概述
本文是基於jdk8_271版本進行分析的。
Hashtable與HashMap一樣,是一個儲存key-value的雙列集合。底層是基於陣列+連結串列實現的,沒有紅黑樹結構。Hashtable預設初始化容量為11,Hashtable也會動態擴容,與HashMap不同的是,每次擴容的容量是原容量2倍+1。Hashtable的key和value都不允許為null。Hashtable在方法上都加了synchronized同步鎖。所以Hashtable是執行緒安全的,同時Hashtable的效率也相對較低。
資料結構
-
實現繼承關係
1 public class Hashtable<K,V> 2 extends Dictionary<K,V> 3 implements Map<K,V>, Cloneable, java.io.Serializable
- Dictionary:
- Map:
- Cloneable:
- Serializable:
-
成員變數
1 // 存放hash表資料 2 private transient Entry<?,?>[] table; 3 4 // 元素數量 5 private transient int count; 6 7 // 閾值。元素數量達到該值,進行擴容 8 private int threshold; 9 10 // 載入因子,預設是0.75 11 private float loadFactor; 12 13 // 修改次數 14 private transient int modCount = 0;
-
建構函式
Hashtable預設初始化容量為11,預設載入因子的值為0.75(與HashMap一樣)。選擇0.75作為預設的載入因子,完全是時間和空間成本上尋求的一種折中選擇。載入因子過高雖然減少了空間開銷,但同時也增加了查詢成本;載入因子過低雖然可以減少查詢時間成本,但是空間利用率很低。
Hashtable初始化容量值使用傳入的值(0除外),不會重新計算(HashMap需要重新計算,使得容量大小為2的指數次冪)。在構造方法建立物件時,會直接初始化陣列,沒有采用懶載入的方式。
1 public Hashtable(int initialCapacity, float loadFactor) { 2 if (initialCapacity < 0) 3 throw new IllegalArgumentException("Illegal Capacity: "+ 4 initialCapacity); 5 if (loadFactor <= 0 || Float.isNaN(loadFactor)) 6 throw new IllegalArgumentException("Illegal Load: "+loadFactor); 7 // 如果初始化容量傳入的是0,則預設使用1 8 if (initialCapacity==0) 9 initialCapacity = 1; 10 this.loadFactor = loadFactor; 11 table = new Entry<?,?>[initialCapacity]; 12 // 計算閾值。預計的閾值為初始化容量*載入因子,預計的閾值如果大於MAX_ARRAY_SIZE + 1,則實際閾值設定為MAX_ARRAY_SIZE + 1 13 threshold = (int)Math.min(initialCapacity * loadFactor, MAX_ARRAY_SIZE + 1); 14 } 15 16 public Hashtable(int initialCapacity) { 17 // 傳入初始化容量,載入因子使用預設值0.75。初始化容量傳入的是多少就初始化多大(0除外;傳入的如果0,預設使用1),不需要再重新計算 18 this(initialCapacity, 0.75f); 19 } 20 21 public Hashtable() { 22 // 預設初始化容量11,預設載入因子0.75 23 this(11, 0.75f); 24 } 25 26 public Hashtable(Map<? extends K, ? extends V> t) { 27 // 初始化容量為傳入集合元素數量的2倍(至少為11),載入因子使用預設值0.75 28 this(Math.max(2*t.size(), 11), 0.75f); 29 putAll(t); 30 }
主要方法解析
-
擴容方法
這裡與HashMap擴容時候有點區別,連結串列資料遷移時候,Hashtable是在連結串列頭部插入(和之前連結串列反過來),HashMap是在尾部插入。
1 protected void rehash() { 2 int oldCapacity = table.length; // 原容量值 3 Entry<?,?>[] oldMap = table; // 原陣列 4 5 // overflow-conscious code 6 int newCapacity = (oldCapacity << 1) + 1; // 預計擴容的容量為原容量的2倍+1 7 if (newCapacity - MAX_ARRAY_SIZE > 0) { 8 if (oldCapacity == MAX_ARRAY_SIZE) // 預計擴容容量如果大於容量最大值,並且原容量為容量最大值,則不進行擴容處理 9 // Keep running with MAX_ARRAY_SIZE buckets 10 return; 11 newCapacity = MAX_ARRAY_SIZE; // 預計擴容容量如果大於容量最大值,則將新容量設定為容量最大值 12 } 13 Entry<?,?>[] newMap = new Entry<?,?>[newCapacity]; // 構建一個新陣列 14 15 modCount++; // 修改次數+1 16 // 計算閾值。新容量值*載入因子,與容量最大值+1,兩個比較取最小值 17 threshold = (int)Math.min(newCapacity * loadFactor, MAX_ARRAY_SIZE + 1); 18 table = newMap; 19 20 for (int i = oldCapacity ; i-- > 0 ;) { 21 // 遍歷原陣列,從後往前 22 for (Entry<K,V> old = (Entry<K,V>)oldMap[i] ; old != null ; ) { 23 // 遍歷該索引位連結串列,這裡與jdk8中hashmap有點區別,這裡是在連結串列頭部插入(和之前連結串列會反過來),hashmap是在尾部插入 24 Entry<K,V> e = old; 25 old = old.next; 26 27 int index = (e.hash & 0x7FFFFFFF) % newCapacity; 28 e.next = (Entry<K,V>)newMap[index]; 29 newMap[index] = e; 30 } 31 } 32 }
