Merge branch 'yt-dlp:master' into master

- Support group /posts/ URLs
- Raise a proper error message if no formats are found

Closes #8669
Authored by: kclauhk

test/test_YoutubeDL.py

@@ -140,6 +140,8 @@ class TestFormatSelection(unittest.TestCase):
         test('example-with-dashes', 'example-with-dashes')
         test('all', '2', '47', '45', 'example-with-dashes', '35')
         test('mergeall', '2+47+45+example-with-dashes+35', multi=True)
+        # See: https://github.com/yt-dlp/yt-dlp/pulls/8797
+        test('7_a/worst', '35')
 
     def test_format_selection_audio(self):
         formats = [

yt_dlp/YoutubeDL.py

@@ -2465,9 +2465,16 @@ class YoutubeDL:
                 return selector_function(ctx_copy)
             return final_selector
 
-        stream = io.BytesIO(format_spec.encode())
+        # HACK: Python 3.12 changed the underlying parser, rendering '7_a' invalid
+        #       Prefix numbers with random letters to avoid it being classified as a number
+        #       See: https://github.com/yt-dlp/yt-dlp/pulls/8797
+        # TODO: Implement parser not reliant on tokenize.tokenize
+        prefix = ''.join(random.choices(string.ascii_letters, k=32))
+        stream = io.BytesIO(re.sub(r'\d[_\d]*', rf'{prefix}\g<0>', format_spec).encode())
         try:
-            tokens = list(_remove_unused_ops(tokenize.tokenize(stream.readline)))
+            tokens = list(_remove_unused_ops(
+                token._replace(string=token.string.replace(prefix, ''))
+                for token in tokenize.tokenize(stream.readline)))
         except tokenize.TokenError:
             raise syntax_error('Missing closing/opening brackets or parenthesis', (0, len(format_spec)))
 

yt_dlp/extractor/ard.py

@@ -292,7 +292,7 @@ class ARDIE(InfoExtractor):
     _TESTS = [{
         # available till 7.12.2023
         'url': 'https://www.daserste.de/information/talk/maischberger/videos/maischberger-video-424.html',
-        'md5': 'a438f671e87a7eba04000336a119ccc4',
+        'md5': '94812e6438488fb923c361a44469614b',
         'info_dict': {
             'id': 'maischberger-video-424',
             'display_id': 'maischberger-video-424',
@@ -403,26 +403,25 @@ class ARDBetaMediathekIE(ARDMediathekBaseIE):
     _VALID_URL = r'''(?x)https://
         (?:(?:beta|www)\.)?ardmediathek\.de/
         (?:(?P<client>[^/]+)/)?
-        (?:player|live|video|(?P<playlist>sendung|sammlung))/
+        (?:player|live|video|(?P<playlist>sendung|serie|sammlung))/
         (?:(?P<display_id>(?(playlist)[^?#]+?|[^?#]+))/)?
         (?P<id>(?(playlist)|Y3JpZDovL)[a-zA-Z0-9]+)
         (?(playlist)/(?P<season>\d+)?/?(?:[?#]|$))'''
 
