[brightcove] Change the 'videoId' or 'videoID' field to '@videoPlayer' (fixes #1697)
[youtube-dl] / youtube_dl / extractor / brightcove.py
index 0d9b87a34c7adf1cebe045c70c298a394612e6db..0c6e13b9cbc33eb7ffecb95b9b84461c3bf5ec2c 100644 (file)
@@ -9,6 +9,7 @@ from ..utils import (
     compat_urllib_parse,
     find_xpath_attr,
     compat_urlparse,
+    compat_str,
 
     ExtractorError,
 )
@@ -41,6 +42,17 @@ class BrightcoveIE(InfoExtractor):
                 u'uploader': u'Oracle',
             },
         },
+        {
+            # From http://mashable.com/2013/10/26/thermoelectric-bracelet-lets-you-control-your-body-temperature/
+            u'url': u'http://c.brightcove.com/services/viewer/federated_f9?&playerID=1265504713001&publisherID=AQ%7E%7E%2CAAABBzUwv1E%7E%2CxP-xFHVUstiMFlNYfvF4G9yFnNaqCw_9&videoID=2750934548001',
+            u'info_dict': {
+                u'id': u'2750934548001',
+                u'ext': u'mp4',
+                u'title': u'This Bracelet Acts as a Personal Thermostat',
+                u'description': u'md5:547b78c64f4112766ccf4e151c20b6a0',
+                u'uploader': u'Mashable',
+            },
+        },
     ]
 
     @classmethod
@@ -71,7 +83,22 @@ class BrightcoveIE(InfoExtractor):
         data = compat_urllib_parse.urlencode(params)
         return cls._FEDERATED_URL_TEMPLATE % data
 
+    @classmethod
+    def _extract_brightcove_url(cls, webpage):
+        """Try to extract the brightcove url from the wepbage, returns None
+        if it can't be found
+        """
+        m_brightcove = re.search(
+            r'<object[^>]+?class=([\'"])[^>]*?BrightcoveExperience.*?\1.+?</object>',
+            webpage, re.DOTALL)
+        if m_brightcove is not None:
+            return cls._build_brighcove_url(m_brightcove.group())
+        else:
+            return None
+
     def _real_extract(self, url):
+        # Change the 'videoId' or 'videoID' field to '@videoPlayer'
+        url = re.sub(r'(?<=[?&])videoI(d|D)', '%40videoPlayer', url)
         mobj = re.match(self._VALID_URL, url)
         query_str = mobj.group('query')
         query = compat_urlparse.parse_qs(query_str)
@@ -109,7 +136,7 @@ class BrightcoveIE(InfoExtractor):
 
     def _extract_video_info(self, video_info):
         info = {
-            'id': video_info['id'],
+            'id': compat_str(video_info['id']),
             'title': video_info['displayName'],
             'description': video_info.get('shortDescription'),
             'thumbnail': video_info.get('videoStillURL') or video_info.get('thumbnailURL'),