]> jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/yle_areena.py
[extractor/yle_areena] Support restricted videos (#5735)
[yt-dlp.git] / 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 'season': None,
50 'season_number': None,
51 'episode': None,
52 'episode_number': None,
53 'thumbnail': 'http://cfvod.kaltura.com/p/1955031/sp/195503100/thumbnail/entry_id/1_l38iz9ur/version/100021',
54 'uploader_id': 'ovp@yle.fi',
55 'duration': 319,
56 'view_count': int,
57 'upload_date': '20211202',
58 'release_date': '20211215',
59 'timestamp': 1638448202,
60 'subtitles': {},
61 'age_limit': 0,
62 'webpage_url': 'https://areena.yle.fi/1-2158940'
63 }
64 }
65 ]
66
67 def _real_extract(self, url):
68 video_id = self._match_id(url)
69 info = self._search_json_ld(self._download_webpage(url, video_id), video_id, default={})
70 video_data = self._download_json(
71 f'https://player.api.yle.fi/v1/preview/{video_id}.json?app_id=player_static_prod&app_key=8930d72170e48303cf5f3867780d549b',
72 video_id, headers={
73 'origin': 'https://areena.yle.fi',
74 'referer': 'https://areena.yle.fi/',
75 'content-type': 'application/json'
76 })
77
78 # Example title: 'K1, J2: Pouchit | Modernit miehet'
79 series, season_number, episode_number, episode = self._search_regex(
80 r'K(?P<season_no>[\d]+),\s*J(?P<episode_no>[\d]+):?\s*\b(?P<episode>[^|]+)\s*|\s*(?P<series>.+)',
81 info.get('title') or '', 'episode metadata', group=('season_no', 'episode_no', 'episode', 'series'),
82 default=(None, None, None, None))
83 description = traverse_obj(video_data, ('data', 'ongoing_ondemand', 'description', 'fin'), expected_type=str)
84
85 subtitles = {}
86 for sub in traverse_obj(video_data, ('data', 'ongoing_ondemand', 'subtitles', ...)):
87 if url_or_none(sub.get('uri')):
88 subtitles.setdefault(sub.get('language') or 'und', []).append({
89 'url': sub['uri'],
90 'ext': 'srt',
91 'name': sub.get('kind'),
92 })
93
94 return {
95 '_type': 'url_transparent',
96 'url': smuggle_url(
97 f'kaltura:1955031:{video_data["data"]["ongoing_ondemand"]["kaltura"]["id"]}',
98 {'source_url': url}),
99 'ie_key': KalturaIE.ie_key(),
100 'title': (traverse_obj(video_data, ('data', 'ongoing_ondemand', 'title', 'fin'), expected_type=str)
101 or episode or info.get('title')),
102 'description': description,
103 'series': (traverse_obj(video_data, ('data', 'ongoing_ondemand', 'series', 'title', 'fin'), expected_type=str)
104 or series),
105 'season_number': (int_or_none(self._search_regex(r'Kausi (\d+)', description, 'season number', default=None))
106 or int_or_none(season_number)),
107 'episode_number': (traverse_obj(video_data, ('data', 'ongoing_ondemand', 'episode_number'), expected_type=int_or_none)
108 or int_or_none(episode_number)),
109 'thumbnails': traverse_obj(info, ('thumbnails', ..., {'url': 'url'})),
110 'age_limit': traverse_obj(video_data, ('data', 'ongoing_ondemand', 'content_rating', 'age_restriction'), expected_type=int_or_none),
111 'subtitles': subtitles,
112 'release_date': unified_strdate(traverse_obj(video_data, ('data', 'ongoing_ondemand', 'start_time'), expected_type=str)),
113 }