/*
Name: 赫夫曼编码
Copyright: 始发于goal00001111的专栏;允许自由转载,但必须注明作者和出处
Author: goal00001111
Date: 16-12-08 21:16
Description: 赫夫曼编码
本程序实现了使用赫夫曼编码压缩数据;输入一串字符串sourceCode——为方便理解,暂时要求字符串只包含大写字母和空格,如果你愿意,
很容易就可以推广到所有的字符——计算出字符串中各个字母的权重,然后对其进行赫夫曼编码,输出赫夫曼树。
将赫夫曼树的叶子结点存储到有序二叉树中,输出原字符串经压缩后得到的用'0'和'1'表示的新字符串destCode;
然后利用赫夫曼树将字符串destCode进行译码,得到目标字符串objCode,比较objCode和sourceCode,发现完全一样!
编码译码成功!
最后销毁有序二叉树和赫夫曼树。
本程序的一个亮点是使用了二叉堆来存储需要合并的赫夫曼树结点,这样在求最小值时时间复杂度可以降低到log(n)。
另外关于赫夫曼编码的详细内容请参考维基百科: http://zh.wikipedia.org/wiki/%E5%93%88%E5%A4%AB%E6%9B%BC%E7%BC%96%E7%A0%81
和数据结构自考网:http://student.zjzk.cn/course_ware/data_structure/web/shu/shu6.6.2.1.htm
关于二叉堆的详细内容请参考百度百科:http://baike.baidu.com/view/668854.html
*/
#include<iostream>
using namespace std;
typedef char ElemType;
typedef struct sNode
{
double weight;
ElemType data;
} *Source;
typedef struct hNode
{
double weight;
ElemType data;
int lc, rc;
} *HuffmanTree;
typedef struct cNode
{
ElemType data;
string str;
struct cNode *lc, *rc;
} *Btree;
HuffmanTree CreateHuffmanTree(const Source w, int n);//创建一棵赫夫曼树
void BuildHeap(HuffmanTree t, int n); //构造一个二叉堆;小顶堆
void PercDown(HuffmanTree t, int pos, int n);//构造二叉堆的功能子函数
void DeleteMin(HuffmanTree t, int len); //删除二叉堆的根,并通过上移使得新得到的序列仍为二叉堆
void InsertHfNode(HuffmanTree t, int len, struct hNode x); //把x插入到原长度为len的二叉堆
void Preorder(HuffmanTree t, int p); //先序遍历赫夫曼树
void Postorder(Btree & t, HuffmanTree a, int n); //后序遍历赫夫曼树,并记录叶子结点编码
bool InsertBtNode(Btree & t, Btree s); //向一个二叉排序树t中插入一个结点s
void Inorder(Btree t); //中序遍历二叉排序树
Btree Search(Btree p, ElemType data); //查找值为data的结点的递归算法
string Coding(string s, Btree t); //利用记录了叶子结点编码的排序二叉树,对sourceCode进行编码,返回编码后的字符串
string Decode(string s, HuffmanTree hT); //利用赫夫曼树对destCode进行解码
void DestroyBTree(Btree & t); //销毁一棵二叉排序树
void DestroyHfmanTree(HuffmanTree & t, int n); //销毁一棵赫夫曼树
int main()
{
string sourceCode;
getline(cin, sourceCode, '\n');
int n = sourceCode.size();
const int MAX = 27; //原码由26个大写字母加空格组成
Source w = new struct sNode[MAX];
//读取各个字母并初始化权重
w[MAX-1].data = ' ';
w[MAX-1].weight = 0;
for (int i=MAX-2; i>=0; i--)
{
w[i].data = 'A' + i;
w[i].weight = 0;
}
//读取各个字母的权重
for (int i=0; i<n; i++)
{
if (sourceCode[i] == ' ')
w[26].weight++;
else
w[sourceCode[i]-'A'].weight++;
}
//获取出现了的大写字母和空格
n = 0;
for (int i=0; i<MAX; i++)
{
if (w[i].weight > 0)
w[n++] = w[i];
}
// //直接输入原码和权重
// for (int i=0; i<n; i++)
// {
// cin >> w[i].weight >> w[i].data;
// }
for (int i=0; i<n; i++)
{
cout << w[i].weight << " " << w[i].data << endl;
