Retrofit缓存网络数据RxCache--RxJava和数据库

RxJava是响应式编程, 在异步处理网络数据时, 使用广泛。

我们也可以使用一些Rx的特性, 优雅地缓存网络数据.

缓存模式: 读取数据库, 显示, 请求数据, 存储到数据库, 再更新页面.

SpikeKing 大神github地址：https://github.com/SpikeKing

代码：https://github.com/SpikeKing/wcl-rx-cache-demo



使用Dagger2+Retrofit+Rx的标准组合, 我来讲解一下如何使用.



框架

常规项目, 包含跳转缓存和非缓存页面, 为了模拟慢速环境, 延迟3秒加载数据.

public class MainActivity extends AppCompatActivity {

@Override
protected void onCreate(Bundle savedInstanceState) {
super.onCreate(savedInstanceState);
setContentView(R.layout.activity_main);
}

// 跳转无缓存
public void gotoNoCache(View view) {
startActivity(new Intent(this, NocacheActivity.class));
}

// 跳转有缓存
public void gotoCache(View view) {
startActivity(new Intent(this, CacheActivity.class));
}
}

无缓存

依赖注入三个关键部分, Application/Component/Module.

public class RcApplication extends Application {
private ApiComponent mApiComponent;

@Override public void onCreate() {
super.onCreate();
mApiComponent = DaggerApiComponent.builder()
.apiModule(new ApiModule(this)).build();
}

public ApiComponent getApiComponent() {
return mApiComponent;
}
}

@Singleton
@Component(modules = ApiModule.class)
public interface ApiComponent {
void inject(NocacheActivity activity);

void inject(CacheActivity activity);
}

@Module
public class ApiModule {
private Application mApplication;

public ApiModule(Application application) {
mApplication = application;
}

@Provides
@Singleton
public Application provideApplication() {
return mApplication;
}

@Provides
@Singleton GitHubClient provideGitHubClient() {
return new GitHubClient();
}

@Provides ObservableRepoDb provideObservableRepoDb() {
return new ObservableRepoDb(mApplication);
}
}

模块提供应用信息, GitHub的网络请求, 数据库.

@Singleton表示单例模式, 全部注入拥有一个实例.

页面, 使用RecyclerView显示列表信息, 在加载时显示ProgressBar.

/**
* 无缓存Activity
* <p>
* Created by wangchenlong on 16/1/18.
*/
public class NocacheActivity extends Activity {

@Bind(R.id.nocache_rv_list) RecyclerView mRvList;
@Bind(R.id.nocache_pb_progress) ProgressBar mPbProgress;

@Inject Application mApplication;
@Inject GitHubClient mGitHubClient;

private ListAdapter mListAdapter;

@Override protected void onCreate(Bundle savedInstanceState) {
super.onCreate(savedInstanceState);
setContentView(R.layout.activity_nocache);
ButterKnife.bind(this);

((RcApplication) getApplication()).getApiComponent().inject(this);

LinearLayoutManager layoutManager = new LinearLayoutManager(mApplication);
mRvList.setLayoutManager(layoutManager);

mListAdapter = new ListAdapter();
mRvList.setAdapter(mListAdapter);
}

@Override protected void onResume() {
super.onResume();

// 延迟3秒, 模拟效果
mGitHubClient.getRepos("SpikeKing")
.delay(3, TimeUnit.SECONDS)
.subscribeOn(Schedulers.io())
.observeOn(AndroidSchedulers.mainThread())
.subscribe(this::onSuccess, this::onError);

mPbProgress.setVisibility(View.VISIBLE);
}

private void onSuccess(ArrayList<Repo> repos) {
mListAdapter.setRepos(repos);
mPbProgress.setVisibility(View.INVISIBLE);
}

private void onError(Throwable throwable) {
mPbProgress.setVisibility(View.INVISIBLE);
}
}

通过观察可以发现, 长时间显示白屏会降低用户体验. 我来看看缓存模式.

缓存

缓存模式: 读取数据库, 显示, 请求数据, 存储到数据库, 再更新页面.

推荐使用脚本生成数据库处理类, 使用方式参考, 自动生成DbHelper的脚本.

