feat(ffmpeg_handling): remove subs with remove_russian

add remove stream function
change remove audio and remove subs to use this function
fix: show_metadata returns default values and return_tags is given as
parameter

media_editing/ffmpeg_handling.py

@@ -520,91 +520,31 @@ def cmd_convert_video(args):
     )
     return True, out
 
-
+from collections.abc import Callable
-def cmd_remove_subs(args):
+def remove_streams(
-    """
+    args,
-    Strips all subtitle data.
+    remove_indices: set[int] | list[int],
-
+    *,
-    FFmpeg Equivalent:
+    output_subtitle_codec: bool = False,
-    ffmpeg -i [input] -sn -c copy [output]
+) -> tuple[bool, str]:
-
-    Parameters:
-    - sn: Disables subtitle recording.
-    """
     probe = ffmpeg.probe(str(args.input))
-
+    remove_indices = set(remove_indices)
-    cmd = [
-        "ffmpeg",
-        "-y",
-        "-i",
-        str(args.input),
-    ]
-
-    if args.num is None:
-        # remove all subtitle streams
-        for stream in probe["streams"]:
-            if stream["codec_type"] != "subtitle":
-                cmd.extend(["-map", f"0:{stream['index']}"])
-    else:
-        # remove only specified subtitle streams
-        for stream in probe["streams"]:
-            if stream["codec_type"] == "subtitle" and stream["index"] in args.num:
-                continue
-
-            cmd.extend(["-map", f"0:{stream['index']}"])
-
-    cmd.extend(
-        [
-            "-c",
-            "copy",
-            "-map_metadata",
-            "0",
-            str(args.output),
-        ]
-    )
-
-    logger.debug("%s", " ".join(cmd))
-    if args.dry_run:
-        logger.debug(f"Dry run: {' '.join(cmd)}")
-        return True, "nooutputneeded"
-    try:
-        subprocess.run(
-            cmd,
-            capture_output=True,
-            text=True,
-            check=True,
-        )
-    except subprocess.CalledProcessError as e:
-        logger.error("Command: %s", " ".join(e.cmd))
-        logger.error("Return code: %s", e.returncode)
-        logger.error("STDOUT:\n%s", e.stdout)
-        logger.error("STDERR:\n%s", e.stderr)
-        raise
-    return True, "nooutputneeded"
-
-
-def cmd_remove_audio_track(args):
-    """
-    1. Removes a specific audio track from the video (Main Output).
-    """
-
-    probe = ffmpeg.probe(str(args.input))
-    remove_indices = []
-    for audio_index in args.audio_index:
-        remove_indices.append(select_audio_stream(args.input, audio_index))
 
     remaining_audio = [
         s
         for s in probe["streams"]
-        if s["codec_type"] == "audio" and s["index"] not in remove_indices
+        if s["codec_type"] == "audio"
+        and s["index"] not in remove_indices
     ]
-    if not remaining_audio:
-        logger.error("No audiotracks remaining")
-        return False, "nooutputneeded"
-    remaining_audio = translate_audio_tracks(remaining_audio)
 
-    removed_was_default = any(
+    if not remaining_audio:
-        s["index"] in remove_indices and s.get("disposition", {}).get("default", 0) == 1
+        logger.warning("No audiotracks remaining")
+        return False, "nooutputneeded"
+
+    removed_default_audio = any(
+        s["codec_type"] == "audio"
+        and s["index"] in remove_indices
+        and s.get("disposition", {}).get("default", 0) == 1
         for s in probe["streams"]
     )
 
@@ -625,13 +565,22 @@ def cmd_remove_audio_track(args):
             "copy",
             "-c:a",
             "copy",
-            "-c:s",
-            get_scodec(args.output),
         ]
     )
 
-    if removed_was_default and remaining_audio:
+    if output_subtitle_codec:
-        # First remaining audio stream in the output becomes default
+        cmd.extend(["-c:s", get_scodec(args.output)])
+    else:
+        cmd.extend(["-c:s", "copy"])
+
+    cmd.extend(
+        [
+            "-map_metadata",
+            "0",
+        ]
+    )
+
+    if removed_default_audio:
         cmd.extend(
             [
                 "-disposition:a",
@@ -642,10 +591,13 @@ def cmd_remove_audio_track(args):
         )
 
     cmd.append(str(args.output))
+
     logger.trace(" ".join(cmd))
+
     if args.dry_run:
-        logger.debug(f"Dry run: {' '.join(cmd)}")
+        logger.debug("Dry run: %s", " ".join(cmd))
         return True, "nooutputneeded"
+
     try:
         result = subprocess.run(
             cmd,
@@ -653,20 +605,47 @@ def cmd_remove_audio_track(args):
             text=True,
             check=True,
         )
+
         logger.debug("FFmpeg subprocess successful")
         logger.trace(result.stdout)
+
         return True, "nooutputneeded"
 
     except subprocess.CalledProcessError as e:
-        logger.error(
+        logger.error("Command: %s", " ".join(e.cmd))
-            "Command failed (%s):\n%s\n\nstderr:\n%s",
+        logger.error("Return code: %s", e.returncode)
-            e.returncode,
+        logger.error("STDOUT:\n%s", e.stdout)
-            " ".join(map(str, e.cmd)),
+        logger.error("STDERR:\n%s", e.stderr)
-            e.stderr,
+        raise
-        )
-        raise RuntimeError(f"Command failed: {' '.join(e.cmd)}") from e
 
+def cmd_remove_subs(args):
+    probe = ffmpeg.probe(str(args.input))
 
