Get E2EE functional

CMakeLists.txt

@@ -160,6 +160,7 @@ add_library(livekit
   include/livekit/audio_source.h
   include/livekit/audio_stream.h
   include/livekit/data_stream.h
+  include/livekit/e2ee.h
   include/livekit/room.h
   include/livekit/room_event_types.h
   include/livekit/room_delegate.h
@@ -186,6 +187,7 @@ add_library(livekit
   src/audio_source.cpp
   src/audio_stream.cpp
   src/data_stream.cpp
+  src/e2ee.cpp
   src/ffi_handle.cpp
   src/ffi_client.cpp
   src/local_audio_track.cpp

README.md

@@ -60,6 +60,19 @@ export LIVEKIT_TOKEN=<jwt-token>
 ./build/examples/SimpleRoom
 ```
 
+**End-to-End Encryption (E2EE)**
+You can enable E2E encryption for the streams via --enable_e2ee and --e2ee_key flags,
+by running the following cmds in two terminals or computers. **Note, jwt_token needs to be different identity**
+```bash
+./build/examples/SimpleRoom --url $URL --token <jwt-token> --enable_e2ee --e2ee_key="your_key"
+```
+**Note**, **all participants must use the exact same E2EE configuration and shared key.**
+If the E2EE keys do not match between participants:
+- Media cannot be decrypted
+- Video tracks will appear as a black screen
+- Audio will be silent
+- No explicit error may be shown at the UI level
+
 Press Ctrl-C to exit the example.
 
 ### SimpleRpc

examples/simple_data_stream/main.cpp

@@ -46,10 +46,10 @@ std::string randomHexId(std::size_t nbytes = 16) {
 }
 
 // Greeting: send text + image
-void greetParticipant(Room &room, const std::string &identity) {
+void greetParticipant(Room *room, const std::string &identity) {
   std::cout << "[DataStream] Greeting participant: " << identity << "\n";
 
-  LocalParticipant *lp = room.localParticipant();
+  LocalParticipant *lp = room->localParticipant();
   if (!lp) {
     std::cerr << "[DataStream] No local participant, cannot greet.\n";
     return;
@@ -209,33 +209,33 @@ int main(int argc, char *argv[]) {
   std::signal(SIGTERM, handleSignal);
 #endif
 
-  Room room{};
+  auto room = std::make_unique<Room>();
   RoomOptions options;
   options.auto_subscribe = true;
   options.dynacast = false;
 
-  bool ok = room.Connect(url, token, options);
+  bool ok = room->Connect(url, token, options);
   std::cout << "[DataStream] Connect result: " << std::boolalpha << ok << "\n";
   if (!ok) {
     std::cerr << "[DataStream] Failed to connect to room\n";
     FfiClient::instance().shutdown();
     return 1;
   }
 
-  auto info = room.room_info();
+  auto info = room->room_info();
   std::cout << "[DataStream] Connected to room '" << info.name
             << "', participants: " << info.num_participants << "\n";
 
   // Register stream handlers
-  room.registerTextStreamHandler(
+  room->registerTextStreamHandler(
       "chat", [](std::shared_ptr<TextStreamReader> reader,
                  const std::string &participant_identity) {
         std::thread t(handleChatMessage, std::move(reader),
                       participant_identity);
         t.detach();
       });
 
-  room.registerByteStreamHandler(
+  room->registerByteStreamHandler(
       "files", [](std::shared_ptr<ByteStreamReader> reader,
                   const std::string &participant_identity) {
         std::thread t(handleWelcomeImage, std::move(reader),
@@ -245,25 +245,25 @@ int main(int argc, char *argv[]) {
 
   // Greet existing participants
   {
-    auto remotes = room.remoteParticipants();
+    auto remotes = room->remoteParticipants();
     for (const auto &rp : remotes) {
       if (!rp)
         continue;
       std::cout << "Remote: " << rp->identity() << "\n";
-      greetParticipant(room, rp->identity());
+      greetParticipant(room.get(), rp->identity());
     }
   }
 
   // Optionally: greet on join
   //
   // If Room API exposes a participant-connected callback, you could do:
   //
-  // room.onParticipantConnected(
+  // room->onParticipantConnected(
   //     [&](RemoteParticipant& participant) {
   //       std::cout << "[DataStream] participant connected: "
   //                 << participant.sid() << " " << participant.identity()
   //                 << "\n";
-  //       greetParticipant(room, participant.identity());
+  //       greetParticipant(room.get(), participant.identity());
   //     });
   //
   // Adjust to your actual event API.
