# 步骤 5：发布和订阅视频
您可以通过以下方式发布/订阅（实时）IVS：
+ 支持 WebRTC 和 RTMPS 的原生 [IVS 广播 SDK](https://docs.aws.amazon.com//ivs/latest/LowLatencyUserGuide/getting-started-set-up-streaming.html#broadcast-sdk)。我们推荐采用此方法，尤其是在生产场景中。有关 [Web](getting-started-pub-sub-web.md)、[Android](getting-started-pub-sub-android.md) 和 [iOS](getting-started-pub-sub-ios.md) 的详细信息，请参阅下文。
+ Amazon IVS 控制台 — 适用于测试流。请参阅下面的。
+ 其他流媒体软件和硬件编码器：您可以使用任何支持 RTMP、RTMPS 或 WHIP 协议的流媒体编码器。有关更多信息，请参阅[流摄取](rt-stream-ingest.md)。
## IVS 控制台
1. 打开 [Amazon IVS 控制台](https://console.aws.amazon.com/ivs)。
（您还可通过 [Amazon 管理控制台](https://console.aws.amazon.com/)访问 Amazon IVS 控制台。）
1. 在导航窗格中，选择**暂存区**。（如果导航窗格已折叠，请选择汉堡图标以将其展开。）
1. 选择您要订阅或发布的暂存区，以转至其详细信息页面。
1. 订阅：如果暂存区有一个或多个发布者，则可以通过按**订阅**选项卡下的**订阅**按钮进行订阅。（这些选项卡在**常规配置**部分下。）
1. 发布：
1. 选择**发布**选项卡。
1. 系统将提示您授予 IVS 控制台访问您摄像头和麦克风的权限；**允许**这些权限。
1. 在**发布**选项卡的底部，使用下拉框为麦克风和摄像头选择输入设备。
1. 要开始发布，请选择**开始发布**。
1. 要查看您发布的内容，请返回**订阅**选项卡。
1. 要停止发布，请前往**发布**选项卡，然后按下底部的**停止发布**按钮。
**注意**：订阅和发布会消耗资源，连接到暂存区的时间将按小时费率收费。要了解更多信息，请参阅 IVS 定价页面上的[实时直播功能](https://aws.amazon.com/ivs/pricing/#Real-Time_Streaming)。
# 使用 IVS Web 广播 SDK 发布和订阅
本部分将引导您完成使用 Web 应用程序发布和订阅舞台所涉及的步骤。
## 创建 HTML 样板
首先，创建 HTML 样板，并将该库作为脚本标签导入：
```
```
## 接受令牌输入并添加“加入/离开”按钮
使用输入控件在此处填充正文。它们将令牌作为输入，然后设置**加入**和**离开**按钮。通常，应用程序会从应用程序的 API 请求令牌，但在本示例中，您需要将令牌复制并粘贴到令牌输入中。
```
IVS Real-Time Streaming
```
## 添加媒体容器元素
这些元素将为本地和远程参与者保留媒体。添加脚本标签来加载在 `app.js` 中定义的应用程序逻辑。
```
```
这样就完成了 HTML 页面，在浏览器中加载 `index.html` 时您应该会看到以下内容：
![\[在浏览器中查看实时流式传输：HTML 设置完成。\]](http://docs.aws.amazon.com/zh_cn/ivs/latest/RealTimeUserGuide/images/RT_Browser_View.png)
## 创建 app.js
开始定义 `app.js` 文件的内容。首先，从 SDK 的全局导入所有必需的属性：
```
const {
Stage,
LocalStageStream,
SubscribeType,
StageEvents,
ConnectionState,
StreamType
} = IVSBroadcastClient;
```
## 创建应用程序变量
建立变量以保存对**加入**和**离开**按钮 HTML 元素的引用，并存储应用程序的状态：
```
let joinButton = document.getElementById("join-button");
let leaveButton = document.getElementById("leave-button");
// Stage management
let stage;
let joining = false;
let connected = false;
let localCamera;
let localMic;
let cameraStageStream;
let micStageStream;
```
## 创建 joinStage 1：定义函数并验证输入
`joinStage` 函数获取输入令牌，创建与舞台的连接，然后开始发布从 `getUserMedia` 中检索的视频和音频。
首先，定义函数并验证状态和令牌输入。我们将在接下来的几个部分中完善此功能。
```
const joinStage = async () => {
if (connected || joining) {
return;
}
joining = true;
const token = document.getElementById("token").value;
if (!token) {
window.alert("Please enter a participant token");
joining = false;
return;
}
// Fill in with the next sections
};
```
## 创建 joinStage 2：发布媒体
以下是将发布到舞台的媒体：
```
async function getCamera() {
// Use Max Width and Height
return navigator.mediaDevices.getUserMedia({
video: true,
audio: false
});
}
async function getMic() {
return navigator.mediaDevices.getUserMedia({
video: false,
audio: true
});
}
// Retrieve the User Media currently set on the page
localCamera = await getCamera();
localMic = await getMic();
// Create StageStreams for Audio and Video
cameraStageStream = new LocalStageStream(localCamera.getVideoTracks()[0]);
micStageStream = new LocalStageStream(localMic.getAudioTracks()[0]);
```
## 创建 joinStage 3：定义舞台策略并创建舞台
这个舞台策略是决策逻辑的核心，SDK 将使用此策略来决定要发布什么内容和订阅哪些参与者。有关此函数用途的更多信息，请参阅 [Strategy](web-publish-subscribe.md#web-publish-subscribe-concepts-strategy)。
这个策略很简单。加入舞台后，发布刚刚检索的流，并订阅每个远程参与者的音频和视频：
```
const strategy = {
stageStreamsToPublish() {
return [cameraStageStream, micStageStream];
},
shouldPublishParticipant() {
return true;
},
shouldSubscribeToParticipant() {
return SubscribeType.AUDIO_VIDEO;
}
};
stage = new Stage(token, strategy);
```
## 创建 JoinStage 4：处理舞台事件和渲染媒体
舞台会发出许多事件。需要侦听 `STAGE_PARTICIPANT_STREAMS_ADDED` 和 `STAGE_PARTICIPANT_LEFT`，以将媒体渲染到页面和从页面中移除媒体。[事件](web-publish-subscribe.md#web-publish-subscribe-concepts-events)中列出了一组更详尽的事件。
请注意，我们在这里创建了四个帮助程序函数，以帮助管理必要的 DOM 元素：`setupParticipant`、`teardownParticipant`、`createVideoEl` 和 `createContainer`。
```
stage.on(StageEvents.STAGE_CONNECTION_STATE_CHANGED, (state) => {
connected = state === ConnectionState.CONNECTED;
if (connected) {
joining = false;
joinButton.style = "display: none";
leaveButton.style = "display: inline-block";
}
});
stage.on(
StageEvents.STAGE_PARTICIPANT_STREAMS_ADDED,
(participant, streams) => {
console.log("Participant Media Added: ", participant, streams);
let streamsToDisplay = streams;
if (participant.isLocal) {
// Ensure to exclude local audio streams, otherwise echo will occur
streamsToDisplay = streams.filter(
(stream) => stream.streamType === StreamType.VIDEO
);
}
const videoEl = setupParticipant(participant);
streamsToDisplay.forEach((stream) =>
videoEl.srcObject.addTrack(stream.mediaStreamTrack)
);
}
);
stage.on(StageEvents.STAGE_PARTICIPANT_LEFT, (participant) => {
console.log("Participant Left: ", participant);
teardownParticipant(participant);
});
// Helper functions for managing DOM
function setupParticipant({ isLocal, id }) {
const groupId = isLocal ? "local-media" : "remote-media";
const groupContainer = document.getElementById(groupId);
const participantContainerId = isLocal ? "local" : id;
const participantContainer = createContainer(participantContainerId);
const videoEl = createVideoEl(participantContainerId);
participantContainer.appendChild(videoEl);
groupContainer.appendChild(participantContainer);
return videoEl;
}
function teardownParticipant({ isLocal, id }) {
const groupId = isLocal ? "local-media" : "remote-media";
const groupContainer = document.getElementById(groupId);
const participantContainerId = isLocal ? "local" : id;
const participantDiv = document.getElementById(
participantContainerId + "-container"
);
if (!participantDiv) {
return;
}
groupContainer.removeChild(participantDiv);
}
function createVideoEl(id) {
const videoEl = document.createElement("video");
videoEl.id = id;
videoEl.autoplay = true;
videoEl.playsInline = true;
videoEl.srcObject = new MediaStream();
return videoEl;
}
function createContainer(id) {
const participantContainer = document.createElement("div");
participantContainer.classList = "participant-container";
participantContainer.id = id + "-container";
return participantContainer;
}
```
## 创建 joinStage 5：加入舞台
通过最终加入舞台来完成 `joinStage` 函数吧！
```
try {
await stage.join();
} catch (err) {
joining = false;
connected = false;
console.error(err.message);
}
```
## 创建 leaveStage
定义 `leaveStage` 函数，以调用离开按钮。
```
const leaveStage = async () => {
stage.leave();
joining = false;
connected = false;
};
```
## 初始化输入事件处理程序
把 `app.js` 文件添加到最后一个函数。加载页面时会立即调用此函数，并建立用于加入和离开舞台的事件处理程序。
```
const init = async () => {
try {
// Prevents issues on Safari/FF so devices are not blank
await navigator.mediaDevices.getUserMedia({ video: true, audio: true });
} catch (e) {
alert(
"Problem retrieving media! Enable camera and microphone permissions."
