src: skip duplicate UTF-8 validation in TextDecoder fatal path

mertcanaltin · addaleax · commit 4639dcb41972 · 2026-05-27T21:31:29.000+02:00
Signed-off-by: Mert Can Altin <mertgold60@gmail.com> PR-URL: #63231 Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com> Reviewed-By: Anna Henningsen <anna@addaleax.net> Reviewed-By: Gürgün Dayıoğlu <hey@gurgun.day> Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
diff --git a/benchmark/util/text-decoder.js b/benchmark/util/text-decoder.js
@@ -6,26 +6,42 @@ const bench = common.createBenchmark(main, {
   encoding: ['utf-8', 'windows-1252', 'iso-8859-3'],
   ignoreBOM: [0, 1],
   fatal: [0, 1],
+  type: ['SharedArrayBuffer', 'ArrayBuffer', 'Buffer'],
+  content: ['ascii', 'one-byte-string', 'two-byte-string'],
   len: [256, 1024 * 16, 1024 * 128],
   n: [1e3],
-  type: ['SharedArrayBuffer', 'ArrayBuffer', 'Buffer'],
 });
 
-function main({ encoding, len, n, ignoreBOM, type, fatal }) {
+function buildContent(content, len) {
+  let base;
+  switch (content) {
+    case 'ascii': base = 'a'; break;
+    case 'one-byte-string': base = '\xff'; break;
+    case 'two-byte-string': base = 'ğ'; break;
+  }
+  const unitBytes = Buffer.byteLength(base, 'utf8');
+  const copies = Math.max(1, Math.floor(len / unitBytes));
+  return Buffer.from(base.repeat(copies));
+}
+
+function main({ encoding, len, n, ignoreBOM, type, fatal, content }) {
   const decoder = new TextDecoder(encoding, { ignoreBOM, fatal });
+  const seed = buildContent(content, len);
   let buf;
 
   switch (type) {
     case 'SharedArrayBuffer': {
-      buf = new SharedArrayBuffer(len);
+      buf = new SharedArrayBuffer(seed.length);
+      new Uint8Array(buf).set(seed);
       break;
     }
     case 'ArrayBuffer': {
-      buf = new ArrayBuffer(len);
+      buf = new ArrayBuffer(seed.length);
+      new Uint8Array(buf).set(seed);
       break;
     }
     case 'Buffer': {
-      buf = Buffer.allocUnsafe(len);
+      buf = seed;
       break;
     }
   }
diff --git a/src/encoding_binding.cc b/src/encoding_binding.cc
@@ -459,14 +459,15 @@ void BindingData::DecodeUTF8(const FunctionCallbackInfo<Value>& args) {
       return node::THROW_ERR_ENCODING_INVALID_ENCODED_DATA(
           env->isolate(), "The encoded data was not valid for encoding utf-8");
     }
-
-    // TODO(chalker): save on utf8 validity recheck in StringBytes::Encode()
   }
 
   if (length == 0) return args.GetReturnValue().SetEmptyString();
 
   Local<Value> ret;
-  if (StringBytes::Encode(env->isolate(), data, length, UTF8).ToLocal(&ret)) {
+  v8::MaybeLocal<Value> encoded =
+      has_fatal ? StringBytes::EncodeValidUtf8(env->isolate(), data, length)
+                : StringBytes::Encode(env->isolate(), data, length, UTF8);
+  if (encoded.ToLocal(&ret)) {
     args.GetReturnValue().Set(ret);
   }
 }
diff --git a/src/string_bytes.cc b/src/string_bytes.cc
@@ -671,6 +671,40 @@ MaybeLocal<Value> StringBytes::Encode(Isolate* isolate,
   }
 }
 
+MaybeLocal<Value> StringBytes::EncodeValidUtf8(Isolate* isolate,
+                                               const char* buf,
+                                               size_t buflen) {
+  CHECK_BUFLEN_IN_RANGE(buflen);
+  if (!buflen) return String::Empty(isolate);
+  buflen = keep_buflen_in_range(buflen);
+
+  // ASCII fast path
+  if (!simdutf::validate_ascii_with_errors(buf, buflen).error) {
+    return ExternOneByteString::NewFromCopy(isolate, buf, buflen);
+  }
+
+  if (buflen >= 32) {
+    size_t u16size = simdutf::utf16_length_from_utf8(buf, buflen);
+    if (u16size > static_cast<size_t>(v8::String::kMaxLength)) {
+      isolate->ThrowException(ERR_STRING_TOO_LONG(isolate));
+      return MaybeLocal<Value>();
+    }
+    return EncodeTwoByteString(
+        isolate, u16size, [buf, buflen, u16size](uint16_t* dst) {
+          size_t written = simdutf::convert_valid_utf8_to_utf16(
+              buf, buflen, reinterpret_cast<char16_t*>(dst));
+          CHECK_EQ(written, u16size);
+        });
+  }
+
+  Local<String> str;
+  if (!String::NewFromUtf8(isolate, buf, v8::NewStringType::kNormal, buflen)
+           .ToLocal(&str)) {
+    isolate->ThrowException(node::ERR_STRING_TOO_LONG(isolate));
+  }
+  return str;
+}
+
 MaybeLocal<Value> StringBytes::Encode(Isolate* isolate,
                                       const uint16_t* buf,
                                       size_t buflen) {
diff --git a/src/string_bytes.h b/src/string_bytes.h
@@ -83,6 +83,11 @@ class StringBytes {
                                           size_t buflen,
                                           enum encoding encoding);
 
+  // Like Encode(..., UTF8) but does not re-validate. Input must be valid UTF-8.
+  static v8::MaybeLocal<v8::Value> EncodeValidUtf8(v8::Isolate* isolate,
+                                                   const char* buf,
+                                                   size_t buflen);
+
   // Warning: This reverses endianness on BE platforms, even though the
   // signature using uint16_t implies that it should not.
   // However, the brokenness is already public API and can't therefore

Original file line number	Diff line number	Diff line change
`@@ -459,14 +459,15 @@ void BindingData::DecodeUTF8(const FunctionCallbackInfo<Value>& args) {`
`459`	`459`	`return node::THROW_ERR_ENCODING_INVALID_ENCODED_DATA(`
`460`	`460`	`env->isolate(), "The encoded data was not valid for encoding utf-8");`
`461`	`461`	`}`
`462`		`-`
`463`		`- // TODO(chalker): save on utf8 validity recheck in StringBytes::Encode()`
`464`	`462`	`}`
`465`	`463`
`466`	`464`	`if (length == 0) return args.GetReturnValue().SetEmptyString();`
`467`	`465`
`468`	`466`	`Local<Value> ret;`
`469`		`- if (StringBytes::Encode(env->isolate(), data, length, UTF8).ToLocal(&ret)) {`
	`467`	`+ v8::MaybeLocal<Value> encoded =`
	`468`	`+ has_fatal ? StringBytes::EncodeValidUtf8(env->isolate(), data, length)`
	`469`	`+ : StringBytes::Encode(env->isolate(), data, length, UTF8);`
	`470`	`+ if (encoded.ToLocal(&ret)) {`
`470`	`471`	`args.GetReturnValue().Set(ret);`
`471`	`472`	`}`
`472`	`473`	`}`