使用分页数据时,您通常需要在加载数据流时对其进行转换。例如,您可能需要过滤一系列数据项,或者将数据项转换为其他类型,才能在界面中呈现它们。数据流转换的另一个常见用例是添加列表分隔符。
概括来说,直接对数据流进行转换可让您将存储库构造和界面构造分开。
由于 PagingData 封装在响应式数据流中,因此您可以在加载数据之后、呈现数据之前逐步对数据执行转换操作。
为了对数据流中的每个 PagingData 对象执行转换,请将转换操作放入对数据流的 map() 操作中:
Kotlin
Coroutines
pager.flow // Type is Flow<PagingData<User>>.
// Map the outer stream so that the transformations are applied to
// each new generation of PagingData.
.map { pagingData ->
// Transformations in this block are applied to the items
// in the paged data.
}
Java
RxJava
PagingRx.getFlowable(pager) // Type is Flowable<PagingData<User>>.
// Map the outer stream so that the transformations are applied to
// each new generation of PagingData.
.map(pagingData -> {
// Transformations in this block are applied to the items
// in the paged data.
});
Java
Guava/LiveData
// Map the outer stream so that the transformations are applied to
// each new generation of PagingData.
Transformations.map(
// Type is LiveData<PagingData<User>>.
PagingLiveData.getLiveData(pager),
pagingData -> {
// Transformations in this block are applied to the items
// in the paged data.
});
可对数据流进行的最基本操作就是将其转换为其他类型。获得对 PagingData 对象的访问权限后,您可以对 PagingData 对象的分页列表中的每个单项执行 map() 操作。
该操作的一个常见用例是将某个网络或数据库层对象映射到界面层中专用的某个对象。以下示例演示了如何执行此类映射操作:
Kotlin
Coroutines
pager.flow // Type is Flow<PagingData<User>>.
.map { pagingData ->
pagingData.map { user -> UiModel(user) }
}
Java
RxJava
// Type is Flowable<PagingData<User>>.
PagingRx.getFlowable(pager)
.map(pagingData ->
pagingData.map(UiModel.UserModel::new)
)
Java
Guava/LiveData
Transformations.map(
// Type is LiveData<PagingData<User>>.
PagingLiveData.getLiveData(pager),
pagingData ->
pagingData.map(UiModel.UserModel::new)
)
另一种常见的数据转换是获取用户输入(例如查询字符串),然后将其转换为要显示的请求输出。若要设置该数据转换,您需要监听并捕获用户查询输入、执行相应请求并将查询结果推送回界面。
您可以使用数据流 API 来监听查询输入。将数据流引用保留在 ViewModel 中。界面层不应直接访问该类;相反,应该定义一个函数来通知 ViewModel 相关用户查询。
Kotlin
Coroutines
private val queryFlow = MutableStateFlow("")
fun onQueryChanged(query: String) {
queryFlow.value = query
}
Java
RxJava
private BehaviorSubject<String> querySubject = BehaviorSubject.create("");
public void onQueryChanged(String query) {
queryFlow.onNext(query)
}
Java
Guava/LiveData
private MutableLiveData<String> queryLiveData = new MutableLiveData("");
public void onQueryChanged(String query) {
queryFlow.setValue(query)
}
使用 flatMapLatest 或 switchMap 等操作可以确保只将最新结果返回到界面。如果用户在数据库操作结束前更改查询输入,这些操作会舍弃旧查询的结果,并立即启动新的搜索。
另一种常见操作是过滤。您可以根据用户条件来过滤数据;如果根据其他条件应该隐藏数据,您也可以从界面中移除相应数据。
您需要将这些过滤操作放入 map() 调用中,因为该过滤条件适用于 PagingData 对象。数据从 PagingData 中过滤掉后,系统会将新的 PagingData 实例传递到界面层进行显示。
Kotlin
Coroutines
pager.flow // Type is Flow<PagingData<User>>.
.map { pagingData ->
pagingData.filter { user -> !user.hiddenFromUi }
}
Java
RxJava
// Type is Flowable<PagingData<User>>.
PagingRx.getFlowable(pager)
.map(pagingData ->
pagingData.filter(user -> !user.isHiddenFromUi())
)
}
Java
Guava/LiveData
Transformations.map(
// Type is LiveData<PagingData<User>>.
