Implémentation de WebAuthn pour les connexions sans mot de passe

Écrit par Oghenetega Denedo✏️

Se souvenir et stocker les mots de passe peut être très compliqué pour nos utilisateurs. Imaginez si la connexion était globalement plus facile pour tout le monde. C'est là qu'intervient WebAuthn, ou Web Authentication API. WebAuthn vise à offrir un avenir sans mots de passe.

Dans cet article, nous expliquerons ce qui fait fonctionner WebAuthn, en expliquant comment il utilise la cryptographie à clé publique pour assurer la sécurité des choses. Nous vous guiderons également dans l'intégration de WebAuthn pour une application Web simple afin d'apprendre à utiliser l'API de manière pratique.

Comme toute solution, WebAuthn a ses bons et ses moins bons côtés. Nous passerons en revue ses avantages et ses inconvénients afin que vous puissiez déterminer s'il correspond le mieux à vos besoins d'authentification. Venez nous rejoindre pour tenter de dire adieu aux problèmes de mots de passe et explorer la promesse d'une expérience de connexion transparente avec WebAuthn.

Choses à savoir avant de découvrir WebAuthn

Avant de passer à la mise en œuvre des connexions sans mot de passe avec WebAuthn, il est essentiel que vous ayez les conditions préalables suivantes en place :

• Node.js installé sur votre machine
• Un appareil Android ou iOS compatible avec WebAuthn à des fins de test
• Connaissance de base de Node.js et Express.js
• Base de données MongoDB pour stocker les informations d'identification et les mots de passe des utilisateurs

Si vous connaissez déjà ce qu'est WebAuthn et comment il fonctionne, n'hésitez pas à passer à la section de mise en œuvre. Si vous sentez que vous avez besoin d'un rappel, ce qui suit devrait vous aider à poser les bases.

Qu’est-ce que WebAuthn ?

WebAuthn est un standard Web né de la nécessité d'une authentification sécurisée et sans mot de passe dans les applications Web pour remédier aux principales lacunes liées à l'utilisation des mots de passe.

Le projet a été publié par le World Wide Web Consortium (W3C) en collaboration avec le FIDO (Fast Identity Online) dans le but de créer une interface standardisée qui fonctionne sur tous les appareils et systèmes d'exploitation pour authentifier les utilisateurs.

Sur le plan pratique, WebAuthn est composé de trois composants essentiels : la partie utilisatrice, le client WebAuthn et l'authentificateur.

La partie utilisatrice est le service ou l'application en ligne qui demande l'authentification de l'utilisateur.

Le client WebAuthn agit comme intermédiaire entre l'utilisateur et la partie utilisatrice : il est intégré dans tout navigateur Web ou application mobile compatible prenant en charge WebAuthn.

L'authentificateur est l'appareil ou la méthode utilisée pour vérifier l'identité de l'utilisateur, comme un scanner d'empreintes digitales, un système de reconnaissance faciale ou une clé de sécurité matérielle.

Comment fonctionne WebAuthn ?

Lorsque vous créez un compte sur un site Web prenant en charge WebAuthn, vous démarrez un processus d'inscription qui implique l'utilisation d'un authentifiant tel que votre scanner d'empreintes digitales sur votre téléphone. Cela aboutit à la génération d’une clé publique stockée dans la base de données de la partie utilisatrice et d’une clé privée stockée en toute sécurité sur votre appareil via une couche matérielle sécurisée.

Étant donné que le site Web ne demandera pas de mot de passe lors de la tentative de connexion. Ce qui se passe réellement, c'est qu'après le lancement de la connexion, un défi est envoyé à votre appareil. Ce défi contient généralement des informations telles que l'adresse du site Web pour confirmer que vous vous connectez à partir du site Web attendu par la partie utilisatrice.

Après avoir reçu le défi du site Web, votre appareil utilise votre clé privée pour créer une réponse signée. Cette réponse montre que vous possédez la clé publique correspondante stockée par le site Web sans divulguer la clé privée elle-même.

