Why can't I use #staticmethod here? - python

I'm using django-fsm to implement a state machine. The code looks like
def user_ok_to_check_me( instance, user):
...
class Job( models.Model):
# ... many screenfulls of code
#transition( field=state, target=BOOKING, source=CHECKING, permission=user_ok_to_check_me)
def fail_checking(self, **kwargs):
...
and it's working. Code readability is impaired by having that little utility function outside the class it belongs with, so I tried
class Job( models.Model):
# ... many screenfulls of code
#staticmethod
def user_ok_to_check_me( instance, user):
...
#transition( field=state, target=BOOKING, source=CHECKING, permission=user_ok_to_check_me)
def fail_checking(self, **kwargs):
which does not work. Not sure what user_ok_to_check_me does now do, it behaves like a no-op function always returning True even when all it does is return False
Why? And is there any way to declare this little function inside the class? (It's just a little bit too long to use lambda instance, user: )

Have found the answer, although I'm not sure I understand it.
class Job( models.Model):
# ... many screenfulls of code
# #staticmethod #NO DECORATOR
def user_ok_to_check_me( instance, user):
...
#transition( field=state, target=BOOKING, source=CHECKING, permission=user_ok_to_check_me)
def fail_checking(self, **kwargs):
The use of ok_to_check_me in #transition occurs during the execution of the code that creates the class, and not during the instantiation thereof. So it needs a reference to the actual function defined above. Application of #staticmethod to that function replaces it by something else, and whatever that is, is not acceptable to the transition decorator.
When the class is instantiated, the function gets bound to the instance. This does not, however, affect the reference to the function which #transition has already stored in its internals. In this case the binding is harmless since instance and self normally refer to the same. In other cases one might want to delete the un-intended bound function from the instance in its __init__ method (or just heavily document not to try to use it as an object method).

Related

Don't break if class instance is passed as initialization argument

I'm trying to add flexibility to a python class, so that it notices when one of the init arguments is already an instance of that class. Skip "Initial situation" if you don't mind, how I got here.
Initial situation
I have this class:
class Pet:
def __init__(self, animal):
self._animal = animal
#property
def present(self):
return "This pet is a " + self._animal
...
and there are many functions which accept an instance of this class as an argument (def f(pet, ...)). Everything worked as expected.
I then wanted to add some flexibility to the usage of these functions: if the caller passes a Pet instance, everything keeps on working as before. In all other cases, a Pet instance is created. One way to achieve that, is like this:
def f(pet_or_animal, ...):
if isinstance(pet_or_animal, Pet): #Pet instance was passed
pet = pet_or_animal
else: #animal string was passed
pet = Pet(pet_or_animal)
...
This also works as expected, but these lines are repeated in every function. Not DRY, not good.
Goal
So, I'd like to extract the if/else from each of the functions, and integrate it into the Pet class itself. I tried changing its __init__ method to
class PetA: #I've changed the name to facilitate discussion here.
def __init__(self, pet_or_animal):
if isinstance(pet_or_animal, PetA):
self = pet_or_animal
else:
self._animal = pet_or_animal
...
and start each function with
def f(pet_or_animal, ...):
pet = PetA(pet_or_animal)
...
However, that is not working. If a Pet instance is passed, everything is good, but if a string is called, a Pet instance is not correctly created.
Current (ugly) solution
What is working, is to add a class method to the class, like so:
class PetB: #I've changed the name to facilitate discussion here.
#classmethod
def init(cls, pet_or_animal):
if isinstance(pet_or_animal, PetB):
return pet_or_animal
else:
return cls(pet_or_animal)
def __init__(self, animal):
self._animal = animal
...
and also change the functions to
def f(pet_or_animal, ...):
pet = PetB.init(pet_or_animal) #ugly
...
Questions
Does anyone know, how to change class PetA so, that it has the intended behavior? To be sure, here is the quick test:
pb1 = PetB.init('dog')
pb2 = PetB.init(pb1) #correctly initialized; points to same instance as pb1 (as desired)
pa1 = PetA('cat')
pa2 = PetA(pa1) #incorrectly initialized; pa1 != pa2
More generally, is this the right way to go about adding this flexibility? Another option I considered was writing a separate function to just do the checking, but this too is rather ugly and yet another thing to keep track of. I'd rather keep everything neat and wrapped in the class itself.
And one final remark: I realize that some people might find the added class method (petB) a more elegant solution. The reason I prefer to add to the __init__ method (petA) is that, in my real-world use, I already allow for many different types of initialization arguments. So, there is already a list of if/elif/elif/... statements that check, just which of the possibilities is used by the creator. I'd like to extend that by one more case, namely, if an initialized instance is passed.
Many thanks
I believe your current "ugly" solution is actually the correct approach.
This pushes the flexibility up as far as possible, since it is messy. Even though python allows for arbitrary types and values to float around, your users and yourself will thank you for keeping that constrained to the outermost levels.
I would think of it as (don't need to implement it this way)
class Pet:
#classmethod
def from_animal(cls, ...):
...
#classmethod
def from_pet(cls, ...):
...
#classmethod
def auto(cls, ...):
if is_pet(...):
return cls.from_pet(...)
def __init__(cls, internal_rep):
...
etc.
It is a code smell if you don't know whether your function is taking an object or an initializer. See if you can do processing as up-front as possible with user input and standardize everything beyond there.
You could use a function instead to get the same behaviour you want:
def make_pet_if_required(pet_or_animal):
if isinstance(pet_or_animal, PetA):
return pet_or_animal
else:
return Pet(pet_or_animal)
And then:
def f(pet_or_animal, ...):
pet = make_pet_if_required(pet_or_animal)
...
For more "beauty" you can try turning that function call into a decorator.

