@craftedxp/voice-js 0.6.0 → 0.9.0

package/dist/browser.js

@@ -1,7 +1,9 @@
 "use strict";
+var __create = Object.create;
 var __defProp = Object.defineProperty;
 var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
 var __getOwnPropNames = Object.getOwnPropertyNames;
+var __getProtoOf = Object.getPrototypeOf;
 var __hasOwnProp = Object.prototype.hasOwnProperty;
 var __export = (target, all) => {
   for (var name in all)
@@ -15,13 +17,21 @@ var __copyProps = (to, from, except, desc) => {
   }
   return to;
 };
+var __toESM = (mod, isNodeMode, target) => (target = mod != null ? __create(__getProtoOf(mod)) : {}, __copyProps(
+  // If the importer is in node compatibility mode or this is not an ESM
+  // file that has been converted to a CommonJS file using a Babel-
+  // compatible transform (i.e. "__esModule" has not been set), then set
+  // "default" to the CommonJS "module.exports" for node compatibility.
+  isNodeMode || !mod || !mod.__esModule ? __defProp(target, "default", { value: mod, enumerable: true }) : target,
+  mod
+));
 var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
 
 // src/browser.ts
 var browser_exports = {};
 __export(browser_exports, {
   buildWsUrl: () => buildWsUrl,
-  configureVoiceClient: () => configureVoiceClient,
+  configureVoiceClient: () => configureVoiceClient2,
   createAudioCapture: () => createAudioCapture,
   createAudioPlayback: () => createAudioPlayback,
   createProtocolState: () => createProtocolState,
@@ -1050,177 +1060,6 @@ async function createWebRtcCall(opts) {
   };
 }
 
-// src/room.ts
-var import_livekit_client = require("livekit-client");
-
-// src/roomProtocol.ts
-var SYSTEM_TOPIC = "system";
-var TRANSCRIPT_TOPIC = "transcript";
-var decodeSystem = (bytes) => {
-  try {
-    const v = JSON.parse(new TextDecoder().decode(bytes));
-    if (v && typeof v.kind === "string") return v;
-    return null;
-  } catch {
-    return null;
-  }
-};
-var decodeTranscript = (bytes) => {
-  try {
-    const v = JSON.parse(new TextDecoder().decode(bytes));
-    if (v && v.kind === "partial") return v;
-    return null;
-  } catch {
-    return null;
-  }
-};
-
-// src/room.ts
-var identityToPid = (identity) => identity.startsWith("guest:") ? identity.slice("guest:".length) : identity;
-var joinRoom = async (opts) => {
-  const exchangeUrl = `${opts.apiBase.replace(/\/+$/, "")}/v1/rooms/${encodeURIComponent(
-    opts.roomId
-  )}/join`;
-  const exchangeRes = await fetch(exchangeUrl, {
-    method: "POST",
-    headers: { "Content-Type": "application/json" },
-    body: JSON.stringify({ code: opts.joinCode, name: opts.name })
-  });
-  if (!exchangeRes.ok) {
-    const err = await exchangeRes.json().catch(() => ({}));
-    throw new Error(err.error?.code ?? `join_failed_${exchangeRes.status}`);
-  }
-  const exchange = await exchangeRes.json();
-  const handlers = /* @__PURE__ */ new Map();
-  const emit = (e, payload) => {
-    handlers.get(e)?.forEach((h) => {
-      try {
-        h(payload);
-      } catch {
-      }
-    });
-  };
-  const room = new import_livekit_client.Room({ adaptiveStream: true, dynacast: true });
-  room.on(
-    import_livekit_client.RoomEvent.ParticipantConnected,
-    (p) => emit("participant.joined", {
-      participantId: identityToPid(p.identity),
-      name: p.name ?? ""
-    })
-  );
-  room.on(
-    import_livekit_client.RoomEvent.ParticipantDisconnected,
-    (p) => emit("participant.left", {
-      participantId: identityToPid(p.identity),
-      name: p.name ?? ""
-    })
-  );
-  room.on(import_livekit_client.RoomEvent.Disconnected, () => emit("room.ended", void 0));
-  room.on(import_livekit_client.RoomEvent.DataReceived, (data, _participant, _kind, topic) => {
-    if (topic === SYSTEM_TOPIC) {
-      const m = decodeSystem(data);
-      if (m) emit("system.message", m);
-    } else if (topic === TRANSCRIPT_TOPIC) {
-      const m = decodeTranscript(data);
-      if (m) emit("transcript.partial", m);
-    }
-  });
