yt_dlp/extractor/duoplay.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     ExtractorError,
   4     extract_attributes,
   5     get_element_text_and_html_by_tag,
   6     int_or_none,
   7     join_nonempty,
   8     parse_qs,
   9     str_or_none,
  10     try_call,
  11     unified_timestamp,
  12 )
  13 from ..utils.traversal import traverse_obj, value
  14
  15
  16 class DuoplayIE(InfoExtractor):
  17     _VALID_URL = r'https?://duoplay\.ee/(?P<id>\d+)(?:[/?#]|$)'
  18     _TESTS = [{
  19         'note': 'Siberi võmm S02E12',
  20         'url': 'https://duoplay.ee/4312/siberi-vomm?ep=24',
  21         'md5': '1ff59d535310ac9c5cf5f287d8f91b2d',
  22         'info_dict': {
  23             'id': '4312_24',
  24             'ext': 'mp4',
  25             'title': 'Operatsioon "Öö"',
  26             'thumbnail': r're:https://.+\.jpg(?:\?c=\d+)?$',
  27             'description': 'md5:8ef98f38569d6b8b78f3d350ccc6ade8',
  28             'upload_date': '20170523',
  29             'timestamp': 1495567800,
  30             'series': 'Siberi võmm',
  31             'series_id': '4312',
  32             'season': 'Season 2',
  33             'season_number': 2,
  34             'episode': 'Operatsioon "Öö"',
  35             'episode_number': 12,
  36             'episode_id': '24',
  37         },
  38         'skip': 'No video found',
  39     }, {
  40         'note': 'Empty title',
  41         'url': 'https://duoplay.ee/17/uhikarotid?ep=14',
  42         'md5': 'cba9f5dabf2582b224d80ac44fb80e47',
  43         'info_dict': {
  44             'id': '17_14',
  45             'ext': 'mp4',
  46             'title': 'Episode 14',
  47             'thumbnail': r're:https?://.+\.jpg',
  48             'description': 'md5:4719b418e058c209def41d48b601276e',
  49             'upload_date': '20100916',
  50             'timestamp': 1284661800,
  51             'series': 'Ühikarotid',
  52             'series_id': '17',
  53             'season': 'Season 2',
  54             'season_number': 2,
  55             'episode_id': '14',
  56             'release_year': 2010,
  57             'episode': 'Episode 14',
  58             'episode_number': 14,
  59         },
  60     }, {
  61         'note': 'Movie without expiry',
  62         'url': 'https://duoplay.ee/5501/pilvede-all.-neljas-ode',
  63         'md5': '7abf63d773a49ef7c39f2c127842b8fd',
  64         'info_dict': {
  65             'id': '5501',
  66             'ext': 'mp4',
  67             'title': 'Pilvede all. Neljas õde',
  68             'thumbnail': r're:https://.+\.jpg(?:\?c=\d+)?$',
  69             'description': 'md5:d86a70f8f31e82c369d4d4f4c79b1279',
  70             'cast': 'count:9',
  71             'upload_date': '20221214',
  72             'timestamp': 1671054000,
  73             'release_year': 2018,
  74         },
  75         'skip': 'No video found',
  76     }, {
  77         'note': 'Episode url without show name',
  78         'url': 'https://duoplay.ee/9644?ep=185',
  79         'md5': '63f324b4fe2dbd8194dca16a6d52184a',
  80         'info_dict': {
  81             'id': '9644_185',
  82             'ext': 'mp4',
  83             'title': 'Episode 185',
  84             'thumbnail': r're:https?://.+\.jpg',
  85             'description': 'md5:ed25ba4e9e5d54bc291a4a0cdd241467',
  86             'upload_date': '20241120',
  87             'timestamp': 1732077000,
  88             'episode': 'Episode 63',
  89             'episode_id': '185',
  90             'episode_number': 63,
  91             'season': 'Season 2',
  92             'season_number': 2,
  93             'series': 'Telehommik',
  94             'series_id': '9644',
  95         },
  96     }]
  97
  98     def _real_extract(self, url):
  99         telecast_id = self._match_id(url)
 100         episode = traverse_obj(parse_qs(url), ('ep', 0, {int_or_none}, {str_or_none}))
 101         video_id = join_nonempty(telecast_id, episode, delim='_')
 102         webpage = self._download_webpage(url, video_id)
 103         video_player = try_call(lambda: extract_attributes(
 104             get_element_text_and_html_by_tag('video-player', webpage)[1]))
 105         if not video_player or not video_player.get('manifest-url'):
 106             raise ExtractorError('No video found', expected=True)
 107
 108         manifest_url = video_player['manifest-url']
 109         session_token = self._download_json(
 110             'https://sts.postimees.ee/session/register', video_id, 'Registering session',
 111             'Unable to register session', headers={
 112                 'Accept': 'application/json',
 113                 'X-Original-URI': manifest_url,
 114             })['session']
 115
 116         episode_attr = self._parse_json(video_player.get(':episode') or '', video_id, fatal=False) or {}
 117
 118         return {
 119             'id': video_id,
 120             'formats': self._extract_m3u8_formats(manifest_url, video_id, 'mp4', query={'s': session_token}),
 121             **traverse_obj(episode_attr, {
 122                 'title': ('title', {str}),
 123                 'description': ('synopsis', {str}),
 124                 'thumbnail': ('images', 'original'),
 125                 'timestamp': ('airtime', {lambda x: unified_timestamp(x + ' +0200')}),
 126                 'cast': ('cast', filter, {lambda x: x.split(', ')}),
 127                 'release_year': ('year', {int_or_none}),
 128             }),
 129             **(traverse_obj(episode_attr, {
 130                 'title': (None, (('subtitle', {str}, filter), {value(f'Episode {episode}' if episode else None)})),
 131                 'series': ('title', {str}),
 132                 'series_id': ('telecast_id', {str_or_none}),
 133                 'season_number': ('season_id', {int_or_none}),
 134                 'episode': ('subtitle', {str}, filter),
 135                 'episode_number': ('episode_nr', {int_or_none}),
 136                 'episode_id': ('episode_id', {str_or_none}),
 137             }, get_all=False) if episode_attr.get('category') != 'movies' else {}),
 138         }