-
Notifications
You must be signed in to change notification settings - Fork 6.3k
Paging Library Guide
A common feature is to load data automatically as a user scrolls through the items (i.e. infinite scroll). Previously there were only third party options that could accomplish this goal. Google's new Paging Library now provides this support.
Make sure to add the Google Maven repository in your root build.gradle
:
allprojects {
repositories {
google()
jcenter()
}
}
Next, add the Paging library to your app/build.gradle
:
dependencies {
implementation "android.arch.paging:runtime:1.0.0"
}
The Paging Library can only be used with RecyclerView. If you are using ListView to display your lists, you should first migrate to using the ViewHolder pattern before transitioning to RecyclerView. Moving to RecyclerView may also require changing how click handling is performed.
Once your lists are using RecyclerView, the next step is to change your adapters to inherit from the ListAdapter
class instead of the RecyclerView.Adapter
class. See this guide for more information.
Once the ListAdapter is made, we will modify the ListAdapter
to be a PagedListAdapter
. The PagedListAdapter
enables for data to be loaded in chunks and does not require all the data to be loaded in memory.
Change from:
public class TweetAdapter extends RecyclerView.Adapter<PostAdapter.ViewHolder> {
To:
public class TweetAdapter extends PagedListAdapter<Post, PostAdapter.ViewHolder> {
In addition, the adapter should no longer retain a copy of its current list. The getItem()
method should be used instead.
override fun onBindViewHolder(holder: PostAdapter.ViewHolder, position: Int) {
// getItem() should be used with ListAdapter
val tweet = getItem(position)
// null placeholders if the PagedList is configured to use them
// only works for data sets that have total count provided (i.e. PositionalDataSource)
if (tweet == null)
{
return;
}
// Handle remaining work here
// ...
}
The next step is to define our data sources. If you are already making existing network or database calls somewhere else, you most likely will need to move this code into a data source. You will also need to first pick what kind of data source to use:
- ItemKeyedDataSource - provides some type of key to use to fetch the next data size and unknown list size (see Twitter example below)
- PageKeyedDataSource - provides a next/before link and has unknown list size (example: Facebook Graph API)
- PositionalDataSource - Useful for sources that provide a fixed size list that can be fetched with arbitrary positions and sizes (see Parse example)
The Twitter API provides an example of how ItemKeyedDataSource. The size of the list is unknown, and usually fetching the next set of the data depends on the last known Twitter Post ID. The last post ID can be used to retrieve the next set of posts, thanks to the max_id parameter in the Twitter API:
The first step is to define the key that will be used to determine the next page of data. The ItemKeyedDataSource in the example below will use a Long type and rely on the Twitter Post ID. The last seen Post ID will be used to fetch the next set of data.
public class TweetDataSource extends ItemKeyedDataSource<Long, Tweet> {
// First type of ItemKeyedDataSource should match return type of getKey()
@NonNull
@Override
public Long getKey(@NonNull Tweet item) {
// item.getPostId() is a Long type
return item.getPostId();
}
Next, we should make sure to pass into the data constructor whatever dependencies are needed. In the example below, we are showing how the use of the Android Async Http library can be used. You can use other networking libraries as well.
// Pass whatever dependencies are needed to make the network call
TwitterClient mClient;
// Define the type of data that will be emitted by this datasource
public TweetDataSource(TwitterClient client) {
mClient = client;
}
Next, we need to define inside the data source the loadInitial()
and loadAfter()
.
public void loadInitial(@NonNull LoadInitialParams<Long> params, @NonNull final LoadInitialCallback<Tweet> callback) {
// Fetch data synchronously (second parameter is set to true)
// load an initial data set so the paged list is not empty.
// See https://issuetracker.google.com/u/2/issues/110843692?pli=1
JsonHttpResponseHandler jsonHttpResponseHandler = createTweetHandler(callback, true);
// No max_id should be passed on initial load
mClient.getHomeTimeline(0, params.requestedLoadSize, jsonHttpResponseHandler);
}
// Called repeatedly when more data needs to be set
@Override
public void loadAfter(@NonNull LoadParams<Long> params, @NonNull LoadCallback<Tweet> callback) {
// This network call can be asynchronous (second parameter is set to false)
JsonHttpResponseHandler jsonHttpResponseHandler = createTweetHandler(callback, false);
// params.key & requestedLoadSize should be used
// params.key will be the lowest Twitter post ID retrieved and should be used for the max_id= parameter in Twitter API.
