Mirroring
/
yt-dlp
mirror of https://github.com/yt-dlp/yt-dlp
1
0
Fork 0
Code Issues Releases Wiki Activity

Merge branch 'yt-dlp:master' into misc-2025-08

Browse Source
pull/13852/head
bashonly 3 weeks ago
parent f09ea0f871 1c6068af99
commit 8ec0bb3359
No known key found for this signature in database
GPG Key ID: 783F096F253D15B0
59 changed files with 4767 additions and 5477 deletions
Show Stats Download Patch File Download Diff File

44
yt_dlp/extractor/adobetv.py
Unescape Escape View File

@ -84,9 +84,10 @@ class AdobeTVBaseIE(InfoExtractor):
class AdobeTVEmbedIE(AdobeTVBaseIE): class AdobeTVEmbedIE(AdobeTVBaseIE):
_WORKING = False
IE_NAME = 'adobetv:embed' IE_NAME = 'adobetv:embed'
_VALID_URL = r'https?://tv\.adobe\.com/embed/\d+/(?P<id>\d+)' _VALID_URL = r'https?://tv\.adobe\.com/embed/\d+/(?P<id>\d+)'
_TEST = { _TESTS = [{
'url': 'https://tv.adobe.com/embed/22/4153', 'url': 'https://tv.adobe.com/embed/22/4153',
'md5': 'c8c0461bf04d54574fc2b4d07ac6783a', 'md5': 'c8c0461bf04d54574fc2b4d07ac6783a',
'info_dict': { 'info_dict': {
@ -94,12 +95,12 @@ class AdobeTVEmbedIE(AdobeTVBaseIE):
'ext': 'flv', 'ext': 'flv',
'title': 'Creating Graphics Optimized for BlackBerry', 'title': 'Creating Graphics Optimized for BlackBerry',
'description': 'md5:eac6e8dced38bdaae51cd94447927459', 'description': 'md5:eac6e8dced38bdaae51cd94447927459',
'thumbnail': r're:https?://.*\.jpg$', 'thumbnail': r're:https?://.+\.jpg',
'upload_date': '20091109', 'upload_date': '20091109',
'duration': 377, 'duration': 377,
'view_count': int, 'view_count': int,
}, },
} }]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)
@ -110,10 +111,10 @@ class AdobeTVEmbedIE(AdobeTVBaseIE):
class AdobeTVIE(AdobeTVBaseIE): class AdobeTVIE(AdobeTVBaseIE):
_WORKING = False
IE_NAME = 'adobetv' IE_NAME = 'adobetv'
_VALID_URL = r'https?://tv\.adobe\.com/(?:(?P<language>fr|de|es|jp)/)?watch/(?P<show_urlname>[^/]+)/(?P<id>[^/]+)' _VALID_URL = r'https?://tv\.adobe\.com/(?:(?P<language>fr|de|es|jp)/)?watch/(?P<show_urlname>[^/]+)/(?P<id>[^/]+)'
_TESTS = [{
_TEST = {
'url': 'http://tv.adobe.com/watch/the-complete-picture-with-julieanne-kost/quick-tip-how-to-draw-a-circle-around-an-object-in-photoshop/', 'url': 'http://tv.adobe.com/watch/the-complete-picture-with-julieanne-kost/quick-tip-how-to-draw-a-circle-around-an-object-in-photoshop/',
'md5': '9bc5727bcdd55251f35ad311ca74fa1e', 'md5': '9bc5727bcdd55251f35ad311ca74fa1e',
'info_dict': { 'info_dict': {
@ -121,12 +122,12 @@ class AdobeTVIE(AdobeTVBaseIE):
'ext': 'mp4', 'ext': 'mp4',
'title': 'Quick Tip - How to Draw a Circle Around an Object in Photoshop', 'title': 'Quick Tip - How to Draw a Circle Around an Object in Photoshop',
'description': 'md5:99ec318dc909d7ba2a1f2b038f7d2311', 'description': 'md5:99ec318dc909d7ba2a1f2b038f7d2311',
'thumbnail': r're:https?://.*\.jpg$', 'thumbnail': r're:https?://.+\.jpg',
'upload_date': '20110914', 'upload_date': '20110914',
'duration': 60, 'duration': 60,
'view_count': int, 'view_count': int,
}, },
} }]
def _real_extract(self, url): def _real_extract(self, url):
language, show_urlname, urlname = self._match_valid_url(url).groups() language, show_urlname, urlname = self._match_valid_url(url).groups()
@ -159,10 +160,10 @@ class AdobeTVPlaylistBaseIE(AdobeTVBaseIE):
class AdobeTVShowIE(AdobeTVPlaylistBaseIE): class AdobeTVShowIE(AdobeTVPlaylistBaseIE):
_WORKING = False
IE_NAME = 'adobetv:show' IE_NAME = 'adobetv:show'
_VALID_URL = r'https?://tv\.adobe\.com/(?:(?P<language>fr|de|es|jp)/)?show/(?P<id>[^/]+)' _VALID_URL = r'https?://tv\.adobe\.com/(?:(?P<language>fr|de|es|jp)/)?show/(?P<id>[^/]+)'
_TESTS = [{
_TEST = {
'url': 'http://tv.adobe.com/show/the-complete-picture-with-julieanne-kost', 'url': 'http://tv.adobe.com/show/the-complete-picture-with-julieanne-kost',
'info_dict': { 'info_dict': {
'id': '36', 'id': '36',
@ -170,7 +171,7 @@ class AdobeTVShowIE(AdobeTVPlaylistBaseIE):
'description': 'md5:fa50867102dcd1aa0ddf2ab039311b27', 'description': 'md5:fa50867102dcd1aa0ddf2ab039311b27',
}, },
'playlist_mincount': 136, 'playlist_mincount': 136,
} }]
_RESOURCE = 'episode' _RESOURCE = 'episode'
_process_data = AdobeTVBaseIE._parse_video_data _process_data = AdobeTVBaseIE._parse_video_data
@ -195,16 +196,16 @@ class AdobeTVShowIE(AdobeTVPlaylistBaseIE):
class AdobeTVChannelIE(AdobeTVPlaylistBaseIE): class AdobeTVChannelIE(AdobeTVPlaylistBaseIE):
_WORKING = False
IE_NAME = 'adobetv:channel' IE_NAME = 'adobetv:channel'
_VALID_URL = r'https?://tv\.adobe\.com/(?:(?P<language>fr|de|es|jp)/)?channel/(?P<id>[^/]+)(?:/(?P<category_urlname>[^/]+))?' _VALID_URL = r'https?://tv\.adobe\.com/(?:(?P<language>fr|de|es|jp)/)?channel/(?P<id>[^/]+)(?:/(?P<category_urlname>[^/]+))?'
_TESTS = [{
_TEST = {
'url': 'http://tv.adobe.com/channel/development', 'url': 'http://tv.adobe.com/channel/development',
'info_dict': { 'info_dict': {
'id': 'development', 'id': 'development',
}, },
'playlist_mincount': 96, 'playlist_mincount': 96,
} }]
_RESOURCE = 'show' _RESOURCE = 'show'
def _process_data(self, show_data): def _process_data(self, show_data):
@ -231,8 +232,7 @@ class AdobeTVVideoIE(AdobeTVBaseIE):
IE_NAME = 'adobetv:video' IE_NAME = 'adobetv:video'
_VALID_URL = r'https?://video\.tv\.adobe\.com/v/(?P<id>\d+)' _VALID_URL = r'https?://video\.tv\.adobe\.com/v/(?P<id>\d+)'
_EMBED_REGEX = [r'<iframe[^>]+src=[\'"](?P<url>(?:https?:)?//video\.tv\.adobe\.com/v/\d+[^"]+)[\'"]'] _EMBED_REGEX = [r'<iframe[^>]+src=[\'"](?P<url>(?:https?:)?//video\.tv\.adobe\.com/v/\d+[^"]+)[\'"]']
_TESTS = [{
_TEST = {
# From https://helpx.adobe.com/acrobat/how-to/new-experience-acrobat-dc.html?set=acrobat--get-started--essential-beginners # From https://helpx.adobe.com/acrobat/how-to/new-experience-acrobat-dc.html?set=acrobat--get-started--essential-beginners
'url': 'https://video.tv.adobe.com/v/2456/', 'url': 'https://video.tv.adobe.com/v/2456/',
'md5': '43662b577c018ad707a63766462b1e87', 'md5': '43662b577c018ad707a63766462b1e87',
@ -242,8 +242,20 @@ class AdobeTVVideoIE(AdobeTVBaseIE):
'title': 'New experience with Acrobat DC', 'title': 'New experience with Acrobat DC',
'description': 'New experience with Acrobat DC', 'description': 'New experience with Acrobat DC',
'duration': 248.667, 'duration': 248.667,
'thumbnail': r're:https?://images-tv\.adobe\.com/.+\.jpg',
},
}]
_WEBPAGE_TESTS = [{
# FIXME: Invalid extension
'url': 'https://www.adobe.com/learn/acrobat/web/customize-toolbar',
'info_dict': {
'id': '3463980',
'ext': 'm3u8',
'title': 'Adobe Acrobat: How to Customize the Toolbar for Faster PDF Editing',
'description': 'md5:94368ab95ae24f9c1bee0cb346e03dc3',
'duration': 97.557,
}, },
} }]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)

12
yt_dlp/extractor/apa.py
Unescape Escape View File

@ -11,12 +11,11 @@ class APAIE(InfoExtractor):
_EMBED_REGEX = [r'<iframe[^>]+\bsrc=(["\'])(?P<url>(?:https?:)?//[^/]+\.apa\.at/embed/[\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12}.*?)\1'] _EMBED_REGEX = [r'<iframe[^>]+\bsrc=(["\'])(?P<url>(?:https?:)?//[^/]+\.apa\.at/embed/[\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12}.*?)\1']
_TESTS = [{ _TESTS = [{
'url': 'http://uvp.apa.at/embed/293f6d17-692a-44e3-9fd5-7b178f3a1029', 'url': 'http://uvp.apa.at/embed/293f6d17-692a-44e3-9fd5-7b178f3a1029',
'md5': '2b12292faeb0a7d930c778c7a5b4759b',
'info_dict': { 'info_dict': {
'id': '293f6d17-692a-44e3-9fd5-7b178f3a1029', 'id': '293f6d17-692a-44e3-9fd5-7b178f3a1029',
'ext': 'mp4', 'ext': 'mp4',
'title': '293f6d17-692a-44e3-9fd5-7b178f3a1029', 'title': '293f6d17-692a-44e3-9fd5-7b178f3a1029',
'thumbnail': r're:^https?://.*\.jpg$', 'thumbnail': r're:https?://kf-vn\.sf\.apa\.at/vn/.+\.jpg',
}, },
}, { }, {
'url': 'https://uvp-apapublisher.sf.apa.at/embed/2f94e9e6-d945-4db2-9548-f9a41ebf7b78', 'url': 'https://uvp-apapublisher.sf.apa.at/embed/2f94e9e6-d945-4db2-9548-f9a41ebf7b78',
@ -28,6 +27,15 @@ class APAIE(InfoExtractor):
'url': 'http://uvp-kleinezeitung.sf.apa.at/embed/f1c44979-dba2-4ebf-b021-e4cf2cac3c81', 'url': 'http://uvp-kleinezeitung.sf.apa.at/embed/f1c44979-dba2-4ebf-b021-e4cf2cac3c81',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
'url': 'https://www.vol.at/blue-man-group/5593454',
'info_dict': {
'id': '293f6d17-692a-44e3-9fd5-7b178f3a1029',
'ext': 'mp4',
'title': '293f6d17-692a-44e3-9fd5-7b178f3a1029',
'thumbnail': r're:https?://kf-vn\.sf\.apa\.at/vn/.+\.jpg',
},
}]
def _real_extract(self, url): def _real_extract(self, url):
mobj = self._match_valid_url(url) mobj = self._match_valid_url(url)

14
yt_dlp/extractor/arcpublishing.py
Unescape Escape View File

@ -62,6 +62,20 @@ class ArcPublishingIE(InfoExtractor):
'url': 'arcpublishing:tronc:460f2931-8130-4719-8ea1-ffcb2d7cb685', 'url': 'arcpublishing:tronc:460f2931-8130-4719-8ea1-ffcb2d7cb685',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
'url': 'https://www.uppermichiganssource.com/2025/07/18/scattered-showers-storms-bring-heavy-rain-potential/',
'info_dict': {
'id': '508116f7-e999-48db-b7c2-60a04842679b',
'ext': 'mp4',
'title': 'Scattered showers & storms bring heavy rain potential',
'description': 'Scattered showers & storms bring heavy rain potential',
'duration': 2016,
'thumbnail': r're:https?://.+\.jpg',
'timestamp': 1752881287,
'upload_date': '20250718',
},
'expected_warnings': ['Ignoring subtitle tracks found in the HLS manifest'],
}]
_POWA_DEFAULTS = [ _POWA_DEFAULTS = [
(['cmg', 'prisa'], '%s-config-prod.api.cdn.arcpublishing.com/video'), (['cmg', 'prisa'], '%s-config-prod.api.cdn.arcpublishing.com/video'),
([ ([

33
yt_dlp/extractor/arte.py
Unescape Escape View File

@ -51,8 +51,8 @@ class ArteTVIE(ArteTVBaseIE):
'id': '109067-000-A', 'id': '109067-000-A',
'ext': 'mp4', 'ext': 'mp4',
'description': 'md5:d2ca367b8ecee028dddaa8bd1aebc739', 'description': 'md5:d2ca367b8ecee028dddaa8bd1aebc739',
'thumbnail': r're:https?://api-cdn\.arte\.tv/img/v2/image/.+',
'timestamp': 1713927600, 'timestamp': 1713927600,
'thumbnail': 'https://api-cdn.arte.tv/img/v2/image/3rR6PLzfbigSkkeHtkCZNF/940x530',
'duration': 7599, 'duration': 7599,
'title': 'La loi de Téhéran', 'title': 'La loi de Téhéran',
'upload_date': '20240424', 'upload_date': '20240424',
@ -62,6 +62,7 @@ class ArteTVIE(ArteTVBaseIE):
'fr-forced': 'mincount:1', 'fr-forced': 'mincount:1',
}, },
}, },
'skip': 'Invalid URL',
}, { }, {
'note': 'age-restricted', 'note': 'age-restricted',
'url': 'https://www.arte.tv/de/videos/006785-000-A/the-element-of-crime/', 'url': 'https://www.arte.tv/de/videos/006785-000-A/the-element-of-crime/',
@ -69,9 +70,9 @@ class ArteTVIE(ArteTVBaseIE):
'id': '006785-000-A', 'id': '006785-000-A',
'description': 'md5:c2f94fdfefc8a280e4dab68ab96ab0ba', 'description': 'md5:c2f94fdfefc8a280e4dab68ab96ab0ba',
'title': 'The Element of Crime', 'title': 'The Element of Crime',
'thumbnail': r're:https?://api-cdn\.arte\.tv/img/v2/image/.+',
'timestamp': 1696111200, 'timestamp': 1696111200,
'duration': 5849, 'duration': 5849,
'thumbnail': 'https://api-cdn.arte.tv/img/v2/image/q82dTTfyuCXupPsGxXsd7B/940x530',
'upload_date': '20230930', 'upload_date': '20230930',
'ext': 'mp4', 'ext': 'mp4',
}, },
@ -252,6 +253,30 @@ class ArteTVEmbedIE(InfoExtractor):
'url': 'https://www.arte.tv/player/v3/index.php?json_url=https://api.arte.tv/api/player/v2/config/de/100605-013-A', 'url': 'https://www.arte.tv/player/v3/index.php?json_url=https://api.arte.tv/api/player/v2/config/de/100605-013-A',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
# FIXME: Embed detection
'url': 'https://timesofmalta.com/article/watch-sunken-warships-north-sea-arte.1108358',
'info_dict': {
'id': '110288-000-A',
'ext': 'mp4',
'title': 'Danger on the Seabed',
'alt_title': 'Sunken Warships in the North Sea',
'description': 'md5:a2c84cbad37d280bddb6484087120add',
'duration': 3148,
'thumbnail': r're:https?://api-cdn\.arte\.tv/img/v2/image/.+',
'timestamp': 1741686820,
'upload_date': '20250311',
},
'params': {'skip_download': 'm3u8'},
}, {
# FIXME: Embed detection
'url': 'https://www.eurockeennes.fr/en-live/',
'info_dict': {
'id': 'en-live',
'title': 'Les Eurocks en live | Les Eurockéennes de Belfort – 3-4-5-6 juillet 2025 sur la Presqu&#039;Île du Malsaucy',
},
'playlist_count': 4,
}]
def _real_extract(self, url): def _real_extract(self, url):
qs = parse_qs(url) qs = parse_qs(url)
@ -304,9 +329,9 @@ class ArteTVCategoryIE(ArteTVBaseIE):
'info_dict': { 'info_dict': {
'id': 'politics-and-society', 'id': 'politics-and-society',
'title': 'Politics and society', 'title': 'Politics and society',
'description': 'Investigative documentary series, geopolitical analysis, and international commentary', 'description': 'Watch documentaries and reportage about politics, society and current affairs.',
}, },
'playlist_mincount': 13, 'playlist_mincount': 3,
}] }]
@classmethod @classmethod

45
yt_dlp/extractor/bandcamp.py
Unescape Escape View File

@ -36,14 +36,12 @@ class BandcampIE(InfoExtractor):
'duration': 9.8485, 'duration': 9.8485,
'uploader': 'youtube-dl "\'/\\ä↭', 'uploader': 'youtube-dl "\'/\\ä↭',
'upload_date': '20121129', 'upload_date': '20121129',
'thumbnail': r're:https?://f4\.bcbits\.com/img/.+\.jpg',
'timestamp': 1354224127, 'timestamp': 1354224127,
'track': 'youtube-dl "\'/\\ä↭ - youtube-dl test song "\'/\\ä↭', 'track': 'youtube-dl "\'/\\ä↭ - youtube-dl test song "\'/\\ä↭',
'album_artist': 'youtube-dl "\'/\\ä↭',
'track_id': '1812978515', 'track_id': '1812978515',
'artist': 'youtube-dl "\'/\\ä↭',
'uploader_url': 'https://youtube-dl.bandcamp.com', 'uploader_url': 'https://youtube-dl.bandcamp.com',
'uploader_id': 'youtube-dl', 'uploader_id': 'youtube-dl',
'thumbnail': 'https://f4.bcbits.com/img/a3216802731_5.jpg',
'artists': ['youtube-dl "\'/\\ä↭'], 'artists': ['youtube-dl "\'/\\ä↭'],
'album_artists': ['youtube-dl "\'/\\ä↭'], 'album_artists': ['youtube-dl "\'/\\ä↭'],
}, },
@ -54,10 +52,9 @@ class BandcampIE(InfoExtractor):
'info_dict': { 'info_dict': {
'id': '2650410135', 'id': '2650410135',
'ext': 'm4a', 'ext': 'm4a',
'acodec': r're:[fa]lac',
'title': 'Ben Prunty - Lanius (Battle)', 'title': 'Ben Prunty - Lanius (Battle)',
'thumbnail': r're:^https?://.*\.jpg$',
'uploader': 'Ben Prunty', 'uploader': 'Ben Prunty',
'thumbnail': r're:https?://f4\.bcbits\.com/img/.+\.jpg',
'timestamp': 1396508491, 'timestamp': 1396508491,
'upload_date': '20140403', 'upload_date': '20140403',
'release_timestamp': 1396483200, 'release_timestamp': 1396483200,
@ -66,8 +63,6 @@ class BandcampIE(InfoExtractor):
'track': 'Lanius (Battle)', 'track': 'Lanius (Battle)',
'track_number': 1, 'track_number': 1,
'track_id': '2650410135', 'track_id': '2650410135',
'artist': 'Ben Prunty',
'album_artist': 'Ben Prunty',
'album': 'FTL: Advanced Edition Soundtrack', 'album': 'FTL: Advanced Edition Soundtrack',
'uploader_url': 'https://benprunty.bandcamp.com', 'uploader_url': 'https://benprunty.bandcamp.com',
'uploader_id': 'benprunty', 'uploader_id': 'benprunty',
@ -83,8 +78,8 @@ class BandcampIE(InfoExtractor):
'id': '2584466013', 'id': '2584466013',
'ext': 'mp3', 'ext': 'mp3',
'title': 'Mastodon - Hail to Fire', 'title': 'Mastodon - Hail to Fire',
'thumbnail': r're:^https?://.*\.jpg$',
'uploader': 'Mastodon', 'uploader': 'Mastodon',
'thumbnail': r're:https?://f4\.bcbits\.com/img/.+\.jpg',
'timestamp': 1322005399, 'timestamp': 1322005399,
'upload_date': '20111122', 'upload_date': '20111122',
'release_timestamp': 1076112000, 'release_timestamp': 1076112000,
@ -93,8 +88,6 @@ class BandcampIE(InfoExtractor):
'track': 'Hail to Fire', 'track': 'Hail to Fire',
'track_number': 5, 'track_number': 5,
'track_id': '2584466013', 'track_id': '2584466013',
'artist': 'Mastodon',
'album_artist': 'Mastodon',
'album': 'Call of the Mastodon', 'album': 'Call of the Mastodon',
'uploader_url': 'https://relapsealumni.bandcamp.com', 'uploader_url': 'https://relapsealumni.bandcamp.com',
'uploader_id': 'relapsealumni', 'uploader_id': 'relapsealumni',
@ -110,8 +103,8 @@ class BandcampIE(InfoExtractor):
'id': '1978174799', 'id': '1978174799',
'ext': 'mp3', 'ext': 'mp3',
'title': 'submerse - submerse - Safehouse', 'title': 'submerse - submerse - Safehouse',
'thumbnail': r're:^https?://.*\.jpg$',
'uploader': 'submerse', 'uploader': 'submerse',
'thumbnail': r're:https?://f4\.bcbits\.com/img/.+\.jpg',
'timestamp': 1480779297, 'timestamp': 1480779297,
'upload_date': '20161203', 'upload_date': '20161203',
'release_timestamp': 1481068800, 'release_timestamp': 1481068800,
@ -120,8 +113,6 @@ class BandcampIE(InfoExtractor):
'track': 'submerse - Safehouse', 'track': 'submerse - Safehouse',
'track_number': 3, 'track_number': 3,
'track_id': '1978174799', 'track_id': '1978174799',
'artist': 'submerse',
'album_artist': 'Diskotopia',
'album': 'DSK F/W 2016-2017 Free Compilation', 'album': 'DSK F/W 2016-2017 Free Compilation',
'uploader_url': 'https://diskotopia.bandcamp.com', 'uploader_url': 'https://diskotopia.bandcamp.com',
'uploader_id': 'diskotopia', 'uploader_id': 'diskotopia',
@ -130,6 +121,30 @@ class BandcampIE(InfoExtractor):
'album_artists': ['Diskotopia'], 'album_artists': ['Diskotopia'],
}, },
}] }]
_WEBPAGE_TESTS = [{
# FIXME: Embed detection
'url': 'https://www.punknews.org/article/85809/stay-inside-super-sonic',
'info_dict': {
'id': '2475540375',
'ext': 'mp3',
'title': 'Stay Inside - Super Sonic',
'album': 'Lunger',
'album_artists': ['Stay Inside'],
'artists': ['Stay Inside'],
'duration': 166.157,
'release_date': '20251003',
'release_timestamp': 1759449600.0,
'thumbnail': r're:https?://f4\.bcbits\.com/img/.+\.jpg',
'timestamp': 1749473029.0,
'track': 'Super Sonic',
'track_id': '2475540375',
'track_number': 3,
'upload_date': '20250609',
'uploader': 'Stay Inside',
'uploader_id': 'stayinside',
'uploader_url': 'https://stayinside.bandcamp.com',
},
}]
def _extract_data_attr(self, webpage, video_id, attr='tralbum', fatal=True): def _extract_data_attr(self, webpage, video_id, attr='tralbum', fatal=True):
return self._parse_json(self._html_search_regex( return self._parse_json(self._html_search_regex(
@ -279,10 +294,10 @@ class BandcampAlbumIE(BandcampIE): # XXX: Do not subclass from concrete IE
'id': '1353101989', 'id': '1353101989',
'ext': 'mp3', 'ext': 'mp3',
'title': 'Blazo - Intro', 'title': 'Blazo - Intro',
'thumbnail': r're:https?://f4\.bcbits\.com/img/.+\.jpg',
'timestamp': 1311756226, 'timestamp': 1311756226,
'upload_date': '20110727', 'upload_date': '20110727',
'uploader': 'Blazo', 'uploader': 'Blazo',
'thumbnail': 'https://f4.bcbits.com/img/a1721150828_5.jpg',
'album_artists': ['Blazo'], 'album_artists': ['Blazo'],
'uploader_url': 'https://blazo.bandcamp.com', 'uploader_url': 'https://blazo.bandcamp.com',
'release_date': '20110727', 'release_date': '20110727',
@ -302,6 +317,7 @@ class BandcampAlbumIE(BandcampIE): # XXX: Do not subclass from concrete IE
'id': '38097443', 'id': '38097443',
'ext': 'mp3', 'ext': 'mp3',
'title': 'Blazo - Kero One - Keep It Alive (Blazo remix)', 'title': 'Blazo - Kero One - Keep It Alive (Blazo remix)',
'thumbnail': r're:https?://f4\.bcbits\.com/img/.+\.jpg',
'timestamp': 1311757238, 'timestamp': 1311757238,
'upload_date': '20110727', 'upload_date': '20110727',
'uploader': 'Blazo', 'uploader': 'Blazo',
@ -315,7 +331,6 @@ class BandcampAlbumIE(BandcampIE): # XXX: Do not subclass from concrete IE
'uploader_id': 'blazo', 'uploader_id': 'blazo',
'album_artists': ['Blazo'], 'album_artists': ['Blazo'],
'artists': ['Blazo'], 'artists': ['Blazo'],
'thumbnail': 'https://f4.bcbits.com/img/a1721150828_5.jpg',
'release_timestamp': 1311724800.0, 'release_timestamp': 1311724800.0,
}, },
}, },

13
yt_dlp/extractor/blogger.py
Unescape Escape View File

@ -19,8 +19,19 @@ class BloggerIE(InfoExtractor):
'id': 'BLOGGER-video-3c740e3a49197e16-796', 'id': 'BLOGGER-video-3c740e3a49197e16-796',
'title': 'BLOGGER-video-3c740e3a49197e16-796', 'title': 'BLOGGER-video-3c740e3a49197e16-796',
'ext': 'mp4', 'ext': 'mp4',
'thumbnail': r're:^https?://.*',
'duration': 76.068, 'duration': 76.068,
'thumbnail': r're:https?://i9\.ytimg\.com/vi_blogger/.+',
},
}]
_WEBPAGE_TESTS = [{
'url': 'https://blog.tomeuvizoso.net/2019/01/a-panfrost-milestone.html',
'md5': 'f1bc19b6ea1b0fd1d81e84ca9ec467ac',
'info_dict': {
'id': 'BLOGGER-video-3c740e3a49197e16-12203',
'ext': 'mp4',
'title': 'BLOGGER-video-3c740e3a49197e16-12203',
'duration': 76.068,
'thumbnail': r're:https?://i9\.ytimg\.com/vi_blogger/.+',
}, },
}] }]

20
yt_dlp/extractor/cloudflarestream.py
Unescape Escape View File

@ -19,18 +19,16 @@ class CloudflareStreamIE(InfoExtractor):
'id': '31c9291ab41fac05471db4e73aa11717', 'id': '31c9291ab41fac05471db4e73aa11717',
'ext': 'mp4', 'ext': 'mp4',
'title': '31c9291ab41fac05471db4e73aa11717', 'title': '31c9291ab41fac05471db4e73aa11717',
'thumbnail': 'https://cloudflarestream.com/31c9291ab41fac05471db4e73aa11717/thumbnails/thumbnail.jpg', 'thumbnail': r're:https?://cloudflarestream\.com/.+\.jpg',
},
'params': {
'skip_download': 'm3u8',
}, },
'params': {'skip_download': 'm3u8'},
}, { }, {
'url': 'https://watch.cloudflarestream.com/embed/sdk-iframe-integration.fla9.latest.js?video=0e8e040aec776862e1d632a699edf59e', 'url': 'https://watch.cloudflarestream.com/embed/sdk-iframe-integration.fla9.latest.js?video=0e8e040aec776862e1d632a699edf59e',
'info_dict': { 'info_dict': {
'id': '0e8e040aec776862e1d632a699edf59e', 'id': '0e8e040aec776862e1d632a699edf59e',
'ext': 'mp4', 'ext': 'mp4',
'title': '0e8e040aec776862e1d632a699edf59e', 'title': '0e8e040aec776862e1d632a699edf59e',
'thumbnail': 'https://cloudflarestream.com/0e8e040aec776862e1d632a699edf59e/thumbnails/thumbnail.jpg', 'thumbnail': r're:https?://cloudflarestream\.com/.+\.jpg',
}, },
}, { }, {
'url': 'https://watch.cloudflarestream.com/9df17203414fd1db3e3ed74abbe936c1', 'url': 'https://watch.cloudflarestream.com/9df17203414fd1db3e3ed74abbe936c1',
@ -54,11 +52,21 @@ class CloudflareStreamIE(InfoExtractor):
'id': 'eaef9dea5159cf968be84241b5cedfe7', 'id': 'eaef9dea5159cf968be84241b5cedfe7',
'ext': 'mp4', 'ext': 'mp4',
'title': 'eaef9dea5159cf968be84241b5cedfe7', 'title': 'eaef9dea5159cf968be84241b5cedfe7',
'thumbnail': 'https://cloudflarestream.com/eaef9dea5159cf968be84241b5cedfe7/thumbnails/thumbnail.jpg', 'thumbnail': r're:https?://cloudflarestream\.com/.+\.jpg',
}, },
'params': { 'params': {
'extractor_args': {'generic': {'impersonate': ['chrome']}},
'skip_download': 'm3u8', 'skip_download': 'm3u8',
}, },
}, {
# FIXME: Embed detection
'url': 'https://www.cloudflare.com/developer-platform/products/cloudflare-stream/',
'info_dict': {
'id': 'e7bd2dd67e0f8860b4ae81e33a966049',
'ext': 'mp4',
'title': 'e7bd2dd67e0f8860b4ae81e33a966049',
'thumbnail': r're:https?://cloudflarestream\.com/.+\.jpg',
},
}] }]
def _real_extract(self, url): def _real_extract(self, url):

18
yt_dlp/extractor/condenast.py
Unescape Escape View File

@ -96,6 +96,24 @@ class CondeNastIE(InfoExtractor):
'upload_date': '20150916', 'upload_date': '20150916',
'timestamp': 1442434920, 'timestamp': 1442434920,
}, },
}, {
# FIXME: Subtitles
'url': 'https://www.vanityfair.com/video/watch/vf-quiz-show-squid-game-s3',
'info_dict': {
'id': '6862f999c1afbc5ff06b4803',
'ext': 'mp4',
'title': '\'Squid Game\' Cast Tests How Well They Know Each Other',
'categories': ['Arts & Culture', 'Hollywood'],
'description': 'md5:7a9c668a1fc87648e77da13842ec1534',
'duration': 955,
'season': 'Season 1',
'series': 'Quizzing Each Other',
'tags': 'count:2',
'thumbnail': r're:https?://dwgyu36up6iuz\.cloudfront\.net/.+\.jpg',
'timestamp': 1751341306,
'upload_date': '20250701',
'uploader': 'vanityfair',
},
}, { }, {
'url': 'https://player.cnevids.com/inline/video/59138decb57ac36b83000005.js?target=js-cne-player', 'url': 'https://player.cnevids.com/inline/video/59138decb57ac36b83000005.js?target=js-cne-player',
'only_matching': True, 'only_matching': True,

17
yt_dlp/extractor/crooksandliars.py
Unescape Escape View File

@ -8,7 +8,6 @@ from ..utils import (
class CrooksAndLiarsIE(InfoExtractor): class CrooksAndLiarsIE(InfoExtractor):
_VALID_URL = r'https?://embed\.crooksandliars\.com/(?:embed|v)/(?P<id>[A-Za-z0-9]+)' _VALID_URL = r'https?://embed\.crooksandliars\.com/(?:embed|v)/(?P<id>[A-Za-z0-9]+)'
_EMBED_REGEX = [r'<(?:iframe[^>]+src|param[^>]+value)=(["\'])(?P<url>(?:https?:)?//embed\.crooksandliars\.com/(?:embed|v)/.+?)\1'] _EMBED_REGEX = [r'<(?:iframe[^>]+src|param[^>]+value)=(["\'])(?P<url>(?:https?:)?//embed\.crooksandliars\.com/(?:embed|v)/.+?)\1']
_TESTS = [{ _TESTS = [{
'url': 'https://embed.crooksandliars.com/embed/8RUoRhRi', 'url': 'https://embed.crooksandliars.com/embed/8RUoRhRi',
'info_dict': { 'info_dict': {
@ -16,7 +15,7 @@ class CrooksAndLiarsIE(InfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'title': 'Fox & Friends Says Protecting Atheists From Discrimination Is Anti-Christian!', 'title': 'Fox & Friends Says Protecting Atheists From Discrimination Is Anti-Christian!',
'description': 'md5:e1a46ad1650e3a5ec7196d432799127f', 'description': 'md5:e1a46ad1650e3a5ec7196d432799127f',
'thumbnail': r're:^https?://.*\.jpg', 'thumbnail': r're:https?://crooksandliars\.com/files/.+',
'timestamp': 1428207000, 'timestamp': 1428207000,
'upload_date': '20150405', 'upload_date': '20150405',
'uploader': 'Heather', 'uploader': 'Heather',
@ -26,6 +25,20 @@ class CrooksAndLiarsIE(InfoExtractor):
'url': 'http://embed.crooksandliars.com/v/MTE3MjUtMzQ2MzA', 'url': 'http://embed.crooksandliars.com/v/MTE3MjUtMzQ2MzA',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
'url': 'https://crooksandliars.com/2015/04/fox-friends-says-protecting-atheists',
'info_dict': {
'id': '8RUoRhRi',
'ext': 'mp4',
'title': 'Fox & Friends Says Protecting Atheists From Discrimination Is Anti-Christian!',
'description': 'md5:e1a46ad1650e3a5ec7196d432799127f',
'duration': 236,
'thumbnail': r're:https?://crooksandliars\.com/files/.+',
'timestamp': 1428207000,
'upload_date': '20150405',
'uploader': 'Heather',
},
}]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)

11
yt_dlp/extractor/dailymail.py
Unescape Escape View File

@ -19,11 +19,22 @@ class DailyMailIE(InfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'title': 'The Mountain appears in sparkling water ad for \'Heavy Bubbles\'', 'title': 'The Mountain appears in sparkling water ad for \'Heavy Bubbles\'',
'description': 'md5:a93d74b6da172dd5dc4d973e0b766a84', 'description': 'md5:a93d74b6da172dd5dc4d973e0b766a84',
'thumbnail': r're:https?://i\.dailymail\.co\.uk/.+\.jpg',
}, },
}, { }, {
'url': 'http://www.dailymail.co.uk/embed/video/1295863.html', 'url': 'http://www.dailymail.co.uk/embed/video/1295863.html',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
'url': 'https://www.daily-news.gr/lifestyle/%ce%b7-%cf%84%cf%81%ce%b1%ce%b3%ce%bf%cf%85%ce%b4%ce%af%cf%83%cf%84%cf%81%ce%b9%ce%b1-jessie-j-%ce%bc%ce%bf%ce%b9%cf%81%ce%ac%cf%83%cf%84%ce%b7%ce%ba%ce%b5-%cf%83%cf%85%ce%b3%ce%ba%ce%bb%ce%bf%ce%bd/',
'info_dict': {
'id': '3463585',
'ext': 'mp4',
'title': 'Jessie J reveals she has undergone surgery as she shares clips',
'description': 'md5:9fa9a25feca5b656b0b4a39c922fad1e',
'thumbnail': r're:https?://i\.dailymail\.co\.uk/.+\.jpg',
},
}]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)

