HashMap源码简单分析

  1 还是老习惯,一边看,一边添加注释,希望坚持下去,HashMap的基本源码进行了分析,内部一些接口和设计还没来得及看
  2 
  3 一、成员
  4 
  5 1、transient Entry[] table;
  6 
  7 HashMap内部维护了一个内部类-Entry,用来存放键值对,这个Entry实现了Map.Entry这一Map的内部接口Entry,HashMap本质上来讲是由数组和Entry链表组成的数据结构
  8 
  9 2、 static final float DEFAULT_LOAD_FACTOR = 0.75f;
 10 
 11 加载因子,加载因子越大,hash表(即Entry数组)所占空间越少,但会影响查询性能(因为需要通过链表一个挨一个向下查询),加载因子越小,hash表(即Entry数组)所占空间越多,这时查询效率较高,但是hash表所占空间较多
 12 
 13 3、static final int DEFAULT_INITIAL_CAPACITY = 16;
 14 
 15 4、/**
 16 
 17   * The next size value at which to resize (capacity * load factor).
 18   * @serial
 19   */
 20   int threshold;
 21 
 22 5、static final int MAXIMUM_CAPACITY = 1 << 30;
 23 
 24 6、static final int DEFAULT_INITIAL_CAPACITY = 16;
 25 
 26 7、final float loadFactor
 27 
 28 决定什么时候进行扩容
 29 
 30 二、方法
 31 
 32 1、核心构造方法
 33 
 34 public HashMap(int initialCapacity, float loadFactor) {
 35 if (initialCapacity < 0)
 36 throw new IllegalArgumentException("Illegal initial capacity: " +
 37 initialCapacity);
 38 if (initialCapacity > MAXIMUM_CAPACITY)
 39 initialCapacity = MAXIMUM_CAPACITY;
 40 if (loadFactor <= 0 || Float.isNaN(loadFactor))
 41 throw new IllegalArgumentException("Illegal load factor: " +
 42 loadFactor);
 43 
 44 // Find a power of 2 >= initialCapacity
 45 int capacity = 1;
 46 while (capacity < initialCapacity)
 47 capacity <<= 1;
 48 
 49 this.loadFactor = loadFactor;
 50 threshold = (int)(capacity * loadFactor);
 51 table = new Entry[capacity];  //capacity代表数组的长度
 52 init();
 53 }
 54 
 55 2、在key对象的hashCodr()方法的基础上再做hash,避免一些不好的hashCode()方法
 56 
 57 //Null keys always map to hash 0,  如果key为null,那么hash()方法的到的hash值为0,再调用indexFor方法得到的数组的索引值也为0,所以key为null的Entry存在数组下标为0的位置
 58 
 59 static int hash(int h) {
 60 // This function ensures that hashCodes that differ only by
 61 // constant multiples at each bit position have a bounded
 62 // number of collisions (approximately 8 at default load factor).
 63 h ^= (h >>> 20) ^ (h >>> 12);
 64 return h ^ (h >>> 7) ^ (h >>> 4);
 65 }
 66 
 67 3、根据2中获得的hash值和数组的长度得到Entry对应的数组的索引
 68 
 69 static int indexFor(int h, int length) {
 70 return h & (length-1);   //屏蔽高位,保证与操作后最大值为length-1
 71 }
 72 
 73 4、根据key获取value
 74 
 75 public V get(Object key) {
 76 if (key == null)
 77 return getForNullKey();  //如果key为null则直接取index为0的Entry对应的value值
 78 int hash = hash(key.hashCode());  //生成Entryhash值
 79 for (Entry<K,V> e = table[indexFor(hash, table.length)];  //获取header
 80 e != null;
 81 e = e.next) {  //链表向下走
 82 Object k;
 83 if (e.hash == hash && ((k = e.key) == key || key.equals(k)))  //判断key是否相同
 84 return e.value;
 85 }
 86 return null;
 87 }
 88 
 89 5、获取key为null的key对应的值(注意:这里使用在链表中查找的方式,因为index为0的链表上不是只有key为null的Entry)
 90 
 91 private V getForNullKey() {
 92 for (Entry<K,V> e = table[0]; e != null; e = e.next) {
 93 if (e.key == null)
 94 return e.value;
 95 }
 96 return null;
 97 }
 98 
 99 6、添加键值对
100 
101 public V put(K key, V value) {
102 
103 /*如果key存在则对value进行修改并将原value返回*/
104 if (key == null)
105 return putForNullKey(value);
106 int hash = hash(key.hashCode());
107 int i = indexFor(hash, table.length);
108 for (Entry<K,V> e = table[i]; e != null; e = e.next) {
109 Object k;
110 if (e.hash == hash && ((k = e.key) == key || key.equals(k))) {
111 V oldValue = e.value;
112 e.value = value;
113 e.recordAccess(this);
114 return oldValue;
115 }
116 }
117 
118 /*如果key不存在则新增键值对*/
119 
120 modCount++;
121 addEntry(hash, key, value, i);
122 return null;
123 }
124 
125 7、新增加键值对
126 
127 void addEntry(int hash, K key, V value, int bucketIndex) {
128 Entry<K,V> e = table[bucketIndex];  //获取header
129 table[bucketIndex] = new Entry<K,V>(hash, key, value, e);  //new一个新Entry,并将后指针指向原header,新加的Entry成为新header
130 if (size++ >= threshold)  //如果元素个数超过阈值,则进行扩容
131 resize(2 * table.length);  //扩容为原来的2倍
132 }
133 
134 8、扩容
135 
136 void resize(int newCapacity) {
137 Entry[] oldTable = table;
138 int oldCapacity = oldTable.length;
139 if (oldCapacity == MAXIMUM_CAPACITY) {
140 threshold = Integer.MAX_VALUE;
141 return;
142 }
143 
144 Entry[] newTable = new Entry[newCapacity];  //扩容为新capacity
145 transfer(newTable);  //将所有的Entry迁移到新的数组中去
146 table = newTable;
147 threshold = (int)(newCapacity * loadFactor);  //重新计算阈值
148 }
149 
150 9、将所有Entry迁移到新数组中
151 
152 void transfer(Entry[] newTable) {
153 Entry[] src = table;
154 int newCapacity = newTable.length;
155 
156 /*遍历Entry数组的0-(size-1)的索引对应的Entry链表,并将链表上的Entry重新计算在新数组中的索引并迁移到新数组的Entry链中*/
157 for (int j = 0; j < src.length; j++) {
158 Entry<K,V> e = src[j];
159 if (e != null) {
160 src[j] = null;  //for GC
161 do {    //遍历处理某个索引上的Entry链
162 Entry<K,V> next = e.next;
163 int i = indexFor(e.hash, newCapacity);  //重新计算索引 
164 
165 /*将所有Entry分别放到应该放到的indexEntry链上*/
166 e.next = newTable[i];
167 newTable[i] = e;
168 e = next;
169 } while (e != null);
170 }
171 }
172 }

 

posted @ 2017-08-22 18:53  薏米仁儿  阅读(225)  评论(0编辑  收藏  举报