]> Raphaƫl G. Git Repositories - youtubedl/blobdiff - youtube_dl/extractor/funnyordie.py
Merge tag 'upstream/2013.10.23'
[youtubedl] / youtube_dl / extractor / funnyordie.py
index 3045978f1b35cff1552b9b0dbf12f7a7ff522989..2ccdb70733e9b07cc2fc925630cde0a48a739b4b 100644 (file)
@@ -5,6 +5,15 @@ from .common import InfoExtractor
 
 class FunnyOrDieIE(InfoExtractor):
     _VALID_URL = r'^(?:https?://)?(?:www\.)?funnyordie\.com/videos/(?P<id>[0-9a-f]+)/.*$'
 
 class FunnyOrDieIE(InfoExtractor):
     _VALID_URL = r'^(?:https?://)?(?:www\.)?funnyordie\.com/videos/(?P<id>[0-9a-f]+)/.*$'
+    _TEST = {
+        u'url': u'http://www.funnyordie.com/videos/0732f586d7/heart-shaped-box-literal-video-version',
+        u'file': u'0732f586d7.mp4',
+        u'md5': u'f647e9e90064b53b6e046e75d0241fbd',
+        u'info_dict': {
+            u"description": u"Lyrics changed to match the video. Spoken cameo by Obscurus Lupa (from ThatGuyWithTheGlasses.com). Based on a concept by Dustin McLean (DustFilms.com). Performed, edited, and written by David A. Scott.", 
+            u"title": u"Heart-Shaped Box: Literal Video Version"
+        }
+    }
 
     def _real_extract(self, url):
         mobj = re.match(self._VALID_URL, url)
 
     def _real_extract(self, url):
         mobj = re.match(self._VALID_URL, url)
@@ -12,20 +21,15 @@ class FunnyOrDieIE(InfoExtractor):
         video_id = mobj.group('id')
         webpage = self._download_webpage(url, video_id)
 
         video_id = mobj.group('id')
         webpage = self._download_webpage(url, video_id)
 
-        video_url = self._html_search_regex(r'<video[^>]*>\s*<source[^>]*>\s*<source src="(?P<url>[^"]+)"',
+        video_url = self._search_regex(
+            [r'type="video/mp4" src="(.*?)"', r'src="([^>]*?)" type=\'video/mp4\''],
             webpage, u'video URL', flags=re.DOTALL)
 
             webpage, u'video URL', flags=re.DOTALL)
 
-        title = self._html_search_regex((r"<h1 class='player_page_h1'.*?>(?P<title>.*?)</h1>",
-            r'<title>(?P<title>[^<]+?)</title>'), webpage, 'title', flags=re.DOTALL)
-
-        video_description = self._html_search_regex(r'<meta property="og:description" content="(?P<desc>.*?)"',
-            webpage, u'description', fatal=False, flags=re.DOTALL)
-
         info = {
             'id': video_id,
             'url': video_url,
             'ext': 'mp4',
         info = {
             'id': video_id,
             'url': video_url,
             'ext': 'mp4',
-            'title': title,
-            'description': video_description,
+            'title': self._og_search_title(webpage),
+            'description': self._og_search_description(webpage),
         }
         return [info]
         }
         return [info]