fix(storage): 优化 SQL 存储兼容性

README.md

@@ -16,10 +16,10 @@ tRPC-Agent-Python provides an end-to-end foundation for agent building, orchestr
 
 - **Multi-paradigm agent orchestration**: Built-in orchestration supports `ChainAgent` / `ParallelAgent` / `CycleAgent` / `TransferAgent`, with `GraphAgent` for graph-based orchestration.
 - **Graph orchestration capability (`GraphAgent`)**: Use DSL to orchestrate `Agent` / `Tool` / `MCP` / `Knowledge` / `CodeExecutor` in one unified flow.
-- **Efficient integration with Python AI ecosystems**: Agent ecosystem extensions (`claude-agent-sdk` / `LangGraph`, etc.) / Tool ecosystem extensions (`mcp`, etc.) / Knowledge ecosystem extensions (`LangChain`, etc.) / Model ecosystem extensions (`LiteLLM`, etc.) / Memory ecosystem extensions (`Mem0`, etc.).
+- **Efficient integration with Python AI ecosystems**: Agent ecosystem extensions (`claude-agent-sdk` / `LangGraph`, etc.) / Tool ecosystem extensions (`mcp`, etc.) / Knowledge ecosystem extensions (`LangChain`, etc.) / Model ecosystem extensions (`LiteLLM`, etc.) / Memory ecosystem extensions (`Mem0`, `Mempalace`, etc.).
 - **Agent ecosystem extensions**: Supports `LangGraphAgent` / `ClaudeAgent` / `TeamAgent` (Agno-Like).
 - **Tool ecosystem extensions**: `FunctionTool` / File tools / `MCPToolset` / LangChain Tool / Agent-as-Tool.
-- **Complete memory capability (`Session` / `Memory`)**: `Session` manages messages and state within a single session, while `Memory` manages cross-session long-term memory and personalization. Persistence supports `InMemory` / `Redis` / `SQL`; `Memory` also supports `Mem0`.
+- **Complete memory capability (`Session` / `Memory`)**: `Session` manages messages and state within a single session, while `Memory` manages cross-session long-term memory and personalization. Persistence supports `InMemory` / `Redis` / `SQL`; `Memory` also supports `Mem0`、`Mempalace`.
 - **Production-grade knowledge capability**: Built on LangChain components with first-class RAG support.
 - **CodeExecutor extension capability**: Supports local / container executors for code execution and task grounding.
 - **Skills extension capability**: Supports `SKILL.md`-based skill systems for reusable capabilities and dynamic tooling.
@@ -84,7 +84,7 @@ pip install trpc-agent-py
 Install optional capabilities as needed:
 
 ```bash
-pip install trpc-agent-py[a2a,ag-ui,knowledge,agent-claude,mem0,langfuse]
+pip install trpc-agent-py[a2a,ag-ui,knowledge,agent-claude,mem0, Mempalace, langfuse]
 ```
 
 ### Develop Weather Agent
@@ -457,7 +457,7 @@ Related docs:
 This group helps you:
 
 - Session: manage per-session messages, summaries, and state
-- Memory: manage cross-session long-term memory (including Mem0)
+- Memory: manage cross-session long-term memory (including Mem0, Mempalace)
 - Knowledge: cover document loading, retrieval, RAG, and prompt templates
 
 ### 10. Serving and Protocols
@@ -545,12 +545,12 @@ The framework is organized in an event-driven architecture where each layer can
 - **Runner layer**: Unified execution entry, coordinating Session / Memory / Artifact services
 - **Tool layer**: FunctionTool / file tools / MCPToolset / Skill tools
 - **Model layer**: OpenAIModel / AnthropicModel / LiteLLMModel
-- **Memory layer**: SessionService / MemoryService / SessionSummarizer / Mem0MemoryService
+- **Memory layer**: SessionService / MemoryService / SessionSummarizer / Mem0MemoryService / MempalaceMemoryService
 - **Knowledge layer**: Production-grade LangChain-based knowledge and RAG capability
 - **Execution and skill layer**: CodeExecutor (local / container) / Skills
 - **Service layer**: FastAPI / A2A / AG-UI
 - **Observability layer**: OpenTelemetry tracing/metrics, integrable with platforms like Langfuse
-- **Ecosystem adapter layer**: claude-agent-sdk / mcp / LangChain / LiteLLM / Mem0 plugged into the main chain through model/tool/memory adapters
+- **Ecosystem adapter layer**: claude-agent-sdk / mcp / LangChain / LiteLLM / Mem0 / Mempalace plugged into the main chain through model/tool/memory adapters
 
