Preference or avoidance per language type (hearing impaired, foreign only)

changelog.d/1175.change.rst

@@ -0,0 +1 @@
+Add cli option to prefer or disfavor hearing impaired (-hi/-HI) or foreign only (-fo/-FO) subtitles.

docs/config.toml

@@ -22,6 +22,7 @@ provider = ["addic7ed", "opensubtitlescom", "opensubtitles"]
 refiner = ["metadata", "hash", "omdb"]
 ignore_refiner = ["tmdb"]
 language = ["fr", "en", "pt-br"]
+foreign_only = false
 encoding = "utf-8"
 min_score = 50
 archives = true

pyproject.toml

@@ -321,6 +321,8 @@ extend-ignore-re = [
     "(?Rm)^.*#\\s*spellchecker:\\s*disable-line$",
     "#\\s*spellchecker:off\\s*\\n.*\\n\\s*#\\s*spellchecker:on"
 ]
+[tool.typos.default.extend-words]
+fo = "fo"
 [tool.typos.default.extend-identifiers]
 tha = "tha"
 bre = "bre"

subliminal/cli.py

@@ -40,7 +40,6 @@
 )
 from subliminal.core import ARCHIVE_EXTENSIONS, scan_name, search_external_subtitles
 from subliminal.extensions import get_default_providers, get_default_refiners
-from subliminal.score import match_hearing_impaired
 from subliminal.utils import merge_extend_and_ignore_unions
 
 if TYPE_CHECKING:
@@ -142,6 +141,12 @@ def configure(ctx: click.Context, param: click.Parameter | None, filename: str |
 
     # make download options
     download_dict = toml_dict.setdefault('download', {})
+    # handle language types
+    for lt in ('hearing_impaired', 'foreign_only'):
+        # if an option was defined in the config file, make it a tuple, the expected type
+        if lt in download_dict and (isinstance(download_dict[lt], bool) or download_dict[lt] is None):
+            download_dict[lt] = (download_dict[lt],)
+
     # remove the provider and refiner lists to select, extend and ignore
     provider_lists = {
         'select': download_dict.pop('provider', []),
@@ -411,7 +416,42 @@ def cache(ctx: click.Context, clear_subliminal: bool) -> None:
     ),
 )
 @click.option('-f', '--force', is_flag=True, default=False, help='Force download even if a subtitle already exist.')
-@click.option('-hi', '--hearing-impaired', is_flag=True, default=False, help='Prefer hearing impaired subtitles.')
+@click.option(
+    '-fo',
+    '--foreign-only',
+    'foreign_only',
+    is_flag=True,
+    flag_value=True,
+    multiple=True,
+    help='Prefer foreign-only subtitles.',
+)
+@click.option(
+    '-FO',
+    '--no-foreign-only',
+    'foreign_only',
+    is_flag=True,
+    flag_value=False,
+    multiple=True,
+    help='Disfavor foreign-only subtitles.',
+)
+@click.option(
+    '-hi',
+    '--hearing-impaired',
+    'hearing_impaired',
+    is_flag=True,
+    flag_value=True,
+    multiple=True,
+    help='Prefer hearing-impaired subtitles.',
+)
+@click.option(
+    '-HI',
+    '--no-hearing-impaired',
+    'hearing_impaired',
+    is_flag=True,
+    flag_value=False,
+    multiple=True,
+    help='Disfavor hearing-impaired subtitles.',
+)
 @click.option(
     '-m',
     '--min-score',
@@ -423,7 +463,7 @@ def cache(ctx: click.Context, clear_subliminal: bool) -> None:
     '--language-type-suffix',
     is_flag=True,
     default=False,
-    help='Add a suffix to the saved subtitle name to indicate a hearing impaired or foreign part subtitle.',
+    help='Add a suffix to the saved subtitle name to indicate a hearing impaired or foreign only subtitle.',
 )
 @click.option(
     '--language-format',
@@ -468,7 +508,8 @@ def download(
     original_encoding: bool,
     single: bool,
     force: bool,
-    hearing_impaired: bool,
+    hearing_impaired: tuple[bool | None, ...],
+    foreign_only: tuple[bool | None, ...],
     min_score: int,
     language_type_suffix: bool,
     language_format: str,
@@ -496,6 +537,14 @@ def download(
     elif encoding is None:
         encoding = 'utf-8'
 
+    # language_type
+    hearing_impaired_flag: bool | None = None
+    if len(hearing_impaired) > 0:
+        hearing_impaired_flag = hearing_impaired[-1]
+    foreign_only_flag: bool | None = None
+    if len(foreign_only) > 0:
+        foreign_only_flag = foreign_only[-1]
+
     debug = obj.get('debug', False)
     if debug:
         verbose = 3
@@ -649,7 +698,8 @@ def download(
                     v,
                     language_set,
                     min_score=scores['hash'] * min_score // 100,
-                    hearing_impaired=hearing_impaired,
+                    hearing_impaired=hearing_impaired_flag,
+                    foreign_only=foreign_only_flag,
                     only_one=single,
                     ignore_subtitles=ignore_subtitles,
                 )
@@ -701,11 +751,8 @@ def download(
                     else:
                         score_color = 'green'
 