+    if args.num is None:
+        remove_indices = {
+            s["index"]
+            for s in probe["streams"]
+            if s["codec_type"] == "subtitle"
+        }
+    else:
+        remove_indices = set(args.num)
+
+    return remove_streams(
+        args=args,
+        remove_indices=remove_indices,
+    )
+
+def cmd_remove_audio_track(args):
+    remove_indices = {
+        select_audio_stream(args.input, audio_index)
+        for audio_index in args.audio_index
+    }
+
+    return remove_streams(
+        args=args,
+        remove_indices=remove_indices,
+        output_subtitle_codec=True,
+    )
 def cmd_extract_audio_track(args):
     """
     1. Saves that specific audio track as a standalone MP3 (Extra Output).
@@ -950,20 +929,20 @@ def cmd_set_primary_audio(args):
 
     return True, out
 
-
+def cmd_show_metadata(args, show_metadata=False, return_tags=None):
-def cmd_show_metadata(args, show_metadata=False):
-    """
-    Prints metadata directly from ffprobe output.
-    """
-
     probe = ffmpeg.probe(str(args.input))
     format_tags = probe.get("format", {}).get("tags", {})
 
+    if return_tags is not None:
+        return_tags.clear()
+        return_tags.update(format_tags)
+
     if show_metadata:
         logger.info("=== FORMAT METADATA ===")
         for k, v in format_tags.items():
             logger.info(f"{k}: {v}")
 
+
     # logger.info(f"\n=== STREAM METADATA ===")
     # for i, stream in enumerate(probe.get("streams", [])):
     #    logger.info(f"\n--- Stream {i} ({stream.get('codec_type')}) ---")
@@ -974,7 +953,7 @@ def cmd_show_metadata(args, show_metadata=False):
     #    else:
     #        for k, v in tags.items():
     #            logger.info(f"{k}: {v}")
-    return False, format_tags
+    return False, "nooutputneeded"
 
 
 def cmd_set_metadata(args):
@@ -1121,34 +1100,55 @@ def cmd_apply_chapters(args):
     return True, "nooutputneeded"
 
 
-def cmd_remove_russian_from_db(args_in):
+def cmd_remove_russian_from_db(args):
     """
     Find all media with russian audio
     remove russian audio tracks
     """
-    file = args_in.input
+
+    file = args.input
+
     logger.trace("Run remove russian cmd")
+
     try:
         audio_tracks = translate_audio_tracks(get_audio_tracks(file))
+        subtitle_tracks = translate_audio_tracks(get_subtitle_tracks(file))
     except ValueError as e:
         logger.warning("%s", e)
         return False, "nooutputneeded"
 
-    logger.trace(f"Remove Russian from File: {file.name}")
+    logger.trace("Remove Russian from File: %s", file.name)
-    logger.trace(f"With Audio tracks: {audio_tracks}")
+
-    remove_tracks = []
+    remove_indices = {
+        track["number"]
+        for track in audio_tracks + subtitle_tracks
+        if is_russian_track(track)
+    }
+
     for track in audio_tracks:
-        if is_russian_track(track):
+        if track["number"] in remove_indices:
-            remove_tracks.append(track["number"])
+            logger.debug(
-            logger.debug(f"Removing russian track {track['number']} from {file}")
+                "Removing russian audio track %s from %s",
-    args = args_in
+                track["number"],
-    args.audio_index = remove_tracks
+                file,
-    if remove_tracks:
+            )
-        cmd_remove_audio_track(args)
+
-        return True, "nooutputneeded"
+    for track in subtitle_tracks:
-    else:
+        if track["number"] in remove_indices:
+            logger.debug(
+                "Removing russian subtitle track %s from %s",
+                track["number"],
+                file,
+            )
+
+    if not remove_indices:
         return False, "nooutputneeded"
 
+    return remove_streams(
+        args=args,
+        remove_indices=remove_indices,
+        )
+
 
 def cmd_test(args):
     """
@@ -1289,7 +1289,8 @@ def cmd_test(args):
         )
         args = ArgsBuffer()
         args.input = file_path
-        tags = cmd_show_metadata(args)
+        tags = {}
+        cmd_show_metadata(args,return_tags=tags)
 
         return_value = True
 
@@ -1398,17 +1399,18 @@ def cmd_test(args):
             audio_tracks = translate_audio_tracks(get_audio_tracks(file_path))
             logger.debug(f"Audio tracks: {audio_tracks}")
             for audio_track in audio_tracks:
-                if is_russian_track(audio_track["description"]):
+                if is_russian_track(audio_track):
                     logger.error(
                         f"Found russian track: {audio_track} in file {file_path}"
                     )
                     return return_value and False
 
-            # logger.debug(f"Subtitle tracks: {subtitle_tracks}")
+            subtitle_tracks = translate_audio_tracks(get_audio_tracks(file_path))
-            # for subtitle_track in subtitle_tracks:
+            logger.debug(f"Subtitle tracks: {subtitle_tracks}")
-            #    if is_russian_track(audio_track["description"]):
+            for subtitle_track in subtitle_tracks:
-            #        logger.error(f"Found russian track: {subtitle_track} in file {file}")
+               if is_russian_track(subtitle_track):
-            #        return return_value and False
+                   logger.error(f"Found russian track: {subtitle_track} in file {file_path}")
+                   return return_value and False
 
         logger.info(f"{green_tag('PASS')} {file_path}")
         return return_value
@@ -1471,7 +1473,7 @@ def cmd_test(args):
                 Path(test_dir)
                 / f"out_{str(cmd['name']).replace('-', '_')}_{Path(test_file).stem}{Path(test_file).suffix}"
             )
-            
+
             ret = ret and run_and_probe(
                 cmd["name"], cmd_remove_audio_track, args_test, test_dict
             )