 Key packages:
 

README.zh_CN.md

@@ -16,10 +16,10 @@ tRPC-Agent-Python 提供从 Agent 构建、编排、工具接入、会话记忆
 
 - **多范式 Agent 编排**：预设编排支持 ChainAgent / ParallelAgent / CycleAgent / TransferAgent，同时支持 GraphAgent 图编排
 - **图编排能力（GraphAgent）**：通过 DSL 统一编排 Agent / Tool / MCP / Knowledge / CodeExecutor
-- **高效接入 Python AI 生态扩展**：Agent 生态扩展（claude-agent-sdk / LangGraph 等）/ 工具生态扩展（mcp 等）/ 知识库生态扩展（LangChain 等）/ 模型生态扩展（LiteLLM 等）/ 记忆生态扩展（Mem0 等）
+- **高效接入 Python AI 生态扩展**：Agent 生态扩展（claude-agent-sdk / LangGraph 等）/ 工具生态扩展（mcp 等）/ 知识库生态扩展（LangChain 等）/ 模型生态扩展（LiteLLM 等）/ 记忆生态扩展（Mem0、Mempalace等）
 - **Agent 生态扩展**：支持 LangGraphAgent / ClaudeAgent / TeamAgent（Agno-Like）
 - **Tool 生态扩展**：FunctionTool / 文件工具 / MCPToolset / LangChain Tool / Agent-as-Tool
-- **完善的记忆能力（Session / Memory）**：Session 负责单会话内的消息与状态管理，Memory 负责跨会话长期记忆与个性化信息沉淀。持久化支持 InMemory / Redis / SQL，Memory 还支持 Mem0
+- **完善的记忆能力（Session / Memory）**：Session 负责单会话内的消息与状态管理，Memory 负责跨会话长期记忆与个性化信息沉淀。持久化支持 InMemory / Redis / SQL，Memory 还支持 Mem0、Mempalace
 - **生产级知识库能力**：知识库能力基于 LangChain 组件构建，支持 RAG 场景
 - **CodeExecutor 扩展能力**：支持本地 / 容器执行器，用于支持 Agent 的代码执行与任务落地能力
 - **Skills 扩展能力**：支持 SKILL.md 技能体系，用于支持 Agent 的技能复用与动态工具化能力
@@ -84,7 +84,7 @@ pip install trpc-agent-py
 按需安装扩展能力：
 
 ```bash
-pip install trpc-agent-py[a2a,ag-ui,knowledge,agent-claude,mem0,langfuse]
+pip install trpc-agent-py[a2a,ag-ui,knowledge,agent-claude,mem0, Mempalace, langfuse]
 ```
 
 
@@ -458,7 +458,7 @@ skill_tool_set = SkillToolSet(repository=repository, run_tool_kwargs=tool_kwargs
 这组示例可以帮你：
 
 - Session：管理单会话的消息、摘要与状态
-- Memory：管理跨会话长期记忆（含 Mem0）
+- Memory：管理跨会话长期记忆（含 Mem0, Mempalace）
 - Knowledge：覆盖文档加载、检索、RAG、提示模板等链路
 
 ### 10. 服务化与协议
@@ -546,12 +546,12 @@ skill_tool_set = SkillToolSet(repository=repository, run_tool_kwargs=tool_kwargs
 - **Runner 层**：统一执行入口，负责 Session/Memory/Artifact 等服务协同
 - **Tool 层**：FunctionTool / 文件工具 / MCPToolset / Skill 工具
 - **Model 层**：OpenAIModel / AnthropicModel / LiteLLMModel
-- **Memory 层**：SessionService / MemoryService / SessionSummarizer / Mem0MemoryService
+- **Memory 层**：SessionService / MemoryService / SessionSummarizer / Mem0MemoryService / MempalaceMemoryService
 - **Knowledge 层**：基于 LangChain 的生产级知识库能力（RAG）
 - **执行与技能层**：CodeExecutor（本地/容器）/ Skills
 - **服务层**：FastAPI / A2A / AG-UI
 - **观测层**：OpenTelemetry tracing/metrics，可对接 Langfuse 等平台
-- **生态适配层**：claude-agent-sdk / mcp / LangChain / LiteLLM / Mem0，通过模型/工具/记忆适配器接入主链路
+- **生态适配层**：claude-agent-sdk / mcp / LangChain / LiteLLM / Mem0 / MemoryService，通过模型/工具/记忆适配器接入主链路
 
 关键包一览：
 

examples/memory_service_with_sql/agent/agent.py

@@ -10,6 +10,8 @@
 from trpc_agent_sdk.models import OpenAIModel
 from trpc_agent_sdk.tools import FunctionTool
 from trpc_agent_sdk.tools import load_memory_tool
+from trpc_agent_sdk.types import GenerateContentConfig
+from trpc_agent_sdk.types import HttpOptions
 
 from .config import get_model_config
 from .prompts import INSTRUCTION
@@ -25,12 +27,18 @@ def _create_model() -> LLMModel:
 
 def create_agent() -> LlmAgent:
     """ Create an agent"""
+    generate_content_config = GenerateContentConfig(
+        http_options=HttpOptions(extra_body={"chat_template_kwargs": {
+                                     "enable_thinking": False
+                                 }}),
+    )
     agent = LlmAgent(
         name="assistant",
         description="A helpful assistant for conversation",
         model=_create_model(),  # You can change this to your preferred model
         instruction=INSTRUCTION,
         tools=[FunctionTool(get_weather_report), load_memory_tool],
+        generate_content_config=generate_content_config,
     )
     return agent
 

examples/session_service_with_sql/agent/agent.py

@@ -9,6 +9,8 @@
 from trpc_agent_sdk.models import LLMModel
 from trpc_agent_sdk.models import OpenAIModel
 from trpc_agent_sdk.tools import FunctionTool
+from trpc_agent_sdk.types import GenerateContentConfig
+from trpc_agent_sdk.types import HttpOptions
 
 from .config import get_model_config
 from .prompts import INSTRUCTION
@@ -24,12 +26,18 @@ def _create_model() -> LLMModel:
 
 def create_agent() -> LlmAgent:
     """ Create an agent"""
+    generate_content_config = GenerateContentConfig(
+        http_options=HttpOptions(extra_body={"chat_template_kwargs": {
+                                     "enable_thinking": False
+                                 }}),
+    )
     agent = LlmAgent(
         name="assistant",
         description="A helpful assistant for conversation",
         model=_create_model(),  # You can change this to your preferred model
         instruction=INSTRUCTION,
         tools=[FunctionTool(get_weather_report)],
+        generate_content_config=generate_content_config,
     )
     return agent
 

examples/session_service_with_sql/run_agent.py

@@ -76,10 +76,9 @@ async def run_weather_agent():
     ]
 
     for query in demo_queries:
-        # Use a new session for each query
-
         user_content = Content(parts=[Part.from_text(text=query)])
 
+        print(f"👤 User: {query}")
         print("🤖 Assistant: ", end="", flush=True)
         async for event in runner.run_async(user_id=user_id, session_id=current_session_id, new_message=user_content):
             # Check if event.content exists

tests/common/test_compatible.py

@@ -7,7 +7,7 @@
 
 Covers:
 - PY_310 version flag
-- checkenum() with standard enums, IntEnum, and fallback path
+- check_enum() with standard enums, IntEnum, and fallback path
 - OSDetector: platform detection, properties, get_os_name, get_os_info, __str__
 - OS_DETECTOR module-level singleton
 """
@@ -21,7 +21,7 @@
 
 import pytest
 
-from trpc_agent_sdk.common._compatible import OS_DETECTOR, OSDetector, PY_310, checkenum
+from trpc_agent_sdk.common._compatible import OS_DETECTOR, OSDetector, PY_310, check_enum
 
 
 # ---------------------------------------------------------------------------
@@ -40,7 +40,7 @@ def test_py310_matches_runtime(self):
 
 
 # ---------------------------------------------------------------------------
-# checkenum
+# check_enum
 # ---------------------------------------------------------------------------
 
