3. Acceptance Testing

Acceptance Testing

Acceptance testing can be performed by a non-technical person. That person can be your tester, manager or even client. If you are developing a web-application (and probably you are) the tester needs nothing more than a web browser to check that your site works correctly. You can reproduce an acceptance tester’s actions in scenarios and run them automatically after each site change. Codeception keeps tests clean and simple, as if they were recorded from the words of an actual acceptance tester.

It makes no difference what CMS or Framework is used on the site. You can even test sites created on different platforms, like Java, .NET, etc. It’s always a good idea to add tests to your web site. At least you will be sure that site features work after the latest changes were made.

Sample Scenario

Let’s say the first test you would want to run would be signing in. In order to write such a test, we still require basic knowledge of PHP and HTML:

$I->fillField('username', 'davert');
$I->fillField('password', 'qwerty');
$I->see('Welcome, Davert!');

This scenario can be performed either by a simple PHP Browser or by a browser with Selenium WebDriver.

PhpBrowser WebDriver
Browser Engine Guzzle + Symfony BrowserKit Chrome or Firefox
JavaScript No Yes
see/seeElement checks if… …text is present in source code …text is actually visible to the user
Read HTTP response headers Yes No
Speed Fast Slow

We will start writing our first acceptance tests with a PhpBrowser.

PHP Browser

This is the fastest way to run acceptance tests, since it doesn’t require running an actual browser. We use a PHP web scraper, which acts like a browser: it sends a request, then receives and parses the response. Codeception uses Guzzle and Symfony BrowserKit to interact with HTML web pages. Please note that you can’t test actual visibility of HTML elements, or JavaScript interactions. Good thing about PhpBrowser is that it can be run in any environment with just PHP and cURL required.

Common PhpBrowser drawbacks:

  • you can click only on links with valid URLs or form submit buttons
  • you can’t fill in fields that are not inside a form
  • you can’t work with JavaScript interactions: modal windows, datepickers, etc.

Before we start, we need a local copy of the site running on your host. We need to specify the url parameter in the acceptance suite config (tests/acceptance.suite.yml):

actor: AcceptanceTester
        - PhpBrowser:
        - \Helper\Acceptance

We should start by creating a ‘Cept’ file in the tests/acceptance directory. Let’s call it SigninCept.php. We will write the first lines into it:

$I = new AcceptanceTester($scenario);
$I->wantTo('sign in');

The $I object is used to write all interactions. The methods of the $I object are taken from the PhpBrowser module. We will briefly describe them here:


We will assume that all actions starting with am and have describe the initial environment. The amOnPage action sets the starting point of a test to the /login page.

With the PhpBrowser you can click the links and fill in the forms. That will probably be the majority of your actions.


Emulates a click on valid anchors. The URL referenced in the href attribute will be opened. As a parameter you can specify the link name or a valid CSS or XPath selector.

$I->click('Log in');
// CSS selector applied
$I->click('#login a');
// XPath
// Using context as second argument
$I->click('Login', '.nav');

Codeception tries to locate an element by its text, name, CSS or XPath. You can specify the locator type manually by passing an array as a parameter. We call this a strict locator. Available strict locator types are:

  • id
  • name
  • css
  • xpath
  • link
  • class
// By specifying locator type
$I->click(['link' => 'Login']);
$I->click(['class' => 'btn']);

There is a special class Codeception\Util\Locator which may help you to generate complex XPath locators. For instance, it can easily allow you to click an element on the last row of a table:

$I->click('Edit' , \Codeception\Util\Locator::elementAt('//table/tr', -1));


Clicking links is probably not what takes the most time during the testing of a web site. The most routine waste of time goes into the testing of forms. Codeception provides several ways of testing forms.

Let’s submit this sample form inside the Codeception test:

<form method="post" action="/update" id="update_form">
     <label for="user_name">Name</label>
     <input type="text" name="user[name]" id="user_name" />
     <label for="user_email">Email</label>
     <input type="text" name="user[email]" id="user_email" />
     <label for="user_gender">Gender</label>
     <select id="user_gender" name="user[gender]">
          <option value="m">Male</option>
          <option value="f">Female</option>
     <input type="submit" name="submitButton" value="Update" />

From a user’s perspective, a form consists of fields which should be filled in, and then a submit button clicked:

// we are using label to match user_name field
$I->fillField('Name', 'Miles');
// we can use input name or id

To match fields by their labels, you should write a for attribute in the label tag.

From the developer’s perspective, submitting a form is just sending a valid POST request to the server. Sometimes it’s easier to fill in all of the fields at once and send the form without clicking a ‘Submit’ button. A similar scenario can be rewritten with only one command:

$I->submitForm('#update_form', array('user' => array(
     'name' => 'Miles',
     'email' => 'Davis',
     'gender' => 'm'

The submitForm is not emulating a user’s actions, but it’s quite useful in situations when the form is not formatted properly, for example to discover that labels aren’t set or that fields have unclean names or badly written IDs, or the form is sent by a JavaScript call.

