updates and backlog

bin/vcmp

@@ -13,6 +13,7 @@ import psutil
 from tqdm import tqdm
 
 video_extensions: list[str] = ["mp4", "mkv", "avi", "wmv", "mov", "m4v", "ts", "flv", "mpg", "webm"]
+embedded_sub_extensions: list[str] = ["mkv"]
 
 total_files: int = 0
 
@@ -111,6 +112,38 @@ def ffmpeg_command(input: pathlib.Path, output: pathlib.Path, nice: int = 20) ->
         str(output)
     ]
 
+def extract_sub_command(input: pathlib.Path, output: pathlib.Path) -> list[str]:
+    return [
+        "ffmpeg",
+        "-progress", "pipe:1",  # Send progress info to stdout
+        "-i", str(input),
+        "-map", "0:s:0",
+        str(output)
+    ]
+
+def has_subs(input: pathlib.Path) -> bool:
+    cmd = [
+        "ffprobe",
+        "-loglevel", "error",
+        "-select_streams", "s",
+        "-show_entries", "stream=index:stream_tags=language",
+        "-of", "csv=p=0",
+        str(input)
+    ]
+    try:
+        result = subprocess.run(cmd, capture_output=True, text=True, check=True)
+        print(input)
+        print(result.stdout)
+        print("---")
+        if result.stdout:
+            return True
+        return False
+    except (subprocess.CalledProcessError, ValueError) as e:
+        logging.error(f"Error getting subs for {input}: {e}")
+        return False
+    except:
+        return False
+
 def clean_directory(directory_path: pathlib.Path) -> None:
     """Yeet all files and directories from a directory."""
     for item in directory_path.iterdir():
@@ -122,6 +155,9 @@ def clean_directory(directory_path: pathlib.Path) -> None:
 def is_video(path: pathlib.Path) -> bool:
     return path.suffix.lower()[1:] in video_extensions
 
+def is_video_with_embedded_subs(path: pathlib.Path) -> bool:
+    return path.suffix.lower()[1:] in embedded_sub_extensions
+
 def get_video_duration(input_file: pathlib.Path) -> float:
     cmd = [
         "ffprobe",
@@ -313,6 +349,26 @@ def process_directory(
         logging.info(f"Processed {count} files, {successful} successful")
     return count
 
+def process_subtitles(
+        *,
+        source_dir: pathlib.Path,
+        destination_dir: pathlib.Path,
+        temp_dir: pathlib.Path):
+    for dirpath, _dirnames, filenames in os.walk(source_dir):
+        for filename in filenames:
+            filepath = pathlib.Path(dirpath) / filename
+            # File might have been deleted since starting script
+            if not filepath.exists() or not is_video_with_embedded_subs(filepath):
+                logging.debug(f"Skipping {filepath}: not a video file or doesn't exist")
+                continue
+            if has_subs(filepath):
+                target_path = destination_dir / filepath.relative_to(source_dir)
+                target_path = target_path.with_suffix(".cmp.en.srt")
+                if target_path.exists():
+                    logging.debug(f"Skipping {filepath}: target already exists")
+                    continue
+                cmd = extract_sub_command(filepath, target_path)
+                result = subprocess.run(cmd, capture_output=True, text=True, check=True)
 
 def main():
     parser = argparse.ArgumentParser(
@@ -332,6 +388,7 @@ def main():
                       help='Remove source files after successful compression')
     parser.add_argument('-v', '--verbose', action='store_true',
                       help='Enable verbose logging output')
+    parser.add_argument('--sub', action='store_true', help='Process files for subtitle extraction first')
     parser.add_argument('-n', '--nice', type=lambda x: int(x) if 0 <= int(x) <= 39 else exec('raise ValueError("Nice value must be between 0 and 39")'),
                       default=20,
                       help='Nice value for process priority (default: 20). Valid values are 0 (highest priority) to 39 (lowest priority). Lower values use more system resources.')
@@ -352,22 +409,27 @@ def main():
     global ignore_busy
     ignore_busy = bool(args.ignore_busy)
 
-    logging.info(f"Starting video compression from {source_dir} to {destination_dir}")
-
-    global total_files
-    total_files = process_directory(
-        source_dir=source_dir,
+    if args.sub:
+        process_subtitles(source_dir=source_dir,
         destination_dir=destination_dir,
-        temp_dir=temp_dir,
-        count_only=True
-    )
+        temp_dir=temp_dir)
+    else:
+        logging.info(f"Starting video compression from {source_dir} to {destination_dir}")
 
-    process_directory(
-        source_dir=source_dir,
-        destination_dir=destination_dir,
-        temp_dir=temp_dir,
-        remove_source=args.remove_source,
-        nice=args.nice
-    )
+        global total_files
+        total_files = process_directory(
+            source_dir=source_dir,
+            destination_dir=destination_dir,
+            temp_dir=temp_dir,
+            count_only=True
+        )
+
+        process_directory(
+            source_dir=source_dir,
+            destination_dir=destination_dir,
+            temp_dir=temp_dir,
+            remove_source=args.remove_source,
+            nice=args.nice
+        )
 if __name__ == "__main__":
     main()