Authentication and authorization
API docs
You are currently viewing the documentation for the unreleased version 5.0.0.CR2 of Vert.x. Visit the latest stable version of this page.

OAuth2 auth provider

This component contains an out of the box OAuth2 (and to some extent OpenID Connect) relying party implementation. To use this project, add the following dependency to the dependencies section of your build descriptor:

  • Maven (in your pom.xml):

<dependency>
  <groupId>io.vertx</groupId>
  <artifactId>vertx-auth-oauth2</artifactId>
  <version>5.0.0.CR2</version>
</dependency>

  • Gradle (in your build.gradle file):

compile 'io.vertx:vertx-auth-oauth2:5.0.0.CR2'

OAuth2 lets users grant the access to the desired resources to third party applications, giving them the possibility to enable and disable those accesses whenever they want.

Vert.x OAuth2 supports the following flows.

  • Authorization Code Flow (for apps with servers that can store persistent information).

  • Password Credentials Flow (when previous flow can’t be used or during development).

  • Client Credentials Flow (the client can request an access token using only its client credentials)

The same code will work with OpenID Connect https://openid.net/connect/ servers and supports the Discovery protocol as specified in http://openid.net/specs/openid-connect-discovery-1_0.html .

Authorization Code Flow

The authorization code grant type is used to obtain both access tokens and refresh tokens and is optimized for confidential clients. As a redirection-based flow, the client must be capable of interacting with the resource owner’s user-agent (typically a web browser) and capable of receiving incoming requests (via redirection) from the authorization server.

For more details see Oauth2 specification, section 4.1.

Password Credentials Flow

The resource owner password credentials grant type is suitable in cases where the resource owner has a trust relationship with the client, such as the device operating system or a highly privileged application. The authorization server should take special care when enabling this grant type, and only allow it when other flows are not viable.

The grant type is suitable for clients capable of obtaining the resource owner’s credentials (username and password, typically using an interactive form). It is also used to migrate existing clients using direct authentication schemes such as HTTP Basic or Digest authentication to OAuth by converting the stored credentials to an access token.

For more details see Oauth2 specification, section 4.3.

Client Credentials Flow

The client can request an access token using only its client credentials (or other supported means of authentication) when the client is requesting access to the protected resources under its control, or those of another resource owner that have been previously arranged with the authorization server (the method of which is beyond the scope of this specification).

The client credentials grant type MUST only be used by confidential clients.

For more details see Oauth2 specification, section 4.4.

JWT (on behalf of) Flow

The client can request an access token using JWTs also known as "on-behalf-of".

Extensions

The provider supports RFC7523 an extension to allow server to server authorization based on JWT.

Getting Started

An example on how to use this provider and authenticate with GitHub can be implemented as:

OAuth2Auth oauth2 = OAuth2Auth.create(vertx, new OAuth2Options()
  .setClientId("YOUR_CLIENT_ID")
  .setClientSecret("YOUR_CLIENT_SECRET")
  .setSite("https://github.com/login")
  .setTokenPath("/oauth/access_token")
  .setAuthorizationPath("/oauth/authorize")
);

// when there is a need to access a protected resource
// or call a protected method, call the authZ url for
// a challenge

String authorization_uri = oauth2.authorizeURL(new OAuth2AuthorizationURL()
  .setRedirectUri("http://localhost:8080/callback")
  .addScope("notifications")
  .setState("3(#0/!~"));

// when working with web application use the above string as a redirect url

// in this case GitHub will call you back in the callback uri one
// should now complete the handshake as:

// the code is provided as a url parameter by github callback call
String code = "xxxxxxxxxxxxxxxxxxxxxxxx";

oauth2.authenticate(
    new Oauth2Credentials()
      .setCode(code)
      .setRedirectUri("http://localhost:8080/callback"))
  .onSuccess(user -> {
    // save the token and continue...
  })
  .onFailure(err -> {
    // error, the code provided is not valid
  });

Authorization Code flow

The Authorization Code flow is made up from two parts. At first your application asks to the user the permission to access their data. If the user approves the OAuth2 server sends to the client an authorization code. In the second part, the client POST the authorization code along with its client secret to the authority server in order to get the access token.

