Merge pull request #165 from ikmak/main

Enhance the `process_folder_complete` function

Author: LarFii

raganything/batch.py

@@ -104,13 +104,32 @@ async def process_folder_complete(
 
         async def process_single_file(file_path: Path):
             async with semaphore:
+                is_in_subdir = (
+                    lambda file_path, dir_path: len(
+                        file_path.relative_to(dir_path).parents
+                    )
+                    > 1
+                )(file_path, folder_path_obj)
+
                 try:
                     await self.process_document_complete(
                         str(file_path),
-                        output_dir=output_dir,
+                        output_dir=(
+                            output_dir
+                            if not is_in_subdir
+                            else str(
+                                output_path
+                                / file_path.parent.relative_to(folder_path_obj)
+                            )
+                        ),
                         parse_method=parse_method,
                         split_by_character=split_by_character,
                         split_by_character_only=split_by_character_only,
+                        file_name=(
+                            None
+                            if not is_in_subdir
+                            else str(file_path.relative_to(folder_path_obj))
+                        ),
                     )
                     return True, str(file_path), None
                 except Exception as e:

raganything/processor.py

@@ -1422,6 +1422,7 @@ async def process_document_complete(
         split_by_character: str | None = None,
         split_by_character_only: bool = False,
         doc_id: str | None = None,
+        file_name: str | None = None,
         **kwargs,
     ):
         """
@@ -1473,7 +1474,8 @@ async def process_document_complete(
 
         # Step 3: Insert pure text content with all parameters
         if text_content.strip():
-            file_name = os.path.basename(file_path)
+            if file_name is None:
+                file_name = os.path.basename(file_path)
             await insert_text_content(
                 self.lightrag,
                 input=text_content,