Develop Reference

Sticky headers with paging library in Jetpack Compose

I'm currently playing around with the new Jetpack compose UI toolkit and I like it a lot. One thing I could not figure out is how to use stickyHeaders in a LazyColumn which is populated by the paging library. The non-paging example from the documentation is:
val grouped = contacts.groupBy { it.firstName[0] }
fun ContactsList(grouped: Map<Char, List<Contact>>) {
LazyColumn {
grouped.forEach { (initial, contactsForInitial) ->
stickyHeader {
CharacterHeader(initial)
}
items(contactsForInitial) { contact ->
ContactListItem(contact)
}
}
}
}
Since I'm using the paging library I cannot use the groupedBy so I tried to use the insertSeparators function on PagingData and insert/create the headers myself like this (please ignore the legacy Date code, it's just for testing):
// On my flow
.insertSeparators { before, after ->
when {
before == null -> ListItem.HeaderItem(after?.workout?.time ?: 0)
after == null -> ListItem.HeaderItem(before.workout.time)
(Date(before.workout.time).day != Date(after.workout.time).day) ->
ListItem.HeaderItem(before.workout.time)
// Return null to avoid adding a separator between two items.
else -> null
}
}
// In my composeable
LazyColumn {
items(workoutItems) {
when(it) {
is ListItem.HeaderItem -> this#LazyColumn.stickyHeader { Header(it) }
is ListItem.SongItem -> WorkoutItem(it)
}
}
}
But this produces a list of all my items and the header items are appended at the end. Any ideas what is the right way to use the stickyHeader function when using the paging library?

I got it to work by looking into the source code of the items function: You must not call stickyHeader within the items function. No need to modify the PagingData flow at all. Just use peek to get the next item without triggering a reload and then layout it:
LazyColumn {
val itemCount = workoutItems.itemCount
var lastWorkout: Workout? = null
for(index in 0 until itemCount) {
val workout = workoutItems.peek(index)
if(lastWorkout?.time != workout?.time) stickyHeader { Header(workout) }
item { WorkoutItem(workoutItems.getAsState(index).value) } // triggers reload
lastWorkout = workout
}
}

I believe the issue in your code was that you were calling this#LazyColumn from inside an LazyItemScope.
I experimented too with insertSeparators and reached this working LazyColumn code:
LazyColumn {
for (index in 0 until photos.itemCount) {
when (val peekData = photos.peek(index)) {
is String? -> stickyHeader {
Text(
text = (photos.getAsState(index).value as? String).orEmpty(),
)
}
is Photo? -> item(key = { peekData?.id }) {
val photo = photos.getAsState(index).value as? Photo
...
}
}
}
}

Nested transformation instead of merged livedatas

I have two livedatas startedAt & finishedAt and want to use them in a transformation that depend on both these values...i tried combine them using mediatorliveData
fun <A, B> LiveData<A>.combine(other: LiveData<B>): PairLiveData<A, B> {
return PairLiveData(this, other)
}
class PairLiveData<A, B>(first: LiveData<A>, second: LiveData<B>) : MediatorLiveData<Pair<A?, B?>>() {
init {
addSource(first) { value = it to second.value }
addSource(second) { value = first.value to it }
}
}
and observe them
val automaticActivities = Transformations.map(startedAt.combine(finishedAt)) {
Log.v("checkindex",index.toString()+" "+it.first?.size+" "+it.second?.size)
}
but this doesn't work as expected as finishedAt returns null before it get its required value ,
debugged rsult :
V/checkindex: 0 1 null
V/checkindex: 0 1 1
will it be a good idea to use a transformation inside a transformation ? ex:
val automaticActivities = Transformations.map(startedAt) {
val finishedAt = Transformations.map(finishedAt){
}
}

How to detect when an EditText is empty using RxTextView (RxBinding)

