客户端采用 WebRTC 技术(推流),通讯用 websocket。
WebRTC 像是一个面试过程:
第一步:发起方(拨打电话者)点击拨打电话时,获取本地媒体流并推流给接收方同时捕获接收方推过来的流,捕获到后把流设置到 dom 上,监听 ICE 候选确保能点对连接,生成 offer,通过 websocket 告知接收方并拉起等待接听界面。
//获取媒体流
stream.value = await navigator.mediaDevices.getUserMedia({video: true,audio: true
});// 初始化 PeerConnection
peerConnection.value = new RTCPeerConnection({iceServers: [{urls: 'stun:stun.l.google.com:19302'}]
});// 推流给接收方
stream.value.getTracks().forEach((track) => {peerConnection.value.addTrack(track, stream.value);
});// 捕获接收方的流
peerConnection.value.ontrack = (event) => {remoteStream.value = event.streams[0];if (callType.value === TypeVideo) {remoteVideo.value.srcObject = remoteStream.value;} else {remoteAudio.value.srcObject = remoteStream.value;}
};// 监听ICE候选,确保 WebRTC 的点对点连接能够成功建立
peerConnection.value.onicecandidate = (event) => {if (event.candidate) {//发送candidatews.send(event.candidate);}
};
// 创建 offer
const offer = await peerConnection.value.createOffer();
await peerConnection.value.setLocalDescription(offer);//发送offer,这里发送的offer可以理解成是接收方用来捕获发起方流的一个凭证,接收方通过peerConnection.value.ontrack可以捕获到。
ws.send(offer);
//拉起等待接听界面
showCall.value = true;
//状态等待接听
callStatus.value = 'wating';
第二步:接收方收到 offer 后,第一步是拉起来电界面,第二步是选择接听或者挂断。
1)拉起来电接听界面
//拉起来电接听界面
showCall.vue = true;
//状态来电接听
callStatus.value = 'coming';
//初始化来电人信息等
....
2)挂断,就是告诉发起方我挂断了,发起方就把 RTC 关掉、停止推流,dom 置空就好了
//接收方
showCall.value = false;
callStatus.value = 'closing';
ws.send('reject');//发起方
if (peerConnection.value) {peerConnection.value.close();peerConnection.value = null;
}
if (stream.value) {const tracks = stream.value.getTracks();tracks.forEach((track) => track.stop());
}
if (localVideo.value)localVideo.value.srcObject = null;
if (remoteVideo.value)remoteVideo.value.srcObject = null;
if (remoteAudio.value)remoteAudio.value.srcObject = null;
showCall.value = false;
callStatus.value = 'closing';
3)接听,操作跟拨打流程差不多,需要设置远端 SDP(发起方的 offer),添加 ICE 候选(发起方的 ice,这里需要注意的是只有远端 SDP 初始化完毕状态下才能设置 ice)
// 获取本地媒体流
...同发起方
// 初始化 PeerConnection
...同发起方
// 推流给发起方
...同发起方
// 捕获发起方的流
...同发起方
// 监听ICE候选
...同发起方//设置远端SDP
await peerConnection.value.setRemoteDescription(new RTCSessionDescription(caller.value.offer));// 添加发起方发过来的ice
iceCandidateQueue.value.forEach(async (candidate) => {
await peerConnection.value.addIceCandidate(candidate);
});
iceCandidateQueue.value = [];// 创建 answer
const answer = await peerConnection.value.createAnswer();
await peerConnection.value.setLocalDescription(answer);//发送answer给发起方
ws.send(answer);
//状态通话中
callStatus.value = 'calling';
关于 ice 的处理,就是远端 SDP 初始化完毕状态可以直接设置,未初始化完毕就存到 iceCandidateQueue 队列备用
// 处理新的 ICE 候选
const handleNewICECandidate = async (candidate) => {const iceCandidate = new RTCIceCandidate(candidate);if (peerConnection.value?.signalingState === 'have-remote-offer' || peerConnection.value?.signalingState === 'stable') {peerConnection.value.addIceCandidate(iceCandidate);} else {iceCandidateQueue.value.push(iceCandidate);}
};
最后一步:发起方收到接收方的答复(接收方接听了),设置远端 SDP(接收方的 answer), 设置 ICE(接受方的 ice)
//设置远端SDP
await peerConnection.value.setRemoteDescription(new RTCSessionDescription(caller.value.answer));//添加ICE
iceCandidateQueue.value.forEach(async (candidate) => {await peerConnection.value.addIceCandidate(candidate);
});
iceCandidateQueue.value = [];
//状态接听中
callStatus.value = 'calling';
这就是 WebRTC 视频通话的关键代码跟流程!
例图: