Enable group calls without video and audio track by configuration of MatrixClient

spec/unit/webrtc/groupCall.spec.ts

@@ -109,7 +109,7 @@ const mockGetStateEvents = (type: EventType, userId?: string): MatrixEvent[] | M
 const ONE_HOUR = 1000 * 60 * 60;
 
 const createAndEnterGroupCall = async (cli: MatrixClient, room: Room): Promise<GroupCall> => {
-    const groupCall = new GroupCall(cli, room, GroupCallType.Video, false, GroupCallIntent.Prompt, FAKE_CONF_ID);
+    const groupCall = new GroupCall(cli, room, GroupCallType.Video, false, GroupCallIntent.Prompt, false, FAKE_CONF_ID);
 
     await groupCall.create();
     await groupCall.enter();
@@ -135,7 +135,7 @@ describe("Group Call", function () {
             mockClient = typedMockClient as unknown as MatrixClient;
 
             room = new Room(FAKE_ROOM_ID, mockClient, FAKE_USER_ID_1);
-            groupCall = new GroupCall(mockClient, room, GroupCallType.Video, false, GroupCallIntent.Prompt);
+            groupCall = new GroupCall(mockClient, room, GroupCallType.Video, false, GroupCallIntent.Prompt, false);
             room.currentState.members[FAKE_USER_ID_1] = {
                 userId: FAKE_USER_ID_1,
                 membership: "join",
@@ -484,7 +484,7 @@ describe("Group Call", function () {
         describe("PTT calls", () => {
             beforeEach(async () => {
                 // replace groupcall with a PTT one
-                groupCall = new GroupCall(mockClient, room, GroupCallType.Video, true, GroupCallIntent.Prompt);
+                groupCall = new GroupCall(mockClient, room, GroupCallType.Video, true, GroupCallIntent.Prompt, false);
 
                 await groupCall.create();
 
@@ -647,6 +647,7 @@ describe("Group Call", function () {
                 GroupCallType.Video,
                 false,
                 GroupCallIntent.Prompt,
+                false,
                 FAKE_CONF_ID,
             );
 
@@ -656,6 +657,7 @@ describe("Group Call", function () {
                 GroupCallType.Video,
                 false,
                 GroupCallIntent.Prompt,
+                false,
                 FAKE_CONF_ID,
             );
         });
@@ -1465,6 +1467,7 @@ describe("Group Call", function () {
                 GroupCallType.Video,
                 false,
                 GroupCallIntent.Prompt,
+                false,
                 FAKE_CONF_ID,
             );
             await groupCall.create();

src/client.ts

@@ -371,6 +371,13 @@ export interface ICreateClientOpts {
      * Defaults to a built-in English handler with basic pluralisation.
      */
     roomNameGenerator?: (roomId: string, state: RoomNameState) => string | null;
+
+    /**
+     * If true, participant can join group call without video and audio this has to be allowed. By default, a local
+     * media stream is needed to establish a group call.
+     * Default: false.
+     */
+    isVoipWithNoMediaAllowed?: boolean;
 }
 
 export interface IMatrixClientCreateOpts extends ICreateClientOpts {
@@ -1169,6 +1176,7 @@ export class MatrixClient extends TypedEventEmitter<EmittedEvents, ClientEventHa
     public iceCandidatePoolSize = 0; // XXX: Intended private, used in code.
     public idBaseUrl?: string;
     public baseUrl: string;
+    public readonly isVoipWithNoMediaAllowed;
 
     // Note: these are all `protected` to let downstream consumers make mistakes if they want to.
     // We don't technically support this usage, but have reasons to do this.
@@ -1313,6 +1321,7 @@ export class MatrixClient extends TypedEventEmitter<EmittedEvents, ClientEventHa
         this.iceCandidatePoolSize = opts.iceCandidatePoolSize === undefined ? 0 : opts.iceCandidatePoolSize;
         this.supportsCallTransfer = opts.supportsCallTransfer || false;
         this.fallbackICEServerAllowed = opts.fallbackICEServerAllowed || false;
+        this.isVoipWithNoMediaAllowed = opts.isVoipWithNoMediaAllowed || false;
 
         if (opts.useE2eForGroupCall !== undefined) this.useE2eForGroupCall = opts.useE2eForGroupCall;
 
@@ -1880,14 +1889,17 @@ export class MatrixClient extends TypedEventEmitter<EmittedEvents, ClientEventHa
             throw new Error(`Cannot find room ${roomId}`);
         }
 
