修改引用逻辑，修改长期记忆bug

app/rag/__init__.py

@@ -13,7 +13,7 @@ RAG 检索与生成模块
     用户查询 → 多路改写 → 并行检索 → RRF 融合 → 重排序 → 返回父文档
 
 示例用法：
-    >>> from app.rag import RAGPipeline, create_rag_tool
+    >>> from app.rag.rag import RAGPipeline, create_rag_tool
     >>> from rag_indexer.builder import IndexBuilder, IndexBuilderConfig
     >>> from langchain_openai import ChatOpenAI
     >>>
@@ -34,16 +34,16 @@ RAG 检索与生成模块
     >>> rag_tool = create_rag_tool(retriever=retriever, llm=llm)
 """
 
-from .retriever import (
+from app.rag.retriever import (
     create_base_retriever,
     create_hybrid_retriever,
     create_qdrant_client,
 )
-from .reranker import LLaMaCPPReranker
-from .query_transform import MultiQueryGenerator
-from .fusion import reciprocal_rank_fusion
-from .pipeline import RAGPipeline
-from .tools import create_rag_tool, create_rag_tool_sync
+from app.rag.reranker import LLaMaCPPReranker
+from app.rag.query_transform import MultiQueryGenerator
+from app.rag.fusion import reciprocal_rank_fusion
+from app.rag.pipeline import RAGPipeline
+from app.rag.tools import create_rag_tool_sync
 
 
 __all__ = [
@@ -65,6 +65,5 @@ __all__ = [
     "RAGPipeline",
     
     # 工具创建（供 Agent 使用）
-    "create_rag_tool",
     "create_rag_tool_sync",
 ]

app/rag/fusion.py

@@ -1,6 +1,6 @@
 # rag/fusion.py
 
-from typing import List, Dict, Tuple
+from typing import List, Dict
 from langchain_core.documents import Document
 
 def reciprocal_rank_fusion(

app/rag/pipeline.py

@@ -2,15 +2,13 @@
 
 import asyncio
 import os
-from typing import List, Optional
+from typing import List
 from langchain_core.documents import Document
 from langchain_core.language_models import BaseLanguageModel
 
-from .retriever import create_qdrant_client  # 可能不需要直接使用
-from .reranker import LLaMaCPPReranker
-from .query_transform import MultiQueryGenerator
-from .fusion import reciprocal_rank_fusion
-
+from app.rag.reranker import LLaMaCPPReranker
+from app.rag.query_transform import MultiQueryGenerator
+from app.rag.fusion import reciprocal_rank_fusion
 
 class RAGPipeline:
     """

app/rag/reranker.py

@@ -2,9 +2,8 @@
 重排序器模块 (适配版)
 使用远程 llama.cpp 服务 (兼容 OpenAI Rerank API) 替代本地 Cross-Encoder
 """
-import os
 import requests
-from typing import List, Optional
+from typing import List
 from langchain_core.documents import Document
 
 class LLaMaCPPReranker:

app/rag/test.py

@@ -11,7 +11,6 @@ RAG 系统使用示例（重构版）
 import asyncio
 import sys
 import os
-from pathlib import Path
 
 from dotenv import load_dotenv
 
@@ -19,12 +18,12 @@ from dotenv import load_dotenv
 load_dotenv()
 
 # 添加项目根目录到路径
-sys.path.insert(0, os.path.join(os.path.dirname(__file__), ".."))
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), "../.."))
 
-from rag_indexer.index_builder import IndexBuilder, IndexBuilderConfig
+from rag_indexer.index_builder import IndexBuilderConfig
 from rag_indexer.splitters import SplitterType
-from rag.pipeline import RAGPipeline
-from rag.tools import create_rag_tool
+from app.rag.pipeline import RAGPipeline
+from app.rag.tools import create_rag_tool_sync
 from pydantic import SecretStr
 # 使用本地 LLM（通过 OpenAI 兼容接口）
 from langchain_openai import ChatOpenAI
@@ -32,7 +31,6 @@ from rag_core.retriever_factory import create_parent_retriever
 
 load_dotenv()
 
-
 def create_llm():
         """创建本地 vLLM 服务 LLM"""
         vllm_base_url = os.getenv(
@@ -60,8 +58,7 @@ async def demonstrate_full_pipeline():
     print("演示：固定流程 RAG 检索（多路改写 + RRF + 重排序 + 父文档）")
     print("=" * 60)
 
-
-    retriever = retriever = create_parent_retriever(collection_name="my_docs", search_k=5)
+    retriever = create_parent_retriever(collection_name="rag_documents", search_k=5)
 
     if retriever is None:
         print("错误：检索器未初始化，请确保索引已构建。")
@@ -103,7 +100,6 @@ async def demonstrate_full_pipeline():
         import traceback
         traceback.print_exc()
 
-
 async def demonstrate_tool_creation():
     """
     演示创建 RAG 工具（供 Agent 使用）
@@ -119,12 +115,11 @@ async def demonstrate_tool_creation():
     )
     retriever = retriever = create_parent_retriever(collection_name="rag_documents", search_k=5)
 
-
     # 2. 创建 LLM
     llm = create_llm()
 
     # 3. 创建工具
-    rag_tool = create_rag_tool(
+    rag_tool = create_rag_tool_sync(
         retriever=retriever,
         llm=llm,
         num_queries=3,
@@ -136,18 +131,16 @@ async def demonstrate_tool_creation():
     print(f"工具描述: {rag_tool.description[:100]}...")
 
     # 4. 模拟 Agent 调用工具
-    query = "请告诉我 RAG 系统的核心组件有哪些？"
+    query = "请告诉我 打虎英雄是谁？"
     print(f"\n模拟调用: {query}")
     print("-" * 40)
 
     result = await rag_tool.ainvoke({"query": query})
     print(result[:800] + "..." if len(result) > 800 else result)
 
-
 async def main():
     await demonstrate_full_pipeline()
     await demonstrate_tool_creation()
 
-
 if __name__ == "__main__":
     asyncio.run(main())

app/rag/tools.py

@@ -4,11 +4,11 @@ RAG 工具模块
 将检索功能封装为 LangChain Tool，供 Agent 调用。
 采用固定流水线：多路改写 → 并行检索 → RRF 融合 → 重排序 → 返回父文档。
 """
-from typing import Optional, Callable
+from typing import Callable
 from langchain_core.tools import tool
 from langchain_core.language_models import BaseLanguageModel
 from langchain_core.retrievers import BaseRetriever
-from .pipeline import RAGPipeline
+from app.rag.pipeline import RAGPipeline
 
 def create_rag_tool_sync(
     retriever: BaseRetriever,