Behat.phar Mink.phar GitHub

Web acceptance testing

One of the most important parts in the web is a browser. Browser is the window, through which web users interact with web applications and other users. Users are always talking with web applications through browsers.

So, in order to test, that our web application behaves correctly, we need a way to simulate this interaction between browser and web application in our tests. We need a Mink.

Mink is an open source acceptance test framework for web applications, written in PHP 5.3.


Mink is a php 5.3 library that you’ll use inside your test suites or project. Before you begin, ensure that you have at least PHP 5.3.1 installed.

Method #1 (Composer)

The simplest way to install Mink with all its dependencies is through Composer:

Create composer.json file in the project root:

    "require": {
        "behat/mink": "1.4@stable"

Then download composer.phar and run install command:

$ curl http://getcomposer.org/installer | php
$ php composer.phar install

Everything will be installed inside vendor folder. Finally, include Composer autoloading script to your project:

require_once 'vendor/autoload.php';


By default, Mink will be installed with no drivers. In order to be able to use additional drivers, you should install (through composer) them. Add appropriate dependencies to your composer.json require block:

  • GoutteDriver - "behat/mink-goutte-driver": "*"
  • SeleniumDriver - "behat/mink-selenium-driver": "*"
  • WebDriver - "behat/mink-selenium2-driver": "*"
  • SahiDriver - "behat/mink-sahi-driver": "*"
  • ZombieDriver - "behat/mink-zombie-driver": "*"

If you’re newcomer or just don’t know what to choose - configure all available drivers with this composer.json (you will be able to tuneup it later):

    "require": {
        "behat/mink": "*",

        "behat/mink-goutte-driver": "*",
        "behat/mink-selenium-driver": "*",
        "behat/mink-selenium2-driver": "*",
        "behat/mink-sahi-driver": "*",
        "behat/mink-zombie-driver": "*"

Method #2 (PHAR)

Also, you can use mink phar package:

$ wget https://github.com/downloads/Behat/Mink/mink.phar

Now you can require phar package in your project:

require_once 'mink.phar';


Minks PHAR package comes bundles with all drivers, so there’s no need to search or download them manually.

Method #3 (Git)

You can also clone the Mink with Git by running:

$ git clone git://github.com/Behat/Mink.git && cd Mink

Then download composer.phar and run install command:

$ curl http://getcomposer.org/installer | php
$ php composer.phar install

Now, you can use Mink in your projects simply by including it:

require_once '/path/to/Mink/vendor/autoload.php';

Understanding the Mink

There’s huge amount of browser emulators out there, like Goutte, Selenium, Sahi and others. They all do the same job, but do it very differently. They behave differently and have very different API’s. But, what’s more important - there’s actually 2 completely different types of browser emulators out there:

  • Headless browser emulators
  • Browser controllers

First type browsers are simple pure HTTP specification implementations, like Goutte. Those browser emulators send a real HTTP requests against an application and parse the response content. They are very simple to run and configure, because this type of emulators can be written in any available programming language and can be run through console on servers without GUI. Headless emulators have both, advantages and disadvantages. Advantages are simplicity, speed and ability to run it without the need in real browser. But this type of browsers have one big disadvantage - they have no JS/AJAX support. So, you can’t test your rich GUI web applications with headless browsers.

Second browser emulators type are browser controllers. Those emulators aims to control the real browser. That’s right, a program to control another program. Browser controllers simulate user interactions on browser and are able to retrieve actual information from current browser page. Selenium and Sahi are two most famous browser controllers. The main advantage of browser controllers usage is the support for JS/AJAX interactions on page. Disadvantage is that such browser emulators require installed browser, extra configuration are usually much slower than headless counterparts.

So, the easy answer is to choose the best emulator for your project and use its API for testing. But as we’ve already seen, both browser types have both, advantages and disadvantages. If you choose headless browser emulator - you’ll not be able to test your JS/AJAX pages. And if you choose browser controller - your overall test suite will become very slow at some point. So, in real world we should use both! And that’s why you need a Mink.

