yt_dlp/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/%s/%s_default/index.html?videoId=%s'
  16
  17     @staticmethod
  18     def _brightcove_result(brightcove_id, player_id, account_id='6150020952001'):
  19         return {
  20             '_type': 'url_transparent',
  21             'url': smuggle_url(TeleQuebecBaseIE.BRIGHTCOVE_URL_TEMPLATE % (account_id, player_id, brightcove_id), {'geo_countries': ['CA']}),
  22             'ie_key': 'BrightcoveNew',
  23         }
  24
  25
  26 class TeleQuebecIE(TeleQuebecBaseIE):
  27     _VALID_URL = r'''(?x)
  28                     https?://
  29                         (?:
  30                             zonevideo\.telequebec\.tv/media|
  31                             coucou\.telequebec\.tv/videos
  32                         )/(?P<id>\d+)
  33                     '''
  34     _TESTS = [{
  35         # available till 01.01.2023
  36         'url': 'http://zonevideo.telequebec.tv/media/37578/un-petit-choc-et-puis-repart/un-chef-a-la-cabane',
  37         'info_dict': {
  38             'id': '6155972771001',
  39             'ext': 'mp4',
  40             'title': 'Un petit choc et puis repart!',
  41             'description': 'md5:b04a7e6b3f74e32d7b294cffe8658374',
  42             'timestamp': 1589262469,
  43             'uploader_id': '6150020952001',
  44             'upload_date': '20200512',
  45         },
  46         'params': {
  47             'format': 'bestvideo',
  48         },
  49         'add_ie': ['BrightcoveNew'],
  50     }, {
  51         'url': 'https://zonevideo.telequebec.tv/media/55267/le-soleil/passe-partout',
  52         'info_dict': {
  53             'id': '6167180337001',
  54             'ext': 'mp4',
  55             'title': 'Le soleil',
  56             'description': 'md5:64289c922a8de2abbe99c354daffde02',
  57             'uploader_id': '6150020952001',
  58             'upload_date': '20200625',
  59             'timestamp': 1593090307,
  60         },
  61         'params': {
  62             'format': 'bestvideo',
  63         },
  64         'add_ie': ['BrightcoveNew'],
  65     }, {
  66         # no description
  67         'url': 'http://zonevideo.telequebec.tv/media/30261',
  68         'only_matching': True,
  69     }, {
  70         'url': 'https://coucou.telequebec.tv/videos/41788/idee-de-genie/l-heure-du-bain',
  71         'only_matching': True,
  72     }]
  73
  74     def _real_extract(self, url):
  75         media_id = self._match_id(url)
  76         media = self._download_json(
  77             'https://mnmedias.api.telequebec.tv/api/v3/media/' + media_id,
  78             media_id)['media']
  79         source_id = next(source_info['sourceId'] for source_info in media['streamInfos'] if source_info.get('source') == 'Brightcove')
  80         info = self._brightcove_result(source_id, '22gPKdt7f')
  81         product = media.get('product') or {}
  82         season = product.get('season') or {}
  83         info.update({
  84             'description': try_get(media, lambda x: x['descriptions'][-1]['text'], compat_str),
  85             'series': try_get(season, lambda x: x['serie']['titre']),
  86             'season': season.get('name'),
  87             'season_number': int_or_none(season.get('seasonNo')),
  88             'episode': product.get('titre'),
  89             'episode_number': int_or_none(product.get('episodeNo')),
  90         })
  91         return info
  92
  93
  94 class TeleQuebecSquatIE(InfoExtractor):
  95     _VALID_URL = r'https://squat\.telequebec\.tv/videos/(?P<id>\d+)'
  96     _TESTS = [{
  97         'url': 'https://squat.telequebec.tv/videos/9314',
  98         'info_dict': {
  99             'id': 'd59ae78112d542e793d83cc9d3a5b530',
 100             'ext': 'mp4',
 101             'title': 'Poupeflekta',
 102             'description': 'md5:2f0718f8d2f8fece1646ee25fb7bce75',
 103             'duration': 1351,
 104             'timestamp': 1569057600,
 105             'upload_date': '20190921',
 106             'series': 'Miraculous : Les Aventures de Ladybug et Chat Noir',
 107             'season': 'Saison 3',
 108             'season_number': 3,
 109             'episode_number': 57,
 110         },
 111         'params': {
 112             'skip_download': True,
 113         },
 114     }]
 115
 116     def _real_extract(self, url):
 117         video_id = self._match_id(url)
 118
 119         video = self._download_json(
 120             'https://squat.api.telequebec.tv/v1/videos/%s' % video_id,
 121             video_id)
 122
 123         media_id = video['sourceId']
 124
 125         return {
 126             '_type': 'url_transparent',
 127             'url': 'http://zonevideo.telequebec.tv/media/%s' % media_id,
 128             'ie_key': TeleQuebecIE.ie_key(),
 129             'id': media_id,
 130             'title': video.get('titre'),
 131             'description': video.get('description'),
 132             'timestamp': unified_timestamp(video.get('datePublication')),
 133             'series': video.get('container'),
 134             'season': video.get('saison'),