55
yt_dlp/extractor/dailymotion.py
Unescape Escape View File

@ -119,13 +119,14 @@ class DailymotionIE(DailymotionBaseInfoExtractor):
_EMBED_REGEX = [rf'(?ix)<(?:(?:embed|iframe)[^>]+?src=|input[^>]+id=[\'"]dmcloudUrlEmissionSelect[\'"][^>]+value=)["\'](?P<url>{_VALID_URL[5:]})'] _EMBED_REGEX = [rf'(?ix)<(?:(?:embed|iframe)[^>]+?src=|input[^>]+id=[\'"]dmcloudUrlEmissionSelect[\'"][^>]+value=)["\'](?P<url>{_VALID_URL[5:]})']
_TESTS = [{ _TESTS = [{
'url': 'http://www.dailymotion.com/video/x5kesuj_office-christmas-party-review-jason-bateman-olivia-munn-t-j-miller_news', 'url': 'http://www.dailymotion.com/video/x5kesuj_office-christmas-party-review-jason-bateman-olivia-munn-t-j-miller_news',
'md5': '074b95bdee76b9e3654137aee9c79dfe',
'info_dict': { 'info_dict': {
'id': 'x5kesuj', 'id': 'x5kesuj',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Office Christmas Party Review – Jason Bateman, Olivia Munn, T.J. Miller', 'title': 'Office Christmas Party Review – Jason Bateman, Olivia Munn, T.J. Miller',
'description': 'Office Christmas Party Review - Jason Bateman, Olivia Munn, T.J. Miller', 'description': 'Office Christmas Party Review - Jason Bateman, Olivia Munn, T.J. Miller',
'duration': 187, 'duration': 187,
'tags': 'count:5',
'thumbnail': r're:https?://s[12]\.dmcdn\.net/v/.+',
'timestamp': 1493651285, 'timestamp': 1493651285,
'upload_date': '20170501', 'upload_date': '20170501',
'uploader': 'Deadline', 'uploader': 'Deadline',
@ -133,18 +134,17 @@ class DailymotionIE(DailymotionBaseInfoExtractor):
'age_limit': 0, 'age_limit': 0,
'view_count': int, 'view_count': int,
'like_count': int, 'like_count': int,
'tags': ['hollywood', 'celeb', 'celebrity', 'movies', 'red carpet'],
'thumbnail': r're:https://(?:s[12]\.)dmcdn\.net/v/K456B1cmt4ZcZ9KiM/x1080',
}, },
}, { }, {
'url': 'https://geo.dailymotion.com/player.html?video=x89eyek&mute=true', 'url': 'https://geo.dailymotion.com/player.html?video=x89eyek&mute=true',
'md5': 'e2f9717c6604773f963f069ca53a07f8',
'info_dict': { 'info_dict': {
'id': 'x89eyek', 'id': 'x89eyek',
'ext': 'mp4', 'ext': 'mp4',
'title': "En quête d'esprit du 27/03/2022", 'title': 'En quête d\'esprit du 27/03/2022',
'description': 'md5:66542b9f4df2eb23f314fc097488e553', 'description': 'md5:66542b9f4df2eb23f314fc097488e553',
'duration': 2756, 'duration': 2756,
'tags': 'count:1',
'thumbnail': r're:https?://s[12]\.dmcdn\.net/v/.+',
'timestamp': 1648383669, 'timestamp': 1648383669,
'upload_date': '20220327', 'upload_date': '20220327',
'uploader': 'CNEWS', 'uploader': 'CNEWS',
@ -152,8 +152,6 @@ class DailymotionIE(DailymotionBaseInfoExtractor):
'age_limit': 0, 'age_limit': 0,
'view_count': int, 'view_count': int,
'like_count': int, 'like_count': int,
'tags': ['en_quete_d_esprit'],
'thumbnail': r're:https://(?:s[12]\.)dmcdn\.net/v/Tncwi1clTH6StrxMP/x1080',
}, },
}, { }, {
'url': 'https://www.dailymotion.com/video/x2iuewm_steam-machine-models-pricing-listed-on-steam-store-ign-news_videogames', 'url': 'https://www.dailymotion.com/video/x2iuewm_steam-machine-models-pricing-listed-on-steam-store-ign-news_videogames',
@ -163,8 +161,8 @@ class DailymotionIE(DailymotionBaseInfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'title': 'Steam Machine Models, Pricing Listed on Steam Store - IGN News', 'title': 'Steam Machine Models, Pricing Listed on Steam Store - IGN News',
'description': 'Several come bundled with the Steam Controller.', 'description': 'Several come bundled with the Steam Controller.',
'thumbnail': r're:^https?:.*\.(?:jpg|png)$',
'duration': 74, 'duration': 74,
'thumbnail': r're:https?://s[12]\.dmcdn\.net/v/.+',
'timestamp': 1425657362, 'timestamp': 1425657362,
'upload_date': '20150306', 'upload_date': '20150306',
'uploader': 'IGN', 'uploader': 'IGN',
@ -183,10 +181,7 @@ class DailymotionIE(DailymotionBaseInfoExtractor):
'uploader': 'Katy Perry', 'uploader': 'Katy Perry',
'upload_date': '20130905', 'upload_date': '20130905',
}, },
'params': { 'skip': 'Invalid URL',
'skip_download': True,
},
'skip': 'VEVO is only available in some countries',
}, { }, {
# age-restricted video # age-restricted video
'url': 'http://www.dailymotion.com/video/xyh2zz_leanna-decker-cyber-girl-of-the-year-desires-nude-playboy-plus_redband', 'url': 'http://www.dailymotion.com/video/xyh2zz_leanna-decker-cyber-girl-of-the-year-desires-nude-playboy-plus_redband',
@ -259,9 +254,9 @@ class DailymotionIE(DailymotionBaseInfoExtractor):
'uploader_id': 'x2vtgmm', 'uploader_id': 'x2vtgmm',
'age_limit': 0, 'age_limit': 0,
'tags': [], 'tags': [],
'thumbnail': r're:https?://s[12]\.dmcdn\.net/v/.+',
'view_count': int, 'view_count': int,
'like_count': int, 'like_count': int,
'thumbnail': r're:https://\w+.dmcdn.net/v/WnEY61cmvMxt2Fi6d/x1080',
}, },
}, { }, {
# https://geo.dailymotion.com/player/xf7zn.html?playlist=x7wdsj # https://geo.dailymotion.com/player/xf7zn.html?playlist=x7wdsj
@ -276,18 +271,18 @@ class DailymotionIE(DailymotionBaseInfoExtractor):
'info_dict': { 'info_dict': {
'id': 'x8u4owg', 'id': 'x8u4owg',
'ext': 'mp4', 'ext': 'mp4',
'description': 'À bord du « véloto », l’alternative à la voiture pour la campagne',
'like_count': int, 'like_count': int,
'uploader': 'Le Parisien', 'uploader': 'Le Parisien',
'thumbnail': 'https://www.leparisien.fr/resizer/ho_GwveeYftNkLwg_cEta--5Bv4=/1200x675/cloudfront-eu-central-1.images.arcpublishing.com/leparisien/BFXJNEBN75EUNHGYJLORUC3TX4.jpg',
'upload_date': '20240309', 'upload_date': '20240309',
'view_count': int, 'view_count': int,
'tags': 'count:7',
'thumbnail': r're:https?://www\.leparisien\.fr/.+\.jpg',
'timestamp': 1709997866, 'timestamp': 1709997866,
'age_limit': 0, 'age_limit': 0,
'uploader_id': 'x32f7b', 'uploader_id': 'x32f7b',
'title': 'VIDÉO. Le «\xa0véloto\xa0», la voiture à pédales qui aimerait se faire une place sur les routes', 'title': 'VIDÉO. Le «\xa0véloto\xa0», la voiture à pédales qui aimerait se faire une place sur les routes',
'duration': 428.0, 'duration': 428.0,
'description': 'À bord du « véloto », l’alternative à la voiture pour la campagne',
'tags': ['biclou', 'vélo', 'véloto', 'campagne', 'voiture', 'environnement', 'véhicules intermédiaires'],
}, },
}, { }, {
# https://geo.dailymotion.com/player/xry80.html?video=x8vu47w # https://geo.dailymotion.com/player/xry80.html?video=x8vu47w
@ -297,9 +292,9 @@ class DailymotionIE(DailymotionBaseInfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'like_count': int, 'like_count': int,
'uploader': 'Metatube', 'uploader': 'Metatube',
'thumbnail': r're:https://\w+.dmcdn.net/v/W1G_S1coGSFTfkTeR/x1080',
'upload_date': '20240326', 'upload_date': '20240326',
'view_count': int, 'view_count': int,
'thumbnail': r're:https?://s[12]\.dmcdn\.net/v/.+',
'timestamp': 1711496732, 'timestamp': 1711496732,
'age_limit': 0, 'age_limit': 0,
'uploader_id': 'x2xpy74', 'uploader_id': 'x2xpy74',
@ -308,6 +303,7 @@ class DailymotionIE(DailymotionBaseInfoExtractor):
'description': 'Que lindura', 'description': 'Que lindura',
'tags': [], 'tags': [],
}, },
'skip': 'Invalid URL',
}, { }, {
# //geo.dailymotion.com/player/xysxq.html?video=k2Y4Mjp7krAF9iCuINM # //geo.dailymotion.com/player/xysxq.html?video=k2Y4Mjp7krAF9iCuINM
'url': 'https://lcp.fr/programmes/avant-la-catastrophe-la-naissance-de-la-dictature-nazie-1933-1936-346819', 'url': 'https://lcp.fr/programmes/avant-la-catastrophe-la-naissance-de-la-dictature-nazie-1933-1936-346819',
@ -322,11 +318,30 @@ class DailymotionIE(DailymotionBaseInfoExtractor):
'like_count': int, 'like_count': int,
'age_limit': 0, 'age_limit': 0,
'duration': 3220, 'duration': 3220,
'thumbnail': 'https://s1.dmcdn.net/v/Xvumk1djJBUZfjj2a/x1080',
'tags': [], 'tags': [],
'thumbnail': r're:https?://s[12]\.dmcdn\.net/v/.+',
'timestamp': 1739919947, 'timestamp': 1739919947,
'upload_date': '20250218', 'upload_date': '20250218',
}, },
'skip': 'Invalid URL',
}, {
'url': 'https://forum.ionicframework.com/t/ionic-2-jw-player-dailymotion-player/83248',
'info_dict': {
'id': 'xwr14q',
'ext': 'mp4',
'title': 'Macklemore & Ryan Lewis - Thrift Shop (feat. Wanz)',
'age_limit': 0,
'description': 'md5:47fbe168b5a6ddc4a205e20dd6c841b2',
'duration': 234,
'like_count': int,
'tags': 'count:5',
'thumbnail': r're:https?://s[12]\.dmcdn\.net/v/.+',
'timestamp': 1358177670,
'upload_date': '20130114',
'uploader': 'Macklemore Official',
'uploader_id': 'x19qlwr',
'view_count': int,
},
}] }]
_GEO_BYPASS = False _GEO_BYPASS = False
_COMMON_MEDIA_FIELDS = '''description _COMMON_MEDIA_FIELDS = '''description
@ -540,7 +555,7 @@ class DailymotionSearchIE(DailymotionPlaylistBaseIE):
'id': 'king of turtles', 'id': 'king of turtles',
'title': 'king of turtles', 'title': 'king of turtles',
}, },
'playlist_mincount': 90, 'playlist_mincount': 0,
}] }]
_SEARCH_QUERY = 'query SEARCH_QUERY( $query: String! $page: Int $limit: Int ) { search { videos( query: $query first: $limit page: $page ) { edges { node { xid } } } } } ' _SEARCH_QUERY = 'query SEARCH_QUERY( $query: String! $page: Int $limit: Int ) { search { videos( query: $query first: $limit page: $page ) { edges { node { xid } } } } } '
@ -584,7 +599,7 @@ class DailymotionUserIE(DailymotionPlaylistBaseIE):
'info_dict': { 'info_dict': {
'id': 'nqtv', 'id': 'nqtv',
}, },
'playlist_mincount': 152, 'playlist_mincount': 148,
}, { }, {
'url': 'http://www.dailymotion.com/user/UnderProject', 'url': 'http://www.dailymotion.com/user/UnderProject',
'info_dict': { 'info_dict': {

18
yt_dlp/extractor/dbtv.py
Unescape Escape View File

@ -12,13 +12,13 @@ class DBTVIE(InfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'title': 'Skulle teste ut fornøyelsespark, men kollegaen var bare opptatt av bikinikroppen', 'title': 'Skulle teste ut fornøyelsespark, men kollegaen var bare opptatt av bikinikroppen',
'description': 'md5:49cc8370e7d66e8a2ef15c3b4631fd3f', 'description': 'md5:49cc8370e7d66e8a2ef15c3b4631fd3f',
'thumbnail': r're:https?://.*\.jpg', 'thumbnail': r're:https?://.+\.jpg',
'upload_date': '20160916', 'upload_date': '20160916',
'duration': 69, 'duration': 69,
'uploader_id': 'UCk5pvsyZJoYJBd7_oFPTlRQ', 'uploader_id': 'UCk5pvsyZJoYJBd7_oFPTlRQ',
'uploader': 'Dagbladet', 'uploader': 'Dagbladet',
}, },
'add_ie': ['Youtube'], 'skip': 'Invalid URL',
}, { }, {
'url': 'https://www.dagbladet.no/video/embed/xlGmyIeN9Jo/?autoplay=false', 'url': 'https://www.dagbladet.no/video/embed/xlGmyIeN9Jo/?autoplay=false',
'only_matching': True, 'only_matching': True,
@ -26,6 +26,20 @@ class DBTVIE(InfoExtractor):
'url': 'https://www.dagbladet.no/video/truer-iran-bor-passe-dere/PalfB2Cw', 'url': 'https://www.dagbladet.no/video/truer-iran-bor-passe-dere/PalfB2Cw',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
# FIXME: Embed detection
'url': 'https://www.dagbladet.no/nyheter/rekordstort-russisk-angrep/83325693',
'info_dict': {
'id': '1HW7fYry',
'ext': 'mp4',
'title': 'Putin taler - så skjer dette',
'description': 'md5:3e8bacee33de861a9663d9a3fcc54e5e',
'display_id': 'putin-taler-sa-skjer-dette',
'thumbnail': r're:https?://cdn\.jwplayer\.com/v2/media/.+',
'timestamp': 1751043600,
'upload_date': '20250627',
},
}]
def _real_extract(self, url): def _real_extract(self, url):
display_id, video_id = self._match_valid_url(url).groups() display_id, video_id = self._match_valid_url(url).groups()

39
yt_dlp/extractor/ertgr.py
Unescape Escape View File

@ -64,14 +64,12 @@ class ERTFlixCodenameIE(ERTFlixBaseIE):
_VALID_URL = r'ertflix:(?P<id>[\w-]+)' _VALID_URL = r'ertflix:(?P<id>[\w-]+)'
_TESTS = [{ _TESTS = [{
'url': 'ertflix:monogramma-praxitelis-tzanoylinos', 'url': 'ertflix:monogramma-praxitelis-tzanoylinos',
'md5': '5b9c2cd171f09126167e4082fc1dd0ef',
'info_dict': { 'info_dict': {
'id': 'monogramma-praxitelis-tzanoylinos', 'id': 'monogramma-praxitelis-tzanoylinos',
'ext': 'mp4', 'ext': 'mp4',
'title': 'md5:ef0b439902963d56c43ac83c3f41dd0e', 'title': 'monogramma-praxitelis-tzanoylinos',
}, },
}, }]
]
def _extract_formats_and_subs(self, video_id): def _extract_formats_and_subs(self, video_id):
media_info = self._call_api(video_id, codename=video_id) media_info = self._call_api(video_id, codename=video_id)
@ -131,13 +129,14 @@ class ERTFlixIE(ERTFlixBaseIE):
'duration': 3166, 'duration': 3166,
'age_limit': 8, 'age_limit': 8,
}, },
'skip': 'Invalid URL',
}, { }, {
'url': 'https://www.ertflix.gr/series/ser.3448-monogramma', 'url': 'https://www.ertflix.gr/series/ser.3448-monogramma',
'info_dict': { 'info_dict': {
'id': 'ser.3448', 'id': 'ser.3448',
'age_limit': 8, 'age_limit': 8,
'description': 'Η εκπομπή σαράντα ετών που σημάδεψε τον πολιτισμό μας.', 'title': 'Monogramma',
'title': 'Μονόγραμμα', 'description': 'md5:e30cc640e6463da87f210a8ed10b2439',
}, },
'playlist_mincount': 64, 'playlist_mincount': 64,
}, { }, {
@ -145,28 +144,28 @@ class ERTFlixIE(ERTFlixBaseIE):
'info_dict': { 'info_dict': {
'id': 'ser.3448', 'id': 'ser.3448',
'age_limit': 8, 'age_limit': 8,
'description': 'Η εκπομπή σαράντα ετών που σημάδεψε τον πολιτισμό μας.', 'title': 'Monogramma',
'title': 'Μονόγραμμα', 'description': 'md5:e30cc640e6463da87f210a8ed10b2439',
}, },
'playlist_count': 22, 'playlist_mincount': 66,
}, { }, {
'url': 'https://www.ertflix.gr/series/ser.3448-monogramma?season=1&season=2021%20-%202022', 'url': 'https://www.ertflix.gr/series/ser.3448-monogramma?season=1&season=2021%20-%202022',
'info_dict': { 'info_dict': {
'id': 'ser.3448', 'id': 'ser.3448',
'age_limit': 8, 'age_limit': 8,
'description': 'Η εκπομπή σαράντα ετών που σημάδεψε τον πολιτισμό μας.', 'title': 'Monogramma',
'title': 'Μονόγραμμα', 'description': 'md5:e30cc640e6463da87f210a8ed10b2439',
}, },
'playlist_mincount': 36, 'playlist_mincount': 25,
}, { }, {
'url': 'https://www.ertflix.gr/series/ser.164991-to-diktuo-1?season=1-9', 'url': 'https://www.ertflix.gr/series/ser.164991-to-diktuo-1?season=1-9',
'info_dict': { 'info_dict': {
'id': 'ser.164991', 'id': 'ser.164991',
'age_limit': 8, 'age_limit': 8,
'description': 'Η πρώτη ελληνική εκπομπή με θεματολογία αποκλειστικά γύρω από το ίντερνετ.', 'title': 'The Network',
'title': 'Το δίκτυο', 'description': 'The first Greek show featuring topics exclusively around the internet.',
}, },
'playlist_mincount': 9, 'playlist_mincount': 0,
}, { }, {
'url': 'https://www.ertflix.gr/en/vod/vod.127652-ta-kalytera-mas-chronia-ep1-mia-volta-sto-feggari', 'url': 'https://www.ertflix.gr/en/vod/vod.127652-ta-kalytera-mas-chronia-ep1-mia-volta-sto-feggari',
'only_matching': True, 'only_matching': True,
@ -282,6 +281,16 @@ class ERTWebtvEmbedIE(InfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'thumbnail': 'https://program.ert.gr/photos/2022/1/to_diktio_ep09_i_istoria_tou_diadiktiou_stin_Ellada_1021x576.jpg', 'thumbnail': 'https://program.ert.gr/photos/2022/1/to_diktio_ep09_i_istoria_tou_diadiktiou_stin_Ellada_1021x576.jpg',
}, },
'skip': 'Invalid URL',
}]
_WEBPAGE_TESTS = [{
'url': 'https://www.ertnews.gr/video/manolis-goyalles-o-anthropos-piso-apo-ti-diadiktyaki-vasilopita/',
'info_dict': {
'id': '2022/tv/news-themata-ianouarios/20220114-apotis6-gouales-pita.mp4',
'ext': 'mp4',
'title': 'VOD - 2022/tv/news-themata-ianouarios/20220114-apotis6-gouales-pita.mp4',
'thumbnail': r're:https?://www\.ert\.gr/themata/photos/.+\.jpg',
},
}] }]
def _real_extract(self, url): def _real_extract(self, url):

144
yt_dlp/extractor/facebook.py
Unescape Escape View File

@ -81,13 +81,14 @@ class FacebookIE(InfoExtractor):
'description': 'md5:34675bda53336b1d16400265c2bb9b3b', 'description': 'md5:34675bda53336b1d16400265c2bb9b3b',
'uploader': 'RADIO KICKS FM', 'uploader': 'RADIO KICKS FM',
'upload_date': '20230818', 'upload_date': '20230818',
'thumbnail': r're:https?://scontent\.fitm\d-1\.fna\.fbcdn\.net/.+',
'timestamp': 1692346159, 'timestamp': 1692346159,
'thumbnail': r're:^https?://.*',
'uploader_id': '100063551323670', 'uploader_id': '100063551323670',
'duration': 3133.583, 'duration': 3133.583,
'view_count': int, 'view_count': int,
'concurrent_view_count': 0, 'concurrent_view_count': 0,
}, },
'expected_warnings': ['Cannot parse data'],
}, { }, {
'url': 'https://www.facebook.com/video.php?v=637842556329505&fref=nf', 'url': 'https://www.facebook.com/video.php?v=637842556329505&fref=nf',
'md5': '6a40d33c0eccbb1af76cf0485a052659', 'md5': '6a40d33c0eccbb1af76cf0485a052659',
@ -106,17 +107,18 @@ class FacebookIE(InfoExtractor):
'info_dict': { 'info_dict': {
'id': '274175099429670', 'id': '274175099429670',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Asif', 'title': '119 reactions · 1.4K shares | Asif Nawab Butt on Reels',
'description': '', 'description': '',
'uploader': 'Asif Nawab Butt', 'uploader': 'Asif Nawab Butt',
'upload_date': '20140506', 'upload_date': '20140506',
'thumbnail': r're:https?://scontent\.fitm\d-1\.fna\.fbcdn\.net/.+',
'timestamp': 1399398998, 'timestamp': 1399398998,
'thumbnail': r're:^https?://.*', 'uploader_id': 'pfbid028xue38TBXRyNbiqBSV2LFs3QK3yopvKjupbqFoL6U9SKbx4p2SMdJjQSBvnjsHGWl',
'uploader_id': 'pfbid05AzrFTXgY37tqwaSgbFTTEpCLBjjEJHkigogwGiRPtKEpAsJYJpzE94H1RxYXWEtl',
'duration': 131.03, 'duration': 131.03,
'concurrent_view_count': int, 'concurrent_view_count': int,
'view_count': int, 'view_count': int,
}, },
'expected_warnings': ['Cannot parse data'],
}, { }, {
'note': 'Video with DASH manifest', 'note': 'Video with DASH manifest',
'url': 'https://www.facebook.com/video.php?v=957955867617029', 'url': 'https://www.facebook.com/video.php?v=957955867617029',
@ -158,7 +160,7 @@ class FacebookIE(InfoExtractor):
'id': '10153664894881749', 'id': '10153664894881749',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Average time to confirm recent Supreme Court nominees: 67 days Longest it\'s t...', 'title': 'Average time to confirm recent Supreme Court nominees: 67 days Longest it\'s t...',
'thumbnail': r're:^https?://.*', 'thumbnail': r're:https?://scontent\.fitm\d-1\.fna\.fbcdn\.net/.+',
'timestamp': 1456259628, 'timestamp': 1456259628,
'upload_date': '20160223', 'upload_date': '20160223',
'uploader': 'Barack Obama', 'uploader': 'Barack Obama',
@ -168,7 +170,7 @@ class FacebookIE(InfoExtractor):
# have 1080P, but only up to 720p in swf params # have 1080P, but only up to 720p in swf params
# data.video.story.attachments[].media # data.video.story.attachments[].media
'url': 'https://www.facebook.com/cnn/videos/10155529876156509/', 'url': 'https://www.facebook.com/cnn/videos/10155529876156509/',
'md5': '1659aa21fb3dd1585874f668e81a72c8', 'md5': '70b82ebf5f0e9b91b2a49d3db3563611',
'info_dict': { 'info_dict': {
'id': '10155529876156509', 'id': '10155529876156509',
'ext': 'mp4', 'ext': 'mp4',
@ -177,7 +179,7 @@ class FacebookIE(InfoExtractor):
'timestamp': 1477818095, 'timestamp': 1477818095,
'upload_date': '20161030', 'upload_date': '20161030',
'uploader': 'CNN', 'uploader': 'CNN',
'thumbnail': r're:^https?://.*', 'thumbnail': r're:https?://scontent\.fitm\d-1\.fna\.fbcdn\.net/.+',
'view_count': int, 'view_count': int,
'uploader_id': '100059479812265', 'uploader_id': '100059479812265',
'concurrent_view_count': int, 'concurrent_view_count': int,
@ -198,13 +200,11 @@ class FacebookIE(InfoExtractor):
'uploader': 'Yaroslav Korpan', 'uploader': 'Yaroslav Korpan',
'uploader_id': 'pfbid06AScABAWcW91qpiuGrLt99Ef9tvwHoXP6t8KeFYEqkSfreMtfa9nTveh8b2ZEVSWl', 'uploader_id': 'pfbid06AScABAWcW91qpiuGrLt99Ef9tvwHoXP6t8KeFYEqkSfreMtfa9nTveh8b2ZEVSWl',
'concurrent_view_count': int, 'concurrent_view_count': int,
'thumbnail': r're:^https?://.*', 'thumbnail': r're:https?://scontent\.fitm\d-1\.fna\.fbcdn\.net/.+',
'view_count': int, 'view_count': int,
'duration': 11736.446, 'duration': 11736.446,
}, },
'params': { 'skip': 'Invalid URL',
'skip_download': True,
},
}, { }, {
# FIXME: Cannot parse data error # FIXME: Cannot parse data error
'url': 'https://www.facebook.com/LaGuiaDelVaron/posts/1072691702860471', 'url': 'https://www.facebook.com/LaGuiaDelVaron/posts/1072691702860471',
@ -215,7 +215,7 @@ class FacebookIE(InfoExtractor):
'timestamp': 1477305000, 'timestamp': 1477305000,
'upload_date': '20161024', 'upload_date': '20161024',
'uploader': 'La Guía Del Varón', 'uploader': 'La Guía Del Varón',
'thumbnail': r're:^https?://.*', 'thumbnail': r're:https?://scontent\.fitm\d-1\.fna\.fbcdn\.net/.+',
}, },
'skip': 'Requires logging in', 'skip': 'Requires logging in',
}, { }, {
@ -244,9 +244,10 @@ class FacebookIE(InfoExtractor):
'upload_date': '20171124', 'upload_date': '20171124',
'uploader': 'Vickie Gentry', 'uploader': 'Vickie Gentry',
'uploader_id': 'pfbid0FkkycT95ySNNyfCw4Cho6u5G7WbbZEcxT496Hq8rtx1K3LcTCATpR3wnyYhmyGC5l', 'uploader_id': 'pfbid0FkkycT95ySNNyfCw4Cho6u5G7WbbZEcxT496Hq8rtx1K3LcTCATpR3wnyYhmyGC5l',
'thumbnail': r're:^https?://.*', 'thumbnail': r're:https?://scontent\.fitm\d-1\.fna\.fbcdn\.net/.+',
'duration': 148.224, 'duration': 148.224,
}, },
'skip': 'Invalid URL',
}, { }, {
# data.node.comet_sections.content.story.attachments[].styles.attachment.media # data.node.comet_sections.content.story.attachments[].styles.attachment.media
'url': 'https://www.facebook.com/attn/posts/pfbid0j1Czf2gGDVqeQ8KiMLFm3pWN8GxsQmeRrVhimWDzMuKQoR8r4b1knNsejELmUgyhl', 'url': 'https://www.facebook.com/attn/posts/pfbid0j1Czf2gGDVqeQ8KiMLFm3pWN8GxsQmeRrVhimWDzMuKQoR8r4b1knNsejELmUgyhl',
@ -260,7 +261,7 @@ class FacebookIE(InfoExtractor):
'duration': 132.675, 'duration': 132.675,
'uploader_id': '100064451419378', 'uploader_id': '100064451419378',
'view_count': int, 'view_count': int,
'thumbnail': r're:^https?://.*', 'thumbnail': r're:https?://scontent\.fitm\d-1\.fna\.fbcdn\.net/.+',
'timestamp': 1701975646, 'timestamp': 1701975646,
}, },
}, { }, {
@ -271,9 +272,9 @@ class FacebookIE(InfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'title': 'Lela Evans', 'title': 'Lela Evans',
'description': 'Today Makkovik\'s own Pilot Mandy Smith made her inaugural landing on the airstrip in her hometown. What a proud moment as we all cheered and...', 'description': 'Today Makkovik\'s own Pilot Mandy Smith made her inaugural landing on the airstrip in her hometown. What a proud moment as we all cheered and...',
'thumbnail': r're:^https?://.*', 'thumbnail': r're:https?://scontent\.fitm\d-1\.fna\.fbcdn\.net/.+',
'uploader': 'Lela Evans', 'uploader': 'Lela Evans',
'uploader_id': 'pfbid0swT2y7t6TAsZVBvcyeYPdhTMefGaS26mzUwML3vd1ma6ndGZKxsyS4Ssu3jitZLXl', 'uploader_id': 'pfbid02wjMpknobSMnyynK3TNKN4Ww1StcpAKXgowqTyge3bz7LwHZMQ68uiXzzbu7xeryBl',
'upload_date': '20231228', 'upload_date': '20231228',
'timestamp': 1703804085, 'timestamp': 1703804085,
'duration': 394.347, 'duration': 394.347,
@ -326,28 +327,27 @@ class FacebookIE(InfoExtractor):
'uploader_id': '100066514874195', 'uploader_id': '100066514874195',
'duration': 4524.001, 'duration': 4524.001,
'view_count': int, 'view_count': int,
'thumbnail': r're:^https?://.*', 'thumbnail': r're:https?://scontent\.fitm\d-1\.fna\.fbcdn\.net/.+',
'concurrent_view_count': int, 'concurrent_view_count': int,
}, },
'params': { 'params': {'skip_download': True},
'skip_download': True,
},
}, { }, {
# data.node.comet_sections.content.story.attachments[].style_type_renderer.attachment.all_subattachments.nodes[].media # data.node.comet_sections.content.story.attachments[].style_type_renderer.attachment.all_subattachments.nodes[].media
'url': 'https://www.facebook.com/100033620354545/videos/106560053808006/', 'url': 'https://www.facebook.com/100033620354545/videos/106560053808006/',
'info_dict': { 'info_dict': {
'id': '106560053808006', 'id': '106560053808006',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Josef', 'title': 'Josef Novak on Reels',
'thumbnail': r're:^https?://.*', 'thumbnail': r're:https?://scontent\.fitm\d-1\.fna\.fbcdn\.net/.+',
'concurrent_view_count': int, 'concurrent_view_count': int,
'uploader_id': 'pfbid02gpfwRM2XvdEJfsERupwQiNmBiDArc38RMRYZnap372q6Vs7MtFTVy72mmFWpJBTKl', 'uploader_id': 'pfbid0cjYJYXpePWqhZ9DgpB6gKXrN2q3obwducdKm4wT7K5nkhbfKg5cneocYbsdaji7fl',
'timestamp': 1549275572, 'timestamp': 1549275572,
'duration': 3.283, 'duration': 3.283,
'uploader': 'Josef Novak', 'uploader': 'Josef Novak',
'description': '', 'description': '',
'upload_date': '20190204', 'upload_date': '20190204',
}, },
'expected_warnings': ['Cannot parse data'],
}, { }, {
# data.video.story.attachments[].media # data.video.story.attachments[].media
'url': 'https://www.facebook.com/watch/?v=647537299265662', 'url': 'https://www.facebook.com/watch/?v=647537299265662',
@ -406,7 +406,7 @@ class FacebookIE(InfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'title': 'ANALISI IN CAMPO OSCURO " Coaguli nel sangue dei vaccinati"', 'title': 'ANALISI IN CAMPO OSCURO " Coaguli nel sangue dei vaccinati"',
'description': 'Other event by Comitato Liberi Pensatori on Tuesday, October 18 2022', 'description': 'Other event by Comitato Liberi Pensatori on Tuesday, October 18 2022',
'thumbnail': r're:^https?://.*', 'thumbnail': r're:https?://scontent\.fitm\d-1\.fna\.fbcdn\.net/.+',
'uploader': 'Comitato Liberi Pensatori', 'uploader': 'Comitato Liberi Pensatori',
'uploader_id': '100065709540881', 'uploader_id': '100065709540881',
}, },
@ -414,6 +414,56 @@ class FacebookIE(InfoExtractor):
'url': 'https://www.facebook.com/groups/1513990329015294/posts/d41d8cd9/2013209885760000/?app=fbl', 'url': 'https://www.facebook.com/groups/1513990329015294/posts/d41d8cd9/2013209885760000/?app=fbl',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
# <iframe> embed
'url': 'http://www.unique-almeria.com/mini-hollywood.html',
'md5': 'cba5d8c5021e9340dcefe925255e2c3e',
'info_dict': {
'id': '1529066599879',
'ext': 'mp4',
'title': 'Facebook video #1529066599879',
},
'expected_warnings': ['unable to extract uploader'],
}, {
# FIXME: Embed detection
# <iframe> embed, plugin video
'url': 'https://www.newsmemory.com/eedition/e-publishing-solutions/2-in-one-app/',
'md5': 'ae97d4a44f8cc9a8b1a4c03b9ed793af',
'info_dict': {
'id': '10155710648695814',
'ext': 'mp4',
'title': 'Download the all new and improved Trinidad Express App',
'concurrent_view_count': int,
'description': 'md5:4806195c99908e4189b45b1c23bd4f89',
'duration': 69.408,
'thumbnail': r're:https?://scontent\.fitm\d-1\.fna\.fbcdn\.net/.+',
'timestamp': 1533919195,
'upload_date': '20180810',
'uploader': 'Trinidad Express Newspapers',
'uploader_id': '100064446413648',
'view_count': int,
},
'expected_warnings': ['Cannot parse data'],
}, {
# API embed
'url': 'https://www.curs.md/ro',
'md5': '090bae53b9bff2be993c896edc2ea205',
'info_dict': {
'id': '334484292523563',
'ext': 'mp4',
'title': 'md5:9abffe1c86cdd967ffa224e1ccc13b90',
'concurrent_view_count': int,
'description': 'md5:0ba98567a61c640f9fabf1882235b33d',
'duration': 8789.891,
'thumbnail': r're:https?://scontent\.fitm\d-1\.fna\.fbcdn\.net/.+',
'timestamp': 1700603114,
'upload_date': '20231121',
'uploader': 'Istoria Moldovei',
'uploader_id': '100063529778592',
'view_count': int,
},
'params': {'extractor_args': {'generic': {'impersonate': ['chrome']}}},
}]
_SUPPORTED_PAGLETS_REGEX = r'(?:pagelet_group_mall|permalink_video_pagelet|hyperfeed_story_id_[0-9a-f]+)' _SUPPORTED_PAGLETS_REGEX = r'(?:pagelet_group_mall|permalink_video_pagelet|hyperfeed_story_id_[0-9a-f]+)'
_api_config = { _api_config = {
'graphURI': '/api/graphql/', 'graphURI': '/api/graphql/',
@ -898,20 +948,24 @@ class FacebookIE(InfoExtractor):
class FacebookPluginsVideoIE(InfoExtractor): class FacebookPluginsVideoIE(InfoExtractor):
_VALID_URL = r'https?://(?:[\w-]+\.)?facebook\.com/plugins/video\.php\?.*?\bhref=(?P<id>https.+)' _VALID_URL = r'https?://(?:[\w-]+\.)?facebook\.com/plugins/video\.php\?.*?\bhref=(?P<id>https.+)'
_TESTS = [{ _TESTS = [{
'url': 'https://www.facebook.com/plugins/video.php?href=https%3A%2F%2Fwww.facebook.com%2Fgov.sg%2Fvideos%2F10154383743583686%2F&show_text=0&width=560', 'url': 'https://www.facebook.com/plugins/video.php?href=https%3A%2F%2Fwww.facebook.com%2Fgov.sg%2Fvideos%2F10154383743583686%2F&show_text=0&width=560',
'md5': '5954e92cdfe51fe5782ae9bda7058a07', 'md5': 'af83aeae1d595f377c6e47a450828155',
'info_dict': { 'info_dict': {
'id': '10154383743583686', 'id': '10154383743583686',
'ext': 'mp4', 'ext': 'mp4',
# TODO: Fix title, uploader
'title': 'What to do during the haze?', 'title': 'What to do during the haze?',
'uploader': 'Gov.sg', 'concurrent_view_count': int,
'upload_date': '20160826', 'description': 'md5:81839c0979803a014b20798df255ed0b',
'duration': 65.087,
'thumbnail': r're:https?://scontent\.fitm\d-1\.fna\.fbcdn\.net/.+',
'timestamp': 1472184808, 'timestamp': 1472184808,
'upload_date': '20160826',
'uploader': 'gov.sg',
'uploader_id': '100064718678925',
'view_count': int,
}, },
'add_ie': [FacebookIE.ie_key()], 'expected_warnings': ['Cannot parse data'],
}, { }, {
'url': 'https://www.facebook.com/plugins/video.php?href=https%3A%2F%2Fwww.facebook.com%2Fvideo.php%3Fv%3D10204634152394104', 'url': 'https://www.facebook.com/plugins/video.php?href=https%3A%2F%2Fwww.facebook.com%2Fvideo.php%3Fv%3D10204634152394104',
'only_matching': True, 'only_matching': True,
@ -945,7 +999,7 @@ class FacebookRedirectURLIE(InfoExtractor):
'tags': 'count:11', 'tags': 'count:11',
'duration': 3332, 'duration': 3332,
'live_status': 'not_live', 'live_status': 'not_live',
'thumbnail': 'https://i.ytimg.com/vi/pO8h3EaFRdo/maxresdefault.jpg', 'thumbnail': r're:https?://i\.ytimg\.com/vi/.+',
'channel_url': 'https://www.youtube.com/channel/UCGBpxWJr9FNOcFYA5GkKrMg', 'channel_url': 'https://www.youtube.com/channel/UCGBpxWJr9FNOcFYA5GkKrMg',
'availability': 'public', 'availability': 'public',
'uploader_url': 'http://www.youtube.com/user/brtvofficial', 'uploader_url': 'http://www.youtube.com/user/brtvofficial',
@ -954,8 +1008,7 @@ class FacebookRedirectURLIE(InfoExtractor):
'view_count': int, 'view_count': int,
'like_count': int, 'like_count': int,
}, },
'add_ie': ['Youtube'], 'skip': 'Youtube video is now private',
'params': {'skip_download': 'Youtube'},
}] }]
def _real_extract(self, url): def _real_extract(self, url):
@ -968,22 +1021,20 @@ class FacebookRedirectURLIE(InfoExtractor):
class FacebookReelIE(InfoExtractor): class FacebookReelIE(InfoExtractor):
_VALID_URL = r'https?://(?:[\w-]+\.)?facebook\.com/reel/(?P<id>\d+)' _VALID_URL = r'https?://(?:[\w-]+\.)?facebook\.com/reel/(?P<id>\d+)'
IE_NAME = 'facebook:reel' IE_NAME = 'facebook:reel'
_TESTS = [{ _TESTS = [{
'url': 'https://www.facebook.com/reel/1195289147628387', 'url': 'https://www.facebook.com/reel/1195289147628387',
'md5': 'a53256d10fc2105441fe0c4212ed8cea', 'md5': 'aeb0153ecb2eaacdf2dc2bf88f593fef',
'info_dict': { 'info_dict': {
'id': '1195289147628387', 'id': '1195289147628387',
'ext': 'mp4', 'ext': 'mp4',
'title': r're:9\.6K views · 355 reactions .+ Let the “Slapathon” commence!! .+ LL COOL J · Mama Said Knock You Out$', 'title': '9.7K views · 352 reactions | When your trying to help your partner out with an arrest and #FAAFO games begin. Let the “Slapathon” commence!! 👊👋 | Beast Camp Training',
'description': r're:When your trying to help your partner .+ LL COOL J · Mama Said Knock You Out$', 'description': 'md5:5a767dc7e78718667b150a7facc4a34f',
'uploader': 'Beast Camp Training', 'uploader': '9.7K views &#xb7; 352 reactions | When your trying to help your partner out with an arrest and #FAAFO games begin. Let the &#x201c;Slapathon&#x201d; commence!! &#x1f44a;&#x1f44b; | Beast Camp Training',
'uploader_id': '100040874179269', 'uploader_id': '100040874179269',
'duration': 9.579, 'duration': 9.579,
'thumbnail': r're:https?://scontent\.fitm\d-1\.fna\.fbcdn\.net/.+',
'timestamp': 1637502609, 'timestamp': 1637502609,
'upload_date': '20211121', 'upload_date': '20211121',
'thumbnail': r're:^https?://.*',
'like_count': int,
'comment_count': int, 'comment_count': int,
'repost_count': int, 'repost_count': int,
}, },
@ -998,7 +1049,6 @@ class FacebookReelIE(InfoExtractor):
class FacebookAdsIE(InfoExtractor): class FacebookAdsIE(InfoExtractor):
_VALID_URL = r'https?://(?:[\w-]+\.)?facebook\.com/ads/library/?\?(?:[^#]+&)?id=(?P<id>\d+)' _VALID_URL = r'https?://(?:[\w-]+\.)?facebook\.com/ads/library/?\?(?:[^#]+&)?id=(?P<id>\d+)'
IE_NAME = 'facebook:ads' IE_NAME = 'facebook:ads'
_TESTS = [{ _TESTS = [{
'url': 'https://www.facebook.com/ads/library/?id=899206155126718', 'url': 'https://www.facebook.com/ads/library/?id=899206155126718',
'info_dict': { 'info_dict': {
@ -1008,12 +1058,13 @@ class FacebookAdsIE(InfoExtractor):
'description': 'md5:0822724069e3aca97cbed5dabbab282e', 'description': 'md5:0822724069e3aca97cbed5dabbab282e',
'uploader': 'Kandao', 'uploader': 'Kandao',
'uploader_id': '774114102743284', 'uploader_id': '774114102743284',
'uploader_url': r're:^https?://.*', 'uploader_url': 'https://facebook.com/KandaoVR',
'timestamp': 1702548330, 'timestamp': 1702548330,
'thumbnail': r're:^https?://.*', 'thumbnail': r're:https?://scontent\.fitm\d-1\.fna\.fbcdn\.net/.+',
'upload_date': '20231214', 'upload_date': '20231214',
'like_count': int, 'like_count': int,
}, },
'skip': 'Invalid URL',
}, { }, {
# key 'watermarked_video_sd_url' missing # key 'watermarked_video_sd_url' missing
'url': 'https://www.facebook.com/ads/library/?id=501152689226254', 'url': 'https://www.facebook.com/ads/library/?id=501152689226254',
@ -1024,9 +1075,9 @@ class FacebookAdsIE(InfoExtractor):
'description': 'md5:02a446ace7ff8c3c37a2892922492490', 'description': 'md5:02a446ace7ff8c3c37a2892922492490',
'uploader': 'mat.nawrocki', 'uploader': 'mat.nawrocki',
'uploader_id': '148586968341456', 'uploader_id': '148586968341456',
'uploader_url': r're:^https?://.*', 'uploader_url': 'https://www.instagram.com/_u/mat.nawrocki',
'thumbnail': r're:https?://scontent\.fitm\d-1\.fna\.fbcdn\.net/.+',
'timestamp': 1723452305, 'timestamp': 1723452305,
'thumbnail': r're:^https?://.*',
'upload_date': '20240812', 'upload_date': '20240812',
'like_count': int, 'like_count': int,
}, },
@ -1037,12 +1088,13 @@ class FacebookAdsIE(InfoExtractor):
'title': 'Jusqu\u2019\u00e0 -25% sur une s\u00e9lection de vins p\u00e9tillants italiens ', 'title': 'Jusqu\u2019\u00e0 -25% sur une s\u00e9lection de vins p\u00e9tillants italiens ',
'uploader': 'Eataly Paris Marais', 'uploader': 'Eataly Paris Marais',
'uploader_id': '2086668958314152', 'uploader_id': '2086668958314152',
'uploader_url': r're:^https?://.*', 'uploader_url': 'https://facebook.com/EatalyParisMarais',
'timestamp': 1703571529, 'timestamp': 1703571529,
'upload_date': '20231226', 'upload_date': '20231226',
'like_count': int, 'like_count': int,
}, },
'playlist_count': 3, 'playlist_count': 3,
'skip': 'Invalid URL',
}, { }, {
'url': 'https://es-la.facebook.com/ads/library/?id=901230958115569', 'url': 'https://es-la.facebook.com/ads/library/?id=901230958115569',
'only_matching': True, 'only_matching': True,

2573
yt_dlp/extractor/generic.py
View File

File diff suppressed because it is too large Load Diff

25
yt_dlp/extractor/glomex.py
Unescape Escape View File

@ -112,16 +112,17 @@ class GlomexIE(GlomexBaseIE):
_TESTS = [{ _TESTS = [{
'url': 'https://video.glomex.com/sport/v-cb24uwg77hgh-nach-2-0-sieg-guardiola-mit-mancity-vor-naechstem-titel', 'url': 'https://video.glomex.com/sport/v-cb24uwg77hgh-nach-2-0-sieg-guardiola-mit-mancity-vor-naechstem-titel',
'md5': 'cec33a943c4240c9cb33abea8c26242e',
'info_dict': { 'info_dict': {
'id': 'v-cb24uwg77hgh', 'id': 'v-cb24uwg77hgh',
'ext': 'mp4', 'ext': 'mp4',
'title': 'md5:38a90cedcfadd72982c81acf13556e0c', 'title': 'Nach 2:0-Sieg: Guardiola mit ManCity vor nächstem Titel',
'description': 'md5:1ea6b6caff1443fcbbba159e432eedb8', 'description': 'md5:1ea6b6caff1443fcbbba159e432eedb8',
'duration': 29600, 'duration': 29600,
'thumbnail': r're:https?://i[a-z0-9]thumbs\.glomex\.com/.+',
'timestamp': 1619895017, 'timestamp': 1619895017,
'upload_date': '20210501', 'upload_date': '20210501',
}, },
'params': {'skip_download': 'm3u8'},
}] }]
def _real_extract(self, url): def _real_extract(self, url):
@ -140,16 +141,17 @@ class GlomexEmbedIE(GlomexBaseIE):
_TESTS = [{ _TESTS = [{
'url': 'https://player.glomex.com/integration/1/iframe-player.html?integrationId=4059a013k56vb2yd&playlistId=v-cfa6lye0dkdd-sf', 'url': 'https://player.glomex.com/integration/1/iframe-player.html?integrationId=4059a013k56vb2yd&playlistId=v-cfa6lye0dkdd-sf',
'md5': '68f259b98cc01918ac34180142fce287',
'info_dict': { 'info_dict': {
'id': 'v-cfa6lye0dkdd-sf', 'id': 'v-cfa6lye0dkdd-sf',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Φώφη Γεννηματά: Ο επικήδειος λόγος του 17χρονου γιου της, Γιώργου',
'thumbnail': r're:https?://i[a-z0-9]thumbs\.glomex\.com/.+',
'timestamp': 1635337199, 'timestamp': 1635337199,
'duration': 133080, 'duration': 133080,
'upload_date': '20211027', 'upload_date': '20211027',
'description': 'md5:e741185fc309310ff5d0c789b437be66', 'description': 'md5:e741185fc309310ff5d0c789b437be66',
'title': 'md5:35647293513a6c92363817a0fb0a7961',
}, },
'params': {'skip_download': 'm3u8'},
}, { }, {
'url': 'https://player.glomex.com/integration/1/iframe-player.html?origin=fullpage&integrationId=19syy24xjn1oqlpc&playlistId=rl-vcb49w1fb592p&playlistIndex=0', 'url': 'https://player.glomex.com/integration/1/iframe-player.html?origin=fullpage&integrationId=19syy24xjn1oqlpc&playlistId=rl-vcb49w1fb592p&playlistIndex=0',
'info_dict': { 'info_dict': {
@ -157,12 +159,27 @@ class GlomexEmbedIE(GlomexBaseIE):
}, },
'playlist_count': 100, 'playlist_count': 100,
}, { }, {
# Geo-restricted
'url': 'https://player.glomex.com/integration/1/iframe-player.html?playlistId=cl-bgqaata6aw8x&integrationId=19syy24xjn1oqlpc', 'url': 'https://player.glomex.com/integration/1/iframe-player.html?playlistId=cl-bgqaata6aw8x&integrationId=19syy24xjn1oqlpc',
'info_dict': { 'info_dict': {
'id': 'cl-bgqaata6aw8x', 'id': 'cl-bgqaata6aw8x',
}, },
'playlist_mincount': 2, 'playlist_mincount': 2,
}] }]
_WEBPAGE_TESTS = [{
'url': 'https://www.skai.gr/news/world/iatrikos-syllogos-tourkias-to-turkovac-aplo-dialyma-erntogan-eiste-apateones-kai-pseytes',
'info_dict': {
'id': 'v-ch2nkhcirwc9-sf',
'ext': 'mp4',
'title': 'Ιατρικός Σύλλογος Τουρκίας: Το Turkovac είναι ένα απλό διάλυμα –Ερντογάν: Είστε απατεώνες και ψεύτες',
'description': 'md5:8b517a61d577efe7e36fde72fd535995',
'duration': 460000,
'thumbnail': r're:https?://i[a-z0-9]thumbs\.glomex\.com/.+',
'timestamp': 1641885019,
'upload_date': '20220111',
},
'params': {'skip_download': 'm3u8'},
}]
@classmethod @classmethod
def build_player_url(cls, video_id, integration, origin_url=None): def build_player_url(cls, video_id, integration, origin_url=None):

