I've started to use the python descriptor protocol more extensively in the code I've been writing. Typically, the default python lookup magic is what I want to happen, but sometimes I'm finding I want to get the descriptor object itself instead the results of its __get__ method. Wanting to know the type of the descriptor, or access state stored in the descriptor, or somesuch thing.
I wrote the code below to walk the namespaces in what I believe is the correct ordering, and return the attribute raw regardless of whether it is a descriptor or not. I'm surprised though that I can't find a built-in function or something in the standard library to do this -- I figure it has to be there and I just haven't noticed it or googled for the right search term.
Is there functionality somewhere in the python distribution that already does this (or something similar)?
Thanks!
from inspect import isdatadescriptor
def namespaces(obj):
obj_dict = None
if hasattr(obj, '__dict__'):
obj_dict = object.__getattribute__(obj, '__dict__')
obj_class = type(obj)
return obj_dict, [t.__dict__ for t in obj_class.__mro__]
def getattr_raw(obj, name):
# get an attribute in the same resolution order one would normally,
# but do not call __get__ on the attribute even if it has one
obj_dict, class_dicts = namespaces(obj)
# look for a data descriptor in class hierarchy; it takes priority over
# the obj's dict if it exists
for d in class_dicts:
if name in d and isdatadescriptor(d[name]):
return d[name]
# look for the attribute in the object's dictionary
if obj_dict and name in obj_dict:
return obj_dict[name]
# look for the attribute anywhere in the class hierarchy
for d in class_dicts:
if name in d:
return d[name]
raise AttributeError
Edit Wed, Oct 28, 2009.
Denis's answer gave me a convention to use in my descriptor classes to get the descriptor objects themselves. But, I had an entire class hierarchy of descriptor classes, and I didn't want to begin every __get__ function with a boilerplate
def __get__(self, instance, instance_type):
if instance is None:
return self
...
To avoid this, I made the root of the descriptor class tree inherit from the following:
def decorate_get(original_get):
def decorated_get(self, instance, instance_type):
if instance is None:
return self
return original_get(self, instance, instance_type)
return decorated_get
class InstanceOnlyDescriptor(object):
"""All __get__ functions are automatically wrapped with a decorator which
causes them to only be applied to instances. If __get__ is called on a
class, the decorator returns the descriptor itself, and the decorated
__get__ is not called.
"""
class __metaclass__(type):
def __new__(cls, name, bases, attrs):
if '__get__' in attrs:
attrs['__get__'] = decorate_get(attrs['__get__'])
return type.__new__(cls, name, bases, attrs)
Most descriptors do their job when accessed as instance attribute only. So it's convenient to return itself when it's accessed for class:
class FixedValueProperty(object):
def __init__(self, value):
self.value = value
def __get__(self, inst, cls):
if inst is None:
return self
return self.value
This allows you to get descriptor itself:
>>> class C(object):
... prop = FixedValueProperty('abc')
...
>>> o = C()
>>> o.prop
'abc'
>>> C.prop
<__main__.FixedValueProperty object at 0xb7eb290c>
>>> C.prop.value
'abc'
>>> type(o).prop.value
'abc'
Note, that this works for (most?) built-in descriptors too:
>>> class C(object):
... #property
... def prop(self):
... return 'abc'
...
>>> C.prop
<property object at 0xb7eb0b6c>
>>> C.prop.fget
<function prop at 0xb7ea36f4>
Accessing descriptor could be useful when you need to extent it in subclass, but there is a better way to do this.
The inspect library provides a function to retrieve an attribute without any descriptor magic: inspect.getattr_static.
Documentation: https://docs.python.org/3/library/inspect.html#fetching-attributes-statically
(This is an old question, but I keep coming across it when trying to remember how to do this, so I'm posting this answer so I can find it again!)
The above method
class FixedValueProperty(object):
def __init__(self, value):
self.value = value
def __get__(self, inst, cls):
if inst is None:
return self
return self.value
Is a great method whenever you control the code of the property, but there are some cases, such as when the property is part of a library controlled by someone else, where another approach is useful. This alternative approach can also be useful in other situations such as implementing object mapping, walking a name-space as described in the question, or other specialised libraries.
