如需转载请评论或简信,并注明出处,未经允许不得转载
系列文章
- Android布局优化(一)LayoutInflate — 从布局加载原理说起
- Android布局优化(二)优雅获取界面布局耗时
- Android布局优化(三)使用AsyncLayoutInflater异步加载布局
- Android布局优化(四)X2C — 提升布局加载速度200%
- Android布局优化(五)绘制优化—避免过度绘制
目录
前言
在Android布局优化(一)从布局加载原理说起中我们说到了布局加载的两大性能瓶颈,通过IO操作将XML加载到内存中并进行解析和通过反射创建View。当xml文件过大或页面文件过深,布局的加载就会较为耗时。我们知道,当主线程进行一些耗时操作可能就会导致页面卡顿,更严重的可能会产生ANR,所以我们能如何来进行布局加载优化呢?解决这个问题有两种思路,直接解决和侧面缓解。直接解决就是不使用IO和反射等技术(这个我们会在下一节进行介绍)。侧面缓解的就是既然耗时操作难以避免,那我们能不能把耗时操作放在子线程中,等到inflate
操作完成后再将结果回调到主线程呢?答案当然是可以的,Android为我们提供了AsyncLayoutInflater
类来进行异步布局加载
AsyncLayoutInflater用法
AsyncLayoutInflater
的使用非常简单,就是把setContentView
和一些view的初始化操作都放到了onInflateFinished
回调中
@Override
protected void onCreate(Bundle savedInstanceState) {
super.onCreate(savedInstanceState);
new AsyncLayoutInflater(this).inflate(R.layout.activity_main,null, new AsyncLayoutInflater.OnInflateFinishedListener(){
@Override
public void onInflateFinished(View view, int resid, ViewGroup parent) {
setContentView(view);
rv = findViewById(R.id.tv_right);
rv.setLayoutManager(new V7LinearLayoutManager(MainActivity.this));
rv.setAdapter(new RightRvAdapter(MainActivity.this));
}
});
}
AsyncLayoutInflater源码分析
AsyncLayoutInflater
的源码非常短,也比较容易理解,总共只有170行左右
AsyncLayoutInflater构造方法和初始化
构造方法中做了三件事件
创建
BasicInflater
创建
Handler
创建
InflateThread
inflate
方法创建一个InflateRequest
对象,并将resid
、parent
、callback
等变量存储到这个对象中,并调用enqueue
方法向队列中添加一个请求
public final class AsyncLayoutInflater {
private static final String TAG = "AsyncLayoutInflater";
LayoutInflater mInflater;
Handler mHandler;
InflateThread mInflateThread;
public AsyncLayoutInflater(@NonNull Context context) {
mInflater = new BasicInflater(context);
mHandler = new Handler(mHandlerCallback);
mInflateThread = InflateThread.getInstance();
}
@UiThread
public void inflate(@LayoutRes int resid, @Nullable ViewGroup parent,
@NonNull OnInflateFinishedListener callback) {
if (callback == null) {
throw new NullPointerException("callback argument may not be null!");
}
InflateRequest request = mInflateThread.obtainRequest();
request.inflater = this;
request.resid = resid;
request.parent = parent;
request.callback = callback;
mInflateThread.enqueue(request);
}
....
