[ie/sportdeuschland] Support embedded player URLs (#13833)

Closes #13766
Authored by: InvalidUsernameException

test/test_utils.py

@@ -1373,6 +1373,7 @@ class TestUtil(unittest.TestCase):
         self.assertEqual(parse_resolution('pre_1920x1080_post'), {'width': 1920, 'height': 1080})
         self.assertEqual(parse_resolution('ep1x2'), {})
         self.assertEqual(parse_resolution('1920, 1080'), {'width': 1920, 'height': 1080})
+        self.assertEqual(parse_resolution('1920w', lenient=True), {'width': 1920})
 
     def test_parse_bitrate(self):
         self.assertEqual(parse_bitrate(None), None)

yt_dlp/downloader/external.py

@@ -572,7 +572,21 @@ class FFmpegFD(ExternalFD):
             if end_time:
                 args += ['-t', str(end_time - start_time)]
 
-            args += [*self._configuration_args((f'_i{i + 1}', '_i')), '-i', fmt['url']]
+            url = fmt['url']
+            if self.params.get('enable_file_urls') and url.startswith('file:'):
+                # The default protocol_whitelist is 'file,crypto,data' when reading local m3u8 URLs,
+                # so only local segments can be read unless we also include 'http,https,tcp,tls'
+                args += ['-protocol_whitelist', 'file,crypto,data,http,https,tcp,tls']
+                # ffmpeg incorrectly handles 'file:' URLs by only removing the
+                # 'file:' prefix and treating the rest as if it's a normal filepath.
+                # FFmpegPostProcessor also depends on this behavior, so we need to fixup the URLs:
+                # - On Windows/Cygwin, replace 'file:///' and 'file://localhost/' with 'file:'
+                # - On *nix, replace 'file://localhost/' with 'file:/'
+                # Ref: https://github.com/yt-dlp/yt-dlp/issues/13781
+                #      https://trac.ffmpeg.org/ticket/2702
+                url = re.sub(r'^file://(?:localhost)?/', 'file:' if os.name == 'nt' else 'file:/', url)
+
+            args += [*self._configuration_args((f'_i{i + 1}', '_i')), '-i', url]
 
         if not (start_time or end_time) or not self.params.get('force_keyframes_at_cuts'):
             args += ['-c', 'copy']

yt_dlp/extractor/_extractors.py

@@ -640,6 +640,7 @@ from .fancode import (
     FancodeVodIE,
 )
 from .fathom import FathomIE
+from .faulio import FaulioLiveIE
 from .faz import FazIE
 from .fc2 import (
     FC2IE,
@@ -1568,6 +1569,7 @@ from .pluralsight import (
 )
 from .plutotv import PlutoTVIE
 from .plvideo import PlVideoIE
+from .plyr import PlyrEmbedIE
 from .podbayfm import (
     PodbayFMChannelIE,
     PodbayFMIE,

