Merge pull request #9395 from pmrowla/afreecatv

[youtube-dl] / youtube_dl / extractor / periscope.py

diff --git a/youtube_dl/extractor/periscope.py b/youtube_dl/extractor/periscope.py
index 8ad9367584e3c7f5e524baf3fb5333b1b2819a5d..c23b314e79df70e6b115fee6fb91386345d00ad4 100644 (file)
--- a/youtube_dl/extractor/periscope.py
+++ b/youtube_dl/extractor/periscope.py
@@ -2,17 +2,18 @@
 from __future__ import unicode_literals
 
 from .common import InfoExtractor
-from ..compat import (
-    compat_urllib_parse,
-    compat_urllib_request,
+from ..utils import (
+    parse_iso8601,
+    unescapeHTML,
 )
-from ..utils import parse_iso8601
 
 
 class PeriscopeIE(InfoExtractor):
     IE_DESC = 'Periscope'
-    _VALID_URL = r'https?://(?:www\.)?periscope\.tv/w/(?P<id>[^/?#]+)'
-    _TEST = {
+    IE_NAME = 'periscope'
+    _VALID_URL = r'https?://(?:www\.)?periscope\.tv/[^/]+/(?P<id>[^/?#]+)'
+    # Alive example URLs can be found here http://onperiscope.com/
+    _TESTS = [{
         'url': 'https://www.periscope.tv/w/aJUQnjY3MjA3ODF8NTYxMDIyMDl2zCg2pECBgwTqRpQuQD352EMPTKQjT4uqlM3cgWFA-g==',
         'md5': '65b57957972e503fcbbaeed8f4fa04ca',
         'info_dict': {
@@ -25,11 +26,17 @@ class PeriscopeIE(InfoExtractor):
             'uploader_id': '1465763',
         },
         'skip': 'Expires in 24 hours',
-    }
+    }, {
+        'url': 'https://www.periscope.tv/w/1ZkKzPbMVggJv',
+        'only_matching': True,
+    }, {
+        'url': 'https://www.periscope.tv/bastaakanoggano/1OdKrlkZZjOJX',
+        'only_matching': True,
+    }]
 
-    def _call_api(self, method, token):
+    def _call_api(self, method, value):
         return self._download_json(
-            'https://api.periscope.tv/api/v2/%s?token=%s' % (method, token), token)
+            'https://api.periscope.tv/api/v2/%s?broadcast_id=%s' % (method, value), value)
 
     def _real_extract(self, url):
         token = self._match_id(url)
@@ -38,8 +45,11 @@ class PeriscopeIE(InfoExtractor):
         broadcast = broadcast_data['broadcast']
         status = broadcast['status']
 
-        uploader = broadcast.get('user_display_name') or broadcast_data.get('user', {}).get('display_name')
-        uploader_id = broadcast.get('user_id') or broadcast_data.get('user', {}).get('id')
+        user = broadcast_data.get('user', {})
+
+        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'))
 
         title = '%s - %s' % (uploader, status) if uploader else status
         state = broadcast.get('state').lower()
@@ -78,22 +88,41 @@ class PeriscopeIE(InfoExtractor):
         }
 
 
-class QuickscopeIE(InfoExtractor):
-    IE_DESC = 'Quick Scope'
-    _VALID_URL = r'https?://watchonperiscope\.com/broadcast/(?P<id>\d+)'
+class PeriscopeUserIE(InfoExtractor):
+    _VALID_URL = r'https?://www\.periscope\.tv/(?P<id>[^/]+)/?$'
+    IE_DESC = 'Periscope user videos'
+    IE_NAME = 'periscope:user'
+
     _TEST = {
-        'url': 'https://watchonperiscope.com/broadcast/56180087',
-        'only_matching': True,
+        'url': 'https://www.periscope.tv/LularoeHusbandMike/',
+        'info_dict': {
+            'id': 'LularoeHusbandMike',
+            'title': 'LULAROE HUSBAND MIKE',
+            'description': 'md5:6cf4ec8047768098da58e446e82c82f0',
+        },
+        # Periscope only shows videos in the last 24 hours, so it's possible to
+        # get 0 videos
+        'playlist_mincount': 0,
     }
 
     def _real_extract(self, url):
-        broadcast_id = self._match_id(url)
-        request = compat_urllib_request.Request(
-            'https://watchonperiscope.com/api/accessChannel', compat_urllib_parse.urlencode({
-                'broadcast_id': broadcast_id,
-                'entry_ticket': '',
-                'from_push': 'false',
-                'uses_sessions': 'true',
-            }).encode('utf-8'))
-        return self.url_result(
-            self._download_json(request, broadcast_id)['share_url'], 'Periscope')
+        user_id = self._match_id(url)
+
+        webpage = self._download_webpage(url, user_id)
+
+        data_store = self._parse_json(
+            unescapeHTML(self._search_regex(
+                r'data-store=(["\'])(?P<data>.+?)\1',
+                webpage, 'data store', default='{}', group='data')),
+            user_id)
+
+        user = data_store.get('User', {}).get('user', {})
+        title = user.get('display_name') or user.get('username')
+        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', [])]
+
+        return self.playlist_result(entries, user_id, title, description)