LruCache源码解析

本文转载自他人,为本人备忘学习所用,原文链接

上一篇分析了LinkedHashMap源码,这个Map集合除了拥有HashMap的大部分特性之外,还拥有链表的特点,即可以保持遍历顺序与插入顺序一致。另外,当我们将accessOrder设置为true时,可以使遍历顺序和访问顺序一致,其内部双向链表将会按照近期最少访问到近期最多访问的顺序排列Entry对象,这可以用来做缓存。

这篇文章分析的LruCache并不是jdk中的类,而是来自安卓,熟悉安卓内存缓存的必然对这个类不陌生。
LruCache内部维护的就是一个LinkedHashMap。
下面开始分析LruCache。
注:下面LruCache源码来自support.v4包。
首先是这个类的成员变量:

1
2
3
4
5
6
7
8
9
private final LinkedHashMap<K, V> map;
/** Size of this cache in units. Not necessarily the number of elements. */
private int size;//当前大小
private int maxSize;//最大容量
private int putCount;//put次数
private int createCount;//create次数
private int evictionCount;//回收次数
private int hitCount;//命中次数
private int missCount;//丢失次数

LinkedHashMap的初始化放在构造器中:

1
2
3
4
5
6
7
public LruCache(int maxSize) {
if (maxSize <= 0) {
throw new IllegalArgumentException("maxSize <= 0");
}
this.maxSize = maxSize;
this.map = new LinkedHashMap<K, V>(0, 0.75f, true);
}

这里将LinkedHashMap的accessOrder设置为true。
接下来看两个最重要的方法,put和get。首先是put方法:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
public final V put(K key, V value) {
if (key == null || value == null) {//键值不允许为空
throw new NullPointerException("key == null || value == null");
}
V previous;
synchronized (this) {//线程安全
putCount++;
size += safeSizeOf(key, value);
previous = map.put(key, value);
if (previous != null) {//之前已经插入过相同的key
size -= safeSizeOf(key, previous);//那么减去该entry的容量,因为发生覆盖
}
}
if (previous != null) {
entryRemoved(false, key, previous, value);//这个方法默认空实现
}
trimToSize(maxSize);//若容量超过maxsize,将会删除最近很少访问的entry
return previous;
}

put方法无非就是调用LinkedHashMap的put方法,但是这里在调用LinkedHashMap的put方法之前,判断了key和value是否为空,也就是说LruCache不允许空键值。除此之外,put操作被加锁了,所以是线程安全的!
既然是缓存,那么必然能够动态删除一些不常用的键值对,这个工作是由trimToSize方法完成的:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23

public void trimToSize(int maxSize) {
while (true) {//不断删除linkedHashMap头部entry,也就是最近最少访问的条目,直到size小于最大容量
K key;
V value;
synchronized (this) {//线程安全
if (size < 0 || (map.isEmpty() && size != 0)) {
throw new IllegalStateException(getClass().getName()
+ ".sizeOf() is reporting inconsistent results!");
}
if (size <= maxSize || map.isEmpty()) {//直到容量小于最大容量为止
break;
}
Map.Entry<K, V> toEvict = map.entrySet().iterator().next();//指向链表头
key = toEvict.getKey();
value = toEvict.getValue();
map.remove(key);//删除最少访问的entry
size -= safeSizeOf(key, value);
evictionCount++;
}
entryRemoved(true, key, value, null);
}
}

这个方法不断循环删除链表首部元素,也就是最近最少访问的元素,直到容量不超过预先定义的最大值为止。
注:LruCache在android.util包中也有一个LruCache类,但是我发现这个类的trimToSize方法是错误的:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
private void trimToSize(int maxSize) {
while (true) {
K key;
V value;
synchronized (this) {
if (size < 0 || (map.isEmpty() && size != 0)) {
throw new IllegalStateException(getClass().getName()
+ ".sizeOf() is reporting inconsistent results!");
}
if (size <= maxSize) {
break;
}

Map.Entry<K, V> toEvict = null;
for (Map.Entry<K, V> entry : map.entrySet()) {
toEvict = entry;
}

if (toEvict == null) {
break;
}
key = toEvict.getKey();
value = toEvict.getValue();
map.remove(key);
size -= safeSizeOf(key, value);
evictionCount++;
}
entryRemoved(true, key, value, null);
}
}

这里的代码将会循环删除链表尾部,也就是最近访问最多的元素,这是不正确的!所以大家在做内存缓存的时候一定要注意,看trimToSize方法是否有问题。

接下来是get方法:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
public final V get(K key) {
if (key == null) {//不允许空键
throw new NullPointerException("key == null");
}
V mapValue;
synchronized (this) {//线程安全
mapValue = map.get(key);//调用LinkedHashMap的get方法
if (mapValue != null) {
hitCount++;//命中次数加1
return mapValue;//返回value
}
missCount++;//未命中
}

V createdValue = create(key);//默认返回为false
if (createdValue == null) {
return null;
}
synchronized (this) {
createCount++;//如果创建成功,那么create次数加1
mapValue = map.put(key, createdValue);//放到哈希表中
if (mapValue != null) {
// There was a conflict so undo that last put
map.put(key, mapValue);
} else {
size += safeSizeOf(key, createdValue);
}
}
if (mapValue != null) {
entryRemoved(false, key, createdValue, mapValue);
return mapValue;
} else {
trimToSize(maxSize);
return createdValue;
}
}

get方法即根据key在LinkedHashMap中寻找对应的value,此方法也是线程安全的。

以上就是LruCache最重要的部分,下面再看下其他方法:
remove:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
public final V remove(K key) {
if (key == null) {
throw new NullPointerException("key == null");
}
V previous;
synchronized (this) {
previous = map.remove(key);//调用LinkedHashMap的remove方法
if (previous != null) {
size -= safeSizeOf(key, previous);
}
}
if (previous != null) {
entryRemoved(false, key, previous, null);
}
return previous;//返回value
}

sizeof:这个方法用于计算每个条目的大小,子类必须得复写这个类。

1
2
3
protected int sizeOf(K key, V value) {//用于计算每个条目的大小
return 1;
}

snapshot方法,返回当前缓存中所有的条目集合

1
2
3
public synchronized final Map<K, V> snapshot() {
return new LinkedHashMap<K, V>(map);
}

总结:
1.LruCache封装了LinkedHashMap,提供了LRU缓存的功能;
2.LruCache通过trimToSize方法自动删除最近最少访问的键值对;
3.LruCache不允许空键值;
4.LruCache线程安全;
5.继承LruCache时,必须要复写sizeof方法,用于计算每个条目的大小。

当前网速较慢或者你使用的浏览器不支持博客特定功能,请尝试刷新或换用Chrome、Firefox等现代浏览器