From 4f13f8f798be06bc2b3c0c42818bb0785e4cde64 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Sun, 25 Oct 2015 23:12:12 +0600 Subject: [PATCH] [youporn] Improve uploader extraction --- youtube_dl/extractor/youporn.py | 32 +++++++++++++++++++++++++++----- 1 file changed, 27 insertions(+), 5 deletions(-) diff --git a/youtube_dl/extractor/youporn.py b/youtube_dl/extractor/youporn.py index db5d049d2..b39fbb5fc 100644 --- a/youtube_dl/extractor/youporn.py +++ b/youtube_dl/extractor/youporn.py @@ -15,8 +15,9 @@ from ..aes import aes_decrypt_text class YouPornIE(InfoExtractor): _VALID_URL = r'https?://(?:www\.)?youporn\.com/watch/(?P\d+)/(?P[^/?#&]+)' - _TEST = { + _TESTS = [{ 'url': 'http://www.youporn.com/watch/505835/sex-ed-is-it-safe-to-masturbate-daily/', + 'md5': '71ec5fcfddacf80f495efa8b6a8d9a89', 'info_dict': { 'id': '505835', 'display_id': 'sex-ed-is-it-safe-to-masturbate-daily', @@ -31,8 +32,29 @@ class YouPornIE(InfoExtractor): 'categories': list, 'tags': list, 'age_limit': 18, - } - } + }, + }, { + # Anonymous User uploader + 'url': 'http://www.youporn.com/watch/561726/big-tits-awesome-brunette-on-amazing-webcam-show/?from=related3&al=2&from_id=561726&pos=4', + 'info_dict': { + 'id': '561726', + 'display_id': 'big-tits-awesome-brunette-on-amazing-webcam-show', + 'ext': 'mp4', + 'title': 'Big Tits Awesome Brunette On amazing webcam show', + 'description': 'http://sweetlivegirls.com Big Tits Awesome Brunette On amazing webcam show.mp4', + 'thumbnail': 're:^https?://.*\.jpg$', + 'uploader': 'Anonymous User', + 'upload_date': '20111125', + 'average_rating': int, + 'view_count': int, + 'categories': list, + 'tags': list, + 'age_limit': 18, + }, + 'params': { + 'skip_download': True, + }, + }] def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) @@ -97,8 +119,8 @@ class YouPornIE(InfoExtractor): r'(?:imageurl\s*=|poster\s*:)\s*(["\'])(?P.+?)\1', webpage, 'thumbnail', fatal=False, group='thumbnail') - uploader = self._search_regex( - r']+class=["\']videoInfoBy["\'][^>]*>\s*By:\s*\s*]+href="[^"]*">([^<]+)', + uploader = self._html_search_regex( + r'(?s)]+class=["\']videoInfoBy["\'][^>]*>\s*By:\s*(.+?)', webpage, 'uploader', fatal=False) upload_date = unified_strdate(self._html_search_regex( r'(?s)]+class=["\']videoInfoTime["\'][^>]*>(.+?)',