);
}
joinButton.addEventListener("click", () => {
joinStage();
});
leaveButton.addEventListener("click", () => {
leaveStage();
joinButton.style = "display: inline-block";
leaveButton.style = "display: none";
});
};
init(); // call the function
```
## 运行应用程序并提供令牌
这时您可以在本地或与其他人共享网页，只需[打开页面](#getting-started-pub-sub-web-media)，输入参与者令牌并加入舞台即可。
## 接下来做什么？
有关涉及 npm、React 等的更多详细示例，请参阅 [IVS 广播 SDK：网络指南（实时直播功能指南）](broadcast-web.md)。
# 使用 IVS Android 广播 SDK 发布和订阅
本部分将引导您完成使用 Android 应用程序发布和订阅舞台所涉及的步骤。
## 创建视图
首先使用自动创建的 `activity_main.xml` 文件为应用程序创建一个简单的布局。此布局包含要添加到令牌的 `EditText`、加入 `Button`、显示舞台状态的 `TextView` 和切换发布的 `CheckBox`。
![\[为您的 Android 应用程序设置发布布局。\]](http://docs.aws.amazon.com/zh_cn/ivs/latest/RealTimeUserGuide/images/Publish_Android_1.png)
以下是视图背后的 XML：
```
```
我们在此处引用了几个字符串 ID，因此现在将创建整个 `strings.xml` 文件：
```
BasicRealTime
Join
Leave
Participant Token
Publish
State: %1$s
```
将 XML 中的这些视图链接到 `MainActivity.kt`：
```
import android.widget.Button
import android.widget.CheckBox
import android.widget.EditText
import android.widget.TextView
import androidx.recyclerview.widget.RecyclerView
private lateinit var checkboxPublish: CheckBox
private lateinit var recyclerView: RecyclerView
private lateinit var buttonJoin: Button
private lateinit var textViewState: TextView
private lateinit var editTextToken: EditText
override fun onCreate(savedInstanceState: Bundle?) {
super.onCreate(savedInstanceState)
setContentView(R.layout.activity_main)
checkboxPublish = findViewById(R.id.main_publish_checkbox)
recyclerView = findViewById(R.id.main_recycler_view)
buttonJoin = findViewById(R.id.main_join)
textViewState = findViewById(R.id.main_state)
editTextToken = findViewById(R.id.main_token)
}
```
现在我们为 `RecyclerView` 创建一个项目视图。要执行此操作，右键单击 `res/layout` 目录，然后选择**创建 > 布局资源文件**。将此新文件命名为 `item_stage_participant.xml`。
![\[为 Android 应用程序 RecyclerView 创建项目视图。\]](http://docs.aws.amazon.com/zh_cn/ivs/latest/RealTimeUserGuide/images/Publish_Android_2.png)
此项目的布局很简单：它包含用于渲染参与者视频流的视图和用于显示参与者有关信息的标签列表：
![\[为 Android 应用程序 RecyclerView 创建项目视图 – 标签。\]](http://docs.aws.amazon.com/zh_cn/ivs/latest/RealTimeUserGuide/images/Publish_Android_3.png)
以下是 XML：
```
```
这个 XML 文件扩大了还没有创建的类 `ParticipantItem`。由于 XML 包含完整的命名空间，因此请务必将此 XML 文件更新到您的命名空间。创建这个类并设置视图，但暂时将其保留为空。
创建一个新的 Kotlin 类，`ParticipantItem`：
```
package com.amazonaws.ivs.realtime.basicrealtime
import android.content.Context
import android.util.AttributeSet
import android.widget.FrameLayout
import android.widget.TextView
import kotlin.math.roundToInt
class ParticipantItem @JvmOverloads constructor(
context: Context,
attrs: AttributeSet? = null,
defStyleAttr: Int = 0,
defStyleRes: Int = 0,
) : FrameLayout(context, attrs, defStyleAttr, defStyleRes) {
private lateinit var previewContainer: FrameLayout
private lateinit var textViewParticipantId: TextView
private lateinit var textViewPublish: TextView
private lateinit var textViewSubscribe: TextView
private lateinit var textViewVideoMuted: TextView
private lateinit var textViewAudioMuted: TextView
private lateinit var textViewAudioLevel: TextView
override fun onFinishInflate() {
super.onFinishInflate()
previewContainer = findViewById(R.id.participant_preview_container)
textViewParticipantId = findViewById(R.id.participant_participant_id)
textViewPublish = findViewById(R.id.participant_publishing)
textViewSubscribe = findViewById(R.id.participant_subscribed)
textViewVideoMuted = findViewById(R.id.participant_video_muted)
textViewAudioMuted = findViewById(R.id.participant_audio_muted)
textViewAudioLevel = findViewById(R.id.participant_audio_level)
}
}
```
## 权限
要使用相机和麦克风，您需要向用户请求权限。我们为此遵循标准权限流程：
```
override fun onStart() {
super.onStart()
requestPermission()
}
private val requestPermissionLauncher =
registerForActivityResult(ActivityResultContracts.RequestMultiplePermissions()) { permissions ->
if (permissions[Manifest.permission.CAMERA] == true && permissions[Manifest.permission.RECORD_AUDIO] == true) {
viewModel.permissionGranted() // we will add this later
}
}
private val permissions = listOf(
Manifest.permission.CAMERA,
Manifest.permission.RECORD_AUDIO,
)
private fun requestPermission() {
when {
this.hasPermissions(permissions) -> viewModel.permissionGranted() // we will add this later
else -> requestPermissionLauncher.launch(permissions.toTypedArray())
}
}
private fun Context.hasPermissions(permissions: List): Boolean {
return permissions.all {
ContextCompat.checkSelfPermission(this, it) == PackageManager.PERMISSION_GRANTED
}
}
```
## 应用程序状态
应用程序在 `MainViewModel.kt` 中本地跟踪参与者，然后使用 Kotlin 的 [StateFlow](https://kotlinlang.org/api/kotlinx.coroutines/kotlinx-coroutines-core/kotlinx.coroutines.flow/-state-flow/) 将状态传回给 `MainActivity`。
创建一个新的 Kotlin 类 `MainViewModel`：
```
package com.amazonaws.ivs.realtime.basicrealtime
import android.app.Application
import androidx.lifecycle.AndroidViewModel
class MainViewModel(application: Application) : AndroidViewModel(application), Stage.Strategy, StageRenderer {
}
```
在 `MainActivity.kt` 中管理视图模型：
```
import androidx.activity.viewModels
private val viewModel: MainViewModel by viewModels()
```
要使用 `AndroidViewModel` 还有这些 Kotlin `ViewModel` 扩展，您需要将以下内容添加到模块的 `build.gradle` 文件：
```
implementation 'androidx.core:core-ktx:1.10.1'
implementation "androidx.activity:activity-ktx:1.7.2"
implementation 'androidx.appcompat:appcompat:1.6.1'
implementation 'com.google.android.material:material:1.10.0'
implementation "androidx.lifecycle:lifecycle-extensions:2.2.0"
def lifecycle_version = "2.6.1"
implementation "androidx.lifecycle:lifecycle-livedata-ktx:$lifecycle_version"
implementation "androidx.lifecycle:lifecycle-viewmodel-ktx:$lifecycle_version"
implementation 'androidx.constraintlayout:constraintlayout:2.1.4'
```
### RecyclerView 适配器
创建一个简单的 `RecyclerView.Adapter` 子类来跟踪参与者并更新舞台活动中的 `RecyclerView`。但首先，要一个代表参与者的类。创建一个新的 Kotlin 类 `StageParticipant`：
