Capybara

Capybara helps you test web applications by simulating how a real user would interact with your app. It is agnostic about the driver running your tests and comes with Rack::Test and Selenium support built in. WebKit is supported through an external gem.

If you and/or your company find value in Capybara and would like to contribute financially to its ongoing maintenance and development, please visit Patreon

Need help? Ask on the discussions (please do not open an issue): https://github.com/orgs/teamcapybara/discussions/categories/q-a

• Key benefits
• Setup
• Using Capybara with Cucumber
• Using Capybara with RSpec
• Using Capybara with Test::Unit
• Using Capybara with Minitest
• Using Capybara with Minitest::Spec
• Drivers
  • Selecting the Driver
  • RackTest
  • Selenium
• The DSL
  • Navigating
  • Clicking links and buttons
  • Interacting with forms
  • Querying
  • Finding
  • Scoping
  • Working with windows
  • Scripting
  • Modals
  • Debugging
• Selectors
  • Name
  • Locator
  • Filters
• Matching
  • Exactness
  • Strategy
• Transactions and database setup
• Asynchronous JavaScript (Ajax and friends)
• Using the DSL elsewhere
• Calling remote servers
• Using sessions
  • Named sessions
  • Using sessions manually
• XPath, CSS and selectors
• Beware the XPath // trap
• Configuring and adding drivers
• Gotchas:
• "Threadsafe" mode
• Development

• No setup necessary for Rails and Rack application. Works out of the box.
• Intuitive API which mimics the language an actual user would use.
• Switch the backend your tests run against from fast headless mode to an actual browser with no changes to your tests.
• Powerful synchronization features mean you never have to manually wait for asynchronous processes to complete.

Capybara requires Ruby 3.0.0 or later. To install, add this line to your Gemfile and run bundle install:

gem 'capybara'

If the application that you are testing is a Rails app, add this line to your test helper file:

require 'capybara/rails'

If the application that you are testing is a Rack app, but not Rails, set Capybara.app to your Rack app:

Capybara.app = MyRackApp

If you need to test JavaScript, or if your app interacts with (or is located at) a remote URL, you'll need to use a different driver. If using Rails 5.0+, but not using the Rails system tests from 5.1, you'll probably also want to swap the "server" used to launch your app to Puma in order to match Rails defaults.

Capybara.server = :puma # Until your setup is working
Capybara.server = :puma, { Silent: true } # To clean up your test output

The cucumber-rails gem comes with Capybara support built-in. If you are not using Rails, manually load the capybara/cucumber module:

require 'capybara/cucumber'
Capybara.app = MyRackApp

You can use the Capybara DSL in your steps, like so:

When /I sign in/ do
  within("#session") do
    fill_in 'Email', with: 'user@example.com'
    fill_in 'Password', with: 'password'
  end
  click_button 'Sign in'
end

You can switch to the Capybara.javascript_driver (:selenium by default) by tagging scenarios (or features) with @javascript:

@javascript
Scenario: do something Ajaxy
  When I click the Ajax link
  ...

There are also explicit tags for each registered driver set up for you (@selenium, @rack_test, etc).

Load RSpec 3.5+ support by adding the following line (typically to your spec_helper.rb file):

require 'capybara/rspec'

If you are using Rails, put your Capybara specs in spec/features or spec/system (only works if you have it configured in RSpec) and if you have your Capybara specs in a different directory, then tag the example groups with type: :feature or type: :system depending on which type of test you're writing.

If you are using Rails system specs please see their documentation for selecting the driver you wish to use.

If you are not using Rails, tag all the example groups in which you want to use Capybara with type: :feature.