By default, submitForm doesn’t send values for buttons. The last parameter allows specifying what button values should be sent, or button values can be explicitly specified in the second parameter:

$I->submitForm('#update_form', array('user' => array(
     'name' => 'Miles',
     'email' => 'Davis',
     'gender' => 'm'
)), 'submitButton');
// this would have the same effect, but the value has to be explicitly specified
$I->submitForm('#update_form', array('user' => array(
     'name' => 'Miles',
     'email' => 'Davis',
     'gender' => 'm',
     'submitButton' => 'Update'


In the PhpBrowser you can test the page contents. In most cases you just need to check that the required text or element is on the page.

The most useful method for this is see():

// We check that 'Thank you, Miles' is on the page.
$I->see('Thank you, Miles');
// We check that 'Thank you, Miles' is inside an element with 'notice' class.
$I->see('Thank you, Miles', '.notice');
// Or using XPath locators
$I->see('Thank you, Miles', "//table/tr[2]");
// We check this message is *not* on the page.
$I->dontSee('Form is filled incorrectly');

You can check that a specific HTML element exists (or doesn’t) on a page:


We also have other useful commands to perform checks. Please note that they all start with the see prefix:

$I->seeInField('user[name]', 'Miles');

Conditional Assertions

Usually, as soon as any assertion fails, further assertions of this test will be skipped. Sometimes you don’t want this - maybe you have a long-running test and you want it to run to the end. In this case you can use conditional assertions. Each see method has a corresponding canSee method, and dontSee has a cantSee method:

$I->cantSeeInField('user[name]', 'Miles');

Each failed assertion will be shown in the test results, but it won’t stop the test.


Within a long scenario you should describe what actions you are going to perform and what results should be achieved. Comment methods like amGoingTo, expect, expectTo help you in making tests more descriptive:

$I->amGoingTo('submit user form with invalid values');
$I->fillField('user[email]', 'miles');
$I->expect('the form is not submitted');
$I->see('Form is filled incorrectly');


These commands retrieve data that can be used in the test. Imagine your site generates a password for every user and you want to check that the user can log into the site using this password:

$I->fillField('email', 'miles@davis.com')
$I->click('Generate Password');
$password = $I->grabTextFrom('#password');
$I->fillField('email', 'miles@davis.com');
$I->fillField('password', $password);
$I->click('Log in!');

Grabbers allow you to get a single value from the current page with commands:

$token = $I->grabTextFrom('.token');
$password = $I->grabTextFrom("descendant::input/descendant::*[@id = 'password']");
$api_key = $I->grabValueFrom('input[name=api]');

Cookies, URLs, Title, etc

Actions for cookies:

$I->setCookie('auth', '123345');

Actions for checking the page title:


Actions for URLs:

$user_id = $I->grabFromCurrentUrl('~$/user/(\d+)/~');

WebDriver: Selenium Server or PhantomJS

A nice feature of Codeception is that most scenarios are similar, no matter of how they are executed. PhpBrowser was emulating browser requests but how to execute such test in a real browser like Chrome or Firefox? Selenium WebDriver can drive them so in our acceptance tests we can automate scenarios we used to test manually. In such tests we should concentrate more on testing the UI than on testing functionality.

Selenium WebDriver” is the name of an API (specified by the Selenium project) to drive browsers automatically. Codeception supports two implementations of this API: Selenium Standalone Server and PhantomJS

Selenium Standalone Server

Selenium Server is a piece of software that “drives” (i.e. runs) your local browser by sending commands (i.e. your test cases) to it.

Overview of components:

Codeception -> facebook/php-webdriver -> Selenium Server -> Gecko/Chrome Driver -> Browser (Chrome or Firefox)

Installation instructions:

  1. You need Chrome and/or Firefox :-)
  2. You need Java
  3. Download Selenium Standalone Server
  4. Download ChromeDriver and/or geckodriver (for Firefox) and extract the executable to a folder where Selenium Server can find it (best guess: just drop it in the same folder where selenium-server-standalone-xxx.jar lives)
  5. Start Selenium Server with:
    java -jar "/path/to/selenium-server-standalone-xxx.jar"

facebook/php-webdriver comes bundled with Codeception.


PhantomJS is a headless browser based on the WebKit rendering engine that renders your application like a real browser would. It just doesn’t display the outcome to you.

Important notice: PhantomJS is not maintained anymore. Use it at your own risk.

Overview of components:

Codeception -> facebook/php-webdriver -> PhantomJS

Installation instructions:

  1. Download PhantomJS and extract it.
  2. Start the executable (located in the bin folder) with:
    phantomjs --webdriver=4444

Full list of available command line arguments for PhantomJS

facebook/php-webdriver comes bundled with Codeception.

Codeception Configuration

To execute a test in a browser we need to change the suite configuration to use WebDriver instead of PhpBrowser.

Modify your acceptance.suite.yml file:

actor: AcceptanceTester
        - WebDriver:
            browser: chrome
        - \Helper\Acceptance

See WebDriver Module for details.

