#unittests

Apache Maven – How to skip the unit test execution

In Apache Maven, we can pass the parameter -Dmaven.test.skip=true to skip the entire unit test while executing Maven goals such as mvn package. By default, Maven will execute the unit tests automatically and if any unit tests fail then the build process also fails. Following are some scenarios where you might want to skip failures or skip unit test execution:1) Pipeline has a specific stage for…

Test Automation fin-test.net ⚙ Test Strategy and Management 💻 mob/desk📱

Testing types fin-test.net

Test automation is a significant investment in any software testing project. Cost and benefits should be weighed before automation is started as it takes time to develop. As a matter of fact, test automation does not differ from other software development activities. Automation requires an initial investment of time, training, and licenses. There is also ongoing maintenance required, and even application changes to accommodate automation. We help our clients assess automation needs and once a decision is made to proceed, our automation specialists choose the best tools depending on automation objectives and the system under test. Automation takes test cases as input scenarios and is often integrated into a continuous integration process.

Data-driven testing

Keyword-driven Testing

Hybrid Testing

Model-based Testing

Behavior-driven Testing

List of UI Testing Tools

List of Web Testing Tools

Continuous Testing

In this post we'll be introducing a few Python related tools that should aid you in day to day development. Anyone using Python 2.7 will benefit from the below.

Gevent is a library for working with Greenlets, which allows python to use async I/O calls. With network heavy applications (databases, apis) gevent can help execute multiple calls in parallel. It's the same model as used by Node.js and nginx, but in general does not expose callbacks to the end user. As a result, gevent'ed code looks very similar to Python code.

Nose is a test runner. It has a great library of available plugins. One of those is Coverage, a great tool written by Ned Batchelder. Unfortunately it doesn't support gevent out of the box. Fortunately, there's a fork which adds gevent support. You can install the gevent enabled version with the following:

pip install git+https://github.com/newbrough/coverage.git

Run your tests. We use a custom test runner, cleverly named bin/nose, which handles some setup (DB tables, elastic search mappings). Nose has an option to generate XML coverage files, but we need the .coverage file, and it seems the only way to generate seems to be to use the coverage utility. We only care about our own modules, which are all under the top level shift namespace.

coverage run --include 'shift/*' bin/nose

Coverage data is saved to .coverage

Great, now we've got our test coverage, ideally we have a way to visualize it. Duvet is a nice tool for exploring your codebase in a color coded manner to quickly understand where your test coverage is lacking.

pip install duvet

Open duvet. It will give you a nice color coded file browser on the left. The code window on the right indicates code paths which are unreachable.

duvet

Here's a screenshot from the pybee site.

Statt den bisherigen funcargs kann man in py.test ja jetzt sogenannte Fixtures benutzen. Hier mal ein kleines Beispiel aus conftest.py, wo eine solche angelegt wird:

# content of conftest.py import pytest import pymongo @pytest.fixture def db(request): return pymongo.Connection().test

Hier wird also eine Datenbank-Connection aufgebaut.

Scopes

Man kann dabei auch angeben, welchen Scope das Ergebnis haben soll, ob diese Funktion also pro Modul, pro Test oder pro Klasse ausgeführt oder aber wiederverwendet werden soll. Dazu gibt man einfach einen der folgenden Parameter an:

@pytest.fixtures(scope="module")

Weitere Werte wären "function" (default), "cls" oder "session". Man sollte dabei natürlich Seiteneffekte beachten.

Finalizers

In dem obigen Beispiel wäre zudem noch schön, wenn die Datenbank nach jedem Test auch wieder gelöscht oder zumindest auf ein Standard-Setup zurückgeführt würde. Wir können das natürlich einfach in die obige Funktion reinschreiben, schöner jedoch wäre eine Funktion, die das nach dem Lauf des Tests macht.  Das geht wie folgt:

@pytest.fixture def db(request): db = pymongo.Connection().test def fin(): db.users.remove() request.addfinalizer(fin) return db

Der Finalizer wird genau dann aufgerufen, wenn der Scope verlassen wird, für den diese Fixture gilt.

  Nutzung von Fixtures

Genutzt werden die Fixtures ähnlich wie die funcargs zuvor. Man gibt sie in seiner Testfunktion einfach an:

def test_something(db): db.users.insert({_id: 1}) assert db.users.find_one()['_id'] == 1

Auch andere Fixtures können das benutzen:

@pytest.fixture def app(request, db): return Application(db)

Die von einer Fixture benötigten anderen Fixtures werden also einfach nach dem Request-Paramater aufgelistet. Mehr Info