feat: LiveKit voice session, control channel, transcripts (Phase 2)
- VoiceSession: token->connect with fresh-room retry on timeout, agent discovery, agent_ready wait, mic enable/mute, text send over lk.chat - ControlChannel: outbound set_voice/set_model/get_models/set_mode/ media_state; inbound agent_ready/models/model_changed/task_event - TranscriptHandler: transcription segments + transcript topic finals, partial/final dedupe per role - ConversationView/OrbView: status orb, mic/text-mode toggles, composer - ServerConfig: plain ws:// signaling for private IPs (self-signed TLS not injectable into SDK WebSocket); Info.plist local networking
This commit is contained in:
snowspeeder committed
1 parent
378b8cb36f
commit
4986b6d7dc
14 files changed
+900
-50
No files matched your search
@@ -2,33 +2,161 @@ import SwiftUI
|
||||
|
||||
struct ConversationView: View {
|
||||
@ObservedObject var settings: SettingsStore
|
||||
@StateObject private var session: VoiceSession
|
||||
@State private var composerText = ""
|
||||
@State private var connectError: String?
|
||||
|
||||
init(settings: SettingsStore) {
|
||||
self.settings = settings
|
||||
let config = ServerConfig.parse(settings.trimmedServerURL)
|
||||
if case .success(let value) = config {
|
||||
_session = StateObject(wrappedValue: VoiceSession(
|
||||
config: value,
|
||||
tokenClient: TokenClient(config: value)
|
||||
))
|
||||
} else {
|
||||
let fallback = ServerConfig.parse("https://192.168.86.2:9178")
|
||||
if case .success(let value) = fallback {
|
||||
_session = StateObject(wrappedValue: VoiceSession(
|
||||
config: value,
|
||||
tokenClient: TokenClient(config: value)
|
||||
))
|
||||
} else {
|
||||
fatalError("Unreachable: default server URL is valid")
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
var body: some View {
|
||||
NavigationStack {
|
||||
VStack(spacing: 24) {
|
||||
Spacer()
|
||||
Circle()
|
||||
.fill(.blue.gradient)
|
||||
.frame(width: 96, height: 96)
|
||||
Text("Hope Voice")
|
||||
.font(.title.bold())
|
||||
Text(settings.trimmedServerURL)
|
||||
.font(.footnote)
|
||||
.foregroundStyle(.secondary)
|
||||
Spacer()
|
||||
VStack(spacing: 0) {
|
||||
statusHeader
|
||||
TranscriptListView(events: session.transcriptHandler.events)
|
||||
composer
|
||||
}
|
||||
.navigationTitle("Voice")
|
||||
.toolbar {
|
||||
ToolbarItem(placement: .topBarTrailing) {
|
||||
NavigationLink {
|
||||
SettingsView(settings: settings)
|
||||
} label: {
|
||||
Image(systemName: "gearshape")
|
||||
}
|
||||
.navigationTitle("Hope Voice")
|
||||
.toolbar { toolbarContent }
|
||||
.onAppear {
|
||||
if case .failure(let error) = ServerConfig.parse(settings.trimmedServerURL) {
|
||||
connectError = error.localizedDescription
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
private var statusHeader: some View {
|
||||
VStack(spacing: 12) {
|
||||
OrbView(state: orbState)
|
||||
Text(statusText)
|
||||
.font(.headline)
|
||||
if let agent = session.agent {
|
||||
Text("Agent: \(agent.name ?? agent.identity)")
|
||||
.font(.caption)
|
||||
.foregroundStyle(.secondary)
|
||||
}
|
||||
if session.state == .ready {
|
||||
HStack(spacing: 16) {
|
||||
Toggle(isOn: Binding(
|
||||
get: { !session.isMuted },
|
||||
set: { newValue in Task { await session.setMuted(!newValue) } }
|
||||
)) {
|
||||
Label(session.isMuted ? "Muted" : "Mic on", systemImage: session.isMuted ? "mic.slash" : "mic")
|
||||
}
|
||||
.toggleStyle(.button)
|
||||
Toggle("Text mode", isOn: $settings.textMode)
|
||||
.toggleStyle(.button)
|
||||
}
|
||||
}
|
||||
if let connectError {
|
||||
Label(connectError, systemImage: "exclamationmark.triangle")
|
||||
.font(.footnote)
|
||||
.foregroundStyle(.red)
|
||||
}
|
||||
}
|
||||
.padding()
|
||||
}
|
||||
|
||||
private var composer: some View {
|
||||
HStack(spacing: 8) {
|
||||
switch session.state {
|
||||
case .idle, .failed:
|
||||
Button(action: startSession) {
|
||||
Label("Start", systemImage: "phone.fill")
|
||||
}
|
||||
.buttonStyle(.borderedProminent)
|
||||
case .connecting, .waitingAgent:
|
||||
ProgressView()
|
||||
case .ready:
|
||||
TextField(settings.textMode ? "Type a message…" : "Voice mode active",
|
||||
text: $composerText,
|
||||
axis: .vertical)
|
||||
.textFieldStyle(.roundedBorder)
|
||||
.lineLimit(1...4)
|
||||
.disabled(!settings.textMode)
|
||||
Button(action: sendComposer) {
|
||||
Image(systemName: "arrow.up.circle.fill")
|
||||
.font(.title2)
|
||||
}
|
||||
.disabled(!settings.textMode || composerText.isEmpty)
|
||||
Button(role: .destructive, action: endSession) {
|
||||
Image(systemName: "phone.down.fill")