Mink removes API differences between different browser emulators providing different drivers (read in “Different Browsers - Drivers” chapter) for every browser emulator and providing you with the easy way to control the browser (“Control the Browser - Session”), traverse pages (“Traverse the Page - Selectors”) or manipulate page elements (“Manipulate the Page - NodeElement”).

Different Browsers - Drivers

How does Mink provide a consistent API for very different browser library types, often written in different languages? Through drivers! Mink driver is a simple class, that implements Behat\Mink\Driver\DriverInterface. This interface describes bridge methods between Mink and real browser emulators. Mink always talks with browser emulators through its driver - it doesn’t know anything about how to start/stop or traverse page in that particular browser emulator - it only knows what driver method it should call in order to do this.

Mink v1.2 comes with four drivers out of the box:

  • GoutteDriver - provides a bridge for Goutte headless browser. Goutte is a classical pure-php headless browser, written by the creator of the Symfony framework - Fabien Potencier.
  • SahiDriver - provides a bridge for Sahi browser controller. Sahi is a new JS browser controller, that fast replaced old Selenium testing suite. It’s both, easier to setup and to use than classical Selenium. It has a GUI installer for each popular operating system out there and is able to control every systems browser through a special bundled proxy server.

New in version 1.1.

  • ZombieDriver - provides a bridge for Zombie.js browser emulator. Zombie.js is a headless browser emulator, written on node.js. It supports all JS interactions that Sahi does and works almost as fast as Goutte does. It’s best of both worlds, actually, but still limited to only one browser type (Chromium), also it’s still slower than Goutte and requires node.js and npm to be installed on the system.

New in version 1.2.

  • SeleniumDriver - provides a bridge for famous Selenium tool. If you need legacy Selenium - you can use it right out of the box in your Behat/Mink test suites.

New in version 1.3.

  • Selenium2Driver - provides a bridge for Selenium2 (webdriver) tool. If you just love Selenium2 - you can now use it right out of the box too.


In order to talk with Goutte, you should instantiate a Driver\GoutteDriver class:

$driver = new \Behat\Mink\Driver\GoutteDriver();

Also, if you want to configure Goutte more precisely, you could do the full setup by hand:

$clientOptions = array();

$client = new \Behat\Mink\Driver\Goutte\Client();
$client->setClient(new \Guzzle\Http\Client('', $clientOptions));
$driver = new \Behat\Mink\Driver\GoutteDriver($client);


$clientOptions is an array of parameters for Guzzle HTTP client, which Goutte uses internally. You can read about Guzzle client parameters here.


In order to talk with a real browser through Sahi, you should install and configure Sahi first:

  1. Download and run the Sahi jar from the http://sourceforge.net/projects/sahi/files/ and run it. It will run the installer, which will guide you through the installation process.

  2. Run Sahi proxy before your test suites (you can start this proxy during system startup):

    cd $YOUR_PATH_TO_SAHI/bin

After installing Sahi and running Sahi proxy server, you’ll be able to control it with Driver\SahiDriver:

$driver = new \Behat\Mink\Driver\SahiDriver('firefox');


Notice, that first argument to SahiDriver is always a browser name, supported by Sahi.

If you want more control during driver initialization, like for example if you want to configure the driver to talk with a proxy on another machine - use the more verbose version with a second client argument:

$driver = new \Behat\Mink\Driver\SahiDriver('firefox',
    new \Behat\SahiClient\Client(
        new \Behat\SahiClient\Connection($sid, $host, $port)


$sid is a Sahi session ID. It’s a unique string, used by the driver and Sahi proxy in order to be able to talk with each other. You should fill this with null if you want Sahi to start your browser automatically or with some unique string if you want to control an already started browser.

$host simply defines the host on which Sahi is started. It’s localhost by default.

$port defines a Sahi proxy port. Default one is 9999.


New in version 1.1.