La partie utilisatrice valide la clé publique stockée dès réception de votre réponse signée. Si la signature s'aligne, le site Web peut vérifier que vous êtes le véritable utilisateur et vous accorder l'accès. Aucun mot de passe n'a été échangé et votre clé privée est restée en sécurité sur votre appareil.

Comment implémenter l'authentification sans mot de passe avec WebAuthn

Maintenant que nous avons couvert les concepts fondamentaux de WebAuthn, nous pouvons voir comment tout cela se déroule dans la pratique. L'application que nous allons construire sera une simple application Express.js avec quelques points de terminaison API pour gérer l'inscription et la connexion, une page HTML de base contenant le formulaire de connexion et d'inscription.

Configuration du projet

Tout d'abord, vous devrez cloner le projet depuis GitHub, qui contient le code de démarrage afin que nous n'ayons pas beaucoup d'échafaudage à faire.

Dans votre terminal, saisissez les commandes ci-dessous :

git clone https://github.com/josephden16/webauthn-demo.git

git checkout start-here # remarque : assurez-vous d'être sur la branche starter

Si vous souhaitez afficher la solution finale, consultez la solution finale ou la branche principale.

Ensuite, installez les dépendances du projet :

npm install

Ensuite, créez un nouveau fichier, .env, à la racine du projet. Copiez-y le contenu du .env.sample et fournissez les valeurs appropriées :

# .env
PORT=8000
MONGODB_URL=<YOUR MONGODB CONNECTION STRING>

After following these steps, the project should run without throwing errors, but to confirm, enter the command below to start the development server:

npm run dev

With that, we've set up the project. In the next section, we'll add the login and registration form.

Creating the login and registration form

The next step in our process is creating a single form that can handle registration and logging in. To do this, we must create a new directory in our codebase called public. Inside this directory, we will create a new file called index.html. This file will contain the necessary code to build the form we need.

Inside the index.html file, add the following code:

<!DOCTYPE html>
<html lang="en">

<head>
  <meta charset="UTF-8">
  <meta name="viewport" content="width=device-width, initial-scale=1.0">
  <title>WebAuthn Demo</title>
  <link rel="preconnect" href="https://fonts.googleapis.com">
  <link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
  <link href="https://fonts.googleapis.com/css2?family=Inter:wght@100..900&display=swap" rel="stylesheet">
  <script src="https://cdn.tailwindcss.com"></script>
  <script src="https://unpkg.com/@simplewebauthn/browser/dist/bundle/index.umd.min.js"></script>
  <style>
    .font-inter {
      font-family: 'Inter', sans-serif;
    }
  </style>
</head>

<body class="font-inter min-h-screen flex flex-col items-center p-24">
  <h1 class="font-bold text-3xl mb-10">WebAuthn Demo</h1>
  <div id="content">
    <div>
      <div id="error" role="alert"
        class="bg-red-600 text-white p-2 w-full my-3 rounded-md text-center font-bold hidden"></div>
      <div id="loginForm" class="text-center">
        <input id="username" autocomplete="webauthn" type="text" placeholder="Username"
          class="w-[20rem] px-4 py-2 mb-4 border rounded-md focus:outline-none focus:ring-2 focus:ring-blue-500">
        <div class="flex justify-center space-x-4">
          <button id="registerBtn" type="button"
            class="bg-green-500 hover:bg-green-600 text-white font-semibold py-2 px-4 rounded-md">Register</button>
          <button id="loginBtn" type="button"
            class="bg-blue-500 hover:bg-blue-600 text-white font-semibold py-2 px-4 rounded-md">Login</button>
        </div>
      </div>
      <div id="welcomeMessage" class="hidden text-center">
        <h1 class="text-3xl font-bold mb-4">Welcome, <span id="usernameDisplay"></span>!</h1>
        <p class="text-lg text-gray-600">You are now logged in.</p>
      </div>
    </div>
  </div>
</body>

</html>

So, we've just added a simple login and registration form for users to sign in with WebAuthn. Also, if you check the element, we've included the link to the Inter font using Google Fonts, Tailwind CSS for styling, and the SimpleWebAuthn browser package.

