Fix #[ts(rename_all_fields = "...")] on enums containing tuple or unit variants

[escritorio-gustavo]

Goal

Using #[ts(rename_all_fields = "...")] on an enum containing tuple or unit variants would cause the error "rename_all is not applicable to newtype/tuple/unit structs"

Changes

Only add rename_all from rename_all_fields if variant.fields is Fields::Named(_)

Checklist

• I have followed the steps listed in the Contributing guide.
• If necessary, I have added documentation related to the changes made.
• I have added or updated the tests related to the changes made.

[NyxCode]

To summarize my understanding:
The problem is that when creating VariantAttrs in VariantAttr::new, we use the rename_all_fields of the enum as value for rename_all.
These VariantAttrs are then converted into StructAttrs, and when they reach newtype, we get the error.

There are a couple of thinks that stood out to me.
This basically duplicates VariantAttr::new, though we could fix that by re-writing VariantAttr::new to use VariantAttr::from_attrs internally.
Also, the match, deciding between VariantAttr::new and VariantAttr::from_attrs really doesn't feel great.

An alternative way of getting rid of the error would be to remove the check from newtype. It probably wouldn't be a big deal to just ignore it if it's there, to be honest.

If we wanted to keep the error, we could alternatively move these sanity checks, which we have in a lot of places, into the attribute parsing logic. But again, we'd then need to know what we're parsing (a unit struct?). So that'd probably be a larger refactor, and I'm not sure if it'd end up being nice.

So yeah, I'm not super satisfied with this solution, but I kinda don't see a better way of doing it right now.
If we're going with this, I think we should definetely add a comment to the match, and re-write VariantAttr::new to use VariantAttr::from_attrs to not have that logic duplicated.

[escritorio-gustavo]

This basically duplicates VariantAttr::new, though we could fix that by re-writing VariantAttr::new to use VariantAttr::from_attrs internally.

Agreed! I'll push a commit to do that

[NyxCode]

Alright, nice!
I'd be happy with merging this as-is. Maybe we'll find some nice way to clean all the "... is not compatible with ..." errors up at some point in the future.