Develop Reference

How to get data from a Flow datatype, to use it in a suspend function in the viewmodel

I'm a little stuck here.
I'm using Flow in an Android App with Room and MVVM paradigm.
On my repository class I have this method that returns a Flow from the DAO.
suspend fun getAllFilialCruzaByFilialId(filial_id: Int): Flow<List<FilialCruza>> {
return filialCruzaDao.getAllFilialCruzaByFiliaId(filial_id)
}
And I want to use/access the "List" in the viewModel class, because is like a business logic.
So i do this..
...
CoroutineScope(Dispatchers.IO). launch {
for (ensayo in ensayosSelected) {
i++
exportRepeticiones(ensayo.filial, null, null)
}
}
...
then, I have this...
suspend fun exportRepeticiones(filial: Filial, nroRepeticion: Int?, nroParcela: Int?)
{
filialcruzaRepository.getAllFilialCruzaByFilialId(filial.id).collect {
Log.d(TAG, "exportToAT: FilialesCruza : $it")
var fechaUcLong: Long
for (fc in it) {
Log.d(TAG, "exportToAT: FilialCruza : $fc")
}
}
}
The first problem that I have is that "collect" ends the suspend function.
When collects ends processing all items in List, the caller process is ended.
So i try with .map but nothing happens.
My question is:
I don't need to get data from Room to show on the UI, I just want to read records and process in the view model.
Is necessary to use Flow in the dao/Repository ?
If I want to keep that the Dao function return a Flow ... how can I read that data in the Viewmodel?
Thanks in advance!
Best Regards

How can I get the query result at once when I use Room?

I hope to get the total of all records with Room database at once. But, normally Room use background thread to query record asynchronously.
If I use getTotalOfVoiceAsLiveData() in Code A, it will return LiveData<Long>, you know that LiveData variable is lazy, maybe the result is null.
If I use getTotalOfVoice() in Code A, I will get error because I can't use return in viewModelScope.launch{ }.
How can I get the total of all records at once with Room database?
Code A
class HomeViewModel(val mApplication: Application, private val mDBVoiceRepository: DBVoiceRepository) : AndroidViewModel(mApplication) {
fun getTotalOfVoice():Long {
viewModelScope.launch {
return mDBVoiceRepository.getTotalOfVoice() //It will cause error
}
}
fun getTotalOfVoiceAsLiveData(): LiveData<Long>{
return mDBVoiceRepository.getTotalOfVoiceAsLiveData() //It's lazy, maybe the result is null.
}
}
class DBVoiceRepository private constructor(private val mDBVoiceDao: DBVoiceDao){
suspend fun getTotalOfVoice() = mDBVoiceDao.getTotalOfVoice()
fun getTotalOfVoiceAsLiveData() = mDBVoiceDao.getTotalOfVoiceAsLiveData()
}
#Dao
interface DBVoiceDao{
#Query("SELECT count(id) FROM voice_table")
suspend fun getTotalOfVoice(): Long
//When Room queries return LiveData, the queries are automatically run asynchronously on a background thread.
#Query("SELECT count(id) FROM voice_table")
fun getTotalOfVoiceAsLiveData(): LiveData<Long>
}
Add content
To Tobi: Thanks!
Why it is important to you to get the data directly?
I need to generate a filename based the total of the records, such as "untitled0", "untitled1", "untitled2"...
If I can get the query result at once, I can use the following code easyly.
Added again
I hope to record a voice by filename based the total of query records when I click Start button. You know the total of records will change when a reocrd is added or deleted!
Code B
fun getTotalOfVoice():Long {
//Get the query result at once
...
}
fun createdFileanme(){
return "untitled"+getTotalOfVoice().toString()
}
btnStart.setOnClickListener{
recordVoice(createdFileanme()) //I will record voice by filename
}
fun addRecord(){
...
}
fun deleteRecord(){
...
}
New added content
Thanks!
I think 'You should also move all of that into the viewmodel class, without LiveData ' is good way, you can see Image A and How can I get the value of a LivaData<String> at once in Android Studio? .
Do you agree with it?
Image A

