diff --git a/yt_dlp/extractor/embedly.py b/yt_dlp/extractor/embedly.py index 483d018bb4..db5ef055ec 100644 --- a/yt_dlp/extractor/embedly.py +++ b/yt_dlp/extractor/embedly.py @@ -1,13 +1,63 @@ import re import urllib.parse + from .common import InfoExtractor -from ..compat import compat_urllib_parse_unquote +from .youtube import YoutubeTabIE +from ..utils import parse_qs, smuggle_url, traverse_obj class EmbedlyIE(InfoExtractor): - _VALID_URL = r'https?://(?:www|cdn\.)?embedly\.com/widgets/media\.html\?(?:[^#]*?&)?url=(?P[^#&]+)' + _VALID_URL = r'https?://(?:www|cdn\.)?embedly\.com/widgets/media\.html\?(?:[^#]*?&)?(?:src|url)=(?:[^#&]+)' _TESTS = [{ 'url': 'https://cdn.embedly.com/widgets/media.html?src=http%3A%2F%2Fwww.youtube.com%2Fembed%2Fvideoseries%3Flist%3DUUGLim4T2loE5rwCMdpCIPVg&url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DSU4fj_aEMVw%26list%3DUUGLim4T2loE5rwCMdpCIPVg&image=http%3A%2F%2Fi.ytimg.com%2Fvi%2FSU4fj_aEMVw%2Fhqdefault.jpg&key=8ee8a2e6a8cc47aab1a5ee67f9a178e0&type=text%2Fhtml&schema=youtube&autoplay=1', + 'info_dict': { + 'id': 'UUGLim4T2loE5rwCMdpCIPVg', + 'modified_date': '20221225', + 'view_count': int, + 'uploader_url': 'https://www.youtube.com/@TraciHinesMusic', + 'channel_id': 'UCGLim4T2loE5rwCMdpCIPVg', + 'uploader': 'TraciJHines', + 'channel_url': 'https://www.youtube.com/@TraciHinesMusic', + 'channel': 'TraciJHines', + 'availability': 'public', + 'uploader_id': 'UCGLim4T2loE5rwCMdpCIPVg', + 'description': '', + 'tags': [], + 'title': 'Uploads from TraciJHines', + }, + 'playlist_mincount': 10, + }, { + 'url': 'https://cdn.embedly.com/widgets/media.html?src=http%3A%2F%2Fwww.youtube.com%2Fembed%2Fvideoseries%3Flist%3DUUGLim4T2loE5rwCMdpCIPVg&url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DSU4fj_aEMVw%26list%3DUUGLim4T2loE5rwCMdpCIPVg&image=http%3A%2F%2Fi.ytimg.com%2Fvi%2FSU4fj_aEMVw%2Fhqdefault.jpg&key=8ee8a2e6a8cc47aab1a5ee67f9a178e0&type=text%2Fhtml&schema=youtube&autoplay=1', + 'params': {'noplaylist': True}, + 'info_dict': { + 'id': 'SU4fj_aEMVw', + 'ext': 'mp4', + 'title': 'I\'m on Patreon!', + 'age_limit': 0, + 'categories': ['Entertainment'], + 'thumbnail': 'https://i.ytimg.com/vi_webp/SU4fj_aEMVw/maxresdefault.webp', + 'live_status': 'not_live', + 'playable_in_embed': True, + 'channel': 'TraciJHines', + 'uploader_id': 'TraciJHines', + 'channel_url': 'https://www.youtube.com/channel/UCGLim4T2loE5rwCMdpCIPVg', + 'uploader_url': 'http://www.youtube.com/user/TraciJHines', + 'upload_date': '20150211', + 'duration': 282, + 'availability': 'public', + 'channel_follower_count': int, + 'tags': 'count:39', + 'view_count': int, + 'comment_count': int, + 'channel_id': 'UCGLim4T2loE5rwCMdpCIPVg', + 'like_count': int, + 'uploader': 'TraciJHines', + 'description': 'md5:8af6425f50bd46fbf29f3db0fc3a8364', + 'chapters': list, + + }, + }, { + 'url': 'https://cdn.embedly.com/widgets/media.html?src=https://player.vimeo.com/video/1234567?h=abcdefgh', 'only_matching': True, }] @@ -21,4 +71,10 @@ class EmbedlyIE(InfoExtractor): yield urllib.parse.unquote(mobj.group('url')) def _real_extract(self, url): - return self.url_result(compat_urllib_parse_unquote(self._match_id(url))) + qs = parse_qs(url) + src = urllib.parse.unquote(traverse_obj(qs, ('url', 0)) or '') + if src and YoutubeTabIE.suitable(src): + return self.url_result(src, YoutubeTabIE) + return self.url_result(smuggle_url( + urllib.parse.unquote(traverse_obj(qs, ('src', 0), ('url', 0))), + {'http_headers': {'Referer': url}}))