_ Git - youtube-dl/blob - youtube_dl/extractor/telequebec.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..compat import compat_str
   6 from ..utils import (
   7     int_or_none,
   8     smuggle_url,
   9     try_get,
  10     unified_timestamp,
  11 )
  12
  13
  14 class TeleQuebecBaseIE(InfoExtractor):
  15     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/6150020952001/default_default/index.html?videoId=%s'
  16
  17     @staticmethod
  18     def _limelight_result(media_id):
  19         return {
  20             '_type': 'url_transparent',
  21             'url': smuggle_url(
  22                 'limelight:media:' + media_id, {'geo_countries': ['CA']}),
  23             'ie_key': 'LimelightMedia',
  24         }
  25
  26     def _brightcove_result(self, brightcove_id):
  27         return self.url_result(
  28             smuggle_url(
  29                 self.BRIGHTCOVE_URL_TEMPLATE % brightcove_id,
  30                 {'geo_countries': ['CA']}),
  31             'BrightcoveNew', brightcove_id)
  32
  33
  34 class TeleQuebecIE(TeleQuebecBaseIE):
  35     _VALID_URL = r'''(?x)
  36                     https?://
  37                         (?:
  38                             zonevideo\.telequebec\.tv/media|
  39                             coucou\.telequebec\.tv/videos
  40                         )/(?P<id>\d+)
  41                     '''
  42     _TESTS = [{
  43         # available till 01.01.2023
  44         'url': 'http://zonevideo.telequebec.tv/media/37578/un-petit-choc-et-puis-repart/un-chef-a-la-cabane',
  45         'info_dict': {
  46             'id': '577116881b4b439084e6b1cf4ef8b1b3',
  47             'ext': 'mp4',
  48             'title': 'Un petit choc et puis repart!',
  49             'description': 'md5:067bc84bd6afecad85e69d1000730907',
  50         },
  51         'params': {
  52             'skip_download': True,
  53         },
  54     }, {
  55         # no description
  56         'url': 'http://zonevideo.telequebec.tv/media/30261',
  57         'only_matching': True,
  58     }, {
  59         'url': 'https://coucou.telequebec.tv/videos/41788/idee-de-genie/l-heure-du-bain',
  60         'only_matching': True,
  61     }]
  62
  63     def _real_extract(self, url):
  64         media_id = self._match_id(url)
  65
  66         media_data = self._download_json(
  67             'https://mnmedias.api.telequebec.tv/api/v2/media/' + media_id,
  68             media_id)['media']
  69
  70         if media_data['streamInfo']['source'] == 'Brightcove':
  71             info = self._brightcove_result(media_data['streamInfo']['sourceId'])
  72         elif media_data['streamInfo']['source'] == 'Limelight':
  73             info = self._limelight_result(media_data['streamInfo']['sourceId'])
  74         info.update({
  75             'title': media_data.get('title'),
  76             'description': try_get(
  77                 media_data, lambda x: x['descriptions'][0]['text'], compat_str),
  78             'duration': int_or_none(
  79                 media_data.get('durationInMilliseconds'), 1000),
  80         })
  81         return info
  82
  83
  84 class TeleQuebecSquatIE(InfoExtractor):
  85     _VALID_URL = r'https://squat\.telequebec\.tv/videos/(?P<id>\d+)'
  86     _TESTS = [{
  87         'url': 'https://squat.telequebec.tv/videos/9314',
  88         'info_dict': {
  89             'id': 'd59ae78112d542e793d83cc9d3a5b530',
  90             'ext': 'mp4',
  91             'title': 'Poupeflekta',
  92             'description': 'md5:2f0718f8d2f8fece1646ee25fb7bce75',
  93             'duration': 1351,
  94             'timestamp': 1569057600,
  95             'upload_date': '20190921',
  96             'series': 'Miraculous : Les Aventures de Ladybug et Chat Noir',
  97             'season': 'Saison 3',
  98             'season_number': 3,
  99             'episode_number': 57,
 100         },
 101         'params': {
 102             'skip_download': True,
 103         },
 104     }]
 105
 106     def _real_extract(self, url):
 107         video_id = self._match_id(url)
 108
 109         video = self._download_json(
 110             'https://squat.api.telequebec.tv/v1/videos/%s' % video_id,
 111             video_id)
 112
 113         media_id = video['sourceId']
 114
