了解HashMap数据结构，超详细！

{"type":"doc","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"写在前面"}]},{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"小伙伴儿们，大家好！今天来学习HashMap相关内容，作为面试必问的知识点，来深入了解一波！"}]}]},{"type":"heading","attrs":{"align":null,"level":5},"content":[{"type":"text","text":"思维导图："}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/d1/d1a4c006620f1e619dfa7e6a7cfe6ce5.png","alt":null,"title":null,"style":null,"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"学习框架图"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"1，HashMap集合简介"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"HashMap基于哈希表的Map接口实现，是以"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong"}],"text":"key-value"},{"type":"text","text":"存储形式存在，即主要用来存放键值对。HashMap的实现不是同步的，这意味着它不是线程安全的。它的key、value都可以为null。此外，HashMap中的映射不是有序的。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"JDK1.8之前的HashMap由数组+链表组成的，数组是HashMap的主体，链表则是主要为了节解决"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong"}],"text":"哈希碰撞"},{"type":"text","text":"(两个对象调用的hashCode方法计算的哈希码值一致导致计算的数组索引值相同)而存在的（“拉链法”解决冲突）。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"JDK1.8之后在解决哈希冲突时有了较大的变化，当"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong"}],"text":"链表长度大于阈值"},{"type":"text","text":"（或者红黑树的边界值，默认为8）并且当前"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong"}],"text":"数组的长度大于64"},{"type":"text","text":"时，此时此索引位置上的所有数据改为使用红黑树存储。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"数组里面都是key-value的实例，在JDK1.8之前叫做Entry，在JDK1.8之后叫做Node。"}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/1c/1c7f5d4f4839b72908faba89c2ab56d9.png","alt":null,"title":null,"style":null,"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"key-value实例"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"由于它的key、value都为null，所以在插入的时候会根据key的hash去计算一个index索引的值。计算索引的方法如下："}]},{"type":"codeblock","attrs":{"lang":"java"},"content":[{"type":"text","text":"/**\n * 根据key求index的过程\n * 1,先用key求出hash值\n */\nstatic final int hash(Object key) {\n int h;\n return (key == null) ? 0 : (h = key.hashCode()) ^ (h >>> 16);\n}\n//2,再用公式index = (n - 1) & hash（n是数组长度）\nint hash=hash(key);\nindex=(n-1)&hash;\n"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"这里的Hash算法本质上就是三步：取key的hashCode值、高位运算、取模运算。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"这样的话比如说put(\"A\",王炸)，插入了key为\"A\"的元素，这时候通过上述公式计算出插入的位置index，若index为3则结果如下（即hash(\"A\")=3）："}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/7c/7c21871ec7ee8c9f758992d568b6cd3f.png","alt":null,"title":null,"style":null,"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"那么，HashMap中的链表又是干什么用的呢？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"大家都知道数组的长度是有限的，"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong"}],"text":"在有限的长度里面使用哈希函数计算index的值时，很有可能插入的k值不同，但所产生的hash是相同的"},{"type":"text","text":"（也叫做哈希碰撞），这也就是哈希函数存在一定的概率性。就像上面的K值为A的元素，如果再次插入一个K值为a的元素，很有可能所产生的index值也为3，也就是即hash(\"a\")=3；那这就形成了链表，这种解决哈希碰撞的方法也叫做拉链法。"}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/06/06c43bf7aa4b497d858503e7c76c4ba7.png","alt":null,"title":null,"style":null,"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"当这个链表长度大于阈值8并且数组长度大于64则进行将链表变为红黑树。