refact: add batch size for scan cmd

memos/cmds/library.py

@@ -169,11 +169,11 @@ def is_temp_file(filename):
     )
 
 
-async def loop_files(library_id, folder, folder_path, force, plugins):
+async def loop_files(library_id, folder, folder_path, force, plugins, batch_size):
     updated_file_count = 0
     added_file_count = 0
     scanned_files = set()
-    semaphore = asyncio.Semaphore(settings.batchsize)
+    semaphore = asyncio.Semaphore(batch_size)
 
     async with httpx.AsyncClient(timeout=60) as client:
         tasks = []
@@ -377,6 +377,7 @@ def scan(
     force: bool = False,
     plugins: List[int] = typer.Option(None, "--plugin", "-p"),
     folders: List[int] = typer.Option(None, "--folder", "-f"),
+    batch_size: int = typer.Option(1, "--batch-size", "-bs", help="Batch size for processing files"),
 ):
     # Check if both path and folders are provided
     if path and folders:
@@ -426,7 +427,7 @@ def scan(
             continue
 
         added_file_count, updated_file_count, scanned_files = asyncio.run(
-            loop_files(library_id, folder, folder_path, force, plugins)
+            loop_files(library_id, folder, folder_path, force, plugins, batch_size)
         )
         total_files_added += added_file_count
         total_files_updated += updated_file_count

memos/commands.py

@@ -106,6 +106,7 @@ def get_or_create_default_library():
     Ensure the library has at least one folder.
     """
     from .cmds.plugin import bind
+
     response = httpx.get(f"{BASE_URL}/libraries")
     if response.status_code != 200:
         print(f"Failed to retrieve libraries: {response.status_code} - {response.text}")
@@ -162,6 +163,9 @@ def scan_default_library(
     path: str = typer.Argument(None, help="Path to scan within the library"),
     plugins: List[int] = typer.Option(None, "--plugin", "-p"),
     folders: List[int] = typer.Option(None, "--folder", "-f"),
+    batch_size: int = typer.Option(
+        1, "--batch-size", "-bs", help="Batch size for processing files"
+    ),
 ):
     """
     Scan the screenshots directory and add it to the library if empty.
@@ -174,7 +178,12 @@ def scan_default_library(
 
     print(f"Scanning library: {default_library['name']}")
     scan(
-        default_library["id"], path=path, plugins=plugins, folders=folders, force=force
+        default_library["id"],
+        path=path,
+        plugins=plugins,
+        folders=folders,
+        force=force,
+        batch_size=batch_size,
     )
 
 

memos/config.py

@@ -71,8 +71,6 @@ class Settings(BaseSettings):
     # Embedding settings
     embedding: EmbeddingSettings = EmbeddingSettings()
 
-    batchsize: int = 1
-
     auth_username: str = "admin"
     auth_password: SecretStr = SecretStr("changeme")
 

memos/default_config.yaml

@@ -6,7 +6,6 @@ screenshots_dir: screenshots
 server_host: 0.0.0.0
 server_port: 8839
 
-batchsize: 1
 auth_username: admin
 auth_password: changeme
 default_plugins: