Merge pull request #1146 from vitiko98/tusubtitulo-english

English support for TuSubtitulo and minor fixes with Sucha and Subdivx

libs/subliminal_patch/providers/subdivx.py

@@ -86,6 +86,8 @@ class SubdivxSubtitle(Subtitle):
             elif video_codecs[0] == "h.265":
                 video_codecs.append("h265")
                 video_codecs.append("x265")
+            elif video_codecs[0] == "divx":
+                video_codecs.append("divx")
             for vc in video_codecs:
                 if vc in self.description:
                     matches.add('video_codec')
@@ -118,9 +120,15 @@ class SubdivxSubtitlesProvider(Provider):
         if isinstance(video, Episode):
             query = "{} S{:02d}E{:02d}".format(video.series, video.season, video.episode)
         else:
+            # Subdvix has problems searching foreign movies if the year is
+            # appended. For example: if we search "Memories of Murder 2003",
+            # Subdix won't return any results; but if we search "Memories of
+            # Murder", it will. That's because in Subdvix foreign titles have
+            # the year after the original title ("Salinui chueok (2003) aka
+            # Memories of Murder").
+            # A proper solution would be filtering results with the year in
+            # _parse_subtitles_page.
             query = video.title
-            if video.year:
-                query += ' {:4d}'.format(video.year)
 
         params = {
             'q': query,  # search string
@@ -145,7 +153,7 @@ class SubdivxSubtitlesProvider(Provider):
 
             subtitles += page_subtitles
 
-            if len(page_subtitles) < 20:
+            if len(page_subtitles) < 100:
                 break  # this is the last page
 
             params['pg'] += 1  # search next page
@@ -188,6 +196,11 @@ class SubdivxSubtitlesProvider(Provider):
 
             # title
             title = title_soup.find("a").text.replace("Subtitulos de ", "")
+
+            # filter by year
+            if video.year and str(video.year) not in title:
+                continue
+
             page_link = title_soup.find("a")["href"]
 
             # description

libs/subliminal_patch/providers/sucha.py

@@ -1,19 +1,18 @@
 # -*- coding: utf-8 -*-
+import io
 import logging
 import os
-import io
-import rarfile
 import zipfile
 
+import rarfile
 from requests import Session
-from subzero.language import Language
+from subliminal import Episode, Movie
-
-from subliminal import Movie, Episode
-from subliminal_patch.exceptions import APIThrottled
 from subliminal.exceptions import ServiceUnavailable
-from subliminal_patch.subtitle import Subtitle
+from subliminal.subtitle import SUBTITLE_EXTENSIONS, fix_line_ending
-from subliminal.subtitle import fix_line_ending, SUBTITLE_EXTENSIONS
+from subliminal_patch.exceptions import APIThrottled
 from subliminal_patch.providers import Provider
+from subliminal_patch.subtitle import Subtitle
+from subzero.language import Language
 
 logger = logging.getLogger(__name__)
 
@@ -26,13 +25,21 @@ class SuchaSubtitle(Subtitle):
     hash_verifiable = False
 
     def __init__(
-        self, language, page_link, filename, download_link, hearing_impaired, matches
+        self,
+        language,
+        page_link,
+        filename,
+        guessit_dict,
+        download_link,
+        hearing_impaired,
+        matches,
     ):
         super(SuchaSubtitle, self).__init__(
             language, hearing_impaired=hearing_impaired, page_link=page_url
         )
         self.download_link = download_link
         self.referer = page_link
+        self.guessit = guessit_dict
         self.language = language
         self.release_info = filename
         self.filename = filename
@@ -43,23 +50,33 @@ class SuchaSubtitle(Subtitle):
         return self.download_link
 
     def get_matches(self, video):
-        if video.resolution and video.resolution.lower() in self.release_info.lower():
+        if (
-            self.found_matches.add("resolution")
+            video.release_group
+            and str(video.release_group).lower() in self.filename.lower()
+        ):
+            self.found_matches.add("release_group")
 
-        if video.source and video.source.lower() in self.release_info.lower():
+        if video.source and video.source.lower() in self.guessit["source"].lower():
             self.found_matches.add("source")
 
-        if video.video_codec:
+        if (
-            if video.video_codec == "H.264" and "x264" in self.release_info.lower():
+            video.resolution
-                self.found_matches.add("video_codec")
+            and video.resolution.lower() in self.guessit["resolution"].lower()
-            elif video.video_codec == "H.265" and "x265" in self.release_info.lower():
+        ):
-                self.found_matches.add("video_codec")
+            self.found_matches.add("resolution")
-            elif video.video_codec.lower() in self.release_info.lower():
+
-                self.found_matches.add("video_codec")
+        if (
+            video.audio_codec
+            and video.audio_codec.lower() in self.guessit["audio_codec"].lower()
+        ):
+            self.found_matches.add("audio_codec")
+
+        if (
+            video.video_codec
+            and video.video_codec.lower() in self.guessit["video_codec"].lower()
+        ):
+            self.found_matches.add("video_codec")
 
-        if video.audio_codec:
-            if video.audio_codec.lower().replace(" ", ".") in self.release_info.lower():
-                self.found_matches.add("audio_codec")
         return self.found_matches
 