SimpleWebAuthn is an easy-to-use library for integrating WebAuthn into your web applications, as the name suggests. It offers a client and server library to reduce the hassle of implementing Webauthn in your projects.

When you visit http://localhost:8010, the port will be what you're using, you should see a form like the one below:  

Let's create a script.js file that'll store all the code for handling form submissions and interacting with the browser's Web Authentication API for registration and authentication. Users must register on a website before logging in, so we must implement the registration functionality first.

Head to the script.js file and include the following code:

const { startRegistration, browserSupportsWebAuthn } = SimpleWebAuthnBrowser;

document.addEventListener("DOMContentLoaded", function () {
  const usernameInput = document.getElementById("username");
  const registerBtn = document.getElementById("registerBtn");
  const loginBtn = document.getElementById("loginBtn");
  const errorDiv = document.getElementById("error");
  const loginForm = document.getElementById("loginForm");
  const welcomeMessage = document.getElementById("welcomeMessage");
  const usernameDisplay = document.getElementById("usernameDisplay");

  registerBtn.addEventListener("click", handleRegister);
  loginBtn.addEventListener("click", handleLogin);
});

At the start of the code above, we import the necessary functions to work with WebAuthn. The document.addEventListener("DOMContentLoaded", function () { ... }) part ensures that the code inside the curly braces ({...}) executes after the web page is loaded.

It is important to avoid errors that might occur if you try to access elements that haven't been loaded yet.

Within the DOMContentLoaded event handler, we're initializing variables to store specific HTML elements we'll be working with and event listeners for the login and registration buttons.

Next, let's add the handleRegister() function. Inside the DOMContentLoaded event handler, add the code below:

async function handleRegister(evt) {
  errorDiv.textContent = "";
  errorDiv.style.display = "none";
  const userName = usernameInput.value;

  if (!browserSupportsWebAuthn()) {
    return alert("This browser does not support WebAuthn");
  }

  const resp = await fetch(`/api/register/start?username=${userName}`, {
    credentials: "include"
  });
  const registrationOptions = await resp.json();
  let authResponse;
  try {
    authResponse = await startRegistration(registrationOptions);
  } catch (error) {
    if (error.name === "InvalidStateError") {
      errorDiv.textContent =
        "Error: Authenticator was probably already registered by user";
    } else {
      errorDiv.textContent = error.message;
    }
  }
  if (!authResponse) {
    errorDiv.textContent = "Failed to connect with your device";
    return;
  }
  const verificationResp = await fetch(
    `/api/register/verify?username=${userName}`,
    {
      credentials: "include",
      method: "POST",
      headers: {
        "Content-Type": "application/json",
      },
      body: JSON.stringify(authResponse),
    }
  );
  if (!verificationResp.ok) {
    errorDiv.textContent = "Oh no, something went wrong!";
    return;
  }
  const verificationJSON = await verificationResp.json();
  if (verificationJSON && verificationJSON.verified) {
    alert("Registration successful! You can now login");
  } else {
    errorDiv.textContent = "Oh no, something went wrong!";
  }
}

The handleRegister() function initiates the registration process by retrieving the username entered by the user from an input field. If the browser supports WebAuthn, it sends a request to the /api/register/start endpoint to initiate the registration process.

Once the registration options are retrieved, the startRegistration() method initiates the registration process with the received options. If the registration process is successful, it sends a verification request to another API endpoint /api/register/verify with the obtained authentication response and alerts the user that the registration was successful.

Since we haven't built the API endpoint for handling user registration yet, it won't function as expected, so let's head back to the codebase and create it.

Building the registration API endpoints

To finish the registration functionality, we'll need two API endpoints: one for generating the registration options that'll be passed to the authenticator and the other for verifying the response from the authenticator. Then, we'll store the credential data from the authenticator and user data in the database.

