fix(parser): classify Unicode line breaks in block comments (#26536)

Block comments containing U+2028 or U+2029 were marked `SingleLineBlock` unless they also contained LF or CR. Switch to the multiline-comment scanner after either Unicode separator so downstream consumers receive the correct comment kind.

This prevents the formatter from dropping parentheses around a return argument with such a leading comment, which could change the returned value to `undefined`.

crates/oxc_codegen/tests/integration/comments.rs

@@ -1011,3 +1011,20 @@ fn test_comment_on_paren_protected_prologue_boundary() {
     test_same("// leading comment\n(\"use strict\");\nfoo();\n");
     test_same("\"use asm\";\n// leading comment\n(\"use strict\");\nfoo();\n");
 }
+
+#[test]
+fn test_block_comment_line_terminators() {
+    for separator in ["\u{2028}", "\u{2029}", "\n", "\r", "\r\n"] {
+        let source = format!("function f(){{return (/* first{separator}second */ {{}});}}");
+        test(&source, "function f() {\n\treturn (/* first\n\tsecond */ {});\n}\n");
+        test_idempotency(&source);
+    }
+    for separator in ["", "\u{2027}", "\u{202a}", "\u{00a0}"] {
+        let source = format!("function f(){{return (/* first{separator}second */ {{}});}}");
+        test(
+            &source,
+            &format!("function f() {{\n\treturn (/* first{separator}second */ {{}});\n}}\n"),
+        );
+        test_idempotency(&source);
+    }
+}

crates/oxc_parser/src/lexer/comment.rs

@@ -82,9 +82,8 @@ impl<'a, C: Config> Lexer<'a, C> {
 
     /// Section 12.4 Multi Line Comment
     pub(super) fn skip_multi_line_comment(&mut self) -> Kind {
-        // If `is_on_new_line` is already set, go directly to faster search which only looks for `*/`
         // We need to identify if comment contains line breaks or not
-        // (`CommentKind::Block` or `CommentKind::MultilineBlock`).
+        // (`CommentKind::SingleLineBlock` or `CommentKind::MultiLineBlock`).
         // So we have to use the loop below for the first line of the comment even if
         // `Token`'s `is_on_new_line` flag is already set.
         // If the loop finds a line break before end of the comment, we then switch to
@@ -126,10 +125,11 @@ impl<'a, C: Config> Lexer<'a, C> {
                     if matches!(next2, LS_BYTES_2_AND_3 | PS_BYTES_2_AND_3) {
                         // Irregular line break
                         self.token.set_is_on_new_line(true);
-                        // Ideally we'd go on to `skip_multi_line_comment_after_line_break` here,
-                        // but irregular line breaks are rare anyway.
+                        // SAFETY: Consuming this 3-byte UTF-8 char leaves a UTF-8 char boundary.
+                        let after_line_break = unsafe { pos.add(3) };
+                        return self.skip_multi_line_comment_after_line_break(after_line_break);
                     }
-                    // Either way, continue searching.
+                    // Some other Unicode char beginning with `0xE2`.
                     // Skip 3 bytes (macro skips 1 already, so skip 2 here), and continue searching.
                     // SAFETY: `0xE2` is always 1st byte of a 3-byte UTF-8 char,
                     // so consuming 3 bytes will place `pos` on next UTF-8 char boundary.

crates/oxc_parser/src/lexer/trivia_builder.rs

@@ -1100,6 +1100,29 @@ function bar() {}";
         }
     }
 
+    #[test]
+    fn block_comment_line_terminators() {
+        for (separator, kind) in [
+            ("\u{2028}", CommentKind::MultiLineBlock),
+            ("\u{2029}", CommentKind::MultiLineBlock),
+            ("\n", CommentKind::MultiLineBlock),
+            ("\r", CommentKind::MultiLineBlock),
+            ("\r\n", CommentKind::MultiLineBlock),
+            ("", CommentKind::SingleLineBlock),
+            ("\u{2027}", CommentKind::SingleLineBlock),
+            ("\u{202a}", CommentKind::SingleLineBlock),
+            ("\u{00a0}", CommentKind::SingleLineBlock),
+        ] {
+            // A newline before the comment must not affect its kind.
+            for prefix in ["", "let a;", "let a;\n"] {
+                let source = format!("{prefix}/* first{separator}second */ let b;");
+                let comments = get_comments(&source);
+                assert_eq!(comments.len(), 1, "{source:?}");
+                assert_eq!(comments[0].kind, kind, "{source:?}");
+            }
+        }
+    }
+
     #[test]
     fn comment_parsing() {
         let data = [