[extractor/bigo] Fix extractor (#4312)

Closes #4139

Authored by: Lesmiscore

yt_dlp/extractor/_extractors.py

@@ -104,6 +104,10 @@ from .atttechchannel import ATTTechChannelIE
 from .atvat import ATVAtIE
 from .audimedia import AudiMediaIE
 from .audioboom import AudioBoomIE
+from .audiodraft import (
+    AudiodraftCustomIE,
+    AudiodraftGenericIE,
+)
 from .audiomack import AudiomackIE, AudiomackAlbumIE
 from .audius import (
     AudiusIE,

yt_dlp/extractor/audiodraft.py

@@ -0,0 +1,93 @@
+from .common import InfoExtractor
+from ..utils import int_or_none
+
+
+class AudiodraftBaseIE(InfoExtractor):
+    def _audiodraft_extract_from_id(self, player_entry_id):
+        data_json = self._download_json(
+            'https://www.audiodraft.com/scripts/general/player/getPlayerInfoNew.php', player_entry_id,
+            headers={
+                'Content-type': 'application/x-www-form-urlencoded; charset=UTF-8',
+                'X-Requested-With': 'XMLHttpRequest',
+            }, data=f'id={player_entry_id}'.encode('utf-8'))
+
+        return {
+            'id': str(data_json['entry_id']),
+            'title': data_json.get('entry_title'),
+            'url': data_json['path'],
+            'vcodec': 'none',
+            'ext': 'mp3',
+            'uploader': data_json.get('designer_name'),
+            'uploader_id': data_json.get('designer_id'),
+            'webpage_url': data_json.get('entry_url'),
+            'like_count': int_or_none(data_json.get('entry_likes')),
+            'average_rating': int_or_none(data_json.get('entry_rating')),
+        }
+
+
+class AudiodraftCustomIE(AudiodraftBaseIE):
+    IE_NAME = 'Audiodraft:custom'
+    _VALID_URL = r'https?://(?:[-\w]+)\.audiodraft\.com/entry/(?P<id>\d+)'
+
+    _TESTS = [{
+        'url': 'http://nokiatune.audiodraft.com/entry/5874',
+        'info_dict': {
+            'id': '9485',
+            'ext': 'mp3',
+            'title': 'Hula Hula Calls',
+            'uploader': 'unclemaki',
+            'uploader_id': '13512',
+            'average_rating': 5,
+            'like_count': int,
+        },
+    }, {
+        'url': 'http://vikinggrace.audiodraft.com/entry/501',
+        'info_dict': {
+            'id': '22241',
+            'ext': 'mp3',
+            'title': 'MVG Happy',
+            'uploader': 'frog',
+            'uploader_id': '19142',
+            'average_rating': 5,
+            'like_count': int,
+        },
+    }, {
+        'url': 'http://timferriss.audiodraft.com/entry/765',
+        'info_dict': {
+            'id': '19710',
+            'ext': 'mp3',
+            'title': 'ferris03',
+            'uploader': 'malex',
+            'uploader_id': '17335',
+            'average_rating': 5,
+            'like_count': int,
+        },
+    }]
+
+    def _real_extract(self, url):
+        id = self._match_id(url)
+        webpage = self._download_webpage(url, id)
+        player_entry_id = self._search_regex(r'playAudio\(\'(player_entry_\d+)\'\);', webpage, id, 'play entry id')
+        return self._audiodraft_extract_from_id(player_entry_id)
+
+
+class AudiodraftGenericIE(AudiodraftBaseIE):
+    IE_NAME = 'Audiodraft:generic'
+    _VALID_URL = r'https?://www\.audiodraft\.com/contests/[^/#]+#entries&eid=(?P<id>\d+)'
+
+    _TESTS = [{
+        'url': 'https://www.audiodraft.com/contests/570-Score-A-Video-Surprise-Us#entries&eid=30138',
+        'info_dict': {
+            'id': '30138',
+            'ext': 'mp3',
+            'title': 'DROP in sound_V2',
+            'uploader': 'TiagoSilva',
+            'uploader_id': '19452',
+            'average_rating': 4,
+            'like_count': int,
+        },
+    }]
+
+    def _real_extract(self, url):
+        id = self._match_id(url)
+        return self._audiodraft_extract_from_id(f'player_entry_{id}')

yt_dlp/extractor/bigo.py

@@ -28,7 +28,7 @@ class BigoIE(InfoExtractor):
         user_id = self._match_id(url)
 
         info_raw = self._download_json(
-            'https://bigo.tv/studio/getInternalStudioInfo',
+            'https://ta.bigo.tv/official_website/studio/getInternalStudioInfo',
             user_id, data=urlencode_postdata({'siteId': user_id}))
 
         if not isinstance(info_raw, dict):
@@ -41,14 +41,14 @@ class BigoIE(InfoExtractor):
         if not info.get('alive'):
             raise ExtractorError('This user is offline.', expected=True)
 
