I may well be being dumb (it has happened before), but I'm struggling
to fix some code breakage with Python 2.6.
I have some code that looks for the '__lt__' method on a class:
if hasattr(clr, '__lt__'):
However - in Python 2.6 object has grown a default implementation of
'__lt__', so this test always returns True.
>>> class X(object): pass
...
>>> X.__lt__
<method-wrapper '__lt__' of type object at 0xa15cf0>
>>> X.__lt__ == object.__lt__
False
So how do I tell if the X.__lt__ is inherited from object? I can look
in the '__dict__' of the class - but that doesn't tell me if X
inherits '__lt__' from a base class other than object. (Looking inside
the method wrapper repr with a regex is not an acceptable answer...)
Some things I have tried:
>>> X.__lt__.__self__
<class '__main__.X'>
>>> dir(X.__lt__)
['__call__', '__class__', '__cmp__', '__delattr__', '__doc__',
'__format__', '__getattribute__', '__hash__', '__init__', '__name__',
'__new__', '__objclass__', '__reduce__', '__reduce_ex__', '__repr__',
'__self__', '__setattr__', '__sizeof__', '__str__',
'__subclasshook__']
>>> X.__lt__.__func__
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
AttributeError: 'method-wrapper' object has no attribute '__func__'
Hmmm... I can get this working with Python 2.6 with:
if '__lt__' in dir(cls):
The default implementation of '__lt__' doesn't appear in the dir of
classes. However this fails with Python 3 where the default
implementation *does* appear in the output of 'dir'. Any suggestions?
Michael Foord
--
http://www.ironpythoninaction.com/
X.__lt__ is object.__lt__
That doesn't work for me.
>>> class A( object ):
... pass
...
>>> class B( A ):
... def __lt__( self, other ):
... return self
...
>>> a= A()
>>> b= B()
>>> B.__lt__ is object.__lt__
False
>>> A.__lt__ is object.__lt__
False
>>>
Further, it's been noted before that
A().meth is not A().meth
Didn't you see that even an equality test fails - so they are not the
same (that being the problem)...
They are unbound method objects - in Python 3 the unbound method has
gone away, so the problem is with Python 2.6.
Michael
--
http://www.ironpythoninaction.com/
Not tested extensively.
class NoLTException( Exception ): pass
class NoLT( object ):
def __lt__( self, other ):
raise NoLTException()
class A( NoLT ):
pass
class B( A ):
def __lt__( self, other ):
return self
def test_lt( obj ):
try:
obj.__lt__( None )
except NoLTException:
return False
except:
pass
return True
>>> a= A()
>>> b= B()
>>> test_lt( a )
False
>>> test_lt( b )
True
>>>
This method won't work for arbitrary classes, only ones that you
control, that inherit from 'NoLT'. The 'test_lt' function works by
trying to call '__lt__' on its argument. The parameter to it doesn't
matter because of what happens next. If '__lt__' raises a
NoLTException, you know it was inherited from NoLT. Otherwise, even
if another exception occurs, the object you know has '__lt__'.
It's a very object oriented solution. Essentially you're inheriting
all the classes that you want to fail, from a class that does.
But not a very good solution to the problem...
The specific problem is to determine if an arbitrary class implements
a specified comparison method. The general problem (that gives rise to
the specific problem) is to write a class decorator that can implement
all comparison methods from a class that implements only one.
See: http://code.activestate.com/recipes/576529/
Michael
--
http://www.ironpythoninaction.com/
Nope, I'm out of ideas, I'm afraid.
Thankfully that page I pointed you to has the solution I came up with
- walk the method resolution order of the class checking in the
classes' '__dict__' to see what they explicitly implement.
Given that you can get hold of X.__lt__ I was surprised by how hard it
was to tell whether that was an inherited implementation or not.
Michael
--
http://www.ironpythoninaction.com
I don't have Python 2.6 available, but if __lt__ on it works similarly
as __str__ on Python 2.5, you might be able to achieve this either
with inspect.ismethod or by checking methods' im_class attribute
directly:
>>> class C(object):
... pass
...
>>> class D(object):
... def __str__(self):
... return ''
...
>>> class E(D):
... pass
...
>>> import inspect
>>> inspect.ismethod(C().__str__)
False
>>> inspect.ismethod(D().__str__)
True
>>> inspect.ismethod(E().__str__)
True
>>>
>>> C().__str__.im_class
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
AttributeError: 'method-wrapper' object has no attribute 'im_class'
>>> D().__str__.im_class
<class '__main__.D'>
>>> E().__str__.im_class
<class '__main__.E'>
Cheers,
.peke
Ooops, didn't notice this was suggested already. One more attempt,
hopefully this is unique. =)
>>> C().__str__.__objclass__
<type 'object'>
>>> D().__str__.__objclass__
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
AttributeError: 'function' object has no attribute '__objclass__'
>>> 'spam'.__str__.__objclass__
<type 'str'>
Someone who actually knows what __objclas__ does can probably comment
does this make any sense in your case.
Cheers,
.peke