```
package com.amazonaws.ivs.realtime.basicrealtime
import com.amazonaws.ivs.broadcast.Stage
import com.amazonaws.ivs.broadcast.StageStream
class StageParticipant(val isLocal: Boolean, var participantId: String?) {
var publishState = Stage.PublishState.NOT_PUBLISHED
var subscribeState = Stage.SubscribeState.NOT_SUBSCRIBED
var streams = mutableListOf()
val stableID: String
get() {
return if (isLocal) {
"LocalUser"
} else {
requireNotNull(participantId)
}
}
}
```
将在接下来要创建的 `ParticipantAdapter` 中使用此类。首先定义类并创建一个变量来跟踪参与者：
```
package com.amazonaws.ivs.realtime.basicrealtime
import android.view.LayoutInflater
import android.view.ViewGroup
import androidx.recyclerview.widget.RecyclerView
class ParticipantAdapter : RecyclerView.Adapter() {
private val participants = mutableListOf()
```
在实现其余的覆盖之前，还必须定义 `RecyclerView.ViewHolder`：
```
class ViewHolder(val participantItem: ParticipantItem) : RecyclerView.ViewHolder(participantItem)
```
如此，便可以实现标准 `RecyclerView.Adapter` 覆盖：
```
override fun onCreateViewHolder(parent: ViewGroup, viewType: Int): ViewHolder {
val item = LayoutInflater.from(parent.context)
.inflate(R.layout.item_stage_participant, parent, false) as ParticipantItem
return ViewHolder(item)
}
override fun getItemCount(): Int {
return participants.size
}
override fun getItemId(position: Int): Long =
participants[position]
.stableID
.hashCode()
.toLong()
override fun onBindViewHolder(holder: ViewHolder, position: Int) {
return holder.participantItem.bind(participants[position])
}
override fun onBindViewHolder(holder: ViewHolder, position: Int, payloads: MutableList) {
val updates = payloads.filterIsInstance()
if (updates.isNotEmpty()) {
updates.forEach { holder.participantItem.bind(it) // implemented later }
} else {
super.onBindViewHolder(holder, position, payloads)
}
}
```
最后，我们添加了新方法，参与者发生更改时，将从 `MainViewModel` 中调用这些新方法。这些方法是适配器上的标准 CRUD 操作。
```
fun participantJoined(participant: StageParticipant) {
participants.add(participant)
notifyItemInserted(participants.size - 1)
}
fun participantLeft(participantId: String) {
val index = participants.indexOfFirst { it.participantId == participantId }
if (index != -1) {
participants.removeAt(index)
notifyItemRemoved(index)
}
}
fun participantUpdated(participantId: String?, update: (participant: StageParticipant) -> Unit) {
val index = participants.indexOfFirst { it.participantId == participantId }
if (index != -1) {
update(participants[index])
notifyItemChanged(index, participants[index])
}
}
```
返回 `MainViewModel`，需要创建并保留对此适配器的引用：
```
internal val participantAdapter = ParticipantAdapter()
```
## 阶段状态
还需要跟踪 `MainViewModel` 内的某些舞台状态。现在来定义这些属性：
```
private val _connectionState = MutableStateFlow(Stage.ConnectionState.DISCONNECTED)
val connectionState = _connectionState.asStateFlow()
private var publishEnabled: Boolean = false
set(value) {
field = value
// Because the strategy returns the value of `checkboxPublish.isChecked`, just call `refreshStrategy`.
stage?.refreshStrategy()
}
private var deviceDiscovery: DeviceDiscovery? = null
private var stage: Stage? = null
private var streams = mutableListOf()
```
要在加入舞台之前查看自己的预览，立即创建本地参与者：
```
init {
deviceDiscovery = DeviceDiscovery(application)
// Create a local participant immediately to render our camera preview and microphone stats
val localParticipant = StageParticipant(true, null)
participantAdapter.participantJoined(localParticipant)
}
```
确保在清理 `ViewModel` 时也清理这些资源。立即覆盖 `onCleared()`，以便不会忘记清理这些资源。
```
override fun onCleared() {
stage?.release()
deviceDiscovery?.release()
deviceDiscovery = null
super.onCleared()
}
```
现在，一旦授予权限，就会填充本地 `streams` 属性，实施之前调用的 `permissionsGranted` 方法：
```
internal fun permissionGranted() {
val deviceDiscovery = deviceDiscovery ?: return
streams.clear()
val devices = deviceDiscovery.listLocalDevices()
// Camera
devices
.filter { it.descriptor.type == Device.Descriptor.DeviceType.CAMERA }
.maxByOrNull { it.descriptor.position == Device.Descriptor.Position.FRONT }
?.let { streams.add(ImageLocalStageStream(it)) }
// Microphone
devices
.filter { it.descriptor.type == Device.Descriptor.DeviceType.MICROPHONE }
.maxByOrNull { it.descriptor.isDefault }
?.let { streams.add(AudioLocalStageStream(it)) }
stage?.refreshStrategy()
// Update our local participant with these new streams
participantAdapter.participantUpdated(null) {
it.streams.clear()
it.streams.addAll(streams)
}
}
```
## 实施舞台 SDK
三个核心[概念](android-publish-subscribe.md#android-publish-subscribe-concepts)构成了实时功能的基础：舞台、策略和渲染器。设计目标是最大限度地减少构建有效产品所需的客户端逻辑量。
### Stage.Strategy
`Stage.Strategy` 实施很简单：
```
override fun stageStreamsToPublishForParticipant(
stage: Stage,
participantInfo: ParticipantInfo
): MutableList {
// Return the camera and microphone to be published.
// This is only called if `shouldPublishFromParticipant` returns true.
return streams
}
override fun shouldPublishFromParticipant(stage: Stage, participantInfo: ParticipantInfo): Boolean {
return publishEnabled
}
override fun shouldSubscribeToParticipant(stage: Stage, participantInfo: ParticipantInfo): Stage.SubscribeType {
// Subscribe to both audio and video for all publishing participants.
return Stage.SubscribeType.AUDIO_VIDEO
}
```
总而言之，要根据内部 `publishEnabled` 状态发布内容，如果要发布，将发布之前收集的流。最后，对于此示例，我们将始终订阅其他参与者并接收他们的音频和视频。
### StageRenderer
考虑到函数的数量，尽管 `StageRenderer` 包含更多的代码，但是它实施起来也相当简单。此渲染器中的一般方法是，SDK 通知参与者发生更改时更新 `ParticipantAdapter`。在某些情况下，我们会以不同的方式处理本地参与者，因为我们决定自己管理这些参与者，这样他们就可以在加入舞台之前看到自己的相机预览。
```
override fun onError(exception: BroadcastException) {
Toast.makeText(getApplication(), "onError ${exception.localizedMessage}", Toast.LENGTH_LONG).show()
Log.e("BasicRealTime", "onError $exception")
}
override fun onConnectionStateChanged(
stage: Stage,
connectionState: Stage.ConnectionState,
exception: BroadcastException?