yt_dlp/extractor/faulio.py

@@ -0,0 +1,92 @@
+import re
+import urllib.parse
+
+from .common import InfoExtractor
+from ..utils import js_to_json, url_or_none
+from ..utils.traversal import traverse_obj
+
+
+class FaulioLiveIE(InfoExtractor):
+    _DOMAINS = (
+        'aloula.sba.sa',
+        'maraya.sba.net.ae',
+        'sat7plus.org',
+    )
+    _VALID_URL = fr'https?://(?:{"|".join(map(re.escape, _DOMAINS))})/(?:(?:en|ar|fa)/)?live/(?P<id>[a-zA-Z0-9-]+)'
+    _TESTS = [{
+        'url': 'https://aloula.sba.sa/live/saudiatv',
+        'info_dict': {
+            'id': 'aloula.faulio.com_saudiatv',
+            'title': str,
+            'description': str,
+            'ext': 'mp4',
+            'live_status': 'is_live',
+        },
+        'params': {
+            'skip_download': 'Livestream',
+        },
+    }, {
+        'url': 'https://maraya.sba.net.ae/live/1',
+        'info_dict': {
+            'id': 'maraya.faulio.com_1',
+            'title': str,
+            'description': str,
+            'ext': 'mp4',
+            'live_status': 'is_live',
+        },
+        'params': {
+            'skip_download': 'Livestream',
+        },
+    }, {
+        'url': 'https://sat7plus.org/live/pars',
+        'info_dict': {
+            'id': 'sat7.faulio.com_pars',
+            'title': str,
+            'description': str,
+            'ext': 'mp4',
+            'live_status': 'is_live',
+        },
+        'params': {
+            'skip_download': 'Livestream',
+        },
+    }, {
+        'url': 'https://sat7plus.org/fa/live/arabic',
+        'only_matching': True,
+    }]
+
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
+        webpage = self._download_webpage(url, video_id)
+
+        config_data = self._search_json(
+            r'window\.__NUXT__\.config=', webpage, 'config', video_id, transform_source=js_to_json)
+        api_base = config_data['public']['TRANSLATIONS_API_URL']
+
+        channel = traverse_obj(
+            self._download_json(f'{api_base}/channels', video_id),
+            (lambda k, v: v['url'] == video_id, any))
+
+        formats = []
+        subtitles = {}
+        if hls_url := traverse_obj(channel, ('streams', 'hls', {url_or_none})):
+            fmts, subs = self._extract_m3u8_formats_and_subtitles(
+                hls_url, video_id, 'mp4', m3u8_id='hls', live=True, fatal=False)
+            formats.extend(fmts)
+            self._merge_subtitles(subs, target=subtitles)
+
+        if mpd_url := traverse_obj(channel, ('streams', 'mpd', {url_or_none})):
+            fmts, subs = self._extract_mpd_formats_and_subtitles(
+                mpd_url, video_id, mpd_id='dash', fatal=False)
+            formats.extend(fmts)
+            self._merge_subtitles(subs, target=subtitles)
+
+        return {
+            'id': f'{urllib.parse.urlparse(api_base).hostname}_{video_id}',
+            **traverse_obj(channel, {
+                'title': ('title', {str}),
+                'description': ('description', {str}),
+            }),
+            'formats': formats,
+            'subtitles': subtitles,
+            'is_live': True,
+        }