     _TESTS = [{
-        'url': 'https://www.ardmediathek.de/video/filme-im-mdr/wolfsland-die-traurigen-schwestern/mdr-fernsehen/Y3JpZDovL21kci5kZS9iZWl0cmFnL2Ntcy8xZGY0ZGJmZS00ZWQwLTRmMGItYjhhYy0wOGQ4ZmYxNjVhZDI',
-        'md5': '3fd5fead7a370a819341129c8d713136',
+        'url': 'https://www.ardmediathek.de/video/filme-im-mdr/liebe-auf-vier-pfoten/mdr-fernsehen/Y3JpZDovL21kci5kZS9zZW5kdW5nLzI4MjA0MC80MjIwOTEtNDAyNTM0',
+        'md5': 'b6e8ab03f2bcc6e1f9e6cef25fcc03c4',
         'info_dict': {
-            'display_id': 'filme-im-mdr/wolfsland-die-traurigen-schwestern/mdr-fernsehen',
-            'id': '12172961',
-            'title': 'Wolfsland - Die traurigen Schwestern',
-            'description': r're:^Als der Polizeiobermeister Raaben',
-            'duration': 5241,
-            'thumbnail': 'https://api.ardmediathek.de/image-service/images/urn:ard:image:efa186f7b0054957',
-            'timestamp': 1670710500,
-            'upload_date': '20221210',
+            'display_id': 'filme-im-mdr/liebe-auf-vier-pfoten/mdr-fernsehen',
+            'id': '12939099',
+            'title': 'Liebe auf vier Pfoten',
+            'description': r're:^Claudia Schmitt, Anwältin in Salzburg',
+            'duration': 5222,
+            'thumbnail': 'https://api.ardmediathek.de/image-service/images/urn:ard:image:aee7cbf8f06de976?w=960&ch=ae4d0f2ee47d8b9b',
+            'timestamp': 1701343800,
+            'upload_date': '20231130',
             'ext': 'mp4',
-            'age_limit': 12,
-            'episode': 'Wolfsland - Die traurigen Schwestern',
+            'episode': 'Liebe auf vier Pfoten',
             'series': 'Filme im MDR'
         },
     }, {
@@ -454,7 +453,7 @@ class ARDBetaMediathekIE(ARDMediathekBaseIE):
             'duration': 915,
             'episode': 'tagesschau, 20:00 Uhr',
             'series': 'tagesschau',
-            'thumbnail': 'https://api.ardmediathek.de/image-service/images/urn:ard:image:fbb21142783b0a49',
+            'thumbnail': 'https://api.ardmediathek.de/image-service/images/urn:ard:image:fbb21142783b0a49?w=960&ch=ee69108ae344f678',
         },
     }, {
         'url': 'https://beta.ardmediathek.de/ard/video/Y3JpZDovL2Rhc2Vyc3RlLmRlL3RhdG9ydC9mYmM4NGM1NC0xNzU4LTRmZGYtYWFhZS0wYzcyZTIxNGEyMDE',
@@ -475,6 +474,10 @@ class ARDBetaMediathekIE(ARDMediathekBaseIE):
         # playlist of type 'sendung'
         'url': 'https://www.ardmediathek.de/ard/sendung/doctor-who/Y3JpZDovL3dkci5kZS9vbmUvZG9jdG9yIHdobw/',
         'only_matching': True,
+    }, {
+        # playlist of type 'serie'
+        'url': 'https://www.ardmediathek.de/serie/nachtstreife/staffel-1/Y3JpZDovL3N3ci5kZS9zZGIvc3RJZC8xMjQy/1',
+        'only_matching': True,
     }, {
         # playlist of type 'sammlung'
         'url': 'https://www.ardmediathek.de/ard/sammlung/team-muenster/5JpTzLSbWUAK8184IOvEir/',
@@ -487,10 +490,11 @@ class ARDBetaMediathekIE(ARDMediathekBaseIE):
         'only_matching': True,
     }]
 
