哈夫曼编码压缩解压缩实现&不同类型文件压缩比的测试

压缩原理及步骤&&压缩比的计算

压缩原理及步骤

压缩的第一步：

压缩第二步：

压缩第三步：

压缩比的计算

强调

考虑ascii码外的其它语言

需要注意的地方

代码对不同类型文档的压缩比测试情况

英语文章

源代码

压缩程序源文件 compress.cpp

#include <iostream>
#include <locale>
#include <cstdlib>
#include <fstream>
#include <vector>
#include <queue>

using namespace std;

const long long MAX_SIZE = 10000000000;//
const int MAX_TYPE = 300;
unsigned int *f = new unsigned int[MAX_TYPE];//计数
unsigned int *p = new unsigned int[MAX_TYPE];//计下标
char *v = new char[MAX_TYPE];
char filename[20];
char *s[MAX_TYPE];

struct Node
{
unsigned int weight, parent, lson, rson;
Node(){};
}HuffmanTree[MAX_TYPE<<1];

struct NodeCmp
{
bool operator()(int a, int b)
{
return HuffmanTree[a].weight > HuffmanTree.weight;
}
};

int CreatTree(char *str, long long len)
{
int num = 1;
for(int i=0;i<len;i++)
f[str[i]]++;
cout<<"len::"<<len<<endl;
for(int i=0;i<len;i++)
{
if(f[str[i]])
{
HuffmanTree[num].weight = f[str[i]];
HuffmanTree[num].lson = 0;
HuffmanTree[num].rson = 0;
f[str[i]] = 0;
if(p[str[i]] == 0)
p[str[i]] = num;
v[num] = str[i];
++num;
}
}
cout<<"num::"<<num<<endl;
return num;
}

void CodingTree(int num)
{
priority_queue<int, vector<int>, NodeCmp> q;
for(int i=1;i<num;i++)
q.push(i);
int len = num;
for(int i=0;i<num-2;i++)
{
int x = q.top(); q.pop();
int y = q.top(); q.pop();
HuffmanTree[len].weight = HuffmanTree[x].weight + HuffmanTree[y].weight;
HuffmanTree[x].parent = HuffmanTree[y].parent = len;
HuffmanTree[len].lson = y;
HuffmanTree[len].rson = x;
q.push(len++);
}
}

void FindPath(int num)
{
char *t = new char[num];
t[num-1] = '\0';
for(int i=1;i<num;i++)
{
int son = i, father = HuffmanTree[i].parent;
int start = num-1;

while(father != 0)
{
--start;
if(HuffmanTree[father].rson == son)
t[start] = '1';
else
t[start] = '0';
son = father;
father = HuffmanTree[father].parent;
}
s[i] = new char[num - start];
strcpy(s[i], &t[start]);
}
}

void print(int num, long long len, char *str)
{
ofstream fout(filename, ios::out);
fout<<num<<endl;
for(int i=1;i<num;i++)
{
fout<<s[i]<<endl;
fout<<v[i]<<endl;
}
long long pos = 0;
char *ans = new char[MAX_SIZE];

int now = 7;
for(long long i=0;i<len;i++)
{
int k = 0;
while(s[p[str[i]]][k] != '\0')
{
ans[pos] |= (s[p[str[i]]][k]-'0')<<now--;
if(now < 0)
{
now = 7;
pos++;
}
++k;
}
}

int zero = 0;
if(now != 7) zero = now%7+1, pos++;

fout<<zero<<" "<<pos<<endl;
fout.write(ans, sizeof(char)*pos);
fout.close();

cout<<"zero::"<<zero<<endl;
}

int main(int argc, char **argv)
{
sprintf(filename, "%s.temp", argv[1]);
ifstream fin(argv[1],ios::ate | ios::in);
if(!fin)
{
cout<<"File open error!"<<endl;
return 0;
}

long long size = fin.tellg();
if(size > MAX_SIZE)
{
cout<<"Too long!"<<endl;
return 0;
}
fin.seekg(0, ios::beg);

char *str = new char[size+1];
fin.read(str,size);
fin.close();

int num = CreatTree(str, size);
CodingTree(num);
FindPath(num);
print(num, size, str);

return 0;
}

#include <iostream>
#include <locale>
#include <cstdlib>
#include <fstream>
#include <vector>
#include <queue>

using namespace std;
char filename[20];
const long long MAX_SIZE = 10000000000;//
const int MAX_TYPE = 300;
struct Node
{
char v;
int parent, lson, rson;
Node(){};
}HuffmanTree[MAX_TYPE<<1];

char *str = new char[MAX_SIZE];
char *ans = new char[MAX_SIZE];

void CreatTree(char *t, char v, int &pos)
{
int root = 0;
for(int i=0;t[i]!='\0';i++)
{
if(t[i] == '1')
{
if(HuffmanTree[root].rson == 0)
HuffmanTree[root].rson = pos++;
root = HuffmanTree[root].rson;
}
else
{
if(HuffmanTree[root].lson == 0)
HuffmanTree[root].lson = pos++;
root = HuffmanTree[root].lson;
}
}
HuffmanTree[root].v = v;
}

void print(int zero, int len, char *str)
{
long long start = 0;
int root = 0;
int end = 0;
for(int i=0;i<len;i++)
{
char t = str[i];
if(i == len-1)
end = zero;
for(int j=7;j>=end;j--)
{
if((1<<j) & t)
root = HuffmanTree[root].rson;
else
root = HuffmanTree[root].lson;
if(HuffmanTree[root].lson == 0 && HuffmanTree[root].rson == 0)
{
ans[start++] = HuffmanTree[root].v;
root = 0;
}
}
}
cout<<"len::"<<start<<endl;
ofstream out(filename, ios::out);
out.write(ans, sizeof(char)*(start));
out.close();
}

int main(int argc, char **argv)
{
strcpy(filename, argv[1]);
filename[strlen(filename)-4] = 'o';
filename[strlen(filename)-3] = 'u';
filename[strlen(filename)-2] = 't';
filename[strlen(filename)-1] = '\0';

ifstream fin(argv[1], ios::in);
if(!fin)
{
cout<<"File open error!"<<endl;
return 0;
}
int num;
char *t = new char[num];
char *v = new char[3];
fin>>num;
fin.getline(t,num);
cout<<"size::"<<num<<endl;
int pos = 1;
for(int i=1;i<num;i++)
{
fin.getline(t,num);
fin.getline(v,num);
if(v[0] == '\0')
{
fin.getline(v,num);
v[0] = '\n';
}
CreatTree(t, v[0], pos);
v[0]=0;
}

int zero;
long long size;
fin>>zero; fin>>size;
fin.getline(t,num);
fin.read(str,sizeof(char)*size);
print(zero, size, str);

cout<<"zero::"<<zero<<endl;

return 0;
}