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"补充："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"将链表转换成红黑树前会判断，"},{"type":"text","marks":[{"type":"strong"}],"text":"如果阈值大于8，但是数组长度小64，此时并不会将链表变为红黑树。而是选择进行数组扩容。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"这样做的目的是因为数组比较小，尽量避开红黑树结构，这种情况下变为红黑树结构，反而会降低效率，因为红黑树需要进行左旋，右旋，变色这些操作来保持平衡。同事数组长度小于64时，搜索时间相对快一些。所以综上所述为了提高性能和减少搜索时间，底层在阈值大于8并且数组长度大于64时，链表才转换为红黑树。具体可以参考treeifyBin方法。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"当然虽然增了红黑树作为底层数据结构，结构变得复杂了，但是阈值大于8并且数组长度大于64时，链表转换为红黑树时，效率也变得更高效。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"特点："}]},{"type":"numberedlist","attrs":{"start":null,"normalizeStart":1},"content":[{"type":"listitem","content":[{"type":"paragraph","attrs":{"indent":0,"number":1,"align":null,"origin":null},"content":[{"type":"text","text":"存取无序的"}]}]},{"type":"listitem","content":[{"type":"paragraph","attrs":{"indent":0,"number":2,"align":null,"origin":null},"content":[{"type":"text","text":"键和值位置都可以是null，但是键位置只能是一个null"}]}]},{"type":"listitem","content":[{"type":"paragraph","attrs":{"indent":0,"number":3,"align":null,"origin":null},"content":[{"type":"text","text":"键位置是唯一的，底层的数据结构控制键的"}]}]},{"type":"listitem","content":[{"type":"paragraph","attrs":{"indent":0,"number":4,"align":null,"origin":null},"content":[{"type":"text","text":"jdk1.8前数据结构是：链表 + 数组  jdk1.8之后是 ：链表 + 数组  + 红黑树"}]}]},{"type":"listitem","content":[{"type":"paragraph","attrs":{"indent":0,"number":5,"align":null,"origin":null},"content":[{"type":"text","text":"阈值(边界值) > 8 并且数组长度大于64，才将链表转换为红黑树，变为红黑树的目的是为了高效的查询。"}]}]}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"2，HsahMap底层数据结构"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"2.1，HashMap存储数据的过程"}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/33/3372df90aec7531aa0993b6c5fb5ce3e.png","alt":null,"title":null,"style":null,"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"每一个Node结点都包含"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong"}],"text":"键值对的key，value"},{"type":"text","text":"还有计算出来的"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong"}],"text":"hash值"},{"type":"text","text":"，还保存着下一个"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}}],"text":" "},{"type":"text","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong"}],"text":"Node 的引用 next"},{"type":"text","text":"（如果没有下一个 Node，next = null），来看看Node的源码："}]},{"type":"codeblock","attrs":{"lang":"java"},"content":[{"type":"text","text":"static class Node implements Map.Entry {\n final int hash;\n final K key;\n V value;\n Node next;\n ...\n }"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"HashMap存储数据需要用到put()方法，关于这些方法的详解，我们下节再说，这里简要说一下；"}]},{"type":"codeblock","attrs":{"lang":"java"},"content":[{"type":"text","text":"public static void main(String[] args) {\n HashMap hmap=new HashMap<>();\n hmap.put(\"斑\",55);\n hmap.put(\"镜\",63);\n hmap.put(\"带土\",25);\n hmap.put(\"鼬\",9);\n hmap.put(\"佐助\",43);\n hmap.put(\"斑\",88);\n System.out.println(hmap);\n }"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"当创建HashMap集合对象的时候，在jdk1.8之前，构造方法中会创建很多"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong"}],"text":"长度是16的Entry[] table"},{"type":"text","text":"用来存储键值对数据的。在jdk1.8之后不是在HashMap的构造方法底层创建数组了，是在"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong"}],"text":"第一次调用put方法"},{"type":"text","text":"时创建的数组，"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong"}],"text":"Node[] table"},{"type":"text","text":"用来存储键值对数据的。"}]},{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"比方说我们向哈希表中存储\"斑\"-55的数据，根据K值(\"斑\")调用String类中重写之后的hashCode()方法计算出值（数量级很大），然后结合数组长度采用取余（(n-1)&hash）操作或者其他操作方法来计算出向Node数组中存储数据的空间的索引值。如果计算出来的索引空间没有数据，则直接将\"斑\"-55数据存储到数组中。跟上面的\"A-王炸\"数据差不多。"}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我们回到上方的数组图，如果此时再插入\"A-蘑菇\"元素，那么首先根据Key值(\"A\")调用hashCode()方法结合数组长度计算出索引肯定也是3，此时比较后存储的\"A-蘑菇\"和已经存在的数据\"A-王炸\"的hash值是否相等，如果hash相等，此时发生hash碰撞。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"那么底层会调用\"A\"所属"},{"type":"text","marks":[{"type":"color","attrs":{"color":"#F5222D","name":"red"}},{"type":"strong"}],"text":"类String中的equals方法"},{"type":"text","text":"比较两个key内容是否相等，若相等，则后添加的数据直接覆盖已经存在的Value，也就是\"蘑菇\"直接覆盖\"王炸\"；若不相等，继续向下和其他数据的key进行比较，如果都不相等，则规划出一个节点存储数据。"}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/7d/7dce818b614372db22deffaffcba10c2.png","alt":null,"title":null,"style":null,"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"两个结点key值比较，是否覆盖"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"2.2，哈希碰撞相关的问题"}]},{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"哈希表底层采用何种算法计算hash值？还有哪些算法可以计算出hash值？"}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"底层是采用key的hashCode方法的值结合数组长度进行"},{"type":"text","marks":[{"type":"strong"}],"text":"无符号右移（>>>）"},{"type":"text","text":"、按位异或（^）、按位与（&）计算出索引的"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"还可以采用：平方取中法，取余数，伪随机数法。这三种效率都比较低。而无符号右移16位异或运算效率是最高的。"}]},{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"当两个对象的hashCode相等时会怎么样？"}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"会产生哈希碰撞，若"},{"type":"text","marks":[{"type":"strong"}],"text":"key值内容相同则替换旧的value"},{"type":"text","text":".否则连接到链表后面，"},{"type":"text","marks":[{"type":"strong"}],"text":"链表长度超过阈值8"},{"type":"text","text":"就转换为红黑树存储。"}]},{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"何时发生哈希碰撞和什么是哈希碰撞,如何解决哈希碰撞？"}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"只要"},{"type":"text","marks":[{"type":"strong"}],"text":"两个元素的key计算的哈希值相同"},{"type":"text","text":"就会发生哈希碰撞。jdk8前使用链表解决哈希碰撞。jdk8之后使用链表+红黑树解决哈希碰撞。"}]},{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"如果两个键的hashcode相同，如何存储键值对？"}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"hashcode相同，通过equals比较内容是否相同。"},{"type":"text","marks":[{"type":"strong"}],"text":"相同：则新的value覆盖之前的value 不相同：则将新的键值对添加到哈希表中"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"2.3，红黑树结构"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"当位于一个链表中的元素较多，即hash值相等但是内容不相等的元素较多时，通过key值依次查找的效率较低。而jdk1.8中，哈希表存储采用数组+链表+红黑树实现，当链表长度(阀值)超过 8 时且当前数组的长度 > 64时，将链表转换为红黑树，这样大大减少了查找时间。jdk8在哈希表中"},{"type":"text","marks":[{"type":"strong"}],"text":"引入红黑树的原因只是为了查找效率更高。"