[twitter:spaces] Add downloader

cited 56ba5e11cc (diff-a949010439cefd1cf130901ae40a60fc3c33d3e191667b9fc8dc40a58d3e5310R271-R289) by @ef1500
TODO: Rename the concatenated `.m4a` to `.aac` and then `FixupM4a`

yt_dlp/downloader/__init__.py

@@ -36,6 +36,7 @@ def get_suitable_downloader(info_dict, params={}, default=NO_DEFAULT, protocol=N
 from .websocket import WebSocketFragmentFD
 from .youtube_live_chat import YoutubeLiveChatFD
 from .bunnycdn import BunnyCdnFD
+from .twitter import TwitterSpacesFD
 
 PROTOCOL_MAP = {
     'rtmp': RtmpFD,
@@ -56,6 +57,7 @@ def get_suitable_downloader(info_dict, params={}, default=NO_DEFAULT, protocol=N
     'youtube_live_chat': YoutubeLiveChatFD,
     'youtube_live_chat_replay': YoutubeLiveChatFD,
     'bunnycdn': BunnyCdnFD,
+    'twitter_spaces': TwitterSpacesFD,
 }
 
 

yt_dlp/downloader/twitter.py

@@ -0,0 +1,15 @@
+from .hls import HlsFD
+
+
+class TwitterSpacesFD(HlsFD):
+    FD_NAME = 'twitterspaces'
+
+    def _read_fragment(self, ctx):
+        frag_content = super()._read_fragment(ctx)
+        idx = frag_content.find(bytes(
+            [0x49, 0x44, 0x33, 0x04, 0x00, 0x00, 0x00, 0x00, 0x00, 0x3F, 0x50, 0x52, 0x49, 0x56]))
+        if idx > 0:
+            self.to_screen(f'[{self.FD_NAME}] Removing partial header ({idx} bytes)')
+            return frag_content[idx:]
+        else:
+            return frag_content

yt_dlp/extractor/twitter.py

@@ -1853,8 +1853,8 @@ def _real_extract(self, url):
             is_audio_space = source and 'audio-space' in source
             formats = self._extract_m3u8_formats(
                 source, metadata['media_key'], 'm4a' if is_audio_space else 'mp4',
-                # XXX: Some audio-only Spaces need ffmpeg as downloader
+                # XXX: Some audio-only Spaces need to truncate heads of their fragments, not only ffmpeg as downloader
-                entry_protocol='m3u8' if is_audio_space else 'm3u8_native',
+                entry_protocol='twitter_spaces' if is_audio_space else 'm3u8_native',
                 live=is_live, headers=headers, fatal=False) if source else []
             if is_audio_space:
                 for fmt in formats: