yt_dlp/extractor/tva.py

   1 import re
   2
   3 from .brightcove import BrightcoveNewIE
   4 from .common import InfoExtractor
   5 from ..utils import float_or_none, int_or_none, smuggle_url, strip_or_none
   6 from ..utils.traversal import traverse_obj
   7
   8
   9 class TVAIE(InfoExtractor):
  10     IE_NAME = 'tvaplus'
  11     IE_DESC = 'TVA+'
  12     _VALID_URL = r'https?://(?:www\.)?tvaplus\.ca/(?:[^/?#]+/)*[\w-]+-(?P<id>\d+)(?:$|[#?])'
  13     _TESTS = [{
  14         'url': 'https://www.tvaplus.ca/tva/alerte-amber/saison-1/episode-01-1000036619',
  15         'md5': '949490fd0e7aee11d0543777611fbd53',
  16         'info_dict': {
  17             'id': '6084352463001',
  18             'ext': 'mp4',
  19             'title': 'Mon dernier jour',
  20             'uploader_id': '5481942443001',
  21             'upload_date': '20190907',
  22             'timestamp': 1567899756,
  23             'description': 'md5:9c0d7fbb90939420c651fd977df90145',
  24             'thumbnail': r're:https://.+\.jpg',
  25             'episode': 'Mon dernier jour',
  26             'episode_number': 1,
  27             'tags': ['alerte amber', 'alerte amber saison 1', 'surdemande'],
  28             'duration': 2625.963,
  29             'season': 'Season 1',
  30             'season_number': 1,
  31             'series': 'Alerte Amber',
  32             'channel': 'TVA',
  33         },
  34     }, {
  35         'url': 'https://www.tvaplus.ca/tva/le-baiser-du-barbu/le-baiser-du-barbu-886644190',
  36         'info_dict': {
  37             'id': '6354448043112',
  38             'ext': 'mp4',
  39             'title': 'Le Baiser du barbu',
  40             'uploader_id': '5481942443001',
  41             'upload_date': '20240606',
  42             'timestamp': 1717694023,
  43             'description': 'md5:025b1219086c1cbf4bc27e4e034e8b57',
  44             'thumbnail': r're:https://.+\.jpg',
  45             'episode': 'Le Baiser du barbu',
  46             'tags': ['fullepisode', 'films'],
  47             'duration': 6053.504,
  48             'series': 'Le Baiser du barbu',
  49             'channel': 'TVA',
  50         },
  51     }]
  52     _BC_URL_TMPL = 'https://players.brightcove.net/5481942443001/default_default/index.html?videoId={}'
  53
  54     def _real_extract(self, url):
  55         entity_id = self._match_id(url)
  56         webpage = self._download_webpage(url, entity_id)
  57         entity = self._search_nextjs_data(webpage, entity_id)['props']['pageProps']['staticEntity']
  58         video_id = entity['videoId']
  59         episode = strip_or_none(entity.get('name'))
  60
  61         return {
  62             '_type': 'url_transparent',
  63             'url': smuggle_url(self._BC_URL_TMPL.format(video_id), {'geo_countries': ['CA']}),
  64             'ie_key': BrightcoveNewIE.ie_key(),
  65             'id': video_id,
  66             'title': episode,
  67             'episode': episode,
  68             **traverse_obj(entity, {
  69                 'description': ('longDescription', {str}),
  70                 'duration': ('durationMillis', {float_or_none(scale=1000)}),
  71                 'channel': ('knownEntities', 'channel', 'name', {str}),
  72                 'series': ('knownEntities', 'videoShow', 'name', {str}),
  73                 'season_number': ('slug', {lambda x: re.search(r'/s(?:ai|ea)son-(\d+)/', x)}, 1, {int_or_none}),
  74                 'episode_number': ('episodeNumber', {int_or_none}),
  75             }),
  76         }