In order to talk with zombie.js server, you should install and configure zombie.js first:

  1. Install node.js by following instructions from the official site: http://nodejs.org/.

  2. Install npm (node package manager) by following instructions from the http://npmjs.org/.

  3. Install zombie.js with npm:

    $ npm install -g zombie

After installing npm and zombie.js, you’ll need to add npm libs to your NODE_PATH. The easiest way to do this is to add:

export NODE_PATH="/PATH/TO/NPM/node_modules"

into your .bashrc.

After that, you’ll be able to just use ZombieDriver without manual server setup - the driver will do all that for you automatically:

$driver = new \Behat\Mink\Driver\ZombieDriver();

If you want more control during driver initialization, like for example if you want to configure the driver to init the server on a specific port - use the more verbose version with 2 arguments:

$driver = new \Behat\Mink\Driver\ZombieDriver(
    new \Behat\Mink\Driver\Zombie\Connection($host, $port),
    new \Behat\Mink\Driver\Zombie\Server($host, $port, $nodeBin, $script)


$host simply defines the host on which zombie.js will be started. It’s by default.

$port defines a zombie.js port. Default one is 8124.

$nodeBin defines full path to node.js binary. Default one is just node.

$script defines a node.js script to start zombie.js server. If you pass a null the default script will be used. Use this option carefully!


New in version 1.2.

In order to talk with selenium server, you should install and configure it first:

  1. Download Selenium Server (formerly the Selenium RC Server) from here: http://seleniumhq.org/download/.

  2. Run server with command:

    $ java -jar selenium-server-standalone-2.11.0.jar

That’s it, now you can use SeleniumDriver:

$client = new \Selenium\Client($host, $port);
$driver = new \Behat\Mink\Driver\SeleniumDriver(
    'firefox', 'base_url', $client


New in version 1.3.

In order to talk with selenium server, you should install and configure it first:

  1. Download Selenium Server (formerly the Selenium RC Server) from here: http://seleniumhq.org/download/.

  2. Run server with command:

    $ java -jar selenium-server-standalone-2.11.0.jar

That’s it, now you can use Selenium2Driver:

$driver = new \Behat\Mink\Driver\Selenium2Driver(
    'firefox', 'base_url'

Control the Browser - Session

Ok. Now we know how to create the browser driver to talk with a specific browser emulator. Although we can use drivers directly to call some actions on the emulator, Mink provides a better way - Session:

// init session:
$session = new \Behat\Mink\Session($driver);

// start session:


As you can see, the first argument to the session ($driver) is just a simple driver instance, which we created in the previous chapter.

start() call is required in order to configure the browser emulator or controller to be fully functional.

Basic Browser Interaction

After you’ve instantiated the $session object, you can control the actual browser emulator with it:

// open some page in browser:

// get the current page URL:
echo $session->getCurrentUrl();

// get the response status code:
echo $session->getStatusCode();

// get page content:
echo $session->getPage()->getContent();

// open another page:

// use history controls:

// evaluate JS expression:
echo $session->evaluateScript(
    "(function(){ return 'something from browser'; })()"

// wait for n milliseconds or
// till JS expression becomes true:
    "$('.suggestions-results').children().length > 0"


Although Mink does its best on removing browser differences between different browser emulators - it can’t do much in some cases. For example, GoutteDriver can’t evaluate JavaScript and SahiDriver can’t get the response status code. In such cases, the driver will always throw meaningful Behat\Mink\Exception\UnsupportedDriverActionException.

Cookies and Headers management

With Mink\Session you can control your browsers cookies and headers:

// setting browser language:
$session->setRequestHeader('Accept-Language', 'fr');

// retrieving response headers:

// set cookie:
$session->setCookie('cookie name', 'value');

// get cookie:
echo $session->getCookie('cookie name');

// delete cookie:
$session->setCookie('cookie name', null);


Headers handling is not supported by Driver\SahiDriver. Because there’s no way Sahi can get such information out of the browser.