 135             'season_number': int_or_none(video.get('noSaison')),
 136             'episode_number': int_or_none(video.get('episode')),
 137         }
 138
 139
 140 class TeleQuebecEmissionIE(InfoExtractor):
 141     _VALID_URL = r'''(?x)
 142                     https?://
 143                         (?:
 144                             [^/]+\.telequebec\.tv/emissions/|
 145                             (?:www\.)?telequebec\.tv/
 146                         )
 147                         (?P<id>[^?#&]+)
 148                     '''
 149     _TESTS = [{
 150         'url': 'http://lindicemcsween.telequebec.tv/emissions/100430013/des-soins-esthetiques-a-377-d-interets-annuels-ca-vous-tente',
 151         'info_dict': {
 152             'id': '6154476028001',
 153             'ext': 'mp4',
 154             'title': 'Des soins esthétiques à 377 % d’intérêts annuels, ça vous tente?',
 155             'description': 'md5:cb4d378e073fae6cce1f87c00f84ae9f',
 156             'upload_date': '20200505',
 157             'timestamp': 1588713424,
 158             'uploader_id': '6150020952001',
 159         },
 160         'params': {
 161             'format': 'bestvideo',
 162         },
 163     }, {
 164         'url': 'http://bancpublic.telequebec.tv/emissions/emission-49/31986/jeunes-meres-sous-pression',
 165         'only_matching': True,
 166     }, {
 167         'url': 'http://www.telequebec.tv/masha-et-michka/epi059masha-et-michka-3-053-078',
 168         'only_matching': True,
 169     }, {
 170         'url': 'http://www.telequebec.tv/documentaire/bebes-sur-mesure/',
 171         'only_matching': True,
 172     }]
 173
 174     def _real_extract(self, url):
 175         display_id = self._match_id(url)
 176
 177         webpage = self._download_webpage(url, display_id)
 178
 179         media_id = self._search_regex(
 180             r'mediaId\s*:\s*(?P<id>\d+)', webpage, 'media id')
 181
 182         return self.url_result(
 183             'http://zonevideo.telequebec.tv/media/' + media_id,
 184             TeleQuebecIE.ie_key())
 185
 186
 187 class TeleQuebecLiveIE(TeleQuebecBaseIE):
 188     _VALID_URL = r'https?://zonevideo\.telequebec\.tv/(?P<id>endirect)'
 189     _TEST = {
 190         'url': 'http://zonevideo.telequebec.tv/endirect/',
 191         'info_dict': {
 192             'id': '6159095684001',
 193             'ext': 'mp4',
 194             'title': 're:^Télé-Québec [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
 195             'is_live': True,
 196             'description': 'Canal principal de Télé-Québec',
 197             'uploader_id': '6150020952001',
 198             'timestamp': 1590439901,
 199             'upload_date': '20200525',
 200         },
 201         'params': {
 202             'skip_download': True,
 203         },
 204     }
 205
 206     def _real_extract(self, url):
 207         return self._brightcove_result('6159095684001', 'skCsmi2Uw')
 208
 209
 210 class TeleQuebecVideoIE(TeleQuebecBaseIE):
 211     _VALID_URL = r'https?://video\.telequebec\.tv/player(?:-live)?/(?P<id>\d+)'
 212     _TESTS = [{
 213         'url': 'https://video.telequebec.tv/player/31110/stream',
 214         'info_dict': {
 215             'id': '6202570652001',
 216             'ext': 'mp4',
 217             'title': 'Le coût du véhicule le plus vendu au Canada / Tous les frais liés à la procréation assistée',
 218             'description': 'md5:685a7e4c450ba777c60adb6e71e41526',
 219             'upload_date': '20201019',
 220             'timestamp': 1603115930,
 221             'uploader_id': '6101674910001',
 222         },
 223         'params': {
 224             'format': 'bestvideo',
 225         },
 226     }, {
 227         'url': 'https://video.telequebec.tv/player-live/28527',
 228         'only_matching': True,
 229     }]
 230
 231     def _call_api(self, path, video_id):
 232         return self._download_json(
 233             'http://beacon.playback.api.brightcove.com/telequebec/api/assets/' + path,
 234             video_id, query={'device_layout': 'web', 'device_type': 'web'})['data']
 235
 236     def _real_extract(self, url):
 237         asset_id = self._match_id(url)
 238         asset = self._call_api(asset_id, asset_id)['asset']
 239         stream = self._call_api(
 240             asset_id + '/streams/' + asset['streams'][0]['id'], asset_id)['stream']
 241         stream_url = stream['url']
 242         account_id = try_get(
 243             stream, lambda x: x['video_provider_details']['account_id']) or '6101674910001'
 244         info = self._brightcove_result(stream_url, 'default', account_id)
 245         info.update({
 246             'description': asset.get('long_description') or asset.get('short_description'),
 247             'series': asset.get('series_original_name'),
 248             'season_number': int_or_none(asset.get('season_number')),
 249             'episode': asset.get('original_name'),
 250             'episode_number': int_or_none(asset.get('episode_number')),
 251         })
 252         return info