Implemented words/regex ban list for subtitles

11 files changed, 141 insertions, 19 deletions

diff --git a/bazarr/api/episodes/episodes_subtitles.py b/bazarr/api/episodes/episodes_subtitles.py
index 86f8b0024..70f59857c 100644
--- a/bazarr/api/episodes/episodes_subtitles.py
+++ b/bazarr/api/episodes/episodes_subtitles.py
@@ -6,7 +6,7 @@ from flask import request
 from flask_restful import Resource
 from subliminal_patch.core import SUBTITLE_EXTENSIONS
 
-from database import TableEpisodes, get_audio_profile_languages
+from database import TableEpisodes, get_audio_profile_languages, get_profile_id
 from ..utils import authenticate
 from helper import path_mappings
 from get_providers import get_providers, get_providers_auth
@@ -55,7 +55,8 @@ class EpisodesSubtitles(Resource):
 
         try:
             result = download_subtitle(episodePath, language, audio_language, hi, forced, providers_list,
-                                       providers_auth, sceneName, title, 'series')
+                                       providers_auth, sceneName, title, 'series',
+                                       profile_id=get_profile_id(episode_id=sonarrEpisodeId))
             if result is not None:
                 message = result[0]
                 path = result[1]
diff --git a/bazarr/api/movies/movies_subtitles.py b/bazarr/api/movies/movies_subtitles.py
index cc2a07e75..c4f1a1e2a 100644
--- a/bazarr/api/movies/movies_subtitles.py
+++ b/bazarr/api/movies/movies_subtitles.py
@@ -6,7 +6,7 @@ from flask import request
 from flask_restful import Resource
 from subliminal_patch.core import SUBTITLE_EXTENSIONS
 
-from database import TableMovies, get_audio_profile_languages
+from database import TableMovies, get_audio_profile_languages, get_profile_id
 from ..utils import authenticate
 from helper import path_mappings
 from get_providers import get_providers, get_providers_auth
@@ -57,7 +57,8 @@ class MoviesSubtitles(Resource):
 
         try:
             result = download_subtitle(moviePath, language, audio_language, hi, forced, providers_list,
-                                       providers_auth, sceneName, title, 'movie')
+                                       providers_auth, sceneName, title, 'movie',
+                                       profile_id=get_profile_id(movie_id=radarrId))
             if result is not None:
                 message = result[0]
                 path = result[1]
diff --git a/bazarr/api/providers/providers_episodes.py b/bazarr/api/providers/providers_episodes.py
index feedc0134..283cc160d 100644
--- a/bazarr/api/providers/providers_episodes.py
+++ b/bazarr/api/providers/providers_episodes.py
@@ -3,7 +3,7 @@
 from flask import request, jsonify
 from flask_restful import Resource
 
-from database import TableEpisodes, TableShows, get_audio_profile_languages
+from database import TableEpisodes, TableShows, get_audio_profile_languages, get_profile_id
 from helper import path_mappings
 from get_providers import get_providers, get_providers_auth
 from get_subtitle import manual_search, manual_download_subtitle
@@ -76,7 +76,8 @@ class ProviderEpisodes(Resource):
 
         try:
             result = manual_download_subtitle(episodePath, language, audio_language, hi, forced, subtitle,
-                                              selected_provider, providers_auth, sceneName, title, 'series')
+                                              selected_provider, providers_auth, sceneName, title, 'series',
+                                              profile_id=get_profile_id(episode_id=sonarrEpisodeId))
             if result is not None:
                 message = result[0]
                 path = result[1]
diff --git a/bazarr/api/providers/providers_movies.py b/bazarr/api/providers/providers_movies.py
index c189ab2c4..d5c31c1c4 100644
--- a/bazarr/api/providers/providers_movies.py
+++ b/bazarr/api/providers/providers_movies.py
@@ -3,7 +3,7 @@
 from flask import request, jsonify
 from flask_restful import Resource
 
