Fix extraction from youtube.

[youtubedl] / youtube_dl / extractor / periscope.py

diff --git a/youtube_dl/extractor/periscope.py b/youtube_dl/extractor/periscope.py
index 1add6b84035041176f77f2bf454ca6bbfb8c7137..8afe541ec3ee7be90571f4ce5cfc39436ae26007 100644 (file)
--- a/youtube_dl/extractor/periscope.py
+++ b/youtube_dl/extractor/periscope.py
@@ -49,25 +49,23 @@ class PeriscopeIE(PeriscopeBaseIE):
     @staticmethod
     def _extract_url(webpage):
         mobj = re.search(
-            r'<iframe[^>]+src=([\'"])(?P<url>(?:https?:)?//(?:www\.)?periscope\.tv/(?:(?!\1).)+)\1', webpage)
+            r'<iframe[^>]+src=([\'"])(?P<url>(?:https?:)?//(?:www\.)?(?:periscope|pscp)\.tv/(?:(?!\1).)+)\1', webpage)
         if mobj:
             return mobj.group('url')
 
     def _real_extract(self, url):
         token = self._match_id(url)
 
-        broadcast_data = self._call_api(
-            'getBroadcastPublic', {'broadcast_id': token}, token)
-        broadcast = broadcast_data['broadcast']
-        status = broadcast['status']
+        stream = self._call_api(
+            'accessVideoPublic', {'broadcast_id': token}, token)
 
-        user = broadcast_data.get('user', {})
+        broadcast = stream['broadcast']
+        title = broadcast['status']
 
-        uploader = broadcast.get('user_display_name') or user.get('display_name')
-        uploader_id = (broadcast.get('username') or user.get('username') or
-                       broadcast.get('user_id') or user.get('id'))
+        uploader = broadcast.get('user_display_name') or broadcast.get('username')
+        uploader_id = (broadcast.get('user_id') or broadcast.get('username'))
 
-        title = '%s - %s' % (uploader, status) if uploader else status
+        title = '%s - %s' % (uploader, title) if uploader else title
         state = broadcast.get('state').lower()
         if state == 'running':
             title = self._live_title(title)
@@ -77,21 +75,24 @@ class PeriscopeIE(PeriscopeBaseIE):
             'url': broadcast[image],
         } for image in ('image_url', 'image_url_small') if broadcast.get(image)]
 
-        stream = self._call_api(
-            'getAccessPublic', {'broadcast_id': token}, token)
-
+        video_urls = set()
         formats = []
-        for format_id in ('replay', 'rtmp', 'hls', 'https_hls'):
+        for format_id in ('replay', 'rtmp', 'hls', 'https_hls', 'lhls', 'lhlsweb'):
             video_url = stream.get(format_id + '_url')
-            if not video_url:
+            if not video_url or video_url in video_urls:
                 continue
-            f = {
+            video_urls.add(video_url)
+            if format_id != 'rtmp':
+                formats.extend(self._extract_m3u8_formats(
+                    video_url, token, 'mp4',
+                    entry_protocol='m3u8_native'
+                    if state in ('ended', 'timed_out') else 'm3u8',
+                    m3u8_id=format_id, fatal=False))
+                continue
+            formats.append({
                 'url': video_url,
                 'ext': 'flv' if format_id == 'rtmp' else 'mp4',
-            }
-            if format_id != 'rtmp':
-                f['protocol'] = 'm3u8_native' if state in ('ended', 'timed_out') else 'm3u8'
-            formats.append(f)
+            })
         self._sort_formats(formats)
 
         return {