Listview异步加载图片之优化篇

在 APP应用中,listview的异步加载图片方式能够带来很好的用户体验,同时也是考量程序性能的一个重要指标。关于listview的异步加载,网上 其实很多示例了,中心思想都差不多,不过很多版本或是有bug,或是有性能问题有待优化。有鉴于此,本人在网上找了个相对理想的版本并在此基础上进行改 造,下面就让在下阐述其原理以探索个中奥秘

贴张效果图先:

ss.png

         异步加载图片基本思想:

1.      先从内存缓存中获取图片显示(内存缓冲)

2.      获取不到的话从SD卡里获取(SD卡缓冲)

3.      都获取不到的话从网络下载图片并保存到SD卡同时加入内存并显示(视情况看是否要显示)

OK,先上adapter的代码:

public class LoaderAdapter extends BaseAdapter{
  
        private static final String TAG = "LoaderAdapter";
        private boolean mBusy = false;
  
        public void setFlagBusy(boolean busy) {
                this.mBusy = busy;
        }
  
          
        private ImageLoader mImageLoader;
        private int mCount;
        private Context mContext;
        private String\[\] urlArrays;
          
          
        public LoaderAdapter(int count, Context context, String \[\]url) {
                this.mCount = count;
                this.mContext = context;
                urlArrays = url;
                mImageLoader = new ImageLoader(context);
        }
          
        public ImageLoader getImageLoader(){
                return mImageLoader;
        }
  
        @Override
        public int getCount() {
                return mCount;
        }
  
        @Override
        public Object getItem(int position) {
                return position;
        }
  
        @Override
        public long getItemId(int position) {
                return position;
        }
  
        @Override
        public View getView(int position, View convertView, ViewGroup parent) {
  
                ViewHolder viewHolder = null;
                if (convertView == null) {
                        convertView = LayoutInflater.from(mContext).inflate(
                                        R.layout.list_item, null);
                        viewHolder = new ViewHolder();
                        viewHolder.mTextView = (TextView) convertView
                                        .findViewById(R.id.tv_tips);
                        viewHolder.mImageView = (ImageView) convertView
                                        .findViewById(R.id.iv_image);
                        convertView.setTag(viewHolder);
                } else {
                        viewHolder = (ViewHolder) convertView.getTag();
                }
                String url = "";
                url = urlArrays\[position % urlArrays.length\];
                  
                viewHolder.mImageView.setImageResource(R.drawable.ic_launcher);
                  
  
                if (!mBusy) {
                        mImageLoader.DisplayImage(url, viewHolder.mImageView, false);
                        viewHolder.mTextView.setText("--" + position
                                        + "--IDLE ||TOUCH_SCROLL");
                } else {
                        mImageLoader.DisplayImage(url, viewHolder.mImageView, true);                
                        viewHolder.mTextView.setText("--" + position + "--FLING");
                }
                return convertView;
        }
  
        static class ViewHolder {
                TextView mTextView;
                ImageView mImageView;
        }
}

关键代码是ImageLoader的DisplayImage方法,再看ImageLoader的实现

public class ImageLoader {
  
        private MemoryCache memoryCache = new MemoryCache();
        private AbstractFileCache fileCache;
        private Map<ImageView, String> imageViews = Collections
                        .synchronizedMap(new WeakHashMap<ImageView, String>());
        // 线程池
        private ExecutorService executorService;
  
        public ImageLoader(Context context) {
                fileCache = new FileCache(context);
                executorService = Executors.newFixedThreadPool(5);
        }
  
        // 最主要的方法
        public void DisplayImage(String url, ImageView imageView, boolean isLoadOnlyFromCache) {
                imageViews.put(imageView, url);
                // 先从内存缓存中查找
  
                Bitmap bitmap = memoryCache.get(url);
                if (bitmap != null)
                        imageView.setImageBitmap(bitmap);
                else if (!isLoadOnlyFromCache){
                          
                        // 若没有的话则开启新线程加载图片
                        queuePhoto(url, imageView);
                }
        }
  
        private void queuePhoto(String url, ImageView imageView) {
                PhotoToLoad p = new PhotoToLoad(url, imageView);
                executorService.submit(new PhotosLoader(p));
        }
  
        private Bitmap getBitmap(String url) {
                File f = fileCache.getFile(url);
                  
