fix(minifier): fix KATAKANA MIDDLE DOT syntax error for unicode 4.1 to 15 (#12829)

fixes rolldown/rolldown#5591

Author: Boshen

crates/oxc_minifier/src/ctx.rs

@@ -9,7 +9,10 @@ use oxc_ecmascript::{
 };
 use oxc_semantic::{IsGlobalReference, Scoping, SymbolId};
 use oxc_span::format_atom;
-use oxc_syntax::reference::ReferenceId;
+use oxc_syntax::{
+    identifier::{is_identifier_part, is_identifier_start},
+    reference::ReferenceId,
+};
 
 use crate::{options::CompressOptions, state::MinifierState, symbol_value::SymbolValue};
 
@@ -241,4 +244,14 @@ impl<'a> Ctx<'a, '_> {
         }
         Some(f64::from(int_value))
     }
+
+    /// `is_identifier_name` patched with KATAKANA MIDDLE DOT and HALFWIDTH KATAKANA MIDDLE DOT
+    /// Otherwise `({ 'x・': 0 })` gets converted to `({ x・: 0 })`, which breaks in Unicode 4.1 to
+    /// 15.
+    /// <https://github.com/oxc-project/unicode-id-start/pull/3>
+    pub fn is_identifier_name_patched(s: &str) -> bool {
+        let mut chars = s.chars();
+        chars.next().is_some_and(is_identifier_start)
+            && chars.all(|c| is_identifier_part(c) && c != '・' && c != '･')
+    }
 }

crates/oxc_minifier/src/peephole/convert_to_dotted_properties.rs

@@ -1,6 +1,5 @@
 use oxc_allocator::TakeIn;
 use oxc_ast::ast::*;
-use oxc_syntax::identifier::is_identifier_name;
 
 use crate::ctx::Ctx;
 
@@ -17,7 +16,7 @@ impl<'a> LatePeepholeOptimizations {
     pub fn convert_to_dotted_properties(expr: &mut MemberExpression<'a>, ctx: &mut Ctx<'a, '_>) {
         let MemberExpression::ComputedMemberExpression(e) = expr else { return };
         let Expression::StringLiteral(s) = &e.expression else { return };
-        if is_identifier_name(&s.value) {
+        if Ctx::is_identifier_name_patched(&s.value) {
             let property = ctx.ast.identifier_name(s.span, s.value);
             *expr =
                 MemberExpression::StaticMemberExpression(ctx.ast.alloc_static_member_expression(

crates/oxc_minifier/src/peephole/substitute_alternate_syntax.rs

@@ -8,7 +8,6 @@ use oxc_span::GetSpan;
 use oxc_span::SPAN;
 use oxc_syntax::{
     es_target::ESTarget,
-    identifier::is_identifier_name,
     number::NumberBase,
     operator::{BinaryOperator, UnaryOperator},
 };
@@ -877,35 +876,40 @@ impl<'a> PeepholeOptimizations {
         computed: &mut bool,
         ctx: &mut Ctx<'a, '_>,
     ) {
-        if let PropertyKey::NumericLiteral(_) = key {
-            if *computed {
-                *computed = false;
+        match key {
+            PropertyKey::NumericLiteral(_) => {
+                if *computed {
+                    *computed = false;
+                }
             }
-            return;
-        }
-        let PropertyKey::StringLiteral(s) = key else { return };
-        let value = s.value.as_str();
-        if is_identifier_name(value) {
-            *computed = false;
-            *key = PropertyKey::StaticIdentifier(ctx.ast.alloc_identifier_name(s.span, s.value));
-            ctx.state.changed = true;
-            return;
-        }
-        if let Some(value) = Ctx::string_to_equivalent_number_value(value) {
-            if value >= 0.0 {
-                *computed = false;
-                *key = PropertyKey::NumericLiteral(ctx.ast.alloc_numeric_literal(
-                    s.span,
-                    value,
-                    None,
-                    NumberBase::Decimal,
-                ));
-                ctx.state.changed = true;
-                return;
+            PropertyKey::StringLiteral(s) => {
+                let value = s.value.as_str();
+                if Ctx::is_identifier_name_patched(value) {
+                    *computed = false;
+                    *key = PropertyKey::StaticIdentifier(
+                        ctx.ast.alloc_identifier_name(s.span, s.value),
+                    );
+                    ctx.state.changed = true;
+                    return;
+                }
+                if let Some(value) = Ctx::string_to_equivalent_number_value(value) {
+                    if value >= 0.0 {
+                        *computed = false;
+                        *key = PropertyKey::NumericLiteral(ctx.ast.alloc_numeric_literal(
+                            s.span,
+                            value,
+                            None,
+                            NumberBase::Decimal,
+                        ));
+                        ctx.state.changed = true;
+                        return;
+                    }
+                }
+                if *computed {
+                    *computed = false;
+                }
             }
-        }
-        if *computed {
-            *computed = false;
+            _ => {}
         }
     }
 
@@ -1744,6 +1748,13 @@ mod test {
             "class C { static accessor __proto__ = 0 }",
         );
 
+        // Patch KATAKANA MIDDLE DOT and HALFWIDTH KATAKANA MIDDLE DOT
+        // <https://github.com/oxc-project/unicode-id-start/pull/3>
+        test_same("x = { 'x・': 0 };");
+        test_same("x = { 'x･': 0 };");
+        test_same("x = y['x・'];");
+        test_same("x = y['x･'];");
+
         // <https://tc39.es/ecma262/2024/multipage/ecmascript-language-functions-and-classes.html#sec-static-semantics-classelementkind>
         // <https://tc39.es/ecma262/2024/multipage/ecmascript-language-functions-and-classes.html#sec-class-definitions-static-semantics-early-errors>
         // <https://arai-a.github.io/ecma262-compare/?pr=2417&id=sec-class-definitions-static-semantics-early-errors>