Please note that actions executed in a browser will behave differently. For instance, seeElement won’t just check that the element exists on a page, but it will also check that element is actually visible to the user:


While WebDriver duplicates the functionality of PhpBrowser, it has its limitations: It can’t check headers, since browsers don’t provide APIs for that. WebDriver also adds browser-specific functionality:


While testing web application, you may need to wait for JavaScript events to occur. Due to its asynchronous nature, complex JavaScript interactions are hard to test. That’s why you may need to use waiters, actions with wait prefix. They can be used to specify what event you expect to occur on a page, before continuing the test.

For example:

$I->waitForElement('#agree_button', 30); // secs

In this case we are waiting for the ‘agree’ button to appear and then clicking it. If it didn’t appear after 30 seconds, the test will fail. There are other wait methods you may use, like waitForText, waitForElementVisible and others.

If you don’t know what exact element you need to wait for, you can simply pause execution with using $I->wait()

$I->wait(3); // wait for 3 secs

Wait and Act

To combine waitForElement with actions inside that element you can use the performOn method. Let’s see how you can perform some actions inside an HTML popup:

$I->performOn('.confirm', \Codeception\Util\ActionSequence::build()
    ->see('Are you sure you want to delete this?')

Alternatively, this can be executed using a callback, in this case the WebDriver instance is passed as argument

$I->performOn('.confirm', function(\Codeception\Module\WebDriver $I) {
    $I->see('Are you sure you want to delete this?');

For more options see performOn() reference.

Multi Session Testing

Codeception allows you to execute actions in concurrent sessions. The most obvious case for this is testing realtime messaging between users on a site. In order to do it, you will need to launch two browser windows at the same time for the same test. Codeception has a very smart concept for doing this. It is called Friends:

$nick = $I->haveFriend('nick');
$nick->does(function(AcceptanceTester $I) {
    $I->fillField('body', 'Hello all!');
    $I->see('Hello all!', '.message');
$I->see('Hello all!', '.message');

In this case we performed, or ‘did’, some actions in the second window with the does method on a friend object.

Sometimes you may want to close a web page before the end of the test. For such cases you may use leave(). You can also specify roles for a friend:

$nickAdmin = $I->haveFriend('nickAdmin', adminStep::class);
$nickAdmin->does(function(adminStep $I) {
    // Admin does ...

Cloud Testing

Selenium WebDriver allows us to execute tests in real browsers on different platforms. Some environments are hard to be reproduced manually, testing Internet Explorer 6-8 on Windows XP may be a hard thing, especially if you don’t have Windows XP installed. This is where Cloud Testing services come to help you. Services such as SauceLabs, BrowserStack and others can create virtual machines on demand and set up Selenium Server and the desired browser. Tests are executed on a remote machine in a cloud, to access local files cloud testing services provide a special application called Tunnel. Tunnel operates on a secured protocol and allows browsers executed in a cloud to connect to a local web server.

Cloud Testing services work with the standard WebDriver protocol. This makes setting up cloud testing really easy. You just need to set the WebDriver configuration to:

  • specify the host to connect to (depends on the cloud provider)
  • authentication details (to use your account)
  • browser
  • OS

We recommend using params to provide authorization credentials.

It should be mentioned that Cloud Testing services are not free. You should investigate their pricing models and choose one that fits your needs. They also may work painfully slowly if ping times between the local server and the cloud is too high. This may lead to random failures in acceptance tests.

AngularJS Testing

In the modern era of Single Page Applications, the browser replaces the server in creating the user interface. Unlike traditional web applications, web pages are not reloaded on user actions. All interactions with the server is done in JavaScript with XHR requests. However, testing Single Page Applications can be a hard task. There could be no information of the application state: e.g. has it completed rendering or not? What is possible to do in this case is to use more wait* methods or execute JavaScript that checks the application state.

For applications built with the AngularJS v1.x framework we implemented AngularJS module which is based on Protractor (an official tool for testing Angular apps). Under the hood, it pauses step execution before the previous actions are completed and uses the AngularJS API to check the application state.

The AngularJS module extends WebDriver so that all the configuration options from it are available.


Codeception modules can print valuable information while running. Just execute tests with the --debug option to see running details. For any custom output use the codecept_debug function:


On each failure, the snapshot of the last shown page will be stored in the tests/_output directory. PhpBrowser will store the HTML code and WebDriver will save a screenshot of the page.

Sometimes you may want to inspect a web page opened by a running test. For such cases you may use the pauseExecution method of the WebDriver module.

You can also record your tests step by step and review the execution flow as a slideshow with the help of the Recorder extension.


Writing acceptance tests with Codeception and PhpBrowser is a good start. You can easily test your Joomla, Drupal, WordPress sites, as well as those made with frameworks. Writing acceptance tests is like describing a tester’s actions in PHP. They are quite readable and very easy to write. If you need to access the database, you can use the Db Module.

© 2011–2017 Michael Bodnarchuk and contributors
Licensed under the MIT License.