yt_dlp/extractor/telequebec.py

   1 from .common import InfoExtractor
   2 from ..compat import compat_str
   3 from ..utils import (
   4     int_or_none,
   5     smuggle_url,
   6     try_get,
   7     unified_timestamp,
   8 )
   9
  10
  11 class TeleQuebecBaseIE(InfoExtractor):
  12     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/%s/%s_default/index.html?videoId=%s'
  13
  14     @staticmethod
  15     def _brightcove_result(brightcove_id, player_id, account_id='6150020952001'):
  16         return {
  17             '_type': 'url_transparent',
  18             'url': smuggle_url(TeleQuebecBaseIE.BRIGHTCOVE_URL_TEMPLATE % (account_id, player_id, brightcove_id), {'geo_countries': ['CA']}),
  19             'ie_key': 'BrightcoveNew',
  20         }
  21
  22
  23 class TeleQuebecIE(TeleQuebecBaseIE):
  24     _VALID_URL = r'''(?x)
  25                     https?://
  26                         (?:
  27                             zonevideo\.telequebec\.tv/media|
  28                             coucou\.telequebec\.tv/videos
  29                         )/(?P<id>\d+)
  30                     '''
  31     _TESTS = [{
  32         # available till 01.01.2023
  33         'url': 'http://zonevideo.telequebec.tv/media/37578/un-petit-choc-et-puis-repart/un-chef-a-la-cabane',
  34         'info_dict': {
  35             'id': '6155972771001',
  36             'ext': 'mp4',
  37             'title': 'Un petit choc et puis repart!',
  38             'description': 'md5:b04a7e6b3f74e32d7b294cffe8658374',
  39             'timestamp': 1589262469,
  40             'uploader_id': '6150020952001',
  41             'upload_date': '20200512',
  42         },
  43         'add_ie': ['BrightcoveNew'],
  44     }, {
  45         'url': 'https://zonevideo.telequebec.tv/media/55267/le-soleil/passe-partout',
  46         'info_dict': {
  47             'id': '6167180337001',
  48             'ext': 'mp4',
  49             'title': 'Le soleil',
  50             'description': 'md5:64289c922a8de2abbe99c354daffde02',
  51             'uploader_id': '6150020952001',
  52             'upload_date': '20200625',
  53             'timestamp': 1593090307,
  54         },
  55         'add_ie': ['BrightcoveNew'],
  56     }, {
  57         # no description
  58         'url': 'http://zonevideo.telequebec.tv/media/30261',
  59         'only_matching': True,
  60     }, {
  61         'url': 'https://coucou.telequebec.tv/videos/41788/idee-de-genie/l-heure-du-bain',
  62         'only_matching': True,
  63     }]
  64
  65     def _real_extract(self, url):
  66         media_id = self._match_id(url)
  67         media = self._download_json(
  68             'https://mnmedias.api.telequebec.tv/api/v3/media/' + media_id,
  69             media_id)['media']
  70         source_id = next(source_info['sourceId'] for source_info in media['streamInfos'] if source_info.get('source') == 'Brightcove')
  71         info = self._brightcove_result(source_id, '22gPKdt7f')
  72         product = media.get('product') or {}
  73         season = product.get('season') or {}
  74         info.update({
  75             'description': try_get(media, lambda x: x['descriptions'][-1]['text'], compat_str),
  76             'series': try_get(season, lambda x: x['serie']['titre']),
  77             'season': season.get('name'),
  78             'season_number': int_or_none(season.get('seasonNo')),
  79             'episode': product.get('titre'),
  80             'episode_number': int_or_none(product.get('episodeNo')),
  81         })
  82         return info
  83
  84
  85 class TeleQuebecSquatIE(InfoExtractor):
  86     _VALID_URL = r'https://squat\.telequebec\.tv/videos/(?P<id>\d+)'
  87     _TESTS = [{
  88         'url': 'https://squat.telequebec.tv/videos/9314',
  89         'info_dict': {
  90             'id': 'd59ae78112d542e793d83cc9d3a5b530',
  91             'ext': 'mp4',
  92             'title': 'Poupeflekta',
  93             'description': 'md5:2f0718f8d2f8fece1646ee25fb7bce75',
  94             'duration': 1351,
  95             'timestamp': 1569057600,
  96             'upload_date': '20190921',
  97             'series': 'Miraculous : Les Aventures de Ladybug et Chat Noir',
  98             'season': 'Saison 3',
  99             'season_number': 3,
 100             'episode_number': 57,
 101         },
 102         'params': {
 103             'skip_download': True,
 104         },
 105     }]
 106
 107     def _real_extract(self, url):
 108         video_id = self._match_id(url)
 109
 110         video = self._download_json(
 111             'https://squat.api.telequebec.tv/v1/videos/%s' % video_id,
 112             video_id)
 113
 114         media_id = video['sourceId']
 115
 116         return {
 117             '_type': 'url_transparent',
 118             'url': 'http://zonevideo.telequebec.tv/media/%s' % media_id,
 119             'ie_key': TeleQuebecIE.ie_key(),
 120             'id': media_id,
 121             'title': video.get('titre'),
 122             'description': video.get('description'),
