]> Raphaƫl G. Git Repositories - youtubedl/blobdiff - youtube_dl/extractor/howcast.py
debian/copyright: Fix missing-license-paragraph-in-dep5-copyright.
[youtubedl] / youtube_dl / extractor / howcast.py
index 7b94f85adc2730cdf13304bcbb47b84de6ecc244..3f7d6666c0810e545c0f285dcf70689806c4dac7 100644 (file)
@@ -1,37 +1,41 @@
+from __future__ import unicode_literals
+
 import re
 
 from .common import InfoExtractor
 
 
 class HowcastIE(InfoExtractor):
-    _VALID_URL = r'(?:https?://)?(?:www\.)?howcast\.com/videos/(?P<id>\d+)'
+    _VALID_URL = r'https?://(?:www\.)?howcast\.com/videos/(?P<id>\d+)'
+    _TEST = {
+        'url': 'http://www.howcast.com/videos/390161-How-to-Tie-a-Square-Knot-Properly',
+        'md5': '8b743df908c42f60cf6496586c7f12c3',
+        'info_dict': {
+            'id': '390161',
+            'ext': 'mp4',
+            'description': 'The square knot, also known as the reef knot, is one of the oldest, most basic knots to tie, and can be used in many different ways. Here\'s the proper way to tie a square knot.',
+            'title': 'How to Tie a Square Knot Properly',
+        }
+    }
 
     def _real_extract(self, url):
         mobj = re.match(self._VALID_URL, url)
 
         video_id = mobj.group('id')
-        webpage_url = 'http://www.howcast.com/videos/' + video_id
-        webpage = self._download_webpage(webpage_url, video_id)
+        webpage = self._download_webpage(url, video_id)
 
         self.report_extraction(video_id)
 
         video_url = self._search_regex(r'\'?file\'?: "(http://mobile-media\.howcast\.com/[0-9]+\.mp4)',
-            webpage, u'video URL')
-
-        video_title = self._html_search_regex(r'<meta content=(?:"([^"]+)"|\'([^\']+)\') property=\'og:title\'',
-            webpage, u'title')
+                                       webpage, 'video URL')
 
         video_description = self._html_search_regex(r'<meta content=(?:"([^"]+)"|\'([^\']+)\') name=\'description\'',
-            webpage, u'description', fatal=False)
-
-        thumbnail = self._html_search_regex(r'<meta content=\'(.+?)\' property=\'og:image\'',
-            webpage, u'thumbnail', fatal=False)
+                                                    webpage, 'description', fatal=False)
 
-        return [{
-            'id':       video_id,
-            'url':      video_url,
-            'ext':      'mp4',
-            'title':    video_title,
+        return {
+            'id': video_id,
+            'url': video_url,
+            'title': self._og_search_title(webpage),
             'description': video_description,
-            'thumbnail': thumbnail,
-        }]
+            'thumbnail': self._og_search_thumbnail(webpage),
+        }