Ruby on Rails Tutorial

Learn Web Development with Rails

Michael Hartl

Contents

Foreword

My former company (CD Baby) was one of the first to loudly switch to Ruby on Rails, and then even more loudly switch back to PHP (Google me to read about the drama). This book by Michael Hartl came so highly recommended that I had to try it, and the Ruby on Rails Tutorial is what I used to switch back to Rails again.

Though I’ve worked my way through many Rails books, this is the one that finally made me “get” it. Everything is done very much “the Rails way”—a way that felt very unnatural to me before, but now after doing this book finally feels natural. This is also the only Rails book that does test-driven development the entire time, an approach highly recommended by the experts but which has never been so clearly demonstrated before. Finally, by including Git, GitHub, and Heroku in the demo examples, the author really gives you a feel for what it’s like to do a real-world project. The tutorial’s code examples are not in isolation.

The linear narrative is such a great format. Personally, I powered through the Rails Tutorial in three long days, doing all the examples and challenges at the end of each chapter. Do it from start to finish, without jumping around, and you’ll get the ultimate benefit.

Enjoy!

Derek Sivers (sivers.org)
Founder, CD Baby

Acknowledgments

The Ruby on Rails Tutorial owes a lot to my previous Rails book, RailsSpace, and hence to my coauthor Aurelius Prochazka. I’d like to thank Aure both for the work he did on that book and for his support of this one. I’d also like to thank Debra Williams Cauley, my editor on both RailsSpace and the Ruby on Rails Tutorial; as long as she keeps taking me to baseball games, I’ll keep writing books for her.

I’d like to acknowledge a long list of Rubyists who have taught and inspired me over the years: David Heinemeier Hansson, Yehuda Katz, Carl Lerche, Jeremy Kemper, Xavier Noria, Ryan Bates, Geoffrey Grosenbach, Peter Cooper, Matt Aimonetti, Gregg Pollack, Wayne E. Seguin, Amy Hoy, Dave Chelimsky, Pat Maddox, Tom Preston-Werner, Chris Wanstrath, Chad Fowler, Josh Susser, Obie Fernandez, Ian McFarland, Steven Bristol, Pratik Naik, Sarah Mei, Sarah Allen, Wolfram Arnold, Alex Chaffee, Giles Bowkett, Evan Dorn, Long Nguyen, James Lindenbaum, Adam Wiggins, Tikhon Bernstam, Ron Evans, Wyatt Greene, Miles Forrest, the good people at Pivotal Labs, the Heroku gang, the thoughtbot guys, and the GitHub crew. Finally, many, many readers—far too many to list—have contributed a huge number of bug reports and suggestions during the writing of this book, and I gratefully acknowledge their help in making it as good as it can be.

About the author

Michael Hartl is the author of the Ruby on Rails Tutorial, the leading introduction to web development with Ruby on Rails. His prior experience includes writing and developing RailsSpace, an extremely obsolete Rails tutorial book, and developing Insoshi, a once-popular and now-obsolete social networking platform in Ruby on Rails. In 2011, Michael received a Ruby Hero Award for his contributions to the Ruby community. He is a graduate of Harvard College, has a Ph.D. in Physics from Caltech, and is an alumnus of the Y Combinator entrepreneur program.

Copyright and license

Ruby on Rails Tutorial: Learn Web Development with Rails. Copyright © 2012 by Michael Hartl. All source code in the Ruby on Rails Tutorial is available jointly under the MIT License and the Beerware License.

The MIT License

Copyright (c) 2012 Michael Hartl

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.

/*
 * ----------------------------------------------------------------------------
 * "THE BEER-WARE LICENSE" (Revision 42):
 * Michael Hartl wrote this code. As long as you retain this notice you
 * can do whatever you want with this stuff. If we meet some day, and you think
 * this stuff is worth it, you can buy me a beer in return.
 * ----------------------------------------------------------------------------
 */

Chapter 9 Updating, showing, and deleting users

In this chapter, we will complete the REST actions for the Users resource (Table 7.1) by adding edit, update, index, and destroy actions. We’ll start by giving users the ability to update their profiles, which will also provide a natural opportunity to enforce a security model (made possible by the authorization code in Chapter 8). Then we’ll make a listing of all users (also requiring authorization), which will motivate the introduction of sample data and pagination. Finally, we’ll add the ability to destroy users, wiping them clear from the database. Since we can’t allow just any user to have such dangerous powers, we’ll take care to create a privileged class of administrative users (admins) authorized to delete other users.

To get started, let’s start work on an updating-users topic branch:

$ git checkout -b updating-users

9.1 Updating users

The pattern for editing user information closely parallels that for creating new users (Chapter 7). Instead of a new action rendering a view for new users, we have an edit action rendering a view to edit users; instead of create responding to a POST request, we have an update action responding to a PUT request (Box 3.2). The biggest difference is that, while anyone can sign up, only the current user should be able to update his information. This means that we need to enforce access control so that only authorized users can edit and update; the authentication machinery from Chapter 8 will allow us to use a before filter to ensure that this is the case.

9.1.1 Edit form

We start with the edit form, whose mockup appears in Figure 9.1.¹ As usual, we’ll begin with some tests. First, note the link to change the Gravatar image; if you poke around the Gravatar site, you’ll see that the page to add or edit images is located at http://gravatar.com/emails, so we test the edit page for a link with that URI.²

Figure 9.1: A mockup of the user edit page. (full size)

The tests for the edit user form are analogous to the test for the new user form in Listing 7.30 from the Chapter 7 exercises, which added a test for the error message on invalid submission. The result appears in Listing 9.1.

require 'spec_helper'

describe "User pages" do
  .
  .
  .
  describe "edit" do
    let(:user) { FactoryGirl.create(:user) }
    before { visit edit_user_path(user) }

    describe "page" do
      it { should have_selector('h1',    text: "Update your profile") }
      it { should have_selector('title', text: "Edit user") }
      it { should have_link('change', href: 'http://gravatar.com/emails') }
    end

    describe "with invalid information" do
      before { click_button "Save changes" }

      it { should have_content('error') }
    end
  end
end

To write the application code, we need to fill in the edit action in the Users controller. Note from Table 7.1 that the proper URI for a user’s edit page is /users/1/edit (assuming the user’s id is 1). Recall that the id of the user is available in the params[:id] variable, which means that we can find the user with the code in Listing 9.2.

class UsersController < ApplicationController
  .
  .
  .
  def edit
    @user = User.find(params[:id])
  end
end

Getting the tests to pass requires making the actual edit view, shown in Listing 9.3. Note how closely this resembles the new user view from Listing 7.17; the large overlap suggests factoring the repeated code into a partial, which is left as an exercise (Section 9.6).

<% provide(:title, "Edit user") %> 
<h1>Update your profile</h1>

<div class="row">
  <div class="span6 offset3">
    <%= form_for(@user) do |f| %>
      <%= render 'shared/error_messages' %>

      <%= f.label :name %>
      <%= f.text_field :name %>

      <%= f.label :email %>
      <%= f.text_field :email %>

      <%= f.label :password %>
      <%= f.password_field :password %>

      <%= f.label :password_confirmation, "Confirm Password" %>
      <%= f.password_field :password_confirmation %>

      <%= f.submit "Save changes", class: "btn btn-large btn-primary" %>
    <% end %>

    <%= gravatar_for @user %>
    <a href="http://gravatar.com/emails">change</a>
  </div>
</div>

Here we have reused the shared error_messages partial introduced in Section 7.3.2.