OAuth2Options credentials = new OAuth2Options()
  .setClientId("<client-id>")
  .setClientSecret("<client-secret>")
  .setSite("https://api.oauth.com");


// Initialize the OAuth2 Library
OAuth2Auth oauth2 = OAuth2Auth.create(vertx, credentials);

// Authorization oauth2 URI
String authorization_uri = oauth2.authorizeURL(new OAuth2AuthorizationURL()
  .setRedirectUri("http://localhost:8080/callback")
  .addScope("<scope>")
  .setState("<state>"));

// Redirect example using Vert.x
response.putHeader("Location", authorization_uri)
  .setStatusCode(302)
  .end();

Credentials tokenConfig = new Oauth2Credentials()
  .setCode("<code>")
  .setRedirectUri("http://localhost:3000/callback");

// Callbacks
// Save the access token
oauth2.authenticate(tokenConfig)
  .onSuccess(user -> {
    // Get the access token object
    // (the authorization code is given from the previous step).
  })
  .onFailure(err -> {
    System.err.println("Access Token Error: " + err.getMessage());
  });

Password Credentials Flow

This flow is suitable when the resource owner has a trust relationship with the client, such as its computer operating system or a highly privileged application. Use this flow only when other flows are not viable or when you need a fast way to test your application.

OAuth2Auth oauth2 = OAuth2Auth.create(vertx);

Credentials tokenConfig = new UsernamePasswordCredentials(
  "username", "password");

oauth2.authenticate(tokenConfig)
  .onSuccess(user -> {
    // Get the access token object
    // (the authorization code is given from the previous step).

    // you can now make requests using the
    // `Authorization` header and the value:
    String httpAuthorizationHeader = user.principal()
      .getString("access_token");

  })
  .onFailure(err -> {
    System.err.println("Access Token Error: " + err.getMessage());
  });

Client Credentials Flow

This flow is suitable when client is requesting access to the protected resources under its control.

OAuth2Options credentials = new OAuth2Options()
  .setClientId("<client-id>")
  .setClientSecret("<client-secret>")
  .setSite("https://api.oauth.com");


// Initialize the OAuth2 Library
OAuth2Auth oauth2 = OAuth2Auth.create(vertx, credentials);

Credentials tokenConfig = new TokenCredentials("<token>");

oauth2.authenticate(tokenConfig)
  .onSuccess(user -> {
    // Success
  })
  .onFailure(err -> {
    System.err.println("Access Token Error: " + err.getMessage());
  });

OpenID Connect Discovery

There is limited support for OpenID Discovery servers. Using OIDC Discovery will simplify the configuration of your auth module into a single line of code, for example, consider setting up your auth using Google:

OpenIDConnectAuth.discover(
    vertx,
    new OAuth2Options()
      .setClientId("clientId")
      .setClientSecret("clientSecret")
      .setSite("https://accounts.google.com"))
  .onSuccess(oauth2 -> {
    // the setup call succeeded.
    // at this moment your auth is ready to use and
    // google signature keys are loaded so tokens can be decoded and verified.
  })
  .onFailure(err -> {
    // the setup failed.
  });

Behind the scenes a couple of actions are performed:

  1. HTTP get request to the .well-known/openid-configuration resource

  2. Validation of the response issuer field as mandated by the spec (the issuer value must match the request one)

  3. If the JWK uri is present, keys are loaded from the server and added to the auth keychain

  4. the auth module is configure and returned to the user.

A couple of well known OpenID Connect Discovery providers are:

This and the given client id/client secret is enough to configure your auth provider object.

