A simple library that allows you to add database fixtures for your unit tests using nothing but JSON or YAML.

Installing Flask-Fixtures is simple, just do a typical pip install like so:

pip install flask-fixtures

If you are going to use JSON as your data serialization format, you
should also consider installing the dateutil package since it will
add much more powerful and flexible parsing of dates and times.

To install the library from source simply download the source code, or check it out if you have git installed on your system, then just run the install command.

git clone https://github.com/croach/Flask-Fixtures.git
cd /path/to/flask-fixtures
python setup.py install

To setup the library, you simply need to tell Flask-Fixtures where it can find the fixtures files for your tests. Fixtures can reside anywhere on the file system, but by default, Flask-Fixtures looks for these files in a directory called fixtures in your app's root directory. To add more directories to the list to be searched, just add an attribute called FIXTURES_DIRS to your app's config object. This attribute should be a list of strings, where each string is a path to a fixtures directory. Absolute paths are added as is, but reltative paths will be relative to your app's root directory.

Once you have configured the extension, you can begin adding fixtures for your tests.

To add a set of fixtures, you simply add any number of JSON or YAML files describing the individual fixtures to be added to your test database into one of the directories you specified in the FIXTURES_DIRS attribute, or into the default fixtures directory. As an example, I'm going to assume we have a Flask application with the following directory structure.

/myapp
    __init__.py
    config.py
    models.py
    /fixtures
        authors.json

The __init__.py file will be responsible for creating our Flask application object.

# myapp/__init__.py

from flask import Flask

app = Flask(__name__)

The config.py object holds our test configuration file.

# myapp/config.py

class TestConfig(object):
    SQLALCHEMY_DATABASE_URI = 'sqlite://'
    testing = True
    debug = True

And, finally, inside of the models.py files we have the following database models.

# myapp/models.py

from flask.ext.sqlalchemy import SQLAlchemy

from myapp import app

db = SQLAlchemy(app)

class Author(db.Model):
    id = db.Column(db.Integer, primary_key=True)
    first_name = db.Column(db.String(30))
    last_name = db.Column(db.String(30))

class Book(db.Model):
    id = db.Column(db.Integer, primary_key=True)
    title = db.Column(db.String(200))
    author_id = db.Column(db.Integer, db.ForeignKey('author.id'))
    author = db.relationship('Author', backref='books')

Given the model classes above, if we wanted to mock up some data for our database, we could do so in single file, or we could even split our fixtures into multiple files each corresponding to a single model class. For this simple example, we'll go with one file that we'll call authors.json.

A fixtures file contains a list of objects. Each object contains a key called records that holds another list of objects each representing either a row in a table, or an instance of a model. If you wish to work with tables, you'll need to specify the name of the table with the table key. If you'd prefer to work with models, specify the fully-qualified class name of the model using the model key. Once you've specified the table or model you want to work with, you'll need to specify the data associated with each table row, or model instance. Each object in the records list will hold the data for a single row or model. The example below is the JSON for a single author record and a few books associated with that author. Create a file called myapp/fixtures/authors.json and copy and paste the fixtures JSON below into that file.

[
    {
        "table": "author",
        "records": [{
            "id": 1,
            "first_name": "William",
            "last_name": "Gibson",
        }]
    },
    {
        "model": "myapp.models.Book",
        "records": [{
            "title": "Neuromancer",
            "author_id": 1
        },
        {
            "title": "Count Zero",
            "author_id": 1
        },
        {
            "title": "Mona Lisa Overdrive",
            "author_id": 1
        }]
    }
]

Another option, if you have PyYAML installed, is to write your fixtures using the YAML syntax instead of JSON. Personally, I prefer to use YAML; I find its syntax is easier to read, and I find the ability to add comments to my fixtures to be invaluable.

If you'd prefer to use YAML, I've added a version of the authors.json file written in YAML below. Just copy and paste it into a file called myapp/fixtures/authors.yaml in place of creating the JSON file above.

- table: author
  records:
    - id: 1
      first_name: William
      last_name: Gibson

- model: myapp.models.Book
  records:
    - title: Neuromancer
      author_id: 1
      published_date: 1984-07-01
    - title: Count Zero
      author_id: 1
      published_date: 1986-03-01
    - title: Neuromancer
      author_id: 1
      published_date: 1988-10-01

After reading over the previous section, you might be asking yourself why the library supports two methods for adding records to the database. There are a few good reasons for supporting both tables and models when creating fixtures. Using tables is faster, since we can take advantage of SQLAlchemy's bulk insert to add several records at once. However, to do so, you must first make sure that the records list is homegenous. In other words, every object in the ``records`` list must have the same set of key/value pairs, otherwise the bulk insert will not work. Using models, however, allows you to have a heterogenous list of record objects.

The other reason you may want to use models instead of tables is that you'll be able to take advantage of any python-level defaults, checks, etc. that you have setup on the model. Using a table, bypasses the model completely and inserts the data directly into the database, which means you'll need to think on a lower level when creating table-based fixtures.

To use Flask-Fixtures in your unit tests, you'll need to make sure your test class inherits from FixturesMixin and that you've specified a list of fixtures files to load. The sample code below shows how to do each these steps.

First, make sure the app that you're testing is initialized with the proper configuration. Then import and initialize the FixturesMixin class, create a new test class, and inherit from FixturesMixin. Now you just need to tell Flask-Fixtures which fixtures files to use for your tests. You can do so by setting the fixtures class variable. Doing so will setup and tear down fixtures between each test. To persist fixtures across tests, i.e., to setup fixtures only when the class is first created and tear them down after all tests have finished executing, you'll need to set the persist_fixtures variable to True. The fixtures variable should be set to a list of strings, each of which is the name of a fixtures file to load. Flask-Fixtures will then search the default fixtures directory followed by each directory in the FIXTURES_DIRS config variable, in order, for a file matching each name in the list and load each into the test database.

# myapp/fixtures/test_fixtures.py

import unittest

from myapp import app
from myapp.models import db, Book, Author

from flask.ext.fixtures import FixturesMixin

# Configure the app with the testing configuration
app.config.from_object('myapp.config.TestConfig')

# Initialize the Flask-Fixtures mixin class
FixturesMixin.init_app(app, db)

# Make sure to inherit from the FixturesMixin class
class TestFoo(unittest.TestCase, FixturesMixin):

    # Specify the fixtures file(s) you want to load.
    # Change the list below to ['authors.yaml'] if you created your fixtures
    # file using YAML instead of JSON.
    fixtures = ['authors.json']

    # Your tests go here

    def test_authors(self):
        authors = Author.query.all()
        assert len(authors) == Author.query.count() == 1
        assert len(authors[0].books) == 3

    def test_books(self):
        books = Book.query.all()
        assert len(books) == Book.query.count() == 3
        gibson = Author.query.filter(Author.last_name=='Gibson').one()
        for book in books:
            assert book.author == gibson

To see the library in action, you can find a simple Flask application and set of unit tests matching the ones in the example above in the tests/myapp directory. To run these examples yourself, just follow the directions below for "Contributing to Flask-Fixtures".

Currently, Flask-Fixtures supports python versions 2.6 and 2.7 and the py.test, nose, and unittest (included in the python standard library) libraries. To contribute bug fixes and features to Flask-Fixtures, you'll need to make sure that any code you contribute does not break any of the existing unit tests in any of these environments.

To run unit tests in all six of the supported environments, I suggest you install tox and simply run the tox command. If, however, you insist on running things by hand, you'll need to create a virtualenv for both python 2.6 and python 2.7. Then, install nose and py.test in each virtualenv. Finally, you can run the tests with the commands in the table below.