Model Mommy: Smart fixtures for better tests

Model-mommy offers you a smart way to create fixtures for testing in Django. With a simple and powerful API you can create many objects with a single line of code.

https://travis-ci.org/vandersonmota/model_mommy.png?branch=master

Install

pip install model_mommy

Basic usage

Let's say you have an app family with a model like this:

class Kid(models.Model):
    happy = models.BooleanField()
    name = models.CharField(max_length=30)
    age = models.IntegerField()
    bio = models.TextField()
    wanted_games_qtd = models.BigIntegerField()
    birthday = models.DateField()
    appointment = models.DateTimeField()

To create a persisted instance, just call Mommy:

from model_mommy import mommy
from family.models import Kid

kid = mommy.make(Kid)

No need to pass attributes every damn time.

Importing every model over and over again is boring. So let Mommy import them for you:

from model_mommy import mommy

# 1st form: app_label.model_name
kid = mommy.make('family.Kid')

# 2nd form: model_name
dog = mommy.make('Dog')

[1]	You can only use the 2nd form on unique model names. If you have an app family with a Dog, and an app farm with a Dog, you must use the app_label.model_name form.

[2]	model_name is case insensitive.

Model Relationships

Mommy also handles relationships. Say the kid has a dog:

class Dog(models.Model):
    owner = models.ForeignKey('Kid')

when you ask Mommy:

from model_mommy import mommy

rex = mommy.make('family.Dog')

She will also create the Kid, automagically.

Defining some attributes

Of course it's possible to explicitly set values for attributes.

from model_mommy import mommy

another_kid = mommy.make('family.Kid', age=3)

Related objects attributes are also reachable:

from model_mommy import mommy

bobs_dog = mommy.make('family.Dog', owner__name='Bob')

Non persistent objects

If don't need a persisted object, Mommy can handle this for you as well:

from model_mommy import mommy

kid = mommy.prepare('family.Kid')

It works like make, but it doesn't persist the instance.

More than one instance

If you need to create more than one instance of the model, you can use the _quantity parameter for it:

from model_mommy import mommy

kids = mommy.make('family.Kid', _quantity=3)
assert len(kids) == 3

It also works with prepare:

from model_mommy import mommy

kids = mommy.prepare('family.Kid', _quantity=3)
assert len(kids) == 3

How mommy behaves?

By default, model-mommy skips fields with null=True or blank=True. Also if a field has a default value, it will be used.

You can override this behavior by explicitly defining values.

When shouldn't you let mommy generate things for you?

If you have fields with special validation, you should set their values by yourself.

Model-mommy should handle fields that:

don't matter for the test you're writing;
don't require special validation (like unique, etc);
are required to create the object.

Currently supported fields

BooleanField, IntegerField, BigIntegerField, SmallIntegerField, PositiveIntegerField, PositiveSmallIntegerField, FloatField, DecimalField
CharField, TextField, SlugField, URLField, EmailField
ForeignKey, OneToOneField, ManyToManyField (even with through model)
DateField, DateTimeField, TimeField
FileField, ImageField

Custom fields

Model-mommy allows you to define generators methods for your custom fields or overrides its default generators. This could be achieved by specifing a dict on settings that its keys are the field paths and the values their generators functions, as the example bellow:

# on your settings.py file:
def gen_func():
    return 'value'

MOMMY_CUSTOM_FIELDS_GEN = {
    'test.generic.fields.CustomField': gen_func,
}

Recipes

If you're not comfortable with random data or even you just want to improve the semantics of the generated data, there's hope for you.

You can define a recipe, which is a set of rules to generate data for your models. Create a module called mommy_recipes.py at your app's root directory:

from model_mommy.recipe import Recipe
from family.models import Person

person = Recipe(Person,
    name = 'John Doe',
    nickname = 'joe',
    age = 18,
    birthday = date.today(),
    appointment = datetime.now()
)

Note you don't have to declare all the fields if you don't want to. Omitted fields will be generated automatically.

The variable person serves as the recipe name:

from model_mommy import mommy

mommy.make_recipe('family.person')

Or if you don't want a persisted instance:

from model_mommy import mommy

mommy.prepare_recipe('family.person')

You can use the _quantity parameter as well if you want to create more than one object from a single recipe.

You can define recipes locally to your module or test case as well. This can be useful for cases where a particular set of values may be unique to a particular test case, but used repeatedly there.

company_recipe = Recipe(Company, name='WidgetCo')

class EmployeeTest(TestCase):
    def setUp(self):
        self.employee_recipe = Recipe(
            Employee, name=seq('Employee '),
            company=company_recipe.make())

    def test_employee_list(self):
        self.employee_recipe.make(_quantity=3)
        # test stuff....

    def test_employee_tasks(self):
        employee1 = self.employee_recipe.make()
        task_recipe = Recipe(Task, employee=employee1)
        task_recipe.make(status='done')
        task_recipe.make(due_date=datetime(2014, 1, 1))
        # test stuff....

