[duped]

You almost never want to compute a substring by codepoint boundaries, which is why this operation isn't in the standard library.

You should be looking at this crate: https://docs.rs/unicode-segmentation/