Android Datastore 动态创建与源码解析

涉及到的知识点

1、协程原理---->很好的博客介绍，一个小故事讲明白进程、线程、Kotlin 协程到底啥关系？
2、Channel知识点---->Android—kotlin-Channel超详细讲解
3、Coroutines : CompletableDeferred and structured concurrency

封装的DataStoreUtils工具—>gitHub

本篇博客目的

公司使用SharedPreferences容易导致ANR，调研能否使用DataStore替换公司目前的SharedPreferences解决ANR问题，所以需要先研究一下源码

版本引入

implementation "androidx.datastore:datastore-preferences:1.0.0"

迁移SharedPreferences数据到dataStore

既然是迁移数据，那么需要将SharedPreferences已存储的数据迁移到dataStore，所以需要先构建dataStore。
目前网上构建迁移DataStore的案例Demo如下

//迁移使用
private val Context.dataStore: DataStore<Preferences> by preferencesDataStore(
    name = "userSharePreFile",
    produceMigrations = { context ->
        listOf(
            SharedPreferencesMigration(
                context,
                 "userSharePreFile"
            )
        )
    }
)

//或 
//这种构建DataStore写法是alpha版本有的，在1.0.0版本就找不到了
var dataStore: DataStore<Preferences> = context.createDataStore(
        name = "userSharePreFile"
 )
//或
//直接构建
private val Context.dataStore: DataStore<Preferences> by preferencesDataStore(
        name = "userSharePreFile"
)

上面3种写法都是对Context进行扩展创建的DataStore，所以上面创建的方式，都有一个缺点，就是需要提前知道name才能创建，如果你之前创建SharedPreferences的方式，是通过外部传递进来name构建的话，上面直接创建DataStore方式就显然不适合你了。

翻阅旧版本(alpha版本)源码，一探究竟如何构建DataStore

//alpha版本构建方式
var dataStore: DataStore<Preferences> = context.createDataStore(
        name = "userSharePreFile"
 )

fun Context.createDataStore(
    name: String,
    corruptionHandler: ReplaceFileCorruptionHandler<Preferences>? = null,
    //①
    migrations: List<DataMigration<Preferences>> = listOf(),
    //②
    scope: CoroutineScope = CoroutineScope(Dispatchers.IO + SupervisorJob())
): DataStore<Preferences> =
    PreferenceDataStoreFactory.create(
        //③
        produceFile = {
            File(this.filesDir, "datastore/$name.preferences_pb")
        },
        corruptionHandler = corruptionHandler,
        migrations = migrations,
        scope = scope
    )

可以明显看到是使用PreferenceDataStoreFactory.create返回DataStore
① 是构建需要迁移SharedPreferences文件名称
② 指明协程是在IO运行
③ 新文件存储的位置
再看看另外一种通过 by preferencesDataStore 创建DataStore方式

private val Context.dataStore: DataStore<Preferences> by preferencesDataStore(
        name = "userSharePreFile"
)

public fun preferencesDataStore(
    name: String,
    corruptionHandler: ReplaceFileCorruptionHandler<Preferences>? = null,
    //①
    produceMigrations: (Context) -> List<DataMigration<Preferences>> = { listOf() },
    //②
    scope: CoroutineScope = CoroutineScope(Dispatchers.IO + SupervisorJob())
): ReadOnlyProperty<Context, DataStore<Preferences>> {
    return PreferenceDataStoreSingletonDelegate(name, corruptionHandler, produceMigrations, scope)
}

internal class PreferenceDataStoreSingletonDelegate internal constructor(
    private val name: String,
    private val corruptionHandler: ReplaceFileCorruptionHandler<Preferences>?,
    private val produceMigrations: (Context) -> List<DataMigration<Preferences>>,
    private val scope: CoroutineScope
) : ReadOnlyProperty<Context, DataStore<Preferences>> {

    private val lock = Any()

    @GuardedBy("lock")
    @Volatile
    private var INSTANCE: DataStore<Preferences>? = null

    override fun getValue(thisRef: Context, property: KProperty<*>): DataStore<Preferences> {
        return INSTANCE ?: synchronized(lock) {
            if (INSTANCE == null) {
                val applicationContext = thisRef.applicationContext

