TL;DR: just execute
./run.py from inside the automation-tests directory.
If you're missing pip or virtualenv, it'll tell you what to do.
If you're missing test dependencies, it'll install them for you.
If all that is OK, it'll connect to dev.123done.org and try to create a fake user, login, and logout.
If you want to run that single test against your ephemeral instance called 'foo', just do
If you want to run all the tests, create a dummy user, put its info in credentials.yaml, then do
run.py --all to run all the tests, including 123done and myfavoritebeer tests.
If you want to run all the tests against all the browsers, using any browsers the script can find locally, then do
how to run selenium tests inside the automation-tests directory against ephemeral, stage, or prod environments
Node bindings aren't as mature as python for Selenium 2 API (webdriver), so we're using python bindings instead. This requires some python-centric setup, but it shouldn't take more than 15 minutes or so to get up and running, unless you're running Windows. See the bottom of this page for Windows setup instructions.
These tests currently only hit myfavoritebeers and 123done domains. For example, to test an ephemeral install named foo.personatest.org, you can pass 'foo.123done.org' into the py.test baseurl parameter (this is covered again in the examples section).
You should have python 2.7 on your system (check python --version).
We have to install a bunch of python libraries. pip fetches packages; virtualenv sandboxes them. If pip and virtualenv aren't on your system already, you'll need to do this once (once per computer, not once per repo):
# only do this if pip and virtualenv aren't on your computer already # might need to use sudo easy_install pip pip install virtualenv
From the automated-tests directory, create a sandboxed python environment to install python dependencies (only need to do this once per clone):
# only do this once per clone virtualenv bid_selenium
Be sure you do not accidentally add the virtualenv directory (here, bid_selenium) to git.
You can activate the sandbox, meaning link installed programs, via:
And when you want to stop using the sandbox, you can exit via
deactivate. Deactivating the virtualenv doesn't destroy it.
In order to install python dependencies into the sandbox, activate the virtualenv, then install the python requirements in requirements.txt:
pip install -Ur requirements.txt
Sweet. Your environment is now ready.
Some of the automation tests verify that existing accounts work, so create a test account, and put the info into credentials.yaml.
When you want to run the tests, make sure the virtualenv is active:
Then, run the tests by calling py.test on the command line with some options. Here is the most relevant documentation: command-line options added to py.test by the mozwebqa plugin, which is awesome. Here is the documentation for the upstream pytest project.
Use local Firefox to run the 123done tests (in the 123done directory) against dev.123done.org:
python -m py.test --destructive --credentials=credentials.yaml \ --baseurl=http://dev.123done.org \ --driver=firefox \ -q 123done
Use local Chrome (assuming you've downloaded Chromedriver to /usr/local/bin/chromedriver) to run just one of the the myfavoritebeer tests against myfavoritebeer.org:
python -m py.test --destructive --credentials=credentials.yaml \ --baseurl=http://www.myfavoritebeer.org \ --driver=chrome --chromepath=/usr/local/bin/chromedriver \ -q myfavoritebeer/tests/test_logout.py
Use Sauce Labs (assuming you've got credentials in saucelabs.yaml) to run IE 8 against an ephemeral instance called 'foo':
python -m py.test --destructive --credentials=credentials.yaml \ --baseurl=http://foo.123done.org \ --platform=XP --browsername="internet explorer" --browserver=8 \ --saucelabs=saucelabs.yaml \ -q 123done
note, your saucelabs.yaml file should be of the form:
# example sauce_labs.yaml config file username: <username> password: <password> api-key: <api-key>
The tests create a /results directory, which contains an index.html file with test results, screenshots, and videos if you used sauce labs. In case of a failure, you'll also see the backtrace. Totally sweet.
TODO: some idioms from the existing test code to help people quickly express "find this" and "click this" idiomatically.
Refer to mozilla's pytest_mozwebqa documentation on writing tests for the time being.
A note about upstreaming bidpom changes: this codebase contains mozilla's bidpom as git-subtree. This allows us to pull in changes from upstream, while easily tracking the bidpom code to branches. It's unlikely that we'll need to push or pull to upstream frequently, but for details on doing so, see also apenwarr's blog post.
Note: this post talks about python 2.5, but you need to install 2.6 or 2.7, and not 3.x.
Alternately, think about running under cygwin instead.