diff options
| author | Sergey M․ <dstftw@gmail.com> | 2018-04-18 01:24:02 +0700 |
|---|---|---|
| committer | Sergey M․ <dstftw@gmail.com> | 2018-04-18 01:26:15 +0700 |
| commit | e30991f9206f98605ba6c4880ed40ad5556fa0b8 (patch) | |
| tree | 11421daf6777a150ecffde2abd6ac5b55fd95020 | |
| parent | 518d5ba5191e3cc26c81e346ba5117e94db51469 (diff) | |
| download | youtube-dl-e30991f9206f98605ba6c4880ed40ad5556fa0b8.tar.xz | |
[kaltura] Improve embeds detection (closes #16201)
| -rw-r--r-- | youtube_dl/extractor/generic.py | 18 | ||||
| -rw-r--r-- | youtube_dl/extractor/kaltura.py | 6 |
2 files changed, 20 insertions, 4 deletions
diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py index e3cb5c5ce..af1322e00 100644 --- a/youtube_dl/extractor/generic.py +++ b/youtube_dl/extractor/generic.py @@ -1220,7 +1220,7 @@ class GenericIE(InfoExtractor): 'title': '35871', 'timestamp': 1355743100, 'upload_date': '20121217', - 'uploader_id': 'batchUser', + 'uploader_id': 'cplapp@learn360.com', }, 'add_ie': ['Kaltura'], }, @@ -1271,6 +1271,22 @@ class GenericIE(InfoExtractor): }, 'add_ie': ['Kaltura'], }, + { + # meta twitter:player + 'url': 'http://thechive.com/2017/12/08/all-i-want-for-christmas-is-more-twerk/', + 'info_dict': { + 'id': '0_01b42zps', + 'ext': 'mp4', + 'title': 'Main Twerk (Video)', + 'upload_date': '20171208', + 'uploader_id': 'sebastian.salinas@thechive.com', + 'timestamp': 1512713057, + }, + 'params': { + 'skip_download': True, + }, + 'add_ie': ['Kaltura'], + }, # referrer protected EaglePlatform embed { 'url': 'https://tvrain.ru/lite/teleshow/kak_vse_nachinalos/namin-418921/', diff --git a/youtube_dl/extractor/kaltura.py b/youtube_dl/extractor/kaltura.py index 562e25f6d..0ea89e4d6 100644 --- a/youtube_dl/extractor/kaltura.py +++ b/youtube_dl/extractor/kaltura.py @@ -135,10 +135,10 @@ class KalturaIE(InfoExtractor): ''', webpage) or re.search( r'''(?xs) - <iframe[^>]+src=(?P<q1>["']) - (?:https?:)?//(?:www\.)?kaltura\.com/(?:(?!(?P=q1)).)*\b(?:p|partner_id)/(?P<partner_id>\d+) + <(?:iframe[^>]+src|meta[^>]+\bcontent)=(?P<q1>["']) + (?:https?:)?//(?:(?:www|cdnapi)\.)?kaltura\.com/(?:(?!(?P=q1)).)*\b(?:p|partner_id)/(?P<partner_id>\d+) (?:(?!(?P=q1)).)* - [?&]entry_id=(?P<id>(?:(?!(?P=q1))[^&])+) + [?&;]entry_id=(?P<id>(?:(?!(?P=q1))[^&])+) (?P=q1) ''', webpage) ) |