Writing and Running Barbican Client Tests¶
As a part of every code review that is submitted to the python-barbicanclient project there are a number of gating jobs which aid in the prevention of regression issues within python-barbicanclient. As a result, a python-barbicanclient developer should be familiar with running python-barbicanclient tests locally.
For your convenience we provide the ability to run all tests through
the tox
utility. If you are unfamiliar with tox please see
refer to the tox documentation for assistance.
Unit Tests¶
Currently, we provide tox environments for Python 2.7. By default
all available test environments within the tox configuration will execute
when calling tox
. If you want to run them independently, you can do so
with the following command:
# Executes tests on Python 2.7
tox -e py27
Note
If you do not have the appropriate Python versions available, consider
setting up PyEnv to install multiple versions of Python. See the
documentation regarding /setup/dev
for more information.
Note
Individual unit tests can also be run, using the following commands:
# runs a single test with the function named
# test_should_entity_str
tox -e py27 -- test_should_entity_str
# runs only tests in the WhenTestingSecrets class and
# the WhenTestingOrderManager class
tox -e py27 -- '(WhenTestingSecrets|WhenTestingOrderManager)'
The function name or class specified must be one located in the barbicanclient/tests directory.
Groups of tests can also be run with a regex match after the --
.
For more information on what can be done with testr
, please see:
http://testrepository.readthedocs.org/en/latest/MANUAL.html
You can also setup breakpoints in the unit tests. This can be done by
adding import pdb; pdb.set_trace()
to the line of the unit test you
want to examine, then running the following command:
# Executes tests on Python 2.7
tox -e debug
Note
For a list of pdb commands, please see: https://docs.python.org/2/library/pdb.html
Functional Tests¶
Unlike running unit tests, the functional tests require Barbican and
Keystone services to be running in order to execute. For more
information on setting up a Barbican development environment
and using Keystone with Barbican
,
see our accompanying project documentation.
A configuration file for functional tests must be edited before the tests
can be run. In the top-level directory of the python-barbicanclient, edit
/etc/functional_tests.conf
to the values you setup in Keystone.
[DEFAULT]
# Leaving this as a placeholder
[keymanager]
# Replace values that represent barbican server and user information
url=http://localhost:9311
username=barbican
password=secretservice
project_name=service
project_id=service
#max_payload_size=10000
project_domain_name=Default
[identity]
# Replace these with values that represent your identity configuration
uri=http://localhost:5000/v2.0
uri_v3=http://localhost:5000/v3
auth_version=v3
username=admin
tenant_name=admin
password=password
domain_name=Default
admin_username=admin
admin_tenant_name=admin
admin_password=password
admin_domain_name=Default
[identity-feature-enabled]
# Leaving this as a placeholder
Once you have the appropriate services running and configured you can execute the functional tests through tox.
# Execute Barbican Functional Tests
tox -e functional
By default, the functional tox job will use nosetests to execute the functional tests. This is primarily due to nose being a very well known and common workflow among developers.
Note
In order to run individual functional test functions, you must use the following commands:
# runs only tests in the test_secrets.py file
tox -e functional -- client/v1/functional/test_secrets.py
# runs only tests in the SecretsTestCase class
tox -e functional -- client/v1/functional/test_secrets.py:\
SecretsTestCase
# runs a single test with the function named
# test_secret_create_defaults_check_content_types
tox -e functional -- client/v1/functional/test_secrets.py:\
SecretsTestCase.test_secret_create_defaults_check_content_types
The path specified must be one located in the functionaltests directory.
Remote Debugging¶
In order to be able to hit break-points on API calls, you must use remote
debugging. This can be done by adding import rpdb; rpdb.set_trace()
to
the line of the API call you wish to test. For example, adding the breakpoint
in def create
in barbicanclient.secrets.py
will allow you to hit the
breakpoint whenever the create
function is called.
Note
After performing the POST
the application will freeze. In order to use
rpdb
, you must open up another terminal and run the following:
# enter rpdb using telnet
telnet localhost 4444
Once in rpdb, you can use the same commands as pdb, as seen here: https://docs.python.org/2/library/pdb.html