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>

tests/test_extract_text.py

@@ -46,6 +46,19 @@ def test_default_language_is_original_audio(fake_yt_dlp, tmp_path, monkeypatch,
     assert [p.name for p in tmp_path.iterdir()] == ["output.txt"]
 
 
+def test_default_language_ignores_dubbed_tracks(fake_yt_dlp, tmp_path, monkeypatch):
+    info = {
+        "id": "dub", "language": "en-US", "subtitles": {"en-US": []},
+        "automatic_captions": {"ar-orig": [], "en-orig": []},
+    }
+    run = fake_yt_dlp(info, vtt=VTT)
+    monkeypatch.setattr(yt.subprocess, "run", run)
+    run_main(monkeypatch, "-u", URL, "-o", tmp_path / "out.txt")
+    download = run.calls[1]
+    assert download[download.index("--sub-langs") + 1] == "en-US"
+    assert "--write-subs" in download
+
+
 def test_fliesstext(fake_yt_dlp, tmp_path, monkeypatch, capsys):
     long_vtt = "WEBVTT\n\n" + "".join(
         f"00:00:{i:02d}.000 --> 00:00:{i + 1:02d}.000\nSatz Nummer {i} zum EU-Gipfel.\n\n"

tests/test_yt_untertitel.py

@@ -47,6 +47,49 @@ def test_original_language(info, expected):
     assert yt.original_language(info) == expected
 
 
+# English original with AI-dubbed audio tracks: every dub has its own "-orig"
+DUBBED = {
+    "id": "dub",
+    "language": "en-US",
+    "subtitles": {"en-US": []},
+    "automatic_captions": {"ar-orig": [], "en-orig": [], "de-DE-orig": [], "de": []},
+}
+DUBBED_NO_METADATA = {
+    "formats": [
+        {"format_note": "Arabic, low", "language": "ar"},
+        {"format_note": "English (US) original (default), low", "language": "en-US"},
+    ],
+    "subtitles": {},
+    "automatic_captions": {"ar-orig": [], "en-orig": [], "de-DE-orig": []},
+}
+
+
+@pytest.mark.parametrize("info, lang, expected", [
+    (DUBBED, "en", ("en-US", False)),          # region variant of a manual track
+    (DUBBED, "en-US", ("en-US", False)),
+    (DUBBED, "de", None),                      # only the German AI dub
+    (DUBBED, "ar", None),
+    (DUBBED_NO_METADATA, "en-US", ("en-orig", True)),
+    (DUBBED_NO_METADATA, "en", ("en-orig", True)),
+    (DUBBED_NO_METADATA, "de", None),
+])
+def test_choose_track_with_dubbed_audio(info, lang, expected):
+    assert yt.choose_track(info, lang) == expected
+
+
+@pytest.mark.parametrize("info, expected", [
+    (DUBBED, "en-US"),
+    (DUBBED_NO_METADATA, "en-US"),             # from the audio track marked original
+    ({"automatic_captions": {"ar-orig": [], "en-orig": []}}, None),
+])
+def test_original_language_with_dubbed_audio(info, expected):
+    assert yt.original_language(info) == expected
+
+
+def test_available_tracks_hides_dubs():
+    assert yt.available_tracks(DUBBED) == (["en-US"], ["en"])
+
+
 def test_available_tracks():
     info = {"subtitles": {"fr": []}, "automatic_captions": {"de-orig": [], "de": []}}
     assert yt.available_tracks(info) == (["fr"], ["de"])