A Node.js REST API example that uses Firebase Admin, built with Express and Typescript that can be used as template for the creation of new servers.
The main aspects of this sample are:
- A project structure that fits well for new API projects that uses Firebase Authentication and Firestore
- Access Control: Restricting routes access with custom claims and checking nuances
- Reject a request outside the controller easily by throwing
new HttpResponseError(status, codeString, message)
- Logs: winston module is preconfigured to write
.log
files - Serialization of objects, so you can easily perform write operations on Firestore of custom prototypes objects
- Caching: A wrapper function which helps to avoid identical queries on the Firestore Database
📱 Check also the Flutter side example which interacts with this API example.
- Getting Started
- Authentication
- Access Control
- Errors and permissions
- Logs
- Serialization of objects
- Caching Firestore results
- Reference
- Go to your Firebase Project
- Click on the engine icon (on right of "Project Overview")
- "Project Settings"
- "Service Accounts"
- "Firebase Admin SDK" and make sure the "Node.js" option is selected
- Click on "Generate new private key". Rename the downloaded file to
firebase-credentials.json
and move it to inside theenvironment
folder.
firebase-credentials.json
and environment.ts
local,
don't commit these files, keep both on .gitignore
As any Firebase server, the API has adminstrative priviliges, that means the API has full permission to perform changes on the Firestore Database (and other Firebase Resources) regardless how the Firestore Security Rules is configured.
This command will start and restart your server as code changes are made, do not use on production
npm run dev
Let's run npm install
to install the dependencies and npm run dev
to start your server locally on port 3000.
npm run build
npm run start
📱 Check this Flutter project to interact with the server, you can also create your own client that uses the Firebase Authentication library, like React, Angular, Vue, etcetera.
We are done! Customize this API on your way!
Firebase Authentication is used to verify
if the client is authenticated on Firebase Authentication,
to do so, the client side should inform the Authorization
header:
The client's token on Firebase Authentication in the format Bearer <token>
,
it can be obtained in the client side after the authentication is performed with the
Firebase Authentication library for the client side.
final dioLoggedIn = Dio(BaseOptions(
baseUrl: 'https://myapi.example.com',
headers: {
"Authorization": "Bearer ${(await FirebaseAuth.instance.currentUser!.getIdToken())}"
}
));
// dioLoggedIn.get('/user').then(...);
📱 Click here to check a Flutter client example for this API
This project uses custom claims on Firebase Authentication to define which routes the users have access to.
This can be done in the server like bellow:
await admin.auth().setCustomUserClaims(user.uid, {
storeOwner: true,
buyer: false
});
You can set a param (array of strings) on the httpServer.<method>
function, like:
httpServer.get ('/product/:productId/full-details',
this.getProductByIdFull.bind(this), ['storeOwner']);
In the example above, only users with the storeOwner
custom claim will
have access to the /product/:productId/full-details
path.
Is this enough? Not always, so let's check the next section Errors and permissions.
You can easily send an HTTP response with code between 400 and 500 to the client
by simply throwing a new HttpResponseError(...)
on your controller, service or repository,
for example:
throw new HttpResponseError(400, 'BAD_REQUEST', "Missing 'name'");
Sometimes defining roles isn't enough to assure that a user can't
access or modify a specific data,
let's imagine if a storeOwner tries to get full details
of a product he is not selling, like a product of another storeOwner,
he still has access to the route because of his storeOwner
custom claim,
but an additional verification is needed.
if (product.storeOwnerUid != req.auth!.uid) {
throw new HttpResponseError(
403,
'FORBIDDEN',
`Even though you are a storeOwner,
you are a owner of another store,
so you can't see full details of this product`
);
}
This project adds 3 new fields to the request object on the express request handler:
type: boolean
Is true only if the client is authenticated, that means, the client
informed Authorization
on the headers, and these
values were successfully validated.
type: UserRecord | null
If authenticated: Contains user data of Firebase Authentication.
type: DecodedIdToken | null
If authenticated: Contains token data of Firebase Authentication.
You can save logs into a file by importing these functions of the src/utils/logger.ts
file
and using like:
log("this is a info", "info");
logDebug("this is a debug");
logInfo("this is a info");
logWarn("this is a warn");
logError("this is a error");
By default, a logs
folder will be generated
aside this project folder, in this structure:
- /api-example-firebase-nodejs
--- /node_modules/*
--- /src/*
--- (and more)
- /logs
--- /api-example-firebase-nodejs
------ 2022-8-21.log
------ 2022-8-22.log
------ 2022-8-23.log
------ (and more)
Each .log
file contains the logs of the respective day.
You can also go to src/utils/logger.ts
and check logsFilename
and logsPathAndFilename
fields
to change the default path and filename so the logs can be saved with a different file name and
in a different location.
By default, regardless of the log level, all logs will be saved in the same file,
you can also change this behavior on the winston.createLogger(transports: ...)
line of
the src/utils/logger.ts
file.
You will get an error if you create your own class, instantiate an object of it and try to save directly on Firestore:
await db().collection('products').doc().set(new ProductEntity(...));
Error: Value for argument "data" is not a valid Firestore document.
Couldn't serialize object of type "ProductEntity".
Firestore doesn't support JavaScript objects with custom prototypes
(i.e. objects that were created via the "new" operator).
To fix this problem, this project has a function called serializeFS(object)
which
accepts an object as param.
await db().collection('products').doc().set(serializeFS(new ProductEntity(...)));
Sometimes you need to fetch for the same data on the database in two or more
functions, you may need to fetch for product
in the database
to check if the client has permission to read it, and if so, you may want to
return the exact same product
as response.
The commom solution is to pass the cached data as param on different functions. This project also offers an alternative way of caching:
You can use req.cacheOf(cacheId, function)
in the request handler to wrap a function into
a new function that will cache the result, in this way, a cache will be created in the
first time this function is called and will be used as result when this function is called
again.
// If there's a cache: it will use the cache, otherwise: it will wait for the getProductById result and cache it
const getProductByIdCached = req
.cacheOf(req.params['productId'], productsRepository.getProductById); // <-- Wrapped with a cache function
const product = await getProductByIdCached(req.params['productId']);
The cache will be valid for a single request handler, so you will not have problems of inconsistent cache on different requests, because each request has its own cache.
But if the data changes, and you want to invalidate the cache on that
request handler, you can call req.invalidateCache(cacheId)
, for example:
req.invalidateCache(req.params['productId']);
Calling req.invalidateCache
will not affect the other requests.
This project based part of the structure of the GitHub project node-typescript-restify. Thank you developer!