}
InflateThread
这个类的主要作用就是创建一个子线程,将inflate
请求添加到阻塞队列中,并按顺序执行BasicInflater.inflate
操作(BasicInflater
实际上就是LayoutInflater
的子类)。不管infalte
成功或失败后,都会将request消息发送给主线程做处理
private static class InflateThread extends Thread {
private static final InflateThread sInstance;
static {
sInstance = new InflateThread();
sInstance.start();
}
public static InflateThread getInstance() {
return sInstance;
}
//生产者-消费者模型,阻塞队列
private ArrayBlockingQueue<InflateRequest> mQueue = new ArrayBlockingQueue<>(10);
//使用了对象池来缓存InflateThread对象,减少对象重复多次创建,避免内存抖动
private SynchronizedPool<InflateRequest> mRequestPool = new SynchronizedPool<>(10);
public void runInner() {
InflateRequest request;
try {
//从队列中取出一条请求,如果没有则阻塞
request = mQueue.take();
} catch (InterruptedException ex) {
// Odd, just continue
Log.w(TAG, ex);
return;
}
try {
//inflate操作(通过调用BasicInflater类)
request.view = request.inflater.mInflater.inflate(
request.resid, request.parent, false);
} catch (RuntimeException ex) {
// 回退机制:如果inflate失败,回到主线程去inflate
Log.w(TAG, "Failed to inflate resource in the background! Retrying on the UI"
+ " thread", ex);
}
//inflate成功或失败,都将request发送到主线程去处理
Message.obtain(request.inflater.mHandler, 0, request)
.sendToTarget();
}
@Override
public void run() {
//死循环(实际不会一直执行,内部是会阻塞等待的)
while (true) {
runInner();
}
}
//从对象池缓存中取出一个InflateThread对象
public InflateRequest obtainRequest() {
InflateRequest obj = mRequestPool.acquire();
if (obj == null) {
obj = new InflateRequest();
}
return obj;
}
//对象池缓存中的对象的数据清空,便于对象复用
public void releaseRequest(InflateRequest obj) {
obj.callback = null;
obj.inflater = null;
obj.parent = null;
obj.resid = 0;
obj.view = null;
mRequestPool.release(obj);
}
//将inflate请求添加到ArrayBlockingQueue(阻塞队列)中
public void enqueue(InflateRequest request) {
try {
mQueue.put(request);
} catch (InterruptedException e) {
throw new RuntimeException(
"Failed to enqueue async inflate request", e);
}
}
}
InflateRequest
InflateRequest
其实就可以理解为主线程和子线程之间传递的数据模型,类似Message
的作用
private static class InflateRequest {
AsyncLayoutInflater inflater;
ViewGroup parent;
int resid;
View view;
OnInflateFinishedListener callback;
InflateRequest() {
}
}
BasicInflater
BasicInflater
继承自 LayoutInflater
,只是覆写了 onCreateView
:优先加载这三个前缀的 Layout
,然后才按照默认的流程去加载,因为大多数情况下我们 Layout
中使用的View
都在这三个 package
下
private static class BasicInflater extends LayoutInflater {
private static final String[] sClassPrefixList = {
"android.widget.",
"android.webkit.",
"android.app."
};
BasicInflater(Context context) {
super(context);
}
@Override
public LayoutInflater cloneInContext(Context newContext) {
return new BasicInflater(newContext);
}
@Override
protected View onCreateView(String name, AttributeSet attrs) throws ClassNotFoundException {
for (String prefix : sClassPrefixList) {
try {
//优先加载"android.widget.”、 "android.webkit."、"android.app."
