refactor(voice): Strict type checking in voice internals & DAVE Support (rec)

CHANGELOG.md

@@ -14,6 +14,10 @@ These changes are available on the `master` branch, but have not yet been releas
 
 ### Changed
 
+- Added support for Discord DAVE (Audio & Video E2EE) for voice-receive related features
+  and refactored the voice-reception system.
+  ([#3159](https://github.com/Pycord-Development/pycord/pull/3159))
+
 ### Fixed
 
 - Fixed internal use of deprecated role type methods.

discord/opus.py

@@ -708,7 +708,15 @@ def _decode_packet(self, packet: Packet) -> tuple[Packet, bytes]:
 
         if packet:
             other_code = False
-            pcm = self._decoder.decode(packet.decrypted_data, fec=False)
+            try:
+                pcm = self._decoder.decode(packet.decrypted_data, fec=False)
+            except OpusError:
+                _log.warning(
+                    "Opus decode failed for packet seq=%s, using PLC",
+                    packet.sequence,
+                    exc_info=True,
+                )
+                pcm = self._decoder.decode(None, fec=False)
 
         if other_code:
             next_packet = self._buffer.peek_next()
@@ -725,9 +733,4 @@ def _decode_packet(self, packet: Packet) -> tuple[Packet, bytes]:
             else:
                 pcm = self._decoder.decode(None, fec=False)
 
-        if HAS_DAVEY:
-            if user_id is not None and in_dave and dave.can_passthrough(user_id):
-                _log.debug("User ID %s can passthrough, decrypting with DAVE", user_id)
-                pcm = dave.decrypt(user_id, davey.MediaType.audio, pcm)
-
         return packet, pcm

discord/sinks/core.py

@@ -37,7 +37,10 @@
 from .errors import SinkException
 
 if TYPE_CHECKING:
+    from ..member import Member
+    from ..user import User
     from ..voice import VoiceClient
+    from ..voice.packets import VoiceData
 
 __all__ = (
     "Filters",
@@ -210,6 +213,8 @@ class Sink(Filters):
         Audio may only be formatted after recording is finished.
     """
 
+    __sink_listeners__: list[tuple[str, str]] = []
+
     def __init__(self, *, filters=None):
         if filters is None:
             filters = default_filters
@@ -222,18 +227,38 @@ def __init__(self, *, filters=None):
     def client(self) -> VoiceClient | None:
         return self.vc
 
+    @property
+    def recording(self) -> bool:
+        """Whether the voice client is currently recording."""
+        return self.vc is not None and self.vc.is_recording()
+
+    def is_opus(self) -> bool:
+        """Whether this sink accepts raw opus packets instead of decoded PCM."""
+        return False
+
+    def walk_children(self):
+        """Yields child sinks. Base implementation yields nothing."""
+        return
+        yield  # make it a generator
+
     def init(self, vc: VoiceClient):  # called under listen
         self.vc = vc
         super().init()
 
     @Filters.container
-    def write(self, data, user):
+    def write(self, data: VoiceData | bytes, user: User | Member | None) -> None:
+        from ..voice.packets import VoiceData
+
+        if isinstance(data, VoiceData):
+            pcm_data = data.pcm
+        else:
+            pcm_data = data
+
         if user not in self.audio_data:
             file = io.BytesIO()
             self.audio_data.update({user: AudioData(file)})
 
-        file = self.audio_data[user]
-        file.write(data)
+        self.audio_data[user].write(pcm_data)
 
     def cleanup(self):
         self.finished = True

discord/sinks/m4a.py

@@ -57,7 +57,7 @@ def format_audio(self, audio):
         M4ASinkError
             Formatting the audio failed.
         """
-        if self.vc.recording:
+        if self.recording:
             raise M4ASinkError(
                 "Audio may only be formatted after recording is finished."
             )

discord/sinks/mka.py

@@ -55,7 +55,7 @@ def format_audio(self, audio):
         MKASinkError
             Formatting the audio failed.
         """
-        if self.vc.recording:
+        if self.recording:
             raise MKASinkError(
                 "Audio may only be formatted after recording is finished."
             )