主页逻辑.

public class CacheActivity extends Activity {

@Bind(R.id.cache_rv_list) RecyclerView mRvList; // 列表
@Bind(R.id.cache_srl_swipe) SwipeRefreshLayout mSrlSwipe; // 刷新

@Inject Application mApplication;
@Inject ObservableRepoDb mRepoDb;
@Inject GitHubClient mGitHubClient;

private ListAdapter mListAdapter; // RecyclerView适配器

@Override protected void onCreate(Bundle savedInstanceState) {
super.onCreate(savedInstanceState);
setContentView(R.layout.activity_cache);
ButterKnife.bind(this);

// 注入类
((RcApplication) getApplication()).getApiComponent().inject(this);

LinearLayoutManager layoutManager = new LinearLayoutManager(mApplication);
mRvList.setLayoutManager(layoutManager);

mListAdapter = new ListAdapter();
mRvList.setAdapter(mListAdapter);

mSrlSwipe.setOnRefreshListener(this::fetchUpdates);
}

@Override protected void onResume() {
super.onResume();
mRepoDb.getObservable()
.subscribeOn(Schedulers.io())
.observeOn(AndroidSchedulers.mainThread())
.subscribe(this::setData);

fetchUpdates();
Toast.makeText(mApplication, "正在更新", Toast.LENGTH_SHORT).show();
}

// 设置数据, 更新完成会调用
private void setData(ArrayList<Repo> repos) {
mListAdapter.setRepos(repos);
Toast.makeText(mApplication, "更新完成", Toast.LENGTH_SHORT).show();
}

private void fetchUpdates() {
// 延迟3秒, 模拟效果
mGitHubClient.getRepos("SpikeKing")
.delay(3, TimeUnit.SECONDS)
.subscribeOn(Schedulers.io())
.observeOn(AndroidSchedulers.mainThread())
.subscribe(mRepoDb::insertRepoList, this::fetchError, this::fetchComplete);
}

private void fetchError(Throwable throwable) {
mSrlSwipe.setRefreshing(false);
}

private void fetchComplete() {
mSrlSwipe.setRefreshing(false);
}
}

数据库的观察者

/**
* Redo的观察者
* <p>
* Created by wangchenlong on 16/1/18.
*/
public class ObservableRepoDb {
private PublishSubject<ArrayList<Repo>> mPublishSubject; // 发表主题
private RepoDbHelper mDbHelper; // 数据库

public ObservableRepoDb(Context context) {
mDbHelper = new RepoDbHelper(context);
mPublishSubject = PublishSubject.create();
}

// 返回观察者
public Observable<ArrayList<Repo>> getObservable() {
Observable<ArrayList<Repo>> firstObservable = Observable.fromCallable(this::getRepoList);
return firstObservable.concatWith(mPublishSubject); // 连接发表主题
}

// 从数据库获得数据
private ArrayList<Repo> getRepoList() {
mDbHelper.openForRead();
ArrayList<Repo> repos = new ArrayList<>();
Cursor c = mDbHelper.getAllRepo();
if (!c.moveToFirst()) {
return repos; // 返回空
}

do {
// 添加数据
repos.add(new Repo(
c.getString(RepoDbHelper.REPO_ID_COLUMN_POSITION),
c.getString(RepoDbHelper.REPO_NAME_COLUMN_POSITION),
c.getString(RepoDbHelper.REPO_DESCRIPTION_COLUMN_POSITION),
new Repo.Owner(c.getString(RepoDbHelper.REPO_OWNER_COLUMN_POSITION), "", "", "")));
} while (c.moveToNext());
c.close();
mDbHelper.close();
return repos;
}

// 插入Repo列表
public void insertRepoList(ArrayList<Repo> repos) {
mDbHelper.open();
mDbHelper.removeAllRepo();
for (Repo repo : repos) {
mDbHelper.addRepo(
repo.getId(),
repo.getName(),
repo.getDescription(),
repo.getOwner().getLogin()
);
}
mDbHelper.close();
mPublishSubject.onNext(repos); // 会调用更新数据
}
}

这一部分是关键, 实现网络请求同步插入数据库和更新页面.

关联PublishSubject, 在插入数据完成后, 调用绑定观察者, 更新页面.

即.concatWith(mPublishSubject)和mPublishSubject.onNext(repos).

Rx在处理网络请求方面, 确实非常优雅, 值得喜欢完美的人使用.

最后感谢作者分享

原文链接：http://blog.csdn.net/caroline_wendy/article/details/50540272

代码：https://github.com/SpikeKing/wcl-rx-cache-demo

关于缓存有一个rxcache也可以试试。

RxCache - Android和Java的Reactive缓存库

RxCache项目地址：https://github.com/VictorAlbertos/RxCache