Proper way to handle errors in a module

[Roy Smith]

In article <mailman.1415.1305134998.9059.python-list at python.org>,
 Andrew Berg <bahamutzero8825 at gmail.com> wrote:

> I'm a bit new to programming outside of shell scripts (and I'm no expert
> there), so I was wondering what is considered the best way to handle
> errors when writing a module. Do I just let exceptions go and raise
> custom exceptions for errors that don't trigger a standard one? Have the
> function/method return nothing or a default value and show an error
> message? I'm sure there's not a clear-cut answer, but I was just
> wondering what most developers would expect a module to do in certain
> situations.

In general, raise an exception when there is no valid value that can be 
returned from a function.  Sometimes it makes more sense to return 0, 
None, an empty list, etc.  So:

count_vowels("banana") ==> 3
count_vowels("xyzzy") ==> 0  # counting "y" as not a vowel
count_vowels("") ==> 0
count_vowels(None) ==> raises TypeError

You want to raise specific errors.  Let's say you've got a function like 
this:

def airspeed(swallow):
   speeds = {"european": 42,
             "african", 196}
   return speeds[swallow]

If somebody passes an invalid string, it will raise KeyError as written.  
Better to do something like:

def airspeed(swallow):
   speeds = {"european": 42,
             "african", 196}
   try:
       return speeds[swallow]
   except KeyError:
        raise UnknownBirdError(swallow)

This lets somebody catch UnknownBirdError at some higher level and do 
something useful.  If you let KeyError escape, that's a lot harder to 
deal with because it could come from almost anywhere.