) {
_connectionState.value = connectionState
}
override fun onParticipantJoined(stage: Stage, participantInfo: ParticipantInfo) {
if (participantInfo.isLocal) {
// If this is the local participant joining the stage, update the participant with a null ID because we
// manually added that participant when setting up our preview
participantAdapter.participantUpdated(null) {
it.participantId = participantInfo.participantId
}
} else {
// If they are not local, add them normally
participantAdapter.participantJoined(
StageParticipant(
participantInfo.isLocal,
participantInfo.participantId
)
)
}
}
override fun onParticipantLeft(stage: Stage, participantInfo: ParticipantInfo) {
if (participantInfo.isLocal) {
// If this is the local participant leaving the stage, update the ID but keep it around because
// we want to keep the camera preview active
participantAdapter.participantUpdated(participantInfo.participantId) {
it.participantId = null
}
} else {
// If they are not local, have them leave normally
participantAdapter.participantLeft(participantInfo.participantId)
}
}
override fun onParticipantPublishStateChanged(
stage: Stage,
participantInfo: ParticipantInfo,
publishState: Stage.PublishState
) {
// Update the publishing state of this participant
participantAdapter.participantUpdated(participantInfo.participantId) {
it.publishState = publishState
}
}
override fun onParticipantSubscribeStateChanged(
stage: Stage,
participantInfo: ParticipantInfo,
subscribeState: Stage.SubscribeState
) {
// Update the subscribe state of this participant
participantAdapter.participantUpdated(participantInfo.participantId) {
it.subscribeState = subscribeState
}
}
override fun onStreamsAdded(stage: Stage, participantInfo: ParticipantInfo, streams: MutableList) {
// We don't want to take any action for the local participant because we track those streams locally
if (participantInfo.isLocal) {
return
}
// For remote participants, add these new streams to that participant's streams array.
participantAdapter.participantUpdated(participantInfo.participantId) {
it.streams.addAll(streams)
}
}
override fun onStreamsRemoved(stage: Stage, participantInfo: ParticipantInfo, streams: MutableList) {
// We don't want to take any action for the local participant because we track those streams locally
if (participantInfo.isLocal) {
return
}
// For remote participants, remove these streams from that participant's streams array.
participantAdapter.participantUpdated(participantInfo.participantId) {
it.streams.removeAll(streams)
}
}
override fun onStreamsMutedChanged(
stage: Stage,
participantInfo: ParticipantInfo,
streams: MutableList
) {
// We don't want to take any action for the local participant because we track those streams locally
if (participantInfo.isLocal) {
return
}
// For remote participants, notify the adapter that the participant has been updated. There is no need to modify
// the `streams` property on the `StageParticipant` because it is the same `StageStream` instance. Just
// query the `isMuted` property again.
participantAdapter.participantUpdated(participantInfo.participantId) {}
}
```
## 实施自定义 RecyclerView LayoutManager
安排不同数量的参与者可能很复杂。您希望参与者占据整个父视图的框架，但是不想单独处理每个参与者配置。为了简单起见，将逐步实施 `RecyclerView.LayoutManager`。
创建另一个新类 `StageLayoutManager`，它应该扩展 `GridLayoutManager`。此类旨在根据基于流的行/列布局中的参与者数量计算每个参与者的布局。每行的高度与其他行相同，但每行列的宽度各不相同。有关如何自定义该行为的说明，请参阅 `layouts` 变量上方的代码注释。
```
package com.amazonaws.ivs.realtime.basicrealtime
import android.content.Context
import androidx.recyclerview.widget.GridLayoutManager
import androidx.recyclerview.widget.RecyclerView
class StageLayoutManager(context: Context?) : GridLayoutManager(context, 6) {
companion object {
/**
* This 2D array contains the description of how the grid of participants should be rendered
* The index of the 1st dimension is the number of participants needed to active that configuration
* Meaning if there is 1 participant, index 0 will be used. If there are 5 participants, index 4 will be used.
*
* The 2nd dimension is a description of the layout. The length of the array is the number of rows that
* will exist, and then each number within that array is the number of columns in each row.
*
* See the code comments next to each index for concrete examples.
*
* This can be customized to fit any layout configuration needed.
*/
val layouts: List> = listOf(
// 1 participant
listOf(1), // 1 row, full width
// 2 participants
listOf(1, 1), // 2 rows, all columns are full width
// 3 participants
listOf(1, 2), // 2 rows, first row's column is full width then 2nd row's columns are 1/2 width
// 4 participants
listOf(2, 2), // 2 rows, all columns are 1/2 width
// 5 participants
listOf(1, 2, 2), // 3 rows, first row's column is full width, 2nd and 3rd row's columns are 1/2 width
// 6 participants
listOf(2, 2, 2), // 3 rows, all column are 1/2 width
// 7 participants
listOf(2, 2, 3), // 3 rows, 1st and 2nd row's columns are 1/2 width, 3rd row's columns are 1/3rd width
// 8 participants
listOf(2, 3, 3),
// 9 participants
listOf(3, 3, 3),
// 10 participants
listOf(2, 3, 2, 3),
// 11 participants
listOf(2, 3, 3, 3),
// 12 participants
listOf(3, 3, 3, 3),
)
}
init {
spanSizeLookup = object : SpanSizeLookup() {
override fun getSpanSize(position: Int): Int {
if (itemCount <= 0) {
return 1
}
// Calculate the row we're in
val config = layouts[itemCount - 1]
var row = 0
var curPosition = position
while (curPosition - config[row] >= 0) {
curPosition -= config[row]
row++
}
// spanCount == max spans, config[row] = number of columns we want
// So spanCount / config[row] would be something like 6 / 3 if we want 3 columns.
// So this will take up 2 spans, with a max of 6 is 1/3rd of the view.
return spanCount / config[row]
}
}
}
override fun onLayoutChildren(recycler: RecyclerView.Recycler?, state: RecyclerView.State?) {
if (itemCount <= 0 || state?.isPreLayout == true) return
val parentHeight = height
val itemHeight = parentHeight / layouts[itemCount - 1].size // height divided by number of rows.
// Set the height of each view based on how many rows exist for the current participant count.
for (i in 0 until childCount) {
val child = getChildAt(i) ?: continue
val layoutParams = child.layoutParams as RecyclerView.LayoutParams
if (layoutParams.height != itemHeight) {
layoutParams.height = itemHeight
child.layoutParams = layoutParams
}
}
// After we set the height for all our views, call super.
// This works because our RecyclerView can not scroll and all views are always visible with stable IDs.
super.onLayoutChildren(recycler, state)
}
override fun canScrollVertically(): Boolean = false
override fun canScrollHorizontally(): Boolean = false
}
```
回到 `MainActivity.kt` 中，我们需要为 `RecyclerView` 设置适配器和布局管理器：
```
// In onCreate after setting recyclerView.
recyclerView.layoutManager = StageLayoutManager(this)
recyclerView.adapter = viewModel.participantAdapter
```
## 挂接 UI 操作
即将完成所有操作；只需要挂接几个 UI 操作。
首先让 `MainActivity` 观察 `MainViewModel` 的 `StateFlow` 变更：
```
// At the end of your onCreate method
lifecycleScope.launch {
repeatOnLifecycle(Lifecycle.State.CREATED) {
viewModel.connectionState.collect { state ->
buttonJoin.setText(if (state == ConnectionState.DISCONNECTED) R.string.join else R.string.leave)
textViewState.text = getString(R.string.state, state.name)
}
}
}
```
接下来，将侦听器添加到“加入”按钮和“发布”复选框中：
```
buttonJoin.setOnClickListener {
viewModel.joinStage(editTextToken.text.toString())
}
checkboxPublish.setOnCheckedChangeListener { _, isChecked ->
viewModel.setPublishEnabled(isChecked)
}
```
上述两个事件调用正在实施的 `MainViewModel` 中的功能：
```
internal fun joinStage(token: String) {
if (_connectionState.value != Stage.ConnectionState.DISCONNECTED) {
// If we're already connected to a stage, leave it.
stage?.leave()
} else {
if (token.isEmpty()) {
Toast.makeText(getApplication(), "Empty Token", Toast.LENGTH_SHORT).show()
return
}
try {
// Destroy the old stage first before creating a new one.
