fix(study): address latest review feedback

MomiJiSan · MomiJiSan · commit 4d7ae335f690 · 2026-06-04T20:41:00.000+08:00
diff --git a/main_logic/core.py b/main_logic/core.py
@@ -844,6 +844,7 @@ def __init__(self, sync_message_queue, lanlan_name, lanlan_prompt):
         self._takeover_input_dispatcher: Optional[
             Callable[..., Awaitable[bool]]
         ] = None
+        self._latest_voice_transcript_request_id = ""
         # 由前端控制的Agent相关开关
         self.agent_flags = {
             'agent_enabled': False,
@@ -1975,6 +1976,30 @@ async def _dispatch_voice_transcript_bridge(
     ) -> str:
         """Let plugin-side voice filters decide whether to cancel or prime context."""
         session_snapshot = self.session
+        request_id_snapshot = str(request_id or "")
+
+        def _session_changed() -> bool:
+            if self.session is session_snapshot:
+                return False
+            logger.debug("[%s] voice bridge result ignored after session change", self.lanlan_name)
+            return True
+
+        def _request_stale() -> bool:
+            if not request_id_snapshot:
+                return False
+            latest_request_id = str(
+                getattr(self, "_latest_voice_transcript_request_id", "") or ""
+            )
+            if latest_request_id == request_id_snapshot:
+                return False
+            logger.debug(
+                "[%s] voice bridge result ignored after newer request latest=%s current=%s",
+                self.lanlan_name,
+                latest_request_id,
+                request_id_snapshot,
+            )
+            return True
+
         metadata: dict[str, Any] = {
             "session_type": type(session_snapshot).__name__ if session_snapshot else "",
             "voice_source": True,
@@ -1983,6 +2008,9 @@ async def _dispatch_voice_transcript_bridge(
         if request_id:
             metadata["request_id"] = request_id
 
+        if _session_changed() or _request_stale():
+            return ""
+
         try:
             result = await publish_voice_transcript_request_reliably(
                 self.lanlan_name,
@@ -1997,13 +2025,7 @@ async def _dispatch_voice_transcript_bridge(
         if not isinstance(result, dict) or not result:
             return ""
 
-        def _session_changed() -> bool:
-            if self.session is session_snapshot:
-                return False
-            logger.debug("[%s] voice bridge result ignored after session change", self.lanlan_name)
-            return True
-
-        if _session_changed():
+        if _session_changed() or _request_stale():
             return ""
 
         action = str(result.get("action") or "").strip()
@@ -2013,10 +2035,10 @@ def _session_changed() -> bool:
                 logger.debug("[%s] voice bridge cancel skipped: session has no cancel_response", self.lanlan_name)
                 return ""
             try:
-                if _session_changed():
+                if _session_changed() or _request_stale():
                     return ""
                 await cancel_response()
-                if _session_changed():
+                if _session_changed() or _request_stale():
                     return ""
                 logger.debug("[%s] voice bridge cancelled current response", self.lanlan_name)
             except asyncio.CancelledError:
@@ -2039,9 +2061,11 @@ def _session_changed() -> bool:
                 return ""
             skipped = bool(result.get("skipped", False))
             try:
-                if _session_changed():
+                if _session_changed() or _request_stale():
                     return ""
                 await prime_context(context_text, skipped=skipped)
+                if _session_changed() or _request_stale():
+                    return ""
                 logger.debug(
                     "[%s] voice bridge primed context len=%d skipped=%s",
                     self.lanlan_name,
@@ -2207,6 +2231,8 @@ async def handle_input_transcript(self, transcript: str, *, is_voice_source: boo
             if is_voice_source and transcript_text
             else ""
         )
+        if realtime_voice_request_id:
+            self._latest_voice_transcript_request_id = realtime_voice_request_id
 
         # 更新用户活动时间戳（用于主动搭话检测）
         self.last_user_activity_time = time.time()
diff --git a/plugin/plugins/study_companion/models.py b/plugin/plugins/study_companion/models.py
@@ -239,14 +239,6 @@ def to_dict(self) -> dict[str, Any]:
         return asdict(self)
 
