Embedded Subtitles provider: add support for unknown language tags

bazarr/app/config.py

@@ -206,6 +206,7 @@ defaults = {
         'included_codecs': '[]',
         'hi_fallback': 'False',
         'timeout': '600',
+        'unknown_as_english': 'False',
     },
     'karagarga': {
         'username': '',

bazarr/app/get_providers.py

@@ -232,6 +232,7 @@ def get_providers_auth():
             'ffprobe_path': _FFPROBE_BINARY,
             'ffmpeg_path': _FFMPEG_BINARY,
             'timeout': settings.embeddedsubtitles.timeout,
+            'unknown_as_english': settings.embeddedsubtitles.getboolean('unknown_as_english'),
         },
         'karagarga': {
             'username': settings.karagarga.username,

frontend/src/pages/Settings/Providers/list.ts

@@ -103,6 +103,11 @@ export const ProviderList: Readonly<ProviderInfo[]> = [
         key: "hi_fallback",
         name: "Use HI subtitles as a fallback (don't enable it if you have a HI language profile)",
       },
+      {
+        type: "switch",
+        key: "unknown_as_english",
+        name: "Use subtitles with unknown info/language as english",
+      },
     ],
     message:
       "Warning for cloud users: this provider needs to read the entire file in order to extract subtitles.",

libs/fese/__init__.py

@@ -4,4 +4,4 @@
 from .container import FFprobeVideoContainer
 from .stream import FFprobeSubtitleStream
 
-__version__ = "0.2"
+__version__ = "0.2.2"

libs/fese/tags.py

@@ -8,12 +8,21 @@ from .exceptions import LanguageNotFound
 
 logger = logging.getLogger(__name__)
 
+LANGUAGE_FALLBACK = None
+
 
 class FFprobeGenericSubtitleTags:
     _DETECTABLE_TAGS = None
 
     def __init__(self, data: dict):
-        self.language = _get_language(data)
+        try:
+            self.language = _get_language(data)
+        except LanguageNotFound:
+            if LANGUAGE_FALLBACK is not None:
+                self.language = Language.fromietf(LANGUAGE_FALLBACK)
+            else:
+                raise
+
         self._data = data
 
     @classmethod

libs/subliminal_patch/providers/embeddedsubtitles.py

@@ -79,9 +79,7 @@ class EmbeddedSubtitlesProvider(Provider):
         ffmpeg_path=None,
         hi_fallback=False,
         timeout=600,
-        include_ass=None,
-        include_srt=None,
-        mergerfs_mode=None,
+        unknown_as_english=False,
     ):
         self._included_codecs = set(included_codecs or _ALLOWED_CODECS)
 
@@ -93,6 +91,7 @@ class EmbeddedSubtitlesProvider(Provider):
             cache_dir or tempfile.gettempdir(), self.__class__.__name__.lower()
         )
         self._hi_fallback = hi_fallback
+        self._unknown_as_english = unknown_as_english
         self._cached_paths = {}
         self._timeout = int(timeout)
 
@@ -105,6 +104,9 @@ class EmbeddedSubtitlesProvider(Provider):
             # Default is True
             container.FFMPEG_STATS = False
 
+        tags.LANGUAGE_FALLBACK = "en" if self._unknown_as_english else None
+        logger.debug("Language fallback set: %s", tags.LANGUAGE_FALLBACK)
+
     def initialize(self):
         os.makedirs(self._cache_dir, exist_ok=True)
 

tests/bazarr/app/test_get_providers.py

@@ -34,6 +34,7 @@ def test_get_providers_auth_embeddedsubtitles():
     assert isinstance(item["ffprobe_path"], str)
     assert isinstance(item["ffmpeg_path"], str)
     assert isinstance(item["timeout"], str)
+    assert isinstance(item["unknown_as_english"], bool)
 
 
 def test_get_providers_auth_karagarga():

tests/subliminal_patch/test_embeddedsubtitles.py

@@ -4,6 +4,7 @@ import os
 from fese import FFprobeSubtitleStream
 from fese import FFprobeVideoContainer
 from fese import tags
+from fese.exceptions import LanguageNotFound
 import pytest
 from subliminal_patch.core import Episode
 from subliminal_patch.core import Movie
@@ -123,13 +124,36 @@ def fake_streams():
     }
 
 
+@pytest.mark.parametrize("tags_", [{}, {"language": "und", "title": "Unknown"}])
+def test_list_subtitles_unknown_as_english(mocker, tags_):
+    with EmbeddedSubtitlesProvider(unknown_as_english=True):
+        fake = FFprobeSubtitleStream(
+            {"index": 3, "codec_name": "subrip", "tags": tags_}
+        )
+        mocker.patch(
+            "subliminal_patch.providers.embeddedsubtitles._MemoizedFFprobeVideoContainer.get_subtitles",
+            return_value=[fake],
+        )
+        streams = _MemoizedFFprobeVideoContainer.get_subtitles("")
+        assert len(streams) == 1
+        assert streams[0].language == Language.fromietf("en")
+
+
+@pytest.mark.parametrize("tags_", [{}, {"language": "und", "title": "Unknown"}])
+def test_list_subtitles_unknown_as_english_disabled(tags_):
+    with EmbeddedSubtitlesProvider(unknown_as_english=False):
+        with pytest.raises(LanguageNotFound):
+            assert FFprobeSubtitleStream(
+                {"index": 3, "codec_name": "subrip", "tags": tags_}
+            )
+
+
 def test_list_subtitles_hi_fallback_one_stream(
     video_single_language, fake_streams, mocker
 ):
     with EmbeddedSubtitlesProvider(hi_fallback=True) as provider:
         language = Language.fromalpha2("en")
         mocker.patch(
-            #            "fese.FFprobeVideoContainer.get_subtitles",
             "subliminal_patch.providers.embeddedsubtitles._MemoizedFFprobeVideoContainer.get_subtitles",
             return_value=[fake_streams["en_hi"]],
         )