I'm receiving data through bluetooth with a custom protocol so I need to "decode" the packet and then show the result. I'm using a function of an object (singleton) to decode the bytes received but the communication is async so it's possible that I'm decoding the bytes/packet while I'm receiving other bytes. I need a solution to process the incoming data sequentially and not call the decode function while I'm already decoding a previous packet.
Here is my actual code. I tried to use mutex without success. Using a defined coroutineScope I should work on the same coroutine. Obviously the decode process have to run in background.
// Activity
bluetoothService.incomingData.observe(this, {
viewModel.getData(it) })
// ViewModel
private var coroutineJob = Job()
private val uiScope = CoroutineScope(Dispatchers.Default + coroutineJob)
fun getData(bytes: ByteArray) {
getDataSafeCall(bytes)
}
private fun getDataSafeCall(bytes: ByteArray) {
uiScopeJob = uiScope.launch {
Packet.processMessage(bytes.toUByteArray()).collect {
updateData(it)
}
}
}
// Packet Object
object Packet {
private val mutex = Mutex()
suspend fun processMessage(buffer: UByteArray) = flow {
mutex.withLock {
decode(buffer).collect { emit(it) }
}
}
private fun decode(buffer: UByteArray) = channelFlow {
// Here I'm processing my packet and then
offer()
}
}
EDIT: This is part of the Bluetooth Service if it helps to understand more, even if there's no problem receiving data.
// BluetoothService
class BluetoothService : Service() {
private val btChannel: BtChannel by inject()
private var socket: BluetoothSocket? = null
private val receivedLive = MutableLiveData<ByteArray>()
fun incomingData(): LiveData<ByteArray> {
if (socket != null) {
socket.let {
btChannel.read(socket).onEach
{
withContext(Dispatchers.Main) {
receivedLive.value = it
}
}.launchIn(CoroutineScope(Dispatchers.IO))
}
}
return receivedLive
}
}
Related
i'm trying to create a component that stream data from remote service continuously. The component starts and stops according to spring container lifecycle. I'm not sure how to test this component as the subscription is done inside my component so i was wondering wether this is the correct way to implement this kind of component with webflux or not. Does anybody know any similar component in any framework from where i might take some ideas?
Regards
class StreamingTaskAdapter(
private val streamEventsUseCase: StreamEventsUseCase,
private val subscriptionProperties: subscriptionProperties,
) : SmartLifecycle, DisposableBean, BeanNameAware {
private lateinit var disposable: Disposable
private var running: Boolean = false
private var beanName: String = "StreamingTaskAdapter"
private val logger = KotlinLogging.logger {}
override fun start() {
logger.info { "Starting container with name $beanName" }
running = true
doStart()
}
private fun doStart() {
disposable = Mono.just(
CreateSubscriptionCommand(
subscriptionProperties.events,
subscriptionProperties.owningApplication
)
)
.flatMap(streamEventsUseCase::createSubscription)
.flatMap { subscription ->
Mono.just(subscription)
.map(::ConsumeSubscriptionCommand)
.flatMap(streamEventsUseCase::consumeSubscription)
}
.repeat()
.retryWhen(Retry.backoff(MAX_ATTEMPTS, Duration.ofSeconds(2)).jitter(0.75))
.doOnSubscribe { logger.info { "Started event streaming" } }
.doOnTerminate { logger.info { "Stopped event streaming" } }
.subscribe()
}
override fun stop() {
logger.info("Stopping container with name $beanName")
doStop()
}
override fun isRunning(): Boolean = running
private fun doStop() {
running = false
disposable.dispose()
}
override fun destroy() {
logger.info("Destroying container with name $beanName")
doStop()
}
override fun setBeanName(name: String) {
this.beanName = name
}
companion object {
const val MAX_ATTEMPTS: Long = 3
}
}
I am trying to create a polling mechanism with kotlin coroutines using sharedFlow and want to stop when there are no subscribers and active when there is at least one subscriber. My question is, is sharedFlow the right choice in this scenario or should I use channel. I tried using channelFlow but I am unaware how to close the channel (not cancel the job) outside the block body. Can someone help? Here's the snippet.