You can now write your specs like so:

describe "the signin process", type: :feature do
  before :each do
    User.create(email: 'user@example.com', password: 'password')
  end

  it "signs me in" do
    visit '/sessions/new'
    within("#session") do
      fill_in 'Email', with: 'user@example.com'
      fill_in 'Password', with: 'password'
    end
    click_button 'Sign in'
    expect(page).to have_content 'Success'
  end
end

Use js: true to switch to the Capybara.javascript_driver (:selenium by default), or provide a :driver option to switch to one specific driver. For example:

describe 'some stuff which requires js', js: true do
  it 'will use the default js driver'
  it 'will switch to one specific driver', driver: :selenium
end

Capybara also comes with a built in DSL for creating descriptive acceptance tests:

feature "Signing in" do
  background do
    User.create(email: 'user@example.com', password: 'caplin')
  end

  scenario "Signing in with correct credentials" do
    visit '/sessions/new'
    within("#session") do
      fill_in 'Email', with: 'user@example.com'
      fill_in 'Password', with: 'caplin'
    end
    click_button 'Sign in'
    expect(page).to have_content 'Success'
  end

  given(:other_user) { User.create(email: 'other@example.com', password: 'rous') }

  scenario "Signing in as another user" do
    visit '/sessions/new'
    within("#session") do
      fill_in 'Email', with: other_user.email
      fill_in 'Password', with: other_user.password
    end
    click_button 'Sign in'
    expect(page).to have_content 'Invalid email or password'
  end
end

feature is in fact just an alias for describe ..., type: :feature, background is an alias for before, scenario for it, and given/given! aliases for let/let!, respectively.

Finally, Capybara matchers are also supported in view specs:

RSpec.describe "todos/show.html.erb", type: :view do
  it "displays the todo title" do
    assign :todo, Todo.new(title: "Buy milk")

    render

    expect(rendered).to have_css("header h1", text: "Buy milk")
  end
end

Note: When you require 'capybara/rspec' proxy methods are installed to work around name collisions between Capybara::DSL methods all/within and the identically named built-in RSpec matchers. If you opt not to require 'capybara/rspec' you can install the proxy methods by requiring 'capybara/rspec/matcher_proxies' after requiring RSpec and 'capybara/dsl'

• If you are using Test::Unit, define a base class for your Capybara tests like so:

  require 'capybara/dsl'
  
  class CapybaraTestCase < Test::Unit::TestCase
    include Capybara::DSL
  
    def teardown
      Capybara.reset_sessions!
      Capybara.use_default_driver
    end
  end

• If you are using Rails system tests please see their documentation for information on selecting the driver you wish to use.

• If you are using Rails, but not using Rails system tests, add the following code in your test_helper.rb file to make Capybara available in all test cases deriving from ActionDispatch::IntegrationTest:

  require 'capybara/rails'
  require 'capybara/minitest'
  
  class ActionDispatch::IntegrationTest
    # Make the Capybara DSL available in all integration tests
    include Capybara::DSL
    # Make `assert_*` methods behave like Minitest assertions
    include Capybara::Minitest::Assertions
  
    # Reset sessions and driver between tests
    teardown do
      Capybara.reset_sessions!
      Capybara.use_default_driver
    end
  end

• If you are not using Rails, define a base class for your Capybara tests like so:

  require 'capybara/minitest'
  
  class CapybaraTestCase < Minitest::Test
    include Capybara::DSL
    include Capybara::Minitest::Assertions
  
    def teardown
      Capybara.reset_sessions!
      Capybara.use_default_driver
    end
  end

  Remember to call super in any subclasses that override teardown.

To switch the driver, set Capybara.current_driver. For instance,

class BlogTest < ActionDispatch::IntegrationTest
  setup do
    Capybara.current_driver = Capybara.javascript_driver # :selenium by default
  end

  test 'shows blog posts' do
    # ... this test is run with Selenium ...
  end
end

Follow the above instructions for Minitest and additionally require capybara/minitest/spec

page.must_have_content('Important!')

Capybara uses the same DSL to drive a variety of browser and headless drivers.

