0
0
Fork 0
mirror of https://github.com/yt-dlp/yt-dlp.git synced 2024-12-05 04:17:27 +00:00
yt-dlp/yt_dlp/extractor/groupon.py

64 lines
2.4 KiB
Python
Raw Normal View History

2014-12-12 16:39:33 +00:00
from .common import InfoExtractor
class GrouponIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?groupon\.com/deals/(?P<id>[^/?#&]+)'
2014-12-12 16:39:33 +00:00
_TEST = {
'url': 'https://www.groupon.com/deals/bikram-yoga-huntington-beach-2#ooid=tubGNycTo_9Uxg82uESj4i61EYX8nyuf',
'info_dict': {
'id': 'bikram-yoga-huntington-beach-2',
'title': '$49 for 10 Yoga Classes or One Month of Unlimited Classes at Bikram Yoga Huntington Beach ($180 Value)',
'description': 'Studio kept at 105 degrees and 40% humidity with anti-microbial and anti-slip Flotex flooring; certified instructors',
},
'playlist': [{
'md5': '42428ce8a00585f9bc36e49226eae7a1',
2014-12-12 16:39:33 +00:00
'info_dict': {
'id': 'fk6OhWpXgIQ',
'ext': 'mp4',
'title': 'Bikram Yoga Huntington Beach | Orange County !tubGNycTo@9Uxg82uESj4i61EYX8nyuf',
'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
'duration': 45,
'upload_date': '20160405',
'uploader_id': 'groupon',
'uploader': 'Groupon',
2014-12-12 16:39:33 +00:00
},
'add_ie': ['Youtube'],
2016-05-24 15:42:22 +00:00
}],
'params': {
'skip_download': True,
},
2014-12-12 16:39:33 +00:00
}
_PROVIDERS = {
'youtube': ('%s', 'Youtube'),
}
2014-12-12 16:39:33 +00:00
def _real_extract(self, url):
playlist_id = self._match_id(url)
webpage = self._download_webpage(url, playlist_id)
payload = self._parse_json(self._search_regex(
2016-04-12 09:26:09 +00:00
r'(?:var\s+|window\.)payload\s*=\s*(.*?);\n', webpage, 'payload'), playlist_id)
2014-12-12 16:39:33 +00:00
videos = payload['carousel'].get('dealVideos', [])
entries = []
for v in videos:
provider = v.get('provider')
video_id = v.get('media') or v.get('id') or v.get('baseURL')
if not provider or not video_id:
continue
url_pattern, ie_key = self._PROVIDERS.get(provider.lower())
if not url_pattern:
2014-12-12 16:39:33 +00:00
self.report_warning(
f'{playlist_id}: Unsupported video provider {provider}, skipping video')
2014-12-12 16:39:33 +00:00
continue
entries.append(self.url_result(url_pattern % video_id, ie_key))
2014-12-12 16:39:33 +00:00
return {
'_type': 'playlist',
'id': playlist_id,
'entries': entries,
'title': self._og_search_title(webpage),
'description': self._og_search_description(webpage),
}