yt_dlp/extractor/tviplayer.py

   1 from .common import InfoExtractor
   2 from ..utils import traverse_obj
   3
   4
   5 class TVIPlayerIE(InfoExtractor):
   6     _VALID_URL = r'https?://tviplayer\.iol\.pt(/programa/[\w-]+/[a-f0-9]+)?/\w+/(?P<id>\w+)'
   7     _TESTS = [{
   8         'url': 'https://tviplayer.iol.pt/programa/jornal-das-8/53c6b3903004dc006243d0cf/video/61c8e8b90cf2c7ea0f0f71a9',
   9         'info_dict': {
  10             'id': '61c8e8b90cf2c7ea0f0f71a9',
  11             'ext': 'mp4',
  12             'duration': 4167,
  13             'title': 'Jornal das 8 - 26 de dezembro de 2021',
  14             'thumbnail': 'https://www.iol.pt/multimedia/oratvi/multimedia/imagem/id/61c8ee630cf2cc58e7d98d9f/',
  15             'season_number': 8,
  16             'season': 'Season 8',
  17         },
  18     }, {
  19         'url': 'https://tviplayer.iol.pt/programa/isabel/62b471090cf26256cd2a8594/video/62be445f0cf2ea4f0a5218e5',
  20         'info_dict': {
  21             'id': '62be445f0cf2ea4f0a5218e5',
  22             'ext': 'mp4',
  23             'duration': 3255,
  24             'season': 'Season 1',
  25             'title': 'Isabel - Episódio 1',
  26             'thumbnail': 'https://www.iol.pt/multimedia/oratvi/multimedia/imagem/id/62beac200cf2f9a86eab856b/',
  27             'season_number': 1,
  28         },
  29     }, {
  30         # no /programa/
  31         'url': 'https://tviplayer.iol.pt/video/62c4131c0cf2f9a86eac06bb',
  32         'info_dict': {
  33             'id': '62c4131c0cf2f9a86eac06bb',
  34             'ext': 'mp4',
  35             'title': 'David e Mickael Carreira respondem: «Qual é o próximo a ser pai?»',
  36             'thumbnail': 'https://www.iol.pt/multimedia/oratvi/multimedia/imagem/id/62c416490cf2ea367d4433fd/',
  37             'season': 'Season 2',
  38             'duration': 148,
  39             'season_number': 2,
  40         },
  41     }, {
  42         # episodio url
  43         'url': 'https://tviplayer.iol.pt/programa/para-sempre/61716c360cf2365a5ed894c4/episodio/t1e187',
  44         'info_dict': {
  45             'id': 't1e187',
  46             'ext': 'mp4',
  47             'season': 'Season 1',
  48             'title': 'Quem denunciou Pedro?',
  49             'thumbnail': 'https://www.iol.pt/multimedia/oratvi/multimedia/imagem/id/62eda30b0cf2ea367d48973b/',
  50             'duration': 1250,
  51             'season_number': 1,
  52         },
  53     }]
  54
  55     def _real_initialize(self):
  56         self.wms_auth_sign_token = self._download_webpage(
  57             'https://services.iol.pt/matrix?userId=', 'wmsAuthSign',
  58             note='Trying to get wmsAuthSign token')
  59
  60     def _real_extract(self, url):
  61         video_id = self._match_id(url)
  62         webpage = self._download_webpage(url, video_id)
  63
  64         json_data = self._search_json(
  65             r'<script>\s*jsonData\s*=', webpage, 'json_data', video_id)
  66
  67         formats, subtitles = self._extract_m3u8_formats_and_subtitles(
  68             f'{json_data["videoUrl"]}?wmsAuthSign={self.wms_auth_sign_token}',
  69             video_id, ext='mp4')
  70         return {
  71             'id': video_id,
  72             'title': json_data.get('title') or self._og_search_title(webpage),
  73             'thumbnail': json_data.get('cover') or self._og_search_thumbnail(webpage),
  74             'duration': json_data.get('duration'),
  75             'formats': formats,
  76             'subtitles': subtitles,
  77             'season_number': traverse_obj(json_data, ('program', 'seasonNum')),
  78         }