fix(gateway): keep close reasons valid UTF-8

MonkeyLeeT · MonkeyLeeT · commit 3da2b1450435 · 2026-07-04T11:08:11.000-07:00
diff --git a/src/gateway/server/close-reason.test.ts b/src/gateway/server/close-reason.test.ts
@@ -0,0 +1,36 @@
+import { Buffer } from "node:buffer";
+import { describe, expect, test } from "vitest";
+import { truncateCloseReason } from "./close-reason.js";
+
+describe("truncateCloseReason", () => {
+  const multibyte = "\u{1f600}";
+
+  test("returns the fallback for an empty reason", () => {
+    expect(truncateCloseReason("")).toBe("invalid handshake");
+  });
+
+  test("keeps close reasons that are already within the byte limit", () => {
+    expect(truncateCloseReason("invalid connect params")).toBe("invalid connect params");
+  });
+
+  test("keeps a complete multibyte character when it fits at the byte limit", () => {
+    const reason = `${"x".repeat(116)}${multibyte} trailing text`;
+    const truncated = truncateCloseReason(reason);
+
+    expect(truncated).toBe(`${"x".repeat(116)}${multibyte}`);
+    expect(Buffer.byteLength(truncated)).toBe(120);
+  });
+
+  test("backs up before a split multibyte character", () => {
+    const reason = `${"x".repeat(118)}${multibyte} trailing text`;
+    const truncated = truncateCloseReason(reason);
+
+    expect(truncated).toBe("x".repeat(118));
+    expect(Buffer.byteLength(truncated)).toBeLessThanOrEqual(120);
+    expect(truncated).not.toContain("\uFFFD");
+  });
+
+  test("returns an empty reason when the byte limit cannot fit one character", () => {
+    expect(truncateCloseReason(multibyte, 3)).toBe("");
+  });
+});
diff --git a/src/gateway/server/close-reason.ts b/src/gateway/server/close-reason.ts
@@ -15,5 +15,35 @@ export function truncateCloseReason(reason: string, maxBytes = CLOSE_REASON_MAX_
   if (buf.length <= maxBytes) {
     return reason;
   }
-  return buf.subarray(0, maxBytes).toString();
+  const end = utf8BoundaryAtOrBefore(buf, maxBytes);
+  return buf.subarray(0, end).toString();
+}
+
+function utf8BoundaryAtOrBefore(buf: Buffer, maxBytes: number): number {
+  const limit = Math.max(0, Math.min(Math.floor(maxBytes), buf.length));
+  let end = 0;
+
+  while (end < limit) {
+    const byte = buf[end];
+    const charBytes = utf8SequenceLength(byte);
+    if (end + charBytes > limit) {
+      break;
+    }
+    end += charBytes;
+  }
+
+  return end;
+}
+
+function utf8SequenceLength(byte: number): number {
+  if (byte < 0x80) {
+    return 1;
+  }
+  if (byte < 0xe0) {
+    return 2;
+  }
+  if (byte < 0xf0) {
+    return 3;
+  }
+  return 4;
 }