Consider a class with a simple property:
class ClassWithProp:
#property
def value(self):
return 3
>>>test=ClassWithProp()
>>>test.value
3
>>>test.__class__.__dict__.['value']
<property object at 0x00000216A39D0778>
When accessed from the container objects class dict, the 'descriptor magic' is bypassed. Note also that if we assign the property to a new class variable, it behaves just like the original with 'descriptor magic', but if assigned to an instance variable, the property behaves as any normal object and also bypasses 'descriptor magic'.
>>> test.__class__.classvar = test.__class__.__dict__['value']
>>> test.classvar
3
>>> test.instvar = test.__class__.__dict__['value']
>>> test.instvar
<property object at 0x00000216A39D0778>
Let's say we want to get the descriptor for obj.prop where type(obj) is C.
C.prop usually works because the descriptor usually returns itself when accessed via C (i.e., bound to C). But C.prop may trigger a descriptor in its metaclass. If prop were not present in obj, obj.prop would raise AttributeError while C.prop might not. So it's better to use inspect.getattr_static(obj, 'prop').
If you are not satisfied with that, here's a CPython-specific method (from _PyObject_GenericGetAttrWithDict in Objects/object.c):
import ctypes, _ctypes
_PyType_Lookup = ctypes.pythonapi._PyType_Lookup
_PyType_Lookup.argtypes = (ctypes.py_object, ctypes.py_object)
_PyType_Lookup.restype = ctypes.c_void_p
def type_lookup(ty, name):
"""look for a name through the MRO of a type."""
if not isinstance(ty, type):
raise TypeError('ty must be a type')
result = _PyType_Lookup(ty, name)
if result is None:
raise AttributeError(name)
return _ctypes.PyObj_FromPtr(result)
type_lookup(type(obj), 'prop') returns the descriptor in the same way when CPython uses it at obj.prop if obj is a usual object (not class, for example).
Related
Often in Python it is helpful to make use of duck typing, for instance, imagine I have an object spam, whose prompt attribute controls the prompt text in my application. Normally, I would say something like:
spam.prompt = "fixed"
for a fixed prompt. However, a dynamic prompt can also be achived - while I can't change the spam class to use a function as the prompt, thanks to duck typing, because the userlying spam object calls str, I can create a dynamic prompt like so:
class MyPrompt:
def __str__( self ):
return eggs.get_user_name() + ">"
spam.prompt = MyPrompt()
This principal could be extended to make any attribute dynamic, for instance:
class MyEnabled:
def __bool__( self ):
return eggs.is_logged_in()
spam.enabled = MyEnabled()
Sometimes though, it would be more succinct to have this inline, i.e.
spam.prompt = lambda: eggs.get_user_name() + ">"
spam.enabled = eggs.is_logged_in
These of course don't work, because neither the __str__ of the lambda or the __bool__ of the function return the actual value of the call.
I feel like a solution for this should be simple, am I missing something, or do I need to wrap my function in a class every time?
What you want are computed attributes. Python's support for computed attributes is the descriptor protocol, which has a generic implementation as the builtin property type.
Now the trick is that, as documented (cf link above), descriptors only work when they are class attributes. Your code snippet is incomplete as it doesn't contains the definition of the spam object but I assume it's a class instance, so you cannot just do spam.something = property(...) - as the descriptor protocol wouldn't then be invoked on property().