Let's start by creating the MongoDB database models to store user data and passkey. At the project's root, create a new folder called models and within that same folder, create two new files: User.js for the user data and PassKey.js for the passkey.

In the User.js file, add the following code:

import mongoose from "mongoose";

const UserSchema = new mongoose.Schema(
  {
    username: {
      type: String,
      unique: true,
      required: true,
    },
    authenticators: [],
  },
  { timestamps: true }
);

const User = mongoose.model("User", UserSchema);

export default User;

We're defining a simple schema for the user model that'll store the data of registered users. Next, in the PassKey.js file, add the following code:

import mongoose from "mongoose";

const PassKeySchema = new mongoose.Schema(
  {
    user: {
      type: mongoose.Schema.ObjectId,
      ref: "User",
      required: true,
    },
    webAuthnUserID: {
      type: String,
      required: true,
    },
    credentialID: {
      type: String,
      required: true,
    },
    publicKey: {
      type: String,
      required: true,
    },
    counter: {
      type: Number,
      required: true,
    },
    deviceType: {
      type: String,
      enum: ["singleDevice", "multiDevice"],
      required: true,
    },
    backedUp: {
      type: Boolean,
      required: true,
    },
    authenticators: [],
    transports: [],
  },
  { timestamps: true }
);
const PassKey = mongoose.model("PassKey", PassKeySchema);

export default PassKey;

We have created a schema for the PassKey model that stores all the necessary data of the authenticator after a successful registration. This schema will be used to identify the authenticator for all future authentications.

Having defined our data models, we can now set up the registration API endpoints. Within the root of the project, create two new folders: routes and controllers. Within each of the newly created folders, add a file named index.js. Within the routes/index.js file, add the code below:

import express from "express";
import {
  generateRegistrationOptionsCtrl,
  verifyRegistrationCtrl,
} from "../controllers/index.js";

const router = express.Router();

router.get("/register/start", generateRegistrationOptionsCtrl);
router.post("/register/verify", verifyRegistrationCtrl);

export default router;

We're defining the routes we used earlier for user registration using Express.js. It imports two controller functions for generating registration options and verifying the response from the startRegistration() method that'll be called in the browser.

Let's start by adding the generateRegistrationOptionsCtrl() controller to generate the registration options. In the controllers/index.js file, add the following code:

// Import necessary modules and functions
import {
  generateRegistrationOptions,
  verifyRegistrationResponse,
} from "@simplewebauthn/server";
import {
  bufferToBase64URLString,
  base64URLStringToBuffer,
} from "@simplewebauthn/browser";
import { v4 } from "uuid";
import User from "../models/User.js";
import PassKey from "../models/PassKey.js";

// Human-readable title for your website
const relyingPartyName = "WebAuthn Demo";
// A unique identifier for your website
const relyingPartyID = "localhost";
// The URL at which registrations and authentications should occur
const origin = `http://${relyingPartyID}`;

// Controller function to generate registration options
export const generateRegistrationOptionsCtrl = async (req, res) => {
  const { username } = req.query;
  const user = await User.findOne({ username });
  let userAuthenticators = [];

  // Retrieve authenticators used by the user before, if any
  if (user) {
    userAuthenticators = [...user.authenticators];
  }

  // Generate a unique ID for the current user session
  let currentUserId;
  if (!req.session.currentUserId) {
    currentUserId = v4();
    req.session.currentUserId = currentUserId;
  } else {
    currentUserId = req.session.currentUserId;
  }

  // Generate registration options
  const options = await generateRegistrationOptions({
    rpName: relyingPartyName,
    rpID: relyingPartyID,
    userID: currentUserId,
    userName: username,
    timeout: 60000,
    attestationType: "none", // Don't prompt users for additional information
    excludeCredentials: userAuthenticators.map((authenticator) => ({
      id: authenticator.credentialID,
      type: "public-key",
      transports: authenticator.transports,
    })),
    supportedAlgorithmIDs: [-7, -257],
    authenticatorSelection: {
      residentKey: "preferred",
      userVerification: "preferred",
    },
  });

  // Save the challenge to the session
  req.session.challenge = options.challenge;
  res.send(options);
};

First, we import the necessary functions and modules from libraries like @simplewebauthn/server and uuid. These help us handle the authentication process smoothly.