I'm doing validation on an EditText. I want the CharSequence to be invalid if it's empty or it doesn't begin with "https://". I'm also using RxBinding, specifically RxTextView. The problem is that when there is one character left, and I then delete it leaving no characters left in the the CharSequence the map operator doesn't fire off an emission. In other words I want my map operator to return false when the EditText is empty. I'm beginning to think this may not be possible the way I'm doing it. What would be an alternative?
Here is my Observable / Disposable:
val systemIdDisposable = RxTextView.textChanges(binding.etSystemId)
.skipInitialValue()
.map { charSeq ->
if (charSeq.isEmpty()) {
false
} else {
viewModel.isSystemIdValid(charSeq.toString())
}
}
.subscribe { isValid ->
if (!isValid) {
binding.systemIdTextInputLayout.isErrorEnabled = true
binding.systemIdTextInputLayout.error = viewModel.authErrorFields.value?.systemId
} else {
binding.systemIdTextInputLayout.isErrorEnabled = false
binding.systemIdTextInputLayout.error = viewModel.authErrorFields.value?.systemId
}
}
And here is a function in my ViewModel that I pass the CharSequence to for validation:
fun isSystemIdValid(systemId: String?): Boolean {
return if (systemId != null && systemId.isNotEmpty()) {
_authErrors.value?.systemId = null
true
} else {
_authErrors.value?.systemId =
getApplication<Application>().resources.getString(R.string.field_empty_error)
false
}
}

After sleeping on it, I figured it out.
I changed RxTextView.textChanges to RxTextView.textChangeEvents. This allowed me to query the CharSequence's text value (using text() method provided by textChangeEvents) even if it's empty. Due to some other changes (not really relevant to what I was asking in this question) I was also able to reduce some of the conditional code too. I'm just putting that out there in case someone comes across this and is curious about these changes. The takeaway is that you can get that empty emission using RxTextView.textChangeEvents.
Here is my new Observer:
val systemIdDisposable = RxTextView.textChangeEvents(binding.etSystemId)
.skipInitialValue()
.map { charSeq -> viewModel.isSystemIdValid(charSeq.text().toString()) }
.subscribe {
binding.systemIdTextInputLayout.error = viewModel.authErrors.value?.systemId
}
And here is my validation code from the ViewModel:
fun isSystemIdValid(systemId: String?): Boolean {
val auth = _authErrors.value
return if (systemId != null && systemId.isNotEmpty()) {
auth?.systemId = null
_authErrors.value = auth
true
} else {
auth?.systemId =
getApplication<Application>().resources.getString(R.string.field_empty_error)
_authErrors.value = auth
false
}
}
Lastly, if anyone is curious about how I'm using my LiveData / MutableLiveData objects; I create a private MutableLiveData object and only expose an immutable LiveData object that returns the values of the first object. I do this for better encapsulation / data hiding. Here is an example:
private val _authErrors: MutableLiveData<AuthErrorFields> by lazy {
MutableLiveData<AuthErrorFields>()
}
val authErrors: LiveData<AuthErrorFields>
get() { return _authErrors }
Hope this helps someone! 🤗

Remove data from list while iterating kotlin

I am new to kotlin programming. What I want is that I want to remove a particular data from a list while iterating through it, but when I am doing that my app is crashing.
for ((pos, i) in listTotal!!.withIndex()) {
if (pos != 0 && pos != listTotal!!.size - 1) {
if (paymentsAndTagsModel.tagName == i.header) {
//listTotal!!.removeAt(pos)
listTotal!!.remove(i)
}
}
}
OR
for ((pos,i) in listTotal!!.listIterator().withIndex()){
if (i.header == paymentsAndTagsModel.tagName){
listTotal!!.listIterator(pos).remove()
}
}
The exception which I am getting
java.lang.IllegalStateException

use removeAll
pushList?.removeAll { TimeUnit.MILLISECONDS.toMinutes(
System.currentTimeMillis() - it.date) > THRESHOLD }

val numbers = mutableListOf(1,2,3,4,5,6)
val numberIterator = numbers.iterator()
while (numberIterator.hasNext()) {
val integer = numberIterator.next()
if (integer < 3) {
numberIterator.remove()
}
}