PagingLiveData.getLiveData(pager),
pagingData ->
pagingData.filter(user -> !user.isHiddenFromUi())
)
Paging 库支持动态列表分隔符。您可以通过将分隔符作为 RecyclerView 列表项直接插入到数据流中来提高列表的可读性。因此,分隔符是功能完备的 ViewHolder 对象,可支持互动、无障碍功能焦点以及 View 提供的所有其他功能。
将分隔符插入到分页列表中需要执行以下三个步骤:
注意:如果您不需要列表分隔符支持互动或实现无障碍功能聚焦,则使用 RecyclerView.ItemDecoration 来创建静态列表分隔符会更简单。
Paging 库将列表分隔符作为实际的列表项插入到 RecyclerView 中,但分隔符必须与列表中的数据项区分开来,以便其能够绑定到具有不同界面的其他 ViewHolder 类型。解决方案是创建一个具有表示数据和分隔符的子类的 Kotlin 密封类。或者,您也可以创建一个可由您的列表项类和分隔符类扩展的基类。
假设您想要向 User 项的分页列表添加分隔符。以下代码段展示了如何创建基类,其中的实例可以是 UserModel 或 SeparatorModel:
Kotlin
Coroutines
sealed class UiModel {
class UserModel(val id: String, val label: String) : UiModel() {
constructor(user: User) : this(user.id, user.label)
}
class SeparatorModel(val description: String) : UiModel()
}
Java
RxJava
class UiModel {
private UiModel() {}
static class UserModel extends UiModel {
@NonNull
private String mId;
@NonNull
private String mLabel;
UserModel(@NonNull String id, @NonNull String label) {
mId = id;
mLabel = label;
}
UserModel(@NonNull User user) {
mId = user.id;
mLabel = user.label;
}
@NonNull
public String getId() {
return mId;
}
@NonNull
public String getLabel() {
return mLabel;
}
}
static class SeparatorModel extends UiModel {
@NonNull
private String mDescription;
SeparatorModel(@NonNull String description) {
mDescription = description;
}
@NonNull
public String getDescription() {
return mDescription;
}
}
}
Java
Guava/LiveData
class UiModel {
private UiModel() {}
static class UserModel extends UiModel {
@NonNull
private String mId;
@NonNull
private String mLabel;
UserModel(@NonNull String id, @NonNull String label) {
mId = id;
mLabel = label;
}
UserModel(@NonNull User user) {
mId = user.id;
mLabel = user.label;
}
@NonNull
public String getId() {
return mId;
}
@NonNull
public String getLabel() {
return mLabel;
}
}
static class SeparatorModel extends UiModel {
@NonNull
private String mDescription;
SeparatorModel(@NonNull String description) {
mDescription = description;
}
@NonNull
public String getDescription() {
return mDescription;
}
}
}
在加载数据流之后、呈现数据流之前必须对数据流进行转换。转换过程应该执行以下操作:
以下示例演示了用于将 PagingData 流更新为添加了分隔符的 PagingData 流的转换操作:
Kotlin
Coroutines
pager.flow.map { pagingData: PagingData<User> ->
// Map outer stream, so you can perform transformations on
// each paging generation.
pagingData
.map { user ->
// Convert items in stream to UiModel.UserModel.
UiModel.UserModel(user)
}
.insertSeparators<UiModel.UserModel, UiModel> { before, after ->
when {
before == null -> UiModel.SeparatorModel("HEADER")
after == null -> UiModel.SeparatorModel("FOOTER")
shouldSeparate(before, after) -> UiModel.SeparatorModel(
"BETWEEN ITEMS $before AND $after"
)
// Return null to avoid adding a separator between two items.
else -> null
}
}
}
Java
RxJava
// Map outer stream, so you can perform transformations on each
// paging generation.
PagingRx.getFlowable(pager).map(pagingData -> {
// First convert items in stream to UiModel.UserModel.
PagingData<UiModel> uiModelPagingData = pagingData.map(
UiModel.UserModel::new);
// Insert UiModel.SeparatorModel, which produces PagingData of
// generic type UiModel.
return PagingData.insertSeparators(uiModelPagingData,
(@Nullable UiModel before, @Nullable UiModel after) -> {
if (before == null) {
return new UiModel.SeparatorModel("HEADER");
} else if (after == null) {
return new UiModel.SeparatorModel("FOOTER");
} else if (shouldSeparate(before, after)) {
return new UiModel.SeparatorModel("BETWEEN ITEMS "
+ before.toString() + " AND " + after.toString());
} else {
// Return null to avoid adding a separator between two
// items.
return null;
}
});
});
Java
Guava/LiveData
// Map outer stream, so you can perform transformations on each
// paging generation.
Transformations.map(PagingLiveData.getLiveData(pager),
pagingData -> {
// First convert items in stream to UiModel.UserModel.