-from database import TableMovies, get_audio_profile_languages
+from database import TableMovies, get_audio_profile_languages, get_profile_id
 from helper import path_mappings
 from get_providers import get_providers, get_providers_auth
 from get_subtitle import manual_search, manual_download_subtitle
@@ -77,7 +77,8 @@ class ProviderMovies(Resource):
 
         try:
             result = manual_download_subtitle(moviePath, language, audio_language, hi, forced, subtitle,
-                                              selected_provider, providers_auth, sceneName, title, 'movie')
+                                              selected_provider, providers_auth, sceneName, title, 'movie',
+                                              profile_id=get_profile_id(movie_id=radarrId))
             if result is not None:
                 message = result[0]
                 path = result[1]
diff --git a/bazarr/api/system/settings.py b/bazarr/api/system/settings.py
index c8d9574fc..50f5d7473 100644
--- a/bazarr/api/system/settings.py
+++ b/bazarr/api/system/settings.py
@@ -56,7 +56,9 @@ class SystemSettings(Resource):
                     TableLanguagesProfiles.update({
                         TableLanguagesProfiles.name: item['name'],
                         TableLanguagesProfiles.cutoff: item['cutoff'] if item['cutoff'] != 'null' else None,
-                        TableLanguagesProfiles.items: json.dumps(item['items'])
+                        TableLanguagesProfiles.items: json.dumps(item['items']),
+                        TableLanguagesProfiles.mustContain: item['mustContain'],
+                        TableLanguagesProfiles.mustNotContain: item['mustNotContain'],
                     })\
                         .where(TableLanguagesProfiles.profileId == item['profileId'])\
                         .execute()
@@ -67,7 +69,9 @@ class SystemSettings(Resource):
                         TableLanguagesProfiles.profileId: item['profileId'],
                         TableLanguagesProfiles.name: item['name'],
                         TableLanguagesProfiles.cutoff: item['cutoff'] if item['cutoff'] != 'null' else None,
-                        TableLanguagesProfiles.items: json.dumps(item['items'])
+                        TableLanguagesProfiles.items: json.dumps(item['items']),
+                        TableLanguagesProfiles.mustContain: item['must_contain'],
+                        TableLanguagesProfiles.mustNotContain: item['must_not_contain'],
                     }).execute()
             for profileId in existing:
                 # Unassign this profileId from series and movies
diff --git a/bazarr/database.py b/bazarr/database.py
index 85f420110..b80c1ee7d 100644
--- a/bazarr/database.py
+++ b/bazarr/database.py
@@ -136,6 +136,8 @@ class TableLanguagesProfiles(BaseModel):
     items = TextField()
     name = TextField()
     profileId = AutoField()
+    mustContain = TextField(null=True)
+    mustNotContain = TextField(null=True)
 
     class Meta:
         table_name = 'table_languages_profiles'
@@ -329,7 +331,9 @@ def migrate_db():
         migrator.add_column('table_history_movie', 'provider', TextField(null=True)),
         migrator.add_column('table_history_movie', 'score', TextField(null=True)),
         migrator.add_column('table_history_movie', 'subs_id', TextField(null=True)),
-        migrator.add_column('table_history_movie', 'subtitles_path', TextField(null=True))
+        migrator.add_column('table_history_movie', 'subtitles_path', TextField(null=True)),
+        migrator.add_column('table_languages_profiles', 'mustContain', TextField(null=True)),
+        migrator.add_column('table_languages_profiles', 'mustNotContain', TextField(null=True)),
     )
 