It's forbidden to modify a collection through its interface while iterating over it. The only way to mutate the collection contents is to use Iterator.remove.
However using Iterators can be unwieldy and in vast majority of cases it's better to treat the collections as immutable which Kotlin encourages. You can use a filter to create a new collections like so:
listTotal = listTotal.filterIndexed { ix, element ->
ix != 0 && ix != listTotal.lastIndex && element.header == paymentsAndTagsModel.tagName
}

The answer by miensol seems perfect.
However, I don't understand the context for using the withIndex function or filteredIndex. You can use the filter function just by itself.
You don't need access to the index the list is at, if you're using
lists.
Also, I'd strongly recommend working with a data class if you already aren't. Your code would look something like this
Data Class
data class Event(
var eventCode : String,
var header : String
)
Filtering Logic
fun main(args:Array<String>){
val eventList : MutableList<Event> = mutableListOf(
Event(eventCode = "123",header = "One"),
Event(eventCode = "456",header = "Two"),
Event(eventCode = "789",header = "Three")
)
val filteredList = eventList.filter { !it.header.equals("Two") }
}

The following code works for me:
val iterator = listTotal.iterator()
for(i in iterator){
if(i.haer== paymentsAndTagsModel.tagName){
iterator.remove()
}
}
You can also read this article.

People didn't break iteration in previous posts dont know why. It can be simple but also with extensions and also for Map:
fun <T> MutableCollection<T>.removeFirst(filter: (T) -> Boolean) =
iterator().removeIf(filter)
fun <K, V> MutableMap<K, V>.removeFirst(filter: (K, V) -> Boolean) =
iterator().removeIf { filter(it.key, it.value) }
fun <T> MutableIterator<T>.removeFirst(filter: (T) -> Boolean): Boolean {
for (item in this) if (filter.invoke(item)) {
remove()
return true
}
return false
}

Use a while loop, here is the kotlin extension function:
fun <E> MutableList<E>.removeIfMatch(isMatchConsumer: (existingItem: E) -> Boolean) {
var index = 0
var lastIndex = this.size -1
while(index <= lastIndex && lastIndex >= 0){
when {
isMatchConsumer.invoke(this[index]) -> {
this.removeAt(index)
lastIndex-- // max is decreased by 1
}
else -> index++ // only increment if we do not remove
}
}
}

Typically you can use:
yourMutableCollection.removeIf { someLogic == true }
However, I'm working with an Android app that must support APIs older than 24.
In this case removeIf can't be used.
Here's a solution that is nearly identical to that implemented in Kotlin Collections that doesn't rely on Predicate.test - which is why API 24+ is required in the first place
//This function is in Kotlin Collections but only for Android API 24+
fun <E> MutableCollection<E>.removeIff(filter: (E) -> Boolean): Boolean {
var removed = false
val iterator: MutableIterator<E> = this.iterator()
while (iterator.hasNext()) {
val value = iterator.next()
if (filter.invoke(value)) {
iterator.remove()
removed = true
}
}
return removed
}

Another solution that will suit small collections. For example set of listeners in some controller.
inline fun <T> MutableCollection<T>.forEachSafe(action: (T) -> Unit) {
val listCopy = ArrayList<T>(this)
for (element: T in listCopy) {
if (this.contains(element)) {
action(element)
}
}
}
It makes sure that elements of collection can be removed safely even from outside code.

