The run_tests.sh Script

Horizon ships with a script called run_tests.sh at the root of the repository. This script provides many crucial functions for the project, and also makes several otherwise complex tasks trivial for you as a developer.

First Run

If you start with a clean copy of the Horizon repository, the first thing you should do is to run ./run_tests.sh from the root of the repository. This will do two things for you:

  1. Set up a virtual environment for both the horizon module and the openstack_dashboard project using ./tools/install_venv.py.
  2. Run the tests for both horizon and openstack_dashboard using their respective environments and verify that everything is working.

Setting up the environment the first time can take several minutes, but only needs to be done once. If dependencies are added in the future, updating the environments will be necessary but not as time consuming.

I just want to run the tests!

Running the full set of unit tests quickly and easily is the main goal of this script. All you need to do is:

./run_tests.sh

Yep, that’s it. However, for a more thorough test run you can include the Selenium tests by using the --with-selenium flag:

./run_tests.sh --with-selenium

If you run horizon in a minimal installation VM, you will probably need the following (steps for Fedora 18 minimal installation):

  1. Install these packages in the VM: yum install xorg-x11-xauth xorg-x11-fonts-Type1.noarch.
  2. Install firefox in the VM: yum install firefox.
  3. Connect to the VM by ssh -X (if you run set|grep DISP, you should see that the DISPLAY is set).
  4. Run ./run_tests.sh --with-selenium.

Running a subset of tests

Instead of running all tests, you can specify an individual directory, file, class, or method that contains test code.

To run the tests in the horizon/test/tests/tables.py file:

./run_tests.sh horizon.test.tests.tables

To run the tests in the WorkflowsTests class in horizon/test/tests/workflows:

./run_tests.sh horizon.test.tests.workflows:WorkflowsTests

To run just the WorkflowsTests.test_workflow_view test method:

./run_tests.sh horizon.test.tests.workflows:WorkflowsTests.test_workflow_view

Running the integration tests

The Horizon integration tests treat Horizon as a black box, and similar to Tempest must be run against an existing OpenStack system. These tests are not run by default.

  1. Update the configuration file openstack_dashboard/test/integration_tests/horizon.conf as required (the format is similar to the Tempest configuration file).

  2. Run the tests with the following command:

    $ ./run_tests.sh --integration
    

Like for the unit tests, you can choose to only run a subset.

$ ./run_tests.sh --integration openstack_dashboard.test.integration_tests.tests.test_login

Using Dashboard and Panel Templates

Horizon has a set of convenient management commands for creating new dashboards and panels based on basic templates.

Dashboards

To create a new dashboard, run the following:

./run_tests.sh -m startdash <dash_name>

This will create a directory with the given dashboard name, a dashboard.py module with the basic dashboard code filled in, and various other common “boilerplate” code.

Available options:

  • --target: the directory in which the dashboard files should be created. Default: A new directory within the current directory.

Panels

To create a new panel, run the following:

./run_tests -m startpanel <panel_name>

This will create a directory with the given panel name, and panel.py module with the basic panel code filled in, and various other common “boilerplate” code.

Available options:

  • -d, --dashboard: The dotted python path to your dashboard app (the module which contains the dashboard.py file.). If not specified, the target dashboard should be specified in a pluggable settings file for the panel.
  • --target: the directory in which the panel files should be created. If the value is auto the panel will be created as a new directory inside the dashboard module’s directory structure. Default: A new directory within the current directory.

JavaScript Tests

You can also run JavaScript unit tests using Karma. Karma is a test environment that allows for multiple test runners and reporters, including such features as code coverage. Karma allows developer to run tests live, as it can watch source and test files for changes.

The default configuration also performs coverage reports, which are saved to horizon/coverage-karma/ and openstack_dashboard/coverage-karma/.

To run the Karma tests for Horizon and Dashboard using the run_tests.sh script:

./run_tests.sh --karma

To run the Karma tests for Horizon and Dashboard using npm:

npm install # You only need to execute this once.
npm test

Note

These two methods are equivalent. The former merely executes the latter.

JavaScript Code Style Checks

You can run the JavaScript code style checks, or linting, using eslint. ESLint is a permissively licensed, sophisticated language parser and linter that confirms both our style guidelines, and checks the code for common errors that may create unexpected behavior.

To run eslint for Horizon and Dashboard using the run_tests.sh script:

./run_tests.sh --eslint

To run eslint for Horizon and Dashboard using npm:

npm install # You only need to execute this once.
npm run lint

Note

These two methods are equivalent. The former merely executes the latter.

Give me metrics!

You can generate various reports and metrics using command line arguments to run_tests.sh.

ESLint

To run ESLint, a JavaScript code style checker:

./run_tests.sh --eslint

Coverage

To run coverage reports:

./run_tests.sh --coverage

The reports are saved to ./reports/ and ./coverage.xml.

PEP8

You can check for PEP8 violations as well:

./run_tests.sh --pep8

The results are saved to ./pep8.txt.

PyLint

For more detailed code analysis you can run:

./run_tests.sh --pylint

The output will be saved in ./pylint.txt.

Tab Characters

For those who dislike having a mix of tab characters and spaces for indentation there’s a command to check for that in Python, CSS, JavaScript and HTML files:

./run_tests.sh --tabs

This will output a total “tab count” and a list of the offending files.

Running the development server

As an added bonus, you can run Django’s development server directly from the root of the repository with run_tests.sh like so:

./run_tests.sh --runserver

This is effectively just an alias for:

./tools/with_venv.sh ./manage.py runserver

Generating the documentation

You can build Horizon’s documentation automatically by running:

./run_tests.sh --docs

The output is stored in ./doc/build/html/.

Updating the translation files

You can update all of the translation files for both the horizon app and openstack_dashboard project with a single command:

./run_tests.sh --makemessages

or, more compactly:

./run_tests.sh --m

Starting clean

If you ever want to start clean with a new environment for Horizon, you can run:

./run_tests.sh --force

That will blow away the existing environments and create new ones for you.

Non-interactive Mode

There is an optional flag which will run the script in a non-interactive (and eventually less verbose) mode:

./run_tests.sh --quiet

This will automatically take the default action for actions which would normally prompt for user input such as installing/updating the environment.

Environment Backups

To speed up the process of doing clean checkouts, running continuous integration tests, etc. there are options for backing up the current environment and restoring from a backup:

./run_tests.sh --restore-environment
./run_tests.sh --backup-environment

The environment backup is stored in /tmp/.horizon_environment/.

Environment Versioning

Horizon keeps track of changes to the environment by comparing the current requirements files (requirements.txt and test-requirements.txt) and the files last time the virtual environment was created or updated. If there is any difference, the virtual environment will be update automatically when you run run_tests.sh.