当前位置：首页 > news >正文

LRU缓存淘汰算法Java实现

news 2025/10/22 12:41:29

在这里插入图片描述

LRU的核心思想是当缓存空间满了,优先淘汰最久未被使用的数据。

这种思想其实在生活中也有，比如想象一下你的书桌只能放5本书。每次看完一本书,你就把它放在最右边。当需要放新书但桌面满了时,你会扔掉最左边那本(最久没看的)。这其实就是LRU的思想!

代码逻辑

1. addNode

public void addNode(DoubleNode newNode) {if (head == null) {// 链表为空,新节点既是头也是尾head = newNode;tail = newNode;} else {// 追加到尾部tail.next = newNode;newNode.last = tail;tail = newNode;}
}

2. moveNodeToTail

每次访问一个节点,都要把它移到尾部,表示"最近刚用过"。

public void moveNodeToTail(DoubleNode node) {if (tail == node) return;  // 已经在尾部// 从原位置摘除节点if (head == node) {head = node.next;head.last = null;} else {node.last.next = node.next;node.next.last = node.last;}// 追加到尾部node.last = tail;node.next = null;tail.next = node;tail = node;
}

3. removeHead

当缓存满了需要淘汰数据时,删除头部(最久未使用的)节点。

public DoubleNode removeHead() {if (head == null) return null;DoubleNode ans = head;if (head == tail) {// 只有一个节点head = null;tail = null;} else {head = ans.next;ans.next = null;head.last = null;}return ans;
}

4.get - 查询数据

public int get(int key) {if (keyNodeMap.containsKey(key)) {DoubleNode ans = keyNodeMap.get(key);nodeList.moveNodeToTail(ans);  // 标记为最近使用return ans.val;}return -1;
}

5.put - 插入或更新数据

public void put(int key, int value) {if (keyNodeMap.containsKey(key)) {// key已存在,更新valueDoubleNode node = keyNodeMap.get(key);node.val = value;nodeList.moveNodeToTail(node);  // 移到尾部} else {// key不存在,插入新节点if (keyNodeMap.size() == capacity) {// 缓存已满,先淘汰最久未使用的DoubleNode removed = nodeList.removeHead();keyNodeMap.remove(removed.key);}DoubleNode newNode = new DoubleNode(key, value);keyNodeMap.put(key, newNode);nodeList.addNode(newNode);}
}

完整代码

class LRUCache{class DoubleNode {public int key;      // 键public int val;      // 值public DoubleNode last;  // 前驱节点public DoubleNode next;  // 后继节点public DoubleNode(int k, int v) {key = k;val = v;}}class DoubleList{private DoubleNode head;private DoubleNode tail;public DoubleList(){head = null;tail = null;}public void addNode(DoubleNode newNode) {if (head == null) {// 链表为空,新节点既是头也是尾head = newNode;tail = newNode;} else {// 追加到尾部tail.next = newNode;newNode.last = tail;tail = newNode;}}public void moveNodeToTail(DoubleNode node) {if (tail == node) return;  // 已经在尾部// 第一步: 从原位置摘除节点if (head == node) {head = node.next;head.last = null;} else {node.last.next = node.next;node.next.last = node.last;}// 第二步: 追加到尾部node.last = tail;node.next = null;tail.next = node;tail = node;}public DoubleNode removeHead() {if (head == null) return null;DoubleNode ans = head;if (head == tail) {// 只有一个节点head = null;tail = null;} else {head = ans.next;ans.next = null;head.last = null;}return ans;}}private HashMap<Integer, DoubleNode> keyNodeMap;private DoubleList nodeList;private final int capacity;public LRUCache(int cap){keyNodeMap = new HashMap<>();nodeList = new DoubleList();capacity = cap;}public int get(int key) {if (keyNodeMap.containsKey(key)) {DoubleNode ans = keyNodeMap.get(key);nodeList.moveNodeToTail(ans);  // 标记为最近使用return ans.val;}return -1;}public void put(int key, int value) {if (keyNodeMap.containsKey(key)) {// key已存在,更新valueDoubleNode node = keyNodeMap.get(key);node.val = value;nodeList.moveNodeToTail(node);  // 移到尾部} else {// key不存在,插入新节点if (keyNodeMap.size() == capacity) {// 缓存已满,先淘汰最久未使用的DoubleNode removed = nodeList.removeHead();keyNodeMap.remove(removed.key);}DoubleNode newNode = new DoubleNode(key, value);keyNodeMap.put(key, newNode);nodeList.addNode(newNode);}}}