Fix handling of item names for HIR

[jyn514]

• Handle variants, fields, macros in Node::ident()
• Handle the crate root in opt_item_name
• Rewrite item_name in terms of opt_item_name

I need this for both #77820 and #78082, so splitting it out into a separate PR so it can land early.

[rust-highfive]

r? @varkor

(rust_highfive has picked a reviewer for you, use r? to override)

[jyn514]

I don't understand this failure:

---- [incremental] incremental/spans_in_type_debuginfo.rs stdout ----

error in revision `rpass2`: compilation failed!
status: exit code: 1
command: "/checkout/obj/build/x86_64-unknown-linux-gnu/stage2/bin/rustc" "/checkout/src/test/incremental/spans_in_type_debuginfo.rs" "-Zthreads=1" "--target=x86_64-unknown-linux-gnu" "--cfg" "rpass2" "-C" "incremental=/checkout/obj/build/x86_64-unknown-linux-gnu/test/incremental/spans_in_type_debuginfo/spans_in_type_debuginfo.inc" "-Z" "incremental-verify-ich" "--error-format" "json" "-Zui-testing" "-Zdeduplicate-diagnostics=no" "-C" "prefer-dynamic" "-o" "/checkout/obj/build/x86_64-unknown-linux-gnu/test/incremental/spans_in_type_debuginfo/a" "-Crpath" "-O" "-Zunstable-options" "-Lnative=/checkout/obj/build/x86_64-unknown-linux-gnu/native/rust-test-helpers" "-Z" "query-dep-graph" "-g" "-L" "/checkout/obj/build/x86_64-unknown-linux-gnu/test/incremental/spans_in_type_debuginfo/auxiliary"
stdout:
------------------------------------------

------------------------------------------
stderr:
------------------------------------------
warning: variant is never constructed: `B`
  --> /checkout/src/test/incremental/spans_in_type_debuginfo.rs:39:9
   |
LL |         B(u32),
   |         ^^^^^^
   |
   = note: `#[warn(dead_code)]` on by default

error: CGU-reuse for `spans_in_type_debuginfo-enums` is `No` but should be at least `PreLto`
  --> /checkout/src/test/incremental/spans_in_type_debuginfo.rs:9:1
   |
LL | #![rustc_partition_reused(module="spans_in_type_debuginfo-enums", cfg="rpass2")]
   | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

error: CGU-reuse for `spans_in_type_debuginfo-structs` is `No` but should be at least `PreLto`
  --> /checkout/src/test/incremental/spans_in_type_debuginfo.rs:8:1
   |
LL | #![rustc_partition_reused(module="spans_in_type_debuginfo-structs", cfg="rpass2")]
   | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
error: aborting due to 2 previous errors; 1 warning emitted

------------------------------------------

Is the test maybe wrong and depending that both the names are None?

[jyn514]

Oh - maybe the issue is that item_name is a query and can't depend on spans? But it removes the span from opt_item_name ... I'm not sure what's going wrong.

[jyn514]

ping @varkor - do you know what's going wrong here?

[jyn514]

Ok I think I see the issue - if item_name is called, we need to use the DefPath first, which doesn't have an associated span. But if opt_item_name is called we need to use hir().get_if_local() so the span will be right. I'll write that up.

[jyn514]

Ugh - this still doesn't do what I want because it still panics if you call opt_item_name on the crate root.

---- [rustdoc] rustdoc/without-redirect.rs stdout ----

error: rustdoc failed!
status: exit code: 101
------------------------------------------
stderr:
------------------------------------------
error: internal compiler error: compiler/rustc_middle/src/hir/map/mod.rs:477:41: couldn't find hir id HirId { owner: DefId(0:0 ~ foo[8787]), local_id: 0 } in the HIR map

[jyn514]

Ok, tested this by rebasing #77820 on top and it finally works 🎉 now I only have the problems from #77820 to solve 😆

[jyn514]

ping @varkor - is this waiting on anything from me?

[varkor]

I thought you were waiting for #77820 to land if this is rebased on top of it?

[jyn514]

@varkor it's the other way around - #77820 is rebased on top of this PR and needs it to land.

[jyn514]

The thing is there's no tests for this outside of #77820, which is why I've been constantly rebasing it.

[varkor]

Ah, my mistake. If the extra line can be removed, that'd be preferable. In either case r=me after checking that.

[jyn514]

If CI passes this is good to go, otherwise I need to add back the call to item_name_from_hir.

[jyn514]

@bors r=varkor

[bors]

📌 Commit f60fd49 has been approved by varkor