Add `std::ops::range::Bound::map_all`

[IoaNNUwU]

Proposal

Problem statement

std and user crates have a lot of places with this pattern:

let start = match range.start_bound() {
    Bound::Included(n) => n,
    Bound::Excluded(n) => n + 1,
    Bound::Unbounded => 0,
}

Solution sketch

Add std::ops::range::Bound::map_all:

    /// Maps a `Bound<T>` to a `R` by applying a selected function `FnOnce(T) -> R` 
    /// to the matched contained value.
    pub fn map_all<R, I: FnOnce(T) -> R, E: FnOnce(T) -> R>(
        self,
        included: I,
        excluded: E,
        unbounded: R,
    ) -> R {
        match self {
            Bound::Included(x) => included(x),
            Bound::Excluded(x) => excluded(x),
            Bound::Unbounded => unbounded,
        }
    }

This will allow:

use std::range::Bound;

let start_bound: Bound<usize> = Bound::Included(0);
let start = start_bound.map_all(|i| i, |e| e + 1, 0);
assert_eq!(start, 0);

let start_bound: Bound<usize> = Bound::Excluded(0);
let start = start_bound.map_all(|i| i, |e| e + 1, 0);
assert_eq!(start, 1);

let start_bound: Bound<usize> = Bound::Unbounded;
let start = start_bound.map_all(|i| i, |e| e + 1, 0);
assert_eq!(start, 0);

Alternatives

• alternative name map_idx

This function makes order of arguments less explicit. It may cause more bugs.

This feature is very small, niche and doesn't add much, but I think its reasonable to have something like this.

[BurntSushi]

Where is n defined? What is the signature of the method you are proposing?

[IoaNNUwU]

Sorry, I didn't put enough thought to this😮 Edited the proposal, added example function signature.

[BurntSushi]

I'm not sure this pattern is really common enough to warrant a special combinator for it. In particular, I think it would be difficult to know---when reading the code---which closure is for the included case and which for the excluded case.

[scottmcm]

Any time I see a multi-closure API where the closures are disjoint, I think it probably shouldn't exist. (There's no way to tell things like move checking that the closures are disjoint, so these APIs work worse than one might wish.) And this one makes it even worse because it's always the same argument type for both, so caller-side it'd be even easier to mix up which is which than usual.

Just write the match.

---

Also, that particular match sounds like it, insteads, wants something more like https://std.noratrieb.dev/x86_64-unknown-linux-gnu/std/slice/fn.range.html. Converting the impl RangeBounds to a Range is useful, but I don't think map_all is.