Quickly create flexible Python decorators


Keywords
decorators, decorator, @, at-syntax
License
MIT
Install
pip install decorators==0.1

Documentation

Decorators

The problem

Python has a few forms for decorators, you can have a plain simple decorator, with no arguments:

@mydecorator
def foo(): pass

Or a decorator with some arguments:

@mydecorator(1, 2)
def foo(): pass

You can even decorate a class:

@mydecorator
class Foo(object): pass

and each form is a little different to implement. This was frustrating if you wanted to create easy to use decorators where the developer didn't need to worry about @mydecorator() working differently than @mydecorator.

decorators module

The decorators.Decorator class allows you to easily create broad decorators that encompass all forms and all types (functions, methods, classes) using the same interface:

import decorators

class mydecorator(decorators.Decorator):
    def decorate_func(self, func, *dec_args, **dec_kwargs):
        def decorator(*args, *kwargs):
            print("You passed into the decorator these arguments", dec_args, dec_kwargs)
            print("You passed into your function these arguments", args, kwargs)
            print("Your function is", func)
            return func(*args, **kwargs)

        return decorator

    def decorate_class(self, klass, *dec_args, **dec_kwargs):
        print("You passed into the decorator these arguments", dec_args, dec_kwargs)
        print("Your class is", klass)
        return klass

You can then use this decorator:

@mydecorator
def foo(): print "foo()"

@mydecorator(1, 2, boom="blam")
def bar(*args, **kwargs): print "bar()"

@mydecorator
class Baz(object): pass

@mydecorator(1, 2, boom="blam")
class Che(object): pass

Now, your decorator can decorate functions or classes, pass in arguments, or not, and you never have to worry about the subtle differences between the decorators, and best of all, you don't have to duplicate code.

Other decorators

FuncDecorator, ClassDecorator, and InstanceDecorator

The Decorator class is good if you want to create a decorator that is totally flexible, if you want to enforce your decorator only being used for a function/method, you can use FuncDecorator. If you want to only decorate a class, use ClassDecorator, and if you want to decorate every instance of a class, use InstanceDecorator.

Whatever child class you use, you override the decorate method to return your decorator function:

import decorators

class only_func(FuncDecorator):
    def decorate(self, func, *dec_a, **dec_kw):
        def decorator(*args, **kwargs):
            return func(*args, **kwargs)
        return decorator

# this will work
@only_func
def foo(): pass

# this will fail
@only_func
class Foo(object): pass

Property Decorator

The property decorator is a drop-in replacement for Python's built-in property decorator, with additional functionality:

from decorators import property

class Foo(object):
    @property
    def bar(self):
        """This will act just like python's built-in @property decorator"""
        return 1
        
    @property(cached="_che")
    def che(self):
        """This will cache the return value into _che and add a setter/deleter"""
        return 1

Classproperty Decorator

Allows you to create a property on the class:

from decorators import classproperty

class Foo(object):
    @classproperty
    def bar(cls):
        """Available as Foo.bar"""
        return 1
        
print(Foo.bar) # 1

Installation

Use pip:

pip install decorators

Or, to get the latest and greatest from source:

pip install -U "git+https://github.com/Jaymon/decorators#egg=decorators"