js_to_json,
parse_age_limit,
parse_duration,
+ try_get,
)
class ViewLiftBaseIE(InfoExtractor):
- _DOMAINS_REGEX = r'(?:snagfilms|snagxtreme|funnyforfree|kiddovid|winnersview|(?:monumental|lax)sportsnetwork|vayafilm)\.com|hoichoi\.tv'
+ _DOMAINS_REGEX = r'(?:(?:main\.)?snagfilms|snagxtreme|funnyforfree|kiddovid|winnersview|(?:monumental|lax)sportsnetwork|vayafilm)\.com|hoichoi\.tv'
class ViewLiftEmbedIE(ViewLiftBaseIE):
class ViewLiftIE(ViewLiftBaseIE):
- _VALID_URL = r'https?://(?:www\.)?(?P<domain>%s)/(?:films/title|show|(?:news/)?videos?)/(?P<id>[^?#]+)' % ViewLiftBaseIE._DOMAINS_REGEX
+ _VALID_URL = r'https?://(?:www\.)?(?P<domain>%s)(?:/(?:films/title|show|(?:news/)?videos?))?/(?P<id>[^?#]+)' % ViewLiftBaseIE._DOMAINS_REGEX
_TESTS = [{
'url': 'http://www.snagfilms.com/films/title/lost_for_life',
'md5': '19844f897b35af219773fd63bdec2942',
'categories': 'mincount:3',
'age_limit': 14,
'upload_date': '20150421',
- 'timestamp': 1429656819,
+ 'timestamp': 1429656820,
}
}, {
'url': 'http://www.snagfilms.com/show/the_world_cut_project/india',
'description': 'md5:5c168c5a8f4719c146aad2e0dfac6f5f',
'thumbnail': r're:^https?://.*\.jpg',
'duration': 979,
- 'categories': 'mincount:2',
'timestamp': 1399478279,
'upload_date': '20140507',
}
+ }, {
+ 'url': 'http://main.snagfilms.com/augie_alone/s_2_ep_12_love',
+ 'info_dict': {
+ 'id': '00000148-7b53-de26-a9fb-fbf306f70020',
+ 'display_id': 'augie_alone/s_2_ep_12_love',
+ 'ext': 'mp4',
+ 'title': 'Augie, Alone:S. 2 Ep. 12 - Love',
+ 'description': 'md5:db2a5c72d994f16a780c1eb353a8f403',
+ 'thumbnail': r're:^https?://.*\.jpg',
+ 'duration': 107,
+ },
+ 'params': {
+ 'skip_download': True,
+ },
+ }, {
+ 'url': 'http://main.snagfilms.com/films/title/the_freebie',
+ 'only_matching': True,
}, {
# Film is not playable in your area.
'url': 'http://www.snagfilms.com/films/title/inside_mecca',
gist = content_data['gist']
film_id = gist['id']
title = gist['title']
- video_assets = content_data['streamingInfo']['videoAssets']
+ video_assets = try_get(
+ content_data, lambda x: x['streamingInfo']['videoAssets'], dict)
+ if not video_assets:
+ token = self._download_json(
+ 'https://prod-api.viewlift.com/identity/anonymous-token',
+ film_id, 'Downloading authorization token',
+ query={'site': 'snagfilms'})['authorizationToken']
+ video_assets = self._download_json(
+ 'https://prod-api.viewlift.com/entitlement/video/status',
+ film_id, headers={
+ 'Authorization': token,
+ 'Referer': url,
+ }, query={
+ 'id': film_id
+ })['video']['streamingInfo']['videoAssets']
formats = []
mpeg_video_assets = video_assets.get('mpeg') or []
if category.get('title')]
break
else:
- title = self._search_regex(
- r'itemprop="title">([^<]+)<', webpage, 'title')
+ title = self._html_search_regex(
+ (r'itemprop="title">([^<]+)<',
+ r'(?s)itemprop="title">(.+?)<div'), webpage, 'title')
description = self._html_search_regex(
r'(?s)<div itemprop="description" class="film-synopsis-inner ">(.+?)</div>',
webpage, 'description', default=None) or self._og_search_description(webpage)