}
HuffmanTree hT = CreateHuffmanTree(w, n);//构造赫夫曼树
// for (int i=1; i<2*n; i++)
// cout << hT[i].weight << " ";
// cout << endl;
//先序遍历赫夫曼树,并输出结点权重和叶子结点的data
Preorder(hT, 1);
cout << endl;
//后序遍历赫夫曼树,并记录叶子结点编码
Btree bT = NULL;
Postorder(bT, hT, n);
//中序遍历记录了叶子结点编码的排序二叉树
Inorder(bT);
//利用记录了叶子结点编码的排序二叉树,对sourceCode进行编码
string destCode = Coding(sourceCode, bT);
cout << destCode << endl;
//利用赫夫曼树对destCode进行解码
string objCode = Decode(destCode, hT);
cout << objCode << endl;
DestroyBTree(bT); //销毁二叉排序树
//Inorder(bT); //再输出试试看
DestroyHfmanTree(hT, n); //销毁赫夫曼树
//Preorder(hT, 1); //再输出试试看
system("pause");
return 0;
}
//创建一棵赫夫曼树
HuffmanTree CreateHuffmanTree(const Source w, int n)
{
HuffmanTree hT = new struct hNode[2*n]; //第一个结点不用
for (int i=0; i<n; i++)
{
hT[i+1].data = w[i].data;
hT[i+1].weight = w[i].weight;
hT[i+1].lc = hT[i+1].rc = 0;
}
BuildHeap(hT, n);//构造一个二叉堆;小顶堆
struct hNode add;
int left = n;
int right = n;
while (left > 1)
{
hT[++right] = hT[1];
add.weight = hT[1].weight;
add.lc = right; //存储左孩子下标
DeleteMin(hT, left--);
hT[left+1] = hT[1];
add.weight += hT[1].weight;
add.rc = left+1; //存储右孩子下标
DeleteMin(hT, left--);
InsertHfNode(hT, ++left, add);
//for (int i=1; i<=right; i++)
// cout << hT[i].weight << " ";
// cout << endl;
// system("pause");
}
return hT;
}
//构造一个二叉堆;小顶堆
void BuildHeap(HuffmanTree t, int len)
{
for (int i=len/2; i>0; i--)
{
PercDown(t, i, len);
}
}
//构造二叉堆的功能子函数
void PercDown(HuffmanTree t, int pos, int len)
{
int child;
struct hNode min = t[pos];
while (pos * 2 <= len)
{
child = pos * 2;
if (child != len && t[child+1].weight < t[child].weight)
child++;
if (min.weight > t[child].weight)
t[pos] = t[child];
else
break;
pos = child;
}
t[pos] = min;
}
//删除二叉堆的根,并通过上移使得新得到的序列仍为二叉堆
void DeleteMin(HuffmanTree t, int len)
{
struct hNode last = t[len--];//二叉堆的最后一个元素
int child, pos = 1;
while (pos * 2 <= len) //把二叉堆的某些元素往前移,使得新得到的序列仍为二叉堆
{
child = pos * 2;
if (child != len && t[child+1].weight < t[child].weight) //若i有右儿子,且右儿子小于左儿子,c指向右儿子
child++;
if (last.weight > t[child].weight) //若i的小儿子小于二叉堆的最后一个元素,把其移到i的位置
t[pos] = t[child];
else
break;
pos = child;
}
t[pos] = last; //把二叉堆的最后一个元素放到适当的空位,此时得到的序列仍为二叉堆
}
//把x插入到原长度为len的二叉堆
void InsertHfNode(HuffmanTree t, int len, struct hNode x)
{
int i;
for (i=len; i/2>0 && t[i/2].weight>x.weight; i/=2)
t[i] = t[i/2];
t[i] = x;
}
//后序遍历赫夫曼树,并记录叶子结点编码
void Postorder(Btree & t, HuffmanTree a, int n)
{
int *stack = new int[n];
int *tag = new int[n];
char *buf = new char[n];
