Prepare to release.

[youtubedl] / youtube_dl / extractor / periscope.py

diff --git a/youtube_dl/extractor/periscope.py b/youtube_dl/extractor/periscope.py
index c23b314e79df70e6b115fee6fb91386345d00ad4..1add6b84035041176f77f2bf454ca6bbfb8c7137 100644 (file)
--- a/youtube_dl/extractor/periscope.py
+++ b/youtube_dl/extractor/periscope.py
@@ -1,6 +1,8 @@
 # coding: utf-8
 from __future__ import unicode_literals
 
 # coding: utf-8
 from __future__ import unicode_literals
 

+import re
+

 from .common import InfoExtractor
 from ..utils import (
     parse_iso8601,
 from .common import InfoExtractor
 from ..utils import (
     parse_iso8601,


@@ -8,10 +10,17 @@ from ..utils import (
 )
 
 
 )
 
 

-class PeriscopeIE(InfoExtractor):
+class PeriscopeBaseIE(InfoExtractor):
+    def _call_api(self, method, query, item_id):
+        return self._download_json(
+            'https://api.periscope.tv/api/v2/%s' % method,
+            item_id, query=query)
+
+
+class PeriscopeIE(PeriscopeBaseIE):

     IE_DESC = 'Periscope'
     IE_NAME = 'periscope'
     IE_DESC = 'Periscope'
     IE_NAME = 'periscope'

-    _VALID_URL = r'https?://(?:www\.)?periscope\.tv/[^/]+/(?P<id>[^/?#]+)'
+    _VALID_URL = r'https?://(?:www\.)?(?:periscope|pscp)\.tv/[^/]+/(?P<id>[^/?#]+)'

     # Alive example URLs can be found here http://onperiscope.com/
     _TESTS = [{
         'url': 'https://www.periscope.tv/w/aJUQnjY3MjA3ODF8NTYxMDIyMDl2zCg2pECBgwTqRpQuQD352EMPTKQjT4uqlM3cgWFA-g==',
     # Alive example URLs can be found here http://onperiscope.com/
     _TESTS = [{
         'url': 'https://www.periscope.tv/w/aJUQnjY3MjA3ODF8NTYxMDIyMDl2zCg2pECBgwTqRpQuQD352EMPTKQjT4uqlM3cgWFA-g==',


@@ -32,16 +41,23 @@ class PeriscopeIE(InfoExtractor):
     }, {
         'url': 'https://www.periscope.tv/bastaakanoggano/1OdKrlkZZjOJX',
         'only_matching': True,
     }, {
         'url': 'https://www.periscope.tv/bastaakanoggano/1OdKrlkZZjOJX',
         'only_matching': True,

+    }, {
+        'url': 'https://www.periscope.tv/w/1ZkKzPbMVggJv',
+        'only_matching': True,

     }]
 
     }]
 

-    def _call_api(self, method, value):
-        return self._download_json(
-            'https://api.periscope.tv/api/v2/%s?broadcast_id=%s' % (method, value), value)
+    @staticmethod
+    def _extract_url(webpage):
+        mobj = re.search(
+            r'<iframe[^>]+src=([\'"])(?P<url>(?:https?:)?//(?:www\.)?periscope\.tv/(?:(?!\1).)+)\1', webpage)
+        if mobj:
+            return mobj.group('url')

 
     def _real_extract(self, url):
         token = self._match_id(url)
 
 
     def _real_extract(self, url):
         token = self._match_id(url)
 

-        broadcast_data = self._call_api('getBroadcastPublic', token)
+        broadcast_data = self._call_api(
+            'getBroadcastPublic', {'broadcast_id': token}, token)

         broadcast = broadcast_data['broadcast']
         status = broadcast['status']
 
         broadcast = broadcast_data['broadcast']
         status = broadcast['status']
 


@@ -61,7 +77,8 @@ class PeriscopeIE(InfoExtractor):
             'url': broadcast[image],
         } for image in ('image_url', 'image_url_small') if broadcast.get(image)]
 
