yt_dlp/extractor/toutv.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import json
   5
   6 from .radiocanada import RadioCanadaIE
   7 from ..compat import compat_HTTPError
   8 from ..utils import (
   9     ExtractorError,
  10     int_or_none,
  11     merge_dicts,
  12 )
  13
  14
  15 class TouTvIE(RadioCanadaIE):
  16     _NETRC_MACHINE = 'toutv'
  17     IE_NAME = 'tou.tv'
  18     _VALID_URL = r'https?://ici\.tou\.tv/(?P<id>[a-zA-Z0-9_-]+(?:/S[0-9]+[EC][0-9]+)?)'
  19
  20     _TESTS = [{
  21         'url': 'http://ici.tou.tv/garfield-tout-court/S2015E17',
  22         'info_dict': {
  23             'id': '122017',
  24             'ext': 'mp4',
  25             'title': 'Saison 2015 Épisode 17',
  26             'description': 'La photo de famille 2',
  27             'upload_date': '20100717',
  28         },
  29         'params': {
  30             # m3u8 download
  31             'skip_download': True,
  32         },
  33         'skip': '404 Not Found',
  34     }, {
  35         'url': 'http://ici.tou.tv/hackers',
  36         'only_matching': True,
  37     }, {
  38         'url': 'https://ici.tou.tv/l-age-adulte/S01C501',
  39         'only_matching': True,
  40     }]
  41     _CLIENT_KEY = '90505c8d-9c34-4f34-8da1-3a85bdc6d4f4'
  42
  43     def _perform_login(self, username, password):
  44         try:
  45             self._access_token = self._download_json(
  46                 'https://services.radio-canada.ca/toutv/profiling/accounts/login',
  47                 None, 'Logging in', data=json.dumps({
  48                     'ClientId': self._CLIENT_KEY,
  49                     'ClientSecret': '34026772-244b-49b6-8b06-317b30ac9a20',
  50                     'Email': username,
  51                     'Password': password,
  52                     'Scope': 'id.write media-validation.read',
  53                 }).encode(), headers={
  54                     'Authorization': 'client-key ' + self._CLIENT_KEY,
  55                     'Content-Type': 'application/json;charset=utf-8',
  56                 })['access_token']
  57         except ExtractorError as e:
  58             if isinstance(e.cause, compat_HTTPError) and e.cause.code == 401:
  59                 error = self._parse_json(e.cause.read().decode(), None)['Message']
  60                 raise ExtractorError(error, expected=True)
  61             raise
  62         self._claims = self._call_api('validation/v2/getClaims')['claims']
  63
  64     def _real_extract(self, url):
  65         path = self._match_id(url)
  66         metadata = self._download_json(
  67             'https://services.radio-canada.ca/toutv/presentation/%s' % path, path, query={
  68                 'client_key': self._CLIENT_KEY,
  69                 'device': 'web',
  70                 'version': 4,
  71             })
  72         # IsDrm does not necessarily mean the video is DRM protected (see
  73         # https://github.com/ytdl-org/youtube-dl/issues/13994).
  74         if not self.get_param('allow_unplayable_formats') and metadata.get('IsDrm'):
  75             self.report_warning('This video is probably DRM protected.', path)
  76         video_id = metadata['IdMedia']
  77         details = metadata['Details']
  78
  79         return merge_dicts({
  80             'id': video_id,
  81             'title': details.get('OriginalTitle'),
  82             'description': details.get('Description'),
  83             'thumbnail': details.get('ImageUrl'),
  84             'duration': int_or_none(details.get('LengthInSeconds')),
  85             'series': metadata.get('ProgramTitle'),
  86             'season_number': int_or_none(metadata.get('SeasonNumber')),
  87             'season': metadata.get('SeasonTitle'),
  88             'episode_number': int_or_none(metadata.get('EpisodeNumber')),
  89             'episode': metadata.get('EpisodeTitle'),
  90         }, self._extract_info(metadata.get('AppCode', 'toutv'), video_id))