Behat

Vortex uses Behat for Behavior-Driven Development (BDD) testing. This page is for the developers who write and run those tests. After reading it, you can run any feature or tag, use the shipped steps, find the screenshots, and tell how continuous integration (CI) splits and retries your scenarios. For unit, kernel, and functional tests of PHP code, see PHPUnit.

You describe how the site should behave in human-readable stories, written in the Gherkin language, and Behat runs every step against the provisioned site. These are end-to-end tests, so spend them on your critical user journeys, such as logging in or resetting a password.

Vortex includes a pre-configured behat.php with profiles for Drupal projects, and a browser container running Chrome, which you can watch through a VNC viewer while the tests run.

Usage

Behat needs a provisioned Drupal site, so run it inside the container environment with the following commands:

Ahoy

# Run all Behat tests
ahoy test-bdd
# Run specific feature file
ahoy test-bdd tests/behat/features/homepage.feature
# Run scenarios with specific tag
ahoy test-bdd -- --tags=@smoke

Docker Compose

# Run all Behat tests. The `-d memory_limit=-1` flag lifts the PHP
# memory limit for long runs, matching what `ahoy test-bdd` does.
docker compose exec cli php -d memory_limit=-1 vendor/bin/behat
# Run specific feature file
docker compose exec cli php -d memory_limit=-1 vendor/bin/behat \
  tests/behat/features/homepage.feature
# Run scenarios with specific tag
docker compose exec cli php -d memory_limit=-1 vendor/bin/behat \
  --tags=@smoke

The -- before Behat's options is optional: Ahoy drops it and passes the rest of the line to Behat. See Ahoy > Usage.

Discovering available step definitions

Behat Steps, Mink, and Behat Screenshot each bring their own steps. To list every step definition your suite can use, run the following command:

Ahoy

# Generate step definitions reference
ahoy test-bdd -- --definitions=l

Docker Compose

# Generate step definitions reference
docker compose exec cli php -d memory_limit=-1 vendor/bin/behat \
  --definitions=l

Writing tests

Put each feature file in the tests/behat/features directory, where the default suite finds it, and structure it like the following example:

@tag_describing_feature_group
Feature: Short feature description

  As a site owner
  I want to make sure that my feature does what it is supposed to do
  So that I can be sure that my site is working as expected

  Scenario: Anonymous user uses a feature
    Given I go to the homepage
    Then the path should be "<front>"
    And I save screenshot

  @javascript
  Scenario: Anonymous user uses a feature with AJAX
    Given I go to the homepage
    When I click on the element ".button"
    Then I save screenshot

Untagged scenarios run through an HTTP client that doesn't execute JavaScript. To run a scenario in a real browser, tag it @javascript: it then runs in Chrome in the chrome container, driven through Selenium. Each @javascript scenario drives a real browser and runs slower than an untagged one, so add the tag only when a scenario needs JavaScript.

Skipping tests

To exclude a feature or scenario from the test run, add the @skipped tag to it. Every profile in behat.php filters it out.

Boilerplate features

Vortex provides the following boilerplate features:

• Core user journeys: homepage, login, password reset, and search
• The shipped module and service integrations: Redis, ClamAV, redirects, robots.txt, XML sitemap, and accessibility
• behat.feature, which checks that Behat itself is configured correctly
• The demo features counter.feature and pages.feature, tagged @demo

The installer removes the features for the services and modules you don't select. The homepage, login, and Behat configuration features are tagged @smoke.

These boilerplate tests run in the continuous integration pipeline from the moment you install Vortex, and they make a good starting point for your own.

Email

To capture outgoing messages instead of delivering them, tag a scenario @email. Then assert on their recipients and content with the steps from EmailTrait. The password reset feature shows the pattern.

To capture the messages that a module sends through its own mail interface, tag the scenario @email:TYPE, where TYPE is the interface's key in the system.mail configuration. A scenario with only typed tags collects only those interfaces, so add @email:default to collect the default interface too.

In the ci environment, settings.system.php makes the same mail collector the default backend for every message sent through interface.default. CI therefore stores those messages instead of delivering them, even when a scenario isn't tagged. The collector is a settings override, not stored configuration, so nothing running inside the site can switch it off. Reroute Email is disabled in ci, so each stored message keeps its intended recipient instead of the rerouting address.

The collector covers interface.default, the only interface the shipped configuration defines. A site that routes a module through its own interface (interface.<module> or interface.<module>_<key>, as Symfony Mailer and Mail System do) bypasses the collector for those messages. For that case, docker-compose.yml also points SSMTP_MAILHUB at a closed port in CI, so delivery fails at the transport no matter which mail plugin produced the message.

The bare When the test email system queue is cleared step still needs the @email tag or an earlier When the test email system is enabled step. Each tagged scenario also clears the queue as it starts, so messages from an earlier scenario never reach your assertions.

Project conventions

Your project's own Behat conventions live in its docs/testing.md file, which your project team maintains to record the testing practices you agree on. It covers the following topics:

• User story format
• Standard user types
• Test data
• An example feature file
• The process for testing a content type

Vortex ships the file in the project's docs/ directory, and the installer keeps it unless you decline its Preserve project documentation? prompt. Declining it removes the whole docs/ directory.