For these well known providers a shortcut is provided:

KeycloakAuth.discover(
    vertx,
    new OAuth2Options()
      .setClientId("clientId")
      .setClientSecret("clientSecret")
      .setSite("https://keycloakhost:keycloakport/auth/realms/{realm}")
      .setTenant("your-realm"))
  .onSuccess(oauth2 -> {
    // ...
  });

// Google example
GoogleAuth.discover(
    vertx,
    new OAuth2Options()
      .setClientId("clientId")
      .setClientSecret("clientSecret"))
  .onSuccess(oauth2 -> {
    // ...
  });

// Salesforce example
SalesforceAuth.discover(
    vertx,
    new OAuth2Options()
      .setClientId("clientId")
      .setClientSecret("clientSecret"))
  .onSuccess(oauth2 -> {
    // ...
  });

// Azure AD example
AzureADAuth.discover(
    vertx,
    new OAuth2Options()
      .setClientId("clientId")
      .setClientSecret("clientSecret")
      .setTenant("your-app-guid"))
  .onSuccess(oauth2 -> {
    // ...
  });

// IBM Cloud example
IBMCloudAuth.discover(
    vertx,
    new OAuth2Options()
      .setClientId("clientId")
      .setClientSecret("clientSecret")
      .setSite("https://<region-id>.appid.cloud.ibm.com/oauth/v4/{tenant}")
      .setTenant("your-tenant-id"))
  .onSuccess(oauth2 -> {
    // ...
  });

User object

When a token expires we need to refresh it. OAuth2 offers the AccessToken class that add a couple of useful methods to refresh the access token when it is expired.

if (user.expired()) {
  // Callbacks
  oauth2.refresh(user)
    .onSuccess(refreshedUser -> {
      // the refreshed user is now available
    })
    .onFailure(err -> {
      // error handling...
    });
}

When you’ve done with the token or you want to log out, you can revoke the access token and refresh token.

oauth2.revoke(user, "access_token")
  .onSuccess(v -> {
    // Session ended. But the refresh_token is still valid.

    // Revoke the refresh_token
    oauth2.revoke(user, "refresh_token")
      .onSuccess(v2 -> {
        System.out.println("token revoked.");
      });
  });

Example configuration for common OAuth2 providers

For convenience there are several helpers to assist your with your configuration. Currently we provide:

JBoss Keycloak

When using this Keycloak the provider has knowledge on how to parse access tokens and extract grants from inside. This information is quite valuable since it allows to do authorization at the API level, for example:

JsonObject keycloakJson = new JsonObject()
  .put("realm", "master")
  .put("realm-public-key", "MIIBIjANBgkqhk...wIDAQAB")
  .put("auth-server-url", "http://localhost:9000/auth")
  .put("ssl-required", "external")
  .put("resource", "frontend")
  .put("credentials", new JsonObject()
    .put("secret", "2fbf5e18-b923-4a83-9657-b4ebd5317f60"));

// Initialize the OAuth2 Library
OAuth2Auth oauth2 = KeycloakAuth
  .create(vertx, OAuth2FlowType.PASSWORD, keycloakJson);

// first get a token (authenticate)
oauth2.authenticate(
    new UsernamePasswordCredentials("user", "secret"))
  .onSuccess(user -> {
    // now check for permissions
    AuthorizationProvider authz = KeycloakAuthorization.create();

    authz.getAuthorizations(user)
      .onSuccess(v -> {
        if (
          RoleBasedAuthorization.create("manage-account")
            .setResource("account")
            .match(user)) {
          // this user is authorized to manage its account
        }
      });
  });

We also provide a helper class for Keycloak so that we can we can easily retrieve decoded token and some necessary data (e.g. preferred_username) from the Keycloak principal. For example:

JsonObject idToken = user.attributes().getJsonObject("idToken");

// you can also retrieve some properties directly from the Keycloak principal
// e.g. `preferred_username`
String username = user.principal().getString("preferred_username");

Please remember that Keycloak does implement OpenID Connect, so you can configure it just by using it’s discovery url:

OpenIDConnectAuth.discover(
    vertx,
    new OAuth2Options()
      .setClientId("clientId")
      .setTenant("your_realm")
      .setSite("https://server:port/auth/realms/{tenant}"))
  .onSuccess(oauth2 -> {
    // the setup call succeeded.
    // at this moment your auth is ready to use
  });

Since you can deploy your Keycloak server anywhere, just replace server:port with the correct value and the your_realm value with your application realm.

