mirror of
https://github.com/ytdl-org/youtube-dl
synced 2024-11-17 02:52:38 +09:00
82 lines
2.7 KiB
Python
82 lines
2.7 KiB
Python
from __future__ import unicode_literals
|
|
|
|
import re
|
|
|
|
from .common import InfoExtractor
|
|
from ..compat import compat_urllib_parse_unquote
|
|
from ..utils import (
|
|
clean_html,
|
|
determine_ext,
|
|
ExtractorError,
|
|
int_or_none,
|
|
parse_duration,
|
|
)
|
|
|
|
|
|
class XVideosIE(InfoExtractor):
|
|
_VALID_URL = r'https?://(?:www\.)?xvideos\.com/video(?P<id>[0-9]+)(?:.*)'
|
|
_TEST = {
|
|
'url': 'http://www.xvideos.com/video4588838/biker_takes_his_girl',
|
|
'md5': '14cea69fcb84db54293b1e971466c2e1',
|
|
'info_dict': {
|
|
'id': '4588838',
|
|
'ext': 'mp4',
|
|
'title': 'Biker Takes his Girl',
|
|
'duration': 108,
|
|
'age_limit': 18,
|
|
}
|
|
}
|
|
|
|
def _real_extract(self, url):
|
|
video_id = self._match_id(url)
|
|
webpage = self._download_webpage(url, video_id)
|
|
|
|
mobj = re.search(r'<h1 class="inlineError">(.+?)</h1>', webpage)
|
|
if mobj:
|
|
raise ExtractorError('%s said: %s' % (self.IE_NAME, clean_html(mobj.group(1))), expected=True)
|
|
|
|
video_title = self._html_search_regex(
|
|
r'<title>(.*?)\s+-\s+XVID', webpage, 'title')
|
|
video_thumbnail = self._search_regex(
|
|
r'url_bigthumb=(.+?)&', webpage, 'thumbnail', fatal=False)
|
|
video_duration = int_or_none(self._og_search_property(
|
|
'duration', webpage, default=None)) or parse_duration(
|
|
self._search_regex(
|
|
r'<span[^>]+class=["\']duration["\'][^>]*>.*?(\d[^<]+)',
|
|
webpage, 'duration', fatal=False))
|
|
|
|
formats = []
|
|
|
|
video_url = compat_urllib_parse_unquote(self._search_regex(
|
|
r'flv_url=(.+?)&', webpage, 'video URL', default=''))
|
|
if video_url:
|
|
formats.append({
|
|
'url': video_url,
|
|
'format_id': 'flv',
|
|
})
|
|
|
|
for kind, _, format_url in re.findall(
|
|
r'setVideo([^(]+)\((["\'])(http.+?)\2\)', webpage):
|
|
format_id = kind.lower()
|
|
if format_id == 'hls':
|
|
formats.extend(self._extract_m3u8_formats(
|
|
format_url, video_id, 'mp4',
|
|
entry_protocol='m3u8_native', m3u8_id='hls', fatal=False))
|
|
elif format_id in ('urllow', 'urlhigh'):
|
|
formats.append({
|
|
'url': format_url,
|
|
'format_id': '%s-%s' % (determine_ext(format_url, 'mp4'), format_id[3:]),
|
|
'quality': -2 if format_id.endswith('low') else None,
|
|
})
|
|
|
|
self._sort_formats(formats)
|
|
|
|
return {
|
|
'id': video_id,
|
|
'formats': formats,
|
|
'title': video_title,
|
|
'duration': video_duration,
|
|
'thumbnail': video_thumbnail,
|
|
'age_limit': 18,
|
|
}
|