yt_dlp/extractor/cwtv.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     ExtractorError,
   4     int_or_none,
   5     parse_age_limit,
   6     parse_iso8601,
   7     smuggle_url,
   8     str_or_none,
   9 )
  10
  11
  12 class CWTVIE(InfoExtractor):
  13     _VALID_URL = r'https?://(?:www\.)?cw(?:tv(?:pr)?|seed)\.com/(?:shows/)?(?:[^/]+/)+[^?]*\?.*\b(?:play|watch)=(?P<id>[a-z0-9]{8}-[a-z0-9]{4}-[a-z0-9]{4}-[a-z0-9]{4}-[a-z0-9]{12})'
  14     _TESTS = [{
  15         'url': 'http://cwtv.com/shows/arrow/legends-of-yesterday/?play=6b15e985-9345-4f60-baf8-56e96be57c63',
  16         'info_dict': {
  17             'id': '6b15e985-9345-4f60-baf8-56e96be57c63',
  18             'ext': 'mp4',
  19             'title': 'Legends of Yesterday',
  20             'description': 'Oliver and Barry Allen take Kendra Saunders and Carter Hall to a remote location to keep them hidden from Vandal Savage while they figure out how to defeat him.',
  21             'duration': 2665,
  22             'series': 'Arrow',
  23             'season_number': 4,
  24             'season': '4',
  25             'episode_number': 8,
  26             'upload_date': '20151203',
  27             'timestamp': 1449122100,
  28         },
  29         'params': {
  30             # m3u8 download
  31             'skip_download': True,
  32         },
  33         'skip': 'redirect to http://cwtv.com/shows/arrow/',
  34     }, {
  35         'url': 'http://www.cwseed.com/shows/whose-line-is-it-anyway/jeff-davis-4/?play=24282b12-ead2-42f2-95ad-26770c2c6088',
  36         'info_dict': {
  37             'id': '24282b12-ead2-42f2-95ad-26770c2c6088',
  38             'ext': 'mp4',
  39             'title': 'Jeff Davis 4',
  40             'description': 'Jeff Davis is back to make you laugh.',
  41             'duration': 1263,
  42             'series': 'Whose Line Is It Anyway?',
  43             'season_number': 11,
  44             'episode_number': 20,
  45             'upload_date': '20151006',
  46             'timestamp': 1444107300,
  47             'age_limit': 14,
  48             'uploader': 'CWTV',
  49             'thumbnail': r're:^https?://.*\.jpe?g$',
  50             'chapters': 'count:4',
  51             'episode': 'Episode 20',
  52             'season': 'Season 11',
  53         },
  54         'params': {
  55             # m3u8 download
  56             'skip_download': True,
  57         },
  58     }, {
  59         'url': 'http://cwtv.com/thecw/chroniclesofcisco/?play=8adebe35-f447-465f-ab52-e863506ff6d6',
  60         'only_matching': True,
  61     }, {
  62         'url': 'http://cwtvpr.com/the-cw/video?watch=9eee3f60-ef4e-440b-b3b2-49428ac9c54e',
  63         'only_matching': True,
  64     }, {
  65         'url': 'http://cwtv.com/shows/arrow/legends-of-yesterday/?watch=6b15e985-9345-4f60-baf8-56e96be57c63',
  66         'only_matching': True,
  67     }]
  68
  69     def _real_extract(self, url):
  70         video_id = self._match_id(url)
  71         data = self._download_json(
  72             'http://images.cwtv.com/feed/mobileapp/video-meta/apiversion_8/guid_' + video_id,
  73             video_id)
  74         if data.get('result') != 'ok':
  75             raise ExtractorError(data['msg'], expected=True)
  76         video_data = data['video']
  77         title = video_data['title']
  78         mpx_url = video_data.get('mpx_url') or f'http://link.theplatform.com/s/cwtv/media/guid/2703454149/{video_id}?formats=M3U'
  79
  80         season = str_or_none(video_data.get('season'))
  81         episode = str_or_none(video_data.get('episode'))
  82         if episode and season:
  83             episode = episode[len(season):]
  84
  85         return {
  86             '_type': 'url_transparent',
  87             'id': video_id,
  88             'title': title,
  89             'url': smuggle_url(mpx_url, {'force_smil_url': True}),
  90             'description': video_data.get('description_long'),
  91             'duration': int_or_none(video_data.get('duration_secs')),
  92             'series': video_data.get('series_name'),
  93             'season_number': int_or_none(season),
  94             'episode_number': int_or_none(episode),
  95             'timestamp': parse_iso8601(video_data.get('start_time')),
  96             'age_limit': parse_age_limit(video_data.get('rating')),
  97             'ie_key': 'ThePlatform',
  98             'thumbnail': video_data.get('large_thumbnail'),
  99         }