mutants
allows to create Python objects that mutate on access.
It works by creating proxy objects
that change underlying objects on every access.
import random
import mutants
n = mutants.OnAccessMutant(0, lambda n: n + 1)
print(n) # prints 1
print(n) # prints 2
print(n) # prints 3
class Duck:
feathers = True
def quack(self):
print('quack')
class Wolf:
teeth = 'sharp'
def quack(self):
print('no quack')
def random_animal():
return random.choice([Duck(), Wolf()])
randy = mutants.ImmutableMutant(random_animal)
randy.quack() # prints 'quack' or 'no quack'
randy.quack() # prints 'quack' or 'no quack'
print(hasattr(randy, 'feathers')) # prints 'True' or 'False'
randy.name = 'Randy'
print(hasattr(randy, 'name')) # prints 'False', see below
def class_toggler(animal):
if isinstance(animal, Duck):
return Wolf
return Duck
tracy = mutants.ClassHopperMutant(Duck(), class_toggler)
tracy.quack() # prints 'no quack' as it's a Wolf
tracy.quack() # prints 'quack' as it's a Duck
print(tracy.teeth) # prints 'sharp' as it's a Wolf
tracy.name = 'Tracy'
print(tracy.name) # prints 'Tracy'
def class_extender(animal):
class SleepyAnimal(animal.__class__):
def quack(self):
super().quack()
print('zzz')
return SleepyAnimal
zetta = mutants.ClassHopperMutant(Duck(), class_extender)
zetta.quack() # prints 'quack' and 'zzz'
Depending on what you want, you can choose one of two mutant kinds: ImmutableMutant and ClassyMutant or make a custom OnAccessMutant.
OnAccessMutant
is the core class of the library.
It wraps an object much like wrapt.ObjectProxy
does.
But there's a callback that is called before each access
and has the ability to modify or replace the proxied object.
Usage: OnAccessMutant(initial_object, callable_mutator)
where: callable_mutator(wrapped_object) -> new_wrapped_object
ImmutableMutant
can impersonate different objects.
Its constructor takes a callable.
Before each access, this callable is called to provide an object
that ImmutableMutant
will impersonate.
Usage: ImmutableMutant(callable_returning_objects_to_be_proxied)
Modifying ImmutableMutant
is probably a strange idea,
because it doesn't remember the objects that it impersonates
and the callable will probably return something else next time.
In this pure Python implementation it's implemented as:
def ImmutableMutant(mutator):
return OnAccessMutant(None, lambda _: mutator())
Future C extensions may implement it separately for performance benefits.
ClassHopper
reevaluates the class of the wrapped object on every access.
It's like obj.__class__ = callable_returning_a_class()
,
but magically happening before every manipulation with the object.
Usage: mutant.ClassHopper(initial_object, callable_returning_a_class)
In this pure Python implementation it's implemented as:
def ClassHopperMutant(initial_object, class_returning_callable, copy=True):
from copy import copy as _copy
def class_mutator(obj):
obj.__class__ = class_returning_callable(obj)
return obj
if copy:
initial_object = _copy(initial_object)
return OnAccessMutant(initial_object, class_mutator)
Future C extensions may implement it separately for performance benefits.
mutants
were born to serve the needs of another library, hacks
,
that aids modifying object, function or class behaviour,
stacking such modifications
and switching `currently active modification stacks' easily on the fly.
Check it out: https://github.com/t184256/hacks
mutants
is similar to wrapt.ObjectProxy
or lazy-object-proxy
but with bugs and flexibility instead of laziness, caching and performance.
CPython/Python guys, please give us something cleaner to pull off our tricks!
mutants
is currently in alpha state,
so send in pull requests if something is broken!
mutants
is distributed under the terms of the MIT License;
see LICENSE.txt.