-    def _ARD_load_playlist_snipped(self, playlist_id, display_id, client, mode, pageNumber):
+    def _ARD_load_playlist_snippet(self, playlist_id, display_id, client, mode, page_number):
         """ Query the ARD server for playlist information
         and returns the data in "raw" format """
-        if mode == 'sendung':
+        assert mode in ('sendung', 'serie', 'sammlung')
+        if mode in ('sendung', 'serie'):
             graphQL = json.dumps({
                 'query': '''{
                     showPage(
@@ -507,7 +511,7 @@ class ARDBetaMediathekIE(ARDMediathekBaseIE):
                             links { target { id href title } }
                             type
                         }
-                    }}''' % (client, playlist_id, pageNumber),
+                    }}''' % (client, playlist_id, page_number),
             }).encode()
         else:  # mode == 'sammlung'
             graphQL = json.dumps({
@@ -528,7 +532,7 @@ class ARDBetaMediathekIE(ARDMediathekBaseIE):
                                 type
                             }
                         }
-                    }}''' % (client, playlist_id, pageNumber),
+                    }}''' % (client, playlist_id, page_number),
             }).encode()
         # Ressources for ARD graphQL debugging:
         # https://api-test.ardmediathek.de/public-gateway
@@ -538,7 +542,7 @@ class ARDBetaMediathekIE(ARDMediathekBaseIE):
             data=graphQL,
             headers={'Content-Type': 'application/json'})['data']
         # align the structure of the returned data:
-        if mode == 'sendung':
+        if mode in ('sendung', 'serie'):
             show_page = show_page['showPage']
         else:  # mode == 'sammlung'
             show_page = show_page['morePage']['widget']
@@ -546,12 +550,12 @@ class ARDBetaMediathekIE(ARDMediathekBaseIE):
 
     def _ARD_extract_playlist(self, url, playlist_id, display_id, client, mode):
         """ Collects all playlist entries and returns them as info dict.
-        Supports playlists of mode 'sendung' and 'sammlung', and also nested
-        playlists. """
+        Supports playlists of mode 'sendung', 'serie', and 'sammlung',
+        as well as nested playlists. """
         entries = []
         pageNumber = 0
         while True:  # iterate by pageNumber
-            show_page = self._ARD_load_playlist_snipped(
+            show_page = self._ARD_load_playlist_snippet(
                 playlist_id, display_id, client, mode, pageNumber)
             for teaser in show_page['teasers']:  # process playlist items
                 if '/compilation/' in teaser['links']['target']['href']:

yt_dlp/extractor/facebook.py

@@ -52,7 +52,7 @@ class FacebookIE(InfoExtractor):
                             )\?(?:.*?)(?:v|video_id|story_fbid)=|
                             [^/]+/videos/(?:[^/]+/)?|
                             [^/]+/posts/|
-                            groups/[^/]+/permalink/|
+                            groups/[^/]+/(?:permalink|posts)/|
                             watchparty/
                         )|
                     facebook:
@@ -232,6 +232,21 @@ class FacebookIE(InfoExtractor):
             'uploader_id': '100013949973717',
         },
         'skip': 'Requires logging in',
+    }, {
+        # data.node.comet_sections.content.story.attachments[].throwbackStyles.attachment_target_renderer.attachment.target.attachments[].styles.attachment.media
+        'url': 'https://www.facebook.com/groups/1645456212344334/posts/3737828833107051/',
+        'info_dict': {
+            'id': '1569199726448814',
+            'ext': 'mp4',
+            'title': 'Pence MUST GO!',
+            'description': 'Vickie Gentry shared a memory.',
+            'timestamp': 1511548260,
+            'upload_date': '20171124',
+            'uploader': 'Vickie Gentry',
+            'uploader_id': 'pfbid0FuZhHCeWDAxWxEbr3yKPFaRstXvRxgsp9uCPG6GjD4J2AitB35NUAuJ4Q75KcjiDl',
+            'thumbnail': r're:^https?://.*',
+            'duration': 148.435,
+        },
     }, {
         'url': 'https://www.facebook.com/video.php?v=10204634152394104',
         'only_matching': True,
@@ -612,9 +627,11 @@ class FacebookIE(InfoExtractor):
                 nodes = variadic(traverse_obj(data, 'nodes', 'node') or [])
                 attachments = traverse_obj(nodes, (
                     ..., 'comet_sections', 'content', 'story', (None, 'attached_story'), 'attachments',
-                    ..., ('styles', 'style_type_renderer'), 'attachment'), expected_type=dict) or []
+                    ..., ('styles', 'style_type_renderer', ('throwbackStyles', 'attachment_target_renderer')),
+                    'attachment', {dict}))
                 for attachment in attachments:
-                    ns = try_get(attachment, lambda x: x['all_subattachments']['nodes'], list) or []
+                    ns = traverse_obj(attachment, ('all_subattachments', 'nodes', ..., {dict}),
+                                      ('target', 'attachments', ..., 'styles', 'attachment', {dict}))
                     for n in ns:
                         parse_attachment(n)
                     parse_attachment(attachment)
@@ -637,7 +654,7 @@ class FacebookIE(InfoExtractor):
                 if len(entries) > 1:
                     return self.playlist_result(entries, video_id)
 
