I am trying to define a python decorator (my_decorator) for a class method (f), shown below in a simplified scenario. my_decorator is parametrized by param, which depends on the class attribute (in this case level).
class my_decorator:
def __init__(self, param):
self.param = param
def __call__(self, f):
def f_decorated(instance, c):
print("decorated with param = %d" % self.param)
return f(c)
return f_decorated
class A:
def __init__(self, level):
self.level = level
#my_decorator(param=self.level) # Here is the problematic line!
def f(x):
return x
if __name__ == "__main__":
a = A(level=2)
a.f(1) # name "self" is not defined
The above code does not work, and I get a "self" is not defined error. So my question is, is there any way to achieve the goal of context-parametrized decorator?
BTW, the use case is: I am trying to achieve persistent memoization technique (described at
memoize to disk - python - persistent memoization)
The file where the cache persists to depends on the class A, specifically 'level'. For instance, I would like to persist to the file cache_%d.txt % self.level .
Chen,
Decorator are executed during the compiled time or during the import as the class body is executed during import. So, if you execute your snippet without creating an instance of that class also will throw error.
And more over that parameter self.level inside decorator doesn't make much sense to me as its an instance variable so you can directly use inside the function f(x):
Here is some more details:
Python decorator function called at compile time
As the error says, self doesn't exist at that point. That should be clear to you: self only exists as a parameter to a method, and you're not even in a method at that time. Decorators, like all class-level code are evaluated at define time.
I'm not totally sure what you want to achieve, but you could use a string along with getattr:
class my_decorator:
def __init__(self, param_name):
self.param_name = param_name
def __call__(self, f):
def f_decorated(instance, c):
param = getattr(instance, self.param_name)
print("decorated with param = %d" % param)
return f(c)
...
class A:
def __init__(self, level):
self.level = level
#my_decorator(param_name='level')
def f(x):
return x
self is a variable as any. It's only defined inside of methods, the decorator is outside. If you need attributes of a object inside an decorator, you have the possibility to access it by string-name:
class my_decorator:
def __init__(self, param):
self.param = param
def __call__(self, f):
def f_decorated(instance, c):
print("decorated with param = %d" % getattr(instance, self.param))
return f(instance, c)
return f_decorated
class A:
def __init__(self, level):
self.level = level
#my_decorator(param='level') # Here is the problematic line!
def f(self, x):
return x
if __name__ == "__main__":
a = A(level=2)
a.f(1) # name "self" is not defined
Related
I'm having the following basic problem in Python.
I want to create a constant static instance of a class within the class itself, to be used in methods of the class.
class MyClass(object):
def __init__(self, i : int):
self.i_c = i
newinstance = MyClass(0)
def method(self):
if self == newinstance:
return 'blaba'
else:
return self.i_c
Of course in this example, i could define the instance separately in the module containing my class, but then I could not use it in the methods of the class.
I don't know if it's feasible and I just don't know the right syntax, or if I cannot do this.
Edit: Based on the comments below one workaround is the following
class MyClass(object):
def __init__(self, i : int):
self.i_c = i
MyClass.newinstance = MyClass(0)
def func(thing):
if thing == MyClass.newinstance:
return 'blaba'
else:
return thing.i_c
This works, but func is now not a method of the class MyClass. This is not really problematic in my case, but I guess I can probably define func as a method of MyClass, I just don't really know how.
Ok, I got it, finally thanks to the comments below. This works
class MyClass(object):
def __init__(self, i : int):
self.i_c = i
def method(self):
if self == MyClass.newinstance:
return 'blaba'
else:
return self.i_c
MyClass.newinstance = MyClass(0)
Is it possible to write a decorator that acts upon a class's method and uses the class's attributes? For example, I would like to add a decorator to functions that will return an error if one of the class's attributes (which is set when the user calls the function) is False.
For example, my attempt (broken code since is_active can't access MyClass's methods):
def is_active(active):
if active == False:
raise Exception("ERROR: Class is inactive")
class MyClass():
def __init__(self, active):
self.active = active
#is_active
def foo(self, variable):
print("foo")
return variable
#is_active
def bar(self, variable):
print("bar")
return variable
where the expected behaviour is:
cls = MyClass(active=True)
cls.foo(42)
---> function prints "foo" and returns 42
cls = MyClass(active=False)
cls.foo(42)
---> function raises an exception as the active flag is False
The above is a dummy example and the actual use case is more complex, but hopefully this shows the problem I'm facing.
If the above is possible, my extra question is: is it possible to "hide"/delete the methods from the instantiated class based on this flag. For example, if the user instantiates the class with a active=False then when they're using iPython and press <tab>, they can only see the methods which are permitted to be used?