Question: at once meaning synchronous or what ? if yes, what happens if the function to get the result has to take a longer time? like network call? well you can decide to do that on another thread.
What I think is for you to use a mutable Object and use the postValue function to dispatch the result to the observers. It should look something like below:
class HomeViewModel(val mApplication: Application, private val mDBVoiceRepository: DBVoiceRepository) : AndroidViewModel(mApplication) {
private val voices = MutableLiveData<Long>()
fun getTotalOfVoiceAsLiveData(): LiveData<Long> {
voices.postValue(mDBVoiceRepository.getTotalOfVoiceAsLiveData().value)
return voices;
}
}
Making use of it in your Fragment will look like below:
override fun onActivityCreated(savedInstanceState: Bundle?) {
super.onActivityCreated(savedInstanceState)
if (activity != null) {
val viewModel = ViewModelProvider(requireActivity())
viewModel.get(HomeViewModel::class.java).getTotalOfVoiceAsLiveData().observe(viewLifecycleOwner, Observer { voices: Long ? ->
voices // Sound of music ? be very free to use ...
})
}
}
Happy Coding.

I hope to get the result at once, but LiveData is lazy
Sorry to tell, but this is how the Room interface is designed.
You are right with the lazyness of the returned LiveData object. But this allows you to handle it on a different thread without having to manually handle different threads.
Based on your new information!
You basically have two options:
A) you could do the following:
load data from Room via LivaData
add observer that stores the current total amount
when the button is clicked you just read the local copy
In your View: (only one observer and one clickListener)
val totalVoiceCount: long
val viewModel = ViewModelProvider(requireActivity()).get(HomeViewModel::class.java)
viewModel.getTotalOfVoiceAsLiveData().observe(viewLifecycleOwner, Observer { totalOfVoice : Long ? ->
if (totalOfVoice != null)
totalVoiceCount = totalOfVoice
})
btnStart.setOnClickListener{
viewModel.recordVoice(totalVoiceCount)
}
In your ViewModel: (the logic and everything else)
fun recordVoice(totalVoiceCount : long){
val fileName = createdFileanme(totalVoiceCount)
// create your recording // depending on how you do this, it probably runs on a background thread anyways
}
fun createdFileName(totalVoiceCount : long){
return "untitled"+ String.valueOf(totalVoiceCount)
}
This works reliably because the LiveData has enough time to update the local copy of totalVoiceCount before the user has the chance to click the button.
B) Based on the answer in your parallel question you can of course outsource even more to a background thread. Then you also have the option to call the DAO query with a non-livedata return (as room returns non-livedata queries only on background threads). Is it worth to implement the threading suggestion of Ridcully? Not possible to answer without knowing what else is going on simultaneously... To me it seems like an overkill, but he is right that the more you do on background threads the better for your refresh rate..

You can return Deferred<Long> from viewModelScope.async. I recommend you to use:
val deferred = viewModelScope.async(Dispatchers.IO) {
return#async mDBVoiceRepository.getTotalOfVoice()
}
val value = deferred.await()
await() is suspend
Edit:
If you want to get a getter which will use in your activity or fragment
you need to write a suspend function like this:
suspend fun getTotalOfVoice(): Long {
return viewModelScope.async(Dispatchers.IO) {
return#async mDBVoiceRepository.getTotalOfVoice()
}.await()
}
But mvvm pattern allows you to create LiveData inside your ViewModel, which gives your fragment an observer.
In view model:
private val _totalOfVoiceLD: MutableLiveData<Long> = MutableLiveData()
val totalOfVoiceLD: LiveData<Long>
get() = _totalOfVoiceLD
fun updateTotalOfVoice() {
viewModelScope.launch(Dispatchers.IO) {
val totalOfVoice = mDBVoiceRepository.getTotalOfVoice()
_totalOfVoiceLD.postValue(totalOfVoice)
}
}
and in your fragment:
override fun onViewCreated(view: View, savedInstanceState: Bundle?) {
super.onViewCreated(view, savedInstanceState)
viewModel.totalOfVoiceLD.observe(viewLifecycleOwner, Observer { totalOfVoice ->
totalOfVoiceTextView.text = totalOfVoice.toString()
})
}