-                video_info = entries[0]
+                video_info = entries[0] if entries else {'id': video_id}
                 webpage_info = extract_metadata(webpage)
                 # honor precise duration in video info
                 if video_info.get('duration'):

yt_dlp/extractor/instagram.py

@@ -10,6 +10,7 @@ from ..utils import (
     ExtractorError,
     decode_base_n,
     encode_base_n,
+    filter_dict,
     float_or_none,
     format_field,
     get_element_by_attribute,
@@ -703,28 +704,31 @@ class InstagramStoryIE(InstagramBaseIE):
         user_info = self._search_json(r'"user":', story_info, 'user info', story_id, fatal=False)
         if not user_info:
             self.raise_login_required('This content is unreachable')
-        user_id = user_info.get('id')
 
+        user_id = traverse_obj(user_info, 'pk', 'id', expected_type=str)
         story_info_url = user_id if username != 'highlights' else f'highlight:{story_id}'
+        if not story_info_url:  # user id is only mandatory for non-highlights
+            raise ExtractorError('Unable to extract user id')
+
         videos = traverse_obj(self._download_json(
             f'{self._API_BASE_URL}/feed/reels_media/?reel_ids={story_info_url}',
             story_id, errnote=False, fatal=False, headers=self._API_HEADERS), 'reels')
         if not videos:
             self.raise_login_required('You need to log in to access this content')
 
-        full_name = traverse_obj(videos, (f'highlight:{story_id}', 'user', 'full_name'), (str(user_id), 'user', 'full_name'))
+        full_name = traverse_obj(videos, (f'highlight:{story_id}', 'user', 'full_name'), (user_id, 'user', 'full_name'))
         story_title = traverse_obj(videos, (f'highlight:{story_id}', 'title'))
         if not story_title:
             story_title = f'Story by {username}'
 
-        highlights = traverse_obj(videos, (f'highlight:{story_id}', 'items'), (str(user_id), 'items'))
+        highlights = traverse_obj(videos, (f'highlight:{story_id}', 'items'), (user_id, 'items'))
         info_data = []
         for highlight in highlights:
             highlight_data = self._extract_product(highlight)
             if highlight_data.get('formats'):
                 info_data.append({
-                    **highlight_data,
                     'uploader': full_name,
                     'uploader_id': user_id,
+                    **filter_dict(highlight_data),
                 })
         return self.playlist_result(info_data, playlist_id=story_id, playlist_title=story_title)

yt_dlp/extractor/litv.py

@@ -6,6 +6,7 @@ from ..utils import (
     int_or_none,
     smuggle_url,
     traverse_obj,
+    try_call,
     unsmuggle_url,
 )
 
@@ -96,13 +97,22 @@ class LiTVIE(InfoExtractor):
             r'uiHlsUrl\s*=\s*testBackendData\(([^;]+)\);',
             webpage, 'video data', default='{}'), video_id)
         if not video_data:
-            payload = {
-                'assetId': program_info['assetId'],
+            payload = {'assetId': program_info['assetId']}
+            puid = try_call(lambda: self._get_cookies('https://www.litv.tv/')['PUID'].value)
+            if puid:
+                payload.update({
+                    'type': 'auth',
+                    'puid': puid,
+                })
+                endpoint = 'getUrl'
+            else:
+                payload.update({
                     'watchDevices': program_info['watchDevices'],
                     'contentType': program_info['contentType'],
-            }
+                })
+                endpoint = 'getMainUrlNoAuth'
             video_data = self._download_json(
-                'https://www.litv.tv/vod/ajax/getMainUrlNoAuth', video_id,
+                f'https://www.litv.tv/vod/ajax/{endpoint}', video_id,
                 data=json.dumps(payload).encode('utf-8'),
                 headers={'Content-Type': 'application/json'})
 