 
@@ -394,10 +398,16 @@ def update_profile_id_list():
     profile_id_list = TableLanguagesProfiles.select(TableLanguagesProfiles.profileId,
                                                     TableLanguagesProfiles.name,
                                                     TableLanguagesProfiles.cutoff,
-                                                    TableLanguagesProfiles.items).dicts()
+                                                    TableLanguagesProfiles.items,
+                                                    TableLanguagesProfiles.mustContain,
+                                                    TableLanguagesProfiles.mustNotContain).dicts()
     profile_id_list = list(profile_id_list)
     for profile in profile_id_list:
         profile['items'] = json.loads(profile['items'])
+        profile['mustContain'] = ast.literal_eval(profile['mustContain']) if profile['mustContain'] else \
+            profile['mustContain']
+        profile['mustNotContain'] = ast.literal_eval(profile['mustNotContain']) if profile['mustNotContain'] else \
+            profile['mustNotContain']
 
 
 def get_profiles_list(profile_id=None):
@@ -422,7 +432,7 @@ def get_desired_languages(profile_id):
 
     if profile_id and profile_id != 'null':
         for profile in profile_id_list:
-            profileId, name, cutoff, items = profile.values()
+            profileId, name, cutoff, items, mustContain, mustNotContain = profile.values()
             if profileId == int(profile_id):
                 languages = [x['language'] for x in items]
                 break
@@ -438,7 +448,7 @@ def get_profile_id_name(profile_id):
 
     if profile_id and profile_id != 'null':
         for profile in profile_id_list:
-            profileId, name, cutoff, items = profile.values()
+            profileId, name, cutoff, items, mustContain, mustNotContain = profile.values()
             if profileId == int(profile_id):
                 name_from_id = name
                 break
@@ -455,7 +465,7 @@ def get_profile_cutoff(profile_id):
     if profile_id and profile_id != 'null':
         cutoff_language = []
         for profile in profile_id_list:
-            profileId, name, cutoff, items = profile.values()
+            profileId, name, cutoff, items, mustContain, mustNotContain = profile.values()
             if cutoff:
                 if profileId == int(profile_id):
                     for item in items:
@@ -498,6 +508,22 @@ def get_audio_profile_languages(series_id=None, episode_id=None, movie_id=None):
     return audio_languages
 
 
+def get_profile_id(series_id=None, episode_id=None, movie_id=None):
+    if series_id:
+        profileId = TableShows.get(TableShows.sonarrSeriesId == series_id).profileId
+    elif episode_id:
+        profileId = TableShows.select(TableShows.profileId)\
+            .join(TableEpisodes, on=(TableShows.sonarrSeriesId == TableEpisodes.sonarrSeriesId))\
+            .where(TableEpisodes.sonarrEpisodeId == episode_id)\
+            .get().profileId
+    elif movie_id:
+        profileId = TableMovies.get(TableMovies.radarrId == movie_id).profileId
+    else:
+        return None
+
+    return profileId
+
+
 def convert_list_to_clause(arr: list):
     if isinstance(arr, list):
         return f"({','.join(str(x) for x in arr)})"
diff --git a/bazarr/get_subtitle.py b/bazarr/get_subtitle.py
index 8e5d35db9..ff8c0cb9a 100644
--- a/bazarr/get_subtitle.py
+++ b/bazarr/get_subtitle.py
@@ -84,7 +84,7 @@ def get_video(path, title, sceneName, providers=None, media_type="movie"):
 
 
 def download_subtitle(path, language, audio_language, hi, forced, providers, providers_auth, sceneName, title,
-                      media_type, forced_minimum_score=None, is_upgrade=False):
+                      media_type, forced_minimum_score=None, is_upgrade=False, profile_id=None):
     # fixme: supply all missing languages, not only one, to hit providers only once who support multiple languages in
     #  one query
 
@@ -158,6 +158,7 @@ def download_subtitle(path, language, audio_language, hi, forced, providers, pro
                                                            compute_score=compute_score,
                                                            throttle_time=None,  # fixme
                                                            blacklist=get_blacklist(media_type=media_type),
+                                                           ban_list=get_ban_list(profile_id),
                                                            throttle_callback=provider_throttle,
                                                            score_obj=handler,
                                                            pre_download_hook=None,  # fixme
@@ -361,6 +362,7 @@ def manual_search(path, profileId, providers, providers_auth, sceneName, title,
                                                providers=providers,
                                                provider_configs=providers_auth,
                                                blacklist=get_blacklist(media_type=media_type),
+                                               ban_list=get_ban_list(profileId),
                                                throttle_callback=provider_throttle,
                                                language_hook=None)  # fixme
 