                INSTANCE = PreferenceDataStoreFactory.create(
                    corruptionHandler = corruptionHandler,
                    migrations = produceMigrations(applicationContext),
                    scope = scope
                ) {
                    applicationContext.preferencesDataStoreFile(name)
                }
            }
            INSTANCE!!
        }
    }
}

//文件存储位置
public fun Context.preferencesDataStoreFile(name: String): File =
    this.dataStoreFile("$name.preferences_pb")

题外话：这里有利用kotlin委托属性by关键字语法
① 需要迁移的SharedPreferences文件
② 协程运行在IO

可以看出旧版本(alpha) 与 by preferencesDataStore 2种方案，都最终通过PreferenceDataStoreFactory.create，返回DataStore，我们就继续再看看PreferenceDataStoreFactory.kt的具体实现逻辑

//PreferenceDataStoreFactory.kt
 public fun create(
        corruptionHandler: ReplaceFileCorruptionHandler<Preferences>? = null,
        //迁移的share文件集合
        migrations: List<DataMigration<Preferences>> = listOf(),
         //IO
        scope: CoroutineScope = CoroutineScope(Dispatchers.IO + SupervisorJob()),
        //dataStore文件存储的目录位置
        produceFile: () -> File 
    ): DataStore<Preferences> {
        val delegate = DataStoreFactory.create(//创建SingleProcessDataStore
            serializer = PreferencesSerializer,
            corruptionHandler = corruptionHandler,
            migrations = migrations,
            scope = scope
        ) {
            //省略代码
        } 
        //传入SingleProcessDataStore
        return PreferenceDataStore(delegate)
    }

//这里有主动的去调用updateData 方法，如果不去主动调用，就不会触发迁移的逻辑
//下文的扩展函数DataStore<Preferences>.edit会说到这里
internal class PreferenceDataStore(private val delegate: DataStore<Preferences>) :
    DataStore<Preferences> by delegate {
    override suspend fun updateData(transform: suspend (t: Preferences) -> Preferences):
        Preferences {
            return delegate.updateData {
                val transformed = transform(it)
                (transformed as MutablePreferences).freeze()
                transformed
            }
        }
}

继续看DataStoreFactory.create

//DataStoreFactory.kt
fun <T> create(
        produceFile: () -> File,
        serializer: Serializer<T>,
        corruptionHandler: ReplaceFileCorruptionHandler<T>? = null,
        migrations: List<DataMigration<T>> = listOf(),
        scope: CoroutineScope = CoroutineScope(Dispatchers.IO + SupervisorJob())
    ): DataStore<T> =
        //找到最终创建的类
        SingleProcessDataStore(
            produceFile = produceFile,
            serializer = serializer,
            corruptionHandler = corruptionHandler ?: NoOpCorruptionHandler(),
            initTasksList = listOf(DataMigrationInitializer.getInitializer(migrations)),
            scope = scope
        )

到目前为止已经知道真相了，最终是通过SingleProcessDataStore返回DataStore。

下面我们通过一张图片来小结一下，旧版本alpha版本的创建与新版本 by preferencesDataStore的调用逻辑链

好，已经知道这么多了，那么我们就开始动态构建DataStore

动态创建DataStore

 fun preferencesMigrationDataStore(sharedPreferName: String) {
    val dataStore = PreferenceDataStoreFactory.create(
      corruptionHandler =  ReplaceFileCorruptionHandler<Preferences>(
        produceNewData = { emptyPreferences() }
      ),
    //需要迁移的sharePrefer文件的名称
     migrations = listOf(SharedPreferencesMigration(mContext, sharedPreferName)),
    //IO
     scope = CoroutineScope(Dispatchers.IO + SupervisorJob())) {
    //dataStore文件名称
     mContext.preferencesDataStoreFile(sharedPreferName)
     }
  
    runBlocking {
        //必须要执行这行代码，否是不会走迁移的逻辑
         dataStore.updateData {
              it.toPreferences()
           }
      }
    }

migrations:表示你要迁移的sharedPreference文件
scope :表示写数据是在IO
执行完上述代码后，.xml就会消失，然后会在files目录下多出一个/datastore/xxx.preferences_pb文件
切勿重复对某个SharedPreferences执行文件迁移方案，否则会报错。比如你前一秒在执行迁移，后一秒又继续执行迁移

