Skip to content

The agent that grows with you

fix(gateway): keep plain audio MEDIA out of WhatsApp/BlueBubbles voice notes

Teknium183cd6cc35fb
diff --git a/plugins/platforms/bluebubbles/adapter.py b/plugins/platforms/bluebubbles/adapter.py
index eb6c8987bd6..628ee8e379f 100644
--- a/plugins/platforms/bluebubbles/adapter.py
+++ b/plugins/platforms/bluebubbles/adapter.py
@@ -428,8 +428,10 @@ class BlueBubblesAdapter(BasePlatformAdapter):
async def send_image_file(self, chat_id, image_path, caption=None, reply_to=None, **kw) -> SendResult:
return await self._send_attachment(chat_id, image_path, caption=caption)
- async def send_voice(self, chat_id, audio_path, caption=None, reply_to=None, **kw) -> SendResult:
- return await self._send_attachment(chat_id, audio_path, caption=caption, is_audio_message=True)
+ async def send_voice(self, chat_id, audio_path, caption=None, reply_to=None, is_voice=None, **kw) -> SendResult:
+ # is_voice=False is a plain audio MEDIA file (music, a sound effect): a normal attachment,
+ # never an iMessage voice memo.
+ return await self._send_attachment(chat_id, audio_path, caption=caption, is_audio_message=is_voice is not False)
async def send_video(self, chat_id, video_path, caption=None, reply_to=None, **kw) -> SendResult:
return await self._send_attachment(chat_id, video_path, caption=caption)
diff --git a/plugins/platforms/whatsapp/adapter.py b/plugins/platforms/whatsapp/adapter.py
index f40f8e6375d..04177d1f063 100644
--- a/plugins/platforms/whatsapp/adapter.py
+++ b/plugins/platforms/whatsapp/adapter.py
@@ -873,7 +873,12 @@ class WhatsAppAdapter(WhatsAppBehaviorMixin, BasePlatformAdapter):
async def send_video(self, chat_id: str, video_path: str, caption: Optional[str] = None, reply_to: Optional[str] = None, **kwargs) -> SendResult:
return await self._send_media_to_bridge(chat_id, video_path, "video", caption)
- async def send_voice(self, chat_id: str, audio_path: str, caption: Optional[str] = None, reply_to: Optional[str] = None, **kwargs) -> SendResult:
+ async def send_voice(self, chat_id: str, audio_path: str, caption: Optional[str] = None, reply_to: Optional[str] = None,
+ is_voice: Optional[bool] = None, **kwargs) -> SendResult:
+ # The bridge turns every "audio" send into a voice note (ptt, transcoded to Opus); a plain
+ # audio MEDIA file (is_voice=False: music, a sound effect) goes as a named document instead.
+ if is_voice is False:
+ return await self.send_document(chat_id, audio_path, caption=caption, reply_to=reply_to)
return await self._send_media_to_bridge(chat_id, audio_path, "audio", caption)
async def send_document(self, chat_id: str, file_path: str, caption: Optional[str] = None, file_name: Optional[str] = None,
diff --git a/plugins/platforms/whatsapp_cloud/adapter.py b/plugins/platforms/whatsapp_cloud/adapter.py
index ce81898f130..0bc2b619635 100644
--- a/plugins/platforms/whatsapp_cloud/adapter.py
+++ b/plugins/platforms/whatsapp_cloud/adapter.py
@@ -612,11 +612,16 @@ class WhatsAppCloudAdapter(WhatsAppBehaviorMixin, BasePlatformAdapter):
chat_id, file_path, "document", caption=caption, filename=file_name or os.path.basename(file_path), reply_to=reply_to,
)
- async def send_voice(self, chat_id: str, audio_path: str, caption: Optional[str] = None, reply_to: Optional[str] = None, **kwargs) -> SendResult:
+ async def send_voice(self, chat_id: str, audio_path: str, caption: Optional[str] = None, reply_to: Optional[str] = None,
+ is_voice: Optional[bool] = None, **kwargs) -> SendResult:
"""Voice message: ``audio/ogg; codecs=opus`` renders as a voice bubble, so a
- local MP3 (Hermes TTS output) is converted via ffmpeg first; other audio is sent as-is."""
+ local MP3 (Hermes TTS output) is converted via ffmpeg first; other audio is sent as-is.
+ A plain audio MEDIA file (``is_voice=False``) is never converted into a voice bubble."""
