feat: 更新首页信息处理模型，支持内存缓存与过期机制；添加全量目录同步脚本

backend/tests/test_home_information.py

@@ -35,6 +35,7 @@ async def setup(app, monkeypatch):
     async with app.state.sessions.begin() as db:
         account = await db.get(Account, 1)
         account.connection_status, account.wq_user_id = "connected", "TEST_USER"
+    monkeypatch.setattr(home, "now", lambda: datetime.fromisoformat("2026-09-12T12:00:00+00:00"))
     calls, state = [], {"rank": 25, "fail": set(), "agreement": "Participants must use GLOBAL region. Delay must be 1."}
 
     async def get(path, params=None, headers=None):
@@ -43,7 +44,7 @@ async def setup(app, monkeypatch):
             raise WqError("资源暂不可用")
         if path == "/users/self/messages":
             offset = params["offset"]
-            return page([{"id": f"msg{offset}", "title": "Update", "description": "<p>News</p><script>secret()</script><a href='javascript:alert(1)'>bad</a>", "dateCreated": "2026-09-11T12:00:00-04:00"}],
+            return page([{"id": f"msg{i}", "title": "Update", "description": "<p>News</p><script>secret()</script><a href='javascript:alert(1)'>bad</a>", "dateCreated": "2026-09-11T12:00:00-04:00"} for i in (range(10) if offset == 0 else [10])],
                         "https://api.worldquantbrain.com/users/self/messages?offset=1" if offset == 0 else None)
         if path == "/consultant/boards/leader":
             assert params == {"user": "TEST_USER"}
@@ -64,12 +65,12 @@ async def setup(app, monkeypatch):
 async def test_cache_pagination_failure_and_account_isolation(app, logged_in, monkeypatch):
     calls, state = await setup(app, monkeypatch)
     first = (await logged_in.get(PREFIX + "/messages")).json()
-    assert first["content"]["next_offset"] == 1
+    assert first["content"]["next_offset"] == 10
     assert "secret()" not in str(first) and "javascript:" not in str(first)
     assert first["analysis"] is None and not first["can_generate"]
-    assert (await logged_in.get(PREFIX + "/messages")).json() == first and len(calls) == 1
-    second = (await logged_in.get(PREFIX + "/messages?offset=1")).json()
-    assert second["content"]["items"][0]["id"] == "msg1"
+    assert (await logged_in.get(PREFIX + "/messages")).json() == first and len(calls) == 2
+    second = (await logged_in.get(PREFIX + "/messages?offset=10")).json()
+    assert second["content"]["items"][0]["id"] == "msg10"
     state["fail"].add("/users/self/messages")
     failed = (await logged_in.post(PREFIX + "/messages/refresh")).json()
     assert failed["stale"] and failed["error"]
@@ -193,7 +194,7 @@ async def test_failed_grounding_preserves_prior_analysis(app, logged_in, monkeyp
 async def test_concurrent_first_reads_are_coalesced(app, logged_in, monkeypatch):
     calls, _ = await setup(app, monkeypatch)
     results = await asyncio.gather(*[logged_in.get(PREFIX + "/messages") for _ in range(4)])
-    assert all(r.status_code == 200 for r in results) and len(calls) == 1
+    assert all(r.status_code == 200 for r in results) and len(calls) == 2
 
