josedario87
65303df96a
Add Samsung proprietary Face Widget (lock screen/AOD) with terminal status display. Add voice interaction services (AgentVoiceInteractionService, RecognitionService) for digital assistant registration. Add PiP mode with voice/expand actions. Add session-state proxy, voice transcript routes, window controls component. Ignore installers/ directory.
106 lines
3.5 KiB
TypeScript
106 lines
3.5 KiB
TypeScript
import { jsonResponse } from '../utils/cors'
|
||
import { PORT_TERMINAL } from '../config'
|
||
|
||
export async function handleVoiceTranscript(req: Request): Promise<Response | null> {
|
||
if (req.method !== 'POST') return null
|
||
|
||
try {
|
||
const body = await req.json() as { text?: string; timestamp?: string; source?: string }
|
||
const { text, timestamp, source } = body
|
||
|
||
if (!text) {
|
||
return jsonResponse({ error: 'Missing "text" field' }, 400)
|
||
}
|
||
|
||
const ts = timestamp || new Date().toISOString()
|
||
const src = source || 'android-voice'
|
||
|
||
console.log(`\n🎙️ [VOICE TRANSCRIPT] ────────────────────────`)
|
||
console.log(` Source: ${src}`)
|
||
console.log(` Time: ${ts}`)
|
||
console.log(` Text: "${text}"`)
|
||
|
||
// Find first alive terminal and send the text as input
|
||
const result = await sendToFirstTerminal(text)
|
||
|
||
console.log(` Terminal: ${result.terminal || 'none found'}`)
|
||
console.log(` Status: ${result.sent ? 'sent ✓' : result.error || 'no terminal'}`)
|
||
console.log(` ──────────────────────────────────────────────\n`)
|
||
|
||
return jsonResponse({
|
||
ok: true,
|
||
received: text,
|
||
timestamp: ts,
|
||
source: src,
|
||
sentToTerminal: result.sent,
|
||
terminal: result.terminal || null,
|
||
ephemeralSessionId: result.ephemeralSessionId || null
|
||
})
|
||
} catch (e: any) {
|
||
console.error('[voice-transcript] Parse error:', e.message)
|
||
return jsonResponse({ error: 'Invalid JSON body' }, 400)
|
||
}
|
||
}
|
||
|
||
async function sendToFirstTerminal(text: string): Promise<{ sent: boolean; terminal?: string; ephemeralSessionId?: string; error?: string }> {
|
||
try {
|
||
// Fetch terminal registry to find alive terminals
|
||
const res = await fetch(`http://localhost:${PORT_TERMINAL}/terminal-registry`)
|
||
if (!res.ok) {
|
||
return { sent: false, error: `registry fetch failed: ${res.status}` }
|
||
}
|
||
|
||
const registry = await res.json() as Array<{
|
||
ephemeralSessionId: string
|
||
agent: string
|
||
label: string
|
||
alive: boolean
|
||
}>
|
||
|
||
// Find first alive terminal
|
||
const target = registry.find(t => t.alive)
|
||
if (!target) {
|
||
return { sent: false, error: 'no alive terminals' }
|
||
}
|
||
|
||
// Connect via WebSocket and send the text as input
|
||
const wsUrl = `ws://localhost:${PORT_TERMINAL}/ws/terminal?session=${target.ephemeralSessionId}`
|
||
|
||
return new Promise((resolve) => {
|
||
const ws = new WebSocket(wsUrl)
|
||
const timeout = setTimeout(() => {
|
||
try { ws.close() } catch {}
|
||
resolve({ sent: false, terminal: target.ephemeralSessionId, error: 'ws timeout' })
|
||
}, 5000)
|
||
|
||
ws.onopen = () => {
|
||
// Send the transcribed text
|
||
ws.send(JSON.stringify({ type: 'input', data: text }))
|
||
|
||
// Send Enter after a short delay
|
||
setTimeout(() => {
|
||
ws.send(JSON.stringify({ type: 'input', data: '\r' }))
|
||
|
||
// Close after sending
|
||
setTimeout(() => {
|
||
clearTimeout(timeout)
|
||
ws.close()
|
||
resolve({
|
||
sent: true,
|
||
terminal: `${target.ephemeralSessionId} (${target.agent})`,
|
||
ephemeralSessionId: target.ephemeralSessionId
|
||
})
|
||
}, 150)
|
||
}, 80)
|
||
}
|
||
|
||
ws.onerror = (err) => {
|
||
clearTimeout(timeout)
|
||
resolve({ sent: false, terminal: target.ephemeralSessionId, error: 'ws connection error' })
|
||
}
|
||
})
|
||
} catch (e: any) {
|
||
return { sent: false, error: e.message }
|
||
}
|
||
}
|