-
新增元素
新增元素時,Hashtable與HashMap有3點區別:
- Hashtable的key-value都不允許為null。
- Hashtable是在連結串列頭部插入(和之前連結串列反過來),HashMap是在尾部插入。
- Hashtable是先判斷是否需要擴容,再插入元素;jdk8HashMap是先插入元素再判斷是否需要擴容。
1 public synchronized V put(K key, V value) { 2 // Make sure the value is not null 3 if (value == null) { 4 // value為空,會丟擲空指標異常 5 throw new NullPointerException(); 6 } 7 8 // Makes sure the key is not already in the hashtable. 9 Entry<?,?> tab[] = table; 10 int hash = key.hashCode(); 11 int index = (hash & 0x7FFFFFFF) % tab.length; 12 @SuppressWarnings("unchecked") 13 Entry<K,V> entry = (Entry<K,V>)tab[index]; 14 for(; entry != null ; entry = entry.next) { 15 if ((entry.hash == hash) && entry.key.equals(key)) { 16 // 該key已經存在,直接替換原值 17 V old = entry.value; 18 entry.value = value; 19 return old; 20 } 21 } 22 // 新增元素 23 addEntry(hash, key, value, index); 24 return null; 25 } 26 private void addEntry(int hash, K key, V value, int index) { 27 modCount++; 28 29 Entry<?,?> tab[] = table; 30 if (count >= threshold) { // 判斷是否元素數量是否達到閾值,如果達到先進行擴容處理 31 // Rehash the table if the threshold is exceeded 32 rehash(); 33 34 tab = table; 35 hash = key.hashCode(); 36 index = (hash & 0x7FFFFFFF) % tab.length; 37 } 38 39 // Creates the new entry. 40 @SuppressWarnings("unchecked") 41 Entry<K,V> e = (Entry<K,V>) tab[index]; 42 // 插入元素是在連結串列頭部插入 43 tab[index] = new Entry<>(hash, key, value, e); 44 count++; 45 }
-
刪除元素
1 public synchronized V remove(Object key) { 2 Entry<?,?> tab[] = table; 3 int hash = key.hashCode(); 4 int index = (hash & 0x7FFFFFFF) % tab.length; 5 @SuppressWarnings("unchecked") 6 Entry<K,V> e = (Entry<K,V>)tab[index]; 7 for(Entry<K,V> prev = null ; e != null ; prev = e, e = e.next) { 8 if ((e.hash == hash) && e.key.equals(key)) { 9 modCount++; 10 if (prev != null) { 11 prev.next = e.next; 12 } else { 13 tab[index] = e.next; 14 } 15 count--; 16 V oldValue = e.value; 17 e.value = null; 18 return oldValue; 19 } 20 } 21 return null; 22 } 23 public synchronized boolean remove(Object key, Object value) { 24 Objects.requireNonNull(value); 25 26 Entry<?,?> tab[] = table; 27 int hash = key.hashCode(); 28 int index = (hash & 0x7FFFFFFF) % tab.length; 29 @SuppressWarnings("unchecked") 30 Entry<K,V> e = (Entry<K,V>)tab[index]; 31 for (Entry<K,V> prev = null; e != null; prev = e, e = e.next) { 32 if ((e.hash == hash) && e.key.equals(key) && e.value.equals(value)) { 33 modCount++; 34 if (prev != null) { 35 prev.next = e.next; 36 } else { 37 tab[index] = e.next; 38 } 39 count--; 40 e.value = null; 41 return true; 42 } 43 } 44 return false; 45 }
-
序列化/反序列化方法