stage?.release()
val stage = Stage(getApplication(), token, this)
stage.addRenderer(this)
stage.join()
this.stage = stage
} catch (e: BroadcastException) {
Toast.makeText(getApplication(), "Failed to join stage ${e.localizedMessage}", Toast.LENGTH_LONG).show()
e.printStackTrace()
}
}
}
internal fun setPublishEnabled(enabled: Boolean) {
publishEnabled = enabled
}
```
## 渲染参与者
最后，需要将从 SDK 收到的数据渲染到之前创建的参与者项目上。我们已经完成了 `RecyclerView` 逻辑，所以只需要在 `ParticipantItem` 中实施 `bind` API。
首先添加空函数，然后逐步进行操作：
```
fun bind(participant: StageParticipant) {
}
```
首先，处理简易状态、参与者 ID、发布状态和订阅状态。对于这些，直接更新 `TextViews`：
```
val participantId = if (participant.isLocal) {
"You (${participant.participantId ?: "Disconnected"})"
} else {
participant.participantId
}
textViewParticipantId.text = participantId
textViewPublish.text = participant.publishState.name
textViewSubscribe.text = participant.subscribeState.name
```
接下来，更新音频和视频的静音状态。要进入静音状态，需要找到来自流数组的 `ImageDevice` 和 `AudioDevice`。要优化性能，需要记住最后连接的设备 ID。
```
// This belongs outside the `bind` API.
private var imageDeviceUrn: String? = null
private var audioDeviceUrn: String? = null
// This belongs inside the `bind` API.
val newImageStream = participant
.streams
.firstOrNull { it.device is ImageDevice }
textViewVideoMuted.text = if (newImageStream != null) {
if (newImageStream.muted) "Video muted" else "Video not muted"
} else {
"No video stream"
}
val newAudioStream = participant
.streams
.firstOrNull { it.device is AudioDevice }
textViewAudioMuted.text = if (newAudioStream != null) {
if (newAudioStream.muted) "Audio muted" else "Audio not muted"
} else {
"No audio stream"
}
```
最后，渲染 `imageDevice` 的预览：
```
if (newImageStream?.device?.descriptor?.urn != imageDeviceUrn) {
// If the device has changed, remove all subviews from the preview container
previewContainer.removeAllViews()
(newImageStream?.device as? ImageDevice)?.let {
val preview = it.getPreviewView(BroadcastConfiguration.AspectMode.FIT)
previewContainer.addView(preview)
preview.layoutParams = FrameLayout.LayoutParams(
FrameLayout.LayoutParams.MATCH_PARENT,
FrameLayout.LayoutParams.MATCH_PARENT
)
}
}
imageDeviceUrn = newImageStream?.device?.descriptor?.urn
```
然后显示来自 `audioDevice` 的音频统计数据：
```
if (newAudioStream?.device?.descriptor?.urn != audioDeviceUrn) {
(newAudioStream?.device as? AudioDevice)?.let {
it.setStatsCallback { _, rms ->
textViewAudioLevel.text = "Audio Level: ${rms.roundToInt()} dB"
}
}
}
audioDeviceUrn = newAudioStream?.device?.descriptor?.urn
```
# 使用 IVS iOS 广播 SDK 发布和订阅
本部分将引导您完成使用 iOS 应用程序发布和订阅舞台所涉及的步骤。
## 创建视图
首先使用自动创建的 `ViewController.swift` 文件来导入 `AmazonIVSBroadcast`，然后添加一些要链接的 `@IBOutlets`：
```
import AmazonIVSBroadcast
class ViewController: UIViewController {
@IBOutlet private var textFieldToken: UITextField!
@IBOutlet private var buttonJoin: UIButton!
@IBOutlet private var labelState: UILabel!
@IBOutlet private var switchPublish: UISwitch!
@IBOutlet private var collectionViewParticipants: UICollectionView!
```
现在创建这些视图，然后在 `Main.storyboard` 中将其链接起来。以下是将使用的视图结构：
![\[使用 Main.storyboard 创建 iOS 视图。\]](http://docs.aws.amazon.com/zh_cn/ivs/latest/RealTimeUserGuide/images/Publish_iOS_1.png)
对于 AutoLayout 配置，需要自定义三个视图。第一个视图是**集合视图参与者** (`UICollectionView`)。将**开头**、**结尾**和**底部**绑定到**安全区域**。同时将**顶部**绑定到**控件容器**。
![\[自定义 iOS 集合视图参与者视图。\]](http://docs.aws.amazon.com/zh_cn/ivs/latest/RealTimeUserGuide/images/Publish_iOS_2.png)
第二个视图是**控件容器**。将**开头**、**结尾**和**顶部**绑定到**安全区域**：
![\[自定义 iOS 控件容器视图。\]](http://docs.aws.amazon.com/zh_cn/ivs/latest/RealTimeUserGuide/images/Publish_iOS_3.png)
第三个也是最后一个视图是**垂直堆栈视图**。将**顶部**、**开头**、**结尾**和**底部**绑定到**超级视图**。对于样式，将间距设置为 8 而不是 0。
![\[自定义 iOS 垂直堆栈视图。\]](http://docs.aws.amazon.com/zh_cn/ivs/latest/RealTimeUserGuide/images/Publish_iOS_4.png)
**UIStackViews** 将处理剩余视图的布局。对于所有三个 **UIStackViews**，使用**填充**作为**对齐**和**分布**。
![\[使用 UIStackViews 自定义剩余的 iOS 视图。\]](http://docs.aws.amazon.com/zh_cn/ivs/latest/RealTimeUserGuide/images/Publish_iOS_5.png)
最后，将这些观点链接到 `ViewController`。从上面绘制以下视图：
+ 将**文本字段加入**绑定到 `textFieldToken`。
+ 将**按钮加入**绑定到 `buttonJoin`。
+ 将**标签状态**绑定到 `labelState`。
+ 将**切换发布**绑定到 `switchPublish`。
+ 将**集合视图参与者**绑定到 `collectionViewParticipants`。
也可以利用这段时间将**集合视图参与者**项的 `dataSource` 设置为所属 `ViewController`：
![\[设置 iOS 应用程序的集合视图参与者的 dataSource。\]](http://docs.aws.amazon.com/zh_cn/ivs/latest/RealTimeUserGuide/images/Publish_iOS_6.png)
现在创建 `UICollectionViewCell` 子类以在其中渲染参与者。首先创建一个新的 **Cocoa Touch 类**文件：
![\[创建 UICollectionViewCell 来渲染 iOS 的实时参与者。\]](http://docs.aws.amazon.com/zh_cn/ivs/latest/RealTimeUserGuide/images/Publish_iOS_7.png)
将其命名为 `ParticipantUICollectionViewCell` 并使其成为 Swift 中 `UICollectionViewCell` 的子类。再次从 Swift 文件开始，创建要链接的 `@IBOutlets`：
```
import AmazonIVSBroadcast
class ParticipantCollectionViewCell: UICollectionViewCell {
@IBOutlet private var viewPreviewContainer: UIView!
@IBOutlet private var labelParticipantId: UILabel!
@IBOutlet private var labelSubscribeState: UILabel!
@IBOutlet private var labelPublishState: UILabel!
@IBOutlet private var labelVideoMuted: UILabel!
@IBOutlet private var labelAudioMuted: UILabel!
@IBOutlet private var labelAudioVolume: UILabel!
```
在关联的 XIB 文件中，创建此视图层次结构：
![\[在关联的 XIB 文件中，创建 iOS 视图层次结构。\]](http://docs.aws.amazon.com/zh_cn/ivs/latest/RealTimeUserGuide/images/Publish_iOS_8.png)
对于 AutoLayout，再次修改三个视图。第一个视图是**视图预览容器**。将**结尾**、**开头**、**顶部**和**底部**设置为**参与者集合视图单元格**。
![\[自定义 iOS 视图预览容器视图。\]](http://docs.aws.amazon.com/zh_cn/ivs/latest/RealTimeUserGuide/images/Publish_iOS_9.png)
第二个视图是**视图**。将**开头**和**顶部**设置为**参与者集合视图单元格**并将值更改为 4。
![\[自定义 iOS 视图的视图。\]](http://docs.aws.amazon.com/zh_cn/ivs/latest/RealTimeUserGuide/images/Publish_iOS_10.png)
第三个视图是**堆栈视图**。将**结尾**、**开头**、**顶部**和**底部**设置为**超级视图**并将值更改为 4。
![\[自定义 iOS 堆栈视图的视图。\]](http://docs.aws.amazon.com/zh_cn/ivs/latest/RealTimeUserGuide/images/Publish_iOS_11.png)
## 权限和空闲计时器
返回 `ViewController`，禁用系统空闲计时器，以防止设备在使用应用程序时进入睡眠状态：
```
override func viewDidAppear(_ animated: Bool) {
super.viewDidAppear(animated)
// Prevent the screen from turning off during a call.