####存储参数

/**
 * @key 参数
 * @value 具体的值
 */
 private fun putInt(key:String, value: Int) {
    runBlocking {
         dataStore.edit {//①
                it[intPreferencesKey(key)] = value
          }
       }
   }
//类似的还有如下，这些都是google提供的参数
intPreferencesKey
doublePreferencesKey
stringPreferencesKey
....

看①详情，点击edit，发现他是一个扩展函数

public suspend fun DataStore<Preferences>.edit(
    transform: suspend (MutablePreferences) -> Unit
): Preferences {
    return this.updateData {//调用的是PreferenceDataStore.updateData()
        //it.toMutablePreferences() 返回类似map
        it.toMutablePreferences().apply { transform(this) }
    }
}

transform 就是调用者{}里面的内容，接下来我们看看 PreferenceDataStore 类的代码

//由前部分的代码，可以得知，delegate = SingleProcessDataStore 
internal class PreferenceDataStore(private val delegate: DataStore<Preferences>) :
    DataStore<Preferences> by delegate {
    override suspend fun updateData(transform: suspend (t: Preferences) -> Preferences):
        Preferences {
            return delegate.updateData {//调用SingleProcessDataStore.updateData 
                //返回给上一个{}也就是  it.toMutablePreferences().apply { transform(this) }
                val transformed = transform(it)
                (transformed as MutablePreferences).freeze()
                transformed //拿到用户的需要更改的内容数据
            }
        }
}

代码里调用了delegate.updateData()，所以继续看SingleProcessDataStore的updateData

SingleProcessDataStore.kt
 override suspend fun updateData(transform: suspend (t: T) -> T): T {
        val ack = CompletableDeferred<T>()
        val currentDownStreamFlowState = downstreamFlow.value
        //协程体封装进Message.Update，coroutineContext 是协程的上下文，就是我们的 runBlocking 启动的线程，我这里是main
        val updateMsg = Message.Update(transform, ack, currentDownStreamFlowState, coroutineContext)
        //对消息进行分发，他的类是 SimpleActor
        actor.offer(updateMsg)
        //这里会拿到Preferences，如何拿？后面会有一个update.ack.completeWith方法，会返回回来
        return ack.await()
    }

internal class SimpleActor<T>(
    private val scope: CoroutineScope,//Dispatchers.IO + SupervisorJob()
    onComplete: (Throwable?) -> Unit,
    onUndeliveredElement: (T, Throwable?) -> Unit,
    private val consumeMessage: suspend (T) -> Unit
) {
    private val messageQueue = Channel<T>(capacity = UNLIMITED)
    private val remainingMessages = AtomicInteger(0)
    //......  省去
    //这里就是将刚刚封装的消息体，添加进这里
    fun offer(msg: T) {
        check(
            //发送封装的消息体
            messageQueue.trySend(msg)
                .onClosed { throw it ?: ClosedSendChannelException("Channel was closed normally") }
                .isSuccess
        )
        if (remainingMessages.getAndIncrement() == 0) {
            scope.launch {
                check(remainingMessages.get() > 0)
                do {
                   // scope = Dispatchers.IO + SupervisorJob()
                    scope.ensureActive()
                    //取出封装的消息体，然后进行任务处理
                    consumeMessage(messageQueue.receive())
                } while (remainingMessages.decrementAndGet() != 0)
            }
        }
    }
}

tip:这里有利用Channel进行协程通信，Channel是可以处理并发的情况
到这里，我们可以知道，我们由runBlocking(main主线程) 协程到 Dispatchers.IO的任务分发

