youtube_dlc/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     @staticmethod
  16     def _result(url, ie_key):
  17         return {
  18             '_type': 'url_transparent',
  19             'url': smuggle_url(url, {'geo_countries': ['CA']}),
  20             'ie_key': ie_key,
  21         }
  22
  23     @staticmethod
  24     def _limelight_result(media_id):
  25         return TeleQuebecBaseIE._result(
  26             'limelight:media:' + media_id, 'LimelightMedia')
  27
  28     @staticmethod
  29     def _brightcove_result(brightcove_id):
  30         return TeleQuebecBaseIE._result(
  31             'http://players.brightcove.net/6150020952001/default_default/index.html?videoId=%s'
  32             % brightcove_id, 'BrightcoveNew')
  33
  34
  35 class TeleQuebecIE(TeleQuebecBaseIE):
  36     _VALID_URL = r'''(?x)
  37                     https?://
  38                         (?:
  39                             zonevideo\.telequebec\.tv/media|
  40                             coucou\.telequebec\.tv/videos
  41                         )/(?P<id>\d+)
  42                     '''
  43     _TESTS = [{
  44         # available till 01.01.2023
  45         'url': 'http://zonevideo.telequebec.tv/media/37578/un-petit-choc-et-puis-repart/un-chef-a-la-cabane',
  46         'info_dict': {
  47             'id': '577116881b4b439084e6b1cf4ef8b1b3',
  48             'ext': 'mp4',
  49             'title': 'Un petit choc et puis repart!',
  50             'description': 'md5:067bc84bd6afecad85e69d1000730907',
  51         },
  52         'params': {
  53             'skip_download': True,
  54         },
  55     }, {
  56         'url': 'https://zonevideo.telequebec.tv/media/55267/le-soleil/passe-partout',
  57         'info_dict': {
  58             'id': '6167180337001',
  59             'ext': 'mp4',
  60             'title': 'Le soleil',
  61             'description': 'md5:64289c922a8de2abbe99c354daffde02',
  62             'uploader_id': '6150020952001',
  63             'upload_date': '20200625',
  64             'timestamp': 1593090307,
  65         },
  66         'params': {
  67             'format': 'bestvideo',
  68             'skip_download': True,
  69         },
  70         'add_ie': ['BrightcoveNew'],
  71     }, {
  72         # no description
  73         'url': 'http://zonevideo.telequebec.tv/media/30261',
  74         'only_matching': True,
  75     }, {
  76         'url': 'https://coucou.telequebec.tv/videos/41788/idee-de-genie/l-heure-du-bain',
  77         'only_matching': True,
  78     }]
  79
  80     def _real_extract(self, url):
  81         media_id = self._match_id(url)
  82
  83         media_data = self._download_json(
  84             'https://mnmedias.api.telequebec.tv/api/v2/media/' + media_id,
  85             media_id)['media']
  86
  87         source_id = media_data['streamInfo']['sourceId']
  88         source = (try_get(
  89             media_data, lambda x: x['streamInfo']['source'],
  90             compat_str) or 'limelight').lower()
  91         if source == 'brightcove':
  92             info = self._brightcove_result(source_id)
  93         else:
  94             info = self._limelight_result(source_id)
  95         info.update({
  96             'title': media_data.get('title'),
  97             'description': try_get(
  98                 media_data, lambda x: x['descriptions'][0]['text'], compat_str),
  99             'duration': int_or_none(
 100                 media_data.get('durationInMilliseconds'), 1000),
 101         })
 102         return info
 103
 104
 105 class TeleQuebecSquatIE(InfoExtractor):
 106     _VALID_URL = r'https://squat\.telequebec\.tv/videos/(?P<id>\d+)'
 107     _TESTS = [{
 108         'url': 'https://squat.telequebec.tv/videos/9314',
 109         'info_dict': {
 110             'id': 'd59ae78112d542e793d83cc9d3a5b530',
 111             'ext': 'mp4',
 112             'title': 'Poupeflekta',
 113             'description': 'md5:2f0718f8d2f8fece1646ee25fb7bce75',
 114             'duration': 1351,
 115             'timestamp': 1569057600,
 116             'upload_date': '20190921',
 117             'series': 'Miraculous : Les Aventures de Ladybug et Chat Noir',
 118             'season': 'Saison 3',
 119             'season_number': 3,
 120             'episode_number': 57,
 121         },
 122         'params': {
 123             'skip_download': True,
 124         },
 125     }]
 126
 127     def _real_extract(self, url):
 128         video_id = self._match_id(url)
 129
 130         video = self._download_json(
 131             'https://squat.api.telequebec.tv/v1/videos/%s' % video_id,
 132             video_id)
 133
 134         media_id = video['sourceId']
 135
 136         return {
 137             '_type': 'url_transparent',
 138             'url': 'http://zonevideo.telequebec.tv/media/%s' % media_id,
 139             'ie_key': TeleQuebecIE.ie_key(),
 140             'id': media_id,
 141             'title': video.get('titre'),
 142             'description': video.get('description'),
 143             'timestamp': unified_timestamp(video.get('datePublication')),
 144             'series': video.get('container'),
 145             'season': video.get('saison'),
 146             'season_number': int_or_none(video.get('noSaison')),
 147             'episode_number': int_or_none(video.get('episode')),
 148         }
 149
 150
 151 class TeleQuebecEmissionIE(TeleQuebecBaseIE):
 152     _VALID_URL = r'''(?x)
 153                     https?://
 154                         (?:
 155                             [^/]+\.telequebec\.tv/emissions/|
 156                             (?:www\.)?telequebec\.tv/
 157                         )
 158                         (?P<id>[^?#&]+)
 159                     '''
 160     _TESTS = [{
 161         'url': 'http://lindicemcsween.telequebec.tv/emissions/100430013/des-soins-esthetiques-a-377-d-interets-annuels-ca-vous-tente',
 162         'info_dict': {
 163             'id': '66648a6aef914fe3badda25e81a4d50a',
 164             'ext': 'mp4',
 165             'title': "Des soins esthétiques à 377 % d'intérêts annuels, ça vous tente?",
 166             'description': 'md5:369e0d55d0083f1fc9b71ffb640ea014',
 167             'upload_date': '20171024',
 168             'timestamp': 1508862118,
 169         },
 170         'params': {
 171             'skip_download': True,
 172         },
 173     }, {
 174         'url': 'http://bancpublic.telequebec.tv/emissions/emission-49/31986/jeunes-meres-sous-pression',
 175         'only_matching': True,
 176     }, {
 177         'url': 'http://www.telequebec.tv/masha-et-michka/epi059masha-et-michka-3-053-078',
 178         'only_matching': True,
 179     }, {
 180         'url': 'http://www.telequebec.tv/documentaire/bebes-sur-mesure/',
 181         'only_matching': True,
 182     }]
 183
 184     def _real_extract(self, url):
 185         display_id = self._match_id(url)
 186
 187         webpage = self._download_webpage(url, display_id)
 188
 189         media_id = self._search_regex(
 190             r'mediaUID\s*:\s*["\'][Ll]imelight_(?P<id>[a-z0-9]{32})', webpage,
 191             'limelight id')
 192
 193         info = self._limelight_result(media_id)
 194         info.update({
 195             'title': self._og_search_title(webpage, default=None),
 196             'description': self._og_search_description(webpage, default=None),
 197         })
 198         return info
 199
 200
 201 class TeleQuebecLiveIE(InfoExtractor):
 202     _VALID_URL = r'https?://zonevideo\.telequebec\.tv/(?P<id>endirect)'
 203     _TEST = {
 204         'url': 'http://zonevideo.telequebec.tv/endirect/',
 205         'info_dict': {
 206             'id': 'endirect',
 207             'ext': 'mp4',
 208             'title': 're:^Télé-Québec - En direct [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
 209             'is_live': True,
 210         },
 211         'params': {
 212             'skip_download': True,
 213         },
 214     }
 215
 216     def _real_extract(self, url):
 217         video_id = self._match_id(url)
 218
 219         m3u8_url = None
 220         webpage = self._download_webpage(
 221             'https://player.telequebec.tv/Tq_VideoPlayer.js', video_id,
 222             fatal=False)
 223         if webpage:
 224             m3u8_url = self._search_regex(
 225                 r'm3U8Url\s*:\s*(["\'])(?P<url>(?:(?!\1).)+)\1', webpage,
 226                 'm3u8 url', default=None, group='url')
 227         if not m3u8_url:
 228             m3u8_url = 'https://teleqmmd.mmdlive.lldns.net/teleqmmd/f386e3b206814e1f8c8c1c71c0f8e748/manifest.m3u8'
 229         formats = self._extract_m3u8_formats(
 230             m3u8_url, video_id, 'mp4', m3u8_id='hls')
 231         self._sort_formats(formats)
 232
 233         return {
 234             'id': video_id,
 235             'title': self._live_title('Télé-Québec - En direct'),
 236             'is_live': True,
 237             'formats': formats,
 238         }