Hypothesis, randomised testing for Django

Rae Knowler @raeknowler CKAN, Symfony, Django slides - https://speakerdeck.com/bellisk/hypothesis-randomised-testing-for-django

@drmaciver - hypothesis author @sinister_katze - fuzzy, loves catching bugs

• slides -
• http://hypothesis.readthedocs.org
• https://github.com/bellisk/hypothesis-django-example
• Property based testing library
• inspired by Haskell’s Quickcheck

Why is it so great?

• lots of randomised data
  • say name is string, email, number
  • come up with 2000 customers!
  • if fail, work down to simplest string with bug (eg ! mark)
• can deal with complex data types - eg Django Models
• works with pytest, unittest - just a library

hypothesis-django

• tests look similar to Django tests
• custom field types
• generate child models
• can generate fixtures for Django (new!)

pip install hypothesis hypothesis-django

Example

• StockSubscription model
  • fields: symbol, last notified, NotificationsPerDay
  • methods - __str__, next_notification_time
• tests - use hypothesis version of models, TestCase, and given
• given decorator - tells hypothesis type of fields, in combination with hypothesis.strategies.integers
• hypothesis finds a falsifying example (0 values -> divide by 0)

Example 2

• Portfolio model
• want to find median value
• test generates list of StockSubscription, max len 100
• test asserts median value is from 0 to 100
• finds bug - if 2 in list, this causes error in median function

Hypothesis is awesome

• it makes your tests way simpler - hypothesis will find the edge cases
• it finds subtle bugs
• simple tests that are insanely effective