@@ -274,6 +274,9 @@ int main(int argc, char *argv[]) {
   }
 
   std::cout << "[DataStream] Shutting down...\n";
+  // It is important to clean up the delegate and room in order.
+  room->setDelegate(nullptr);
+  room.reset();
   FfiClient::instance().shutdown();
   return 0;
 }

examples/simple_room/main.cpp

@@ -43,27 +43,40 @@ namespace {
 std::atomic<bool> g_running{true};
 
 void printUsage(const char *prog) {
-  std::cerr << "Usage:\n"
-            << "  " << prog << " <ws-url> <token>\n"
-            << "or:\n"
-            << "  " << prog << " --url=<ws-url> --token=<token>\n"
-            << "  " << prog << " --url <ws-url> --token <token>\n\n"
-            << "Env fallbacks:\n"
-            << "  LIVEKIT_URL, LIVEKIT_TOKEN\n";
+  std::cerr
+      << "Usage:\n"
+      << "  " << prog
+      << " <ws-url> <token> [--enable_e2ee] [--e2ee_key <key>]\n"
+      << "or:\n"
+      << "  " << prog
+      << " --url=<ws-url> --token=<token> [--enable_e2ee] [--e2ee_key=<key>]\n"
+      << "  " << prog
+      << " --url <ws-url> --token <token> [--enable_e2ee] [--e2ee_key "
+         "<key>]\n\n"
+      << "E2EE:\n"
+      << "  --enable_e2ee          Enable end-to-end encryption (E2EE)\n"
+      << "  --e2ee_key <key>       Optional shared key (UTF-8). If omitted, "
+         "E2EE is enabled\n"
+      << "                         but no shared key is set (advanced "
+         "usage).\n\n"
+      << "Env fallbacks:\n"
+      << "  LIVEKIT_URL, LIVEKIT_TOKEN, LIVEKIT_E2EE_KEY\n";
 }
 
 void handleSignal(int) { g_running.store(false); }
 
-bool parseArgs(int argc, char *argv[], std::string &url, std::string &token) {
-  // 1) --help
+bool parseArgs(int argc, char *argv[], std::string &url, std::string &token,
+               bool &enable_e2ee, std::string &e2ee_key) {
+  enable_e2ee = false;
+  // --help
   for (int i = 1; i < argc; ++i) {
     std::string a = argv[i];
     if (a == "-h" || a == "--help") {
       return false;
     }
   }
 
-  // 2) flags: --url= / --token= or split form
+  // flags: --url= / --token= or split form
   auto get_flag_value = [&](const std::string &name, int &i) -> std::string {
     std::string arg = argv[i];
     const std::string eq = name + "=";
@@ -79,18 +92,24 @@ bool parseArgs(int argc, char *argv[], std::string &url, std::string &token) {
 
   for (int i = 1; i < argc; ++i) {
     const std::string a = argv[i];
-    if (a.rfind("--url", 0) == 0) {
+    if (a == "--enable_e2ee") {
+      enable_e2ee = true;
+    } else if (a.rfind("--url", 0) == 0) {
       auto v = get_flag_value("--url", i);
       if (!v.empty())
         url = v;
     } else if (a.rfind("--token", 0) == 0) {
       auto v = get_flag_value("--token", i);
       if (!v.empty())
         token = v;
+    } else if (a.rfind("--e2ee_key", 0) == 0) {
+      auto v = get_flag_value("--e2ee_key", i);
+      if (!v.empty())
+        e2ee_key = v;
     }
   }
 
-  // 3) positional if still empty
+  // positional if still empty
   if (url.empty() || token.empty()) {
     std::vector<std::string> pos;
     for (int i = 1; i < argc; ++i) {
@@ -118,6 +137,11 @@ bool parseArgs(int argc, char *argv[], std::string &url, std::string &token) {
     if (e)
       token = e;
   }
+  if (e2ee_key.empty()) {
+    const char *e = std::getenv("LIVEKIT_E2EE_KEY");
+    if (e)
+      e2ee_key = e;
+  }
 
   return !(url.empty() || token.empty());
 }
@@ -211,11 +235,17 @@ class SimpleRoomDelegate : public livekit::RoomDelegate {