View view = createView(name, prefix, attrs);
if (view != null) {
return view;
}
} catch (ClassNotFoundException e) {
}
}
return super.onCreateView(name, attrs);
}
}
mHandlerCallback
这里就是在主线程中handleMessage
的操作,这里有一个回退机制,就是当子线程中inflate
失败后,会继续再主线程中进行inflate
操作,最终通过OnInflateFinishedListener
接口将view
回调到主线程
private Callback mHandlerCallback = new Callback() {
@Override
public boolean handleMessage(Message msg) {
InflateRequest request = (InflateRequest) msg.obj;
if (request.view == null) {
//view == null说明inflate失败
//继续再主线程中进行inflate操作
request.view = mInflater.inflate(
request.resid, request.parent, false);
}
//回调到主线程
request.callback.onInflateFinished(
request.view, request.resid, request.parent);
mInflateThread.releaseRequest(request);
return true;
}
};
OnInflateFinishedListener
布局加载完成后,通过OnInflateFinishedListener
将加载完成后的view
回调出来
public interface OnInflateFinishedListener {
void onInflateFinished(View view, int resid, ViewGroup parent);
}
AsyncLayoutInflater的局限性及改进
使用AsyncLayoutInflate
主要有如下几个局限性:
所有构建的
View
中必须不能直接使用Handler
或者是调用Looper.myLooper()
,因为异步线程默认没有调用Looper.prepare ()
异步转换出来的 View 并没有被加到 parent view中,
AsyncLayoutInflater
是调用了LayoutInflater.inflate(int, ViewGroup, false)
,因此如果需要加到 parent view 中,就需要我们自己手动添加;AsyncLayoutInflater
不支持设置LayoutInflater.Factory
或者LayoutInflater.Factory2
同时缓存队列默认 10 的大小限制如果超过了10个则会导致主线程的等待
使用单线程来做全部的
inflate
工作,如果一个界面中layout
很多不一定能满足需求
那我们如何来解决这些问题呢?AsyncLayoutInflate
类修饰为 final
,所以不能通过继承重写父类来实现。庆幸的是AsyncLayoutInflate
的代码非常短而且相对简单,所以我们可以直接把AsyncLayoutInflate
的代码复制出来一份,然后在这基础之上进行改进优化
接下来我们主要从两个方面来进行优化
引入线程池,减少单线程等待
手动设置setFactory2
public class AsyncLayoutInflatePlus {
private static final String TAG = "AsyncLayoutInflatePlus";
private Pools.SynchronizedPool<InflateRequest> mRequestPool = new Pools.SynchronizedPool<>(10);
LayoutInflater mInflater;
Handler mHandler;
Dispather mDispatcher;
public AsyncLayoutInflatePlus(@NonNull Context context) {
mInflater = new BasicInflater(context);
mHandler = new Handler(mHandlerCallback);
mDispatcher = new Dispather();
}
@UiThread
public void inflate(@LayoutRes int resid, @Nullable ViewGroup parent,
@NonNull OnInflateFinishedListener callback) {
if (callback == null) {
throw new NullPointerException("callback argument may not be null!");
}
InflateRequest request = obtainRequest();
request.inflater = this;
request.resid = resid;
request.parent = parent;
request.callback = callback;
mDispatcher.enqueue(request);
}
private Handler.Callback mHandlerCallback = new Handler.Callback() {
@Override
public boolean handleMessage(Message msg) {
InflateRequest request = (InflateRequest) msg.obj;
if (request.view == null) {
request.view = mInflater.inflate(
request.resid, request.parent, false);
}
request.callback.onInflateFinished(
request.view, request.resid, request.parent);
releaseRequest(request);
return true;
}
};
public interface OnInflateFinishedListener {
void onInflateFinished(@NonNull View view, @LayoutRes int resid,
@Nullable ViewGroup parent);
}
private static class InflateRequest {
AsyncLayoutInflatePlus inflater;
ViewGroup parent;
int resid;
View view;
OnInflateFinishedListener callback;
InflateRequest() {
}
}
private static class Dispather {
//获得当前CPU的核心数
private static final int CPU_COUNT = Runtime.getRuntime().availableProcessors();
//设置线程池的核心线程数2-4之间,但是取决于CPU核数
private static final int CORE_POOL_SIZE = Math.max(2, Math.min(CPU_COUNT - 1, 4));
//设置线程池的最大线程数为 CPU核数 * 2 + 1
private static final int MAXIMUM_POOL_SIZE = CPU_COUNT * 2 + 1;
//设置线程池空闲线程存活时间30s
private static final int KEEP_ALIVE_SECONDS = 30;
private static final ThreadFactory sThreadFactory = new ThreadFactory() {
private final AtomicInteger mCount = new AtomicInteger(1);
public Thread newThread(Runnable r) {
return new Thread(r, "AsyncLayoutInflatePlus #" + mCount.getAndIncrement());
}
};
//LinkedBlockingQueue 默认构造器,队列容量是Integer.MAX_VALUE
private static final BlockingQueue<Runnable> sPoolWorkQueue =
new LinkedBlockingQueue<Runnable>();
/**
* An {@link Executor} that can be used to execute tasks in parallel.
