[youtube] Fix extraction (closes #17457, closes #17464)
[youtube-dl] / youtube_dl / extractor / youporn.py
index 64e3c3c9405ac158cd4d5468a1897027044e8a60..ea0bce784c5fbe91d904428670377d3e92414453 100644 (file)
@@ -3,13 +3,13 @@ from __future__ import unicode_literals
 import re
 
 from .common import InfoExtractor
-from ..compat import compat_str
 from ..utils import (
     int_or_none,
     sanitized_Request,
     str_to_int,
     unescapeHTML,
     unified_strdate,
+    url_or_none,
 )
 from ..aes import aes_decrypt_text
 
@@ -27,7 +27,7 @@ class YouPornIE(InfoExtractor):
             'description': 'Love & Sex Answers: http://bit.ly/DanAndJenn -- Is It Unhealthy To Masturbate Daily?',
             'thumbnail': r're:^https?://.*\.jpg$',
             'uploader': 'Ask Dan And Jennifer',
-            'upload_date': '20101221',
+            'upload_date': '20101217',
             'average_rating': int,
             'view_count': int,
             'comment_count': int,
@@ -46,7 +46,7 @@ class YouPornIE(InfoExtractor):
             'description': 'http://sweetlivegirls.com Big Tits Awesome Brunette On amazing webcam show.mp4',
             'thumbnail': r're:^https?://.*\.jpg$',
             'uploader': 'Unknown',
-            'upload_date': '20111125',
+            'upload_date': '20110418',
             'average_rating': int,
             'view_count': int,
             'comment_count': int,
@@ -88,8 +88,8 @@ class YouPornIE(InfoExtractor):
             for definition in definitions:
                 if not isinstance(definition, dict):
                     continue
-                video_url = definition.get('videoUrl')
-                if isinstance(video_url, compat_str) and video_url:
+                video_url = url_or_none(definition.get('videoUrl'))
+                if video_url:
                     links.append(video_url)
 
         # Fallback #1, this also contains extra low quality 180p format
@@ -143,7 +143,8 @@ class YouPornIE(InfoExtractor):
             r'(?s)<div[^>]+class=["\']submitByLink["\'][^>]*>(.+?)</div>',
             webpage, 'uploader', fatal=False)
         upload_date = unified_strdate(self._html_search_regex(
-            r'(?s)<div[^>]+class=["\']videoInfo(?:Date|Time)["\'][^>]*>(.+?)</div>',
+            [r'Date\s+[Aa]dded:\s*<span>([^<]+)',
+             r'(?s)<div[^>]+class=["\']videoInfo(?:Date|Time)["\'][^>]*>(.+?)</div>'],
             webpage, 'upload date', fatal=False))
 
         age_limit = self._rta_search(webpage)