             'url': broadcast[image],
         } for image in ('image_url', 'image_url_small') if broadcast.get(image)]
 

-        stream = self._call_api('getAccessPublic', token)
+        stream = self._call_api(
+            'getAccessPublic', {'broadcast_id': token}, token)

 
         formats = []
         for format_id in ('replay', 'rtmp', 'hls', 'https_hls'):
 
         formats = []
         for format_id in ('replay', 'rtmp', 'hls', 'https_hls'):


@@ -73,7 +90,7 @@ class PeriscopeIE(InfoExtractor):
                 'ext': 'flv' if format_id == 'rtmp' else 'mp4',
             }
             if format_id != 'rtmp':
                 'ext': 'flv' if format_id == 'rtmp' else 'mp4',
             }
             if format_id != 'rtmp':

-                f['protocol'] = 'm3u8_native' if state == 'ended' else 'm3u8'
+                f['protocol'] = 'm3u8_native' if state in ('ended', 'timed_out') else 'm3u8'

             formats.append(f)
         self._sort_formats(formats)
 
             formats.append(f)
         self._sort_formats(formats)
 


@@ -88,8 +105,8 @@ class PeriscopeIE(InfoExtractor):
         }
 
 
         }
 
 

-class PeriscopeUserIE(InfoExtractor):
-    _VALID_URL = r'https?://www\.periscope\.tv/(?P<id>[^/]+)/?$'
+class PeriscopeUserIE(PeriscopeBaseIE):
+    _VALID_URL = r'https?://(?:www\.)?(?:periscope|pscp)\.tv/(?P<id>[^/]+)/?$'

     IE_DESC = 'Periscope user videos'
     IE_NAME = 'periscope:user'
 
     IE_DESC = 'Periscope user videos'
     IE_NAME = 'periscope:user'
 


@@ -106,23 +123,34 @@ class PeriscopeUserIE(InfoExtractor):
     }
 
     def _real_extract(self, url):
     }
 
     def _real_extract(self, url):

-        user_id = self._match_id(url)
+        user_name = self._match_id(url)

 
 

-        webpage = self._download_webpage(url, user_id)
+        webpage = self._download_webpage(url, user_name)

 
         data_store = self._parse_json(
             unescapeHTML(self._search_regex(
                 r'data-store=(["\'])(?P<data>.+?)\1',
                 webpage, 'data store', default='{}', group='data')),
 
         data_store = self._parse_json(
             unescapeHTML(self._search_regex(
                 r'data-store=(["\'])(?P<data>.+?)\1',
                 webpage, 'data store', default='{}', group='data')),

-            user_id)
+            user_name)
+
+        user = list(data_store['UserCache']['users'].values())[0]['user']
+        user_id = user['id']
+        session_id = data_store['SessionToken']['public']['broadcastHistory']['token']['session_id']
+
+        broadcasts = self._call_api(
+            'getUserBroadcastsPublic',
+            {'user_id': user_id, 'session_id': session_id},
+            user_name)['broadcasts']
+
+        broadcast_ids = [
+            broadcast['id'] for broadcast in broadcasts if broadcast.get('id')]

 
 

-        user = data_store.get('User', {}).get('user', {})
-        title = user.get('display_name') or user.get('username')
+        title = user.get('display_name') or user.get('username') or user_name

         description = user.get('description')
 
         entries = [
             self.url_result(
         description = user.get('description')
 
         entries = [
             self.url_result(

-                'https://www.periscope.tv/%s/%s' % (user_id, broadcast['id']))
-            for broadcast in data_store.get('UserBroadcastHistory', {}).get('broadcasts', [])]
+                'https://www.periscope.tv/%s/%s' % (user_name, broadcast_id))
+            for broadcast_id in broadcast_ids]

 
         return self.playlist_result(entries, user_id, title, description)
 
         return self.playlist_result(entries, user_id, title, description)