A SwiftUI Grid and GridRow Tutorial

The previous chapter introduced LazyHGrid, LazyVGrid, and GridItem views and explored how they can be used to create scrollable multicolumn layouts. While these views can handle large numbers of rows, they lack flexibility, particularly in grid cell arrangement and positioning.

In this chapter, we will work with two grid layout views (Grid and GridRow) that were introduced in iOS 16. While lacking support for large grid layouts, these two views provide several features that are not available when using the lazy grid views including column spanning cells, empty cells, and a range of alignment and spacing options.

Grid and GridRow Views

A grid layout is defined using the Grid view with each row represented by a GridRow child, and direct child views of a GridRow instance represent the column cells in that row.

The syntax for declaring a grid using Grid and GridRow is as follows:

Grid {
    
    GridRow {
        // Cell views here
    }
    
    GridRow {
        // Cell views here
    }
.
.
}

Creating the GridRowDemo Project

Launch Xcode and select the option to create a new Multiplatform App project named GridRowDemo. Once the project is ready, edit the ContentView.swift file to add a custom view to be used as the content for the grid cells in later examples:

struct CellContent: View {

    var index: Int
    var color: Color

    var body: some View {
        Text("\(index)")
            .frame(minWidth: 50, maxWidth: .infinity, minHeight: 100)
            .background(color)
            .cornerRadius(8)
            .font(.system(.largeTitle))
    }
}

A Simple Grid Layout

As a first step, we will create a simple grid 5 x 3 grid by modifying the body of the ContentView structure in the ContentView.swift file as follows:

struct ContentView: View {
    var body: some View {

        Grid {
            GridRow {
                ForEach(1...5, id: \.self) { index in
                    CellContent(index: index, color: .red)
                }
            }
            
            GridRow {
                ForEach(6...10, id: \.self) { index in
                    CellContent(index: index, color: .blue)
                }
            }
            
            GridRow {
                ForEach(11...15, id: \.self) { index in
                    CellContent(index: index, color: .green)
                }
            }
        }
        .padding()
    }
}

The above example consists of a Grid view parent containing three GridRow children. Each GridRow contains a ForEach loop that generates three CellContent views. After making these changes, the layout should appear within the Preview panel, in turn, as shown in Figure 1-1:

Figure 1-1

Non-GridRow Children

So far in this chapter, we have implied that the direct children of a Grid view must be GridRows. While this is the most common use of the Grid view, it is also possible to include children outside the scope of a GridRow. Grid children not contained within a GridRow will expand to occupy an entire row within the grid layout.

The following changes, for example, add a fourth row to the grid containing a single CellContent view that fills the row:

struct ContentView: View {
    var body: some View {
        Grid {
            GridRow {
                ForEach(1...5, id: \.self) { index in
                    CellContent(index: index, color: .red)
                }
            }
            
            GridRow {
                ForEach(6...10, id: \.self) { index in
                    CellContent(index: index, color: .blue)
                }
            }
            
            GridRow {
                ForEach(11...15, id: \.self) { index in
                    CellContent(index: index, color: .green)
                }
            }
            
            CellContent(index: 16, color: .blue)
        }
        .padding()
    }
}

Within the Preview panel, the grid should appear as shown in Figure 1-2 below:

Figure 1-2

Automatic Empty Grid Cells

When creating grids, we generally assume that each row must contain the same number of columns. This is not, however, a requirement when using the Grid and GridRow views. When the Grid view is required to display rows containing different cell counts, it will automatically add empty cells to shorter rows so that they match the longest row. To experience this in our example, change the ForEach loop ranges as follows:

.
.
GridRow {
    ForEach(1...5, id: \.self) { index in
        CellContent(index: index, color: .red)
    }
}

GridRow {
    ForEach(6...8, id: \.self) { index in
        CellContent(index: index, color: .blue)
    }
}

GridRow {
    ForEach(11...12, id: \.self) { index in
        CellContent(index: index, color: .green)
    }
}
.
.

When the grid is rendered, it will place empty cells within the rows containing fewer cells as shown in Figure 1-3:

Figure 1-3

Adding Empty Cells

In addition to allowing GridRow to add empty cells, you can also insert empty cells into fixed positions in a grid layout. Empty cells are represented by a Color view configured with the “clear” color value. Applying the .gridCellUnsizedAxes() modifier to the Color view ensures that the empty cell matches the default height and width of the occupied cells. Modify the first grid row in our example so that even-numbered columns contain empty cells:

GridRow {
    ForEach(1...5, id: \.self) { index in
        if (index % 2 == 1) {
            CellContent(index: index, color: .red)
        } else {
            Color.clear
                .gridCellUnsizedAxes([.horizontal, .vertical])
        }
    }
}

Refer to the Live Preview to verify that the empty cells appear in the first row of the grid as illustrated in Figure 1-4:

Figure 1-4

Column Spanning

A key feature of Grid and GridRow is the ability for a single cell to span a specified number of columns. We can achieve this by applying the .gridCellColumns() modifier to individual content cell views within GridRow declarations. Add another row to the grid containing two cells configured to span two and three columns respectively:

.
.
CellContent(index: 16, color: .blue)

GridRow {
    CellContent(index: 17, color: .orange)
        .gridCellColumns(2)
    CellContent(index: 18, color: .indigo)
        .gridCellColumns(3)
}
.
.

The layout will now appear as shown below:

Figure 1-5

Grid Alignment and Spacing

Spacing between rows and columns can be applied using the Grid view’s verticalSpacing and horizontalSpacing parameters, for example:

Grid(horizontalSpacing: 30, verticalSpacing: 0) {
    GridRow {
        ForEach(1...5, id: \.self) { index in
.
.

The above changes increase the spacing between columns while removing the spacing between rows so that the grid appears as shown in the figure below:

Figure 1-6

We designed CellContent view used throughout this chapter to fill the available space within a grid cell. As this makes it impossible to see changes in alignment, we need to add cells containing content that will demonstrate alignment settings. Begin by inserting two new rows at the top of the grid as outlined below. Also, remove the code that placed empty cells in the row containing cells 1 through 5 so that all cells are displayed:

struct ContentView: View {
    var body: some View {
        Grid {

            GridRow {
                CellContent(index: 0, color: .orange)
                Image(systemName: "record.circle.fill")
                Image(systemName: "record.circle.fill")
                Image(systemName: "record.circle.fill")
                CellContent(index: 0, color: .yellow)
                
            }
            .font(.largeTitle)
            
            GridRow {
                CellContent(index: 0, color: .orange)
                Image(systemName: "record.circle.fill")
                Image(systemName: "record.circle.fill")
                Image(systemName: "record.circle.fill")
                CellContent(index: 0, color: .yellow)
                
            }
            .font(.largeTitle)

            GridRow {
                ForEach(1...5, id: \.self) { index in
                        CellContent(index: index, color: .red)
                }
            }
.
.

After making these changes, refer to the preview and verify that the top three rows of the grid match that shown in Figure 1-7:

Figure 1-7

We can see from the positioning of the circle symbols that the Grid and GridRow views default to centering content within grid cells. The default alignment for all cells within a grid can be changed by assigning one of the following values to the alignment parameter of the Grid view:

  • .trailing
  • .leading • .top
  • .bottom
  • .topLeading
  • .topTrailing
  • .bottomLeading
  • .bottomTrailing
  • .center

Cell content can also be aligned with the baselines of text contained in adjoining cells using the following alignment values:

  • .centerFirstTextBaseline
  • .centerLastTextBaseline
  • .leadingFirstTextBaseline
  • .leadingLastTextBaseline
  • .trailingFirstTextBaseline
  • .trailingLastTextBaseline

Modify the Grid declaration in the example code so that all content is aligned at the leading top of the containing cell:

struct ContentView: View {
    var body: some View {
        Grid(alignment: .topLeading) {
.
.

Review the preview panel and confirm that the positioning of the circle symbols matches the layout shown in Figure 1-8:

Figure 1-8

It is also possible to override the default vertical alignment setting on individual rows using the alignment property of the GridRow view. The following code modifications, for example, change the second row of symbols to use bottom alignment:

struct ContentView: View {
    var body: some View {
        Grid(alignment: .topLeading) {
                
        GridRow(alignment: .bottom) {
            CellContent(index: 0, color: .orange)
            Image(systemName: "record.circle.fill")
            Image(systemName: "record.circle.fill")
            Image(systemName: "record.circle.fill")
            CellContent(index: 0, color: .yellow)
            
        }
        .font(.largeTitle)
.
.

The circles in the first row are now positioned along the bottom of the row while the second row continues to adopt the default alignment specified by the parent grid view:

Figure 1-9

Note that GridRow alignment will only adjust the vertical positioning of cell content. As illustrated above, the first row of circles has continued to use the leading alignment applied to the parent Grid view.

Horizontal content alignment for the cells in individual columns can be changed by applying the .gridColumnAlignment() modifier to any cell within the corresponding column. The following code change, for example, applies trailing alignment to the second grid column:

struct ContentView: View {
    var body: some View {
        Grid(alignment: .topLeading) {
                
        GridRow(alignment: .bottom) {
            CellContent(index: 0, color: .orange)
            Image(systemName: "record.circle.fill")
                .gridColumnAlignment(.trailing)
            Image(systemName: "record.circle.fill")
            Image(systemName: "record.circle.fill")
            CellContent(index: 0, color: .yellow)
            
        }
        .font(.largeTitle)
.
.

When previewed, the first grid rows will appear as illustrated in Figure 1-10:

Figure 1-10

Finally, you can override content alignment in an individual using the .gridCellAnchor() modifier as follows:

Grid(alignment: .topLeading) {
        
GridRow(alignment: .bottom) {
    CellContent(index: 0, color: .orange)
    Image(systemName: "record.circle.fill")
        .gridColumnAlignment(.trailing)
    Image(systemName: "record.circle.fill")
        .gridCellAnchor(.center)
    Image(systemName: "record.circle.fill")
        .gridCellAnchor(.top)
    CellContent(index: 0, color: .yellow)  
}
.font(.largeTitle)

Once the preview updates to reflect the above changes, the circle symbol rows should appear as shown below:

Figure 1-11

Summary

The Grid and GridRow views combine to provide highly flexible grid layout options when working with SwiftUI. While these views are unsuitable for displaying scrolling grids containing a large number of views, they have several advantages over the LazyVGrid and LazyHGrid views covered in the previous chapter. Particular strengths include the ability for a single cell to span multiple columns, support for empty cells, automatic addition of empty cells to maintain matching column counts, and the ability to adjust content alignment at the grid, row, and individual cell levels.

 

A Jetpack Compose In-App Purchasing Tutorial

In the previous chapter, we explored how to integrate in-app purchasing into an Android project and also looked at some code samples that can be used when working on your own projects. This chapter will put this theory into practice by creating an example project that demonstrates how to add a consumable in-app product to an Android app using Jetpack Compose. The tutorial will also show how in-app products are added and managed within the Google Play Console and explain how to enable test payments so that purchases can be made during testing without having to spend real money.

About the In-App Purchasing Example Project

The simple concept behind this project is an app in which an in-app product must be purchased before a button can be clicked. This in-app product is consumed each time the button is clicked, requiring the user to re-purchase the product each time they want to be able to click the button. On initialization, the app will connect to the app store, obtain details of the product, and display the product name. Once the app has established that the product is available, a purchase button will be enabled which, when clicked, will step through the purchase process. On completion of the purchase, a second button will be enabled so that the user can click on it and consume the purchase.

Creating the InAppPurchase Project

The first step in this exercise is to create a new project. Begin by launching Android Studio and selecting the New Project option from the welcome screen. In the new project dialog, choose the Empty Compose Activity template before clicking on the Next button.

Enter InAppPurchase into the Name field and specify a package name that will uniquely identify your app within the Google Play ecosystem (for example com.<your company>.InAppPurchase). Before clicking on the Finish button, change the Minimum API level setting to API 26: Android 8.0 (Oreo).

Within the MainActivity.kt file, delete the Greeting function and add a new empty composable named MainScreen:

@Composable
fun MainScreen() {
 
}

Next, edit the onCreateActivity() method function to call MainScreen instead of Greeting. Since this project will be using features that are not supported by the Preview panel, also delete the DefaultPreview composable from the file. To test the project we will be running it on a device or emulator session.

Adding Libraries to the Project

Before we start writing code, some libraries need to be added to the project build configuration, including the standard Android billing client libraries. Later in the project, we will also need to use the ImmutableList class which is part of Google’s Guava Core Java libraries. Add these libraries now by modifying the Gradle Scripts -> build.gradle (Module: InAppPurchase.app) file with the following changes:

.
.
dependencies {
.
.
    implementation 'com.android.billingclient:billing:5.0.0'
    implementation 'com.android.billingclient:billing-ktx:5.0.0'
    implementation 'com.google.guava:guava:24.1-jre'
    implementation 'com.google.guava:guava:27.0.1-android'
.
.

Click on the Sync Now link at the top of the editor panel to commit these changes.

Adding the App to the Google Play Store

Using the steps outlined in the chapter entitled Creating, Testing, and Uploading an Android App Bundle, sign into the Play Console, create a new app, and set up a new internal testing track including the email addresses of designated testers. Return to Android Studio and generate a signed release app bundle for the project. Once the bundle file has been generated, upload it to the internal testing track and roll it out for testing.