}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/2d/2d6bf6020091655bf7dade250e72cbc8.png","alt":null,"title":null,"style":null,"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"红黑树结构"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"JDK 1.8 以前 HashMap 的实现是 数组+链表，即使哈希函数取得再好，也很难达到元素百分百均匀分布。当 HashMap 中有大量的元素都存放到同一个桶中时，这个桶下有一条长长的链表，这个时候 HashMap 就相当于一个单链表，假如单链表有 n 个元素，遍历的时间复杂度就是 O(n)，完全失去了它的优势。针对这种情况，JDK 1.8 中引入了 红黑树（"},{"type":"text","marks":[{"type":"strong"}],"text":"查找时间复杂度为 O(logn)"},{"type":"text","text":"）来优化这个问题。当链表长度很小的时候，即使遍历，速度也非常快，但是当链表长度不断变长，肯定会对查询性能有一定的影响，所以才需要转成树。"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"2.4，存储流程图"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"HashMap存放数据是用的put方法，put 方法内部调用的是 putVal() 方法，所以对 put 方法的分析也是对 putVal 方法的分析，整个过程比较复杂，流程图如下："}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/e4/e40955237c51d77cbddd5d04b53a66f6.png","alt":null,"title":null,"style":null,"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"来看看put()源码："}]},{"type":"codeblock","attrs":{"lang":"java"},"content":[{"type":"text","text":"public V put(K key, V value) {\n //对key的hashCode()做hash，调用的是putVal方法\n return putVal(hash(key), key, value, false, true);\n }\n\n final V putVal(int hash, K key, V value, boolean onlyIfAbsent,\n boolean evict) {\n Node[] tab; Node p; int n, i;\n /*\n 1，tab为空则开始创建，\n 2，(tab = table) == null 表示将空的table赋值给tab,然后判断tab是否等于null，第一次肯定是null\n 3，(n = tab.length) == 0 表示没有为table分配内存\n 4，tab为空，执行代码 n = (tab = resize()).length; 进行扩容。并将初始化好的数组长度赋值给n.\n 5，执行完n = (tab = resize()).length，数组tab每个空间都是null\n */\n \n if ((tab = table) == null || (n = tab.length) == 0)\n //调用resize()方法进行扩容\n n = (tab = resize()).length;\n /*\n 1，i = (n - 1) & hash 表示计算数组的索引赋值给i，即确定元素存放在哪个桶中\n 2，p = tab[i = (n - 1) & hash]表示获取计算出的位置的数据赋值给节点p\n 3，(p = tab[i = (n - 1) & hash]) == null 判断节点位置是否等于null，\n 如果为null，则执行代码：tab[i] = newNode(hash, key, value, null);根据键值对创建新的节点放入该位置的桶中\n 小结：如果当前桶没有哈希碰撞冲突，则直接把键值对插入空间位置\n */ \n if ((p = tab[i = (n - 1) & hash]) == null)\n //节点位置为null，则直接进行插入操作\n tab[i] = newNode(hash, key, value, null);\n //节点位置不为null，表示这个位置已经有值了，于是需要进行比较hash值是否相等\n else {\n Node e; K k;\n /*\n 比较桶中第一个元素(数组中的结点)的hash值和key是否相等\n 1，p.hash == hash 中的p.hash表示原来存在数据的hash值 hash表示后添加数据的hash值 比较两个hash值是否相等\n 2，(k = p.key) == key ：p.key获取原来数据的key赋值给k key表示后添加数据的key 比较两个key的地址值是否相等\n 3，key != null && key.equals(k)：能够执行到这里说明两个key的地址值不相等，那么先判断后添加的key是否等于null，如果不等于null再调用equals方法判断两个key的内容是否相等\n */\n if (p.hash == hash &&\n ((k = p.key) == key || (key != null && key.equals(k))))\n /*\n 说明：两个元素哈希值相等（哈希碰撞），并且key的值也相等\n 将旧的元素整体对象赋值给e，用e来记录\n */ \n e = p;\n // hash值不相等或者key不相等；判断p是否为红黑树结点\n else if (p instanceof TreeNode)\n // 是红黑树，调用树的插入方法\n e = ((TreeNode)p).putTreeVal(this, tab, hash, key, value);\n // 说明是链表节点，这时进行插入操作\n else {\n /*\n 1，如果是链表的话需要遍历到最后节点然后插入\n 2，采用循环遍历的方式，判断链表中是否有重复的key\n */\n for (int binCount = 0; ; ++binCount) {\n /*\n 1)e = p.next 获取p的下一个元素赋值给e\n 2)(e = p.next) == null 判断p.next是否等于null，等于null，说明p没有下一个元 素，那么此时到达了链表的尾部，还没有找到重复的key,则说明HashMap没有包含该键\n 将该键值对插入链表中\n */\n if ((e = p.next) == null) {\n p.next = newNode(hash, key, value, null);\n //插入后发现链表长度大于8，转换成红黑树结构\n