mime_type: Optional[str] = None
- if not audio_path.startswith(_HTTP_PREFIXES) and audio_path.lower().endswith(".mp3") and os.path.exists(audio_path):
+ if is_voice is False and audio_path.lower().endswith(".mp3"):
+ mime_type = "audio/mpeg"
+ elif (not audio_path.startswith(_HTTP_PREFIXES) and audio_path.lower().endswith(".mp3")
+ and await asyncio.to_thread(os.path.exists, audio_path)):
opus_path = await self._convert_to_opus(audio_path)
if opus_path:
try:
diff --git a/tests/gateway/test_plain_audio_not_voice_note.py b/tests/gateway/test_plain_audio_not_voice_note.py
new file mode 100644
index 00000000000..4321821b6c3
--- /dev/null
+++ b/tests/gateway/test_plain_audio_not_voice_note.py
@@ -0,0 +1,68 @@
+"""A plain audio MEDIA file (``is_voice=False``: music, a sound effect) is never sent as a voice note.
+
+The media dispatch routes every audio extension to ``send_voice(..., is_voice=<bool>)``; adapters
+whose ``send_voice`` is a native voice-note transport must honour the flag. Each case records the
+adapter's own wire boundary.
+"""
+from types import SimpleNamespace
+from unittest.mock import AsyncMock, MagicMock
+
+import pytest
+
+from gateway.config import PlatformConfig
+
+
[email protected]("is_voice, expected", [(False, "document"), (None, "audio")], ids=["plain", "voice"])
+async def test_whatsapp_bridge_sends_plain_audio_as_a_document(tmp_path, is_voice, expected):
+ from plugins.platforms.whatsapp.adapter import WhatsAppAdapter
+
+ adapter = WhatsAppAdapter(PlatformConfig(enabled=True, extra={"session_path": str(tmp_path / "wa")}))
+ posted = []
+
+ async def post_bridge_message(route, payload, timeout=0):
+ posted.append(payload)
+ return SimpleNamespace(success=True, message_id=None)
+
+ adapter._post_bridge_message = post_bridge_message
+ adapter._bridge_unavailable = AsyncMock(return_value=None)
+ path = tmp_path / "door_knock.mp3"
+ path.write_bytes(b"ID3\x04" + b"\x00" * 32)
+
+ assert (await adapter.send_voice("15551234567", str(path), is_voice=is_voice)).success
+ assert [p["mediaType"] for p in posted] == [expected]
+
+
[email protected]("is_voice, audio_message", [(False, False), (None, True)], ids=["plain", "voice"])
+async def test_bluebubbles_sends_plain_audio_without_the_voice_memo_flag(tmp_path, monkeypatch, is_voice, audio_message):
+ from plugins.platforms.bluebubbles.adapter import BlueBubblesAdapter
+
+ adapter = BlueBubblesAdapter(PlatformConfig(enabled=True, extra={"server_url": "http://localhost:1234", "password": "x"}))
+ response = MagicMock()
+ response.json.return_value = {"status": 200, "data": {"guid": "g1"}}
+ adapter.client = SimpleNamespace(post=AsyncMock(return_value=response))
+ monkeypatch.setattr(adapter, "_resolve_chat_guid", AsyncMock(return_value="iMessage;-;+15551234567"))
+ path = tmp_path / "door_knock.mp3"
+ path.write_bytes(b"ID3\x04" + b"\x00" * 32)
+
+ assert (await adapter.send_voice("+15551234567", str(path), is_voice=is_voice)).success
+ data = adapter.client.post.await_args.kwargs["data"]
+ assert ("isAudioMessage" in data) is audio_message
+ assert data["name"] == "door_knock.mp3"
+
+
+async def test_whatsapp_cloud_never_converts_plain_mp3_into_a_voice_bubble(tmp_path):
+ from plugins.platforms.whatsapp_cloud.adapter import WhatsAppCloudAdapter
+
+ adapter = WhatsAppCloudAdapter.__new__(WhatsAppCloudAdapter)
+ adapter._convert_to_opus = AsyncMock(side_effect=AssertionError("plain audio must not become Opus"))
+ adapter._send_media_from_path_or_link = AsyncMock(return_value=SimpleNamespace(success=True))
+ path = tmp_path / "door_knock.mp3"
+ path.write_bytes(b"ID3\x04" + b"\x00" * 32)
+
+ assert (await adapter.send_voice("15551234567", str(path), is_voice=False)).success
+ call = adapter._send_media_from_path_or_link.await_args
+ assert call.args[1] == str(path)
+ assert call.kwargs["mime_type"] == "audio/mpeg"