[ie/twitter:broadcast] Improve metadata extraction (#8383)

Authored by: HitomaruKonpaku

diff --git a/yt_dlp/extractor/periscope.py b/yt_dlp/extractor/periscope.py
index 84bcf1573ac04ffab523c6efa5c2cbf12df7cb77..dcd021926694f3106a18bf8e468dd90072f0616f 100644 (file)
--- a/yt_dlp/extractor/periscope.py
+++ b/yt_dlp/extractor/periscope.py
@@ -24,12 +24,13 @@ def _parse_broadcast_data(self, broadcast, video_id):
 
         thumbnails = [{
             'url': broadcast[image],
-        } for image in ('image_url', 'image_url_small') if broadcast.get(image)]
+        } for image in ('image_url', 'image_url_medium', 'image_url_small') if broadcast.get(image)]
 
         return {
             'id': broadcast.get('id') or video_id,
             'title': title,
-            'timestamp': parse_iso8601(broadcast.get('created_at')),
+            'timestamp': parse_iso8601(broadcast.get('created_at')) or int_or_none(
+                broadcast.get('created_at_ms'), scale=1000),
             'uploader': uploader,
             'uploader_id': broadcast.get('user_id') or broadcast.get('username'),
             'thumbnails': thumbnails,

diff --git a/yt_dlp/extractor/twitter.py b/yt_dlp/extractor/twitter.py
index b6386214d9c71a4f50f84acdb64000055fbfd3a1..7bd78eb487e1f87e4ac321bb78afc4921059bc30 100644 (file)
--- a/yt_dlp/extractor/twitter.py
+++ b/yt_dlp/extractor/twitter.py
@@ -1563,7 +1563,7 @@ class TwitterBroadcastIE(TwitterBaseIE, PeriscopeBaseIE):
     IE_NAME = 'twitter:broadcast'
     _VALID_URL = TwitterBaseIE._BASE_REGEX + r'i/broadcasts/(?P<id>[0-9a-zA-Z]{13})'
 
-    _TEST = {
+    _TESTS = [{
         # untitled Periscope video
         'url': 'https://twitter.com/i/broadcasts/1yNGaQLWpejGj',
         'info_dict': {
@@ -1571,11 +1571,42 @@ class TwitterBroadcastIE(TwitterBaseIE, PeriscopeBaseIE):
             'ext': 'mp4',
             'title': 'Andrea May Sahouri - Periscope Broadcast',
             'uploader': 'Andrea May Sahouri',
-            'uploader_id': '1PXEdBZWpGwKe',
+            'uploader_id': 'andreamsahouri',
+            'uploader_url': 'https://twitter.com/andreamsahouri',
+            'timestamp': 1590973638,
+            'upload_date': '20200601',
             'thumbnail': r're:^https?://[^?#]+\.jpg\?token=',
             'view_count': int,
         },
-    }
+    }, {
+        'url': 'https://twitter.com/i/broadcasts/1ZkKzeyrPbaxv',
+        'info_dict': {
+            'id': '1ZkKzeyrPbaxv',
+            'ext': 'mp4',
+            'title': 'Starship | SN10 | High-Altitude Flight Test',
+            'uploader': 'SpaceX',
+            'uploader_id': 'SpaceX',
+            'uploader_url': 'https://twitter.com/SpaceX',
+            'timestamp': 1614812942,
+            'upload_date': '20210303',
+            'thumbnail': r're:^https?://[^?#]+\.jpg\?token=',
+            'view_count': int,
+        },
+    }, {
+        'url': 'https://twitter.com/i/broadcasts/1OyKAVQrgzwGb',
+        'info_dict': {
+            'id': '1OyKAVQrgzwGb',
+            'ext': 'mp4',
+            'title': 'Starship Flight Test',
+            'uploader': 'SpaceX',
+            'uploader_id': 'SpaceX',
+            'uploader_url': 'https://twitter.com/SpaceX',
+            'timestamp': 1681993964,
+            'upload_date': '20230420',
+            'thumbnail': r're:^https?://[^?#]+\.jpg\?token=',
+            'view_count': int,
+        },
+    }]
 
     def _real_extract(self, url):
         broadcast_id = self._match_id(url)
@@ -1585,6 +1616,9 @@ def _real_extract(self, url):
         if not broadcast:
             raise ExtractorError('Broadcast no longer exists', expected=True)
         info = self._parse_broadcast_data(broadcast, broadcast_id)
+        info['title'] = broadcast.get('status') or info.get('title')
+        info['uploader_id'] = broadcast.get('twitter_username') or info.get('uploader_id')
+        info['uploader_url'] = format_field(broadcast, 'twitter_username', 'https://twitter.com/%s', default=None)
         media_key = broadcast['media_key']
         source = self._call_api(
             f'live_video_stream/status/{media_key}', media_key)['source']