-                # scale score from 0 to 100 taking out preferences
-                scaled_score = score
-                if match_hearing_impaired(s, hearing_impaired=hearing_impaired):
-                    scaled_score -= scores['hearing_impaired']
-                scaled_score *= 100 / scores['hash']
+                # scale score from 0 to 100
+                scaled_score = score * 100 / scores['hash']
 
                 # echo some nice colored output
                 language_str = (

subliminal/core.py

@@ -24,7 +24,7 @@
     refiner_manager,
 )
 from .score import compute_score as default_compute_score
-from .subtitle import SUBTITLE_EXTENSIONS, Subtitle
+from .subtitle import SUBTITLE_EXTENSIONS, LanguageType, Subtitle
 from .utils import get_age, handle_exception
 from .video import VIDEO_EXTENSIONS, Episode, Movie, Video
 
@@ -148,7 +148,7 @@ def list_subtitles_provider(self, provider: str, video: Video, languages: Set[La
         try:
             return self[provider].list_subtitles(video, provider_languages)
         except Exception as e:  # noqa: BLE001
-            handle_exception(e, 'Provider {provider}')
+            handle_exception(e, f'Provider {provider}')
 
         return []
 
@@ -220,7 +220,8 @@ def download_best_subtitles(
         languages: Set[Language],
         *,
         min_score: int = 0,
-        hearing_impaired: bool = False,
+        hearing_impaired: bool | None = None,
+        foreign_only: bool | None = None,
         only_one: bool = False,
         compute_score: ComputeScore | None = None,
         ignore_subtitles: Sequence[str] | None = None,
@@ -234,10 +235,11 @@ def download_best_subtitles(
         :param languages: languages to download.
         :type languages: set of :class:`~babelfish.language.Language`
         :param int min_score: minimum score for a subtitle to be downloaded.
-        :param bool hearing_impaired: hearing impaired preference.
+        :param (bool | None) hearing_impaired: hearing impaired preference (yes/no/indifferent).
+        :param (bool | None) foreign_only: foreign only preference (yes/no/indifferent).
         :param bool only_one: download only one subtitle, not one per language.
         :param compute_score: function that takes `subtitle` and `video` as positional arguments,
-            `hearing_impaired` as keyword argument and returns the score.
+            and returns the score.
         :param ignore_subtitles: list of subtitle ids to ignore (None defaults to an empty list).
         :return: downloaded subtitles.
         :rtype: list of :class:`~subliminal.subtitle.Subtitle`
@@ -249,9 +251,19 @@ def download_best_subtitles(
         # ignore subtitles
         subtitles = [s for s in subtitles if s.id not in ignore_subtitles]
 
+        # sort by hearing impaired and foreign only
+        language_type = LanguageType.from_flags(hearing_impaired=hearing_impaired, foreign_only=foreign_only)
+        if language_type != LanguageType.UNKNOWN:
+            logger.info('Sort subtitles by %s types first', language_type.value)
+            subtitles = sorted(
+                subtitles,
+                key=lambda s: s.language_type == language_type,
+                reverse=True,
+            )
+
         # sort subtitles by score
         scored_subtitles = sorted(
-            [(s, compute_score(s, video, hearing_impaired=hearing_impaired)) for s in subtitles],
+            [(s, compute_score(s, video)) for s in subtitles],
             key=operator.itemgetter(1),
             reverse=True,
         )
@@ -411,7 +423,7 @@ def parse_subtitle_filename(subtitle_filename: str, video_filename: str) -> Subt
         except (ValueError, LanguageReverseError):
             logger.exception('Cannot parse language code %r', language_code)
 
-    # TODO: extract the hearing_impaired or forced attribute
+    # TODO: extract the hearing_impaired or foreign_only attribute
 
     return Subtitle(language, subtitle_id=subtitle_filename)
 
@@ -775,7 +787,8 @@ def download_best_subtitles(
     languages: Set[Language],
     *,
     min_score: int = 0,
-    hearing_impaired: bool = False,
+    hearing_impaired: bool | None = None,
+    foreign_only: bool | None = None,
     only_one: bool = False,
     compute_score: ComputeScore | None = None,
     pool_class: type[ProviderPool] = ProviderPool,
@@ -790,7 +803,8 @@ def download_best_subtitles(
     :param languages: languages to download.
     :type languages: set of :class:`~babelfish.language.Language`
     :param int min_score: minimum score for a subtitle to be downloaded.
-    :param bool hearing_impaired: hearing impaired preference.
+    :param (bool | None) hearing_impaired: hearing impaired preference (yes/no/indifferent).
+    :param (bool | None) foreign_only: foreign only preference (yes/no/indifferent).
     :param bool only_one: download only one subtitle, not one per language.
     :param compute_score: function that takes `subtitle` and `video` as positional arguments,
         `hearing_impaired` as keyword argument and returns the score.
@@ -825,6 +839,7 @@ def download_best_subtitles(
                 languages,
                 min_score=min_score,
                 hearing_impaired=hearing_impaired,
+                foreign_only=foreign_only,
                 only_one=only_one,
                 compute_score=compute_score,
             )
@@ -861,7 +876,7 @@ def save_subtitles(
     :param str directory: path to directory where to save the subtitles, default is next to the video.
     :param str encoding: encoding in which to save the subtitles, default is to keep original encoding.
     :param (str | None) extension: the subtitle extension, default is to match to the subtitle format.
-    :param bool language_type_suffix: add a suffix 'hi' or 'forced' if needed. Default to False.
+    :param bool language_type_suffix: add a suffix 'hi' or 'fo' if needed. Default to False.
     :param str language_format: format of the language suffix. Default to 'alpha2'.
     :return: the saved subtitles
     :rtype: list of :class:`~subliminal.subtitle.Subtitle`

