yt-dlp/yt_dlp/extractor/golem.py
bashonly c6387abc1a
[cleanup] Misc (#10807)
Closes #10751, Closes #10769, Closes #10791
Authored by: bashonly, Codenade, pzhlkj6612, seproDev, coletdjnz, grqz, Grub4K

Co-authored-by: Codenade <amadeus.dorian04@gmail.com>
Co-authored-by: Mozi <29089388+pzhlkj6612@users.noreply.github.com>
Co-authored-by: sepro <4618135+seproDev@users.noreply.github.com>
Co-authored-by: coletdjnz <coletdjnz@protonmail.com>
Co-authored-by: N/Ame <173015200+grqz@users.noreply.github.com>
Co-authored-by: Simon Sawicki <contact@grub4k.xyz>
2024-09-27 22:46:22 +00:00

66 lines
2 KiB
Python

import urllib.parse
from .common import InfoExtractor
from ..utils import (
determine_ext,
)
class GolemIE(InfoExtractor):
_VALID_URL = r'https?://video\.golem\.de/.+?/(?P<id>.+?)/'
_TEST = {
'url': 'http://video.golem.de/handy/14095/iphone-6-und-6-plus-test.html',
'md5': 'c1a2c0a3c863319651c7c992c5ee29bf',
'info_dict': {
'id': '14095',
'format_id': 'high',
'ext': 'mp4',
'title': 'iPhone 6 und 6 Plus - Test',
'duration': 300.44,
'filesize': 65309548,
},
}
_PREFIX = 'http://video.golem.de'
def _real_extract(self, url):
video_id = self._match_id(url)
config = self._download_xml(
f'https://video.golem.de/xml/{video_id}.xml', video_id)
info = {
'id': video_id,
'title': config.findtext('./title', 'golem'),
'duration': self._float(config.findtext('./playtime'), 'duration'),
}
formats = []
for e in config:
url = e.findtext('./url')
if not url:
continue
formats.append({
'format_id': str(e.tag),
'url': urllib.parse.urljoin(self._PREFIX, url),
'height': self._int(e.get('height'), 'height'),
'width': self._int(e.get('width'), 'width'),
'filesize': self._int(e.findtext('filesize'), 'filesize'),
'ext': determine_ext(e.findtext('./filename')),
})
info['formats'] = formats
thumbnails = []
for e in config.findall('.//teaser'):
url = e.findtext('./url')
if not url:
continue
thumbnails.append({
'url': urllib.parse.urljoin(self._PREFIX, url),
'width': self._int(e.get('width'), 'thumbnail width'),
'height': self._int(e.get('height'), 'thumbnail height'),
})
info['thumbnails'] = thumbnails
return info