JDK1.8源码分析:ArrayList（jdk 源码）

数据结构

ArrayList相对于数组Array只是提供了动态拓容的功能，在内部也是使用一个数组来存储数据的。

ArrayList也不是线程安全的，如果需要线程安全，则需要使用Collections.synchronziedList来进行包装，如：List list = Collections.synchronizedList(new ArrayList(…));

/**
 * Default initial capacity.
 */
private static final int DEFAULT_CAPACITY = 10;

/**
 * Shared empty array instance used for empty instances.
 */
private static final Object[] EMPTY_ELEMENTDATA = {};

/**
 * Shared empty array instance used for default sized empty instances. We
 * distinguish this from EMPTY_ELEMENTDATA to know how much to inflate when
 * first element is added.
 */
private static final Object[] DEFAULTCAPACITY_EMPTY_ELEMENTDATA = {};

/**
 * The array buffer into which the elements of the ArrayList are stored.
 * The capacity of the ArrayList is the length of this array buffer. Any
 * empty ArrayList with elementData == DEFAULTCAPACITY_EMPTY_ELEMENTDATA
 * will be expanded to DEFAULT_CAPACITY when the first element is added.
 */
transient Object[] elementData; // non-private to simplify nested class access

元素个数size维护：size记录了当前list的内部数组存放了多少个元素，所以可以通过size方法来通过O(1)时间复杂度返回size，而不需要每次都对数组进行一次遍历计算。主要是在添加和删除元素时，执行size++或size–来同步更新：

/**
 * Appends the specified element to the end of this list.
 *
 * @param e element to be appended to this list
 * @return <tt>true</tt> (as specified by {@link Collection#add})
 */
public boolean add(E e) {
    ensureCapacityInternal(size + 1); // Increments modCount!!
    
    // 递增size
    elementData[size++] = e;
    return true;
}

/**
 * Removes the element at the specified position in this list.
 * Shifts any subsequent elements to the left (subtracts one from their
 * indices).
 *
 * @param index the index of the element to be removed
 * @return the element that was removed from the list
 * @throws IndexOutOfBoundsException {@inheritDoc}
 */
public E remove(int index) {
    rangeCheck(index);

    modCount++;
    E oldValue = elementData(index);

    int numMoved = size - index - 1;
    if (numMoved > 0)
        System.arraycopy(elementData, index+1, elementData, index,
                         numMoved);
    
    // 递减size
    elementData[--size] = null; // clear to let GC do its work

    return oldValue;
}

拓容规则

内部数组拓容规则：每次在调用add添加元素时，如果内部数组大小满了，则会进行拓容。拓容规则为：初始大小为10，之后每次拓容为当前数组容量的1.5倍，实现源码如下：主要在grow方法实现。

/**
 * The maximum size of array to allocate.
 * Some VMs reserve some header words in an array.
 * Attempts to allocate larger arrays may result in
 * OutOfMemoryError: Requested array size exceeds VM limit
 */
private static final int MAX_ARRAY_SIZE = Integer.MAX_VALUE - 8;

/**
 * Increases the capacity to ensure that it can hold at least the
 * number of elements specified by the minimum capacity argument.
 *
 * @param minCapacity the desired minimum capacity
 */
private void grow(int minCapacity) {
    // overflow-conscious code
    int oldCapacity = elementData.length;
    
    // 默认拓容为原来的1.5倍
    int newCapacity = oldCapacity + (oldCapacity >> 1);
    
    // 如果原来的容量加一比原来的1.5倍大，则使用该加一的版本
    if (newCapacity - minCapacity < 0)
        newCapacity = minCapacity;
        
    // 最大不超过Integer.MAX_VALUE - 8
    if (newCapacity - MAX_ARRAY_SIZE > 0)
        newCapacity = hugeCapacity(minCapacity);
        
    // minCapacity is usually close to size, so this is a win:
    elementData = Arrays.copyOf(elementData, newCapacity);
}

private static int hugeCapacity(int minCapacity) {
    if (minCapacity < 0) // overflow
        throw new OutOfMemoryError();
    return (minCapacity > MAX_ARRAY_SIZE) ?
        Integer.MAX_VALUE :
        MAX_ARRAY_SIZE;
}

