refactor: 重构RAG核心组件，简化代码结构和测试文件

rag_indexer/cli.py

@@ -6,10 +6,6 @@ import asyncio
 import logging
 import sys
 from pathlib import Path
-from dotenv import load_dotenv
-
-# 加载 .env 文件
-load_dotenv()
 
 from rag_indexer.index_builder import IndexBuilder, IndexBuilderConfig
 from rag_indexer.splitters import SplitterType
@@ -38,7 +34,7 @@ def get_input_path() -> Path:
     if len(sys.argv) > 1:
         return Path(sys.argv[1])
     # 默认测试路径（可按需修改）
-    return Path("data/user_docs/doublestory.txt")
+    return Path("data/corpus/三国演义.txt")
 
 
 async def main():

rag_indexer/index_builder.py

@@ -45,6 +45,11 @@ class IndexBuilderConfig:
     child_chunk_size: int = 200
     child_chunk_overlap: int = 20
     child_splitter_type: SplitterType = SplitterType.SEMANTIC   # 子块默认语义切分
+    # 子块语义切分参数
+    child_buffer_size: int = 1
+    child_breakpoint_threshold_type: str = "percentile"
+    child_breakpoint_threshold_amount: float = 90  # 降低阈值，让切分更激进
+    child_min_chunk_size: int = 50  # 降低最小块大小
 
     # 检索参数
     search_k: int = 5
@@ -86,7 +91,6 @@ class IndexBuilder:
         # 初始化向量存储（自动支持稠密+稀疏混合检索）
         self.vector_store = QdrantHybridStore(
             collection_name=config.collection_name,
-            embeddings=self.embeddings,
         )
         logger.info("✅ 混合检索向量存储初始化成功（稠密+BM25稀疏）")
 
@@ -125,6 +129,10 @@ class IndexBuilder:
             self.child_splitter = get_splitter(
                 SplitterType.SEMANTIC,
                 embeddings=self.embeddings,
+                buffer_size=cfg.child_buffer_size,
+                breakpoint_threshold_type=cfg.child_breakpoint_threshold_type,
+                breakpoint_threshold_amount=cfg.child_breakpoint_threshold_amount,
+                min_chunk_size=cfg.child_min_chunk_size,
                 **cfg.extra_splitter_kwargs
             )
         else:

rag_indexer/reset_qdrant.py

@@ -0,0 +1,34 @@
+#!/usr/bin/env python3
+"""
+删除 Qdrant 集合并重新索引
+"""
+
+import asyncio
+import os
+import sys
+
+from backend.rag_core import QdrantHybridStore
+
+
+async def delete_and_recreate():
+    """删除并重新创建集合"""
+    print("="*70)
+    print("删除旧集合并重新创建...")
+    print("="*70)
+    
+    vs = QdrantHybridStore(collection_name="rag_documents")
+    
+    # 删除旧集合
+    try:
+        vs.delete_collection()
+        print("✅ 旧集合已删除")
+    except Exception as e:
+        print(f"⚠️ 删除集合时出错（可能不存在）: {e}")
+    
+    # 重新创建
+    vs.create_collection()
+    print("✅ 新集合已创建")
+
+
+if __name__ == "__main__":
+    asyncio.run(delete_and_recreate())