You can use coroutineContext.async to get data from DB and wait for getting it's response with data by using .await function for a async dispatch.
suspend fun getAllVoices() : Long{
val awatingResults = viewModelScope.async(Dispatchers.IO) {
mDBVoiceRepository.getTotalOfVoice()
}
val records = awatingResults.await()
return records
}
It is necessary to call a Suspend function from a coroutine and
async.await() is always called in a suspended function so,
val voiceLiveData: MutableLiveData<Long> = MutableLiveData()
fun getAllVoicesFromDB() {
viewModelScope.launch(Dispatchers.IO) {
voiceLiveData.postValue(mDBVoiceRepository.getTotalOfVoice())
}
}
Now call it where ever you want to get your voice data from database and also remember do your further work inside your voiceLiveData observer where you get your response of voices :)

Live data is designed to be lazy, when the value of the live data changes internally it emits and wherever you are observing it, the onChange function will be invoked. It is designed to fire and forget.
Because room uses background thread to run the query.
You can't expect live data to behave like sharedpreference where you store key value pair.
If you want to achieve something like that.
I would suggest you to use
Paper Db or Realm.

If you need your Room result synchronously, your code should be execute in IO thread. In case of coroutines, you can use Dispatchers.IO. Your code can be changed to this to pass the error.
class HomeViewModel(val mApplication: Application, private val mDBVoiceRepository: DBVoiceRepository) : AndroidViewModel(mApplication) {
fun getTotalOfVoice():Long {
viewModelScope.launch(Dispatchers.IO) { // here
return mDBVoiceRepository.getTotalOfVoice()
}
}
}

If you must run the queries in the main thread, then:
Allow android room to execute queries in main thread.
val dbInstance = Room
.databaseBuilder(ctx, YourDBClass::class.java, "YourDBName")
.allowMainThreadQueries()
.build()
Define the dao method as follows
#Dao
interface DBVoiceDao{
#Query("SELECT count(id) FROM voice_table")
fun getTotalOfVoice(): Long
}
Access the method in the repository
fun getTotalOfVoice():Long {
return dao.getTotalOfVoice()
}

Emit coroutine Flow from Room while backfilling via network request

I have my architecture like so:
Dao methods returning Flow<T>:
#Query("SELECT * FROM table WHERE id = :id")
fun itemById(id: Int): Flow<Item>
Repository layer returning items from DB but also backfilling from network:
(* Need help here -- this is not working as intended **)
fun items(): Flow<Item> = flow {
// Immediately emit values from DB
emitAll(itemDao.itemById(1))
// Backfill DB via network request without blocking coroutine
itemApi.makeRequest()
.also { insert(it) }
}
ViewModel layer taking the flow, applying any transformations, and converting it into a LiveData using .asLiveData():
fun observeItem(): LiveData<Item> = itemRepository.getItemFlow()
.map { // apply transformation to view model }
.asLiveData()
Fragment observing LiveData emissions and updating UI:
viewModel.item().observeNotNull(viewLifecycleOwner) {
renderUI(it)
}
The issue I'm having is at step 2. I can't seem to figure out a way to structure the logic so that I can emit the items from Flow immediately, but also perform the network fetch without waiting.
Since the fetch from network logic is in the same suspend function it'll wait for the network request to finish before emitting the results downstream. But I just want to fire that request independently since I'm not interested in waiting for a result (when it comes back, it'll update Room and I'll get the results naturally).
Any thoughts?
EDIT
Marko's solution works well for me, but I did attempt a similar approach like so:
suspend fun items(): Flow<List<Cryptocurrency>> = coroutineScope {
launch {
itemApi.makeRequest().also { insert(it) }
}
itemDao.itemById(1)
}

It sounds like you're describing a background task that you want to launch. For that you need access to your coroutine scope, so items() should be an extension function on CoroutineScope:
fun CoroutineScope.items(): Flow<Item> {
launch {
itemApi.makeRequest().also { insert(it) }
}
return flow {
emitAll(itemDao.itemById(1))
}
}
On the other hand, if you'd like to start a remote fetch whose result will also become a part of the response, you can do it as follows:
fun items(): Flow<Item> = flow {
coroutineScope {
val lateItem = async { itemApi.makeRequest().also { insert(it) } }
emitAll(itemDao.itemById(1))
emit(lateItem.await())
}
}