Configuration

See the Behat user guide and its configuration reference.

All global configuration takes place in the behat.php file, written in PHP, the only configuration format that Behat 4 reads.

By default, Behat runs all tests defined in the tests/behat/features directory.

To add or remove test targets, change the suite's paths:

behat.php

$suite = (new Suite('default'))
  ->withPaths('%paths.base%/tests/behat/features', '%paths.base%/web/modules/custom/mymodule/tests/behat/features')

Chrome session flags

The behat.php configuration passes the following flags to Chrome through goog:chromeOptions, so the tests run the same way every time in the container environment:

Flag	Purpose
--disable-extensions	Prevents interference from browser extensions
--disable-popup-blocking	Allows tests to open popups without being blocked
--disable-translate	Prevents the translation bar from appearing on non-English pages
--force-prefers-reduced-motion	Disables CSS animations and transitions for test stability
--test-type	Suppresses error dialogs and crash recovery prompts
--window-size=1920,1080	Sets a deterministic viewport size for consistent screenshots

--window-size only sets the size Chrome opens at. The Behat Screenshot context resizes the window to 1440 by 900 pixels at the start of every @javascript scenario.

Contexts and extensions

The configuration uses the following contexts and extensions:

• Behat Steps: a library of reusable Behat steps for Drupal. It also supplies the Drupal integration layer, registered as the DrevOps\BehatSteps\Extension\ServiceContainer\BehatStepsExtension extension with the Drupal API, Drush, and black-box backends, the theme's region map, and the message and login selectors.
• Mink Extension: the browser sessions and the base browser vocabulary, registered as the Behat\MinkExtension\ServiceContainer\MinkExtension extension and the Behat\MinkExtension\Context\MinkContext context. Untagged scenarios use a BrowserKit session over plain HTTP, and @javascript scenarios use the selenium2 session, which drives Chrome at http://chrome:4444/wd/hub.
• Behat Screenshot: extension and step definitions to create HTML and image screenshots on demand or when tests fail, registered as the DrevOps\BehatScreenshotExtension\ServiceContainer\BehatScreenshotExtension extension and the DrevOps\BehatScreenshotExtension\Context\ScreenshotContext context.

FeatureContext

The FeatureContext.php file is your project's own Behat context. The default suite loads it, and it's where your custom step definitions and hooks go. Behat 4 reads them from PHP attributes, such as #[Given] and #[BeforeScenario], and doesn't read docblock annotations such as @Given.

FeatureContext extends DrevOps\BehatSteps\DrupalContext from the Behat Steps package. That context already carries every step the package provides, so the file holds only the project's own step definitions and hooks. Don't also add the package's WebContext to the suite: DrupalContext already includes its steps, and registering both is a fatal error, because every web step would be defined twice.

Reports

Behat writes test run reports in JUnit format to the .logs/test_results/behat directory.

Screenshots

Behat Screenshot saves a screenshot whenever a step fails, and whenever a step such as I save screenshot runs. Each capture is an HTML file of the page, headed by its URL, feature, step, and time, plus a full-page PNG image for @javascript scenarios. The HTTP client behind untagged scenarios can't take images, so those get the HTML file only. To capture a screenshot after every passed step of a scenario, tag the scenario or its feature @screenshots.

Screenshots land in the .logs/screenshots directory by default, and screenshots from earlier runs stay there, because behat.php sets purge to FALSE. To change either behavior, set the matching environment variable inside the cli container, for example with docker compose exec -e as the CI step does. ahoy test-bdd doesn't pass these variables through from your host:

• BEHAT_SCREENSHOT_DIR stores the screenshots in another directory.
• BEHAT_SCREENSHOT_PURGE set to 1 empties the directory at the start of a run.

For example, the following command empties the directory, then runs Behat:

docker compose exec -e BEHAT_SCREENSHOT_PURGE=1 cli \
  php -d memory_limit=-1 vendor/bin/behat

Animated screenshots

Behat Screenshot combines the per-step screenshots into an animated GIF for each scenario, so a failing test is easier to follow. behat.php turns this on under the animation key of the BehatScreenshotExtension extension, with 500 milliseconds between frames (frame_delay). The frames are the PNG images, so only @javascript scenarios produce a GIF.

Animation captures a screenshot after every passed step, and with always_fullscreen on, each one is a full-page capture. Its cost grows with the number of steps and roughly doubles the wall time of a run. To turn animation off everywhere, set animation.enabled to FALSE in behat.php. To exempt a single scenario or feature, tag it @screenshots:animated:skip.

With animation off, the @screenshots:animated tag opts a single scenario or feature back in. A tag on a scenario overrides a tag on its feature. See Artifacts for how the continuous integration pipeline handles animation.

Output format

By default, Vortex uses Behat's built-in progress formatter with its inlineFailures option turned on in behat.php. It prints a dot for each passed step and the details of each failed, pending, or undefined step as soon as it happens. Without the option, those details wait for the summary at the end of the run. The test run continues after a failure while the output stays minimal.

