fix: unnested_or_patterns suggests wrongly in let

[profetia]

Closes #9952

changelog: [unnested_or_patterns] fix wrong suggestion in let

[rustbot]

r? @y21

rustbot has assigned @y21.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[samueltardieu]

This is one way to fix it up, but there might be a simpler one. By fixing the visitor in insert_necessary_parens() to not remove parens (if any) on the outermost pattern, you get the same result without having to introduce PatternSource. This could be something like (with minimal diffs to the existing visitor):

fn remove_all_parens(pat: &mut P<Pat>) {
    #[derive(Default)]
    struct Visitor {
        is_inner: bool,
    }
    impl MutVisitor for Visitor {
        fn visit_pat(&mut self, pat: &mut P<Pat>) {
            let is_inner = mem::replace(&mut self.is_inner, true);
            walk_pat(self, pat);
            pat.kind = match &mut pat.kind {
                Paren(i) if is_inner => mem::replace(&mut i.kind, Wild),
                _ => return,
            };
        }
    }
    Visitor::default().visit_pat(pat);
}

The only difference in output with what you are proposing is that this wouldn't "fix" the outermost parens if the user chose to use them. For example, this would transform:

if let (0 | (1 | 2)) = …

into

if let (0 | 1 | 2) = …

whereas your change also drops the outermost parens. Even if dropping them is correct, this is not what the lint advertises, and this might be considered redundant with rustc's unused_parens lint.

[profetia]

Good advice! Thanks!

[flip1995]

Thanks, just a NIT.

[y21]

Also looks good to me, just one small suggestion