youtube_dl/extractor/cbc.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..utils import js_to_json
   8
   9
  10 class CBCIE(InfoExtractor):
  11     _VALID_URL = r'https?://(?:www\.)?cbc\.ca/(?:[^/]+/)+(?P<id>[^/?#]+)'
  12     _TESTS = [{
  13         # with mediaId
  14         'url': 'http://www.cbc.ca/22minutes/videos/clips-season-23/don-cherry-play-offs',
  15         'info_dict': {
  16             'id': '2682904050',
  17             'ext': 'flv',
  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',
  22         },
  23         'params': {
  24             # rtmp download
  25             'skip_download': True,
  26         },
  27     }, {
  28         # with clipId
  29         'url': 'http://www.cbc.ca/archives/entry/1978-robin-williams-freestyles-on-90-minutes-live',
  30         'info_dict': {
  31             'id': '2487345465',
  32             'ext': 'flv',
  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',
  37         },
  38         'params': {
  39             # rtmp download
  40             'skip_download': True,
  41         },
  42     }, {
  43         # multiple iframes
  44         'url': 'http://www.cbc.ca/natureofthings/blog/birds-eye-view-from-vancouvers-burrard-street-bridge-how-we-got-the-shot',
  45         'playlist': [{
  46             'info_dict': {
  47                 'id': '2680832926',
  48                 'ext': 'flv',
  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',
  52             },
  53         }, {
  54             'info_dict': {
  55                 'id': '2658915080',
  56                 'ext': 'flv',
  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',
  60             },
  61         }],
  62         'params': {
  63             # rtmp download
  64             'skip_download': True,
  65         },
  66     }]
  67
  68     @classmethod
  69     def suitable(cls, url):
  70         return False if CBCPlayerIE.suitable(url) else super(CBCIE, cls).suitable(url)
  71
  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',
  77             default=None)
  78         if player_init:
  79             player_info = self._parse_json(player_init, display_id, js_to_json)
  80             media_id = player_info.get('mediaId')
  81             if not media_id:
  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)
  87         else:
  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)
  90
  91
  92 class CBCPlayerIE(InfoExtractor):
  93     _VALID_URL = r'(?:cbcplayer:|https?://(?:www\.)?cbc\.ca/(?:player/play/|i/caffeine/syndicate/\?mediaId=))(?P<id>\d+)'
  94     _TEST = {
  95         'url': 'http://www.cbc.ca/player/play/2683190193',
  96         'info_dict': {
  97             'id': '2683190193',
  98             'ext': 'flv',
  99             'title': 'Gerry Runs a Sweat Shop',
 100             'description': 'md5:b457e1c01e8ff408d9d801c1c2cd29b0',
 101             'timestamp': 1455067800,
 102             'upload_date': '20160210',
 103         },
 104         'params': {
 105             # rtmp download
 106             'skip_download': True,
 107         },
 108     }
 109
 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)