Observing the browser

For @javascript tests, Behat drives Chrome through Selenium in the chrome container. That container also runs a VNC server with noVNC, so you can watch the browser from a browser on your host machine. To open it, follow the steps for your setup:

Ahoy

1. Print the project information:

ahoy info

2. Open the URL on the Selenium VNC URL on host line in a browser on your host machine. noVNC connects with the password secret.

Docker Compose

1. Print the host port that maps to the container's noVNC port, 7900:

docker compose port chrome 7900

2. Replace <port> with the port number after the colon in the output, and open http://localhost:<port>/?autoconnect=1&password=secret in a browser on your host machine.

Confirm that noVNC shows the Chrome window: run ahoy test-bdd tests/behat/features/homepage.feature, and watch Chrome open the site's homepage during the feature's @javascript scenario.

To work with the page yourself in the middle of a test, add an And I wait for 60 seconds step to the scenario. The step pauses the test for a minute, so you can use the browser in the container through noVNC on your host machine.

Continuous integration

In the continuous integration pipeline, Behat runs with --strict, so an undefined or pending step fails the run there. ahoy test-bdd doesn't pass --strict, so to match CI locally, run ahoy test-bdd -- --strict.

Profiles

Without a --profile option, Behat uses the default profile defined in behat.php: it runs every scenario except those tagged @skipped.

When the pipeline has more than 1 runner, Behat tests run in parallel across the runners, so the suite finishes sooner. Tags mark which runner a feature or scenario lands on, and the CI step picks the profile named after the runner index, starting from p0. Vortex supports any number of parallel runners but ships only 2 parallel profiles, p0 and p1, to match the 2 runners the pipeline starts with:

• p0 (the first runner) is the catch-all: it runs every scenario not tagged @p1, plus all @smoke scenarios, and skips anything tagged @skipped.
• p1 (the second runner) runs scenarios tagged @p1, plus all @smoke scenarios, and skips anything tagged @skipped.

In practice, leave a feature untagged to run it on the first runner. Tag it @p1 to move it to the second runner, or tag it @smoke to run it on every runner. A feature you forget to tag still runs on the first runner.

In the following example, with 1 runner, Behat runs both scenarios. With 2 runners, the first scenario runs on the first runner (untagged scenarios stay there) and the second scenario runs on the second runner:

@homepage
Feature: Homepage

  Ensure that homepage is displayed as expected.

  Scenario: Anonymous user visits homepage
    Given I go to the homepage
    Then the path should be "<front>"
    And I save screenshot

  @javascript @p1
  Scenario: Anonymous user visits homepage
    Given I go to the homepage
    Then the path should be "<front>"
    And I save screenshot

To add more p* profiles, copy the existing p1 profile in your behat.php, change its name and tag filter, and register it in the returned Config. Then exclude the new tag from the p0 catch-all, or its scenarios run on the first runner as well. See Adding more containers for the full steps.

Profile numbers count from the first runner that runs Behat. The CI configuration subtracts VORTEX_CI_BEHAT_PROFILE_OFFSET (0 by default) from the runner index to pick the profile. As a result, dedicating leading runners to other tools requires no profile renumbering or feature re-tagging. See Test parallelism for the container mechanics.

To force a profile, set VORTEX_CI_BEHAT_PROFILE in the continuous integration pipeline configuration (a repository variable in GitHub Actions, an environment variable in CircleCI). Every runner then uses that profile.

If the pipeline has only 1 runner and VORTEX_CI_BEHAT_PROFILE is unset, the default profile is used and all tests run there except those tagged @skipped. An explicitly set VORTEX_CI_BEHAT_PROFILE stays active even on a single runner.

Artifacts

The continuous integration pipeline stores the test results and screenshots as artifacts, and the JUnit reports track test performance and stability. In GitHub Actions, each runner uploads its .logs directory as the test-artifacts-<runner index> artifact, which you download from the run's Summary page. Unless VORTEX_CI_TEST_RESULTS_SKIP is 1, the JUnit reports also appear among the run's checks as Test results (<runner index>). In CircleCI, the screenshots are in the job's Artifacts tab and the JUnit reports are in its test results.

Vortex passes BEHAT_SCREENSHOT_ANIMATION_SKIP=1 into the Behat step in continuous integration to skip animation there, while keeping it available for local runs where a recording is worth the wait. The variable overrides both behat.php and the animation tags.

Flaky tests

A Behat scenario can fail now and then without a real bug behind it, on a slow page load or a timing-sensitive step. How often depends on the size of your site and the resources of the continuous integration runner.

The CI step therefore retries once: when the first run fails, it runs Behat again with --rerun, which repeats only the scenarios that failed. The step fails only when a scenario fails in both runs. If the first run failed without recording a failed scenario, --rerun runs the whole profile again.

A scenario that passes only on the second run is still flaky, so check the first run's output in the job log.

Ignoring failures

Set VORTEX_CI_BEHAT_IGNORE_FAILURE to 1 to let the build pass despite failing scenarios. Behat still runs, retries as described in Flaky tests, and reports every failure in the job log. See Ignore tool failures.