bodhi-realtime-agent / ClientSenderAdapter

Class: ClientSenderAdapter

Defined in: src/transport/client-sender-adapter.ts:11

Adapts a SessionClientSender (e.g. multi-user WebSocket) to the IClientChannel interface expected by VoiceSession. Used when the server owns the client connection and feeds input explicitly via feedAudioFromClient / feedJsonFromClient.

Implements

• IClientChannel

Constructors

Constructor

new ClientSenderAdapter(sender, supportsPlaybackStateProtocol?): ClientSenderAdapter

Defined in: src/transport/client-sender-adapter.ts:20

Parameters

sender

SessionClientSender

supportsPlaybackStateProtocol?

boolean = false

Returns

ClientSenderAdapter

Properties

supportsPlaybackStateProtocol

readonly supportsPlaybackStateProtocol: boolean

Defined in: src/transport/client-sender-adapter.ts:18

Whether this channel participates in the playback-state protocol. Set by the channel producer (the factory) — not inferred here, since the adapter only sees a SessionClientSender and cannot tell PCM from avatar/RTC.

Implementation of

IClientChannel.supportsPlaybackStateProtocol

Methods

discardBuffered()

discardBuffered(): void

Defined in: src/transport/client-sender-adapter.ts:77

Leave buffering mode and drop the buffered assistant audio without sending it to the client (audio from an abandoned turn).

Returns

void

Implementation of

IClientChannel.discardBuffered

---

sendAudioToClient()

sendAudioToClient(data): void

Defined in: src/transport/client-sender-adapter.ts:34

Parameters

data

Buffer

Returns

void

Implementation of

IClientChannel.sendAudioToClient

---

sendJsonAfterAudio()

sendJsonAfterAudio(message): void

Defined in: src/transport/client-sender-adapter.ts:52

Playback-state protocol: deliver JSON in order after the turn's audio. The underlying sender writes audio and JSON to the one client socket in call order, so this is sendJsonToClient after the audio sends. During a reconnect-buffering window audio is buffered for replay while JSON would send immediately — sending audio.done then would let it overtake the buffered audio, so it is dropped and the turn falls to the fallback.

Parameters

message

AnyServerToClientMessage

Returns

void

Implementation of

IClientChannel.sendJsonAfterAudio

---

sendJsonToClient()

sendJsonToClient(message): void

Defined in: src/transport/client-sender-adapter.ts:42

Parameters

message

AnyServerToClientMessage

Returns

void

Implementation of

IClientChannel.sendJsonToClient

---

start()

start(): Promise<void>

Defined in: src/transport/client-sender-adapter.ts:25

Returns

Promise<void>

Implementation of

IClientChannel.start

---

startBuffering()

startBuffering(): void

Defined in: src/transport/client-sender-adapter.ts:57

Returns

void

Implementation of

IClientChannel.startBuffering

---

stop()

stop(): Promise<void>

Defined in: src/transport/client-sender-adapter.ts:29

Returns

Promise<void>

Implementation of

IClientChannel.stop

---

stopBuffering()

stopBuffering(): Buffer<ArrayBufferLike>[]

Defined in: src/transport/client-sender-adapter.ts:62

Returns

Buffer<ArrayBufferLike>[]

Implementation of

IClientChannel.stopBuffering