private val actor = SimpleActor<Message<T>>(
        scope = scope,// CoroutineScope(Dispatchers.IO + SupervisorJob())
        onComplete = {//.....省略},
        onUndeliveredElement = { msg, ex ->
          //.....省略
      ) { msg ->
        //处理分发的任务，msg 为刚刚封装的updateMsg 
        when (msg) { 
            is Message.Read -> {//读取
                handleRead(msg)
            }
            is Message.Update -> {//更新
                handleUpdate(msg)
            }
        }
    }

 private suspend fun handleUpdate(update: Message.Update<T>) {
        update.ack.completeWith(
            runCatching {
                when (val currentState = downstreamFlow.value) {
                    is Data -> {
                        //写数据到file
                        transformAndWrite(update.transform, update.callerContext)
                    }
                    is ReadException, is UnInitialized -> {
                        if (currentState === update.lastState) {           
                            //读取file文件      ①          
                            readAndInitOrPropagateAndThrowFailure()
                            //写数据到file       ②
                            transformAndWrite(update.transform, update.callerContext)
                        } else {
                            throw (currentState as ReadException).readException
                        }
                    }

                    is Final -> throw currentState.finalException // won't happen
                }
            }
        )
    }

第一次使用 downstreamFlow.value = UnInitialized 。
这里要注意一下update.ack.completeWith这个函数，他是拿到结果成功返回

这里再次展示出来，是告诉大家，在哪里会等待结果返回
 override suspend fun updateData(transform: suspend (t: T) -> T): T {
        val ack = CompletableDeferred<T>()
        val currentDownStreamFlowState = downstreamFlow.value
        val updateMsg =
            Message.Update(transform, ack, currentDownStreamFlowState, coroutineContext)
        actor.offer(updateMsg)
        return ack.await() //这里就是等待 update.ack.completeWith的结果返回，所以如果不加这行，是不会卡主线程的
    }

所以使用runBlocking是会卡主线程的，如果你还有UI刷新情况，严重的情况会导致ANR问题

不扯之前的了，我们继续继续，看① 的读取

 private suspend fun readAndInitOrPropagateAndThrowFailure() {
        try {
            readAndInit()
        } catch (throwable: Throwable) {
            downstreamFlow.value = ReadException(throwable)
            throw throwable
        }
    }

 private suspend fun readAndInit() {
        check(downstreamFlow.value == UnInitialized || downstreamFlow.value is ReadException)
        //这个是锁，协程里面专有的，详情可以看 https://www.kotlincn.net/docs/reference/coroutines/shared-mutable-state-and-concurrency.html
        val updateLock = Mutex()
        //读取dataStore文件
        var initData = readDataOrHandleCorruption()
        var initializationComplete: Boolean = false
        
        //这里就是shareprefence转dataStore
        val api = object : InitializerApi<T> {
            override suspend fun updateData(transform: suspend (t: T) -> T): T {
                return updateLock.withLock() {
                    if (initializationComplete) {
                        throw IllegalStateException(
                            "InitializerApi.updateData should not be " +
                                "called after initialization is complete."
                        )
                    }
                    //transform里面就是去迁移数据的方法
                    val newData = transform(initData)
                    //这里有做，新 旧值比较，如果不同，就去写入
                    if (newData != initData) {
                        //写文件
                        writeData(newData)
                        initData = newData
                    }
                    initData
                }
            }
        }
        //initTasks 里面装的就是需要转换的 SharedPreferences集合
        initTasks?.forEach { it(api) }
        initTasks = null
        updateLock.withLock {
            initializationComplete = true
        }
        //这里有将迁移完成后的数据，存储在flow.value里面
        downstreamFlow.value = Data(initData, initData.hashCode())
    }

//读取dataStore文件
private suspend fun readDataOrHandleCorruption(): T {
        try {
            return readData()
        } catch (ex: CorruptionException) {
            val newData: T = corruptionHandler.handleCorruption(ex)
            try {
                writeData(newData)
            } catch (writeEx: IOException) {
                ex.addSuppressed(writeEx)
                throw ex
            }
            return newData
        }
    }

 private suspend fun readData(): T {
        try {
            FileInputStream(file).use { stream ->
                return serializer.readFrom(stream)
            }
        } catch (ex: FileNotFoundException) {
            if (file.exists()) {
                throw ex
            }
            return serializer.defaultValue
        }
    }

file就是我们存储的dataStore，目录是在 “datastore/$name.preferences_pb”