By default, Capybara uses the :rack_test driver, which is fast but limited: it does not support JavaScript, nor is it able to access HTTP resources outside of your Rack application, such as remote APIs and OAuth services. To get around these limitations, you can set up a different default driver for your features. For example, if you'd prefer to run everything in Selenium, you could do:

Capybara.default_driver = :selenium # :selenium_chrome and :selenium_chrome_headless are also registered

However, if you are using RSpec or Cucumber (and your app runs correctly without JS), you may instead want to consider leaving the faster :rack_test as the default_driver, and marking only those tests that require a JavaScript-capable driver using js: true or @javascript, respectively. By default, JavaScript tests are run using the :selenium driver. You can change this by setting Capybara.javascript_driver.

You can also change the driver temporarily (typically in the Before/setup and After/teardown blocks):

Capybara.current_driver = :selenium # temporarily select different driver
# tests here
Capybara.use_default_driver       # switch back to default driver

Note: switching the driver creates a new session, so you may not be able to switch in the middle of a test.

RackTest is Capybara's default driver. It is written in pure Ruby and does not have any support for executing JavaScript. Since the RackTest driver interacts directly with Rack interfaces, it does not require a server to be started. However, this means that if your application is not a Rack application (Rails, Sinatra and most other Ruby frameworks are Rack applications) then you cannot use this driver. Furthermore, you cannot use the RackTest driver to test a remote application, or to access remote URLs (e.g., redirects to external sites, external APIs, or OAuth services) that your application might interact with.

capybara-mechanize provides a similar driver that can access remote servers.

RackTest can be configured with a set of headers like this:

Capybara.register_driver :rack_test do |app|
  Capybara::RackTest::Driver.new(app, headers: { 'HTTP_USER_AGENT' => 'Capybara' })
end

See the section on adding and configuring drivers.

Capybara supports Selenium 3.5+ (Webdriver). In order to use Selenium, you'll need to install the selenium-webdriver gem, and add it to your Gemfile if you're using bundler.

Capybara pre-registers a number of named drivers that use Selenium - they are:

• :selenium => Selenium driving Firefox
• :selenium_headless => Selenium driving Firefox in a headless configuration
• :selenium_chrome => Selenium driving Chrome
• :selenium_chrome_headless => Selenium driving Chrome in a headless configuration

These should work (with relevant software installation) in a local desktop configuration but you may need to customize them if using in a CI environment where additional options may need to be passed to the browsers. See the section on adding and configuring drivers.

Note: drivers which run the server in a different thread may not share the same transaction as your tests, causing data not to be shared between your test and test server, see Transactions and database setup below.

A complete reference is available at rubydoc.info.

Note: By default Capybara will only locate visible elements. This is because a real user would not be able to interact with non-visible elements.

Note: All searches in Capybara are case sensitive. This is because Capybara heavily uses XPath, which doesn't support case insensitivity.

You can use the visit method to navigate to other pages:

visit('/projects')
visit(post_comments_path(post))

The visit method only takes a single parameter, the request method is always GET.

You can get the current path of the browsing session, and test it using the have_current_path matcher:

expect(page).to have_current_path(post_comments_path(post))

Note: You can also assert the current path by testing the value of current_path directly. However, using the have_current_path matcher is safer since it uses Capybara's waiting behaviour to ensure that preceding actions (such as a click_link) have completed.

Full reference: Capybara::Node::Actions

You can interact with the webapp by following links and buttons. Capybara automatically follows any redirects, and submits forms associated with buttons.

click_link('id-of-link')
click_link('Link Text')
click_button('Save')
click_on('Link Text') # clicks on either links or buttons
click_on('Button Value')

Full reference: Capybara::Node::Actions

There are a number of tools for interacting with form elements:

fill_in('First Name', with: 'John')
fill_in('Password', with: 'Seekrit')
fill_in('Description', with: 'Really Long Text...')
choose('A Radio Button')
check('A Checkbox')
uncheck('A Checkbox')
attach_file('Image', '/path/to/image.jpg')
select('Option', from: 'Select Box')

Full reference: Capybara::Node::Matchers

Capybara has a rich set of options for querying the page for the existence of certain elements, and working with and manipulating those elements.

page.has_selector?('table tr')
page.has_selector?(:xpath, './/table/tr')

page.has_xpath?('.//table/tr')
page.has_css?('table tr.foo')
page.has_content?('foo')

Note: The negative forms like has_no_selector? are different from not has_selector?. Read the section on asynchronous JavaScript for an explanation.