+        formats, subs = self._extract_m3u8_formats_and_subtitles(
+            info.get('hls_src'), user_id, 'mp4', 'm3u8')
+
         return {
             'id': info.get('roomId') or user_id,
             'title': info.get('roomTopic') or info.get('nick_name') or user_id,
-            'formats': [{
-                'url': info.get('hls_src'),
-                'ext': 'mp4',
-                'protocol': 'm3u8',
-            }],
+            'formats': formats,
+            'subtitles': subs,
             'thumbnail': info.get('snapshot'),
             'uploader': info.get('nick_name'),
             'uploader_id': user_id,

yt_dlp/extractor/fifa.py

@@ -16,21 +16,21 @@ class FifaIE(InfoExtractor):
             'title': 'Italy v France | Final | 2006 FIFA World Cup Germany™ | Full Match Replay',
             'description': 'md5:f4520d0ee80529c8ba4134a7d692ff8b',
             'ext': 'mp4',
-            'categories': ['FIFA Tournaments', 'Replay'],
+            'categories': ['FIFA Tournaments'],
             'thumbnail': 'https://digitalhub.fifa.com/transform/fa6f0b3e-a2e9-4cf7-9f32-53c57bcb7360/2006_Final_ITA_FRA',
-            'duration': 8164,
+            'duration': 8165,
         },
         'params': {'skip_download': 'm3u8'},
     }, {
         'url': 'https://www.fifa.com/fifaplus/pt/watch/1cg5r5Qt6Qt12ilkDgb1sV',
         'info_dict': {
             'id': '1cg5r5Qt6Qt12ilkDgb1sV',
-            'title': 'Brasil x Alemanha | Semifinais | Copa do Mundo FIFA Brasil 2014 | Compacto',
-            'description': 'md5:ba4ffcc084802b062beffc3b4c4b19d6',
+            'title': 'Brazil v Germany | Semi-finals | 2014 FIFA World Cup Brazil™ | Extended Highlights',
+            'description': 'md5:d908c74ee66322b804ae2e521b02a855',
             'ext': 'mp4',
             'categories': ['FIFA Tournaments', 'Highlights'],
             'thumbnail': 'https://digitalhub.fifa.com/transform/d8fe6f61-276d-4a73-a7fe-6878a35fd082/FIFAPLS_100EXTHL_2014BRAvGER_TMB',
-            'duration': 901,
+            'duration': 902,
             'release_timestamp': 1404777600,
             'release_date': '20140708',
         },
@@ -39,8 +39,8 @@ class FifaIE(InfoExtractor):
         'url': 'https://www.fifa.com/fifaplus/fr/watch/3C6gQH9C2DLwzNx7BMRQdp',
         'info_dict': {
             'id': '3C6gQH9C2DLwzNx7BMRQdp',
-            'title': 'Le but de Josimar contre le Irlande du Nord | Buts classiques',
-            'description': 'md5:16f9f789f09960bfe7220fe67af31f34',
+            'title': 'Josimar goal against Northern Ireland | Classic Goals',
+            'description': 'md5:cbe7e7bb52f603c9f1fe9a4780fe983b',
             'ext': 'mp4',
             'categories': ['FIFA Tournaments', 'Goal'],
             'duration': 28,
@@ -56,27 +56,13 @@ class FifaIE(InfoExtractor):
         preconnect_link = self._search_regex(
             r'<link[^>]+rel\s*=\s*"preconnect"[^>]+href\s*=\s*"([^"]+)"', webpage, 'Preconnect Link')
 
-        json_data = self._download_json(
-            f'{preconnect_link}/video/GetVideoPlayerData/{video_id}', video_id,
-            'Downloading Video Player Data', query={'includeIdents': True, 'locale': locale})
-
         video_details = self._download_json(
             f'{preconnect_link}/sections/videoDetails/{video_id}', video_id, 'Downloading Video Details', fatal=False)
 
         preplay_parameters = self._download_json(
-            f'{preconnect_link}/video/GetVerizonPreplayParameters', video_id, 'Downloading Preplay Parameters', query={
-                'entryId': video_id,
-                'assetId': json_data['verizonAssetId'],
-                'useExternalId': False,
-                'requiresToken': json_data['requiresToken'],
-                'adConfig': 'fifaplusvideo',
-                'prerollAds': True,
-                'adVideoId': json_data['externalVerizonAssetId'],
-                'preIdentId': json_data['preIdentId'],
-                'postIdentId': json_data['postIdentId'],
-            })
+            f'{preconnect_link}/video/GetVerizonPreplayParameters/{video_id}', video_id, 'Downloading Preplay Parameters')['preplayParameters']
 
-        cid = f'{json_data["preIdentId"]},{json_data["verizonAssetId"]},{json_data["postIdentId"]}'
+        cid = preplay_parameters['contentId']
         content_data = self._download_json(
             f'https://content.uplynk.com/preplay/{cid}/multiple.json', video_id, 'Downloading Content Data', query={
                 'v': preplay_parameters['preplayAPIVersion'],
@@ -98,9 +84,9 @@ class FifaIE(InfoExtractor):
 
         return {
             'id': video_id,
-            'title': json_data.get('title'),
-            'description': json_data.get('description'),
-            'duration': int_or_none(json_data.get('duration')),
+            'title': video_details.get('title'),
+            'description': video_details.get('description'),
+            'duration': int_or_none(video_details.get('duration')),
             'release_timestamp': unified_timestamp(video_details.get('dateOfRelease')),
             'categories': traverse_obj(video_details, (('videoCategory', 'videoSubcategory'),)),
             'thumbnail': traverse_obj(video_details, ('backgroundImage', 'src')),