]> jfr.im git - yt-dlp.git/blame - yt_dlp/extractor/golem.py
[cleanup] Add more ruff rules (#10149)
[yt-dlp.git] / yt_dlp / extractor / golem.py
CommitLineData
add96eb9 1import urllib.parse
2
6a5af6ac 3from .common import InfoExtractor
1cc79574 4from ..utils import (
b14f3a4c
PH
5 determine_ext,
6)
6a5af6ac
M
7
8
9class GolemIE(InfoExtractor):
10 _VALID_URL = r'^https?://video\.golem\.de/.+?/(?P<id>.+?)/'
11 _TEST = {
12 'url': 'http://video.golem.de/handy/14095/iphone-6-und-6-plus-test.html',
13 'md5': 'c1a2c0a3c863319651c7c992c5ee29bf',
14 'info_dict': {
15 'id': '14095',
16 'format_id': 'high',
17 'ext': 'mp4',
18 'title': 'iPhone 6 und 6 Plus - Test',
b14f3a4c 19 'duration': 300.44,
6a5af6ac 20 'filesize': 65309548,
add96eb9 21 },
6a5af6ac
M
22 }
23
6a5af6ac
M
24 _PREFIX = 'http://video.golem.de'
25
6a5af6ac 26 def _real_extract(self, url):
b14f3a4c 27 video_id = self._match_id(url)
6a5af6ac 28
b14f3a4c 29 config = self._download_xml(
add96eb9 30 f'https://video.golem.de/xml/{video_id}.xml', video_id)
6a5af6ac
M
31
32 info = {
b14f3a4c
PH
33 'id': video_id,
34 'title': config.findtext('./title', 'golem'),
35 'duration': self._float(config.findtext('./playtime'), 'duration'),
6a5af6ac
M
36 }
37
38 formats = []
8157ae39 39 for e in config:
b14f3a4c
PH
40 url = e.findtext('./url')
41 if not url:
b14f3a4c
PH
42 continue
43
44 formats.append({
add96eb9 45 'format_id': str(e.tag),
46 'url': urllib.parse.urljoin(self._PREFIX, url),
b14f3a4c
PH
47 'height': self._int(e.get('height'), 'height'),
48 'width': self._int(e.get('width'), 'width'),
49 'filesize': self._int(e.findtext('filesize'), 'filesize'),
50 'ext': determine_ext(e.findtext('./filename')),
51 })
6a5af6ac
M
52 info['formats'] = formats
53
54 thumbnails = []
8157ae39 55 for e in config.findall('.//teaser'):
b14f3a4c
PH
56 url = e.findtext('./url')
57 if not url:
58 continue
59 thumbnails.append({
add96eb9 60 'url': urllib.parse.urljoin(self._PREFIX, url),
b14f3a4c
PH
61 'width': self._int(e.get('width'), 'thumbnail width'),
62 'height': self._int(e.get('height'), 'thumbnail height'),
63 })
6a5af6ac
M
64 info['thumbnails'] = thumbnails
65
6a5af6ac 66 return info