25
yt_dlp/extractor/indavideo.py
Unescape Escape View File

@ -11,7 +11,6 @@ from ..utils import (
class IndavideoEmbedIE(InfoExtractor): class IndavideoEmbedIE(InfoExtractor):
_VALID_URL = r'https?://(?:(?:embed\.)?indavideo\.hu/player/video/|assets\.indavideo\.hu/swf/player\.swf\?.*\b(?:v(?:ID|id))=)(?P<id>[\da-f]+)' _VALID_URL = r'https?://(?:(?:embed\.)?indavideo\.hu/player/video/|assets\.indavideo\.hu/swf/player\.swf\?.*\b(?:v(?:ID|id))=)(?P<id>[\da-f]+)'
# Some example URLs covered by generic extractor: # Some example URLs covered by generic extractor:
# https://indavideo.hu/video/Vicces_cica_1
# https://index.indavideo.hu/video/Hod_Nemetorszagban # https://index.indavideo.hu/video/Hod_Nemetorszagban
# https://auto.indavideo.hu/video/Sajat_utanfutoban_a_kis_tacsko # https://auto.indavideo.hu/video/Sajat_utanfutoban_a_kis_tacsko
# https://film.indavideo.hu/video/f_farkaslesen # https://film.indavideo.hu/video/f_farkaslesen
@ -25,14 +24,14 @@ class IndavideoEmbedIE(InfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'title': 'Cicatánc', 'title': 'Cicatánc',
'description': '', 'description': '',
'thumbnail': r're:^https?://.*\.jpg$',
'uploader': 'cukiajanlo', 'uploader': 'cukiajanlo',
'uploader_id': '83729', 'uploader_id': '83729',
'thumbnail': r're:https?://pics\.indavideo\.hu/videos/.+\.jpg',
'timestamp': 1439193826, 'timestamp': 1439193826,
'upload_date': '20150810', 'upload_date': '20150810',
'duration': 72, 'duration': 72,
'age_limit': 0, 'age_limit': 0,
'tags': ['tánc', 'cica', 'cuki', 'cukiajanlo', 'newsroom'], 'tags': 'count:5',
}, },
}, { }, {
'url': 'https://embed.indavideo.hu/player/video/1bdc3c6d80?autostart=1&hide=1', 'url': 'https://embed.indavideo.hu/player/video/1bdc3c6d80?autostart=1&hide=1',
@ -45,14 +44,30 @@ class IndavideoEmbedIE(InfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'title': 'Vicces cica', 'title': 'Vicces cica',
'description': 'Játszik a tablettel. :D', 'description': 'Játszik a tablettel. :D',
'thumbnail': r're:^https?://.*\.jpg$', 'thumbnail': r're:https?://pics\.indavideo\.hu/videos/.+\.jpg',
'uploader': 'Jet_Pack', 'uploader': 'Jet_Pack',
'uploader_id': '491217', 'uploader_id': '491217',
'timestamp': 1390821212, 'timestamp': 1390821212,
'upload_date': '20140127', 'upload_date': '20140127',
'duration': 7, 'duration': 7,
'age_limit': 0, 'age_limit': 0,
'tags': ['cica', 'Jet_Pack'], 'tags': 'count:2',
},
}, {
'url': 'https://palyazat.indavideo.hu/video/RUSH_1',
'info_dict': {
'id': '3808180',
'ext': 'mp4',
'title': 'RUSH',
'age_limit': 0,
'description': '',
'duration': 650,
'tags': 'count:2',
'thumbnail': r're:https?://pics\.indavideo\.hu/videos/.+\.jpg',
'timestamp': 1729136266,
'upload_date': '20241017',
'uploader': '7summerfilms',
'uploader_id': '1628496',
}, },
}] }]

14
yt_dlp/extractor/joj.py
Unescape Escape View File

@ -22,18 +22,17 @@ class JojIE(InfoExtractor):
'id': 'a388ec4c-6019-4a4a-9312-b1bee194e932', 'id': 'a388ec4c-6019-4a4a-9312-b1bee194e932',
'ext': 'mp4', 'ext': 'mp4',
'title': 'NOVÉ BÝVANIE', 'title': 'NOVÉ BÝVANIE',
'thumbnail': r're:^https?://.*?$',
'duration': 3118, 'duration': 3118,
'thumbnail': r're:https?://img\.joj\.sk/.+',
}, },
}, { }, {
'url': 'https://media.joj.sk/embed/CSM0Na0l0p1', 'url': 'https://media.joj.sk/embed/CSM0Na0l0p1',
'info_dict': { 'info_dict': {
'id': 'CSM0Na0l0p1', 'id': 'CSM0Na0l0p1',
'ext': 'mp4', 'ext': 'mp4',
'height': 576,
'title': 'Extrémne rodiny 2 - POKRAČOVANIE (2012/04/09 21:30:00)', 'title': 'Extrémne rodiny 2 - POKRAČOVANIE (2012/04/09 21:30:00)',
'duration': 3937, 'duration': 3937,
'thumbnail': r're:^https?://.*?$', 'thumbnail': r're:https?://img\.joj\.sk/.+',
}, },
}, { }, {
'url': 'https://media.joj.sk/embed/9i1cxv', 'url': 'https://media.joj.sk/embed/9i1cxv',
@ -45,6 +44,15 @@ class JojIE(InfoExtractor):
'url': 'joj:9i1cxv', 'url': 'joj:9i1cxv',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
# FIXME: Embed detection
'url': 'https://www.noviny.sk/slovensko/238543-slovenskom-sa-prehnala-vlna-silnych-burok',
'info_dict': {
'id': '238543-slovenskom-sa-prehnala-vlna-silnych-burok',
'title': 'Slovenskom sa prehnala vlna silných búrok',
},
'playlist_mincount': 5,
}]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)

26
yt_dlp/extractor/jwplatform.py
Unescape Escape View File

@ -8,7 +8,6 @@ class JWPlatformIE(InfoExtractor):
_VALID_URL = r'(?:https?://(?:content\.jwplatform|cdn\.jwplayer)\.com/(?:(?:feed|player|thumb|preview|manifest)s|jw6|v2/media)/|jwplatform:)(?P<id>[a-zA-Z0-9]{8})' _VALID_URL = r'(?:https?://(?:content\.jwplatform|cdn\.jwplayer)\.com/(?:(?:feed|player|thumb|preview|manifest)s|jw6|v2/media)/|jwplatform:)(?P<id>[a-zA-Z0-9]{8})'
_TESTS = [{ _TESTS = [{
'url': 'http://content.jwplatform.com/players/nPripu9l-ALJ3XQCI.js', 'url': 'http://content.jwplatform.com/players/nPripu9l-ALJ3XQCI.js',
'md5': '3aa16e4f6860e6e78b7df5829519aed3',
'info_dict': { 'info_dict': {
'id': 'nPripu9l', 'id': 'nPripu9l',
'ext': 'mp4', 'ext': 'mp4',
@ -17,13 +16,12 @@ class JWPlatformIE(InfoExtractor):
'upload_date': '20081127', 'upload_date': '20081127',
'timestamp': 1227796140, 'timestamp': 1227796140,
'duration': 32.0, 'duration': 32.0,
'thumbnail': 'https://cdn.jwplayer.com/v2/media/nPripu9l/poster.jpg?width=720', 'thumbnail': r're:https?://cdn\.jwplayer\.com/v2/media/.+',
}, },
}, { }, {
'url': 'https://cdn.jwplayer.com/players/nPripu9l-ALJ3XQCI.js', 'url': 'https://cdn.jwplayer.com/players/nPripu9l-ALJ3XQCI.js',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{ _WEBPAGE_TESTS = [{
# JWPlatform iframe # JWPlatform iframe
'url': 'https://www.covermagazine.co.uk/feature/2465255/business-protection-involved', 'url': 'https://www.covermagazine.co.uk/feature/2465255/business-protection-involved',
@ -33,10 +31,11 @@ class JWPlatformIE(InfoExtractor):
'upload_date': '20160719', 'upload_date': '20160719',
'timestamp': 1468923808, 'timestamp': 1468923808,
'title': '2016_05_18 Cover L&G Business Protection V1 FINAL.mp4', 'title': '2016_05_18 Cover L&G Business Protection V1 FINAL.mp4',
'thumbnail': 'https://cdn.jwplayer.com/v2/media/AG26UQXM/poster.jpg?width=720', 'thumbnail': r're:https?://cdn\.jwplayer\.com/v2/media/.+',
'description': '', 'description': '',
'duration': 294.0, 'duration': 294.0,
}, },
'skip': 'Site no longer embeds JWPlatform',
}, { }, {
# Player url not surrounded by quotes # Player url not surrounded by quotes
'url': 'https://www.deutsche-kinemathek.de/en/online/streaming/school-trip', 'url': 'https://www.deutsche-kinemathek.de/en/online/streaming/school-trip',
@ -45,12 +44,12 @@ class JWPlatformIE(InfoExtractor):
'title': 'Klassenfahrt', 'title': 'Klassenfahrt',
'ext': 'mp4', 'ext': 'mp4',
'upload_date': '20230109', 'upload_date': '20230109',
'thumbnail': 'https://cdn.jwplayer.com/v2/media/jUxh5uin/poster.jpg?width=720', 'thumbnail': r're:https?://cdn\.jwplayer\.com/v2/media/.+',
'timestamp': 1673270298, 'timestamp': 1673270298,
'description': '', 'description': '',
'duration': 5193.0, 'duration': 5193.0,
}, },
'params': {'allowed_extractors': ['generic', 'jwplatform']}, 'skip': 'Site no longer embeds JWPlatform',
}, { }, {
# iframe src attribute includes backslash before URL string # iframe src attribute includes backslash before URL string
'url': 'https://www.elespectador.com/colombia/video-asi-se-evito-la-fuga-de-john-poulos-presunto-feminicida-de-valentina-trespalacios-explicacion', 'url': 'https://www.elespectador.com/colombia/video-asi-se-evito-la-fuga-de-john-poulos-presunto-feminicida-de-valentina-trespalacios-explicacion',
@ -59,11 +58,24 @@ class JWPlatformIE(InfoExtractor):
'title': 'Así se evitó la fuga de John Poulos, presunto feminicida de Valentina Trespalacios', 'title': 'Así se evitó la fuga de John Poulos, presunto feminicida de Valentina Trespalacios',
'ext': 'mp4', 'ext': 'mp4',
'upload_date': '20230127', 'upload_date': '20230127',
'thumbnail': 'https://cdn.jwplayer.com/v2/media/QD3gsexj/poster.jpg?width=720', 'thumbnail': r're:https?://cdn\.jwplayer\.com/v2/media/.+',
'timestamp': 1674862986, 'timestamp': 1674862986,
'description': 'md5:128fd74591c4e1fc2da598c5cb6f5ce4', 'description': 'md5:128fd74591c4e1fc2da598c5cb6f5ce4',
'duration': 263.0, 'duration': 263.0,
}, },
}, {
'url': 'https://www.skimag.com/video/ski-people-1980',
'info_dict': {
'id': 'YTmgRiNU',
'ext': 'mp4',
'title': 'Ski People (1980)',
'channel': 'snow',
'description': 'md5:cf9c3d101452c91e141f292b19fe4843',
'duration': 5688.0,
'thumbnail': r're:https?://cdn\.jwplayer\.com/v2/media/.+',
'timestamp': 1610407738,
'upload_date': '20210111',
},
}] }]
@classmethod @classmethod

305
yt_dlp/extractor/kaltura.py
Unescape Escape View File

@ -41,149 +41,188 @@ class KalturaIE(InfoExtractor):
2: 'ttml', 2: 'ttml',
3: 'vtt', 3: 'vtt',
} }
_TESTS = [ _TESTS = [{
{ 'url': 'kaltura:269692:1_1jc2y3e4',
'url': 'kaltura:269692:1_1jc2y3e4', 'md5': '3adcbdb3dcc02d647539e53f284ba171',
'md5': '3adcbdb3dcc02d647539e53f284ba171', 'info_dict': {
'info_dict': { 'id': '1_1jc2y3e4',
'id': '1_1jc2y3e4', 'ext': 'mp4',
'ext': 'mp4', 'title': 'Straight from the Heart',
'title': 'Straight from the Heart', 'upload_date': '20131219',
'upload_date': '20131219', 'uploader_id': 'mlundberg@wolfgangsvault.com',
'uploader_id': 'mlundberg@wolfgangsvault.com', 'description': 'The Allman Brothers Band, 12/16/1981',
'description': 'The Allman Brothers Band, 12/16/1981', 'thumbnail': r're:https?://.+/thumbnail/.+',
'thumbnail': 're:^https?://.*/thumbnail/.*', 'timestamp': int,
'timestamp': int,
},
'skip': 'The access to this service is forbidden since the specified partner is blocked',
},
{
'url': 'http://www.kaltura.com/index.php/kwidget/cache_st/1300318621/wid/_269692/uiconf_id/3873291/entry_id/1_1jc2y3e4',
'only_matching': True,
}, },
{ 'skip': 'The access to this service is forbidden since the specified partner is blocked',
'url': 'https://cdnapisec.kaltura.com/index.php/kwidget/wid/_557781/uiconf_id/22845202/entry_id/1_plr1syf3', }, {
'only_matching': True, 'url': 'http://www.kaltura.com/index.php/kwidget/cache_st/1300318621/wid/_269692/uiconf_id/3873291/entry_id/1_1jc2y3e4',
'only_matching': True,
}, {
'url': 'https://cdnapisec.kaltura.com/index.php/kwidget/wid/_557781/uiconf_id/22845202/entry_id/1_plr1syf3',
'only_matching': True,
}, {
'url': 'https://cdnapisec.kaltura.com/html5/html5lib/v2.30.2/mwEmbedFrame.php/p/1337/uiconf_id/20540612/entry_id/1_sf5ovm7u?wid=_243342',
'only_matching': True,
}, {
# video with subtitles
'url': 'kaltura:111032:1_cw786r8q',
'only_matching': True,
}, {
# video with ttml subtitles (no fileExt)
'url': 'kaltura:1926081:0_l5ye1133',
'info_dict': {
'id': '0_l5ye1133',
'ext': 'mp4',
'title': 'What Can You Do With Python?',
'upload_date': '20160221',
'uploader_id': 'stork',
'thumbnail': r're:https?://.+/thumbnail/.+',
'timestamp': int,
'subtitles': {
'en': [{
'ext': 'ttml',
}],
},
}, },
{ 'skip': 'Gone. Maybe https://www.safaribooksonline.com/library/tutorials/introduction-to-python-anon/3469/',
'url': 'https://cdnapisec.kaltura.com/html5/html5lib/v2.30.2/mwEmbedFrame.php/p/1337/uiconf_id/20540612/entry_id/1_sf5ovm7u?wid=_243342', 'params': {'skip_download': True},
'only_matching': True, }, {
'url': 'https://www.kaltura.com/index.php/extwidget/preview/partner_id/1770401/uiconf_id/37307382/entry_id/0_58u8kme7/embed/iframe?&flashvars[streamerType]=auto',
'only_matching': True,
}, {
'url': 'https://www.kaltura.com:443/index.php/extwidget/preview/partner_id/1770401/uiconf_id/37307382/entry_id/0_58u8kme7/embed/iframe?&flashvars[streamerType]=auto',
'only_matching': True,
}, {
# unavailable source format
'url': 'kaltura:513551:1_66x4rg7o',
'only_matching': True,
}, {
# html5lib URL using kwidget player
'url': 'https://cdnapisec.kaltura.com/html5/html5lib/v2.46/mwEmbedFrame.php/p/691292/uiconf_id/20499062/entry_id/0_c076mna6?wid=_691292&iframeembed=true&playerId=kaltura_player_1420508608&entry_id=0_c076mna6&flashvars%5BakamaiHD.loadingPolicy%5D=preInitialize&flashvars%5BakamaiHD.asyncInit%5D=true&flashvars%5BstreamerType%5D=hdnetwork',
'info_dict': {
'id': '0_c076mna6',
'ext': 'mp4',
'title': 'md5:4883e7acbcbf42583a2dddc97dee4855',
'duration': 3608,
'uploader_id': 'commons@swinburne.edu.au',
'timestamp': 1408086874,
'view_count': int,
'upload_date': '20140815',
'thumbnail': r're:https?://cfvod\.kaltura\.com/.+',
}, },
{ }, {
# video with subtitles # html5lib playlist URL using kwidget player
'url': 'kaltura:111032:1_cw786r8q', 'url': 'https://cdnapisec.kaltura.com/html5/html5lib/v2.89/mwEmbedFrame.php/p/2019031/uiconf_id/40436601?wid=1_4j3m32cv&iframeembed=true&playerId=kaltura_player_&flashvars[playlistAPI.kpl0Id]=1_jovey5nu&flashvars[ks]=&&flashvars[imageDefaultDuration]=30&flashvars[localizationCode]=en&flashvars[leadWithHTML5]=true&flashvars[forceMobileHTML5]=true&flashvars[nextPrevBtn.plugin]=true&flashvars[hotspots.plugin]=true&flashvars[sideBarContainer.plugin]=true&flashvars[sideBarContainer.position]=left&flashvars[sideBarContainer.clickToClose]=true&flashvars[chapters.plugin]=true&flashvars[chapters.layout]=vertical&flashvars[chapters.thumbnailRotator]=false&flashvars[streamSelector.plugin]=true&flashvars[EmbedPlayer.SpinnerTarget]=videoHolder&flashvars[dualScreen.plugin]=true&flashvars[playlistAPI.playlistUrl]=https://canvasgatechtest.kaf.kaltura.com/playlist/details/{playlistAPI.kpl0Id}/categoryid/126428551',
'only_matching': True, 'info_dict': {
'id': '1_jovey5nu',
'title': '00-00 Introduction',
}, },
{ 'playlist': [
# video with ttml subtitles (no fileExt) {
'url': 'kaltura:1926081:0_l5ye1133', 'info_dict': {
'info_dict': { 'id': '1_b1y5hlvx',
'id': '0_l5ye1133', 'ext': 'mp4',
'ext': 'mp4', 'title': 'CS7646_00-00 Introductio_Introduction',
'title': 'What Can You Do With Python?', 'duration': 91,
'upload_date': '20160221', 'thumbnail': r're:https?://cfvod\.kaltura\.com/.+',
'uploader_id': 'stork', 'view_count': int,
'thumbnail': 're:^https?://.*/thumbnail/.*', 'timestamp': 1533154447,
'timestamp': int, 'upload_date': '20180801',
'subtitles': { 'uploader_id': 'djoyner3',
'en': [{ },
'ext': 'ttml', }, {
}], 'info_dict': {
'id': '1_jfb7mdpn',
'ext': 'mp4',
'title': 'CS7646_00-00 Introductio_Three parts to the course',
'duration': 63,
'thumbnail': r're:https?://cfvod\.kaltura\.com/.+',
'view_count': int,
'timestamp': 1533154489,
'upload_date': '20180801',
'uploader_id': 'djoyner3',
},
}, {
'info_dict': {
'id': '1_8xflxdp7',
'ext': 'mp4',
'title': 'CS7646_00-00 Introductio_Textbooks',
'duration': 37,
'thumbnail': r're:https?://cfvod\.kaltura\.com/.+',
'view_count': int,
'timestamp': 1533154512,
'upload_date': '20180801',
'uploader_id': 'djoyner3',
},
}, {
'info_dict': {
'id': '1_3hqew8kn',
'ext': 'mp4',
'title': 'CS7646_00-00 Introductio_Prerequisites',
'duration': 49,
'thumbnail': r're:https?://cfvod\.kaltura\.com/.+',
'view_count': int,
'timestamp': 1533154536,
'upload_date': '20180801',
'uploader_id': 'djoyner3',
}, },
}, },
'skip': 'Gone. Maybe https://www.safaribooksonline.com/library/tutorials/introduction-to-python-anon/3469/', ],
'params': { }]
'skip_download': True, _WEBPAGE_TESTS = [{
}, 'url': 'https://www.cornell.edu/VIDEO/nima-arkani-hamed-standard-models-of-particle-physics',
}, 'info_dict': {
{ 'id': '1_sgtvehim',
'url': 'https://www.kaltura.com/index.php/extwidget/preview/partner_id/1770401/uiconf_id/37307382/entry_id/0_58u8kme7/embed/iframe?&flashvars[streamerType]=auto', 'ext': 'mp4',
'only_matching': True, 'title': 'Our "Standard Models" of particle physics and cosmology',
}, 'duration': 5420,
{ 'thumbnail': r're:https?://cdnsecakmi\.kaltura\.com/.+',
'url': 'https://www.kaltura.com:443/index.php/extwidget/preview/partner_id/1770401/uiconf_id/37307382/entry_id/0_58u8kme7/embed/iframe?&flashvars[streamerType]=auto', 'timestamp': 1321158993,
'only_matching': True, 'upload_date': '20111113',
'uploader_id': 'kps1',
'view_count': int,
}, },
{ }, {
# unavailable source format 'url': 'https://www.oreilly.com/ideas/my-cloud-makes-pretty-pictures',
'url': 'kaltura:513551:1_66x4rg7o', 'info_dict': {
'only_matching': True, 'id': '0_utuok90b',
'ext': 'mp4',
'title': '06_matthew_brender_raj_dutt',
'duration': 331,
'thumbnail': r're:https?://cfvod\.kaltura\.com/.+',
'timestamp': 1466638791,
'upload_date': '20160622',
'uploader_id': '',
'view_count': int,
}, },
{ }, {
# html5lib URL using kwidget player 'url': 'https://fod.infobase.com/p_ViewPlaylist.aspx?AssignmentID=NUN8ZY',
'url': 'https://cdnapisec.kaltura.com/html5/html5lib/v2.46/mwEmbedFrame.php/p/691292/uiconf_id/20499062/entry_id/0_c076mna6?wid=_691292&iframeembed=true&playerId=kaltura_player_1420508608&entry_id=0_c076mna6&flashvars%5BakamaiHD.loadingPolicy%5D=preInitialize&flashvars%5BakamaiHD.asyncInit%5D=true&flashvars%5BstreamerType%5D=hdnetwork', 'info_dict': {
'info_dict': { 'id': '0_izeg5utt',
'id': '0_c076mna6', 'ext': 'mp4',
'ext': 'mp4', 'title': '35871',
'title': 'md5:4883e7acbcbf42583a2dddc97dee4855', 'duration': 3403,
'duration': 3608, 'thumbnail': r're:https?://cfvod\.kaltura\.com/.+',
'uploader_id': 'commons@swinburne.edu.au', 'timestamp': 1355743100,
'timestamp': 1408086874, 'upload_date': '20121217',
'view_count': int, 'uploader_id': 'cplapp@learn360.com',
'upload_date': '20140815', 'view_count': int,
'thumbnail': 'http://cfvod.kaltura.com/p/691292/sp/69129200/thumbnail/entry_id/0_c076mna6/version/100022',
},
}, },
{ }, {
# html5lib playlist URL using kwidget player 'url': 'https://www.cns.nyu.edu/~eero/math-tools17/Videos/lecture-05sep2017.html',
'url': 'https://cdnapisec.kaltura.com/html5/html5lib/v2.89/mwEmbedFrame.php/p/2019031/uiconf_id/40436601?wid=1_4j3m32cv&iframeembed=true&playerId=kaltura_player_&flashvars[playlistAPI.kpl0Id]=1_jovey5nu&flashvars[ks]=&&flashvars[imageDefaultDuration]=30&flashvars[localizationCode]=en&flashvars[leadWithHTML5]=true&flashvars[forceMobileHTML5]=true&flashvars[nextPrevBtn.plugin]=true&flashvars[hotspots.plugin]=true&flashvars[sideBarContainer.plugin]=true&flashvars[sideBarContainer.position]=left&flashvars[sideBarContainer.clickToClose]=true&flashvars[chapters.plugin]=true&flashvars[chapters.layout]=vertical&flashvars[chapters.thumbnailRotator]=false&flashvars[streamSelector.plugin]=true&flashvars[EmbedPlayer.SpinnerTarget]=videoHolder&flashvars[dualScreen.plugin]=true&flashvars[playlistAPI.playlistUrl]=https://canvasgatechtest.kaf.kaltura.com/playlist/details/{playlistAPI.kpl0Id}/categoryid/126428551', 'info_dict': {
'info_dict': { 'id': '1_9gzouybz',
'id': '1_jovey5nu', 'ext': 'mp4',
'title': '00-00 Introduction', 'title': 'lecture-05sep2017',
}, 'duration': 7219,
'playlist': [ 'thumbnail': r're:https?://cfvod\.kaltura\.com/.+',
{ 'timestamp': 1505340777,
'info_dict': { 'upload_date': '20170913',
'id': '1_b1y5hlvx', 'uploader_id': 'eps2',
'ext': 'mp4', 'view_count': int,
'title': 'CS7646_00-00 Introductio_Introduction',
'duration': 91,
'thumbnail': 'http://cfvod.kaltura.com/p/2019031/sp/201903100/thumbnail/entry_id/1_b1y5hlvx/version/100001',
'view_count': int,
'timestamp': 1533154447,
'upload_date': '20180801',
'uploader_id': 'djoyner3',
},
}, {
'info_dict': {
'id': '1_jfb7mdpn',
'ext': 'mp4',
'title': 'CS7646_00-00 Introductio_Three parts to the course',
'duration': 63,
'thumbnail': 'http://cfvod.kaltura.com/p/2019031/sp/201903100/thumbnail/entry_id/1_jfb7mdpn/version/100001',
'view_count': int,
'timestamp': 1533154489,
'upload_date': '20180801',
'uploader_id': 'djoyner3',
},
}, {
'info_dict': {
'id': '1_8xflxdp7',
'ext': 'mp4',
'title': 'CS7646_00-00 Introductio_Textbooks',
'duration': 37,
'thumbnail': 'http://cfvod.kaltura.com/p/2019031/sp/201903100/thumbnail/entry_id/1_8xflxdp7/version/100001',
'view_count': int,
'timestamp': 1533154512,
'upload_date': '20180801',
'uploader_id': 'djoyner3',
},
}, {
'info_dict': {
'id': '1_3hqew8kn',
'ext': 'mp4',
'title': 'CS7646_00-00 Introductio_Prerequisites',
'duration': 49,
'thumbnail': 'http://cfvod.kaltura.com/p/2019031/sp/201903100/thumbnail/entry_id/1_3hqew8kn/version/100001',
'view_count': int,
'timestamp': 1533154536,
'upload_date': '20180801',
'uploader_id': 'djoyner3',
},
},
],
}, },
] }]
@classmethod @classmethod
def _extract_embed_urls(cls, url, webpage): def _extract_embed_urls(cls, url, webpage):

9
yt_dlp/extractor/kinja.py
Unescape Escape View File

@ -89,6 +89,15 @@ class KinjaEmbedIE(InfoExtractor):
'url': 'https://kinja.com/ajax/inset/iframe?id=youtube-video-00QyL0AgPAE', 'url': 'https://kinja.com/ajax/inset/iframe?id=youtube-video-00QyL0AgPAE',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
'url': 'http://www.clickhole.com/video/dont-understand-bitcoin-man-will-mumble-explanatio-2537',
'info_dict': {
'id': '106351',
'ext': 'mp4',
'title': 'Don’t Understand Bitcoin? This Man Will Mumble An Explanation At You',
},
'skip': 'Invalid URL',
}]
_JWPLATFORM_PROVIDER = ('cdn.jwplayer.com/v2/media/', 'JWPlatform') _JWPLATFORM_PROVIDER = ('cdn.jwplayer.com/v2/media/', 'JWPlatform')
_PROVIDER_MAP = { _PROVIDER_MAP = {
'fb': ('facebook.com/video.php?v=', 'Facebook'), 'fb': ('facebook.com/video.php?v=', 'Facebook'),

34
yt_dlp/extractor/libsyn.py
Unescape Escape View File

@ -18,12 +18,10 @@ class LibsynIE(InfoExtractor):
'info_dict': { 'info_dict': {
'id': '6385796', 'id': '6385796',
'ext': 'mp3', 'ext': 'mp3',
'title': 'Champion Minded - Developing a Growth Mindset', 'title': 'The Allistair McCaw Podcast - Developing a Growth Mindset',
# description fetched using another request: 'duration': 834.0,
# http://html5-player.libsyn.com/embed/getitemdetails?item_id=6385796 'thumbnail': r're:https?://assets\.libsyn\.com/.+',
# 'description': 'In this episode, Allistair talks about the importance of developing a growth mindset, not only in sports, but in life too.',
'upload_date': '20180320', 'upload_date': '20180320',
'thumbnail': 're:^https?://.*',
}, },
}, { }, {
'url': 'https://html5-player.libsyn.com/embed/episode/id/3727166/height/75/width/200/theme/standard/direction/no/autoplay/no/autonext/no/thumbnail/no/preload/no/no_addthis/no/', 'url': 'https://html5-player.libsyn.com/embed/episode/id/3727166/height/75/width/200/theme/standard/direction/no/autoplay/no/autonext/no/thumbnail/no/preload/no/no_addthis/no/',
@ -32,8 +30,32 @@ class LibsynIE(InfoExtractor):
'id': '3727166', 'id': '3727166',
'ext': 'mp3', 'ext': 'mp3',
'title': 'Clients From Hell Podcast - How a Sex Toy Company Kickstarted my Freelance Career', 'title': 'Clients From Hell Podcast - How a Sex Toy Company Kickstarted my Freelance Career',
'thumbnail': r're:https?://assets\.libsyn\.com/.+',
'upload_date': '20150818', 'upload_date': '20150818',
'thumbnail': 're:^https?://.*', },
'skip': 'Invalid URL',
}]
_WEBPAGE_TESTS = [{
'url': 'https://html5-player.libsyn.com/',
'md5': '50cff329596b8f674d4449ed077ef2f9',
'info_dict': {
'id': '2378831',
'ext': 'mp3',
'title': 'md5:54108b15f98e1b4056612c10b50106b2',
'duration': 3561.0,
'thumbnail': r're:https?://assets\.libsyn\.com/.+',
'upload_date': '20130630',
},
}, {
'url': 'https://undergroundwellness.com/podcasts/306-5-steps-to-permanent-gut-healing/',
'md5': '23576952577f9604520a730d90371761',
'info_dict': {
'id': '3793998',
'ext': 'mp3',
'title': 'Underground Wellness Radio - Jack Tips: 5 Steps to Permanent Gut Healing',
'duration': 3989.0,
'thumbnail': r're:https?://assets\.libsyn\.com/.+',
'upload_date': '20141126',
}, },
}] }]

180
yt_dlp/extractor/mainstreaming.py
Unescape Escape View File

@ -16,91 +16,103 @@ class MainStreamingIE(InfoExtractor):
_EMBED_REGEX = [rf'<iframe[^>]+?src=["\']?(?P<url>{_VALID_URL})["\']?'] _EMBED_REGEX = [rf'<iframe[^>]+?src=["\']?(?P<url>{_VALID_URL})["\']?']
IE_DESC = 'MainStreaming Player' IE_DESC = 'MainStreaming Player'
_TESTS = [ _TESTS = [{
{ # Live stream offline, has alternative content id
# Live stream offline, has alternative content id 'url': 'https://webtools-e18da6642b684f8aa9ae449862783a56.msvdn.net/embed/53EN6GxbWaJC',
'url': 'https://webtools-e18da6642b684f8aa9ae449862783a56.msvdn.net/embed/53EN6GxbWaJC', 'info_dict': {
'info_dict': { 'id': '53EN6GxbWaJC',
'id': '53EN6GxbWaJC', 'title': 'Diretta homepage 2021-12-31 12:00',
'title': 'Diretta homepage 2021-12-31 12:00', 'description': '',
'description': '', 'live_status': 'was_live',
'live_status': 'was_live', 'ext': 'mp4',
'ext': 'mp4', 'thumbnail': r're:https?://[\w-]+\.msvdn\.net/image/\w+/poster',
'thumbnail': r're:https?://[A-Za-z0-9-]*\.msvdn.net/image/\w+/poster',
},
'expected_warnings': [
'Ignoring alternative content ID: WDAF1KOWUpH3',
'MainStreaming said: Live event is OFFLINE',
],
'skip': 'live stream offline',
}, {
# playlist
'url': 'https://webtools-e18da6642b684f8aa9ae449862783a56.msvdn.net/embed/WDAF1KOWUpH3',
'info_dict': {
'id': 'WDAF1KOWUpH3',
'title': 'Playlist homepage',
},
'playlist_mincount': 2,
}, {
# livestream
'url': 'https://webtools-859c1818ed614cc5b0047439470927b0.msvdn.net/embed/tDoFkZD3T1Lw',
'info_dict': {
'id': 'tDoFkZD3T1Lw',
'title': r're:Class CNBC Live \d{4}-\d{2}-\d{2} \d{2}:\d{2}$',
'live_status': 'is_live',
'ext': 'mp4',
'thumbnail': r're:https?://[A-Za-z0-9-]*\.msvdn.net/image/\w+/poster',
},
'skip': 'live stream',
}, {
'url': 'https://webtools-f5842579ff984c1c98d63b8d789673eb.msvdn.net/embed/EUlZfGWkGpOd?autoPlay=false',
'info_dict': {
'id': 'EUlZfGWkGpOd',
'title': 'La Settimana ',
'description': '03 Ottobre ore 02:00',
'ext': 'mp4',
'live_status': 'not_live',
'thumbnail': r're:https?://[A-Za-z0-9-]*\.msvdn.net/image/\w+/poster',
'duration': 1512,
},
}, {
# video without webtools- prefix
'url': 'https://f5842579ff984c1c98d63b8d789673eb.msvdn.net/embed/MfuWmzL2lGkA?autoplay=false&T=1635860445',
'info_dict': {
'id': 'MfuWmzL2lGkA',
'title': 'TG Mattina',
'description': '06 Ottobre ore 08:00',
'ext': 'mp4',
'live_status': 'not_live',
'thumbnail': r're:https?://[A-Za-z0-9-]*\.msvdn.net/image/\w+/poster',
'duration': 789.04,
},
}, {
# always-on livestream with DVR
'url': 'https://webtools-f5842579ff984c1c98d63b8d789673eb.msvdn.net/embed/HVvPMzy',
'info_dict': {
'id': 'HVvPMzy',
'title': r're:^Diretta LaC News24 \d{4}-\d{2}-\d{2} \d{2}:\d{2}$',
'description': 'canale all news',
'live_status': 'is_live',
'ext': 'mp4',
'thumbnail': r're:https?://[A-Za-z0-9-]*\.msvdn.net/image/\w+/poster',
},
'params': {
'skip_download': True,
},
}, {
# no host
'url': 'https://webtools.msvdn.net/embed/MfuWmzL2lGkA',
'only_matching': True,
}, {
'url': 'https://859c1818ed614cc5b0047439470927b0.msvdn.net/amp_embed/tDoFkZD3T1Lw',
'only_matching': True,
}, {
'url': 'https://859c1818ed614cc5b0047439470927b0.msvdn.net/content/tDoFkZD3T1Lw#',
'only_matching': True,
}, },
] 'expected_warnings': [
'Ignoring alternative content ID: WDAF1KOWUpH3',
'MainStreaming said: Live event is OFFLINE',
],
'skip': 'live stream offline',
}, {
# playlist
'url': 'https://webtools-e18da6642b684f8aa9ae449862783a56.msvdn.net/embed/WDAF1KOWUpH3',
'info_dict': {
'id': 'WDAF1KOWUpH3',
'title': 'Playlist homepage',
},
'playlist_mincount': 2,
}, {
# livestream
'url': 'https://webtools-859c1818ed614cc5b0047439470927b0.msvdn.net/embed/tDoFkZD3T1Lw',
'info_dict': {
'id': 'tDoFkZD3T1Lw',
'title': str,
'live_status': 'is_live',
'ext': 'mp4',
'thumbnail': r're:https?://[\w-]+\.msvdn\.net/image/\w+/poster',
},
'skip': 'live stream',
}, {
'url': 'https://webtools-f5842579ff984c1c98d63b8d789673eb.msvdn.net/embed/EUlZfGWkGpOd?autoPlay=false',
'info_dict': {
'id': 'EUlZfGWkGpOd',
'title': 'La Settimana ',
'description': '03 Ottobre ore 02:00',
'ext': 'mp4',
'live_status': 'not_live',
'thumbnail': r're:https?://[\w-]+\.msvdn\.net/image/\w+/poster',
'duration': 1512,
},
'skip': 'Invalid URL',
}, {
# video without webtools- prefix
'url': 'https://f5842579ff984c1c98d63b8d789673eb.msvdn.net/embed/MfuWmzL2lGkA?autoplay=false&T=1635860445',
'info_dict': {
'id': 'MfuWmzL2lGkA',
'title': 'TG Mattina',
'description': '06 Ottobre ore 08:00',
'ext': 'mp4',
'live_status': 'not_live',
'thumbnail': r're:https?://[\w-]+\.msvdn\.net/image/\w+/poster',
'duration': 789.04,
},
'skip': 'Invalid URL',
}, {
# always-on livestream with DVR
'url': 'https://webtools-f5842579ff984c1c98d63b8d789673eb.msvdn.net/embed/HVvPMzy',
'info_dict': {
'id': 'HVvPMzy',
'title': str,
'description': 'canale all news',
'live_status': 'is_live',
'ext': 'mp4',
'thumbnail': r're:https?://[\w-]+\.msvdn\.net/image/\w+/poster',
},
'params': {'skip_download': 'm3u8'},
}, {
# no host
'url': 'https://webtools.msvdn.net/embed/MfuWmzL2lGkA',
'only_matching': True,
}, {
'url': 'https://859c1818ed614cc5b0047439470927b0.msvdn.net/amp_embed/tDoFkZD3T1Lw',
'only_matching': True,
}, {
'url': 'https://859c1818ed614cc5b0047439470927b0.msvdn.net/content/tDoFkZD3T1Lw#',
'only_matching': True,
}]
_WEBPAGE_TESTS = [{
# FIXME: Embed detection
'url': 'https://www.lacplay.it/video/in-evidenza_728/lac-storie-p-250-i-santi-pietro-e-paolo_77297/',
'info_dict': {
'id': 'u7kiX5DUaHYr',
'ext': 'mp4',
'title': 'I Santi Pietro e Paolo',
'description': 'md5:ff6be24916ba6b9ae990bf5f3df4911e',
'duration': 1700.0,
'thumbnail': r're:https?://.+',
'tags': '06/07/2025',
'live_status': 'not_live',
},
}]
def _playlist_entries(self, host, playlist_content): def _playlist_entries(self, host, playlist_content):
for entry in playlist_content: for entry in playlist_content:

14
yt_dlp/extractor/medialaan.py
Unescape Escape View File

@ -42,6 +42,7 @@ class MedialaanIE(InfoExtractor):
'id': '193993', 'id': '193993',
'ext': 'mp4', 'ext': 'mp4',
'title': 'De terugkeer van Ally de Aap en wie vertrekt er nog bij NAC?', 'title': 'De terugkeer van Ally de Aap en wie vertrekt er nog bij NAC?',
'thumbnail': r're:https?://images\.mychannels\.video/imgix/.+',
'timestamp': 1611663540, 'timestamp': 1611663540,
'upload_date': '20210126', 'upload_date': '20210126',
'duration': 238, 'duration': 238,
@ -68,6 +69,19 @@ class MedialaanIE(InfoExtractor):
'url': 'https://embed.mychannels.video/embed/193993', 'url': 'https://embed.mychannels.video/embed/193993',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
'url': 'https://www.demorgen.be/snelnieuws/tom-waes-promoot-alcoholtesten-op-werchter-ik-ben-de-laatste-persoon-die-met-de-vinger-moet-wijzen~b7457c0d/',
'info_dict': {
'id': '1576607',
'ext': 'mp4',
'title': 'Tom Waes blaastest',
'duration': 62,
'thumbnail': r're:https?://video-images\.persgroep\.be/aws_generated.+\.jpg',
'timestamp': 1751730795,
'upload_date': '20250705',
},
'params': {'extractor_args': {'generic': {'impersonate': ['chrome']}}},
}]
@classmethod @classmethod
def _extract_embed_urls(cls, url, webpage): def _extract_embed_urls(cls, url, webpage):