 
 async def test_event_sources_include_competitions_and_current_day(app, logged_in, monkeypatch):
@@ -298,3 +299,93 @@ async def test_shared_connection_change_invalidates_ai(app, logged_in, monkeypat
         setattr(await db.get(AISettings, 1), field, value)
     assert (await logged_in.get(PREFIX + '/competition' + QUERY)).json()['analysis']['outdated']
     assert calls == ['basic-model']
+
+
+@pytest.mark.parametrize('at,expected', [
+    ('2026-03-31T12:00:00-04:00', '2026-02-28T12:00:00-05:00'),
+    ('2024-03-31T12:00:00-04:00', '2024-02-29T12:00:00-05:00'),
+    ('2026-01-31T12:00:00-05:00', '2025-12-31T12:00:00-05:00'),
+])
+def test_recent_month_uses_eastern_calendar_month(monkeypatch, at, expected):
+    monkeypatch.setattr(home, 'now', lambda: datetime.fromisoformat(at))
+    assert home.month_start().isoformat() == expected
+
+
+async def test_recent_messages_scan_pages_filter_unknown_future_and_boundary(monkeypatch):
+    monkeypatch.setattr(home, 'now', lambda: datetime.fromisoformat('2026-03-31T12:00:00-04:00'))
+    calls = []
+    class Client:
+        async def get(self, path, params):
+            calls.append(params)
+            if params['offset'] == 0:
+                return page([{'id': 'old', 'dateCreated': '2026-02-28T11:59:59-05:00'},
+                             {'id': 'unknown'}, {'id': 'future', 'dateCreated': '2026-04-01T00:00:00Z'}],
+                            'https://api.worldquantbrain.com/users/self/messages?offset=3&limit=100')
+            return page([{'id': 'boundary', 'dateCreated': '2026-02-28T12:00:00-05:00'},
+                         {'id': 'today', 'dateCreated': '2026-03-31T12:00:00-04:00'}])
+    data = await home.source(Client(), 'messages', 'TEST_USER', 0, '')
+    assert [item['id'] for item in data['items']] == ['today', 'boundary']
+    assert calls == [{'offset': 0, 'limit': 100}, {'offset': 3, 'limit': 100}]
+
+
+async def test_messages_and_ai_never_persist_and_expire_without_auto_generation(app, logged_in, monkeypatch):
+    platform_calls, _ = await setup(app, monkeypatch)
+    await model_setup(app)
+    model_calls = []
+    def respond(messages, info):
+        return ModelResponse(parts=[ToolCallPart(info.output_tools[0].name, {'items': [], 'suggestions': []})])
+    @asynccontextmanager
+    async def factory(config, settings):
+        model_calls.append(config.model)
+        yield FunctionModel(respond)
+    app.state.ai.model_factory = factory
+    # Legacy rows must never be used; migration will remove them on deployment.
+    async with app.state.sessions.begin() as db:
+        db.add(HomeInformation(user_id='TEST_USER', resource='messages:0', content={'items': [{'title': 'legacy'}]}))
+    first = (await logged_in.get(PREFIX + '/messages')).json()
+    assert first['ephemeral'] and first['content']['total'] == 11
+    generated = await logged_in.post(PREFIX + '/messages/generate')
+    assert generated.status_code == 200, generated.text
+    assert generated.json()['analysis']['model'] == 'basic-model'
+    await logged_in.get(PREFIX + '/messages')
+    await logged_in.get(PREFIX + '/messages?offset=10')
+    assert len(platform_calls) == 2 and model_calls == ['basic-model']
+    async with app.state.sessions() as db:
+        saved = (await db.scalars(sa.select(HomeInformation))).all()
+        assert len(saved) == 1 and saved[0].content == {'items': [{'title': 'legacy'}]}
+        assert saved[0].analysis is None
+    monkeypatch.setattr(home, 'now', lambda: datetime.fromisoformat('2026-09-12T12:15:00+00:00'))
+    assert (await logged_in.post(PREFIX + '/messages/generate')).status_code == 409
+    reloaded = (await logged_in.get(PREFIX + '/messages')).json()
+    assert reloaded['analysis'] is None and len(platform_calls) == 4
+    assert model_calls == ['basic-model']
+
+
+async def test_message_window_rechecks_memory_and_removes_expired_interpretation(app, logged_in, monkeypatch):
+    await setup(app, monkeypatch)
+    monkeypatch.setattr(home, 'now', lambda: datetime.fromisoformat('2026-09-12T12:00:00+00:00'))
+    async def upstream(path, params=None):
+        return page([{'id': 'boundary', 'dateCreated': '2026-08-12T08:00:00-04:00'}])
+    monkeypatch.setattr(app.state.runner.client, 'get', upstream)
+    assert (await logged_in.get(PREFIX + '/messages')).json()['content']['total'] == 1
+    app.state.home_message_cache['TEST_USER'].analyses[0] = {'old': 'interpretation'}
+    monkeypatch.setattr(home, 'now', lambda: datetime.fromisoformat('2026-09-12T12:00:01+00:00'))
+    result = (await logged_in.get(PREFIX + '/messages')).json()
+    assert result['content']['total'] == 0 and result['analysis'] is None
+
+
+def test_message_cleanup_migration_only_removes_message_cache(tmp_path):
+    path = Path(__file__).parents[1] / 'migrations/versions/0017_remove_message_cache.py'
+    spec = importlib.util.spec_from_file_location('message_cleanup', path)
+    migration = importlib.util.module_from_spec(spec)
+    spec.loader.exec_module(migration)
+    engine = sa.create_engine(f'sqlite:///{tmp_path}/cleanup.db')
+    with engine.begin() as conn:
+        conn.exec_driver_sql('CREATE TABLE home_information (user_id TEXT, resource TEXT, content TEXT, analysis TEXT)')
+        for user, resource in [('A', 'messages:0'), ('B', 'messages:10'), ('A', 'events:'), ('A', 'competition:ARC2026')]:
+            conn.exec_driver_sql('INSERT INTO home_information VALUES (?, ?, ?, ?)', (user, resource, 'source', 'ai'))
+        with Operations.context(MigrationContext.configure(conn)):
+            migration.upgrade()
+            migration.upgrade()
+        assert conn.exec_driver_sql('SELECT resource FROM home_information ORDER BY resource').scalars().all() == ['competition:ARC2026', 'events:']
+    engine.dispose()