Access the python class from method while defining it

I wanted to access the class on which method is to be defined. This can be used, for example, to create alias for methods with decorator. This particular case could be implemented without using decorator (alias = original_name), but I would like to use decorator, primarily so because the aliasing will be visible along side the method definition at the top, useful when the method definition is long.
def method_alias(*aliases):
def aliased(m):
class_of_m = ??? # GET class of this method
for alias in aliases:
setattr(class_of_m, alias, m)
return m
return aliased
class Test():
#method_alias('check', 'examine')
def test():
print('I am implemented with name "test"')
Later, I found here that the above could be implemented by using two decorators (first store the aliases as method attributes, later when the class is already created, add the attributes to class). Can it be done without decorating the class, i.e. only decorating the method? This requires getting access to the class name in the decorator.
The short answer is no. The contents of the class body are evaluated before the class object is created, i.e. the function test is created and passed to the decorator without class Test already existing. The decorator is therefore unable to obtain a reference to it.
To solve the problem of method aliasing, I reckon three approaches:
Using a class decorator as described by your link.
Using a metaclass, which lets you modifies the class' __dict__ before the class object is created. (Implementing a metaclass class is acutally overriding the default constructor for class objects, see here. Also the metaclass usage syntax has changed in Python 3.)
Creating the aliases in the __init__ method for each instance of Test.
The first approach is probably the most straightforward. I wrote another example. It basically does the same as your link, but is more stripped down to make it a bit clearer.
def alias(*aliases):
def decorator(f):
f.aliases = set(aliases)
return f
return decorator
def apply_aliases(cls):
for name, elem in list(cls.__dict__.items()):
if not hasattr(elem, 'aliases'):
continue
for alias in elem.aliases:
setattr(cls, alias, elem)
return cls
#apply_aliases
class Test(object):
#alias('check', 'examine')
def test(self):
print('I am implemented with name "test"')
Test().test()
Test().check()
Test().examine()

Why is get_default_prefix #classmethod instead of just a regular method

In formsets.py, you find this code snippet
class BaseFormSet(StrAndUnicode):
"""
A collection of instances of the same Form class.
"""
def __init__(self, data=None, files=None, auto_id='id_%s', prefix=None,
initial=None, error_class=ErrorList):
...
self.prefix = prefix or self.get_default_prefix() # Note the self.get_default_prefix
...
...
#classmethod # Note the #classmethod
def get_default_prefix(cls):
return 'form'
Why is get_default_prefix declared this way and then called with self.? Is there something gained by doing it this way? get_default_prefix has another definition in BaseInlineFormSet (forms/models.py)
class BaseInlineFormSet(BaseModelFormSet):
...
#classmethod
def get_default_prefix(cls):
from django.db.models.fields.related import RelatedObject
return RelatedObject(cls.fk.rel.to, cls.model, cls.fk).get_accessor_name().replace('+','')
and another in BaseGenericInlineFormset again using the #classmethod, so it doesn't appear to be a typo. I just don't understand why it would be done this way and then called with self.
The only clue I see (which I don't understand) is that the admin seems to call it with FormSet.get_default_prefix()
I'm wondering if there is something I'm just not understanding about python.
Calling a class method from an instance is perfectly legal, as you can see in the code. A related stackoverflow post said there was no benefit, (and it is bad practice) to call from an instance; because if you are only calling from instance your method should probably not be a classmethod.
I think you answer your own question, though. If django is calling FormSet.get_default_prefix() from somewhere, then they probably didn't want to instantiate a formset object

Is it bad practice to put stuff into new function properties?

