pytest-setup

The pytest-setup plugin provides a way to mark your tests with data required to run the tests. It moves the data creation out of the actual test-function in an effort to keep the test clean and specific.

There are a lot of moving parts, but hopefully this user guide will help you get up and running.

It also provides a simple test database to manage all the created data.

Getting started

In your project you'll need a directory that contains the different artifacts that you'll be creating. This directory needs to be a python package, ie. it needs to contain an __init__.py file. Also, that __init__-file needs to contain all the imports to the individual representations.

from .user import User
from .project import Project

In our examples we'll call the directory representations.

The relative path to representations needs to go into a pytest INI-style file. For details refer to the pytest documentation.

[pytest]
representation_path = <from project root relative path to>/representations

The pytest.ini file will also need the name of your class that all other representations will be based on. If you only have the one then give the name of that class.

Next we need to create a class for the artifact we want to create using our marker. Let's create a user artifact.

Requirements

There are a couple of requirements for the plugin to work.

SIGNATURE

This is a dictionary containing the parameter-names of the create-function. The names are the keys and the type of the name are the values. For example our User has a name, this is basestring (that will handle both unicode and str).

Note: SIGNATURE is a keyword and so must be in all caps.

Property identifier

Each class/artifact needs to provide a property called identifier, this is used to get data from the test database and is also used by the database to make sure there are no duplicates.

create()

This is the create-function of our user-artifact. It has to be a @classmethod and has to be called create. Furthermore it has to return an instantiated object of the artifact.

Example:

import .representations.project as project

class User(object):

    def __init__(self, user_name):
        self._user_name = user_name

    @property
    def identifier(self):
        return self._user_name

    SIGNATURE = {'name': basestring, 'project': project.Project}

    @classmethod
    def create(cls, name, project):
        # Whatever code is needed to setup the user and add him to the project
        return cls(name)

Usage

When the User artifact is in place you can now use the marker to create as many users as you want. There is currently two ways of creating objects: on the module level and on the function level.

Module level

Objects created on the module level will be available for every test function in the module. The syntax is:

module_setup_data = [{'Class Name1': [{'key': 'value', 'key': 'value'},  # Class Name1 obj one
                                      {'key': 'value', 'key': 'value'}]},  # Class Name2 obj one
                     {'Class Name2': [{'key': 'value'}]}]

This will create two objects of type Class Name1 and one object of type Class Name2.

Function level

Objects created on the function level will be available for only that decorated test function. The syntax is:

@pytest.mark.setup_data({'Class Name1': [{'key': 'value', 'key': 'value'},  # Class Name1 obj one
                                         {'key': 'value', 'key': 'value'}]},  # Class Name2 obj one
                        {'Class Name2': [{'key': 'value'}]})

This will create two objects of type Class Name1 and one object of type Class Name2.

Example of using both:

import pytest

module_setup_data = [{'Project': [{'name': 'MyProject'}]}]

@pytest.mark.setup_data({'User': [{'name': 'Tom Jones', 'project': 'MyProject'}]})
def test_login(test_db):
    login_user(user=test_db.get('User', 'Tom Jones',
               password='111111',
               project=test_db.get('Project', 'MyProject'))

Test Database

To easily manage all the created data the plugin provides a simple key-value database.

The database is accessed via the pytest fixture test_db.

def test_login(test_db):
    user = test_db.get('User', 'Tommy')

The first argument to get is the object as a string you want to get, the second argument is the identifier mentioned earlier.

If you so wish, you can also add to the database. The first argument is the object you wish to add and the second, optional, argument is the time-to-live (ttl) which can be either "function" or "module".

The ttl corresponds to the scope of the setup-data. For "function" that data is only available during the scope of that decorated test function. For "module" it's available for all the test functions within that module.

Advanced Usage

In some cases an object you create in turn creates its own objects. If you want those objects available in the test database, you need to provide the creating object with a default _representations property.

Let's say we have an account object that also creates a user object.

from .representations.user import User

class Account(object):

    def __init__(self, account_name):
        self._account_name = account_name
        self._user = User.create("Account Owner")

    @property
    def identifier(self):
        return self._account_name

    @property
    def default_representations
        return [self._user]

The default_representations property has to return a list with the object(s) it creates.