Nested try statements in python?

Question

Is there a nicer way of doing the following:

try:
    a.method1()
except AttributeError:
    try:
        a.method2()
    except AttributeError:
        try:
            a.method3()
        except AttributeError:
            raise

It looks pretty nasty and I'd rather not do:

if hasattr(a, 'method1'):
    a.method1()
else if hasattr(a, 'method2'):
    a.method2()
else if hasattr(a, 'method3'):
    a.method3()
else:
    raise AttributeError

to maintain maximum efficiency...

Answer 1

How about encapsulating the calls in a function?

def method_1_2_or_3():
    try:
        a.method1()
        return
    except AttributeError:
        pass
    try:
        a.method2()
        return
    except AttributeError:
        pass
    try:
        a.method3()
    except AttributeError:
        raise

Answer 2

Perhaps you could try something like this:

def call_attrs(obj, attrs_list, *args):
    for attr in attrs_list:
        if hasattr(obj, attr):
            bound_method = getattr(obj, attr)
            return bound_method(*args)

    raise AttributeError

You would call it like this:

call_attrs(a, ['method1', 'method2', 'method3'])

This will try to call the methods in the order they are in in the list. If you wanted to pass any arguments, you could just pass them along after the list like so:

call_attrs(a, ['method1', 'method2', 'method3'], arg1, arg2)

Answer 3

If you are using new-style object:

methods = ('method1','method2','method3')
for method in methods:
    try:
        b = a.__getattribute__(method)
    except AttributeError:
        continue
    else:
        b()
        break
else:
    # re-raise the AttributeError if nothing has worked
    raise AttributeError

Of course, if you aren't using a new-style object, you may try __dict__ instead of __getattribute__.

EDIT: This code might prove to be a screaming mess. If __getattribute__ or __dict__ is not found, take a wild guess what kind of error is raised.

Answer 4

A slight change to the second looks pretty nice and simple. I really doubt you'll notice any performance difference between the two, and this is a bit nicer than a nested try/excepts

def something(a):
    for methodname in ['method1', 'method2', 'method3']:
        try:
            m = getattr(a, methodname)
        except AttributeError:
            pass
        else:
            return m()
    raise AttributeError

The other very readable way is to do..

def something(a):
    try:
        return a.method1()
    except:
        pass

    try:
        return a.method2()
    except:
        pass

    try:
        return a.method3()
    except:
        pass

    raise AttributeError

While long, it's very obvious what the function is doing.. Performance really shouldn't be an issue (if a few try/except statements slow your script down noticeably, there is probably a bigger issue with the script structure)