-  const trackKind = (t) => t.kind === import_livekit_client.Track.Kind.Video ? "video" : "audio";
-  const trackSource = (s) => {
-    switch (s) {
-      case import_livekit_client.Track.Source.Camera:
-        return "camera";
-      case import_livekit_client.Track.Source.Microphone:
-        return "microphone";
-      case import_livekit_client.Track.Source.ScreenShare:
-        return "screen_share";
-      case import_livekit_client.Track.Source.ScreenShareAudio:
-        return "screen_share_audio";
-      default:
-        return "unknown";
-    }
-  };
-  room.on(
-    import_livekit_client.RoomEvent.TrackSubscribed,
-    (track, pub, participant) => emit("track.subscribed", {
-      participantId: identityToPid(participant.identity),
-      kind: trackKind(track),
-      source: trackSource(pub.source),
-      track
-    })
-  );
-  room.on(
-    import_livekit_client.RoomEvent.TrackUnsubscribed,
-    (track, pub, participant) => emit("track.unsubscribed", {
-      participantId: identityToPid(participant.identity),
-      kind: trackKind(track),
-      source: trackSource(pub.source),
-      track
-    })
-  );
-  room.on(
-    import_livekit_client.RoomEvent.ActiveSpeakersChanged,
-    (speakers) => emit(
-      "active.speakers",
-      speakers.map((p) => identityToPid(p.identity))
-    )
-  );
-  await room.connect(exchange.livekit.url, exchange.livekit.token);
-  return {
-    participantId: exchange.participantId,
-    get participants() {
-      return [...room.remoteParticipants.values()].map((p) => ({
-        participantId: identityToPid(p.identity),
-        name: p.name ?? ""
-      }));
-    },
-    on(event, handler) {
-      const set = handlers.get(event) ?? /* @__PURE__ */ new Set();
-      set.add(handler);
-      handlers.set(event, set);
-    },
-    publishMic: async () => {
-      await room.localParticipant.setMicrophoneEnabled(true);
-    },
-    publishCamera: async () => {
-      await room.localParticipant.setCameraEnabled(true);
-    },
-    setMicEnabled: async (on) => {
-      await room.localParticipant.setMicrophoneEnabled(on);
-    },
-    setCameraEnabled: async (on) => {
-      await room.localParticipant.setCameraEnabled(on);
-    },
-    isMicEnabled: () => room.localParticipant.isMicrophoneEnabled,
-    isCameraEnabled: () => room.localParticipant.isCameraEnabled,
-    getLocalCameraTrack: () => room.localParticipant.getTrackPublication(import_livekit_client.Track.Source.Camera)?.videoTrack ?? null,
-    getRemoteTracks: () => {
-      const out = [];
-      for (const p of room.remoteParticipants.values()) {
-        for (const pub of p.trackPublications.values()) {
-          const track = pub.track;
-          if (!track) continue;
-          out.push({
-            participantId: identityToPid(p.identity),
-            kind: trackKind(track),
-            source: trackSource(pub.source),
-            track
-          });
-        }
-      }
-      return out;
-    },
-    setScreenShareEnabled: async (on, opts2) => {
-      await room.localParticipant.setScreenShareEnabled(on, { audio: opts2?.audio ?? false });
-    },
-    isScreenShareEnabled: () => room.localParticipant.isScreenShareEnabled,
-    getLocalScreenTrack: () => room.localParticipant.getTrackPublication(import_livekit_client.Track.Source.ScreenShare)?.videoTrack ?? null,
-    leave: async () => {
-      await room.disconnect();
-    }
-  };
-};
-
 // src/textSession.ts
 async function startTextSession(opts) {
   const f = opts.fetch ?? fetch;
@@ -1328,9 +1167,9 @@ var parseIncomingCall = (raw) => {
   return out;
 };
 
-// src/browser.ts
+// src/assistant.ts
 var browserWsFactory = (url) => new globalThis.WebSocket(url);
-var BrowserVoiceFactory = class {
+var AssistantVoiceFactory = class {
   constructor(config) {
     this.startCall = async (options) => {
       if (!options.agentId) {
@@ -1385,14 +1224,6 @@ var BrowserVoiceFactory = class {
       await client.start();
       return client;
     };
-    // Multi-party rooms (Phase 7 video).