拓容在确定好新的容量之后，会通过Arrays.copyOf方法来新建一个该新容量的数组，然后在Arrays.copyOf内部通过System.arraycopy方法将旧数组的内容拷贝到新数组，然后返回该新数组作为内部数组。其中System.arraycopy为本地native方法，性能较高。

private void grow(int minCapacity) {
        // overflow-conscious code
        int oldCapacity = elementData.length;
        int newCapacity = oldCapacity + (oldCapacity >> 1);
        if (newCapacity - minCapacity < 0)
            newCapacity = minCapacity;
        if (newCapacity - MAX_ARRAY_SIZE > 0)
            newCapacity = hugeCapacity(minCapacity);
        // minCapacity is usually close to size, so this is a win:
        
        // 数组拓容
        elementData = Arrays.copyOf(elementData, newCapacity);
    }
    
    
    // Arrays.copyOf的实现：
    
    public static <T> T[] copyOf(T[] original, int newLength) {
        return (T[]) copyOf(original, newLength, original.getClass());
    }
    
    public static <T,U> T[] copyOf(U[] original, int newLength, Class<? extends T[]> newType) {
        @SuppressWarnings("unchecked")
        T[] copy = ((Object)newType == (Object)Object[].class)
            ? (T[]) new Object[newLength]
            : (T[]) Array.newInstance(newType.getComponentType(), newLength);
        System.arraycopy(original, 0, copy, 0,
                         Math.min(original.length, newLength));
        return copy;
    }

子集SubList

子集合SubList：SubList为List列表的一个子集，在底层与List列表公用一个内部数组来存储数据，所以对suList返回的子列表进行操作也会对主列表List产生影响。如下：set方法其实是对主列表的内部数组elementData进行操作的。

private class SubList extends AbstractList<E> implements RandomAccess {
    private final AbstractList<E> parent;
    private final int parentOffset;
    private final int offset;
    int size;

    SubList(AbstractList<E> parent,
            int offset, int fromIndex, int toIndex) {
        this.parent = parent;
        this.parentOffset = fromIndex;
        this.offset = offset + fromIndex;
        this.size = toIndex - fromIndex;
        this.modCount = ArrayList.this.modCount;
    }

    public E set(int index, E e) {
        rangeCheck(index);
        checkForComodification();
        E oldValue = ArrayList.this.elementData(offset + index);
        ArrayList.this.elementData[offset + index] = e;
        return oldValue;
    }
    
    ...
    
}

迭代器

迭代器iterator：迭代器也是fail-fast的，注意判断fail-fast的条件是在当遍历到的数组下标i大于当前内部数组的size或者数组容量capacity时，才抛出异常。所以在使用迭代器的时候，不能依赖数组的位置关系，否则可能会出错，如在遍历第5个位置之前，其他线程在1到5之间插入了一个元素，则第5个其实是产生该迭代器的时候的第4个。

如果需要依赖位置关系，即线程安全方面的考虑，则需要使用Collections.synchronizedList来包装成线程安全的List。

public Iterator<E> iterator() {
    return new Itr();
}

/**
 * An optimized version of AbstractList.Itr
 */
private class Itr implements Iterator<E> {
    int cursor; // index of next element to return
    int lastRet = -1; // index of last element returned; -1 if no such
    int expectedModCount = modCount;

    public boolean hasNext() {
        return cursor != size;
    }

    @SuppressWarnings("unchecked")
    public E next() {
        checkForComodification();
        int i = cursor;
        
        // i 大于数组大小
        if (i >= size)
            throw new NoSuchElementException();
        Object[] elementData = ArrayList.this.elementData;
        
        // i 大于数组容量
        if (i >= elementData.length)
            throw new ConcurrentModificationException();
        cursor = i + 1;
        return (E) elementData[lastRet = i];
    }
    
    ...