看完了①，再来看看② 写入数据到file，写数据的方法是 transformAndWrite()

//....
transformAndWrite(update.transform, update.callerContext)
//...

 private suspend fun transformAndWrite(
         //来源于 Message.Update.transform封装
        transform: suspend (t: T) -> T,
        //来源于 Message.Update.callerContext封装
        callerContext: CoroutineContext
    ): T {
        val curDataAndHash = downstreamFlow.value as Data<T>
        curDataAndHash.checkHashCode()

        val curData = curDataAndHash.value
        //这里callerContext  就是我们的 runBlocking，main(主线程)
        //这里是将旧的值给回调用者，然后从调用者获取到新参数
        val newData = withContext(callerContext) { transform(curData) }

        curDataAndHash.checkHashCode()
        //这里有做数据比较
        return if (curData == newData) {
            curData
        } else {
            //写入数据
            writeData(newData)
            //保存到flow.value里面
            downstreamFlow.value = Data(newData, newData.hashCode())
            newData
        }
    }

private val SCRATCH_SUFFIX = ".tmp"
//写入数据
internal suspend fun writeData(newData: T) {
        file.createParentDirectories()
        //这里创建出来的文件是"datastore/$name.preferences_pb.tmp"
        val scratchFile = File(file.absolutePath + SCRATCH_SUFFIX)
        try {
            FileOutputStream(scratchFile).use { stream ->
                serializer.writeTo(newData, UncloseableOutputStream(stream))
                stream.fd.sync()
            }
            //重新命名回去file，这里的file是我们目标的文件dataStore名称
            if (!scratchFile.renameTo(file)) {
                //重新命名失败，抛出异常
                throw IOException(
                    "Unable to rename $scratchFile." +
                        "This likely means that there are multiple instances of DataStore " +
                        "for this file. Ensure that you are only creating a single instance of " +
                        "datastore for this file."
                )
            }
        } catch (ex: IOException) {
            if (scratchFile.exists()) {
                scratchFile.delete() 
            }
            throw ex
        }
    }

到此，更新值的操作，我们已经全部走完了流程

总结

1、文件的写入是发生在IO层面
2、使用runBlocking是会卡主线程，如果此时存在需要刷新UI的情况，严重会ANR


/**
 * @key 参数
 * @value 具体的值
 */
 private fun putInt(key:String, value: Int) {
    runBlocking {
         dataStore.edit {
                it[intPreferencesKey(key)] = value
          }
       }
   }

public suspend fun DataStore<Preferences>.edit(
    transform: suspend (MutablePreferences) -> Unit
): Preferences {
    return this.updateData {
        it.toMutablePreferences().apply { transform(this) }
    }
}

//更新逻辑
 private suspend fun handleUpdate(update: Message.Update<T>) {
        update.ack.completeWith(//通知结果回调
            //.....省去
        )
    }

//transform 就是上面的{}里面的内容
 override suspend fun updateData(transform: suspend (t: T) -> T): T {
        val ack = CompletableDeferred<T>()
        val currentDownStreamFlowState = downstreamFlow.value
        val updateMsg =
            Message.Update(transform, ack, currentDownStreamFlowState, coroutineContext)
        actor.offer(updateMsg)
        return ack.await() //这里就是等待 update.ack.completeWith的结果返回，所以如果不加这行，是不会卡主线程的
        //题外话不加ack.await() 代码也会执行
    }

所以，可以考虑使用withContext(IO){读取/更新等待操作}

3、更新参数的时候，是会跟旧的值比较，如果值相同就不写入了，否则就写入到文件里面，并且更新flow.value的值

 return if (curData == newData) {
            curData
        } else {
            writeData(newData)
            downstreamFlow.value = Data(newData, newData.hashCode())
            newData
        }

4、解决并发问题，使用channel解决协程之间沟通与并发，单线程的IO更新文件与并发

5、如果已将SharedPreference迁移到DataStore，你就不要继续使用SharedPreferences了，如果继续使用SharedPreferences，会与DataStore的值不同了