从源码看Android常用的数据结构 ( 七, SDK28下的HashMap )相关衔接此系列文章放在了我的专栏里,欢迎

前言

Java1.8 和 Android SDK28 对 HashMap 做了大量优化, 与之前版本的主要区别在于以下两点:

hash碰撞超过8时把原链表转化为红黑树, 提高查询效率;
resize时对链表节点的重定位算法无需重hash计算, 而是通过判断&操作的高位来判断是放在低位cap还是高位cap;

新版本的HashMap结合了链表和红黑树的优化, 以达到最优的时间空间效率平衡, 这种精雕细琢的精神值得好好学习!

下面详细说明

先看关键参数

    /**
     * The table, initialized on first use, and resized as
     * necessary. When allocated, length is always a power of two.
     * 
     * 数据存储集合, 必要时会resize, 重分配后大小为2的n次方.
     * 添加transient关键字以屏蔽默认serializable, 自主实现read/write.
     */
    transient Node<K,V>[] table;

    /**
     * Holds cached entrySet(). Note that AbstractMap fields are used
     * for keySet() and values().
     * 
     * 缓存entrySet.
     */
    transient Set<Map.Entry<K,V>> entrySet;

    /**
     * The number of key-value mappings contained in this map.
     * 
     * map内实际存储的元素数目
     */
    transient int size;

    /**
     * The number of times this HashMap has been structurally modified
     * Structural modifications are those that change the number of mappings in
     * the HashMap or otherwise modify its internal structure (e.g.,
     * rehash).  This field is used to make iterators on Collection-views of
     * the HashMap fail-fast.  (See ConcurrentModificationException).
     * 
     * HashMap结构修改的次数. 结构修改是指元素数量改变或者以其他方式修改其内部
     * 结构等操作(例如rehash). 
     * 执行结构修改时modCount++, 在执行遍历等非原子操作前缓存modCount, 遍历过程
     * 中或之后如果modCount与原缓存值不同, 说明出现了并发修改, 抛出异常
     */
    transient int modCount;

    /**
     * The next size value at which to resize (capacity * load factor).
     * 阈值, 触发resize的临界容量, 初始化时为最小的超过容量的2的指数, 后续更改为
     * newCap * loadFactor
     */
    int threshold;

    /**
     * The load factor for the hash table.
     * 加载因子, 用于判断threshold
     */
    final float loadFactor;

核心方法Get/Put/Remove

GET方法

    public V get(Object key) {
        Node<K,V> e;
        return (e = getNode(hash(key), key)) == null ? null : e.value;
    }
    
    /**
     * 计算key的hash值, 使用hashcode的高16位异或低16位. 看注释的意思是可能会发生碰撞,
     * 但是考虑到碰撞的元素使用较高效率的tree来存储, 所以hash算法采用了较轻量的方式.
     */
    static final int hash(Object key) {
        int h;
        return (key == null) ? 0 : (h = key.hashCode()) ^ (h >>> 16);
    }
    
    /**
     * Implements Map.get and related methods
     *
     * @param hash hash for key
     * @param key  the key
     * @return the node, or null if none
     */
    final Node<K, V> getNode(int hash, Object key) {
        Node<K, V>[] tab;
        Node<K, V> first, e;
        int n;
        K k;
        // 如果(n-1)&hash位置存在元素
        if ((tab = table) != null && (n = tab.length) > 0 && (first = tab[(n - 1) & hash]) != null) {
            // 如果hash相等, key相等, 返回第一个
            if (first.hash == hash && ((k = first.key) == key || (key != null && key.equals(k))))
                return first;
            // 不是第一个的话遍历链表或者树
            if ((e = first.next) != null) {
                // 查询树
                if (first instanceof TreeNode)
                    return ((TreeNode<K, V>) first).getTreeNode(hash, key);
                // 遍历链表
                do {
                    if (e.hash == hash && ((k = e.key) == key || (key != null && key.equals(k))))
                        return e;
                } while ((e = e.next) != null);
            }
        }
        return null;
    }

Put相关

    public V put(K key, V value) {
        return putVal(hash(key), key, value, false, true);
    }
    
    /**
     * Implements Map.put and related methods
     *
     * @param hash         hash for key
     * @param key          the key
     * @param value        the value to put
     * @param onlyIfAbsent if true, don't change existing value
     * @param evict        if false, the table is in creation mode.
     * @return previous value, or null if none
     */
    final V putVal(int hash, K key, V value, boolean onlyIfAbsent, boolean evict) {
        Node<K, V>[] tab;
        Node<K, V> parent;
        int tabLength, position;
        // 如果还未初始化, 调用resize来初始化一下, resize见下方方法
        if ((tab = table) == null || (tabLength = tab.length) == 0)
            tabLength = (tab = resize()).length;
        // 如果该位置未存有元素, 第一位新建node存入
        if ((parent = tab[position = (tabLength - 1) & hash]) == null)
            tab[position] = newNode(hash, key, value, null);
            // 如果该位置已有元素, 分为三种情况:
        else {
            Node<K, V> exitMapping;
            K k;
            // 该位置头节点就是对应节点
            if (parent.hash == hash && ((k = parent.key) == key || (key != null && key.equals(k))))
                exitMapping = parent;
                // 如果是树, 去树内查找
            else if (parent instanceof TreeNode)
                exitMapping = ((TreeNode<K, V>) parent).putTreeVal(this, tab, hash, key, value);
                // 链表
            else {
                for (int binCount = 0; ; ++binCount) {
                    // 如果直到链表末尾也没找到已存在的点, 则直接插入到末尾, 退出循环
                    if ((exitMapping = parent.next) == null) {
                        parent.next = newNode(hash, key, value, null);
                        // 如果碰撞次数超过了转化为tree的阈值(8), 转化为红黑树
                        if (binCount >= TREEIFY_THRESHOLD - 1) // -1 for 1st
                            treeifyBin(tab, hash);
                        break;
                    }
                    // 找到了对应节点
                    if (exitMapping.hash == hash && ((k = exitMapping.key) == key || (key != null && key.equals(k))))
                        break;
                    parent = exitMapping;
                }
            }
            // 如果已存在节点, 在这里做value替换; 不存在对应节点的情况在前面for循环中已经插入了新节点
            if (exitMapping != null) { // existing mapping for key
                V oldValue = exitMapping.value;
                // 如果允许替换值, 替换
                if (!onlyIfAbsent || oldValue == null)
                    exitMapping.value = value;
                // 用于linkedHashMap
                afterNodeAccess(exitMapping);
                return oldValue;
            }
        }
        // 发生了结构化修改
        ++modCount;
        // 如果大小超过了tab阈值, resize为double
        if (++size > threshold)
            resize();
        // 用于linkedHashMap
        afterNodeInsertion(evict);
        return null;
    }
    
    /**
     * Initializes or doubles table size.  If null, allocates in
     * accord with initial capacity target held in field threshold.
     * Otherwise, because we are using power-of-two expansion, the
     * elements from each bin must either stay at same index, or move
     * with a power of two offset in the new table.
     * 某元素只可能保持在当前位置或者2次方位置
     *
     * @return the table
     */
    final Node<K, V>[] resize() {
        Node<K, V>[] oldTab = table;
        // 旧表的容量
        int oldCap = (oldTab == null) ? 0 : oldTab.length;
        // 旧表的阈值
        int oldThr = threshold;
        int newCap, newThr = 0;
        // 如果有旧值, 不是初始化
        if (oldCap > 0) {
            // 如果旧容量超过了2的30次方, 则阈值*2即int的最大值, 也就是2的31次方, 也不重新分配bin了, resize直接完成
            if (oldCap >= MAXIMUM_CAPACITY) {
                threshold = Integer.MAX_VALUE;
                return oldTab;
            }
            // 新容量首先变为两倍, 如果旧容量超过了默认容量16, 新阈值double
            else if ((newCap = oldCap << 1) < MAXIMUM_CAPACITY && oldCap >= DEFAULT_INITIAL_CAPACITY)
                newThr = oldThr << 1; // double threshold
        }
        // 以下两种情况均是初始化
        else if (oldThr > 0) {
            // 旧容量为0, 阈值不为0, 就把新的容量被设置为旧阈值
            newCap = oldThr;
        } else {
            // 容量阈值均使用默认值
            newCap = DEFAULT_INITIAL_CAPACITY;
            newThr = (int) (DEFAULT_LOAD_FACTOR * DEFAULT_INITIAL_CAPACITY);
        }
        // 上面设置好了新的容量, 如果新阈值还是0的话(主要是针对上面三种情况的第二种), 设置为newCap * 重载因子loadFactor
        if (newThr == 0) {
            float ft = (float) newCap * loadFactor;
            newThr = (newCap < MAXIMUM_CAPACITY && ft < (float) MAXIMUM_CAPACITY ? (int) ft : Integer.MAX_VALUE);
        }
        threshold = newThr;