fun poll(id: String) = channelFlow {
while (!isClosedForSend) {
try {
send(repository.getDetails(id))
delay(MIN_REFRESH_TIME_MS)
} catch (throwable: Throwable) {
Timber.e("error -> ${throwable.message}")
}
invokeOnClose { Timber.e("channel flow closed.") }
}
}
You can use SharedFlow which emits values in a broadcast fashion (won't emit new value until the previous one is consumed by all the collectors).
val sharedFlow = MutableSharedFlow<String>()
val scope = CoroutineScope(Job() + Dispatchers.IO)
var producer: Job()
scope.launch {
val producer = launch() {
sharedFlow.emit(...)
}
sharedFlow.subscriptionCount
.map {count -> count > 0}
.distinctUntilChanged()
.collect { isActive -> if (isActive) stopProducing() else startProducing()
}
fun CoroutineScope.startProducing() {
producer = launch() {
sharedFlow.emit(...)
}
}
fun stopProducing() {
producer.cancel()
}
First of all, when you call channelFlow(block), there is no need to close the channel manually. The channel will be closed automatically after the execution of block is done.
I think the "produce" coroutine builder function may be what you need. But unfortunately, it's still an experimental api.
fun poll(id: String) = someScope.produce {
invokeOnClose { Timber.e("channel flow closed.") }
while (true) {
try {
send(repository.getDetails(id))
// delay(MIN_REFRESH_TIME_MS) //no need
} catch (throwable: Throwable) {
Timber.e("error -> ${throwable.message}")
}
}
}
fun main() = runBlocking {
val channel = poll("hello")
channel.receive()
channel.cancel()
}
The produce function will suspended when you don't call the returned channel's receive() method, so there is no need to delay.
UPDATE: Use broadcast for sharing values across multiple ReceiveChannel.
fun poll(id: String) = someScope.broadcast {
invokeOnClose { Timber.e("channel flow closed.") }
while (true) {
try {
send(repository.getDetails(id))
// delay(MIN_REFRESH_TIME_MS) //no need
} catch (throwable: Throwable) {
Timber.e("error -> ${throwable.message}")
}
}
}
fun main() = runBlocking {
val broadcast = poll("hello")
val channel1 = broadcast.openSubscription()
val channel2 = broadcast.openSubscription()
channel1.receive()
channel2.receive()
broadcast.cancel()
}
I'm following the google billing integration instructions and have got stuck with how to await for the billing client connection result.
Whenever I need to query sku details or purchases I need to make sure that the billing client is initialized and connected. There are querySkuDetails and queryPurchasesAsync awaitable kotlin extension functions, but startConnection is based on listeners instead. Here are code samples from the docs.
private var billingClient = BillingClient.newBuilder(activity)
.setListener(purchasesUpdatedListener)
.enablePendingPurchases()
.build()
billingClient.startConnection(object : BillingClientStateListener {
override fun onBillingSetupFinished(billingResult: BillingResult) {
if (billingResult.responseCode == BillingResponseCode.OK) {
// The BillingClient is ready. You can query purchases here.
}
}
override fun onBillingServiceDisconnected() {
// Try to restart the connection on the next request to
// Google Play by calling the startConnection() method.
}
})
suspend fun querySkuDetails() {
// prepare params
// leverage querySkuDetails Kotlin extension function
val skuDetailsResult = withContext(Dispatchers.IO) {
billingClient.querySkuDetails(params.build())
}
// Process the result.
}
How to put all this together using suspend functions?
One way to create a suspending version of startConnection is the following:
/**
* Returns immediately if this BillingClient is already connected, otherwise
* initiates the connection and suspends until this client is connected.
