I need a class that works like this:
>>> a=Foo()
>>> b=Foo()
>>> c=Foo()
>>> c.i
3
Here is my try:
class Foo(object):
i = 0
def __init__(self):
Foo.i += 1
It works as required, but I wonder if there is a more pythonic way to do it.
Nope. That's pretty good.
From The Zen of Python: "Simple is better than complex."
That works fine and is clear on what you're doing, don't complicate it. Maybe name it counter
or something, but other than that you're good to go as far as pythonic goes.
Abuse of decorators and metaclasses.
def counting(cls):
class MetaClass(getattr(cls, '__class__', type)):
__counter = 0
def __new__(meta, name, bases, attrs):
old_init = attrs.get('__init__')
def __init__(*args, **kwargs):
MetaClass.__counter += 1
if old_init: return old_init(*args, **kwargs)
@classmethod
def get_counter(cls):
return MetaClass.__counter
new_attrs = dict(attrs)
new_attrs.update({'__init__': __init__, 'get_counter': get_counter})
return super(MetaClass, meta).__new__(meta, name, bases, new_attrs)
return MetaClass(cls.__name__, cls.__bases__, cls.__dict__)
@counting
class Foo(object):
pass
class Bar(Foo):
pass
print Foo.get_counter() # ==> 0
print Foo().get_counter() # ==> 1
print Bar.get_counter() # ==> 1
print Bar().get_counter() # ==> 2
print Foo.get_counter() # ==> 2
print Foo().get_counter() # ==> 3
You can tell it's Pythonic by the frequent use of double underscored names. (Kidding, kidding...)
If you want to worry about thread safety (so that the class variable can be modified from multiple threads that are instantiating Foo
s), the above answer is in correct. I asked this question about thread safety here. In summary, you would have to do something like this:
from __future__ import with_statement # for python 2.5
import threading
class Foo(object):
lock = threading.Lock()
instance_count = 0
def __init__(self):
with Foo.lock:
Foo.instance_count += 1
Now Foo
may be instantiated from multiple threads.