464 lines
9.1 KiB
Markdown
464 lines
9.1 KiB
Markdown
# BackendServer API 文档
|
||
|
||
本文档描述后端服务器的 API 接口，供开发者扩展其他连接方式（如网络接口、WebSocket 等）。
|
||
|
||
## 概述
|
||
|
||
TrulyMEM 后端采用 **Packet 通信协议**，通过 `queue.Queue` 实现线程安全通信。后端在独立线程中运行，处理来自客户端的请求。
|
||
|
||
### 核心组件
|
||
|
||
| 组件 | 说明 |
|
||
|------|------|
|
||
| `BackendServer` | 后端服务器，独立线程运行 |
|
||
| `BackendClient` | 客户端封装，提供便捷方法 |
|
||
| `PacketType` | 请求类型枚举 |
|
||
| `Packet` | 数据包（请求） |
|
||
| `PacketResponse` | 数据包响应 |
|
||
|
||
---
|
||
|
||
## 请求类型 (PacketType)
|
||
|
||
```python
|
||
class PacketType(Enum):
|
||
PROCESS_MESSAGE = "process_message" # 处理消息
|
||
EXECUTE_TOOL = "execute_tool" # 执行工具
|
||
GET_STATUS = "get_status" # 获取状态
|
||
GET_CONFIG = "get_config" # 获取配置
|
||
SET_CONFIG = "set_config" # 设置配置
|
||
GET_HISTORY = "get_history" # 获取历史
|
||
SAVE_HISTORY = "save_history" # 保存历史
|
||
SHUTDOWN = "shutdown" # 关闭服务
|
||
```
|
||
|
||
---
|
||
|
||
## 数据包格式
|
||
|
||
### Packet
|
||
|
||
```python
|
||
@dataclass
|
||
class Packet:
|
||
id: str # 唯一标识
|
||
type: PacketType # 请求类型
|
||
body: Dict[str, Any] # 请求参数
|
||
response_queue: queue.Queue # 响应队列（可选）
|
||
created_at: float # 创建时间
|
||
```
|
||
|
||
### PacketResponse
|
||
|
||
```python
|
||
@dataclass
|
||
class PacketResponse:
|
||
id: str # 对应的请求ID
|
||
success: bool # 是否成功
|
||
data: Any = None # 返回数据
|
||
error: Optional[str] = None # 错误信息
|
||
```
|
||
|
||
---
|
||
|
||
## API 接口详情
|
||
|
||
### 1. PROCESS_MESSAGE - 处理消息
|
||
|
||
发送用户消息，AI 将处理并返回回复（可能包含工具调用）。
|
||
|
||
**请求参数：**
|
||
```python
|
||
body = {
|
||
"user_input": str # 用户输入的消息
|
||
}
|
||
```
|
||
|
||
**响应数据：**
|
||
```python
|
||
{
|
||
"success": True,
|
||
"content": str, # AI 回复内容
|
||
"tool_calls": [ # 工具调用记录
|
||
{
|
||
"name": str, # 工具名称
|
||
"arguments": dict,# 工具参数
|
||
"result": str # 工具执行结果
|
||
}
|
||
],
|
||
"rejected_tools": [ # 被拒绝的工具调用
|
||
(str, str) # (工具名, 拒绝原因)
|
||
]
|
||
}
|
||
```
|
||
|
||
**示例：**
|
||
```python
|
||
from core import BackendServer, BackendClient
|
||
|
||
server = BackendServer(db_path="graph_memory.db", use_embedded_db=True)
|
||
server.start(api_key="your-api-key")
|
||
|
||
client = BackendClient(server)
|
||
result = client.process_message("你好，请记住我的名字是小明")
|
||
|
||
if result.get("success"):
|
||
print(result["content"])
|
||
```
|
||
|
||
---
|
||
|
||
### 2. EXECUTE_TOOL - 执行工具
|
||
|
||
直接执行指定的记忆工具。
|
||
|
||
> **注意**：前端直接调用的工具**不受次数限制**，只有模型发起的工具调用才受限制。
|
||
|
||
**请求参数：**
|
||
```python
|
||
body = {
|
||
"tool_name": str, # 工具名称
|
||
"arguments": dict # 工具参数
|
||
}
|
||
```
|
||
|
||
**响应数据：**
|
||
```python
|
||
{
|
||
"success": True,
|
||
"result": str # 工具执行结果
|
||
}
|
||
```
|
||
|
||
**示例：**
|
||
```python
|
||
result = client.execute_tool("memory_recall", {"query_intent": "用户信息"})
|
||
```
|
||
|
||
---
|
||
|
||
### 3. GET_STATUS - 获取状态
|
||
|
||
获取后端运行状态。
|
||
|
||
**请求参数：**
|
||
```python
|
||
body = {} # 无参数
|
||
```
|
||
|
||
**响应数据：**
|
||
```python
|
||
{
|
||
"running": bool, # 后端是否运行中
