util: add fast path for utf8 encoding

Co-authored-by: Anna Henningsen <[email protected]>

Author: anonrig

lib/internal/encoding.js

@@ -4,6 +4,7 @@
 // https://encoding.spec.whatwg.org
 
 const {
+  Boolean,
   ObjectCreate,
   ObjectDefineProperties,
   ObjectGetOwnPropertyDescriptors,
@@ -29,6 +30,8 @@ const kFlags = Symbol('flags');
 const kEncoding = Symbol('encoding');
 const kDecoder = Symbol('decoder');
 const kEncoder = Symbol('encoder');
+const kUTF8FastPath = Symbol('kUTF8FastPath');
+const kIgnoreBOM = Symbol('kIgnoreBOM');
 
 const {
   getConstructorOf,
@@ -50,7 +53,8 @@ const {
 
 const {
   encodeInto,
-  encodeUtf8String
+  encodeUtf8String,
+  decodeUTF8,
 } = internalBinding('buffer');
 
 let Buffer;
@@ -398,19 +402,40 @@ function makeTextDecoderICU() {
         flags |= options.ignoreBOM ? CONVERTER_FLAGS_IGNORE_BOM : 0;
       }
 
-      const handle = getConverter(enc, flags);
-      if (handle === undefined)
-        throw new ERR_ENCODING_NOT_SUPPORTED(encoding);
+      // Only support fast path for UTF-8 without FATAL flag
+      const fastPathAvailable = enc === 'utf-8' && !(options?.fatal);
 
       this[kDecoder] = true;
-      this[kHandle] = handle;
       this[kFlags] = flags;
       this[kEncoding] = enc;
+      this[kIgnoreBOM] = Boolean(options?.ignoreBOM);
+      this[kUTF8FastPath] = fastPathAvailable;
+      this[kHandle] = undefined;
+
+      if (!fastPathAvailable) {
+        this.#prepareConverter();
+      }
     }
 
+    #prepareConverter() {
+      if (this[kHandle] !== undefined) return;
+      const handle = getConverter(this[kEncoding], this[kFlags]);
+      if (handle === undefined)
+        throw new ERR_ENCODING_NOT_SUPPORTED(this[kEncoding]);
+      this[kHandle] = handle;
+    }
 
     decode(input = empty, options = kEmptyObject) {
       validateDecoder(this);
+
+      this[kUTF8FastPath] &&= !(options?.stream);
+
+      if (this[kUTF8FastPath]) {
+        return decodeUTF8(input, this[kIgnoreBOM]);
+      }
+
+      this.#prepareConverter();
+
       if (!isAnyArrayBuffer(input) && !isArrayBufferView(input)) {
         throw new ERR_INVALID_ARG_TYPE('input',
                                        ['ArrayBuffer', 'ArrayBufferView'],

src/node_buffer.cc

@@ -24,6 +24,7 @@
 #include "node_blob.h"
 #include "node_errors.h"
 #include "node_external_reference.h"
+#include "node_i18n.h"
 #include "node_internals.h"
 
 #include "env-inl.h"
@@ -565,6 +566,53 @@ void StringSlice(const FunctionCallbackInfo<Value>& args) {
   args.GetReturnValue().Set(ret);
 }
 
+// Convert the input into an encoded string
+void DecodeUTF8(const FunctionCallbackInfo<Value>& args) {
+  Environment* env = Environment::GetCurrent(args);  // list, flags
+
+  if (!(args[0]->IsArrayBuffer() || args[0]->IsSharedArrayBuffer() ||
+        args[0]->IsArrayBufferView())) {
+    return node::THROW_ERR_INVALID_ARG_TYPE(
+        env->isolate(),
+        "The \"list\" argument must be an instance of SharedArrayBuffer, "
+        "ArrayBuffer or ArrayBufferView.");
+  }
+
+  ArrayBufferViewContents<char> buffer(args[0]);
+
+  CHECK(args[1]->IsBoolean());
+  bool ignore_bom = args[1]->BooleanValue(env->isolate());
+
+  const char* data = buffer.data();
+  size_t beginning = 0;
+  size_t length = buffer.length();
+
+  if (length == 0) return args.GetReturnValue().SetEmptyString();
+
+  if (!ignore_bom) {
+    char bom[] = "\xEF\xBB\xBF";
+
+    if (strncmp(data, bom, 3) == 0) {
+      beginning += 3;
+      length -= 3;
+    }
+  }
+
+  auto output = data + beginning;
+
+  Local<Value> error;
+  MaybeLocal<Value> maybe_ret =
+      StringBytes::Encode(env->isolate(), output, length, UTF8, &error);
+  Local<Value> ret;
+
+  if (!maybe_ret.ToLocal(&ret)) {
+    CHECK(!error.IsEmpty());
+    env->isolate()->ThrowException(error);
+    return;
+  }
+
+  args.GetReturnValue().Set(ret);
+}
 
 // bytesCopied = copy(buffer, target[, targetStart][, sourceStart][, sourceEnd])
 void Copy(const FunctionCallbackInfo<Value> &args) {
@@ -1282,6 +1330,7 @@ void Initialize(Local<Object> target,
 
   SetMethod(context, target, "setBufferPrototype", SetBufferPrototype);
   SetMethodNoSideEffect(context, target, "createFromString", CreateFromString);
+  SetMethodNoSideEffect(context, target, "decodeUTF8", DecodeUTF8);
 
   SetMethodNoSideEffect(context, target, "byteLengthUtf8", ByteLengthUtf8);
   SetMethod(context, target, "copy", Copy);
@@ -1339,6 +1388,7 @@ void Initialize(Local<Object> target,
 void RegisterExternalReferences(ExternalReferenceRegistry* registry) {
   registry->Register(SetBufferPrototype);
   registry->Register(CreateFromString);
+  registry->Register(DecodeUTF8);
 
   registry->Register(ByteLengthUtf8);
   registry->Register(Copy);

test/parallel/test-whatwg-encoding-custom-textdecoder.js

@@ -113,7 +113,7 @@ if (common.hasIntl) {
       '  fatal: false,\n' +
       '  ignoreBOM: true,\n' +
       '  [Symbol(flags)]: 4,\n' +
-      '  [Symbol(handle)]: Converter {}\n' +
+      '  [Symbol(handle)]: undefined\n' +
       '}'
     );
   } else {