 
-def _clamp_int(value: object, minimum: int, maximum: int, default: int) -> int:
-    try:
-        number = int(value)
-    except (TypeError, ValueError, OverflowError):
-        number = default
-    return max(minimum, min(maximum, number))
-
-
 @dataclass(slots=True)
 class AwarenessConfig:
     enabled: bool = False
@@ -261,18 +253,20 @@ class AwarenessConfig:
 
     def __post_init__(self) -> None:
         self.enabled = bool(self.enabled)
-        self.snapshot_interval_seconds = _clamp_int(
+        self.snapshot_interval_seconds = _clamp_int_or_default(
             self.snapshot_interval_seconds, 1, 60, 5
         )
-        self.context_window_minutes = _clamp_int(
+        self.context_window_minutes = _clamp_int_or_default(
             self.context_window_minutes, 1, 60, 5
         )
         mode = str(self.classify_mode or "title_first").strip().lower()
         self.classify_mode = (
             mode if mode in {"title_first", "ocr_text", "both"} else "title_first"
         )
-        self.image_max_bytes = _clamp_int(self.image_max_bytes, 10_240, 512_000, 65_536)
-        self.push_to_llm_interval_seconds = _clamp_int(
+        self.image_max_bytes = _clamp_int_or_default(
+            self.image_max_bytes, 10_240, 512_000, 65_536
+        )
+        self.push_to_llm_interval_seconds = _clamp_int_or_default(
             self.push_to_llm_interval_seconds, 30, 300, 300
         )
         push_mode = str(self.push_to_llm_mode or "read").strip().lower()
diff --git a/plugin/plugins/study_companion/static/i18n.js b/plugin/plugins/study_companion/static/i18n.js
@@ -143,8 +143,8 @@ const I18n = {
         el.setAttribute('placeholder', this.t(key, el.getAttribute('placeholder') || ''));
       }
     });
-    root.querySelectorAll('[data-i18n-aria-label]').forEach((el) => {
-      const key = el.getAttribute('data-i18n-aria-label');
+    root.querySelectorAll('[data-i18n-aria], [data-i18n-aria-label]').forEach((el) => {
+      const key = el.getAttribute('data-i18n-aria') || el.getAttribute('data-i18n-aria-label');
       if (key) {
         el.setAttribute('aria-label', this.t(key, el.getAttribute('aria-label') || ''));
       }
diff --git a/plugin/plugins/study_companion/static/index.html b/plugin/plugins/study_companion/static/index.html
@@ -9,7 +9,7 @@
   </head>
   <body>
     <main class="study-shell">
-      <section class="study-toolbar" data-i18n-aria-label="ui.label.study_controls" aria-label="Study companion controls">
+      <section class="study-toolbar" data-i18n-aria="ui.label.study_controls" aria-label="Study companion controls">
         <div>
           <h1 data-i18n="ui.title">Study Companion</h1>
           <p id="statusLine" data-i18n="ui.status.starting">Starting...</p>
@@ -20,13 +20,13 @@ <h1 data-i18n="ui.title">Study Companion</h1>
         </div>
       </section>
 
-      <section class="mode-strip" data-i18n-aria-label="ui.label.mode">
+      <section class="mode-strip" data-i18n-aria="ui.label.mode">
         <button id="modeCompanionBtn" type="button" data-mode="companion" data-i18n="status.mode.companion">Companion</button>
         <button id="modeInteractiveBtn" type="button" data-mode="interactive" data-i18n="status.mode.interactive">Interactive</button>
         <button id="modeTeachingBtn" type="button" data-mode="teaching" data-i18n="status.mode.teaching">Teaching</button>
       </section>
 
-      <section class="study-meta" data-i18n-aria-label="ui.label.study_state" aria-label="Study state">
+      <section class="study-meta" data-i18n-aria="ui.label.study_state" aria-label="Study state">
         <div class="meta-item">
           <span data-i18n="ui.label.screen">Screen</span>
           <strong id="screenType">-</strong>
@@ -41,7 +41,7 @@ <h1 data-i18n="ui.title">Study Companion</h1>
         </div>
       </section>
 
