faunauth

v2.5.1

Published

2 years ago

Add email-based authentication to your Fauna database

Downloads

0High
0Medium
0Low

alexnitta

fauna faunaDB schema migration authentication magic link typescript

`faunauth`

This library helps you add email-based authentication to your Fauna database, allowing you to support three login patterns:

Faunauth is MIT-licensed. Contributions are welcome; please see ./CONTRIBUTING.md for details on how to contribute.

The faunuath npm package follows Semantic Versioning by using semantic-release. This means that the version number is determined automatically using commit messages. A change log is also automatically created in ./docs/CHANGELOG.md.

The main functions of faunauth are implemented in TypeScript, and type definitions are included. As with any TypeScript library, you don't need to use TypeScript in your own project in order to use faunauth.

Caveats

Before using this library, you should understand the different options available for authentication in Fauna. You can configure external authentication via an identity provider like Auth0, and this might be a good choice for your needs. On the other hand, if you wish to use the internal Fauna authentication features, then this library will allow you to add them to your database.

This library is built on top of fauna-schema-migrate from Fauna Labs. Note that fauna-schema-migrate is experimental; its README lists this disclaimer:

This repository contains unofficial patterns, sample code, or tools to help developers build more effectively with Fauna. All Fauna Labs repositories are provided “as-is” and without support. By using this repository or its contents, you agree that this repository may never be officially supported and moved to the Fauna organization.

This means faunauth is also experimental; use it at your own discretion.

You are likely using Fauna because you want to take advantage of its GraphQL API. At this point, fauna-schema-migrate does not provide integration with resources defined in a GraphQL schema (this might be added at some point). This means you cannot use fauna-schema-migrate to refer to indexes or collections created via GraphQL schemas. The one place this really matters is when provisioning a user role in order to grant access to your resources from your GraphQL schema. To help with this, you can create a script that calls the createOrUpdateUserRole utility function - more on this below.

Getting started

faunauth provides you with a set of FQL resources - collections, functions, indexes and roles - that will enable various authentication tasks. To add these resources to your Fauna database, you will complete a "schema migration" using fauna-schema-migrate. In this context, "schema migration" means that your current database schema will be migrated to a new schema that contains the FQL resources provided by faunauth. As noted in the fauna-schema-migrate readme, this is not a data migration tool.

You must complete a schema migration before the functions exposed by faunauth will work properly. The faunauth CLI will help you accomplish this.

Step 1: add Fauna resources via schema migration

npm i faunauth to install this package
npm i -D @fauna-labs/fauna-schema-migrate to install fauna-schema-migrate as a dev dependency
npx faunauth init to copy over files that will be used by fauna-schema-migrate to complete the schema migration process. These files should be committed to version control. They will be copied over to your project at fauna/src and fauna/resources/faunauth. Note that the npx faunauth init command will not overwrite any existing files - if you see output indicating that any of the files already exist, you will need to delete or rename your existing files and run npx faunauth init again. If you need to update your existing files, you can run npx faunauth init -o to overwrite any existing files with the version from faunauth.
If you're using yarn 1.x workspaces, you can avoid hoisting by setting up the workspaces.nohoist option in your root package.json file.
```
{
    "workspaces": {
        "packages": [ // this will vary depending on where you keep your packages
            "libs/*",
            "apps/*"
        ],
        "nohoist": [ // this tells yarn not to hoist faunauth
            "**/faunauth"
        ]
    }
}
```
Using the Fauna dashboard, select your database (or create it if it doesn't yet exist). Once you have selected the database, click Security at the left and use the Keys feature to add a new key with the built-in "Admin" role. Save this key somewhere secure (i.e. in a password manager); you won't see it again.
In a terminal window, run export FAUNA_ADMIN_KEY=<your_fauna_admin_key>, using the admin key you just created. This sets the environment variable that fauna-schema-migrate will use to authenticate you and identify which database to apply your migrations to. If you are using something other than the "Classic" region group, you will need to also run export FAUNADB_DOMAIN=<domain> using the domain for your region group. As an example, the command for the US region group is: export FAUNADB_DOMAIN="db.us.fauna.com". Other domain values are listed here.
Run npx fauna-schema-migrate generate to analyze the contents of the /fauna folder and determine what needs to be added to your database. The results of this analysis are saved in a new folder in /fauna/migrations. Review the migration that was created to make sure it looks correct.
Use npx fauna-schema-migrate apply to apply the migration to your database.
If needed, you can use npx fauna-schema-migrate rollback to revert your migrations.

Once you complete this process, your Fauna database will include the resources that were copied into your fauna folder when you ran npx faunauth init.

The fauna-schema-migrate commands are documented more thoroughly in the fauna-schema-migrate readme.

Step 2: create a public Fauna key for your client application

If your terminal does not yet have the FAUNA_ADMIN_KEY environment variable set, run: export FAUNA_ADMIN_KEY=<your_fauna_admin_key> with the admin key you created in Step 1 above.
npx faunauth create-public-key to create a key for the role named "public" and print it to the console. This role is created during schema migration, so you will see an error if you have not completed Step 1. If you are using a region group other than "Classic", you need to provide it after the -r flag, like this: npx faunauth create-public-key -r us for the "us" region. Run npx faunauth create-public-key --help to see other options. Save this key somewhere secure (i.e. in a password manager).