The solution here is the good old "strategy" design pattern: use properties (or custom descriptors, but if you only have a couple of such attributes the builtin property will work just fine) that delegates to a "strategy" function:
def default_prompt_strategy(obj):
return "fixed"
def default_enabled_strategy(obj):
return False
class Spam(object):
def __init__(self, prompt_strategy=default_prompt_strategy, enabled_strategy=default_enabled_strategy):
self.prompt = prompt_strategy
self.enabled = enabled_strategy
#property
def prompt(self):
return self._prompt_strategy(self)
#prompt.setter
def prompt(self, value):
if not callable(value):
raise TypeError("PromptStrategy must be a callable")
self._prompt_strategy = value
#property
def enabled(self):
return self._enabled_strategy(self)
#enabled.setter
def enabled(self, value):
if not callable(value):
raise TypeError("EnabledtStrategy must be a callable")
self._enabled_strategy = value
class Eggs(object):
def is_logged_in(self):
return True
def get_user_name(self):
return "DeadParrot"
eggs = Eggs()
spam = Spam(enabled_strategy=lambda obj: eggs.is_logged_in())
spam.prompt = lambda obj: "{}>".format(eggs.get_user_name())
Consider this small example:
import datetime as dt
class Timed(object):
def __init__(self, f):
self.func = f
def __call__(self, *args, **kwargs):
start = dt.datetime.now()
ret = self.func(*args, **kwargs)
time = dt.datetime.now() - start
ret["time"] = time
return ret
class Test(object):
def __init__(self):
super(Test, self).__init__()
#Timed
def decorated(self, *args, **kwargs):
print(self)
print(args)
print(kwargs)
return dict()
def call_deco(self):
self.decorated("Hello", world="World")
if __name__ == "__main__":
t = Test()
ret = t.call_deco()
which prints
Hello
()
{'world': 'World'}
Why is the self parameter (which should be the Test obj instance) not passed as first argument to the decorated function decorated?
If I do it manually, like :
def call_deco(self):
self.decorated(self, "Hello", world="World")
it works as expected. But if I must know in advance if a function is decorated or not, it defeats the whole purpose of decorators. What is the pattern to go here, or do I misunderstood something?
tl;dr
You can fix this problem by making the Timed class a descriptor and returning a partially applied function from __get__ which applies the Test object as one of the arguments, like this
class Timed(object):
def __init__(self, f):
self.func = f
def __call__(self, *args, **kwargs):
print(self)
start = dt.datetime.now()
ret = self.func(*args, **kwargs)
time = dt.datetime.now() - start
ret["time"] = time
return ret
def __get__(self, instance, owner):
from functools import partial
return partial(self.__call__, instance)
The actual problem
Quoting Python documentation for decorator,
The decorator syntax is merely syntactic sugar, the following two function definitions are semantically equivalent:
def f(...):
...
f = staticmethod(f)
#staticmethod
def f(...):
...
So, when you say,
#Timed
def decorated(self, *args, **kwargs):
it is actually
decorated = Timed(decorated)
only the function object is passed to the Timed, the object to which it is actually bound is not passed on along with it. So, when you invoke it like this
ret = self.func(*args, **kwargs)
self.func will refer to the unbound function object and it is invoked with Hello as the first argument. That is why self prints as Hello.
How can I fix this?
Since you have no reference to the Test instance in the Timed, the only way to do this would be to convert Timed as a descriptor class. Quoting the documentation, Invoking descriptors section,
In general, a descriptor is an object attribute with “binding behavior”, one whose attribute access has been overridden by methods in the descriptor protocol: __get__(), __set__(), and __delete__(). If any of those methods are defined for an object, it is said to be a descriptor.
The default behavior for attribute access is to get, set, or delete the attribute from an object’s dictionary. For instance, a.x has a lookup chain starting with a.__dict__['x'], then type(a).__dict__['x'], and continuing through the base classes of type(a) excluding metaclasses.
However, if the looked-up value is an object defining one of the descriptor methods, then Python may override the default behavior and invoke the descriptor method instead.
We can make Timed a descriptor, by simply defining a method like this
def __get__(self, instance, owner):
...
Here, self refers to the Timed object itself, instance refers to the actual object on which the attribute lookup is happening and owner refers to the class corresponding to the instance.
Now, when __call__ is invoked on Timed, the __get__ method will be invoked. Now, somehow, we need to pass the first argument as the instance of Test class (even before Hello). So, we create another partially applied function, whose first parameter will be the Test instance, like this
def __get__(self, instance, owner):
from functools import partial
return partial(self.__call__, instance)
Now, self.__call__ is a bound method (bound to Timed instance) and the second parameter to partial is the first argument to the self.__call__ call.
So, all these effectively translate like this
t.call_deco()
self.decorated("Hello", world="World")
Now self.decorated is actually Timed(decorated) (this will be referred as TimedObject from now on) object. Whenever we access it, the __get__ method defined in it will be invoked and it returns a partial function. You can confirm that like this
def call_deco(self):
print(self.decorated)
self.decorated("Hello", world="World")
would print
<functools.partial object at 0x7fecbc59ad60>
...