HTTP Authentication

Also, Mink session has a special method to perform HTTP Basic authentication:

$session->setBasicAuth($user, $password);


Automatic HTTP authentication is not supported by SahiDriver. Because HTTP authentication in browser requires manual user action, that can’t be done remotely.

Resetting the Session

The primary aim for Mink is to provide a single consistent web browsing API for acceptance tests. But most important part in testing is isolation. We need a way to isolate our tests from each other. And Mink provides two very useful methods for you to use in your teardown() methods:

// soft-reset:

// hard-reset:

Both methods do exactly the same job for headless browsers - they clear browser’s cookies and history. The difference appears with Driver\SahiDriver:

  • $session->reset() will try to clean all available from browser side cookies. It’s very fast and doesn’t require the physical reload of the browser between tests, making them much faster. But it has a disadvantage - it clears only the cookies, available to clean from browser side. And we also have http-only cookies. In such case, resetting simply won’t work. Also, browsing history will state the same after this call. So, it’s very fast, but limited in complex cases.
  • $session->restart() will physically restart the browser. This action will physically clean all your cookies and browsing history by cost of browser reloading.

Taking all this into account, it would be the best way to use reset() by default and to call restart() in cases when we need really full isolation.

Sessions Manager

Although $session object is already usable enough, it’s not as easy to write multisession (multidriver/multibrowser) code. Yep, you’ve heard me right, with Mink you can manipulate multiple browser emulators simultaneously with a single consistent API:

// init sessions
$session1 = new \Behat\Mink\Session($driver1);
$session2 = new \Behat\Mink\Session($driver2);

// start sessions


Isn’t it cool? But Mink makes it even cooler:

$mink = new \Behat\Mink\Mink();
$mink->registerSession('goutte', $goutteSession);
$mink->registerSession('sahi', $sahiSession);

With such configuration, you can talk with your sessions by name through one single container object:



Mink will even lazy-start your sessions when needed (on first getSession() call). So, the browser will not be started till you really need it!

Or you could even omit the session name in default cases:


This call is possible thanks to $mink->setDefaultSessionName('goutte') setting previously. We’ve set the default session, that would be returned on getSession() call without arguments.


Mink class also provides an easy way to reset or restart your started sessions (and only started ones):

// reset started sessions

// restart started sessions

Traverse the Page - Selectors

Now you know how to control the browser itself. But what about traversing the current page content? Mink talks to its drivers with XPath selectors, but you also have access to named selectors and css selectors. Mink will transform such selectors into XPath queries internally for you.

The main class of Mink’s selectors engine is Behat\Mink\Selector\SelectorsHandler. It handles different selector types, which implements Behat\Mink\Selector\SelectorInterface:

$cssSelector = new \Behat\Mink\Selector\CssSelector();

// generate XPath query out of CSS:
echo $cssSelector->translateToXPath('h1 > a');

$handler = new \Behat\Mink\Selector\SelectorsHandler();
$handler->registerSelector('css', $cssSelector);

// generate XPath query out of CSS:
echo $handler->selectorToXpath('css', 'h1 > a');

When you initialize Selector\SelectorsHandler it already has XPath selectors, named selectors and css selectors registered in it.

You can provide a custom selectors handler as a second argument to your session instances:

$session = new \Behat\Mink\Session($driver,
    new \Behat\Mink\Selector\SelectorsHandler()

Mink will use this handler internally in find* methods.

Named Selectors

Named selectors provide a way to get named XPath queries:

$selector = new \Behat\Mink\Selector\NamedSelector();
$handler  = new \Behat\Mink\Selector\SelectorsHandler(array(
    'named' => $selector

// XPath query to find the fieldset:
$xpath1 = $selector->translateToXPath(
    array('fieldset', 'id|legend')
$xpath1 = $handler->selectorToXpath('named',
    array('fieldset', 'id|legend')

// XPath query to find the field:
$xpath2 = $selector->translateToXPath(
    array('field', 'id|name|value|label')
$xpath2 = $handler->selectorToXpath('named',
    array('field', 'id|name|value|label')

