feat: add voice co-creation session skeleton

backend/tests/conftest.py

@@ -187,6 +187,18 @@ def isolated_story_audio_cache(tmp_path, monkeypatch):
     yield
 
 
+@pytest.fixture(autouse=True)
+def isolated_voice_session_storage(tmp_path, monkeypatch):
+    """Use an isolated directory for voice session assets."""
+
+    monkeypatch.setattr(
+        settings,
+        "voice_session_storage_dir",
+        str(tmp_path / "voice_sessions"),
+    )
+    yield
+
+
 @pytest.fixture
 def mock_text_provider():
     """Mock text generation."""

backend/tests/test_voice_sessions.py

@@ -0,0 +1,201 @@
+from unittest.mock import AsyncMock, patch
+
+from httpx import ASGITransport, AsyncClient
+
+from app.db.database import get_db
+from app.main import app
+from app.services.adapters.text.models import StoryOutput
+
+
+async def test_voice_session_create_and_fallback_turn_returns_audio(
+    db_session,
+    auth_token,
+):
+    async def override_get_db():
+        yield db_session
+
+    app.dependency_overrides[get_db] = override_get_db
+
+    with (
+        patch(
+            "app.services.voice_session_service.generate_story_content",
+            new_callable=AsyncMock,
+        ) as mock_generate,
+        patch(
+            "app.services.voice_session_service.text_to_speech",
+            new_callable=AsyncMock,
+        ) as mock_tts,
+    ):
+        mock_generate.return_value = StoryOutput(
+            mode="generated",
+            title="小猫去太空",
+            story_text="小猫跳上纸飞机，朝着月亮轻轻挥手。",
+            cover_prompt_suggestion="温暖儿童绘本封面，小猫与月亮",
+        )
+        mock_tts.return_value = b"fake-turn-audio"
+
+        transport = ASGITransport(app=app)
+        try:
+            async with AsyncClient(transport=transport, base_url="http://test") as client:
+                client.cookies.set("access_token", auth_token)
+
+                response = await client.post("/api/voice-sessions", json={})
+                assert response.status_code == 201
+                session_id = response.json()["id"]
+
+                response = await client.post(
+                    f"/api/voice-sessions/{session_id}/turns/fallback",
+                    json={"transcript_text": "我想听一个小猫去太空的故事"},
+                )
+                assert response.status_code == 202
+                turn_id = response.json()["turn_id"]
+
+                response = await client.get(
+                    f"/api/voice-sessions/{session_id}/turns/{turn_id}"
+                )
+                assert response.status_code == 200
+                turn_data = response.json()
+                assert turn_data["status"] == "audio_ready"
+                assert turn_data["detected_intent"] == "start_story"
+                assert turn_data["assistant_audio_ready"] is True
+                assert turn_data["assistant_audio_url"].endswith("/audio")
+
+                response = await client.get(turn_data["assistant_audio_url"])
+                assert response.status_code == 200
+                assert response.content == b"fake-turn-audio"
+                assert response.headers["content-type"] == "audio/mpeg"
+
+                response = await client.get(f"/api/voice-sessions/{session_id}")
+                assert response.status_code == 200
+                session_data = response.json()
+                assert session_data["status"] == "waiting_user"
+                assert session_data["working_title"] == "小猫去太空"
+                assert session_data["can_continue"] is True
+                assert session_data["can_finalize"] is True
+                assert len(session_data["recent_turns"]) == 1
+        finally:
+            app.dependency_overrides.clear()
+
+
+async def test_voice_session_correct_turn_and_finalize_to_story(
+    db_session,
+    auth_token,
+):
+    async def override_get_db():
+        yield db_session
+
+    app.dependency_overrides[get_db] = override_get_db
+
+    with (
+        patch(
+            "app.services.voice_session_service.generate_story_content",
+            new_callable=AsyncMock,
+        ) as mock_generate,
+        patch(
+            "app.services.voice_session_service.text_to_speech",
+            new_callable=AsyncMock,
+        ) as mock_tts,
+    ):