-    //
-    // The guest's browser calls this with the roomId + joinCode it parsed
-    // out of the invite link. The SDK exchanges the code for a LiveKit
-    // JWT against `${apiBase}/v1/rooms/:roomId/join` (an AUTH-EXEMPT
-    // endpoint — the opaque code is the only credential), then connects
-    // to LiveKit and returns a typed event surface.
-    this.joinRoom = (opts) => joinRoom({ apiBase: this.config.apiBase, ...opts });
     // Text-channel chat session (no microphone / audio).
     // Mint a `ct_` token with `channel: 'text'` on your backend, then call
     // this to open an SSE stream against the chat API.
@@ -1404,7 +1235,215 @@ var BrowserVoiceFactory = class {
   }
 };
 function configureVoiceClient(config) {
-  return new BrowserVoiceFactory(normalizeConfig(config));
+  return new AssistantVoiceFactory(normalizeConfig(config));
+}
+
+// src/roomProtocol.ts
+var SYSTEM_TOPIC = "system";
+var TRANSCRIPT_TOPIC = "transcript";
+var ANALYSIS_TOPIC = "analysis";
+var decodeSystem = (bytes) => {
+  try {
+    const v = JSON.parse(new TextDecoder().decode(bytes));
+    if (v && typeof v.kind === "string") return v;
+    return null;
+  } catch {
+    return null;
+  }
+};
+var decodeTranscript = (bytes) => {
+  try {
+    const v = JSON.parse(new TextDecoder().decode(bytes));
+    if (v && v.kind === "partial") return v;
+    return null;
+  } catch {
+    return null;
+  }
+};
+var decodeAnalysis = (bytes) => {
+  try {
+    const v = JSON.parse(new TextDecoder().decode(bytes));
+    if (v && v.kind === "analysis" && typeof v.participantId === "string" && typeof v.label === "string") {
+      return v;
+    }
+    return null;
+  } catch {
+    return null;
+  }
+};
+
+// src/room.ts
+var lkPromise;
+var loadLiveKit = () => {
+  lkPromise ?? (lkPromise = import("livekit-client").catch((e) => {
+    lkPromise = void 0;
+    const msg = "joinRoom requires the optional peer dependency 'livekit-client' \u2014 npm install livekit-client";
+    const wrapped = new Error(msg);
+    wrapped.cause = e;
+    throw wrapped;
+  }));
+  return lkPromise;
+};
+var identityToPid = (identity) => identity.startsWith("guest:") ? identity.slice("guest:".length) : identity;
+var joinRoom = async (opts) => {
+  const exchangeUrl = `${opts.apiBase.replace(/\/+$/, "")}/v1/rooms/${encodeURIComponent(
+    opts.roomId
+  )}/join`;
+  const exchangeRes = await fetch(exchangeUrl, {
+    method: "POST",
+    headers: { "Content-Type": "application/json" },
+    body: JSON.stringify({ code: opts.joinCode, name: opts.name })
+  });
+  if (!exchangeRes.ok) {
+    const err = await exchangeRes.json().catch(() => ({}));
+    throw new Error(err.error?.code ?? `join_failed_${exchangeRes.status}`);
+  }
+  const exchange = await exchangeRes.json();
+  const handlers = /* @__PURE__ */ new Map();
+  const emit = (e, payload) => {
+    handlers.get(e)?.forEach((h) => {
+      try {
+        h(payload);
+      } catch {
+      }
+    });
+  };
+  const lk = await loadLiveKit();
+  const trackKind = (t) => t.kind === lk.Track.Kind.Video ? "video" : "audio";
+  const trackSource = (s) => {
+    switch (s) {
+      case lk.Track.Source.Camera:
+        return "camera";
+      case lk.Track.Source.Microphone:
+        return "microphone";
+      case lk.Track.Source.ScreenShare:
+        return "screen_share";
+      case lk.Track.Source.ScreenShareAudio:
+        return "screen_share_audio";
+      default:
+        return "unknown";
+    }
+  };
+  const room = new lk.Room({ adaptiveStream: true, dynacast: true });
+  room.on(
+    lk.RoomEvent.ParticipantConnected,
+    (p) => emit("participant.joined", {
+      participantId: identityToPid(p.identity),
+      name: p.name ?? ""
+    })
+  );
+  room.on(
+    lk.RoomEvent.ParticipantDisconnected,
+    (p) => emit("participant.left", {
