feat(ast): add SymbolId and ReferenceId (#755)

Closes #510

crates/oxc_ast/src/ast/js.rs

@@ -1,9 +1,13 @@
-use std::fmt;
+use std::{cell::Cell, fmt, hash::Hash};
 
 use oxc_allocator::{Box, Vec};
 use oxc_span::{Atom, SourceType, Span};
-use oxc_syntax::operator::{
-    AssignmentOperator, BinaryOperator, LogicalOperator, UnaryOperator, UpdateOperator,
+use oxc_syntax::{
+    operator::{
+        AssignmentOperator, BinaryOperator, LogicalOperator, UnaryOperator, UpdateOperator,
+    },
+    reference::ReferenceId,
+    symbol::SymbolId,
 };
 #[cfg(feature = "serde")]
 use serde::Serialize;
@@ -254,21 +258,51 @@ pub struct IdentifierName {
 }
 
 /// Identifier Reference
-#[derive(Debug, Clone, Hash)]
+#[derive(Debug, Clone)]
 #[cfg_attr(feature = "serde", derive(Serialize), serde(tag = "type"))]
 pub struct IdentifierReference {
     #[cfg_attr(feature = "serde", serde(flatten))]
     pub span: Span,
     pub name: Atom,
+    #[cfg_attr(feature = "serde", serde(skip))]
+    pub reference_id: Cell<Option<ReferenceId>>,
+}
+
+impl Hash for IdentifierReference {
+    fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
+        self.span.hash(state);
+        self.name.hash(state);
+    }
+}
+
+impl IdentifierReference {
+    pub fn new(name: Atom, span: Span) -> Self {
+        Self { name, span, reference_id: Cell::default() }
+    }
 }
 
 /// Binding Identifier
-#[derive(Debug, Clone, Hash)]
+#[derive(Debug, Clone)]
 #[cfg_attr(feature = "serde", derive(Serialize), serde(tag = "type"))]
 pub struct BindingIdentifier {
     #[cfg_attr(feature = "serde", serde(flatten))]
     pub span: Span,
     pub name: Atom,
+    #[cfg_attr(feature = "serde", serde(skip))]
+    pub symbol_id: Cell<Option<SymbolId>>,
+}
+
+impl Hash for BindingIdentifier {
+    fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
+        self.span.hash(state);
+        self.name.hash(state);
+    }
+}
+
+impl BindingIdentifier {
+    pub fn new(name: Atom, span: Span) -> Self {
+        Self { name, span, symbol_id: Cell::default() }
+    }
 }
 
 /// Label Identifier

crates/oxc_ast_lower/src/lib.rs

@@ -371,6 +371,7 @@ impl<'a> AstLower<'a> {
         self.hir.with_statement(stmt.span, object, body)
     }
 
+    #[allow(clippy::too_many_lines)]
     fn lower_expression(&mut self, expr: &ast::Expression<'a>) -> hir::Expression<'a> {
         ensure_sufficient_stack(|| {
             match expr {
@@ -444,7 +445,7 @@ impl<'a> AstLower<'a> {
                 ast::Expression::JSXElement(elem) => {
                     // TODO: implement JSX
                     let ident = self.lower_identifier_reference(
-                        &ast::IdentifierReference { span: elem.span, name: "undefined".into() },
+                        &ast::IdentifierReference::new("undefined".into(), elem.span),
                         ReferenceFlag::Read,
                     );
                     self.hir.identifier_reference_expression(ident)
@@ -452,7 +453,7 @@ impl<'a> AstLower<'a> {
                 ast::Expression::JSXFragment(elem) => {
                     // TODO: implement JSX
                     let ident = self.lower_identifier_reference(
-                        &ast::IdentifierReference { span: elem.span, name: "undefined".into() },
+                        &ast::IdentifierReference::new("undefined".into(), elem.span),
                         ReferenceFlag::Read,
                     );
                     self.hir.identifier_reference_expression(ident)
@@ -850,7 +851,7 @@ impl<'a> AstLower<'a> {
             expr => {
                 // return undefined because this is invalid syntax
                 let ident = self.lower_identifier_reference(
-                    &ast::IdentifierReference { span: expr.span(), name: "undefined".into() },
+                    &ast::IdentifierReference::new("undefined".into(), expr.span()),
                     ReferenceFlag::Write,
                 );
                 self.hir.assignment_target_identifier(ident)

crates/oxc_parser/src/js/binding.rs

@@ -90,8 +90,7 @@ impl<'a> Parser<'a> {
             //       ^ BindingIdentifier
             if let PropertyKey::Identifier(ident) = &key {
                 shorthand = true;
-                let binding_identifier =
-                    BindingIdentifier { span: ident.span, name: ident.name.clone() };
+                let binding_identifier = BindingIdentifier::new(ident.name.clone(), ident.span);
                 let identifier = self.ast.binding_identifier(binding_identifier);
                 let left = self.ast.binding_pattern(identifier, None, false);
                 self.parse_initializer(span, left)?

