yt_dlp/extractor/yle_areena.py

   1 from .common import InfoExtractor
   2 from .kaltura import KalturaIE
   3 from ..utils import (
   4     int_or_none,
   5     smuggle_url,
   6     traverse_obj,
   7     unified_strdate,
   8     url_or_none,
   9 )
  10
  11
  12 class YleAreenaIE(InfoExtractor):
  13     _VALID_URL = r'https?://areena\.yle\.fi/(?P<id>[\d-]+)'
  14     _TESTS = [
  15         {
  16             'url': 'https://areena.yle.fi/1-4371942',
  17             'md5': '932edda0ecf5dfd6423804182d32f8ac',
  18             'info_dict': {
  19                 'id': '0_a3tjk92c',
  20                 'ext': 'mp4',
  21                 'title': 'Pouchit',
  22                 'description': 'md5:d487309c3abbe5650265bbd1742d2f82',
  23                 'series': 'Modernit miehet',
  24                 'season': 'Season 1',
  25                 'season_number': 1,
  26                 'episode': 'Episode 2',
  27                 'episode_number': 2,
  28                 'thumbnail': 'http://cfvod.kaltura.com/p/1955031/sp/195503100/thumbnail/entry_id/0_a3tjk92c/version/100061',
  29                 'uploader_id': 'ovp@yle.fi',
  30                 'duration': 1435,
  31                 'view_count': int,
  32                 'upload_date': '20181204',
  33                 'release_date': '20190106',
  34                 'timestamp': 1543916210,
  35                 'subtitles': {'fin': [{'url': r're:^https?://', 'ext': 'srt'}]},
  36                 'age_limit': 7,
  37                 'webpage_url': 'https://areena.yle.fi/1-4371942'
  38             }
  39         },
  40         {
  41             'url': 'https://areena.yle.fi/1-2158940',
  42             'md5': 'cecb603661004e36af8c5188b5212b12',
  43             'info_dict': {
  44                 'id': '1_l38iz9ur',
  45                 'ext': 'mp4',
  46                 'title': 'Albi haluaa vessan',
  47                 'description': 'md5:15236d810c837bed861fae0e88663c33',
  48                 'series': 'Albi Lumiukko',
  49                 'thumbnail': 'http://cfvod.kaltura.com/p/1955031/sp/195503100/thumbnail/entry_id/1_l38iz9ur/version/100021',
  50                 'uploader_id': 'ovp@yle.fi',
  51                 'duration': 319,
  52                 'view_count': int,
  53                 'upload_date': '20211202',
  54                 'release_date': '20211215',
  55                 'timestamp': 1638448202,
  56                 'subtitles': {},
  57                 'age_limit': 0,
  58                 'webpage_url': 'https://areena.yle.fi/1-2158940'
  59             }
  60         },
  61         {
  62             'url': 'https://areena.yle.fi/1-64829589',
  63             'info_dict': {
  64                 'id': '1-64829589',
  65                 'ext': 'mp4',
  66                 'title': 'HKO & Mälkki & Tanner',
  67                 'description': 'md5:b4f1b1af2c6569b33f75179a86eea156',
  68                 'series': 'Helsingin kaupunginorkesterin konsertteja',
  69                 'thumbnail': r're:^https?://.+\.jpg$',
  70                 'release_date': '20230120',
  71             },
  72             'params': {
  73                 'skip_download': 'm3u8',
  74             },
  75         },
  76     ]
  77
  78     def _real_extract(self, url):
  79         video_id = self._match_id(url)
  80         info = self._search_json_ld(self._download_webpage(url, video_id), video_id, default={})
  81         video_data = self._download_json(
  82             f'https://player.api.yle.fi/v1/preview/{video_id}.json?app_id=player_static_prod&app_key=8930d72170e48303cf5f3867780d549b',
  83             video_id, headers={
  84                 'origin': 'https://areena.yle.fi',
  85                 'referer': 'https://areena.yle.fi/',
  86                 'content-type': 'application/json'
  87             })
  88
  89         # Example title: 'K1, J2: Pouchit | Modernit miehet'
  90         series, season_number, episode_number, episode = self._search_regex(
  91             r'K(?P<season_no>[\d]+),\s*J(?P<episode_no>[\d]+):?\s*\b(?P<episode>[^|]+)\s*|\s*(?P<series>.+)',
  92             info.get('title') or '', 'episode metadata', group=('season_no', 'episode_no', 'episode', 'series'),
  93             default=(None, None, None, None))
  94         description = traverse_obj(video_data, ('data', 'ongoing_ondemand', 'description', 'fin'), expected_type=str)
  95
  96         subtitles = {}
  97         for sub in traverse_obj(video_data, ('data', 'ongoing_ondemand', 'subtitles', ...)):
  98             if url_or_none(sub.get('uri')):
  99                 subtitles.setdefault(sub.get('language') or 'und', []).append({
 100                     'url': sub['uri'],
 101                     'ext': 'srt',
 102                     'name': sub.get('kind'),
 103                 })
 104
 105         kaltura_id = traverse_obj(video_data, ('data', 'ongoing_ondemand', 'kaltura', 'id'), expected_type=str)
 106         if kaltura_id:
 107             info_dict = {
 108                 '_type': 'url_transparent',
 109                 'url': smuggle_url(f'kaltura:1955031:{kaltura_id}', {'source_url': url}),
 110                 'ie_key': KalturaIE.ie_key(),
 111             }
 112         else:
 113             info_dict = {
 114                 'id': video_id,
 115                 'formats': self._extract_m3u8_formats(
 116                     video_data['data']['ongoing_ondemand']['manifest_url'], video_id, 'mp4', m3u8_id='hls'),
 117             }
 118
 119         return {
 120             **info_dict,
 121             'title': (traverse_obj(video_data, ('data', 'ongoing_ondemand', 'title', 'fin'), expected_type=str)
 122                       or episode or info.get('title')),
 123             'description': description,
 124             'series': (traverse_obj(video_data, ('data', 'ongoing_ondemand', 'series', 'title', 'fin'), expected_type=str)
 125                        or series),
 126             'season_number': (int_or_none(self._search_regex(r'Kausi (\d+)', description, 'season number', default=None))
 127                               or int_or_none(season_number)),
 128             'episode_number': (traverse_obj(video_data, ('data', 'ongoing_ondemand', 'episode_number'), expected_type=int_or_none)
 129                                or int_or_none(episode_number)),
 130             'thumbnails': traverse_obj(info, ('thumbnails', ..., {'url': 'url'})),
 131             'age_limit': traverse_obj(video_data, ('data', 'ongoing_ondemand', 'content_rating', 'age_restriction'), expected_type=int_or_none),
 132             'subtitles': subtitles,
 133             'release_date': unified_strdate(traverse_obj(video_data, ('data', 'ongoing_ondemand', 'start_time'), expected_type=str)),
 134         }