django-dbdiff 0.9.7

Database state assertion

A nice way to test a data import script is to create a source data fixture with a subset of data, ie. with only 10 cities instead of 28K or only 3 european parliament representatives instead of 3600, feed the import function with that and then compare the database state with a django fixture. This looks like what I was used to do:

• use such a command to create a small data extract shuf -n3 cities15000.txt > cities_light/tests/cities_test_fixture.txt,

• use it against the import script on a clean database,

• verify the database manually, and run django-admin dumpdata –indent=4 cities_light > cities_light/tests/cities_test_expected.txt

• then, make a test case that calls the import script against the fixture,

• write and maintain some funny (fuzzy ?) repetitive test code to ensure that the database is in the expected state.

When a bug is fixed, just add the case to the fixture and repeat the process to create new expected data dumps, use coverage to ensure no case is missed.

With django-dbdiff, I just need to maintain to initial data extract, and test it with Fixture('appname/path/to/fixture', models=[YourModelToTest]).assertNoDiff() in a django.test.TransactionTestCase which has reset_sequences=True:

• if the fixture in question doesn’t exist, it’ll be automatically created on with dumpdata for the concerned models on the first run, raising “FixtureCreated” exception to fail the test and inform of the path of the created fixture, so that it doesn’t mislead the user in thinking the test passed with an existing fixture,

• if the fixture exists, it’ll run dumpdata on the models concerned and GNU diff it against the fixture, if there’s any output it’ll be raised in the “DiffFound” exception, failing the test and printing the diff.

Usage

Example:

from django.test import TransactionTestCase
from dbdiff.fixture import Fixture


class YourImportTest(TransactionTestCase):
    reset_sequences = True

    def test_your_import(self):
        your_import()

        Fixture('yourapp/tests/yourtest.json',
                models=[YourModel]).assertNoDiff()

The first time, it will raise a FixtureCreated exception, and the test will fail. This is to inform the user that the test didn’t really run. On the next run though, it will pass.

If any difference is found between the database and the test fixture, then diff() will return the diff as outputed by GNU diff.

If you need to ignore fields globally, set the class-level variable exclude as such:

Fixture.exclude = {'mrsrequest.mrsrequest': ['token']}

If your import produces records with non-deterministic primary keys (e.g. UUIDs or sequence gaps), pass ignore_pk=True to match records by their field content instead of by pk:

Fixture('yourapp/tests/yourtest.json',
        models=[YourModel],
        ignore_pk=True).assertNoDiff()

Instead of deleting the fixtures manually before running the tests to regenerate them, just run your tests with FIXTURE_REWRITE=1 environment variable. This will overwrite the fixtures and make the tests look like it passed.

See tests and docstrings for crunchy details.

Requirements

MySQL, SQLite and PostgreSQL, Python 3.10 to 3.14 are supported along with Django 4.2, 5.2, and 6.0 - it’s always better to support django’s master so that we can upgrade easily when it is released, which is one of the selling points for having 100% coverage.

Install

Install django-dbdiff with pip and add dbdiff to INSTALLED_APPS.

Django model observer

It is interresting to note that a related, perhaps sort-of similar app exists: https://github.com/Griffosx/djmo