X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fspankbang.py;h=e6c2dcfc438b758bf43080799817fb8f6a3e2075;hb=cf7259bc931874318f8abe142a89914b41b49236;hp=50433d0f678f27c348031dbe0d6fcc3774d021b7;hpb=d9e543b68030c8bd24d1b0352938d54312823345;p=youtube-dl diff --git a/youtube_dl/extractor/spankbang.py b/youtube_dl/extractor/spankbang.py index 50433d0f6..e6c2dcfc4 100644 --- a/youtube_dl/extractor/spankbang.py +++ b/youtube_dl/extractor/spankbang.py @@ -3,10 +3,11 @@ from __future__ import unicode_literals import re from .common import InfoExtractor +from ..utils import ExtractorError class SpankBangIE(InfoExtractor): - _VALID_URL = r'https?://(?:(?:www|[a-z]{2})\.)?spankbang\.com/(?P[\da-z]+)/video' + _VALID_URL = r'https?://(?:(?:www|m|[a-z]{2})\.)?spankbang\.com/(?P[\da-z]+)/video' _TESTS = [{ 'url': 'http://spankbang.com/3vvn/video/fantasy+solo', 'md5': '1cc433e1d6aa14bc376535b8679302f7', @@ -14,8 +15,8 @@ class SpankBangIE(InfoExtractor): 'id': '3vvn', 'ext': 'mp4', 'title': 'fantasy solo', - 'description': 'dillion harper masturbates on a bed', - 'thumbnail': 're:^https?://.*\.jpg$', + 'description': 'Watch fantasy solo free HD porn video - 05 minutes - Babe,Masturbation,Solo,Toy - dillion harper masturbates on a bed free adult movies sexy clips.', + 'thumbnail': r're:^https?://.*\.jpg$', 'uploader': 'silly2587', 'age_limit': 18, } @@ -23,12 +24,24 @@ class SpankBangIE(InfoExtractor): # 480p only 'url': 'http://spankbang.com/1vt0/video/solvane+gangbang', 'only_matching': True, + }, { + # no uploader + 'url': 'http://spankbang.com/lklg/video/sex+with+anyone+wedding+edition+2', + 'only_matching': True, + }, { + # mobile page + 'url': 'http://m.spankbang.com/1o2de/video/can+t+remember+her+name', + 'only_matching': True, }] def _real_extract(self, url): video_id = self._match_id(url) webpage = self._download_webpage(url, video_id) + if re.search(r'<[^>]+\bid=["\']video_removed', webpage): + raise ExtractorError( + 'Video %s is not available' % video_id, expected=True) + stream_key = self._html_search_regex( r'''var\s+stream_key\s*=\s*['"](.+?)['"]''', webpage, 'stream key') @@ -44,13 +57,11 @@ class SpankBangIE(InfoExtractor): title = self._html_search_regex( r'(?s)]*>(.+?)', webpage, 'title') - description = self._search_regex( - r'class="desc"[^>]*>([^<]+)', - webpage, 'description', default=None) + description = self._og_search_description(webpage) thumbnail = self._og_search_thumbnail(webpage) uploader = self._search_regex( - r'class="user"[^>]*>([^<]+)', - webpage, 'uploader', fatal=False) + r'class="user"[^>]*>]+>([^<]+)', + webpage, 'uploader', default=None) age_limit = self._rta_search(webpage)