Improve some _VALID_URLs

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

diff --git a/youtube_dl/extractor/periscope.py b/youtube_dl/extractor/periscope.py
index 514e9b4339be43b509f9c9a8a6d2b87187e5f056..eb1aeba46848cee3ea652c01b284ff44fe25292e 100644 (file)
--- a/youtube_dl/extractor/periscope.py
+++ b/youtube_dl/extractor/periscope.py
@@ -2,11 +2,22 @@
 from __future__ import unicode_literals
 
 from .common import InfoExtractor
-from ..utils import parse_iso8601
+from ..utils import (
+    parse_iso8601,
+    unescapeHTML,
+)
 
 
-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'
     _VALID_URL = r'https?://(?:www\.)?periscope\.tv/[^/]+/(?P<id>[^/?#]+)'
     # Alive example URLs can be found here http://onperiscope.com/
     _TESTS = [{
@@ -30,19 +41,19 @@ class PeriscopeIE(InfoExtractor):
         '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)
-
     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']
 
-        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()
@@ -54,7 +65,8 @@ class PeriscopeIE(InfoExtractor):
             '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'):
@@ -79,3 +91,54 @@ class PeriscopeIE(InfoExtractor):
             'thumbnails': thumbnails,
             'formats': formats,
         }
+
+
+class PeriscopeUserIE(PeriscopeBaseIE):
+    _VALID_URL = r'https?://(?:www\.)?periscope\.tv/(?P<id>[^/]+)/?$'
+    IE_DESC = 'Periscope user videos'
+    IE_NAME = 'periscope:user'
+
+    _TEST = {
+        '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):
+        user_name = self._match_id(url)
+
+        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')),
+            user_name)
+
+        user = list(data_store['UserCache']['users'].values())[0]['user']
+        user_id = user['id']
+        session_id = data_store['SessionToken']['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')]
+
+        title = user.get('display_name') or user.get('username') or user_name
+        description = user.get('description')
+
+        entries = [
+            self.url_result(
+                '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)