Simon Forman wrote:
> I've got a function that I'd like to improve.
>
> It takes a list of lists and a "target" element, and it returns the set
> of the items in the lists that appear either before or after the target
> item. (Actually, it's a generator, and I use the set class outside of
> it to collect the unique items, but you get the idea. ;-) )
>
> data = [
> ['this', 'string', 'is', 'nice'],
> ['this', 'string', 'sucks'],
> ['string', 'not', 'good'],
> ['what', 'a', 'string']
> ]
>
> def f(target, ListOfLists):
> for N in ListOfLists:
> try:
> i = N.index(target)
> except ValueError:
> continue
>
> # item before target
> if i: yield N[i - 1]
>
> # item after target
> try:
> yield N[i + 1]
> except IndexError:
> pass
>
> print set(n for n in f('string', data))
>
> # Prints set(['this', 'not', 'is', 'sucks', 'a'])
>
>
> Now, this works and normally I'd be happy with this and not try to
> improve it unless I found that it was a bottleneck. However, in this
> case I know that when a try..except statement fails (i.e. executes the
> except part of the statement) it's "slow", *and* I know that the real
> list of lists will have many lists in it in which the target item does
> not appear.
>
> That means that the first try..except statement will be executing it's
> except clause more than half the time, and probably *much* more than
> half the time.
Nevermind, it turns out that each list in the list of lists will
*always* contain the target item, making this task much less
interesting.
~Simon
--
http://mail.python.org/mailman/listinfo/python-list