Thank you.
Decorators can be confusing. Note a function is passed as a parameter and the decorator expects that a function (or callable object) is returned. So you just need to return a different function. You have everything else you need since self is passed as the first argument to a class method. You just need to add a new function in your decorator that does what you want.
def is_active_method(func):
def new_func(*args, **kwargs):
self_arg = args[0] # First argument is the self
if not self_arg.active:
raise Exception("ERROR: Class is inactive")
return func(*args, **kwargs)
return new_func
class MyClass():
def __init__(self, active):
self.active = active
#is_active_method
def foo(self, variable):
print("foo")
return variable
#is_active_method
def bar(self, variable):
print("bar")
return variable
m = MyClass(True) # Prints foo from the method
m.foo(2)
m = MyClass(False) # Outputs the exception
m.foo(2)
I have a situation where I'm using #classmethod to create a constructor for a class. Within this constructor, a function gets called, which then in turn calls another function. But either this doesn't work or (more probably) I'm doing something to make it not work. Here's an example in miniature:
class testclass:
def __init__(self, x):
self.x = x
#classmethod
def constructor(cls, x):
adj_x = cls.outer_adjust(cls, x)
return testclass(adj_x)
def outer_adjust(self, x):
return self.inner_adjust(x)
def inner_adjust(self, x):
return x + 1
test_instance = testclass.constructor(4)
This produces an error message:
inner_adjust() missing 1 required positional argument: 'x'
I can make it work by explicitly passing self to inner_adjust, eg
def outer_adjust(self, x):
return self.inner_adjust(self, x)
But this then means that the outer_adjust method can't be used outside of the constructor, which is not what I want.
Any assistance gratefully received.
Here's a more detailed example, with two constructors shown. I'm trying to follow the approach to constructors described in
What is a clean, pythonic way to have multiple constructors in Python?
Which is essentially that the constructors do some processing to figure out what variables they should pass to init when instantiating the class.
Both constructors give the same error:
if_char_is_z_make_it_a() missing 1 required positional argument: 'char_input'
As before, I need to be able to use the if_char_is_make_it_a function outside of the constructor (ie, when using the class normally).
class testclass:
def __init__(self, char):
self.char = char
#classmethod
def constructor_from_int(cls, int_input):
as_char = chr(int_input)
char = cls.process_char(cls, as_char)
return testclass(char)
#classmethod
def constructor_from_char(cls, char_input):
char = cls.process_char(cls, char_input)
return testclass(char)
def process_char(self, char_input):
processed_char = '(' + char_input + ')'
output_char = self.if_char_is_z_make_it_a(processed_char)
return output_char
def if_char_is_z_make_it_a(self, char_input):
if char_input == '(z)':
return '(a)'
return char_input
test_instance = testclass.constructor_from_char('a')
When you call cls.outer_adjust from constructor you are calling the unbound outer_adjust method.
Thus, you pass the class itself as self and not an instance to a method that expects to receive an instance as argument.
Although, there is no real reason to have a constructor method. This is exactly what __init__ is for.
class testclass:
def __init__(self, x):
self.x = self.outer_adjust(x)
def outer_adjust(self, x):
return self.inner_adjust(x)
def inner_adjust(self, x):
return x + 1
test_instance = testclass(4)
If you absolutely need the transformation on x to be done before the instantiation, then use __new__ instead. Although, this is generally not necessary.
Multiple constructors
If for some reason you still need to have a constructor method, by example if you want multiple constructors. Then keep in mind that outer_adjust and inner_adjust are instance methods, this means they must be called after you have created an instance.
class testclass:
def __init__(self, x):
self.x = x
#classmethod
def constructor1(cls, x):
instance = cls(x)
instance.outer_adjust()
return instance
#classmethod
def constructor2(cls, x):
instance = cls(x)
instance.inner_adjust()
return instance
def outer_adjust(self):
print('Do something else')
return self.inner_adjust()
def inner_adjust(self):
self.x += 1
As a sidenote, notice how I did not need to call testclass, but simply called cls in the constructor methods. Since this is a class method, we do not need to explicitly name the class. This is better, especially if you are to use inheritance.
Basically what you are doing here shall be done via the __new__ which serve as constructor.
class testclass:
def __init__(self, x):
self.x = x
def __new__(cls, *args, **kwargs):
instance = super(testclass, cls).__new__(cls, *args, **kwargs)
instance.outer_adjust(args[0])
return instance
def outer_adjust(self, x):
return self.inner_adjust(x)
def inner_adjust(self, x):
self.x = x + 1
test_instance = testclass(4)
You are abusing self. The point of the class method is to use the cls argument as constructor, instead of explicitly naming the class by testclass(adj_x). Also, during the cls.outer_adjust(cls, x) call, you are passing the class instead of the instance, which happens to work because you are not using any instance attributes.