crates/oxc_parser/src/js/expression.rs

@@ -1,3 +1,5 @@
+use std::cell::Cell;
+
 use oxc_allocator::Box;
 use oxc_ast::ast::*;
 use oxc_diagnostics::Result;
@@ -55,7 +57,7 @@ impl<'a> Parser<'a> {
         }
         let (span, name) = self.parse_identifier_kind(Kind::Ident);
         self.check_identifier(span, &name);
-        Ok(IdentifierReference { span, name })
+        Ok(IdentifierReference { span, name, reference_id: Cell::default() })
     }
 
     /// `BindingIdentifier` : Identifier
@@ -65,7 +67,7 @@ impl<'a> Parser<'a> {
         }
         let (span, name) = self.parse_identifier_kind(Kind::Ident);
         self.check_identifier(span, &name);
-        Ok(BindingIdentifier { span, name })
+        Ok(BindingIdentifier { span, name, symbol_id: Cell::default() })
     }
 
     pub(crate) fn parse_label_identifier(&mut self) -> Result<LabelIdentifier> {

crates/oxc_parser/src/js/function.rs

@@ -1,3 +1,5 @@
+use std::cell::Cell;
+
 use oxc_allocator::Box;
 use oxc_ast::{ast::*, AstBuilder};
 use oxc_diagnostics::Result;
@@ -332,7 +334,7 @@ impl<'a> Parser<'a> {
         let id = self.cur_kind().is_binding_identifier().then(|| {
             let (span, name) = self.parse_identifier_kind(Kind::Ident);
             self.check_identifier(span, &name);
-            BindingIdentifier { span, name }
+            BindingIdentifier { span, name, symbol_id: Cell::default() }
         });
         self.ctx = ctx;
 

crates/oxc_parser/src/js/grammar.rs

@@ -151,7 +151,8 @@ impl<'a> CoverGrammar<'a, ObjectProperty<'a>> for AssignmentTargetProperty<'a> {
         if property.shorthand {
             let binding = match property.key {
                 PropertyKey::Identifier(ident) => {
-                    IdentifierReference { span: ident.span, name: ident.unbox().name }
+                    let ident = ident.unbox();
+                    IdentifierReference::new(ident.name, ident.span)
                 }
                 _ => return Err(p.unexpected()),
             };

crates/oxc_parser/src/ts/statement.rs

@@ -431,7 +431,7 @@ impl<'a> Parser<'a> {
         let span = self.start_span();
         let (ident_span, name) = self.parse_identifier_kind(Kind::This);
         let type_annotation = self.parse_ts_type_annotation()?;
-        let kind = self.ast.binding_identifier(BindingIdentifier { span: ident_span, name });
+        let kind = self.ast.binding_identifier(BindingIdentifier::new(name, ident_span));
         let binding = self.ast.binding_pattern(kind, type_annotation, /* optional */ false);
         Ok(self.ast.formal_parameter(
             self.end_span(span),

crates/oxc_semantic/src/binder.rs

@@ -28,6 +28,7 @@ impl<'a> Binder for VariableDeclarator<'a> {
         };
         self.id.bound_names(&mut |ident| {
             let symbol_id = builder.declare_symbol(ident.span, &ident.name, includes, excludes);
+            ident.symbol_id.set(Some(symbol_id));
             if self.kind == VariableDeclarationKind::Var
                 && !builder.scope.get_flags(current_scope_id).is_var()
             {
@@ -59,12 +60,13 @@ impl<'a> Binder for Class<'a> {
     fn bind(&self, builder: &mut SemanticBuilder) {
         let Some(ident) = &self.id else { return };
         if !self.modifiers.contains(ModifierKind::Declare) {
-            builder.declare_symbol(
+            let symbol_id = builder.declare_symbol(
                 ident.span,
                 &ident.name,
                 SymbolFlags::Class,
                 SymbolFlags::ClassExcludes,
             );
+            ident.symbol_id.set(Some(symbol_id));
         }
     }
 }
@@ -108,13 +110,14 @@ impl<'a> Binder for Function<'a> {
                         )
                     };
 