-      <section class="memory-panel" aria-live="polite" data-i18n-aria-label="ui.memory.title" aria-label="Memory Deck">
+      <section class="memory-panel" aria-live="polite" data-i18n-aria="ui.memory.title" aria-label="Memory Deck">
         <div class="section-heading">
           <h2 data-i18n="ui.memory.title">Memory Deck</h2>
           <span id="memoryDeckStatus">-</span>
@@ -57,15 +57,15 @@ <h2 data-i18n="ui.memory.title">Memory Deck</h2>
           <button id="memoryAddBtn" type="button" data-i18n="ui.button.save_card">Save Card</button>
         </div>
         <div id="memoryDueCard" class="memory-card">-</div>
-        <div class="memory-review-actions" data-i18n-aria-label="ui.memory.ratings_aria">
+        <div class="memory-review-actions" data-i18n-aria="ui.memory.ratings_aria">
           <button type="button" data-memory-rating="again" data-i18n="ui.button.rating.again">Again</button>
           <button type="button" data-memory-rating="hard" data-i18n="ui.button.rating.hard">Hard</button>
           <button type="button" data-memory-rating="good" data-i18n="ui.button.rating.good">Good</button>
           <button type="button" data-memory-rating="easy" data-i18n="ui.button.rating.easy">Easy</button>
         </div>
       </section>
 
-      <section class="study-workspace" data-i18n-aria-label="ui.label.study_workspace" aria-label="Study workspace">
+      <section class="study-workspace" data-i18n-aria="ui.label.study_workspace" aria-label="Study workspace">
         <label class="input-label" for="studyInput" data-i18n="ui.label.text">Text</label>
         <textarea id="studyInput" spellcheck="true" placeholder="Paste a concept, problem statement, or OCR text here." data-i18n-placeholder="ui.placeholder.input"></textarea>
         <div class="action-row">
@@ -74,7 +74,7 @@ <h2 data-i18n="ui.memory.title">Memory Deck</h2>
         </div>
       </section>
 
-      <section class="question-panel" aria-live="polite" data-i18n-aria-label="ui.label.question_panel" aria-label="Question panel">
+      <section class="question-panel" aria-live="polite" data-i18n-aria="ui.label.question_panel" aria-label="Question panel">
         <h2 data-i18n="ui.label.question">Question</h2>
         <pre id="questionText"></pre>
         <label class="input-label" for="answerInput" data-i18n="ui.label.answer">Answer</label>
@@ -85,7 +85,7 @@ <h2 data-i18n="ui.label.question">Question</h2>
         </div>
       </section>
 
-      <section class="reply-panel" aria-live="polite" data-i18n-aria-label="ui.label.reply_panel" aria-label="Reply panel">
+      <section class="reply-panel" aria-live="polite" data-i18n-aria="ui.label.reply_panel" aria-label="Reply panel">
         <h2 data-i18n="ui.label.reply">Reply</h2>
         <pre id="replyText"></pre>
       </section>
diff --git a/plugin/tests/unit/plugins/test_study_companion_phase9_ux.py b/plugin/tests/unit/plugins/test_study_companion_phase9_ux.py
@@ -37,6 +37,26 @@ def test_phase9_static_math_assets_are_local_and_registered() -> None:
     assert "trust: false" in renderer
 
 
+def test_phase9_static_ui_uses_standard_aria_i18n_attribute() -> None:
+    index = (PLUGIN_DIR / "static" / "index.html").read_text(encoding="utf-8")
+    i18n = (PLUGIN_DIR / "static" / "i18n.js").read_text(encoding="utf-8")
+
+    assert "data-i18n-aria-label" not in index
+    for key in (
+        "ui.label.study_controls",
+        "ui.label.mode",
+        "ui.label.study_state",
+        "ui.memory.title",
+        "ui.memory.ratings_aria",
+        "ui.label.study_workspace",
+        "ui.label.question_panel",
+        "ui.label.reply_panel",
+    ):
+        assert f'data-i18n-aria="{key}"' in index
+    assert "[data-i18n-aria]" in i18n
+    assert "getAttribute('data-i18n-aria')" in i18n
+
+
 def test_phase9_hosted_study_panel_uses_span_based_katex_rendering() -> None:
     source = (PLUGIN_DIR / "surfaces" / "study_panel.tsx").read_text(encoding="utf-8")
 
diff --git a/tests/unit/test_core_game_route_memory_contract.py b/tests/unit/test_core_game_route_memory_contract.py
@@ -122,6 +122,7 @@ def _make_manager():
     mgr.tts_handler_task = None
     mgr._takeover_active = False
     mgr._takeover_input_dispatcher = None
+    mgr._latest_voice_transcript_request_id = ""
     mgr.sent_responses = []
     mgr.user_activity = []
 