yt_dlp/extractor/plyr.py

@@ -0,0 +1,104 @@
+import re
+
+from .common import InfoExtractor
+from .vimeo import VimeoIE
+
+
+class PlyrEmbedIE(InfoExtractor):
+    _VALID_URL = False
+    _WEBPAGE_TESTS = [{
+        # data-plyr-embed-id="https://player.vimeo.com/video/522319456/90e5c96063?dnt=1"
+        'url': 'https://www.dhm.de/zeughauskino/filmreihen/online-filmreihen/filme-des-marshall-plans/200000000-mouths/',
+        'info_dict': {
+            'id': '522319456',
+            'ext': 'mp4',
+            'title': '200.000.000 Mouths (1950–51)',
+            'uploader': 'Zeughauskino',
+            'uploader_url': '',
+            'comment_count': int,
+            'like_count': int,
+            'duration': 963,
+            'thumbnail': 'https://i.vimeocdn.com/video/1081797161-9f09ddb4b7faa86e834e006b8e4b9c2cbaa0baa7da493211bf0796ae133a5ab8-d',
+            'timestamp': 1615467405,
+            'upload_date': '20210311',
+            'release_timestamp': 1615467405,
+            'release_date': '20210311',
+        },
+        'params': {'skip_download': 'm3u8'},
+        'expected_warnings': ['Failed to parse XML: not well-formed'],
+    }, {
+        # data-plyr-provider="vimeo" data-plyr-embed-id="803435276"
+        'url': 'https://www.inarcassa.it/',
+        'info_dict': {
+            'id': '803435276',
+            'ext': 'mp4',
+            'title': 'HOME_Moto_Perpetuo',
+            'uploader': 'Inarcassa',
+            'uploader_url': '',
+            'duration': 38,
+            'thumbnail': 'https://i.vimeocdn.com/video/1663734769-945ad7ffabb16dbca009c023fd1d7b36bdb426a3dbae8345ed758136fe28f89a-d',
+        },
+        'params': {'skip_download': 'm3u8'},
+        'expected_warnings': ['Failed to parse XML: not well-formed'],
+    }, {
+        # data-plyr-embed-id="https://youtu.be/GF-BjYKoAqI"
+        'url': 'https://www.profile.nl',
+        'info_dict': {
+            'id': 'GF-BjYKoAqI',
+            'ext': 'mp4',
+            'title': 'PROFILE: Recruitment Profile',
+            'description': '',
+            'media_type': 'video',
+            'uploader': 'Profile Nederland',
+            'uploader_id': '@profilenederland',
+            'uploader_url': 'https://www.youtube.com/@profilenederland',
+            'channel': 'Profile Nederland',
+            'channel_id': 'UC9AUkB0Tv39-TBYjs05n3vg',
+            'channel_url': 'https://www.youtube.com/channel/UC9AUkB0Tv39-TBYjs05n3vg',
+            'channel_follower_count': int,
+            'view_count': int,
+            'like_count': int,
+            'age_limit': 0,
+            'duration': 39,
+            'thumbnail': 'https://i.ytimg.com/vi/GF-BjYKoAqI/maxresdefault.jpg',
+            'categories': ['Autos & Vehicles'],
+            'tags': [],
+            'timestamp': 1675692990,
+            'upload_date': '20230206',
+            'playable_in_embed': True,
+            'availability': 'public',
+            'live_status': 'not_live',
+        },
+    }, {
+        # data-plyr-embed-id="B1TZV8rNZoc" data-plyr-provider="youtube"
+        'url': 'https://www.vnis.edu.vn',
+        'info_dict': {
+            'id': 'vnis.edu',
+            'title': 'VNIS Education - Master Agent các Trường hàng đầu Bắc Mỹ',
+            'description': 'md5:4dafcf7335bb018780e4426da8ab8e4e',
+            'age_limit': 0,
+            'thumbnail': 'https://vnis.edu.vn/wp-content/uploads/2021/05/ve-welcome-en.png',
+            'timestamp': 1753233356,
+            'upload_date': '20250723',
+        },
+        'playlist_count': 3,
+    }]
+
+    @classmethod
+    def _extract_embed_urls(cls, url, webpage):
+        plyr_embeds = re.finditer(r'''(?x)
+            <div[^>]+(?:
+                data-plyr-embed-id="(?P<id1>[^"]+)"[^>]+data-plyr-provider="(?P<provider1>[^"]+)"|
+                data-plyr-provider="(?P<provider2>[^"]+)"[^>]+data-plyr-embed-id="(?P<id2>[^"]+)"
+            )[^>]*>''', webpage)
+        for mobj in plyr_embeds:
+            embed_id = mobj.group('id1') or mobj.group('id2')
+            provider = mobj.group('provider1') or mobj.group('provider2')
+            if provider == 'vimeo':
+                if not re.match(r'https?://', embed_id):
+                    embed_id = f'https://player.vimeo.com/video/{embed_id}'
+                yield VimeoIE._smuggle_referrer(embed_id, url)
+            elif provider == 'youtube':
+                if not re.match(r'https?://', embed_id):
+                    embed_id = f'https://youtube.com/watch?v={embed_id}'
+                yield embed_id

