util: add fast path for Latin1 decoding

nodejs · Oct 5, 2024 · 2dd0714 · 2dd0714
1 parent e1852b5
commit 2dd0714
Show file tree

Hide file tree

Showing 3 changed files with 46 additions and 0 deletions.
diff --git a/lib/internal/encoding.js b/lib/internal/encoding.js
@@ -55,6 +55,7 @@ const {
   encodeIntoResults,
   encodeUtf8String,
   decodeUTF8,
+  decodeLatin1,
 } = binding;
 
 const { Buffer } = require('buffer');
@@ -443,6 +444,10 @@ function makeTextDecoderICU() {
         return decodeUTF8(input, this[kIgnoreBOM], this[kFatal]);
       }
 
+      if (this[kEncoding] === 'windows-1252') {
+        return decodeLatin1(input);
+      }
+
       this.#prepareConverter();
 
       validateObject(options, 'options', kValidateObjectAllowObjectsAndNull);

diff --git a/src/encoding_binding.cc b/src/encoding_binding.cc
@@ -1,6 +1,7 @@
 #include "encoding_binding.h"
 #include "ada.h"
 #include "env-inl.h"
+#include "node_buffer.h"
 #include "node_errors.h"
 #include "node_external_reference.h"
 #include "simdutf.h"
@@ -226,6 +227,7 @@ void BindingData::CreatePerIsolateProperties(IsolateData* isolate_data,
   SetMethodNoSideEffect(isolate, target, "decodeUTF8", DecodeUTF8);
   SetMethodNoSideEffect(isolate, target, "toASCII", ToASCII);
   SetMethodNoSideEffect(isolate, target, "toUnicode", ToUnicode);
+  SetMethodNoSideEffect(isolate, target, "decodeLatin1", DecodeLatin1);
 }
 
 void BindingData::CreatePerContextProperties(Local<Object> target,
@@ -243,6 +245,44 @@ void BindingData::RegisterTimerExternalReferences(
   registry->Register(DecodeUTF8);
   registry->Register(ToASCII);
   registry->Register(ToUnicode);
+  registry->Register(DecodeLatin1);
+}
+
+void BindingData::DecodeLatin1(const FunctionCallbackInfo<Value>& args) {
+  Environment* env = Environment::GetCurrent(args);
+
+  CHECK_GE(args.Length(), 1);
+  if (!(args[0]->IsArrayBuffer() || args[0]->IsSharedArrayBuffer() ||
+        args[0]->IsArrayBufferView())) {
+    return node::THROW_ERR_INVALID_ARG_TYPE(
+        env->isolate(),
+        "The \"input\" argument must be an instance of ArrayBuffer, "
+        "SharedArrayBuffer, or ArrayBufferView.");
+  }
+
+  ArrayBufferViewContents<uint8_t> buffer(args[0]);
+  const uint8_t* data = buffer.data();
+  size_t length = buffer.length();
+
+  if (length == 0) {
+    return args.GetReturnValue().SetEmptyString();
+  }
+
+  std::string result(length * 2, '\0');
+
+  size_t written = simdutf::convert_latin1_to_utf8(
+      reinterpret_cast<const char*>(data), length, &result[0]);
+
+  if (written == 0) {
+    return node::THROW_ERR_ENCODING_INVALID_ENCODED_DATA(
+        env->isolate(), "The encoded data was not valid for encoding latin1");
+  }
+
+  result.resize(written);
+
+  Local<Object> buffer_result =
+      node::Buffer::Copy(env, result.c_str(), result.length()).ToLocalChecked();
+  args.GetReturnValue().Set(buffer_result);
 }
 
 }  // namespace encoding_binding

diff --git a/src/encoding_binding.h b/src/encoding_binding.h
@@ -31,6 +31,7 @@ class BindingData : public SnapshotableObject {
   static void EncodeInto(const v8::FunctionCallbackInfo<v8::Value>& args);
   static void EncodeUtf8String(const v8::FunctionCallbackInfo<v8::Value>& args);
   static void DecodeUTF8(const v8::FunctionCallbackInfo<v8::Value>& args);
+  static void DecodeLatin1(const v8::FunctionCallbackInfo<v8::Value>& args);
 
   static void ToASCII(const v8::FunctionCallbackInfo<v8::Value>& args);
   static void ToUnicode(const v8::FunctionCallbackInfo<v8::Value>& args);