fix: Originalsprache bei Videos mit KI-synchronisierten Tonspuren

Gefunden mit einem englischen Video (0LmCveTM788): ohne -l wurde Arabisch
gewählt, denn jede der 20 Synchron-Tonspuren hat eine eigene „-orig“-Spur.
- original_language(): aus „language“, sonst Tonspur mit „original“, sonst
  die einzige „-orig“-Spur
- Automatische Untertitel nur in der Originalsprache (keine Synchronisation)
- Sprachen auch mit Region: „en“ findet manuelle Untertitel „en-US“

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

yt_untertitel.py

@@ -31,39 +31,66 @@ def fetch_info(url):
     return json.loads(result.stdout)
 
 
-def choose_track(info, lang):
-    """Pick the subtitle track for `lang`: (track name, is_auto) or None.
-
-    Manual subtitles come first. Of the automatic captions only the original
-    language ("<lang>-orig") is used: plain "<lang>" there is a machine
-    translation whenever the video is in another language.
-    """
-    if lang in (info.get("subtitles") or {}):
-        return lang, False
-    if f"{lang}-orig" in (info.get("automatic_captions") or {}):
-        return f"{lang}-orig", True
-    return None
+def base_language(code):
+    """"en-US" -> "en", "de-orig" -> "de"."""
+    return code.removesuffix("-orig").split("-")[0].lower()
 
 
 def original_language(info):
-    """Language of the video's audio, or None if yt-dlp does not know it.
+    """Language of the video's original audio, or None if unknown.
 
-    YouTube marks the automatic captions in the spoken language with "-orig".
+    Not simply the first "<lang>-orig" caption track: videos with AI-dubbed
+    audio tracks have such a track for every dubbed language.
     """
-    for name in info.get("automatic_captions") or {}:
-        if name.endswith("-orig"):
-            return name.removesuffix("-orig")
-    return info.get("language")
+    if info.get("language"):
+        return info["language"]
+    for fmt in info.get("formats") or []:
+        if "original" in (fmt.get("format_note") or "") and fmt.get("language"):
+            return fmt["language"]
+    originals = [n for n in info.get("automatic_captions") or {} if n.endswith("-orig")]
+    return originals[0].removesuffix("-orig") if len(originals) == 1 else None
+
+
+def choose_track(info, lang):
+    """Pick the subtitle track for `lang`: (track name, is_auto) or None.
+
+    Manual subtitles come first, also with a region ("en" finds "en-US").
+    Automatic captions are used only in the language of the original audio
+    ("<lang>-orig"): plain "<lang>" is a machine translation, and "-orig"
+    tracks of other languages belong to AI-dubbed audio tracks.
+    """
+    manual = info.get("subtitles") or {}
+    if lang in manual:
+        return lang, False
+    same_language = sorted(n for n in manual if base_language(n) == base_language(lang))
+    if same_language:
+        return same_language[0], False
+
+    original = original_language(info)
+    if original and base_language(original) != base_language(lang):
+        return None
+    auto = info.get("automatic_captions") or {}
+    candidates = [f"{lang}-orig", f"{base_language(lang)}-orig"] + sorted(
+        n for n in auto
+        if n.endswith("-orig") and base_language(n) == base_language(lang)
+    )
+    for name in candidates:
+        if name in auto:
+            return name, True
+    return None
 
 
 def available_tracks(info):
+    """Manual tracks and the automatic ones usable for this program."""
     manual = sorted((info.get("subtitles") or {}).keys())
-    original = sorted(
+    original = original_language(info)
+    automatic = sorted(
         name.removesuffix("-orig")
         for name in (info.get("automatic_captions") or {})
         if name.endswith("-orig")
+        and (original is None or base_language(name) == base_language(original))
     )
-    return manual, original
+    return manual, automatic
 
 
 def no_track_message(info, lang):