UIApplication.shared.isIdleTimerDisabled = true
}
override func viewDidDisappear(_ animated: Bool) {
super.viewDidDisappear(animated)
UIApplication.shared.isIdleTimerDisabled = false
}
```
接下来，向系统请求相机和麦克风权限：
```
private func checkPermissions() {
checkOrGetPermission(for: .video) { [weak self] granted in
guard granted else {
print("Video permission denied")
return
}
self?.checkOrGetPermission(for: .audio) { [weak self] granted in
guard granted else {
print("Audio permission denied")
return
}
self?.setupLocalUser() // we will cover this later
}
}
}
private func checkOrGetPermission(for mediaType: AVMediaType, _ result: @escaping (Bool) -> Void) {
func mainThreadResult(_ success: Bool) {
DispatchQueue.main.async {
result(success)
}
}
switch AVCaptureDevice.authorizationStatus(for: mediaType) {
case .authorized: mainThreadResult(true)
case .notDetermined:
AVCaptureDevice.requestAccess(for: mediaType) { granted in
mainThreadResult(granted)
}
case .denied, .restricted: mainThreadResult(false)
@unknown default: mainThreadResult(false)
}
}
```
## 应用程序状态
需要使用之前创建的布局文件配置 `collectionViewParticipants`：
```
override func viewDidLoad() {
super.viewDidLoad()
// We render everything to exactly the frame, so don't allow scrolling.
collectionViewParticipants.isScrollEnabled = false
collectionViewParticipants.register(UINib(nibName: "ParticipantCollectionViewCell", bundle: .main), forCellWithReuseIdentifier: "ParticipantCollectionViewCell")
}
```
为了代表每个参与者，创建了一个名为 `StageParticipant` 的简单结构。这可以包含在 `ViewController.swift` 文件中，也可以创建一个新文件。
```
import Foundation
import AmazonIVSBroadcast
struct StageParticipant {
let isLocal: Bool
var participantId: String?
var publishState: IVSParticipantPublishState = .notPublished
var subscribeState: IVSParticipantSubscribeState = .notSubscribed
var streams: [IVSStageStream] = []
init(isLocal: Bool, participantId: String?) {
self.isLocal = isLocal
self.participantId = participantId
}
}
```
为了追踪这些参与者，我们将他们的数组作为私有财产保留在 `ViewController` 中：
```
private var participants = [StageParticipant]()
```
此属性将用于为之前从故事情节链接的 `UICollectionViewDataSource` 提供支持：
```
extension ViewController: UICollectionViewDataSource {
func collectionView(_ collectionView: UICollectionView, numberOfItemsInSection section: Int) -> Int {
return participants.count
}
func collectionView(_ collectionView: UICollectionView, cellForItemAt indexPath: IndexPath) -> UICollectionViewCell {
if let cell = collectionView.dequeueReusableCell(withReuseIdentifier: "ParticipantCollectionViewCell", for: indexPath) as? ParticipantCollectionViewCell {
cell.set(participant: participants[indexPath.row])
return cell
} else {
fatalError("Couldn't load custom cell type 'ParticipantCollectionViewCell'")
}
}
}
```
要在加入舞台之前查看自己的预览，立即创建本地参与者：
```
override func viewDidLoad() {
/* existing UICollectionView code */
participants.append(StageParticipant(isLocal: true, participantId: nil))
}
```
这会导致在应用程序运行后立即渲染代表本地参与者的参与者单元格。
用户希望在加入舞台之前能够看到自己，所以接下来要实施之前从权限处理代码中调用的 `setupLocalUser()` 方法。将相机和麦克风引用存储为 `IVSLocalStageStream` 对象。
```
private var streams = [IVSLocalStageStream]()
private let deviceDiscovery = IVSDeviceDiscovery()
private func setupLocalUser() {
// Gather our camera and microphone once permissions have been granted
let devices = deviceDiscovery.listLocalDevices()
streams.removeAll()
if let camera = devices.compactMap({ $0 as? IVSCamera }).first {
streams.append(IVSLocalStageStream(device: camera))
// Use a front camera if available.
if let frontSource = camera.listAvailableInputSources().first(where: { $0.position == .front }) {
camera.setPreferredInputSource(frontSource)
}
}
if let mic = devices.compactMap({ $0 as? IVSMicrophone }).first {
streams.append(IVSLocalStageStream(device: mic))
}
participants[0].streams = streams
participantsChanged(index: 0, changeType: .updated)
}
```
在这里，通过 SDK 找到设备的相机和麦克风，并将它们存储在本地 `streams` 对象中，然后将第一个参与者（之前创建的本地参与者）的 `streams` 数组分配给 `streams`。最后使用 `index` 0 和 `updated` 的 `changeType` 调用 `participantsChanged`。此函数是一个帮助程序函数，用于使用精美的动画更新 `UICollectionView`。它看起来像这样：
```
private func participantsChanged(index: Int, changeType: ChangeType) {
switch changeType {
case .joined:
collectionViewParticipants?.insertItems(at: [IndexPath(item: index, section: 0)])
case .updated:
// Instead of doing reloadItems, just grab the cell and update it ourselves. It saves a create/destroy of a cell
// and more importantly fixes some UI flicker. We disable scrolling so the index path per cell
// never changes.
if let cell = collectionViewParticipants?.cellForItem(at: IndexPath(item: index, section: 0)) as? ParticipantCollectionViewCell {
cell.set(participant: participants[index])
}
case .left:
collectionViewParticipants?.deleteItems(at: [IndexPath(item: index, section: 0)])
}
}
```
暂时不用担心 `cell.set`；稍后会讨论这个问题，但这就是我们将根据参与者渲染单元格内容的地方。
`ChangeType` 是简单的枚举：
```
enum ChangeType {
case joined, updated, left
}
```
最后，要跟踪舞台是否已连接。我们使用简单的 `bool` 进行跟踪，其将在用户界面自行更新时自动更新。
```
private var connectingOrConnected = false {
didSet {
buttonJoin.setTitle(connectingOrConnected ? "Leave" : "Join", for: .normal)
buttonJoin.tintColor = connectingOrConnected ? .systemRed : .systemBlue
}
}
```
## 实施舞台 SDK
三个核心[概念](ios-publish-subscribe.md#ios-publish-subscribe-concepts)构成了实时功能的基础：舞台、策略和渲染器。设计目标是最大限度地减少构建有效产品所需的客户端逻辑量。
### IVSStageStrategy
`IVSStageStrategy` 实施很简单：
```
extension ViewController: IVSStageStrategy {
func stage(_ stage: IVSStage, streamsToPublishForParticipant participant: IVSParticipantInfo) -> [IVSLocalStageStream] {
// Return the camera and microphone to be published.
// This is only called if `shouldPublishParticipant` returns true.
return streams
}
func stage(_ stage: IVSStage, shouldPublishParticipant participant: IVSParticipantInfo) -> Bool {
// Our publish status is based directly on the UISwitch view
return switchPublish.isOn
}
func stage(_ stage: IVSStage, shouldSubscribeToParticipant participant: IVSParticipantInfo) -> IVSStageSubscribeType {
// Subscribe to both audio and video for all publishing participants.
return .audioVideo
}
}
```
总而言之，只有在发布开关处于“打开”位置时才会发布，如果要发布，将发布之前收集的流。最后，对于此示例，我们将始终订阅其他参与者并接收他们的音频和视频。
### IVSStageRenderer
考虑到函数的数量，尽管 `IVSStageRenderer` 包含更多的代码，但是它实施起来也相当简单。此渲染器中的一般方法是，SDK 通知参与者发生更改时更新 `participants` 数组。在某些情况下，我们会以不同的方式处理本地参与者，因为我们决定自己管理这些参与者，这样他们就可以在加入舞台之前看到自己的相机预览。
```
extension ViewController: IVSStageRenderer {
func stage(_ stage: IVSStage, didChange connectionState: IVSStageConnectionState, withError error: Error?) {
labelState.text = connectionState.text
connectingOrConnected = connectionState != .disconnected
}
func stage(_ stage: IVSStage, participantDidJoin participant: IVSParticipantInfo) {
if participant.isLocal {
// If this is the local participant joining the Stage, update the first participant in our array because we
// manually added that participant when setting up our preview
participants[0].participantId = participant.participantId
participantsChanged(index: 0, changeType: .updated)
} else {
// If they are not local, add them to the array as a newly joined participant.
participants.append(StageParticipant(isLocal: false, participantId: participant.participantId))
participantsChanged(index: (participants.count - 1), changeType: .joined)
}
}
func stage(_ stage: IVSStage, participantDidLeave participant: IVSParticipantInfo) {
if participant.isLocal {
// If this is the local participant leaving the Stage, update the first participant in our array because
// we want to keep the camera preview active
participants[0].participantId = nil
participantsChanged(index: 0, changeType: .updated)
} else {
// If they are not local, find their index and remove them from the array.
if let index = participants.firstIndex(where: { $0.participantId == participant.participantId }) {
participants.remove(at: index)
participantsChanged(index: index, changeType: .left)
}
}
}
func stage(_ stage: IVSStage, participant: IVSParticipantInfo, didChange publishState: IVSParticipantPublishState) {
// Update the publishing state of this participant
mutatingParticipant(participant.participantId) { data in
data.publishState = publishState
}
}
func stage(_ stage: IVSStage, participant: IVSParticipantInfo, didChange subscribeState: IVSParticipantSubscribeState) {
// Update the subscribe state of this participant
mutatingParticipant(participant.participantId) { data in
data.subscribeState = subscribeState
}
}
func stage(_ stage: IVSStage, participant: IVSParticipantInfo, didChangeMutedStreams streams: [IVSStageStream]) {
// We don't want to take any action for the local participant because we track those streams locally
if participant.isLocal { return }
// For remote participants, notify the UICollectionView that they have updated. There is no need to modify
// the `streams` property on the `StageParticipant` because it is the same `IVSStageStream` instance. Just
// query the `isMuted` property again.
if let index = participants.firstIndex(where: { $0.participantId == participant.participantId }) {
participantsChanged(index: index, changeType: .updated)
}
}
func stage(_ stage: IVSStage, participant: IVSParticipantInfo, didAdd streams: [IVSStageStream]) {
// We don't want to take any action for the local participant because we track those streams locally
if participant.isLocal { return }
// For remote participants, add these new streams to that participant's streams array.
mutatingParticipant(participant.participantId) { data in
data.streams.append(contentsOf: streams)
}
}
func stage(_ stage: IVSStage, participant: IVSParticipantInfo, didRemove streams: [IVSStageStream]) {
// We don't want to take any action for the local participant because we track those streams locally
if participant.isLocal { return }
// For remote participants, remove these streams from that participant's streams array.
mutatingParticipant(participant.participantId) { data in
let oldUrns = streams.map { $0.device.descriptor().urn }
data.streams.removeAll(where: { stream in
return oldUrns.contains(stream.device.descriptor().urn)
})
}
}
// A helper function to find a participant by its ID, mutate that participant, and then update the UICollectionView accordingly.
private func mutatingParticipant(_ participantId: String?, modifier: (inout StageParticipant) -> Void) {
guard let index = participants.firstIndex(where: { $0.participantId == participantId }) else {
fatalError("Something is out of sync, investigate if this was a sample app or SDK issue.")
}
var participant = participants[index]
modifier(&participant)
participants[index] = participant
participantsChanged(index: index, changeType: .updated)
}
}
```
此代码使用扩展将连接状态转换为用户友好文本：
```
extension IVSStageConnectionState {
var text: String {
switch self {
case .disconnected: return "Disconnected"
case .connecting: return "Connecting"
case .connected: return "Connected"
@unknown default: fatalError()
}
}
}
```
## 实施自定义 UICollectionViewLayout
安排不同数量的参与者可能很复杂。您希望参与者占据整个父视图的框架，但是不想单独处理每个参与者配置。为了简单起见，将逐步实施 `UICollectionViewLayout`。
创建另一个新文件 `ParticipantCollectionViewLayout.swift`，它应该扩展 `UICollectionViewLayout`。此类将使用另一个名为 `StageLayoutCalculator` 的类，我们很快就会介绍它。类接收每个参与者的计算框架值，然后生成必要的 `UICollectionViewLayoutAttributes` 对象。
```
import Foundation
import UIKit
/**
Code modified from https://developer.apple.com/documentation/uikit/views_and_controls/collection_views/layouts/customizing_collection_view_layouts?language=objc
*/
class ParticipantCollectionViewLayout: UICollectionViewLayout {
private let layoutCalculator = StageLayoutCalculator()
private var contentBounds = CGRect.zero
private var cachedAttributes = [UICollectionViewLayoutAttributes]()
override func prepare() {
super.prepare()
guard let collectionView = collectionView else { return }
cachedAttributes.removeAll()
contentBounds = CGRect(origin: .zero, size: collectionView.bounds.size)
layoutCalculator.calculateFrames(participantCount: collectionView.numberOfItems(inSection: 0),
width: collectionView.bounds.size.width,
height: collectionView.bounds.size.height,
padding: 4)
.enumerated()
.forEach { (index, frame) in
let attributes = UICollectionViewLayoutAttributes(forCellWith: IndexPath(item: index, section: 0))
attributes.frame = frame
cachedAttributes.append(attributes)
contentBounds = contentBounds.union(frame)
}
}
override var collectionViewContentSize: CGSize {
return contentBounds.size
}
override func shouldInvalidateLayout(forBoundsChange newBounds: CGRect) -> Bool {
guard let collectionView = collectionView else { return false }
return !newBounds.size.equalTo(collectionView.bounds.size)
}
override func layoutAttributesForItem(at indexPath: IndexPath) -> UICollectionViewLayoutAttributes? {
return cachedAttributes[indexPath.item]
}
override func layoutAttributesForElements(in rect: CGRect) -> [UICollectionViewLayoutAttributes]? {
var attributesArray = [UICollectionViewLayoutAttributes]()
// Find any cell that sits within the query rect.
guard let lastIndex = cachedAttributes.indices.last, let firstMatchIndex = binSearch(rect, start: 0, end: lastIndex) else {
return attributesArray
}
// Starting from the match, loop up and down through the array until all the attributes
// have been added within the query rect.
for attributes in cachedAttributes[..= rect.minY else { break }
attributesArray.append(attributes)
}
for attributes in cachedAttributes[firstMatchIndex...] {
guard attributes.frame.minY <= rect.maxY else { break }
attributesArray.append(attributes)
}
return attributesArray
}
// Perform a binary search on the cached attributes array.
func binSearch(_ rect: CGRect, start: Int, end: Int) -> Int? {
if end < start { return nil }
let mid = (start + end) / 2
let attr = cachedAttributes[mid]
if attr.frame.intersects(rect) {
return mid
} else {
if attr.frame.maxY < rect.minY {
return binSearch(rect, start: (mid + 1), end: end)
} else {
return binSearch(rect, start: start, end: (mid - 1))
}
}
}
}
```
更重要的是 `StageLayoutCalculator.swift` 类。此类旨在根据基于流的行/列布局中的参与者数量计算每个参与者的框架。每行的高度与其他行相同，但每行列的宽度各不相同。有关如何自定义该行为的说明，请参阅 `layouts` 变量上方的代码注释。
```
import Foundation
import UIKit
class StageLayoutCalculator {
/// This 2D array contains the description of how the grid of participants should be rendered
/// The index of the 1st dimension is the number of participants needed to active that configuration
/// Meaning if there is 1 participant, index 0 will be used. If there are 5 participants, index 4 will be used.
///
/// The 2nd dimension is a description of the layout. The length of the array is the number of rows that
/// will exist, and then each number within that array is the number of columns in each row.
///
/// See the code comments next to each index for concrete examples.