This public key has limited permissions that are appropriate for un-authenticated users, i.e. user registration, login, and resetting passwords. It is safe to store in the browser in your client-side code and your client will need to provide it to your server when calling the various faunauth functions in Step 4 below.

Step 3: add a role so that users can access your resources

Fauna does not allow access to a resource unless it is specifically granted. You have two options for this:

If you are not declaring your non-user resources with a GraphQL schema, you can add your own files to the /fauna folder in order to use fauna-schema-migrate with them. You must do the following:
- add FQL, JS or TS files to /fauna for all of your database resources (collections, indexes, roles)
- add an FQL, JS or TS file to /fauna to define a role that grants privileges to the User document for accessing your non-user resources
If you are declaring your resources with a GraphQL schema, or don't want to manage your non-user resources with fauna-schema-migrate there's a utility function called createOrUpdateUserRole that will help with this; see the docs here.

Step 4: use `faunauth` functions in your server-side code

After completing the first three steps, you can use the faunauth functions in your server. Complete documentation is available in ./docs/modules.md#functions. Each of these functions is available as a top-level import - for example:

import { login } from 'faunauth';

To create a full set of authentication features, you will need to implement server-side logic that does the following:

Sign up a new user with register. If you want users to be able to login with a username, they must provide both email, username and password. If you just want them to login with email, then only email and password are required.
Confirm a new user's email address with setPassword
Log in a user with login (works with either email address or username)
Log out a user with logout
Initiate a "forgot password" flow with sendConfirmationEmail
Finish a "forgot password" flow with setPassword
Change a password for a user that knows their password with changePassword
Rotate the accessSecret / refreshSecret pair with rotateTokens
Update non-password user details (email, etc.) with updateUser

Optionally, you can also implement login with magic link by doing:

Initiate a "magic link" flow with sendConfirmationEmail
Finish a "magic link" flow with loginWithMagicLink

Other available functions include:

registerAdmin: allows you to create a user for testing purposes without confirming their email address. This uses a Fauna admin key secret and does not authenticate the user because it doesn't require email confirmation, so use it with caution and do not expose it in your public API.
deleteUser: deletes a user account. This function works with a secret that is either a Fauna admin key secret or an accessSecret. If it is an accessSecret, it must be associated with the same user that is being deleted.

Secrets vs. tokens

You'll notice that faunauth uses the names accessSecret and refreshSecret to refer to the strings that you will use for various authentication tasks. This is because Fauna exposes a document type which is named Token. Each Token has a secret which can be used to authenticate queries. From a frontend developer perspective, the secret behaves like a token; in other words, you set the Authorization header to Bearer ${accessSecret} to get access to the Fauna API.

To reduce confusion between the Fauna definition of a token (instance of Token document) vs. a conventional token, we use the Fauna term "secret", hence the names accessSecret and refreshSecret. This was a breaking change in version 2.0 of faunauth.

Example login code

Here's an example of how you might use the login function from faunauth in an Express handler. Express is only used here for illustrative purposes; faunauth is framework-agnostic.

Note that you'll need to pass in the publicFaunaKey that you created in Step 2 above.

// This function handles login requests for an Express server route.
// You would typically use bodyparser to make sure responses are handled
// as JSON. Other server frameworks will work differently.
import { login } from 'faunauth';

/**
 * Log in a user in with an email and password, thereby getting access to
 * an accessSecret, refreshSecret and user data.
 */
export const loginHandler = async (req, res) => {
    // The `publicFaunaKey` should be created using the faunauth CLI as
    // described above.
    // Your client-side application will need to store this public key
    // in the browser and send it to the server for various
    // un-authenticated requests.
    const { email, password, publicFaunaKey } = req.body;

    try {
        // Here is where you would validate the input or throw an error
        // if it is invalid

        const { accessSecret, refreshSecret, user } = await login({
            email,
            password,
            publicFaunaKey,
        });

        // Here is where you would save the refreshSecret in a session
        // cookie.

        // Send the `accessSecret` and `user` back to the client so it can
        // use the `accessSecret` to authenticate Fauna requests
        res.send({ accessSecret, user });
    } catch (error) {
        res.status(400).send({ error });
    }
};

Confirming a user's identity

As with any other email-based authentication system, there is are a few steps involved when confirming a user's identity. Generally, these steps are as follows:

User states a wish to authenticate by attempting to sign up for a new account, reset the password for an existing account, or login via a "magic link" for an existing account.
A confirmation email is sent to the provided email address that contains a link. The link leads to a page on your site. You pass this in as the callbackUrl and faunauth appends a URL search parameter to it called data that contains both the email address and a unique token created in order to confirm the user's identity.
Your code calls a faunauth function that checks if the token is valid, i.e. the token exists, is associated with the given email address, and has not already been used. If the token is valid, the rest of the authentication flow is allowed to proceed.