yt_dlp/extractor/twitter.py

@@ -10,6 +10,7 @@ from ..compat import (
     compat_urllib_parse_unquote,
     compat_urllib_parse_urlparse,
 )
+from ..networking.exceptions import HTTPError
 from ..utils import (
     ExtractorError,
     dict_get,
@@ -1317,20 +1318,7 @@ class TwitterIE(TwitterBaseIE):
             }
         }
 
-    def _extract_status(self, twid):
-        if self.is_logged_in or self._selected_api == 'graphql':
-            status = self._graphql_to_legacy(self._call_graphql_api(self._GRAPHQL_ENDPOINT, twid), twid)
-
-        elif self._selected_api == 'legacy':
-            status = self._call_api(f'statuses/show/{twid}.json', twid, {
-                'cards_platform': 'Web-12',
-                'include_cards': 1,
-                'include_reply_count': 1,
-                'include_user_entities': 0,
-                'tweet_mode': 'extended',
-            })
-
-        elif self._selected_api == 'syndication':
+    def _call_syndication_api(self, twid):
         self.report_warning(
             'Not all metadata or media is available via syndication endpoint', twid, only_once=True)
         status = self._download_json(
@@ -1350,8 +1338,31 @@ class TwitterIE(TwitterBaseIE):
             media.append(detail)
         status['extended_entities'] = {'media': media}
 
-        else:
-            raise ExtractorError(f'"{self._selected_api}" is not a valid API selection', expected=True)
+        return status
+
+    def _extract_status(self, twid):
+        if self._selected_api not in ('graphql', 'legacy', 'syndication'):
+            raise ExtractorError(f'{self._selected_api!r} is not a valid API selection', expected=True)
+
+        try:
+            if self.is_logged_in or self._selected_api == 'graphql':
+                status = self._graphql_to_legacy(self._call_graphql_api(self._GRAPHQL_ENDPOINT, twid), twid)
+            elif self._selected_api == 'legacy':
+                status = self._call_api(f'statuses/show/{twid}.json', twid, {
+                    'cards_platform': 'Web-12',
+                    'include_cards': 1,
+                    'include_reply_count': 1,
+                    'include_user_entities': 0,
+                    'tweet_mode': 'extended',
+                })
+        except ExtractorError as e:
+            if not isinstance(e.cause, HTTPError) or not e.cause.status == 429:
+                raise
+            self.report_warning('Rate-limit exceeded; falling back to syndication endpoint')
+            status = self._call_syndication_api(twid)
+
+        if self._selected_api == 'syndication':
+            status = self._call_syndication_api(twid)
 
         return traverse_obj(status, 'retweeted_status', None, expected_type=dict) or {}
 
@@ -1416,8 +1427,8 @@ class TwitterIE(TwitterBaseIE):
                 'thumbnails': thumbnails,
                 'view_count': traverse_obj(media, ('mediaStats', 'viewCount', {int_or_none})),  # No longer available
                 'duration': float_or_none(traverse_obj(media, ('video_info', 'duration_millis')), 1000),
-                # The codec of http formats are unknown
-                '_format_sort_fields': ('res', 'br', 'size', 'proto'),
+                # Prioritize m3u8 formats for compat, see https://github.com/yt-dlp/yt-dlp/issues/8117
+                '_format_sort_fields': ('res', 'proto:m3u8', 'br', 'size'),  # http format codec is unknown
             }
 
         def extract_from_card_info(card):