if (binCount >= TREEIFY_THRESHOLD - 1) \n //转换为红黑树\n treeifyBin(tab, hash);\n break;\n }\n //key值以及存在直接覆盖value\n if (e.hash == hash &&\n ((k = e.key) == key || (key != null && key.equals(k))))\n break;\n p = e;\n }\n }\n //若结点为null，则不进行插入操作\n if (e != null) { \n V oldValue = e.value;\n if (!onlyIfAbsent || oldValue == null)\n e.value = value;\n afterNodeAccess(e);\n return oldValue;\n }\n }\n //修改记录次数\n ++modCount;\n // 判断实际大小是否大于threshold阈值，如果超过则扩容\n if (++size > threshold)\n resize();\n // 插入后回调\n afterNodeInsertion(evict);\n return null;\n }"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"小结："}]},{"type":"numberedlist","attrs":{"start":null,"normalizeStart":1},"content":[{"type":"listitem","content":[{"type":"paragraph","attrs":{"indent":0,"number":1,"align":null,"origin":null},"content":[{"type":"text","text":"根据哈希表中元素个数确定是"},{"type":"text","marks":[{"type":"strong"}],"text":"扩容还是树形化"}]}]},{"type":"listitem","content":[{"type":"paragraph","attrs":{"indent":0,"number":2,"align":null,"origin":null},"content":[{"type":"text","text":"如果是树形化遍历桶中的元素，创建相同个数的树形节点，复制内容，建立起联系"}]}]},{"type":"listitem","content":[{"type":"paragraph","attrs":{"indent":0,"number":3,"align":null,"origin":null},"content":[{"type":"text","text":"然后让桶中的第一个元素指向新创建的树根节点，替换桶的链表内容为树形化内容"}]}]}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"3，HashMap的扩容机制"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我们知道，数组的容量是有限的，多次插入数据的话，到达一定数量就会进行扩容；先来看两个问题"}]},{"type":"heading","attrs":{"align":null,"level":5},"content":[{"type":"text","text":"什么时候需要扩容？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"当HashMap中的"},{"type":"text","marks":[{"type":"strong"}],"text":"元素个数超过数组长度loadFactor(负载因子)"},{"type":"text","text":"时，就会进行数组扩容，loadFactor的默认值是0.75,这是一个折中的取值。也就是说，默认情况下，数组大小为16，那么当HashMap中的元素个数超过16×0.75=12(这个值就是阈值)的时候，就把数组的大小扩展为2×16=32，即扩大一倍，然后"},{"type":"text","marks":[{"type":"strong"}],"text":"重新计算每个元素在数组中的位置"},{"type":"text","text":"，而这是一个非常耗性能的操作，所以如果我们已经预知HashMap中元素的个数，那么预知元素的个数能够有效的提高HashMap的性能。"}]},{"type":"heading","attrs":{"align":null,"level":5},"content":[{"type":"text","text":"怎么进行扩容的？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"HashMap在进行扩容时使用 resize() 方法，计算 table 数组的新容量和 Node 在新数组中的新位置，将旧数组中的值复制到新数组中，从而实现自动扩容。因为每次扩容都是翻倍，与原来计算的 (n-1)&hash的结果相比，只是多了一个bit位，所以节点要么就在原来的位置，要么就被分配到\""},{"type":"text","marks":[{"type":"strong"}],"text":"原位置+旧容量"},{"type":"text","text":"\"这个位置。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"因此，我们在扩充HashMap的时候，不需要重新计算hash，只需要看看原来hash值新增的那个bit是1还是0就可以了，是0的话索引没变，是1的话索引变成“原索引+oldCap("},{"type":"text","marks":[{"type":"strong"}],"text":"原位置+旧容量"},{"type":"text","text":")”。这里不再详细赘述，可以看看下图为16扩充为32的resize示意图："}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/fc/fc1b19df3ec331e181c92fa047b08dff.png","alt":null,"title":null,"style":null,"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"hashmap扩容"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"4，HashMap数组长度为什么是2的次幂"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我们先看看它的成员变量："}]},{"type":"heading","attrs":{"align":null,"level":5},"content":[{"type":"text","text":"序列化版本号"}]},{"type":"codeblock","attrs":{"lang":"java"},"content":[{"type":"text","text":"private