chore: snapshot current project updates

src/vlm/scanner.py

@@ -6,6 +6,7 @@ directory structure.
 """
 
 import csv
+from concurrent.futures import ThreadPoolExecutor, as_completed
 import json
 import logging
 import os
@@ -72,26 +73,61 @@ def scan_library(
     if progress_callback is not None:
         progress_callback(0, total_paths)
 
-    for index, file_path in enumerate(discovered_paths, start=1):
-        video_file = _create_video_file(
-            file_path,
-            root,
-            config.categories,
-            include_video_metadata=include_video_metadata,
-            metadata_cache=metadata_cache
-        )
-        if video_file is None:
+    max_workers = max(1, min(config.enrichment_max_concurrency, total_paths or 1))
+    use_parallel = include_video_metadata and total_paths > 1 and max_workers > 1
+
+    if use_parallel:
+        indexed_results: list[tuple[int, VideoFile]] = []
+        with ThreadPoolExecutor(max_workers=max_workers) as pool:
+            future_to_index = {
+                pool.submit(
+                    _create_video_file,
+                    file_path,
+                    root,
+                    config.categories,
+                    include_video_metadata,
+                    metadata_cache,
+                ): index
+                for index, file_path in enumerate(discovered_paths, start=1)
+            }
+            for completed_count, future in enumerate(as_completed(future_to_index), start=1):
+                index = future_to_index[future]
+                try:
+                    video_file = future.result()
+                except Exception as exc:
+                    logger.error(f"Failed to scan file #{index}: {exc}")
+                    video_file = None
+                if video_file is not None:
+                    indexed_results.append((index, video_file))
+                    file_count += 1
+                if progress_callback is not None:
+                    progress_callback(completed_count, total_paths)
+                if file_count % 100 == 0 and file_count > 0:
+                    logger.debug(f"Scanned {file_count} files so far...")
+
+        indexed_results.sort(key=lambda item: item[0])
+        video_files = [item[1] for item in indexed_results]
+    else:
+        for index, file_path in enumerate(discovered_paths, start=1):
+            video_file = _create_video_file(
+                file_path,
+                root,
+                config.categories,
+                include_video_metadata=include_video_metadata,
+                metadata_cache=metadata_cache
+            )
+            if video_file is None:
+                if progress_callback is not None:
+                    progress_callback(index, total_paths)
+                continue
+            video_files.append(video_file)
+            file_count += 1
+
             if progress_callback is not None:
                 progress_callback(index, total_paths)
-            continue
-        video_files.append(video_file)
-        file_count += 1
 
-        if progress_callback is not None:
-            progress_callback(index, total_paths)
-        
-        if file_count % 100 == 0:
-            logger.debug(f"Scanned {file_count} files so far...")
+            if file_count % 100 == 0:
+                logger.debug(f"Scanned {file_count} files so far...")
     
     logger.info(f"Scan complete. Found {file_count} video files")
     return video_files