yt_dlp/extractor/paramountplus.py

   1 from __future__ import unicode_literals
   2
   3 from .common import InfoExtractor
   4 from .cbs import CBSBaseIE
   5 from ..utils import (
   6     int_or_none,
   7     url_or_none,
   8 )
   9
  10
  11 class ParamountPlusIE(CBSBaseIE):
  12     _VALID_URL = r'''(?x)
  13         (?:
  14             paramountplus:|
  15             https?://(?:www\.)?(?:
  16                 paramountplus\.com/(?:shows/[^/]+/video|movies/[^/]+)/
  17         )(?P<id>[\w-]+))'''
  18
  19     # All tests are blocked outside US
  20     _TESTS = [{
  21         'url': 'https://www.paramountplus.com/shows/catdog/video/Oe44g5_NrlgiZE3aQVONleD6vXc8kP0k/catdog-climb-every-catdog-the-canine-mutiny/',
  22         'info_dict': {
  23             'id': 'Oe44g5_NrlgiZE3aQVONleD6vXc8kP0k',
  24             'ext': 'mp4',
  25             'title': 'CatDog - Climb Every CatDog/The Canine Mutiny',
  26             'description': 'md5:7ac835000645a69933df226940e3c859',
  27             'duration': 1418,
  28             'timestamp': 920264400,
  29             'upload_date': '19990301',
  30             'uploader': 'CBSI-NEW',
  31         },
  32         'params': {
  33             'skip_download': 'm3u8',
  34         },
  35     }, {
  36         'url': 'https://www.paramountplus.com/shows/tooning-out-the-news/video/6hSWYWRrR9EUTz7IEe5fJKBhYvSUfexd/7-23-21-week-in-review-rep-jahana-hayes-howard-fineman-sen-michael-bennet-sheera-frenkel-cecilia-kang-/',
  37         'info_dict': {
  38             'id': '6hSWYWRrR9EUTz7IEe5fJKBhYvSUfexd',
  39             'ext': 'mp4',
  40             'title': '7/23/21 WEEK IN REVIEW (Rep. Jahana Hayes/Howard Fineman/Sen. Michael Bennet/Sheera Frenkel & Cecilia Kang)',
  41             'description': 'md5:f4adcea3e8b106192022e121f1565bae',
  42             'duration': 2506,
  43             'timestamp': 1627063200,
  44             'upload_date': '20210723',
  45             'uploader': 'CBSI-NEW',
  46         },
  47         'params': {
  48             'skip_download': 'm3u8',
  49         },
  50     }, {
  51         'url': 'https://www.paramountplus.com/movies/daddys-home/vM2vm0kE6vsS2U41VhMRKTOVHyQAr6pC',
  52         'info_dict': {
  53             'id': 'vM2vm0kE6vsS2U41VhMRKTOVHyQAr6pC',
  54             'ext': 'mp4',
  55             'title': 'Daddy\'s Home',
  56             'upload_date': '20151225',
  57             'description': 'md5:a0beaf24e8d3b0e81b2ee41d47c06f33',
  58             'uploader': 'CBSI-NEW',
  59             'timestamp': 1451030400,
  60         },
  61         'params': {
  62             'skip_download': 'm3u8',
  63         },
  64         'expected_warnings': ['Ignoring subtitle tracks'],  # TODO: Investigate this
  65     }, {
  66         'url': 'https://www.paramountplus.com/movies/sonic-the-hedgehog/5EKDXPOzdVf9voUqW6oRuocyAEeJGbEc',
  67         'info_dict': {
  68             'id': '5EKDXPOzdVf9voUqW6oRuocyAEeJGbEc',
  69             'ext': 'mp4',
  70             'uploader': 'CBSI-NEW',
  71             'description': 'md5:bc7b6fea84ba631ef77a9bda9f2ff911',
  72             'timestamp': 1577865600,
  73             'title': 'Sonic the Hedgehog',
  74             'upload_date': '20200101',
  75         },
  76         'params': {
  77             'skip_download': 'm3u8',
  78         },
  79         'expected_warnings': ['Ignoring subtitle tracks'],
  80     }, {
  81         'url': 'https://www.paramountplus.com/shows/all-rise/video/QmR1WhNkh1a_IrdHZrbcRklm176X_rVc/all-rise-space/',
  82         'only_matching': True,
  83     }, {
  84         'url': 'https://www.paramountplus.com/movies/million-dollar-american-princesses-meghan-and-harry/C0LpgNwXYeB8txxycdWdR9TjxpJOsdCq',
  85         'only_matching': True,
  86     }]
  87
  88     def _extract_video_info(self, content_id, mpx_acc=2198311517):
  89         items_data = self._download_json(
  90             'https://www.paramountplus.com/apps-api/v2.0/androidtv/video/cid/%s.json' % content_id,
  91             content_id, query={'locale': 'en-us', 'at': 'ABCqWNNSwhIqINWIIAG+DFzcFUvF8/vcN6cNyXFFfNzWAIvXuoVgX+fK4naOC7V8MLI='}, headers=self.geo_verification_headers())
  92
  93         asset_types = {
  94             item.get('assetType'): {
  95                 'format': 'SMIL',
  96                 'formats': 'MPEG4,M3U',
  97             } for item in items_data['itemList']
  98         }
  99         item = items_data['itemList'][-1]
 100         return self._extract_common_video_info(content_id, asset_types, mpx_acc, extra_info={
 101             'title': item.get('title'),
 102             'series': item.get('seriesTitle'),
 103             'season_number': int_or_none(item.get('seasonNum')),
 104             'episode_number': int_or_none(item.get('episodeNum')),
 105             'duration': int_or_none(item.get('duration')),
 106             'thumbnail': url_or_none(item.get('thumbnail')),
 107         })
 108
 109
 110 class ParamountPlusSeriesIE(InfoExtractor):
 111     _VALID_URL = r'https?://(?:www\.)?paramountplus\.com/shows/(?P<id>[a-zA-Z0-9-_]+)/?(?:[#?]|$)'
 112     _TESTS = [{
 113         'url': 'https://www.paramountplus.com/shows/drake-josh',
 114         'playlist_mincount': 50,
 115         'info_dict': {
 116             'id': 'drake-josh',
 117         }
 118     }, {
 119         'url': 'https://www.paramountplus.com/shows/hawaii_five_0/',
 120         'playlist_mincount': 240,
 121         'info_dict': {
 122             'id': 'hawaii_five_0',
 123         }
 124     }, {
 125         'url': 'https://www.paramountplus.com/shows/spongebob-squarepants/',
 126         'playlist_mincount': 248,
 127         'info_dict': {
 128             'id': 'spongebob-squarepants',
 129         }
 130     }]
 131     _API_URL = 'https://www.paramountplus.com/shows/{}/xhr/episodes/page/0/size/100000/xs/0/season/0/'
 132
 133     def _entries(self, show_name):
 134         show_json = self._download_json(self._API_URL.format(show_name), video_id=show_name)
 135         if show_json.get('success'):
 136             for episode in show_json['result']['data']:
 137                 yield self.url_result(
 138                     'https://www.paramountplus.com%s' % episode['url'],
 139                     ie=ParamountPlusIE.ie_key(), video_id=episode['content_id'])
 140
 141     def _real_extract(self, url):
 142         show_name = self._match_id(url)
 143         return self.playlist_result(self._entries(show_name), playlist_id=show_name)