Raphaël G. Git Repositories - youtubedl/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     @staticmethod
  16     def _limelight_result(media_id):
  17         return {
  18             '_type': 'url_transparent',
  19             'url': smuggle_url(
  20                 'limelight:media:' + media_id, {'geo_countries': ['CA']}),
  21             'ie_key': 'LimelightMedia',
  22         }
  23
  24
  25 class TeleQuebecIE(TeleQuebecBaseIE):
  26     _VALID_URL = r'''(?x)
  27                     https?://
  28                         (?:
  29                             zonevideo\.telequebec\.tv/media|
  30                             coucou\.telequebec\.tv/videos
  31                         )/(?P<id>\d+)
  32                     '''
  33     _TESTS = [{
  34         # available till 01.01.2023
  35         'url': 'http://zonevideo.telequebec.tv/media/37578/un-petit-choc-et-puis-repart/un-chef-a-la-cabane',
  36         'info_dict': {
  37             'id': '577116881b4b439084e6b1cf4ef8b1b3',
  38             'ext': 'mp4',
  39             'title': 'Un petit choc et puis repart!',
  40             'description': 'md5:b04a7e6b3f74e32d7b294cffe8658374',
  41             'upload_date': '20180222',
  42             'timestamp': 1519326631,
  43         },
  44         'params': {
  45             'skip_download': True,
  46         },
  47     }, {
  48         # no description
  49         'url': 'http://zonevideo.telequebec.tv/media/30261',
  50         'only_matching': True,
  51     }, {
  52         'url': 'https://coucou.telequebec.tv/videos/41788/idee-de-genie/l-heure-du-bain',
  53         'only_matching': True,
  54     }]
  55
  56     def _real_extract(self, url):
  57         media_id = self._match_id(url)
  58
  59         media_data = self._download_json(
  60             'https://mnmedias.api.telequebec.tv/api/v2/media/' + media_id,
  61             media_id)['media']
  62
  63         info = self._limelight_result(media_data['streamInfo']['sourceId'])
  64         info.update({
  65             'title': media_data.get('title'),
  66             'description': try_get(
  67                 media_data, lambda x: x['descriptions'][0]['text'], compat_str),
  68             'duration': int_or_none(
  69                 media_data.get('durationInMilliseconds'), 1000),
  70         })
  71         return info
  72
  73
  74 class TeleQuebecSquatIE(InfoExtractor):
  75     _VALID_URL = r'https://squat\.telequebec\.tv/videos/(?P<id>\d+)'
  76     _TESTS = [{
  77         'url': 'https://squat.telequebec.tv/videos/9314',
  78         'info_dict': {
  79             'id': 'd59ae78112d542e793d83cc9d3a5b530',
  80             'ext': 'mp4',
  81             'title': 'Poupeflekta',
  82             'description': 'md5:2f0718f8d2f8fece1646ee25fb7bce75',
  83             'duration': 1351,
  84             'timestamp': 1569057600,
  85             'upload_date': '20190921',
  86             'series': 'Miraculous : Les Aventures de Ladybug et Chat Noir',
  87             'season': 'Saison 3',
  88             'season_number': 3,
  89             'episode_number': 57,
  90         },
  91         'params': {
  92             'skip_download': True,
  93         },
  94     }]
  95
  96     def _real_extract(self, url):
  97         video_id = self._match_id(url)
  98
  99         video = self._download_json(
 100             'https://squat.api.telequebec.tv/v1/videos/%s' % video_id,
 101             video_id)
 102
 103         media_id = video['sourceId']
 104
 105         return {
 106             '_type': 'url_transparent',
 107             'url': 'http://zonevideo.telequebec.tv/media/%s' % media_id,
 108             'ie_key': TeleQuebecIE.ie_key(),
 109             'id': media_id,
 110             'title': video.get('titre'),
 111             'description': video.get('description'),
 112             'timestamp': unified_timestamp(video.get('datePublication')),
 113             'series': video.get('container'),
 114             'season': video.get('saison'),
 115             'season_number': int_or_none(video.get('noSaison')),
 116             'episode_number': int_or_none(video.get('episode')),
 117         }
 118
 119
 120 class TeleQuebecEmissionIE(TeleQuebecBaseIE):
 121     _VALID_URL = r'''(?x)
 122                     https?://
 123                         (?:
 124                             [^/]+\.telequebec\.tv/emissions/|
 125                             (?:www\.)?telequebec\.tv/
 126                         )
 127                         (?P<id>[^?#&]+)
 128                     '''
 129     _TESTS = [{
 130         'url': 'http://lindicemcsween.telequebec.tv/emissions/100430013/des-soins-esthetiques-a-377-d-interets-annuels-ca-vous-tente',
 131         'info_dict': {
 132             'id': '66648a6aef914fe3badda25e81a4d50a',
 133             'ext': 'mp4',
 134             'title': "Des soins esthétiques à 377 % d'intérêts annuels, ça vous tente?",
 135             'description': 'md5:369e0d55d0083f1fc9b71ffb640ea014',
 136             'upload_date': '20171024',
 137             'timestamp': 1508862118,
 138         },
 139         'params': {
 140             'skip_download': True,
 141         },
 142     }, {
 143         'url': 'http://bancpublic.telequebec.tv/emissions/emission-49/31986/jeunes-meres-sous-pression',
 144         'only_matching': True,
 145     }, {
 146         'url': 'http://www.telequebec.tv/masha-et-michka/epi059masha-et-michka-3-053-078',
 147         'only_matching': True,
 148     }, {
 149         'url': 'http://www.telequebec.tv/documentaire/bebes-sur-mesure/',
 150         'only_matching': True,
 151     }]
 152
 153     def _real_extract(self, url):
 154         display_id = self._match_id(url)
 155
 156         webpage = self._download_webpage(url, display_id)
 157
 158         media_id = self._search_regex(
 159             r'mediaUID\s*:\s*["\'][Ll]imelight_(?P<id>[a-z0-9]{32})', webpage,
 160             'limelight id')
 161
 162         info = self._limelight_result(media_id)
 163         info.update({
 164             'title': self._og_search_title(webpage, default=None),
 165             'description': self._og_search_description(webpage, default=None),
 166         })
 167         return info
 168
 169
 170 class TeleQuebecLiveIE(InfoExtractor):
 171     _VALID_URL = r'https?://zonevideo\.telequebec\.tv/(?P<id>endirect)'
 172     _TEST = {
 173         'url': 'http://zonevideo.telequebec.tv/endirect/',
 174         'info_dict': {
 175             'id': 'endirect',
 176             'ext': 'mp4',
 177             'title': 're:^Télé-Québec - En direct [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
 178             'is_live': True,
 179         },
 180         'params': {
 181             'skip_download': True,
 182         },
 183     }
 184
 185     def _real_extract(self, url):
 186         video_id = self._match_id(url)
 187
 188         m3u8_url = None
 189         webpage = self._download_webpage(
 190             'https://player.telequebec.tv/Tq_VideoPlayer.js', video_id,
 191             fatal=False)
 192         if webpage:
 193             m3u8_url = self._search_regex(
 194                 r'm3U8Url\s*:\s*(["\'])(?P<url>(?:(?!\1).)+)\1', webpage,
 195                 'm3u8 url', default=None, group='url')
 196         if not m3u8_url:
 197             m3u8_url = 'https://teleqmmd.mmdlive.lldns.net/teleqmmd/f386e3b206814e1f8c8c1c71c0f8e748/manifest.m3u8'
 198         formats = self._extract_m3u8_formats(
 199             m3u8_url, video_id, 'mp4', m3u8_id='hls')
 200         self._sort_formats(formats)
 201
 202         return {
 203             'id': video_id,
 204             'title': self._live_title('Télé-Québec - En direct'),
 205             'is_live': True,
 206             'formats': formats,
 207         }