_ Git - youtube-dl/blob - 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 (
   8     js_to_json,
   9     smuggle_url,
  10 )
  11
  12
  13 class CBCIE(InfoExtractor):
  14     _VALID_URL = r'https?://(?:www\.)?cbc\.ca/(?!player/)(?:[^/]+/)+(?P<id>[^/?#]+)'
  15     _TESTS = [{
  16         # with mediaId
  17         'url': 'http://www.cbc.ca/22minutes/videos/clips-season-23/don-cherry-play-offs',
  18         'md5': '97e24d09672fc4cf56256d6faa6c25bc',
  19         'info_dict': {
  20             'id': '2682904050',
  21             'ext': 'mp4',
  22             'title': 'Don Cherry – All-Stars',
  23             'description': 'Don Cherry has a bee in his bonnet about AHL player John Scott because that guy’s got heart.',
  24             'timestamp': 1454463000,
  25             'upload_date': '20160203',
  26             'uploader': 'CBCC-NEW',
  27         },
  28     }, {
  29         # with clipId
  30         'url': 'http://www.cbc.ca/archives/entry/1978-robin-williams-freestyles-on-90-minutes-live',
  31         'info_dict': {
  32             'id': '2487345465',
  33             'ext': 'mp4',
  34             'title': 'Robin Williams freestyles on 90 Minutes Live',
  35             'description': 'Wacky American comedian Robin Williams shows off his infamous "freestyle" comedic talents while being interviewed on CBC\'s 90 Minutes Live.',
  36             'upload_date': '19780210',
  37             'uploader': 'CBCC-NEW',
  38             'timestamp': 255977160,
  39         },
  40     }, {
  41         # multiple iframes
  42         'url': 'http://www.cbc.ca/natureofthings/blog/birds-eye-view-from-vancouvers-burrard-street-bridge-how-we-got-the-shot',
  43         'playlist': [{
  44             'md5': '377572d0b49c4ce0c9ad77470e0b96b4',
  45             'info_dict': {
  46                 'id': '2680832926',
  47                 'ext': 'mp4',
  48                 'title': 'An Eagle\'s-Eye View Off Burrard Bridge',
  49                 'description': 'Hercules the eagle flies from Vancouver\'s Burrard Bridge down to a nearby park with a mini-camera strapped to his back.',
  50                 'upload_date': '20160201',
  51                 'timestamp': 1454342820,
  52                 'uploader': 'CBCC-NEW',
  53             },
  54         }, {
  55             'md5': '415a0e3f586113894174dfb31aa5bb1a',
  56             'info_dict': {
  57                 'id': '2658915080',
  58                 'ext': 'mp4',
  59                 'title': 'Fly like an eagle!',
  60                 'description': 'Eagle equipped with a mini camera flies from the world\'s tallest tower',
  61                 'upload_date': '20150315',
  62                 'timestamp': 1426443984,
  63                 'uploader': 'CBCC-NEW',
  64             },
  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     _TESTS = [{
  95         'url': 'http://www.cbc.ca/player/play/2683190193',
  96         'info_dict': {
  97             'id': '2683190193',
  98             'ext': 'mp4',
  99             'title': 'Gerry Runs a Sweat Shop',
 100             'description': 'md5:b457e1c01e8ff408d9d801c1c2cd29b0',
 101             'timestamp': 1455071400,
 102             'upload_date': '20160210',
 103             'uploader': 'CBCC-NEW',
 104         },
 105     }, {
 106         # Redirected from http://www.cbc.ca/player/AudioMobile/All%20in%20a%20Weekend%20Montreal/ID/2657632011/
 107         'url': 'http://www.cbc.ca/player/play/2657631896',
 108         'md5': 'e5e708c34ae6fca156aafe17c43e8b75',
 109         'info_dict': {
 110             'id': '2657631896',
 111             'ext': 'mp3',
 112             'title': 'CBC Montreal is organizing its first ever community hackathon!',
 113             'description': 'The modern technology we tend to depend on so heavily, is never without it\'s share of hiccups and headaches. Next weekend - CBC Montreal will be getting members of the public for its first Hackathon.',
 114             'timestamp': 1425704400,
 115             'upload_date': '20150307',
 116             'uploader': 'CBCC-NEW',
 117         },
 118     }]
 119
 120     def _real_extract(self, url):
 121         video_id = self._match_id(url)
 122         return {
 123             '_type': 'url_transparent',
 124             'ie_key': 'ThePlatform',
 125             'url': smuggle_url(
 126                 'http://link.theplatform.com/s/ExhSPC/media/guid/2655402169/%s?mbr=true' % video_id, {
 127                     'force_smil_url': True
 128                 }),
 129             'id': video_id,
 130         }