*/
public static final ThreadPoolExecutor THREAD_POOL_EXECUTOR;
static {
Log.i(TAG, "static initializer: " + " CPU_COUNT = " + CPU_COUNT + " CORE_POOL_SIZE = " + CORE_POOL_SIZE + " MAXIMUM_POOL_SIZE = " + MAXIMUM_POOL_SIZE);
ThreadPoolExecutor threadPoolExecutor = new ThreadPoolExecutor(
CORE_POOL_SIZE, MAXIMUM_POOL_SIZE, KEEP_ALIVE_SECONDS, TimeUnit.SECONDS,
sPoolWorkQueue, sThreadFactory);
threadPoolExecutor.allowCoreThreadTimeOut(true);
THREAD_POOL_EXECUTOR = threadPoolExecutor;
}
public void enqueue(InflateRequest request) {
THREAD_POOL_EXECUTOR.execute((new InflateRunnable(request)));
}
}
private static class BasicInflater extends LayoutInflater {
private static final String[] sClassPrefixList = {
"android.widget.",
"android.webkit.",
"android.app."
};
BasicInflater(Context context) {
super(context);
if (context instanceof AppCompatActivity) {
// 手动setFactory2,兼容AppCompatTextView等控件
AppCompatDelegate appCompatDelegate = ((AppCompatActivity) context).getDelegate();
if (appCompatDelegate instanceof LayoutInflater.Factory2) {
LayoutInflaterCompat.setFactory2(this, (LayoutInflater.Factory2) appCompatDelegate);
}
}
}
@Override
public LayoutInflater cloneInContext(Context newContext) {
return new BasicInflater(newContext);
}
@Override
protected View onCreateView(String name, AttributeSet attrs) throws ClassNotFoundException {
for (String prefix : sClassPrefixList) {
try {
View view = createView(name, prefix, attrs);
if (view != null) {
return view;
}
} catch (ClassNotFoundException e) {
// In this case we want to let the base class take a crack
// at it.
}
}
return super.onCreateView(name, attrs);
}
}
private static class InflateRunnable implements Runnable {
private InflateRequest request;
private boolean isRunning;
public InflateRunnable(InflateRequest request) {
this.request = request;
}
@Override
public void run() {
isRunning = true;
try {
request.view = request.inflater.mInflater.inflate(
request.resid, request.parent, false);
} catch (RuntimeException ex) {
// Probably a Looper failure, retry on the UI thread
Log.w(TAG, "Failed to inflate resource in the background! Retrying on the UI"
+ " thread", ex);
}
Message.obtain(request.inflater.mHandler, 0, request)
.sendToTarget();
}
public boolean isRunning() {
return isRunning;
}
}
public InflateRequest obtainRequest() {
InflateRequest obj = mRequestPool.acquire();
if (obj == null) {
obj = new InflateRequest();
}
return obj;
}
public void releaseRequest(InflateRequest obj) {
obj.callback = null;
obj.inflater = null;
obj.parent = null;
obj.resid = 0;
obj.view = null;
mRequestPool.release(obj);
}
public void cancel() {
mHandler.removeCallbacksAndMessages(null);
mHandlerCallback = null;
}
}
总结
本文介绍了通过异步的方式进行布局加载,缓解了主线程的压力。同时也介绍了AsyncLayoutInflate
的实现原理以及如何定制自己的AsyncLayoutInflate。本文的定制方式仅仅只是作为一个参考,具体的实现方式可以根据自己项目的实际情况来定制