Next, we define some constants. relyingPartyName is a friendly name for our website. In this case, it's set to "WebAuthn Demo." relyingPartyID is a unique identifier for our website. Here, it's set to "localhost". Then, we construct the origin variable, the URL where registrations and authentications will happen. In this case, it's constructed using the relying party ID.

Moving on to the main part of the code, we have the controller generateRegistrationOptionsCtrl(). It's responsible for generating user registration options.

Inside this function, we first extract the username from the request. Then, we try to find the user in our database using this username. If we find the user, we retrieve the authenticators they've used before. Otherwise, we initialize an empty array for user authenticators.

Next, we generate a unique ID for the current user session. If there's no ID stored in the session yet, we generate a new one using the v4 function from the uuid library and store it in the session. Otherwise, we retrieve the ID from the session.

Then, we use the generateRegistrationOptions() function to create user registration options. After generating these options, we save the challenge to the session and send the options back as a response.

Next, we'll need to add the verifyRegistrationCtrl() controller to handle verifying the response sent from the browser after the user has initiated the registration:

// Controller function to verify registration
export const verifyRegistrationCtrl = async (req, res) => {
  const body = req.body;
  const { username } = req.query;
  const user = await User.findOne({ username });
  const expectedChallenge = req.session.challenge;

  // Check if the expected challenge exists
  if (!expectedChallenge) {
    return res.status(400).send({ error: "Failed: No challenge found" });
  }

  let verification;

  try {
    const verificationOptions = {
      response: body,
      expectedChallenge: `${expectedChallenge}`,
      expectedOrigin: origin,
      expectedRPID: relyingPartyID,
      requireUserVerification: false,
    };
    verification = await verifyRegistrationResponse(verificationOptions);
  } catch (error) {
    console.error(error);
    return res.status(400).send({ error: error.message });
  }

  const { verified, registrationInfo } = verification;

  // If registration is verified, update user data
  if (verified && registrationInfo) {
    const {
      credentialPublicKey,
      credentialID,
      counter,
      credentialBackedUp,
      credentialDeviceType,
    } = registrationInfo;

    const credId = bufferToBase64URLString(credentialID);
    const credPublicKey = bufferToBase64URLString(credentialPublicKey);

    const newDevice = {
      credentialPublicKey: credPublicKey,
      credentialID: credId,
      counter,
      transports: body.response.transports,
    };

    // Check if the device already exists for the user
    const existingDevice = user?.authenticators.find(
      (authenticator) => authenticator.credentialID === credId
    );

    if (!existingDevice && user) {
      // Add the new device to the user's list of devices
      await User.updateOne(
        { _id: user._id },
        { $push: { authenticators: newDevice } }
      );
      await PassKey.create({
        counter,
        credentialID: credId,
        user: user._id,
        webAuthnUserID: req.session.currentUserId,
        publicKey: credPublicKey,
        backedUp: credentialBackedUp,
        deviceType: credentialDeviceType,
        transports: body.response.transports,
        authenticators: [newDevice],
      });
    } else {
      const newUser = await User.create({
        username,
        authenticators: [newDevice],
      });
      await PassKey.create({
        counter,
        credentialID: credId,
        user: newUser._id,
        webAuthnUserID: req.session.currentUserId,
        publicKey: credPublicKey,
        backedUp: credentialBackedUp,
        deviceType: credentialDeviceType,
        transports: body.response.transports,
        authenticators: [newDevice],
      });
    }
  }

  // Clear the challenge from the session
  req.session.challenge = undefined;
  res.send({ verified });
};

The verifyRegistrationCtrl() controller searches for a user in the database based on the provided username. If found, it retrieves the expected challenge from the session data. If there's no expected challenge, the function returns an error. It then sets up verification options and calls a function named verifyRegistrationResponse.