1 private void writeObject(java.io.ObjectOutputStream s) 2 throws IOException { 3 Entry<Object, Object> entryStack = null; 4 5 synchronized (this) { 6 // Write out the threshold and loadFactor 7 s.defaultWriteObject(); 8 9 // Write out the length and count of elements 10 s.writeInt(table.length); 11 s.writeInt(count); 12 13 // Stack copies of the entries in the table 14 for (int index = 0; index < table.length; index++) { 15 Entry<?,?> entry = table[index]; 16 17 while (entry != null) { 18 entryStack = 19 new Entry<>(0, entry.key, entry.value, entryStack); 20 entry = entry.next; 21 } 22 } 23 } 24 25 // Write out the key/value objects from the stacked entries 26 while (entryStack != null) { 27 s.writeObject(entryStack.key); 28 s.writeObject(entryStack.value); 29 entryStack = entryStack.next; 30 } 31 } 32 33 private void readObject(java.io.ObjectInputStream s) 34 throws IOException, ClassNotFoundException 35 { 36 // Read in the threshold and loadFactor 37 s.defaultReadObject(); 38 39 // Validate loadFactor (ignore threshold - it will be re-computed) 40 if (loadFactor <= 0 || Float.isNaN(loadFactor)) 41 throw new StreamCorruptedException("Illegal Load: " + loadFactor); 42 43 // Read the original length of the array and number of elements 44 int origlength = s.readInt(); 45 int elements = s.readInt(); 46 47 // Validate # of elements 48 if (elements < 0) 49 throw new StreamCorruptedException("Illegal # of Elements: " + elements); 50 51 // Clamp original length to be more than elements / loadFactor 52 // (this is the invariant enforced with auto-growth) 53 origlength = Math.max(origlength, (int)(elements / loadFactor) + 1); 54 55 // Compute new length with a bit of room 5% + 3 to grow but 56 // no larger than the clamped original length. Make the length 57 // odd if it's large enough, this helps distribute the entries. 58 // Guard against the length ending up zero, that's not valid. 59 int length = (int)((elements + elements / 20) / loadFactor) + 3; 60 if (length > elements && (length & 1) == 0) 61 length--; 62 length = Math.min(length, origlength); 63 64 if (length < 0) { // overflow 65 length = origlength; 66 } 67 68 // Check Map.Entry[].class since it's the nearest public type to 69 // what we're actually creating. 70 SharedSecrets.getJavaOISAccess().checkArray(s, Map.Entry[].class, length); 71 table = new Entry<?,?>[length]; 72 threshold = (int)Math.min(length * loadFactor, MAX_ARRAY_SIZE + 1); 73 count = 0; 74 75 // Read the number of elements and then all the key/value objects 76 for (; elements > 0; elements--) { 77 @SuppressWarnings("unchecked") 78 K key = (K)s.readObject(); 79 @SuppressWarnings("unchecked") 80 V value = (V)s.readObject(); 81 // sync is eliminated for performance 82 reconstitutionPut(table, key, value); 83 } 84 }
附錄
HashMap原始碼詳細註釋Github地址:https://github.com/y2ex/jdk-source/blob/jdk1.8.0_271/src/main/java/java/util/Hashtable.java
jdk1.8原始碼Github地址:https://github.com/y2ex/jdk-source/tree/jdk1.8.0_271