Retrieve or generate random instances of Django models.
Switch branches/tags
Nothing to show
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.

Random Instances

This module exports a get_or_create_random function that improves Django's get_or_create on two aspects:

  • invoking get_or_create_random with parameters that match MULTIPLE instances does not raise an error, but rather returns one of those instances at random

  • invoking get_or_create_random with parameters that do not match ANY instance returns a NEW instance of that model (the same occurs with get_or_create). The improvement is that get_or_create_random can be invoked without passing a value for all the 'required' fields of the model. If these fields are not passed, they are automatically filled with random values (e.g.: CharFields are filled with random strings, ImageFields with random images).

The goal is to make prototyping faster, as model instances can be obtained and created by specifying just the minimum set of desired fields. This is useful when writing tests and can avoid having to write complex fixtures.


Download and install the package from GitHub, or do it the easy way:

pip install random_instances

Example usages

Say you have the following Subject model defined in the subjects application, and no instances of Subject in the database:

from django.db import models
class Subject(models.Model):
    word             = models.CharField(unique=True, max_length=255)
    is_current       = models.BooleanField(default=False)

To get or create a random subject, type:

from random_instances import get_or_create_random
from subjects.models import Subject

This will generate an instance of Subject with the following values:

  • word => a random string (no longer than 255 characters)
  • is_current => a random Boolean value (True or False)

To get or create a subject with a specific value for the word field, type:

get_or_create_random(Subject, word='Peace')

This will generate an instance of Subject with the following values:

  • word => 'Peace'
  • is_current => a random Boolean value (True or False)

To get a subject or create one with some default values, type:

get_or_create_random(Subject, word='Peace', defaults={'is_current': True})

This will get a Subject instance with word='Peace', if any exists. Otherwise, a new one will be created with the following values:

  • word => 'Peace'
  • is_current => True

In short, get_or_create_random creates a new instance whenever an existing appropriate one cannot be found. The fields of the newly created instance are filled according to these ordered rules:

  • if the field has been assigned a value in the function parameters, that value is used
  • if the field is required (blank=False, null=False), a random value is used
  • otherwise, the field is left unassigned

Testing and contributing

The module contains a suite of tests that can be executed by importing random_instances into the INSTALLED_APPS of a Django project and running:

python test random_instances

To contribute to the module and report bugs, add the appropriate tests, check that they pass, then send a pull request with the patch or the fixed bug.

To do

As of v0.0.2, add relationship fields (ForeignKey, ManyToMany, OneToOne) and add tests to check that all the fields parameters are respected (e.g., maximum length for CharField, number of digits for Decimals).


This project was partly inspired by Dilla.

How random_instances contributes to Test-Driven Development

After a couple of weeks using lettuce I found out that one of the steps I use most often is:

Given there is a Photographer 

(substitute Photographer for any Django model you have). For instance, one typical scenario is:

Given there is a Photographer 
And I navigate to his detail page 
Then I should see his pictures 

In this kind of story, I don't care which photographer the test will run against: any will (or should) work. So the most straightforward implementation of this step would be:


that either returns an existing photographer or creates a new one. The problem with [Django's get_or_create](http:// is that it will fail if the query returns multiple instances (in the previous case, if there are multiple photographers). What I need is a function that returns a random instance if any exists, otherwise creates a new one. random_instances serves this purpose.

Basic Usage

Using random_instances, the step:

Given there is a Photographer 

can be easily implemented as:

from random_instances import get_or_create_random 

Optional Fields

This function follows the prototype of [Django's get_or_create](http://, so you can pass desired and default values for the model's field. For instance, say that Photographer has a "name" CharField, an "is_famous" BooleanField and an "avatar" ImageField, then you can have the step:

Given there is a Photographer named "Helmut Newton" 

easily implemented as:

get_or_create_random(Photographer, name="Helmut Newton") 

which will either return that photographer (if existing) or create a new one. If you have this step as the Background of a Scenario (that is, executed before each scenario), then the instance will only be created the first time, and retrieved the successive ones, resulting in better performance (shorter time) for the test suite to complete.

Required fields

What happens if get_or_create_random is invoked without specifying a value for a required field? For instance, imagine that the Photographer model has an "is_famous" BooleanField that is required (blank=False, null=False) and does not have a default value. If your step states:

Given there is a Photographer named "Helmut Newton" 

then you do not really care about the value of "is_famous", so it would be a waste of time to specify one just because the field is required. The solution provided by get_or_create_random is quite intuitive: every field that is required and does not have a default value is automatically assign a random value, coherently with the field type (e.g., a random string for CharField, a random number for IntegerField, a random image for ImageField). For this reason, the implementation

get_or_create_random(Photographer, name="Helmut Newton") 

would still work, even without specifying a value for "is_famous", which would be randomly set to True or False. If you really care about this value, then you can always set it explicitly, e.g.:

Given there is a famous Photographer named "Helmut Newton" 

which would be implemented as:

get_or_create_random(Photographer, name="Helmut Newton", is_famous=True) 

Summing up

random_instances can be beneficial to lettuce users, although it has a broader extent (can be actually used inside the same Django application to generate random instances of models... anyone sees a nice usage for that?). It's a faster and more meaningful way to create fixtures, since you only specify the fields that you are interested in.