As part of these user flows, you will need to expose a page within your app at the callbackUrl that you pass in. A URL search parameter called data will be appended to the callback URL which will include a Base64-encoded string containing an object with the email and token. Your app needs to read the data param, decode the email and token from it, and pass them to the appropriate faunauth function to complete the user flow.

Here's how you would parse the encoded data URL search parameter:

// If you're using react-router, you can get search params with
// useSearchParams()
const search = window.location.search
const urlQuery = new URLSearchParams(search);
const data = urlQuery.get('data');

try {
    const { email, token } = JSON.parse(atob(data));

    // Now that we have the email and token, we can call `setPassword()`
    // or `loginWithMagicLink()`
} catch {
    // Show an error message - we could not read data from URL, so either
    // the user came to this page by accident or they are trying to
    // impersonate someone else to gain access.
}

Here are the three user flows that implement this pattern:

Register a new user

The new user visits your sign up page and enters an email and password into a form. You may include an optional username field to allow your users to log in with a username and password. Your frontend app hits an API endpoint that calls the register function, which creates an entity in the User collection, creates a email confirmation token for that entity, and sends the confirmation email as described above.
The new user opens the confirmation email and clicks the link, which opens the callback URL with the added data URL parameter. Your frontend app must decode the email and token from this data parameter, as shown above, then hit an API endpoint that calls the setPassword function. This function returns an object containing the accessSecret, refreshSecret and user object. The endpoint should set the refreshSecret on a session cookie and return the accessSecret and user data back to the frontend.
You frontend should store a reference to the accessSecret and user data, then redirect the user as appropriate, usually to the main dashboard of the app. The accessSecret should be included in an Authorization header as the Bearer ${accessSecret} when making requests at the Fauna GraphQL endpoint.

Reset a password

An existing user visits your password reset page and enters the email address they registered previously with. Your frontend app hits an API endpoint that calls the sendConfirmationEmail function, which creates an email confirmation token for the User entity that matches the given email address and sends the confirmation email as described above.

Step 2 and 3 are the same as when signing up a new user.

Sign in with magic link

An existing user visits your sign in page and enters the email address they registered previously with. Your frontend app hits an API endpoint that calls the sendConfirmationEmail function, which creates an email confirmation token for the User entity that matches the given email address and sends the confirmation email as described above.
The new user opens the confirmation email and clicks the link, which opens the callback URL with the added data URL parameter. Your frontend app must decode the email and token from this data parameter, as shown above, then hit an API endpoint that calls the loginWithMagicLink function. This function returns an object containing the accessSecret, refreshSecret and user object. The endpoint should set the refreshSecret on a session cookie and return the accessSecret and user data back to the frontend.

Step 3 is the same as when signing up a new user.

Tokens

When a user logs in with the login function, they receive an accessSecret, refreshSecret and user object. The accessSecret provides identity-based access to your Fauna database, which means you can use it to authenticate requests to the Fauna GraphQL endpoint as well as to authenticate a Fauna client from the faunadb JavaScript driver, or any of the other drivers. In Fauna's terminology, the accessSecret is technically a "token secret." You can read more on Fauna tokens here.

Access tokens expire in 10 minutes. When a request to the Fauna GraphQL endpoint fails due to an expired token, your frontend app should hit an API endpoint that calls the rotateTokens function. This function takes the refreshSecret and uses it to get a new pair of accessSecret and refreshSecret values. Your endpoint handler should set the new refreshSecret on the session cookie so that it can be used in the future to repeat the token rotation process. It should return the accessSecret to the frontend app so that it can be used to authenticate further Fauna requests.

Sending emails

When calling either the register or sendConfirmationEmail functions, you have two options for sending the user an email that will confirm their identity:

Pass an input that conforms to AuthInputWithEmailTemplate in order to use the built-in faunauth email template. When using this method, you provide a config object that includes details about the styles and text content for the email, as well as an async function that will be called to send the email. Typically this function will be a wrapper around something like @sendgrid/mail.
Pass an input that confirms to AuthInputWithCustomEmail in order to provide your own email template logic. When using this method, you also need to provide an async function that sends the email, similarly to the function passed when using the faunauth email template. However, the only input to this email sending function is the final callback URL; you are responsible for actually building the rest of the email including all styling and text content.

Error handling

This library exports an errors object that is a map of error messages. Every error thrown within faunauth uses one of these messages. In your application, you can import the errors object and check if an Error instance has a .message property that exists on it. This allows you to deterministically show custom error messages to the user, which is useful when localizing your application. For example:

import { login, errors } from 'faunuath';

// In your application code...

login({
    publicFaunaKey: process.env.PUBLIC_FAUNA_KEY,
    password,
    email,
}).then(() => {
    // Do something on successful login
}).catch(e => {
    // In JavaScript, any expression can be thrown, so we have to check if
    // it's an error.
    if (e instanceof Error && e.message === errors.invalidEmailOrPassword) {
        // Show the user a message about invalid credentials
    } else {
        // Show the user some other error message
    }
})

To check for the possible errors in each function exported by faunauth, you could:

Browse the source code and check which errors are used in each function, then handle each of them separately
Create a function that takes an Error instance and returns a custom error message to show to the user. This function would need to import the errors object and use it to map the incoming Error to a custom message.