   SDLMediaManager &media_;
 };
 
+static std::vector<std::uint8_t> toBytes(const std::string &s) {
+  return std::vector<std::uint8_t>(s.begin(), s.end());
+}
+
 } // namespace
 
 int main(int argc, char *argv[]) {
   std::string url, token;
-  if (!parseArgs(argc, argv, url, token)) {
+  bool enable_e2ee = false;
+  std::string e2ee_key;
+  if (!parseArgs(argc, argv, url, token, enable_e2ee, e2ee_key)) {
     printUsage(argv[0]);
     return 1;
   }
@@ -240,22 +270,41 @@ int main(int argc, char *argv[]) {
   // Handle Ctrl-C to exit the idle loop
   std::signal(SIGINT, handleSignal);
 
-  livekit::Room room{};
+  auto room = std::make_unique<livekit::Room>();
   SimpleRoomDelegate delegate(media);
-  room.setDelegate(&delegate);
+  room->setDelegate(&delegate);
 
   RoomOptions options;
   options.auto_subscribe = true;
   options.dynacast = false;
-  bool res = room.Connect(url, token, options);
+
+  if (enable_e2ee) {
+    livekit::E2EEOptions e2ee;
+    e2ee.encryption_type = livekit::EncryptionType::GCM;
+    // Optional shared key: if empty, we enable E2EE without setting a shared
+    // key. (Advanced use: keys can be set/ratcheted later via
+    // E2EEManager/KeyProvider.)
+    if (!e2ee_key.empty()) {
+      e2ee.shared_key = toBytes(e2ee_key);
+    }
+    options.e2ee = e2ee;
+    if (!e2ee_key.empty()) {
+      std::cout << "[E2EE] enabled : (shared key length=" << e2ee_key.size()
+                << ")\n";
+    } else {
+      std::cout << "[E2EE] enabled: (no shared key set)\n";
+    }
+  }
+
+  bool res = room->Connect(url, token, options);
   std::cout << "Connect result is " << std::boolalpha << res << std::endl;
   if (!res) {
     std::cerr << "Failed to connect to room\n";
     FfiClient::instance().shutdown();
     return 1;
   }
 
-  auto info = room.room_info();
+  auto info = room->room_info();
   std::cout << "Connected to room:\n"
             << "  SID: " << (info.sid ? *info.sid : "(none)") << "\n"
             << "  Name: " << info.name << "\n"
@@ -286,7 +335,7 @@ int main(int argc, char *argv[]) {
   try {
     // publishTrack takes std::shared_ptr<Track>, LocalAudioTrack derives from
     // Track
-    audioPub = room.localParticipant()->publishTrack(audioTrack, audioOpts);
+    audioPub = room->localParticipant()->publishTrack(audioTrack, audioOpts);
 
     std::cout << "Published track:\n"
               << "  SID: " << audioPub->sid() << "\n"
@@ -314,7 +363,7 @@ int main(int argc, char *argv[]) {
   try {
     // publishTrack takes std::shared_ptr<Track>, LocalAudioTrack derives from
     // Track
-    videoPub = room.localParticipant()->publishTrack(videoTrack, videoOpts);
+    videoPub = room->localParticipant()->publishTrack(videoTrack, videoOpts);
 
     std::cout << "Published track:\n"
               << "  SID: " << videoPub->sid() << "\n"
@@ -341,12 +390,16 @@ int main(int argc, char *argv[]) {
   media.stopMic();
 
   // Clean up the audio track publishment
-  room.localParticipant()->unpublishTrack(audioPub->sid());
+  room->localParticipant()->unpublishTrack(audioPub->sid());
 
   media.stopCamera();
 
   // Clean up the video track publishment
-  room.localParticipant()->unpublishTrack(videoPub->sid());
+  room->localParticipant()->unpublishTrack(videoPub->sid());
+
+  // Must be cleaned up before FfiClient::instance().shutdown();
+  room->setDelegate(nullptr);
+  room.reset();
 
   FfiClient::instance().shutdown();
   std::cout << "Exiting.\n";