Add TextEncoder polyfill (WHATWG Encoding Standard)

CMakeLists.txt

@@ -82,6 +82,7 @@ option(JSRUNTIMEHOST_POLYFILL_WEBSOCKET "Include JsRuntimeHost Polyfill WebSocke
 option(JSRUNTIMEHOST_POLYFILL_BLOB "Include JsRuntimeHost Polyfill Blob." ON)
 option(JSRUNTIMEHOST_POLYFILL_PERFORMANCE "Include JsRuntimeHost Polyfill Performance." ON)
 option(JSRUNTIMEHOST_POLYFILL_TEXTDECODER "Include JsRuntimeHost Polyfill TextDecoder." ON)
+option(JSRUNTIMEHOST_POLYFILL_TEXTENCODER "Include JsRuntimeHost Polyfill TextEncoder." ON)
 
 # Sanitizers
 option(ENABLE_SANITIZERS "Enable AddressSanitizer and UBSan" OFF)

Polyfills/CMakeLists.txt

@@ -32,4 +32,8 @@ endif()
 
 if(JSRUNTIMEHOST_POLYFILL_TEXTDECODER)
     add_subdirectory(TextDecoder)
+endif()
+
+if(JSRUNTIMEHOST_POLYFILL_TEXTENCODER)
+    add_subdirectory(TextEncoder)
 endif()

Polyfills/TextEncoder/CMakeLists.txt

@@ -0,0 +1,15 @@
+set(SOURCES
+    "Include/Babylon/Polyfills/TextEncoder.h"
+    "Source/TextEncoder.cpp")
+
+add_library(TextEncoder ${SOURCES})
+warnings_as_errors(TextEncoder)
+
+target_include_directories(TextEncoder PUBLIC "Include")
+
+target_link_libraries(TextEncoder
+    PUBLIC napi
+    PUBLIC Foundation)
+
+set_property(TARGET TextEncoder PROPERTY FOLDER Polyfills)
+source_group(TREE ${CMAKE_CURRENT_SOURCE_DIR} FILES ${SOURCES})

Polyfills/TextEncoder/Include/Babylon/Polyfills/TextEncoder.h

@@ -0,0 +1,9 @@
+#pragma once
+
+#include <napi/env.h>
+#include <Babylon/Api.h>
+
+namespace Babylon::Polyfills::TextEncoder
+{
+    void BABYLON_API Initialize(Napi::Env env);
+}

Polyfills/TextEncoder/README.md

@@ -0,0 +1,28 @@
+# TextEncoder Polyfill
+
+A C++ implementation of the [WHATWG Encoding API](https://encoding.spec.whatwg.org/) `TextEncoder` interface for use in Babylon Native JavaScript runtimes via [Napi](https://github.com/nodejs/node-addon-api).
+
+This is the encoding counterpart to the `TextDecoder` polyfill in this same repository. Both polyfills exist primarily for older Chakra-based runtimes where the WHATWG Encoding Standard globals are not built in. On modern V8 / JSC / Hermes runtimes the constructor is already exposed and `Initialize()` is a no-op.
+
+## Current State
+
+### Supported
+
+- Constructing `TextEncoder` with no argument (UTF-8 is the only encoding the spec mandates).
+- The `encoding` accessor (always returns `"utf-8"`).
+- `encode(input)` — returns a `Uint8Array` containing the UTF-8 bytes of `input`. Calling with no argument or `undefined` returns an empty `Uint8Array` (matches the spec, which defaults `input` to `""`).
+
+### Not Supported
+
+- Encodings other than UTF-8 — the `TextEncoder` constructor in the spec only accepts UTF-8 anyway, so this is not a deviation.
+- `encodeInto(source, destination)` — not implemented. Babylon.js does not call this entry point; the (substantially more involved) UTF-16-code-unit accounting it would require is not justified by any current consumer. If a future consumer needs it, it can be added at that time.
+
+## Usage
+
+```javascript
+const encoder = new TextEncoder();
+encoder.encoding;                 // "utf-8"
+
+encoder.encode("Hello");          // Uint8Array(5) [72,101,108,108,111]
+encoder.encode();                 // Uint8Array(0) []
+```