As to your questions, there's no way to avoid the x argument. inner_adjust increases some value by 1, so you must give it something to increase. The idea would be to have
def constructor(cls, x):
return cls(x)
def inner_adjust(self):
return self.x += 1
and then do something like
object= testclass.constructor(12)
object.inner_adjust()
I know there is similar question, but my scenario is somehow different: refer to codes:
class MyClass(object):
def __init__(self, log_location)
self.logs = logging(log_location) # create log object by the log_location, this object should be used by the decorator fucntion
def record_log(log_object):
""" this is the decorator function
"""
def deco(func):
def wrap(*args, **kwargs):
rs = func()
# use log object to record log
if rs:
log_object.record('success')
else:
log_object.record('fail')
return wrap
return deco
#record_log(self.logs)
def test(self):
rs = do_some_thing
if rs:
return True
return False
def main():
my_class = MyClass()
my_class.test()
But, there is an error like this:
#record_log(self.logs)
NameError: name 'self' is not defined
Hos should I use the instance attribute self.logs in a decorator function in such scenario like this??
Thanks very much!
You can not pass a reference to self or any attribute of self at this point. The #record_log line is executed (the method is decorated) before the code in main is executed, i.e. before any instance of MyClass is created at all -- in fact, even before the definition of MyClass has been completed! But remember that
#record_log(self.logs)
def test(self, n):
is actually just syntactic sugar for
test = record_log(self.logs)(test)
So one way to work around your problem would be to redefine test in your __init__, i.e.
def __init__(self, log_location)
self.logs = logging(log_location)
self.test = record_log(self.logs)(self.test)
Also note that your decorator is not passing any parameters to func and not returning the results. Also, it should probably be defined on module level (before the class).
def record_log(log_object):
def deco(func):
def wrap(*args, **kwargs):
rs = func(*args, **kwargs) # pass parameters
if rs:
log_object.record('success')
else:
log_object.record('fail')
return rs # return result
return wrap
return deco
There are several objections about your code:
deco() is redundant. You can directly return wrap from record_log().
If you only plan to decorate MyClass's methods, then there is no point in passing log_object to the decorator, as self.logs will always be used. Otherwise, consider moving the decorator to module level, as already suggested by others.
The decorated method's return value is currently lost.
The call to the decorated function does not pass self to it.
The proper code would therefore be:
class MyClass(object):
def __init__(self, log_location):
self.logs = logging(log_location)
def record_log(func):
""" this is the decorator function
"""
def wrap(self):
rs = func(self)
# use log object to record log
if rs:
print 1
self.logs.record('success')
else:
print 2
self.logs.record('fail')
return rs
return wrap
#record_log
def test(self):
rs = do_some_thing
if rs:
return True
return False
I'm just getting to grips with decorators in Python and using them to add callbacks to some instance variables using the following simple pattern:
class A(object):
def __init__(self):
self._var = 0
self.var_callbacks = []
#property
def var(self):
return self._var
#var.setter
def var(self, x):
self._var = x
for f in self.var_callbacks:
f(x)
The property decorator is a neat way of allowing me to introduce callbacks where necessary without changing the class interface. However, after the third or fourth variable it's making the code a bit repetitive.
Is there a way to refactor this pattern into something along the following:
class A(object):
def __init__(self):
self.var = 0
enable_callback(self, 'var', 'var_callbacks')
You'll need to set the property on the class (since it is a descriptor), so using a enable_callback call in the initializer is not going to work.
You could use a class decorator to set the properties from a pattern:
def callback_properties(callbacks_attribute, *names):
def create_callback_property(name):
def getter(self):
return getattr(self, '_' + name)
def setter(self, value):
setattr(self, '_' + name, value)
for f in getattr(self, callbacks_attribute):
f(value)
return property(getter, setter)
def add_callback_properties(cls):
for name in names:
setattr(cls, name, create_callback_property(name)
return cls
return add_callback_properties
Then use that as:
#add_callback_properties('var_callbacks', 'var1', 'var2')
class A(object):
# everything else
Have a look at the Python descriptor protocol. In essence, you can define a class that handles the getting, setting and deleting of a property. So you could define a descriptor that runs your callbacks on setting the attribute.
Descriptors are regular classes, and can be parameterized. So you could implement a descriptor that takes the destination variable its constructor. Something like the following:
class A(object):
var = CallbackDescriptor('var')
foo = CallbackDescriptor('foo')