Say I have a class and a function:
class AddressValidator(self):
def __init__(self):
pass
def validate(address):
# ...
def validate_address(addr):
validator = AddressValidator()
return validator.validate(addr)
The function is a shortcut for using the class, if you will. Now, what if this function has to be run thousands of times? If the validator class actually has to do something on instantiation, like connecting to a database, creating it over and over thousands of times is pretty wasteful. I was wondering if I could perhaps do something like this:
def validate_address(addr):
if not hasattr(validate_address, 'validator'):
validate_address.validator = AddressValidator()
validator = validate_address.validator
return validator.validate(addr)
Now the validator class is only instantiated once and saved "in the function", to put it that way. I've never seen this done though, so I'm guessing it's bad practice. If so, why?
Note: I know I can just cache the validator object in a module global. I'm just curious if this is a viable solution when I want to avoid littering my module.
Despite "everithing is an object", not everithing work as nice as instances of well controlled class.
This problem looks like typical case for "functor" or "callable object" as it called in python.
the code will be look something like
class AddressValidator(self):
def __init__(self):
pass
def __call__(self,address):
# ...
validate_address = AdressValidator()
or you could just define your function as shortcut to bound method
class AddressValidator(self):
def __init__(self):
pass
def validate(self,address):
# ...
validate_adress = AdressValidator().validate
I'd go with a default argument (evaluated once at function definition time and bound to the function):
def validate_address(addr, validator=AddressValidator())
return validator.validate(addr)
This is perfectly acceptable if instances of AddressValidator are considered immutable (i.e. they don't contain methods that modify their internal state), and it also allows you to later override the choice of validator should you find the need to (e.g. to provide a validator specialized for a particular country).

Why does "self" outside a function's parameters give a "not defined" error?

