X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/ced7488f6d3a519b2c1b1cbd31048743fb8285bd..0bec2c9f91a3b4f834e00c8ea4790481e257d0a2:/youtube_dl/extractor/drtuber.py diff --git a/youtube_dl/extractor/drtuber.py b/youtube_dl/extractor/drtuber.py index e8870c4..5c41c80 100644 --- a/youtube_dl/extractor/drtuber.py +++ b/youtube_dl/extractor/drtuber.py @@ -10,8 +10,8 @@ from ..utils import ( class DrTuberIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?drtuber\.com/video/(?P\d+)/(?P[\w-]+)' - _TEST = { + _VALID_URL = r'https?://(?:(?:www|m)\.)?drtuber\.com/(?:video|embed)/(?P\d+)(?:/(?P[\w-]+))?' + _TESTS = [{ 'url': 'http://www.drtuber.com/video/1740434/hot-perky-blonde-naked-golf', 'md5': '93e680cf2536ad0dfb7e74d94a89facd', 'info_dict': { @@ -22,23 +22,53 @@ class DrTuberIE(InfoExtractor): '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']+?src=["\'](?P(?: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']*>

([^<]+)<', + (r']+class=["\']title[^>]+>([^<]+)', + r'([^<]+)\s*@\s+DrTuber', + r'class="title_watch"[^>]*><(?:p|h\d+)[^>]*>([^<]+)<', r'<p[^>]+class="title_substrate">([^<]+)</p>', r'<title>([^<]+) - \d+'), webpage, 'title') @@ -65,7 +95,7 @@ class DrTuberIE(InfoExtractor): return { 'id': video_id, 'display_id': display_id, - 'url': video_url, + 'formats': formats, 'title': title, 'thumbnail': thumbnail, 'like_count': like_count,