 
@@ -115,6 +132,7 @@ class SuchaProvider(Provider):
                     if q["query"].lower() in i["title"].lower():
                         matches.add("title")
                         matches.add("series")
+                        matches.add("imdb_id")
                         matches.add("season")
                         matches.add("episode")
                         matches.add("year")
@@ -122,11 +140,24 @@ class SuchaProvider(Provider):
                     matches.add("year")
                 if imdb_id:
                     matches.add("imdb_id")
+
+                # We'll add release group info (if found) to the pseudo filename 
+                # in order to show it in the manual search
+                filename = i["pseudo_file"]
+                if (
+                    video.release_group
+                    and str(video.release_group).lower() in i["original_description"]
+                ):
+                    filename = i["pseudo_file"].replace(
+                        ".es.srt", "-" + str(video.release_group) + ".es.srt"
+                    )
+
                 subtitles.append(
                     SuchaSubtitle(
                         language,
                         i["referer"],
-                        i["pseudo_file"],
+                        filename,
+                        i["guessit"],
                         i["download_url"],
                         i["hearing_impaired"],
                         matches,

libs/subliminal_patch/providers/tusubtitulo.py

@@ -21,6 +21,7 @@ BASE = "https://www.tusubtitulo.com/series.php?/"
 
 class TuSubtituloSubtitle(Subtitle):
     provider_name = "tusubtitulo"
+    hash_verifiable = False
 
     def __init__(self, language, filename, download_link, page_link, matches):
         super(TuSubtituloSubtitle, self).__init__(
@@ -52,9 +53,16 @@ class TuSubtituloSubtitle(Subtitle):
             elif video.video_codec.lower() in self.release_info.lower():
                 self.found_matches.add("video_codec")
 
+        if (
+            video.release_group
+            and video.release_group.lower() in self.release_info.lower()
+        ):
+            self.found_matches.add("release_group")
+
         if video.audio_codec:
             if video.audio_codec.lower().replace(" ", ".") in self.release_info.lower():
                 self.found_matches.add("audio_codec")
+
         return self.found_matches
 
 
@@ -62,8 +70,7 @@ class TuSubtituloProvider(Provider):
     """TuSubtitulo.com Provider"""
 
     BASE = "https://www.tusubtitulo.com/series.php?/"
-    languages = {Language.fromalpha2(l) for l in ["es"]}
+    languages = {Language.fromietf(lang) for lang in ["en", "es"]}
-    language_list = list(languages)
     logger.debug(languages)
     video_types = (Episode,)
 
@@ -146,7 +153,13 @@ class TuSubtituloProvider(Provider):
                     try:
                         content = tables[tr + inc].find_all("td")
                         language = content[4].text
-                        completed = content[5]
+                        if "eng" in language.lower():
+                            language = "en"
+                        elif "esp" in language.lower():
+                            language = "es"
+                        else:
+                            language = None
+                        completed = True if not "%" in content[5].text else False
                         url = content[6].find_all("a")[0].get("href")
                         sub_id = parse.parse_qs(parse.urlparse(url).query)["id"][0]
                         lang_id = parse.parse_qs(parse.urlparse(url).query)["lang"][0]
@@ -158,12 +171,13 @@ class TuSubtituloProvider(Provider):
                                 lang_id, sub_id, version_
                             )
                         )
-                        if "esp" in language.lower():
+                        if language and completed:
                             season_subs.append(
                                 {
                                     "episode_id": sub_id,
                                     "metadata": source_var,
                                     "download_url": download_url,
+                                    "language": language,
                                 }
                             )
                         inc += 1
@@ -180,6 +194,7 @@ class TuSubtituloProvider(Provider):
                             "episode_url": i["episode_url"],
                             "metadata": t["metadata"],
                             "download_url": t["download_url"],
+                            "language": t["language"],
                         }
                     )
         return final_list
@@ -206,7 +221,6 @@ class TuSubtituloProvider(Provider):
         logger.debug("Episode not found")
 
     def query(self, languages, video):
-        language = self.language_list[0]
         query = "{} {} {}".format(video.series, video.season, video.episode)
         logger.debug("Searching subtitles: {}".format(query))
         results = self.search(video.series, str(video.season), str(video.episode))
@@ -223,7 +237,7 @@ class TuSubtituloProvider(Provider):
                 matches.add("year")
                 subtitles.append(
                     TuSubtituloSubtitle(
-                        language,
+                        Language.fromietf(i["language"]),
                         i["metadata"],
                         i["download_url"],
                         i["episode_url"],

views/settingsproviders.html

@@ -516,7 +516,7 @@
                 <div class="form-group col-sm-8">
                     <label class="custom-control custom-checkbox">
                         <input type="checkbox" class="custom-control-input provider" id="subtitulamostv">
-                        <span class="custom-control-label">Spanish subtitles provider.</span>
+                        <span class="custom-control-label">Spanish Subtitles Provider.</span>
                     </label>
                 </div>
             </div>
@@ -528,7 +528,7 @@
                 <div class="form-group col-sm-8">
                     <label class="custom-control custom-checkbox">
                         <input type="checkbox" class="custom-control-input provider" id="sucha">
-                        <span class="custom-control-label">Spanish subtitles provider.</span>
+                        <span class="custom-control-label">Spanish Subtitles Provider.</span>
                     </label>
                 </div>
             </div>
@@ -596,7 +596,7 @@
                 <div class="form-group col-sm-8">
                     <label class="custom-control custom-checkbox">
                         <input type="checkbox" class="custom-control-input provider" id="tusubtitulo">
-                        <span class="custom-control-label">Spanish subtitles provider.</span>
+                        <span class="custom-control-label">Spanish/English Subtitles Provider for TV Shows.</span>
                     </label>
                 </div>
             </div>