Recipes with foreign keys

You can define foreign_key relations:

from model_mommy.recipe import Recipe, foreign_key
from family.models import Person, Dog


person = Recipe(Person,
    name = 'John Doe',
    nickname = 'joe',
    age = 18,
    birthday = date.today(),
    appointment = datetime.now()
)

dog = Recipe(Dog,
    breed = 'Pug',
    owner = foreign_key(person)
)

Notice that person is a recipe.

You may be thinking: "I can put the Person model instance directly in the owner field". That's not recommended.

Using the foreign_key is important for 2 reasons:

Semantics. You'll know that attribute is a foreign key when you're reading;
The associated instance will be created only when you call make_recipe and not during recipe definition;

You can also use related, when you want two or more models to share the same parent:

from model_mommy.recipe import related, Recipe

dog = Recipe(Dog,
    breed = 'Pug',
)
other_dog = Recipe(Dog,
    breed = 'Boxer',
)
person_with_three_dogs = Recipe(Person,
    dog_set = related('dog', 'other_dog')
)

Note this will only work when calling make_recipe because the related manager requires the objects in the related_set to be persisted. That said, calling prepare_recipe the related_set will be empty.

Recipes with callables

It's possible to use callables as recipe's attribute value.

from datetime import date
from model_mommy.recipe import Recipe
from family.models import Person

person = Recipe(Person,
    birthday = date.today,
)

When you call make_recipe, Mommy will set the attribute to the value returned by the callable.

Recipes with iterators

You can also use iterators (including generators) to provide multiple values to a recipe.

from itertools import cycle

colors = ['red', 'green', 'blue', 'yellow']
person = Recipe(Person,
    favorite_color = cycle(colors)
)

Mommy will use the next value in the iterator every time you create a model from the recipe.

Sequences in recipes

Sometimes, you have a field with an unique value and using make can cause random errors. Also, passing an attribute value just to avoid uniqueness validation problems can be tedious. To solve this you can define a sequence with seq

from model_mommy.recipe import Recipe, seq
from family.models import Person

person = Recipe(Person,
    name = seq('Joe'),
    age = seq(15)
)

p = mommy.make_recipe('myapp.person')
p.name
>>> 'Joe1'
p.age
>>> 16

p = mommy.make_recipe('myapp.person')
p.name
>>> 'Joe2'
p.age
>>> 17

This will append a counter to strings to avoid uniqueness problems and it will sum the counter with numerical values.

You can also provide an optional increment_by argument which will modify incrementing behaviour. This can be an integer, float or Decimal.

person = Recipe(Person,
    age = seq(15, increment_by=3)
    height_ft = seq(5.5, increment_by=.25)
)

p = mommy.make_recipe('myapp.person')
p.age
>>> 18
p.height_ft
>>> 5.75

p = mommy.make_recipe('myapp.person')
p.age
>>> 21
p.height_ft
>>> 6.0

Overriding recipe definitions

Passing values when calling make_recipe or prepare_recipe will override the recipe rule.

from model_mommy import mommy

mommy.make_recipe('model_mommy.person', name='Peter Parker')

This is useful when you have to create multiple objects and you have some unique field, for instance.

Deprecation Warnings

Because of the changes of model_mommy's API, the following methods are deprecated and will be removed in one of the future releases:

mommy.make_one -> should use the method mommy.make instead

mommy.prepare_one -> should use the method mommy.prepare instead

mommy.make_many -> should use the method mommy.make with the _quantity parameter instead

mommy.make_many_from_recipe -> should use the method mommy.make_recipe with the _quantity parameter instead

Known Issues

django-taggit

Model-mommy identifies django-taggit's TaggableManager as a normal Django field, which can lead to errors:

TypeError: <class 'taggit.managers.TaggableManager'> is not supported by mommy.

The fix for this is to set blank=True on your TaggableManager.

Extensions

GeoDjango

Works with it? This project has some custom generators for it: https://github.com/sigma-consultoria/mommy_spatial_generators

Contributing

Prepare a virtual environment.

pip install virtualenvwrapper
mkvirtualenv --no-site-packages --distribute

Install the requirements.

pip install -r requirements.txt

Run the tests.

make test

Inspiration

Model-mommy was inspired by many great open source software like ruby's ObjectDaddy and FactoryGirl.

Doubts? Loved it? Hated it? Suggestions?

Join our mailing list for support, development and ideas!

https://groups.google.com/group/model-mommy

jsatt/model_mommy