You can use these with RSpec's magic matchers:

expect(page).to have_selector('table tr')
expect(page).to have_selector(:xpath, './/table/tr')

expect(page).to have_xpath('.//table/tr')
expect(page).to have_css('table tr.foo')
expect(page).to have_content('foo')

Full reference: Capybara::Node::Finders

You can also find specific elements, in order to manipulate them:

find_field('First Name').value
find_field(id: 'my_field').value
find_link('Hello', :visible => :all).visible?
find_link(class: ['some_class', 'some_other_class'], :visible => :all).visible?

find_button('Send').click
find_button(value: '1234').click

find(:xpath, ".//table/tr").click
find("#overlay").find("h1").click
all('a').each { |a| a[:href] }

If you need to find elements by additional attributes/properties you can also pass a filter block, which will be checked inside the normal waiting behavior. If you find yourself needing to use this a lot you may be better off adding a custom selector or adding a filter to an existing selector.

find_field('First Name'){ |el| el['data-xyz'] == '123' }
find("#img_loading"){ |img| img['complete'] == true }

Note: find will wait for an element to appear on the page, as explained in the Ajax section. If the element does not appear it will raise an error.

These elements all have all the Capybara DSL methods available, so you can restrict them to specific parts of the page:

find('#navigation').click_link('Home')
expect(find('#navigation')).to have_button('Sign out')

Capybara makes it possible to restrict certain actions, such as interacting with forms or clicking links and buttons, to within a specific area of the page. For this purpose you can use the generic within method. Optionally you can specify which kind of selector to use.

within("li#employee") do
  fill_in 'Name', with: 'Jimmy'
end

within(:xpath, ".//li[@id='employee']") do
  fill_in 'Name', with: 'Jimmy'
end

There are special methods for restricting the scope to a specific fieldset, identified by either an id or the text of the fieldset's legend tag, and to a specific table, identified by either id or text of the table's caption tag.

within_fieldset('Employee') do
  fill_in 'Name', with: 'Jimmy'
end

within_table('Employee') do
  fill_in 'Name', with: 'Jimmy'
end

Capybara provides some methods to ease finding and switching windows:

facebook_window = window_opened_by do
  click_button 'Like'
end
within_window facebook_window do
  find('#login_email').set('a@example.com')
  find('#login_password').set('qwerty')
  click_button 'Submit'
end

In drivers which support it, you can easily execute JavaScript:

page.execute_script("$('body').empty()")

For simple expressions, you can return the result of the script.

result = page.evaluate_script('4 + 4');

For more complicated scripts you'll need to write them as one expression.

result = page.evaluate_script(<<~JS, 3, element)
  (function(n, el){
    var val = parseInt(el.value, 10);
    return n+val;
  })(arguments[0], arguments[1])
JS

In drivers which support it, you can accept, dismiss and respond to alerts, confirms, and prompts.

You can accept alert messages by wrapping the code that produces an alert in a block:

accept_alert 'optional text or regex' do
  click_link('Show Alert')
end

You can accept or dismiss a confirmation by wrapping it in a block, as well:

accept_confirm 'optional text' do
  click_link('Show Confirm')
end

dismiss_confirm 'optional text' do
  click_link('Show Confirm')
end

You can accept or dismiss prompts as well, and also provide text to fill in for the response:

accept_prompt('optional text', with: 'Linus Torvalds') do
  click_link('Show Prompt About Linux')
end

dismiss_prompt('optional text') do
  click_link('Show Prompt About Linux')
end

All modal methods return the message that was presented. So, you can access the prompt message by assigning the return to a variable:

message = accept_prompt(with: 'Linus Torvalds') do
  click_link('Show Prompt About Linux')
end
expect(message).to eq('Who is the chief architect of Linux?')

