Instrumenting a class to see how it is used

[Paul Moore]

I'm trying to reverse-engineer some pretty complex code. One thing that would 
help me a lot is if I could instrument a key class, so that I'd get a report of 
when and how each method was called and any properties or attributes were 
accessed during a typical run.

I could do this relatively easily by just adding print calls to each 
method/attribute, but I was wondering - is there a library that does this sort 
of wrapping already? I tried writing my own but got bogged down in infinite 
recursion in _getattribute__ and couldn't see an easy way out.

If anyone has any suggestions, I'd be interested.
Thanks,
Paul.
-- 
http://mail.python.org/mailman/listinfo/python-list

Re: Instrumenting a class to see how it is used

[Michele Simionato]

This may get you started (warning: not really tested).

$ echo instr.py
from warnings import warn

oget = object.__getattribute__
tget = type.__getattribute__

class Instr(object):

class __metaclass__(type):
def __getattribute__(cls, name):
clsname = tget(cls, '__name__')
warn('accessing %s.%s' % (clsname, name), stacklevel=2)
return tget(cls, name)

def __getattribute__(self, name):
warn('accessing %s.%s' % (self, name), stacklevel=2)
return oget(self, name)


Then change the base classes of the class you want to instrument, i.e.

class MyClass(MyBase) - class MyClass(MyBase, Instr)

and see what happens. It should work in simple cases.
It will log a lot, so will have to adapt this.
-- 
http://mail.python.org/mailman/listinfo/python-list