feat(ast, ast_macros): apply stable repr to all #[ast] enums (#4373)

closes #4296

Cargo.lock

@@ -1355,6 +1355,11 @@ dependencies = [
 [[package]]
 name = "oxc_ast_macros"
 version = "0.23.0"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
 
 [[package]]
 name = "oxc_benchmark"

crates/oxc_ast/src/ast/js.rs

@@ -52,7 +52,6 @@ inherit_variants! {
 ///
 /// [`ast` module docs]: `super`
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged)]
@@ -269,7 +268,6 @@ inherit_variants! {
 ///
 /// [`ast` module docs]: `super`
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize))]
 #[serde(untagged)]
@@ -359,7 +357,6 @@ inherit_variants! {
 ///
 /// [`ast` module docs]: `super`
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged)]
@@ -436,7 +433,6 @@ pub struct TemplateElementValue<'a> {
 
 /// <https://tc39.es/ecma262/#prod-MemberExpression>
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged)]
@@ -597,7 +593,6 @@ inherit_variants! {
 ///
 /// [`ast` module docs]: `super`
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged)]
@@ -706,7 +701,6 @@ inherit_variants! {
 ///
 /// [`ast` module docs]: `super`
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged)]
@@ -725,7 +719,6 @@ inherit_variants! {
 ///
 /// [`ast` module docs]: `super`
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged)]
@@ -780,7 +773,6 @@ macro_rules! match_simple_assignment_target {
 pub use match_simple_assignment_target;
 
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged)]
@@ -846,7 +838,6 @@ inherit_variants! {
 ///
 /// [`ast` module docs]: `super`
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged)]
@@ -949,7 +940,6 @@ inherit_variants! {
 ///
 /// [`ast` module docs]: `super`
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged)]
@@ -979,7 +969,6 @@ inherit_variants! {
 ///
 /// [`ast` module docs]: `super`
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged)]
@@ -1050,7 +1039,6 @@ pub struct BlockStatement<'a> {
 
 /// Declarations and the Variable Statement
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged)]
@@ -1217,7 +1205,6 @@ inherit_variants! {
 ///
 /// [`ast` module docs]: `super`
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged)]
@@ -1251,7 +1238,6 @@ inherit_variants! {
 ///
 /// [`ast` module docs]: `super`
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged)]
@@ -2000,7 +1986,6 @@ pub struct StaticBlock<'a> {
 /// export as namespace d;    // TSNamespaceExportDeclaration
 /// ```
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged)]
@@ -2303,7 +2288,6 @@ inherit_variants! {
 ///
 /// [`ast` module docs]: `super`
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged)]

crates/oxc_ast/src/ast/jsx.rs

@@ -240,7 +240,6 @@ inherit_variants! {
 ///
 /// [`ast` module docs]: `super`
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged)]

crates/oxc_ast/src/ast/macros.rs

@@ -20,7 +20,7 @@
 ///
 /// ```
 /// inherit_variants! {
-///     #[repr(C, u8)]
+///     #[ast]
 ///     enum Statement<'a> {
 ///         pub enum Statement<'a> {
 ///             BlockStatement(Box<'a, BlockStatement<'a>>) = 0,
@@ -35,7 +35,7 @@
 /// expands to:
 ///
 /// ```
-/// #[repr(C, u8)]
+/// #[ast]
 /// enum Statement<'a> {
 ///     pub enum Statement<'a> {
 ///         BlockStatement(Box<'a, BlockStatement<'a>>) = 0,

crates/oxc_ast/src/ast/ts.rs

@@ -108,7 +108,6 @@ inherit_variants! {
 ///
 /// [`ast` module docs]: `super`
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged)]
@@ -191,7 +190,6 @@ pub enum TSLiteral<'a> {
 /// This is the root-level type for TypeScript types, kind of like [`Expression`] is for
 /// expressions.
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged, rename_all = "camelCase")]
@@ -483,7 +481,6 @@ inherit_variants! {
 ///
 /// [`ast` module docs]: `super`
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged, rename_all = "camelCase")]
@@ -642,7 +639,6 @@ pub struct TSTypeReference<'a> {
 ///     IdentifierReference
 ///     NamespaceName . IdentifierReference
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged)]
@@ -1035,7 +1031,6 @@ inherit_variants! {
 ///
 /// [`ast` module docs]: `super`
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged)]
@@ -1208,7 +1203,6 @@ inherit_variants! {
 ///
 /// [`ast` module docs]: `super`
 #[ast(visit)]
-#[repr(C, u8)]
 #[derive(Debug, Hash)]
 #[cfg_attr(feature = "serialize", derive(Serialize, Tsify))]
 #[serde(untagged, rename_all = "camelCase")]

