Raphaël G. Git Repositories - youtubedl/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         'md5': '0274a90b51a9b4971fe005c63f592f12',
  32         'info_dict': {
  33             'id': '2487345465',
  34             'ext': 'mp4',
  35             'title': 'Robin Williams freestyles on 90 Minutes Live',
  36             'description': 'Wacky American comedian Robin Williams shows off his infamous "freestyle" comedic talents while being interviewed on CBC\'s 90 Minutes Live.',
  37             'upload_date': '19780210',
  38             'uploader': 'CBCC-NEW',
  39             'timestamp': 255977160,
  40         },
  41     }, {
  42         # multiple iframes
  43         'url': 'http://www.cbc.ca/natureofthings/blog/birds-eye-view-from-vancouvers-burrard-street-bridge-how-we-got-the-shot',
  44         'playlist': [{
  45             'md5': '377572d0b49c4ce0c9ad77470e0b96b4',
  46             'info_dict': {
  47                 'id': '2680832926',
  48                 'ext': 'mp4',
  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': '20160201',
  52                 'timestamp': 1454342820,
  53                 'uploader': 'CBCC-NEW',
  54             },
  55         }, {
  56             'md5': '415a0e3f586113894174dfb31aa5bb1a',
  57             'info_dict': {
  58                 'id': '2658915080',
  59                 'ext': 'mp4',
  60                 'title': 'Fly like an eagle!',
  61                 'description': 'Eagle equipped with a mini camera flies from the world\'s tallest tower',
  62                 'upload_date': '20150315',
  63                 'timestamp': 1426443984,
  64                 'uploader': 'CBCC-NEW',
  65             },
  66         }],
  67     }]
  68
  69     @classmethod
  70     def suitable(cls, url):
  71         return False if CBCPlayerIE.suitable(url) else super(CBCIE, cls).suitable(url)
  72
  73     def _real_extract(self, url):
  74         display_id = self._match_id(url)
  75         webpage = self._download_webpage(url, display_id)
  76         player_init = self._search_regex(
  77             r'CBC\.APP\.Caffeine\.initInstance\(({.+?})\);', webpage, 'player init',
  78             default=None)
  79         if player_init:
  80             player_info = self._parse_json(player_init, display_id, js_to_json)
  81             media_id = player_info.get('mediaId')
  82             if not media_id:
  83                 clip_id = player_info['clipId']
  84                 media_id = self._download_json(
  85                     'http://feed.theplatform.com/f/h9dtGB/punlNGjMlc1F?fields=id&byContent=byReleases%3DbyId%253D' + clip_id,
  86                     clip_id)['entries'][0]['id'].split('/')[-1]
  87             return self.url_result('cbcplayer:%s' % media_id, 'CBCPlayer', media_id)
  88         else:
  89             entries = [self.url_result('cbcplayer:%s' % media_id, 'CBCPlayer', media_id) for media_id in re.findall(r'<iframe[^>]+src="[^"]+?mediaId=(\d+)"', webpage)]
  90             return self.playlist_result(entries)
  91
  92
  93 class CBCPlayerIE(InfoExtractor):
  94     _VALID_URL = r'(?:cbcplayer:|https?://(?:www\.)?cbc\.ca/(?:player/play/|i/caffeine/syndicate/\?mediaId=))(?P<id>\d+)'
  95     _TESTS = [{
  96         'url': 'http://www.cbc.ca/player/play/2683190193',
  97         'md5': '64d25f841ddf4ddb28a235338af32e2c',
  98         'info_dict': {
  99             'id': '2683190193',
 100             'ext': 'mp4',
 101             'title': 'Gerry Runs a Sweat Shop',
 102             'description': 'md5:b457e1c01e8ff408d9d801c1c2cd29b0',
 103             'timestamp': 1455071400,
 104             'upload_date': '20160210',
 105             'uploader': 'CBCC-NEW',
 106         },
 107     }, {
 108         # Redirected from http://www.cbc.ca/player/AudioMobile/All%20in%20a%20Weekend%20Montreal/ID/2657632011/
 109         'url': 'http://www.cbc.ca/player/play/2657631896',
 110         'md5': 'e5e708c34ae6fca156aafe17c43e8b75',
 111         'info_dict': {
 112             'id': '2657631896',
 113             'ext': 'mp3',
 114             'title': 'CBC Montreal is organizing its first ever community hackathon!',
 115             '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.',
 116             'timestamp': 1425704400,
 117             'upload_date': '20150307',
 118             'uploader': 'CBCC-NEW',
 119         },
 120     }, {
 121         # available only when we add `formats=MPEG4,FLV,MP3` to theplatform url
 122         'url': 'http://www.cbc.ca/player/play/2164402062',
 123         'md5': '17a61eb813539abea40618d6323a7f82',
 124         'info_dict': {
 125             'id': '2164402062',
 126             'ext': 'flv',
 127             'title': 'Cancer survivor four times over',
 128             'description': 'Tim Mayer has beaten three different forms of cancer four times in five years.',
 129             'timestamp': 1320410746,
 130             'upload_date': '20111104',
 131             'uploader': 'CBCC-NEW',
 132         },
 133     }]
 134
 135     def _real_extract(self, url):
 136         video_id = self._match_id(url)
 137         return {
 138             '_type': 'url_transparent',
 139             'ie_key': 'ThePlatform',
 140             'url': smuggle_url(
 141                 'http://link.theplatform.com/s/ExhSPC/media/guid/2655402169/%s?mbr=true&formats=MPEG4,FLV,MP3' % video_id, {
 142                     'force_smil_url': True
 143                 }),
 144             'id': video_id,
 145         }