Typed repositories for Node around Firestore providing a very simple API to write and read documents. Including a simple to use query builder and an in-memory storage implementation for running blazing fast tests
Firestore Storage provides a thin layer of abstraction to accessing data in Firestore. It follows the repository pattern, for more information about it you can read this short article
const userRepo = new UserRepository();
// Saving a user
const user = await userRepo.save({
name: 'John Doe',
active: true
});
console.log(user);
/* prints
{
id: '0vdxYqEisf5vwJLhyLjA',
name: 'John Doe',
active: true,
createdAt: '2019-04-29T16:35:33.195Z',
updatedAt: '2019-04-29T16:35:33.195Z'
}*/
// Listing all documents
const allUsers = await userRepo.list();
// Filtering documents based on attributes
const activeUsers = await userRepo.list({
active: true
});
// More complex queries
const date = new Date('2019-02-01');
const asd = await userRepo.query((qb) => {
return qb
.where('signUpDate', '<=', date)
.orderBy('signUpDate', 'asc');
});
The firestore-storage package is available via npm
$ npm install firestore-storage
# or
$ yarn add firestore-storage
To run all tests using only the MemoryStorage implementation run:
$ yarn test
It's also possible to run all tests using the FirestoreStorage implementation. To do this
you need to create a Firebase project and download the Admin SDK credentials file.
Copy the .env.sample to .env and add the absolute path to the FIREBASE_CREDENTIALS variable.
To execute the tests run:
$ yarn test:firestore
To run tests using both MemoryStorage and FirestoreStorage run
$ yarn test:all
This packages provides utilities to run tests against your Firestore rules.
You need the @firebase/testing package and the local emulator installed.
To install the emulator run $ firebase setup:emulators:firestore
Below is an example of how to run tests against the rules. Create a new instance
of FirestoreRuleTest for each test. Add some test data, load the rules and run your assertions.
The constructor of FirestoreRuleTest takes the uid of the authenticated user as an argument.
This will be the request.auth.uid property which you can read in your rules.
Passing no uid will send unauthenticated requests to the emulator.
import {FirestoreRuleTest} from 'firestore-storage';
import * as firebase from "@firebase/testing";
describe('Rules', function () {
const pathToRules = `${__dirname}/../../../../firestore.rules`;
before(async () => {
await FirestoreRuleTest.start();
});
after(async () => {
await FirestoreRuleTest.stop();
});
describe('Unauthenticated', function () {
it('should not be able to read from users', async () => {
const tc = new FirestoreRuleTest();
const userId = 'alice';
const userDoc = tc.firestore.collection('users').doc(userId);
await userDoc.set({});
await tc.loadRules(pathToRules);
await firebase.assertFails(userDoc.get())
});
});
describe('Authenticated', function () {
it('should not be able to read reservations from different account', async () => {
const userId1 = 'alice';
const accountId1 = `account-${userId1}`;
const userId2 = 'bob';
const accountId2 = `account-${userId2}`;
const tc = new FirestoreRuleTest(userId1);
const userDoc1 = tc.firestore.collection('users').doc(userId1);
const userDoc2 = tc.firestore.collection('users').doc(userId2);
await userDoc1.set({accountId: accountId1});
await userDoc2.set({accountId: accountId2});
const resColl1 = tc.firestore.collection('accounts').doc(accountId1).collection('reservations');
const resColl2 = tc.firestore.collection('accounts').doc(accountId2).collection('reservations');
await resColl1.add({});
await resColl2.add({});
await tc.loadRules(pathToRules);
await firebase.assertSucceeds(resColl1.get());
await firebase.assertFails(resColl2.get());
});
});
});
Firestore Storage can be used with the dependency injection library Inversify as well as without it.
// In another file (storage.ts)
export const storage = new FirestoreStorage(admin.firestore());
// OR
export const storage = new MemoryStorage();
// user_repository.ts
import {storage} from './storage';
class UserRepository extends BaseRepository<User> {
constructor() {
super(storage);
}
getCollectionPath(...documentIds: string[]): string {
return 'users';
}
listAllActive() {
return this.list({active: true});
}
}
const repo = new UserRepository();
This library fully supports Inversify To use it you have load the FirestoreStorageModule module
into your container and add the @injectable() decorator to each repository class. This will automatically
inject the correct storage implementation (Firestore or In-Memory) into your repositories
if (process.env.NODE_ENV === 'test') {
container.load(FirestoreStorageModule.createWithMemoryStorage());
} else {
container.load(FirestoreStorageModule.createWithFirestore(admin.firestore()));
}
container.bind(UserRepository).toSelf().inSingletonScope();
Your models should extend or implement the interface BaseModel which contains the id and
modification timestamps.
interface BaseModel {
id?: string;
createdAt?: Date;
updatedAt?: Date;
}
Since those values are not in the document itself, they will be added to the
returning object when reading from Firestore. You can pass objects with those attributes to
the save() function. They will always be omitted and the id will be used as the document id
when writing data.