22
yt_dlp/extractor/megatvcom.py
Unescape Escape View File

@ -31,10 +31,9 @@ class MegaTVComIE(MegaTVComBaseIE):
IE_NAME = 'megatvcom' IE_NAME = 'megatvcom'
IE_DESC = 'megatv.com videos' IE_DESC = 'megatv.com videos'
_VALID_URL = r'https?://(?:www\.)?megatv\.com/(?:\d{4}/\d{2}/\d{2}|[^/]+/(?P<id>\d+))/(?P<slug>[^/]+)' _VALID_URL = r'https?://(?:www\.)?megatv\.com/(?:\d{4}/\d{2}/\d{2}|[^/]+/(?P<id>\d+))/(?P<slug>[^/]+)'
_TESTS = [{ _TESTS = [{
# FIXME: Unable to extract article id
'url': 'https://www.megatv.com/2021/10/23/egkainia-gia-ti-nea-skini-omega-tou-dimotikou-theatrou-peiraia/', 'url': 'https://www.megatv.com/2021/10/23/egkainia-gia-ti-nea-skini-omega-tou-dimotikou-theatrou-peiraia/',
'md5': '6546a1a37fff0dd51c9dce5f490b7d7d',
'info_dict': { 'info_dict': {
'id': '520979', 'id': '520979',
'ext': 'mp4', 'ext': 'mp4',
@ -43,20 +42,19 @@ class MegaTVComIE(MegaTVComBaseIE):
'timestamp': 1634975747, 'timestamp': 1634975747,
'upload_date': '20211023', 'upload_date': '20211023',
'display_id': 'egkainia-gia-ti-nea-skini-omega-tou-dimotikou-theatrou-peiraia', 'display_id': 'egkainia-gia-ti-nea-skini-omega-tou-dimotikou-theatrou-peiraia',
'thumbnail': 'https://www.megatv.com/wp-content/uploads/2021/10/ΠΕΙΡΑΙΑΣ-1024x450.jpg', 'thumbnail': r're:https?://www\.megatv\.com/wp-content/uploads/.+\.jpg',
}, },
}, { }, {
'url': 'https://www.megatv.com/tvshows/527800/epeisodio-65-12/', 'url': 'https://www.megatv.com/tvshows/527800/epeisodio-65-12/',
'md5': 'cba2085d45c1abeb8e7e9b7e1d6c0072',
'info_dict': { 'info_dict': {
'id': '527800', 'id': '527800',
'ext': 'mp4', 'ext': 'mp4',
'title': 'md5:fc322cb51f682eecfe2f54cd5ab3a157', 'title': 'Η Γη της Ελιάς: Επεισόδιο 65 - A\' ΚΥΚΛΟΣ ',
'description': 'md5:b2b7ed3690a78f2a0156eb790fdc00df', 'description': 'md5:b2b7ed3690a78f2a0156eb790fdc00df',
'timestamp': 1636048859, 'timestamp': 1636048859,
'upload_date': '20211104', 'upload_date': '20211104',
'display_id': 'epeisodio-65-12', 'display_id': 'epeisodio-65-12',
'thumbnail': 'https://www.megatv.com/wp-content/uploads/2021/11/16-1-1.jpg', 'thumbnail': r're:https?://www\.megatv\.com/wp-content/uploads/.+\.jpg',
}, },
}] }]
@ -104,8 +102,8 @@ class MegaTVComEmbedIE(MegaTVComBaseIE):
IE_DESC = 'megatv.com embedded videos' IE_DESC = 'megatv.com embedded videos'
_VALID_URL = r'(?:https?:)?//(?:www\.)?megatv\.com/embed/?\?p=(?P<id>\d+)' _VALID_URL = r'(?:https?:)?//(?:www\.)?megatv\.com/embed/?\?p=(?P<id>\d+)'
_EMBED_REGEX = [rf'''<iframe[^>]+?src=(?P<_q1>["'])(?P<url>{_VALID_URL})(?P=_q1)'''] _EMBED_REGEX = [rf'''<iframe[^>]+?src=(?P<_q1>["'])(?P<url>{_VALID_URL})(?P=_q1)''']
_TESTS = [{ _TESTS = [{
# FIXME: Unable to extract article id
'url': 'https://www.megatv.com/embed/?p=2020520979', 'url': 'https://www.megatv.com/embed/?p=2020520979',
'md5': '6546a1a37fff0dd51c9dce5f490b7d7d', 'md5': '6546a1a37fff0dd51c9dce5f490b7d7d',
'info_dict': { 'info_dict': {
@ -119,6 +117,7 @@ class MegaTVComEmbedIE(MegaTVComBaseIE):
'thumbnail': 'https://www.megatv.com/wp-content/uploads/2021/10/ΠΕΙΡΑΙΑΣ-1024x450.jpg', 'thumbnail': 'https://www.megatv.com/wp-content/uploads/2021/10/ΠΕΙΡΑΙΑΣ-1024x450.jpg',
}, },
}, { }, {
# FIXME: Unable to extract article id
'url': 'https://www.megatv.com/embed/?p=2020534081', 'url': 'https://www.megatv.com/embed/?p=2020534081',
'md5': '6ac8b3ce4dc6120c802f780a1e6b3812', 'md5': '6ac8b3ce4dc6120c802f780a1e6b3812',
'info_dict': { 'info_dict': {
@ -132,6 +131,15 @@ class MegaTVComEmbedIE(MegaTVComBaseIE):
'thumbnail': 'https://www.megatv.com/wp-content/uploads/2021/11/Capture-266.jpg', 'thumbnail': 'https://www.megatv.com/wp-content/uploads/2021/11/Capture-266.jpg',
}, },
}] }]
_WEBPAGE_TESTS = [{
# FIXME: Unable to extract article id
'url': 'https://www.in.gr/2021/12/18/greece/apokalypsi-mega-poios-parelave-tin-ereyna-tsiodra-ek-merous-tis-kyvernisis-o-prothypourgos-telika-gnorize/',
'info_dict': {
'id': 'apokalypsi-mega-poios-parelave-tin-ereyna-tsiodra-ek-merous-tis-kyvernisis-o-prothypourgos-telika-gnorize',
'title': 'md5:5e569cf996ec111057c2764ec272848f',
},
'playlist_count': 2,
}]
def _match_canonical_url(self, webpage): def _match_canonical_url(self, webpage):
LINK_RE = r'''(?x) LINK_RE = r'''(?x)

161
yt_dlp/extractor/mlb.py
Unescape Escape View File

@ -105,89 +105,85 @@ class MLBIE(MLBBaseIE):
r'<iframe[^>]+?src=(["\'])(?P<url>https?://m(?:lb)?\.mlb\.com/shared/video/embed/embed\.html\?.+?)\1', r'<iframe[^>]+?src=(["\'])(?P<url>https?://m(?:lb)?\.mlb\.com/shared/video/embed/embed\.html\?.+?)\1',
r'data-video-link=["\'](?P<url>http://m\.mlb\.com/video/[^"\']+)', r'data-video-link=["\'](?P<url>http://m\.mlb\.com/video/[^"\']+)',
] ]
_TESTS = [ _TESTS = [{
{ 'url': 'https://www.mlb.com/mariners/video/ackleys-spectacular-catch/c-34698933',
'url': 'https://www.mlb.com/mariners/video/ackleys-spectacular-catch/c-34698933', 'info_dict': {
'md5': '632358dacfceec06bad823b83d21df2d', 'id': '34698933',
'info_dict': { 'ext': 'mp4',
'id': '34698933', 'title': 'Ackley\'s spectacular catch',
'ext': 'mp4', 'description': 'md5:7f5a981eb4f3cbc8daf2aeffa2215bf0',
'title': "Ackley's spectacular catch", 'duration': 66,
'description': 'md5:7f5a981eb4f3cbc8daf2aeffa2215bf0', 'timestamp': 1405995000,
'duration': 66, 'upload_date': '20140722',
'timestamp': 1405995000, 'thumbnail': r're:https?://.+\.jpg',
'upload_date': '20140722',
'thumbnail': r're:^https?://.*\.jpg$',
},
},
{
'url': 'https://www.mlb.com/video/stanton-prepares-for-derby/c-34496663',
'md5': 'bf2619bf9cacc0a564fc35e6aeb9219f',
'info_dict': {
'id': '34496663',
'ext': 'mp4',
'title': 'Stanton prepares for Derby',
'description': 'md5:d00ce1e5fd9c9069e9c13ab4faedfa57',
'duration': 46,
'timestamp': 1405120200,
'upload_date': '20140711',
'thumbnail': r're:^https?://.*\.jpg$',
},
},
{
'url': 'https://www.mlb.com/video/cespedes-repeats-as-derby-champ/c-34578115',
'md5': '99bb9176531adc600b90880fb8be9328',
'info_dict': {
'id': '34578115',
'ext': 'mp4',
'title': 'Cespedes repeats as Derby champ',
'description': 'md5:08df253ce265d4cf6fb09f581fafad07',
'duration': 488,
'timestamp': 1405414336,
'upload_date': '20140715',
'thumbnail': r're:^https?://.*\.jpg$',
},
},
{
'url': 'https://www.mlb.com/video/bautista-on-home-run-derby/c-34577915',
'md5': 'da8b57a12b060e7663ee1eebd6f330ec',
'info_dict': {
'id': '34577915',
'ext': 'mp4',
'title': 'Bautista on Home Run Derby',
'description': 'md5:b80b34031143d0986dddc64a8839f0fb',
'duration': 52,
'timestamp': 1405405122,
'upload_date': '20140715',
'thumbnail': r're:^https?://.*\.jpg$',
},
},
{
'url': 'https://www.mlb.com/video/hargrove-homers-off-caldwell/c-1352023483?tid=67793694',
'only_matching': True,
},
{
'url': 'http://m.mlb.com/shared/video/embed/embed.html?content_id=35692085&topic_id=6479266&width=400&height=224&property=mlb',
'only_matching': True,
}, },
{ }, {
'url': 'http://mlb.mlb.com/shared/video/embed/embed.html?content_id=36599553', 'url': 'https://www.mlb.com/video/stanton-prepares-for-derby/c-34496663',
'only_matching': True, 'info_dict': {
'id': '34496663',
'ext': 'mp4',
'title': 'Stanton prepares for Derby',
'description': 'md5:d00ce1e5fd9c9069e9c13ab4faedfa57',
'duration': 46,
'timestamp': 1405120200,
'upload_date': '20140711',
'thumbnail': r're:https?://.+\.jpg',
}, },
{ }, {
'url': 'http://mlb.mlb.com/es/video/play.jsp?content_id=36599553', 'url': 'https://www.mlb.com/video/cespedes-repeats-as-derby-champ/c-34578115',
'only_matching': True, 'info_dict': {
'id': '34578115',
'ext': 'mp4',
'title': 'Cespedes repeats as Derby champ',
'description': 'md5:08df253ce265d4cf6fb09f581fafad07',
'duration': 488,
'timestamp': 1405414336,
'upload_date': '20140715',
'thumbnail': r're:https?://.+\.jpg',
}, },
{ }, {
'url': 'https://www.mlb.com/cardinals/video/piscottys-great-sliding-catch/c-51175783', 'url': 'https://www.mlb.com/video/bautista-on-home-run-derby/c-34577915',
'only_matching': True, 'info_dict': {
'id': '34577915',
'ext': 'mp4',
'title': 'Bautista on Home Run Derby',
'description': 'md5:b80b34031143d0986dddc64a8839f0fb',
'duration': 52,
'timestamp': 1405405122,
'upload_date': '20140715',
'thumbnail': r're:https?://.+\.jpg',
}, },
{ }, {
# From http://m.mlb.com/news/article/118550098/blue-jays-kevin-pillar-goes-spidey-up-the-wall-to-rob-tim-beckham-of-a-homer 'url': 'https://www.mlb.com/video/hargrove-homers-off-caldwell/c-1352023483?tid=67793694',
'url': 'http://mlb.mlb.com/shared/video/embed/m-internal-embed.html?content_id=75609783&property=mlb&autoplay=true&hashmode=false&siteSection=mlb/multimedia/article_118550098/article_embed&club=mlb', 'only_matching': True,
'only_matching': True, }, {
'url': 'http://m.mlb.com/shared/video/embed/embed.html?content_id=35692085&topic_id=6479266&width=400&height=224&property=mlb',
'only_matching': True,
}, {
'url': 'http://mlb.mlb.com/shared/video/embed/embed.html?content_id=36599553',
'only_matching': True,
}, {
'url': 'http://mlb.mlb.com/es/video/play.jsp?content_id=36599553',
'only_matching': True,
}, {
'url': 'https://www.mlb.com/cardinals/video/piscottys-great-sliding-catch/c-51175783',
'only_matching': True,
}, {
# From http://m.mlb.com/news/article/118550098/blue-jays-kevin-pillar-goes-spidey-up-the-wall-to-rob-tim-beckham-of-a-homer
'url': 'http://mlb.mlb.com/shared/video/embed/m-internal-embed.html?content_id=75609783&property=mlb&autoplay=true&hashmode=false&siteSection=mlb/multimedia/article_118550098/article_embed&club=mlb',
'only_matching': True,
}]
_WEBPAGE_TESTS = [{
'url': 'https://www.mlbdailydish.com/2013/2/25/4028804/mlb-classic-video-vault-open-watch-embed-share',
'info_dict': {
'id': 'mlb-classic-video-vault-open-watch-embed-share',
'title': 'MLB Classic vault is open! Don\'t avert your eyes!',
'age_limit': 0,
'description': 'All the video needed to hold you over until real baseball starts next month.',
'thumbnail': r're:https?://cdn\.vox-cdn\.com/thumbor/.+\.jpg',
}, },
] 'playlist_count': 3,
}]
_TIMESTAMP_KEY = 'date' _TIMESTAMP_KEY = 'date'
@staticmethod @staticmethod
@ -215,20 +211,19 @@ class MLBIE(MLBBaseIE):
class MLBVideoIE(MLBBaseIE): class MLBVideoIE(MLBBaseIE):
_VALID_URL = r'https?://(?:www\.)?mlb\.com/(?:[^/]+/)*video/(?P<id>[^/?&#]+)' _VALID_URL = r'https?://(?:www\.)?mlb\.com/(?:[^/]+/)*video/(?P<id>[^/?&#]+)'
_TEST = { _TESTS = [{
'url': 'https://www.mlb.com/mariners/video/ackley-s-spectacular-catch-c34698933', 'url': 'https://www.mlb.com/mariners/video/ackley-s-spectacular-catch-c34698933',
'md5': '632358dacfceec06bad823b83d21df2d',
'info_dict': { 'info_dict': {
'id': 'c04a8863-f569-42e6-9f87-992393657614', 'id': 'c04a8863-f569-42e6-9f87-992393657614',
'ext': 'mp4', 'ext': 'mp4',
'title': "Ackley's spectacular catch", 'title': 'Ackley\'s spectacular catch',
'description': 'md5:7f5a981eb4f3cbc8daf2aeffa2215bf0', 'description': 'md5:7f5a981eb4f3cbc8daf2aeffa2215bf0',
'duration': 66, 'duration': 66,
'timestamp': 1405995000, 'timestamp': 1405995000,
'upload_date': '20140722', 'upload_date': '20140722',
'thumbnail': r're:^https?://.+', 'thumbnail': r're:https?://.+',
}, },
} }]
_TIMESTAMP_KEY = 'timestamp' _TIMESTAMP_KEY = 'timestamp'
@classmethod @classmethod

402
yt_dlp/extractor/nbc.py
Unescape Escape View File

@ -138,95 +138,88 @@ class NBCUniversalBaseIE(ThePlatformBaseIE):
class NBCIE(NBCUniversalBaseIE): class NBCIE(NBCUniversalBaseIE):
_VALID_URL = r'https?(?P<permalink>://(?:www\.)?nbc\.com/(?:classic-tv/)?[^/?#]+/video/[^/?#]+/(?P<id>\w+))' _VALID_URL = r'https?(?P<permalink>://(?:www\.)?nbc\.com/(?:classic-tv/)?[^/?#]+/video/[^/?#]+/(?P<id>\w+))'
_TESTS = [ _TESTS = [{
{ 'url': 'http://www.nbc.com/the-tonight-show/video/jimmy-fallon-surprises-fans-at-ben-jerrys/2848237',
'url': 'http://www.nbc.com/the-tonight-show/video/jimmy-fallon-surprises-fans-at-ben-jerrys/2848237', 'info_dict': {
'info_dict': { 'id': '2848237',
'id': '2848237', 'ext': 'mp4',
'ext': 'mp4', 'title': 'Jimmy Fallon Surprises Fans at Ben & Jerry\'s',
'title': 'Jimmy Fallon Surprises Fans at Ben & Jerry\'s', 'description': 'Jimmy gives out free scoops of his new "Tonight Dough" ice cream flavor by surprising customers at the Ben & Jerry\'s scoop shop.',
'description': 'Jimmy gives out free scoops of his new "Tonight Dough" ice cream flavor by surprising customers at the Ben & Jerry\'s scoop shop.', 'timestamp': 1424246400,
'timestamp': 1424246400, 'upload_date': '20150218',
'upload_date': '20150218', 'uploader': 'NBCU-COM',
'uploader': 'NBCU-COM', 'episode': 'Jimmy Fallon Surprises Fans at Ben & Jerry\'s',
'episode': 'Jimmy Fallon Surprises Fans at Ben & Jerry\'s', 'episode_number': 86,
'episode_number': 86, 'season': 'Season 2',
'season': 'Season 2', 'season_number': 2,
'season_number': 2, 'series': 'Tonight',
'series': 'Tonight', 'duration': 236.504,
'duration': 236.504, 'tags': 'count:2',
'tags': 'count:2', 'thumbnail': r're:https?://.+\.jpg',
'thumbnail': r're:https?://.+\.jpg', 'categories': ['Series/The Tonight Show Starring Jimmy Fallon'],
'categories': ['Series/The Tonight Show Starring Jimmy Fallon'], 'media_type': 'Full Episode',
'media_type': 'Full Episode', 'age_limit': 14,
'age_limit': 14, '_old_archive_ids': ['theplatform 2848237'],
'_old_archive_ids': ['theplatform 2848237'],
},
'params': {
'skip_download': 'm3u8',
},
}, },
{ 'params': {
'url': 'https://www.nbc.com/the-golden-globe-awards/video/oprah-winfrey-receives-cecil-b-de-mille-award-at-the-2018-golden-globes/3646439', 'skip_download': 'm3u8',
'info_dict': {
'id': '3646439',
'ext': 'mp4',
'title': 'Oprah Winfrey Receives Cecil B. de Mille Award at the 2018 Golden Globes',
'episode': 'Oprah Winfrey Receives Cecil B. de Mille Award at the 2018 Golden Globes',
'episode_number': 1,
'season': 'Season 75',
'season_number': 75,
'series': 'Golden Globes',
'description': 'Oprah Winfrey receives the Cecil B. de Mille Award at the 75th Annual Golden Globe Awards.',
'uploader': 'NBCU-COM',
'upload_date': '20180107',
'timestamp': 1515312000,
'duration': 569.703,
'tags': 'count:8',
'thumbnail': r're:https?://.+\.jpg',
'media_type': 'Highlight',
'age_limit': 0,
'categories': ['Series/The Golden Globe Awards'],
'_old_archive_ids': ['theplatform 3646439'],
},
'params': {
'skip_download': 'm3u8',
},
}, },
{ }, {
# Needs to be extracted from webpage instead of GraphQL 'url': 'https://www.nbc.com/the-golden-globe-awards/video/oprah-winfrey-receives-cecil-b-de-mille-award-at-the-2018-golden-globes/3646439',
'url': 'https://www.nbc.com/paris2024/video/ali-truwit-found-purpose-pool-after-her-life-changed/para24_sww_alitruwittodayshow_240823', 'info_dict': {
'info_dict': { 'id': '3646439',
'id': 'para24_sww_alitruwittodayshow_240823', 'ext': 'mp4',
'ext': 'mp4', 'title': 'Oprah Winfrey Receives Cecil B. de Mille Award at the 2018 Golden Globes',
'title': 'Ali Truwit found purpose in the pool after her life changed', 'episode': 'Oprah Winfrey Receives Cecil B. de Mille Award at the 2018 Golden Globes',
'description': 'md5:c16d7489e1516593de1cc5d3f39b9bdb', 'episode_number': 1,
'uploader': 'NBCU-SPORTS', 'season': 'Season 75',
'duration': 311.077, 'season_number': 75,
'thumbnail': r're:https?://.+\.jpg', 'series': 'Golden Globes',
'episode': 'Ali Truwit found purpose in the pool after her life changed', 'description': 'Oprah Winfrey receives the Cecil B. de Mille Award at the 75th Annual Golden Globe Awards.',
'timestamp': 1724435902.0, 'uploader': 'NBCU-COM',
'upload_date': '20240823', 'upload_date': '20180107',
'_old_archive_ids': ['theplatform para24_sww_alitruwittodayshow_240823'], 'timestamp': 1515312000,
}, 'duration': 569.703,
'params': { 'tags': 'count:8',
'skip_download': 'm3u8', 'thumbnail': r're:https?://.+\.jpg',
}, 'media_type': 'Highlight',
'age_limit': 0,
'categories': ['Series/The Golden Globe Awards'],
'_old_archive_ids': ['theplatform 3646439'],
}, },
{ 'params': {
'url': 'https://www.nbc.com/quantum-leap/video/bens-first-leap-nbcs-quantum-leap/NBCE125189978', 'skip_download': 'm3u8',
'only_matching': True,
}, },
{ }, {
'url': 'https://www.nbc.com/classic-tv/charles-in-charge/video/charles-in-charge-pilot/n3310', # Needs to be extracted from webpage instead of GraphQL
'only_matching': True, 'url': 'https://www.nbc.com/paris2024/video/ali-truwit-found-purpose-pool-after-her-life-changed/para24_sww_alitruwittodayshow_240823',
'info_dict': {
'id': 'para24_sww_alitruwittodayshow_240823',
'ext': 'mp4',
'title': 'Ali Truwit found purpose in the pool after her life changed',
'description': 'md5:c16d7489e1516593de1cc5d3f39b9bdb',
'uploader': 'NBCU-SPORTS',
'duration': 311.077,
'thumbnail': r're:https?://.+\.jpg',
'episode': 'Ali Truwit found purpose in the pool after her life changed',
'timestamp': 1724435902.0,
'upload_date': '20240823',
'_old_archive_ids': ['theplatform para24_sww_alitruwittodayshow_240823'],
}, },
{ 'params': {
# Percent escaped url 'skip_download': 'm3u8',
'url': 'https://www.nbc.com/up-all-night/video/day-after-valentine%27s-day/n2189',
'only_matching': True,
}, },
] }, {
'url': 'https://www.nbc.com/quantum-leap/video/bens-first-leap-nbcs-quantum-leap/NBCE125189978',
'only_matching': True,
}, {
'url': 'https://www.nbc.com/classic-tv/charles-in-charge/video/charles-in-charge-pilot/n3310',
'only_matching': True,
}, {
# Percent escaped url
'url': 'https://www.nbc.com/up-all-night/video/day-after-valentine%27s-day/n2189',
'only_matching': True,
}]
_SOFTWARE_STATEMENT = 'eyJhbGciOiJSUzI1NiJ9.eyJzdWIiOiI1Yzg2YjdkYy04NDI3LTRjNDUtOGQwZi1iNDkzYmE3MmQwYjQiLCJuYmYiOjE1Nzg3MDM2MzEsImlzcyI6ImF1dGguYWRvYmUuY29tIiwiaWF0IjoxNTc4NzAzNjMxfQ.QQKIsBhAjGQTMdAqRTqhcz2Cddr4Y2hEjnSiOeKKki4nLrkDOsjQMmqeTR0hSRarraxH54wBgLvsxI7LHwKMvr7G8QpynNAxylHlQD3yhN9tFhxt4KR5wW3as02B-W2TznK9bhNWPKIyHND95Uo2Mi6rEQoq8tM9O09WPWaanE5BX_-r6Llr6dPq5F0Lpx2QOn2xYRb1T4nFxdFTNoss8GBds8OvChTiKpXMLHegLTc1OS4H_1a8tO_37jDwSdJuZ8iTyRLV4kZ2cpL6OL5JPMObD4-HQiec_dfcYgMKPiIfP9ZqdXpec2SVaCLsWEk86ZYvD97hLIQrK5rrKd1y-A' _SOFTWARE_STATEMENT = 'eyJhbGciOiJSUzI1NiJ9.eyJzdWIiOiI1Yzg2YjdkYy04NDI3LTRjNDUtOGQwZi1iNDkzYmE3MmQwYjQiLCJuYmYiOjE1Nzg3MDM2MzEsImlzcyI6ImF1dGguYWRvYmUuY29tIiwiaWF0IjoxNTc4NzAzNjMxfQ.QQKIsBhAjGQTMdAqRTqhcz2Cddr4Y2hEjnSiOeKKki4nLrkDOsjQMmqeTR0hSRarraxH54wBgLvsxI7LHwKMvr7G8QpynNAxylHlQD3yhN9tFhxt4KR5wW3as02B-W2TznK9bhNWPKIyHND95Uo2Mi6rEQoq8tM9O09WPWaanE5BX_-r6Llr6dPq5F0Lpx2QOn2xYRb1T4nFxdFTNoss8GBds8OvChTiKpXMLHegLTc1OS4H_1a8tO_37jDwSdJuZ8iTyRLV4kZ2cpL6OL5JPMObD4-HQiec_dfcYgMKPiIfP9ZqdXpec2SVaCLsWEk86ZYvD97hLIQrK5rrKd1y-A'
def _real_extract(self, url): def _real_extract(self, url):
@ -378,6 +371,15 @@ class NBCSportsIE(InfoExtractor):
'url': 'https://www.nbcsports.com/boston/video/report-card-pats-secondary-no-match-josh-allen', 'url': 'https://www.nbcsports.com/boston/video/report-card-pats-secondary-no-match-josh-allen',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
'url': 'http://www.riderfans.com/forum/showthread.php?121827-Freeman&s=e98fa1ea6dc08e886b1678d35212494a',
'info_dict': {
'id': 'ln7x1qSThw4k',
'ext': 'flv',
'title': "PFT Live: New leader in the 'new-look' defense",
},
'skip': 'Invalid URL',
}]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)
@ -389,7 +391,7 @@ class NBCSportsIE(InfoExtractor):
class NBCSportsStreamIE(AdobePassIE): class NBCSportsStreamIE(AdobePassIE):
_WORKING = False _WORKING = False
_VALID_URL = r'https?://stream\.nbcsports\.com/.+?\bpid=(?P<id>\d+)' _VALID_URL = r'https?://stream\.nbcsports\.com/.+?\bpid=(?P<id>\d+)'
_TEST = { _TESTS = [{
'url': 'http://stream.nbcsports.com/nbcsn/generic?pid=206559', 'url': 'http://stream.nbcsports.com/nbcsn/generic?pid=206559',
'info_dict': { 'info_dict': {
'id': '206559', 'id': '206559',
@ -402,7 +404,7 @@ class NBCSportsStreamIE(AdobePassIE):
'skip_download': True, 'skip_download': True,
}, },
'skip': 'Requires Adobe Pass Authentication', 'skip': 'Requires Adobe Pass Authentication',
} }]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)
@ -449,98 +451,100 @@ class NBCNewsIE(ThePlatformIE): # XXX: Do not subclass from concrete IE
_VALID_URL = r'(?x)https?://(?:www\.)?(?:nbcnews|today|msnbc)\.com/([^/]+/)*(?:.*-)?(?P<id>[^/?]+)' _VALID_URL = r'(?x)https?://(?:www\.)?(?:nbcnews|today|msnbc)\.com/([^/]+/)*(?:.*-)?(?P<id>[^/?]+)'
_EMBED_REGEX = [r'<iframe[^>]+src=(["\'])(?P<url>(?:https?:)?//www\.nbcnews\.com/widget/video-embed/[^"\']+)\1'] _EMBED_REGEX = [r'<iframe[^>]+src=(["\'])(?P<url>(?:https?:)?//www\.nbcnews\.com/widget/video-embed/[^"\']+)\1']
_TESTS = [ _TESTS = [{
{ 'url': 'http://www.nbcnews.com/watch/nbcnews-com/how-twitter-reacted-to-the-snowden-interview-269389891880',
'url': 'http://www.nbcnews.com/watch/nbcnews-com/how-twitter-reacted-to-the-snowden-interview-269389891880', 'md5': 'fb3dcd2d7b1dd9804305fa2fc95ab610', # md5 tends to fluctuate
'md5': 'fb3dcd2d7b1dd9804305fa2fc95ab610', # md5 tends to fluctuate 'info_dict': {
'info_dict': { 'id': '269389891880',
'id': '269389891880', 'ext': 'mp4',
'ext': 'mp4', 'title': 'How Twitter Reacted To The Snowden Interview',
'title': 'How Twitter Reacted To The Snowden Interview', 'description': 'md5:65a0bd5d76fe114f3c2727aa3a81fe64',
'description': 'md5:65a0bd5d76fe114f3c2727aa3a81fe64', 'timestamp': 1401363060,
'timestamp': 1401363060, 'upload_date': '20140529',
'upload_date': '20140529', 'duration': 46.0,
'duration': 46.0, 'thumbnail': 'https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/140529/p_tweet_snow_140529.jpg',
'thumbnail': 'https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/140529/p_tweet_snow_140529.jpg',
},
},
{
'url': 'http://www.nbcnews.com/feature/dateline-full-episodes/full-episode-family-business-n285156',
'md5': 'fdbf39ab73a72df5896b6234ff98518a',
'info_dict': {
'id': '529953347624',
'ext': 'mp4',
'title': 'FULL EPISODE: Family Business',
'description': 'md5:757988edbaae9d7be1d585eb5d55cc04',
},
'skip': 'This page is unavailable.',
}, },
{ }, {
'url': 'http://www.nbcnews.com/nightly-news/video/nightly-news-with-brian-williams-full-broadcast-february-4-394064451844', 'url': 'http://www.nbcnews.com/feature/dateline-full-episodes/full-episode-family-business-n285156',
'md5': '40d0e48c68896359c80372306ece0fc3', 'md5': 'fdbf39ab73a72df5896b6234ff98518a',
'info_dict': { 'info_dict': {
'id': '394064451844', 'id': '529953347624',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Nightly News with Brian Williams Full Broadcast (February 4)', 'title': 'FULL EPISODE: Family Business',
'description': 'md5:1c10c1eccbe84a26e5debb4381e2d3c5', 'description': 'md5:757988edbaae9d7be1d585eb5d55cc04',
'timestamp': 1423104900,
'upload_date': '20150205',
'duration': 1236.0,
'thumbnail': 'https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/__NEW/nn_netcast_150204.jpg',
},
}, },
{ 'skip': 'This page is unavailable.',
'url': 'http://www.nbcnews.com/business/autos/volkswagen-11-million-vehicles-could-have-suspect-software-emissions-scandal-n431456', }, {
'md5': 'ffb59bcf0733dc3c7f0ace907f5e3939', 'url': 'http://www.nbcnews.com/nightly-news/video/nightly-news-with-brian-williams-full-broadcast-february-4-394064451844',
'info_dict': { 'md5': '40d0e48c68896359c80372306ece0fc3',
'id': 'n431456', 'info_dict': {
'ext': 'mp4', 'id': '394064451844',
'title': "Volkswagen U.S. Chief: We 'Totally Screwed Up'", 'ext': 'mp4',
'description': 'md5:d22d1281a24f22ea0880741bb4dd6301', 'title': 'Nightly News with Brian Williams Full Broadcast (February 4)',
'upload_date': '20150922', 'description': 'md5:1c10c1eccbe84a26e5debb4381e2d3c5',
'timestamp': 1442917800, 'timestamp': 1423104900,
'duration': 37.0, 'upload_date': '20150205',
'thumbnail': 'https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/__NEW/x_lon_vwhorn_150922.jpg', 'duration': 1236.0,
}, 'thumbnail': 'https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/__NEW/nn_netcast_150204.jpg',
}, },
{ }, {
'url': 'http://www.today.com/video/see-the-aurora-borealis-from-space-in-stunning-new-nasa-video-669831235788', 'url': 'http://www.nbcnews.com/business/autos/volkswagen-11-million-vehicles-could-have-suspect-software-emissions-scandal-n431456',
'md5': '693d1fa21d23afcc9b04c66b227ed9ff', 'md5': 'ffb59bcf0733dc3c7f0ace907f5e3939',
'info_dict': { 'info_dict': {
'id': '669831235788', 'id': 'n431456',
'ext': 'mp4', 'ext': 'mp4',
'title': 'See the aurora borealis from space in stunning new NASA video', 'title': "Volkswagen U.S. Chief: We 'Totally Screwed Up'",
'description': 'md5:74752b7358afb99939c5f8bb2d1d04b1', 'description': 'md5:d22d1281a24f22ea0880741bb4dd6301',
'upload_date': '20160420', 'upload_date': '20150922',
'timestamp': 1461152093, 'timestamp': 1442917800,
'duration': 69.0, 'duration': 37.0,
'thumbnail': 'https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/201604/2016-04-20T11-35-09-133Z--1280x720.jpg', 'thumbnail': 'https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/__NEW/x_lon_vwhorn_150922.jpg',
},
}, },
{ }, {
'url': 'http://www.msnbc.com/all-in-with-chris-hayes/watch/the-chaotic-gop-immigration-vote-314487875924', 'url': 'http://www.today.com/video/see-the-aurora-borealis-from-space-in-stunning-new-nasa-video-669831235788',
'md5': '6d236bf4f3dddc226633ce6e2c3f814d', 'md5': '693d1fa21d23afcc9b04c66b227ed9ff',
'info_dict': { 'info_dict': {
'id': '314487875924', 'id': '669831235788',
'ext': 'mp4', 'ext': 'mp4',
'title': 'The chaotic GOP immigration vote', 'title': 'See the aurora borealis from space in stunning new NASA video',
'description': 'The Republican House votes on a border bill that has no chance of getting through the Senate or signed by the President and is drawing criticism from all sides.', 'description': 'md5:74752b7358afb99939c5f8bb2d1d04b1',
'thumbnail': r're:^https?://.*\.jpg$', 'upload_date': '20160420',
'timestamp': 1406937606, 'timestamp': 1461152093,
'upload_date': '20140802', 'duration': 69.0,
'duration': 940.0, 'thumbnail': 'https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/201604/2016-04-20T11-35-09-133Z--1280x720.jpg',
},
}, },
{ 'skip': 'Invalid URL',
'url': 'http://www.nbcnews.com/watch/dateline/full-episode--deadly-betrayal-386250819952', }, {
'only_matching': True, 'url': 'http://www.msnbc.com/all-in-with-chris-hayes/watch/the-chaotic-gop-immigration-vote-314487875924',
'md5': '6d236bf4f3dddc226633ce6e2c3f814d',
'info_dict': {
'id': '314487875924',
'ext': 'mp4',
'title': 'The chaotic GOP immigration vote',
'description': 'The Republican House votes on a border bill that has no chance of getting through the Senate or signed by the President and is drawing criticism from all sides.',
'thumbnail': r're:https?://.+\.jpg',
'timestamp': 1406937606,
'upload_date': '20140802',
'duration': 940.0,
}, },
{ 'skip': 'Invalid URL',
# From http://www.vulture.com/2016/06/letterman-couldnt-care-less-about-late-night.html }, {
'url': 'http://www.nbcnews.com/widget/video-embed/701714499682', 'url': 'http://www.nbcnews.com/watch/dateline/full-episode--deadly-betrayal-386250819952',
'only_matching': True, 'only_matching': True,
}, {
# From http://www.vulture.com/2016/06/letterman-couldnt-care-less-about-late-night.html
'url': 'http://www.nbcnews.com/widget/video-embed/701714499682',
'only_matching': True,
}]
_WEBPAGE_TESTS = [{
'url': 'http://www.vulture.com/2016/06/letterman-couldnt-care-less-about-late-night.html',
'info_dict': {
'id': 'x_dtl_oa_LettermanliftPR_160608',
'ext': 'mp4',
'title': 'David Letterman: A Preview',
}, },
] 'skip': 'Invalid URL',
}]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)
@ -610,10 +614,10 @@ class NBCOlympicsIE(InfoExtractor):
'display_id': 'watch-final-minutes-team-usas-mens-basketball-gold', 'display_id': 'watch-final-minutes-team-usas-mens-basketball-gold',
'title': 'Watch the final minutes of Team USA\'s men\'s basketball gold', 'title': 'Watch the final minutes of Team USA\'s men\'s basketball gold',
'description': 'md5:f704f591217305c9559b23b877aa8d31', 'description': 'md5:f704f591217305c9559b23b877aa8d31',
'episode': 'Watch the final minutes of Team USA\'s men\'s basketball gold',
'uploader': 'NBCU-SPORTS', 'uploader': 'NBCU-SPORTS',
'duration': 387.053, 'duration': 387.053,
'thumbnail': r're:https://.+/.+\.jpg', 'thumbnail': r're:https?://.+\.jpg',
'chapters': [],
'timestamp': 1723346984, 'timestamp': 1723346984,
'upload_date': '20240811', 'upload_date': '20240811',
}, },
@ -652,33 +656,31 @@ class NBCOlympicsStreamIE(AdobePassIE):
_WORKING = False _WORKING = False
IE_NAME = 'nbcolympics:stream' IE_NAME = 'nbcolympics:stream'
_VALID_URL = r'https?://stream\.nbcolympics\.com/(?P<id>[0-9a-z-]+)' _VALID_URL = r'https?://stream\.nbcolympics\.com/(?P<id>[0-9a-z-]+)'
_TESTS = [ _TESTS = [{
{ 'note': 'Tokenized m3u8 source URL',
'note': 'Tokenized m3u8 source URL', 'url': 'https://stream.nbcolympics.com/womens-soccer-group-round-11',
'url': 'https://stream.nbcolympics.com/womens-soccer-group-round-11', 'info_dict': {
'info_dict': { 'id': '2019740',
'id': '2019740', 'ext': 'mp4',
'ext': 'mp4', 'title': r"re:Women's Group Stage - Netherlands vs\. Brazil [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$",
'title': r"re:Women's Group Stage - Netherlands vs\. Brazil [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$", },
}, 'params': {
'params': { 'skip_download': 'm3u8',
'skip_download': 'm3u8',
},
'skip': 'Livestream',
}, {
'note': 'Plain m3u8 source URL',
'url': 'https://stream.nbcolympics.com/gymnastics-event-finals-mens-floor-pommel-horse-womens-vault-bars',
'info_dict': {
'id': '2021729',
'ext': 'mp4',
'title': r're:Event Finals: M Floor, W Vault, M Pommel, W Uneven Bars [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
},
'params': {
'skip_download': 'm3u8',
},
'skip': 'Livestream',
}, },
] 'skip': 'Livestream',
}, {
'note': 'Plain m3u8 source URL',
'url': 'https://stream.nbcolympics.com/gymnastics-event-finals-mens-floor-pommel-horse-womens-vault-bars',
'info_dict': {
'id': '2021729',
'ext': 'mp4',
'title': r're:Event Finals: M Floor, W Vault, M Pommel, W Uneven Bars [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
},
'params': {
'skip_download': 'm3u8',
},
'skip': 'Livestream',
}]
def _real_extract(self, url): def _real_extract(self, url):
display_id = self._match_id(url) display_id = self._match_id(url)
@ -758,9 +760,7 @@ class NBCStationsIE(InfoExtractor):
'channel_id': 'KNBC', 'channel_id': 'KNBC',
'channel': 'nbclosangeles', 'channel': 'nbclosangeles',
}, },
'params': { 'skip': 'Site changed',
'skip_download': 'm3u8',
},
}, { }, {
'url': 'https://www.telemundoarizona.com/responde/huracan-complica-reembolso-para-televidente-de-tucson/2247002/', 'url': 'https://www.telemundoarizona.com/responde/huracan-complica-reembolso-para-televidente-de-tucson/2247002/',
'info_dict': { 'info_dict': {