bool flag = true;
int top = -1;
int p = 1;
while (a[p].lc > 0 || top >= 0)
{
while (a[p].lc > 0) //先一直寻找左孩子
{
flag = true; //此时p指向的是新叶子(未输出过的叶子)
stack[++top] = p; //结点入栈
p = a[p].lc;
tag[top] = 0; //表示右孩子没有被访问
buf[top] = '0'; //左孩子标记'0'
}
if (flag) //如果p指向的是新叶子
{
//cout << a[p].data << " : "; //输出叶子结点
// for (int i=0; i<=top; i++)
// cout << buf[i];
// cout << endl;
Btree s = new struct cNode;
s->data = a[p].data;
for (int i=0; i<=top; i++)
s->str += buf[i];
s->lc = s->rc = NULL;
if (!(InsertBtNode(t, s))) //插入一个结点s
delete s;
}
if (top >= 0) //所有左孩子处理完毕后
{
if (tag[top] == 0) //如果右孩子没有被访问
{
flag = true; //此时p指向的是新叶子(未输出过的叶子)
p = stack[top]; //读取栈顶元素,但不退栈 ,因为要先输出其右孩子结点
p = a[p].rc;
tag[top] = 1; //表示右孩子被访问,下次直接退栈
buf[top] = '1'; //右孩子标记'1'
}
else //栈顶元素出栈
{
flag = false; //此时p指向的是旧叶子(已输出过的叶子),不再输出
top--;
}
}
}
}
//先序遍历赫夫曼树
void Preorder(HuffmanTree t, int p)
{
if (t == NULL)
return;
if (t[p].lc > 0)
{
cout << t[p].weight << endl;
Preorder(t, t[p].lc); //遍历左子树
Preorder(t, t[p].rc); //遍历右子树
}
else
cout << t[p].weight << " " << t[p].data << endl;
}
//向一个二叉排序树t中插入一个结点s
bool InsertBtNode(Btree & t, Btree s)
{
if (t == NULL)
{
t = s;
return true;
}
else if (t->data > s->data) //把s所指结点插入到左子树中
return InsertBtNode(t->lc, s);
else if (t->data < s->data) //把s所指结点插入到右子树中
return InsertBtNode(t->rc, s);
else //若s->data等于b的根结点的数据域之值,则什么也不做
return false;
}
//中序遍历二叉排序树
void Inorder(Btree t)
{
if (t)
{
Inorder(t->lc); //遍历左子树
cout << t->data << " : " << t->str << endl; //输出该结点
Inorder(t->rc); //遍历右子树
}
}
//查找值为data的结点的递归算法
Btree Search(Btree p, ElemType data)
{
if (p == NULL || p->data == data) //空树或找到结点
return p;
if (p->data > data)
return Search(p->lc, data); //在左孩子中寻找
else
return Search(p->rc, data); //在右孩子中寻找
}
//利用记录了叶子结点编码的排序二叉树,对sourceCode进行编码,返回编码后的字符串
string Coding(string s, Btree t)
{
Btree p = NULL;
string dest;
for (int i=0; i<s.size(); i++)
{
p = Search(t, s[i]);
if (p != NULL)
{
dest += p->str;
//dest += ' ';
}
}
return dest;
}
//利用赫夫曼树对destCode进行解码
string Decode(string s, HuffmanTree hT)
{
string dest;
int p = 1;
int i = 0;
while (i < s.size())
{
while (hT[p].lc > 0)//非叶子结点
{
if (s[i++] == '0')
p = hT[p].lc; //向左结点前进
else
p = hT[p].rc; //向右结点前进
}
dest += hT[p].data; //存储叶子结点
p = 1;
}
return dest;
}
//销毁一棵二叉排序树
void DestroyBTree(Btree & t)
{
if (t != NULL)
{
DestroyBTree(t->lc);
DestroyBTree(t->rc);
delete t;
t = NULL;
}
}
//销毁一棵赫夫曼树
void DestroyHfmanTree(HuffmanTree & t, int n)
{
for (int i=n-1; i>=0; i--)
{
delete &t[i];
}
t = NULL;
}