yt_dlp/extractor/sportdeutschland.py

@@ -8,84 +8,9 @@ from ..utils import (
 
 
 class SportDeutschlandIE(InfoExtractor):
-    _VALID_URL = r'https?://sportdeutschland\.tv/(?P<id>(?:[^/]+/)?[^?#/&]+)'
+    _VALID_URL = r'https?://(?:player\.)?sportdeutschland\.tv/(?P<id>(?:[^/?#]+/)?[^?#/&]+)'
     _TESTS = [{
-        'url': 'https://sportdeutschland.tv/blauweissbuchholztanzsport/buchholzer-formationswochenende-2023-samstag-1-bundesliga-landesliga',
-        'info_dict': {
-            'id': '9839a5c7-0dbb-48a8-ab63-3b408adc7b54',
-            'ext': 'mp4',
-            'title': 'Buchholzer Formationswochenende 2023 - Samstag - 1. Bundesliga / Landesliga',
-            'display_id': 'blauweissbuchholztanzsport/buchholzer-formationswochenende-2023-samstag-1-bundesliga-landesliga',
-            'description': 'md5:a288c794a5ee69e200d8f12982f81a87',
-            'live_status': 'was_live',
-            'channel': 'Blau-Weiss Buchholz Tanzsport',
-            'channel_url': 'https://sportdeutschland.tv/blauweissbuchholztanzsport',
-            'channel_id': '93ec33c9-48be-43b6-b404-e016b64fdfa3',
-            'duration': 32447,
-            'upload_date': '20230114',
-            'timestamp': 1673733618,
-        },
-        'skip': 'not found',
-    }, {
-        'url': 'https://sportdeutschland.tv/deutscherbadmintonverband/bwf-tour-1-runde-feld-1-yonex-gainward-german-open-2022-0',
-        'info_dict': {
-            'id': '95c80c52-6b9a-4ae9-9197-984145adfced',
-            'ext': 'mp4',
-            'title': 'BWF Tour: 1. Runde Feld 1 - YONEX GAINWARD German Open 2022',
-            'display_id': 'deutscherbadmintonverband/bwf-tour-1-runde-feld-1-yonex-gainward-german-open-2022-0',
-            'description': 'md5:2afb5996ceb9ac0b2ac81f563d3a883e',
-            'live_status': 'was_live',
-            'channel': 'Deutscher Badminton Verband',
-            'channel_url': 'https://sportdeutschland.tv/deutscherbadmintonverband',
-            'channel_id': '93ca5866-2551-49fc-8424-6db35af58920',
-            'duration': 41097,
-            'upload_date': '20220309',
-            'timestamp': 1646860727.0,
-        },
-        'skip': 'not found',
-    }, {
-        'url': 'https://sportdeutschland.tv/ggcbremen/formationswochenende-latein-2023',
-        'info_dict': {
-            'id': '9889785e-55b0-4d97-a72a-ce9a9f157cce',
-            'title': 'Formationswochenende Latein 2023 - Samstag',
-            'display_id': 'ggcbremen/formationswochenende-latein-2023',
-            'description': 'md5:6e4060d40ff6a8f8eeb471b51a8f08b2',
-            'live_status': 'was_live',
-            'channel': 'Grün-Gold-Club Bremen e.V.',
-            'channel_id': '9888f04e-bb46-4c7f-be47-df960a4167bb',
-            'channel_url': 'https://sportdeutschland.tv/ggcbremen',
-        },
-        'playlist_count': 3,
-        'playlist': [{
-            'info_dict': {
-                'id': '988e1fea-9d44-4fab-8c72-3085fb667547',
-                'ext': 'mp4',
-                'channel_url': 'https://sportdeutschland.tv/ggcbremen',
-                'channel_id': '9888f04e-bb46-4c7f-be47-df960a4167bb',
-                'channel': 'Grün-Gold-Club Bremen e.V.',
-                'duration': 86,
-                'title': 'Formationswochenende Latein 2023 - Samstag Part 1',
-                'upload_date': '20230225',
-                'timestamp': 1677349909,
-                'live_status': 'was_live',
-            },
-        }],
-        'skip': 'not found',
-    }, {
-        'url': 'https://sportdeutschland.tv/dtb/gymnastik-international-tag-1',
-        'info_dict': {
-            'id': '95d71b8a-370a-4b87-ad16-94680da18528',
-            'ext': 'mp4',
-            'title': r're:Gymnastik International - Tag 1 .+',
-            'display_id': 'dtb/gymnastik-international-tag-1',
-            'channel_id': '936ecef1-2f4a-4e08-be2f-68073cb7ecab',
-            'channel': 'Deutscher Turner-Bund',
-            'channel_url': 'https://sportdeutschland.tv/dtb',
-            'description': 'md5:07a885dde5838a6f0796ee21dc3b0c52',
-            'live_status': 'is_live',
-        },
-        'skip': 'live',
-    }, {
+        # Single-part video, direct link
         'url': 'https://sportdeutschland.tv/rostock-griffins/gfl2-rostock-griffins-vs-elmshorn-fighting-pirates',
         'md5': '35c11a19395c938cdd076b93bda54cde',