With the @user instance variable from Listing 9.2, the edit page tests from Listing 9.1 should pass:

$ bundle exec rspec spec/requests/user_pages_spec.rb -e "edit page"

The corresponding page appears in Figure 9.2, which shows how Rails automatically pre-fills the Name and Email fields using the attributes of the @user variable.

Figure 9.2: The initial user edit page with pre-filled name & email. (full size)

Looking at the HTML source for Figure 9.2, we see a form tag as expected (Listing 9.4).

<form action="/users/1" class="edit_user" id="edit_user_1" method="post">
  <input name="_method" type="hidden" value="put" />
  .
  .
  .
</form>

Note here the hidden input field

<input name="_method" type="hidden" value="put" />

Since web browsers can’t natively send PUT requests (as required by the REST conventions from Table 7.1), Rails fakes it with a POST request and a hidden input field.³

There’s another subtlety to address here: the code form_for(@user) in Listing 9.3 is exactly the same as the code in Listing 7.17—so how does Rails know to use a POST request for new users and a PUT for editing users? The answer is that it is possible to tell whether a user is new or already exists in the database via Active Record’s new_record? boolean method:

$ rails console
>> User.new.new_record?
=> true
>> User.first.new_record?
=> false

When constructing a form using form_for(@user), Rails uses POST if @user.new_record? is true and PUT if it is false.

As a final touch, we’ll add a URI to the user settings link to the site navigation. Since it depends on the signin status of the user, the test for the “Settings” link belongs with the other authentication tests, as shown in Listing 9.5. (It would be nice to have additional tests verifying that such links don’t appear for users who aren’t signed in; writing these tests is left as an exercise (Section 9.6).)

require 'spec_helper'

describe "Authentication" do
    .
    .
    .
    describe "with valid information" do
      let(:user) { FactoryGirl.create(:user) }
      before { sign_in user }

      it { should have_selector('title', text: user.name) }
      it { should have_link('Profile',  href: user_path(user)) }
      it { should have_link('Settings', href: edit_user_path(user)) }
      it { should have_link('Sign out', href: signout_path) }
      it { should_not have_link('Sign in', href: signin_path) }
      .
      .
      .
    end
  end
end

For convenience, the code in Listing 9.5 uses a helper to sign in a user inside the tests. The method is to visit the signin page and submit valid information, as shown in Listing 9.6.

.
.
.
def sign_in(user)
  visit signin_path
  fill_in "Email",    with: user.email
  fill_in "Password", with: user.password
  click_button "Sign in"
  # Sign in when not using Capybara as well.
  cookies[:remember_token] = user.remember_token
end

As noted in the comment line, filling in the form doesn’t work when not using Capybara, so to cover this case we also add the user’s remember token to the cookies:

# Sign in when not using Capybara as well.
cookies[:remember_token] = user.remember_token

This is necessary when using one of the HTTP request methods directly (get, post, put, or delete), as we’ll see in Listing 9.47. (Note that the test cookies object isn’t a perfect simulation of the real cookies object; in particular, the cookies.permanent method seen in Listing 8.19 doesn’t work inside tests.) As you might suspect, the sign_in method will prove useful in future tests, and in fact it can already be used to eliminate some duplication (Section 9.6).

The application code to add the URI for the “Settings” link is simple: we just use the named route edit_user_path from Table 7.1, together with the handy current_user helper method defined in Listing 8.22:

<%= link_to "Settings", edit_user_path(current_user) %>

The full application code appears in Listing 9.7).

<header class="navbar navbar-fixed-top navbar-inverse">
  <div class="navbar-inner">
    <div class="container">
      <%= link_to "sample app", root_path, id: "logo" %>
      <nav>
        <ul class="nav pull-right">
          <li><%= link_to "Home", root_path %></li>
          <li><%= link_to "Help", help_path %></li>
          <% if signed_in? %>
            <li><%= link_to "Users", '#' %></li>
            <li id="fat-menu" class="dropdown">
              <a href="#" class="dropdown-toggle" data-toggle="dropdown">
                Account <b class="caret"></b>
              </a>
              <ul class="dropdown-menu">
                <li><%= link_to "Profile", current_user %></li>
                <li><%= link_to "Settings", edit_user_path(current_user) %></li>
                <li class="divider"></li>
                <li>
                  <%= link_to "Sign out", signout_path, method: "delete" %>
                </li>
              </ul>
            </li>
          <% else %>
            <li><%= link_to "Sign in", signin_path %></li>
          <% end %>
        </ul>
      </nav>
    </div>
  </div>
</header>

9.1.2 Unsuccessful edits

In this section we’ll handle unsuccessful edits and get the error messages test in Listing 9.1 to pass. The application code creates an update action that uses update_attributes (Section 6.1.5) to update the user based on the submitted params hash, as shown in Listing 9.8. With invalid information, the update attempt returns false, so the else branch re-renders the edit page. We’ve seen this pattern before; the structure closely parallels the first version of the create action (Listing 7.21).

class UsersController < ApplicationController
  .
  .
  .
  def edit
    @user = User.find(params[:id])
  end

  def update
    @user = User.find(params[:id])
    if @user.update_attributes(params[:user])
      # Handle a successful update.
    else
      render 'edit'
    end
  end
end

The resulting error message (Figure 9.3) is the one needed to get the error message test to pass, as you should verify by running the test suite:

$ bundle exec rspec spec/

Figure 9.3: Error message from submitting the update form. (full size)

9.1.3 Successful edits

Now it’s time to get the edit form to work. Editing the profile images is already functional since we’ve outsourced image upload to Gravatar; we can edit gravatars by clicking on the “change” link from Figure 9.2, as shown in Figure 9.4. Let’s get the rest of the user edit functionality working as well.

Figure 9.4: The Gravatar image-cropping interface, with a picture of some dude.

The tests for the update action are similar to those for create. Listing 9.9 show how to use Capybara to fill in the form fields with valid information and then test that the resulting behavior is correct. This is a lot of code; see if you can work through it by referring back to the tests in Chapter 7.

require 'spec_helper'

describe "User pages" do
  .
  .
  .
  describe "edit" do
    let(:user) { FactoryGirl.create(:user) }
    before { visit edit_user_path(user) }
    .
    .
    .
    describe "with valid information" do
      let(:new_name)  { "New Name" }
      let(:new_email) { "new@example.com" }
      before do
        fill_in "Name",             with: new_name
        fill_in "Email",            with: new_email
        fill_in "Password",         with: user.password
        fill_in "Confirm Password", with: user.password
        click_button "Save changes"
      end

      it { should have_selector('title', text: new_name) }
      it { should have_selector('div.alert.alert-success') }
      it { should have_link('Sign out', href: signout_path) }
      specify { user.reload.name.should  == new_name }
      specify { user.reload.email.should == new_email }
    end
  end
end

The only real novelty in Listing 9.9 is the reload method, which appears in the test for changing the user’s attributes:

specify { user.reload.name.should  == new_name }
specify { user.reload.email.should == new_email }

This reloads the user variable from the test database using user.reload, and then verifies that the user’s new name and email match the new values.

