Add circular_tuple_windows

[ed-bassett]

As described in #349, I think circular_tuple_windows would be broadly useful enough to be added to Itertools. It relies on tuple_windows and cloning, so I'm sure it could not be included in the standard library.

I'm not sure if this is the best way to implement the function, but I thought I would give it a shot so that at the very least you'd have something functional to work from.

[jswrenn]

Are there any issues with eliminating the ExactSizeIterator bound?

[ed-bassett]

I added ExactSizeIterator so that the call to len() on line 198 on in tuple_impl.rs would be performant. Using tuple_windows(iter.cycle()).take(len) was a neat way of getting the extra elements wrapped onto the end, but if removing the ExactSizeIterator constraint is preferable we could potentially go with something like the following, at the expense of a slight reduction in readability:

pub struct CircularTupleWindows<I, T: Clone>
    where I: Iterator<Item = T::Item> + Clone,
          T: TupleCollect + Clone,
{
    iter: TupleWindows<Chain<Chain<IntoIter<<T as TupleCollect>::Item>, I>, IntoIter<<T as TupleCollect>::Item>>, T>,
}

pub fn circular_tuple_windows<I, T>(mut iter: I) -> CircularTupleWindows<I, T>
    where I: Iterator<Item = T::Item> + Clone,
          T: TupleCollect + Clone,
          T::Item: Clone,
{
    let first_elements = iter.by_ref().take(<T as TupleCollect>::num_items() - 1).collect::<Vec<_>>();
    let iter = tuple_windows(first_elements.clone().into_iter().chain(iter).chain(first_elements.into_iter()));

    CircularTupleWindows {
        iter: iter,
    }
}

impl<I, T> Iterator for CircularTupleWindows<I, T>
    where I: Iterator<Item = T::Item> + Clone,
          T: TupleCollect + Clone,
          T::Item: Clone,
{
    type Item = T;

    fn next(&mut self) -> Option<T> {
        self.iter.next()
    }
}

what do you think?

[Boscop]

Any update on this? :)

[ed-bassett]

Any update on this? :)

Just waiting on feedback from @jswrenn on which way to go regarding the ExactSizeIterator bound

[jswrenn]

Sorry for dropping the ball on this!

I think I can merge this as-is. There's no reason we can't drop the bound later if it's unduly prohibitive.

bors r+

[ed-bassett]

Any chance I can squash in the fixup before it's merged? I was holding off while it was waiting on review, but I don't really know the process around getting things merged

[jswrenn]

Sure! Just let me know when you're ready.

[ed-bassett]

Good to go

[jswrenn]

bors r+

[bors]

Build succeeded

• continuous-integration/travis-ci/push