Huffman Coding 哈夫曼编码

作者：jostree 转载请注明出处 http://www.cnblogs.com/jostree/p/4096079.html

使用优先队列实现，需要注意以下几点：

1.在使用priority_queue时，内部需要存储哈夫曼树节点的指针，而不能是节点。因为构建哈夫曼树时，需要把其左右指针指向孩子，而如果储存的是节点，那么孩子的地址是会改变的。同理节点应当使用new在内存中开辟，而不能使用vector，原因是vector在数组大小为2整数次幂时，大小会倍增，开辟新数组并把老数组的数字copy过去，从而也会导致地址变化。

2.优先队列对指针的排列，需要额外写一个比较函数来比较指针指向的节点的大小。bool operator () (wcnode * node1, wcnode * node2) return node1->lessthan(node2);并在定义优先队列时使用这种方法：    priority_queue <wcnode*, vector<wcnode*>, compare> 第一个参数是节点类型，第二个参数是优先队列的储存结构，第三个参数是比较函数。

3.C++在写入文件时，由于只能按字节写入，因此需要把8个bit位转化为一个字节，最后不足8位用0补齐，并记录文件总bit数，便于解码。然后写入文件。另写入二进制文件可以使用ofstream out("output.txt",std::ofstream::binary);

4.哈夫曼编码信息包括每种字符的映射，和该文件的总bit数。

其代码如下：

  1 #include <cstdio>
  2 #include <cstdlib>
  3 #include <iostream>
  4 #include <cstring>
  5 #include  <fstream>
  6 #include  <queue>
  7 #include  <map>
  8 #include  <vector>
  9 using namespace std;
 10 class compare;
 11
 12 class wcnode
 13 {
 14     public:
 15         friend class compare;
 16         char word;
 17         int count;
 18         wcnode* left;
 19         wcnode* right;
 20         bool lessthan (const wcnode *w)const
 21         {
 22             return count > w->count;
 23         }
 24         wcnode(char w=‘\0‘, int c=0, wcnode* l=NULL, wcnode * r=NULL)
 25         {
 26             word = w; count = c; left = l; right = r;
 27         }
 28 };
 29
 30 class compare
 31 {
 32     public:
 33         bool operator () (wcnode * node1, wcnode * node2)
 34         {
 35             return node1->lessthan(node2);
 36         }
 37 };
 38
 39 void preorder(wcnode *head, vector<bool> rec, map<char, vector<bool> > & res)
 40 {
 41     if( head->left == NULL && head->right == NULL )
 42     {
 43         res[head->word] = rec;
 44         return;
 45     }
 46     vector<bool> l = rec;
 47     l.push_back(0);
 48     vector<bool> r = rec;
 49     r.push_back(1);
 50     if(head->left != NULL) preorder(head->left, l, res);
 51     if(head->right != NULL) preorder(head->right, r, res);
 52 }
 53 map<char, vector<bool> > encode(map<char, int> &wordcount)
 54 {
 55     map<char, vector<bool> > res;
 56     priority_queue <wcnode*, vector<wcnode*>, compare> pq;
 57     map<char, int>::iterator t;
 58     wcnode *tmp;
 59     wcnode *t1, *t2, *t3;
 60
 61     for( t = wordcount.begin() ; t != wordcount.end() ; t++ )
 62     {
 63         tmp = new wcnode();
 64         tmp->word = t->first;
 65         tmp->count = t->second;
 66         pq.push(tmp);
 67     }
 68     while( pq.size() > 1 )
 69     {
 70         t1 = pq.top();
 71         pq.pop();
 72         t2 = pq.top();
 73         pq.pop();
 74         t3 = new wcnode();
 75         t3->count = t1->count + t2->count;
 76         t3->left = t1;
 77         t3->right = t2;
 78         pq.push(t3);
 79     }
 80     wcnode *huffmanhead = pq.top();
 81     vector<bool> rec;
 82     preorder(huffmanhead, rec, res);
 83     map<char, vector<bool>  >::iterator it;
 84     for( it = res.begin() ; it != res.end() ; it++ )
 85     {
 86         cout<<it->first<<":";
 87         for( int i = 0; i < it->second.size() ; i++ )
 88         {
 89             cout<<it->second[i];
 90         }
 91         cout<<", ";
 92     }
 93     return res;
 94 }
 95
 96 void output(string s, string passage, map<char, vector<bool> > res)
 97 {
 98     ofstream out(s.c_str());
 99     vector<bool> bit;
100     for( int i = 0 ; i < passage.size() ; i++ )
101     {
102         vector<bool> tmp = res[passage[i]];
103         for( int i = 0 ; i < tmp.size(); i++ )
104         {
105             bit.push_back(tmp[i]);
106         }
107     }
108     char outputchar = 0;
109     for( int i = 0 ; i < bit.size() ; i++ )
110     {
111         if( i % 8 == 7 )
112         {
113            out.write(&outputchar, sizeof(outputchar));
114            outputchar = 0;
115         }
116         outputchar = outputchar + bit[i];
117         outputchar = outputchar * 2;
118     }
119     if( outputchar != 0 )
120     {
121         out.write(&outputchar, sizeof(outputchar));
122     }
123     out.close();
124 }
125 int main(int argc, char *argv[])
126 {
127     char tmp;
128     ifstream in("Aesop_Fables.txt");
129     map <char, int> wordcount;
130     map <char, vector<bool> > res;
131     string passage;
132     while( in.get(tmp) )
133     {
134         passage += tmp;
135         if( wordcount.count(tmp) == 0  )
136         {
137             wordcount[tmp] = 1;
138         }
139         else
140         {
141             wordcount[tmp]++;
142         }
143     }
144     res = encode(wordcount);
145     output("outAesop.txt", passage, res);
146     in.close();
147 }