So,
self.decorated("Hello", world="World")
gets translated to
Timed.__get__(TimedObject, <Test obj>, Test.__class__)("Hello", world="World")
Since we return a partial function,
partial(TimedObject.__call__, <Test obj>)("Hello", world="World"))
which is actually
TimedObject.__call__(<Test obj>, 'Hello', world="World")
So, <Test obj> also becomes a part of *args, and when self.func is invoked, the first argument will be the <Test obj>.
You first have to understand how function become methods and how self is "automagically" injected.
Once you know that, the "problem" is obvious: you are decorating the decorated function with a Timed instance - IOW, Test.decorated is a Timed instance, not a function instance - and your Timed class does not mimick the function type's implementation of the descriptor protocol. What you want looks like this:
import types
class Timed(object):
def __init__(self, f):
self.func = f
def __call__(self, *args, **kwargs):
start = dt.datetime.now()
ret = self.func(*args, **kwargs)
time = dt.datetime.now() - start
ret["time"] = time
return ret
def __get__(self, instance, cls):
return types.MethodType(self, instance, cls)
The IPython source code includes a getattr check that tests for the existence of '_ipython_canary_method_should_not_exist_' at the beginning of the get_real_method function:
def get_real_method(obj, name):
"""Like getattr, but with a few extra sanity checks:
- If obj is a class, ignore everything except class methods
- Check if obj is a proxy that claims to have all attributes
- Catch attribute access failing with any exception
- Check that the attribute is a callable object
Returns the method or None.
"""
try:
canary = getattr(obj, '_ipython_canary_method_should_not_exist_', None)
except Exception:
return None
if canary is not None:
# It claimed to have an attribute it should never have
return None
And although it's easy enough to find other coders special-casing this name, it's a harder to find any meaningful explanation of why.
Given these two classes:
from __future__ import print_function
class Parrot(object):
def __getattr__(self, attr):
print(attr)
return lambda *a, **kw: print(attr, a, kw)
class DeadParrot(object):
def __getattr__(self, attr):
print(attr)
if attr == '_ipython_canary_method_should_not_exist_':
raise AttributeError(attr)
return lambda *a, **kw: print(attr, a, kw)
It seems that IPython is using the existence or lack of this method to decide whether to use repr or one of its rich display methods. Intentionally thwarting the test in DeadParrot causes IPython to look up and invoke _repr_mimebundle_.
I'm writing an object that pretends all attrs exist. How do I decide whether to special-case this?
This question already has answers here:
How to use __setattr__ correctly, avoiding infinite recursion
(5 answers)
Closed 7 years ago.
I have this:
class MySession:
def __init__(self, session):
session['my-data'] = {} # my data is here
self._session = session
def __getattr__(self, name):
return self._session['my-data'][name]
def __setattr__(self, name, value):
my_data = self._session['my-data']
my_data[name] = value
self._session['my-data'] = my_data
obj = MySession({})
obj.x = 3
Basically I want to encapsulate access to the session (sub-)dictionary with an object attribute access. But I can not do it, since this causes infinite recursion, I guess because doing this:
self._session = session
calls setattr, which in turn calls getattr, which in turn calls getattr, etc
How can I pre-initialize some (normal) attributes in a class implementing getattr / setattr?
The __getattr__ method is only called for attributes that don't exist in the normal attribute dictionary. __setattr__ however is called unconditionally (it's mirror is really __getattribute__ rather than __getattr__). If you can get your _session attribute set up properly in __init__, you won't need to worry about anything in the other methods.
To add an attribute without running into any recursion, use super(MySession, self).__setattr__ to call the version of the method you inherited from object (you should always inherit from object in Python 2, to make your class a new-style class, in Python 3, it's the default). You could also call object.__setattr__ directly, but using super is better if you ever end up using multiple inheritance.
class MySession(object):
def __init__(self, session):
session['my-data'] = {}
super(MySession, self).__setattr__("_session", session) # avoid our __setattr__
def __getattr__(self, name):
return self._session['my-data'][name] # this doesn't recurse if _session exists
def __setattr__(self, name, value):
my_data = self._session['my-data']
my_data[name] = value
self._session['my-data'] = my_data
You could initialize first and change the setters getters later on:
def __init__(self, session):
session['my-data'] = {}
self._session = session
self.__setattr__ = self._setattr
self.__getattr__ = self._getattr
assuming self._setattr and self._getattr are implemented of course :)
Recently, I faced a problem which was similar to this question:
Accessing the class that owns a decorated method from the decorator
My rep was not high enough to comment there, so I am starting a new question to address some improvements to the answer to that problem.