394
yt_dlp/extractor/panopto.py
Unescape Escape View File

@ -73,163 +73,179 @@ class PanoptoBaseIE(InfoExtractor):
class PanoptoIE(PanoptoBaseIE): class PanoptoIE(PanoptoBaseIE):
_VALID_URL = PanoptoBaseIE.BASE_URL_RE + r'/Pages/(Viewer|Embed)\.aspx.*(?:\?|&)id=(?P<id>[a-f0-9-]+)' _VALID_URL = PanoptoBaseIE.BASE_URL_RE + r'/Pages/(Viewer|Embed)\.aspx.*(?:\?|&)id=(?P<id>[a-f0-9-]+)'
_EMBED_REGEX = [rf'<iframe[^>]+src=["\'](?P<url>{PanoptoBaseIE.BASE_URL_RE}/Pages/(Viewer|Embed|Sessions/List)\.aspx[^"\']+)'] _EMBED_REGEX = [rf'<iframe[^>]+src=["\'](?P<url>{PanoptoBaseIE.BASE_URL_RE}/Pages/(Viewer|Embed|Sessions/List)\.aspx[^"\']+)']
_TESTS = [ _TESTS = [{
{ 'url': 'https://demo.hosted.panopto.com/Panopto/Pages/Viewer.aspx?id=26b3ae9e-4a48-4dcc-96ba-0befba08a0fb',
'url': 'https://demo.hosted.panopto.com/Panopto/Pages/Viewer.aspx?id=26b3ae9e-4a48-4dcc-96ba-0befba08a0fb', 'info_dict': {
'info_dict': { 'id': '26b3ae9e-4a48-4dcc-96ba-0befba08a0fb',
'id': '26b3ae9e-4a48-4dcc-96ba-0befba08a0fb', 'title': 'Panopto for Business - Use Cases',
'title': 'Panopto for Business - Use Cases', 'timestamp': 1459184200,
'timestamp': 1459184200, 'thumbnail': r're:https?://demo\.hosted\.panopto\.com/.+',
'thumbnail': r're:https://demo\.hosted\.panopto\.com/.+', 'upload_date': '20160328',
'upload_date': '20160328', 'ext': 'mp4',
'ext': 'mp4', 'cast': [],
'cast': [], 'chapters': [],
'chapters': [], 'duration': 88.17099999999999,
'duration': 88.17099999999999, 'average_rating': int,
'average_rating': int, 'tags': [],
'uploader_id': '2db6b718-47a0-4b0b-9e17-ab0b00f42b1e', 'uploader_id': '2db6b718-47a0-4b0b-9e17-ab0b00f42b1e',
'channel_id': 'e4c6a2fc-1214-4ca0-8fb7-aef2e29ff63a', 'channel_id': 'bb0b58ff-b31b-47a0-9aa2-af6f0113613a',
'channel': 'Showcase Videos', 'channel': 'Product',
},
}, },
{ }, {
'url': 'https://demo.hosted.panopto.com/Panopto/Pages/Viewer.aspx?id=ed01b077-c9e5-4c7b-b8ff-15fa306d7a59', 'url': 'https://demo.hosted.panopto.com/Panopto/Pages/Viewer.aspx?id=ed01b077-c9e5-4c7b-b8ff-15fa306d7a59',
'info_dict': { 'info_dict': {
'id': 'ed01b077-c9e5-4c7b-b8ff-15fa306d7a59', 'id': 'ed01b077-c9e5-4c7b-b8ff-15fa306d7a59',
'title': 'Overcoming Top 4 Challenges of Enterprise Video', 'title': 'Overcoming Top 4 Challenges of Enterprise Video',
'uploader': 'Panopto Support', 'uploader': 'Panopto Support',
'timestamp': 1449409251, 'timestamp': 1449409251,
'thumbnail': r're:https://demo\.hosted\.panopto\.com/.+', 'thumbnail': r're:https?://demo\.hosted\.panopto\.com/.+',
'upload_date': '20151206', 'upload_date': '20151206',
'ext': 'mp4', 'ext': 'mp4',
'chapters': 'count:12', 'chapters': 'count:13',
'cast': ['Panopto Support'], 'cast': ['Panopto Support'],
'uploader_id': 'a96d1a31-b4de-489b-9eee-b4a5b414372c', 'tags': [],
'average_rating': int, 'uploader_id': 'a96d1a31-b4de-489b-9eee-b4a5b414372c',
'description': 'md5:4391837802b3fc856dadf630c4b375d1', 'average_rating': int,
'duration': 1088.2659999999998, 'description': 'md5:4391837802b3fc856dadf630c4b375d1',
'channel_id': '9f3c1921-43bb-4bda-8b3a-b8d2f05a8546', 'duration': 1088.2659999999998,
'channel': 'Webcasts', 'channel_id': '9f3c1921-43bb-4bda-8b3a-b8d2f05a8546',
}, 'channel': 'Webcasts',
}, },
{ }, {
# Extra params in URL # Extra params in URL
'url': 'https://howtovideos.hosted.panopto.com/Panopto/Pages/Viewer.aspx?randomparam=thisisnotreal&id=5fa74e93-3d87-4694-b60e-aaa4012214ed&advance=true', 'url': 'https://howtovideos.hosted.panopto.com/Panopto/Pages/Viewer.aspx?randomparam=thisisnotreal&id=5fa74e93-3d87-4694-b60e-aaa4012214ed&advance=true',
'info_dict': { 'info_dict': {
'id': '5fa74e93-3d87-4694-b60e-aaa4012214ed', 'id': '5fa74e93-3d87-4694-b60e-aaa4012214ed',
'ext': 'mp4', 'ext': 'mp4',
'duration': 129.513, 'duration': 129.513,
'cast': ['Kathryn Kelly'], 'cast': ['Kathryn Kelly'],
'uploader_id': '316a0a58-7fa2-4cd9-be1c-64270d284a56', 'uploader_id': '316a0a58-7fa2-4cd9-be1c-64270d284a56',
'timestamp': 1569845768, 'timestamp': 1569845768,
'tags': ['Viewer', 'Enterprise'], 'tags': ['Viewer', 'Enterprise'],
'chapters': [], 'chapters': [],
'upload_date': '20190930', 'upload_date': '20190930',
'thumbnail': r're:https://howtovideos\.hosted\.panopto\.com/.+', 'thumbnail': r're:https?://howtovideos\.hosted\.panopto\.com/.+',
'description': 'md5:2d844aaa1b1a14ad0e2601a0993b431f', 'description': 'md5:2d844aaa1b1a14ad0e2601a0993b431f',
'title': 'Getting Started: View a Video', 'title': 'Getting Started: View a Video',
'average_rating': int, 'average_rating': int,
'uploader': 'Kathryn Kelly', 'uploader': 'Kathryn Kelly',
'channel_id': 'fb93bc3c-6750-4b80-a05b-a921013735d3', 'channel_id': 'fb93bc3c-6750-4b80-a05b-a921013735d3',
'channel': 'Getting Started', 'channel': 'Getting Started',
},
}, },
{ 'skip': 'Invalid URL',
# Does not allow normal Viewer.aspx. AUDIO livestream has no url, so should be skipped and only give one stream. }, {
'url': 'https://unisa.au.panopto.com/Panopto/Pages/Embed.aspx?id=9d9a0fa3-e99a-4ebd-a281-aac2017f4da4', # Does not allow normal Viewer.aspx. AUDIO livestream has no url, so should be skipped and only give one stream.
'info_dict': { 'url': 'https://unisa.au.panopto.com/Panopto/Pages/Embed.aspx?id=9d9a0fa3-e99a-4ebd-a281-aac2017f4da4',
'id': '9d9a0fa3-e99a-4ebd-a281-aac2017f4da4', 'info_dict': {
'ext': 'mp4', 'id': '9d9a0fa3-e99a-4ebd-a281-aac2017f4da4',
'cast': ['LTS CLI Script'], 'ext': 'mp4',
'chapters': [], 'cast': ['LTS CLI Script'],
'duration': 2178.45, 'chapters': [],
'description': 'md5:ee5cf653919f55b72bce2dbcf829c9fa', 'duration': 2178.45,
'channel_id': 'b23e673f-c287-4cb1-8344-aae9005a69f8', 'description': 'md5:ee5cf653919f55b72bce2dbcf829c9fa',
'average_rating': int, 'channel_id': 'b23e673f-c287-4cb1-8344-aae9005a69f8',
'uploader_id': '38377323-6a23-41e2-9ff6-a8e8004bf6f7', 'average_rating': int,
'uploader': 'LTS CLI Script', 'uploader_id': '38377323-6a23-41e2-9ff6-a8e8004bf6f7',
'timestamp': 1572458134, 'uploader': 'LTS CLI Script',
'title': 'WW2 Vets Interview 3 Ronald Stanley George', 'tags': [],
'thumbnail': r're:https://unisa\.au\.panopto\.com/.+', 'timestamp': 1572458134,
'channel': 'World War II Veteran Interviews', 'title': 'WW2 Vets Interview 3 Ronald Stanley George',
'upload_date': '20191030', 'thumbnail': r're:https?://unisa\.au\.panopto\.com/.+',
}, 'channel': 'World War II Veteran Interviews',
'upload_date': '20191030',
}, },
{ }, {
# Slides/storyboard # Slides/storyboard
'url': 'https://demo.hosted.panopto.com/Panopto/Pages/Viewer.aspx?id=a7f12f1d-3872-4310-84b0-f8d8ab15326b', 'url': 'https://demo.hosted.panopto.com/Panopto/Pages/Viewer.aspx?id=a7f12f1d-3872-4310-84b0-f8d8ab15326b',
'info_dict': { 'info_dict': {
'id': 'a7f12f1d-3872-4310-84b0-f8d8ab15326b', 'id': 'a7f12f1d-3872-4310-84b0-f8d8ab15326b',
'ext': 'mhtml', 'ext': 'mhtml',
'timestamp': 1448798857, 'timestamp': 1448798857,
'duration': 4712.681, 'duration': 4712.681,
'title': 'Cache Memory - CompSci 15-213, Lecture 12', 'title': 'Cache Memory - CompSci 15-213, Lecture 12',
'channel_id': 'e4c6a2fc-1214-4ca0-8fb7-aef2e29ff63a', 'channel_id': '0202d932-6d28-4fb2-b373-af6f0121c8f0',
'uploader_id': 'a96d1a31-b4de-489b-9eee-b4a5b414372c', 'uploader_id': 'a96d1a31-b4de-489b-9eee-b4a5b414372c',
'upload_date': '20151129', 'upload_date': '20151129',
'average_rating': 0, 'average_rating': 0,
'uploader': 'Panopto Support', 'uploader': 'Panopto Support',
'channel': 'Showcase Videos', 'channel': 'Customer Demonstrations',
'description': 'md5:55e51d54233ddb0e6c2ed388ca73822c', 'description': 'md5:55e51d54233ddb0e6c2ed388ca73822c',
'cast': ['ISR Videographer', 'Panopto Support'], 'cast': ['ISR Videographer', 'Panopto Support'],
'chapters': 'count:28', 'chapters': 'count:28',
'thumbnail': r're:https://demo\.hosted\.panopto\.com/.+', 'tags': [],
}, 'thumbnail': r're:https?://demo\.hosted\.panopto\.com/.+',
'params': {'format': 'mhtml', 'skip_download': True},
}, },
{ 'params': {'format': 'mhtml', 'skip_download': True},
'url': 'https://na-training-1.hosted.panopto.com/Panopto/Pages/Viewer.aspx?id=8285224a-9a2b-4957-84f2-acb0000c4ea9', }, {
'info_dict': { 'url': 'https://na-training-1.hosted.panopto.com/Panopto/Pages/Viewer.aspx?id=8285224a-9a2b-4957-84f2-acb0000c4ea9',
'id': '8285224a-9a2b-4957-84f2-acb0000c4ea9', 'info_dict': {
'ext': 'mp4', 'id': '8285224a-9a2b-4957-84f2-acb0000c4ea9',
'chapters': [], 'ext': 'mp4',
'title': 'Company Policy', 'chapters': [],
'average_rating': 0, 'title': 'Company Policy',
'timestamp': 1615058901, 'average_rating': 0,
'channel': 'Human Resources', 'timestamp': 1615058901,
'tags': ['HumanResources'], 'channel': 'Human Resources',
'duration': 1604.243, 'tags': ['HumanResources'],
'thumbnail': r're:https://na-training-1\.hosted\.panopto\.com/.+', 'duration': 1604.243,
'uploader_id': '8e8ba0a3-424f-40df-a4f1-ab3a01375103', 'thumbnail': r're:https?://na-training-1\.hosted\.panopto\.com/.+',
'uploader': 'Cait M.', 'uploader_id': '8e8ba0a3-424f-40df-a4f1-ab3a01375103',
'upload_date': '20210306', 'uploader': 'Cait M.',
'cast': ['Cait M.'], 'upload_date': '20210306',
'subtitles': {'en-US': [{'ext': 'srt', 'data': 'md5:a3f4d25963fdeace838f327097c13265'}], 'cast': ['Cait M.'],
'es-ES': [{'ext': 'srt', 'data': 'md5:57e9dad365fd0fbaf0468eac4949f189'}]},
},
'params': {'writesubtitles': True, 'skip_download': True},
}, {
# On Panopto there are two subs: "Default" and en-US. en-US is blank and should be skipped.
'url': 'https://na-training-1.hosted.panopto.com/Panopto/Pages/Viewer.aspx?id=940cbd41-f616-4a45-b13e-aaf1000c915b',
'info_dict': {
'id': '940cbd41-f616-4a45-b13e-aaf1000c915b',
'ext': 'mp4',
'subtitles': 'count:1',
'title': 'HR Benefits Review Meeting*',
'cast': ['Panopto Support'],
'chapters': [],
'timestamp': 1575024251,
'thumbnail': r're:https://na-training-1\.hosted\.panopto\.com/.+',
'channel': 'Zoom',
'description': 'md5:04f90a9c2c68b7828144abfb170f0106',
'uploader': 'Panopto Support',
'average_rating': 0,
'duration': 409.34499999999997,
'uploader_id': 'b6ac04ad-38b8-4724-a004-a851004ea3df',
'upload_date': '20191129',
},
'params': {'writesubtitles': True, 'skip_download': True},
}, },
{ 'params': {'writesubtitles': True, 'skip_download': True},
'url': 'https://ucc.cloud.panopto.eu/Panopto/Pages/Viewer.aspx?id=0e8484a4-4ceb-4d98-a63f-ac0200b455cb', }, {
'only_matching': True, # On Panopto there are two subs: "Default" and en-US. en-US is blank and should be skipped.
'url': 'https://na-training-1.hosted.panopto.com/Panopto/Pages/Viewer.aspx?id=940cbd41-f616-4a45-b13e-aaf1000c915b',
'info_dict': {
'id': '940cbd41-f616-4a45-b13e-aaf1000c915b',
'ext': 'mp4',
'subtitles': 'count:1',
'title': 'HR Benefits Review Meeting*',
'cast': ['Panopto Support'],
'chapters': [],
'timestamp': 1575024251,
'thumbnail': r're:https://na-training-1\.hosted\.panopto\.com/.+',
'channel': 'Zoom',
'description': 'md5:04f90a9c2c68b7828144abfb170f0106',
'uploader': 'Panopto Support',
'average_rating': 0,
'duration': 409.34499999999997,
'tags': [],
'uploader_id': 'b6ac04ad-38b8-4724-a004-a851004ea3df',
'upload_date': '20191129',
}, },
{ 'params': {'writesubtitles': True, 'skip_download': True},
'url': 'https://brown.hosted.panopto.com/Panopto/Pages/Embed.aspx?id=0b3ff73b-36a0-46c5-8455-aadf010a3638', }, {
'only_matching': True, 'url': 'https://ucc.cloud.panopto.eu/Panopto/Pages/Viewer.aspx?id=0e8484a4-4ceb-4d98-a63f-ac0200b455cb',
'only_matching': True,
}, {
'url': 'https://brown.hosted.panopto.com/Panopto/Pages/Embed.aspx?id=0b3ff73b-36a0-46c5-8455-aadf010a3638',
'only_matching': True,
}]
_WEBPAGE_TESTS = [{
'url': 'https://www.monash.edu/learning-teaching/teachhq/learning-technologies/panopto/how-to/insert-a-quiz-into-a-panopto-video',
'info_dict': {
'id': '0bd3f16c-824a-436a-8486-ac5900693aef',
'ext': 'mp4',
'title': 'Quizzes in Panopto',
'average_rating': 0,
'cast': ['Stephanie Luo'],
'chapters': 'count:8',
'channel': 'Panopto',
'description': 'md5:731ce802eee75808b1181db1ff1b5002',
'duration': 185.833,
'tags': [],
'thumbnail': r're:https?://monash\.au\.panopto\.com/.+',
'timestamp': 1607562188,
'upload_date': '20201210',
'uploader': 'Stephanie Luo',
'uploader_id': 'b18ca46d-20df-4ff5-b0b3-aa7a00085617',
}, },
] 'params': {'extractor_args': {'generic': {'impersonate': ['chrome']}}},
}]
@classmethod @classmethod
def suitable(cls, url): def suitable(cls, url):
@ -423,27 +439,23 @@ class PanoptoIE(PanoptoBaseIE):
class PanoptoPlaylistIE(PanoptoBaseIE): class PanoptoPlaylistIE(PanoptoBaseIE):
_VALID_URL = PanoptoBaseIE.BASE_URL_RE + r'/Pages/(Viewer|Embed)\.aspx.*(?:\?|&)pid=(?P<id>[a-f0-9-]+)' _VALID_URL = PanoptoBaseIE.BASE_URL_RE + r'/Pages/(Viewer|Embed)\.aspx.*(?:\?|&)pid=(?P<id>[a-f0-9-]+)'
_TESTS = [ _TESTS = [{
{ 'url': 'https://howtovideos.hosted.panopto.com/Panopto/Pages/Viewer.aspx?pid=f3b39fcf-882f-4849-93d6-a9f401236d36&id=5fa74e93-3d87-4694-b60e-aaa4012214ed&advance=true',
'url': 'https://howtovideos.hosted.panopto.com/Panopto/Pages/Viewer.aspx?pid=f3b39fcf-882f-4849-93d6-a9f401236d36&id=5fa74e93-3d87-4694-b60e-aaa4012214ed&advance=true', 'info_dict': {
'info_dict': { 'id': 'f3b39fcf-882f-4849-93d6-a9f401236d36',
'title': 'Featured Video Tutorials', 'title': 'Featured Video Tutorials',
'id': 'f3b39fcf-882f-4849-93d6-a9f401236d36', 'description': '',
'description': '',
},
'playlist_mincount': 36,
}, },
{ 'playlist_mincount': 19,
'url': 'https://utsa.hosted.panopto.com/Panopto/Pages/Viewer.aspx?pid=e2900555-3ad4-4bdb-854d-ad2401686190', }, {
'info_dict': { 'url': 'https://utsa.hosted.panopto.com/Panopto/Pages/Viewer.aspx?pid=e2900555-3ad4-4bdb-854d-ad2401686190',
'title': 'Library Website Introduction Playlist', 'info_dict': {
'id': 'e2900555-3ad4-4bdb-854d-ad2401686190', 'id': 'e2900555-3ad4-4bdb-854d-ad2401686190',
'description': 'md5:f958bca50a1cbda15fdc1e20d32b3ecb', 'title': 'Library Website Introduction Playlist',
}, 'description': 'md5:f958bca50a1cbda15fdc1e20d32b3ecb',
'playlist_mincount': 4,
}, },
'playlist_mincount': 4,
] }]
def _entries(self, base_url, playlist_id, session_list_id): def _entries(self, base_url, playlist_id, session_list_id):
session_list_info = self._call_api( session_list_info = self._call_api(
@ -486,35 +498,29 @@ class PanoptoPlaylistIE(PanoptoBaseIE):
class PanoptoListIE(PanoptoBaseIE): class PanoptoListIE(PanoptoBaseIE):
_VALID_URL = PanoptoBaseIE.BASE_URL_RE + r'/Pages/Sessions/List\.aspx' _VALID_URL = PanoptoBaseIE.BASE_URL_RE + r'/Pages/Sessions/List\.aspx'
_PAGE_SIZE = 250 _PAGE_SIZE = 250
_TESTS = [ _TESTS = [{
{ 'url': 'https://demo.hosted.panopto.com/Panopto/Pages/Sessions/List.aspx#folderID=%22e4c6a2fc-1214-4ca0-8fb7-aef2e29ff63a%22',
'url': 'https://demo.hosted.panopto.com/Panopto/Pages/Sessions/List.aspx#folderID=%22e4c6a2fc-1214-4ca0-8fb7-aef2e29ff63a%22', 'info_dict': {
'info_dict': { 'id': 'e4c6a2fc-1214-4ca0-8fb7-aef2e29ff63a',
'id': 'e4c6a2fc-1214-4ca0-8fb7-aef2e29ff63a', 'title': 'Showcase Videos',
'title': 'Showcase Videos',
},
'playlist_mincount': 140,
}, },
{ 'playlist_mincount': 8,
'url': 'https://demo.hosted.panopto.com/Panopto/Pages/Sessions/List.aspx#view=2&maxResults=250', }, {
'info_dict': { 'url': 'https://demo.hosted.panopto.com/Panopto/Pages/Sessions/List.aspx#view=2&maxResults=250',
'id': 'panopto_list', 'info_dict': {
'title': 'panopto_list', 'id': 'panopto_list',
}, 'title': 'panopto_list',
'playlist_mincount': 300,
}, },
{ 'playlist_mincount': 300,
# Folder that contains 8 folders and a playlist }, {
'url': 'https://howtovideos.hosted.panopto.com/Panopto/Pages/Sessions/List.aspx?noredirect=true#folderID=%224b9de7ae-0080-4158-8496-a9ba01692c2e%22', # Folder that contains 8 folders and a playlist
'info_dict': { 'url': 'https://howtovideos.hosted.panopto.com/Panopto/Pages/Sessions/List.aspx?noredirect=true#folderID=%224b9de7ae-0080-4158-8496-a9ba01692c2e%22',
'id': '4b9de7ae-0080-4158-8496-a9ba01692c2e', 'info_dict': {
'title': 'Video Tutorials', 'id': '4b9de7ae-0080-4158-8496-a9ba01692c2e',
}, 'title': 'Video Tutorials',
'playlist_mincount': 9,
}, },
'playlist_mincount': 9,
] }]
def _fetch_page(self, base_url, query_params, display_id, page): def _fetch_page(self, base_url, query_params, display_id, page):

103
yt_dlp/extractor/peertube.py
Unescape Escape View File

@ -1331,7 +1331,7 @@ class PeerTubeIE(InfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'title': 'What is PeerTube?', 'title': 'What is PeerTube?',
'description': 'md5:3fefb8dde2b189186ce0719fda6f7b10', 'description': 'md5:3fefb8dde2b189186ce0719fda6f7b10',
'thumbnail': r're:https?://.*\.(?:jpg|png)', 'thumbnail': r're:https?://framatube\.org/lazy-static/thumbnails/.+\.jpg',
'timestamp': 1538391166, 'timestamp': 1538391166,
'upload_date': '20181001', 'upload_date': '20181001',
'uploader': 'Framasoft', 'uploader': 'Framasoft',
@ -1346,19 +1346,34 @@ class PeerTubeIE(InfoExtractor):
'view_count': int, 'view_count': int,
'like_count': int, 'like_count': int,
'dislike_count': int, 'dislike_count': int,
'tags': ['framasoft', 'peertube'], 'tags': 'count:2',
'categories': ['Science & Technology'], 'categories': ['Science & Technology'],
}, },
'expected_warnings': ['HTTP Error 400: Bad Request'],
'params': {'skip_download': 'm3u8'},
}, { }, {
'url': 'https://peertube2.cpy.re/w/122d093a-1ede-43bd-bd34-59d2931ffc5e', 'url': 'https://peertube2.cpy.re/w/122d093a-1ede-43bd-bd34-59d2931ffc5e',
'info_dict': { 'info_dict': {
'id': '122d093a-1ede-43bd-bd34-59d2931ffc5e', 'id': '122d093a-1ede-43bd-bd34-59d2931ffc5e',
'ext': 'mp4', 'ext': 'mp4',
'title': 'E2E tests', 'title': 'E2E tests',
'uploader_id': '37855', 'categories': ['Unknown'],
'channel': 'Main chocobozzz channel',
'channel_id': '5187',
'channel_url': 'https://peertube2.cpy.re/video-channels/chocobozzz_channel',
'description': 'md5:67daf92c833c41c95db874e18fcb2786',
'dislike_count': int,
'duration': 52,
'license': 'Unknown',
'like_count': int,
'tags': [],
'thumbnail': r're:https?://peertube2\.cpy\.re/lazy-static/thumbnails/.+\.jpg',
'timestamp': 1589276219, 'timestamp': 1589276219,
'upload_date': '20200512', 'upload_date': '20200512',
'uploader': 'chocobozzz', 'uploader': 'chocobozzz',
'uploader_id': '37855',
'uploader_url': 'https://peertube2.cpy.re/accounts/chocobozzz',
'view_count': int,
}, },
}, { }, {
'url': 'https://peertube2.cpy.re/w/3fbif9S3WmtTP8gGsC5HBd', 'url': 'https://peertube2.cpy.re/w/3fbif9S3WmtTP8gGsC5HBd',
@ -1366,10 +1381,23 @@ class PeerTubeIE(InfoExtractor):
'id': '3fbif9S3WmtTP8gGsC5HBd', 'id': '3fbif9S3WmtTP8gGsC5HBd',
'ext': 'mp4', 'ext': 'mp4',
'title': 'E2E tests', 'title': 'E2E tests',
'uploader_id': '37855', 'categories': ['Unknown'],
'channel': 'Main chocobozzz channel',
'channel_id': '5187',
'channel_url': 'https://peertube2.cpy.re/video-channels/chocobozzz_channel',
'description': 'md5:67daf92c833c41c95db874e18fcb2786',
'dislike_count': int,
'duration': 52,
'license': 'Unknown',
'like_count': int,
'tags': [],
'thumbnail': r're:https?://peertube2\.cpy\.re/lazy-static/thumbnails/.+\.jpg',
'timestamp': 1589276219, 'timestamp': 1589276219,
'upload_date': '20200512', 'upload_date': '20200512',
'uploader': 'chocobozzz', 'uploader': 'chocobozzz',
'uploader_id': '37855',
'uploader_url': 'https://peertube2.cpy.re/accounts/chocobozzz',
'view_count': int,
}, },
}, { }, {
'url': 'https://peertube2.cpy.re/api/v1/videos/3fbif9S3WmtTP8gGsC5HBd', 'url': 'https://peertube2.cpy.re/api/v1/videos/3fbif9S3WmtTP8gGsC5HBd',
@ -1377,13 +1405,26 @@ class PeerTubeIE(InfoExtractor):
'id': '3fbif9S3WmtTP8gGsC5HBd', 'id': '3fbif9S3WmtTP8gGsC5HBd',
'ext': 'mp4', 'ext': 'mp4',
'title': 'E2E tests', 'title': 'E2E tests',
'uploader_id': '37855', 'categories': ['Unknown'],
'channel': 'Main chocobozzz channel',
'channel_id': '5187',
'channel_url': 'https://peertube2.cpy.re/video-channels/chocobozzz_channel',
'description': 'md5:67daf92c833c41c95db874e18fcb2786',
'dislike_count': int,
'duration': 52,
'license': 'Unknown',
'like_count': int,
'tags': [],
'thumbnail': r're:https?://peertube2\.cpy\.re/lazy-static/thumbnails/.+\.jpg',
'timestamp': 1589276219, 'timestamp': 1589276219,
'upload_date': '20200512', 'upload_date': '20200512',
'uploader': 'chocobozzz', 'uploader': 'chocobozzz',
'uploader_id': '37855',
'uploader_url': 'https://peertube2.cpy.re/accounts/chocobozzz',
'view_count': int,
}, },
}, { }, {
# Issue #26002 # https://github.com/ytdl-org/youtube-dl/issues/26002
'url': 'peertube:spacepub.space:d8943b2d-8280-497b-85ec-bc282ec2afdc', 'url': 'peertube:spacepub.space:d8943b2d-8280-497b-85ec-bc282ec2afdc',
'info_dict': { 'info_dict': {
'id': 'd8943b2d-8280-497b-85ec-bc282ec2afdc', 'id': 'd8943b2d-8280-497b-85ec-bc282ec2afdc',
@ -1394,6 +1435,7 @@ class PeerTubeIE(InfoExtractor):
'upload_date': '20200420', 'upload_date': '20200420',
'uploader': 'Drew DeVault', 'uploader': 'Drew DeVault',
}, },
'skip': 'Invalid URL',
}, { }, {
'url': 'https://peertube.debian.social/videos/watch/0b04f13d-1e18-4f1d-814e-4979aa7c9c44', 'url': 'https://peertube.debian.social/videos/watch/0b04f13d-1e18-4f1d-814e-4979aa7c9c44',
'only_matching': True, 'only_matching': True,
@ -1411,6 +1453,33 @@ class PeerTubeIE(InfoExtractor):
'url': 'peertube:framatube.org:b37a5b9f-e6b5-415c-b700-04a5cd6ec205', 'url': 'peertube:framatube.org:b37a5b9f-e6b5-415c-b700-04a5cd6ec205',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
'url': 'https://video.macver.org/w/6gvhZpUGQVd4SQ6oYDc9pC',
'info_dict': {
'id': '6gvhZpUGQVd4SQ6oYDc9pC',
'ext': 'mp4',
'title': 'Minecraft, but if you say a block, it gets deleted',
'categories': ['Gaming'],
'channel': 'Waffle Irons Gaming',
'channel_id': '4',
'channel_url': 'https://video.macver.org/video-channels/waffle_irons',
'description': 'md5:eda8daf64b0dadd00cc248f28eef213c',
'dislike_count': int,
'duration': 1643,
'license': 'Attribution - Non Commercial',
'like_count': int,
'tags': 'count:1',
'thumbnail': r're:https?://video\.macver\.org/lazy-static/thumbnails/.+\.jpg',
'timestamp': 1751142352,
'upload_date': '20250628',
'uploader': 'Bog',
'uploader_id': '3',
'uploader_url': 'https://video.macver.org/accounts/bog',
'view_count': int,
},
'expected_warnings': ['HTTP Error 400: Bad Request', 'Ignoring subtitle tracks found in the HLS manifest'],
'params': {'skip_download': 'm3u8'},
}]
@staticmethod @staticmethod
def _extract_peertube_url(webpage, source_url): def _extract_peertube_url(webpage, source_url):
@ -1580,31 +1649,47 @@ class PeerTubePlaylistIE(InfoExtractor):
'id': 'hFdJoTuyhNJVa1cDWd1d12', 'id': 'hFdJoTuyhNJVa1cDWd1d12',
'description': 'Diversas palestras do Richard Stallman no Brasil.', 'description': 'Diversas palestras do Richard Stallman no Brasil.',
'title': 'Richard Stallman no Brasil', 'title': 'Richard Stallman no Brasil',
'channel': 'debianbrazilteam',
'channel_id': 1522,
'thumbnail': r're:https?://peertube\.debian\.social/lazy-static/thumbnails/.+\.jpg',
'timestamp': 1599676222, 'timestamp': 1599676222,
'upload_date': '20200909',
}, },
'playlist_mincount': 9, 'playlist_mincount': 9,
}, { }, {
'url': 'https://peertube2.cpy.re/a/chocobozzz/videos', 'url': 'https://peertube2.cpy.re/a/chocobozzz/videos',
'info_dict': { 'info_dict': {
'id': 'chocobozzz', 'id': 'chocobozzz',
'timestamp': 1553874564,
'title': 'chocobozzz', 'title': 'chocobozzz',
'channel': 'chocobozzz',
'channel_id': 37855,
'thumbnail': '',
'timestamp': 1553874564,
'upload_date': '20190329',
}, },
'playlist_mincount': 2, 'playlist_mincount': 2,
}, { }, {
'url': 'https://framatube.org/c/bf54d359-cfad-4935-9d45-9d6be93f63e8/videos', 'url': 'https://framatube.org/c/bf54d359-cfad-4935-9d45-9d6be93f63e8/videos',
'info_dict': { 'info_dict': {
'id': 'bf54d359-cfad-4935-9d45-9d6be93f63e8', 'id': 'bf54d359-cfad-4935-9d45-9d6be93f63e8',
'timestamp': 1519917377,
'title': 'Les vidéos de Framasoft', 'title': 'Les vidéos de Framasoft',
'channel': 'framasoft',
'channel_id': 3,
'thumbnail': '',
'timestamp': 1519917377,
'upload_date': '20180301',
}, },
'playlist_mincount': 345, 'playlist_mincount': 345,
}, { }, {
'url': 'https://peertube2.cpy.re/c/blender_open_movies@video.blender.org/videos', 'url': 'https://peertube2.cpy.re/c/blender_open_movies@video.blender.org/videos',
'info_dict': { 'info_dict': {
'id': 'blender_open_movies@video.blender.org', 'id': 'blender_open_movies@video.blender.org',
'timestamp': 1542287810,
'title': 'Official Blender Open Movies', 'title': 'Official Blender Open Movies',
'channel': 'blender',
'channel_id': 1926,
'thumbnail': '',
'timestamp': 1540472902,
'upload_date': '20181025',
}, },
'playlist_mincount': 11, 'playlist_mincount': 11,
}] }]

15
yt_dlp/extractor/playwire.py
Unescape Escape View File

@ -19,6 +19,7 @@ class PlaywireIE(InfoExtractor):
'thumbnail': r're:^https?://.*\.png$', 'thumbnail': r're:^https?://.*\.png$',
'duration': 145.94, 'duration': 145.94,
}, },
'skip': 'Invalid URL',
}, { }, {
# m3u8 in f4m # m3u8 in f4m
'url': 'http://config.playwire.com/21772/videos/v2/4840492/zeus.json', 'url': 'http://config.playwire.com/21772/videos/v2/4840492/zeus.json',
@ -27,10 +28,7 @@ class PlaywireIE(InfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'title': 'ITV EL SHOW FULL', 'title': 'ITV EL SHOW FULL',
}, },
'params': { 'skip': 'Invalid URL',
# m3u8 download
'skip_download': True,
},
}, { }, {
# Multiple resolutions while bitrates missing # Multiple resolutions while bitrates missing
'url': 'http://cdn.playwire.com/11625/embed/85228.html', 'url': 'http://cdn.playwire.com/11625/embed/85228.html',
@ -42,6 +40,15 @@ class PlaywireIE(InfoExtractor):
'url': 'http://cdn.playwire.com/v2/12342/config/1532636.json', 'url': 'http://cdn.playwire.com/v2/12342/config/1532636.json',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
'url': 'https://www.cinemablend.com/new/First-Joe-Dirt-2-Trailer-Teaser-Stupid-Greatness-70874.html',
'info_dict': {
'id': '3519514',
'ext': 'mp4',
'title': 'Joe Dirt 2 Beautiful Loser Teaser Trailer',
},
'skip': 'Site no longer embeds Playwire',
}]
def _real_extract(self, url): def _real_extract(self, url):
mobj = self._match_valid_url(url) mobj = self._match_valid_url(url)

43
yt_dlp/extractor/rutube.py
Unescape Escape View File

@ -115,7 +115,6 @@ class RutubeIE(RutubeBaseIE):
_TESTS = [{ _TESTS = [{
'url': 'https://rutube.ru/video/3eac3b4561676c17df9132a9a1e62e3e/', 'url': 'https://rutube.ru/video/3eac3b4561676c17df9132a9a1e62e3e/',
'md5': '3d73fdfe5bb81b9aef139e22ef3de26a',
'info_dict': { 'info_dict': {
'id': '3eac3b4561676c17df9132a9a1e62e3e', 'id': '3eac3b4561676c17df9132a9a1e62e3e',
'ext': 'mp4', 'ext': 'mp4',
@ -128,10 +127,11 @@ class RutubeIE(RutubeBaseIE):
'upload_date': '20131016', 'upload_date': '20131016',
'age_limit': 0, 'age_limit': 0,
'view_count': int, 'view_count': int,
'thumbnail': 'https://pic.rutubelist.ru/video/d2/a0/d2a0aec998494a396deafc7ba2c82add.jpg', 'thumbnail': r're:https?://pic\.rutubelist\.ru/video/.+\.(?:jpg|png)',
'categories': ['Новости и СМИ'], 'categories': ['Новости и СМИ'],
'chapters': [], 'chapters': [],
}, },
'params': {'skip_download': 'm3u8'},
}, { }, {
'url': 'https://rutube.ru/play/embed/a10e53b86e8f349080f718582ce4c661', 'url': 'https://rutube.ru/play/embed/a10e53b86e8f349080f718582ce4c661',
'only_matching': True, 'only_matching': True,
@ -146,7 +146,6 @@ class RutubeIE(RutubeBaseIE):
'only_matching': True, 'only_matching': True,
}, { }, {
'url': 'https://rutube.ru/video/private/884fb55f07a97ab673c7d654553e0f48/?p=x2QojCumHTS3rsKHWXN8Lg', 'url': 'https://rutube.ru/video/private/884fb55f07a97ab673c7d654553e0f48/?p=x2QojCumHTS3rsKHWXN8Lg',
'md5': '4fce7b4fcc7b1bcaa3f45eb1e1ad0dd7',
'info_dict': { 'info_dict': {
'id': '884fb55f07a97ab673c7d654553e0f48', 'id': '884fb55f07a97ab673c7d654553e0f48',
'ext': 'mp4', 'ext': 'mp4',
@ -163,6 +162,7 @@ class RutubeIE(RutubeBaseIE):
'categories': ['Видеоигры'], 'categories': ['Видеоигры'],
'chapters': [], 'chapters': [],
}, },
'params': {'skip_download': 'm3u8'},
}, { }, {
'url': 'https://rutube.ru/video/c65b465ad0c98c89f3b25cb03dcc87c6/', 'url': 'https://rutube.ru/video/c65b465ad0c98c89f3b25cb03dcc87c6/',
'info_dict': { 'info_dict': {
@ -171,7 +171,7 @@ class RutubeIE(RutubeBaseIE):
'chapters': 'count:4', 'chapters': 'count:4',
'categories': ['Бизнес и предпринимательство'], 'categories': ['Бизнес и предпринимательство'],
'description': 'md5:252feac1305257d8c1bab215cedde75d', 'description': 'md5:252feac1305257d8c1bab215cedde75d',
'thumbnail': 'https://pic.rutubelist.ru/video/71/8f/718f27425ea9706073eb80883dd3787b.png', 'thumbnail': r're:https?://pic\.rutubelist\.ru/video/.+\.(?:jpg|png)',
'duration': 782, 'duration': 782,
'age_limit': 0, 'age_limit': 0,
'uploader_id': '23491359', 'uploader_id': '23491359',
@ -181,6 +181,7 @@ class RutubeIE(RutubeBaseIE):
'title': 'Бизнес с нуля: найм сотрудников. Интервью с директором строительной компании #1', 'title': 'Бизнес с нуля: найм сотрудников. Интервью с директором строительной компании #1',
'uploader': 'Стас Быков', 'uploader': 'Стас Быков',
}, },
'params': {'skip_download': 'm3u8'},
}, { }, {
'url': 'https://rutube.ru/live/video/c58f502c7bb34a8fcdd976b221fca292/', 'url': 'https://rutube.ru/live/video/c58f502c7bb34a8fcdd976b221fca292/',
'info_dict': { 'info_dict': {
@ -188,16 +189,17 @@ class RutubeIE(RutubeBaseIE):
'ext': 'mp4', 'ext': 'mp4',
'categories': ['Телепередачи'], 'categories': ['Телепередачи'],
'description': '', 'description': '',
'thumbnail': 'https://pic.rutubelist.ru/video/14/19/14190807c0c48b40361aca93ad0867c7.jpg', 'thumbnail': r're:https?://pic\.rutubelist\.ru/video/.+\.(?:jpg|png)',
'live_status': 'is_live', 'live_status': 'is_live',
'age_limit': 0, 'age_limit': 0,
'uploader_id': '23460655', 'uploader_id': '23460655',
'timestamp': 1652972968, 'timestamp': 1652972968,
'view_count': int, 'view_count': int,
'upload_date': '20220519', 'upload_date': '20220519',
'title': r're:Первый канал. Прямой эфир \d{4}-\d{2}-\d{2} \d{2}:\d{2}$', 'title': str,
'uploader': 'Первый канал', 'uploader': 'Первый канал',
}, },
'skip': 'Invalid URL',
}, { }, {
'url': 'https://rutube.ru/play/embed/03a9cb54bac3376af4c5cb0f18444e01/', 'url': 'https://rutube.ru/play/embed/03a9cb54bac3376af4c5cb0f18444e01/',
'info_dict': { 'info_dict': {
@ -211,11 +213,12 @@ class RutubeIE(RutubeBaseIE):
'duration': 293, 'duration': 293,
'uploader': 'MOEX - Московская биржа', 'uploader': 'MOEX - Московская биржа',
'timestamp': 1724946628, 'timestamp': 1724946628,
'thumbnail': 'https://pic.rutubelist.ru/video/2e/24/2e241fddb459baf0fa54acfca44874f4.jpg', 'thumbnail': r're:https?://pic\.rutubelist\.ru/video/.+\.(?:jpg|png)',
'view_count': int, 'view_count': int,
'uploader_id': '38420507', 'uploader_id': '38420507',
'categories': ['Интервью'], 'categories': ['Интервью'],
}, },
'params': {'skip_download': 'm3u8'},
}, { }, {
'url': 'https://rutube.ru/video/5ab908fccfac5bb43ef2b1e4182256b0/', 'url': 'https://rutube.ru/video/5ab908fccfac5bb43ef2b1e4182256b0/',
'only_matching': True, 'only_matching': True,
@ -223,6 +226,26 @@ class RutubeIE(RutubeBaseIE):
'url': 'https://rutube.ru/live/video/private/c58f502c7bb34a8fcdd976b221fca292/', 'url': 'https://rutube.ru/live/video/private/c58f502c7bb34a8fcdd976b221fca292/',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
'url': 'https://novate.ru/blogs/170625/73644/',
'info_dict': {
'id': 'b0c96c75a4e5b274721bbced6ed8fb64',
'ext': 'mp4',
'title': 'Где в России находится единственная в своем роде скальная торпедная батарея',
'age_limit': 0,
'categories': ['Наука'],
'chapters': [],
'description': 'md5:2ed82e6b81958a43da6fb4d56f949e1f',
'duration': 182,
'thumbnail': r're:https?://pic\.rutubelist\.ru/video/.+\.(?:jpg|png)',
'timestamp': 1749950158,
'upload_date': '20250615',
'uploader': 'Novate',
'uploader_id': '24044809',
'view_count': int,
},
'params': {'skip_download': 'm3u8'},
}]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)
@ -256,12 +279,10 @@ class RutubeEmbedIE(RutubeBaseIE):
'chapters': [], 'chapters': [],
'description': 'md5:a5acea57bbc3ccdc3cacd1f11a014b5b', 'description': 'md5:a5acea57bbc3ccdc3cacd1f11a014b5b',
'view_count': int, 'view_count': int,
'thumbnail': 'https://pic.rutubelist.ru/video/d3/03/d3031f4670a6e6170d88fb3607948418.jpg', 'thumbnail': r're:https?://pic\.rutubelist\.ru/video/.+\.(?:jpg|png)',
'categories': ['Сериалы'], 'categories': ['Сериалы'],
}, },
'params': { 'params': {'skip_download': 'm3u8'},
'skip_download': True,
},
}, { }, {
'url': 'https://rutube.ru/play/embed/8083783', 'url': 'https://rutube.ru/play/embed/8083783',
'only_matching': True, 'only_matching': True,

