1
0
mirror of https://github.com/l1ving/youtube-dl synced 2024-11-25 13:55:29 +08:00
youtube-dl/youtube_dl/extractor/extremetube.py

76 lines
2.5 KiB
Python
Raw Normal View History

2014-04-20 01:42:51 +08:00
from __future__ import unicode_literals
import re
from .common import InfoExtractor
from ..compat import (
compat_parse_qs,
compat_urllib_request,
)
from ..utils import (
qualities,
2014-09-27 23:36:53 +08:00
str_to_int,
)
2014-04-20 01:42:51 +08:00
class ExtremeTubeIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?(?P<url>extremetube\.com/.*?video/.+?(?P<id>[0-9]+))(?:[/?&]|$)'
_TESTS = [{
2014-04-20 01:42:51 +08:00
'url': 'http://www.extremetube.com/video/music-video-14-british-euro-brit-european-cumshots-swallow-652431',
'md5': '344d0c6d50e2f16b06e49ca011d8ac69',
2014-04-20 01:42:51 +08:00
'info_dict': {
'id': '652431',
'ext': 'mp4',
'title': 'Music Video 14 british euro brit european cumshots swallow',
'uploader': 'unknown',
2014-09-27 23:36:53 +08:00
'view_count': int,
2014-04-20 01:42:51 +08:00
'age_limit': 18,
}
}, {
'url': 'http://www.extremetube.com/gay/video/abcde-1234',
'only_matching': True,
}]
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('id')
url = 'http://www.' + mobj.group('url')
req = compat_urllib_request.Request(url)
req.add_header('Cookie', 'age_verified=1')
webpage = self._download_webpage(req, video_id)
2014-04-20 01:42:51 +08:00
video_title = self._html_search_regex(
r'<h1 [^>]*?title="([^"]+)"[^>]*>', webpage, 'title')
2014-04-20 01:42:51 +08:00
uploader = self._html_search_regex(
2014-09-27 23:36:53 +08:00
r'Uploaded by:\s*</strong>\s*(.+?)\s*</div>',
webpage, 'uploader', fatal=False)
view_count = str_to_int(self._html_search_regex(
r'Views:\s*</strong>\s*<span>([\d,\.]+)</span>',
webpage, 'view count', fatal=False))
flash_vars = compat_parse_qs(self._search_regex(
r'<param[^>]+?name="flashvars"[^>]+?value="([^"]+)"', webpage, 'flash vars'))
formats = []
quality = qualities(['180p', '240p', '360p', '480p', '720p', '1080p'])
for k, vals in flash_vars.items():
m = re.match(r'quality_(?P<quality>[0-9]+p)$', k)
if m is not None:
formats.append({
'format_id': m.group('quality'),
'quality': quality(m.group('quality')),
'url': vals[0],
})
self._sort_formats(formats)
return {
'id': video_id,
'title': video_title,
'formats': formats,
'uploader': uploader,
2014-09-27 23:36:53 +08:00
'view_count': view_count,
'age_limit': 18,
}