It can be useful to take a snapshot of the page as it currently is and take a look at it:

save_and_open_page

You can also retrieve the current state of the DOM as a string using page.html.

print page.html

This is mostly useful for debugging. You should avoid testing against the contents of page.html and use the more expressive finder methods instead.

Finally, in drivers that support it, you can save a screenshot:

page.save_screenshot('screenshot.png')

Or have it save and automatically open:

save_and_open_screenshot

Screenshots are saved to Capybara.save_path, relative to the app directory. If you have required capybara/rails, Capybara.save_path will default to tmp/capybara.

Helpers and matchers that accept Selectors share a common method signature that includes:

1. a positional Name argument
2. a positional Locator argument
3. keyword Filter arguments
4. a predicate Filter block argument

These arguments are usually optional in one way or another.

The name argument determines the Selector to use. The argument is optional when a helper explicitly conveys the selector name (for example, find_field uses :field, find_link uses :link, etc):

page.html # => '<a href="/">Home</a>'

page.find(:link) == page.find_link

page.html # => '<input>'

page.find(:field) == page.find_field

The locator argument usually represents information that can most meaningfully distinguish an element that matches the selector from an element that does not:

page.html # => '<div id="greeting">Hello world</div>'

page.find(:css, 'div').text     # => 'Hello world'
page.find(:xpath, './/div').text # => 'Hello world'

General purpose finder methods like find and all can accept the locator as their first positional argument when the method can infer the default value from the Capybara.default_selector configuration:

page.html # => '<div id="greeting">Hello world</div>'

Capybara.default_selector = :css

page.find('div').text     # => 'Hello world'

Capybara.default_selector = :xpath

page.find('.//div').text # => 'Hello world'

The locator argument's semantics are context-specific, and depend on the selector. The types of arguments are varied. Some selectors support String or Regexp arguments, while others like :table_row support Array<String> and Hash<String, String>:

page.html # => '<label for="greeting">Greeting</label>
          #     <input id="greeting" name="content">'

# find by the <input> element's [id] attribute
page.find(:id, 'greeting') == page.find_by_id('greeting') # => true

# find by the <input> element's [id] attribute
page.find(:field, 'greeting') == page.find_field('greeting') # => true

# find by the <input> element's [name] attribute
page.find(:field, 'content') == page.find_field('content') # => true

# find by the <label> element's text
page.find(:field, 'Greeting') == page.find_field('Greeting') # => true

page.html # => '<table>
          #       <tr>
          #         <th>A</th>
          #         <th>B</th>
          #       </tr>
          #       <tr>
          #         <td>1</td>
          #         <td>2</td>
          #       </tr>
          #     </table>'

# find by <td> content
page.find(:table_row, ['1', '2']) == page.find(:css, 'tr:last-of-type') # => true

# find by <th> content paired with corresponding <td> content
page.find(:table_row, 'A' => '1') == page.find(:table_row, 'B' => '2') # => true

All filters are optional. The supported set of keys is a mixture of both global and context-specific filters.The supported types of values depend on the context:

page.html # => '<a href="/">Home</a>'

# find by the [href] attribute
page.find_link(href: '/') == page.find_link(text: 'Home') # => true

page.html # => '<div id="element" data-attribute="value">Content</div>'

# find by the [id] attribute
page.find(id: 'element') == page.find(text: 'Content') # => true

# find by the [data-attribute] attribute
page.find(:element, 'data-attribute': /value/) == page.find(text: 'Content') # => true

page.html # => '<input type="checkbox">'

# find by the absence of the [checked] attribute
page.find_field(checked: false) == page.find_field(unchecked: true) # => true

The predicate block is always optional. When there are results for a selector query, the block is called with each item in the result set. When the block evaluates to true, the item is included from the result set. Otherwise, the item is excluded:

page.html # => '<input role="switch" type="checkbox" checked>'

switch = page.find_field { |input| input["role"] == "switch" }
field = page.find_field(checked: true)

switch == field # => true

It is possible to customize how Capybara finds elements. At your disposal are two options, Capybara.exact and Capybara.match.

