Exceptions

Exceptions can be a powerful tool for you and for code development and usage, but can also be a drawback (they can have a huge impact on performance) when not being used properly.

Probably you are using exceptions even without noticing. For example, for lists raise a StopIteration exception when you iterate over them (e.g. in a for loop) and there are no more items left.

You should be aware when to use and when not use exceptions. Do not use exceptions for flow-control. Exceptions exist for exceptional situations: events that are not a part of normal execution.

No:

def is_multiple(x, y):
    if x % y == 0:
        return True
    else:
        raise Exception()

Yes:

def is_multiple(x, y):
    if y == 0:
        raise Exception()
    if x % y == 0:
        return True
    else:
        return False

If your code is going to capture exceptions that you would like to manage or to set a particular message to help users or developers what went wrong and where, a good practice is to define your own exception class.

User exceptions should typically derive from, at least, Exception:

class MyPackageError(Exception):
    pass

Most exceptions are defined with names that end in “Error”.

If you want to create a set of exceptions you can create a base class and derive from that:

class MyPackageError(Exception):
   pass

class UserInputError(MyPackage):
   pass

class ClusterError(MyPackage):
   pass

It is also useful to give a meaningful error message to your exceptions:

In [1]: class MyPackage(Exception):
   ...:     pass
   ...: 

In [2]: raise MyPackage('Exception raised')
---------------------------------------------------------------------------
MyPackage                                 Traceback (most recent call last)
<ipython-input-2-e2bcd5645bf9> in <module>
----> 1 raise MyPackage('Exception raised')

MyPackage: Exception raised

No:

def f():
    if something_goes_wrong:
        raise Exception()

Yes:

def f():
    if something_goes_wrong:
        raise Exception('This has occurred')

Handling exceptions

In Python exceptions are handled with the try-except clause.

• Avoid try-except whenever is possible

• Limit the try clause to the minimum possible lines

• Indicate the exception you want to capture

No:

def to_int(v):
    try:
        return int(v)
    except:
        None

Yes:

def to_int(v):
    try:
       v = int(v)
    except ValueError:
        return None
    else:
        return v