The update action needed to get the tests in Listing 9.9 to pass is similar to the final form of the create action (Listing 8.27), as seen in Listing 9.10. All this does is add

flash[:success] = "Profile updated"
sign_in @user
redirect_to @user

to the code in Listing 9.8. Note that we sign in the user as part of a successful profile update; this is because the remember token gets reset when the user is saved (Listing 8.18), which invalidates the user’s session (Listing 8.22). This is a nice security feature, as it means that any hijacked sessions will automatically expire when the user information is changed.

class UsersController < ApplicationController
  .
  .
  .
  def update
    @user = User.find(params[:id])
    if @user.update_attributes(params[:user])
      flash[:success] = "Profile updated"
      sign_in @user
      redirect_to @user
    else
      render 'edit'
    end
  end
end

Note that, as currently constructed, every edit requires the user to reconfirm the password (as implied by the empty confirmation text box in Figure 9.2), which is a minor annoyance but makes updates much more secure.

With the code in this section, the user edit page should be working, as you can double-check by re-running the test suite, which should now be green:

$ bundle exec rspec spec/

9.2 Authorization

One nice effect of building the authentication machinery in Chapter 8 is that we are now in a position to implement authorization as well: authentication allows us to identify users of our site, and authorization lets us control what they can do.

Although the edit and update actions from Section 9.1 are functionally complete, they suffer from a ridiculous security flaw: they allow anyone (even non-signed-in users) to access either action, and any signed-in user can update the information for any other user. In this section, we’ll implement a security model that requires users to be signed in and prevents them from updating any information other than their own. Users who aren’t signed in and who try to access protected pages will be forwarded to the signin page with a helpful message, as mocked up in Figure 9.5.

Figure 9.5: A mockup of the result of visiting a protected page (full size)

9.2.1 Requiring signed-in users

Since the security restrictions for the edit and update actions are identical, we’ll handle them in a single RSpec describe block. Starting with the sign-in requirement, our initial tests verify that non-signed-in users attempting to access either action are simply sent to the signin page, as seen in Listing 9.11.

require 'spec_helper'

describe "Authentication" do
  .
  .
  .
  describe "authorization" do

    describe "for non-signed-in users" do
      let(:user) { FactoryGirl.create(:user) }

      describe "in the Users controller" do

        describe "visiting the edit page" do
          before { visit edit_user_path(user) }
          it { should have_selector('title', text: 'Sign in') }
        end

        describe "submitting to the update action" do
          before { put user_path(user) }
          specify { response.should redirect_to(signin_path) }
        end
      end
    end
  end
end

The code in Listing 9.11 introduces a second way, apart from Capybara’s visit method, to access a controller action: by issuing the appropriate HTTP request directly, in this case using the put method to issue a PUT request:

describe "submitting to the update action" do
  before { put user_path(user) }
  specify { response.should redirect_to(signin_path) }
end

This issues a PUT request directly to /users/1, which gets routed to the update action of the Users controller (Table 7.1). This is necessary because there is no way for a browser to visit the update action directly—it can only get there indirectly by submitting the edit form—so Capybara can’t do it either. But visiting the edit page only tests the authorization for the edit action, not for update. As a result, the only way to test the proper authorization for the update action itself is to issue a direct request. (As you might guess, in addition to put Rails tests support get, post, and delete as well.)

When using one of the methods to issue HTTP requests directly, we get access to the low-level response object. Unlike the Capybara page object, response lets us test for the server response itself, in this case verifying that the update action responds by redirecting to the signin page:

specify { response.should redirect_to(signin_path) }

The authorization application code uses a before filter, which arranges for a particular method to be called before the given actions. To require users to be signed in, we define a signed_in_user method and invoke it using before_filter :signed_in_user, as shown in Listing 9.12.

class UsersController < ApplicationController
  before_filter :signed_in_user, only: [:edit, :update]
  .
  .
  .
  private

    def signed_in_user
      redirect_to signin_url, notice: "Please sign in." unless signed_in?
    end
end

By default, before filters apply to every action in a controller, so here we restrict the filter to act only on the :edit and :update actions by passing the appropriate :only options hash.

Note that Listing 9.12 uses a shortcut for setting flash[:notice] by passing an options hash to the redirect_to function. The code in Listing 9.12 is equivalent to the more verbose

flash[:notice] = "Please sign in."
redirect_to signin_url

(The same construction works for the :error key, but not for :success.)

Together with :success and :error, the :notice key completes our triumvirate of flash styles, all of which are supported natively by Bootstrap CSS. By signing out and attempting to access the user edit page /users/1/edit, we can see the resulting yellow “notice” box, as seen in Figure 9.6.

Figure 9.6: The signin form after trying to access a protected page. (full size)

Unfortunately, in the process of getting the authorization tests from Listing 9.11 to pass, we’ve broken the tests in Listing 9.1. Code like

describe "edit" do
  let(:user) { FactoryGirl.create(:user) }
  before { visit edit_user_path(user) }
  .
  .
  .

no longer works because visiting the edit user path requires a signed-in user. The solution is to sign in the user with the sign_in utility from Listing 9.6, as shown in Listing 9.13.

require 'spec_helper'

describe "User pages" do
  .
  .
  .
  describe "edit" do
    let(:user) { FactoryGirl.create(:user) }
    before do
      sign_in user
      visit edit_user_path(user)
    end
    .
    .
    .
  end
end

At this point our, test suite should be green:

$ bundle exec rspec spec/ 

9.2.2 Requiring the right user

Of course, requiring users to sign in isn’t quite enough; users should only be allowed to edit their own information. We can test for this by first signing in as an incorrect user and then hitting the edit and update actions (Listing 9.14). Note that, since users should never even try to edit another user’s profile, we redirect not to the signin page but to the root URL.

require 'spec_helper'

describe "Authentication" do
  .
  .
  .
  describe "authorization" do
    .
    .
    .
    describe "as wrong user" do
      let(:user) { FactoryGirl.create(:user) }
      let(:wrong_user) { FactoryGirl.create(:user, email: "wrong@example.com") }
      before { sign_in user }

      describe "visiting Users#edit page" do
        before { visit edit_user_path(wrong_user) }
        it { should_not have_selector('title', text: full_title('Edit user')) }
      end

      describe "submitting a PUT request to the Users#update action" do
        before { put user_path(wrong_user) }
        specify { response.should redirect_to(root_path) }
      end
    end
  end
end

Note here that a factory can take an option:

FactoryGirl.create(:user, email: "wrong@example.com")

This creates a user with a different email address from the default. The tests specify that the original user should not have access to the wrong user’s edit or update actions.

