[syfy] fix extraction(closes #9087)(closes #3820)(closes #2388)

This commit is contained in:
Remita Amine 2016-07-14 23:59:12 +01:00
parent ad27649be3
commit 224db034ab

View file

@ -1,46 +1,56 @@
from __future__ import unicode_literals from __future__ import unicode_literals
import re from .theplatform import ThePlatformIE
from ..utils import (
from .common import InfoExtractor update_url_query,
smuggle_url,
)
class SyfyIE(InfoExtractor): class SyfyIE(ThePlatformIE):
_VALID_URL = r'https?://www\.syfy\.com/(?:videos/.+?vid:(?P<id>[0-9]+)|(?!videos)(?P<video_name>[^/]+)(?:$|[?#]))' _VALID_URL = r'https?://www\.syfy\.com/(?:[^/]+/)?videos/(?P<id>[^/?#]+)'
_TESTS = [{ _TESTS = [{
'url': 'http://www.syfy.com/videos/Robot%20Combat%20League/Behind%20the%20Scenes/vid:2631458', 'url': 'http://www.syfy.com/theinternetruinedmylife/videos/the-internet-ruined-my-life-season-1-trailer',
'info_dict': { 'info_dict': {
'id': 'NmqMrGnXvmO1', 'id': '2968097',
'ext': 'flv', 'ext': 'mp4',
'title': 'George Lucas has Advice for his Daughter', 'title': 'The Internet Ruined My Life: Season 1 Trailer',
'description': 'Listen to what insights George Lucas give his daughter Amanda.', 'description': 'One tweet, one post, one click, can destroy everything.',
'uploader': 'NBCU-MPAT',
'upload_date': '20170113',
'timestamp': 1484345640,
},
'params': {
# m3u8 download
'skip_download': True,
}, },
'add_ie': ['ThePlatform'], 'add_ie': ['ThePlatform'],
}, {
'url': 'http://www.syfy.com/wilwheaton',
'md5': '94dfa54ee3ccb63295b276da08c415f6',
'info_dict': {
'id': '4yoffOOXC767',
'ext': 'flv',
'title': 'The Wil Wheaton Project - Premiering May 27th at 10/9c.',
'description': 'The Wil Wheaton Project premieres May 27th at 10/9c. Don\'t miss it.',
},
'add_ie': ['ThePlatform'],
'skip': 'Blocked outside the US',
}] }]
def _real_extract(self, url): def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url) display_id = self._match_id(url)
video_name = mobj.group('video_name') webpage = self._download_webpage(url, display_id)
if video_name: syfy_mpx = list(self._parse_json(self._search_regex(
generic_webpage = self._download_webpage(url, video_name) r'jQuery\.extend\([^,]+,\s*({.+})\);', webpage, 'drupal settings'),
video_id = self._search_regex( display_id)['syfy']['syfy_mpx'].values())[0]
r'<iframe.*?class="video_iframe_page"\s+src="/_utils/video/thP_video_controller.php.*?_vid([0-9]+)">', video_id = syfy_mpx['mpxGUID']
generic_webpage, 'video ID') title = syfy_mpx['episodeTitle']
url = 'http://www.syfy.com/videos/%s/%s/vid:%s' % ( query = {
video_name, video_name, video_id) 'mbr': 'true',
else: 'manifest': 'm3u',
video_id = mobj.group('id') }
webpage = self._download_webpage(url, video_id) if syfy_mpx.get('entitlement') == 'auth':
return self.url_result(self._og_search_video_url(webpage)) resource = '<rss version="2.0" xmlns:media="http://search.yahoo.com/mrss/"><channel><title>syfy</title><item><title><![CDATA[%s]]></title><guid>%s</guid><media:rating scheme="urn:v-chip">%s</media:rating></item></channel></rss>' % (title, video_id, syfy_mpx.get('mpxRating', 'TV-14'))
query['auth'] = self._extract_mvpd_auth(
url, video_id, 'syfy', resource)
return {
'_type': 'url_transparent',
'ie_key': 'ThePlatform',
'url': smuggle_url(update_url_query(
self._proto_relative_url(syfy_mpx['releaseURL']), query),
{'force_smil_url': True}),
'title': title,
'id': video_id,
'display_id': display_id,
}