[ie/twitch] Support `--live-from-start` (#13202)

Closes #10520
Authored by: bashonly

yt_dlp/extractor/twitch.py

@@ -187,7 +187,7 @@ class TwitchBaseIE(InfoExtractor):
             'url': thumbnail,
         }] if thumbnail else None
 
-    def _extract_twitch_m3u8_formats(self, path, video_id, token, signature):
+    def _extract_twitch_m3u8_formats(self, path, video_id, token, signature, live_from_start=False):
         formats = self._extract_m3u8_formats(
             f'{self._USHER_BASE}/{path}/{video_id}.m3u8', video_id, 'mp4', query={
                 'allow_source': 'true',
@@ -204,7 +204,10 @@ class TwitchBaseIE(InfoExtractor):
         for fmt in formats:
             if fmt.get('vcodec') and fmt['vcodec'].startswith('av01'):
                 # mpegts does not yet have proper support for av1
-                fmt['downloader_options'] = {'ffmpeg_args_out': ['-f', 'mp4']}
+                fmt.setdefault('downloader_options', {}).update({'ffmpeg_args_out': ['-f', 'mp4']})
+            if live_from_start:
+                fmt.setdefault('downloader_options', {}).update({'ffmpeg_args': ['-live_start_index', '0']})
+                fmt['is_from_start'] = True
 
         return formats
 
@@ -550,7 +553,8 @@ class TwitchVodIE(TwitchBaseIE):
         access_token = self._download_access_token(vod_id, 'video', 'id')
 
         formats = self._extract_twitch_m3u8_formats(
-            'vod', vod_id, access_token['value'], access_token['signature'])
+            'vod', vod_id, access_token['value'], access_token['signature'],
+            live_from_start=self.get_param('live_from_start'))
         formats.extend(self._extract_storyboard(vod_id, video.get('storyboard'), info.get('duration')))
 
         self._prefer_source(formats)
@@ -633,6 +637,10 @@ class TwitchPlaylistBaseIE(TwitchBaseIE):
     _PAGE_LIMIT = 100
 
     def _entries(self, channel_name, *args):
+        """
+        Subclasses must define _make_variables() and _extract_entry(),
+        as well as set _OPERATION_NAME, _ENTRY_KIND, _EDGE_KIND, and _NODE_KIND
+        """
         cursor = None
         variables_common = self._make_variables(channel_name, *args)
         entries_key = f'{self._ENTRY_KIND}s'
@@ -672,7 +680,22 @@ class TwitchPlaylistBaseIE(TwitchBaseIE):
                 break
 
 
-class TwitchVideosIE(TwitchPlaylistBaseIE):
+class TwitchVideosBaseIE(TwitchPlaylistBaseIE):
+    _OPERATION_NAME = 'FilterableVideoTower_Videos'
+    _ENTRY_KIND = 'video'
+    _EDGE_KIND = 'VideoEdge'
+    _NODE_KIND = 'Video'
+
+    @staticmethod
+    def _make_variables(channel_name, broadcast_type, sort):
+        return {
+            'channelOwnerLogin': channel_name,
+            'broadcastType': broadcast_type,
+            'videoSort': sort.upper(),
+        }
+
+
+class TwitchVideosIE(TwitchVideosBaseIE):
     _VALID_URL = r'https?://(?:(?:www|go|m)\.)?twitch\.tv/(?P<id>[^/]+)/(?:videos|profile)'
 
     _TESTS = [{
@@ -751,11 +774,6 @@ class TwitchVideosIE(TwitchPlaylistBaseIE):
         'views': 'Popular',
     }
 
-    _OPERATION_NAME = 'FilterableVideoTower_Videos'
-    _ENTRY_KIND = 'video'
-    _EDGE_KIND = 'VideoEdge'
-    _NODE_KIND = 'Video'
-
     @classmethod
     def suitable(cls, url):
         return (False
@@ -764,14 +782,6 @@ class TwitchVideosIE(TwitchPlaylistBaseIE):
                     TwitchVideosCollectionsIE))
                 else super().suitable(url))
 
-    @staticmethod
-    def _make_variables(channel_name, broadcast_type, sort):
-        return {
-            'channelOwnerLogin': channel_name,
-            'broadcastType': broadcast_type,
-            'videoSort': sort.upper(),
-        }
-
     @staticmethod
     def _extract_entry(node):
         return _make_video_result(node)
@@ -919,7 +929,7 @@ class TwitchVideosCollectionsIE(TwitchPlaylistBaseIE):
             playlist_title=f'{channel_name} - Collections')
 
 
-class TwitchStreamIE(TwitchBaseIE):
+class TwitchStreamIE(TwitchVideosBaseIE):
     IE_NAME = 'twitch:stream'
     _VALID_URL = r'''(?x)
                     https?://
@@ -982,6 +992,7 @@ class TwitchStreamIE(TwitchBaseIE):
             'skip_download': 'Livestream',
         },
     }]
+    _PAGE_LIMIT = 1
 
     @classmethod
     def suitable(cls, url):
@@ -995,6 +1006,20 @@ class TwitchStreamIE(TwitchBaseIE):
                     TwitchClipsIE))
                 else super().suitable(url))
 
+    @staticmethod
+    def _extract_entry(node):
+        if not isinstance(node, dict) or not node.get('id'):
+            return None
+        video_id = node['id']
+        return {
+            '_type': 'url',
+            'ie_key': TwitchVodIE.ie_key(),
+            'id': 'v' + video_id,
+            'url': f'https://www.twitch.tv/videos/{video_id}',
+            'title': node.get('title'),
+            'timestamp': unified_timestamp(node.get('publishedAt')) or 0,
+        }
+
     def _real_extract(self, url):
         channel_name = self._match_id(url).lower()
 
@@ -1029,6 +1054,16 @@ class TwitchStreamIE(TwitchBaseIE):
         if not stream:
             raise UserNotLive(video_id=channel_name)
 
+        timestamp = unified_timestamp(stream.get('createdAt'))
+
+        if self.get_param('live_from_start'):
+            self.to_screen(f'{channel_name}: Extracting VOD to download live from start')
+            entry = next(self._entries(channel_name, None, 'time'), None)
+            if entry and entry.pop('timestamp') >= (timestamp or float('inf')):
+                return entry
+            self.report_warning(
+                'Unable to extract the VOD associated with this livestream', video_id=channel_name)
+
         access_token = self._download_access_token(
             channel_name, 'stream', 'channelName')
 
@@ -1038,7 +1073,6 @@ class TwitchStreamIE(TwitchBaseIE):
         self._prefer_source(formats)
 
         view_count = stream.get('viewers')
-        timestamp = unified_timestamp(stream.get('createdAt'))
 
         sq_user = try_get(gql, lambda x: x[1]['data']['user'], dict) or {}
         uploader = sq_user.get('displayName')

yt_dlp/options.py

@@ -470,7 +470,7 @@ def create_parser():
     general.add_option(
         '--live-from-start',
         action='store_true', dest='live_from_start',
-        help='Download livestreams from the start. Currently only supported for YouTube (Experimental)')
+        help='Download livestreams from the start. Currently only supported for YouTube (experimental) and Twitch')
     general.add_option(
         '--no-live-from-start',
         action='store_false', dest='live_from_start',