The application code adds a second before filter to call the correct_user method, as shown in Listing 9.15.

class UsersController < ApplicationController
  before_filter :signed_in_user, only: [:edit, :update]
  before_filter :correct_user,   only: [:edit, :update]
  .
  .
  .
  def edit
  end

  def update
    if @user.update_attributes(params[:user])
      flash[:success] = "Profile updated"
      sign_in @user
      redirect_to @user
    else
      render 'edit'
    end
  end
  .
  .
  .
  private

    def signed_in_user
      redirect_to signin_url, notice: "Please sign in." unless signed_in?
    end

    def correct_user
      @user = User.find(params[:id])
      redirect_to(root_path) unless current_user?(@user)
    end
end

The correct_user filter uses the current_user? boolean method, which we define in the Sessions helper (Listing 9.16).

module SessionsHelper
  .
  .
  .
  def current_user
    @current_user ||= User.find_by_remember_token(cookies[:remember_token])
  end

  def current_user?(user)
    user == current_user
  end
  .
  .
  .
end

Listing 9.15 also shows the updated edit and update actions. Before, in Listing 9.2, we had

def edit
  @user = User.find(params[:id])
end

and similarly for update. Now that the correct_user before filter defines @user, we can omit it from both actions.

Before moving on, you should verify that the test suite passes:

$ bundle exec rspec spec/

9.2.3 Friendly forwarding

Our site authorization is complete as written, but there is one minor blemish: when users try to access a protected page, they are currently redirected to their profile pages regardless of where they were trying to go. In other words, if a non-logged-in user tries to visit the edit page, after signing in the user will be redirected to /users/1 instead of /users/1/edit. It would be much friendlier to redirect them to their intended destination instead.

To test for such “friendly forwarding”, we first visit the user edit page, which redirects to the signin page. We then enter valid signin information and click the “Sign in” button. The resulting page, which by default is the user’s profile, should in this case be the “Edit user” page. The test for this sequence appears in Listing 9.17.

require 'spec_helper'

describe "Authentication" do
  .
  .
  .
  describe "authorization" do

    describe "for non-signed-in users" do
      let(:user) { FactoryGirl.create(:user) }

      describe "when attempting to visit a protected page" do
        before do
          visit edit_user_path(user)
          fill_in "Email",    with: user.email
          fill_in "Password", with: user.password
          click_button "Sign in"
        end

        describe "after signing in" do

          it "should render the desired protected page" do
            page.should have_selector('title', text: 'Edit user')
          end
        end
      end
      .
      .
      .
    end
    .
    .
    .
  end
end

Now for the implementation.⁴ In order to forward users to their intended destination, we need to store the location of the requested page somewhere, and then redirect to that location instead. We accomplish this with a pair of methods, store_location and redirect_back_or, both defined in the Sessions helper (Listing 9.18).

module SessionsHelper
  .
  .
  .
  def redirect_back_or(default)
    redirect_to(session[:return_to] || default)
    session.delete(:return_to)
  end

  def store_location
    session[:return_to] = request.url
  end
end

The storage mechanism is the session facility provided by Rails, which you can think of as being like an instance of the cookies variable from Section 8.2.1 that automatically expires upon browser close. We also use the request object to get the url, i.e., the URI/URL of the requested page. The store_location method puts the requested URI in the session variable under the key :return_to.

To make use of store_location, we need to add it to the signed_in_user before filter, as shown in Listing 9.19.

class UsersController < ApplicationController
  before_filter :signed_in_user, only: [:edit, :update]
  before_filter :correct_user,   only: [:edit, :update]
  .
  .
  .
  def edit
  end
  .
  .
  .
  private

    def signed_in_user
      unless signed_in?
        store_location
        redirect_to signin_url, notice: "Please sign in."
      end
    end

    def correct_user
      @user = User.find(params[:id])
      redirect_to(root_path) unless current_user?(@user)
    end
end

To implement the forwarding itself, we use the redirect_back_or method to redirect to the requested URI if it exists, or some default URI otherwise, which we add to the Sessions controller create action to redirect after successful signin (Listing 9.20). The redirect_back_or method uses the or operator || through

session[:return_to] || default

This evaluates to session[:return_to] unless it’s nil, in which case it evaluates to the given default URI. Note that Listing 9.18 is careful to remove the forwarding URI; otherwise, subsequent signin attempts would forward to the protected page until the user closed his browser. (Testing for this is left as an exercise (Section 9.6.)

class SessionsController < ApplicationController
  .
  .
  .
  def create
    user = User.find_by_email(params[:session][:email].downcase)
    if user && user.authenticate(params[:session][:password])
      sign_in user
      redirect_back_or user
    else
      flash.now[:error] = 'Invalid email/password combination'
      render 'new'
    end
  end
  .
  .
  .
end

(If you completed the first exercise in Chapter 8, be sure to use the proper params hash in Listing 9.20.)

With that, the friendly forwarding integration test in Listing 9.17 should pass, and the basic user authentication and page protection implementation is complete. As usual, it’s a good idea to verify that the test suite is green before proceeding:

$ bundle exec rspec spec/

9.3 Showing all users

In this section, we’ll add the penultimate user action, the index action, which is designed to display all the users instead of just one. Along the way, we’ll learn about populating the database with sample users and paginating the user output so that the index page can scale up to display a potentially large number of users. A mockup of the result—users, pagination links, and a “Users” navigation link—appears in Figure 9.7.⁵ In Section 9.4, we’ll add an administrative interface to the user index so that (presumably troublesome) users can be destroyed.

Figure 9.7: A mockup of the user index, with pagination and a “Users” nav link. (full size)

9.3.1 User index

Although we’ll keep individual user show pages visible to all site visitors, the user index will be restricted to signed-in users so that there’s a limit to how much unregistered users can see by default. We’ll start by testing that the index action is protected by visiting the users_path (Table 7.1) and verifying that we are redirected to the signin page. As with other authorization tests, we’ll put this example in the authentication integration test, as shown in Listing 9.21.

require 'spec_helper'

describe "Authentication" do
  .
  .
  .
  describe "authorization" do

    describe "for non-signed-in users" do
      .
      .
      .
      describe "in the Users controller" do
        .
        .
        .
        describe "visiting the user index" do
          before { visit users_path }
          it { should have_selector('title', text: 'Sign in') }
        end
      end
      .
      .
      .
    end
  end
end

The corresponding application code simply involves adding index to the list of actions protected by the signed_in_user before filter, as shown in Listing 9.22.

class UsersController < ApplicationController
  before_filter :signed_in_user, only: [:index, :edit, :update]
  .
  .
  .
  def index
  end
  .
  .
  .
end

The next set of tests makes sure that, for signed-in users, the index page has the right title/heading and lists all of the site’s users. The method is to make three factory users (signing in as the first one) and then verify that the index page has a list element (li) tag for the name of each one. Note that we’ve taken care to give the users different names so that each element in the list of users has a unique entry, as shown in Listing 9.23.

require 'spec_helper'

describe "User pages" do

  subject { page }

  describe "index" do
    before do
      sign_in FactoryGirl.create(:user)
      FactoryGirl.create(:user, name: "Bob", email: "bob@example.com")
      FactoryGirl.create(:user, name: "Ben", email: "ben@example.com")
      visit users_path
    end

    it { should have_selector('title', text: 'All users') }
    it { should have_selector('h1',    text: 'All users') }

    it "should list each user" do
      User.all.each do |user|
        page.should have_selector('li', text: user.name)
      end
    end
  end
  .
  .
  .
end

As you may recall from the corresponding action in the demo app (Listing 2.4), the application code uses User.all to pull all the users out of the database, assigning them to an @users instance variable for use in the view, as seen in Listing 9.24. (If displaying all the users at once seems like a bad idea, you’re right, and we’ll remove this blemish in Section 9.3.3.)

class UsersController < ApplicationController
  before_filter :signed_in_user, only: [:index, :edit, :update]
  .
  .
  .
  def index
    @users = User.all
  end
  .
  .
  .
end

To make the actual index page, we need to make a view that iterates through the users and wraps each one in an li tag. We do this with the each method, displaying each user’s Gravatar and name, while wrapping the whole thing in an unordered list (ul) tag (Listing 9.25). The code in Listing 9.25 uses the result of Listing 7.29 from Section 7.6, which allows us to pass an option to the Gravatar helper specifying a size other than the default. If you didn’t do that exercise, update your Users helper file with the contents of Listing 7.29 before proceeding.