///
/// This can be customized to fit any layout configuration needed.
private let layouts: [[Int]] = [
// 1 participant
[ 1 ], // 1 row, full width
// 2 participants
[ 1, 1 ], // 2 rows, all columns are full width
// 3 participants
[ 1, 2 ], // 2 rows, first row's column is full width then 2nd row's columns are 1/2 width
// 4 participants
[ 2, 2 ], // 2 rows, all columns are 1/2 width
// 5 participants
[ 1, 2, 2 ], // 3 rows, first row's column is full width, 2nd and 3rd row's columns are 1/2 width
// 6 participants
[ 2, 2, 2 ], // 3 rows, all column are 1/2 width
// 7 participants
[ 2, 2, 3 ], // 3 rows, 1st and 2nd row's columns are 1/2 width, 3rd row's columns are 1/3rd width
// 8 participants
[ 2, 3, 3 ],
// 9 participants
[ 3, 3, 3 ],
// 10 participants
[ 2, 3, 2, 3 ],
// 11 participants
[ 2, 3, 3, 3 ],
// 12 participants
[ 3, 3, 3, 3 ],
]
// Given a frame (this could be for a UICollectionView, or a Broadcast Mixer's canvas), calculate the frames for each
// participant, with optional padding.
func calculateFrames(participantCount: Int, width: CGFloat, height: CGFloat, padding: CGFloat) -> [CGRect] {
if participantCount > layouts.count {
fatalError("Only \(layouts.count) participants are supported at this time")
}
if participantCount == 0 {
return []
}
var currentIndex = 0
var lastFrame: CGRect = .zero
// If the height is less than the width, the rows and columns will be flipped.
// Meaning for 6 participants, there will be 2 rows of 3 columns each.
let isVertical = height > width
let halfPadding = padding / 2.0
let layout = layouts[participantCount - 1] // 1 participant is in index 0, so `-1`.
let rowHeight = (isVertical ? height : width) / CGFloat(layout.count)
var frames = [CGRect]()
for row in 0 ..< layout.count {
// layout[row] is the number of columns in a layout
let itemWidth = (isVertical ? width : height) / CGFloat(layout[row])
let segmentFrame = CGRect(x: (isVertical ? 0 : lastFrame.maxX) + halfPadding,
y: (isVertical ? lastFrame.maxY : 0) + halfPadding,
width: (isVertical ? itemWidth : rowHeight) - padding,
height: (isVertical ? rowHeight : itemWidth) - padding)
for column in 0 ..< layout[row] {
var frame = segmentFrame
if isVertical {
frame.origin.x = (itemWidth * CGFloat(column)) + halfPadding
} else {
frame.origin.y = (itemWidth * CGFloat(column)) + halfPadding
}
frames.append(frame)
currentIndex += 1
}
lastFrame = segmentFrame
lastFrame.origin.x += halfPadding
lastFrame.origin.y += halfPadding
}
return frames
}
}
```
返回 `Main.storyboard`，确保将 `UICollectionView` 的布局类设置为刚刚创建的类：
![\[Xcode interface showing storyboard with UICollectionView and its layout settings.\]](http://docs.aws.amazon.com/zh_cn/ivs/latest/RealTimeUserGuide/images/Publish_iOS_12.png)
## 挂接 UI 操作
即将完成所有操作，还需要创建几个 `IBActions`。
首先处理加入按钮。它根据 `connectingOrConnected` 的值做出不同响应。已连接时，它就会离开舞台。如果断开连接，它会从令牌 `UITextField` 中读取文本，并使用此 `IVSStage` 文本创建一个新文本。然后，添加 `ViewController` 作为 `strategy`、`errorDelegate` 和 `IVSStage` 的渲染器，最后异步加入舞台。
```
@IBAction private func joinTapped(_ sender: UIButton) {
if connectingOrConnected {
// If we're already connected to a Stage, leave it.
stage?.leave()
} else {
guard let token = textFieldToken.text else {
print("No token")
return
}
// Hide the keyboard after tapping Join
textFieldToken.resignFirstResponder()
do {
// Destroy the old Stage first before creating a new one.
self.stage = nil
let stage = try IVSStage(token: token, strategy: self)
stage.errorDelegate = self
stage.addRenderer(self)
try stage.join()
self.stage = stage
} catch {
print("Failed to join stage - \(error)")
}
}
}
```
需要挂接的另一个 UI 操作是发布开关：
```
@IBAction private func publishToggled(_ sender: UISwitch) {
// Because the strategy returns the value of `switchPublish.isOn`, just call `refreshStrategy`.
stage?.refreshStrategy()
}
```
## 渲染参与者
最后，需要将从 SDK 收到的数据渲染到之前创建的参与者单元格上。我们已经完成了 `UICollectionView` 逻辑，所以只需要在 `ParticipantCollectionViewCell.swift` 中实施 `set` API。
首先添加 `empty` 函数，然后逐步进行操作：
```
func set(participant: StageParticipant) {
}
```
首先，处理简易状态、参与者 ID、发布状态和订阅状态。对于这些，直接更新 `UILabels`：
```
labelParticipantId.text = participant.isLocal ? "You (\(participant.participantId ?? "Disconnected"))" : participant.participantId
labelPublishState.text = participant.publishState.text
labelSubscribeState.text = participant.subscribeState.text
```
发布和订阅枚举的文本属性来自本地扩展：
```
extension IVSParticipantPublishState {
var text: String {
switch self {
case .notPublished: return "Not Published"
case .attemptingPublish: return "Attempting to Publish"
case .published: return "Published"
@unknown default: fatalError()
}
}
}
extension IVSParticipantSubscribeState {
var text: String {
switch self {
case .notSubscribed: return "Not Subscribed"
case .attemptingSubscribe: return "Attempting to Subscribe"
case .subscribed: return "Subscribed"
@unknown default: fatalError()
}
}
}
```
接下来，更新音频和视频的静音状态。要进入静音状态，需要找到来自 `streams` 数组的 `IVSImageDevice` 和 `IVSAudioDevice`。要优化性能，需要记住最后连接的设备。
```
// This belongs outside `set(participant:)`
private var registeredStreams: Set = []
private var imageDevice: IVSImageDevice? {
return registeredStreams.lazy.compactMap { $0.device as? IVSImageDevice }.first
}
private var audioDevice: IVSAudioDevice? {
return registeredStreams.lazy.compactMap { $0.device as? IVSAudioDevice }.first
}
// This belongs inside `set(participant:)`
let existingAudioStream = registeredStreams.first { $0.device is IVSAudioDevice }
let existingImageStream = registeredStreams.first { $0.device is IVSImageDevice }
registeredStreams = Set(participant.streams)
let newAudioStream = participant.streams.first { $0.device is IVSAudioDevice }
let newImageStream = participant.streams.first { $0.device is IVSImageDevice }
// `isMuted != false` covers the stream not existing, as well as being muted.
labelVideoMuted.text = "Video Muted: \(newImageStream?.isMuted != false)"
labelAudioMuted.text = "Audio Muted: \(newAudioStream?.isMuted != false)"
```
最后，要渲染 `imageDevice` 的预览并显示 `audioDevice` 的音频统计数据：
```
if existingImageStream !== newImageStream {
// The image stream has changed
updatePreview() // We’ll cover this next
}
if existingAudioStream !== newAudioStream {
(existingAudioStream?.device as? IVSAudioDevice)?.setStatsCallback(nil)
audioDevice?.setStatsCallback( { [weak self] stats in
self?.labelAudioVolume.text = String(format: "Audio Level: %.0f dB", stats.rms)
})
// When the audio stream changes, it will take some time to receive new stats. Reset the value temporarily.
self.labelAudioVolume.text = "Audio Level: -100 dB"
}
```
要创建的最后一个函数是 `updatePreview()`，它将参与者的预览添加到视图中：
```
private func updatePreview() {
// Remove any old previews from the preview container
viewPreviewContainer.subviews.forEach { $0.removeFromSuperview() }
if let imageDevice = self.imageDevice {
if let preview = try? imageDevice.previewView(with: .fit) {
viewPreviewContainer.addSubviewMatchFrame(preview)
}
}
}
```
上面的步骤在 `UIView` 上使用了帮助程序函数来让嵌入子视图变得更易于操作：
```
extension UIView {
func addSubviewMatchFrame(_ view: UIView) {
view.translatesAutoresizingMaskIntoConstraints = false
self.addSubview(view)
NSLayoutConstraint.activate([
view.topAnchor.constraint(equalTo: self.topAnchor, constant: 0),
view.bottomAnchor.constraint(equalTo: self.bottomAnchor, constant: 0),
view.leadingAnchor.constraint(equalTo: self.leadingAnchor, constant: 0),
view.trailingAnchor.constraint(equalTo: self.trailingAnchor, constant: 0),
])
}
}
```