Correctly handle #[doc(alias = "...")] attribute on inlined reexports

[GuillaumeGomez]

Fixes #152939.

During the doc attributing migration to the new API, this information got lost. At least now we have a test for it. :)

r? @lolbinarycat

[lolbinarycat]

1. there's a bunch of other fields being discarded here, do we have tests ensuring we handle stuff like search_unbox and masked properly with re-exports?
2. I kinda wish we could avoid .. here to force new fields to be handled explicitly, but at the same time there's so many fields so I understand that's not the most practical.
3. As for the test, I think the filename for the inner an outer library should be different, just for clarity. Personally I would just add -inner to the filename of the auxiliary library.

[GuillaumeGomez]

I agree with 1 and 2. As for 3, that's how we usually do: we give the same name as the test to the auxiliary one and rename it on import.

[GuillaumeGomez]

We already have a test for masked (tests/rustdoc-html/masked.rs). I added one for search_unbox. Remains the potential following ones:

• fake_variadic
• keyword
• attribute
• notable_trait

I'll open an issue once this PR is merged to check them and add a regression test if they don't have one already.

I also improved the code as you suggested (very good idea, love when code enforces that nothing is forgotten :3 ).

Does that sound ok to you @lolbinarycat ?

[lolbinarycat]

Everything looks good, addressing the rest of those in a followup sounds fine to me, just one final nit: reexport-search_unbox.js should have a comment explaining what it is testing.

Once that's fixed, you can r=me.

[GuillaumeGomez]

Arf, forgot to add it indeed. Fixing that right away.

[lolbinarycat]

@bors r+

[rust-bors]

📌 Commit ab9e1da has been approved by lolbinarycat

It is now in the queue for this repository.