<% provide(:title, 'All users') %>
<h1>All users</h1>

<ul class="users">
  <% @users.each do |user| %>
    <li>
      <%= gravatar_for user, size: 52 %>
      <%= link_to user.name, user %>
    </li>
  <% end %>
</ul>

Let’s also add a little CSS (or, rather, SCSS) for style (Listing 9.26).

.
.
.

/* users index */

.users {
  list-style: none;
  margin: 0;
  li {
    overflow: auto;
    padding: 10px 0;
    border-top: 1px solid $grayLighter;
    &:last-child {
      border-bottom: 1px solid $grayLighter;
    }
  }
}

Finally, we’ll add the URI to the users link in the site’s navigation header using users_path, thereby using the last of the unused named routes in Table 7.1. The test (Listing 9.27) and application code (Listing 9.28) are both straightforward.

require 'spec_helper'

describe "Authentication" do
    .
    .
    .
    describe "with valid information" do
      let(:user) { FactoryGirl.create(:user) }
      before { sign_in user }

      it { should have_selector('title', text: user.name) }

      it { should have_link('Users',    href: users_path) }
      it { should have_link('Profile',  href: user_path(user)) }
      it { should have_link('Settings', href: edit_user_path(user)) }
      it { should have_link('Sign out', href: signout_path) }

      it { should_not have_link('Sign in', href: signin_path) }
      .
      .
      .
    end
  end
end

<header class="navbar navbar-fixed-top navbar-inverse">
  <div class="navbar-inner">
    <div class="container">
      <%= link_to "sample app", root_path, id: "logo" %>
      <nav>
        <ul class="nav pull-right">
          <li><%= link_to "Home", root_path %></li>
          <li><%= link_to "Help", help_path %></li>
          <% if signed_in? %>
            <li><%= link_to "Users", users_path %></li>
            <li id="fat-menu" class="dropdown">
              <a href="#" class="dropdown-toggle" data-toggle="dropdown">
                Account <b class="caret"></b>
              </a>
              <ul class="dropdown-menu">
                <li><%= link_to "Profile", current_user %></li>
                <li><%= link_to "Settings", edit_user_path(current_user) %></li>
                <li class="divider"></li>
                <li>
                  <%= link_to "Sign out", signout_path, method: "delete" %>
                </li>
              </ul>
            </li>
          <% else %>
            <li><%= link_to "Sign in", signin_path %></li>
          <% end %>
        </ul>
      </nav>
    </div>
  </div>
</header>

With that, the user index is fully functional, with all tests passing:

$ bundle exec rspec spec/

On the other hand, as seen in Figure 9.8, it is a bit… lonely. Let’s remedy this sad situation.

Figure 9.8: The user index page /users with only one user. (full size)

9.3.2 Sample users

In this section, we’ll give our lonely sample user some company. Of course, to create enough users to make a decent user index, we could use our web browser to visit the signup page and make the new users one by one, but far a better solution is to use Ruby (and Rake) to make the users for us.

First, we’ll add the Faker gem to the Gemfile, which will allow us to make sample users with semi-realistic names and email addresses (Listing 9.29).

source 'https://rubygems.org'

gem 'rails', '3.2.13'
gem 'bootstrap-sass', '2.1'
gem 'bcrypt-ruby', '3.0.1'
gem 'faker', '1.0.1'
.
.
.

Then install as usual:

$ bundle install

Next, we’ll add a Rake task to create sample users. Rake tasks live in the lib/tasks directory, and are defined using namespaces (in this case, :db), as seen in Listing 9.30. (This is a bit advanced, so don’t worry too much about the details.)

namespace :db do
  desc "Fill database with sample data"
  task populate: :environment do
    User.create!(name: "Example User",
                 email: "example@railstutorial.org",
                 password: "foobar",
                 password_confirmation: "foobar")
    99.times do |n|
      name  = Faker::Name.name
      email = "example-#{n+1}@railstutorial.org"
      password  = "password"
      User.create!(name: name,
                   email: email,
                   password: password,
                   password_confirmation: password)
    end
  end
end

This defines a task db:populate that creates an example user with name and email address replicating our previous one, and then makes 99 more. The line

task populate: :environment do

ensures that the Rake task has access to the local Rails environment, including the User model (and hence User.create!). Here create! is just like the create method, except it raises an exception (Section 6.1.4) for an invalid user rather than returning false. This noisier construction makes debugging easier by avoiding silent errors.

With the :db namespace as in Listing 9.30, we can invoke the Rake task as follows:

$ bundle exec rake db:reset
$ bundle exec rake db:populate
$ bundle exec rake db:test:prepare

After running the Rake task, our application has 100 sample users, as seen in Figure 9.9. (I’ve taken the liberty of associating the first few sample addresses with photos so that they’re not all the default Gravatar image.)

Figure 9.9: The user index page /users with 100 sample users. (full size)

9.3.3 Pagination

Our original user doesn’t suffer from loneliness any more, but now we have the opposite problem: our user has too many companions, and they all appear on the same page. Right now there are a hundred, which is already a reasonably large number, and on a real site it could be thousands. The solution is to paginate the users, so that (for example) only 30 show up on a page at any one time.

There are several pagination methods in Rails; we’ll use one of the simplest and most robust, called will_paginate. To use it, we need to include both the will_paginate gem and bootstrap-will_paginate, which configures will_paginate to use Bootstrap’s pagination styles. The updated Gemfile appears in Listing 9.31.

source 'https://rubygems.org'

gem 'rails', '3.2.13'
gem 'bootstrap-sass', '2.1'
gem 'bcrypt-ruby', '3.0.1'
gem 'faker', '1.0.1'
gem 'will_paginate', '3.0.3'
gem 'bootstrap-will_paginate', '0.0.6'
.
.
.

Then run bundle install:

$ bundle install

You should also restart the web server to insure that the new gems are loaded properly.

Because the will_paginate gem is in wide use, there’s no need to test it thoroughly, so we’ll take a lightweight approach. First, we’ll test for a div with CSS class “pagination”, which is what gets output by will_paginate. Then we’ll verify that the correct users appear on the first page of results. This requires the use of the paginate method, which we’ll cover shortly.

