]>
Commit | Line | Data |
---|---|---|
a83333c4 TI |
1 | from .common import InfoExtractor |
2 | from ..utils import js_to_json, traverse_obj | |
3 | ||
4 | ||
5 | class IltalehtiIE(InfoExtractor): | |
6 | _VALID_URL = r'https?://(?:www\.)?iltalehti\.fi/[^/?#]+/a/(?P<id>[^/?#])' | |
7 | _TESTS = [ | |
8 | # jwplatform embed main_media | |
9 | { | |
10 | 'url': 'https://www.iltalehti.fi/ulkomaat/a/9fbd067f-94e4-46cd-8748-9d958eb4dae2', | |
11 | 'md5': 'af12d42c539f1f49f0b62d231fe72dcd', | |
12 | 'info_dict': { | |
13 | 'id': 'gYjjaf1L', | |
14 | 'ext': 'mp4', | |
15 | 'title': 'Sensuroimaton Päivärinta, jakso 227: Vieraana Suomen Venäjän ex-suurlähettiläs René Nyberg ja Kenraalimajuri evp Pekka Toveri', | |
16 | 'description': '', | |
17 | 'upload_date': '20220928', | |
18 | 'timestamp': 1664360878, | |
19 | 'duration': 2089, | |
20 | 'thumbnail': r're:^https?://.*\.jpg', | |
21 | }, | |
22 | }, | |
23 | # jwplatform embed body | |
24 | { | |
25 | 'url': 'https://www.iltalehti.fi/politiikka/a/1ce49d85-1670-428b-8db8-d2479b9950a4', | |
26 | 'md5': '9e50334b8f8330ce8828b567a82a3c65', | |
27 | 'info_dict': { | |
28 | 'id': '18R6zkLi', | |
29 | 'ext': 'mp4', | |
30 | 'title': 'Pekka Toverin arvio: Näin Nord Stream -kaasuputken räjäyttäminen on saatettu toteuttaa', | |
31 | 'description': 'md5:3d1302c9e17e7ffd564143ff58f8de35', | |
32 | 'upload_date': '20220929', | |
33 | 'timestamp': 1664435867, | |
34 | 'duration': 165.0, | |
35 | 'thumbnail': r're:^https?://.*\.jpg', | |
36 | }, | |
37 | }, | |
38 | ] | |
39 | ||
40 | def _real_extract(self, url): | |
41 | article_id = self._match_id(url) | |
42 | webpage = self._download_webpage(url, article_id) | |
43 | info = self._search_json( | |
304ad45a | 44 | r'<script>\s*window.App\s*=', webpage, 'json', article_id, |
a83333c4 TI |
45 | transform_source=js_to_json) |
46 | props = traverse_obj(info, ( | |
47 | 'state', 'articles', ..., 'items', (('main_media', 'properties'), ('body', ..., 'properties')))) | |
48 | video_ids = traverse_obj(props, (lambda _, v: v['provider'] == 'jwplayer', 'id')) | |
49 | return self.playlist_from_matches( | |
50 | video_ids, article_id, ie='JWPlatform', getter=lambda id: f'jwplatform:{id}', | |
51 | title=traverse_obj(info, ('state', 'articles', ..., 'items', 'canonical_title'), get_all=False)) |