If an error occurs, it logs the error and sends a response with the error message. If the registration is successfully verified, the function updates the user's data with the information provided in the registration response. It adds the new device to the user's list of devices if it does not exist.

Finally, the challenge is cleared from the session, and a response indicates whether the registration was successfully verified.

Before we head back to the browser to test what we've done so far, return to the app.js file and add the following code to register the routes:

import router from "./routes/index.js"; // place this at the start of the file

app.use("/api", router); // place this before the call to `app.listen()`

Now that we've assembled all the pieces for the registration functionality, we can return to the browser to test it out.

When you enter a username and click the "register" button, you should see a prompt similar to the one shown below:  

  To create a new passkey, you can now scan the QR code with your Android or iOS device. Upon successfully creating the passkey, a response is sent from the startRegistration() method to the /register/verify endpoint. Still, you'll notice it fails because of the error sent from the API:

{
    "error": "Unexpected registration response origin \"http://localhost:8030\", expected \"http://localhost\""
}

Why this is happening is because the origin that the verifyRegistrationResponse() method expected, which is http://localhost, is different from http://localhost:8010, was sent.

So, you might wonder why we can't just change it to http://localhost:8010. That’s because when we defined the origin in the controllers/index.js file, the relyingPartyID was set to "localhost", and we can't explicitly specify the port for the relying party ID.

An approach to get around this issue is to use a web tunneling service like tunnelmole or ngrok to expose our local server to the internet with a publicly accessible URL so we don't have to specify the port when defining the relyingPartyID.

Exposing your local server to the internet

Let's quickly set up tunnelmole to share the server on our local machine to a live URL.

First, let's install tunnelmole by entering the command below in your terminal:

sudo npm install -g tunnelmole

Next, enter the command below to make the server running locally available on the internet:

tmole <port>

You should see an output like this from your terminal if it was successful: You can now use the tunnelmole URL as the origin:

const relyingPartyID = "randomstring.tunnelmole.net"; // use output from your terminal
const origin = `https://${relyingPartyID}`; // webauthn only works with https

Everything should work as expected, so head back to your browser to start the registration process. Once you're done, an alert should pop up informing you that the registration was successful and that you can now log in:  

We've successfully set up the user registration feature. The only thing left to do is implement the logging-in functionality.

Building the login functionality

The login process will follow a similar flow to the registration process. First, we’ll request authentication options from the server to be passed to the authenticator on your device.

Afterward, a request will be sent to the server to verify the authenticator's response. If all the criteria are met, the user can log in successfully.

Head back to the public/script.js file, and include the function to handle when the "login" button is clicked:

async function handleLogin(evt) {
  errorDiv.textContent = "";
  errorDiv.style.display = "none";
  const userName = usernameInput.value;
  if (!browserSupportsWebAuthn()) {
    return alert("This browser does not support WebAuthn");
  }
  const resp = await fetch(`/api/login/start?username=${userName}`, {
    credentials: "include",
    headers: {
      "ngrok-skip-browser-warning": "69420",
    },
  });
  if (!resp.ok) {
    const error = (await resp.json()).error;
    errorDiv.textContent = error;
    errorDiv.style.display = "block";
    return;
  }
  let asseResp;
  try {
    asseResp = await startAuthentication(await resp.json());
  } catch (error) {
    errorDiv.textContent = error.message;
    errorDiv.style.display = "block";
  }
  if (!asseResp) {
    errorDiv.textContent = "Failed to connect with your device";
    errorDiv.style.display = "block";
    return;
  }
  const verificationResp = await fetch(
    `/api/login/verify?username=${userName}`,
    {
      credentials: "include",
      method: "POST",
      headers: {
        "Content-Type": "application/json",
        "ngrok-skip-browser-warning": "69420",
      },
      body: JSON.stringify(asseResp),
    }
  );
  const verificationJSON = await verificationResp.json();
  if (verificationJSON && verificationJSON.verified) {
    const userName = verificationJSON.username;
    // Hide login form and show welcome message
    loginForm.style.display = "none";
    welcomeMessage.style.display = "block";
    usernameDisplay.textContent = userName;
  } else {
    errorDiv.textContent = "Oh no, something went wrong!";
    errorDiv.style.display = "block";
  }
}

The function starts by clearing error messages and retrieving the user's username from the form. It checks if the browser supports WebAuthn; if it does, it sends a request to the server to initiate the login process.

