2 from __future__ import unicode_literals
6 from .common import InfoExtractor
7 from ..utils import js_to_json
10 class CBCIE(InfoExtractor):
11 _VALID_URL = r'https?://(?:www\.)?cbc\.ca/(?:[^/]+/)+(?P<id>[^/?#]+)'
14 'url': 'http://www.cbc.ca/22minutes/videos/clips-season-23/don-cherry-play-offs',
18 'title': 'Don Cherry – All-Stars',
19 'description': 'Don Cherry has a bee in his bonnet about AHL player John Scott because that guy’s got heart.',
20 'timestamp': 1454475540,
21 'upload_date': '20160203',
25 'skip_download': True,
29 'url': 'http://www.cbc.ca/archives/entry/1978-robin-williams-freestyles-on-90-minutes-live',
33 'title': 'Robin Williams freestyles on 90 Minutes Live',
34 'description': 'Wacky American comedian Robin Williams shows off his infamous "freestyle" comedic talents while being interviewed on CBC\'s 90 Minutes Live.',
35 'upload_date': '19700101',
36 'uploader': 'CBCC-NEW',
40 'skip_download': True,
44 'url': 'http://www.cbc.ca/natureofthings/blog/birds-eye-view-from-vancouvers-burrard-street-bridge-how-we-got-the-shot',
49 'title': 'An Eagle\'s-Eye View Off Burrard Bridge',
50 'description': 'Hercules the eagle flies from Vancouver\'s Burrard Bridge down to a nearby park with a mini-camera strapped to his back.',
51 'upload_date': '19700101',
57 'title': 'Fly like an eagle!',
58 'description': 'Eagle equipped with a mini camera flies from the world\'s tallest tower',
59 'upload_date': '19700101',
64 'skip_download': True,
69 def suitable(cls, url):
70 return False if CBCPlayerIE.suitable(url) else super(CBCIE, cls).suitable(url)
72 def _real_extract(self, url):
73 display_id = self._match_id(url)
74 webpage = self._download_webpage(url, display_id)
75 player_init = self._search_regex(
76 r'CBC\.APP\.Caffeine\.initInstance\(({.+?})\);', webpage, 'player init',
79 player_info = self._parse_json(player_init, display_id, js_to_json)
80 media_id = player_info.get('mediaId')
82 clip_id = player_info['clipId']
83 media_id = self._download_json(
84 'http://feed.theplatform.com/f/h9dtGB/punlNGjMlc1F?fields=id&byContent=byReleases%3DbyId%253D' + clip_id,
85 clip_id)['entries'][0]['id'].split('/')[-1]
86 return self.url_result('cbcplayer:%s' % media_id, 'CBCPlayer', media_id)
88 entries = [self.url_result('cbcplayer:%s' % media_id, 'CBCPlayer', media_id) for media_id in re.findall(r'<iframe[^>]+src="[^"]+?mediaId=(\d+)"', webpage)]
89 return self.playlist_result(entries)
92 class CBCPlayerIE(InfoExtractor):
93 _VALID_URL = r'(?:cbcplayer:|https?://(?:www\.)?cbc\.ca/(?:player/play/|i/caffeine/syndicate/\?mediaId=))(?P<id>\d+)'
95 'url': 'http://www.cbc.ca/player/play/2683190193',
99 'title': 'Gerry Runs a Sweat Shop',
100 'description': 'md5:b457e1c01e8ff408d9d801c1c2cd29b0',
101 'timestamp': 1455067800,
102 'upload_date': '20160210',
106 'skip_download': True,
110 def _real_extract(self, url):
111 video_id = self._match_id(url)
112 return self.url_result(
113 'http://feed.theplatform.com/f/ExhSPC/vms_5akSXx4Ng_Zn?byGuid=%s' % video_id,
114 'ThePlatformFeed', video_id)