+      participantId: identityToPid(p.identity),
+      name: p.name ?? ""
+    })
+  );
+  room.on(lk.RoomEvent.Disconnected, () => emit("room.ended", void 0));
+  room.on(lk.RoomEvent.DataReceived, (data, _participant, _kind, topic) => {
+    if (topic === SYSTEM_TOPIC) {
+      const m = decodeSystem(data);
+      if (m) emit("system.message", m);
+    } else if (topic === TRANSCRIPT_TOPIC) {
+      const m = decodeTranscript(data);
+      if (m) emit("transcript.partial", m);
+    } else if (topic === ANALYSIS_TOPIC) {
+      const m = decodeAnalysis(data);
+      if (m) emit("analysis", m);
+    }
+  });
+  room.on(
+    lk.RoomEvent.TrackSubscribed,
+    (track, pub, participant) => emit("track.subscribed", {
+      participantId: identityToPid(participant.identity),
+      kind: trackKind(track),
+      source: trackSource(pub.source),
+      track
+    })
+  );
+  room.on(
+    lk.RoomEvent.TrackUnsubscribed,
+    (track, pub, participant) => emit("track.unsubscribed", {
+      participantId: identityToPid(participant.identity),
+      kind: trackKind(track),
+      source: trackSource(pub.source),
+      track
+    })
+  );
+  room.on(
+    lk.RoomEvent.ActiveSpeakersChanged,
+    (speakers) => emit(
+      "active.speakers",
+      speakers.map((p) => identityToPid(p.identity))
+    )
+  );
+  await room.connect(exchange.livekit.url, exchange.livekit.token);
+  return {
+    participantId: exchange.participantId,
+    kind: exchange.kind ?? "video",
+    get participants() {
+      return [...room.remoteParticipants.values()].map((p) => ({
+        participantId: identityToPid(p.identity),
+        name: p.name ?? ""
+      }));
+    },
+    on(event, handler) {
+      const set = handlers.get(event) ?? /* @__PURE__ */ new Set();
+      set.add(handler);
+      handlers.set(event, set);
+    },
+    publishMic: async () => {
+      await room.localParticipant.setMicrophoneEnabled(true);
+    },
+    publishCamera: async () => {
+      await room.localParticipant.setCameraEnabled(true);
+    },
+    setMicEnabled: async (on) => {
+      await room.localParticipant.setMicrophoneEnabled(on);
+    },
+    setCameraEnabled: async (on) => {
+      await room.localParticipant.setCameraEnabled(on);
+    },
+    isMicEnabled: () => room.localParticipant.isMicrophoneEnabled,
+    isCameraEnabled: () => room.localParticipant.isCameraEnabled,
+    getLocalCameraTrack: () => room.localParticipant.getTrackPublication(lk.Track.Source.Camera)?.videoTrack ?? null,
+    getRemoteTracks: () => {
+      const out = [];
+      for (const p of room.remoteParticipants.values()) {
+        for (const pub of p.trackPublications.values()) {
+          const track = pub.track;
+          if (!track) continue;
+          out.push({
+            participantId: identityToPid(p.identity),
+            kind: trackKind(track),
+            source: trackSource(pub.source),
+            track
+          });
+        }
+      }
+      return out;
+    },
+    setScreenShareEnabled: async (on, opts2) => {
+      await room.localParticipant.setScreenShareEnabled(on, { audio: opts2?.audio ?? false });
+    },
+    isScreenShareEnabled: () => room.localParticipant.isScreenShareEnabled,
+    getLocalScreenTrack: () => room.localParticipant.getTrackPublication(lk.Track.Source.ScreenShare)?.videoTrack ?? null,
+    leave: async () => {
+      await room.disconnect();
+    }
+  };
+};
+
+// src/browser.ts
+function configureVoiceClient2(config) {
+  const normalized = normalizeConfig(config);
+  const factory = configureVoiceClient(normalized);
+  return {
+    ...factory,
+    // Re-compose joinRoom onto the factory so barrel consumers who call
+    // `voice.joinRoom(...)` keep working unchanged.
+    joinRoom: (opts) => joinRoom({ apiBase: normalized.apiBase, ...opts })
+  };
 }
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {