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