break or continue jump across class boundary kotlin

is anyone facing this problem.
break or continue jump across class boundary kotlin
this problem appears when i am going to use break or continue. inside lambda with receiver i create 'letIn'
lambda with receiver code
fun letIn(componentName: String?, values: List<LifeService.Value?>?,
body: (String, List<LifeService.Value?>) -> Unit) {
if (!TextUtils.isEmpty(componentName) && (values != null && values.isNotEmpty())) {
body(componentName!!, values)
}
}
this sample code for it.
for (option in 0 until optionsSize) {
val component = optionsGroup?.options?.get(option)
component?.let {
with(component) {
letIn(presentation, values, { componentName, values ->
if (componentName == LifeComponentViewType.CHECKBOX) {
letIn(transformCheckBoxValues(optionsGroup), { data ->
dataSource?.push(componentName, ComponentDataCheckBoxCollection(name, data))
view.buildComponent(componentName)
// break or continue didnt work
})
} else {
dataSource?.push(componentName, ComponentDataCollection(name, values))
view.buildComponent(componentName)
}
})
}
}
}
because above code didnt work so i use imperative way.
for (option in 0 until optionsSize) {
val component = optionsGroup?.options?.get(option)
if (component != null) {
val presentation: String? = component.presentation
val values = component.values
if (!TextUtils.isEmpty(presentation)) {
if (presentation == LifeComponentViewType.CHECKBOX) {
val data = transformCheckBoxValues(optionsGroup)
if (data.isNotEmpty()) {
dataSource?.push(presentation, ComponentDataCheckBoxCollection(optionsGroup.name, data))
view.buildComponent(presentation)
return
}
} else {
dataSource?.push(presentation!!, ComponentDataCollection(component.name, values))
view.buildComponent(presentation!!)
}
} else {
return
}
}
}
does anyone have suggestions?
UPDATE
i've been fix this issue by inlining high order function.

(Other coding errors aside) You are seeing the error because inside your lambda, you cannot use break or continue to jump out of the lambda to the nearest loop. Instead, you can use a qualified return to jump out of the lambda to a label.
Referring to the language reference
The return-expression returns from the nearest enclosing function, i.e. foo. (Note that such non-local returns are supported only for lambda expressions passed to inline functions.) If we need to return from a lambda expression, we have to label it and qualify the return:
(Emphasis mine)
Your second example shows that you want your lambdas to do a non-local return from the enclosing function. Therefore, you do not need to qualify your return, but your function letIn must be declared inline (else you can only do a local, qualified return).
inline fun letIn(componentName: String?, values: List<LifeService.Value?>?,
body: (String, List<LifeService.Value?>) -> Unit) {
if (!TextUtils.isEmpty(componentName) && (values != null && values.isNotEmpty())) {
body(componentName!!, values)
}
}
... or if you want it to have receivers...
inline fun String?.letIn(values: List<LifeService.Value?>?,
body: String.(List<LifeService.Value?>) -> Unit) {
if (!TextUtils.isEmpty(this) && (values != null && values.isNotEmpty())) {
this!!.body(values)
}
}
When you declare letIn as inline, then you can place return in your lambdas without the compiler complaining. Your function would not need to be inline if your lambdas are only doing local returns, but it would need to have a qualified return (for example return#letIn).
Your first example would then look like this...
for (option in 0 until optionsSize) {
val component = optionsGroup?.options?.get(option)
component?.let {
with(component) {
presentation.letIn(values, { values ->
if (this == LifeComponentViewType.CHECKBOX) {
this.letIn(transformCheckBoxValues(optionsGroup), { data ->
dataSource?.push(this, ComponentDataCheckBoxCollection(this, data))
view.buildComponent(this)
return //returns from function
})
} else {
dataSource?.push(this, ComponentDataCollection(name, values))
view.buildComponent(this)
return //returns from function
}
})
}
}
}
Lastly, note that if you wanted to jump out of the lambda early, but continue an outer loop as in:
fun test1() {
val list = listOf("a", "b", "c")
val optionsSize = 2
for(i in 0..optionsSize) loop# {
println("calliing list.forEach")
list.forEach lit# {
if(it == "a") return#lit
if(it == "c") return#loop
println(it)
}
}
}
It won't work. Intelli-sense doesn't complain about it, but the compiler throws an internal error. But you can convert the outer loop to a lambda, and it does work...
fun test() {
val list = listOf("a", "b", "c")
val optionsSize = 2
(0..optionsSize).forEach() loop# {
println("calliing list.forEach")
list.forEach lit# {
if(it == "a") return#lit
if(it == "c") return#loop
println(it)
}
}
}
Again, this only works if the function to which the lambda is passed is declared inline (like forEach is declared inline).