修改配置

rag_indexer/loaders.py

@@ -11,6 +11,9 @@ from langchain_core.documents import Document
 from unstructured.documents.elements import Element
 from unstructured.partition.auto import partition
 
+# 相对导入配置
+from .config import RAG_OCR_LANGUAGES, RAG_DOC_LANGUAGES
+
 logger = logging.getLogger(__name__)
 
 # 模块加载时设置一次环境变量，避免重复设置
@@ -47,8 +50,8 @@ class DocumentLoader:
         """
         self.extract_images = extract_images
         self.strategy = strategy
-        self.ocr_languages = ocr_languages or ["chi_sim", "eng"]
-        self.languages = languages or ["zh"]
+        self.ocr_languages = ocr_languages or RAG_OCR_LANGUAGES
+        self.languages = languages or RAG_DOC_LANGUAGES
         self.include_page_breaks = include_page_breaks
         self.pdf_infer_table_structure = pdf_infer_table_structure
         self.partition_kwargs = partition_kwargs or {}