[pornhub] Modernize and fix test definition
authorPhilipp Hagemeister <phihag@phihag.de>
Sun, 26 Oct 2014 23:33:35 +0000 (00:33 +0100)
committerPhilipp Hagemeister <phihag@phihag.de>
Sun, 26 Oct 2014 23:33:35 +0000 (00:33 +0100)
youtube_dl/extractor/pornhub.py
youtube_dl/extractor/rtlnow.py

index 4118ee9560e03d2fa1eea171766ef4893e274aa5..618e8f5dd18deff8ffe3f907ce2f97d1256ba204 100644 (file)
@@ -16,13 +16,14 @@ from ..aes import (
 
 
 class PornHubIE(InfoExtractor):
-    _VALID_URL = r'^(?:https?://)?(?:www\.)?(?P<url>pornhub\.com/view_video\.php\?viewkey=(?P<videoid>[0-9a-f]+))'
+    _VALID_URL = r'^https?://(?:www\.)?pornhub\.com/view_video\.php\?viewkey=(?P<id>[0-9a-f]+)'
     _TEST = {
         'url': 'http://www.pornhub.com/view_video.php?viewkey=648719015',
-        'file': '648719015.mp4',
         'md5': '882f488fa1f0026f023f33576004a2ed',
         'info_dict': {
-            "uploader": "BABES-COM",
+            'id': '648719015',
+            'ext': 'mp4',
+            "uploader": "Babes",
             "title": "Seductive Indian beauty strips down and fingers her pink pussy",
             "age_limit": 18
         }
@@ -35,9 +36,7 @@ class PornHubIE(InfoExtractor):
         return count
 
     def _real_extract(self, url):
-        mobj = re.match(self._VALID_URL, url)
-        video_id = mobj.group('videoid')
-        url = 'http://www.' + mobj.group('url')
+        video_id = self._match_id(url)
 
         req = compat_urllib_request.Request(url)
         req.add_header('Cookie', 'age_verified=1')
@@ -45,7 +44,7 @@ class PornHubIE(InfoExtractor):
 
         video_title = self._html_search_regex(r'<h1 [^>]+>([^<]+)', webpage, 'title')
         video_uploader = self._html_search_regex(
-            r'(?s)From:&nbsp;.+?<(?:a href="/users/|<span class="username)[^>]+>(.+?)<',
+            r'(?s)From:&nbsp;.+?<(?:a href="/users/|a href="/channels/|<span class="username)[^>]+>(.+?)<',
             webpage, 'uploader', fatal=False)
         thumbnail = self._html_search_regex(r'"image_url":"([^"]+)', webpage, 'thumbnail', fatal=False)
         if thumbnail:
index a45884b251fa355e04c65de554f0e9cbfb5406bb..1a41cbe409acc666a30a056c4d07c63afbadd863 100644 (file)
@@ -81,7 +81,7 @@ class RTLnowIE(InfoExtractor):
                 'id': '99205',
                 'ext': 'flv',
                 'title': 'Medicopter 117 - Angst!',
-                'description': 'md5:895b1df01639b5f61a04fc305a5cb94d',
+                'description': 're:^Im Therapiezentrum \'Sonnalm\' kommen durch eine Unachtsamkeit die für die B.handlung mit Phobikern gehaltenen Voglespinnen frei\. Eine Ausreißerin',
                 'thumbnail': 'http://autoimg.static-fra.de/superrtlnow/287529/1500x1500/image2.jpg',
                 'upload_date': '20080928',
                 'duration': 2691,