java 实现哈夫曼编码

核心算法是哈夫曼树的构建以及哈夫曼最小两个权重子树的选择。其中最小权重子树的选择有待优化。如有其它瑕疵，欢迎留言斧正，本人将感激不尽。完整源码请移步github:github.com/VeinXuan/hu…

package com.huawei.entity.tree.huffman;

import java.util.ArrayList; import java.util.List;

import com.alibaba.fastjson.JSONArray; import com.huawei.entity.tree.binary.BinaryNode; import com.huawei.entity.tree.node.SimpleLinkedStack;

/** *

• @author vein

• @date 2019年12月2日 */ public class HuffmanTreeBuilder { public static void main(String[] args) { List nums = new ArrayList(); int[] ints = { -2, -0, 2 }; for (int i = 0; i < ints.length; i++) { nums.add(ints[i]); } HuffmanNode<?> hf = buildNode(nums); System.out.println(JSONArray.toJSONString(hf)); buildHuffmanCode(hf); System.out.println(JSONArray.toJSONString(hf)); }

  /**

  • 构建哈夫曼编码
  • @param huffmanTree */ public static <T extends Comparable<? super T>> void buildHuffmanCode(HuffmanNode huffmanTree) { if (huffmanTree == null) { return; } HuffmanNode parent = null, curNode = huffmanTree; char[] huffmanCode; SimpleLinkedStack<BinaryNode> stack = new SimpleLinkedStack<BinaryNode>(); do { parent = (HuffmanNode) curNode.getParent(); curNode.setLevel(parent == null ? 0 : parent.getLevel() + 1); huffmanCode = new char[curNode.getLevel()]; if (curNode.getLevel() > 0) { System.arraycopy(parent.getHmCode(), 0, huffmanCode, 0, parent == null ? 0 : parent.getHmCode().length); if (curNode == parent.getLchild()) {// 左子树 huffmanCode[curNode.getLevel() - 1] = '0'; } else {// 右子树 huffmanCode[curNode.getLevel() - 1] = '1'; } } curNode.setHmCode(huffmanCode); if (curNode.getLchild() != null) { if (curNode.getRchild() != null) { stack.push(curNode.getRchild()); } curNode = (HuffmanNode) curNode.getLchild(); } else if (curNode.getRchild() != null) { curNode = (HuffmanNode) curNode.getRchild(); } else { curNode = (HuffmanNode) stack.pop(); } } while (curNode != null); }

  /**

  • 构建哈夫曼树
  • @param data
  • @return / public static <T extends Comparable<? super T>> HuffmanNode buildNode(List data) { HuffmanNode hn = new HuffmanNode(); if (data == null || data.size() == 0) { return hn; } if (data.size() == 1) { hn.setWeight(0).setParent(new HuffmanNode().setHmCode(new char[0]).setLchild(hn).setLevel(0)) .setData(data.get(0)); return hn; } /
    • 由于哈夫曼树是两两组合在一起的， 所以当节点数大于1时，整个哈夫曼树只存在叶子节点和根节点，
    • 由二叉树的性质可知，叶子节点恰比所有出度为2的节点多一个 因此整个哈夫曼树可以用一个2n-1个节点的一维数组表示 */ List<HuffmanNode> hnList = new ArrayList<HuffmanNode>(data.size() * 2 - 1); // 初始化所有节点 for (int i = 0, j = data.size() * 2 - 1; i < j; i++) { hn = new HuffmanNode(); if (i < data.size()) {// 前n个数组空间用于存储数据相关的节点 hn.setData(data.get(i)); hn.setWeight(data.get(i) == null ? 0 : data.get(i).hashCode()); } else {// 第n+1至第2n-1个数组空间用于存储所有子树的根节点 hn.setWeight(Integer.MAX_VALUE);// 默认所有子树的根节点权重为无限大 } hnList.add(hn); } // 用于保存最小的两个元素 @SuppressWarnings("unchecked") HuffmanNode[] selectedNode = new HuffmanNode[2]; HuffmanNode parent; for (int i = 0, j = data.size() - 1; i < j; i++) { select(hnList, selectedNode);// 选择权重最小的两个子树，进行合并 parent = hnList.get(data.size() + i);// 取n+i个空间作为根节点的存储空间 parent.setWeight(selectedNode[0].getWeight() + selectedNode[1].getWeight());// 新的根节点为左右子树的权重之和 // 双亲与子节点的关联关系的建立 parent.setLchild(selectedNode[0]); parent.setRchild(selectedNode[1]); selectedNode[0].setParent(parent); selectedNode[1].setParent(parent); } return hn = hnList.get(hnList.size() - 1);// 获取最后一个构成的哈夫曼树 }

  private static <T extends Comparable<? super T>> void select(List<HuffmanNode> hnList, HuffmanNode[] selectedNode) { if(hnList==null||hnList.size()<1){ return; } HuffmanNode curNode; selectedNode[0] = selectedNode[1] = hnList.get(hnList.size() - 1);// 默认最后两个元素的权重是最大的 for (int i = 0, j = hnList.size(); i < j; i++) { curNode = hnList.get(i); if (curNode.getParent() != null) { continue; } if (curNode.getWeight() < selectedNode[0].getWeight()) { selectedNode[1] = selectedNode[0]; selectedNode[0] = curNode; } else if (curNode.getWeight() < selectedNode[1].getWeight()) { selectedNode[1] = curNode; } } }

}