问题:

I would like to define a single object that can be iterated over without having to create a class and then an instance. Something like this:

class Thing(object):
    stuff = ["foo", "bar", "baz"]

    @classmethod
    def __iter__(cls):
        return iter(cls.stuff)

for thing in Thing:
    print thing

However this doesn't actually work. Is there any way to do this?

回答1:

What Ashwini correctly suggested in his comment is the following. This works in Python 2.

class ThingType(type):
    __stuff__ = ["foo", "bar", "baz"]

    @classmethod
    def __iter__(cls):
        return iter(cls.__stuff__)

class Thing(object):
    __metaclass__ = ThingType

for thing in Thing:
    print thing

And this works in Python 3:

class ThingType(type):
    __stuff__ = ["foo", "bar", "baz"]

    @classmethod
    def __iter__(cls):
        return iter(cls.__stuff__)

class Thing(object, metaclass=ThingType):
    pass

for thing in Thing:
    print(thing)

回答2:

Does Thing actually need to be a type? You could make it an object that has type-like callable behavior, which would probably be simpler:

class RealThing(object):
  pass

class ThingFactory(object):
  def __iter__(self):
    return iter(["foo", "bar", "baz"])

  def __call__(self):
    return RealThing()

Thing = ThingFactory()