]> jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/expressen.py
[spotify] Detect iframe embeds (#3430)
[yt-dlp.git] / yt_dlp / extractor / expressen.py
1 import re
2
3 from .common import InfoExtractor
4 from ..utils import (
5 determine_ext,
6 int_or_none,
7 unescapeHTML,
8 unified_timestamp,
9 )
10
11
12 class ExpressenIE(InfoExtractor):
13 _VALID_URL = r'''(?x)
14 https?://
15 (?:www\.)?(?:expressen|di)\.se/
16 (?:(?:tvspelare/video|videoplayer/embed)/)?
17 tv/(?:[^/]+/)*
18 (?P<id>[^/?#&]+)
19 '''
20 _TESTS = [{
21 'url': 'https://www.expressen.se/tv/ledare/ledarsnack/ledarsnack-om-arbetslosheten-bland-kvinnor-i-speciellt-utsatta-omraden/',
22 'md5': '2fbbe3ca14392a6b1b36941858d33a45',
23 'info_dict': {
24 'id': '8690962',
25 'ext': 'mp4',
26 'title': 'Ledarsnack: Om arbetslösheten bland kvinnor i speciellt utsatta områden',
27 'description': 'md5:f38c81ff69f3de4d269bbda012fcbbba',
28 'thumbnail': r're:^https?://.*\.jpg$',
29 'duration': 788,
30 'timestamp': 1526639109,
31 'upload_date': '20180518',
32 },
33 }, {
34 'url': 'https://www.expressen.se/tv/kultur/kulturdebatt-med-expressens-karin-olsson/',
35 'only_matching': True,
36 }, {
37 'url': 'https://www.expressen.se/tvspelare/video/tv/ditv/ekonomistudion/experterna-har-ar-fragorna-som-avgor-valet/?embed=true&external=true&autoplay=true&startVolume=0&partnerId=di',
38 'only_matching': True,
39 }, {
40 'url': 'https://www.expressen.se/videoplayer/embed/tv/ditv/ekonomistudion/experterna-har-ar-fragorna-som-avgor-valet/?embed=true&external=true&autoplay=true&startVolume=0&partnerId=di',
41 'only_matching': True,
42 }, {
43 'url': 'https://www.di.se/videoplayer/embed/tv/ditv/borsmorgon/implantica-rusar-70--under-borspremiaren-hor-styrelsemedlemmen/?embed=true&external=true&autoplay=true&startVolume=0&partnerId=di',
44 'only_matching': True,
45 }]
46
47 @staticmethod
48 def _extract_urls(webpage):
49 return [
50 mobj.group('url') for mobj in re.finditer(
51 r'<iframe[^>]+\bsrc=(["\'])(?P<url>(?:https?:)?//(?:www\.)?(?:expressen|di)\.se/(?:tvspelare/video|videoplayer/embed)/tv/.+?)\1',
52 webpage)]
53
54 def _real_extract(self, url):
55 display_id = self._match_id(url)
56
57 webpage = self._download_webpage(url, display_id)
58
59 def extract_data(name):
60 return self._parse_json(
61 self._search_regex(
62 r'data-%s=(["\'])(?P<value>(?:(?!\1).)+)\1' % name,
63 webpage, 'info', group='value'),
64 display_id, transform_source=unescapeHTML)
65
66 info = extract_data('video-tracking-info')
67 video_id = info['videoId']
68
69 data = extract_data('article-data')
70 stream = data['stream']
71
72 if determine_ext(stream) == 'm3u8':
73 formats = self._extract_m3u8_formats(
74 stream, display_id, 'mp4', entry_protocol='m3u8_native',
75 m3u8_id='hls')
76 else:
77 formats = [{
78 'url': stream,
79 }]
80 self._sort_formats(formats)
81
82 title = info.get('titleRaw') or data['title']
83 description = info.get('descriptionRaw')
84 thumbnail = info.get('socialMediaImage') or data.get('image')
85 duration = int_or_none(info.get('videoTotalSecondsDuration')
86 or data.get('totalSecondsDuration'))
87 timestamp = unified_timestamp(info.get('publishDate'))
88
89 return {
90 'id': video_id,
91 'display_id': display_id,
92 'title': title,
93 'description': description,
94 'thumbnail': thumbnail,
95 'duration': duration,
96 'timestamp': timestamp,
97 'formats': formats,
98 }