*/
suspend fun BillingClient.ensureReady() {
if (isReady) {
return // fast path if already connected
}
return suspendCoroutine { cont ->
startConnection(object : BillingClientStateListener {
override fun onBillingSetupFinished(billingResult: BillingResult) {
if (billingResult.responseCode == BillingResponseCode.OK) {
cont.resume(Unit)
} else {
// you could also use a custom, more precise exception
cont.resumeWithException(RuntimeException("Billing setup failed: ${billingResult.debugMessage} (code ${billingResult.responseCode})"))
}
}
override fun onBillingServiceDisconnected() {
// no need to setup reconnection logic here, call ensureReady()
// before each purchase to reconnect as necessary
}
})
}
}
This will fail if another coroutine already initiated a connection.
If you want to deal with potential concurrent calls to this method, you can use a mutex to protect the connection part:
val billingConnectionMutex = Mutex()
/**
* Returns immediately if this BillingClient is already connected, otherwise
* initiates the connection and suspends until this client is connected.
* If a connection is already in the process of being established, this
* method just suspends until the billing client is ready.
*/
suspend fun BillingClient.ensureReady() {
billingConnectionMutex.withLock {
// fast path: avoid suspension if another coroutine already connected
if (isReady) {
return
}
connectOrThrow()
}
}
private suspend fun BillingClient.connectOrThrow() = suspendCoroutine<Unit> { cont ->
startConnection(object : BillingClientStateListener {
override fun onBillingSetupFinished(billingResult: BillingResult) {
if (billingResult.responseCode == BillingClient.BillingResponseCode.OK) {
cont.resume(Unit)
} else {
cont.resumeWithException(RuntimeException("Billing setup failed: ${billingResult.debugMessage} (code ${billingResult.responseCode})"))
}
}
override fun onBillingServiceDisconnected() {
// no need to setup reconnection logic here, call ensureReady()
// before each purchase to reconnect as necessary
}
})
}
Here, the release of the mutex corresponds to the end of connectOrThrow() for whichever other coroutine was holding it, so it's released as soon as the connection succeeds. If that other connection fails, this method will attempt the connection itself, and will succeed or fail on its own so the caller will be notified in case of error.
If you prefer to deal with result codes directly in if statements, you can return results instead of throwing:
private val billingConnectionMutex = Mutex()
private val resultAlreadyConnected = BillingResult.newBuilder()
.setResponseCode(BillingClient.BillingResponseCode.OK)
.setDebugMessage("Billing client is already connected")
.build()
/**
* Returns immediately if this BillingClient is already connected, otherwise
* initiates the connection and suspends until this client is connected.
* If a connection is already in the process of being established, this
* method just suspends until the billing client is ready.
*/
suspend fun BillingClient.connect(): BillingResult = billingConnectionMutex.withLock {
if (isReady) {
// fast path: avoid suspension if already connected
resultAlreadyConnected
} else {
unsafeConnect()
}
}
private suspend fun BillingClient.unsafeConnect() = suspendCoroutine<BillingResult> { cont ->
startConnection(object : BillingClientStateListener {
override fun onBillingSetupFinished(billingResult: BillingResult) {
cont.resume(billingResult)
}
override fun onBillingServiceDisconnected() {
// no need to setup reconnection logic here, call ensureReady()
// before each purchase to reconnect as necessary
}
})
}
As discussed in BillingClient.BillingClientStateListener.onBillingSetupFinished is called multiple times, it is not possible to translate startConnection into the Coroutine world, because its callback will be called multiple times.
Thanks to Joffrey's answer and comments I have managed to construct a solution with BillingResult returned
val billingConnectionMutex = Mutex()
suspend fun BillingClient.connect(): BillingResult =
billingConnectionMutex.withLock {
suspendCoroutine { cont ->
startConnection(object : BillingClientStateListener {
override fun onBillingSetupFinished(billingResult: BillingResult) {
if (billingResult.responseCode != BillingResponseCode.OK)
Log.d(TAG, "Billing setup failed: ${billingResult.debugMessage} (code ${billingResult.responseCode})")
cont.resume(billingResult)
}
override fun onBillingServiceDisconnected() {}
})
}
}
and then call it like this within the billing flow
if(billingClient!!.connect().responseCode != BillingResponseCode.OK)
return
If billingClient is already connected it will return responseCode.OK, so no need to check for isReady.
