refactor(syntax): add #[ast] attr to semantic ID types (#8867)

Add `#[ast]` attr to `ScopeId`, `SymbolId` and `ReferenceId` types. This makes them an "official" part of the AST, and removes the hacky "special case" logic from `oxc_ast_tools` which was previously working around that they weren't registered as part of the AST.

This uses the `#[ast(foreign = ...)]` feature introduced in #8866 to inform `oxc_ast_tools` about the foreign `NonMaxU32` type, enabling the ID types to be parsed and understood.

.github/.generated_ast_watch_list.yml

@@ -32,7 +32,11 @@ src:
   - 'crates/oxc_syntax/src/generated/derive_clone_in.rs'
   - 'crates/oxc_syntax/src/generated/derive_content_eq.rs'
   - 'crates/oxc_syntax/src/generated/derive_estree.rs'
+  - 'crates/oxc_syntax/src/lib.rs'
   - 'crates/oxc_syntax/src/number.rs'
   - 'crates/oxc_syntax/src/operator.rs'
+  - 'crates/oxc_syntax/src/reference.rs'
+  - 'crates/oxc_syntax/src/scope.rs'
+  - 'crates/oxc_syntax/src/symbol.rs'
   - 'npm/oxc-types/types.d.ts'
   - 'tasks/ast_tools/src/**'

crates/oxc_ast/Cargo.toml

@@ -28,6 +28,7 @@ oxc_syntax = { workspace = true }
 
 bitflags = { workspace = true }
 cow-utils = { workspace = true }
+nonmax = { workspace = true }
 num-bigint = { workspace = true }
 num-traits = { workspace = true }
 

crates/oxc_ast/src/generated/assert_layouts.rs

@@ -3,7 +3,10 @@
 
 use std::mem::{align_of, offset_of, size_of};
 
+use nonmax::NonMaxU32;
+
 use oxc_regular_expression::ast::*;
+use oxc_syntax::{reference::ReferenceId, scope::ScopeId, symbol::SymbolId};
 
 use crate::ast::*;
 
@@ -1393,6 +1396,9 @@ const _: () = {
     assert!(offset_of!(Comment, preceded_by_newline) == 14);
     assert!(offset_of!(Comment, followed_by_newline) == 15);
 
+    assert!(size_of::<NonMaxU32>() == 4);
+    assert!(align_of::<NonMaxU32>() == 4);
+
     assert!(size_of::<NumberBase>() == 1);
     assert!(align_of::<NumberBase>() == 1);
 
@@ -1414,6 +1420,15 @@ const _: () = {
     assert!(size_of::<UpdateOperator>() == 1);
     assert!(align_of::<UpdateOperator>() == 1);
 
+    assert!(size_of::<ScopeId>() == 4);
+    assert!(align_of::<ScopeId>() == 4);
+
+    assert!(size_of::<SymbolId>() == 4);
+    assert!(align_of::<SymbolId>() == 4);
+
+    assert!(size_of::<ReferenceId>() == 4);
+    assert!(align_of::<ReferenceId>() == 4);
+
     assert!(size_of::<Span>() == 8);
     assert!(align_of::<Span>() == 8);
     assert!(offset_of!(Span, start) == 0);
@@ -2959,6 +2974,9 @@ const _: () = {
     assert!(offset_of!(Comment, preceded_by_newline) == 14);
     assert!(offset_of!(Comment, followed_by_newline) == 15);
 
+    assert!(size_of::<NonMaxU32>() == 4);
+    assert!(align_of::<NonMaxU32>() == 4);
+
     assert!(size_of::<NumberBase>() == 1);
     assert!(align_of::<NumberBase>() == 1);
 
@@ -2980,6 +2998,15 @@ const _: () = {
     assert!(size_of::<UpdateOperator>() == 1);
     assert!(align_of::<UpdateOperator>() == 1);
 
+    assert!(size_of::<ScopeId>() == 4);
+    assert!(align_of::<ScopeId>() == 4);
+
+    assert!(size_of::<SymbolId>() == 4);
+    assert!(align_of::<SymbolId>() == 4);
+
+    assert!(size_of::<ReferenceId>() == 4);
+    assert!(align_of::<ReferenceId>() == 4);
+
     assert!(size_of::<Span>() == 8);
     assert!(align_of::<Span>() == 4);
     assert!(offset_of!(Span, start) == 0);

crates/oxc_syntax/src/lib.rs

@@ -1,5 +1,10 @@
 //! Common code for JavaScript Syntax
 #![warn(missing_docs)]
+
+use std::num::NonZeroU32;
+
+use oxc_ast_macros::ast;
+
 pub mod class;
 pub mod es_target;
 pub mod identifier;
@@ -19,3 +24,8 @@ mod generated {
     #[cfg(feature = "serialize")]
     mod derive_estree;
 }
+
+/// Dummy type to communicate the content of `nonmax::NonMaxU32` to `oxc_ast_tools`.
+#[ast(foreign = NonMaxU32)]
+#[expect(dead_code)]
+struct NonMaxU32Alias(NonZeroU32);

crates/oxc_syntax/src/reference.rs

@@ -8,6 +8,9 @@ use serde::{Serialize, Serializer};
 
 use crate::{node::NodeId, symbol::SymbolId};
 
