yt_dlp/extractor/bilibili.py

@@ -1649,6 +1649,7 @@ class BiliIntlBaseIE(InfoExtractor):
         return data
 
     def _get_subtitles(self, *, ep_id=None, aid=None):
+        subtitles = {}
         sub_json = self._call_api(
             '/web/v2/subtitle', ep_id or aid, fatal=False,
             note='Downloading subtitles list', errnote='Unable to download subtitles list',
@@ -1658,34 +1659,43 @@ class BiliIntlBaseIE(InfoExtractor):
                 'episode_id': ep_id,
                 'aid': aid,
             })) or {}
-        subtitles = {}
-        fetched_urls = set()
-        for sub in traverse_obj(sub_json, (('subtitles', 'video_subtitle'), ..., {dict})):
-            for url in traverse_obj(sub, ((None, 'ass', 'srt'), 'url', {url_or_none})):
-                if url in fetched_urls:
-                    continue
-                fetched_urls.add(url)
-                sub_ext = determine_ext(url)
-                sub_lang = sub.get('lang_key') or 'en'
 
-                if sub_ext == 'ass':
+        for sub in traverse_obj(sub_json, ('subtitles', lambda _, v: v['url'])):
+            sub_ext = determine_ext(sub['url'])
+            sub_lang = sub.get('lang_key') or 'en'
+
+            sub_data = self._download_webpage(
+                sub['url'], ep_id or aid, fatal=False,
+                encoding='utf-8-sig' if sub_ext == 'ass' else None,
+                note=f'Downloading subtitles{format_field(sub, "lang", " for %s")} ({sub_lang})',
+                errnote='Unable to download subtitles')
+            if sub_ext != 'ass':
+                sub_ext, sub_data = 'srt', self.json2srt(sub_data)
+
+            if sub_data:
+                subtitles.setdefault(sub_lang, []).append({
+                    'ext': sub_ext,
+                    'data': sub_data
+                })
+
+        for sub in traverse_obj(sub_json, ('video_subtitle', ..., {dict})):
+            sub_lang = sub.get('lang_key') or 'en'
+            if sub.get('ass'):
+                subtitles.setdefault(sub_lang, []).append({
+                    'ext': 'ass',
+                    'url': traverse_obj(sub, ('ass', 'url'))
+                })
+
+            if sub.get('srt'):
+                sub_data = self._download_json(
+                    traverse_obj(sub, ('srt', 'url')), ep_id or aid, fatal=False,
+                    note=f'Downloading subtitles{format_field(sub, "lang", " for %s")} ({sub_lang})',
+                    errnote='Unable to download subtitles')
+                if sub_data:
                     subtitles.setdefault(sub_lang, []).append({
-                        'ext': 'ass',
+                        'ext': 'srt',
-                        'url': url,
+                        'data': self.json2srt(sub_data)
                     })
-                elif sub_ext == 'json':
-                    sub_data = self._download_json(
-                        url, ep_id or aid, fatal=False,
-                        note=f'Downloading subtitles{format_field(sub, "lang", " for %s")} ({sub_lang})',
-                        errnote='Unable to download subtitles')
-
-                    if sub_data:
-                        subtitles.setdefault(sub_lang, []).append({
-                            'ext': 'srt',
-                            'data': self.json2srt(sub_data),
-                        })
-                else:
-                    self.report_warning('Unexpected subtitle extension', ep_id or aid)
 
         return subtitles