Room with Kotlin-coroutines observe db changes

So, I recently started experimentation with coroutines, I switched from Rxjava2 to coroutines, I haven't got a grasp of it yet but still, I ran into a condition where I needed to observe my database change and update the UI corresponding to that.
RxJava used to provide me with Flowables, Completeable etc. using that I would be able to observe changes in Db.
abstract fun insert(data: SomeData): Long
#Query("SELECT * FROM somedata_table")
abstract fun getData(): Flowable<List<SomeData>>
So here now I used to subscribe to getData and always used to observe changes
Now Enter coroutines, I am using a suspended function with a deferred result to return my responses
#Insert(onConflict = OnConflictStrategy.IGNORE)
abstract fun insert(data: SomeData): Long
#Query("SELECT * FROM somedata_table")
abstract fun getData(): List<SomeData>
suspend fun getAllSomeData():Deferred<List<SomeData>>{
return GlobalScope.async (context= coroutineContext){
database.myDao().getData()
}
}
Now I have no way to listen for updates, Channels in coroutines might be the right answer? but I am not sure how to use it with Room.

Use Room 2.2.0 Flows and kotlin coroutines. It's contentious but I dislike LiveData as it gives you results on the UI thread. If you have to do any data parsing you'll have to push everything back to another IO thread. It's also cleaner than using channels directly as you have to do extra openSubscription().consumeEach { .. } calls every time you want to listen to events.
Flow approach Requires the following versions:
// this version uses coroutines and flows in their non-experimental version
org.jetbrains.kotlinx:kotlinx-coroutines-android:1.3.2
androidx.room:room-runtime:2.2.0
androidx.room:room-compiler:2.2.0
Dao:
#Dao
interface MyDao {
#Query("SELECT * FROM somedata_table")
fun getData(): Flow<List<SomeData>>
}
class to do observation:
launch {
dao.getData().collect { data ->
//handle data here
}
}
if your calling class is not itself a CoroutineScope you'd have to call launch with the context of something that is. That can be GlobalScope or some other class you create. Here I'm using lifecycleScope assuming we're in an Activity class.
lifecycleScope.launch {
dao.getData().collect { data ->
//handle data here
}
}
the collect lambda will receive every udpate to the table much like an Rx onNext call.

Currently, there are two different ways of doing that. The first is to use a liveData builder function. To make this work, you need to update lifecycle to androidx.lifecycle:*:2.2.0-alpha01 or any newer version. The LiveData builder function will be used to call getData() asynchronously, and then use emit() to emit the result. Using this method, you will modify your Room getData() function to a suspend function and make the return type wrapped as a LiveData, replacing the Flowable used before.
#Query("SELECT * FROM somedata_table")
abstract suspend fun getData(): LiveData<List<SomeData>>
In your viewmodel you create a liveData which references your Room database
val someData: LiveData<SomeData> = liveData {
val data = database.myDao().getData()
emit(data)
}
The second approach is to get data from our DB as Flow. To use this, you need to update Room to androidx.room:room-*:2.2.0-alpha02 (currently the latest) or a newer version. This update enables #Query DAO methods to be of return type Flow The returned Flow will re-emit a new set of values if the observing tables in the query are invalidated. Declaring a DAO function with a Channel return type is an error
#Query("SELECT * FROM somedata_table")
abstract fun getData(): Flow<List<SomeData>?>
The return type is a flow of a nullable list. The list is nullable because Room will return null when the query has no data fetched.
To fetch data from the flow we will use the terminal operator collect{ } in our Presenter/ViewModel. It is preferable to do this in the ViewModel since it comes with a ViewModelScope. The solution given below assumes we are doing this in a ViewModel where we have a provided viewModelScope.
fun loadData(){
viewModelScope.launch {
database.myDao()
.getData()
.distinctUntilChanged().
.collect{
it?.let{ /** Update your obsevable data here **/
}
}

