python access to attributes redefined in derived classes

Question

I'm puzzled with python the more I get into it.

For example the following code:

class A:
    def __init__ (self):
        self.test = "a"

    def dump(self):
        print("A.test: %s" % (self.test,))
        print("A.test: %s" % (self.__dict__["test"],))
        print("A.test: %s" % (getattr(self, "test"),))
        print()

class B (A):
    def __init__ (self):
        self.test = "b"

    def dump(self):
        print("B.test: %s" % (self.test,))
        print("B.test: %s" % (self.__dict__["test"],))
        print("B.test: %s" % (getattr(self, "test"),))
        print()


o = B ()
o.dump()
A.dump(o)
super(B, o).dump()

prints:

B.test: b
B.test: b
B.test: b

A.test: b
A.test: b
A.test: b

A.test: b
A.test: b
A.test: b

which seems to show that you can call a function of a base class but if that class has an attribute which was also used in some derived class you can't access this attribute by using the normal object.attribute notation or perhaps you can't access it at all.

Is that really true? If so it would kill - IMHO - the whole python object model.

No, you can't access it at all. Perhaps you should delve in the whole class ideology of Python. It is a bit different than languages such as C++ and C# — dmg
– dmg, Commented Feb 20, 2013 at 17:02
The fact you observe does not in any way harm the Python object model. It is, in fact, rather central. — user395760
– user395760, Commented Feb 20, 2013 at 17:26

Daniel Roseman · Accepted Answer · 2013-02-20 16:59:42Z

6

o is an instance of B, not A. When it is initialised, it sets the value of its test attribute to "b". At no point do you even call the superclass __init__, so it never gets set to "a" - but even if you did, it can only have one of the values "a" or "b", depending on whether you called super first or second.

No idea why you think this breaks the Python object model.

answered Feb 20, 2013 at 16:59

Daniel Roseman

602k68 gold badges911 silver badges924 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

eyquem · Accepted Answer · 2013-02-20 17:40:13Z

0

You say

if that class has an attribute which was also used in some derived class

In the definition of A

class A:
    def __init__ (self):
        self.test = "a"

doesn't mean that an object test of name test is created as an attribute of the class A
It means that each time an object will be instantiated, an attribute test will be creted in THE INSTANCE

So, you can't have hope to access to an attribute of A that doesn't exist.

def A:
    attA = 13

does creates the object 13 as an attribute of A : print(A.attA) will print the value of 13

answered Feb 20, 2013 at 17:40

eyquem

27.7k7 gold badges43 silver badges46 bronze badges

Comments

thebjorn · Accepted Answer · 2013-02-20 17:59:41Z

0

Is this what you are looking for?

class A(object):
    def __init__ (self):
        self.__test = "a"

    def dump(self):
        print "A.test:", self.__test

class B (A):
    def __init__ (self):
        super(B, self).__init__()
        self.__test = "b"

    def dump(self):
        print "B.test:", self.__test


o = B()
o.dump()
super(B, o).dump()

output:

B.test: b
A.test: a

(using Python 2, but you get the idea..)

This works because Python does some fancy name-mangling with pseudo-private __variables.

answered Feb 20, 2013 at 17:59

thebjorn

27.6k12 gold badges107 silver badges152 bronze badges

Collectives™ on Stack Overflow

python access to attributes redefined in derived classes

3 Answers 3

Comments

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

Comments

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related