In python, instances of classes can have custom attributes set dynamically:
class A(object):
pass
a=A()
a.x=1
This, however, doesn't work if you instantiate object
object().x=1
#raises AttributeError: 'object' object has no attribute 'x'
object has no __dict__ attribute to store the custom attributes, so this doesn't work.
What's the easiest way to create a object where we can dynamically create custom attributes?
Seems that your example with creating an "empty" class is the fastest solution.
Nevertheless I wouldn't recommend doing this. When you access the attribute 'x' before assignment, you'll get an AttributeError. In other places inside your application you will not be able to determine how to handle objects, because you can't use type-checking to determine the object's features (attributes etc.). You'll always have to use hasattr() etc.
If you still want an object accepting any attribute names, consider overriding __getattr__ and __setattr__ in your class. See this python doc: http://docs.python.org/2/reference/datamodel.html#object.__getattr__
__getattr__ would be a bit of overkill, since I just need to make sure two attributes exist, and I create them immediately after creating the mock object
class A(object): pass approach. Or you create a class creating the attributes you need in its __init__().
A trick I've seen is:
x = lambda: magic
x.foo = 1
x.bar = 2
Which takes advantage of the fact functions have a __dict__
The way you propose in your question is generally fine (and often used) for mocking. However, if you'd doing it a lot, you might want to look at the mock library. In Python3, its part of the stdlib (import unittest.mock)
a=A()approach?