+        mock_generate.side_effect = [
+            StoryOutput(
+                mode="generated",
+                title="小猫去太空",
+                story_text="第一段故事：小猫坐着纸飞机飞向月亮。",
+                cover_prompt_suggestion="温暖儿童绘本封面，小猫飞向月亮",
+            ),
+            StoryOutput(
+                mode="generated",
+                title="小猫去太空",
+                story_text="第二段故事：它在月亮上遇见了会发光的新朋友。",
+                cover_prompt_suggestion="温暖儿童绘本封面，小猫与月亮朋友",
+            ),
+        ]
+        mock_tts.side_effect = [b"turn-1-audio", b"turn-2-audio"]
+
+        transport = ASGITransport(app=app)
+        try:
+            async with AsyncClient(transport=transport, base_url="http://test") as client:
+                client.cookies.set("access_token", auth_token)
+
+                response = await client.post("/api/voice-sessions", json={})
+                assert response.status_code == 201
+                session_id = response.json()["id"]
+
+                response = await client.post(
+                    f"/api/voice-sessions/{session_id}/turns/fallback",
+                    json={"transcript_text": "我想听一个小猫去太空的故事"},
+                )
+                assert response.status_code == 202
+
+                response = await client.post(
+                    f"/api/voice-sessions/{session_id}/turns/fallback",
+                    json={"transcript_text": "不要让它哭了，我想让它找到一个朋友"},
+                )
+                assert response.status_code == 202
+                turn_id = response.json()["turn_id"]
+
+                response = await client.get(
+                    f"/api/voice-sessions/{session_id}/turns/{turn_id}"
+                )
+                assert response.status_code == 200
+                assert response.json()["detected_intent"] == "correct_story"
+
+                response = await client.post(
+                    f"/api/voice-sessions/{session_id}/finalize",
+                    json={"save_story": True, "generate_cover": True},
+                )
+                assert response.status_code == 200
+                finalize_data = response.json()
+                story_id = finalize_data["story_id"]
+                assert finalize_data["status"] == "completed"
+
+                response = await client.get(f"/api/stories/{story_id}")
+                assert response.status_code == 200
+                story_data = response.json()
+                assert story_data["title"] == "小猫去太空"
+                assert "第一段故事" in story_data["story_text"]
+                assert "第二段故事" in story_data["story_text"]
+                assert story_data["generation_status"] == "partial_ready"
+
+                response = await client.get(f"/api/voice-sessions/{session_id}")
+                assert response.status_code == 200
+                session_data = response.json()
+                assert session_data["status"] == "completed"
+                assert session_data["final_story_id"] == story_id
+                assert session_data["can_continue"] is False
+        finally:
+            app.dependency_overrides.clear()
+
+
+async def test_voice_session_abandon_blocks_future_turns(
+    db_session,
+    auth_token,
+):
+    async def override_get_db():
+        yield db_session
+
+    app.dependency_overrides[get_db] = override_get_db
+
+    transport = ASGITransport(app=app)
+    try:
+        async with AsyncClient(transport=transport, base_url="http://test") as client:
+            client.cookies.set("access_token", auth_token)
+
+            response = await client.post("/api/voice-sessions", json={})
+            assert response.status_code == 201
+            session_id = response.json()["id"]
+
+            response = await client.post(
+                f"/api/voice-sessions/{session_id}/abandon",
+                json={"reason": "孩子先去吃饭了"},
+            )
+            assert response.status_code == 200
+            assert response.json()["status"] == "abandoned"
+
+            response = await client.post(
+                f"/api/voice-sessions/{session_id}/turns/fallback",
+                json={"transcript_text": "我们继续讲吧"},
+            )
+            assert response.status_code == 409
+    finally:
+        app.dependency_overrides.clear()