Each repository as well as the FirestoreStorage and MemoryStorage implementations provide a transaction() function.
Create repository classes for each collection you want to query documents from. For example
if you want to query documents to query from the users collection you create a class UserRepository extending BaseRepository.
Each repository provides a list of functions for saving, querying and deleting documents
and you can extend each repository based on your needs.
When extending BaseRepository you have to implement the function
getCollectionPath(...ids: string[]). For root collections the ids[] will be empty.
For sub-collections this parameter will contain an hierarchically ordered list of parent
document ids.
Each function takes multiple ids as its last arguments. Those are the hierarchically
ordered list of parent document ids passed to the getCollectionPath(...) function.
The following examples are based on the UserRepository and TodoRepository
created below
Takes a hierarchical ordered list of document ids. Returns the document when found or null
const todo = await todoRepo.findById(userId, todoId);
Queries the collection to match the given arguments, returns the first result or null if none is found.
const doneTodo = await todoRepo.find({
done: true
}, userId);
Works exactly like findById but throws an error if no document was found
Works exactly like find but throws an error if no document was found
Query a list of documents with a set of given arguments. This function always returns an array. If no results were found the array will be empty
const allDoneTodos = await todoRepo.list({
done: true
}, userId);
Do more complex queries like greater than and lower than comparisons.
const passedDeadlineTodos = await todoRepo.query(() => {
return qb
.where('done', '==', false)
.where('deadlineDate', '<', new Date());
});
Valid operators are == | < | <= | > | >=
qb.where(fieldName, operator, value)
qb.orderBy(fieldName, direction) // 'asc' or 'desc'
qb.offset(number)
qb.limit(number)
Returns an array of documents for a given array of ids. The array will contain null values if some documents aren't found
const users = await userRepo.batchGet([userId1, userId2]);
Saves a document into Firestore.
const user = await userRepo.save({
name: 'John',
email: 'john@exmaple.com'
});
If you want to update data you just have to pass the id of the document.
const user = await userRepo.save({
id: '8zCW4UszD0wmdrpBNswp',
name: 'John',
email: 'john@exmaple.com'
});
By default this will create the document with this id if it doesn't exist or merge the properties into the existing document. If you want to write a document and instead of don't merge use the [write()][write] function
Sets the passed data. If the document exists it will be overwritten.
const user = await userRepo.write({
name: 'John',
email: 'john@exmaple.com'
});
Deletes a document by a given id
// For a nested collection
await todoRepo.delete(userId, todoId);
// For a root level collection
await userRepo.delete(userId);
Takes an update function and an array of ids. Find more about transactions at the Firestore documentation
const result = await userRepo.transaction((trx) => {
const u = trx.get('some-id');
u.name = 'John';
trx.set(u);
return 'worked';
})
export class UserRepository extends BaseRepository<User> {
getCollectionPath(...documentIds: string[]): string {
return 'users';
}
}
When creating repositories for nested collection it's always a good idea to check if the correct ids are passed into
getCollectionPath(...).
export class TodoRepository<T> extends BaseRepository<Todo> {
getCollectionPath(...documentIds): string {
const id = documentIds.shift();
if (!id) {
throw new Error('User id is missing');
}
return `users/${id}/todos`;
}
}
This will throw an error when trying to save or query without passing the user id.
await todoRepo.save({...}); // Throws and error
await todoRepo.save({...}, '<userId>'); // Succeeds
This package provides a base class to migrate data in Firestore. For more info look at this example
The query functions get and getById will throw an error if the document doesn't exist. If you want to throw an custom error you can do that by passing an error factory.
export class HttpError extends Error {
constructor(msg: string, public code: number) {
super(msg)
}
}
const errorFactory = (msg) => {
return new HttpError(msg, 404);
};
FirestoreStorageModule.createWithFirestore(admin.firestore(), errorFactory)
class UserRepository extends BaseRepository<User> {
constructor() {
super(storage, errorFactory);
}
}