-                builder.declare_symbol_on_scope(
+                let symbol_id = builder.declare_symbol_on_scope(
                     ident.span,
                     &ident.name,
                     parent_scope_id,
                     includes,
                     excludes,
                 );
+                ident.symbol_id.set(Some(symbol_id));
             }
         }
 
@@ -152,7 +155,8 @@ impl<'a> Binder for FormalParameters<'a> {
         let is_signature = self.kind == FormalParameterKind::Signature;
         self.bound_names(&mut |ident| {
             if !is_signature {
-                builder.declare_symbol(ident.span, &ident.name, includes, excludes);
+                let symbol_id = builder.declare_symbol(ident.span, &ident.name, includes, excludes);
+                ident.symbol_id.set(Some(symbol_id));
             }
         });
     }
@@ -167,15 +171,22 @@ impl<'a> Binder for CatchClause<'a> {
             // unless CatchParameter is CatchParameter : BindingIdentifier
             if let BindingPatternKind::BindingIdentifier(ident) = &param.kind {
                 let includes = SymbolFlags::FunctionScopedVariable | SymbolFlags::CatchVariable;
-                builder.declare_shadow_symbol(&ident.name, ident.span, current_scope_id, includes);
+                let symbol_id = builder.declare_shadow_symbol(
+                    &ident.name,
+                    ident.span,
+                    current_scope_id,
+                    includes,
+                );
+                ident.symbol_id.set(Some(symbol_id));
             } else {
                 param.bound_names(&mut |ident| {
-                    builder.declare_symbol(
+                    let symbol_id = builder.declare_symbol(
                         ident.span,
                         &ident.name,
                         SymbolFlags::BlockScopedVariable | SymbolFlags::CatchVariable,
                         SymbolFlags::BlockScopedVariableExcludes,
                     );
+                    ident.symbol_id.set(Some(symbol_id));
                 });
             }
         }
@@ -185,35 +196,38 @@ impl<'a> Binder for CatchClause<'a> {
 impl<'a> Binder for ModuleDeclaration<'a> {
     fn bind(&self, builder: &mut SemanticBuilder) {
         self.bound_names(&mut |ident| {
-            builder.declare_symbol(
+            let symbol_id = builder.declare_symbol(
                 ident.span,
                 &ident.name,
                 SymbolFlags::ImportBinding,
                 SymbolFlags::ImportBindingExcludes,
             );
+            ident.symbol_id.set(Some(symbol_id));
         });
     }
 }
 
 impl<'a> Binder for TSTypeAliasDeclaration<'a> {
     fn bind(&self, builder: &mut SemanticBuilder) {
-        builder.declare_symbol(
+        let symbol_id = builder.declare_symbol(
             self.id.span,
             &self.id.name,
             SymbolFlags::TypeAlias,
             SymbolFlags::TypeAliasExcludes,
         );
+        self.id.symbol_id.set(Some(symbol_id));
     }
 }
 
 impl<'a> Binder for TSInterfaceDeclaration<'a> {
     fn bind(&self, builder: &mut SemanticBuilder) {
-        builder.declare_symbol(
+        let symbol_id = builder.declare_symbol(
             self.id.span,
             &self.id.name,
             SymbolFlags::Interface,
             SymbolFlags::InterfaceExcludes,
         );
+        self.id.symbol_id.set(Some(symbol_id));
     }
 }
 
@@ -226,7 +240,8 @@ impl<'a> Binder for TSEnumDeclaration<'a> {
         } else {
             SymbolFlags::RegularEnumExcludes
         };
-        builder.declare_symbol(self.id.span, &self.id.name, includes, excludes);
+        let symbol_id = builder.declare_symbol(self.id.span, &self.id.name, includes, excludes);
+        self.id.symbol_id.set(Some(symbol_id));
     }
 }
 