Now that the app has a presence in the Google Play Store, we are ready to create an in-app product for the project.

Creating an In-App Product

With the app selected in the Play Console, scroll down the list of options in the left-hand panel until the Monetize section comes into view. Within this section, select the In-app products option listed under Products as shown in Figure 1-1:

Figure 1-1

On the In-app products page, click on the Create product button:

Figure 1-2

On the new product screen, enter the following information before saving the new product:

  • Product ID: one_button_click
  • Name: A Button Click
  • Description: This is a test in-app product that allows a button to be clicked once.
  • Default price: Set to the lowest possible price in your preferred currency.

Enabling License Testers

When testing in-app billing it is useful to be able to make test purchases without spending any money. This can be achieved by enabling license testing for the internal track testers. License testers can use a test payment card when making purchases so that they are not charged.

Within the Play Console, return to the main home screen and select the Setup -> License testing option:

Figure 1-3

Within the license testing screen, add the testers that were added for the internal testing track, change the License response setting to RESPOND_NORMALLY, and save the changes:

Figure 1-4

Now that both the app and the in-app product have been set up in the Play Console, we can start adding code to the project.

Creating a Purchase Helper Class

To establish a clean separation between the user interface and billing code, we will create a new helper class that will handle all of the purchasing tasks and use StateFlow instances to update the user interface with status changes. While it may be tempting to create this helper class as a view model, doing so will result in unstable code. The problem is that the billing client will need a reference to the main activity to process purchase transactions. This means that we will need to pass this reference to our helper class when an instance is created. As we know from previous chapters, activities are subject to being destroyed and recreated during the lifecycle of an app. Since view models are, by definition, designed to survive the destruction and recreation of activities we run the risk within our billing code of relying on a reference to an activity that no longer exists. To avoid this problem we will declare our purchase helper as a standard Kotlin data class that will be destroyed and recreated along with the activity.

Within the Project tool window, right-click on the com.<your company>.InAppPurchase entry, select the New -> Kotlin Class/File menu option and create a new class named PurchaseHelper. With the new class file created, edit it so that it reads as follows:

.
.
import android.app.Activity
import android.util.Log
import com.android.billingclient.api.*
import com.google.common.collect.ImmutableList
import kotlinx.coroutines.CoroutineScope
import kotlinx.coroutines.Dispatchers
import kotlinx.coroutines.flow.MutableStateFlow
import kotlinx.coroutines.flow.asStateFlow
import kotlinx.coroutines.launch
 
data class PurchaseHelper(val activity: Activity) {

}

These changes import a set of libraries that will be needed later in the chapter and configure the class to expect an Activity initialization parameter. Next, add variable declarations to store values related to the billing process together with the id of the product created in the Google Play Console:

.
.
data class PurchaseHelper(val activity: Activity)  {

    private val coroutineScope = CoroutineScope(Dispatchers.IO)

    private lateinit var billingClient: BillingClient
    private lateinit var productDetails: ProductDetails
    private lateinit var purchase: Purchase

    private val demoProductId = "one_button_click"
.
.

Adding the StateFlow Streams

Communication between the purchase process and the user interface will be performed using StateFlow streams. Specifically, the user interface will use these to display status information on Text components and to ensure that Buttons are appropriately enabled and disabled. Using the techniques outlined in the chapter titled Kotlin Flow with Jetpack Compose, add the following StateFlow declarations to the PurchaseHelper class:

data class PurchaseHelper(val activity: Activity)  {
.
.
    private val _productName = MutableStateFlow("Searching...")
    val productName = _productName.asStateFlow()

    private val _buyEnabled = MutableStateFlow(false)
    val buyEnabled = _buyEnabled.asStateFlow()

    private val _consumeEnabled = MutableStateFlow(false)
    val consumeEnabled = _consumeEnabled.asStateFlow()

    private val _statusText = MutableStateFlow("Initializing...")
    val statusText = _statusText.asStateFlow()
}

Initializing the Billing Client

Next, the PurchaseHelper class needs a method that can be called from the MainActivity to initialize the billing client. Remaining within the PurchaseHelper.kt file, add this new method as follows:

fun billingSetup() {
    billingClient = BillingClient.newBuilder(activity)
        .setListener(purchasesUpdatedListener)
        .enablePendingPurchases()
        .build()

    billingClient.startConnection(object : BillingClientStateListener {
        override fun onBillingSetupFinished(
            billingResult: BillingResult
        ) {
            if (billingResult.responseCode ==
                BillingClient.BillingResponseCode.OK
            ) {
                _statusText.value = "Billing Client Connected"
                queryProduct(demoProductId)
            } else {
                _statusText.value = "Billing Client Connection Failure"
            }
        }

        override fun onBillingServiceDisconnected() {
            _statusText.value = "Billing Client Connection Lost"
        }
    })
}

When this method is called, it will create a new billing client instance and attempt to connect to the Google Play Billing Library. The onBillingSetupFinished() listener will be called when the connection attempt completes and update the statusText state flow indicating the success or otherwise of the connection attempt. Finally, we have also implemented the onBillingServiceDisconnected() callback which will be called if the Google Play Billing Library connection is lost.

If the connection is successful a method named queryProduct() is called. Both this method and the purchasesUpdatedListener assigned to the billing client now need to be added.

Querying the Product

To make sure the product is available for purchase, we need to create a QueryProductDetailsParams instance configured with the product ID that was specified in the Play Console, and pass it to the queryProductDetailsAsync() method of the billing client. This will require that we also add the onProductDetailsResponse() callback method where we will check that the product exists, extract the product name, and assign it to the statusText state. Now that we have obtained the product details, we can also safely enable the purchase button via the buyEnabled flow. Within the PurchaseHelper.kt file, add the queryProduct() method so that it reads as follows:

fun queryProduct(productId: String) {
    val queryProductDetailsParams = QueryProductDetailsParams.newBuilder()
        .setProductList(
            ImmutableList.of(
                QueryProductDetailsParams.Product.newBuilder()
                    .setProductId(productId)
                    .setProductType(
                        BillingClient.ProductType.INAPP
                    )
                    .build()
            )
        )
        .build()

    billingClient.queryProductDetailsAsync(
        queryProductDetailsParams
    ) { billingResult, productDetailsList ->
        if (productDetailsList.isNotEmpty()) {
            productDetails = productDetailsList[0]
            _productName.value = "Product: " + productDetails.name
        } else {
            _statusText.value = "No Matching Products Found"
            _buyEnabled.value = false
        }
    }
}

Much of the code used here should be familiar from the previous chapter. The listener code checks that at least one product was found that matches the query criteria. The ProductDetails object is then extracted from the first matching product, stored in the productDetails variable, and the product name property assigned to the productName state flow.

Handling Purchase Updates

The results of the purchase process will be reported to the app via the PurchaseUpdatedListener that was assigned to the billing client during the initialization phase. Add this handler now as follows:

private val purchasesUpdatedListener =
    PurchasesUpdatedListener { billingResult, purchases ->
        if (billingResult.responseCode ==
            BillingClient.BillingResponseCode.OK
            && purchases != null
        ) {
            for (purchase in purchases) {
                completePurchase(purchase)
            }
        } else if (billingResult.responseCode ==
            BillingClient.BillingResponseCode.USER_CANCELED
        ) {
            _statusText.value = "Purchase Canceled"
        } else {
            _statusText.value = "Purchase Error"
        }
    }

The handler will update the status text if the user cancels the purchase or another error occurs. A successful purchase, however, results in a call to a method named completePurchase() which is passed the current Purchase object. Add this method as outlined below:

private fun completePurchase(item: Purchase) {
    purchase = item
    if (purchase.purchaseState == Purchase.PurchaseState.PURCHASED) {
        _buyEnabled.value = false
        _consumeEnabled.value = true
        _statusText.value = "Purchase Completed"
    }
}

This method stores the purchase before verifying that the product has indeed been purchased and that payment is not still pending. The consume button is enabled, the purchase button disabled, and the user is notified that the purchase was successful.

Launching the Purchase Flow

We now need to add the following method which will be called from the purchase button in the user interface to start the purchase process:

fun makePurchase() {
    val billingFlowParams = BillingFlowParams.newBuilder()
        .setProductDetailsParamsList(
            ImmutableList.of(
                BillingFlowParams.ProductDetailsParams.newBuilder()
                    .setProductDetails(productDetails)
                    .build()
            )
        )
        .build()

    billingClient.launchBillingFlow(activity, billingFlowParams)
}

Consuming the Product

With the user now able to click on the “consume” button, the next step is to make sure the product is consumed so that only one click can be performed before another button click is purchased. This requires that we now write the consumePurchase() method:

fun consumePurchase() {
    val consumeParams = ConsumeParams.newBuilder()
        .setPurchaseToken(purchase.purchaseToken)
        .build()

    coroutineScope.launch {
        val result = billingClient.consumePurchase(consumeParams)

        if (result.billingResult.responseCode ==
            BillingClient.BillingResponseCode.OK) {
            _statusText.value = "Purchase Consumed"
            _buyEnabled.value = true
            _consumeEnabled.value = false
        }
    }
}

This method creates a ConsumeParams instance and configures it with the purchase token for the current purchase (obtained from the Purchase object previously saved in the completePurchase() method). This is passed to the consumePurchase() method which is launched within a coroutine using the IO dispatcher. If the product is successfully consumed, the consume button is disabled and the status text updated.

Restoring a Previous Purchase

With the code added so far, we can purchase a product and consume it within a single session. If we were to make a purchase and then exit the app before consuming it the purchase would currently be lost when the app restarts. We can solve this problem by configuring a QueryPurchasesParams instance to search for the unconsumed In-App product and passing it to the queryPurchasesAsync() method of the billing client together with a reference to a listener that will be called with the results. Add a new method and the listener to the MainActivity.kt file as follows:

private fun reloadPurchase() {
    val queryPurchasesParams = QueryPurchasesParams.newBuilder()
        .setProductType(BillingClient.ProductType.INAPP)
        .build()

    billingClient.queryPurchasesAsync(
        queryPurchasesParams,
        purchasesListener
    )
}
 
private val purchasesListener =
    PurchasesResponseListener { billingResult, purchases ->
        if (purchases.isNotEmpty()) {
            purchase = purchases.first()
            _buyEnabled.value = false
            _consumeEnabled.value = true
            _statusText.value = "Previous Purchase Found"
        } else {
            _buyEnabled.value = true
            _consumeEnabled.value = false
        }
    }

If the list of purchases passed to the listener is not empty, the first purchase in the list is assigned to the purchase variable, and the consume button enabled (in a more complete implementation code should be added to check this is the correct product by comparing the product id and to handle the return of multiple purchases). If no purchases are found, the consume button is disabled until another purchase is made. All that remains is to call our new reloadPurchase() method during the billing setup process as follows:

fun billingSetup() {
.
.
            if (billingResult.responseCode ==
                BillingClient.BillingResponseCode.OK
            ) {
                _statusText.value = "Billing Client Connected"
                queryProduct(demoProductId)
                reloadPurchase()
            } else {
                _statusText.value = "Billing Client Connection Failure"
            }
        }
.
.
}

Completing the MainActivity

Now that the helper class is completed, changes need to be made to the MainActivity.kt file. The first step is to modify the onCreate() function to create an instance of our PurchaseHelper class and pass it to the MainScreen composable:

override fun onCreate(savedInstanceState: Bundle?) {
    super.onCreate(savedInstanceState)
    setContent {
        InAppPurchaseTheme {
            // A surface container using the 'background' color from the theme
            Surface(
                modifier = Modifier.fillMaxSize(),
                color = MaterialTheme.colorScheme.background
            ) {
                val purchaseHelper = PurchaseHelper(this)
                purchaseHelper.billingSetup()
                MainScreen(purchaseHelper)
            }
        }
    }
}

Remaining in the MainActivity.kt file, modify the MainScreen function as follows to accept the purchase handler instance and to collect from the state flow instances:

.
.
import androidx.compose.runtime.*
.
.
@Composable
fun MainScreen(purchaseHelper: PurchaseHelper) {
 
    val buyEnabled by purchaseHelper.buyEnabled.collectAsState(false)
    val consumeEnabled by purchaseHelper.consumeEnabled.collectAsState(false)
    val productName by purchaseHelper.productName.collectAsState("")
    val statusText by purchaseHelper.statusText.collectAsState("")
}

The final task before testing the app is to call the composables that make up the user interface. This will consist of a Column containing two Text components and an embedded Row containing two Buttons configured to call the makePurchase() and consumePurchase() methods of the purchase handler. The content displayed by the Text composables and the status of the buttons will be controlled by the state flow values. Make the following changes to complete the MainScreen composable:

.
.
import androidx.compose.foundation.layout.*
import androidx.compose.ui.Alignment
import androidx.compose.material.Button
import androidx.compose.ui.unit.dp
import androidx.compose.ui.unit.sp
.
.
@Composable
fun MainScreen(purchaseHelper: PurchaseHelper) {
.
.
    Column(
        Modifier.padding(20.dp),
        horizontalAlignment = Alignment.CenterHorizontally,
        verticalArrangement = Arrangement.Center
    ) {

        Text(
            productName,
            Modifier.padding(20.dp),
            fontSize = 30.sp)

        Text(statusText)

        Row(Modifier.padding(20.dp)) {

           Button(
               onClick = { purchaseHelper.makePurchase() },
               Modifier.padding(20.dp),
               enabled = buyEnabled
           ) {
               Text("Purchase")
           }

           Button(
               onClick = { purchaseHelper.consumePurchase() },
               Modifier.padding(20.dp),
               enabled = consumeEnabled
           ) {
               Text("Consume")
           }
       }
    }
}

Testing the App

Before we can test the app we need to upload this latest version to the Play Console. As we already have version 1 uploaded, we first need to increase the version number in the build.gradle (Module: InAppPurchase.app) file:

.
.
defaultConfig {
    applicationId "com.ebookfrenzy.inapppurchase"
    minSdk 26
    targetSdk 32
    versionCode 2
    versionName "2.0"
.
.