|
||||
}
|
||||
}
|
||||
}
|
||||
.padding()
|
||||
}
|
||||
|
||||
private var toolbarContent: some ToolbarContent {
|
||||
ToolbarItem(placement: .topBarTrailing) {
|
||||
NavigationLink {
|
||||
SettingsView(settings: settings)
|
||||
} label: {
|
||||
Image(systemName: "gearshape")
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
private var orbState: OrbView.OrbState {
|
||||
switch session.state {
|
||||
case .idle: .idle
|
||||
case .connecting, .waitingAgent: .connecting
|
||||
case .ready: settings.textMode ? .idle : .listening
|
||||
case .failed: .failed
|
||||
}
|
||||
}
|
||||
|
||||
private var statusText: String {
|
||||
switch session.state {
|
||||
case .idle: "Ready to connect"
|
||||
case .connecting: "Connecting…"
|
||||
case .waitingAgent: "Waiting for agent…"
|
||||
case .ready: settings.textMode ? "Text mode" : "Listening"
|
||||
case .failed(let message): "Failed: \(message)"
|
||||
}
|
||||
}
|
||||
|
||||
private func startSession() {
|
||||
connectError = nil
|
||||
session.transcriptHandler.clear()
|
||||
Task { await session.connect() }
|
||||
}
|
||||
|
||||
private func endSession() {
|
||||
Task { await session.disconnect() }
|
||||
}
|
||||
|
||||
private func sendComposer() {
|
||||
let text = composerText.trimmingCharacters(in: .whitespacesAndNewlines)
|
||||
guard !text.isEmpty else { return }
|
||||
composerText = ""
|
||||
Task {
|
||||
do {
|
||||
try await session.sendText(text)
|
||||
} catch {
|
||||
connectError = error.localizedDescription
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
#Preview {
|
||||
|
||||
@@ -0,0 +1,101 @@
|
||||
import SwiftUI
|
||||
|
||||
struct OrbView: View {
|
||||
enum OrbState {
|
||||
case idle
|
||||
case connecting
|
||||
case listening
|
||||
case speaking
|
||||
case thinking
|
||||
case failed
|
||||
}
|
||||
|
||||
let state: OrbState
|
||||
|
||||
@State private var animating = false
|
||||
|
||||
private var baseColor: Color {
|
||||
switch state {
|
||||
case .idle, .connecting: .blue
|
||||
case .listening: .green
|
||||
case .speaking, .thinking: .orange
|
||||
case .failed: .red
|
||||
}
|
||||
}
|
||||
|
||||
private var scale: CGFloat {
|
||||
switch state {
|
||||
case .connecting, .listening, .speaking, .thinking: 1.1
|
||||
default: 1.0
|
||||
}
|
||||
}
|
||||
|
||||
var body: some View {
|
||||
let shape = Circle()
|
||||
.fill(baseColor.gradient)
|
||||
.frame(width: 120, height: 120)
|
||||
let animated = shape
|
||||
.scaleEffect(animating ? scale : 0.95)
|
||||
.opacity(state == .idle || state == .failed ? 0.7 : 1.0)
|
||||
let anim: Animation? = state == .idle || state == .failed
|
||||
? nil
|
||||
: .easeInOut(duration: 0.8).repeatForever(autoreverses: true)
|
||||
animated
|
||||
.animation(anim, value: animating)
|
||||
.onAppear { animating = true }
|
||||
.onChange(of: state) { _, _ in animating = true }
|
||||
}
|
||||
}
|
||||
|
||||
struct TranscriptListView: View {
|
||||
let events: [TranscriptEvent]
|
||||
|
||||
var body: some View {
|
||||
ScrollViewReader { proxy in
|
||||
ScrollView {
|
||||
LazyVStack(alignment: .leading, spacing: 8) {
|
||||
ForEach(events) { event in
|
||||
transcriptRow(event)
|
||||
}
|
||||
}
|
||||
.padding(.horizontal)
|
||||
}
|
||||
.onChange(of: events.count) { _, count in
|
||||
if let last = events.last {
|
||||
withAnimation { proxy.scrollTo(last.id, anchor: .bottom) }
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
private func transcriptRow(_ event: TranscriptEvent) -> some View {
|
||||
let roleName = event.role == .user ? "You" : "Hope"
|
||||
let nameColor: Color = event.role == .user ? .secondary : .blue
|
||||
let rowBackground: Color = event.role == .user
|
||||
? Color(.secondarySystemBackground)
|
||||
: Color.blue.opacity(0.08)
|
||||
let textOpacity: Double = event.isFinal ? 1.0 : 0.55
|
||||
return VStack(alignment: .leading, spacing: 2) {
|
||||
Text(roleName)
|
||||
.font(.caption2.bold())
|
||||
.foregroundStyle(nameColor)
|
||||
Text(stripped(event.text))
|
||||
.font(.body)
|
||||
.opacity(textOpacity)
|
||||
}
|
||||
.frame(maxWidth: .infinity, alignment: .leading)
|
||||
.padding(8)
|
||||
.background(rowBackground)
|
||||
.cornerRadius(10)
|
||||
.id(event.id)
|
||||
}
|
||||
|
||||
private func stripped(_ text: String) -> String {
|
||||
// Strip delivery-style markers like [appreciative], [pauses] etc.
|
||||
text.replacingOccurrences(
|
||||
of: #"\[[a-zA-Z ,'-]+\]"#,
|
||||
with: "",
|
||||
options: .regularExpression
|
||||
).trimmingCharacters(in: .whitespaces)
|
||||
}
|
||||
}
|
||||
Reference in new issue
Block a user