There’s whole lot more named selectors for you to use:

  • link - for searching a link by its href, id, title, img alt or value
  • button - for searching a button by its name, id, value, img alt or title
  • link_or_button - for searching for both, links and buttons
  • content - for searching a specific page content (text)
  • select - for searching a select field by its id, name or label
  • checkbox - for searching a checkbox by its id, name, or label
  • radio - for searching a radio button by its id, name, or label
  • file - for searching a file input by its id, name, or label
  • optgroup - for searching optgroup by its label
  • option - for searching an option by its content
  • table - for searching a table by its id or caption

CSS Selectors

With Selector\CssSelector, you can use CSS expressions to search page elements:

$selector = new \Behat\Mink\Selector\CssSelector();
$handler  = new \Behat\Mink\Selector\SelectorsHandler(array(
    'css' => $selector

// XPath query to find the link by ID:
$xpath1 = $selector->translateToXPath('a#ID');
$xpath1 = $handler->selectorToXpath('css', 'a#ID');

XPath Selectors

And of course, you can use clean XPath queries:

$xpath = $handler->selectorToXpath('xpath', '//html');

It’s like a proxy method, which will return the same expression you give to it. It’s used internally in find* methods.

find* Methods

So, now we know how to generate XPath queries for specific elements search. But how we actually make this search? The answer is find* methods, available on DocumentElement object. You can get this object from session:

$page = $session->getPage();
$page = $mink->getSession('sahi')->getPage();

This object provides two very useful traversing methods:

  • find() - evaluates specific selector on the page content and returns the last matched element or null:

    $fieldElement = $page->find('named',
        array('field', 'id|name|value|label')
    $elementByCss = $page->find('css', 'h3 > a');
  • findAll() - evaluates specific selector on the page content and returns an array of matched elements:

    $fieldElements = $page->findAll('named',
        array('field', 'id|name|value|label')
    $elementsByCss = $page->findAll('css', 'h3 > a');

Also, there’s a bunch of shortcut methods:

  • findById() - will search for an element by its ID
  • findLink() - will search for a link with link named selector
  • findButton() - will search for a button with button named selector
  • findField() - will search for a field with field named selector

Nested Traversing

Every find*() method will return Behat\Mink\Element\NodeElement instance and findAll() will return an array of such instances. The fun part is you can make same old traversing on such elements too:

$registerForm = $page->find('css', 'form.register');

// find some field INSIDE form with class="register"
$field = $registerForm->findField('id|name|value|label');

Manipulate the Page - NodeElement

Ok, you’ve got an interesting page element. Now you’ll want to do something with it. Behat\Mink\Element\NodeElement provides bunch a of useful methods for you:

$el = $page->find('css', '.something');

// get tag name:
echo $el->getTagName();

// check that element has href attribute:

// get element's href attribute:
echo $el->getAttribute('href');

Element Content and Text

To retrieve HTML content or plain text from out of the element, you can use:

$plainText = $el->getText();
$html = $el->getHtml();


getText() will strip tags and unprinted characters out of the response, including newlines. So it’ll basically return the text, that user sees on the page.

Form Field Manipulations

You can fill form fields/retrieve its values with form manipulation actions:

// check/uncheck checkbox:
if ($el->isChecked()) {

// select option in select:
$el->selectOption('optin value');

// attach file to file input:

// get input value:
echo $el->getValue();

// set intput value:
$el->setValue('some val');

// press the button:

Mouse Manipulations

You can perform mouse manipulations on an element:



All methods except click() are not supported by Driver\GoutteDriver, because there’s no way how it can perform them without actual browser window.


Mink even supports drag’n’drop of one field onto another:

$el1 = $page->find(...);
$el2 = $page->find(...);



Drag’n’drop is not supported by Driver\GoutteDriver, because there’s no way how it can perform this action without actual browser window.

Testing Tools Integration

Mink comes with PHPUnit, Behat and Symfony2 support out of the box.

Mink API

Find out all available functions in Mink API.