Sync the build configuration, then follow the steps in the Creating, Testing, and Uploading an Android App Bundle chapter to generate a new app bundle, upload it to the internal test track and roll it out to the testers. Using the testing track link, install the app on a device or emulator on which one of the test accounts is signed in.

After the app starts the user interface should appear as shown in Figure 1-5 below with the billing client connected, the product name displayed, and the Purchase button enabled:

Figure 1-5

Clicking the Purchase button will begin the purchase flow as shown in Figure 1-6:

Figure 1-6

Tap the buy button to complete the purchase using the test card and wait for the Consume button to be enabled.

Tap the Consume button and wait for the “Purchase Consumed” status message to appear. With the product consumed, it should now be possible to purchase it again. Make another purchase, then terminate and restart the app. The app should locate the previous unconsumed purchase and enable the consume button.

Troubleshooting

For additional information about failures, a useful trick is to access the debug message from BillingResult instances, for example:

.
.
} else if (billingResult.responseCode ==
    BillingClient.BillingResponseCode.USER_CANCELED
) {
    _statusText.value = "Purchase Canceled"
} else {
    _statusText.value = "Purchase Error"
    Log.i("InAppPurchase", billingResult.getDebugMessage())
}

After adding the debug code, make sure the device is attached to Android Studio, either via a USB cable or WiFi, and select it from within the Logcat panel. Enter InAppPurchaseTag into the Logcat search bar and check the diagnostic output, adding additional Log calls in the code if necessary.

Note that as long as you leave the app version number unchanged in the module-level build.gradle file, you should now be able to run modified versions of the app directly on the device or emulator without having to re-bundle and upload it to the console.

If the test payment card is not listed, make sure the user account on the device has been added to the license testers list. If the app is running on a physical device, try running it on an emulator. If all else fails, you can enter a valid payment method to make test purchases, and then refund yourself using the Order management screen accessible from the Play Console home page.

Summary

In this chapter, we created a project that demonstrated how to add an in-app product to an Android app. This included the creation of the product within the Google Play Console and the writing of code to initialize and connect to the billing client, querying of available products, and, finally, the purchase and consumption of the product. We also explained how to add license testers using the Play Console so that purchases can be made during testing without spending money.

A Kotlin Android In-App Purchasing Tutorial

In the previous chapter, we explored how to integrate in-app purchasing into an Android project and also looked at some code samples that can be used when working on your own projects. This chapter will put this theory into practice by creating an example project that demonstrates how to add a consumable in-app product to an Android app. The tutorial will also show how in-app products are added and managed within the Google Play Console and explain how to enable test payments so that purchases can be made during testing without having to spend real money.

About the In-App Purchasing Example Project

The simple concept behind this project is an app in which an in-app product must be purchased before a button can be clicked. This in-app product is consumed each time the button is clicked, requiring the user to re-purchase the product each time they want to be able to click the button. On initialization, the app will connect to the app store, obtain details of the product, and display the product name. Once the app has established that the product is available, a purchase button will be enabled which, when clicked, will step through the purchase process. On completion of the purchase, the main button will be enabled so that the user can click on it and consume the purchase.

Creating the InAppPurchase Project

The first step in this exercise is to create a new project. Begin by launching Android Studio and selecting the New Project option from the welcome screen. In the new project dialog, choose the Empty Activity template before clicking on the Next button.

Enter InAppPurchase into the Name field and specify a package name that will uniquely identify your app within the Google Play ecosystem (for example com.<your company>.InAppPurchase). Before clicking on the Finish button, change the Minimum API level setting to API 26: Android 8.0 (Oreo) and the Language menu to Kotlin. Once the project has been created, use the steps outlined in section 11.8 Migrating a Project to View Binding to convert the project to use view binding.

Adding Libraries to the Project

Before we start writing code, some libraries need to be added to the project build configuration, one of which is the standard Android billing client library. Later in the project, we will also need to use the ImmutableList class which is part of Google’s Guava Core Java libraries. Add these libraries now by modifying the Gradle Scripts -> build.gradle (Module: InAppPurchase.app) file with the following changes:

.
.
dependencies {
.
.
    implementation 'com.android.billingclient:billing:5.0.0'
    implementation 'com.android.billingclient:billing-ktx:5.0.0'
    implementation 'com.google.guava:guava:24.1-jre'
    implementation 'com.google.guava:guava:27.0.1-android'
.
.

Click on the Sync Now link at the top of the editor panel to commit these changes.

Designing the User Interface

The user interface will consist of the existing TextView and two Buttons. With the activity_main.xml file loaded into the editor, drag and drop two Button views onto the layout so that one is above and the other below the TextView. Select the TextView and change the id attribute to statusText.

Click on the Clear all Constraints button in the toolbar and shift-click to select all three views. Right-click on the top-most Button view and select the Center -> Horizontally in Parent menu option. Repeat this step once more, this time selecting Chains -> Create Vertical Chain. Change the text attribute of the top button so that it reads “Consume Purchase” and the id to consumeButton. Also, configure the onClick property to call a method named consumePurchase.

Select the bottom-most button and repeat the above steps, this time setting the text to “Buy Product”, the id to buyButton, and the onClick callback to makePurchase. Once completed, the layout should match that shown in Figure 84-1:

Figure 1-1

Adding the App to the Google Play Store

Using the steps outlined in the chapter entitled “Creating, Testing and Uploading an Android App Bundle”, sign into the Play Console, create a new app, and set up a new internal testing track including the email addresses of designated testers. Return to Android Studio and generate a signed release app bundle for the project. Once the bundle file has been generated, upload it to the internal testing track and roll it out for testing.

Now that the app has a presence in the Google Play Store, we are ready to create an in-app product for the project.

Creating an In-App Product

With the app selected in the Play Console, scroll down the list of options in the left-hand panel until the Monetize section comes into view. Within this section, select the In-app products option listed under Products as shown in Figure 84-2:

Figure 1-2

On the In-app products page, click on the Create product button:

Figure 1-3

On the new product screen, enter the following information before saving the new product:

  • Product ID: one_button_click
  • Name: A Button Click
  • Description: This is a test in-app product that allows a button to be clicked once.
  • Default price: Set to the lowest possible price in your preferred currency.

Enabling License Testers

When testing in-app billing it is useful to be able to make test purchases without spending any money. This can be achieved by enabling license testing for the internal track testers. License testers can use a test payment card when making purchases so that they are not charged.

Within the Play Console, return to the main home screen and select the Setup -> License testing option:

Figure 1-4

Within the license testing screen, add the testers that were added for the internal testing track, change the License response setting to RESPOND_NORMALLY, and save the changes:

Figure 1-5

Now that both the app and the in-app product have been set up in the Play Console, we can start adding code to the project.

Initializing the Billing Client

Edit the MainActivity.kt file and make the following changes to begin implementing the in-app purchase functionality:

.
.
import android.util.Log
import com.android.billingclient.api.*
.
.
class MainActivity : AppCompatActivity() {
 
    private lateinit var binding: ActivityMainBinding
    private lateinit var billingClient: BillingClient
    private lateinit var productDetails: ProductDetails
    private lateinit var purchase: Purchase
    private val demo_product = "one_button_click"
 
    val TAG = "InAppPurchaseTag"
 
    override fun onCreate(savedInstanceState: Bundle?) {
        super.onCreate(savedInstanceState)
        binding = ActivityMainBinding.inflate(layoutInflater)
        setContentView(binding.root)
        billingSetup()
    }
 
    private fun billingSetup() {
        billingClient = BillingClient.newBuilder(this)
            .setListener(purchasesUpdatedListener)
            .enablePendingPurchases()
            .build()
            
        billingClient.startConnection(object : BillingClientStateListener {
            override fun onBillingSetupFinished(
                billingResult: BillingResult
            ) {
                if (billingResult.responseCode ==
                    BillingClient.BillingResponseCode.OK
                ) {
                    Log.i(TAG, "OnBillingSetupFinish connected")
                    queryProduct(demo_product)
                } else {
                    Log.i(TAG, "OnBillingSetupFinish failed")
                }
            }
 
            override fun onBillingServiceDisconnected() {
                Log.i(TAG, "OnBillingSetupFinish connection lost")
            }
        })
    }
.
.

When the app starts, the onCreate() method will now call billingSetup() which will, in turn, create a new billing client instance and attempt to connect to the Google Play Billing Library. The onBillingSetupFinished() listener will be called when the connection attempt completes and output Logcat messages indicating the success or otherwise of the connection attempt. Finally, we have also implemented the onBillingServiceDisconnected() callback which will be called if the Google Play Billing Library connection is lost.

If the connection is successful a method named queryProduct() is called. Both this method and the purchasesUpdatedListener assigned to the billing client now need to be added.

Querying the Product

To make sure the product is available for purchase, we need to create a QueryProductDetailsParams instance configured with the product ID that was specified in the Play Console, and pass it to the queryProductDetailsAsync() method of the billing client. This will require that we also add the onProductDetailsResponse() callback method where we will check that the product exists, extract the product name, and display it on the status TextView. Now that we have obtained the product details, we can also safely enable the buy button. Within the MainActivity.kt file, add the queryProduct() method so that it reads as follows:

.
.
import com.google.common.collect.ImmutableList
.
.
private fun queryProduct(productId: String) {
    val queryProductDetailsParams = QueryProductDetailsParams.newBuilder()
        .setProductList(
            ImmutableList.of(
                Product.newBuilder()
                    .setProductId(productId)
                    .setProductType(
                        BillingClient.ProductType.INAPP
                    )
                    .build()
            )
        )
        .build()
        
    billingClient.queryProductDetailsAsync(
        queryProductDetailsParams
    ) { billingResult, productDetailsList ->
        if (!productDetailsList.isEmpty()) {
            productDetails = productDetailsList[0]
            runOnUiThread {
                binding.statusText.text = productDetails.getName()
            }
        } else {
            Log.i(TAG, "onProductDetailsResponse: No products")
        }
    }
}

Much of the code used here should be familiar from the previous chapter. The listener code checks that at least one product was found that matches the query criteria. The ProductDetails object is then extracted from the first matching product, stored in the productDetails variable, and the product name property is displayed on the TextView.

One point of note is that when we display the product name on the status TextView we do so by calling runOnUiThread(). This is necessary because the listener is not running on the main thread so cannot safely make direct changes to the user interface. The runOnUiThread() method provides a quick and convenient way to execute code on the main thread without having to use coroutines.

Launching the Purchase Flow

When the user clicks the buy button, a method named makePurchase() will be called to start the purchase process. We can now add this method as follows:

.
.
import android.view.View
.
.
fun makePurchase(view: View?) {
    val billingFlowParams = BillingFlowParams.newBuilder()
        .setProductDetailsParamsList(
            ImmutableList.of(
                ProductDetailsParams.newBuilder()
                    .setProductDetails(productDetails)
                    .build()
            )
        )
        .build()
    billingClient.launchBillingFlow(this, billingFlowParams)
}

Handling Purchase Updates

The results of the purchase process will be reported to the app via the PurchaseUpdatedListener that was assigned to the billing client during the initialization phase. Add this handler now as follows:

private val purchasesUpdatedListener =
    PurchasesUpdatedListener { billingResult, purchases ->
        if (billingResult.responseCode ==
            BillingClient.BillingResponseCode.OK
            && purchases != null
        ) {
            for (purchase in purchases) {
                completePurchase(purchase)
            }
        } else if (billingResult.responseCode ==
            BillingClient.BillingResponseCode.USER_CANCELED
        ) {
            Log.i(TAG, "onPurchasesUpdated: Purchase Canceled")
        } else {
            Log.i(TAG, "onPurchasesUpdated: Error")
        }
    }

The handler will output log messages if the user cancels the purchase or another error occurs. A successful purchase, however, results in a call to a method named completePurchase() which is passed the current Purchase object. Add this method as outlined below:

private fun completePurchase(item: Purchase) {
    purchase = item
    if (purchase.getPurchaseState() == Purchase.PurchaseState.PURCHASED)
        runOnUiThread {
            binding.consumeButton.isEnabled = true
            binding.statusText.text = "Purchase Complete"
    }
}

This method stores the purchase before verifying that the product has indeed been purchased and that payment is not still pending. The “consume” button is then enabled and the user is notified that the purchase was successful.