150
yt_dlp/extractor/rutv.py
Unescape Escape View File

@ -16,96 +16,88 @@ class RUTVIE(InfoExtractor):
) )
(?P<id>\d+) (?P<id>\d+)
''' '''
_EMBED_URLS = [ _EMBED_REGEX = [
r'<iframe[^>]+?src=(["\'])(?P<url>https?://(?:test)?player\.(?:rutv\.ru|vgtrk\.com)/(?:iframe/(?:swf|video|live)/id|index/iframe/cast_id)/.+?)\1', r'<iframe[^>]+?src=(["\'])(?P<url>https?://(?:test)?player\.(?:rutv\.ru|vgtrk\.com)/(?:iframe/(?:swf|video|live)/id|index/iframe/cast_id)/.+?)\1',
r'<meta[^>]+?property=(["\'])og:video\1[^>]+?content=(["\'])(?P<url>https?://(?:test)?player\.(?:rutv\.ru|vgtrk\.com)/flash\d+v/container\.swf\?id=.+?\2)', r'<meta[^>]+?property=(["\'])og:video\1[^>]+?content=(["\'])(?P<url>https?://(?:test)?player\.(?:rutv\.ru|vgtrk\.com)/flash\d+v/container\.swf\?id=.+?\2)',
] ]
_TESTS = [ _TESTS = [{
{ 'url': 'http://player.rutv.ru/flash2v/container.swf?id=774471&sid=kultura&fbv=true&isPlay=true&ssl=false&i=560&acc_video_id=episode_id/972347/video_id/978186/brand_id/31724',
'url': 'http://player.rutv.ru/flash2v/container.swf?id=774471&sid=kultura&fbv=true&isPlay=true&ssl=false&i=560&acc_video_id=episode_id/972347/video_id/978186/brand_id/31724', 'info_dict': {
'info_dict': { 'id': '774471',
'id': '774471', 'ext': 'mp4',
'ext': 'mp4', 'title': 'Монологи на все времена. Концерт',
'title': 'Монологи на все времена', 'description': 'md5:18d8b5e6a41fb1faa53819471852d5d5',
'description': 'md5:18d8b5e6a41fb1faa53819471852d5d5', 'duration': 2906,
'duration': 2906, 'thumbnail': r're:https?://cdn-st2\.smotrim\.ru/.+\.jpg',
},
'params': {
# m3u8 download
'skip_download': True,
},
}, },
{ 'params': {'skip_download': 'm3u8'},
'url': 'https://player.vgtrk.com/flash2v/container.swf?id=774016&sid=russiatv&fbv=true&isPlay=true&ssl=false&i=560&acc_video_id=episode_id/972098/video_id/977760/brand_id/57638', }, {
'info_dict': { 'url': 'https://player.vgtrk.com/flash2v/container.swf?id=774016&sid=russiatv&fbv=true&isPlay=true&ssl=false&i=560&acc_video_id=episode_id/972098/video_id/977760/brand_id/57638',
'id': '774016', 'info_dict': {
'ext': 'mp4', 'id': '774016',
'title': 'Чужой в семье Сталина', 'ext': 'mp4',
'description': '', 'title': 'Чужой в семье Сталина',
'duration': 2539, 'description': '',
}, 'duration': 2539,
'params': {
# m3u8 download
'skip_download': True,
},
}, },
{ 'skip': 'Invalid URL',
'url': 'http://player.rutv.ru/iframe/swf/id/766888/sid/hitech/?acc_video_id=4000', }, {
'info_dict': { 'url': 'http://player.rutv.ru/iframe/swf/id/766888/sid/hitech/?acc_video_id=4000',
'id': '766888', 'info_dict': {
'ext': 'mp4', 'id': '766888',
'title': 'Вести.net: интернет-гиганты начали перетягивание программных "одеял"', 'ext': 'mp4',
'description': 'md5:65ddd47f9830c4f42ed6475f8730c995', 'title': 'Вести.net: интернет-гиганты начали перетягивание программных "одеял"',
'duration': 279, 'description': 'md5:65ddd47f9830c4f42ed6475f8730c995',
}, 'duration': 279,
'params': { 'thumbnail': r're:https?://cdn-st2\.smotrim\.ru/.+\.jpg',
# m3u8 download
'skip_download': True,
},
}, },
{ 'params': {'skip_download': 'm3u8'},
'url': 'http://player.rutv.ru/iframe/video/id/771852/start_zoom/true/showZoomBtn/false/sid/russiatv/?acc_video_id=episode_id/970443/video_id/975648/brand_id/5169', }, {
'info_dict': { 'url': 'http://player.rutv.ru/iframe/video/id/771852/start_zoom/true/showZoomBtn/false/sid/russiatv/?acc_video_id=episode_id/970443/video_id/975648/brand_id/5169',
'id': '771852', 'info_dict': {
'ext': 'mp4', 'id': '771852',
'title': 'Прямой эфир. Жертвы загадочной болезни: смерть от старости в 17 лет', 'ext': 'mp4',
'description': 'md5:b81c8c55247a4bd996b43ce17395b2d8', 'title': 'Прямой эфир. Жертвы загадочной болезни: смерть от старости в 17 лет',
'duration': 3096, 'description': 'md5:b81c8c55247a4bd996b43ce17395b2d8',
}, 'duration': 3096,
'params': { 'thumbnail': r're:https?://cdn-st2\.smotrim\.ru/.+\.jpg',
# m3u8 download
'skip_download': True,
},
}, },
{ 'params': {'skip_download': 'm3u8'},
'url': 'http://player.rutv.ru/iframe/live/id/51499/showZoomBtn/false/isPlay/true/sid/sochi2014', }, {
'info_dict': { 'url': 'http://player.rutv.ru/iframe/live/id/51499/showZoomBtn/false/isPlay/true/sid/sochi2014',
'id': '51499', 'info_dict': {
'ext': 'flv', 'id': '51499',
'title': 'Сочи-2014. Биатлон. Индивидуальная гонка. Мужчины ', 'ext': 'flv',
'description': 'md5:9e0ed5c9d2fa1efbfdfed90c9a6d179c', 'title': 'Сочи-2014. Биатлон. Индивидуальная гонка. Мужчины ',
}, 'description': 'md5:9e0ed5c9d2fa1efbfdfed90c9a6d179c',
'skip': 'Translation has finished',
}, },
{ 'skip': 'Invalid URL',
'url': 'http://player.rutv.ru/iframe/live/id/21/showZoomBtn/false/isPlay/true/', }, {
'info_dict': { 'url': 'http://player.rutv.ru/iframe/live/id/21/showZoomBtn/false/isPlay/true/',
'id': '21', 'info_dict': {
'ext': 'mp4', 'id': '21',
'title': 're:^Россия 24. Прямой эфир [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$', 'ext': 'mp4',
'is_live': True, 'title': str,
}, 'is_live': True,
'params': {
# m3u8 download
'skip_download': True,
},
}, },
{ 'skip': 'Invalid URL',
'url': 'https://testplayer.vgtrk.com/iframe/live/id/19201/showZoomBtn/false/isPlay/true/', }, {
'only_matching': True, 'url': 'https://testplayer.vgtrk.com/iframe/live/id/19201/showZoomBtn/false/isPlay/true/',
'only_matching': True,
}]
_WEBPAGE_TESTS = [{
'url': 'http://istoriya-teatra.ru/news/item/f00/s05/n0000545/index.shtml',
'info_dict': {
'id': '1952012',
'ext': 'mp4',
'title': 'Новости культуры. Эфир от 10.10.2019 (23:30). Театр Сатиры отмечает день рождения премьерой',
'description': 'md5:fced27112ff01ff8fc4a452fc088bad6',
'duration': 191,
'thumbnail': r're:https?://cdn-st2\.smotrim\.ru/.+\.jpg',
}, },
] 'params': {'skip_download': 'm3u8'},
}]
def _real_extract(self, url): def _real_extract(self, url):
mobj = self._match_valid_url(url) mobj = self._match_valid_url(url)

194
yt_dlp/extractor/ruutu.py
Unescape Escape View File

@ -18,6 +18,7 @@ from ..utils import (
class RuutuIE(InfoExtractor): class RuutuIE(InfoExtractor):
_WORKING = False
_VALID_URL = r'''(?x) _VALID_URL = r'''(?x)
https?:// https?://
(?: (?:
@ -26,112 +27,111 @@ class RuutuIE(InfoExtractor):
) )
(?P<id>\d+) (?P<id>\d+)
''' '''
_TESTS = [ _TESTS = [{
{ 'url': 'http://www.ruutu.fi/video/2058907',
'url': 'http://www.ruutu.fi/video/2058907', 'md5': 'ab2093f39be1ca8581963451b3c0234f',
'md5': 'ab2093f39be1ca8581963451b3c0234f', 'info_dict': {
'info_dict': { 'id': '2058907',
'id': '2058907', 'ext': 'mp4',
'ext': 'mp4', 'title': 'Oletko aina halunnut tietää mitä tapahtuu vain hetki ennen lähetystä? - Nyt se selvisi!',
'title': 'Oletko aina halunnut tietää mitä tapahtuu vain hetki ennen lähetystä? - Nyt se selvisi!', 'description': 'md5:cfc6ccf0e57a814360df464a91ff67d6',
'description': 'md5:cfc6ccf0e57a814360df464a91ff67d6', 'thumbnail': r're:^https?://.*\.jpg$',
'thumbnail': r're:^https?://.*\.jpg$', 'duration': 114,
'duration': 114, 'age_limit': 0,
'age_limit': 0, 'upload_date': '20150508',
'upload_date': '20150508',
},
}, },
{ }, {
'url': 'http://www.ruutu.fi/video/2057306', 'url': 'http://www.ruutu.fi/video/2057306',
'md5': '065a10ae4d5b8cfd9d0c3d332465e3d9', 'md5': '065a10ae4d5b8cfd9d0c3d332465e3d9',
'info_dict': { 'info_dict': {
'id': '2057306', 'id': '2057306',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Superpesis: katso koko kausi Ruudussa', 'title': 'Superpesis: katso koko kausi Ruudussa',
'description': 'md5:bfb7336df2a12dc21d18fa696c9f8f23', 'description': 'md5:bfb7336df2a12dc21d18fa696c9f8f23',
'thumbnail': r're:^https?://.*\.jpg$', 'thumbnail': r're:^https?://.*\.jpg$',
'duration': 40, 'duration': 40,
'age_limit': 0, 'age_limit': 0,
'upload_date': '20150507', 'upload_date': '20150507',
'series': 'Superpesis', 'series': 'Superpesis',
'categories': ['Urheilu'], 'categories': ['Urheilu'],
},
}, },
{ }, {
'url': 'http://www.supla.fi/supla/2231370', 'url': 'http://www.supla.fi/supla/2231370',
'md5': 'df14e782d49a2c0df03d3be2a54ef949', 'md5': 'df14e782d49a2c0df03d3be2a54ef949',
'info_dict': { 'info_dict': {
'id': '2231370', 'id': '2231370',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Osa 1: Mikael Jungner', 'title': 'Osa 1: Mikael Jungner',
'description': 'md5:7d90f358c47542e3072ff65d7b1bcffe', 'description': 'md5:7d90f358c47542e3072ff65d7b1bcffe',
'thumbnail': r're:^https?://.*\.jpg$', 'thumbnail': r're:^https?://.*\.jpg$',
'age_limit': 0, 'age_limit': 0,
'upload_date': '20151012', 'upload_date': '20151012',
'series': 'Läpivalaisu', 'series': 'Läpivalaisu',
},
}, },
}, {
# Episode where <SourceFile> is "NOT-USED", but has other # Episode where <SourceFile> is "NOT-USED", but has other
# downloadable sources available. # downloadable sources available.
{ 'url': 'http://www.ruutu.fi/video/3193728',
'url': 'http://www.ruutu.fi/video/3193728', 'only_matching': True,
'only_matching': True, }, {
# audio podcast
'url': 'https://www.supla.fi/supla/3382410',
'md5': 'b9d7155fed37b2ebf6021d74c4b8e908',
'info_dict': {
'id': '3382410',
'ext': 'mp3',
'title': 'Mikä ihmeen poltergeist?',
'description': 'md5:bbb6963df17dfd0ecd9eb9a61bf14b52',
'thumbnail': r're:^https?://.*\.jpg$',
'age_limit': 0,
'upload_date': '20190320',
'series': 'Mysteeritarinat',
'duration': 1324,
}, },
{ 'expected_warnings': [
# audio podcast 'HTTP Error 502: Bad Gateway',
'url': 'https://www.supla.fi/supla/3382410', 'Failed to download m3u8 information',
'md5': 'b9d7155fed37b2ebf6021d74c4b8e908', ],
'info_dict': { }, {
'id': '3382410', 'url': 'http://www.supla.fi/audio/2231370',
'ext': 'mp3', 'only_matching': True,
'title': 'Mikä ihmeen poltergeist?', }, {
'description': 'md5:bbb6963df17dfd0ecd9eb9a61bf14b52', 'url': 'https://static.nelonenmedia.fi/player/misc/embed_player.html?nid=3618790',
'thumbnail': r're:^https?://.*\.jpg$', 'only_matching': True,
'age_limit': 0, }, {
'upload_date': '20190320', # episode
'series': 'Mysteeritarinat', 'url': 'https://www.ruutu.fi/video/3401964',
'duration': 1324, 'info_dict': {
}, 'id': '3401964',
'expected_warnings': [ 'ext': 'mp4',
'HTTP Error 502: Bad Gateway', 'title': 'Temptation Island Suomi - Kausi 5 - Jakso 17',
'Failed to download m3u8 information', 'description': 'md5:87cf01d5e1e88adf0c8a2937d2bd42ba',
], 'thumbnail': r're:^https?://.*\.jpg$',
'duration': 2582,
'age_limit': 12,
'upload_date': '20190508',
'series': 'Temptation Island Suomi',
'season_number': 5,
'episode_number': 17,
'categories': ['Reality ja tositapahtumat', 'Kotimaiset suosikit', 'Romantiikka ja parisuhde'],
}, },
{ 'params': {
'url': 'http://www.supla.fi/audio/2231370', 'skip_download': True,
'only_matching': True,
}, },
{ }, {
'url': 'https://static.nelonenmedia.fi/player/misc/embed_player.html?nid=3618790', # premium
'only_matching': True, 'url': 'https://www.ruutu.fi/video/3618715',
'only_matching': True,
}]
_WEBPAGE_TESTS = [{
# FIXME: Broken IE
'url': 'https://www.hs.fi/maailma/art-2000011353059.html',
'info_dict': {
'id': '4746675',
'ext': 'mp4',
'title': 'Yhdysvaltojen Texasin osavaltiota ovat koetelleet tuhoisat tulvat',
}, },
{ }]
# episode
'url': 'https://www.ruutu.fi/video/3401964',
'info_dict': {
'id': '3401964',
'ext': 'mp4',
'title': 'Temptation Island Suomi - Kausi 5 - Jakso 17',
'description': 'md5:87cf01d5e1e88adf0c8a2937d2bd42ba',
'thumbnail': r're:^https?://.*\.jpg$',
'duration': 2582,
'age_limit': 12,
'upload_date': '20190508',
'series': 'Temptation Island Suomi',
'season_number': 5,
'episode_number': 17,
'categories': ['Reality ja tositapahtumat', 'Kotimaiset suosikit', 'Romantiikka ja parisuhde'],
},
'params': {
'skip_download': True,
},
},
{
# premium
'url': 'https://www.ruutu.fi/video/3618715',
'only_matching': True,
},
]
_API_BASE = 'https://gatling.nelonenmedia.fi' _API_BASE = 'https://gatling.nelonenmedia.fi'
@classmethod @classmethod

32
yt_dlp/extractor/senategov.py
Unescape Escape View File

@ -23,13 +23,10 @@ class SenateISVPIE(InfoExtractor):
'id': 'judiciary031715', 'id': 'judiciary031715',
'ext': 'mp4', 'ext': 'mp4',
'title': 'ISVP', 'title': 'ISVP',
'thumbnail': r're:^https?://.*\.(?:jpg|png)$', 'thumbnail': r're:https?://.+\.(?:jpe?g|png)',
'_old_archive_ids': ['senategov judiciary031715'], '_old_archive_ids': ['senategov judiciary031715'],
}, },
'params': { 'params': {'skip_download': 'm3u8'},
# m3u8 download
'skip_download': True,
},
'expected_warnings': ['Failed to download m3u8 information'], 'expected_warnings': ['Failed to download m3u8 information'],
}, { }, {
'url': 'http://www.senate.gov/isvp/?type=live&comm=commerce&filename=commerce011514.mp4&auto_play=false', 'url': 'http://www.senate.gov/isvp/?type=live&comm=commerce&filename=commerce011514.mp4&auto_play=false',
@ -39,10 +36,6 @@ class SenateISVPIE(InfoExtractor):
'title': 'Integrated Senate Video Player', 'title': 'Integrated Senate Video Player',
'_old_archive_ids': ['senategov commerce011514'], '_old_archive_ids': ['senategov commerce011514'],
}, },
'params': {
# m3u8 download
'skip_download': True,
},
'skip': 'This video is not available.', 'skip': 'This video is not available.',
}, { }, {
'url': 'http://www.senate.gov/isvp/?type=arch&comm=intel&filename=intel090613&hc_location=ufi', 'url': 'http://www.senate.gov/isvp/?type=arch&comm=intel&filename=intel090613&hc_location=ufi',
@ -60,7 +53,7 @@ class SenateISVPIE(InfoExtractor):
'id': 'help090920', 'id': 'help090920',
'ext': 'mp4', 'ext': 'mp4',
'title': 'ISVP', 'title': 'ISVP',
'thumbnail': 'https://www.help.senate.gov/assets/images/video-poster.png', 'thumbnail': r're:https?://.+\.(?:jpe?g|png)',
'_old_archive_ids': ['senategov help090920'], '_old_archive_ids': ['senategov help090920'],
}, },
}, { }, {
@ -68,6 +61,17 @@ class SenateISVPIE(InfoExtractor):
'url': 'http://www.senate.gov/isvp?type=live&comm=banking&filename=banking012715', 'url': 'http://www.senate.gov/isvp?type=live&comm=banking&filename=banking012715',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
# FIXME: Embed detection
'url': 'https://www.hsgac.senate.gov/subcommittees/bmfwra/hearings/match-ready-oversight-of-the-federal-governments-border-management-and-personnel-readiness-efforts-for-the-decade-of-sports/',
'info_dict': {
'id': 'govtaff061025',
'ext': 'mp4',
'title': 'ISVP',
'thumbnail': r're:https?://.+\.(?:jpe?g|png)',
'_old_archive_ids': ['senategov govtaff061025'],
},
}]
_COMMITTEES = { _COMMITTEES = {
'ag': ('76440', 'https://ag-f.akamaihd.net', '2036803', 'agriculture'), 'ag': ('76440', 'https://ag-f.akamaihd.net', '2036803', 'agriculture'),
@ -150,10 +154,10 @@ class SenateGovIE(InfoExtractor):
'id': 'help090920', 'id': 'help090920',
'display_id': 'vaccines-saving-lives-ensuring-confidence-and-protecting-public-health', 'display_id': 'vaccines-saving-lives-ensuring-confidence-and-protecting-public-health',
'title': 'Vaccines: Saving Lives, Ensuring Confidence, and Protecting Public Health', 'title': 'Vaccines: Saving Lives, Ensuring Confidence, and Protecting Public Health',
'description': 'The U.S. Senate Committee on Health, Education, Labor & Pensions', 'description': 'Full Committee Hearing on September 9, 2020 at 6:00 AM',
'ext': 'mp4', 'ext': 'mp4',
'age_limit': 0, 'age_limit': 0,
'thumbnail': 'https://www.help.senate.gov/assets/images/sharelogo.jpg', 'thumbnail': r're:https?://.+\.(?:jpe?g|png)',
'_old_archive_ids': ['senategov help090920'], '_old_archive_ids': ['senategov help090920'],
}, },
'params': {'skip_download': 'm3u8'}, 'params': {'skip_download': 'm3u8'},
@ -165,7 +169,7 @@ class SenateGovIE(InfoExtractor):
'title': 'Review of the FY2019 Budget Request for the U.S. Army', 'title': 'Review of the FY2019 Budget Request for the U.S. Army',
'ext': 'mp4', 'ext': 'mp4',
'age_limit': 0, 'age_limit': 0,
'thumbnail': 'https://www.appropriations.senate.gov/themes/appropriations/images/video-poster-flash-fit.png', 'thumbnail': r're:https?://.+\.(?:jpe?g|png)',
'_old_archive_ids': ['senategov appropsA051518'], '_old_archive_ids': ['senategov appropsA051518'],
}, },
'params': {'skip_download': 'm3u8'}, 'params': {'skip_download': 'm3u8'},
@ -178,7 +182,7 @@ class SenateGovIE(InfoExtractor):
'title': '21st Century Communities: Public Transportation Infrastructure Investment and FAST Act Reauthorization', 'title': '21st Century Communities: Public Transportation Infrastructure Investment and FAST Act Reauthorization',
'description': 'The Official website of The United States Committee on Banking, Housing, and Urban Affairs', 'description': 'The Official website of The United States Committee on Banking, Housing, and Urban Affairs',
'ext': 'mp4', 'ext': 'mp4',
'thumbnail': 'https://www.banking.senate.gov/themes/banking/images/sharelogo.jpg', 'thumbnail': r're:https?://.+\.(?:jpe?g|png)',
'age_limit': 0, 'age_limit': 0,
'_old_archive_ids': ['senategov banking041521'], '_old_archive_ids': ['senategov banking041521'],
}, },

34
yt_dlp/extractor/simplecast.py
Unescape Escape View File

@ -76,17 +76,18 @@ class SimplecastIE(SimplecastBaseIE):
'id': 'b6dc49a2-9404-4853-9aa9-9cfc097be876', 'id': 'b6dc49a2-9404-4853-9aa9-9cfc097be876',
'ext': 'mp3', 'ext': 'mp3',
'title': 'Errant Signal - Chris Franklin & New Wave Video Essays', 'title': 'Errant Signal - Chris Franklin & New Wave Video Essays',
'channel_url': 'https://the-re-bind-io-podcast.simplecast.com',
'episode': 'Episode 1',
'episode_number': 1, 'episode_number': 1,
'episode_id': 'b6dc49a2-9404-4853-9aa9-9cfc097be876', 'episode_id': 'b6dc49a2-9404-4853-9aa9-9cfc097be876',
'description': 'md5:34752789d3d2702e2d2c975fbd14f357', 'description': 'md5:34752789d3d2702e2d2c975fbd14f357',
'season': 'Season 1',
'season_number': 1, 'season_number': 1,
'season_id': 'e23df0da-bae4-4531-8bbf-71364a88dc13', 'season_id': 'e23df0da-bae4-4531-8bbf-71364a88dc13',
'series': 'The RE:BIND.io Podcast', 'series': 'The RE:BIND.io Podcast',
'duration': 5343, 'duration': 5343,
'timestamp': 1580979475, 'timestamp': 1580979475,
'upload_date': '20200206', 'upload_date': '20200206',
'webpage_url': r're:^https?://the-re-bind-io-podcast\.simplecast\.com/episodes/errant-signal-chris-franklin-new-wave-video-essays',
'channel_url': r're:^https?://the-re-bind-io-podcast\.simplecast\.com$',
} }
_TESTS = [{ _TESTS = [{
'url': 'https://api.simplecast.com/episodes/b6dc49a2-9404-4853-9aa9-9cfc097be876', 'url': 'https://api.simplecast.com/episodes/b6dc49a2-9404-4853-9aa9-9cfc097be876',
@ -96,6 +97,29 @@ class SimplecastIE(SimplecastBaseIE):
'url': 'https://player.simplecast.com/b6dc49a2-9404-4853-9aa9-9cfc097be876', 'url': 'https://player.simplecast.com/b6dc49a2-9404-4853-9aa9-9cfc097be876',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
# FIXME: Embed detection
'url': 'https://poddtoppen.se/podcast/1498417306/the-rebindio-podcast/errant-signal-chris-franklin-new-wave-video-essays',
'md5': '8c93be7be54251bf29ee97464eabe61c',
'info_dict': {
'id': 'b6dc49a2-9404-4853-9aa9-9cfc097be876',
'ext': 'mp3',
'title': 'Errant Signal - Chris Franklin & New Wave Video Essays',
'channel_url': 'https://the-re-bind-io-podcast.simplecast.com',
'description': 'md5:34752789d3d2702e2d2c975fbd14f357',
'display_id': 'errant-signal-chris-franklin-new-wave-video-essays',
'duration': 5343,
'episode': 'Episode 1',
'episode_id': 'b6dc49a2-9404-4853-9aa9-9cfc097be876',
'episode_number': 1,
'season': 'Season 1',
'season_id': 'e23df0da-bae4-4531-8bbf-71364a88dc13',
'season_number': 1,
'series': 'The RE:BIND.io Podcast',
'timestamp': 1580979475,
'upload_date': '20200206',
},
}]
def _real_extract(self, url): def _real_extract(self, url):
episode_id = self._match_id(url) episode_id = self._match_id(url)
@ -106,11 +130,11 @@ class SimplecastIE(SimplecastBaseIE):
class SimplecastEpisodeIE(SimplecastBaseIE): class SimplecastEpisodeIE(SimplecastBaseIE):
IE_NAME = 'simplecast:episode' IE_NAME = 'simplecast:episode'
_VALID_URL = r'https?://(?!api\.)[^/]+\.simplecast\.com/episodes/(?P<id>[^/?&#]+)' _VALID_URL = r'https?://(?!api\.)[^/]+\.simplecast\.com/episodes/(?P<id>[^/?&#]+)'
_TEST = { _TESTS = [{
'url': 'https://the-re-bind-io-podcast.simplecast.com/episodes/errant-signal-chris-franklin-new-wave-video-essays', 'url': 'https://the-re-bind-io-podcast.simplecast.com/episodes/errant-signal-chris-franklin-new-wave-video-essays',
'md5': '8c93be7be54251bf29ee97464eabe61c', 'md5': '8c93be7be54251bf29ee97464eabe61c',
'info_dict': SimplecastIE._COMMON_TEST_INFO, 'info_dict': SimplecastIE._COMMON_TEST_INFO,
} }]
def _real_extract(self, url): def _real_extract(self, url):
mobj = self._match_valid_url(url) mobj = self._match_valid_url(url)
@ -124,7 +148,7 @@ class SimplecastPodcastIE(SimplecastBaseIE):
_VALID_URL = r'https?://(?!(?:api|cdn|embed|feeds|player)\.)(?P<id>[^/]+)\.simplecast\.com(?!/episodes/[^/?&#]+)' _VALID_URL = r'https?://(?!(?:api|cdn|embed|feeds|player)\.)(?P<id>[^/]+)\.simplecast\.com(?!/episodes/[^/?&#]+)'
_TESTS = [{ _TESTS = [{
'url': 'https://the-re-bind-io-podcast.simplecast.com', 'url': 'https://the-re-bind-io-podcast.simplecast.com',
'playlist_mincount': 33, 'playlist_mincount': 32,
'info_dict': { 'info_dict': {
'id': '07d28d26-7522-42eb-8c53-2bdcfc81c43c', 'id': '07d28d26-7522-42eb-8c53-2bdcfc81c43c',
'title': 'The RE:BIND.io Podcast', 'title': 'The RE:BIND.io Podcast',

527
yt_dlp/extractor/soundcloud.py
Unescape Escape View File

@ -26,11 +26,47 @@ from ..utils.traversal import traverse_obj
class SoundcloudEmbedIE(InfoExtractor): class SoundcloudEmbedIE(InfoExtractor):
_VALID_URL = r'https?://(?:w|player|p)\.soundcloud\.com/player/?.*?\burl=(?P<id>.+)' _VALID_URL = r'https?://(?:w|player|p)\.soundcloud\.com/player/?.*?\burl=(?P<id>.+)'
_EMBED_REGEX = [r'<iframe[^>]+src=(["\'])(?P<url>(?:https?://)?(?:w\.)?soundcloud\.com/player.+?)\1'] _EMBED_REGEX = [r'<iframe[^>]+src=(["\'])(?P<url>(?:https?://)?(?:w\.)?soundcloud\.com/player.+?)\1']
_TEST = { _TESTS = [{
# from https://www.soundi.fi/uutiset/ennakkokuuntelussa-timo-kaukolammen-station-to-station-to-station-julkaisua-juhlitaan-tanaan-g-livelabissa/ # from https://www.soundi.fi/uutiset/ennakkokuuntelussa-timo-kaukolammen-station-to-station-to-station-julkaisua-juhlitaan-tanaan-g-livelabissa/
'url': 'https://w.soundcloud.com/player/?visual=true&url=https%3A%2F%2Fapi.soundcloud.com%2Fplaylists%2F922213810&show_artwork=true&maxwidth=640&maxheight=960&dnt=1&secret_token=s-ziYey', 'url': 'https://w.soundcloud.com/player/?visual=true&url=https%3A%2F%2Fapi.soundcloud.com%2Fplaylists%2F922213810&show_artwork=true&maxwidth=640&maxheight=960&dnt=1&secret_token=s-ziYey',
'only_matching': True, 'only_matching': True,
} }]
_WEBPAGE_TESTS = [{
'url': 'https://news.sophos.com/en-us/2023/08/10/s3-ep147-what-if-you-type-in-your-password-during-a-meeting/',
'info_dict': {
'id': '1588847423',
'ext': 'm4a',
'title': 'S3 Ep147: What if you type in your password during a meeting?',
'artists': ['Naked Security'],
'description': 'md5:6931a0630b920413c8c904407bf4b3b2',
'duration': 942.762,
'genres': ['Technology'],
'license': 'all-rights-reserved',
'repost_count': int,
'tags': 'count:4',
'thumbnail': r're:https?://[ai]1\.sndcdn\.com/.+\.(?:jpg|png)',
'timestamp': 1691624365,
'track': 'S3 Ep147: What if you type in your password during a meeting?',
'upload_date': '20230809',
'uploader': 'Naked Security',
'uploader_id': '61390843',
'uploader_url': 'https://soundcloud.com/sophossecurity',
},
'params': {'skip_download': 'm3u8'},
}, {
'url': 'https://www.guitarplayer.com/lessons/november-2023-guitar-player-lesson-audio',
'info_dict': {
'id': '1695754080',
'title': 'A Tribute to Brian Setzer’s Guitar Mastery',
'album': 'A Tribute to Brian Setzer’s Guitar Mastery',
'album_artists': ['Guitar Player'],
'album_type': 'playlist',
'description': '',
'uploader': 'Guitar Player',
'uploader_id': '489924156',
},
'playlist_mincount': 7,
}]
def _real_extract(self, url): def _real_extract(self, url):
query = parse_qs(url) query = parse_qs(url)
@ -407,269 +443,256 @@ class SoundcloudIE(SoundcloudBaseIE):
) )
''' '''
IE_NAME = 'soundcloud' IE_NAME = 'soundcloud'
_TESTS = [ _TESTS = [{
{ 'url': 'http://soundcloud.com/ethmusic/lostin-powers-she-so-heavy',
'url': 'http://soundcloud.com/ethmusic/lostin-powers-she-so-heavy', 'md5': 'de9bac153e7427a7333b4b0c1b6a18d2',
'md5': 'de9bac153e7427a7333b4b0c1b6a18d2', 'info_dict': {
'info_dict': { 'id': '62986583',
'id': '62986583', 'ext': 'opus',
'ext': 'opus', 'title': 'Lostin Powers - She so Heavy (SneakPreview) Adrian Ackers Blueprint 1',
'title': 'Lostin Powers - She so Heavy (SneakPreview) Adrian Ackers Blueprint 1', 'track': 'Lostin Powers - She so Heavy (SneakPreview) Adrian Ackers Blueprint 1',
'track': 'Lostin Powers - She so Heavy (SneakPreview) Adrian Ackers Blueprint 1', 'description': 'md5:7b6074e00887ad79f59b647c8fb6d5ae',
'description': 'No Downloads untill we record the finished version this weekend, i was too pumped n i had to post it , earl is prolly gonna b hella p.o\'d', 'uploader': 'E.T. ExTerrestrial Music',
'uploader': 'E.T. ExTerrestrial Music', 'uploader_id': '1571244',
'uploader_id': '1571244', 'timestamp': 1349920598,
'timestamp': 1349920598, 'upload_date': '20121011',
'upload_date': '20121011', 'duration': 143.216,
'duration': 143.216, 'license': 'all-rights-reserved',
'license': 'all-rights-reserved', 'view_count': int,
'view_count': int, 'like_count': int,
'like_count': int, 'comment_count': int,
'comment_count': int, 'repost_count': int,
'repost_count': int, 'thumbnail': r're:https?://[ai]1\.sndcdn\.com/.+\.(?:jpg|png)',
'thumbnail': 'https://i1.sndcdn.com/artworks-000031955188-rwb18x-original.jpg', 'uploader_url': 'https://soundcloud.com/ethmusic',
'uploader_url': 'https://soundcloud.com/ethmusic', 'tags': 'count:14',
'tags': 'count:14',
},
}, },
# geo-restricted }, {
{ # Geo-restricted
'url': 'https://soundcloud.com/the-concept-band/goldrushed-mastered?in=the-concept-band/sets/the-royal-concept-ep', 'url': 'https://soundcloud.com/the-concept-band/goldrushed-mastered?in=the-concept-band/sets/the-royal-concept-ep',
'info_dict': { 'info_dict': {
'id': '47127627', 'id': '47127627',
'ext': 'opus', 'ext': 'opus',
'title': 'Goldrushed', 'title': 'Goldrushed',
'track': 'Goldrushed', 'track': 'Goldrushed',
'description': 'From Stockholm Sweden\r\nPovel / Magnus / Filip / David\r\nwww.theroyalconcept.com', 'description': 'md5:c0080b79a3710811d60234f94f391a40',
'uploader': 'The Royal Concept', 'uploader': 'The Royal Concept',
'uploader_id': '9615865', 'uploader_id': '9615865',
'timestamp': 1337635207, 'timestamp': 1337635207,
'upload_date': '20120521', 'upload_date': '20120521',
'duration': 227.103, 'duration': 227.103,
'license': 'all-rights-reserved', 'license': 'all-rights-reserved',
'view_count': int, 'view_count': int,
'like_count': int, 'like_count': int,
'comment_count': int, 'comment_count': int,
'repost_count': int, 'repost_count': int,
'uploader_url': 'https://soundcloud.com/the-concept-band', 'uploader_url': 'https://soundcloud.com/the-concept-band',
'thumbnail': 'https://i1.sndcdn.com/artworks-v8bFHhXm7Au6-0-original.jpg', 'thumbnail': r're:https?://[ai]1\.sndcdn\.com/.+\.(?:jpg|png)',
'genres': ['Alternative'], 'genres': ['Alternative'],
'artists': ['The Royal Concept'], 'artists': ['The Royal Concept'],
'tags': [], 'tags': [],
},
}, },
}, {
# private link # private link
{ 'url': 'https://soundcloud.com/jaimemf/youtube-dl-test-video-a-y-baw/s-8Pjrp',
'url': 'https://soundcloud.com/jaimemf/youtube-dl-test-video-a-y-baw/s-8Pjrp', 'md5': 'aa0dd32bfea9b0c5ef4f02aacd080604',
'md5': 'aa0dd32bfea9b0c5ef4f02aacd080604', 'info_dict': {
'info_dict': { 'id': '123998367',
'id': '123998367', 'ext': 'mp3',
'ext': 'mp3', 'title': 'Youtube - Dl Test Video \'\' Ä↭',
'title': 'Youtube - Dl Test Video \'\' Ä↭', 'track': 'Youtube - Dl Test Video \'\' Ä↭',
'track': 'Youtube - Dl Test Video \'\' Ä↭', 'description': 'md5:610b729ee06ac4cedaa28607212948f3',
'description': 'test chars: "\'/\\ä↭', 'uploader': 'jaimeMF',
'uploader': 'jaimeMF', 'uploader_id': '69767071',
'uploader_id': '69767071', 'timestamp': 1386604920,
'timestamp': 1386604920, 'upload_date': '20131209',
'upload_date': '20131209', 'duration': 9.927,
'duration': 9.927, 'license': 'all-rights-reserved',
'license': 'all-rights-reserved', 'view_count': int,
'view_count': int, 'like_count': int,
'like_count': int, 'comment_count': int,
'comment_count': int, 'repost_count': int,
'repost_count': int, 'uploader_url': 'https://soundcloud.com/jaimemf',
'uploader_url': 'https://soundcloud.com/jaimemf', 'thumbnail': r're:https?://[ai]1\.sndcdn\.com/.+\.(?:jpg|png)',
'thumbnail': 'https://a1.sndcdn.com/images/default_avatar_large.png', 'genres': ['youtubedl'],
'genres': ['youtubedl'], 'tags': [],
'tags': [],
},
}, },
}, {
# private link (alt format) # private link (alt format)
{ 'url': 'https://api.soundcloud.com/tracks/123998367?secret_token=s-8Pjrp',
'url': 'https://api.soundcloud.com/tracks/123998367?secret_token=s-8Pjrp', 'md5': 'aa0dd32bfea9b0c5ef4f02aacd080604',
'md5': 'aa0dd32bfea9b0c5ef4f02aacd080604', 'info_dict': {
'info_dict': { 'id': '123998367',
'id': '123998367', 'ext': 'mp3',
'ext': 'mp3', 'title': 'Youtube - Dl Test Video \'\' Ä↭',
'title': 'Youtube - Dl Test Video \'\' Ä↭', 'track': 'Youtube - Dl Test Video \'\' Ä↭',
'track': 'Youtube - Dl Test Video \'\' Ä↭', 'description': 'md5:610b729ee06ac4cedaa28607212948f3',
'description': 'test chars: "\'/\\ä↭', 'uploader': 'jaimeMF',
'uploader': 'jaimeMF', 'uploader_id': '69767071',
'uploader_id': '69767071', 'timestamp': 1386604920,
'timestamp': 1386604920, 'upload_date': '20131209',
'upload_date': '20131209', 'duration': 9.927,
'duration': 9.927, 'license': 'all-rights-reserved',
'license': 'all-rights-reserved', 'view_count': int,
'view_count': int, 'like_count': int,
'like_count': int, 'comment_count': int,
'comment_count': int, 'repost_count': int,
'repost_count': int, 'uploader_url': 'https://soundcloud.com/jaimemf',
'uploader_url': 'https://soundcloud.com/jaimemf', 'thumbnail': r're:https?://[ai]1\.sndcdn\.com/.+\.(?:jpg|png)',
'thumbnail': 'https://a1.sndcdn.com/images/default_avatar_large.png', 'genres': ['youtubedl'],
'genres': ['youtubedl'], 'tags': [],
'tags': [],
},
}, },
}, {
# downloadable song # downloadable song
{ 'url': 'https://soundcloud.com/the80m/the-following',
'url': 'https://soundcloud.com/the80m/the-following', 'md5': 'ecb87d7705d5f53e6c02a63760573c75', # wav: '9ffcddb08c87d74fb5808a3c183a1d04'
'md5': 'ecb87d7705d5f53e6c02a63760573c75', # wav: '9ffcddb08c87d74fb5808a3c183a1d04' 'info_dict': {
'info_dict': { 'id': '343609555',
'id': '343609555', 'ext': 'opus', # wav original available with auth
'ext': 'opus', # wav original available with auth 'title': 'The Following',
'title': 'The Following', 'track': 'The Following',
'track': 'The Following', 'description': '',
'description': '', 'uploader': '80M',
'uploader': '80M', 'uploader_id': '312384765',
'uploader_id': '312384765', 'uploader_url': 'https://soundcloud.com/the80m',
'uploader_url': 'https://soundcloud.com/the80m', 'upload_date': '20170922',
'upload_date': '20170922', 'timestamp': 1506120436,
'timestamp': 1506120436, 'duration': 397.228,
'duration': 397.228, 'thumbnail': r're:https?://[ai]1\.sndcdn\.com/.+\.(?:jpg|png)',
'thumbnail': 'https://i1.sndcdn.com/artworks-000243916348-ktoo7d-original.jpg', 'license': 'all-rights-reserved',
'license': 'all-rights-reserved', 'like_count': int,
'like_count': int, 'comment_count': int,
'comment_count': int, 'repost_count': int,
'repost_count': int, 'view_count': int,
'view_count': int, 'genres': ['Dance & EDM'],
'genres': ['Dance & EDM'], 'artists': ['80M'],
'artists': ['80M'], 'tags': 'count:4',
'tags': ['80M', 'EDM', 'Dance', 'Music'],
},
'expected_warnings': ['Original download format is only available for registered users'],
}, },
'expected_warnings': ['Original download format is only available for registered users'],
}, {
# private link, downloadable format # private link, downloadable format
# tags with spaces (e.g. "Uplifting Trance", "Ori Uplift") # tags with spaces (e.g. "Uplifting Trance", "Ori Uplift")
{ 'url': 'https://soundcloud.com/oriuplift/uponly-238-no-talking-wav/s-AyZUd',
'url': 'https://soundcloud.com/oriuplift/uponly-238-no-talking-wav/s-AyZUd', 'md5': '2e1530d0e9986a833a67cb34fc90ece0', # wav: '64a60b16e617d41d0bef032b7f55441e'
'md5': '2e1530d0e9986a833a67cb34fc90ece0', # wav: '64a60b16e617d41d0bef032b7f55441e' 'info_dict': {
'info_dict': { 'id': '340344461',
'id': '340344461', 'ext': 'opus', # wav original available with auth
'ext': 'opus', # wav original available with auth 'title': 'Uplifting Only 238 [No Talking] (incl. Alex Feed Guestmix) (Aug 31, 2017) [wav]',
'title': 'Uplifting Only 238 [No Talking] (incl. Alex Feed Guestmix) (Aug 31, 2017) [wav]', 'track': 'Uplifting Only 238 [No Talking] (incl. Alex Feed Guestmix) (Aug 31, 2017) [wav]',
'track': 'Uplifting Only 238 [No Talking] (incl. Alex Feed Guestmix) (Aug 31, 2017) [wav]', 'description': 'md5:fa20ee0fca76a3d6df8c7e57f3715366',
'description': 'md5:fa20ee0fca76a3d6df8c7e57f3715366', 'uploader': 'Ori Uplift Music',
'uploader': 'Ori Uplift Music', 'uploader_id': '12563093',
'uploader_id': '12563093', 'timestamp': 1504206263,
'timestamp': 1504206263, 'upload_date': '20170831',
'upload_date': '20170831', 'duration': 7449.096,
'duration': 7449.096, 'license': 'all-rights-reserved',
'license': 'all-rights-reserved', 'view_count': int,
'view_count': int, 'like_count': int,
'like_count': int, 'comment_count': int,
'comment_count': int, 'repost_count': int,
'repost_count': int, 'thumbnail': r're:https?://[ai]1\.sndcdn\.com/.+\.(?:jpg|png)',
'thumbnail': 'https://i1.sndcdn.com/artworks-000240712245-kedn4p-original.jpg', 'uploader_url': 'https://soundcloud.com/oriuplift',
'uploader_url': 'https://soundcloud.com/oriuplift', 'genres': ['Trance'],
'genres': ['Trance'], 'artists': ['Ori Uplift'],
'artists': ['Ori Uplift'], 'tags': 'count:6',
'tags': ['Orchestral', 'Emotional', 'Uplifting Trance', 'Trance', 'Ori Uplift', 'UpOnly'],
},
'expected_warnings': ['Original download format is only available for registered users'],
}, },
'expected_warnings': ['Original download format is only available for registered users'],
}, {
# no album art, use avatar pic for thumbnail # no album art, use avatar pic for thumbnail
{ 'url': 'https://soundcloud.com/garyvee/sideways-prod-mad-real',
'url': 'https://soundcloud.com/garyvee/sideways-prod-mad-real', 'md5': '59c7872bc44e5d99b7211891664760c2',
'md5': '59c7872bc44e5d99b7211891664760c2', 'info_dict': {
'info_dict': { 'id': '309699954',
'id': '309699954', 'ext': 'mp3',
'ext': 'mp3', 'title': 'Sideways (Prod. Mad Real)',
'title': 'Sideways (Prod. Mad Real)', 'track': 'Sideways (Prod. Mad Real)',
'track': 'Sideways (Prod. Mad Real)', 'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
'description': 'md5:d41d8cd98f00b204e9800998ecf8427e', 'uploader': 'garyvee',
'uploader': 'garyvee', 'uploader_id': '2366352',
'uploader_id': '2366352', 'timestamp': 1488152409,
'timestamp': 1488152409, 'upload_date': '20170226',
'upload_date': '20170226', 'duration': 207.012,
'duration': 207.012, 'thumbnail': r're:https?://[ai]1\.sndcdn\.com/.+\.(?:jpg|png)',
'thumbnail': r're:https?://.*\.jpg', 'license': 'all-rights-reserved',
'license': 'all-rights-reserved', 'view_count': int,
'view_count': int, 'like_count': int,
'like_count': int, 'comment_count': int,
'comment_count': int, 'repost_count': int,
'repost_count': int, 'uploader_url': 'https://soundcloud.com/garyvee',
'uploader_url': 'https://soundcloud.com/garyvee', 'artists': ['MadReal'],
'artists': ['MadReal'], 'tags': [],
'tags': [],
},
'params': {
'skip_download': True,
},
}, },
{ 'params': {'skip_download': 'm3u8'},
'url': 'https://soundcloud.com/giovannisarani/mezzo-valzer', }, {
'md5': '8227c3473a4264df6b02ad7e5b7527ac', 'url': 'https://soundcloud.com/giovannisarani/mezzo-valzer',
'info_dict': { 'md5': '8227c3473a4264df6b02ad7e5b7527ac',
'id': '583011102', 'info_dict': {
'ext': 'opus', 'id': '583011102',
'title': 'Mezzo Valzer', 'ext': 'm4a',
'track': 'Mezzo Valzer', 'title': 'Mezzo Valzer',
'description': 'md5:f4d5f39d52e0ccc2b4f665326428901a', 'track': 'Mezzo Valzer',
'uploader': 'Giovanni Sarani', 'description': 'md5:f4d5f39d52e0ccc2b4f665326428901a',
'uploader_id': '3352531', 'uploader': 'Giovanni Sarani',
'timestamp': 1551394171, 'uploader_id': '3352531',
'upload_date': '20190228', 'timestamp': 1551394171,
'duration': 180.157, 'upload_date': '20190228',
'thumbnail': r're:https?://.*\.jpg', 'duration': 180.134,
'license': 'all-rights-reserved', 'thumbnail': r're:https?://[ai]1\.sndcdn\.com/.+\.(?:jpg|png)',
'view_count': int, 'license': 'all-rights-reserved',
'like_count': int, 'view_count': int,
'comment_count': int, 'like_count': int,
'repost_count': int, 'comment_count': int,
'genres': ['Piano'], 'repost_count': int,
'uploader_url': 'https://soundcloud.com/giovannisarani', 'genres': ['Piano'],
'tags': 'count:10', 'uploader_url': 'https://soundcloud.com/giovannisarani',
}, 'tags': 'count:10',
}, },
'params': {'skip_download': 'm3u8'},
}, {
# .png "original" artwork, 160kbps m4a HLS format # .png "original" artwork, 160kbps m4a HLS format
{ 'url': 'https://soundcloud.com/skorxh/audio-dealer',
'url': 'https://soundcloud.com/skorxh/audio-dealer', 'info_dict': {
'info_dict': { 'id': '2011421339',
'id': '2011421339', 'ext': 'm4a',
'ext': 'm4a', 'title': 'audio dealer',
'title': 'audio dealer', 'description': '',
'description': '', 'uploader': '$KORCH',
'uploader': '$KORCH', 'uploader_id': '150292288',
'uploader_id': '150292288', 'uploader_url': 'https://soundcloud.com/skorxh',
'uploader_url': 'https://soundcloud.com/skorxh', 'comment_count': int,
'comment_count': int, 'view_count': int,
'view_count': int, 'like_count': int,
'like_count': int, 'repost_count': int,
'repost_count': int, 'duration': 213.469,
'duration': 213.469, 'tags': [],
'tags': [], 'artists': ['$KORXH'],
'artists': ['$KORXH'], 'track': 'audio dealer',
'track': 'audio dealer', 'timestamp': 1737143201,
'timestamp': 1737143201, 'upload_date': '20250117',
'upload_date': '20250117', 'license': 'all-rights-reserved',
'license': 'all-rights-reserved', 'thumbnail': r're:https?://[ai]1\.sndcdn\.com/.+\.(?:jpg|png)',
'thumbnail': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-original.png', 'thumbnails': [
'thumbnails': [ {'id': 'mini', 'url': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-mini.jpg'},
{'id': 'mini', 'url': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-mini.jpg'}, {'id': 'tiny', 'url': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-tiny.jpg'},
{'id': 'tiny', 'url': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-tiny.jpg'}, {'id': 'small', 'url': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-small.jpg'},
{'id': 'small', 'url': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-small.jpg'}, {'id': 'badge', 'url': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-badge.jpg'},
{'id': 'badge', 'url': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-badge.jpg'}, {'id': 't67x67', 'url': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-t67x67.jpg'},
{'id': 't67x67', 'url': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-t67x67.jpg'}, {'id': 'large', 'url': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-large.jpg'},
{'id': 'large', 'url': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-large.jpg'}, {'id': 't300x300', 'url': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-t300x300.jpg'},
{'id': 't300x300', 'url': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-t300x300.jpg'}, {'id': 'crop', 'url': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-crop.jpg'},
{'id': 'crop', 'url': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-crop.jpg'}, {'id': 't500x500', 'url': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-t500x500.jpg'},
{'id': 't500x500', 'url': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-t500x500.jpg'}, {'id': 'original', 'url': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-original.png'},
{'id': 'original', 'url': 'https://i1.sndcdn.com/artworks-a1wKGMYNreDLTMrT-fGjRiw-original.png'}, ],
],
},
'params': {'skip_download': 'm3u8', 'format': 'hls_aac_160k'},
},
{
# AAC HQ format available (account with active subscription needed)
'url': 'https://soundcloud.com/wandw/the-chainsmokers-ft-daya-dont-let-me-down-ww-remix-1',
'only_matching': True,
},
{
# Go+ (account with active subscription needed)
'url': 'https://soundcloud.com/taylorswiftofficial/look-what-you-made-me-do',
'only_matching': True,
}, },
] 'params': {'skip_download': 'm3u8', 'format': 'hls_aac_160k'},
}, {
# AAC HQ format available (account with active subscription needed)
'url': 'https://soundcloud.com/wandw/the-chainsmokers-ft-daya-dont-let-me-down-ww-remix-1',
'only_matching': True,
}, {
# Go+ (account with active subscription needed)
'url': 'https://soundcloud.com/taylorswiftofficial/look-what-you-made-me-do',
'only_matching': True,
}]
def _real_extract(self, url): def _real_extract(self, url):
mobj = self._match_valid_url(url) mobj = self._match_valid_url(url)
@ -907,7 +930,7 @@ class SoundcloudUserIE(SoundcloudPagedPlaylistBaseIE):
'id': '7098329', 'id': '7098329',
'title': 'Grynpyret (Spotlight)', 'title': 'Grynpyret (Spotlight)',
}, },
'playlist_mincount': 1, 'playlist_mincount': 0,
}, { }, {
'url': 'https://soundcloud.com/one-thousand-and-one/comments', 'url': 'https://soundcloud.com/one-thousand-and-one/comments',
'info_dict': { 'info_dict': {
@ -998,7 +1021,7 @@ class SoundcloudRelatedIE(SoundcloudPagedPlaylistBaseIE):
'id': '1084577272', 'id': '1084577272',
'title': 'Sexapil - Pingers 5 (Recommended)', 'title': 'Sexapil - Pingers 5 (Recommended)',
}, },
'playlist_mincount': 50, 'playlist_mincount': 49,
}, { }, {
'url': 'https://soundcloud.com/wajang/sexapil-pingers-5/albums', 'url': 'https://soundcloud.com/wajang/sexapil-pingers-5/albums',
'info_dict': { 'info_dict': {
@ -1045,7 +1068,7 @@ class SoundcloudPlaylistIE(SoundcloudPlaylistBaseIE):
'info_dict': { 'info_dict': {
'id': '4110309', 'id': '4110309',
'title': 'TILT Brass - Bowery Poetry Club, August \'03 [Non-Site SCR 02]', 'title': 'TILT Brass - Bowery Poetry Club, August \'03 [Non-Site SCR 02]',
'description': 're:.*?TILT Brass - Bowery Poetry Club', 'description': 'md5:e4373f7177fe3db292a8552b4ec41bc6',
'uploader': 'Non-Site Records', 'uploader': 'Non-Site Records',
'uploader_id': '33660914', 'uploader_id': '33660914',
'album_artists': ['Non-Site Records'], 'album_artists': ['Non-Site Records'],

