itertools.islice and slice objects

[Steven Bethard]

Is there a reason that itertools.islice doesn't support None arguments
for start and step? This would be handy for use with slice objects:

>>> r = range(20)
>>> s1 = slice(2, 10, 2)
>>> s2 = slice(2, 10)
>>> s3 = slice(10)
>>> list(itertools.islice(r, s1.start, s1.stop, s1.step)) [2, 4, 6, 8]
>>> list(itertools.islice(r, s2.start, s2.stop, s2.step))

Traceback (most recent call last):

>>> list(itertools.islice(r, s2.start, s2.stop, s2.step or 1)) [2, 3, 4, 5, 6, 7, 8, 9]
>>> list(itertools.islice(r, s3.start, s3.stop, s3.step))

Traceback (most recent call last):
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9]

Of course, I can get (start, stop, step) tuples with slice.indices, but
only if I know the length of the iterable, which kinda defeats the
purpose of using itertools...

>>> list(itertools.islice(r, *s2.indices(10))) [2, 3, 4, 5, 6, 7, 8, 9]
>>> list(itertools.islice(r, *s3.indices(10)))

[0, 1, 2, 3, 4, 5, 6, 7, 8, 9]

Steve

 

[Raymond Hettinger]

[Steven Bethard]

Is there a reason that itertools.islice doesn't support None arguments
for start and step? This would be handy for use with slice objects

Offhand, this seems like a reasonable idea. Please file a Py2.5 feature request
and assign to me.

Of course, I can get (start, stop, step) tuples with slice.indices, but
only if I know the length of the iterable, which kinda defeats the
purpose of using itertools...

The main purpose is to extract elements one at time instead of all at once.
This memory friendly approach results in better scalability.


Raymond Hettinger