As before, we’ll use Factory Girl to simulate users, but immediately we have a problem: user email addresses must be unique, which would appear to require creating more than 30 users by hand—a terribly cumbersome job. In addition, when testing for user listings it would be convenient for them all to have different names. Fortunately, Factory Girl anticipates this issue, and provides sequences to solve it. Our original factory (Listing 7.8) hard-coded the name and email address:

FactoryGirl.define do
  factory :user do
    name     "Michael Hartl"
    email    "michael@example.com"
    password "foobar"
    password_confirmation "foobar"
  end
end

Instead, we can arrange for a sequence of names and email addresses using the sequence method:

factory :user do
  sequence(:name)  { |n| "Person #{n}" }
  sequence(:email) { |n| "person_#{n}@example.com"}   
  .
  .
  .

Here sequence takes a symbol corresponding to the desired attribute (such as :name) and a block with one variable, which we have called n. Upon successive invocations of the FactoryGirl method,

FactoryGirl.create(:user)

The block variable n is automatically incremented, so that the first user has name “Person 1” and email address “person_1@example.com”, the second user has name “Person 2” and email address “person_2@example.com”, and so on. The full code appears in Listing 9.32.

FactoryGirl.define do
  factory :user do
    sequence(:name)  { |n| "Person #{n}" }
    sequence(:email) { |n| "person_#{n}@example.com"}   
    password "foobar"
    password_confirmation "foobar"
  end
end

Applying the idea of factory sequences, we can make 30 users in our test, which (as we will see) will be sufficient to invoke pagination:

before(:all) { 30.times { FactoryGirl.create(:user) } }
after(:all)  { User.delete_all }    

Note here the use of before(:all), which ensures that the sample users are created once, before all the tests in the block. This is an optimization for speed, as creating 30 users can be slow on some systems. We use the complementary method after(:all) to delete the users once we’re done.

The tests for the appearance of the pagination div and the right users appears in Listing 9.33. Note the replacement of the User.all array from Listing 9.23 with User.paginate(page: 1), which (as we’ll see momentarily) is how to pull out the first page of users from the database. Note also that Listing 9.33 uses before(:each) to emphasize the contrast with before(:all).

require 'spec_helper'

describe "User pages" do

  subject { page }

  describe "index" do

    let(:user) { FactoryGirl.create(:user) }

    before(:each) do
      sign_in user
      visit users_path
    end

    it { should have_selector('title', text: 'All users') }
    it { should have_selector('h1',    text: 'All users') }

    describe "pagination" do

      before(:all) { 30.times { FactoryGirl.create(:user) } }
      after(:all)  { User.delete_all }

      it { should have_selector('div.pagination') }

      it "should list each user" do
        User.paginate(page: 1).each do |user|
          page.should have_selector('li', text: user.name)
        end
      end
    end
  end
  .
  .
  .
end

To get pagination working, we need to add some code to the index view telling Rails to paginate the users, and we need to replace User.all in the index action with an object that knows about pagination. We’ll start by adding the special will_paginate method in the view (Listing 9.34); we’ll see in a moment why the code appears both above and below the user list.

<% provide(:title, 'All users') %>
<h1>All users</h1>

<%= will_paginate %>

<ul class="users">
  <% @users.each do |user| %>
    <li>
      <%= gravatar_for user, size: 52 %>
      <%= link_to user.name, user %>
    </li>
  <% end %>
</ul>

<%= will_paginate %>

The will_paginate method is a little magical; inside a users view, it automatically looks for an @users object, and then displays pagination links to access other pages. The view in Listing 9.34 doesn’t work yet, though, because currently @users contains the results of User.all (Listing 9.24), which is of class Array, whereas will_paginate expects an object of class ActiveRecord::Relation. Happily, this is just the kind of object returned by the paginate method added by the will_paginate gem to all Active Record objects:

$ rails console
>> User.all.class
=> Array
>> User.paginate(page: 1).class
=> ActiveRecord::Relation

Note that paginate takes a hash argument with key :page and value equal to the page requested. User.paginate pulls the users out of the database one chunk at a time (30 by default), based on the :page parameter. So, for example, page 1 is users 1–30, page 2 is users 31–60, etc. If the page is nil, paginate simply returns the first page.

Using the paginate method, we can paginate the users in the sample application by using paginate in place of all in the index action (Listing 9.35). Here the :page parameter comes from params[:page], which is generated automatically by will_paginate.

class UsersController < ApplicationController
  before_filter :signed_in_user, only: [:index, :edit, :update]
  .
  .
  .
  def index
    @users = User.paginate(page: params[:page])
  end
  .
  .
  .
end

The user index page should now be working, appearing as in Figure 9.10. (On some systems, you may have to restart the Rails server at this point.) Because we included will_paginate both above and below the user list, the pagination links appear in both places.

Figure 9.10: The user index page /users with pagination. (full size)

If you now click on either the 2 link or Next link, you’ll get the second page of results, as shown in Figure 9.11.

Figure 9.11: Page 2 of the user index (/users?page=2). (full size)

You should also verify that the tests are passing:

$ bundle exec rspec spec/

9.3.4 Partial refactoring

The paginated user index is now complete, but there’s one improvement I can’t resist including: Rails has some incredibly slick tools for making compact views, and in this section we’ll refactor the index page to use them. Because our code is well-tested, we can refactor with confidence, assured that we are unlikely to break our site’s functionality.

The first step in our refactoring is to replace the user li from Listing 9.34 with a render call (Listing 9.36).

<% provide(:title, 'All users') %>
<h1>All users</h1>

<%= will_paginate %>

<ul class="users">
  <% @users.each do |user| %>
    <%= render user %>
  <% end %>
</ul>

<%= will_paginate %>

Here we call render not on a string with the name of a partial, but rather on a user variable of class User;⁶ in this context, Rails automatically looks for a partial called _user.html.erb, which we must create (Listing 9.37).

<li>
  <%= gravatar_for user, size: 52 %>
  <%= link_to user.name, user %>
</li>

This is a definite improvement, but we can do even better: we can call render directly on the @users variable (Listing 9.38).

<% provide(:title, 'All users') %>
<h1>All users</h1>

<%= will_paginate %>

<ul class="users">
  <%= render @users %>
</ul>

<%= will_paginate %>

Here Rails infers that @users is a list of User objects; moreover, when called with a collection of users, Rails automatically iterates through them and renders each one with the _user.html.erb partial. The result is the impressively compact code in Listing 9.38. As with any refactoring, you should verify that the test suite is still green after changing the application code:

$ bundle exec rspec spec/

9.4 Deleting users

Now that the user index is complete, there’s only one canonical REST action left: destroy. In this section, we’ll add links to delete users, as mocked up in Figure 9.12, and define the destroy action necessary to accomplish the deletion. But first, we’ll create the class of administrative users authorized to do so.

Figure 9.12: A mockup of the user index with delete links. (full size)

9.4.1 Administrative users

We will identify privileged administrative users with a boolean admin attribute in the User model, which, as we’ll see, will automatically lead to an admin? boolean method to test for admin status. We can write tests for this attribute as in Listing 9.39.

require 'spec_helper'

describe User do
  .
  .
  .
  it { should respond_to(:admin) }
  it { should respond_to(:authenticate) }

  it { should be_valid }
  it { should_not be_admin }

  describe "with admin attribute set to 'true'" do
    before do
      @user.save!
      @user.toggle!(:admin)
    end

    it { should be_admin }
  end
  .
  .
  .
end

Here we’ve used the toggle! method to flip the admin attribute from false to true. Also note that the line

it { should be_admin }

implies (via the RSpec boolean convention) that the user should have an admin? boolean method.