Consuming the Product

With the user now able to click on the “consume” button, the next step is to make sure the product is consumed so that only one click can be performed before another button click is purchased. This requires that we now write the consumePurchase() method:

.
.
import kotlinx.coroutines.*
.
.
class MainActivity : AppCompatActivity() {
 
    private val coroutineScope = CoroutineScope(Dispatchers.IO)
.
.
    fun consumePurchase(view: View?) {
 
        val consumeParams = ConsumeParams.newBuilder()
            .setPurchaseToken(purchase.purchaseToken)
            .build()
 
         coroutineScope.launch {
             val result = billingClient.consumePurchase(consumeParams)
 
             if (result.billingResult.responseCode == 
                      BillingClient.BillingResponseCode.OK) {
                 runOnUiThread() {
                     binding.consumeButton.isEnabled = false
                     binding.statusText.text = "Purchase consumed"
                 }
             }
         }
    }
.
.

This method creates a ConsumeParams instance and configures it with the purchase token for the current purchase (obtained from the Purchase object previously saved in the completePurchase() method). This is passed to the consumePurchase() method which is launched within a coroutine using the IO dispatcher. If the product is successfully consumed, code is executed in the main thread to disable the consume button and to update the status text.

Restoring a Previous Purchase

With the code added so far, we can purchase a product and consume it within a single session. If we were to make a purchase and then exit the app before consuming it the purchase would currently be lost when the app restarts. We can solve this problem by configuring a QueryPurchasesParams instance to search for the unconsumed In-App product and passing it to the queryPurchasesAsync() method of the billing client together with a reference to a listener that will be called with the results. Add a new function and the listener to the MainActivity.kt file as follows:

private fun reloadPurchase() {
 
    val queryPurchasesParams = QueryPurchasesParams.newBuilder()
        .setProductType(BillingClient.ProductType.INAPP)
        .build()
 
    billingClient.queryPurchasesAsync(
        queryPurchasesParams,
            purchasesListener
    )
}

private val purchasesListener =
    PurchasesResponseListener { billingResult, purchases ->
        if (!purchases.isEmpty()) {
            purchase = purchases.first()
            binding.consumeButton.isEnabled = true
        } else {
            binding.consumeButton.isEnabled = false
        }
    }

If the list of purchases passed to the listener is not empty, the first purchase in the list is assigned to the purchase variable, and the consume button enabled (in a more complete implementation code should be added to check this is the correct product by comparing the product id and to handle the return of multiple purchases). If no purchases are found, the consume button is disabled until another purchase is made. All that remains is to call our new reloadPurchase() method during the billing setup process as follows:

private fun billingSetup() {
.
.
            if (billingResult.responseCode ==
                BillingClient.BillingResponseCode.OK
            ) {
                Log.i(TAG, "OnBillingSetupFinish connected")
                queryProduct(demo_product)
                reloadPurchase()
            } else {
                Log.i(TAG, "OnBillingSetupFinish failed")
            }
.
.
}

Testing the App

Before we can test the app we need to upload this latest version to the Play Console. As we already have version 1 uploaded, we first need to increase the version number in the build.gradle (Module: InAppPurchase.app) file:

.
.
defaultConfig {
    applicationId "com.ebookfrenzy.inapppurchase"
    minSdk 26
    targetSdk 32
    versionCode 2
    versionName "2.0"
.
.

Sync the build configuration, then follow the steps in the “Creating, Testing and Uploading an Android App Bundle” chapter to generate a new app bundle, upload it to the internal test track and roll it out to the testers. Using the testing track link, install the app on a device or emulator on which one of the test accounts is signed in.

After the app starts it should, after a short delay, display the product name on the TextView. Clicking the buy button will begin the purchase flow as shown in Figure 84-6:

Figure 1-6

Tap the buy button to complete the purchase using the test card and wait for the Consume Purchase button to be enabled. Before tapping this button, attempt to purchase the product again and verify that it is not possible to do so because you already own the product.

Tap the Consume Purchase button and wait for the “Purchase consumed” message to appear on the TextView. With the product consumed, it should now be possible to purchase it again. Make another purchase, then terminate and restart the app. The app should locate the previous unconsumed purchase and enable the consume button.

Troubleshooting

If you encounter problems with the purchase, make sure the device is attached to Android Studio, either via a USB cable or WiFi, and select it from within the Logcat panel. Enter InAppPurchaseTag into the Logcat search bar and check the diagnostic output, adding additional Log calls in the code if necessary. For additional information about failures, a useful trick is to access the debug message from BillingResult instances, for example:

.
.
} else if (billingResult.responseCode ==
    BillingClient.BillingResponseCode.USER_CANCELED
) {
    Log.i(TAG, "onPurchasesUpdated: Purchase Canceled")
} else {
    Log.i(TAG, billingResult.getDebugMessage())
}

Note that as long as you leave the app version number unchanged in the module-level build.gradle file, you should now be able to run modified versions of the app directly on the device or emulator without having to re-bundle and upload it to the console.

If the test payment card is not listed, make sure the user account on the device has been added to the license testers list. If the app is running on a physical device, try running it on an emulator. If all else fails, you can enter a valid payment method to make test purchases, and then refund yourself using the Order management screen accessible from the Play Console home page.

Summary

In this chapter, we created a project that demonstrated how to add an in-app product to an Android app. This included the creation of the product within the Google Play Console and the writing of code to initialize and connect to the billing client, querying of available products, and, finally, the purchase and consumption of the product. We also explained how to add license testers with the Play Console so that purchases can be made during testing without spending money.

How to Add Android In-App Billing using Kotlin

In the early days of mobile applications for operating systems such as Android and iOS, the most common method for earning revenue was to charge an upfront fee to download and install the application. Another revenue opportunity was soon introduced in the form of embedding advertising within applications. Perhaps the most common and lucrative option is now to charge the user for purchasing items from within the application after it has been installed. This typically takes the form of access to a higher level in a game, acquiring virtual goods or currency, or subscribing to premium content in the digital edition of a magazine or newspaper.

Google provides support for the integration of in-app purchasing through the Google Play In-App Billing API and the Play Console. This chapter will provide an overview of in-app billing and outline how to integrate in-app billing into your Android projects. Once these topics have been explored, the next chapter will walk you through creating an example app that includes in-app purchasing features.

Preparing a Project for In-App Purchasing

Building in-app purchasing into an app will require a Google Play Developer Console account, details of which were covered previously in the “Creating, Testing and Uploading an Android App Bundle” chapter. You will also need to register a Google merchant account and configure your payment settings. These settings can be found by navigating to Setup -> Payments profile in the Play Console. Note that merchant registration is not available in all countries. For details, refer to the following page:

https://support.google.com/googleplay/android-developer/answer/9306917

The app will then need to be uploaded to the console and enabled for in-app purchasing. The console will not activate in-app purchasing support for an app, however, unless the Google Play Billing Library has been added to the module-level build.gradle file. When working with Kotlin, the Google Play Kotlin Extensions Library is also recommended:

dependencies {
.
.
    implementation 'com.android.billingclient:billing:<latest version>'
    implementation 'com.android.billingclient:billing-ktx:<latest version>'
.
.
}

Once the build file has been modified and the app bundle uploaded to the console, the next step is to add in-app products or subscriptions for the user to purchase.

Creating In-App Products and Subscriptions

Products and subscriptions are created and managed using the options listed beneath the Monetize section of the Play Console navigation panel as highlighted in Figure 83-1 below:

Figure 1-1

Each product or subscription needs an ID, title, description, and pricing information. Purchases fall into the categories of consumable (the item must be purchased each time it is required by the user such as virtual currency in a game), non-consumable (only needs to be purchased once by the user such as content access), and subscription-based. Consumable and non-consumable products are collectively referred to as managed products.

Subscriptions are useful for selling an item that needs to be renewed on a regular schedule such as access to news content or the premium features of an app. When creating a subscription, a base plan is defined specifying the price, renewal period (monthly, annually, etc.), and whether the subscription auto-renews. Users can also be provided with discount offers and given the option of pre-purchasing a subscription.

Billing Client Initialization

Communication between your app and the Google Play Billing Library is handled by a BillingClient instance. In addition, BillingClient includes a set of methods that can be called to perform both synchronous and asynchronous billing-related activities. When the billing client is initialized, it will need to be provided with a reference to a PurchasesUpdatedListener callback handler. The client will call this handler to notify your app of the results of any purchasing activity. To avoid duplicate notifications, it is recommended to have only one BillingClient instance per app.

A BillingClient instance can be created using the newBuilder() method, passing through the current activity or fragment context. The purchase update handler is then assigned to the client via the setListener() method:

private val purchasesUpdatedListener =
    PurchasesUpdatedListener { billingResult, purchases ->
        if (billingResult.responseCode ==
            BillingClient.BillingResponseCode.OK
            && purchases != null
        ) {
            for (purchase in purchases) {
                // Process the purchases
            }
        } else if (billingResult.responseCode ==
            BillingClient.BillingResponseCode.USER_CANCELED
        ) {
            // Purchase cancelled by user
        } else {
            // Handle errors here
        }
    }
 
billingClient = BillingClient.newBuilder(this)
    .setListener(purchasesUpdatedListener)
    .enablePendingPurchases()
    .build()

Connecting to the Google Play Billing Library

After the successful creation of the Billing Client, the next step is to initialize a connection to the Google Play Billing Library. To establish this connection, a call needs to be made to the startConnection() method of the billing client instance. Since the connection is performed asynchronously, a BillingClientStateListener handler needs to be implemented to receive a callback indicating whether the connection was successful. Code should also be added to override the onBillingServiceDisconnected() method. This is called if the connection to the Billing Library is lost and can be used to report the problem to the user and retry the connection.

Once the setup and connection tasks are complete, the BillingClient instance will make a call to the onBillingSetupFinished() method which can be used to check that the client is ready:

billingClient.startConnection(object : BillingClientStateListener {
    override fun onBillingSetupFinished(
        billingResult: BillingResult
    ) {
        if (billingResult.responseCode ==
            BillingClient.BillingResponseCode.OK
        ) {
            // Connection successful
        } else {
            // Connection failed 
        }
    }
 
    override fun onBillingServiceDisconnected() {
        // Connection to billing service lost
    }
})

Querying Available Products

Once the billing environment is initialized and ready to go, the next step is to request the details of the products or subscriptions that are available for purchase. This is achieved by making a call to the queryProductDetailsAsync() method of the BillingClient and passing through an appropriately configured QueryProductDetailsParams instance containing the product ID and type (ProductType.SUBS for a subscription or ProductType.INAPP for a managed product):

val queryProductDetailsParams = QueryProductDetailsParams.newBuilder()
    .setProductList(
        ImmutableList.of(
            QueryProductDetailsParams.Product.newBuilder()
                .setProductId(productId)
                .setProductType(
                    BillingClient.ProductType.INAPP
                )
                .build()
        )
    )
    .build()
 
billingClient.queryProductDetailsAsync(
    queryProductDetailsParams
) { billingResult, productDetailsList ->
    if (!productDetailsList.isEmpty()) {
        // Process list of matching products
    } else {
        // No product matches found
    }
}

The queryProductDetailsAsync() method is passed a ProductDetailsResponseListener handler (in this case in the form of a lambda code block) which, in turn, is called and passed a list of ProductDetail objects containing information about the matching products. For example, we can call methods on these objects to get information such as the product name, title, description, price, and offer details.

Starting the Purchase Process

Once a product or subscription has been queried and selected for purchase by the user, the purchase process is ready to be launched. We do this by calling the launchBillingFlow() method of the BillingClient, passing through as arguments the current activity and a BillingFlowParams instance configured with the ProductDetail object for the item being purchased.

val billingFlowParams = BillingFlowParams.newBuilder()
    .setProductDetailsParamsList(
        ImmutableList.of(
            BillingFlowParams.ProductDetailsParams.newBuilder()
                .setProductDetails(productDetails)
                .build()
        )
    )
    .build()
 
billingClient.launchBillingFlow(this, billingFlowParams)

The success or otherwise of the purchase operation will be reported via a call to the PurchasesUpdatedListener callback handler outlined earlier in the chapter.

Completing the Purchase

When purchases are successful, the PurchasesUpdatedListener handler will be passed a list containing a Purchase object for each item. You can verify that the item has been purchased by calling the getPurchaseState() method of the Purchase instance as follows:

if (purchase.getPurchaseState() == Purchase.PurchaseState.PURCHASED) {
    // Purchase completed. 
} else if (purchase.getPurchaseState() == Purchase.PurchaseState.PENDING) {
    // Payment is still pending
}

Note that your app will only support pending purchases if a call is made to the enablePendingPurchases() method during initialization. A pending purchase will remain so until the user completes the payment process.

