u

shulaoda · shulaoda · commit 177a0929f110 · 2026-03-20T10:49:58.000+08:00
diff --git a/crates/rolldown_utils/src/hash_placeholder.rs b/crates/rolldown_utils/src/hash_placeholder.rs
@@ -55,11 +55,10 @@ impl<'a> Iterator for HashPlaceholderIter<'a> {
       // Use byte-level slicing to avoid panics when `search_end` falls inside a
       // multi-byte UTF-8 character (e.g. Chinese/Japanese characters in the source).
       let search_end = (left_pos + MAX_HASH_SIZE + HASH_PLACEHOLDER_OVERHEAD).min(self.s.len());
-      let search_bytes = &self.s.as_bytes()[left_pos..search_end];
-      if let Some(right_pos) = memchr::memmem::find(search_bytes, HASH_PLACEHOLDER_RIGHT.as_bytes())
-      {
-        // `right_pos` points to the start of `}~` within `search_bytes`; advance
-        // past those two ASCII bytes so the end index is always char-boundary-safe.
+      if let Some(right_pos) = memchr::memmem::find(
+        &self.s.as_bytes()[left_pos..search_end],
+        HASH_PLACEHOLDER_RIGHT.as_bytes(),
+      ) {
         let right_pos = left_pos + right_pos + HASH_PLACEHOLDER_RIGHT.len();
         let placeholder = &self.s[left_pos..right_pos];
         self.start = right_pos;
@@ -253,25 +252,10 @@ fn test_find_hash_placeholders() {
 }
 
 #[test]
-fn test_find_hash_placeholders_multibyte_chars() {
-  // Regression test: Chinese/Japanese characters (multi-byte UTF-8) in the
-  // surrounding source text must not cause a panic when `search_end` lands
-  // inside a multi-byte character.  The placeholder '!~{001}~' is still found
-  // correctly.
+fn test_find_hash_placeholders_multi_byte_chars() {
+  // Multi-byte UTF-8 chars near placeholders must not cause a panic.
   let s = "import{C as e}from\"./vue.runtime.esm-bundler-!~{001}~.js\";// 中文级别文字";
   let placeholders: Vec<_> = find_hash_placeholders(s, &HASH_PLACEHOLDER_LEFT_FINDER).collect();
   assert_eq!(placeholders.len(), 1);
   assert_eq!(placeholders[0].2, "!~{001}~");
-
-  // Test where the multi-byte character falls right at the search boundary
-  // (26 bytes after `!~{`).  Construct a string where a 3-byte Chinese char
-  // straddles byte offset `left_pos + MAX_HASH_SIZE + OVERHEAD`.
-  let prefix = "!~{001}~"; // placeholder itself
-  let filler = "x".repeat(20); // padding so next `!~{` is far enough
-  let chinese = "级"; // 3-byte UTF-8 char
-  let s = format!("{prefix}{filler}!~{{002}}{chinese}");
-  // No `}~` closing the second `!~{`, so only one placeholder should be found.
-  let placeholders: Vec<_> = find_hash_placeholders(&s, &HASH_PLACEHOLDER_LEFT_FINDER).collect();
-  assert_eq!(placeholders.len(), 1);
-  assert_eq!(placeholders[0].2, "!~{001}~");
 }
