Add get/set/config messages for voice assistant (#956)

aioesphomeapi/api.proto

@@ -1589,6 +1589,36 @@ message VoiceAssistantAnnounceFinished {
   bool success = 1;
 }
 
+message VoiceAssistantWakeWord {
+  uint32 id = 1;
+  string wake_word = 2;
+  repeated string trained_languages = 3;
+}
+
+message VoiceAssistantConfigurationRequest {
+  option (id) = 121;
+  option (source) = SOURCE_CLIENT;
+  option (ifdef) = "USE_VOICE_ASSISTANT";
+}
+
+message VoiceAssistantConfigurationResponse {
+  option (id) = 122;
+  option (source) = SOURCE_SERVER;
+  option (ifdef) = "USE_VOICE_ASSISTANT";
+
+  repeated VoiceAssistantWakeWord available_wake_words = 1;
+  repeated uint32 active_wake_words = 2;
+  uint32 max_active_wake_words = 3;
+}
+
+message VoiceAssistantSetConfiguration {
+  option (id) = 123;
+  option (source) = SOURCE_CLIENT;
+  option (ifdef) = "USE_VOICE_ASSISTANT";
+
+  repeated uint32 active_wake_words = 1;
+}
+
 // ==================== ALARM CONTROL PANEL ====================
 enum AlarmControlPanelState {
   ALARM_STATE_DISARMED = 0;

aioesphomeapi/client.py

@@ -73,10 +73,13 @@
     VoiceAssistantAnnounceFinished,
     VoiceAssistantAnnounceRequest,
     VoiceAssistantAudio,
+    VoiceAssistantConfigurationRequest,
+    VoiceAssistantConfigurationResponse,
     VoiceAssistantEventData,
     VoiceAssistantEventResponse,
     VoiceAssistantRequest,
     VoiceAssistantResponse,
+    VoiceAssistantSetConfiguration,
     VoiceAssistantTimerEventResponse,
 )
 from .client_callbacks import (
@@ -133,6 +136,7 @@
     VoiceAssistantAudioData,
     VoiceAssistantAudioSettings as VoiceAssistantAudioSettingsModel,
     VoiceAssistantCommand,
+    VoiceAssistantConfigurationResponse as VoiceAssistantConfigurationResponseModel,
     VoiceAssistantEventType,
     VoiceAssistantSubscriptionFlag,
     VoiceAssistantTimerEventType,
@@ -1459,6 +1463,22 @@ async def send_voice_assistant_announcement_await_response(
         )
         return VoiceAssistantAnnounceFinishedModel.from_pb(resp)
 
+    async def get_voice_assistant_configuration(
+        self, timeout: float
+    ) -> VoiceAssistantConfigurationResponseModel:
+        resp = await self._get_connection().send_message_await_response(
+            VoiceAssistantConfigurationRequest(),
+            VoiceAssistantConfigurationResponse,
+            timeout,
+        )
+        return VoiceAssistantConfigurationResponseModel.from_pb(resp)
+
+    async def set_voice_assistant_configuration(
+        self, active_wake_words: list[int]
+    ) -> None:
+        req = VoiceAssistantSetConfiguration(active_wake_words=active_wake_words)
+        self._get_connection().send_message(req)
+
     def alarm_control_panel_command(
         self,
         key: int,

aioesphomeapi/core.py

@@ -121,9 +121,12 @@
     VoiceAssistantAnnounceFinished,
     VoiceAssistantAnnounceRequest,
     VoiceAssistantAudio,
+    VoiceAssistantConfigurationRequest,
+    VoiceAssistantConfigurationResponse,
     VoiceAssistantEventResponse,
     VoiceAssistantRequest,
     VoiceAssistantResponse,
+    VoiceAssistantSetConfiguration,
     VoiceAssistantTimerEventResponse,
 )
 
@@ -396,6 +399,9 @@ def __init__(self, error: BluetoothGATTError) -> None:
     118: UpdateCommandRequest,
     119: VoiceAssistantAnnounceRequest,
     120: VoiceAssistantAnnounceFinished,
+    121: VoiceAssistantConfigurationRequest,
+    122: VoiceAssistantConfigurationResponse,
+    123: VoiceAssistantSetConfiguration,
 }
 
 MESSAGE_NUMBER_TO_PROTO = tuple(MESSAGE_TYPE_TO_PROTO.values())

aioesphomeapi/model.py

@@ -1298,6 +1298,42 @@ class VoiceAssistantAnnounceFinished(APIModelBase):
     success: bool = False
 