As usual, we add the admin attribute with a migration, indicating the boolean type on the command line:

$ rails generate migration add_admin_to_users admin:boolean

The migration simply adds the admin column to the users table (Listing 9.40), yielding the data model in Figure 9.13.

class AddAdminToUsers < ActiveRecord::Migration
  def change
    add_column :users, :admin, :boolean, default: false
  end
end

Note that we’ve added the argument default: false to add_column in Listing 9.40, which means that users will not be administrators by default. (Without the default: false argument, admin will be nil by default, which is still false, so this step is not strictly necessary. It is more explicit, though, and communicates our intentions more clearly both to Rails and to readers of our code.)

Figure 9.13: The User model with an added admin boolean attribute.

Finally, we migrate the development database and prepare the test database:

$ bundle exec rake db:migrate
$ bundle exec rake db:test:prepare

As expected, Rails figures out the boolean nature of the admin attribute and automatically adds the question-mark method admin?:

$ rails console --sandbox
>> user = User.first
>> user.admin?
=> false
>> user.toggle!(:admin)
=> true
>> user.admin?
=> true

As a result, the admin tests should pass:

$ bundle exec rspec spec/models/user_spec.rb

As a final step, let’s update our sample data populator to make the first user an admin by default (Listing 9.41).

namespace :db do
  desc "Fill database with sample data"
  task populate: :environment do
    admin = User.create!(name: "Example User",
                         email: "example@railstutorial.org",
                         password: "foobar",
                         password_confirmation: "foobar")
    admin.toggle!(:admin)
    .
    .
    .
  end
end

Then reset the database and re-populate the sample data:

$ bundle exec rake db:reset
$ bundle exec rake db:populate
$ bundle exec rake db:test:prepare

Revisiting attr_accessible

You might have noticed that Listing 9.41 makes the user an admin with toggle!(:admin), but why not just add admin: true to the initialization hash? The answer is, it won’t work, and this is by design: only attr_accessible attributes can be assigned through mass assignment (that is, using an initialization hash, as in User.new(name: "Foo", ...)), and the admin attribute isn’t accessible. Listing 9.42 reproduces the most recent list of attr_accessible attributes—note that :admin is not on the list.

class User < ActiveRecord::Base
  attr_accessible :name, :email, :password, :password_confirmation
  .
  .
  .
end

Explicitly defining accessible attributes is crucial for good site security. If we omitted the attr_accessible list in the User model (or foolishly added :admin to the list), a malicious user could send a PUT request as follows:⁷

put /users/17?admin=1

This request would make user 17 an admin, which would be a potentially serious security breach, to say the least. Because of this danger, it is a good practice to define attr_accessible for every model. In fact, it’s a good idea to write a test for any attribute that isn’t accessible; writing such a test for the admin attribute is left as an exercise (Section 9.6).

9.4.2 The destroy action

The final step needed to complete the Users resource is to add delete links and a destroy action. We’ll start by adding a delete link for each user on the user index page, restricting access to administrative users.

To write tests for the delete functionality, it’s helpful to be able to have a factory to create admins. We can accomplish this by adding an :admin block to our factories, as shown in Listing 9.43.

FactoryGirl.define do
  factory :user do
    sequence(:name)  { |n| "Person #{n}" }
    sequence(:email) { |n| "person_#{n}@example.com"}   
    password "foobar"
    password_confirmation "foobar"

    factory :admin do
      admin true
    end
  end
end

With the code in Listing 9.43, we can now use FactoryGirl.create(:admin) to create an administrative user in our tests.

Our security model requires that ordinary users not see delete links:

it { should_not have_link('delete') }

But administrative users should see such links, and by clicking on a delete link we expect an admin to delete the user, i.e., to change the User count by -1:

it { should have_link('delete', href: user_path(User.first)) }
it "should be able to delete another user" do
  expect { click_link('delete') }.to change(User, :count).by(-1)
end
it { should_not have_link('delete', href: user_path(admin)) }

Note that we have added a test to verify that the admin does not see a link to delete himself. The full set of delete link tests appears in Listing 9.44.

require 'spec_helper'

describe "User pages" do

  subject { page }

  describe "index" do

    let(:user) { FactoryGirl.create(:user) }

    before do
      sign_in user
      visit users_path
    end

    it { should have_selector('title', text: 'All users') }
    it { should have_selector('h1',    text: 'All users') }

    describe "pagination" do
      .
      .
      .
    end

    describe "delete links" do

      it { should_not have_link('delete') }

      describe "as an admin user" do
        let(:admin) { FactoryGirl.create(:admin) }
        before do
          sign_in admin
          visit users_path
        end

        it { should have_link('delete', href: user_path(User.first)) }
        it "should be able to delete another user" do
          expect { click_link('delete') }.to change(User, :count).by(-1)
        end
        it { should_not have_link('delete', href: user_path(admin)) }
      end
    end
  end
  .
  .
  .
end

The application code links to "delete" if the current user is an admin (Listing 9.45). Note the method: :delete argument, which arranges for the link to issue the necessary DELETE request. We’ve also wrapped each link inside an if statement so that only admins can see them. The result for our admin user appears in Figure 9.14.

<li>
  <%= gravatar_for user, size: 52 %>
  <%= link_to user.name, user %>
  <% if current_user.admin? && !current_user?(user) %>
    | <%= link_to "delete", user, method: :delete,
                                  data: { confirm: "You sure?" } %>
  <% end %>
</li>

Web browsers can’t send DELETE requests natively, so Rails fakes them with JavaScript. This means that the delete links won’t work if the user has JavaScript disabled. If you must support non-JavaScript-enabled browsers you can fake a DELETE request using a form and a POST request, which works even without JavaScript; see the RailsCast on “Destroy Without JavaScript” for details.

Figure 9.14: The user index /users with delete links. (full size)

To get the delete links to work, we need to add a destroy action (Table 7.1), which finds the corresponding user and destroys it with the Active Record destroy method, finally redirecting to the user index, as seen in Listing 9.46. Note that we also add :destroy to the signed_in_user before filter.

class UsersController < ApplicationController
  before_filter :signed_in_user, only: [:index, :edit, :update, :destroy]
  before_filter :correct_user,   only: [:edit, :update]
  .
  .
  .
  def destroy
    User.find(params[:id]).destroy
    flash[:success] = "User destroyed."
    redirect_to users_url
  end
  .
  .
  .
end

Note that the destroy action uses method chaining to combine the find and destroy into one line:

User.find(params[:id]).destroy

As constructed, only admins can destroy users through the web, because only admins can see the delete links. Unfortunately, there’s still a terrible security hole: any sufficiently sophisticated attacker could simply issue DELETE requests directly from the command line to delete any user on the site. To secure the site properly, we also need access control on the destroy action, so our tests should check not only that admins can delete users, but also that other users can’t. The results appear in Listing 9.47. Note that, in analogy with the put method from Listing 9.11, we use delete to issue a DELETE request directly to the specified URI (in this case, the user path, as required by Table 7.1).

require 'spec_helper'

describe "Authentication" do
  .
  .
  .
  describe "authorization" do
    .
    .
    .
    describe "as non-admin user" do
      let(:user) { FactoryGirl.create(:user) }
      let(:non_admin) { FactoryGirl.create(:user) }

      before { sign_in non_admin }

      describe "submitting a DELETE request to the Users#destroy action" do
        before { delete user_path(user) }
        specify { response.should redirect_to(root_path) }        
      end
    end
  end
end

In principle, there’s still a minor security hole, which is that an admin could delete himself by issuing a DELETE request directly. One might argue that such an admin is only getting what he deserves, but it would be nice to prevent such an occurrence, and doing so is left as an exercise (Section 9.6).