+use oxc_ast_macros::ast;
+
+#[ast]
 #[derive(Debug, Clone, Copy, Eq, PartialEq, Ord, PartialOrd, Hash)]
 pub struct ReferenceId(NonMaxU32);
 

crates/oxc_syntax/src/scope.rs

@@ -5,6 +5,9 @@ use oxc_index::Idx;
 #[cfg(feature = "serialize")]
 use serde::{Serialize, Serializer};
 
+use oxc_ast_macros::ast;
+
+#[ast]
 #[derive(Debug, Clone, Copy, Eq, PartialEq, Ord, PartialOrd, Hash)]
 pub struct ScopeId(NonMaxU32);
 

crates/oxc_syntax/src/symbol.rs

@@ -5,6 +5,9 @@ use oxc_index::Idx;
 #[cfg(feature = "serialize")]
 use serde::{Serialize, Serializer};
 
+use oxc_ast_macros::ast;
+
+#[ast]
 #[derive(Debug, Clone, Copy, Eq, PartialEq, Ord, PartialOrd, Hash)]
 pub struct SymbolId(NonMaxU32);
 

tasks/ast_tools/src/generators/assert_layouts.rs

@@ -46,8 +46,12 @@ impl Generator for AssertLayouts {
         let output = quote! {
             use std::mem::{align_of, offset_of, size_of};
 
+            ///@@line_break
+            use nonmax::NonMaxU32;
+
             ///@@line_break
             use oxc_regular_expression::ast::*;
+            use oxc_syntax::{reference::ReferenceId, scope::ScopeId, symbol::SymbolId};
 
             ///@@line_break
             use crate::ast::*;
@@ -337,8 +341,6 @@ fn calculate_layout_for_cell(type_id: TypeId, schema: &mut Schema) -> Layout {
 ///
 /// Primitives have varying layouts. Some have niches, most don't.
 fn calculate_layout_for_primitive(primitive_def: &PrimitiveDef) -> Layout {
-    // `ScopeId`, `SymbolId` and `ReferenceId` are a `NonZeroU32`, with a niche for 0
-    let semantic_id_layout = Layout::from_size_align_niche(4, 4, Niche::new(0, 4, true, 1));
     // `&str` and `Atom` are a `NonNull` pointer + `usize` pair. Niche for 0 on the pointer field
     let str_layout = Layout {
         layout_64: PlatformLayout::from_size_align_niche(16, 8, Niche::new(0, 8, true, 1)),
@@ -394,9 +396,6 @@ fn calculate_layout_for_primitive(primitive_def: &PrimitiveDef) -> Layout {
             panic!("Cannot calculate alignment for `NonZeroI128`. It differs depending on Rust version.")
         }
         "NonZeroIsize" => non_zero_usize_layout,
-        "ScopeId" => semantic_id_layout.clone(),
-        "SymbolId" => semantic_id_layout.clone(),
-        "ReferenceId" => semantic_id_layout,
         "PointerAlign" => Layout {
             layout_64: PlatformLayout::from_size_align(0, 8),
             layout_32: PlatformLayout::from_size_align(0, 4),

tasks/ast_tools/src/main.rs

@@ -198,8 +198,12 @@ static SOURCE_PATHS: &[&str] = &[
     "crates/oxc_ast/src/ast/jsx.rs",
     "crates/oxc_ast/src/ast/ts.rs",
     "crates/oxc_ast/src/ast/comment.rs",
+    "crates/oxc_syntax/src/lib.rs",
     "crates/oxc_syntax/src/number.rs",
     "crates/oxc_syntax/src/operator.rs",
+    "crates/oxc_syntax/src/scope.rs",
+    "crates/oxc_syntax/src/symbol.rs",
+    "crates/oxc_syntax/src/reference.rs",
     "crates/oxc_span/src/span/types.rs",
     "crates/oxc_span/src/source_type/mod.rs",
     "crates/oxc_regular_expression/src/ast.rs",

tasks/ast_tools/src/parse/parse.rs

@@ -135,9 +135,6 @@ impl<'c> Parser<'c> {
             "NonZeroIsize" => primitive("NonZeroIsize"),
             "&str" => primitive("&str"),
             "Atom" => primitive("Atom"),
-            "ScopeId" => primitive("ScopeId"),
-            "SymbolId" => primitive("SymbolId"),
-            "ReferenceId" => primitive("ReferenceId"),
             // TODO: Remove the need for this by adding
             // `#[cfg_attr(target_pointer_width = "64", repr(align(8)))]` to all AST types
             "PointerAlign" => primitive("PointerAlign"),