Demystifying the Patch Function

Speaker

Lisa Roach

Material

video

Note

• Patch: temporarily replace your target with MagicMock()
• Target: 'package.module.ClassName'
• When should you mock?
  • When you don't want to actually call an object
    • e.g., database, system call
• What makes it so confusing?
  1. Identifying the target
  2. Multiple ways to call

Patch

• Target must be importable
• Patch where the object is used

# module.py
from db import db_write

# test_module.py
@patch("module.db_write")

• How to call Patch
  • Context Manager
  • Function / Class Decorators
  • Manual start/stop

Spec

• The spec arguments: spec, autospec, spec_set
• Mock Problems that spec can solve
  1. Misspelled asserts look like attributes
  2. Mocked objects called incorrectly pass silently
• A MagicMock
  • dir()
    • assert_any_call
    • assert_called
    • assert_called_once
    • assert_called_once_with
    • assert_called_with
    • ...
    • return_value
    • side_effect
• A Specced MagicMock
  • spec=True
    • The attribute of patched object is now in MagicMock.
    • We can now only get attribute that exists.
    • It does not know the attribute of your attribute → autospec=True
      • However, autospec can be dangerous. It will run the code that will be triggered when introspected.
    • It doesn't know dynamically created attribute (including the ones in __init__) → manually create it
  • spec_set=True prevent you from setting attribute that does not exist

Other arguments

• new_callable
  • Common Use case: Mocking property object
• new
• create
• kwargs
  • return_value
  • set attribute

Readmore

• patch.object
• patch.dict
• patch.multiple

Share on: Twitter ❄ Facebook ❄ Email