重排，多路查询

rag_indexer/__init__.py

@@ -23,7 +23,7 @@ Offline RAG Indexer module.
     >>> await builder.build_from_file("document.pdf")
 """
 
-from .IndexBuilder import IndexBuilder, IndexBuilderConfig, DocstoreConfig
+from .index_builder import IndexBuilder, IndexBuilderConfig, DocstoreConfig
 from .loaders import DocumentLoader
 from .splitters import SplitterType, get_splitter
 
@@ -39,7 +39,7 @@ __version__ = "2.0.0"
 
 __all__ = [
     # 核心构建器与配置
-    "IndexBuilder",
+    "index_builder",
     "IndexBuilderConfig",
     "DocstoreConfig",
     

rag_indexer/cli.py

@@ -7,7 +7,7 @@ import logging
 import sys
 from pathlib import Path
 
-from rag_indexer.IndexBuilder import IndexBuilder, IndexBuilderConfig
+from rag_indexer.index_builder import IndexBuilder, IndexBuilderConfig
 from rag_indexer.splitters import SplitterType
 
 logging.basicConfig(

rag_indexer/index_builder.py

@@ -19,7 +19,8 @@ from langchain_classic.retrievers import ParentDocumentRetriever
 
 from .loaders import DocumentLoader
 from .splitters import SplitterType, get_splitter, SemanticChunkerAdapter
-from rag_core import LlamaCppEmbedder, QdrantVectorStore, create_docstore
+from rag_core import LlamaCppEmbedder, QdrantVectorStore, create_docstore, create_parent_retriever
+
 
 logger = logging.getLogger(__name__)
 
@@ -113,43 +114,40 @@ class IndexBuilder:
         logger.info("使用单一 %s 切分器", self.config.splitter_type.value)
 
     def _init_parent_child_mode(self) -> None:
-        """父子块切分模式，初始化父块/子块切分器、文档存储和检索器。"""
         cfg = self.config
 
-        # 父块切分器（始终使用递归切分）
+        # 父块切分器（索引构建需要，必须保留）
         self.parent_splitter = RecursiveCharacterTextSplitter(
             chunk_size=cfg.parent_chunk_size,
             chunk_overlap=cfg.parent_chunk_overlap,
         )
 
-        # 子块切分器
+        # 子块切分器（索引构建需要）
         if cfg.child_splitter_type == SplitterType.SEMANTIC:
             self.child_splitter = get_splitter(
                 SplitterType.SEMANTIC,
                 embeddings=self.embeddings,
                 **cfg.extra_splitter_kwargs
             )
-            logger.info("子块使用语义切分器")
         else:
             self.child_splitter = RecursiveCharacterTextSplitter(
                 chunk_size=cfg.child_chunk_size,
                 chunk_overlap=cfg.child_chunk_overlap,
             )
-            logger.info("子块使用递归切分器，块大小=%d，重叠=%d",
-                       cfg.child_chunk_size, cfg.child_chunk_overlap)
 
-        # 初始化文档存储（用于父块）
+        # 文档存储
         self.docstore = self._create_or_use_docstore()
 
-        # 创建检索器
-        self.retriever = ParentDocumentRetriever(
-            vectorstore=self.vector_store.get_langchain_vectorstore(),
-            docstore=self.docstore,
-            child_splitter=self.child_splitter,   # type: ignore[arg-type]
+        # 使用工厂函数创建检索器，避免重复代码
+        self.retriever = create_parent_retriever(
+            collection_name=cfg.collection_name,
+            embeddings=self.embeddings,
             parent_splitter=self.parent_splitter,
-            search_kwargs={"k": cfg.search_k},
+            child_splitter=self.child_splitter,
+            docstore=self.docstore,
+            search_k=cfg.search_k,
         )
-        logger.info("ParentDocumentRetriever 初始化完成，父块大小=%d", cfg.parent_chunk_size)
+        logger.info("ParentDocumentRetriever 初始化完成")
 
     def _create_or_use_docstore(self) -> BaseStore:
         """创建或获取文档存储实例。"""

rag_indexer/test/test_refactored.py

@@ -10,7 +10,7 @@ import sys
 # 添加项目根目录到 Python 路径
 sys.path.insert(0, os.path.join(os.path.dirname(__file__), "..", ".."))
 
-from rag_indexer.IndexBuilder import IndexBuilder
+from rag_indexer.index_builder import IndexBuilder
 from rag_indexer.splitters import SplitterType
 
 async def test_index_builder():

rag_indexer/test/test_validate_index.py

@@ -129,7 +129,7 @@ async def check_postgres():
 
 async def test_search():
     """测试检索功能。"""
-    from rag_indexer.IndexBuilder import IndexBuilder, IndexBuilderConfig
+    from rag_indexer.index_builder import IndexBuilder, IndexBuilderConfig
     from rag_indexer.splitters import SplitterType
 
     print("\n" + "=" * 60)