src: improve StringBytes::Encode perf on ASCII

ChALkeR · aduh95 · commit b6d3caeda8fe · 2026-01-19T19:19:29.000+01:00
PR-URL: #61119 Reviewed-By: Daniel Lemire <daniel@lemire.me> Reviewed-By: Matteo Collina <matteo.collina@gmail.com> Reviewed-By: Colin Ihrig <cjihrig@gmail.com> Reviewed-By: Rafael Gonzaga <rafael.nunu@hotmail.com> Reviewed-By: Gürgün Dayıoğlu <hey@gurgun.day> Reviewed-By: Xuguang Mei <meixuguang@gmail.com> Reviewed-By: Juan José Arboleda <soyjuanarbol@gmail.com> Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
diff --git a/src/encoding_binding.cc b/src/encoding_binding.cc
@@ -163,7 +163,24 @@ void BindingData::DecodeUTF8(const FunctionCallbackInfo<Value>& args) {
   const char* data = buffer.data();
   size_t length = buffer.length();
 
+  if (!ignore_bom && length >= 3) {
+    if (memcmp(data, "\xEF\xBB\xBF", 3) == 0) {
+      data += 3;
+      length -= 3;
+    }
+  }
+
   if (has_fatal) {
+    // Are we perhaps ASCII? Then we won't have to check for UTF-8
+    if (!simdutf::validate_ascii_with_errors(data, length).error) {
+      Local<Value> ret;
+      if (StringBytes::Encode(env->isolate(), data, length, LATIN1)
+              .ToLocal(&ret)) {
+        args.GetReturnValue().Set(ret);
+      }
+      return;
+    }
+
     auto result = simdutf::validate_utf8_with_errors(data, length);
 
     if (result.error) {
@@ -172,13 +189,6 @@ void BindingData::DecodeUTF8(const FunctionCallbackInfo<Value>& args) {
     }
   }
 
-  if (!ignore_bom && length >= 3) {
-    if (memcmp(data, "\xEF\xBB\xBF", 3) == 0) {
-      data += 3;
-      length -= 3;
-    }
-  }
-
   if (length == 0) return args.GetReturnValue().SetEmptyString();
 
   Local<Value> ret;
diff --git a/src/string_bytes.cc b/src/string_bytes.cc
@@ -531,6 +531,14 @@ MaybeLocal<Value> StringBytes::Encode(Isolate* isolate,
 
     case UTF8: {
       buflen = keep_buflen_in_range(buflen);
+
+      // ASCII fast path
+      // TODO(chalker): remove when String::NewFromUtf8 is fast enough itself
+      // This is cheap compared to the benefits though
+      if (!simdutf::validate_ascii_with_errors(buf, buflen).error) {
+        return ExternOneByteString::NewFromCopy(isolate, buf, buflen);
+      }
+
       val =
           String::NewFromUtf8(isolate, buf, v8::NewStringType::kNormal, buflen);
       Local<String> str;