If the response from the server is successful, it attempts to authenticate the user. Upon successful authentication, it hides the login form and displays a welcome message with the user's name. Otherwise, it displays an error message to the user.

Next, head back to the routes/index.js file and add the routes for logging in:

router.get("/login/start", generateAuthenticationOptionsCtrl);
router.post("/login/verify", verifyAuthenticationCtrl);

Don't forget to update the imports, as you're including the code above. Let's continue by adding the code to generate the authentication options. Go to the controllers/index.js file and add the following code:

// Controller function to generate authentication options
export const generateAuthenticationOptionsCtrl = async (req, res) => {
  const { username } = req.query;
  const user = await User.findOne({ username });
  if (!user) {
    return res
      .status(404)
      .send({ error: "User with this username does not exist" });
  }
  const options = await generateAuthenticationOptions({
    rpID: relyingPartyID,
    timeout: 60000,
    allowCredentials: user.authenticators.map((authenticator) => ({
      id: base64URLStringToBuffer(authenticator.credentialID),
      transports: authenticator.transports,
      type: "public-key",
    })),
    userVerification: "preferred",
  });
  req.session.challenge = options.challenge;
  res.send(options);
};

The generateAuthenticationOptionsCtrl() controller starts by extracting the username from the request query and searching for the user in the database. If found, it proceeds to generate authentication options crucial for the process.

These options include the relying party ID (rpID), timeout, allowed credentials derived from stored authenticators, and user verification option set to preferred. Then, it stores the challenge from the options in the session for authentication verification and sends them as a response to the browser.

Let's add the controller for verifying the authenticator's response for the final part of the auth flow:

// Controller function to verify authentication
export const verifyAuthenticationCtrl = async (req, res) => {
  const body = req.body;
  const { username } = req.query;
  const user = await User.findOne({ username });
  if (!user) {
    return res
      .status(404)
      .send({ error: "User with this username does not exist" });
  }
  const passKey = await PassKey.findOne({
    user: user._id,
    credentialID: body.id,
  });
  if (!passKey) {
    return res
      .status(400)
      .send({ error: "Could not find passkey for this user" });
  }
  const expectedChallenge = req.session.challenge;
  let dbAuthenticator;
  // Check if the authenticator exists in the user's data
  for (const authenticator of user.authenticators) {
    if (authenticator.credentialID === body.id) {
      dbAuthenticator = authenticator;
      dbAuthenticator.credentialPublicKey = base64URLStringToBuffer(
        authenticator.credentialPublicKey
      );
      break;
    }
  }
  // If the authenticator is not found, return an error
  if (!dbAuthenticator) {
    return res.status(400).send({
      error: "This authenticator is not registered with this site",
    });
  }
  let verification;
  try {
    const verificationOptions = {
      response: body,
      expectedChallenge: `${expectedChallenge}`,
      expectedOrigin: origin,
      expectedRPID: relyingPartyID,
      authenticator: dbAuthenticator,
      requireUserVerification: false,
    };
    verification = await verifyAuthenticationResponse(verificationOptions);
  } catch (error) {
    console.error(error);
    return res.status(400).send({ error: error.message });
  }
  const { verified, authenticationInfo } = verification;
  if (verified) {
    // Update the authenticator's counter in the DB to the newest count in the authentication
    dbAuthenticator.counter = authenticationInfo.newCounter;
    const filter = { username };
    const update = {
      $set: {
        "authenticators.$[element].counter": authenticationInfo.newCounter,
      },
    };
    const options = {
      arrayFilters: [{ "element.credentialID": dbAuthenticator.credentialID }],
    };
    await User.updateOne(filter, update, options);
  }
  // Clear the challenge from the session
  req.session.challenge = undefined;
  res.send({ verified, username: user.username });
};

The verifyAuthenticationCtrl() controller first extracts data from the request body and query, including the username and authentication details. It then searches for the user in the database. If not found, it returns a 404 error.