         'info_dict': {
@@ -100,7 +25,82 @@ class SportDeutschlandIE(InfoExtractor):
             'channel_id': '9635f21c-3f67-4584-9ce4-796e9a47276b',
             'timestamp': 1749913117,
             'upload_date': '20250614',
+            'duration': 12287.0,
         },
+    }, {
+        # Single-part video, embedded player link
+        'url': 'https://player.sportdeutschland.tv/9e9619c4-7d77-43c4-926d-49fb57dc06dc',
+        'info_dict': {
+            'id': '9f27a97d-1544-4d0b-aa03-48d92d17a03a',
+            'ext': 'mp4',
+            'title': 'GFL2: Rostock Griffins vs. Elmshorn Fighting Pirates',
+            'display_id': '9e9619c4-7d77-43c4-926d-49fb57dc06dc',
+            'channel': 'Rostock Griffins',
+            'channel_url': 'https://sportdeutschland.tv/rostock-griffins',
+            'live_status': 'was_live',
+            'description': 'md5:60cb00067e55dafa27b0933a43d72862',
+            'channel_id': '9635f21c-3f67-4584-9ce4-796e9a47276b',
+            'timestamp': 1749913117,
+            'upload_date': '20250614',
+            'duration': 12287.0,
+        },
+        'params': {'skip_download': True},
+    }, {
+        # Multi-part video
+        'url': 'https://sportdeutschland.tv/rhine-ruhr-2025-fisu-world-university-games/volleyball-w-japan-vs-brasilien-halbfinale-2',
+        'info_dict': {
+            'id': '9f63d737-2444-4e3a-a1ea-840df73fd481',
+            'display_id': 'rhine-ruhr-2025-fisu-world-university-games/volleyball-w-japan-vs-brasilien-halbfinale-2',
+            'title': 'Volleyball w: Japan vs. Braslien - Halbfinale 2',
+            'description': 'md5:0a17da15e48a687e6019639c3452572b',
+            'channel': 'Rhine-Ruhr 2025 FISU World University Games',
+            'channel_id': '9f5216be-a49d-470b-9a30-4fe9df993334',
+            'channel_url': 'https://sportdeutschland.tv/rhine-ruhr-2025-fisu-world-university-games',
+            'live_status': 'was_live',
+        },
+        'playlist_count': 2,
+        'playlist': [{
+            'info_dict': {
+                'id': '9f725a94-d43e-40ff-859d-13da3081bb04',
+                'ext': 'mp4',
+                'title': 'Volleyball w: Japan vs. Braslien - Halbfinale 2 Part 1',
+                'channel': 'Rhine-Ruhr 2025 FISU World University Games',
+                'channel_id': '9f5216be-a49d-470b-9a30-4fe9df993334',
+                'channel_url': 'https://sportdeutschland.tv/rhine-ruhr-2025-fisu-world-university-games',
+                'duration': 14773.0,
+                'timestamp': 1753085197,
+                'upload_date': '20250721',
+                'live_status': 'was_live',
+            },
+        }, {
+            'info_dict': {
+                'id': '9f725a94-370e-4477-89ac-1751098e3217',
+                'ext': 'mp4',
+                'title': 'Volleyball w: Japan vs. Braslien - Halbfinale 2 Part 2',
+                'channel': 'Rhine-Ruhr 2025 FISU World University Games',
+                'channel_id': '9f5216be-a49d-470b-9a30-4fe9df993334',
+                'channel_url': 'https://sportdeutschland.tv/rhine-ruhr-2025-fisu-world-university-games',
+                'duration': 14773.0,
+                'timestamp': 1753128421,
+                'upload_date': '20250721',
+                'live_status': 'was_live',
+            },
+        }],
+    }, {
+        # Livestream
+        'url': 'https://sportdeutschland.tv/dtb/gymnastik-international-tag-1',
+        'info_dict': {
+            'id': '95d71b8a-370a-4b87-ad16-94680da18528',
+            'ext': 'mp4',
+            'title': r're:Gymnastik International - Tag 1 .+',
+            'display_id': 'dtb/gymnastik-international-tag-1',
+            'channel_id': '936ecef1-2f4a-4e08-be2f-68073cb7ecab',
+            'channel': 'Deutscher Turner-Bund',
+            'channel_url': 'https://sportdeutschland.tv/dtb',
+            'description': 'md5:07a885dde5838a6f0796ee21dc3b0c52',
+            'live_status': 'is_live',
+        },
+        'skip': 'live',
     }]
 
     def _process_video(self, asset_id, video):

yt_dlp/utils/_utils.py

@@ -1875,6 +1875,11 @@ def parse_resolution(s, *, lenient=False):
     if mobj:
         return {'height': int(mobj.group(1)) * 540}
 
+    if lenient:
+        mobj = re.search(r'(?<!\d)(\d{2,5})w(?![a-zA-Z0-9])', s)
+        if mobj:
+            return {'width': int(mobj.group(1))}
+
     return {}