pretheme

s2s-client/ws/s2s-ws-client.js

@@ -162,6 +162,20 @@ export class S2sWsRealtimeClient extends EventTarget {
     this._joinTimer = 0;
     /** @type {NoiseGate} Mic noise gate; off by default. */
     this._noiseGate = options.noiseGate ?? { enabled: false, thresholdDb: -45 };
+    /** @type {boolean} If true, right after the initial session.updated ack we inject
+     *  the wake-word-style bootstrap phrase (default "hola vocero") as if the user
+     *  had spoken it, then trigger `requestResponse()` — so the assistant
+     *  behaves EXACTLY as it does when the "Hola Vocero" hot-word is detected
+     *  (same code path, same greeting style). Used when S2S_START_SESSION_MODE=TALK. */
+    this._startWithResponse = options.startWithResponse === true;
+    /** @type {string} Phrase injected into the first turn (mirrors the wake-word
+     *  confirmation flow. Defaults to "hola vocero". Used ONLY on first session start.
+     *  Overridable via options.startWithResponseText for test/CI flows. */
+    this._startWithResponseText = (options.startWithResponseText || "").trim() || "hola vocero";
+    /** @type {boolean} Guards the initial bootstrap turn from firing more than once
+     *  (e.g. after a live `updateSession()` that re-emits session.updated later
+     *  while the session is already running). */
+    this._initialResponseSent = false;
     /** @type {WebSocket | null} */
     this._ws = null;
     /** @type {AudioContext | null} */
@@ -671,6 +685,32 @@ export class S2sWsRealtimeClient extends EventTarget {
             `tools=${toolNames || "none"}`
           );
         }
+        // Primo ack del server: se startup mode=TALK vogliamo il comportamento
+        // IDENTICO a quando arriva la frase di wake word "Hola Vocero".
+        // Quindi NON mandiamo shaped response.create con istruzioni di greet
+        // hardcodate, ma invece:
+        //   1. injectiamo "hola vocero" nella conversazione come se l'utente
+        //      l'avesse detto (sendUserText → conversation.item.create user)
+        //   2. requestResponse() VUOTO → il modello risponde esattamente come
+        //      nel codice wake-word in main.js.
+        // Stesso identico risultato, nessuna istruzione custom.
+        if (this._startWithResponse && !this._initialResponseSent) {
+          this._initialResponseSent = true;
+          const phrase = this._startWithResponseText.trim() || "hola vocero";
+          // #region debug-point B:start-session-initial-response
+          reportWsDebug("B", "s2s-ws-client.js:session.updated",
+            "emitting initial turn via wake-word style sendUserText + requestResponse() (S2S_START_SESSION_MODE=TALK)",
+            {
+              toolCount: Array.isArray(event.session?.tools) ? event.session.tools.length : 0,
+              phrase,
+            });
+          // #endregion
+          if (this._debug) {
+            console.debug(`[Vocero WS] bootstrap turn: sendUserText(${JSON.stringify(phrase)}) → requestResponse()`);
+          }
+          this.sendUserText(phrase);
+          this.requestResponse();
+        }
         // Acknowledged by server, nothing to do.
         break;