Creating a Record

Create a record like this

const blog: Blog = store.createRecord(Blog, {
    id: '1234567890',
    name: 'The Blog Name,

If an id is not provided a Firebase push id will be assigned. Optionally you can configure the store to use a UUID instead, see above.

You can get/set attributes using the models properties eg

const oldName = = 'An updated Name'

It can then be saved by calling


Finding a Record

const blog: Blog = await store.findRecord(Blog, '1234567890');

store.findRecord returns an object of ModelOrPromise<T> meaning it either immediately returns a model, or it returns a promise that will resolve to the model. If the model is not found an exception will be thrown

Relating Records

Two way relationships in Firebase require an entry in the record on both sides of the relationship. ninjafire ensures these records stay in sync by performing the Firebase updates in a single atomic operation.

ninjafire tracks which relationships have changed and will save all related record when any one of the records is saved.

For example

const blog: Blog = store.createRecord(Blog, { name: 'The Blog Name', });
const post1: Post = store.createRecord(Post, { name: 'Post 1', });
const post2: Post = store.createRecord(Post, { name: 'Post 2', });
const user: Post = store.createRecord(Post, { name: 'User', });

blog.posts = [post1, post];
blog.owner = user;

Then saving any of the 4 objects will cause all 4 to save in one atomic update eg


For this to work records that have had relationship changes must already be loaded into the store, either with findRecord, createRecord, or through traversing relationships. An exception will be thrown if they are not.

ninjafire will also unlink relationships when they change. Eg continuing with the above example:

const newUser: Post = store.createRecord(Post, { name: 'Another User', });
blog.owner = newUser;


This will update blog, user and newUser. Again for this to work all 3 records must be in the store. An exception will be thrown if they are not.

Deleting Records

To delete a record call the .deleteRecord() method on the record. For relationships that have an inverse the related record will also be updated to remove the reference back to the record.

Note that this is an async operation, unlike in Ember Data. This is to support the retrieval and update of related records.

await blog.deleteRecord();

Similarly to Ember Data deleteRecord will not automatically save that change. You will later need to perform a save() on the record.

If you want to immediately delete the record then perform destroyRecord(). This will delete and commit the deletion.

await blog.destroyRecord();

Embedded Records

ninjafire supports the same 'embedded records' style as used in emberfire (details). This means that a record can contain a belongsTo or hasMany relationship where the data for the target is nested under the parent record.

Embedded records can be useful when you want to include an array like structure in a record and always want to retrieve the members in the same request as the parent record. It can result in fewer calls to Firebase. However embedded records can't be referenced in a relationship from anything other than their parent record. The embedded record can however reference other records via relationships.

To embed a record the model must be marked as embeddable like so (see test/models/blog/post/photos.ts)

public static embedded: boolean = true;

Then in the parent record the relationship must contains an embedded: true option like so (see test/models/blog/post.ts):

heroImage: belongsTo(Photo, { embedded: true }),
photos: hasMany(Photo, { embedded: true }),

Grouping Records

In addition to being able to set a base path for all records, you can 'group' types under an additional prefix. This is useful for, say, grouping all records relating to a team under a single path. This can help with performance and makes Firebase rules easier to write.

To set a group define an additional static property on the model, eg:

public static pathPrefixGroup: string = 'team';

Then tell the store what path to use for this group: = `/team/1234-1234-1234`;

Not all requests for models with the 'post' pathPrefixGroup will have /team/1234-1234-1234 prefixed to their references.

See the test/path-prefix-group.ts file for an example of this in action.

Saving all changes

You can commit all outstanding changes in an atomic operation by calling saveAll() on the store. Like this:

await store.saveAll();

It will either commit all changes or error and commit none of them.


The intent is to implement a largely similar API to Ember Data. All the documented features above have been implemented.

The following API has been implemented on the store

  • createRecord
  • findRecord
  • unloadRecord
  • unloadAll

Notably these methods have not yet been implemented

  • query

The following API has been implemented on models

  • save
  • changedAttributes
  • dirtyType
  • hasDirtyAttributes
  • isValid
  • isLoading
  • isSaving
  • isNew
  • rollbackAttributes
  • unloadRecord
  • deleteRecord
  • destroyRecord


This package is not yet published to npm/yarn. So you will need to install it directly form github like so:

With Yarn:

yarn add ""

Or alternatively with npm:

npm install --save-dev ""

Because the package is not yet published to NPM it will compile itself when yarn/npm installs the module. This should happen transparently.


This project currently has 100% code coverage. The tests are currently performed against a live Firebase and need a bit of setup to run:

  • Configure Firebase so that a user called ninjafire can access a path in the database
  • Create a .env file with:
    • FIREBASE_DATABASE_URL set to your Firebase instance.
    • FIREBASE_BASE_PATH the path you want the tests to occur at. The ninjafire user must be able to read/write at this path.
  • Place serviceAccountKey.json in the root of the project
  • Install dependencies with yarn
  • Open the project in VSCode and run either of the test launch configurations
    • Run mocha tests Has source maps, but no code coverage, test output is in the debug console
    • Run mocha tests with code coverage Has code coverage but no source maps, test output in the embedded terminal.
  • Alternatively run yarn test which will perform the tests without code coverage.


Bug reports and pull requests are welcome on GitHub at


This project was created by Lineup Ninja. If you're planning a complex event, we can help you build your timetable!


MIT License.


An ORM for Firebase written in Typescript. Heavily based on the Ember Data API and compatible with data in Firebase created by the emberfire project.








No releases published


No packages published