As you might suspect by now, the application code uses a before filter, this time to restrict access to the destroy action to admins. The resulting admin_user before filter appears in Listing 9.48.

class UsersController < ApplicationController
  before_filter :signed_in_user, only: [:index, :edit, :update, :destroy]
  before_filter :correct_user,   only: [:edit, :update]
  before_filter :admin_user,     only: :destroy
  .
  .
  .
  private
    .
    .
    .
    def admin_user
      redirect_to(root_path) unless current_user.admin?
    end
end

At this point, all the tests should be passing, and the Users resource—with its controller, model, and views—is functionally complete.

$ bundle exec rspec spec/

9.5 Conclusion

We’ve come a long way since introducing the Users controller way back in Section 5.4. Those users couldn’t even sign up; now users can sign up, sign in, sign out, view their profiles, edit their settings, and see an index of all users—and some can even destroy other users.

The rest of this book builds on the foundation of the Users resource (and associated authorization system) to make a site with Twitter-like microposts (Chapter 10) and a status feed of posts from followed users (Chapter 11). These chapters will introduce some of the most powerful features of Rails, including data modeling with has_many and has_many through.

Before moving on, be sure to merge all the changes into the master branch:

$ git add .
$ git commit -m "Finish user edit, update, index, and destroy actions"
$ git checkout master
$ git merge updating-users

You can also deploy the application and even populate the production database with sample users (using the pg:reset task to reset the production database):

$ git push heroku
$ heroku pg:reset DATABASE
$ heroku run rake db:migrate
$ heroku run rake db:populate

You may have to redeploy to Heroku to force an app restart. Here’s a hack that will force Heroku to restart the application:

$ touch foo
$ git add foo
$ git commit -m "foo"
$ git push heroku

It’s also worth noting that this chapter saw the last of the necessary gem installations. For reference, the final Gemfile is shown in Listing 9.49. (Optional gems may be system-dependent and are commented out. You can uncomment them to see if they work on your system.)

source 'https://rubygems.org'

gem 'rails', '3.2.13'
gem 'bootstrap-sass', '2.1'
gem 'bcrypt-ruby', '3.0.1'
gem 'faker', '1.0.1'
gem 'will_paginate', '3.0.3'
gem 'bootstrap-will_paginate', '0.0.6'
gem 'jquery-rails', '2.0.2'

group :development, :test do
  gem 'sqlite3', '1.3.5'
  gem 'rspec-rails', '2.11.0'
  # gem 'guard-rspec', '1.2.1'
  # gem 'guard-spork', '1.2.0'  
  # gem 'childprocess', '0.3.6'
  # gem 'spork', '0.9.2'
end

# Gems used only for assets and not required
# in production environments by default.
group :assets do
  gem 'sass-rails',   '3.2.5'
  gem 'coffee-rails', '3.2.2'
  gem 'uglifier', '1.2.3'
end

group :test do
  gem 'capybara', '1.1.2'
  gem 'factory_girl_rails', '4.1.0'
  gem 'cucumber-rails', '1.2.1', :require => false
  gem 'database_cleaner', '0.7.0'
  # gem 'launchy', '2.1.0'
  # gem 'rb-fsevent', '0.9.1', :require => false
  # gem 'growl', '1.0.3'
end

group :production do
  gem 'pg', '0.12.2'
end

9.6 Exercises

1. Following the model in Listing 10.8, add a test to verify that the User admin attribute isn’t accessible. Be sure to get first to Red, and then to Green. (Hint: Your first step should be to add admin to the accessible list.)
2. Arrange for the Gravatar “change” link in Listing 9.3 to open in a new window (or tab). Hint: Search the web; you should find one particularly robust method involving something called _blank.
3. The current authentication tests check that navigation links such as “Profile” and “Settings” appear when a user is signed in. Add tests to make sure that these links don’t appear when a user isn’t signed in.
4. Use the sign_in test helper from Listing 9.6 in as many places as you can find.
5. Remove the duplicated form code by refactoring the new.html.erb and edit.html.erb views to use the partial in Listing 9.50. Note that you will have to pass the form variable f explicitly as a local variable, as shown in Listing 9.51. You will also have to update the tests, as the forms aren’t currently exactly the same; identify the slight difference and update the tests accordingly.
6. Signed-in users have no reason to access the new and create actions in the Users controller. Arrange for such users to be redirected to the root URL if they do try to hit those pages.
7. Learn about the request object by inserting some of the methods listed in the Rails API⁸ into the site layout. (Refer to Listing 7.1 if you get stuck.)
8. Write a test to make sure that the friendly forwarding only forwards to the given URI the first time. On subsequent signin attempts, the forwarding URI should revert to the default (i.e., the profile page). See Listing 9.52 for a hint (and, by a hint, I mean the solution).
9. Modify the destroy action to prevent admin users from destroying themselves. (Write a test first.)

<%= render 'shared/error_messages' %>

<%= f.label :name %>
<%= f.text_field :name %>

<%= f.label :email %>
<%= f.text_field :email %>

<%= f.label :password %>
<%= f.password_field :password %>

<%= f.label :password_confirmation, "Confirm Password" %>
<%= f.password_field :password_confirmation %>

<% provide(:title, 'Sign up') %>
<h1>Sign up</h1>

<div class="row">
  <div class="span6 offset3">
    <%= form_for(@user) do |f| %>
      <%= render 'fields', f: f %>
      <%= f.submit "Create my account", class: "btn btn-large btn-primary" %>
    <% end %>
  </div>
</div>

require 'spec_helper'

describe "Authentication" do
  .
  .
  .
  describe "authorization" do

    describe "for non-signed-in users" do
      .
      .
      .    
      describe "when attempting to visit a protected page" do
        before do
          visit edit_user_path(user)
          fill_in "Email",    with: user.email
          fill_in "Password", with: user.password
          click_button "Sign in"
        end

        describe "after signing in" do

          it "should render the desired protected page" do
            page.should have_selector('title', text: 'Edit user')
          end

          describe "when signing in again" do
            before do
              delete signout_path
              visit signin_path
              fill_in "Email",    with: user.email
              fill_in "Password", with: user.password
              click_button "Sign in"
            end

            it "should render the default (profile) page" do
              page.should have_selector('title', text: user.name) 
            end
          end
        end
      end
    end
    .
    .
    .
  end
end