// max_id = params.key - 1
mClient.getHomeTimeline(params.key - 1, params.requestedLoadSize, jsonHttpResponseHandler);
}
The createTweetHandler()
method parses the JSON data and posts the data to the RecyclerView adapter and PagedList handler. Keep in mind that all of these network calls are already done in a background thread, so the network call should be forced to run synchronously:
public JsonHttpResponseHandler createTweetHandler(final LoadCallback<Tweet> callback, boolean isAsync) {
JsonHttpResponseHandler handler = new JsonHttpResponseHandler() {
@Override
public void onSuccess(int statusCode, Header[] headers, JSONArray response) {
ArrayList<Tweet> tweets = new ArrayList<Tweet>();
tweets = Tweet.fromJson(response);
// send back to PagedList handler
callback.onResult(tweets);
}
};
if (isAsync) {
// Fetch data synchronously
// For AsyncHttpClient, this workaround forces the callback to be run synchronously
handler.setUseSynchronousMode(true);
handler.setUsePoolThread(true);
}
A data source factory simply creates the data source. Because of the dependency on the Twitter client, we need to pass it here too:
public class TweetDataSourceFactory extends DataSource.Factory<Long, Tweet> {
TwitterClient client;
public TweetDataSourceFactory(TwitterClient client) {
this.client = client;
}
@Override
public DataSource<Long, Tweet> create() {
TweetDataSource dataSource = new TweetDataSource(this.client);
return dataSource;
}
}
Once we have created the data source and data source factory, the final step is to create the PagedList and listen for updates. We can then call the submitList()
on our adapter with this next set of data:
public abstract class MyActivity extends AppCompatActivity {
TweetAdapter tweetAdapter;
// Normally this data should be encapsulated in ViewModels, but shown here for simplicity
LiveData<PagedList<Tweet>> tweets;
public void onCreate(Bundle savedInstanceState) {
tweetAdapter = new TweetAdapter();
// Setup rest of TweetAdapter here (i.e. LayoutManager)
// Initial page size to fetch can also be configured here too
PagedList.Config config = new PagedList.Config.Builder().setPageSize(20).build();
// Pass in dependency
TweetDataSourceFactory factory = new TweetDataSourceFactory(RestClientApp.getRestClient());
tweets = new LivePagedListBuilder(factory, config).build();
tweets.observe(this, new Observer<PagedList<Tweet>>() {
@Override
public void onChanged(@Nullable PagedList<Tweet> tweets) {
tweetAdapter.submitList(tweets);
}
});
}
The Paging Library requires that the data used in the RecyclerView, regardless of whether they are pulled from the network, database, or memory, be defined as data sources. It also allows the flexibility to first pull data from the database or memory, and if there is no more data, to pull additional segments of data from the network. To help showcase how the Paging Library works, you can also try out this sample code from Google.
In order to use the paging library with SwipeRefreshLayout, we need to be able to invalidate the data source to force a refresh. In order to do so, we first need a reference to the data source. We can do so by creating a MutableLiveData
instance, which is lifecycle aware, that will hold this value.
public class TweetDataSourceFactory extends DataSource.Factory<Long, Tweet> {
// Use to hold a reference to the
public MutableLiveData<TweetDataSource> postLiveData;
@Override
public DataSource<Long, Tweet> create() {
TweetDataSource dataSource = new TweetDataSource(this.client);
// Keep reference to the data source with a MutableLiveData reference
postLiveData = new MutableLiveData<>();
postLiveData.postValue(dataSource);
return dataSource;
}
Next, we need to move the TweetDataSourceFactory to be accessible:
public abstract class MyActivity extends AppCompatActivity {
// Should be in the ViewModel but shown here for simplicity
TweetDataSourceFactory factory;
public void onCreate(Bundle savedInstanceState) {
factory = new TweetDataSourceFactory(RestClientApp.getRestClient());
}
}
Finally, we can use the reference to the data source to call invalidate()
, which should trigger the data source to reload the data and call loadInitial()
again:
// Pass in dependency
SwipeRefreshLayout swipeContainer = v.findViewById(R.id.swipeContainer);
swipeContainer.setOnRefreshListener(new SwipeRefreshLayout.OnRefreshListener() {
@Override
public void onRefresh() {
factory.postLiveData.getValue().invalidate();
}
});
Once a swipe to refresh is triggered and a new set of data is retrieved, we simply need to call setRefreshing(false)
on the SwipeRefreshLayout
:
tweets.observe(this, new Observer<PagedList<Tweet>>() {
@Override
public void onChanged(@Nullable PagedList<Tweet> tweets) {
tweetAdapter.submitList(tweets);
swipeContainer.setRefreshing(false);
}
});
Created by CodePath with much help from the community. Contributed content licensed under cc-wiki with attribution required. You are free to remix and reuse, as long as you attribute and use a similar license.
Finding these guides helpful?
We need help from the broader community to improve these guides, add new topics and keep the topics up-to-date. See our contribution guidelines here and our topic issues list for great ways to help out.
Check these same guides through our standalone viewer for a better browsing experience and an improved search. Follow us on twitter @codepath for access to more useful Android development resources.