@@ -381,6 +382,7 @@ async def fake_publish(lanlan_name, transcript, *, metadata):
     assert calls[0][2]["request_id"].startswith("realtime-stt-")
     assert calls[0][2]["source"] == "realtime_stt"
     assert "voice_session_id" not in calls[0][2]
+    assert mgr._latest_voice_transcript_request_id == calls[0][2]["request_id"]
     assert session.cancelled == 1
     assert mgr._activity_tracker.voice_rms_count == 1
     assert mgr._activity_tracker.user_messages == []
@@ -389,6 +391,57 @@ async def fake_publish(lanlan_name, transcript, *, metadata):
     assert mgr.sync_message_queue.messages == []
 
 
+@pytest.mark.unit
+@pytest.mark.asyncio
+async def test_voice_transcript_bridge_skips_publish_for_stale_request(monkeypatch):
+    async def fail_publish(*_args, **_kwargs):
+        raise AssertionError("stale voice bridge request must not publish")
+
+    session = _FakeVoiceBridgeSession()
+    mgr = _make_transcript_manager()
+    mgr.session = session
+    mgr._latest_voice_transcript_request_id = "req-new"
+    monkeypatch.setattr(core_module, "publish_voice_transcript_request_reliably", fail_publish)
+
+    result = await core_module.LLMSessionManager._dispatch_voice_transcript_bridge(
+        mgr,
+        "screen echo",
+        request_id="req-old",
+    )
+
+    assert result == ""
+    assert session.cancelled == 0
+    assert session.prime_context_calls == []
+
+
+@pytest.mark.unit
+@pytest.mark.asyncio
+async def test_voice_transcript_bridge_ignores_result_that_becomes_stale(monkeypatch):
+    session = _FakeVoiceBridgeSession()
+    mgr = _make_transcript_manager()
+    mgr.session = session
+    mgr._latest_voice_transcript_request_id = "req-old"
+    calls = []
+
+    async def fake_publish(lanlan_name, transcript, *, metadata):
+        calls.append((lanlan_name, transcript, metadata))
+        mgr._latest_voice_transcript_request_id = "req-new"
+        return {"action": "cancel_response", "reason": "ocr_overlap"}
+
+    monkeypatch.setattr(core_module, "publish_voice_transcript_request_reliably", fake_publish)
+
+    result = await core_module.LLMSessionManager._dispatch_voice_transcript_bridge(
+        mgr,
+        "screen echo",
+        request_id="req-old",
+    )
+
+    assert result == ""
+    assert calls[0][2]["request_id"] == "req-old"
+    assert session.cancelled == 0
+    assert session.prime_context_calls == []
+
+
 @pytest.mark.unit
 @pytest.mark.asyncio
 async def test_voice_transcript_bridge_cancel_without_session_handler_uses_ordinary_flow(

Original file line number	Diff line number	Diff line change
`@@ -143,8 +143,8 @@ const I18n = {`
`143`	`143`	`el.setAttribute('placeholder', this.t(key, el.getAttribute('placeholder') \|\| ''));`
`144`	`144`	`}`
`145`	`145`	`});`
`146`		`- root.querySelectorAll('[data-i18n-aria-label]').forEach((el) => {`
`147`		`- const key = el.getAttribute('data-i18n-aria-label');`
	`146`	`+ root.querySelectorAll('[data-i18n-aria], [data-i18n-aria-label]').forEach((el) => {`
	`147`	`+ const key = el.getAttribute('data-i18n-aria') \|\| el.getAttribute('data-i18n-aria-label');`
`148`	`148`	`if (key) {`
`149`	`149`	`el.setAttribute('aria-label', this.t(key, el.getAttribute('aria-label') \|\| ''));`
`150`	`150`	`}`