This is what I needed:
def original_decorator(func):
# need to access class here
# for eg, to append the func itself to class variable "a", to register func
# or say, append func's default arg values to class variable "a"
return func
class A(object):
a=[]
#classmethod
#original_decorator
def some_method(self,a=5):
''' hello'''
print "Calling some_method"
#original_decorator
def some_method_2(self):
''' hello again'''
print "Calling some_method_2"
The solution would need to work both with class methods and instance methods, the method returned from the decorator should work and behave just the same way if it was undecorated i.e. method signature should be preserved.
The accepted answer for that question returned a Class from the decorator and the metaclass identified that specific Class, and did the "class-accessing" operations.
The answer did mention itself as a rough solution, but clearly it had a few caveats :
Decorator returned a class and it was not callable. Obviously, it can be made callable easily, but the returned value is still a class - it just behaves the same way while calling, but its properties and behaviors would be different. Essentially, it would not work the same way as the undecorated method.
It forced the decorator to return a custom-type class and all the "class-accessing" code was put inside the metaclass directly. It is simply not nice, writing the decorator should not enforce touching the metaclass directly.
I have tried to come up with a better solution, documented in the answer.
Here is the solution.
It uses a decorator (which would work on "class-accessing" decorators) and a metaclass, which would fulfill all my requirements and address the problems of that answer. Probably the best advantage is that the "class-accessing" decorators can just access the class, without even touching the metaclass.
# Using metaclass and decorator to allow class access during class creation time
# No method defined within the class should have "_process_meta" as arg
# Potential problems: Using closures, function.func_globals is read-only
from functools import partial
import inspect
class meta(type):
def __new__(cls, name, base, clsdict):
temp_cls = type.__new__(cls, name, base, clsdict)
methods = inspect.getmembers(temp_cls, inspect.ismethod)
for (method_name, method_obj) in methods:
tmp_spec = inspect.getargspec(method_obj)
if "__process_meta" in tmp_spec.args:
what_to_do, main_func = tmp_spec.defaults[:-1]
f = method_obj.im_func
f.func_code, f.func_defaults, f.func_dict, f.func_doc, f.func_name = main_func.func_code, main_func.func_defaults, main_func.func_dict, main_func.func_doc, main_func.func_name
mod_func = what_to_do(temp_cls, f)
f.func_code, f.func_defaults, f.func_dict, f.func_doc, f.func_name = mod_func.func_code, mod_func.func_defaults, mod_func.func_dict, mod_func.func_doc, mod_func.func_name
return temp_cls
def do_it(what_to_do, main_func=None):
if main_func is None:
return partial(do_it, what_to_do)
def whatever(what_to_do=what_to_do, main_func=main_func, __process_meta=True):
pass
return whatever
def original_classmethod_decorator(cls, func):
# cls => class of the method
# appends default arg values to class variable "a"
func_defaults = inspect.getargspec(func).defaults
cls.a.append(func_defaults)
func.__doc__ = "This is a class method"
print "Calling original classmethod decorator"
return func
def original_method_decorator(cls, func):
func_defaults = inspect.getargspec(func).defaults
cls.a.append(func_defaults)
func.__doc__ = "This is a instance method" # Can change func properties
print "Calling original method decorator"
return func
class A(object):
__metaclass__ = meta
a = []
#classmethod
#do_it(original_classmethod_decorator)
def some_method(cls, x=1):
''' hello'''
print "Calling original class method"
#do_it(original_method_decorator)
def some_method_2(self, y=2):
''' hello again'''
print "Calling original method"
# signature preserved
print(inspect.getargspec(A.some_method))
print(inspect.getargspec(A.some_method_2))
Open to suggestions on whether this approach has any ceveats.