class DrTuberIE(InfoExtractor):
- _VALID_URL = r'https?://(?:www\.)?drtuber\.com/video/(?P<id>\d+)/(?P<display_id>[\w-]+)'
- _TEST = {
+ _VALID_URL = r'https?://(?:(?:www|m)\.)?drtuber\.com/(?:video|embed)/(?P<id>\d+)(?:/(?P<display_id>[\w-]+))?'
+ _TESTS = [{
'url': 'http://www.drtuber.com/video/1740434/hot-perky-blonde-naked-golf',
'md5': '93e680cf2536ad0dfb7e74d94a89facd',
'info_dict': {
'like_count': int,
'comment_count': int,
'categories': ['Babe', 'Blonde', 'Erotic', 'Outdoor', 'Softcore', 'Solo'],
- 'thumbnail': 're:https?://.*\.jpg$',
+ 'thumbnail': r're:https?://.*\.jpg$',
'age_limit': 18,
}
- }
+ }, {
+ 'url': 'http://www.drtuber.com/embed/489939',
+ 'only_matching': True,
+ }, {
+ 'url': 'http://m.drtuber.com/video/3893529/lingerie-blowjob-from-beautiful-teen',
+ 'only_matching': True,
+ }]
+
+ @staticmethod
+ def _extract_urls(webpage):
+ return re.findall(
+ r'<iframe[^>]+?src=["\'](?P<url>(?:https?:)?//(?:www\.)?drtuber\.com/embed/\d+)',
+ webpage)
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('id')
- display_id = mobj.group('display_id')
+ display_id = mobj.group('display_id') or video_id
+
+ webpage = self._download_webpage(
+ 'http://www.drtuber.com/video/%s' % video_id, display_id)
- webpage = self._download_webpage(url, display_id)
+ video_data = self._download_json(
+ 'http://www.drtuber.com/player_config_json/', video_id, query={
+ 'vid': video_id,
+ 'embed': 0,
+ 'aid': 0,
+ 'domain_id': 0,
+ })
- video_url = self._html_search_regex(
- r'<source src="([^"]+)"', webpage, 'video URL')
+ formats = []
+ for format_id, video_url in video_data['files'].items():
+ if video_url:
+ formats.append({
+ 'format_id': format_id,
+ 'quality': 2 if format_id == 'hq' else 1,
+ 'url': video_url
+ })
+ self._sort_formats(formats)
title = self._html_search_regex(
- (r'class="title_watch"[^>]*><p>([^<]+)<',
+ (r'class="title_watch"[^>]*><(?:p|h\d+)[^>]*>([^<]+)<',
r'<p[^>]+class="title_substrate">([^<]+)</p>',
r'<title>([^<]+) - \d+'),
webpage, 'title')
return {
'id': video_id,
'display_id': display_id,
- 'url': video_url,
+ 'formats': formats,
'title': title,
'thumbnail': thumbnail,
'like_count': like_count,