discord/sinks/mkv.py

@@ -55,7 +55,7 @@ def format_audio(self, audio):
         MKVSinkError
             Formatting the audio failed.
         """
-        if self.vc.recording:
+        if self.recording:
             raise MKVSinkError(
                 "Audio may only be formatted after recording is finished."
             )

discord/sinks/mp3.py

@@ -55,7 +55,7 @@ def format_audio(self, audio):
         MP3SinkError
             Formatting the audio failed.
         """
-        if self.vc.recording:
+        if self.recording:
             raise MP3SinkError(
                 "Audio may only be formatted after recording is finished."
             )

discord/sinks/mp4.py

@@ -57,7 +57,7 @@ def format_audio(self, audio):
         MP4SinkError
             Formatting the audio failed.
         """
-        if self.vc.recording:
+        if self.recording:
             raise MP4SinkError(
                 "Audio may only be formatted after recording is finished."
             )

discord/sinks/ogg.py

@@ -55,7 +55,7 @@ def format_audio(self, audio):
         OGGSinkError
             Formatting the audio failed.
         """
-        if self.vc.recording:
+        if self.recording:
             raise OGGSinkError(
                 "Audio may only be formatted after recording is finished."
             )

discord/sinks/wave.py

@@ -23,7 +23,9 @@
 """
 
 import wave
+from io import BytesIO
 
+from ..opus import Decoder as OpusDecoder
 from .core import Filters, Sink, default_filters
 from .errors import WaveSinkError
 
@@ -54,16 +56,20 @@ def format_audio(self, audio):
         WaveSinkError
             Formatting the audio failed.
         """
-        if self.vc.recording:
+        if self.recording:
             raise WaveSinkError(
                 "Audio may only be formatted after recording is finished."
             )
-        data = audio.file
 
-        with wave.open(data, "wb") as f:
-            f.setnchannels(self.vc.decoder.CHANNELS)
-            f.setsampwidth(self.vc.decoder.SAMPLE_SIZE // self.vc.decoder.CHANNELS)
-            f.setframerate(self.vc.decoder.SAMPLING_RATE)
+        audio.file.seek(0)
+        pcm_data = audio.file.read()
 
+        data = BytesIO()
+        with wave.open(data, "wb") as f:
+            f.setnchannels(OpusDecoder.CHANNELS)
+            f.setsampwidth(OpusDecoder.SAMPLE_SIZE // OpusDecoder.CHANNELS)
+            f.setframerate(OpusDecoder.SAMPLING_RATE)
+            f.writeframes(pcm_data)
         data.seek(0)
+        audio.file = data
         audio.on_format(self.encoding)

discord/utils.py

@@ -58,6 +58,7 @@
     Iterator,
     Literal,
     Mapping,
+    ParamSpec,
     Protocol,
     Sequence,
     TypeVar,
@@ -176,6 +177,8 @@ class _RequestLike(Protocol):
 
 T = TypeVar("T")
 T_co = TypeVar("T_co", covariant=True)
+_MC_P = ParamSpec("_MC_P")
+_MC_T = TypeVar("_MC_T")
 _Iter = Union[Iterator[T], AsyncIterator[T]]
 
 
@@ -880,7 +883,11 @@ def _parse_ratelimit_header(request: Any, *, use_clock: bool = False) -> float:
     return (reset - now).total_seconds()
 
 
-async def maybe_coroutine(f, *args, **kwargs):
+async def maybe_coroutine(
+    f: Callable[_MC_P, _MC_T | Awaitable[_MC_T]],
+    *args: _MC_P.args,
+    **kwargs: _MC_P.kwargs,
+) -> _MC_T:
     value = f(*args, **kwargs)
     if _isawaitable(value):
         return await value

discord/voice/__init__.py

@@ -11,8 +11,8 @@
 from ..errors import MissingVoiceDependenciesError
 from ..utils import get_missing_voice_dependencies
 
-if missing := get_missing_voice_dependencies():
-    raise MissingVoiceDependenciesError(missing=missing)
+if _missing := get_missing_voice_dependencies():
+    raise MissingVoiceDependenciesError(missing=_missing)
 
 from ._types import *
 from .client import *