|
||
"config": dict, # 当前配置
|
||
"graph_initialized": bool, # 图数据库是否初始化
|
||
"client_initialized": bool # API 客户端是否初始化
|
||
}
|
||
```
|
||
|
||
**示例：**
|
||
```python
|
||
status = client.get_status()
|
||
print(status["data"]["running"]) # True
|
||
```
|
||
|
||
---
|
||
|
||
### 4. GET_CONFIG - 获取配置
|
||
|
||
获取当前 API 配置。
|
||
|
||
**请求参数：**
|
||
```python
|
||
body = {} # 无参数
|
||
```
|
||
|
||
**响应数据：**
|
||
```python
|
||
{
|
||
"api_key": str, # API Key
|
||
"base_url": str # API Base URL
|
||
}
|
||
```
|
||
|
||
---
|
||
|
||
### 5. SET_CONFIG - 设置配置
|
||
|
||
更新 API 配置（API Key 和 Base URL）。
|
||
|
||
**请求参数：**
|
||
```python
|
||
body = {
|
||
"api_key": str, # API Key
|
||
"base_url": str, # API Base URL (默认: https://api.deepseek.com)
|
||
"model": str # 模型名称 (默认: deepseek-chat)
|
||
}
|
||
```
|
||
|
||
**响应数据：**
|
||
```python
|
||
{
|
||
"status": "config_updated"
|
||
}
|
||
```
|
||
|
||
**示例：**
|
||
```python
|
||
result = client.update_config(
|
||
api_key="sk-xxxxx",
|
||
base_url="https://api.deepseek.com",
|
||
model="deepseek-chat"
|
||
)
|
||
```
|
||
|
||
---
|
||
|
||
### 6. GET_HISTORY - 获取消息历史
|
||
|
||
获取保存的消息历史。
|
||
|
||
**请求参数：**
|
||
```python
|
||
body = {} # 无参数
|
||
```
|
||
|
||
**响应数据：**
|
||
```python
|
||
{
|
||
"history": list # 消息历史列表
|
||
}
|
||
```
|
||
|
||
---
|
||
|
||
### 7. SAVE_HISTORY - 保存消息历史
|
||
|
||
保存消息历史到内存。
|
||
|
||
**请求参数：**
|
||
```python
|
||
body = {
|
||
"messages": list # 消息列表
|
||
}
|
||
```
|
||
|
||
**响应数据：**
|
||
```python
|
||
{
|
||
"status": "history_saved"
|
||
}
|
||
```
|
||
|
||
---
|
||
|
||
### 8. SHUTDOWN - 关闭服务
|
||
|
||
关闭后端服务器。
|
||
|
||
**请求参数：**
|
||
```python
|
||
body = {} # 无参数
|
||
```
|
||
|
||
**响应数据：**
|
||
```python
|
||
{
|
||
"status": "shutdown"
|
||
}
|
||
```
|
||
|
||
---
|
||
|
||
## 使用示例
|
||
|
||
### 基础使用
|
||
|
||
```python
|
||
from core import BackendServer, BackendClient
|
||
|
||
# 1. 创建并启动后端
|
||
# config_file 默认: ~/.trulymem/config.json
|
||
server = BackendServer(
|
||
db_path="graph_memory.db",
|
||
use_embedded_db=True,
|
||
config_file=None # 可选，自定义配置路径
|
||
)
|
||
server.start(
|
||
api_key="your-api-key",
|
||
base_url="https://api.deepseek.com",
|
||
model="deepseek-chat" # 可选，模型名称
|
||
)
|
||
|
||
# 2. 创建客户端
|
||
client = BackendClient(server)
|
||
|
||
# 3. 发送消息
|
||
result = client.process_message("你好")
|
||
if result.get("success"):
|
||
print(result["content"])
|
||
|
||
# 4. 关闭
|
||
client.shutdown()
|
||
```
|
||
|
||
### 使用 Packet 协议
|
||
|
||
```python
|
||
import queue
|
||
from core import BackendServer, Packet, PacketType
|
||
|
||
server = BackendServer(config_file=None)
|
||
server.start(api_key="your-key", model="deepseek-chat")
|
||
|
||
# 创建请求包
|
||
response_queue = queue.Queue()
|
||
packet = Packet(
|
||
id="req-001",
|
||
type=PacketType.PROCESS_MESSAGE,
|
||
body={"user_input": "你好"},
|
||
response_queue=response_queue
|
||
)
|
||
|
||
# 发送请求
|
||
result = server.send(packet)
|
||
print(result.body)