Look at this code:
class MyClass():
# Why does this give me "NameError: name 'self' is not defined":
mySelf = self
# But this does not?
def myFunction(self):
mySelf2 = self
Basically I want a way for a class to refer to itself without needing to name itself specifically, hence I want self to work for the class, not just methods/functions. How can I achieve this?
EDIT: The point of this is that I'm trying to refer to the class name from inside the class itself with something like self.class._name_ so that the class name isn't hardcoded anywhere in the class's code, and thus it's easier to re-use the code.
EDIT 2: From what I've learned from the answers below, what I'm trying to do is impossible. I'll have to find a different way. Mission abandoned.
EDIT 3: Here is specifically what I'm trying to do:
class simpleObject(object):
def __init__(self, request):
self.request = request
#view_defaults(renderer='string')
class Test(simpleObject):
# this line throws an error because of self
myClassName = self.__class__.__name__
#view_config(route_name=myClassName)
def activateTheView(self):
db = self.request.db
foo = 'bar'
return foo
Note that self is not defined at the time when you want the class to refer to itself for the assignment to work. This is because (in addition to being named arbitrarily), self refers to instances and not classes. At the time that the suspect line of code attempts to run, there is as of yet no class for it to refer to. Not that it would refer to the class if there was.
In a method, you can always use type(self). That will get the subclass of MyClass that created the current instance. If you want to hard-code to MyClass, that name will be available in the global scope of the methods. This will allow you to do everything that your example would allow if it actually worked. E.g, you can just do MyClass.some_attribute inside your methods.
You probably want to modify the class attributes after class creation. This can be done with decorators or on an ad-hoc basis. Metaclasses may be a better fit. Without knowing what you actually want to do though, it's impossible to say.
UPDATE:
Here's some code to do what you want. It uses a metaclass AutoViewConfigMeta and a new decorator to mark the methods that you want view_config applied to. I spoofed the view_config decorator. It prints out the class name when it's called though to prove that it has access to it. The metaclass __new__ just loops through the class dictionary and looks for methods that were marked by the auto_view_config decorator. It cleans off the mark and applies the view_config decorator with the appropriate class name.
Here's the code.
# This just spoofs the view_config decorator.
def view_config(route=''):
def dec(f):
def wrapper(*args, **kwargs):
print "route={0}".format(route)
return f(*args, **kwargs)
return wrapper
return dec
# Apply this decorator to methods for which you want to call view_config with
# the class name. It will tag them. The metaclass will apply view_config once it
# has the class name.
def auto_view_config(f):
f.auto_view_config = True
return f
class AutoViewConfigMeta(type):
def __new__(mcls, name, bases, dict_):
#This is called during class creation. _dict is the namespace of the class and
# name is it's name. So the idea is to pull out the methods that need
# view_config applied to them and manually apply them with the class name.
# We'll recognize them because they will have the auto_view_config attribute
# set on them by the `auto_view_config` decorator. Then use type to create
# the class and return it.
for item in dict_:
if hasattr(dict_[item], 'auto_view_config'):
method = dict_[item]
del method.auto_view_config # Clean up after ourselves.
# The next line is the manual form of applying a decorator.
dict_[item] = view_config(route=name)(method)
# Call out to type to actually create the class with the modified dict.
return type.__new__(mcls, name, bases, dict_)
class simpleObject(object):
__metaclass__ = AutoViewConfigMeta
class Test(simpleObject):
#auto_view_config
def activateTheView(self):
foo = 'bar'
print foo
if __name__=='__main__':
t = Test()
t.activateTheView()
Let me know if you have any questions.
Python has an "explict is better than implicit" design philosophy.
Many languages have an implicit pointer or variable in the scope of a method that (e.g. this in C++) that refers to the object through which the method was invoked. Python does not have this. Here, all bound methods will have an extra first argument that is the object through which the method was invoked. You can call it anything you want (self is not a keyword like this in C++). The name self is convention rather than a syntactic rule.
Your method myFunction defines the variable self as a parameter so it works. There's no such variable at the class level so it's erroring out.
So much for the explanation. I'm not aware of a straightforward way for you to do what you want and I've never seen such requirement in Python. Can you detail why you want to do such a thing? Perhaps there's an assumption that you're making which can be handled in another way using Python.
self is just a name, your self in this case is a class variable and not this for the object using which it is called,
self is treated as a normal variable and it is not defined, where as the self in the function comes from the object used for calling.
you want to treat the object reference in self as a class variable which is not possible.
self isn't a keyword, it's just a convention. The methods are attributes of the class object (not the instance), but they receive the instance as their first argument. You could rename the argument to xyzzy if you wanted and it would still work the same way.
But (as should be obvious) you can't refer to a method argument outside the body of the method. Inside a class block but outside of any method, self is undefined. And the concept wouldn't even make sense -- at the time the class block is being evaluated, no instance of the class can possibly exist yet.
Because the name self is explicitly defined as part of the arguments to myFunction. The first argument to a method is the instance that the method was called on; in the class body, there isn't an "instance we're dealing with", because the class body deals with every possible instance of the class (including ones that don't necessarily exist yet) - so, there isn't a particular object that could be called self.
If you want to refer to the class itself, rather than some instance of it, this is spelled self.__class__ (or, for new-style classes in Py2 and all classes in Py3, type(self)) anywhere self exists. If you want to be able to deal with this in situations where self doesn't exist, then you may want to look at class methods which aren't associated with any particular instance, and so take the class itself in place of self. If you really need to do this in the class body (and, you probably don't), you'll just have to call it by name.
You can't refer to the class itself within the class body because the class doesn't exist at the time that the class body is executed. (If the previous sentence is confusing, reading up about metaclasses will either clear this up or make you more confused.)
Within an instance method, you can refer to the class of the instance with self.__class__, but be careful here. This will be the instance's actual class, which through the power of inheritance might not be the class in which the method was defined.
Within a class method, the class is passed in as the first argument, much like instances are the first argument to instance methods:
class MyClass(object):
#classmethod
def foo(cls):
print cls.__name__
MyClass.foo() # Should print "MyClass"
As with instance methods, the actual class might differ due to inheritance.
class OtherClass(MyClass):
pass
OtherClass.foo() # Should print "OtherClass"
If you really need to refer to MyClass within a method of MyClass, you're pretty much going to have to refer to it as MyClass unless you use magic. This sort of magic is more trouble than it is worth.

Categories

Resources