When the purchase of a non-consumable item is complete, it will need to be acknowledged to prevent a refund from being issued to the user. This requires the purchase token for the item which is obtained via a call to the getPurchaseToken() method of the Purchase object. This token is used to create an AcknowledgePurchaseParams instance together with an AcknowledgePurchaseResponseListener handler. Managed product purchases and subscriptions are acknowledged by calling the BillingClient’s acknowledgePurchase() method as follows:

billingClient.acknowledgePurchase(acknowledgePurchaseParams, 
                             acknowledgePurchaseResponseListener);
val acknowledgePurchaseParams = AcknowledgePurchaseParams.newBuilder()
    .setPurchaseToken(purchase.purchaseToken)
    .build()
 
val acknowledgePurchaseResponseListener = AcknowledgePurchaseResponseListener {
    // Check acknowledgement result
}
 
billingClient.acknowledgePurchase(
    acknowledgePurchaseParams,
    acknowledgePurchaseResponseListener
)

For consumable purchases, you will need to notify Google Play when the item has been consumed so that it is available to be repurchased by the user. This requires a configured ConsumeParams instance containing a purchase token and a call to the billing client’s consumePurchase() method:

val consumeParams = ConsumeParams.newBuilder()
    .setPurchaseToken(purchase.purchaseToken)
    .build()
 
coroutineScope.launch {
    val result = billingClient.consumePurchase(consumeParams)
 
    if (result.billingResult.responseCode == 
                     BillingClient.BillingResponseCode.OK) {
        // Purchase successfully consumed
    }
}

Querying Previous Purchases

When working with in-app billing it is a common requirement to check whether a user has already purchased a product or subscription. A list of all the user’s previous purchases of a specific type can be generated by calling the queryPurchasesAsync() method of the BillingClient instance and implementing a PurchaseResponseListener. The following code, for example, obtains a list of all previously purchased items that have not yet been consumed:

val queryPurchasesParams = QueryPurchasesParams.newBuilder()
    .setProductType(BillingClient.ProductType.INAPP)
    .build()
 
billingClient.queryPurchasesAsync(
    queryPurchasesParams,
    purchasesListener
)
.
.
private val purchasesListener =
    PurchasesResponseListener { billingResult, purchases ->
 
        if (!purchases.isEmpty()) {
            // Access existing active purchases
        } else {
            // No 
        }
    }

To obtain a list of active subscriptions, change the ProductType value from INAPP to SUBS.

Alternatively, to obtain a list of the most recent purchases for each product, make a call to the BillingClient queryPurchaseHistoryAsync() method:

val queryPurchaseHistoryParams = QueryPurchaseHistoryParams.newBuilder()
    .setProductType(BillingClient.ProductType.INAPP)
    .build()
 
billingClient.queryPurchaseHistoryAsync(queryPurchaseHistoryParams) { billingResult, historyList ->
    // Process purchase history list
}

Summary

In-app purchases provide a way to generate revenue from within Android apps by selling virtual products and subscriptions to users. In this chapter, we have explored managed products and subscriptions and explained the difference between consumable and non-consumable products. In-app purchasing support is added to an app using the Google Play In-app Billing Library and involves creating and initializing a billing client on which methods are called to perform tasks such as making purchases, listing available products, and consuming existing purchases. The next chapter contains a tutorial demonstrating the addition of in-app purchases to an Android Studio project.

Coroutines and LaunchedEffects in Jetpack Compose

When an Android application is first started, the runtime system creates a single thread in which all application components will run by default. This thread is generally referred to as the main thread. The primary role of the main thread is to handle the user interface in terms of event handling and interaction with views in the user interface. Any additional components that are started within the application will, by default, also run on the main thread.

Any code within an application that performs a time-consuming task using the main thread will cause the entire application to appear to lock up until the task is completed. This will typically result in the operating system displaying an “Application is not responding” warning to the user. This is far from the desired behavior for any application. Fortunately, Kotlin provides a lightweight alternative in the form of Coroutines. In this chapter, we will introduce Coroutines, including terminology such as dispatchers, coroutine scope, suspend functions, coroutine builders, and structured concurrency. The chapter will also explore channel-based communication between coroutines and explain how to safely launch coroutines from within composable functions.

What are coroutines?

Coroutines are blocks of code that execute asynchronously without blocking the thread from which they are launched. Coroutines can be implemented without having to worry about building complex multi-tasking implementations or directly managing multiple threads. Because of the way they are implemented, coroutines are much more efficient and less resource-intensive than using traditional multi-threading options. Coroutines also make for code that is much easier to write, understand and maintain since it allows code to be written sequentially without having to write callbacks to handle thread-related events and results.

Although a relatively recent addition to Kotlin, there is nothing new or innovative about coroutines. Coroutines in one form or another have existed in programming languages since the 1960s and are based on a model known as Communicating Sequential Processes (CSP). In fact, Kotlin still uses multi-threading behind the scenes, though it does so highly efficiently.

Threads vs. coroutines

A problem with threads is that they are a finite resource and expensive in terms of CPU capabilities and system overhead. In the background, a lot of work is involved in creating, scheduling, and destroying a thread. Although modern CPUs can run large numbers of threads, the actual number of threads that can be run in parallel at any one time is limited by the number of CPU cores (though newer CPUs have 8 or more cores, most Android devices contain CPUs with 4 cores). When more threads are required than there are CPU cores, the system has to perform thread scheduling to decide how the execution of these threads is to be shared between the available cores.

To avoid these overheads, instead of starting a new thread for each coroutine and then destroying it when the coroutine exits, Kotlin maintains a pool of active threads and manages how coroutines are assigned to those threads. When an active coroutine is suspended it is saved by the Kotlin runtime and another coroutine resumed to take its place. When the coroutine is resumed, it is simply restored to an existing unoccupied thread within the pool to continue executing until it either completes or is suspended. Using this approach, a limited number of threads are used efficiently to execute asynchronous tasks with the potential to perform large numbers of concurrent tasks without the inherent performance degeneration that would occur using standard multithreading.

Coroutine Scope

All coroutines must run within a specific scope which allows them to be managed as groups instead of as individual coroutines. This is particularly important when canceling and cleaning up coroutines and ensuring that coroutines do not “leak” (in other words continue running in the background when they are no longer needed by the app). By assigning coroutines to a scope they can, for example, all be canceled in bulk when they are no longer needed.

Kotlin and Android provide some built-in scopes as well as the option to create custom scopes using the CoroutineScope class. The built-in scopes can be summarized as follows:

  • GlobalScope – GlobalScope is used to launch top-level coroutines which are tied to the entire lifecycle of the application. Since this has the potential for coroutines in this scope to continue running when not needed (for example when an Activity exits) use of this scope is not recommended for use in Android applications. Coroutines running in GlobalScope are considered to be using unstructured concurrency.
  • ViewModelScope – Provided specifically for use in ViewModel instances when using the Jetpack architecture ViewModel component. Coroutines launched in this scope from within a ViewModel instance are automatically canceled by the Kotlin runtime system when the corresponding ViewModel instance is destroyed.
  • LifecycleScope – Every lifecycle owner has associated with it a LifecycleScope. This scope is canceled when the corresponding lifecycle owner is destroyed making it particularly useful for launching coroutines from within composables and activities.

For most requirements, the best way to access a coroutine scope from within a composable is to make a call to the rememberCoroutineScope() function as follows:

val coroutineScope = rememberCoroutineScope()

The coroutineScope declares the dispatcher that will be used to run coroutines (though this can be overridden) and must be referenced each time a coroutine is started if it is to be included within the scope. All of the running coroutines in a scope can be canceled via a call to the cancel() method of the scope instance:

coroutineScope.cancel()

Suspend functions

A suspend function is a special type of Kotlin function that contains the code of a coroutine. It is declared using the Kotlin suspend keyword which indicates to Kotlin that the function can be paused and resumed later, allowing long-running computations to execute without blocking the main thread. The following is an example suspend function:

suspend fun mySlowTask() {
    // Perform long-running task here    
}

Coroutine dispatchers

Kotlin maintains threads for different types of asynchronous activity and, when launching a coroutine, you have the option to specify a specific dispatcher from the following options:

  • Dispatchers.Main – Runs the coroutine on the main thread and is suitable for coroutines that need to make changes to the UI and as a general-purpose option for performing lightweight tasks.
  • Dispatchers.IO – Recommended for coroutines that perform network, disk, or database operations.
  • Dispatchers.Default – Intended for CPU-intensive tasks such as sorting data or performing complex calculations.

The dispatcher is responsible for assigning coroutines to appropriate threads and suspending and resuming the coroutine during its lifecycle. The following code, for example, launches a coroutine using the IO dispatcher:

.
.
coroutineScope.launch(Dispatchers.IO) {
    performSlowTask()
}
.
.

In addition to the predefined dispatchers, it is also possible to create dispatchers for your own custom thread pools.

Coroutine builders

The coroutine builders bring together all of the components covered so far and launch the coroutines so that they start executing. For this purpose, Kotlin provides the following six builders:

  • launch – Starts a coroutine without blocking the current thread and does not return a result to the caller. Use this builder when calling a suspend function from within a traditional function, and when the results of the coroutine do not need to be handled (sometimes referred to as “fire and forget” coroutines).
  • async – Starts a coroutine and allows the caller to wait for a result using the await() function without blocking the current thread. Use async when you have multiple coroutines that need to run in parallel. The async builder can only be used from within another suspend function.
  • withContext – This allows a coroutine to be launched in a different context from that used by the parent coroutine. A coroutine running using the Main context could, for example, launch a child coroutine in the Default context using this builder. The withContext builder also provides a useful alternative to async when returning results from a coroutine.
  • coroutineScope – The coroutineScope builder is ideal for situations where a suspend function launches multiple coroutines that will run in parallel and where some action needs to take place only when all the coroutines reach completion. If those coroutines are launched using the coroutineScope builder, the calling function will not return until all child coroutines have completed. When using coroutineScope, a failure in any of the coroutines will result in the cancellation of all other coroutines.
  • supervisorScope – Similar to the coroutineScope outlined above, with the exception that a failure in one child does not result in cancellation of the other coroutines.
  • runBlocking – Starts a coroutine and blocks the current thread until the coroutine reaches completion. This is typically the opposite of what is wanted from coroutines but is useful for testing code and integrating legacy code and libraries. Otherwise to be avoided.

Jobs

Each call to a coroutine builder such as launch or async returns a Job instance which can, in turn, be used to track and manage the lifecycle of the corresponding coroutine. Subsequent builder calls from within the coroutine create new Job instances which will become children of the immediate parent Job forming a parentchild relationship tree where canceling a parent Job will recursively cancel all its children. Canceling a child does not, however, cancel the parent, though an uncaught exception within a child created using the launch builder may result in the cancellation of the parent (this is not the case for children created using the async builder which encapsulates the exception in the result returned to the parent).

The status of a coroutine can be identified by accessing the isActive, isCompleted, and isCancelled properties of the associated Job object. In addition to these properties, several methods are also available on a Job instance. A Job and all of its children may, for example, be canceled by calling the cancel() method of the Job object, while a call to the cancelChildren() method will cancel all child coroutines.

The join() method can be called to suspend the coroutine associated with the job until all of its child jobs have completed. To perform this task and cancel the Job once all child jobs have completed, simply call the cancelAndJoin() method.

This hierarchical Job structure together with coroutine scopes form the foundation of structured concurrency, the goal of which is to ensure that coroutines do not run for longer than they are required without the need to manually keep references to each coroutine.

Coroutines – suspending and resuming

To gain a better understanding of coroutine suspension, it helps to see some examples of coroutines in action. To start with, let’s assume a simple Android app containing a button that, when clicked, calls a suspend function named performSlowTask(). The code for this might read as follows:

val coroutineScope = rememberCoroutineScope()

Button(onClick = {
    coroutineScope.launch {
        performSlowTask()
    }
}) {
    Text(text = "Click Me")
}

In the above code, a coroutine scope is obtained and referenced in the call to the launch builder which, in turn, calls the performSlowTask() suspend function. Next, we can declare the performSlowTask() suspend function as follows:

suspend fun performSlowTask() {
    println("performSlowTask before")
    delay(5000) // simulates long-running task
    println("performSlowTask after")
}

As implemented, all the function does is output diagnostic messages before and after performing a 5-second delay, simulating a long-running task. While the 5-second delay is in effect, the user interface will continue to be responsive because the main thread is not being blocked. To understand why it helps to explore what is happening behind the scenes.

A click on the button launches the performSlowTask() suspend function as a coroutine. This function then calls the Kotlin delay() function passing through a time value. In fact, the built-in Kotlin delay() function is itself implemented as a suspend function so is also launched as a coroutine by the Kotlin runtime environment. The code execution has now reached what is referred to as a suspend point which will cause the performSlowTask() coroutine to be suspended while the delay coroutine is running. This frees up the thread on which performSlowTask() was running and returns control to the main thread so that the UI is unaffected.

Once the delay() function reaches completion, the suspended coroutine will be resumed and restored to a thread from the pool where it can display the log message and return.

When working with coroutines in Android Studio suspend points within the code editor are marked as shown in the figure below:

Figure 1-1

We will explore some coroutine examples when we start to look at List composables, starting with the chapter titled Jetpack Compose Lists and Grids.

Coroutine channel communication

Channels provide a simple way to implement communication between coroutines including streams of data. In the simplest form this involves the creation of a Channel instance and calling the send() method to send the data. Once sent, transmitted data can be received in another coroutine via a call to the receive() method of the same Channel instance.