|
||
|
||
# 关闭
|
||
server.shutdown()
|
||
```
|
||
|
||
---
|
||
|
||
## 扩展指南
|
||
|
||
### 扩展为 HTTP API
|
||
|
||
```python
|
||
from flask import Flask, request, jsonify
|
||
from core import BackendServer, BackendClient
|
||
|
||
app = Flask(__name__)
|
||
server = BackendServer()
|
||
client = BackendClient(server)
|
||
|
||
@app.route("/message", methods=["POST"])
|
||
def send_message():
|
||
data = request.json
|
||
result = client.process_message(data["message"])
|
||
return jsonify(result)
|
||
|
||
@app.route("/config", methods=["POST"])
|
||
def update_config():
|
||
data = request.json
|
||
result = client.update_config(data["api_key"], data.get("base_url"))
|
||
return jsonify(result)
|
||
|
||
@app.route("/status", methods=["GET"])
|
||
def get_status():
|
||
result = client.get_status()
|
||
return jsonify(result)
|
||
|
||
if __name__ == "__main__":
|
||
server.start()
|
||
app.run(port=8080)
|
||
```
|
||
|
||
### 扩展为 WebSocket
|
||
|
||
```python
|
||
import asyncio
|
||
import websockets
|
||
import json
|
||
from core import BackendServer, BackendClient
|
||
|
||
server = BackendServer()
|
||
client = BackendClient(server)
|
||
|
||
async def handler(websocket):
|
||
async for message in websocket:
|
||
data = json.loads(message)
|
||
msg_type = data.get("type")
|
||
|
||
if msg_type == "message":
|
||
result = client.process_message(data["content"])
|
||
elif msg_type == "config":
|
||
result = client.update_config(data["api_key"], data.get("base_url"))
|
||
elif msg_type == "status":
|
||
result = client.get_status()
|
||
else:
|
||
result = {"success": False, "error": "unknown type"}
|
||
|
||
await websocket.send(json.dumps(result))
|
||
|
||
async def main():
|
||
server.start()
|
||
async with websockets.serve(handler, "localhost", 8765):
|
||
await asyncio.Future()
|
||
|
||
asyncio.run(main())
|
||
```
|
||
|
||
---
|
||
|
||
## 线程安全说明
|
||
|
||
- `BackendServer` 使用 `threading.Lock` 保护共享资源
|
||
- 所有请求通过 `queue.Queue` 传递，线程安全
|
||
- 响应通过每个请求独立的响应队列返回
|
||
- 默认超时时间：30 秒
|
||
|
||
---
|
||
|
||
## 工具调用限制
|
||
|
||
### 限制范围
|
||
|
||
| 调用方式 | 是否受限 | 说明 |
|
||
|---------|---------|------|
|
||
| 模型发起的工具调用 | ✅ 受限 | 通过 `PROCESS_MESSAGE` 触发，模型自动调用工具 |
|
||
| 前端直接调用工具 | ❌ 不受限 | 通过 `EXECUTE_TOOL` 直接调用 |
|
||
|
||
### 限制规则（仅限模型发起）
|
||
|
||
| 类别 | 操作 | 每轮上限 |
|
||
|------|------|---------|
|
||
| 人设图 | 查询 | 1 次 |
|
||
| 人设图 | 修改 | 1 次 |
|
||
| 工作记忆链 | 查询 | 4 次 |
|
||
| 工作记忆链 | 修改 | 2 次 |
|
||
| 一般记忆 | 查询 | 20 次 |
|
||
| 一般记忆 | 修改 | 10 次 |
|
||
|
||
### 重置机制
|
||
|
||
- 每次调用 `PROCESS_MESSAGE` 时，计数器自动重置
|
||
- 前端直接调用 `EXECUTE_TOOL` 不会重置计数器
|
||
|
||
---
|
||
|
||
## 错误处理
|
||
|
||
所有 API 返回统一格式：
|
||
|
||
```python
|
||
# 成功
|
||
{
|
||
"success": True,
|
||
"data": {...}
|
||
}
|
||
|
||
# 失败
|
||
{
|
||
"success": False,
|
||
"error": "错误描述"
|
||
}
|
||
```
|
||
|
||
常见错误：
|
||
|
||
| 错误信息 | 说明 |
|
||
|---------|------|
|
||
| `API Key 未配置` | 未设置 API Key |
|
||
| `timeout` | 请求超时 |
|
||
| `工具调用被拒绝: ...` | 工具调用频率超限 | |