 115         return {
 116             '_type': 'url_transparent',
 117             'url': 'http://zonevideo.telequebec.tv/media/%s' % media_id,
 118             'ie_key': TeleQuebecIE.ie_key(),
 119             'id': media_id,
 120             'title': video.get('titre'),
 121             'description': video.get('description'),
 122             'timestamp': unified_timestamp(video.get('datePublication')),
 123             'series': video.get('container'),
 124             'season': video.get('saison'),
 125             'season_number': int_or_none(video.get('noSaison')),
 126             'episode_number': int_or_none(video.get('episode')),
 127         }
 128
 129
 130 class TeleQuebecEmissionIE(TeleQuebecBaseIE):
 131     _VALID_URL = r'''(?x)
 132                     https?://
 133                         (?:
 134                             [^/]+\.telequebec\.tv/emissions/|
 135                             (?:www\.)?telequebec\.tv/
 136                         )
 137                         (?P<id>[^?#&]+)
 138                     '''
 139     _TESTS = [{
 140         'url': 'http://lindicemcsween.telequebec.tv/emissions/100430013/des-soins-esthetiques-a-377-d-interets-annuels-ca-vous-tente',
 141         'info_dict': {
 142             'id': '66648a6aef914fe3badda25e81a4d50a',
 143             'ext': 'mp4',
 144             'title': "Des soins esthétiques à 377 % d'intérêts annuels, ça vous tente?",
 145             'description': 'md5:369e0d55d0083f1fc9b71ffb640ea014',
 146             'upload_date': '20171024',
 147             'timestamp': 1508862118,
 148         },
 149         'params': {
 150             'skip_download': True,
 151         },
 152     }, {
 153         'url': 'http://bancpublic.telequebec.tv/emissions/emission-49/31986/jeunes-meres-sous-pression',
 154         'only_matching': True,
 155     }, {
 156         'url': 'http://www.telequebec.tv/masha-et-michka/epi059masha-et-michka-3-053-078',
 157         'only_matching': True,
 158     }, {
 159         'url': 'http://www.telequebec.tv/documentaire/bebes-sur-mesure/',
 160         'only_matching': True,
 161     }]
 162
 163     def _real_extract(self, url):
 164         display_id = self._match_id(url)
 165
 166         webpage = self._download_webpage(url, display_id)
 167
 168         media_id = self._search_regex(
 169             r'mediaUID\s*:\s*["\'][Ll]imelight_(?P<id>[a-z0-9]{32})', webpage,
 170             'limelight id')
 171
 172         info = self._limelight_result(media_id)
 173         info.update({
 174             'title': self._og_search_title(webpage, default=None),
 175             'description': self._og_search_description(webpage, default=None),
 176         })
 177         return info
 178
 179
 180 class TeleQuebecLiveIE(InfoExtractor):
 181     _VALID_URL = r'https?://zonevideo\.telequebec\.tv/(?P<id>endirect)'
 182     _TEST = {
 183         'url': 'http://zonevideo.telequebec.tv/endirect/',
 184         'info_dict': {
 185             'id': 'endirect',
 186             'ext': 'mp4',
 187             'title': 're:^Télé-Québec - En direct [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
 188             'is_live': True,
 189         },
 190         'params': {
 191             'skip_download': True,
 192         },
 193     }
 194
 195     def _real_extract(self, url):
 196         video_id = self._match_id(url)
 197
 198         m3u8_url = None
 199         webpage = self._download_webpage(
 200             'https://player.telequebec.tv/Tq_VideoPlayer.js', video_id,
 201             fatal=False)
 202         if webpage:
 203             m3u8_url = self._search_regex(
 204                 r'm3U8Url\s*:\s*(["\'])(?P<url>(?:(?!\1).)+)\1', webpage,
 205                 'm3u8 url', default=None, group='url')
 206         if not m3u8_url:
 207             m3u8_url = 'https://teleqmmd.mmdlive.lldns.net/teleqmmd/f386e3b206814e1f8c8c1c71c0f8e748/manifest.m3u8'
 208         formats = self._extract_m3u8_formats(
 209             m3u8_url, video_id, 'mp4', m3u8_id='hls')
 210         self._sort_formats(formats)
 211
 212         return {
 213             'id': video_id,
 214             'title': self._live_title('Télé-Québec - En direct'),
 215             'is_live': True,
 216             'formats': formats,
 217         }