]> jfr.im git - yt-dlp.git/blame - yt_dlp/extractor/drbonanza.py
[extractor] Common function `_match_valid_url`
[yt-dlp.git] / yt_dlp / extractor / drbonanza.py
CommitLineData
8e7a9016
JJ
1from __future__ import unicode_literals
2
8e7a9016 3
8e2ec955
PH
4from .common import InfoExtractor
5from ..utils import (
87b5184a
S
6 js_to_json,
7 parse_duration,
8 unescapeHTML,
8e2ec955
PH
9)
10
11
8e7a9016 12class DRBonanzaIE(InfoExtractor):
87b5184a
S
13 _VALID_URL = r'https?://(?:www\.)?dr\.dk/bonanza/[^/]+/\d+/[^/]+/(?P<id>\d+)/(?P<display_id>[^/?#&]+)'
14 _TEST = {
15 'url': 'http://www.dr.dk/bonanza/serie/154/matador/40312/matador---0824-komme-fremmede-',
8e7a9016 16 'info_dict': {
87b5184a
S
17 'id': '40312',
18 'display_id': 'matador---0824-komme-fremmede-',
8e7a9016 19 'ext': 'mp4',
87b5184a
S
20 'title': 'MATADOR - 08:24. "Komme fremmede".',
21 'description': 'md5:77b4c1ac4d4c1b9d610ab4395212ff84',
ec85ded8 22 'thumbnail': r're:^https?://.*\.(?:gif|jpg)$',
87b5184a 23 'duration': 4613,
8e7a9016 24 },
87b5184a 25 }
8e7a9016
JJ
26
27 def _real_extract(self, url):
5ad28e7f 28 mobj = self._match_valid_url(url)
87b5184a 29 video_id, display_id = mobj.group('id', 'display_id')
8e2ec955 30
87b5184a 31 webpage = self._download_webpage(url, display_id)
8e2ec955 32
87b5184a
S
33 info = self._parse_html5_media_entries(
34 url, webpage, display_id, m3u8_id='hls',
35 m3u8_entry_protocol='m3u8_native')[0]
36 self._sort_formats(info['formats'])
8e2ec955 37
87b5184a
S
38 asset = self._parse_json(
39 self._search_regex(
40 r'(?s)currentAsset\s*=\s*({.+?})\s*</script', webpage, 'asset'),
41 display_id, transform_source=js_to_json)
8e7a9016 42
87b5184a 43 title = unescapeHTML(asset['AssetTitle']).strip()
8e2ec955 44
87b5184a
S
45 def extract(field):
46 return self._search_regex(
47 r'<div[^>]+>\s*<p>%s:<p>\s*</div>\s*<div[^>]+>\s*<p>([^<]+)</p>' % field,
48 webpage, field, default=None)
8e2ec955 49
87b5184a
S
50 info.update({
51 'id': asset.get('AssetId') or video_id,
8e7a9016
JJ
52 'display_id': display_id,
53 'title': title,
87b5184a
S
54 'description': extract('Programinfo'),
55 'duration': parse_duration(extract('Tid')),
56 'thumbnail': asset.get('AssetImageUrl'),
57 })
58 return info