From 866781fff6751c8d2964b6f395ecd998f51de03f Mon Sep 17 00:00:00 2001
From: Nikita Skovoroda <chalkerx@gmail.com>
Date: Fri, 19 Dec 2025 01:32:42 +0400
Subject: [PATCH 1/2] src: improve StringBytes::Encode perf on ASCII

---
 src/encoding_binding.cc | 24 +++++++++++++++++-------
 src/string_bytes.cc     |  8 ++++++++
 2 files changed, 25 insertions(+), 7 deletions(-)

diff --git a/src/encoding_binding.cc b/src/encoding_binding.cc
index f68dd9522a0f69..01f10de535bcca 100644
--- a/src/encoding_binding.cc
+++ b/src/encoding_binding.cc
@@ -356,7 +356,24 @@ 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(data, length)) {
+      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) {
@@ -365,13 +382,6 @@ 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
index 03b5fd7ebe3816..550b1fa5aeb898 100644
--- 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(buf, buflen)) {
+        return ExternOneByteString::NewFromCopy(isolate, buf, buflen);
+      }
+
       val =
           String::NewFromUtf8(isolate, buf, v8::NewStringType::kNormal, buflen);
       Local<String> str;

From 42b4ff4a114e6a1d8695aef26e439fac92c16284 Mon Sep 17 00:00:00 2001
From: Nikita Skovoroda <chalkerx@gmail.com>
Date: Fri, 19 Dec 2025 13:30:57 +0400
Subject: [PATCH 2/2] src: use validate_ascii_with_errors

---
 src/encoding_binding.cc | 2 +-
 src/string_bytes.cc     | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/src/encoding_binding.cc b/src/encoding_binding.cc
index 01f10de535bcca..03b2848cb7b948 100644
--- a/src/encoding_binding.cc
+++ b/src/encoding_binding.cc
@@ -365,7 +365,7 @@ void BindingData::DecodeUTF8(const FunctionCallbackInfo<Value>& args) {
 
   if (has_fatal) {
     // Are we perhaps ASCII? Then we won't have to check for UTF-8
-    if (simdutf::validate_ascii(data, length)) {
+    if (!simdutf::validate_ascii_with_errors(data, length).error) {
       Local<Value> ret;
       if (StringBytes::Encode(env->isolate(), data, length, LATIN1)
               .ToLocal(&ret)) {
diff --git a/src/string_bytes.cc b/src/string_bytes.cc
index 550b1fa5aeb898..8f6bedd63e25b3 100644
--- a/src/string_bytes.cc
+++ b/src/string_bytes.cc
@@ -535,7 +535,7 @@ MaybeLocal<Value> StringBytes::Encode(Isolate* isolate,
       // 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(buf, buflen)) {
+      if (!simdutf::validate_ascii_with_errors(buf, buflen).error) {
         return ExternOneByteString::NewFromCopy(isolate, buf, buflen);
       }