@@ -375,6 +377,7 @@ def manual_search(path, profileId, providers, providers_auth, sceneName, title,
                                                                 providers=['subscene'],
                                                                 provider_configs=providers_auth,
                                                                 blacklist=get_blacklist(media_type=media_type),
+                                                                ban_list=get_ban_list(profileId),
                                                                 throttle_callback=provider_throttle,
                                                                 language_hook=None)  # fixme
                         providers_auth['subscene']['only_foreign'] = False
@@ -466,7 +469,7 @@ def manual_search(path, profileId, providers, providers_auth, sceneName, title,
 
 
 def manual_download_subtitle(path, language, audio_language, hi, forced, subtitle, provider, providers_auth, sceneName,
-                             title, media_type):
+                             title, media_type, profile_id):
     logging.debug('BAZARR Manually downloading Subtitles for this file: ' + path)
 
     if settings.general.getboolean('utf8_encode'):
@@ -498,6 +501,7 @@ def manual_download_subtitle(path, language, audio_language, hi, forced, subtitl
                                    provider_configs=providers_auth,
                                    pool_class=provider_pool(),
                                    blacklist=get_blacklist(media_type=media_type),
+                                   ban_list=get_ban_list(profile_id),
                                    throttle_callback=provider_throttle)
                 logging.debug('BAZARR Subtitles file downloaded for this file:' + path)
             else:
@@ -1706,6 +1710,7 @@ def _get_lang_obj(alpha3):
 
     return sub.subzero_language()
 
+
 def _get_scores(media_type, min_movie=None, min_ep=None):
     series = "series" == media_type
     handler = series_score if series else movie_score
@@ -1713,3 +1718,12 @@ def _get_scores(media_type, min_movie=None, min_ep=None):
     min_ep = min_ep or (240 * 100 / handler.max_score)
     min_score_ = int(min_ep if series else min_movie)
     return handler.get_scores(min_score_)
+
+
+def get_ban_list(profile_id):
+    if profile_id:
+        profile = get_profiles_list(profile_id)
+        if profile:
+            return {'must_contain': profile['mustContain'] or [],
+                    'must_not_contain': profile['mustNotContain'] or []}
+    return None
diff --git a/frontend/src/@types/api.d.ts b/frontend/src/@types/api.d.ts
index 043753879..2d820460a 100644
--- a/frontend/src/@types/api.d.ts
+++ b/frontend/src/@types/api.d.ts
@@ -33,6 +33,8 @@ declare namespace Language {
     profileId: number;
     cutoff: number | null;
     items: ProfileItem[];
+    mustContain: string[];
+    mustNotContain: string[];
   }
 }
 
diff --git a/frontend/src/Settings/Languages/modal.tsx b/frontend/src/Settings/Languages/modal.tsx
index 20d1943a9..7371267dc 100644
--- a/frontend/src/Settings/Languages/modal.tsx
+++ b/frontend/src/Settings/Languages/modal.tsx
@@ -13,6 +13,7 @@ import {
   ActionButton,
   BaseModal,
   BaseModalProps,
+  Chips,
   LanguageSelector,
   Selector,
   SimpleTable,
@@ -31,6 +32,8 @@ function createDefaultProfile(): Language.Profile {
     name: "",
     items: [],
     cutoff: null,
+    mustContain: [],
+    mustNotContain: [],
   };
 }
 