The following code, for example, passes six integers from one coroutine to another:

.
.
import kotlinx.coroutines.channels.*
.
.
val channel = Channel<Int>()
 
coroutineScope.launch() {
    coroutineScope.launch(Dispatchers.Main) { performTask1() }
    coroutineScope.launch(Dispatchers.Main) { performTask2() }
}
 
suspend fun performTask1() {
    (1..6).forEach {
        channel.send(it)
    }
}

suspend fun performTask2() {
    repeat(6) {
        println("Received: ${channel.receive()}")
    }
}

When executed, the following logcat output will be generated:

Received: 1
Received: 2
Received: 3
Received: 4
Received: 5
Received: 6

Understanding side effects

So far in this chapter, we have looked at coroutines and explained how to use a coroutine scope to execute code asynchronously. In each case, the coroutine was launched from within the onClick event handler of a Button composable. The reason for this is that while it is possible to launch a coroutine in this way from within the scope of an event handler, it is not safe to do so from within the scope of the parent composable. Consider, for example, the following code:

@Composable
fun Greeting(name: String) {

    val coroutineScope = rememberCoroutineScope()

    coroutineScope.launch() {
        performSlowTask()
    }
}

An attempt to compile the above code will result in an error that reads as follows:

Calls to launch should happen inside a LaunchedEffect and not composition

It is not possible to launch coroutines in this way when working within a composable because it can cause adverse side effects. In the context of Jetpack Compose, a side effect occurs when asynchronous code makes changes to the state of a composable from a different scope without taking into consideration the lifecycle of that composable. The risk here is the potential for a coroutine to continue running after the composable exits, a particular problem if the coroutine is still executing and making state changes the next time the composable runs.

To avoid this problem, we need to launch our coroutines from within the body of either a LaunchedEffect or SideEffect composable. Unlike the above attempt to directly launch a coroutine from within the scope of a composable, these two composables are considered safe to launch coroutines because they are aware of the lifecycle of the parent composable.

When a LaunchedEffect composable containing coroutine launch code is called, the coroutine will immediately launch and begin executing the asynchronous code. As soon as the parent composable completes, the LaunchedEffect instance and coroutine are destroyed.

The syntax for declaring a LaunchedEffect containing a coroutine is as follows:

LaunchedEffect(key1, key2, ...) {
    coroutineScope.launch() {
        // async code here
    }
}

The key parameter values (of which there must be at least one) control the behavior of the coroutine through recompositions. As long as the values of any of the key parameters remain unchanged, LaunchedEffect will keep the same coroutine running through multiple recompositions of the parent composable. If a key value changes, however, LaunchedEffect will cancel the current coroutine and launch a new one.

To call our suspend function from within our composable, we would need to change the code to read as follows:

@Composable
fun Greeting(name: String) {

    val coroutineScope = rememberCoroutineScope()

    LaunchedEffect(key1 = Unit) {
        coroutineScope.launch() {
            performSlowTask()
        }
    }
}

Note that we have passed a Unit instance (the equivalent of a void value) as the key in the above example to indicate that the coroutine does not need to be recreated through recompositions.

In addition to LaunchedEffect, Jetpack Compose also includes the SideEffect composable. Unlike LaunchedEffect, a SideEffect coroutine is executed after composition of the parent completes. SideEffect also does not accept key parameters and relaunches on every recomposition of the parent composable. We will be making use of LaunchedEffect in the chapter entitled “A Jetpack Compose SharedFlow Tutorial”.

Summary

Kotlin coroutines provide a simpler and more efficient approach to performing asynchronous tasks than that offered by traditional multi-threading. Coroutines allow asynchronous tasks to be implemented in a structured way without the need to implement the callbacks associated with typical thread-based tasks. This chapter has introduced the basic concepts of coroutines including jobs, scope, builders, suspend functions, structured concurrency, and channel-based communication.

While it is possible to directly start coroutines from within an event handler such as the onClick handler of a Button, doing so within the main body of a Composable is considered unsafe and results in a syntax error. In this situation, coroutines must be launched using either the LaunchedEffect or SideEffect composable functions.

A Jetpack Compose SharedFlow Tutorial

The previous chapter introduced Kotlin flows and explored how these can be used to return multiple sequential values from within coroutine-based asynchronous code. In this tutorial, we will look at a more detailed flow implementation, this time using SharedFlow. The tutorial will also demonstrate how to ensure that flow collection responds correctly to an app switching between background and foreground modes.

About the project

The app created in this chapter will consist of a user interface containing a List composable. A shared flow located within a ViewModel will be activated as soon as the view model is created and will emit an integer value every two seconds. The Main Activity will collect the values from the flow and display them within the List. The project will then be modified to suspend the collection process while the app is placed in the background.

Creating the SharedFlowDemo project

Launch Android Studio and create a new Empty Compose Activity project named SharedFlowDemo, specifying com.example.sharedflowdemo as the package name, and selecting a minimum API level of API 26: Android 8.0 (Oreo). Within the MainActivity.kt file, delete the Greeting function and add a new empty composable named ScreenSetup which, in turn, calls a function named MainScreen:

@Composable
fun ScreenSetup() {
    MainScreen()
}
 
@Composable
fun MainScreen() {
    
}

Edit the onCreateActivity() method function to call ScreenSetup instead of Greeting and remove the Greeting call from DefaultPreview.

Next, modify the build.gradle (Module: FlowDemo.app) file to add the Compose view model and Kotlin runtime extensions libraries to the dependencies section:

dependencies {
.
.
    implementation 'androidx.lifecycle:lifecycle-viewmodel-compose:2.4.1'
    implementation 'androidx.lifecycle:lifecycle-runtime-ktx:2.4.1'
.
.
}

When prompted, click on the Sync Now button at the top of the editor panel to commit to the change.

Adding a view model to the project

For this project, the flow will once again reside in a view model class. Add this model to the project by locating and right-clicking on the app -> java -> com.example.sharedflowdemo entry in the Project tool window and selecting the New -> Kotlin Class/File menu option. In the resulting dialog, name the class DemoViewModel before tapping the keyboard Enter key. Once created, modify the file so that it reads as follows:

package com.example.flowdemo

import androidx.lifecycle.ViewModel

class DemoViewModel : ViewModel() {
}

Return to the MainActivity.kt file and make changes to access an instance of the view model:

.
.
import androidx.lifecycle.viewmodel.compose.viewModel
.
.
@Composable
fun ScreenSetup(viewModel: DemoViewModel = viewModel()) {
    MainScreen()
}

Declaring the SharedFlow

The next step is to add some code to the view model to create and start the SharedFlow instance. Begin by editing the DemoViewModel.kt file so that it reads as follows:

package com.example.sharedflowdemo
 
import androidx.lifecycle.ViewModel
import androidx.lifecycle.viewModelScope
import kotlinx.coroutines.delay
import kotlinx.coroutines.flow.MutableSharedFlow
import kotlinx.coroutines.flow.asSharedFlow
import kotlinx.coroutines.launch
 
class DemoViewModel  : ViewModel() {
 
    private val _sharedFlow = MutableSharedFlow<Int>()
    val sharedFlow = _sharedFlow.asSharedFlow()
 
    init {
        sharedFlowInit()
    }
 
    fun sharedFlowInit() {
    }
}

When the ViewModel instance is created, the initializer will call the sharedFlowInit() function. The purpose of this function is to launch a new coroutine containing a loop in which new values are emitted using a shared flow.

With the flow declared, code can now be added to the sharedFlowInit() function to launch the flow using the view model’s own scope. This will ensure that the flow ends when the view model is destroyed:

fun sharedFlowInit() {
    viewModelScope.launch {
        for (i in 1..1000) {
            delay(2000)
            _sharedFlow.emit(i)
        }
    }
}

Collecting the flow values

Before testing the app for the first time we need to add some code to perform the flow collection and display those values in a LazyColumn composable. As the values are collected from the flow, they will be added to a mutable list state instance which, in turn, will serve as the data source for the LazyColumn content. We also need to pass a reference to the shared flow down to the MainScreen composable. Edit the MainActivity.kt file and make the following changes:

.
.
import androidx.compose.runtime.*
import androidx.compose.foundation.layout.padding
import androidx.compose.foundation.lazy.*
import androidx.compose.ui.unit.dp
import kotlinx.coroutines.flow.SharedFlow
import androidx.compose.ui.platform.LocalLifecycleOwner
.
.
@Composable
fun ScreenSetup(viewModel: DemoViewModel = viewModel()) {
    MainScreen(viewModel.sharedFlow)
}
 
@Composable
fun MainScreen(sharedFlow: SharedFlow<Int>) {

    val messages = remember { mutableStateListOf<Int>()}

    LazyColumn {
        items(messages) {
            Text(
                "Collected Value = $it",
                style = MaterialTheme.typography.h3,
                modifier = Modifier.padding(5.dp)
            )
        }
    }
}
 
@Preview(showBackground = true)
@Composable
fun DefaultPreview() {
    SharedFlowDemoTheme {
        val viewModel: DemoViewModel = viewModel()
        MainScreen(viewModel.sharedFlow)
    }
}

With these changes made we are ready to collect the values emitted by the shared flow and display them. Since the flow collection will be taking place in a coroutine and outside the scope of the MainScreen composable, the launch code needs to be placed within a LaunchedEffect call (a topic covered in the chapter titled “Coroutines and Side Effects in Jetpack Compose”. Add a LaunchedEffect call to the MainScreen composable as follows to collect from the flow:

.
.
import kotlinx.coroutines.flow.collect
.
.
@Composable
fun MainScreen(sharedFlow: SharedFlow<Int>) {
 
    val messages = remember { mutableStateListOf<Int>()}
    val lifecycleOwner = LocalLifecycleOwner.current
 
    LaunchedEffect(key1 = Unit) {
        sharedFlow.collect {
             messages.add(it)
        }
    }
.
.

This code accesses the shared flow instance within the view model and begins collecting values from the stream. Each collected value is added to the messages mutable list. This will cause a recomposition and the new value will appear at the end of the LazyColumn list.

Testing the SharedFlowDemo app

Compile and run the app on a device or emulator and verify that values appear within the LazyColumn as they are emitted by the shared flow. Rotate the device into landscape orientation to trigger a configuration change and confirm that the count continues without restarting from zero:

Figure 1-1

With the app now working, it is time to look at what happens when it is placed in the background.

Handling flows in the background

In our app, we have a shared flow that feeds values to the user interface in the form of a LazyColumn. By performing the collection in a coroutine scope, the user interface remains responsive while the flow is being collected (you can verify this by scrolling up and down within the list of values while the list is updating). This raises the question of what happens when the app is placed in the background. To find out, we can add some diagnostic output to both the emitter and collector code. First, edit the MainViewModel.kt file and add a println() call within the body of the emission for loop:

fun sharedFlowInit() {
    viewModelScope.launch {
        for (i in 1..1000) {
            delay(2000)
            println("Emitting $i")
            _sharedFlow.emit(i)
        }
    }
}

Make a similar change to the collection code block in the MainFragment.kt file as follows:

.
.
LaunchedEffect(key1 = Unit) {
        sharedFlow.collect {
        println("Collecting $it")
        messages.add(it)
    }
}
.
.

Once these changes have been made, display the Logcat tool window, enter System.out into the search bar, and run the app. As the list of values updates, output similar to the following should appear in the Run tool window:

Emitting 1
Collecting 1
Emitting 2
Collecting 2
Emitting 3
Collecting 3
.
.

Now place the app in the background and note that both the emission and collection operations continue to run, even though the app is no longer visible to the user. The continued emission is to be expected and is the correct behavior for a shared flow residing within a view model. It is wasteful of resources, however, to be collecting data and updating a user interface that is not currently visible to the user. We can resolve this problem by executing the collection using the repeatOnLifecycle function.

The repeatOnLifecycle function is a suspend function that runs a specified block of code each time the current lifecycle reaches or exceeds one of the following states:

  • Lifecycle.State.INITIALIZED
  • Lifecycle.State.CREATED
  • Lifecycle.State.STARTED
  • Lifecycle.State.RESUMED
  • Lifecycle.State.DESTROYED

Conversely, when the lifecycle drops below the target state, the coroutine is canceled.

In this case, we want the collection to start each time Lifecycle.State.STARTED is reached and to stop when the lifecycle is suspended. To implement this, modify the collection code as follows:

.
.
import androidx.lifecycle.Lifecycle
import androidx.lifecycle.repeatOnLifecycle
.
.
LaunchedEffect(key1 = Unit) {
    lifecycleOwner.repeatOnLifecycle(Lifecycle.State.STARTED) {
        sharedFlow.collect {
            println("Collecting $it")
            messages.add(it)
        }
    }
}

Run the app once again, place it in the background and note that only the emission diagnostic messages appear in the Logcat output confirming that the main activity is no longer collecting values and adding them to the

RecyclerView list. When the app is brought to the foreground, the collection will resume at the latest emitted value since replay was not configured on the shared flow.