 
+@_frozen_dataclass_decorator
+class VoiceAssistantWakeWord(APIModelBase):
+    id: int
+    wake_word: str
+    trained_languages: list[str]
+
+    @classmethod
+    def convert_list(cls, value: list[Any]) -> list[VoiceAssistantWakeWord]:
+        ret = []
+        for x in value:
+            if isinstance(x, dict):
+                ret.append(VoiceAssistantWakeWord.from_dict(x))
+            else:
+                ret.append(VoiceAssistantWakeWord.from_pb(x))
+        return ret
+
+
+@_frozen_dataclass_decorator
+class VoiceAssistantConfigurationResponse(APIModelBase):
+    available_wake_words: list[VoiceAssistantWakeWord] = converter_field(
+        default_factory=list, converter=VoiceAssistantWakeWord.convert_list
+    )
+    active_wake_words: list[int] = converter_field(default_factory=list, converter=list)
+    max_active_wake_words: int = 0
+
+
+@_frozen_dataclass_decorator
+class VoiceAssistantConfigurationRequest(APIModelBase):
+    pass
+
+
+@_frozen_dataclass_decorator
+class VoiceAssistantSetConfiguration(APIModelBase):
+    active_wake_words: list[int] = converter_field(default_factory=list, converter=list)
+
+
 class LogLevel(APIIntEnum):
     LOG_LEVEL_NONE = 0
     LOG_LEVEL_ERROR = 1

tests/test_client.py

@@ -71,11 +71,15 @@
     VoiceAssistantAnnounceRequest,
     VoiceAssistantAudio,
     VoiceAssistantAudioSettings,
+    VoiceAssistantConfigurationRequest,
+    VoiceAssistantConfigurationResponse,
     VoiceAssistantEventData,
     VoiceAssistantEventResponse,
     VoiceAssistantRequest,
     VoiceAssistantResponse,
+    VoiceAssistantSetConfiguration,
     VoiceAssistantTimerEventResponse,
+    VoiceAssistantWakeWord,
 )
 from aioesphomeapi.client import APIClient, BluetoothConnectionDroppedError
 from aioesphomeapi.connection import APIConnection
@@ -113,6 +117,7 @@
     UserServiceArgType,
     VoiceAssistantAnnounceFinished as VoiceAssistantAnnounceFinishedModel,
     VoiceAssistantAudioSettings as VoiceAssistantAudioSettingsModel,
+    VoiceAssistantConfigurationResponse as VoiceAssistantConfigurationResponseModel,
     VoiceAssistantEventType as VoiceAssistantEventModelType,
     VoiceAssistantTimerEventType as VoiceAssistantTimerEventModelType,
 )
@@ -2660,6 +2665,57 @@ async def handle_announcement_finished(
     assert len(send.mock_calls) == 0
 
 
+@pytest.mark.asyncio
+async def test_get_voice_assistant_configuration(
+    api_client: tuple[
+        APIClient, APIConnection, asyncio.Transport, APIPlaintextFrameHelper
+    ],
+) -> None:
+    client, connection, _transport, protocol = api_client
+    original_send_message = connection.send_message
+
+    def send_message(msg):
+        assert msg == VoiceAssistantConfigurationRequest()
+        original_send_message(msg)
+
+    with patch.object(connection, "send_message", new=send_message):
+        config_task = asyncio.create_task(
+            client.get_voice_assistant_configuration(timeout=1.0)
+        )
+        await asyncio.sleep(0)
+        response: message.Message = VoiceAssistantConfigurationResponse(
+            available_wake_words=[
+                VoiceAssistantWakeWord(
+                    id=1,
+                    wake_word="okay nabu",
+                    trained_languages=["en"],
+                )
+            ],
+            active_wake_words=[1],
+            max_active_wake_words=1,
+        )
+        mock_data_received(protocol, generate_plaintext_packet(response))
+        config = await config_task
+        assert isinstance(config, VoiceAssistantConfigurationResponseModel)
+
+
+@pytest.mark.asyncio
+async def test_set_voice_assistant_configuration(
+    api_client: tuple[
+        APIClient, APIConnection, asyncio.Transport, APIPlaintextFrameHelper
+    ],
+) -> None:
+    client, connection, _transport, protocol = api_client
+    original_send_message = connection.send_message
+
+    def send_message(msg):
+        assert msg == VoiceAssistantSetConfiguration(active_wake_words=[1])
+        original_send_message(msg)
+
+    with patch.object(connection, "send_message", new=send_message):
+        await client.set_voice_assistant_configuration([1])
+
+
 @pytest.mark.asyncio
 async def test_api_version_after_connection_closed(
     api_client: tuple[

tests/test_model.py

@@ -118,7 +118,9 @@
     UserServiceArgType,
     ValveInfo,
     ValveState,
+    VoiceAssistantConfigurationResponse,
     VoiceAssistantFeature,
+    VoiceAssistantWakeWord,
     build_unique_id,
     converter_field,
 )
@@ -681,3 +683,30 @@ def test_media_player_supported_format_convert_list() -> None:
             )
         ],
     )
+
+
+def test_voice_assistant_wake_word_convert_list() -> None:
+    """Test list conversion for VoiceAssistantWakeWord."""
+    assert VoiceAssistantConfigurationResponse.from_dict(
+        {
+            "available_wake_words": [
+                {
+                    "id": 1,
+                    "wake_word": "okay nabu",
+                    "trained_languages": ["en"],
+                }
+            ],
+            "active_wake_words": [1],
+            "max_active_wake_words": 1,
+        }
+    ) == VoiceAssistantConfigurationResponse(
+        available_wake_words=[
+            VoiceAssistantWakeWord(
+                id=1,
+                wake_word="okay nabu",
+                trained_languages=["en"],
+            )
+        ],
+        active_wake_words=[1],
+        max_active_wake_words=1,
+    )