12
yt_dlp/extractor/sportbox.py
Unescape Escape View File

@ -8,6 +8,7 @@ from ..utils import (
class SportBoxIE(InfoExtractor): class SportBoxIE(InfoExtractor):
_WORKING = False
_VALID_URL = r'https?://(?:news\.sportbox|matchtv)\.ru/vdl/player(?:/[^/]+/|\?.*?\bn?id=)(?P<id>\d+)' _VALID_URL = r'https?://(?:news\.sportbox|matchtv)\.ru/vdl/player(?:/[^/]+/|\?.*?\bn?id=)(?P<id>\d+)'
_EMBED_REGEX = [r'<iframe[^>]+src="(?P<url>https?://(?:news\.sportbox|matchtv)\.ru/vdl/player[^"]+)"'] _EMBED_REGEX = [r'<iframe[^>]+src="(?P<url>https?://(?:news\.sportbox|matchtv)\.ru/vdl/player[^"]+)"']
_TESTS = [{ _TESTS = [{
@ -17,7 +18,7 @@ class SportBoxIE(InfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'title': 'В Новороссийске прошел детский турнир «Поле славы боевой»', 'title': 'В Новороссийске прошел детский турнир «Поле славы боевой»',
'description': 'В Новороссийске прошел детский турнир «Поле славы боевой»', 'description': 'В Новороссийске прошел детский турнир «Поле славы боевой»',
'thumbnail': r're:^https?://.*\.jpg$', 'thumbnail': r're:https?://.+\.jpg',
'duration': 292, 'duration': 292,
'view_count': int, 'view_count': int,
'timestamp': 1426237001, 'timestamp': 1426237001,
@ -40,6 +41,15 @@ class SportBoxIE(InfoExtractor):
'url': 'https://matchtv.ru/vdl/player/media/109158', 'url': 'https://matchtv.ru/vdl/player/media/109158',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
'url': 'http://www.vestifinance.ru/articles/25753',
'info_dict': {
'id': '25753',
'title': 'Прямые трансляции с Форума-выставки "Госзаказ-2013"',
},
'playlist_count': 3,
'skip': 'Invalid URL',
}]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)

12
yt_dlp/extractor/springboardplatform.py
Unescape Escape View File

@ -30,11 +30,12 @@ class SpringboardPlatformIE(InfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'title': 'Redman "BUD like YOU" "Usher Good Kisser" REMIX', 'title': 'Redman "BUD like YOU" "Usher Good Kisser" REMIX',
'description': 'Redman "BUD like YOU" "Usher Good Kisser" REMIX', 'description': 'Redman "BUD like YOU" "Usher Good Kisser" REMIX',
'thumbnail': r're:^https?://.*\.jpg$', 'thumbnail': r're:https?://.+\.jpg',
'timestamp': 1409132328, 'timestamp': 1409132328,
'upload_date': '20140827', 'upload_date': '20140827',
'duration': 193, 'duration': 193,
}, },
'skip': 'Invalid URL',
}, { }, {
'url': 'http://cms.springboardplatform.com/embed_iframe/159/video/981017/rab007/rapbasement.com/1/1', 'url': 'http://cms.springboardplatform.com/embed_iframe/159/video/981017/rab007/rapbasement.com/1/1',
'only_matching': True, 'only_matching': True,
@ -45,6 +46,15 @@ class SpringboardPlatformIE(InfoExtractor):
'url': 'http://cms.springboardplatform.com/xml_feeds_advanced/index/159/rss3/981017/0/0/1/', 'url': 'http://cms.springboardplatform.com/xml_feeds_advanced/index/159/rss3/981017/0/0/1/',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
'url': 'https://www.kidzworld.com/article/30935-trolls-the-beat-goes-on-interview-skylar-astin-and-amanda-leighton',
'info_dict': {
'id': '1731611',
'ext': 'mp4',
'title': 'Official Trailer | TROLLS: THE BEAT GOES ON!',
},
'skip': 'Invalid URL',
}]
def _real_extract(self, url): def _real_extract(self, url):
mobj = self._match_valid_url(url) mobj = self._match_valid_url(url)

23
yt_dlp/extractor/substack.py
Unescape Escape View File

@ -63,6 +63,29 @@ class SubstackIE(InfoExtractor):
'uploader_id': '61579', 'uploader_id': '61579',
}, },
}] }]
_WEBPAGE_TESTS = [{
'url': 'https://www.mollymovieclub.com/p/interstellar',
'info_dict': {
'id': '53602801',
'ext': 'mpga',
'title': 'Interstellar',
'description': 'Listen now | Episode One',
'thumbnail': r're:https?://.+\.jpeg',
'uploader': 'Molly Movie Club',
'uploader_id': '839621',
},
}, {
'url': 'https://www.blockedandreported.org/p/episode-117-lets-talk-about-depp',
'info_dict': {
'id': '57962052',
'ext': 'mpga',
'title': 'md5:855b2756f0ee10f6723fa00b16266f8d',
'description': 'The takes the takes the takes',
'thumbnail': r're:https?://.+\.jpeg',
'uploader': 'Blocked and Reported',
'uploader_id': '500230',
},
}]
@classmethod @classmethod
def _extract_embed_urls(cls, url, webpage): def _extract_embed_urls(cls, url, webpage):

32
yt_dlp/extractor/ted.py
Unescape Escape View File

@ -32,12 +32,12 @@ class TedTalkIE(TedBaseIE):
'title': 'How to break down barriers and not accept limits', 'title': 'How to break down barriers and not accept limits',
'description': 'md5:000707cece219d1e165b11550d612331', 'description': 'md5:000707cece219d1e165b11550d612331',
'view_count': int, 'view_count': int,
'tags': ['personal growth', 'equality', 'activism', 'motivation', 'social change', 'sports'], 'tags': 'count:6',
'uploader': 'Candace Parker', 'uploader': 'Candace Parker',
'duration': 676.0, 'duration': 679,
'thumbnail': r're:https?://pi\.tedcdn\.com/.+\.jpg',
'upload_date': '20220114', 'upload_date': '20220114',
'release_date': '20211201', 'release_date': '20211201',
'thumbnail': r're:http.*\.jpg',
}, },
}] }]
@ -162,7 +162,7 @@ class TedSeriesIE(TedBaseIE):
'id': '8_2', 'id': '8_2',
'title': 'The Way We Work Season 2', 'title': 'The Way We Work Season 2',
'series': 'The Way We Work', 'series': 'The Way We Work',
'description': 'md5:59469256e533e1a48c4aa926a382234c', 'description': 'md5:36678fe045f6ad7f39da80ea9370cbcd',
'season_number': 2, 'season_number': 2,
}, },
'playlist_mincount': 8, 'playlist_mincount': 8,
@ -213,7 +213,6 @@ class TedPlaylistIE(TedBaseIE):
class TedEmbedIE(InfoExtractor): class TedEmbedIE(InfoExtractor):
_VALID_URL = r'https?://embed(?:-ssl)?\.ted\.com/' _VALID_URL = r'https?://embed(?:-ssl)?\.ted\.com/'
_EMBED_REGEX = [rf'<iframe[^>]+?src=(["\'])(?P<url>{_VALID_URL}.+?)\1'] _EMBED_REGEX = [rf'<iframe[^>]+?src=(["\'])(?P<url>{_VALID_URL}.+?)\1']
_TESTS = [{ _TESTS = [{
'url': 'https://embed.ted.com/talks/janet_stovall_how_to_get_serious_about_diversity_and_inclusion_in_the_workplace', 'url': 'https://embed.ted.com/talks/janet_stovall_how_to_get_serious_about_diversity_and_inclusion_in_the_workplace',
'info_dict': { 'info_dict': {
@ -222,14 +221,31 @@ class TedEmbedIE(InfoExtractor):
'title': 'How to get serious about diversity and inclusion in the workplace', 'title': 'How to get serious about diversity and inclusion in the workplace',
'description': 'md5:0978aafe396e05341f8ecc795d22189d', 'description': 'md5:0978aafe396e05341f8ecc795d22189d',
'view_count': int, 'view_count': int,
'tags': list,
'uploader': 'Janet Stovall', 'uploader': 'Janet Stovall',
'duration': 664.0, 'duration': 654,
'tags': 'count:10',
'thumbnail': r're:https?://pi\.tedcdn\.com/.+\.jpg',
'upload_date': '20180822', 'upload_date': '20180822',
'release_date': '20180719', 'release_date': '20180719',
'thumbnail': r're:http.*\.jpg',
}, },
}] }]
_WEBPAGE_TESTS = [{
'url': 'https://ideas.ted.com/6-ways-to-give-that-arent-about-money/',
'info_dict': {
'id': '123235',
'ext': 'mp4',
'title': 'It\'s time for infectious generosity. Here\'s how',
'description': 'md5:0f972eb2b53ad7d1385fb65f519657b4',
'duration': 1172,
'release_date': '20231128',
'tags': 'count:9',
'thumbnail': r're:https?://pi\.tedcdn\.com/.+\.jpg',
'upload_date': '20240109',
'uploader': 'Chris Anderson',
'view_count': int,
},
'params': {'skip_download': 'm3u8'},
}]
def _real_extract(self, url): def _real_extract(self, url):
return self.url_result(re.sub(r'://embed(-ssl)?', '://www', url), TedTalkIE.ie_key()) return self.url_result(re.sub(r'://embed(-ssl)?', '://www', url), TedTalkIE.ie_key())

40
yt_dlp/extractor/tunein.py
Unescape Escape View File

@ -51,9 +51,9 @@ class TuneInStationIE(TuneInBaseIE):
'url': 'https://tunein.com/radio/Jazz24-885-s34682/', 'url': 'https://tunein.com/radio/Jazz24-885-s34682/',
'info_dict': { 'info_dict': {
'id': 's34682', 'id': 's34682',
'title': 're:^Jazz24', 'title': str,
'description': 'md5:d6d0b89063fd68d529fa7058ee98619b', 'description': 'md5:d6d0b89063fd68d529fa7058ee98619b',
'thumbnail': 're:^https?://[^?&]+/s34682', 'thumbnail': r're:https?://cdn-profiles\.tunein\.com/.+',
'location': 'Seattle-Tacoma, US', 'location': 'Seattle-Tacoma, US',
'ext': 'mp3', 'ext': 'mp3',
'live_status': 'is_live', 'live_status': 'is_live',
@ -68,17 +68,26 @@ class TuneInStationIE(TuneInBaseIE):
'url': 'https://tunein.com/radio/BBC-Radio-1-988-s24939/', 'url': 'https://tunein.com/radio/BBC-Radio-1-988-s24939/',
'info_dict': { 'info_dict': {
'id': 's24939', 'id': 's24939',
'title': 're:^BBC Radio 1', 'title': str,
'description': 'md5:f3f75f7423398d87119043c26e7bfb84', 'description': 'md5:ee2c56794844610d045f8caf5ff34d0c',
'thumbnail': 're:^https?://[^?&]+/s24939', 'thumbnail': r're:https?://cdn-profiles\.tunein\.com/.+',
'location': 'London, UK', 'location': 'London, UK',
'ext': 'mp3', 'ext': 'm4a',
'live_status': 'is_live', 'live_status': 'is_live',
}, },
'params': { 'params': {
'skip_download': True, 'skip_download': True,
}, },
}] }]
_WEBPAGE_TESTS = [{
'url': 'https://www.martiniinthemorning.com/',
'info_dict': {
'id': 's55412',
'ext': 'mp3',
'title': 'TuneInStation video #s55412',
},
'expected_warnings': ['unable to extract hydration', 'Extractor failed to obtain "title"'],
}]
def _real_extract(self, url): def _real_extract(self, url):
station_id = self._match_id(url) station_id = self._match_id(url)
@ -123,9 +132,9 @@ class TuneInPodcastIE(TuneInBaseIE):
'info_dict': { 'info_dict': {
'id': 'p14', 'id': 'p14',
'title': 'BBC News', 'title': 'BBC News',
'description': 'md5:1218e575eeaff75f48ed978261fa2068', 'description': 'md5:30b9622bcc4bd101d4acd6f38f284aed',
}, },
'playlist_mincount': 200, 'playlist_mincount': 36,
}] }]
_PAGE_SIZE = 30 _PAGE_SIZE = 30
@ -164,9 +173,9 @@ class TuneInPodcastEpisodeIE(TuneInBaseIE):
'url': 'https://tunein.com/podcasts/Technology-Podcasts/Artificial-Intelligence-p1153019/?topicId=236404354', 'url': 'https://tunein.com/podcasts/Technology-Podcasts/Artificial-Intelligence-p1153019/?topicId=236404354',
'info_dict': { 'info_dict': {
'id': 't236404354', 'id': 't236404354',
'title': '#351 \u2013 MrBeast: Future of YouTube, Twitter, TikTok, and Instagram', 'title': '#351 – MrBeast: Future of YouTube, Twitter, TikTok, and Instagram',
'description': 'md5:e1734db6f525e472c0c290d124a2ad77', 'description': 'md5:2784533b98f8ac45c0820b1e4a8d8bb2',
'thumbnail': 're:^https?://[^?&]+/p1153019', 'thumbnail': r're:https?://cdn-profiles\.tunein\.com/.+',
'timestamp': 1673458571, 'timestamp': 1673458571,
'upload_date': '20230111', 'upload_date': '20230111',
'series_id': 'p1153019', 'series_id': 'p1153019',
@ -198,18 +207,19 @@ class TuneInPodcastEpisodeIE(TuneInBaseIE):
class TuneInShortenerIE(InfoExtractor): class TuneInShortenerIE(InfoExtractor):
_WORKING = False
IE_NAME = 'tunein:shortener' IE_NAME = 'tunein:shortener'
IE_DESC = False # Do not list IE_DESC = False # Do not list
_VALID_URL = r'https?://tun\.in/(?P<id>[A-Za-z0-9]+)' _VALID_URL = r'https?://tun\.in/(?P<id>[A-Za-z0-9]+)'
_TEST = { _TESTS = [{
# test redirection # test redirection
'url': 'http://tun.in/ser7s', 'url': 'http://tun.in/ser7s',
'info_dict': { 'info_dict': {
'id': 's34682', 'id': 's34682',
'title': 're:^Jazz24', 'title': str,
'description': 'md5:d6d0b89063fd68d529fa7058ee98619b', 'description': 'md5:d6d0b89063fd68d529fa7058ee98619b',
'thumbnail': 're:^https?://[^?&]+/s34682', 'thumbnail': r're:https?://cdn-profiles\.tunein\.com/.+',
'location': 'Seattle-Tacoma, US', 'location': 'Seattle-Tacoma, US',
'ext': 'mp3', 'ext': 'mp3',
'live_status': 'is_live', 'live_status': 'is_live',
@ -217,7 +227,7 @@ class TuneInShortenerIE(InfoExtractor):
'params': { 'params': {
'skip_download': True, # live stream 'skip_download': True, # live stream
}, },
} }]
def _real_extract(self, url): def _real_extract(self, url):
redirect_id = self._match_id(url) redirect_id = self._match_id(url)

22
yt_dlp/extractor/tvc.py
Unescape Escape View File

@ -8,17 +8,29 @@ from ..utils import (
class TVCIE(InfoExtractor): class TVCIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?tvc\.ru/video/iframe/id/(?P<id>\d+)' _VALID_URL = r'https?://(?:www\.)?tvc\.ru/video/iframe/id/(?P<id>\d+)'
_EMBED_REGEX = [r'<iframe[^>]+?src=(["\'])(?P<url>(?:http:)?//(?:www\.)?tvc\.ru/video/iframe/id/[^"]+)\1'] _EMBED_REGEX = [r'<iframe[^>]+?src=(["\'])(?P<url>(?:http:)?//(?:www\.)?tvc\.ru/video/iframe/id/[^"]+)\1']
_TEST = { _TESTS = [{
'url': 'http://www.tvc.ru/video/iframe/id/74622/isPlay/false/id_stat/channel/?acc_video_id=/channel/brand/id/17/show/episodes/episode_id/39702', 'url': 'http://www.tvc.ru/video/iframe/id/74622/isPlay/false/id_stat/channel/?acc_video_id=/channel/brand/id/17/show/episodes/episode_id/39702',
'md5': 'bbc5ff531d1e90e856f60fc4b3afd708', 'md5': 'aa6fb3cf384e18a0ad3b30ee2898beba',
'info_dict': { 'info_dict': {
'id': '74622', 'id': '74622',
'ext': 'mp4', 'ext': 'mp4',
'title': 'События. "События". Эфир от 22.05.2015 14:30', 'title': 'TVC video #74622',
'thumbnail': r're:^https?://.*\.jpg$',
'duration': 1122, 'duration': 1122,
'thumbnail': r're:https?://cdn\.tvc\.ru/pictures/.+\.jpg',
},
}]
_WEBPAGE_TESTS = [{
# FIXME: Embed detection
'url': 'https://krizis-centr.ru/informatsiya/smi-o-tsentre/liniya-zashchity-bitye-zhjony-tv-tsentr',
'md5': '43b8eee579a5cd2b85c9ed5b73d1c671',
'info_dict': {
'id': '123378',
'ext': 'mp4',
'title': 'TVC video #123378',
'duration': 1526,
'thumbnail': r're:https?://cdn\.tvc\.ru/pictures/.+\.jpg',
}, },
} }]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)

20
yt_dlp/extractor/tvopengr.py
Unescape Escape View File

@ -22,12 +22,12 @@ class TVOpenGrWatchIE(TVOpenGrBaseIE):
_TESTS = [{ _TESTS = [{
'url': 'https://www.ethnos.gr/watch/101009/nikoskaprabelosdenexoymekanenanasthenhsemethmethmetallaxhomikron', 'url': 'https://www.ethnos.gr/watch/101009/nikoskaprabelosdenexoymekanenanasthenhsemethmethmetallaxhomikron',
'md5': '8728570e3a72e0f8d9475ba94859fdc1',
'info_dict': { 'info_dict': {
'id': '101009', 'id': '101009',
'title': 'md5:51f68773dcb6c70498cd326f45fefdf0', 'title': 'md5:51f68773dcb6c70498cd326f45fefdf0',
'display_id': 'nikoskaprabelosdenexoymekanenanasthenhsemethmethmetallaxhomikron', 'display_id': 'nikoskaprabelosdenexoymekanenanasthenhsemethmethmetallaxhomikron',
'description': 'md5:78fff49f18fb3effe41b070e5c7685d6', 'description': 'md5:78fff49f18fb3effe41b070e5c7685d6',
'duration': 246.0,
'thumbnail': 'https://opentv-static.siliconweb.com/imgHandler/1920/d573ba71-ec5f-43c6-b4cb-d181f327d3a8.jpg', 'thumbnail': 'https://opentv-static.siliconweb.com/imgHandler/1920/d573ba71-ec5f-43c6-b4cb-d181f327d3a8.jpg',
'ext': 'mp4', 'ext': 'mp4',
'upload_date': '20220109', 'upload_date': '20220109',
@ -35,12 +35,12 @@ class TVOpenGrWatchIE(TVOpenGrBaseIE):
}, },
}, { }, {
'url': 'https://www.tvopen.gr/watch/100979/se28099agapaomenalla7cepeisodio267cmhthrargiapashskakias', 'url': 'https://www.tvopen.gr/watch/100979/se28099agapaomenalla7cepeisodio267cmhthrargiapashskakias',
'md5': '38f98a1be0c577db4ea2d1b1c0770c48',
'info_dict': { 'info_dict': {
'id': '100979', 'id': '100979',
'title': 'md5:e021f3001e16088ee40fa79b20df305b', 'title': 'md5:e021f3001e16088ee40fa79b20df305b',
'display_id': 'se28099agapaomenalla7cepeisodio267cmhthrargiapashskakias', 'display_id': 'se28099agapaomenalla7cepeisodio267cmhthrargiapashskakias',
'description': 'md5:ba17db53954134eb8d625d199e2919fb', 'description': 'md5:ba17db53954134eb8d625d199e2919fb',
'duration': 2420.0,
'thumbnail': 'https://opentv-static.siliconweb.com/imgHandler/1920/9bb71cf1-21da-43a9-9d65-367950fde4e3.jpg', 'thumbnail': 'https://opentv-static.siliconweb.com/imgHandler/1920/9bb71cf1-21da-43a9-9d65-367950fde4e3.jpg',
'ext': 'mp4', 'ext': 'mp4',
'upload_date': '20220108', 'upload_date': '20220108',
@ -98,18 +98,32 @@ class TVOpenGrEmbedIE(TVOpenGrBaseIE):
_TESTS = [{ _TESTS = [{
'url': 'https://cdn.ethnos.gr/embed/100963', 'url': 'https://cdn.ethnos.gr/embed/100963',
'md5': '2da147881f45571d81662d94d086628b',
'info_dict': { 'info_dict': {
'id': '100963', 'id': '100963',
'display_id': 'koronoiosapotoysdieythyntestonsxoleionselftestgiaosoysdenbrhkan', 'display_id': 'koronoiosapotoysdieythyntestonsxoleionselftestgiaosoysdenbrhkan',
'title': 'md5:2c71876fadf0cda6043da0da5fca2936', 'title': 'md5:2c71876fadf0cda6043da0da5fca2936',
'description': 'md5:17482b4432e5ed30eccd93b05d6ea509', 'description': 'md5:17482b4432e5ed30eccd93b05d6ea509',
'duration': 118.0,
'thumbnail': 'https://opentv-static.siliconweb.com/imgHandler/1920/5804e07f-799a-4247-a696-33842c94ca37.jpg', 'thumbnail': 'https://opentv-static.siliconweb.com/imgHandler/1920/5804e07f-799a-4247-a696-33842c94ca37.jpg',
'ext': 'mp4', 'ext': 'mp4',
'upload_date': '20220108', 'upload_date': '20220108',
'timestamp': 1641600000, 'timestamp': 1641600000,
}, },
}] }]
_WEBPAGE_TESTS = [{
'url': 'https://www.ethnos.gr/World/article/190604/hparosiaxekinoynoisynomiliessthgeneyhmethskiatoypolemoypanoapothnoykrania',
'info_dict': {
'id': '101119',
'ext': 'mp4',
'title': 'Οι καρποί των διαπραγματεύσεων ΗΠΑ-Ρωσίας | Ώρα Ελλάδος 7:00 > Ρεπορτάζ',
'description': 'Ξεκινούν οι διαπραγματεύσεις ανάμεσα σε Ηνωμένες Πολιτείες και Ρωσία για την Ουκρανία.',
'display_id': 'oikarpoitondiapragmateyseonhparosias',
'duration': 421.0,
'thumbnail': r're:https?://opentv-static\.siliconweb\.com/imgHandler/.+\.jpg',
'timestamp': 1641772800,
'upload_date': '20220110',
},
}]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)

15
yt_dlp/extractor/twentymin.py
Unescape Escape View File

@ -6,6 +6,7 @@ from ..utils import (
class TwentyMinutenIE(InfoExtractor): class TwentyMinutenIE(InfoExtractor):
_WORKING = False
IE_NAME = '20min' IE_NAME = '20min'
_VALID_URL = r'''(?x) _VALID_URL = r'''(?x)
https?:// https?://
@ -24,7 +25,7 @@ class TwentyMinutenIE(InfoExtractor):
'id': '469148', 'id': '469148',
'ext': 'mp4', 'ext': 'mp4',
'title': '85 000 Franken für 15 perfekte Minuten', 'title': '85 000 Franken für 15 perfekte Minuten',
'thumbnail': r're:https?://.*\.jpg$', 'thumbnail': r're:https?://.+\.jpg',
}, },
}, { }, {
'url': 'http://www.20min.ch/videoplayer/videoplayer.html?params=client@twentyDE|videoId@523629', 'url': 'http://www.20min.ch/videoplayer/videoplayer.html?params=client@twentyDE|videoId@523629',
@ -33,7 +34,7 @@ class TwentyMinutenIE(InfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'title': 'So kommen Sie bei Eis und Schnee sicher an', 'title': 'So kommen Sie bei Eis und Schnee sicher an',
'description': 'md5:117c212f64b25e3d95747e5276863f7d', 'description': 'md5:117c212f64b25e3d95747e5276863f7d',
'thumbnail': r're:https?://.*\.jpg$', 'thumbnail': r're:https?://.+\.jpg',
}, },
'params': { 'params': {
'skip_download': True, 'skip_download': True,
@ -42,6 +43,16 @@ class TwentyMinutenIE(InfoExtractor):
'url': 'http://www.20min.ch/videotv/?cid=44&vid=468738', 'url': 'http://www.20min.ch/videotv/?cid=44&vid=468738',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
# FIXME: Update _VALID_URL
'url': 'https://www.20min.ch/story/so-kommen-sie-bei-eis-und-schnee-sicher-an-557858045456',
'info_dict': {
'id': '523629',
'ext': 'mp4',
'title': 'So kommen Sie bei Eis und Schnee sicher an',
'description': 'md5:117c212f64b25e3d95747e5276863f7d',
},
}]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)

18
yt_dlp/extractor/udn.py
Unescape Escape View File

@ -22,11 +22,7 @@ class UDNEmbedIE(InfoExtractor):
'title': '生物老師男變女 全校挺"做自己"', 'title': '生物老師男變女 全校挺"做自己"',
'thumbnail': r're:^https?://.*\.jpg$', 'thumbnail': r're:^https?://.*\.jpg$',
}, },
'params': { 'skip': 'Invalid URL',
# m3u8 download
'skip_download': True,
},
'expected_warnings': ['Failed to parse JSON Expecting value'],
}, { }, {
'url': 'https://video.udn.com/embed/news/300040', 'url': 'https://video.udn.com/embed/news/300040',
'only_matching': True, 'only_matching': True,
@ -35,6 +31,18 @@ class UDNEmbedIE(InfoExtractor):
'url': 'https://video.udn.com/play/news/303776', 'url': 'https://video.udn.com/play/news/303776',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
# FIXME: Update _VALID_URL
'url': 'https://video.udn.com/news/1308561',
'info_dict': {
'id': '1308561',
'ext': 'mp4',
'title': '影／丹娜絲颱風暴風圈擴大 上午8:30發布海警',
'thumbnail': r're:https?://cdn\.udn\.com/img/.+\.jpg',
},
'expected_warnings': ['Failed to parse JSON'],
'params': {'skip_download': 'm3u8'},
}]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)

19
yt_dlp/extractor/vbox7.py
Unescape Escape View File

@ -26,7 +26,7 @@ class Vbox7IE(InfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'title': 'Борисов: Притеснен съм за бъдещето на България', 'title': 'Борисов: Притеснен съм за бъдещето на България',
'description': 'По думите му е опасно страната ни да бъде обявена за "сигурна"', 'description': 'По думите му е опасно страната ни да бъде обявена за "сигурна"',
'thumbnail': r're:^https?://.*\.jpg$', 'thumbnail': r're:https?://.+\.jpg',
'timestamp': 1470982814, 'timestamp': 1470982814,
'upload_date': '20160812', 'upload_date': '20160812',
'uploader': 'zdraveibulgaria', 'uploader': 'zdraveibulgaria',
@ -48,7 +48,7 @@ class Vbox7IE(InfoExtractor):
'upload_date': '20130207', 'upload_date': '20130207',
'duration': 83, 'duration': 83,
}, },
'expected_warnings': ['Failed to download m3u8 information'], 'skip': 'Invalid URL',
}, { }, {
'url': 'http://vbox7.com/emb/external.php?vid=a240d20f9c&autoplay=1', 'url': 'http://vbox7.com/emb/external.php?vid=a240d20f9c&autoplay=1',
'only_matching': True, 'only_matching': True,
@ -56,6 +56,21 @@ class Vbox7IE(InfoExtractor):
'url': 'http://i49.vbox7.com/player/ext.swf?vid=0946fff23c&autoplay=1', 'url': 'http://i49.vbox7.com/player/ext.swf?vid=0946fff23c&autoplay=1',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
'url': 'https://nova.bg/news/view/2016/08/16/156543/%D0%BD%D0%B0-%D0%BA%D0%BE%D1%81%D1%8A%D0%BC-%D0%BE%D1%82-%D0%B2%D0%B7%D1%80%D0%B8%D0%B2-%D0%BE%D1%82%D1%86%D0%B5%D0%BF%D0%B8%D1%85%D0%B0-%D1%86%D1%8F%D0%BB-%D0%BA%D0%B2%D0%B0%D1%80%D1%82%D0%B0%D0%BB-%D0%B7%D0%B0%D1%80%D0%B0%D0%B4%D0%B8-%D0%B8%D0%B7%D1%82%D0%B8%D1%87%D0%B0%D0%BD%D0%B5-%D0%BD%D0%B0-%D0%B3%D0%B0%D0%B7-%D0%B2-%D0%BF%D0%BB%D0%BE%D0%B2%D0%B4%D0%B8%D0%B2/',
'info_dict': {
'id': '5a4d12166d',
'ext': 'mp4',
'title': 'НА КОСЪМ ОТ ВЗРИВ: Отцепиха цял квартал заради изтичане на газ в Пловдив',
'description': 'Инцидентът е станал на бензиностанция',
'duration': 200,
'thumbnail': r're:https?://.+\.jpg',
'timestamp': 1471353501,
'upload_date': '20160816',
'uploader': 'novinitenanova',
'view_count': int,
},
}]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)