Polyfills/TextEncoder/Source/TextEncoder.cpp

@@ -0,0 +1,72 @@
+#include <Babylon/Polyfills/TextEncoder.h>
+
+#include <napi/napi.h>
+
+#include <cstring>
+#include <string>
+
+namespace
+{
+    class TextEncoder final : public Napi::ObjectWrap<TextEncoder>
+    {
+    public:
+        static void Initialize(Napi::Env env)
+        {
+            Napi::HandleScope scope{env};
+
+            static constexpr auto JS_TEXTENCODER_CONSTRUCTOR_NAME = "TextEncoder";
+            if (env.Global().Get(JS_TEXTENCODER_CONSTRUCTOR_NAME).IsUndefined())
+            {
+                Napi::Function func = DefineClass(
+                    env,
+                    JS_TEXTENCODER_CONSTRUCTOR_NAME,
+                    {
+                        InstanceAccessor("encoding", &TextEncoder::Encoding, nullptr),
+                        InstanceMethod("encode", &TextEncoder::Encode),
+                    });
+
+                env.Global().Set(JS_TEXTENCODER_CONSTRUCTOR_NAME, func);
+            }
+        }
+
+        explicit TextEncoder(const Napi::CallbackInfo& info)
+            : Napi::ObjectWrap<TextEncoder>{info}
+        {
+            // The TextEncoder constructor takes no arguments. The encoding is
+            // always UTF-8 per the WHATWG Encoding Standard.
+        }
+
+    private:
+        Napi::Value Encoding(const Napi::CallbackInfo& info)
+        {
+            return Napi::String::New(info.Env(), "utf-8");
+        }
+
+        // encode(input = "") - returns a Uint8Array containing the UTF-8 bytes.
+        // Per spec, undefined input defaults to the empty string (not "undefined").
+        Napi::Value Encode(const Napi::CallbackInfo& info)
+        {
+            auto env = info.Env();
+            std::string utf8;
+            if (info.Length() > 0 && !info[0].IsUndefined())
+            {
+                utf8 = info[0].ToString().Utf8Value();
+            }
+
+            auto bytes = Napi::Uint8Array::New(env, utf8.size());
+            if (!utf8.empty())
+            {
+                std::memcpy(bytes.Data(), utf8.data(), utf8.size());
+            }
+            return bytes;
+        }
+    };
+}
+
+namespace Babylon::Polyfills::TextEncoder
+{
+    void BABYLON_API Initialize(Napi::Env env)
+    {
+        ::TextEncoder::Initialize(env);
+    }
+}

Tests/UnitTests/Android/app/src/main/cpp/CMakeLists.txt

@@ -42,4 +42,5 @@ target_link_libraries(UnitTestsJNI
     PRIVATE gtest_main
     PRIVATE Blob
     PRIVATE TextDecoder
+    PRIVATE TextEncoder
     PRIVATE Performance)

Tests/UnitTests/CMakeLists.txt

@@ -61,6 +61,7 @@ target_link_libraries(UnitTests
     PRIVATE Blob
     PRIVATE Performance
     PRIVATE TextDecoder
+    PRIVATE TextEncoder
     ${ADDITIONAL_LIBRARIES})
 
 # See https://gitlab.kitware.com/cmake/cmake/-/issues/23543

Tests/UnitTests/Scripts/tests.ts

@@ -1356,6 +1356,44 @@ describe("TextDecoder", function () {
     });
 });
 
+describe("TextEncoder", function () {
+    it("should expose encoding === 'utf-8'", function () {
+        const encoder = new TextEncoder();
+        expect(encoder.encoding).to.equal("utf-8");
+    });
+
+    it("should encode an ASCII string into UTF-8 bytes", function () {
+        const encoder = new TextEncoder();
+        const bytes = encoder.encode("Hello");
+        expect(Array.from(bytes)).to.eql([72, 101, 108, 108, 111]);
+    });
+
+    it("should return an empty Uint8Array when called with no argument", function () {
+        const encoder = new TextEncoder();
+        const bytes = encoder.encode();
+        expect(bytes.length).to.equal(0);
+    });
+
+    it("should return an empty Uint8Array for undefined input", function () {
+        const encoder = new TextEncoder();
+        const bytes = encoder.encode(undefined);
+        expect(bytes.length).to.equal(0);
+    });
+
+    it("should encode a multi-byte UTF-8 string", function () {
+        const encoder = new TextEncoder();
+        // "é" is U+00E9 -> 0xC3 0xA9 in UTF-8
+        const bytes = encoder.encode("é");
+        expect(Array.from(bytes)).to.eql([0xC3, 0xA9]);
+    });
+
+    it("should encode a string containing a null byte", function () {
+        const encoder = new TextEncoder();
+        const bytes = encoder.encode("H\0i");
+        expect(Array.from(bytes)).to.eql([72, 0, 105]);
+    });
+});
+
 function runTests() {
     mocha.run((failures: number) => {
         // Test program will wait for code to be set before exiting

Tests/UnitTests/Shared/Shared.cpp

@@ -10,6 +10,7 @@
 #include <Babylon/Polyfills/XMLHttpRequest.h>
 #include <Babylon/Polyfills/Blob.h>
 #include <Babylon/Polyfills/TextDecoder.h>
+#include <Babylon/Polyfills/TextEncoder.h>
 #include <gtest/gtest.h>
 #include <arcana/threading/blocking_concurrent_queue.h>
 #include <atomic>
@@ -83,6 +84,7 @@ TEST(JavaScript, All)
         Babylon::Polyfills::XMLHttpRequest::Initialize(env);
         Babylon::Polyfills::Blob::Initialize(env);
         Babylon::Polyfills::TextDecoder::Initialize(env);
+        Babylon::Polyfills::TextEncoder::Initialize(env);
 
         auto setExitCodeCallback = Napi::Function::New(
             env, [&exitCodePromise](const Napi::CallbackInfo& info) {