[ie/mlbtv] Fix extraction (#10296)

[yt-dlp.git] / yt_dlp / extractor / bitchute.py

diff --git a/yt_dlp/extractor/bitchute.py b/yt_dlp/extractor/bitchute.py
index 41367c5b9b971346d11e1a2edb79513b48c1c3ec..c83222ea5b4f4a6ab2b576132d50c99523bdf007 100644 (file)
--- a/yt_dlp/extractor/bitchute.py
+++ b/yt_dlp/extractor/bitchute.py
@@ -24,7 +24,7 @@
 
 
 class BitChuteIE(InfoExtractor):
-    _VALID_URL = r'https?://(?:www\.)?bitchute\.com/(?:video|embed|torrent/[^/]+)/(?P<id>[^/?#&]+)'
+    _VALID_URL = r'https?://(?:(?:www|old)\.)?bitchute\.com/(?:video|embed|torrent/[^/]+)/(?P<id>[^/?#&]+)'
     _EMBED_REGEX = [rf'<(?:script|iframe)[^>]+\bsrc=(["\'])(?P<url>{_VALID_URL})']
     _TESTS = [{
         'url': 'https://www.bitchute.com/video/UGlrF9o9b-Q/',
@@ -39,7 +39,7 @@ class BitChuteIE(InfoExtractor):
             'upload_date': '20170103',
             'uploader_url': 'https://www.bitchute.com/profile/I5NgtHZn9vPj/',
             'channel': 'BitChute',
-            'channel_url': 'https://www.bitchute.com/channel/bitchute/'
+            'channel_url': 'https://www.bitchute.com/channel/bitchute/',
         },
     }, {
         # test case: video with different channel and uploader
@@ -55,7 +55,7 @@ class BitChuteIE(InfoExtractor):
             'upload_date': '20231106',
             'uploader_url': 'https://www.bitchute.com/profile/9K0kUWA9zmd9/',
             'channel': 'Full Measure with Sharyl Attkisson',
-            'channel_url': 'https://www.bitchute.com/channel/sharylattkisson/'
+            'channel_url': 'https://www.bitchute.com/channel/sharylattkisson/',
         },
     }, {
         # video not downloadable in browser, but we can recover it
@@ -72,7 +72,7 @@ class BitChuteIE(InfoExtractor):
             'upload_date': '20181113',
             'uploader_url': 'https://www.bitchute.com/profile/I5NgtHZn9vPj/',
             'channel': 'BitChute',
-            'channel_url': 'https://www.bitchute.com/channel/bitchute/'
+            'channel_url': 'https://www.bitchute.com/channel/bitchute/',
         },
         'params': {'check_formats': None},
     }, {
@@ -91,6 +91,9 @@ class BitChuteIE(InfoExtractor):
     }, {
         'url': 'https://www.bitchute.com/torrent/Zee5BE49045h/szoMrox2JEI.webtorrent',
         'only_matching': True,
+    }, {
+        'url': 'https://old.bitchute.com/video/UGlrF9o9b-Q/',
+        'only_matching': True,
     }]
     _GEO_BYPASS = False
 
@@ -115,7 +118,7 @@ def _check_format(self, video_url, video_id):
                 continue
             return {
                 'url': url,
-                'filesize': int_or_none(response.headers.get('Content-Length'))
+                'filesize': int_or_none(response.headers.get('Content-Length')),
             }
 
     def _raise_if_restricted(self, webpage):
@@ -132,7 +135,7 @@ def _make_url(html):
     def _real_extract(self, url):
         video_id = self._match_id(url)
         webpage = self._download_webpage(
-            f'https://www.bitchute.com/video/{video_id}', video_id, headers=self._HEADERS)
+            f'https://old.bitchute.com/video/{video_id}', video_id, headers=self._HEADERS)
 
         self._raise_if_restricted(webpage)
         publish_date = clean_html(get_element_by_class('video-publish-date', webpage))
@@ -171,13 +174,13 @@ def _real_extract(self, url):
 
 
 class BitChuteChannelIE(InfoExtractor):
-    _VALID_URL = r'https?://(?:www\.)?bitchute\.com/(?P<type>channel|playlist)/(?P<id>[^/?#&]+)'
+    _VALID_URL = r'https?://(?:(?:www|old)\.)?bitchute\.com/(?P<type>channel|playlist)/(?P<id>[^/?#&]+)'
     _TESTS = [{
         'url': 'https://www.bitchute.com/channel/bitchute/',
         'info_dict': {
             'id': 'bitchute',
             'title': 'BitChute',
-            'description': 'md5:5329fb3866125afa9446835594a9b138',
+            'description': 'md5:2134c37d64fc3a4846787c402956adac',
         },
         'playlist': [
             {
@@ -185,7 +188,6 @@ class BitChuteChannelIE(InfoExtractor):
                 'info_dict': {
                     'id': 'UGlrF9o9b-Q',
                     'ext': 'mp4',
-                    'filesize': None,
                     'title': 'This is the first video on #BitChute !',
                     'description': 'md5:a0337e7b1fe39e32336974af8173a034',
                     'thumbnail': r're:^https?://.*\.jpg$',
@@ -197,7 +199,7 @@ class BitChuteChannelIE(InfoExtractor):
                     'duration': 16,
                     'view_count': int,
                 },
-            }
+            },
         ],
         'params': {
             'skip_download': True,
@@ -210,7 +212,10 @@ class BitChuteChannelIE(InfoExtractor):
             'id': 'wV9Imujxasw9',
             'title': 'Bruce MacDonald and "The Light of Darkness"',
             'description': 'md5:747724ef404eebdfc04277714f81863e',
-        }
+        },
+    }, {
+        'url': 'https://old.bitchute.com/playlist/wV9Imujxasw9/',
+        'only_matching': True,
     }]
 
     _TOKEN = 'zyG6tQcGPE5swyAEFLqKUwMuMMuF6IO2DZ6ZDQjGfsL0e4dcTLwqkTTul05Jdve7'
@@ -225,13 +230,13 @@ class BitChuteChannelIE(InfoExtractor):
             'container': 'playlist-video',
             'title': 'title',
             'description': 'description',
-        }
+        },
 
     }
 
     @staticmethod
     def _make_url(playlist_id, playlist_type):
-        return f'https://www.bitchute.com/{playlist_type}/{playlist_id}/'
+        return f'https://old.bitchute.com/{playlist_type}/{playlist_id}/'
 
     def _fetch_page(self, playlist_id, playlist_type, page_num):
         playlist_url = self._make_url(playlist_id, playlist_type)