This gem provides a toolkit for adding OAuth2 provider capabilities to a Ruby web app. It handles most of the protocol for you: it is designed to provide a sufficient level of abstraction that it can implement updates to the protocol without affecting your application code at all. All you have to deal with is authenticating your users and letting them grant access to client apps.
It is also designed to be usable within any web frontend, at least those of Rails and Sinatra. It assumes very little about the request objects in your environment, namely they:
Hashof request parameters
#urlwith the full URL string of the request
#env, returning the HTTP environment
It stores the clients and authorizations using ActiveRecord; the schema is in
lib/oauth2/model/schema.rb. Run it to update your database to store OAuth data:
The current imeplementation is based on draft-10.
A basic example is in
example/application.rb. To implement OAuth, you need to provide four things:
Some UI to register client applications
The OAuth request endpoint
A flow for logged-in users to grant access to clients
Resources protected by access tokens
OAuth2::Provider requires very little configuration. The only thing it needs to know about your app is its name, which is used in the headers for some authentication errors. To load the library, just do this:
require 'oauth2/provider' ::. = 'My OAuth app'
You may also need to configure assertion handlers if your application supports third-party access credentials. See 'Using Assertions' below.
Registering client applications
Clients are modelled by the
OAuth2::Model::Client class, which is an ActiveRecord model. You just need to implement a UI for creating them, for example in a Sinatra app:
get '/oauth/apps/new' do @client = ::::.new erb :new_client end post '/oauth/apps' do @client = ::::.new(params) @client.save ? erb(:show_client) : erb(:new_client) end
Client applications must have a
name and a
redirect_uri: provide fields for editing these but do not allow the other fields to be edited, since they are the client's access credentials. When you've created the client, you should show its details to the user registering the client: its
client_secret (the last two are generated for you).
client_secret is not stored in plain text so you can only read it when you initially create the client object.
OAuth request endpoint
This is a path that your application exposes in order for clients to communicate with your application. It is also the page that the client will send users to so they can authenticate and grant access. Many requests to this endpoint will be protocol-level requests that do not involve the user, and
OAuth2::Provider gives you a generic way to handle all that.
You should use this to get the right response, status code and headers to send to the client. In the event that
OAuth2::Provider does not provide a response, you should render a page that lets the user begin to authenticate and grant access.
This endpoint must be accessible via GET and POST. In this example we will expose the OAuth service through the path
/oauth/authorize. We check if there is a logged-in resource owner and give this to
OAuth::Provider, since we may be able to immediately redirect if the user has already authorized the client:
[:get, :post].each do |method| __send__ method, '/oauth/authorize' do @owner = User.find_by_id(session[:user_id]) @oauth2 = ::.(@owner, request) redirect @oauth2.redirect_uri if @oauth2.redirect? headers @oauth2.response_headers status @oauth2.response_status @oauth2.response_body || erb(:login) end end
There is a set of parameters that you will need to hold on to for when your app needs to redirect back to the client. You could store them in the session, or pass them through forms as the user completes the flow. For example to embed them in the login form, do this:
<% @oauth2.params.each do |key, value| %> <input type="hidden" name="<%= key %>" value="<%= value %>"> <% end %>
You may also want to use scopes to provide granular access to your domain using scopes. The
@oauth2 object exposes the scopes the client has asked for so you can display them to the user:
<p>The application <%= @oauth2.client.name %> wants the following permissions:</p> <ul> <% @oauth2.scopes.each do |scope| %> <li><%= PERMISSION_UI_STRINGS[scope] %></li> <% end %> </ul>
You can also use the method
@oauth2.unauthorized_scopes to get the list of scopes the user has not already granted to the client, in the case where the client already has some authorization. If no prior authorization exists between the user and the client,
@oauth2.unauthorized_scopes just returns all the scopes the client has asked for.