Summary

In this chapter, we created a SharedFlow instance within a view model. We then collected the streamed values within the main activity and used that data to update the user interface. We also outlined the importance of avoiding unnecessary flow-driven user interface updates when an app is placed in the background, a problem that can easily be resolved using the repeatOnLifecycle function. This function can be used to cancel and restart asynchronous tasks such as flow collection when a target lifecycle state is reached by the containing lifecycle.

Kotlin Flow with Jetpack Compose

The earlier chapter titled “Coroutines and Side Effects in Jetpack Compose” taught us about Kotlin Coroutines and explained how they can be used to perform multiple tasks concurrently without blocking the main thread. A shortcoming of suspend functions is that they are typically only useful for performing tasks that either do not return a result or only return a single value. In this chapter, we will introduce Kotlin Flows and explore how these can be used to return sequential streams of results from coroutine-based tasks.

By the end of the chapter, you should have a good understanding of the Flow, StateFlow, and SharedFlow Kotlin types, and appreciate the difference between hot and cold flow streams. In the next chapter (“A Jetpack Compose SharedFlow Tutorial”), we will look more closely at using SharedFlow within the context of an example Android app project.

Understanding Flows

Flows are a part of the Kotlin programming language and are designed to allow multiple values to be returned sequentially from coroutine-based asynchronous tasks. A stream of data arriving over time via a network connection would, for example, be an ideal situation for using a Kotlin flow.

Flows are comprised of producers, intermediaries, and consumers. Producers are responsible for providing the data that makes up the flow. The code that retrieves the stream of data from our hypothetical network connection, for example, would be considered a producer. As each data value becomes available, the producer emits that value to the flow. The consumer sits at the opposite end of the flow stream and collects the values as they are emitted by the producer.

Intermediaries may be placed between the producer and consumer to perform additional operations on the data such as filtering the stream, performing additional processing, or transforming the data in other ways before it reaches the consumer. Figure 1-1 illustrates the typical structure of a Kotlin flow:

Figure 1-1

The flow shown in the above diagram consists of a single producer and consumer. In practice, it is possible both for multiple consumers to collect emissions from a single producer, and for a single consumer to collect data from multiple producers.

The remainder of this chapter will demonstrate many of the key features of Kotlin flows within the context of Jetpack Compose-based development.

Creating the Sample Project

Launch Android Studio and create a new Empty Compose Activity project named FlowDemo, specifying com. example.flowdemo as the package name, and selecting a minimum API level of API 26: Android 8.0 (Oreo). Within the MainActivity.kt file, delete the Greeting function and add a new empty composable named ScreenSetup which, in turn, calls a function named MainScreen:

@Composable
fun ScreenSetup() {
    MainScreen()
}
 
@Composable
fun MainScreen() {
    
}

Edit the onCreate() method function to call ScreenSetup instead of Greeting (we will modify the DefaultPreview composable later).

Next, modify the build.gradle (Module: FlowDemo.app) file to add the Compose view model and Kotlin runtime extensions libraries to the dependencies section:

dependencies {
.
.
    implementation 'androidx.lifecycle:lifecycle-viewmodel-compose:2.4.1'
    implementation 'androidx.lifecycle:lifecycle-runtime-ktx:2.4.1'
.
.
}

When prompted, click on the Sync Now button at the top of the editor panel to commit the change.

Adding a view model to the project

For this project, the flow will reside in a view model class. Add this model to the project by locating and rightclicking on the app -> java -> com.example.flowdemo entry in the project tool window and selecting the New -> Kotlin Class/File menu option. In the resulting dialog, name the class DemoViewModel before tapping the keyboard Enter key. Once created, modify the file so that it reads as follows:

package com.example.flowdemo
 
import androidx.lifecycle.ViewModel
 
class DemoViewModel : ViewModel() {
}

Return to the MainActivity.kt file and make changes to access an instance of the view model:

.
.
import androidx.lifecycle.viewmodel.compose.viewModel
.
.
@Composable
fun ScreenSetup(viewModel: DemoViewModel = viewModel()) {
    MainScreen()
}

Declaring the flow

The most basic form of flow is represented by the Kotlin Flow type. Each flow is only able to emit data of a single type which must be specified when the flow is declared. The following declaration, for example, declares a Flow instance designed to stream String-based data:

Flow<String>

When declaring a flow, we need to assign to it the code that will generate the data stream. This code is referred to as the producer block. This can be achieved using the flow() builder which takes as a parameter a coroutine suspend block containing the producer block code. Add the following code to the DemoViewModel.kt file to declare a flow named myFlow designed to emit a stream of integer values:

package com.example.flowdemo
 
import androidx.lifecycle.ViewModel
import kotlinx.coroutines.*
import kotlinx.coroutines.flow.*
 
class DemoViewModel : ViewModel() {
 
    val myFlow: Flow<Int> = flow {
        // Producer block
    }
}

As an alternative to the flow builder, the flowOf() builder can be used to convert a fixed set of values into a flow:

val myFlow2 = flowOf(2, 4, 6, 8)

Also, many Kotlin collection types now include an asFlow() extension function that can be called to convert the contained data to a flow. The following code, for example, converts an array of string values to a flow:

val myArrayFlow = arrayOf<String>("Red", "Green", "Blue").asFlow()

Emitting flow data

Once a flow has been built, the next step is to make sure the data is emitted so that it reaches any consumers that are observing the flow. Of the three flow builders we looked at in the previous section, only the flowOf() and asFlow() builders create flows that automatically emit the data as soon as a consumer starts collecting. In the case of the flow builder, however, we need to write code to manually emit each value as it becomes available. We achieve this by making calls to the emit() function and passing through as an argument the current value to be streamed. The following changes to our myFlow declaration implement a loop that emits the value of an incrementing counter. To demonstrate the asynchronous nature of flow streams, a 2-second delay is performed on each loop iteration:

val myFlow: Flow<Int> = flow {
    for (i in 0..9) {
        emit(i)
        delay(2000)
    }
}

Collecting flow data as state

As we will see later in the chapter, one way to collect data from a flow within a consumer is to call the collect() method on the flow instance. When working with Compose, however, a less flexible, but more convenient option is to convert the flow to state by calling the collectAsState() function on the flow instance. This allows us to treat the data just as we would any other state within our code. To see this in action, edit the MainActivity.kt file and make the following changes:

.
.
import androidx.compose.runtime.*
import kotlinx.coroutines.flow.*
.
.
@Composable
fun ScreenSetup(viewModel: DemoViewModel = viewModel()) {
    MainScreen(viewModel.myFlow)
}
 
@Composable
fun MainScreen(flow: Flow<Int>) {
   val count by flow.collectAsState(initial = 0)
}
.
.
@Preview(showBackground = true)
@Composable
fun DefaultPreview() {
    FlowDemoTheme {
        ScreenSetup(viewModel())
    }
}

The changes pass a myFlow reference to the MainScreen composable where it is converted to a State with an initial value of 0. Next, we need to design a simple user interface to display the count values as they are emitted to the flow:

.
.
import androidx.compose.foundation.layout.*
import androidx.compose.ui.Alignment
import androidx.compose.ui.text.TextStyle
import androidx.compose.ui.unit.sp
.
.
@Composable
fun MainScreen(myFlow: Flow<Int>) {
    val count by myFlow.collectAsState(initial = 0)

    Column(
        modifier = Modifier.fillMaxSize(),
        verticalArrangement = Arrangement.Center,
        horizontalAlignment = Alignment.CenterHorizontally
    ) {
        Text(text = "$count", style = TextStyle(fontSize = 40.sp))
    }
}

Try out the app either using the preview panel in interactive mode or by running it on a device or emulator. Once the app starts, the count value displayed on the Text component should increment as each new value is emitted by the flow.

Transforming data with intermediaries

In the previous example, we passed the data values to the consumer without any modifications. Changes to the data can be made between the producer and consumer by applying one or more intermediate flow operators. In this section, we will look at some of these operators.

The map() operator can be used to convert the value to some other value. We can use map(), for example, to convert our integer value to a string and add some additional text. Edit the DemoViewModel.kt file and create a modified version of our flow as follows:

.
.
class DemoViewModel : ViewModel() {
    
    val myFlow: Flow<Int> = flow {
        for (i in 0..9) {
            emit(i)
            delay(2000)
        }
    }
 
    val newFlow = myFlow.map {
        "Current value = $it"
    }
}

Before we can test this operator, some changes are needed within the MainActivity.kt file to use this new flow:

Composable
fun ScreenSetup(viewModel: DemoViewModel = viewModel()) {
    MainScreen(viewModel.newFlow)
}
 
@Composable
fun MainScreen(flow: Flow<String>) {
    val count by flow.collectAsState(initial = "Current value =")
.
.

When the code is executed, the text will display the text string updated with the count:

Current value = 1
Current value = 2
.
.

The map() operator will perform the conversion on every collected value. The filter() operator can be used to control which values get collected. The filter code block needs to contain an expression that returns a Boolean value. Only if the expression evaluates to true does the value pass through to the collection. The following code filters odd numbers out of the data flow (note that we’ve left the map() operator in place to demonstrate the chaining of operators):

val newFlow = myFlow
    .filter {
        it % 2 == 0
    }
    .map {
        "Current value = $it"
    }

The above changes will display count updates only for even numbers.

The transform() operator serves a similar purpose to map() but provides more flexibility. The transform() operator also needs to manually emit the modified result. A particular advantage of transform() is that it can emit multiple values, for example:

val newFlow = myFlow
    .transform {
        emit("Value = $it")
        delay(1000)
        val doubled = it * 2
        emit("Value doubled = $doubled")
    }
 
// Output
Value = 0
Value doubled = 0
Value = 1
Value doubled = 2
Value = 2
Value doubled = 4
Value = 3
.
.

Before moving to the next step, revert the newFlow declaration to its original form:

val newFlow = myFlow.map {
    "Current value = $it"
}

Collecting flow data

So far in this chapter we have used the collectAsState() function to convert a flow to a State instance. Behind the scenes, this method is using the collect() function to initiate the collection of data. Although collectAsState() works well most of the time, there will be situations where you may need to call collect(). In fact, collect() is just one of several so-called terminal flow operators that can be called directly to achieve results that aren’t possible using collectAsState().

These operators are suspend functions so can only be called from within a coroutine scope. In the chapter entitled “Coroutines and Side Effects in Jetpack Compose”, we looked at coroutines and explained how to use LaunchedEffect to execute asynchronous code safely from within a composable function. Once we have implemented the LaunchedEffect call, we still need the streamed values to be stored as state, so we also need a mutable state into which to store the latest value. Bringing these requirements together, modify the MainScreen function so that it reads as follows:

@Composable
fun MainScreen(flow: Flow<String>) {
    
    var count by remember { mutableStateOf<String>("Current value =")}
 
    LaunchedEffect(Unit) {
        flow.collect {
            count = it
        }
    }
 
    Column(
        modifier = Modifier.fillMaxSize(),
.
.

Test the app and verify that the text component updates as expected. Now that we are using the collect() function we can begin to explore some options that were not available to us when we were using collectAsState().

For example, to add code to be executed when the stream ends, the collection can be performed in a try/finally construct, for example:

LaunchedEffect(Unit) {
    try {
        flow.collect {
            count = it
        }
    } finally {
        count = "Flow stream ended."
    }
}

The collect() operator will collect every value emitted by the producer, even if new values are emitted while the last value is still being processed in the consumer. For example, our producer is configured to emit a new value every two seconds. Suppose, however, that we simulate our consumer taking 2.5 seconds to process each collected value. When executed, we will still see all of the values listed in the output because collect() does not discard any uncollected values regardless of whether more recent values have been emitted since the last collection. This type of behavior is essential to avoid data loss within the flow. In some situations, however, the consumer may be uninterested in any intermediate values emitted between the most recently processed value and the latest emitted value. In this case, the collectLatest() operator can be called on the flow instance. This operator works by canceling the current collection if a new value arrives before processing completes on the previous value and restarts the process on the latest value.

The conflate() operator is similar to the collectLatest() operator except that instead of canceling the current collection operation when a new value arrives, conflate() allows the current operation to complete, but discards intermediate values that arrive during this process. When the current operation completes, the most recent value is then collected.

Another collection operator is the single() operator. This operator collects a single value from the flow and throws an exception if it finds another value in the stream. This operator is useful where the appearance of a second stream value indicates that something else has gone wrong somewhere in the app or data source.