14
yt_dlp/extractor/viddler.py
Unescape Escape View File

@ -6,6 +6,7 @@ from ..utils import (
class ViddlerIE(InfoExtractor): class ViddlerIE(InfoExtractor):
_WORKING = False
_VALID_URL = r'https?://(?:www\.)?viddler\.com/(?:v|embed|player)/(?P<id>[a-z0-9]+)(?:.+?\bsecret=(\d+))?' _VALID_URL = r'https?://(?:www\.)?viddler\.com/(?:v|embed|player)/(?P<id>[a-z0-9]+)(?:.+?\bsecret=(\d+))?'
_EMBED_REGEX = [r'<(?:iframe[^>]+?src|param[^>]+?value)=(["\'])(?P<url>(?:https?:)?//(?:www\.)?viddler\.com/(?:embed|player)/.+?)\1'] _EMBED_REGEX = [r'<(?:iframe[^>]+?src|param[^>]+?value)=(["\'])(?P<url>(?:https?:)?//(?:www\.)?viddler\.com/(?:embed|player)/.+?)\1']
@ -21,11 +22,12 @@ class ViddlerIE(InfoExtractor):
'timestamp': 1335371429, 'timestamp': 1335371429,
'upload_date': '20120425', 'upload_date': '20120425',
'duration': 100.89, 'duration': 100.89,
'thumbnail': r're:^https?://.*\.jpg$', 'thumbnail': r're:https?://.+\.jpg',
'view_count': int, 'view_count': int,
'comment_count': int, 'comment_count': int,
'categories': ['video content', 'high quality video', 'video made easy', 'how to produce video with limited resources', 'viddler'], 'categories': ['video content', 'high quality video', 'video made easy', 'how to produce video with limited resources', 'viddler'],
}, },
'skip': 'Invalid URL',
}, { }, {
'url': 'http://www.viddler.com/v/4d03aad9/', 'url': 'http://www.viddler.com/v/4d03aad9/',
'md5': 'f12c5a7fa839c47a79363bfdf69404fb', 'md5': 'f12c5a7fa839c47a79363bfdf69404fb',
@ -53,6 +55,7 @@ class ViddlerIE(InfoExtractor):
'view_count': int, 'view_count': int,
'comment_count': int, 'comment_count': int,
}, },
'skip': 'Invalid URL',
}, { }, {
# secret protected # secret protected
'url': 'http://www.viddler.com/v/890c0985?secret=34051570', 'url': 'http://www.viddler.com/v/890c0985?secret=34051570',
@ -71,6 +74,15 @@ class ViddlerIE(InfoExtractor):
'skip_download': True, 'skip_download': True,
}, },
}] }]
_WEBPAGE_TESTS = [{
'url': 'https://deadspin.com/i-cant-stop-watching-john-wall-chop-the-nuggets-with-th-1681801597/',
'info_dict': {
'id': '4d03aad9',
'ext': 'mp4',
'title': 'WALL-TO-GORTAT',
},
'skip': 'Site no longer embeds Viddler',
}]
def _real_extract(self, url): def _real_extract(self, url):
video_id, secret = self._match_valid_url(url).groups() video_id, secret = self._match_valid_url(url).groups()

25
yt_dlp/extractor/videa.py
Unescape Escape View File

@ -37,7 +37,7 @@ class VideaIE(InfoExtractor):
'id': '8YfIAjxwWGwT8HVQ', 'id': '8YfIAjxwWGwT8HVQ',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Az őrült kígyász 285 kígyót enged szabadon', 'title': 'Az őrült kígyász 285 kígyót enged szabadon',
'thumbnail': r're:^https?://.*', 'thumbnail': r're:https?://videa\.hu/static/still/.+',
'duration': 21, 'duration': 21,
'age_limit': 0, 'age_limit': 0,
}, },
@ -48,7 +48,7 @@ class VideaIE(InfoExtractor):
'id': 'jAHDWfWSJH5XuFhH', 'id': 'jAHDWfWSJH5XuFhH',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Supercars előzés', 'title': 'Supercars előzés',
'thumbnail': r're:^https?://.*', 'thumbnail': r're:https?://videa\.hu/static/still/.+',
'duration': 64, 'duration': 64,
'age_limit': 0, 'age_limit': 0,
}, },
@ -59,7 +59,7 @@ class VideaIE(InfoExtractor):
'id': '8YfIAjxwWGwT8HVQ', 'id': '8YfIAjxwWGwT8HVQ',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Az őrült kígyász 285 kígyót enged szabadon', 'title': 'Az őrült kígyász 285 kígyót enged szabadon',
'thumbnail': r're:^https?://.*', 'thumbnail': r're:https?://videa\.hu/static/still/.+',
'duration': 21, 'duration': 21,
'age_limit': 0, 'age_limit': 0,
}, },
@ -76,6 +76,25 @@ class VideaIE(InfoExtractor):
'url': 'https://videakid.hu/player/v/8YfIAjxwWGwT8HVQ?autoplay=1', 'url': 'https://videakid.hu/player/v/8YfIAjxwWGwT8HVQ?autoplay=1',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
'url': 'https://www.kapucziner.hu/',
'info_dict': {
'id': '95yhJCdK2dX1T5Nh',
'ext': 'mp4',
'title': 'Nemzetközi díjat kapott a győri kávémanufaktúra',
'age_limit': 0,
'duration': 207,
'thumbnail': r're:https?://videa\.hu/static/still/.+',
},
}, {
# FIXME: No video formats found
'url': 'https://hirtv.hu/hirtv_kesleltetett',
'info_dict': {
'id': 'IDRqF7W9X0GXHGj1',
'ext': 'mp4',
'title': 'Hír TV - 60 perccel késleltetett adás',
},
}]
_STATIC_SECRET = 'xHb0ZvME5q8CBcoQi6AngerDu3FGO9fkUlwPmLVY_RTzj2hJIS4NasXWKy1td7p' _STATIC_SECRET = 'xHb0ZvME5q8CBcoQi6AngerDu3FGO9fkUlwPmLVY_RTzj2hJIS4NasXWKy1td7p'
@staticmethod @staticmethod

19
yt_dlp/extractor/videopress.py
Unescape Escape View File

@ -23,8 +23,9 @@ class VideoPressIE(InfoExtractor):
'id': 'kUJmAcSf', 'id': 'kUJmAcSf',
'ext': 'mp4', 'ext': 'mp4',
'title': 'VideoPress Demo', 'title': 'VideoPress Demo',
'thumbnail': r're:^https?://.*\.jpg', 'description': '',
'duration': 634.6, 'duration': 635.0,
'thumbnail': r're:https?://videos\.files\.wordpress\.com/.+\.jpg',
'timestamp': 1434983935, 'timestamp': 1434983935,
'upload_date': '20150622', 'upload_date': '20150622',
'age_limit': 0, 'age_limit': 0,
@ -37,6 +38,20 @@ class VideoPressIE(InfoExtractor):
'url': 'https://video.wordpress.com/embed/kUJmAcSf', 'url': 'https://video.wordpress.com/embed/kUJmAcSf',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
'url': 'https://wordpress.com/support/videopress/',
'info_dict': {
'id': 'BZHMfMfN',
'ext': 'mp4',
'title': 'videopress example',
'age_limit': 0,
'description': '',
'duration': 19.796,
'thumbnail': r're:https?://videos\.files\.wordpress\.com/.+\.jpg',
'timestamp': 1748969554,
'upload_date': '20250603',
},
}]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)

49
yt_dlp/extractor/viewlift.py
Unescape Escape View File

@ -76,6 +76,7 @@ class ViewLiftEmbedIE(ViewLiftBaseIE):
'timestamp': 1334350096, 'timestamp': 1334350096,
'upload_date': '20120413', 'upload_date': '20120413',
}, },
'skip': 'Invalid URL',
}, { }, {
# invalid labels, 360p is better that 480p # invalid labels, 360p is better that 480p
'url': 'http://www.snagfilms.com/embed/player?filmId=17ca0950-a74a-11e0-a92a-0026bb61d036', 'url': 'http://www.snagfilms.com/embed/player?filmId=17ca0950-a74a-11e0-a92a-0026bb61d036',
@ -90,6 +91,15 @@ class ViewLiftEmbedIE(ViewLiftBaseIE):
'url': 'http://www.snagfilms.com/embed/player?filmId=0000014c-de2f-d5d6-abcf-ffef58af0017', 'url': 'http://www.snagfilms.com/embed/player?filmId=0000014c-de2f-d5d6-abcf-ffef58af0017',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
'url': 'http://whilewewatch.blogspot.ru/2012/06/whilewewatch-whilewewatch-gripping.html',
'info_dict': {
'id': '74849a00-85a9-11e1-9660-123139220831',
'ext': 'mp4',
'title': '#whilewewatch',
},
'skip': 'Dead embed URL',
}]
def _real_extract(self, url): def _real_extract(self, url):
domain, film_id = self._match_valid_url(url).groups() domain, film_id = self._match_valid_url(url).groups()
@ -164,13 +174,14 @@ class ViewLiftIE(ViewLiftBaseIE):
'ext': 'mp4', 'ext': 'mp4',
'title': 'Lost for Life', 'title': 'Lost for Life',
'description': 'md5:ea10b5a50405ae1f7b5269a6ec594102', 'description': 'md5:ea10b5a50405ae1f7b5269a6ec594102',
'thumbnail': r're:^https?://.*\.jpg', 'thumbnail': r're:https?://.+\.jpg',
'duration': 4489, 'duration': 4489,
'categories': 'mincount:3', 'categories': 'mincount:3',
'age_limit': 14, 'age_limit': 14,
'upload_date': '20150421', 'upload_date': '20150421',
'timestamp': 1429656820, 'timestamp': 1429656820,
}, },
'skip': 'Invalid URL',
}, { }, {
'url': 'http://www.snagfilms.com/show/the_world_cut_project/india', 'url': 'http://www.snagfilms.com/show/the_world_cut_project/india',
'md5': 'e6292e5b837642bbda82d7f8bf3fbdfd', 'md5': 'e6292e5b837642bbda82d7f8bf3fbdfd',
@ -180,11 +191,12 @@ class ViewLiftIE(ViewLiftBaseIE):
'ext': 'mp4', 'ext': 'mp4',
'title': 'India', 'title': 'India',
'description': 'md5:5c168c5a8f4719c146aad2e0dfac6f5f', 'description': 'md5:5c168c5a8f4719c146aad2e0dfac6f5f',
'thumbnail': r're:^https?://.*\.jpg', 'thumbnail': r're:https?://.+\.jpg',
'duration': 979, 'duration': 979,
'timestamp': 1399478279, 'timestamp': 1399478279,
'upload_date': '20140507', 'upload_date': '20140507',
}, },
'skip': 'Invalid URL',
}, { }, {
'url': 'http://main.snagfilms.com/augie_alone/s_2_ep_12_love', 'url': 'http://main.snagfilms.com/augie_alone/s_2_ep_12_love',
'info_dict': { 'info_dict': {
@ -193,15 +205,13 @@ class ViewLiftIE(ViewLiftBaseIE):
'ext': 'mp4', 'ext': 'mp4',
'title': 'S. 2 Ep. 12 - Love', 'title': 'S. 2 Ep. 12 - Love',
'description': 'Augie finds love.', 'description': 'Augie finds love.',
'thumbnail': r're:^https?://.*\.jpg', 'thumbnail': r're:https?://.+\.jpg',
'duration': 107, 'duration': 107,
'upload_date': '20141012', 'upload_date': '20141012',
'timestamp': 1413129540, 'timestamp': 1413129540,
'age_limit': 17, 'age_limit': 17,
}, },
'params': { 'skip': 'Invalid URL',
'skip_download': True,
},
}, { }, {
'url': 'http://main.snagfilms.com/films/title/the_freebie', 'url': 'http://main.snagfilms.com/films/title/the_freebie',
'only_matching': True, 'only_matching': True,
@ -230,10 +240,10 @@ class ViewLiftIE(ViewLiftBaseIE):
'ext': 'mp4', 'ext': 'mp4',
'title': 'Shuyopoka', 'title': 'Shuyopoka',
'description': 'md5:e28f2fb8680096a69c944d37c1fa5ffc', 'description': 'md5:e28f2fb8680096a69c944d37c1fa5ffc',
'thumbnail': r're:^https?://.*\.jpg$', 'thumbnail': r're:https?://.+\.jpg',
'upload_date': '20211006', 'upload_date': '20211006',
}, },
'params': {'skip_download': True}, 'skip': 'Subscription required',
}, { # Free film }, { # Free film
'url': 'https://www.hoichoi.tv/films/title/dadu-no1', 'url': 'https://www.hoichoi.tv/films/title/dadu-no1',
'info_dict': { 'info_dict': {
@ -241,10 +251,10 @@ class ViewLiftIE(ViewLiftBaseIE):
'ext': 'mp4', 'ext': 'mp4',
'title': 'Dadu No.1', 'title': 'Dadu No.1',
'description': 'md5:605cba408e51a79dafcb824bdeded51e', 'description': 'md5:605cba408e51a79dafcb824bdeded51e',
'thumbnail': r're:^https?://.*\.jpg$', 'thumbnail': r're:https?://.+\.jpg',
'upload_date': '20210827', 'upload_date': '20210827',
}, },
'params': {'skip_download': True}, 'skip': 'Subscription required',
}, { # Free episode }, { # Free episode
'url': 'https://www.hoichoi.tv/webseries/case-jaundice-s01-e01', 'url': 'https://www.hoichoi.tv/webseries/case-jaundice-s01-e01',
'info_dict': { 'info_dict': {
@ -252,11 +262,11 @@ class ViewLiftIE(ViewLiftBaseIE):
'ext': 'mp4', 'ext': 'mp4',
'title': 'Humans Vs. Corona', 'title': 'Humans Vs. Corona',
'description': 'md5:ca30a682b4528d02a3eb6d0427dd0f87', 'description': 'md5:ca30a682b4528d02a3eb6d0427dd0f87',
'thumbnail': r're:^https?://.*\.jpg$', 'thumbnail': r're:https?://.+\.jpg',
'upload_date': '20210830', 'upload_date': '20210830',
'series': 'Case Jaundice', 'series': 'Case Jaundice',
}, },
'params': {'skip_download': True}, 'skip': 'Invalid URL',
}, { # Free video }, { # Free video
'url': 'https://www.hoichoi.tv/videos/1549072415320-six-episode-02-hindi', 'url': 'https://www.hoichoi.tv/videos/1549072415320-six-episode-02-hindi',
'info_dict': { 'info_dict': {
@ -264,11 +274,11 @@ class ViewLiftIE(ViewLiftBaseIE):
'ext': 'mp4', 'ext': 'mp4',
'title': 'Woman in red - Hindi', 'title': 'Woman in red - Hindi',
'description': 'md5:9d21edc1827d32f8633eb67c2054fc31', 'description': 'md5:9d21edc1827d32f8633eb67c2054fc31',
'thumbnail': r're:^https?://.*\.jpg$', 'thumbnail': r're:https?://.+\.jpg',
'upload_date': '20211006', 'upload_date': '20211006',
'series': 'Six (Hindi)', 'series': 'Six (Hindi)',
}, },
'params': {'skip_download': True}, 'skip': 'Invalid URL',
}, { # Free episode }, { # Free episode
'url': 'https://www.hoichoi.tv/shows/watch-asian-paints-moner-thikana-online-season-1-episode-1', 'url': 'https://www.hoichoi.tv/shows/watch-asian-paints-moner-thikana-online-season-1-episode-1',
'info_dict': { 'info_dict': {
@ -276,23 +286,25 @@ class ViewLiftIE(ViewLiftBaseIE):
'ext': 'mp4', 'ext': 'mp4',
'title': 'Jisshu Sengupta', 'title': 'Jisshu Sengupta',
'description': 'md5:ef6ffae01a3d83438597367400f824ed', 'description': 'md5:ef6ffae01a3d83438597367400f824ed',
'thumbnail': r're:^https?://.*\.jpg$', 'thumbnail': r're:https?://.+\.jpg',
'upload_date': '20211004', 'upload_date': '20211004',
'series': 'Asian Paints Moner Thikana', 'series': 'Asian Paints Moner Thikana',
}, },
'params': {'skip_download': True}, 'skip': 'Invalid URL',
}, { # Free series }, { # Free series
'url': 'https://www.hoichoi.tv/shows/watch-moner-thikana-bengali-web-series-online', 'url': 'https://www.hoichoi.tv/shows/watch-moner-thikana-bengali-web-series-online',
'playlist_mincount': 5, 'playlist_mincount': 5,
'info_dict': { 'info_dict': {
'id': 'watch-moner-thikana-bengali-web-series-online', 'id': 'watch-moner-thikana-bengali-web-series-online',
}, },
'skip': 'Subscription required',
}, { # Premium series }, { # Premium series
'url': 'https://www.hoichoi.tv/shows/watch-byomkesh-bengali-web-series-online', 'url': 'https://www.hoichoi.tv/shows/watch-byomkesh-bengali-web-series-online',
'playlist_mincount': 14, 'playlist_mincount': 14,
'info_dict': { 'info_dict': {
'id': 'watch-byomkesh-bengali-web-series-online', 'id': 'watch-byomkesh-bengali-web-series-online',
}, },
'skip': 'Subscription required',
}, { # Premium movie }, { # Premium movie
'url': 'https://www.hoichoi.tv/movies/detective-2020', 'url': 'https://www.hoichoi.tv/movies/detective-2020',
'only_matching': True, 'only_matching': True,
@ -302,6 +314,7 @@ class ViewLiftIE(ViewLiftBaseIE):
'info_dict': { 'info_dict': {
'id': 'bn/series/sinpaat', 'id': 'bn/series/sinpaat',
}, },
'skip': 'Subscription required',
}, { # Chorki free movie }, { # Chorki free movie
'url': 'https://www.chorki.com/bn/videos/bangla-movie-bikkhov', 'url': 'https://www.chorki.com/bn/videos/bangla-movie-bikkhov',
'info_dict': { 'info_dict': {
@ -317,9 +330,7 @@ class ViewLiftIE(ViewLiftBaseIE):
'description': 'md5:71492b086450625f4374a3eb824f27dc', 'description': 'md5:71492b086450625f4374a3eb824f27dc',
'duration': 8002, 'duration': 8002,
}, },
'params': { 'skip': 'Invalid URL',
'skip_download': True,
},
}, { # Chorki Premium movie }, { # Chorki Premium movie
'url': 'https://www.chorki.com/bn/videos/something-like-an-autobiography', 'url': 'https://www.chorki.com/bn/videos/something-like-an-autobiography',
'only_matching': True, 'only_matching': True,

994
yt_dlp/extractor/vimeo.py
View File

File diff suppressed because it is too large Load Diff

15
yt_dlp/extractor/viqeo.py
Unescape Escape View File

@ -24,7 +24,7 @@ class ViqeoIE(InfoExtractor):
'id': 'cde96f09d25f39bee837', 'id': 'cde96f09d25f39bee837',
'ext': 'mp4', 'ext': 'mp4',
'title': 'cde96f09d25f39bee837', 'title': 'cde96f09d25f39bee837',
'thumbnail': r're:^https?://.*\.jpg$', 'thumbnail': r're:https?://.+\.jpg',
'duration': 76, 'duration': 76,
}, },
}, { }, {
@ -34,6 +34,19 @@ class ViqeoIE(InfoExtractor):
'url': 'https://api.viqeo.tv/v1/data/startup?video%5B%5D=71bbec412ade45c3216c&profile=112', 'url': 'https://api.viqeo.tv/v1/data/startup?video%5B%5D=71bbec412ade45c3216c&profile=112',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
'url': 'https://viqeo.tv/',
'info_dict': {
'id': 'viqeo',
'title': 'Viqeo video platform',
'age_limit': 0,
'description': 'md5:e8e06e20df92ed66febeaef2533a0d5d',
'thumbnail': r're:https?://static\.tildacdn\.com/.+\.png',
'timestamp': 1751479769,
'upload_date': '20250702',
},
'playlist_count': 3,
}]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)

10
yt_dlp/extractor/voxmedia.py
Unescape Escape View File

@ -72,6 +72,7 @@ class VoxMediaIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?(?:(?:theverge|vox|sbnation|eater|polygon|curbed|racked|funnyordie)\.com|recode\.net)/(?:[^/]+/)*(?P<id>[^/?]+)' _VALID_URL = r'https?://(?:www\.)?(?:(?:theverge|vox|sbnation|eater|polygon|curbed|racked|funnyordie)\.com|recode\.net)/(?:[^/]+/)*(?P<id>[^/?]+)'
_EMBED_REGEX = [r'<iframe[^>]+?src="(?P<url>https?://(?:www\.)?funnyordie\.com/embed/[^"]+)"'] _EMBED_REGEX = [r'<iframe[^>]+?src="(?P<url>https?://(?:www\.)?funnyordie\.com/embed/[^"]+)"']
_TESTS = [{ _TESTS = [{
# FIXME: Unsupported iframe embed
# Volume embed, Youtube # Volume embed, Youtube
'url': 'http://www.theverge.com/2014/6/27/5849272/material-world-how-google-discovered-what-software-is-made-of', 'url': 'http://www.theverge.com/2014/6/27/5849272/material-world-how-google-discovered-what-software-is-made-of',
'info_dict': { 'info_dict': {
@ -156,6 +157,15 @@ class VoxMediaIE(InfoExtractor):
}], }],
'skip': 'Page no longer contain videos', 'skip': 'Page no longer contain videos',
}] }]
_WEBPAGE_TESTS = [{
'url': 'http://www.theguardian.com/world/2014/mar/11/obama-zach-galifianakis-between-two-ferns',
'info_dict': {
'id': '18e820ec3f',
'ext': 'mp4',
'title': 'Between Two Ferns with Zach Galifianakis: President Barack Obama',
},
'skip': 'Invalid URL',
}]
def _real_extract(self, url): def _real_extract(self, url):
display_id = self._match_id(url) display_id = self._match_id(url)

19
yt_dlp/extractor/wimtv.py
Unescape Escape View File

@ -30,9 +30,7 @@ class WimTVIE(InfoExtractor):
'duration': 6481, 'duration': 6481,
'thumbnail': r're:https?://.+?/thumbnail/.+?/720$', 'thumbnail': r're:https?://.+?/thumbnail/.+?/720$',
}, },
'params': { 'skip': 'Invalid URL',
'skip_download': True,
},
}, { }, {
# live stream # live stream
'url': 'https://platform.wim.tv/embed/?live=28e22c22-49db-40f3-8c37-8cbb0ff44556&autostart=true', 'url': 'https://platform.wim.tv/embed/?live=28e22c22-49db-40f3-8c37-8cbb0ff44556&autostart=true',
@ -42,9 +40,7 @@ class WimTVIE(InfoExtractor):
'title': 'Streaming MSmotorTV', 'title': 'Streaming MSmotorTV',
'is_live': True, 'is_live': True,
}, },
'params': { 'skip': 'Invalid URL',
'skip_download': True,
},
}, { }, {
'url': 'https://platform.wim.tv/#/webtv/automotornews/vod/422492b6-539e-474d-9c6b-68c9d5893365', 'url': 'https://platform.wim.tv/#/webtv/automotornews/vod/422492b6-539e-474d-9c6b-68c9d5893365',
'only_matching': True, 'only_matching': True,
@ -52,6 +48,17 @@ class WimTVIE(InfoExtractor):
'url': 'https://platform.wim.tv/#/webtv/renzoarborechannel/cast/f47e0d15-5b45-455e-bf0d-dba8ffa96365', 'url': 'https://platform.wim.tv/#/webtv/renzoarborechannel/cast/f47e0d15-5b45-455e-bf0d-dba8ffa96365',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
'url': 'http://www.renzoarborechannel.tv/50_sorrisi_da_napoli.htm',
'info_dict': {
'id': '50_sorrisi_da_napoli',
'title': 'Renzo Arbore Channel . TV - 50 Sorrisi da Napoli',
'age_limit': 0,
'timestamp': 1612226372,
'upload_date': '20210202',
},
'playlist_count': 40,
}]
def _real_initialize(self): def _real_initialize(self):
if not self._player: if not self._player:

37
yt_dlp/extractor/wistia.py
Unescape Escape View File

@ -193,12 +193,12 @@ class WistiaIE(WistiaBaseIE):
'info_dict': { 'info_dict': {
'id': 'a6ndpko1wg', 'id': 'a6ndpko1wg',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Episode 2: Boxed Water\'s retention is thirsty', 'title': 'BXO-S02-E02-Boxed_Water-v4.mp4',
'upload_date': '20210324', 'upload_date': '20210324',
'description': 'md5:da5994c2c2d254833b412469d9666b7a', 'description': 'md5:3b9296a45aa46010767451b3691b1105',
'duration': 966.0, 'duration': 966.0,
'timestamp': 1616614369, 'timestamp': 1616614369,
'thumbnail': 'https://embed-ssl.wistia.com/deliveries/53dc60239348dc9b9fba3755173ea4c2.png', 'thumbnail': r're:https?://embed(?:-ssl)?\.wistia\.com/.+\.(?:jpg|png)',
}, },
}, { }, {
'url': 'wistia:5vd7p4bct5', 'url': 'wistia:5vd7p4bct5',
@ -211,7 +211,7 @@ class WistiaIE(WistiaBaseIE):
'upload_date': '20220915', 'upload_date': '20220915',
'timestamp': 1663258727, 'timestamp': 1663258727,
'duration': 623.019, 'duration': 623.019,
'thumbnail': r're:https?://embed(?:-ssl)?.wistia.com/.+\.jpg$', 'thumbnail': r're:https?://embed(?:-ssl)?\.wistia\.com/.+\.(?:jpg|png)',
}, },
}, { }, {
'url': 'wistia:sh7fpupwlt', 'url': 'wistia:sh7fpupwlt',
@ -226,7 +226,6 @@ class WistiaIE(WistiaBaseIE):
'url': 'http://fast.wistia.net/embed/medias/sh7fpupwlt.json', 'url': 'http://fast.wistia.net/embed/medias/sh7fpupwlt.json',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{ _WEBPAGE_TESTS = [{
'url': 'https://www.weidert.com/blog/wistia-channels-video-marketing-tool', 'url': 'https://www.weidert.com/blog/wistia-channels-video-marketing-tool',
'info_dict': { 'info_dict': {
@ -237,8 +236,9 @@ class WistiaIE(WistiaBaseIE):
'timestamp': 1618974400, 'timestamp': 1618974400,
'description': 'md5:27abc99a758573560be72600ef95cece', 'description': 'md5:27abc99a758573560be72600ef95cece',
'upload_date': '20210421', 'upload_date': '20210421',
'thumbnail': 'https://embed-ssl.wistia.com/deliveries/6c551820ae950cdee2306d6cbe9ef742.jpg', 'thumbnail': r're:https?://embed(?:-ssl)?\.wistia\.com/.+\.(?:jpg|png)',
}, },
'skip': 'Invalid URL',
}, { }, {
'url': 'https://study.com/academy/lesson/north-american-exploration-failed-colonies-of-spain-france-england.html#lesson', 'url': 'https://study.com/academy/lesson/north-american-exploration-failed-colonies-of-spain-france-england.html#lesson',
'md5': 'b9676d24bf30945d97060638fbfe77f0', 'md5': 'b9676d24bf30945d97060638fbfe77f0',
@ -249,9 +249,19 @@ class WistiaIE(WistiaBaseIE):
'upload_date': '20220915', 'upload_date': '20220915',
'timestamp': 1663258727, 'timestamp': 1663258727,
'duration': 623.019, 'duration': 623.019,
'thumbnail': 'https://embed-ssl.wistia.com/deliveries/83e6ec693e2c05a0ce65809cbaead86a.jpg', 'thumbnail': r're:https?://embed(?:-ssl)?\.wistia\.com/.+\.(?:jpg|png)',
'description': 'a Paywall Videos video', 'description': 'a Paywall Videos video',
}, },
}, {
'url': 'https://support.wistia.com/en/articles/8233354-embedding-your-media',
'info_dict': {
'id': '8233354-embedding-your-media',
'title': 'Embedding Your Media | Wistia Help Center',
'age_limit': 0,
'description': 'md5:32a5edc0e266cd61e2d15be28873d614',
'thumbnail': r're:https?://downloads\.intercomcdn\.com/.+\.jpg',
},
'playlist_count': 2,
}] }]
def _real_extract(self, url): def _real_extract(self, url):
@ -278,13 +288,13 @@ class WistiaIE(WistiaBaseIE):
class WistiaPlaylistIE(WistiaBaseIE): class WistiaPlaylistIE(WistiaBaseIE):
_VALID_URL = rf'{WistiaBaseIE._VALID_URL_BASE}playlists/{WistiaBaseIE._VALID_ID_REGEX}' _VALID_URL = rf'{WistiaBaseIE._VALID_URL_BASE}playlists/{WistiaBaseIE._VALID_ID_REGEX}'
_TEST = { _TESTS = [{
'url': 'https://fast.wistia.net/embed/playlists/aodt9etokc', 'url': 'https://fast.wistia.net/embed/playlists/aodt9etokc',
'info_dict': { 'info_dict': {
'id': 'aodt9etokc', 'id': 'aodt9etokc',
}, },
'playlist_count': 3, 'playlist_count': 3,
} }]
def _real_extract(self, url): def _real_extract(self, url):
playlist_id = self._match_id(url) playlist_id = self._match_id(url)
@ -312,7 +322,7 @@ class WistiaChannelIE(WistiaBaseIE):
'description': 'Learn all things Copysmith via short and informative videos!', 'description': 'Learn all things Copysmith via short and informative videos!',
}, },
'playlist_mincount': 7, 'playlist_mincount': 7,
'expected_warnings': ['falling back to webpage'], 'skip': 'Invalid URL',
}, { }, {
'url': 'https://fast.wistia.net/embed/channel/3802iirk0l', 'url': 'https://fast.wistia.net/embed/channel/3802iirk0l',
'info_dict': { 'info_dict': {
@ -327,7 +337,7 @@ class WistiaChannelIE(WistiaBaseIE):
'id': 'sp5dqjzw3n', 'id': 'sp5dqjzw3n',
'ext': 'mp4', 'ext': 'mp4',
'title': 'The Roof S2: The Modern CRO', 'title': 'The Roof S2: The Modern CRO',
'thumbnail': 'https://embed-ssl.wistia.com/deliveries/dadfa9233eaa505d5e0c85c23ff70741.png', 'thumbnail': r're:https?://embed(?:-ssl)?\.wistia\.com/.+\.(?:jpg|png)',
'duration': 86.487, 'duration': 86.487,
'description': 'A sales leader on The Roof? Man, they really must be letting anyone up here this season.\n', 'description': 'A sales leader on The Roof? Man, they really must be letting anyone up here this season.\n',
'timestamp': 1619790290, 'timestamp': 1619790290,
@ -343,6 +353,7 @@ class WistiaChannelIE(WistiaBaseIE):
'description': 'md5:14a8a93a1dbe236718e6a59f8c8c7bae', 'description': 'md5:14a8a93a1dbe236718e6a59f8c8c7bae',
}, },
'playlist_mincount': 30, 'playlist_mincount': 30,
'skip': 'Site no longer embeds Wistia playlists',
}, { }, {
# section instead of div # section instead of div
'url': 'https://360learning.com/studio/onboarding-joei/', 'url': 'https://360learning.com/studio/onboarding-joei/',
@ -362,9 +373,9 @@ class WistiaChannelIE(WistiaBaseIE):
'upload_date': '20220530', 'upload_date': '20220530',
'description': 'Learn how to help your company improve and achieve your product related goals.', 'description': 'Learn how to help your company improve and achieve your product related goals.',
'duration': 1854.39, 'duration': 1854.39,
'thumbnail': 'https://embed-ssl.wistia.com/deliveries/12fd19e56413d9d6f04e2185c16a6f8854e25226.png', 'thumbnail': r're:https?://embed(?:-ssl)?\.wistia\.com/.+\.(?:jpg|png)',
}, },
'params': {'noplaylist': True, 'skip_download': True}, 'skip': 'Invalid URL',
}] }]
def _real_extract(self, url): def _real_extract(self, url):

50
yt_dlp/extractor/xhamster.py
Unescape Escape View File

@ -43,7 +43,7 @@ class XHamsterIE(InfoExtractor):
'uploader_id': 'ruseful2011', 'uploader_id': 'ruseful2011',
'duration': 893, 'duration': 893,
'age_limit': 18, 'age_limit': 18,
'thumbnail': 'https://thumb-nss.xhcdn.com/a/u3Vr5F2vvcU3yK59_jJqVA/001/509/445/1280x720.8.jpg', 'thumbnail': r're:https?://.+\.jpg',
'uploader_url': 'https://xhamster.com/users/ruseful2011', 'uploader_url': 'https://xhamster.com/users/ruseful2011',
'description': '', 'description': '',
'view_count': int, 'view_count': int,
@ -63,11 +63,12 @@ class XHamsterIE(InfoExtractor):
'age_limit': 18, 'age_limit': 18,
'description': '', 'description': '',
'view_count': int, 'view_count': int,
'thumbnail': 'https://thumb-nss.xhcdn.com/a/kk5nio_iR-h4Z3frfVtoDw/002/221/348/1280x720.4.jpg', 'thumbnail': r're:https?://.+\.jpg',
'comment_count': int, 'comment_count': int,
}, },
'params': { 'params': {
'skip_download': True, 'extractor_args': {'generic': {'impersonate': ['chrome']}},
'skip_download': 'm3u8',
}, },
}, { }, {
# empty seo, unavailable via new URL schema # empty seo, unavailable via new URL schema
@ -86,11 +87,9 @@ class XHamsterIE(InfoExtractor):
'uploader_url': 'https://xhamster.com/users/parejafree', 'uploader_url': 'https://xhamster.com/users/parejafree',
'description': '', 'description': '',
'view_count': int, 'view_count': int,
'thumbnail': 'https://thumb-nss.xhcdn.com/a/xc8MSwVKcsQeRRiTT-saMQ/005/667/973/1280x720.2.jpg', 'thumbnail': r're:https?://.+\.jpg',
},
'params': {
'skip_download': True,
}, },
'skip': 'Invalid URL',
}, { }, {
# mobile site # mobile site
'url': 'https://m.xhamster.com/videos/cute-teen-jacqueline-solo-masturbation-8559111', 'url': 'https://m.xhamster.com/videos/cute-teen-jacqueline-solo-masturbation-8559111',
@ -390,19 +389,48 @@ class XHamsterIE(InfoExtractor):
class XHamsterEmbedIE(InfoExtractor): class XHamsterEmbedIE(InfoExtractor):
_VALID_URL = rf'https?://(?:[^/?#]+\.)?{XHamsterIE._DOMAINS}/xembed\.php\?video=(?P<id>\d+)' _VALID_URL = rf'https?://(?:[^/?#]+\.)?{XHamsterIE._DOMAINS}/xembed\.php\?video=(?P<id>\d+)'
_EMBED_REGEX = [r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?xhamster\.com/xembed\.php\?video=\d+)\1'] _EMBED_REGEX = [r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?xhamster\.com/xembed\.php\?video=\d+)\1']
_TEST = { _TESTS = [{
'url': 'http://xhamster.com/xembed.php?video=3328539', 'url': 'http://xhamster.com/xembed.php?video=3328539',
'info_dict': { 'info_dict': {
'id': '3328539', 'id': '3328539',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Pen Masturbation', 'title': 'Pen Masturbation',
'comment_count': int,
'description': '',
'display_id': 'pen-masturbation',
'timestamp': 1406581861, 'timestamp': 1406581861,
'upload_date': '20140728', 'upload_date': '20140728',
'uploader': 'ManyakisArt', 'uploader': 'ManyakisArt',
'duration': 5, 'duration': 5,
'age_limit': 18, 'age_limit': 18,
'thumbnail': r're:https?://.+\.jpg',
'uploader_id': 'manyakisart',
'uploader_url': 'https://xhamster.com/users/manyakisart',
'view_count': int,
}, },
} }]
_WEBPAGE_TESTS = [{
# FIXME: Embed detection
'url': 'https://xhamster.com/awards/2023',
'info_dict': {
'id': 'xh2VnYn',
'ext': 'mp4',
'title': 'xHamster Awards 2023 - The Winners',
'age_limit': 18,
'comment_count': int,
'description': '',
'display_id': 'xhamster-awards-2023-the-winners',
'duration': 292,
'thumbnail': r're:https?://ic-vt-nss\.xhcdn\.com/.+',
'timestamp': 1700122082,
'upload_date': '20231116',
'uploader': 'xHamster',
'uploader_id': 'xhamster',
'uploader_url': 'https://xhamster.com/users/xhamster',
'view_count': int,
},
'params': {'skip_download': 'm3u8'},
}]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)
@ -437,13 +465,13 @@ class XHamsterUserIE(InfoExtractor):
'info_dict': { 'info_dict': {
'id': 'firatkaan', 'id': 'firatkaan',
}, },
'playlist_mincount': 1, 'playlist_mincount': 0,
}, { }, {
'url': 'https://xhamster.com/creators/squirt-orgasm-69', 'url': 'https://xhamster.com/creators/squirt-orgasm-69',
'info_dict': { 'info_dict': {
'id': 'squirt-orgasm-69', 'id': 'squirt-orgasm-69',
}, },
'playlist_mincount': 150, 'playlist_mincount': 46,
}, { }, {
'url': 'https://xhday.com/users/mobhunter', 'url': 'https://xhday.com/users/mobhunter',
'only_matching': True, 'only_matching': True,

9
yt_dlp/extractor/yapfiles.py
Unescape Escape View File

@ -28,6 +28,15 @@ class YapFilesIE(InfoExtractor):
'url': 'https://api.yapfiles.ru/get_player/?uid=video_player_1872528&plroll=1&adv=1&v=vMDE4NzI1Mjgt690b', 'url': 'https://api.yapfiles.ru/get_player/?uid=video_player_1872528&plroll=1&adv=1&v=vMDE4NzI1Mjgt690b',
'only_matching': True, 'only_matching': True,
}] }]
_WEBPAGE_TESTS = [{
# FIXME: Update _VALID_URL
'url': 'https://www.yapfiles.ru/show/3397030/e34b69aa03829d513d7dc3ace6ec9631.mp4.html',
'info_dict': {
'id': 'vMDE4NzI1Mjgt690b',
'ext': 'mp4',
'title': 'Котята',
},
}]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)

2967
yt_dlp/extractor/youtube/_video.py
View File

File diff suppressed because it is too large Load Diff

58
yt_dlp/extractor/zapiks.py
Unescape Escape View File

@ -13,35 +13,39 @@ from ..utils import (
class ZapiksIE(InfoExtractor): class ZapiksIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?zapiks\.(?:fr|com)/(?:(?:[a-z]{2}/)?(?P<display_id>.+?)\.html|index\.php\?.*\bmedia_id=(?P<id>\d+))' _VALID_URL = r'https?://(?:www\.)?zapiks\.(?:fr|com)/(?:(?:[a-z]{2}/)?(?P<display_id>.+?)\.html|index\.php\?.*\bmedia_id=(?P<id>\d+))'
_EMBED_REGEX = [r'<iframe[^>]+src="(?P<url>https?://(?:www\.)?zapiks\.fr/index\.php\?.+?)"'] _EMBED_REGEX = [r'<iframe[^>]+src="(?P<url>https?://(?:www\.)?zapiks\.fr/index\.php\?.+?)"']
_TESTS = [ _TESTS = [{
{ 'url': 'http://www.zapiks.fr/ep2s3-bon-appetit-eh-be-viva.html',
'url': 'http://www.zapiks.fr/ep2s3-bon-appetit-eh-be-viva.html', 'md5': 'aeb3c473b2d564b2d46d664d28d5f050',
'md5': 'aeb3c473b2d564b2d46d664d28d5f050', 'info_dict': {
'info_dict': { 'id': '80798',
'id': '80798', 'ext': 'mp4',
'ext': 'mp4', 'title': 'EP2S3 - Bon Appétit - Eh bé viva les pyrénées con!',
'title': 'EP2S3 - Bon Appétit - Eh bé viva les pyrénées con!', 'description': 'md5:7054d6f6f620c6519be1fe710d4da847',
'description': 'md5:7054d6f6f620c6519be1fe710d4da847', 'thumbnail': r're:https?://zpks\.com/.+\.jpg',
'thumbnail': r're:^https?://.*\.jpg$', 'duration': 528,
'duration': 528, 'timestamp': 1359044972,
'timestamp': 1359044972, 'upload_date': '20130124',
'upload_date': '20130124', 'view_count': int,
'view_count': int,
},
}, },
{ }, {
'url': 'http://www.zapiks.com/ep3s5-bon-appetit-baqueira-m-1.html', 'url': 'http://www.zapiks.com/ep3s5-bon-appetit-baqueira-m-1.html',
'only_matching': True, 'only_matching': True,
}, {
'url': 'http://www.zapiks.com/nl/ep3s5-bon-appetit-baqueira-m-1.html',
'only_matching': True,
}, {
'url': 'http://www.zapiks.fr/index.php?action=playerIframe&amp;media_id=118046&amp;width=640&amp;height=360&amp;autoStart=false&amp;language=fr',
'only_matching': True,
}]
_WEBPAGE_TESTS = [{
'url': 'https://www.skipass.com/news/116090-bon-appetit-s5ep3-baqueira-mi-cor.html',
'info_dict': {
'id': '118046',
'ext': 'mp4',
'title': 'EP3S5 - Bon Appétit - Baqueira Mi Corazon !',
'thumbnail': r're:https?://zpks\.com/.+\.jpg',
}, },
{ }]
'url': 'http://www.zapiks.com/nl/ep3s5-bon-appetit-baqueira-m-1.html',
'only_matching': True,
},
{
'url': 'http://www.zapiks.fr/index.php?action=playerIframe&amp;media_id=118046&amp;width=640&amp;height=360&amp;autoStart=false&amp;language=fr',
'only_matching': True,
},
]
def _real_extract(self, url): def _real_extract(self, url):
mobj = self._match_valid_url(url) mobj = self._match_valid_url(url)

Write Preview
Loading…
Cancel
Save