Defining __setattr__ overrides all setter methods / properties I define in a class. I want to use the defined setter methods in the property, if a property exists
You need to rewrite your __setattr__ function. As per the docs, new style classes should use baseclass.__setattr__(self, attr, value) instead of self.__dict__[attr] = value. The former will lookup any descriptors whereas the latter will assign directly to the dict.
So rewrite your method as
def __setattr__(self, name, value):
object.__setattr__(self, name, value)
or
def __setattr__(self, name, value):
super(Test, self).__setattr__(name, value)
and you'll be fine. The code
class Test(object):
@property
def gx(self):
print "getting gx"
return self.__dict__['gx']
@gx.setter
def gx(self, value):
print "setting gx"
self.__dict__['gx'] = value
def __setattr__(self, attr, value):
print "using setattr"
object.__setattr__(self, attr, value)
t = Test()
t.gx = 4
t.dummy = 5
print t.gx
print t.dummy
print dir(Test)
outputs
using setattr
setting gx
getting gx
using setattr
4
5
['__class__', '__delattr__', '__dict__', '__doc__', '__format__', '__getattribute__', '__hash__', '__init__', '__module__', '__new__', '__reduce__', '__reduce_ex__', '__repr__', '__setattr__', '__sizeof__', '__str__', '__subclasshook__', '__weakref__', 'gx']
I don't know why your version is calling the getter twice. This one doesn't. Also, to answer your question about where descriptors live, you can plainly see it as the last entry in the class dict.
It's worth noting that you don't need __setattr__ to do what you want in your example. Python will always write an assignment foo.bar = x to foo.__dict__['bar'] = x regardless of if there's an entry in foo.__dict__ or not. __setattr__ is for when you want to transform the value or log the assignment or something like that.