@@ -260,6 +263,28 @@ const LanguagesProfileModal: FunctionComponent<Props & BaseModalProps> = (
         ></Selector>
         <Message>Ignore others if existing</Message>
       </Input>
+      <Input name="Release info must contain">
+        <Chips
+          value={current.mustContain}
+          onChange={(mc) => updateProfile("mustContain", mc)}
+        ></Chips>
+        <Message>
+          Subtitles release info must include one of those words or they will be
+          excluded from search results (regex supported).
+        </Message>
+      </Input>
+      <Input name="Release info must not contain">
+        <Chips
+          value={current.mustNotContain}
+          onChange={(mnc: string[]) => {
+            updateProfile("mustNotContain", mnc);
+          }}
+        ></Chips>
+        <Message>
+          Subtitles release info including one of those words (case insensitive)
+          will be excluded from search results (regex supported).
+        </Message>
+      </Input>
     </BaseModal>
   );
 };
diff --git a/frontend/src/Settings/Languages/table.tsx b/frontend/src/Settings/Languages/table.tsx
index 4547e3198..09c75d138 100644
--- a/frontend/src/Settings/Languages/table.tsx
+++ b/frontend/src/Settings/Languages/table.tsx
@@ -95,6 +95,40 @@ const Table: FunctionComponent = () => {
         },
       },
       {
+        Header: "Must contain",
+        accessor: "mustContain",
+        Cell: (row) => {
+          const items = row.value;
+          if (!items) {
+            return false;
+          }
+          return items.map((v) => {
+            return (
+              <Badge className={"mx-1"} variant={"secondary"}>
+                {v}
+              </Badge>
+            );
+          });
+        },
+      },
+      {
+        Header: "Must not contain",
+        accessor: "mustNotContain",
+        Cell: (row) => {
+          const items = row.value;
+          if (!items) {
+            return false;
+          }
+          return items.map((v) => {
+            return (
+              <Badge className={"mx-1"} variant={"secondary"}>
+                {v}
+              </Badge>
+            );
+          });
+        },
+      },
+      {
         accessor: "profileId",
         Cell: ({ row, update }) => {
           const profile = row.original;
diff --git a/libs/subliminal_patch/core.py b/libs/subliminal_patch/core.py
index b6629526a..aa82db1f9 100644
--- a/libs/subliminal_patch/core.py
+++ b/libs/subliminal_patch/core.py
@@ -66,7 +66,7 @@ def remove_crap_from_fn(fn):
 
 
 class SZProviderPool(ProviderPool):
-    def __init__(self, providers=None, provider_configs=None, blacklist=None, throttle_callback=None,
+    def __init__(self, providers=None, provider_configs=None, blacklist=None, ban_list=None, throttle_callback=None,
                  pre_download_hook=None, post_download_hook=None, language_hook=None):
         #: Name of providers to use
         self.providers = providers
@@ -82,6 +82,9 @@ class SZProviderPool(ProviderPool):
 
         self.blacklist = blacklist or []
 
+        #: Should be a dict of 2 lists of strings
+        self.ban_list = ban_list or {'must_contain': [], 'must_not_contain': []}
+
         self.throttle_callback = throttle_callback
 
         self.pre_download_hook = pre_download_hook
@@ -184,6 +187,16 @@ class SZProviderPool(ProviderPool):
                 if (str(provider), str(s.id)) in self.blacklist:
                     logger.info("Skipping blacklisted subtitle: %s", s)
                     continue
+                if hasattr(s, 'release_info'):
+                    if s.release_info is not None:
+                        if any([x for x in self.ban_list["must_not_contain"]
+                                if re.search(x, s.release_info, flags=re.IGNORECASE) is not None]):
+                            logger.info("Skipping subtitle because release name contains prohibited string: %s", s)
+                            continue
+                        if any([x for x in self.ban_list["must_contain"]
+                                if re.search(x, s.release_info, flags=re.IGNORECASE) is None]):
+                            logger.info("Skipping subtitle because release name does not contains required string: %s", s)
+                            continue
                 if s.id in seen:
                     continue
                 s.plex_media_fps = float(video.fps) if video.fps else None
@@ -506,7 +519,7 @@ class SZAsyncProviderPool(SZProviderPool):
 
         return provider, provider_subtitles
 
-    def list_subtitles(self, video, languages, blacklist=None):
+    def list_subtitles(self, video, languages, blacklist=None, ban_list=None):
         if is_windows_special_path:
             return super(SZAsyncProviderPool, self).list_subtitles(video, languages)