@@ -271,11 +286,12 @@ impl<'a> Binder for TSModuleDeclaration<'a> {
 
 impl<'a> Binder for TSTypeParameter<'a> {
     fn bind(&self, builder: &mut SemanticBuilder) {
-        builder.declare_symbol(
+        let symbol_id = builder.declare_symbol(
             self.name.span,
             &self.name.name,
             SymbolFlags::TypeParameter,
             SymbolFlags::TypeParameterExcludes,
         );
+        self.name.symbol_id.set(Some(symbol_id));
     }
 }

crates/oxc_semantic/src/builder.rs

@@ -573,7 +573,8 @@ impl<'a> SemanticBuilder<'a> {
     fn reference_identifier(&mut self, ident: &IdentifierReference) {
         let flag = self.resolve_reference_usages();
         let reference = Reference::new(ident.span, ident.name.clone(), self.current_node_id, flag);
-        self.declare_reference(reference);
+        let reference_id = self.declare_reference(reference);
+        ident.reference_id.set(Some(reference_id));
     }
 
     /// Resolve reference flags for the current ast node.

crates/oxc_semantic/src/reference.rs

@@ -1,12 +1,9 @@
 use bitflags::bitflags;
-use oxc_index::define_index_type;
 use oxc_span::{Atom, Span};
 
 use crate::{symbol::SymbolId, AstNodeId};
 
-define_index_type! {
-    pub struct ReferenceId = u32;
-}
+pub use oxc_syntax::reference::ReferenceId;
 
 #[derive(Debug, Clone)]
 pub struct Reference {

crates/oxc_syntax/src/lib.rs

@@ -4,6 +4,7 @@ pub mod identifier;
 pub mod module_record;
 pub mod operator;
 pub mod precedence;
+pub mod reference;
 pub mod scope;
 pub mod symbol;
 

crates/oxc_syntax/src/reference.rs

@@ -0,0 +1,5 @@
+use oxc_index::define_index_type;
+
+define_index_type! {
+    pub struct ReferenceId = u32;
+}

tasks/coverage/src/suite.rs

@@ -285,6 +285,9 @@ pub trait Case: Sized + Sync + Send + UnwindSafe {
             .with_module_record_builder(true)
             .with_check_syntax_error(true)
             .build(program);
+        if let Some(res) = self.check_semantic(&semantic_ret.semantic) {
+            return res;
+        }
         let errors = parser_ret.errors.into_iter().chain(semantic_ret.errors).collect::<Vec<_>>();
 
         let result = if errors.is_empty() {
@@ -356,4 +359,8 @@ pub trait Case: Sized + Sync + Send + UnwindSafe {
         }
         Ok(())
     }
+
+    fn check_semantic(&self, _semantic: &oxc_semantic::Semantic<'_>) -> Option<TestResult> {
+        None
+    }
 }

tasks/coverage/src/test262.rs

@@ -190,4 +190,53 @@ impl Case for Test262Case {
             }
         };
     }
+
+    fn check_semantic(&self, semantic: &oxc_semantic::Semantic<'_>) -> Option<TestResult> {
+        if are_all_identifiers_resolved(semantic) {
+            None
+        } else {
+            Some(TestResult::ParseError("Unset symbol / reference".to_string(), true))
+        }
+    }
+}
+
+fn are_all_identifiers_resolved(semantic: &oxc_semantic::Semantic<'_>) -> bool {
+    use oxc_ast::{ast, AstKind};
+    use oxc_semantic::AstNode;
+
+    let ast_nodes = semantic.nodes();
+    let has_non_resolved = ast_nodes.iter().any(|node| {
+        match node.kind() {
+            AstKind::BindingIdentifier(id) => {
+                let mut parents = ast_nodes.iter_parents(node.id()).map(AstNode::kind);
+                parents.next(); // Exclude BindingIdentifier itself
+                match parents.next() {
+                    Some(AstKind::Function(func))
+                        if func.r#type == ast::FunctionType::FunctionExpression =>
+                    {
+                        // FIXME: Currently, the name of `FunctionExpression` won't be assigned a `SymbolId`
+                        return false;
+                    }
+                    _ => {}
+                }
+                let mut parents = ast_nodes.iter_parents(node.id()).map(AstNode::kind);
+                parents.next(); // Exclude BindingIdentifier itself
+                match (parents.next(), parents.next()) {
+                    // FIXME: case like `if (xx) ; else function test() {}`
+                    (Some(AstKind::Function(func)), Some(AstKind::IfStatement(_)))
+                        if func.r#type == ast::FunctionType::FunctionDeclaration =>
+                    {
+                        return false;
+                    }
+                    _ => {}
+                }
+
+                id.symbol_id.get().is_none()
+            }
+            AstKind::IdentifierReference(ref_id) => ref_id.reference_id.get().is_none(),
+            _ => false,
+        }
+    });
+
+    !has_non_resolved
 }