 123             'timestamp': unified_timestamp(video.get('datePublication')),
 124             'series': video.get('container'),
 125             'season': video.get('saison'),
 126             'season_number': int_or_none(video.get('noSaison')),
 127             'episode_number': int_or_none(video.get('episode')),
 128         }
 129
 130
 131 class TeleQuebecEmissionIE(InfoExtractor):
 132     _VALID_URL = r'''(?x)
 133                     https?://
 134                         (?:
 135                             [^/]+\.telequebec\.tv/emissions/|
 136                             (?:www\.)?telequebec\.tv/
 137                         )
 138                         (?P<id>[^?#&]+)
 139                     '''
 140     _TESTS = [{
 141         'url': 'http://lindicemcsween.telequebec.tv/emissions/100430013/des-soins-esthetiques-a-377-d-interets-annuels-ca-vous-tente',
 142         'info_dict': {
 143             'id': '6154476028001',
 144             'ext': 'mp4',
 145             'title': 'Des soins esthétiques à 377 % d’intérêts annuels, ça vous tente?',
 146             'description': 'md5:cb4d378e073fae6cce1f87c00f84ae9f',
 147             'upload_date': '20200505',
 148             'timestamp': 1588713424,
 149             'uploader_id': '6150020952001',
 150         },
 151     }, {
 152         'url': 'http://bancpublic.telequebec.tv/emissions/emission-49/31986/jeunes-meres-sous-pression',
 153         'only_matching': True,
 154     }, {
 155         'url': 'http://www.telequebec.tv/masha-et-michka/epi059masha-et-michka-3-053-078',
 156         'only_matching': True,
 157     }, {
 158         'url': 'http://www.telequebec.tv/documentaire/bebes-sur-mesure/',
 159         'only_matching': True,
 160     }]
 161
 162     def _real_extract(self, url):
 163         display_id = self._match_id(url)
 164
 165         webpage = self._download_webpage(url, display_id)
 166
 167         media_id = self._search_regex(
 168             r'mediaId\s*:\s*(?P<id>\d+)', webpage, 'media id')
 169
 170         return self.url_result(
 171             'http://zonevideo.telequebec.tv/media/' + media_id,
 172             TeleQuebecIE.ie_key())
 173
 174
 175 class TeleQuebecLiveIE(TeleQuebecBaseIE):
 176     _VALID_URL = r'https?://zonevideo\.telequebec\.tv/(?P<id>endirect)'
 177     _TEST = {
 178         'url': 'http://zonevideo.telequebec.tv/endirect/',
 179         'info_dict': {
 180             'id': '6159095684001',
 181             'ext': 'mp4',
 182             'title': 're:^Télé-Québec [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
 183             'is_live': True,
 184             'description': 'Canal principal de Télé-Québec',
 185             'uploader_id': '6150020952001',
 186             'timestamp': 1590439901,
 187             'upload_date': '20200525',
 188         },
 189         'params': {
 190             'skip_download': True,
 191         },
 192     }
 193
 194     def _real_extract(self, url):
 195         return self._brightcove_result('6159095684001', 'skCsmi2Uw')
 196
 197
 198 class TeleQuebecVideoIE(TeleQuebecBaseIE):
 199     _VALID_URL = r'https?://video\.telequebec\.tv/player(?:-live)?/(?P<id>\d+)'
 200     _TESTS = [{
 201         'url': 'https://video.telequebec.tv/player/31110/stream',
 202         'info_dict': {
 203             'id': '6202570652001',
 204             'ext': 'mp4',
 205             'title': 'Le coût du véhicule le plus vendu au Canada / Tous les frais liés à la procréation assistée',
 206             'description': 'md5:685a7e4c450ba777c60adb6e71e41526',
 207             'upload_date': '20201019',
 208             'timestamp': 1603115930,
 209             'uploader_id': '6101674910001',
 210         },
 211     }, {
 212         'url': 'https://video.telequebec.tv/player-live/28527',
 213         'only_matching': True,
 214     }]
 215
 216     def _call_api(self, path, video_id):
 217         return self._download_json(
 218             'http://beacon.playback.api.brightcove.com/telequebec/api/assets/' + path,
 219             video_id, query={'device_layout': 'web', 'device_type': 'web'})['data']
 220
 221     def _real_extract(self, url):
 222         asset_id = self._match_id(url)
 223         asset = self._call_api(asset_id, asset_id)['asset']
 224         stream = self._call_api(
 225             asset_id + '/streams/' + asset['streams'][0]['id'], asset_id)['stream']
 226         stream_url = stream['url']
 227         account_id = try_get(
 228             stream, lambda x: x['video_provider_details']['account_id']) or '6101674910001'
 229         info = self._brightcove_result(stream_url, 'default', account_id)
 230         info.update({
 231             'description': asset.get('long_description') or asset.get('short_description'),
 232             'series': asset.get('series_original_name'),
 233             'season_number': int_or_none(asset.get('season_number')),
 234             'episode': asset.get('original_name'),
 235             'episode_number': int_or_none(asset.get('episode_number')),
 236         })
 237         return info