]>
Commit | Line | Data |
---|---|---|
734d461c S |
1 | from .common import InfoExtractor |
2 | from ..utils import ( | |
3 | determine_ext, | |
4 | int_or_none, | |
5 | unescapeHTML, | |
6 | unified_timestamp, | |
7 | ) | |
8 | ||
9 | ||
10 | class ExpressenIE(InfoExtractor): | |
57c68ec4 S |
11 | _VALID_URL = r'''(?x) |
12 | https?:// | |
b7770046 | 13 | (?:www\.)?(?:expressen|di)\.se/ |
a5e264d7 | 14 | (?:(?:tvspelare/video|video-?player/embed)/)? |
15 | (?:tv|nyheter)/(?:[^/?#]+/)* | |
57c68ec4 S |
16 | (?P<id>[^/?#&]+) |
17 | ''' | |
bfd973ec | 18 | _EMBED_REGEX = [r'<iframe[^>]+\bsrc=(["\'])(?P<url>(?:https?:)?//(?:www\.)?(?:expressen|di)\.se/(?:tvspelare/video|videoplayer/embed)/tv/.+?)\1'] |
734d461c S |
19 | _TESTS = [{ |
20 | 'url': 'https://www.expressen.se/tv/ledare/ledarsnack/ledarsnack-om-arbetslosheten-bland-kvinnor-i-speciellt-utsatta-omraden/', | |
0d6bafbf | 21 | 'md5': 'deb2ca62e7b1dcd19fa18ba37523f66e', |
734d461c | 22 | 'info_dict': { |
0d6bafbf DL |
23 | 'id': 'ba90f5a9-78d1-4511-aa02-c177b9c99136', |
24 | 'display_id': 'ledarsnack-om-arbetslosheten-bland-kvinnor-i-speciellt-utsatta-omraden', | |
734d461c S |
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, | |
57c68ec4 S |
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, | |
b7770046 U |
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, | |
a5e264d7 | 45 | }, { |
46 | 'url': 'https://www.expressen.se/video-player/embed/tv/nyheter/ekero-fodda-olof-gustafsson-forvaltar-knarkbaronen-pablo-escobars-namn', | |
47 | 'only_matching': True, | |
48 | }, { | |
49 | 'url': 'https://www.expressen.se/nyheter/efter-egna-telefonbluffen-escobar-stammer-klarna/', | |
50 | 'only_matching': True, | |
734d461c S |
51 | }] |
52 | ||
53 | def _real_extract(self, url): | |
54 | display_id = self._match_id(url) | |
55 | ||
56 | webpage = self._download_webpage(url, display_id) | |
57 | ||
58 | def extract_data(name): | |
59 | return self._parse_json( | |
60 | self._search_regex( | |
61 | r'data-%s=(["\'])(?P<value>(?:(?!\1).)+)\1' % name, | |
62 | webpage, 'info', group='value'), | |
63 | display_id, transform_source=unescapeHTML) | |
64 | ||
65 | info = extract_data('video-tracking-info') | |
0d6bafbf | 66 | video_id = info['contentId'] |
734d461c S |
67 | |
68 | data = extract_data('article-data') | |
69 | stream = data['stream'] | |
70 | ||
71 | if determine_ext(stream) == 'm3u8': | |
72 | formats = self._extract_m3u8_formats( | |
73 | stream, display_id, 'mp4', entry_protocol='m3u8_native', | |
74 | m3u8_id='hls') | |
75 | else: | |
76 | formats = [{ | |
77 | 'url': stream, | |
78 | }] | |
734d461c S |
79 | |
80 | title = info.get('titleRaw') or data['title'] | |
81 | description = info.get('descriptionRaw') | |
82 | thumbnail = info.get('socialMediaImage') or data.get('image') | |
3089bc74 S |
83 | duration = int_or_none(info.get('videoTotalSecondsDuration') |
84 | or data.get('totalSecondsDuration')) | |
734d461c S |
85 | timestamp = unified_timestamp(info.get('publishDate')) |
86 | ||
87 | return { | |
88 | 'id': video_id, | |
89 | 'display_id': display_id, | |
90 | 'title': title, | |
91 | 'description': description, | |
92 | 'thumbnail': thumbnail, | |
93 | 'duration': duration, | |
94 | 'timestamp': timestamp, | |
95 | 'formats': formats, | |
96 | } |