fix: TextEncoder#encodeInto spec compliance + perf gains (#10129)

2 files changed, 90 insertions, 0 deletions

diff --git a/cli/tests/text_encoder_into_perf.js b/cli/tests/text_encoder_into_perf.js
new file mode 100644
index 000000000..8d60e9f00
--- /dev/null
+++ b/cli/tests/text_encoder_into_perf.js
@@ -0,0 +1,34 @@
+const mixed = "@Ā๐😀";
+
+function generateRandom(bytes) {
+  let result = "";
+  let i = 0;
+  while (i < bytes) {
+    const toAdd = Math.floor(Math.random() * Math.min(4, bytes - i));
+    switch (toAdd) {
+      case 0:
+        result += mixed[0];
+        i++;
+        break;
+      case 1:
+        result += mixed[1];
+        i++;
+        break;
+      case 2:
+        result += mixed[2];
+        i++;
+        break;
+      case 3:
+        result += mixed[3];
+        result += mixed[4];
+        i += 2;
+        break;
+    }
+  }
+  return result;
+}
+
+const randomData = generateRandom(1024);
+const encoder = new TextEncoder();
+const targetBuffer = new Uint8Array(randomData.length * 4);
+for (let i = 0; i < 10_000; i++) encoder.encodeInto(randomData, targetBuffer);
diff --git a/cli/tests/unit/text_encoding_test.ts b/cli/tests/unit/text_encoding_test.ts
index 7a15c9376..42c221cb2 100644
--- a/cli/tests/unit/text_encoding_test.ts
+++ b/cli/tests/unit/text_encoding_test.ts
@@ -157,6 +157,62 @@ unitTest(function textEncodeInto3(): void {
   ]);
 });
 
+unitTest(function loneSurrogateEncodeInto(): void {
+  const fixture = "lone𝄞\ud888surrogate";
+  const encoder = new TextEncoder();
+  const bytes = new Uint8Array(20);
+  const result = encoder.encodeInto(fixture, bytes);
+  assertEquals(result.read, 16);
+  assertEquals(result.written, 20);
+  // deno-fmt-ignore
+  assertEquals(Array.from(bytes), [
+    0x6c, 0x6f, 0x6e, 0x65,
+    0xf0, 0x9d, 0x84, 0x9e,
+    0xef, 0xbf, 0xbd, 0x73,
+    0x75, 0x72, 0x72, 0x6f,
+    0x67, 0x61, 0x74, 0x65
+  ]);
+});
+
+unitTest(function loneSurrogateEncodeInto2(): void {
+  const fixture = "\ud800";
+  const encoder = new TextEncoder();
+  const bytes = new Uint8Array(3);
+  const result = encoder.encodeInto(fixture, bytes);
+  assertEquals(result.read, 1);
+  assertEquals(result.written, 3);
+  // deno-fmt-ignore
+  assertEquals(Array.from(bytes), [
+    0xef, 0xbf, 0xbd
+  ]);
+});
+
+unitTest(function loneSurrogateEncodeInto3(): void {
+  const fixture = "\udc00";
+  const encoder = new TextEncoder();
+  const bytes = new Uint8Array(3);
+  const result = encoder.encodeInto(fixture, bytes);
+  assertEquals(result.read, 1);
+  assertEquals(result.written, 3);
+  // deno-fmt-ignore
+  assertEquals(Array.from(bytes), [
+    0xef, 0xbf, 0xbd
+  ]);
+});
+
+unitTest(function swappedSurrogatePairEncodeInto4(): void {
+  const fixture = "\udc00\ud800";
+  const encoder = new TextEncoder();
+  const bytes = new Uint8Array(8);
+  const result = encoder.encodeInto(fixture, bytes);
+  assertEquals(result.read, 2);
+  assertEquals(result.written, 6);
+  // deno-fmt-ignore
+  assertEquals(Array.from(bytes), [
+    0xef, 0xbf, 0xbd, 0xef, 0xbf, 0xbd, 0x00, 0x00
+  ]);
+});
+
 unitTest(function textDecoderSharedUint8Array(): void {
   const ab = new SharedArrayBuffer(6);
   const dataView = new DataView(ab);