Adjust enum variant spans to exclude any explicit discriminant

[ytmimi]

Fixes #5686

For reference, explicit discriminants were proposed in RFC-2363, and I believe the feature is stabilized.

ast::Variant spans extend to include explicit discriminants when they are present.

Now we'll adjust the span of enum variants to exclude any explicit discriminant.

[calebcartwright]

Would suggest including the search needle in a few other pathological positions, e.g. within a comment or an attribute (assuming the latter is at least parseable, even if invalid syntactically), to make sure that span_after will suffice vs. something more positionally explicit span_after_last

[ytmimi]

Just want to make sure I'm understanding you correctly. Are you referring to a scenario like this:

enum Animal {
    // Closing tuple struct needle --------|
    //                                     v
    Dog(/* Coment with closing parenthesis ) */) = 1,
    Cat(/* Coment with */ #[hello(world)] String) = 2,
    //                                 ^
    // Closing tuple struct needle ----|

}

[calebcartwright]

Just want to make sure I'm understanding you correctly. Are you referring to a scenario like this:

enum Animal {
    // Closing tuple struct needle --------|
    //                                     v
    Dog(/* Coment with closing parenthesis ) */) = 1,
    Cat(/* Coment with */ #[hello(world)] String) = 2,
    //                                 ^
    // Closing tuple struct needle ----|

}

Yup exactly, anywhere some sneaky needles could reside in a parseable position

[ytmimi]

@calebcartwright I made some updates to the PR. A combination of shrinking the variant.span to exclude the discriminant, before using span_after_last to pinpoint the end of the variant. I decided to move the span calculation to its own function since I think we'll be able to reuse it when fixing #5713

[calebcartwright]

lol 🔥