I need to make a sync call to reauthenticate the user and get a new token, but I haven't found a way that works. The code below blocks the thread and it is never unblocked, ie. I have an infinite loop
class ApolloAuthenticator(private val authenticated: Boolean) : Authenticator {
#Throws(IOException::class)
override fun authenticate(route: Route, response: Response): Request? {
// Refresh your access_token using a synchronous api request
if (response.request().header(HEADER_KEY_APOLLO_AUTHORIZATION) != null) {
return null //if you've tried to authorize and failed, give up
}
synchronized(this) {
refreshTokenSync() // This is blocked and never unblocked
val newToken = getApolloTokenFromSharedPreference()
return response.request().newBuilder()
.header(HEADER_KEY_APOLLO_AUTHORIZATION, newToken)
.build()
}
private fun refreshTokenSync(): EmptyResult {
//Refresh token, synchronously
val repository = Injection.provideSignInRepository()
return repository
.signInGraphQL()
.toBlocking()
.first()
}
fun signInGraphQL() : Observable<EmptyResult> =
sharedPreferencesDataSource.identifier
.flatMap { result -> graphqlAuthenticationDataSource.getAuth(result) }
.flatMap { result -> sharedPreferencesDataSource.saveApolloToken(result) }
.onErrorReturn { EmptyResult() }
}
---------- Use of it
val apollAuthenticator = ApolloAuthenticator(authenticated)
val okHttpBuilder =
OkHttpClient.Builder()
.authenticator(apollAuthenticator)
I haven't found a way to make a sync call using RxJava, but I can make it by using kotlin coutorine runBlocking, which will block the thread until the request is finished:
synchronized(this) {
runBlocking {
val subscription = ApolloReauthenticator.signInGraphQl() // await until it's finished
subscription.unsubscribe()
}
}
fun signInGraphQl(): Subscription {
return repository.refreshToken()
.subscribe(
{ Observable.just(EmptyResult()) },
{ Observable.just(EmptyResult()) }
)
}
So I have a method that binds to the service.
fun bindService() {
val intent = Intent(this, BluetoothService::class.java)
bindService(intent, serviceConnection, Context.BIND_AUTO_CREATE)
}
Inside onCreate method I use this code:
bindService()
launch {
delay(500L)
service = serviceConnection.serviceBinder?.getService() as BluetoothService
}
Is there more elegant way to wait for the service to be bound than using delay()?
I wrote this just now, and haven't tried it, but hopefully something like it could work. The magic is in suspendCoroutine, which pauses the current coroutine and then gives you a continuation thingy you can use to resume it later. In our case we resume it when the onServiceConnected is called.
// helper class which holds data
class BoundService(
private val context: Context,
val name: ComponentName?,
val service: IBinder?,
val conn: ServiceConnection) {
fun unbind() {
context.unbindService(conn)
}
}
// call within a coroutine to bind service, waiting for onServiceConnected
// before the coroutine resumes
suspend fun bindServiceAndWait(context: Context, intent: Intent, flags: Int) = suspendCoroutine<BoundService> { continuation ->
val conn = object: ServiceConnection {
override fun onServiceConnected(name: ComponentName?, service: IBinder?) {
continuation.resume(BoundService(context, name, service, this))
}
override fun onServiceDisconnected(name: ComponentName?) {
// ignore, not much we can do
}
}
context.bindService(intent, conn, flags)
}
// just an example
suspend fun exampleUsage() {
val bs = bindServiceAndWait(context, intent, Context.BIND_AUTO_CREATE)
try {
// ...do something with bs.service...
} finally {
bs.unbind()
}
}