Google Server to Server

The provider also supports Server to Server or the RFC7523 extension. This is a feature present on Google with their service account.

Token Introspection

Tokens can be introspected in order to assert that they are still valid. Although there is RFC7662 for this purpose not many providers implement it. Instead there are variations also known as TokenInfo end points. The OAuth2 provider will accept both end points as a configuration. Currently we are known to work with Google and Keycloak.

Token introspection assumes that tokens are opaque, so they need to be validated on the provider server. Every time a token is validated it requires a round trip to the provider. Introspection can be performed at the OAuth2 level or at the User level:

oauth2.authenticate(new TokenCredentials("opaque string"))
  .onSuccess(theUser -> {
    // token is valid!
  });

// User level
oauth2.authenticate(new TokenCredentials(user.<String>get("access_token")))
  .onSuccess(authenticatedUser -> {
    // Token is valid!
  });

Verifying JWT tokens

We’ve just covered how to introspect a token however when dealing with JWT tokens one can reduce the amount of trips to the provider server thus enhancing your overall response times. In this case tokens will be verified using the JWT protocol at your application side only. Verifying JWT tokens is cheaper and offers better performance, however due to the stateless nature of JWTs it is not possible to know if a user is logged out and a token is invalid. For this specific case one needs to use the token introspection if the provider supports it.

oauth2.authenticate(new TokenCredentials("jwt-token"))
  .onSuccess(theUser -> {
    // token is valid!
  });

Until now we covered mostly authentication, although the implementation is relying party (that means that the real authentication happens somewhere else), there is more you can do with the handler. For example you can also do authorization if the provider is known to support JSON web tokens. This is a common feature if your provider is a OpenId Connect provider or if the provider does support `access_token`s as JWTs.

Such provider is Keycloak that is a OpenId Connect implementation. In that case you will be able to perform authorization in a very easy way.

Role Based Access Control

OAuth2 is an AuthN protocol, however OpenId Connect adds JWTs to the token format which means that AuthZ can be encoded at the token level. Currently there are 2 known JWT AuthZ known formats:

  • Keycloak

  • MicroProfile JWT 1.1 spec (from the auth-jwt module)

Keycloak JWT

Given that Keycloak does provide JWT access_tokens one can authorize at two distinct levels:

  • role

  • authority

To distinct the two, the auth provider follows the same recommendations from the base user class, i.e.: use the`:` as a separator for the two. It should be noted that both role and authorities do not need to be together, in the most simple case an authority is enough.

In order to map to keycloak’s token format the following checks are performed:

  1. If no role is provided, it is assumed to the the provider realm name

  2. If the role is realm then the lookup happens in realm_access list

  3. If a role is provided then the lookup happends in the resource_access list under the role name

Check for a specific authorities

Here is one example how you can perform authorization after the user has been loaded from the oauth2 handshake, for example you want to see if the user can print in the current application:

if (PermissionBasedAuthorization.create("print").match(user)) {
  // Yes the user can print
}

However this is quite specific, you might want to verify if the user can add-user to the whole system (the realm):

if (
  PermissionBasedAuthorization.create("add-user")
    .setResource("realm")
    .match(user)) {
  // Yes the user can add users to the application
}

Or if the user can access the year-report in the finance department:

if (
  PermissionBasedAuthorization.create("year-report")
    .setResource("finance")
    .match(user)) {
  // Yes the user can access the year report from the finance department
}

MicroProfile JWT 1.1 spec

Another format in the form of a spec is the MP-JWT 1.1. This spec defines a JSON array of strings under the property name groups that define the "groups" the token has an authority over.