static final long serialVersionUID = 362498820763181265L;"}]},{"type":"heading","attrs":{"align":null,"level":5},"content":[{"type":"text","text":"集合的初始化容量initCapacity"}]},{"type":"codeblock","attrs":{"lang":"java"},"content":[{"type":"text","text":"//默认的初始容量是16 -- 1<<4相当于1*2的4次方---1*16\nstatic final int DEFAULT_INITIAL_CAPACITY = 1 << 4; "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"初始化容量默认是16，容量过大，遍历时会减慢速度，效率低；容量过小，那么扩容的次数变多，非常耗费性能。"}]},{"type":"heading","attrs":{"align":null,"level":5},"content":[{"type":"text","text":"负载因子"}]},{"type":"codeblock","attrs":{"lang":"java"},"content":[{"type":"text","text":"/**\n * The load factor used when none specified in constructor.\n */\n static final float DEFAULT_LOAD_FACTOR = 0.75f;"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"初始默认值为0.75，若过大，会导致哈希冲突的可能性更大；若过小，扩容的次数也会提高。"}]},{"type":"heading","attrs":{"align":null,"level":5},"content":[{"type":"text","text":"为什么必须是2的n次幂？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"当向HashMap中添加一个元素的时候，需要根据key的hash值，去确定其在数组中的具体位置。HashMap为了提高存取效率，要尽量较少碰撞，就是要尽量把数据分配均匀，每个链表长度大致相同，这个实现就在把数据存到哪个链表中的算法。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"这个算法实际就是取模，hash%length，计算机中直接求余效率不如位移运算。所以源码中做了优化,使用 "},{"type":"codeinline","content":[{"type":"text","text":"hash&(length-1)"}]},{"type":"text","text":"，而实际上"},{"type":"codeinline","content":[{"type":"text","text":"hash%length"}]},{"type":"text","text":"等于"},{"type":"codeinline","content":[{"type":"text","text":"hash&(length-1)"}]},{"type":"text","text":"的前提是length是2的n次幂。"}]},{"type":"heading","attrs":{"align":null,"level":5},"content":[{"type":"text","text":"如果输入值不是2的幂会怎么样？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"如果数组长度不是2的n次幂，计算出的索引特别容易相同，及其容易发生hash碰撞，导致其余数组空间很大程度上并没有存储数据，链表或者红黑树过长，效率降低。"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"小结："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"1，当根据key的hash确定其在数组的位置时，如果n为2的幂次方"},{"type":"text","marks":[{"type":"strong"}],"text":"，可以保证数据的均匀插入"},{"type":"text","text":"，如果n不是2的幂次方，可能数组的一些位置永远不会插入数据，"},{"type":"text","marks":[{"type":"strong"}],"text":"浪费数组的空间，加大hash冲突。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"2，一般可能会想通过 % 求余来确定位置，这样也可以，只不过性能不如 & 运算。而且当n是2的幂次方时：hash & (length - 1) == hash % length"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"3，因此，HashMap 容量为2次幂的原因，就是"},{"type":"text","marks":[{"type":"strong"}],"text":"为了数据的的均匀分布，减少hash冲突，"},{"type":"text","text":"毕竟hash冲突越大，代表数组中一个链的长度越大，这样的话会降低hashmap的性能"}]},{"type":"horizontalrule"},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"微信搜索公众号《"},{"type":"text","marks":[{"type":"size","attrs":{"size":12}},{"type":"color","attrs":{"color":"#F5222D","name":"red"}}],"text":"程序员的时光"},{"type":"text","text":"》"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#000000","name":"black"}}],"text":"好了，今天就先分享到这里了，下期继续给大家带来HashMap面试内容！"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#000000","name":"black"}}],"text":"更多干货、优质文章，欢迎关注我的原创技术公众号~"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}}]}