Gradle dependencies:
dependencies {
compile group: 'org.jetbrains.kotlinx', name: 'kotlinx-coroutines-reactive', version: '1.1.1'
}
Room Dao
#Dao
interface HistoryDao : BaseDao<HistoryEntity> {
#Query("select * from History order by time desc")
fun observe(): Flowable<List<HistoryEntity>>
...
}
Interactor (browserHistoryInteractor below) (layer between dao and Fragment/Presenter)
// To get channel of List<HistoryEntity>:
import kotlinx.coroutines.reactive.openSubscription
fun observe() = historyDao.observe().openSubscription() // convert list to Coroutines channel
Presenter/Fragment/Activity (end point (in my case it is lifecycle-aware presenter))
import kotlinx.coroutines.Job
import kotlinx.coroutines.launch
private val compositeJob = Job() // somewhat equivalent "compositeDisposable" in rx
override fun onCreate() {
super.onCreate()
launch(compositeJob) { // start coroutine
val channel = browserHistoryInteractor.observe()
for (items in channel) { // waits for next list of items (suspended)
showInView { view?.setItems(items) }
}
}
}
override fun onDestroy() {
compositeJob.cancel() // as in rx you need to cancel all jobs
super.onDestroy()
}
https://www.youtube.com/watch?v=lh2Vqt4DpHU&list=PLdb5m83JnoaBqMWF-qqhZY_01SNEhG5Qs&index=5 at 29:25

Room with Flowable: initialize database if it's empty

I have following #Dao, that provides Flowable<User> stream:
#Dao
interface UsersDao {
#Query("SELECT * FROM users")
fun loadUsers(): Flowable<List<User>>
}
I want the subscriber of the stream to receive updates of the database as soon as some change happens there. Subscribing to Room's Flowable I will get that feature out of the box.
What I want is following: if database is empty I want to perform a web request and save users into database. The subscriber will automatically receive new updates that had just happened.
Now I want the client of the repository not to be aware all of the initialization logics: all he does - he performs usersRepository.loadUsers(). And all of these magic should take place inside the repository class:
class UsersRepository #Inject constructor(
private val api: Api,
private val db: UsersDao
) {
fun loadUsers(): Flowable<List<User>> {
...
}
}
Of course I can use following approach:
fun loadUsers(): Flowable<List<User>> {
return db.loadTables()
.doOnSubscribe {
if (db.getCount() == 0) {
val list = api.getTables().blockingGet()
db.insert(list)
}
}
}
But I would like to construct the stream without using side-effects (doOn... operators). I've tried composing() but that didn't help much. Been stuck on this for a while.

You could apply some conditional flatMaps:
#Dao
interface UsersDao {
#Query("SELECT * FROM users")
fun loadUsers(): Flowable<List<User>>
#Query("SELECT COUNT(1) FROM users")
fun userCount() : Flowable<List<Integer>>
#Insert // I don't know Room btw.
fun insertUsers(List<User> users) : Flowable<Object>
}
interface RemoteUsers {
fun getUsers() : Flowable<List<User>>
}
fun getUsers() : Flowable<List<User>> {
return
db.userCount()
.take(1)
.flatMap({ counts ->
if (counts.isEmpty() || counts.get(0) == 0) {
return remote.getUsers()
.flatMap({ users -> db.insertUsers(users) })
.ignoreElements()
.andThen(db.loadUsers())
}
return db.loadUsers()
})
}
Disclaimer: I don't know Room so please adapt the example above as the features of it allow.

Assuming your insert() call is async and also handles updates, you could do something like this:
fun loadUsers(): Flowable<List<User>> = userDao.getAllUsers().switchIfEmpty { api.getAllUsers().doOnNext { userDao.insert(it) } }
You could also use some:
fun loadUsers(): Flowable<List<User>> = userDao.getAllUsers().flatMap { it-> if (it.isEmpty()) api.getAllUsers().doOnNext { userDao.insert(it) } else Flowable.just(it)}
Advice:
You should consider the case when the data is stale, therefore you need to go another way around, do a network request and database call at the same time. Whichever observable finish first, take the result and display it. Updating database should be right after network call is done.