PagingData<UiModel> uiModelPagingData = pagingData.map(
UiModel.UserModel::new);
// Insert UiModel.SeparatorModel, which produces PagingData of
// generic type UiModel.
return PagingData.insertSeparators(uiModelPagingData,
(@Nullable UiModel before, @Nullable UiModel after) -> {
if (before == null) {
return new UiModel.SeparatorModel("HEADER");
} else if (after == null) {
return new UiModel.SeparatorModel("FOOTER");
} else if (shouldSeparate(before, after)) {
return new UiModel.SeparatorModel("BETWEEN ITEMS "
+ before.toString() + " AND " + after.toString());
} else {
// Return null to avoid adding a separator between two
// items.
return null;
}
});
});
最后一步是更改界面以适应分隔符项类型。为您的分隔符项创建布局和 ViewHolder,并将列表 Adapter 更改为使用 RecyclerView.ViewHolder 作为其 ViewHolder 类型,以便 Adapter 可以处理多种类型的 ViewHolder。或者,您也可以定义项和分隔符 ViewHolder 类都可扩展的通用基类。
您还必须对列表 Adapter 进行以下更改:
Kotlin
Coroutines
class UiModelAdapter :
PagingDataAdapter<UiModel, RecyclerView.ViewHolder>(UiModelComparator) {
override fun onCreateViewHolder(
parent: ViewGroup,
viewType: Int
) = when (viewType) {
R.layout.item -> UserModelViewHolder(parent)
else -> SeparatorModelViewHolder(parent)
}
override fun getItemViewType(position: Int) {
// Use peek over getItem to avoid triggering page fetch / drops, since
// recycling views is not indicative of the user's current scroll position.
return when (peek(position)) {
is UiModel.UserModel -> R.layout.item
is UiModel.SeparatorModel -> R.layout.separator_item
null -> throw IllegalStateException("Unknown view")
}
}
override fun onBindViewHolder(
holder: RecyclerView.ViewHolder,
position: Int
) {
val item = getItem(position)
if (holder is UserModelViewHolder) {
holder.bind(item as UserModel)
} else if (holder is SeparatorModelViewHolder) {
holder.bind(item as SeparatorModel)
}
}
}
object UiModelComparator : DiffUtil.ItemCallback<UiModel>() {
override fun areItemsTheSame(
oldItem: UiModel,
newItem: UiModel
): Boolean {
val isSameRepoItem = oldItem is UiModel.UserModel
&& newItem is UiModel.UserModel
&& oldItem.id == newItem.id
val isSameSeparatorItem = oldItem is UiModel.SeparatorModel
&& newItem is UiModel.SeparatorModel
&& oldItem.description == newItem.description
return isSameRepoItem || isSameSeparatorItem
}
override fun areContentsTheSame(
oldItem: UiModel,
newItem: UiModel
) = oldItem == newItem
}
Java
RxJava
class UiModelAdapter extends PagingDataAdapter<UiModel, RecyclerView.ViewHolder> {
UiModelAdapter() {
super(new UiModelComparator(), Dispatchers.getMain(),
Dispatchers.getDefault());
}
@NonNull
@Override
public RecyclerView.ViewHolder onCreateViewHolder(@NonNull ViewGroup parent,
int viewType) {
if (viewType == R.layout.item) {
return new UserModelViewHolder(parent);
} else {
return new SeparatorModelViewHolder(parent);
}
}
@Override
public int getItemViewType(int position) {
// Use peek over getItem to avoid triggering page fetch / drops, since
// recycling views is not indicative of the user's current scroll position.
UiModel item = peek(position);
if (item instanceof UiModel.UserModel) {
return R.layout.item;
} else if (item instanceof UiModel.SeparatorModel) {
return R.layout.separator_item;
} else {
throw new IllegalStateException("Unknown view");
}
}
@Override
public void onBindViewHolder(@NonNull RecyclerView.ViewHolder holder,
int position) {
if (holder instanceOf UserModelViewHolder) {
UserModel userModel = (UserModel) getItem(position);
((UserModelViewHolder) holder).bind(userModel);
} else {
SeparatorModel separatorModel = (SeparatorModel) getItem(position);
((SeparatorModelViewHolder) holder).bind(separatorModel);
}
}
}
class UiModelComparator extends DiffUtil.ItemCallback<UiModel> {
@Override
public boolean areItemsTheSame(@NonNull UiModel oldItem,
@NonNull UiModel newItem) {
boolean isSameRepoItem = oldItem instanceof UserModel
&& newItem instanceof UserModel
&& ((UserModel) oldItem).getId().equals(((UserModel) newItem).getId());
boolean isSameSeparatorItem = oldItem instanceof SeparatorModel
&& newItem instanceof SeparatorModel
&& ((SeparatorModel) oldItem).getDescription().equals(
((SeparatorModel) newItem).getDescription());
return isSameRepoItem || isSameSeparatorItem;
}
@Override
public boolean areContentsTheSame(@NonNull UiModel oldItem,
@NonNull UiModel newItem) {
return oldItem.equals(newItem);
}
}
Java
Guava/LiveData
class UiModelAdapter extends PagingDataAdapter<UiModel, RecyclerView.ViewHolder> {
UiModelAdapter() {
super(new UiModelComparator(), Dispatchers.getMain(),
Dispatchers.getDefault());
}
@NonNull
@Override
public RecyclerView.ViewHolder onCreateViewHolder(@NonNull ViewGroup parent,
int viewType) {
if (viewType == R.layout.item) {
return new UserModelViewHolder(parent);
} else {
return new SeparatorModelViewHolder(parent);
}
}
@Override
public int getItemViewType(int position) {
// Use peek over getItem to avoid triggering page fetch / drops, since
// recycling views is not indicative of the user's current scroll position.