Capybara.exact and the exact option work together with the is expression inside the XPath gem. When exact is true, all is expressions match exactly, when it is false, they allow substring matches. Many of the selectors built into Capybara use the is expression. This way you can specify whether you want to allow substring matches or not. Capybara.exact is false by default.

For example:

click_link("Password") # also matches "Password confirmation"
Capybara.exact = true
click_link("Password") # does not match "Password confirmation"
click_link("Password", exact: false) # can be overridden

Using Capybara.match and the equivalent match option, you can control how Capybara behaves when multiple elements all match a query. There are currently four different strategies built into Capybara:

1. first: Just picks the first element that matches.
2. one: Raises an error if more than one element matches.
3. smart: If exact is true, raises an error if more than one element matches, just like one. If exact is false, it will first try to find an exact match. An error is raised if more than one element is found. If no element is found, a new search is performed which allows partial matches. If that search returns multiple matches, an error is raised.
4. prefer_exact: If multiple matches are found, some of which are exact, and some of which are not, then the first exactly matching element is returned.

The default for Capybara.match is :smart. To emulate the behaviour in Capybara 2.0.x, set Capybara.match to :one. To emulate the behaviour in Capybara 1.x, set Capybara.match to :prefer_exact.

Note: Rails 5.1+ "safely" shares the database connection between the app and test threads. Therefore, if using Rails 5.1+ you SHOULD be able to ignore this section.

Some Capybara drivers need to run against an actual HTTP server. Capybara takes care of this and starts one for you in the same process as your test, but on another thread. Selenium is one of those drivers, whereas RackTest is not.

If you are using a SQL database, it is common to run every test in a transaction, which is rolled back at the end of the test, rspec-rails does this by default out of the box for example. Since transactions are usually not shared across threads, this will cause data you have put into the database in your test code to be invisible to Capybara.

Cucumber handles this by using truncation instead of transactions, i.e. they empty out the entire database after each test. You can get the same behaviour by using a gem such as database_cleaner.

When working with asynchronous JavaScript, you might come across situations where you are attempting to interact with an element which is not yet present on the page. Capybara automatically deals with this by waiting for elements to appear on the page.

When issuing instructions to the DSL such as:

click_link('foo')
click_link('bar')
expect(page).to have_content('baz')

If clicking on the foo link triggers an asynchronous process, such as an Ajax request, which, when complete will add the bar link to the page, clicking on the bar link would be expected to fail, since that link doesn't exist yet. However, Capybara is smart enough to retry finding the link for a brief period of time before giving up and throwing an error. The same is true of the next line, which looks for the content baz on the page; it will retry looking for that content for a brief time. You can adjust how long this period is (the default is 2 seconds):

Capybara.default_max_wait_time = 5

Be aware that because of this behaviour, the following two statements are not equivalent, and you should always use the latter!

# Given use of a driver where the page is loaded when visit returns
# and that Capybara.predicates_wait is `true`
# consider a page where the `a` tag is removed through AJAX after 1s
visit(some_path)
!page.has_xpath?('a')  # is false
page.has_no_xpath?('a')  # is true

First expression:

• has_xpath?('a') is called right after visit returns. It is true because the link has not yet been removed
• Capybara does not wait upon successful predicates/assertions, therefore has_xpath? returns true immediately
• The expression returns false (because it is negated with the leading !)

