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     _GEO_COUNTRIES = ['FI']
  15     _TESTS = [
  16         {
  17             'url': 'https://areena.yle.fi/1-4371942',
  18             'md5': '932edda0ecf5dfd6423804182d32f8ac',
  19             'info_dict': {
  20                 'id': '0_a3tjk92c',
  21                 'ext': 'mp4',
  22                 'title': 'Pouchit',
  23                 'description': 'md5:01071d7056ceec375f63960f90c35366',
  24                 'series': 'Modernit miehet',
  25                 'season': 'Season 1',
  26                 'season_number': 1,
  27                 'episode': 'Episode 2',
  28                 'episode_number': 2,
  29                 'thumbnail': 'http://cfvod.kaltura.com/p/1955031/sp/195503100/thumbnail/entry_id/0_a3tjk92c/version/100061',
  30                 'uploader_id': 'ovp@yle.fi',
  31                 'duration': 1435,
  32                 'view_count': int,
  33                 'upload_date': '20181204',
  34                 'release_date': '20190106',
  35                 'timestamp': 1543916210,
  36                 'subtitles': {'fin': [{'url': r're:^https?://', 'ext': 'srt'}]},
  37                 'age_limit': 7,
  38                 'webpage_url': 'https://areena.yle.fi/1-4371942',
  39             },
  40         },
  41         {
  42             'url': 'https://areena.yle.fi/1-2158940',
  43             'md5': 'cecb603661004e36af8c5188b5212b12',
  44             'info_dict': {
  45                 'id': '1_l38iz9ur',
  46                 'ext': 'mp4',
  47                 'title': 'Albi haluaa vessan',
  48                 'description': 'md5:15236d810c837bed861fae0e88663c33',
  49                 'series': 'Albi Lumiukko',
  50                 'thumbnail': 'http://cfvod.kaltura.com/p/1955031/sp/195503100/thumbnail/entry_id/1_l38iz9ur/version/100021',
  51                 'uploader_id': 'ovp@yle.fi',
  52                 'duration': 319,
  53                 'view_count': int,
  54                 'upload_date': '20211202',
  55                 'release_date': '20211215',
  56                 'timestamp': 1638448202,
  57                 'subtitles': {},
  58                 'age_limit': 0,
  59                 'webpage_url': 'https://areena.yle.fi/1-2158940',
  60             },
  61         },
  62         {
  63             'url': 'https://areena.yle.fi/1-64829589',
  64             'info_dict': {
  65                 'id': '1-64829589',
  66                 'ext': 'mp4',
  67                 'title': 'HKO & Mälkki & Tanner',
  68                 'description': 'md5:b4f1b1af2c6569b33f75179a86eea156',
  69                 'series': 'Helsingin kaupunginorkesterin konsertteja',
  70                 'thumbnail': r're:^https?://.+\.jpg$',
  71                 'release_date': '20230120',
  72             },
  73             'params': {
  74                 'skip_download': 'm3u8',
  75             },
  76         },
  77     ]
  78
  79     def _real_extract(self, url):
  80         video_id = self._match_id(url)
  81         info = self._search_json_ld(self._download_webpage(url, video_id), video_id, default={})
  82         video_data = self._download_json(
  83             f'https://player.api.yle.fi/v1/preview/{video_id}.json?app_id=player_static_prod&app_key=8930d72170e48303cf5f3867780d549b',
  84             video_id, headers={
  85                 'origin': 'https://areena.yle.fi',
  86                 'referer': 'https://areena.yle.fi/',
  87                 'content-type': 'application/json',
  88             })
  89
  90         # Example title: 'K1, J2: Pouchit | Modernit miehet'
  91         season_number, episode_number, episode, series = self._search_regex(
  92             r'K(?P<season_no>\d+),\s*J(?P<episode_no>\d+):?\s*\b(?P<episode>[^|]+)\s*|\s*(?P<series>.+)',
  93             info.get('title') or '', 'episode metadata', group=('season_no', 'episode_no', 'episode', 'series'),
  94             default=(None, None, None, None))
  95         description = traverse_obj(video_data, ('data', 'ongoing_ondemand', 'description', 'fin'), expected_type=str)
  96
  97         subtitles = {}
  98         for sub in traverse_obj(video_data, ('data', 'ongoing_ondemand', 'subtitles', ...)):
  99             if url_or_none(sub.get('uri')):
 100                 subtitles.setdefault(sub.get('language') or 'und', []).append({
 101                     'url': sub['uri'],
 102                     'ext': 'srt',
 103                     'name': sub.get('kind'),
 104                 })
 105
 106         kaltura_id = traverse_obj(video_data, ('data', 'ongoing_ondemand', 'kaltura', 'id'), expected_type=str)
 107         if kaltura_id:
 108             info_dict = {
 109                 '_type': 'url_transparent',
 110                 'url': smuggle_url(f'kaltura:1955031:{kaltura_id}', {'source_url': url}),
 111                 'ie_key': KalturaIE.ie_key(),
 112             }
 113         else:
 114             formats, subs = self._extract_m3u8_formats_and_subtitles(
 115                 video_data['data']['ongoing_ondemand']['manifest_url'], video_id, 'mp4', m3u8_id='hls')
 116             self._merge_subtitles(subs, target=subtitles)
 117             info_dict = {
 118                 'id': video_id,
 119                 'formats': formats,
 120             }
 121
 122         return {
 123             **info_dict,
 124             'title': (traverse_obj(video_data, ('data', 'ongoing_ondemand', 'title', 'fin'), expected_type=str)
 125                       or episode or info.get('title')),
 126             'description': description,
 127             'series': (traverse_obj(video_data, ('data', 'ongoing_ondemand', 'series', 'title', 'fin'), expected_type=str)
 128                        or series),
 129             'season_number': (int_or_none(self._search_regex(r'Kausi (\d+)', description, 'season number', default=None))
 130                               or int_or_none(season_number)),
 131             'episode_number': (traverse_obj(video_data, ('data', 'ongoing_ondemand', 'episode_number'), expected_type=int_or_none)
 132                                or int_or_none(episode_number)),
 133             'thumbnails': traverse_obj(info, ('thumbnails', ..., {'url': 'url'})),
 134             'age_limit': traverse_obj(video_data, ('data', 'ongoing_ondemand', 'content_rating', 'age_restriction'), expected_type=int_or_none),
 135             'subtitles': subtitles or None,
 136             'release_date': unified_strdate(traverse_obj(video_data, ('data', 'ongoing_ondemand', 'start_time'), expected_type=str)),
 137         }