init: RAG 知识库服务初始提交

- 后端 API（Flask + Gunicorn）
- RAG 引擎（混合检索 + 云端 Reranker + 引用溯源）
- 文档解析（MinerU + 多格式支持）
- Docker 生产部署配置
- 排除前端项目、敏感配置、模型文件

repositories/__init__.py

@@ -0,0 +1,5 @@
+"""
+Repository 模块
+
+提供数据访问层抽象，支持开发环境和生产环境的不同实现。
+"""

repositories/session_repo.py

@@ -0,0 +1,65 @@
+"""
+会话存储接口
+
+定义会话管理的抽象接口，支持不同的存储实现。
+"""
+
+from abc import ABC, abstractmethod
+from typing import List, Dict, Optional
+
+
+class BaseSessionRepo(ABC):
+    """会话存储接口"""
+
+    @abstractmethod
+    def get_history(self, session_id: str) -> List[Dict]:
+        """
+        获取会话历史
+
+        Args:
+            session_id: 会话ID
+
+        Returns:
+            消息列表，每条消息包含 role、content 和可选的 metadata
+        """
+        pass
+
+    @abstractmethod
+    def add_message(self, session_id: str, role: str, content: str, metadata: Optional[Dict] = None) -> None:
+        """
+        添加消息到会话
+
+        Args:
+            session_id: 会话ID
+            role: 角色（user/assistant）
+            content: 消息内容
+            metadata: 可选的元数据（如图片、来源等）
+        """
+        pass
+
+    @abstractmethod
+    def create_session(self, user_id: str, title: str = "新对话") -> str:
+        """
+        创建新会话
+
+        Args:
+            user_id: 用户ID
+            title: 会话标题
+
+        Returns:
+            会话ID
+        """
+        pass
+
+    @abstractmethod
+    def get_user_sessions(self, user_id: str) -> List[Dict]:
+        """
+        获取用户的会话列表
+
+        Args:
+            user_id: 用户ID
+
+        Returns:
+            会话列表
+        """
+        pass

repositories/sqlite_session_repo.py

@@ -0,0 +1,82 @@
+"""
+SQLite 会话存储实现（开发环境）
+
+使用本地 SQLite 数据库存储会话数据。
+"""
+
+from .session_repo import BaseSessionRepo
+from typing import List, Dict, Optional
+import uuid
+from datetime import datetime
+import json
+
+
+class SQLiteSessionRepo(BaseSessionRepo):
+    """开发环境：SQLite存储"""
+
+    def __init__(self):
+        from data.db import get_connection
+        self.get_connection = get_connection
+
+    def get_history(self, session_id: str) -> List[Dict]:
+        """获取会话历史（包含 metadata）"""
+        with self.get_connection("session") as conn:
+            cursor = conn.execute(
+                "SELECT role, content, metadata FROM messages WHERE session_id = ? ORDER BY created_at",
+                (session_id,)
+            )
+            history = []
+            for row in cursor.fetchall():
+                msg = {"role": row[0], "content": row[1]}
+                # 解析 metadata
+                if row[2]:
+                    try:
+                        msg["metadata"] = json.loads(row[2])
+                    except (json.JSONDecodeError, TypeError):
+                        msg["metadata"] = {}
+                history.append(msg)
+            return history
+
+    def add_message(self, session_id: str, role: str, content: str, metadata: Optional[Dict] = None) -> None:
+        """添加消息到会话（支持 metadata）"""
+        metadata_str = json.dumps(metadata, ensure_ascii=False) if metadata else None
+        with self.get_connection("session") as conn:
+            conn.execute(
+                "INSERT INTO messages (session_id, role, content, metadata, created_at) VALUES (?, ?, ?, ?, ?)",
+                (session_id, role, content, metadata_str, datetime.now().isoformat())
+            )
+
+    def update_last_active(self, session_id: str) -> None:
+        """更新会话最后活跃时间"""
+        with self.get_connection("session") as conn:
+            conn.execute(
+                "UPDATE sessions SET last_active = ? WHERE session_id = ?",
+                (datetime.now().isoformat(), session_id)
+            )
+
+    def create_session(self, user_id: str, title: str = "新对话") -> str:
+        """创建新会话"""
+        session_id = str(uuid.uuid4())
+        with self.get_connection("session") as conn:
+            conn.execute(
+                "INSERT INTO sessions (session_id, user_id, created_at, last_active) VALUES (?, ?, ?, ?)",
+                (session_id, user_id, datetime.now().isoformat(), datetime.now().isoformat())
+            )
+        return session_id
+
+    def get_user_sessions(self, user_id: str) -> List[Dict]:
+        """获取用户的会话列表"""
+        with self.get_connection("session") as conn:
+            cursor = conn.execute(
+                "SELECT session_id, created_at, last_active FROM sessions WHERE user_id = ? ORDER BY last_active DESC",
+                (user_id,)
+            )
+            return [
+                {
+                    "session_id": row[0],
+                    "title": "对话",  # 默认标题
+                    "created_at": row[1],
+                    "last_active": row[2]
+                }
+                for row in cursor.fetchall()
+            ]

repositories/stateless_session_repo.py

@@ -0,0 +1,28 @@
+"""
+无状态会话存储实现（生产环境）
+
+生产环境不存储会话数据，历史由后端传入。
+"""
+
+from .session_repo import BaseSessionRepo
+from typing import List, Dict, Optional
+
+
+class StatelessSessionRepo(BaseSessionRepo):
+    """生产环境：无状态，不存储"""
+
+    def get_history(self, session_id: str) -> List[Dict]:
+        """生产环境：历史由后端传入，不查询"""
+        return []
+
+    def add_message(self, session_id: str, role: str, content: str, metadata: Optional[Dict] = None) -> None:
+        """生产环境：不存储消息"""
+        pass
+
+    def create_session(self, user_id: str, title: str = "新对话") -> str:
+        """生产环境：不创建会话，返回占位符"""
+        return "stateless"
+
+    def get_user_sessions(self, user_id: str) -> List[Dict]:
+        """生产环境：会话列表由后端管理"""
+        return []