Second expression:

• has_no_xpath?('a') is called right after visit returns. It is false because the link has not yet been removed.
• Capybara waits upon failed predicates/assertions, therefore has_no_xpath? does not return false immediately
• Capybara will periodically re-check the predicate/assertion up to the default_max_wait_time defined
• after 1s, the predicate becomes true (because the link has been removed)
• The expression returns true

Capybara's RSpec matchers, however, are smart enough to handle either form. The two following statements are functionally equivalent:

expect(page).not_to have_xpath('a')
expect(page).to have_no_xpath('a')

Capybara's waiting behaviour is quite advanced, and can deal with situations such as the following line of code:

expect(find('#sidebar').find('h1')).to have_content('Something')

Even if JavaScript causes #sidebar to disappear off the page, Capybara will automatically reload it and any elements it contains. So if an AJAX request causes the contents of #sidebar to change, which would update the text of the h1 to "Something", and this happened, this test would pass. If you do not want this behaviour, you can set Capybara.automatic_reload to false.

You can mix the DSL into any context by including Capybara::DSL:

require 'capybara/dsl'

Capybara.default_driver = :webkit

module MyModule
  include Capybara::DSL

  def login!
    within(:xpath, ".//form[@id='session']") do
      fill_in 'Email', with: 'user@example.com'
      fill_in 'Password', with: 'password'
    end
    click_button 'Sign in'
  end
end

This enables its use in unsupported testing frameworks, and for general-purpose scripting.

Normally Capybara expects to be testing an in-process Rack application, but you can also use it to talk to a web server running anywhere on the internet, by setting app_host:

Capybara.current_driver = :selenium
Capybara.app_host = 'http://www.google.com'
...
visit('/')

Note: the default driver (:rack_test) does not support running against a remote server. With drivers that support it, you can also visit any URL directly:

visit('http://www.google.com')

By default Capybara will try to boot a rack application automatically. You might want to switch off Capybara's rack server if you are running against a remote application:

Capybara.run_server = false

Capybara manages named sessions (:default if not specified) allowing multiple sessions using the same driver and test app instance to be interacted with. A new session will be created using the current driver if a session with the given name using the current driver and test app instance is not found.

To perform operations in a different session and then revert to the previous session

Capybara.using_session("Bob's session") do
   #do something in Bob's browser session
end
 #reverts to previous session

To permanently switch the current session to a different session

Capybara.session_name = "some other session"

For ultimate control, you can instantiate and use a Session manually.

require 'capybara'

session = Capybara::Session.new(:webkit, my_rack_app)
session.within("form#session") do
  session.fill_in 'Email', with: 'user@example.com'
  session.fill_in 'Password', with: 'password'
end
session.click_button 'Sign in'

Capybara does not try to guess what kind of selector you are going to give it, and will always use CSS by default. If you want to use XPath, you'll need to do:

within(:xpath, './/ul/li') { ... }
find(:xpath, './/ul/li').text
find(:xpath, './/li[contains(.//a[@href = "#"]/text(), "foo")]').value

Alternatively you can set the default selector to XPath:

Capybara.default_selector = :xpath
find('.//ul/li').text

Capybara provides a number of other built-in selector types. The full list, along with applicable filters, can be seen at built-in selectors

Capybara also allows you to add custom selectors, which can be very useful if you find yourself using the same kinds of selectors very often. The examples below are very simple, and there are many available features not demonstrated. For more in-depth examples please see Capybaras built-in selector definitions.

Capybara.add_selector(:my_attribute) do
  xpath { |id| XPath.descendant[XPath.attr(:my_attribute) == id.to_s] }
end

Capybara.add_selector(:row) do
  xpath { |num| ".//tbody/tr[#{num}]" }
end

Capybara.add_selector(:flash_type) do
  css { |type| "#flash.#{type}" }
end

The block given to xpath must always return an XPath expression as a String, or an XPath expression generated through the XPath gem. You can now use these selectors like this:

find(:my_attribute, 'post_123') # find element with matching attribute
find(:row, 3) # find 3rd row in table body
find(:flash_type, :notice) # find element with id of 'flash' and class of 'notice'