        // 通过上面的处理, 新的容量和阈值已经配置好了, 下面的逻辑是把旧表里的节点放入新表中
        Node<K, V>[] newTab = (Node<K, V>[]) new Node[newCap];
        table = newTab;
        if (oldTab != null) {
            for (int j = 0; j < oldCap; ++j) {
                Node<K, V> header; // 某位置的链表头节点
                if ((header = oldTab[j]) != null) {
                    oldTab[j] = null;
                    // 如果只有头节点, 就重hash放入新表中
                    if (header.next == null) {
                        newTab[header.hash & (newCap - 1)] = header;
                    }
                    // 树形结构的修剪
                    else if (header instanceof TreeNode) {
                        // 把树切为低位和高位两部分, 如果某一部分size小于6了, 转化为链表
                        ((TreeNode<K, V>) header).split(this, newTab, j, oldCap);
                    }
                    // 链表节点切分(!精髓!), (header.hash & oldCap) == 0 的节点放在前oldCap位, 其余的放在后半部分
                    // 原来的hash计算是 header.hash & (oldCap- 1), doubleSize后计算 header.hash & oldCap,
                    //
                    // 因为oldCap肯定是2的幂次, 所以只有一个高位1, 而计算 hash&(oldCap-1)时忽略了header.hash的最高位, 所以这一堆链表节点才会在这一个链表里
                    // 而header.hash & oldCap的结果反应的就是hash值的对应高位是否为1, 如果为1则插入高位, 为0说明本来就该在低位
                    else {
                        // 放在新table低位的链表头尾
                        Node<K, V> loHead = null, loTail = null;
                        // 放在新table高位的链表头尾
                        Node<K, V> hiHead = null, hiTail = null;
                        Node<K, V> next;
                        do {
                            next = header.next;
                            // 无需重hash的精髓!!
                            if ((header.hash & oldCap) == 0) {
                                if (loTail == null)
                                    loHead = header;
                                else
                                    loTail.next = header;
                                loTail = header;
                            } else {
                                if (hiTail == null)
                                    hiHead = header;
                                else
                                    hiTail.next = header;
                                hiTail = header;
                            }
                        } while ((header = next) != null);
                        // 低尾节点不为null, 把低头节点放在新table的低位
                        if (loTail != null) {
                            loTail.next = null;
                            newTab[j] = loHead;
                        }
                        // 高尾节点不为null, 把高头节点放在新table的高位
                        if (hiTail != null) {
                            hiTail.next = null;
                            newTab[j + oldCap] = hiHead;
                        }
                    }
                }
            }
        }
        return newTab;
    }  

    /**
    * 把链表转化为树
    */ 
    final void treeifyBin(Node<K, V>[] tab, int hash) {
        int tabLength, index;
        Node<K, V> e; // 遍历node
        // 如果tab太小, 直接调用resize, 可能就不用转化为tree了
        if (tab == null || (tabLength = tab.length) < MIN_TREEIFY_CAPACITY) {
            resize();
        }
        // 正常tableSize, 需要转化. 首先找到对应位置
        else if ((e = tab[index = (tabLength - 1) & hash]) != null) {
            TreeNode<K, V> header = null, tail = null;
            do {
                // 依次转化为TreeNode
                TreeNode<K, V> p = replacementTreeNode(e, null);
                if (tail == null)
                    header = p;
                else {
                    p.prev = tail;
                    tail.next = p;
                }
                tail = p;
            } while ((e = e.next) != null);
            if ((tab[index] = header) != null)
                // 格式化treeNode, 令其满足红黑树规则
                header.treeify(tab);
        }
    }

Remove相关

    /**
     * Removes the mapping for the specified key from this map if present.
     *
     * @param key key whose mapping is to be removed from the map
     * @return the previous value associated with <tt>key</tt>, or
     * <tt>null</tt> if there was no mapping for <tt>key</tt>.
     * (A <tt>null</tt> return can also indicate that the map
     * previously associated <tt>null</tt> with <tt>key</tt>.)
     */
    public V remove(Object key) {
        Node<K, V> e;
        return (e = removeNode(hash(key), key, null, false, true)) == null ? null : e.value;
    }

    /**
     * Implements Map.remove and related methods
     *
     * @param hash       hash for key
     * @param key        the key
     * @param value      the value to match if matchValue, else ignored
     * @param matchValue if true only remove if value is equal
     * @param movable    if false do not move other nodes while removing
     * @return the node, or null if none
     */
    final Node<K, V> removeNode(int hash, Object key, Object value, boolean matchValue, boolean movable) {
        Node<K, V>[] tab;
        Node<K, V> header; // 对应位置的头节点
        int n, index;
        // 如果有该hash对应的节点
        if ((tab = table) != null && (n = tab.length) > 0 && (header = tab[index = (n - 1) & hash]) != null) {
            // 待删除的点
            Node<K, V> node = null;
            // 用于遍历的node
            Node<K, V> e;
            K k;
            V v;
            // 恰好是第一位
            if (header.hash == hash && ((k = header.key) == key || (key != null && key.equals(k))))
                node = header;
            else if ((e = header.next) != null) {
                // 树结构查找
                if (header instanceof TreeNode)
                    node = ((TreeNode<K, V>) header).getTreeNode(hash, key);
                else {
                    // 链表遍历
                    do {
                        if (e.hash == hash && ((k = e.key) == key || (key != null && key.equals(k)))) {
                            node = e;
                            break;
                        }
                        header = e;
                    } while ((e = e.next) != null);
                }
            }
            // 找到了这个点
            if (node != null && (!matchValue || (v = node.value) == value || (value != null && value.equals(v)))) {
                // 调用红黑树的remove
                if (node instanceof TreeNode)
                    ((TreeNode<K, V>) node).removeTreeNode(this, tab, movable);
                // 以下两种情况均是链表前移
                else if (node == header)
                    tab[index] = node.next;
                else
                    header.next = node.next;
                ++modCount;
                --size;
                afterNodeRemoval(node);
                return node;
            }
        }
        return null;
    }