Adding a flow buffer

When a consumer takes time to process the values emitted by a producer, there is the potential for execution time inefficiencies to occur. Suppose, for example, that in addition to the two-second delay between each emission from our newFlow producer, the collection process in our consumer takes an additional second to complete. We can simulate this behavior as follows:

.
.
import kotlin.system.measureTimeMillis
import kotlinx.coroutines.delay
.
.
LaunchedEffect(Unit) {
 
    val elapsedTime = measureTimeMillis {
        flow.collect {
                count = it
                delay(1000)
        }
    }
    count = "Duration = $elapsedTime"
}

To allow us to measure the total time to fully process the flow, the consumer code has been placed in the closure of a call to the Kotlin measureTimeMillis() function. Next we need to make the following changes to the newFlow declaration in the DemoViewController class:

val newFlow = myFlow
    .map {
        "Current value = $it"
    }

After execution completes, a duration similar to the following will be reported:

Duration = 30044

This accounts for approximately 20 seconds to process the 10 values within newFlow and an additional 10 seconds for those values to be collected. There is an inefficiency here because the producer is waiting for the consumer to process each value before starting on the next value. This would be much more efficient if the producer did not have to wait for the consumer. We could, of course, use the collectLatest() or conflate() operators, but only if the loss of intermediate values is not a concern. To speed up the processing while also collecting every emitted value we can make use of the buffer() operator. This operator buffers values as they are emitted and passes them to the consumer when it is ready to receive them. This allows the producer to continue emitting values while the consumer is processing preceding values while also ensuring that every emitted value is collected. The buffer() operator may be applied to a flow as follows:

LaunchedEffect("Unit") {

    val elapsedTime = measureTimeMillis {
        flow
            .buffer()
            .collect {
                count = it
                delay(1000)
        }
    }
    count = "Duration = $elapsedTime"
}

Execution of the above code indicates that we have now reclaimed the 10 seconds previously lost in the collection code:

Duration = 20052

More terminal flow operators

The reduce() operator is one of several other terminal flow operators that can be used in place of a collection operator to make changes to the flow data. The reduce() operator takes two parameters in the form of an accumulator and a value. The first flow value is placed in the accumulator and a specified operation is performed between the accumulator and the current value (with the result stored in the accumulator). To try this out we need to revert to using myFlow instead of newFlow in addition to adding the reduce() operator call:

@Composable
fun ScreenSetup(viewModel: DemoViewModel = viewModel()) {
    MainScreen(viewModel.myFlow)
}

@Composable
fun MainScreen(flow: Flow<Int>) {
    
    var count by remember { mutableStateOf<Int>(0)}


    LaunchedEffect(Unit) {
    
        flow
            .reduce { accumulator, value ->
                count = accumulator
                accumulator + value
            }
    }
.
.

The fold() operator works similarly to the reduce() operator, with the exception that it is passed an initial accumulator value:

.
.
LaunchedEffect(Unit) {

    flow
        .fold(10) { accumulator, value ->
            count = accumulator
            accumulator + value
        }

}
.
.

Flow flattening

As we have seen in earlier examples, we can use operators to perform tasks on values collected from a flow. An interesting situation occurs, however, when that task itself creates one or more flows resulting in a “flow of flows”. In situations where this occurs, these streams can be flattened into a single stream. Consider the following example code which declares two flows:

val myFlow: Flow<Int> = flow {
    for (i in 1..5) {
        delay(1000)
        emit(i)
    }
}

fun doubleIt(value: Int) = flow {
    emit(value)
    delay(1000)
    emit(value + value)
}

If we were to call doubleIt() for each value in the myFlow stream we would end up with a separate flow for each value. This problem can be solved by concatenating the doubleIt() streams into a single flow using the flatMapConcat() operator as follows:

@Composable
fun ScreenSetup(viewModel: DemoViewModel = viewModel()) {
    MainScreen(viewModel)
}

@Composable
fun MainScreen(viewModel: DemoViewModel) {

    var count by remember { mutableStateOf<Int>(0)}

    LaunchedEffect(Unit) {

        viewModel.myFlow
            .flatMapConcat { viewModel.doubleIt(it) }
            .collect { 
              count = it 
              println("Count = $it")
            }
    }
.
.

When this modified code executes we will see the following output from the collect() operator:

1
2
2
4
3
6
4
8
5
10

As we can see from the output, the doubleIt() flow has emitted the value provided by myFlow followed by the doubled value. When using the flatMapConcat() operator, the doubleIt() calls are being performed synchronously, causing execution to wait until doubleIt() has emitted both values before processing the next flow value. The emitted values can instead be collected asynchronously using the flatMapMerge() operator as follows:

viewModel.myFlow
    .flatMapMerge { viewModel.doubleIt(it) }
    .collect { 
        count = it 
        println("Count = $it")
    }
}

Because the collection is being performed asynchronously the displayed value change too quickly to see all of the count values. Display the Run tool window to see the full list of collected values generated by the println() call:

I/System.out: Count = 1
I/System.out: Count = 2
I/System.out: Count = 2
I/System.out: Count = 4
I/System.out: Count = 3
I/System.out: Count = 6
I/System.out: Count = 4
I/System.out: Count = 8
I/System.out: Count = 5
I/System.out: Count = 10

Combining multiple flows

Multiple flows can be combined into a single flow using the zip() and combine() operators. The following code demonstrates the zip() operator being used to convert two flows into a single flow:

var count by remember { mutableStateOf<String>("")}

LaunchedEffect(Unit) {

    val flow1 = (1..5).asFlow()
        .onEach { delay(1000) }
    val flow2 = flowOf("one", "two", "three", "four")
        .onEach { delay(1500) }
    flow1.zip(flow2) { value, string -> "$value, $string" }
        .collect { count = it }
}
// Output
1, one
2, two
3, three
4, four

Note that we have applied the onEach() operator to both flows in the above code. This is a useful operator for performing a task on receipt of each stream value.

The zip() operator will wait until both flows have emitted a new value before performing the collection. The combine() operator works slightly differently in that it proceeds as soon as either flow emits a new value, using the last value emitted by the other flow in the absence of a new value:

.
.
    val flow1 = (1..5).asFlow()
        .onEach { delay(1000) }
    val flow2 = flowOf("one", "two", "three", "four")
        .onEach { delay(1500) }
    flow1.combine(flow2) { value, string -> "$value, $string" }
        .collect { count = it }
.
.
// Output
1, one
2, one
3, one
3, two
4, two
4, three
5, three
5, four

As we can see from the output, multiple instances have occurred where the last value has been reused on a flow because a new value was emitted on the other.

Hot and cold flows

So far in this chapter, we have looked exclusively at the Kotlin Flow type. Kotlin also provides additional types in the form of StateFlow and SharedFlow. Before exploring these, however, it is important to understand the concept of hot and cold flows.

A stream declared using the Flow type is referred to as a cold flow because the code within the producer does not begin executing until a consumer begins collecting values. StateFlow and SharedFlow, on the other hand, are referred to as hot flows because they begin emitting values immediately, regardless of whether any consumers are collecting the values.

Once a consumer begins collecting from a hot flow, it will receive the latest value emitted by the producer followed by any subsequent values. Unless steps are taken to implement caching, any previous values emitted before the collection starts will be lost.

Another important difference between Flow, StateFlow, and SharedFlow is that a Flow-based stream cannot have multiple collectors. Each Flow collector launches a new flow with its own independent data stream. With StateFlow and SharedFlow, on the other hand, multiple collectors share access to the same flow.

StateFlow

StateFlow, as the name suggests, is primarily used as a way to observe a change in state within an app such as the current setting of a counter, toggle button, or slider. Each StateFlow instance is used to store a single value that is likely to change over time and to notify all consumers when those changes occur. This enables you to write code that reacts to changes in state instead of code that has to continually check whether or not a state value has changed. StateFlow behaves the same way as LiveData with the exception that LiveData has lifecycle awareness and does not require an initial value (LiveData was covered previously in the chapter titled “Working with ViewModels in Compose”).

To create a StateFlow stream, begin by creating an instance of MutableStateFlow, passing through a mandatory initial value. This is the variable that will be used to change the current state value from within the app code:

private val _stateFlow = MutableStateFlow(0)

Next, call asStateFlow() on the MutableStateFlow instance to convert it into a StateFlow from which changes in state can be collected:

val stateFlow = _stateFlow.asStateFlow()

Once created, any changes to the state are made via the value property of the mutable state instance. The following code, for example, increments the state value:

_stateFlow.value += 1

Once the flow is active, the state can be consumed using collectAsState() or directly using a collection function, though it is generally recommended to collect from StateFlow using the collectLatest() operator. To try out an example, begin by making the following modifications to the DemoViewModel.kt file:

.
.
class DemoViewModel : ViewModel() {

    private val _stateFlow = MutableStateFlow(0)
    val stateFlow = _stateFlow.asStateFlow()

    fun increaseValue() {
        _stateFlow.value += 1
    }
.
.

Next, edit the MainActivity.kt file and change MainScreen so that it collects from the new state flow and add a button configured to call the view model increaseValue() function:

.
.
import androidx.compose.material.Button
.
.
@Composable
fun MainScreen(viewModel: DemoViewModel) {

    val count by viewModel.stateFlow.collectAsState()

    Column(
        modifier = Modifier.fillMaxSize(),
        verticalArrangement = Arrangement.Center,
        horizontalAlignment = Alignment.CenterHorizontally
    ) {
        Text(text = "$count", style = TextStyle(fontSize = 40.sp))
        Button(onClick = { viewModel.increaseValue() }) {
            Text("Click Me")
        }
    }
}

Run the app and verify that the button updates the count Text component with the incremented count value each time it is clicked.

SharedFlow

SharedFlow provides a more general-purpose streaming option than that offered by StateFlow. Some of the key differences between StateFlow and SharedFlow are as follows:

  • Consumers are generally referred to as subscribers.
  • An initial value is not provided when creating a SharedFlow instance.
  • SharedFlow allows values that were emitted prior to collection starting to be “replayed” to the collector.
  • SharedFlow emits values instead of using a value property.

SharedFlow instances are created using MutableSharedFlow as the backing property on which we call the asSharedFlow() function to obtain a SharedFlow reference. For example, make the following changes to the DemoViewModel class to declare a shared flow:

.
.
import androidx.lifecycle.viewModelScope
import kotlinx.coroutines.channels.BufferOverflow
.
.
class DemoViewModel : ViewModel() {

    private val _sharedFlow = MutableSharedFlow<Int>(
        replay = 10,
        onBufferOverflow = BufferOverflow.DROP_OLDEST
    )

    val sharedFlow = _sharedFlow.asSharedFlow()
.
.

As configured above, new flow subscribers will receive the last 10 values before receiving any new values. The above flow is also configured to discard the oldest value when more than 10 values are buffered. The full set of options for handling buffer overflows are as follows:

  • DROP_LATEST – The latest value is dropped when the buffer is full leaving the buffer unchanged as new values are processed.
  • DROP_OLDEST – Treats the buffer as a “last-in, first-out” stack where the oldest value is dropped to make room for a new value when the buffer is full.
  • SUSPEND – The flow is suspended when the buffer is full.

Values are emitted on a SharedFlow stream by calling the emit() method of the MutableSharedFlow instance from within a coroutine. Remaining in the DemoViewModel.kt file, add a new method that can be called from the main activity to start the shared flow:

fun startSharedFlow() {
    
    viewModelScope.launch {
        for (i in 1..5) {
            _sharedFlow.emit(i)
            delay(2000)
        }
    }
}

Finally, make the following changes to the MainScreen composable:

@Composable
fun MainScreen(viewModel: DemoViewModel) {

    val count by viewModel.sharedFlow.collectAsState(initial = 0)

    Column(
        modifier = Modifier.fillMaxSize(),
        verticalArrangement = Arrangement.Center,
        horizontalAlignment = Alignment.CenterHorizontally
    ) {
        Text(text = "$count", style = TextStyle(fontSize = 40.sp))
        Button(onClick = { viewModel.startSharedFlow() }) {
            Text("Click Me")
        }
    }
}

Run the app on a device or emulator (shared flow code does not always work in the interactive preview) and verify that clicking the button causes the count to begin updating. Note that since new values are being emitted from within a coroutine you can click on the button repeatedly and collect values from multiple flows.

One final point to note about shared flows is that the current number of subscribers to a SharedFlow stream can be obtained via the subscriptionCount property of the mutable instance:

val subCount = _sharedFlow.subscriptionCount

Converting a flow from cold to hot

A cold flow can be made hot by calling the shareIn() function on the flow. This call requires a coroutine scope in which to execute the flow, a replay value, and start policy setting indicating the conditions under which the flow is to start and stop. The available start policy options are as follows:

  • SharingStarted.WhileSubscribed() – The flow is kept alive as long as it has active subscribers.
  • SharingStarted.Eagerly() – The flow begins immediately and remains active even in the absence of active subscribers.
  • SharingStarted.Lazily() – The flow begins only after the first consumer subscribes and remains active even in the absence of active subscribers.

We could, for example, make one of our earlier cold flows hot using the following code:

val hotFlow = myFlow.shareIn(
    viewModelScope, 
    replay = 1, 
    started = SharingStarted.WhileSubscribed()
)

Summary

Kotlin flows allow sequential data or state changes to be returned over time from asynchronous tasks. A flow consists of a producer that emits a sequence of values and consumers that collect and process those values. The flow stream can be manipulated between the producer and consumer by applying one or more intermediary operators including transformations and filtering. Flows are created based on the Flow, StateFlow, and SharedFlow types. A Flow-based stream can only have a single collector while StateFlow and SharedFlow can have multiple collectors.

Flows are categorized as being hot or cold. A cold flow does not begin emitting values until a consumer begins collection. Hot flows, on the other hand, begin emitting values as soon as they are created, regardless of whether or not the values are being collected. In the case of SharedFlow, a predefined number of values may be buffered and subsequently replayed to new subscribers when they begin collecting values. A cold flow can be made hot via a call to the flow’s shareIn() function.