In XPath the expression // means something very specific, and it might not be what you think. Contrary to common belief, // means "anywhere in the document" not "anywhere in the current context". As an example:

page.find(:xpath, '//body').all(:xpath, '//script')

You might expect this to find all script tags in the body, but actually, it finds all script tags in the entire document, not only those in the body! What you're looking for is the .// expression which means "any descendant of the current node":

page.find(:xpath, '//body').all(:xpath, './/script')

The same thing goes for within:

within(:xpath, '//body') do
  page.find(:xpath, './/script')
  within(:xpath, './/table/tbody') do
  ...
  end
end

Capybara makes it convenient to switch between different drivers. It also exposes an API to tweak those drivers with whatever settings you want, or to add your own drivers. This is how to override the selenium driver configuration to use chrome:

Capybara.register_driver :selenium do |app|
  Capybara::Selenium::Driver.new(app, :browser => :chrome)
end

However, it's also possible to give this configuration a different name.

# Note: Capybara registers this by default
Capybara.register_driver :selenium_chrome do |app|
  Capybara::Selenium::Driver.new(app, :browser => :chrome)
end

Then tests can switch between using different browsers effortlessly:

Capybara.current_driver = :selenium_chrome

Whatever is returned from the block should conform to the API described by Capybara::Driver::Base, it does not however have to inherit from this class. Gems can use this API to add their own drivers to Capybara.

The Selenium wiki has additional info about how the underlying driver can be configured.

• Access to session and request is not possible from the test, Access to response is limited. Some drivers allow access to response headers and HTTP status code, but this kind of functionality is not provided by some drivers, such as Selenium.

• Access to Rails specific stuff (such as controller) is unavailable, since we're not using Rails' integration testing.

• Freezing time: It's common practice to mock out the Time so that features that depend on the current Date work as expected. This can be problematic on ruby/platform combinations that don't support access to a monotonic process clock, since Capybara's Ajax timing uses the system time, resulting in Capybara never timing out and just hanging when a failure occurs. It's still possible to use gems which allow you to travel in time, rather than freeze time. One such gem is Timecop.

• When using Rack::Test, beware if attempting to visit absolute URLs. For example, a session might not be shared between visits to posts_path and posts_url. If testing an absolute URL in an Action Mailer email, set default_url_options to match the Rails default of www.example.com.

• Server errors will only be raised in the session that initiates the server thread. If you are testing for specific server errors and using multiple sessions make sure to test for the errors using the initial session (usually :default)

• If WebMock is enabled, you may encounter a "Too many open files" error. A simple page.find call may cause thousands of HTTP requests until the timeout occurs. By default, WebMock will cause each of these requests to spawn a new connection. To work around this problem, you may need to enable WebMock's net_http_connect_on_start: true parameter.

In normal mode most of Capybara's configuration options are global settings which can cause issues if using multiple sessions and wanting to change a setting for only one of the sessions. To provide support for this type of usage Capybara now provides a "threadsafe" mode which can be enabled by setting

Capybara.threadsafe = true

This setting can only be changed before any sessions have been created. In "threadsafe" mode the following behaviors of Capybara change

• Most options can now be set on a session. These can either be set at session creation time or after, and default to the global options at the time of session creation. Options which are NOT session specific are app, reuse_server, default_driver, javascript_driver, and (obviously) threadsafe. Any drivers and servers registered through register_driver and register_server are also global.

  my_session = Capybara::Session.new(:driver, some_app) do |config|
    config.automatic_label_click = true # only set for my_session
  end
  my_session.config.default_max_wait_time = 10 # only set for my_session
  Capybara.default_max_wait_time = 2 # will not change the default_max_wait in my_session

• current_driver and session_name are thread specific. This means that using_session and using_driver also only affect the current thread.

To set up a development environment, simply do:

bundle install
bundle exec rake  # run the test suite with Firefox - requires `geckodriver` to be installed
bundle exec rake spec_chrome # run the test suite with Chrome - require `chromedriver` to be installed

See CONTRIBUTING.md for how to send issues and pull requests.