From 07822aaefdb249c2678bea03cca6c8db4863a650 Mon Sep 17 00:00:00 2001
From: Vincent Koc <vincentkoc@ieee.org>
Date: Sun, 2 Aug 2026 03:19:48 +0800
Subject: [PATCH] fix(ui): preserve Talk transcript surrogate bounds

---
 .../chat/realtime-talk-conversation.test.ts   | 23 +++++++++++++++++++
 .../pages/chat/realtime-talk-conversation.ts  |  7 +++---
 2 files changed, 27 insertions(+), 3 deletions(-)

diff --git a/ui/src/pages/chat/realtime-talk-conversation.test.ts b/ui/src/pages/chat/realtime-talk-conversation.test.ts
index 08a44cae000c..49cb5e1ab315 100644
--- a/ui/src/pages/chat/realtime-talk-conversation.test.ts
+++ b/ui/src/pages/chat/realtime-talk-conversation.test.ts
@@ -217,6 +217,29 @@ describe("realtime Talk conversation", () => {
     expect(state.entries[0]?.text.endsWith("NEWEST")).toBe(true);
   });
 
+  it.each([255, 256])(
+    "does not retain a lone high surrogate before a natural marker at offset %i",
+    (markerOffset) => {
+      let state = createRealtimeTalkConversationState();
+      const retainedText = "a".repeat(markerOffset - 1);
+
+      state = updateRealtimeTalkConversation(state, {
+        role: "assistant",
+        text: `${retainedText}\uD800\n…\n${"b".repeat(8_000)}NEWEST`,
+        final: true,
+        nowMs: 1,
+      });
+
+      const text = state.entries[0]?.text ?? "";
+      expect(text.length).toBeLessThanOrEqual(8_000);
+      expect(text.startsWith(`${retainedText}\n…\n`)).toBe(true);
+      expect(text.endsWith("NEWEST")).toBe(true);
+      expect(text).not.toMatch(
+        /(?:[\uD800-\uDBFF](?![\uDC00-\uDFFF])|(?<![\uD800-\uDBFF])[\uDC00-\uDFFF])/,
+      );
+    },
+  );
+
   it.each(["user", "assistant"] as const)(
     "bounds oversized final %s entries while retaining the newest text",
     (role) => {
diff --git a/ui/src/pages/chat/realtime-talk-conversation.ts b/ui/src/pages/chat/realtime-talk-conversation.ts
index 04fbd493b041..74c275db500c 100644
--- a/ui/src/pages/chat/realtime-talk-conversation.ts
+++ b/ui/src/pages/chat/realtime-talk-conversation.ts
@@ -285,9 +285,10 @@ function boundRealtimeConversationText(text: string): string {
   const hasBoundedPrefix =
     markerIndex >= CONVERSATION_ENTRY_PREFIX_CHARS - 1 &&
     markerIndex <= CONVERSATION_ENTRY_PREFIX_CHARS;
-  const prefix = hasBoundedPrefix
-    ? text.slice(0, markerIndex)
-    : sliceUtf16Safe(text, 0, CONVERSATION_ENTRY_PREFIX_CHARS);
+  const prefixEnd = hasBoundedPrefix ? markerIndex : CONVERSATION_ENTRY_PREFIX_CHARS;
+  // A natural marker can follow malformed provider text ending in a lone high
+  // surrogate. Keep that code unit out of the retained truncation boundary.
+  const prefix = sliceUtf16Safe(text, 0, prefixEnd).replace(/[\uD800-\uDBFF]$/, "");
   const tailChars =
     MAX_CONVERSATION_ENTRY_CHARS - prefix.length - CONVERSATION_ENTRY_TRUNCATION_MARKER.length;
   const tail = sliceUtf16Safe(text, -tailChars);