                // 先从文件缓存中查找是否有
                Bitmap b = null;
                if (f != null && f.exists()){
                        b = decodeFile(f);
                }
                if (b != null){
                        return b;
                }
                // 最后从指定的url中下载图片
                try {
                        Bitmap bitmap = null;
                        URL imageUrl = new URL(url);
                        HttpURLConnection conn = (HttpURLConnection) imageUrl
                                        .openConnection();
                        conn.setConnectTimeout(30000);
                        conn.setReadTimeout(30000);
                        conn.setInstanceFollowRedirects(true);
                        InputStream is = conn.getInputStream();
                        OutputStream os = new FileOutputStream(f);
                        CopyStream(is, os);
                        os.close();
                        bitmap = decodeFile(f);
                        return bitmap;
                } catch (Exception ex) {
                        Log.e("", "getBitmap catch Exception...\\nmessage = " + ex.getMessage());
                        return null;
                }
        }
  
        // decode这个图片并且按比例缩放以减少内存消耗,虚拟机对每张图片的缓存大小也是有限制的
        private Bitmap decodeFile(File f) {
                try {
                        // decode image size
                        BitmapFactory.Options o = new BitmapFactory.Options();
                        o.inJustDecodeBounds = true;
                        BitmapFactory.decodeStream(new FileInputStream(f), null, o);
  
                        // Find the correct scale value. It should be the power of 2.
                        final int REQUIRED_SIZE = 100;
                        int width_tmp = o.outWidth, height_tmp = o.outHeight;
                        int scale = 1;
                        while (true) {
                                if (width_tmp / 2 < REQUIRED_SIZE
                                                || height_tmp / 2 < REQUIRED_SIZE)
                                        break;
                                width_tmp /= 2;
                                height_tmp /= 2;
                                scale *= 2;
                        }
  
                        // decode with inSampleSize
                        BitmapFactory.Options o2 = new BitmapFactory.Options();
                        o2.inSampleSize = scale;
                        return BitmapFactory.decodeStream(new FileInputStream(f), null, o2);
                } catch (FileNotFoundException e) {
                }
                return null;
        }
  
        // Task for the queue
        private class PhotoToLoad {
                public String url;
                public ImageView imageView;
  
                public PhotoToLoad(String u, ImageView i) {
                        url = u;
                        imageView = i;
                }
        }
  
        class PhotosLoader implements Runnable {
                PhotoToLoad photoToLoad;
  
                PhotosLoader(PhotoToLoad photoToLoad) {
                        this.photoToLoad = photoToLoad;
                }
  
                @Override
                public void run() {
                        if (imageViewReused(photoToLoad))
                                return;
                        Bitmap bmp = getBitmap(photoToLoad.url);
                        memoryCache.put(photoToLoad.url, bmp);
                        if (imageViewReused(photoToLoad))
                                return;
                        BitmapDisplayer bd = new BitmapDisplayer(bmp, photoToLoad);
                        // 更新的操作放在UI线程中
                        Activity a = (Activity) photoToLoad.imageView.getContext();
                        a.runOnUiThread(bd);
                }
        }
  
        /**
         * 防止图片错位
         * 
         * @param photoToLoad
         * @return
         */
        boolean imageViewReused(PhotoToLoad photoToLoad) {
                String tag = imageViews.get(photoToLoad.imageView);
                if (tag == null || !tag.equals(photoToLoad.url))
                        return true;
                return false;
        }
  
        // 用于在UI线程中更新界面
        class BitmapDisplayer implements Runnable {
                Bitmap bitmap;
                PhotoToLoad photoToLoad;
  
                public BitmapDisplayer(Bitmap b, PhotoToLoad p) {
                        bitmap = b;
                        photoToLoad = p;
                }
  
                public void run() {
                        if (imageViewReused(photoToLoad))
                                return;
                        if (bitmap != null)
                                photoToLoad.imageView.setImageBitmap(bitmap);
          
                }
        }
  
        public void clearCache() {
                memoryCache.clear();
                fileCache.clear();
        }
  
        public static void CopyStream(InputStream is, OutputStream os) {
                final int buffer_size = 1024;
                try {
                        byte\[\] bytes = new byte\[buffer_size\];
                        for (;;) {
                                int count = is.read(bytes, 0, buffer_size);
                                if (count == -1)
                                        break;
                                os.write(bytes, 0, count);
                        }
                } catch (Exception ex) {
                        Log.e("", "CopyStream catch Exception...");
                }
        }
}

先从内存中加载,没有则开启线程从SD卡或网络中获取,这里注意从 SD卡获取图片是放在子线程里执行的,否则快速滑屏的话会不够流畅,这是优化一。于此同时,在adapter里有个busy变量,表示listview是 否处于滑动状态,如果是滑动状态则仅从内存中获取图片,没有的话无需再开启线程去外存或网络获取图片,这是优化二。ImageLoader里的线程使用了 线程池,从而避免了过多线程频繁创建和销毁,有的童鞋每次总是new一个线程去执行这是非常不可取的,好一点的用的AsyncTask类,其实内部也是用 到了线程池。在从网络获取图片时,先是将其保存到sd卡,然后再加载到内存,这么做的好处是在加载到内存时可以做个压缩处理,以减少图片所占内存,这是优 化三。