UiModel item = peek(position);
if (item instanceof UiModel.UserModel) {
return R.layout.item;
} else if (item instanceof UiModel.SeparatorModel) {
return R.layout.separator_item;
} else {
throw new IllegalStateException("Unknown view");
}
}
@Override
public void onBindViewHolder(@NonNull RecyclerView.ViewHolder holder,
int position) {
if (holder instanceOf UserModelViewHolder) {
UserModel userModel = (UserModel) getItem(position);
((UserModelViewHolder) holder).bind(userModel);
} else {
SeparatorModel separatorModel = (SeparatorModel) getItem(position);
((SeparatorModelViewHolder) holder).bind(separatorModel);
}
}
}
class UiModelComparator extends DiffUtil.ItemCallback<UiModel> {
@Override
public boolean areItemsTheSame(@NonNull UiModel oldItem,
@NonNull UiModel newItem) {
boolean isSameRepoItem = oldItem instanceof UserModel
&& newItem instanceof UserModel
&& ((UserModel) oldItem).getId().equals(((UserModel) newItem).getId());
boolean isSameSeparatorItem = oldItem instanceof SeparatorModel
&& newItem instanceof SeparatorModel
&& ((SeparatorModel) oldItem).getDescription().equals(
((SeparatorModel) newItem).getDescription());
return isSameRepoItem || isSameSeparatorItem;
}
@Override
public boolean areContentsTheSame(@NonNull UiModel oldItem,
@NonNull UiModel newItem) {
return oldItem.equals(newItem);
}
}
需要避免的一个主要问题是让应用执行不必要的工作。提取数据是一项成本高昂的操作,并且数据转换也需要花费宝贵的时间。 一旦数据加载完毕并准备好在界面中显示,系统就应该保存数据,以备在发生配置更改且需要重新创建界面时使用。
cachedIn() 操作会缓存在它执行之前发生的任何转换的结果。因此,cachedIn() 应该是 ViewModel 中的最后一次调用。
Kotlin
Coroutines
pager.flow // Type is Flow<PagingData<User>>.
.map { pagingData ->
pagingData.filter { user -> !user.hiddenFromUi }
.map { user -> UiModel.UserModel(user) }
}
.cachedIn(viewModelScope)
Java
RxJava
// CoroutineScope helper provided by the lifecycle-viewmodel-ktx artifact.
CoroutineScope viewModelScope = ViewModelKt.getViewModelScope(viewModel);
PagingRx.cachedIn(
// Type is Flowable<PagingData<User>>.
PagingRx.getFlowable(pager)
.map(pagingData -> pagingData
.filter(user -> !user.isHiddenFromUi())
.map(UiModel.UserModel::new)),
viewModelScope);
}
Java
Guava/LiveData
// CoroutineScope helper provided by the lifecycle-viewmodel-ktx artifact.
CoroutineScope viewModelScope = ViewModelKt.getViewModelScope(viewModel);
PagingLiveData.cachedIn(
Transformations.map(
// Type is LiveData<PagingData<User>>.
PagingLiveData.getLiveData(pager),
pagingData -> pagingData
.filter(user -> !user.isHiddenFromUi())
.map(UiModel.UserModel::new)),
viewModelScope);
如需详细了解如何将 cachedIn() 用于 PagingData 流,请参阅设置 PagingData 流。
注意:在 submitData() 中,每个 PagingData 实例只能使用一次。如需绕过此限制,请考虑使用 cachedIn() 运算符,它会在缓存结果的过程中多播该数据流。如此一来,后续观察器便可以收到 PagingData 的新有效实例。如果采用重复使用最新发出的 PagingData 的运算符,例如流的 .combine() 运算符,这种方法尤其有用。