subliminal/providers/opensubtitlescom.py

@@ -179,6 +179,7 @@ def __init__(
         subtitle_id: str,
         *,
         hearing_impaired: bool = False,
+        foreign_only: bool = False,
         movie_kind: str | None = None,
         release: str | None = None,
         movie_title: str | None = None,
@@ -199,7 +200,14 @@ def __init__(
         file_id: int = 0,
         file_name: str = '',
     ) -> None:
-        super().__init__(language, subtitle_id, hearing_impaired=hearing_impaired, page_link=None, encoding='utf-8')
+        super().__init__(
+            language,
+            subtitle_id,
+            hearing_impaired=hearing_impaired,
+            foreign_only=foreign_only,
+            page_link=None,
+            encoding='utf-8',
+        )
         self.movie_kind = movie_kind
         self.release = release
         self.movie_title = movie_title
@@ -235,6 +243,7 @@ def from_response(
         attributes = response.get('attributes', {})
         language = Language.fromopensubtitlescom(str(attributes.get('language')))
         hearing_impaired = bool(int(attributes.get('hearing_impaired')))
+        foreign_only = bool(int(attributes.get('foreign_parts_only')))
         release = str(attributes.get('release'))
         moviehash_match = bool(attributes.get('moviehash_match', False))
         download_count = int(attributes.get('download_count'))
@@ -266,6 +275,7 @@ def from_response(
             language,
             subtitle_id,
             hearing_impaired=hearing_impaired,
+            foreign_only=foreign_only,
             movie_kind=movie_kind,
             release=release,
             movie_title=movie_title,

subliminal/score.py

@@ -49,7 +49,7 @@
     class ComputeScore(Protocol):
         """Compute the score of a subtitle matching a video."""
 
-        def __call__(self, subtitle: Subtitle, video: Video, *, hearing_impaired: bool | None) -> int: ...  # noqa: D102
+        def __call__(self, subtitle: Subtitle, video: Video) -> int: ...  # noqa: D102
 
 
 # Check if sympy is installed (for tests)
@@ -141,8 +141,8 @@ def match_hearing_impaired(subtitle: Subtitle, *, hearing_impaired: bool | None
     )
 
 
-def compute_score(subtitle: Subtitle, video: Video, *, hearing_impaired: bool | None = None) -> int:
-    """Compute the score of the `subtitle` against the `video` with `hearing_impaired` preference.
+def compute_score(subtitle: Subtitle, video: Video, **kwargs: Any) -> int:
+    """Compute the score of the `subtitle` against the `video`.
 
     :func:`compute_score` uses the :meth:`Subtitle.get_matches <subliminal.subtitle.Subtitle.get_matches>` method and
     applies the scores (either from :data:`episode_scores` or :data:`movie_scores`) after some processing.
@@ -151,12 +151,11 @@ def compute_score(subtitle: Subtitle, video: Video, *, hearing_impaired: bool |
     :type subtitle: :class:`~subliminal.subtitle.Subtitle`
     :param video: the video to compute the score against.
     :type video: :class:`~subliminal.video.Video`
-    :param (bool | None) hearing_impaired: hearing impaired preference (None if no preference).
     :return: score of the subtitle.
     :rtype: int
 
     """
-    logger.info('Computing score of %r for video %r with %r', subtitle, video, {'hearing_impaired': hearing_impaired})
+    logger.info('Computing score of %r for video %r', subtitle, video)
 
     # get the scores dict
     scores = get_scores(video)
@@ -193,17 +192,12 @@ def compute_score(subtitle: Subtitle, video: Video, *, hearing_impaired: bool |
             logger.debug('Adding imdb_id match equivalents')
             matches |= {'title', 'year', 'country'}
 
-    # handle hearing impaired
-    if match_hearing_impaired(subtitle, hearing_impaired=hearing_impaired):
-        logger.debug('Matched hearing_impaired')
-        matches.add('hearing_impaired')
-
     # compute the score
     score = int(sum(scores.get(match, 0) for match in matches))
     logger.info('Computed score %r with final matches %r', score, matches)
 
     # ensure score is within valid bounds
-    max_score = scores['hash'] + scores['hearing_impaired']
+    max_score = scores['hash']
     if not (0 <= score <= max_score):  # pragma: no cover
         logger.info('Clip score between 0 and %d: %d', max_score, score)
         score = int(clip(score, 0, max_score))