而 图片错位问题的本质源于我们的listview使用了缓存convertView,假设一种场景,一个listview一屏显示九个item,那么在拉出 第十个item的时候,事实上该item是重复使用了第一个item,也就是说在第一个item从网络中下载图片并最终要显示的时候其实该item已经不 在当前显示区域内了,此时显示的后果将是在可能在第十个item上输出图像,这就导致了图片错位的问题。所以解决之道在于可见则显示,不可见则不显示。在 ImageLoader里有个imageViews的map对象,就是用于保存当前显示区域图像对应的url集,在显示前判断处理一下即可。

下面再说下内存缓冲机制,本例采用的是LRU算法,先看看MemoryCache的实现

public class MemoryCache {
  
        private static final String TAG = "MemoryCache";
        // 放入缓存时是个同步操作
        // LinkedHashMap构造方法的最后一个参数true代表这个map里的元素将按照最近使用次数由少到多排列,即LRU
        // 这样的好处是如果要将缓存中的元素替换,则先遍历出最近最少使用的元素来替换以提高效率
        private Map<String, Bitmap> cache = Collections
                        .synchronizedMap(new LinkedHashMap<String, Bitmap>(10, 1.5f, true));
        // 缓存中图片所占用的字节,初始0,将通过此变量严格控制缓存所占用的堆内存
        private long size = 0;// current allocated size
        // 缓存只能占用的最大堆内存
        private long limit = 1000000;// max memory in bytes
  
        public MemoryCache() {
                // use 25% of available heap size
                setLimit(Runtime.getRuntime().maxMemory() / 10);
        }
  
        public void setLimit(long new_limit) {
                limit = new_limit;
                Log.i(TAG, "MemoryCache will use up to " + limit / 1024. / 1024. + "MB");
        }
  
        public Bitmap get(String id) {
                try {
                        if (!cache.containsKey(id))
                                return null;
                        return cache.get(id);
                } catch (NullPointerException ex) {
                        return null;
                }
        }
  
        public void put(String id, Bitmap bitmap) {
                try {
                        if (cache.containsKey(id))
                                size -= getSizeInBytes(cache.get(id));
                        cache.put(id, bitmap);
                        size += getSizeInBytes(bitmap);
                        checkSize();
                } catch (Throwable th) {
                        th.printStackTrace();
                }
        }
  
        /**
         * 严格控制堆内存,如果超过将首先替换最近最少使用的那个图片缓存
         * 
         */
        private void checkSize() {
                Log.i(TAG, "cache size=" + size + " length=" + cache.size());
                if (size > limit) {
                        // 先遍历最近最少使用的元素
                        Iterator<Entry<String, Bitmap>> iter = cache.entrySet().iterator();
                        while (iter.hasNext()) {
                                Entry<String, Bitmap> entry = iter.next();
                                size -= getSizeInBytes(entry.getValue());
                                iter.remove();
                                if (size <= limit)
                                        break;
                        }
                        Log.i(TAG, "Clean cache. New size " + cache.size());
                }
        }
  
        public void clear() {
                cache.clear();
        }
  
        /**
         * 图片占用的内存
         * 
         * <a href="\\"http://www.jcodecraeer.com/\\"" target="\\"_blank\\"">@Param</a> bitmap
         * 
         * @return
         */
        long getSizeInBytes(Bitmap bitmap) {
                if (bitmap == null)
                        return 0;
                return bitmap.getRowBytes() * bitmap.getHeight();
        }
}

首 先限制内存图片缓冲的堆内存大小,每次有图片往缓存里加时判断是否超过限制大小,超过的话就从中取出最少使用的图片并将其移除,当然这里如果不采用这种方 式,换做软引用也是可行的,二者目的皆是最大程度的利用已存在于内存中的图片缓存,避免重复制造垃圾增加GC负担,OOM溢出往往皆因内存瞬时大量增加而 垃圾回收不及时造成的。只不过二者区别在于LinkedHashMap里的图片缓存在没有移除出去之前是不会被GC回收的,而SoftReference 里的图片缓存在没有其他引用保存时随时都会被GC回收。所以在使用LinkedHashMap这种LRU算法缓存更有利于图片的有效命中,当然二者配合使 用的话效果更佳,即从LinkedHashMap里移除出的缓存放到SoftReference里,这就是内存的二级缓存,有兴趣的童鞋不凡一试。

下面附上工程链接:

LazyLoaderDemo.zip(937.26 KB)