In node tree, parent.addChild(self) is adding self to self

I'm building a simple node tree in python. But when my constructor tries to add the current node as a child of a given parent, the current node also adds itself as a child of itself.

Here's my constructor method (and the creation of children):

children = []

def __init__(self, parent=None, tag="[ROOT]", attrs=None):
        self.parent = parent
        self.tag = tag
        self.attrs = attrs
        print "\n", "self:%s ... children:%s" % (self.tag, self.children)
        if parent != None:
            parent.addChild(self)
        print "self:%s开发者_JAVA百科 ... children:%s" % (self.tag, self.children)

Here's my addChild method in the same class (which is supposed to be called for the parent, not the node currently under construction):

def addChild(self, child):
        self.children.append(child)

Here's the output:

foo []
foo [foo]

The two lines of output should be the same because the line of code between them should only affect the parent node, not the node currently under construction.

What am I doing wrong?

When you initialize children at the class level every instance of your class ends up sharing the same list object.

>>> class C:
...  children = []
... 
>>> a = C()
>>> b = C()
>>> id(a.children)
144349644
>>> id(b.children)
144349644

Try initializing it in your constructor instead:

def __init__(self, parent=None, tag="[ROOT]", attrs=None):
    self.children = []

I suspect children is somehow shared across all of your node objects; we can't see the declaration so I can't say exactly what you're doing wrong.