 
@@ -63,32 +63,32 @@ class _FlagEnum(enum.Flag):
 
 
 class TestCheckenum:
-    """Tests for checkenum()."""
+    """Tests for check_enum()."""
 
     def test_valid_enum_member(self):
-        assert checkenum(_Color.RED, _Color) is True
+        assert check_enum(_Color.RED, _Color) is True
 
     def test_invalid_enum_member(self):
-        assert checkenum("yellow", _Color) is False
+        assert check_enum("yellow", _Color) is False
 
     def test_valid_int_enum_member(self):
-        assert checkenum(_Priority.HIGH, _Priority) is True
+        assert check_enum(_Priority.HIGH, _Priority) is True
 
     def test_invalid_int_enum_member(self):
-        assert checkenum(99, _Priority) is False
+        assert check_enum(99, _Priority) is False
 
     def test_valid_flag_enum_member(self):
-        assert checkenum(_FlagEnum.READ, _FlagEnum) is True
+        assert check_enum(_FlagEnum.READ, _FlagEnum) is True
 
     def test_string_value_is_found_by_value(self):
         # Python 3.12+ enum __contains__ matches by value
-        assert checkenum("red", _Color) is True
+        assert check_enum("red", _Color) is True
 
     def test_string_not_matching_any_value(self):
-        assert checkenum("magenta", _Color) is False
+        assert check_enum("magenta", _Color) is False
 
     def test_none_is_not_member(self):
-        assert checkenum(None, _Color) is False
+        assert check_enum(None, _Color) is False
 
     def test_fallback_to_members_values(self):
         """When ``in`` raises, falls back to __members__.values()."""
@@ -108,8 +108,8 @@ def __contains__(self, item):
             def __iter__(self):
                 raise TypeError("broken __iter__")
 
-        assert checkenum("a", _BadContains()) is True
-        assert checkenum("c", _BadContains()) is False
+        assert check_enum("a", _BadContains()) is True
+        assert check_enum("c", _BadContains()) is False
 
 
 # ---------------------------------------------------------------------------

tests/common/test_init.py

@@ -11,19 +11,19 @@
 from __future__ import annotations
 
 import trpc_agent_sdk.common as common_mod
-from trpc_agent_sdk.common import OS_DETECTOR, OSDetector, checkenum
+from trpc_agent_sdk.common import OS_DETECTOR, OSDetector, check_enum
 from trpc_agent_sdk.common._compatible import (
     OS_DETECTOR as _ORIG_OS_DETECTOR,
     OSDetector as _OrigOSDetector,
-    checkenum as _orig_checkenum,
+    check_enum as _orig_check_enum,
 )
 
 
 class TestPublicExports:
     """Ensure __init__.py re-exports the right objects."""
 
     def test_all_contains_expected_names(self):
-        assert set(common_mod.__all__) == {"OSDetector", "OS_DETECTOR", "checkenum"}
+        assert set(common_mod.__all__) == {"OSDetector", "OS_DETECTOR", "check_enum"}
 
     def test_os_detector_class_is_same_object(self):
         assert OSDetector is _OrigOSDetector
@@ -32,4 +32,4 @@ def test_os_detector_instance_is_same_object(self):
         assert OS_DETECTOR is _ORIG_OS_DETECTOR
 
     def test_checkenum_is_same_function(self):
-        assert checkenum is _orig_checkenum
+        assert check_enum is _orig_check_enum

tests/memory/test_sql_memory_service.py

@@ -278,17 +278,6 @@ async def test_store_skips_events_without_content(self):
         svc._sql_storage.add.assert_not_called()
         svc._sql_storage.commit.assert_not_called()
 
-    async def test_store_raises_on_non_session(self):
-        svc = SqlMemoryService.__new__(SqlMemoryService)
-        svc._memory_service_config = _make_config_no_ttl()
-        svc._sql_storage = _patch_sql_storage()
-        svc._SqlMemoryService__cleanup_task = None
-        svc._SqlMemoryService__cleanup_stop_event = None
-
-        with pytest.raises(TypeError, match="Content must be a Session"):
-            await svc.store_session("not a session")
-
-
 # ---------------------------------------------------------------------------
 # SqlMemoryService — search_memory
 # ---------------------------------------------------------------------------

tests/models/test_anthropic_model_ext.py

@@ -519,14 +519,5 @@ def test_part_with_inline_data_is_valid(self):
         request = LlmRequest(contents=[Content(parts=[part], role="user")])
         model.validate_request(request)
 
