itertools.cycle(iterable) vs while True

Question

I was recently asked to do this task (school) :

Write a loop generator, which takes as parameter a finite iterable, and generates in infinite loop the iterable

So I did :

import itertools
def loop(l):
    for eleme‍‌​‍‌nt in itertools.cycle(l):
        yield element

and one of my classmate did :

def loop(l):
    while True:
​‍         for element in l:
            yield element

I'd like to know what are the main differences between the two and if there is a more "pythonic" way to write something simple as this.

Answer 1

I'd like to know what are the main differences between the two ...

The main difference is that these code snippets are not entirely equivalent in behaviour. Using cycle, you can accept and repeat an exhaustible iterator, whereas the while loop can not.

>>> def gen():
...     yield 1
...     yield 2
...     
>>> def loop_it(it):
...     for element in itertools.cycle(it):
...         yield element
...         
>>> g = loop_it(gen())
>>> next(g)
1
>>> next(g)
2
>>> next(g)
1

Contrast:

>>> def loop_while(it):
...     while True:
...         for element in it:
...             yield element
...             
>>> g = loop_while(gen())
>>> next(g)
1
>>> next(g)
2
>>> next(g)
# ... hangs forever

... and if there is a more "pythonic" way to write something simple as this

My recommendation is for the while loop, exactly as written. If you are asked in a school task to write the generator, it will likely be frowned upon to use "one prepared earlier" from itertools. The while loop is also more Pythonic. An "itertoolsthonic" approach would instead be using the cycle directly, like this:

items = itertools.cycle(l)
# do something with `items`

There is no point to write the extra scaffolding of a generator function and for loop yielding from an itertools.cycle - since the cycle is already an iterator, you would just use it directly.

Answer 2

You're right, itertools.cycle isn't of great interest here over the classical while True loop.

On the other hand, it's of great help in infinite generator comprehensions, where you cannot create an infinite loop because it only allows for, tests and function calls. Example to generate squared value of a list indefinitely:

generator = (x*x for x in itertools.cycle(l))

Of course you could always shorten your current code with:

def loop(l):
    yield from itertools.cycle(l)

or even:

loop = itertools.cycle

Answer 3

As I see it, the purpose of itertools.cycle is that you wouldn't have to write it yourself at all :)

The most pythonic way would be to simply write loop = itertools.cycle.

If you truly need to write it as a generator for your school assignment, the second form would probably perform faster, because the first form basically does the same but also has additional overhead from re-yielding the values from cycle in your generator.