In order to use this spec to assert AuthZ use the AuthorizationProvider factory provided on the auth-jwt module.

Token Management

Check if it is expired

Tokens are usually fetched from the server and cached, in this case when used later they might have already expired and be invalid, you can verify if the token is still valid like this:

boolean isExpired = user.expired();

This call is totally offline, it could still happen that the Oauth2 server invalidated your token but you get a non expired token result. The reason behind this is that the expiration is checked against the token expiration dates, not before date and such values.

Refresh token

There are times you know the token is about to expire and would like to avoid to redirect the user again to the login screen. In this case you can refresh the token. To refresh a token you need to have already a user and call:

oauth2.refresh(user)
  .onSuccess(refreshedUser -> {
    // the refresh call succeeded
  })
  .onFailure(err -> {
    // the token was not refreshed, a best practise would be
    // to forcefully logout the user since this could be a
    // symptom that you're logged out by the server and this
    // token is not valid anymore.
  });

Revoke token

Since tokens can be shared across various applications you might want to disallow the usage of the current token by any application. In order to do this one needs to revoke the token against the Oauth2 server:

oauth2.revoke(user, "access_token")
  .onSuccess(v -> {
    // the revoke call succeeded
  })
  .onFailure(err -> {
    // the token was not revoked.
  });

It is important to note that this call requires a token type. The reason is because some providers will return more than one token e.g.:

  • id_token

  • refresh_token

  • access_token

So one needs to know what token to invalidate. It should be obvious that if you invalidate the refresh_token you’re still logged in but you won’t be able to refresh anymore, which means that once the token expires you need to redirect the user again to the login page.

Introspect

Introspect a token is similar to a expiration check, however one needs to note that this check is fully online. This means that the check happens on the OAuth2 server.

oauth2.authenticate(new TokenCredentials(user.<String>get("access_token")))
  .onSuccess(validUser -> {
    // the introspection call succeeded
  })
  .onFailure(err -> {
    // the token failed the introspection. You should proceed
    // to logout the user since this means that this token is
    // not valid anymore.
  });

Important note is that even if the expired() call is true the return from the introspect call can still be an error. This is because the OAuth2 might have received a request to invalidate the token or a loggout in between.

Logging out

Logging out is not a Oauth2 feature but it is present on OpenID Connect and most providers do support some sort of logging out. This provider also covers this area if the configuration is enough to let it make the call. For the user this is as simple as:

String logoutUrl = oauth2.endSessionURL(user);

Key Management

When the provider is configured with a jwks path. Either manually or using the discovery mechanism, there are events when keys must be rotated. For this reason this provider implements the 2 recommended ways by the openid connect core spec.

When calling the refresh method, if the server returns the recommended cache header as described on https://openid.net/specs/openid-connect-core-1_0.html#RotateEncKeys then a periodic task will run at the recommeneded time by the server to reload the keys.

boolean isExpired = user.expired();

However there are times when servers change keys and this provider isn’t aware. For example, to mitigate a leak or an expired certificate. In such event the server will start emitting tokens with a different kid than the ones on the store as described: https://openid.net/specs/openid-connect-core-1_0.html#RotateSigKeys For this situation and to avoid DDoS attacks the provider will notify you that a unknown key is missing:

oauth2.refresh(user)
  .onSuccess(refreshedUser -> {
    // the refresh call succeeded
  })
  .onFailure(err -> {
    // the token was not refreshed, a best practise would be
    // to forcefully logout the user since this could be a
    // symptom that you're logged out by the server and this
    // token is not valid anymore.
  });

A special note on this is that if a user will send many requests with a missing key, your handler should throttle the calls to refresh the new key set, or you might end up DDoS your IdP server.