yt_dlp/extractor/polsatgo.py

   1 from uuid import uuid4
   2 import json
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     int_or_none,
   7     try_get,
   8     url_or_none,
   9     ExtractorError,
  10 )
  11
  12
  13 class PolsatGoIE(InfoExtractor):
  14     _VALID_URL = r'https?://(?:www\.)?polsat(?:box)?go\.pl/.+/(?P<id>[0-9a-fA-F]+)(?:[/#?]|$)'
  15     _TESTS = [{
  16         'url': 'https://polsatgo.pl/wideo/seriale/swiat-wedlug-kiepskich/5024045/sezon-1/5028300/swiat-wedlug-kiepskich-odcinek-88/4121',
  17         'info_dict': {
  18             'id': '4121',
  19             'ext': 'mp4',
  20             'title': 'Świat według Kiepskich - Odcinek 88',
  21             'age_limit': 12,
  22         },
  23     }]
  24
  25     def _extract_formats(self, sources, video_id):
  26         for source in sources or []:
  27             if not source.get('id'):
  28                 continue
  29             url = url_or_none(self._call_api(
  30                 'drm', video_id, 'getPseudoLicense',
  31                 {'mediaId': video_id, 'sourceId': source['id']}).get('url'))
  32             if not url:
  33                 continue
  34             yield {
  35                 'url': url,
  36                 'height': int_or_none(try_get(source, lambda x: x['quality'][:-1]))
  37             }
  38
  39     def _real_extract(self, url):
  40         video_id = self._match_id(url)
  41         media = self._call_api('navigation', video_id, 'prePlayData', {'mediaId': video_id})['mediaItem']
  42
  43         formats = list(self._extract_formats(
  44             try_get(media, lambda x: x['playback']['mediaSources']), video_id))
  45         self._sort_formats(formats)
  46
  47         return {
  48             'id': video_id,
  49             'title': media['displayInfo']['title'],
  50             'formats': formats,
  51             'age_limit': int_or_none(media['displayInfo']['ageGroup'])
  52         }
  53
  54     def _call_api(self, endpoint, media_id, method, params):
  55         rand_uuid = str(uuid4())
  56         res = self._download_json(
  57             f'https://b2c-mobile.redefine.pl/rpc/{endpoint}/', media_id,
  58             note=f'Downloading {method} JSON metadata',
  59             data=json.dumps({
  60                 'method': method,
  61                 'id': '2137',
  62                 'jsonrpc': '2.0',
  63                 'params': {
  64                     **params,
  65                     'userAgentData': {
  66                         'deviceType': 'mobile',
  67                         'application': 'native',
  68                         'os': 'android',
  69                         'build': 10003,
  70                         'widevine': False,
  71                         'portal': 'pg',
  72                         'player': 'cpplayer',
  73                     },
  74                     'deviceId': {
  75                         'type': 'other',
  76                         'value': rand_uuid,
  77                     },
  78                     'clientId': rand_uuid,
  79                     'cpid': 1,
  80                 },
  81             }).encode('utf-8'),
  82             headers={'Content-type': 'application/json'})
  83         if not res.get('result'):
  84             if res['error']['code'] == 13404:
  85                 raise ExtractorError('This video is either unavailable in your region or is DRM protected', expected=True)
  86             raise ExtractorError(f'Solorz said: {res["error"]["message"]} - {res["error"]["data"]["userMessage"]}')
  87         return res['result']