Granting access to clients
Your application will probably have some concept of a user, or a resource owner in OAuth lingo. Add this mixin to the model that represents your users:
class User < ActiveRecord::Base include :::: end
This just adds a couple of relations and methods to the model to let it interact with the
Once the user has authenticated you should show them a page to let them grant or deny access to the client application. This is straightforward; let's say the user checks a box before posting a form to indicate their intent:
post '/oauth/allow' do @user = User.find_by_id(session[:user_id]) @auth = ::::.(@user, params) if params['allow'] == '1' @auth.grant_access! else @auth.deny_access! end redirect @auth.redirect_uri end
After granting or denying access, we just redirect back to the client using a URI that
OAuth2::Provider will provide for you.
Using password credentials
If you like, OAuth lets you use a user's login credentials to authenticate with a provider. In this case the client application must request these credentials directly from the user and then post them to the exchange endpoint. On the provider side you can handle this using the
grant_access! API methods, for example:
::. do |client, username, password| user = User.find_by_username(username) if user.authenticate?(password) user.grant_access!(client) else nil end end
The block must return
user.grant_access!(client) if you want to allow access, otherwise it should return
Assertions provide a way to access your OAuth services using user credentials from another service. When using assertions, the user will not authenticate on your web site; the OAuth client will authenticate the user using some other framework and obtain a token, then exchange this token for an access token on your domain.
For example, a client application may let a user authenticate using Facebook, so the application obtains a Facebook access token from the user. The client would then pass this token to your OAuth endpoint and exchange it for an access token from your site. You will typically create an account in your database to represent this, then have that new account grant access to the client.
To use assertions, you must tell
OAuth2::Provider how to handle assertions based on their type. An assertion type must be a valid URI. For the Facebook example we'd do the following. The block yields the
Client object making the exchange request, and the value of the assertion, which in this example will be a Facebook access token.
::. 'https://graph.facebook.com/me' do |client, assertion| facebook = URI.parse('https://graph.facebook.com/me?access_token=' + assertion) response = Net::HTTP.get_response(facebook) user_data = JSON.parse(response.body) account = User.from_facebook_data(user_data) account.grant_access!(client) end
This code should run when your app boots, not during a request handler - think of it as configuration for
OAuth2::Provider. The framework will invoke it when a client attempts to use assertions with your OAuth endpoint.
The final call in your handler should be to
grant_access!; this returns an
Authorization object that the framework then uses to complete the response to the client. If you want to deny the request for whatever reason, the block must return
nil. If a client tries to use an assertion type you have no handler for, the client will get an error response.
Protecting resources with access tokens
To protect the user's resources you need to check for access tokens. This is simple, for example a call to get a user's notes:
get '/user/:username/notes' do user = User.find_by_username(params[:username]) token = ::.(user, ['read_notes'], request) headers token.response_headers status token.response_status if token.valid? JSON.unparse('notes' => user.notes) else JSON.unparse('error' => 'No notes for you!') end end
OAuth2::Provider.access_token() takes a
ResourceOwner, a list of scopes required to access the resource, and a request object. If the token was not granted for the required scopes, has expired or is simply invalid, headers and a status code are set to indicate this to the client.
token.valid? is the call you should use to determine whether to server the request or not.
It is also common to provide a dynamic resource for getting some basic data about a user by supplying their access token. This can be done by passing
nil as the resource owner:
get '/me' do token = ::.(nil, , request) if token.valid? JSON.unparse('username' => token.owner.username) else JSON.unparse('error' => 'Keep out!') end end
token.owner returns the
ResourceOwner that issued the token. A token represents the fact that a single owner gave a single client a set of permissions.
Your application should ensure that any endpoint that receives or returns OAuth data is only accessible over a secure transport such as the
OAuth2::Provider can enforces this to make it easier to keep your users' data secure. If you want to enable these behaviours, set
OAuth2::Provider.enforce_ssl = true.
OAuth2::Provider.parsemethod will produce error responses and will not process the incoming request unless the request was made using the
An access token constructed using
#valid?unless the request was made using the
Any access token received over an insecure connection is immediately destroyed to prevent eavesdroppers getting access to the user's resources. A client making an insecure request will have to send the user through the authorization process again to get a new token.
Copyright © 2010-2011 Songkick.com
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the “Software”), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.