import java.util.ArrayList;
import java.util.Arrays;
import java.util.Comparator;
import java.util.List;import com.dm.core.structure.tupler.StrDoubleTuple;/*** 最大堆，用作优先队列的TOPK查找<br>* 原理：每个节点的值都>=其左右孩子（如果有的话）值的完全二叉树* * @author Anthony* @param <T>*/
public class MaxHeap<T> {/*** 堆数据*/private List<T> heap;/*** 堆数据的比较对象*/private Comparator<T> comparator;public MaxHeap(List<T> heap) {this(heap, new Comparator<T>() {@SuppressWarnings("unchecked")@Overridepublic int compare(T o1, T o2) {return ((Comparable<T>) o1).compareTo(o2);}});}public MaxHeap(List<T> heap, Comparator<T> comparator) {super();this.heap = heap;this.comparator = comparator;}/*** 向最大堆中插入元素，添加到数组的尾部，然后上升操作* * @param value*/public void insert(T value) {// 数组下标为0的位置不放元素if (heap.size() == 0)heap.add(null);heap.add(value);up(heap.size() - 1);}/*** 节点上升递归实现<br>* 由于新插入的数是在数组尾部，所以需要做上升操作，让插入的数和父节点的值比较，当大于父节点的时候交换* * @param index*/private void up(int index) {// 注意堆是从下标为1开始，当index=1的时候，已经是根节点了if (index <= 1)return;int parent = index / 2; // 父节点T parentValue = heap.get(parent);T indexValue = heap.get(index);if (comparator.compare(parentValue, indexValue) < 0) {swap(parent, index);up(parent);}}/*** 节点上升非递归实现* * @param index*/@SuppressWarnings("unused")private void up2(int index) {int parent = 0;for (; index > 1; index /= 2) {parent = index / 2;T parentValue = heap.get(parent);T indexValue = heap.get(index);if (comparator.compare(parentValue, indexValue) < 0)swap(parent, index);}}/*** 交换a和b的位置* * @param a* @param b*/private void swap(int a, int b) {T temp = heap.get(a);heap.set(a, heap.get(b));heap.set(b, temp);}/*** 删除堆中位置是index处的值<br>* 原理是：当删除节点时，原来的位置就会出现一个孔，填充这个孔的方法就是，把最后的叶子赋给该孔，然后把该叶子删除* * @param index*/public void delete(int index) {heap.set(index, heap.get(heap.size() - 1));down(index);heap.remove(heap.size() - 1);}/*** 节点下沉递归实现<br>* 删除数据的时候，由于是用的尾部的数据（基本上是最小值）填充，所以需要做下沉操作* * @param index*/public void down(int index) {int n = heap.size() - 2; // 因为最后一个节点已经挪至index位置，所以已经是废弃叶子节点，不再考虑int child = 2 * index;// 说明该节点没有左右儿子节点了，那么无须下沉，直接返回if (child > n)return;// 如果左右儿子都存在，取值较大的那个儿子节点if (child < n&& comparator.compare(heap.get(child), heap.get(child + 1)) < 0)child++;// 如果该节点小于较大的那个儿子，那么下沉if (comparator.compare(heap.get(index), heap.get(child)) < 0) {swap(child, index);down(child);}}/*** 节点下沉非递归实现* * @param index*/public void down2(int index) {T temp = heap.get(index);int n = heap.size() - 2;int child = 0;for (; 2 * index <= n; index = child) {child = 2 * index;if (child < n&& comparator.compare(heap.get(child), heap.get(child + 1)) < 0)child++;if (comparator.compare(temp, heap.get(child)) < 0)swap(child, index);elsebreak;}}/*** 根据树的性质建堆，树节点前一半一定是分支节点，即有孩子的，所以我们从这里开始调整出初始堆* * @param heap*/public void adjust() {for (int i = heap.size() / 2; i > 0; i--)adjust(i, heap.size() - 1);}/*** 调整堆，使其满足最大堆得定义<br>* 具体调整过程为: 从最后一个分支结点（n/2）开始，到根（1）为止，依次对每个分支结点进行调整（下沉）<br>* 以便形成以每个分支结点为根的堆，当最后对树根结点进行调整后，整个树就变成了一个堆* * @param i* @param n*/public void adjust(int i, int n) {int child = 0;for (; i <= n / 2;) {child = i * 2;if (child < n&& comparator.compare(heap.get(child), heap.get(child + 1)) < 0)child++;if (comparator.compare(heap.get(i), heap.get(child)) < 0) {swap(i, child);i = child; // 交换后，以child+1为根的子树不一定满足堆定义，所以从child处开始调整} elsebreak;}}/*** 堆排序，从尾部开始，将每个节点和根节点交换，然后调整节点之上的子堆*/public void sort() {for (int i = heap.size() - 1; i > 0; i--) {swap(1, i);adjust(1, i - 1);}}public static void main(String args[]) {List<Integer> array = new ArrayList<Integer>(Arrays.asList(null, 1, 2,5, 10, 3, 7, 11, 15, 17, 20, 9, 15, 8, 16));MaxHeap<Integer> mh = new MaxHeap<Integer>(array);mh.adjust();System.out.println("调整后的初始堆:" + array);mh.delete(8);System.out.println("删除下标8之后的堆:" + array);mh.insert(99);System.out.println("添加值99之后的堆:" + array);mh.sort();System.out.println("排序后的堆:" + array);List<StrDoubleTuple> list = new ArrayList<StrDoubleTuple>();list.add(null);list.add(new StrDoubleTuple("a", 1.0));list.add(new StrDoubleTuple("a", 2.0));list.add(new StrDoubleTuple("a", 5.0));list.add(new StrDoubleTuple("a", 10.0));list.add(new StrDoubleTuple("a", 3.0));list.add(new StrDoubleTuple("a", 7.0));list.add(new StrDoubleTuple("a", 11.0));list.add(new StrDoubleTuple("a", 15.0));list.add(new StrDoubleTuple("a", 17.0));list.add(new StrDoubleTuple("a", 20.0));list.add(new StrDoubleTuple("a", 9.0));list.add(new StrDoubleTuple("b", 15.0));list.add(new StrDoubleTuple("a", 8.0));list.add(new StrDoubleTuple("a", 16.0));MaxHeap<StrDoubleTuple> mho = new MaxHeap<StrDoubleTuple>(list);mho.adjust();System.out.println("调整后的初始堆:" + list);mho.delete(8);System.out.println("删除下标8之后的堆:" + list);mho.insert(new StrDoubleTuple("a", 99.0));System.out.println("添加值99之后的堆:" + list);mho.sort();System.out.println("排序后的堆:" + list);}
}