Assuming the user exists, it proceeds to find the passkey associated with the user and provides authentication details. If no passkey is found, it returns a 400 error.

Then, the expected challenge value is retrieved from the session data and iterates over the user's authenticators to find a match.

After attempting the verification, if an error occurs, the error is logged to the console and a 400 error is returned. If the verification is successful, the authenticator's counter is updated in the database, and the challenge is cleared from the session. Finally, the response includes the verification status and the username.

Return to your browser to ensure that everything functions as expected. Below is a GIF demonstrating the entire authentication process:  

We've successfully implemented the WebAuthn authentication, providing our users with a fast, secure, and password-less way to authenticate themselves. With biometric information or physical security keys, users can access their accounts securely.

Benefits and limitations of WebAuthn

While WebAuthn presents a solution to modern authentication challenges, it's essential to understand its strengths and weaknesses. Below, we highlight the key advantages and potential drawbacks of adopting WebAuthn in your authentication strategy.

Benefits of WebAuthn

WebAuthn offers a higher security level than traditional password-based authentication methods because of how it leverages public key cryptography to mitigate the risks associated with password breaches and phishing attacks.

Ainsi, même en cas de cyberattaque, les auteurs n'auront accès qu'à votre clé publique qui, à elle seule, est insuffisante pour accéder à votre compte.

La prise en charge de divers facteurs d'authentification tels que les données biométriques et les clés de sécurité physiques offre le type de flexibilité qui vous permet de mettre en œuvre une authentification multifacteur pour une sécurité accrue.

Étant donné que WebAuthn est actuellement pris en charge par la plupart des navigateurs et plates-formes Web modernes, cela le rend accessible à de nombreux utilisateurs. L'expérience d'authentification est également la même sur différents appareils et systèmes d'exploitation pour garantir la cohérence.

Limites de WebAuthn

L'intégration de WebAuthn peut s'avérer techniquement difficile pour les organisations disposant de systèmes complexes ou existants. Imaginez ensuite tous les types d'appareils que vos utilisateurs peuvent utiliser et toutes les autres limitations techniques associées.

Une autre limitation importante est l'aspect humain : dans quelle mesure le processus d'authentification est-il accessible pour vos utilisateurs ? La méconnaissance de la technologie peut soit décourager les utilisateurs, soit nécessiter la création de ressources éducatives et pédagogiques.

Conclusion

Dans cet article, nous avons vu comment WebAuthn fournit un processus d'authentification sans mot de passe qui utilise la cryptographie à clé publique sous le capot pour une expérience de connexion sécurisée et pratique. Avec un exemple pratique et des explications claires, nous avons expliqué comment configurer WebAuthn dans une application Web pour profiter d'un moyen plus fluide et plus sûr de s'authentifier dans nos applications.

---

LogRocket : déboguez plus facilement les erreurs JavaScript en comprenant le contexte

Le débogage du code est toujours une tâche fastidieuse. Mais plus vous comprenez vos erreurs, plus il est facile de les corriger.

LogRocket vous permet de comprendre ces erreurs de manière nouvelle et unique. Notre solution de surveillance front-end suit l'engagement des utilisateurs avec vos frontends JavaScript pour vous donner la possibilité de voir exactement ce que l'utilisateur a fait qui a conduit à une erreur.

LogRocket enregistre les journaux de la console, les temps de chargement des pages, les traces de pile, les requêtes/réponses réseau lentes avec en-têtes + corps, les métadonnées du navigateur et les journaux personnalisés. Comprendre l'impact de votre code JavaScript n'aura jamais été aussi simple !

Essayez-le gratuitement.

Ce qui précède est le contenu détaillé de. pour plus d'informations, suivez d'autres articles connexes sur le site Web de PHP en chinois!