+        // Because without Media section a WebRTC connection is not possible, so need a RTCDataChannel to set up a
+        // no media WebRTC connection anyway.
         return new GroupCall(
             this,
             room,
             type,
             isPtt,
             intent,
+            this.isVoipWithNoMediaAllowed,
             undefined,
-            dataChannelsEnabled,
+            dataChannelsEnabled || this.isVoipWithNoMediaAllowed,
             dataChannelOptions,
         ).create();
     }

src/webrtc/call.ts

@@ -390,6 +390,9 @@ export class MatrixCall extends TypedEventEmitter<CallEvent, CallEventHandlerMap
     // Used to keep the timer for the delay before actually stopping our
     // video track after muting (see setLocalVideoMuted)
     private stopVideoTrackTimer?: ReturnType<typeof setTimeout>;
+    // Used to allow connection without Video and Audio. To establish a webrtc connection without media a Data channel is
+    // needed At the moment this property is true if we allow MatrixClient with isVoipWithNoMediaAllowed = true
+    private readonly isOnlyDataChannelAllowed: boolean;
 
     /**
      * Construct a new Matrix Call.
@@ -420,6 +423,8 @@ export class MatrixCall extends TypedEventEmitter<CallEvent, CallEventHandlerMap
             utils.checkObjectHasKeys(server, ["urls"]);
         }
         this.callId = genCallID();
+        // If the Client provides calls without audio and video we need a datachannel for a webrtc connection
+        this.isOnlyDataChannelAllowed = this.client.isVoipWithNoMediaAllowed;
     }
 
     /**
@@ -944,7 +949,11 @@ export class MatrixCall extends TypedEventEmitter<CallEvent, CallEventHandlerMap
         // According to previous comments in this file, firefox at some point did not
         // add streams until media started arriving on them. Testing latest firefox
         // (81 at time of writing), this is no longer a problem, so let's do it the correct way.
-        if (!remoteStream || remoteStream.getTracks().length === 0) {
+        //
+        // For example in case of no media webrtc connections like screen share only call we have to allow webrtc
+        // connections without remote media. In this case we always use a data channel. At the moment we allow as well
+        // only data channel as media in the WebRTC connection with this setup here.
+        if (!this.isOnlyDataChannelAllowed && (!remoteStream || remoteStream.getTracks().length === 0)) {
             logger.error(
                 `Call ${this.callId} initWithInvite() no remote stream or no tracks after setting remote description!`,
             );

src/webrtc/groupCall.ts

@@ -216,6 +216,7 @@ export class GroupCall extends TypedEventEmitter<
         public type: GroupCallType,
         public isPtt: boolean,
         public intent: GroupCallIntent,
+        public readonly initCallWithoutVideoAndAudio: boolean,
         groupCallId?: string,
         private dataChannelsEnabled?: boolean,
         private dataChannelOptions?: IGroupCallDataChannelOptions,
@@ -374,8 +375,15 @@ export class GroupCall extends TypedEventEmitter<
         try {
             stream = await this.client.getMediaHandler().getUserMediaStream(true, this.type === GroupCallType.Video);
         } catch (error) {
-            this.state = GroupCallState.LocalCallFeedUninitialized;
-            throw error;
+            // If is allowed to join a call without a media stream, then we
+            // don't throw an error here. But we need an empty Local Feed to establish
+            // a connection later.
+            if (this.initCallWithoutVideoAndAudio) {
+                stream = new MediaStream();
+            } else {
+                this.state = GroupCallState.LocalCallFeedUninitialized;
+                throw error;
+            }
         }
 
         // The call could've been disposed while we were waiting, and could
@@ -609,6 +617,11 @@ export class GroupCall extends TypedEventEmitter<
      * @returns Whether muting/unmuting was successful
      */
     public async setLocalVideoMuted(muted: boolean): Promise<boolean> {
+        // Because we need a Local Call Feed to establish a call connection, we avoid muting video in case of empty
+        // video track. In this way we go sure if a client implements muting we don't raise an error.
+        if (this.initCallWithoutVideoAndAudio && this.localCallFeed?.stream.getVideoTracks().length === 0) {
+            return false;
+        }
         // hasAudioDevice can block indefinitely if the window has lost focus,
         // and it doesn't make much sense to keep a device from being muted, so
         // we always allow muted = true changes to go through

src/webrtc/groupCallEventHandler.ts

@@ -183,8 +183,11 @@ export class GroupCallEventHandler {
             callType,
             isPtt,
             callIntent,
+            this.client.isVoipWithNoMediaAllowed,
             groupCallId,
-            content?.dataChannelsEnabled,
+            // Because without Media section a WebRTC connection is not possible, so need a RTCDataChannel to set up a
+            // no media WebRTC connection anyway.
+            content?.dataChannelsEnabled || this.client.isVoipWithNoMediaAllowed,
             dataChannelOptions,
         );