crates/oxc_ast_macros/Cargo.toml

@@ -19,3 +19,8 @@ workspace = true
 [lib]
 proc-macro = true
 doctest    = false
+
+[dependencies]
+quote = { workspace = true }
+syn = { workspace = true, features = ["full"] }
+proc-macro2 = { workspace = true }

crates/oxc_ast_macros/src/lib.rs

@@ -1,30 +1,54 @@
 use proc_macro::TokenStream;
-use std::str::FromStr;
+use proc_macro2::TokenStream as TokenStream2;
+use quote::quote;
 
-/// Attach to AST node type (struct or enum), to signal to codegen to create visitor for this type.
+/// returns `#[repr(C, u8)]` if `enum_` has any non-unit variant,
+/// Otherwise it would return `#[repr(u8)]`.
+fn enum_repr(enum_: &syn::ItemEnum) -> TokenStream2 {
+    if enum_.variants.iter().any(|var| !matches!(var.fields, syn::Fields::Unit)) {
+        quote!(#[repr(C, u8)])
+    } else {
+        quote!(#[repr(u8)])
+    }
+}
+
+/// This attribute serves two purposes,
+/// First, it is a marker for our codegen to detect AST types. Furthermore.
+/// It is also a lightweight macro; All of its computation is cached and
+/// it only applies the following changes without any complex operation:
 ///
-/// Macro's role is not to generate code - it's purely a means to communicate information to the codegen.
+/// * Prepend `#[repr(C, u8)]` to fieldful enums e.g. `enum E { X: u32, Y: u8 }`
+/// * Prepend `#[repr(u8)]` to unit (fieldless) enums e.g. `enum E { X, Y, Z, }`
+/// * Prepend `#[derive(oxc_ast_macros::Ast)]`
 ///
-/// Only thing macro does is add `#[derive(Ast)]` to the item.
-/// Deriving `Ast` does nothing, but supports `#[scope]`, `#[visit]`, and other attrs on struct fields.
-/// These "helper" attributes are also signals to the codegen, and do nothing in themselves.
-///
-/// This is a workaround for Rust not supporting helper attributes for `proc_macro_attribute` macros,
-/// so we need to use a derive macro to get that support.
-///
-/// Use native Rust `TokenStream`, to avoid dependency on slow-compiling crates like `syn` and `quote`.
 #[proc_macro_attribute]
 #[allow(clippy::missing_panics_doc)]
 pub fn ast(_args: TokenStream, input: TokenStream) -> TokenStream {
-    let mut stream = TokenStream::from_str("#[derive(::oxc_ast_macros::Ast)]").unwrap();
-    stream.extend(input);
-    stream
+    let input = syn::parse_macro_input!(input as syn::Item);
+
+    let repr = match input {
+        syn::Item::Enum(ref enum_) => enum_repr(enum_),
+        // In future, we'll add `#[repr(C)]` to structs, but at present this is disabled
+        syn::Item::Struct(_) => TokenStream2::default(),
+
+        _ => {
+            unreachable!()
+        }
+    };
+
+    let expanded = quote! {
+        #[derive(::oxc_ast_macros::Ast)]
+        #repr
+        #input
+    };
+    TokenStream::from(expanded)
 }
 
 /// Dummy derive macro for a non-existent trait `Ast`.
 ///
 /// Does not generate any code.
 /// Only purpose is to allow using `#[scope]`, `#[visit]`, and other attrs in the AST node type defs.
+/// These "marker" attributes are used in codegen.
 #[proc_macro_derive(Ast, attributes(scope, visit, span, serde, tsify))]
 pub fn ast_derive(_item: TokenStream) -> TokenStream {
     TokenStream::new()