-    def test_part_with_no_content_raises(self):
-        """Part with no meaningful content raises ValueError."""
-        model = _model()
-        part = Part()
-        request = LlmRequest(contents=[Content(parts=[part], role="user")])
-        with pytest.raises(ValueError, match="Content parts must have"):
-            model.validate_request(request)
-
-
 if __name__ == "__main__":
     pytest.main([__file__, "-v"])

tests/models/test_openai_model.py

@@ -148,18 +148,6 @@ def test_validate_request_with_multiple_contents(self):
         # Should not raise
         model.validate_request(request)
 
-    def test_validate_request_with_parts_without_content(self):
-        """Test validating request with parts that have no actual content raises ValueError."""
-        model = OpenAIModel(model_name="gpt-4", api_key="test_key")
-
-        # Create a part with all None fields
-        part = Part()
-        content = Content(parts=[part], role="user")
-        request = LlmRequest(contents=[content], config=None, tools_dict={})
-
-        with pytest.raises(ValueError, match="Content parts must have"):
-            model.validate_request(request)
-
     def test_properties_and_config(self):
         """Test model properties and config."""
         model = OpenAIModel(

tests/sessions/test_sql_session_service.py

@@ -90,12 +90,39 @@ def test_from_event_with_function_call(self):
         storage_event = SessionStorageEvent.from_event(session, event)
         assert storage_event.content is not None
 
+    def test_from_event_drops_empty_parts(self):
+        session = Session(id="s1", app_name="app", user_id="user", save_key="k")
+        event = Event(
+            invocation_id="inv-1",
+            author="agent",
+            content=Content(parts=[Part()]),
+        )
+        storage_event = SessionStorageEvent.from_event(session, event)
+        assert storage_event.content is None
+
     def test_from_event_no_content(self):
         session = Session(id="s1", app_name="app", user_id="user", save_key="k")
         event = Event(invocation_id="inv-1", author="agent", actions=EventActions())
         storage_event = SessionStorageEvent.from_event(session, event)
         assert storage_event.content is None
 
+    def test_to_event_drops_legacy_empty_parts(self):
+        storage_event = SessionStorageEvent(
+            id="e1",
+            app_name="app",
+            user_id="user",
+            session_id="s1",
+            invocation_id="inv-1",
+            author="agent",
+            actions=EventActions(),
+            long_running_tool_ids=set(),
+            timestamp=datetime.now(),
+            model_flags=1,
+            content={"parts": [{}], "role": "model"},
+        )
+        event = storage_event.to_event()
+        assert event.content is None
+
     def test_long_running_tool_ids_property(self):
         session = Session(id="s1", app_name="app", user_id="user", save_key="k")
         event = _make_event()

tests/storage/test_sql_common.py

@@ -332,6 +332,14 @@ def test_process_bind_param_spanner(self):
         result = dpt.process_bind_param(value, dialect)
         assert pickle.loads(result) == value
 
+    def test_process_bind_param_mysql(self):
+        dpt = DynamicPickleType()
+        dialect = _make_dialect("mysql")
+        value = {"key": "value", "nums": [1, 2, 3]}
+        result = dpt.process_bind_param(value, dialect)
+        assert isinstance(result, bytes)
+        assert pickle.loads(result) == value
+
     def test_process_bind_param_non_spanner(self):
         dpt = DynamicPickleType()
         dialect = _make_dialect("sqlite")
@@ -352,6 +360,14 @@ def test_process_result_value_spanner(self):
         result = dpt.process_result_value(pickled, dialect)
         assert result == original
 
+    def test_process_result_value_mysql(self):
+        dpt = DynamicPickleType()
+        dialect = _make_dialect("mysql")
+        original = {"key": "value", "nums": [1, 2, 3]}
+        pickled = pickle.dumps(original)
+        result = dpt.process_result_value(pickled, dialect)
+        assert result == original
+
     def test_process_result_value_non_spanner(self):
         dpt = DynamicPickleType()
         dialect = _make_dialect("sqlite")